109 files changed, 51525 insertions, 32573 deletions

diff --git a/fs/btrfs/Kconfig b/fs/btrfs/Kconfig
index 575636f6491e..183e5c4aed34 100644
--- a/fs/btrfs/Kconfig
+++ b/fs/btrfs/Kconfig
@@ -14,9 +14,11 @@ config BTRFS_FS
 	select LZO_DECOMPRESS
 	select ZSTD_COMPRESS
 	select ZSTD_DECOMPRESS
+	select FS_IOMAP
 	select RAID6_PQ
 	select XOR_BLOCKS
 	select SRCU
+	depends on PAGE_SIZE_LESS_THAN_256KB
 
 	help
 	  Btrfs is a general purpose copy-on-write filesystem with extents,
diff --git a/fs/btrfs/Makefile b/fs/btrfs/Makefile
index 9a0ff3384381..fa9ddcc9eb0b 100644
--- a/fs/btrfs/Makefile
+++ b/fs/btrfs/Makefile
@@ -1,21 +1,43 @@
 # SPDX-License-Identifier: GPL-2.0
 
+# Subset of W=1 warnings
+subdir-ccflags-y += -Wextra -Wunused -Wno-unused-parameter
+subdir-ccflags-y += -Wmissing-declarations
+subdir-ccflags-y += -Wmissing-format-attribute
+subdir-ccflags-y += -Wmissing-prototypes
+subdir-ccflags-y += -Wold-style-definition
+subdir-ccflags-y += -Wmissing-include-dirs
+condflags := \
+	$(call cc-option, -Wunused-but-set-variable)		\
+	$(call cc-option, -Wunused-const-variable)		\
+	$(call cc-option, -Wpacked-not-aligned)			\
+	$(call cc-option, -Wstringop-truncation)
+subdir-ccflags-y += $(condflags)
+# The following turn off the warnings enabled by -Wextra
+subdir-ccflags-y += -Wno-missing-field-initializers
+subdir-ccflags-y += -Wno-sign-compare
+subdir-ccflags-y += -Wno-type-limits
+subdir-ccflags-y += -Wno-shift-negative-value
+
 obj-$(CONFIG_BTRFS_FS) := btrfs.o
 
 btrfs-y += super.o ctree.o extent-tree.o print-tree.o root-tree.o dir-item.o \
-	   file-item.o inode-item.o inode-map.o disk-io.o \
+	   file-item.o inode-item.o disk-io.o \
 	   transaction.o inode.o file.o tree-defrag.o \
 	   extent_map.o sysfs.o struct-funcs.o xattr.o ordered-data.o \
 	   extent_io.o volumes.o async-thread.o ioctl.o locking.o orphan.o \
 	   export.o tree-log.o free-space-cache.o zlib.o lzo.o zstd.o \
 	   compression.o delayed-ref.o relocation.o delayed-inode.o scrub.o \
-	   reada.o backref.o ulist.o qgroup.o send.o dev-replace.o raid56.o \
+	   backref.o ulist.o qgroup.o send.o dev-replace.o raid56.o \
 	   uuid-tree.o props.o free-space-tree.o tree-checker.o space-info.o \
-	   block-rsv.o delalloc-space.o block-group.o discard.o
+	   block-rsv.o delalloc-space.o block-group.o discard.o reflink.o \
+	   subpage.o tree-mod-log.o extent-io-tree.o
 
 btrfs-$(CONFIG_BTRFS_FS_POSIX_ACL) += acl.o
 btrfs-$(CONFIG_BTRFS_FS_CHECK_INTEGRITY) += check-integrity.o
 btrfs-$(CONFIG_BTRFS_FS_REF_VERIFY) += ref-verify.o
+btrfs-$(CONFIG_BLK_DEV_ZONED) += zoned.o
+btrfs-$(CONFIG_FS_VERITY) += verity.o
 
 btrfs-$(CONFIG_BTRFS_FS_RUN_SANITY_TESTS) += tests/free-space-tests.o \
 	tests/extent-buffer-tests.o tests/btrfs-tests.o \
diff --git a/fs/btrfs/acl.c b/fs/btrfs/acl.c
index a0af1b952c4d..548d6a5477b4 100644
--- a/fs/btrfs/acl.c
+++ b/fs/btrfs/acl.c
@@ -16,13 +16,16 @@
 #include "btrfs_inode.h"
 #include "xattr.h"
 
-struct posix_acl *btrfs_get_acl(struct inode *inode, int type)
+struct posix_acl *btrfs_get_acl(struct inode *inode, int type, bool rcu)
 {
 	int size;
 	const char *name;
 	char *value = NULL;
 	struct posix_acl *acl;
 
+	if (rcu)
+		return ERR_PTR(-ECHILD);
+
 	switch (type) {
 	case ACL_TYPE_ACCESS:
 		name = XATTR_NAME_POSIX_ACL_ACCESS;
@@ -52,8 +55,8 @@ struct posix_acl *btrfs_get_acl(struct inode *inode, int type)
 	return acl;
 }
 
-static int __btrfs_set_acl(struct btrfs_trans_handle *trans,
-			 struct inode *inode, struct posix_acl *acl, int type)
+int __btrfs_set_acl(struct btrfs_trans_handle *trans, struct inode *inode,
+		    struct posix_acl *acl, int type)
 {
 	int ret, size = 0;
 	const char *name;
@@ -107,13 +110,15 @@ out:
 	return ret;
 }
 
-int btrfs_set_acl(struct inode *inode, struct posix_acl *acl, int type)
+int btrfs_set_acl(struct user_namespace *mnt_userns, struct inode *inode,
+		  struct posix_acl *acl, int type)
 {
 	int ret;
 	umode_t old_mode = inode->i_mode;
 
 	if (type == ACL_TYPE_ACCESS && acl) {
-		ret = posix_acl_update_mode(inode, &inode->i_mode, &acl);
+		ret = posix_acl_update_mode(mnt_userns, inode,
+					    &inode->i_mode, &acl);
 		if (ret)
 			return ret;
 	}
@@ -122,35 +127,3 @@ int btrfs_set_acl(struct inode *inode, struct posix_acl *acl, int type)
 		inode->i_mode = old_mode;
 	return ret;
 }
-
-int btrfs_init_acl(struct btrfs_trans_handle *trans,
-		   struct inode *inode, struct inode *dir)
-{
-	struct posix_acl *default_acl, *acl;
-	int ret = 0;
-
-	/* this happens with subvols */
-	if (!dir)
-		return 0;
-
-	ret = posix_acl_create(dir, &inode->i_mode, &default_acl, &acl);
-	if (ret)
-		return ret;
-
-	if (default_acl) {
-		ret = __btrfs_set_acl(trans, inode, default_acl,
-				      ACL_TYPE_DEFAULT);
-		posix_acl_release(default_acl);
-	}
-
-	if (acl) {
-		if (!ret)
-			ret = __btrfs_set_acl(trans, inode, acl,
-					      ACL_TYPE_ACCESS);
-		posix_acl_release(acl);
-	}
-
-	if (!default_acl && !acl)
-		cache_no_acl(inode);
-	return ret;
-}
diff --git a/fs/btrfs/async-thread.c b/fs/btrfs/async-thread.c
index 1d32a07bb2d1..aac240430efe 100644
--- a/fs/btrfs/async-thread.c
+++ b/fs/btrfs/async-thread.c
@@ -15,13 +15,12 @@
 enum {
 	WORK_DONE_BIT,
 	WORK_ORDER_DONE_BIT,
-	WORK_HIGH_PRIO_BIT,
 };
 
 #define NO_THRESHOLD (-1)
 #define DFT_THRESHOLD (32)
 
-struct __btrfs_workqueue {
+struct btrfs_workqueue {
 	struct workqueue_struct *normal_wq;
 
 	/* File system this workqueue services */
@@ -48,12 +47,7 @@ struct __btrfs_workqueue {
 	spinlock_t thres_lock;
 };
 
-struct btrfs_workqueue {
-	struct __btrfs_workqueue *normal;
-	struct __btrfs_workqueue *high;
-};
-
-struct btrfs_fs_info * __pure btrfs_workqueue_owner(const struct __btrfs_workqueue *wq)
+struct btrfs_fs_info * __pure btrfs_workqueue_owner(const struct btrfs_workqueue *wq)
 {
 	return wq->fs_info;
 }
@@ -66,22 +60,22 @@ struct btrfs_fs_info * __pure btrfs_work_owner(const struct btrfs_work *work)
 bool btrfs_workqueue_normal_congested(const struct btrfs_workqueue *wq)
 {
 	/*
-	 * We could compare wq->normal->pending with num_online_cpus()
+	 * We could compare wq->pending with num_online_cpus()
 	 * to support "thresh == NO_THRESHOLD" case, but it requires
 	 * moving up atomic_inc/dec in thresh_queue/exec_hook. Let's
 	 * postpone it until someone needs the support of that case.
 	 */
-	if (wq->normal->thresh == NO_THRESHOLD)
+	if (wq->thresh == NO_THRESHOLD)
 		return false;
 
-	return atomic_read(&wq->normal->pending) > wq->normal->thresh * 2;
+	return atomic_read(&wq->pending) > wq->thresh * 2;
 }
 
-static struct __btrfs_workqueue *
-__btrfs_alloc_workqueue(struct btrfs_fs_info *fs_info, const char *name,
-			unsigned int flags, int limit_active, int thresh)
+struct btrfs_workqueue *btrfs_alloc_workqueue(struct btrfs_fs_info *fs_info,
+					      const char *name, unsigned int flags,
+					      int limit_active, int thresh)
 {
-	struct __btrfs_workqueue *ret = kzalloc(sizeof(*ret), GFP_KERNEL);
+	struct btrfs_workqueue *ret = kzalloc(sizeof(*ret), GFP_KERNEL);
 
 	if (!ret)
 		return NULL;
@@ -105,12 +99,8 @@ __btrfs_alloc_workqueue(struct btrfs_fs_info *fs_info, const char *name,
 		ret->thresh = thresh;
 	}
 
-	if (flags & WQ_HIGHPRI)
-		ret->normal_wq = alloc_workqueue("btrfs-%s-high", flags,
-						 ret->current_active, name);
-	else
-		ret->normal_wq = alloc_workqueue("btrfs-%s", flags,
-						 ret->current_active, name);
+	ret->normal_wq = alloc_workqueue("btrfs-%s", flags, ret->current_active,
+					 name);
 	if (!ret->normal_wq) {
 		kfree(ret);
 		return NULL;
@@ -119,41 +109,7 @@ __btrfs_alloc_workqueue(struct btrfs_fs_info *fs_info, const char *name,
 	INIT_LIST_HEAD(&ret->ordered_list);
 	spin_lock_init(&ret->list_lock);
 	spin_lock_init(&ret->thres_lock);
-	trace_btrfs_workqueue_alloc(ret, name, flags & WQ_HIGHPRI);
-	return ret;
-}
-
-static inline void
-__btrfs_destroy_workqueue(struct __btrfs_workqueue *wq);
-
-struct btrfs_workqueue *btrfs_alloc_workqueue(struct btrfs_fs_info *fs_info,
-					      const char *name,
-					      unsigned int flags,
-					      int limit_active,
-					      int thresh)
-{
-	struct btrfs_workqueue *ret = kzalloc(sizeof(*ret), GFP_KERNEL);
-
-	if (!ret)
-		return NULL;
-
-	ret->normal = __btrfs_alloc_workqueue(fs_info, name,
-					      flags & ~WQ_HIGHPRI,
-					      limit_active, thresh);
-	if (!ret->normal) {
-		kfree(ret);
-		return NULL;
-	}
-
-	if (flags & WQ_HIGHPRI) {
-		ret->high = __btrfs_alloc_workqueue(fs_info, name, flags,
-						    limit_active, thresh);
-		if (!ret->high) {
-			__btrfs_destroy_workqueue(ret->normal);
-			kfree(ret);
-			return NULL;
-		}
-	}
+	trace_btrfs_workqueue_alloc(ret, name);
 	return ret;
 }
 
@@ -162,7 +118,7 @@ struct btrfs_workqueue *btrfs_alloc_workqueue(struct btrfs_fs_info *fs_info,
  * This hook WILL be called in IRQ handler context,
  * so workqueue_set_max_active MUST NOT be called in this hook
  */
-static inline void thresh_queue_hook(struct __btrfs_workqueue *wq)
+static inline void thresh_queue_hook(struct btrfs_workqueue *wq)
 {
 	if (wq->thresh == NO_THRESHOLD)
 		return;
@@ -174,7 +130,7 @@ static inline void thresh_queue_hook(struct __btrfs_workqueue *wq)
  * This hook is called in kthread content.
  * So workqueue_set_max_active is called here.
  */
-static inline void thresh_exec_hook(struct __btrfs_workqueue *wq)
+static inline void thresh_exec_hook(struct btrfs_workqueue *wq)
 {
 	int new_current_active;
 	long pending;
@@ -217,7 +173,7 @@ out:
 	}
 }
 
-static void run_ordered_work(struct __btrfs_workqueue *wq,
+static void run_ordered_work(struct btrfs_workqueue *wq,
 			     struct btrfs_work *self)
 {
 	struct list_head *list = &wq->ordered_list;
@@ -234,6 +190,13 @@ static void run_ordered_work(struct __btrfs_workqueue *wq,
 				  ordered_list);
 		if (!test_bit(WORK_DONE_BIT, &work->flags))
 			break;
+		/*
+		 * Orders all subsequent loads after reading WORK_DONE_BIT,
+		 * paired with the smp_mb__before_atomic in btrfs_work_helper
+		 * this guarantees that the ordered function will see all
+		 * updates from ordinary work function.
+		 */
+		smp_rmb();
 
 		/*
 		 * we are going to call the ordered done function, but
@@ -298,7 +261,7 @@ static void btrfs_work_helper(struct work_struct *normal_work)
 {
 	struct btrfs_work *work = container_of(normal_work, struct btrfs_work,
 					       normal_work);
-	struct __btrfs_workqueue *wq;
+	struct btrfs_workqueue *wq = work->wq;
 	int need_order = 0;
 
 	/*
@@ -311,12 +274,18 @@ static void btrfs_work_helper(struct work_struct *normal_work)
 	 */
 	if (work->ordered_func)
 		need_order = 1;
-	wq = work->wq;
 
 	trace_btrfs_work_sched(work);
 	thresh_exec_hook(wq);
 	work->func(work);
 	if (need_order) {
+		/*
+		 * Ensures all memory accesses done in the work function are
+		 * ordered before setting the WORK_DONE_BIT. Ensuring the thread
+		 * which is going to executed the ordered work sees them.
+		 * Pairs with the smp_rmb in run_ordered_work.
+		 */
+		smp_mb__before_atomic();
 		set_bit(WORK_DONE_BIT, &work->flags);
 		run_ordered_work(wq, work);
 	} else {
@@ -336,8 +305,7 @@ void btrfs_init_work(struct btrfs_work *work, btrfs_func_t func,
 	work->flags = 0;
 }
 
-static inline void __btrfs_queue_work(struct __btrfs_workqueue *wq,
-				      struct btrfs_work *work)
+void btrfs_queue_work(struct btrfs_workqueue *wq, struct btrfs_work *work)
 {
 	unsigned long flags;
 
@@ -352,46 +320,22 @@ static inline void __btrfs_queue_work(struct __btrfs_workqueue *wq,
 	queue_work(wq->normal_wq, &work->normal_work);
 }
 
-void btrfs_queue_work(struct btrfs_workqueue *wq,
-		      struct btrfs_work *work)
-{
-	struct __btrfs_workqueue *dest_wq;
-
-	if (test_bit(WORK_HIGH_PRIO_BIT, &work->flags) && wq->high)
-		dest_wq = wq->high;
-	else
-		dest_wq = wq->normal;
-	__btrfs_queue_work(dest_wq, work);
-}
-
-static inline void
-__btrfs_destroy_workqueue(struct __btrfs_workqueue *wq)
-{
-	destroy_workqueue(wq->normal_wq);
-	trace_btrfs_workqueue_destroy(wq);
-	kfree(wq);
-}
-
 void btrfs_destroy_workqueue(struct btrfs_workqueue *wq)
 {
 	if (!wq)
 		return;
-	if (wq->high)
-		__btrfs_destroy_workqueue(wq->high);
-	__btrfs_destroy_workqueue(wq->normal);
+	destroy_workqueue(wq->normal_wq);
+	trace_btrfs_workqueue_destroy(wq);
 	kfree(wq);
 }
 
 void btrfs_workqueue_set_max(struct btrfs_workqueue *wq, int limit_active)
 {
-	if (!wq)
-		return;
-	wq->normal->limit_active = limit_active;
-	if (wq->high)
-		wq->high->limit_active = limit_active;
+	if (wq)
+		wq->limit_active = limit_active;
 }
 
-void btrfs_set_work_high_priority(struct btrfs_work *work)
+void btrfs_flush_workqueue(struct btrfs_workqueue *wq)
 {
-	set_bit(WORK_HIGH_PRIO_BIT, &work->flags);
+	flush_workqueue(wq->normal_wq);
 }
diff --git a/fs/btrfs/async-thread.h b/fs/btrfs/async-thread.h
index a4434301d84d..6e2596ddae10 100644
--- a/fs/btrfs/async-thread.h
+++ b/fs/btrfs/async-thread.h
@@ -11,11 +11,8 @@
 
 struct btrfs_fs_info;
 struct btrfs_workqueue;
-/* Internal use only */
-struct __btrfs_workqueue;
 struct btrfs_work;
 typedef void (*btrfs_func_t)(struct btrfs_work *arg);
-typedef void (*btrfs_work_func_t)(struct work_struct *arg);
 
 struct btrfs_work {
 	btrfs_func_t func;
@@ -25,7 +22,7 @@ struct btrfs_work {
 	/* Don't touch things below */
 	struct work_struct normal_work;
 	struct list_head ordered_list;
-	struct __btrfs_workqueue *wq;
+	struct btrfs_workqueue *wq;
 	unsigned long flags;
 };
 
@@ -40,9 +37,9 @@ void btrfs_queue_work(struct btrfs_workqueue *wq,
 		      struct btrfs_work *work);
 void btrfs_destroy_workqueue(struct btrfs_workqueue *wq);
 void btrfs_workqueue_set_max(struct btrfs_workqueue *wq, int max);
-void btrfs_set_work_high_priority(struct btrfs_work *work);
 struct btrfs_fs_info * __pure btrfs_work_owner(const struct btrfs_work *work);
-struct btrfs_fs_info * __pure btrfs_workqueue_owner(const struct __btrfs_workqueue *wq);
+struct btrfs_fs_info * __pure btrfs_workqueue_owner(const struct btrfs_workqueue *wq);
 bool btrfs_workqueue_normal_congested(const struct btrfs_workqueue *wq);
+void btrfs_flush_workqueue(struct btrfs_workqueue *wq);
 
 #endif
diff --git a/fs/btrfs/backref.c b/fs/btrfs/backref.c
index e5d85311d5d5..18374a6d05bd 100644
--- a/fs/btrfs/backref.c
+++ b/fs/btrfs/backref.c
@@ -13,6 +13,8 @@
 #include "transaction.h"
 #include "delayed-ref.h"
 #include "locking.h"
+#include "misc.h"
+#include "tree-mod-log.h"
 
 /* Just an arbitrary number so we can be sure this happened */
 #define BACKREF_FOUND_SHARED 6
@@ -136,6 +138,7 @@ struct share_check {
 	u64 root_objectid;
 	u64 inum;
 	int share_count;
+	bool have_delayed_delete_refs;
 };
 
 static inline int extent_is_shared(struct share_check *sc)
@@ -286,8 +289,10 @@ static void prelim_release(struct preftree *preftree)
 	struct prelim_ref *ref, *next_ref;
 
 	rbtree_postorder_for_each_entry_safe(ref, next_ref,
-					     &preftree->root.rb_root, rbnode)
+					     &preftree->root.rb_root, rbnode) {
+		free_inode_elem_list(ref->inode_list);
 		free_pref(ref);
+	}
 
 	preftree->root = RB_ROOT_CACHED;
 	preftree->count = 0;
@@ -347,33 +352,10 @@ static int add_prelim_ref(const struct btrfs_fs_info *fs_info,
 		return -ENOMEM;
 
 	ref->root_id = root_id;
-	if (key) {
+	if (key)
 		ref->key_for_search = *key;
-		/*
-		 * We can often find data backrefs with an offset that is too
-		 * large (>= LLONG_MAX, maximum allowed file offset) due to
-		 * underflows when subtracting a file's offset with the data
-		 * offset of its corresponding extent data item. This can
-		 * happen for example in the clone ioctl.
-		 * So if we detect such case we set the search key's offset to
-		 * zero to make sure we will find the matching file extent item
-		 * at add_all_parents(), otherwise we will miss it because the
-		 * offset taken form the backref is much larger then the offset
-		 * of the file extent item. This can make us scan a very large
-		 * number of file extent items, but at least it will not make
-		 * us miss any.
-		 * This is an ugly workaround for a behaviour that should have
-		 * never existed, but it does and a fix for the clone ioctl
-		 * would touch a lot of places, cause backwards incompatibility
-		 * and would not fix the problem for extents cloned with older
-		 * kernels.
-		 */
-		if (ref->key_for_search.type == BTRFS_EXTENT_DATA_KEY &&
-		    ref->key_for_search.offset >= LLONG_MAX)
-			ref->key_for_search.offset = 0;
-	} else {
+	else
 		memset(&ref->key_for_search, 0, sizeof(ref->key_for_search));
-	}
 
 	ref->inode_list = NULL;
 	ref->level = level;
@@ -409,10 +391,36 @@ static int add_indirect_ref(const struct btrfs_fs_info *fs_info,
 			      wanted_disk_byte, count, sc, gfp_mask);
 }
 
+static int is_shared_data_backref(struct preftrees *preftrees, u64 bytenr)
+{
+	struct rb_node **p = &preftrees->direct.root.rb_root.rb_node;
+	struct rb_node *parent = NULL;
+	struct prelim_ref *ref = NULL;
+	struct prelim_ref target = {};
+	int result;
+
+	target.parent = bytenr;
+
+	while (*p) {
+		parent = *p;
+		ref = rb_entry(parent, struct prelim_ref, rbnode);
+		result = prelim_ref_compare(ref, &target);
+
+		if (result < 0)
+			p = &(*p)->rb_left;
+		else if (result > 0)
+			p = &(*p)->rb_right;
+		else
+			return 1;
+	}
+	return 0;
+}
+
 static int add_all_parents(struct btrfs_root *root, struct btrfs_path *path,
-			   struct ulist *parents, struct prelim_ref *ref,
+			   struct ulist *parents,
+			   struct preftrees *preftrees, struct prelim_ref *ref,
 			   int level, u64 time_seq, const u64 *extent_item_pos,
-			   u64 total_refs, bool ignore_offset)
+			   bool ignore_offset)
 {
 	int ret = 0;
 	int slot;
@@ -424,6 +432,7 @@ static int add_all_parents(struct btrfs_root *root, struct btrfs_path *path,
 	u64 disk_byte;
 	u64 wanted_disk_byte = ref->wanted_disk_byte;
 	u64 count = 0;
+	u64 data_offset;
 
 	if (level != 0) {
 		eb = path->nodes[level];
@@ -434,18 +443,26 @@ static int add_all_parents(struct btrfs_root *root, struct btrfs_path *path,
 	}
 
 	/*
-	 * We normally enter this function with the path already pointing to
-	 * the first item to check. But sometimes, we may enter it with
-	 * slot==nritems. In that case, go to the next leaf before we continue.
+	 * 1. We normally enter this function with the path already pointing to
+	 *    the first item to check. But sometimes, we may enter it with
+	 *    slot == nritems.
+	 * 2. We are searching for normal backref but bytenr of this leaf
+	 *    matches shared data backref
+	 * 3. The leaf owner is not equal to the root we are searching
+	 *
+	 * For these cases, go to the next leaf before we continue.
 	 */
-	if (path->slots[0] >= btrfs_header_nritems(path->nodes[0])) {
-		if (time_seq == SEQ_LAST)
+	eb = path->nodes[0];
+	if (path->slots[0] >= btrfs_header_nritems(eb) ||
+	    is_shared_data_backref(preftrees, eb->start) ||
+	    ref->root_id != btrfs_header_owner(eb)) {
+		if (time_seq == BTRFS_SEQ_LAST)
 			ret = btrfs_next_leaf(root, path);
 		else
 			ret = btrfs_next_old_leaf(root, path, time_seq);
 	}
 
-	while (!ret && count < total_refs) {
+	while (!ret && count < ref->count) {
 		eb = path->nodes[0];
 		slot = path->slots[0];
 
@@ -455,13 +472,31 @@ static int add_all_parents(struct btrfs_root *root, struct btrfs_path *path,
 		    key.type != BTRFS_EXTENT_DATA_KEY)
 			break;
 
+		/*
+		 * We are searching for normal backref but bytenr of this leaf
+		 * matches shared data backref, OR
+		 * the leaf owner is not equal to the root we are searching for
+		 */
+		if (slot == 0 &&
+		    (is_shared_data_backref(preftrees, eb->start) ||
+		     ref->root_id != btrfs_header_owner(eb))) {
+			if (time_seq == BTRFS_SEQ_LAST)
+				ret = btrfs_next_leaf(root, path);
+			else
+				ret = btrfs_next_old_leaf(root, path, time_seq);
+			continue;
+		}
 		fi = btrfs_item_ptr(eb, slot, struct btrfs_file_extent_item);
 		disk_byte = btrfs_file_extent_disk_bytenr(eb, fi);
+		data_offset = btrfs_file_extent_offset(eb, fi);
 
 		if (disk_byte == wanted_disk_byte) {
 			eie = NULL;
 			old = NULL;
-			count++;
+			if (ref->key_for_search.offset == key.offset - data_offset)
+				count++;
+			else
+				goto next;
 			if (extent_item_pos) {
 				ret = check_extent_in_eb(&key, eb, fi,
 						*extent_item_pos,
@@ -483,7 +518,7 @@ static int add_all_parents(struct btrfs_root *root, struct btrfs_path *path,
 			eie = NULL;
 		}
 next:
-		if (time_seq == SEQ_LAST)
+		if (time_seq == BTRFS_SEQ_LAST)
 			ret = btrfs_next_item(root, path);
 		else
 			ret = btrfs_next_old_item(root, path, time_seq);
@@ -502,59 +537,82 @@ next:
  */
 static int resolve_indirect_ref(struct btrfs_fs_info *fs_info,
 				struct btrfs_path *path, u64 time_seq,
+				struct preftrees *preftrees,
 				struct prelim_ref *ref, struct ulist *parents,
-				const u64 *extent_item_pos, u64 total_refs,
-				bool ignore_offset)
+				const u64 *extent_item_pos, bool ignore_offset)
 {
 	struct btrfs_root *root;
-	struct btrfs_key root_key;
 	struct extent_buffer *eb;
 	int ret = 0;
 	int root_level;
 	int level = ref->level;
-	int index;
+	struct btrfs_key search_key = ref->key_for_search;
 
-	root_key.objectid = ref->root_id;
-	root_key.type = BTRFS_ROOT_ITEM_KEY;
-	root_key.offset = (u64)-1;
-
-	index = srcu_read_lock(&fs_info->subvol_srcu);
-
-	root = btrfs_get_fs_root(fs_info, &root_key, false);
+	/*
+	 * If we're search_commit_root we could possibly be holding locks on
+	 * other tree nodes.  This happens when qgroups does backref walks when
+	 * adding new delayed refs.  To deal with this we need to look in cache
+	 * for the root, and if we don't find it then we need to search the
+	 * tree_root's commit root, thus the btrfs_get_fs_root_commit_root usage
+	 * here.
+	 */
+	if (path->search_commit_root)
+		root = btrfs_get_fs_root_commit_root(fs_info, path, ref->root_id);
+	else
+		root = btrfs_get_fs_root(fs_info, ref->root_id, false);
 	if (IS_ERR(root)) {
-		srcu_read_unlock(&fs_info->subvol_srcu, index);
 		ret = PTR_ERR(root);
+		goto out_free;
+	}
+
+	if (!path->search_commit_root &&
+	    test_bit(BTRFS_ROOT_DELETING, &root->state)) {
+		ret = -ENOENT;
 		goto out;
 	}
 
 	if (btrfs_is_testing(fs_info)) {
-		srcu_read_unlock(&fs_info->subvol_srcu, index);
 		ret = -ENOENT;
 		goto out;
 	}
 
 	if (path->search_commit_root)
 		root_level = btrfs_header_level(root->commit_root);
-	else if (time_seq == SEQ_LAST)
+	else if (time_seq == BTRFS_SEQ_LAST)
 		root_level = btrfs_header_level(root->node);
 	else
 		root_level = btrfs_old_root_level(root, time_seq);
 
-	if (root_level + 1 == level) {
-		srcu_read_unlock(&fs_info->subvol_srcu, index);
+	if (root_level + 1 == level)
 		goto out;
-	}
 
+	/*
+	 * We can often find data backrefs with an offset that is too large
+	 * (>= LLONG_MAX, maximum allowed file offset) due to underflows when
+	 * subtracting a file's offset with the data offset of its
+	 * corresponding extent data item. This can happen for example in the
+	 * clone ioctl.
+	 *
+	 * So if we detect such case we set the search key's offset to zero to
+	 * make sure we will find the matching file extent item at
+	 * add_all_parents(), otherwise we will miss it because the offset
+	 * taken form the backref is much larger then the offset of the file
+	 * extent item. This can make us scan a very large number of file
+	 * extent items, but at least it will not make us miss any.
+	 *
+	 * This is an ugly workaround for a behaviour that should have never
+	 * existed, but it does and a fix for the clone ioctl would touch a lot
+	 * of places, cause backwards incompatibility and would not fix the
+	 * problem for extents cloned with older kernels.
+	 */
+	if (search_key.type == BTRFS_EXTENT_DATA_KEY &&
+	    search_key.offset >= LLONG_MAX)
+		search_key.offset = 0;
 	path->lowest_level = level;
-	if (time_seq == SEQ_LAST)
-		ret = btrfs_search_slot(NULL, root, &ref->key_for_search, path,
-					0, 0);
+	if (time_seq == BTRFS_SEQ_LAST)
+		ret = btrfs_search_slot(NULL, root, &search_key, path, 0, 0);
 	else
-		ret = btrfs_search_old_slot(root, &ref->key_for_search, path,
-					    time_seq);
-
-	/* root node has been locked, we can release @subvol_srcu safely here */
-	srcu_read_unlock(&fs_info->subvol_srcu, index);
+		ret = btrfs_search_old_slot(root, &search_key, path, time_seq);
 
 	btrfs_debug(fs_info,
 		"search slot in root %llu (level %d, ref count %d) returned %d for key (%llu %u %llu)",
@@ -574,9 +632,11 @@ static int resolve_indirect_ref(struct btrfs_fs_info *fs_info,
 		eb = path->nodes[level];
 	}
 
-	ret = add_all_parents(root, path, parents, ref, level, time_seq,
-			      extent_item_pos, total_refs, ignore_offset);
+	ret = add_all_parents(root, path, parents, preftrees, ref, level,
+			      time_seq, extent_item_pos, ignore_offset);
 out:
+	btrfs_put_root(root);
+out_free:
 	path->lowest_level = 0;
 	btrfs_release_path(path);
 	return ret;
@@ -590,6 +650,18 @@ unode_aux_to_inode_list(struct ulist_node *node)
 	return (struct extent_inode_elem *)(uintptr_t)node->aux;
 }
 
+static void free_leaf_list(struct ulist *ulist)
+{
+	struct ulist_node *node;
+	struct ulist_iterator uiter;
+
+	ULIST_ITER_INIT(&uiter);
+	while ((node = ulist_next(ulist, &uiter)))
+		free_inode_elem_list(unode_aux_to_inode_list(node));
+
+	ulist_free(ulist);
+}
+
 /*
  * We maintain three separate rbtrees: one for direct refs, one for
  * indirect refs which have a key, and one for indirect refs which do not
@@ -609,7 +681,7 @@ unode_aux_to_inode_list(struct ulist_node *node)
 static int resolve_indirect_refs(struct btrfs_fs_info *fs_info,
 				 struct btrfs_path *path, u64 time_seq,
 				 struct preftrees *preftrees,
-				 const u64 *extent_item_pos, u64 total_refs,
+				 const u64 *extent_item_pos,
 				 struct share_check *sc, bool ignore_offset)
 {
 	int err;
@@ -653,9 +725,9 @@ static int resolve_indirect_refs(struct btrfs_fs_info *fs_info,
 			ret = BACKREF_FOUND_SHARED;
 			goto out;
 		}
-		err = resolve_indirect_ref(fs_info, path, time_seq, ref,
-					   parents, extent_item_pos,
-					   total_refs, ignore_offset);
+		err = resolve_indirect_ref(fs_info, path, time_seq, preftrees,
+					   ref, parents, extent_item_pos,
+					   ignore_offset);
 		/*
 		 * we can only tolerate ENOENT,otherwise,we should catch error
 		 * and return directly.
@@ -704,7 +776,11 @@ static int resolve_indirect_refs(struct btrfs_fs_info *fs_info,
 		cond_resched();
 	}
 out:
-	ulist_free(parents);
+	/*
+	 * We may have inode lists attached to refs in the parents ulist, so we
+	 * must free them before freeing the ulist and its refs.
+	 */
+	free_leaf_list(parents);
 	return ret;
 }
 
@@ -727,16 +803,18 @@ static int add_missing_keys(struct btrfs_fs_info *fs_info,
 		BUG_ON(ref->key_for_search.type);
 		BUG_ON(!ref->wanted_disk_byte);
 
-		eb = read_tree_block(fs_info, ref->wanted_disk_byte, 0,
-				     ref->level - 1, NULL);
+		eb = read_tree_block(fs_info, ref->wanted_disk_byte,
+				     ref->root_id, 0, ref->level - 1, NULL);
 		if (IS_ERR(eb)) {
 			free_pref(ref);
 			return PTR_ERR(eb);
-		} else if (!extent_buffer_uptodate(eb)) {
+		}
+		if (!extent_buffer_uptodate(eb)) {
 			free_pref(ref);
 			free_extent_buffer(eb);
 			return -EIO;
 		}
+
 		if (lock)
 			btrfs_tree_read_lock(eb);
 		if (btrfs_header_level(eb) == 0)
@@ -758,20 +836,14 @@ static int add_missing_keys(struct btrfs_fs_info *fs_info,
  */
 static int add_delayed_refs(const struct btrfs_fs_info *fs_info,
 			    struct btrfs_delayed_ref_head *head, u64 seq,
-			    struct preftrees *preftrees, u64 *total_refs,
-			    struct share_check *sc)
+			    struct preftrees *preftrees, struct share_check *sc)
 {
 	struct btrfs_delayed_ref_node *node;
-	struct btrfs_delayed_extent_op *extent_op = head->extent_op;
 	struct btrfs_key key;
-	struct btrfs_key tmp_op_key;
 	struct rb_node *n;
 	int count;
 	int ret = 0;
 
-	if (extent_op && extent_op->update_key)
-		btrfs_disk_key_to_cpu(&tmp_op_key, &extent_op->key);
-
 	spin_lock(&head->lock);
 	for (n = rb_first_cached(&head->ref_tree); n; n = rb_next(n)) {
 		node = rb_entry(n, struct btrfs_delayed_ref_node,
@@ -793,15 +865,20 @@ static int add_delayed_refs(const struct btrfs_fs_info *fs_info,
 		default:
 			BUG();
 		}
-		*total_refs += count;
 		switch (node->type) {
 		case BTRFS_TREE_BLOCK_REF_KEY: {
 			/* NORMAL INDIRECT METADATA backref */
 			struct btrfs_delayed_tree_ref *ref;
+			struct btrfs_key *key_ptr = NULL;
+
+			if (head->extent_op && head->extent_op->update_key) {
+				btrfs_disk_key_to_cpu(&key, &head->extent_op->key);
+				key_ptr = &key;
+			}
 
 			ref = btrfs_delayed_node_to_tree_ref(node);
 			ret = add_indirect_ref(fs_info, preftrees, ref->root,
-					       &tmp_op_key, ref->level + 1,
+					       key_ptr, ref->level + 1,
 					       node->bytenr, count, sc,
 					       GFP_ATOMIC);
 			break;
@@ -827,13 +904,22 @@ static int add_delayed_refs(const struct btrfs_fs_info *fs_info,
 			key.offset = ref->offset;
 
 			/*
-			 * Found a inum that doesn't match our known inum, we
-			 * know it's shared.
+			 * If we have a share check context and a reference for
+			 * another inode, we can't exit immediately. This is
+			 * because even if this is a BTRFS_ADD_DELAYED_REF
+			 * reference we may find next a BTRFS_DROP_DELAYED_REF
+			 * which cancels out this ADD reference.
+			 *
+			 * If this is a DROP reference and there was no previous
+			 * ADD reference, then we need to signal that when we
+			 * process references from the extent tree (through
+			 * add_inline_refs() and add_keyed_refs()), we should
+			 * not exit early if we find a reference for another
+			 * inode, because one of the delayed DROP references
+			 * may cancel that reference in the extent tree.
 			 */
-			if (sc && sc->inum && ref->objectid != sc->inum) {
-				ret = BACKREF_FOUND_SHARED;
-				goto out;
-			}
+			if (sc && count < 0)
+				sc->have_delayed_delete_refs = true;
 
 			ret = add_indirect_ref(fs_info, preftrees, ref->root,
 					       &key, 0, node->bytenr, count, sc,
@@ -863,7 +949,7 @@ static int add_delayed_refs(const struct btrfs_fs_info *fs_info,
 	}
 	if (!ret)
 		ret = extent_is_shared(sc);
-out:
+
 	spin_unlock(&head->lock);
 	return ret;
 }
@@ -876,7 +962,7 @@ out:
 static int add_inline_refs(const struct btrfs_fs_info *fs_info,
 			   struct btrfs_path *path, u64 bytenr,
 			   int *info_level, struct preftrees *preftrees,
-			   u64 *total_refs, struct share_check *sc)
+			   struct share_check *sc)
 {
 	int ret = 0;
 	int slot;
@@ -895,12 +981,11 @@ static int add_inline_refs(const struct btrfs_fs_info *fs_info,
 	leaf = path->nodes[0];
 	slot = path->slots[0];
 
-	item_size = btrfs_item_size_nr(leaf, slot);
+	item_size = btrfs_item_size(leaf, slot);
 	BUG_ON(item_size < sizeof(*ei));
 
 	ei = btrfs_item_ptr(leaf, slot, struct btrfs_extent_item);
 	flags = btrfs_extent_flags(leaf, ei);
-	*total_refs += btrfs_extent_refs(leaf, ei);
 	btrfs_item_key_to_cpu(leaf, &found_key, slot);
 
 	ptr = (unsigned long)(ei + 1);
@@ -967,7 +1052,8 @@ static int add_inline_refs(const struct btrfs_fs_info *fs_info,
 			key.type = BTRFS_EXTENT_DATA_KEY;
 			key.offset = btrfs_extent_data_ref_offset(leaf, dref);
 
-			if (sc && sc->inum && key.objectid != sc->inum) {
+			if (sc && sc->inum && key.objectid != sc->inum &&
+			    !sc->have_delayed_delete_refs) {
 				ret = BACKREF_FOUND_SHARED;
 				break;
 			}
@@ -977,6 +1063,7 @@ static int add_inline_refs(const struct btrfs_fs_info *fs_info,
 			ret = add_indirect_ref(fs_info, preftrees, root,
 					       &key, 0, bytenr, count,
 					       sc, GFP_NOFS);
+
 			break;
 		}
 		default:
@@ -995,12 +1082,12 @@ static int add_inline_refs(const struct btrfs_fs_info *fs_info,
  *
  * Returns 0 on success, <0 on error, or BACKREF_FOUND_SHARED.
  */
-static int add_keyed_refs(struct btrfs_fs_info *fs_info,
+static int add_keyed_refs(struct btrfs_root *extent_root,
 			  struct btrfs_path *path, u64 bytenr,
 			  int info_level, struct preftrees *preftrees,
 			  struct share_check *sc)
 {
-	struct btrfs_root *extent_root = fs_info->extent_root;
+	struct btrfs_fs_info *fs_info = extent_root->fs_info;
 	int ret;
 	int slot;
 	struct extent_buffer *leaf;
@@ -1066,7 +1153,8 @@ static int add_keyed_refs(struct btrfs_fs_info *fs_info,
 			key.type = BTRFS_EXTENT_DATA_KEY;
 			key.offset = btrfs_extent_data_ref_offset(leaf, dref);
 
-			if (sc && sc->inum && key.objectid != sc->inum) {
+			if (sc && sc->inum && key.objectid != sc->inum &&
+			    !sc->have_delayed_delete_refs) {
 				ret = BACKREF_FOUND_SHARED;
 				break;
 			}
@@ -1094,8 +1182,8 @@ static int add_keyed_refs(struct btrfs_fs_info *fs_info,
  * indirect refs to their parent bytenr.
  * When roots are found, they're added to the roots list
  *
- * If time_seq is set to SEQ_LAST, it will not search delayed_refs, and behave
- * much like trans == NULL case, the difference only lies in it will not
+ * If time_seq is set to BTRFS_SEQ_LAST, it will not search delayed_refs, and
+ * behave much like trans == NULL case, the difference only lies in it will not
  * commit root.
  * The special case is for qgroup to search roots in commit_transaction().
  *
@@ -1116,6 +1204,7 @@ static int find_parent_nodes(struct btrfs_trans_handle *trans,
 			     struct ulist *roots, const u64 *extent_item_pos,
 			     struct share_check *sc, bool ignore_offset)
 {
+	struct btrfs_root *root = btrfs_extent_root(fs_info, bytenr);
 	struct btrfs_key key;
 	struct btrfs_path *path;
 	struct btrfs_delayed_ref_root *delayed_refs = NULL;
@@ -1125,8 +1214,6 @@ static int find_parent_nodes(struct btrfs_trans_handle *trans,
 	struct prelim_ref *ref;
 	struct rb_node *node;
 	struct extent_inode_elem *eie = NULL;
-	/* total of both direct AND indirect refs! */
-	u64 total_refs = 0;
 	struct preftrees preftrees = {
 		.direct = PREFTREE_INIT,
 		.indirect = PREFTREE_INIT,
@@ -1148,31 +1235,29 @@ static int find_parent_nodes(struct btrfs_trans_handle *trans,
 		path->skip_locking = 1;
 	}
 
-	if (time_seq == SEQ_LAST)
+	if (time_seq == BTRFS_SEQ_LAST)
 		path->skip_locking = 1;
 
-	/*
-	 * grab both a lock on the path and a lock on the delayed ref head.
-	 * We need both to get a consistent picture of how the refs look
-	 * at a specified point in time
-	 */
 again:
 	head = NULL;
 
-	ret = btrfs_search_slot(trans, fs_info->extent_root, &key, path, 0, 0);
+	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
 	if (ret < 0)
 		goto out;
-	BUG_ON(ret == 0);
+	if (ret == 0) {
+		/* This shouldn't happen, indicates a bug or fs corruption. */
+		ASSERT(ret != 0);
+		ret = -EUCLEAN;
+		goto out;
+	}
 
-#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
 	if (trans && likely(trans->type != __TRANS_DUMMY) &&
-	    time_seq != SEQ_LAST) {
-#else
-	if (trans && time_seq != SEQ_LAST) {
-#endif
+	    time_seq != BTRFS_SEQ_LAST) {
 		/*
-		 * look if there are updates for this ref queued and lock the
-		 * head
+		 * We have a specific time_seq we care about and trans which
+		 * means we have the path lock, we need to grab the ref head and
+		 * lock it so we have a consistent view of the refs at the given
+		 * time.
 		 */
 		delayed_refs = &trans->transaction->delayed_refs;
 		spin_lock(&delayed_refs->lock);
@@ -1195,7 +1280,7 @@ again:
 			}
 			spin_unlock(&delayed_refs->lock);
 			ret = add_delayed_refs(fs_info, head, time_seq,
-					       &preftrees, &total_refs, sc);
+					       &preftrees, sc);
 			mutex_unlock(&head->mutex);
 			if (ret)
 				goto out;
@@ -1216,11 +1301,10 @@ again:
 		    (key.type == BTRFS_EXTENT_ITEM_KEY ||
 		     key.type == BTRFS_METADATA_ITEM_KEY)) {
 			ret = add_inline_refs(fs_info, path, bytenr,
-					      &info_level, &preftrees,
-					      &total_refs, sc);
+					      &info_level, &preftrees, sc);
 			if (ret)
 				goto out;
-			ret = add_keyed_refs(fs_info, path, bytenr, info_level,
+			ret = add_keyed_refs(root, path, bytenr, info_level,
 					     &preftrees, sc);
 			if (ret)
 				goto out;
@@ -1236,7 +1320,7 @@ again:
 	WARN_ON(!RB_EMPTY_ROOT(&preftrees.indirect_missing_keys.root.rb_root));
 
 	ret = resolve_indirect_refs(fs_info, path, time_seq, &preftrees,
-				    extent_item_pos, total_refs, sc, ignore_offset);
+				    extent_item_pos, sc, ignore_offset);
 	if (ret)
 		goto out;
 
@@ -1281,28 +1365,33 @@ again:
 				struct extent_buffer *eb;
 
 				eb = read_tree_block(fs_info, ref->parent, 0,
-						     ref->level, NULL);
+						     0, ref->level, NULL);
 				if (IS_ERR(eb)) {
 					ret = PTR_ERR(eb);
 					goto out;
-				} else if (!extent_buffer_uptodate(eb)) {
+				}
+				if (!extent_buffer_uptodate(eb)) {
 					free_extent_buffer(eb);
 					ret = -EIO;
 					goto out;
 				}
 
-				if (!path->skip_locking) {
+				if (!path->skip_locking)
 					btrfs_tree_read_lock(eb);
-					btrfs_set_lock_blocking_read(eb);
-				}
 				ret = find_extent_in_eb(eb, bytenr,
 							*extent_item_pos, &eie, ignore_offset);
 				if (!path->skip_locking)
-					btrfs_tree_read_unlock_blocking(eb);
+					btrfs_tree_read_unlock(eb);
 				free_extent_buffer(eb);
 				if (ret < 0)
 					goto out;
 				ref->inode_list = eie;
+				/*
+				 * We transferred the list ownership to the ref,
+				 * so set to NULL to avoid a double free in case
+				 * an error happens after this.
+				 */
+				eie = NULL;
 			}
 			ret = ulist_add_merge_ptr(refs, ref->parent,
 						  ref->inode_list,
@@ -1311,15 +1400,31 @@ again:
 				goto out;
 			if (!ret && extent_item_pos) {
 				/*
-				 * we've recorded that parent, so we must extend
-				 * its inode list here
+				 * We've recorded that parent, so we must extend
+				 * its inode list here.
+				 *
+				 * However if there was corruption we may not
+				 * have found an eie, return an error in this
+				 * case.
 				 */
-				BUG_ON(!eie);
+				ASSERT(eie);
+				if (!eie) {
+					ret = -EUCLEAN;
+					goto out;
+				}
 				while (eie->next)
 					eie = eie->next;
 				eie->next = ref->inode_list;
 			}
 			eie = NULL;
+			/*
+			 * We have transferred the inode list ownership from
+			 * this ref to the ref we added to the 'refs' ulist.
+			 * So set this ref's inode list to NULL to avoid
+			 * use-after-free when our caller uses it or double
+			 * frees in case an error happens before we return.
+			 */
+			ref->inode_list = NULL;
 		}
 		cond_resched();
 	}
@@ -1336,24 +1441,6 @@ out:
 	return ret;
 }
 
-static void free_leaf_list(struct ulist *blocks)
-{
-	struct ulist_node *node = NULL;
-	struct extent_inode_elem *eie;
-	struct ulist_iterator uiter;
-
-	ULIST_ITER_INIT(&uiter);
-	while ((node = ulist_next(blocks, &uiter))) {
-		if (!node->aux)
-			continue;
-		eie = unode_aux_to_inode_list(node);
-		free_inode_elem_list(eie);
-		node->aux = 0;
-	}
-
-	ulist_free(blocks);
-}
-
 /*
  * Finds all leafs with a reference to the specified combination of bytenr and
  * offset. key_list_head will point to a list of corresponding keys (caller must
@@ -1362,10 +1449,10 @@ static void free_leaf_list(struct ulist *blocks)
  *
  * returns 0 on success, <0 on error
  */
-static int btrfs_find_all_leafs(struct btrfs_trans_handle *trans,
-				struct btrfs_fs_info *fs_info, u64 bytenr,
-				u64 time_seq, struct ulist **leafs,
-				const u64 *extent_item_pos, bool ignore_offset)
+int btrfs_find_all_leafs(struct btrfs_trans_handle *trans,
+			 struct btrfs_fs_info *fs_info, u64 bytenr,
+			 u64 time_seq, struct ulist **leafs,
+			 const u64 *extent_item_pos, bool ignore_offset)
 {
 	int ret;
 
@@ -1422,6 +1509,7 @@ static int btrfs_find_all_roots_safe(struct btrfs_trans_handle *trans,
 		if (ret < 0 && ret != -ENOENT) {
 			ulist_free(tmp);
 			ulist_free(*roots);
+			*roots = NULL;
 			return ret;
 		}
 		node = ulist_next(tmp, &uiter);
@@ -1438,23 +1526,150 @@ static int btrfs_find_all_roots_safe(struct btrfs_trans_handle *trans,
 int btrfs_find_all_roots(struct btrfs_trans_handle *trans,
 			 struct btrfs_fs_info *fs_info, u64 bytenr,
 			 u64 time_seq, struct ulist **roots,
-			 bool ignore_offset)
+			 bool skip_commit_root_sem)
 {
 	int ret;
 
-	if (!trans)
+	if (!trans && !skip_commit_root_sem)
 		down_read(&fs_info->commit_root_sem);
 	ret = btrfs_find_all_roots_safe(trans, fs_info, bytenr,
-					time_seq, roots, ignore_offset);
-	if (!trans)
+					time_seq, roots, false);
+	if (!trans && !skip_commit_root_sem)
 		up_read(&fs_info->commit_root_sem);
 	return ret;
 }
 
-/**
- * btrfs_check_shared - tell us whether an extent is shared
+/*
+ * The caller has joined a transaction or is holding a read lock on the
+ * fs_info->commit_root_sem semaphore, so no need to worry about the root's last
+ * snapshot field changing while updating or checking the cache.
+ */
+static bool lookup_backref_shared_cache(struct btrfs_backref_shared_cache *cache,
+					struct btrfs_root *root,
+					u64 bytenr, int level, bool *is_shared)
+{
+	struct btrfs_backref_shared_cache_entry *entry;
+
+	if (!cache->use_cache)
+		return false;
+
+	if (WARN_ON_ONCE(level >= BTRFS_MAX_LEVEL))
+		return false;
+
+	/*
+	 * Level -1 is used for the data extent, which is not reliable to cache
+	 * because its reference count can increase or decrease without us
+	 * realizing. We cache results only for extent buffers that lead from
+	 * the root node down to the leaf with the file extent item.
+	 */
+	ASSERT(level >= 0);
+
+	entry = &cache->entries[level];
+
+	/* Unused cache entry or being used for some other extent buffer. */
+	if (entry->bytenr != bytenr)
+		return false;
+
+	/*
+	 * We cached a false result, but the last snapshot generation of the
+	 * root changed, so we now have a snapshot. Don't trust the result.
+	 */
+	if (!entry->is_shared &&
+	    entry->gen != btrfs_root_last_snapshot(&root->root_item))
+		return false;
+
+	/*
+	 * If we cached a true result and the last generation used for dropping
+	 * a root changed, we can not trust the result, because the dropped root
+	 * could be a snapshot sharing this extent buffer.
+	 */
+	if (entry->is_shared &&
+	    entry->gen != btrfs_get_last_root_drop_gen(root->fs_info))
+		return false;
+
+	*is_shared = entry->is_shared;
+	/*
+	 * If the node at this level is shared, than all nodes below are also
+	 * shared. Currently some of the nodes below may be marked as not shared
+	 * because we have just switched from one leaf to another, and switched
+	 * also other nodes above the leaf and below the current level, so mark
+	 * them as shared.
+	 */
+	if (*is_shared) {
+		for (int i = 0; i < level; i++) {
+			cache->entries[i].is_shared = true;
+			cache->entries[i].gen = entry->gen;
+		}
+	}
+
+	return true;
+}
+
+/*
+ * The caller has joined a transaction or is holding a read lock on the
+ * fs_info->commit_root_sem semaphore, so no need to worry about the root's last
+ * snapshot field changing while updating or checking the cache.
+ */
+static void store_backref_shared_cache(struct btrfs_backref_shared_cache *cache,
+				       struct btrfs_root *root,
+				       u64 bytenr, int level, bool is_shared)
+{
+	struct btrfs_backref_shared_cache_entry *entry;
+	u64 gen;
+
+	if (!cache->use_cache)
+		return;
+
+	if (WARN_ON_ONCE(level >= BTRFS_MAX_LEVEL))
+		return;
+
+	/*
+	 * Level -1 is used for the data extent, which is not reliable to cache
+	 * because its reference count can increase or decrease without us
+	 * realizing. We cache results only for extent buffers that lead from
+	 * the root node down to the leaf with the file extent item.
+	 */
+	ASSERT(level >= 0);
+
+	if (is_shared)
+		gen = btrfs_get_last_root_drop_gen(root->fs_info);
+	else
+		gen = btrfs_root_last_snapshot(&root->root_item);
+
+	entry = &cache->entries[level];
+	entry->bytenr = bytenr;
+	entry->is_shared = is_shared;
+	entry->gen = gen;
+
+	/*
+	 * If we found an extent buffer is shared, set the cache result for all
+	 * extent buffers below it to true. As nodes in the path are COWed,
+	 * their sharedness is moved to their children, and if a leaf is COWed,
+	 * then the sharedness of a data extent becomes direct, the refcount of
+	 * data extent is increased in the extent item at the extent tree.
+	 */
+	if (is_shared) {
+		for (int i = 0; i < level; i++) {
+			entry = &cache->entries[i];
+			entry->is_shared = is_shared;
+			entry->gen = gen;
+		}
+	}
+}
+
+/*
+ * Check if a data extent is shared or not.
  *
- * btrfs_check_shared uses the backref walking code but will short
+ * @root:        The root the inode belongs to.
+ * @inum:        Number of the inode whose extent we are checking.
+ * @bytenr:      Logical bytenr of the extent we are checking.
+ * @extent_gen:  Generation of the extent (file extent item) or 0 if it is
+ *               not known.
+ * @roots:       List of roots this extent is shared among.
+ * @tmp:         Temporary list used for iteration.
+ * @cache:       A backref lookup result cache.
+ *
+ * btrfs_is_data_extent_shared uses the backref walking code but will short
  * circuit as soon as it finds a root or inode that doesn't match the
  * one passed in. This provides a significant performance benefit for
  * callers (such as fiemap) which want to know whether the extent is
@@ -1465,20 +1680,24 @@ int btrfs_find_all_roots(struct btrfs_trans_handle *trans,
  *
  * Return: 0 if extent is not shared, 1 if it is shared, < 0 on error.
  */
-int btrfs_check_shared(struct btrfs_root *root, u64 inum, u64 bytenr,
-		struct ulist *roots, struct ulist *tmp)
+int btrfs_is_data_extent_shared(struct btrfs_root *root, u64 inum, u64 bytenr,
+				u64 extent_gen,
+				struct ulist *roots, struct ulist *tmp,
+				struct btrfs_backref_shared_cache *cache)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct btrfs_trans_handle *trans;
 	struct ulist_iterator uiter;
 	struct ulist_node *node;
-	struct seq_list elem = SEQ_LIST_INIT(elem);
+	struct btrfs_seq_list elem = BTRFS_SEQ_LIST_INIT(elem);
 	int ret = 0;
 	struct share_check shared = {
 		.root_objectid = root->root_key.objectid,
 		.inum = inum,
 		.share_count = 0,
+		.have_delayed_delete_refs = false,
 	};
+	int level;
 
 	ulist_init(roots);
 	ulist_init(tmp);
@@ -1495,23 +1714,73 @@ int btrfs_check_shared(struct btrfs_root *root, u64 inum, u64 bytenr,
 		btrfs_get_tree_mod_seq(fs_info, &elem);
 	}
 
+	/* -1 means we are in the bytenr of the data extent. */
+	level = -1;
 	ULIST_ITER_INIT(&uiter);
+	cache->use_cache = true;
 	while (1) {
+		bool is_shared;
+		bool cached;
+
 		ret = find_parent_nodes(trans, fs_info, bytenr, elem.seq, tmp,
 					roots, NULL, &shared, false);
 		if (ret == BACKREF_FOUND_SHARED) {
 			/* this is the only condition under which we return 1 */
 			ret = 1;
+			if (level >= 0)
+				store_backref_shared_cache(cache, root, bytenr,
+							   level, true);
 			break;
 		}
 		if (ret < 0 && ret != -ENOENT)
 			break;
 		ret = 0;
+		/*
+		 * If our data extent is not shared through reflinks and it was
+		 * created in a generation after the last one used to create a
+		 * snapshot of the inode's root, then it can not be shared
+		 * indirectly through subtrees, as that can only happen with
+		 * snapshots. In this case bail out, no need to check for the
+		 * sharedness of extent buffers.
+		 */
+		if (level == -1 &&
+		    extent_gen > btrfs_root_last_snapshot(&root->root_item))
+			break;
+
+		/*
+		 * If our data extent was not directly shared (without multiple
+		 * reference items), than it might have a single reference item
+		 * with a count > 1 for the same offset, which means there are 2
+		 * (or more) file extent items that point to the data extent -
+		 * this happens when a file extent item needs to be split and
+		 * then one item gets moved to another leaf due to a b+tree leaf
+		 * split when inserting some item. In this case the file extent
+		 * items may be located in different leaves and therefore some
+		 * of the leaves may be referenced through shared subtrees while
+		 * others are not. Since our extent buffer cache only works for
+		 * a single path (by far the most common case and simpler to
+		 * deal with), we can not use it if we have multiple leaves
+		 * (which implies multiple paths).
+		 */
+		if (level == -1 && tmp->nnodes > 1)
+			cache->use_cache = false;
+
+		if (level >= 0)
+			store_backref_shared_cache(cache, root, bytenr,
+						   level, false);
 		node = ulist_next(tmp, &uiter);
 		if (!node)
 			break;
 		bytenr = node->val;
+		level++;
+		cached = lookup_backref_shared_cache(cache, root, bytenr, level,
+						     &is_shared);
+		if (cached) {
+			ret = (is_shared ? 1 : 0);
+			break;
+		}
 		shared.share_count = 0;
+		shared.have_delayed_delete_refs = false;
 		cond_resched();
 	}
 
@@ -1620,13 +1889,11 @@ char *btrfs_ref_to_path(struct btrfs_root *fs_root, struct btrfs_path *path,
 	s64 bytes_left = ((s64)size) - 1;
 	struct extent_buffer *eb = eb_in;
 	struct btrfs_key found_key;
-	int leave_spinning = path->leave_spinning;
 	struct btrfs_inode_ref *iref;
 
 	if (bytes_left >= 0)
 		dest[bytes_left] = '\0';
 
-	path->leave_spinning = 1;
 	while (1) {
 		bytes_left -= name_len;
 		if (bytes_left >= 0)
@@ -1634,7 +1901,7 @@ char *btrfs_ref_to_path(struct btrfs_root *fs_root, struct btrfs_path *path,
 					   name_off, name_len);
 		if (eb != eb_in) {
 			if (!path->skip_locking)
-				btrfs_tree_read_unlock_blocking(eb);
+				btrfs_tree_read_unlock(eb);
 			free_extent_buffer(eb);
 		}
 		ret = btrfs_find_item(fs_root, path, parent, 0,
@@ -1654,8 +1921,6 @@ char *btrfs_ref_to_path(struct btrfs_root *fs_root, struct btrfs_path *path,
 		eb = path->nodes[0];
 		/* make sure we can use eb after releasing the path */
 		if (eb != eb_in) {
-			if (!path->skip_locking)
-				btrfs_set_lock_blocking_read(eb);
 			path->nodes[0] = NULL;
 			path->locks[0] = 0;
 		}
@@ -1672,7 +1937,6 @@ char *btrfs_ref_to_path(struct btrfs_root *fs_root, struct btrfs_path *path,
 	}
 
 	btrfs_release_path(path);
-	path->leave_spinning = leave_spinning;
 
 	if (ret)
 		return ERR_PTR(ret);
@@ -1689,6 +1953,7 @@ int extent_from_logical(struct btrfs_fs_info *fs_info, u64 logical,
 			struct btrfs_path *path, struct btrfs_key *found_key,
 			u64 *flags_ret)
 {
+	struct btrfs_root *extent_root = btrfs_extent_root(fs_info, logical);
 	int ret;
 	u64 flags;
 	u64 size = 0;
@@ -1704,11 +1969,11 @@ int extent_from_logical(struct btrfs_fs_info *fs_info, u64 logical,
 	key.objectid = logical;
 	key.offset = (u64)-1;
 
-	ret = btrfs_search_slot(NULL, fs_info->extent_root, &key, path, 0, 0);
+	ret = btrfs_search_slot(NULL, extent_root, &key, path, 0, 0);
 	if (ret < 0)
 		return ret;
 
-	ret = btrfs_previous_extent_item(fs_info->extent_root, path, 0);
+	ret = btrfs_previous_extent_item(extent_root, path, 0);
 	if (ret) {
 		if (ret > 0)
 			ret = -ENOENT;
@@ -1728,7 +1993,7 @@ int extent_from_logical(struct btrfs_fs_info *fs_info, u64 logical,
 	}
 
 	eb = path->nodes[0];
-	item_size = btrfs_item_size_nr(eb, path->slots[0]);
+	item_size = btrfs_item_size(eb, path->slots[0]);
 	BUG_ON(item_size < sizeof(*ei));
 
 	ei = btrfs_item_ptr(eb, path->slots[0], struct btrfs_extent_item);
@@ -1903,7 +2168,7 @@ int iterate_extent_inodes(struct btrfs_fs_info *fs_info,
 	struct ulist *roots = NULL;
 	struct ulist_node *ref_node = NULL;
 	struct ulist_node *root_node = NULL;
-	struct seq_list tree_mod_seq_elem = SEQ_LIST_INIT(tree_mod_seq_elem);
+	struct btrfs_seq_list seq_elem = BTRFS_SEQ_LIST_INIT(seq_elem);
 	struct ulist_iterator ref_uiter;
 	struct ulist_iterator root_uiter;
 
@@ -1911,7 +2176,7 @@ int iterate_extent_inodes(struct btrfs_fs_info *fs_info,
 			extent_item_objectid);
 
 	if (!search_commit_root) {
-		trans = btrfs_attach_transaction(fs_info->extent_root);
+		trans = btrfs_attach_transaction(fs_info->tree_root);
 		if (IS_ERR(trans)) {
 			if (PTR_ERR(trans) != -ENOENT &&
 			    PTR_ERR(trans) != -EROFS)
@@ -1921,12 +2186,12 @@ int iterate_extent_inodes(struct btrfs_fs_info *fs_info,
 	}
 
 	if (trans)
-		btrfs_get_tree_mod_seq(fs_info, &tree_mod_seq_elem);
+		btrfs_get_tree_mod_seq(fs_info, &seq_elem);
 	else
 		down_read(&fs_info->commit_root_sem);
 
 	ret = btrfs_find_all_leafs(trans, fs_info, extent_item_objectid,
-				   tree_mod_seq_elem.seq, &refs,
+				   seq_elem.seq, &refs,
 				   &extent_item_pos, ignore_offset);
 	if (ret)
 		goto out;
@@ -1934,7 +2199,7 @@ int iterate_extent_inodes(struct btrfs_fs_info *fs_info,
 	ULIST_ITER_INIT(&ref_uiter);
 	while (!ret && (ref_node = ulist_next(refs, &ref_uiter))) {
 		ret = btrfs_find_all_roots_safe(trans, fs_info, ref_node->val,
-						tree_mod_seq_elem.seq, &roots,
+						seq_elem.seq, &roots,
 						ignore_offset);
 		if (ret)
 			break;
@@ -1957,7 +2222,7 @@ int iterate_extent_inodes(struct btrfs_fs_info *fs_info,
 	free_leaf_list(refs);
 out:
 	if (trans) {
-		btrfs_put_tree_mod_seq(fs_info, &tree_mod_seq_elem);
+		btrfs_put_tree_mod_seq(fs_info, &seq_elem);
 		btrfs_end_transaction(trans);
 	} else {
 		up_read(&fs_info->commit_root_sem);
@@ -1966,10 +2231,29 @@ out:
 	return ret;
 }
 
+static int build_ino_list(u64 inum, u64 offset, u64 root, void *ctx)
+{
+	struct btrfs_data_container *inodes = ctx;
+	const size_t c = 3 * sizeof(u64);
+
+	if (inodes->bytes_left >= c) {
+		inodes->bytes_left -= c;
+		inodes->val[inodes->elem_cnt] = inum;
+		inodes->val[inodes->elem_cnt + 1] = offset;
+		inodes->val[inodes->elem_cnt + 2] = root;
+		inodes->elem_cnt += 3;
+	} else {
+		inodes->bytes_missing += c - inodes->bytes_left;
+		inodes->bytes_left = 0;
+		inodes->elem_missed += 3;
+	}
+
+	return 0;
+}
+
 int iterate_inodes_from_logical(u64 logical, struct btrfs_fs_info *fs_info,
 				struct btrfs_path *path,
-				iterate_extent_inodes_t *iterate, void *ctx,
-				bool ignore_offset)
+				void *ctx, bool ignore_offset)
 {
 	int ret;
 	u64 extent_item_pos;
@@ -1987,17 +2271,15 @@ int iterate_inodes_from_logical(u64 logical, struct btrfs_fs_info *fs_info,
 	extent_item_pos = logical - found_key.objectid;
 	ret = iterate_extent_inodes(fs_info, found_key.objectid,
 					extent_item_pos, search_commit_root,
-					iterate, ctx, ignore_offset);
+					build_ino_list, ctx, ignore_offset);
 
 	return ret;
 }
 
-typedef int (iterate_irefs_t)(u64 parent, u32 name_len, unsigned long name_off,
-			      struct extent_buffer *eb, void *ctx);
+static int inode_to_path(u64 inum, u32 name_len, unsigned long name_off,
+			 struct extent_buffer *eb, struct inode_fs_paths *ipath);
 
-static int iterate_inode_refs(u64 inum, struct btrfs_root *fs_root,
-			      struct btrfs_path *path,
-			      iterate_irefs_t *iterate, void *ctx)
+static int iterate_inode_refs(u64 inum, struct inode_fs_paths *ipath)
 {
 	int ret = 0;
 	int slot;
@@ -2006,8 +2288,9 @@ static int iterate_inode_refs(u64 inum, struct btrfs_root *fs_root,
 	u32 name_len;
 	u64 parent = 0;
 	int found = 0;
+	struct btrfs_root *fs_root = ipath->fs_root;
+	struct btrfs_path *path = ipath->btrfs_path;
 	struct extent_buffer *eb;
-	struct btrfs_item *item;
 	struct btrfs_inode_ref *iref;
 	struct btrfs_key found_key;
 
@@ -2033,18 +2316,17 @@ static int iterate_inode_refs(u64 inum, struct btrfs_root *fs_root,
 		}
 		btrfs_release_path(path);
 
-		item = btrfs_item_nr(slot);
 		iref = btrfs_item_ptr(eb, slot, struct btrfs_inode_ref);
 
-		for (cur = 0; cur < btrfs_item_size(eb, item); cur += len) {
+		for (cur = 0; cur < btrfs_item_size(eb, slot); cur += len) {
 			name_len = btrfs_inode_ref_name_len(eb, iref);
 			/* path must be released before calling iterate()! */
 			btrfs_debug(fs_root->fs_info,
 				"following ref at offset %u for inode %llu in tree %llu",
 				cur, found_key.objectid,
 				fs_root->root_key.objectid);
-			ret = iterate(parent, name_len,
-				      (unsigned long)(iref + 1), eb, ctx);
+			ret = inode_to_path(parent, name_len,
+				      (unsigned long)(iref + 1), eb, ipath);
 			if (ret)
 				break;
 			len = sizeof(*iref) + name_len;
@@ -2058,15 +2340,15 @@ static int iterate_inode_refs(u64 inum, struct btrfs_root *fs_root,
 	return ret;
 }
 
-static int iterate_inode_extrefs(u64 inum, struct btrfs_root *fs_root,
-				 struct btrfs_path *path,
-				 iterate_irefs_t *iterate, void *ctx)
+static int iterate_inode_extrefs(u64 inum, struct inode_fs_paths *ipath)
 {
 	int ret;
 	int slot;
 	u64 offset = 0;
 	u64 parent;
 	int found = 0;
+	struct btrfs_root *fs_root = ipath->fs_root;
+	struct btrfs_path *path = ipath->btrfs_path;
 	struct extent_buffer *eb;
 	struct btrfs_inode_extref *extref;
 	u32 item_size;
@@ -2092,7 +2374,7 @@ static int iterate_inode_extrefs(u64 inum, struct btrfs_root *fs_root,
 		}
 		btrfs_release_path(path);
 
-		item_size = btrfs_item_size_nr(eb, slot);
+		item_size = btrfs_item_size(eb, slot);
 		ptr = btrfs_item_ptr_offset(eb, slot);
 		cur_offset = 0;
 
@@ -2102,8 +2384,8 @@ static int iterate_inode_extrefs(u64 inum, struct btrfs_root *fs_root,
 			extref = (struct btrfs_inode_extref *)(ptr + cur_offset);
 			parent = btrfs_inode_extref_parent(eb, extref);
 			name_len = btrfs_inode_extref_name_len(eb, extref);
-			ret = iterate(parent, name_len,
-				      (unsigned long)&extref->name, eb, ctx);
+			ret = inode_to_path(parent, name_len,
+				      (unsigned long)&extref->name, eb, ipath);
 			if (ret)
 				break;
 
@@ -2120,34 +2402,13 @@ static int iterate_inode_extrefs(u64 inum, struct btrfs_root *fs_root,
 	return ret;
 }
 
-static int iterate_irefs(u64 inum, struct btrfs_root *fs_root,
-			 struct btrfs_path *path, iterate_irefs_t *iterate,
-			 void *ctx)
-{
-	int ret;
-	int found_refs = 0;
-
-	ret = iterate_inode_refs(inum, fs_root, path, iterate, ctx);
-	if (!ret)
-		++found_refs;
-	else if (ret != -ENOENT)
-		return ret;
-
-	ret = iterate_inode_extrefs(inum, fs_root, path, iterate, ctx);
-	if (ret == -ENOENT && found_refs)
-		return 0;
-
-	return ret;
-}
-
 /*
  * returns 0 if the path could be dumped (probably truncated)
  * returns <0 in case of an error
  */
 static int inode_to_path(u64 inum, u32 name_len, unsigned long name_off,
-			 struct extent_buffer *eb, void *ctx)
+			 struct extent_buffer *eb, struct inode_fs_paths *ipath)
 {
-	struct inode_fs_paths *ipath = ctx;
 	char *fspath;
 	char *fspath_min;
 	int i = ipath->fspath->elem_cnt;
@@ -2188,8 +2449,20 @@ static int inode_to_path(u64 inum, u32 name_len, unsigned long name_off,
  */
 int paths_from_inode(u64 inum, struct inode_fs_paths *ipath)
 {
-	return iterate_irefs(inum, ipath->fs_root, ipath->btrfs_path,
-			     inode_to_path, ipath);
+	int ret;
+	int found_refs = 0;
+
+	ret = iterate_inode_refs(inum, ipath);
+	if (!ret)
+		++found_refs;
+	else if (ret != -ENOENT)
+		return ret;
+
+	ret = iterate_inode_extrefs(inum, ipath);
+	if (ret == -ENOENT && found_refs)
+		return 0;
+
+	return ret;
 }
 
 struct btrfs_data_container *init_data_container(u32 total_bytes)
@@ -2252,3 +2525,827 @@ void free_ipath(struct inode_fs_paths *ipath)
 	kvfree(ipath->fspath);
 	kfree(ipath);
 }
+
+struct btrfs_backref_iter *btrfs_backref_iter_alloc(
+		struct btrfs_fs_info *fs_info, gfp_t gfp_flag)
+{
+	struct btrfs_backref_iter *ret;
+
+	ret = kzalloc(sizeof(*ret), gfp_flag);
+	if (!ret)
+		return NULL;
+
+	ret->path = btrfs_alloc_path();
+	if (!ret->path) {
+		kfree(ret);
+		return NULL;
+	}
+
+	/* Current backref iterator only supports iteration in commit root */
+	ret->path->search_commit_root = 1;
+	ret->path->skip_locking = 1;
+	ret->fs_info = fs_info;
+
+	return ret;
+}
+
+int btrfs_backref_iter_start(struct btrfs_backref_iter *iter, u64 bytenr)
+{
+	struct btrfs_fs_info *fs_info = iter->fs_info;
+	struct btrfs_root *extent_root = btrfs_extent_root(fs_info, bytenr);
+	struct btrfs_path *path = iter->path;
+	struct btrfs_extent_item *ei;
+	struct btrfs_key key;
+	int ret;
+
+	key.objectid = bytenr;
+	key.type = BTRFS_METADATA_ITEM_KEY;
+	key.offset = (u64)-1;
+	iter->bytenr = bytenr;
+
+	ret = btrfs_search_slot(NULL, extent_root, &key, path, 0, 0);
+	if (ret < 0)
+		return ret;
+	if (ret == 0) {
+		ret = -EUCLEAN;
+		goto release;
+	}
+	if (path->slots[0] == 0) {
+		WARN_ON(IS_ENABLED(CONFIG_BTRFS_DEBUG));
+		ret = -EUCLEAN;
+		goto release;
+	}
+	path->slots[0]--;
+
+	btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
+	if ((key.type != BTRFS_EXTENT_ITEM_KEY &&
+	     key.type != BTRFS_METADATA_ITEM_KEY) || key.objectid != bytenr) {
+		ret = -ENOENT;
+		goto release;
+	}
+	memcpy(&iter->cur_key, &key, sizeof(key));
+	iter->item_ptr = (u32)btrfs_item_ptr_offset(path->nodes[0],
+						    path->slots[0]);
+	iter->end_ptr = (u32)(iter->item_ptr +
+			btrfs_item_size(path->nodes[0], path->slots[0]));
+	ei = btrfs_item_ptr(path->nodes[0], path->slots[0],
+			    struct btrfs_extent_item);
+
+	/*
+	 * Only support iteration on tree backref yet.
+	 *
+	 * This is an extra precaution for non skinny-metadata, where
+	 * EXTENT_ITEM is also used for tree blocks, that we can only use
+	 * extent flags to determine if it's a tree block.
+	 */
+	if (btrfs_extent_flags(path->nodes[0], ei) & BTRFS_EXTENT_FLAG_DATA) {
+		ret = -ENOTSUPP;
+		goto release;
+	}
+	iter->cur_ptr = (u32)(iter->item_ptr + sizeof(*ei));
+
+	/* If there is no inline backref, go search for keyed backref */
+	if (iter->cur_ptr >= iter->end_ptr) {
+		ret = btrfs_next_item(extent_root, path);
+
+		/* No inline nor keyed ref */
+		if (ret > 0) {
+			ret = -ENOENT;
+			goto release;
+		}
+		if (ret < 0)
+			goto release;
+
+		btrfs_item_key_to_cpu(path->nodes[0], &iter->cur_key,
+				path->slots[0]);
+		if (iter->cur_key.objectid != bytenr ||
+		    (iter->cur_key.type != BTRFS_SHARED_BLOCK_REF_KEY &&
+		     iter->cur_key.type != BTRFS_TREE_BLOCK_REF_KEY)) {
+			ret = -ENOENT;
+			goto release;
+		}
+		iter->cur_ptr = (u32)btrfs_item_ptr_offset(path->nodes[0],
+							   path->slots[0]);
+		iter->item_ptr = iter->cur_ptr;
+		iter->end_ptr = (u32)(iter->item_ptr + btrfs_item_size(
+				      path->nodes[0], path->slots[0]));
+	}
+
+	return 0;
+release:
+	btrfs_backref_iter_release(iter);
+	return ret;
+}
+
+/*
+ * Go to the next backref item of current bytenr, can be either inlined or
+ * keyed.
+ *
+ * Caller needs to check whether it's inline ref or not by iter->cur_key.
+ *
+ * Return 0 if we get next backref without problem.
+ * Return >0 if there is no extra backref for this bytenr.
+ * Return <0 if there is something wrong happened.
+ */
+int btrfs_backref_iter_next(struct btrfs_backref_iter *iter)
+{
+	struct extent_buffer *eb = btrfs_backref_get_eb(iter);
+	struct btrfs_root *extent_root;
+	struct btrfs_path *path = iter->path;
+	struct btrfs_extent_inline_ref *iref;
+	int ret;
+	u32 size;
+
+	if (btrfs_backref_iter_is_inline_ref(iter)) {
+		/* We're still inside the inline refs */
+		ASSERT(iter->cur_ptr < iter->end_ptr);
+
+		if (btrfs_backref_has_tree_block_info(iter)) {
+			/* First tree block info */
+			size = sizeof(struct btrfs_tree_block_info);
+		} else {
+			/* Use inline ref type to determine the size */
+			int type;
+
+			iref = (struct btrfs_extent_inline_ref *)
+				((unsigned long)iter->cur_ptr);
+			type = btrfs_extent_inline_ref_type(eb, iref);
+
+			size = btrfs_extent_inline_ref_size(type);
+		}
+		iter->cur_ptr += size;
+		if (iter->cur_ptr < iter->end_ptr)
+			return 0;
+
+		/* All inline items iterated, fall through */
+	}
+
+	/* We're at keyed items, there is no inline item, go to the next one */
+	extent_root = btrfs_extent_root(iter->fs_info, iter->bytenr);
+	ret = btrfs_next_item(extent_root, iter->path);
+	if (ret)
+		return ret;
+
+	btrfs_item_key_to_cpu(path->nodes[0], &iter->cur_key, path->slots[0]);
+	if (iter->cur_key.objectid != iter->bytenr ||
+	    (iter->cur_key.type != BTRFS_TREE_BLOCK_REF_KEY &&
+	     iter->cur_key.type != BTRFS_SHARED_BLOCK_REF_KEY))
+		return 1;
+	iter->item_ptr = (u32)btrfs_item_ptr_offset(path->nodes[0],
+					path->slots[0]);
+	iter->cur_ptr = iter->item_ptr;
+	iter->end_ptr = iter->item_ptr + (u32)btrfs_item_size(path->nodes[0],
+						path->slots[0]);
+	return 0;
+}
+
+void btrfs_backref_init_cache(struct btrfs_fs_info *fs_info,
+			      struct btrfs_backref_cache *cache, int is_reloc)
+{
+	int i;
+
+	cache->rb_root = RB_ROOT;
+	for (i = 0; i < BTRFS_MAX_LEVEL; i++)
+		INIT_LIST_HEAD(&cache->pending[i]);
+	INIT_LIST_HEAD(&cache->changed);
+	INIT_LIST_HEAD(&cache->detached);
+	INIT_LIST_HEAD(&cache->leaves);
+	INIT_LIST_HEAD(&cache->pending_edge);
+	INIT_LIST_HEAD(&cache->useless_node);
+	cache->fs_info = fs_info;
+	cache->is_reloc = is_reloc;
+}
+
+struct btrfs_backref_node *btrfs_backref_alloc_node(
+		struct btrfs_backref_cache *cache, u64 bytenr, int level)
+{
+	struct btrfs_backref_node *node;
+
+	ASSERT(level >= 0 && level < BTRFS_MAX_LEVEL);
+	node = kzalloc(sizeof(*node), GFP_NOFS);
+	if (!node)
+		return node;
+
+	INIT_LIST_HEAD(&node->list);
+	INIT_LIST_HEAD(&node->upper);
+	INIT_LIST_HEAD(&node->lower);
+	RB_CLEAR_NODE(&node->rb_node);
+	cache->nr_nodes++;
+	node->level = level;
+	node->bytenr = bytenr;
+
+	return node;
+}
+
+struct btrfs_backref_edge *btrfs_backref_alloc_edge(
+		struct btrfs_backref_cache *cache)
+{
+	struct btrfs_backref_edge *edge;
+
+	edge = kzalloc(sizeof(*edge), GFP_NOFS);
+	if (edge)
+		cache->nr_edges++;
+	return edge;
+}
+
+/*
+ * Drop the backref node from cache, also cleaning up all its
+ * upper edges and any uncached nodes in the path.
+ *
+ * This cleanup happens bottom up, thus the node should either
+ * be the lowest node in the cache or a detached node.
+ */
+void btrfs_backref_cleanup_node(struct btrfs_backref_cache *cache,
+				struct btrfs_backref_node *node)
+{
+	struct btrfs_backref_node *upper;
+	struct btrfs_backref_edge *edge;
+
+	if (!node)
+		return;
+
+	BUG_ON(!node->lowest && !node->detached);
+	while (!list_empty(&node->upper)) {
+		edge = list_entry(node->upper.next, struct btrfs_backref_edge,
+				  list[LOWER]);
+		upper = edge->node[UPPER];
+		list_del(&edge->list[LOWER]);
+		list_del(&edge->list[UPPER]);
+		btrfs_backref_free_edge(cache, edge);
+
+		/*
+		 * Add the node to leaf node list if no other child block
+		 * cached.
+		 */
+		if (list_empty(&upper->lower)) {
+			list_add_tail(&upper->lower, &cache->leaves);
+			upper->lowest = 1;
+		}
+	}
+
+	btrfs_backref_drop_node(cache, node);
+}
+
+/*
+ * Release all nodes/edges from current cache
+ */
+void btrfs_backref_release_cache(struct btrfs_backref_cache *cache)
+{
+	struct btrfs_backref_node *node;
+	int i;
+
+	while (!list_empty(&cache->detached)) {
+		node = list_entry(cache->detached.next,
+				  struct btrfs_backref_node, list);
+		btrfs_backref_cleanup_node(cache, node);
+	}
+
+	while (!list_empty(&cache->leaves)) {
+		node = list_entry(cache->leaves.next,
+				  struct btrfs_backref_node, lower);
+		btrfs_backref_cleanup_node(cache, node);
+	}
+
+	cache->last_trans = 0;
+
+	for (i = 0; i < BTRFS_MAX_LEVEL; i++)
+		ASSERT(list_empty(&cache->pending[i]));
+	ASSERT(list_empty(&cache->pending_edge));
+	ASSERT(list_empty(&cache->useless_node));
+	ASSERT(list_empty(&cache->changed));
+	ASSERT(list_empty(&cache->detached));
+	ASSERT(RB_EMPTY_ROOT(&cache->rb_root));
+	ASSERT(!cache->nr_nodes);
+	ASSERT(!cache->nr_edges);
+}
+
+/*
+ * Handle direct tree backref
+ *
+ * Direct tree backref means, the backref item shows its parent bytenr
+ * directly. This is for SHARED_BLOCK_REF backref (keyed or inlined).
+ *
+ * @ref_key:	The converted backref key.
+ *		For keyed backref, it's the item key.
+ *		For inlined backref, objectid is the bytenr,
+ *		type is btrfs_inline_ref_type, offset is
+ *		btrfs_inline_ref_offset.
+ */
+static int handle_direct_tree_backref(struct btrfs_backref_cache *cache,
+				      struct btrfs_key *ref_key,
+				      struct btrfs_backref_node *cur)
+{
+	struct btrfs_backref_edge *edge;
+	struct btrfs_backref_node *upper;
+	struct rb_node *rb_node;
+
+	ASSERT(ref_key->type == BTRFS_SHARED_BLOCK_REF_KEY);
+
+	/* Only reloc root uses backref pointing to itself */
+	if (ref_key->objectid == ref_key->offset) {
+		struct btrfs_root *root;
+
+		cur->is_reloc_root = 1;
+		/* Only reloc backref cache cares about a specific root */
+		if (cache->is_reloc) {
+			root = find_reloc_root(cache->fs_info, cur->bytenr);
+			if (!root)
+				return -ENOENT;
+			cur->root = root;
+		} else {
+			/*
+			 * For generic purpose backref cache, reloc root node
+			 * is useless.
+			 */
+			list_add(&cur->list, &cache->useless_node);
+		}
+		return 0;
+	}
+
+	edge = btrfs_backref_alloc_edge(cache);
+	if (!edge)
+		return -ENOMEM;
+
+	rb_node = rb_simple_search(&cache->rb_root, ref_key->offset);
+	if (!rb_node) {
+		/* Parent node not yet cached */
+		upper = btrfs_backref_alloc_node(cache, ref_key->offset,
+					   cur->level + 1);
+		if (!upper) {
+			btrfs_backref_free_edge(cache, edge);
+			return -ENOMEM;
+		}
+
+		/*
+		 *  Backrefs for the upper level block isn't cached, add the
+		 *  block to pending list
+		 */
+		list_add_tail(&edge->list[UPPER], &cache->pending_edge);
+	} else {
+		/* Parent node already cached */
+		upper = rb_entry(rb_node, struct btrfs_backref_node, rb_node);
+		ASSERT(upper->checked);
+		INIT_LIST_HEAD(&edge->list[UPPER]);
+	}
+	btrfs_backref_link_edge(edge, cur, upper, LINK_LOWER);
+	return 0;
+}
+
+/*
+ * Handle indirect tree backref
+ *
+ * Indirect tree backref means, we only know which tree the node belongs to.
+ * We still need to do a tree search to find out the parents. This is for
+ * TREE_BLOCK_REF backref (keyed or inlined).
+ *
+ * @ref_key:	The same as @ref_key in  handle_direct_tree_backref()
+ * @tree_key:	The first key of this tree block.
+ * @path:	A clean (released) path, to avoid allocating path every time
+ *		the function get called.
+ */
+static int handle_indirect_tree_backref(struct btrfs_backref_cache *cache,
+					struct btrfs_path *path,
+					struct btrfs_key *ref_key,
+					struct btrfs_key *tree_key,
+					struct btrfs_backref_node *cur)
+{
+	struct btrfs_fs_info *fs_info = cache->fs_info;
+	struct btrfs_backref_node *upper;
+	struct btrfs_backref_node *lower;
+	struct btrfs_backref_edge *edge;
+	struct extent_buffer *eb;
+	struct btrfs_root *root;
+	struct rb_node *rb_node;
+	int level;
+	bool need_check = true;
+	int ret;
+
+	root = btrfs_get_fs_root(fs_info, ref_key->offset, false);
+	if (IS_ERR(root))
+		return PTR_ERR(root);
+	if (!test_bit(BTRFS_ROOT_SHAREABLE, &root->state))
+		cur->cowonly = 1;
+
+	if (btrfs_root_level(&root->root_item) == cur->level) {
+		/* Tree root */
+		ASSERT(btrfs_root_bytenr(&root->root_item) == cur->bytenr);
+		/*
+		 * For reloc backref cache, we may ignore reloc root.  But for
+		 * general purpose backref cache, we can't rely on
+		 * btrfs_should_ignore_reloc_root() as it may conflict with
+		 * current running relocation and lead to missing root.
+		 *
+		 * For general purpose backref cache, reloc root detection is
+		 * completely relying on direct backref (key->offset is parent
+		 * bytenr), thus only do such check for reloc cache.
+		 */
+		if (btrfs_should_ignore_reloc_root(root) && cache->is_reloc) {
+			btrfs_put_root(root);
+			list_add(&cur->list, &cache->useless_node);
+		} else {
+			cur->root = root;
+		}
+		return 0;
+	}
+
+	level = cur->level + 1;
+
+	/* Search the tree to find parent blocks referring to the block */
+	path->search_commit_root = 1;
+	path->skip_locking = 1;
+	path->lowest_level = level;
+	ret = btrfs_search_slot(NULL, root, tree_key, path, 0, 0);
+	path->lowest_level = 0;
+	if (ret < 0) {
+		btrfs_put_root(root);
+		return ret;
+	}
+	if (ret > 0 && path->slots[level] > 0)
+		path->slots[level]--;
+
+	eb = path->nodes[level];
+	if (btrfs_node_blockptr(eb, path->slots[level]) != cur->bytenr) {
+		btrfs_err(fs_info,
+"couldn't find block (%llu) (level %d) in tree (%llu) with key (%llu %u %llu)",
+			  cur->bytenr, level - 1, root->root_key.objectid,
+			  tree_key->objectid, tree_key->type, tree_key->offset);
+		btrfs_put_root(root);
+		ret = -ENOENT;
+		goto out;
+	}
+	lower = cur;
+
+	/* Add all nodes and edges in the path */
+	for (; level < BTRFS_MAX_LEVEL; level++) {
+		if (!path->nodes[level]) {
+			ASSERT(btrfs_root_bytenr(&root->root_item) ==
+			       lower->bytenr);
+			/* Same as previous should_ignore_reloc_root() call */
+			if (btrfs_should_ignore_reloc_root(root) &&
+			    cache->is_reloc) {
+				btrfs_put_root(root);
+				list_add(&lower->list, &cache->useless_node);
+			} else {
+				lower->root = root;
+			}
+			break;
+		}
+
+		edge = btrfs_backref_alloc_edge(cache);
+		if (!edge) {
+			btrfs_put_root(root);
+			ret = -ENOMEM;
+			goto out;
+		}
+
+		eb = path->nodes[level];
+		rb_node = rb_simple_search(&cache->rb_root, eb->start);
+		if (!rb_node) {
+			upper = btrfs_backref_alloc_node(cache, eb->start,
+							 lower->level + 1);
+			if (!upper) {
+				btrfs_put_root(root);
+				btrfs_backref_free_edge(cache, edge);
+				ret = -ENOMEM;
+				goto out;
+			}
+			upper->owner = btrfs_header_owner(eb);
+			if (!test_bit(BTRFS_ROOT_SHAREABLE, &root->state))
+				upper->cowonly = 1;
+
+			/*
+			 * If we know the block isn't shared we can avoid
+			 * checking its backrefs.
+			 */
+			if (btrfs_block_can_be_shared(root, eb))
+				upper->checked = 0;
+			else
+				upper->checked = 1;
+
+			/*
+			 * Add the block to pending list if we need to check its
+			 * backrefs, we only do this once while walking up a
+			 * tree as we will catch anything else later on.
+			 */
+			if (!upper->checked && need_check) {
+				need_check = false;
+				list_add_tail(&edge->list[UPPER],
+					      &cache->pending_edge);
+			} else {
+				if (upper->checked)
+					need_check = true;
+				INIT_LIST_HEAD(&edge->list[UPPER]);
+			}
+		} else {
+			upper = rb_entry(rb_node, struct btrfs_backref_node,
+					 rb_node);
+			ASSERT(upper->checked);
+			INIT_LIST_HEAD(&edge->list[UPPER]);
+			if (!upper->owner)
+				upper->owner = btrfs_header_owner(eb);
+		}
+		btrfs_backref_link_edge(edge, lower, upper, LINK_LOWER);
+
+		if (rb_node) {
+			btrfs_put_root(root);
+			break;
+		}
+		lower = upper;
+		upper = NULL;
+	}
+out:
+	btrfs_release_path(path);
+	return ret;
+}
+
+/*
+ * Add backref node @cur into @cache.
+ *
+ * NOTE: Even if the function returned 0, @cur is not yet cached as its upper
+ *	 links aren't yet bi-directional. Needs to finish such links.
+ *	 Use btrfs_backref_finish_upper_links() to finish such linkage.
+ *
+ * @path:	Released path for indirect tree backref lookup
+ * @iter:	Released backref iter for extent tree search
+ * @node_key:	The first key of the tree block
+ */
+int btrfs_backref_add_tree_node(struct btrfs_backref_cache *cache,
+				struct btrfs_path *path,
+				struct btrfs_backref_iter *iter,
+				struct btrfs_key *node_key,
+				struct btrfs_backref_node *cur)
+{
+	struct btrfs_fs_info *fs_info = cache->fs_info;
+	struct btrfs_backref_edge *edge;
+	struct btrfs_backref_node *exist;
+	int ret;
+
+	ret = btrfs_backref_iter_start(iter, cur->bytenr);
+	if (ret < 0)
+		return ret;
+	/*
+	 * We skip the first btrfs_tree_block_info, as we don't use the key
+	 * stored in it, but fetch it from the tree block
+	 */
+	if (btrfs_backref_has_tree_block_info(iter)) {
+		ret = btrfs_backref_iter_next(iter);
+		if (ret < 0)
+			goto out;
+		/* No extra backref? This means the tree block is corrupted */
+		if (ret > 0) {
+			ret = -EUCLEAN;
+			goto out;
+		}
+	}
+	WARN_ON(cur->checked);
+	if (!list_empty(&cur->upper)) {
+		/*
+		 * The backref was added previously when processing backref of
+		 * type BTRFS_TREE_BLOCK_REF_KEY
+		 */
+		ASSERT(list_is_singular(&cur->upper));
+		edge = list_entry(cur->upper.next, struct btrfs_backref_edge,
+				  list[LOWER]);
+		ASSERT(list_empty(&edge->list[UPPER]));
+		exist = edge->node[UPPER];
+		/*
+		 * Add the upper level block to pending list if we need check
+		 * its backrefs
+		 */
+		if (!exist->checked)
+			list_add_tail(&edge->list[UPPER], &cache->pending_edge);
+	} else {
+		exist = NULL;
+	}
+
+	for (; ret == 0; ret = btrfs_backref_iter_next(iter)) {
+		struct extent_buffer *eb;
+		struct btrfs_key key;
+		int type;
+
+		cond_resched();
+		eb = btrfs_backref_get_eb(iter);
+
+		key.objectid = iter->bytenr;
+		if (btrfs_backref_iter_is_inline_ref(iter)) {
+			struct btrfs_extent_inline_ref *iref;
+
+			/* Update key for inline backref */
+			iref = (struct btrfs_extent_inline_ref *)
+				((unsigned long)iter->cur_ptr);
+			type = btrfs_get_extent_inline_ref_type(eb, iref,
+							BTRFS_REF_TYPE_BLOCK);
+			if (type == BTRFS_REF_TYPE_INVALID) {
+				ret = -EUCLEAN;
+				goto out;
+			}
+			key.type = type;
+			key.offset = btrfs_extent_inline_ref_offset(eb, iref);
+		} else {
+			key.type = iter->cur_key.type;
+			key.offset = iter->cur_key.offset;
+		}
+
+		/*
+		 * Parent node found and matches current inline ref, no need to
+		 * rebuild this node for this inline ref
+		 */
+		if (exist &&
+		    ((key.type == BTRFS_TREE_BLOCK_REF_KEY &&
+		      exist->owner == key.offset) ||
+		     (key.type == BTRFS_SHARED_BLOCK_REF_KEY &&
+		      exist->bytenr == key.offset))) {
+			exist = NULL;
+			continue;
+		}
+
+		/* SHARED_BLOCK_REF means key.offset is the parent bytenr */
+		if (key.type == BTRFS_SHARED_BLOCK_REF_KEY) {
+			ret = handle_direct_tree_backref(cache, &key, cur);
+			if (ret < 0)
+				goto out;
+			continue;
+		} else if (unlikely(key.type == BTRFS_EXTENT_REF_V0_KEY)) {
+			ret = -EINVAL;
+			btrfs_print_v0_err(fs_info);
+			btrfs_handle_fs_error(fs_info, ret, NULL);
+			goto out;
+		} else if (key.type != BTRFS_TREE_BLOCK_REF_KEY) {
+			continue;
+		}
+
+		/*
+		 * key.type == BTRFS_TREE_BLOCK_REF_KEY, inline ref offset
+		 * means the root objectid. We need to search the tree to get
+		 * its parent bytenr.
+		 */
+		ret = handle_indirect_tree_backref(cache, path, &key, node_key,
+						   cur);
+		if (ret < 0)
+			goto out;
+	}
+	ret = 0;
+	cur->checked = 1;
+	WARN_ON(exist);
+out:
+	btrfs_backref_iter_release(iter);
+	return ret;
+}
+
+/*
+ * Finish the upwards linkage created by btrfs_backref_add_tree_node()
+ */
+int btrfs_backref_finish_upper_links(struct btrfs_backref_cache *cache,
+				     struct btrfs_backref_node *start)
+{
+	struct list_head *useless_node = &cache->useless_node;
+	struct btrfs_backref_edge *edge;
+	struct rb_node *rb_node;
+	LIST_HEAD(pending_edge);
+
+	ASSERT(start->checked);
+
+	/* Insert this node to cache if it's not COW-only */
+	if (!start->cowonly) {
+		rb_node = rb_simple_insert(&cache->rb_root, start->bytenr,
+					   &start->rb_node);
+		if (rb_node)
+			btrfs_backref_panic(cache->fs_info, start->bytenr,
+					    -EEXIST);
+		list_add_tail(&start->lower, &cache->leaves);
+	}
+
+	/*
+	 * Use breadth first search to iterate all related edges.
+	 *
+	 * The starting points are all the edges of this node
+	 */
+	list_for_each_entry(edge, &start->upper, list[LOWER])
+		list_add_tail(&edge->list[UPPER], &pending_edge);
+
+	while (!list_empty(&pending_edge)) {
+		struct btrfs_backref_node *upper;
+		struct btrfs_backref_node *lower;
+
+		edge = list_first_entry(&pending_edge,
+				struct btrfs_backref_edge, list[UPPER]);
+		list_del_init(&edge->list[UPPER]);
+		upper = edge->node[UPPER];
+		lower = edge->node[LOWER];
+
+		/* Parent is detached, no need to keep any edges */
+		if (upper->detached) {
+			list_del(&edge->list[LOWER]);
+			btrfs_backref_free_edge(cache, edge);
+
+			/* Lower node is orphan, queue for cleanup */
+			if (list_empty(&lower->upper))
+				list_add(&lower->list, useless_node);
+			continue;
+		}
+
+		/*
+		 * All new nodes added in current build_backref_tree() haven't
+		 * been linked to the cache rb tree.
+		 * So if we have upper->rb_node populated, this means a cache
+		 * hit. We only need to link the edge, as @upper and all its
+		 * parents have already been linked.
+		 */
+		if (!RB_EMPTY_NODE(&upper->rb_node)) {
+			if (upper->lowest) {
+				list_del_init(&upper->lower);
+				upper->lowest = 0;
+			}
+
+			list_add_tail(&edge->list[UPPER], &upper->lower);
+			continue;
+		}
+
+		/* Sanity check, we shouldn't have any unchecked nodes */
+		if (!upper->checked) {
+			ASSERT(0);
+			return -EUCLEAN;
+		}
+
+		/* Sanity check, COW-only node has non-COW-only parent */
+		if (start->cowonly != upper->cowonly) {
+			ASSERT(0);
+			return -EUCLEAN;
+		}
+
+		/* Only cache non-COW-only (subvolume trees) tree blocks */
+		if (!upper->cowonly) {
+			rb_node = rb_simple_insert(&cache->rb_root, upper->bytenr,
+						   &upper->rb_node);
+			if (rb_node) {
+				btrfs_backref_panic(cache->fs_info,
+						upper->bytenr, -EEXIST);
+				return -EUCLEAN;
+			}
+		}
+
+		list_add_tail(&edge->list[UPPER], &upper->lower);
+
+		/*
+		 * Also queue all the parent edges of this uncached node
+		 * to finish the upper linkage
+		 */
+		list_for_each_entry(edge, &upper->upper, list[LOWER])
+			list_add_tail(&edge->list[UPPER], &pending_edge);
+	}
+	return 0;
+}
+
+void btrfs_backref_error_cleanup(struct btrfs_backref_cache *cache,
+				 struct btrfs_backref_node *node)
+{
+	struct btrfs_backref_node *lower;
+	struct btrfs_backref_node *upper;
+	struct btrfs_backref_edge *edge;
+
+	while (!list_empty(&cache->useless_node)) {
+		lower = list_first_entry(&cache->useless_node,
+				   struct btrfs_backref_node, list);
+		list_del_init(&lower->list);
+	}
+	while (!list_empty(&cache->pending_edge)) {
+		edge = list_first_entry(&cache->pending_edge,
+				struct btrfs_backref_edge, list[UPPER]);
+		list_del(&edge->list[UPPER]);
+		list_del(&edge->list[LOWER]);
+		lower = edge->node[LOWER];
+		upper = edge->node[UPPER];
+		btrfs_backref_free_edge(cache, edge);
+
+		/*
+		 * Lower is no longer linked to any upper backref nodes and
+		 * isn't in the cache, we can free it ourselves.
+		 */
+		if (list_empty(&lower->upper) &&
+		    RB_EMPTY_NODE(&lower->rb_node))
+			list_add(&lower->list, &cache->useless_node);
+
+		if (!RB_EMPTY_NODE(&upper->rb_node))
+			continue;
+
+		/* Add this guy's upper edges to the list to process */
+		list_for_each_entry(edge, &upper->upper, list[LOWER])
+			list_add_tail(&edge->list[UPPER],
+				      &cache->pending_edge);
+		if (list_empty(&upper->upper))
+			list_add(&upper->list, &cache->useless_node);
+	}
+
+	while (!list_empty(&cache->useless_node)) {
+		lower = list_first_entry(&cache->useless_node,
+				   struct btrfs_backref_node, list);
+		list_del_init(&lower->list);
+		if (lower == node)
+			node = NULL;
+		btrfs_backref_drop_node(cache, lower);
+	}
+
+	btrfs_backref_cleanup_node(cache, node);
+	ASSERT(list_empty(&cache->useless_node) &&
+	       list_empty(&cache->pending_edge));
+}
diff --git a/fs/btrfs/backref.h b/fs/btrfs/backref.h
index 777f61dc081e..8e69584d538d 100644
--- a/fs/btrfs/backref.h
+++ b/fs/btrfs/backref.h
@@ -8,6 +8,7 @@
 
 #include <linux/btrfs.h>
 #include "ulist.h"
+#include "disk-io.h"
 #include "extent_io.h"
 
 struct inode_fs_paths {
@@ -16,6 +17,21 @@ struct inode_fs_paths {
 	struct btrfs_data_container	*fspath;
 };
 
+struct btrfs_backref_shared_cache_entry {
+	u64 bytenr;
+	u64 gen;
+	bool is_shared;
+};
+
+struct btrfs_backref_shared_cache {
+	/*
+	 * A path from a root to a leaf that has a file extent item pointing to
+	 * a given data extent should never exceed the maximum b+tree height.
+	 */
+	struct btrfs_backref_shared_cache_entry entries[BTRFS_MAX_LEVEL];
+	bool use_cache;
+};
+
 typedef int (iterate_extent_inodes_t)(u64 inum, u64 offset, u64 root,
 		void *ctx);
 
@@ -34,15 +50,19 @@ int iterate_extent_inodes(struct btrfs_fs_info *fs_info,
 				bool ignore_offset);
 
 int iterate_inodes_from_logical(u64 logical, struct btrfs_fs_info *fs_info,
-				struct btrfs_path *path,
-				iterate_extent_inodes_t *iterate, void *ctx,
+				struct btrfs_path *path, void *ctx,
 				bool ignore_offset);
 
 int paths_from_inode(u64 inum, struct inode_fs_paths *ipath);
 
+int btrfs_find_all_leafs(struct btrfs_trans_handle *trans,
+			 struct btrfs_fs_info *fs_info, u64 bytenr,
+			 u64 time_seq, struct ulist **leafs,
+			 const u64 *extent_item_pos, bool ignore_offset);
 int btrfs_find_all_roots(struct btrfs_trans_handle *trans,
 			 struct btrfs_fs_info *fs_info, u64 bytenr,
-			 u64 time_seq, struct ulist **roots, bool ignore_offset);
+			 u64 time_seq, struct ulist **roots,
+			 bool skip_commit_root_sem);
 char *btrfs_ref_to_path(struct btrfs_root *fs_root, struct btrfs_path *path,
 			u32 name_len, unsigned long name_off,
 			struct extent_buffer *eb_in, u64 parent,
@@ -57,8 +77,10 @@ int btrfs_find_one_extref(struct btrfs_root *root, u64 inode_objectid,
 			  u64 start_off, struct btrfs_path *path,
 			  struct btrfs_inode_extref **ret_extref,
 			  u64 *found_off);
-int btrfs_check_shared(struct btrfs_root *root, u64 inum, u64 bytenr,
-		struct ulist *roots, struct ulist *tmp_ulist);
+int btrfs_is_data_extent_shared(struct btrfs_root *root, u64 inum, u64 bytenr,
+				u64 extent_gen,
+				struct ulist *roots, struct ulist *tmp,
+				struct btrfs_backref_shared_cache *cache);
 
 int __init btrfs_prelim_ref_init(void);
 void __cold btrfs_prelim_ref_exit(void);
@@ -74,4 +96,303 @@ struct prelim_ref {
 	u64 wanted_disk_byte;
 };
 
+/*
+ * Iterate backrefs of one extent.
+ *
+ * Now it only supports iteration of tree block in commit root.
+ */
+struct btrfs_backref_iter {
+	u64 bytenr;
+	struct btrfs_path *path;
+	struct btrfs_fs_info *fs_info;
+	struct btrfs_key cur_key;
+	u32 item_ptr;
+	u32 cur_ptr;
+	u32 end_ptr;
+};
+
+struct btrfs_backref_iter *btrfs_backref_iter_alloc(
+		struct btrfs_fs_info *fs_info, gfp_t gfp_flag);
+
+static inline void btrfs_backref_iter_free(struct btrfs_backref_iter *iter)
+{
+	if (!iter)
+		return;
+	btrfs_free_path(iter->path);
+	kfree(iter);
+}
+
+static inline struct extent_buffer *btrfs_backref_get_eb(
+		struct btrfs_backref_iter *iter)
+{
+	if (!iter)
+		return NULL;
+	return iter->path->nodes[0];
+}
+
+/*
+ * For metadata with EXTENT_ITEM key (non-skinny) case, the first inline data
+ * is btrfs_tree_block_info, without a btrfs_extent_inline_ref header.
+ *
+ * This helper determines if that's the case.
+ */
+static inline bool btrfs_backref_has_tree_block_info(
+		struct btrfs_backref_iter *iter)
+{
+	if (iter->cur_key.type == BTRFS_EXTENT_ITEM_KEY &&
+	    iter->cur_ptr - iter->item_ptr == sizeof(struct btrfs_extent_item))
+		return true;
+	return false;
+}
+
+int btrfs_backref_iter_start(struct btrfs_backref_iter *iter, u64 bytenr);
+
+int btrfs_backref_iter_next(struct btrfs_backref_iter *iter);
+
+static inline bool btrfs_backref_iter_is_inline_ref(
+		struct btrfs_backref_iter *iter)
+{
+	if (iter->cur_key.type == BTRFS_EXTENT_ITEM_KEY ||
+	    iter->cur_key.type == BTRFS_METADATA_ITEM_KEY)
+		return true;
+	return false;
+}
+
+static inline void btrfs_backref_iter_release(struct btrfs_backref_iter *iter)
+{
+	iter->bytenr = 0;
+	iter->item_ptr = 0;
+	iter->cur_ptr = 0;
+	iter->end_ptr = 0;
+	btrfs_release_path(iter->path);
+	memset(&iter->cur_key, 0, sizeof(iter->cur_key));
+}
+
+/*
+ * Backref cache related structures
+ *
+ * The whole objective of backref_cache is to build a bi-directional map
+ * of tree blocks (represented by backref_node) and all their parents.
+ */
+
+/*
+ * Represent a tree block in the backref cache
+ */
+struct btrfs_backref_node {
+	struct {
+		struct rb_node rb_node;
+		u64 bytenr;
+	}; /* Use rb_simple_node for search/insert */
+
+	u64 new_bytenr;
+	/* Objectid of tree block owner, can be not uptodate */
+	u64 owner;
+	/* Link to pending, changed or detached list */
+	struct list_head list;
+
+	/* List of upper level edges, which link this node to its parents */
+	struct list_head upper;
+	/* List of lower level edges, which link this node to its children */
+	struct list_head lower;
+
+	/* NULL if this node is not tree root */
+	struct btrfs_root *root;
+	/* Extent buffer got by COWing the block */
+	struct extent_buffer *eb;
+	/* Level of the tree block */
+	unsigned int level:8;
+	/* Is the block in a non-shareable tree */
+	unsigned int cowonly:1;
+	/* 1 if no child node is in the cache */
+	unsigned int lowest:1;
+	/* Is the extent buffer locked */
+	unsigned int locked:1;
+	/* Has the block been processed */
+	unsigned int processed:1;
+	/* Have backrefs of this block been checked */
+	unsigned int checked:1;
+	/*
+	 * 1 if corresponding block has been COWed but some upper level block
+	 * pointers may not point to the new location
+	 */
+	unsigned int pending:1;
+	/* 1 if the backref node isn't connected to any other backref node */
+	unsigned int detached:1;
+
+	/*
+	 * For generic purpose backref cache, where we only care if it's a reloc
+	 * root, doesn't care the source subvolid.
+	 */
+	unsigned int is_reloc_root:1;
+};
+
+#define LOWER	0
+#define UPPER	1
+
+/*
+ * Represent an edge connecting upper and lower backref nodes.
+ */
+struct btrfs_backref_edge {
+	/*
+	 * list[LOWER] is linked to btrfs_backref_node::upper of lower level
+	 * node, and list[UPPER] is linked to btrfs_backref_node::lower of
+	 * upper level node.
+	 *
+	 * Also, build_backref_tree() uses list[UPPER] for pending edges, before
+	 * linking list[UPPER] to its upper level nodes.
+	 */
+	struct list_head list[2];
+
+	/* Two related nodes */
+	struct btrfs_backref_node *node[2];
+};
+
+struct btrfs_backref_cache {
+	/* Red black tree of all backref nodes in the cache */
+	struct rb_root rb_root;
+	/* For passing backref nodes to btrfs_reloc_cow_block */
+	struct btrfs_backref_node *path[BTRFS_MAX_LEVEL];
+	/*
+	 * List of blocks that have been COWed but some block pointers in upper
+	 * level blocks may not reflect the new location
+	 */
+	struct list_head pending[BTRFS_MAX_LEVEL];
+	/* List of backref nodes with no child node */
+	struct list_head leaves;
+	/* List of blocks that have been COWed in current transaction */
+	struct list_head changed;
+	/* List of detached backref node. */
+	struct list_head detached;
+
+	u64 last_trans;
+
+	int nr_nodes;
+	int nr_edges;
+
+	/* List of unchecked backref edges during backref cache build */
+	struct list_head pending_edge;
+
+	/* List of useless backref nodes during backref cache build */
+	struct list_head useless_node;
+
+	struct btrfs_fs_info *fs_info;
+
+	/*
+	 * Whether this cache is for relocation
+	 *
+	 * Reloction backref cache require more info for reloc root compared
+	 * to generic backref cache.
+	 */
+	unsigned int is_reloc;
+};
+
+void btrfs_backref_init_cache(struct btrfs_fs_info *fs_info,
+			      struct btrfs_backref_cache *cache, int is_reloc);
+struct btrfs_backref_node *btrfs_backref_alloc_node(
+		struct btrfs_backref_cache *cache, u64 bytenr, int level);
+struct btrfs_backref_edge *btrfs_backref_alloc_edge(
+		struct btrfs_backref_cache *cache);
+
+#define		LINK_LOWER	(1 << 0)
+#define		LINK_UPPER	(1 << 1)
+static inline void btrfs_backref_link_edge(struct btrfs_backref_edge *edge,
+					   struct btrfs_backref_node *lower,
+					   struct btrfs_backref_node *upper,
+					   int link_which)
+{
+	ASSERT(upper && lower && upper->level == lower->level + 1);
+	edge->node[LOWER] = lower;
+	edge->node[UPPER] = upper;
+	if (link_which & LINK_LOWER)
+		list_add_tail(&edge->list[LOWER], &lower->upper);
+	if (link_which & LINK_UPPER)
+		list_add_tail(&edge->list[UPPER], &upper->lower);
+}
+
+static inline void btrfs_backref_free_node(struct btrfs_backref_cache *cache,
+					   struct btrfs_backref_node *node)
+{
+	if (node) {
+		ASSERT(list_empty(&node->list));
+		ASSERT(list_empty(&node->lower));
+		ASSERT(node->eb == NULL);
+		cache->nr_nodes--;
+		btrfs_put_root(node->root);
+		kfree(node);
+	}
+}
+
+static inline void btrfs_backref_free_edge(struct btrfs_backref_cache *cache,
+					   struct btrfs_backref_edge *edge)
+{
+	if (edge) {
+		cache->nr_edges--;
+		kfree(edge);
+	}
+}
+
+static inline void btrfs_backref_unlock_node_buffer(
+		struct btrfs_backref_node *node)
+{
+	if (node->locked) {
+		btrfs_tree_unlock(node->eb);
+		node->locked = 0;
+	}
+}
+
+static inline void btrfs_backref_drop_node_buffer(
+		struct btrfs_backref_node *node)
+{
+	if (node->eb) {
+		btrfs_backref_unlock_node_buffer(node);
+		free_extent_buffer(node->eb);
+		node->eb = NULL;
+	}
+}
+
+/*
+ * Drop the backref node from cache without cleaning up its children
+ * edges.
+ *
+ * This can only be called on node without parent edges.
+ * The children edges are still kept as is.
+ */
+static inline void btrfs_backref_drop_node(struct btrfs_backref_cache *tree,
+					   struct btrfs_backref_node *node)
+{
+	ASSERT(list_empty(&node->upper));
+
+	btrfs_backref_drop_node_buffer(node);
+	list_del_init(&node->list);
+	list_del_init(&node->lower);
+	if (!RB_EMPTY_NODE(&node->rb_node))
+		rb_erase(&node->rb_node, &tree->rb_root);
+	btrfs_backref_free_node(tree, node);
+}
+
+void btrfs_backref_cleanup_node(struct btrfs_backref_cache *cache,
+				struct btrfs_backref_node *node);
+
+void btrfs_backref_release_cache(struct btrfs_backref_cache *cache);
+
+static inline void btrfs_backref_panic(struct btrfs_fs_info *fs_info,
+				       u64 bytenr, int errno)
+{
+	btrfs_panic(fs_info, errno,
+		    "Inconsistency in backref cache found at offset %llu",
+		    bytenr);
+}
+
+int btrfs_backref_add_tree_node(struct btrfs_backref_cache *cache,
+				struct btrfs_path *path,
+				struct btrfs_backref_iter *iter,
+				struct btrfs_key *node_key,
+				struct btrfs_backref_node *cur);
+
+int btrfs_backref_finish_upper_links(struct btrfs_backref_cache *cache,
+				     struct btrfs_backref_node *start);
+
+void btrfs_backref_error_cleanup(struct btrfs_backref_cache *cache,
+				 struct btrfs_backref_node *node);
+
 #endif
diff --git a/fs/btrfs/block-group.c b/fs/btrfs/block-group.c
index 7f09147872dc..deebc8ddbd93 100644
--- a/fs/btrfs/block-group.c
+++ b/fs/btrfs/block-group.c
@@ -1,5 +1,6 @@
 // SPDX-License-Identifier: GPL-2.0
 
+#include <linux/list_sort.h>
 #include "misc.h"
 #include "ctree.h"
 #include "block-group.h"
@@ -7,7 +8,6 @@
 #include "disk-io.h"
 #include "free-space-cache.h"
 #include "free-space-tree.h"
-#include "disk-io.h"
 #include "volumes.h"
 #include "transaction.h"
 #include "ref-verify.h"
@@ -16,6 +16,7 @@
 #include "delalloc-space.h"
 #include "discard.h"
 #include "raid56.h"
+#include "zoned.h"
 
 /*
  * Return target flags in extended format or 0 if restripe for this chunk_type
@@ -66,11 +67,8 @@ static u64 btrfs_reduce_alloc_profile(struct btrfs_fs_info *fs_info, u64 flags)
 	spin_lock(&fs_info->balance_lock);
 	target = get_restripe_target(fs_info, flags);
 	if (target) {
-		/* Pick target profile only if it's already available */
-		if ((flags & target) & BTRFS_EXTENDED_PROFILE_MASK) {
-			spin_unlock(&fs_info->balance_lock);
-			return extended_to_chunk(target);
-		}
+		spin_unlock(&fs_info->balance_lock);
+		return extended_to_chunk(target);
 	}
 	spin_unlock(&fs_info->balance_lock);
 
@@ -119,14 +117,23 @@ u64 btrfs_get_alloc_profile(struct btrfs_fs_info *fs_info, u64 orig_flags)
 
 void btrfs_get_block_group(struct btrfs_block_group *cache)
 {
-	atomic_inc(&cache->count);
+	refcount_inc(&cache->refs);
 }
 
 void btrfs_put_block_group(struct btrfs_block_group *cache)
 {
-	if (atomic_dec_and_test(&cache->count)) {
+	if (refcount_dec_and_test(&cache->refs)) {
 		WARN_ON(cache->pinned > 0);
-		WARN_ON(cache->reserved > 0);
+		/*
+		 * If there was a failure to cleanup a log tree, very likely due
+		 * to an IO failure on a writeback attempt of one or more of its
+		 * extent buffers, we could not do proper (and cheap) unaccounting
+		 * of their reserved space, so don't warn on reserved > 0 in that
+		 * case.
+		 */
+		if (!(cache->flags & BTRFS_BLOCK_GROUP_METADATA) ||
+		    !BTRFS_FS_LOG_CLEANUP_ERROR(cache->fs_info))
+			WARN_ON(cache->reserved > 0);
 
 		/*
 		 * A block_group shouldn't be on the discard_list anymore.
@@ -147,6 +154,7 @@ void btrfs_put_block_group(struct btrfs_block_group *cache)
 		 */
 		WARN_ON(!RB_EMPTY_ROOT(&cache->full_stripe_locks_root.root));
 		kfree(cache->free_space_ctl);
+		kfree(cache->physical_map);
 		kfree(cache);
 	}
 }
@@ -160,9 +168,12 @@ static int btrfs_add_block_group_cache(struct btrfs_fs_info *info,
 	struct rb_node **p;
 	struct rb_node *parent = NULL;
 	struct btrfs_block_group *cache;
+	bool leftmost = true;
 
-	spin_lock(&info->block_group_cache_lock);
-	p = &info->block_group_cache_tree.rb_node;
+	ASSERT(block_group->length != 0);
+
+	write_lock(&info->block_group_cache_lock);
+	p = &info->block_group_cache_tree.rb_root.rb_node;
 
 	while (*p) {
 		parent = *p;
@@ -171,20 +182,18 @@ static int btrfs_add_block_group_cache(struct btrfs_fs_info *info,
 			p = &(*p)->rb_left;
 		} else if (block_group->start > cache->start) {
 			p = &(*p)->rb_right;
+			leftmost = false;
 		} else {
-			spin_unlock(&info->block_group_cache_lock);
+			write_unlock(&info->block_group_cache_lock);
 			return -EEXIST;
 		}
 	}
 
 	rb_link_node(&block_group->cache_node, parent, p);
-	rb_insert_color(&block_group->cache_node,
-			&info->block_group_cache_tree);
-
-	if (info->first_logical_byte > block_group->start)
-		info->first_logical_byte = block_group->start;
+	rb_insert_color_cached(&block_group->cache_node,
+			       &info->block_group_cache_tree, leftmost);
 
-	spin_unlock(&info->block_group_cache_lock);
+	write_unlock(&info->block_group_cache_lock);
 
 	return 0;
 }
@@ -200,8 +209,8 @@ static struct btrfs_block_group *block_group_cache_tree_search(
 	struct rb_node *n;
 	u64 end, start;
 
-	spin_lock(&info->block_group_cache_lock);
-	n = info->block_group_cache_tree.rb_node;
+	read_lock(&info->block_group_cache_lock);
+	n = info->block_group_cache_tree.rb_root.rb_node;
 
 	while (n) {
 		cache = rb_entry(n, struct btrfs_block_group, cache_node);
@@ -223,12 +232,9 @@ static struct btrfs_block_group *block_group_cache_tree_search(
 			break;
 		}
 	}
-	if (ret) {
+	if (ret)
 		btrfs_get_block_group(ret);
-		if (bytenr == 0 && info->first_logical_byte > ret->start)
-			info->first_logical_byte = ret->start;
-	}
-	spin_unlock(&info->block_group_cache_lock);
+	read_unlock(&info->block_group_cache_lock);
 
 	return ret;
 }
@@ -257,15 +263,15 @@ struct btrfs_block_group *btrfs_next_block_group(
 	struct btrfs_fs_info *fs_info = cache->fs_info;
 	struct rb_node *node;
 
-	spin_lock(&fs_info->block_group_cache_lock);
+	read_lock(&fs_info->block_group_cache_lock);
 
 	/* If our block group was removed, we need a full search. */
 	if (RB_EMPTY_NODE(&cache->cache_node)) {
 		const u64 next_bytenr = cache->start + cache->length;
 
-		spin_unlock(&fs_info->block_group_cache_lock);
+		read_unlock(&fs_info->block_group_cache_lock);
 		btrfs_put_block_group(cache);
-		cache = btrfs_lookup_first_block_group(fs_info, next_bytenr); return cache;
+		return btrfs_lookup_first_block_group(fs_info, next_bytenr);
 	}
 	node = rb_next(&cache->cache_node);
 	btrfs_put_block_group(cache);
@@ -274,46 +280,70 @@ struct btrfs_block_group *btrfs_next_block_group(
 		btrfs_get_block_group(cache);
 	} else
 		cache = NULL;
-	spin_unlock(&fs_info->block_group_cache_lock);
+	read_unlock(&fs_info->block_group_cache_lock);
 	return cache;
 }
 
-bool btrfs_inc_nocow_writers(struct btrfs_fs_info *fs_info, u64 bytenr)
+/**
+ * Check if we can do a NOCOW write for a given extent.
+ *
+ * @fs_info:       The filesystem information object.
+ * @bytenr:        Logical start address of the extent.
+ *
+ * Check if we can do a NOCOW write for the given extent, and increments the
+ * number of NOCOW writers in the block group that contains the extent, as long
+ * as the block group exists and it's currently not in read-only mode.
+ *
+ * Returns: A non-NULL block group pointer if we can do a NOCOW write, the caller
+ *          is responsible for calling btrfs_dec_nocow_writers() later.
+ *
+ *          Or NULL if we can not do a NOCOW write
+ */
+struct btrfs_block_group *btrfs_inc_nocow_writers(struct btrfs_fs_info *fs_info,
+						  u64 bytenr)
 {
 	struct btrfs_block_group *bg;
-	bool ret = true;
+	bool can_nocow = true;
 
 	bg = btrfs_lookup_block_group(fs_info, bytenr);
 	if (!bg)
-		return false;
+		return NULL;
 
 	spin_lock(&bg->lock);
 	if (bg->ro)
-		ret = false;
+		can_nocow = false;
 	else
 		atomic_inc(&bg->nocow_writers);
 	spin_unlock(&bg->lock);
 
-	/* No put on block group, done by btrfs_dec_nocow_writers */
-	if (!ret)
+	if (!can_nocow) {
 		btrfs_put_block_group(bg);
+		return NULL;
+	}
 
-	return ret;
+	/* No put on block group, done by btrfs_dec_nocow_writers(). */
+	return bg;
 }
 
-void btrfs_dec_nocow_writers(struct btrfs_fs_info *fs_info, u64 bytenr)
+/**
+ * Decrement the number of NOCOW writers in a block group.
+ *
+ * @bg:       The block group.
+ *
+ * This is meant to be called after a previous call to btrfs_inc_nocow_writers(),
+ * and on the block group returned by that call. Typically this is called after
+ * creating an ordered extent for a NOCOW write, to prevent races with scrub and
+ * relocation.
+ *
+ * After this call, the caller should not use the block group anymore. It it wants
+ * to use it, then it should get a reference on it before calling this function.
+ */
+void btrfs_dec_nocow_writers(struct btrfs_block_group *bg)
 {
-	struct btrfs_block_group *bg;
-
-	bg = btrfs_lookup_block_group(fs_info, bytenr);
-	ASSERT(bg);
 	if (atomic_dec_and_test(&bg->nocow_writers))
 		wake_up_var(&bg->nocow_writers);
-	/*
-	 * Once for our lookup and once for the lookup done by a previous call
-	 * to btrfs_inc_nocow_writers()
-	 */
-	btrfs_put_block_group(bg);
+
+	/* For the lookup done by a previous call to btrfs_inc_nocow_writers(). */
 	btrfs_put_block_group(bg);
 }
 
@@ -410,18 +440,22 @@ void btrfs_wait_block_group_cache_progress(struct btrfs_block_group *cache,
 	btrfs_put_caching_control(caching_ctl);
 }
 
-int btrfs_wait_block_group_cache_done(struct btrfs_block_group *cache)
+static int btrfs_caching_ctl_wait_done(struct btrfs_block_group *cache,
+				       struct btrfs_caching_control *caching_ctl)
+{
+	wait_event(caching_ctl->wait, btrfs_block_group_done(cache));
+	return cache->cached == BTRFS_CACHE_ERROR ? -EIO : 0;
+}
+
+static int btrfs_wait_block_group_cache_done(struct btrfs_block_group *cache)
 {
 	struct btrfs_caching_control *caching_ctl;
-	int ret = 0;
+	int ret;
 
 	caching_ctl = btrfs_get_caching_control(cache);
 	if (!caching_ctl)
 		return (cache->cached == BTRFS_CACHE_ERROR) ? -EIO : 0;
-
-	wait_event(caching_ctl->wait, btrfs_block_group_done(cache));
-	if (cache->cached == BTRFS_CACHE_ERROR)
-		ret = -EIO;
+	ret = btrfs_caching_ctl_wait_done(cache, caching_ctl);
 	btrfs_put_caching_control(caching_ctl);
 	return ret;
 }
@@ -460,7 +494,7 @@ u64 add_new_free_space(struct btrfs_block_group *block_group, u64 start, u64 end
 	int ret;
 
 	while (start < end) {
-		ret = find_first_extent_bit(info->pinned_extents, start,
+		ret = find_first_extent_bit(&info->excluded_extents, start,
 					    &extent_start, &extent_end,
 					    EXTENT_DIRTY | EXTENT_UPTODATE,
 					    NULL);
@@ -496,7 +530,7 @@ static int load_extent_tree_free(struct btrfs_caching_control *caching_ctl)
 {
 	struct btrfs_block_group *block_group = caching_ctl->block_group;
 	struct btrfs_fs_info *fs_info = block_group->fs_info;
-	struct btrfs_root *extent_root = fs_info->extent_root;
+	struct btrfs_root *extent_root;
 	struct btrfs_path *path;
 	struct extent_buffer *leaf;
 	struct btrfs_key key;
@@ -511,6 +545,7 @@ static int load_extent_tree_free(struct btrfs_caching_control *caching_ctl)
 		return -ENOMEM;
 
 	last = max_t(u64, block_group->start, BTRFS_SUPER_INFO_OFFSET);
+	extent_root = btrfs_extent_root(fs_info, last);
 
 #ifdef CONFIG_BTRFS_DEBUG
 	/*
@@ -558,8 +593,6 @@ next:
 
 			if (need_resched() ||
 			    rwsem_is_contended(&fs_info->commit_root_sem)) {
-				if (wakeup)
-					caching_ctl->progress = last;
 				btrfs_release_path(path);
 				up_read(&fs_info->commit_root_sem);
 				mutex_unlock(&caching_ctl->mutex);
@@ -583,9 +616,6 @@ next:
 			key.objectid = last;
 			key.offset = 0;
 			key.type = BTRFS_EXTENT_ITEM_KEY;
-
-			if (wakeup)
-				caching_ctl->progress = last;
 			btrfs_release_path(path);
 			goto next;
 		}
@@ -620,7 +650,6 @@ next:
 
 	total_found += add_new_free_space(block_group, last,
 				block_group->start + block_group->length);
-	caching_ctl->progress = (u64)-1;
 
 out:
 	btrfs_free_path(path);
@@ -641,11 +670,36 @@ static noinline void caching_thread(struct btrfs_work *work)
 	mutex_lock(&caching_ctl->mutex);
 	down_read(&fs_info->commit_root_sem);
 
-	if (btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE))
+	if (btrfs_test_opt(fs_info, SPACE_CACHE)) {
+		ret = load_free_space_cache(block_group);
+		if (ret == 1) {
+			ret = 0;
+			goto done;
+		}
+
+		/*
+		 * We failed to load the space cache, set ourselves to
+		 * CACHE_STARTED and carry on.
+		 */
+		spin_lock(&block_group->lock);
+		block_group->cached = BTRFS_CACHE_STARTED;
+		spin_unlock(&block_group->lock);
+		wake_up(&caching_ctl->wait);
+	}
+
+	/*
+	 * If we are in the transaction that populated the free space tree we
+	 * can't actually cache from the free space tree as our commit root and
+	 * real root are the same, so we could change the contents of the blocks
+	 * while caching.  Instead do the slow caching in this case, and after
+	 * the transaction has committed we will be safe.
+	 */
+	if (btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE) &&
+	    !(test_bit(BTRFS_FS_FREE_SPACE_TREE_UNTRUSTED, &fs_info->flags)))
 		ret = load_free_space_tree(caching_ctl);
 	else
 		ret = load_extent_tree_free(caching_ctl);
-
+done:
 	spin_lock(&block_group->lock);
 	block_group->caching_ctl = NULL;
 	block_group->cached = ret ? BTRFS_CACHE_ERROR : BTRFS_CACHE_FINISHED;
@@ -665,8 +719,6 @@ static noinline void caching_thread(struct btrfs_work *work)
 	}
 #endif
 
-	caching_ctl->progress = (u64)-1;
-
 	up_read(&fs_info->commit_root_sem);
 	btrfs_free_excluded_extents(block_group);
 	mutex_unlock(&caching_ctl->mutex);
@@ -677,13 +729,16 @@ static noinline void caching_thread(struct btrfs_work *work)
 	btrfs_put_block_group(block_group);
 }
 
-int btrfs_cache_block_group(struct btrfs_block_group *cache, int load_cache_only)
+int btrfs_cache_block_group(struct btrfs_block_group *cache, bool wait)
 {
-	DEFINE_WAIT(wait);
 	struct btrfs_fs_info *fs_info = cache->fs_info;
-	struct btrfs_caching_control *caching_ctl;
+	struct btrfs_caching_control *caching_ctl = NULL;
 	int ret = 0;
 
+	/* Allocator for zoned filesystems does not use the cache at all */
+	if (btrfs_is_zoned(fs_info))
+		return 0;
+
 	caching_ctl = kzalloc(sizeof(*caching_ctl), GFP_NOFS);
 	if (!caching_ctl)
 		return -ENOMEM;
@@ -692,120 +747,37 @@ int btrfs_cache_block_group(struct btrfs_block_group *cache, int load_cache_only
 	mutex_init(&caching_ctl->mutex);
 	init_waitqueue_head(&caching_ctl->wait);
 	caching_ctl->block_group = cache;
-	caching_ctl->progress = cache->start;
-	refcount_set(&caching_ctl->count, 1);
+	refcount_set(&caching_ctl->count, 2);
 	btrfs_init_work(&caching_ctl->work, caching_thread, NULL, NULL);
 
 	spin_lock(&cache->lock);
-	/*
-	 * This should be a rare occasion, but this could happen I think in the
-	 * case where one thread starts to load the space cache info, and then
-	 * some other thread starts a transaction commit which tries to do an
-	 * allocation while the other thread is still loading the space cache
-	 * info.  The previous loop should have kept us from choosing this block
-	 * group, but if we've moved to the state where we will wait on caching
-	 * block groups we need to first check if we're doing a fast load here,
-	 * so we can wait for it to finish, otherwise we could end up allocating
-	 * from a block group who's cache gets evicted for one reason or
-	 * another.
-	 */
-	while (cache->cached == BTRFS_CACHE_FAST) {
-		struct btrfs_caching_control *ctl;
-
-		ctl = cache->caching_ctl;
-		refcount_inc(&ctl->count);
-		prepare_to_wait(&ctl->wait, &wait, TASK_UNINTERRUPTIBLE);
-		spin_unlock(&cache->lock);
-
-		schedule();
-
-		finish_wait(&ctl->wait, &wait);
-		btrfs_put_caching_control(ctl);
-		spin_lock(&cache->lock);
-	}
-
 	if (cache->cached != BTRFS_CACHE_NO) {
-		spin_unlock(&cache->lock);
 		kfree(caching_ctl);
-		return 0;
+
+		caching_ctl = cache->caching_ctl;
+		if (caching_ctl)
+			refcount_inc(&caching_ctl->count);
+		spin_unlock(&cache->lock);
+		goto out;
 	}
 	WARN_ON(cache->caching_ctl);
 	cache->caching_ctl = caching_ctl;
-	cache->cached = BTRFS_CACHE_FAST;
+	cache->cached = BTRFS_CACHE_STARTED;
 	spin_unlock(&cache->lock);
 
-	if (btrfs_test_opt(fs_info, SPACE_CACHE)) {
-		mutex_lock(&caching_ctl->mutex);
-		ret = load_free_space_cache(cache);
-
-		spin_lock(&cache->lock);
-		if (ret == 1) {
-			cache->caching_ctl = NULL;
-			cache->cached = BTRFS_CACHE_FINISHED;
-			cache->last_byte_to_unpin = (u64)-1;
-			caching_ctl->progress = (u64)-1;
-		} else {
-			if (load_cache_only) {
-				cache->caching_ctl = NULL;
-				cache->cached = BTRFS_CACHE_NO;
-			} else {
-				cache->cached = BTRFS_CACHE_STARTED;
-				cache->has_caching_ctl = 1;
-			}
-		}
-		spin_unlock(&cache->lock);
-#ifdef CONFIG_BTRFS_DEBUG
-		if (ret == 1 &&
-		    btrfs_should_fragment_free_space(cache)) {
-			u64 bytes_used;
-
-			spin_lock(&cache->space_info->lock);
-			spin_lock(&cache->lock);
-			bytes_used = cache->length - cache->used;
-			cache->space_info->bytes_used += bytes_used >> 1;
-			spin_unlock(&cache->lock);
-			spin_unlock(&cache->space_info->lock);
-			fragment_free_space(cache);
-		}
-#endif
-		mutex_unlock(&caching_ctl->mutex);
-
-		wake_up(&caching_ctl->wait);
-		if (ret == 1) {
-			btrfs_put_caching_control(caching_ctl);
-			btrfs_free_excluded_extents(cache);
-			return 0;
-		}
-	} else {
-		/*
-		 * We're either using the free space tree or no caching at all.
-		 * Set cached to the appropriate value and wakeup any waiters.
-		 */
-		spin_lock(&cache->lock);
-		if (load_cache_only) {
-			cache->caching_ctl = NULL;
-			cache->cached = BTRFS_CACHE_NO;
-		} else {
-			cache->cached = BTRFS_CACHE_STARTED;
-			cache->has_caching_ctl = 1;
-		}
-		spin_unlock(&cache->lock);
-		wake_up(&caching_ctl->wait);
-	}
-
-	if (load_cache_only) {
-		btrfs_put_caching_control(caching_ctl);
-		return 0;
-	}
-
-	down_write(&fs_info->commit_root_sem);
+	write_lock(&fs_info->block_group_cache_lock);
 	refcount_inc(&caching_ctl->count);
 	list_add_tail(&caching_ctl->list, &fs_info->caching_block_groups);
-	up_write(&fs_info->commit_root_sem);
+	write_unlock(&fs_info->block_group_cache_lock);
 
 	btrfs_get_block_group(cache);
 
 	btrfs_queue_work(fs_info->caching_workers, &caching_ctl->work);
+out:
+	if (wait && caching_ctl)
+		ret = btrfs_caching_ctl_wait_done(cache, caching_ctl);
+	if (caching_ctl)
+		btrfs_put_caching_control(caching_ctl);
 
 	return ret;
 }
@@ -863,16 +835,37 @@ static void clear_incompat_bg_bits(struct btrfs_fs_info *fs_info, u64 flags)
 	}
 }
 
+static int remove_block_group_item(struct btrfs_trans_handle *trans,
+				   struct btrfs_path *path,
+				   struct btrfs_block_group *block_group)
+{
+	struct btrfs_fs_info *fs_info = trans->fs_info;
+	struct btrfs_root *root;
+	struct btrfs_key key;
+	int ret;
+
+	root = btrfs_block_group_root(fs_info);
+	key.objectid = block_group->start;
+	key.type = BTRFS_BLOCK_GROUP_ITEM_KEY;
+	key.offset = block_group->length;
+
+	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
+	if (ret > 0)
+		ret = -ENOENT;
+	if (ret < 0)
+		return ret;
+
+	ret = btrfs_del_item(trans, root, path);
+	return ret;
+}
+
 int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
 			     u64 group_start, struct extent_map *em)
 {
 	struct btrfs_fs_info *fs_info = trans->fs_info;
-	struct btrfs_root *root = fs_info->extent_root;
 	struct btrfs_path *path;
 	struct btrfs_block_group *block_group;
 	struct btrfs_free_cluster *cluster;
-	struct btrfs_root *tree_root = fs_info->tree_root;
-	struct btrfs_key key;
 	struct inode *inode;
 	struct kobject *kobj = NULL;
 	int ret;
@@ -913,6 +906,9 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
 	btrfs_return_cluster_to_free_space(block_group, cluster);
 	spin_unlock(&cluster->refill_lock);
 
+	btrfs_clear_treelog_bg(block_group);
+	btrfs_clear_data_reloc_bg(block_group);
+
 	path = btrfs_alloc_path();
 	if (!path) {
 		ret = -ENOMEM;
@@ -950,51 +946,19 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
 	spin_unlock(&trans->transaction->dirty_bgs_lock);
 	mutex_unlock(&trans->transaction->cache_write_mutex);
 
-	if (!IS_ERR(inode)) {
-		ret = btrfs_orphan_add(trans, BTRFS_I(inode));
-		if (ret) {
-			btrfs_add_delayed_iput(inode);
-			goto out;
-		}
-		clear_nlink(inode);
-		/* One for the block groups ref */
-		spin_lock(&block_group->lock);
-		if (block_group->iref) {
-			block_group->iref = 0;
-			block_group->inode = NULL;
-			spin_unlock(&block_group->lock);
-			iput(inode);
-		} else {
-			spin_unlock(&block_group->lock);
-		}
-		/* One for our lookup ref */
-		btrfs_add_delayed_iput(inode);
-	}
-
-	key.objectid = BTRFS_FREE_SPACE_OBJECTID;
-	key.type = 0;
-	key.offset = block_group->start;
-
-	ret = btrfs_search_slot(trans, tree_root, &key, path, -1, 1);
-	if (ret < 0)
+	ret = btrfs_remove_free_space_inode(trans, inode, block_group);
+	if (ret)
 		goto out;
-	if (ret > 0)
-		btrfs_release_path(path);
-	if (ret == 0) {
-		ret = btrfs_del_item(trans, tree_root, path);
-		if (ret)
-			goto out;
-		btrfs_release_path(path);
-	}
 
-	spin_lock(&fs_info->block_group_cache_lock);
-	rb_erase(&block_group->cache_node,
-		 &fs_info->block_group_cache_tree);
+	write_lock(&fs_info->block_group_cache_lock);
+	rb_erase_cached(&block_group->cache_node,
+			&fs_info->block_group_cache_tree);
 	RB_CLEAR_NODE(&block_group->cache_node);
 
-	if (fs_info->first_logical_byte == block_group->start)
-		fs_info->first_logical_byte = (u64)-1;
-	spin_unlock(&fs_info->block_group_cache_lock);
+	/* Once for the block groups rbtree */
+	btrfs_put_block_group(block_group);
+
+	write_unlock(&fs_info->block_group_cache_lock);
 
 	down_write(&block_group->space_info->groups_sem);
 	/*
@@ -1014,32 +978,31 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
 		kobject_put(kobj);
 	}
 
-	if (block_group->has_caching_ctl)
-		caching_ctl = btrfs_get_caching_control(block_group);
 	if (block_group->cached == BTRFS_CACHE_STARTED)
 		btrfs_wait_block_group_cache_done(block_group);
-	if (block_group->has_caching_ctl) {
-		down_write(&fs_info->commit_root_sem);
-		if (!caching_ctl) {
-			struct btrfs_caching_control *ctl;
-
-			list_for_each_entry(ctl,
-				    &fs_info->caching_block_groups, list)
-				if (ctl->block_group == block_group) {
-					caching_ctl = ctl;
-					refcount_inc(&caching_ctl->count);
-					break;
-				}
-		}
-		if (caching_ctl)
-			list_del_init(&caching_ctl->list);
-		up_write(&fs_info->commit_root_sem);
-		if (caching_ctl) {
-			/* Once for the caching bgs list and once for us. */
-			btrfs_put_caching_control(caching_ctl);
-			btrfs_put_caching_control(caching_ctl);
+
+	write_lock(&fs_info->block_group_cache_lock);
+	caching_ctl = btrfs_get_caching_control(block_group);
+	if (!caching_ctl) {
+		struct btrfs_caching_control *ctl;
+
+		list_for_each_entry(ctl, &fs_info->caching_block_groups, list) {
+			if (ctl->block_group == block_group) {
+				caching_ctl = ctl;
+				refcount_inc(&caching_ctl->count);
+				break;
+			}
 		}
 	}
+	if (caching_ctl)
+		list_del_init(&caching_ctl->list);
+	write_unlock(&fs_info->block_group_cache_lock);
+
+	if (caching_ctl) {
+		/* Once for the caching bgs list and once for us. */
+		btrfs_put_caching_control(caching_ctl);
+		btrfs_put_caching_control(caching_ctl);
+	}
 
 	spin_lock(&trans->transaction->dirty_bgs_lock);
 	WARN_ON(!list_empty(&block_group->dirty_list));
@@ -1055,36 +1018,65 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
 		WARN_ON(block_group->space_info->total_bytes
 			< block_group->length);
 		WARN_ON(block_group->space_info->bytes_readonly
-			< block_group->length);
+			< block_group->length - block_group->zone_unusable);
+		WARN_ON(block_group->space_info->bytes_zone_unusable
+			< block_group->zone_unusable);
 		WARN_ON(block_group->space_info->disk_total
 			< block_group->length * factor);
+		WARN_ON(test_bit(BLOCK_GROUP_FLAG_ZONE_IS_ACTIVE,
+				 &block_group->runtime_flags) &&
+			block_group->space_info->active_total_bytes
+			< block_group->length);
 	}
 	block_group->space_info->total_bytes -= block_group->length;
-	block_group->space_info->bytes_readonly -= block_group->length;
+	if (test_bit(BLOCK_GROUP_FLAG_ZONE_IS_ACTIVE, &block_group->runtime_flags))
+		block_group->space_info->active_total_bytes -= block_group->length;
+	block_group->space_info->bytes_readonly -=
+		(block_group->length - block_group->zone_unusable);
+	block_group->space_info->bytes_zone_unusable -=
+		block_group->zone_unusable;
 	block_group->space_info->disk_total -= block_group->length * factor;
 
 	spin_unlock(&block_group->space_info->lock);
 
-	key.objectid = block_group->start;
-	key.type = BTRFS_BLOCK_GROUP_ITEM_KEY;
-	key.offset = block_group->length;
+	/*
+	 * Remove the free space for the block group from the free space tree
+	 * and the block group's item from the extent tree before marking the
+	 * block group as removed. This is to prevent races with tasks that
+	 * freeze and unfreeze a block group, this task and another task
+	 * allocating a new block group - the unfreeze task ends up removing
+	 * the block group's extent map before the task calling this function
+	 * deletes the block group item from the extent tree, allowing for
+	 * another task to attempt to create another block group with the same
+	 * item key (and failing with -EEXIST and a transaction abort).
+	 */
+	ret = remove_block_group_free_space(trans, block_group);
+	if (ret)
+		goto out;
+
+	ret = remove_block_group_item(trans, path, block_group);
+	if (ret < 0)
+		goto out;
 
-	mutex_lock(&fs_info->chunk_mutex);
 	spin_lock(&block_group->lock);
-	block_group->removed = 1;
+	set_bit(BLOCK_GROUP_FLAG_REMOVED, &block_group->runtime_flags);
+
 	/*
-	 * At this point trimming can't start on this block group, because we
-	 * removed the block group from the tree fs_info->block_group_cache_tree
-	 * so no one can't find it anymore and even if someone already got this
-	 * block group before we removed it from the rbtree, they have already
-	 * incremented block_group->trimming - if they didn't, they won't find
-	 * any free space entries because we already removed them all when we
-	 * called btrfs_remove_free_space_cache().
+	 * At this point trimming or scrub can't start on this block group,
+	 * because we removed the block group from the rbtree
+	 * fs_info->block_group_cache_tree so no one can't find it anymore and
+	 * even if someone already got this block group before we removed it
+	 * from the rbtree, they have already incremented block_group->frozen -
+	 * if they didn't, for the trimming case they won't find any free space
+	 * entries because we already removed them all when we called
+	 * btrfs_remove_free_space_cache().
 	 *
 	 * And we must not remove the extent map from the fs_info->mapping_tree
 	 * to prevent the same logical address range and physical device space
-	 * ranges from being reused for a new block group. This is because our
-	 * fs trim operation (btrfs_trim_fs() / btrfs_ioctl_fitrim()) is
+	 * ranges from being reused for a new block group. This is needed to
+	 * avoid races with trimming and scrub.
+	 *
+	 * An fs trim operation (btrfs_trim_fs() / btrfs_ioctl_fitrim()) is
 	 * completely transactionless, so while it is trimming a range the
 	 * currently running transaction might finish and a new one start,
 	 * allowing for new block groups to be created that can reuse the same
@@ -1095,28 +1087,9 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
 	 * in place until the extents have been discarded completely when
 	 * the transaction commit has completed.
 	 */
-	remove_em = (atomic_read(&block_group->trimming) == 0);
+	remove_em = (atomic_read(&block_group->frozen) == 0);
 	spin_unlock(&block_group->lock);
 
-	mutex_unlock(&fs_info->chunk_mutex);
-
-	ret = remove_block_group_free_space(trans, block_group);
-	if (ret)
-		goto out;
-
-	btrfs_put_block_group(block_group);
-	btrfs_put_block_group(block_group);
-
-	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
-	if (ret > 0)
-		ret = -EIO;
-	if (ret < 0)
-		goto out;
-
-	ret = btrfs_del_item(trans, root, path);
-	if (ret)
-		goto out;
-
 	if (remove_em) {
 		struct extent_map_tree *em_tree;
 
@@ -1127,7 +1100,10 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
 		/* once for the tree */
 		free_extent_map(em);
 	}
+
 out:
+	/* Once for the lookup reference */
+	btrfs_put_block_group(block_group);
 	if (remove_rsv)
 		btrfs_delayed_refs_rsv_release(fs_info, 1);
 	btrfs_free_path(path);
@@ -1137,6 +1113,7 @@ out:
 struct btrfs_trans_handle *btrfs_start_trans_remove_block_group(
 		struct btrfs_fs_info *fs_info, const u64 chunk_offset)
 {
+	struct btrfs_root *root = btrfs_block_group_root(fs_info);
 	struct extent_map_tree *em_tree = &fs_info->mapping_tree;
 	struct extent_map *em;
 	struct map_lookup *map;
@@ -1170,8 +1147,7 @@ struct btrfs_trans_handle *btrfs_start_trans_remove_block_group(
 	num_items = 3 + map->num_stripes;
 	free_extent_map(em);
 
-	return btrfs_start_transaction_fallback_global_rsv(fs_info->extent_root,
-							   num_items, 1);
+	return btrfs_start_transaction_fallback_global_rsv(root, num_items);
 }
 
 /*
@@ -1196,6 +1172,11 @@ static int inc_block_group_ro(struct btrfs_block_group *cache, int force)
 	spin_lock(&sinfo->lock);
 	spin_lock(&cache->lock);
 
+	if (cache->swap_extents) {
+		ret = -ETXTBSY;
+		goto out;
+	}
+
 	if (cache->ro) {
 		cache->ro++;
 		ret = 0;
@@ -1203,7 +1184,7 @@ static int inc_block_group_ro(struct btrfs_block_group *cache, int force)
 	}
 
 	num_bytes = cache->length - cache->reserved - cache->pinned -
-		    cache->bytes_super - cache->used;
+		    cache->bytes_super - cache->zone_unusable - cache->used;
 
 	/*
 	 * Data never overcommits, even in mixed mode, so do just the straight
@@ -1234,6 +1215,12 @@ static int inc_block_group_ro(struct btrfs_block_group *cache, int force)
 
 	if (!ret) {
 		sinfo->bytes_readonly += num_bytes;
+		if (btrfs_is_zoned(cache->fs_info)) {
+			/* Migrate zone_unusable bytes to readonly */
+			sinfo->bytes_readonly += cache->zone_unusable;
+			sinfo->bytes_zone_unusable -= cache->zone_unusable;
+			cache->zone_unusable = 0;
+		}
 		cache->ro++;
 		list_add_tail(&cache->ro_list, &sinfo->ro_bgs);
 	}
@@ -1248,6 +1235,51 @@ out:
 	return ret;
 }
 
+static bool clean_pinned_extents(struct btrfs_trans_handle *trans,
+				 struct btrfs_block_group *bg)
+{
+	struct btrfs_fs_info *fs_info = bg->fs_info;
+	struct btrfs_transaction *prev_trans = NULL;
+	const u64 start = bg->start;
+	const u64 end = start + bg->length - 1;
+	int ret;
+
+	spin_lock(&fs_info->trans_lock);
+	if (trans->transaction->list.prev != &fs_info->trans_list) {
+		prev_trans = list_last_entry(&trans->transaction->list,
+					     struct btrfs_transaction, list);
+		refcount_inc(&prev_trans->use_count);
+	}
+	spin_unlock(&fs_info->trans_lock);
+
+	/*
+	 * Hold the unused_bg_unpin_mutex lock to avoid racing with
+	 * btrfs_finish_extent_commit(). If we are at transaction N, another
+	 * task might be running finish_extent_commit() for the previous
+	 * transaction N - 1, and have seen a range belonging to the block
+	 * group in pinned_extents before we were able to clear the whole block
+	 * group range from pinned_extents. This means that task can lookup for
+	 * the block group after we unpinned it from pinned_extents and removed
+	 * it, leading to a BUG_ON() at unpin_extent_range().
+	 */
+	mutex_lock(&fs_info->unused_bg_unpin_mutex);
+	if (prev_trans) {
+		ret = clear_extent_bits(&prev_trans->pinned_extents, start, end,
+					EXTENT_DIRTY);
+		if (ret)
+			goto out;
+	}
+
+	ret = clear_extent_bits(&trans->transaction->pinned_extents, start, end,
+				EXTENT_DIRTY);
+out:
+	mutex_unlock(&fs_info->unused_bg_unpin_mutex);
+	if (prev_trans)
+		btrfs_put_transaction(prev_trans);
+
+	return ret == 0;
+}
+
 /*
  * Process the unused_bgs list and remove any that don't have any allocated
  * space inside of them.
@@ -1263,9 +1295,18 @@ void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
 	if (!test_bit(BTRFS_FS_OPEN, &fs_info->flags))
 		return;
 
+	if (btrfs_fs_closing(fs_info))
+		return;
+
+	/*
+	 * Long running balances can keep us blocked here for eternity, so
+	 * simply skip deletion if we're unable to get the mutex.
+	 */
+	if (!mutex_trylock(&fs_info->reclaim_bgs_lock))
+		return;
+
 	spin_lock(&fs_info->unused_bgs_lock);
 	while (!list_empty(&fs_info->unused_bgs)) {
-		u64 start, end;
 		int trimming;
 
 		block_group = list_first_entry(&fs_info->unused_bgs,
@@ -1283,8 +1324,6 @@ void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
 
 		btrfs_discard_cancel_work(&fs_info->discard_ctl, block_group);
 
-		mutex_lock(&fs_info->delete_unused_bgs_mutex);
-
 		/* Don't want to race with allocators so take the groups_sem */
 		down_write(&space_info->groups_sem);
 
@@ -1328,6 +1367,14 @@ void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
 			goto next;
 		}
 
+		ret = btrfs_zone_finish(block_group);
+		if (ret < 0) {
+			btrfs_dec_block_group_ro(block_group);
+			if (ret == -EAGAIN)
+				ret = 0;
+			goto next;
+		}
+
 		/*
 		 * Want to do this before we do anything else so we can recover
 		 * properly if we fail to join the transaction.
@@ -1344,35 +1391,10 @@ void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
 		 * We could have pending pinned extents for this block group,
 		 * just delete them, we don't care about them anymore.
 		 */
-		start = block_group->start;
-		end = start + block_group->length - 1;
-		/*
-		 * Hold the unused_bg_unpin_mutex lock to avoid racing with
-		 * btrfs_finish_extent_commit(). If we are at transaction N,
-		 * another task might be running finish_extent_commit() for the
-		 * previous transaction N - 1, and have seen a range belonging
-		 * to the block group in freed_extents[] before we were able to
-		 * clear the whole block group range from freed_extents[]. This
-		 * means that task can lookup for the block group after we
-		 * unpinned it from freed_extents[] and removed it, leading to
-		 * a BUG_ON() at btrfs_unpin_extent_range().
-		 */
-		mutex_lock(&fs_info->unused_bg_unpin_mutex);
-		ret = clear_extent_bits(&fs_info->freed_extents[0], start, end,
-				  EXTENT_DIRTY);
-		if (ret) {
-			mutex_unlock(&fs_info->unused_bg_unpin_mutex);
+		if (!clean_pinned_extents(trans, block_group)) {
 			btrfs_dec_block_group_ro(block_group);
 			goto end_trans;
 		}
-		ret = clear_extent_bits(&fs_info->freed_extents[1], start, end,
-				  EXTENT_DIRTY);
-		if (ret) {
-			mutex_unlock(&fs_info->unused_bg_unpin_mutex);
-			btrfs_dec_block_group_ro(block_group);
-			goto end_trans;
-		}
-		mutex_unlock(&fs_info->unused_bg_unpin_mutex);
 
 		/*
 		 * At this point, the block_group is read only and should fail
@@ -1398,9 +1420,6 @@ void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
 		btrfs_space_info_update_bytes_pinned(fs_info, space_info,
 						     -block_group->pinned);
 		space_info->bytes_readonly += block_group->pinned;
-		percpu_counter_add_batch(&space_info->total_bytes_pinned,
-				   -block_group->pinned,
-				   BTRFS_TOTAL_BYTES_PINNED_BATCH);
 		block_group->pinned = 0;
 
 		spin_unlock(&block_group->lock);
@@ -1416,12 +1435,16 @@ void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
 		if (!async_trim_enabled && btrfs_test_opt(fs_info, DISCARD_ASYNC))
 			goto flip_async;
 
-		/* DISCARD can flip during remount */
-		trimming = btrfs_test_opt(fs_info, DISCARD_SYNC);
+		/*
+		 * DISCARD can flip during remount. On zoned filesystems, we
+		 * need to reset sequential-required zones.
+		 */
+		trimming = btrfs_test_opt(fs_info, DISCARD_SYNC) ||
+				btrfs_is_zoned(fs_info);
 
 		/* Implicit trim during transaction commit. */
 		if (trimming)
-			btrfs_get_block_group_trimming(block_group);
+			btrfs_freeze_block_group(block_group);
 
 		/*
 		 * Btrfs_remove_chunk will abort the transaction if things go
@@ -1431,7 +1454,7 @@ void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
 
 		if (ret) {
 			if (trimming)
-				btrfs_put_block_group_trimming(block_group);
+				btrfs_unfreeze_block_group(block_group);
 			goto end_trans;
 		}
 
@@ -1455,16 +1478,16 @@ void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
 end_trans:
 		btrfs_end_transaction(trans);
 next:
-		mutex_unlock(&fs_info->delete_unused_bgs_mutex);
 		btrfs_put_block_group(block_group);
 		spin_lock(&fs_info->unused_bgs_lock);
 	}
 	spin_unlock(&fs_info->unused_bgs_lock);
+	mutex_unlock(&fs_info->reclaim_bgs_lock);
 	return;
 
 flip_async:
 	btrfs_end_transaction(trans);
-	mutex_unlock(&fs_info->delete_unused_bgs_mutex);
+	mutex_unlock(&fs_info->reclaim_bgs_lock);
 	btrfs_put_block_group(block_group);
 	btrfs_discard_punt_unused_bgs_list(fs_info);
 }
@@ -1482,83 +1505,223 @@ void btrfs_mark_bg_unused(struct btrfs_block_group *bg)
 	spin_unlock(&fs_info->unused_bgs_lock);
 }
 
-static int find_first_block_group(struct btrfs_fs_info *fs_info,
-				  struct btrfs_path *path,
-				  struct btrfs_key *key)
+/*
+ * We want block groups with a low number of used bytes to be in the beginning
+ * of the list, so they will get reclaimed first.
+ */
+static int reclaim_bgs_cmp(void *unused, const struct list_head *a,
+			   const struct list_head *b)
 {
-	struct btrfs_root *root = fs_info->extent_root;
-	int ret = 0;
-	struct btrfs_key found_key;
-	struct extent_buffer *leaf;
+	const struct btrfs_block_group *bg1, *bg2;
+
+	bg1 = list_entry(a, struct btrfs_block_group, bg_list);
+	bg2 = list_entry(b, struct btrfs_block_group, bg_list);
+
+	return bg1->used > bg2->used;
+}
+
+static inline bool btrfs_should_reclaim(struct btrfs_fs_info *fs_info)
+{
+	if (btrfs_is_zoned(fs_info))
+		return btrfs_zoned_should_reclaim(fs_info);
+	return true;
+}
+
+void btrfs_reclaim_bgs_work(struct work_struct *work)
+{
+	struct btrfs_fs_info *fs_info =
+		container_of(work, struct btrfs_fs_info, reclaim_bgs_work);
+	struct btrfs_block_group *bg;
+	struct btrfs_space_info *space_info;
+
+	if (!test_bit(BTRFS_FS_OPEN, &fs_info->flags))
+		return;
+
+	if (btrfs_fs_closing(fs_info))
+		return;
+
+	if (!btrfs_should_reclaim(fs_info))
+		return;
+
+	sb_start_write(fs_info->sb);
+
+	if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_BALANCE)) {
+		sb_end_write(fs_info->sb);
+		return;
+	}
+
+	/*
+	 * Long running balances can keep us blocked here for eternity, so
+	 * simply skip reclaim if we're unable to get the mutex.
+	 */
+	if (!mutex_trylock(&fs_info->reclaim_bgs_lock)) {
+		btrfs_exclop_finish(fs_info);
+		sb_end_write(fs_info->sb);
+		return;
+	}
+
+	spin_lock(&fs_info->unused_bgs_lock);
+	/*
+	 * Sort happens under lock because we can't simply splice it and sort.
+	 * The block groups might still be in use and reachable via bg_list,
+	 * and their presence in the reclaim_bgs list must be preserved.
+	 */
+	list_sort(NULL, &fs_info->reclaim_bgs, reclaim_bgs_cmp);
+	while (!list_empty(&fs_info->reclaim_bgs)) {
+		u64 zone_unusable;
+		int ret = 0;
+
+		bg = list_first_entry(&fs_info->reclaim_bgs,
+				      struct btrfs_block_group,
+				      bg_list);
+		list_del_init(&bg->bg_list);
+
+		space_info = bg->space_info;
+		spin_unlock(&fs_info->unused_bgs_lock);
+
+		/* Don't race with allocators so take the groups_sem */
+		down_write(&space_info->groups_sem);
+
+		spin_lock(&bg->lock);
+		if (bg->reserved || bg->pinned || bg->ro) {
+			/*
+			 * We want to bail if we made new allocations or have
+			 * outstanding allocations in this block group.  We do
+			 * the ro check in case balance is currently acting on
+			 * this block group.
+			 */
+			spin_unlock(&bg->lock);
+			up_write(&space_info->groups_sem);
+			goto next;
+		}
+		spin_unlock(&bg->lock);
+
+		/* Get out fast, in case we're unmounting the filesystem */
+		if (btrfs_fs_closing(fs_info)) {
+			up_write(&space_info->groups_sem);
+			goto next;
+		}
+
+		/*
+		 * Cache the zone_unusable value before turning the block group
+		 * to read only. As soon as the blog group is read only it's
+		 * zone_unusable value gets moved to the block group's read-only
+		 * bytes and isn't available for calculations anymore.
+		 */
+		zone_unusable = bg->zone_unusable;
+		ret = inc_block_group_ro(bg, 0);
+		up_write(&space_info->groups_sem);
+		if (ret < 0)
+			goto next;
+
+		btrfs_info(fs_info,
+			"reclaiming chunk %llu with %llu%% used %llu%% unusable",
+				bg->start, div_u64(bg->used * 100, bg->length),
+				div64_u64(zone_unusable * 100, bg->length));
+		trace_btrfs_reclaim_block_group(bg);
+		ret = btrfs_relocate_chunk(fs_info, bg->start);
+		if (ret) {
+			btrfs_dec_block_group_ro(bg);
+			btrfs_err(fs_info, "error relocating chunk %llu",
+				  bg->start);
+		}
+
+next:
+		btrfs_put_block_group(bg);
+		spin_lock(&fs_info->unused_bgs_lock);
+	}
+	spin_unlock(&fs_info->unused_bgs_lock);
+	mutex_unlock(&fs_info->reclaim_bgs_lock);
+	btrfs_exclop_finish(fs_info);
+	sb_end_write(fs_info->sb);
+}
+
+void btrfs_reclaim_bgs(struct btrfs_fs_info *fs_info)
+{
+	spin_lock(&fs_info->unused_bgs_lock);
+	if (!list_empty(&fs_info->reclaim_bgs))
+		queue_work(system_unbound_wq, &fs_info->reclaim_bgs_work);
+	spin_unlock(&fs_info->unused_bgs_lock);
+}
+
+void btrfs_mark_bg_to_reclaim(struct btrfs_block_group *bg)
+{
+	struct btrfs_fs_info *fs_info = bg->fs_info;
+
+	spin_lock(&fs_info->unused_bgs_lock);
+	if (list_empty(&bg->bg_list)) {
+		btrfs_get_block_group(bg);
+		trace_btrfs_add_reclaim_block_group(bg);
+		list_add_tail(&bg->bg_list, &fs_info->reclaim_bgs);
+	}
+	spin_unlock(&fs_info->unused_bgs_lock);
+}
+
+static int read_bg_from_eb(struct btrfs_fs_info *fs_info, struct btrfs_key *key,
+			   struct btrfs_path *path)
+{
+	struct extent_map_tree *em_tree;
+	struct extent_map *em;
 	struct btrfs_block_group_item bg;
-	u64 flags;
+	struct extent_buffer *leaf;
 	int slot;
+	u64 flags;
+	int ret = 0;
 
-	ret = btrfs_search_slot(NULL, root, key, path, 0, 0);
-	if (ret < 0)
-		goto out;
+	slot = path->slots[0];
+	leaf = path->nodes[0];
 
-	while (1) {
-		slot = path->slots[0];
-		leaf = path->nodes[0];
-		if (slot >= btrfs_header_nritems(leaf)) {
-			ret = btrfs_next_leaf(root, path);
-			if (ret == 0)
-				continue;
-			if (ret < 0)
-				goto out;
-			break;
-		}
-		btrfs_item_key_to_cpu(leaf, &found_key, slot);
+	em_tree = &fs_info->mapping_tree;
+	read_lock(&em_tree->lock);
+	em = lookup_extent_mapping(em_tree, key->objectid, key->offset);
+	read_unlock(&em_tree->lock);
+	if (!em) {
+		btrfs_err(fs_info,
+			  "logical %llu len %llu found bg but no related chunk",
+			  key->objectid, key->offset);
+		return -ENOENT;
+	}
+
+	if (em->start != key->objectid || em->len != key->offset) {
+		btrfs_err(fs_info,
+			"block group %llu len %llu mismatch with chunk %llu len %llu",
+			key->objectid, key->offset, em->start, em->len);
+		ret = -EUCLEAN;
+		goto out_free_em;
+	}
 
+	read_extent_buffer(leaf, &bg, btrfs_item_ptr_offset(leaf, slot),
+			   sizeof(bg));
+	flags = btrfs_stack_block_group_flags(&bg) &
+		BTRFS_BLOCK_GROUP_TYPE_MASK;
+
+	if (flags != (em->map_lookup->type & BTRFS_BLOCK_GROUP_TYPE_MASK)) {
+		btrfs_err(fs_info,
+"block group %llu len %llu type flags 0x%llx mismatch with chunk type flags 0x%llx",
+			  key->objectid, key->offset, flags,
+			  (BTRFS_BLOCK_GROUP_TYPE_MASK & em->map_lookup->type));
+		ret = -EUCLEAN;
+	}
+
+out_free_em:
+	free_extent_map(em);
+	return ret;
+}
+
+static int find_first_block_group(struct btrfs_fs_info *fs_info,
+				  struct btrfs_path *path,
+				  struct btrfs_key *key)
+{
+	struct btrfs_root *root = btrfs_block_group_root(fs_info);
+	int ret;
+	struct btrfs_key found_key;
+
+	btrfs_for_each_slot(root, key, &found_key, path, ret) {
 		if (found_key.objectid >= key->objectid &&
 		    found_key.type == BTRFS_BLOCK_GROUP_ITEM_KEY) {
-			struct extent_map_tree *em_tree;
-			struct extent_map *em;
-
-			em_tree = &root->fs_info->mapping_tree;
-			read_lock(&em_tree->lock);
-			em = lookup_extent_mapping(em_tree, found_key.objectid,
-						   found_key.offset);
-			read_unlock(&em_tree->lock);
-			if (!em) {
-				btrfs_err(fs_info,
-			"logical %llu len %llu found bg but no related chunk",
-					  found_key.objectid, found_key.offset);
-				ret = -ENOENT;
-			} else if (em->start != found_key.objectid ||
-				   em->len != found_key.offset) {
-				btrfs_err(fs_info,
-		"block group %llu len %llu mismatch with chunk %llu len %llu",
-					  found_key.objectid, found_key.offset,
-					  em->start, em->len);
-				ret = -EUCLEAN;
-			} else {
-				read_extent_buffer(leaf, &bg,
-					btrfs_item_ptr_offset(leaf, slot),
-					sizeof(bg));
-				flags = btrfs_stack_block_group_flags(&bg) &
-					BTRFS_BLOCK_GROUP_TYPE_MASK;
-
-				if (flags != (em->map_lookup->type &
-					      BTRFS_BLOCK_GROUP_TYPE_MASK)) {
-					btrfs_err(fs_info,
-"block group %llu len %llu type flags 0x%llx mismatch with chunk type flags 0x%llx",
-						found_key.objectid,
-						found_key.offset, flags,
-						(BTRFS_BLOCK_GROUP_TYPE_MASK &
-						 em->map_lookup->type));
-					ret = -EUCLEAN;
-				} else {
-					ret = 0;
-				}
-			}
-			free_extent_map(em);
-			goto out;
+			return read_bg_from_eb(fs_info, &found_key, path);
 		}
-		path->slots[0]++;
 	}
-out:
 	return ret;
 }
 
@@ -1578,8 +1741,11 @@ static void set_avail_alloc_bits(struct btrfs_fs_info *fs_info, u64 flags)
 }
 
 /**
- * btrfs_rmap_block - Map a physical disk address to a list of logical addresses
+ * Map a physical disk address to a list of logical addresses
+ *
+ * @fs_info:       the filesystem
  * @chunk_start:   logical address of block group
+ * @bdev:	   physical device to resolve, can be NULL to indicate any device
  * @physical:	   physical address to map to logical addresses
  * @logical:	   return array of logical addresses which map to @physical
  * @naddrs:	   length of @logical
@@ -1589,9 +1755,9 @@ static void set_avail_alloc_bits(struct btrfs_fs_info *fs_info, u64 flags)
  * Used primarily to exclude those portions of a block group that contain super
  * block copies.
  */
-EXPORT_FOR_TESTS
 int btrfs_rmap_block(struct btrfs_fs_info *fs_info, u64 chunk_start,
-		     u64 physical, u64 **logical, int *naddrs, int *stripe_len)
+		     struct block_device *bdev, u64 physical, u64 **logical,
+		     int *naddrs, int *stripe_len)
 {
 	struct extent_map *em;
 	struct map_lookup *map;
@@ -1607,19 +1773,13 @@ int btrfs_rmap_block(struct btrfs_fs_info *fs_info, u64 chunk_start,
 		return -EIO;
 
 	map = em->map_lookup;
-	data_stripe_length = em->len;
+	data_stripe_length = em->orig_block_len;
 	io_stripe_size = map->stripe_len;
+	chunk_start = em->start;
 
-	if (map->type & BTRFS_BLOCK_GROUP_RAID10)
-		data_stripe_length = div_u64(data_stripe_length,
-					     map->num_stripes / map->sub_stripes);
-	else if (map->type & BTRFS_BLOCK_GROUP_RAID0)
-		data_stripe_length = div_u64(data_stripe_length, map->num_stripes);
-	else if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) {
-		data_stripe_length = div_u64(data_stripe_length,
-					     nr_data_stripes(map));
+	/* For RAID5/6 adjust to a full IO stripe length */
+	if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK)
 		io_stripe_size = map->stripe_len * nr_data_stripes(map);
-	}
 
 	buf = kcalloc(map->num_stripes, sizeof(u64), GFP_NOFS);
 	if (!buf) {
@@ -1630,20 +1790,23 @@ int btrfs_rmap_block(struct btrfs_fs_info *fs_info, u64 chunk_start,
 	for (i = 0; i < map->num_stripes; i++) {
 		bool already_inserted = false;
 		u64 stripe_nr;
+		u64 offset;
 		int j;
 
 		if (!in_range(physical, map->stripes[i].physical,
 			      data_stripe_length))
 			continue;
 
+		if (bdev && map->stripes[i].dev->bdev != bdev)
+			continue;
+
 		stripe_nr = physical - map->stripes[i].physical;
-		stripe_nr = div64_u64(stripe_nr, map->stripe_len);
+		stripe_nr = div64_u64_rem(stripe_nr, map->stripe_len, &offset);
 
-		if (map->type & BTRFS_BLOCK_GROUP_RAID10) {
+		if (map->type & (BTRFS_BLOCK_GROUP_RAID0 |
+				 BTRFS_BLOCK_GROUP_RAID10)) {
 			stripe_nr = stripe_nr * map->num_stripes + i;
 			stripe_nr = div_u64(stripe_nr, map->sub_stripes);
-		} else if (map->type & BTRFS_BLOCK_GROUP_RAID0) {
-			stripe_nr = stripe_nr * map->num_stripes + i;
 		}
 		/*
 		 * The remaining case would be for RAID56, multiply by
@@ -1651,7 +1814,7 @@ int btrfs_rmap_block(struct btrfs_fs_info *fs_info, u64 chunk_start,
 		 * instead of map->stripe_len
 		 */
 
-		bytenr = chunk_start + stripe_nr * io_stripe_size;
+		bytenr = chunk_start + stripe_nr * io_stripe_size + offset;
 
 		/* Ensure we don't add duplicate addresses */
 		for (j = 0; j < nr; j++) {
@@ -1676,6 +1839,7 @@ out:
 static int exclude_super_stripes(struct btrfs_block_group *cache)
 {
 	struct btrfs_fs_info *fs_info = cache->fs_info;
+	const bool zoned = btrfs_is_zoned(fs_info);
 	u64 bytenr;
 	u64 *logical;
 	int stripe_len;
@@ -1692,31 +1856,26 @@ static int exclude_super_stripes(struct btrfs_block_group *cache)
 
 	for (i = 0; i < BTRFS_SUPER_MIRROR_MAX; i++) {
 		bytenr = btrfs_sb_offset(i);
-		ret = btrfs_rmap_block(fs_info, cache->start,
+		ret = btrfs_rmap_block(fs_info, cache->start, NULL,
 				       bytenr, &logical, &nr, &stripe_len);
 		if (ret)
 			return ret;
 
-		while (nr--) {
-			u64 start, len;
-
-			if (logical[nr] > cache->start + cache->length)
-				continue;
-
-			if (logical[nr] + stripe_len <= cache->start)
-				continue;
+		/* Shouldn't have super stripes in sequential zones */
+		if (zoned && nr) {
+			btrfs_err(fs_info,
+			"zoned: block group %llu must not contain super block",
+				  cache->start);
+			return -EUCLEAN;
+		}
 
-			start = logical[nr];
-			if (start < cache->start) {
-				start = cache->start;
-				len = (logical[nr] + stripe_len) - start;
-			} else {
-				len = min_t(u64, stripe_len,
-					    cache->start + cache->length - start);
-			}
+		while (nr--) {
+			u64 len = min_t(u64, stripe_len,
+				cache->start + cache->length - logical[nr]);
 
 			cache->bytes_super += len;
-			ret = btrfs_add_excluded_extent(fs_info, start, len);
+			ret = btrfs_add_excluded_extent(fs_info, logical[nr],
+							len);
 			if (ret) {
 				kfree(logical);
 				return ret;
@@ -1728,24 +1887,8 @@ static int exclude_super_stripes(struct btrfs_block_group *cache)
 	return 0;
 }
 
-static void link_block_group(struct btrfs_block_group *cache)
-{
-	struct btrfs_space_info *space_info = cache->space_info;
-	int index = btrfs_bg_flags_to_raid_index(cache->flags);
-	bool first = false;
-
-	down_write(&space_info->groups_sem);
-	if (list_empty(&space_info->block_groups[index]))
-		first = true;
-	list_add_tail(&cache->list, &space_info->block_groups[index]);
-	up_write(&space_info->groups_sem);
-
-	if (first)
-		btrfs_sysfs_add_block_group_type(cache);
-}
-
 static struct btrfs_block_group *btrfs_create_block_group_cache(
-		struct btrfs_fs_info *fs_info, u64 start, u64 size)
+		struct btrfs_fs_info *fs_info, u64 start)
 {
 	struct btrfs_block_group *cache;
 
@@ -1761,15 +1904,13 @@ static struct btrfs_block_group *btrfs_create_block_group_cache(
 	}
 
 	cache->start = start;
-	cache->length = size;
 
 	cache->fs_info = fs_info;
 	cache->full_stripe_len = btrfs_full_stripe_len(fs_info, start);
-	set_free_space_tree_thresholds(cache);
 
 	cache->discard_index = BTRFS_DISCARD_INDEX_UNUSED;
 
-	atomic_set(&cache->count, 1);
+	refcount_set(&cache->refs, 1);
 	spin_lock_init(&cache->lock);
 	init_rwsem(&cache->data_rwsem);
 	INIT_LIST_HEAD(&cache->list);
@@ -1779,10 +1920,12 @@ static struct btrfs_block_group *btrfs_create_block_group_cache(
 	INIT_LIST_HEAD(&cache->discard_list);
 	INIT_LIST_HEAD(&cache->dirty_list);
 	INIT_LIST_HEAD(&cache->io_list);
-	btrfs_init_free_space_ctl(cache);
-	atomic_set(&cache->trimming, 0);
+	INIT_LIST_HEAD(&cache->active_bg_list);
+	btrfs_init_free_space_ctl(cache, cache->free_space_ctl);
+	atomic_set(&cache->frozen, 0);
 	mutex_init(&cache->free_space_lock);
-	btrfs_init_full_stripe_locks_tree(&cache->full_stripe_locks_root);
+	cache->full_stripe_locks_root.root = RB_ROOT;
+	mutex_init(&cache->full_stripe_locks_root.lock);
 
 	return cache;
 }
@@ -1842,24 +1985,27 @@ static int check_chunk_block_group_mappings(struct btrfs_fs_info *fs_info)
 }
 
 static int read_one_block_group(struct btrfs_fs_info *info,
-				struct btrfs_path *path,
+				struct btrfs_block_group_item *bgi,
 				const struct btrfs_key *key,
 				int need_clear)
 {
-	struct extent_buffer *leaf = path->nodes[0];
 	struct btrfs_block_group *cache;
-	struct btrfs_space_info *space_info;
-	struct btrfs_block_group_item bgi;
 	const bool mixed = btrfs_fs_incompat(info, MIXED_GROUPS);
-	int slot = path->slots[0];
 	int ret;
 
 	ASSERT(key->type == BTRFS_BLOCK_GROUP_ITEM_KEY);
 
-	cache = btrfs_create_block_group_cache(info, key->objectid, key->offset);
+	cache = btrfs_create_block_group_cache(info, key->objectid);
 	if (!cache)
 		return -ENOMEM;
 
+	cache->length = key->offset;
+	cache->used = btrfs_stack_block_group_used(bgi);
+	cache->flags = btrfs_stack_block_group_flags(bgi);
+	cache->global_root_id = btrfs_stack_block_group_chunk_objectid(bgi);
+
+	set_free_space_tree_thresholds(cache);
+
 	if (need_clear) {
 		/*
 		 * When we mount with old space cache, we need to
@@ -1874,10 +2020,6 @@ static int read_one_block_group(struct btrfs_fs_info *info,
 		if (btrfs_test_opt(info, SPACE_CACHE))
 			cache->disk_cache_state = BTRFS_DC_CLEAR;
 	}
-	read_extent_buffer(leaf, &bgi, btrfs_item_ptr_offset(leaf, slot),
-			   sizeof(bgi));
-	cache->used = btrfs_stack_block_group_used(&bgi);
-	cache->flags = btrfs_stack_block_group_flags(&bgi);
 	if (!mixed && ((cache->flags & BTRFS_BLOCK_GROUP_METADATA) &&
 	    (cache->flags & BTRFS_BLOCK_GROUP_DATA))) {
 			btrfs_err(info,
@@ -1887,6 +2029,13 @@ static int read_one_block_group(struct btrfs_fs_info *info,
 			goto error;
 	}
 
+	ret = btrfs_load_block_group_zone_info(cache, false);
+	if (ret) {
+		btrfs_err(info, "zoned: failed to load zone info of bg %llu",
+			  cache->start);
+		goto error;
+	}
+
 	/*
 	 * We need to exclude the super stripes now so that the space info has
 	 * super bytes accounted for, otherwise we'll think we have more space
@@ -1900,20 +2049,28 @@ static int read_one_block_group(struct btrfs_fs_info *info,
 	}
 
 	/*
-	 * Check for two cases, either we are full, and therefore don't need
-	 * to bother with the caching work since we won't find any space, or we
-	 * are empty, and we can just add all the space in and be done with it.
-	 * This saves us _a_lot_ of time, particularly in the full case.
+	 * For zoned filesystem, space after the allocation offset is the only
+	 * free space for a block group. So, we don't need any caching work.
+	 * btrfs_calc_zone_unusable() will set the amount of free space and
+	 * zone_unusable space.
+	 *
+	 * For regular filesystem, check for two cases, either we are full, and
+	 * therefore don't need to bother with the caching work since we won't
+	 * find any space, or we are empty, and we can just add all the space
+	 * in and be done with it.  This saves us _a_lot_ of time, particularly
+	 * in the full case.
 	 */
-	if (key->offset == cache->used) {
-		cache->last_byte_to_unpin = (u64)-1;
+	if (btrfs_is_zoned(info)) {
+		btrfs_calc_zone_unusable(cache);
+		/* Should not have any excluded extents. Just in case, though. */
+		btrfs_free_excluded_extents(cache);
+	} else if (cache->length == cache->used) {
 		cache->cached = BTRFS_CACHE_FINISHED;
 		btrfs_free_excluded_extents(cache);
 	} else if (cache->used == 0) {
-		cache->last_byte_to_unpin = (u64)-1;
 		cache->cached = BTRFS_CACHE_FINISHED;
-		add_new_free_space(cache, key->objectid,
-				   key->objectid + key->offset);
+		add_new_free_space(cache, cache->start,
+				   cache->start + cache->length);
 		btrfs_free_excluded_extents(cache);
 	}
 
@@ -1923,31 +2080,81 @@ static int read_one_block_group(struct btrfs_fs_info *info,
 		goto error;
 	}
 	trace_btrfs_add_block_group(info, cache, 0);
-	btrfs_update_space_info(info, cache->flags, key->offset,
-				cache->used, cache->bytes_super, &space_info);
-
-	cache->space_info = space_info;
-
-	link_block_group(cache);
+	btrfs_add_bg_to_space_info(info, cache);
 
 	set_avail_alloc_bits(info, cache->flags);
-	if (btrfs_chunk_readonly(info, cache->start)) {
+	if (btrfs_chunk_writeable(info, cache->start)) {
+		if (cache->used == 0) {
+			ASSERT(list_empty(&cache->bg_list));
+			if (btrfs_test_opt(info, DISCARD_ASYNC))
+				btrfs_discard_queue_work(&info->discard_ctl, cache);
+			else
+				btrfs_mark_bg_unused(cache);
+		}
+	} else {
 		inc_block_group_ro(cache, 1);
-	} else if (cache->used == 0) {
-		ASSERT(list_empty(&cache->bg_list));
-		if (btrfs_test_opt(info, DISCARD_ASYNC))
-			btrfs_discard_queue_work(&info->discard_ctl, cache);
-		else
-			btrfs_mark_bg_unused(cache);
 	}
+
 	return 0;
 error:
 	btrfs_put_block_group(cache);
 	return ret;
 }
 
+static int fill_dummy_bgs(struct btrfs_fs_info *fs_info)
+{
+	struct extent_map_tree *em_tree = &fs_info->mapping_tree;
+	struct rb_node *node;
+	int ret = 0;
+
+	for (node = rb_first_cached(&em_tree->map); node; node = rb_next(node)) {
+		struct extent_map *em;
+		struct map_lookup *map;
+		struct btrfs_block_group *bg;
+
+		em = rb_entry(node, struct extent_map, rb_node);
+		map = em->map_lookup;
+		bg = btrfs_create_block_group_cache(fs_info, em->start);
+		if (!bg) {
+			ret = -ENOMEM;
+			break;
+		}
+
+		/* Fill dummy cache as FULL */
+		bg->length = em->len;
+		bg->flags = map->type;
+		bg->cached = BTRFS_CACHE_FINISHED;
+		bg->used = em->len;
+		bg->flags = map->type;
+		ret = btrfs_add_block_group_cache(fs_info, bg);
+		/*
+		 * We may have some valid block group cache added already, in
+		 * that case we skip to the next one.
+		 */
+		if (ret == -EEXIST) {
+			ret = 0;
+			btrfs_put_block_group(bg);
+			continue;
+		}
+
+		if (ret) {
+			btrfs_remove_free_space_cache(bg);
+			btrfs_put_block_group(bg);
+			break;
+		}
+
+		btrfs_add_bg_to_space_info(fs_info, bg);
+
+		set_avail_alloc_bits(fs_info, bg->flags);
+	}
+	if (!ret)
+		btrfs_init_global_block_rsv(fs_info);
+	return ret;
+}
+
 int btrfs_read_block_groups(struct btrfs_fs_info *info)
 {
+	struct btrfs_root *root = btrfs_block_group_root(info);
 	struct btrfs_path *path;
 	int ret;
 	struct btrfs_block_group *cache;
@@ -1956,13 +2163,24 @@ int btrfs_read_block_groups(struct btrfs_fs_info *info)
 	int need_clear = 0;
 	u64 cache_gen;
 
+	/*
+	 * Either no extent root (with ibadroots rescue option) or we have
+	 * unsupported RO options. The fs can never be mounted read-write, so no
+	 * need to waste time searching block group items.
+	 *
+	 * This also allows new extent tree related changes to be RO compat,
+	 * no need for a full incompat flag.
+	 */
+	if (!root || (btrfs_super_compat_ro_flags(info->super_copy) &
+		      ~BTRFS_FEATURE_COMPAT_RO_SUPP))
+		return fill_dummy_bgs(info);
+
 	key.objectid = 0;
 	key.offset = 0;
 	key.type = BTRFS_BLOCK_GROUP_ITEM_KEY;
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
-	path->reada = READA_FORWARD;
 
 	cache_gen = btrfs_super_cache_generation(info->super_copy);
 	if (btrfs_test_opt(info, SPACE_CACHE) &&
@@ -1972,22 +2190,44 @@ int btrfs_read_block_groups(struct btrfs_fs_info *info)
 		need_clear = 1;
 
 	while (1) {
+		struct btrfs_block_group_item bgi;
+		struct extent_buffer *leaf;
+		int slot;
+
 		ret = find_first_block_group(info, path, &key);
 		if (ret > 0)
 			break;
 		if (ret != 0)
 			goto error;
 
-		btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
-		ret = read_one_block_group(info, path, &key, need_clear);
+		leaf = path->nodes[0];
+		slot = path->slots[0];
+
+		read_extent_buffer(leaf, &bgi, btrfs_item_ptr_offset(leaf, slot),
+				   sizeof(bgi));
+
+		btrfs_item_key_to_cpu(leaf, &key, slot);
+		btrfs_release_path(path);
+		ret = read_one_block_group(info, &bgi, &key, need_clear);
 		if (ret < 0)
 			goto error;
 		key.objectid += key.offset;
 		key.offset = 0;
-		btrfs_release_path(path);
 	}
+	btrfs_release_path(path);
+
+	list_for_each_entry(space_info, &info->space_info, list) {
+		int i;
+
+		for (i = 0; i < BTRFS_NR_RAID_TYPES; i++) {
+			if (list_empty(&space_info->block_groups[i]))
+				continue;
+			cache = list_first_entry(&space_info->block_groups[i],
+						 struct btrfs_block_group,
+						 list);
+			btrfs_sysfs_add_block_group_type(cache);
+		}
 
-	list_for_each_entry_rcu(space_info, &info->space_info, list) {
 		if (!(btrfs_get_alloc_profile(info, space_info->flags) &
 		      (BTRFS_BLOCK_GROUP_RAID10 |
 		       BTRFS_BLOCK_GROUP_RAID1_MASK |
@@ -2012,46 +2252,184 @@ int btrfs_read_block_groups(struct btrfs_fs_info *info)
 	ret = check_chunk_block_group_mappings(info);
 error:
 	btrfs_free_path(path);
+	/*
+	 * We've hit some error while reading the extent tree, and have
+	 * rescue=ibadroots mount option.
+	 * Try to fill the tree using dummy block groups so that the user can
+	 * continue to mount and grab their data.
+	 */
+	if (ret && btrfs_test_opt(info, IGNOREBADROOTS))
+		ret = fill_dummy_bgs(info);
 	return ret;
 }
 
-void btrfs_create_pending_block_groups(struct btrfs_trans_handle *trans)
+/*
+ * This function, insert_block_group_item(), belongs to the phase 2 of chunk
+ * allocation.
+ *
+ * See the comment at btrfs_chunk_alloc() for details about the chunk allocation
+ * phases.
+ */
+static int insert_block_group_item(struct btrfs_trans_handle *trans,
+				   struct btrfs_block_group *block_group)
 {
 	struct btrfs_fs_info *fs_info = trans->fs_info;
-	struct btrfs_block_group *block_group;
-	struct btrfs_root *extent_root = fs_info->extent_root;
-	struct btrfs_block_group_item item;
+	struct btrfs_block_group_item bgi;
+	struct btrfs_root *root = btrfs_block_group_root(fs_info);
+	struct btrfs_key key;
+
+	spin_lock(&block_group->lock);
+	btrfs_set_stack_block_group_used(&bgi, block_group->used);
+	btrfs_set_stack_block_group_chunk_objectid(&bgi,
+						   block_group->global_root_id);
+	btrfs_set_stack_block_group_flags(&bgi, block_group->flags);
+	key.objectid = block_group->start;
+	key.type = BTRFS_BLOCK_GROUP_ITEM_KEY;
+	key.offset = block_group->length;
+	spin_unlock(&block_group->lock);
+
+	return btrfs_insert_item(trans, root, &key, &bgi, sizeof(bgi));
+}
+
+static int insert_dev_extent(struct btrfs_trans_handle *trans,
+			    struct btrfs_device *device, u64 chunk_offset,
+			    u64 start, u64 num_bytes)
+{
+	struct btrfs_fs_info *fs_info = device->fs_info;
+	struct btrfs_root *root = fs_info->dev_root;
+	struct btrfs_path *path;
+	struct btrfs_dev_extent *extent;
+	struct extent_buffer *leaf;
 	struct btrfs_key key;
+	int ret;
+
+	WARN_ON(!test_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &device->dev_state));
+	WARN_ON(test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state));
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+
+	key.objectid = device->devid;
+	key.type = BTRFS_DEV_EXTENT_KEY;
+	key.offset = start;
+	ret = btrfs_insert_empty_item(trans, root, path, &key, sizeof(*extent));
+	if (ret)
+		goto out;
+
+	leaf = path->nodes[0];
+	extent = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_dev_extent);
+	btrfs_set_dev_extent_chunk_tree(leaf, extent, BTRFS_CHUNK_TREE_OBJECTID);
+	btrfs_set_dev_extent_chunk_objectid(leaf, extent,
+					    BTRFS_FIRST_CHUNK_TREE_OBJECTID);
+	btrfs_set_dev_extent_chunk_offset(leaf, extent, chunk_offset);
+
+	btrfs_set_dev_extent_length(leaf, extent, num_bytes);
+	btrfs_mark_buffer_dirty(leaf);
+out:
+	btrfs_free_path(path);
+	return ret;
+}
+
+/*
+ * This function belongs to phase 2.
+ *
+ * See the comment at btrfs_chunk_alloc() for details about the chunk allocation
+ * phases.
+ */
+static int insert_dev_extents(struct btrfs_trans_handle *trans,
+				   u64 chunk_offset, u64 chunk_size)
+{
+	struct btrfs_fs_info *fs_info = trans->fs_info;
+	struct btrfs_device *device;
+	struct extent_map *em;
+	struct map_lookup *map;
+	u64 dev_offset;
+	u64 stripe_size;
+	int i;
 	int ret = 0;
 
-	if (!trans->can_flush_pending_bgs)
-		return;
+	em = btrfs_get_chunk_map(fs_info, chunk_offset, chunk_size);
+	if (IS_ERR(em))
+		return PTR_ERR(em);
+
+	map = em->map_lookup;
+	stripe_size = em->orig_block_len;
+
+	/*
+	 * Take the device list mutex to prevent races with the final phase of
+	 * a device replace operation that replaces the device object associated
+	 * with the map's stripes, because the device object's id can change
+	 * at any time during that final phase of the device replace operation
+	 * (dev-replace.c:btrfs_dev_replace_finishing()), so we could grab the
+	 * replaced device and then see it with an ID of BTRFS_DEV_REPLACE_DEVID,
+	 * resulting in persisting a device extent item with such ID.
+	 */
+	mutex_lock(&fs_info->fs_devices->device_list_mutex);
+	for (i = 0; i < map->num_stripes; i++) {
+		device = map->stripes[i].dev;
+		dev_offset = map->stripes[i].physical;
+
+		ret = insert_dev_extent(trans, device, chunk_offset, dev_offset,
+				       stripe_size);
+		if (ret)
+			break;
+	}
+	mutex_unlock(&fs_info->fs_devices->device_list_mutex);
+
+	free_extent_map(em);
+	return ret;
+}
+
+/*
+ * This function, btrfs_create_pending_block_groups(), belongs to the phase 2 of
+ * chunk allocation.
+ *
+ * See the comment at btrfs_chunk_alloc() for details about the chunk allocation
+ * phases.
+ */
+void btrfs_create_pending_block_groups(struct btrfs_trans_handle *trans)
+{
+	struct btrfs_fs_info *fs_info = trans->fs_info;
+	struct btrfs_block_group *block_group;
+	int ret = 0;
 
 	while (!list_empty(&trans->new_bgs)) {
+		int index;
+
 		block_group = list_first_entry(&trans->new_bgs,
 					       struct btrfs_block_group,
 					       bg_list);
 		if (ret)
 			goto next;
 
-		spin_lock(&block_group->lock);
-		btrfs_set_stack_block_group_used(&item, block_group->used);
-		btrfs_set_stack_block_group_chunk_objectid(&item,
-				BTRFS_FIRST_CHUNK_TREE_OBJECTID);
-		btrfs_set_stack_block_group_flags(&item, block_group->flags);
-		key.objectid = block_group->start;
-		key.type = BTRFS_BLOCK_GROUP_ITEM_KEY;
-		key.offset = block_group->length;
-		spin_unlock(&block_group->lock);
+		index = btrfs_bg_flags_to_raid_index(block_group->flags);
 
-		ret = btrfs_insert_item(trans, extent_root, &key, &item,
-					sizeof(item));
+		ret = insert_block_group_item(trans, block_group);
 		if (ret)
 			btrfs_abort_transaction(trans, ret);
-		ret = btrfs_finish_chunk_alloc(trans, key.objectid, key.offset);
+		if (!test_bit(BLOCK_GROUP_FLAG_CHUNK_ITEM_INSERTED,
+			      &block_group->runtime_flags)) {
+			mutex_lock(&fs_info->chunk_mutex);
+			ret = btrfs_chunk_alloc_add_chunk_item(trans, block_group);
+			mutex_unlock(&fs_info->chunk_mutex);
+			if (ret)
+				btrfs_abort_transaction(trans, ret);
+		}
+		ret = insert_dev_extents(trans, block_group->start,
+					 block_group->length);
 		if (ret)
 			btrfs_abort_transaction(trans, ret);
 		add_block_group_free_space(trans, block_group);
+
+		/*
+		 * If we restriped during balance, we may have added a new raid
+		 * type, so now add the sysfs entries when it is safe to do so.
+		 * We don't have to worry about locking here as it's handled in
+		 * btrfs_sysfs_add_block_group_type.
+		 */
+		if (block_group->space_info->block_group_kobjs[index] == NULL)
+			btrfs_sysfs_add_block_group_type(block_group);
+
 		/* Already aborted the transaction if it failed. */
 next:
 		btrfs_delayed_refs_rsv_release(fs_info, 1);
@@ -2060,8 +2438,30 @@ next:
 	btrfs_trans_release_chunk_metadata(trans);
 }
 
-int btrfs_make_block_group(struct btrfs_trans_handle *trans, u64 bytes_used,
-			   u64 type, u64 chunk_offset, u64 size)
+/*
+ * For extent tree v2 we use the block_group_item->chunk_offset to point at our
+ * global root id.  For v1 it's always set to BTRFS_FIRST_CHUNK_TREE_OBJECTID.
+ */
+static u64 calculate_global_root_id(struct btrfs_fs_info *fs_info, u64 offset)
+{
+	u64 div = SZ_1G;
+	u64 index;
+
+	if (!btrfs_fs_incompat(fs_info, EXTENT_TREE_V2))
+		return BTRFS_FIRST_CHUNK_TREE_OBJECTID;
+
+	/* If we have a smaller fs index based on 128MiB. */
+	if (btrfs_super_total_bytes(fs_info->super_copy) <= (SZ_1G * 10ULL))
+		div = SZ_128M;
+
+	offset = div64_u64(offset, div);
+	div64_u64_rem(offset, fs_info->nr_global_roots, &index);
+	return index;
+}
+
+struct btrfs_block_group *btrfs_make_block_group(struct btrfs_trans_handle *trans,
+						 u64 bytes_used, u64 type,
+						 u64 chunk_offset, u64 size)
 {
 	struct btrfs_fs_info *fs_info = trans->fs_info;
 	struct btrfs_block_group *cache;
@@ -2069,35 +2469,38 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans, u64 bytes_used,
 
 	btrfs_set_log_full_commit(trans);
 
-	cache = btrfs_create_block_group_cache(fs_info, chunk_offset, size);
+	cache = btrfs_create_block_group_cache(fs_info, chunk_offset);
 	if (!cache)
-		return -ENOMEM;
+		return ERR_PTR(-ENOMEM);
 
+	cache->length = size;
+	set_free_space_tree_thresholds(cache);
 	cache->used = bytes_used;
 	cache->flags = type;
-	cache->last_byte_to_unpin = (u64)-1;
 	cache->cached = BTRFS_CACHE_FINISHED;
-	cache->needs_free_space = 1;
+	cache->global_root_id = calculate_global_root_id(fs_info, cache->start);
+
+	if (btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE))
+		cache->needs_free_space = 1;
+
+	ret = btrfs_load_block_group_zone_info(cache, true);
+	if (ret) {
+		btrfs_put_block_group(cache);
+		return ERR_PTR(ret);
+	}
+
 	ret = exclude_super_stripes(cache);
 	if (ret) {
 		/* We may have excluded something, so call this just in case */
 		btrfs_free_excluded_extents(cache);
 		btrfs_put_block_group(cache);
-		return ret;
+		return ERR_PTR(ret);
 	}
 
 	add_new_free_space(cache, chunk_offset, chunk_offset + size);
 
 	btrfs_free_excluded_extents(cache);
 
-#ifdef CONFIG_BTRFS_DEBUG
-	if (btrfs_should_fragment_free_space(cache)) {
-		u64 new_bytes_used = size - bytes_used;
-
-		bytes_used += new_bytes_used >> 1;
-		fragment_free_space(cache);
-	}
-#endif
 	/*
 	 * Ensure the corresponding space_info object is created and
 	 * assigned to our block group. We want our bg to be added to the rbtree
@@ -2110,7 +2513,7 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans, u64 bytes_used,
 	if (ret) {
 		btrfs_remove_free_space_cache(cache);
 		btrfs_put_block_group(cache);
-		return ret;
+		return ERR_PTR(ret);
 	}
 
 	/*
@@ -2118,66 +2521,24 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans, u64 bytes_used,
 	 * the rbtree, update the space info's counters.
 	 */
 	trace_btrfs_add_block_group(fs_info, cache, 1);
-	btrfs_update_space_info(fs_info, cache->flags, size, bytes_used,
-				cache->bytes_super, &cache->space_info);
+	btrfs_add_bg_to_space_info(fs_info, cache);
 	btrfs_update_global_block_rsv(fs_info);
 
-	link_block_group(cache);
+#ifdef CONFIG_BTRFS_DEBUG
+	if (btrfs_should_fragment_free_space(cache)) {
+		u64 new_bytes_used = size - bytes_used;
+
+		cache->space_info->bytes_used += new_bytes_used >> 1;
+		fragment_free_space(cache);
+	}
+#endif
 
 	list_add_tail(&cache->bg_list, &trans->new_bgs);
 	trans->delayed_ref_updates++;
 	btrfs_update_delayed_refs_rsv(trans);
 
 	set_avail_alloc_bits(fs_info, type);
-	return 0;
-}
-
-static u64 update_block_group_flags(struct btrfs_fs_info *fs_info, u64 flags)
-{
-	u64 num_devices;
-	u64 stripped;
-
-	/*
-	 * if restripe for this chunk_type is on pick target profile and
-	 * return, otherwise do the usual balance
-	 */
-	stripped = get_restripe_target(fs_info, flags);
-	if (stripped)
-		return extended_to_chunk(stripped);
-
-	num_devices = fs_info->fs_devices->rw_devices;
-
-	stripped = BTRFS_BLOCK_GROUP_RAID0 | BTRFS_BLOCK_GROUP_RAID56_MASK |
-		BTRFS_BLOCK_GROUP_RAID1_MASK | BTRFS_BLOCK_GROUP_RAID10;
-
-	if (num_devices == 1) {
-		stripped |= BTRFS_BLOCK_GROUP_DUP;
-		stripped = flags & ~stripped;
-
-		/* turn raid0 into single device chunks */
-		if (flags & BTRFS_BLOCK_GROUP_RAID0)
-			return stripped;
-
-		/* turn mirroring into duplication */
-		if (flags & (BTRFS_BLOCK_GROUP_RAID1_MASK |
-			     BTRFS_BLOCK_GROUP_RAID10))
-			return stripped | BTRFS_BLOCK_GROUP_DUP;
-	} else {
-		/* they already had raid on here, just return */
-		if (flags & stripped)
-			return flags;
-
-		stripped |= BTRFS_BLOCK_GROUP_DUP;
-		stripped = flags & ~stripped;
-
-		/* switch duplicated blocks with raid1 */
-		if (flags & BTRFS_BLOCK_GROUP_DUP)
-			return stripped | BTRFS_BLOCK_GROUP_RAID1;
-
-		/* this is drive concat, leave it alone */
-	}
-
-	return flags;
+	return cache;
 }
 
 /*
@@ -2194,38 +2555,56 @@ int btrfs_inc_block_group_ro(struct btrfs_block_group *cache,
 {
 	struct btrfs_fs_info *fs_info = cache->fs_info;
 	struct btrfs_trans_handle *trans;
+	struct btrfs_root *root = btrfs_block_group_root(fs_info);
 	u64 alloc_flags;
 	int ret;
-
-again:
-	trans = btrfs_join_transaction(fs_info->extent_root);
-	if (IS_ERR(trans))
-		return PTR_ERR(trans);
+	bool dirty_bg_running;
 
 	/*
-	 * we're not allowed to set block groups readonly after the dirty
-	 * block groups cache has started writing.  If it already started,
-	 * back off and let this transaction commit
+	 * This can only happen when we are doing read-only scrub on read-only
+	 * mount.
+	 * In that case we should not start a new transaction on read-only fs.
+	 * Thus here we skip all chunk allocations.
 	 */
-	mutex_lock(&fs_info->ro_block_group_mutex);
-	if (test_bit(BTRFS_TRANS_DIRTY_BG_RUN, &trans->transaction->flags)) {
-		u64 transid = trans->transid;
-
+	if (sb_rdonly(fs_info->sb)) {
+		mutex_lock(&fs_info->ro_block_group_mutex);
+		ret = inc_block_group_ro(cache, 0);
 		mutex_unlock(&fs_info->ro_block_group_mutex);
-		btrfs_end_transaction(trans);
-
-		ret = btrfs_wait_for_commit(fs_info, transid);
-		if (ret)
-			return ret;
-		goto again;
+		return ret;
 	}
 
+	do {
+		trans = btrfs_join_transaction(root);
+		if (IS_ERR(trans))
+			return PTR_ERR(trans);
+
+		dirty_bg_running = false;
+
+		/*
+		 * We're not allowed to set block groups readonly after the dirty
+		 * block group cache has started writing.  If it already started,
+		 * back off and let this transaction commit.
+		 */
+		mutex_lock(&fs_info->ro_block_group_mutex);
+		if (test_bit(BTRFS_TRANS_DIRTY_BG_RUN, &trans->transaction->flags)) {
+			u64 transid = trans->transid;
+
+			mutex_unlock(&fs_info->ro_block_group_mutex);
+			btrfs_end_transaction(trans);
+
+			ret = btrfs_wait_for_commit(fs_info, transid);
+			if (ret)
+				return ret;
+			dirty_bg_running = true;
+		}
+	} while (dirty_bg_running);
+
 	if (do_chunk_alloc) {
 		/*
 		 * If we are changing raid levels, try to allocate a
 		 * corresponding block group with the new raid level.
 		 */
-		alloc_flags = update_block_group_flags(fs_info, cache->flags);
+		alloc_flags = btrfs_get_alloc_profile(fs_info, cache->flags);
 		if (alloc_flags != cache->flags) {
 			ret = btrfs_chunk_alloc(trans, alloc_flags,
 						CHUNK_ALLOC_FORCE);
@@ -2241,7 +2620,7 @@ again:
 	}
 
 	ret = inc_block_group_ro(cache, 0);
-	if (!do_chunk_alloc)
+	if (!do_chunk_alloc || ret == -ETXTBSY)
 		goto unlock_out;
 	if (!ret)
 		goto out;
@@ -2249,10 +2628,20 @@ again:
 	ret = btrfs_chunk_alloc(trans, alloc_flags, CHUNK_ALLOC_FORCE);
 	if (ret < 0)
 		goto out;
+	/*
+	 * We have allocated a new chunk. We also need to activate that chunk to
+	 * grant metadata tickets for zoned filesystem.
+	 */
+	ret = btrfs_zoned_activate_one_bg(fs_info, cache->space_info, true);
+	if (ret < 0)
+		goto out;
+
 	ret = inc_block_group_ro(cache, 0);
+	if (ret == -ETXTBSY)
+		goto unlock_out;
 out:
 	if (cache->flags & BTRFS_BLOCK_GROUP_SYSTEM) {
-		alloc_flags = update_block_group_flags(fs_info, cache->flags);
+		alloc_flags = btrfs_get_alloc_profile(fs_info, cache->flags);
 		mutex_lock(&fs_info->chunk_mutex);
 		check_system_chunk(trans, alloc_flags);
 		mutex_unlock(&fs_info->chunk_mutex);
@@ -2274,8 +2663,17 @@ void btrfs_dec_block_group_ro(struct btrfs_block_group *cache)
 	spin_lock(&sinfo->lock);
 	spin_lock(&cache->lock);
 	if (!--cache->ro) {
+		if (btrfs_is_zoned(cache->fs_info)) {
+			/* Migrate zone_unusable bytes back */
+			cache->zone_unusable =
+				(cache->alloc_offset - cache->used) +
+				(cache->length - cache->zone_capacity);
+			sinfo->bytes_zone_unusable += cache->zone_unusable;
+			sinfo->bytes_readonly -= cache->zone_unusable;
+		}
 		num_bytes = cache->length - cache->reserved -
-			    cache->pinned - cache->bytes_super - cache->used;
+			    cache->pinned - cache->bytes_super -
+			    cache->zone_unusable - cache->used;
 		sinfo->bytes_readonly -= num_bytes;
 		list_del_init(&cache->ro_list);
 	}
@@ -2283,13 +2681,13 @@ void btrfs_dec_block_group_ro(struct btrfs_block_group *cache)
 	spin_unlock(&sinfo->lock);
 }
 
-static int write_one_cache_group(struct btrfs_trans_handle *trans,
-				 struct btrfs_path *path,
-				 struct btrfs_block_group *cache)
+static int update_block_group_item(struct btrfs_trans_handle *trans,
+				   struct btrfs_path *path,
+				   struct btrfs_block_group *cache)
 {
 	struct btrfs_fs_info *fs_info = trans->fs_info;
 	int ret;
-	struct btrfs_root *extent_root = fs_info->extent_root;
+	struct btrfs_root *root = btrfs_block_group_root(fs_info);
 	unsigned long bi;
 	struct extent_buffer *leaf;
 	struct btrfs_block_group_item bgi;
@@ -2299,7 +2697,7 @@ static int write_one_cache_group(struct btrfs_trans_handle *trans,
 	key.type = BTRFS_BLOCK_GROUP_ITEM_KEY;
 	key.offset = cache->length;
 
-	ret = btrfs_search_slot(trans, extent_root, &key, path, 0, 1);
+	ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
 	if (ret) {
 		if (ret > 0)
 			ret = -ENOENT;
@@ -2310,7 +2708,7 @@ static int write_one_cache_group(struct btrfs_trans_handle *trans,
 	bi = btrfs_item_ptr_offset(leaf, path->slots[0]);
 	btrfs_set_stack_block_group_used(&bgi, cache->used);
 	btrfs_set_stack_block_group_chunk_objectid(&bgi,
-			BTRFS_FIRST_CHUNK_TREE_OBJECTID);
+						   cache->global_root_id);
 	btrfs_set_stack_block_group_flags(&bgi, cache->flags);
 	write_extent_buffer(leaf, &bgi, bi, sizeof(bgi));
 	btrfs_mark_buffer_dirty(leaf);
@@ -2330,10 +2728,13 @@ static int cache_save_setup(struct btrfs_block_group *block_group,
 	struct extent_changeset *data_reserved = NULL;
 	u64 alloc_hint = 0;
 	int dcs = BTRFS_DC_ERROR;
-	u64 num_pages = 0;
+	u64 cache_size = 0;
 	int retries = 0;
 	int ret = 0;
 
+	if (!btrfs_test_opt(fs_info, SPACE_CACHE))
+		return 0;
+
 	/*
 	 * If this block group is smaller than 100 megs don't bother caching the
 	 * block group.
@@ -2345,7 +2746,7 @@ static int cache_save_setup(struct btrfs_block_group *block_group,
 		return 0;
 	}
 
-	if (trans->aborted)
+	if (TRANS_ABORTED(trans))
 		return 0;
 again:
 	inode = lookup_free_space_inode(block_group, path);
@@ -2374,7 +2775,7 @@ again:
 	 * time.
 	 */
 	BTRFS_I(inode)->generation = 0;
-	ret = btrfs_update_inode(trans, root, inode);
+	ret = btrfs_update_inode(trans, root, BTRFS_I(inode));
 	if (ret) {
 		/*
 		 * So theoretically we could recover from this, simply set the
@@ -2439,19 +2840,20 @@ again:
 	 * taking up quite a bit since it's not folded into the other space
 	 * cache.
 	 */
-	num_pages = div_u64(block_group->length, SZ_256M);
-	if (!num_pages)
-		num_pages = 1;
+	cache_size = div_u64(block_group->length, SZ_256M);
+	if (!cache_size)
+		cache_size = 1;
 
-	num_pages *= 16;
-	num_pages *= PAGE_SIZE;
+	cache_size *= 16;
+	cache_size *= fs_info->sectorsize;
 
-	ret = btrfs_check_data_free_space(inode, &data_reserved, 0, num_pages);
+	ret = btrfs_check_data_free_space(BTRFS_I(inode), &data_reserved, 0,
+					  cache_size, false);
 	if (ret)
 		goto out_put;
 
-	ret = btrfs_prealloc_file_range_trans(inode, trans, 0, 0, num_pages,
-					      num_pages, num_pages,
+	ret = btrfs_prealloc_file_range_trans(inode, trans, 0, 0, cache_size,
+					      cache_size, cache_size,
 					      &alloc_hint);
 	/*
 	 * Our cache requires contiguous chunks so that we don't modify a bunch
@@ -2529,7 +2931,6 @@ int btrfs_start_dirty_block_groups(struct btrfs_trans_handle *trans)
 	struct btrfs_path *path = NULL;
 	LIST_HEAD(dirty);
 	struct list_head *io = &cur_trans->io_bgs;
-	int num_started = 0;
 	int loops = 0;
 
 	spin_lock(&cur_trans->dirty_bgs_lock);
@@ -2546,8 +2947,10 @@ again:
 
 	if (!path) {
 		path = btrfs_alloc_path();
-		if (!path)
-			return -ENOMEM;
+		if (!path) {
+			ret = -ENOMEM;
+			goto out;
+		}
 	}
 
 	/*
@@ -2593,7 +2996,6 @@ again:
 			cache->io_ctl.inode = NULL;
 			ret = btrfs_write_out_cache(trans, cache, path);
 			if (ret == 0 && cache->io_ctl.inode) {
-				num_started++;
 				should_put = 0;
 
 				/*
@@ -2611,7 +3013,7 @@ again:
 			}
 		}
 		if (!ret) {
-			ret = write_one_cache_group(trans, path, cache);
+			ret = update_block_group_item(trans, path, cache);
 			/*
 			 * Our block group might still be attached to the list
 			 * of new block groups in the transaction handle of some
@@ -2641,16 +3043,14 @@ again:
 			btrfs_put_block_group(cache);
 		if (drop_reserve)
 			btrfs_delayed_refs_rsv_release(fs_info, 1);
-
-		if (ret)
-			break;
-
 		/*
 		 * Avoid blocking other tasks for too long. It might even save
 		 * us from writing caches for block groups that are going to be
 		 * removed.
 		 */
 		mutex_unlock(&trans->transaction->cache_write_mutex);
+		if (ret)
+			goto out;
 		mutex_lock(&trans->transaction->cache_write_mutex);
 	}
 	mutex_unlock(&trans->transaction->cache_write_mutex);
@@ -2659,7 +3059,8 @@ again:
 	 * Go through delayed refs for all the stuff we've just kicked off
 	 * and then loop back (just once)
 	 */
-	ret = btrfs_run_delayed_refs(trans, 0);
+	if (!ret)
+		ret = btrfs_run_delayed_refs(trans, 0);
 	if (!ret && loops == 0) {
 		loops++;
 		spin_lock(&cur_trans->dirty_bgs_lock);
@@ -2673,7 +3074,12 @@ again:
 			goto again;
 		}
 		spin_unlock(&cur_trans->dirty_bgs_lock);
-	} else if (ret < 0) {
+	}
+out:
+	if (ret < 0) {
+		spin_lock(&cur_trans->dirty_bgs_lock);
+		list_splice_init(&dirty, &cur_trans->dirty_bgs);
+		spin_unlock(&cur_trans->dirty_bgs_lock);
 		btrfs_cleanup_dirty_bgs(cur_trans, fs_info);
 	}
 
@@ -2690,7 +3096,6 @@ int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans)
 	int should_put;
 	struct btrfs_path *path;
 	struct list_head *io = &cur_trans->io_bgs;
-	int num_started = 0;
 
 	path = btrfs_alloc_path();
 	if (!path)
@@ -2748,7 +3153,6 @@ int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans)
 			cache->io_ctl.inode = NULL;
 			ret = btrfs_write_out_cache(trans, cache, path);
 			if (ret == 0 && cache->io_ctl.inode) {
-				num_started++;
 				should_put = 0;
 				list_add_tail(&cache->io_list, io);
 			} else {
@@ -2760,7 +3164,7 @@ int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans)
 			}
 		}
 		if (!ret) {
-			ret = write_one_cache_group(trans, path, cache);
+			ret = update_block_group_item(trans, path, cache);
 			/*
 			 * One of the free space endio workers might have
 			 * created a new block group while updating a free space
@@ -2771,13 +3175,13 @@ int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans)
 			 * finished yet (no block group item in the extent tree
 			 * yet, etc). If this is the case, wait for all free
 			 * space endio workers to finish and retry. This is a
-			 * a very rare case so no need for a more efficient and
+			 * very rare case so no need for a more efficient and
 			 * complex approach.
 			 */
 			if (ret == -ENOENT) {
 				wait_event(cur_trans->writer_wait,
 				   atomic_read(&cur_trans->num_writers) == 1);
-				ret = write_one_cache_group(trans, path, cache);
+				ret = update_block_group_item(trans, path, cache);
 			}
 			if (ret)
 				btrfs_abort_transaction(trans, ret);
@@ -2807,8 +3211,33 @@ int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans)
 	return ret;
 }
 
+static inline bool should_reclaim_block_group(struct btrfs_block_group *bg,
+					      u64 bytes_freed)
+{
+	const struct btrfs_space_info *space_info = bg->space_info;
+	const int reclaim_thresh = READ_ONCE(space_info->bg_reclaim_threshold);
+	const u64 new_val = bg->used;
+	const u64 old_val = new_val + bytes_freed;
+	u64 thresh;
+
+	if (reclaim_thresh == 0)
+		return false;
+
+	thresh = div_factor_fine(bg->length, reclaim_thresh);
+
+	/*
+	 * If we were below the threshold before don't reclaim, we are likely a
+	 * brand new block group and we don't want to relocate new block groups.
+	 */
+	if (old_val < thresh)
+		return false;
+	if (new_val >= thresh)
+		return false;
+	return true;
+}
+
 int btrfs_update_block_group(struct btrfs_trans_handle *trans,
-			     u64 bytenr, u64 num_bytes, int alloc)
+			     u64 bytenr, u64 num_bytes, bool alloc)
 {
 	struct btrfs_fs_info *info = trans->fs_info;
 	struct btrfs_block_group *cache = NULL;
@@ -2829,6 +3258,8 @@ int btrfs_update_block_group(struct btrfs_trans_handle *trans,
 	spin_unlock(&info->delalloc_root_lock);
 
 	while (total) {
+		bool reclaim;
+
 		cache = btrfs_lookup_block_group(info, bytenr);
 		if (!cache) {
 			ret = -ENOENT;
@@ -2843,7 +3274,7 @@ int btrfs_update_block_group(struct btrfs_trans_handle *trans,
 		 * space back to the block group, otherwise we will leak space.
 		 */
 		if (!alloc && !btrfs_block_group_done(cache))
-			btrfs_cache_block_group(cache, 1);
+			btrfs_cache_block_group(cache, true);
 
 		byte_in_group = bytenr - cache->start;
 		WARN_ON(byte_in_group > cache->length);
@@ -2874,14 +3305,12 @@ int btrfs_update_block_group(struct btrfs_trans_handle *trans,
 					cache->space_info, num_bytes);
 			cache->space_info->bytes_used -= num_bytes;
 			cache->space_info->disk_used -= num_bytes * factor;
+
+			reclaim = should_reclaim_block_group(cache, num_bytes);
 			spin_unlock(&cache->lock);
 			spin_unlock(&cache->space_info->lock);
 
-			percpu_counter_add_batch(
-					&cache->space_info->total_bytes_pinned,
-					num_bytes,
-					BTRFS_TOTAL_BYTES_PINNED_BATCH);
-			set_extent_dirty(info->pinned_extents,
+			set_extent_dirty(&trans->transaction->pinned_extents,
 					 bytenr, bytenr + num_bytes - 1,
 					 GFP_NOFS | __GFP_NOFAIL);
 		}
@@ -2904,6 +3333,8 @@ int btrfs_update_block_group(struct btrfs_trans_handle *trans,
 		if (!alloc && old_val == 0) {
 			if (!btrfs_test_opt(info, DISCARD_ASYNC))
 				btrfs_mark_bg_unused(cache);
+		} else if (!alloc && reclaim) {
+			btrfs_mark_bg_to_reclaim(cache);
 		}
 
 		btrfs_put_block_group(cache);
@@ -2947,6 +3378,13 @@ int btrfs_add_reserved_bytes(struct btrfs_block_group *cache,
 						      space_info, -ram_bytes);
 		if (delalloc)
 			cache->delalloc_bytes += num_bytes;
+
+		/*
+		 * Compression can use less space than we reserved, so wake
+		 * tickets if that happens
+		 */
+		if (num_bytes < ram_bytes)
+			btrfs_try_granting_tickets(cache->fs_info, space_info);
 	}
 	spin_unlock(&cache->lock);
 	spin_unlock(&space_info->lock);
@@ -2980,6 +3418,8 @@ void btrfs_free_reserved_bytes(struct btrfs_block_group *cache,
 	if (delalloc)
 		cache->delalloc_bytes -= num_bytes;
 	spin_unlock(&cache->lock);
+
+	btrfs_try_granting_tickets(cache->fs_info, space_info);
 	spin_unlock(&space_info->lock);
 }
 
@@ -2988,12 +3428,10 @@ static void force_metadata_allocation(struct btrfs_fs_info *info)
 	struct list_head *head = &info->space_info;
 	struct btrfs_space_info *found;
 
-	rcu_read_lock();
-	list_for_each_entry_rcu(found, head, list) {
+	list_for_each_entry(found, head, list) {
 		if (found->flags & BTRFS_BLOCK_GROUP_METADATA)
 			found->force_alloc = CHUNK_ALLOC_FORCE;
 	}
-	rcu_read_unlock();
 }
 
 static int should_alloc_chunk(struct btrfs_fs_info *fs_info,
@@ -3029,11 +3467,203 @@ int btrfs_force_chunk_alloc(struct btrfs_trans_handle *trans, u64 type)
 	return btrfs_chunk_alloc(trans, alloc_flags, CHUNK_ALLOC_FORCE);
 }
 
+static struct btrfs_block_group *do_chunk_alloc(struct btrfs_trans_handle *trans, u64 flags)
+{
+	struct btrfs_block_group *bg;
+	int ret;
+
+	/*
+	 * Check if we have enough space in the system space info because we
+	 * will need to update device items in the chunk btree and insert a new
+	 * chunk item in the chunk btree as well. This will allocate a new
+	 * system block group if needed.
+	 */
+	check_system_chunk(trans, flags);
+
+	bg = btrfs_create_chunk(trans, flags);
+	if (IS_ERR(bg)) {
+		ret = PTR_ERR(bg);
+		goto out;
+	}
+
+	ret = btrfs_chunk_alloc_add_chunk_item(trans, bg);
+	/*
+	 * Normally we are not expected to fail with -ENOSPC here, since we have
+	 * previously reserved space in the system space_info and allocated one
+	 * new system chunk if necessary. However there are three exceptions:
+	 *
+	 * 1) We may have enough free space in the system space_info but all the
+	 *    existing system block groups have a profile which can not be used
+	 *    for extent allocation.
+	 *
+	 *    This happens when mounting in degraded mode. For example we have a
+	 *    RAID1 filesystem with 2 devices, lose one device and mount the fs
+	 *    using the other device in degraded mode. If we then allocate a chunk,
+	 *    we may have enough free space in the existing system space_info, but
+	 *    none of the block groups can be used for extent allocation since they
+	 *    have a RAID1 profile, and because we are in degraded mode with a
+	 *    single device, we are forced to allocate a new system chunk with a
+	 *    SINGLE profile. Making check_system_chunk() iterate over all system
+	 *    block groups and check if they have a usable profile and enough space
+	 *    can be slow on very large filesystems, so we tolerate the -ENOSPC and
+	 *    try again after forcing allocation of a new system chunk. Like this
+	 *    we avoid paying the cost of that search in normal circumstances, when
+	 *    we were not mounted in degraded mode;
+	 *
+	 * 2) We had enough free space info the system space_info, and one suitable
+	 *    block group to allocate from when we called check_system_chunk()
+	 *    above. However right after we called it, the only system block group
+	 *    with enough free space got turned into RO mode by a running scrub,
+	 *    and in this case we have to allocate a new one and retry. We only
+	 *    need do this allocate and retry once, since we have a transaction
+	 *    handle and scrub uses the commit root to search for block groups;
+	 *
+	 * 3) We had one system block group with enough free space when we called
+	 *    check_system_chunk(), but after that, right before we tried to
+	 *    allocate the last extent buffer we needed, a discard operation came
+	 *    in and it temporarily removed the last free space entry from the
+	 *    block group (discard removes a free space entry, discards it, and
+	 *    then adds back the entry to the block group cache).
+	 */
+	if (ret == -ENOSPC) {
+		const u64 sys_flags = btrfs_system_alloc_profile(trans->fs_info);
+		struct btrfs_block_group *sys_bg;
+
+		sys_bg = btrfs_create_chunk(trans, sys_flags);
+		if (IS_ERR(sys_bg)) {
+			ret = PTR_ERR(sys_bg);
+			btrfs_abort_transaction(trans, ret);
+			goto out;
+		}
+
+		ret = btrfs_chunk_alloc_add_chunk_item(trans, sys_bg);
+		if (ret) {
+			btrfs_abort_transaction(trans, ret);
+			goto out;
+		}
+
+		ret = btrfs_chunk_alloc_add_chunk_item(trans, bg);
+		if (ret) {
+			btrfs_abort_transaction(trans, ret);
+			goto out;
+		}
+	} else if (ret) {
+		btrfs_abort_transaction(trans, ret);
+		goto out;
+	}
+out:
+	btrfs_trans_release_chunk_metadata(trans);
+
+	if (ret)
+		return ERR_PTR(ret);
+
+	btrfs_get_block_group(bg);
+	return bg;
+}
+
 /*
- * If force is CHUNK_ALLOC_FORCE:
+ * Chunk allocation is done in 2 phases:
+ *
+ * 1) Phase 1 - through btrfs_chunk_alloc() we allocate device extents for
+ *    the chunk, the chunk mapping, create its block group and add the items
+ *    that belong in the chunk btree to it - more specifically, we need to
+ *    update device items in the chunk btree and add a new chunk item to it.
+ *
+ * 2) Phase 2 - through btrfs_create_pending_block_groups(), we add the block
+ *    group item to the extent btree and the device extent items to the devices
+ *    btree.
+ *
+ * This is done to prevent deadlocks. For example when COWing a node from the
+ * extent btree we are holding a write lock on the node's parent and if we
+ * trigger chunk allocation and attempted to insert the new block group item
+ * in the extent btree right way, we could deadlock because the path for the
+ * insertion can include that parent node. At first glance it seems impossible
+ * to trigger chunk allocation after starting a transaction since tasks should
+ * reserve enough transaction units (metadata space), however while that is true
+ * most of the time, chunk allocation may still be triggered for several reasons:
+ *
+ * 1) When reserving metadata, we check if there is enough free space in the
+ *    metadata space_info and therefore don't trigger allocation of a new chunk.
+ *    However later when the task actually tries to COW an extent buffer from
+ *    the extent btree or from the device btree for example, it is forced to
+ *    allocate a new block group (chunk) because the only one that had enough
+ *    free space was just turned to RO mode by a running scrub for example (or
+ *    device replace, block group reclaim thread, etc), so we can not use it
+ *    for allocating an extent and end up being forced to allocate a new one;
+ *
+ * 2) Because we only check that the metadata space_info has enough free bytes,
+ *    we end up not allocating a new metadata chunk in that case. However if
+ *    the filesystem was mounted in degraded mode, none of the existing block
+ *    groups might be suitable for extent allocation due to their incompatible
+ *    profile (for e.g. mounting a 2 devices filesystem, where all block groups
+ *    use a RAID1 profile, in degraded mode using a single device). In this case
+ *    when the task attempts to COW some extent buffer of the extent btree for
+ *    example, it will trigger allocation of a new metadata block group with a
+ *    suitable profile (SINGLE profile in the example of the degraded mount of
+ *    the RAID1 filesystem);
+ *
+ * 3) The task has reserved enough transaction units / metadata space, but when
+ *    it attempts to COW an extent buffer from the extent or device btree for
+ *    example, it does not find any free extent in any metadata block group,
+ *    therefore forced to try to allocate a new metadata block group.
+ *    This is because some other task allocated all available extents in the
+ *    meanwhile - this typically happens with tasks that don't reserve space
+ *    properly, either intentionally or as a bug. One example where this is
+ *    done intentionally is fsync, as it does not reserve any transaction units
+ *    and ends up allocating a variable number of metadata extents for log
+ *    tree extent buffers;
+ *
+ * 4) The task has reserved enough transaction units / metadata space, but right
+ *    before it tries to allocate the last extent buffer it needs, a discard
+ *    operation comes in and, temporarily, removes the last free space entry from
+ *    the only metadata block group that had free space (discard starts by
+ *    removing a free space entry from a block group, then does the discard
+ *    operation and, once it's done, it adds back the free space entry to the
+ *    block group).
+ *
+ * We also need this 2 phases setup when adding a device to a filesystem with
+ * a seed device - we must create new metadata and system chunks without adding
+ * any of the block group items to the chunk, extent and device btrees. If we
+ * did not do it this way, we would get ENOSPC when attempting to update those
+ * btrees, since all the chunks from the seed device are read-only.
+ *
+ * Phase 1 does the updates and insertions to the chunk btree because if we had
+ * it done in phase 2 and have a thundering herd of tasks allocating chunks in
+ * parallel, we risk having too many system chunks allocated by many tasks if
+ * many tasks reach phase 1 without the previous ones completing phase 2. In the
+ * extreme case this leads to exhaustion of the system chunk array in the
+ * superblock. This is easier to trigger if using a btree node/leaf size of 64K
+ * and with RAID filesystems (so we have more device items in the chunk btree).
+ * This has happened before and commit eafa4fd0ad0607 ("btrfs: fix exhaustion of
+ * the system chunk array due to concurrent allocations") provides more details.
+ *
+ * Allocation of system chunks does not happen through this function. A task that
+ * needs to update the chunk btree (the only btree that uses system chunks), must
+ * preallocate chunk space by calling either check_system_chunk() or
+ * btrfs_reserve_chunk_metadata() - the former is used when allocating a data or
+ * metadata chunk or when removing a chunk, while the later is used before doing
+ * a modification to the chunk btree - use cases for the later are adding,
+ * removing and resizing a device as well as relocation of a system chunk.
+ * See the comment below for more details.
+ *
+ * The reservation of system space, done through check_system_chunk(), as well
+ * as all the updates and insertions into the chunk btree must be done while
+ * holding fs_info->chunk_mutex. This is important to guarantee that while COWing
+ * an extent buffer from the chunks btree we never trigger allocation of a new
+ * system chunk, which would result in a deadlock (trying to lock twice an
+ * extent buffer of the chunk btree, first time before triggering the chunk
+ * allocation and the second time during chunk allocation while attempting to
+ * update the chunks btree). The system chunk array is also updated while holding
+ * that mutex. The same logic applies to removing chunks - we must reserve system
+ * space, update the chunk btree and the system chunk array in the superblock
+ * while holding fs_info->chunk_mutex.
+ *
+ * This function, btrfs_chunk_alloc(), belongs to phase 1.
+ *
+ * If @force is CHUNK_ALLOC_FORCE:
  *    - return 1 if it successfully allocates a chunk,
  *    - return errors including -ENOSPC otherwise.
- * If force is NOT CHUNK_ALLOC_FORCE:
+ * If @force is NOT CHUNK_ALLOC_FORCE:
  *    - return 0 if it doesn't need to allocate a new chunk,
  *    - return 1 if it successfully allocates a chunk,
  *    - return errors including -ENOSPC otherwise.
@@ -3043,13 +3673,43 @@ int btrfs_chunk_alloc(struct btrfs_trans_handle *trans, u64 flags,
 {
 	struct btrfs_fs_info *fs_info = trans->fs_info;
 	struct btrfs_space_info *space_info;
+	struct btrfs_block_group *ret_bg;
 	bool wait_for_alloc = false;
 	bool should_alloc = false;
+	bool from_extent_allocation = false;
 	int ret = 0;
 
+	if (force == CHUNK_ALLOC_FORCE_FOR_EXTENT) {
+		from_extent_allocation = true;
+		force = CHUNK_ALLOC_FORCE;
+	}
+
 	/* Don't re-enter if we're already allocating a chunk */
 	if (trans->allocating_chunk)
 		return -ENOSPC;
+	/*
+	 * Allocation of system chunks can not happen through this path, as we
+	 * could end up in a deadlock if we are allocating a data or metadata
+	 * chunk and there is another task modifying the chunk btree.
+	 *
+	 * This is because while we are holding the chunk mutex, we will attempt
+	 * to add the new chunk item to the chunk btree or update an existing
+	 * device item in the chunk btree, while the other task that is modifying
+	 * the chunk btree is attempting to COW an extent buffer while holding a
+	 * lock on it and on its parent - if the COW operation triggers a system
+	 * chunk allocation, then we can deadlock because we are holding the
+	 * chunk mutex and we may need to access that extent buffer or its parent
+	 * in order to add the chunk item or update a device item.
+	 *
+	 * Tasks that want to modify the chunk tree should reserve system space
+	 * before updating the chunk btree, by calling either
+	 * btrfs_reserve_chunk_metadata() or check_system_chunk().
+	 * It's possible that after a task reserves the space, it still ends up
+	 * here - this happens in the cases described above at do_chunk_alloc().
+	 * The task will have to either retry or fail.
+	 */
+	if (flags & BTRFS_BLOCK_GROUP_SYSTEM)
+		return -ENOSPC;
 
 	space_info = btrfs_find_space_info(fs_info, flags);
 	ASSERT(space_info);
@@ -3078,6 +3738,7 @@ int btrfs_chunk_alloc(struct btrfs_trans_handle *trans, u64 flags,
 			 * attempt.
 			 */
 			wait_for_alloc = true;
+			force = CHUNK_ALLOC_NO_FORCE;
 			spin_unlock(&space_info->lock);
 			mutex_lock(&fs_info->chunk_mutex);
 			mutex_unlock(&fs_info->chunk_mutex);
@@ -3113,15 +3774,22 @@ int btrfs_chunk_alloc(struct btrfs_trans_handle *trans, u64 flags,
 			force_metadata_allocation(fs_info);
 	}
 
-	/*
-	 * Check if we have enough space in SYSTEM chunk because we may need
-	 * to update devices.
-	 */
-	check_system_chunk(trans, flags);
-
-	ret = btrfs_alloc_chunk(trans, flags);
+	ret_bg = do_chunk_alloc(trans, flags);
 	trans->allocating_chunk = false;
 
+	if (IS_ERR(ret_bg)) {
+		ret = PTR_ERR(ret_bg);
+	} else if (from_extent_allocation) {
+		/*
+		 * New block group is likely to be used soon. Try to activate
+		 * it now. Failure is OK for now.
+		 */
+		btrfs_zone_activate(ret_bg);
+	}
+
+	if (!ret)
+		btrfs_put_block_group(ret_bg);
+
 	spin_lock(&space_info->lock);
 	if (ret < 0) {
 		if (ret == -ENOSPC)
@@ -3138,22 +3806,6 @@ out:
 	space_info->chunk_alloc = 0;
 	spin_unlock(&space_info->lock);
 	mutex_unlock(&fs_info->chunk_mutex);
-	/*
-	 * When we allocate a new chunk we reserve space in the chunk block
-	 * reserve to make sure we can COW nodes/leafs in the chunk tree or
-	 * add new nodes/leafs to it if we end up needing to do it when
-	 * inserting the chunk item and updating device items as part of the
-	 * second phase of chunk allocation, performed by
-	 * btrfs_finish_chunk_alloc(). So make sure we don't accumulate a
-	 * large number of new block groups to create in our transaction
-	 * handle's new_bgs list to avoid exhausting the chunk block reserve
-	 * in extreme cases - like having a single transaction create many new
-	 * block groups when starting to write out the free space caches of all
-	 * the block groups that were made dirty during the lifetime of the
-	 * transaction.
-	 */
-	if (trans->chunk_bytes_reserved >= (u64)SZ_2M)
-		btrfs_create_pending_block_groups(trans);
 
 	return ret;
 }
@@ -3169,17 +3821,14 @@ static u64 get_profile_num_devs(struct btrfs_fs_info *fs_info, u64 type)
 	return num_dev;
 }
 
-/*
- * Reserve space in the system space for allocating or removing a chunk
- */
-void check_system_chunk(struct btrfs_trans_handle *trans, u64 type)
+static void reserve_chunk_space(struct btrfs_trans_handle *trans,
+				u64 bytes,
+				u64 type)
 {
 	struct btrfs_fs_info *fs_info = trans->fs_info;
 	struct btrfs_space_info *info;
 	u64 left;
-	u64 thresh;
 	int ret = 0;
-	u64 num_devs;
 
 	/*
 	 * Needed because we can end up allocating a system chunk and for an
@@ -3192,20 +3841,15 @@ void check_system_chunk(struct btrfs_trans_handle *trans, u64 type)
 	left = info->total_bytes - btrfs_space_info_used(info, true);
 	spin_unlock(&info->lock);
 
-	num_devs = get_profile_num_devs(fs_info, type);
-
-	/* num_devs device items to update and 1 chunk item to add or remove */
-	thresh = btrfs_calc_metadata_size(fs_info, num_devs) +
-		btrfs_calc_insert_metadata_size(fs_info, 1);
-
-	if (left < thresh && btrfs_test_opt(fs_info, ENOSPC_DEBUG)) {
+	if (left < bytes && btrfs_test_opt(fs_info, ENOSPC_DEBUG)) {
 		btrfs_info(fs_info, "left=%llu, need=%llu, flags=%llu",
-			   left, thresh, type);
+			   left, bytes, type);
 		btrfs_dump_space_info(fs_info, info, 0, 0);
 	}
 
-	if (left < thresh) {
+	if (left < bytes) {
 		u64 flags = btrfs_system_alloc_profile(fs_info);
+		struct btrfs_block_group *bg;
 
 		/*
 		 * Ignore failure to create system chunk. We might end up not
@@ -3213,50 +3857,111 @@ void check_system_chunk(struct btrfs_trans_handle *trans, u64 type)
 		 * the paths we visit in the chunk tree (they were already COWed
 		 * or created in the current transaction for example).
 		 */
-		ret = btrfs_alloc_chunk(trans, flags);
+		bg = btrfs_create_chunk(trans, flags);
+		if (IS_ERR(bg)) {
+			ret = PTR_ERR(bg);
+		} else {
+			/*
+			 * We have a new chunk. We also need to activate it for
+			 * zoned filesystem.
+			 */
+			ret = btrfs_zoned_activate_one_bg(fs_info, info, true);
+			if (ret < 0)
+				return;
+
+			/*
+			 * If we fail to add the chunk item here, we end up
+			 * trying again at phase 2 of chunk allocation, at
+			 * btrfs_create_pending_block_groups(). So ignore
+			 * any error here. An ENOSPC here could happen, due to
+			 * the cases described at do_chunk_alloc() - the system
+			 * block group we just created was just turned into RO
+			 * mode by a scrub for example, or a running discard
+			 * temporarily removed its free space entries, etc.
+			 */
+			btrfs_chunk_alloc_add_chunk_item(trans, bg);
+		}
 	}
 
 	if (!ret) {
-		ret = btrfs_block_rsv_add(fs_info->chunk_root,
+		ret = btrfs_block_rsv_add(fs_info,
 					  &fs_info->chunk_block_rsv,
-					  thresh, BTRFS_RESERVE_NO_FLUSH);
+					  bytes, BTRFS_RESERVE_NO_FLUSH);
 		if (!ret)
-			trans->chunk_bytes_reserved += thresh;
+			trans->chunk_bytes_reserved += bytes;
 	}
 }
 
+/*
+ * Reserve space in the system space for allocating or removing a chunk.
+ * The caller must be holding fs_info->chunk_mutex.
+ */
+void check_system_chunk(struct btrfs_trans_handle *trans, u64 type)
+{
+	struct btrfs_fs_info *fs_info = trans->fs_info;
+	const u64 num_devs = get_profile_num_devs(fs_info, type);
+	u64 bytes;
+
+	/* num_devs device items to update and 1 chunk item to add or remove. */
+	bytes = btrfs_calc_metadata_size(fs_info, num_devs) +
+		btrfs_calc_insert_metadata_size(fs_info, 1);
+
+	reserve_chunk_space(trans, bytes, type);
+}
+
+/*
+ * Reserve space in the system space, if needed, for doing a modification to the
+ * chunk btree.
+ *
+ * @trans:		A transaction handle.
+ * @is_item_insertion:	Indicate if the modification is for inserting a new item
+ *			in the chunk btree or if it's for the deletion or update
+ *			of an existing item.
+ *
+ * This is used in a context where we need to update the chunk btree outside
+ * block group allocation and removal, to avoid a deadlock with a concurrent
+ * task that is allocating a metadata or data block group and therefore needs to
+ * update the chunk btree while holding the chunk mutex. After the update to the
+ * chunk btree is done, btrfs_trans_release_chunk_metadata() should be called.
+ *
+ */
+void btrfs_reserve_chunk_metadata(struct btrfs_trans_handle *trans,
+				  bool is_item_insertion)
+{
+	struct btrfs_fs_info *fs_info = trans->fs_info;
+	u64 bytes;
+
+	if (is_item_insertion)
+		bytes = btrfs_calc_insert_metadata_size(fs_info, 1);
+	else
+		bytes = btrfs_calc_metadata_size(fs_info, 1);
+
+	mutex_lock(&fs_info->chunk_mutex);
+	reserve_chunk_space(trans, bytes, BTRFS_BLOCK_GROUP_SYSTEM);
+	mutex_unlock(&fs_info->chunk_mutex);
+}
+
 void btrfs_put_block_group_cache(struct btrfs_fs_info *info)
 {
 	struct btrfs_block_group *block_group;
-	u64 last = 0;
 
-	while (1) {
-		struct inode *inode;
+	block_group = btrfs_lookup_first_block_group(info, 0);
+	while (block_group) {
+		btrfs_wait_block_group_cache_done(block_group);
+		spin_lock(&block_group->lock);
+		if (test_and_clear_bit(BLOCK_GROUP_FLAG_IREF,
+				       &block_group->runtime_flags)) {
+			struct inode *inode = block_group->inode;
 
-		block_group = btrfs_lookup_first_block_group(info, last);
-		while (block_group) {
-			btrfs_wait_block_group_cache_done(block_group);
-			spin_lock(&block_group->lock);
-			if (block_group->iref)
-				break;
+			block_group->inode = NULL;
 			spin_unlock(&block_group->lock);
-			block_group = btrfs_next_block_group(block_group);
-		}
-		if (!block_group) {
-			if (last == 0)
-				break;
-			last = 0;
-			continue;
-		}
 
-		inode = block_group->inode;
-		block_group->iref = 0;
-		block_group->inode = NULL;
-		spin_unlock(&block_group->lock);
-		ASSERT(block_group->io_ctl.inode == NULL);
-		iput(inode);
-		last = block_group->start + block_group->length;
-		btrfs_put_block_group(block_group);
+			ASSERT(block_group->io_ctl.inode == NULL);
+			iput(inode);
+		} else {
+			spin_unlock(&block_group->lock);
+		}
+		block_group = btrfs_next_block_group(block_group);
 	}
 }
 
@@ -3272,14 +3977,14 @@ int btrfs_free_block_groups(struct btrfs_fs_info *info)
 	struct btrfs_caching_control *caching_ctl;
 	struct rb_node *n;
 
-	down_write(&info->commit_root_sem);
+	write_lock(&info->block_group_cache_lock);
 	while (!list_empty(&info->caching_block_groups)) {
 		caching_ctl = list_entry(info->caching_block_groups.next,
 					 struct btrfs_caching_control, list);
 		list_del(&caching_ctl->list);
 		btrfs_put_caching_control(caching_ctl);
 	}
-	up_write(&info->commit_root_sem);
+	write_unlock(&info->block_group_cache_lock);
 
 	spin_lock(&info->unused_bgs_lock);
 	while (!list_empty(&info->unused_bgs)) {
@@ -3289,16 +3994,34 @@ int btrfs_free_block_groups(struct btrfs_fs_info *info)
 		list_del_init(&block_group->bg_list);
 		btrfs_put_block_group(block_group);
 	}
+
+	while (!list_empty(&info->reclaim_bgs)) {
+		block_group = list_first_entry(&info->reclaim_bgs,
+					       struct btrfs_block_group,
+					       bg_list);
+		list_del_init(&block_group->bg_list);
+		btrfs_put_block_group(block_group);
+	}
 	spin_unlock(&info->unused_bgs_lock);
 
-	spin_lock(&info->block_group_cache_lock);
-	while ((n = rb_last(&info->block_group_cache_tree)) != NULL) {
+	spin_lock(&info->zone_active_bgs_lock);
+	while (!list_empty(&info->zone_active_bgs)) {
+		block_group = list_first_entry(&info->zone_active_bgs,
+					       struct btrfs_block_group,
+					       active_bg_list);
+		list_del_init(&block_group->active_bg_list);
+		btrfs_put_block_group(block_group);
+	}
+	spin_unlock(&info->zone_active_bgs_lock);
+
+	write_lock(&info->block_group_cache_lock);
+	while ((n = rb_last(&info->block_group_cache_tree.rb_root)) != NULL) {
 		block_group = rb_entry(n, struct btrfs_block_group,
 				       cache_node);
-		rb_erase(&block_group->cache_node,
-			 &info->block_group_cache_tree);
+		rb_erase_cached(&block_group->cache_node,
+				&info->block_group_cache_tree);
 		RB_CLEAR_NODE(&block_group->cache_node);
-		spin_unlock(&info->block_group_cache_lock);
+		write_unlock(&info->block_group_cache_lock);
 
 		down_write(&block_group->space_info->groups_sem);
 		list_del(&block_group->list);
@@ -3317,20 +4040,13 @@ int btrfs_free_block_groups(struct btrfs_fs_info *info)
 		ASSERT(list_empty(&block_group->dirty_list));
 		ASSERT(list_empty(&block_group->io_list));
 		ASSERT(list_empty(&block_group->bg_list));
-		ASSERT(atomic_read(&block_group->count) == 1);
+		ASSERT(refcount_read(&block_group->refs) == 1);
+		ASSERT(block_group->swap_extents == 0);
 		btrfs_put_block_group(block_group);
 
-		spin_lock(&info->block_group_cache_lock);
+		write_lock(&info->block_group_cache_lock);
 	}
-	spin_unlock(&info->block_group_cache_lock);
-
-	/*
-	 * Now that all the block groups are freed, go through and free all the
-	 * space_info structs.  This is only called during the final stages of
-	 * unmount, and so we know nobody is using them.  We call
-	 * synchronize_rcu() once before we start, just to be on the safe side.
-	 */
-	synchronize_rcu();
+	write_unlock(&info->block_group_cache_lock);
 
 	btrfs_release_global_block_rsv(info);
 
@@ -3344,11 +4060,87 @@ int btrfs_free_block_groups(struct btrfs_fs_info *info)
 		 * important and indicates a real bug if this happens.
 		 */
 		if (WARN_ON(space_info->bytes_pinned > 0 ||
-			    space_info->bytes_reserved > 0 ||
 			    space_info->bytes_may_use > 0))
 			btrfs_dump_space_info(info, space_info, 0, 0);
+
+		/*
+		 * If there was a failure to cleanup a log tree, very likely due
+		 * to an IO failure on a writeback attempt of one or more of its
+		 * extent buffers, we could not do proper (and cheap) unaccounting
+		 * of their reserved space, so don't warn on bytes_reserved > 0 in
+		 * that case.
+		 */
+		if (!(space_info->flags & BTRFS_BLOCK_GROUP_METADATA) ||
+		    !BTRFS_FS_LOG_CLEANUP_ERROR(info)) {
+			if (WARN_ON(space_info->bytes_reserved > 0))
+				btrfs_dump_space_info(info, space_info, 0, 0);
+		}
+
+		WARN_ON(space_info->reclaim_size > 0);
 		list_del(&space_info->list);
 		btrfs_sysfs_remove_space_info(space_info);
 	}
 	return 0;
 }
+
+void btrfs_freeze_block_group(struct btrfs_block_group *cache)
+{
+	atomic_inc(&cache->frozen);
+}
+
+void btrfs_unfreeze_block_group(struct btrfs_block_group *block_group)
+{
+	struct btrfs_fs_info *fs_info = block_group->fs_info;
+	struct extent_map_tree *em_tree;
+	struct extent_map *em;
+	bool cleanup;
+
+	spin_lock(&block_group->lock);
+	cleanup = (atomic_dec_and_test(&block_group->frozen) &&
+		   test_bit(BLOCK_GROUP_FLAG_REMOVED, &block_group->runtime_flags));
+	spin_unlock(&block_group->lock);
+
+	if (cleanup) {
+		em_tree = &fs_info->mapping_tree;
+		write_lock(&em_tree->lock);
+		em = lookup_extent_mapping(em_tree, block_group->start,
+					   1);
+		BUG_ON(!em); /* logic error, can't happen */
+		remove_extent_mapping(em_tree, em);
+		write_unlock(&em_tree->lock);
+
+		/* once for us and once for the tree */
+		free_extent_map(em);
+		free_extent_map(em);
+
+		/*
+		 * We may have left one free space entry and other possible
+		 * tasks trimming this block group have left 1 entry each one.
+		 * Free them if any.
+		 */
+		btrfs_remove_free_space_cache(block_group);
+	}
+}
+
+bool btrfs_inc_block_group_swap_extents(struct btrfs_block_group *bg)
+{
+	bool ret = true;
+
+	spin_lock(&bg->lock);
+	if (bg->ro)
+		ret = false;
+	else
+		bg->swap_extents++;
+	spin_unlock(&bg->lock);
+
+	return ret;
+}
+
+void btrfs_dec_block_group_swap_extents(struct btrfs_block_group *bg, int amount)
+{
+	spin_lock(&bg->lock);
+	ASSERT(!bg->ro);
+	ASSERT(bg->swap_extents >= amount);
+	bg->swap_extents -= amount;
+	spin_unlock(&bg->lock);
+}
diff --git a/fs/btrfs/block-group.h b/fs/btrfs/block-group.h
index 107bb557ca8d..8fb14b99a1d1 100644
--- a/fs/btrfs/block-group.h
+++ b/fs/btrfs/block-group.h
@@ -35,11 +35,33 @@ enum btrfs_discard_state {
  * the FS with empty chunks
  *
  * CHUNK_ALLOC_FORCE means it must try to allocate one
+ *
+ * CHUNK_ALLOC_FORCE_FOR_EXTENT like CHUNK_ALLOC_FORCE but called from
+ * find_free_extent() that also activaes the zone
  */
 enum btrfs_chunk_alloc_enum {
 	CHUNK_ALLOC_NO_FORCE,
 	CHUNK_ALLOC_LIMITED,
 	CHUNK_ALLOC_FORCE,
+	CHUNK_ALLOC_FORCE_FOR_EXTENT,
+};
+
+/* Block group flags set at runtime */
+enum btrfs_block_group_flags {
+	BLOCK_GROUP_FLAG_IREF,
+	BLOCK_GROUP_FLAG_REMOVED,
+	BLOCK_GROUP_FLAG_TO_COPY,
+	BLOCK_GROUP_FLAG_RELOCATING_REPAIR,
+	BLOCK_GROUP_FLAG_CHUNK_ITEM_INSERTED,
+	BLOCK_GROUP_FLAG_ZONE_IS_ACTIVE,
+	BLOCK_GROUP_FLAG_ZONED_DATA_RELOC,
+};
+
+enum btrfs_caching_type {
+	BTRFS_CACHE_NO,
+	BTRFS_CACHE_STARTED,
+	BTRFS_CACHE_FINISHED,
+	BTRFS_CACHE_ERROR,
 };
 
 struct btrfs_caching_control {
@@ -48,13 +70,20 @@ struct btrfs_caching_control {
 	wait_queue_head_t wait;
 	struct btrfs_work work;
 	struct btrfs_block_group *block_group;
-	u64 progress;
 	refcount_t count;
 };
 
 /* Once caching_thread() finds this much free space, it will wake up waiters. */
 #define CACHING_CTL_WAKE_UP SZ_2M
 
+/*
+ * Tree to record all locked full stripes of a RAID5/6 block group
+ */
+struct btrfs_full_stripe_locks_tree {
+	struct rb_root root;
+	struct mutex lock;
+};
+
 struct btrfs_block_group {
 	struct btrfs_fs_info *fs_info;
 	struct inode *inode;
@@ -68,6 +97,7 @@ struct btrfs_block_group {
 	u64 bytes_super;
 	u64 flags;
 	u64 cache_generation;
+	u64 global_root_id;
 
 	/*
 	 * If the free space extent count exceeds this number, convert the block
@@ -90,18 +120,15 @@ struct btrfs_block_group {
 
 	/* For raid56, this is a full stripe, without parity */
 	unsigned long full_stripe_len;
+	unsigned long runtime_flags;
 
 	unsigned int ro;
-	unsigned int iref:1;
-	unsigned int has_caching_ctl:1;
-	unsigned int removed:1;
 
 	int disk_cache_state;
 
 	/* Cache tracking stuff */
 	int cached;
 	struct btrfs_caching_control *caching_ctl;
-	u64 last_byte_to_unpin;
 
 	struct btrfs_space_info *space_info;
 
@@ -114,8 +141,7 @@ struct btrfs_block_group {
 	/* For block groups in the same raid type */
 	struct list_head list;
 
-	/* Usage count */
-	atomic_t count;
+	refcount_t refs;
 
 	/*
 	 * List of struct btrfs_free_clusters for this block group.
@@ -129,8 +155,17 @@ struct btrfs_block_group {
 	/* For read-only block groups */
 	struct list_head ro_list;
 
+	/*
+	 * When non-zero it means the block group's logical address and its
+	 * device extents can not be reused for future block group allocations
+	 * until the counter goes down to 0. This is to prevent them from being
+	 * reused while some task is still using the block group after it was
+	 * deleted - we want to make sure they can only be reused for new block
+	 * groups after that task is done with the deleted block group.
+	 */
+	atomic_t frozen;
+
 	/* For discard operations */
-	atomic_t trimming;
 	struct list_head discard_list;
 	int discard_index;
 	u64 discard_eligible_time;
@@ -173,8 +208,30 @@ struct btrfs_block_group {
 	 */
 	int needs_free_space;
 
+	/* Flag indicating this block group is placed on a sequential zone */
+	bool seq_zone;
+
+	/*
+	 * Number of extents in this block group used for swap files.
+	 * All accesses protected by the spinlock 'lock'.
+	 */
+	int swap_extents;
+
 	/* Record locked full stripes for RAID5/6 block group */
 	struct btrfs_full_stripe_locks_tree full_stripe_locks_root;
+
+	/*
+	 * Allocation offset for the block group to implement sequential
+	 * allocation. This is used only on a zoned filesystem.
+	 */
+	u64 alloc_offset;
+	u64 zone_unusable;
+	u64 zone_capacity;
+	u64 meta_write_pointer;
+	struct map_lookup *physical_map;
+	struct list_head active_bg_list;
+	struct work_struct zone_finish_work;
+	struct extent_buffer *last_eb;
 };
 
 static inline u64 btrfs_block_group_end(struct btrfs_block_group *block_group)
@@ -217,14 +274,13 @@ void btrfs_put_block_group(struct btrfs_block_group *cache);
 void btrfs_dec_block_group_reservations(struct btrfs_fs_info *fs_info,
 					const u64 start);
 void btrfs_wait_block_group_reservations(struct btrfs_block_group *bg);
-bool btrfs_inc_nocow_writers(struct btrfs_fs_info *fs_info, u64 bytenr);
-void btrfs_dec_nocow_writers(struct btrfs_fs_info *fs_info, u64 bytenr);
+struct btrfs_block_group *btrfs_inc_nocow_writers(struct btrfs_fs_info *fs_info,
+						  u64 bytenr);
+void btrfs_dec_nocow_writers(struct btrfs_block_group *bg);
 void btrfs_wait_nocow_writers(struct btrfs_block_group *bg);
 void btrfs_wait_block_group_cache_progress(struct btrfs_block_group *cache,
 				           u64 num_bytes);
-int btrfs_wait_block_group_cache_done(struct btrfs_block_group *cache);
-int btrfs_cache_block_group(struct btrfs_block_group *cache,
-			    int load_cache_only);
+int btrfs_cache_block_group(struct btrfs_block_group *cache, bool wait);
 void btrfs_put_caching_control(struct btrfs_caching_control *ctl);
 struct btrfs_caching_control *btrfs_get_caching_control(
 		struct btrfs_block_group *cache);
@@ -237,9 +293,13 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
 			     u64 group_start, struct extent_map *em);
 void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info);
 void btrfs_mark_bg_unused(struct btrfs_block_group *bg);
+void btrfs_reclaim_bgs_work(struct work_struct *work);
+void btrfs_reclaim_bgs(struct btrfs_fs_info *fs_info);
+void btrfs_mark_bg_to_reclaim(struct btrfs_block_group *bg);
 int btrfs_read_block_groups(struct btrfs_fs_info *info);
-int btrfs_make_block_group(struct btrfs_trans_handle *trans, u64 bytes_used,
-			   u64 type, u64 chunk_offset, u64 size);
+struct btrfs_block_group *btrfs_make_block_group(struct btrfs_trans_handle *trans,
+						 u64 bytes_used, u64 type,
+						 u64 chunk_offset, u64 size);
 void btrfs_create_pending_block_groups(struct btrfs_trans_handle *trans);
 int btrfs_inc_block_group_ro(struct btrfs_block_group *cache,
 			     bool do_chunk_alloc);
@@ -248,7 +308,7 @@ int btrfs_start_dirty_block_groups(struct btrfs_trans_handle *trans);
 int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans);
 int btrfs_setup_space_cache(struct btrfs_trans_handle *trans);
 int btrfs_update_block_group(struct btrfs_trans_handle *trans,
-			     u64 bytenr, u64 num_bytes, int alloc);
+			     u64 bytenr, u64 num_bytes, bool alloc);
 int btrfs_add_reserved_bytes(struct btrfs_block_group *cache,
 			     u64 ram_bytes, u64 num_bytes, int delalloc);
 void btrfs_free_reserved_bytes(struct btrfs_block_group *cache,
@@ -257,9 +317,14 @@ int btrfs_chunk_alloc(struct btrfs_trans_handle *trans, u64 flags,
 		      enum btrfs_chunk_alloc_enum force);
 int btrfs_force_chunk_alloc(struct btrfs_trans_handle *trans, u64 type);
 void check_system_chunk(struct btrfs_trans_handle *trans, const u64 type);
+void btrfs_reserve_chunk_metadata(struct btrfs_trans_handle *trans,
+				  bool is_item_insertion);
 u64 btrfs_get_alloc_profile(struct btrfs_fs_info *fs_info, u64 orig_flags);
 void btrfs_put_block_group_cache(struct btrfs_fs_info *info);
 int btrfs_free_block_groups(struct btrfs_fs_info *info);
+int btrfs_rmap_block(struct btrfs_fs_info *fs_info, u64 chunk_start,
+		       struct block_device *bdev, u64 physical, u64 **logical,
+		       int *naddrs, int *stripe_len);
 
 static inline u64 btrfs_data_alloc_profile(struct btrfs_fs_info *fs_info)
 {
@@ -283,9 +348,10 @@ static inline int btrfs_block_group_done(struct btrfs_block_group *cache)
 		cache->cached == BTRFS_CACHE_ERROR;
 }
 
-#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
-int btrfs_rmap_block(struct btrfs_fs_info *fs_info, u64 chunk_start,
-		     u64 physical, u64 **logical, int *naddrs, int *stripe_len);
-#endif
+void btrfs_freeze_block_group(struct btrfs_block_group *cache);
+void btrfs_unfreeze_block_group(struct btrfs_block_group *cache);
+
+bool btrfs_inc_block_group_swap_extents(struct btrfs_block_group *bg);
+void btrfs_dec_block_group_swap_extents(struct btrfs_block_group *bg, int amount);
 
 #endif /* BTRFS_BLOCK_GROUP_H */
diff --git a/fs/btrfs/block-rsv.c b/fs/btrfs/block-rsv.c
index d07bd41a7c1e..ec96285357e0 100644
--- a/fs/btrfs/block-rsv.c
+++ b/fs/btrfs/block-rsv.c
@@ -5,6 +5,100 @@
 #include "block-rsv.h"
 #include "space-info.h"
 #include "transaction.h"
+#include "block-group.h"
+#include "disk-io.h"
+
+/*
+ * HOW DO BLOCK RESERVES WORK
+ *
+ *   Think of block_rsv's as buckets for logically grouped metadata
+ *   reservations.  Each block_rsv has a ->size and a ->reserved.  ->size is
+ *   how large we want our block rsv to be, ->reserved is how much space is
+ *   currently reserved for this block reserve.
+ *
+ *   ->failfast exists for the truncate case, and is described below.
+ *
+ * NORMAL OPERATION
+ *
+ *   -> Reserve
+ *     Entrance: btrfs_block_rsv_add, btrfs_block_rsv_refill
+ *
+ *     We call into btrfs_reserve_metadata_bytes() with our bytes, which is
+ *     accounted for in space_info->bytes_may_use, and then add the bytes to
+ *     ->reserved, and ->size in the case of btrfs_block_rsv_add.
+ *
+ *     ->size is an over-estimation of how much we may use for a particular
+ *     operation.
+ *
+ *   -> Use
+ *     Entrance: btrfs_use_block_rsv
+ *
+ *     When we do a btrfs_alloc_tree_block() we call into btrfs_use_block_rsv()
+ *     to determine the appropriate block_rsv to use, and then verify that
+ *     ->reserved has enough space for our tree block allocation.  Once
+ *     successful we subtract fs_info->nodesize from ->reserved.
+ *
+ *   -> Finish
+ *     Entrance: btrfs_block_rsv_release
+ *
+ *     We are finished with our operation, subtract our individual reservation
+ *     from ->size, and then subtract ->size from ->reserved and free up the
+ *     excess if there is any.
+ *
+ *     There is some logic here to refill the delayed refs rsv or the global rsv
+ *     as needed, otherwise the excess is subtracted from
+ *     space_info->bytes_may_use.
+ *
+ * TYPES OF BLOCK RESERVES
+ *
+ * BLOCK_RSV_TRANS, BLOCK_RSV_DELOPS, BLOCK_RSV_CHUNK
+ *   These behave normally, as described above, just within the confines of the
+ *   lifetime of their particular operation (transaction for the whole trans
+ *   handle lifetime, for example).
+ *
+ * BLOCK_RSV_GLOBAL
+ *   It is impossible to properly account for all the space that may be required
+ *   to make our extent tree updates.  This block reserve acts as an overflow
+ *   buffer in case our delayed refs reserve does not reserve enough space to
+ *   update the extent tree.
+ *
+ *   We can steal from this in some cases as well, notably on evict() or
+ *   truncate() in order to help users recover from ENOSPC conditions.
+ *
+ * BLOCK_RSV_DELALLOC
+ *   The individual item sizes are determined by the per-inode size
+ *   calculations, which are described with the delalloc code.  This is pretty
+ *   straightforward, it's just the calculation of ->size encodes a lot of
+ *   different items, and thus it gets used when updating inodes, inserting file
+ *   extents, and inserting checksums.
+ *
+ * BLOCK_RSV_DELREFS
+ *   We keep a running tally of how many delayed refs we have on the system.
+ *   We assume each one of these delayed refs are going to use a full
+ *   reservation.  We use the transaction items and pre-reserve space for every
+ *   operation, and use this reservation to refill any gap between ->size and
+ *   ->reserved that may exist.
+ *
+ *   From there it's straightforward, removing a delayed ref means we remove its
+ *   count from ->size and free up reservations as necessary.  Since this is
+ *   the most dynamic block reserve in the system, we will try to refill this
+ *   block reserve first with any excess returned by any other block reserve.
+ *
+ * BLOCK_RSV_EMPTY
+ *   This is the fallback block reserve to make us try to reserve space if we
+ *   don't have a specific bucket for this allocation.  It is mostly used for
+ *   updating the device tree and such, since that is a separate pool we're
+ *   content to just reserve space from the space_info on demand.
+ *
+ * BLOCK_RSV_TEMP
+ *   This is used by things like truncate and iput.  We will temporarily
+ *   allocate a block reserve, set it to some size, and then truncate bytes
+ *   until we have no space left.  With ->failfast set we'll simply return
+ *   ENOSPC from btrfs_use_block_rsv() to signal that we need to unwind and try
+ *   to make a new reservation.  This is because these operations are
+ *   unbounded, so we want to do as much work as we can, and then back off and
+ *   re-reserve.
+ */
 
 static u64 block_rsv_release_bytes(struct btrfs_fs_info *fs_info,
 				    struct btrfs_block_rsv *block_rsv,
@@ -24,7 +118,7 @@ static u64 block_rsv_release_bytes(struct btrfs_fs_info *fs_info,
 	if (block_rsv->reserved >= block_rsv->size) {
 		num_bytes = block_rsv->reserved - block_rsv->size;
 		block_rsv->reserved = block_rsv->size;
-		block_rsv->full = 1;
+		block_rsv->full = true;
 	} else {
 		num_bytes = 0;
 	}
@@ -48,7 +142,7 @@ static u64 block_rsv_release_bytes(struct btrfs_fs_info *fs_info,
 				bytes_to_add = min(num_bytes, bytes_to_add);
 				dest->reserved += bytes_to_add;
 				if (dest->reserved >= dest->size)
-					dest->full = 1;
+					dest->full = true;
 				num_bytes -= bytes_to_add;
 			}
 			spin_unlock(&dest->lock);
@@ -77,7 +171,7 @@ int btrfs_block_rsv_migrate(struct btrfs_block_rsv *src,
 	return 0;
 }
 
-void btrfs_init_block_rsv(struct btrfs_block_rsv *rsv, unsigned short type)
+void btrfs_init_block_rsv(struct btrfs_block_rsv *rsv, enum btrfs_rsv_type type)
 {
 	memset(rsv, 0, sizeof(*rsv));
 	spin_lock_init(&rsv->lock);
@@ -86,7 +180,7 @@ void btrfs_init_block_rsv(struct btrfs_block_rsv *rsv, unsigned short type)
 
 void btrfs_init_metadata_block_rsv(struct btrfs_fs_info *fs_info,
 				   struct btrfs_block_rsv *rsv,
-				   unsigned short type)
+				   enum btrfs_rsv_type type)
 {
 	btrfs_init_block_rsv(rsv, type);
 	rsv->space_info = btrfs_find_space_info(fs_info,
@@ -94,7 +188,7 @@ void btrfs_init_metadata_block_rsv(struct btrfs_fs_info *fs_info,
 }
 
 struct btrfs_block_rsv *btrfs_alloc_block_rsv(struct btrfs_fs_info *fs_info,
-					      unsigned short type)
+					      enum btrfs_rsv_type type)
 {
 	struct btrfs_block_rsv *block_rsv;
 
@@ -111,11 +205,11 @@ void btrfs_free_block_rsv(struct btrfs_fs_info *fs_info,
 {
 	if (!rsv)
 		return;
-	btrfs_block_rsv_release(fs_info, rsv, (u64)-1);
+	btrfs_block_rsv_release(fs_info, rsv, (u64)-1, NULL);
 	kfree(rsv);
 }
 
-int btrfs_block_rsv_add(struct btrfs_root *root,
+int btrfs_block_rsv_add(struct btrfs_fs_info *fs_info,
 			struct btrfs_block_rsv *block_rsv, u64 num_bytes,
 			enum btrfs_reserve_flush_enum flush)
 {
@@ -124,7 +218,7 @@ int btrfs_block_rsv_add(struct btrfs_root *root,
 	if (num_bytes == 0)
 		return 0;
 
-	ret = btrfs_reserve_metadata_bytes(root, block_rsv, num_bytes, flush);
+	ret = btrfs_reserve_metadata_bytes(fs_info, block_rsv, num_bytes, flush);
 	if (!ret)
 		btrfs_block_rsv_add_bytes(block_rsv, num_bytes, true);
 
@@ -148,7 +242,7 @@ int btrfs_block_rsv_check(struct btrfs_block_rsv *block_rsv, int min_factor)
 	return ret;
 }
 
-int btrfs_block_rsv_refill(struct btrfs_root *root,
+int btrfs_block_rsv_refill(struct btrfs_fs_info *fs_info,
 			   struct btrfs_block_rsv *block_rsv, u64 min_reserved,
 			   enum btrfs_reserve_flush_enum flush)
 {
@@ -169,7 +263,7 @@ int btrfs_block_rsv_refill(struct btrfs_root *root,
 	if (!ret)
 		return 0;
 
-	ret = btrfs_reserve_metadata_bytes(root, block_rsv, num_bytes, flush);
+	ret = btrfs_reserve_metadata_bytes(fs_info, block_rsv, num_bytes, flush);
 	if (!ret) {
 		btrfs_block_rsv_add_bytes(block_rsv, num_bytes, false);
 		return 0;
@@ -178,9 +272,9 @@ int btrfs_block_rsv_refill(struct btrfs_root *root,
 	return ret;
 }
 
-u64 __btrfs_block_rsv_release(struct btrfs_fs_info *fs_info,
-			      struct btrfs_block_rsv *block_rsv,
-			      u64 num_bytes, u64 *qgroup_to_release)
+u64 btrfs_block_rsv_release(struct btrfs_fs_info *fs_info,
+			    struct btrfs_block_rsv *block_rsv, u64 num_bytes,
+			    u64 *qgroup_to_release)
 {
 	struct btrfs_block_rsv *global_rsv = &fs_info->global_block_rsv;
 	struct btrfs_block_rsv *delayed_rsv = &fs_info->delayed_refs_rsv;
@@ -192,7 +286,7 @@ u64 __btrfs_block_rsv_release(struct btrfs_fs_info *fs_info,
 	 */
 	if (block_rsv == delayed_rsv)
 		target = global_rsv;
-	else if (block_rsv != global_rsv && !delayed_rsv->full)
+	else if (block_rsv != global_rsv && !btrfs_block_rsv_full(delayed_rsv))
 		target = delayed_rsv;
 
 	if (target && block_rsv->space_info != target->space_info)
@@ -210,7 +304,7 @@ int btrfs_block_rsv_use_bytes(struct btrfs_block_rsv *block_rsv, u64 num_bytes)
 	if (block_rsv->reserved >= num_bytes) {
 		block_rsv->reserved -= num_bytes;
 		if (block_rsv->reserved < block_rsv->size)
-			block_rsv->full = 0;
+			block_rsv->full = false;
 		ret = 0;
 	}
 	spin_unlock(&block_rsv->lock);
@@ -225,7 +319,7 @@ void btrfs_block_rsv_add_bytes(struct btrfs_block_rsv *block_rsv,
 	if (update_size)
 		block_rsv->size += num_bytes;
 	else if (block_rsv->reserved >= block_rsv->size)
-		block_rsv->full = 1;
+		block_rsv->full = true;
 	spin_unlock(&block_rsv->lock);
 }
 
@@ -247,7 +341,7 @@ int btrfs_cond_migrate_bytes(struct btrfs_fs_info *fs_info,
 	}
 	global_rsv->reserved -= num_bytes;
 	if (global_rsv->reserved < global_rsv->size)
-		global_rsv->full = 0;
+		global_rsv->full = false;
 	spin_unlock(&global_rsv->lock);
 
 	btrfs_block_rsv_add_bytes(dest, num_bytes, true);
@@ -258,23 +352,29 @@ void btrfs_update_global_block_rsv(struct btrfs_fs_info *fs_info)
 {
 	struct btrfs_block_rsv *block_rsv = &fs_info->global_block_rsv;
 	struct btrfs_space_info *sinfo = block_rsv->space_info;
-	u64 num_bytes;
-	unsigned min_items;
+	struct btrfs_root *root, *tmp;
+	u64 num_bytes = btrfs_root_used(&fs_info->tree_root->root_item);
+	unsigned int min_items = 1;
 
 	/*
 	 * The global block rsv is based on the size of the extent tree, the
 	 * checksum tree and the root tree.  If the fs is empty we want to set
 	 * it to a minimal amount for safety.
+	 *
+	 * We also are going to need to modify the minimum of the tree root and
+	 * any global roots we could touch.
 	 */
-	num_bytes = btrfs_root_used(&fs_info->extent_root->root_item) +
-		btrfs_root_used(&fs_info->csum_root->root_item) +
-		btrfs_root_used(&fs_info->tree_root->root_item);
-
-	/*
-	 * We at a minimum are going to modify the csum root, the tree root, and
-	 * the extent root.
-	 */
-	min_items = 3;
+	read_lock(&fs_info->global_root_lock);
+	rbtree_postorder_for_each_entry_safe(root, tmp, &fs_info->global_root_tree,
+					     rb_node) {
+		if (root->root_key.objectid == BTRFS_EXTENT_TREE_OBJECTID ||
+		    root->root_key.objectid == BTRFS_CSUM_TREE_OBJECTID ||
+		    root->root_key.objectid == BTRFS_FREE_SPACE_TREE_OBJECTID) {
+			num_bytes += btrfs_root_used(&root->root_item);
+			min_items++;
+		}
+	}
+	read_unlock(&fs_info->global_root_lock);
 
 	/*
 	 * But we also want to reserve enough space so we can do the fallback
@@ -297,9 +397,9 @@ void btrfs_update_global_block_rsv(struct btrfs_fs_info *fs_info)
 
 	if (block_rsv->reserved < block_rsv->size) {
 		num_bytes = block_rsv->size - block_rsv->reserved;
-		block_rsv->reserved += num_bytes;
 		btrfs_space_info_update_bytes_may_use(fs_info, sinfo,
 						      num_bytes);
+		block_rsv->reserved = block_rsv->size;
 	} else if (block_rsv->reserved > block_rsv->size) {
 		num_bytes = block_rsv->reserved - block_rsv->size;
 		btrfs_space_info_update_bytes_may_use(fs_info, sinfo,
@@ -308,15 +408,39 @@ void btrfs_update_global_block_rsv(struct btrfs_fs_info *fs_info)
 		btrfs_try_granting_tickets(fs_info, sinfo);
 	}
 
-	if (block_rsv->reserved == block_rsv->size)
-		block_rsv->full = 1;
-	else
-		block_rsv->full = 0;
+	block_rsv->full = (block_rsv->reserved == block_rsv->size);
 
+	if (block_rsv->size >= sinfo->total_bytes)
+		sinfo->force_alloc = CHUNK_ALLOC_FORCE;
 	spin_unlock(&block_rsv->lock);
 	spin_unlock(&sinfo->lock);
 }
 
+void btrfs_init_root_block_rsv(struct btrfs_root *root)
+{
+	struct btrfs_fs_info *fs_info = root->fs_info;
+
+	switch (root->root_key.objectid) {
+	case BTRFS_CSUM_TREE_OBJECTID:
+	case BTRFS_EXTENT_TREE_OBJECTID:
+	case BTRFS_FREE_SPACE_TREE_OBJECTID:
+	case BTRFS_BLOCK_GROUP_TREE_OBJECTID:
+		root->block_rsv = &fs_info->delayed_refs_rsv;
+		break;
+	case BTRFS_ROOT_TREE_OBJECTID:
+	case BTRFS_DEV_TREE_OBJECTID:
+	case BTRFS_QUOTA_TREE_OBJECTID:
+		root->block_rsv = &fs_info->global_block_rsv;
+		break;
+	case BTRFS_CHUNK_TREE_OBJECTID:
+		root->block_rsv = &fs_info->chunk_block_rsv;
+		break;
+	default:
+		root->block_rsv = NULL;
+		break;
+	}
+}
+
 void btrfs_init_global_block_rsv(struct btrfs_fs_info *fs_info)
 {
 	struct btrfs_space_info *space_info;
@@ -331,20 +455,13 @@ void btrfs_init_global_block_rsv(struct btrfs_fs_info *fs_info)
 	fs_info->delayed_block_rsv.space_info = space_info;
 	fs_info->delayed_refs_rsv.space_info = space_info;
 
-	fs_info->extent_root->block_rsv = &fs_info->delayed_refs_rsv;
-	fs_info->csum_root->block_rsv = &fs_info->delayed_refs_rsv;
-	fs_info->dev_root->block_rsv = &fs_info->global_block_rsv;
-	fs_info->tree_root->block_rsv = &fs_info->global_block_rsv;
-	if (fs_info->quota_root)
-		fs_info->quota_root->block_rsv = &fs_info->global_block_rsv;
-	fs_info->chunk_root->block_rsv = &fs_info->chunk_block_rsv;
-
 	btrfs_update_global_block_rsv(fs_info);
 }
 
 void btrfs_release_global_block_rsv(struct btrfs_fs_info *fs_info)
 {
-	btrfs_block_rsv_release(fs_info, &fs_info->global_block_rsv, (u64)-1);
+	btrfs_block_rsv_release(fs_info, &fs_info->global_block_rsv, (u64)-1,
+				NULL);
 	WARN_ON(fs_info->trans_block_rsv.size > 0);
 	WARN_ON(fs_info->trans_block_rsv.reserved > 0);
 	WARN_ON(fs_info->chunk_block_rsv.size > 0);
@@ -362,9 +479,10 @@ static struct btrfs_block_rsv *get_block_rsv(
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct btrfs_block_rsv *block_rsv = NULL;
 
-	if (test_bit(BTRFS_ROOT_REF_COWS, &root->state) ||
-	    (root == fs_info->csum_root && trans->adding_csums) ||
-	    (root == fs_info->uuid_root))
+	if (test_bit(BTRFS_ROOT_SHAREABLE, &root->state) ||
+	    (root == fs_info->uuid_root) ||
+	    (trans->adding_csums &&
+	     root->root_key.objectid == BTRFS_CSUM_TREE_OBJECTID))
 		block_rsv = trans->block_rsv;
 
 	if (!block_rsv)
@@ -415,10 +533,11 @@ again:
 				/*DEFAULT_RATELIMIT_BURST*/ 1);
 		if (__ratelimit(&_rs))
 			WARN(1, KERN_DEBUG
-				"BTRFS: block rsv returned %d\n", ret);
+				"BTRFS: block rsv %d returned %d\n",
+				block_rsv->type, ret);
 	}
 try_reserve:
-	ret = btrfs_reserve_metadata_bytes(root, block_rsv, blocksize,
+	ret = btrfs_reserve_metadata_bytes(fs_info, block_rsv, blocksize,
 					   BTRFS_RESERVE_NO_FLUSH);
 	if (!ret)
 		return block_rsv;
diff --git a/fs/btrfs/block-rsv.h b/fs/btrfs/block-rsv.h
index d1428bb73fc5..578c3497a455 100644
--- a/fs/btrfs/block-rsv.h
+++ b/fs/btrfs/block-rsv.h
@@ -9,7 +9,7 @@ enum btrfs_reserve_flush_enum;
 /*
  * Types of block reserves
  */
-enum {
+enum btrfs_rsv_type {
 	BTRFS_BLOCK_RSV_GLOBAL,
 	BTRFS_BLOCK_RSV_DELALLOC,
 	BTRFS_BLOCK_RSV_TRANS,
@@ -25,9 +25,10 @@ struct btrfs_block_rsv {
 	u64 reserved;
 	struct btrfs_space_info *space_info;
 	spinlock_t lock;
-	unsigned short full;
-	unsigned short type;
-	unsigned short failfast;
+	bool full;
+	bool failfast;
+	/* Block reserve type, one of BTRFS_BLOCK_RSV_* */
+	enum btrfs_rsv_type type:8;
 
 	/*
 	 * Qgroup equivalent for @size @reserved
@@ -49,19 +50,20 @@ struct btrfs_block_rsv {
 	u64 qgroup_rsv_reserved;
 };
 
-void btrfs_init_block_rsv(struct btrfs_block_rsv *rsv, unsigned short type);
+void btrfs_init_block_rsv(struct btrfs_block_rsv *rsv, enum btrfs_rsv_type type);
+void btrfs_init_root_block_rsv(struct btrfs_root *root);
 struct btrfs_block_rsv *btrfs_alloc_block_rsv(struct btrfs_fs_info *fs_info,
-					      unsigned short type);
+					      enum btrfs_rsv_type type);
 void btrfs_init_metadata_block_rsv(struct btrfs_fs_info *fs_info,
 				   struct btrfs_block_rsv *rsv,
-				   unsigned short type);
+				   enum btrfs_rsv_type type);
 void btrfs_free_block_rsv(struct btrfs_fs_info *fs_info,
 			  struct btrfs_block_rsv *rsv);
-int btrfs_block_rsv_add(struct btrfs_root *root,
+int btrfs_block_rsv_add(struct btrfs_fs_info *fs_info,
 			struct btrfs_block_rsv *block_rsv, u64 num_bytes,
 			enum btrfs_reserve_flush_enum flush);
 int btrfs_block_rsv_check(struct btrfs_block_rsv *block_rsv, int min_factor);
-int btrfs_block_rsv_refill(struct btrfs_root *root,
+int btrfs_block_rsv_refill(struct btrfs_fs_info *fs_info,
 			   struct btrfs_block_rsv *block_rsv, u64 min_reserved,
 			   enum btrfs_reserve_flush_enum flush);
 int btrfs_block_rsv_migrate(struct btrfs_block_rsv *src_rsv,
@@ -73,7 +75,7 @@ int btrfs_cond_migrate_bytes(struct btrfs_fs_info *fs_info,
 			     int min_factor);
 void btrfs_block_rsv_add_bytes(struct btrfs_block_rsv *block_rsv,
 			       u64 num_bytes, bool update_size);
-u64 __btrfs_block_rsv_release(struct btrfs_fs_info *fs_info,
+u64 btrfs_block_rsv_release(struct btrfs_fs_info *fs_info,
 			      struct btrfs_block_rsv *block_rsv,
 			      u64 num_bytes, u64 *qgroup_to_release);
 void btrfs_update_global_block_rsv(struct btrfs_fs_info *fs_info);
@@ -82,20 +84,21 @@ void btrfs_release_global_block_rsv(struct btrfs_fs_info *fs_info);
 struct btrfs_block_rsv *btrfs_use_block_rsv(struct btrfs_trans_handle *trans,
 					    struct btrfs_root *root,
 					    u32 blocksize);
-
-static inline void btrfs_block_rsv_release(struct btrfs_fs_info *fs_info,
-					   struct btrfs_block_rsv *block_rsv,
-					   u64 num_bytes)
-{
-	__btrfs_block_rsv_release(fs_info, block_rsv, num_bytes, NULL);
-}
-
 static inline void btrfs_unuse_block_rsv(struct btrfs_fs_info *fs_info,
 					 struct btrfs_block_rsv *block_rsv,
 					 u32 blocksize)
 {
 	btrfs_block_rsv_add_bytes(block_rsv, blocksize, false);
-	btrfs_block_rsv_release(fs_info, block_rsv, 0);
+	btrfs_block_rsv_release(fs_info, block_rsv, 0, NULL);
+}
+
+/*
+ * Fast path to check if the reserve is full, may be carefully used outside of
+ * locks.
+ */
+static inline bool btrfs_block_rsv_full(const struct btrfs_block_rsv *rsv)
+{
+	return data_race(rsv->full);
 }
 
 #endif /* BTRFS_BLOCK_RSV_H */
diff --git a/fs/btrfs/btrfs_inode.h b/fs/btrfs/btrfs_inode.h
index 4e12a477d32e..54c2ccb36b61 100644
--- a/fs/btrfs/btrfs_inode.h
+++ b/fs/btrfs/btrfs_inode.h
@@ -7,12 +7,20 @@
 #define BTRFS_INODE_H
 
 #include <linux/hash.h>
+#include <linux/refcount.h>
 #include "extent_map.h"
 #include "extent_io.h"
 #include "ordered-data.h"
 #include "delayed-inode.h"
 
 /*
+ * Since we search a directory based on f_pos (struct dir_context::pos) we have
+ * to start at 2 since '.' and '..' have f_pos of 0 and 1 respectively, so
+ * everybody else has to start at 2 (see btrfs_real_readdir() and dir_emit_dots()).
+ */
+#define BTRFS_DIR_START_INDEX 2
+
+/*
  * ordered_data_close is set by truncate when a file that used
  * to have good data has been truncated to zero.  When it is set
  * the btrfs file release call will add this inode to the
@@ -20,16 +28,45 @@
  * new data the application may have written before commit.
  */
 enum {
-	BTRFS_INODE_ORDERED_DATA_CLOSE,
+	BTRFS_INODE_FLUSH_ON_CLOSE,
 	BTRFS_INODE_DUMMY,
 	BTRFS_INODE_IN_DEFRAG,
 	BTRFS_INODE_HAS_ASYNC_EXTENT,
+	 /*
+	  * Always set under the VFS' inode lock, otherwise it can cause races
+	  * during fsync (we start as a fast fsync and then end up in a full
+	  * fsync racing with ordered extent completion).
+	  */
 	BTRFS_INODE_NEEDS_FULL_SYNC,
 	BTRFS_INODE_COPY_EVERYTHING,
 	BTRFS_INODE_IN_DELALLOC_LIST,
-	BTRFS_INODE_READDIO_NEED_LOCK,
 	BTRFS_INODE_HAS_PROPS,
 	BTRFS_INODE_SNAPSHOT_FLUSH,
+	/*
+	 * Set and used when logging an inode and it serves to signal that an
+	 * inode does not have xattrs, so subsequent fsyncs can avoid searching
+	 * for xattrs to log. This bit must be cleared whenever a xattr is added
+	 * to an inode.
+	 */
+	BTRFS_INODE_NO_XATTRS,
+	/*
+	 * Set when we are in a context where we need to start a transaction and
+	 * have dirty pages with the respective file range locked. This is to
+	 * ensure that when reserving space for the transaction, if we are low
+	 * on available space and need to flush delalloc, we will not flush
+	 * delalloc for this inode, because that could result in a deadlock (on
+	 * the file range, inode's io_tree).
+	 */
+	BTRFS_INODE_NO_DELALLOC_FLUSH,
+	/*
+	 * Set when we are working on enabling verity for a file. Computing and
+	 * writing the whole Merkle tree can take a while so we want to prevent
+	 * races where two separate tasks attempt to simultaneously start verity
+	 * on the same file.
+	 */
+	BTRFS_INODE_VERITY_IN_PROGRESS,
+	/* Set when this inode is a free space inode. */
+	BTRFS_INODE_FREE_SPACE_INODE,
 };
 
 /* in memory btrfs inode */
@@ -45,7 +82,8 @@ struct btrfs_inode {
 	/*
 	 * Lock for counters and all fields used to determine if the inode is in
 	 * the log or not (last_trans, last_sub_trans, last_log_commit,
-	 * logged_trans).
+	 * logged_trans), to access/update new_delalloc_bytes and to update the
+	 * VFS' inode number of bytes used.
 	 */
 	spinlock_t lock;
 
@@ -58,7 +96,14 @@ struct btrfs_inode {
 	/* special utility tree used to record which mirrors have already been
 	 * tried when checksums fail for a given block
 	 */
-	struct extent_io_tree io_failure_tree;
+	struct rb_root io_failure_tree;
+	spinlock_t io_failure_lock;
+
+	/*
+	 * Keep track of where the inode has extent items mapped in order to
+	 * make sure the i_size adjustments are accurate
+	 */
+	struct extent_io_tree file_extent_tree;
 
 	/* held while logging the inode in tree-log.c */
 	struct mutex log_mutex;
@@ -103,17 +148,26 @@ struct btrfs_inode {
 	/* a local copy of root's last_log_commit */
 	int last_log_commit;
 
-	/* total number of bytes pending delalloc, used by stat to calc the
-	 * real block usage of the file
+	/*
+	 * Total number of bytes pending delalloc, used by stat to calculate the
+	 * real block usage of the file. This is used only for files.
 	 */
 	u64 delalloc_bytes;
 
-	/*
-	 * Total number of bytes pending delalloc that fall within a file
-	 * range that is either a hole or beyond EOF (and no prealloc extent
-	 * exists in the range). This is always <= delalloc_bytes.
-	 */
-	u64 new_delalloc_bytes;
+	union {
+		/*
+		 * Total number of bytes pending delalloc that fall within a file
+		 * range that is either a hole or beyond EOF (and no prealloc extent
+		 * exists in the range). This is always <= delalloc_bytes and this
+		 * is used only for files.
+		 */
+		u64 new_delalloc_bytes;
+		/*
+		 * The offset of the last dir index key that was logged.
+		 * This is used only for directories.
+		 */
+		u64 last_dir_index_offset;
+	};
 
 	/*
 	 * total number of bytes pending defrag, used by stat to check whether
@@ -129,8 +183,9 @@ struct btrfs_inode {
 	u64 disk_i_size;
 
 	/*
-	 * if this is a directory then index_cnt is the counter for the index
-	 * number for new files that are created
+	 * If this is a directory then index_cnt is the counter for the index
+	 * number for new files that are created. For an empty directory, this
+	 * must be initialized to BTRFS_DIR_START_INDEX.
 	 */
 	u64 index_cnt;
 
@@ -145,13 +200,26 @@ struct btrfs_inode {
 	u64 last_unlink_trans;
 
 	/*
+	 * The id/generation of the last transaction where this inode was
+	 * either the source or the destination of a clone/dedupe operation.
+	 * Used when logging an inode to know if there are shared extents that
+	 * need special care when logging checksum items, to avoid duplicate
+	 * checksum items in a log (which can lead to a corruption where we end
+	 * up with missing checksum ranges after log replay).
+	 * Protected by the vfs inode lock.
+	 */
+	u64 last_reflink_trans;
+
+	/*
 	 * Number of bytes outstanding that are going to need csums.  This is
 	 * used in ENOSPC accounting.
 	 */
 	u64 csum_bytes;
 
-	/* flags field from the on disk inode */
+	/* Backwards incompatible flags, lower half of inode_item::flags  */
 	u32 flags;
+	/* Read-only compatibility flags, upper half of inode_item::flags */
+	u32 ro_flags;
 
 	/*
 	 * Counters to keep track of the number of extent item's we may use due
@@ -181,16 +249,7 @@ struct btrfs_inode {
 	/* Hook into fs_info->delayed_iputs */
 	struct list_head delayed_iput;
 
-	/*
-	 * To avoid races between lockless (i_mutex not held) direct IO writes
-	 * and concurrent fsync requests. Direct IO writes must acquire read
-	 * access on this semaphore for creating an extent map and its
-	 * corresponding ordered extent. The fast fsync path must acquire write
-	 * access on this semaphore before it collects ordered extents and
-	 * extent maps.
-	 */
-	struct rw_semaphore dio_sem;
-
+	struct rw_semaphore i_mmap_lock;
 	struct inode vfs_inode;
 };
 
@@ -211,26 +270,31 @@ static inline unsigned long btrfs_inode_hash(u64 objectid,
 	return (unsigned long)h;
 }
 
-static inline void btrfs_insert_inode_hash(struct inode *inode)
-{
-	unsigned long h = btrfs_inode_hash(inode->i_ino, BTRFS_I(inode)->root);
-
-	__insert_inode_hash(inode, h);
-}
+#if BITS_PER_LONG == 32
 
+/*
+ * On 32 bit systems the i_ino of struct inode is 32 bits (unsigned long), so
+ * we use the inode's location objectid which is a u64 to avoid truncation.
+ */
 static inline u64 btrfs_ino(const struct btrfs_inode *inode)
 {
 	u64 ino = inode->location.objectid;
 
-	/*
-	 * !ino: btree_inode
-	 * type == BTRFS_ROOT_ITEM_KEY: subvol dir
-	 */
-	if (!ino || inode->location.type == BTRFS_ROOT_ITEM_KEY)
+	/* type == BTRFS_ROOT_ITEM_KEY: subvol dir */
+	if (inode->location.type == BTRFS_ROOT_ITEM_KEY)
 		ino = inode->vfs_inode.i_ino;
 	return ino;
 }
 
+#else
+
+static inline u64 btrfs_ino(const struct btrfs_inode *inode)
+{
+	return inode->vfs_inode.i_ino;
+}
+
+#endif
+
 static inline void btrfs_i_size_write(struct btrfs_inode *inode, u64 size)
 {
 	i_size_write(&inode->vfs_inode, size);
@@ -239,14 +303,7 @@ static inline void btrfs_i_size_write(struct btrfs_inode *inode, u64 size)
 
 static inline bool btrfs_is_free_space_inode(struct btrfs_inode *inode)
 {
-	struct btrfs_root *root = inode->root;
-
-	if (root == root->fs_info->tree_root &&
-	    btrfs_ino(inode) != BTRFS_BTREE_INODE_OBJECTID)
-		return true;
-	if (inode->location.objectid == BTRFS_FREE_INO_OBJECTID)
-		return true;
-	return false;
+	return test_bit(BTRFS_INODE_FREE_SPACE_INODE, &inode->runtime_flags);
 }
 
 static inline bool is_data_inode(struct inode *inode)
@@ -265,73 +322,89 @@ static inline void btrfs_mod_outstanding_extents(struct btrfs_inode *inode,
 						  mod);
 }
 
-static inline int btrfs_inode_in_log(struct btrfs_inode *inode, u64 generation)
+/*
+ * Called every time after doing a buffered, direct IO or memory mapped write.
+ *
+ * This is to ensure that if we write to a file that was previously fsynced in
+ * the current transaction, then try to fsync it again in the same transaction,
+ * we will know that there were changes in the file and that it needs to be
+ * logged.
+ */
+static inline void btrfs_set_inode_last_sub_trans(struct btrfs_inode *inode)
+{
+	spin_lock(&inode->lock);
+	inode->last_sub_trans = inode->root->log_transid;
+	spin_unlock(&inode->lock);
+}
+
+/*
+ * Should be called while holding the inode's VFS lock in exclusive mode or in a
+ * context where no one else can access the inode concurrently (during inode
+ * creation or when loading an inode from disk).
+ */
+static inline void btrfs_set_inode_full_sync(struct btrfs_inode *inode)
+{
+	set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &inode->runtime_flags);
+	/*
+	 * The inode may have been part of a reflink operation in the last
+	 * transaction that modified it, and then a fsync has reset the
+	 * last_reflink_trans to avoid subsequent fsyncs in the same
+	 * transaction to do unnecessary work. So update last_reflink_trans
+	 * to the last_trans value (we have to be pessimistic and assume a
+	 * reflink happened).
+	 *
+	 * The ->last_trans is protected by the inode's spinlock and we can
+	 * have a concurrent ordered extent completion update it. Also set
+	 * last_reflink_trans to ->last_trans only if the former is less than
+	 * the later, because we can be called in a context where
+	 * last_reflink_trans was set to the current transaction generation
+	 * while ->last_trans was not yet updated in the current transaction,
+	 * and therefore has a lower value.
+	 */
+	spin_lock(&inode->lock);
+	if (inode->last_reflink_trans < inode->last_trans)
+		inode->last_reflink_trans = inode->last_trans;
+	spin_unlock(&inode->lock);
+}
+
+static inline bool btrfs_inode_in_log(struct btrfs_inode *inode, u64 generation)
 {
-	int ret = 0;
+	bool ret = false;
 
 	spin_lock(&inode->lock);
 	if (inode->logged_trans == generation &&
 	    inode->last_sub_trans <= inode->last_log_commit &&
-	    inode->last_sub_trans <= inode->root->last_log_commit) {
-		/*
-		 * After a ranged fsync we might have left some extent maps
-		 * (that fall outside the fsync's range). So return false
-		 * here if the list isn't empty, to make sure btrfs_log_inode()
-		 * will be called and process those extent maps.
-		 */
-		smp_mb();
-		if (list_empty(&inode->extent_tree.modified_extents))
-			ret = 1;
-	}
+	    inode->last_sub_trans <= inode->root->last_log_commit)
+		ret = true;
 	spin_unlock(&inode->lock);
 	return ret;
 }
 
-#define BTRFS_DIO_ORIG_BIO_SUBMITTED	0x1
-
-struct btrfs_dio_private {
-	struct inode *inode;
-	unsigned long flags;
-	u64 logical_offset;
-	u64 disk_bytenr;
-	u64 bytes;
-	void *private;
-
-	/* number of bios pending for this dio */
-	atomic_t pending_bios;
-
-	/* IO errors */
-	int errors;
-
-	/* orig_bio is our btrfs_io_bio */
-	struct bio *orig_bio;
-
-	/* dio_bio came from fs/direct-io.c */
-	struct bio *dio_bio;
-
-	/*
-	 * The original bio may be split to several sub-bios, this is
-	 * done during endio of sub-bios
-	 */
-	blk_status_t (*subio_endio)(struct inode *, struct btrfs_io_bio *,
-			blk_status_t);
-};
+/*
+ * Check if the inode has flags compatible with compression
+ */
+static inline bool btrfs_inode_can_compress(const struct btrfs_inode *inode)
+{
+	if (inode->flags & BTRFS_INODE_NODATACOW ||
+	    inode->flags & BTRFS_INODE_NODATASUM)
+		return false;
+	return true;
+}
 
 /*
- * Disable DIO read nolock optimization, so new dio readers will be forced
- * to grab i_mutex. It is used to avoid the endless truncate due to
- * nonlocked dio read.
+ * btrfs_inode_item stores flags in a u64, btrfs_inode stores them in two
+ * separate u32s. These two functions convert between the two representations.
  */
-static inline void btrfs_inode_block_unlocked_dio(struct btrfs_inode *inode)
+static inline u64 btrfs_inode_combine_flags(u32 flags, u32 ro_flags)
 {
-	set_bit(BTRFS_INODE_READDIO_NEED_LOCK, &inode->runtime_flags);
-	smp_mb();
+	return (flags | ((u64)ro_flags << 32));
 }
 
-static inline void btrfs_inode_resume_unlocked_dio(struct btrfs_inode *inode)
+static inline void btrfs_inode_split_flags(u64 inode_item_flags,
+					   u32 *flags, u32 *ro_flags)
 {
-	smp_mb__before_atomic();
-	clear_bit(BTRFS_INODE_READDIO_NEED_LOCK, &inode->runtime_flags);
+	*flags = (u32)inode_item_flags;
+	*ro_flags = (u32)(inode_item_flags >> 32);
 }
 
 /* Array of bytes with variable length, hexadecimal format 0x1234 */
@@ -342,8 +415,7 @@ static inline void btrfs_print_data_csum_error(struct btrfs_inode *inode,
 		u64 logical_start, u8 *csum, u8 *csum_expected, int mirror_num)
 {
 	struct btrfs_root *root = inode->root;
-	struct btrfs_super_block *sb = root->fs_info->super_copy;
-	const u16 csum_size = btrfs_super_csum_size(sb);
+	const u32 csum_size = root->fs_info->csum_size;
 
 	/* Output minus objectid, which is more meaningful */
 	if (root->root_key.objectid >= BTRFS_LAST_FREE_OBJECTID)
diff --git a/fs/btrfs/check-integrity.c b/fs/btrfs/check-integrity.c
index a0ce69f2d27c..98c6e5feab19 100644
--- a/fs/btrfs/check-integrity.c
+++ b/fs/btrfs/check-integrity.c
@@ -77,9 +77,7 @@
 
 #include <linux/sched.h>
 #include <linux/slab.h>
-#include <linux/buffer_head.h>
 #include <linux/mutex.h>
-#include <linux/genhd.h>
 #include <linux/blkdev.h>
 #include <linux/mm.h>
 #include <linux/string.h>
@@ -152,12 +150,9 @@ struct btrfsic_block {
 	struct list_head ref_to_list;	/* list */
 	struct list_head ref_from_list;	/* list */
 	struct btrfsic_block *next_in_same_bio;
-	void *orig_bio_bh_private;
-	union {
-		bio_end_io_t *bio;
-		bh_end_io_t *bh;
-	} orig_bio_bh_end_io;
-	int submit_bio_bh_rw;
+	void *orig_bio_private;
+	bio_end_io_t *orig_bio_end_io;
+	blk_opf_t submit_bio_bh_rw;
 	u64 flush_gen; /* only valid if !never_written */
 };
 
@@ -190,7 +185,6 @@ struct btrfsic_dev_state {
 	struct list_head collision_resolving_node;	/* list node */
 	struct btrfsic_block dummy_block_for_bio_bh_flush;
 	u64 last_flush_gen;
-	char name[BDEVNAME_SIZE];
 };
 
 struct btrfsic_block_hashtable {
@@ -237,7 +231,6 @@ struct btrfsic_stack_frame {
 struct btrfsic_state {
 	u32 print_mask;
 	int include_extent_data;
-	int csum_size;
 	struct list_head all_blocks_list;
 	struct btrfsic_block_hashtable block_hashtable;
 	struct btrfsic_block_link_hashtable block_link_hashtable;
@@ -248,47 +241,6 @@ struct btrfsic_state {
 	u32 datablock_size;
 };
 
-static void btrfsic_block_init(struct btrfsic_block *b);
-static struct btrfsic_block *btrfsic_block_alloc(void);
-static void btrfsic_block_free(struct btrfsic_block *b);
-static void btrfsic_block_link_init(struct btrfsic_block_link *n);
-static struct btrfsic_block_link *btrfsic_block_link_alloc(void);
-static void btrfsic_block_link_free(struct btrfsic_block_link *n);
-static void btrfsic_dev_state_init(struct btrfsic_dev_state *ds);
-static struct btrfsic_dev_state *btrfsic_dev_state_alloc(void);
-static void btrfsic_dev_state_free(struct btrfsic_dev_state *ds);
-static void btrfsic_block_hashtable_init(struct btrfsic_block_hashtable *h);
-static void btrfsic_block_hashtable_add(struct btrfsic_block *b,
-					struct btrfsic_block_hashtable *h);
-static void btrfsic_block_hashtable_remove(struct btrfsic_block *b);
-static struct btrfsic_block *btrfsic_block_hashtable_lookup(
-		struct block_device *bdev,
-		u64 dev_bytenr,
-		struct btrfsic_block_hashtable *h);
-static void btrfsic_block_link_hashtable_init(
-		struct btrfsic_block_link_hashtable *h);
-static void btrfsic_block_link_hashtable_add(
-		struct btrfsic_block_link *l,
-		struct btrfsic_block_link_hashtable *h);
-static void btrfsic_block_link_hashtable_remove(struct btrfsic_block_link *l);
-static struct btrfsic_block_link *btrfsic_block_link_hashtable_lookup(
-		struct block_device *bdev_ref_to,
-		u64 dev_bytenr_ref_to,
-		struct block_device *bdev_ref_from,
-		u64 dev_bytenr_ref_from,
-		struct btrfsic_block_link_hashtable *h);
-static void btrfsic_dev_state_hashtable_init(
-		struct btrfsic_dev_state_hashtable *h);
-static void btrfsic_dev_state_hashtable_add(
-		struct btrfsic_dev_state *ds,
-		struct btrfsic_dev_state_hashtable *h);
-static void btrfsic_dev_state_hashtable_remove(struct btrfsic_dev_state *ds);
-static struct btrfsic_dev_state *btrfsic_dev_state_hashtable_lookup(dev_t dev,
-		struct btrfsic_dev_state_hashtable *h);
-static struct btrfsic_stack_frame *btrfsic_stack_frame_alloc(void);
-static void btrfsic_stack_frame_free(struct btrfsic_stack_frame *sf);
-static int btrfsic_process_superblock(struct btrfsic_state *state,
-				      struct btrfs_fs_devices *fs_devices);
 static int btrfsic_process_metablock(struct btrfsic_state *state,
 				     struct btrfsic_block *block,
 				     struct btrfsic_block_data_ctx *block_ctx,
@@ -318,21 +270,11 @@ static int btrfsic_map_block(struct btrfsic_state *state, u64 bytenr, u32 len,
 static void btrfsic_release_block_ctx(struct btrfsic_block_data_ctx *block_ctx);
 static int btrfsic_read_block(struct btrfsic_state *state,
 			      struct btrfsic_block_data_ctx *block_ctx);
-static void btrfsic_dump_database(struct btrfsic_state *state);
-static int btrfsic_test_for_metadata(struct btrfsic_state *state,
-				     char **datav, unsigned int num_pages);
-static void btrfsic_process_written_block(struct btrfsic_dev_state *dev_state,
-					  u64 dev_bytenr, char **mapped_datav,
-					  unsigned int num_pages,
-					  struct bio *bio, int *bio_is_patched,
-					  struct buffer_head *bh,
-					  int submit_bio_bh_rw);
 static int btrfsic_process_written_superblock(
 		struct btrfsic_state *state,
 		struct btrfsic_block *const block,
 		struct btrfs_super_block *const super_hdr);
 static void btrfsic_bio_end_io(struct bio *bp);
-static void btrfsic_bh_end_io(struct buffer_head *bh, int uptodate);
 static int btrfsic_is_block_ref_by_superblock(const struct btrfsic_state *state,
 					      const struct btrfsic_block *block,
 					      int recursion_level);
@@ -399,8 +341,8 @@ static void btrfsic_block_init(struct btrfsic_block *b)
 	b->never_written = 0;
 	b->mirror_num = 0;
 	b->next_in_same_bio = NULL;
-	b->orig_bio_bh_private = NULL;
-	b->orig_bio_bh_end_io.bio = NULL;
+	b->orig_bio_private = NULL;
+	b->orig_bio_end_io = NULL;
 	INIT_LIST_HEAD(&b->collision_resolving_node);
 	INIT_LIST_HEAD(&b->all_blocks_node);
 	INIT_LIST_HEAD(&b->ref_to_list);
@@ -459,7 +401,6 @@ static void btrfsic_dev_state_init(struct btrfsic_dev_state *ds)
 	ds->magic_num = BTRFSIC_DEV2STATE_MAGIC_NUMBER;
 	ds->bdev = NULL;
 	ds->state = NULL;
-	ds->name[0] = '\0';
 	INIT_LIST_HEAD(&ds->collision_resolving_node);
 	ds->last_flush_gen = 0;
 	btrfsic_block_init(&ds->dummy_block_for_bio_bh_flush);
@@ -637,10 +578,8 @@ static int btrfsic_process_superblock(struct btrfsic_state *state,
 	int pass;
 
 	selected_super = kzalloc(sizeof(*selected_super), GFP_NOFS);
-	if (NULL == selected_super) {
-		pr_info("btrfsic: error, kmalloc failed!\n");
+	if (!selected_super)
 		return -ENOMEM;
-	}
 
 	list_for_each_entry(device, dev_head, dev_list) {
 		int i;
@@ -668,8 +607,6 @@ static int btrfsic_process_superblock(struct btrfsic_state *state,
 		return -1;
 	}
 
-	state->csum_size = btrfs_super_csum_size(selected_super);
-
 	for (pass = 0; pass < 3; pass++) {
 		int num_copies;
 		int mirror_num;
@@ -767,29 +704,31 @@ static int btrfsic_process_superblock_dev_mirror(
 	struct btrfs_fs_info *fs_info = state->fs_info;
 	struct btrfs_super_block *super_tmp;
 	u64 dev_bytenr;
-	struct buffer_head *bh;
 	struct btrfsic_block *superblock_tmp;
 	int pass;
 	struct block_device *const superblock_bdev = device->bdev;
+	struct page *page;
+	struct address_space *mapping = superblock_bdev->bd_inode->i_mapping;
+	int ret = 0;
 
 	/* super block bytenr is always the unmapped device bytenr */
 	dev_bytenr = btrfs_sb_offset(superblock_mirror_num);
 	if (dev_bytenr + BTRFS_SUPER_INFO_SIZE > device->commit_total_bytes)
 		return -1;
-	bh = __bread(superblock_bdev, dev_bytenr / BTRFS_BDEV_BLOCKSIZE,
-		     BTRFS_SUPER_INFO_SIZE);
-	if (NULL == bh)
+
+	page = read_cache_page_gfp(mapping, dev_bytenr >> PAGE_SHIFT, GFP_NOFS);
+	if (IS_ERR(page))
 		return -1;
-	super_tmp = (struct btrfs_super_block *)
-	    (bh->b_data + (dev_bytenr & (BTRFS_BDEV_BLOCKSIZE - 1)));
+
+	super_tmp = page_address(page);
 
 	if (btrfs_super_bytenr(super_tmp) != dev_bytenr ||
 	    btrfs_super_magic(super_tmp) != BTRFS_MAGIC ||
 	    memcmp(device->uuid, super_tmp->dev_item.uuid, BTRFS_UUID_SIZE) ||
 	    btrfs_super_nodesize(super_tmp) != state->metablock_size ||
 	    btrfs_super_sectorsize(super_tmp) != state->datablock_size) {
-		brelse(bh);
-		return 0;
+		ret = 0;
+		goto out;
 	}
 
 	superblock_tmp =
@@ -799,9 +738,8 @@ static int btrfsic_process_superblock_dev_mirror(
 	if (NULL == superblock_tmp) {
 		superblock_tmp = btrfsic_block_alloc();
 		if (NULL == superblock_tmp) {
-			pr_info("btrfsic: error, kmalloc failed!\n");
-			brelse(bh);
-			return -1;
+			ret = -1;
+			goto out;
 		}
 		/* for superblock, only the dev_bytenr makes sense */
 		superblock_tmp->dev_bytenr = dev_bytenr;
@@ -815,10 +753,10 @@ static int btrfsic_process_superblock_dev_mirror(
 		superblock_tmp->mirror_num = 1 + superblock_mirror_num;
 		if (state->print_mask & BTRFSIC_PRINT_MASK_SUPERBLOCK_WRITE)
 			btrfs_info_in_rcu(fs_info,
-				"new initial S-block (bdev %p, %s) @%llu (%s/%llu/%d)",
+			"new initial S-block (bdev %p, %s) @%llu (%pg/%llu/%d)",
 				     superblock_bdev,
 				     rcu_str_deref(device->name), dev_bytenr,
-				     dev_state->name, dev_bytenr,
+				     dev_state->bdev, dev_bytenr,
 				     superblock_mirror_num);
 		list_add(&superblock_tmp->all_blocks_node,
 			 &state->all_blocks_list);
@@ -885,8 +823,8 @@ static int btrfsic_process_superblock_dev_mirror(
 					      mirror_num)) {
 				pr_info("btrfsic: btrfsic_map_block(bytenr @%llu, mirror %d) failed!\n",
 				       next_bytenr, mirror_num);
-				brelse(bh);
-				return -1;
+				ret = -1;
+				goto out;
 			}
 
 			next_block = btrfsic_block_lookup_or_add(
@@ -895,8 +833,8 @@ static int btrfsic_process_superblock_dev_mirror(
 					mirror_num, NULL);
 			if (NULL == next_block) {
 				btrfsic_release_block_ctx(&tmp_next_block_ctx);
-				brelse(bh);
-				return -1;
+				ret = -1;
+				goto out;
 			}
 
 			next_block->disk_key = tmp_disk_key;
@@ -907,16 +845,17 @@ static int btrfsic_process_superblock_dev_mirror(
 					BTRFSIC_GENERATION_UNKNOWN);
 			btrfsic_release_block_ctx(&tmp_next_block_ctx);
 			if (NULL == l) {
-				brelse(bh);
-				return -1;
+				ret = -1;
+				goto out;
 			}
 		}
 	}
 	if (state->print_mask & BTRFSIC_PRINT_MASK_INITIAL_ALL_TREES)
 		btrfsic_dump_tree_sub(state, superblock_tmp, 0);
 
-	brelse(bh);
-	return 0;
+out:
+	put_page(page);
+	return ret;
 }
 
 static struct btrfsic_stack_frame *btrfsic_stack_frame_alloc(void)
@@ -924,9 +863,7 @@ static struct btrfsic_stack_frame *btrfsic_stack_frame_alloc(void)
 	struct btrfsic_stack_frame *sf;
 
 	sf = kzalloc(sizeof(*sf), GFP_NOFS);
-	if (NULL == sf)
-		pr_info("btrfsic: alloc memory failed!\n");
-	else
+	if (sf)
 		sf->magic = BTRFSIC_BLOCK_STACK_FRAME_MAGIC_NUMBER;
 	return sf;
 }
@@ -962,7 +899,7 @@ static noinline_for_stack int btrfsic_process_metablock(
 	sf->prev = NULL;
 
 continue_with_new_stack_frame:
-	sf->block->generation = le64_to_cpu(sf->hdr->generation);
+	sf->block->generation = btrfs_stack_header_generation(sf->hdr);
 	if (0 == sf->hdr->level) {
 		struct btrfs_leaf *const leafhdr =
 		    (struct btrfs_leaf *)sf->hdr;
@@ -998,9 +935,10 @@ continue_with_current_leaf_stack_frame:
 			if (disk_item_offset + sizeof(struct btrfs_item) >
 			    sf->block_ctx->len) {
 leaf_item_out_of_bounce_error:
-				pr_info("btrfsic: leaf item out of bounce at logical %llu, dev %s\n",
+				pr_info(
+		"btrfsic: leaf item out of bounce at logical %llu, dev %pg\n",
 				       sf->block_ctx->start,
-				       sf->block_ctx->dev->name);
+				       sf->block_ctx->dev->bdev);
 				goto one_stack_frame_backwards;
 			}
 			btrfsic_read_from_block_data(sf->block_ctx,
@@ -1118,9 +1056,10 @@ continue_with_current_node_stack_frame:
 					  (uintptr_t)nodehdr;
 			if (key_ptr_offset + sizeof(struct btrfs_key_ptr) >
 			    sf->block_ctx->len) {
-				pr_info("btrfsic: node item out of bounce at logical %llu, dev %s\n",
+				pr_info(
+		"btrfsic: node item out of bounce at logical %llu, dev %pg\n",
 				       sf->block_ctx->start,
-				       sf->block_ctx->dev->name);
+				       sf->block_ctx->dev->bdev);
 				goto one_stack_frame_backwards;
 			}
 			btrfsic_read_from_block_data(
@@ -1288,15 +1227,17 @@ static int btrfsic_create_link_to_next_block(
 			if (next_block->logical_bytenr != next_bytenr &&
 			    !(!next_block->is_metadata &&
 			      0 == next_block->logical_bytenr))
-				pr_info("Referenced block @%llu (%s/%llu/%d) found in hash table, %c, bytenr mismatch (!= stored %llu).\n",
-				       next_bytenr, next_block_ctx->dev->name,
+				pr_info(
+"referenced block @%llu (%pg/%llu/%d) found in hash table, %c, bytenr mismatch (!= stored %llu)\n",
+				       next_bytenr, next_block_ctx->dev->bdev,
 				       next_block_ctx->dev_bytenr, *mirror_nump,
 				       btrfsic_get_block_type(state,
 							      next_block),
 				       next_block->logical_bytenr);
 			else
-				pr_info("Referenced block @%llu (%s/%llu/%d) found in hash table, %c.\n",
-				       next_bytenr, next_block_ctx->dev->name,
+				pr_info(
+		"referenced block @%llu (%pg/%llu/%d) found in hash table, %c\n",
+				       next_bytenr, next_block_ctx->dev->bdev,
 				       next_block_ctx->dev_bytenr, *mirror_nump,
 				       btrfsic_get_block_type(state,
 							      next_block));
@@ -1316,7 +1257,6 @@ static int btrfsic_create_link_to_next_block(
 	if (NULL == l) {
 		l = btrfsic_block_link_alloc();
 		if (NULL == l) {
-			pr_info("btrfsic: error, kmalloc failed!\n");
 			btrfsic_release_block_ctx(next_block_ctx);
 			*next_blockp = NULL;
 			return -1;
@@ -1385,8 +1325,8 @@ static int btrfsic_handle_extent_data(
 	if (file_extent_item_offset +
 	    offsetof(struct btrfs_file_extent_item, disk_num_bytes) >
 	    block_ctx->len) {
-		pr_info("btrfsic: file item out of bounce at logical %llu, dev %s\n",
-		       block_ctx->start, block_ctx->dev->name);
+		pr_info("btrfsic: file item out of bounce at logical %llu, dev %pg\n",
+		       block_ctx->start, block_ctx->dev->bdev);
 		return -1;
 	}
 
@@ -1405,8 +1345,8 @@ static int btrfsic_handle_extent_data(
 
 	if (file_extent_item_offset + sizeof(struct btrfs_file_extent_item) >
 	    block_ctx->len) {
-		pr_info("btrfsic: file item out of bounce at logical %llu, dev %s\n",
-		       block_ctx->start, block_ctx->dev->name);
+		pr_info("btrfsic: file item out of bounce at logical %llu, dev %pg\n",
+		       block_ctx->start, block_ctx->dev->bdev);
 		return -1;
 	}
 	btrfsic_read_from_block_data(block_ctx, &file_extent_item,
@@ -1473,7 +1413,6 @@ static int btrfsic_handle_extent_data(
 					mirror_num,
 					&block_was_created);
 			if (NULL == next_block) {
-				pr_info("btrfsic: error, kmalloc failed!\n");
 				btrfsic_release_block_ctx(&next_block_ctx);
 				return -1;
 			}
@@ -1483,9 +1422,10 @@ static int btrfsic_handle_extent_data(
 				    next_block->logical_bytenr != next_bytenr &&
 				    !(!next_block->is_metadata &&
 				      0 == next_block->logical_bytenr)) {
-					pr_info("Referenced block @%llu (%s/%llu/%d) found in hash table, D, bytenr mismatch (!= stored %llu).\n",
+					pr_info(
+"referenced block @%llu (%pg/%llu/%d) found in hash table, D, bytenr mismatch (!= stored %llu)\n",
 					       next_bytenr,
-					       next_block_ctx.dev->name,
+					       next_block_ctx.dev->bdev,
 					       next_block_ctx.dev_bytenr,
 					       mirror_num,
 					       next_block->logical_bytenr);
@@ -1517,7 +1457,7 @@ static int btrfsic_map_block(struct btrfsic_state *state, u64 bytenr, u32 len,
 	struct btrfs_fs_info *fs_info = state->fs_info;
 	int ret;
 	u64 length;
-	struct btrfs_bio *multi = NULL;
+	struct btrfs_io_context *multi = NULL;
 	struct btrfs_device *device;
 
 	length = len;
@@ -1568,12 +1508,11 @@ static void btrfsic_release_block_ctx(struct btrfsic_block_data_ctx *block_ctx)
 		BUG_ON(!block_ctx->pagev);
 		num_pages = (block_ctx->len + (u64)PAGE_SIZE - 1) >>
 			    PAGE_SHIFT;
+		/* Pages must be unmapped in reverse order */
 		while (num_pages > 0) {
 			num_pages--;
-			if (block_ctx->datav[num_pages]) {
-				kunmap(block_ctx->pagev[num_pages]);
+			if (block_ctx->datav[num_pages])
 				block_ctx->datav[num_pages] = NULL;
-			}
 			if (block_ctx->pagev[num_pages]) {
 				__free_page(block_ctx->pagev[num_pages]);
 				block_ctx->pagev[num_pages] = NULL;
@@ -1613,21 +1552,18 @@ static int btrfsic_read_block(struct btrfsic_state *state,
 		return -ENOMEM;
 	block_ctx->datav = block_ctx->mem_to_free;
 	block_ctx->pagev = (struct page **)(block_ctx->datav + num_pages);
-	for (i = 0; i < num_pages; i++) {
-		block_ctx->pagev[i] = alloc_page(GFP_NOFS);
-		if (!block_ctx->pagev[i])
-			return -1;
-	}
+	ret = btrfs_alloc_page_array(num_pages, block_ctx->pagev);
+	if (ret)
+		return ret;
 
 	dev_bytenr = block_ctx->dev_bytenr;
 	for (i = 0; i < num_pages;) {
 		struct bio *bio;
 		unsigned int j;
 
-		bio = btrfs_io_bio_alloc(num_pages - i);
-		bio_set_dev(bio, block_ctx->dev->bdev);
+		bio = bio_alloc(block_ctx->dev->bdev, num_pages - i,
+				REQ_OP_READ, GFP_NOFS);
 		bio->bi_iter.bi_sector = dev_bytenr >> 9;
-		bio->bi_opf = REQ_OP_READ;
 
 		for (j = i; j < num_pages; j++) {
 			ret = bio_add_page(bio, block_ctx->pagev[j],
@@ -1640,8 +1576,8 @@ static int btrfsic_read_block(struct btrfsic_state *state,
 			return -1;
 		}
 		if (submit_bio_wait(bio)) {
-			pr_info("btrfsic: read error at logical %llu dev %s!\n",
-			       block_ctx->start, block_ctx->dev->name);
+			pr_info("btrfsic: read error at logical %llu dev %pg!\n",
+			       block_ctx->start, block_ctx->dev->bdev);
 			bio_put(bio);
 			return -1;
 		}
@@ -1650,7 +1586,7 @@ static int btrfsic_read_block(struct btrfsic_state *state,
 		i = j;
 	}
 	for (i = 0; i < num_pages; i++)
-		block_ctx->datav[i] = kmap(block_ctx->pagev[i]);
+		block_ctx->datav[i] = page_address(block_ctx->pagev[i]);
 
 	return block_ctx->len;
 }
@@ -1665,33 +1601,35 @@ static void btrfsic_dump_database(struct btrfsic_state *state)
 	list_for_each_entry(b_all, &state->all_blocks_list, all_blocks_node) {
 		const struct btrfsic_block_link *l;
 
-		pr_info("%c-block @%llu (%s/%llu/%d)\n",
+		pr_info("%c-block @%llu (%pg/%llu/%d)\n",
 		       btrfsic_get_block_type(state, b_all),
-		       b_all->logical_bytenr, b_all->dev_state->name,
+		       b_all->logical_bytenr, b_all->dev_state->bdev,
 		       b_all->dev_bytenr, b_all->mirror_num);
 
 		list_for_each_entry(l, &b_all->ref_to_list, node_ref_to) {
-			pr_info(" %c @%llu (%s/%llu/%d) refers %u* to %c @%llu (%s/%llu/%d)\n",
+			pr_info(
+		" %c @%llu (%pg/%llu/%d) refers %u* to %c @%llu (%pg/%llu/%d)\n",
 			       btrfsic_get_block_type(state, b_all),
-			       b_all->logical_bytenr, b_all->dev_state->name,
+			       b_all->logical_bytenr, b_all->dev_state->bdev,
 			       b_all->dev_bytenr, b_all->mirror_num,
 			       l->ref_cnt,
 			       btrfsic_get_block_type(state, l->block_ref_to),
 			       l->block_ref_to->logical_bytenr,
-			       l->block_ref_to->dev_state->name,
+			       l->block_ref_to->dev_state->bdev,
 			       l->block_ref_to->dev_bytenr,
 			       l->block_ref_to->mirror_num);
 		}
 
 		list_for_each_entry(l, &b_all->ref_from_list, node_ref_from) {
-			pr_info(" %c @%llu (%s/%llu/%d) is ref %u* from %c @%llu (%s/%llu/%d)\n",
+			pr_info(
+		" %c @%llu (%pg/%llu/%d) is ref %u* from %c @%llu (%pg/%llu/%d)\n",
 			       btrfsic_get_block_type(state, b_all),
-			       b_all->logical_bytenr, b_all->dev_state->name,
+			       b_all->logical_bytenr, b_all->dev_state->bdev,
 			       b_all->dev_bytenr, b_all->mirror_num,
 			       l->ref_cnt,
 			       btrfsic_get_block_type(state, l->block_ref_from),
 			       l->block_ref_from->logical_bytenr,
-			       l->block_ref_from->dev_state->name,
+			       l->block_ref_from->dev_state->bdev,
 			       l->block_ref_from->dev_bytenr,
 			       l->block_ref_from->mirror_num);
 		}
@@ -1733,7 +1671,7 @@ static noinline_for_stack int btrfsic_test_for_metadata(
 		crypto_shash_update(shash, data, sublen);
 	}
 	crypto_shash_final(shash, csum);
-	if (memcmp(csum, h->csum, state->csum_size))
+	if (memcmp(csum, h->csum, fs_info->csum_size))
 		return 1;
 
 	return 0; /* is metadata */
@@ -1743,8 +1681,7 @@ static void btrfsic_process_written_block(struct btrfsic_dev_state *dev_state,
 					  u64 dev_bytenr, char **mapped_datav,
 					  unsigned int num_pages,
 					  struct bio *bio, int *bio_is_patched,
-					  struct buffer_head *bh,
-					  int submit_bio_bh_rw)
+					  blk_opf_t submit_bio_bh_rw)
 {
 	int is_metadata;
 	struct btrfsic_block *block;
@@ -1807,16 +1744,18 @@ again:
 				if (block->logical_bytenr != bytenr &&
 				    !(!block->is_metadata &&
 				      block->logical_bytenr == 0))
-					pr_info("Written block @%llu (%s/%llu/%d) found in hash table, %c, bytenr mismatch (!= stored %llu).\n",
-					       bytenr, dev_state->name,
+					pr_info(
+"written block @%llu (%pg/%llu/%d) found in hash table, %c, bytenr mismatch (!= stored %llu)\n",
+					       bytenr, dev_state->bdev,
 					       dev_bytenr,
 					       block->mirror_num,
 					       btrfsic_get_block_type(state,
 								      block),
 					       block->logical_bytenr);
 				else
-					pr_info("Written block @%llu (%s/%llu/%d) found in hash table, %c.\n",
-					       bytenr, dev_state->name,
+					pr_info(
+		"written block @%llu (%pg/%llu/%d) found in hash table, %c\n",
+					       bytenr, dev_state->bdev,
 					       dev_bytenr, block->mirror_num,
 					       btrfsic_get_block_type(state,
 								      block));
@@ -1831,8 +1770,9 @@ again:
 			processed_len = state->datablock_size;
 			bytenr = block->logical_bytenr;
 			if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE)
-				pr_info("Written block @%llu (%s/%llu/%d) found in hash table, %c.\n",
-				       bytenr, dev_state->name, dev_bytenr,
+				pr_info(
+		"written block @%llu (%pg/%llu/%d) found in hash table, %c\n",
+				       bytenr, dev_state->bdev, dev_bytenr,
 				       block->mirror_num,
 				       btrfsic_get_block_type(state, block));
 		}
@@ -1842,9 +1782,10 @@ again:
 			       list_empty(&block->ref_to_list) ? ' ' : '!',
 			       list_empty(&block->ref_from_list) ? ' ' : '!');
 		if (btrfsic_is_block_ref_by_superblock(state, block, 0)) {
-			pr_info("btrfs: attempt to overwrite %c-block @%llu (%s/%llu/%d), old(gen=%llu, objectid=%llu, type=%d, offset=%llu), new(gen=%llu), which is referenced by most recent superblock (superblockgen=%llu)!\n",
+			pr_info(
+"btrfs: attempt to overwrite %c-block @%llu (%pg/%llu/%d), old(gen=%llu, objectid=%llu, type=%d, offset=%llu), new(gen=%llu), which is referenced by most recent superblock (superblockgen=%llu)!\n",
 			       btrfsic_get_block_type(state, block), bytenr,
-			       dev_state->name, dev_bytenr, block->mirror_num,
+			       dev_state->bdev, dev_bytenr, block->mirror_num,
 			       block->generation,
 			       btrfs_disk_key_objectid(&block->disk_key),
 			       block->disk_key.type,
@@ -1856,9 +1797,10 @@ again:
 		}
 
 		if (!block->is_iodone && !block->never_written) {
-			pr_info("btrfs: attempt to overwrite %c-block @%llu (%s/%llu/%d), oldgen=%llu, newgen=%llu, which is not yet iodone!\n",
+			pr_info(
+"btrfs: attempt to overwrite %c-block @%llu (%pg/%llu/%d), oldgen=%llu, newgen=%llu, which is not yet iodone!\n",
 			       btrfsic_get_block_type(state, block), bytenr,
-			       dev_state->name, dev_bytenr, block->mirror_num,
+			       dev_state->bdev, dev_bytenr, block->mirror_num,
 			       block->generation,
 			       btrfs_stack_header_generation(
 				       (struct btrfs_header *)
@@ -1902,9 +1844,9 @@ again:
 				block->is_iodone = 0;
 				BUG_ON(NULL == bio_is_patched);
 				if (!*bio_is_patched) {
-					block->orig_bio_bh_private =
+					block->orig_bio_private =
 					    bio->bi_private;
-					block->orig_bio_bh_end_io.bio =
+					block->orig_bio_end_io =
 					    bio->bi_end_io;
 					block->next_in_same_bio = NULL;
 					bio->bi_private = block;
@@ -1916,25 +1858,17 @@ again:
 					    bio->bi_private;
 
 					BUG_ON(NULL == chained_block);
-					block->orig_bio_bh_private =
-					    chained_block->orig_bio_bh_private;
-					block->orig_bio_bh_end_io.bio =
-					    chained_block->orig_bio_bh_end_io.
-					    bio;
+					block->orig_bio_private =
+					    chained_block->orig_bio_private;
+					block->orig_bio_end_io =
+					    chained_block->orig_bio_end_io;
 					block->next_in_same_bio = chained_block;
 					bio->bi_private = block;
 				}
-			} else if (NULL != bh) {
-				block->is_iodone = 0;
-				block->orig_bio_bh_private = bh->b_private;
-				block->orig_bio_bh_end_io.bh = bh->b_end_io;
-				block->next_in_same_bio = NULL;
-				bh->b_private = block;
-				bh->b_end_io = btrfsic_bh_end_io;
 			} else {
 				block->is_iodone = 1;
-				block->orig_bio_bh_private = NULL;
-				block->orig_bio_bh_end_io.bio = NULL;
+				block->orig_bio_private = NULL;
+				block->orig_bio_end_io = NULL;
 				block->next_in_same_bio = NULL;
 			}
 		}
@@ -1993,8 +1927,9 @@ again:
 		if (!is_metadata) {
 			processed_len = state->datablock_size;
 			if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE)
-				pr_info("Written block (%s/%llu/?) !found in hash table, D.\n",
-				       dev_state->name, dev_bytenr);
+				pr_info(
+			"written block (%pg/%llu/?) !found in hash table, D\n",
+				       dev_state->bdev, dev_bytenr);
 			if (!state->include_extent_data) {
 				/* ignore that written D block */
 				goto continue_loop;
@@ -2011,8 +1946,9 @@ again:
 			btrfsic_cmp_log_and_dev_bytenr(state, bytenr, dev_state,
 						       dev_bytenr);
 			if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE)
-				pr_info("Written block @%llu (%s/%llu/?) !found in hash table, M.\n",
-				       bytenr, dev_state->name, dev_bytenr);
+				pr_info(
+			"written block @%llu (%pg/%llu/?) !found in hash table, M\n",
+				       bytenr, dev_state->bdev, dev_bytenr);
 		}
 
 		block_ctx.dev = dev_state;
@@ -2025,7 +1961,6 @@ again:
 
 		block = btrfsic_block_alloc();
 		if (NULL == block) {
-			pr_info("btrfsic: error, kmalloc failed!\n");
 			btrfsic_release_block_ctx(&block_ctx);
 			goto continue_loop;
 		}
@@ -2042,8 +1977,8 @@ again:
 			block->is_iodone = 0;
 			BUG_ON(NULL == bio_is_patched);
 			if (!*bio_is_patched) {
-				block->orig_bio_bh_private = bio->bi_private;
-				block->orig_bio_bh_end_io.bio = bio->bi_end_io;
+				block->orig_bio_private = bio->bi_private;
+				block->orig_bio_end_io = bio->bi_end_io;
 				block->next_in_same_bio = NULL;
 				bio->bi_private = block;
 				bio->bi_end_io = btrfsic_bio_end_io;
@@ -2054,30 +1989,23 @@ again:
 				    bio->bi_private;
 
 				BUG_ON(NULL == chained_block);
-				block->orig_bio_bh_private =
-				    chained_block->orig_bio_bh_private;
-				block->orig_bio_bh_end_io.bio =
-				    chained_block->orig_bio_bh_end_io.bio;
+				block->orig_bio_private =
+				    chained_block->orig_bio_private;
+				block->orig_bio_end_io =
+				    chained_block->orig_bio_end_io;
 				block->next_in_same_bio = chained_block;
 				bio->bi_private = block;
 			}
-		} else if (NULL != bh) {
-			block->is_iodone = 0;
-			block->orig_bio_bh_private = bh->b_private;
-			block->orig_bio_bh_end_io.bh = bh->b_end_io;
-			block->next_in_same_bio = NULL;
-			bh->b_private = block;
-			bh->b_end_io = btrfsic_bh_end_io;
 		} else {
 			block->is_iodone = 1;
-			block->orig_bio_bh_private = NULL;
-			block->orig_bio_bh_end_io.bio = NULL;
+			block->orig_bio_private = NULL;
+			block->orig_bio_end_io = NULL;
 			block->next_in_same_bio = NULL;
 		}
 		if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE)
-			pr_info("New written %c-block @%llu (%s/%llu/%d)\n",
+			pr_info("new written %c-block @%llu (%pg/%llu/%d)\n",
 			       is_metadata ? 'M' : 'D',
-			       block->logical_bytenr, block->dev_state->name,
+			       block->logical_bytenr, block->dev_state->bdev,
 			       block->dev_bytenr, block->mirror_num);
 		list_add(&block->all_blocks_node, &state->all_blocks_list);
 		btrfsic_block_hashtable_add(block, &state->block_hashtable);
@@ -2102,7 +2030,7 @@ continue_loop:
 
 static void btrfsic_bio_end_io(struct bio *bp)
 {
-	struct btrfsic_block *block = (struct btrfsic_block *)bp->bi_private;
+	struct btrfsic_block *block = bp->bi_private;
 	int iodone_w_error;
 
 	/* mutex is not held! This is not save if IO is not yet completed
@@ -2112,8 +2040,8 @@ static void btrfsic_bio_end_io(struct bio *bp)
 		iodone_w_error = 1;
 
 	BUG_ON(NULL == block);
-	bp->bi_private = block->orig_bio_bh_private;
-	bp->bi_end_io = block->orig_bio_bh_end_io.bio;
+	bp->bi_private = block->orig_bio_private;
+	bp->bi_end_io = block->orig_bio_end_io;
 
 	do {
 		struct btrfsic_block *next_block;
@@ -2121,10 +2049,10 @@ static void btrfsic_bio_end_io(struct bio *bp)
 
 		if ((dev_state->state->print_mask &
 		     BTRFSIC_PRINT_MASK_END_IO_BIO_BH))
-			pr_info("bio_end_io(err=%d) for %c @%llu (%s/%llu/%d)\n",
+			pr_info("bio_end_io(err=%d) for %c @%llu (%pg/%llu/%d)\n",
 			       bp->bi_status,
 			       btrfsic_get_block_type(dev_state->state, block),
-			       block->logical_bytenr, dev_state->name,
+			       block->logical_bytenr, dev_state->bdev,
 			       block->dev_bytenr, block->mirror_num);
 		next_block = block->next_in_same_bio;
 		block->iodone_w_error = iodone_w_error;
@@ -2132,8 +2060,8 @@ static void btrfsic_bio_end_io(struct bio *bp)
 			dev_state->last_flush_gen++;
 			if ((dev_state->state->print_mask &
 			     BTRFSIC_PRINT_MASK_END_IO_BIO_BH))
-				pr_info("bio_end_io() new %s flush_gen=%llu\n",
-				       dev_state->name,
+				pr_info("bio_end_io() new %pg flush_gen=%llu\n",
+				       dev_state->bdev,
 				       dev_state->last_flush_gen);
 		}
 		if (block->submit_bio_bh_rw & REQ_FUA)
@@ -2146,38 +2074,6 @@ static void btrfsic_bio_end_io(struct bio *bp)
 	bp->bi_end_io(bp);
 }
 
-static void btrfsic_bh_end_io(struct buffer_head *bh, int uptodate)
-{
-	struct btrfsic_block *block = (struct btrfsic_block *)bh->b_private;
-	int iodone_w_error = !uptodate;
-	struct btrfsic_dev_state *dev_state;
-
-	BUG_ON(NULL == block);
-	dev_state = block->dev_state;
-	if ((dev_state->state->print_mask & BTRFSIC_PRINT_MASK_END_IO_BIO_BH))
-		pr_info("bh_end_io(error=%d) for %c @%llu (%s/%llu/%d)\n",
-		       iodone_w_error,
-		       btrfsic_get_block_type(dev_state->state, block),
-		       block->logical_bytenr, block->dev_state->name,
-		       block->dev_bytenr, block->mirror_num);
-
-	block->iodone_w_error = iodone_w_error;
-	if (block->submit_bio_bh_rw & REQ_PREFLUSH) {
-		dev_state->last_flush_gen++;
-		if ((dev_state->state->print_mask &
-		     BTRFSIC_PRINT_MASK_END_IO_BIO_BH))
-			pr_info("bh_end_io() new %s flush_gen=%llu\n",
-			       dev_state->name, dev_state->last_flush_gen);
-	}
-	if (block->submit_bio_bh_rw & REQ_FUA)
-		block->flush_gen = 0; /* FUA completed means block is on disk */
-
-	bh->b_private = block->orig_bio_bh_private;
-	bh->b_end_io = block->orig_bio_bh_end_io.bh;
-	block->is_iodone = 1; /* for FLUSH, this releases the block */
-	bh->b_end_io(bh, uptodate);
-}
-
 static int btrfsic_process_written_superblock(
 		struct btrfsic_state *state,
 		struct btrfsic_block *const superblock,
@@ -2190,17 +2086,19 @@ static int btrfsic_process_written_superblock(
 	if (!(superblock->generation > state->max_superblock_generation ||
 	      0 == state->max_superblock_generation)) {
 		if (state->print_mask & BTRFSIC_PRINT_MASK_SUPERBLOCK_WRITE)
-			pr_info("btrfsic: superblock @%llu (%s/%llu/%d) with old gen %llu <= %llu\n",
+			pr_info(
+	"btrfsic: superblock @%llu (%pg/%llu/%d) with old gen %llu <= %llu\n",
 			       superblock->logical_bytenr,
-			       superblock->dev_state->name,
+			       superblock->dev_state->bdev,
 			       superblock->dev_bytenr, superblock->mirror_num,
 			       btrfs_super_generation(super_hdr),
 			       state->max_superblock_generation);
 	} else {
 		if (state->print_mask & BTRFSIC_PRINT_MASK_SUPERBLOCK_WRITE)
-			pr_info("btrfsic: got new superblock @%llu (%s/%llu/%d) with new gen %llu > %llu\n",
+			pr_info(
+	"btrfsic: got new superblock @%llu (%pg/%llu/%d) with new gen %llu > %llu\n",
 			       superblock->logical_bytenr,
-			       superblock->dev_state->name,
+			       superblock->dev_state->bdev,
 			       superblock->dev_bytenr, superblock->mirror_num,
 			       btrfs_super_generation(super_hdr),
 			       state->max_superblock_generation);
@@ -2285,7 +2183,6 @@ static int btrfsic_process_written_superblock(
 					mirror_num,
 					&was_created);
 			if (NULL == next_block) {
-				pr_info("btrfsic: error, kmalloc failed!\n");
 				btrfsic_release_block_ctx(&tmp_next_block_ctx);
 				return -1;
 			}
@@ -2345,38 +2242,42 @@ static int btrfsic_check_all_ref_blocks(struct btrfsic_state *state,
 	 */
 	list_for_each_entry(l, &block->ref_to_list, node_ref_to) {
 		if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE)
-			pr_info("rl=%d, %c @%llu (%s/%llu/%d) %u* refers to %c @%llu (%s/%llu/%d)\n",
+			pr_info(
+		"rl=%d, %c @%llu (%pg/%llu/%d) %u* refers to %c @%llu (%pg/%llu/%d)\n",
 			       recursion_level,
 			       btrfsic_get_block_type(state, block),
-			       block->logical_bytenr, block->dev_state->name,
+			       block->logical_bytenr, block->dev_state->bdev,
 			       block->dev_bytenr, block->mirror_num,
 			       l->ref_cnt,
 			       btrfsic_get_block_type(state, l->block_ref_to),
 			       l->block_ref_to->logical_bytenr,
-			       l->block_ref_to->dev_state->name,
+			       l->block_ref_to->dev_state->bdev,
 			       l->block_ref_to->dev_bytenr,
 			       l->block_ref_to->mirror_num);
 		if (l->block_ref_to->never_written) {
-			pr_info("btrfs: attempt to write superblock which references block %c @%llu (%s/%llu/%d) which is never written!\n",
+			pr_info(
+"btrfs: attempt to write superblock which references block %c @%llu (%pg/%llu/%d) which is never written!\n",
 			       btrfsic_get_block_type(state, l->block_ref_to),
 			       l->block_ref_to->logical_bytenr,
-			       l->block_ref_to->dev_state->name,
+			       l->block_ref_to->dev_state->bdev,
 			       l->block_ref_to->dev_bytenr,
 			       l->block_ref_to->mirror_num);
 			ret = -1;
 		} else if (!l->block_ref_to->is_iodone) {
-			pr_info("btrfs: attempt to write superblock which references block %c @%llu (%s/%llu/%d) which is not yet iodone!\n",
+			pr_info(
+"btrfs: attempt to write superblock which references block %c @%llu (%pg/%llu/%d) which is not yet iodone!\n",
 			       btrfsic_get_block_type(state, l->block_ref_to),
 			       l->block_ref_to->logical_bytenr,
-			       l->block_ref_to->dev_state->name,
+			       l->block_ref_to->dev_state->bdev,
 			       l->block_ref_to->dev_bytenr,
 			       l->block_ref_to->mirror_num);
 			ret = -1;
 		} else if (l->block_ref_to->iodone_w_error) {
-			pr_info("btrfs: attempt to write superblock which references block %c @%llu (%s/%llu/%d) which has write error!\n",
+			pr_info(
+"btrfs: attempt to write superblock which references block %c @%llu (%pg/%llu/%d) which has write error!\n",
 			       btrfsic_get_block_type(state, l->block_ref_to),
 			       l->block_ref_to->logical_bytenr,
-			       l->block_ref_to->dev_state->name,
+			       l->block_ref_to->dev_state->bdev,
 			       l->block_ref_to->dev_bytenr,
 			       l->block_ref_to->mirror_num);
 			ret = -1;
@@ -2386,10 +2287,11 @@ static int btrfsic_check_all_ref_blocks(struct btrfsic_state *state,
 			   l->parent_generation &&
 			   BTRFSIC_GENERATION_UNKNOWN !=
 			   l->block_ref_to->generation) {
-			pr_info("btrfs: attempt to write superblock which references block %c @%llu (%s/%llu/%d) with generation %llu != parent generation %llu!\n",
+			pr_info(
+"btrfs: attempt to write superblock which references block %c @%llu (%pg/%llu/%d) with generation %llu != parent generation %llu!\n",
 			       btrfsic_get_block_type(state, l->block_ref_to),
 			       l->block_ref_to->logical_bytenr,
-			       l->block_ref_to->dev_state->name,
+			       l->block_ref_to->dev_state->bdev,
 			       l->block_ref_to->dev_bytenr,
 			       l->block_ref_to->mirror_num,
 			       l->block_ref_to->generation,
@@ -2397,10 +2299,11 @@ static int btrfsic_check_all_ref_blocks(struct btrfsic_state *state,
 			ret = -1;
 		} else if (l->block_ref_to->flush_gen >
 			   l->block_ref_to->dev_state->last_flush_gen) {
-			pr_info("btrfs: attempt to write superblock which references block %c @%llu (%s/%llu/%d) which is not flushed out of disk's write cache (block flush_gen=%llu, dev->flush_gen=%llu)!\n",
+			pr_info(
+"btrfs: attempt to write superblock which references block %c @%llu (%pg/%llu/%d) which is not flushed out of disk's write cache (block flush_gen=%llu, dev->flush_gen=%llu)!\n",
 			       btrfsic_get_block_type(state, l->block_ref_to),
 			       l->block_ref_to->logical_bytenr,
-			       l->block_ref_to->dev_state->name,
+			       l->block_ref_to->dev_state->bdev,
 			       l->block_ref_to->dev_bytenr,
 			       l->block_ref_to->mirror_num, block->flush_gen,
 			       l->block_ref_to->dev_state->last_flush_gen);
@@ -2437,15 +2340,16 @@ static int btrfsic_is_block_ref_by_superblock(
 	 */
 	list_for_each_entry(l, &block->ref_from_list, node_ref_from) {
 		if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE)
-			pr_info("rl=%d, %c @%llu (%s/%llu/%d) is ref %u* from %c @%llu (%s/%llu/%d)\n",
+			pr_info(
+	"rl=%d, %c @%llu (%pg/%llu/%d) is ref %u* from %c @%llu (%pg/%llu/%d)\n",
 			       recursion_level,
 			       btrfsic_get_block_type(state, block),
-			       block->logical_bytenr, block->dev_state->name,
+			       block->logical_bytenr, block->dev_state->bdev,
 			       block->dev_bytenr, block->mirror_num,
 			       l->ref_cnt,
 			       btrfsic_get_block_type(state, l->block_ref_from),
 			       l->block_ref_from->logical_bytenr,
-			       l->block_ref_from->dev_state->name,
+			       l->block_ref_from->dev_state->bdev,
 			       l->block_ref_from->dev_bytenr,
 			       l->block_ref_from->mirror_num);
 		if (l->block_ref_from->is_superblock &&
@@ -2467,30 +2371,30 @@ static int btrfsic_is_block_ref_by_superblock(
 static void btrfsic_print_add_link(const struct btrfsic_state *state,
 				   const struct btrfsic_block_link *l)
 {
-	pr_info("Add %u* link from %c @%llu (%s/%llu/%d) to %c @%llu (%s/%llu/%d).\n",
+	pr_info("add %u* link from %c @%llu (%pg/%llu/%d) to %c @%llu (%pg/%llu/%d)\n",
 	       l->ref_cnt,
 	       btrfsic_get_block_type(state, l->block_ref_from),
 	       l->block_ref_from->logical_bytenr,
-	       l->block_ref_from->dev_state->name,
+	       l->block_ref_from->dev_state->bdev,
 	       l->block_ref_from->dev_bytenr, l->block_ref_from->mirror_num,
 	       btrfsic_get_block_type(state, l->block_ref_to),
 	       l->block_ref_to->logical_bytenr,
-	       l->block_ref_to->dev_state->name, l->block_ref_to->dev_bytenr,
+	       l->block_ref_to->dev_state->bdev, l->block_ref_to->dev_bytenr,
 	       l->block_ref_to->mirror_num);
 }
 
 static void btrfsic_print_rem_link(const struct btrfsic_state *state,
 				   const struct btrfsic_block_link *l)
 {
-	pr_info("Rem %u* link from %c @%llu (%s/%llu/%d) to %c @%llu (%s/%llu/%d).\n",
+	pr_info("rem %u* link from %c @%llu (%pg/%llu/%d) to %c @%llu (%pg/%llu/%d)\n",
 	       l->ref_cnt,
 	       btrfsic_get_block_type(state, l->block_ref_from),
 	       l->block_ref_from->logical_bytenr,
-	       l->block_ref_from->dev_state->name,
+	       l->block_ref_from->dev_state->bdev,
 	       l->block_ref_from->dev_bytenr, l->block_ref_from->mirror_num,
 	       btrfsic_get_block_type(state, l->block_ref_to),
 	       l->block_ref_to->logical_bytenr,
-	       l->block_ref_to->dev_state->name, l->block_ref_to->dev_bytenr,
+	       l->block_ref_to->dev_state->bdev, l->block_ref_to->dev_bytenr,
 	       l->block_ref_to->mirror_num);
 }
 
@@ -2532,9 +2436,9 @@ static void btrfsic_dump_tree_sub(const struct btrfsic_state *state,
 	 * This algorithm is recursive because the amount of used stack space
 	 * is very small and the max recursion depth is limited.
 	 */
-	indent_add = sprintf(buf, "%c-%llu(%s/%llu/%u)",
+	indent_add = sprintf(buf, "%c-%llu(%pg/%llu/%u)",
 			     btrfsic_get_block_type(state, block),
-			     block->logical_bytenr, block->dev_state->name,
+			     block->logical_bytenr, block->dev_state->bdev,
 			     block->dev_bytenr, block->mirror_num);
 	if (indent_level + indent_add > BTRFSIC_TREE_DUMP_MAX_INDENT_LEVEL) {
 		printk("[...]\n");
@@ -2593,10 +2497,8 @@ static struct btrfsic_block_link *btrfsic_block_link_lookup_or_add(
 						&state->block_link_hashtable);
 	if (NULL == l) {
 		l = btrfsic_block_link_alloc();
-		if (NULL == l) {
-			pr_info("btrfsic: error, kmalloc failed!\n");
+		if (!l)
 			return NULL;
-		}
 
 		l->block_ref_to = next_block;
 		l->block_ref_from = from_block;
@@ -2640,10 +2542,9 @@ static struct btrfsic_block *btrfsic_block_lookup_or_add(
 		struct btrfsic_dev_state *dev_state;
 
 		block = btrfsic_block_alloc();
-		if (NULL == block) {
-			pr_info("btrfsic: error, kmalloc failed!\n");
+		if (!block)
 			return NULL;
-		}
+
 		dev_state = btrfsic_dev_state_lookup(block_ctx->dev->bdev->bd_dev);
 		if (NULL == dev_state) {
 			pr_info("btrfsic: error, lookup dev_state failed!\n");
@@ -2658,10 +2559,10 @@ static struct btrfsic_block *btrfsic_block_lookup_or_add(
 		block->never_written = never_written;
 		block->mirror_num = mirror_num;
 		if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE)
-			pr_info("New %s%c-block @%llu (%s/%llu/%d)\n",
+			pr_info("New %s%c-block @%llu (%pg/%llu/%d)\n",
 			       additional_string,
 			       btrfsic_get_block_type(state, block),
-			       block->logical_bytenr, dev_state->name,
+			       block->logical_bytenr, dev_state->bdev,
 			       block->dev_bytenr, mirror_num);
 		list_add(&block->all_blocks_node, &state->all_blocks_list);
 		btrfsic_block_hashtable_add(block, &state->block_hashtable);
@@ -2708,8 +2609,9 @@ static void btrfsic_cmp_log_and_dev_bytenr(struct btrfsic_state *state,
 	}
 
 	if (WARN_ON(!match)) {
-		pr_info("btrfs: attempt to write M-block which contains logical bytenr that doesn't map to dev+physical bytenr of submit_bio, buffer->log_bytenr=%llu, submit_bio(bdev=%s, phys_bytenr=%llu)!\n",
-		       bytenr, dev_state->name, dev_bytenr);
+		pr_info(
+"btrfs: attempt to write M-block which contains logical bytenr that doesn't map to dev+physical bytenr of submit_bio, buffer->log_bytenr=%llu, submit_bio(bdev=%pg, phys_bytenr=%llu)!\n",
+		       bytenr, dev_state->bdev, dev_bytenr);
 		for (mirror_num = 1; mirror_num <= num_copies; mirror_num++) {
 			ret = btrfsic_map_block(state, bytenr,
 						state->metablock_size,
@@ -2717,8 +2619,8 @@ static void btrfsic_cmp_log_and_dev_bytenr(struct btrfsic_state *state,
 			if (ret)
 				continue;
 
-			pr_info("Read logical bytenr @%llu maps to (%s/%llu/%d)\n",
-			       bytenr, block_ctx.dev->name,
+			pr_info("read logical bytenr @%llu maps to (%pg/%llu/%d)\n",
+			       bytenr, block_ctx.dev->bdev,
 			       block_ctx.dev_bytenr, mirror_num);
 		}
 	}
@@ -2730,161 +2632,95 @@ static struct btrfsic_dev_state *btrfsic_dev_state_lookup(dev_t dev)
 						  &btrfsic_dev_state_hashtable);
 }
 
-int btrfsic_submit_bh(int op, int op_flags, struct buffer_head *bh)
+static void btrfsic_check_write_bio(struct bio *bio, struct btrfsic_dev_state *dev_state)
 {
-	struct btrfsic_dev_state *dev_state;
+	unsigned int segs = bio_segments(bio);
+	u64 dev_bytenr = 512 * bio->bi_iter.bi_sector;
+	u64 cur_bytenr = dev_bytenr;
+	struct bvec_iter iter;
+	struct bio_vec bvec;
+	char **mapped_datav;
+	int bio_is_patched = 0;
+	int i = 0;
+
+	if (dev_state->state->print_mask & BTRFSIC_PRINT_MASK_SUBMIT_BIO_BH)
+		pr_info(
+"submit_bio(rw=%d,0x%x, bi_vcnt=%u, bi_sector=%llu (bytenr %llu), bi_bdev=%p)\n",
+		       bio_op(bio), bio->bi_opf, segs,
+		       bio->bi_iter.bi_sector, dev_bytenr, bio->bi_bdev);
+
+	mapped_datav = kmalloc_array(segs, sizeof(*mapped_datav), GFP_NOFS);
+	if (!mapped_datav)
+		return;
 
-	if (!btrfsic_is_initialized)
-		return submit_bh(op, op_flags, bh);
+	bio_for_each_segment(bvec, bio, iter) {
+		BUG_ON(bvec.bv_len != PAGE_SIZE);
+		mapped_datav[i] = page_address(bvec.bv_page);
+		i++;
 
-	mutex_lock(&btrfsic_mutex);
-	/* since btrfsic_submit_bh() might also be called before
-	 * btrfsic_mount(), this might return NULL */
-	dev_state = btrfsic_dev_state_lookup(bh->b_bdev->bd_dev);
+		if (dev_state->state->print_mask &
+		    BTRFSIC_PRINT_MASK_SUBMIT_BIO_BH_VERBOSE)
+			pr_info("#%u: bytenr=%llu, len=%u, offset=%u\n",
+			       i, cur_bytenr, bvec.bv_len, bvec.bv_offset);
+		cur_bytenr += bvec.bv_len;
+	}
 
-	/* Only called to write the superblock (incl. FLUSH/FUA) */
-	if (NULL != dev_state &&
-	    (op == REQ_OP_WRITE) && bh->b_size > 0) {
-		u64 dev_bytenr;
+	btrfsic_process_written_block(dev_state, dev_bytenr, mapped_datav, segs,
+				      bio, &bio_is_patched, bio->bi_opf);
+	kfree(mapped_datav);
+}
 
-		dev_bytenr = BTRFS_BDEV_BLOCKSIZE * bh->b_blocknr;
-		if (dev_state->state->print_mask &
-		    BTRFSIC_PRINT_MASK_SUBMIT_BIO_BH)
-			pr_info("submit_bh(op=0x%x,0x%x, blocknr=%llu (bytenr %llu), size=%zu, data=%p, bdev=%p)\n",
-			       op, op_flags, (unsigned long long)bh->b_blocknr,
-			       dev_bytenr, bh->b_size, bh->b_data, bh->b_bdev);
-		btrfsic_process_written_block(dev_state, dev_bytenr,
-					      &bh->b_data, 1, NULL,
-					      NULL, bh, op_flags);
-	} else if (NULL != dev_state && (op_flags & REQ_PREFLUSH)) {
-		if (dev_state->state->print_mask &
-		    BTRFSIC_PRINT_MASK_SUBMIT_BIO_BH)
-			pr_info("submit_bh(op=0x%x,0x%x FLUSH, bdev=%p)\n",
-			       op, op_flags, bh->b_bdev);
-		if (!dev_state->dummy_block_for_bio_bh_flush.is_iodone) {
-			if ((dev_state->state->print_mask &
-			     (BTRFSIC_PRINT_MASK_SUBMIT_BIO_BH |
-			      BTRFSIC_PRINT_MASK_VERBOSE)))
-				pr_info("btrfsic_submit_bh(%s) with FLUSH but dummy block already in use (ignored)!\n",
-				       dev_state->name);
-		} else {
-			struct btrfsic_block *const block =
-				&dev_state->dummy_block_for_bio_bh_flush;
+static void btrfsic_check_flush_bio(struct bio *bio, struct btrfsic_dev_state *dev_state)
+{
+	if (dev_state->state->print_mask & BTRFSIC_PRINT_MASK_SUBMIT_BIO_BH)
+		pr_info("submit_bio(rw=%d,0x%x FLUSH, bdev=%p)\n",
+		       bio_op(bio), bio->bi_opf, bio->bi_bdev);
 
-			block->is_iodone = 0;
-			block->never_written = 0;
-			block->iodone_w_error = 0;
-			block->flush_gen = dev_state->last_flush_gen + 1;
-			block->submit_bio_bh_rw = op_flags;
-			block->orig_bio_bh_private = bh->b_private;
-			block->orig_bio_bh_end_io.bh = bh->b_end_io;
-			block->next_in_same_bio = NULL;
-			bh->b_private = block;
-			bh->b_end_io = btrfsic_bh_end_io;
-		}
+	if (dev_state->dummy_block_for_bio_bh_flush.is_iodone) {
+		struct btrfsic_block *const block =
+			&dev_state->dummy_block_for_bio_bh_flush;
+
+		block->is_iodone = 0;
+		block->never_written = 0;
+		block->iodone_w_error = 0;
+		block->flush_gen = dev_state->last_flush_gen + 1;
+		block->submit_bio_bh_rw = bio->bi_opf;
+		block->orig_bio_private = bio->bi_private;
+		block->orig_bio_end_io = bio->bi_end_io;
+		block->next_in_same_bio = NULL;
+		bio->bi_private = block;
+		bio->bi_end_io = btrfsic_bio_end_io;
+	} else if ((dev_state->state->print_mask &
+		   (BTRFSIC_PRINT_MASK_SUBMIT_BIO_BH |
+		    BTRFSIC_PRINT_MASK_VERBOSE))) {
+		pr_info(
+"btrfsic_submit_bio(%pg) with FLUSH but dummy block already in use (ignored)!\n",
+		       dev_state->bdev);
 	}
-	mutex_unlock(&btrfsic_mutex);
-	return submit_bh(op, op_flags, bh);
 }
 
-static void __btrfsic_submit_bio(struct bio *bio)
+void btrfsic_check_bio(struct bio *bio)
 {
 	struct btrfsic_dev_state *dev_state;
 
 	if (!btrfsic_is_initialized)
 		return;
 
+	/*
+	 * We can be called before btrfsic_mount, so there might not be a
+	 * dev_state.
+	 */
+	dev_state = btrfsic_dev_state_lookup(bio->bi_bdev->bd_dev);
 	mutex_lock(&btrfsic_mutex);
-	/* since btrfsic_submit_bio() is also called before
-	 * btrfsic_mount(), this might return NULL */
-	dev_state = btrfsic_dev_state_lookup(bio_dev(bio) + bio->bi_partno);
-	if (NULL != dev_state &&
-	    (bio_op(bio) == REQ_OP_WRITE) && bio_has_data(bio)) {
-		unsigned int i = 0;
-		u64 dev_bytenr;
-		u64 cur_bytenr;
-		struct bio_vec bvec;
-		struct bvec_iter iter;
-		int bio_is_patched;
-		char **mapped_datav;
-		unsigned int segs = bio_segments(bio);
-
-		dev_bytenr = 512 * bio->bi_iter.bi_sector;
-		bio_is_patched = 0;
-		if (dev_state->state->print_mask &
-		    BTRFSIC_PRINT_MASK_SUBMIT_BIO_BH)
-			pr_info("submit_bio(rw=%d,0x%x, bi_vcnt=%u, bi_sector=%llu (bytenr %llu), bi_disk=%p)\n",
-			       bio_op(bio), bio->bi_opf, segs,
-			       (unsigned long long)bio->bi_iter.bi_sector,
-			       dev_bytenr, bio->bi_disk);
-
-		mapped_datav = kmalloc_array(segs,
-					     sizeof(*mapped_datav), GFP_NOFS);
-		if (!mapped_datav)
-			goto leave;
-		cur_bytenr = dev_bytenr;
-
-		bio_for_each_segment(bvec, bio, iter) {
-			BUG_ON(bvec.bv_len != PAGE_SIZE);
-			mapped_datav[i] = kmap(bvec.bv_page);
-			i++;
-
-			if (dev_state->state->print_mask &
-			    BTRFSIC_PRINT_MASK_SUBMIT_BIO_BH_VERBOSE)
-				pr_info("#%u: bytenr=%llu, len=%u, offset=%u\n",
-				       i, cur_bytenr, bvec.bv_len, bvec.bv_offset);
-			cur_bytenr += bvec.bv_len;
-		}
-		btrfsic_process_written_block(dev_state, dev_bytenr,
-					      mapped_datav, segs,
-					      bio, &bio_is_patched,
-					      NULL, bio->bi_opf);
-		bio_for_each_segment(bvec, bio, iter)
-			kunmap(bvec.bv_page);
-		kfree(mapped_datav);
-	} else if (NULL != dev_state && (bio->bi_opf & REQ_PREFLUSH)) {
-		if (dev_state->state->print_mask &
-		    BTRFSIC_PRINT_MASK_SUBMIT_BIO_BH)
-			pr_info("submit_bio(rw=%d,0x%x FLUSH, disk=%p)\n",
-			       bio_op(bio), bio->bi_opf, bio->bi_disk);
-		if (!dev_state->dummy_block_for_bio_bh_flush.is_iodone) {
-			if ((dev_state->state->print_mask &
-			     (BTRFSIC_PRINT_MASK_SUBMIT_BIO_BH |
-			      BTRFSIC_PRINT_MASK_VERBOSE)))
-				pr_info("btrfsic_submit_bio(%s) with FLUSH but dummy block already in use (ignored)!\n",
-				       dev_state->name);
-		} else {
-			struct btrfsic_block *const block =
-				&dev_state->dummy_block_for_bio_bh_flush;
-
-			block->is_iodone = 0;
-			block->never_written = 0;
-			block->iodone_w_error = 0;
-			block->flush_gen = dev_state->last_flush_gen + 1;
-			block->submit_bio_bh_rw = bio->bi_opf;
-			block->orig_bio_bh_private = bio->bi_private;
-			block->orig_bio_bh_end_io.bio = bio->bi_end_io;
-			block->next_in_same_bio = NULL;
-			bio->bi_private = block;
-			bio->bi_end_io = btrfsic_bio_end_io;
-		}
+	if (dev_state) {
+		if (bio_op(bio) == REQ_OP_WRITE && bio_has_data(bio))
+			btrfsic_check_write_bio(bio, dev_state);
+		else if (bio->bi_opf & REQ_PREFLUSH)
+			btrfsic_check_flush_bio(bio, dev_state);
 	}
-leave:
 	mutex_unlock(&btrfsic_mutex);
 }
 
-void btrfsic_submit_bio(struct bio *bio)
-{
-	__btrfsic_submit_bio(bio);
-	submit_bio(bio);
-}
-
-int btrfsic_submit_bio_wait(struct bio *bio)
-{
-	__btrfsic_submit_bio(bio);
-	return submit_bio_wait(bio);
-}
-
 int btrfsic_mount(struct btrfs_fs_info *fs_info,
 		  struct btrfs_fs_devices *fs_devices,
 		  int including_extent_data, u32 print_mask)
@@ -2905,10 +2741,8 @@ int btrfsic_mount(struct btrfs_fs_info *fs_info,
 		return -1;
 	}
 	state = kvzalloc(sizeof(*state), GFP_KERNEL);
-	if (!state) {
-		pr_info("btrfs check-integrity: allocation failed!\n");
+	if (!state)
 		return -ENOMEM;
-	}
 
 	if (!btrfsic_is_initialized) {
 		mutex_init(&btrfsic_mutex);
@@ -2919,7 +2753,6 @@ int btrfsic_mount(struct btrfs_fs_info *fs_info,
 	state->fs_info = fs_info;
 	state->print_mask = print_mask;
 	state->include_extent_data = including_extent_data;
-	state->csum_size = 0;
 	state->metablock_size = fs_info->nodesize;
 	state->datablock_size = fs_info->sectorsize;
 	INIT_LIST_HEAD(&state->all_blocks_list);
@@ -2930,23 +2763,17 @@ int btrfsic_mount(struct btrfs_fs_info *fs_info,
 
 	list_for_each_entry(device, dev_head, dev_list) {
 		struct btrfsic_dev_state *ds;
-		const char *p;
 
 		if (!device->bdev || !device->name)
 			continue;
 
 		ds = btrfsic_dev_state_alloc();
 		if (NULL == ds) {
-			pr_info("btrfs check-integrity: kmalloc() failed!\n");
 			mutex_unlock(&btrfsic_mutex);
 			return -ENOMEM;
 		}
 		ds->bdev = device->bdev;
 		ds->state = state;
-		bdevname(ds->bdev, ds->name);
-		ds->name[BDEVNAME_SIZE - 1] = '\0';
-		p = kbasename(ds->name);
-		strlcpy(ds->name, p, sizeof(ds->name));
 		btrfsic_dev_state_hashtable_add(ds,
 						&btrfsic_dev_state_hashtable);
 	}
@@ -3024,9 +2851,10 @@ void btrfsic_unmount(struct btrfs_fs_devices *fs_devices)
 		if (b_all->is_iodone || b_all->never_written)
 			btrfsic_block_free(b_all);
 		else
-			pr_info("btrfs: attempt to free %c-block @%llu (%s/%llu/%d) on umount which is not yet iodone!\n",
+			pr_info(
+"btrfs: attempt to free %c-block @%llu (%pg/%llu/%d) on umount which is not yet iodone!\n",
 			       btrfsic_get_block_type(state, b_all),
-			       b_all->logical_bytenr, b_all->dev_state->name,
+			       b_all->logical_bytenr, b_all->dev_state->bdev,
 			       b_all->dev_bytenr, b_all->mirror_num);
 	}
 
diff --git a/fs/btrfs/check-integrity.h b/fs/btrfs/check-integrity.h
index 9bf4359cc44c..e4c8aed7996f 100644
--- a/fs/btrfs/check-integrity.h
+++ b/fs/btrfs/check-integrity.h
@@ -7,13 +7,9 @@
 #define BTRFS_CHECK_INTEGRITY_H
 
 #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
-int btrfsic_submit_bh(int op, int op_flags, struct buffer_head *bh);
-void btrfsic_submit_bio(struct bio *bio);
-int btrfsic_submit_bio_wait(struct bio *bio);
+void btrfsic_check_bio(struct bio *bio);
 #else
-#define btrfsic_submit_bh submit_bh
-#define btrfsic_submit_bio submit_bio
-#define btrfsic_submit_bio_wait submit_bio_wait
+static inline void btrfsic_check_bio(struct bio *bio) { }
 #endif
 
 int btrfsic_mount(struct btrfs_fs_info *fs_info,
diff --git a/fs/btrfs/compression.c b/fs/btrfs/compression.c
index 9ab610cc9114..e6635fe70067 100644
--- a/fs/btrfs/compression.c
+++ b/fs/btrfs/compression.c
@@ -8,12 +8,15 @@
 #include <linux/file.h>
 #include <linux/fs.h>
 #include <linux/pagemap.h>
+#include <linux/pagevec.h>
 #include <linux/highmem.h>
+#include <linux/kthread.h>
 #include <linux/time.h>
 #include <linux/init.h>
 #include <linux/string.h>
 #include <linux/backing-dev.h>
 #include <linux/writeback.h>
+#include <linux/psi.h>
 #include <linux/slab.h>
 #include <linux/sched/mm.h>
 #include <linux/log2.h>
@@ -28,41 +31,8 @@
 #include "compression.h"
 #include "extent_io.h"
 #include "extent_map.h"
-
-int zlib_compress_pages(struct list_head *ws, struct address_space *mapping,
-		u64 start, struct page **pages, unsigned long *out_pages,
-		unsigned long *total_in, unsigned long *total_out);
-int zlib_decompress_bio(struct list_head *ws, struct compressed_bio *cb);
-int zlib_decompress(struct list_head *ws, unsigned char *data_in,
-		struct page *dest_page, unsigned long start_byte, size_t srclen,
-		size_t destlen);
-struct list_head *zlib_alloc_workspace(unsigned int level);
-void zlib_free_workspace(struct list_head *ws);
-struct list_head *zlib_get_workspace(unsigned int level);
-
-int lzo_compress_pages(struct list_head *ws, struct address_space *mapping,
-		u64 start, struct page **pages, unsigned long *out_pages,
-		unsigned long *total_in, unsigned long *total_out);
-int lzo_decompress_bio(struct list_head *ws, struct compressed_bio *cb);
-int lzo_decompress(struct list_head *ws, unsigned char *data_in,
-		struct page *dest_page, unsigned long start_byte, size_t srclen,
-		size_t destlen);
-struct list_head *lzo_alloc_workspace(unsigned int level);
-void lzo_free_workspace(struct list_head *ws);
-
-int zstd_compress_pages(struct list_head *ws, struct address_space *mapping,
-		u64 start, struct page **pages, unsigned long *out_pages,
-		unsigned long *total_in, unsigned long *total_out);
-int zstd_decompress_bio(struct list_head *ws, struct compressed_bio *cb);
-int zstd_decompress(struct list_head *ws, unsigned char *data_in,
-		struct page *dest_page, unsigned long start_byte, size_t srclen,
-		size_t destlen);
-void zstd_init_workspace_manager(void);
-void zstd_cleanup_workspace_manager(void);
-struct list_head *zstd_alloc_workspace(unsigned int level);
-void zstd_free_workspace(struct list_head *ws);
-struct list_head *zstd_get_workspace(unsigned int level);
-void zstd_put_workspace(struct list_head *ws);
+#include "subpage.h"
+#include "zoned.h"
 
 static const char* const btrfs_compress_types[] = { "", "zlib", "lzo", "zstd" };
 
@@ -115,18 +85,23 @@ static int compression_compress_pages(int type, struct list_head *ws,
 	case BTRFS_COMPRESS_NONE:
 	default:
 		/*
-		 * This can't happen, the type is validated several times
-		 * before we get here. As a sane fallback, return what the
-		 * callers will understand as 'no compression happened'.
+		 * This can happen when compression races with remount setting
+		 * it to 'no compress', while caller doesn't call
+		 * inode_need_compress() to check if we really need to
+		 * compress.
+		 *
+		 * Not a big deal, just need to inform caller that we
+		 * haven't allocated any pages yet.
 		 */
+		*out_pages = 0;
 		return -E2BIG;
 	}
 }
 
-static int compression_decompress_bio(int type, struct list_head *ws,
-		struct compressed_bio *cb)
+static int compression_decompress_bio(struct list_head *ws,
+				      struct compressed_bio *cb)
 {
-	switch (type) {
+	switch (cb->compress_type) {
 	case BTRFS_COMPRESS_ZLIB: return zlib_decompress_bio(ws, cb);
 	case BTRFS_COMPRESS_LZO:  return lzo_decompress_bio(ws, cb);
 	case BTRFS_COMPRESS_ZSTD: return zstd_decompress_bio(ws, cb);
@@ -163,146 +138,76 @@ static int compression_decompress(int type, struct list_head *ws,
 
 static int btrfs_decompress_bio(struct compressed_bio *cb);
 
-static inline int compressed_bio_size(struct btrfs_fs_info *fs_info,
-				      unsigned long disk_size)
-{
-	u16 csum_size = btrfs_super_csum_size(fs_info->super_copy);
-
-	return sizeof(struct compressed_bio) +
-		(DIV_ROUND_UP(disk_size, fs_info->sectorsize)) * csum_size;
-}
-
-static int check_compressed_csum(struct btrfs_inode *inode,
-				 struct compressed_bio *cb,
-				 u64 disk_start)
-{
-	struct btrfs_fs_info *fs_info = inode->root->fs_info;
-	SHASH_DESC_ON_STACK(shash, fs_info->csum_shash);
-	const u16 csum_size = btrfs_super_csum_size(fs_info->super_copy);
-	int ret;
-	struct page *page;
-	unsigned long i;
-	char *kaddr;
-	u8 csum[BTRFS_CSUM_SIZE];
-	u8 *cb_sum = cb->sums;
-
-	if (inode->flags & BTRFS_INODE_NODATASUM)
-		return 0;
-
-	shash->tfm = fs_info->csum_shash;
-
-	for (i = 0; i < cb->nr_pages; i++) {
-		page = cb->compressed_pages[i];
-
-		crypto_shash_init(shash);
-		kaddr = kmap_atomic(page);
-		crypto_shash_update(shash, kaddr, PAGE_SIZE);
-		kunmap_atomic(kaddr);
-		crypto_shash_final(shash, (u8 *)&csum);
-
-		if (memcmp(&csum, cb_sum, csum_size)) {
-			btrfs_print_data_csum_error(inode, disk_start,
-					csum, cb_sum, cb->mirror_num);
-			ret = -EIO;
-			goto fail;
-		}
-		cb_sum += csum_size;
-
-	}
-	ret = 0;
-fail:
-	return ret;
-}
-
-/* when we finish reading compressed pages from the disk, we
- * decompress them and then run the bio end_io routines on the
- * decompressed pages (in the inode address space).
- *
- * This allows the checksumming and other IO error handling routines
- * to work normally
- *
- * The compressed pages are freed here, and it must be run
- * in process context
- */
-static void end_compressed_bio_read(struct bio *bio)
+static void finish_compressed_bio_read(struct compressed_bio *cb)
 {
-	struct compressed_bio *cb = bio->bi_private;
-	struct inode *inode;
+	unsigned int index;
 	struct page *page;
-	unsigned long index;
-	unsigned int mirror = btrfs_io_bio(bio)->mirror_num;
-	int ret = 0;
-
-	if (bio->bi_status)
-		cb->errors = 1;
 
-	/* if there are more bios still pending for this compressed
-	 * extent, just exit
-	 */
-	if (!refcount_dec_and_test(&cb->pending_bios))
-		goto out;
+	if (cb->status == BLK_STS_OK)
+		cb->status = errno_to_blk_status(btrfs_decompress_bio(cb));
 
-	/*
-	 * Record the correct mirror_num in cb->orig_bio so that
-	 * read-repair can work properly.
-	 */
-	ASSERT(btrfs_io_bio(cb->orig_bio));
-	btrfs_io_bio(cb->orig_bio)->mirror_num = mirror;
-	cb->mirror_num = mirror;
-
-	/*
-	 * Some IO in this cb have failed, just skip checksum as there
-	 * is no way it could be correct.
-	 */
-	if (cb->errors == 1)
-		goto csum_failed;
-
-	inode = cb->inode;
-	ret = check_compressed_csum(BTRFS_I(inode), cb,
-				    (u64)bio->bi_iter.bi_sector << 9);
-	if (ret)
-		goto csum_failed;
-
-	/* ok, we're the last bio for this extent, lets start
-	 * the decompression.
-	 */
-	ret = btrfs_decompress_bio(cb);
-
-csum_failed:
-	if (ret)
-		cb->errors = 1;
-
-	/* release the compressed pages */
-	index = 0;
+	/* Release the compressed pages */
 	for (index = 0; index < cb->nr_pages; index++) {
 		page = cb->compressed_pages[index];
 		page->mapping = NULL;
 		put_page(page);
 	}
 
-	/* do io completion on the original bio */
-	if (cb->errors) {
-		bio_io_error(cb->orig_bio);
-	} else {
-		struct bio_vec *bvec;
-		struct bvec_iter_all iter_all;
+	/* Do io completion on the original bio */
+	btrfs_bio_end_io(btrfs_bio(cb->orig_bio), cb->status);
 
-		/*
-		 * we have verified the checksum already, set page
-		 * checked so the end_io handlers know about it
-		 */
-		ASSERT(!bio_flagged(bio, BIO_CLONED));
-		bio_for_each_segment_all(bvec, cb->orig_bio, iter_all)
-			SetPageChecked(bvec->bv_page);
+	/* Finally free the cb struct */
+	kfree(cb->compressed_pages);
+	kfree(cb);
+}
 
-		bio_endio(cb->orig_bio);
+/*
+ * Verify the checksums and kick off repair if needed on the uncompressed data
+ * before decompressing it into the original bio and freeing the uncompressed
+ * pages.
+ */
+static void end_compressed_bio_read(struct btrfs_bio *bbio)
+{
+	struct compressed_bio *cb = bbio->private;
+	struct inode *inode = cb->inode;
+	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+	struct btrfs_inode *bi = BTRFS_I(inode);
+	bool csum = !(bi->flags & BTRFS_INODE_NODATASUM) &&
+		    !test_bit(BTRFS_FS_STATE_NO_CSUMS, &fs_info->fs_state);
+	blk_status_t status = bbio->bio.bi_status;
+	struct bvec_iter iter;
+	struct bio_vec bv;
+	u32 offset;
+
+	btrfs_bio_for_each_sector(fs_info, bv, bbio, iter, offset) {
+		u64 start = bbio->file_offset + offset;
+
+		if (!status &&
+		    (!csum || !btrfs_check_data_csum(inode, bbio, offset,
+						     bv.bv_page, bv.bv_offset))) {
+			btrfs_clean_io_failure(bi, start, bv.bv_page,
+					       bv.bv_offset);
+		} else {
+			int ret;
+
+			refcount_inc(&cb->pending_ios);
+			ret = btrfs_repair_one_sector(inode, bbio, offset,
+						      bv.bv_page, bv.bv_offset,
+						      btrfs_submit_data_read_bio);
+			if (ret) {
+				refcount_dec(&cb->pending_ios);
+				status = errno_to_blk_status(ret);
+			}
+		}
 	}
 
-	/* finally free the cb struct */
-	kfree(cb->compressed_pages);
-	kfree(cb);
-out:
-	bio_put(bio);
+	if (status)
+		cb->status = status;
+
+	if (refcount_dec_and_test(&cb->pending_ios))
+		finish_compressed_bio_read(cb);
+	btrfs_bio_free_csum(bbio);
+	bio_put(&bbio->bio);
 }
 
 /*
@@ -312,90 +217,151 @@ out:
 static noinline void end_compressed_writeback(struct inode *inode,
 					      const struct compressed_bio *cb)
 {
+	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	unsigned long index = cb->start >> PAGE_SHIFT;
 	unsigned long end_index = (cb->start + cb->len - 1) >> PAGE_SHIFT;
-	struct page *pages[16];
-	unsigned long nr_pages = end_index - index + 1;
+	struct folio_batch fbatch;
+	const int errno = blk_status_to_errno(cb->status);
 	int i;
 	int ret;
 
-	if (cb->errors)
-		mapping_set_error(inode->i_mapping, -EIO);
+	if (errno)
+		mapping_set_error(inode->i_mapping, errno);
+
+	folio_batch_init(&fbatch);
+	while (index <= end_index) {
+		ret = filemap_get_folios(inode->i_mapping, &index, end_index,
+				&fbatch);
+
+		if (ret == 0)
+			return;
 
-	while (nr_pages > 0) {
-		ret = find_get_pages_contig(inode->i_mapping, index,
-				     min_t(unsigned long,
-				     nr_pages, ARRAY_SIZE(pages)), pages);
-		if (ret == 0) {
-			nr_pages -= 1;
-			index += 1;
-			continue;
-		}
 		for (i = 0; i < ret; i++) {
-			if (cb->errors)
-				SetPageError(pages[i]);
-			end_page_writeback(pages[i]);
-			put_page(pages[i]);
+			struct folio *folio = fbatch.folios[i];
+
+			if (errno)
+				folio_set_error(folio);
+			btrfs_page_clamp_clear_writeback(fs_info, &folio->page,
+							 cb->start, cb->len);
 		}
-		nr_pages -= ret;
-		index += ret;
+		folio_batch_release(&fbatch);
 	}
 	/* the inode may be gone now */
 }
 
-/*
- * do the cleanup once all the compressed pages hit the disk.
- * This will clear writeback on the file pages and free the compressed
- * pages.
- *
- * This also calls the writeback end hooks for the file pages so that
- * metadata and checksums can be updated in the file.
- */
-static void end_compressed_bio_write(struct bio *bio)
+static void finish_compressed_bio_write(struct compressed_bio *cb)
 {
-	struct compressed_bio *cb = bio->bi_private;
-	struct inode *inode;
-	struct page *page;
-	unsigned long index;
-
-	if (bio->bi_status)
-		cb->errors = 1;
+	struct inode *inode = cb->inode;
+	unsigned int index;
 
-	/* if there are more bios still pending for this compressed
-	 * extent, just exit
-	 */
-	if (!refcount_dec_and_test(&cb->pending_bios))
-		goto out;
-
-	/* ok, we're the last bio for this extent, step one is to
-	 * call back into the FS and do all the end_io operations
+	/*
+	 * Ok, we're the last bio for this extent, step one is to call back
+	 * into the FS and do all the end_io operations.
 	 */
-	inode = cb->inode;
-	cb->compressed_pages[0]->mapping = cb->inode->i_mapping;
-	btrfs_writepage_endio_finish_ordered(cb->compressed_pages[0],
+	btrfs_writepage_endio_finish_ordered(BTRFS_I(inode), NULL,
 			cb->start, cb->start + cb->len - 1,
-			bio->bi_status == BLK_STS_OK);
-	cb->compressed_pages[0]->mapping = NULL;
+			cb->status == BLK_STS_OK);
 
-	end_compressed_writeback(inode, cb);
-	/* note, our inode could be gone now */
+	if (cb->writeback)
+		end_compressed_writeback(inode, cb);
+	/* Note, our inode could be gone now */
 
 	/*
-	 * release the compressed pages, these came from alloc_page and
+	 * Release the compressed pages, these came from alloc_page and
 	 * are not attached to the inode at all
 	 */
-	index = 0;
 	for (index = 0; index < cb->nr_pages; index++) {
-		page = cb->compressed_pages[index];
+		struct page *page = cb->compressed_pages[index];
+
 		page->mapping = NULL;
 		put_page(page);
 	}
 
-	/* finally free the cb struct */
+	/* Finally free the cb struct */
 	kfree(cb->compressed_pages);
 	kfree(cb);
-out:
-	bio_put(bio);
+}
+
+static void btrfs_finish_compressed_write_work(struct work_struct *work)
+{
+	struct compressed_bio *cb =
+		container_of(work, struct compressed_bio, write_end_work);
+
+	finish_compressed_bio_write(cb);
+}
+
+/*
+ * Do the cleanup once all the compressed pages hit the disk.  This will clear
+ * writeback on the file pages and free the compressed pages.
+ *
+ * This also calls the writeback end hooks for the file pages so that metadata
+ * and checksums can be updated in the file.
+ */
+static void end_compressed_bio_write(struct btrfs_bio *bbio)
+{
+	struct compressed_bio *cb = bbio->private;
+
+	if (bbio->bio.bi_status)
+		cb->status = bbio->bio.bi_status;
+
+	if (refcount_dec_and_test(&cb->pending_ios)) {
+		struct btrfs_fs_info *fs_info = btrfs_sb(cb->inode->i_sb);
+
+		btrfs_record_physical_zoned(cb->inode, cb->start, &bbio->bio);
+		queue_work(fs_info->compressed_write_workers, &cb->write_end_work);
+	}
+	bio_put(&bbio->bio);
+}
+
+/*
+ * Allocate a compressed_bio, which will be used to read/write on-disk
+ * (aka, compressed) * data.
+ *
+ * @cb:                 The compressed_bio structure, which records all the needed
+ *                      information to bind the compressed data to the uncompressed
+ *                      page cache.
+ * @disk_byten:         The logical bytenr where the compressed data will be read
+ *                      from or written to.
+ * @endio_func:         The endio function to call after the IO for compressed data
+ *                      is finished.
+ * @next_stripe_start:  Return value of logical bytenr of where next stripe starts.
+ *                      Let the caller know to only fill the bio up to the stripe
+ *                      boundary.
+ */
+
+
+static struct bio *alloc_compressed_bio(struct compressed_bio *cb, u64 disk_bytenr,
+					blk_opf_t opf,
+					btrfs_bio_end_io_t endio_func,
+					u64 *next_stripe_start)
+{
+	struct btrfs_fs_info *fs_info = btrfs_sb(cb->inode->i_sb);
+	struct btrfs_io_geometry geom;
+	struct extent_map *em;
+	struct bio *bio;
+	int ret;
+
+	bio = btrfs_bio_alloc(BIO_MAX_VECS, opf, endio_func, cb);
+	bio->bi_iter.bi_sector = disk_bytenr >> SECTOR_SHIFT;
+
+	em = btrfs_get_chunk_map(fs_info, disk_bytenr, fs_info->sectorsize);
+	if (IS_ERR(em)) {
+		bio_put(bio);
+		return ERR_CAST(em);
+	}
+
+	if (bio_op(bio) == REQ_OP_ZONE_APPEND)
+		bio_set_dev(bio, em->map_lookup->stripes[0].dev->bdev);
+
+	ret = btrfs_get_io_geometry(fs_info, em, btrfs_op(bio), disk_bytenr, &geom);
+	free_extent_map(em);
+	if (ret < 0) {
+		bio_put(bio);
+		return ERR_PTR(ret);
+	}
+	*next_stripe_start = disk_bytenr + geom.len;
+	refcount_inc(&cb->pending_ios);
+	return bio;
 }
 
 /*
@@ -407,122 +373,122 @@ out:
  * This also checksums the file bytes and gets things ready for
  * the end io hooks.
  */
-blk_status_t btrfs_submit_compressed_write(struct inode *inode, u64 start,
-				 unsigned long len, u64 disk_start,
-				 unsigned long compressed_len,
+blk_status_t btrfs_submit_compressed_write(struct btrfs_inode *inode, u64 start,
+				 unsigned int len, u64 disk_start,
+				 unsigned int compressed_len,
 				 struct page **compressed_pages,
-				 unsigned long nr_pages,
-				 unsigned int write_flags,
-				 struct cgroup_subsys_state *blkcg_css)
+				 unsigned int nr_pages,
+				 blk_opf_t write_flags,
+				 struct cgroup_subsys_state *blkcg_css,
+				 bool writeback)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
 	struct bio *bio = NULL;
 	struct compressed_bio *cb;
-	unsigned long bytes_left;
-	int pg_index = 0;
-	struct page *page;
-	u64 first_byte = disk_start;
-	blk_status_t ret;
-	int skip_sum = BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM;
-
-	WARN_ON(!PAGE_ALIGNED(start));
-	cb = kmalloc(compressed_bio_size(fs_info, compressed_len), GFP_NOFS);
+	u64 cur_disk_bytenr = disk_start;
+	u64 next_stripe_start;
+	blk_status_t ret = BLK_STS_OK;
+	int skip_sum = inode->flags & BTRFS_INODE_NODATASUM;
+	const bool use_append = btrfs_use_zone_append(inode, disk_start);
+	const enum req_op bio_op = use_append ? REQ_OP_ZONE_APPEND : REQ_OP_WRITE;
+
+	ASSERT(IS_ALIGNED(start, fs_info->sectorsize) &&
+	       IS_ALIGNED(len, fs_info->sectorsize));
+	cb = kmalloc(sizeof(struct compressed_bio), GFP_NOFS);
 	if (!cb)
 		return BLK_STS_RESOURCE;
-	refcount_set(&cb->pending_bios, 0);
-	cb->errors = 0;
-	cb->inode = inode;
+	refcount_set(&cb->pending_ios, 1);
+	cb->status = BLK_STS_OK;
+	cb->inode = &inode->vfs_inode;
 	cb->start = start;
 	cb->len = len;
-	cb->mirror_num = 0;
 	cb->compressed_pages = compressed_pages;
 	cb->compressed_len = compressed_len;
-	cb->orig_bio = NULL;
+	cb->writeback = writeback;
+	INIT_WORK(&cb->write_end_work, btrfs_finish_compressed_write_work);
 	cb->nr_pages = nr_pages;
 
-	bio = btrfs_bio_alloc(first_byte);
-	bio->bi_opf = REQ_OP_WRITE | write_flags;
-	bio->bi_private = cb;
-	bio->bi_end_io = end_compressed_bio_write;
-
-	if (blkcg_css) {
-		bio->bi_opf |= REQ_CGROUP_PUNT;
+	if (blkcg_css)
 		kthread_associate_blkcg(blkcg_css);
-	}
-	refcount_set(&cb->pending_bios, 1);
 
-	/* create and submit bios for the compressed pages */
-	bytes_left = compressed_len;
-	for (pg_index = 0; pg_index < cb->nr_pages; pg_index++) {
-		int submit = 0;
-
-		page = compressed_pages[pg_index];
-		page->mapping = inode->i_mapping;
-		if (bio->bi_iter.bi_size)
-			submit = btrfs_bio_fits_in_stripe(page, PAGE_SIZE, bio,
-							  0);
-
-		page->mapping = NULL;
-		if (submit || bio_add_page(bio, page, PAGE_SIZE, 0) <
-		    PAGE_SIZE) {
-			/*
-			 * inc the count before we submit the bio so
-			 * we know the end IO handler won't happen before
-			 * we inc the count.  Otherwise, the cb might get
-			 * freed before we're done setting it up
-			 */
-			refcount_inc(&cb->pending_bios);
-			ret = btrfs_bio_wq_end_io(fs_info, bio,
-						  BTRFS_WQ_ENDIO_DATA);
-			BUG_ON(ret); /* -ENOMEM */
-
-			if (!skip_sum) {
-				ret = btrfs_csum_one_bio(inode, bio, start, 1);
-				BUG_ON(ret); /* -ENOMEM */
-			}
-
-			ret = btrfs_map_bio(fs_info, bio, 0);
-			if (ret) {
-				bio->bi_status = ret;
-				bio_endio(bio);
+	while (cur_disk_bytenr < disk_start + compressed_len) {
+		u64 offset = cur_disk_bytenr - disk_start;
+		unsigned int index = offset >> PAGE_SHIFT;
+		unsigned int real_size;
+		unsigned int added;
+		struct page *page = compressed_pages[index];
+		bool submit = false;
+
+		/* Allocate new bio if submitted or not yet allocated */
+		if (!bio) {
+			bio = alloc_compressed_bio(cb, cur_disk_bytenr,
+				bio_op | write_flags, end_compressed_bio_write,
+				&next_stripe_start);
+			if (IS_ERR(bio)) {
+				ret = errno_to_blk_status(PTR_ERR(bio));
+				break;
 			}
-
-			bio = btrfs_bio_alloc(first_byte);
-			bio->bi_opf = REQ_OP_WRITE | write_flags;
-			bio->bi_private = cb;
-			bio->bi_end_io = end_compressed_bio_write;
 			if (blkcg_css)
 				bio->bi_opf |= REQ_CGROUP_PUNT;
-			bio_add_page(bio, page, PAGE_SIZE, 0);
-		}
-		if (bytes_left < PAGE_SIZE) {
-			btrfs_info(fs_info,
-					"bytes left %lu compress len %lu nr %lu",
-			       bytes_left, cb->compressed_len, cb->nr_pages);
 		}
-		bytes_left -= PAGE_SIZE;
-		first_byte += PAGE_SIZE;
-		cond_resched();
-	}
-
-	ret = btrfs_bio_wq_end_io(fs_info, bio, BTRFS_WQ_ENDIO_DATA);
-	BUG_ON(ret); /* -ENOMEM */
+		/*
+		 * We should never reach next_stripe_start start as we will
+		 * submit comp_bio when reach the boundary immediately.
+		 */
+		ASSERT(cur_disk_bytenr != next_stripe_start);
 
-	if (!skip_sum) {
-		ret = btrfs_csum_one_bio(inode, bio, start, 1);
-		BUG_ON(ret); /* -ENOMEM */
-	}
+		/*
+		 * We have various limits on the real read size:
+		 * - stripe boundary
+		 * - page boundary
+		 * - compressed length boundary
+		 */
+		real_size = min_t(u64, U32_MAX, next_stripe_start - cur_disk_bytenr);
+		real_size = min_t(u64, real_size, PAGE_SIZE - offset_in_page(offset));
+		real_size = min_t(u64, real_size, compressed_len - offset);
+		ASSERT(IS_ALIGNED(real_size, fs_info->sectorsize));
+
+		if (use_append)
+			added = bio_add_zone_append_page(bio, page, real_size,
+					offset_in_page(offset));
+		else
+			added = bio_add_page(bio, page, real_size,
+					offset_in_page(offset));
+		/* Reached zoned boundary */
+		if (added == 0)
+			submit = true;
+
+		cur_disk_bytenr += added;
+		/* Reached stripe boundary */
+		if (cur_disk_bytenr == next_stripe_start)
+			submit = true;
+
+		/* Finished the range */
+		if (cur_disk_bytenr == disk_start + compressed_len)
+			submit = true;
+
+		if (submit) {
+			if (!skip_sum) {
+				ret = btrfs_csum_one_bio(inode, bio, start, true);
+				if (ret) {
+					btrfs_bio_end_io(btrfs_bio(bio), ret);
+					break;
+				}
+			}
 
-	ret = btrfs_map_bio(fs_info, bio, 0);
-	if (ret) {
-		bio->bi_status = ret;
-		bio_endio(bio);
+			ASSERT(bio->bi_iter.bi_size);
+			btrfs_submit_bio(fs_info, bio, 0);
+			bio = NULL;
+		}
+		cond_resched();
 	}
 
 	if (blkcg_css)
 		kthread_associate_blkcg(NULL);
 
-	return 0;
+	if (refcount_dec_and_test(&cb->pending_ios))
+		finish_compressed_bio_write(cb);
+	return ret;
 }
 
 static u64 bio_end_offset(struct bio *bio)
@@ -532,45 +498,75 @@ static u64 bio_end_offset(struct bio *bio)
 	return page_offset(last->bv_page) + last->bv_len + last->bv_offset;
 }
 
+/*
+ * Add extra pages in the same compressed file extent so that we don't need to
+ * re-read the same extent again and again.
+ *
+ * NOTE: this won't work well for subpage, as for subpage read, we lock the
+ * full page then submit bio for each compressed/regular extents.
+ *
+ * This means, if we have several sectors in the same page points to the same
+ * on-disk compressed data, we will re-read the same extent many times and
+ * this function can only help for the next page.
+ */
 static noinline int add_ra_bio_pages(struct inode *inode,
 				     u64 compressed_end,
-				     struct compressed_bio *cb)
+				     struct compressed_bio *cb,
+				     int *memstall, unsigned long *pflags)
 {
+	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	unsigned long end_index;
-	unsigned long pg_index;
-	u64 last_offset;
+	u64 cur = bio_end_offset(cb->orig_bio);
 	u64 isize = i_size_read(inode);
 	int ret;
 	struct page *page;
-	unsigned long nr_pages = 0;
 	struct extent_map *em;
 	struct address_space *mapping = inode->i_mapping;
 	struct extent_map_tree *em_tree;
 	struct extent_io_tree *tree;
-	u64 end;
-	int misses = 0;
+	int sectors_missed = 0;
 
-	last_offset = bio_end_offset(cb->orig_bio);
 	em_tree = &BTRFS_I(inode)->extent_tree;
 	tree = &BTRFS_I(inode)->io_tree;
 
 	if (isize == 0)
 		return 0;
 
+	/*
+	 * For current subpage support, we only support 64K page size,
+	 * which means maximum compressed extent size (128K) is just 2x page
+	 * size.
+	 * This makes readahead less effective, so here disable readahead for
+	 * subpage for now, until full compressed write is supported.
+	 */
+	if (btrfs_sb(inode->i_sb)->sectorsize < PAGE_SIZE)
+		return 0;
+
 	end_index = (i_size_read(inode) - 1) >> PAGE_SHIFT;
 
-	while (last_offset < compressed_end) {
-		pg_index = last_offset >> PAGE_SHIFT;
+	while (cur < compressed_end) {
+		u64 page_end;
+		u64 pg_index = cur >> PAGE_SHIFT;
+		u32 add_size;
 
 		if (pg_index > end_index)
 			break;
 
 		page = xa_load(&mapping->i_pages, pg_index);
 		if (page && !xa_is_value(page)) {
-			misses++;
-			if (misses > 4)
+			sectors_missed += (PAGE_SIZE - offset_in_page(cur)) >>
+					  fs_info->sectorsize_bits;
+
+			/* Beyond threshold, no need to continue */
+			if (sectors_missed > 4)
 				break;
-			goto next;
+
+			/*
+			 * Jump to next page start as we already have page for
+			 * current offset.
+			 */
+			cur = (pg_index << PAGE_SHIFT) + PAGE_SIZE;
+			continue;
 		}
 
 		page = __page_cache_alloc(mapping_gfp_constraint(mapping,
@@ -580,27 +576,39 @@ static noinline int add_ra_bio_pages(struct inode *inode,
 
 		if (add_to_page_cache_lru(page, mapping, pg_index, GFP_NOFS)) {
 			put_page(page);
-			goto next;
+			/* There is already a page, skip to page end */
+			cur = (pg_index << PAGE_SHIFT) + PAGE_SIZE;
+			continue;
 		}
 
-		end = last_offset + PAGE_SIZE - 1;
-		/*
-		 * at this point, we have a locked page in the page cache
-		 * for these bytes in the file.  But, we have to make
-		 * sure they map to this compressed extent on disk.
-		 */
-		set_page_extent_mapped(page);
-		lock_extent(tree, last_offset, end);
+		if (!*memstall && PageWorkingset(page)) {
+			psi_memstall_enter(pflags);
+			*memstall = 1;
+		}
+
+		ret = set_page_extent_mapped(page);
+		if (ret < 0) {
+			unlock_page(page);
+			put_page(page);
+			break;
+		}
+
+		page_end = (pg_index << PAGE_SHIFT) + PAGE_SIZE - 1;
+		lock_extent(tree, cur, page_end, NULL);
 		read_lock(&em_tree->lock);
-		em = lookup_extent_mapping(em_tree, last_offset,
-					   PAGE_SIZE);
+		em = lookup_extent_mapping(em_tree, cur, page_end + 1 - cur);
 		read_unlock(&em_tree->lock);
 
-		if (!em || last_offset < em->start ||
-		    (last_offset + PAGE_SIZE > extent_map_end(em)) ||
+		/*
+		 * At this point, we have a locked page in the page cache for
+		 * these bytes in the file.  But, we have to make sure they map
+		 * to this compressed extent on disk.
+		 */
+		if (!em || cur < em->start ||
+		    (cur + fs_info->sectorsize > extent_map_end(em)) ||
 		    (em->block_start >> 9) != cb->orig_bio->bi_iter.bi_sector) {
 			free_extent_map(em);
-			unlock_extent(tree, last_offset, end);
+			unlock_extent(tree, cur, page_end, NULL);
 			unlock_page(page);
 			put_page(page);
 			break;
@@ -608,33 +616,32 @@ static noinline int add_ra_bio_pages(struct inode *inode,
 		free_extent_map(em);
 
 		if (page->index == end_index) {
-			char *userpage;
 			size_t zero_offset = offset_in_page(isize);
 
 			if (zero_offset) {
 				int zeros;
 				zeros = PAGE_SIZE - zero_offset;
-				userpage = kmap_atomic(page);
-				memset(userpage + zero_offset, 0, zeros);
-				flush_dcache_page(page);
-				kunmap_atomic(userpage);
+				memzero_page(page, zero_offset, zeros);
 			}
 		}
 
-		ret = bio_add_page(cb->orig_bio, page,
-				   PAGE_SIZE, 0);
-
-		if (ret == PAGE_SIZE) {
-			nr_pages++;
-			put_page(page);
-		} else {
-			unlock_extent(tree, last_offset, end);
+		add_size = min(em->start + em->len, page_end + 1) - cur;
+		ret = bio_add_page(cb->orig_bio, page, add_size, offset_in_page(cur));
+		if (ret != add_size) {
+			unlock_extent(tree, cur, page_end, NULL);
 			unlock_page(page);
 			put_page(page);
 			break;
 		}
-next:
-		last_offset += PAGE_SIZE;
+		/*
+		 * If it's subpage, we also need to increase its
+		 * subpage::readers number, as at endio we will decrease
+		 * subpage::readers and to unlock the page.
+		 */
+		if (fs_info->sectorsize < PAGE_SIZE)
+			btrfs_subpage_start_reader(fs_info, page, cur, add_size);
+		put_page(page);
+		cur += add_size;
 	}
 	return 0;
 }
@@ -650,171 +657,179 @@ next:
  * After the compressed pages are read, we copy the bytes into the
  * bio we were passed and then call the bio end_io calls
  */
-blk_status_t btrfs_submit_compressed_read(struct inode *inode, struct bio *bio,
-				 int mirror_num, unsigned long bio_flags)
+void btrfs_submit_compressed_read(struct inode *inode, struct bio *bio,
+				  int mirror_num)
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	struct extent_map_tree *em_tree;
 	struct compressed_bio *cb;
-	unsigned long compressed_len;
-	unsigned long nr_pages;
-	unsigned long pg_index;
-	struct page *page;
-	struct bio *comp_bio;
-	u64 cur_disk_byte = (u64)bio->bi_iter.bi_sector << 9;
+	unsigned int compressed_len;
+	struct bio *comp_bio = NULL;
+	const u64 disk_bytenr = bio->bi_iter.bi_sector << SECTOR_SHIFT;
+	u64 cur_disk_byte = disk_bytenr;
+	u64 next_stripe_start;
+	u64 file_offset;
 	u64 em_len;
 	u64 em_start;
 	struct extent_map *em;
-	blk_status_t ret = BLK_STS_RESOURCE;
-	int faili = 0;
-	const u16 csum_size = btrfs_super_csum_size(fs_info->super_copy);
-	u8 *sums;
+	unsigned long pflags;
+	int memstall = 0;
+	blk_status_t ret;
+	int ret2;
+	int i;
 
 	em_tree = &BTRFS_I(inode)->extent_tree;
 
+	file_offset = bio_first_bvec_all(bio)->bv_offset +
+		      page_offset(bio_first_page_all(bio));
+
 	/* we need the actual starting offset of this extent in the file */
 	read_lock(&em_tree->lock);
-	em = lookup_extent_mapping(em_tree,
-				   page_offset(bio_first_page_all(bio)),
-				   PAGE_SIZE);
+	em = lookup_extent_mapping(em_tree, file_offset, fs_info->sectorsize);
 	read_unlock(&em_tree->lock);
-	if (!em)
-		return BLK_STS_IOERR;
+	if (!em) {
+		ret = BLK_STS_IOERR;
+		goto out;
+	}
 
+	ASSERT(em->compress_type != BTRFS_COMPRESS_NONE);
 	compressed_len = em->block_len;
-	cb = kmalloc(compressed_bio_size(fs_info, compressed_len), GFP_NOFS);
-	if (!cb)
+	cb = kmalloc(sizeof(struct compressed_bio), GFP_NOFS);
+	if (!cb) {
+		ret = BLK_STS_RESOURCE;
 		goto out;
+	}
 
-	refcount_set(&cb->pending_bios, 0);
-	cb->errors = 0;
+	refcount_set(&cb->pending_ios, 1);
+	cb->status = BLK_STS_OK;
 	cb->inode = inode;
-	cb->mirror_num = mirror_num;
-	sums = cb->sums;
 
 	cb->start = em->orig_start;
 	em_len = em->len;
 	em_start = em->start;
 
-	free_extent_map(em);
-	em = NULL;
-
 	cb->len = bio->bi_iter.bi_size;
 	cb->compressed_len = compressed_len;
-	cb->compress_type = extent_compress_type(bio_flags);
+	cb->compress_type = em->compress_type;
 	cb->orig_bio = bio;
 
-	nr_pages = DIV_ROUND_UP(compressed_len, PAGE_SIZE);
-	cb->compressed_pages = kcalloc(nr_pages, sizeof(struct page *),
-				       GFP_NOFS);
-	if (!cb->compressed_pages)
-		goto fail1;
-
-	for (pg_index = 0; pg_index < nr_pages; pg_index++) {
-		cb->compressed_pages[pg_index] = alloc_page(GFP_NOFS |
-							      __GFP_HIGHMEM);
-		if (!cb->compressed_pages[pg_index]) {
-			faili = pg_index - 1;
-			ret = BLK_STS_RESOURCE;
-			goto fail2;
-		}
+	free_extent_map(em);
+	em = NULL;
+
+	cb->nr_pages = DIV_ROUND_UP(compressed_len, PAGE_SIZE);
+	cb->compressed_pages = kcalloc(cb->nr_pages, sizeof(struct page *), GFP_NOFS);
+	if (!cb->compressed_pages) {
+		ret = BLK_STS_RESOURCE;
+		goto fail;
 	}
-	faili = nr_pages - 1;
-	cb->nr_pages = nr_pages;
 
-	add_ra_bio_pages(inode, em_start + em_len, cb);
+	ret2 = btrfs_alloc_page_array(cb->nr_pages, cb->compressed_pages);
+	if (ret2) {
+		ret = BLK_STS_RESOURCE;
+		goto fail;
+	}
+
+	add_ra_bio_pages(inode, em_start + em_len, cb, &memstall, &pflags);
 
 	/* include any pages we added in add_ra-bio_pages */
 	cb->len = bio->bi_iter.bi_size;
 
-	comp_bio = btrfs_bio_alloc(cur_disk_byte);
-	comp_bio->bi_opf = REQ_OP_READ;
-	comp_bio->bi_private = cb;
-	comp_bio->bi_end_io = end_compressed_bio_read;
-	refcount_set(&cb->pending_bios, 1);
-
-	for (pg_index = 0; pg_index < nr_pages; pg_index++) {
-		int submit = 0;
+	while (cur_disk_byte < disk_bytenr + compressed_len) {
+		u64 offset = cur_disk_byte - disk_bytenr;
+		unsigned int index = offset >> PAGE_SHIFT;
+		unsigned int real_size;
+		unsigned int added;
+		struct page *page = cb->compressed_pages[index];
+		bool submit = false;
+
+		/* Allocate new bio if submitted or not yet allocated */
+		if (!comp_bio) {
+			comp_bio = alloc_compressed_bio(cb, cur_disk_byte,
+					REQ_OP_READ, end_compressed_bio_read,
+					&next_stripe_start);
+			if (IS_ERR(comp_bio)) {
+				cb->status = errno_to_blk_status(PTR_ERR(comp_bio));
+				break;
+			}
+		}
+		/*
+		 * We should never reach next_stripe_start start as we will
+		 * submit comp_bio when reach the boundary immediately.
+		 */
+		ASSERT(cur_disk_byte != next_stripe_start);
+		/*
+		 * We have various limit on the real read size:
+		 * - stripe boundary
+		 * - page boundary
+		 * - compressed length boundary
+		 */
+		real_size = min_t(u64, U32_MAX, next_stripe_start - cur_disk_byte);
+		real_size = min_t(u64, real_size, PAGE_SIZE - offset_in_page(offset));
+		real_size = min_t(u64, real_size, compressed_len - offset);
+		ASSERT(IS_ALIGNED(real_size, fs_info->sectorsize));
 
-		page = cb->compressed_pages[pg_index];
-		page->mapping = inode->i_mapping;
-		page->index = em_start >> PAGE_SHIFT;
+		added = bio_add_page(comp_bio, page, real_size, offset_in_page(offset));
+		/*
+		 * Maximum compressed extent is smaller than bio size limit,
+		 * thus bio_add_page() should always success.
+		 */
+		ASSERT(added == real_size);
+		cur_disk_byte += added;
 
-		if (comp_bio->bi_iter.bi_size)
-			submit = btrfs_bio_fits_in_stripe(page, PAGE_SIZE,
-							  comp_bio, 0);
+		/* Reached stripe boundary, need to submit */
+		if (cur_disk_byte == next_stripe_start)
+			submit = true;
 
-		page->mapping = NULL;
-		if (submit || bio_add_page(comp_bio, page, PAGE_SIZE, 0) <
-		    PAGE_SIZE) {
-			unsigned int nr_sectors;
+		/* Has finished the range, need to submit */
+		if (cur_disk_byte == disk_bytenr + compressed_len)
+			submit = true;
 
-			ret = btrfs_bio_wq_end_io(fs_info, comp_bio,
-						  BTRFS_WQ_ENDIO_DATA);
-			BUG_ON(ret); /* -ENOMEM */
+		if (submit) {
+			/* Save the original iter for read repair */
+			if (bio_op(comp_bio) == REQ_OP_READ)
+				btrfs_bio(comp_bio)->iter = comp_bio->bi_iter;
 
 			/*
-			 * inc the count before we submit the bio so
-			 * we know the end IO handler won't happen before
-			 * we inc the count.  Otherwise, the cb might get
-			 * freed before we're done setting it up
+			 * Save the initial offset of this chunk, as there
+			 * is no direct correlation between compressed pages and
+			 * the original file offset.  The field is only used for
+			 * priting error messages.
 			 */
-			refcount_inc(&cb->pending_bios);
-
-			if (!(BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM)) {
-				ret = btrfs_lookup_bio_sums(inode, comp_bio,
-							    (u64)-1, sums);
-				BUG_ON(ret); /* -ENOMEM */
-			}
+			btrfs_bio(comp_bio)->file_offset = file_offset;
 
-			nr_sectors = DIV_ROUND_UP(comp_bio->bi_iter.bi_size,
-						  fs_info->sectorsize);
-			sums += csum_size * nr_sectors;
-
-			ret = btrfs_map_bio(fs_info, comp_bio, mirror_num);
+			ret = btrfs_lookup_bio_sums(inode, comp_bio, NULL);
 			if (ret) {
-				comp_bio->bi_status = ret;
-				bio_endio(comp_bio);
+				btrfs_bio_end_io(btrfs_bio(comp_bio), ret);
+				break;
 			}
 
-			comp_bio = btrfs_bio_alloc(cur_disk_byte);
-			comp_bio->bi_opf = REQ_OP_READ;
-			comp_bio->bi_private = cb;
-			comp_bio->bi_end_io = end_compressed_bio_read;
-
-			bio_add_page(comp_bio, page, PAGE_SIZE, 0);
+			ASSERT(comp_bio->bi_iter.bi_size);
+			btrfs_submit_bio(fs_info, comp_bio, mirror_num);
+			comp_bio = NULL;
 		}
-		cur_disk_byte += PAGE_SIZE;
 	}
 
-	ret = btrfs_bio_wq_end_io(fs_info, comp_bio, BTRFS_WQ_ENDIO_DATA);
-	BUG_ON(ret); /* -ENOMEM */
+	if (memstall)
+		psi_memstall_leave(&pflags);
 
-	if (!(BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM)) {
-		ret = btrfs_lookup_bio_sums(inode, comp_bio, (u64)-1, sums);
-		BUG_ON(ret); /* -ENOMEM */
-	}
+	if (refcount_dec_and_test(&cb->pending_ios))
+		finish_compressed_bio_read(cb);
+	return;
 
-	ret = btrfs_map_bio(fs_info, comp_bio, mirror_num);
-	if (ret) {
-		comp_bio->bi_status = ret;
-		bio_endio(comp_bio);
-	}
-
-	return 0;
-
-fail2:
-	while (faili >= 0) {
-		__free_page(cb->compressed_pages[faili]);
-		faili--;
+fail:
+	if (cb->compressed_pages) {
+		for (i = 0; i < cb->nr_pages; i++) {
+			if (cb->compressed_pages[i])
+				__free_page(cb->compressed_pages[i]);
+		}
 	}
 
 	kfree(cb->compressed_pages);
-fail1:
 	kfree(cb);
 out:
 	free_extent_map(em);
-	return ret;
+	btrfs_bio_end_io(btrfs_bio(bio), ret);
+	return;
 }
 
 /*
@@ -1142,6 +1157,22 @@ static void put_workspace(int type, struct list_head *ws)
 }
 
 /*
+ * Adjust @level according to the limits of the compression algorithm or
+ * fallback to default
+ */
+static unsigned int btrfs_compress_set_level(int type, unsigned level)
+{
+	const struct btrfs_compress_op *ops = btrfs_compress_op[type];
+
+	if (level == 0)
+		level = ops->default_level;
+	else
+		level = min(level, ops->max_level);
+
+	return level;
+}
+
+/*
  * Given an address space and start and length, compress the bytes into @pages
  * that are allocated on demand.
  *
@@ -1160,9 +1191,6 @@ static void put_workspace(int type, struct list_head *ws)
  *
  * @total_out is an in/out parameter, must be set to the input length and will
  * be also used to return the total number of compressed bytes
- *
- * @max_out tells us the max number of bytes that we're allowed to
- * stuff into pages
  */
 int btrfs_compress_pages(unsigned int type_level, struct address_space *mapping,
 			 u64 start, struct page **pages,
@@ -1183,20 +1211,6 @@ int btrfs_compress_pages(unsigned int type_level, struct address_space *mapping,
 	return ret;
 }
 
-/*
- * pages_in is an array of pages with compressed data.
- *
- * disk_start is the starting logical offset of this array in the file
- *
- * orig_bio contains the pages from the file that we want to decompress into
- *
- * srclen is the number of bytes in pages_in
- *
- * The basic idea is that we have a bio that was created by readpages.
- * The pages in the bio are for the uncompressed data, and they may not
- * be contiguous.  They all correspond to the range of bytes covered by
- * the compressed extent.
- */
 static int btrfs_decompress_bio(struct compressed_bio *cb)
 {
 	struct list_head *workspace;
@@ -1204,7 +1218,7 @@ static int btrfs_decompress_bio(struct compressed_bio *cb)
 	int type = cb->compress_type;
 
 	workspace = get_workspace(type, 0);
-	ret = compression_decompress_bio(type, workspace, cb);
+	ret = compression_decompress_bio(workspace, cb);
 	put_workspace(type, workspace);
 
 	return ret;
@@ -1246,98 +1260,81 @@ void __cold btrfs_exit_compress(void)
 }
 
 /*
- * Copy uncompressed data from working buffer to pages.
+ * Copy decompressed data from working buffer to pages.
+ *
+ * @buf:		The decompressed data buffer
+ * @buf_len:		The decompressed data length
+ * @decompressed:	Number of bytes that are already decompressed inside the
+ * 			compressed extent
+ * @cb:			The compressed extent descriptor
+ * @orig_bio:		The original bio that the caller wants to read for
+ *
+ * An easier to understand graph is like below:
+ *
+ * 		|<- orig_bio ->|     |<- orig_bio->|
+ * 	|<-------      full decompressed extent      ----->|
+ * 	|<-----------    @cb range   ---->|
+ * 	|			|<-- @buf_len -->|
+ * 	|<--- @decompressed --->|
  *
- * buf_start is the byte offset we're of the start of our workspace buffer.
+ * Note that, @cb can be a subpage of the full decompressed extent, but
+ * @cb->start always has the same as the orig_file_offset value of the full
+ * decompressed extent.
  *
- * total_out is the last byte of the buffer
+ * When reading compressed extent, we have to read the full compressed extent,
+ * while @orig_bio may only want part of the range.
+ * Thus this function will ensure only data covered by @orig_bio will be copied
+ * to.
+ *
+ * Return 0 if we have copied all needed contents for @orig_bio.
+ * Return >0 if we need continue decompress.
  */
-int btrfs_decompress_buf2page(const char *buf, unsigned long buf_start,
-			      unsigned long total_out, u64 disk_start,
-			      struct bio *bio)
+int btrfs_decompress_buf2page(const char *buf, u32 buf_len,
+			      struct compressed_bio *cb, u32 decompressed)
 {
-	unsigned long buf_offset;
-	unsigned long current_buf_start;
-	unsigned long start_byte;
-	unsigned long prev_start_byte;
-	unsigned long working_bytes = total_out - buf_start;
-	unsigned long bytes;
-	char *kaddr;
-	struct bio_vec bvec = bio_iter_iovec(bio, bio->bi_iter);
-
-	/*
-	 * start byte is the first byte of the page we're currently
-	 * copying into relative to the start of the compressed data.
-	 */
-	start_byte = page_offset(bvec.bv_page) - disk_start;
+	struct bio *orig_bio = cb->orig_bio;
+	/* Offset inside the full decompressed extent */
+	u32 cur_offset;
+
+	cur_offset = decompressed;
+	/* The main loop to do the copy */
+	while (cur_offset < decompressed + buf_len) {
+		struct bio_vec bvec;
+		size_t copy_len;
+		u32 copy_start;
+		/* Offset inside the full decompressed extent */
+		u32 bvec_offset;
+
+		bvec = bio_iter_iovec(orig_bio, orig_bio->bi_iter);
+		/*
+		 * cb->start may underflow, but subtracting that value can still
+		 * give us correct offset inside the full decompressed extent.
+		 */
+		bvec_offset = page_offset(bvec.bv_page) + bvec.bv_offset - cb->start;
 
-	/* we haven't yet hit data corresponding to this page */
-	if (total_out <= start_byte)
-		return 1;
+		/* Haven't reached the bvec range, exit */
+		if (decompressed + buf_len <= bvec_offset)
+			return 1;
 
-	/*
-	 * the start of the data we care about is offset into
-	 * the middle of our working buffer
-	 */
-	if (total_out > start_byte && buf_start < start_byte) {
-		buf_offset = start_byte - buf_start;
-		working_bytes -= buf_offset;
-	} else {
-		buf_offset = 0;
-	}
-	current_buf_start = buf_start;
-
-	/* copy bytes from the working buffer into the pages */
-	while (working_bytes > 0) {
-		bytes = min_t(unsigned long, bvec.bv_len,
-				PAGE_SIZE - (buf_offset % PAGE_SIZE));
-		bytes = min(bytes, working_bytes);
-
-		kaddr = kmap_atomic(bvec.bv_page);
-		memcpy(kaddr + bvec.bv_offset, buf + buf_offset, bytes);
-		kunmap_atomic(kaddr);
-		flush_dcache_page(bvec.bv_page);
-
-		buf_offset += bytes;
-		working_bytes -= bytes;
-		current_buf_start += bytes;
-
-		/* check if we need to pick another page */
-		bio_advance(bio, bytes);
-		if (!bio->bi_iter.bi_size)
-			return 0;
-		bvec = bio_iter_iovec(bio, bio->bi_iter);
-		prev_start_byte = start_byte;
-		start_byte = page_offset(bvec.bv_page) - disk_start;
+		copy_start = max(cur_offset, bvec_offset);
+		copy_len = min(bvec_offset + bvec.bv_len,
+			       decompressed + buf_len) - copy_start;
+		ASSERT(copy_len);
 
 		/*
-		 * We need to make sure we're only adjusting
-		 * our offset into compression working buffer when
-		 * we're switching pages.  Otherwise we can incorrectly
-		 * keep copying when we were actually done.
+		 * Extra range check to ensure we didn't go beyond
+		 * @buf + @buf_len.
 		 */
-		if (start_byte != prev_start_byte) {
-			/*
-			 * make sure our new page is covered by this
-			 * working buffer
-			 */
-			if (total_out <= start_byte)
-				return 1;
-
-			/*
-			 * the next page in the biovec might not be adjacent
-			 * to the last page, but it might still be found
-			 * inside this working buffer. bump our offset pointer
-			 */
-			if (total_out > start_byte &&
-			    current_buf_start < start_byte) {
-				buf_offset = start_byte - buf_start;
-				working_bytes = total_out - start_byte;
-				current_buf_start = buf_start + buf_offset;
-			}
-		}
+		ASSERT(copy_start - decompressed < buf_len);
+		memcpy_to_page(bvec.bv_page, bvec.bv_offset,
+			       buf + copy_start - decompressed, copy_len);
+		cur_offset += copy_len;
+
+		bio_advance(orig_bio, copy_len);
+		/* Finished the bio */
+		if (!orig_bio->bi_iter.bi_size)
+			return 0;
 	}
-
 	return 1;
 }
 
@@ -1614,7 +1611,7 @@ static void heuristic_collect_sample(struct inode *inode, u64 start, u64 end,
 	curr_sample_pos = 0;
 	while (index < index_end) {
 		page = find_get_page(inode->i_mapping, index);
-		in_data = kmap(page);
+		in_data = kmap_local_page(page);
 		/* Handle case where the start is not aligned to PAGE_SIZE */
 		i = start % PAGE_SIZE;
 		while (i < PAGE_SIZE - SAMPLING_READ_SIZE) {
@@ -1627,7 +1624,7 @@ static void heuristic_collect_sample(struct inode *inode, u64 start, u64 end,
 			start += SAMPLING_INTERVAL;
 			curr_sample_pos += SAMPLING_READ_SIZE;
 		}
-		kunmap(page);
+		kunmap_local(in_data);
 		put_page(page);
 
 		index++;
@@ -1748,19 +1745,3 @@ unsigned int btrfs_compress_str2level(unsigned int type, const char *str)
 
 	return level;
 }
-
-/*
- * Adjust @level according to the limits of the compression algorithm or
- * fallback to default
- */
-unsigned int btrfs_compress_set_level(int type, unsigned level)
-{
-	const struct btrfs_compress_op *ops = btrfs_compress_op[type];
-
-	if (level == 0)
-		level = ops->default_level;
-	else
-		level = min(level, ops->max_level);
-
-	return level;
-}
diff --git a/fs/btrfs/compression.h b/fs/btrfs/compression.h
index d253f7aa8ed5..1aa02903de69 100644
--- a/fs/btrfs/compression.h
+++ b/fs/btrfs/compression.h
@@ -8,6 +8,8 @@
 
 #include <linux/sizes.h>
 
+struct btrfs_inode;
+
 /*
  * We want to make sure that amount of RAM required to uncompress an extent is
  * reasonable, so we limit the total size in ram of a compressed extent to
@@ -20,14 +22,19 @@
 
 /* Maximum length of compressed data stored on disk */
 #define BTRFS_MAX_COMPRESSED		(SZ_128K)
+static_assert((BTRFS_MAX_COMPRESSED % PAGE_SIZE) == 0);
+
 /* Maximum size of data before compression */
 #define BTRFS_MAX_UNCOMPRESSED		(SZ_128K)
 
 #define	BTRFS_ZLIB_DEFAULT_LEVEL		3
 
 struct compressed_bio {
-	/* number of bios pending for this compressed extent */
-	refcount_t pending_bios;
+	/* Number of outstanding bios */
+	refcount_t pending_ios;
+
+	/* Number of compressed pages in the array */
+	unsigned int nr_pages;
 
 	/* the pages with the compressed data on them */
 	struct page **compressed_pages;
@@ -38,30 +45,26 @@ struct compressed_bio {
 	/* starting offset in the inode for our pages */
 	u64 start;
 
-	/* number of bytes in the inode we're working on */
-	unsigned long len;
+	/* Number of bytes in the inode we're working on */
+	unsigned int len;
 
-	/* number of bytes on disk */
-	unsigned long compressed_len;
+	/* Number of bytes on disk */
+	unsigned int compressed_len;
 
-	/* the compression algorithm for this bio */
-	int compress_type;
+	/* The compression algorithm for this bio */
+	u8 compress_type;
 
-	/* number of compressed pages in the array */
-	unsigned long nr_pages;
+	/* Whether this is a write for writeback. */
+	bool writeback;
 
 	/* IO errors */
-	int errors;
-	int mirror_num;
+	blk_status_t status;
 
-	/* for reads, this is the bio we are copying the data into */
-	struct bio *orig_bio;
-
-	/*
-	 * the start of a variable length array of checksums only
-	 * used by reads
-	 */
-	u8 sums[];
+	union {
+		/* For reads, this is the bio we are copying the data into */
+		struct bio *orig_bio;
+		struct work_struct write_end_work;
+	};
 };
 
 static inline unsigned int btrfs_compress_type(unsigned int type_level)
@@ -84,19 +87,19 @@ int btrfs_compress_pages(unsigned int type_level, struct address_space *mapping,
 			 unsigned long *total_out);
 int btrfs_decompress(int type, unsigned char *data_in, struct page *dest_page,
 		     unsigned long start_byte, size_t srclen, size_t destlen);
-int btrfs_decompress_buf2page(const char *buf, unsigned long buf_start,
-			      unsigned long total_out, u64 disk_start,
-			      struct bio *bio);
+int btrfs_decompress_buf2page(const char *buf, u32 buf_len,
+			      struct compressed_bio *cb, u32 decompressed);
 
-blk_status_t btrfs_submit_compressed_write(struct inode *inode, u64 start,
-				  unsigned long len, u64 disk_start,
-				  unsigned long compressed_len,
+blk_status_t btrfs_submit_compressed_write(struct btrfs_inode *inode, u64 start,
+				  unsigned int len, u64 disk_start,
+				  unsigned int compressed_len,
 				  struct page **compressed_pages,
-				  unsigned long nr_pages,
-				  unsigned int write_flags,
-				  struct cgroup_subsys_state *blkcg_css);
-blk_status_t btrfs_submit_compressed_read(struct inode *inode, struct bio *bio,
-				 int mirror_num, unsigned long bio_flags);
+				  unsigned int nr_pages,
+				  blk_opf_t write_flags,
+				  struct cgroup_subsys_state *blkcg_css,
+				  bool writeback);
+void btrfs_submit_compressed_read(struct inode *inode, struct bio *bio,
+				  int mirror_num);
 
 unsigned int btrfs_compress_str2level(unsigned int type, const char *str);
 
@@ -140,8 +143,41 @@ extern const struct btrfs_compress_op btrfs_zstd_compress;
 const char* btrfs_compress_type2str(enum btrfs_compression_type type);
 bool btrfs_compress_is_valid_type(const char *str, size_t len);
 
-unsigned int btrfs_compress_set_level(int type, unsigned level);
-
 int btrfs_compress_heuristic(struct inode *inode, u64 start, u64 end);
 
+int zlib_compress_pages(struct list_head *ws, struct address_space *mapping,
+		u64 start, struct page **pages, unsigned long *out_pages,
+		unsigned long *total_in, unsigned long *total_out);
+int zlib_decompress_bio(struct list_head *ws, struct compressed_bio *cb);
+int zlib_decompress(struct list_head *ws, unsigned char *data_in,
+		struct page *dest_page, unsigned long start_byte, size_t srclen,
+		size_t destlen);
+struct list_head *zlib_alloc_workspace(unsigned int level);
+void zlib_free_workspace(struct list_head *ws);
+struct list_head *zlib_get_workspace(unsigned int level);
+
+int lzo_compress_pages(struct list_head *ws, struct address_space *mapping,
+		u64 start, struct page **pages, unsigned long *out_pages,
+		unsigned long *total_in, unsigned long *total_out);
+int lzo_decompress_bio(struct list_head *ws, struct compressed_bio *cb);
+int lzo_decompress(struct list_head *ws, unsigned char *data_in,
+		struct page *dest_page, unsigned long start_byte, size_t srclen,
+		size_t destlen);
+struct list_head *lzo_alloc_workspace(unsigned int level);
+void lzo_free_workspace(struct list_head *ws);
+
+int zstd_compress_pages(struct list_head *ws, struct address_space *mapping,
+		u64 start, struct page **pages, unsigned long *out_pages,
+		unsigned long *total_in, unsigned long *total_out);
+int zstd_decompress_bio(struct list_head *ws, struct compressed_bio *cb);
+int zstd_decompress(struct list_head *ws, unsigned char *data_in,
+		struct page *dest_page, unsigned long start_byte, size_t srclen,
+		size_t destlen);
+void zstd_init_workspace_manager(void);
+void zstd_cleanup_workspace_manager(void);
+struct list_head *zstd_alloc_workspace(unsigned int level);
+void zstd_free_workspace(struct list_head *ws);
+struct list_head *zstd_get_workspace(unsigned int level);
+void zstd_put_workspace(struct list_head *ws);
+
 #endif
diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c
index f2ec1a9bae28..a9543f01184c 100644
--- a/fs/btrfs/ctree.c
+++ b/fs/btrfs/ctree.c
@@ -7,6 +7,7 @@
 #include <linux/slab.h>
 #include <linux/rbtree.h>
 #include <linux/mm.h>
+#include <linux/error-injection.h>
 #include "ctree.h"
 #include "disk-io.h"
 #include "transaction.h"
@@ -14,6 +15,8 @@
 #include "locking.h"
 #include "volumes.h"
 #include "qgroup.h"
+#include "tree-mod-log.h"
+#include "tree-checker.h"
 
 static int split_node(struct btrfs_trans_handle *trans, struct btrfs_root
 		      *root, struct btrfs_path *path, int level);
@@ -31,8 +34,8 @@ static void del_ptr(struct btrfs_root *root, struct btrfs_path *path,
 
 static const struct btrfs_csums {
 	u16		size;
-	const char	*name;
-	const char	*driver;
+	const char	name[10];
+	const char	driver[12];
 } btrfs_csums[] = {
 	[BTRFS_CSUM_TYPE_CRC32] = { .size = 4, .name = "crc32c" },
 	[BTRFS_CSUM_TYPE_XXHASH] = { .size = 8, .name = "xxhash64" },
@@ -63,11 +66,12 @@ const char *btrfs_super_csum_name(u16 csum_type)
 const char *btrfs_super_csum_driver(u16 csum_type)
 {
 	/* csum type is validated at mount time */
-	return btrfs_csums[csum_type].driver ?:
+	return btrfs_csums[csum_type].driver[0] ?
+		btrfs_csums[csum_type].driver :
 		btrfs_csums[csum_type].name;
 }
 
-size_t __const btrfs_get_num_csums(void)
+size_t __attribute_const__ btrfs_get_num_csums(void)
 {
 	return ARRAY_SIZE(btrfs_csums);
 }
@@ -110,6 +114,22 @@ noinline void btrfs_release_path(struct btrfs_path *p)
 }
 
 /*
+ * We want the transaction abort to print stack trace only for errors where the
+ * cause could be a bug, eg. due to ENOSPC, and not for common errors that are
+ * caused by external factors.
+ */
+bool __cold abort_should_print_stack(int errno)
+{
+	switch (errno) {
+	case -EIO:
+	case -EROFS:
+	case -ENOMEM:
+		return false;
+	}
+	return true;
+}
+
+/*
  * safely gets a reference on the root node of a tree.  A lock
  * is not taken, so a concurrent writer may put a different node
  * at the root of the tree.  See btrfs_lock_root_node for the
@@ -143,47 +163,10 @@ struct extent_buffer *btrfs_root_node(struct btrfs_root *root)
 	return eb;
 }
 
-/* loop around taking references on and locking the root node of the
- * tree until you end up with a lock on the root.  A locked buffer
- * is returned, with a reference held.
- */
-struct extent_buffer *btrfs_lock_root_node(struct btrfs_root *root)
-{
-	struct extent_buffer *eb;
-
-	while (1) {
-		eb = btrfs_root_node(root);
-		btrfs_tree_lock(eb);
-		if (eb == root->node)
-			break;
-		btrfs_tree_unlock(eb);
-		free_extent_buffer(eb);
-	}
-	return eb;
-}
-
-/* loop around taking references on and locking the root node of the
- * tree until you end up with a lock on the root.  A locked buffer
- * is returned, with a reference held.
- */
-struct extent_buffer *btrfs_read_lock_root_node(struct btrfs_root *root)
-{
-	struct extent_buffer *eb;
-
-	while (1) {
-		eb = btrfs_root_node(root);
-		btrfs_tree_read_lock(eb);
-		if (eb == root->node)
-			break;
-		btrfs_tree_read_unlock(eb);
-		free_extent_buffer(eb);
-	}
-	return eb;
-}
-
-/* cowonly root (everything not a reference counted cow subvolume), just get
- * put onto a simple dirty list.  transaction.c walks this to make sure they
- * get properly updated on disk.
+/*
+ * Cowonly root (not-shareable trees, everything not subvolume or reloc roots),
+ * just get put onto a simple dirty list.  Transaction walks this list to make
+ * sure they get properly updated on disk.
  */
 static void add_root_to_dirty_list(struct btrfs_root *root)
 {
@@ -222,9 +205,9 @@ int btrfs_copy_root(struct btrfs_trans_handle *trans,
 	int level;
 	struct btrfs_disk_key disk_key;
 
-	WARN_ON(test_bit(BTRFS_ROOT_REF_COWS, &root->state) &&
+	WARN_ON(test_bit(BTRFS_ROOT_SHAREABLE, &root->state) &&
 		trans->transid != fs_info->running_transaction->transid);
-	WARN_ON(test_bit(BTRFS_ROOT_REF_COWS, &root->state) &&
+	WARN_ON(test_bit(BTRFS_ROOT_SHAREABLE, &root->state) &&
 		trans->transid != root->last_trans);
 
 	level = btrfs_header_level(buf);
@@ -234,7 +217,8 @@ int btrfs_copy_root(struct btrfs_trans_handle *trans,
 		btrfs_node_key(buf, &disk_key, 0);
 
 	cow = btrfs_alloc_tree_block(trans, root, 0, new_root_objectid,
-			&disk_key, level, buf->start, 0);
+				     &disk_key, level, buf->start, 0,
+				     BTRFS_NESTING_NEW_ROOT);
 	if (IS_ERR(cow))
 		return PTR_ERR(cow);
 
@@ -256,605 +240,18 @@ int btrfs_copy_root(struct btrfs_trans_handle *trans,
 		ret = btrfs_inc_ref(trans, root, cow, 1);
 	else
 		ret = btrfs_inc_ref(trans, root, cow, 0);
-
-	if (ret)
+	if (ret) {
+		btrfs_tree_unlock(cow);
+		free_extent_buffer(cow);
+		btrfs_abort_transaction(trans, ret);
 		return ret;
+	}
 
 	btrfs_mark_buffer_dirty(cow);
 	*cow_ret = cow;
 	return 0;
 }
 
-enum mod_log_op {
-	MOD_LOG_KEY_REPLACE,
-	MOD_LOG_KEY_ADD,
-	MOD_LOG_KEY_REMOVE,
-	MOD_LOG_KEY_REMOVE_WHILE_FREEING,
-	MOD_LOG_KEY_REMOVE_WHILE_MOVING,
-	MOD_LOG_MOVE_KEYS,
-	MOD_LOG_ROOT_REPLACE,
-};
-
-struct tree_mod_root {
-	u64 logical;
-	u8 level;
-};
-
-struct tree_mod_elem {
-	struct rb_node node;
-	u64 logical;
-	u64 seq;
-	enum mod_log_op op;
-
-	/* this is used for MOD_LOG_KEY_* and MOD_LOG_MOVE_KEYS operations */
-	int slot;
-
-	/* this is used for MOD_LOG_KEY* and MOD_LOG_ROOT_REPLACE */
-	u64 generation;
-
-	/* those are used for op == MOD_LOG_KEY_{REPLACE,REMOVE} */
-	struct btrfs_disk_key key;
-	u64 blockptr;
-
-	/* this is used for op == MOD_LOG_MOVE_KEYS */
-	struct {
-		int dst_slot;
-		int nr_items;
-	} move;
-
-	/* this is used for op == MOD_LOG_ROOT_REPLACE */
-	struct tree_mod_root old_root;
-};
-
-/*
- * Pull a new tree mod seq number for our operation.
- */
-static inline u64 btrfs_inc_tree_mod_seq(struct btrfs_fs_info *fs_info)
-{
-	return atomic64_inc_return(&fs_info->tree_mod_seq);
-}
-
-/*
- * This adds a new blocker to the tree mod log's blocker list if the @elem
- * passed does not already have a sequence number set. So when a caller expects
- * to record tree modifications, it should ensure to set elem->seq to zero
- * before calling btrfs_get_tree_mod_seq.
- * Returns a fresh, unused tree log modification sequence number, even if no new
- * blocker was added.
- */
-u64 btrfs_get_tree_mod_seq(struct btrfs_fs_info *fs_info,
-			   struct seq_list *elem)
-{
-	write_lock(&fs_info->tree_mod_log_lock);
-	if (!elem->seq) {
-		elem->seq = btrfs_inc_tree_mod_seq(fs_info);
-		list_add_tail(&elem->list, &fs_info->tree_mod_seq_list);
-	}
-	write_unlock(&fs_info->tree_mod_log_lock);
-
-	return elem->seq;
-}
-
-void btrfs_put_tree_mod_seq(struct btrfs_fs_info *fs_info,
-			    struct seq_list *elem)
-{
-	struct rb_root *tm_root;
-	struct rb_node *node;
-	struct rb_node *next;
-	struct seq_list *cur_elem;
-	struct tree_mod_elem *tm;
-	u64 min_seq = (u64)-1;
-	u64 seq_putting = elem->seq;
-
-	if (!seq_putting)
-		return;
-
-	write_lock(&fs_info->tree_mod_log_lock);
-	list_del(&elem->list);
-	elem->seq = 0;
-
-	list_for_each_entry(cur_elem, &fs_info->tree_mod_seq_list, list) {
-		if (cur_elem->seq < min_seq) {
-			if (seq_putting > cur_elem->seq) {
-				/*
-				 * blocker with lower sequence number exists, we
-				 * cannot remove anything from the log
-				 */
-				write_unlock(&fs_info->tree_mod_log_lock);
-				return;
-			}
-			min_seq = cur_elem->seq;
-		}
-	}
-
-	/*
-	 * anything that's lower than the lowest existing (read: blocked)
-	 * sequence number can be removed from the tree.
-	 */
-	tm_root = &fs_info->tree_mod_log;
-	for (node = rb_first(tm_root); node; node = next) {
-		next = rb_next(node);
-		tm = rb_entry(node, struct tree_mod_elem, node);
-		if (tm->seq >= min_seq)
-			continue;
-		rb_erase(node, tm_root);
-		kfree(tm);
-	}
-	write_unlock(&fs_info->tree_mod_log_lock);
-}
-
-/*
- * key order of the log:
- *       node/leaf start address -> sequence
- *
- * The 'start address' is the logical address of the *new* root node
- * for root replace operations, or the logical address of the affected
- * block for all other operations.
- */
-static noinline int
-__tree_mod_log_insert(struct btrfs_fs_info *fs_info, struct tree_mod_elem *tm)
-{
-	struct rb_root *tm_root;
-	struct rb_node **new;
-	struct rb_node *parent = NULL;
-	struct tree_mod_elem *cur;
-
-	lockdep_assert_held_write(&fs_info->tree_mod_log_lock);
-
-	tm->seq = btrfs_inc_tree_mod_seq(fs_info);
-
-	tm_root = &fs_info->tree_mod_log;
-	new = &tm_root->rb_node;
-	while (*new) {
-		cur = rb_entry(*new, struct tree_mod_elem, node);
-		parent = *new;
-		if (cur->logical < tm->logical)
-			new = &((*new)->rb_left);
-		else if (cur->logical > tm->logical)
-			new = &((*new)->rb_right);
-		else if (cur->seq < tm->seq)
-			new = &((*new)->rb_left);
-		else if (cur->seq > tm->seq)
-			new = &((*new)->rb_right);
-		else
-			return -EEXIST;
-	}
-
-	rb_link_node(&tm->node, parent, new);
-	rb_insert_color(&tm->node, tm_root);
-	return 0;
-}
-
-/*
- * Determines if logging can be omitted. Returns 1 if it can. Otherwise, it
- * returns zero with the tree_mod_log_lock acquired. The caller must hold
- * this until all tree mod log insertions are recorded in the rb tree and then
- * write unlock fs_info::tree_mod_log_lock.
- */
-static inline int tree_mod_dont_log(struct btrfs_fs_info *fs_info,
-				    struct extent_buffer *eb) {
-	smp_mb();
-	if (list_empty(&(fs_info)->tree_mod_seq_list))
-		return 1;
-	if (eb && btrfs_header_level(eb) == 0)
-		return 1;
-
-	write_lock(&fs_info->tree_mod_log_lock);
-	if (list_empty(&(fs_info)->tree_mod_seq_list)) {
-		write_unlock(&fs_info->tree_mod_log_lock);
-		return 1;
-	}
-
-	return 0;
-}
-
-/* Similar to tree_mod_dont_log, but doesn't acquire any locks. */
-static inline int tree_mod_need_log(const struct btrfs_fs_info *fs_info,
-				    struct extent_buffer *eb)
-{
-	smp_mb();
-	if (list_empty(&(fs_info)->tree_mod_seq_list))
-		return 0;
-	if (eb && btrfs_header_level(eb) == 0)
-		return 0;
-
-	return 1;
-}
-
-static struct tree_mod_elem *
-alloc_tree_mod_elem(struct extent_buffer *eb, int slot,
-		    enum mod_log_op op, gfp_t flags)
-{
-	struct tree_mod_elem *tm;
-
-	tm = kzalloc(sizeof(*tm), flags);
-	if (!tm)
-		return NULL;
-
-	tm->logical = eb->start;
-	if (op != MOD_LOG_KEY_ADD) {
-		btrfs_node_key(eb, &tm->key, slot);
-		tm->blockptr = btrfs_node_blockptr(eb, slot);
-	}
-	tm->op = op;
-	tm->slot = slot;
-	tm->generation = btrfs_node_ptr_generation(eb, slot);
-	RB_CLEAR_NODE(&tm->node);
-
-	return tm;
-}
-
-static noinline int tree_mod_log_insert_key(struct extent_buffer *eb, int slot,
-		enum mod_log_op op, gfp_t flags)
-{
-	struct tree_mod_elem *tm;
-	int ret;
-
-	if (!tree_mod_need_log(eb->fs_info, eb))
-		return 0;
-
-	tm = alloc_tree_mod_elem(eb, slot, op, flags);
-	if (!tm)
-		return -ENOMEM;
-
-	if (tree_mod_dont_log(eb->fs_info, eb)) {
-		kfree(tm);
-		return 0;
-	}
-
-	ret = __tree_mod_log_insert(eb->fs_info, tm);
-	write_unlock(&eb->fs_info->tree_mod_log_lock);
-	if (ret)
-		kfree(tm);
-
-	return ret;
-}
-
-static noinline int tree_mod_log_insert_move(struct extent_buffer *eb,
-		int dst_slot, int src_slot, int nr_items)
-{
-	struct tree_mod_elem *tm = NULL;
-	struct tree_mod_elem **tm_list = NULL;
-	int ret = 0;
-	int i;
-	int locked = 0;
-
-	if (!tree_mod_need_log(eb->fs_info, eb))
-		return 0;
-
-	tm_list = kcalloc(nr_items, sizeof(struct tree_mod_elem *), GFP_NOFS);
-	if (!tm_list)
-		return -ENOMEM;
-
-	tm = kzalloc(sizeof(*tm), GFP_NOFS);
-	if (!tm) {
-		ret = -ENOMEM;
-		goto free_tms;
-	}
-
-	tm->logical = eb->start;
-	tm->slot = src_slot;
-	tm->move.dst_slot = dst_slot;
-	tm->move.nr_items = nr_items;
-	tm->op = MOD_LOG_MOVE_KEYS;
-
-	for (i = 0; i + dst_slot < src_slot && i < nr_items; i++) {
-		tm_list[i] = alloc_tree_mod_elem(eb, i + dst_slot,
-		    MOD_LOG_KEY_REMOVE_WHILE_MOVING, GFP_NOFS);
-		if (!tm_list[i]) {
-			ret = -ENOMEM;
-			goto free_tms;
-		}
-	}
-
-	if (tree_mod_dont_log(eb->fs_info, eb))
-		goto free_tms;
-	locked = 1;
-
-	/*
-	 * When we override something during the move, we log these removals.
-	 * This can only happen when we move towards the beginning of the
-	 * buffer, i.e. dst_slot < src_slot.
-	 */
-	for (i = 0; i + dst_slot < src_slot && i < nr_items; i++) {
-		ret = __tree_mod_log_insert(eb->fs_info, tm_list[i]);
-		if (ret)
-			goto free_tms;
-	}
-
-	ret = __tree_mod_log_insert(eb->fs_info, tm);
-	if (ret)
-		goto free_tms;
-	write_unlock(&eb->fs_info->tree_mod_log_lock);
-	kfree(tm_list);
-
-	return 0;
-free_tms:
-	for (i = 0; i < nr_items; i++) {
-		if (tm_list[i] && !RB_EMPTY_NODE(&tm_list[i]->node))
-			rb_erase(&tm_list[i]->node, &eb->fs_info->tree_mod_log);
-		kfree(tm_list[i]);
-	}
-	if (locked)
-		write_unlock(&eb->fs_info->tree_mod_log_lock);
-	kfree(tm_list);
-	kfree(tm);
-
-	return ret;
-}
-
-static inline int
-__tree_mod_log_free_eb(struct btrfs_fs_info *fs_info,
-		       struct tree_mod_elem **tm_list,
-		       int nritems)
-{
-	int i, j;
-	int ret;
-
-	for (i = nritems - 1; i >= 0; i--) {
-		ret = __tree_mod_log_insert(fs_info, tm_list[i]);
-		if (ret) {
-			for (j = nritems - 1; j > i; j--)
-				rb_erase(&tm_list[j]->node,
-					 &fs_info->tree_mod_log);
-			return ret;
-		}
-	}
-
-	return 0;
-}
-
-static noinline int tree_mod_log_insert_root(struct extent_buffer *old_root,
-			 struct extent_buffer *new_root, int log_removal)
-{
-	struct btrfs_fs_info *fs_info = old_root->fs_info;
-	struct tree_mod_elem *tm = NULL;
-	struct tree_mod_elem **tm_list = NULL;
-	int nritems = 0;
-	int ret = 0;
-	int i;
-
-	if (!tree_mod_need_log(fs_info, NULL))
-		return 0;
-
-	if (log_removal && btrfs_header_level(old_root) > 0) {
-		nritems = btrfs_header_nritems(old_root);
-		tm_list = kcalloc(nritems, sizeof(struct tree_mod_elem *),
-				  GFP_NOFS);
-		if (!tm_list) {
-			ret = -ENOMEM;
-			goto free_tms;
-		}
-		for (i = 0; i < nritems; i++) {
-			tm_list[i] = alloc_tree_mod_elem(old_root, i,
-			    MOD_LOG_KEY_REMOVE_WHILE_FREEING, GFP_NOFS);
-			if (!tm_list[i]) {
-				ret = -ENOMEM;
-				goto free_tms;
-			}
-		}
-	}
-
-	tm = kzalloc(sizeof(*tm), GFP_NOFS);
-	if (!tm) {
-		ret = -ENOMEM;
-		goto free_tms;
-	}
-
-	tm->logical = new_root->start;
-	tm->old_root.logical = old_root->start;
-	tm->old_root.level = btrfs_header_level(old_root);
-	tm->generation = btrfs_header_generation(old_root);
-	tm->op = MOD_LOG_ROOT_REPLACE;
-
-	if (tree_mod_dont_log(fs_info, NULL))
-		goto free_tms;
-
-	if (tm_list)
-		ret = __tree_mod_log_free_eb(fs_info, tm_list, nritems);
-	if (!ret)
-		ret = __tree_mod_log_insert(fs_info, tm);
-
-	write_unlock(&fs_info->tree_mod_log_lock);
-	if (ret)
-		goto free_tms;
-	kfree(tm_list);
-
-	return ret;
-
-free_tms:
-	if (tm_list) {
-		for (i = 0; i < nritems; i++)
-			kfree(tm_list[i]);
-		kfree(tm_list);
-	}
-	kfree(tm);
-
-	return ret;
-}
-
-static struct tree_mod_elem *
-__tree_mod_log_search(struct btrfs_fs_info *fs_info, u64 start, u64 min_seq,
-		      int smallest)
-{
-	struct rb_root *tm_root;
-	struct rb_node *node;
-	struct tree_mod_elem *cur = NULL;
-	struct tree_mod_elem *found = NULL;
-
-	read_lock(&fs_info->tree_mod_log_lock);
-	tm_root = &fs_info->tree_mod_log;
-	node = tm_root->rb_node;
-	while (node) {
-		cur = rb_entry(node, struct tree_mod_elem, node);
-		if (cur->logical < start) {
-			node = node->rb_left;
-		} else if (cur->logical > start) {
-			node = node->rb_right;
-		} else if (cur->seq < min_seq) {
-			node = node->rb_left;
-		} else if (!smallest) {
-			/* we want the node with the highest seq */
-			if (found)
-				BUG_ON(found->seq > cur->seq);
-			found = cur;
-			node = node->rb_left;
-		} else if (cur->seq > min_seq) {
-			/* we want the node with the smallest seq */
-			if (found)
-				BUG_ON(found->seq < cur->seq);
-			found = cur;
-			node = node->rb_right;
-		} else {
-			found = cur;
-			break;
-		}
-	}
-	read_unlock(&fs_info->tree_mod_log_lock);
-
-	return found;
-}
-
-/*
- * this returns the element from the log with the smallest time sequence
- * value that's in the log (the oldest log item). any element with a time
- * sequence lower than min_seq will be ignored.
- */
-static struct tree_mod_elem *
-tree_mod_log_search_oldest(struct btrfs_fs_info *fs_info, u64 start,
-			   u64 min_seq)
-{
-	return __tree_mod_log_search(fs_info, start, min_seq, 1);
-}
-
-/*
- * this returns the element from the log with the largest time sequence
- * value that's in the log (the most recent log item). any element with
- * a time sequence lower than min_seq will be ignored.
- */
-static struct tree_mod_elem *
-tree_mod_log_search(struct btrfs_fs_info *fs_info, u64 start, u64 min_seq)
-{
-	return __tree_mod_log_search(fs_info, start, min_seq, 0);
-}
-
-static noinline int tree_mod_log_eb_copy(struct extent_buffer *dst,
-		     struct extent_buffer *src, unsigned long dst_offset,
-		     unsigned long src_offset, int nr_items)
-{
-	struct btrfs_fs_info *fs_info = dst->fs_info;
-	int ret = 0;
-	struct tree_mod_elem **tm_list = NULL;
-	struct tree_mod_elem **tm_list_add, **tm_list_rem;
-	int i;
-	int locked = 0;
-
-	if (!tree_mod_need_log(fs_info, NULL))
-		return 0;
-
-	if (btrfs_header_level(dst) == 0 && btrfs_header_level(src) == 0)
-		return 0;
-
-	tm_list = kcalloc(nr_items * 2, sizeof(struct tree_mod_elem *),
-			  GFP_NOFS);
-	if (!tm_list)
-		return -ENOMEM;
-
-	tm_list_add = tm_list;
-	tm_list_rem = tm_list + nr_items;
-	for (i = 0; i < nr_items; i++) {
-		tm_list_rem[i] = alloc_tree_mod_elem(src, i + src_offset,
-		    MOD_LOG_KEY_REMOVE, GFP_NOFS);
-		if (!tm_list_rem[i]) {
-			ret = -ENOMEM;
-			goto free_tms;
-		}
-
-		tm_list_add[i] = alloc_tree_mod_elem(dst, i + dst_offset,
-		    MOD_LOG_KEY_ADD, GFP_NOFS);
-		if (!tm_list_add[i]) {
-			ret = -ENOMEM;
-			goto free_tms;
-		}
-	}
-
-	if (tree_mod_dont_log(fs_info, NULL))
-		goto free_tms;
-	locked = 1;
-
-	for (i = 0; i < nr_items; i++) {
-		ret = __tree_mod_log_insert(fs_info, tm_list_rem[i]);
-		if (ret)
-			goto free_tms;
-		ret = __tree_mod_log_insert(fs_info, tm_list_add[i]);
-		if (ret)
-			goto free_tms;
-	}
-
-	write_unlock(&fs_info->tree_mod_log_lock);
-	kfree(tm_list);
-
-	return 0;
-
-free_tms:
-	for (i = 0; i < nr_items * 2; i++) {
-		if (tm_list[i] && !RB_EMPTY_NODE(&tm_list[i]->node))
-			rb_erase(&tm_list[i]->node, &fs_info->tree_mod_log);
-		kfree(tm_list[i]);
-	}
-	if (locked)
-		write_unlock(&fs_info->tree_mod_log_lock);
-	kfree(tm_list);
-
-	return ret;
-}
-
-static noinline int tree_mod_log_free_eb(struct extent_buffer *eb)
-{
-	struct tree_mod_elem **tm_list = NULL;
-	int nritems = 0;
-	int i;
-	int ret = 0;
-
-	if (btrfs_header_level(eb) == 0)
-		return 0;
-
-	if (!tree_mod_need_log(eb->fs_info, NULL))
-		return 0;
-
-	nritems = btrfs_header_nritems(eb);
-	tm_list = kcalloc(nritems, sizeof(struct tree_mod_elem *), GFP_NOFS);
-	if (!tm_list)
-		return -ENOMEM;
-
-	for (i = 0; i < nritems; i++) {
-		tm_list[i] = alloc_tree_mod_elem(eb, i,
-		    MOD_LOG_KEY_REMOVE_WHILE_FREEING, GFP_NOFS);
-		if (!tm_list[i]) {
-			ret = -ENOMEM;
-			goto free_tms;
-		}
-	}
-
-	if (tree_mod_dont_log(eb->fs_info, eb))
-		goto free_tms;
-
-	ret = __tree_mod_log_free_eb(eb->fs_info, tm_list, nritems);
-	write_unlock(&eb->fs_info->tree_mod_log_lock);
-	if (ret)
-		goto free_tms;
-	kfree(tm_list);
-
-	return 0;
-
-free_tms:
-	for (i = 0; i < nritems; i++)
-		kfree(tm_list[i]);
-	kfree(tm_list);
-
-	return ret;
-}
-
 /*
  * check if the tree block can be shared by multiple trees
  */
@@ -862,12 +259,11 @@ int btrfs_block_can_be_shared(struct btrfs_root *root,
 			      struct extent_buffer *buf)
 {
 	/*
-	 * Tree blocks not in reference counted trees and tree roots
-	 * are never shared. If a block was allocated after the last
-	 * snapshot and the block was not allocated by tree relocation,
-	 * we know the block is not shared.
+	 * Tree blocks not in shareable trees and tree roots are never shared.
+	 * If a block was allocated after the last snapshot and the block was
+	 * not allocated by tree relocation, we know the block is not shared.
 	 */
-	if (test_bit(BTRFS_ROOT_REF_COWS, &root->state) &&
+	if (test_bit(BTRFS_ROOT_SHAREABLE, &root->state) &&
 	    buf != root->node && buf != root->commit_root &&
 	    (btrfs_header_generation(buf) <=
 	     btrfs_root_last_snapshot(&root->root_item) ||
@@ -962,10 +358,8 @@ static noinline int update_ref_for_cow(struct btrfs_trans_handle *trans,
 		if (new_flags != 0) {
 			int level = btrfs_header_level(buf);
 
-			ret = btrfs_set_disk_extent_flags(trans,
-							  buf->start,
-							  buf->len,
-							  new_flags, level, 0);
+			ret = btrfs_set_disk_extent_flags(trans, buf,
+							  new_flags, level);
 			if (ret)
 				return ret;
 		}
@@ -988,48 +382,6 @@ static noinline int update_ref_for_cow(struct btrfs_trans_handle *trans,
 	return 0;
 }
 
-static struct extent_buffer *alloc_tree_block_no_bg_flush(
-					  struct btrfs_trans_handle *trans,
-					  struct btrfs_root *root,
-					  u64 parent_start,
-					  const struct btrfs_disk_key *disk_key,
-					  int level,
-					  u64 hint,
-					  u64 empty_size)
-{
-	struct btrfs_fs_info *fs_info = root->fs_info;
-	struct extent_buffer *ret;
-
-	/*
-	 * If we are COWing a node/leaf from the extent, chunk, device or free
-	 * space trees, make sure that we do not finish block group creation of
-	 * pending block groups. We do this to avoid a deadlock.
-	 * COWing can result in allocation of a new chunk, and flushing pending
-	 * block groups (btrfs_create_pending_block_groups()) can be triggered
-	 * when finishing allocation of a new chunk. Creation of a pending block
-	 * group modifies the extent, chunk, device and free space trees,
-	 * therefore we could deadlock with ourselves since we are holding a
-	 * lock on an extent buffer that btrfs_create_pending_block_groups() may
-	 * try to COW later.
-	 * For similar reasons, we also need to delay flushing pending block
-	 * groups when splitting a leaf or node, from one of those trees, since
-	 * we are holding a write lock on it and its parent or when inserting a
-	 * new root node for one of those trees.
-	 */
-	if (root == fs_info->extent_root ||
-	    root == fs_info->chunk_root ||
-	    root == fs_info->dev_root ||
-	    root == fs_info->free_space_root)
-		trans->can_flush_pending_bgs = false;
-
-	ret = btrfs_alloc_tree_block(trans, root, parent_start,
-				     root->root_key.objectid, disk_key, level,
-				     hint, empty_size);
-	trans->can_flush_pending_bgs = true;
-
-	return ret;
-}
-
 /*
  * does the dirty work in cow of a single block.  The parent block (if
  * supplied) is updated to point to the new cow copy.  The new buffer is marked
@@ -1047,7 +399,8 @@ static noinline int __btrfs_cow_block(struct btrfs_trans_handle *trans,
 			     struct extent_buffer *buf,
 			     struct extent_buffer *parent, int parent_slot,
 			     struct extent_buffer **cow_ret,
-			     u64 search_start, u64 empty_size)
+			     u64 search_start, u64 empty_size,
+			     enum btrfs_lock_nesting nest)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct btrfs_disk_key disk_key;
@@ -1060,11 +413,11 @@ static noinline int __btrfs_cow_block(struct btrfs_trans_handle *trans,
 	if (*cow_ret == buf)
 		unlock_orig = 1;
 
-	btrfs_assert_tree_locked(buf);
+	btrfs_assert_tree_write_locked(buf);
 
-	WARN_ON(test_bit(BTRFS_ROOT_REF_COWS, &root->state) &&
+	WARN_ON(test_bit(BTRFS_ROOT_SHAREABLE, &root->state) &&
 		trans->transid != fs_info->running_transaction->transid);
-	WARN_ON(test_bit(BTRFS_ROOT_REF_COWS, &root->state) &&
+	WARN_ON(test_bit(BTRFS_ROOT_SHAREABLE, &root->state) &&
 		trans->transid != root->last_trans);
 
 	level = btrfs_header_level(buf);
@@ -1077,8 +430,9 @@ static noinline int __btrfs_cow_block(struct btrfs_trans_handle *trans,
 	if ((root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID) && parent)
 		parent_start = parent->start;
 
-	cow = alloc_tree_block_no_bg_flush(trans, root, parent_start, &disk_key,
-					   level, search_start, empty_size);
+	cow = btrfs_alloc_tree_block(trans, root, parent_start,
+				     root->root_key.objectid, &disk_key, level,
+				     search_start, empty_size, nest);
 	if (IS_ERR(cow))
 		return PTR_ERR(cow);
 
@@ -1099,13 +453,17 @@ static noinline int __btrfs_cow_block(struct btrfs_trans_handle *trans,
 
 	ret = update_ref_for_cow(trans, root, buf, cow, &last_ref);
 	if (ret) {
+		btrfs_tree_unlock(cow);
+		free_extent_buffer(cow);
 		btrfs_abort_transaction(trans, ret);
 		return ret;
 	}
 
-	if (test_bit(BTRFS_ROOT_REF_COWS, &root->state)) {
+	if (test_bit(BTRFS_ROOT_SHAREABLE, &root->state)) {
 		ret = btrfs_reloc_cow_block(trans, root, buf, cow);
 		if (ret) {
+			btrfs_tree_unlock(cow);
+			free_extent_buffer(cow);
 			btrfs_abort_transaction(trans, ret);
 			return ret;
 		}
@@ -1118,32 +476,34 @@ static noinline int __btrfs_cow_block(struct btrfs_trans_handle *trans,
 			parent_start = buf->start;
 
 		atomic_inc(&cow->refs);
-		ret = tree_mod_log_insert_root(root->node, cow, 1);
+		ret = btrfs_tree_mod_log_insert_root(root->node, cow, true);
 		BUG_ON(ret < 0);
 		rcu_assign_pointer(root->node, cow);
 
-		btrfs_free_tree_block(trans, root, buf, parent_start,
-				      last_ref);
+		btrfs_free_tree_block(trans, btrfs_root_id(root), buf,
+				      parent_start, last_ref);
 		free_extent_buffer(buf);
 		add_root_to_dirty_list(root);
 	} else {
 		WARN_ON(trans->transid != btrfs_header_generation(parent));
-		tree_mod_log_insert_key(parent, parent_slot,
-					MOD_LOG_KEY_REPLACE, GFP_NOFS);
+		btrfs_tree_mod_log_insert_key(parent, parent_slot,
+					      BTRFS_MOD_LOG_KEY_REPLACE, GFP_NOFS);
 		btrfs_set_node_blockptr(parent, parent_slot,
 					cow->start);
 		btrfs_set_node_ptr_generation(parent, parent_slot,
 					      trans->transid);
 		btrfs_mark_buffer_dirty(parent);
 		if (last_ref) {
-			ret = tree_mod_log_free_eb(buf);
+			ret = btrfs_tree_mod_log_free_eb(buf);
 			if (ret) {
+				btrfs_tree_unlock(cow);
+				free_extent_buffer(cow);
 				btrfs_abort_transaction(trans, ret);
 				return ret;
 			}
 		}
-		btrfs_free_tree_block(trans, root, buf, parent_start,
-				      last_ref);
+		btrfs_free_tree_block(trans, btrfs_root_id(root), buf,
+				      parent_start, last_ref);
 	}
 	if (unlock_orig)
 		btrfs_tree_unlock(buf);
@@ -1153,295 +513,6 @@ static noinline int __btrfs_cow_block(struct btrfs_trans_handle *trans,
 	return 0;
 }
 
-/*
- * returns the logical address of the oldest predecessor of the given root.
- * entries older than time_seq are ignored.
- */
-static struct tree_mod_elem *__tree_mod_log_oldest_root(
-		struct extent_buffer *eb_root, u64 time_seq)
-{
-	struct tree_mod_elem *tm;
-	struct tree_mod_elem *found = NULL;
-	u64 root_logical = eb_root->start;
-	int looped = 0;
-
-	if (!time_seq)
-		return NULL;
-
-	/*
-	 * the very last operation that's logged for a root is the
-	 * replacement operation (if it is replaced at all). this has
-	 * the logical address of the *new* root, making it the very
-	 * first operation that's logged for this root.
-	 */
-	while (1) {
-		tm = tree_mod_log_search_oldest(eb_root->fs_info, root_logical,
-						time_seq);
-		if (!looped && !tm)
-			return NULL;
-		/*
-		 * if there are no tree operation for the oldest root, we simply
-		 * return it. this should only happen if that (old) root is at
-		 * level 0.
-		 */
-		if (!tm)
-			break;
-
-		/*
-		 * if there's an operation that's not a root replacement, we
-		 * found the oldest version of our root. normally, we'll find a
-		 * MOD_LOG_KEY_REMOVE_WHILE_FREEING operation here.
-		 */
-		if (tm->op != MOD_LOG_ROOT_REPLACE)
-			break;
-
-		found = tm;
-		root_logical = tm->old_root.logical;
-		looped = 1;
-	}
-
-	/* if there's no old root to return, return what we found instead */
-	if (!found)
-		found = tm;
-
-	return found;
-}
-
-/*
- * tm is a pointer to the first operation to rewind within eb. then, all
- * previous operations will be rewound (until we reach something older than
- * time_seq).
- */
-static void
-__tree_mod_log_rewind(struct btrfs_fs_info *fs_info, struct extent_buffer *eb,
-		      u64 time_seq, struct tree_mod_elem *first_tm)
-{
-	u32 n;
-	struct rb_node *next;
-	struct tree_mod_elem *tm = first_tm;
-	unsigned long o_dst;
-	unsigned long o_src;
-	unsigned long p_size = sizeof(struct btrfs_key_ptr);
-
-	n = btrfs_header_nritems(eb);
-	read_lock(&fs_info->tree_mod_log_lock);
-	while (tm && tm->seq >= time_seq) {
-		/*
-		 * all the operations are recorded with the operator used for
-		 * the modification. as we're going backwards, we do the
-		 * opposite of each operation here.
-		 */
-		switch (tm->op) {
-		case MOD_LOG_KEY_REMOVE_WHILE_FREEING:
-			BUG_ON(tm->slot < n);
-			/* Fallthrough */
-		case MOD_LOG_KEY_REMOVE_WHILE_MOVING:
-		case MOD_LOG_KEY_REMOVE:
-			btrfs_set_node_key(eb, &tm->key, tm->slot);
-			btrfs_set_node_blockptr(eb, tm->slot, tm->blockptr);
-			btrfs_set_node_ptr_generation(eb, tm->slot,
-						      tm->generation);
-			n++;
-			break;
-		case MOD_LOG_KEY_REPLACE:
-			BUG_ON(tm->slot >= n);
-			btrfs_set_node_key(eb, &tm->key, tm->slot);
-			btrfs_set_node_blockptr(eb, tm->slot, tm->blockptr);
-			btrfs_set_node_ptr_generation(eb, tm->slot,
-						      tm->generation);
-			break;
-		case MOD_LOG_KEY_ADD:
-			/* if a move operation is needed it's in the log */
-			n--;
-			break;
-		case MOD_LOG_MOVE_KEYS:
-			o_dst = btrfs_node_key_ptr_offset(tm->slot);
-			o_src = btrfs_node_key_ptr_offset(tm->move.dst_slot);
-			memmove_extent_buffer(eb, o_dst, o_src,
-					      tm->move.nr_items * p_size);
-			break;
-		case MOD_LOG_ROOT_REPLACE:
-			/*
-			 * this operation is special. for roots, this must be
-			 * handled explicitly before rewinding.
-			 * for non-roots, this operation may exist if the node
-			 * was a root: root A -> child B; then A gets empty and
-			 * B is promoted to the new root. in the mod log, we'll
-			 * have a root-replace operation for B, a tree block
-			 * that is no root. we simply ignore that operation.
-			 */
-			break;
-		}
-		next = rb_next(&tm->node);
-		if (!next)
-			break;
-		tm = rb_entry(next, struct tree_mod_elem, node);
-		if (tm->logical != first_tm->logical)
-			break;
-	}
-	read_unlock(&fs_info->tree_mod_log_lock);
-	btrfs_set_header_nritems(eb, n);
-}
-
-/*
- * Called with eb read locked. If the buffer cannot be rewound, the same buffer
- * is returned. If rewind operations happen, a fresh buffer is returned. The
- * returned buffer is always read-locked. If the returned buffer is not the
- * input buffer, the lock on the input buffer is released and the input buffer
- * is freed (its refcount is decremented).
- */
-static struct extent_buffer *
-tree_mod_log_rewind(struct btrfs_fs_info *fs_info, struct btrfs_path *path,
-		    struct extent_buffer *eb, u64 time_seq)
-{
-	struct extent_buffer *eb_rewin;
-	struct tree_mod_elem *tm;
-
-	if (!time_seq)
-		return eb;
-
-	if (btrfs_header_level(eb) == 0)
-		return eb;
-
-	tm = tree_mod_log_search(fs_info, eb->start, time_seq);
-	if (!tm)
-		return eb;
-
-	btrfs_set_path_blocking(path);
-	btrfs_set_lock_blocking_read(eb);
-
-	if (tm->op == MOD_LOG_KEY_REMOVE_WHILE_FREEING) {
-		BUG_ON(tm->slot != 0);
-		eb_rewin = alloc_dummy_extent_buffer(fs_info, eb->start);
-		if (!eb_rewin) {
-			btrfs_tree_read_unlock_blocking(eb);
-			free_extent_buffer(eb);
-			return NULL;
-		}
-		btrfs_set_header_bytenr(eb_rewin, eb->start);
-		btrfs_set_header_backref_rev(eb_rewin,
-					     btrfs_header_backref_rev(eb));
-		btrfs_set_header_owner(eb_rewin, btrfs_header_owner(eb));
-		btrfs_set_header_level(eb_rewin, btrfs_header_level(eb));
-	} else {
-		eb_rewin = btrfs_clone_extent_buffer(eb);
-		if (!eb_rewin) {
-			btrfs_tree_read_unlock_blocking(eb);
-			free_extent_buffer(eb);
-			return NULL;
-		}
-	}
-
-	btrfs_tree_read_unlock_blocking(eb);
-	free_extent_buffer(eb);
-
-	btrfs_tree_read_lock(eb_rewin);
-	__tree_mod_log_rewind(fs_info, eb_rewin, time_seq, tm);
-	WARN_ON(btrfs_header_nritems(eb_rewin) >
-		BTRFS_NODEPTRS_PER_BLOCK(fs_info));
-
-	return eb_rewin;
-}
-
-/*
- * get_old_root() rewinds the state of @root's root node to the given @time_seq
- * value. If there are no changes, the current root->root_node is returned. If
- * anything changed in between, there's a fresh buffer allocated on which the
- * rewind operations are done. In any case, the returned buffer is read locked.
- * Returns NULL on error (with no locks held).
- */
-static inline struct extent_buffer *
-get_old_root(struct btrfs_root *root, u64 time_seq)
-{
-	struct btrfs_fs_info *fs_info = root->fs_info;
-	struct tree_mod_elem *tm;
-	struct extent_buffer *eb = NULL;
-	struct extent_buffer *eb_root;
-	u64 eb_root_owner = 0;
-	struct extent_buffer *old;
-	struct tree_mod_root *old_root = NULL;
-	u64 old_generation = 0;
-	u64 logical;
-	int level;
-
-	eb_root = btrfs_read_lock_root_node(root);
-	tm = __tree_mod_log_oldest_root(eb_root, time_seq);
-	if (!tm)
-		return eb_root;
-
-	if (tm->op == MOD_LOG_ROOT_REPLACE) {
-		old_root = &tm->old_root;
-		old_generation = tm->generation;
-		logical = old_root->logical;
-		level = old_root->level;
-	} else {
-		logical = eb_root->start;
-		level = btrfs_header_level(eb_root);
-	}
-
-	tm = tree_mod_log_search(fs_info, logical, time_seq);
-	if (old_root && tm && tm->op != MOD_LOG_KEY_REMOVE_WHILE_FREEING) {
-		btrfs_tree_read_unlock(eb_root);
-		free_extent_buffer(eb_root);
-		old = read_tree_block(fs_info, logical, 0, level, NULL);
-		if (WARN_ON(IS_ERR(old) || !extent_buffer_uptodate(old))) {
-			if (!IS_ERR(old))
-				free_extent_buffer(old);
-			btrfs_warn(fs_info,
-				   "failed to read tree block %llu from get_old_root",
-				   logical);
-		} else {
-			eb = btrfs_clone_extent_buffer(old);
-			free_extent_buffer(old);
-		}
-	} else if (old_root) {
-		eb_root_owner = btrfs_header_owner(eb_root);
-		btrfs_tree_read_unlock(eb_root);
-		free_extent_buffer(eb_root);
-		eb = alloc_dummy_extent_buffer(fs_info, logical);
-	} else {
-		btrfs_set_lock_blocking_read(eb_root);
-		eb = btrfs_clone_extent_buffer(eb_root);
-		btrfs_tree_read_unlock_blocking(eb_root);
-		free_extent_buffer(eb_root);
-	}
-
-	if (!eb)
-		return NULL;
-	btrfs_tree_read_lock(eb);
-	if (old_root) {
-		btrfs_set_header_bytenr(eb, eb->start);
-		btrfs_set_header_backref_rev(eb, BTRFS_MIXED_BACKREF_REV);
-		btrfs_set_header_owner(eb, eb_root_owner);
-		btrfs_set_header_level(eb, old_root->level);
-		btrfs_set_header_generation(eb, old_generation);
-	}
-	if (tm)
-		__tree_mod_log_rewind(fs_info, eb, time_seq, tm);
-	else
-		WARN_ON(btrfs_header_level(eb) != 0);
-	WARN_ON(btrfs_header_nritems(eb) > BTRFS_NODEPTRS_PER_BLOCK(fs_info));
-
-	return eb;
-}
-
-int btrfs_old_root_level(struct btrfs_root *root, u64 time_seq)
-{
-	struct tree_mod_elem *tm;
-	int level;
-	struct extent_buffer *eb_root = btrfs_root_node(root);
-
-	tm = __tree_mod_log_oldest_root(eb_root, time_seq);
-	if (tm && tm->op == MOD_LOG_ROOT_REPLACE) {
-		level = tm->old_root.level;
-	} else {
-		level = btrfs_header_level(eb_root);
-	}
-	free_extent_buffer(eb_root);
-
-	return level;
-}
-
 static inline int should_cow_block(struct btrfs_trans_handle *trans,
 				   struct btrfs_root *root,
 				   struct extent_buffer *buf)
@@ -1480,7 +551,8 @@ static inline int should_cow_block(struct btrfs_trans_handle *trans,
 noinline int btrfs_cow_block(struct btrfs_trans_handle *trans,
 		    struct btrfs_root *root, struct extent_buffer *buf,
 		    struct extent_buffer *parent, int parent_slot,
-		    struct extent_buffer **cow_ret)
+		    struct extent_buffer **cow_ret,
+		    enum btrfs_lock_nesting nest)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	u64 search_start;
@@ -1500,17 +572,12 @@ noinline int btrfs_cow_block(struct btrfs_trans_handle *trans,
 		       trans->transid, fs_info->generation);
 
 	if (!should_cow_block(trans, root, buf)) {
-		trans->dirty = true;
 		*cow_ret = buf;
 		return 0;
 	}
 
 	search_start = buf->start & ~((u64)SZ_1G - 1);
 
-	if (parent)
-		btrfs_set_lock_blocking_write(parent);
-	btrfs_set_lock_blocking_write(buf);
-
 	/*
 	 * Before CoWing this block for later modification, check if it's
 	 * the subtree root and do the delayed subtree trace if needed.
@@ -1519,12 +586,13 @@ noinline int btrfs_cow_block(struct btrfs_trans_handle *trans,
 	 */
 	btrfs_qgroup_trace_subtree_after_cow(trans, root, buf);
 	ret = __btrfs_cow_block(trans, root, buf, parent,
-				 parent_slot, cow_ret, search_start, 0);
+				 parent_slot, cow_ret, search_start, 0, nest);
 
 	trace_btrfs_cow_block(root, buf, *cow_ret);
 
 	return ret;
 }
+ALLOW_ERROR_INJECTION(btrfs_cow_block, ERRNO);
 
 /*
  * helper function for defrag to decide if two blocks pointed to by a
@@ -1539,6 +607,22 @@ static int close_blocks(u64 blocknr, u64 other, u32 blocksize)
 	return 0;
 }
 
+#ifdef __LITTLE_ENDIAN
+
+/*
+ * Compare two keys, on little-endian the disk order is same as CPU order and
+ * we can avoid the conversion.
+ */
+static int comp_keys(const struct btrfs_disk_key *disk_key,
+		     const struct btrfs_key *k2)
+{
+	const struct btrfs_key *k1 = (const struct btrfs_key *)disk_key;
+
+	return btrfs_comp_cpu_keys(k1, k2);
+}
+
+#else
+
 /*
  * compare two keys in a memcmp fashion
  */
@@ -1551,6 +635,7 @@ static int comp_keys(const struct btrfs_disk_key *disk,
 
 	return btrfs_comp_cpu_keys(&k1, k2);
 }
+#endif
 
 /*
  * same as comp_keys only with two btrfs_key's
@@ -1585,7 +670,6 @@ int btrfs_realloc_node(struct btrfs_trans_handle *trans,
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct extent_buffer *cur;
 	u64 blocknr;
-	u64 gen;
 	u64 search_start = *last_ret;
 	u64 last_block = 0;
 	u64 other;
@@ -1593,14 +677,10 @@ int btrfs_realloc_node(struct btrfs_trans_handle *trans,
 	int end_slot;
 	int i;
 	int err = 0;
-	int parent_level;
-	int uptodate;
 	u32 blocksize;
 	int progress_passed = 0;
 	struct btrfs_disk_key disk_key;
 
-	parent_level = btrfs_header_level(parent);
-
 	WARN_ON(trans->transaction != fs_info->running_transaction);
 	WARN_ON(trans->transid != fs_info->generation);
 
@@ -1611,10 +691,7 @@ int btrfs_realloc_node(struct btrfs_trans_handle *trans,
 	if (parent_nritems <= 1)
 		return 0;
 
-	btrfs_set_lock_blocking_write(parent);
-
 	for (i = start_slot; i <= end_slot; i++) {
-		struct btrfs_key first_key;
 		int close = 1;
 
 		btrfs_node_key(parent, &disk_key, i);
@@ -1623,8 +700,6 @@ int btrfs_realloc_node(struct btrfs_trans_handle *trans,
 
 		progress_passed = 1;
 		blocknr = btrfs_node_blockptr(parent, i);
-		gen = btrfs_node_ptr_generation(parent, i);
-		btrfs_node_key_to_cpu(parent, &first_key, i);
 		if (last_block == 0)
 			last_block = blocknr;
 
@@ -1641,40 +716,18 @@ int btrfs_realloc_node(struct btrfs_trans_handle *trans,
 			continue;
 		}
 
-		cur = find_extent_buffer(fs_info, blocknr);
-		if (cur)
-			uptodate = btrfs_buffer_uptodate(cur, gen, 0);
-		else
-			uptodate = 0;
-		if (!cur || !uptodate) {
-			if (!cur) {
-				cur = read_tree_block(fs_info, blocknr, gen,
-						      parent_level - 1,
-						      &first_key);
-				if (IS_ERR(cur)) {
-					return PTR_ERR(cur);
-				} else if (!extent_buffer_uptodate(cur)) {
-					free_extent_buffer(cur);
-					return -EIO;
-				}
-			} else if (!uptodate) {
-				err = btrfs_read_buffer(cur, gen,
-						parent_level - 1,&first_key);
-				if (err) {
-					free_extent_buffer(cur);
-					return err;
-				}
-			}
-		}
+		cur = btrfs_read_node_slot(parent, i);
+		if (IS_ERR(cur))
+			return PTR_ERR(cur);
 		if (search_start == 0)
 			search_start = last_block;
 
 		btrfs_tree_lock(cur);
-		btrfs_set_lock_blocking_write(cur);
 		err = __btrfs_cow_block(trans, root, cur, parent, i,
 					&cur, search_start,
 					min(16 * blocksize,
-					    (end_slot - i) * blocksize));
+					    (end_slot - i) * blocksize),
+					BTRFS_NESTING_COW);
 		if (err) {
 			btrfs_tree_unlock(cur);
 			free_extent_buffer(cur);
@@ -1690,31 +743,26 @@ int btrfs_realloc_node(struct btrfs_trans_handle *trans,
 }
 
 /*
- * search for key in the extent_buffer.  The items start at offset p,
- * and they are item_size apart.  There are 'max' items in p.
+ * Search for a key in the given extent_buffer.
  *
- * the slot in the array is returned via slot, and it points to
- * the place where you would insert key if it is not found in
- * the array.
+ * The lower boundary for the search is specified by the slot number @low. Use a
+ * value of 0 to search over the whole extent buffer.
  *
- * slot may point to max if the key is bigger than all of the keys
+ * The slot in the extent buffer is returned via @slot. If the key exists in the
+ * extent buffer, then @slot will point to the slot where the key is, otherwise
+ * it points to the slot where you would insert the key.
+ *
+ * Slot may point to the total number of items (i.e. one position beyond the last
+ * key) if the key is bigger than the last key in the extent buffer.
  */
-static noinline int generic_bin_search(struct extent_buffer *eb,
-				       unsigned long p, int item_size,
-				       const struct btrfs_key *key,
-				       int max, int *slot)
+static noinline int generic_bin_search(struct extent_buffer *eb, int low,
+				       const struct btrfs_key *key, int *slot)
 {
-	int low = 0;
-	int high = max;
-	int mid;
+	unsigned long p;
+	int item_size;
+	int high = btrfs_header_nritems(eb);
 	int ret;
-	struct btrfs_disk_key *tmp = NULL;
-	struct btrfs_disk_key unaligned;
-	unsigned long offset;
-	char *kaddr = NULL;
-	unsigned long map_start = 0;
-	unsigned long map_len = 0;
-	int err;
+	const int key_size = sizeof(struct btrfs_disk_key);
 
 	if (low > high) {
 		btrfs_err(eb->fs_info,
@@ -1724,33 +772,36 @@ static noinline int generic_bin_search(struct extent_buffer *eb,
 		return -EINVAL;
 	}
 
+	if (btrfs_header_level(eb) == 0) {
+		p = offsetof(struct btrfs_leaf, items);
+		item_size = sizeof(struct btrfs_item);
+	} else {
+		p = offsetof(struct btrfs_node, ptrs);
+		item_size = sizeof(struct btrfs_key_ptr);
+	}
+
 	while (low < high) {
+		unsigned long oip;
+		unsigned long offset;
+		struct btrfs_disk_key *tmp;
+		struct btrfs_disk_key unaligned;
+		int mid;
+
 		mid = (low + high) / 2;
 		offset = p + mid * item_size;
+		oip = offset_in_page(offset);
 
-		if (!kaddr || offset < map_start ||
-		    (offset + sizeof(struct btrfs_disk_key)) >
-		    map_start + map_len) {
-
-			err = map_private_extent_buffer(eb, offset,
-						sizeof(struct btrfs_disk_key),
-						&kaddr, &map_start, &map_len);
-
-			if (!err) {
-				tmp = (struct btrfs_disk_key *)(kaddr + offset -
-							map_start);
-			} else if (err == 1) {
-				read_extent_buffer(eb, &unaligned,
-						   offset, sizeof(unaligned));
-				tmp = &unaligned;
-			} else {
-				return err;
-			}
+		if (oip + key_size <= PAGE_SIZE) {
+			const unsigned long idx = get_eb_page_index(offset);
+			char *kaddr = page_address(eb->pages[idx]);
 
+			oip = get_eb_offset_in_page(eb, offset);
+			tmp = (struct btrfs_disk_key *)(kaddr + oip);
 		} else {
-			tmp = (struct btrfs_disk_key *)(kaddr + offset -
-							map_start);
+			read_extent_buffer(eb, &unaligned, offset, key_size);
+			tmp = &unaligned;
 		}
+
 		ret = comp_keys(tmp, key);
 
 		if (ret < 0)
@@ -1767,24 +818,13 @@ static noinline int generic_bin_search(struct extent_buffer *eb,
 }
 
 /*
- * simple bin_search frontend that does the right thing for
- * leaves vs nodes
+ * Simple binary search on an extent buffer. Works for both leaves and nodes, and
+ * always searches over the whole range of keys (slot 0 to slot 'nritems - 1').
  */
 int btrfs_bin_search(struct extent_buffer *eb, const struct btrfs_key *key,
-		     int level, int *slot)
+		     int *slot)
 {
-	if (level == 0)
-		return generic_bin_search(eb,
-					  offsetof(struct btrfs_leaf, items),
-					  sizeof(struct btrfs_item),
-					  key, btrfs_header_nritems(eb),
-					  slot);
-	else
-		return generic_bin_search(eb,
-					  offsetof(struct btrfs_node, ptrs),
-					  sizeof(struct btrfs_key_ptr),
-					  key, btrfs_header_nritems(eb),
-					  slot);
+	return generic_bin_search(eb, 0, key, slot);
 }
 
 static void root_add_used(struct btrfs_root *root, u32 size)
@@ -1820,11 +860,14 @@ struct extent_buffer *btrfs_read_node_slot(struct extent_buffer *parent,
 
 	btrfs_node_key_to_cpu(parent, &first_key, slot);
 	eb = read_tree_block(parent->fs_info, btrfs_node_blockptr(parent, slot),
+			     btrfs_header_owner(parent),
 			     btrfs_node_ptr_generation(parent, slot),
 			     level - 1, &first_key);
-	if (!IS_ERR(eb) && !extent_buffer_uptodate(eb)) {
+	if (IS_ERR(eb))
+		return eb;
+	if (!extent_buffer_uptodate(eb)) {
 		free_extent_buffer(eb);
-		eb = ERR_PTR(-EIO);
+		return ERR_PTR(-EIO);
 	}
 
 	return eb;
@@ -1854,8 +897,7 @@ static noinline int balance_level(struct btrfs_trans_handle *trans,
 
 	mid = path->nodes[level];
 
-	WARN_ON(path->locks[level] != BTRFS_WRITE_LOCK &&
-		path->locks[level] != BTRFS_WRITE_LOCK_BLOCKING);
+	WARN_ON(path->locks[level] != BTRFS_WRITE_LOCK);
 	WARN_ON(btrfs_header_generation(mid) != trans->transid);
 
 	orig_ptr = btrfs_node_blockptr(mid, orig_slot);
@@ -1884,15 +926,15 @@ static noinline int balance_level(struct btrfs_trans_handle *trans,
 		}
 
 		btrfs_tree_lock(child);
-		btrfs_set_lock_blocking_write(child);
-		ret = btrfs_cow_block(trans, root, child, mid, 0, &child);
+		ret = btrfs_cow_block(trans, root, child, mid, 0, &child,
+				      BTRFS_NESTING_COW);
 		if (ret) {
 			btrfs_tree_unlock(child);
 			free_extent_buffer(child);
 			goto enospc;
 		}
 
-		ret = tree_mod_log_insert_root(root->node, child, 1);
+		ret = btrfs_tree_mod_log_insert_root(root->node, child, true);
 		BUG_ON(ret < 0);
 		rcu_assign_pointer(root->node, child);
 
@@ -1907,7 +949,7 @@ static noinline int balance_level(struct btrfs_trans_handle *trans,
 		free_extent_buffer(mid);
 
 		root_sub_used(root, mid->len);
-		btrfs_free_tree_block(trans, root, mid, 0, 1);
+		btrfs_free_tree_block(trans, btrfs_root_id(root), mid, 0, 1);
 		/* once for the root ptr */
 		free_extent_buffer_stale(mid);
 		return 0;
@@ -1921,10 +963,10 @@ static noinline int balance_level(struct btrfs_trans_handle *trans,
 		left = NULL;
 
 	if (left) {
-		btrfs_tree_lock(left);
-		btrfs_set_lock_blocking_write(left);
+		__btrfs_tree_lock(left, BTRFS_NESTING_LEFT);
 		wret = btrfs_cow_block(trans, root, left,
-				       parent, pslot - 1, &left);
+				       parent, pslot - 1, &left,
+				       BTRFS_NESTING_LEFT_COW);
 		if (wret) {
 			ret = wret;
 			goto enospc;
@@ -1936,10 +978,10 @@ static noinline int balance_level(struct btrfs_trans_handle *trans,
 		right = NULL;
 
 	if (right) {
-		btrfs_tree_lock(right);
-		btrfs_set_lock_blocking_write(right);
+		__btrfs_tree_lock(right, BTRFS_NESTING_RIGHT);
 		wret = btrfs_cow_block(trans, root, right,
-				       parent, pslot + 1, &right);
+				       parent, pslot + 1, &right,
+				       BTRFS_NESTING_RIGHT_COW);
 		if (wret) {
 			ret = wret;
 			goto enospc;
@@ -1966,14 +1008,15 @@ static noinline int balance_level(struct btrfs_trans_handle *trans,
 			btrfs_tree_unlock(right);
 			del_ptr(root, path, level + 1, pslot + 1);
 			root_sub_used(root, right->len);
-			btrfs_free_tree_block(trans, root, right, 0, 1);
+			btrfs_free_tree_block(trans, btrfs_root_id(root), right,
+					      0, 1);
 			free_extent_buffer_stale(right);
 			right = NULL;
 		} else {
 			struct btrfs_disk_key right_key;
 			btrfs_node_key(right, &right_key, 0);
-			ret = tree_mod_log_insert_key(parent, pslot + 1,
-					MOD_LOG_KEY_REPLACE, GFP_NOFS);
+			ret = btrfs_tree_mod_log_insert_key(parent, pslot + 1,
+					BTRFS_MOD_LOG_KEY_REPLACE, GFP_NOFS);
 			BUG_ON(ret < 0);
 			btrfs_set_node_key(parent, &right_key, pslot + 1);
 			btrfs_mark_buffer_dirty(parent);
@@ -2011,15 +1054,15 @@ static noinline int balance_level(struct btrfs_trans_handle *trans,
 		btrfs_tree_unlock(mid);
 		del_ptr(root, path, level + 1, pslot);
 		root_sub_used(root, mid->len);
-		btrfs_free_tree_block(trans, root, mid, 0, 1);
+		btrfs_free_tree_block(trans, btrfs_root_id(root), mid, 0, 1);
 		free_extent_buffer_stale(mid);
 		mid = NULL;
 	} else {
 		/* update the parent key to reflect our changes */
 		struct btrfs_disk_key mid_key;
 		btrfs_node_key(mid, &mid_key, 0);
-		ret = tree_mod_log_insert_key(parent, pslot,
-				MOD_LOG_KEY_REPLACE, GFP_NOFS);
+		ret = btrfs_tree_mod_log_insert_key(parent, pslot,
+				BTRFS_MOD_LOG_KEY_REPLACE, GFP_NOFS);
 		BUG_ON(ret < 0);
 		btrfs_set_node_key(parent, &mid_key, pslot);
 		btrfs_mark_buffer_dirty(parent);
@@ -2099,15 +1142,15 @@ static noinline int push_nodes_for_insert(struct btrfs_trans_handle *trans,
 	if (left) {
 		u32 left_nr;
 
-		btrfs_tree_lock(left);
-		btrfs_set_lock_blocking_write(left);
+		__btrfs_tree_lock(left, BTRFS_NESTING_LEFT);
 
 		left_nr = btrfs_header_nritems(left);
 		if (left_nr >= BTRFS_NODEPTRS_PER_BLOCK(fs_info) - 1) {
 			wret = 1;
 		} else {
 			ret = btrfs_cow_block(trans, root, left, parent,
-					      pslot - 1, &left);
+					      pslot - 1, &left,
+					      BTRFS_NESTING_LEFT_COW);
 			if (ret)
 				wret = 1;
 			else {
@@ -2120,8 +1163,8 @@ static noinline int push_nodes_for_insert(struct btrfs_trans_handle *trans,
 			struct btrfs_disk_key disk_key;
 			orig_slot += left_nr;
 			btrfs_node_key(mid, &disk_key, 0);
-			ret = tree_mod_log_insert_key(parent, pslot,
-					MOD_LOG_KEY_REPLACE, GFP_NOFS);
+			ret = btrfs_tree_mod_log_insert_key(parent, pslot,
+					BTRFS_MOD_LOG_KEY_REPLACE, GFP_NOFS);
 			BUG_ON(ret < 0);
 			btrfs_set_node_key(parent, &disk_key, pslot);
 			btrfs_mark_buffer_dirty(parent);
@@ -2153,8 +1196,7 @@ static noinline int push_nodes_for_insert(struct btrfs_trans_handle *trans,
 	if (right) {
 		u32 right_nr;
 
-		btrfs_tree_lock(right);
-		btrfs_set_lock_blocking_write(right);
+		__btrfs_tree_lock(right, BTRFS_NESTING_RIGHT);
 
 		right_nr = btrfs_header_nritems(right);
 		if (right_nr >= BTRFS_NODEPTRS_PER_BLOCK(fs_info) - 1) {
@@ -2162,7 +1204,7 @@ static noinline int push_nodes_for_insert(struct btrfs_trans_handle *trans,
 		} else {
 			ret = btrfs_cow_block(trans, root, right,
 					      parent, pslot + 1,
-					      &right);
+					      &right, BTRFS_NESTING_RIGHT_COW);
 			if (ret)
 				wret = 1;
 			else {
@@ -2175,8 +1217,8 @@ static noinline int push_nodes_for_insert(struct btrfs_trans_handle *trans,
 			struct btrfs_disk_key disk_key;
 
 			btrfs_node_key(right, &disk_key, 0);
-			ret = tree_mod_log_insert_key(parent, pslot + 1,
-					MOD_LOG_KEY_REPLACE, GFP_NOFS);
+			ret = btrfs_tree_mod_log_insert_key(parent, pslot + 1,
+					BTRFS_MOD_LOG_KEY_REPLACE, GFP_NOFS);
 			BUG_ON(ret < 0);
 			btrfs_set_node_key(parent, &disk_key, pslot + 1);
 			btrfs_mark_buffer_dirty(parent);
@@ -2214,12 +1256,12 @@ static void reada_for_search(struct btrfs_fs_info *fs_info,
 	u64 search;
 	u64 target;
 	u64 nread = 0;
-	struct extent_buffer *eb;
+	u64 nread_max;
 	u32 nr;
 	u32 blocksize;
 	u32 nscan = 0;
 
-	if (level != 1)
+	if (level != 1 && path->reada != READA_FORWARD_ALWAYS)
 		return;
 
 	if (!path->nodes[level])
@@ -2227,12 +1269,30 @@ static void reada_for_search(struct btrfs_fs_info *fs_info,
 
 	node = path->nodes[level];
 
+	/*
+	 * Since the time between visiting leaves is much shorter than the time
+	 * between visiting nodes, limit read ahead of nodes to 1, to avoid too
+	 * much IO at once (possibly random).
+	 */
+	if (path->reada == READA_FORWARD_ALWAYS) {
+		if (level > 1)
+			nread_max = node->fs_info->nodesize;
+		else
+			nread_max = SZ_128K;
+	} else {
+		nread_max = SZ_64K;
+	}
+
 	search = btrfs_node_blockptr(node, slot);
 	blocksize = fs_info->nodesize;
-	eb = find_extent_buffer(fs_info, search);
-	if (eb) {
-		free_extent_buffer(eb);
-		return;
+	if (path->reada != READA_FORWARD_ALWAYS) {
+		struct extent_buffer *eb;
+
+		eb = find_extent_buffer(fs_info, search);
+		if (eb) {
+			free_extent_buffer(eb);
+			return;
+		}
 	}
 
 	target = search;
@@ -2245,7 +1305,8 @@ static void reada_for_search(struct btrfs_fs_info *fs_info,
 			if (nr == 0)
 				break;
 			nr--;
-		} else if (path->reada == READA_FORWARD) {
+		} else if (path->reada == READA_FORWARD ||
+			   path->reada == READA_FORWARD_ALWAYS) {
 			nr++;
 			if (nr >= nritems)
 				break;
@@ -2256,27 +1317,23 @@ static void reada_for_search(struct btrfs_fs_info *fs_info,
 				break;
 		}
 		search = btrfs_node_blockptr(node, nr);
-		if ((search <= target && target - search <= 65536) ||
+		if (path->reada == READA_FORWARD_ALWAYS ||
+		    (search <= target && target - search <= 65536) ||
 		    (search > target && search - target <= 65536)) {
-			readahead_tree_block(fs_info, search);
+			btrfs_readahead_node_child(node, nr);
 			nread += blocksize;
 		}
 		nscan++;
-		if ((nread > 65536 || nscan > 32))
+		if (nread > nread_max || nscan > 32)
 			break;
 	}
 }
 
-static noinline void reada_for_balance(struct btrfs_fs_info *fs_info,
-				       struct btrfs_path *path, int level)
+static noinline void reada_for_balance(struct btrfs_path *path, int level)
 {
+	struct extent_buffer *parent;
 	int slot;
 	int nritems;
-	struct extent_buffer *parent;
-	struct extent_buffer *eb;
-	u64 gen;
-	u64 block1 = 0;
-	u64 block2 = 0;
 
 	parent = path->nodes[level + 1];
 	if (!parent)
@@ -2285,32 +1342,10 @@ static noinline void reada_for_balance(struct btrfs_fs_info *fs_info,
 	nritems = btrfs_header_nritems(parent);
 	slot = path->slots[level + 1];
 
-	if (slot > 0) {
-		block1 = btrfs_node_blockptr(parent, slot - 1);
-		gen = btrfs_node_ptr_generation(parent, slot - 1);
-		eb = find_extent_buffer(fs_info, block1);
-		/*
-		 * if we get -eagain from btrfs_buffer_uptodate, we
-		 * don't want to return eagain here.  That will loop
-		 * forever
-		 */
-		if (eb && btrfs_buffer_uptodate(eb, gen, 1) != 0)
-			block1 = 0;
-		free_extent_buffer(eb);
-	}
-	if (slot + 1 < nritems) {
-		block2 = btrfs_node_blockptr(parent, slot + 1);
-		gen = btrfs_node_ptr_generation(parent, slot + 1);
-		eb = find_extent_buffer(fs_info, block2);
-		if (eb && btrfs_buffer_uptodate(eb, gen, 1) != 0)
-			block2 = 0;
-		free_extent_buffer(eb);
-	}
-
-	if (block1)
-		readahead_tree_block(fs_info, block1);
-	if (block2)
-		readahead_tree_block(fs_info, block2);
+	if (slot > 0)
+		btrfs_readahead_node_child(parent, slot - 1);
+	if (slot + 1 < nritems)
+		btrfs_readahead_node_child(parent, slot + 1);
 }
 
 
@@ -2333,33 +1368,34 @@ static noinline void unlock_up(struct btrfs_path *path, int level,
 {
 	int i;
 	int skip_level = level;
-	int no_skips = 0;
-	struct extent_buffer *t;
+	bool check_skip = true;
 
 	for (i = level; i < BTRFS_MAX_LEVEL; i++) {
 		if (!path->nodes[i])
 			break;
 		if (!path->locks[i])
 			break;
-		if (!no_skips && path->slots[i] == 0) {
-			skip_level = i + 1;
-			continue;
-		}
-		if (!no_skips && path->keep_locks) {
-			u32 nritems;
-			t = path->nodes[i];
-			nritems = btrfs_header_nritems(t);
-			if (nritems < 1 || path->slots[i] >= nritems - 1) {
+
+		if (check_skip) {
+			if (path->slots[i] == 0) {
 				skip_level = i + 1;
 				continue;
 			}
+
+			if (path->keep_locks) {
+				u32 nritems;
+
+				nritems = btrfs_header_nritems(path->nodes[i]);
+				if (nritems < 1 || path->slots[i] >= nritems - 1) {
+					skip_level = i + 1;
+					continue;
+				}
+			}
 		}
-		if (skip_level < i && i >= lowest_unlock)
-			no_skips = 1;
 
-		t = path->nodes[i];
 		if (i >= lowest_unlock && i > skip_level) {
-			btrfs_tree_unlock_rw(t, path->locks[i]);
+			check_skip = false;
+			btrfs_tree_unlock_rw(path->nodes[i], path->locks[i]);
 			path->locks[i] = 0;
 			if (write_lock_level &&
 			    i > min_write_lock_level &&
@@ -2371,12 +1407,13 @@ static noinline void unlock_up(struct btrfs_path *path, int level,
 }
 
 /*
- * helper function for btrfs_search_slot.  The goal is to find a block
- * in cache without setting the path to blocking.  If we find the block
- * we return zero and the path is unchanged.
+ * Helper function for btrfs_search_slot() and other functions that do a search
+ * on a btree. The goal is to find a tree block in the cache (the radix tree at
+ * fs_info->buffer_radix), but if we can't find it, or it's not up to date, read
+ * its pages from disk.
  *
- * If we can't find the block, we set the path blocking and do some
- * reada.  -EAGAIN is returned and the search must be repeated.
+ * Returns -EAGAIN, with the path unlocked, if the caller needs to repeat the
+ * whole btree search, starting again from the current root node.
  */
 static int
 read_block_for_search(struct btrfs_root *root, struct btrfs_path *p,
@@ -2386,19 +1423,30 @@ read_block_for_search(struct btrfs_root *root, struct btrfs_path *p,
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	u64 blocknr;
 	u64 gen;
-	struct extent_buffer *b = *eb_ret;
 	struct extent_buffer *tmp;
 	struct btrfs_key first_key;
 	int ret;
 	int parent_level;
+	bool unlock_up;
 
-	blocknr = btrfs_node_blockptr(b, slot);
-	gen = btrfs_node_ptr_generation(b, slot);
-	parent_level = btrfs_header_level(b);
-	btrfs_node_key_to_cpu(b, &first_key, slot);
+	unlock_up = ((level + 1 < BTRFS_MAX_LEVEL) && p->locks[level + 1]);
+	blocknr = btrfs_node_blockptr(*eb_ret, slot);
+	gen = btrfs_node_ptr_generation(*eb_ret, slot);
+	parent_level = btrfs_header_level(*eb_ret);
+	btrfs_node_key_to_cpu(*eb_ret, &first_key, slot);
 
+	/*
+	 * If we need to read an extent buffer from disk and we are holding locks
+	 * on upper level nodes, we unlock all the upper nodes before reading the
+	 * extent buffer, and then return -EAGAIN to the caller as it needs to
+	 * restart the search. We don't release the lock on the current level
+	 * because we need to walk this node to figure out which blocks to read.
+	 */
 	tmp = find_extent_buffer(fs_info, blocknr);
 	if (tmp) {
+		if (p->reada == READA_FORWARD_ALWAYS)
+			reada_for_search(fs_info, p, level, slot, key->objectid);
+
 		/* first we do an atomic uptodate check */
 		if (btrfs_buffer_uptodate(tmp, gen, 1) > 0) {
 			/*
@@ -2415,56 +1463,68 @@ read_block_for_search(struct btrfs_root *root, struct btrfs_path *p,
 			return 0;
 		}
 
-		/* the pages were up to date, but we failed
-		 * the generation number check.  Do a full
-		 * read for the generation number that is correct.
-		 * We must do this without dropping locks so
-		 * we can trust our generation number
-		 */
-		btrfs_set_path_blocking(p);
+		if (p->nowait) {
+			free_extent_buffer(tmp);
+			return -EAGAIN;
+		}
+
+		if (unlock_up)
+			btrfs_unlock_up_safe(p, level + 1);
 
 		/* now we're allowed to do a blocking uptodate check */
-		ret = btrfs_read_buffer(tmp, gen, parent_level - 1, &first_key);
-		if (!ret) {
-			*eb_ret = tmp;
-			return 0;
+		ret = btrfs_read_extent_buffer(tmp, gen, parent_level - 1, &first_key);
+		if (ret) {
+			free_extent_buffer(tmp);
+			btrfs_release_path(p);
+			return -EIO;
 		}
-		free_extent_buffer(tmp);
-		btrfs_release_path(p);
-		return -EIO;
+		if (btrfs_check_eb_owner(tmp, root->root_key.objectid)) {
+			free_extent_buffer(tmp);
+			btrfs_release_path(p);
+			return -EUCLEAN;
+		}
+
+		if (unlock_up)
+			ret = -EAGAIN;
+
+		goto out;
+	} else if (p->nowait) {
+		return -EAGAIN;
 	}
 
-	/*
-	 * reduce lock contention at high levels
-	 * of the btree by dropping locks before
-	 * we read.  Don't release the lock on the current
-	 * level because we need to walk this node to figure
-	 * out which blocks to read.
-	 */
-	btrfs_unlock_up_safe(p, level + 1);
-	btrfs_set_path_blocking(p);
+	if (unlock_up) {
+		btrfs_unlock_up_safe(p, level + 1);
+		ret = -EAGAIN;
+	} else {
+		ret = 0;
+	}
 
 	if (p->reada != READA_NONE)
 		reada_for_search(fs_info, p, level, slot, key->objectid);
 
-	ret = -EAGAIN;
-	tmp = read_tree_block(fs_info, blocknr, gen, parent_level - 1,
-			      &first_key);
-	if (!IS_ERR(tmp)) {
-		/*
-		 * If the read above didn't mark this buffer up to date,
-		 * it will never end up being up to date.  Set ret to EIO now
-		 * and give up so that our caller doesn't loop forever
-		 * on our EAGAINs.
-		 */
-		if (!extent_buffer_uptodate(tmp))
-			ret = -EIO;
-		free_extent_buffer(tmp);
+	tmp = read_tree_block(fs_info, blocknr, root->root_key.objectid,
+			      gen, parent_level - 1, &first_key);
+	if (IS_ERR(tmp)) {
+		btrfs_release_path(p);
+		return PTR_ERR(tmp);
+	}
+	/*
+	 * If the read above didn't mark this buffer up to date,
+	 * it will never end up being up to date.  Set ret to EIO now
+	 * and give up so that our caller doesn't loop forever
+	 * on our EAGAINs.
+	 */
+	if (!extent_buffer_uptodate(tmp))
+		ret = -EIO;
+
+out:
+	if (ret == 0) {
+		*eb_ret = tmp;
 	} else {
-		ret = PTR_ERR(tmp);
+		free_extent_buffer(tmp);
+		btrfs_release_path(p);
 	}
 
-	btrfs_release_path(p);
 	return ret;
 }
 
@@ -2484,74 +1544,45 @@ setup_nodes_for_search(struct btrfs_trans_handle *trans,
 		       int *write_lock_level)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
-	int ret;
+	int ret = 0;
 
 	if ((p->search_for_split || ins_len > 0) && btrfs_header_nritems(b) >=
 	    BTRFS_NODEPTRS_PER_BLOCK(fs_info) - 3) {
-		int sret;
 
 		if (*write_lock_level < level + 1) {
 			*write_lock_level = level + 1;
 			btrfs_release_path(p);
-			goto again;
+			return -EAGAIN;
 		}
 
-		btrfs_set_path_blocking(p);
-		reada_for_balance(fs_info, p, level);
-		sret = split_node(trans, root, p, level);
+		reada_for_balance(p, level);
+		ret = split_node(trans, root, p, level);
 
-		BUG_ON(sret > 0);
-		if (sret) {
-			ret = sret;
-			goto done;
-		}
 		b = p->nodes[level];
 	} else if (ins_len < 0 && btrfs_header_nritems(b) <
 		   BTRFS_NODEPTRS_PER_BLOCK(fs_info) / 2) {
-		int sret;
 
 		if (*write_lock_level < level + 1) {
 			*write_lock_level = level + 1;
 			btrfs_release_path(p);
-			goto again;
+			return -EAGAIN;
 		}
 
-		btrfs_set_path_blocking(p);
-		reada_for_balance(fs_info, p, level);
-		sret = balance_level(trans, root, p, level);
+		reada_for_balance(p, level);
+		ret = balance_level(trans, root, p, level);
+		if (ret)
+			return ret;
 
-		if (sret) {
-			ret = sret;
-			goto done;
-		}
 		b = p->nodes[level];
 		if (!b) {
 			btrfs_release_path(p);
-			goto again;
+			return -EAGAIN;
 		}
 		BUG_ON(btrfs_header_nritems(b) == 1);
 	}
-	return 0;
-
-again:
-	ret = -EAGAIN;
-done:
 	return ret;
 }
 
-static int key_search(struct extent_buffer *b, const struct btrfs_key *key,
-		      int level, int *prev_cmp, int *slot)
-{
-	if (*prev_cmp != 0) {
-		*prev_cmp = btrfs_bin_search(b, key, level, slot);
-		return *prev_cmp;
-	}
-
-	*slot = 0;
-
-	return 0;
-}
-
 int btrfs_find_item(struct btrfs_root *fs_root, struct btrfs_path *path,
 		u64 iobjectid, u64 ioff, u8 key_type,
 		struct btrfs_key *found_key)
@@ -2591,35 +1622,13 @@ static struct extent_buffer *btrfs_search_slot_get_root(struct btrfs_root *root,
 							struct btrfs_path *p,
 							int write_lock_level)
 {
-	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct extent_buffer *b;
-	int root_lock;
+	int root_lock = 0;
 	int level = 0;
 
-	/* We try very hard to do read locks on the root */
-	root_lock = BTRFS_READ_LOCK;
-
 	if (p->search_commit_root) {
-		/*
-		 * The commit roots are read only so we always do read locks,
-		 * and we always must hold the commit_root_sem when doing
-		 * searches on them, the only exception is send where we don't
-		 * want to block transaction commits for a long time, so
-		 * we need to clone the commit root in order to avoid races
-		 * with transaction commits that create a snapshot of one of
-		 * the roots used by a send operation.
-		 */
-		if (p->need_commit_sem) {
-			down_read(&fs_info->commit_root_sem);
-			b = btrfs_clone_extent_buffer(root->commit_root);
-			up_read(&fs_info->commit_root_sem);
-			if (!b)
-				return ERR_PTR(-ENOMEM);
-
-		} else {
-			b = root->commit_root;
-			atomic_inc(&b->refs);
-		}
+		b = root->commit_root;
+		atomic_inc(&b->refs);
 		level = btrfs_header_level(b);
 		/*
 		 * Ensure that all callers have set skip_locking when
@@ -2636,6 +1645,9 @@ static struct extent_buffer *btrfs_search_slot_get_root(struct btrfs_root *root,
 		goto out;
 	}
 
+	/* We try very hard to do read locks on the root */
+	root_lock = BTRFS_READ_LOCK;
+
 	/*
 	 * If the level is set to maximum, we can skip trying to get the read
 	 * lock.
@@ -2645,7 +1657,13 @@ static struct extent_buffer *btrfs_search_slot_get_root(struct btrfs_root *root,
 		 * We don't know the level of the root node until we actually
 		 * have it read locked
 		 */
-		b = btrfs_read_lock_root_node(root);
+		if (p->nowait) {
+			b = btrfs_try_read_lock_root_node(root);
+			if (IS_ERR(b))
+				return b;
+		} else {
+			b = btrfs_read_lock_root_node(root);
+		}
 		level = btrfs_header_level(b);
 		if (level > write_lock_level)
 			goto out;
@@ -2662,6 +1680,17 @@ static struct extent_buffer *btrfs_search_slot_get_root(struct btrfs_root *root,
 	level = btrfs_header_level(b);
 
 out:
+	/*
+	 * The root may have failed to write out at some point, and thus is no
+	 * longer valid, return an error in this case.
+	 */
+	if (!extent_buffer_uptodate(b)) {
+		if (root_lock)
+			btrfs_tree_unlock_rw(b, root_lock);
+		free_extent_buffer(b);
+		return ERR_PTR(-EIO);
+	}
+
 	p->nodes[level] = b;
 	if (!p->skip_locking)
 		p->locks[level] = root_lock;
@@ -2671,6 +1700,191 @@ out:
 	return b;
 }
 
+/*
+ * Replace the extent buffer at the lowest level of the path with a cloned
+ * version. The purpose is to be able to use it safely, after releasing the
+ * commit root semaphore, even if relocation is happening in parallel, the
+ * transaction used for relocation is committed and the extent buffer is
+ * reallocated in the next transaction.
+ *
+ * This is used in a context where the caller does not prevent transaction
+ * commits from happening, either by holding a transaction handle or holding
+ * some lock, while it's doing searches through a commit root.
+ * At the moment it's only used for send operations.
+ */
+static int finish_need_commit_sem_search(struct btrfs_path *path)
+{
+	const int i = path->lowest_level;
+	const int slot = path->slots[i];
+	struct extent_buffer *lowest = path->nodes[i];
+	struct extent_buffer *clone;
+
+	ASSERT(path->need_commit_sem);
+
+	if (!lowest)
+		return 0;
+
+	lockdep_assert_held_read(&lowest->fs_info->commit_root_sem);
+
+	clone = btrfs_clone_extent_buffer(lowest);
+	if (!clone)
+		return -ENOMEM;
+
+	btrfs_release_path(path);
+	path->nodes[i] = clone;
+	path->slots[i] = slot;
+
+	return 0;
+}
+
+static inline int search_for_key_slot(struct extent_buffer *eb,
+				      int search_low_slot,
+				      const struct btrfs_key *key,
+				      int prev_cmp,
+				      int *slot)
+{
+	/*
+	 * If a previous call to btrfs_bin_search() on a parent node returned an
+	 * exact match (prev_cmp == 0), we can safely assume the target key will
+	 * always be at slot 0 on lower levels, since each key pointer
+	 * (struct btrfs_key_ptr) refers to the lowest key accessible from the
+	 * subtree it points to. Thus we can skip searching lower levels.
+	 */
+	if (prev_cmp == 0) {
+		*slot = 0;
+		return 0;
+	}
+
+	return generic_bin_search(eb, search_low_slot, key, slot);
+}
+
+static int search_leaf(struct btrfs_trans_handle *trans,
+		       struct btrfs_root *root,
+		       const struct btrfs_key *key,
+		       struct btrfs_path *path,
+		       int ins_len,
+		       int prev_cmp)
+{
+	struct extent_buffer *leaf = path->nodes[0];
+	int leaf_free_space = -1;
+	int search_low_slot = 0;
+	int ret;
+	bool do_bin_search = true;
+
+	/*
+	 * If we are doing an insertion, the leaf has enough free space and the
+	 * destination slot for the key is not slot 0, then we can unlock our
+	 * write lock on the parent, and any other upper nodes, before doing the
+	 * binary search on the leaf (with search_for_key_slot()), allowing other
+	 * tasks to lock the parent and any other upper nodes.
+	 */
+	if (ins_len > 0) {
+		/*
+		 * Cache the leaf free space, since we will need it later and it
+		 * will not change until then.
+		 */
+		leaf_free_space = btrfs_leaf_free_space(leaf);
+
+		/*
+		 * !path->locks[1] means we have a single node tree, the leaf is
+		 * the root of the tree.
+		 */
+		if (path->locks[1] && leaf_free_space >= ins_len) {
+			struct btrfs_disk_key first_key;
+
+			ASSERT(btrfs_header_nritems(leaf) > 0);
+			btrfs_item_key(leaf, &first_key, 0);
+
+			/*
+			 * Doing the extra comparison with the first key is cheap,
+			 * taking into account that the first key is very likely
+			 * already in a cache line because it immediately follows
+			 * the extent buffer's header and we have recently accessed
+			 * the header's level field.
+			 */
+			ret = comp_keys(&first_key, key);
+			if (ret < 0) {
+				/*
+				 * The first key is smaller than the key we want
+				 * to insert, so we are safe to unlock all upper
+				 * nodes and we have to do the binary search.
+				 *
+				 * We do use btrfs_unlock_up_safe() and not
+				 * unlock_up() because the later does not unlock
+				 * nodes with a slot of 0 - we can safely unlock
+				 * any node even if its slot is 0 since in this
+				 * case the key does not end up at slot 0 of the
+				 * leaf and there's no need to split the leaf.
+				 */
+				btrfs_unlock_up_safe(path, 1);
+				search_low_slot = 1;
+			} else {
+				/*
+				 * The first key is >= then the key we want to
+				 * insert, so we can skip the binary search as
+				 * the target key will be at slot 0.
+				 *
+				 * We can not unlock upper nodes when the key is
+				 * less than the first key, because we will need
+				 * to update the key at slot 0 of the parent node
+				 * and possibly of other upper nodes too.
+				 * If the key matches the first key, then we can
+				 * unlock all the upper nodes, using
+				 * btrfs_unlock_up_safe() instead of unlock_up()
+				 * as stated above.
+				 */
+				if (ret == 0)
+					btrfs_unlock_up_safe(path, 1);
+				/*
+				 * ret is already 0 or 1, matching the result of
+				 * a btrfs_bin_search() call, so there is no need
+				 * to adjust it.
+				 */
+				do_bin_search = false;
+				path->slots[0] = 0;
+			}
+		}
+	}
+
+	if (do_bin_search) {
+		ret = search_for_key_slot(leaf, search_low_slot, key,
+					  prev_cmp, &path->slots[0]);
+		if (ret < 0)
+			return ret;
+	}
+
+	if (ins_len > 0) {
+		/*
+		 * Item key already exists. In this case, if we are allowed to
+		 * insert the item (for example, in dir_item case, item key
+		 * collision is allowed), it will be merged with the original
+		 * item. Only the item size grows, no new btrfs item will be
+		 * added. If search_for_extension is not set, ins_len already
+		 * accounts the size btrfs_item, deduct it here so leaf space
+		 * check will be correct.
+		 */
+		if (ret == 0 && !path->search_for_extension) {
+			ASSERT(ins_len >= sizeof(struct btrfs_item));
+			ins_len -= sizeof(struct btrfs_item);
+		}
+
+		ASSERT(leaf_free_space >= 0);
+
+		if (leaf_free_space < ins_len) {
+			int err;
+
+			err = split_leaf(trans, root, key, path, ins_len,
+					 (ret == 0));
+			ASSERT(err <= 0);
+			if (WARN_ON(err > 0))
+				err = -EUCLEAN;
+			if (err)
+				ret = err;
+		}
+	}
+
+	return ret;
+}
 
 /*
  * btrfs_search_slot - look for a key in a tree and perform necessary
@@ -2680,8 +1894,14 @@ out:
  * @p:		Holds all btree nodes along the search path
  * @root:	The root node of the tree
  * @key:	The key we are looking for
- * @ins_len:	Indicates purpose of search, for inserts it is 1, for
- *		deletions it's -1. 0 for plain searches
+ * @ins_len:	Indicates purpose of search:
+ *              >0  for inserts it's size of item inserted (*)
+ *              <0  for deletions
+ *               0  for plain searches, not modifying the tree
+ *
+ *              (*) If size of item inserted doesn't include
+ *              sizeof(struct btrfs_item), then p->search_for_extension must
+ *              be set.
  * @cow:	boolean should CoW operations be performed. Must always be 1
  *		when modifying the tree.
  *
@@ -2701,6 +1921,7 @@ int btrfs_search_slot(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 		      const struct btrfs_key *key, struct btrfs_path *p,
 		      int ins_len, int cow)
 {
+	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct extent_buffer *b;
 	int slot;
 	int ret;
@@ -2718,6 +1939,13 @@ int btrfs_search_slot(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 	WARN_ON(p->nodes[0] != NULL);
 	BUG_ON(!cow && ins_len);
 
+	/*
+	 * For now only allow nowait for read only operations.  There's no
+	 * strict reason why we can't, we just only need it for reads so it's
+	 * only implemented for reads.
+	 */
+	ASSERT(!p->nowait || !cow);
+
 	if (ins_len < 0) {
 		lowest_unlock = 2;
 
@@ -2742,6 +1970,16 @@ int btrfs_search_slot(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 
 	min_write_lock_level = write_lock_level;
 
+	if (p->need_commit_sem) {
+		ASSERT(p->search_commit_root);
+		if (p->nowait) {
+			if (!down_read_trylock(&fs_info->commit_root_sem))
+				return -EAGAIN;
+		} else {
+			down_read(&fs_info->commit_root_sem);
+		}
+	}
+
 again:
 	prev_cmp = -1;
 	b = btrfs_search_slot_get_root(root, p, write_lock_level);
@@ -2763,10 +2001,8 @@ again:
 			 * then we don't want to set the path blocking,
 			 * so we test it here
 			 */
-			if (!should_cow_block(trans, root, b)) {
-				trans->dirty = true;
+			if (!should_cow_block(trans, root, b))
 				goto cow_done;
-			}
 
 			/*
 			 * must have write locks on this node and the
@@ -2781,14 +2017,15 @@ again:
 				goto again;
 			}
 
-			btrfs_set_path_blocking(p);
 			if (last_level)
 				err = btrfs_cow_block(trans, root, b, NULL, 0,
-						      &b);
+						      &b,
+						      BTRFS_NESTING_COW);
 			else
 				err = btrfs_cow_block(trans, root, b,
 						      p->nodes[level + 1],
-						      p->slots[level + 1], &b);
+						      p->slots[level + 1], &b,
+						      BTRFS_NESTING_COW);
 			if (err) {
 				ret = err;
 				goto done;
@@ -2796,10 +2033,6 @@ again:
 		}
 cow_done:
 		p->nodes[level] = b;
-		/*
-		 * Leave path with blocking locks to avoid massive
-		 * lock context switch, this is made on purpose.
-		 */
 
 		/*
 		 * we have a lock on b and as long as we aren't changing
@@ -2821,35 +2054,22 @@ cow_done:
 			}
 		}
 
-		ret = key_search(b, key, level, &prev_cmp, &slot);
-		if (ret < 0)
-			goto done;
-
 		if (level == 0) {
-			p->slots[level] = slot;
-			if (ins_len > 0 &&
-			    btrfs_leaf_free_space(b) < ins_len) {
-				if (write_lock_level < 1) {
-					write_lock_level = 1;
-					btrfs_release_path(p);
-					goto again;
-				}
+			if (ins_len > 0)
+				ASSERT(write_lock_level >= 1);
 
-				btrfs_set_path_blocking(p);
-				err = split_leaf(trans, root, key,
-						 p, ins_len, ret == 0);
-
-				BUG_ON(err > 0);
-				if (err) {
-					ret = err;
-					goto done;
-				}
-			}
+			ret = search_leaf(trans, root, key, p, ins_len, prev_cmp);
 			if (!p->search_for_split)
 				unlock_up(p, level, lowest_unlock,
 					  min_write_lock_level, NULL);
 			goto done;
 		}
+
+		ret = search_for_key_slot(b, 0, key, prev_cmp, &slot);
+		if (ret < 0)
+			goto done;
+		prev_cmp = ret;
+
 		if (ret && slot > 0) {
 			dec = 1;
 			slot--;
@@ -2896,15 +2116,20 @@ cow_done:
 
 		if (!p->skip_locking) {
 			level = btrfs_header_level(b);
+
+			btrfs_maybe_reset_lockdep_class(root, b);
+
 			if (level <= write_lock_level) {
-				if (!btrfs_try_tree_write_lock(b)) {
-					btrfs_set_path_blocking(p);
-					btrfs_tree_lock(b);
-				}
+				btrfs_tree_lock(b);
 				p->locks[level] = BTRFS_WRITE_LOCK;
 			} else {
-				if (!btrfs_tree_read_lock_atomic(b)) {
-					btrfs_set_path_blocking(p);
+				if (p->nowait) {
+					if (!btrfs_try_tree_read_lock(b)) {
+						free_extent_buffer(b);
+						ret = -EAGAIN;
+						goto done;
+					}
+				} else {
 					btrfs_tree_read_lock(b);
 				}
 				p->locks[level] = BTRFS_READ_LOCK;
@@ -2914,16 +2139,21 @@ cow_done:
 	}
 	ret = 1;
 done:
-	/*
-	 * we don't really know what they plan on doing with the path
-	 * from here on, so for now just mark it as blocking
-	 */
-	if (!p->leave_spinning)
-		btrfs_set_path_blocking(p);
 	if (ret < 0 && !p->skip_release_on_error)
 		btrfs_release_path(p);
+
+	if (p->need_commit_sem) {
+		int ret2;
+
+		ret2 = finish_need_commit_sem_search(p);
+		up_read(&fs_info->commit_root_sem);
+		if (ret2)
+			ret = ret2;
+	}
+
 	return ret;
 }
+ALLOW_ERROR_INJECTION(btrfs_search_slot, ERRNO);
 
 /*
  * Like btrfs_search_slot, this looks for a key in the given tree. It uses the
@@ -2947,10 +2177,10 @@ int btrfs_search_old_slot(struct btrfs_root *root, const struct btrfs_key *key,
 	int level;
 	int lowest_unlock = 1;
 	u8 lowest_level = 0;
-	int prev_cmp = -1;
 
 	lowest_level = p->lowest_level;
 	WARN_ON(p->nodes[0] != NULL);
+	ASSERT(!p->nowait);
 
 	if (p->search_commit_root) {
 		BUG_ON(time_seq);
@@ -2958,7 +2188,7 @@ int btrfs_search_old_slot(struct btrfs_root *root, const struct btrfs_key *key,
 	}
 
 again:
-	b = get_old_root(root, time_seq);
+	b = btrfs_get_old_root(root, time_seq);
 	if (!b) {
 		ret = -EIO;
 		goto done;
@@ -2980,12 +2210,7 @@ again:
 		 */
 		btrfs_unlock_up_safe(p, level + 1);
 
-		/*
-		 * Since we can unwind ebs we want to do a real search every
-		 * time.
-		 */
-		prev_cmp = -1;
-		ret = key_search(b, key, level, &prev_cmp, &slot);
+		ret = btrfs_bin_search(b, key, &slot);
 		if (ret < 0)
 			goto done;
 
@@ -3017,11 +2242,8 @@ again:
 		}
 
 		level = btrfs_header_level(b);
-		if (!btrfs_tree_read_lock_atomic(b)) {
-			btrfs_set_path_blocking(p);
-			btrfs_tree_read_lock(b);
-		}
-		b = tree_mod_log_rewind(fs_info, p, b, time_seq);
+		btrfs_tree_read_lock(b);
+		b = btrfs_tree_mod_log_rewind(fs_info, p, b, time_seq);
 		if (!b) {
 			ret = -ENOMEM;
 			goto done;
@@ -3031,8 +2253,6 @@ again:
 	}
 	ret = 1;
 done:
-	if (!p->leave_spinning)
-		btrfs_set_path_blocking(p);
 	if (ret < 0)
 		btrfs_release_path(p);
 
@@ -3117,6 +2337,64 @@ again:
 }
 
 /*
+ * Execute search and call btrfs_previous_item to traverse backwards if the item
+ * was not found.
+ *
+ * Return 0 if found, 1 if not found and < 0 if error.
+ */
+int btrfs_search_backwards(struct btrfs_root *root, struct btrfs_key *key,
+			   struct btrfs_path *path)
+{
+	int ret;
+
+	ret = btrfs_search_slot(NULL, root, key, path, 0, 0);
+	if (ret > 0)
+		ret = btrfs_previous_item(root, path, key->objectid, key->type);
+
+	if (ret == 0)
+		btrfs_item_key_to_cpu(path->nodes[0], key, path->slots[0]);
+
+	return ret;
+}
+
+/**
+ * Search for a valid slot for the given path.
+ *
+ * @root:	The root node of the tree.
+ * @key:	Will contain a valid item if found.
+ * @path:	The starting point to validate the slot.
+ *
+ * Return: 0  if the item is valid
+ *         1  if not found
+ *         <0 if error.
+ */
+int btrfs_get_next_valid_item(struct btrfs_root *root, struct btrfs_key *key,
+			      struct btrfs_path *path)
+{
+	while (1) {
+		int ret;
+		const int slot = path->slots[0];
+		const struct extent_buffer *leaf = path->nodes[0];
+
+		/* This is where we start walking the path. */
+		if (slot >= btrfs_header_nritems(leaf)) {
+			/*
+			 * If we've reached the last slot in this leaf we need
+			 * to go to the next leaf and reset the path.
+			 */
+			ret = btrfs_next_leaf(root, path);
+			if (ret)
+				return ret;
+			continue;
+		}
+		/* Store the found, valid item in @key. */
+		btrfs_item_key_to_cpu(leaf, key, slot);
+		break;
+	}
+	return 0;
+}
+
+/*
  * adjust the pointers going up the tree, starting at level
  * making sure the right key of each node is points to 'key'.
  * This is used after shifting pointers to the left, so it stops
@@ -3137,8 +2415,8 @@ static void fixup_low_keys(struct btrfs_path *path,
 		if (!path->nodes[i])
 			break;
 		t = path->nodes[i];
-		ret = tree_mod_log_insert_key(t, tslot, MOD_LOG_KEY_REPLACE,
-				GFP_ATOMIC);
+		ret = btrfs_tree_mod_log_insert_key(t, tslot,
+				BTRFS_MOD_LOG_KEY_REPLACE, GFP_ATOMIC);
 		BUG_ON(ret < 0);
 		btrfs_set_node_key(t, key, tslot);
 		btrfs_mark_buffer_dirty(path->nodes[i]);
@@ -3200,6 +2478,58 @@ void btrfs_set_item_key_safe(struct btrfs_fs_info *fs_info,
 }
 
 /*
+ * Check key order of two sibling extent buffers.
+ *
+ * Return true if something is wrong.
+ * Return false if everything is fine.
+ *
+ * Tree-checker only works inside one tree block, thus the following
+ * corruption can not be detected by tree-checker:
+ *
+ * Leaf @left			| Leaf @right
+ * --------------------------------------------------------------
+ * | 1 | 2 | 3 | 4 | 5 | f6 |   | 7 | 8 |
+ *
+ * Key f6 in leaf @left itself is valid, but not valid when the next
+ * key in leaf @right is 7.
+ * This can only be checked at tree block merge time.
+ * And since tree checker has ensured all key order in each tree block
+ * is correct, we only need to bother the last key of @left and the first
+ * key of @right.
+ */
+static bool check_sibling_keys(struct extent_buffer *left,
+			       struct extent_buffer *right)
+{
+	struct btrfs_key left_last;
+	struct btrfs_key right_first;
+	int level = btrfs_header_level(left);
+	int nr_left = btrfs_header_nritems(left);
+	int nr_right = btrfs_header_nritems(right);
+
+	/* No key to check in one of the tree blocks */
+	if (!nr_left || !nr_right)
+		return false;
+
+	if (level) {
+		btrfs_node_key_to_cpu(left, &left_last, nr_left - 1);
+		btrfs_node_key_to_cpu(right, &right_first, 0);
+	} else {
+		btrfs_item_key_to_cpu(left, &left_last, nr_left - 1);
+		btrfs_item_key_to_cpu(right, &right_first, 0);
+	}
+
+	if (btrfs_comp_cpu_keys(&left_last, &right_first) >= 0) {
+		btrfs_crit(left->fs_info,
+"bad key order, sibling blocks, left last (%llu %u %llu) right first (%llu %u %llu)",
+			   left_last.objectid, left_last.type,
+			   left_last.offset, right_first.objectid,
+			   right_first.type, right_first.offset);
+		return true;
+	}
+	return false;
+}
+
+/*
  * try to push data from one node into the next node left in the
  * tree.
  *
@@ -3243,7 +2573,13 @@ static int push_node_left(struct btrfs_trans_handle *trans,
 	} else
 		push_items = min(src_nritems - 8, push_items);
 
-	ret = tree_mod_log_eb_copy(dst, src, dst_nritems, 0, push_items);
+	/* dst is the left eb, src is the middle eb */
+	if (check_sibling_keys(dst, src)) {
+		ret = -EUCLEAN;
+		btrfs_abort_transaction(trans, ret);
+		return ret;
+	}
+	ret = btrfs_tree_mod_log_eb_copy(dst, src, dst_nritems, 0, push_items);
 	if (ret) {
 		btrfs_abort_transaction(trans, ret);
 		return ret;
@@ -3255,8 +2591,8 @@ static int push_node_left(struct btrfs_trans_handle *trans,
 
 	if (push_items < src_nritems) {
 		/*
-		 * Don't call tree_mod_log_insert_move here, key removal was
-		 * already fully logged by tree_mod_log_eb_copy above.
+		 * Don't call btrfs_tree_mod_log_insert_move() here, key removal
+		 * was already fully logged by btrfs_tree_mod_log_eb_copy() above.
 		 */
 		memmove_extent_buffer(src, btrfs_node_key_ptr_offset(0),
 				      btrfs_node_key_ptr_offset(push_items),
@@ -3311,15 +2647,21 @@ static int balance_node_right(struct btrfs_trans_handle *trans,
 	if (max_push < push_items)
 		push_items = max_push;
 
-	ret = tree_mod_log_insert_move(dst, push_items, 0, dst_nritems);
+	/* dst is the right eb, src is the middle eb */
+	if (check_sibling_keys(src, dst)) {
+		ret = -EUCLEAN;
+		btrfs_abort_transaction(trans, ret);
+		return ret;
+	}
+	ret = btrfs_tree_mod_log_insert_move(dst, push_items, 0, dst_nritems);
 	BUG_ON(ret < 0);
 	memmove_extent_buffer(dst, btrfs_node_key_ptr_offset(push_items),
 				      btrfs_node_key_ptr_offset(0),
 				      (dst_nritems) *
 				      sizeof(struct btrfs_key_ptr));
 
-	ret = tree_mod_log_eb_copy(dst, src, 0, src_nritems - push_items,
-				   push_items);
+	ret = btrfs_tree_mod_log_eb_copy(dst, src, 0, src_nritems - push_items,
+					 push_items);
 	if (ret) {
 		btrfs_abort_transaction(trans, ret);
 		return ret;
@@ -3366,8 +2708,9 @@ static noinline int insert_new_root(struct btrfs_trans_handle *trans,
 	else
 		btrfs_node_key(lower, &lower_key, 0);
 
-	c = alloc_tree_block_no_bg_flush(trans, root, 0, &lower_key, level,
-					 root->node->start, 0);
+	c = btrfs_alloc_tree_block(trans, root, 0, root->root_key.objectid,
+				   &lower_key, level, root->node->start, 0,
+				   BTRFS_NESTING_NEW_ROOT);
 	if (IS_ERR(c))
 		return PTR_ERR(c);
 
@@ -3384,7 +2727,7 @@ static noinline int insert_new_root(struct btrfs_trans_handle *trans,
 	btrfs_mark_buffer_dirty(c);
 
 	old = root->node;
-	ret = tree_mod_log_insert_root(root->node, c, 0);
+	ret = btrfs_tree_mod_log_insert_root(root->node, c, false);
 	BUG_ON(ret < 0);
 	rcu_assign_pointer(root->node, c);
 
@@ -3394,7 +2737,7 @@ static noinline int insert_new_root(struct btrfs_trans_handle *trans,
 	add_root_to_dirty_list(root);
 	atomic_inc(&c->refs);
 	path->nodes[level] = c;
-	path->locks[level] = BTRFS_WRITE_LOCK_BLOCKING;
+	path->locks[level] = BTRFS_WRITE_LOCK;
 	path->slots[level] = 0;
 	return 0;
 }
@@ -3416,15 +2759,15 @@ static void insert_ptr(struct btrfs_trans_handle *trans,
 	int ret;
 
 	BUG_ON(!path->nodes[level]);
-	btrfs_assert_tree_locked(path->nodes[level]);
+	btrfs_assert_tree_write_locked(path->nodes[level]);
 	lower = path->nodes[level];
 	nritems = btrfs_header_nritems(lower);
 	BUG_ON(slot > nritems);
 	BUG_ON(nritems == BTRFS_NODEPTRS_PER_BLOCK(trans->fs_info));
 	if (slot != nritems) {
 		if (level) {
-			ret = tree_mod_log_insert_move(lower, slot + 1, slot,
-					nritems - slot);
+			ret = btrfs_tree_mod_log_insert_move(lower, slot + 1,
+					slot, nritems - slot);
 			BUG_ON(ret < 0);
 		}
 		memmove_extent_buffer(lower,
@@ -3433,8 +2776,8 @@ static void insert_ptr(struct btrfs_trans_handle *trans,
 			      (nritems - slot) * sizeof(struct btrfs_key_ptr));
 	}
 	if (level) {
-		ret = tree_mod_log_insert_key(lower, slot, MOD_LOG_KEY_ADD,
-				GFP_NOFS);
+		ret = btrfs_tree_mod_log_insert_key(lower, slot,
+					    BTRFS_MOD_LOG_KEY_ADD, GFP_NOFS);
 		BUG_ON(ret < 0);
 	}
 	btrfs_set_node_key(lower, key, slot);
@@ -3475,9 +2818,9 @@ static noinline int split_node(struct btrfs_trans_handle *trans,
 		 * tree mod log: We don't log_removal old root in
 		 * insert_new_root, because that root buffer will be kept as a
 		 * normal node. We are going to log removal of half of the
-		 * elements below with tree_mod_log_eb_copy. We're holding a
-		 * tree lock on the buffer, which is why we cannot race with
-		 * other tree_mod_log users.
+		 * elements below with btrfs_tree_mod_log_eb_copy(). We're
+		 * holding a tree lock on the buffer, which is why we cannot
+		 * race with other tree_mod_log users.
 		 */
 		ret = insert_new_root(trans, root, path, level + 1);
 		if (ret)
@@ -3496,15 +2839,16 @@ static noinline int split_node(struct btrfs_trans_handle *trans,
 	mid = (c_nritems + 1) / 2;
 	btrfs_node_key(c, &disk_key, mid);
 
-	split = alloc_tree_block_no_bg_flush(trans, root, 0, &disk_key, level,
-					     c->start, 0);
+	split = btrfs_alloc_tree_block(trans, root, 0, root->root_key.objectid,
+				       &disk_key, level, c->start, 0,
+				       BTRFS_NESTING_SPLIT);
 	if (IS_ERR(split))
 		return PTR_ERR(split);
 
 	root_add_used(root, fs_info->nodesize);
 	ASSERT(btrfs_header_level(c) == level);
 
-	ret = tree_mod_log_eb_copy(split, c, 0, mid, c_nritems - mid);
+	ret = btrfs_tree_mod_log_eb_copy(split, c, 0, mid, c_nritems - mid);
 	if (ret) {
 		btrfs_abort_transaction(trans, ret);
 		return ret;
@@ -3515,7 +2859,6 @@ static noinline int split_node(struct btrfs_trans_handle *trans,
 			   (c_nritems - mid) * sizeof(struct btrfs_key_ptr));
 	btrfs_set_header_nritems(split, c_nritems - mid);
 	btrfs_set_header_nritems(c, mid);
-	ret = 0;
 
 	btrfs_mark_buffer_dirty(c);
 	btrfs_mark_buffer_dirty(split);
@@ -3533,7 +2876,7 @@ static noinline int split_node(struct btrfs_trans_handle *trans,
 		btrfs_tree_unlock(split);
 		free_extent_buffer(split);
 	}
-	return ret;
+	return 0;
 }
 
 /*
@@ -3543,21 +2886,14 @@ static noinline int split_node(struct btrfs_trans_handle *trans,
  */
 static int leaf_space_used(struct extent_buffer *l, int start, int nr)
 {
-	struct btrfs_item *start_item;
-	struct btrfs_item *end_item;
-	struct btrfs_map_token token;
 	int data_len;
 	int nritems = btrfs_header_nritems(l);
 	int end = min(nritems, start + nr) - 1;
 
 	if (!nr)
 		return 0;
-	btrfs_init_map_token(&token, l);
-	start_item = btrfs_item_nr(start);
-	end_item = btrfs_item_nr(end);
-	data_len = btrfs_token_item_offset(l, start_item, &token) +
-		btrfs_token_item_size(l, start_item, &token);
-	data_len = data_len - btrfs_token_item_offset(l, end_item, &token);
+	data_len = btrfs_item_offset(l, start) + btrfs_item_size(l, start);
+	data_len = data_len - btrfs_item_offset(l, end);
 	data_len += sizeof(struct btrfs_item) * nr;
 	WARN_ON(data_len < 0);
 	return data_len;
@@ -3604,7 +2940,6 @@ static noinline int __push_leaf_right(struct btrfs_path *path,
 	u32 i;
 	int push_space = 0;
 	int push_items = 0;
-	struct btrfs_item *item;
 	u32 nr;
 	u32 right_nritems;
 	u32 data_end;
@@ -3621,8 +2956,6 @@ static noinline int __push_leaf_right(struct btrfs_path *path,
 	slot = path->slots[1];
 	i = left_nritems - 1;
 	while (i >= nr) {
-		item = btrfs_item_nr(i);
-
 		if (!empty && push_items > 0) {
 			if (path->slots[0] > i)
 				break;
@@ -3637,12 +2970,13 @@ static noinline int __push_leaf_right(struct btrfs_path *path,
 		if (path->slots[0] == i)
 			push_space += data_size;
 
-		this_item_size = btrfs_item_size(left, item);
-		if (this_item_size + sizeof(*item) + push_space > free_space)
+		this_item_size = btrfs_item_size(left, i);
+		if (this_item_size + sizeof(struct btrfs_item) +
+		    push_space > free_space)
 			break;
 
 		push_items++;
-		push_space += this_item_size + sizeof(*item);
+		push_space += this_item_size + sizeof(struct btrfs_item);
 		if (i == 0)
 			break;
 		i--;
@@ -3656,7 +2990,7 @@ static noinline int __push_leaf_right(struct btrfs_path *path,
 	/* push left to right */
 	right_nritems = btrfs_header_nritems(right);
 
-	push_space = btrfs_item_end_nr(left, left_nritems - push_items);
+	push_space = btrfs_item_data_end(left, left_nritems - push_items);
 	push_space -= leaf_data_end(left);
 
 	/* make room in the right data area */
@@ -3687,9 +3021,8 @@ static noinline int __push_leaf_right(struct btrfs_path *path,
 	btrfs_set_header_nritems(right, right_nritems);
 	push_space = BTRFS_LEAF_DATA_SIZE(fs_info);
 	for (i = 0; i < right_nritems; i++) {
-		item = btrfs_item_nr(i);
-		push_space -= btrfs_token_item_size(right, item, &token);
-		btrfs_set_token_item_offset(right, item, push_space, &token);
+		push_space -= btrfs_token_item_size(&token, i);
+		btrfs_set_token_item_offset(&token, i, push_space);
 	}
 
 	left_nritems -= push_items;
@@ -3758,7 +3091,7 @@ static int push_leaf_right(struct btrfs_trans_handle *trans, struct btrfs_root
 	if (slot >= btrfs_header_nritems(upper) - 1)
 		return 1;
 
-	btrfs_assert_tree_locked(path->nodes[1]);
+	btrfs_assert_tree_write_locked(path->nodes[1]);
 
 	right = btrfs_read_node_slot(upper, slot + 1);
 	/*
@@ -3768,27 +3101,27 @@ static int push_leaf_right(struct btrfs_trans_handle *trans, struct btrfs_root
 	if (IS_ERR(right))
 		return 1;
 
-	btrfs_tree_lock(right);
-	btrfs_set_lock_blocking_write(right);
+	__btrfs_tree_lock(right, BTRFS_NESTING_RIGHT);
 
 	free_space = btrfs_leaf_free_space(right);
 	if (free_space < data_size)
 		goto out_unlock;
 
-	/* cow and double check */
 	ret = btrfs_cow_block(trans, root, right, upper,
-			      slot + 1, &right);
+			      slot + 1, &right, BTRFS_NESTING_RIGHT_COW);
 	if (ret)
 		goto out_unlock;
 
-	free_space = btrfs_leaf_free_space(right);
-	if (free_space < data_size)
-		goto out_unlock;
-
 	left_nritems = btrfs_header_nritems(left);
 	if (left_nritems == 0)
 		goto out_unlock;
 
+	if (check_sibling_keys(left, right)) {
+		ret = -EUCLEAN;
+		btrfs_tree_unlock(right);
+		free_extent_buffer(right);
+		return ret;
+	}
 	if (path->slots[0] == left_nritems && !empty) {
 		/* Key greater than all keys in the leaf, right neighbor has
 		 * enough room for it and we're not emptying our leaf to delete
@@ -3829,7 +3162,6 @@ static noinline int __push_leaf_left(struct btrfs_path *path, int data_size,
 	int i;
 	int push_space = 0;
 	int push_items = 0;
-	struct btrfs_item *item;
 	u32 old_left_nritems;
 	u32 nr;
 	int ret = 0;
@@ -3843,8 +3175,6 @@ static noinline int __push_leaf_left(struct btrfs_path *path, int data_size,
 		nr = min(right_nritems - 1, max_slot);
 
 	for (i = 0; i < nr; i++) {
-		item = btrfs_item_nr(i);
-
 		if (!empty && push_items > 0) {
 			if (path->slots[0] < i)
 				break;
@@ -3859,12 +3189,13 @@ static noinline int __push_leaf_left(struct btrfs_path *path, int data_size,
 		if (path->slots[0] == i)
 			push_space += data_size;
 
-		this_item_size = btrfs_item_size(right, item);
-		if (this_item_size + sizeof(*item) + push_space > free_space)
+		this_item_size = btrfs_item_size(right, i);
+		if (this_item_size + sizeof(struct btrfs_item) + push_space >
+		    free_space)
 			break;
 
 		push_items++;
-		push_space += this_item_size + sizeof(*item);
+		push_space += this_item_size + sizeof(struct btrfs_item);
 	}
 
 	if (push_items == 0) {
@@ -3880,27 +3211,24 @@ static noinline int __push_leaf_left(struct btrfs_path *path, int data_size,
 			   push_items * sizeof(struct btrfs_item));
 
 	push_space = BTRFS_LEAF_DATA_SIZE(fs_info) -
-		     btrfs_item_offset_nr(right, push_items - 1);
+		     btrfs_item_offset(right, push_items - 1);
 
 	copy_extent_buffer(left, right, BTRFS_LEAF_DATA_OFFSET +
 		     leaf_data_end(left) - push_space,
 		     BTRFS_LEAF_DATA_OFFSET +
-		     btrfs_item_offset_nr(right, push_items - 1),
+		     btrfs_item_offset(right, push_items - 1),
 		     push_space);
 	old_left_nritems = btrfs_header_nritems(left);
 	BUG_ON(old_left_nritems <= 0);
 
 	btrfs_init_map_token(&token, left);
-	old_left_item_size = btrfs_item_offset_nr(left, old_left_nritems - 1);
+	old_left_item_size = btrfs_item_offset(left, old_left_nritems - 1);
 	for (i = old_left_nritems; i < old_left_nritems + push_items; i++) {
 		u32 ioff;
 
-		item = btrfs_item_nr(i);
-
-		ioff = btrfs_token_item_offset(left, item, &token);
-		btrfs_set_token_item_offset(left, item,
-		      ioff - (BTRFS_LEAF_DATA_SIZE(fs_info) - old_left_item_size),
-		      &token);
+		ioff = btrfs_token_item_offset(&token, i);
+		btrfs_set_token_item_offset(&token, i,
+		      ioff - (BTRFS_LEAF_DATA_SIZE(fs_info) - old_left_item_size));
 	}
 	btrfs_set_header_nritems(left, old_left_nritems + push_items);
 
@@ -3910,7 +3238,7 @@ static noinline int __push_leaf_left(struct btrfs_path *path, int data_size,
 		       right_nritems);
 
 	if (push_items < right_nritems) {
-		push_space = btrfs_item_offset_nr(right, push_items - 1) -
+		push_space = btrfs_item_offset(right, push_items - 1) -
 						  leaf_data_end(right);
 		memmove_extent_buffer(right, BTRFS_LEAF_DATA_OFFSET +
 				      BTRFS_LEAF_DATA_SIZE(fs_info) - push_space,
@@ -3928,11 +3256,8 @@ static noinline int __push_leaf_left(struct btrfs_path *path, int data_size,
 	btrfs_set_header_nritems(right, right_nritems);
 	push_space = BTRFS_LEAF_DATA_SIZE(fs_info);
 	for (i = 0; i < right_nritems; i++) {
-		item = btrfs_item_nr(i);
-
-		push_space = push_space - btrfs_token_item_size(right,
-								item, &token);
-		btrfs_set_token_item_offset(right, item, push_space, &token);
+		push_space = push_space - btrfs_token_item_size(&token, i);
+		btrfs_set_token_item_offset(&token, i, push_space);
 	}
 
 	btrfs_mark_buffer_dirty(left);
@@ -3993,7 +3318,7 @@ static int push_leaf_left(struct btrfs_trans_handle *trans, struct btrfs_root
 	if (right_nritems == 0)
 		return 1;
 
-	btrfs_assert_tree_locked(path->nodes[1]);
+	btrfs_assert_tree_write_locked(path->nodes[1]);
 
 	left = btrfs_read_node_slot(path->nodes[1], slot - 1);
 	/*
@@ -4003,8 +3328,7 @@ static int push_leaf_left(struct btrfs_trans_handle *trans, struct btrfs_root
 	if (IS_ERR(left))
 		return 1;
 
-	btrfs_tree_lock(left);
-	btrfs_set_lock_blocking_write(left);
+	__btrfs_tree_lock(left, BTRFS_NESTING_LEFT);
 
 	free_space = btrfs_leaf_free_space(left);
 	if (free_space < data_size) {
@@ -4012,9 +3336,9 @@ static int push_leaf_left(struct btrfs_trans_handle *trans, struct btrfs_root
 		goto out;
 	}
 
-	/* cow and double check */
 	ret = btrfs_cow_block(trans, root, left,
-			      path->nodes[1], slot - 1, &left);
+			      path->nodes[1], slot - 1, &left,
+			      BTRFS_NESTING_LEFT_COW);
 	if (ret) {
 		/* we hit -ENOSPC, but it isn't fatal here */
 		if (ret == -ENOSPC)
@@ -4022,12 +3346,10 @@ static int push_leaf_left(struct btrfs_trans_handle *trans, struct btrfs_root
 		goto out;
 	}
 
-	free_space = btrfs_leaf_free_space(left);
-	if (free_space < data_size) {
-		ret = 1;
+	if (check_sibling_keys(left, right)) {
+		ret = -EUCLEAN;
 		goto out;
 	}
-
 	return __push_leaf_left(path, min_data_size,
 			       empty, left, free_space, right_nritems,
 			       max_slot);
@@ -4056,7 +3378,7 @@ static noinline void copy_for_split(struct btrfs_trans_handle *trans,
 
 	nritems = nritems - mid;
 	btrfs_set_header_nritems(right, nritems);
-	data_copy_size = btrfs_item_end_nr(l, mid) - leaf_data_end(l);
+	data_copy_size = btrfs_item_data_end(l, mid) - leaf_data_end(l);
 
 	copy_extent_buffer(right, l, btrfs_item_nr_offset(0),
 			   btrfs_item_nr_offset(mid),
@@ -4067,16 +3389,14 @@ static noinline void copy_for_split(struct btrfs_trans_handle *trans,
 		     data_copy_size, BTRFS_LEAF_DATA_OFFSET +
 		     leaf_data_end(l), data_copy_size);
 
-	rt_data_off = BTRFS_LEAF_DATA_SIZE(fs_info) - btrfs_item_end_nr(l, mid);
+	rt_data_off = BTRFS_LEAF_DATA_SIZE(fs_info) - btrfs_item_data_end(l, mid);
 
 	btrfs_init_map_token(&token, right);
 	for (i = 0; i < nritems; i++) {
-		struct btrfs_item *item = btrfs_item_nr(i);
 		u32 ioff;
 
-		ioff = btrfs_token_item_offset(right, item, &token);
-		btrfs_set_token_item_offset(right, item,
-					    ioff + rt_data_off, &token);
+		ioff = btrfs_token_item_offset(&token, i);
+		btrfs_set_token_item_offset(&token, i, ioff + rt_data_off);
 	}
 
 	btrfs_set_header_nritems(l, mid);
@@ -4192,7 +3512,7 @@ static noinline int split_leaf(struct btrfs_trans_handle *trans,
 
 	l = path->nodes[0];
 	slot = path->slots[0];
-	if (extend && data_size + btrfs_item_size_nr(l, slot) +
+	if (extend && data_size + btrfs_item_size(l, slot) +
 	    sizeof(struct btrfs_item) > BTRFS_LEAF_DATA_SIZE(fs_info))
 		return -EOVERFLOW;
 
@@ -4277,8 +3597,18 @@ again:
 	else
 		btrfs_item_key(l, &disk_key, mid);
 
-	right = alloc_tree_block_no_bg_flush(trans, root, 0, &disk_key, 0,
-					     l->start, 0);
+	/*
+	 * We have to about BTRFS_NESTING_NEW_ROOT here if we've done a double
+	 * split, because we're only allowed to have MAX_LOCKDEP_SUBCLASSES
+	 * subclasses, which is 8 at the time of this patch, and we've maxed it
+	 * out.  In the future we could add a
+	 * BTRFS_NESTING_SPLIT_THE_SPLITTENING if we need to, but for now just
+	 * use BTRFS_NESTING_NEW_ROOT.
+	 */
+	right = btrfs_alloc_tree_block(trans, root, 0, root->root_key.objectid,
+				       &disk_key, 0, l->start, 0,
+				       num_doubles ? BTRFS_NESTING_NEW_ROOT :
+				       BTRFS_NESTING_SPLIT);
 	if (IS_ERR(right))
 		return PTR_ERR(right);
 
@@ -4351,7 +3681,7 @@ static noinline int setup_leaf_for_split(struct btrfs_trans_handle *trans,
 	if (btrfs_leaf_free_space(leaf) >= ins_len)
 		return 0;
 
-	item_size = btrfs_item_size_nr(leaf, path->slots[0]);
+	item_size = btrfs_item_size(leaf, path->slots[0]);
 	if (key.type == BTRFS_EXTENT_DATA_KEY) {
 		fi = btrfs_item_ptr(leaf, path->slots[0],
 				    struct btrfs_file_extent_item);
@@ -4371,7 +3701,7 @@ static noinline int setup_leaf_for_split(struct btrfs_trans_handle *trans,
 	ret = -EAGAIN;
 	leaf = path->nodes[0];
 	/* if our item isn't there, return now */
-	if (item_size != btrfs_item_size_nr(leaf, path->slots[0]))
+	if (item_size != btrfs_item_size(leaf, path->slots[0]))
 		goto err;
 
 	/* the leaf has  changed, it now has room.  return now */
@@ -4385,7 +3715,6 @@ static noinline int setup_leaf_for_split(struct btrfs_trans_handle *trans,
 			goto err;
 	}
 
-	btrfs_set_path_blocking(path);
 	ret = split_leaf(trans, root, &key, path, ins_len, 1);
 	if (ret)
 		goto err;
@@ -4403,9 +3732,7 @@ static noinline int split_item(struct btrfs_path *path,
 			       unsigned long split_offset)
 {
 	struct extent_buffer *leaf;
-	struct btrfs_item *item;
-	struct btrfs_item *new_item;
-	int slot;
+	int orig_slot, slot;
 	char *buf;
 	u32 nritems;
 	u32 item_size;
@@ -4415,11 +3742,9 @@ static noinline int split_item(struct btrfs_path *path,
 	leaf = path->nodes[0];
 	BUG_ON(btrfs_leaf_free_space(leaf) < sizeof(struct btrfs_item));
 
-	btrfs_set_path_blocking(path);
-
-	item = btrfs_item_nr(path->slots[0]);
-	orig_offset = btrfs_item_offset(leaf, item);
-	item_size = btrfs_item_size(leaf, item);
+	orig_slot = path->slots[0];
+	orig_offset = btrfs_item_offset(leaf, path->slots[0]);
+	item_size = btrfs_item_size(leaf, path->slots[0]);
 
 	buf = kmalloc(item_size, GFP_NOFS);
 	if (!buf)
@@ -4440,14 +3765,12 @@ static noinline int split_item(struct btrfs_path *path,
 	btrfs_cpu_key_to_disk(&disk_key, new_key);
 	btrfs_set_item_key(leaf, &disk_key, slot);
 
-	new_item = btrfs_item_nr(slot);
+	btrfs_set_item_offset(leaf, slot, orig_offset);
+	btrfs_set_item_size(leaf, slot, item_size - split_offset);
 
-	btrfs_set_item_offset(leaf, new_item, orig_offset);
-	btrfs_set_item_size(leaf, new_item, item_size - split_offset);
-
-	btrfs_set_item_offset(leaf, item,
-			      orig_offset + item_size - split_offset);
-	btrfs_set_item_size(leaf, item, split_offset);
+	btrfs_set_item_offset(leaf, orig_slot,
+				 orig_offset + item_size - split_offset);
+	btrfs_set_item_size(leaf, orig_slot, split_offset);
 
 	btrfs_set_header_nritems(leaf, nritems + 1);
 
@@ -4499,42 +3822,6 @@ int btrfs_split_item(struct btrfs_trans_handle *trans,
 }
 
 /*
- * This function duplicate a item, giving 'new_key' to the new item.
- * It guarantees both items live in the same tree leaf and the new item
- * is contiguous with the original item.
- *
- * This allows us to split file extent in place, keeping a lock on the
- * leaf the entire time.
- */
-int btrfs_duplicate_item(struct btrfs_trans_handle *trans,
-			 struct btrfs_root *root,
-			 struct btrfs_path *path,
-			 const struct btrfs_key *new_key)
-{
-	struct extent_buffer *leaf;
-	int ret;
-	u32 item_size;
-
-	leaf = path->nodes[0];
-	item_size = btrfs_item_size_nr(leaf, path->slots[0]);
-	ret = setup_leaf_for_split(trans, root, path,
-				   item_size + sizeof(struct btrfs_item));
-	if (ret)
-		return ret;
-
-	path->slots[0]++;
-	setup_items_for_insert(root, path, new_key, &item_size,
-			       item_size, item_size +
-			       sizeof(struct btrfs_item), 1);
-	leaf = path->nodes[0];
-	memcpy_extent_buffer(leaf,
-			     btrfs_item_ptr_offset(leaf, path->slots[0]),
-			     btrfs_item_ptr_offset(leaf, path->slots[0] - 1),
-			     item_size);
-	return 0;
-}
-
-/*
  * make the item pointed to by the path smaller.  new_size indicates
  * how small to make it, and from_end tells us if we just chop bytes
  * off the end of the item or if we shift the item to chop bytes off
@@ -4544,7 +3831,6 @@ void btrfs_truncate_item(struct btrfs_path *path, u32 new_size, int from_end)
 {
 	int slot;
 	struct extent_buffer *leaf;
-	struct btrfs_item *item;
 	u32 nritems;
 	unsigned int data_end;
 	unsigned int old_data_start;
@@ -4556,14 +3842,14 @@ void btrfs_truncate_item(struct btrfs_path *path, u32 new_size, int from_end)
 	leaf = path->nodes[0];
 	slot = path->slots[0];
 
-	old_size = btrfs_item_size_nr(leaf, slot);
+	old_size = btrfs_item_size(leaf, slot);
 	if (old_size == new_size)
 		return;
 
 	nritems = btrfs_header_nritems(leaf);
 	data_end = leaf_data_end(leaf);
 
-	old_data_start = btrfs_item_offset_nr(leaf, slot);
+	old_data_start = btrfs_item_offset(leaf, slot);
 
 	size_diff = old_size - new_size;
 
@@ -4577,11 +3863,9 @@ void btrfs_truncate_item(struct btrfs_path *path, u32 new_size, int from_end)
 	btrfs_init_map_token(&token, leaf);
 	for (i = slot; i < nritems; i++) {
 		u32 ioff;
-		item = btrfs_item_nr(i);
 
-		ioff = btrfs_token_item_offset(leaf, item, &token);
-		btrfs_set_token_item_offset(leaf, item,
-					    ioff + size_diff, &token);
+		ioff = btrfs_token_item_offset(&token, i);
+		btrfs_set_token_item_offset(&token, i, ioff + size_diff);
 	}
 
 	/* shift the data */
@@ -4624,8 +3908,7 @@ void btrfs_truncate_item(struct btrfs_path *path, u32 new_size, int from_end)
 			fixup_low_keys(path, &disk_key, 1);
 	}
 
-	item = btrfs_item_nr(slot);
-	btrfs_set_item_size(leaf, item, new_size);
+	btrfs_set_item_size(leaf, slot, new_size);
 	btrfs_mark_buffer_dirty(leaf);
 
 	if (btrfs_leaf_free_space(leaf) < 0) {
@@ -4641,7 +3924,6 @@ void btrfs_extend_item(struct btrfs_path *path, u32 data_size)
 {
 	int slot;
 	struct extent_buffer *leaf;
-	struct btrfs_item *item;
 	u32 nritems;
 	unsigned int data_end;
 	unsigned int old_data;
@@ -4659,7 +3941,7 @@ void btrfs_extend_item(struct btrfs_path *path, u32 data_size)
 		BUG();
 	}
 	slot = path->slots[0];
-	old_data = btrfs_item_end_nr(leaf, slot);
+	old_data = btrfs_item_data_end(leaf, slot);
 
 	BUG_ON(slot < 0);
 	if (slot >= nritems) {
@@ -4676,11 +3958,9 @@ void btrfs_extend_item(struct btrfs_path *path, u32 data_size)
 	btrfs_init_map_token(&token, leaf);
 	for (i = slot; i < nritems; i++) {
 		u32 ioff;
-		item = btrfs_item_nr(i);
 
-		ioff = btrfs_token_item_offset(leaf, item, &token);
-		btrfs_set_token_item_offset(leaf, item,
-					    ioff - data_size, &token);
+		ioff = btrfs_token_item_offset(&token, i);
+		btrfs_set_token_item_offset(&token, i, ioff - data_size);
 	}
 
 	/* shift the data */
@@ -4689,9 +3969,8 @@ void btrfs_extend_item(struct btrfs_path *path, u32 data_size)
 		      data_end, old_data - data_end);
 
 	data_end = old_data;
-	old_size = btrfs_item_size_nr(leaf, slot);
-	item = btrfs_item_nr(slot);
-	btrfs_set_item_size(leaf, item, old_size + data_size);
+	old_size = btrfs_item_size(leaf, slot);
+	btrfs_set_item_size(leaf, slot, old_size + data_size);
 	btrfs_mark_buffer_dirty(leaf);
 
 	if (btrfs_leaf_free_space(leaf) < 0) {
@@ -4700,17 +3979,19 @@ void btrfs_extend_item(struct btrfs_path *path, u32 data_size)
 	}
 }
 
-/*
- * this is a helper for btrfs_insert_empty_items, the main goal here is
- * to save stack depth by doing the bulk of the work in a function
- * that doesn't call btrfs_search_slot
+/**
+ * setup_items_for_insert - Helper called before inserting one or more items
+ * to a leaf. Main purpose is to save stack depth by doing the bulk of the work
+ * in a function that doesn't call btrfs_search_slot
+ *
+ * @root:	root we are inserting items to
+ * @path:	points to the leaf/slot where we are going to insert new items
+ * @batch:      information about the batch of items to insert
  */
-void setup_items_for_insert(struct btrfs_root *root, struct btrfs_path *path,
-			    const struct btrfs_key *cpu_key, u32 *data_size,
-			    u32 total_data, u32 total_size, int nr)
+static void setup_items_for_insert(struct btrfs_root *root, struct btrfs_path *path,
+				   const struct btrfs_item_batch *batch)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
-	struct btrfs_item *item;
 	int i;
 	u32 nritems;
 	unsigned int data_end;
@@ -4718,9 +3999,15 @@ void setup_items_for_insert(struct btrfs_root *root, struct btrfs_path *path,
 	struct extent_buffer *leaf;
 	int slot;
 	struct btrfs_map_token token;
+	u32 total_size;
 
+	/*
+	 * Before anything else, update keys in the parent and other ancestors
+	 * if needed, then release the write locks on them, so that other tasks
+	 * can use them while we modify the leaf.
+	 */
 	if (path->slots[0] == 0) {
-		btrfs_cpu_key_to_disk(&disk_key, cpu_key);
+		btrfs_cpu_key_to_disk(&disk_key, &batch->keys[0]);
 		fixup_low_keys(path, &disk_key, 1);
 	}
 	btrfs_unlock_up_safe(path, 1);
@@ -4730,6 +4017,7 @@ void setup_items_for_insert(struct btrfs_root *root, struct btrfs_path *path,
 
 	nritems = btrfs_header_nritems(leaf);
 	data_end = leaf_data_end(leaf);
+	total_size = batch->total_data_size + (batch->nr * sizeof(struct btrfs_item));
 
 	if (btrfs_leaf_free_space(leaf) < total_size) {
 		btrfs_print_leaf(leaf);
@@ -4740,11 +4028,12 @@ void setup_items_for_insert(struct btrfs_root *root, struct btrfs_path *path,
 
 	btrfs_init_map_token(&token, leaf);
 	if (slot != nritems) {
-		unsigned int old_data = btrfs_item_end_nr(leaf, slot);
+		unsigned int old_data = btrfs_item_data_end(leaf, slot);
 
 		if (old_data < data_end) {
 			btrfs_print_leaf(leaf);
-			btrfs_crit(fs_info, "slot %d old_data %d data_end %d",
+			btrfs_crit(fs_info,
+		"item at slot %d with data offset %u beyond data end of leaf %u",
 				   slot, old_data, data_end);
 			BUG();
 		}
@@ -4755,35 +4044,33 @@ void setup_items_for_insert(struct btrfs_root *root, struct btrfs_path *path,
 		for (i = slot; i < nritems; i++) {
 			u32 ioff;
 
-			item = btrfs_item_nr(i);
-			ioff = btrfs_token_item_offset(leaf, item, &token);
-			btrfs_set_token_item_offset(leaf, item,
-						    ioff - total_data, &token);
+			ioff = btrfs_token_item_offset(&token, i);
+			btrfs_set_token_item_offset(&token, i,
+						       ioff - batch->total_data_size);
 		}
 		/* shift the items */
-		memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot + nr),
+		memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot + batch->nr),
 			      btrfs_item_nr_offset(slot),
 			      (nritems - slot) * sizeof(struct btrfs_item));
 
 		/* shift the data */
 		memmove_extent_buffer(leaf, BTRFS_LEAF_DATA_OFFSET +
-			      data_end - total_data, BTRFS_LEAF_DATA_OFFSET +
-			      data_end, old_data - data_end);
+				      data_end - batch->total_data_size,
+				      BTRFS_LEAF_DATA_OFFSET + data_end,
+				      old_data - data_end);
 		data_end = old_data;
 	}
 
 	/* setup the item for the new data */
-	for (i = 0; i < nr; i++) {
-		btrfs_cpu_key_to_disk(&disk_key, cpu_key + i);
+	for (i = 0; i < batch->nr; i++) {
+		btrfs_cpu_key_to_disk(&disk_key, &batch->keys[i]);
 		btrfs_set_item_key(leaf, &disk_key, slot + i);
-		item = btrfs_item_nr(slot + i);
-		btrfs_set_token_item_offset(leaf, item,
-					    data_end - data_size[i], &token);
-		data_end -= data_size[i];
-		btrfs_set_token_item_size(leaf, item, data_size[i], &token);
+		data_end -= batch->data_sizes[i];
+		btrfs_set_token_item_offset(&token, slot + i, data_end);
+		btrfs_set_token_item_size(&token, slot + i, batch->data_sizes[i]);
 	}
 
-	btrfs_set_header_nritems(leaf, nritems + nr);
+	btrfs_set_header_nritems(leaf, nritems + batch->nr);
 	btrfs_mark_buffer_dirty(leaf);
 
 	if (btrfs_leaf_free_space(leaf) < 0) {
@@ -4793,26 +4080,43 @@ void setup_items_for_insert(struct btrfs_root *root, struct btrfs_path *path,
 }
 
 /*
+ * Insert a new item into a leaf.
+ *
+ * @root:      The root of the btree.
+ * @path:      A path pointing to the target leaf and slot.
+ * @key:       The key of the new item.
+ * @data_size: The size of the data associated with the new key.
+ */
+void btrfs_setup_item_for_insert(struct btrfs_root *root,
+				 struct btrfs_path *path,
+				 const struct btrfs_key *key,
+				 u32 data_size)
+{
+	struct btrfs_item_batch batch;
+
+	batch.keys = key;
+	batch.data_sizes = &data_size;
+	batch.total_data_size = data_size;
+	batch.nr = 1;
+
+	setup_items_for_insert(root, path, &batch);
+}
+
+/*
  * Given a key and some data, insert items into the tree.
  * This does all the path init required, making room in the tree if needed.
  */
 int btrfs_insert_empty_items(struct btrfs_trans_handle *trans,
 			    struct btrfs_root *root,
 			    struct btrfs_path *path,
-			    const struct btrfs_key *cpu_key, u32 *data_size,
-			    int nr)
+			    const struct btrfs_item_batch *batch)
 {
 	int ret = 0;
 	int slot;
-	int i;
-	u32 total_size = 0;
-	u32 total_data = 0;
-
-	for (i = 0; i < nr; i++)
-		total_data += data_size[i];
+	u32 total_size;
 
-	total_size = total_data + (nr * sizeof(struct btrfs_item));
-	ret = btrfs_search_slot(trans, root, cpu_key, path, total_size, 1);
+	total_size = batch->total_data_size + (batch->nr * sizeof(struct btrfs_item));
+	ret = btrfs_search_slot(trans, root, &batch->keys[0], path, total_size, 1);
 	if (ret == 0)
 		return -EEXIST;
 	if (ret < 0)
@@ -4821,8 +4125,7 @@ int btrfs_insert_empty_items(struct btrfs_trans_handle *trans,
 	slot = path->slots[0];
 	BUG_ON(slot < 0);
 
-	setup_items_for_insert(root, path, cpu_key, data_size,
-			       total_data, total_size, nr);
+	setup_items_for_insert(root, path, batch);
 	return 0;
 }
 
@@ -4854,6 +4157,40 @@ int btrfs_insert_item(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 }
 
 /*
+ * This function duplicates an item, giving 'new_key' to the new item.
+ * It guarantees both items live in the same tree leaf and the new item is
+ * contiguous with the original item.
+ *
+ * This allows us to split a file extent in place, keeping a lock on the leaf
+ * the entire time.
+ */
+int btrfs_duplicate_item(struct btrfs_trans_handle *trans,
+			 struct btrfs_root *root,
+			 struct btrfs_path *path,
+			 const struct btrfs_key *new_key)
+{
+	struct extent_buffer *leaf;
+	int ret;
+	u32 item_size;
+
+	leaf = path->nodes[0];
+	item_size = btrfs_item_size(leaf, path->slots[0]);
+	ret = setup_leaf_for_split(trans, root, path,
+				   item_size + sizeof(struct btrfs_item));
+	if (ret)
+		return ret;
+
+	path->slots[0]++;
+	btrfs_setup_item_for_insert(root, path, new_key, item_size);
+	leaf = path->nodes[0];
+	memcpy_extent_buffer(leaf,
+			     btrfs_item_ptr_offset(leaf, path->slots[0]),
+			     btrfs_item_ptr_offset(leaf, path->slots[0] - 1),
+			     item_size);
+	return 0;
+}
+
+/*
  * delete the pointer from a given node.
  *
  * the tree should have been previously balanced so the deletion does not
@@ -4869,8 +4206,8 @@ static void del_ptr(struct btrfs_root *root, struct btrfs_path *path,
 	nritems = btrfs_header_nritems(parent);
 	if (slot != nritems - 1) {
 		if (level) {
-			ret = tree_mod_log_insert_move(parent, slot, slot + 1,
-					nritems - slot - 1);
+			ret = btrfs_tree_mod_log_insert_move(parent, slot,
+					slot + 1, nritems - slot - 1);
 			BUG_ON(ret < 0);
 		}
 		memmove_extent_buffer(parent,
@@ -4879,8 +4216,8 @@ static void del_ptr(struct btrfs_root *root, struct btrfs_path *path,
 			      sizeof(struct btrfs_key_ptr) *
 			      (nritems - slot - 1));
 	} else if (level) {
-		ret = tree_mod_log_insert_key(parent, slot, MOD_LOG_KEY_REMOVE,
-				GFP_NOFS);
+		ret = btrfs_tree_mod_log_insert_key(parent, slot,
+				BTRFS_MOD_LOG_KEY_REMOVE, GFP_NOFS);
 		BUG_ON(ret < 0);
 	}
 
@@ -4926,7 +4263,7 @@ static noinline void btrfs_del_leaf(struct btrfs_trans_handle *trans,
 	root_sub_used(root, leaf->len);
 
 	atomic_inc(&leaf->refs);
-	btrfs_free_tree_block(trans, root, leaf, 0, 1);
+	btrfs_free_tree_block(trans, btrfs_root_id(root), leaf, 0, 1);
 	free_extent_buffer_stale(leaf);
 }
 /*
@@ -4938,25 +4275,22 @@ int btrfs_del_items(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct extent_buffer *leaf;
-	struct btrfs_item *item;
-	u32 last_off;
-	u32 dsize = 0;
 	int ret = 0;
 	int wret;
-	int i;
 	u32 nritems;
 
 	leaf = path->nodes[0];
-	last_off = btrfs_item_offset_nr(leaf, slot + nr - 1);
-
-	for (i = 0; i < nr; i++)
-		dsize += btrfs_item_size_nr(leaf, slot + i);
-
 	nritems = btrfs_header_nritems(leaf);
 
 	if (slot + nr != nritems) {
-		int data_end = leaf_data_end(leaf);
+		const u32 last_off = btrfs_item_offset(leaf, slot + nr - 1);
+		const int data_end = leaf_data_end(leaf);
 		struct btrfs_map_token token;
+		u32 dsize = 0;
+		int i;
+
+		for (i = 0; i < nr; i++)
+			dsize += btrfs_item_size(leaf, slot + i);
 
 		memmove_extent_buffer(leaf, BTRFS_LEAF_DATA_OFFSET +
 			      data_end + dsize,
@@ -4967,10 +4301,8 @@ int btrfs_del_items(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 		for (i = slot + nr; i < nritems; i++) {
 			u32 ioff;
 
-			item = btrfs_item_nr(i);
-			ioff = btrfs_token_item_offset(leaf, item, &token);
-			btrfs_set_token_item_offset(leaf, item,
-						    ioff + dsize, &token);
+			ioff = btrfs_token_item_offset(&token, i);
+			btrfs_set_token_item_offset(&token, i, ioff + dsize);
 		}
 
 		memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot),
@@ -4986,7 +4318,6 @@ int btrfs_del_items(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 		if (leaf == root->node) {
 			btrfs_set_header_level(leaf, 0);
 		} else {
-			btrfs_set_path_blocking(path);
 			btrfs_clean_tree_block(leaf);
 			btrfs_del_leaf(trans, root, path, leaf);
 		}
@@ -4999,25 +4330,50 @@ int btrfs_del_items(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 			fixup_low_keys(path, &disk_key, 1);
 		}
 
-		/* delete the leaf if it is mostly empty */
+		/*
+		 * Try to delete the leaf if it is mostly empty. We do this by
+		 * trying to move all its items into its left and right neighbours.
+		 * If we can't move all the items, then we don't delete it - it's
+		 * not ideal, but future insertions might fill the leaf with more
+		 * items, or items from other leaves might be moved later into our
+		 * leaf due to deletions on those leaves.
+		 */
 		if (used < BTRFS_LEAF_DATA_SIZE(fs_info) / 3) {
+			u32 min_push_space;
+
 			/* push_leaf_left fixes the path.
 			 * make sure the path still points to our leaf
 			 * for possible call to del_ptr below
 			 */
 			slot = path->slots[1];
 			atomic_inc(&leaf->refs);
-
-			btrfs_set_path_blocking(path);
-			wret = push_leaf_left(trans, root, path, 1, 1,
-					      1, (u32)-1);
+			/*
+			 * We want to be able to at least push one item to the
+			 * left neighbour leaf, and that's the first item.
+			 */
+			min_push_space = sizeof(struct btrfs_item) +
+				btrfs_item_size(leaf, 0);
+			wret = push_leaf_left(trans, root, path, 0,
+					      min_push_space, 1, (u32)-1);
 			if (wret < 0 && wret != -ENOSPC)
 				ret = wret;
 
 			if (path->nodes[0] == leaf &&
 			    btrfs_header_nritems(leaf)) {
-				wret = push_leaf_right(trans, root, path, 1,
-						       1, 1, 0);
+				/*
+				 * If we were not able to push all items from our
+				 * leaf to its left neighbour, then attempt to
+				 * either push all the remaining items to the
+				 * right neighbour or none. There's no advantage
+				 * in pushing only some items, instead of all, as
+				 * it's pointless to end up with a leaf having
+				 * too few items while the neighbours can be full
+				 * or nearly full.
+				 */
+				nritems = btrfs_header_nritems(leaf);
+				min_push_space = leaf_space_used(leaf, 0, nritems);
+				wret = push_leaf_right(trans, root, path, 0,
+						       min_push_space, 1, 0);
 				if (wret < 0 && wret != -ENOSPC)
 					ret = wret;
 			}
@@ -5126,6 +4482,7 @@ int btrfs_search_forward(struct btrfs_root *root, struct btrfs_key *min_key,
 	int ret = 1;
 	int keep_locks = path->keep_locks;
 
+	ASSERT(!path->nowait);
 	path->keep_locks = 1;
 again:
 	cur = btrfs_read_lock_root_node(root);
@@ -5141,7 +4498,7 @@ again:
 	while (1) {
 		nritems = btrfs_header_nritems(cur);
 		level = btrfs_header_level(cur);
-		sret = btrfs_bin_search(cur, min_key, level, &slot);
+		sret = btrfs_bin_search(cur, min_key, &slot);
 		if (sret < 0) {
 			ret = sret;
 			goto out;
@@ -5160,7 +4517,7 @@ again:
 			slot--;
 		/*
 		 * check this node pointer against the min_trans parameters.
-		 * If it is too old, old, skip to the next one.
+		 * If it is too old, skip to the next one.
 		 */
 		while (slot < nritems) {
 			u64 gen;
@@ -5179,7 +4536,6 @@ find_next_key:
 		 */
 		if (slot >= nritems) {
 			path->slots[level] = slot;
-			btrfs_set_path_blocking(path);
 			sret = btrfs_find_next_key(root, path, min_key, level,
 						  min_trans);
 			if (sret == 0) {
@@ -5196,7 +4552,6 @@ find_next_key:
 			ret = 0;
 			goto out;
 		}
-		btrfs_set_path_blocking(path);
 		cur = btrfs_read_node_slot(cur, slot);
 		if (IS_ERR(cur)) {
 			ret = PTR_ERR(cur);
@@ -5213,7 +4568,6 @@ out:
 	path->keep_locks = keep_locks;
 	if (ret == 0) {
 		btrfs_unlock_up_safe(path, path->lowest_level + 1);
-		btrfs_set_path_blocking(path);
 		memcpy(min_key, &found_key, sizeof(found_key));
 	}
 	return ret;
@@ -5295,16 +4649,6 @@ next:
 	return 1;
 }
 
-/*
- * search the tree again to find a leaf with greater keys
- * returns 0 if it found something or 1 if there are no greater leaves.
- * returns < 0 on io errors.
- */
-int btrfs_next_leaf(struct btrfs_root *root, struct btrfs_path *path)
-{
-	return btrfs_next_old_leaf(root, path, 0);
-}
-
 int btrfs_next_old_leaf(struct btrfs_root *root, struct btrfs_path *path,
 			u64 time_seq)
 {
@@ -5312,11 +4656,14 @@ int btrfs_next_old_leaf(struct btrfs_root *root, struct btrfs_path *path,
 	int level;
 	struct extent_buffer *c;
 	struct extent_buffer *next;
+	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct btrfs_key key;
+	bool need_commit_sem = false;
 	u32 nritems;
 	int ret;
-	int old_spinning = path->leave_spinning;
-	int next_rw_lock = 0;
+	int i;
+
+	ASSERT(!path->nowait);
 
 	nritems = btrfs_header_nritems(path->nodes[0]);
 	if (nritems == 0)
@@ -5326,20 +4673,24 @@ int btrfs_next_old_leaf(struct btrfs_root *root, struct btrfs_path *path,
 again:
 	level = 1;
 	next = NULL;
-	next_rw_lock = 0;
 	btrfs_release_path(path);
 
 	path->keep_locks = 1;
-	path->leave_spinning = 1;
 
-	if (time_seq)
+	if (time_seq) {
 		ret = btrfs_search_old_slot(root, &key, path, time_seq);
-	else
+	} else {
+		if (path->need_commit_sem) {
+			path->need_commit_sem = 0;
+			need_commit_sem = true;
+			down_read(&fs_info->commit_root_sem);
+		}
 		ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
+	}
 	path->keep_locks = 0;
 
 	if (ret < 0)
-		return ret;
+		goto done;
 
 	nritems = btrfs_header_nritems(path->nodes[0]);
 	/*
@@ -5390,13 +4741,22 @@ again:
 			continue;
 		}
 
-		if (next) {
-			btrfs_tree_unlock_rw(next, next_rw_lock);
-			free_extent_buffer(next);
+
+		/*
+		 * Our current level is where we're going to start from, and to
+		 * make sure lockdep doesn't complain we need to drop our locks
+		 * and nodes from 0 to our current level.
+		 */
+		for (i = 0; i < level; i++) {
+			if (path->locks[level]) {
+				btrfs_tree_read_unlock(path->nodes[i]);
+				path->locks[i] = 0;
+			}
+			free_extent_buffer(path->nodes[i]);
+			path->nodes[i] = NULL;
 		}
 
 		next = c;
-		next_rw_lock = path->locks[level];
 		ret = read_block_for_search(root, path, &next, level,
 					    slot, &key);
 		if (ret == -EAGAIN)
@@ -5422,26 +4782,18 @@ again:
 				cond_resched();
 				goto again;
 			}
-			if (!ret) {
-				btrfs_set_path_blocking(path);
+			if (!ret)
 				btrfs_tree_read_lock(next);
-			}
-			next_rw_lock = BTRFS_READ_LOCK;
 		}
 		break;
 	}
 	path->slots[level] = slot;
 	while (1) {
 		level--;
-		c = path->nodes[level];
-		if (path->locks[level])
-			btrfs_tree_unlock_rw(c, path->locks[level]);
-
-		free_extent_buffer(c);
 		path->nodes[level] = next;
 		path->slots[level] = 0;
 		if (!path->skip_locking)
-			path->locks[level] = next_rw_lock;
+			path->locks[level] = BTRFS_READ_LOCK;
 		if (!level)
 			break;
 
@@ -5455,21 +4807,21 @@ again:
 			goto done;
 		}
 
-		if (!path->skip_locking) {
-			ret = btrfs_try_tree_read_lock(next);
-			if (!ret) {
-				btrfs_set_path_blocking(path);
-				btrfs_tree_read_lock(next);
-			}
-			next_rw_lock = BTRFS_READ_LOCK;
-		}
+		if (!path->skip_locking)
+			btrfs_tree_read_lock(next);
 	}
 	ret = 0;
 done:
 	unlock_up(path, 0, 1, 0, NULL);
-	path->leave_spinning = old_spinning;
-	if (!old_spinning)
-		btrfs_set_path_blocking(path);
+	if (need_commit_sem) {
+		int ret2;
+
+		path->need_commit_sem = 1;
+		ret2 = finish_need_commit_sem_search(path);
+		up_read(&fs_info->commit_root_sem);
+		if (ret2)
+			ret = ret2;
+	}
 
 	return ret;
 }
@@ -5491,7 +4843,6 @@ int btrfs_previous_item(struct btrfs_root *root,
 
 	while (1) {
 		if (path->slots[0] == 0) {
-			btrfs_set_path_blocking(path);
 			ret = btrfs_prev_leaf(root, path);
 			if (ret != 0)
 				return ret;
@@ -5533,7 +4884,6 @@ int btrfs_previous_extent_item(struct btrfs_root *root,
 
 	while (1) {
 		if (path->slots[0] == 0) {
-			btrfs_set_path_blocking(path);
 			ret = btrfs_prev_leaf(root, path);
 			if (ret != 0)
 				return ret;
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 36df977b64d9..9e6d48ff4597 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -17,7 +17,6 @@
 #include <linux/wait.h>
 #include <linux/slab.h>
 #include <trace/events/btrfs.h>
-#include <asm/kmap_types.h>
 #include <asm/unaligned.h>
 #include <linux/pagemap.h>
 #include <linux/btrfs.h>
@@ -28,11 +27,13 @@
 #include <linux/dynamic_debug.h>
 #include <linux/refcount.h>
 #include <linux/crc32c.h>
+#include <linux/iomap.h>
 #include "extent-io-tree.h"
 #include "extent_io.h"
 #include "extent_map.h"
 #include "async-thread.h"
 #include "block-rsv.h"
+#include "locking.h"
 
 struct btrfs_trans_handle;
 struct btrfs_transaction;
@@ -41,12 +42,18 @@ struct btrfs_delayed_ref_root;
 struct btrfs_space_info;
 struct btrfs_block_group;
 extern struct kmem_cache *btrfs_trans_handle_cachep;
-extern struct kmem_cache *btrfs_bit_radix_cachep;
 extern struct kmem_cache *btrfs_path_cachep;
 extern struct kmem_cache *btrfs_free_space_cachep;
 extern struct kmem_cache *btrfs_free_space_bitmap_cachep;
 struct btrfs_ordered_sum;
 struct btrfs_ref;
+struct btrfs_bio;
+struct btrfs_ioctl_encoded_io_args;
+struct btrfs_device;
+struct btrfs_fs_devices;
+struct btrfs_balance_control;
+struct btrfs_delayed_root;
+struct reloc_control;
 
 #define BTRFS_MAGIC 0x4D5F53665248425FULL /* ascii _BHRfS_M, no null */
 
@@ -66,12 +73,6 @@ struct btrfs_ref;
 #define BTRFS_OLDEST_GENERATION	0ULL
 
 /*
- * the max metadata block size.  This limit is somewhat artificial,
- * but the memmove costs go through the roof for larger blocks.
- */
-#define BTRFS_MAX_METADATA_BLOCKSIZE 65536
-
-/*
  * we can actually store much bigger names, but lets not confuse the rest
  * of linux
  */
@@ -110,14 +111,6 @@ struct btrfs_ref;
 #define BTRFS_STAT_CURR		0
 #define BTRFS_STAT_PREV		1
 
-/*
- * Count how many BTRFS_MAX_EXTENT_SIZE cover the @size
- */
-static inline u32 count_max_extents(u64 size)
-{
-	return div_u64(size + BTRFS_MAX_EXTENT_SIZE - 1, BTRFS_MAX_EXTENT_SIZE);
-}
-
 static inline unsigned long btrfs_chunk_item_size(int num_stripes)
 {
 	BUG_ON(num_stripes == 0);
@@ -136,6 +129,8 @@ enum {
 	 * defrag
 	 */
 	BTRFS_FS_STATE_REMOUNTING,
+	/* Filesystem in RO mode */
+	BTRFS_FS_STATE_RO,
 	/* Track if a transaction abort has been reported on this filesystem */
 	BTRFS_FS_STATE_TRANS_ABORTED,
 	/*
@@ -145,6 +140,13 @@ enum {
 	BTRFS_FS_STATE_DEV_REPLACING,
 	/* The btrfs_fs_info created for self-tests */
 	BTRFS_FS_STATE_DUMMY_FS_INFO,
+
+	BTRFS_FS_STATE_NO_CSUMS,
+
+	/* Indicates there was an error cleaning up a log tree. */
+	BTRFS_FS_STATE_LOG_CLEANUP_ERROR,
+
+	BTRFS_FS_STATE_COUNT
 };
 
 #define BTRFS_BACKREF_REV_MAX		256
@@ -220,6 +222,16 @@ struct btrfs_root_backup {
 	u8 unused_8[10];
 } __attribute__ ((__packed__));
 
+#define BTRFS_SUPER_INFO_OFFSET			SZ_64K
+#define BTRFS_SUPER_INFO_SIZE			4096
+
+/*
+ * The reserved space at the beginning of each device.
+ * It covers the primary super block and leaves space for potential use by other
+ * tools like bootloaders or to lower potential damage of accidental overwrite.
+ */
+#define BTRFS_DEVICE_RANGE_RESERVED			(SZ_1M)
+
 /*
  * the super block basically lists the main trees of the FS
  * it currently lacks any block count etc etc
@@ -239,8 +251,12 @@ struct btrfs_super_block {
 	__le64 chunk_root;
 	__le64 log_root;
 
-	/* this will help find the new super based on the log root */
-	__le64 log_root_transid;
+	/*
+	 * This member has never been utilized since the very beginning, thus
+	 * it's always 0 regardless of kernel version.  We always use
+	 * generation + 1 to read log tree root.  So here we mark it deprecated.
+	 */
+	__le64 __unused_log_root_transid;
 	__le64 total_bytes;
 	__le64 bytes_used;
 	__le64 root_dir_objectid;
@@ -269,10 +285,15 @@ struct btrfs_super_block {
 	u8 metadata_uuid[BTRFS_FSID_SIZE];
 
 	/* future expansion */
-	__le64 reserved[28];
+	u8 reserved8[8];
+	__le64 reserved[27];
 	u8 sys_chunk_array[BTRFS_SYSTEM_CHUNK_ARRAY_SIZE];
 	struct btrfs_root_backup super_roots[BTRFS_NUM_BACKUP_ROOTS];
+
+	/* Padded to 4096 bytes */
+	u8 padding[565];
 } __attribute__ ((__packed__));
+static_assert(sizeof(struct btrfs_super_block) == BTRFS_SUPER_INFO_SIZE);
 
 /*
  * Compat flags that we support.  If any incompat flags are set other than the
@@ -284,11 +305,33 @@ struct btrfs_super_block {
 
 #define BTRFS_FEATURE_COMPAT_RO_SUPP			\
 	(BTRFS_FEATURE_COMPAT_RO_FREE_SPACE_TREE |	\
-	 BTRFS_FEATURE_COMPAT_RO_FREE_SPACE_TREE_VALID)
+	 BTRFS_FEATURE_COMPAT_RO_FREE_SPACE_TREE_VALID | \
+	 BTRFS_FEATURE_COMPAT_RO_VERITY |		\
+	 BTRFS_FEATURE_COMPAT_RO_BLOCK_GROUP_TREE)
 
 #define BTRFS_FEATURE_COMPAT_RO_SAFE_SET	0ULL
 #define BTRFS_FEATURE_COMPAT_RO_SAFE_CLEAR	0ULL
 
+#ifdef CONFIG_BTRFS_DEBUG
+/*
+ * Extent tree v2 supported only with CONFIG_BTRFS_DEBUG
+ */
+#define BTRFS_FEATURE_INCOMPAT_SUPP			\
+	(BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF |		\
+	 BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL |	\
+	 BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS |		\
+	 BTRFS_FEATURE_INCOMPAT_BIG_METADATA |		\
+	 BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO |		\
+	 BTRFS_FEATURE_INCOMPAT_COMPRESS_ZSTD |		\
+	 BTRFS_FEATURE_INCOMPAT_RAID56 |		\
+	 BTRFS_FEATURE_INCOMPAT_EXTENDED_IREF |		\
+	 BTRFS_FEATURE_INCOMPAT_SKINNY_METADATA |	\
+	 BTRFS_FEATURE_INCOMPAT_NO_HOLES	|	\
+	 BTRFS_FEATURE_INCOMPAT_METADATA_UUID	|	\
+	 BTRFS_FEATURE_INCOMPAT_RAID1C34	|	\
+	 BTRFS_FEATURE_INCOMPAT_ZONED		|	\
+	 BTRFS_FEATURE_INCOMPAT_EXTENT_TREE_V2)
+#else
 #define BTRFS_FEATURE_INCOMPAT_SUPP			\
 	(BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF |		\
 	 BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL |	\
@@ -301,7 +344,9 @@ struct btrfs_super_block {
 	 BTRFS_FEATURE_INCOMPAT_SKINNY_METADATA |	\
 	 BTRFS_FEATURE_INCOMPAT_NO_HOLES	|	\
 	 BTRFS_FEATURE_INCOMPAT_METADATA_UUID	|	\
-	 BTRFS_FEATURE_INCOMPAT_RAID1C34)
+	 BTRFS_FEATURE_INCOMPAT_RAID1C34	|	\
+	 BTRFS_FEATURE_INCOMPAT_ZONED)
+#endif
 
 #define BTRFS_FEATURE_INCOMPAT_SAFE_SET			\
 	(BTRFS_FEATURE_INCOMPAT_EXTENDED_IREF)
@@ -344,6 +389,27 @@ struct btrfs_node {
 	struct btrfs_key_ptr ptrs[];
 } __attribute__ ((__packed__));
 
+/* Read ahead values for struct btrfs_path.reada */
+enum {
+	READA_NONE,
+	READA_BACK,
+	READA_FORWARD,
+	/*
+	 * Similar to READA_FORWARD but unlike it:
+	 *
+	 * 1) It will trigger readahead even for leaves that are not close to
+	 *    each other on disk;
+	 * 2) It also triggers readahead for nodes;
+	 * 3) During a search, even when a node or leaf is already in memory, it
+	 *    will still trigger readahead for other nodes and leaves that follow
+	 *    it.
+	 *
+	 * This is meant to be used only when we know we are iterating over the
+	 * entire tree or a very large part of it.
+	 */
+	READA_FORWARD_ALWAYS,
+};
+
 /*
  * btrfs_paths remember the path taken from the root down to the leaf.
  * level 0 is always the leaf, and nodes[1...BTRFS_MAX_LEVEL] will point
@@ -352,7 +418,6 @@ struct btrfs_node {
  * The slots array records the index of the item or block pointer
  * used while walking the tree.
  */
-enum { READA_NONE, READA_BACK, READA_FORWARD };
 struct btrfs_path {
 	struct extent_buffer *nodes[BTRFS_MAX_LEVEL];
 	int slots[BTRFS_MAX_LEVEL];
@@ -369,13 +434,19 @@ struct btrfs_path {
 	unsigned int search_for_split:1;
 	unsigned int keep_locks:1;
 	unsigned int skip_locking:1;
-	unsigned int leave_spinning:1;
 	unsigned int search_commit_root:1;
 	unsigned int need_commit_sem:1;
 	unsigned int skip_release_on_error:1;
+	/*
+	 * Indicate that new item (btrfs_search_slot) is extending already
+	 * existing item and ins_len contains only the data size and not item
+	 * header (ie. sizeof(struct btrfs_item) is not included).
+	 */
+	unsigned int search_for_extension:1;
+	/* Stop search if any locks need to be taken (for read) */
+	unsigned int nowait:1;
 };
-#define BTRFS_MAX_EXTENT_ITEM_SIZE(r) ((BTRFS_LEAF_DATA_SIZE(r->fs_info) >> 4) - \
-					sizeof(struct btrfs_item))
+
 struct btrfs_dev_replace {
 	u64 replace_state;	/* see #define above */
 	time64_t time_started;	/* seconds since 1-Jan-1970 */
@@ -432,22 +503,6 @@ struct btrfs_free_cluster {
 	struct list_head block_group_list;
 };
 
-enum btrfs_caching_type {
-	BTRFS_CACHE_NO,
-	BTRFS_CACHE_STARTED,
-	BTRFS_CACHE_FAST,
-	BTRFS_CACHE_FINISHED,
-	BTRFS_CACHE_ERROR,
-};
-
-/*
- * Tree to record all locked full stripes of a RAID5/6 block group
- */
-struct btrfs_full_stripe_locks_tree {
-	struct rb_root root;
-	struct mutex lock;
-};
-
 /* Discard control. */
 /*
  * Async discard uses multiple lists to differentiate the discard filter
@@ -467,10 +522,11 @@ struct btrfs_discard_ctl {
 	struct btrfs_block_group *block_group;
 	struct list_head discard_list[BTRFS_NR_DISCARD_LISTS];
 	u64 prev_discard;
+	u64 prev_discard_time;
 	atomic_t discardable_extents;
 	atomic64_t discardable_bytes;
 	u64 max_discard_size;
-	unsigned long delay;
+	u64 delay_ms;
 	u32 iops_limit;
 	u32 kbps_limit;
 	u64 discard_extent_bytes;
@@ -478,54 +534,7 @@ struct btrfs_discard_ctl {
 	atomic64_t discard_bytes_saved;
 };
 
-/* delayed seq elem */
-struct seq_list {
-	struct list_head list;
-	u64 seq;
-};
-
-#define SEQ_LIST_INIT(name)	{ .list = LIST_HEAD_INIT((name).list), .seq = 0 }
-
-#define SEQ_LAST	((u64)-1)
-
-enum btrfs_orphan_cleanup_state {
-	ORPHAN_CLEANUP_STARTED	= 1,
-	ORPHAN_CLEANUP_DONE	= 2,
-};
-
-void btrfs_init_async_reclaim_work(struct work_struct *work);
-
-/* fs_info */
-struct reloc_control;
-struct btrfs_device;
-struct btrfs_fs_devices;
-struct btrfs_balance_control;
-struct btrfs_delayed_root;
-
-/*
- * Block group or device which contains an active swapfile. Used for preventing
- * unsafe operations while a swapfile is active.
- *
- * These are sorted on (ptr, inode) (note that a block group or device can
- * contain more than one swapfile). We compare the pointer values because we
- * don't actually care what the object is, we just need a quick check whether
- * the object exists in the rbtree.
- */
-struct btrfs_swapfile_pin {
-	struct rb_node node;
-	void *ptr;
-	struct inode *inode;
-	/*
-	 * If true, ptr points to a struct btrfs_block_group. Otherwise, ptr
-	 * points to a struct btrfs_device.
-	 */
-	bool is_block_group;
-};
-
-bool btrfs_pinned_by_swapfile(struct btrfs_fs_info *fs_info, void *ptr);
-
 enum {
-	BTRFS_FS_BARRIER,
 	BTRFS_FS_CLOSING_START,
 	BTRFS_FS_CLOSING_DONE,
 	BTRFS_FS_LOG_RECOVERING,
@@ -540,22 +549,17 @@ enum {
 	/* Used to record internally whether fs has been frozen */
 	BTRFS_FS_FROZEN,
 	/*
-	 * Indicate that a whole-filesystem exclusive operation is running
-	 * (device replace, resize, device add/delete, balance)
-	 */
-	BTRFS_FS_EXCL_OP,
-	/*
-	 * To info transaction_kthread we need an immediate commit so it
-	 * doesn't need to wait for commit_interval
-	 */
-	BTRFS_FS_NEED_ASYNC_COMMIT,
-	/*
 	 * Indicate that balance has been set up from the ioctl and is in the
 	 * main phase. The fs_info::balance_ctl is initialized.
-	 * Set and cleared while holding fs_info::balance_mutex.
 	 */
 	BTRFS_FS_BALANCE_RUNNING,
 
+	/*
+	 * Indicate that relocation of a chunk has started, it's set per chunk
+	 * and is toggled between chunks.
+	 */
+	BTRFS_FS_RELOC_RUNNING,
+
 	/* Indicate that the cleaner thread is awake and doing something. */
 	BTRFS_FS_CLEANER_RUNNING,
 
@@ -567,37 +571,89 @@ enum {
 
 	/* Indicate that the discard workqueue can service discards. */
 	BTRFS_FS_DISCARD_RUNNING,
+
+	/* Indicate that we need to cleanup space cache v1 */
+	BTRFS_FS_CLEANUP_SPACE_CACHE_V1,
+
+	/* Indicate that we can't trust the free space tree for caching yet */
+	BTRFS_FS_FREE_SPACE_TREE_UNTRUSTED,
+
+	/* Indicate whether there are any tree modification log users */
+	BTRFS_FS_TREE_MOD_LOG_USERS,
+
+	/* Indicate that we want the transaction kthread to commit right now. */
+	BTRFS_FS_COMMIT_TRANS,
+
+	/* Indicate we have half completed snapshot deletions pending. */
+	BTRFS_FS_UNFINISHED_DROPS,
+
+	/* Indicate we have to finish a zone to do next allocation. */
+	BTRFS_FS_NEED_ZONE_FINISH,
+
+#if BITS_PER_LONG == 32
+	/* Indicate if we have error/warn message printed on 32bit systems */
+	BTRFS_FS_32BIT_ERROR,
+	BTRFS_FS_32BIT_WARN,
+#endif
+};
+
+/*
+ * Exclusive operations (device replace, resize, device add/remove, balance)
+ */
+enum btrfs_exclusive_operation {
+	BTRFS_EXCLOP_NONE,
+	BTRFS_EXCLOP_BALANCE_PAUSED,
+	BTRFS_EXCLOP_BALANCE,
+	BTRFS_EXCLOP_DEV_ADD,
+	BTRFS_EXCLOP_DEV_REMOVE,
+	BTRFS_EXCLOP_DEV_REPLACE,
+	BTRFS_EXCLOP_RESIZE,
+	BTRFS_EXCLOP_SWAP_ACTIVATE,
+};
+
+/* Store data about transaction commits, exported via sysfs. */
+struct btrfs_commit_stats {
+	/* Total number of commits */
+	u64 commit_count;
+	/* The maximum commit duration so far in ns */
+	u64 max_commit_dur;
+	/* The last commit duration in ns */
+	u64 last_commit_dur;
+	/* The total commit duration in ns */
+	u64 total_commit_dur;
 };
 
 struct btrfs_fs_info {
 	u8 chunk_tree_uuid[BTRFS_UUID_SIZE];
 	unsigned long flags;
-	struct btrfs_root *extent_root;
 	struct btrfs_root *tree_root;
 	struct btrfs_root *chunk_root;
 	struct btrfs_root *dev_root;
 	struct btrfs_root *fs_root;
-	struct btrfs_root *csum_root;
 	struct btrfs_root *quota_root;
 	struct btrfs_root *uuid_root;
-	struct btrfs_root *free_space_root;
+	struct btrfs_root *data_reloc_root;
+	struct btrfs_root *block_group_root;
 
 	/* the log root tree is a directory of all the other log roots */
 	struct btrfs_root *log_root_tree;
 
+	/* The tree that holds the global roots (csum, extent, etc) */
+	rwlock_t global_root_lock;
+	struct rb_root global_root_tree;
+
 	spinlock_t fs_roots_radix_lock;
 	struct radix_tree_root fs_roots_radix;
 
 	/* block group cache stuff */
-	spinlock_t block_group_cache_lock;
-	u64 first_logical_byte;
-	struct rb_root block_group_cache_tree;
+	rwlock_t block_group_cache_lock;
+	struct rb_root_cached block_group_cache_tree;
 
 	/* keep track of unallocated space */
 	atomic64_t free_chunk_space;
 
-	struct extent_io_tree freed_extents[2];
-	struct extent_io_tree *pinned_extents;
+	/* Track ranges which are used by log trees blocks/logged data extents */
+	struct extent_io_tree excluded_extents;
 
 	/* logical->physical extent mapping */
 	struct extent_map_tree mapping_tree;
@@ -620,6 +676,12 @@ struct btrfs_fs_info {
 
 	u64 generation;
 	u64 last_trans_committed;
+	/*
+	 * Generation of the last transaction used for block group relocation
+	 * since the filesystem was last mounted (or 0 if none happened yet).
+	 * Must be written and read while holding btrfs_fs_info::commit_root_sem.
+	 */
+	u64 last_reloc_trans;
 	u64 avg_delayed_ref_runtime;
 
 	/*
@@ -696,7 +758,6 @@ struct btrfs_fs_info {
 	struct rw_semaphore cleanup_work_sem;
 
 	struct rw_semaphore subvol_sem;
-	struct srcu_struct subvol_srcu;
 
 	spinlock_t trans_lock;
 	/*
@@ -753,18 +814,17 @@ struct btrfs_fs_info {
 	 * two
 	 */
 	struct btrfs_workqueue *workers;
+	struct btrfs_workqueue *hipri_workers;
 	struct btrfs_workqueue *delalloc_workers;
 	struct btrfs_workqueue *flush_workers;
-	struct btrfs_workqueue *endio_workers;
-	struct btrfs_workqueue *endio_meta_workers;
-	struct btrfs_workqueue *endio_raid56_workers;
-	struct btrfs_workqueue *endio_repair_workers;
-	struct btrfs_workqueue *rmw_workers;
-	struct btrfs_workqueue *endio_meta_write_workers;
+	struct workqueue_struct *endio_workers;
+	struct workqueue_struct *endio_meta_workers;
+	struct workqueue_struct *endio_raid56_workers;
+	struct workqueue_struct *rmw_workers;
+	struct workqueue_struct *compressed_write_workers;
 	struct btrfs_workqueue *endio_write_workers;
 	struct btrfs_workqueue *endio_freespace_worker;
 	struct btrfs_workqueue *caching_workers;
-	struct btrfs_workqueue *readahead_workers;
 
 	/*
 	 * fixup workers take dirty pages that didn't properly go through
@@ -779,13 +839,13 @@ struct btrfs_fs_info {
 	u32 thread_pool_size;
 
 	struct kobject *space_info_kobj;
-
-	u64 total_pinned;
+	struct kobject *qgroups_kobj;
+	struct kobject *discard_kobj;
 
 	/* used to keep from writing metadata until there is a nice batch */
 	struct percpu_counter dirty_metadata_bytes;
 	struct percpu_counter delalloc_bytes;
-	struct percpu_counter dio_bytes;
+	struct percpu_counter ordered_bytes;
 	s32 dirty_metadata_batch;
 	s32 delalloc_batch;
 
@@ -834,6 +894,9 @@ struct btrfs_fs_info {
 	struct btrfs_balance_control *balance_ctl;
 	wait_queue_head_t balance_wait_q;
 
+	/* Cancellation requests for chunk relocation */
+	atomic_t reloc_cancel_req;
+
 	u32 data_chunk_allocations;
 	u32 metadata_ratio;
 
@@ -851,9 +914,10 @@ struct btrfs_fs_info {
 	 * running.
 	 */
 	refcount_t scrub_workers_refcnt;
-	struct btrfs_workqueue *scrub_workers;
-	struct btrfs_workqueue *scrub_wr_completion_workers;
-	struct btrfs_workqueue *scrub_parity_workers;
+	struct workqueue_struct *scrub_workers;
+	struct workqueue_struct *scrub_wr_completion_workers;
+	struct workqueue_struct *scrub_parity_workers;
+	struct btrfs_subpage_info *subpage_info;
 
 	struct btrfs_discard_ctl discard_ctl;
 
@@ -873,7 +937,10 @@ struct btrfs_fs_info {
 	 */
 	struct ulist *qgroup_ulist;
 
-	/* protect user change for quota operations */
+	/*
+	 * Protect user change for quota operations. If a transaction is needed,
+	 * it must be started before locking this lock.
+	 */
 	struct mutex qgroup_ioctl_lock;
 
 	/* list of dirty qgroups to be written at next commit */
@@ -889,21 +956,16 @@ struct btrfs_fs_info {
 	struct completion qgroup_rescan_completion;
 	struct btrfs_work qgroup_rescan_work;
 	bool qgroup_rescan_running;	/* protected by qgroup_rescan_lock */
+	u8 qgroup_drop_subtree_thres;
 
 	/* filesystem state */
 	unsigned long fs_state;
 
 	struct btrfs_delayed_root *delayed_root;
 
-	/* readahead tree */
-	spinlock_t reada_lock;
-	struct radix_tree_root reada_tree;
-
-	/* readahead works cnt */
-	atomic_t reada_works_cnt;
-
 	/* Extent buffer radix tree */
 	spinlock_t buffer_lock;
+	/* Entries are eb->start / sectorsize */
 	struct radix_tree_root buffer_radix;
 
 	/* next backup root to be overwritten */
@@ -916,28 +978,88 @@ struct btrfs_fs_info {
 
 	/* Used to reclaim the metadata space in the background. */
 	struct work_struct async_reclaim_work;
+	struct work_struct async_data_reclaim_work;
+	struct work_struct preempt_reclaim_work;
+
+	/* Reclaim partially filled block groups in the background */
+	struct work_struct reclaim_bgs_work;
+	struct list_head reclaim_bgs;
+	int bg_reclaim_threshold;
 
 	spinlock_t unused_bgs_lock;
 	struct list_head unused_bgs;
 	struct mutex unused_bg_unpin_mutex;
-	struct mutex delete_unused_bgs_mutex;
+	/* Protect block groups that are going to be deleted */
+	struct mutex reclaim_bgs_lock;
 
 	/* Cached block sizes */
 	u32 nodesize;
 	u32 sectorsize;
+	/* ilog2 of sectorsize, use to avoid 64bit division */
+	u32 sectorsize_bits;
+	u32 csum_size;
+	u32 csums_per_leaf;
 	u32 stripesize;
 
+	/*
+	 * Maximum size of an extent. BTRFS_MAX_EXTENT_SIZE on regular
+	 * filesystem, on zoned it depends on the device constraints.
+	 */
+	u64 max_extent_size;
+
 	/* Block groups and devices containing active swapfiles. */
 	spinlock_t swapfile_pins_lock;
 	struct rb_root swapfile_pins;
 
 	struct crypto_shash *csum_shash;
 
+	/* Type of exclusive operation running, protected by super_lock */
+	enum btrfs_exclusive_operation exclusive_operation;
+
 	/*
-	 * Number of send operations in progress.
-	 * Updated while holding fs_info::balance_mutex.
+	 * Zone size > 0 when in ZONED mode, otherwise it's used for a check
+	 * if the mode is enabled
 	 */
-	int send_in_progress;
+	u64 zone_size;
+
+	/* Max size to emit ZONE_APPEND write command */
+	u64 max_zone_append_size;
+	struct mutex zoned_meta_io_lock;
+	spinlock_t treelog_bg_lock;
+	u64 treelog_bg;
+
+	/*
+	 * Start of the dedicated data relocation block group, protected by
+	 * relocation_bg_lock.
+	 */
+	spinlock_t relocation_bg_lock;
+	u64 data_reloc_bg;
+	struct mutex zoned_data_reloc_io_lock;
+
+	u64 nr_global_roots;
+
+	spinlock_t zone_active_bgs_lock;
+	struct list_head zone_active_bgs;
+
+	/* Updates are not protected by any lock */
+	struct btrfs_commit_stats commit_stats;
+
+	/*
+	 * Last generation where we dropped a non-relocation root.
+	 * Use btrfs_set_last_root_drop_gen() and btrfs_get_last_root_drop_gen()
+	 * to change it and to read it, respectively.
+	 */
+	u64 last_root_drop_gen;
+
+	/*
+	 * Annotations for transaction events (structures are empty when
+	 * compiled without lockdep).
+	 */
+	struct lockdep_map btrfs_trans_num_writers_map;
+	struct lockdep_map btrfs_trans_num_extwriters_map;
+	struct lockdep_map btrfs_state_change_map[4];
+	struct lockdep_map btrfs_trans_pending_ordered_map;
+	struct lockdep_map btrfs_ordered_extent_map;
 
 #ifdef CONFIG_BTRFS_FS_REF_VERIFY
 	spinlock_t ref_verify_lock;
@@ -946,19 +1068,90 @@ struct btrfs_fs_info {
 
 #ifdef CONFIG_BTRFS_DEBUG
 	struct kobject *debug_kobj;
-	struct kobject *discard_debug_kobj;
+	struct list_head allocated_roots;
+
+	spinlock_t eb_leak_lock;
+	struct list_head allocated_ebs;
 #endif
 };
 
+static inline void btrfs_set_last_root_drop_gen(struct btrfs_fs_info *fs_info,
+						u64 gen)
+{
+	WRITE_ONCE(fs_info->last_root_drop_gen, gen);
+}
+
+static inline u64 btrfs_get_last_root_drop_gen(const struct btrfs_fs_info *fs_info)
+{
+	return READ_ONCE(fs_info->last_root_drop_gen);
+}
+
 static inline struct btrfs_fs_info *btrfs_sb(struct super_block *sb)
 {
 	return sb->s_fs_info;
 }
 
-struct btrfs_subvolume_writers {
-	struct percpu_counter	counter;
-	wait_queue_head_t	wait;
-};
+/*
+ * Take the number of bytes to be checksummed and figure out how many leaves
+ * it would require to store the csums for that many bytes.
+ */
+static inline u64 btrfs_csum_bytes_to_leaves(
+			const struct btrfs_fs_info *fs_info, u64 csum_bytes)
+{
+	const u64 num_csums = csum_bytes >> fs_info->sectorsize_bits;
+
+	return DIV_ROUND_UP_ULL(num_csums, fs_info->csums_per_leaf);
+}
+
+/*
+ * Use this if we would be adding new items, as we could split nodes as we cow
+ * down the tree.
+ */
+static inline u64 btrfs_calc_insert_metadata_size(struct btrfs_fs_info *fs_info,
+						  unsigned num_items)
+{
+	return (u64)fs_info->nodesize * BTRFS_MAX_LEVEL * 2 * num_items;
+}
+
+/*
+ * Doing a truncate or a modification won't result in new nodes or leaves, just
+ * what we need for COW.
+ */
+static inline u64 btrfs_calc_metadata_size(struct btrfs_fs_info *fs_info,
+						 unsigned num_items)
+{
+	return (u64)fs_info->nodesize * BTRFS_MAX_LEVEL * num_items;
+}
+
+#define BTRFS_MAX_EXTENT_ITEM_SIZE(r) ((BTRFS_LEAF_DATA_SIZE(r->fs_info) >> 4) - \
+					sizeof(struct btrfs_item))
+
+static inline bool btrfs_is_zoned(const struct btrfs_fs_info *fs_info)
+{
+	return fs_info->zone_size > 0;
+}
+
+/*
+ * Count how many fs_info->max_extent_size cover the @size
+ */
+static inline u32 count_max_extents(struct btrfs_fs_info *fs_info, u64 size)
+{
+#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
+	if (!fs_info)
+		return div_u64(size + BTRFS_MAX_EXTENT_SIZE - 1, BTRFS_MAX_EXTENT_SIZE);
+#endif
+
+	return div_u64(size + fs_info->max_extent_size - 1, fs_info->max_extent_size);
+}
+
+bool btrfs_exclop_start(struct btrfs_fs_info *fs_info,
+			enum btrfs_exclusive_operation type);
+bool btrfs_exclop_start_try_lock(struct btrfs_fs_info *fs_info,
+				 enum btrfs_exclusive_operation type);
+void btrfs_exclop_start_unlock(struct btrfs_fs_info *fs_info);
+void btrfs_exclop_finish(struct btrfs_fs_info *fs_info);
+void btrfs_exclop_balance(struct btrfs_fs_info *fs_info,
+			  enum btrfs_exclusive_operation op);
 
 /*
  * The state of btrfs root
@@ -971,7 +1164,28 @@ enum {
 	 * is used to tell us when more checks are required
 	 */
 	BTRFS_ROOT_IN_TRANS_SETUP,
-	BTRFS_ROOT_REF_COWS,
+
+	/*
+	 * Set if tree blocks of this root can be shared by other roots.
+	 * Only subvolume trees and their reloc trees have this bit set.
+	 * Conflicts with TRACK_DIRTY bit.
+	 *
+	 * This affects two things:
+	 *
+	 * - How balance works
+	 *   For shareable roots, we need to use reloc tree and do path
+	 *   replacement for balance, and need various pre/post hooks for
+	 *   snapshot creation to handle them.
+	 *
+	 *   While for non-shareable trees, we just simply do a tree search
+	 *   with COW.
+	 *
+	 * - How dirty roots are tracked
+	 *   For shareable roots, btrfs_record_root_in_trans() is needed to
+	 *   track them, while non-subvolume roots have TRACK_DIRTY bit, they
+	 *   don't need to set this manually.
+	 */
+	BTRFS_ROOT_SHAREABLE,
 	BTRFS_ROOT_TRACK_DIRTY,
 	BTRFS_ROOT_IN_RADIX,
 	BTRFS_ROOT_ORPHAN_ITEM_INSERTED,
@@ -989,9 +1203,100 @@ enum {
 	BTRFS_ROOT_DEAD_RELOC_TREE,
 	/* Mark dead root stored on device whose cleanup needs to be resumed */
 	BTRFS_ROOT_DEAD_TREE,
+	/* The root has a log tree. Used for subvolume roots and the tree root. */
+	BTRFS_ROOT_HAS_LOG_TREE,
+	/* Qgroup flushing is in progress */
+	BTRFS_ROOT_QGROUP_FLUSHING,
+	/* We started the orphan cleanup for this root. */
+	BTRFS_ROOT_ORPHAN_CLEANUP,
+	/* This root has a drop operation that was started previously. */
+	BTRFS_ROOT_UNFINISHED_DROP,
+	/* This reloc root needs to have its buffers lockdep class reset. */
+	BTRFS_ROOT_RESET_LOCKDEP_CLASS,
+};
+
+enum btrfs_lockdep_trans_states {
+	BTRFS_LOCKDEP_TRANS_COMMIT_START,
+	BTRFS_LOCKDEP_TRANS_UNBLOCKED,
+	BTRFS_LOCKDEP_TRANS_SUPER_COMMITTED,
+	BTRFS_LOCKDEP_TRANS_COMPLETED,
 };
 
 /*
+ * Lockdep annotation for wait events.
+ *
+ * @owner:  The struct where the lockdep map is defined
+ * @lock:   The lockdep map corresponding to a wait event
+ *
+ * This macro is used to annotate a wait event. In this case a thread acquires
+ * the lockdep map as writer (exclusive lock) because it has to block until all
+ * the threads that hold the lock as readers signal the condition for the wait
+ * event and release their locks.
+ */
+#define btrfs_might_wait_for_event(owner, lock)					\
+	do {									\
+		rwsem_acquire(&owner->lock##_map, 0, 0, _THIS_IP_);		\
+		rwsem_release(&owner->lock##_map, _THIS_IP_);			\
+	} while (0)
+
+/*
+ * Protection for the resource/condition of a wait event.
+ *
+ * @owner:  The struct where the lockdep map is defined
+ * @lock:   The lockdep map corresponding to a wait event
+ *
+ * Many threads can modify the condition for the wait event at the same time
+ * and signal the threads that block on the wait event. The threads that modify
+ * the condition and do the signaling acquire the lock as readers (shared
+ * lock).
+ */
+#define btrfs_lockdep_acquire(owner, lock)					\
+	rwsem_acquire_read(&owner->lock##_map, 0, 0, _THIS_IP_)
+
+/*
+ * Used after signaling the condition for a wait event to release the lockdep
+ * map held by a reader thread.
+ */
+#define btrfs_lockdep_release(owner, lock)					\
+	rwsem_release(&owner->lock##_map, _THIS_IP_)
+
+/*
+ * Macros for the transaction states wait events, similar to the generic wait
+ * event macros.
+ */
+#define btrfs_might_wait_for_state(owner, i)					\
+	do {									\
+		rwsem_acquire(&owner->btrfs_state_change_map[i], 0, 0, _THIS_IP_); \
+		rwsem_release(&owner->btrfs_state_change_map[i], _THIS_IP_);	\
+	} while (0)
+
+#define btrfs_trans_state_lockdep_acquire(owner, i)				\
+	rwsem_acquire_read(&owner->btrfs_state_change_map[i], 0, 0, _THIS_IP_)
+
+#define btrfs_trans_state_lockdep_release(owner, i)				\
+	rwsem_release(&owner->btrfs_state_change_map[i], _THIS_IP_)
+
+/* Initialization of the lockdep map */
+#define btrfs_lockdep_init_map(owner, lock)					\
+	do {									\
+		static struct lock_class_key lock##_key;			\
+		lockdep_init_map(&owner->lock##_map, #lock, &lock##_key, 0);	\
+	} while (0)
+
+/* Initialization of the transaction states lockdep maps. */
+#define btrfs_state_lockdep_init_map(owner, lock, state)			\
+	do {									\
+		static struct lock_class_key lock##_key;			\
+		lockdep_init_map(&owner->btrfs_state_change_map[state], #lock,	\
+				 &lock##_key, 0);				\
+	} while (0)
+
+static inline void btrfs_wake_unfinished_drop(struct btrfs_fs_info *fs_info)
+{
+	clear_and_wake_up_bit(BTRFS_FS_UNFINISHED_DROPS, &fs_info->flags);
+}
+
+/*
  * Record swapped tree blocks of a subvolume tree for delayed subtree trace
  * code. For detail check comment in fs/btrfs/qgroup.c.
  */
@@ -1007,6 +1312,8 @@ struct btrfs_qgroup_swapped_blocks {
  * and for the extent tree extent_root root.
  */
 struct btrfs_root {
+	struct rb_node rb_node;
+
 	struct extent_buffer *node;
 
 	struct extent_buffer *commit_root;
@@ -1024,21 +1331,14 @@ struct btrfs_root {
 	spinlock_t accounting_lock;
 	struct btrfs_block_rsv *block_rsv;
 
-	/* free ino cache stuff */
-	struct btrfs_free_space_ctl *free_ino_ctl;
-	enum btrfs_caching_type ino_cache_state;
-	spinlock_t ino_cache_lock;
-	wait_queue_head_t ino_cache_wait;
-	struct btrfs_free_space_ctl *free_ino_pinned;
-	u64 ino_cache_progress;
-	struct inode *ino_cache_inode;
-
 	struct mutex log_mutex;
 	wait_queue_head_t log_writer_wait;
 	wait_queue_head_t log_commit_wait[2];
 	struct list_head log_ctxs[2];
+	/* Used only for log trees of subvolumes, not for the log root tree */
 	atomic_t log_writers;
 	atomic_t log_commit[2];
+	/* Used only for log trees of subvolumes, not for the log root tree */
 	atomic_t log_batch;
 	int log_transid;
 	/* No matter the commit succeeds or not*/
@@ -1051,13 +1351,12 @@ struct btrfs_root {
 
 	u32 type;
 
-	u64 highest_objectid;
+	u64 free_objectid;
 
-	u64 defrag_trans_start;
 	struct btrfs_key defrag_progress;
 	struct btrfs_key defrag_max;
 
-	/* the dirty list is only used by non-reference counted roots */
+	/* The dirty list is only used by non-shareable roots */
 	struct list_head dirty_list;
 
 	struct list_head root_list;
@@ -1065,8 +1364,6 @@ struct btrfs_root {
 	spinlock_t log_extents_lock[2];
 	struct list_head logged_list[2];
 
-	int orphan_cleanup_state;
-
 	spinlock_t inode_lock;
 	/* red-black tree that keeps track of in-memory inodes */
 	struct rb_root inode_tree;
@@ -1131,14 +1428,16 @@ struct btrfs_root {
 	 * root_item_lock.
 	 */
 	int dedupe_in_progress;
-	struct btrfs_subvolume_writers *subv_writers;
-	atomic_t will_be_snapshotted;
+	/* For exclusion of snapshot creation and nocow writes */
+	struct btrfs_drew_lock snapshot_lock;
+
 	atomic_t snapshot_force_cow;
 
 	/* For qgroup metadata reserved space */
 	spinlock_t qgroup_meta_rsv_lock;
 	u64 qgroup_meta_rsv_pertrans;
 	u64 qgroup_meta_rsv_prealloc;
+	wait_queue_head_t qgroup_flush_wait;
 
 	/* Number of active swapfiles */
 	atomic_t nr_swapfiles;
@@ -1146,29 +1445,111 @@ struct btrfs_root {
 	/* Record pairs of swapped blocks for qgroup */
 	struct btrfs_qgroup_swapped_blocks swapped_blocks;
 
+	/* Used only by log trees, when logging csum items */
+	struct extent_io_tree log_csum_range;
+
 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
 	u64 alloc_bytenr;
 #endif
+
+#ifdef CONFIG_BTRFS_DEBUG
+	struct list_head leak_list;
+#endif
 };
 
-struct btrfs_clone_extent_info {
+/*
+ * Structure that conveys information about an extent that is going to replace
+ * all the extents in a file range.
+ */
+struct btrfs_replace_extent_info {
 	u64 disk_offset;
 	u64 disk_len;
 	u64 data_offset;
 	u64 data_len;
 	u64 file_offset;
+	/* Pointer to a file extent item of type regular or prealloc. */
 	char *extent_buf;
-	u32 item_size;
+	/*
+	 * Set to true when attempting to replace a file range with a new extent
+	 * described by this structure, set to false when attempting to clone an
+	 * existing extent into a file range.
+	 */
+	bool is_new_extent;
+	/* Indicate if we should update the inode's mtime and ctime. */
+	bool update_times;
+	/* Meaningful only if is_new_extent is true. */
+	int qgroup_reserved;
+	/*
+	 * Meaningful only if is_new_extent is true.
+	 * Used to track how many extent items we have already inserted in a
+	 * subvolume tree that refer to the extent described by this structure,
+	 * so that we know when to create a new delayed ref or update an existing
+	 * one.
+	 */
+	int insertions;
+};
+
+/* Arguments for btrfs_drop_extents() */
+struct btrfs_drop_extents_args {
+	/* Input parameters */
+
+	/*
+	 * If NULL, btrfs_drop_extents() will allocate and free its own path.
+	 * If 'replace_extent' is true, this must not be NULL. Also the path
+	 * is always released except if 'replace_extent' is true and
+	 * btrfs_drop_extents() sets 'extent_inserted' to true, in which case
+	 * the path is kept locked.
+	 */
+	struct btrfs_path *path;
+	/* Start offset of the range to drop extents from */
+	u64 start;
+	/* End (exclusive, last byte + 1) of the range to drop extents from */
+	u64 end;
+	/* If true drop all the extent maps in the range */
+	bool drop_cache;
+	/*
+	 * If true it means we want to insert a new extent after dropping all
+	 * the extents in the range. If this is true, the 'extent_item_size'
+	 * parameter must be set as well and the 'extent_inserted' field will
+	 * be set to true by btrfs_drop_extents() if it could insert the new
+	 * extent.
+	 * Note: when this is set to true the path must not be NULL.
+	 */
+	bool replace_extent;
+	/*
+	 * Used if 'replace_extent' is true. Size of the file extent item to
+	 * insert after dropping all existing extents in the range
+	 */
+	u32 extent_item_size;
+
+	/* Output parameters */
+
+	/*
+	 * Set to the minimum between the input parameter 'end' and the end
+	 * (exclusive, last byte + 1) of the last dropped extent. This is always
+	 * set even if btrfs_drop_extents() returns an error.
+	 */
+	u64 drop_end;
+	/*
+	 * The number of allocated bytes found in the range. This can be smaller
+	 * than the range's length when there are holes in the range.
+	 */
+	u64 bytes_found;
+	/*
+	 * Only set if 'replace_extent' is true. Set to true if we were able
+	 * to insert a replacement extent after dropping all extents in the
+	 * range, otherwise set to false by btrfs_drop_extents().
+	 * Also, if btrfs_drop_extents() has set this to true it means it
+	 * returned with the path locked, otherwise if it has set this to
+	 * false it has returned with the path released.
+	 */
+	bool extent_inserted;
 };
 
 struct btrfs_file_private {
 	void *filldir_buf;
 };
 
-static inline u32 btrfs_inode_sectorsize(const struct inode *inode)
-{
-	return btrfs_sb(inode->i_sb)->sectorsize;
-}
 
 static inline u32 BTRFS_LEAF_DATA_SIZE(const struct btrfs_fs_info *info)
 {
@@ -1206,36 +1587,39 @@ static inline u32 BTRFS_MAX_XATTR_SIZE(const struct btrfs_fs_info *info)
  *
  * Note: don't forget to add new options to btrfs_show_options()
  */
-#define BTRFS_MOUNT_NODATASUM		(1 << 0)
-#define BTRFS_MOUNT_NODATACOW		(1 << 1)
-#define BTRFS_MOUNT_NOBARRIER		(1 << 2)
-#define BTRFS_MOUNT_SSD			(1 << 3)
-#define BTRFS_MOUNT_DEGRADED		(1 << 4)
-#define BTRFS_MOUNT_COMPRESS		(1 << 5)
-#define BTRFS_MOUNT_NOTREELOG           (1 << 6)
-#define BTRFS_MOUNT_FLUSHONCOMMIT       (1 << 7)
-#define BTRFS_MOUNT_SSD_SPREAD		(1 << 8)
-#define BTRFS_MOUNT_NOSSD		(1 << 9)
-#define BTRFS_MOUNT_DISCARD_SYNC	(1 << 10)
-#define BTRFS_MOUNT_FORCE_COMPRESS      (1 << 11)
-#define BTRFS_MOUNT_SPACE_CACHE		(1 << 12)
-#define BTRFS_MOUNT_CLEAR_CACHE		(1 << 13)
-#define BTRFS_MOUNT_USER_SUBVOL_RM_ALLOWED (1 << 14)
-#define BTRFS_MOUNT_ENOSPC_DEBUG	 (1 << 15)
-#define BTRFS_MOUNT_AUTO_DEFRAG		(1 << 16)
-#define BTRFS_MOUNT_INODE_MAP_CACHE	(1 << 17)
-#define BTRFS_MOUNT_USEBACKUPROOT	(1 << 18)
-#define BTRFS_MOUNT_SKIP_BALANCE	(1 << 19)
-#define BTRFS_MOUNT_CHECK_INTEGRITY	(1 << 20)
-#define BTRFS_MOUNT_CHECK_INTEGRITY_INCLUDING_EXTENT_DATA (1 << 21)
-#define BTRFS_MOUNT_PANIC_ON_FATAL_ERROR	(1 << 22)
-#define BTRFS_MOUNT_RESCAN_UUID_TREE	(1 << 23)
-#define BTRFS_MOUNT_FRAGMENT_DATA	(1 << 24)
-#define BTRFS_MOUNT_FRAGMENT_METADATA	(1 << 25)
-#define BTRFS_MOUNT_FREE_SPACE_TREE	(1 << 26)
-#define BTRFS_MOUNT_NOLOGREPLAY		(1 << 27)
-#define BTRFS_MOUNT_REF_VERIFY		(1 << 28)
-#define BTRFS_MOUNT_DISCARD_ASYNC	(1 << 29)
+enum {
+	BTRFS_MOUNT_NODATASUM			= (1UL << 0),
+	BTRFS_MOUNT_NODATACOW			= (1UL << 1),
+	BTRFS_MOUNT_NOBARRIER			= (1UL << 2),
+	BTRFS_MOUNT_SSD				= (1UL << 3),
+	BTRFS_MOUNT_DEGRADED			= (1UL << 4),
+	BTRFS_MOUNT_COMPRESS			= (1UL << 5),
+	BTRFS_MOUNT_NOTREELOG   		= (1UL << 6),
+	BTRFS_MOUNT_FLUSHONCOMMIT		= (1UL << 7),
+	BTRFS_MOUNT_SSD_SPREAD			= (1UL << 8),
+	BTRFS_MOUNT_NOSSD			= (1UL << 9),
+	BTRFS_MOUNT_DISCARD_SYNC		= (1UL << 10),
+	BTRFS_MOUNT_FORCE_COMPRESS      	= (1UL << 11),
+	BTRFS_MOUNT_SPACE_CACHE			= (1UL << 12),
+	BTRFS_MOUNT_CLEAR_CACHE			= (1UL << 13),
+	BTRFS_MOUNT_USER_SUBVOL_RM_ALLOWED	= (1UL << 14),
+	BTRFS_MOUNT_ENOSPC_DEBUG		= (1UL << 15),
+	BTRFS_MOUNT_AUTO_DEFRAG			= (1UL << 16),
+	BTRFS_MOUNT_USEBACKUPROOT		= (1UL << 17),
+	BTRFS_MOUNT_SKIP_BALANCE		= (1UL << 18),
+	BTRFS_MOUNT_CHECK_INTEGRITY		= (1UL << 19),
+	BTRFS_MOUNT_CHECK_INTEGRITY_DATA	= (1UL << 20),
+	BTRFS_MOUNT_PANIC_ON_FATAL_ERROR	= (1UL << 21),
+	BTRFS_MOUNT_RESCAN_UUID_TREE		= (1UL << 22),
+	BTRFS_MOUNT_FRAGMENT_DATA		= (1UL << 23),
+	BTRFS_MOUNT_FRAGMENT_METADATA		= (1UL << 24),
+	BTRFS_MOUNT_FREE_SPACE_TREE		= (1UL << 25),
+	BTRFS_MOUNT_NOLOGREPLAY			= (1UL << 26),
+	BTRFS_MOUNT_REF_VERIFY			= (1UL << 27),
+	BTRFS_MOUNT_DISCARD_ASYNC		= (1UL << 28),
+	BTRFS_MOUNT_IGNOREBADROOTS		= (1UL << 29),
+	BTRFS_MOUNT_IGNOREDATACSUMS		= (1UL << 30),
+};
 
 #define BTRFS_DEFAULT_COMMIT_INTERVAL	(30)
 #define BTRFS_DEFAULT_MAX_INLINE	(2048)
@@ -1247,18 +1631,18 @@ static inline u32 BTRFS_MAX_XATTR_SIZE(const struct btrfs_fs_info *info)
 					 BTRFS_MOUNT_##opt)
 
 #define btrfs_set_and_info(fs_info, opt, fmt, args...)			\
-{									\
+do {									\
 	if (!btrfs_test_opt(fs_info, opt))				\
 		btrfs_info(fs_info, fmt, ##args);			\
 	btrfs_set_opt(fs_info->mount_opt, opt);				\
-}
+} while (0)
 
 #define btrfs_clear_and_info(fs_info, opt, fmt, args...)		\
-{									\
+do {									\
 	if (btrfs_test_opt(fs_info, opt))				\
 		btrfs_info(fs_info, fmt, ##args);			\
 	btrfs_clear_opt(fs_info->mount_opt, opt);			\
-}
+} while (0)
 
 /*
  * Requests for changes that need to be done during transaction commit.
@@ -1268,9 +1652,7 @@ static inline u32 BTRFS_MAX_XATTR_SIZE(const struct btrfs_fs_info *info)
  * transaction commit)
  */
 
-#define BTRFS_PENDING_SET_INODE_MAP_CACHE	(0)
-#define BTRFS_PENDING_CLEAR_INODE_MAP_CACHE	(1)
-#define BTRFS_PENDING_COMMIT			(2)
+#define BTRFS_PENDING_COMMIT			(0)
 
 #define btrfs_test_pending(info, opt)	\
 	test_bit(BTRFS_PENDING_##opt, &(info)->pending_changes)
@@ -1306,20 +1688,20 @@ do {                                                                   \
 /*
  * Inode flags
  */
-#define BTRFS_INODE_NODATASUM		(1 << 0)
-#define BTRFS_INODE_NODATACOW		(1 << 1)
-#define BTRFS_INODE_READONLY		(1 << 2)
-#define BTRFS_INODE_NOCOMPRESS		(1 << 3)
-#define BTRFS_INODE_PREALLOC		(1 << 4)
-#define BTRFS_INODE_SYNC		(1 << 5)
-#define BTRFS_INODE_IMMUTABLE		(1 << 6)
-#define BTRFS_INODE_APPEND		(1 << 7)
-#define BTRFS_INODE_NODUMP		(1 << 8)
-#define BTRFS_INODE_NOATIME		(1 << 9)
-#define BTRFS_INODE_DIRSYNC		(1 << 10)
-#define BTRFS_INODE_COMPRESS		(1 << 11)
-
-#define BTRFS_INODE_ROOT_ITEM_INIT	(1 << 31)
+#define BTRFS_INODE_NODATASUM		(1U << 0)
+#define BTRFS_INODE_NODATACOW		(1U << 1)
+#define BTRFS_INODE_READONLY		(1U << 2)
+#define BTRFS_INODE_NOCOMPRESS		(1U << 3)
+#define BTRFS_INODE_PREALLOC		(1U << 4)
+#define BTRFS_INODE_SYNC		(1U << 5)
+#define BTRFS_INODE_IMMUTABLE		(1U << 6)
+#define BTRFS_INODE_APPEND		(1U << 7)
+#define BTRFS_INODE_NODUMP		(1U << 8)
+#define BTRFS_INODE_NOATIME		(1U << 9)
+#define BTRFS_INODE_DIRSYNC		(1U << 10)
+#define BTRFS_INODE_COMPRESS		(1U << 11)
+
+#define BTRFS_INODE_ROOT_ITEM_INIT	(1U << 31)
 
 #define BTRFS_INODE_FLAG_MASK						\
 	(BTRFS_INODE_NODATASUM |					\
@@ -1336,20 +1718,25 @@ do {                                                                   \
 	 BTRFS_INODE_COMPRESS |						\
 	 BTRFS_INODE_ROOT_ITEM_INIT)
 
+#define BTRFS_INODE_RO_VERITY		(1U << 0)
+
+#define BTRFS_INODE_RO_FLAG_MASK	(BTRFS_INODE_RO_VERITY)
+
 struct btrfs_map_token {
-	const struct extent_buffer *eb;
+	struct extent_buffer *eb;
 	char *kaddr;
 	unsigned long offset;
 };
 
 #define BTRFS_BYTES_TO_BLKS(fs_info, bytes) \
-				((bytes) >> (fs_info)->sb->s_blocksize_bits)
+				((bytes) >> (fs_info)->sectorsize_bits)
 
 static inline void btrfs_init_map_token(struct btrfs_map_token *token,
 					struct extent_buffer *eb)
 {
 	token->eb = eb;
-	token->kaddr = NULL;
+	token->kaddr = page_address(eb->pages[0]);
+	token->offset = 0;
 }
 
 /* some macros to generate set/get functions for the struct fields.  This
@@ -1360,6 +1747,16 @@ static inline void btrfs_init_map_token(struct btrfs_map_token *token,
 #define cpu_to_le8(v) (v)
 #define __le8 u8
 
+static inline u8 get_unaligned_le8(const void *p)
+{
+       return *(u8 *)p;
+}
+
+static inline void put_unaligned_le8(u8 val, void *p)
+{
+       *(u8 *)p = val;
+}
+
 #define read_eb_member(eb, ptr, type, member, result) (\
 	read_extent_buffer(eb, (char *)(result),			\
 			   ((unsigned long)(ptr)) +			\
@@ -1373,15 +1770,14 @@ static inline void btrfs_init_map_token(struct btrfs_map_token *token,
 			   sizeof(((type *)0)->member)))
 
 #define DECLARE_BTRFS_SETGET_BITS(bits)					\
-u##bits btrfs_get_token_##bits(const struct extent_buffer *eb,		\
-			       const void *ptr, unsigned long off,	\
-			       struct btrfs_map_token *token);		\
-void btrfs_set_token_##bits(struct extent_buffer *eb, const void *ptr,	\
-			    unsigned long off, u##bits val,		\
-			    struct btrfs_map_token *token);		\
+u##bits btrfs_get_token_##bits(struct btrfs_map_token *token,		\
+			       const void *ptr, unsigned long off);	\
+void btrfs_set_token_##bits(struct btrfs_map_token *token,		\
+			    const void *ptr, unsigned long off,		\
+			    u##bits val);				\
 u##bits btrfs_get_##bits(const struct extent_buffer *eb,		\
 			 const void *ptr, unsigned long off);		\
-void btrfs_set_##bits(struct extent_buffer *eb, void *ptr,		\
+void btrfs_set_##bits(const struct extent_buffer *eb, void *ptr,	\
 		      unsigned long off, u##bits val);
 
 DECLARE_BTRFS_SETGET_BITS(8)
@@ -1393,69 +1789,66 @@ DECLARE_BTRFS_SETGET_BITS(64)
 static inline u##bits btrfs_##name(const struct extent_buffer *eb,	\
 				   const type *s)			\
 {									\
-	BUILD_BUG_ON(sizeof(u##bits) != sizeof(((type *)0))->member);	\
+	static_assert(sizeof(u##bits) == sizeof(((type *)0))->member);	\
 	return btrfs_get_##bits(eb, s, offsetof(type, member));		\
 }									\
-static inline void btrfs_set_##name(struct extent_buffer *eb, type *s,	\
+static inline void btrfs_set_##name(const struct extent_buffer *eb, type *s, \
 				    u##bits val)			\
 {									\
-	BUILD_BUG_ON(sizeof(u##bits) != sizeof(((type *)0))->member);	\
+	static_assert(sizeof(u##bits) == sizeof(((type *)0))->member);	\
 	btrfs_set_##bits(eb, s, offsetof(type, member), val);		\
 }									\
-static inline u##bits btrfs_token_##name(const struct extent_buffer *eb,\
-					 const type *s,			\
-					 struct btrfs_map_token *token)	\
+static inline u##bits btrfs_token_##name(struct btrfs_map_token *token,	\
+					 const type *s)			\
 {									\
-	BUILD_BUG_ON(sizeof(u##bits) != sizeof(((type *)0))->member);	\
-	return btrfs_get_token_##bits(eb, s, offsetof(type, member), token); \
+	static_assert(sizeof(u##bits) == sizeof(((type *)0))->member);	\
+	return btrfs_get_token_##bits(token, s, offsetof(type, member));\
 }									\
-static inline void btrfs_set_token_##name(struct extent_buffer *eb,	\
-					  type *s, u##bits val,		\
-                                         struct btrfs_map_token *token)	\
+static inline void btrfs_set_token_##name(struct btrfs_map_token *token,\
+					  type *s, u##bits val)		\
 {									\
-	BUILD_BUG_ON(sizeof(u##bits) != sizeof(((type *)0))->member);	\
-	btrfs_set_token_##bits(eb, s, offsetof(type, member), val, token); \
+	static_assert(sizeof(u##bits) == sizeof(((type *)0))->member);	\
+	btrfs_set_token_##bits(token, s, offsetof(type, member), val);	\
 }
 
 #define BTRFS_SETGET_HEADER_FUNCS(name, type, member, bits)		\
 static inline u##bits btrfs_##name(const struct extent_buffer *eb)	\
 {									\
-	const type *p = page_address(eb->pages[0]);			\
-	u##bits res = le##bits##_to_cpu(p->member);			\
-	return res;							\
+	const type *p = page_address(eb->pages[0]) +			\
+			offset_in_page(eb->start);			\
+	return get_unaligned_le##bits(&p->member);			\
 }									\
-static inline void btrfs_set_##name(struct extent_buffer *eb,		\
+static inline void btrfs_set_##name(const struct extent_buffer *eb,	\
 				    u##bits val)			\
 {									\
-	type *p = page_address(eb->pages[0]);				\
-	p->member = cpu_to_le##bits(val);				\
+	type *p = page_address(eb->pages[0]) + offset_in_page(eb->start); \
+	put_unaligned_le##bits(val, &p->member);			\
 }
 
 #define BTRFS_SETGET_STACK_FUNCS(name, type, member, bits)		\
 static inline u##bits btrfs_##name(const type *s)			\
 {									\
-	return le##bits##_to_cpu(s->member);				\
+	return get_unaligned_le##bits(&s->member);			\
 }									\
 static inline void btrfs_set_##name(type *s, u##bits val)		\
 {									\
-	s->member = cpu_to_le##bits(val);				\
+	put_unaligned_le##bits(val, &s->member);			\
 }
 
-
-static inline u64 btrfs_device_total_bytes(struct extent_buffer *eb,
+static inline u64 btrfs_device_total_bytes(const struct extent_buffer *eb,
 					   struct btrfs_dev_item *s)
 {
-	BUILD_BUG_ON(sizeof(u64) !=
-		     sizeof(((struct btrfs_dev_item *)0))->total_bytes);
+	static_assert(sizeof(u64) ==
+		      sizeof(((struct btrfs_dev_item *)0))->total_bytes);
 	return btrfs_get_64(eb, s, offsetof(struct btrfs_dev_item,
 					    total_bytes));
 }
-static inline void btrfs_set_device_total_bytes(struct extent_buffer *eb,
+static inline void btrfs_set_device_total_bytes(const struct extent_buffer *eb,
 						struct btrfs_dev_item *s,
 						u64 val)
 {
-	BUILD_BUG_ON(sizeof(u64) !=
-		     sizeof(((struct btrfs_dev_item *)0))->total_bytes);
+	static_assert(sizeof(u64) ==
+		      sizeof(((struct btrfs_dev_item *)0))->total_bytes);
 	WARN_ON(!IS_ALIGNED(val, eb->fs_info->sectorsize));
 	btrfs_set_64(eb, s, offsetof(struct btrfs_dev_item, total_bytes), val);
 }
@@ -1554,13 +1947,13 @@ static inline char *btrfs_stripe_dev_uuid_nr(struct btrfs_chunk *c, int nr)
 	return btrfs_stripe_dev_uuid(btrfs_stripe_nr(c, nr));
 }
 
-static inline u64 btrfs_stripe_offset_nr(struct extent_buffer *eb,
+static inline u64 btrfs_stripe_offset_nr(const struct extent_buffer *eb,
 					 struct btrfs_chunk *c, int nr)
 {
 	return btrfs_stripe_offset(eb, btrfs_stripe_nr(c, nr));
 }
 
-static inline u64 btrfs_stripe_devid_nr(struct extent_buffer *eb,
+static inline u64 btrfs_stripe_devid_nr(const struct extent_buffer *eb,
 					 struct btrfs_chunk *c, int nr)
 {
 	return btrfs_stripe_devid(eb, btrfs_stripe_nr(c, nr));
@@ -1640,31 +2033,21 @@ BTRFS_SETGET_FUNCS(dev_extent_chunk_objectid, struct btrfs_dev_extent,
 BTRFS_SETGET_FUNCS(dev_extent_chunk_offset, struct btrfs_dev_extent,
 		   chunk_offset, 64);
 BTRFS_SETGET_FUNCS(dev_extent_length, struct btrfs_dev_extent, length, 64);
-
-static inline unsigned long btrfs_dev_extent_chunk_tree_uuid(struct btrfs_dev_extent *dev)
-{
-	unsigned long ptr = offsetof(struct btrfs_dev_extent, chunk_tree_uuid);
-	return (unsigned long)dev + ptr;
-}
-
 BTRFS_SETGET_FUNCS(extent_refs, struct btrfs_extent_item, refs, 64);
 BTRFS_SETGET_FUNCS(extent_generation, struct btrfs_extent_item,
 		   generation, 64);
 BTRFS_SETGET_FUNCS(extent_flags, struct btrfs_extent_item, flags, 64);
 
-BTRFS_SETGET_FUNCS(extent_refs_v0, struct btrfs_extent_item_v0, refs, 32);
-
-
 BTRFS_SETGET_FUNCS(tree_block_level, struct btrfs_tree_block_info, level, 8);
 
-static inline void btrfs_tree_block_key(struct extent_buffer *eb,
+static inline void btrfs_tree_block_key(const struct extent_buffer *eb,
 					struct btrfs_tree_block_info *item,
 					struct btrfs_disk_key *key)
 {
 	read_eb_member(eb, item, struct btrfs_tree_block_info, key, key);
 }
 
-static inline void btrfs_set_tree_block_key(struct extent_buffer *eb,
+static inline void btrfs_set_tree_block_key(const struct extent_buffer *eb,
 					    struct btrfs_tree_block_info *item,
 					    struct btrfs_disk_key *key)
 {
@@ -1702,12 +2085,6 @@ static inline u32 btrfs_extent_inline_ref_size(int type)
 	return 0;
 }
 
-BTRFS_SETGET_FUNCS(ref_root_v0, struct btrfs_extent_ref_v0, root, 64);
-BTRFS_SETGET_FUNCS(ref_generation_v0, struct btrfs_extent_ref_v0,
-		   generation, 64);
-BTRFS_SETGET_FUNCS(ref_objectid_v0, struct btrfs_extent_ref_v0, objectid, 64);
-BTRFS_SETGET_FUNCS(ref_count_v0, struct btrfs_extent_ref_v0, count, 32);
-
 /* struct btrfs_node */
 BTRFS_SETGET_FUNCS(key_blockptr, struct btrfs_key_ptr, blockptr, 64);
 BTRFS_SETGET_FUNCS(key_generation, struct btrfs_key_ptr, generation, 64);
@@ -1716,7 +2093,7 @@ BTRFS_SETGET_STACK_FUNCS(stack_key_blockptr, struct btrfs_key_ptr,
 BTRFS_SETGET_STACK_FUNCS(stack_key_generation, struct btrfs_key_ptr,
 			 generation, 64);
 
-static inline u64 btrfs_node_blockptr(struct extent_buffer *eb, int nr)
+static inline u64 btrfs_node_blockptr(const struct extent_buffer *eb, int nr)
 {
 	unsigned long ptr;
 	ptr = offsetof(struct btrfs_node, ptrs) +
@@ -1724,7 +2101,7 @@ static inline u64 btrfs_node_blockptr(struct extent_buffer *eb, int nr)
 	return btrfs_key_blockptr(eb, (struct btrfs_key_ptr *)ptr);
 }
 
-static inline void btrfs_set_node_blockptr(struct extent_buffer *eb,
+static inline void btrfs_set_node_blockptr(const struct extent_buffer *eb,
 					   int nr, u64 val)
 {
 	unsigned long ptr;
@@ -1733,7 +2110,7 @@ static inline void btrfs_set_node_blockptr(struct extent_buffer *eb,
 	btrfs_set_key_blockptr(eb, (struct btrfs_key_ptr *)ptr, val);
 }
 
-static inline u64 btrfs_node_ptr_generation(struct extent_buffer *eb, int nr)
+static inline u64 btrfs_node_ptr_generation(const struct extent_buffer *eb, int nr)
 {
 	unsigned long ptr;
 	ptr = offsetof(struct btrfs_node, ptrs) +
@@ -1741,7 +2118,7 @@ static inline u64 btrfs_node_ptr_generation(struct extent_buffer *eb, int nr)
 	return btrfs_key_generation(eb, (struct btrfs_key_ptr *)ptr);
 }
 
-static inline void btrfs_set_node_ptr_generation(struct extent_buffer *eb,
+static inline void btrfs_set_node_ptr_generation(const struct extent_buffer *eb,
 						 int nr, u64 val)
 {
 	unsigned long ptr;
@@ -1759,7 +2136,7 @@ static inline unsigned long btrfs_node_key_ptr_offset(int nr)
 void btrfs_node_key(const struct extent_buffer *eb,
 		    struct btrfs_disk_key *disk_key, int nr);
 
-static inline void btrfs_set_node_key(struct extent_buffer *eb,
+static inline void btrfs_set_node_key(const struct extent_buffer *eb,
 				      struct btrfs_disk_key *disk_key, int nr)
 {
 	unsigned long ptr;
@@ -1769,8 +2146,8 @@ static inline void btrfs_set_node_key(struct extent_buffer *eb,
 }
 
 /* struct btrfs_item */
-BTRFS_SETGET_FUNCS(item_offset, struct btrfs_item, offset, 32);
-BTRFS_SETGET_FUNCS(item_size, struct btrfs_item, size, 32);
+BTRFS_SETGET_FUNCS(raw_item_offset, struct btrfs_item, offset, 32);
+BTRFS_SETGET_FUNCS(raw_item_size, struct btrfs_item, size, 32);
 BTRFS_SETGET_STACK_FUNCS(stack_item_offset, struct btrfs_item, offset, 32);
 BTRFS_SETGET_STACK_FUNCS(stack_item_size, struct btrfs_item, size, 32);
 
@@ -1785,25 +2162,36 @@ static inline struct btrfs_item *btrfs_item_nr(int nr)
 	return (struct btrfs_item *)btrfs_item_nr_offset(nr);
 }
 
-static inline u32 btrfs_item_end(const struct extent_buffer *eb,
-				 struct btrfs_item *item)
-{
-	return btrfs_item_offset(eb, item) + btrfs_item_size(eb, item);
+#define BTRFS_ITEM_SETGET_FUNCS(member)						\
+static inline u32 btrfs_item_##member(const struct extent_buffer *eb,		\
+				      int slot)					\
+{										\
+	return btrfs_raw_item_##member(eb, btrfs_item_nr(slot));		\
+}										\
+static inline void btrfs_set_item_##member(const struct extent_buffer *eb,	\
+					   int slot, u32 val)			\
+{										\
+	btrfs_set_raw_item_##member(eb, btrfs_item_nr(slot), val);		\
+}										\
+static inline u32 btrfs_token_item_##member(struct btrfs_map_token *token,	\
+					    int slot)				\
+{										\
+	struct btrfs_item *item = btrfs_item_nr(slot);				\
+	return btrfs_token_raw_item_##member(token, item);			\
+}										\
+static inline void btrfs_set_token_item_##member(struct btrfs_map_token *token,	\
+						 int slot, u32 val)		\
+{										\
+	struct btrfs_item *item = btrfs_item_nr(slot);				\
+	btrfs_set_token_raw_item_##member(token, item, val);			\
 }
 
-static inline u32 btrfs_item_end_nr(const struct extent_buffer *eb, int nr)
-{
-	return btrfs_item_end(eb, btrfs_item_nr(nr));
-}
-
-static inline u32 btrfs_item_offset_nr(const struct extent_buffer *eb, int nr)
-{
-	return btrfs_item_offset(eb, btrfs_item_nr(nr));
-}
+BTRFS_ITEM_SETGET_FUNCS(offset)
+BTRFS_ITEM_SETGET_FUNCS(size);
 
-static inline u32 btrfs_item_size_nr(const struct extent_buffer *eb, int nr)
+static inline u32 btrfs_item_data_end(const struct extent_buffer *eb, int nr)
 {
-	return btrfs_item_size(eb, btrfs_item_nr(nr));
+	return btrfs_item_offset(eb, nr) + btrfs_item_size(eb, nr);
 }
 
 static inline void btrfs_item_key(const struct extent_buffer *eb,
@@ -1883,6 +2271,52 @@ BTRFS_SETGET_STACK_FUNCS(disk_key_objectid, struct btrfs_disk_key,
 BTRFS_SETGET_STACK_FUNCS(disk_key_offset, struct btrfs_disk_key, offset, 64);
 BTRFS_SETGET_STACK_FUNCS(disk_key_type, struct btrfs_disk_key, type, 8);
 
+#ifdef __LITTLE_ENDIAN
+
+/*
+ * Optimized helpers for little-endian architectures where CPU and on-disk
+ * structures have the same endianness and we can skip conversions.
+ */
+
+static inline void btrfs_disk_key_to_cpu(struct btrfs_key *cpu_key,
+					 const struct btrfs_disk_key *disk_key)
+{
+	memcpy(cpu_key, disk_key, sizeof(struct btrfs_key));
+}
+
+static inline void btrfs_cpu_key_to_disk(struct btrfs_disk_key *disk_key,
+					 const struct btrfs_key *cpu_key)
+{
+	memcpy(disk_key, cpu_key, sizeof(struct btrfs_key));
+}
+
+static inline void btrfs_node_key_to_cpu(const struct extent_buffer *eb,
+					 struct btrfs_key *cpu_key, int nr)
+{
+	struct btrfs_disk_key *disk_key = (struct btrfs_disk_key *)cpu_key;
+
+	btrfs_node_key(eb, disk_key, nr);
+}
+
+static inline void btrfs_item_key_to_cpu(const struct extent_buffer *eb,
+					 struct btrfs_key *cpu_key, int nr)
+{
+	struct btrfs_disk_key *disk_key = (struct btrfs_disk_key *)cpu_key;
+
+	btrfs_item_key(eb, disk_key, nr);
+}
+
+static inline void btrfs_dir_item_key_to_cpu(const struct extent_buffer *eb,
+					     const struct btrfs_dir_item *item,
+					     struct btrfs_key *cpu_key)
+{
+	struct btrfs_disk_key *disk_key = (struct btrfs_disk_key *)cpu_key;
+
+	btrfs_dir_item_key(eb, item, disk_key);
+}
+
+#else
+
 static inline void btrfs_disk_key_to_cpu(struct btrfs_key *cpu,
 					 const struct btrfs_disk_key *disk)
 {
@@ -1924,6 +2358,8 @@ static inline void btrfs_dir_item_key_to_cpu(const struct extent_buffer *eb,
 	btrfs_disk_key_to_cpu(key, &disk_key);
 }
 
+#endif
+
 /* struct btrfs_header */
 BTRFS_SETGET_HEADER_FUNCS(header_bytenr, struct btrfs_header, bytenr, 64);
 BTRFS_SETGET_HEADER_FUNCS(header_generation, struct btrfs_header,
@@ -1971,16 +2407,6 @@ static inline void btrfs_set_header_backref_rev(struct extent_buffer *eb,
 	btrfs_set_header_flags(eb, flags);
 }
 
-static inline unsigned long btrfs_header_fsid(void)
-{
-	return offsetof(struct btrfs_header, fsid);
-}
-
-static inline unsigned long btrfs_header_chunk_tree_uuid(const struct extent_buffer *eb)
-{
-	return offsetof(struct btrfs_header, chunk_tree_uuid);
-}
-
 static inline int btrfs_is_leaf(const struct extent_buffer *eb)
 {
 	return btrfs_header_level(eb) == 0;
@@ -1996,6 +2422,7 @@ BTRFS_SETGET_FUNCS(disk_root_level, struct btrfs_root_item, level, 8);
 BTRFS_SETGET_STACK_FUNCS(root_generation, struct btrfs_root_item,
 			 generation, 64);
 BTRFS_SETGET_STACK_FUNCS(root_bytenr, struct btrfs_root_item, bytenr, 64);
+BTRFS_SETGET_STACK_FUNCS(root_drop_level, struct btrfs_root_item, drop_level, 8);
 BTRFS_SETGET_STACK_FUNCS(root_level, struct btrfs_root_item, level, 8);
 BTRFS_SETGET_STACK_FUNCS(root_dirid, struct btrfs_root_item, root_dirid, 64);
 BTRFS_SETGET_STACK_FUNCS(root_refs, struct btrfs_root_item, refs, 32);
@@ -2017,14 +2444,21 @@ BTRFS_SETGET_STACK_FUNCS(root_rtransid, struct btrfs_root_item,
 
 static inline bool btrfs_root_readonly(const struct btrfs_root *root)
 {
+	/* Byte-swap the constant at compile time, root_item::flags is LE */
 	return (root->root_item.flags & cpu_to_le64(BTRFS_ROOT_SUBVOL_RDONLY)) != 0;
 }
 
 static inline bool btrfs_root_dead(const struct btrfs_root *root)
 {
+	/* Byte-swap the constant at compile time, root_item::flags is LE */
 	return (root->root_item.flags & cpu_to_le64(BTRFS_ROOT_SUBVOL_DEAD)) != 0;
 }
 
+static inline u64 btrfs_root_id(const struct btrfs_root *root)
+{
+	return root->root_key.objectid;
+}
+
 /* struct btrfs_root_backup */
 BTRFS_SETGET_STACK_FUNCS(backup_tree_root, struct btrfs_root_backup,
 		   tree_root, 64);
@@ -2177,8 +2611,6 @@ BTRFS_SETGET_STACK_FUNCS(super_chunk_root_level, struct btrfs_super_block,
 			 chunk_root_level, 8);
 BTRFS_SETGET_STACK_FUNCS(super_log_root, struct btrfs_super_block,
 			 log_root, 64);
-BTRFS_SETGET_STACK_FUNCS(super_log_root_transid, struct btrfs_super_block,
-			 log_root_transid, 64);
 BTRFS_SETGET_STACK_FUNCS(super_log_root_level, struct btrfs_super_block,
 			 log_root_level, 8);
 BTRFS_SETGET_STACK_FUNCS(super_total_bytes, struct btrfs_super_block,
@@ -2212,7 +2644,7 @@ BTRFS_SETGET_STACK_FUNCS(super_uuid_tree_generation, struct btrfs_super_block,
 int btrfs_super_csum_size(const struct btrfs_super_block *s);
 const char *btrfs_super_csum_name(u16 csum_type);
 const char *btrfs_super_csum_driver(u16 csum_type);
-size_t __const btrfs_get_num_csums(void);
+size_t __attribute_const__ btrfs_get_num_csums(void);
 
 
 /*
@@ -2226,11 +2658,12 @@ static inline unsigned int leaf_data_end(const struct extent_buffer *leaf)
 
 	if (nr == 0)
 		return BTRFS_LEAF_DATA_SIZE(leaf->fs_info);
-	return btrfs_item_offset_nr(leaf, nr - 1);
+	return btrfs_item_offset(leaf, nr - 1);
 }
 
 /* struct btrfs_file_extent_item */
-BTRFS_SETGET_FUNCS(file_extent_type, struct btrfs_file_extent_item, type, 8);
+BTRFS_SETGET_STACK_FUNCS(stack_file_extent_type, struct btrfs_file_extent_item,
+			 type, 8);
 BTRFS_SETGET_STACK_FUNCS(stack_file_extent_disk_bytenr,
 			 struct btrfs_file_extent_item, disk_bytenr, 64);
 BTRFS_SETGET_STACK_FUNCS(stack_file_extent_offset,
@@ -2239,6 +2672,8 @@ BTRFS_SETGET_STACK_FUNCS(stack_file_extent_generation,
 			 struct btrfs_file_extent_item, generation, 64);
 BTRFS_SETGET_STACK_FUNCS(stack_file_extent_num_bytes,
 			 struct btrfs_file_extent_item, num_bytes, 64);
+BTRFS_SETGET_STACK_FUNCS(stack_file_extent_ram_bytes,
+			 struct btrfs_file_extent_item, ram_bytes, 64);
 BTRFS_SETGET_STACK_FUNCS(stack_file_extent_disk_num_bytes,
 			 struct btrfs_file_extent_item, disk_num_bytes, 64);
 BTRFS_SETGET_STACK_FUNCS(stack_file_extent_compression,
@@ -2255,6 +2690,7 @@ static inline u32 btrfs_file_extent_calc_inline_size(u32 datasize)
 	return BTRFS_FILE_EXTENT_INLINE_DATA_START + datasize;
 }
 
+BTRFS_SETGET_FUNCS(file_extent_type, struct btrfs_file_extent_item, type, 8);
 BTRFS_SETGET_FUNCS(file_extent_disk_bytenr, struct btrfs_file_extent_item,
 		   disk_bytenr, 64);
 BTRFS_SETGET_FUNCS(file_extent_generation, struct btrfs_file_extent_item,
@@ -2281,9 +2717,9 @@ BTRFS_SETGET_FUNCS(file_extent_other_encoding, struct btrfs_file_extent_item,
  */
 static inline u32 btrfs_file_extent_inline_item_len(
 						const struct extent_buffer *eb,
-						struct btrfs_item *e)
+						int nr)
 {
-	return btrfs_item_size(eb, e) - BTRFS_FILE_EXTENT_INLINE_DATA_START;
+	return btrfs_item_size(eb, nr) - BTRFS_FILE_EXTENT_INLINE_DATA_START;
 }
 
 /* btrfs_qgroup_status_item */
@@ -2375,11 +2811,11 @@ BTRFS_SETGET_STACK_FUNCS(stack_dev_replace_cursor_right,
 /* helper function to cast into the data area of the leaf. */
 #define btrfs_item_ptr(leaf, slot, type) \
 	((type *)(BTRFS_LEAF_DATA_OFFSET + \
-	btrfs_item_offset_nr(leaf, slot)))
+	btrfs_item_offset(leaf, slot)))
 
 #define btrfs_item_ptr_offset(leaf, slot) \
 	((unsigned long)(BTRFS_LEAF_DATA_OFFSET + \
-	btrfs_item_offset_nr(leaf, slot)))
+	btrfs_item_offset(leaf, slot)))
 
 static inline u32 btrfs_crc32c(u32 crc, const void *address, unsigned length)
 {
@@ -2424,27 +2860,6 @@ int btrfs_get_extent_inline_ref_type(const struct extent_buffer *eb,
 				     enum btrfs_inline_ref_type is_data);
 u64 hash_extent_data_ref(u64 root_objectid, u64 owner, u64 offset);
 
-u64 btrfs_csum_bytes_to_leaves(struct btrfs_fs_info *fs_info, u64 csum_bytes);
-
-/*
- * Use this if we would be adding new items, as we could split nodes as we cow
- * down the tree.
- */
-static inline u64 btrfs_calc_insert_metadata_size(struct btrfs_fs_info *fs_info,
-						  unsigned num_items)
-{
-	return (u64)fs_info->nodesize * BTRFS_MAX_LEVEL * 2 * num_items;
-}
-
-/*
- * Doing a truncate or a modification won't result in new nodes or leaves, just
- * what we need for COW.
- */
-static inline u64 btrfs_calc_metadata_size(struct btrfs_fs_info *fs_info,
-						 unsigned num_items)
-{
-	return (u64)fs_info->nodesize * BTRFS_MAX_LEVEL * num_items;
-}
 
 int btrfs_add_excluded_extent(struct btrfs_fs_info *fs_info,
 			      u64 start, u64 num_bytes);
@@ -2458,21 +2873,23 @@ int btrfs_lookup_data_extent(struct btrfs_fs_info *fs_info, u64 start, u64 len);
 int btrfs_lookup_extent_info(struct btrfs_trans_handle *trans,
 			     struct btrfs_fs_info *fs_info, u64 bytenr,
 			     u64 offset, int metadata, u64 *refs, u64 *flags);
-int btrfs_pin_extent(struct btrfs_fs_info *fs_info,
-		     u64 bytenr, u64 num, int reserved);
-int btrfs_pin_extent_for_log_replay(struct btrfs_fs_info *fs_info,
+int btrfs_pin_extent(struct btrfs_trans_handle *trans, u64 bytenr, u64 num,
+		     int reserved);
+int btrfs_pin_extent_for_log_replay(struct btrfs_trans_handle *trans,
 				    u64 bytenr, u64 num_bytes);
 int btrfs_exclude_logged_extents(struct extent_buffer *eb);
 int btrfs_cross_ref_exist(struct btrfs_root *root,
-			  u64 objectid, u64 offset, u64 bytenr);
+			  u64 objectid, u64 offset, u64 bytenr, bool strict,
+			  struct btrfs_path *path);
 struct extent_buffer *btrfs_alloc_tree_block(struct btrfs_trans_handle *trans,
 					     struct btrfs_root *root,
 					     u64 parent, u64 root_objectid,
 					     const struct btrfs_disk_key *key,
 					     int level, u64 hint,
-					     u64 empty_size);
+					     u64 empty_size,
+					     enum btrfs_lock_nesting nest);
 void btrfs_free_tree_block(struct btrfs_trans_handle *trans,
-			   struct btrfs_root *root,
+			   u64 root_id,
 			   struct extent_buffer *buf,
 			   u64 parent, int last_ref);
 int btrfs_alloc_reserved_file_extent(struct btrfs_trans_handle *trans,
@@ -2490,34 +2907,62 @@ int btrfs_inc_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 int btrfs_dec_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 		  struct extent_buffer *buf, int full_backref);
 int btrfs_set_disk_extent_flags(struct btrfs_trans_handle *trans,
-				u64 bytenr, u64 num_bytes, u64 flags,
-				int level, int is_data);
+				struct extent_buffer *eb, u64 flags, int level);
 int btrfs_free_extent(struct btrfs_trans_handle *trans, struct btrfs_ref *ref);
 
 int btrfs_free_reserved_extent(struct btrfs_fs_info *fs_info,
 			       u64 start, u64 len, int delalloc);
-int btrfs_pin_reserved_extent(struct btrfs_fs_info *fs_info, u64 start,
+int btrfs_pin_reserved_extent(struct btrfs_trans_handle *trans, u64 start,
 			      u64 len);
-void btrfs_prepare_extent_commit(struct btrfs_fs_info *fs_info);
 int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans);
 int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
 			 struct btrfs_ref *generic_ref);
 
-int btrfs_extent_readonly(struct btrfs_fs_info *fs_info, u64 bytenr);
-void btrfs_get_block_group_trimming(struct btrfs_block_group *cache);
-void btrfs_put_block_group_trimming(struct btrfs_block_group *cache);
 void btrfs_clear_space_info_full(struct btrfs_fs_info *info);
 
+/*
+ * Different levels for to flush space when doing space reservations.
+ *
+ * The higher the level, the more methods we try to reclaim space.
+ */
 enum btrfs_reserve_flush_enum {
 	/* If we are in the transaction, we can't flush anything.*/
 	BTRFS_RESERVE_NO_FLUSH,
+
 	/*
-	 * Flushing delalloc may cause deadlock somewhere, in this
-	 * case, use FLUSH LIMIT
+	 * Flush space by:
+	 * - Running delayed inode items
+	 * - Allocating a new chunk
 	 */
 	BTRFS_RESERVE_FLUSH_LIMIT,
+
+	/*
+	 * Flush space by:
+	 * - Running delayed inode items
+	 * - Running delayed refs
+	 * - Running delalloc and waiting for ordered extents
+	 * - Allocating a new chunk
+	 */
 	BTRFS_RESERVE_FLUSH_EVICT,
+
+	/*
+	 * Flush space by above mentioned methods and by:
+	 * - Running delayed iputs
+	 * - Committing transaction
+	 *
+	 * Can be interrupted by a fatal signal.
+	 */
+	BTRFS_RESERVE_FLUSH_DATA,
+	BTRFS_RESERVE_FLUSH_FREE_SPACE_INODE,
 	BTRFS_RESERVE_FLUSH_ALL,
+
+	/*
+	 * Pretty much the same as FLUSH_ALL, but can also steal space from
+	 * global rsv.
+	 *
+	 * Can be interrupted by a fatal signal.
+	 */
+	BTRFS_RESERVE_FLUSH_ALL_STEAL,
 };
 
 enum btrfs_flush_state {
@@ -2527,20 +2972,22 @@ enum btrfs_flush_state {
 	FLUSH_DELAYED_REFS	=	4,
 	FLUSH_DELALLOC		=	5,
 	FLUSH_DELALLOC_WAIT	=	6,
-	ALLOC_CHUNK		=	7,
-	ALLOC_CHUNK_FORCE	=	8,
-	RUN_DELAYED_IPUTS	=	9,
-	COMMIT_TRANS		=	10,
+	FLUSH_DELALLOC_FULL	=	7,
+	ALLOC_CHUNK		=	8,
+	ALLOC_CHUNK_FORCE	=	9,
+	RUN_DELAYED_IPUTS	=	10,
+	COMMIT_TRANS		=	11,
 };
 
 int btrfs_subvolume_reserve_metadata(struct btrfs_root *root,
 				     struct btrfs_block_rsv *rsv,
 				     int nitems, bool use_global_rsv);
-void btrfs_subvolume_release_metadata(struct btrfs_fs_info *fs_info,
+void btrfs_subvolume_release_metadata(struct btrfs_root *root,
 				      struct btrfs_block_rsv *rsv);
 void btrfs_delalloc_release_extents(struct btrfs_inode *inode, u64 num_bytes);
 
-int btrfs_delalloc_reserve_metadata(struct btrfs_inode *inode, u64 num_bytes);
+int btrfs_delalloc_reserve_metadata(struct btrfs_inode *inode, u64 num_bytes,
+				    u64 disk_num_bytes, bool noflush);
 u64 btrfs_account_ro_block_groups_free_space(struct btrfs_space_info *sinfo);
 int btrfs_error_unpin_extent_range(struct btrfs_fs_info *fs_info,
 				   u64 start, u64 end);
@@ -2557,7 +3004,7 @@ void btrfs_wait_for_snapshot_creation(struct btrfs_root *root);
 
 /* ctree.c */
 int btrfs_bin_search(struct extent_buffer *eb, const struct btrfs_key *key,
-		     int level, int *slot);
+		     int *slot);
 int __pure btrfs_comp_cpu_keys(const struct btrfs_key *k1, const struct btrfs_key *k2);
 int btrfs_previous_item(struct btrfs_root *root,
 			struct btrfs_path *path, u64 min_objectid,
@@ -2568,8 +3015,6 @@ void btrfs_set_item_key_safe(struct btrfs_fs_info *fs_info,
 			     struct btrfs_path *path,
 			     const struct btrfs_key *new_key);
 struct extent_buffer *btrfs_root_node(struct btrfs_root *root);
-struct extent_buffer *btrfs_lock_root_node(struct btrfs_root *root);
-struct extent_buffer *btrfs_read_lock_root_node(struct btrfs_root *root);
 int btrfs_find_next_key(struct btrfs_root *root, struct btrfs_path *path,
 			struct btrfs_key *key, int lowest_level,
 			u64 min_trans);
@@ -2582,7 +3027,8 @@ struct extent_buffer *btrfs_read_node_slot(struct extent_buffer *parent,
 int btrfs_cow_block(struct btrfs_trans_handle *trans,
 		    struct btrfs_root *root, struct extent_buffer *buf,
 		    struct extent_buffer *parent, int parent_slot,
-		    struct extent_buffer **cow_ret);
+		    struct extent_buffer **cow_ret,
+		    enum btrfs_lock_nesting nest);
 int btrfs_copy_root(struct btrfs_trans_handle *trans,
 		      struct btrfs_root *root,
 		      struct extent_buffer *buf,
@@ -2628,16 +3074,42 @@ static inline int btrfs_del_item(struct btrfs_trans_handle *trans,
 	return btrfs_del_items(trans, root, path, path->slots[0], 1);
 }
 
-void setup_items_for_insert(struct btrfs_root *root, struct btrfs_path *path,
-			    const struct btrfs_key *cpu_key, u32 *data_size,
-			    u32 total_data, u32 total_size, int nr);
+/*
+ * Describes a batch of items to insert in a btree. This is used by
+ * btrfs_insert_empty_items().
+ */
+struct btrfs_item_batch {
+	/*
+	 * Pointer to an array containing the keys of the items to insert (in
+	 * sorted order).
+	 */
+	const struct btrfs_key *keys;
+	/* Pointer to an array containing the data size for each item to insert. */
+	const u32 *data_sizes;
+	/*
+	 * The sum of data sizes for all items. The caller can compute this while
+	 * setting up the data_sizes array, so it ends up being more efficient
+	 * than having btrfs_insert_empty_items() or setup_item_for_insert()
+	 * doing it, as it would avoid an extra loop over a potentially large
+	 * array, and in the case of setup_item_for_insert(), we would be doing
+	 * it while holding a write lock on a leaf and often on upper level nodes
+	 * too, unnecessarily increasing the size of a critical section.
+	 */
+	u32 total_data_size;
+	/* Size of the keys and data_sizes arrays (number of items in the batch). */
+	int nr;
+};
+
+void btrfs_setup_item_for_insert(struct btrfs_root *root,
+				 struct btrfs_path *path,
+				 const struct btrfs_key *key,
+				 u32 data_size);
 int btrfs_insert_item(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 		      const struct btrfs_key *key, void *data, u32 data_size);
 int btrfs_insert_empty_items(struct btrfs_trans_handle *trans,
 			     struct btrfs_root *root,
 			     struct btrfs_path *path,
-			     const struct btrfs_key *cpu_key, u32 *data_size,
-			     int nr);
+			     const struct btrfs_item_batch *batch);
 
 static inline int btrfs_insert_empty_item(struct btrfs_trans_handle *trans,
 					  struct btrfs_root *root,
@@ -2645,13 +3117,52 @@ static inline int btrfs_insert_empty_item(struct btrfs_trans_handle *trans,
 					  const struct btrfs_key *key,
 					  u32 data_size)
 {
-	return btrfs_insert_empty_items(trans, root, path, key, &data_size, 1);
+	struct btrfs_item_batch batch;
+
+	batch.keys = key;
+	batch.data_sizes = &data_size;
+	batch.total_data_size = data_size;
+	batch.nr = 1;
+
+	return btrfs_insert_empty_items(trans, root, path, &batch);
 }
 
-int btrfs_next_leaf(struct btrfs_root *root, struct btrfs_path *path);
 int btrfs_prev_leaf(struct btrfs_root *root, struct btrfs_path *path);
 int btrfs_next_old_leaf(struct btrfs_root *root, struct btrfs_path *path,
 			u64 time_seq);
+
+int btrfs_search_backwards(struct btrfs_root *root, struct btrfs_key *key,
+			   struct btrfs_path *path);
+
+int btrfs_get_next_valid_item(struct btrfs_root *root, struct btrfs_key *key,
+			      struct btrfs_path *path);
+
+/*
+ * Search in @root for a given @key, and store the slot found in @found_key.
+ *
+ * @root:	The root node of the tree.
+ * @key:	The key we are looking for.
+ * @found_key:	Will hold the found item.
+ * @path:	Holds the current slot/leaf.
+ * @iter_ret:	Contains the value returned from btrfs_search_slot or
+ * 		btrfs_get_next_valid_item, whichever was executed last.
+ *
+ * The @iter_ret is an output variable that will contain the return value of
+ * btrfs_search_slot, if it encountered an error, or the value returned from
+ * btrfs_get_next_valid_item otherwise. That return value can be 0, if a valid
+ * slot was found, 1 if there were no more leaves, and <0 if there was an error.
+ *
+ * It's recommended to use a separate variable for iter_ret and then use it to
+ * set the function return value so there's no confusion of the 0/1/errno
+ * values stemming from btrfs_search_slot.
+ */
+#define btrfs_for_each_slot(root, key, found_key, path, iter_ret)		\
+	for (iter_ret = btrfs_search_slot(NULL, (root), (key), (path), 0, 0);	\
+		(iter_ret) >= 0 &&						\
+		(iter_ret = btrfs_get_next_valid_item((root), (found_key), (path))) == 0; \
+		(path)->slots[0]++						\
+	)
+
 static inline int btrfs_next_old_item(struct btrfs_root *root,
 				      struct btrfs_path *p, u64 time_seq)
 {
@@ -2660,14 +3171,25 @@ static inline int btrfs_next_old_item(struct btrfs_root *root,
 		return btrfs_next_old_leaf(root, p, time_seq);
 	return 0;
 }
+
+/*
+ * Search the tree again to find a leaf with greater keys.
+ *
+ * Returns 0 if it found something or 1 if there are no greater leaves.
+ * Returns < 0 on error.
+ */
+static inline int btrfs_next_leaf(struct btrfs_root *root, struct btrfs_path *path)
+{
+	return btrfs_next_old_leaf(root, path, 0);
+}
+
 static inline int btrfs_next_item(struct btrfs_root *root, struct btrfs_path *p)
 {
 	return btrfs_next_old_item(root, p, 0);
 }
 int btrfs_leaf_free_space(struct extent_buffer *leaf);
-int __must_check btrfs_drop_snapshot(struct btrfs_root *root,
-				     struct btrfs_block_rsv *block_rsv,
-				     int update_ref, int for_reloc);
+int __must_check btrfs_drop_snapshot(struct btrfs_root *root, int update_ref,
+				     int for_reloc);
 int btrfs_drop_subtree(struct btrfs_trans_handle *trans,
 			struct btrfs_root *root,
 			struct extent_buffer *node,
@@ -2689,35 +3211,27 @@ static inline int btrfs_fs_closing(struct btrfs_fs_info *fs_info)
  * If we remount the fs to be R/O or umount the fs, the cleaner needn't do
  * anything except sleeping. This function is used to check the status of
  * the fs.
+ * We check for BTRFS_FS_STATE_RO to avoid races with a concurrent remount,
+ * since setting and checking for SB_RDONLY in the superblock's flags is not
+ * atomic.
  */
 static inline int btrfs_need_cleaner_sleep(struct btrfs_fs_info *fs_info)
 {
-	return fs_info->sb->s_flags & SB_RDONLY || btrfs_fs_closing(fs_info);
+	return test_bit(BTRFS_FS_STATE_RO, &fs_info->fs_state) ||
+		btrfs_fs_closing(fs_info);
 }
 
-static inline void free_fs_info(struct btrfs_fs_info *fs_info)
+static inline void btrfs_set_sb_rdonly(struct super_block *sb)
 {
-	kfree(fs_info->balance_ctl);
-	kfree(fs_info->delayed_root);
-	kfree(fs_info->extent_root);
-	kfree(fs_info->tree_root);
-	kfree(fs_info->chunk_root);
-	kfree(fs_info->dev_root);
-	kfree(fs_info->csum_root);
-	kfree(fs_info->quota_root);
-	kfree(fs_info->uuid_root);
-	kfree(fs_info->free_space_root);
-	kfree(fs_info->super_copy);
-	kfree(fs_info->super_for_commit);
-	kvfree(fs_info);
+	sb->s_flags |= SB_RDONLY;
+	set_bit(BTRFS_FS_STATE_RO, &btrfs_sb(sb)->fs_state);
 }
 
-/* tree mod log functions from ctree.c */
-u64 btrfs_get_tree_mod_seq(struct btrfs_fs_info *fs_info,
-			   struct seq_list *elem);
-void btrfs_put_tree_mod_seq(struct btrfs_fs_info *fs_info,
-			    struct seq_list *elem);
-int btrfs_old_root_level(struct btrfs_root *root, u64 time_seq);
+static inline void btrfs_clear_sb_rdonly(struct super_block *sb)
+{
+	sb->s_flags &= ~SB_RDONLY;
+	clear_bit(BTRFS_FS_STATE_RO, &btrfs_sb(sb)->fs_state);
+}
 
 /* root-item.c */
 int btrfs_add_root_ref(struct btrfs_trans_handle *trans, u64 root_id,
@@ -2750,9 +3264,7 @@ int btrfs_uuid_tree_add(struct btrfs_trans_handle *trans, u8 *uuid, u8 type,
 			u64 subid);
 int btrfs_uuid_tree_remove(struct btrfs_trans_handle *trans, u8 *uuid, u8 type,
 			u64 subid);
-int btrfs_uuid_tree_iterate(struct btrfs_fs_info *fs_info,
-			    int (*check_func)(struct btrfs_fs_info *, u8 *, u8,
-					      u64));
+int btrfs_uuid_tree_iterate(struct btrfs_fs_info *fs_info);
 
 /* dir-item.c */
 int btrfs_check_dir_item_collision(struct btrfs_root *root, u64 dir,
@@ -2769,7 +3281,7 @@ struct btrfs_dir_item *
 btrfs_lookup_dir_index_item(struct btrfs_trans_handle *trans,
 			    struct btrfs_root *root,
 			    struct btrfs_path *path, u64 dir,
-			    u64 objectid, const char *name, int name_len,
+			    u64 index, const char *name, int name_len,
 			    int mod);
 struct btrfs_dir_item *
 btrfs_search_dir_index_item(struct btrfs_root *root,
@@ -2801,48 +3313,13 @@ int btrfs_del_orphan_item(struct btrfs_trans_handle *trans,
 			  struct btrfs_root *root, u64 offset);
 int btrfs_find_orphan_item(struct btrfs_root *root, u64 offset);
 
-/* inode-item.c */
-int btrfs_insert_inode_ref(struct btrfs_trans_handle *trans,
-			   struct btrfs_root *root,
-			   const char *name, int name_len,
-			   u64 inode_objectid, u64 ref_objectid, u64 index);
-int btrfs_del_inode_ref(struct btrfs_trans_handle *trans,
-			   struct btrfs_root *root,
-			   const char *name, int name_len,
-			   u64 inode_objectid, u64 ref_objectid, u64 *index);
-int btrfs_insert_empty_inode(struct btrfs_trans_handle *trans,
-			     struct btrfs_root *root,
-			     struct btrfs_path *path, u64 objectid);
-int btrfs_lookup_inode(struct btrfs_trans_handle *trans, struct btrfs_root
-		       *root, struct btrfs_path *path,
-		       struct btrfs_key *location, int mod);
-
-struct btrfs_inode_extref *
-btrfs_lookup_inode_extref(struct btrfs_trans_handle *trans,
-			  struct btrfs_root *root,
-			  struct btrfs_path *path,
-			  const char *name, int name_len,
-			  u64 inode_objectid, u64 ref_objectid, int ins_len,
-			  int cow);
-
-struct btrfs_inode_ref *btrfs_find_name_in_backref(struct extent_buffer *leaf,
-						   int slot, const char *name,
-						   int name_len);
-struct btrfs_inode_extref *btrfs_find_name_in_ext_backref(
-		struct extent_buffer *leaf, int slot, u64 ref_objectid,
-		const char *name, int name_len);
 /* file-item.c */
-struct btrfs_dio_private;
 int btrfs_del_csums(struct btrfs_trans_handle *trans,
 		    struct btrfs_root *root, u64 bytenr, u64 len);
-blk_status_t btrfs_lookup_bio_sums(struct inode *inode, struct bio *bio,
-				   u64 offset, u8 *dst);
-int btrfs_insert_file_extent(struct btrfs_trans_handle *trans,
-			     struct btrfs_root *root,
-			     u64 objectid, u64 pos,
-			     u64 disk_offset, u64 disk_num_bytes,
-			     u64 num_bytes, u64 offset, u64 ram_bytes,
-			     u8 compression, u8 encryption, u16 other_encoding);
+blk_status_t btrfs_lookup_bio_sums(struct inode *inode, struct bio *bio, u8 *dst);
+int btrfs_insert_hole_extent(struct btrfs_trans_handle *trans,
+			     struct btrfs_root *root, u64 objectid, u64 pos,
+			     u64 num_bytes);
 int btrfs_lookup_file_extent(struct btrfs_trans_handle *trans,
 			     struct btrfs_root *root,
 			     struct btrfs_path *path, u64 objectid,
@@ -2850,88 +3327,113 @@ int btrfs_lookup_file_extent(struct btrfs_trans_handle *trans,
 int btrfs_csum_file_blocks(struct btrfs_trans_handle *trans,
 			   struct btrfs_root *root,
 			   struct btrfs_ordered_sum *sums);
-blk_status_t btrfs_csum_one_bio(struct inode *inode, struct bio *bio,
-		       u64 file_start, int contig);
+blk_status_t btrfs_csum_one_bio(struct btrfs_inode *inode, struct bio *bio,
+				u64 offset, bool one_ordered);
 int btrfs_lookup_csums_range(struct btrfs_root *root, u64 start, u64 end,
-			     struct list_head *list, int search_commit);
+			     struct list_head *list, int search_commit,
+			     bool nowait);
 void btrfs_extent_item_to_extent_map(struct btrfs_inode *inode,
 				     const struct btrfs_path *path,
 				     struct btrfs_file_extent_item *fi,
 				     const bool new_inline,
 				     struct extent_map *em);
+int btrfs_inode_clear_file_extent_range(struct btrfs_inode *inode, u64 start,
+					u64 len);
+int btrfs_inode_set_file_extent_range(struct btrfs_inode *inode, u64 start,
+				      u64 len);
+void btrfs_inode_safe_disk_i_size_write(struct btrfs_inode *inode, u64 new_i_size);
+u64 btrfs_file_extent_end(const struct btrfs_path *path);
 
 /* inode.c */
-struct extent_map *btrfs_get_extent_fiemap(struct btrfs_inode *inode,
-					   u64 start, u64 len);
+void btrfs_submit_data_write_bio(struct inode *inode, struct bio *bio, int mirror_num);
+void btrfs_submit_data_read_bio(struct inode *inode, struct bio *bio,
+			int mirror_num, enum btrfs_compression_type compress_type);
+int btrfs_check_sector_csum(struct btrfs_fs_info *fs_info, struct page *page,
+			    u32 pgoff, u8 *csum, const u8 * const csum_expected);
+int btrfs_check_data_csum(struct inode *inode, struct btrfs_bio *bbio,
+			  u32 bio_offset, struct page *page, u32 pgoff);
+unsigned int btrfs_verify_data_csum(struct btrfs_bio *bbio,
+				    u32 bio_offset, struct page *page,
+				    u64 start, u64 end);
+int btrfs_check_data_csum(struct inode *inode, struct btrfs_bio *bbio,
+			  u32 bio_offset, struct page *page, u32 pgoff);
 noinline int can_nocow_extent(struct inode *inode, u64 offset, u64 *len,
 			      u64 *orig_start, u64 *orig_block_len,
-			      u64 *ram_bytes);
+			      u64 *ram_bytes, bool nowait, bool strict);
 
 void __btrfs_del_delalloc_inode(struct btrfs_root *root,
 				struct btrfs_inode *inode);
 struct inode *btrfs_lookup_dentry(struct inode *dir, struct dentry *dentry);
 int btrfs_set_inode_index(struct btrfs_inode *dir, u64 *index);
 int btrfs_unlink_inode(struct btrfs_trans_handle *trans,
-		       struct btrfs_root *root,
 		       struct btrfs_inode *dir, struct btrfs_inode *inode,
 		       const char *name, int name_len);
 int btrfs_add_link(struct btrfs_trans_handle *trans,
 		   struct btrfs_inode *parent_inode, struct btrfs_inode *inode,
 		   const char *name, int name_len, int add_backref, u64 index);
 int btrfs_delete_subvolume(struct inode *dir, struct dentry *dentry);
-int btrfs_truncate_block(struct inode *inode, loff_t from, loff_t len,
-			int front);
-int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans,
-			       struct btrfs_root *root,
-			       struct inode *inode, u64 new_size,
-			       u32 min_type);
-
-int btrfs_start_delalloc_snapshot(struct btrfs_root *root);
-int btrfs_start_delalloc_roots(struct btrfs_fs_info *fs_info, int nr);
-int btrfs_set_extent_delalloc(struct inode *inode, u64 start, u64 end,
+int btrfs_truncate_block(struct btrfs_inode *inode, loff_t from, loff_t len,
+			 int front);
+
+int btrfs_start_delalloc_snapshot(struct btrfs_root *root, bool in_reclaim_context);
+int btrfs_start_delalloc_roots(struct btrfs_fs_info *fs_info, long nr,
+			       bool in_reclaim_context);
+int btrfs_set_extent_delalloc(struct btrfs_inode *inode, u64 start, u64 end,
 			      unsigned int extra_bits,
 			      struct extent_state **cached_state);
-int btrfs_create_subvol_root(struct btrfs_trans_handle *trans,
-			     struct btrfs_root *new_root,
-			     struct btrfs_root *parent_root,
-			     u64 new_dirid);
+struct btrfs_new_inode_args {
+	/* Input */
+	struct inode *dir;
+	struct dentry *dentry;
+	struct inode *inode;
+	bool orphan;
+	bool subvol;
+
+	/*
+	 * Output from btrfs_new_inode_prepare(), input to
+	 * btrfs_create_new_inode().
+	 */
+	struct posix_acl *default_acl;
+	struct posix_acl *acl;
+};
+int btrfs_new_inode_prepare(struct btrfs_new_inode_args *args,
+			    unsigned int *trans_num_items);
+int btrfs_create_new_inode(struct btrfs_trans_handle *trans,
+			   struct btrfs_new_inode_args *args);
+void btrfs_new_inode_args_destroy(struct btrfs_new_inode_args *args);
+struct inode *btrfs_new_subvol_inode(struct user_namespace *mnt_userns,
+				     struct inode *dir);
  void btrfs_set_delalloc_extent(struct inode *inode, struct extent_state *state,
-			       unsigned *bits);
+			        u32 bits);
 void btrfs_clear_delalloc_extent(struct inode *inode,
-				 struct extent_state *state, unsigned *bits);
+				 struct extent_state *state, u32 bits);
 void btrfs_merge_delalloc_extent(struct inode *inode, struct extent_state *new,
 				 struct extent_state *other);
 void btrfs_split_delalloc_extent(struct inode *inode,
 				 struct extent_state *orig, u64 split);
-int btrfs_bio_fits_in_stripe(struct page *page, size_t size, struct bio *bio,
-			     unsigned long bio_flags);
-void btrfs_set_range_writeback(struct extent_io_tree *tree, u64 start, u64 end);
+void btrfs_set_range_writeback(struct btrfs_inode *inode, u64 start, u64 end);
 vm_fault_t btrfs_page_mkwrite(struct vm_fault *vmf);
-int btrfs_readpage(struct file *file, struct page *page);
 void btrfs_evict_inode(struct inode *inode);
-int btrfs_write_inode(struct inode *inode, struct writeback_control *wbc);
 struct inode *btrfs_alloc_inode(struct super_block *sb);
 void btrfs_destroy_inode(struct inode *inode);
 void btrfs_free_inode(struct inode *inode);
 int btrfs_drop_inode(struct inode *inode);
 int __init btrfs_init_cachep(void);
 void __cold btrfs_destroy_cachep(void);
-struct inode *btrfs_iget_path(struct super_block *s, struct btrfs_key *location,
+struct inode *btrfs_iget_path(struct super_block *s, u64 ino,
 			      struct btrfs_root *root, struct btrfs_path *path);
-struct inode *btrfs_iget(struct super_block *s, struct btrfs_key *location,
-			 struct btrfs_root *root);
+struct inode *btrfs_iget(struct super_block *s, u64 ino, struct btrfs_root *root);
 struct extent_map *btrfs_get_extent(struct btrfs_inode *inode,
 				    struct page *page, size_t pg_offset,
 				    u64 start, u64 end);
 int btrfs_update_inode(struct btrfs_trans_handle *trans,
-			      struct btrfs_root *root,
-			      struct inode *inode);
+		       struct btrfs_root *root, struct btrfs_inode *inode);
 int btrfs_update_inode_fallback(struct btrfs_trans_handle *trans,
-				struct btrfs_root *root, struct inode *inode);
+				struct btrfs_root *root, struct btrfs_inode *inode);
 int btrfs_orphan_add(struct btrfs_trans_handle *trans,
 		struct btrfs_inode *inode);
 int btrfs_orphan_cleanup(struct btrfs_root *root);
-int btrfs_cont_expand(struct inode *inode, loff_t oldsize, loff_t size);
+int btrfs_cont_expand(struct btrfs_inode *inode, loff_t oldsize, loff_t size);
 void btrfs_add_delayed_iput(struct inode *inode);
 void btrfs_run_delayed_iputs(struct btrfs_fs_info *fs_info);
 int btrfs_wait_on_delayed_iputs(struct btrfs_fs_info *fs_info);
@@ -2942,23 +3444,55 @@ int btrfs_prealloc_file_range_trans(struct inode *inode,
 				    struct btrfs_trans_handle *trans, int mode,
 				    u64 start, u64 num_bytes, u64 min_size,
 				    loff_t actual_len, u64 *alloc_hint);
-int btrfs_run_delalloc_range(struct inode *inode, struct page *locked_page,
+int btrfs_run_delalloc_range(struct btrfs_inode *inode, struct page *locked_page,
 		u64 start, u64 end, int *page_started, unsigned long *nr_written,
 		struct writeback_control *wbc);
-int btrfs_writepage_cow_fixup(struct page *page, u64 start, u64 end);
-void btrfs_writepage_endio_finish_ordered(struct page *page, u64 start,
-					  u64 end, int uptodate);
+int btrfs_writepage_cow_fixup(struct page *page);
+void btrfs_writepage_endio_finish_ordered(struct btrfs_inode *inode,
+					  struct page *page, u64 start,
+					  u64 end, bool uptodate);
+int btrfs_encoded_io_compression_from_extent(struct btrfs_fs_info *fs_info,
+					     int compress_type);
+int btrfs_encoded_read_regular_fill_pages(struct btrfs_inode *inode,
+					  u64 file_offset, u64 disk_bytenr,
+					  u64 disk_io_size,
+					  struct page **pages);
+ssize_t btrfs_encoded_read(struct kiocb *iocb, struct iov_iter *iter,
+			   struct btrfs_ioctl_encoded_io_args *encoded);
+ssize_t btrfs_do_encoded_write(struct kiocb *iocb, struct iov_iter *from,
+			     const struct btrfs_ioctl_encoded_io_args *encoded);
+
+ssize_t btrfs_dio_read(struct kiocb *iocb, struct iov_iter *iter,
+		       size_t done_before);
+struct iomap_dio *btrfs_dio_write(struct kiocb *iocb, struct iov_iter *iter,
+				  size_t done_before);
+
 extern const struct dentry_operations btrfs_dentry_operations;
 
+/* Inode locking type flags, by default the exclusive lock is taken */
+#define BTRFS_ILOCK_SHARED	(1U << 0)
+#define BTRFS_ILOCK_TRY 	(1U << 1)
+#define BTRFS_ILOCK_MMAP	(1U << 2)
+
+int btrfs_inode_lock(struct inode *inode, unsigned int ilock_flags);
+void btrfs_inode_unlock(struct inode *inode, unsigned int ilock_flags);
+void btrfs_update_inode_bytes(struct btrfs_inode *inode,
+			      const u64 add_bytes,
+			      const u64 del_bytes);
+void btrfs_assert_inode_range_clean(struct btrfs_inode *inode, u64 start, u64 end);
+
 /* ioctl.c */
 long btrfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
 long btrfs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
+int btrfs_fileattr_get(struct dentry *dentry, struct fileattr *fa);
+int btrfs_fileattr_set(struct user_namespace *mnt_userns,
+		       struct dentry *dentry, struct fileattr *fa);
 int btrfs_ioctl_get_supported_features(void __user *arg);
 void btrfs_sync_inode_flags_to_i_flags(struct inode *inode);
 int __pure btrfs_is_empty_uuid(u8 *uuid);
-int btrfs_defrag_file(struct inode *inode, struct file *file,
+int btrfs_defrag_file(struct inode *inode, struct file_ra_state *ra,
 		      struct btrfs_ioctl_defrag_range_args *range,
-		      u64 newer_than, unsigned long max_pages);
+		      u64 newer_than, unsigned long max_to_defrag);
 void btrfs_get_block_group_info(struct list_head *groups_list,
 				struct btrfs_ioctl_space_info *space);
 void btrfs_update_ioctl_balance_args(struct btrfs_fs_info *fs_info,
@@ -2968,37 +3502,33 @@ void btrfs_update_ioctl_balance_args(struct btrfs_fs_info *fs_info,
 int __init btrfs_auto_defrag_init(void);
 void __cold btrfs_auto_defrag_exit(void);
 int btrfs_add_inode_defrag(struct btrfs_trans_handle *trans,
-			   struct btrfs_inode *inode);
+			   struct btrfs_inode *inode, u32 extent_thresh);
 int btrfs_run_defrag_inodes(struct btrfs_fs_info *fs_info);
 void btrfs_cleanup_defrag_inodes(struct btrfs_fs_info *fs_info);
 int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync);
-void btrfs_drop_extent_cache(struct btrfs_inode *inode, u64 start, u64 end,
-			     int skip_pinned);
 extern const struct file_operations btrfs_file_operations;
-int __btrfs_drop_extents(struct btrfs_trans_handle *trans,
-			 struct btrfs_root *root, struct inode *inode,
-			 struct btrfs_path *path, u64 start, u64 end,
-			 u64 *drop_end, int drop_cache,
-			 int replace_extent,
-			 u32 extent_item_size,
-			 int *key_inserted);
 int btrfs_drop_extents(struct btrfs_trans_handle *trans,
-		       struct btrfs_root *root, struct inode *inode, u64 start,
-		       u64 end, int drop_cache);
-int btrfs_punch_hole_range(struct inode *inode, struct btrfs_path *path,
-			   const u64 start, const u64 end,
-			   struct btrfs_clone_extent_info *clone_info,
+		       struct btrfs_root *root, struct btrfs_inode *inode,
+		       struct btrfs_drop_extents_args *args);
+int btrfs_replace_file_extents(struct btrfs_inode *inode,
+			   struct btrfs_path *path, const u64 start,
+			   const u64 end,
+			   struct btrfs_replace_extent_info *extent_info,
 			   struct btrfs_trans_handle **trans_out);
 int btrfs_mark_extent_written(struct btrfs_trans_handle *trans,
 			      struct btrfs_inode *inode, u64 start, u64 end);
+ssize_t btrfs_do_write_iter(struct kiocb *iocb, struct iov_iter *from,
+			    const struct btrfs_ioctl_encoded_io_args *encoded);
 int btrfs_release_file(struct inode *inode, struct file *file);
-int btrfs_dirty_pages(struct inode *inode, struct page **pages,
+int btrfs_dirty_pages(struct btrfs_inode *inode, struct page **pages,
 		      size_t num_pages, loff_t pos, size_t write_bytes,
-		      struct extent_state **cached);
+		      struct extent_state **cached, bool noreserve);
 int btrfs_fdatawrite_range(struct inode *inode, loff_t start, loff_t end);
-loff_t btrfs_remap_file_range(struct file *file_in, loff_t pos_in,
-			      struct file *file_out, loff_t pos_out,
-			      loff_t len, unsigned int remap_flags);
+int btrfs_check_nocow_lock(struct btrfs_inode *inode, loff_t pos,
+			   size_t *write_bytes, bool nowait);
+void btrfs_check_nocow_unlock(struct btrfs_inode *inode);
+bool btrfs_find_delalloc_in_range(struct btrfs_inode *inode, u64 start, u64 end,
+				  u64 *delalloc_start_ret, u64 *delalloc_end_ret);
 
 /* tree-defrag.c */
 int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
@@ -3008,17 +3538,37 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 int btrfs_parse_options(struct btrfs_fs_info *info, char *options,
 			unsigned long new_flags);
 int btrfs_sync_fs(struct super_block *sb, int wait);
+char *btrfs_get_subvol_name_from_objectid(struct btrfs_fs_info *fs_info,
+					  u64 subvol_objectid);
 
 static inline __printf(2, 3) __cold
 void btrfs_no_printk(const struct btrfs_fs_info *fs_info, const char *fmt, ...)
 {
 }
 
-#ifdef CONFIG_PRINTK
+#ifdef CONFIG_PRINTK_INDEX
+
+#define btrfs_printk(fs_info, fmt, args...)					\
+do {										\
+	printk_index_subsys_emit("%sBTRFS %s (device %s): ", NULL, fmt);	\
+	_btrfs_printk(fs_info, fmt, ##args);					\
+} while (0)
+
+__printf(2, 3)
+__cold
+void _btrfs_printk(const struct btrfs_fs_info *fs_info, const char *fmt, ...);
+
+#elif defined(CONFIG_PRINTK)
+
+#define btrfs_printk(fs_info, fmt, args...)				\
+	_btrfs_printk(fs_info, fmt, ##args)
+
 __printf(2, 3)
 __cold
-void btrfs_printk(const struct btrfs_fs_info *fs_info, const char *fmt, ...);
+void _btrfs_printk(const struct btrfs_fs_info *fs_info, const char *fmt, ...);
+
 #else
+
 #define btrfs_printk(fs_info, fmt, args...) \
 	btrfs_no_printk(fs_info, fmt, ##args)
 #endif
@@ -3171,6 +3721,52 @@ static inline void assertfail(const char *expr, const char* file, int line) { }
 #define ASSERT(expr)	(void)(expr)
 #endif
 
+#if BITS_PER_LONG == 32
+#define BTRFS_32BIT_MAX_FILE_SIZE (((u64)ULONG_MAX + 1) << PAGE_SHIFT)
+/*
+ * The warning threshold is 5/8th of the MAX_LFS_FILESIZE that limits the logical
+ * addresses of extents.
+ *
+ * For 4K page size it's about 10T, for 64K it's 160T.
+ */
+#define BTRFS_32BIT_EARLY_WARN_THRESHOLD (BTRFS_32BIT_MAX_FILE_SIZE * 5 / 8)
+void btrfs_warn_32bit_limit(struct btrfs_fs_info *fs_info);
+void btrfs_err_32bit_limit(struct btrfs_fs_info *fs_info);
+#endif
+
+/*
+ * Get the correct offset inside the page of extent buffer.
+ *
+ * @eb:		target extent buffer
+ * @start:	offset inside the extent buffer
+ *
+ * Will handle both sectorsize == PAGE_SIZE and sectorsize < PAGE_SIZE cases.
+ */
+static inline size_t get_eb_offset_in_page(const struct extent_buffer *eb,
+					   unsigned long offset)
+{
+	/*
+	 * For sectorsize == PAGE_SIZE case, eb->start will always be aligned
+	 * to PAGE_SIZE, thus adding it won't cause any difference.
+	 *
+	 * For sectorsize < PAGE_SIZE, we must only read the data that belongs
+	 * to the eb, thus we have to take the eb->start into consideration.
+	 */
+	return offset_in_page(offset + eb->start);
+}
+
+static inline unsigned long get_eb_page_index(unsigned long offset)
+{
+	/*
+	 * For sectorsize == PAGE_SIZE case, plain >> PAGE_SHIFT is enough.
+	 *
+	 * For sectorsize < PAGE_SIZE case, we only support 64K PAGE_SIZE,
+	 * and have ensured that all tree blocks are contained in one page,
+	 * thus we always get index == 0.
+	 */
+	return offset >> PAGE_SHIFT;
+}
+
 /*
  * Use that for functions that are conditionally exported for sanity tests but
  * otherwise static
@@ -3198,21 +3794,26 @@ const char * __attribute_const__ btrfs_decode_error(int errno);
 __cold
 void __btrfs_abort_transaction(struct btrfs_trans_handle *trans,
 			       const char *function,
-			       unsigned int line, int errno);
+			       unsigned int line, int errno, bool first_hit);
+
+bool __cold abort_should_print_stack(int errno);
 
 /*
  * Call btrfs_abort_transaction as early as possible when an error condition is
- * detected, that way the exact line number is reported.
+ * detected, that way the exact stack trace is reported for some errors.
  */
 #define btrfs_abort_transaction(trans, errno)		\
 do {								\
+	bool first = false;					\
 	/* Report first abort since mount */			\
 	if (!test_and_set_bit(BTRFS_FS_STATE_TRANS_ABORTED,	\
 			&((trans)->fs_info->fs_state))) {	\
-		if ((errno) != -EIO) {				\
-			WARN(1, KERN_DEBUG				\
+		first = true;					\
+		if (WARN(abort_should_print_stack(errno), 	\
+			KERN_DEBUG				\
 			"BTRFS: Transaction aborted (error %d)\n",	\
-			(errno));					\
+			(errno))) {					\
+			/* Stack trace printed. */			\
 		} else {						\
 			btrfs_debug((trans)->fs_info,			\
 				    "Transaction aborted (error %d)", \
@@ -3220,15 +3821,34 @@ do {								\
 		}						\
 	}							\
 	__btrfs_abort_transaction((trans), __func__,		\
-				  __LINE__, (errno));		\
+				  __LINE__, (errno), first);	\
 } while (0)
 
+#ifdef CONFIG_PRINTK_INDEX
+
 #define btrfs_handle_fs_error(fs_info, errno, fmt, args...)		\
-do {								\
-	__btrfs_handle_fs_error((fs_info), __func__, __LINE__,	\
-			  (errno), fmt, ##args);		\
+do {									\
+	printk_index_subsys_emit(					\
+		"BTRFS: error (device %s%s) in %s:%d: errno=%d %s",	\
+		KERN_CRIT, fmt);					\
+	__btrfs_handle_fs_error((fs_info), __func__, __LINE__,		\
+				(errno), fmt, ##args);			\
 } while (0)
 
+#else
+
+#define btrfs_handle_fs_error(fs_info, errno, fmt, args...)		\
+	__btrfs_handle_fs_error((fs_info), __func__, __LINE__,		\
+				(errno), fmt, ##args)
+
+#endif
+
+#define BTRFS_FS_ERROR(fs_info)	(unlikely(test_bit(BTRFS_FS_STATE_ERROR, \
+						   &(fs_info)->fs_state)))
+#define BTRFS_FS_LOG_CLEANUP_ERROR(fs_info)				\
+	(unlikely(test_bit(BTRFS_FS_STATE_LOG_CLEANUP_ERROR,		\
+			   &(fs_info)->fs_state)))
+
 __printf(5, 6)
 __cold
 void __btrfs_panic(struct btrfs_fs_info *fs_info, const char *function,
@@ -3372,17 +3992,19 @@ static inline int __btrfs_fs_compat_ro(struct btrfs_fs_info *fs_info, u64 flag)
 
 /* acl.c */
 #ifdef CONFIG_BTRFS_FS_POSIX_ACL
-struct posix_acl *btrfs_get_acl(struct inode *inode, int type);
-int btrfs_set_acl(struct inode *inode, struct posix_acl *acl, int type);
-int btrfs_init_acl(struct btrfs_trans_handle *trans,
-		   struct inode *inode, struct inode *dir);
+struct posix_acl *btrfs_get_acl(struct inode *inode, int type, bool rcu);
+int btrfs_set_acl(struct user_namespace *mnt_userns, struct inode *inode,
+		  struct posix_acl *acl, int type);
+int __btrfs_set_acl(struct btrfs_trans_handle *trans, struct inode *inode,
+		    struct posix_acl *acl, int type);
 #else
 #define btrfs_get_acl NULL
 #define btrfs_set_acl NULL
-static inline int btrfs_init_acl(struct btrfs_trans_handle *trans,
-				 struct inode *inode, struct inode *dir)
+static inline int __btrfs_set_acl(struct btrfs_trans_handle *trans,
+				  struct inode *inode, struct posix_acl *acl,
+				  int type)
 {
-	return 0;
+	return -EOPNOTSUPP;
 }
 #endif
 
@@ -3392,8 +4014,8 @@ int btrfs_init_reloc_root(struct btrfs_trans_handle *trans,
 			  struct btrfs_root *root);
 int btrfs_update_reloc_root(struct btrfs_trans_handle *trans,
 			    struct btrfs_root *root);
-int btrfs_recover_relocation(struct btrfs_root *root);
-int btrfs_reloc_clone_csums(struct inode *inode, u64 file_pos, u64 len);
+int btrfs_recover_relocation(struct btrfs_fs_info *fs_info);
+int btrfs_reloc_clone_csums(struct btrfs_inode *inode, u64 file_pos, u64 len);
 int btrfs_reloc_cow_block(struct btrfs_trans_handle *trans,
 			  struct btrfs_root *root, struct extent_buffer *buf,
 			  struct extent_buffer *cow);
@@ -3401,6 +4023,10 @@ void btrfs_reloc_pre_snapshot(struct btrfs_pending_snapshot *pending,
 			      u64 *bytes_to_reserve);
 int btrfs_reloc_post_snapshot(struct btrfs_trans_handle *trans,
 			      struct btrfs_pending_snapshot *pending);
+int btrfs_should_cancel_balance(struct btrfs_fs_info *fs_info);
+struct btrfs_root *find_reloc_root(struct btrfs_fs_info *fs_info,
+				   u64 bytenr);
+int btrfs_should_ignore_reloc_root(struct btrfs_root *root);
 
 /* scrub.c */
 int btrfs_scrub_dev(struct btrfs_fs_info *fs_info, u64 devid, u64 start,
@@ -3412,16 +4038,9 @@ int btrfs_scrub_cancel(struct btrfs_fs_info *info);
 int btrfs_scrub_cancel_dev(struct btrfs_device *dev);
 int btrfs_scrub_progress(struct btrfs_fs_info *fs_info, u64 devid,
 			 struct btrfs_scrub_progress *progress);
-static inline void btrfs_init_full_stripe_locks_tree(
-			struct btrfs_full_stripe_locks_tree *locks_root)
-{
-	locks_root->root = RB_ROOT;
-	mutex_init(&locks_root->lock);
-}
 
 /* dev-replace.c */
 void btrfs_bio_counter_inc_blocked(struct btrfs_fs_info *fs_info);
-void btrfs_bio_counter_inc_noblocked(struct btrfs_fs_info *fs_info);
 void btrfs_bio_counter_sub(struct btrfs_fs_info *fs_info, s64 amount);
 
 static inline void btrfs_bio_counter_dec(struct btrfs_fs_info *fs_info)
@@ -3429,21 +4048,6 @@ static inline void btrfs_bio_counter_dec(struct btrfs_fs_info *fs_info)
 	btrfs_bio_counter_sub(fs_info, 1);
 }
 
-/* reada.c */
-struct reada_control {
-	struct btrfs_fs_info	*fs_info;		/* tree to prefetch */
-	struct btrfs_key	key_start;
-	struct btrfs_key	key_end;	/* exclusive */
-	atomic_t		elems;
-	struct kref		refcnt;
-	wait_queue_head_t	wait;
-};
-struct reada_control *btrfs_reada_add(struct btrfs_root *root,
-			      struct btrfs_key *start, struct btrfs_key *end);
-int btrfs_reada_wait(void *handle);
-void btrfs_reada_detach(void *handle);
-int btree_readahead_hook(struct extent_buffer *eb, int err);
-
 static inline int is_fstree(u64 rootid)
 {
 	if (rootid == BTRFS_FS_TREE_OBJECTID ||
@@ -3458,13 +4062,40 @@ static inline int btrfs_defrag_cancelled(struct btrfs_fs_info *fs_info)
 	return signal_pending(current);
 }
 
-#define in_range(b, first, len) ((b) >= (first) && (b) < (first) + (len))
+/* verity.c */
+#ifdef CONFIG_FS_VERITY
+
+extern const struct fsverity_operations btrfs_verityops;
+int btrfs_drop_verity_items(struct btrfs_inode *inode);
+int btrfs_get_verity_descriptor(struct inode *inode, void *buf, size_t buf_size);
+
+BTRFS_SETGET_FUNCS(verity_descriptor_encryption, struct btrfs_verity_descriptor_item,
+		   encryption, 8);
+BTRFS_SETGET_FUNCS(verity_descriptor_size, struct btrfs_verity_descriptor_item,
+		   size, 64);
+BTRFS_SETGET_STACK_FUNCS(stack_verity_descriptor_encryption,
+			 struct btrfs_verity_descriptor_item, encryption, 8);
+BTRFS_SETGET_STACK_FUNCS(stack_verity_descriptor_size,
+			 struct btrfs_verity_descriptor_item, size, 64);
+
+#else
+
+static inline int btrfs_drop_verity_items(struct btrfs_inode *inode)
+{
+	return 0;
+}
+
+static inline int btrfs_get_verity_descriptor(struct inode *inode, void *buf,
+					      size_t buf_size)
+{
+	return -EPERM;
+}
+
+#endif
 
 /* Sanity test specific functions */
 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
-void btrfs_test_inode_set_ops(struct inode *inode);
 void btrfs_test_destroy_inode(struct inode *inode);
-
 static inline int btrfs_is_testing(struct btrfs_fs_info *fs_info)
 {
 	return test_bit(BTRFS_FS_STATE_DUMMY_FS_INFO, &fs_info->fs_state);
@@ -3476,4 +4107,22 @@ static inline int btrfs_is_testing(struct btrfs_fs_info *fs_info)
 }
 #endif
 
+static inline bool btrfs_is_data_reloc_root(const struct btrfs_root *root)
+{
+	return root->root_key.objectid == BTRFS_DATA_RELOC_TREE_OBJECTID;
+}
+
+/*
+ * We use page status Private2 to indicate there is an ordered extent with
+ * unfinished IO.
+ *
+ * Rename the Private2 accessors to Ordered, to improve readability.
+ */
+#define PageOrdered(page)		PagePrivate2(page)
+#define SetPageOrdered(page)		SetPagePrivate2(page)
+#define ClearPageOrdered(page)		ClearPagePrivate2(page)
+#define folio_test_ordered(folio)	folio_test_private_2(folio)
+#define folio_set_ordered(folio)	folio_set_private_2(folio)
+#define folio_clear_ordered(folio)	folio_clear_private_2(folio)
+
 #endif
diff --git a/fs/btrfs/delalloc-space.c b/fs/btrfs/delalloc-space.c
index 4cdac4d834f5..118b2e20b2e1 100644
--- a/fs/btrfs/delalloc-space.c
+++ b/fs/btrfs/delalloc-space.c
@@ -9,136 +9,129 @@
 #include "qgroup.h"
 #include "block-group.h"
 
+/*
+ * HOW DOES THIS WORK
+ *
+ * There are two stages to data reservations, one for data and one for metadata
+ * to handle the new extents and checksums generated by writing data.
+ *
+ *
+ * DATA RESERVATION
+ *   The general flow of the data reservation is as follows
+ *
+ *   -> Reserve
+ *     We call into btrfs_reserve_data_bytes() for the user request bytes that
+ *     they wish to write.  We make this reservation and add it to
+ *     space_info->bytes_may_use.  We set EXTENT_DELALLOC on the inode io_tree
+ *     for the range and carry on if this is buffered, or follow up trying to
+ *     make a real allocation if we are pre-allocating or doing O_DIRECT.
+ *
+ *   -> Use
+ *     At writepages()/prealloc/O_DIRECT time we will call into
+ *     btrfs_reserve_extent() for some part or all of this range of bytes.  We
+ *     will make the allocation and subtract space_info->bytes_may_use by the
+ *     original requested length and increase the space_info->bytes_reserved by
+ *     the allocated length.  This distinction is important because compression
+ *     may allocate a smaller on disk extent than we previously reserved.
+ *
+ *   -> Allocation
+ *     finish_ordered_io() will insert the new file extent item for this range,
+ *     and then add a delayed ref update for the extent tree.  Once that delayed
+ *     ref is written the extent size is subtracted from
+ *     space_info->bytes_reserved and added to space_info->bytes_used.
+ *
+ *   Error handling
+ *
+ *   -> By the reservation maker
+ *     This is the simplest case, we haven't completed our operation and we know
+ *     how much we reserved, we can simply call
+ *     btrfs_free_reserved_data_space*() and it will be removed from
+ *     space_info->bytes_may_use.
+ *
+ *   -> After the reservation has been made, but before cow_file_range()
+ *     This is specifically for the delalloc case.  You must clear
+ *     EXTENT_DELALLOC with the EXTENT_CLEAR_DATA_RESV bit, and the range will
+ *     be subtracted from space_info->bytes_may_use.
+ *
+ * METADATA RESERVATION
+ *   The general metadata reservation lifetimes are discussed elsewhere, this
+ *   will just focus on how it is used for delalloc space.
+ *
+ *   We keep track of two things on a per inode bases
+ *
+ *   ->outstanding_extents
+ *     This is the number of file extent items we'll need to handle all of the
+ *     outstanding DELALLOC space we have in this inode.  We limit the maximum
+ *     size of an extent, so a large contiguous dirty area may require more than
+ *     one outstanding_extent, which is why count_max_extents() is used to
+ *     determine how many outstanding_extents get added.
+ *
+ *   ->csum_bytes
+ *     This is essentially how many dirty bytes we have for this inode, so we
+ *     can calculate the number of checksum items we would have to add in order
+ *     to checksum our outstanding data.
+ *
+ *   We keep a per-inode block_rsv in order to make it easier to keep track of
+ *   our reservation.  We use btrfs_calculate_inode_block_rsv_size() to
+ *   calculate the current theoretical maximum reservation we would need for the
+ *   metadata for this inode.  We call this and then adjust our reservation as
+ *   necessary, either by attempting to reserve more space, or freeing up excess
+ *   space.
+ *
+ * OUTSTANDING_EXTENTS HANDLING
+ *
+ *  ->outstanding_extents is used for keeping track of how many extents we will
+ *  need to use for this inode, and it will fluctuate depending on where you are
+ *  in the life cycle of the dirty data.  Consider the following normal case for
+ *  a completely clean inode, with a num_bytes < our maximum allowed extent size
+ *
+ *  -> reserve
+ *    ->outstanding_extents += 1 (current value is 1)
+ *
+ *  -> set_delalloc
+ *    ->outstanding_extents += 1 (current value is 2)
+ *
+ *  -> btrfs_delalloc_release_extents()
+ *    ->outstanding_extents -= 1 (current value is 1)
+ *
+ *    We must call this once we are done, as we hold our reservation for the
+ *    duration of our operation, and then assume set_delalloc will update the
+ *    counter appropriately.
+ *
+ *  -> add ordered extent
+ *    ->outstanding_extents += 1 (current value is 2)
+ *
+ *  -> btrfs_clear_delalloc_extent
+ *    ->outstanding_extents -= 1 (current value is 1)
+ *
+ *  -> finish_ordered_io/btrfs_remove_ordered_extent
+ *    ->outstanding_extents -= 1 (current value is 0)
+ *
+ *  Each stage is responsible for their own accounting of the extent, thus
+ *  making error handling and cleanup easier.
+ */
+
 int btrfs_alloc_data_chunk_ondemand(struct btrfs_inode *inode, u64 bytes)
 {
 	struct btrfs_root *root = inode->root;
 	struct btrfs_fs_info *fs_info = root->fs_info;
-	struct btrfs_space_info *data_sinfo = fs_info->data_sinfo;
-	u64 used;
-	int ret = 0;
-	int need_commit = 2;
-	int have_pinned_space;
+	enum btrfs_reserve_flush_enum flush = BTRFS_RESERVE_FLUSH_DATA;
 
 	/* Make sure bytes are sectorsize aligned */
 	bytes = ALIGN(bytes, fs_info->sectorsize);
 
-	if (btrfs_is_free_space_inode(inode)) {
-		need_commit = 0;
-		ASSERT(current->journal_info);
-	}
-
-again:
-	/* Make sure we have enough space to handle the data first */
-	spin_lock(&data_sinfo->lock);
-	used = btrfs_space_info_used(data_sinfo, true);
-
-	if (used + bytes > data_sinfo->total_bytes) {
-		struct btrfs_trans_handle *trans;
-
-		/*
-		 * If we don't have enough free bytes in this space then we need
-		 * to alloc a new chunk.
-		 */
-		if (!data_sinfo->full) {
-			u64 alloc_target;
-
-			data_sinfo->force_alloc = CHUNK_ALLOC_FORCE;
-			spin_unlock(&data_sinfo->lock);
-
-			alloc_target = btrfs_data_alloc_profile(fs_info);
-			/*
-			 * It is ugly that we don't call nolock join
-			 * transaction for the free space inode case here.
-			 * But it is safe because we only do the data space
-			 * reservation for the free space cache in the
-			 * transaction context, the common join transaction
-			 * just increase the counter of the current transaction
-			 * handler, doesn't try to acquire the trans_lock of
-			 * the fs.
-			 */
-			trans = btrfs_join_transaction(root);
-			if (IS_ERR(trans))
-				return PTR_ERR(trans);
-
-			ret = btrfs_chunk_alloc(trans, alloc_target,
-						CHUNK_ALLOC_NO_FORCE);
-			btrfs_end_transaction(trans);
-			if (ret < 0) {
-				if (ret != -ENOSPC)
-					return ret;
-				else {
-					have_pinned_space = 1;
-					goto commit_trans;
-				}
-			}
-
-			goto again;
-		}
-
-		/*
-		 * If we don't have enough pinned space to deal with this
-		 * allocation, and no removed chunk in current transaction,
-		 * don't bother committing the transaction.
-		 */
-		have_pinned_space = __percpu_counter_compare(
-			&data_sinfo->total_bytes_pinned,
-			used + bytes - data_sinfo->total_bytes,
-			BTRFS_TOTAL_BYTES_PINNED_BATCH);
-		spin_unlock(&data_sinfo->lock);
-
-		/* Commit the current transaction and try again */
-commit_trans:
-		if (need_commit) {
-			need_commit--;
-
-			if (need_commit > 0) {
-				btrfs_start_delalloc_roots(fs_info, -1);
-				btrfs_wait_ordered_roots(fs_info, U64_MAX, 0,
-							 (u64)-1);
-			}
-
-			trans = btrfs_join_transaction(root);
-			if (IS_ERR(trans))
-				return PTR_ERR(trans);
-			if (have_pinned_space >= 0 ||
-			    test_bit(BTRFS_TRANS_HAVE_FREE_BGS,
-				     &trans->transaction->flags) ||
-			    need_commit > 0) {
-				ret = btrfs_commit_transaction(trans);
-				if (ret)
-					return ret;
-				/*
-				 * The cleaner kthread might still be doing iput
-				 * operations. Wait for it to finish so that
-				 * more space is released.  We don't need to
-				 * explicitly run the delayed iputs here because
-				 * the commit_transaction would have woken up
-				 * the cleaner.
-				 */
-				ret = btrfs_wait_on_delayed_iputs(fs_info);
-				if (ret)
-					return ret;
-				goto again;
-			} else {
-				btrfs_end_transaction(trans);
-			}
-		}
-
-		trace_btrfs_space_reservation(fs_info,
-					      "space_info:enospc",
-					      data_sinfo->flags, bytes, 1);
-		return -ENOSPC;
-	}
-	btrfs_space_info_update_bytes_may_use(fs_info, data_sinfo, bytes);
-	spin_unlock(&data_sinfo->lock);
+	if (btrfs_is_free_space_inode(inode))
+		flush = BTRFS_RESERVE_FLUSH_FREE_SPACE_INODE;
 
-	return 0;
+	return btrfs_reserve_data_bytes(fs_info, bytes, flush);
 }
 
-int btrfs_check_data_free_space(struct inode *inode,
-			struct extent_changeset **reserved, u64 start, u64 len)
+int btrfs_check_data_free_space(struct btrfs_inode *inode,
+				struct extent_changeset **reserved, u64 start,
+				u64 len, bool noflush)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
+	enum btrfs_reserve_flush_enum flush = BTRFS_RESERVE_FLUSH_DATA;
 	int ret;
 
 	/* align the range */
@@ -146,16 +139,24 @@ int btrfs_check_data_free_space(struct inode *inode,
 	      round_down(start, fs_info->sectorsize);
 	start = round_down(start, fs_info->sectorsize);
 
-	ret = btrfs_alloc_data_chunk_ondemand(BTRFS_I(inode), len);
+	if (noflush)
+		flush = BTRFS_RESERVE_NO_FLUSH;
+	else if (btrfs_is_free_space_inode(inode))
+		flush = BTRFS_RESERVE_FLUSH_FREE_SPACE_INODE;
+
+	ret = btrfs_reserve_data_bytes(fs_info, len, flush);
 	if (ret < 0)
 		return ret;
 
 	/* Use new btrfs_qgroup_reserve_data to reserve precious data space. */
 	ret = btrfs_qgroup_reserve_data(inode, reserved, start, len);
-	if (ret < 0)
-		btrfs_free_reserved_data_space_noquota(inode, start, len);
-	else
+	if (ret < 0) {
+		btrfs_free_reserved_data_space_noquota(fs_info, len);
+		extent_changeset_free(*reserved);
+		*reserved = NULL;
+	} else {
 		ret = 0;
+	}
 	return ret;
 }
 
@@ -167,21 +168,15 @@ int btrfs_check_data_free_space(struct inode *inode,
  * which we can't sleep and is sure it won't affect qgroup reserved space.
  * Like clear_bit_hook().
  */
-void btrfs_free_reserved_data_space_noquota(struct inode *inode, u64 start,
+void btrfs_free_reserved_data_space_noquota(struct btrfs_fs_info *fs_info,
 					    u64 len)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	struct btrfs_space_info *data_sinfo;
 
-	/* Make sure the range is aligned to sectorsize */
-	len = round_up(start + len, fs_info->sectorsize) -
-	      round_down(start, fs_info->sectorsize);
-	start = round_down(start, fs_info->sectorsize);
+	ASSERT(IS_ALIGNED(len, fs_info->sectorsize));
 
 	data_sinfo = fs_info->data_sinfo;
-	spin_lock(&data_sinfo->lock);
-	btrfs_space_info_update_bytes_may_use(fs_info, data_sinfo, -len);
-	spin_unlock(&data_sinfo->lock);
+	btrfs_space_info_free_bytes_may_use(fs_info, data_sinfo, len);
 }
 
 /*
@@ -191,27 +186,29 @@ void btrfs_free_reserved_data_space_noquota(struct inode *inode, u64 start,
  * This one will handle the per-inode data rsv map for accurate reserved
  * space framework.
  */
-void btrfs_free_reserved_data_space(struct inode *inode,
+void btrfs_free_reserved_data_space(struct btrfs_inode *inode,
 			struct extent_changeset *reserved, u64 start, u64 len)
 {
-	struct btrfs_root *root = BTRFS_I(inode)->root;
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
 
 	/* Make sure the range is aligned to sectorsize */
-	len = round_up(start + len, root->fs_info->sectorsize) -
-	      round_down(start, root->fs_info->sectorsize);
-	start = round_down(start, root->fs_info->sectorsize);
+	len = round_up(start + len, fs_info->sectorsize) -
+	      round_down(start, fs_info->sectorsize);
+	start = round_down(start, fs_info->sectorsize);
 
-	btrfs_free_reserved_data_space_noquota(inode, start, len);
+	btrfs_free_reserved_data_space_noquota(fs_info, len);
 	btrfs_qgroup_free_data(inode, reserved, start, len);
 }
 
 /**
- * btrfs_inode_rsv_release - release any excessive reservation.
- * @inode - the inode we need to release from.
- * @qgroup_free - free or convert qgroup meta.
- *   Unlike normal operation, qgroup meta reservation needs to know if we are
- *   freeing qgroup reservation or just converting it into per-trans.  Normally
- *   @qgroup_free is true for error handling, and false for normal release.
+ * Release any excessive reservation
+ *
+ * @inode:       the inode we need to release from
+ * @qgroup_free: free or convert qgroup meta. Unlike normal operation, qgroup
+ *               meta reservation needs to know if we are freeing qgroup
+ *               reservation or just converting it into per-trans.  Normally
+ *               @qgroup_free is true for error handling, and false for normal
+ *               release.
  *
  * This is the same as btrfs_block_rsv_release, except that it handles the
  * tracepoint for the reservation.
@@ -228,8 +225,8 @@ static void btrfs_inode_rsv_release(struct btrfs_inode *inode, bool qgroup_free)
 	 * are releasing 0 bytes, and then we'll just get the reservation over
 	 * the size free'd.
 	 */
-	released = __btrfs_block_rsv_release(fs_info, block_rsv, 0,
-					     &qgroup_to_release);
+	released = btrfs_block_rsv_release(fs_info, block_rsv, 0,
+					   &qgroup_to_release);
 	if (released > 0)
 		trace_btrfs_space_reservation(fs_info, "delalloc",
 					      btrfs_ino(inode), released, 0);
@@ -280,11 +277,11 @@ static void btrfs_calculate_inode_block_rsv_size(struct btrfs_fs_info *fs_info,
 }
 
 static void calc_inode_reservations(struct btrfs_fs_info *fs_info,
-				    u64 num_bytes, u64 *meta_reserve,
-				    u64 *qgroup_reserve)
+				    u64 num_bytes, u64 disk_num_bytes,
+				    u64 *meta_reserve, u64 *qgroup_reserve)
 {
-	u64 nr_extents = count_max_extents(num_bytes);
-	u64 csum_leaves = btrfs_csum_bytes_to_leaves(fs_info, num_bytes);
+	u64 nr_extents = count_max_extents(fs_info, num_bytes);
+	u64 csum_leaves = btrfs_csum_bytes_to_leaves(fs_info, disk_num_bytes);
 	u64 inode_update = btrfs_calc_metadata_size(fs_info, 1);
 
 	*meta_reserve = btrfs_calc_insert_metadata_size(fs_info,
@@ -298,7 +295,8 @@ static void calc_inode_reservations(struct btrfs_fs_info *fs_info,
 	*qgroup_reserve = nr_extents * fs_info->nodesize;
 }
 
-int btrfs_delalloc_reserve_metadata(struct btrfs_inode *inode, u64 num_bytes)
+int btrfs_delalloc_reserve_metadata(struct btrfs_inode *inode, u64 num_bytes,
+				    u64 disk_num_bytes, bool noflush)
 {
 	struct btrfs_root *root = inode->root;
 	struct btrfs_fs_info *fs_info = root->fs_info;
@@ -317,7 +315,7 @@ int btrfs_delalloc_reserve_metadata(struct btrfs_inode *inode, u64 num_bytes)
 	 * If we have a transaction open (can happen if we call truncate_block
 	 * from truncate), then we need FLUSH_LIMIT so we don't deadlock.
 	 */
-	if (btrfs_is_free_space_inode(inode)) {
+	if (noflush || btrfs_is_free_space_inode(inode)) {
 		flush = BTRFS_RESERVE_NO_FLUSH;
 	} else {
 		if (current->journal_info)
@@ -328,6 +326,7 @@ int btrfs_delalloc_reserve_metadata(struct btrfs_inode *inode, u64 num_bytes)
 	}
 
 	num_bytes = ALIGN(num_bytes, fs_info->sectorsize);
+	disk_num_bytes = ALIGN(disk_num_bytes, fs_info->sectorsize);
 
 	/*
 	 * We always want to do it this way, every other way is wrong and ends
@@ -339,12 +338,13 @@ int btrfs_delalloc_reserve_metadata(struct btrfs_inode *inode, u64 num_bytes)
 	 * everything out and try again, which is bad.  This way we just
 	 * over-reserve slightly, and clean up the mess when we are done.
 	 */
-	calc_inode_reservations(fs_info, num_bytes, &meta_reserve,
-				&qgroup_reserve);
-	ret = btrfs_qgroup_reserve_meta_prealloc(root, qgroup_reserve, true);
+	calc_inode_reservations(fs_info, num_bytes, disk_num_bytes,
+				&meta_reserve, &qgroup_reserve);
+	ret = btrfs_qgroup_reserve_meta_prealloc(root, qgroup_reserve, true,
+						 noflush);
 	if (ret)
 		return ret;
-	ret = btrfs_reserve_metadata_bytes(root, block_rsv, meta_reserve, flush);
+	ret = btrfs_reserve_metadata_bytes(fs_info, block_rsv, meta_reserve, flush);
 	if (ret) {
 		btrfs_qgroup_free_meta_prealloc(root, qgroup_reserve);
 		return ret;
@@ -357,9 +357,9 @@ int btrfs_delalloc_reserve_metadata(struct btrfs_inode *inode, u64 num_bytes)
 	 * needs to free the reservation we just made.
 	 */
 	spin_lock(&inode->lock);
-	nr_extents = count_max_extents(num_bytes);
+	nr_extents = count_max_extents(fs_info, num_bytes);
 	btrfs_mod_outstanding_extents(inode, nr_extents);
-	inode->csum_bytes += num_bytes;
+	inode->csum_bytes += disk_num_bytes;
 	btrfs_calculate_inode_block_rsv_size(fs_info, inode);
 	spin_unlock(&inode->lock);
 
@@ -376,7 +376,8 @@ int btrfs_delalloc_reserve_metadata(struct btrfs_inode *inode, u64 num_bytes)
 }
 
 /**
- * btrfs_delalloc_release_metadata - release a metadata reservation for an inode
+ * Release a metadata reservation for an inode
+ *
  * @inode: the inode to release the reservation for.
  * @num_bytes: the number of bytes we are releasing.
  * @qgroup_free: free qgroup reservation or convert it to per-trans reservation
@@ -419,7 +420,7 @@ void btrfs_delalloc_release_extents(struct btrfs_inode *inode, u64 num_bytes)
 	unsigned num_extents;
 
 	spin_lock(&inode->lock);
-	num_extents = count_max_extents(num_bytes);
+	num_extents = count_max_extents(fs_info, num_bytes);
 	btrfs_mod_outstanding_extents(inode, -num_extents);
 	btrfs_calculate_inode_block_rsv_size(fs_info, inode);
 	spin_unlock(&inode->lock);
@@ -455,36 +456,41 @@ void btrfs_delalloc_release_extents(struct btrfs_inode *inode, u64 num_bytes)
  * Return 0 for success
  * Return <0 for error(-ENOSPC or -EQUOT)
  */
-int btrfs_delalloc_reserve_space(struct inode *inode,
+int btrfs_delalloc_reserve_space(struct btrfs_inode *inode,
 			struct extent_changeset **reserved, u64 start, u64 len)
 {
 	int ret;
 
-	ret = btrfs_check_data_free_space(inode, reserved, start, len);
+	ret = btrfs_check_data_free_space(inode, reserved, start, len, false);
 	if (ret < 0)
 		return ret;
-	ret = btrfs_delalloc_reserve_metadata(BTRFS_I(inode), len);
-	if (ret < 0)
+	ret = btrfs_delalloc_reserve_metadata(inode, len, len, false);
+	if (ret < 0) {
 		btrfs_free_reserved_data_space(inode, *reserved, start, len);
+		extent_changeset_free(*reserved);
+		*reserved = NULL;
+	}
 	return ret;
 }
 
 /**
- * btrfs_delalloc_release_space - release data and metadata space for delalloc
- * @inode: inode we're releasing space for
- * @start: start position of the space already reserved
- * @len: the len of the space already reserved
- * @release_bytes: the len of the space we consumed or didn't use
+ * Release data and metadata space for delalloc
+ *
+ * @inode:       inode we're releasing space for
+ * @reserved:    list of changed/reserved ranges
+ * @start:       start position of the space already reserved
+ * @len:         length of the space already reserved
+ * @qgroup_free: should qgroup reserved-space also be freed
  *
  * This function will release the metadata space that was not used and will
  * decrement ->delalloc_bytes and remove it from the fs_info delalloc_inodes
  * list if there are no delalloc bytes left.
  * Also it will handle the qgroup reserved space.
  */
-void btrfs_delalloc_release_space(struct inode *inode,
+void btrfs_delalloc_release_space(struct btrfs_inode *inode,
 				  struct extent_changeset *reserved,
 				  u64 start, u64 len, bool qgroup_free)
 {
-	btrfs_delalloc_release_metadata(BTRFS_I(inode), len, qgroup_free);
+	btrfs_delalloc_release_metadata(inode, len, qgroup_free);
 	btrfs_free_reserved_data_space(inode, reserved, start, len);
 }
diff --git a/fs/btrfs/delalloc-space.h b/fs/btrfs/delalloc-space.h
index 54466fbd7075..e07d46043455 100644
--- a/fs/btrfs/delalloc-space.h
+++ b/fs/btrfs/delalloc-space.h
@@ -6,18 +6,19 @@
 struct extent_changeset;
 
 int btrfs_alloc_data_chunk_ondemand(struct btrfs_inode *inode, u64 bytes);
-int btrfs_check_data_free_space(struct inode *inode,
-			struct extent_changeset **reserved, u64 start, u64 len);
-void btrfs_free_reserved_data_space(struct inode *inode,
+int btrfs_check_data_free_space(struct btrfs_inode *inode,
+			struct extent_changeset **reserved, u64 start, u64 len,
+			bool noflush);
+void btrfs_free_reserved_data_space(struct btrfs_inode *inode,
 			struct extent_changeset *reserved, u64 start, u64 len);
-void btrfs_delalloc_release_space(struct inode *inode,
+void btrfs_delalloc_release_space(struct btrfs_inode *inode,
 				  struct extent_changeset *reserved,
 				  u64 start, u64 len, bool qgroup_free);
-void btrfs_free_reserved_data_space_noquota(struct inode *inode, u64 start,
+void btrfs_free_reserved_data_space_noquota(struct btrfs_fs_info *fs_info,
 					    u64 len);
 void btrfs_delalloc_release_metadata(struct btrfs_inode *inode, u64 num_bytes,
 				     bool qgroup_free);
-int btrfs_delalloc_reserve_space(struct inode *inode,
+int btrfs_delalloc_reserve_space(struct btrfs_inode *inode,
 			struct extent_changeset **reserved, u64 start, u64 len);
 
 #endif /* BTRFS_DELALLOC_SPACE_H */
diff --git a/fs/btrfs/delayed-inode.c b/fs/btrfs/delayed-inode.c
index d3e15e1d4a91..cac5169eaf8d 100644
--- a/fs/btrfs/delayed-inode.c
+++ b/fs/btrfs/delayed-inode.c
@@ -13,6 +13,7 @@
 #include "ctree.h"
 #include "qgroup.h"
 #include "locking.h"
+#include "inode-item.h"
 
 #define BTRFS_DELAYED_WRITEBACK		512
 #define BTRFS_DELAYED_BACKGROUND	128
@@ -51,18 +52,6 @@ static inline void btrfs_init_delayed_node(
 	INIT_LIST_HEAD(&delayed_node->p_list);
 }
 
-static inline int btrfs_is_continuous_delayed_item(
-					struct btrfs_delayed_item *item1,
-					struct btrfs_delayed_item *item2)
-{
-	if (item1->key.type == BTRFS_DIR_INDEX_KEY &&
-	    item1->key.objectid == item2->key.objectid &&
-	    item1->key.type == item2->key.type &&
-	    item1->key.offset + 1 == item2->key.offset)
-		return 1;
-	return 0;
-}
-
 static struct btrfs_delayed_node *btrfs_get_delayed_node(
 		struct btrfs_inode *btrfs_inode)
 {
@@ -313,15 +302,21 @@ static inline void btrfs_release_prepared_delayed_node(
 	__btrfs_release_delayed_node(node, 1);
 }
 
-static struct btrfs_delayed_item *btrfs_alloc_delayed_item(u32 data_len)
+static struct btrfs_delayed_item *btrfs_alloc_delayed_item(u16 data_len,
+					   struct btrfs_delayed_node *node,
+					   enum btrfs_delayed_item_type type)
 {
 	struct btrfs_delayed_item *item;
+
 	item = kmalloc(sizeof(*item) + data_len, GFP_NOFS);
 	if (item) {
 		item->data_len = data_len;
-		item->ins_or_del = 0;
+		item->type = type;
 		item->bytes_reserved = 0;
-		item->delayed_node = NULL;
+		item->delayed_node = node;
+		RB_CLEAR_NODE(&item->rb_node);
+		INIT_LIST_HEAD(&item->log_list);
+		item->logged = false;
 		refcount_set(&item->refs, 1);
 	}
 	return item;
@@ -330,89 +325,46 @@ static struct btrfs_delayed_item *btrfs_alloc_delayed_item(u32 data_len)
 /*
  * __btrfs_lookup_delayed_item - look up the delayed item by key
  * @delayed_node: pointer to the delayed node
- * @key:	  the key to look up
- * @prev:	  used to store the prev item if the right item isn't found
- * @next:	  used to store the next item if the right item isn't found
+ * @index:	  the dir index value to lookup (offset of a dir index key)
  *
  * Note: if we don't find the right item, we will return the prev item and
  * the next item.
  */
 static struct btrfs_delayed_item *__btrfs_lookup_delayed_item(
 				struct rb_root *root,
-				struct btrfs_key *key,
-				struct btrfs_delayed_item **prev,
-				struct btrfs_delayed_item **next)
+				u64 index)
 {
-	struct rb_node *node, *prev_node = NULL;
+	struct rb_node *node = root->rb_node;
 	struct btrfs_delayed_item *delayed_item = NULL;
-	int ret = 0;
-
-	node = root->rb_node;
 
 	while (node) {
 		delayed_item = rb_entry(node, struct btrfs_delayed_item,
 					rb_node);
-		prev_node = node;
-		ret = btrfs_comp_cpu_keys(&delayed_item->key, key);
-		if (ret < 0)
+		if (delayed_item->index < index)
 			node = node->rb_right;
-		else if (ret > 0)
+		else if (delayed_item->index > index)
 			node = node->rb_left;
 		else
 			return delayed_item;
 	}
 
-	if (prev) {
-		if (!prev_node)
-			*prev = NULL;
-		else if (ret < 0)
-			*prev = delayed_item;
-		else if ((node = rb_prev(prev_node)) != NULL) {
-			*prev = rb_entry(node, struct btrfs_delayed_item,
-					 rb_node);
-		} else
-			*prev = NULL;
-	}
-
-	if (next) {
-		if (!prev_node)
-			*next = NULL;
-		else if (ret > 0)
-			*next = delayed_item;
-		else if ((node = rb_next(prev_node)) != NULL) {
-			*next = rb_entry(node, struct btrfs_delayed_item,
-					 rb_node);
-		} else
-			*next = NULL;
-	}
 	return NULL;
 }
 
-static struct btrfs_delayed_item *__btrfs_lookup_delayed_insertion_item(
-					struct btrfs_delayed_node *delayed_node,
-					struct btrfs_key *key)
-{
-	return __btrfs_lookup_delayed_item(&delayed_node->ins_root.rb_root, key,
-					   NULL, NULL);
-}
-
 static int __btrfs_add_delayed_item(struct btrfs_delayed_node *delayed_node,
-				    struct btrfs_delayed_item *ins,
-				    int action)
+				    struct btrfs_delayed_item *ins)
 {
 	struct rb_node **p, *node;
 	struct rb_node *parent_node = NULL;
 	struct rb_root_cached *root;
 	struct btrfs_delayed_item *item;
-	int cmp;
 	bool leftmost = true;
 
-	if (action == BTRFS_DELAYED_INSERTION_ITEM)
+	if (ins->type == BTRFS_DELAYED_INSERTION_ITEM)
 		root = &delayed_node->ins_root;
-	else if (action == BTRFS_DELAYED_DELETION_ITEM)
-		root = &delayed_node->del_root;
 	else
-		BUG();
+		root = &delayed_node->del_root;
+
 	p = &root->rb_root.rb_node;
 	node = &ins->rb_node;
 
@@ -421,11 +373,10 @@ static int __btrfs_add_delayed_item(struct btrfs_delayed_node *delayed_node,
 		item = rb_entry(parent_node, struct btrfs_delayed_item,
 				 rb_node);
 
-		cmp = btrfs_comp_cpu_keys(&item->key, &ins->key);
-		if (cmp < 0) {
+		if (item->index < ins->index) {
 			p = &(*p)->rb_right;
 			leftmost = false;
-		} else if (cmp > 0) {
+		} else if (item->index > ins->index) {
 			p = &(*p)->rb_left;
 		} else {
 			return -EEXIST;
@@ -434,33 +385,16 @@ static int __btrfs_add_delayed_item(struct btrfs_delayed_node *delayed_node,
 
 	rb_link_node(node, parent_node, p);
 	rb_insert_color_cached(node, root, leftmost);
-	ins->delayed_node = delayed_node;
-	ins->ins_or_del = action;
 
-	if (ins->key.type == BTRFS_DIR_INDEX_KEY &&
-	    action == BTRFS_DELAYED_INSERTION_ITEM &&
-	    ins->key.offset >= delayed_node->index_cnt)
-			delayed_node->index_cnt = ins->key.offset + 1;
+	if (ins->type == BTRFS_DELAYED_INSERTION_ITEM &&
+	    ins->index >= delayed_node->index_cnt)
+		delayed_node->index_cnt = ins->index + 1;
 
 	delayed_node->count++;
 	atomic_inc(&delayed_node->root->fs_info->delayed_root->items);
 	return 0;
 }
 
-static int __btrfs_add_delayed_insertion_item(struct btrfs_delayed_node *node,
-					      struct btrfs_delayed_item *item)
-{
-	return __btrfs_add_delayed_item(node, item,
-					BTRFS_DELAYED_INSERTION_ITEM);
-}
-
-static int __btrfs_add_delayed_deletion_item(struct btrfs_delayed_node *node,
-					     struct btrfs_delayed_item *item)
-{
-	return __btrfs_add_delayed_item(node, item,
-					BTRFS_DELAYED_DELETION_ITEM);
-}
-
 static void finish_one_item(struct btrfs_delayed_root *delayed_root)
 {
 	int seq = atomic_inc_return(&delayed_root->items_seq);
@@ -476,21 +410,21 @@ static void __btrfs_remove_delayed_item(struct btrfs_delayed_item *delayed_item)
 	struct rb_root_cached *root;
 	struct btrfs_delayed_root *delayed_root;
 
-	/* Not associated with any delayed_node */
-	if (!delayed_item->delayed_node)
+	/* Not inserted, ignore it. */
+	if (RB_EMPTY_NODE(&delayed_item->rb_node))
 		return;
+
 	delayed_root = delayed_item->delayed_node->root->fs_info->delayed_root;
 
 	BUG_ON(!delayed_root);
-	BUG_ON(delayed_item->ins_or_del != BTRFS_DELAYED_DELETION_ITEM &&
-	       delayed_item->ins_or_del != BTRFS_DELAYED_INSERTION_ITEM);
 
-	if (delayed_item->ins_or_del == BTRFS_DELAYED_INSERTION_ITEM)
+	if (delayed_item->type == BTRFS_DELAYED_INSERTION_ITEM)
 		root = &delayed_item->delayed_node->ins_root;
 	else
 		root = &delayed_item->delayed_node->del_root;
 
 	rb_erase_cached(&delayed_item->rb_node, root);
+	RB_CLEAR_NODE(&delayed_item->rb_node);
 	delayed_item->delayed_node->count--;
 
 	finish_one_item(delayed_root);
@@ -545,12 +479,11 @@ static struct btrfs_delayed_item *__btrfs_next_delayed_item(
 }
 
 static int btrfs_delayed_item_reserve_metadata(struct btrfs_trans_handle *trans,
-					       struct btrfs_root *root,
 					       struct btrfs_delayed_item *item)
 {
 	struct btrfs_block_rsv *src_rsv;
 	struct btrfs_block_rsv *dst_rsv;
-	struct btrfs_fs_info *fs_info = root->fs_info;
+	struct btrfs_fs_info *fs_info = trans->fs_info;
 	u64 num_bytes;
 	int ret;
 
@@ -570,9 +503,15 @@ static int btrfs_delayed_item_reserve_metadata(struct btrfs_trans_handle *trans,
 	ret = btrfs_block_rsv_migrate(src_rsv, dst_rsv, num_bytes, true);
 	if (!ret) {
 		trace_btrfs_space_reservation(fs_info, "delayed_item",
-					      item->key.objectid,
+					      item->delayed_node->inode_id,
 					      num_bytes, 1);
-		item->bytes_reserved = num_bytes;
+		/*
+		 * For insertions we track reserved metadata space by accounting
+		 * for the number of leaves that will be used, based on the delayed
+		 * node's index_items_size field.
+		 */
+		if (item->type == BTRFS_DELAYED_DELETION_ITEM)
+			item->bytes_reserved = num_bytes;
 	}
 
 	return ret;
@@ -593,16 +532,29 @@ static void btrfs_delayed_item_release_metadata(struct btrfs_root *root,
 	 * to release/reserve qgroup space.
 	 */
 	trace_btrfs_space_reservation(fs_info, "delayed_item",
-				      item->key.objectid, item->bytes_reserved,
-				      0);
-	btrfs_block_rsv_release(fs_info, rsv,
-				item->bytes_reserved);
+				      item->delayed_node->inode_id,
+				      item->bytes_reserved, 0);
+	btrfs_block_rsv_release(fs_info, rsv, item->bytes_reserved, NULL);
+}
+
+static void btrfs_delayed_item_release_leaves(struct btrfs_delayed_node *node,
+					      unsigned int num_leaves)
+{
+	struct btrfs_fs_info *fs_info = node->root->fs_info;
+	const u64 bytes = btrfs_calc_insert_metadata_size(fs_info, num_leaves);
+
+	/* There are no space reservations during log replay, bail out. */
+	if (test_bit(BTRFS_FS_LOG_RECOVERING, &fs_info->flags))
+		return;
+
+	trace_btrfs_space_reservation(fs_info, "delayed_item", node->inode_id,
+				      bytes, 0);
+	btrfs_block_rsv_release(fs_info, &fs_info->delayed_block_rsv, bytes, NULL);
 }
 
 static int btrfs_delayed_inode_reserve_metadata(
 					struct btrfs_trans_handle *trans,
 					struct btrfs_root *root,
-					struct btrfs_inode *inode,
 					struct btrfs_delayed_node *node)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
@@ -627,38 +579,23 @@ static int btrfs_delayed_inode_reserve_metadata(
 	 */
 	if (!src_rsv || (!trans->bytes_reserved &&
 			 src_rsv->type != BTRFS_BLOCK_RSV_DELALLOC)) {
-		ret = btrfs_qgroup_reserve_meta_prealloc(root,
-				fs_info->nodesize, true);
+		ret = btrfs_qgroup_reserve_meta(root, num_bytes,
+					  BTRFS_QGROUP_RSV_META_PREALLOC, true);
 		if (ret < 0)
 			return ret;
-		ret = btrfs_block_rsv_add(root, dst_rsv, num_bytes,
+		ret = btrfs_block_rsv_add(fs_info, dst_rsv, num_bytes,
 					  BTRFS_RESERVE_NO_FLUSH);
-		/*
-		 * Since we're under a transaction reserve_metadata_bytes could
-		 * try to commit the transaction which will make it return
-		 * EAGAIN to make us stop the transaction we have, so return
-		 * ENOSPC instead so that btrfs_dirty_inode knows what to do.
-		 */
-		if (ret == -EAGAIN) {
-			ret = -ENOSPC;
+		/* NO_FLUSH could only fail with -ENOSPC */
+		ASSERT(ret == 0 || ret == -ENOSPC);
+		if (ret)
 			btrfs_qgroup_free_meta_prealloc(root, num_bytes);
-		}
-		if (!ret) {
-			node->bytes_reserved = num_bytes;
-			trace_btrfs_space_reservation(fs_info,
-						      "delayed_inode",
-						      btrfs_ino(inode),
-						      num_bytes, 1);
-		} else {
-			btrfs_qgroup_free_meta_prealloc(root, fs_info->nodesize);
-		}
-		return ret;
+	} else {
+		ret = btrfs_block_rsv_migrate(src_rsv, dst_rsv, num_bytes, true);
 	}
 
-	ret = btrfs_block_rsv_migrate(src_rsv, dst_rsv, num_bytes, true);
 	if (!ret) {
 		trace_btrfs_space_reservation(fs_info, "delayed_inode",
-					      btrfs_ino(inode), num_bytes, 1);
+					      node->inode_id, num_bytes, 1);
 		node->bytes_reserved = num_bytes;
 	}
 
@@ -677,8 +614,7 @@ static void btrfs_delayed_inode_release_metadata(struct btrfs_fs_info *fs_info,
 	rsv = &fs_info->delayed_block_rsv;
 	trace_btrfs_space_reservation(fs_info, "delayed_inode",
 				      node->inode_id, node->bytes_reserved, 0);
-	btrfs_block_rsv_release(fs_info, rsv,
-				node->bytes_reserved);
+	btrfs_block_rsv_release(fs_info, rsv, node->bytes_reserved, NULL);
 	if (qgroup_free)
 		btrfs_qgroup_free_meta_prealloc(node->root,
 				node->bytes_reserved);
@@ -689,182 +625,201 @@ static void btrfs_delayed_inode_release_metadata(struct btrfs_fs_info *fs_info,
 }
 
 /*
- * This helper will insert some continuous items into the same leaf according
- * to the free space of the leaf.
+ * Insert a single delayed item or a batch of delayed items, as many as possible
+ * that fit in a leaf. The delayed items (dir index keys) are sorted by their key
+ * in the rbtree, and if there's a gap between two consecutive dir index items,
+ * then it means at some point we had delayed dir indexes to add but they got
+ * removed (by btrfs_delete_delayed_dir_index()) before we attempted to flush them
+ * into the subvolume tree. Dir index keys also have their offsets coming from a
+ * monotonically increasing counter, so we can't get new keys with an offset that
+ * fits within a gap between delayed dir index items.
  */
-static int btrfs_batch_insert_items(struct btrfs_root *root,
-				    struct btrfs_path *path,
-				    struct btrfs_delayed_item *item)
+static int btrfs_insert_delayed_item(struct btrfs_trans_handle *trans,
+				     struct btrfs_root *root,
+				     struct btrfs_path *path,
+				     struct btrfs_delayed_item *first_item)
 {
-	struct btrfs_delayed_item *curr, *next;
-	int free_space;
-	int total_data_size = 0, total_size = 0;
-	struct extent_buffer *leaf;
-	char *data_ptr;
-	struct btrfs_key *keys;
-	u32 *data_size;
-	struct list_head head;
-	int slot;
-	int nitems;
-	int i;
-	int ret = 0;
-
-	BUG_ON(!path->nodes[0]);
+	struct btrfs_fs_info *fs_info = root->fs_info;
+	struct btrfs_delayed_node *node = first_item->delayed_node;
+	LIST_HEAD(item_list);
+	struct btrfs_delayed_item *curr;
+	struct btrfs_delayed_item *next;
+	const int max_size = BTRFS_LEAF_DATA_SIZE(fs_info);
+	struct btrfs_item_batch batch;
+	struct btrfs_key first_key;
+	const u32 first_data_size = first_item->data_len;
+	int total_size;
+	char *ins_data = NULL;
+	int ret;
+	bool continuous_keys_only = false;
 
-	leaf = path->nodes[0];
-	free_space = btrfs_leaf_free_space(leaf);
-	INIT_LIST_HEAD(&head);
+	lockdep_assert_held(&node->mutex);
 
-	next = item;
-	nitems = 0;
+	/*
+	 * During normal operation the delayed index offset is continuously
+	 * increasing, so we can batch insert all items as there will not be any
+	 * overlapping keys in the tree.
+	 *
+	 * The exception to this is log replay, where we may have interleaved
+	 * offsets in the tree, so our batch needs to be continuous keys only in
+	 * order to ensure we do not end up with out of order items in our leaf.
+	 */
+	if (test_bit(BTRFS_FS_LOG_RECOVERING, &fs_info->flags))
+		continuous_keys_only = true;
 
 	/*
-	 * count the number of the continuous items that we can insert in batch
+	 * For delayed items to insert, we track reserved metadata bytes based
+	 * on the number of leaves that we will use.
+	 * See btrfs_insert_delayed_dir_index() and
+	 * btrfs_delayed_item_reserve_metadata()).
 	 */
-	while (total_size + next->data_len + sizeof(struct btrfs_item) <=
-	       free_space) {
-		total_data_size += next->data_len;
-		total_size += next->data_len + sizeof(struct btrfs_item);
-		list_add_tail(&next->tree_list, &head);
-		nitems++;
+	ASSERT(first_item->bytes_reserved == 0);
+
+	list_add_tail(&first_item->tree_list, &item_list);
+	batch.total_data_size = first_data_size;
+	batch.nr = 1;
+	total_size = first_data_size + sizeof(struct btrfs_item);
+	curr = first_item;
+
+	while (true) {
+		int next_size;
 
-		curr = next;
 		next = __btrfs_next_delayed_item(curr);
 		if (!next)
 			break;
 
-		if (!btrfs_is_continuous_delayed_item(curr, next))
+		/*
+		 * We cannot allow gaps in the key space if we're doing log
+		 * replay.
+		 */
+		if (continuous_keys_only && (next->index != curr->index + 1))
 			break;
-	}
 
-	if (!nitems) {
-		ret = 0;
-		goto out;
-	}
+		ASSERT(next->bytes_reserved == 0);
 
-	/*
-	 * we need allocate some memory space, but it might cause the task
-	 * to sleep, so we set all locked nodes in the path to blocking locks
-	 * first.
-	 */
-	btrfs_set_path_blocking(path);
+		next_size = next->data_len + sizeof(struct btrfs_item);
+		if (total_size + next_size > max_size)
+			break;
 
-	keys = kmalloc_array(nitems, sizeof(struct btrfs_key), GFP_NOFS);
-	if (!keys) {
-		ret = -ENOMEM;
-		goto out;
+		list_add_tail(&next->tree_list, &item_list);
+		batch.nr++;
+		total_size += next_size;
+		batch.total_data_size += next->data_len;
+		curr = next;
 	}
 
-	data_size = kmalloc_array(nitems, sizeof(u32), GFP_NOFS);
-	if (!data_size) {
-		ret = -ENOMEM;
-		goto error;
+	if (batch.nr == 1) {
+		first_key.objectid = node->inode_id;
+		first_key.type = BTRFS_DIR_INDEX_KEY;
+		first_key.offset = first_item->index;
+		batch.keys = &first_key;
+		batch.data_sizes = &first_data_size;
+	} else {
+		struct btrfs_key *ins_keys;
+		u32 *ins_sizes;
+		int i = 0;
+
+		ins_data = kmalloc(batch.nr * sizeof(u32) +
+				   batch.nr * sizeof(struct btrfs_key), GFP_NOFS);
+		if (!ins_data) {
+			ret = -ENOMEM;
+			goto out;
+		}
+		ins_sizes = (u32 *)ins_data;
+		ins_keys = (struct btrfs_key *)(ins_data + batch.nr * sizeof(u32));
+		batch.keys = ins_keys;
+		batch.data_sizes = ins_sizes;
+		list_for_each_entry(curr, &item_list, tree_list) {
+			ins_keys[i].objectid = node->inode_id;
+			ins_keys[i].type = BTRFS_DIR_INDEX_KEY;
+			ins_keys[i].offset = curr->index;
+			ins_sizes[i] = curr->data_len;
+			i++;
+		}
 	}
 
-	/* get keys of all the delayed items */
-	i = 0;
-	list_for_each_entry(next, &head, tree_list) {
-		keys[i] = next->key;
-		data_size[i] = next->data_len;
-		i++;
+	ret = btrfs_insert_empty_items(trans, root, path, &batch);
+	if (ret)
+		goto out;
+
+	list_for_each_entry(curr, &item_list, tree_list) {
+		char *data_ptr;
+
+		data_ptr = btrfs_item_ptr(path->nodes[0], path->slots[0], char);
+		write_extent_buffer(path->nodes[0], &curr->data,
+				    (unsigned long)data_ptr, curr->data_len);
+		path->slots[0]++;
 	}
 
-	/* insert the keys of the items */
-	setup_items_for_insert(root, path, keys, data_size,
-			       total_data_size, total_size, nitems);
+	/*
+	 * Now release our path before releasing the delayed items and their
+	 * metadata reservations, so that we don't block other tasks for more
+	 * time than needed.
+	 */
+	btrfs_release_path(path);
 
-	/* insert the dir index items */
-	slot = path->slots[0];
-	list_for_each_entry_safe(curr, next, &head, tree_list) {
-		data_ptr = btrfs_item_ptr(leaf, slot, char);
-		write_extent_buffer(leaf, &curr->data,
-				    (unsigned long)data_ptr,
-				    curr->data_len);
-		slot++;
+	ASSERT(node->index_item_leaves > 0);
 
-		btrfs_delayed_item_release_metadata(root, curr);
+	/*
+	 * For normal operations we will batch an entire leaf's worth of delayed
+	 * items, so if there are more items to process we can decrement
+	 * index_item_leaves by 1 as we inserted 1 leaf's worth of items.
+	 *
+	 * However for log replay we may not have inserted an entire leaf's
+	 * worth of items, we may have not had continuous items, so decrementing
+	 * here would mess up the index_item_leaves accounting.  For this case
+	 * only clean up the accounting when there are no items left.
+	 */
+	if (next && !continuous_keys_only) {
+		/*
+		 * We inserted one batch of items into a leaf a there are more
+		 * items to flush in a future batch, now release one unit of
+		 * metadata space from the delayed block reserve, corresponding
+		 * the leaf we just flushed to.
+		 */
+		btrfs_delayed_item_release_leaves(node, 1);
+		node->index_item_leaves--;
+	} else if (!next) {
+		/*
+		 * There are no more items to insert. We can have a number of
+		 * reserved leaves > 1 here - this happens when many dir index
+		 * items are added and then removed before they are flushed (file
+		 * names with a very short life, never span a transaction). So
+		 * release all remaining leaves.
+		 */
+		btrfs_delayed_item_release_leaves(node, node->index_item_leaves);
+		node->index_item_leaves = 0;
+	}
 
+	list_for_each_entry_safe(curr, next, &item_list, tree_list) {
 		list_del(&curr->tree_list);
 		btrfs_release_delayed_item(curr);
 	}
-
-error:
-	kfree(data_size);
-	kfree(keys);
 out:
+	kfree(ins_data);
 	return ret;
 }
 
-/*
- * This helper can just do simple insertion that needn't extend item for new
- * data, such as directory name index insertion, inode insertion.
- */
-static int btrfs_insert_delayed_item(struct btrfs_trans_handle *trans,
-				     struct btrfs_root *root,
-				     struct btrfs_path *path,
-				     struct btrfs_delayed_item *delayed_item)
-{
-	struct extent_buffer *leaf;
-	char *ptr;
-	int ret;
-
-	ret = btrfs_insert_empty_item(trans, root, path, &delayed_item->key,
-				      delayed_item->data_len);
-	if (ret < 0 && ret != -EEXIST)
-		return ret;
-
-	leaf = path->nodes[0];
-
-	ptr = btrfs_item_ptr(leaf, path->slots[0], char);
-
-	write_extent_buffer(leaf, delayed_item->data, (unsigned long)ptr,
-			    delayed_item->data_len);
-	btrfs_mark_buffer_dirty(leaf);
-
-	btrfs_delayed_item_release_metadata(root, delayed_item);
-	return 0;
-}
-
-/*
- * we insert an item first, then if there are some continuous items, we try
- * to insert those items into the same leaf.
- */
 static int btrfs_insert_delayed_items(struct btrfs_trans_handle *trans,
 				      struct btrfs_path *path,
 				      struct btrfs_root *root,
 				      struct btrfs_delayed_node *node)
 {
-	struct btrfs_delayed_item *curr, *prev;
 	int ret = 0;
 
-do_again:
-	mutex_lock(&node->mutex);
-	curr = __btrfs_first_delayed_insertion_item(node);
-	if (!curr)
-		goto insert_end;
-
-	ret = btrfs_insert_delayed_item(trans, root, path, curr);
-	if (ret < 0) {
-		btrfs_release_path(path);
-		goto insert_end;
-	}
+	while (ret == 0) {
+		struct btrfs_delayed_item *curr;
 
-	prev = curr;
-	curr = __btrfs_next_delayed_item(prev);
-	if (curr && btrfs_is_continuous_delayed_item(prev, curr)) {
-		/* insert the continuous items into the same leaf */
-		path->slots[0]++;
-		btrfs_batch_insert_items(root, path, curr);
+		mutex_lock(&node->mutex);
+		curr = __btrfs_first_delayed_insertion_item(node);
+		if (!curr) {
+			mutex_unlock(&node->mutex);
+			break;
+		}
+		ret = btrfs_insert_delayed_item(trans, root, path, curr);
+		mutex_unlock(&node->mutex);
 	}
-	btrfs_release_delayed_item(prev);
-	btrfs_mark_buffer_dirty(path->nodes[0]);
 
-	btrfs_release_path(path);
-	mutex_unlock(&node->mutex);
-	goto do_again;
-
-insert_end:
-	mutex_unlock(&node->mutex);
 	return ret;
 }
 
@@ -873,62 +828,77 @@ static int btrfs_batch_delete_items(struct btrfs_trans_handle *trans,
 				    struct btrfs_path *path,
 				    struct btrfs_delayed_item *item)
 {
+	const u64 ino = item->delayed_node->inode_id;
+	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct btrfs_delayed_item *curr, *next;
-	struct extent_buffer *leaf;
-	struct btrfs_key key;
-	struct list_head head;
-	int nitems, i, last_item;
-	int ret = 0;
+	struct extent_buffer *leaf = path->nodes[0];
+	LIST_HEAD(batch_list);
+	int nitems, slot, last_slot;
+	int ret;
+	u64 total_reserved_size = item->bytes_reserved;
 
-	BUG_ON(!path->nodes[0]);
+	ASSERT(leaf != NULL);
 
-	leaf = path->nodes[0];
+	slot = path->slots[0];
+	last_slot = btrfs_header_nritems(leaf) - 1;
+	/*
+	 * Our caller always gives us a path pointing to an existing item, so
+	 * this can not happen.
+	 */
+	ASSERT(slot <= last_slot);
+	if (WARN_ON(slot > last_slot))
+		return -ENOENT;
 
-	i = path->slots[0];
-	last_item = btrfs_header_nritems(leaf) - 1;
-	if (i > last_item)
-		return -ENOENT;	/* FIXME: Is errno suitable? */
+	nitems = 1;
+	curr = item;
+	list_add_tail(&curr->tree_list, &batch_list);
 
-	next = item;
-	INIT_LIST_HEAD(&head);
-	btrfs_item_key_to_cpu(leaf, &key, i);
-	nitems = 0;
 	/*
-	 * count the number of the dir index items that we can delete in batch
+	 * Keep checking if the next delayed item matches the next item in the
+	 * leaf - if so, we can add it to the batch of items to delete from the
+	 * leaf.
 	 */
-	while (btrfs_comp_cpu_keys(&next->key, &key) == 0) {
-		list_add_tail(&next->tree_list, &head);
-		nitems++;
+	while (slot < last_slot) {
+		struct btrfs_key key;
 
-		curr = next;
 		next = __btrfs_next_delayed_item(curr);
 		if (!next)
 			break;
 
-		if (!btrfs_is_continuous_delayed_item(curr, next))
-			break;
-
-		i++;
-		if (i > last_item)
+		slot++;
+		btrfs_item_key_to_cpu(leaf, &key, slot);
+		if (key.objectid != ino ||
+		    key.type != BTRFS_DIR_INDEX_KEY ||
+		    key.offset != next->index)
 			break;
-		btrfs_item_key_to_cpu(leaf, &key, i);
+		nitems++;
+		curr = next;
+		list_add_tail(&curr->tree_list, &batch_list);
+		total_reserved_size += curr->bytes_reserved;
 	}
 
-	if (!nitems)
-		return 0;
-
 	ret = btrfs_del_items(trans, root, path, path->slots[0], nitems);
 	if (ret)
-		goto out;
+		return ret;
 
-	list_for_each_entry_safe(curr, next, &head, tree_list) {
-		btrfs_delayed_item_release_metadata(root, curr);
+	/* In case of BTRFS_FS_LOG_RECOVERING items won't have reserved space */
+	if (total_reserved_size > 0) {
+		/*
+		 * Check btrfs_delayed_item_reserve_metadata() to see why we
+		 * don't need to release/reserve qgroup space.
+		 */
+		trace_btrfs_space_reservation(fs_info, "delayed_item", ino,
+					      total_reserved_size, 0);
+		btrfs_block_rsv_release(fs_info, &fs_info->delayed_block_rsv,
+					total_reserved_size, NULL);
+	}
+
+	list_for_each_entry_safe(curr, next, &batch_list, tree_list) {
 		list_del(&curr->tree_list);
 		btrfs_release_delayed_item(curr);
 	}
 
-out:
-	return ret;
+	return 0;
 }
 
 static int btrfs_delete_delayed_items(struct btrfs_trans_handle *trans,
@@ -936,43 +906,57 @@ static int btrfs_delete_delayed_items(struct btrfs_trans_handle *trans,
 				      struct btrfs_root *root,
 				      struct btrfs_delayed_node *node)
 {
-	struct btrfs_delayed_item *curr, *prev;
+	struct btrfs_key key;
 	int ret = 0;
 
-do_again:
-	mutex_lock(&node->mutex);
-	curr = __btrfs_first_delayed_deletion_item(node);
-	if (!curr)
-		goto delete_fail;
+	key.objectid = node->inode_id;
+	key.type = BTRFS_DIR_INDEX_KEY;
+
+	while (ret == 0) {
+		struct btrfs_delayed_item *item;
+
+		mutex_lock(&node->mutex);
+		item = __btrfs_first_delayed_deletion_item(node);
+		if (!item) {
+			mutex_unlock(&node->mutex);
+			break;
+		}
+
+		key.offset = item->index;
+		ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
+		if (ret > 0) {
+			/*
+			 * There's no matching item in the leaf. This means we
+			 * have already deleted this item in a past run of the
+			 * delayed items. We ignore errors when running delayed
+			 * items from an async context, through a work queue job
+			 * running btrfs_async_run_delayed_root(), and don't
+			 * release delayed items that failed to complete. This
+			 * is because we will retry later, and at transaction
+			 * commit time we always run delayed items and will
+			 * then deal with errors if they fail to run again.
+			 *
+			 * So just release delayed items for which we can't find
+			 * an item in the tree, and move to the next item.
+			 */
+			btrfs_release_path(path);
+			btrfs_release_delayed_item(item);
+			ret = 0;
+		} else if (ret == 0) {
+			ret = btrfs_batch_delete_items(trans, root, path, item);
+			btrfs_release_path(path);
+		}
 
-	ret = btrfs_search_slot(trans, root, &curr->key, path, -1, 1);
-	if (ret < 0)
-		goto delete_fail;
-	else if (ret > 0) {
 		/*
-		 * can't find the item which the node points to, so this node
-		 * is invalid, just drop it.
+		 * We unlock and relock on each iteration, this is to prevent
+		 * blocking other tasks for too long while we are being run from
+		 * the async context (work queue job). Those tasks are typically
+		 * running system calls like creat/mkdir/rename/unlink/etc which
+		 * need to add delayed items to this delayed node.
 		 */
-		prev = curr;
-		curr = __btrfs_next_delayed_item(prev);
-		btrfs_release_delayed_item(prev);
-		ret = 0;
-		btrfs_release_path(path);
-		if (curr) {
-			mutex_unlock(&node->mutex);
-			goto do_again;
-		} else
-			goto delete_fail;
+		mutex_unlock(&node->mutex);
 	}
 
-	btrfs_batch_delete_items(trans, root, path, curr);
-	btrfs_release_path(path);
-	mutex_unlock(&node->mutex);
-	goto do_again;
-
-delete_fail:
-	btrfs_release_path(path);
-	mutex_unlock(&node->mutex);
 	return ret;
 }
 
@@ -993,14 +977,16 @@ static void btrfs_release_delayed_inode(struct btrfs_delayed_node *delayed_node)
 
 static void btrfs_release_delayed_iref(struct btrfs_delayed_node *delayed_node)
 {
-	struct btrfs_delayed_root *delayed_root;
 
-	ASSERT(delayed_node->root);
-	clear_bit(BTRFS_DELAYED_NODE_DEL_IREF, &delayed_node->flags);
-	delayed_node->count--;
+	if (test_and_clear_bit(BTRFS_DELAYED_NODE_DEL_IREF, &delayed_node->flags)) {
+		struct btrfs_delayed_root *delayed_root;
 
-	delayed_root = delayed_node->root->fs_info->delayed_root;
-	finish_one_item(delayed_root);
+		ASSERT(delayed_node->root);
+		delayed_node->count--;
+
+		delayed_root = delayed_node->root->fs_info->delayed_root;
+		finish_one_item(delayed_root);
+	}
 }
 
 static int __btrfs_update_delayed_inode(struct btrfs_trans_handle *trans,
@@ -1025,12 +1011,10 @@ static int __btrfs_update_delayed_inode(struct btrfs_trans_handle *trans,
 		mod = 1;
 
 	ret = btrfs_lookup_inode(trans, root, path, &key, mod);
-	if (ret > 0) {
-		btrfs_release_path(path);
-		return -ENOENT;
-	} else if (ret < 0) {
-		return ret;
-	}
+	if (ret > 0)
+		ret = -ENOENT;
+	if (ret < 0)
+		goto out;
 
 	leaf = path->nodes[0];
 	inode_item = btrfs_item_ptr(leaf, path->slots[0],
@@ -1040,7 +1024,7 @@ static int __btrfs_update_delayed_inode(struct btrfs_trans_handle *trans,
 	btrfs_mark_buffer_dirty(leaf);
 
 	if (!test_bit(BTRFS_DELAYED_NODE_DEL_IREF, &node->flags))
-		goto no_iref;
+		goto out;
 
 	path->slots[0]++;
 	if (path->slots[0] >= btrfs_header_nritems(leaf))
@@ -1062,12 +1046,19 @@ again:
 	btrfs_del_item(trans, root, path);
 out:
 	btrfs_release_delayed_iref(node);
-no_iref:
 	btrfs_release_path(path);
 err_out:
 	btrfs_delayed_inode_release_metadata(fs_info, node, (ret < 0));
 	btrfs_release_delayed_inode(node);
 
+	/*
+	 * If we fail to update the delayed inode we need to abort the
+	 * transaction, because we could leave the inode with the improper
+	 * counts behind.
+	 */
+	if (ret && ret != -ENOENT)
+		btrfs_abort_transaction(trans, ret);
+
 	return ret;
 
 search:
@@ -1075,6 +1066,7 @@ search:
 
 	key.type = BTRFS_INODE_EXTREF_KEY;
 	key.offset = -1;
+
 	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
 	if (ret < 0)
 		goto err_out;
@@ -1139,13 +1131,12 @@ static int __btrfs_run_delayed_items(struct btrfs_trans_handle *trans, int nr)
 	int ret = 0;
 	bool count = (nr > 0);
 
-	if (trans->aborted)
+	if (TRANS_ABORTED(trans))
 		return -EIO;
 
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
-	path->leave_spinning = 1;
 
 	block_rsv = trans->block_rsv;
 	trans->block_rsv = &fs_info->delayed_block_rsv;
@@ -1153,7 +1144,7 @@ static int __btrfs_run_delayed_items(struct btrfs_trans_handle *trans, int nr)
 	delayed_root = fs_info->delayed_root;
 
 	curr_node = btrfs_first_delayed_node(delayed_root);
-	while (curr_node && (!count || (count && nr--))) {
+	while (curr_node && (!count || nr--)) {
 		ret = __btrfs_commit_inode_delayed_items(trans, path,
 							 curr_node);
 		if (ret) {
@@ -1210,7 +1201,6 @@ int btrfs_commit_inode_delayed_items(struct btrfs_trans_handle *trans,
 		btrfs_release_delayed_node(delayed_node);
 		return -ENOMEM;
 	}
-	path->leave_spinning = 1;
 
 	block_rsv = trans->block_rsv;
 	trans->block_rsv = &delayed_node->root->fs_info->delayed_block_rsv;
@@ -1255,7 +1245,6 @@ int btrfs_commit_inode_delayed_inode(struct btrfs_inode *inode)
 		ret = -ENOMEM;
 		goto trans_out;
 	}
-	path->leave_spinning = 1;
 
 	block_rsv = trans->block_rsv;
 	trans->block_rsv = &fs_info->delayed_block_rsv;
@@ -1324,7 +1313,6 @@ static void btrfs_async_run_delayed_root(struct btrfs_work *work)
 		if (!delayed_node)
 			break;
 
-		path->leave_spinning = 1;
 		root = delayed_node->root;
 
 		trans = btrfs_join_transaction(root);
@@ -1427,24 +1415,28 @@ int btrfs_insert_delayed_dir_index(struct btrfs_trans_handle *trans,
 				   struct btrfs_disk_key *disk_key, u8 type,
 				   u64 index)
 {
+	struct btrfs_fs_info *fs_info = trans->fs_info;
+	const unsigned int leaf_data_size = BTRFS_LEAF_DATA_SIZE(fs_info);
 	struct btrfs_delayed_node *delayed_node;
 	struct btrfs_delayed_item *delayed_item;
 	struct btrfs_dir_item *dir_item;
+	bool reserve_leaf_space;
+	u32 data_len;
 	int ret;
 
 	delayed_node = btrfs_get_or_create_delayed_node(dir);
 	if (IS_ERR(delayed_node))
 		return PTR_ERR(delayed_node);
 
-	delayed_item = btrfs_alloc_delayed_item(sizeof(*dir_item) + name_len);
+	delayed_item = btrfs_alloc_delayed_item(sizeof(*dir_item) + name_len,
+						delayed_node,
+						BTRFS_DELAYED_INSERTION_ITEM);
 	if (!delayed_item) {
 		ret = -ENOMEM;
 		goto release_node;
 	}
 
-	delayed_item->key.objectid = btrfs_ino(dir);
-	delayed_item->key.type = BTRFS_DIR_INDEX_KEY;
-	delayed_item->key.offset = index;
+	delayed_item->index = index;
 
 	dir_item = (struct btrfs_dir_item *)delayed_item->data;
 	dir_item->location = *disk_key;
@@ -1454,15 +1446,51 @@ int btrfs_insert_delayed_dir_index(struct btrfs_trans_handle *trans,
 	btrfs_set_stack_dir_type(dir_item, type);
 	memcpy((char *)(dir_item + 1), name, name_len);
 
-	ret = btrfs_delayed_item_reserve_metadata(trans, dir->root, delayed_item);
-	/*
-	 * we have reserved enough space when we start a new transaction,
-	 * so reserving metadata failure is impossible
-	 */
-	BUG_ON(ret);
+	data_len = delayed_item->data_len + sizeof(struct btrfs_item);
 
 	mutex_lock(&delayed_node->mutex);
-	ret = __btrfs_add_delayed_insertion_item(delayed_node, delayed_item);
+
+	if (delayed_node->index_item_leaves == 0 ||
+	    delayed_node->curr_index_batch_size + data_len > leaf_data_size) {
+		delayed_node->curr_index_batch_size = data_len;
+		reserve_leaf_space = true;
+	} else {
+		delayed_node->curr_index_batch_size += data_len;
+		reserve_leaf_space = false;
+	}
+
+	if (reserve_leaf_space) {
+		ret = btrfs_delayed_item_reserve_metadata(trans, delayed_item);
+		/*
+		 * Space was reserved for a dir index item insertion when we
+		 * started the transaction, so getting a failure here should be
+		 * impossible.
+		 */
+		if (WARN_ON(ret)) {
+			mutex_unlock(&delayed_node->mutex);
+			btrfs_release_delayed_item(delayed_item);
+			goto release_node;
+		}
+
+		delayed_node->index_item_leaves++;
+	} else if (!test_bit(BTRFS_FS_LOG_RECOVERING, &fs_info->flags)) {
+		const u64 bytes = btrfs_calc_insert_metadata_size(fs_info, 1);
+
+		/*
+		 * Adding the new dir index item does not require touching another
+		 * leaf, so we can release 1 unit of metadata that was previously
+		 * reserved when starting the transaction. This applies only to
+		 * the case where we had a transaction start and excludes the
+		 * transaction join case (when replaying log trees).
+		 */
+		trace_btrfs_space_reservation(fs_info, "transaction",
+					      trans->transid, bytes, 0);
+		btrfs_block_rsv_release(fs_info, trans->block_rsv, bytes, NULL);
+		ASSERT(trans->bytes_reserved >= bytes);
+		trans->bytes_reserved -= bytes;
+	}
+
+	ret = __btrfs_add_delayed_item(delayed_node, delayed_item);
 	if (unlikely(ret)) {
 		btrfs_err(trans->fs_info,
 			  "err add delayed dir index item(name: %.*s) into the insertion tree of the delayed node(root id: %llu, inode id: %llu, errno: %d)",
@@ -1479,19 +1507,48 @@ release_node:
 
 static int btrfs_delete_delayed_insertion_item(struct btrfs_fs_info *fs_info,
 					       struct btrfs_delayed_node *node,
-					       struct btrfs_key *key)
+					       u64 index)
 {
 	struct btrfs_delayed_item *item;
 
 	mutex_lock(&node->mutex);
-	item = __btrfs_lookup_delayed_insertion_item(node, key);
+	item = __btrfs_lookup_delayed_item(&node->ins_root.rb_root, index);
 	if (!item) {
 		mutex_unlock(&node->mutex);
 		return 1;
 	}
 
-	btrfs_delayed_item_release_metadata(node->root, item);
+	/*
+	 * For delayed items to insert, we track reserved metadata bytes based
+	 * on the number of leaves that we will use.
+	 * See btrfs_insert_delayed_dir_index() and
+	 * btrfs_delayed_item_reserve_metadata()).
+	 */
+	ASSERT(item->bytes_reserved == 0);
+	ASSERT(node->index_item_leaves > 0);
+
+	/*
+	 * If there's only one leaf reserved, we can decrement this item from the
+	 * current batch, otherwise we can not because we don't know which leaf
+	 * it belongs to. With the current limit on delayed items, we rarely
+	 * accumulate enough dir index items to fill more than one leaf (even
+	 * when using a leaf size of 4K).
+	 */
+	if (node->index_item_leaves == 1) {
+		const u32 data_len = item->data_len + sizeof(struct btrfs_item);
+
+		ASSERT(node->curr_index_batch_size >= data_len);
+		node->curr_index_batch_size -= data_len;
+	}
+
 	btrfs_release_delayed_item(item);
+
+	/* If we now have no more dir index items, we can release all leaves. */
+	if (RB_EMPTY_ROOT(&node->ins_root.rb_root)) {
+		btrfs_delayed_item_release_leaves(node, node->index_item_leaves);
+		node->index_item_leaves = 0;
+	}
+
 	mutex_unlock(&node->mutex);
 	return 0;
 }
@@ -1501,31 +1558,25 @@ int btrfs_delete_delayed_dir_index(struct btrfs_trans_handle *trans,
 {
 	struct btrfs_delayed_node *node;
 	struct btrfs_delayed_item *item;
-	struct btrfs_key item_key;
 	int ret;
 
 	node = btrfs_get_or_create_delayed_node(dir);
 	if (IS_ERR(node))
 		return PTR_ERR(node);
 
-	item_key.objectid = btrfs_ino(dir);
-	item_key.type = BTRFS_DIR_INDEX_KEY;
-	item_key.offset = index;
-
-	ret = btrfs_delete_delayed_insertion_item(trans->fs_info, node,
-						  &item_key);
+	ret = btrfs_delete_delayed_insertion_item(trans->fs_info, node, index);
 	if (!ret)
 		goto end;
 
-	item = btrfs_alloc_delayed_item(0);
+	item = btrfs_alloc_delayed_item(0, node, BTRFS_DELAYED_DELETION_ITEM);
 	if (!item) {
 		ret = -ENOMEM;
 		goto end;
 	}
 
-	item->key = item_key;
+	item->index = index;
 
-	ret = btrfs_delayed_item_reserve_metadata(trans, dir->root, item);
+	ret = btrfs_delayed_item_reserve_metadata(trans, item);
 	/*
 	 * we have reserved enough space when we start a new transaction,
 	 * so reserving metadata failure is impossible.
@@ -1538,7 +1589,7 @@ int btrfs_delete_delayed_dir_index(struct btrfs_trans_handle *trans,
 	}
 
 	mutex_lock(&node->mutex);
-	ret = __btrfs_add_delayed_deletion_item(node, item);
+	ret = __btrfs_add_delayed_item(node, item);
 	if (unlikely(ret)) {
 		btrfs_err(trans->fs_info,
 			  "err add delayed dir index item(index: %llu) into the deletion tree of the delayed node(root id: %llu, inode id: %llu, errno: %d)",
@@ -1590,8 +1641,8 @@ bool btrfs_readdir_get_delayed_items(struct inode *inode,
 	 * We can only do one readdir with delayed items at a time because of
 	 * item->readdir_list.
 	 */
-	inode_unlock_shared(inode);
-	inode_lock(inode);
+	btrfs_inode_unlock(inode, BTRFS_ILOCK_SHARED);
+	btrfs_inode_lock(inode, 0);
 
 	mutex_lock(&delayed_node->mutex);
 	item = __btrfs_first_delayed_insertion_item(delayed_node);
@@ -1654,9 +1705,9 @@ int btrfs_should_delete_dir_index(struct list_head *del_list,
 	int ret = 0;
 
 	list_for_each_entry(curr, del_list, readdir_list) {
-		if (curr->key.offset > index)
+		if (curr->index > index)
 			break;
-		if (curr->key.offset == index) {
+		if (curr->index == index) {
 			ret = 1;
 			break;
 		}
@@ -1690,13 +1741,13 @@ int btrfs_readdir_delayed_dir_index(struct dir_context *ctx,
 	list_for_each_entry_safe(curr, next, ins_list, readdir_list) {
 		list_del(&curr->readdir_list);
 
-		if (curr->key.offset < ctx->pos) {
+		if (curr->index < ctx->pos) {
 			if (refcount_dec_and_test(&curr->refs))
 				kfree(curr);
 			continue;
 		}
 
-		ctx->pos = curr->key.offset;
+		ctx->pos = curr->index;
 
 		di = (struct btrfs_dir_item *)curr->data;
 		name = (char *)(di + 1);
@@ -1722,6 +1773,8 @@ static void fill_stack_inode_item(struct btrfs_trans_handle *trans,
 				  struct btrfs_inode_item *inode_item,
 				  struct inode *inode)
 {
+	u64 flags;
+
 	btrfs_set_stack_inode_uid(inode_item, i_uid_read(inode));
 	btrfs_set_stack_inode_gid(inode_item, i_gid_read(inode));
 	btrfs_set_stack_inode_size(inode_item, BTRFS_I(inode)->disk_i_size);
@@ -1734,7 +1787,9 @@ static void fill_stack_inode_item(struct btrfs_trans_handle *trans,
 				       inode_peek_iversion(inode));
 	btrfs_set_stack_inode_transid(inode_item, trans->transid);
 	btrfs_set_stack_inode_rdev(inode_item, inode->i_rdev);
-	btrfs_set_stack_inode_flags(inode_item, BTRFS_I(inode)->flags);
+	flags = btrfs_inode_combine_flags(BTRFS_I(inode)->flags,
+					  BTRFS_I(inode)->ro_flags);
+	btrfs_set_stack_inode_flags(inode_item, flags);
 	btrfs_set_stack_inode_block_group(inode_item, 0);
 
 	btrfs_set_stack_timespec_sec(&inode_item->atime,
@@ -1760,6 +1815,7 @@ static void fill_stack_inode_item(struct btrfs_trans_handle *trans,
 
 int btrfs_fill_inode(struct inode *inode, u32 *rdev)
 {
+	struct btrfs_fs_info *fs_info = BTRFS_I(inode)->root->fs_info;
 	struct btrfs_delayed_node *delayed_node;
 	struct btrfs_inode_item *inode_item;
 
@@ -1779,6 +1835,8 @@ int btrfs_fill_inode(struct inode *inode, u32 *rdev)
 	i_uid_write(inode, btrfs_stack_inode_uid(inode_item));
 	i_gid_write(inode, btrfs_stack_inode_gid(inode_item));
 	btrfs_i_size_write(BTRFS_I(inode), btrfs_stack_inode_size(inode_item));
+	btrfs_inode_set_file_extent_range(BTRFS_I(inode), 0,
+			round_up(i_size_read(inode), fs_info->sectorsize));
 	inode->i_mode = btrfs_stack_inode_mode(inode_item);
 	set_nlink(inode, btrfs_stack_inode_nlink(inode_item));
 	inode_set_bytes(inode, btrfs_stack_inode_nbytes(inode_item));
@@ -1789,7 +1847,8 @@ int btrfs_fill_inode(struct inode *inode, u32 *rdev)
 				   btrfs_stack_inode_sequence(inode_item));
 	inode->i_rdev = 0;
 	*rdev = btrfs_stack_inode_rdev(inode_item);
-	BTRFS_I(inode)->flags = btrfs_stack_inode_flags(inode_item);
+	btrfs_inode_split_flags(btrfs_stack_inode_flags(inode_item),
+				&BTRFS_I(inode)->flags, &BTRFS_I(inode)->ro_flags);
 
 	inode->i_atime.tv_sec = btrfs_stack_timespec_sec(&inode_item->atime);
 	inode->i_atime.tv_nsec = btrfs_stack_timespec_nsec(&inode_item->atime);
@@ -1814,27 +1873,28 @@ int btrfs_fill_inode(struct inode *inode, u32 *rdev)
 }
 
 int btrfs_delayed_update_inode(struct btrfs_trans_handle *trans,
-			       struct btrfs_root *root, struct inode *inode)
+			       struct btrfs_root *root,
+			       struct btrfs_inode *inode)
 {
 	struct btrfs_delayed_node *delayed_node;
 	int ret = 0;
 
-	delayed_node = btrfs_get_or_create_delayed_node(BTRFS_I(inode));
+	delayed_node = btrfs_get_or_create_delayed_node(inode);
 	if (IS_ERR(delayed_node))
 		return PTR_ERR(delayed_node);
 
 	mutex_lock(&delayed_node->mutex);
 	if (test_bit(BTRFS_DELAYED_NODE_INODE_DIRTY, &delayed_node->flags)) {
-		fill_stack_inode_item(trans, &delayed_node->inode_item, inode);
+		fill_stack_inode_item(trans, &delayed_node->inode_item,
+				      &inode->vfs_inode);
 		goto release_node;
 	}
 
-	ret = btrfs_delayed_inode_reserve_metadata(trans, root, BTRFS_I(inode),
-						   delayed_node);
+	ret = btrfs_delayed_inode_reserve_metadata(trans, root, delayed_node);
 	if (ret)
 		goto release_node;
 
-	fill_stack_inode_item(trans, &delayed_node->inode_item, inode);
+	fill_stack_inode_item(trans, &delayed_node->inode_item, &inode->vfs_inode);
 	set_bit(BTRFS_DELAYED_NODE_INODE_DIRTY, &delayed_node->flags);
 	delayed_node->count++;
 	atomic_inc(&root->fs_info->delayed_root->items);
@@ -1897,12 +1957,17 @@ static void __btrfs_kill_delayed_node(struct btrfs_delayed_node *delayed_node)
 	mutex_lock(&delayed_node->mutex);
 	curr_item = __btrfs_first_delayed_insertion_item(delayed_node);
 	while (curr_item) {
-		btrfs_delayed_item_release_metadata(root, curr_item);
 		prev_item = curr_item;
 		curr_item = __btrfs_next_delayed_item(prev_item);
 		btrfs_release_delayed_item(prev_item);
 	}
 
+	if (delayed_node->index_item_leaves > 0) {
+		btrfs_delayed_item_release_leaves(delayed_node,
+					  delayed_node->index_item_leaves);
+		delayed_node->index_item_leaves = 0;
+	}
+
 	curr_item = __btrfs_first_delayed_deletion_item(delayed_node);
 	while (curr_item) {
 		btrfs_delayed_item_release_metadata(root, curr_item);
@@ -1911,8 +1976,7 @@ static void __btrfs_kill_delayed_node(struct btrfs_delayed_node *delayed_node)
 		btrfs_release_delayed_item(prev_item);
 	}
 
-	if (test_bit(BTRFS_DELAYED_NODE_DEL_IREF, &delayed_node->flags))
-		btrfs_release_delayed_iref(delayed_node);
+	btrfs_release_delayed_iref(delayed_node);
 
 	if (test_bit(BTRFS_DELAYED_NODE_INODE_DIRTY, &delayed_node->flags)) {
 		btrfs_delayed_inode_release_metadata(fs_info, delayed_node, false);
@@ -1983,3 +2047,113 @@ void btrfs_destroy_delayed_inodes(struct btrfs_fs_info *fs_info)
 	}
 }
 
+void btrfs_log_get_delayed_items(struct btrfs_inode *inode,
+				 struct list_head *ins_list,
+				 struct list_head *del_list)
+{
+	struct btrfs_delayed_node *node;
+	struct btrfs_delayed_item *item;
+
+	node = btrfs_get_delayed_node(inode);
+	if (!node)
+		return;
+
+	mutex_lock(&node->mutex);
+	item = __btrfs_first_delayed_insertion_item(node);
+	while (item) {
+		/*
+		 * It's possible that the item is already in a log list. This
+		 * can happen in case two tasks are trying to log the same
+		 * directory. For example if we have tasks A and task B:
+		 *
+		 * Task A collected the delayed items into a log list while
+		 * under the inode's log_mutex (at btrfs_log_inode()), but it
+		 * only releases the items after logging the inodes they point
+		 * to (if they are new inodes), which happens after unlocking
+		 * the log mutex;
+		 *
+		 * Task B enters btrfs_log_inode() and acquires the log_mutex
+		 * of the same directory inode, before task B releases the
+		 * delayed items. This can happen for example when logging some
+		 * inode we need to trigger logging of its parent directory, so
+		 * logging two files that have the same parent directory can
+		 * lead to this.
+		 *
+		 * If this happens, just ignore delayed items already in a log
+		 * list. All the tasks logging the directory are under a log
+		 * transaction and whichever finishes first can not sync the log
+		 * before the other completes and leaves the log transaction.
+		 */
+		if (!item->logged && list_empty(&item->log_list)) {
+			refcount_inc(&item->refs);
+			list_add_tail(&item->log_list, ins_list);
+		}
+		item = __btrfs_next_delayed_item(item);
+	}
+
+	item = __btrfs_first_delayed_deletion_item(node);
+	while (item) {
+		/* It may be non-empty, for the same reason mentioned above. */
+		if (!item->logged && list_empty(&item->log_list)) {
+			refcount_inc(&item->refs);
+			list_add_tail(&item->log_list, del_list);
+		}
+		item = __btrfs_next_delayed_item(item);
+	}
+	mutex_unlock(&node->mutex);
+
+	/*
+	 * We are called during inode logging, which means the inode is in use
+	 * and can not be evicted before we finish logging the inode. So we never
+	 * have the last reference on the delayed inode.
+	 * Also, we don't use btrfs_release_delayed_node() because that would
+	 * requeue the delayed inode (change its order in the list of prepared
+	 * nodes) and we don't want to do such change because we don't create or
+	 * delete delayed items.
+	 */
+	ASSERT(refcount_read(&node->refs) > 1);
+	refcount_dec(&node->refs);
+}
+
+void btrfs_log_put_delayed_items(struct btrfs_inode *inode,
+				 struct list_head *ins_list,
+				 struct list_head *del_list)
+{
+	struct btrfs_delayed_node *node;
+	struct btrfs_delayed_item *item;
+	struct btrfs_delayed_item *next;
+
+	node = btrfs_get_delayed_node(inode);
+	if (!node)
+		return;
+
+	mutex_lock(&node->mutex);
+
+	list_for_each_entry_safe(item, next, ins_list, log_list) {
+		item->logged = true;
+		list_del_init(&item->log_list);
+		if (refcount_dec_and_test(&item->refs))
+			kfree(item);
+	}
+
+	list_for_each_entry_safe(item, next, del_list, log_list) {
+		item->logged = true;
+		list_del_init(&item->log_list);
+		if (refcount_dec_and_test(&item->refs))
+			kfree(item);
+	}
+
+	mutex_unlock(&node->mutex);
+
+	/*
+	 * We are called during inode logging, which means the inode is in use
+	 * and can not be evicted before we finish logging the inode. So we never
+	 * have the last reference on the delayed inode.
+	 * Also, we don't use btrfs_release_delayed_node() because that would
+	 * requeue the delayed inode (change its order in the list of prepared
+	 * nodes) and we don't want to do such change because we don't create or
+	 * delete delayed items.
+	 */
+	ASSERT(refcount_read(&node->refs) > 1);
+	refcount_dec(&node->refs);
+}
diff --git a/fs/btrfs/delayed-inode.h b/fs/btrfs/delayed-inode.h
index 74ae226ffaf0..0163ca637a96 100644
--- a/fs/btrfs/delayed-inode.h
+++ b/fs/btrfs/delayed-inode.h
@@ -16,9 +16,10 @@
 #include <linux/refcount.h>
 #include "ctree.h"
 
-/* types of the delayed item */
-#define BTRFS_DELAYED_INSERTION_ITEM	1
-#define BTRFS_DELAYED_DELETION_ITEM	2
+enum btrfs_delayed_item_type {
+	BTRFS_DELAYED_INSERTION_ITEM,
+	BTRFS_DELAYED_DELETION_ITEM
+};
 
 struct btrfs_delayed_root {
 	spinlock_t lock;
@@ -58,19 +59,43 @@ struct btrfs_delayed_node {
 	u64 index_cnt;
 	unsigned long flags;
 	int count;
+	/*
+	 * The size of the next batch of dir index items to insert (if this
+	 * node is from a directory inode). Protected by @mutex.
+	 */
+	u32 curr_index_batch_size;
+	/*
+	 * Number of leaves reserved for inserting dir index items (if this
+	 * node belongs to a directory inode). This may be larger then the
+	 * actual number of leaves we end up using. Protected by @mutex.
+	 */
+	u32 index_item_leaves;
 };
 
 struct btrfs_delayed_item {
 	struct rb_node rb_node;
-	struct btrfs_key key;
+	/* Offset value of the corresponding dir index key. */
+	u64 index;
 	struct list_head tree_list;	/* used for batch insert/delete items */
 	struct list_head readdir_list;	/* used for readdir items */
+	/*
+	 * Used when logging a directory.
+	 * Insertions and deletions to this list are protected by the parent
+	 * delayed node's mutex.
+	 */
+	struct list_head log_list;
 	u64 bytes_reserved;
 	struct btrfs_delayed_node *delayed_node;
 	refcount_t refs;
-	int ins_or_del;
-	u32 data_len;
-	char data[0];
+	enum btrfs_delayed_item_type type:8;
+	/*
+	 * Track if this delayed item was already logged.
+	 * Protected by the mutex of the parent delayed inode.
+	 */
+	bool logged;
+	/* The maximum leaf size is 64K, so u16 is more than enough. */
+	u16 data_len;
+	char data[];
 };
 
 static inline void btrfs_init_delayed_root(
@@ -110,7 +135,8 @@ int btrfs_commit_inode_delayed_inode(struct btrfs_inode *inode);
 
 
 int btrfs_delayed_update_inode(struct btrfs_trans_handle *trans,
-			       struct btrfs_root *root, struct inode *inode);
+			       struct btrfs_root *root,
+			       struct btrfs_inode *inode);
 int btrfs_fill_inode(struct inode *inode, u32 *rdev);
 int btrfs_delayed_delete_inode_ref(struct btrfs_inode *inode);
 
@@ -132,6 +158,14 @@ int btrfs_should_delete_dir_index(struct list_head *del_list,
 int btrfs_readdir_delayed_dir_index(struct dir_context *ctx,
 				    struct list_head *ins_list);
 
+/* Used during directory logging. */
+void btrfs_log_get_delayed_items(struct btrfs_inode *inode,
+				 struct list_head *ins_list,
+				 struct list_head *del_list);
+void btrfs_log_put_delayed_items(struct btrfs_inode *inode,
+				 struct list_head *ins_list,
+				 struct list_head *del_list);
+
 /* for init */
 int __init btrfs_delayed_inode_init(void);
 void __cold btrfs_delayed_inode_exit(void);
diff --git a/fs/btrfs/delayed-ref.c b/fs/btrfs/delayed-ref.c
index dfdb7d4f8406..36a3debe9493 100644
--- a/fs/btrfs/delayed-ref.c
+++ b/fs/btrfs/delayed-ref.c
@@ -11,6 +11,7 @@
 #include "transaction.h"
 #include "qgroup.h"
 #include "space-info.h"
+#include "tree-mod-log.h"
 
 struct kmem_cache *btrfs_delayed_ref_head_cachep;
 struct kmem_cache *btrfs_delayed_tree_ref_cachep;
@@ -69,9 +70,10 @@ int btrfs_should_throttle_delayed_refs(struct btrfs_trans_handle *trans)
 }
 
 /**
- * btrfs_delayed_refs_rsv_release - release a ref head's reservation.
- * @fs_info - the fs_info for our fs.
- * @nr - the number of items to drop.
+ * Release a ref head's reservation
+ *
+ * @fs_info:  the filesystem
+ * @nr:       number of items to drop
  *
  * This drops the delayed ref head's count from the delayed refs rsv and frees
  * any excess reservation we had.
@@ -82,8 +84,18 @@ void btrfs_delayed_refs_rsv_release(struct btrfs_fs_info *fs_info, int nr)
 	u64 num_bytes = btrfs_calc_insert_metadata_size(fs_info, nr);
 	u64 released = 0;
 
-	released = __btrfs_block_rsv_release(fs_info, block_rsv, num_bytes,
-					     NULL);
+	/*
+	 * We have to check the mount option here because we could be enabling
+	 * the free space tree for the first time and don't have the compat_ro
+	 * option set yet.
+	 *
+	 * We need extra reservations if we have the free space tree because
+	 * we'll have to modify that tree as well.
+	 */
+	if (btrfs_test_opt(fs_info, FREE_SPACE_TREE))
+		num_bytes *= 2;
+
+	released = btrfs_block_rsv_release(fs_info, block_rsv, num_bytes, NULL);
 	if (released)
 		trace_btrfs_space_reservation(fs_info, "delayed_refs_rsv",
 					      0, released, 0);
@@ -107,18 +119,30 @@ void btrfs_update_delayed_refs_rsv(struct btrfs_trans_handle *trans)
 
 	num_bytes = btrfs_calc_insert_metadata_size(fs_info,
 						    trans->delayed_ref_updates);
+	/*
+	 * We have to check the mount option here because we could be enabling
+	 * the free space tree for the first time and don't have the compat_ro
+	 * option set yet.
+	 *
+	 * We need extra reservations if we have the free space tree because
+	 * we'll have to modify that tree as well.
+	 */
+	if (btrfs_test_opt(fs_info, FREE_SPACE_TREE))
+		num_bytes *= 2;
+
 	spin_lock(&delayed_rsv->lock);
 	delayed_rsv->size += num_bytes;
-	delayed_rsv->full = 0;
+	delayed_rsv->full = false;
 	spin_unlock(&delayed_rsv->lock);
 	trans->delayed_ref_updates = 0;
 }
 
 /**
- * btrfs_migrate_to_delayed_refs_rsv - transfer bytes to our delayed refs rsv.
- * @fs_info - the fs info for our fs.
- * @src - the source block rsv to transfer from.
- * @num_bytes - the number of bytes to transfer.
+ * Transfer bytes to our delayed refs rsv
+ *
+ * @fs_info:   the filesystem
+ * @src:       source block rsv to transfer from
+ * @num_bytes: number of bytes to transfer
  *
  * This transfers up to the num_bytes amount from the src rsv to the
  * delayed_refs_rsv.  Any extra bytes are returned to the space info.
@@ -151,7 +175,7 @@ void btrfs_migrate_to_delayed_refs_rsv(struct btrfs_fs_info *fs_info,
 	if (num_bytes)
 		delayed_refs_rsv->reserved += num_bytes;
 	if (delayed_refs_rsv->reserved >= delayed_refs_rsv->size)
-		delayed_refs_rsv->full = 1;
+		delayed_refs_rsv->full = true;
 	spin_unlock(&delayed_refs_rsv->lock);
 
 	if (num_bytes)
@@ -163,9 +187,10 @@ void btrfs_migrate_to_delayed_refs_rsv(struct btrfs_fs_info *fs_info,
 }
 
 /**
- * btrfs_delayed_refs_rsv_refill - refill based on our delayed refs usage.
- * @fs_info - the fs_info for our fs.
- * @flush - control how we can flush for this reservation.
+ * Refill based on our delayed refs usage
+ *
+ * @fs_info: the filesystem
+ * @flush:   control how we can flush for this reservation.
  *
  * This will refill the delayed block_rsv up to 1 items size worth of space and
  * will return -ENOSPC if we can't make the reservation.
@@ -188,8 +213,7 @@ int btrfs_delayed_refs_rsv_refill(struct btrfs_fs_info *fs_info,
 	if (!num_bytes)
 		return 0;
 
-	ret = btrfs_reserve_metadata_bytes(fs_info->extent_root, block_rsv,
-					   num_bytes, flush);
+	ret = btrfs_reserve_metadata_bytes(fs_info, block_rsv, num_bytes, flush);
 	if (ret)
 		return ret;
 	btrfs_block_rsv_add_bytes(block_rsv, num_bytes, 0);
@@ -492,16 +516,7 @@ void btrfs_merge_delayed_refs(struct btrfs_trans_handle *trans,
 	if (head->is_data)
 		return;
 
-	read_lock(&fs_info->tree_mod_log_lock);
-	if (!list_empty(&fs_info->tree_mod_seq_list)) {
-		struct seq_list *elem;
-
-		elem = list_first_entry(&fs_info->tree_mod_seq_list,
-					struct seq_list, list);
-		seq = elem->seq;
-	}
-	read_unlock(&fs_info->tree_mod_log_lock);
-
+	seq = btrfs_tree_mod_log_lowest_seq(fs_info);
 again:
 	for (node = rb_first_cached(&head->ref_tree); node;
 	     node = rb_next(node)) {
@@ -515,23 +530,16 @@ again:
 
 int btrfs_check_delayed_seq(struct btrfs_fs_info *fs_info, u64 seq)
 {
-	struct seq_list *elem;
 	int ret = 0;
+	u64 min_seq = btrfs_tree_mod_log_lowest_seq(fs_info);
 
-	read_lock(&fs_info->tree_mod_log_lock);
-	if (!list_empty(&fs_info->tree_mod_seq_list)) {
-		elem = list_first_entry(&fs_info->tree_mod_seq_list,
-					struct seq_list, list);
-		if (seq >= elem->seq) {
-			btrfs_debug(fs_info,
-				"holding back delayed_ref %#x.%x, lowest is %#x.%x",
-				(u32)(seq >> 32), (u32)seq,
-				(u32)(elem->seq >> 32), (u32)elem->seq);
-			ret = 1;
-		}
+	if (min_seq != 0 && seq >= min_seq) {
+		btrfs_debug(fs_info,
+			    "holding back delayed_ref %llu, lowest is %llu",
+			    seq, min_seq);
+		ret = 1;
 	}
 
-	read_unlock(&fs_info->tree_mod_log_lock);
 	return ret;
 }
 
@@ -649,8 +657,7 @@ inserted:
  */
 static noinline void update_existing_head_ref(struct btrfs_trans_handle *trans,
 			 struct btrfs_delayed_ref_head *existing,
-			 struct btrfs_delayed_ref_head *update,
-			 int *old_ref_mod_ret)
+			 struct btrfs_delayed_ref_head *update)
 {
 	struct btrfs_delayed_ref_root *delayed_refs =
 		&trans->transaction->delayed_refs;
@@ -702,8 +709,6 @@ static noinline void update_existing_head_ref(struct btrfs_trans_handle *trans,
 	 * currently, for refs we just added we know we're a-ok.
 	 */
 	old_ref_mod = existing->total_ref_mod;
-	if (old_ref_mod_ret)
-		*old_ref_mod_ret = old_ref_mod;
 	existing->ref_mod += update->ref_mod;
 	existing->total_ref_mod += update->ref_mod;
 
@@ -725,6 +730,7 @@ static noinline void update_existing_head_ref(struct btrfs_trans_handle *trans,
 			trans->delayed_ref_updates += csum_leaves;
 		}
 	}
+
 	spin_unlock(&existing->lock);
 }
 
@@ -799,8 +805,7 @@ static noinline struct btrfs_delayed_ref_head *
 add_delayed_ref_head(struct btrfs_trans_handle *trans,
 		     struct btrfs_delayed_ref_head *head_ref,
 		     struct btrfs_qgroup_extent_record *qrecord,
-		     int action, int *qrecord_inserted_ret,
-		     int *old_ref_mod, int *new_ref_mod)
+		     int action, int *qrecord_inserted_ret)
 {
 	struct btrfs_delayed_ref_head *existing;
 	struct btrfs_delayed_ref_root *delayed_refs;
@@ -822,8 +827,7 @@ add_delayed_ref_head(struct btrfs_trans_handle *trans,
 	existing = htree_insert(&delayed_refs->href_root,
 				&head_ref->href_node);
 	if (existing) {
-		update_existing_head_ref(trans, existing, head_ref,
-					 old_ref_mod);
+		update_existing_head_ref(trans, existing, head_ref);
 		/*
 		 * we've updated the existing ref, free the newly
 		 * allocated ref
@@ -831,8 +835,6 @@ add_delayed_ref_head(struct btrfs_trans_handle *trans,
 		kmem_cache_free(btrfs_delayed_ref_head_cachep, head_ref);
 		head_ref = existing;
 	} else {
-		if (old_ref_mod)
-			*old_ref_mod = 0;
 		if (head_ref->is_data && head_ref->ref_mod < 0) {
 			delayed_refs->pending_csums += head_ref->num_bytes;
 			trans->delayed_ref_updates +=
@@ -846,8 +848,6 @@ add_delayed_ref_head(struct btrfs_trans_handle *trans,
 	}
 	if (qrecord_inserted_ret)
 		*qrecord_inserted_ret = qrecord_inserted;
-	if (new_ref_mod)
-		*new_ref_mod = head_ref->total_ref_mod;
 
 	return head_ref;
 }
@@ -910,8 +910,7 @@ static void init_delayed_ref_common(struct btrfs_fs_info *fs_info,
  */
 int btrfs_add_delayed_tree_ref(struct btrfs_trans_handle *trans,
 			       struct btrfs_ref *generic_ref,
-			       struct btrfs_delayed_extent_op *extent_op,
-			       int *old_ref_mod, int *new_ref_mod)
+			       struct btrfs_delayed_extent_op *extent_op)
 {
 	struct btrfs_fs_info *fs_info = trans->fs_info;
 	struct btrfs_delayed_tree_ref *ref;
@@ -928,10 +927,9 @@ int btrfs_add_delayed_tree_ref(struct btrfs_trans_handle *trans,
 	u64 parent = generic_ref->parent;
 	u8 ref_type;
 
-	is_system = (generic_ref->real_root == BTRFS_CHUNK_TREE_OBJECTID);
+	is_system = (generic_ref->tree_ref.owning_root == BTRFS_CHUNK_TREE_OBJECTID);
 
 	ASSERT(generic_ref->type == BTRFS_REF_METADATA && generic_ref->action);
-	BUG_ON(extent_op && extent_op->is_data);
 	ref = kmem_cache_alloc(btrfs_delayed_tree_ref_cachep, GFP_NOFS);
 	if (!ref)
 		return -ENOMEM;
@@ -943,8 +941,6 @@ int btrfs_add_delayed_tree_ref(struct btrfs_trans_handle *trans,
 	}
 
 	if (test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags) &&
-	    is_fstree(generic_ref->real_root) &&
-	    is_fstree(generic_ref->tree_ref.root) &&
 	    !generic_ref->skip_qgroup) {
 		record = kzalloc(sizeof(*record), GFP_NOFS);
 		if (!record) {
@@ -960,14 +956,15 @@ int btrfs_add_delayed_tree_ref(struct btrfs_trans_handle *trans,
 		ref_type = BTRFS_TREE_BLOCK_REF_KEY;
 
 	init_delayed_ref_common(fs_info, &ref->node, bytenr, num_bytes,
-				generic_ref->tree_ref.root, action, ref_type);
-	ref->root = generic_ref->tree_ref.root;
+				generic_ref->tree_ref.owning_root, action,
+				ref_type);
+	ref->root = generic_ref->tree_ref.owning_root;
 	ref->parent = parent;
 	ref->level = level;
 
 	init_delayed_ref_head(head_ref, record, bytenr, num_bytes,
-			      generic_ref->tree_ref.root, 0, action, false,
-			      is_system);
+			      generic_ref->tree_ref.owning_root, 0, action,
+			      false, is_system);
 	head_ref->extent_op = extent_op;
 
 	delayed_refs = &trans->transaction->delayed_refs;
@@ -978,8 +975,7 @@ int btrfs_add_delayed_tree_ref(struct btrfs_trans_handle *trans,
 	 * the spin lock
 	 */
 	head_ref = add_delayed_ref_head(trans, head_ref, record,
-					action, &qrecord_inserted,
-					old_ref_mod, new_ref_mod);
+					action, &qrecord_inserted);
 
 	ret = insert_delayed_ref(trans, delayed_refs, head_ref, &ref->node);
 	spin_unlock(&delayed_refs->lock);
@@ -997,7 +993,7 @@ int btrfs_add_delayed_tree_ref(struct btrfs_trans_handle *trans,
 		kmem_cache_free(btrfs_delayed_tree_ref_cachep, ref);
 
 	if (qrecord_inserted)
-		btrfs_qgroup_trace_extent_post(fs_info, record);
+		btrfs_qgroup_trace_extent_post(trans, record);
 
 	return 0;
 }
@@ -1007,8 +1003,7 @@ int btrfs_add_delayed_tree_ref(struct btrfs_trans_handle *trans,
  */
 int btrfs_add_delayed_data_ref(struct btrfs_trans_handle *trans,
 			       struct btrfs_ref *generic_ref,
-			       u64 reserved, int *old_ref_mod,
-			       int *new_ref_mod)
+			       u64 reserved)
 {
 	struct btrfs_fs_info *fs_info = trans->fs_info;
 	struct btrfs_delayed_data_ref *ref;
@@ -1021,7 +1016,7 @@ int btrfs_add_delayed_data_ref(struct btrfs_trans_handle *trans,
 	u64 bytenr = generic_ref->bytenr;
 	u64 num_bytes = generic_ref->len;
 	u64 parent = generic_ref->parent;
-	u64 ref_root = generic_ref->data_ref.ref_root;
+	u64 ref_root = generic_ref->data_ref.owning_root;
 	u64 owner = generic_ref->data_ref.ino;
 	u64 offset = generic_ref->data_ref.offset;
 	u8 ref_type;
@@ -1050,8 +1045,6 @@ int btrfs_add_delayed_data_ref(struct btrfs_trans_handle *trans,
 	}
 
 	if (test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags) &&
-	    is_fstree(ref_root) &&
-	    is_fstree(generic_ref->real_root) &&
 	    !generic_ref->skip_qgroup) {
 		record = kzalloc(sizeof(*record), GFP_NOFS);
 		if (!record) {
@@ -1074,8 +1067,7 @@ int btrfs_add_delayed_data_ref(struct btrfs_trans_handle *trans,
 	 * the spin lock
 	 */
 	head_ref = add_delayed_ref_head(trans, head_ref, record,
-					action, &qrecord_inserted,
-					old_ref_mod, new_ref_mod);
+					action, &qrecord_inserted);
 
 	ret = insert_delayed_ref(trans, delayed_refs, head_ref, &ref->node);
 	spin_unlock(&delayed_refs->lock);
@@ -1094,7 +1086,7 @@ int btrfs_add_delayed_data_ref(struct btrfs_trans_handle *trans,
 
 
 	if (qrecord_inserted)
-		return btrfs_qgroup_trace_extent_post(fs_info, record);
+		return btrfs_qgroup_trace_extent_post(trans, record);
 	return 0;
 }
 
@@ -1110,15 +1102,14 @@ int btrfs_add_delayed_extent_op(struct btrfs_trans_handle *trans,
 		return -ENOMEM;
 
 	init_delayed_ref_head(head_ref, NULL, bytenr, num_bytes, 0, 0,
-			      BTRFS_UPDATE_DELAYED_HEAD, extent_op->is_data,
-			      false);
+			      BTRFS_UPDATE_DELAYED_HEAD, false, false);
 	head_ref->extent_op = extent_op;
 
 	delayed_refs = &trans->transaction->delayed_refs;
 	spin_lock(&delayed_refs->lock);
 
 	add_delayed_ref_head(trans, head_ref, NULL, BTRFS_UPDATE_DELAYED_HEAD,
-			     NULL, NULL, NULL);
+			     NULL);
 
 	spin_unlock(&delayed_refs->lock);
 
diff --git a/fs/btrfs/delayed-ref.h b/fs/btrfs/delayed-ref.h
index 1c977e6d45dc..d6304b690ec4 100644
--- a/fs/btrfs/delayed-ref.h
+++ b/fs/btrfs/delayed-ref.h
@@ -58,7 +58,6 @@ struct btrfs_delayed_extent_op {
 	u8 level;
 	bool update_key;
 	bool update_flags;
-	bool is_data;
 	u64 flags_to_set;
 };
 
@@ -135,6 +134,11 @@ struct btrfs_delayed_data_ref {
 	u64 offset;
 };
 
+enum btrfs_delayed_ref_flags {
+	/* Indicate that we are flushing delayed refs for the commit */
+	BTRFS_DELAYED_REFS_FLUSHING,
+};
+
 struct btrfs_delayed_ref_root {
 	/* head ref rbtree */
 	struct rb_root_cached href_root;
@@ -158,12 +162,7 @@ struct btrfs_delayed_ref_root {
 
 	u64 pending_csums;
 
-	/*
-	 * set when the tree is flushing before a transaction commit,
-	 * used by the throttling code to decide if new updates need
-	 * to be run right away
-	 */
-	int flushing;
+	unsigned long flags;
 
 	u64 run_delayed_start;
 
@@ -186,8 +185,8 @@ enum btrfs_ref_type {
 struct btrfs_data_ref {
 	/* For EXTENT_DATA_REF */
 
-	/* Root which refers to this data extent */
-	u64 ref_root;
+	/* Original root this data extent belongs to */
+	u64 owning_root;
 
 	/* Inode which refers to this data extent */
 	u64 ino;
@@ -210,11 +209,11 @@ struct btrfs_tree_ref {
 	int level;
 
 	/*
-	 * Root which refers to this tree block.
+	 * Root which owns this tree block.
 	 *
 	 * For TREE_BLOCK_REF (skinny metadata, either inline or keyed)
 	 */
-	u64 root;
+	u64 owning_root;
 
 	/* For non-skinny metadata, no special member needed */
 };
@@ -231,17 +230,10 @@ struct btrfs_ref {
 	 */
 	bool skip_qgroup;
 
-	/*
-	 * Optional. For which root is this modification.
-	 * Mostly used for qgroup optimization.
-	 *
-	 * When unset, data/tree ref init code will populate it.
-	 * In certain cases, we're modifying reference for a different root.
-	 * E.g. COW fs tree blocks for balance.
-	 * In that case, tree_ref::root will be fs tree, but we're doing this
-	 * for reloc tree, then we should set @real_root to reloc tree.
-	 */
+#ifdef CONFIG_BTRFS_FS_REF_VERIFY
+	/* Through which root is this modification. */
 	u64 real_root;
+#endif
 	u64 bytenr;
 	u64 len;
 
@@ -271,26 +263,40 @@ static inline void btrfs_init_generic_ref(struct btrfs_ref *generic_ref,
 }
 
 static inline void btrfs_init_tree_ref(struct btrfs_ref *generic_ref,
-				int level, u64 root)
+				int level, u64 root, u64 mod_root, bool skip_qgroup)
 {
+#ifdef CONFIG_BTRFS_FS_REF_VERIFY
 	/* If @real_root not set, use @root as fallback */
-	if (!generic_ref->real_root)
-		generic_ref->real_root = root;
+	generic_ref->real_root = mod_root ?: root;
+#endif
 	generic_ref->tree_ref.level = level;
-	generic_ref->tree_ref.root = root;
+	generic_ref->tree_ref.owning_root = root;
 	generic_ref->type = BTRFS_REF_METADATA;
+	if (skip_qgroup || !(is_fstree(root) &&
+			     (!mod_root || is_fstree(mod_root))))
+		generic_ref->skip_qgroup = true;
+	else
+		generic_ref->skip_qgroup = false;
+
 }
 
 static inline void btrfs_init_data_ref(struct btrfs_ref *generic_ref,
-				u64 ref_root, u64 ino, u64 offset)
+				u64 ref_root, u64 ino, u64 offset, u64 mod_root,
+				bool skip_qgroup)
 {
+#ifdef CONFIG_BTRFS_FS_REF_VERIFY
 	/* If @real_root not set, use @root as fallback */
-	if (!generic_ref->real_root)
-		generic_ref->real_root = ref_root;
-	generic_ref->data_ref.ref_root = ref_root;
+	generic_ref->real_root = mod_root ?: ref_root;
+#endif
+	generic_ref->data_ref.owning_root = ref_root;
 	generic_ref->data_ref.ino = ino;
 	generic_ref->data_ref.offset = offset;
 	generic_ref->type = BTRFS_REF_DATA;
+	if (skip_qgroup || !(is_fstree(ref_root) &&
+			     (!mod_root || is_fstree(mod_root))))
+		generic_ref->skip_qgroup = true;
+	else
+		generic_ref->skip_qgroup = false;
 }
 
 static inline struct btrfs_delayed_extent_op *
@@ -326,6 +332,16 @@ static inline void btrfs_put_delayed_ref(struct btrfs_delayed_ref_node *ref)
 	}
 }
 
+static inline u64 btrfs_ref_head_to_space_flags(
+				struct btrfs_delayed_ref_head *head_ref)
+{
+	if (head_ref->is_data)
+		return BTRFS_BLOCK_GROUP_DATA;
+	else if (head_ref->is_system)
+		return BTRFS_BLOCK_GROUP_SYSTEM;
+	return BTRFS_BLOCK_GROUP_METADATA;
+}
+
 static inline void btrfs_put_delayed_ref_head(struct btrfs_delayed_ref_head *head)
 {
 	if (refcount_dec_and_test(&head->refs))
@@ -334,12 +350,10 @@ static inline void btrfs_put_delayed_ref_head(struct btrfs_delayed_ref_head *hea
 
 int btrfs_add_delayed_tree_ref(struct btrfs_trans_handle *trans,
 			       struct btrfs_ref *generic_ref,
-			       struct btrfs_delayed_extent_op *extent_op,
-			       int *old_ref_mod, int *new_ref_mod);
+			       struct btrfs_delayed_extent_op *extent_op);
 int btrfs_add_delayed_data_ref(struct btrfs_trans_handle *trans,
 			       struct btrfs_ref *generic_ref,
-			       u64 reserved, int *old_ref_mod,
-			       int *new_ref_mod);
+			       u64 reserved);
 int btrfs_add_delayed_extent_op(struct btrfs_trans_handle *trans,
 				u64 bytenr, u64 num_bytes,
 				struct btrfs_delayed_extent_op *extent_op);
diff --git a/fs/btrfs/dev-replace.c b/fs/btrfs/dev-replace.c
index 2ca2a09d0e23..61e58066b5fd 100644
--- a/fs/btrfs/dev-replace.c
+++ b/fs/btrfs/dev-replace.c
@@ -21,17 +21,56 @@
 #include "rcu-string.h"
 #include "dev-replace.h"
 #include "sysfs.h"
+#include "zoned.h"
+#include "block-group.h"
+
+/*
+ * Device replace overview
+ *
+ * [Objective]
+ * To copy all extents (both new and on-disk) from source device to target
+ * device, while still keeping the filesystem read-write.
+ *
+ * [Method]
+ * There are two main methods involved:
+ *
+ * - Write duplication
+ *
+ *   All new writes will be written to both target and source devices, so even
+ *   if replace gets canceled, sources device still contains up-to-date data.
+ *
+ *   Location:		handle_ops_on_dev_replace() from __btrfs_map_block()
+ *   Start:		btrfs_dev_replace_start()
+ *   End:		btrfs_dev_replace_finishing()
+ *   Content:		Latest data/metadata
+ *
+ * - Copy existing extents
+ *
+ *   This happens by re-using scrub facility, as scrub also iterates through
+ *   existing extents from commit root.
+ *
+ *   Location:		scrub_write_block_to_dev_replace() from
+ *   			scrub_block_complete()
+ *   Content:		Data/meta from commit root.
+ *
+ * Due to the content difference, we need to avoid nocow write when dev-replace
+ * is happening.  This is done by marking the block group read-only and waiting
+ * for NOCOW writes.
+ *
+ * After replace is done, the finishing part is done by swapping the target and
+ * source devices.
+ *
+ *   Location:		btrfs_dev_replace_update_device_in_mapping_tree() from
+ *   			btrfs_dev_replace_finishing()
+ */
 
 static int btrfs_dev_replace_finishing(struct btrfs_fs_info *fs_info,
 				       int scrub_ret);
-static void btrfs_dev_replace_update_device_in_mapping_tree(
-						struct btrfs_fs_info *fs_info,
-						struct btrfs_device *srcdev,
-						struct btrfs_device *tgtdev);
 static int btrfs_dev_replace_kthread(void *data);
 
 int btrfs_init_dev_replace(struct btrfs_fs_info *fs_info)
 {
+	struct btrfs_dev_lookup_args args = { .devid = BTRFS_DEV_REPLACE_DEVID };
 	struct btrfs_key key;
 	struct btrfs_root *dev_root = fs_info->dev_root;
 	struct btrfs_dev_replace *dev_replace = &fs_info->dev_replace;
@@ -43,6 +82,9 @@ int btrfs_init_dev_replace(struct btrfs_fs_info *fs_info)
 	struct btrfs_dev_replace_item *ptr;
 	u64 src_devid;
 
+	if (!dev_root)
+		return 0;
+
 	path = btrfs_alloc_path();
 	if (!path) {
 		ret = -ENOMEM;
@@ -55,6 +97,16 @@ int btrfs_init_dev_replace(struct btrfs_fs_info *fs_info)
 	ret = btrfs_search_slot(NULL, dev_root, &key, path, 0, 0);
 	if (ret) {
 no_valid_dev_replace_entry_found:
+		/*
+		 * We don't have a replace item or it's corrupted.  If there is
+		 * a replace target, fail the mount.
+		 */
+		if (btrfs_find_device(fs_info->fs_devices, &args)) {
+			btrfs_err(fs_info,
+			"found replace target device without a valid replace item");
+			ret = -EUCLEAN;
+			goto out;
+		}
 		ret = 0;
 		dev_replace->replace_state =
 			BTRFS_IOCTL_DEV_REPLACE_STATE_NEVER_STARTED;
@@ -76,7 +128,7 @@ no_valid_dev_replace_entry_found:
 	}
 	slot = path->slots[0];
 	eb = path->nodes[0];
-	item_size = btrfs_item_size_nr(eb, slot);
+	item_size = btrfs_item_size(eb, slot);
 	ptr = btrfs_item_ptr(eb, slot, struct btrfs_dev_replace_item);
 
 	if (item_size != sizeof(struct btrfs_dev_replace_item)) {
@@ -107,16 +159,25 @@ no_valid_dev_replace_entry_found:
 	case BTRFS_IOCTL_DEV_REPLACE_STATE_NEVER_STARTED:
 	case BTRFS_IOCTL_DEV_REPLACE_STATE_FINISHED:
 	case BTRFS_IOCTL_DEV_REPLACE_STATE_CANCELED:
-		dev_replace->srcdev = NULL;
-		dev_replace->tgtdev = NULL;
+		/*
+		 * We don't have an active replace item but if there is a
+		 * replace target, fail the mount.
+		 */
+		if (btrfs_find_device(fs_info->fs_devices, &args)) {
+			btrfs_err(fs_info,
+"replace without active item, run 'device scan --forget' on the target device");
+			ret = -EUCLEAN;
+		} else {
+			dev_replace->srcdev = NULL;
+			dev_replace->tgtdev = NULL;
+		}
 		break;
 	case BTRFS_IOCTL_DEV_REPLACE_STATE_STARTED:
 	case BTRFS_IOCTL_DEV_REPLACE_STATE_SUSPENDED:
-		dev_replace->srcdev = btrfs_find_device(fs_info->fs_devices,
-						src_devid, NULL, NULL, true);
-		dev_replace->tgtdev = btrfs_find_device(fs_info->fs_devices,
-							BTRFS_DEV_REPLACE_DEVID,
-							NULL, NULL, true);
+		dev_replace->tgtdev = btrfs_find_device(fs_info->fs_devices, &args);
+		args.devid = src_devid;
+		dev_replace->srcdev = btrfs_find_device(fs_info->fs_devices, &args);
+
 		/*
 		 * allow 'btrfs dev replace_cancel' if src/tgt device is
 		 * missing
@@ -182,15 +243,15 @@ static int btrfs_init_dev_replace_tgtdev(struct btrfs_fs_info *fs_info,
 				  struct btrfs_device *srcdev,
 				  struct btrfs_device **device_out)
 {
+	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
 	struct btrfs_device *device;
 	struct block_device *bdev;
-	struct list_head *devices;
 	struct rcu_string *name;
 	u64 devid = BTRFS_DEV_REPLACE_DEVID;
 	int ret = 0;
 
 	*device_out = NULL;
-	if (fs_info->fs_devices->seeding) {
+	if (srcdev->fs_devices->seeding) {
 		btrfs_err(fs_info, "the filesystem is a seed filesystem!");
 		return -EINVAL;
 	}
@@ -202,10 +263,16 @@ static int btrfs_init_dev_replace_tgtdev(struct btrfs_fs_info *fs_info,
 		return PTR_ERR(bdev);
 	}
 
+	if (!btrfs_check_device_zone_type(fs_info, bdev)) {
+		btrfs_err(fs_info,
+		"dev-replace: zoned type of target device mismatch with filesystem");
+		ret = -EINVAL;
+		goto error;
+	}
+
 	sync_blockdev(bdev);
 
-	devices = &fs_info->fs_devices->devices;
-	list_for_each_entry(device, devices, dev_list) {
+	list_for_each_entry(device, &fs_devices->devices, dev_list) {
 		if (device->bdev == bdev) {
 			btrfs_err(fs_info,
 				  "target device is in the filesystem!");
@@ -215,8 +282,7 @@ static int btrfs_init_dev_replace_tgtdev(struct btrfs_fs_info *fs_info,
 	}
 
 
-	if (i_size_read(bdev->bd_inode) <
-	    btrfs_device_get_total_bytes(srcdev)) {
+	if (bdev_nr_bytes(bdev) < btrfs_device_get_total_bytes(srcdev)) {
 		btrfs_err(fs_info,
 			  "target device is smaller than source device!");
 		ret = -EINVAL;
@@ -237,6 +303,9 @@ static int btrfs_init_dev_replace_tgtdev(struct btrfs_fs_info *fs_info,
 		goto error;
 	}
 	rcu_assign_pointer(device->name, name);
+	ret = lookup_bdev(device_path, &device->devt);
+	if (ret)
+		goto error;
 
 	set_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state);
 	device->generation = 0;
@@ -255,13 +324,17 @@ static int btrfs_init_dev_replace_tgtdev(struct btrfs_fs_info *fs_info,
 	device->mode = FMODE_EXCL;
 	device->dev_stats_valid = 1;
 	set_blocksize(device->bdev, BTRFS_BDEV_BLOCKSIZE);
-	device->fs_devices = fs_info->fs_devices;
+	device->fs_devices = fs_devices;
+
+	ret = btrfs_get_dev_zone_info(device, false);
+	if (ret)
+		goto error;
 
-	mutex_lock(&fs_info->fs_devices->device_list_mutex);
-	list_add(&device->dev_list, &fs_info->fs_devices->devices);
-	fs_info->fs_devices->num_devices++;
-	fs_info->fs_devices->open_devices++;
-	mutex_unlock(&fs_info->fs_devices->device_list_mutex);
+	mutex_lock(&fs_devices->device_list_mutex);
+	list_add(&device->dev_list, &fs_devices->devices);
+	fs_devices->num_devices++;
+	fs_devices->open_devices++;
+	mutex_unlock(&fs_devices->device_list_mutex);
 
 	*device_out = device;
 	return 0;
@@ -312,7 +385,7 @@ int btrfs_run_dev_replace(struct btrfs_trans_handle *trans)
 	}
 
 	if (ret == 0 &&
-	    btrfs_item_size_nr(path->nodes[0], path->slots[0]) < sizeof(*ptr)) {
+	    btrfs_item_size(path->nodes[0], path->slots[0]) < sizeof(*ptr)) {
 		/*
 		 * need to delete old one and insert a new one.
 		 * Since no attempt is made to recover any old state, if the
@@ -391,6 +464,154 @@ static char* btrfs_dev_name(struct btrfs_device *device)
 		return rcu_str_deref(device->name);
 }
 
+static int mark_block_group_to_copy(struct btrfs_fs_info *fs_info,
+				    struct btrfs_device *src_dev)
+{
+	struct btrfs_path *path;
+	struct btrfs_key key;
+	struct btrfs_key found_key;
+	struct btrfs_root *root = fs_info->dev_root;
+	struct btrfs_dev_extent *dev_extent = NULL;
+	struct btrfs_block_group *cache;
+	struct btrfs_trans_handle *trans;
+	int iter_ret = 0;
+	int ret = 0;
+	u64 chunk_offset;
+
+	/* Do not use "to_copy" on non zoned filesystem for now */
+	if (!btrfs_is_zoned(fs_info))
+		return 0;
+
+	mutex_lock(&fs_info->chunk_mutex);
+
+	/* Ensure we don't have pending new block group */
+	spin_lock(&fs_info->trans_lock);
+	while (fs_info->running_transaction &&
+	       !list_empty(&fs_info->running_transaction->dev_update_list)) {
+		spin_unlock(&fs_info->trans_lock);
+		mutex_unlock(&fs_info->chunk_mutex);
+		trans = btrfs_attach_transaction(root);
+		if (IS_ERR(trans)) {
+			ret = PTR_ERR(trans);
+			mutex_lock(&fs_info->chunk_mutex);
+			if (ret == -ENOENT) {
+				spin_lock(&fs_info->trans_lock);
+				continue;
+			} else {
+				goto unlock;
+			}
+		}
+
+		ret = btrfs_commit_transaction(trans);
+		mutex_lock(&fs_info->chunk_mutex);
+		if (ret)
+			goto unlock;
+
+		spin_lock(&fs_info->trans_lock);
+	}
+	spin_unlock(&fs_info->trans_lock);
+
+	path = btrfs_alloc_path();
+	if (!path) {
+		ret = -ENOMEM;
+		goto unlock;
+	}
+
+	path->reada = READA_FORWARD;
+	path->search_commit_root = 1;
+	path->skip_locking = 1;
+
+	key.objectid = src_dev->devid;
+	key.type = BTRFS_DEV_EXTENT_KEY;
+	key.offset = 0;
+
+	btrfs_for_each_slot(root, &key, &found_key, path, iter_ret) {
+		struct extent_buffer *leaf = path->nodes[0];
+
+		if (found_key.objectid != src_dev->devid)
+			break;
+
+		if (found_key.type != BTRFS_DEV_EXTENT_KEY)
+			break;
+
+		if (found_key.offset < key.offset)
+			break;
+
+		dev_extent = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_dev_extent);
+
+		chunk_offset = btrfs_dev_extent_chunk_offset(leaf, dev_extent);
+
+		cache = btrfs_lookup_block_group(fs_info, chunk_offset);
+		if (!cache)
+			continue;
+
+		set_bit(BLOCK_GROUP_FLAG_TO_COPY, &cache->runtime_flags);
+		btrfs_put_block_group(cache);
+	}
+	if (iter_ret < 0)
+		ret = iter_ret;
+
+	btrfs_free_path(path);
+unlock:
+	mutex_unlock(&fs_info->chunk_mutex);
+
+	return ret;
+}
+
+bool btrfs_finish_block_group_to_copy(struct btrfs_device *srcdev,
+				      struct btrfs_block_group *cache,
+				      u64 physical)
+{
+	struct btrfs_fs_info *fs_info = cache->fs_info;
+	struct extent_map *em;
+	struct map_lookup *map;
+	u64 chunk_offset = cache->start;
+	int num_extents, cur_extent;
+	int i;
+
+	/* Do not use "to_copy" on non zoned filesystem for now */
+	if (!btrfs_is_zoned(fs_info))
+		return true;
+
+	spin_lock(&cache->lock);
+	if (test_bit(BLOCK_GROUP_FLAG_REMOVED, &cache->runtime_flags)) {
+		spin_unlock(&cache->lock);
+		return true;
+	}
+	spin_unlock(&cache->lock);
+
+	em = btrfs_get_chunk_map(fs_info, chunk_offset, 1);
+	ASSERT(!IS_ERR(em));
+	map = em->map_lookup;
+
+	num_extents = 0;
+	cur_extent = 0;
+	for (i = 0; i < map->num_stripes; i++) {
+		/* We have more device extent to copy */
+		if (srcdev != map->stripes[i].dev)
+			continue;
+
+		num_extents++;
+		if (physical == map->stripes[i].physical)
+			cur_extent = i;
+	}
+
+	free_extent_map(em);
+
+	if (num_extents > 1 && cur_extent < num_extents - 1) {
+		/*
+		 * Has more stripes on this device. Keep this block group
+		 * readonly until we finish all the stripes.
+		 */
+		return false;
+	}
+
+	/* Last stripe on this device */
+	clear_bit(BLOCK_GROUP_FLAG_TO_COPY, &cache->runtime_flags);
+
+	return true;
+}
+
 static int btrfs_dev_replace_start(struct btrfs_fs_info *fs_info,
 		const char *tgtdev_name, u64 srcdevid, const char *srcdev_name,
 		int read_src)
@@ -432,6 +653,10 @@ static int btrfs_dev_replace_start(struct btrfs_fs_info *fs_info,
 	if (ret)
 		return ret;
 
+	ret = mark_block_group_to_copy(fs_info, src_device);
+	if (ret)
+		return ret;
+
 	down_write(&dev_replace->rwsem);
 	switch (dev_replace->replace_state) {
 	case BTRFS_IOCTL_DEV_REPLACE_STATE_NEVER_STARTED:
@@ -472,13 +697,18 @@ static int btrfs_dev_replace_start(struct btrfs_fs_info *fs_info,
 	atomic64_set(&dev_replace->num_uncorrectable_read_errors, 0);
 	up_write(&dev_replace->rwsem);
 
-	ret = btrfs_sysfs_add_device_link(tgt_device->fs_devices, tgt_device);
+	ret = btrfs_sysfs_add_device(tgt_device);
 	if (ret)
 		btrfs_err(fs_info, "kobj add dev failed %d", ret);
 
 	btrfs_wait_ordered_roots(fs_info, U64_MAX, 0, (u64)-1);
 
-	/* Commit dev_replace state and reserve 1 item for it. */
+	/*
+	 * Commit dev_replace state and reserve 1 item for it.
+	 * This is crucial to ensure we won't miss copying extents for new block
+	 * groups that are allocated after we started the device replace, and
+	 * must be done after setting up the device replace state.
+	 */
 	trans = btrfs_start_transaction(root, 1);
 	if (IS_ERR(trans)) {
 		ret = PTR_ERR(trans);
@@ -559,10 +789,68 @@ static void btrfs_rm_dev_replace_unblocked(struct btrfs_fs_info *fs_info)
 	wake_up(&fs_info->dev_replace.replace_wait);
 }
 
+/*
+ * When finishing the device replace, before swapping the source device with the
+ * target device we must update the chunk allocation state in the target device,
+ * as it is empty because replace works by directly copying the chunks and not
+ * through the normal chunk allocation path.
+ */
+static int btrfs_set_target_alloc_state(struct btrfs_device *srcdev,
+					struct btrfs_device *tgtdev)
+{
+	struct extent_state *cached_state = NULL;
+	u64 start = 0;
+	u64 found_start;
+	u64 found_end;
+	int ret = 0;
+
+	lockdep_assert_held(&srcdev->fs_info->chunk_mutex);
+
+	while (!find_first_extent_bit(&srcdev->alloc_state, start,
+				      &found_start, &found_end,
+				      CHUNK_ALLOCATED, &cached_state)) {
+		ret = set_extent_bits(&tgtdev->alloc_state, found_start,
+				      found_end, CHUNK_ALLOCATED);
+		if (ret)
+			break;
+		start = found_end + 1;
+	}
+
+	free_extent_state(cached_state);
+	return ret;
+}
+
+static void btrfs_dev_replace_update_device_in_mapping_tree(
+						struct btrfs_fs_info *fs_info,
+						struct btrfs_device *srcdev,
+						struct btrfs_device *tgtdev)
+{
+	struct extent_map_tree *em_tree = &fs_info->mapping_tree;
+	struct extent_map *em;
+	struct map_lookup *map;
+	u64 start = 0;
+	int i;
+
+	write_lock(&em_tree->lock);
+	do {
+		em = lookup_extent_mapping(em_tree, start, (u64)-1);
+		if (!em)
+			break;
+		map = em->map_lookup;
+		for (i = 0; i < map->num_stripes; i++)
+			if (srcdev == map->stripes[i].dev)
+				map->stripes[i].dev = tgtdev;
+		start = em->start + em->len;
+		free_extent_map(em);
+	} while (start);
+	write_unlock(&em_tree->lock);
+}
+
 static int btrfs_dev_replace_finishing(struct btrfs_fs_info *fs_info,
 				       int scrub_ret)
 {
 	struct btrfs_dev_replace *dev_replace = &fs_info->dev_replace;
+	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
 	struct btrfs_device *tgt_device;
 	struct btrfs_device *src_device;
 	struct btrfs_root *root = fs_info->tree_root;
@@ -590,7 +878,7 @@ static int btrfs_dev_replace_finishing(struct btrfs_fs_info *fs_info,
 	 * flush all outstanding I/O and inode extent mappings before the
 	 * copy operation is declared as being finished
 	 */
-	ret = btrfs_start_delalloc_roots(fs_info, -1);
+	ret = btrfs_start_delalloc_roots(fs_info, LONG_MAX, false);
 	if (ret) {
 		mutex_unlock(&dev_replace->lock_finishing_cancel_unmount);
 		return ret;
@@ -612,12 +900,12 @@ static int btrfs_dev_replace_finishing(struct btrfs_fs_info *fs_info,
 		WARN_ON(ret);
 
 		/* Prevent write_all_supers() during the finishing procedure */
-		mutex_lock(&fs_info->fs_devices->device_list_mutex);
+		mutex_lock(&fs_devices->device_list_mutex);
 		/* Prevent new chunks being allocated on the source device */
 		mutex_lock(&fs_info->chunk_mutex);
 
 		if (!list_empty(&src_device->post_commit_list)) {
-			mutex_unlock(&fs_info->fs_devices->device_list_mutex);
+			mutex_unlock(&fs_devices->device_list_mutex);
 			mutex_unlock(&fs_info->chunk_mutex);
 		} else {
 			break;
@@ -633,8 +921,14 @@ static int btrfs_dev_replace_finishing(struct btrfs_fs_info *fs_info,
 	dev_replace->time_stopped = ktime_get_real_seconds();
 	dev_replace->item_needs_writeback = 1;
 
-	/* replace old device with new one in mapping tree */
+	/*
+	 * Update allocation state in the new device and replace the old device
+	 * with the new one in the mapping tree.
+	 */
 	if (!scrub_ret) {
+		scrub_ret = btrfs_set_target_alloc_state(src_device, tgt_device);
+		if (scrub_ret)
+			goto error;
 		btrfs_dev_replace_update_device_in_mapping_tree(fs_info,
 								src_device,
 								tgt_device);
@@ -645,9 +939,10 @@ static int btrfs_dev_replace_finishing(struct btrfs_fs_info *fs_info,
 				 btrfs_dev_name(src_device),
 				 src_device->devid,
 				 rcu_str_deref(tgt_device->name), scrub_ret);
+error:
 		up_write(&dev_replace->rwsem);
 		mutex_unlock(&fs_info->chunk_mutex);
-		mutex_unlock(&fs_info->fs_devices->device_list_mutex);
+		mutex_unlock(&fs_devices->device_list_mutex);
 		btrfs_rm_dev_replace_blocked(fs_info);
 		if (tgt_device)
 			btrfs_destroy_dev_replace_tgtdev(tgt_device);
@@ -676,8 +971,8 @@ static int btrfs_dev_replace_finishing(struct btrfs_fs_info *fs_info,
 
 	btrfs_assign_next_active_device(src_device, tgt_device);
 
-	list_add(&tgt_device->dev_alloc_list, &fs_info->fs_devices->alloc_list);
-	fs_info->fs_devices->rw_devices++;
+	list_add(&tgt_device->dev_alloc_list, &fs_devices->alloc_list);
+	fs_devices->rw_devices++;
 
 	up_write(&dev_replace->rwsem);
 	btrfs_rm_dev_replace_blocked(fs_info);
@@ -700,12 +995,14 @@ static int btrfs_dev_replace_finishing(struct btrfs_fs_info *fs_info,
 	 * belong to this filesystem.
 	 */
 	mutex_unlock(&fs_info->chunk_mutex);
-	mutex_unlock(&fs_info->fs_devices->device_list_mutex);
+	mutex_unlock(&fs_devices->device_list_mutex);
 
 	/* replace the sysfs entry */
-	btrfs_sysfs_rm_device_link(fs_info->fs_devices, src_device);
+	btrfs_sysfs_remove_device(src_device);
 	btrfs_sysfs_update_devid(tgt_device);
-	btrfs_rm_dev_replace_free_srcdev(src_device);
+	if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &src_device->dev_state))
+		btrfs_scratch_superblocks(fs_info, src_device->bdev,
+					  src_device->name->str);
 
 	/* write back the superblocks */
 	trans = btrfs_start_transaction(root, 0);
@@ -714,33 +1011,9 @@ static int btrfs_dev_replace_finishing(struct btrfs_fs_info *fs_info,
 
 	mutex_unlock(&dev_replace->lock_finishing_cancel_unmount);
 
-	return 0;
-}
-
-static void btrfs_dev_replace_update_device_in_mapping_tree(
-						struct btrfs_fs_info *fs_info,
-						struct btrfs_device *srcdev,
-						struct btrfs_device *tgtdev)
-{
-	struct extent_map_tree *em_tree = &fs_info->mapping_tree;
-	struct extent_map *em;
-	struct map_lookup *map;
-	u64 start = 0;
-	int i;
+	btrfs_rm_dev_replace_free_srcdev(src_device);
 
-	write_lock(&em_tree->lock);
-	do {
-		em = lookup_extent_mapping(em_tree, start, (u64)-1);
-		if (!em)
-			break;
-		map = em->map_lookup;
-		for (i = 0; i < map->num_stripes; i++)
-			if (srcdev == map->stripes[i].dev)
-				map->stripes[i].dev = tgtdev;
-		start = em->start + em->len;
-		free_extent_map(em);
-	} while (start);
-	write_unlock(&em_tree->lock);
+	return 0;
 }
 
 /*
@@ -851,8 +1124,7 @@ int btrfs_dev_replace_cancel(struct btrfs_fs_info *fs_info)
 		up_write(&dev_replace->rwsem);
 
 		/* Scrub for replace must not be running in suspended state */
-		ret = btrfs_scrub_cancel(fs_info);
-		ASSERT(ret != -ENOTCONN);
+		btrfs_scrub_cancel(fs_info);
 
 		trans = btrfs_start_transaction(root, 0);
 		if (IS_ERR(trans)) {
@@ -943,7 +1215,7 @@ int btrfs_resume_dev_replace_async(struct btrfs_fs_info *fs_info)
 	 * should never allow both to start and pause. We don't want to allow
 	 * dev-replace to start anyway.
 	 */
-	if (test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags)) {
+	if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_DEV_REPLACE)) {
 		down_write(&dev_replace->rwsem);
 		dev_replace->replace_state =
 					BTRFS_IOCTL_DEV_REPLACE_STATE_SUSPENDED;
@@ -980,7 +1252,7 @@ static int btrfs_dev_replace_kthread(void *data)
 	ret = btrfs_dev_replace_finishing(fs_info, ret);
 	WARN_ON(ret && ret != -ECANCELED);
 
-	clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
+	btrfs_exclop_finish(fs_info);
 	return 0;
 }
 
@@ -1011,11 +1283,6 @@ int __pure btrfs_dev_replace_is_ongoing(struct btrfs_dev_replace *dev_replace)
 	return 1;
 }
 
-void btrfs_bio_counter_inc_noblocked(struct btrfs_fs_info *fs_info)
-{
-	percpu_counter_inc(&fs_info->dev_replace.bio_counter);
-}
-
 void btrfs_bio_counter_sub(struct btrfs_fs_info *fs_info, s64 amount)
 {
 	percpu_counter_sub(&fs_info->dev_replace.bio_counter, amount);
diff --git a/fs/btrfs/dev-replace.h b/fs/btrfs/dev-replace.h
index 60b70dacc299..6084b313056a 100644
--- a/fs/btrfs/dev-replace.h
+++ b/fs/btrfs/dev-replace.h
@@ -7,6 +7,10 @@
 #define BTRFS_DEV_REPLACE_H
 
 struct btrfs_ioctl_dev_replace_args;
+struct btrfs_fs_info;
+struct btrfs_trans_handle;
+struct btrfs_dev_replace;
+struct btrfs_block_group;
 
 int btrfs_init_dev_replace(struct btrfs_fs_info *fs_info);
 int btrfs_run_dev_replace(struct btrfs_trans_handle *trans);
@@ -18,5 +22,8 @@ int btrfs_dev_replace_cancel(struct btrfs_fs_info *fs_info);
 void btrfs_dev_replace_suspend_for_unmount(struct btrfs_fs_info *fs_info);
 int btrfs_resume_dev_replace_async(struct btrfs_fs_info *fs_info);
 int __pure btrfs_dev_replace_is_ongoing(struct btrfs_dev_replace *dev_replace);
+bool btrfs_finish_block_group_to_copy(struct btrfs_device *srcdev,
+				      struct btrfs_block_group *cache,
+				      u64 physical);
 
 #endif
diff --git a/fs/btrfs/dir-item.c b/fs/btrfs/dir-item.c
index 863367c2c620..72fb2c518a2b 100644
--- a/fs/btrfs/dir-item.c
+++ b/fs/btrfs/dir-item.c
@@ -27,7 +27,6 @@ static struct btrfs_dir_item *insert_with_overflow(struct btrfs_trans_handle
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	int ret;
 	char *ptr;
-	struct btrfs_item *item;
 	struct extent_buffer *leaf;
 
 	ret = btrfs_insert_empty_item(trans, root, path, cpu_key, data_size);
@@ -41,10 +40,9 @@ static struct btrfs_dir_item *insert_with_overflow(struct btrfs_trans_handle
 		return ERR_PTR(ret);
 	WARN_ON(ret > 0);
 	leaf = path->nodes[0];
-	item = btrfs_item_nr(path->slots[0]);
 	ptr = btrfs_item_ptr(leaf, path->slots[0], char);
-	BUG_ON(data_size > btrfs_item_size(leaf, item));
-	ptr += btrfs_item_size(leaf, item) - data_size;
+	ASSERT(data_size <= btrfs_item_size(leaf, path->slots[0]));
+	ptr += btrfs_item_size(leaf, path->slots[0]) - data_size;
 	return (struct btrfs_dir_item *)ptr;
 }
 
@@ -127,7 +125,6 @@ int btrfs_insert_dir_item(struct btrfs_trans_handle *trans, const char *name,
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
-	path->leave_spinning = 1;
 
 	btrfs_cpu_key_to_disk(&disk_key, location);
 
@@ -171,10 +168,40 @@ out_free:
 	return 0;
 }
 
+static struct btrfs_dir_item *btrfs_lookup_match_dir(
+			struct btrfs_trans_handle *trans,
+			struct btrfs_root *root, struct btrfs_path *path,
+			struct btrfs_key *key, const char *name,
+			int name_len, int mod)
+{
+	const int ins_len = (mod < 0 ? -1 : 0);
+	const int cow = (mod != 0);
+	int ret;
+
+	ret = btrfs_search_slot(trans, root, key, path, ins_len, cow);
+	if (ret < 0)
+		return ERR_PTR(ret);
+	if (ret > 0)
+		return ERR_PTR(-ENOENT);
+
+	return btrfs_match_dir_item_name(root->fs_info, path, name, name_len);
+}
+
 /*
- * lookup a directory item based on name.  'dir' is the objectid
- * we're searching in, and 'mod' tells us if you plan on deleting the
- * item (use mod < 0) or changing the options (use mod > 0)
+ * Lookup for a directory item by name.
+ *
+ * @trans:	The transaction handle to use. Can be NULL if @mod is 0.
+ * @root:	The root of the target tree.
+ * @path:	Path to use for the search.
+ * @dir:	The inode number (objectid) of the directory.
+ * @name:	The name associated to the directory entry we are looking for.
+ * @name_len:	The length of the name.
+ * @mod:	Used to indicate if the tree search is meant for a read only
+ *		lookup, for a modification lookup or for a deletion lookup, so
+ *		its value should be 0, 1 or -1, respectively.
+ *
+ * Returns: NULL if the dir item does not exists, an error pointer if an error
+ * happened, or a pointer to a dir item if a dir item exists for the given name.
  */
 struct btrfs_dir_item *btrfs_lookup_dir_item(struct btrfs_trans_handle *trans,
 					     struct btrfs_root *root,
@@ -182,23 +209,18 @@ struct btrfs_dir_item *btrfs_lookup_dir_item(struct btrfs_trans_handle *trans,
 					     const char *name, int name_len,
 					     int mod)
 {
-	int ret;
 	struct btrfs_key key;
-	int ins_len = mod < 0 ? -1 : 0;
-	int cow = mod != 0;
+	struct btrfs_dir_item *di;
 
 	key.objectid = dir;
 	key.type = BTRFS_DIR_ITEM_KEY;
-
 	key.offset = btrfs_name_hash(name, name_len);
 
-	ret = btrfs_search_slot(trans, root, &key, path, ins_len, cow);
-	if (ret < 0)
-		return ERR_PTR(ret);
-	if (ret > 0)
+	di = btrfs_lookup_match_dir(trans, root, path, &key, name, name_len, mod);
+	if (IS_ERR(di) && PTR_ERR(di) == -ENOENT)
 		return NULL;
 
-	return btrfs_match_dir_item_name(root->fs_info, path, name, name_len);
+	return di;
 }
 
 int btrfs_check_dir_item_collision(struct btrfs_root *root, u64 dir,
@@ -212,7 +234,6 @@ int btrfs_check_dir_item_collision(struct btrfs_root *root, u64 dir,
 	int slot;
 	struct btrfs_path *path;
 
-
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
@@ -221,20 +242,20 @@ int btrfs_check_dir_item_collision(struct btrfs_root *root, u64 dir,
 	key.type = BTRFS_DIR_ITEM_KEY;
 	key.offset = btrfs_name_hash(name, name_len);
 
-	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
-
-	/* return back any errors */
-	if (ret < 0)
-		goto out;
+	di = btrfs_lookup_match_dir(NULL, root, path, &key, name, name_len, 0);
+	if (IS_ERR(di)) {
+		ret = PTR_ERR(di);
+		/* Nothing found, we're safe */
+		if (ret == -ENOENT) {
+			ret = 0;
+			goto out;
+		}
 
-	/* nothing found, we're safe */
-	if (ret > 0) {
-		ret = 0;
-		goto out;
+		if (ret < 0)
+			goto out;
 	}
 
 	/* we found an item, look for our name in the item */
-	di = btrfs_match_dir_item_name(root->fs_info, path, name, name_len);
 	if (di) {
 		/* our exact name was found */
 		ret = -EEXIST;
@@ -248,7 +269,7 @@ int btrfs_check_dir_item_collision(struct btrfs_root *root, u64 dir,
 	data_size = sizeof(*di) + name_len;
 	leaf = path->nodes[0];
 	slot = path->slots[0];
-	if (data_size + btrfs_item_size_nr(leaf, slot) +
+	if (data_size + btrfs_item_size(leaf, slot) +
 	    sizeof(struct btrfs_item) > BTRFS_LEAF_DATA_SIZE(root->fs_info)) {
 		ret = -EOVERFLOW;
 	} else {
@@ -261,35 +282,42 @@ out:
 }
 
 /*
- * lookup a directory item based on index.  'dir' is the objectid
- * we're searching in, and 'mod' tells us if you plan on deleting the
- * item (use mod < 0) or changing the options (use mod > 0)
+ * Lookup for a directory index item by name and index number.
+ *
+ * @trans:	The transaction handle to use. Can be NULL if @mod is 0.
+ * @root:	The root of the target tree.
+ * @path:	Path to use for the search.
+ * @dir:	The inode number (objectid) of the directory.
+ * @index:	The index number.
+ * @name:	The name associated to the directory entry we are looking for.
+ * @name_len:	The length of the name.
+ * @mod:	Used to indicate if the tree search is meant for a read only
+ *		lookup, for a modification lookup or for a deletion lookup, so
+ *		its value should be 0, 1 or -1, respectively.
  *
- * The name is used to make sure the index really points to the name you were
- * looking for.
+ * Returns: NULL if the dir index item does not exists, an error pointer if an
+ * error happened, or a pointer to a dir item if the dir index item exists and
+ * matches the criteria (name and index number).
  */
 struct btrfs_dir_item *
 btrfs_lookup_dir_index_item(struct btrfs_trans_handle *trans,
 			    struct btrfs_root *root,
 			    struct btrfs_path *path, u64 dir,
-			    u64 objectid, const char *name, int name_len,
+			    u64 index, const char *name, int name_len,
 			    int mod)
 {
-	int ret;
+	struct btrfs_dir_item *di;
 	struct btrfs_key key;
-	int ins_len = mod < 0 ? -1 : 0;
-	int cow = mod != 0;
 
 	key.objectid = dir;
 	key.type = BTRFS_DIR_INDEX_KEY;
-	key.offset = objectid;
+	key.offset = index;
 
-	ret = btrfs_search_slot(trans, root, &key, path, ins_len, cow);
-	if (ret < 0)
-		return ERR_PTR(ret);
-	if (ret > 0)
-		return ERR_PTR(-ENOENT);
-	return btrfs_match_dir_item_name(root->fs_info, path, name, name_len);
+	di = btrfs_lookup_match_dir(trans, root, path, &key, name, name_len, mod);
+	if (di == ERR_PTR(-ENOENT))
+		return NULL;
+
+	return di;
 }
 
 struct btrfs_dir_item *
@@ -297,36 +325,15 @@ btrfs_search_dir_index_item(struct btrfs_root *root,
 			    struct btrfs_path *path, u64 dirid,
 			    const char *name, int name_len)
 {
-	struct extent_buffer *leaf;
 	struct btrfs_dir_item *di;
 	struct btrfs_key key;
-	u32 nritems;
 	int ret;
 
 	key.objectid = dirid;
 	key.type = BTRFS_DIR_INDEX_KEY;
 	key.offset = 0;
 
-	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
-	if (ret < 0)
-		return ERR_PTR(ret);
-
-	leaf = path->nodes[0];
-	nritems = btrfs_header_nritems(leaf);
-
-	while (1) {
-		if (path->slots[0] >= nritems) {
-			ret = btrfs_next_leaf(root, path);
-			if (ret < 0)
-				return ERR_PTR(ret);
-			if (ret > 0)
-				break;
-			leaf = path->nodes[0];
-			nritems = btrfs_header_nritems(leaf);
-			continue;
-		}
-
-		btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
+	btrfs_for_each_slot(root, &key, &key, path, ret) {
 		if (key.objectid != dirid || key.type != BTRFS_DIR_INDEX_KEY)
 			break;
 
@@ -334,10 +341,12 @@ btrfs_search_dir_index_item(struct btrfs_root *root,
 					       name, name_len);
 		if (di)
 			return di;
-
-		path->slots[0]++;
 	}
-	return NULL;
+	/* Adjust return code if the key was not found in the next leaf. */
+	if (ret > 0)
+		ret = 0;
+
+	return ERR_PTR(ret);
 }
 
 struct btrfs_dir_item *btrfs_lookup_xattr(struct btrfs_trans_handle *trans,
@@ -346,21 +355,18 @@ struct btrfs_dir_item *btrfs_lookup_xattr(struct btrfs_trans_handle *trans,
 					  const char *name, u16 name_len,
 					  int mod)
 {
-	int ret;
 	struct btrfs_key key;
-	int ins_len = mod < 0 ? -1 : 0;
-	int cow = mod != 0;
+	struct btrfs_dir_item *di;
 
 	key.objectid = dir;
 	key.type = BTRFS_XATTR_ITEM_KEY;
 	key.offset = btrfs_name_hash(name, name_len);
-	ret = btrfs_search_slot(trans, root, &key, path, ins_len, cow);
-	if (ret < 0)
-		return ERR_PTR(ret);
-	if (ret > 0)
+
+	di = btrfs_lookup_match_dir(trans, root, path, &key, name, name_len, mod);
+	if (IS_ERR(di) && PTR_ERR(di) == -ENOENT)
 		return NULL;
 
-	return btrfs_match_dir_item_name(root->fs_info, path, name, name_len);
+	return di;
 }
 
 /*
@@ -382,7 +388,7 @@ struct btrfs_dir_item *btrfs_match_dir_item_name(struct btrfs_fs_info *fs_info,
 	leaf = path->nodes[0];
 	dir_item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_dir_item);
 
-	total_len = btrfs_item_size_nr(leaf, path->slots[0]);
+	total_len = btrfs_item_size(leaf, path->slots[0]);
 	while (cur < total_len) {
 		this_len = sizeof(*dir_item) +
 			btrfs_dir_name_len(leaf, dir_item) +
@@ -418,7 +424,7 @@ int btrfs_delete_one_dir_name(struct btrfs_trans_handle *trans,
 	leaf = path->nodes[0];
 	sub_item_len = sizeof(*di) + btrfs_dir_name_len(leaf, di) +
 		btrfs_dir_data_len(leaf, di);
-	item_len = btrfs_item_size_nr(leaf, path->slots[0]);
+	item_len = btrfs_item_size(leaf, path->slots[0]);
 	if (sub_item_len == item_len) {
 		ret = btrfs_del_item(trans, root, path);
 	} else {
diff --git a/fs/btrfs/discard.c b/fs/btrfs/discard.c
index 5615320fa659..e1b7bd927d69 100644
--- a/fs/btrfs/discard.c
+++ b/fs/btrfs/discard.c
@@ -185,10 +185,12 @@ static struct btrfs_block_group *find_next_block_group(
 }
 
 /**
- * peek_discard_list - wrap find_next_block_group()
- * @discard_ctl: discard control
+ * Wrap find_next_block_group()
+ *
+ * @discard_ctl:   discard control
  * @discard_state: the discard_state of the block_group after state management
  * @discard_index: the discard_index of the block_group after state management
+ * @now:           time when discard was invoked, in ns
  *
  * This wraps find_next_block_group() and sets the block_group to be in use.
  * discard_state's control flow is managed here.  Variables related to
@@ -199,16 +201,15 @@ static struct btrfs_block_group *find_next_block_group(
 static struct btrfs_block_group *peek_discard_list(
 					struct btrfs_discard_ctl *discard_ctl,
 					enum btrfs_discard_state *discard_state,
-					int *discard_index)
+					int *discard_index, u64 now)
 {
 	struct btrfs_block_group *block_group;
-	const u64 now = ktime_get_ns();
 
 	spin_lock(&discard_ctl->lock);
 again:
 	block_group = find_next_block_group(discard_ctl, now);
 
-	if (block_group && now > block_group->discard_eligible_time) {
+	if (block_group && now >= block_group->discard_eligible_time) {
 		if (block_group->discard_index == BTRFS_DISCARD_INDEX_UNUSED &&
 		    block_group->used != 0) {
 			if (btrfs_is_block_group_data_only(block_group))
@@ -222,12 +223,11 @@ again:
 			block_group->discard_state = BTRFS_DISCARD_EXTENTS;
 		}
 		discard_ctl->block_group = block_group;
+	}
+	if (block_group) {
 		*discard_state = block_group->discard_state;
 		*discard_index = block_group->discard_index;
-	} else {
-		block_group = NULL;
 	}
-
 	spin_unlock(&discard_ctl->lock);
 
 	return block_group;
@@ -330,32 +330,19 @@ void btrfs_discard_queue_work(struct btrfs_discard_ctl *discard_ctl,
 		btrfs_discard_schedule_work(discard_ctl, false);
 }
 
-/**
- * btrfs_discard_schedule_work - responsible for scheduling the discard work
- * @discard_ctl: discard control
- * @override: override the current timer
- *
- * Discards are issued by a delayed workqueue item.  @override is used to
- * update the current delay as the baseline delay interval is reevaluated on
- * transaction commit.  This is also maxed with any other rate limit.
- */
-void btrfs_discard_schedule_work(struct btrfs_discard_ctl *discard_ctl,
-				 bool override)
+static void __btrfs_discard_schedule_work(struct btrfs_discard_ctl *discard_ctl,
+					  u64 now, bool override)
 {
 	struct btrfs_block_group *block_group;
-	const u64 now = ktime_get_ns();
-
-	spin_lock(&discard_ctl->lock);
 
 	if (!btrfs_run_discard_work(discard_ctl))
-		goto out;
-
+		return;
 	if (!override && delayed_work_pending(&discard_ctl->work))
-		goto out;
+		return;
 
 	block_group = find_next_block_group(discard_ctl, now);
 	if (block_group) {
-		unsigned long delay = discard_ctl->delay;
+		u64 delay = discard_ctl->delay_ms * NSEC_PER_MSEC;
 		u32 kbps_limit = READ_ONCE(discard_ctl->kbps_limit);
 
 		/*
@@ -366,9 +353,9 @@ void btrfs_discard_schedule_work(struct btrfs_discard_ctl *discard_ctl,
 		if (kbps_limit && discard_ctl->prev_discard) {
 			u64 bps_limit = ((u64)kbps_limit) * SZ_1K;
 			u64 bps_delay = div64_u64(discard_ctl->prev_discard *
-						  MSEC_PER_SEC, bps_limit);
+						  NSEC_PER_SEC, bps_limit);
 
-			delay = max(delay, msecs_to_jiffies(bps_delay));
+			delay = max(delay, bps_delay);
 		}
 
 		/*
@@ -378,13 +365,39 @@ void btrfs_discard_schedule_work(struct btrfs_discard_ctl *discard_ctl,
 		if (now < block_group->discard_eligible_time) {
 			u64 bg_timeout = block_group->discard_eligible_time - now;
 
-			delay = max(delay, nsecs_to_jiffies(bg_timeout));
+			delay = max(delay, bg_timeout);
+		}
+
+		if (override && discard_ctl->prev_discard) {
+			u64 elapsed = now - discard_ctl->prev_discard_time;
+
+			if (delay > elapsed)
+				delay -= elapsed;
+			else
+				delay = 0;
 		}
 
 		mod_delayed_work(discard_ctl->discard_workers,
-				 &discard_ctl->work, delay);
+				 &discard_ctl->work, nsecs_to_jiffies(delay));
 	}
-out:
+}
+
+/*
+ * btrfs_discard_schedule_work - responsible for scheduling the discard work
+ * @discard_ctl:  discard control
+ * @override:     override the current timer
+ *
+ * Discards are issued by a delayed workqueue item.  @override is used to
+ * update the current delay as the baseline delay interval is reevaluated on
+ * transaction commit.  This is also maxed with any other rate limit.
+ */
+void btrfs_discard_schedule_work(struct btrfs_discard_ctl *discard_ctl,
+				 bool override)
+{
+	const u64 now = ktime_get_ns();
+
+	spin_lock(&discard_ctl->lock);
+	__btrfs_discard_schedule_work(discard_ctl, now, override);
 	spin_unlock(&discard_ctl->lock);
 }
 
@@ -429,13 +442,18 @@ static void btrfs_discard_workfn(struct work_struct *work)
 	int discard_index = 0;
 	u64 trimmed = 0;
 	u64 minlen = 0;
+	u64 now = ktime_get_ns();
 
 	discard_ctl = container_of(work, struct btrfs_discard_ctl, work.work);
 
 	block_group = peek_discard_list(discard_ctl, &discard_state,
-					&discard_index);
+					&discard_index, now);
 	if (!block_group || !btrfs_run_discard_work(discard_ctl))
 		return;
+	if (now < block_group->discard_eligible_time) {
+		btrfs_discard_schedule_work(discard_ctl, false);
+		return;
+	}
 
 	/* Perform discarding */
 	minlen = discard_minlen[discard_index];
@@ -465,8 +483,6 @@ static void btrfs_discard_workfn(struct work_struct *work)
 		discard_ctl->discard_extent_bytes += trimmed;
 	}
 
-	discard_ctl->prev_discard = trimmed;
-
 	/* Determine next steps for a block_group */
 	if (block_group->discard_cursor >= btrfs_block_group_end(block_group)) {
 		if (discard_state == BTRFS_DISCARD_BITMAPS) {
@@ -482,11 +498,13 @@ static void btrfs_discard_workfn(struct work_struct *work)
 		}
 	}
 
+	now = ktime_get_ns();
 	spin_lock(&discard_ctl->lock);
+	discard_ctl->prev_discard = trimmed;
+	discard_ctl->prev_discard_time = now;
 	discard_ctl->block_group = NULL;
+	__btrfs_discard_schedule_work(discard_ctl, now, false);
 	spin_unlock(&discard_ctl->lock);
-
-	btrfs_discard_schedule_work(discard_ctl, false);
 }
 
 /**
@@ -519,7 +537,6 @@ void btrfs_discard_calc_delay(struct btrfs_discard_ctl *discard_ctl)
 	s64 discardable_bytes;
 	u32 iops_limit;
 	unsigned long delay;
-	unsigned long lower_limit = BTRFS_DISCARD_MIN_DELAY_MSEC;
 
 	discardable_extents = atomic_read(&discard_ctl->discardable_extents);
 	if (!discardable_extents)
@@ -550,12 +567,13 @@ void btrfs_discard_calc_delay(struct btrfs_discard_ctl *discard_ctl)
 
 	iops_limit = READ_ONCE(discard_ctl->iops_limit);
 	if (iops_limit)
-		lower_limit = max_t(unsigned long, lower_limit,
-				    MSEC_PER_SEC / iops_limit);
+		delay = MSEC_PER_SEC / iops_limit;
+	else
+		delay = BTRFS_DISCARD_TARGET_MSEC / discardable_extents;
 
-	delay = BTRFS_DISCARD_TARGET_MSEC / discardable_extents;
-	delay = clamp(delay, lower_limit, BTRFS_DISCARD_MAX_DELAY_MSEC);
-	discard_ctl->delay = msecs_to_jiffies(delay);
+	delay = clamp(delay, BTRFS_DISCARD_MIN_DELAY_MSEC,
+		      BTRFS_DISCARD_MAX_DELAY_MSEC);
+	discard_ctl->delay_ms = delay;
 
 	spin_unlock(&discard_ctl->lock);
 }
@@ -563,15 +581,14 @@ void btrfs_discard_calc_delay(struct btrfs_discard_ctl *discard_ctl)
 /**
  * btrfs_discard_update_discardable - propagate discard counters
  * @block_group: block_group of interest
- * @ctl: free_space_ctl of @block_group
  *
  * This propagates deltas of counters up to the discard_ctl.  It maintains a
  * current counter and a previous counter passing the delta up to the global
  * stat.  Then the current counter value becomes the previous counter value.
  */
-void btrfs_discard_update_discardable(struct btrfs_block_group *block_group,
-				      struct btrfs_free_space_ctl *ctl)
+void btrfs_discard_update_discardable(struct btrfs_block_group *block_group)
 {
+	struct btrfs_free_space_ctl *ctl;
 	struct btrfs_discard_ctl *discard_ctl;
 	s32 extents_delta;
 	s64 bytes_delta;
@@ -581,8 +598,10 @@ void btrfs_discard_update_discardable(struct btrfs_block_group *block_group,
 	    !btrfs_is_block_group_data_only(block_group))
 		return;
 
+	ctl = block_group->free_space_ctl;
 	discard_ctl = &block_group->fs_info->discard_ctl;
 
+	lockdep_assert_held(&ctl->tree_lock);
 	extents_delta = ctl->discardable_extents[BTRFS_STAT_CURR] -
 			ctl->discardable_extents[BTRFS_STAT_PREV];
 	if (extents_delta) {
@@ -605,7 +624,7 @@ void btrfs_discard_update_discardable(struct btrfs_block_group *block_group,
  * @fs_info: fs_info of interest
  *
  * The unused_bgs list needs to be punted to the discard lists because the
- * order of operations is changed.  In the normal sychronous discard path, the
+ * order of operations is changed.  In the normal synchronous discard path, the
  * block groups are trimmed via a single large trim in transaction commit.  This
  * is ultimately what we are trying to avoid with asynchronous discard.  Thus,
  * it must be done before going down the unused_bgs path.
@@ -619,6 +638,7 @@ void btrfs_discard_punt_unused_bgs_list(struct btrfs_fs_info *fs_info)
 	list_for_each_entry_safe(block_group, next, &fs_info->unused_bgs,
 				 bg_list) {
 		list_del_init(&block_group->bg_list);
+		btrfs_put_block_group(block_group);
 		btrfs_discard_queue_work(&fs_info->discard_ctl, block_group);
 	}
 	spin_unlock(&fs_info->unused_bgs_lock);
@@ -683,10 +703,11 @@ void btrfs_discard_init(struct btrfs_fs_info *fs_info)
 		INIT_LIST_HEAD(&discard_ctl->discard_list[i]);
 
 	discard_ctl->prev_discard = 0;
+	discard_ctl->prev_discard_time = 0;
 	atomic_set(&discard_ctl->discardable_extents, 0);
 	atomic64_set(&discard_ctl->discardable_bytes, 0);
 	discard_ctl->max_discard_size = BTRFS_ASYNC_DISCARD_DEFAULT_MAX_SIZE;
-	discard_ctl->delay = BTRFS_DISCARD_MAX_DELAY_MSEC;
+	discard_ctl->delay_ms = BTRFS_DISCARD_MAX_DELAY_MSEC;
 	discard_ctl->iops_limit = BTRFS_DISCARD_MAX_IOPS;
 	discard_ctl->kbps_limit = 0;
 	discard_ctl->discard_extent_bytes = 0;
diff --git a/fs/btrfs/discard.h b/fs/btrfs/discard.h
index 21a15776dac4..57b9202f427f 100644
--- a/fs/btrfs/discard.h
+++ b/fs/btrfs/discard.h
@@ -1,4 +1,4 @@
-// SPDX-License-Identifier: GPL-2.0
+/* SPDX-License-Identifier: GPL-2.0 */
 
 #ifndef BTRFS_DISCARD_H
 #define BTRFS_DISCARD_H
@@ -28,8 +28,7 @@ bool btrfs_run_discard_work(struct btrfs_discard_ctl *discard_ctl);
 
 /* Update operations */
 void btrfs_discard_calc_delay(struct btrfs_discard_ctl *discard_ctl);
-void btrfs_discard_update_discardable(struct btrfs_block_group *block_group,
-				      struct btrfs_free_space_ctl *ctl);
+void btrfs_discard_update_discardable(struct btrfs_block_group *block_group);
 
 /* Setup/cleanup operations */
 void btrfs_discard_punt_unused_bgs_list(struct btrfs_fs_info *fs_info);
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index c6c9a6a8e6c8..d99bf7c64611 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -7,7 +7,6 @@
 #include <linux/blkdev.h>
 #include <linux/radix-tree.h>
 #include <linux/writeback.h>
-#include <linux/buffer_head.h>
 #include <linux/workqueue.h>
 #include <linux/kthread.h>
 #include <linux/slab.h>
@@ -30,7 +29,6 @@
 #include "tree-log.h"
 #include "free-space-cache.h"
 #include "free-space-tree.h"
-#include "inode-map.h"
 #include "check-integrity.h"
 #include "rcu-string.h"
 #include "dev-replace.h"
@@ -42,6 +40,9 @@
 #include "ref-verify.h"
 #include "block-group.h"
 #include "discard.h"
+#include "space-info.h"
+#include "zoned.h"
+#include "subpage.h"
 
 #define BTRFS_SUPER_FLAG_SUPP	(BTRFS_HEADER_FLAG_WRITTEN |\
 				 BTRFS_HEADER_FLAG_RELOC |\
@@ -50,8 +51,6 @@
 				 BTRFS_SUPER_FLAG_METADUMP |\
 				 BTRFS_SUPER_FLAG_METADUMP_V2)
 
-static const struct extent_io_ops btree_extent_io_ops;
-static void end_workqueue_fn(struct btrfs_work *work);
 static void btrfs_destroy_ordered_extents(struct btrfs_root *root);
 static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
 				      struct btrfs_fs_info *fs_info);
@@ -64,38 +63,10 @@ static int btrfs_destroy_pinned_extent(struct btrfs_fs_info *fs_info,
 static int btrfs_cleanup_transaction(struct btrfs_fs_info *fs_info);
 static void btrfs_error_commit_super(struct btrfs_fs_info *fs_info);
 
-/*
- * btrfs_end_io_wq structs are used to do processing in task context when an IO
- * is complete.  This is used during reads to verify checksums, and it is used
- * by writes to insert metadata for new file extents after IO is complete.
- */
-struct btrfs_end_io_wq {
-	struct bio *bio;
-	bio_end_io_t *end_io;
-	void *private;
-	struct btrfs_fs_info *info;
-	blk_status_t status;
-	enum btrfs_wq_endio_type metadata;
-	struct btrfs_work work;
-};
-
-static struct kmem_cache *btrfs_end_io_wq_cache;
-
-int __init btrfs_end_io_wq_init(void)
-{
-	btrfs_end_io_wq_cache = kmem_cache_create("btrfs_end_io_wq",
-					sizeof(struct btrfs_end_io_wq),
-					0,
-					SLAB_MEM_SPREAD,
-					NULL);
-	if (!btrfs_end_io_wq_cache)
-		return -ENOMEM;
-	return 0;
-}
-
-void __cold btrfs_end_io_wq_exit(void)
+static void btrfs_free_csum_hash(struct btrfs_fs_info *fs_info)
 {
-	kmem_cache_destroy(btrfs_end_io_wq_cache);
+	if (fs_info->csum_shash)
+		crypto_free_shash(fs_info->csum_shash);
 }
 
 /*
@@ -104,190 +75,41 @@ void __cold btrfs_end_io_wq_exit(void)
  * just before they are sent down the IO stack.
  */
 struct async_submit_bio {
-	void *private_data;
+	struct inode *inode;
 	struct bio *bio;
 	extent_submit_bio_start_t *submit_bio_start;
 	int mirror_num;
-	/*
-	 * bio_offset is optional, can be used if the pages in the bio
-	 * can't tell us where in the file the bio should go
-	 */
-	u64 bio_offset;
+
+	/* Optional parameter for submit_bio_start used by direct io */
+	u64 dio_file_offset;
 	struct btrfs_work work;
 	blk_status_t status;
 };
 
 /*
- * Lockdep class keys for extent_buffer->lock's in this root.  For a given
- * eb, the lockdep key is determined by the btrfs_root it belongs to and
- * the level the eb occupies in the tree.
- *
- * Different roots are used for different purposes and may nest inside each
- * other and they require separate keysets.  As lockdep keys should be
- * static, assign keysets according to the purpose of the root as indicated
- * by btrfs_root->root_key.objectid.  This ensures that all special purpose
- * roots have separate keysets.
- *
- * Lock-nesting across peer nodes is always done with the immediate parent
- * node locked thus preventing deadlock.  As lockdep doesn't know this, use
- * subclass to avoid triggering lockdep warning in such cases.
- *
- * The key is set by the readpage_end_io_hook after the buffer has passed
- * csum validation but before the pages are unlocked.  It is also set by
- * btrfs_init_new_buffer on freshly allocated blocks.
- *
- * We also add a check to make sure the highest level of the tree is the
- * same as our lockdep setup here.  If BTRFS_MAX_LEVEL changes, this code
- * needs update as well.
- */
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
-# if BTRFS_MAX_LEVEL != 8
-#  error
-# endif
-
-static struct btrfs_lockdep_keyset {
-	u64			id;		/* root objectid */
-	const char		*name_stem;	/* lock name stem */
-	char			names[BTRFS_MAX_LEVEL + 1][20];
-	struct lock_class_key	keys[BTRFS_MAX_LEVEL + 1];
-} btrfs_lockdep_keysets[] = {
-	{ .id = BTRFS_ROOT_TREE_OBJECTID,	.name_stem = "root"	},
-	{ .id = BTRFS_EXTENT_TREE_OBJECTID,	.name_stem = "extent"	},
-	{ .id = BTRFS_CHUNK_TREE_OBJECTID,	.name_stem = "chunk"	},
-	{ .id = BTRFS_DEV_TREE_OBJECTID,	.name_stem = "dev"	},
-	{ .id = BTRFS_FS_TREE_OBJECTID,		.name_stem = "fs"	},
-	{ .id = BTRFS_CSUM_TREE_OBJECTID,	.name_stem = "csum"	},
-	{ .id = BTRFS_QUOTA_TREE_OBJECTID,	.name_stem = "quota"	},
-	{ .id = BTRFS_TREE_LOG_OBJECTID,	.name_stem = "log"	},
-	{ .id = BTRFS_TREE_RELOC_OBJECTID,	.name_stem = "treloc"	},
-	{ .id = BTRFS_DATA_RELOC_TREE_OBJECTID,	.name_stem = "dreloc"	},
-	{ .id = BTRFS_UUID_TREE_OBJECTID,	.name_stem = "uuid"	},
-	{ .id = BTRFS_FREE_SPACE_TREE_OBJECTID,	.name_stem = "free-space" },
-	{ .id = 0,				.name_stem = "tree"	},
-};
-
-void __init btrfs_init_lockdep(void)
-{
-	int i, j;
-
-	/* initialize lockdep class names */
-	for (i = 0; i < ARRAY_SIZE(btrfs_lockdep_keysets); i++) {
-		struct btrfs_lockdep_keyset *ks = &btrfs_lockdep_keysets[i];
-
-		for (j = 0; j < ARRAY_SIZE(ks->names); j++)
-			snprintf(ks->names[j], sizeof(ks->names[j]),
-				 "btrfs-%s-%02d", ks->name_stem, j);
-	}
-}
-
-void btrfs_set_buffer_lockdep_class(u64 objectid, struct extent_buffer *eb,
-				    int level)
-{
-	struct btrfs_lockdep_keyset *ks;
-
-	BUG_ON(level >= ARRAY_SIZE(ks->keys));
-
-	/* find the matching keyset, id 0 is the default entry */
-	for (ks = btrfs_lockdep_keysets; ks->id; ks++)
-		if (ks->id == objectid)
-			break;
-
-	lockdep_set_class_and_name(&eb->lock,
-				   &ks->keys[level], ks->names[level]);
-}
-
-#endif
-
-/*
- * extents on the btree inode are pretty simple, there's one extent
- * that covers the entire device
- */
-struct extent_map *btree_get_extent(struct btrfs_inode *inode,
-				    struct page *page, size_t pg_offset,
-				    u64 start, u64 len)
-{
-	struct extent_map_tree *em_tree = &inode->extent_tree;
-	struct extent_map *em;
-	int ret;
-
-	read_lock(&em_tree->lock);
-	em = lookup_extent_mapping(em_tree, start, len);
-	if (em) {
-		read_unlock(&em_tree->lock);
-		goto out;
-	}
-	read_unlock(&em_tree->lock);
-
-	em = alloc_extent_map();
-	if (!em) {
-		em = ERR_PTR(-ENOMEM);
-		goto out;
-	}
-	em->start = 0;
-	em->len = (u64)-1;
-	em->block_len = (u64)-1;
-	em->block_start = 0;
-
-	write_lock(&em_tree->lock);
-	ret = add_extent_mapping(em_tree, em, 0);
-	if (ret == -EEXIST) {
-		free_extent_map(em);
-		em = lookup_extent_mapping(em_tree, start, len);
-		if (!em)
-			em = ERR_PTR(-EIO);
-	} else if (ret) {
-		free_extent_map(em);
-		em = ERR_PTR(ret);
-	}
-	write_unlock(&em_tree->lock);
-
-out:
-	return em;
-}
-
-/*
  * Compute the csum of a btree block and store the result to provided buffer.
- *
- * Returns error if the extent buffer cannot be mapped.
  */
-static int csum_tree_block(struct extent_buffer *buf, u8 *result)
+static void csum_tree_block(struct extent_buffer *buf, u8 *result)
 {
 	struct btrfs_fs_info *fs_info = buf->fs_info;
+	const int num_pages = num_extent_pages(buf);
+	const int first_page_part = min_t(u32, PAGE_SIZE, fs_info->nodesize);
 	SHASH_DESC_ON_STACK(shash, fs_info->csum_shash);
-	unsigned long len;
-	unsigned long cur_len;
-	unsigned long offset = BTRFS_CSUM_SIZE;
 	char *kaddr;
-	unsigned long map_start;
-	unsigned long map_len;
-	int err;
+	int i;
 
 	shash->tfm = fs_info->csum_shash;
 	crypto_shash_init(shash);
+	kaddr = page_address(buf->pages[0]) + offset_in_page(buf->start);
+	crypto_shash_update(shash, kaddr + BTRFS_CSUM_SIZE,
+			    first_page_part - BTRFS_CSUM_SIZE);
 
-	len = buf->len - offset;
-
-	while (len > 0) {
-		/*
-		 * Note: we don't need to check for the err == 1 case here, as
-		 * with the given combination of 'start = BTRFS_CSUM_SIZE (32)'
-		 * and 'min_len = 32' and the currently implemented mapping
-		 * algorithm we cannot cross a page boundary.
-		 */
-		err = map_private_extent_buffer(buf, offset, 32,
-					&kaddr, &map_start, &map_len);
-		if (WARN_ON(err))
-			return err;
-		cur_len = min(len, map_len - (offset - map_start));
-		crypto_shash_update(shash, kaddr + offset - map_start, cur_len);
-		len -= cur_len;
-		offset += cur_len;
+	for (i = 1; i < num_pages; i++) {
+		kaddr = page_address(buf->pages[i]);
+		crypto_shash_update(shash, kaddr, PAGE_SIZE);
 	}
 	memset(result, 0, BTRFS_CSUM_SIZE);
-
 	crypto_shash_final(shash, result);
-
-	return 0;
 }
 
 /*
@@ -302,7 +124,6 @@ static int verify_parent_transid(struct extent_io_tree *io_tree,
 {
 	struct extent_state *cached_state = NULL;
 	int ret;
-	bool need_lock = (current->journal_info == BTRFS_SEND_TRANS_STUB);
 
 	if (!parent_transid || btrfs_header_generation(eb) == parent_transid)
 		return 0;
@@ -310,39 +131,21 @@ static int verify_parent_transid(struct extent_io_tree *io_tree,
 	if (atomic)
 		return -EAGAIN;
 
-	if (need_lock) {
-		btrfs_tree_read_lock(eb);
-		btrfs_set_lock_blocking_read(eb);
-	}
-
-	lock_extent_bits(io_tree, eb->start, eb->start + eb->len - 1,
-			 &cached_state);
+	lock_extent(io_tree, eb->start, eb->start + eb->len - 1, &cached_state);
 	if (extent_buffer_uptodate(eb) &&
 	    btrfs_header_generation(eb) == parent_transid) {
 		ret = 0;
 		goto out;
 	}
 	btrfs_err_rl(eb->fs_info,
-		"parent transid verify failed on %llu wanted %llu found %llu",
-			eb->start,
+"parent transid verify failed on logical %llu mirror %u wanted %llu found %llu",
+			eb->start, eb->read_mirror,
 			parent_transid, btrfs_header_generation(eb));
 	ret = 1;
-
-	/*
-	 * Things reading via commit roots that don't have normal protection,
-	 * like send, can have a really old block in cache that may point at a
-	 * block that has been freed and re-allocated.  So don't clear uptodate
-	 * if we find an eb that is under IO (dirty/writeback) because we could
-	 * end up reading in the stale data and then writing it back out and
-	 * making everybody very sad.
-	 */
-	if (!extent_buffer_under_io(eb))
-		clear_extent_buffer_uptodate(eb);
+	clear_extent_buffer_uptodate(eb);
 out:
-	unlock_extent_cached(io_tree, eb->start, eb->start + eb->len - 1,
-			     &cached_state);
-	if (need_lock)
-		btrfs_tree_read_unlock_blocking(eb);
+	unlock_extent(io_tree, eb->start, eb->start + eb->len - 1,
+		      &cached_state);
 	return ret;
 }
 
@@ -363,27 +166,23 @@ static bool btrfs_supported_super_csum(u16 csum_type)
  * Return 0 if the superblock checksum type matches the checksum value of that
  * algorithm. Pass the raw disk superblock data.
  */
-static int btrfs_check_super_csum(struct btrfs_fs_info *fs_info,
-				  char *raw_disk_sb)
+int btrfs_check_super_csum(struct btrfs_fs_info *fs_info,
+			   const struct btrfs_super_block *disk_sb)
 {
-	struct btrfs_super_block *disk_sb =
-		(struct btrfs_super_block *)raw_disk_sb;
 	char result[BTRFS_CSUM_SIZE];
 	SHASH_DESC_ON_STACK(shash, fs_info->csum_shash);
 
 	shash->tfm = fs_info->csum_shash;
-	crypto_shash_init(shash);
 
 	/*
 	 * The super_block structure does not span the whole
 	 * BTRFS_SUPER_INFO_SIZE range, we expect that the unused space is
 	 * filled with zeros and is included in the checksum.
 	 */
-	crypto_shash_update(shash, raw_disk_sb + BTRFS_CSUM_SIZE,
-			    BTRFS_SUPER_INFO_SIZE - BTRFS_CSUM_SIZE);
-	crypto_shash_final(shash, result);
+	crypto_shash_digest(shash, (const u8 *)disk_sb + BTRFS_CSUM_SIZE,
+			    BTRFS_SUPER_INFO_SIZE - BTRFS_CSUM_SIZE, result);
 
-	if (memcmp(disk_sb->csum, result, btrfs_super_csum_size(disk_sb)))
+	if (memcmp(disk_sb->csum, result, fs_info->csum_size))
 		return 1;
 
 	return 0;
@@ -455,9 +254,9 @@ int btrfs_verify_level_key(struct extent_buffer *eb, int level,
  * @level:		expected level, mandatory check
  * @first_key:		expected key of first slot, skip check if NULL
  */
-static int btree_read_extent_buffer_pages(struct extent_buffer *eb,
-					  u64 parent_transid, int level,
-					  struct btrfs_key *first_key)
+int btrfs_read_extent_buffer(struct extent_buffer *eb,
+			     u64 parent_transid, int level,
+			     struct btrfs_key *first_key)
 {
 	struct btrfs_fs_info *fs_info = eb->fs_info;
 	struct extent_io_tree *io_tree;
@@ -506,164 +305,202 @@ static int btree_read_extent_buffer_pages(struct extent_buffer *eb,
 	return ret;
 }
 
-/*
- * checksum a dirty tree block before IO.  This has extra checks to make sure
- * we only fill in the checksum field in the first page of a multi-page block
- */
-
-static int csum_dirty_buffer(struct btrfs_fs_info *fs_info, struct page *page)
+static int csum_one_extent_buffer(struct extent_buffer *eb)
 {
-	u64 start = page_offset(page);
-	u64 found_start;
+	struct btrfs_fs_info *fs_info = eb->fs_info;
 	u8 result[BTRFS_CSUM_SIZE];
-	u16 csum_size = btrfs_super_csum_size(fs_info->super_copy);
-	struct extent_buffer *eb;
 	int ret;
 
-	eb = (struct extent_buffer *)page->private;
-	if (page != eb->pages[0])
-		return 0;
-
-	found_start = btrfs_header_bytenr(eb);
-	/*
-	 * Please do not consolidate these warnings into a single if.
-	 * It is useful to know what went wrong.
-	 */
-	if (WARN_ON(found_start != start))
-		return -EUCLEAN;
-	if (WARN_ON(!PageUptodate(page)))
-		return -EUCLEAN;
-
 	ASSERT(memcmp_extent_buffer(eb, fs_info->fs_devices->metadata_uuid,
-			btrfs_header_fsid(), BTRFS_FSID_SIZE) == 0);
-
-	if (csum_tree_block(eb, result))
-		return -EINVAL;
+				    offsetof(struct btrfs_header, fsid),
+				    BTRFS_FSID_SIZE) == 0);
+	csum_tree_block(eb, result);
 
 	if (btrfs_header_level(eb))
 		ret = btrfs_check_node(eb);
 	else
 		ret = btrfs_check_leaf_full(eb);
 
-	if (ret < 0) {
-		btrfs_print_tree(eb, 0);
+	if (ret < 0)
+		goto error;
+
+	/*
+	 * Also check the generation, the eb reached here must be newer than
+	 * last committed. Or something seriously wrong happened.
+	 */
+	if (unlikely(btrfs_header_generation(eb) <= fs_info->last_trans_committed)) {
+		ret = -EUCLEAN;
 		btrfs_err(fs_info,
-		"block=%llu write time tree block corruption detected",
-			  eb->start);
-		WARN_ON(IS_ENABLED(CONFIG_BTRFS_DEBUG));
-		return ret;
+			"block=%llu bad generation, have %llu expect > %llu",
+			  eb->start, btrfs_header_generation(eb),
+			  fs_info->last_trans_committed);
+		goto error;
 	}
-	write_extent_buffer(eb, result, 0, csum_size);
+	write_extent_buffer(eb, result, 0, fs_info->csum_size);
 
 	return 0;
+
+error:
+	btrfs_print_tree(eb, 0);
+	btrfs_err(fs_info, "block=%llu write time tree block corruption detected",
+		  eb->start);
+	WARN_ON(IS_ENABLED(CONFIG_BTRFS_DEBUG));
+	return ret;
 }
 
-static int check_tree_block_fsid(struct extent_buffer *eb)
+/* Checksum all dirty extent buffers in one bio_vec */
+static int csum_dirty_subpage_buffers(struct btrfs_fs_info *fs_info,
+				      struct bio_vec *bvec)
 {
-	struct btrfs_fs_info *fs_info = eb->fs_info;
-	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
-	u8 fsid[BTRFS_FSID_SIZE];
-	int ret = 1;
+	struct page *page = bvec->bv_page;
+	u64 bvec_start = page_offset(page) + bvec->bv_offset;
+	u64 cur;
+	int ret = 0;
 
-	read_extent_buffer(eb, fsid, btrfs_header_fsid(), BTRFS_FSID_SIZE);
-	while (fs_devices) {
-		u8 *metadata_uuid;
+	for (cur = bvec_start; cur < bvec_start + bvec->bv_len;
+	     cur += fs_info->nodesize) {
+		struct extent_buffer *eb;
+		bool uptodate;
 
-		/*
-		 * Checking the incompat flag is only valid for the current
-		 * fs. For seed devices it's forbidden to have their uuid
-		 * changed so reading ->fsid in this case is fine
-		 */
-		if (fs_devices == fs_info->fs_devices &&
-		    btrfs_fs_incompat(fs_info, METADATA_UUID))
-			metadata_uuid = fs_devices->metadata_uuid;
-		else
-			metadata_uuid = fs_devices->fsid;
+		eb = find_extent_buffer(fs_info, cur);
+		uptodate = btrfs_subpage_test_uptodate(fs_info, page, cur,
+						       fs_info->nodesize);
 
-		if (!memcmp(fsid, metadata_uuid, BTRFS_FSID_SIZE)) {
-			ret = 0;
-			break;
+		/* A dirty eb shouldn't disappear from buffer_radix */
+		if (WARN_ON(!eb))
+			return -EUCLEAN;
+
+		if (WARN_ON(cur != btrfs_header_bytenr(eb))) {
+			free_extent_buffer(eb);
+			return -EUCLEAN;
+		}
+		if (WARN_ON(!uptodate)) {
+			free_extent_buffer(eb);
+			return -EUCLEAN;
 		}
-		fs_devices = fs_devices->seed;
+
+		ret = csum_one_extent_buffer(eb);
+		free_extent_buffer(eb);
+		if (ret < 0)
+			return ret;
 	}
 	return ret;
 }
 
-static int btree_readpage_end_io_hook(struct btrfs_io_bio *io_bio,
-				      u64 phy_offset, struct page *page,
-				      u64 start, u64 end, int mirror)
+/*
+ * Checksum a dirty tree block before IO.  This has extra checks to make sure
+ * we only fill in the checksum field in the first page of a multi-page block.
+ * For subpage extent buffers we need bvec to also read the offset in the page.
+ */
+static int csum_dirty_buffer(struct btrfs_fs_info *fs_info, struct bio_vec *bvec)
 {
+	struct page *page = bvec->bv_page;
+	u64 start = page_offset(page);
 	u64 found_start;
-	int found_level;
 	struct extent_buffer *eb;
-	struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;
-	struct btrfs_fs_info *fs_info = root->fs_info;
-	u16 csum_size = btrfs_super_csum_size(fs_info->super_copy);
-	int ret = 0;
-	u8 result[BTRFS_CSUM_SIZE];
-	int reads_done;
 
-	if (!page->private)
-		goto out;
+	if (fs_info->nodesize < PAGE_SIZE)
+		return csum_dirty_subpage_buffers(fs_info, bvec);
 
 	eb = (struct extent_buffer *)page->private;
+	if (page != eb->pages[0])
+		return 0;
 
-	/* the pending IO might have been the only thing that kept this buffer
-	 * in memory.  Make sure we have a ref for all this other checks
+	found_start = btrfs_header_bytenr(eb);
+
+	if (test_bit(EXTENT_BUFFER_NO_CHECK, &eb->bflags)) {
+		WARN_ON(found_start != 0);
+		return 0;
+	}
+
+	/*
+	 * Please do not consolidate these warnings into a single if.
+	 * It is useful to know what went wrong.
 	 */
-	atomic_inc(&eb->refs);
+	if (WARN_ON(found_start != start))
+		return -EUCLEAN;
+	if (WARN_ON(!PageUptodate(page)))
+		return -EUCLEAN;
 
-	reads_done = atomic_dec_and_test(&eb->io_pages);
-	if (!reads_done)
-		goto err;
+	return csum_one_extent_buffer(eb);
+}
 
-	eb->read_mirror = mirror;
-	if (test_bit(EXTENT_BUFFER_READ_ERR, &eb->bflags)) {
-		ret = -EIO;
-		goto err;
-	}
+static int check_tree_block_fsid(struct extent_buffer *eb)
+{
+	struct btrfs_fs_info *fs_info = eb->fs_info;
+	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices, *seed_devs;
+	u8 fsid[BTRFS_FSID_SIZE];
+	u8 *metadata_uuid;
+
+	read_extent_buffer(eb, fsid, offsetof(struct btrfs_header, fsid),
+			   BTRFS_FSID_SIZE);
+	/*
+	 * Checking the incompat flag is only valid for the current fs. For
+	 * seed devices it's forbidden to have their uuid changed so reading
+	 * ->fsid in this case is fine
+	 */
+	if (btrfs_fs_incompat(fs_info, METADATA_UUID))
+		metadata_uuid = fs_devices->metadata_uuid;
+	else
+		metadata_uuid = fs_devices->fsid;
+
+	if (!memcmp(fsid, metadata_uuid, BTRFS_FSID_SIZE))
+		return 0;
+
+	list_for_each_entry(seed_devs, &fs_devices->seed_list, seed_list)
+		if (!memcmp(fsid, seed_devs->fsid, BTRFS_FSID_SIZE))
+			return 0;
+
+	return 1;
+}
+
+/* Do basic extent buffer checks at read time */
+static int validate_extent_buffer(struct extent_buffer *eb)
+{
+	struct btrfs_fs_info *fs_info = eb->fs_info;
+	u64 found_start;
+	const u32 csum_size = fs_info->csum_size;
+	u8 found_level;
+	u8 result[BTRFS_CSUM_SIZE];
+	const u8 *header_csum;
+	int ret = 0;
 
 	found_start = btrfs_header_bytenr(eb);
 	if (found_start != eb->start) {
-		btrfs_err_rl(fs_info, "bad tree block start, want %llu have %llu",
-			     eb->start, found_start);
+		btrfs_err_rl(fs_info,
+			"bad tree block start, mirror %u want %llu have %llu",
+			     eb->read_mirror, eb->start, found_start);
 		ret = -EIO;
-		goto err;
+		goto out;
 	}
 	if (check_tree_block_fsid(eb)) {
-		btrfs_err_rl(fs_info, "bad fsid on block %llu",
-			     eb->start);
+		btrfs_err_rl(fs_info, "bad fsid on logical %llu mirror %u",
+			     eb->start, eb->read_mirror);
 		ret = -EIO;
-		goto err;
+		goto out;
 	}
 	found_level = btrfs_header_level(eb);
 	if (found_level >= BTRFS_MAX_LEVEL) {
-		btrfs_err(fs_info, "bad tree block level %d on %llu",
-			  (int)btrfs_header_level(eb), eb->start);
+		btrfs_err(fs_info,
+			"bad tree block level, mirror %u level %d on logical %llu",
+			eb->read_mirror, btrfs_header_level(eb), eb->start);
 		ret = -EIO;
-		goto err;
+		goto out;
 	}
 
-	btrfs_set_buffer_lockdep_class(btrfs_header_owner(eb),
-				       eb, found_level);
-
-	ret = csum_tree_block(eb, result);
-	if (ret)
-		goto err;
-
-	if (memcmp_extent_buffer(eb, result, 0, csum_size)) {
-		u32 val;
-		u32 found = 0;
-
-		memcpy(&found, result, csum_size);
+	csum_tree_block(eb, result);
+	header_csum = page_address(eb->pages[0]) +
+		get_eb_offset_in_page(eb, offsetof(struct btrfs_header, csum));
 
-		read_extent_buffer(eb, &val, 0, csum_size);
+	if (memcmp(result, header_csum, csum_size) != 0) {
 		btrfs_warn_rl(fs_info,
-		"%s checksum verify failed on %llu wanted %x found %x level %d",
-			      fs_info->sb->s_id, eb->start,
-			      val, found, btrfs_header_level(eb));
+"checksum verify failed on logical %llu mirror %u wanted " CSUM_FMT " found " CSUM_FMT " level %d",
+			      eb->start, eb->read_mirror,
+			      CSUM_FMT_VALUE(csum_size, header_csum),
+			      CSUM_FMT_VALUE(csum_size, result),
+			      btrfs_header_level(eb));
 		ret = -EUCLEAN;
-		goto err;
+		goto out;
 	}
 
 	/*
@@ -683,79 +520,106 @@ static int btree_readpage_end_io_hook(struct btrfs_io_bio *io_bio,
 		set_extent_buffer_uptodate(eb);
 	else
 		btrfs_err(fs_info,
-			  "block=%llu read time tree block corruption detected",
-			  eb->start);
-err:
-	if (reads_done &&
-	    test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
-		btree_readahead_hook(eb, ret);
-
-	if (ret) {
-		/*
-		 * our io error hook is going to dec the io pages
-		 * again, we have to make sure it has something
-		 * to decrement
-		 */
-		atomic_inc(&eb->io_pages);
-		clear_extent_buffer_uptodate(eb);
-	}
-	free_extent_buffer(eb);
+		"read time tree block corruption detected on logical %llu mirror %u",
+			  eb->start, eb->read_mirror);
 out:
 	return ret;
 }
 
-static void end_workqueue_bio(struct bio *bio)
+static int validate_subpage_buffer(struct page *page, u64 start, u64 end,
+				   int mirror)
 {
-	struct btrfs_end_io_wq *end_io_wq = bio->bi_private;
-	struct btrfs_fs_info *fs_info;
-	struct btrfs_workqueue *wq;
-
-	fs_info = end_io_wq->info;
-	end_io_wq->status = bio->bi_status;
-
-	if (bio_op(bio) == REQ_OP_WRITE) {
-		if (end_io_wq->metadata == BTRFS_WQ_ENDIO_METADATA)
-			wq = fs_info->endio_meta_write_workers;
-		else if (end_io_wq->metadata == BTRFS_WQ_ENDIO_FREE_SPACE)
-			wq = fs_info->endio_freespace_worker;
-		else if (end_io_wq->metadata == BTRFS_WQ_ENDIO_RAID56)
-			wq = fs_info->endio_raid56_workers;
-		else
-			wq = fs_info->endio_write_workers;
-	} else {
-		if (unlikely(end_io_wq->metadata == BTRFS_WQ_ENDIO_DIO_REPAIR))
-			wq = fs_info->endio_repair_workers;
-		else if (end_io_wq->metadata == BTRFS_WQ_ENDIO_RAID56)
-			wq = fs_info->endio_raid56_workers;
-		else if (end_io_wq->metadata)
-			wq = fs_info->endio_meta_workers;
-		else
-			wq = fs_info->endio_workers;
+	struct btrfs_fs_info *fs_info = btrfs_sb(page->mapping->host->i_sb);
+	struct extent_buffer *eb;
+	bool reads_done;
+	int ret = 0;
+
+	/*
+	 * We don't allow bio merge for subpage metadata read, so we should
+	 * only get one eb for each endio hook.
+	 */
+	ASSERT(end == start + fs_info->nodesize - 1);
+	ASSERT(PagePrivate(page));
+
+	eb = find_extent_buffer(fs_info, start);
+	/*
+	 * When we are reading one tree block, eb must have been inserted into
+	 * the radix tree. If not, something is wrong.
+	 */
+	ASSERT(eb);
+
+	reads_done = atomic_dec_and_test(&eb->io_pages);
+	/* Subpage read must finish in page read */
+	ASSERT(reads_done);
+
+	eb->read_mirror = mirror;
+	if (test_bit(EXTENT_BUFFER_READ_ERR, &eb->bflags)) {
+		ret = -EIO;
+		goto err;
 	}
+	ret = validate_extent_buffer(eb);
+	if (ret < 0)
+		goto err;
 
-	btrfs_init_work(&end_io_wq->work, end_workqueue_fn, NULL, NULL);
-	btrfs_queue_work(wq, &end_io_wq->work);
+	set_extent_buffer_uptodate(eb);
+
+	free_extent_buffer(eb);
+	return ret;
+err:
+	/*
+	 * end_bio_extent_readpage decrements io_pages in case of error,
+	 * make sure it has something to decrement.
+	 */
+	atomic_inc(&eb->io_pages);
+	clear_extent_buffer_uptodate(eb);
+	free_extent_buffer(eb);
+	return ret;
 }
 
-blk_status_t btrfs_bio_wq_end_io(struct btrfs_fs_info *info, struct bio *bio,
-			enum btrfs_wq_endio_type metadata)
+int btrfs_validate_metadata_buffer(struct btrfs_bio *bbio,
+				   struct page *page, u64 start, u64 end,
+				   int mirror)
 {
-	struct btrfs_end_io_wq *end_io_wq;
+	struct extent_buffer *eb;
+	int ret = 0;
+	int reads_done;
 
-	end_io_wq = kmem_cache_alloc(btrfs_end_io_wq_cache, GFP_NOFS);
-	if (!end_io_wq)
-		return BLK_STS_RESOURCE;
+	ASSERT(page->private);
 
-	end_io_wq->private = bio->bi_private;
-	end_io_wq->end_io = bio->bi_end_io;
-	end_io_wq->info = info;
-	end_io_wq->status = 0;
-	end_io_wq->bio = bio;
-	end_io_wq->metadata = metadata;
+	if (btrfs_sb(page->mapping->host->i_sb)->nodesize < PAGE_SIZE)
+		return validate_subpage_buffer(page, start, end, mirror);
 
-	bio->bi_private = end_io_wq;
-	bio->bi_end_io = end_workqueue_bio;
-	return 0;
+	eb = (struct extent_buffer *)page->private;
+
+	/*
+	 * The pending IO might have been the only thing that kept this buffer
+	 * in memory.  Make sure we have a ref for all this other checks
+	 */
+	atomic_inc(&eb->refs);
+
+	reads_done = atomic_dec_and_test(&eb->io_pages);
+	if (!reads_done)
+		goto err;
+
+	eb->read_mirror = mirror;
+	if (test_bit(EXTENT_BUFFER_READ_ERR, &eb->bflags)) {
+		ret = -EIO;
+		goto err;
+	}
+	ret = validate_extent_buffer(eb);
+err:
+	if (ret) {
+		/*
+		 * our io error hook is going to dec the io pages
+		 * again, we have to make sure it has something
+		 * to decrement
+		 */
+		atomic_inc(&eb->io_pages);
+		clear_extent_buffer_uptodate(eb);
+	}
+	free_extent_buffer(eb);
+
+	return ret;
 }
 
 static void run_one_async_start(struct btrfs_work *work)
@@ -764,8 +628,8 @@ static void run_one_async_start(struct btrfs_work *work)
 	blk_status_t ret;
 
 	async = container_of(work, struct  async_submit_bio, work);
-	ret = async->submit_bio_start(async->private_data, async->bio,
-				      async->bio_offset);
+	ret = async->submit_bio_start(async->inode, async->bio,
+				      async->dio_file_offset);
 	if (ret)
 		async->status = ret;
 }
@@ -780,17 +644,14 @@ static void run_one_async_start(struct btrfs_work *work)
  */
 static void run_one_async_done(struct btrfs_work *work)
 {
-	struct async_submit_bio *async;
-	struct inode *inode;
-	blk_status_t ret;
-
-	async = container_of(work, struct  async_submit_bio, work);
-	inode = async->private_data;
+	struct async_submit_bio *async =
+		container_of(work, struct  async_submit_bio, work);
+	struct inode *inode = async->inode;
+	struct btrfs_bio *bbio = btrfs_bio(async->bio);
 
 	/* If an error occurred we just want to clean up the bio and move on */
 	if (async->status) {
-		async->bio->bi_status = async->status;
-		bio_endio(async->bio);
+		btrfs_bio_end_io(bbio, async->status);
 		return;
 	}
 
@@ -800,11 +661,7 @@ static void run_one_async_done(struct btrfs_work *work)
 	 * This changes nothing when cgroups aren't in use.
 	 */
 	async->bio->bi_opf |= REQ_CGROUP_PUNT;
-	ret = btrfs_map_bio(btrfs_sb(inode->i_sb), async->bio, async->mirror_num);
-	if (ret) {
-		async->bio->bi_status = ret;
-		bio_endio(async->bio);
-	}
+	btrfs_submit_bio(btrfs_sb(inode->i_sb), async->bio, async->mirror_num);
 }
 
 static void run_one_async_free(struct btrfs_work *work)
@@ -815,18 +672,25 @@ static void run_one_async_free(struct btrfs_work *work)
 	kfree(async);
 }
 
-blk_status_t btrfs_wq_submit_bio(struct btrfs_fs_info *fs_info, struct bio *bio,
-				 int mirror_num, unsigned long bio_flags,
-				 u64 bio_offset, void *private_data,
-				 extent_submit_bio_start_t *submit_bio_start)
+/*
+ * Submit bio to an async queue.
+ *
+ * Retrun:
+ * - true if the work has been succesfuly submitted
+ * - false in case of error
+ */
+bool btrfs_wq_submit_bio(struct inode *inode, struct bio *bio, int mirror_num,
+			 u64 dio_file_offset,
+			 extent_submit_bio_start_t *submit_bio_start)
 {
+	struct btrfs_fs_info *fs_info = BTRFS_I(inode)->root->fs_info;
 	struct async_submit_bio *async;
 
 	async = kmalloc(sizeof(*async), GFP_NOFS);
 	if (!async)
-		return BLK_STS_RESOURCE;
+		return false;
 
-	async->private_data = private_data;
+	async->inode = inode;
 	async->bio = bio;
 	async->mirror_num = mirror_num;
 	async->submit_bio_start = submit_bio_start;
@@ -834,15 +698,15 @@ blk_status_t btrfs_wq_submit_bio(struct btrfs_fs_info *fs_info, struct bio *bio,
 	btrfs_init_work(&async->work, run_one_async_start, run_one_async_done,
 			run_one_async_free);
 
-	async->bio_offset = bio_offset;
+	async->dio_file_offset = dio_file_offset;
 
 	async->status = 0;
 
 	if (op_is_sync(bio->bi_opf))
-		btrfs_set_work_high_priority(&async->work);
-
-	btrfs_queue_work(fs_info->workers, &async->work);
-	return 0;
+		btrfs_queue_work(fs_info->hipri_workers, &async->work);
+	else
+		btrfs_queue_work(fs_info->workers, &async->work);
+	return true;
 }
 
 static blk_status_t btree_csum_one_bio(struct bio *bio)
@@ -855,7 +719,7 @@ static blk_status_t btree_csum_one_bio(struct bio *bio)
 	ASSERT(!bio_flagged(bio, BIO_CLONED));
 	bio_for_each_segment_all(bvec, bio, iter_all) {
 		root = BTRFS_I(bvec->bv_page->mapping->host)->root;
-		ret = csum_dirty_buffer(root->fs_info, bvec->bv_page);
+		ret = csum_dirty_buffer(root->fs_info, bvec);
 		if (ret)
 			break;
 	}
@@ -863,91 +727,81 @@ static blk_status_t btree_csum_one_bio(struct bio *bio)
 	return errno_to_blk_status(ret);
 }
 
-static blk_status_t btree_submit_bio_start(void *private_data, struct bio *bio,
-					     u64 bio_offset)
+static blk_status_t btree_submit_bio_start(struct inode *inode, struct bio *bio,
+					   u64 dio_file_offset)
 {
 	/*
 	 * when we're called for a write, we're already in the async
-	 * submission context.  Just jump into btrfs_map_bio
+	 * submission context.  Just jump into btrfs_submit_bio.
 	 */
 	return btree_csum_one_bio(bio);
 }
 
-static int check_async_write(struct btrfs_fs_info *fs_info,
+static bool should_async_write(struct btrfs_fs_info *fs_info,
 			     struct btrfs_inode *bi)
 {
+	if (btrfs_is_zoned(fs_info))
+		return false;
 	if (atomic_read(&bi->sync_writers))
-		return 0;
+		return false;
 	if (test_bit(BTRFS_FS_CSUM_IMPL_FAST, &fs_info->flags))
-		return 0;
-	return 1;
+		return false;
+	return true;
 }
 
-static blk_status_t btree_submit_bio_hook(struct inode *inode, struct bio *bio,
-					  int mirror_num,
-					  unsigned long bio_flags)
+void btrfs_submit_metadata_bio(struct inode *inode, struct bio *bio, int mirror_num)
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	int async = check_async_write(fs_info, BTRFS_I(inode));
+	struct btrfs_bio *bbio = btrfs_bio(bio);
 	blk_status_t ret;
 
-	if (bio_op(bio) != REQ_OP_WRITE) {
-		/*
-		 * called for a read, do the setup so that checksum validation
-		 * can happen in the async kernel threads
-		 */
-		ret = btrfs_bio_wq_end_io(fs_info, bio,
-					  BTRFS_WQ_ENDIO_METADATA);
-		if (ret)
-			goto out_w_error;
-		ret = btrfs_map_bio(fs_info, bio, mirror_num);
-	} else if (!async) {
-		ret = btree_csum_one_bio(bio);
-		if (ret)
-			goto out_w_error;
-		ret = btrfs_map_bio(fs_info, bio, mirror_num);
-	} else {
-		/*
-		 * kthread helpers are used to submit writes so that
-		 * checksumming can happen in parallel across all CPUs
-		 */
-		ret = btrfs_wq_submit_bio(fs_info, bio, mirror_num, 0,
-					  0, inode, btree_submit_bio_start);
+	bio->bi_opf |= REQ_META;
+
+	if (btrfs_op(bio) != BTRFS_MAP_WRITE) {
+		btrfs_submit_bio(fs_info, bio, mirror_num);
+		return;
 	}
 
-	if (ret)
-		goto out_w_error;
-	return 0;
+	/*
+	 * Kthread helpers are used to submit writes so that checksumming can
+	 * happen in parallel across all CPUs.
+	 */
+	if (should_async_write(fs_info, BTRFS_I(inode)) &&
+	    btrfs_wq_submit_bio(inode, bio, mirror_num, 0, btree_submit_bio_start))
+		return;
 
-out_w_error:
-	bio->bi_status = ret;
-	bio_endio(bio);
-	return ret;
+	ret = btree_csum_one_bio(bio);
+	if (ret) {
+		btrfs_bio_end_io(bbio, ret);
+		return;
+	}
+
+	btrfs_submit_bio(fs_info, bio, mirror_num);
 }
 
 #ifdef CONFIG_MIGRATION
-static int btree_migratepage(struct address_space *mapping,
-			struct page *newpage, struct page *page,
-			enum migrate_mode mode)
+static int btree_migrate_folio(struct address_space *mapping,
+		struct folio *dst, struct folio *src, enum migrate_mode mode)
 {
 	/*
 	 * we can't safely write a btree page from here,
 	 * we haven't done the locking hook
 	 */
-	if (PageDirty(page))
+	if (folio_test_dirty(src))
 		return -EAGAIN;
 	/*
 	 * Buffers may be managed in a filesystem specific way.
 	 * We must have no buffers or drop them.
 	 */
-	if (page_has_private(page) &&
-	    !try_to_release_page(page, GFP_KERNEL))
+	if (folio_get_private(src) &&
+	    !filemap_release_folio(src, GFP_KERNEL))
 		return -EAGAIN;
-	return migrate_page(mapping, newpage, page, mode);
+	return migrate_folio(mapping, dst, src, mode);
 }
+#else
+#define btree_migrate_folio NULL
 #endif
 
-
 static int btree_writepages(struct address_space *mapping,
 			    struct writeback_control *wbc)
 {
@@ -970,114 +824,126 @@ static int btree_writepages(struct address_space *mapping,
 	return btree_write_cache_pages(mapping, wbc);
 }
 
-static int btree_readpage(struct file *file, struct page *page)
+static bool btree_release_folio(struct folio *folio, gfp_t gfp_flags)
 {
-	struct extent_io_tree *tree;
-	tree = &BTRFS_I(page->mapping->host)->io_tree;
-	return extent_read_full_page(tree, page, btree_get_extent, 0);
-}
-
-static int btree_releasepage(struct page *page, gfp_t gfp_flags)
-{
-	if (PageWriteback(page) || PageDirty(page))
-		return 0;
+	if (folio_test_writeback(folio) || folio_test_dirty(folio))
+		return false;
 
-	return try_release_extent_buffer(page);
+	return try_release_extent_buffer(&folio->page);
 }
 
-static void btree_invalidatepage(struct page *page, unsigned int offset,
-				 unsigned int length)
+static void btree_invalidate_folio(struct folio *folio, size_t offset,
+				 size_t length)
 {
 	struct extent_io_tree *tree;
-	tree = &BTRFS_I(page->mapping->host)->io_tree;
-	extent_invalidatepage(tree, page, offset);
-	btree_releasepage(page, GFP_NOFS);
-	if (PagePrivate(page)) {
-		btrfs_warn(BTRFS_I(page->mapping->host)->root->fs_info,
-			   "page private not zero on page %llu",
-			   (unsigned long long)page_offset(page));
-		ClearPagePrivate(page);
-		set_page_private(page, 0);
-		put_page(page);
+	tree = &BTRFS_I(folio->mapping->host)->io_tree;
+	extent_invalidate_folio(tree, folio, offset);
+	btree_release_folio(folio, GFP_NOFS);
+	if (folio_get_private(folio)) {
+		btrfs_warn(BTRFS_I(folio->mapping->host)->root->fs_info,
+			   "folio private not zero on folio %llu",
+			   (unsigned long long)folio_pos(folio));
+		folio_detach_private(folio);
 	}
 }
 
-static int btree_set_page_dirty(struct page *page)
-{
 #ifdef DEBUG
+static bool btree_dirty_folio(struct address_space *mapping,
+		struct folio *folio)
+{
+	struct btrfs_fs_info *fs_info = btrfs_sb(mapping->host->i_sb);
+	struct btrfs_subpage *subpage;
 	struct extent_buffer *eb;
+	int cur_bit = 0;
+	u64 page_start = folio_pos(folio);
+
+	if (fs_info->sectorsize == PAGE_SIZE) {
+		eb = folio_get_private(folio);
+		BUG_ON(!eb);
+		BUG_ON(!test_bit(EXTENT_BUFFER_DIRTY, &eb->bflags));
+		BUG_ON(!atomic_read(&eb->refs));
+		btrfs_assert_tree_write_locked(eb);
+		return filemap_dirty_folio(mapping, folio);
+	}
+	subpage = folio_get_private(folio);
+
+	ASSERT(subpage->dirty_bitmap);
+	while (cur_bit < BTRFS_SUBPAGE_BITMAP_SIZE) {
+		unsigned long flags;
+		u64 cur;
+		u16 tmp = (1 << cur_bit);
+
+		spin_lock_irqsave(&subpage->lock, flags);
+		if (!(tmp & subpage->dirty_bitmap)) {
+			spin_unlock_irqrestore(&subpage->lock, flags);
+			cur_bit++;
+			continue;
+		}
+		spin_unlock_irqrestore(&subpage->lock, flags);
+		cur = page_start + cur_bit * fs_info->sectorsize;
 
-	BUG_ON(!PagePrivate(page));
-	eb = (struct extent_buffer *)page->private;
-	BUG_ON(!eb);
-	BUG_ON(!test_bit(EXTENT_BUFFER_DIRTY, &eb->bflags));
-	BUG_ON(!atomic_read(&eb->refs));
-	btrfs_assert_tree_locked(eb);
-#endif
-	return __set_page_dirty_nobuffers(page);
+		eb = find_extent_buffer(fs_info, cur);
+		ASSERT(eb);
+		ASSERT(test_bit(EXTENT_BUFFER_DIRTY, &eb->bflags));
+		ASSERT(atomic_read(&eb->refs));
+		btrfs_assert_tree_write_locked(eb);
+		free_extent_buffer(eb);
+
+		cur_bit += (fs_info->nodesize >> fs_info->sectorsize_bits);
+	}
+	return filemap_dirty_folio(mapping, folio);
 }
+#else
+#define btree_dirty_folio filemap_dirty_folio
+#endif
 
 static const struct address_space_operations btree_aops = {
-	.readpage	= btree_readpage,
 	.writepages	= btree_writepages,
-	.releasepage	= btree_releasepage,
-	.invalidatepage = btree_invalidatepage,
-#ifdef CONFIG_MIGRATION
-	.migratepage	= btree_migratepage,
-#endif
-	.set_page_dirty = btree_set_page_dirty,
+	.release_folio	= btree_release_folio,
+	.invalidate_folio = btree_invalidate_folio,
+	.migrate_folio	= btree_migrate_folio,
+	.dirty_folio	= btree_dirty_folio,
 };
 
-void readahead_tree_block(struct btrfs_fs_info *fs_info, u64 bytenr)
-{
-	struct extent_buffer *buf = NULL;
-	int ret;
-
-	buf = btrfs_find_create_tree_block(fs_info, bytenr);
-	if (IS_ERR(buf))
-		return;
-
-	ret = read_extent_buffer_pages(buf, WAIT_NONE, 0);
-	if (ret < 0)
-		free_extent_buffer_stale(buf);
-	else
-		free_extent_buffer(buf);
-}
-
 struct extent_buffer *btrfs_find_create_tree_block(
 						struct btrfs_fs_info *fs_info,
-						u64 bytenr)
+						u64 bytenr, u64 owner_root,
+						int level)
 {
 	if (btrfs_is_testing(fs_info))
 		return alloc_test_extent_buffer(fs_info, bytenr);
-	return alloc_extent_buffer(fs_info, bytenr);
+	return alloc_extent_buffer(fs_info, bytenr, owner_root, level);
 }
 
 /*
  * Read tree block at logical address @bytenr and do variant basic but critical
  * verification.
  *
+ * @owner_root:		the objectid of the root owner for this block.
  * @parent_transid:	expected transid of this tree block, skip check if 0
  * @level:		expected level, mandatory check
  * @first_key:		expected key in slot 0, skip check if NULL
  */
 struct extent_buffer *read_tree_block(struct btrfs_fs_info *fs_info, u64 bytenr,
-				      u64 parent_transid, int level,
-				      struct btrfs_key *first_key)
+				      u64 owner_root, u64 parent_transid,
+				      int level, struct btrfs_key *first_key)
 {
 	struct extent_buffer *buf = NULL;
 	int ret;
 
-	buf = btrfs_find_create_tree_block(fs_info, bytenr);
+	buf = btrfs_find_create_tree_block(fs_info, bytenr, owner_root, level);
 	if (IS_ERR(buf))
 		return buf;
 
-	ret = btree_read_extent_buffer_pages(buf, parent_transid,
-					     level, first_key);
+	ret = btrfs_read_extent_buffer(buf, parent_transid, level, first_key);
 	if (ret) {
 		free_extent_buffer_stale(buf);
 		return ERR_PTR(ret);
 	}
+	if (btrfs_check_eb_owner(buf, owner_root)) {
+		free_extent_buffer_stale(buf);
+		return ERR_PTR(-EUCLEAN);
+	}
 	return buf;
 
 }
@@ -1087,61 +953,40 @@ void btrfs_clean_tree_block(struct extent_buffer *buf)
 	struct btrfs_fs_info *fs_info = buf->fs_info;
 	if (btrfs_header_generation(buf) ==
 	    fs_info->running_transaction->transid) {
-		btrfs_assert_tree_locked(buf);
+		btrfs_assert_tree_write_locked(buf);
 
 		if (test_and_clear_bit(EXTENT_BUFFER_DIRTY, &buf->bflags)) {
 			percpu_counter_add_batch(&fs_info->dirty_metadata_bytes,
 						 -buf->len,
 						 fs_info->dirty_metadata_batch);
-			/* ugh, clear_extent_buffer_dirty needs to lock the page */
-			btrfs_set_lock_blocking_write(buf);
 			clear_extent_buffer_dirty(buf);
 		}
 	}
 }
 
-static struct btrfs_subvolume_writers *btrfs_alloc_subvolume_writers(void)
-{
-	struct btrfs_subvolume_writers *writers;
-	int ret;
-
-	writers = kmalloc(sizeof(*writers), GFP_NOFS);
-	if (!writers)
-		return ERR_PTR(-ENOMEM);
-
-	ret = percpu_counter_init(&writers->counter, 0, GFP_NOFS);
-	if (ret < 0) {
-		kfree(writers);
-		return ERR_PTR(ret);
-	}
-
-	init_waitqueue_head(&writers->wait);
-	return writers;
-}
-
-static void
-btrfs_free_subvolume_writers(struct btrfs_subvolume_writers *writers)
-{
-	percpu_counter_destroy(&writers->counter);
-	kfree(writers);
-}
-
 static void __setup_root(struct btrfs_root *root, struct btrfs_fs_info *fs_info,
 			 u64 objectid)
 {
 	bool dummy = test_bit(BTRFS_FS_STATE_DUMMY_FS_INFO, &fs_info->fs_state);
+
+	memset(&root->root_key, 0, sizeof(root->root_key));
+	memset(&root->root_item, 0, sizeof(root->root_item));
+	memset(&root->defrag_progress, 0, sizeof(root->defrag_progress));
+	root->fs_info = fs_info;
+	root->root_key.objectid = objectid;
 	root->node = NULL;
 	root->commit_root = NULL;
 	root->state = 0;
-	root->orphan_cleanup_state = 0;
+	RB_CLEAR_NODE(&root->rb_node);
 
 	root->last_trans = 0;
-	root->highest_objectid = 0;
+	root->free_objectid = 0;
 	root->nr_delalloc_inodes = 0;
 	root->nr_ordered_extents = 0;
 	root->inode_tree = RB_ROOT;
 	INIT_RADIX_TREE(&root->delayed_nodes_tree, GFP_ATOMIC);
-	root->block_rsv = NULL;
+
+	btrfs_init_root_block_rsv(root);
 
 	INIT_LIST_HEAD(&root->dirty_list);
 	INIT_LIST_HEAD(&root->root_list);
@@ -1163,6 +1008,7 @@ static void __setup_root(struct btrfs_root *root, struct btrfs_fs_info *fs_info,
 	mutex_init(&root->log_mutex);
 	mutex_init(&root->ordered_extent_mutex);
 	mutex_init(&root->delalloc_mutex);
+	init_waitqueue_head(&root->qgroup_flush_wait);
 	init_waitqueue_head(&root->log_writer_wait);
 	init_waitqueue_head(&root->log_commit_wait[0]);
 	init_waitqueue_head(&root->log_commit_wait[1]);
@@ -1173,36 +1019,35 @@ static void __setup_root(struct btrfs_root *root, struct btrfs_fs_info *fs_info,
 	atomic_set(&root->log_writers, 0);
 	atomic_set(&root->log_batch, 0);
 	refcount_set(&root->refs, 1);
-	atomic_set(&root->will_be_snapshotted, 0);
 	atomic_set(&root->snapshot_force_cow, 0);
 	atomic_set(&root->nr_swapfiles, 0);
 	root->log_transid = 0;
 	root->log_transid_committed = -1;
 	root->last_log_commit = 0;
-	if (!dummy)
+	root->anon_dev = 0;
+	if (!dummy) {
 		extent_io_tree_init(fs_info, &root->dirty_log_pages,
 				    IO_TREE_ROOT_DIRTY_LOG_PAGES, NULL);
-
-	memset(&root->root_key, 0, sizeof(root->root_key));
-	memset(&root->root_item, 0, sizeof(root->root_item));
-	memset(&root->defrag_progress, 0, sizeof(root->defrag_progress));
-	if (!dummy)
-		root->defrag_trans_start = fs_info->generation;
-	else
-		root->defrag_trans_start = 0;
-	root->root_key.objectid = objectid;
-	root->anon_dev = 0;
+		extent_io_tree_init(fs_info, &root->log_csum_range,
+				    IO_TREE_LOG_CSUM_RANGE, NULL);
+	}
 
 	spin_lock_init(&root->root_item_lock);
 	btrfs_qgroup_init_swapped_blocks(&root->swapped_blocks);
+#ifdef CONFIG_BTRFS_DEBUG
+	INIT_LIST_HEAD(&root->leak_list);
+	spin_lock(&fs_info->fs_roots_radix_lock);
+	list_add_tail(&root->leak_list, &fs_info->allocated_roots);
+	spin_unlock(&fs_info->fs_roots_radix_lock);
+#endif
 }
 
 static struct btrfs_root *btrfs_alloc_root(struct btrfs_fs_info *fs_info,
-		gfp_t flags)
+					   u64 objectid, gfp_t flags)
 {
 	struct btrfs_root *root = kzalloc(sizeof(*root), flags);
 	if (root)
-		root->fs_info = fs_info;
+		__setup_root(root, fs_info, objectid);
 	return root;
 }
 
@@ -1215,18 +1060,113 @@ struct btrfs_root *btrfs_alloc_dummy_root(struct btrfs_fs_info *fs_info)
 	if (!fs_info)
 		return ERR_PTR(-EINVAL);
 
-	root = btrfs_alloc_root(fs_info, GFP_KERNEL);
+	root = btrfs_alloc_root(fs_info, BTRFS_ROOT_TREE_OBJECTID, GFP_KERNEL);
 	if (!root)
 		return ERR_PTR(-ENOMEM);
 
 	/* We don't use the stripesize in selftest, set it as sectorsize */
-	__setup_root(root, fs_info, BTRFS_ROOT_TREE_OBJECTID);
 	root->alloc_bytenr = 0;
 
 	return root;
 }
 #endif
 
+static int global_root_cmp(struct rb_node *a_node, const struct rb_node *b_node)
+{
+	const struct btrfs_root *a = rb_entry(a_node, struct btrfs_root, rb_node);
+	const struct btrfs_root *b = rb_entry(b_node, struct btrfs_root, rb_node);
+
+	return btrfs_comp_cpu_keys(&a->root_key, &b->root_key);
+}
+
+static int global_root_key_cmp(const void *k, const struct rb_node *node)
+{
+	const struct btrfs_key *key = k;
+	const struct btrfs_root *root = rb_entry(node, struct btrfs_root, rb_node);
+
+	return btrfs_comp_cpu_keys(key, &root->root_key);
+}
+
+int btrfs_global_root_insert(struct btrfs_root *root)
+{
+	struct btrfs_fs_info *fs_info = root->fs_info;
+	struct rb_node *tmp;
+
+	write_lock(&fs_info->global_root_lock);
+	tmp = rb_find_add(&root->rb_node, &fs_info->global_root_tree, global_root_cmp);
+	write_unlock(&fs_info->global_root_lock);
+	ASSERT(!tmp);
+
+	return tmp ? -EEXIST : 0;
+}
+
+void btrfs_global_root_delete(struct btrfs_root *root)
+{
+	struct btrfs_fs_info *fs_info = root->fs_info;
+
+	write_lock(&fs_info->global_root_lock);
+	rb_erase(&root->rb_node, &fs_info->global_root_tree);
+	write_unlock(&fs_info->global_root_lock);
+}
+
+struct btrfs_root *btrfs_global_root(struct btrfs_fs_info *fs_info,
+				     struct btrfs_key *key)
+{
+	struct rb_node *node;
+	struct btrfs_root *root = NULL;
+
+	read_lock(&fs_info->global_root_lock);
+	node = rb_find(key, &fs_info->global_root_tree, global_root_key_cmp);
+	if (node)
+		root = container_of(node, struct btrfs_root, rb_node);
+	read_unlock(&fs_info->global_root_lock);
+
+	return root;
+}
+
+static u64 btrfs_global_root_id(struct btrfs_fs_info *fs_info, u64 bytenr)
+{
+	struct btrfs_block_group *block_group;
+	u64 ret;
+
+	if (!btrfs_fs_incompat(fs_info, EXTENT_TREE_V2))
+		return 0;
+
+	if (bytenr)
+		block_group = btrfs_lookup_block_group(fs_info, bytenr);
+	else
+		block_group = btrfs_lookup_first_block_group(fs_info, bytenr);
+	ASSERT(block_group);
+	if (!block_group)
+		return 0;
+	ret = block_group->global_root_id;
+	btrfs_put_block_group(block_group);
+
+	return ret;
+}
+
+struct btrfs_root *btrfs_csum_root(struct btrfs_fs_info *fs_info, u64 bytenr)
+{
+	struct btrfs_key key = {
+		.objectid = BTRFS_CSUM_TREE_OBJECTID,
+		.type = BTRFS_ROOT_ITEM_KEY,
+		.offset = btrfs_global_root_id(fs_info, bytenr),
+	};
+
+	return btrfs_global_root(fs_info, &key);
+}
+
+struct btrfs_root *btrfs_extent_root(struct btrfs_fs_info *fs_info, u64 bytenr)
+{
+	struct btrfs_key key = {
+		.objectid = BTRFS_EXTENT_TREE_OBJECTID,
+		.type = BTRFS_ROOT_ITEM_KEY,
+		.offset = btrfs_global_root_id(fs_info, bytenr),
+	};
+
+	return btrfs_global_root(fs_info, &key);
+}
+
 struct btrfs_root *btrfs_create_tree(struct btrfs_trans_handle *trans,
 				     u64 objectid)
 {
@@ -1237,28 +1177,27 @@ struct btrfs_root *btrfs_create_tree(struct btrfs_trans_handle *trans,
 	struct btrfs_key key;
 	unsigned int nofs_flag;
 	int ret = 0;
-	uuid_le uuid = NULL_UUID_LE;
 
 	/*
 	 * We're holding a transaction handle, so use a NOFS memory allocation
 	 * context to avoid deadlock if reclaim happens.
 	 */
 	nofs_flag = memalloc_nofs_save();
-	root = btrfs_alloc_root(fs_info, GFP_KERNEL);
+	root = btrfs_alloc_root(fs_info, objectid, GFP_KERNEL);
 	memalloc_nofs_restore(nofs_flag);
 	if (!root)
 		return ERR_PTR(-ENOMEM);
 
-	__setup_root(root, fs_info, objectid);
 	root->root_key.objectid = objectid;
 	root->root_key.type = BTRFS_ROOT_ITEM_KEY;
 	root->root_key.offset = 0;
 
-	leaf = btrfs_alloc_tree_block(trans, root, 0, objectid, NULL, 0, 0, 0);
+	leaf = btrfs_alloc_tree_block(trans, root, 0, objectid, NULL, 0, 0, 0,
+				      BTRFS_NESTING_NORMAL);
 	if (IS_ERR(leaf)) {
 		ret = PTR_ERR(leaf);
 		leaf = NULL;
-		goto fail;
+		goto fail_unlock;
 	}
 
 	root->node = leaf;
@@ -1267,8 +1206,8 @@ struct btrfs_root *btrfs_create_tree(struct btrfs_trans_handle *trans,
 	root->commit_root = btrfs_root_node(root);
 	set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
 
-	root->root_item.flags = 0;
-	root->root_item.byte_limit = 0;
+	btrfs_set_root_flags(&root->root_item, 0);
+	btrfs_set_root_limit(&root->root_item, 0);
 	btrfs_set_root_bytenr(&root->root_item, leaf->start);
 	btrfs_set_root_generation(&root->root_item, trans->transid);
 	btrfs_set_root_level(&root->root_item, 0);
@@ -1277,9 +1216,12 @@ struct btrfs_root *btrfs_create_tree(struct btrfs_trans_handle *trans,
 	btrfs_set_root_last_snapshot(&root->root_item, 0);
 	btrfs_set_root_dirid(&root->root_item, 0);
 	if (is_fstree(objectid))
-		uuid_le_gen(&uuid);
-	memcpy(root->root_item.uuid, uuid.b, BTRFS_UUID_SIZE);
-	root->root_item.drop_level = 0;
+		generate_random_guid(root->root_item.uuid);
+	else
+		export_guid(root->root_item.uuid, &guid_null);
+	btrfs_set_root_drop_level(&root->root_item, 0);
+
+	btrfs_tree_unlock(leaf);
 
 	key.objectid = objectid;
 	key.type = BTRFS_ROOT_ITEM_KEY;
@@ -1288,17 +1230,13 @@ struct btrfs_root *btrfs_create_tree(struct btrfs_trans_handle *trans,
 	if (ret)
 		goto fail;
 
-	btrfs_tree_unlock(leaf);
-
 	return root;
 
-fail:
-	if (leaf) {
+fail_unlock:
+	if (leaf)
 		btrfs_tree_unlock(leaf);
-		free_extent_buffer(root->commit_root);
-		free_extent_buffer(leaf);
-	}
-	kfree(root);
+fail:
+	btrfs_put_root(root);
 
 	return ERR_PTR(ret);
 }
@@ -1307,39 +1245,44 @@ static struct btrfs_root *alloc_log_tree(struct btrfs_trans_handle *trans,
 					 struct btrfs_fs_info *fs_info)
 {
 	struct btrfs_root *root;
-	struct extent_buffer *leaf;
 
-	root = btrfs_alloc_root(fs_info, GFP_NOFS);
+	root = btrfs_alloc_root(fs_info, BTRFS_TREE_LOG_OBJECTID, GFP_NOFS);
 	if (!root)
 		return ERR_PTR(-ENOMEM);
 
-	__setup_root(root, fs_info, BTRFS_TREE_LOG_OBJECTID);
-
 	root->root_key.objectid = BTRFS_TREE_LOG_OBJECTID;
 	root->root_key.type = BTRFS_ROOT_ITEM_KEY;
 	root->root_key.offset = BTRFS_TREE_LOG_OBJECTID;
 
+	return root;
+}
+
+int btrfs_alloc_log_tree_node(struct btrfs_trans_handle *trans,
+			      struct btrfs_root *root)
+{
+	struct extent_buffer *leaf;
+
 	/*
-	 * DON'T set REF_COWS for log trees
+	 * DON'T set SHAREABLE bit for log trees.
 	 *
-	 * log trees do not get reference counted because they go away
-	 * before a real commit is actually done.  They do store pointers
-	 * to file data extents, and those reference counts still get
-	 * updated (along with back refs to the log tree).
+	 * Log trees are not exposed to user space thus can't be snapshotted,
+	 * and they go away before a real commit is actually done.
+	 *
+	 * They do store pointers to file data extents, and those reference
+	 * counts still get updated (along with back refs to the log tree).
 	 */
 
 	leaf = btrfs_alloc_tree_block(trans, root, 0, BTRFS_TREE_LOG_OBJECTID,
-			NULL, 0, 0, 0);
-	if (IS_ERR(leaf)) {
-		kfree(root);
-		return ERR_CAST(leaf);
-	}
+			NULL, 0, 0, 0, BTRFS_NESTING_NORMAL);
+	if (IS_ERR(leaf))
+		return PTR_ERR(leaf);
 
 	root->node = leaf;
 
 	btrfs_mark_buffer_dirty(root->node);
 	btrfs_tree_unlock(root->node);
-	return root;
+
+	return 0;
 }
 
 int btrfs_init_log_root_tree(struct btrfs_trans_handle *trans,
@@ -1350,6 +1293,16 @@ int btrfs_init_log_root_tree(struct btrfs_trans_handle *trans,
 	log_root = alloc_log_tree(trans, fs_info);
 	if (IS_ERR(log_root))
 		return PTR_ERR(log_root);
+
+	if (!btrfs_is_zoned(fs_info)) {
+		int ret = btrfs_alloc_log_tree_node(trans, log_root);
+
+		if (ret) {
+			btrfs_put_root(log_root);
+			return ret;
+		}
+	}
+
 	WARN_ON(fs_info->log_root_tree);
 	fs_info->log_root_tree = log_root;
 	return 0;
@@ -1361,11 +1314,18 @@ int btrfs_add_log_tree(struct btrfs_trans_handle *trans,
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct btrfs_root *log_root;
 	struct btrfs_inode_item *inode_item;
+	int ret;
 
 	log_root = alloc_log_tree(trans, fs_info);
 	if (IS_ERR(log_root))
 		return PTR_ERR(log_root);
 
+	ret = btrfs_alloc_log_tree_node(trans, log_root);
+	if (ret) {
+		btrfs_put_root(log_root);
+		return ret;
+	}
+
 	log_root->last_trans = trans->transid;
 	log_root->root_key.offset = root->root_key.objectid;
 
@@ -1387,115 +1347,130 @@ int btrfs_add_log_tree(struct btrfs_trans_handle *trans,
 	return 0;
 }
 
-static struct btrfs_root *btrfs_read_tree_root(struct btrfs_root *tree_root,
-					       struct btrfs_key *key)
+static struct btrfs_root *read_tree_root_path(struct btrfs_root *tree_root,
+					      struct btrfs_path *path,
+					      struct btrfs_key *key)
 {
 	struct btrfs_root *root;
 	struct btrfs_fs_info *fs_info = tree_root->fs_info;
-	struct btrfs_path *path;
 	u64 generation;
 	int ret;
 	int level;
 
-	path = btrfs_alloc_path();
-	if (!path)
+	root = btrfs_alloc_root(fs_info, key->objectid, GFP_NOFS);
+	if (!root)
 		return ERR_PTR(-ENOMEM);
 
-	root = btrfs_alloc_root(fs_info, GFP_NOFS);
-	if (!root) {
-		ret = -ENOMEM;
-		goto alloc_fail;
-	}
-
-	__setup_root(root, fs_info, key->objectid);
-
 	ret = btrfs_find_root(tree_root, key, path,
 			      &root->root_item, &root->root_key);
 	if (ret) {
 		if (ret > 0)
 			ret = -ENOENT;
-		goto find_fail;
+		goto fail;
 	}
 
 	generation = btrfs_root_generation(&root->root_item);
 	level = btrfs_root_level(&root->root_item);
 	root->node = read_tree_block(fs_info,
 				     btrfs_root_bytenr(&root->root_item),
-				     generation, level, NULL);
+				     key->objectid, generation, level, NULL);
 	if (IS_ERR(root->node)) {
 		ret = PTR_ERR(root->node);
-		goto find_fail;
-	} else if (!btrfs_buffer_uptodate(root->node, generation, 0)) {
+		root->node = NULL;
+		goto fail;
+	}
+	if (!btrfs_buffer_uptodate(root->node, generation, 0)) {
 		ret = -EIO;
-		free_extent_buffer(root->node);
-		goto find_fail;
+		goto fail;
+	}
+
+	/*
+	 * For real fs, and not log/reloc trees, root owner must
+	 * match its root node owner
+	 */
+	if (!test_bit(BTRFS_FS_STATE_DUMMY_FS_INFO, &fs_info->fs_state) &&
+	    root->root_key.objectid != BTRFS_TREE_LOG_OBJECTID &&
+	    root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID &&
+	    root->root_key.objectid != btrfs_header_owner(root->node)) {
+		btrfs_crit(fs_info,
+"root=%llu block=%llu, tree root owner mismatch, have %llu expect %llu",
+			   root->root_key.objectid, root->node->start,
+			   btrfs_header_owner(root->node),
+			   root->root_key.objectid);
+		ret = -EUCLEAN;
+		goto fail;
 	}
 	root->commit_root = btrfs_root_node(root);
-out:
-	btrfs_free_path(path);
 	return root;
-
-find_fail:
-	kfree(root);
-alloc_fail:
-	root = ERR_PTR(ret);
-	goto out;
+fail:
+	btrfs_put_root(root);
+	return ERR_PTR(ret);
 }
 
-struct btrfs_root *btrfs_read_fs_root(struct btrfs_root *tree_root,
-				      struct btrfs_key *location)
+struct btrfs_root *btrfs_read_tree_root(struct btrfs_root *tree_root,
+					struct btrfs_key *key)
 {
 	struct btrfs_root *root;
+	struct btrfs_path *path;
 
-	root = btrfs_read_tree_root(tree_root, location);
-	if (IS_ERR(root))
-		return root;
-
-	if (root->root_key.objectid != BTRFS_TREE_LOG_OBJECTID) {
-		set_bit(BTRFS_ROOT_REF_COWS, &root->state);
-		btrfs_check_and_init_root_item(&root->root_item);
-	}
+	path = btrfs_alloc_path();
+	if (!path)
+		return ERR_PTR(-ENOMEM);
+	root = read_tree_root_path(tree_root, path, key);
+	btrfs_free_path(path);
 
 	return root;
 }
 
-int btrfs_init_fs_root(struct btrfs_root *root)
+/*
+ * Initialize subvolume root in-memory structure
+ *
+ * @anon_dev:	anonymous device to attach to the root, if zero, allocate new
+ */
+static int btrfs_init_fs_root(struct btrfs_root *root, dev_t anon_dev)
 {
 	int ret;
-	struct btrfs_subvolume_writers *writers;
+	unsigned int nofs_flag;
 
-	root->free_ino_ctl = kzalloc(sizeof(*root->free_ino_ctl), GFP_NOFS);
-	root->free_ino_pinned = kzalloc(sizeof(*root->free_ino_pinned),
-					GFP_NOFS);
-	if (!root->free_ino_pinned || !root->free_ino_ctl) {
-		ret = -ENOMEM;
+	/*
+	 * We might be called under a transaction (e.g. indirect backref
+	 * resolution) which could deadlock if it triggers memory reclaim
+	 */
+	nofs_flag = memalloc_nofs_save();
+	ret = btrfs_drew_lock_init(&root->snapshot_lock);
+	memalloc_nofs_restore(nofs_flag);
+	if (ret)
 		goto fail;
-	}
 
-	writers = btrfs_alloc_subvolume_writers();
-	if (IS_ERR(writers)) {
-		ret = PTR_ERR(writers);
-		goto fail;
+	if (root->root_key.objectid != BTRFS_TREE_LOG_OBJECTID &&
+	    !btrfs_is_data_reloc_root(root)) {
+		set_bit(BTRFS_ROOT_SHAREABLE, &root->state);
+		btrfs_check_and_init_root_item(&root->root_item);
 	}
-	root->subv_writers = writers;
-
-	btrfs_init_free_ino_ctl(root);
-	spin_lock_init(&root->ino_cache_lock);
-	init_waitqueue_head(&root->ino_cache_wait);
 
-	ret = get_anon_bdev(&root->anon_dev);
-	if (ret)
-		goto fail;
+	/*
+	 * Don't assign anonymous block device to roots that are not exposed to
+	 * userspace, the id pool is limited to 1M
+	 */
+	if (is_fstree(root->root_key.objectid) &&
+	    btrfs_root_refs(&root->root_item) > 0) {
+		if (!anon_dev) {
+			ret = get_anon_bdev(&root->anon_dev);
+			if (ret)
+				goto fail;
+		} else {
+			root->anon_dev = anon_dev;
+		}
+	}
 
 	mutex_lock(&root->objectid_mutex);
-	ret = btrfs_find_highest_objectid(root,
-					&root->highest_objectid);
+	ret = btrfs_init_root_free_objectid(root);
 	if (ret) {
 		mutex_unlock(&root->objectid_mutex);
 		goto fail;
 	}
 
-	ASSERT(root->highest_objectid <= BTRFS_LAST_FREE_OBJECTID);
+	ASSERT(root->free_objectid <= BTRFS_LAST_FREE_OBJECTID);
 
 	mutex_unlock(&root->objectid_mutex);
 
@@ -1505,18 +1480,56 @@ fail:
 	return ret;
 }
 
-struct btrfs_root *btrfs_lookup_fs_root(struct btrfs_fs_info *fs_info,
-					u64 root_id)
+static struct btrfs_root *btrfs_lookup_fs_root(struct btrfs_fs_info *fs_info,
+					       u64 root_id)
 {
 	struct btrfs_root *root;
 
 	spin_lock(&fs_info->fs_roots_radix_lock);
 	root = radix_tree_lookup(&fs_info->fs_roots_radix,
 				 (unsigned long)root_id);
+	if (root)
+		root = btrfs_grab_root(root);
 	spin_unlock(&fs_info->fs_roots_radix_lock);
 	return root;
 }
 
+static struct btrfs_root *btrfs_get_global_root(struct btrfs_fs_info *fs_info,
+						u64 objectid)
+{
+	struct btrfs_key key = {
+		.objectid = objectid,
+		.type = BTRFS_ROOT_ITEM_KEY,
+		.offset = 0,
+	};
+
+	if (objectid == BTRFS_ROOT_TREE_OBJECTID)
+		return btrfs_grab_root(fs_info->tree_root);
+	if (objectid == BTRFS_EXTENT_TREE_OBJECTID)
+		return btrfs_grab_root(btrfs_global_root(fs_info, &key));
+	if (objectid == BTRFS_CHUNK_TREE_OBJECTID)
+		return btrfs_grab_root(fs_info->chunk_root);
+	if (objectid == BTRFS_DEV_TREE_OBJECTID)
+		return btrfs_grab_root(fs_info->dev_root);
+	if (objectid == BTRFS_CSUM_TREE_OBJECTID)
+		return btrfs_grab_root(btrfs_global_root(fs_info, &key));
+	if (objectid == BTRFS_QUOTA_TREE_OBJECTID)
+		return btrfs_grab_root(fs_info->quota_root) ?
+			fs_info->quota_root : ERR_PTR(-ENOENT);
+	if (objectid == BTRFS_UUID_TREE_OBJECTID)
+		return btrfs_grab_root(fs_info->uuid_root) ?
+			fs_info->uuid_root : ERR_PTR(-ENOENT);
+	if (objectid == BTRFS_BLOCK_GROUP_TREE_OBJECTID)
+		return btrfs_grab_root(fs_info->block_group_root) ?
+			fs_info->block_group_root : ERR_PTR(-ENOENT);
+	if (objectid == BTRFS_FREE_SPACE_TREE_OBJECTID) {
+		struct btrfs_root *root = btrfs_global_root(fs_info, &key);
+
+		return btrfs_grab_root(root) ? root : ERR_PTR(-ENOENT);
+	}
+	return NULL;
+}
+
 int btrfs_insert_fs_root(struct btrfs_fs_info *fs_info,
 			 struct btrfs_root *root)
 {
@@ -1530,51 +1543,123 @@ int btrfs_insert_fs_root(struct btrfs_fs_info *fs_info,
 	ret = radix_tree_insert(&fs_info->fs_roots_radix,
 				(unsigned long)root->root_key.objectid,
 				root);
-	if (ret == 0)
+	if (ret == 0) {
+		btrfs_grab_root(root);
 		set_bit(BTRFS_ROOT_IN_RADIX, &root->state);
+	}
 	spin_unlock(&fs_info->fs_roots_radix_lock);
 	radix_tree_preload_end();
 
 	return ret;
 }
 
-struct btrfs_root *btrfs_get_fs_root(struct btrfs_fs_info *fs_info,
-				     struct btrfs_key *location,
-				     bool check_ref)
+void btrfs_check_leaked_roots(struct btrfs_fs_info *fs_info)
+{
+#ifdef CONFIG_BTRFS_DEBUG
+	struct btrfs_root *root;
+
+	while (!list_empty(&fs_info->allocated_roots)) {
+		char buf[BTRFS_ROOT_NAME_BUF_LEN];
+
+		root = list_first_entry(&fs_info->allocated_roots,
+					struct btrfs_root, leak_list);
+		btrfs_err(fs_info, "leaked root %s refcount %d",
+			  btrfs_root_name(&root->root_key, buf),
+			  refcount_read(&root->refs));
+		while (refcount_read(&root->refs) > 1)
+			btrfs_put_root(root);
+		btrfs_put_root(root);
+	}
+#endif
+}
+
+static void free_global_roots(struct btrfs_fs_info *fs_info)
+{
+	struct btrfs_root *root;
+	struct rb_node *node;
+
+	while ((node = rb_first_postorder(&fs_info->global_root_tree)) != NULL) {
+		root = rb_entry(node, struct btrfs_root, rb_node);
+		rb_erase(&root->rb_node, &fs_info->global_root_tree);
+		btrfs_put_root(root);
+	}
+}
+
+void btrfs_free_fs_info(struct btrfs_fs_info *fs_info)
+{
+	percpu_counter_destroy(&fs_info->dirty_metadata_bytes);
+	percpu_counter_destroy(&fs_info->delalloc_bytes);
+	percpu_counter_destroy(&fs_info->ordered_bytes);
+	percpu_counter_destroy(&fs_info->dev_replace.bio_counter);
+	btrfs_free_csum_hash(fs_info);
+	btrfs_free_stripe_hash_table(fs_info);
+	btrfs_free_ref_cache(fs_info);
+	kfree(fs_info->balance_ctl);
+	kfree(fs_info->delayed_root);
+	free_global_roots(fs_info);
+	btrfs_put_root(fs_info->tree_root);
+	btrfs_put_root(fs_info->chunk_root);
+	btrfs_put_root(fs_info->dev_root);
+	btrfs_put_root(fs_info->quota_root);
+	btrfs_put_root(fs_info->uuid_root);
+	btrfs_put_root(fs_info->fs_root);
+	btrfs_put_root(fs_info->data_reloc_root);
+	btrfs_put_root(fs_info->block_group_root);
+	btrfs_check_leaked_roots(fs_info);
+	btrfs_extent_buffer_leak_debug_check(fs_info);
+	kfree(fs_info->super_copy);
+	kfree(fs_info->super_for_commit);
+	kfree(fs_info->subpage_info);
+	kvfree(fs_info);
+}
+
+
+/*
+ * Get an in-memory reference of a root structure.
+ *
+ * For essential trees like root/extent tree, we grab it from fs_info directly.
+ * For subvolume trees, we check the cached filesystem roots first. If not
+ * found, then read it from disk and add it to cached fs roots.
+ *
+ * Caller should release the root by calling btrfs_put_root() after the usage.
+ *
+ * NOTE: Reloc and log trees can't be read by this function as they share the
+ *	 same root objectid.
+ *
+ * @objectid:	root id
+ * @anon_dev:	preallocated anonymous block device number for new roots,
+ * 		pass 0 for new allocation.
+ * @check_ref:	whether to check root item references, If true, return -ENOENT
+ *		for orphan roots
+ */
+static struct btrfs_root *btrfs_get_root_ref(struct btrfs_fs_info *fs_info,
+					     u64 objectid, dev_t anon_dev,
+					     bool check_ref)
 {
 	struct btrfs_root *root;
 	struct btrfs_path *path;
 	struct btrfs_key key;
 	int ret;
 
-	if (location->objectid == BTRFS_ROOT_TREE_OBJECTID)
-		return fs_info->tree_root;
-	if (location->objectid == BTRFS_EXTENT_TREE_OBJECTID)
-		return fs_info->extent_root;
-	if (location->objectid == BTRFS_CHUNK_TREE_OBJECTID)
-		return fs_info->chunk_root;
-	if (location->objectid == BTRFS_DEV_TREE_OBJECTID)
-		return fs_info->dev_root;
-	if (location->objectid == BTRFS_CSUM_TREE_OBJECTID)
-		return fs_info->csum_root;
-	if (location->objectid == BTRFS_QUOTA_TREE_OBJECTID)
-		return fs_info->quota_root ? fs_info->quota_root :
-					     ERR_PTR(-ENOENT);
-	if (location->objectid == BTRFS_UUID_TREE_OBJECTID)
-		return fs_info->uuid_root ? fs_info->uuid_root :
-					    ERR_PTR(-ENOENT);
-	if (location->objectid == BTRFS_FREE_SPACE_TREE_OBJECTID)
-		return fs_info->free_space_root ? fs_info->free_space_root :
-						  ERR_PTR(-ENOENT);
+	root = btrfs_get_global_root(fs_info, objectid);
+	if (root)
+		return root;
 again:
-	root = btrfs_lookup_fs_root(fs_info, location->objectid);
+	root = btrfs_lookup_fs_root(fs_info, objectid);
 	if (root) {
-		if (check_ref && btrfs_root_refs(&root->root_item) == 0)
+		/* Shouldn't get preallocated anon_dev for cached roots */
+		ASSERT(!anon_dev);
+		if (check_ref && btrfs_root_refs(&root->root_item) == 0) {
+			btrfs_put_root(root);
 			return ERR_PTR(-ENOENT);
+		}
 		return root;
 	}
 
-	root = btrfs_read_fs_root(fs_info->tree_root, location);
+	key.objectid = objectid;
+	key.type = BTRFS_ROOT_ITEM_KEY;
+	key.offset = (u64)-1;
+	root = btrfs_read_tree_root(fs_info->tree_root, &key);
 	if (IS_ERR(root))
 		return root;
 
@@ -1583,7 +1668,7 @@ again:
 		goto fail;
 	}
 
-	ret = btrfs_init_fs_root(root);
+	ret = btrfs_init_fs_root(root, anon_dev);
 	if (ret)
 		goto fail;
 
@@ -1594,7 +1679,7 @@ again:
 	}
 	key.objectid = BTRFS_ORPHAN_OBJECTID;
 	key.type = BTRFS_ORPHAN_ITEM_KEY;
-	key.offset = location->objectid;
+	key.offset = objectid;
 
 	ret = btrfs_search_slot(NULL, fs_info->tree_root, &key, path, 0, 0);
 	btrfs_free_path(path);
@@ -1606,61 +1691,101 @@ again:
 	ret = btrfs_insert_fs_root(fs_info, root);
 	if (ret) {
 		if (ret == -EEXIST) {
-			btrfs_free_fs_root(root);
+			btrfs_put_root(root);
 			goto again;
 		}
 		goto fail;
 	}
 	return root;
 fail:
-	btrfs_free_fs_root(root);
+	/*
+	 * If our caller provided us an anonymous device, then it's his
+	 * responsibility to free it in case we fail. So we have to set our
+	 * root's anon_dev to 0 to avoid a double free, once by btrfs_put_root()
+	 * and once again by our caller.
+	 */
+	if (anon_dev)
+		root->anon_dev = 0;
+	btrfs_put_root(root);
 	return ERR_PTR(ret);
 }
 
-static int btrfs_congested_fn(void *congested_data, int bdi_bits)
+/*
+ * Get in-memory reference of a root structure
+ *
+ * @objectid:	tree objectid
+ * @check_ref:	if set, verify that the tree exists and the item has at least
+ *		one reference
+ */
+struct btrfs_root *btrfs_get_fs_root(struct btrfs_fs_info *fs_info,
+				     u64 objectid, bool check_ref)
 {
-	struct btrfs_fs_info *info = (struct btrfs_fs_info *)congested_data;
-	int ret = 0;
-	struct btrfs_device *device;
-	struct backing_dev_info *bdi;
+	return btrfs_get_root_ref(fs_info, objectid, 0, check_ref);
+}
 
-	rcu_read_lock();
-	list_for_each_entry_rcu(device, &info->fs_devices->devices, dev_list) {
-		if (!device->bdev)
-			continue;
-		bdi = device->bdev->bd_bdi;
-		if (bdi_congested(bdi, bdi_bits)) {
-			ret = 1;
-			break;
-		}
-	}
-	rcu_read_unlock();
-	return ret;
+/*
+ * Get in-memory reference of a root structure, created as new, optionally pass
+ * the anonymous block device id
+ *
+ * @objectid:	tree objectid
+ * @anon_dev:	if zero, allocate a new anonymous block device or use the
+ *		parameter value
+ */
+struct btrfs_root *btrfs_get_new_fs_root(struct btrfs_fs_info *fs_info,
+					 u64 objectid, dev_t anon_dev)
+{
+	return btrfs_get_root_ref(fs_info, objectid, anon_dev, true);
 }
 
 /*
- * called by the kthread helper functions to finally call the bio end_io
- * functions.  This is where read checksum verification actually happens
+ * btrfs_get_fs_root_commit_root - return a root for the given objectid
+ * @fs_info:	the fs_info
+ * @objectid:	the objectid we need to lookup
+ *
+ * This is exclusively used for backref walking, and exists specifically because
+ * of how qgroups does lookups.  Qgroups will do a backref lookup at delayed ref
+ * creation time, which means we may have to read the tree_root in order to look
+ * up a fs root that is not in memory.  If the root is not in memory we will
+ * read the tree root commit root and look up the fs root from there.  This is a
+ * temporary root, it will not be inserted into the radix tree as it doesn't
+ * have the most uptodate information, it'll simply be discarded once the
+ * backref code is finished using the root.
  */
-static void end_workqueue_fn(struct btrfs_work *work)
+struct btrfs_root *btrfs_get_fs_root_commit_root(struct btrfs_fs_info *fs_info,
+						 struct btrfs_path *path,
+						 u64 objectid)
 {
-	struct bio *bio;
-	struct btrfs_end_io_wq *end_io_wq;
+	struct btrfs_root *root;
+	struct btrfs_key key;
+
+	ASSERT(path->search_commit_root && path->skip_locking);
+
+	/*
+	 * This can return -ENOENT if we ask for a root that doesn't exist, but
+	 * since this is called via the backref walking code we won't be looking
+	 * up a root that doesn't exist, unless there's corruption.  So if root
+	 * != NULL just return it.
+	 */
+	root = btrfs_get_global_root(fs_info, objectid);
+	if (root)
+		return root;
+
+	root = btrfs_lookup_fs_root(fs_info, objectid);
+	if (root)
+		return root;
 
-	end_io_wq = container_of(work, struct btrfs_end_io_wq, work);
-	bio = end_io_wq->bio;
+	key.objectid = objectid;
+	key.type = BTRFS_ROOT_ITEM_KEY;
+	key.offset = (u64)-1;
+	root = read_tree_root_path(fs_info->tree_root, path, &key);
+	btrfs_release_path(path);
 
-	bio->bi_status = end_io_wq->status;
-	bio->bi_private = end_io_wq->private;
-	bio->bi_end_io = end_io_wq->end_io;
-	bio_endio(bio);
-	kmem_cache_free(btrfs_end_io_wq_cache, end_io_wq);
+	return root;
 }
 
 static int cleaner_kthread(void *arg)
 {
-	struct btrfs_root *root = arg;
-	struct btrfs_fs_info *fs_info = root->fs_info;
+	struct btrfs_fs_info *fs_info = arg;
 	int again;
 
 	while (1) {
@@ -1693,7 +1818,7 @@ static int cleaner_kthread(void *arg)
 
 		btrfs_run_delayed_iputs(fs_info);
 
-		again = btrfs_clean_one_deleted_snapshot(root);
+		again = btrfs_clean_one_deleted_snapshot(fs_info);
 		mutex_unlock(&fs_info->cleaner_mutex);
 
 		/*
@@ -1703,16 +1828,23 @@ static int cleaner_kthread(void *arg)
 		btrfs_run_defrag_inodes(fs_info);
 
 		/*
-		 * Acquires fs_info->delete_unused_bgs_mutex to avoid racing
+		 * Acquires fs_info->reclaim_bgs_lock to avoid racing
 		 * with relocation (btrfs_relocate_chunk) and relocation
 		 * acquires fs_info->cleaner_mutex (btrfs_relocate_block_group)
-		 * after acquiring fs_info->delete_unused_bgs_mutex. So we
+		 * after acquiring fs_info->reclaim_bgs_lock. So we
 		 * can't hold, nor need to, fs_info->cleaner_mutex when deleting
 		 * unused block groups.
 		 */
 		btrfs_delete_unused_bgs(fs_info);
+
+		/*
+		 * Reclaim block groups in the reclaim_bgs list after we deleted
+		 * all unused block_groups. This possibly gives us some more free
+		 * space.
+		 */
+		btrfs_reclaim_bgs(fs_info);
 sleep:
-		clear_bit(BTRFS_FS_CLEANER_RUNNING, &fs_info->flags);
+		clear_and_wake_up_bit(BTRFS_FS_CLEANER_RUNNING, &fs_info->flags);
 		if (kthread_should_park())
 			kthread_parkme();
 		if (kthread_should_stop())
@@ -1732,13 +1864,13 @@ static int transaction_kthread(void *arg)
 	struct btrfs_trans_handle *trans;
 	struct btrfs_transaction *cur;
 	u64 transid;
-	time64_t now;
+	time64_t delta;
 	unsigned long delay;
 	bool cannot_commit;
 
 	do {
 		cannot_commit = false;
-		delay = HZ * fs_info->commit_interval;
+		delay = msecs_to_jiffies(fs_info->commit_interval * 1000);
 		mutex_lock(&fs_info->transaction_kthread_mutex);
 
 		spin_lock(&fs_info->trans_lock);
@@ -1748,13 +1880,14 @@ static int transaction_kthread(void *arg)
 			goto sleep;
 		}
 
-		now = ktime_get_seconds();
-		if (cur->state < TRANS_STATE_COMMIT_START &&
-		    !test_bit(BTRFS_FS_NEED_ASYNC_COMMIT, &fs_info->flags) &&
-		    (now < cur->start_time ||
-		     now - cur->start_time < fs_info->commit_interval)) {
+		delta = ktime_get_seconds() - cur->start_time;
+		if (!test_and_clear_bit(BTRFS_FS_COMMIT_TRANS, &fs_info->flags) &&
+		    cur->state < TRANS_STATE_COMMIT_START &&
+		    delta < fs_info->commit_interval) {
 			spin_unlock(&fs_info->trans_lock);
-			delay = HZ * 5;
+			delay -= msecs_to_jiffies((delta - 1) * 1000);
+			delay = min(delay,
+				    msecs_to_jiffies(fs_info->commit_interval * 1000));
 			goto sleep;
 		}
 		transid = cur->transid;
@@ -1776,8 +1909,7 @@ sleep:
 		wake_up_process(fs_info->cleaner_kthread);
 		mutex_unlock(&fs_info->transaction_kthread_mutex);
 
-		if (unlikely(test_bit(BTRFS_FS_STATE_ERROR,
-				      &fs_info->fs_state)))
+		if (BTRFS_FS_ERROR(fs_info))
 			btrfs_cleanup_transaction(fs_info);
 		if (!kthread_should_stop() &&
 				(!btrfs_transaction_blocked(fs_info) ||
@@ -1846,11 +1978,23 @@ static void backup_super_roots(struct btrfs_fs_info *info)
 	btrfs_set_backup_chunk_root_level(root_backup,
 			       btrfs_header_level(info->chunk_root->node));
 
-	btrfs_set_backup_extent_root(root_backup, info->extent_root->node->start);
-	btrfs_set_backup_extent_root_gen(root_backup,
-			       btrfs_header_generation(info->extent_root->node));
-	btrfs_set_backup_extent_root_level(root_backup,
-			       btrfs_header_level(info->extent_root->node));
+	if (!btrfs_fs_compat_ro(info, BLOCK_GROUP_TREE)) {
+		struct btrfs_root *extent_root = btrfs_extent_root(info, 0);
+		struct btrfs_root *csum_root = btrfs_csum_root(info, 0);
+
+		btrfs_set_backup_extent_root(root_backup,
+					     extent_root->node->start);
+		btrfs_set_backup_extent_root_gen(root_backup,
+				btrfs_header_generation(extent_root->node));
+		btrfs_set_backup_extent_root_level(root_backup,
+					btrfs_header_level(extent_root->node));
+
+		btrfs_set_backup_csum_root(root_backup, csum_root->node->start);
+		btrfs_set_backup_csum_root_gen(root_backup,
+					       btrfs_header_generation(csum_root->node));
+		btrfs_set_backup_csum_root_level(root_backup,
+						 btrfs_header_level(csum_root->node));
+	}
 
 	/*
 	 * we might commit during log recovery, which happens before we set
@@ -1871,12 +2015,6 @@ static void backup_super_roots(struct btrfs_fs_info *info)
 	btrfs_set_backup_dev_root_level(root_backup,
 				       btrfs_header_level(info->dev_root->node));
 
-	btrfs_set_backup_csum_root(root_backup, info->csum_root->node->start);
-	btrfs_set_backup_csum_root_gen(root_backup,
-			       btrfs_header_generation(info->csum_root->node));
-	btrfs_set_backup_csum_root_level(root_backup,
-			       btrfs_header_level(info->csum_root->node));
-
 	btrfs_set_backup_total_bytes(root_backup,
 			     btrfs_super_total_bytes(info->super_copy));
 	btrfs_set_backup_bytes_used(root_backup,
@@ -1942,16 +2080,20 @@ static void btrfs_stop_all_workers(struct btrfs_fs_info *fs_info)
 {
 	btrfs_destroy_workqueue(fs_info->fixup_workers);
 	btrfs_destroy_workqueue(fs_info->delalloc_workers);
+	btrfs_destroy_workqueue(fs_info->hipri_workers);
 	btrfs_destroy_workqueue(fs_info->workers);
-	btrfs_destroy_workqueue(fs_info->endio_workers);
-	btrfs_destroy_workqueue(fs_info->endio_raid56_workers);
-	btrfs_destroy_workqueue(fs_info->endio_repair_workers);
-	btrfs_destroy_workqueue(fs_info->rmw_workers);
+	if (fs_info->endio_workers)
+		destroy_workqueue(fs_info->endio_workers);
+	if (fs_info->endio_raid56_workers)
+		destroy_workqueue(fs_info->endio_raid56_workers);
+	if (fs_info->rmw_workers)
+		destroy_workqueue(fs_info->rmw_workers);
+	if (fs_info->compressed_write_workers)
+		destroy_workqueue(fs_info->compressed_write_workers);
 	btrfs_destroy_workqueue(fs_info->endio_write_workers);
 	btrfs_destroy_workqueue(fs_info->endio_freespace_worker);
 	btrfs_destroy_workqueue(fs_info->delayed_workers);
 	btrfs_destroy_workqueue(fs_info->caching_workers);
-	btrfs_destroy_workqueue(fs_info->readahead_workers);
 	btrfs_destroy_workqueue(fs_info->flush_workers);
 	btrfs_destroy_workqueue(fs_info->qgroup_rescan_workers);
 	if (fs_info->discard_ctl.discard_workers)
@@ -1961,8 +2103,8 @@ static void btrfs_stop_all_workers(struct btrfs_fs_info *fs_info)
 	 * the queues used for metadata I/O, since tasks from those other work
 	 * queues can do metadata I/O operations.
 	 */
-	btrfs_destroy_workqueue(fs_info->endio_meta_workers);
-	btrfs_destroy_workqueue(fs_info->endio_meta_write_workers);
+	if (fs_info->endio_meta_workers)
+		destroy_workqueue(fs_info->endio_meta_workers);
 }
 
 static void free_root_extent_buffers(struct btrfs_root *root)
@@ -1975,19 +2117,51 @@ static void free_root_extent_buffers(struct btrfs_root *root)
 	}
 }
 
+static void free_global_root_pointers(struct btrfs_fs_info *fs_info)
+{
+	struct btrfs_root *root, *tmp;
+
+	rbtree_postorder_for_each_entry_safe(root, tmp,
+					     &fs_info->global_root_tree,
+					     rb_node)
+		free_root_extent_buffers(root);
+}
+
 /* helper to cleanup tree roots */
 static void free_root_pointers(struct btrfs_fs_info *info, bool free_chunk_root)
 {
 	free_root_extent_buffers(info->tree_root);
 
+	free_global_root_pointers(info);
 	free_root_extent_buffers(info->dev_root);
-	free_root_extent_buffers(info->extent_root);
-	free_root_extent_buffers(info->csum_root);
 	free_root_extent_buffers(info->quota_root);
 	free_root_extent_buffers(info->uuid_root);
+	free_root_extent_buffers(info->fs_root);
+	free_root_extent_buffers(info->data_reloc_root);
+	free_root_extent_buffers(info->block_group_root);
 	if (free_chunk_root)
 		free_root_extent_buffers(info->chunk_root);
-	free_root_extent_buffers(info->free_space_root);
+}
+
+void btrfs_put_root(struct btrfs_root *root)
+{
+	if (!root)
+		return;
+
+	if (refcount_dec_and_test(&root->refs)) {
+		WARN_ON(!RB_EMPTY_ROOT(&root->inode_tree));
+		WARN_ON(test_bit(BTRFS_ROOT_DEAD_RELOC_TREE, &root->state));
+		if (root->anon_dev)
+			free_anon_bdev(root->anon_dev);
+		btrfs_drew_lock_destroy(&root->snapshot_lock);
+		free_root_extent_buffers(root);
+#ifdef CONFIG_BTRFS_DEBUG
+		spin_lock(&root->fs_info->fs_roots_radix_lock);
+		list_del_init(&root->leak_list);
+		spin_unlock(&root->fs_info->fs_roots_radix_lock);
+#endif
+		kfree(root);
+	}
 }
 
 void btrfs_free_fs_roots(struct btrfs_fs_info *fs_info)
@@ -2001,13 +2175,9 @@ void btrfs_free_fs_roots(struct btrfs_fs_info *fs_info)
 				     struct btrfs_root, root_list);
 		list_del(&gang[0]->root_list);
 
-		if (test_bit(BTRFS_ROOT_IN_RADIX, &gang[0]->state)) {
+		if (test_bit(BTRFS_ROOT_IN_RADIX, &gang[0]->state))
 			btrfs_drop_and_free_fs_root(fs_info, gang[0]);
-		} else {
-			free_extent_buffer(gang[0]->node);
-			free_extent_buffer(gang[0]->commit_root);
-			btrfs_put_fs_root(gang[0]);
-		}
+		btrfs_put_root(gang[0]);
 	}
 
 	while (1) {
@@ -2019,11 +2189,6 @@ void btrfs_free_fs_roots(struct btrfs_fs_info *fs_info)
 		for (i = 0; i < ret; i++)
 			btrfs_drop_and_free_fs_root(fs_info, gang[i]);
 	}
-
-	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
-		btrfs_free_log_root_tree(NULL, fs_info);
-		btrfs_destroy_pinned_extent(fs_info, fs_info->pinned_extents);
-	}
 }
 
 static void btrfs_init_scrub(struct btrfs_fs_info *fs_info)
@@ -2045,11 +2210,14 @@ static void btrfs_init_balance(struct btrfs_fs_info *fs_info)
 	atomic_set(&fs_info->balance_cancel_req, 0);
 	fs_info->balance_ctl = NULL;
 	init_waitqueue_head(&fs_info->balance_wait_q);
+	atomic_set(&fs_info->reloc_cancel_req, 0);
 }
 
 static void btrfs_init_btree_inode(struct btrfs_fs_info *fs_info)
 {
 	struct inode *inode = fs_info->btree_inode;
+	unsigned long hash = btrfs_inode_hash(BTRFS_BTREE_INODE_OBJECTID,
+					      fs_info->tree_root);
 
 	inode->i_ino = BTRFS_BTREE_INODE_OBJECTID;
 	set_nlink(inode, 1);
@@ -2063,16 +2231,15 @@ static void btrfs_init_btree_inode(struct btrfs_fs_info *fs_info)
 
 	RB_CLEAR_NODE(&BTRFS_I(inode)->rb_node);
 	extent_io_tree_init(fs_info, &BTRFS_I(inode)->io_tree,
-			    IO_TREE_INODE_IO, inode);
-	BTRFS_I(inode)->io_tree.track_uptodate = false;
+			    IO_TREE_BTREE_INODE_IO, NULL);
 	extent_map_tree_init(&BTRFS_I(inode)->extent_tree);
 
-	BTRFS_I(inode)->io_tree.ops = &btree_extent_io_ops;
-
-	BTRFS_I(inode)->root = fs_info->tree_root;
-	memset(&BTRFS_I(inode)->location, 0, sizeof(struct btrfs_key));
+	BTRFS_I(inode)->root = btrfs_grab_root(fs_info->tree_root);
+	BTRFS_I(inode)->location.objectid = BTRFS_BTREE_INODE_OBJECTID;
+	BTRFS_I(inode)->location.type = 0;
+	BTRFS_I(inode)->location.offset = 0;
 	set_bit(BTRFS_INODE_DUMMY, &BTRFS_I(inode)->runtime_flags);
-	btrfs_insert_inode_hash(inode);
+	__insert_inode_hash(inode, hash);
 }
 
 static void btrfs_init_dev_replace_locks(struct btrfs_fs_info *fs_info)
@@ -2091,17 +2258,19 @@ static void btrfs_init_qgroup(struct btrfs_fs_info *fs_info)
 	fs_info->qgroup_seq = 1;
 	fs_info->qgroup_ulist = NULL;
 	fs_info->qgroup_rescan_running = false;
+	fs_info->qgroup_drop_subtree_thres = BTRFS_MAX_LEVEL;
 	mutex_init(&fs_info->qgroup_rescan_lock);
 }
 
-static int btrfs_init_workqueues(struct btrfs_fs_info *fs_info,
-		struct btrfs_fs_devices *fs_devices)
+static int btrfs_init_workqueues(struct btrfs_fs_info *fs_info)
 {
 	u32 max_active = fs_info->thread_pool_size;
 	unsigned int flags = WQ_MEM_RECLAIM | WQ_FREEZABLE | WQ_UNBOUND;
 
 	fs_info->workers =
-		btrfs_alloc_workqueue(fs_info, "worker",
+		btrfs_alloc_workqueue(fs_info, "worker", flags, max_active, 16);
+	fs_info->hipri_workers =
+		btrfs_alloc_workqueue(fs_info, "worker-high",
 				      flags | WQ_HIGHPRI, max_active, 16);
 
 	fs_info->delalloc_workers =
@@ -2118,52 +2287,37 @@ static int btrfs_init_workqueues(struct btrfs_fs_info *fs_info,
 	fs_info->fixup_workers =
 		btrfs_alloc_workqueue(fs_info, "fixup", flags, 1, 0);
 
-	/*
-	 * endios are largely parallel and should have a very
-	 * low idle thresh
-	 */
 	fs_info->endio_workers =
-		btrfs_alloc_workqueue(fs_info, "endio", flags, max_active, 4);
+		alloc_workqueue("btrfs-endio", flags, max_active);
 	fs_info->endio_meta_workers =
-		btrfs_alloc_workqueue(fs_info, "endio-meta", flags,
-				      max_active, 4);
-	fs_info->endio_meta_write_workers =
-		btrfs_alloc_workqueue(fs_info, "endio-meta-write", flags,
-				      max_active, 2);
+		alloc_workqueue("btrfs-endio-meta", flags, max_active);
 	fs_info->endio_raid56_workers =
-		btrfs_alloc_workqueue(fs_info, "endio-raid56", flags,
-				      max_active, 4);
-	fs_info->endio_repair_workers =
-		btrfs_alloc_workqueue(fs_info, "endio-repair", flags, 1, 0);
-	fs_info->rmw_workers =
-		btrfs_alloc_workqueue(fs_info, "rmw", flags, max_active, 2);
+		alloc_workqueue("btrfs-endio-raid56", flags, max_active);
+	fs_info->rmw_workers = alloc_workqueue("btrfs-rmw", flags, max_active);
 	fs_info->endio_write_workers =
 		btrfs_alloc_workqueue(fs_info, "endio-write", flags,
 				      max_active, 2);
+	fs_info->compressed_write_workers =
+		alloc_workqueue("btrfs-compressed-write", flags, max_active);
 	fs_info->endio_freespace_worker =
 		btrfs_alloc_workqueue(fs_info, "freespace-write", flags,
 				      max_active, 0);
 	fs_info->delayed_workers =
 		btrfs_alloc_workqueue(fs_info, "delayed-meta", flags,
 				      max_active, 0);
-	fs_info->readahead_workers =
-		btrfs_alloc_workqueue(fs_info, "readahead", flags,
-				      max_active, 2);
 	fs_info->qgroup_rescan_workers =
 		btrfs_alloc_workqueue(fs_info, "qgroup-rescan", flags, 1, 0);
 	fs_info->discard_ctl.discard_workers =
 		alloc_workqueue("btrfs_discard", WQ_UNBOUND | WQ_FREEZABLE, 1);
 
-	if (!(fs_info->workers && fs_info->delalloc_workers &&
-	      fs_info->flush_workers &&
+	if (!(fs_info->workers && fs_info->hipri_workers &&
+	      fs_info->delalloc_workers && fs_info->flush_workers &&
 	      fs_info->endio_workers && fs_info->endio_meta_workers &&
-	      fs_info->endio_meta_write_workers &&
-	      fs_info->endio_repair_workers &&
+	      fs_info->compressed_write_workers &&
 	      fs_info->endio_write_workers && fs_info->endio_raid56_workers &&
 	      fs_info->endio_freespace_worker && fs_info->rmw_workers &&
-	      fs_info->caching_workers && fs_info->readahead_workers &&
-	      fs_info->fixup_workers && fs_info->delayed_workers &&
-	      fs_info->qgroup_rescan_workers &&
+	      fs_info->caching_workers && fs_info->fixup_workers &&
+	      fs_info->delayed_workers && fs_info->qgroup_rescan_workers &&
 	      fs_info->discard_ctl.discard_workers)) {
 		return -ENOMEM;
 	}
@@ -2186,14 +2340,12 @@ static int btrfs_init_csum_hash(struct btrfs_fs_info *fs_info, u16 csum_type)
 
 	fs_info->csum_shash = csum_shash;
 
+	btrfs_info(fs_info, "using %s (%s) checksum algorithm",
+			btrfs_super_csum_name(csum_type),
+			crypto_shash_driver_name(csum_shash));
 	return 0;
 }
 
-static void btrfs_free_csum_hash(struct btrfs_fs_info *fs_info)
-{
-	crypto_free_shash(fs_info->csum_shash);
-}
-
 static int btrfs_replay_log(struct btrfs_fs_info *fs_info,
 			    struct btrfs_fs_devices *fs_devices)
 {
@@ -2208,33 +2360,34 @@ static int btrfs_replay_log(struct btrfs_fs_info *fs_info,
 		return -EIO;
 	}
 
-	log_tree_root = btrfs_alloc_root(fs_info, GFP_KERNEL);
+	log_tree_root = btrfs_alloc_root(fs_info, BTRFS_TREE_LOG_OBJECTID,
+					 GFP_KERNEL);
 	if (!log_tree_root)
 		return -ENOMEM;
 
-	__setup_root(log_tree_root, fs_info, BTRFS_TREE_LOG_OBJECTID);
-
 	log_tree_root->node = read_tree_block(fs_info, bytenr,
-					      fs_info->generation + 1,
-					      level, NULL);
+					      BTRFS_TREE_LOG_OBJECTID,
+					      fs_info->generation + 1, level,
+					      NULL);
 	if (IS_ERR(log_tree_root->node)) {
 		btrfs_warn(fs_info, "failed to read log tree");
 		ret = PTR_ERR(log_tree_root->node);
-		kfree(log_tree_root);
+		log_tree_root->node = NULL;
+		btrfs_put_root(log_tree_root);
 		return ret;
-	} else if (!extent_buffer_uptodate(log_tree_root->node)) {
+	}
+	if (!extent_buffer_uptodate(log_tree_root->node)) {
 		btrfs_err(fs_info, "failed to read log tree");
-		free_extent_buffer(log_tree_root->node);
-		kfree(log_tree_root);
+		btrfs_put_root(log_tree_root);
 		return -EIO;
 	}
+
 	/* returns with log_tree_root freed on success */
 	ret = btrfs_recover_log_trees(log_tree_root);
 	if (ret) {
 		btrfs_handle_fs_error(fs_info, ret,
 				      "Failed to recover log tree");
-		free_extent_buffer(log_tree_root->node);
-		kfree(log_tree_root);
+		btrfs_put_root(log_tree_root);
 		return ret;
 	}
 
@@ -2247,6 +2400,115 @@ static int btrfs_replay_log(struct btrfs_fs_info *fs_info,
 	return 0;
 }
 
+static int load_global_roots_objectid(struct btrfs_root *tree_root,
+				      struct btrfs_path *path, u64 objectid,
+				      const char *name)
+{
+	struct btrfs_fs_info *fs_info = tree_root->fs_info;
+	struct btrfs_root *root;
+	u64 max_global_id = 0;
+	int ret;
+	struct btrfs_key key = {
+		.objectid = objectid,
+		.type = BTRFS_ROOT_ITEM_KEY,
+		.offset = 0,
+	};
+	bool found = false;
+
+	/* If we have IGNOREDATACSUMS skip loading these roots. */
+	if (objectid == BTRFS_CSUM_TREE_OBJECTID &&
+	    btrfs_test_opt(fs_info, IGNOREDATACSUMS)) {
+		set_bit(BTRFS_FS_STATE_NO_CSUMS, &fs_info->fs_state);
+		return 0;
+	}
+
+	while (1) {
+		ret = btrfs_search_slot(NULL, tree_root, &key, path, 0, 0);
+		if (ret < 0)
+			break;
+
+		if (path->slots[0] >= btrfs_header_nritems(path->nodes[0])) {
+			ret = btrfs_next_leaf(tree_root, path);
+			if (ret) {
+				if (ret > 0)
+					ret = 0;
+				break;
+			}
+		}
+		ret = 0;
+
+		btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
+		if (key.objectid != objectid)
+			break;
+		btrfs_release_path(path);
+
+		/*
+		 * Just worry about this for extent tree, it'll be the same for
+		 * everybody.
+		 */
+		if (objectid == BTRFS_EXTENT_TREE_OBJECTID)
+			max_global_id = max(max_global_id, key.offset);
+
+		found = true;
+		root = read_tree_root_path(tree_root, path, &key);
+		if (IS_ERR(root)) {
+			if (!btrfs_test_opt(fs_info, IGNOREBADROOTS))
+				ret = PTR_ERR(root);
+			break;
+		}
+		set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
+		ret = btrfs_global_root_insert(root);
+		if (ret) {
+			btrfs_put_root(root);
+			break;
+		}
+		key.offset++;
+	}
+	btrfs_release_path(path);
+
+	if (objectid == BTRFS_EXTENT_TREE_OBJECTID)
+		fs_info->nr_global_roots = max_global_id + 1;
+
+	if (!found || ret) {
+		if (objectid == BTRFS_CSUM_TREE_OBJECTID)
+			set_bit(BTRFS_FS_STATE_NO_CSUMS, &fs_info->fs_state);
+
+		if (!btrfs_test_opt(fs_info, IGNOREBADROOTS))
+			ret = ret ? ret : -ENOENT;
+		else
+			ret = 0;
+		btrfs_err(fs_info, "failed to load root %s", name);
+	}
+	return ret;
+}
+
+static int load_global_roots(struct btrfs_root *tree_root)
+{
+	struct btrfs_path *path;
+	int ret = 0;
+
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+
+	ret = load_global_roots_objectid(tree_root, path,
+					 BTRFS_EXTENT_TREE_OBJECTID, "extent");
+	if (ret)
+		goto out;
+	ret = load_global_roots_objectid(tree_root, path,
+					 BTRFS_CSUM_TREE_OBJECTID, "csum");
+	if (ret)
+		goto out;
+	if (!btrfs_fs_compat_ro(tree_root->fs_info, FREE_SPACE_TREE))
+		goto out;
+	ret = load_global_roots_objectid(tree_root, path,
+					 BTRFS_FREE_SPACE_TREE_OBJECTID,
+					 "free space");
+out:
+	btrfs_free_path(path);
+	return ret;
+}
+
 static int btrfs_read_roots(struct btrfs_fs_info *fs_info)
 {
 	struct btrfs_root *tree_root = fs_info->tree_root;
@@ -2256,36 +2518,58 @@ static int btrfs_read_roots(struct btrfs_fs_info *fs_info)
 
 	BUG_ON(!fs_info->tree_root);
 
-	location.objectid = BTRFS_EXTENT_TREE_OBJECTID;
+	ret = load_global_roots(tree_root);
+	if (ret)
+		return ret;
+
 	location.type = BTRFS_ROOT_ITEM_KEY;
 	location.offset = 0;
 
-	root = btrfs_read_tree_root(tree_root, &location);
-	if (IS_ERR(root)) {
-		ret = PTR_ERR(root);
-		goto out;
+	if (btrfs_fs_compat_ro(fs_info, BLOCK_GROUP_TREE)) {
+		location.objectid = BTRFS_BLOCK_GROUP_TREE_OBJECTID;
+		root = btrfs_read_tree_root(tree_root, &location);
+		if (IS_ERR(root)) {
+			if (!btrfs_test_opt(fs_info, IGNOREBADROOTS)) {
+				ret = PTR_ERR(root);
+				goto out;
+			}
+		} else {
+			set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
+			fs_info->block_group_root = root;
+		}
 	}
-	set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
-	fs_info->extent_root = root;
 
 	location.objectid = BTRFS_DEV_TREE_OBJECTID;
 	root = btrfs_read_tree_root(tree_root, &location);
 	if (IS_ERR(root)) {
-		ret = PTR_ERR(root);
-		goto out;
+		if (!btrfs_test_opt(fs_info, IGNOREBADROOTS)) {
+			ret = PTR_ERR(root);
+			goto out;
+		}
+	} else {
+		set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
+		fs_info->dev_root = root;
 	}
-	set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
-	fs_info->dev_root = root;
-	btrfs_init_devices_late(fs_info);
+	/* Initialize fs_info for all devices in any case */
+	ret = btrfs_init_devices_late(fs_info);
+	if (ret)
+		goto out;
 
-	location.objectid = BTRFS_CSUM_TREE_OBJECTID;
-	root = btrfs_read_tree_root(tree_root, &location);
+	/*
+	 * This tree can share blocks with some other fs tree during relocation
+	 * and we need a proper setup by btrfs_get_fs_root
+	 */
+	root = btrfs_get_fs_root(tree_root->fs_info,
+				 BTRFS_DATA_RELOC_TREE_OBJECTID, true);
 	if (IS_ERR(root)) {
-		ret = PTR_ERR(root);
-		goto out;
+		if (!btrfs_test_opt(fs_info, IGNOREBADROOTS)) {
+			ret = PTR_ERR(root);
+			goto out;
+		}
+	} else {
+		set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
+		fs_info->data_reloc_root = root;
 	}
-	set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
-	fs_info->csum_root = root;
 
 	location.objectid = BTRFS_QUOTA_TREE_OBJECTID;
 	root = btrfs_read_tree_root(tree_root, &location);
@@ -2298,23 +2582,14 @@ static int btrfs_read_roots(struct btrfs_fs_info *fs_info)
 	location.objectid = BTRFS_UUID_TREE_OBJECTID;
 	root = btrfs_read_tree_root(tree_root, &location);
 	if (IS_ERR(root)) {
-		ret = PTR_ERR(root);
-		if (ret != -ENOENT)
-			goto out;
-	} else {
-		set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
-		fs_info->uuid_root = root;
-	}
-
-	if (btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE)) {
-		location.objectid = BTRFS_FREE_SPACE_TREE_OBJECTID;
-		root = btrfs_read_tree_root(tree_root, &location);
-		if (IS_ERR(root)) {
+		if (!btrfs_test_opt(fs_info, IGNOREBADROOTS)) {
 			ret = PTR_ERR(root);
-			goto out;
+			if (ret != -ENOENT)
+				goto out;
 		}
+	} else {
 		set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
-		fs_info->free_space_root = root;
+		fs_info->uuid_root = root;
 	}
 
 	return 0;
@@ -2334,8 +2609,8 @@ out:
  * 		1, 2	2nd and 3rd backup copy
  * 	       -1	skip bytenr check
  */
-static int validate_super(struct btrfs_fs_info *fs_info,
-			    struct btrfs_super_block *sb, int mirror_num)
+int btrfs_validate_super(struct btrfs_fs_info *fs_info,
+			 struct btrfs_super_block *sb, int mirror_num)
 {
 	u64 nodesize = btrfs_super_nodesize(sb);
 	u64 sectorsize = btrfs_super_sectorsize(sb);
@@ -2375,13 +2650,22 @@ static int validate_super(struct btrfs_fs_info *fs_info,
 		btrfs_err(fs_info, "invalid sectorsize %llu", sectorsize);
 		ret = -EINVAL;
 	}
-	/* Only PAGE SIZE is supported yet */
-	if (sectorsize != PAGE_SIZE) {
+
+	/*
+	 * We only support at most two sectorsizes: 4K and PAGE_SIZE.
+	 *
+	 * We can support 16K sectorsize with 64K page size without problem,
+	 * but such sectorsize/pagesize combination doesn't make much sense.
+	 * 4K will be our future standard, PAGE_SIZE is supported from the very
+	 * beginning.
+	 */
+	if (sectorsize > PAGE_SIZE || (sectorsize != SZ_4K && sectorsize != PAGE_SIZE)) {
 		btrfs_err(fs_info,
-			"sectorsize %llu not supported yet, only support %lu",
+			"sectorsize %llu not yet supported for page size %lu",
 			sectorsize, PAGE_SIZE);
 		ret = -EINVAL;
 	}
+
 	if (!is_power_of_2(nodesize) || nodesize < sectorsize ||
 	    nodesize > BTRFS_MAX_METADATA_BLOCKSIZE) {
 		btrfs_err(fs_info, "invalid nodesize %llu", nodesize);
@@ -2410,6 +2694,36 @@ static int validate_super(struct btrfs_fs_info *fs_info,
 		ret = -EINVAL;
 	}
 
+	if (memcmp(fs_info->fs_devices->fsid, fs_info->super_copy->fsid,
+		   BTRFS_FSID_SIZE)) {
+		btrfs_err(fs_info,
+		"superblock fsid doesn't match fsid of fs_devices: %pU != %pU",
+			fs_info->super_copy->fsid, fs_info->fs_devices->fsid);
+		ret = -EINVAL;
+	}
+
+	if (btrfs_fs_incompat(fs_info, METADATA_UUID) &&
+	    memcmp(fs_info->fs_devices->metadata_uuid,
+		   fs_info->super_copy->metadata_uuid, BTRFS_FSID_SIZE)) {
+		btrfs_err(fs_info,
+"superblock metadata_uuid doesn't match metadata uuid of fs_devices: %pU != %pU",
+			fs_info->super_copy->metadata_uuid,
+			fs_info->fs_devices->metadata_uuid);
+		ret = -EINVAL;
+	}
+
+	/*
+	 * Artificial requirement for block-group-tree to force newer features
+	 * (free-space-tree, no-holes) so the test matrix is smaller.
+	 */
+	if (btrfs_fs_compat_ro(fs_info, BLOCK_GROUP_TREE) &&
+	    (!btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE_VALID) ||
+	     !btrfs_fs_incompat(fs_info, NO_HOLES))) {
+		btrfs_err(fs_info,
+		"block-group-tree feature requires fres-space-tree and no-holes");
+		ret = -EINVAL;
+	}
+
 	if (memcmp(fs_info->fs_devices->metadata_uuid, sb->dev_item.fsid,
 		   BTRFS_FSID_SIZE) != 0) {
 		btrfs_err(fs_info,
@@ -2492,7 +2806,7 @@ static int validate_super(struct btrfs_fs_info *fs_info,
  */
 static int btrfs_validate_mount_super(struct btrfs_fs_info *fs_info)
 {
-	return validate_super(fs_info, fs_info->super_copy, 0);
+	return btrfs_validate_super(fs_info, fs_info->super_copy, 0);
 }
 
 /*
@@ -2506,7 +2820,7 @@ static int btrfs_validate_write_super(struct btrfs_fs_info *fs_info,
 {
 	int ret;
 
-	ret = validate_super(fs_info, sb, -1);
+	ret = btrfs_validate_super(fs_info, sb, -1);
 	if (ret < 0)
 		goto out;
 	if (!btrfs_supported_super_csum(btrfs_super_csum_type(sb))) {
@@ -2530,6 +2844,46 @@ out:
 	return ret;
 }
 
+static int load_super_root(struct btrfs_root *root, u64 bytenr, u64 gen, int level)
+{
+	int ret = 0;
+
+	root->node = read_tree_block(root->fs_info, bytenr,
+				     root->root_key.objectid, gen, level, NULL);
+	if (IS_ERR(root->node)) {
+		ret = PTR_ERR(root->node);
+		root->node = NULL;
+		return ret;
+	}
+	if (!extent_buffer_uptodate(root->node)) {
+		free_extent_buffer(root->node);
+		root->node = NULL;
+		return -EIO;
+	}
+
+	btrfs_set_root_node(&root->root_item, root->node);
+	root->commit_root = btrfs_root_node(root);
+	btrfs_set_root_refs(&root->root_item, 1);
+	return ret;
+}
+
+static int load_important_roots(struct btrfs_fs_info *fs_info)
+{
+	struct btrfs_super_block *sb = fs_info->super_copy;
+	u64 gen, bytenr;
+	int level, ret;
+
+	bytenr = btrfs_super_root(sb);
+	gen = btrfs_super_generation(sb);
+	level = btrfs_super_root_level(sb);
+	ret = load_super_root(fs_info->tree_root, bytenr, gen, level);
+	if (ret) {
+		btrfs_warn(fs_info, "couldn't read tree root");
+		return ret;
+	}
+	return 0;
+}
+
 static int __cold init_tree_roots(struct btrfs_fs_info *fs_info)
 {
 	int backup_index = find_newest_super_backup(fs_info);
@@ -2540,9 +2894,6 @@ static int __cold init_tree_roots(struct btrfs_fs_info *fs_info)
 	int i;
 
 	for (i = 0; i < BTRFS_NUM_BACKUP_ROOTS; i++) {
-		u64 generation;
-		int level;
-
 		if (handle_error) {
 			if (!IS_ERR(tree_root->node))
 				free_extent_buffer(tree_root->node);
@@ -2567,39 +2918,24 @@ static int __cold init_tree_roots(struct btrfs_fs_info *fs_info)
 			if (ret < 0)
 				return ret;
 		}
-		generation = btrfs_super_generation(sb);
-		level = btrfs_super_root_level(sb);
-		tree_root->node = read_tree_block(fs_info, btrfs_super_root(sb),
-						  generation, level, NULL);
-		if (IS_ERR(tree_root->node) ||
-		    !extent_buffer_uptodate(tree_root->node)) {
-			handle_error = true;
 
-			if (IS_ERR(tree_root->node))
-				ret = PTR_ERR(tree_root->node);
-			else if (!extent_buffer_uptodate(tree_root->node))
-				ret = -EUCLEAN;
-
-			btrfs_warn(fs_info, "failed to read tree root");
+		ret = load_important_roots(fs_info);
+		if (ret) {
+			handle_error = true;
 			continue;
 		}
 
-		btrfs_set_root_node(&tree_root->root_item, tree_root->node);
-		tree_root->commit_root = btrfs_root_node(tree_root);
-		btrfs_set_root_refs(&tree_root->root_item, 1);
-
 		/*
 		 * No need to hold btrfs_root::objectid_mutex since the fs
 		 * hasn't been fully initialised and we are the only user
 		 */
-		ret = btrfs_find_highest_objectid(tree_root,
-						&tree_root->highest_objectid);
+		ret = btrfs_init_root_free_objectid(tree_root);
 		if (ret < 0) {
 			handle_error = true;
 			continue;
 		}
 
-		ASSERT(tree_root->highest_objectid <= BTRFS_LAST_FREE_OBJECTID);
+		ASSERT(tree_root->free_objectid <= BTRFS_LAST_FREE_OBJECTID);
 
 		ret = btrfs_read_roots(fs_info);
 		if (ret < 0) {
@@ -2608,8 +2944,9 @@ static int __cold init_tree_roots(struct btrfs_fs_info *fs_info)
 		}
 
 		/* All successful */
-		fs_info->generation = generation;
-		fs_info->last_trans_committed = generation;
+		fs_info->generation = btrfs_header_generation(tree_root->node);
+		fs_info->last_trans_committed = fs_info->generation;
+		fs_info->last_reloc_trans = 0;
 
 		/* Always begin writing backup roots after the one being used */
 		if (backup_index < 0) {
@@ -2624,67 +2961,8 @@ static int __cold init_tree_roots(struct btrfs_fs_info *fs_info)
 	return ret;
 }
 
-int __cold open_ctree(struct super_block *sb,
-	       struct btrfs_fs_devices *fs_devices,
-	       char *options)
+void btrfs_init_fs_info(struct btrfs_fs_info *fs_info)
 {
-	u32 sectorsize;
-	u32 nodesize;
-	u32 stripesize;
-	u64 generation;
-	u64 features;
-	u16 csum_type;
-	struct btrfs_key location;
-	struct buffer_head *bh;
-	struct btrfs_super_block *disk_super;
-	struct btrfs_fs_info *fs_info = btrfs_sb(sb);
-	struct btrfs_root *tree_root;
-	struct btrfs_root *chunk_root;
-	int ret;
-	int err = -EINVAL;
-	int clear_free_space_tree = 0;
-	int level;
-
-	tree_root = fs_info->tree_root = btrfs_alloc_root(fs_info, GFP_KERNEL);
-	chunk_root = fs_info->chunk_root = btrfs_alloc_root(fs_info, GFP_KERNEL);
-	if (!tree_root || !chunk_root) {
-		err = -ENOMEM;
-		goto fail;
-	}
-
-	ret = init_srcu_struct(&fs_info->subvol_srcu);
-	if (ret) {
-		err = ret;
-		goto fail;
-	}
-
-	ret = percpu_counter_init(&fs_info->dio_bytes, 0, GFP_KERNEL);
-	if (ret) {
-		err = ret;
-		goto fail_srcu;
-	}
-
-	ret = percpu_counter_init(&fs_info->dirty_metadata_bytes, 0, GFP_KERNEL);
-	if (ret) {
-		err = ret;
-		goto fail_dio_bytes;
-	}
-	fs_info->dirty_metadata_batch = PAGE_SIZE *
-					(1 + ilog2(nr_cpu_ids));
-
-	ret = percpu_counter_init(&fs_info->delalloc_bytes, 0, GFP_KERNEL);
-	if (ret) {
-		err = ret;
-		goto fail_dirty_metadata_bytes;
-	}
-
-	ret = percpu_counter_init(&fs_info->dev_replace.bio_counter, 0,
-			GFP_KERNEL);
-	if (ret) {
-		err = ret;
-		goto fail_delalloc_bytes;
-	}
-
 	INIT_RADIX_TREE(&fs_info->fs_roots_radix, GFP_ATOMIC);
 	INIT_RADIX_TREE(&fs_info->buffer_radix, GFP_ATOMIC);
 	INIT_LIST_HEAD(&fs_info->trans_list);
@@ -2700,17 +2978,43 @@ int __cold open_ctree(struct super_block *sb,
 	spin_lock_init(&fs_info->super_lock);
 	spin_lock_init(&fs_info->buffer_lock);
 	spin_lock_init(&fs_info->unused_bgs_lock);
+	spin_lock_init(&fs_info->treelog_bg_lock);
+	spin_lock_init(&fs_info->zone_active_bgs_lock);
+	spin_lock_init(&fs_info->relocation_bg_lock);
 	rwlock_init(&fs_info->tree_mod_log_lock);
+	rwlock_init(&fs_info->global_root_lock);
 	mutex_init(&fs_info->unused_bg_unpin_mutex);
-	mutex_init(&fs_info->delete_unused_bgs_mutex);
+	mutex_init(&fs_info->reclaim_bgs_lock);
 	mutex_init(&fs_info->reloc_mutex);
 	mutex_init(&fs_info->delalloc_root_mutex);
+	mutex_init(&fs_info->zoned_meta_io_lock);
+	mutex_init(&fs_info->zoned_data_reloc_io_lock);
 	seqlock_init(&fs_info->profiles_lock);
 
+	btrfs_lockdep_init_map(fs_info, btrfs_trans_num_writers);
+	btrfs_lockdep_init_map(fs_info, btrfs_trans_num_extwriters);
+	btrfs_lockdep_init_map(fs_info, btrfs_trans_pending_ordered);
+	btrfs_lockdep_init_map(fs_info, btrfs_ordered_extent);
+	btrfs_state_lockdep_init_map(fs_info, btrfs_trans_commit_start,
+				     BTRFS_LOCKDEP_TRANS_COMMIT_START);
+	btrfs_state_lockdep_init_map(fs_info, btrfs_trans_unblocked,
+				     BTRFS_LOCKDEP_TRANS_UNBLOCKED);
+	btrfs_state_lockdep_init_map(fs_info, btrfs_trans_super_committed,
+				     BTRFS_LOCKDEP_TRANS_SUPER_COMMITTED);
+	btrfs_state_lockdep_init_map(fs_info, btrfs_trans_completed,
+				     BTRFS_LOCKDEP_TRANS_COMPLETED);
+
 	INIT_LIST_HEAD(&fs_info->dirty_cowonly_roots);
 	INIT_LIST_HEAD(&fs_info->space_info);
 	INIT_LIST_HEAD(&fs_info->tree_mod_seq_list);
 	INIT_LIST_HEAD(&fs_info->unused_bgs);
+	INIT_LIST_HEAD(&fs_info->reclaim_bgs);
+	INIT_LIST_HEAD(&fs_info->zone_active_bgs);
+#ifdef CONFIG_BTRFS_DEBUG
+	INIT_LIST_HEAD(&fs_info->allocated_roots);
+	INIT_LIST_HEAD(&fs_info->allocated_ebs);
+	spin_lock_init(&fs_info->eb_leak_lock);
+#endif
 	extent_map_tree_init(&fs_info->mapping_tree);
 	btrfs_init_block_rsv(&fs_info->global_block_rsv,
 			     BTRFS_BLOCK_RSV_GLOBAL);
@@ -2724,10 +3028,9 @@ int __cold open_ctree(struct super_block *sb,
 
 	atomic_set(&fs_info->async_delalloc_pages, 0);
 	atomic_set(&fs_info->defrag_running, 0);
-	atomic_set(&fs_info->reada_works_cnt, 0);
 	atomic_set(&fs_info->nr_delayed_iputs, 0);
 	atomic64_set(&fs_info->tree_mod_seq, 0);
-	fs_info->sb = sb;
+	fs_info->global_root_tree = RB_ROOT;
 	fs_info->max_inline = BTRFS_DEFAULT_MAX_INLINE;
 	fs_info->metadata_ratio = 0;
 	fs_info->defrag_inodes = RB_ROOT;
@@ -2735,9 +3038,6 @@ int __cold open_ctree(struct super_block *sb,
 	fs_info->tree_mod_log = RB_ROOT;
 	fs_info->commit_interval = BTRFS_DEFAULT_COMMIT_INTERVAL;
 	fs_info->avg_delayed_ref_runtime = NSEC_PER_SEC >> 6; /* div by 64 */
-	/* readahead state */
-	INIT_RADIX_TREE(&fs_info->reada_tree, GFP_NOFS & ~__GFP_DIRECT_RECLAIM);
-	spin_lock_init(&fs_info->reada_lock);
 	btrfs_init_ref_verify(fs_info);
 
 	fs_info->thread_pool_size = min_t(unsigned long,
@@ -2746,43 +3046,18 @@ int __cold open_ctree(struct super_block *sb,
 	INIT_LIST_HEAD(&fs_info->ordered_roots);
 	spin_lock_init(&fs_info->ordered_root_lock);
 
-	fs_info->btree_inode = new_inode(sb);
-	if (!fs_info->btree_inode) {
-		err = -ENOMEM;
-		goto fail_bio_counter;
-	}
-	mapping_set_gfp_mask(fs_info->btree_inode->i_mapping, GFP_NOFS);
-
-	fs_info->delayed_root = kmalloc(sizeof(struct btrfs_delayed_root),
-					GFP_KERNEL);
-	if (!fs_info->delayed_root) {
-		err = -ENOMEM;
-		goto fail_iput;
-	}
-	btrfs_init_delayed_root(fs_info->delayed_root);
-
 	btrfs_init_scrub(fs_info);
 #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
 	fs_info->check_integrity_print_mask = 0;
 #endif
 	btrfs_init_balance(fs_info);
-	btrfs_init_async_reclaim_work(&fs_info->async_reclaim_work);
-
-	sb->s_blocksize = BTRFS_BDEV_BLOCKSIZE;
-	sb->s_blocksize_bits = blksize_bits(BTRFS_BDEV_BLOCKSIZE);
-
-	btrfs_init_btree_inode(fs_info);
+	btrfs_init_async_reclaim_work(fs_info);
 
-	spin_lock_init(&fs_info->block_group_cache_lock);
-	fs_info->block_group_cache_tree = RB_ROOT;
-	fs_info->first_logical_byte = (u64)-1;
+	rwlock_init(&fs_info->block_group_cache_lock);
+	fs_info->block_group_cache_tree = RB_ROOT_CACHED;
 
-	extent_io_tree_init(fs_info, &fs_info->freed_extents[0],
-			    IO_TREE_FS_INFO_FREED_EXTENTS0, NULL);
-	extent_io_tree_init(fs_info, &fs_info->freed_extents[1],
-			    IO_TREE_FS_INFO_FREED_EXTENTS1, NULL);
-	fs_info->pinned_extents = &fs_info->freed_extents[0];
-	set_bit(BTRFS_FS_BARRIER, &fs_info->flags);
+	extent_io_tree_init(fs_info, &fs_info->excluded_extents,
+			    IO_TREE_FS_EXCLUDED_EXTENTS, NULL);
 
 	mutex_init(&fs_info->ordered_operations_mutex);
 	mutex_init(&fs_info->tree_log_mutex);
@@ -2811,48 +3086,392 @@ int __cold open_ctree(struct super_block *sb,
 	/* Usable values until the real ones are cached from the superblock */
 	fs_info->nodesize = 4096;
 	fs_info->sectorsize = 4096;
+	fs_info->sectorsize_bits = ilog2(4096);
 	fs_info->stripesize = 4096;
 
+	fs_info->max_extent_size = BTRFS_MAX_EXTENT_SIZE;
+
 	spin_lock_init(&fs_info->swapfile_pins_lock);
 	fs_info->swapfile_pins = RB_ROOT;
 
-	fs_info->send_in_progress = 0;
+	fs_info->bg_reclaim_threshold = BTRFS_DEFAULT_RECLAIM_THRESH;
+	INIT_WORK(&fs_info->reclaim_bgs_work, btrfs_reclaim_bgs_work);
+}
+
+static int init_mount_fs_info(struct btrfs_fs_info *fs_info, struct super_block *sb)
+{
+	int ret;
+
+	fs_info->sb = sb;
+	sb->s_blocksize = BTRFS_BDEV_BLOCKSIZE;
+	sb->s_blocksize_bits = blksize_bits(BTRFS_BDEV_BLOCKSIZE);
+
+	ret = percpu_counter_init(&fs_info->ordered_bytes, 0, GFP_KERNEL);
+	if (ret)
+		return ret;
+
+	ret = percpu_counter_init(&fs_info->dirty_metadata_bytes, 0, GFP_KERNEL);
+	if (ret)
+		return ret;
+
+	fs_info->dirty_metadata_batch = PAGE_SIZE *
+					(1 + ilog2(nr_cpu_ids));
+
+	ret = percpu_counter_init(&fs_info->delalloc_bytes, 0, GFP_KERNEL);
+	if (ret)
+		return ret;
+
+	ret = percpu_counter_init(&fs_info->dev_replace.bio_counter, 0,
+			GFP_KERNEL);
+	if (ret)
+		return ret;
+
+	fs_info->delayed_root = kmalloc(sizeof(struct btrfs_delayed_root),
+					GFP_KERNEL);
+	if (!fs_info->delayed_root)
+		return -ENOMEM;
+	btrfs_init_delayed_root(fs_info->delayed_root);
+
+	if (sb_rdonly(sb))
+		set_bit(BTRFS_FS_STATE_RO, &fs_info->fs_state);
+
+	return btrfs_alloc_stripe_hash_table(fs_info);
+}
+
+static int btrfs_uuid_rescan_kthread(void *data)
+{
+	struct btrfs_fs_info *fs_info = data;
+	int ret;
+
+	/*
+	 * 1st step is to iterate through the existing UUID tree and
+	 * to delete all entries that contain outdated data.
+	 * 2nd step is to add all missing entries to the UUID tree.
+	 */
+	ret = btrfs_uuid_tree_iterate(fs_info);
+	if (ret < 0) {
+		if (ret != -EINTR)
+			btrfs_warn(fs_info, "iterating uuid_tree failed %d",
+				   ret);
+		up(&fs_info->uuid_tree_rescan_sem);
+		return ret;
+	}
+	return btrfs_uuid_scan_kthread(data);
+}
+
+static int btrfs_check_uuid_tree(struct btrfs_fs_info *fs_info)
+{
+	struct task_struct *task;
+
+	down(&fs_info->uuid_tree_rescan_sem);
+	task = kthread_run(btrfs_uuid_rescan_kthread, fs_info, "btrfs-uuid");
+	if (IS_ERR(task)) {
+		/* fs_info->update_uuid_tree_gen remains 0 in all error case */
+		btrfs_warn(fs_info, "failed to start uuid_rescan task");
+		up(&fs_info->uuid_tree_rescan_sem);
+		return PTR_ERR(task);
+	}
+
+	return 0;
+}
+
+/*
+ * Some options only have meaning at mount time and shouldn't persist across
+ * remounts, or be displayed. Clear these at the end of mount and remount
+ * code paths.
+ */
+void btrfs_clear_oneshot_options(struct btrfs_fs_info *fs_info)
+{
+	btrfs_clear_opt(fs_info->mount_opt, USEBACKUPROOT);
+	btrfs_clear_opt(fs_info->mount_opt, CLEAR_CACHE);
+}
+
+/*
+ * Mounting logic specific to read-write file systems. Shared by open_ctree
+ * and btrfs_remount when remounting from read-only to read-write.
+ */
+int btrfs_start_pre_rw_mount(struct btrfs_fs_info *fs_info)
+{
+	int ret;
+	const bool cache_opt = btrfs_test_opt(fs_info, SPACE_CACHE);
+	bool clear_free_space_tree = false;
+
+	if (btrfs_test_opt(fs_info, CLEAR_CACHE) &&
+	    btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE)) {
+		clear_free_space_tree = true;
+	} else if (btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE) &&
+		   !btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE_VALID)) {
+		btrfs_warn(fs_info, "free space tree is invalid");
+		clear_free_space_tree = true;
+	}
+
+	if (clear_free_space_tree) {
+		btrfs_info(fs_info, "clearing free space tree");
+		ret = btrfs_clear_free_space_tree(fs_info);
+		if (ret) {
+			btrfs_warn(fs_info,
+				   "failed to clear free space tree: %d", ret);
+			goto out;
+		}
+	}
+
+	/*
+	 * btrfs_find_orphan_roots() is responsible for finding all the dead
+	 * roots (with 0 refs), flag them with BTRFS_ROOT_DEAD_TREE and load
+	 * them into the fs_info->fs_roots_radix tree. This must be done before
+	 * calling btrfs_orphan_cleanup() on the tree root. If we don't do it
+	 * first, then btrfs_orphan_cleanup() will delete a dead root's orphan
+	 * item before the root's tree is deleted - this means that if we unmount
+	 * or crash before the deletion completes, on the next mount we will not
+	 * delete what remains of the tree because the orphan item does not
+	 * exists anymore, which is what tells us we have a pending deletion.
+	 */
+	ret = btrfs_find_orphan_roots(fs_info);
+	if (ret)
+		goto out;
+
+	ret = btrfs_cleanup_fs_roots(fs_info);
+	if (ret)
+		goto out;
+
+	down_read(&fs_info->cleanup_work_sem);
+	if ((ret = btrfs_orphan_cleanup(fs_info->fs_root)) ||
+	    (ret = btrfs_orphan_cleanup(fs_info->tree_root))) {
+		up_read(&fs_info->cleanup_work_sem);
+		goto out;
+	}
+	up_read(&fs_info->cleanup_work_sem);
+
+	mutex_lock(&fs_info->cleaner_mutex);
+	ret = btrfs_recover_relocation(fs_info);
+	mutex_unlock(&fs_info->cleaner_mutex);
+	if (ret < 0) {
+		btrfs_warn(fs_info, "failed to recover relocation: %d", ret);
+		goto out;
+	}
+
+	if (btrfs_test_opt(fs_info, FREE_SPACE_TREE) &&
+	    !btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE)) {
+		btrfs_info(fs_info, "creating free space tree");
+		ret = btrfs_create_free_space_tree(fs_info);
+		if (ret) {
+			btrfs_warn(fs_info,
+				"failed to create free space tree: %d", ret);
+			goto out;
+		}
+	}
+
+	if (cache_opt != btrfs_free_space_cache_v1_active(fs_info)) {
+		ret = btrfs_set_free_space_cache_v1_active(fs_info, cache_opt);
+		if (ret)
+			goto out;
+	}
+
+	ret = btrfs_resume_balance_async(fs_info);
+	if (ret)
+		goto out;
+
+	ret = btrfs_resume_dev_replace_async(fs_info);
+	if (ret) {
+		btrfs_warn(fs_info, "failed to resume dev_replace");
+		goto out;
+	}
+
+	btrfs_qgroup_rescan_resume(fs_info);
+
+	if (!fs_info->uuid_root) {
+		btrfs_info(fs_info, "creating UUID tree");
+		ret = btrfs_create_uuid_tree(fs_info);
+		if (ret) {
+			btrfs_warn(fs_info,
+				   "failed to create the UUID tree %d", ret);
+			goto out;
+		}
+	}
+
+out:
+	return ret;
+}
+
+/*
+ * Do various sanity and dependency checks of different features.
+ *
+ * This is the place for less strict checks (like for subpage or artificial
+ * feature dependencies).
+ *
+ * For strict checks or possible corruption detection, see
+ * btrfs_validate_super().
+ *
+ * This should be called after btrfs_parse_options(), as some mount options
+ * (space cache related) can modify on-disk format like free space tree and
+ * screw up certain feature dependencies.
+ */
+int btrfs_check_features(struct btrfs_fs_info *fs_info, struct super_block *sb)
+{
+	struct btrfs_super_block *disk_super = fs_info->super_copy;
+	u64 incompat = btrfs_super_incompat_flags(disk_super);
+	const u64 compat_ro = btrfs_super_compat_ro_flags(disk_super);
+	const u64 compat_ro_unsupp = (compat_ro & ~BTRFS_FEATURE_COMPAT_RO_SUPP);
+
+	if (incompat & ~BTRFS_FEATURE_INCOMPAT_SUPP) {
+		btrfs_err(fs_info,
+		"cannot mount because of unknown incompat features (0x%llx)",
+		    incompat);
+		return -EINVAL;
+	}
+
+	/* Runtime limitation for mixed block groups. */
+	if ((incompat & BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS) &&
+	    (fs_info->sectorsize != fs_info->nodesize)) {
+		btrfs_err(fs_info,
+"unequal nodesize/sectorsize (%u != %u) are not allowed for mixed block groups",
+			fs_info->nodesize, fs_info->sectorsize);
+		return -EINVAL;
+	}
+
+	/* Mixed backref is an always-enabled feature. */
+	incompat |= BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF;
+
+	/* Set compression related flags just in case. */
+	if (fs_info->compress_type == BTRFS_COMPRESS_LZO)
+		incompat |= BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO;
+	else if (fs_info->compress_type == BTRFS_COMPRESS_ZSTD)
+		incompat |= BTRFS_FEATURE_INCOMPAT_COMPRESS_ZSTD;
+
+	/*
+	 * An ancient flag, which should really be marked deprecated.
+	 * Such runtime limitation doesn't really need a incompat flag.
+	 */
+	if (btrfs_super_nodesize(disk_super) > PAGE_SIZE)
+		incompat |= BTRFS_FEATURE_INCOMPAT_BIG_METADATA;
+
+	if (compat_ro_unsupp && !sb_rdonly(sb)) {
+		btrfs_err(fs_info,
+	"cannot mount read-write because of unknown compat_ro features (0x%llx)",
+		       compat_ro);
+		return -EINVAL;
+	}
+
+	/*
+	 * We have unsupported RO compat features, although RO mounted, we
+	 * should not cause any metadata writes, including log replay.
+	 * Or we could screw up whatever the new feature requires.
+	 */
+	if (compat_ro_unsupp && btrfs_super_log_root(disk_super) &&
+	    !btrfs_test_opt(fs_info, NOLOGREPLAY)) {
+		btrfs_err(fs_info,
+"cannot replay dirty log with unsupported compat_ro features (0x%llx), try rescue=nologreplay",
+			  compat_ro);
+		return -EINVAL;
+	}
+
+	/*
+	 * Artificial limitations for block group tree, to force
+	 * block-group-tree to rely on no-holes and free-space-tree.
+	 */
+	if (btrfs_fs_compat_ro(fs_info, BLOCK_GROUP_TREE) &&
+	    (!btrfs_fs_incompat(fs_info, NO_HOLES) ||
+	     !btrfs_test_opt(fs_info, FREE_SPACE_TREE))) {
+		btrfs_err(fs_info,
+"block-group-tree feature requires no-holes and free-space-tree features");
+		return -EINVAL;
+	}
+
+	/*
+	 * Subpage runtime limitation on v1 cache.
+	 *
+	 * V1 space cache still has some hard codeed PAGE_SIZE usage, while
+	 * we're already defaulting to v2 cache, no need to bother v1 as it's
+	 * going to be deprecated anyway.
+	 */
+	if (fs_info->sectorsize < PAGE_SIZE && btrfs_test_opt(fs_info, SPACE_CACHE)) {
+		btrfs_warn(fs_info,
+	"v1 space cache is not supported for page size %lu with sectorsize %u",
+			   PAGE_SIZE, fs_info->sectorsize);
+		return -EINVAL;
+	}
+
+	/* This can be called by remount, we need to protect the super block. */
+	spin_lock(&fs_info->super_lock);
+	btrfs_set_super_incompat_flags(disk_super, incompat);
+	spin_unlock(&fs_info->super_lock);
+
+	return 0;
+}
+
+int __cold open_ctree(struct super_block *sb, struct btrfs_fs_devices *fs_devices,
+		      char *options)
+{
+	u32 sectorsize;
+	u32 nodesize;
+	u32 stripesize;
+	u64 generation;
+	u64 features;
+	u16 csum_type;
+	struct btrfs_super_block *disk_super;
+	struct btrfs_fs_info *fs_info = btrfs_sb(sb);
+	struct btrfs_root *tree_root;
+	struct btrfs_root *chunk_root;
+	int ret;
+	int err = -EINVAL;
+	int level;
 
-	ret = btrfs_alloc_stripe_hash_table(fs_info);
+	ret = init_mount_fs_info(fs_info, sb);
 	if (ret) {
 		err = ret;
-		goto fail_alloc;
+		goto fail;
+	}
+
+	/* These need to be init'ed before we start creating inodes and such. */
+	tree_root = btrfs_alloc_root(fs_info, BTRFS_ROOT_TREE_OBJECTID,
+				     GFP_KERNEL);
+	fs_info->tree_root = tree_root;
+	chunk_root = btrfs_alloc_root(fs_info, BTRFS_CHUNK_TREE_OBJECTID,
+				      GFP_KERNEL);
+	fs_info->chunk_root = chunk_root;
+	if (!tree_root || !chunk_root) {
+		err = -ENOMEM;
+		goto fail;
 	}
 
-	__setup_root(tree_root, fs_info, BTRFS_ROOT_TREE_OBJECTID);
+	fs_info->btree_inode = new_inode(sb);
+	if (!fs_info->btree_inode) {
+		err = -ENOMEM;
+		goto fail;
+	}
+	mapping_set_gfp_mask(fs_info->btree_inode->i_mapping, GFP_NOFS);
+	btrfs_init_btree_inode(fs_info);
 
-	invalidate_bdev(fs_devices->latest_bdev);
+	invalidate_bdev(fs_devices->latest_dev->bdev);
 
 	/*
 	 * Read super block and check the signature bytes only
 	 */
-	bh = btrfs_read_dev_super(fs_devices->latest_bdev);
-	if (IS_ERR(bh)) {
-		err = PTR_ERR(bh);
+	disk_super = btrfs_read_dev_super(fs_devices->latest_dev->bdev);
+	if (IS_ERR(disk_super)) {
+		err = PTR_ERR(disk_super);
 		goto fail_alloc;
 	}
 
 	/*
-	 * Verify the type first, if that or the the checksum value are
+	 * Verify the type first, if that or the checksum value are
 	 * corrupted, we'll find out
 	 */
-	csum_type = btrfs_super_csum_type((struct btrfs_super_block *)bh->b_data);
+	csum_type = btrfs_super_csum_type(disk_super);
 	if (!btrfs_supported_super_csum(csum_type)) {
 		btrfs_err(fs_info, "unsupported checksum algorithm: %u",
 			  csum_type);
 		err = -EINVAL;
-		brelse(bh);
+		btrfs_release_disk_super(disk_super);
 		goto fail_alloc;
 	}
 
+	fs_info->csum_size = btrfs_super_csum_size(disk_super);
+
 	ret = btrfs_init_csum_hash(fs_info, csum_type);
 	if (ret) {
 		err = ret;
+		btrfs_release_disk_super(disk_super);
 		goto fail_alloc;
 	}
 
@@ -2860,11 +3479,11 @@ int __cold open_ctree(struct super_block *sb,
 	 * We want to check superblock checksum, the type is stored inside.
 	 * Pass the whole disk block of size BTRFS_SUPER_INFO_SIZE (4k).
 	 */
-	if (btrfs_check_super_csum(fs_info, bh->b_data)) {
+	if (btrfs_check_super_csum(fs_info, disk_super)) {
 		btrfs_err(fs_info, "superblock checksum mismatch");
 		err = -EINVAL;
-		brelse(bh);
-		goto fail_csum;
+		btrfs_release_disk_super(disk_super);
+		goto fail_alloc;
 	}
 
 	/*
@@ -2872,19 +3491,11 @@ int __cold open_ctree(struct super_block *sb,
 	 * following bytes up to INFO_SIZE, the checksum is calculated from
 	 * the whole block of INFO_SIZE
 	 */
-	memcpy(fs_info->super_copy, bh->b_data, sizeof(*fs_info->super_copy));
-	brelse(bh);
+	memcpy(fs_info->super_copy, disk_super, sizeof(*fs_info->super_copy));
+	btrfs_release_disk_super(disk_super);
 
 	disk_super = fs_info->super_copy;
 
-	ASSERT(!memcmp(fs_info->fs_devices->fsid, fs_info->super_copy->fsid,
-		       BTRFS_FSID_SIZE));
-
-	if (btrfs_fs_incompat(fs_info, METADATA_UUID)) {
-		ASSERT(!memcmp(fs_info->fs_devices->metadata_uuid,
-				fs_info->super_copy->metadata_uuid,
-				BTRFS_FSID_SIZE));
-	}
 
 	features = btrfs_super_flags(disk_super);
 	if (features & BTRFS_SUPER_FLAG_CHANGING_FSID_V2) {
@@ -2901,11 +3512,11 @@ int __cold open_ctree(struct super_block *sb,
 	if (ret) {
 		btrfs_err(fs_info, "superblock contains fatal errors");
 		err = -EINVAL;
-		goto fail_csum;
+		goto fail_alloc;
 	}
 
 	if (!btrfs_super_root(disk_super))
-		goto fail_csum;
+		goto fail_alloc;
 
 	/* check FS state, whether FS is broken. */
 	if (btrfs_super_flags(disk_super) & BTRFS_SUPER_FLAG_ERROR)
@@ -2917,92 +3528,62 @@ int __cold open_ctree(struct super_block *sb,
 	 */
 	fs_info->compress_type = BTRFS_COMPRESS_ZLIB;
 
-	ret = btrfs_parse_options(fs_info, options, sb->s_flags);
-	if (ret) {
-		err = ret;
-		goto fail_csum;
-	}
-
-	features = btrfs_super_incompat_flags(disk_super) &
-		~BTRFS_FEATURE_INCOMPAT_SUPP;
-	if (features) {
-		btrfs_err(fs_info,
-		    "cannot mount because of unsupported optional features (%llx)",
-		    features);
-		err = -EINVAL;
-		goto fail_csum;
-	}
-
-	features = btrfs_super_incompat_flags(disk_super);
-	features |= BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF;
-	if (fs_info->compress_type == BTRFS_COMPRESS_LZO)
-		features |= BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO;
-	else if (fs_info->compress_type == BTRFS_COMPRESS_ZSTD)
-		features |= BTRFS_FEATURE_INCOMPAT_COMPRESS_ZSTD;
-
-	if (features & BTRFS_FEATURE_INCOMPAT_SKINNY_METADATA)
-		btrfs_info(fs_info, "has skinny extents");
-
-	/*
-	 * flag our filesystem as having big metadata blocks if
-	 * they are bigger than the page size
-	 */
-	if (btrfs_super_nodesize(disk_super) > PAGE_SIZE) {
-		if (!(features & BTRFS_FEATURE_INCOMPAT_BIG_METADATA))
-			btrfs_info(fs_info,
-				"flagging fs with big metadata feature");
-		features |= BTRFS_FEATURE_INCOMPAT_BIG_METADATA;
-	}
 
+	/* Set up fs_info before parsing mount options */
 	nodesize = btrfs_super_nodesize(disk_super);
 	sectorsize = btrfs_super_sectorsize(disk_super);
 	stripesize = sectorsize;
 	fs_info->dirty_metadata_batch = nodesize * (1 + ilog2(nr_cpu_ids));
 	fs_info->delalloc_batch = sectorsize * 512 * (1 + ilog2(nr_cpu_ids));
 
-	/* Cache block sizes */
 	fs_info->nodesize = nodesize;
 	fs_info->sectorsize = sectorsize;
+	fs_info->sectorsize_bits = ilog2(sectorsize);
+	fs_info->csums_per_leaf = BTRFS_MAX_ITEM_SIZE(fs_info) / fs_info->csum_size;
 	fs_info->stripesize = stripesize;
 
-	/*
-	 * mixed block groups end up with duplicate but slightly offset
-	 * extent buffers for the same range.  It leads to corruptions
-	 */
-	if ((features & BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS) &&
-	    (sectorsize != nodesize)) {
-		btrfs_err(fs_info,
-"unequal nodesize/sectorsize (%u != %u) are not allowed for mixed block groups",
-			nodesize, sectorsize);
-		goto fail_csum;
+	ret = btrfs_parse_options(fs_info, options, sb->s_flags);
+	if (ret) {
+		err = ret;
+		goto fail_alloc;
 	}
 
-	/*
-	 * Needn't use the lock because there is no other task which will
-	 * update the flag.
-	 */
-	btrfs_set_super_incompat_flags(disk_super, features);
+	ret = btrfs_check_features(fs_info, sb);
+	if (ret < 0) {
+		err = ret;
+		goto fail_alloc;
+	}
 
-	features = btrfs_super_compat_ro_flags(disk_super) &
-		~BTRFS_FEATURE_COMPAT_RO_SUPP;
-	if (!sb_rdonly(sb) && features) {
-		btrfs_err(fs_info,
-	"cannot mount read-write because of unsupported optional features (%llx)",
-		       features);
-		err = -EINVAL;
-		goto fail_csum;
+	if (sectorsize < PAGE_SIZE) {
+		struct btrfs_subpage_info *subpage_info;
+
+		/*
+		 * V1 space cache has some hardcoded PAGE_SIZE usage, and is
+		 * going to be deprecated.
+		 *
+		 * Force to use v2 cache for subpage case.
+		 */
+		btrfs_clear_opt(fs_info->mount_opt, SPACE_CACHE);
+		btrfs_set_and_info(fs_info, FREE_SPACE_TREE,
+			"forcing free space tree for sector size %u with page size %lu",
+			sectorsize, PAGE_SIZE);
+
+		btrfs_warn(fs_info,
+		"read-write for sector size %u with page size %lu is experimental",
+			   sectorsize, PAGE_SIZE);
+		subpage_info = kzalloc(sizeof(*subpage_info), GFP_KERNEL);
+		if (!subpage_info)
+			goto fail_alloc;
+		btrfs_init_subpage_info(subpage_info, sectorsize);
+		fs_info->subpage_info = subpage_info;
 	}
 
-	ret = btrfs_init_workqueues(fs_info, fs_devices);
+	ret = btrfs_init_workqueues(fs_info);
 	if (ret) {
 		err = ret;
 		goto fail_sb_buffer;
 	}
 
-	sb->s_bdi->congested_fn = btrfs_congested_fn;
-	sb->s_bdi->congested_data = fs_info;
-	sb->s_bdi->capabilities |= BDI_CAP_CGROUP_WRITEBACK;
-	sb->s_bdi->ra_pages = VM_READAHEAD_PAGES;
 	sb->s_bdi->ra_pages *= btrfs_super_num_devices(disk_super);
 	sb->s_bdi->ra_pages = max(sb->s_bdi->ra_pages, SZ_4M / PAGE_SIZE);
 
@@ -3020,25 +3601,16 @@ int __cold open_ctree(struct super_block *sb,
 
 	generation = btrfs_super_chunk_root_generation(disk_super);
 	level = btrfs_super_chunk_root_level(disk_super);
-
-	__setup_root(chunk_root, fs_info, BTRFS_CHUNK_TREE_OBJECTID);
-
-	chunk_root->node = read_tree_block(fs_info,
-					   btrfs_super_chunk_root(disk_super),
-					   generation, level, NULL);
-	if (IS_ERR(chunk_root->node) ||
-	    !extent_buffer_uptodate(chunk_root->node)) {
+	ret = load_super_root(chunk_root, btrfs_super_chunk_root(disk_super),
+			      generation, level);
+	if (ret) {
 		btrfs_err(fs_info, "failed to read chunk root");
-		if (!IS_ERR(chunk_root->node))
-			free_extent_buffer(chunk_root->node);
-		chunk_root->node = NULL;
 		goto fail_tree_roots;
 	}
-	btrfs_set_root_node(&chunk_root->root_item, chunk_root->node);
-	chunk_root->commit_root = btrfs_root_node(chunk_root);
 
 	read_extent_buffer(chunk_root->node, fs_info->chunk_tree_uuid,
-	   btrfs_header_chunk_tree_uuid(chunk_root->node), BTRFS_UUID_SIZE);
+			   offsetof(struct btrfs_header, chunk_tree_uuid),
+			   BTRFS_UUID_SIZE);
 
 	ret = btrfs_read_chunk_tree(fs_info);
 	if (ret) {
@@ -3047,12 +3619,14 @@ int __cold open_ctree(struct super_block *sb,
 	}
 
 	/*
-	 * Keep the devid that is marked to be the target device for the
-	 * device replace procedure
+	 * At this point we know all the devices that make this filesystem,
+	 * including the seed devices but we don't know yet if the replace
+	 * target is required. So free devices that are not part of this
+	 * filesystem but skip the replace target device which is checked
+	 * below in btrfs_init_dev_replace().
 	 */
-	btrfs_free_extra_devids(fs_devices, 0);
-
-	if (!fs_devices->latest_bdev) {
+	btrfs_free_extra_devids(fs_devices);
+	if (!fs_devices->latest_dev->bdev) {
 		btrfs_err(fs_info, "failed to read devices");
 		goto fail_tree_roots;
 	}
@@ -3061,6 +3635,31 @@ int __cold open_ctree(struct super_block *sb,
 	if (ret)
 		goto fail_tree_roots;
 
+	/*
+	 * Get zone type information of zoned block devices. This will also
+	 * handle emulation of a zoned filesystem if a regular device has the
+	 * zoned incompat feature flag set.
+	 */
+	ret = btrfs_get_dev_zone_info_all_devices(fs_info);
+	if (ret) {
+		btrfs_err(fs_info,
+			  "zoned: failed to read device zone info: %d",
+			  ret);
+		goto fail_block_groups;
+	}
+
+	/*
+	 * If we have a uuid root and we're not being told to rescan we need to
+	 * check the generation here so we can set the
+	 * BTRFS_FS_UPDATE_UUID_TREE_GEN bit.  Otherwise we could commit the
+	 * transaction during a balance or the log replay without updating the
+	 * uuid generation, and then if we crash we would rescan the uuid tree,
+	 * even though it was perfectly fine.
+	 */
+	if (fs_info->uuid_root && !btrfs_test_opt(fs_info, RESCAN_UUID_TREE) &&
+	    fs_info->generation == btrfs_super_uuid_tree_generation(disk_super))
+		set_bit(BTRFS_FS_UPDATE_UUID_TREE_GEN, &fs_info->flags);
+
 	ret = btrfs_verify_dev_extents(fs_info);
 	if (ret) {
 		btrfs_err(fs_info,
@@ -3086,7 +3685,12 @@ int __cold open_ctree(struct super_block *sb,
 		goto fail_block_groups;
 	}
 
-	btrfs_free_extra_devids(fs_devices, 1);
+	ret = btrfs_check_zoned_mode(fs_info);
+	if (ret) {
+		btrfs_err(fs_info, "failed to initialize zoned mode: %d",
+			  ret);
+		goto fail_block_groups;
+	}
 
 	ret = btrfs_sysfs_add_fsid(fs_devices);
 	if (ret) {
@@ -3113,13 +3717,16 @@ int __cold open_ctree(struct super_block *sb,
 		goto fail_sysfs;
 	}
 
-	if (!sb_rdonly(sb) && !btrfs_check_rw_degradable(fs_info, NULL)) {
+	btrfs_free_zone_cache(fs_info);
+
+	if (!sb_rdonly(sb) && fs_info->fs_devices->missing_devices &&
+	    !btrfs_check_rw_degradable(fs_info, NULL)) {
 		btrfs_warn(fs_info,
 		"writable mount is not allowed due to too many missing devices");
 		goto fail_sysfs;
 	}
 
-	fs_info->cleaner_kthread = kthread_run(cleaner_kthread, tree_root,
+	fs_info->cleaner_kthread = kthread_run(cleaner_kthread, fs_info,
 					       "btrfs-cleaner");
 	if (IS_ERR(fs_info->cleaner_kthread))
 		goto fail_sysfs;
@@ -3145,8 +3752,7 @@ int __cold open_ctree(struct super_block *sb,
 	if (btrfs_test_opt(fs_info, CHECK_INTEGRITY)) {
 		ret = btrfsic_mount(fs_info, fs_devices,
 				    btrfs_test_opt(fs_info,
-					CHECK_INTEGRITY_INCLUDING_EXTENT_DATA) ?
-				    1 : 0,
+					CHECK_INTEGRITY_DATA) ? 1 : 0,
 				    fs_info->check_integrity_print_mask);
 		if (ret)
 			btrfs_warn(fs_info,
@@ -3172,31 +3778,7 @@ int __cold open_ctree(struct super_block *sb,
 		}
 	}
 
-	ret = btrfs_find_orphan_roots(fs_info);
-	if (ret)
-		goto fail_qgroup;
-
-	if (!sb_rdonly(sb)) {
-		ret = btrfs_cleanup_fs_roots(fs_info);
-		if (ret)
-			goto fail_qgroup;
-
-		mutex_lock(&fs_info->cleaner_mutex);
-		ret = btrfs_recover_relocation(tree_root);
-		mutex_unlock(&fs_info->cleaner_mutex);
-		if (ret < 0) {
-			btrfs_warn(fs_info, "failed to recover relocation: %d",
-					ret);
-			err = -EINVAL;
-			goto fail_qgroup;
-		}
-	}
-
-	location.objectid = BTRFS_FS_TREE_OBJECTID;
-	location.type = BTRFS_ROOT_ITEM_KEY;
-	location.offset = 0;
-
-	fs_info->fs_root = btrfs_read_fs_root_no_name(fs_info, &location);
+	fs_info->fs_root = btrfs_get_fs_root(fs_info, BTRFS_FS_TREE_OBJECTID, true);
 	if (IS_ERR(fs_info->fs_root)) {
 		err = PTR_ERR(fs_info->fs_root);
 		btrfs_warn(fs_info, "failed to read fs tree: %d", err);
@@ -3205,78 +3787,18 @@ int __cold open_ctree(struct super_block *sb,
 	}
 
 	if (sb_rdonly(sb))
-		return 0;
-
-	if (btrfs_test_opt(fs_info, CLEAR_CACHE) &&
-	    btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE)) {
-		clear_free_space_tree = 1;
-	} else if (btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE) &&
-		   !btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE_VALID)) {
-		btrfs_warn(fs_info, "free space tree is invalid");
-		clear_free_space_tree = 1;
-	}
+		goto clear_oneshot;
 
-	if (clear_free_space_tree) {
-		btrfs_info(fs_info, "clearing free space tree");
-		ret = btrfs_clear_free_space_tree(fs_info);
-		if (ret) {
-			btrfs_warn(fs_info,
-				   "failed to clear free space tree: %d", ret);
-			close_ctree(fs_info);
-			return ret;
-		}
-	}
-
-	if (btrfs_test_opt(fs_info, FREE_SPACE_TREE) &&
-	    !btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE)) {
-		btrfs_info(fs_info, "creating free space tree");
-		ret = btrfs_create_free_space_tree(fs_info);
-		if (ret) {
-			btrfs_warn(fs_info,
-				"failed to create free space tree: %d", ret);
-			close_ctree(fs_info);
-			return ret;
-		}
-	}
-
-	down_read(&fs_info->cleanup_work_sem);
-	if ((ret = btrfs_orphan_cleanup(fs_info->fs_root)) ||
-	    (ret = btrfs_orphan_cleanup(fs_info->tree_root))) {
-		up_read(&fs_info->cleanup_work_sem);
-		close_ctree(fs_info);
-		return ret;
-	}
-	up_read(&fs_info->cleanup_work_sem);
-
-	ret = btrfs_resume_balance_async(fs_info);
+	ret = btrfs_start_pre_rw_mount(fs_info);
 	if (ret) {
-		btrfs_warn(fs_info, "failed to resume balance: %d", ret);
 		close_ctree(fs_info);
 		return ret;
 	}
-
-	ret = btrfs_resume_dev_replace_async(fs_info);
-	if (ret) {
-		btrfs_warn(fs_info, "failed to resume device replace: %d", ret);
-		close_ctree(fs_info);
-		return ret;
-	}
-
-	btrfs_qgroup_rescan_resume(fs_info);
 	btrfs_discard_resume(fs_info);
 
-	if (!fs_info->uuid_root) {
-		btrfs_info(fs_info, "creating UUID tree");
-		ret = btrfs_create_uuid_tree(fs_info);
-		if (ret) {
-			btrfs_warn(fs_info,
-				"failed to create the UUID tree: %d", ret);
-			close_ctree(fs_info);
-			return ret;
-		}
-	} else if (btrfs_test_opt(fs_info, RESCAN_UUID_TREE) ||
-		   fs_info->generation !=
-				btrfs_super_uuid_tree_generation(disk_super)) {
+	if (fs_info->uuid_root &&
+	    (btrfs_test_opt(fs_info, RESCAN_UUID_TREE) ||
+	     fs_info->generation != btrfs_super_uuid_tree_generation(disk_super))) {
 		btrfs_info(fs_info, "checking UUID tree");
 		ret = btrfs_check_uuid_tree(fs_info);
 		if (ret) {
@@ -3285,17 +3807,16 @@ int __cold open_ctree(struct super_block *sb,
 			close_ctree(fs_info);
 			return ret;
 		}
-	} else {
-		set_bit(BTRFS_FS_UPDATE_UUID_TREE_GEN, &fs_info->flags);
 	}
+
 	set_bit(BTRFS_FS_OPEN, &fs_info->flags);
 
-	/*
-	 * backuproot only affect mount behavior, and if open_ctree succeeded,
-	 * no need to keep the flag
-	 */
-	btrfs_clear_opt(fs_info->mount_opt, USEBACKUPROOT);
+	/* Kick the cleaner thread so it'll start deleting snapshots. */
+	if (test_bit(BTRFS_FS_UNFINISHED_DROPS, &fs_info->flags))
+		wake_up_process(fs_info->cleaner_kthread);
 
+clear_oneshot:
+	btrfs_clear_oneshot_options(fs_info);
 	return 0;
 
 fail_qgroup:
@@ -3323,96 +3844,110 @@ fail_block_groups:
 	btrfs_put_block_group_cache(fs_info);
 
 fail_tree_roots:
+	if (fs_info->data_reloc_root)
+		btrfs_drop_and_free_fs_root(fs_info, fs_info->data_reloc_root);
 	free_root_pointers(fs_info, true);
 	invalidate_inode_pages2(fs_info->btree_inode->i_mapping);
 
 fail_sb_buffer:
 	btrfs_stop_all_workers(fs_info);
 	btrfs_free_block_groups(fs_info);
-fail_csum:
-	btrfs_free_csum_hash(fs_info);
 fail_alloc:
-fail_iput:
 	btrfs_mapping_tree_free(&fs_info->mapping_tree);
 
 	iput(fs_info->btree_inode);
-fail_bio_counter:
-	percpu_counter_destroy(&fs_info->dev_replace.bio_counter);
-fail_delalloc_bytes:
-	percpu_counter_destroy(&fs_info->delalloc_bytes);
-fail_dirty_metadata_bytes:
-	percpu_counter_destroy(&fs_info->dirty_metadata_bytes);
-fail_dio_bytes:
-	percpu_counter_destroy(&fs_info->dio_bytes);
-fail_srcu:
-	cleanup_srcu_struct(&fs_info->subvol_srcu);
 fail:
-	btrfs_free_stripe_hash_table(fs_info);
 	btrfs_close_devices(fs_info->fs_devices);
 	return err;
 }
 ALLOW_ERROR_INJECTION(open_ctree, ERRNO);
 
-static void btrfs_end_buffer_write_sync(struct buffer_head *bh, int uptodate)
+static void btrfs_end_super_write(struct bio *bio)
 {
-	if (uptodate) {
-		set_buffer_uptodate(bh);
-	} else {
-		struct btrfs_device *device = (struct btrfs_device *)
-			bh->b_private;
-
-		btrfs_warn_rl_in_rcu(device->fs_info,
-				"lost page write due to IO error on %s",
-					  rcu_str_deref(device->name));
-		/* note, we don't set_buffer_write_io_error because we have
-		 * our own ways of dealing with the IO errors
-		 */
-		clear_buffer_uptodate(bh);
-		btrfs_dev_stat_inc_and_print(device, BTRFS_DEV_STAT_WRITE_ERRS);
+	struct btrfs_device *device = bio->bi_private;
+	struct bio_vec *bvec;
+	struct bvec_iter_all iter_all;
+	struct page *page;
+
+	bio_for_each_segment_all(bvec, bio, iter_all) {
+		page = bvec->bv_page;
+
+		if (bio->bi_status) {
+			btrfs_warn_rl_in_rcu(device->fs_info,
+				"lost page write due to IO error on %s (%d)",
+				rcu_str_deref(device->name),
+				blk_status_to_errno(bio->bi_status));
+			ClearPageUptodate(page);
+			SetPageError(page);
+			btrfs_dev_stat_inc_and_print(device,
+						     BTRFS_DEV_STAT_WRITE_ERRS);
+		} else {
+			SetPageUptodate(page);
+		}
+
+		put_page(page);
+		unlock_page(page);
 	}
-	unlock_buffer(bh);
-	put_bh(bh);
+
+	bio_put(bio);
 }
 
-int btrfs_read_dev_one_super(struct block_device *bdev, int copy_num,
-			struct buffer_head **bh_ret)
+struct btrfs_super_block *btrfs_read_dev_one_super(struct block_device *bdev,
+						   int copy_num, bool drop_cache)
 {
-	struct buffer_head *bh;
 	struct btrfs_super_block *super;
-	u64 bytenr;
+	struct page *page;
+	u64 bytenr, bytenr_orig;
+	struct address_space *mapping = bdev->bd_inode->i_mapping;
+	int ret;
 
-	bytenr = btrfs_sb_offset(copy_num);
-	if (bytenr + BTRFS_SUPER_INFO_SIZE >= i_size_read(bdev->bd_inode))
-		return -EINVAL;
+	bytenr_orig = btrfs_sb_offset(copy_num);
+	ret = btrfs_sb_log_location_bdev(bdev, copy_num, READ, &bytenr);
+	if (ret == -ENOENT)
+		return ERR_PTR(-EINVAL);
+	else if (ret)
+		return ERR_PTR(ret);
 
-	bh = __bread(bdev, bytenr / BTRFS_BDEV_BLOCKSIZE, BTRFS_SUPER_INFO_SIZE);
-	/*
-	 * If we fail to read from the underlying devices, as of now
-	 * the best option we have is to mark it EIO.
-	 */
-	if (!bh)
-		return -EIO;
+	if (bytenr + BTRFS_SUPER_INFO_SIZE >= bdev_nr_bytes(bdev))
+		return ERR_PTR(-EINVAL);
 
-	super = (struct btrfs_super_block *)bh->b_data;
-	if (btrfs_super_bytenr(super) != bytenr ||
-		    btrfs_super_magic(super) != BTRFS_MAGIC) {
-		brelse(bh);
-		return -EINVAL;
+	if (drop_cache) {
+		/* This should only be called with the primary sb. */
+		ASSERT(copy_num == 0);
+
+		/*
+		 * Drop the page of the primary superblock, so later read will
+		 * always read from the device.
+		 */
+		invalidate_inode_pages2_range(mapping,
+				bytenr >> PAGE_SHIFT,
+				(bytenr + BTRFS_SUPER_INFO_SIZE) >> PAGE_SHIFT);
 	}
 
-	*bh_ret = bh;
-	return 0;
+	page = read_cache_page_gfp(mapping, bytenr >> PAGE_SHIFT, GFP_NOFS);
+	if (IS_ERR(page))
+		return ERR_CAST(page);
+
+	super = page_address(page);
+	if (btrfs_super_magic(super) != BTRFS_MAGIC) {
+		btrfs_release_disk_super(super);
+		return ERR_PTR(-ENODATA);
+	}
+
+	if (btrfs_super_bytenr(super) != bytenr_orig) {
+		btrfs_release_disk_super(super);
+		return ERR_PTR(-EINVAL);
+	}
+
+	return super;
 }
 
 
-struct buffer_head *btrfs_read_dev_super(struct block_device *bdev)
+struct btrfs_super_block *btrfs_read_dev_super(struct block_device *bdev)
 {
-	struct buffer_head *bh;
-	struct buffer_head *latest = NULL;
-	struct btrfs_super_block *super;
+	struct btrfs_super_block *super, *latest = NULL;
 	int i;
 	u64 transid = 0;
-	int ret = -EINVAL;
 
 	/* we would like to check all the supers, but that would make
 	 * a btrfs mount succeed after a mkfs from a different FS.
@@ -3420,48 +3955,42 @@ struct buffer_head *btrfs_read_dev_super(struct block_device *bdev)
 	 * later supers, using BTRFS_SUPER_MIRROR_MAX instead
 	 */
 	for (i = 0; i < 1; i++) {
-		ret = btrfs_read_dev_one_super(bdev, i, &bh);
-		if (ret)
+		super = btrfs_read_dev_one_super(bdev, i, false);
+		if (IS_ERR(super))
 			continue;
 
-		super = (struct btrfs_super_block *)bh->b_data;
-
 		if (!latest || btrfs_super_generation(super) > transid) {
-			brelse(latest);
-			latest = bh;
+			if (latest)
+				btrfs_release_disk_super(super);
+
+			latest = super;
 			transid = btrfs_super_generation(super);
-		} else {
-			brelse(bh);
 		}
 	}
 
-	if (!latest)
-		return ERR_PTR(ret);
-
-	return latest;
+	return super;
 }
 
 /*
  * Write superblock @sb to the @device. Do not wait for completion, all the
- * buffer heads we write are pinned.
+ * pages we use for writing are locked.
  *
  * Write @max_mirrors copies of the superblock, where 0 means default that fit
  * the expected device size at commit time. Note that max_mirrors must be
  * same for write and wait phases.
  *
- * Return number of errors when buffer head is not found or submission fails.
+ * Return number of errors when page is not found or submission fails.
  */
 static int write_dev_supers(struct btrfs_device *device,
 			    struct btrfs_super_block *sb, int max_mirrors)
 {
 	struct btrfs_fs_info *fs_info = device->fs_info;
+	struct address_space *mapping = device->bdev->bd_inode->i_mapping;
 	SHASH_DESC_ON_STACK(shash, fs_info->csum_shash);
-	struct buffer_head *bh;
 	int i;
-	int ret;
 	int errors = 0;
-	u64 bytenr;
-	int op_flags;
+	int ret;
+	u64 bytenr, bytenr_orig;
 
 	if (max_mirrors == 0)
 		max_mirrors = BTRFS_SUPER_MIRROR_MAX;
@@ -3469,48 +3998,73 @@ static int write_dev_supers(struct btrfs_device *device,
 	shash->tfm = fs_info->csum_shash;
 
 	for (i = 0; i < max_mirrors; i++) {
-		bytenr = btrfs_sb_offset(i);
+		struct page *page;
+		struct bio *bio;
+		struct btrfs_super_block *disk_super;
+
+		bytenr_orig = btrfs_sb_offset(i);
+		ret = btrfs_sb_log_location(device, i, WRITE, &bytenr);
+		if (ret == -ENOENT) {
+			continue;
+		} else if (ret < 0) {
+			btrfs_err(device->fs_info,
+				"couldn't get super block location for mirror %d",
+				i);
+			errors++;
+			continue;
+		}
 		if (bytenr + BTRFS_SUPER_INFO_SIZE >=
 		    device->commit_total_bytes)
 			break;
 
-		btrfs_set_super_bytenr(sb, bytenr);
+		btrfs_set_super_bytenr(sb, bytenr_orig);
 
-		crypto_shash_init(shash);
-		crypto_shash_update(shash, (const char *)sb + BTRFS_CSUM_SIZE,
-				    BTRFS_SUPER_INFO_SIZE - BTRFS_CSUM_SIZE);
-		crypto_shash_final(shash, sb->csum);
+		crypto_shash_digest(shash, (const char *)sb + BTRFS_CSUM_SIZE,
+				    BTRFS_SUPER_INFO_SIZE - BTRFS_CSUM_SIZE,
+				    sb->csum);
 
-		/* One reference for us, and we leave it for the caller */
-		bh = __getblk(device->bdev, bytenr / BTRFS_BDEV_BLOCKSIZE,
-			      BTRFS_SUPER_INFO_SIZE);
-		if (!bh) {
+		page = find_or_create_page(mapping, bytenr >> PAGE_SHIFT,
+					   GFP_NOFS);
+		if (!page) {
 			btrfs_err(device->fs_info,
-			    "couldn't get super buffer head for bytenr %llu",
+			    "couldn't get super block page for bytenr %llu",
 			    bytenr);
 			errors++;
 			continue;
 		}
 
-		memcpy(bh->b_data, sb, BTRFS_SUPER_INFO_SIZE);
+		/* Bump the refcount for wait_dev_supers() */
+		get_page(page);
 
-		/* one reference for submit_bh */
-		get_bh(bh);
+		disk_super = page_address(page);
+		memcpy(disk_super, sb, BTRFS_SUPER_INFO_SIZE);
 
-		set_buffer_uptodate(bh);
-		lock_buffer(bh);
-		bh->b_end_io = btrfs_end_buffer_write_sync;
-		bh->b_private = device;
+		/*
+		 * Directly use bios here instead of relying on the page cache
+		 * to do I/O, so we don't lose the ability to do integrity
+		 * checking.
+		 */
+		bio = bio_alloc(device->bdev, 1,
+				REQ_OP_WRITE | REQ_SYNC | REQ_META | REQ_PRIO,
+				GFP_NOFS);
+		bio->bi_iter.bi_sector = bytenr >> SECTOR_SHIFT;
+		bio->bi_private = device;
+		bio->bi_end_io = btrfs_end_super_write;
+		__bio_add_page(bio, page, BTRFS_SUPER_INFO_SIZE,
+			       offset_in_page(bytenr));
 
 		/*
-		 * we fua the first super.  The others we allow
-		 * to go down lazy.
+		 * We FUA only the first super block.  The others we allow to
+		 * go down lazy and there's a short window where the on-disk
+		 * copies might still contain the older version.
 		 */
-		op_flags = REQ_SYNC | REQ_META | REQ_PRIO;
 		if (i == 0 && !btrfs_test_opt(device->fs_info, NOBARRIER))
-			op_flags |= REQ_FUA;
-		ret = btrfsic_submit_bh(REQ_OP_WRITE, op_flags, bh);
-		if (ret)
+			bio->bi_opf |= REQ_FUA;
+
+		btrfsic_check_bio(bio);
+		submit_bio(bio);
+
+		if (btrfs_advance_sb_log(device, i))
 			errors++;
 	}
 	return errors < i ? 0 : -1;
@@ -3520,47 +4074,57 @@ static int write_dev_supers(struct btrfs_device *device,
  * Wait for write completion of superblocks done by write_dev_supers,
  * @max_mirrors same for write and wait phases.
  *
- * Return number of errors when buffer head is not found or not marked up to
+ * Return number of errors when page is not found or not marked up to
  * date.
  */
 static int wait_dev_supers(struct btrfs_device *device, int max_mirrors)
 {
-	struct buffer_head *bh;
 	int i;
 	int errors = 0;
 	bool primary_failed = false;
+	int ret;
 	u64 bytenr;
 
 	if (max_mirrors == 0)
 		max_mirrors = BTRFS_SUPER_MIRROR_MAX;
 
 	for (i = 0; i < max_mirrors; i++) {
-		bytenr = btrfs_sb_offset(i);
+		struct page *page;
+
+		ret = btrfs_sb_log_location(device, i, READ, &bytenr);
+		if (ret == -ENOENT) {
+			break;
+		} else if (ret < 0) {
+			errors++;
+			if (i == 0)
+				primary_failed = true;
+			continue;
+		}
 		if (bytenr + BTRFS_SUPER_INFO_SIZE >=
 		    device->commit_total_bytes)
 			break;
 
-		bh = __find_get_block(device->bdev,
-				      bytenr / BTRFS_BDEV_BLOCKSIZE,
-				      BTRFS_SUPER_INFO_SIZE);
-		if (!bh) {
+		page = find_get_page(device->bdev->bd_inode->i_mapping,
+				     bytenr >> PAGE_SHIFT);
+		if (!page) {
 			errors++;
 			if (i == 0)
 				primary_failed = true;
 			continue;
 		}
-		wait_on_buffer(bh);
-		if (!buffer_uptodate(bh)) {
+		/* Page is submitted locked and unlocked once the IO completes */
+		wait_on_page_locked(page);
+		if (PageError(page)) {
 			errors++;
 			if (i == 0)
 				primary_failed = true;
 		}
 
-		/* drop our reference */
-		brelse(bh);
+		/* Drop our reference */
+		put_page(page);
 
-		/* drop the reference from the writing run */
-		brelse(bh);
+		/* Drop the reference from the writing run */
+		put_page(page);
 	}
 
 	/* log error, force error return */
@@ -3579,6 +4143,7 @@ static int wait_dev_supers(struct btrfs_device *device, int max_mirrors)
  */
 static void btrfs_end_empty_barrier(struct bio *bio)
 {
+	bio_uninit(bio);
 	complete(bio->bi_private);
 }
 
@@ -3588,20 +4153,31 @@ static void btrfs_end_empty_barrier(struct bio *bio)
  */
 static void write_dev_flush(struct btrfs_device *device)
 {
-	struct request_queue *q = bdev_get_queue(device->bdev);
-	struct bio *bio = device->flush_bio;
+	struct bio *bio = &device->flush_bio;
 
-	if (!test_bit(QUEUE_FLAG_WC, &q->queue_flags))
+#ifndef CONFIG_BTRFS_FS_CHECK_INTEGRITY
+	/*
+	 * When a disk has write caching disabled, we skip submission of a bio
+	 * with flush and sync requests before writing the superblock, since
+	 * it's not needed. However when the integrity checker is enabled, this
+	 * results in reports that there are metadata blocks referred by a
+	 * superblock that were not properly flushed. So don't skip the bio
+	 * submission only when the integrity checker is enabled for the sake
+	 * of simplicity, since this is a debug tool and not meant for use in
+	 * non-debug builds.
+	 */
+	if (!bdev_write_cache(device->bdev))
 		return;
+#endif
 
-	bio_reset(bio);
+	bio_init(bio, device->bdev, NULL, 0,
+		 REQ_OP_WRITE | REQ_SYNC | REQ_PREFLUSH);
 	bio->bi_end_io = btrfs_end_empty_barrier;
-	bio_set_dev(bio, device->bdev);
-	bio->bi_opf = REQ_OP_WRITE | REQ_SYNC | REQ_PREFLUSH;
 	init_completion(&device->flush_wait);
 	bio->bi_private = &device->flush_wait;
 
-	btrfsic_submit_bio(bio);
+	btrfsic_check_bio(bio);
+	submit_bio(bio);
 	set_bit(BTRFS_DEV_STATE_FLUSH_SENT, &device->dev_state);
 }
 
@@ -3610,7 +4186,7 @@ static void write_dev_flush(struct btrfs_device *device)
  */
 static blk_status_t wait_dev_flush(struct btrfs_device *device)
 {
-	struct bio *bio = device->flush_bio;
+	struct bio *bio = &device->flush_bio;
 
 	if (!test_bit(BTRFS_DEV_STATE_FLUSH_SENT, &device->dev_state))
 		return BLK_STS_OK;
@@ -3832,44 +4408,25 @@ int write_all_supers(struct btrfs_fs_info *fs_info, int max_mirrors)
 void btrfs_drop_and_free_fs_root(struct btrfs_fs_info *fs_info,
 				  struct btrfs_root *root)
 {
+	bool drop_ref = false;
+
 	spin_lock(&fs_info->fs_roots_radix_lock);
 	radix_tree_delete(&fs_info->fs_roots_radix,
 			  (unsigned long)root->root_key.objectid);
+	if (test_and_clear_bit(BTRFS_ROOT_IN_RADIX, &root->state))
+		drop_ref = true;
 	spin_unlock(&fs_info->fs_roots_radix_lock);
 
-	if (btrfs_root_refs(&root->root_item) == 0)
-		synchronize_srcu(&fs_info->subvol_srcu);
-
-	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
-		btrfs_free_log(NULL, root);
+	if (BTRFS_FS_ERROR(fs_info)) {
+		ASSERT(root->log_root == NULL);
 		if (root->reloc_root) {
-			free_extent_buffer(root->reloc_root->node);
-			free_extent_buffer(root->reloc_root->commit_root);
-			btrfs_put_fs_root(root->reloc_root);
+			btrfs_put_root(root->reloc_root);
 			root->reloc_root = NULL;
 		}
 	}
 
-	if (root->free_ino_pinned)
-		__btrfs_remove_free_space_cache(root->free_ino_pinned);
-	if (root->free_ino_ctl)
-		__btrfs_remove_free_space_cache(root->free_ino_ctl);
-	btrfs_free_fs_root(root);
-}
-
-void btrfs_free_fs_root(struct btrfs_root *root)
-{
-	iput(root->ino_cache_inode);
-	WARN_ON(!RB_EMPTY_ROOT(&root->inode_tree));
-	if (root->anon_dev)
-		free_anon_bdev(root->anon_dev);
-	if (root->subv_writers)
-		btrfs_free_subvolume_writers(root->subv_writers);
-	free_extent_buffer(root->node);
-	free_extent_buffer(root->commit_root);
-	kfree(root->free_ino_ctl);
-	kfree(root->free_ino_pinned);
-	btrfs_put_fs_root(root);
+	if (drop_ref)
+		btrfs_put_root(root);
 }
 
 int btrfs_cleanup_fs_roots(struct btrfs_fs_info *fs_info)
@@ -3879,15 +4436,14 @@ int btrfs_cleanup_fs_roots(struct btrfs_fs_info *fs_info)
 	int i = 0;
 	int err = 0;
 	unsigned int ret = 0;
-	int index;
 
 	while (1) {
-		index = srcu_read_lock(&fs_info->subvol_srcu);
+		spin_lock(&fs_info->fs_roots_radix_lock);
 		ret = radix_tree_gang_lookup(&fs_info->fs_roots_radix,
 					     (void **)gang, root_objectid,
 					     ARRAY_SIZE(gang));
 		if (!ret) {
-			srcu_read_unlock(&fs_info->subvol_srcu, index);
+			spin_unlock(&fs_info->fs_roots_radix_lock);
 			break;
 		}
 		root_objectid = gang[ret - 1]->root_key.objectid + 1;
@@ -3899,9 +4455,9 @@ int btrfs_cleanup_fs_roots(struct btrfs_fs_info *fs_info)
 				continue;
 			}
 			/* grab all the search result for later use */
-			gang[i] = btrfs_grab_fs_root(gang[i]);
+			gang[i] = btrfs_grab_root(gang[i]);
 		}
-		srcu_read_unlock(&fs_info->subvol_srcu, index);
+		spin_unlock(&fs_info->fs_roots_radix_lock);
 
 		for (i = 0; i < ret; i++) {
 			if (!gang[i])
@@ -3910,7 +4466,7 @@ int btrfs_cleanup_fs_roots(struct btrfs_fs_info *fs_info)
 			err = btrfs_orphan_cleanup(gang[i]);
 			if (err)
 				break;
-			btrfs_put_fs_root(gang[i]);
+			btrfs_put_root(gang[i]);
 		}
 		root_objectid++;
 	}
@@ -3918,7 +4474,7 @@ int btrfs_cleanup_fs_roots(struct btrfs_fs_info *fs_info)
 	/* release the uncleaned roots due to error */
 	for (; i < ret; i++) {
 		if (gang[i])
-			btrfs_put_fs_root(gang[i]);
+			btrfs_put_root(gang[i]);
 	}
 	return err;
 }
@@ -3943,11 +4499,75 @@ int btrfs_commit_super(struct btrfs_fs_info *fs_info)
 	return btrfs_commit_transaction(trans);
 }
 
+static void warn_about_uncommitted_trans(struct btrfs_fs_info *fs_info)
+{
+	struct btrfs_transaction *trans;
+	struct btrfs_transaction *tmp;
+	bool found = false;
+
+	if (list_empty(&fs_info->trans_list))
+		return;
+
+	/*
+	 * This function is only called at the very end of close_ctree(),
+	 * thus no other running transaction, no need to take trans_lock.
+	 */
+	ASSERT(test_bit(BTRFS_FS_CLOSING_DONE, &fs_info->flags));
+	list_for_each_entry_safe(trans, tmp, &fs_info->trans_list, list) {
+		struct extent_state *cached = NULL;
+		u64 dirty_bytes = 0;
+		u64 cur = 0;
+		u64 found_start;
+		u64 found_end;
+
+		found = true;
+		while (!find_first_extent_bit(&trans->dirty_pages, cur,
+			&found_start, &found_end, EXTENT_DIRTY, &cached)) {
+			dirty_bytes += found_end + 1 - found_start;
+			cur = found_end + 1;
+		}
+		btrfs_warn(fs_info,
+	"transaction %llu (with %llu dirty metadata bytes) is not committed",
+			   trans->transid, dirty_bytes);
+		btrfs_cleanup_one_transaction(trans, fs_info);
+
+		if (trans == fs_info->running_transaction)
+			fs_info->running_transaction = NULL;
+		list_del_init(&trans->list);
+
+		btrfs_put_transaction(trans);
+		trace_btrfs_transaction_commit(fs_info);
+	}
+	ASSERT(!found);
+}
+
 void __cold close_ctree(struct btrfs_fs_info *fs_info)
 {
 	int ret;
 
 	set_bit(BTRFS_FS_CLOSING_START, &fs_info->flags);
+
+	/*
+	 * If we had UNFINISHED_DROPS we could still be processing them, so
+	 * clear that bit and wake up relocation so it can stop.
+	 * We must do this before stopping the block group reclaim task, because
+	 * at btrfs_relocate_block_group() we wait for this bit, and after the
+	 * wait we stop with -EINTR if btrfs_fs_closing() returns non-zero - we
+	 * have just set BTRFS_FS_CLOSING_START, so btrfs_fs_closing() will
+	 * return 1.
+	 */
+	btrfs_wake_unfinished_drop(fs_info);
+
+	/*
+	 * We may have the reclaim task running and relocating a data block group,
+	 * in which case it may create delayed iputs. So stop it before we park
+	 * the cleaner kthread otherwise we can get new delayed iputs after
+	 * parking the cleaner, and that can make the async reclaim task to hang
+	 * if it's waiting for delayed iputs to complete, since the cleaner is
+	 * parked and can not run delayed iputs - this will make us hang when
+	 * trying to stop the async reclaim task.
+	 */
+	cancel_work_sync(&fs_info->reclaim_bgs_work);
 	/*
 	 * We don't want the cleaner to start new transactions, add more delayed
 	 * iputs, etc. while we're closing. We can't use kthread_stop() yet
@@ -3978,7 +4598,34 @@ void __cold close_ctree(struct btrfs_fs_info *fs_info)
 	/* clear out the rbtree of defraggable inodes */
 	btrfs_cleanup_defrag_inodes(fs_info);
 
+	/*
+	 * After we parked the cleaner kthread, ordered extents may have
+	 * completed and created new delayed iputs. If one of the async reclaim
+	 * tasks is running and in the RUN_DELAYED_IPUTS flush state, then we
+	 * can hang forever trying to stop it, because if a delayed iput is
+	 * added after it ran btrfs_run_delayed_iputs() and before it called
+	 * btrfs_wait_on_delayed_iputs(), it will hang forever since there is
+	 * no one else to run iputs.
+	 *
+	 * So wait for all ongoing ordered extents to complete and then run
+	 * delayed iputs. This works because once we reach this point no one
+	 * can either create new ordered extents nor create delayed iputs
+	 * through some other means.
+	 *
+	 * Also note that btrfs_wait_ordered_roots() is not safe here, because
+	 * it waits for BTRFS_ORDERED_COMPLETE to be set on an ordered extent,
+	 * but the delayed iput for the respective inode is made only when doing
+	 * the final btrfs_put_ordered_extent() (which must happen at
+	 * btrfs_finish_ordered_io() when we are unmounting).
+	 */
+	btrfs_flush_workqueue(fs_info->endio_write_workers);
+	/* Ordered extents for free space inodes. */
+	btrfs_flush_workqueue(fs_info->endio_freespace_worker);
+	btrfs_run_delayed_iputs(fs_info);
+
 	cancel_work_sync(&fs_info->async_reclaim_work);
+	cancel_work_sync(&fs_info->async_data_reclaim_work);
+	cancel_work_sync(&fs_info->preempt_reclaim_work);
 
 	/* Cancel or finish ongoing discard work */
 	btrfs_discard_cleanup(fs_info);
@@ -3990,13 +4637,25 @@ void __cold close_ctree(struct btrfs_fs_info *fs_info)
 		 */
 		btrfs_delete_unused_bgs(fs_info);
 
+		/*
+		 * There might be existing delayed inode workers still running
+		 * and holding an empty delayed inode item. We must wait for
+		 * them to complete first because they can create a transaction.
+		 * This happens when someone calls btrfs_balance_delayed_items()
+		 * and then a transaction commit runs the same delayed nodes
+		 * before any delayed worker has done something with the nodes.
+		 * We must wait for any worker here and not at transaction
+		 * commit time since that could cause a deadlock.
+		 * This is a very rare case.
+		 */
+		btrfs_flush_workqueue(fs_info->delayed_workers);
+
 		ret = btrfs_commit_super(fs_info);
 		if (ret)
 			btrfs_err(fs_info, "commit super ret %d", ret);
 	}
 
-	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state) ||
-	    test_bit(BTRFS_FS_STATE_TRANS_ABORTED, &fs_info->fs_state))
+	if (BTRFS_FS_ERROR(fs_info))
 		btrfs_error_commit_super(fs_info);
 
 	kthread_stop(fs_info->transaction_kthread);
@@ -4005,6 +4664,11 @@ void __cold close_ctree(struct btrfs_fs_info *fs_info)
 	ASSERT(list_empty(&fs_info->delayed_iputs));
 	set_bit(BTRFS_FS_CLOSING_DONE, &fs_info->flags);
 
+	if (btrfs_check_quota_leak(fs_info)) {
+		WARN_ON(IS_ENABLED(CONFIG_BTRFS_DEBUG));
+		btrfs_err(fs_info, "qgroup reserved space leaked");
+	}
+
 	btrfs_free_qgroup_config(fs_info);
 	ASSERT(list_empty(&fs_info->delalloc_roots));
 
@@ -4013,15 +4677,13 @@ void __cold close_ctree(struct btrfs_fs_info *fs_info)
 		       percpu_counter_sum(&fs_info->delalloc_bytes));
 	}
 
-	if (percpu_counter_sum(&fs_info->dio_bytes))
+	if (percpu_counter_sum(&fs_info->ordered_bytes))
 		btrfs_info(fs_info, "at unmount dio bytes count %lld",
-			   percpu_counter_sum(&fs_info->dio_bytes));
+			   percpu_counter_sum(&fs_info->ordered_bytes));
 
 	btrfs_sysfs_remove_mounted(fs_info);
 	btrfs_sysfs_remove_fsid(fs_info->fs_devices);
 
-	btrfs_free_fs_roots(fs_info);
-
 	btrfs_put_block_group_cache(fs_info);
 
 	/*
@@ -4031,8 +4693,12 @@ void __cold close_ctree(struct btrfs_fs_info *fs_info)
 	invalidate_inode_pages2(fs_info->btree_inode->i_mapping);
 	btrfs_stop_all_workers(fs_info);
 
+	/* We shouldn't have any transaction open at this point */
+	warn_about_uncommitted_trans(fs_info);
+
 	clear_bit(BTRFS_FS_OPEN, &fs_info->flags);
 	free_root_pointers(fs_info, true);
+	btrfs_free_fs_roots(fs_info);
 
 	/*
 	 * We must free the block groups after dropping the fs_roots as we could
@@ -4052,16 +4718,6 @@ void __cold close_ctree(struct btrfs_fs_info *fs_info)
 
 	btrfs_mapping_tree_free(&fs_info->mapping_tree);
 	btrfs_close_devices(fs_info->fs_devices);
-
-	percpu_counter_destroy(&fs_info->dirty_metadata_bytes);
-	percpu_counter_destroy(&fs_info->delalloc_bytes);
-	percpu_counter_destroy(&fs_info->dio_bytes);
-	percpu_counter_destroy(&fs_info->dev_replace.bio_counter);
-	cleanup_srcu_struct(&fs_info->subvol_srcu);
-
-	btrfs_free_csum_hash(fs_info);
-	btrfs_free_stripe_hash_table(fs_info);
-	btrfs_free_ref_cache(fs_info);
 }
 
 int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid,
@@ -4083,8 +4739,7 @@ int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid,
 
 void btrfs_mark_buffer_dirty(struct extent_buffer *buf)
 {
-	struct btrfs_fs_info *fs_info;
-	struct btrfs_root *root;
+	struct btrfs_fs_info *fs_info = buf->fs_info;
 	u64 transid = btrfs_header_generation(buf);
 	int was_dirty;
 
@@ -4097,9 +4752,7 @@ void btrfs_mark_buffer_dirty(struct extent_buffer *buf)
 	if (unlikely(test_bit(EXTENT_BUFFER_UNMAPPED, &buf->bflags)))
 		return;
 #endif
-	root = BTRFS_I(buf->pages[0]->mapping->host)->root;
-	fs_info = root->fs_info;
-	btrfs_assert_tree_locked(buf);
+	btrfs_assert_tree_write_locked(buf);
 	if (transid != fs_info->generation)
 		WARN(1, KERN_CRIT "btrfs transid mismatch buffer %llu, found %llu running %llu\n",
 			buf->start, transid, fs_info->generation);
@@ -4155,13 +4808,6 @@ void btrfs_btree_balance_dirty_nodelay(struct btrfs_fs_info *fs_info)
 	__btrfs_btree_balance_dirty(fs_info, 0);
 }
 
-int btrfs_read_buffer(struct extent_buffer *buf, u64 parent_transid, int level,
-		      struct btrfs_key *first_key)
-{
-	return btree_read_extent_buffer_pages(buf, parent_transid,
-					      level, first_key);
-}
-
 static void btrfs_error_commit_super(struct btrfs_fs_info *fs_info)
 {
 	/* cleanup FS via transaction */
@@ -4175,6 +4821,36 @@ static void btrfs_error_commit_super(struct btrfs_fs_info *fs_info)
 	up_write(&fs_info->cleanup_work_sem);
 }
 
+static void btrfs_drop_all_logs(struct btrfs_fs_info *fs_info)
+{
+	struct btrfs_root *gang[8];
+	u64 root_objectid = 0;
+	int ret;
+
+	spin_lock(&fs_info->fs_roots_radix_lock);
+	while ((ret = radix_tree_gang_lookup(&fs_info->fs_roots_radix,
+					     (void **)gang, root_objectid,
+					     ARRAY_SIZE(gang))) != 0) {
+		int i;
+
+		for (i = 0; i < ret; i++)
+			gang[i] = btrfs_grab_root(gang[i]);
+		spin_unlock(&fs_info->fs_roots_radix_lock);
+
+		for (i = 0; i < ret; i++) {
+			if (!gang[i])
+				continue;
+			root_objectid = gang[i]->root_key.objectid;
+			btrfs_free_log(NULL, gang[i]);
+			btrfs_put_root(gang[i]);
+		}
+		root_objectid++;
+		spin_lock(&fs_info->fs_roots_radix_lock);
+	}
+	spin_unlock(&fs_info->fs_roots_radix_lock);
+	btrfs_free_log_root_tree(NULL, fs_info);
+}
+
 static void btrfs_destroy_ordered_extents(struct btrfs_root *root)
 {
 	struct btrfs_ordered_extent *ordered;
@@ -4235,7 +4911,7 @@ static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
 	spin_lock(&delayed_refs->lock);
 	if (atomic_read(&delayed_refs->num_entries) == 0) {
 		spin_unlock(&delayed_refs->lock);
-		btrfs_info(fs_info, "delayed_refs has NO entry");
+		btrfs_debug(fs_info, "delayed_refs has NO entry");
 		return ret;
 	}
 
@@ -4269,9 +4945,27 @@ static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
 		spin_unlock(&delayed_refs->lock);
 		mutex_unlock(&head->mutex);
 
-		if (pin_bytes)
-			btrfs_pin_extent(fs_info, head->bytenr,
-					 head->num_bytes, 1);
+		if (pin_bytes) {
+			struct btrfs_block_group *cache;
+
+			cache = btrfs_lookup_block_group(fs_info, head->bytenr);
+			BUG_ON(!cache);
+
+			spin_lock(&cache->space_info->lock);
+			spin_lock(&cache->lock);
+			cache->pinned += head->num_bytes;
+			btrfs_space_info_update_bytes_pinned(fs_info,
+				cache->space_info, head->num_bytes);
+			cache->reserved -= head->num_bytes;
+			cache->space_info->bytes_reserved -= head->num_bytes;
+			spin_unlock(&cache->lock);
+			spin_unlock(&cache->space_info->lock);
+
+			btrfs_put_block_group(cache);
+
+			btrfs_error_unpin_extent_range(fs_info, head->bytenr,
+				head->bytenr + head->num_bytes - 1);
+		}
 		btrfs_cleanup_ref_head_accounting(fs_info, delayed_refs, head);
 		btrfs_put_delayed_ref_head(head);
 		cond_resched();
@@ -4327,12 +5021,12 @@ static void btrfs_destroy_all_delalloc_inodes(struct btrfs_fs_info *fs_info)
 	while (!list_empty(&splice)) {
 		root = list_first_entry(&splice, struct btrfs_root,
 					 delalloc_root);
-		root = btrfs_grab_fs_root(root);
+		root = btrfs_grab_root(root);
 		BUG_ON(!root);
 		spin_unlock(&fs_info->delalloc_root_lock);
 
 		btrfs_destroy_delalloc_inodes(root);
-		btrfs_put_fs_root(root);
+		btrfs_put_root(root);
 
 		spin_lock(&fs_info->delalloc_root_lock);
 	}
@@ -4373,16 +5067,12 @@ static int btrfs_destroy_marked_extents(struct btrfs_fs_info *fs_info,
 }
 
 static int btrfs_destroy_pinned_extent(struct btrfs_fs_info *fs_info,
-				       struct extent_io_tree *pinned_extents)
+				       struct extent_io_tree *unpin)
 {
-	struct extent_io_tree *unpin;
 	u64 start;
 	u64 end;
 	int ret;
-	bool loop = true;
 
-	unpin = pinned_extents;
-again:
 	while (1) {
 		struct extent_state *cached_state = NULL;
 
@@ -4407,15 +5097,6 @@ again:
 		cond_resched();
 	}
 
-	if (loop) {
-		if (unpin == &fs_info->freed_extents[0])
-			unpin = &fs_info->freed_extents[1];
-		else
-			unpin = &fs_info->freed_extents[0];
-		loop = false;
-		goto again;
-	}
-
 	return 0;
 }
 
@@ -4430,6 +5111,7 @@ static void btrfs_cleanup_bg_io(struct btrfs_block_group *cache)
 		cache->io_ctl.inode = NULL;
 		iput(inode);
 	}
+	ASSERT(cache->io_ctl.pages == NULL);
 	btrfs_put_block_group(cache);
 }
 
@@ -4506,8 +5188,9 @@ void btrfs_cleanup_one_transaction(struct btrfs_transaction *cur_trans,
 
 	btrfs_destroy_marked_extents(fs_info, &cur_trans->dirty_pages,
 				     EXTENT_DIRTY);
-	btrfs_destroy_pinned_extent(fs_info,
-				    fs_info->pinned_extents);
+	btrfs_destroy_pinned_extent(fs_info, &cur_trans->pinned_extents);
+
+	btrfs_free_redirty_list(cur_trans);
 
 	cur_trans->state =TRANS_STATE_COMPLETED;
 	wake_up(&cur_trans->commit_wait);
@@ -4552,22 +5235,71 @@ static int btrfs_cleanup_transaction(struct btrfs_fs_info *fs_info)
 		spin_unlock(&fs_info->trans_lock);
 
 		btrfs_put_transaction(t);
-		trace_btrfs_transaction_commit(fs_info->tree_root);
+		trace_btrfs_transaction_commit(fs_info);
 		spin_lock(&fs_info->trans_lock);
 	}
 	spin_unlock(&fs_info->trans_lock);
 	btrfs_destroy_all_ordered_extents(fs_info);
 	btrfs_destroy_delayed_inodes(fs_info);
 	btrfs_assert_delayed_root_empty(fs_info);
-	btrfs_destroy_pinned_extent(fs_info, fs_info->pinned_extents);
 	btrfs_destroy_all_delalloc_inodes(fs_info);
+	btrfs_drop_all_logs(fs_info);
 	mutex_unlock(&fs_info->transaction_kthread_mutex);
 
 	return 0;
 }
 
-static const struct extent_io_ops btree_extent_io_ops = {
-	/* mandatory callbacks */
-	.submit_bio_hook = btree_submit_bio_hook,
-	.readpage_end_io_hook = btree_readpage_end_io_hook,
-};
+int btrfs_init_root_free_objectid(struct btrfs_root *root)
+{
+	struct btrfs_path *path;
+	int ret;
+	struct extent_buffer *l;
+	struct btrfs_key search_key;
+	struct btrfs_key found_key;
+	int slot;
+
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+
+	search_key.objectid = BTRFS_LAST_FREE_OBJECTID;
+	search_key.type = -1;
+	search_key.offset = (u64)-1;
+	ret = btrfs_search_slot(NULL, root, &search_key, path, 0, 0);
+	if (ret < 0)
+		goto error;
+	BUG_ON(ret == 0); /* Corruption */
+	if (path->slots[0] > 0) {
+		slot = path->slots[0] - 1;
+		l = path->nodes[0];
+		btrfs_item_key_to_cpu(l, &found_key, slot);
+		root->free_objectid = max_t(u64, found_key.objectid + 1,
+					    BTRFS_FIRST_FREE_OBJECTID);
+	} else {
+		root->free_objectid = BTRFS_FIRST_FREE_OBJECTID;
+	}
+	ret = 0;
+error:
+	btrfs_free_path(path);
+	return ret;
+}
+
+int btrfs_get_free_objectid(struct btrfs_root *root, u64 *objectid)
+{
+	int ret;
+	mutex_lock(&root->objectid_mutex);
+
+	if (unlikely(root->free_objectid >= BTRFS_LAST_FREE_OBJECTID)) {
+		btrfs_warn(root->fs_info,
+			   "the objectid of root %llu reaches its highest value",
+			   root->root_key.objectid);
+		ret = -ENOSPC;
+		goto out;
+	}
+
+	*objectid = root->free_objectid++;
+	ret = 0;
+out:
+	mutex_unlock(&root->objectid_mutex);
+	return ret;
+}
diff --git a/fs/btrfs/disk-io.h b/fs/btrfs/disk-io.h
index 8c2d6cf1ce59..9fa923e005a3 100644
--- a/fs/btrfs/disk-io.h
+++ b/fs/btrfs/disk-io.h
@@ -6,9 +6,6 @@
 #ifndef BTRFS_DISK_IO_H
 #define BTRFS_DISK_IO_H
 
-#define BTRFS_SUPER_INFO_OFFSET SZ_64K
-#define BTRFS_SUPER_INFO_SIZE 4096
-
 #define BTRFS_SUPER_MIRROR_MAX	 3
 #define BTRFS_SUPER_MIRROR_SHIFT 12
 
@@ -20,14 +17,6 @@
  */
 #define BTRFS_BDEV_BLOCKSIZE	(4096)
 
-enum btrfs_wq_endio_type {
-	BTRFS_WQ_ENDIO_DATA,
-	BTRFS_WQ_ENDIO_METADATA,
-	BTRFS_WQ_ENDIO_FREE_SPACE,
-	BTRFS_WQ_ENDIO_RAID56,
-	BTRFS_WQ_ENDIO_DIO_REPAIR,
-};
-
 static inline u64 btrfs_sb_offset(int mirror)
 {
 	u64 start = SZ_16K;
@@ -39,51 +28,64 @@ static inline u64 btrfs_sb_offset(int mirror)
 struct btrfs_device;
 struct btrfs_fs_devices;
 
+void btrfs_check_leaked_roots(struct btrfs_fs_info *fs_info);
+void btrfs_init_fs_info(struct btrfs_fs_info *fs_info);
 int btrfs_verify_level_key(struct extent_buffer *eb, int level,
 			   struct btrfs_key *first_key, u64 parent_transid);
 struct extent_buffer *read_tree_block(struct btrfs_fs_info *fs_info, u64 bytenr,
-				      u64 parent_transid, int level,
-				      struct btrfs_key *first_key);
-void readahead_tree_block(struct btrfs_fs_info *fs_info, u64 bytenr);
+				      u64 owner_root, u64 parent_transid,
+				      int level, struct btrfs_key *first_key);
 struct extent_buffer *btrfs_find_create_tree_block(
 						struct btrfs_fs_info *fs_info,
-						u64 bytenr);
+						u64 bytenr, u64 owner_root,
+						int level);
 void btrfs_clean_tree_block(struct extent_buffer *buf);
+void btrfs_clear_oneshot_options(struct btrfs_fs_info *fs_info);
+int btrfs_start_pre_rw_mount(struct btrfs_fs_info *fs_info);
+int btrfs_check_super_csum(struct btrfs_fs_info *fs_info,
+			   const struct btrfs_super_block *disk_sb);
 int __cold open_ctree(struct super_block *sb,
 	       struct btrfs_fs_devices *fs_devices,
 	       char *options);
 void __cold close_ctree(struct btrfs_fs_info *fs_info);
+int btrfs_validate_super(struct btrfs_fs_info *fs_info,
+			 struct btrfs_super_block *sb, int mirror_num);
+int btrfs_check_features(struct btrfs_fs_info *fs_info, struct super_block *sb);
 int write_all_supers(struct btrfs_fs_info *fs_info, int max_mirrors);
-struct buffer_head *btrfs_read_dev_super(struct block_device *bdev);
-int btrfs_read_dev_one_super(struct block_device *bdev, int copy_num,
-			struct buffer_head **bh_ret);
+struct btrfs_super_block *btrfs_read_dev_super(struct block_device *bdev);
+struct btrfs_super_block *btrfs_read_dev_one_super(struct block_device *bdev,
+						   int copy_num, bool drop_cache);
 int btrfs_commit_super(struct btrfs_fs_info *fs_info);
-struct btrfs_root *btrfs_read_fs_root(struct btrfs_root *tree_root,
-				      struct btrfs_key *location);
-int btrfs_init_fs_root(struct btrfs_root *root);
-struct btrfs_root *btrfs_lookup_fs_root(struct btrfs_fs_info *fs_info,
-					u64 root_id);
+struct btrfs_root *btrfs_read_tree_root(struct btrfs_root *tree_root,
+					struct btrfs_key *key);
 int btrfs_insert_fs_root(struct btrfs_fs_info *fs_info,
 			 struct btrfs_root *root);
 void btrfs_free_fs_roots(struct btrfs_fs_info *fs_info);
 
 struct btrfs_root *btrfs_get_fs_root(struct btrfs_fs_info *fs_info,
-				     struct btrfs_key *key,
-				     bool check_ref);
-static inline struct btrfs_root *
-btrfs_read_fs_root_no_name(struct btrfs_fs_info *fs_info,
-			   struct btrfs_key *location)
-{
-	return btrfs_get_fs_root(fs_info, location, true);
-}
-
+				     u64 objectid, bool check_ref);
+struct btrfs_root *btrfs_get_new_fs_root(struct btrfs_fs_info *fs_info,
+					 u64 objectid, dev_t anon_dev);
+struct btrfs_root *btrfs_get_fs_root_commit_root(struct btrfs_fs_info *fs_info,
+						 struct btrfs_path *path,
+						 u64 objectid);
+int btrfs_global_root_insert(struct btrfs_root *root);
+void btrfs_global_root_delete(struct btrfs_root *root);
+struct btrfs_root *btrfs_global_root(struct btrfs_fs_info *fs_info,
+				     struct btrfs_key *key);
+struct btrfs_root *btrfs_csum_root(struct btrfs_fs_info *fs_info, u64 bytenr);
+struct btrfs_root *btrfs_extent_root(struct btrfs_fs_info *fs_info, u64 bytenr);
+
+void btrfs_free_fs_info(struct btrfs_fs_info *fs_info);
 int btrfs_cleanup_fs_roots(struct btrfs_fs_info *fs_info);
 void btrfs_btree_balance_dirty(struct btrfs_fs_info *fs_info);
 void btrfs_btree_balance_dirty_nodelay(struct btrfs_fs_info *fs_info);
 void btrfs_drop_and_free_fs_root(struct btrfs_fs_info *fs_info,
 				 struct btrfs_root *root);
-void btrfs_free_fs_root(struct btrfs_root *root);
-
+int btrfs_validate_metadata_buffer(struct btrfs_bio *bbio,
+				   struct page *page, u64 start, u64 end,
+				   int mirror);
+void btrfs_submit_metadata_bio(struct inode *inode, struct bio *bio, int mirror_num);
 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
 struct btrfs_root *btrfs_alloc_dummy_root(struct btrfs_fs_info *fs_info);
 #endif
@@ -95,32 +97,35 @@ struct btrfs_root *btrfs_alloc_dummy_root(struct btrfs_fs_info *fs_info);
  * If you want to ensure the whole tree is safe, you should use
  * 	fs_info->subvol_srcu
  */
-static inline struct btrfs_root *btrfs_grab_fs_root(struct btrfs_root *root)
+static inline struct btrfs_root *btrfs_grab_root(struct btrfs_root *root)
 {
+	if (!root)
+		return NULL;
 	if (refcount_inc_not_zero(&root->refs))
 		return root;
 	return NULL;
 }
 
-static inline void btrfs_put_fs_root(struct btrfs_root *root)
+static inline struct btrfs_root *btrfs_block_group_root(struct btrfs_fs_info *fs_info)
 {
-	if (refcount_dec_and_test(&root->refs))
-		kfree(root);
+	if (btrfs_fs_compat_ro(fs_info, BLOCK_GROUP_TREE))
+		return fs_info->block_group_root;
+	return btrfs_extent_root(fs_info, 0);
 }
 
+void btrfs_put_root(struct btrfs_root *root);
 void btrfs_mark_buffer_dirty(struct extent_buffer *buf);
 int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid,
 			  int atomic);
-int btrfs_read_buffer(struct extent_buffer *buf, u64 parent_transid, int level,
-		      struct btrfs_key *first_key);
-blk_status_t btrfs_bio_wq_end_io(struct btrfs_fs_info *info, struct bio *bio,
-			enum btrfs_wq_endio_type metadata);
-blk_status_t btrfs_wq_submit_bio(struct btrfs_fs_info *fs_info, struct bio *bio,
-			int mirror_num, unsigned long bio_flags,
-			u64 bio_offset, void *private_data,
-			extent_submit_bio_start_t *submit_bio_start);
+int btrfs_read_extent_buffer(struct extent_buffer *buf, u64 parent_transid,
+			     int level, struct btrfs_key *first_key);
+bool btrfs_wq_submit_bio(struct inode *inode, struct bio *bio, int mirror_num,
+			 u64 dio_file_offset,
+			 extent_submit_bio_start_t *submit_bio_start);
 blk_status_t btrfs_submit_bio_done(void *private_data, struct bio *bio,
 			  int mirror_num);
+int btrfs_alloc_log_tree_node(struct btrfs_trans_handle *trans,
+			      struct btrfs_root *root);
 int btrfs_init_log_root_tree(struct btrfs_trans_handle *trans,
 			     struct btrfs_fs_info *fs_info);
 int btrfs_add_log_tree(struct btrfs_trans_handle *trans,
@@ -133,24 +138,8 @@ struct btrfs_root *btrfs_create_tree(struct btrfs_trans_handle *trans,
 				     u64 objectid);
 int btree_lock_page_hook(struct page *page, void *data,
 				void (*flush_fn)(void *));
-struct extent_map *btree_get_extent(struct btrfs_inode *inode,
-				    struct page *page, size_t pg_offset,
-				    u64 start, u64 len);
 int btrfs_get_num_tolerated_disk_barrier_failures(u64 flags);
-int __init btrfs_end_io_wq_init(void);
-void __cold btrfs_end_io_wq_exit(void);
-
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
-void btrfs_init_lockdep(void);
-void btrfs_set_buffer_lockdep_class(u64 objectid,
-			            struct extent_buffer *eb, int level);
-#else
-static inline void btrfs_init_lockdep(void)
-{ }
-static inline void btrfs_set_buffer_lockdep_class(u64 objectid,
-					struct extent_buffer *eb, int level)
-{
-}
-#endif
+int btrfs_get_free_objectid(struct btrfs_root *root, u64 *objectid);
+int btrfs_init_root_free_objectid(struct btrfs_root *root);
 
 #endif
diff --git a/fs/btrfs/export.c b/fs/btrfs/export.c
index 72e312cae69d..fab7eb76e53b 100644
--- a/fs/btrfs/export.c
+++ b/fs/btrfs/export.c
@@ -57,43 +57,25 @@ static int btrfs_encode_fh(struct inode *inode, u32 *fh, int *max_len,
 	return type;
 }
 
-static struct dentry *btrfs_get_dentry(struct super_block *sb, u64 objectid,
-				       u64 root_objectid, u32 generation,
-				       int check_generation)
+struct dentry *btrfs_get_dentry(struct super_block *sb, u64 objectid,
+				u64 root_objectid, u64 generation,
+				int check_generation)
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(sb);
 	struct btrfs_root *root;
 	struct inode *inode;
-	struct btrfs_key key;
-	int index;
-	int err = 0;
 
 	if (objectid < BTRFS_FIRST_FREE_OBJECTID)
 		return ERR_PTR(-ESTALE);
 
-	key.objectid = root_objectid;
-	key.type = BTRFS_ROOT_ITEM_KEY;
-	key.offset = (u64)-1;
-
-	index = srcu_read_lock(&fs_info->subvol_srcu);
-
-	root = btrfs_read_fs_root_no_name(fs_info, &key);
-	if (IS_ERR(root)) {
-		err = PTR_ERR(root);
-		goto fail;
-	}
-
-	key.objectid = objectid;
-	key.type = BTRFS_INODE_ITEM_KEY;
-	key.offset = 0;
+	root = btrfs_get_fs_root(fs_info, root_objectid, true);
+	if (IS_ERR(root))
+		return ERR_CAST(root);
 
-	inode = btrfs_iget(sb, &key, root);
-	if (IS_ERR(inode)) {
-		err = PTR_ERR(inode);
-		goto fail;
-	}
-
-	srcu_read_unlock(&fs_info->subvol_srcu, index);
+	inode = btrfs_iget(sb, objectid, root);
+	btrfs_put_root(root);
+	if (IS_ERR(inode))
+		return ERR_CAST(inode);
 
 	if (check_generation && generation != inode->i_generation) {
 		iput(inode);
@@ -101,9 +83,6 @@ static struct dentry *btrfs_get_dentry(struct super_block *sb, u64 objectid,
 	}
 
 	return d_obtain_alias(inode);
-fail:
-	srcu_read_unlock(&fs_info->subvol_srcu, index);
-	return ERR_PTR(err);
 }
 
 static struct dentry *btrfs_fh_to_parent(struct super_block *sb, struct fid *fh,
@@ -152,7 +131,7 @@ static struct dentry *btrfs_fh_to_dentry(struct super_block *sb, struct fid *fh,
 	return btrfs_get_dentry(sb, objectid, root_objectid, generation, 1);
 }
 
-static struct dentry *btrfs_get_parent(struct dentry *child)
+struct dentry *btrfs_get_parent(struct dentry *child)
 {
 	struct inode *dir = d_inode(child);
 	struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb);
@@ -212,9 +191,7 @@ static struct dentry *btrfs_get_parent(struct dentry *child)
 					found_key.offset, 0, 0);
 	}
 
-	key.type = BTRFS_INODE_ITEM_KEY;
-	key.offset = 0;
-	return d_obtain_alias(btrfs_iget(fs_info->sb, &key, root));
+	return d_obtain_alias(btrfs_iget(fs_info->sb, key.objectid, root));
 fail:
 	btrfs_free_path(path);
 	return ERR_PTR(ret);
@@ -245,7 +222,6 @@ static int btrfs_get_name(struct dentry *parent, char *name,
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
-	path->leave_spinning = 1;
 
 	if (ino == BTRFS_FIRST_FREE_OBJECTID) {
 		key.objectid = BTRFS_I(inode)->root->root_key.objectid;
diff --git a/fs/btrfs/export.h b/fs/btrfs/export.h
index 57488ecd7d4e..5afb7ca42828 100644
--- a/fs/btrfs/export.h
+++ b/fs/btrfs/export.h
@@ -18,4 +18,9 @@ struct btrfs_fid {
 	u64 parent_root_objectid;
 } __attribute__ ((packed));
 
+struct dentry *btrfs_get_dentry(struct super_block *sb, u64 objectid,
+				u64 root_objectid, u64 generation,
+				int check_generation);
+struct dentry *btrfs_get_parent(struct dentry *child);
+
 #endif
diff --git a/fs/btrfs/extent-io-tree.c b/fs/btrfs/extent-io-tree.c
new file mode 100644
index 000000000000..83cb0378096f
--- /dev/null
+++ b/fs/btrfs/extent-io-tree.c
@@ -0,0 +1,1674 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <linux/slab.h>
+#include <trace/events/btrfs.h>
+#include "ctree.h"
+#include "extent-io-tree.h"
+#include "btrfs_inode.h"
+#include "misc.h"
+
+static struct kmem_cache *extent_state_cache;
+
+static inline bool extent_state_in_tree(const struct extent_state *state)
+{
+	return !RB_EMPTY_NODE(&state->rb_node);
+}
+
+#ifdef CONFIG_BTRFS_DEBUG
+static LIST_HEAD(states);
+static DEFINE_SPINLOCK(leak_lock);
+
+static inline void btrfs_leak_debug_add_state(struct extent_state *state)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&leak_lock, flags);
+	list_add(&state->leak_list, &states);
+	spin_unlock_irqrestore(&leak_lock, flags);
+}
+
+static inline void btrfs_leak_debug_del_state(struct extent_state *state)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&leak_lock, flags);
+	list_del(&state->leak_list);
+	spin_unlock_irqrestore(&leak_lock, flags);
+}
+
+static inline void btrfs_extent_state_leak_debug_check(void)
+{
+	struct extent_state *state;
+
+	while (!list_empty(&states)) {
+		state = list_entry(states.next, struct extent_state, leak_list);
+		pr_err("BTRFS: state leak: start %llu end %llu state %u in tree %d refs %d\n",
+		       state->start, state->end, state->state,
+		       extent_state_in_tree(state),
+		       refcount_read(&state->refs));
+		list_del(&state->leak_list);
+		kmem_cache_free(extent_state_cache, state);
+	}
+}
+
+#define btrfs_debug_check_extent_io_range(tree, start, end)		\
+	__btrfs_debug_check_extent_io_range(__func__, (tree), (start), (end))
+static inline void __btrfs_debug_check_extent_io_range(const char *caller,
+						       struct extent_io_tree *tree,
+						       u64 start, u64 end)
+{
+	struct inode *inode = tree->private_data;
+	u64 isize;
+
+	if (!inode)
+		return;
+
+	isize = i_size_read(inode);
+	if (end >= PAGE_SIZE && (end % 2) == 0 && end != isize - 1) {
+		btrfs_debug_rl(BTRFS_I(inode)->root->fs_info,
+		    "%s: ino %llu isize %llu odd range [%llu,%llu]",
+			caller, btrfs_ino(BTRFS_I(inode)), isize, start, end);
+	}
+}
+#else
+#define btrfs_leak_debug_add_state(state)		do {} while (0)
+#define btrfs_leak_debug_del_state(state)		do {} while (0)
+#define btrfs_extent_state_leak_debug_check()		do {} while (0)
+#define btrfs_debug_check_extent_io_range(c, s, e)	do {} while (0)
+#endif
+
+/*
+ * For the file_extent_tree, we want to hold the inode lock when we lookup and
+ * update the disk_i_size, but lockdep will complain because our io_tree we hold
+ * the tree lock and get the inode lock when setting delalloc.  These two things
+ * are unrelated, so make a class for the file_extent_tree so we don't get the
+ * two locking patterns mixed up.
+ */
+static struct lock_class_key file_extent_tree_class;
+
+struct tree_entry {
+	u64 start;
+	u64 end;
+	struct rb_node rb_node;
+};
+
+void extent_io_tree_init(struct btrfs_fs_info *fs_info,
+			 struct extent_io_tree *tree, unsigned int owner,
+			 void *private_data)
+{
+	tree->fs_info = fs_info;
+	tree->state = RB_ROOT;
+	spin_lock_init(&tree->lock);
+	tree->private_data = private_data;
+	tree->owner = owner;
+	if (owner == IO_TREE_INODE_FILE_EXTENT)
+		lockdep_set_class(&tree->lock, &file_extent_tree_class);
+}
+
+void extent_io_tree_release(struct extent_io_tree *tree)
+{
+	spin_lock(&tree->lock);
+	/*
+	 * Do a single barrier for the waitqueue_active check here, the state
+	 * of the waitqueue should not change once extent_io_tree_release is
+	 * called.
+	 */
+	smp_mb();
+	while (!RB_EMPTY_ROOT(&tree->state)) {
+		struct rb_node *node;
+		struct extent_state *state;
+
+		node = rb_first(&tree->state);
+		state = rb_entry(node, struct extent_state, rb_node);
+		rb_erase(&state->rb_node, &tree->state);
+		RB_CLEAR_NODE(&state->rb_node);
+		/*
+		 * btree io trees aren't supposed to have tasks waiting for
+		 * changes in the flags of extent states ever.
+		 */
+		ASSERT(!waitqueue_active(&state->wq));
+		free_extent_state(state);
+
+		cond_resched_lock(&tree->lock);
+	}
+	spin_unlock(&tree->lock);
+}
+
+static struct extent_state *alloc_extent_state(gfp_t mask)
+{
+	struct extent_state *state;
+
+	/*
+	 * The given mask might be not appropriate for the slab allocator,
+	 * drop the unsupported bits
+	 */
+	mask &= ~(__GFP_DMA32|__GFP_HIGHMEM);
+	state = kmem_cache_alloc(extent_state_cache, mask);
+	if (!state)
+		return state;
+	state->state = 0;
+	RB_CLEAR_NODE(&state->rb_node);
+	btrfs_leak_debug_add_state(state);
+	refcount_set(&state->refs, 1);
+	init_waitqueue_head(&state->wq);
+	trace_alloc_extent_state(state, mask, _RET_IP_);
+	return state;
+}
+
+static struct extent_state *alloc_extent_state_atomic(struct extent_state *prealloc)
+{
+	if (!prealloc)
+		prealloc = alloc_extent_state(GFP_ATOMIC);
+
+	return prealloc;
+}
+
+void free_extent_state(struct extent_state *state)
+{
+	if (!state)
+		return;
+	if (refcount_dec_and_test(&state->refs)) {
+		WARN_ON(extent_state_in_tree(state));
+		btrfs_leak_debug_del_state(state);
+		trace_free_extent_state(state, _RET_IP_);
+		kmem_cache_free(extent_state_cache, state);
+	}
+}
+
+static int add_extent_changeset(struct extent_state *state, u32 bits,
+				 struct extent_changeset *changeset,
+				 int set)
+{
+	int ret;
+
+	if (!changeset)
+		return 0;
+	if (set && (state->state & bits) == bits)
+		return 0;
+	if (!set && (state->state & bits) == 0)
+		return 0;
+	changeset->bytes_changed += state->end - state->start + 1;
+	ret = ulist_add(&changeset->range_changed, state->start, state->end,
+			GFP_ATOMIC);
+	return ret;
+}
+
+static inline struct extent_state *next_state(struct extent_state *state)
+{
+	struct rb_node *next = rb_next(&state->rb_node);
+
+	if (next)
+		return rb_entry(next, struct extent_state, rb_node);
+	else
+		return NULL;
+}
+
+static inline struct extent_state *prev_state(struct extent_state *state)
+{
+	struct rb_node *next = rb_prev(&state->rb_node);
+
+	if (next)
+		return rb_entry(next, struct extent_state, rb_node);
+	else
+		return NULL;
+}
+
+/*
+ * Search @tree for an entry that contains @offset. Such entry would have
+ * entry->start <= offset && entry->end >= offset.
+ *
+ * @tree:       the tree to search
+ * @offset:     offset that should fall within an entry in @tree
+ * @node_ret:   pointer where new node should be anchored (used when inserting an
+ *	        entry in the tree)
+ * @parent_ret: points to entry which would have been the parent of the entry,
+ *               containing @offset
+ *
+ * Return a pointer to the entry that contains @offset byte address and don't change
+ * @node_ret and @parent_ret.
+ *
+ * If no such entry exists, return pointer to entry that ends before @offset
+ * and fill parameters @node_ret and @parent_ret, ie. does not return NULL.
+ */
+static inline struct extent_state *tree_search_for_insert(struct extent_io_tree *tree,
+							  u64 offset,
+							  struct rb_node ***node_ret,
+							  struct rb_node **parent_ret)
+{
+	struct rb_root *root = &tree->state;
+	struct rb_node **node = &root->rb_node;
+	struct rb_node *prev = NULL;
+	struct extent_state *entry = NULL;
+
+	while (*node) {
+		prev = *node;
+		entry = rb_entry(prev, struct extent_state, rb_node);
+
+		if (offset < entry->start)
+			node = &(*node)->rb_left;
+		else if (offset > entry->end)
+			node = &(*node)->rb_right;
+		else
+			return entry;
+	}
+
+	if (node_ret)
+		*node_ret = node;
+	if (parent_ret)
+		*parent_ret = prev;
+
+	/* Search neighbors until we find the first one past the end */
+	while (entry && offset > entry->end)
+		entry = next_state(entry);
+
+	return entry;
+}
+
+/*
+ * Search offset in the tree or fill neighbor rbtree node pointers.
+ *
+ * @tree:      the tree to search
+ * @offset:    offset that should fall within an entry in @tree
+ * @next_ret:  pointer to the first entry whose range ends after @offset
+ * @prev_ret:  pointer to the first entry whose range begins before @offset
+ *
+ * Return a pointer to the entry that contains @offset byte address. If no
+ * such entry exists, then return NULL and fill @prev_ret and @next_ret.
+ * Otherwise return the found entry and other pointers are left untouched.
+ */
+static struct extent_state *tree_search_prev_next(struct extent_io_tree *tree,
+						  u64 offset,
+						  struct extent_state **prev_ret,
+						  struct extent_state **next_ret)
+{
+	struct rb_root *root = &tree->state;
+	struct rb_node **node = &root->rb_node;
+	struct extent_state *orig_prev;
+	struct extent_state *entry = NULL;
+
+	ASSERT(prev_ret);
+	ASSERT(next_ret);
+
+	while (*node) {
+		entry = rb_entry(*node, struct extent_state, rb_node);
+
+		if (offset < entry->start)
+			node = &(*node)->rb_left;
+		else if (offset > entry->end)
+			node = &(*node)->rb_right;
+		else
+			return entry;
+	}
+
+	orig_prev = entry;
+	while (entry && offset > entry->end)
+		entry = next_state(entry);
+	*next_ret = entry;
+	entry = orig_prev;
+
+	while (entry && offset < entry->start)
+		entry = prev_state(entry);
+	*prev_ret = entry;
+
+	return NULL;
+}
+
+/*
+ * Inexact rb-tree search, return the next entry if @offset is not found
+ */
+static inline struct extent_state *tree_search(struct extent_io_tree *tree, u64 offset)
+{
+	return tree_search_for_insert(tree, offset, NULL, NULL);
+}
+
+static void extent_io_tree_panic(struct extent_io_tree *tree, int err)
+{
+	btrfs_panic(tree->fs_info, err,
+	"locking error: extent tree was modified by another thread while locked");
+}
+
+/*
+ * Utility function to look for merge candidates inside a given range.  Any
+ * extents with matching state are merged together into a single extent in the
+ * tree.  Extents with EXTENT_IO in their state field are not merged because
+ * the end_io handlers need to be able to do operations on them without
+ * sleeping (or doing allocations/splits).
+ *
+ * This should be called with the tree lock held.
+ */
+static void merge_state(struct extent_io_tree *tree, struct extent_state *state)
+{
+	struct extent_state *other;
+
+	if (state->state & (EXTENT_LOCKED | EXTENT_BOUNDARY))
+		return;
+
+	other = prev_state(state);
+	if (other && other->end == state->start - 1 &&
+	    other->state == state->state) {
+		if (tree->private_data)
+			btrfs_merge_delalloc_extent(tree->private_data,
+						    state, other);
+		state->start = other->start;
+		rb_erase(&other->rb_node, &tree->state);
+		RB_CLEAR_NODE(&other->rb_node);
+		free_extent_state(other);
+	}
+	other = next_state(state);
+	if (other && other->start == state->end + 1 &&
+	    other->state == state->state) {
+		if (tree->private_data)
+			btrfs_merge_delalloc_extent(tree->private_data, state,
+						    other);
+		state->end = other->end;
+		rb_erase(&other->rb_node, &tree->state);
+		RB_CLEAR_NODE(&other->rb_node);
+		free_extent_state(other);
+	}
+}
+
+static void set_state_bits(struct extent_io_tree *tree,
+			   struct extent_state *state,
+			   u32 bits, struct extent_changeset *changeset)
+{
+	u32 bits_to_set = bits & ~EXTENT_CTLBITS;
+	int ret;
+
+	if (tree->private_data)
+		btrfs_set_delalloc_extent(tree->private_data, state, bits);
+
+	ret = add_extent_changeset(state, bits_to_set, changeset, 1);
+	BUG_ON(ret < 0);
+	state->state |= bits_to_set;
+}
+
+/*
+ * Insert an extent_state struct into the tree.  'bits' are set on the
+ * struct before it is inserted.
+ *
+ * This may return -EEXIST if the extent is already there, in which case the
+ * state struct is freed.
+ *
+ * The tree lock is not taken internally.  This is a utility function and
+ * probably isn't what you want to call (see set/clear_extent_bit).
+ */
+static int insert_state(struct extent_io_tree *tree,
+			struct extent_state *state,
+			u32 bits, struct extent_changeset *changeset)
+{
+	struct rb_node **node;
+	struct rb_node *parent;
+	const u64 end = state->end;
+
+	set_state_bits(tree, state, bits, changeset);
+
+	node = &tree->state.rb_node;
+	while (*node) {
+		struct extent_state *entry;
+
+		parent = *node;
+		entry = rb_entry(parent, struct extent_state, rb_node);
+
+		if (end < entry->start) {
+			node = &(*node)->rb_left;
+		} else if (end > entry->end) {
+			node = &(*node)->rb_right;
+		} else {
+			btrfs_err(tree->fs_info,
+			       "found node %llu %llu on insert of %llu %llu",
+			       entry->start, entry->end, state->start, end);
+			return -EEXIST;
+		}
+	}
+
+	rb_link_node(&state->rb_node, parent, node);
+	rb_insert_color(&state->rb_node, &tree->state);
+
+	merge_state(tree, state);
+	return 0;
+}
+
+/*
+ * Insert state to @tree to the location given by @node and @parent.
+ */
+static void insert_state_fast(struct extent_io_tree *tree,
+			      struct extent_state *state, struct rb_node **node,
+			      struct rb_node *parent, unsigned bits,
+			      struct extent_changeset *changeset)
+{
+	set_state_bits(tree, state, bits, changeset);
+	rb_link_node(&state->rb_node, parent, node);
+	rb_insert_color(&state->rb_node, &tree->state);
+	merge_state(tree, state);
+}
+
+/*
+ * Split a given extent state struct in two, inserting the preallocated
+ * struct 'prealloc' as the newly created second half.  'split' indicates an
+ * offset inside 'orig' where it should be split.
+ *
+ * Before calling,
+ * the tree has 'orig' at [orig->start, orig->end].  After calling, there
+ * are two extent state structs in the tree:
+ * prealloc: [orig->start, split - 1]
+ * orig: [ split, orig->end ]
+ *
+ * The tree locks are not taken by this function. They need to be held
+ * by the caller.
+ */
+static int split_state(struct extent_io_tree *tree, struct extent_state *orig,
+		       struct extent_state *prealloc, u64 split)
+{
+	struct rb_node *parent = NULL;
+	struct rb_node **node;
+
+	if (tree->private_data)
+		btrfs_split_delalloc_extent(tree->private_data, orig, split);
+
+	prealloc->start = orig->start;
+	prealloc->end = split - 1;
+	prealloc->state = orig->state;
+	orig->start = split;
+
+	parent = &orig->rb_node;
+	node = &parent;
+	while (*node) {
+		struct extent_state *entry;
+
+		parent = *node;
+		entry = rb_entry(parent, struct extent_state, rb_node);
+
+		if (prealloc->end < entry->start) {
+			node = &(*node)->rb_left;
+		} else if (prealloc->end > entry->end) {
+			node = &(*node)->rb_right;
+		} else {
+			free_extent_state(prealloc);
+			return -EEXIST;
+		}
+	}
+
+	rb_link_node(&prealloc->rb_node, parent, node);
+	rb_insert_color(&prealloc->rb_node, &tree->state);
+
+	return 0;
+}
+
+/*
+ * Utility function to clear some bits in an extent state struct.  It will
+ * optionally wake up anyone waiting on this state (wake == 1).
+ *
+ * If no bits are set on the state struct after clearing things, the
+ * struct is freed and removed from the tree
+ */
+static struct extent_state *clear_state_bit(struct extent_io_tree *tree,
+					    struct extent_state *state,
+					    u32 bits, int wake,
+					    struct extent_changeset *changeset)
+{
+	struct extent_state *next;
+	u32 bits_to_clear = bits & ~EXTENT_CTLBITS;
+	int ret;
+
+	if (tree->private_data)
+		btrfs_clear_delalloc_extent(tree->private_data, state, bits);
+
+	ret = add_extent_changeset(state, bits_to_clear, changeset, 0);
+	BUG_ON(ret < 0);
+	state->state &= ~bits_to_clear;
+	if (wake)
+		wake_up(&state->wq);
+	if (state->state == 0) {
+		next = next_state(state);
+		if (extent_state_in_tree(state)) {
+			rb_erase(&state->rb_node, &tree->state);
+			RB_CLEAR_NODE(&state->rb_node);
+			free_extent_state(state);
+		} else {
+			WARN_ON(1);
+		}
+	} else {
+		merge_state(tree, state);
+		next = next_state(state);
+	}
+	return next;
+}
+
+/*
+ * Clear some bits on a range in the tree.  This may require splitting or
+ * inserting elements in the tree, so the gfp mask is used to indicate which
+ * allocations or sleeping are allowed.
+ *
+ * Pass 'wake' == 1 to kick any sleepers, and 'delete' == 1 to remove the given
+ * range from the tree regardless of state (ie for truncate).
+ *
+ * The range [start, end] is inclusive.
+ *
+ * This takes the tree lock, and returns 0 on success and < 0 on error.
+ */
+int __clear_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
+		       u32 bits, struct extent_state **cached_state,
+		       gfp_t mask, struct extent_changeset *changeset)
+{
+	struct extent_state *state;
+	struct extent_state *cached;
+	struct extent_state *prealloc = NULL;
+	u64 last_end;
+	int err;
+	int clear = 0;
+	int wake;
+	int delete = (bits & EXTENT_CLEAR_ALL_BITS);
+
+	btrfs_debug_check_extent_io_range(tree, start, end);
+	trace_btrfs_clear_extent_bit(tree, start, end - start + 1, bits);
+
+	if (delete)
+		bits |= ~EXTENT_CTLBITS;
+
+	if (bits & EXTENT_DELALLOC)
+		bits |= EXTENT_NORESERVE;
+
+	wake = (bits & EXTENT_LOCKED) ? 1 : 0;
+	if (bits & (EXTENT_LOCKED | EXTENT_BOUNDARY))
+		clear = 1;
+again:
+	if (!prealloc && gfpflags_allow_blocking(mask)) {
+		/*
+		 * Don't care for allocation failure here because we might end
+		 * up not needing the pre-allocated extent state at all, which
+		 * is the case if we only have in the tree extent states that
+		 * cover our input range and don't cover too any other range.
+		 * If we end up needing a new extent state we allocate it later.
+		 */
+		prealloc = alloc_extent_state(mask);
+	}
+
+	spin_lock(&tree->lock);
+	if (cached_state) {
+		cached = *cached_state;
+
+		if (clear) {
+			*cached_state = NULL;
+			cached_state = NULL;
+		}
+
+		if (cached && extent_state_in_tree(cached) &&
+		    cached->start <= start && cached->end > start) {
+			if (clear)
+				refcount_dec(&cached->refs);
+			state = cached;
+			goto hit_next;
+		}
+		if (clear)
+			free_extent_state(cached);
+	}
+
+	/* This search will find the extents that end after our range starts. */
+	state = tree_search(tree, start);
+	if (!state)
+		goto out;
+hit_next:
+	if (state->start > end)
+		goto out;
+	WARN_ON(state->end < start);
+	last_end = state->end;
+
+	/* The state doesn't have the wanted bits, go ahead. */
+	if (!(state->state & bits)) {
+		state = next_state(state);
+		goto next;
+	}
+
+	/*
+	 *     | ---- desired range ---- |
+	 *  | state | or
+	 *  | ------------- state -------------- |
+	 *
+	 * We need to split the extent we found, and may flip bits on second
+	 * half.
+	 *
+	 * If the extent we found extends past our range, we just split and
+	 * search again.  It'll get split again the next time though.
+	 *
+	 * If the extent we found is inside our range, we clear the desired bit
+	 * on it.
+	 */
+
+	if (state->start < start) {
+		prealloc = alloc_extent_state_atomic(prealloc);
+		BUG_ON(!prealloc);
+		err = split_state(tree, state, prealloc, start);
+		if (err)
+			extent_io_tree_panic(tree, err);
+
+		prealloc = NULL;
+		if (err)
+			goto out;
+		if (state->end <= end) {
+			state = clear_state_bit(tree, state, bits, wake, changeset);
+			goto next;
+		}
+		goto search_again;
+	}
+	/*
+	 * | ---- desired range ---- |
+	 *                        | state |
+	 * We need to split the extent, and clear the bit on the first half.
+	 */
+	if (state->start <= end && state->end > end) {
+		prealloc = alloc_extent_state_atomic(prealloc);
+		BUG_ON(!prealloc);
+		err = split_state(tree, state, prealloc, end + 1);
+		if (err)
+			extent_io_tree_panic(tree, err);
+
+		if (wake)
+			wake_up(&state->wq);
+
+		clear_state_bit(tree, prealloc, bits, wake, changeset);
+
+		prealloc = NULL;
+		goto out;
+	}
+
+	state = clear_state_bit(tree, state, bits, wake, changeset);
+next:
+	if (last_end == (u64)-1)
+		goto out;
+	start = last_end + 1;
+	if (start <= end && state && !need_resched())
+		goto hit_next;
+
+search_again:
+	if (start > end)
+		goto out;
+	spin_unlock(&tree->lock);
+	if (gfpflags_allow_blocking(mask))
+		cond_resched();
+	goto again;
+
+out:
+	spin_unlock(&tree->lock);
+	if (prealloc)
+		free_extent_state(prealloc);
+
+	return 0;
+
+}
+
+static void wait_on_state(struct extent_io_tree *tree,
+			  struct extent_state *state)
+		__releases(tree->lock)
+		__acquires(tree->lock)
+{
+	DEFINE_WAIT(wait);
+	prepare_to_wait(&state->wq, &wait, TASK_UNINTERRUPTIBLE);
+	spin_unlock(&tree->lock);
+	schedule();
+	spin_lock(&tree->lock);
+	finish_wait(&state->wq, &wait);
+}
+
+/*
+ * Wait for one or more bits to clear on a range in the state tree.
+ * The range [start, end] is inclusive.
+ * The tree lock is taken by this function
+ */
+void wait_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, u32 bits)
+{
+	struct extent_state *state;
+
+	btrfs_debug_check_extent_io_range(tree, start, end);
+
+	spin_lock(&tree->lock);
+again:
+	while (1) {
+		/*
+		 * This search will find all the extents that end after our
+		 * range starts.
+		 */
+		state = tree_search(tree, start);
+process_node:
+		if (!state)
+			break;
+		if (state->start > end)
+			goto out;
+
+		if (state->state & bits) {
+			start = state->start;
+			refcount_inc(&state->refs);
+			wait_on_state(tree, state);
+			free_extent_state(state);
+			goto again;
+		}
+		start = state->end + 1;
+
+		if (start > end)
+			break;
+
+		if (!cond_resched_lock(&tree->lock)) {
+			state = next_state(state);
+			goto process_node;
+		}
+	}
+out:
+	spin_unlock(&tree->lock);
+}
+
+static void cache_state_if_flags(struct extent_state *state,
+				 struct extent_state **cached_ptr,
+				 unsigned flags)
+{
+	if (cached_ptr && !(*cached_ptr)) {
+		if (!flags || (state->state & flags)) {
+			*cached_ptr = state;
+			refcount_inc(&state->refs);
+		}
+	}
+}
+
+static void cache_state(struct extent_state *state,
+			struct extent_state **cached_ptr)
+{
+	return cache_state_if_flags(state, cached_ptr,
+				    EXTENT_LOCKED | EXTENT_BOUNDARY);
+}
+
+/*
+ * Find the first state struct with 'bits' set after 'start', and return it.
+ * tree->lock must be held.  NULL will returned if nothing was found after
+ * 'start'.
+ */
+static struct extent_state *find_first_extent_bit_state(struct extent_io_tree *tree,
+							u64 start, u32 bits)
+{
+	struct extent_state *state;
+
+	/*
+	 * This search will find all the extents that end after our range
+	 * starts.
+	 */
+	state = tree_search(tree, start);
+	while (state) {
+		if (state->end >= start && (state->state & bits))
+			return state;
+		state = next_state(state);
+	}
+	return NULL;
+}
+
+/*
+ * Find the first offset in the io tree with one or more @bits set.
+ *
+ * Note: If there are multiple bits set in @bits, any of them will match.
+ *
+ * Return 0 if we find something, and update @start_ret and @end_ret.
+ * Return 1 if we found nothing.
+ */
+int find_first_extent_bit(struct extent_io_tree *tree, u64 start,
+			  u64 *start_ret, u64 *end_ret, u32 bits,
+			  struct extent_state **cached_state)
+{
+	struct extent_state *state;
+	int ret = 1;
+
+	spin_lock(&tree->lock);
+	if (cached_state && *cached_state) {
+		state = *cached_state;
+		if (state->end == start - 1 && extent_state_in_tree(state)) {
+			while ((state = next_state(state)) != NULL) {
+				if (state->state & bits)
+					goto got_it;
+			}
+			free_extent_state(*cached_state);
+			*cached_state = NULL;
+			goto out;
+		}
+		free_extent_state(*cached_state);
+		*cached_state = NULL;
+	}
+
+	state = find_first_extent_bit_state(tree, start, bits);
+got_it:
+	if (state) {
+		cache_state_if_flags(state, cached_state, 0);
+		*start_ret = state->start;
+		*end_ret = state->end;
+		ret = 0;
+	}
+out:
+	spin_unlock(&tree->lock);
+	return ret;
+}
+
+/*
+ * Find a contiguous area of bits
+ *
+ * @tree:      io tree to check
+ * @start:     offset to start the search from
+ * @start_ret: the first offset we found with the bits set
+ * @end_ret:   the final contiguous range of the bits that were set
+ * @bits:      bits to look for
+ *
+ * set_extent_bit and clear_extent_bit can temporarily split contiguous ranges
+ * to set bits appropriately, and then merge them again.  During this time it
+ * will drop the tree->lock, so use this helper if you want to find the actual
+ * contiguous area for given bits.  We will search to the first bit we find, and
+ * then walk down the tree until we find a non-contiguous area.  The area
+ * returned will be the full contiguous area with the bits set.
+ */
+int find_contiguous_extent_bit(struct extent_io_tree *tree, u64 start,
+			       u64 *start_ret, u64 *end_ret, u32 bits)
+{
+	struct extent_state *state;
+	int ret = 1;
+
+	spin_lock(&tree->lock);
+	state = find_first_extent_bit_state(tree, start, bits);
+	if (state) {
+		*start_ret = state->start;
+		*end_ret = state->end;
+		while ((state = next_state(state)) != NULL) {
+			if (state->start > (*end_ret + 1))
+				break;
+			*end_ret = state->end;
+		}
+		ret = 0;
+	}
+	spin_unlock(&tree->lock);
+	return ret;
+}
+
+/*
+ * Find a contiguous range of bytes in the file marked as delalloc, not more
+ * than 'max_bytes'.  start and end are used to return the range,
+ *
+ * True is returned if we find something, false if nothing was in the tree.
+ */
+bool btrfs_find_delalloc_range(struct extent_io_tree *tree, u64 *start,
+			       u64 *end, u64 max_bytes,
+			       struct extent_state **cached_state)
+{
+	struct extent_state *state;
+	u64 cur_start = *start;
+	bool found = false;
+	u64 total_bytes = 0;
+
+	spin_lock(&tree->lock);
+
+	/*
+	 * This search will find all the extents that end after our range
+	 * starts.
+	 */
+	state = tree_search(tree, cur_start);
+	if (!state) {
+		*end = (u64)-1;
+		goto out;
+	}
+
+	while (state) {
+		if (found && (state->start != cur_start ||
+			      (state->state & EXTENT_BOUNDARY))) {
+			goto out;
+		}
+		if (!(state->state & EXTENT_DELALLOC)) {
+			if (!found)
+				*end = state->end;
+			goto out;
+		}
+		if (!found) {
+			*start = state->start;
+			*cached_state = state;
+			refcount_inc(&state->refs);
+		}
+		found = true;
+		*end = state->end;
+		cur_start = state->end + 1;
+		total_bytes += state->end - state->start + 1;
+		if (total_bytes >= max_bytes)
+			break;
+		state = next_state(state);
+	}
+out:
+	spin_unlock(&tree->lock);
+	return found;
+}
+
+/*
+ * Set some bits on a range in the tree.  This may require allocations or
+ * sleeping, so the gfp mask is used to indicate what is allowed.
+ *
+ * If any of the exclusive bits are set, this will fail with -EEXIST if some
+ * part of the range already has the desired bits set.  The start of the
+ * existing range is returned in failed_start in this case.
+ *
+ * [start, end] is inclusive This takes the tree lock.
+ */
+static int __set_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
+			    u32 bits, u64 *failed_start,
+			    struct extent_state **cached_state,
+			    struct extent_changeset *changeset, gfp_t mask)
+{
+	struct extent_state *state;
+	struct extent_state *prealloc = NULL;
+	struct rb_node **p;
+	struct rb_node *parent;
+	int err = 0;
+	u64 last_start;
+	u64 last_end;
+	u32 exclusive_bits = (bits & EXTENT_LOCKED);
+
+	btrfs_debug_check_extent_io_range(tree, start, end);
+	trace_btrfs_set_extent_bit(tree, start, end - start + 1, bits);
+
+	if (exclusive_bits)
+		ASSERT(failed_start);
+	else
+		ASSERT(failed_start == NULL);
+again:
+	if (!prealloc && gfpflags_allow_blocking(mask)) {
+		/*
+		 * Don't care for allocation failure here because we might end
+		 * up not needing the pre-allocated extent state at all, which
+		 * is the case if we only have in the tree extent states that
+		 * cover our input range and don't cover too any other range.
+		 * If we end up needing a new extent state we allocate it later.
+		 */
+		prealloc = alloc_extent_state(mask);
+	}
+
+	spin_lock(&tree->lock);
+	if (cached_state && *cached_state) {
+		state = *cached_state;
+		if (state->start <= start && state->end > start &&
+		    extent_state_in_tree(state))
+			goto hit_next;
+	}
+	/*
+	 * This search will find all the extents that end after our range
+	 * starts.
+	 */
+	state = tree_search_for_insert(tree, start, &p, &parent);
+	if (!state) {
+		prealloc = alloc_extent_state_atomic(prealloc);
+		BUG_ON(!prealloc);
+		prealloc->start = start;
+		prealloc->end = end;
+		insert_state_fast(tree, prealloc, p, parent, bits, changeset);
+		cache_state(prealloc, cached_state);
+		prealloc = NULL;
+		goto out;
+	}
+hit_next:
+	last_start = state->start;
+	last_end = state->end;
+
+	/*
+	 * | ---- desired range ---- |
+	 * | state |
+	 *
+	 * Just lock what we found and keep going
+	 */
+	if (state->start == start && state->end <= end) {
+		if (state->state & exclusive_bits) {
+			*failed_start = state->start;
+			err = -EEXIST;
+			goto out;
+		}
+
+		set_state_bits(tree, state, bits, changeset);
+		cache_state(state, cached_state);
+		merge_state(tree, state);
+		if (last_end == (u64)-1)
+			goto out;
+		start = last_end + 1;
+		state = next_state(state);
+		if (start < end && state && state->start == start &&
+		    !need_resched())
+			goto hit_next;
+		goto search_again;
+	}
+
+	/*
+	 *     | ---- desired range ---- |
+	 * | state |
+	 *   or
+	 * | ------------- state -------------- |
+	 *
+	 * We need to split the extent we found, and may flip bits on second
+	 * half.
+	 *
+	 * If the extent we found extends past our range, we just split and
+	 * search again.  It'll get split again the next time though.
+	 *
+	 * If the extent we found is inside our range, we set the desired bit
+	 * on it.
+	 */
+	if (state->start < start) {
+		if (state->state & exclusive_bits) {
+			*failed_start = start;
+			err = -EEXIST;
+			goto out;
+		}
+
+		/*
+		 * If this extent already has all the bits we want set, then
+		 * skip it, not necessary to split it or do anything with it.
+		 */
+		if ((state->state & bits) == bits) {
+			start = state->end + 1;
+			cache_state(state, cached_state);
+			goto search_again;
+		}
+
+		prealloc = alloc_extent_state_atomic(prealloc);
+		BUG_ON(!prealloc);
+		err = split_state(tree, state, prealloc, start);
+		if (err)
+			extent_io_tree_panic(tree, err);
+
+		prealloc = NULL;
+		if (err)
+			goto out;
+		if (state->end <= end) {
+			set_state_bits(tree, state, bits, changeset);
+			cache_state(state, cached_state);
+			merge_state(tree, state);
+			if (last_end == (u64)-1)
+				goto out;
+			start = last_end + 1;
+			state = next_state(state);
+			if (start < end && state && state->start == start &&
+			    !need_resched())
+				goto hit_next;
+		}
+		goto search_again;
+	}
+	/*
+	 * | ---- desired range ---- |
+	 *     | state | or               | state |
+	 *
+	 * There's a hole, we need to insert something in it and ignore the
+	 * extent we found.
+	 */
+	if (state->start > start) {
+		u64 this_end;
+		if (end < last_start)
+			this_end = end;
+		else
+			this_end = last_start - 1;
+
+		prealloc = alloc_extent_state_atomic(prealloc);
+		BUG_ON(!prealloc);
+
+		/*
+		 * Avoid to free 'prealloc' if it can be merged with the later
+		 * extent.
+		 */
+		prealloc->start = start;
+		prealloc->end = this_end;
+		err = insert_state(tree, prealloc, bits, changeset);
+		if (err)
+			extent_io_tree_panic(tree, err);
+
+		cache_state(prealloc, cached_state);
+		prealloc = NULL;
+		start = this_end + 1;
+		goto search_again;
+	}
+	/*
+	 * | ---- desired range ---- |
+	 *                        | state |
+	 *
+	 * We need to split the extent, and set the bit on the first half
+	 */
+	if (state->start <= end && state->end > end) {
+		if (state->state & exclusive_bits) {
+			*failed_start = start;
+			err = -EEXIST;
+			goto out;
+		}
+
+		prealloc = alloc_extent_state_atomic(prealloc);
+		BUG_ON(!prealloc);
+		err = split_state(tree, state, prealloc, end + 1);
+		if (err)
+			extent_io_tree_panic(tree, err);
+
+		set_state_bits(tree, prealloc, bits, changeset);
+		cache_state(prealloc, cached_state);
+		merge_state(tree, prealloc);
+		prealloc = NULL;
+		goto out;
+	}
+
+search_again:
+	if (start > end)
+		goto out;
+	spin_unlock(&tree->lock);
+	if (gfpflags_allow_blocking(mask))
+		cond_resched();
+	goto again;
+
+out:
+	spin_unlock(&tree->lock);
+	if (prealloc)
+		free_extent_state(prealloc);
+
+	return err;
+
+}
+
+int set_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
+		   u32 bits, struct extent_state **cached_state, gfp_t mask)
+{
+	return __set_extent_bit(tree, start, end, bits, NULL, cached_state,
+				NULL, mask);
+}
+
+/*
+ * Convert all bits in a given range from one bit to another
+ *
+ * @tree:	the io tree to search
+ * @start:	the start offset in bytes
+ * @end:	the end offset in bytes (inclusive)
+ * @bits:	the bits to set in this range
+ * @clear_bits:	the bits to clear in this range
+ * @cached_state:	state that we're going to cache
+ *
+ * This will go through and set bits for the given range.  If any states exist
+ * already in this range they are set with the given bit and cleared of the
+ * clear_bits.  This is only meant to be used by things that are mergeable, ie.
+ * converting from say DELALLOC to DIRTY.  This is not meant to be used with
+ * boundary bits like LOCK.
+ *
+ * All allocations are done with GFP_NOFS.
+ */
+int convert_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
+		       u32 bits, u32 clear_bits,
+		       struct extent_state **cached_state)
+{
+	struct extent_state *state;
+	struct extent_state *prealloc = NULL;
+	struct rb_node **p;
+	struct rb_node *parent;
+	int err = 0;
+	u64 last_start;
+	u64 last_end;
+	bool first_iteration = true;
+
+	btrfs_debug_check_extent_io_range(tree, start, end);
+	trace_btrfs_convert_extent_bit(tree, start, end - start + 1, bits,
+				       clear_bits);
+
+again:
+	if (!prealloc) {
+		/*
+		 * Best effort, don't worry if extent state allocation fails
+		 * here for the first iteration. We might have a cached state
+		 * that matches exactly the target range, in which case no
+		 * extent state allocations are needed. We'll only know this
+		 * after locking the tree.
+		 */
+		prealloc = alloc_extent_state(GFP_NOFS);
+		if (!prealloc && !first_iteration)
+			return -ENOMEM;
+	}
+
+	spin_lock(&tree->lock);
+	if (cached_state && *cached_state) {
+		state = *cached_state;
+		if (state->start <= start && state->end > start &&
+		    extent_state_in_tree(state))
+			goto hit_next;
+	}
+
+	/*
+	 * This search will find all the extents that end after our range
+	 * starts.
+	 */
+	state = tree_search_for_insert(tree, start, &p, &parent);
+	if (!state) {
+		prealloc = alloc_extent_state_atomic(prealloc);
+		if (!prealloc) {
+			err = -ENOMEM;
+			goto out;
+		}
+		prealloc->start = start;
+		prealloc->end = end;
+		insert_state_fast(tree, prealloc, p, parent, bits, NULL);
+		cache_state(prealloc, cached_state);
+		prealloc = NULL;
+		goto out;
+	}
+hit_next:
+	last_start = state->start;
+	last_end = state->end;
+
+	/*
+	 * | ---- desired range ---- |
+	 * | state |
+	 *
+	 * Just lock what we found and keep going.
+	 */
+	if (state->start == start && state->end <= end) {
+		set_state_bits(tree, state, bits, NULL);
+		cache_state(state, cached_state);
+		state = clear_state_bit(tree, state, clear_bits, 0, NULL);
+		if (last_end == (u64)-1)
+			goto out;
+		start = last_end + 1;
+		if (start < end && state && state->start == start &&
+		    !need_resched())
+			goto hit_next;
+		goto search_again;
+	}
+
+	/*
+	 *     | ---- desired range ---- |
+	 * | state |
+	 *   or
+	 * | ------------- state -------------- |
+	 *
+	 * We need to split the extent we found, and may flip bits on second
+	 * half.
+	 *
+	 * If the extent we found extends past our range, we just split and
+	 * search again.  It'll get split again the next time though.
+	 *
+	 * If the extent we found is inside our range, we set the desired bit
+	 * on it.
+	 */
+	if (state->start < start) {
+		prealloc = alloc_extent_state_atomic(prealloc);
+		if (!prealloc) {
+			err = -ENOMEM;
+			goto out;
+		}
+		err = split_state(tree, state, prealloc, start);
+		if (err)
+			extent_io_tree_panic(tree, err);
+		prealloc = NULL;
+		if (err)
+			goto out;
+		if (state->end <= end) {
+			set_state_bits(tree, state, bits, NULL);
+			cache_state(state, cached_state);
+			state = clear_state_bit(tree, state, clear_bits, 0, NULL);
+			if (last_end == (u64)-1)
+				goto out;
+			start = last_end + 1;
+			if (start < end && state && state->start == start &&
+			    !need_resched())
+				goto hit_next;
+		}
+		goto search_again;
+	}
+	/*
+	 * | ---- desired range ---- |
+	 *     | state | or               | state |
+	 *
+	 * There's a hole, we need to insert something in it and ignore the
+	 * extent we found.
+	 */
+	if (state->start > start) {
+		u64 this_end;
+		if (end < last_start)
+			this_end = end;
+		else
+			this_end = last_start - 1;
+
+		prealloc = alloc_extent_state_atomic(prealloc);
+		if (!prealloc) {
+			err = -ENOMEM;
+			goto out;
+		}
+
+		/*
+		 * Avoid to free 'prealloc' if it can be merged with the later
+		 * extent.
+		 */
+		prealloc->start = start;
+		prealloc->end = this_end;
+		err = insert_state(tree, prealloc, bits, NULL);
+		if (err)
+			extent_io_tree_panic(tree, err);
+		cache_state(prealloc, cached_state);
+		prealloc = NULL;
+		start = this_end + 1;
+		goto search_again;
+	}
+	/*
+	 * | ---- desired range ---- |
+	 *                        | state |
+	 *
+	 * We need to split the extent, and set the bit on the first half.
+	 */
+	if (state->start <= end && state->end > end) {
+		prealloc = alloc_extent_state_atomic(prealloc);
+		if (!prealloc) {
+			err = -ENOMEM;
+			goto out;
+		}
+
+		err = split_state(tree, state, prealloc, end + 1);
+		if (err)
+			extent_io_tree_panic(tree, err);
+
+		set_state_bits(tree, prealloc, bits, NULL);
+		cache_state(prealloc, cached_state);
+		clear_state_bit(tree, prealloc, clear_bits, 0, NULL);
+		prealloc = NULL;
+		goto out;
+	}
+
+search_again:
+	if (start > end)
+		goto out;
+	spin_unlock(&tree->lock);
+	cond_resched();
+	first_iteration = false;
+	goto again;
+
+out:
+	spin_unlock(&tree->lock);
+	if (prealloc)
+		free_extent_state(prealloc);
+
+	return err;
+}
+
+/*
+ * Find the first range that has @bits not set. This range could start before
+ * @start.
+ *
+ * @tree:      the tree to search
+ * @start:     offset at/after which the found extent should start
+ * @start_ret: records the beginning of the range
+ * @end_ret:   records the end of the range (inclusive)
+ * @bits:      the set of bits which must be unset
+ *
+ * Since unallocated range is also considered one which doesn't have the bits
+ * set it's possible that @end_ret contains -1, this happens in case the range
+ * spans (last_range_end, end of device]. In this case it's up to the caller to
+ * trim @end_ret to the appropriate size.
+ */
+void find_first_clear_extent_bit(struct extent_io_tree *tree, u64 start,
+				 u64 *start_ret, u64 *end_ret, u32 bits)
+{
+	struct extent_state *state;
+	struct extent_state *prev = NULL, *next;
+
+	spin_lock(&tree->lock);
+
+	/* Find first extent with bits cleared */
+	while (1) {
+		state = tree_search_prev_next(tree, start, &prev, &next);
+		if (!state && !next && !prev) {
+			/*
+			 * Tree is completely empty, send full range and let
+			 * caller deal with it
+			 */
+			*start_ret = 0;
+			*end_ret = -1;
+			goto out;
+		} else if (!state && !next) {
+			/*
+			 * We are past the last allocated chunk, set start at
+			 * the end of the last extent.
+			 */
+			*start_ret = prev->end + 1;
+			*end_ret = -1;
+			goto out;
+		} else if (!state) {
+			state = next;
+		}
+
+		/*
+		 * At this point 'state' either contains 'start' or start is
+		 * before 'state'
+		 */
+		if (in_range(start, state->start, state->end - state->start + 1)) {
+			if (state->state & bits) {
+				/*
+				 * |--range with bits sets--|
+				 *    |
+				 *    start
+				 */
+				start = state->end + 1;
+			} else {
+				/*
+				 * 'start' falls within a range that doesn't
+				 * have the bits set, so take its start as the
+				 * beginning of the desired range
+				 *
+				 * |--range with bits cleared----|
+				 *      |
+				 *      start
+				 */
+				*start_ret = state->start;
+				break;
+			}
+		} else {
+			/*
+			 * |---prev range---|---hole/unset---|---node range---|
+			 *                          |
+			 *                        start
+			 *
+			 *                        or
+			 *
+			 * |---hole/unset--||--first node--|
+			 * 0   |
+			 *    start
+			 */
+			if (prev)
+				*start_ret = prev->end + 1;
+			else
+				*start_ret = 0;
+			break;
+		}
+	}
+
+	/*
+	 * Find the longest stretch from start until an entry which has the
+	 * bits set
+	 */
+	while (state) {
+		if (state->end >= start && !(state->state & bits)) {
+			*end_ret = state->end;
+		} else {
+			*end_ret = state->start - 1;
+			break;
+		}
+		state = next_state(state);
+	}
+out:
+	spin_unlock(&tree->lock);
+}
+
+/*
+ * Count the number of bytes in the tree that have a given bit(s) set.  This
+ * can be fairly slow, except for EXTENT_DIRTY which is cached.  The total
+ * number found is returned.
+ */
+u64 count_range_bits(struct extent_io_tree *tree,
+		     u64 *start, u64 search_end, u64 max_bytes,
+		     u32 bits, int contig)
+{
+	struct extent_state *state;
+	u64 cur_start = *start;
+	u64 total_bytes = 0;
+	u64 last = 0;
+	int found = 0;
+
+	if (WARN_ON(search_end <= cur_start))
+		return 0;
+
+	spin_lock(&tree->lock);
+
+	/*
+	 * This search will find all the extents that end after our range
+	 * starts.
+	 */
+	state = tree_search(tree, cur_start);
+	while (state) {
+		if (state->start > search_end)
+			break;
+		if (contig && found && state->start > last + 1)
+			break;
+		if (state->end >= cur_start && (state->state & bits) == bits) {
+			total_bytes += min(search_end, state->end) + 1 -
+				       max(cur_start, state->start);
+			if (total_bytes >= max_bytes)
+				break;
+			if (!found) {
+				*start = max(cur_start, state->start);
+				found = 1;
+			}
+			last = state->end;
+		} else if (contig && found) {
+			break;
+		}
+		state = next_state(state);
+	}
+	spin_unlock(&tree->lock);
+	return total_bytes;
+}
+
+/*
+ * Searche a range in the state tree for a given mask.  If 'filled' == 1, this
+ * returns 1 only if every extent in the tree has the bits set.  Otherwise, 1
+ * is returned if any bit in the range is found set.
+ */
+int test_range_bit(struct extent_io_tree *tree, u64 start, u64 end,
+		   u32 bits, int filled, struct extent_state *cached)
+{
+	struct extent_state *state = NULL;
+	int bitset = 0;
+
+	spin_lock(&tree->lock);
+	if (cached && extent_state_in_tree(cached) && cached->start <= start &&
+	    cached->end > start)
+		state = cached;
+	else
+		state = tree_search(tree, start);
+	while (state && start <= end) {
+		if (filled && state->start > start) {
+			bitset = 0;
+			break;
+		}
+
+		if (state->start > end)
+			break;
+
+		if (state->state & bits) {
+			bitset = 1;
+			if (!filled)
+				break;
+		} else if (filled) {
+			bitset = 0;
+			break;
+		}
+
+		if (state->end == (u64)-1)
+			break;
+
+		start = state->end + 1;
+		if (start > end)
+			break;
+		state = next_state(state);
+	}
+
+	/* We ran out of states and were still inside of our range. */
+	if (filled && !state)
+		bitset = 0;
+	spin_unlock(&tree->lock);
+	return bitset;
+}
+
+/* Wrappers around set/clear extent bit */
+int set_record_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
+			   u32 bits, struct extent_changeset *changeset)
+{
+	/*
+	 * We don't support EXTENT_LOCKED yet, as current changeset will
+	 * record any bits changed, so for EXTENT_LOCKED case, it will
+	 * either fail with -EEXIST or changeset will record the whole
+	 * range.
+	 */
+	ASSERT(!(bits & EXTENT_LOCKED));
+
+	return __set_extent_bit(tree, start, end, bits, NULL, NULL, changeset,
+				GFP_NOFS);
+}
+
+int clear_record_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
+			     u32 bits, struct extent_changeset *changeset)
+{
+	/*
+	 * Don't support EXTENT_LOCKED case, same reason as
+	 * set_record_extent_bits().
+	 */
+	ASSERT(!(bits & EXTENT_LOCKED));
+
+	return __clear_extent_bit(tree, start, end, bits, NULL, GFP_NOFS,
+				  changeset);
+}
+
+int try_lock_extent(struct extent_io_tree *tree, u64 start, u64 end)
+{
+	int err;
+	u64 failed_start;
+
+	err = __set_extent_bit(tree, start, end, EXTENT_LOCKED, &failed_start,
+			       NULL, NULL, GFP_NOFS);
+	if (err == -EEXIST) {
+		if (failed_start > start)
+			clear_extent_bit(tree, start, failed_start - 1,
+					 EXTENT_LOCKED, NULL);
+		return 0;
+	}
+	return 1;
+}
+
+/*
+ * Either insert or lock state struct between start and end use mask to tell
+ * us if waiting is desired.
+ */
+int lock_extent(struct extent_io_tree *tree, u64 start, u64 end,
+		struct extent_state **cached_state)
+{
+	int err;
+	u64 failed_start;
+
+	err = __set_extent_bit(tree, start, end, EXTENT_LOCKED, &failed_start,
+			       cached_state, NULL, GFP_NOFS);
+	while (err == -EEXIST) {
+		if (failed_start != start)
+			clear_extent_bit(tree, start, failed_start - 1,
+					 EXTENT_LOCKED, cached_state);
+
+		wait_extent_bit(tree, failed_start, end, EXTENT_LOCKED);
+		err = __set_extent_bit(tree, start, end, EXTENT_LOCKED,
+				       &failed_start, cached_state, NULL,
+				       GFP_NOFS);
+	}
+	return err;
+}
+
+void __cold extent_state_free_cachep(void)
+{
+	btrfs_extent_state_leak_debug_check();
+	kmem_cache_destroy(extent_state_cache);
+}
+
+int __init extent_state_init_cachep(void)
+{
+	extent_state_cache = kmem_cache_create("btrfs_extent_state",
+			sizeof(struct extent_state), 0,
+			SLAB_MEM_SPREAD, NULL);
+	if (!extent_state_cache)
+		return -ENOMEM;
+
+	return 0;
+}
diff --git a/fs/btrfs/extent-io-tree.h b/fs/btrfs/extent-io-tree.h
index a3febe746c79..a855f40dd61d 100644
--- a/fs/btrfs/extent-io-tree.h
+++ b/fs/btrfs/extent-io-tree.h
@@ -17,14 +17,35 @@ struct io_failure_record;
 #define EXTENT_NODATASUM	(1U << 7)
 #define EXTENT_CLEAR_META_RESV	(1U << 8)
 #define EXTENT_NEED_WAIT	(1U << 9)
-#define EXTENT_DAMAGED		(1U << 10)
 #define EXTENT_NORESERVE	(1U << 11)
 #define EXTENT_QGROUP_RESERVED	(1U << 12)
 #define EXTENT_CLEAR_DATA_RESV	(1U << 13)
+/*
+ * Must be cleared only during ordered extent completion or on error paths if we
+ * did not manage to submit bios and create the ordered extents for the range.
+ * Should not be cleared during page release and page invalidation (if there is
+ * an ordered extent in flight), that is left for the ordered extent completion.
+ */
 #define EXTENT_DELALLOC_NEW	(1U << 14)
+/*
+ * When an ordered extent successfully completes for a region marked as a new
+ * delalloc range, use this flag when clearing a new delalloc range to indicate
+ * that the VFS' inode number of bytes should be incremented and the inode's new
+ * delalloc bytes decremented, in an atomic way to prevent races with stat(2).
+ */
+#define EXTENT_ADD_INODE_BYTES  (1U << 15)
+
+/*
+ * Set during truncate when we're clearing an entire range and we just want the
+ * extent states to go away.
+ */
+#define EXTENT_CLEAR_ALL_BITS	(1U << 16)
+
 #define EXTENT_DO_ACCOUNTING    (EXTENT_CLEAR_META_RESV | \
 				 EXTENT_CLEAR_DATA_RESV)
-#define EXTENT_CTLBITS		(EXTENT_DO_ACCOUNTING)
+#define EXTENT_CTLBITS		(EXTENT_DO_ACCOUNTING | \
+				 EXTENT_ADD_INODE_BYTES | \
+				 EXTENT_CLEAR_ALL_BITS)
 
 /*
  * Redefined bits above which are used only in the device allocation tree,
@@ -34,30 +55,32 @@ struct io_failure_record;
  */
 #define CHUNK_ALLOCATED				EXTENT_DIRTY
 #define CHUNK_TRIMMED				EXTENT_DEFRAG
+#define CHUNK_STATE_MASK			(CHUNK_ALLOCATED |		\
+						 CHUNK_TRIMMED)
 
 enum {
-	IO_TREE_FS_INFO_FREED_EXTENTS0,
-	IO_TREE_FS_INFO_FREED_EXTENTS1,
+	IO_TREE_FS_PINNED_EXTENTS,
+	IO_TREE_FS_EXCLUDED_EXTENTS,
+	IO_TREE_BTREE_INODE_IO,
 	IO_TREE_INODE_IO,
-	IO_TREE_INODE_IO_FAILURE,
 	IO_TREE_RELOC_BLOCKS,
 	IO_TREE_TRANS_DIRTY_PAGES,
 	IO_TREE_ROOT_DIRTY_LOG_PAGES,
+	IO_TREE_INODE_FILE_EXTENT,
+	IO_TREE_LOG_CSUM_RANGE,
 	IO_TREE_SELFTEST,
+	IO_TREE_DEVICE_ALLOC_STATE,
 };
 
 struct extent_io_tree {
 	struct rb_root state;
 	struct btrfs_fs_info *fs_info;
 	void *private_data;
-	u64 dirty_bytes;
-	bool track_uptodate;
 
 	/* Who owns this io tree, should be one of IO_TREE_* */
 	u8 owner;
 
 	spinlock_t lock;
-	const struct extent_io_ops *ops;
 };
 
 struct extent_state {
@@ -68,110 +91,95 @@ struct extent_state {
 	/* ADD NEW ELEMENTS AFTER THIS */
 	wait_queue_head_t wq;
 	refcount_t refs;
-	unsigned state;
-
-	struct io_failure_record *failrec;
+	u32 state;
 
 #ifdef CONFIG_BTRFS_DEBUG
 	struct list_head leak_list;
 #endif
 };
 
-int __init extent_state_cache_init(void);
-void __cold extent_state_cache_exit(void);
-
 void extent_io_tree_init(struct btrfs_fs_info *fs_info,
 			 struct extent_io_tree *tree, unsigned int owner,
 			 void *private_data);
 void extent_io_tree_release(struct extent_io_tree *tree);
 
-int lock_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
-		     struct extent_state **cached);
-
-static inline int lock_extent(struct extent_io_tree *tree, u64 start, u64 end)
-{
-	return lock_extent_bits(tree, start, end, NULL);
-}
+int lock_extent(struct extent_io_tree *tree, u64 start, u64 end,
+		struct extent_state **cached);
 
 int try_lock_extent(struct extent_io_tree *tree, u64 start, u64 end);
 
-int __init extent_io_init(void);
-void __cold extent_io_exit(void);
+int __init extent_state_init_cachep(void);
+void __cold extent_state_free_cachep(void);
 
 u64 count_range_bits(struct extent_io_tree *tree,
 		     u64 *start, u64 search_end,
-		     u64 max_bytes, unsigned bits, int contig);
+		     u64 max_bytes, u32 bits, int contig);
 
 void free_extent_state(struct extent_state *state);
 int test_range_bit(struct extent_io_tree *tree, u64 start, u64 end,
-		   unsigned bits, int filled,
-		   struct extent_state *cached_state);
+		   u32 bits, int filled, struct extent_state *cached_state);
 int clear_record_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
-		unsigned bits, struct extent_changeset *changeset);
-int clear_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
-		     unsigned bits, int wake, int delete,
-		     struct extent_state **cached);
+			     u32 bits, struct extent_changeset *changeset);
 int __clear_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
-		     unsigned bits, int wake, int delete,
-		     struct extent_state **cached, gfp_t mask,
-		     struct extent_changeset *changeset);
+		       u32 bits, struct extent_state **cached, gfp_t mask,
+		       struct extent_changeset *changeset);
 
-static inline int unlock_extent(struct extent_io_tree *tree, u64 start, u64 end)
+static inline int clear_extent_bit(struct extent_io_tree *tree, u64 start,
+				   u64 end, u32 bits,
+				   struct extent_state **cached)
 {
-	return clear_extent_bit(tree, start, end, EXTENT_LOCKED, 1, 0, NULL);
+	return __clear_extent_bit(tree, start, end, bits, cached,
+				  GFP_NOFS, NULL);
 }
 
-static inline int unlock_extent_cached(struct extent_io_tree *tree, u64 start,
-		u64 end, struct extent_state **cached)
+static inline int unlock_extent(struct extent_io_tree *tree, u64 start, u64 end,
+				struct extent_state **cached)
 {
-	return __clear_extent_bit(tree, start, end, EXTENT_LOCKED, 1, 0, cached,
-				GFP_NOFS, NULL);
+	return __clear_extent_bit(tree, start, end, EXTENT_LOCKED, cached,
+				  GFP_NOFS, NULL);
 }
 
-static inline int unlock_extent_cached_atomic(struct extent_io_tree *tree,
-		u64 start, u64 end, struct extent_state **cached)
+static inline int unlock_extent_atomic(struct extent_io_tree *tree, u64 start,
+				       u64 end, struct extent_state **cached)
 {
-	return __clear_extent_bit(tree, start, end, EXTENT_LOCKED, 1, 0, cached,
-				GFP_ATOMIC, NULL);
+	return __clear_extent_bit(tree, start, end, EXTENT_LOCKED, cached,
+				  GFP_ATOMIC, NULL);
 }
 
 static inline int clear_extent_bits(struct extent_io_tree *tree, u64 start,
-		u64 end, unsigned bits)
+				    u64 end, u32 bits)
 {
-	int wake = 0;
-
-	if (bits & EXTENT_LOCKED)
-		wake = 1;
-
-	return clear_extent_bit(tree, start, end, bits, wake, 0, NULL);
+	return clear_extent_bit(tree, start, end, bits, NULL);
 }
 
 int set_record_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
-			   unsigned bits, struct extent_changeset *changeset);
+			   u32 bits, struct extent_changeset *changeset);
 int set_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
-		   unsigned bits, u64 *failed_start,
-		   struct extent_state **cached_state, gfp_t mask);
-int set_extent_bits_nowait(struct extent_io_tree *tree, u64 start, u64 end,
-			   unsigned bits);
+		   u32 bits, struct extent_state **cached_state, gfp_t mask);
+
+static inline int set_extent_bits_nowait(struct extent_io_tree *tree, u64 start,
+					 u64 end, u32 bits)
+{
+	return set_extent_bit(tree, start, end, bits, NULL, GFP_NOWAIT);
+}
 
 static inline int set_extent_bits(struct extent_io_tree *tree, u64 start,
-		u64 end, unsigned bits)
+		u64 end, u32 bits)
 {
-	return set_extent_bit(tree, start, end, bits, NULL, NULL, GFP_NOFS);
+	return set_extent_bit(tree, start, end, bits, NULL, GFP_NOFS);
 }
 
 static inline int clear_extent_uptodate(struct extent_io_tree *tree, u64 start,
 		u64 end, struct extent_state **cached_state)
 {
-	return __clear_extent_bit(tree, start, end, EXTENT_UPTODATE, 0, 0,
-				cached_state, GFP_NOFS, NULL);
+	return __clear_extent_bit(tree, start, end, EXTENT_UPTODATE,
+				  cached_state, GFP_NOFS, NULL);
 }
 
 static inline int set_extent_dirty(struct extent_io_tree *tree, u64 start,
 		u64 end, gfp_t mask)
 {
-	return set_extent_bit(tree, start, end, EXTENT_DIRTY, NULL,
-			      NULL, mask);
+	return set_extent_bit(tree, start, end, EXTENT_DIRTY, NULL, mask);
 }
 
 static inline int clear_extent_dirty(struct extent_io_tree *tree, u64 start,
@@ -179,70 +187,53 @@ static inline int clear_extent_dirty(struct extent_io_tree *tree, u64 start,
 {
 	return clear_extent_bit(tree, start, end,
 				EXTENT_DIRTY | EXTENT_DELALLOC |
-				EXTENT_DO_ACCOUNTING, 0, 0, cached);
+				EXTENT_DO_ACCOUNTING, cached);
 }
 
 int convert_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
-		       unsigned bits, unsigned clear_bits,
+		       u32 bits, u32 clear_bits,
 		       struct extent_state **cached_state);
 
 static inline int set_extent_delalloc(struct extent_io_tree *tree, u64 start,
-				      u64 end, unsigned int extra_bits,
+				      u64 end, u32 extra_bits,
 				      struct extent_state **cached_state)
 {
 	return set_extent_bit(tree, start, end,
-			      EXTENT_DELALLOC | EXTENT_UPTODATE | extra_bits,
-			      NULL, cached_state, GFP_NOFS);
+			      EXTENT_DELALLOC | extra_bits,
+			      cached_state, GFP_NOFS);
 }
 
 static inline int set_extent_defrag(struct extent_io_tree *tree, u64 start,
 		u64 end, struct extent_state **cached_state)
 {
 	return set_extent_bit(tree, start, end,
-			      EXTENT_DELALLOC | EXTENT_UPTODATE | EXTENT_DEFRAG,
-			      NULL, cached_state, GFP_NOFS);
+			      EXTENT_DELALLOC | EXTENT_DEFRAG,
+			      cached_state, GFP_NOFS);
 }
 
 static inline int set_extent_new(struct extent_io_tree *tree, u64 start,
 		u64 end)
 {
-	return set_extent_bit(tree, start, end, EXTENT_NEW, NULL, NULL,
-			GFP_NOFS);
+	return set_extent_bit(tree, start, end, EXTENT_NEW, NULL, GFP_NOFS);
 }
 
 static inline int set_extent_uptodate(struct extent_io_tree *tree, u64 start,
 		u64 end, struct extent_state **cached_state, gfp_t mask)
 {
-	return set_extent_bit(tree, start, end, EXTENT_UPTODATE, NULL,
+	return set_extent_bit(tree, start, end, EXTENT_UPTODATE,
 			      cached_state, mask);
 }
 
 int find_first_extent_bit(struct extent_io_tree *tree, u64 start,
-			  u64 *start_ret, u64 *end_ret, unsigned bits,
+			  u64 *start_ret, u64 *end_ret, u32 bits,
 			  struct extent_state **cached_state);
 void find_first_clear_extent_bit(struct extent_io_tree *tree, u64 start,
-				 u64 *start_ret, u64 *end_ret, unsigned bits);
-int extent_invalidatepage(struct extent_io_tree *tree,
-			  struct page *page, unsigned long offset);
+				 u64 *start_ret, u64 *end_ret, u32 bits);
+int find_contiguous_extent_bit(struct extent_io_tree *tree, u64 start,
+			       u64 *start_ret, u64 *end_ret, u32 bits);
 bool btrfs_find_delalloc_range(struct extent_io_tree *tree, u64 *start,
 			       u64 *end, u64 max_bytes,
 			       struct extent_state **cached_state);
-
-/* This should be reworked in the future and put elsewhere. */
-int get_state_failrec(struct extent_io_tree *tree, u64 start,
-		      struct io_failure_record **failrec);
-int set_state_failrec(struct extent_io_tree *tree, u64 start,
-		      struct io_failure_record *failrec);
-void btrfs_free_io_failure_record(struct btrfs_inode *inode, u64 start,
-		u64 end);
-int btrfs_get_io_failure_record(struct inode *inode, u64 start, u64 end,
-				struct io_failure_record **failrec_ret);
-int free_io_failure(struct extent_io_tree *failure_tree,
-		    struct extent_io_tree *io_tree,
-		    struct io_failure_record *rec);
-int clean_io_failure(struct btrfs_fs_info *fs_info,
-		     struct extent_io_tree *failure_tree,
-		     struct extent_io_tree *io_tree, u64 start,
-		     struct page *page, u64 ino, unsigned int pg_offset);
+void wait_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, u32 bits);
 
 #endif /* BTRFS_EXTENT_IO_TREE_H */
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index a7bc66121330..2801c991814f 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -33,6 +33,9 @@
 #include "delalloc-space.h"
 #include "block-group.h"
 #include "discard.h"
+#include "rcu-string.h"
+#include "zoned.h"
+#include "dev-replace.h"
 
 #undef SCRAMBLE_DELAYED_REFS
 
@@ -64,10 +67,8 @@ int btrfs_add_excluded_extent(struct btrfs_fs_info *fs_info,
 			      u64 start, u64 num_bytes)
 {
 	u64 end = start + num_bytes - 1;
-	set_extent_bits(&fs_info->freed_extents[0],
-			start, end, EXTENT_UPTODATE);
-	set_extent_bits(&fs_info->freed_extents[1],
-			start, end, EXTENT_UPTODATE);
+	set_extent_bits(&fs_info->excluded_extents, start, end,
+			EXTENT_UPTODATE);
 	return 0;
 }
 
@@ -79,50 +80,14 @@ void btrfs_free_excluded_extents(struct btrfs_block_group *cache)
 	start = cache->start;
 	end = start + cache->length - 1;
 
-	clear_extent_bits(&fs_info->freed_extents[0],
-			  start, end, EXTENT_UPTODATE);
-	clear_extent_bits(&fs_info->freed_extents[1],
-			  start, end, EXTENT_UPTODATE);
-}
-
-static u64 generic_ref_to_space_flags(struct btrfs_ref *ref)
-{
-	if (ref->type == BTRFS_REF_METADATA) {
-		if (ref->tree_ref.root == BTRFS_CHUNK_TREE_OBJECTID)
-			return BTRFS_BLOCK_GROUP_SYSTEM;
-		else
-			return BTRFS_BLOCK_GROUP_METADATA;
-	}
-	return BTRFS_BLOCK_GROUP_DATA;
-}
-
-static void add_pinned_bytes(struct btrfs_fs_info *fs_info,
-			     struct btrfs_ref *ref)
-{
-	struct btrfs_space_info *space_info;
-	u64 flags = generic_ref_to_space_flags(ref);
-
-	space_info = btrfs_find_space_info(fs_info, flags);
-	ASSERT(space_info);
-	percpu_counter_add_batch(&space_info->total_bytes_pinned, ref->len,
-		    BTRFS_TOTAL_BYTES_PINNED_BATCH);
-}
-
-static void sub_pinned_bytes(struct btrfs_fs_info *fs_info,
-			     struct btrfs_ref *ref)
-{
-	struct btrfs_space_info *space_info;
-	u64 flags = generic_ref_to_space_flags(ref);
-
-	space_info = btrfs_find_space_info(fs_info, flags);
-	ASSERT(space_info);
-	percpu_counter_add_batch(&space_info->total_bytes_pinned, -ref->len,
-		    BTRFS_TOTAL_BYTES_PINNED_BATCH);
+	clear_extent_bits(&fs_info->excluded_extents, start, end,
+			  EXTENT_UPTODATE);
 }
 
 /* simple helper to search for an existing data extent at a given offset */
 int btrfs_lookup_data_extent(struct btrfs_fs_info *fs_info, u64 start, u64 len)
 {
+	struct btrfs_root *root = btrfs_extent_root(fs_info, start);
 	int ret;
 	struct btrfs_key key;
 	struct btrfs_path *path;
@@ -134,7 +99,7 @@ int btrfs_lookup_data_extent(struct btrfs_fs_info *fs_info, u64 start, u64 len)
 	key.objectid = start;
 	key.offset = len;
 	key.type = BTRFS_EXTENT_ITEM_KEY;
-	ret = btrfs_search_slot(NULL, fs_info->extent_root, &key, path, 0, 0);
+	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
 	btrfs_free_path(path);
 	return ret;
 }
@@ -152,6 +117,7 @@ int btrfs_lookup_extent_info(struct btrfs_trans_handle *trans,
 			     struct btrfs_fs_info *fs_info, u64 bytenr,
 			     u64 offset, int metadata, u64 *refs, u64 *flags)
 {
+	struct btrfs_root *extent_root;
 	struct btrfs_delayed_ref_head *head;
 	struct btrfs_delayed_ref_root *delayed_refs;
 	struct btrfs_path *path;
@@ -189,7 +155,8 @@ search_again:
 	else
 		key.type = BTRFS_EXTENT_ITEM_KEY;
 
-	ret = btrfs_search_slot(trans, fs_info->extent_root, &key, path, 0, 0);
+	extent_root = btrfs_extent_root(fs_info, bytenr);
+	ret = btrfs_search_slot(NULL, extent_root, &key, path, 0, 0);
 	if (ret < 0)
 		goto out_free;
 
@@ -207,7 +174,7 @@ search_again:
 
 	if (ret == 0) {
 		leaf = path->nodes[0];
-		item_size = btrfs_item_size_nr(leaf, path->slots[0]);
+		item_size = btrfs_item_size(leaf, path->slots[0]);
 		if (item_size >= sizeof(*ei)) {
 			ei = btrfs_item_ptr(leaf, path->slots[0],
 					    struct btrfs_extent_item);
@@ -403,12 +370,11 @@ int btrfs_get_extent_inline_ref_type(const struct extent_buffer *eb,
 			if (type == BTRFS_SHARED_BLOCK_REF_KEY) {
 				ASSERT(eb->fs_info);
 				/*
-				 * Every shared one has parent tree
-				 * block, which must be aligned to
-				 * nodesize.
+				 * Every shared one has parent tree block,
+				 * which must be aligned to sector size.
 				 */
 				if (offset &&
-				    IS_ALIGNED(offset, eb->fs_info->nodesize))
+				    IS_ALIGNED(offset, eb->fs_info->sectorsize))
 					return type;
 			}
 		} else if (is_data == BTRFS_REF_TYPE_DATA) {
@@ -417,12 +383,11 @@ int btrfs_get_extent_inline_ref_type(const struct extent_buffer *eb,
 			if (type == BTRFS_SHARED_DATA_REF_KEY) {
 				ASSERT(eb->fs_info);
 				/*
-				 * Every shared one has parent tree
-				 * block, which must be aligned to
-				 * nodesize.
+				 * Every shared one has parent tree block,
+				 * which must be aligned to sector size.
 				 */
 				if (offset &&
-				    IS_ALIGNED(offset, eb->fs_info->nodesize))
+				    IS_ALIGNED(offset, eb->fs_info->sectorsize))
 					return type;
 			}
 		} else {
@@ -432,8 +397,9 @@ int btrfs_get_extent_inline_ref_type(const struct extent_buffer *eb,
 	}
 
 	btrfs_print_leaf((struct extent_buffer *)eb);
-	btrfs_err(eb->fs_info, "eb %llu invalid extent inline ref type %d",
-		  eb->start, type);
+	btrfs_err(eb->fs_info,
+		  "eb %llu iref 0x%lx invalid extent inline ref type %d",
+		  eb->start, (unsigned long)iref, type);
 	WARN_ON(1);
 
 	return BTRFS_REF_TYPE_INVALID;
@@ -480,7 +446,7 @@ static noinline int lookup_extent_data_ref(struct btrfs_trans_handle *trans,
 					   u64 root_objectid,
 					   u64 owner, u64 offset)
 {
-	struct btrfs_root *root = trans->fs_info->extent_root;
+	struct btrfs_root *root = btrfs_extent_root(trans->fs_info, bytenr);
 	struct btrfs_key key;
 	struct btrfs_extent_data_ref *ref;
 	struct extent_buffer *leaf;
@@ -556,7 +522,7 @@ static noinline int insert_extent_data_ref(struct btrfs_trans_handle *trans,
 					   u64 root_objectid, u64 owner,
 					   u64 offset, int refs_to_add)
 {
-	struct btrfs_root *root = trans->fs_info->extent_root;
+	struct btrfs_root *root = btrfs_extent_root(trans->fs_info, bytenr);
 	struct btrfs_key key;
 	struct extent_buffer *leaf;
 	u32 size;
@@ -630,8 +596,9 @@ fail:
 }
 
 static noinline int remove_extent_data_ref(struct btrfs_trans_handle *trans,
+					   struct btrfs_root *root,
 					   struct btrfs_path *path,
-					   int refs_to_drop, int *last_ref)
+					   int refs_to_drop)
 {
 	struct btrfs_key key;
 	struct btrfs_extent_data_ref *ref1 = NULL;
@@ -663,8 +630,7 @@ static noinline int remove_extent_data_ref(struct btrfs_trans_handle *trans,
 	num_refs -= refs_to_drop;
 
 	if (num_refs == 0) {
-		ret = btrfs_del_item(trans, trans->fs_info->extent_root, path);
-		*last_ref = 1;
+		ret = btrfs_del_item(trans, root, path);
 	} else {
 		if (key.type == BTRFS_EXTENT_DATA_REF_KEY)
 			btrfs_set_extent_data_ref_count(leaf, ref1, num_refs);
@@ -722,7 +688,7 @@ static noinline int lookup_tree_block_ref(struct btrfs_trans_handle *trans,
 					  u64 bytenr, u64 parent,
 					  u64 root_objectid)
 {
-	struct btrfs_root *root = trans->fs_info->extent_root;
+	struct btrfs_root *root = btrfs_extent_root(trans->fs_info, bytenr);
 	struct btrfs_key key;
 	int ret;
 
@@ -746,6 +712,7 @@ static noinline int insert_tree_block_ref(struct btrfs_trans_handle *trans,
 					  u64 bytenr, u64 parent,
 					  u64 root_objectid)
 {
+	struct btrfs_root *root = btrfs_extent_root(trans->fs_info, bytenr);
 	struct btrfs_key key;
 	int ret;
 
@@ -758,8 +725,7 @@ static noinline int insert_tree_block_ref(struct btrfs_trans_handle *trans,
 		key.offset = root_objectid;
 	}
 
-	ret = btrfs_insert_empty_item(trans, trans->fs_info->extent_root,
-				      path, &key, 0);
+	ret = btrfs_insert_empty_item(trans, root, path, &key, 0);
 	btrfs_release_path(path);
 	return ret;
 }
@@ -824,7 +790,7 @@ int lookup_inline_extent_backref(struct btrfs_trans_handle *trans,
 				 u64 owner, u64 offset, int insert)
 {
 	struct btrfs_fs_info *fs_info = trans->fs_info;
-	struct btrfs_root *root = fs_info->extent_root;
+	struct btrfs_root *root = btrfs_extent_root(fs_info, bytenr);
 	struct btrfs_key key;
 	struct extent_buffer *leaf;
 	struct btrfs_extent_item *ei;
@@ -848,6 +814,7 @@ int lookup_inline_extent_backref(struct btrfs_trans_handle *trans,
 	want = extent_ref_type(parent, owner);
 	if (insert) {
 		extra_size = btrfs_extent_inline_ref_size(want);
+		path->search_for_extension = 1;
 		path->keep_locks = 1;
 	} else
 		extra_size = -1;
@@ -901,7 +868,7 @@ again:
 	}
 
 	leaf = path->nodes[0];
-	item_size = btrfs_item_size_nr(leaf, path->slots[0]);
+	item_size = btrfs_item_size(leaf, path->slots[0]);
 	if (unlikely(item_size < sizeof(*ei))) {
 		err = -EINVAL;
 		btrfs_print_v0_err(fs_info);
@@ -928,7 +895,13 @@ again:
 	err = -ENOENT;
 	while (1) {
 		if (ptr >= end) {
-			WARN_ON(ptr > end);
+			if (ptr > end) {
+				err = -EUCLEAN;
+				btrfs_print_leaf(path->nodes[0]);
+				btrfs_crit(fs_info,
+"overrun extent record at slot %d while looking for inline extent for root %llu owner %llu offset %llu parent %llu",
+					path->slots[0], root_objectid, owner, offset, parent);
+			}
 			break;
 		}
 		iref = (struct btrfs_extent_inline_ref *)ptr;
@@ -1000,6 +973,7 @@ again:
 out:
 	if (insert) {
 		path->keep_locks = 0;
+		path->search_for_extension = 0;
 		btrfs_unlock_up_safe(path, 1);
 	}
 	return err;
@@ -1042,7 +1016,7 @@ void setup_inline_extent_backref(struct btrfs_fs_info *fs_info,
 		__run_delayed_extent_op(extent_op, leaf, ei);
 
 	ptr = (unsigned long)ei + item_offset;
-	end = (unsigned long)ei + btrfs_item_size_nr(leaf, path->slots[0]);
+	end = (unsigned long)ei + btrfs_item_size(leaf, path->slots[0]);
 	if (ptr < end - size)
 		memmove_extent_buffer(leaf, ptr + size, ptr,
 				      end - size - ptr);
@@ -1103,8 +1077,7 @@ static noinline_for_stack
 void update_inline_extent_backref(struct btrfs_path *path,
 				  struct btrfs_extent_inline_ref *iref,
 				  int refs_to_mod,
-				  struct btrfs_delayed_extent_op *extent_op,
-				  int *last_ref)
+				  struct btrfs_delayed_extent_op *extent_op)
 {
 	struct extent_buffer *leaf = path->nodes[0];
 	struct btrfs_extent_item *ei;
@@ -1152,9 +1125,8 @@ void update_inline_extent_backref(struct btrfs_path *path,
 		else
 			btrfs_set_shared_data_ref_count(leaf, sref, refs);
 	} else {
-		*last_ref = 1;
 		size =  btrfs_extent_inline_ref_size(type);
-		item_size = btrfs_item_size_nr(leaf, path->slots[0]);
+		item_size = btrfs_item_size(leaf, path->slots[0]);
 		ptr = (unsigned long)iref;
 		end = (unsigned long)ei + item_size;
 		if (ptr + size < end)
@@ -1181,9 +1153,23 @@ int insert_inline_extent_backref(struct btrfs_trans_handle *trans,
 					   num_bytes, parent, root_objectid,
 					   owner, offset, 1);
 	if (ret == 0) {
-		BUG_ON(owner < BTRFS_FIRST_FREE_OBJECTID);
-		update_inline_extent_backref(path, iref, refs_to_add,
-					     extent_op, NULL);
+		/*
+		 * We're adding refs to a tree block we already own, this
+		 * should not happen at all.
+		 */
+		if (owner < BTRFS_FIRST_FREE_OBJECTID) {
+			btrfs_crit(trans->fs_info,
+"adding refs to an existing tree ref, bytenr %llu num_bytes %llu root_objectid %llu",
+				   bytenr, num_bytes, root_objectid);
+			if (IS_ENABLED(CONFIG_BTRFS_DEBUG)) {
+				WARN_ON(1);
+				btrfs_crit(trans->fs_info,
+			"path->slots[0]=%d path->nodes[0]:", path->slots[0]);
+				btrfs_print_leaf(path->nodes[0]);
+			}
+			return -EUCLEAN;
+		}
+		update_inline_extent_backref(path, iref, refs_to_add, extent_op);
 	} else if (ret == -ENOENT) {
 		setup_inline_extent_backref(trans->fs_info, path, iref, parent,
 					    root_objectid, owner, offset,
@@ -1193,42 +1179,21 @@ int insert_inline_extent_backref(struct btrfs_trans_handle *trans,
 	return ret;
 }
 
-static int insert_extent_backref(struct btrfs_trans_handle *trans,
-				 struct btrfs_path *path,
-				 u64 bytenr, u64 parent, u64 root_objectid,
-				 u64 owner, u64 offset, int refs_to_add)
-{
-	int ret;
-	if (owner < BTRFS_FIRST_FREE_OBJECTID) {
-		BUG_ON(refs_to_add != 1);
-		ret = insert_tree_block_ref(trans, path, bytenr, parent,
-					    root_objectid);
-	} else {
-		ret = insert_extent_data_ref(trans, path, bytenr, parent,
-					     root_objectid, owner, offset,
-					     refs_to_add);
-	}
-	return ret;
-}
-
 static int remove_extent_backref(struct btrfs_trans_handle *trans,
+				 struct btrfs_root *root,
 				 struct btrfs_path *path,
 				 struct btrfs_extent_inline_ref *iref,
-				 int refs_to_drop, int is_data, int *last_ref)
+				 int refs_to_drop, int is_data)
 {
 	int ret = 0;
 
 	BUG_ON(!is_data && refs_to_drop != 1);
-	if (iref) {
-		update_inline_extent_backref(path, iref, -refs_to_drop, NULL,
-					     last_ref);
-	} else if (is_data) {
-		ret = remove_extent_data_ref(trans, path, refs_to_drop,
-					     last_ref);
-	} else {
-		*last_ref = 1;
-		ret = btrfs_del_item(trans, trans->fs_info->extent_root, path);
-	}
+	if (iref)
+		update_inline_extent_backref(path, iref, -refs_to_drop, NULL);
+	else if (is_data)
+		ret = remove_extent_data_ref(trans, root, path, refs_to_drop);
+	else
+		ret = btrfs_del_item(trans, root, path);
 	return ret;
 }
 
@@ -1280,7 +1245,7 @@ static int btrfs_issue_discard(struct block_device *bdev, u64 start, u64 len,
 
 		if (size) {
 			ret = blkdev_issue_discard(bdev, start >> 9, size >> 9,
-						   GFP_NOFS, 0);
+						   GFP_NOFS);
 			if (!ret)
 				*discarded_bytes += size;
 			else if (ret != -EOPNOTSUPP)
@@ -1297,13 +1262,53 @@ static int btrfs_issue_discard(struct block_device *bdev, u64 start, u64 len,
 
 	if (bytes_left) {
 		ret = blkdev_issue_discard(bdev, start >> 9, bytes_left >> 9,
-					   GFP_NOFS, 0);
+					   GFP_NOFS);
 		if (!ret)
 			*discarded_bytes += bytes_left;
 	}
 	return ret;
 }
 
+static int do_discard_extent(struct btrfs_discard_stripe *stripe, u64 *bytes)
+{
+	struct btrfs_device *dev = stripe->dev;
+	struct btrfs_fs_info *fs_info = dev->fs_info;
+	struct btrfs_dev_replace *dev_replace = &fs_info->dev_replace;
+	u64 phys = stripe->physical;
+	u64 len = stripe->length;
+	u64 discarded = 0;
+	int ret = 0;
+
+	/* Zone reset on a zoned filesystem */
+	if (btrfs_can_zone_reset(dev, phys, len)) {
+		u64 src_disc;
+
+		ret = btrfs_reset_device_zone(dev, phys, len, &discarded);
+		if (ret)
+			goto out;
+
+		if (!btrfs_dev_replace_is_ongoing(dev_replace) ||
+		    dev != dev_replace->srcdev)
+			goto out;
+
+		src_disc = discarded;
+
+		/* Send to replace target as well */
+		ret = btrfs_reset_device_zone(dev_replace->tgtdev, phys, len,
+					      &discarded);
+		discarded += src_disc;
+	} else if (bdev_max_discard_sectors(stripe->dev->bdev)) {
+		ret = btrfs_issue_discard(dev->bdev, phys, len, &discarded);
+	} else {
+		ret = 0;
+		*bytes = 0;
+	}
+
+out:
+	*bytes = discarded;
+	return ret;
+}
+
 int btrfs_discard_extent(struct btrfs_fs_info *fs_info, u64 bytenr,
 			 u64 num_bytes, u64 *actual_bytes)
 {
@@ -1311,80 +1316,60 @@ int btrfs_discard_extent(struct btrfs_fs_info *fs_info, u64 bytenr,
 	u64 discarded_bytes = 0;
 	u64 end = bytenr + num_bytes;
 	u64 cur = bytenr;
-	struct btrfs_bio *bbio = NULL;
-
 
 	/*
-	 * Avoid races with device replace and make sure our bbio has devices
-	 * associated to its stripes that don't go away while we are discarding.
+	 * Avoid races with device replace and make sure the devices in the
+	 * stripes don't go away while we are discarding.
 	 */
 	btrfs_bio_counter_inc_blocked(fs_info);
 	while (cur < end) {
-		struct btrfs_bio_stripe *stripe;
+		struct btrfs_discard_stripe *stripes;
+		unsigned int num_stripes;
 		int i;
 
 		num_bytes = end - cur;
-		/* Tell the block device(s) that the sectors can be discarded */
-		ret = btrfs_map_block(fs_info, BTRFS_MAP_DISCARD, cur,
-				      &num_bytes, &bbio, 0);
-		/*
-		 * Error can be -ENOMEM, -ENOENT (no such chunk mapping) or
-		 * -EOPNOTSUPP. For any such error, @num_bytes is not updated,
-		 * thus we can't continue anyway.
-		 */
-		if (ret < 0)
-			goto out;
+		stripes = btrfs_map_discard(fs_info, cur, &num_bytes, &num_stripes);
+		if (IS_ERR(stripes)) {
+			ret = PTR_ERR(stripes);
+			if (ret == -EOPNOTSUPP)
+				ret = 0;
+			break;
+		}
 
-		stripe = bbio->stripes;
-		for (i = 0; i < bbio->num_stripes; i++, stripe++) {
+		for (i = 0; i < num_stripes; i++) {
+			struct btrfs_discard_stripe *stripe = stripes + i;
 			u64 bytes;
-			struct request_queue *req_q;
 
 			if (!stripe->dev->bdev) {
 				ASSERT(btrfs_test_opt(fs_info, DEGRADED));
 				continue;
 			}
-			req_q = bdev_get_queue(stripe->dev->bdev);
-			if (!blk_queue_discard(req_q))
+
+			if (!test_bit(BTRFS_DEV_STATE_WRITEABLE,
+					&stripe->dev->dev_state))
 				continue;
 
-			ret = btrfs_issue_discard(stripe->dev->bdev,
-						  stripe->physical,
-						  stripe->length,
-						  &bytes);
-			if (!ret) {
-				discarded_bytes += bytes;
-			} else if (ret != -EOPNOTSUPP) {
+			ret = do_discard_extent(stripe, &bytes);
+			if (ret) {
 				/*
-				 * Logic errors or -ENOMEM, or -EIO, but
-				 * unlikely to happen.
-				 *
-				 * And since there are two loops, explicitly
-				 * go to out to avoid confusion.
+				 * Keep going if discard is not supported by the
+				 * device.
 				 */
-				btrfs_put_bbio(bbio);
-				goto out;
+				if (ret != -EOPNOTSUPP)
+					break;
+				ret = 0;
+			} else {
+				discarded_bytes += bytes;
 			}
-
-			/*
-			 * Just in case we get back EOPNOTSUPP for some reason,
-			 * just ignore the return value so we don't screw up
-			 * people calling discard_extent.
-			 */
-			ret = 0;
 		}
-		btrfs_put_bbio(bbio);
+		kfree(stripes);
+		if (ret)
+			break;
 		cur += num_bytes;
 	}
-out:
 	btrfs_bio_counter_dec(fs_info);
-
 	if (actual_bytes)
 		*actual_bytes = discarded_bytes;
-
-
-	if (ret == -EOPNOTSUPP)
-		ret = 0;
 	return ret;
 }
 
@@ -1393,32 +1378,29 @@ int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
 			 struct btrfs_ref *generic_ref)
 {
 	struct btrfs_fs_info *fs_info = trans->fs_info;
-	int old_ref_mod, new_ref_mod;
 	int ret;
 
 	ASSERT(generic_ref->type != BTRFS_REF_NOT_SET &&
 	       generic_ref->action);
 	BUG_ON(generic_ref->type == BTRFS_REF_METADATA &&
-	       generic_ref->tree_ref.root == BTRFS_TREE_LOG_OBJECTID);
+	       generic_ref->tree_ref.owning_root == BTRFS_TREE_LOG_OBJECTID);
 
 	if (generic_ref->type == BTRFS_REF_METADATA)
-		ret = btrfs_add_delayed_tree_ref(trans, generic_ref,
-				NULL, &old_ref_mod, &new_ref_mod);
+		ret = btrfs_add_delayed_tree_ref(trans, generic_ref, NULL);
 	else
-		ret = btrfs_add_delayed_data_ref(trans, generic_ref, 0,
-						 &old_ref_mod, &new_ref_mod);
+		ret = btrfs_add_delayed_data_ref(trans, generic_ref, 0);
 
 	btrfs_ref_tree_mod(fs_info, generic_ref);
 
-	if (ret == 0 && old_ref_mod < 0 && new_ref_mod >= 0)
-		sub_pinned_bytes(fs_info, generic_ref);
-
 	return ret;
 }
 
 /*
  * __btrfs_inc_extent_ref - insert backreference for a given extent
  *
+ * The counterpart is in __btrfs_free_extent(), with examples and more details
+ * how it works.
+ *
  * @trans:	    Handle of transaction
  *
  * @node:	    The delayed ref node used to get the bytenr/length for
@@ -1429,7 +1411,7 @@ int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
  *		    bytenr of the parent block. Since new extents are always
  *		    created with indirect references, this will only be the case
  *		    when relocating a shared extent. In that case, root_objectid
- *		    will be BTRFS_TREE_RELOC_OBJECTID. Otheriwse, parent must
+ *		    will be BTRFS_TREE_RELOC_OBJECTID. Otherwise, parent must
  *		    be 0
  *
  * @root_objectid:  The id of the root where this modification has originated,
@@ -1469,8 +1451,6 @@ static int __btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
 	if (!path)
 		return -ENOMEM;
 
-	path->reada = READA_FORWARD;
-	path->leave_spinning = 1;
 	/* this will setup the path even if it fails to insert the back ref */
 	ret = insert_inline_extent_backref(trans, path, bytenr, num_bytes,
 					   parent, root_objectid, owner,
@@ -1494,11 +1474,16 @@ static int __btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
 	btrfs_mark_buffer_dirty(leaf);
 	btrfs_release_path(path);
 
-	path->reada = READA_FORWARD;
-	path->leave_spinning = 1;
 	/* now insert the actual backref */
-	ret = insert_extent_backref(trans, path, bytenr, parent, root_objectid,
-				    owner, offset, refs_to_add);
+	if (owner < BTRFS_FIRST_FREE_OBJECTID) {
+		BUG_ON(refs_to_add != 1);
+		ret = insert_tree_block_ref(trans, path, bytenr, parent,
+					    root_objectid);
+	} else {
+		ret = insert_extent_data_ref(trans, path, bytenr, parent,
+					     root_objectid, owner, offset,
+					     refs_to_add);
+	}
 	if (ret)
 		btrfs_abort_transaction(trans, ret);
 out:
@@ -1574,6 +1559,7 @@ static int run_delayed_extent_op(struct btrfs_trans_handle *trans,
 				 struct btrfs_delayed_extent_op *extent_op)
 {
 	struct btrfs_fs_info *fs_info = trans->fs_info;
+	struct btrfs_root *root;
 	struct btrfs_key key;
 	struct btrfs_path *path;
 	struct btrfs_extent_item *ei;
@@ -1581,12 +1567,12 @@ static int run_delayed_extent_op(struct btrfs_trans_handle *trans,
 	u32 item_size;
 	int ret;
 	int err = 0;
-	int metadata = !extent_op->is_data;
+	int metadata = 1;
 
-	if (trans->aborted)
+	if (TRANS_ABORTED(trans))
 		return 0;
 
-	if (metadata && !btrfs_fs_incompat(fs_info, SKINNY_METADATA))
+	if (!btrfs_fs_incompat(fs_info, SKINNY_METADATA))
 		metadata = 0;
 
 	path = btrfs_alloc_path();
@@ -1603,10 +1589,9 @@ static int run_delayed_extent_op(struct btrfs_trans_handle *trans,
 		key.offset = head->num_bytes;
 	}
 
+	root = btrfs_extent_root(fs_info, key.objectid);
 again:
-	path->reada = READA_FORWARD;
-	path->leave_spinning = 1;
-	ret = btrfs_search_slot(trans, fs_info->extent_root, &key, path, 0, 1);
+	ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
 	if (ret < 0) {
 		err = ret;
 		goto out;
@@ -1638,7 +1623,7 @@ again:
 	}
 
 	leaf = path->nodes[0];
-	item_size = btrfs_item_size_nr(leaf, path->slots[0]);
+	item_size = btrfs_item_size(leaf, path->slots[0]);
 
 	if (unlikely(item_size < sizeof(*ei))) {
 		err = -EINVAL;
@@ -1703,10 +1688,9 @@ static int run_one_delayed_ref(struct btrfs_trans_handle *trans,
 {
 	int ret = 0;
 
-	if (trans->aborted) {
+	if (TRANS_ABORTED(trans)) {
 		if (insert_reserved)
-			btrfs_pin_extent(trans->fs_info, node->bytenr,
-					 node->num_bytes, 1);
+			btrfs_pin_extent(trans, node->bytenr, node->num_bytes, 1);
 		return 0;
 	}
 
@@ -1721,8 +1705,7 @@ static int run_one_delayed_ref(struct btrfs_trans_handle *trans,
 	else
 		BUG();
 	if (ret && insert_reserved)
-		btrfs_pin_extent(trans->fs_info, node->bytenr,
-				 node->num_bytes, 1);
+		btrfs_pin_extent(trans, node->bytenr, node->num_bytes, 1);
 	return ret;
 }
 
@@ -1798,34 +1781,15 @@ void btrfs_cleanup_ref_head_accounting(struct btrfs_fs_info *fs_info,
 {
 	int nr_items = 1;	/* Dropping this ref head update. */
 
-	if (head->total_ref_mod < 0) {
-		struct btrfs_space_info *space_info;
-		u64 flags;
-
-		if (head->is_data)
-			flags = BTRFS_BLOCK_GROUP_DATA;
-		else if (head->is_system)
-			flags = BTRFS_BLOCK_GROUP_SYSTEM;
-		else
-			flags = BTRFS_BLOCK_GROUP_METADATA;
-		space_info = btrfs_find_space_info(fs_info, flags);
-		ASSERT(space_info);
-		percpu_counter_add_batch(&space_info->total_bytes_pinned,
-				   -head->num_bytes,
-				   BTRFS_TOTAL_BYTES_PINNED_BATCH);
-
-		/*
-		 * We had csum deletions accounted for in our delayed refs rsv,
-		 * we need to drop the csum leaves for this update from our
-		 * delayed_refs_rsv.
-		 */
-		if (head->is_data) {
-			spin_lock(&delayed_refs->lock);
-			delayed_refs->pending_csums -= head->num_bytes;
-			spin_unlock(&delayed_refs->lock);
-			nr_items += btrfs_csum_bytes_to_leaves(fs_info,
-				head->num_bytes);
-		}
+	/*
+	 * We had csum deletions accounted for in our delayed refs rsv, we need
+	 * to drop the csum leaves for this update from our delayed_refs_rsv.
+	 */
+	if (head->total_ref_mod < 0 && head->is_data) {
+		spin_lock(&delayed_refs->lock);
+		delayed_refs->pending_csums -= head->num_bytes;
+		spin_unlock(&delayed_refs->lock);
+		nr_items += btrfs_csum_bytes_to_leaves(fs_info, head->num_bytes);
 	}
 
 	btrfs_delayed_refs_rsv_release(fs_info, nr_items);
@@ -1867,11 +1831,13 @@ static int cleanup_ref_head(struct btrfs_trans_handle *trans,
 	spin_unlock(&delayed_refs->lock);
 
 	if (head->must_insert_reserved) {
-		btrfs_pin_extent(fs_info, head->bytenr,
-				 head->num_bytes, 1);
+		btrfs_pin_extent(trans, head->bytenr, head->num_bytes, 1);
 		if (head->is_data) {
-			ret = btrfs_del_csums(trans, fs_info->csum_root,
-					      head->bytenr, head->num_bytes);
+			struct btrfs_root *csum_root;
+
+			csum_root = btrfs_csum_root(fs_info, head->bytenr);
+			ret = btrfs_del_csums(trans, csum_root, head->bytenr,
+					      head->num_bytes);
 		}
 	}
 
@@ -1880,7 +1846,7 @@ static int cleanup_ref_head(struct btrfs_trans_handle *trans,
 	trace_run_delayed_ref_head(fs_info, head, 0);
 	btrfs_delayed_ref_unlock(head);
 	btrfs_put_delayed_ref_head(head);
-	return 0;
+	return ret;
 }
 
 static struct btrfs_delayed_ref_head *btrfs_obtain_ref_head(
@@ -2135,41 +2101,6 @@ static u64 find_middle(struct rb_root *root)
 }
 #endif
 
-static inline u64 heads_to_leaves(struct btrfs_fs_info *fs_info, u64 heads)
-{
-	u64 num_bytes;
-
-	num_bytes = heads * (sizeof(struct btrfs_extent_item) +
-			     sizeof(struct btrfs_extent_inline_ref));
-	if (!btrfs_fs_incompat(fs_info, SKINNY_METADATA))
-		num_bytes += heads * sizeof(struct btrfs_tree_block_info);
-
-	/*
-	 * We don't ever fill up leaves all the way so multiply by 2 just to be
-	 * closer to what we're really going to want to use.
-	 */
-	return div_u64(num_bytes, BTRFS_LEAF_DATA_SIZE(fs_info));
-}
-
-/*
- * Takes the number of bytes to be csumm'ed and figures out how many leaves it
- * would require to store the csums for that many bytes.
- */
-u64 btrfs_csum_bytes_to_leaves(struct btrfs_fs_info *fs_info, u64 csum_bytes)
-{
-	u64 csum_size;
-	u64 num_csums_per_leaf;
-	u64 num_csums;
-
-	csum_size = BTRFS_MAX_ITEM_SIZE(fs_info);
-	num_csums_per_leaf = div64_u64(csum_size,
-			(u64)btrfs_super_csum_size(fs_info->super_copy));
-	num_csums = div64_u64(csum_bytes, fs_info->sectorsize);
-	num_csums += num_csums_per_leaf - 1;
-	num_csums = div64_u64(num_csums, num_csums_per_leaf);
-	return num_csums;
-}
-
 /*
  * this starts processing the delayed reference count updates and
  * extent insertions we have queued up so far.  count can be
@@ -2191,7 +2122,7 @@ int btrfs_run_delayed_refs(struct btrfs_trans_handle *trans,
 	int run_all = count == (unsigned long)-1;
 
 	/* We'll clean this up in btrfs_cleanup_transaction */
-	if (trans->aborted)
+	if (TRANS_ABORTED(trans))
 		return 0;
 
 	if (test_bit(BTRFS_FS_CREATING_FREE_SPACE_TREE, &fs_info->flags))
@@ -2199,7 +2130,7 @@ int btrfs_run_delayed_refs(struct btrfs_trans_handle *trans,
 
 	delayed_refs = &trans->transaction->delayed_refs;
 	if (count == 0)
-		count = atomic_read(&delayed_refs->num_entries) * 2;
+		count = delayed_refs->num_heads_ready;
 
 again:
 #ifdef SCRAMBLE_DELAYED_REFS
@@ -2238,8 +2169,8 @@ out:
 }
 
 int btrfs_set_disk_extent_flags(struct btrfs_trans_handle *trans,
-				u64 bytenr, u64 num_bytes, u64 flags,
-				int level, int is_data)
+				struct extent_buffer *eb, u64 flags,
+				int level)
 {
 	struct btrfs_delayed_extent_op *extent_op;
 	int ret;
@@ -2251,10 +2182,9 @@ int btrfs_set_disk_extent_flags(struct btrfs_trans_handle *trans,
 	extent_op->flags_to_set = flags;
 	extent_op->update_flags = true;
 	extent_op->update_key = false;
-	extent_op->is_data = is_data ? true : false;
 	extent_op->level = level;
 
-	ret = btrfs_add_delayed_extent_op(trans, bytenr, num_bytes, extent_op);
+	ret = btrfs_add_delayed_extent_op(trans, eb->start, eb->len, extent_op);
 	if (ret)
 		btrfs_free_delayed_extent_op(extent_op);
 	return ret;
@@ -2290,6 +2220,12 @@ static noinline int check_delayed_ref(struct btrfs_root *root,
 	}
 
 	if (!mutex_trylock(&head->mutex)) {
+		if (path->nowait) {
+			spin_unlock(&delayed_refs->lock);
+			btrfs_put_transaction(cur_trans);
+			return -EAGAIN;
+		}
+
 		refcount_inc(&head->refs);
 		spin_unlock(&delayed_refs->lock);
 
@@ -2342,10 +2278,11 @@ static noinline int check_delayed_ref(struct btrfs_root *root,
 
 static noinline int check_committed_ref(struct btrfs_root *root,
 					struct btrfs_path *path,
-					u64 objectid, u64 offset, u64 bytenr)
+					u64 objectid, u64 offset, u64 bytenr,
+					bool strict)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
-	struct btrfs_root *extent_root = fs_info->extent_root;
+	struct btrfs_root *extent_root = btrfs_extent_root(fs_info, bytenr);
 	struct extent_buffer *leaf;
 	struct btrfs_extent_data_ref *ref;
 	struct btrfs_extent_inline_ref *iref;
@@ -2376,7 +2313,7 @@ static noinline int check_committed_ref(struct btrfs_root *root,
 		goto out;
 
 	ret = 1;
-	item_size = btrfs_item_size_nr(leaf, path->slots[0]);
+	item_size = btrfs_item_size(leaf, path->slots[0]);
 	ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
 
 	/* If extent item has more than 1 inline ref then it's shared */
@@ -2384,9 +2321,13 @@ static noinline int check_committed_ref(struct btrfs_root *root,
 	    btrfs_extent_inline_ref_size(BTRFS_EXTENT_DATA_REF_KEY))
 		goto out;
 
-	/* If extent created before last snapshot => it's definitely shared */
-	if (btrfs_extent_generation(leaf, ei) <=
-	    btrfs_root_last_snapshot(&root->root_item))
+	/*
+	 * If extent created before last snapshot => it's shared unless the
+	 * snapshot has been deleted. Use the heuristic if strict is false.
+	 */
+	if (!strict &&
+	    (btrfs_extent_generation(leaf, ei) <=
+	     btrfs_root_last_snapshot(&root->root_item)))
 		goto out;
 
 	iref = (struct btrfs_extent_inline_ref *)(ei + 1);
@@ -2411,18 +2352,13 @@ out:
 }
 
 int btrfs_cross_ref_exist(struct btrfs_root *root, u64 objectid, u64 offset,
-			  u64 bytenr)
+			  u64 bytenr, bool strict, struct btrfs_path *path)
 {
-	struct btrfs_path *path;
 	int ret;
 
-	path = btrfs_alloc_path();
-	if (!path)
-		return -ENOMEM;
-
 	do {
 		ret = check_committed_ref(root, path, objectid,
-					  offset, bytenr);
+					  offset, bytenr, strict);
 		if (ret && ret != -ENOENT)
 			goto out;
 
@@ -2430,8 +2366,8 @@ int btrfs_cross_ref_exist(struct btrfs_root *root, u64 objectid, u64 offset,
 	} while (ret == -EAGAIN);
 
 out:
-	btrfs_free_path(path);
-	if (root->root_key.objectid == BTRFS_DATA_RELOC_TREE_OBJECTID)
+	btrfs_release_path(path);
+	if (btrfs_is_data_reloc_root(root))
 		WARN_ON(ret > 0);
 	return ret;
 }
@@ -2463,7 +2399,7 @@ static int __btrfs_mod_ref(struct btrfs_trans_handle *trans,
 	nritems = btrfs_header_nritems(buf);
 	level = btrfs_header_level(buf);
 
-	if (!test_bit(BTRFS_ROOT_REF_COWS, &root->state) && level == 0)
+	if (!test_bit(BTRFS_ROOT_SHAREABLE, &root->state) && level == 0)
 		return 0;
 
 	if (full_backref)
@@ -2493,10 +2429,9 @@ static int __btrfs_mod_ref(struct btrfs_trans_handle *trans,
 			key.offset -= btrfs_file_extent_offset(buf, fi);
 			btrfs_init_generic_ref(&generic_ref, action, bytenr,
 					       num_bytes, parent);
-			generic_ref.real_root = root->root_key.objectid;
 			btrfs_init_data_ref(&generic_ref, ref_root, key.objectid,
-					    key.offset);
-			generic_ref.skip_qgroup = for_reloc;
+					    key.offset, root->root_key.objectid,
+					    for_reloc);
 			if (inc)
 				ret = btrfs_inc_extent_ref(trans, &generic_ref);
 			else
@@ -2508,9 +2443,8 @@ static int __btrfs_mod_ref(struct btrfs_trans_handle *trans,
 			num_bytes = fs_info->nodesize;
 			btrfs_init_generic_ref(&generic_ref, action, bytenr,
 					       num_bytes, parent);
-			generic_ref.real_root = root->root_key.objectid;
-			btrfs_init_tree_ref(&generic_ref, level - 1, ref_root);
-			generic_ref.skip_qgroup = for_reloc;
+			btrfs_init_tree_ref(&generic_ref, level - 1, ref_root,
+					    root->root_key.objectid, for_reloc);
 			if (inc)
 				ret = btrfs_inc_extent_ref(trans, &generic_ref);
 			else
@@ -2536,19 +2470,6 @@ int btrfs_dec_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 	return __btrfs_mod_ref(trans, root, buf, full_backref, 0);
 }
 
-int btrfs_extent_readonly(struct btrfs_fs_info *fs_info, u64 bytenr)
-{
-	struct btrfs_block_group *block_group;
-	int readonly = 0;
-
-	block_group = btrfs_lookup_block_group(fs_info, bytenr);
-	if (!block_group || block_group->ro)
-		readonly = 1;
-	if (block_group)
-		btrfs_put_block_group(block_group);
-	return readonly;
-}
-
 static u64 get_alloc_profile_by_root(struct btrfs_root *root, int data)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
@@ -2566,29 +2487,27 @@ static u64 get_alloc_profile_by_root(struct btrfs_root *root, int data)
 	return ret;
 }
 
-static u64 first_logical_byte(struct btrfs_fs_info *fs_info, u64 search_start)
+static u64 first_logical_byte(struct btrfs_fs_info *fs_info)
 {
-	struct btrfs_block_group *cache;
-	u64 bytenr;
+	struct rb_node *leftmost;
+	u64 bytenr = 0;
 
-	spin_lock(&fs_info->block_group_cache_lock);
-	bytenr = fs_info->first_logical_byte;
-	spin_unlock(&fs_info->block_group_cache_lock);
+	read_lock(&fs_info->block_group_cache_lock);
+	/* Get the block group with the lowest logical start address. */
+	leftmost = rb_first_cached(&fs_info->block_group_cache_tree);
+	if (leftmost) {
+		struct btrfs_block_group *bg;
 
-	if (bytenr < (u64)-1)
-		return bytenr;
-
-	cache = btrfs_lookup_first_block_group(fs_info, search_start);
-	if (!cache)
-		return 0;
-
-	bytenr = cache->start;
-	btrfs_put_block_group(cache);
+		bg = rb_entry(leftmost, struct btrfs_block_group, cache_node);
+		bytenr = bg->start;
+	}
+	read_unlock(&fs_info->block_group_cache_lock);
 
 	return bytenr;
 }
 
-static int pin_down_extent(struct btrfs_block_group *cache,
+static int pin_down_extent(struct btrfs_trans_handle *trans,
+			   struct btrfs_block_group *cache,
 			   u64 bytenr, u64 num_bytes, int reserved)
 {
 	struct btrfs_fs_info *fs_info = cache->fs_info;
@@ -2605,24 +2524,20 @@ static int pin_down_extent(struct btrfs_block_group *cache,
 	spin_unlock(&cache->lock);
 	spin_unlock(&cache->space_info->lock);
 
-	percpu_counter_add_batch(&cache->space_info->total_bytes_pinned,
-		    num_bytes, BTRFS_TOTAL_BYTES_PINNED_BATCH);
-	set_extent_dirty(fs_info->pinned_extents, bytenr,
+	set_extent_dirty(&trans->transaction->pinned_extents, bytenr,
 			 bytenr + num_bytes - 1, GFP_NOFS | __GFP_NOFAIL);
 	return 0;
 }
 
-int btrfs_pin_extent(struct btrfs_fs_info *fs_info,
+int btrfs_pin_extent(struct btrfs_trans_handle *trans,
 		     u64 bytenr, u64 num_bytes, int reserved)
 {
 	struct btrfs_block_group *cache;
 
-	ASSERT(fs_info->running_transaction);
-
-	cache = btrfs_lookup_block_group(fs_info, bytenr);
+	cache = btrfs_lookup_block_group(trans->fs_info, bytenr);
 	BUG_ON(!cache); /* Logic error */
 
-	pin_down_extent(cache, bytenr, num_bytes, reserved);
+	pin_down_extent(trans, cache, bytenr, num_bytes, reserved);
 
 	btrfs_put_block_group(cache);
 	return 0;
@@ -2631,28 +2546,29 @@ int btrfs_pin_extent(struct btrfs_fs_info *fs_info,
 /*
  * this function must be called within transaction
  */
-int btrfs_pin_extent_for_log_replay(struct btrfs_fs_info *fs_info,
+int btrfs_pin_extent_for_log_replay(struct btrfs_trans_handle *trans,
 				    u64 bytenr, u64 num_bytes)
 {
 	struct btrfs_block_group *cache;
 	int ret;
 
-	cache = btrfs_lookup_block_group(fs_info, bytenr);
+	cache = btrfs_lookup_block_group(trans->fs_info, bytenr);
 	if (!cache)
 		return -EINVAL;
 
 	/*
-	 * pull in the free space cache (if any) so that our pin
-	 * removes the free space from the cache.  We have load_only set
-	 * to one because the slow code to read in the free extents does check
-	 * the pinned extents.
+	 * Fully cache the free space first so that our pin removes the free space
+	 * from the cache.
 	 */
-	btrfs_cache_block_group(cache, 1);
+	ret = btrfs_cache_block_group(cache, true);
+	if (ret)
+		goto out;
 
-	pin_down_extent(cache, bytenr, num_bytes, 0);
+	pin_down_extent(trans, cache, bytenr, num_bytes, 0);
 
 	/* remove us from the free space cache (if we're there at all) */
 	ret = btrfs_remove_free_space(cache, bytenr, num_bytes);
+out:
 	btrfs_put_block_group(cache);
 	return ret;
 }
@@ -2662,45 +2578,17 @@ static int __exclude_logged_extent(struct btrfs_fs_info *fs_info,
 {
 	int ret;
 	struct btrfs_block_group *block_group;
-	struct btrfs_caching_control *caching_ctl;
 
 	block_group = btrfs_lookup_block_group(fs_info, start);
 	if (!block_group)
 		return -EINVAL;
 
-	btrfs_cache_block_group(block_group, 0);
-	caching_ctl = btrfs_get_caching_control(block_group);
-
-	if (!caching_ctl) {
-		/* Logic error */
-		BUG_ON(!btrfs_block_group_done(block_group));
-		ret = btrfs_remove_free_space(block_group, start, num_bytes);
-	} else {
-		mutex_lock(&caching_ctl->mutex);
-
-		if (start >= caching_ctl->progress) {
-			ret = btrfs_add_excluded_extent(fs_info, start,
-							num_bytes);
-		} else if (start + num_bytes <= caching_ctl->progress) {
-			ret = btrfs_remove_free_space(block_group,
-						      start, num_bytes);
-		} else {
-			num_bytes = caching_ctl->progress - start;
-			ret = btrfs_remove_free_space(block_group,
-						      start, num_bytes);
-			if (ret)
-				goto out_lock;
+	ret = btrfs_cache_block_group(block_group, true);
+	if (ret)
+		goto out;
 
-			num_bytes = (start + num_bytes) -
-				caching_ctl->progress;
-			start = caching_ctl->progress;
-			ret = btrfs_add_excluded_extent(fs_info, start,
-							num_bytes);
-		}
-out_lock:
-		mutex_unlock(&caching_ctl->mutex);
-		btrfs_put_caching_control(caching_ctl);
-	}
+	ret = btrfs_remove_free_space(block_group, start, num_bytes);
+out:
 	btrfs_put_block_group(block_group);
 	return ret;
 }
@@ -2743,36 +2631,6 @@ btrfs_inc_block_group_reservations(struct btrfs_block_group *bg)
 	atomic_inc(&bg->reservations);
 }
 
-void btrfs_prepare_extent_commit(struct btrfs_fs_info *fs_info)
-{
-	struct btrfs_caching_control *next;
-	struct btrfs_caching_control *caching_ctl;
-	struct btrfs_block_group *cache;
-
-	down_write(&fs_info->commit_root_sem);
-
-	list_for_each_entry_safe(caching_ctl, next,
-				 &fs_info->caching_block_groups, list) {
-		cache = caching_ctl->block_group;
-		if (btrfs_block_group_done(cache)) {
-			cache->last_byte_to_unpin = (u64)-1;
-			list_del_init(&caching_ctl->list);
-			btrfs_put_caching_control(caching_ctl);
-		} else {
-			cache->last_byte_to_unpin = caching_ctl->progress;
-		}
-	}
-
-	if (fs_info->pinned_extents == &fs_info->freed_extents[0])
-		fs_info->pinned_extents = &fs_info->freed_extents[1];
-	else
-		fs_info->pinned_extents = &fs_info->freed_extents[0];
-
-	up_write(&fs_info->commit_root_sem);
-
-	btrfs_update_global_block_rsv(fs_info);
-}
-
 /*
  * Returns the free cluster for the given space info and sets empty_cluster to
  * what it should be based on the mount options.
@@ -2834,11 +2692,8 @@ static int unpin_extent_range(struct btrfs_fs_info *fs_info,
 		len = cache->start + cache->length - start;
 		len = min(len, end + 1 - start);
 
-		if (start < cache->last_byte_to_unpin) {
-			len = min(len, cache->last_byte_to_unpin - start);
-			if (return_free_space)
-				btrfs_add_free_space(cache, start, len);
-		}
+		if (return_free_space)
+			btrfs_add_free_space(cache, start, len);
 
 		start += len;
 		total_unpinned += len;
@@ -2862,21 +2717,22 @@ static int unpin_extent_range(struct btrfs_fs_info *fs_info,
 		cache->pinned -= len;
 		btrfs_space_info_update_bytes_pinned(fs_info, space_info, -len);
 		space_info->max_extent_size = 0;
-		percpu_counter_add_batch(&space_info->total_bytes_pinned,
-			    -len, BTRFS_TOTAL_BYTES_PINNED_BATCH);
 		if (cache->ro) {
 			space_info->bytes_readonly += len;
 			readonly = true;
+		} else if (btrfs_is_zoned(fs_info)) {
+			/* Need reset before reusing in a zoned block group */
+			space_info->bytes_zone_unusable += len;
+			readonly = true;
 		}
 		spin_unlock(&cache->lock);
 		if (!readonly && return_free_space &&
 		    global_rsv->space_info == space_info) {
-			u64 to_add = len;
-
 			spin_lock(&global_rsv->lock);
 			if (!global_rsv->full) {
-				to_add = min(len, global_rsv->size -
-					     global_rsv->reserved);
+				u64 to_add = min(len, global_rsv->size -
+						      global_rsv->reserved);
+
 				global_rsv->reserved += to_add;
 				btrfs_space_info_update_bytes_may_use(fs_info,
 						space_info, to_add);
@@ -2885,11 +2741,10 @@ static int unpin_extent_range(struct btrfs_fs_info *fs_info,
 				len -= to_add;
 			}
 			spin_unlock(&global_rsv->lock);
-			/* Add to any tickets we may have */
-			if (len)
-				btrfs_try_granting_tickets(fs_info,
-							   space_info);
 		}
+		/* Add to any tickets we may have */
+		if (!readonly && return_free_space && len)
+			btrfs_try_granting_tickets(fs_info, space_info);
 		spin_unlock(&space_info->lock);
 	}
 
@@ -2908,12 +2763,9 @@ int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans)
 	u64 end;
 	int ret;
 
-	if (fs_info->pinned_extents == &fs_info->freed_extents[0])
-		unpin = &fs_info->freed_extents[1];
-	else
-		unpin = &fs_info->freed_extents[0];
+	unpin = &trans->transaction->pinned_extents;
 
-	while (!trans->aborted) {
+	while (!TRANS_ABORTED(trans)) {
 		struct extent_state *cached_state = NULL;
 
 		mutex_lock(&fs_info->unused_bg_unpin_mutex);
@@ -2950,14 +2802,14 @@ int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans)
 		u64 trimmed = 0;
 
 		ret = -EROFS;
-		if (!trans->aborted)
+		if (!TRANS_ABORTED(trans))
 			ret = btrfs_discard_extent(fs_info,
 						   block_group->start,
 						   block_group->length,
 						   &trimmed);
 
 		list_del_init(&block_group->bg_list);
-		btrfs_put_block_group_trimming(block_group);
+		btrfs_unfreeze_block_group(block_group);
 		btrfs_put_block_group(block_group);
 
 		if (ret) {
@@ -2971,6 +2823,94 @@ int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans)
 	return 0;
 }
 
+static int do_free_extent_accounting(struct btrfs_trans_handle *trans,
+				     u64 bytenr, u64 num_bytes, bool is_data)
+{
+	int ret;
+
+	if (is_data) {
+		struct btrfs_root *csum_root;
+
+		csum_root = btrfs_csum_root(trans->fs_info, bytenr);
+		ret = btrfs_del_csums(trans, csum_root, bytenr, num_bytes);
+		if (ret) {
+			btrfs_abort_transaction(trans, ret);
+			return ret;
+		}
+	}
+
+	ret = add_to_free_space_tree(trans, bytenr, num_bytes);
+	if (ret) {
+		btrfs_abort_transaction(trans, ret);
+		return ret;
+	}
+
+	ret = btrfs_update_block_group(trans, bytenr, num_bytes, false);
+	if (ret)
+		btrfs_abort_transaction(trans, ret);
+
+	return ret;
+}
+
+/*
+ * Drop one or more refs of @node.
+ *
+ * 1. Locate the extent refs.
+ *    It's either inline in EXTENT/METADATA_ITEM or in keyed SHARED_* item.
+ *    Locate it, then reduce the refs number or remove the ref line completely.
+ *
+ * 2. Update the refs count in EXTENT/METADATA_ITEM
+ *
+ * Inline backref case:
+ *
+ * in extent tree we have:
+ *
+ * 	item 0 key (13631488 EXTENT_ITEM 1048576) itemoff 16201 itemsize 82
+ *		refs 2 gen 6 flags DATA
+ *		extent data backref root FS_TREE objectid 258 offset 0 count 1
+ *		extent data backref root FS_TREE objectid 257 offset 0 count 1
+ *
+ * This function gets called with:
+ *
+ *    node->bytenr = 13631488
+ *    node->num_bytes = 1048576
+ *    root_objectid = FS_TREE
+ *    owner_objectid = 257
+ *    owner_offset = 0
+ *    refs_to_drop = 1
+ *
+ * Then we should get some like:
+ *
+ * 	item 0 key (13631488 EXTENT_ITEM 1048576) itemoff 16201 itemsize 82
+ *		refs 1 gen 6 flags DATA
+ *		extent data backref root FS_TREE objectid 258 offset 0 count 1
+ *
+ * Keyed backref case:
+ *
+ * in extent tree we have:
+ *
+ *	item 0 key (13631488 EXTENT_ITEM 1048576) itemoff 3971 itemsize 24
+ *		refs 754 gen 6 flags DATA
+ *	[...]
+ *	item 2 key (13631488 EXTENT_DATA_REF <HASH>) itemoff 3915 itemsize 28
+ *		extent data backref root FS_TREE objectid 866 offset 0 count 1
+ *
+ * This function get called with:
+ *
+ *    node->bytenr = 13631488
+ *    node->num_bytes = 1048576
+ *    root_objectid = FS_TREE
+ *    owner_objectid = 866
+ *    owner_offset = 0
+ *    refs_to_drop = 1
+ *
+ * Then we should get some like:
+ *
+ *	item 0 key (13631488 EXTENT_ITEM 1048576) itemoff 3971 itemsize 24
+ *		refs 753 gen 6 flags DATA
+ *
+ * And that (13631488 EXTENT_DATA_REF <HASH>) gets removed.
+ */
 static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 			       struct btrfs_delayed_ref_node *node, u64 parent,
 			       u64 root_objectid, u64 owner_objectid,
@@ -2980,7 +2920,7 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 	struct btrfs_fs_info *info = trans->fs_info;
 	struct btrfs_key key;
 	struct btrfs_path *path;
-	struct btrfs_root *extent_root = info->extent_root;
+	struct btrfs_root *extent_root;
 	struct extent_buffer *leaf;
 	struct btrfs_extent_item *ei;
 	struct btrfs_extent_inline_ref *iref;
@@ -2993,18 +2933,25 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 	u64 refs;
 	u64 bytenr = node->bytenr;
 	u64 num_bytes = node->num_bytes;
-	int last_ref = 0;
 	bool skinny_metadata = btrfs_fs_incompat(info, SKINNY_METADATA);
 
+	extent_root = btrfs_extent_root(info, bytenr);
+	ASSERT(extent_root);
+
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
 
-	path->reada = READA_FORWARD;
-	path->leave_spinning = 1;
-
 	is_data = owner_objectid >= BTRFS_FIRST_FREE_OBJECTID;
-	BUG_ON(!is_data && refs_to_drop != 1);
+
+	if (!is_data && refs_to_drop != 1) {
+		btrfs_crit(info,
+"invalid refs_to_drop, dropping more than 1 refs for tree block %llu refs_to_drop %u",
+			   node->bytenr, refs_to_drop);
+		ret = -EINVAL;
+		btrfs_abort_transaction(trans, ret);
+		goto out;
+	}
 
 	if (is_data)
 		skinny_metadata = false;
@@ -3013,6 +2960,13 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 				    parent, root_objectid, owner_objectid,
 				    owner_offset);
 	if (ret == 0) {
+		/*
+		 * Either the inline backref or the SHARED_DATA_REF/
+		 * SHARED_BLOCK_REF is found
+		 *
+		 * Here is a quick path to locate EXTENT/METADATA_ITEM.
+		 * It's possible the EXTENT/METADATA_ITEM is near current slot.
+		 */
 		extent_slot = path->slots[0];
 		while (extent_slot >= 0) {
 			btrfs_item_key_to_cpu(path->nodes[0], &key,
@@ -3029,23 +2983,30 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 				found_extent = 1;
 				break;
 			}
+
+			/* Quick path didn't find the EXTEMT/METADATA_ITEM */
 			if (path->slots[0] - extent_slot > 5)
 				break;
 			extent_slot--;
 		}
 
 		if (!found_extent) {
-			BUG_ON(iref);
-			ret = remove_extent_backref(trans, path, NULL,
-						    refs_to_drop,
-						    is_data, &last_ref);
+			if (iref) {
+				btrfs_crit(info,
+"invalid iref, no EXTENT/METADATA_ITEM found but has inline extent ref");
+				btrfs_abort_transaction(trans, -EUCLEAN);
+				goto err_dump;
+			}
+			/* Must be SHARED_* item, remove the backref first */
+			ret = remove_extent_backref(trans, extent_root, path,
+						    NULL, refs_to_drop, is_data);
 			if (ret) {
 				btrfs_abort_transaction(trans, ret);
 				goto out;
 			}
 			btrfs_release_path(path);
-			path->leave_spinning = 1;
 
+			/* Slow path to locate EXTENT/METADATA_ITEM */
 			key.objectid = bytenr;
 			key.type = BTRFS_EXTENT_ITEM_KEY;
 			key.offset = num_bytes;
@@ -3108,7 +3069,7 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 	}
 
 	leaf = path->nodes[0];
-	item_size = btrfs_item_size_nr(leaf, extent_slot);
+	item_size = btrfs_item_size(leaf, extent_slot);
 	if (unlikely(item_size < sizeof(*ei))) {
 		ret = -EINVAL;
 		btrfs_print_v0_err(info);
@@ -3120,19 +3081,26 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 	if (owner_objectid < BTRFS_FIRST_FREE_OBJECTID &&
 	    key.type == BTRFS_EXTENT_ITEM_KEY) {
 		struct btrfs_tree_block_info *bi;
-		BUG_ON(item_size < sizeof(*ei) + sizeof(*bi));
+		if (item_size < sizeof(*ei) + sizeof(*bi)) {
+			btrfs_crit(info,
+"invalid extent item size for key (%llu, %u, %llu) owner %llu, has %u expect >= %zu",
+				   key.objectid, key.type, key.offset,
+				   owner_objectid, item_size,
+				   sizeof(*ei) + sizeof(*bi));
+			btrfs_abort_transaction(trans, -EUCLEAN);
+			goto err_dump;
+		}
 		bi = (struct btrfs_tree_block_info *)(ei + 1);
 		WARN_ON(owner_objectid != btrfs_tree_block_level(leaf, bi));
 	}
 
 	refs = btrfs_extent_refs(leaf, ei);
 	if (refs < refs_to_drop) {
-		btrfs_err(info,
-			  "trying to drop %d refs but we only have %Lu for bytenr %Lu",
+		btrfs_crit(info,
+		"trying to drop %d refs but we only have %llu for bytenr %llu",
 			  refs_to_drop, refs, bytenr);
-		ret = -EINVAL;
-		btrfs_abort_transaction(trans, ret);
-		goto out;
+		btrfs_abort_transaction(trans, -EUCLEAN);
+		goto err_dump;
 	}
 	refs -= refs_to_drop;
 
@@ -3144,34 +3112,63 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 		 * be updated by remove_extent_backref
 		 */
 		if (iref) {
-			BUG_ON(!found_extent);
+			if (!found_extent) {
+				btrfs_crit(info,
+"invalid iref, got inlined extent ref but no EXTENT/METADATA_ITEM found");
+				btrfs_abort_transaction(trans, -EUCLEAN);
+				goto err_dump;
+			}
 		} else {
 			btrfs_set_extent_refs(leaf, ei, refs);
 			btrfs_mark_buffer_dirty(leaf);
 		}
 		if (found_extent) {
-			ret = remove_extent_backref(trans, path, iref,
-						    refs_to_drop, is_data,
-						    &last_ref);
+			ret = remove_extent_backref(trans, extent_root, path,
+						    iref, refs_to_drop, is_data);
 			if (ret) {
 				btrfs_abort_transaction(trans, ret);
 				goto out;
 			}
 		}
 	} else {
+		/* In this branch refs == 1 */
 		if (found_extent) {
-			BUG_ON(is_data && refs_to_drop !=
-			       extent_data_ref_count(path, iref));
+			if (is_data && refs_to_drop !=
+			    extent_data_ref_count(path, iref)) {
+				btrfs_crit(info,
+		"invalid refs_to_drop, current refs %u refs_to_drop %u",
+					   extent_data_ref_count(path, iref),
+					   refs_to_drop);
+				btrfs_abort_transaction(trans, -EUCLEAN);
+				goto err_dump;
+			}
 			if (iref) {
-				BUG_ON(path->slots[0] != extent_slot);
+				if (path->slots[0] != extent_slot) {
+					btrfs_crit(info,
+"invalid iref, extent item key (%llu %u %llu) doesn't have wanted iref",
+						   key.objectid, key.type,
+						   key.offset);
+					btrfs_abort_transaction(trans, -EUCLEAN);
+					goto err_dump;
+				}
 			} else {
-				BUG_ON(path->slots[0] != extent_slot + 1);
+				/*
+				 * No inline ref, we must be at SHARED_* item,
+				 * And it's single ref, it must be:
+				 * |	extent_slot	  ||extent_slot + 1|
+				 * [ EXTENT/METADATA_ITEM ][ SHARED_* ITEM ]
+				 */
+				if (path->slots[0] != extent_slot + 1) {
+					btrfs_crit(info,
+	"invalid SHARED_* item, previous item is not EXTENT/METADATA_ITEM");
+					btrfs_abort_transaction(trans, -EUCLEAN);
+					goto err_dump;
+				}
 				path->slots[0] = extent_slot;
 				num_to_del = 2;
 			}
 		}
 
-		last_ref = 1;
 		ret = btrfs_del_items(trans, extent_root, path, path->slots[0],
 				      num_to_del);
 		if (ret) {
@@ -3180,32 +3177,26 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 		}
 		btrfs_release_path(path);
 
-		if (is_data) {
-			ret = btrfs_del_csums(trans, info->csum_root, bytenr,
-					      num_bytes);
-			if (ret) {
-				btrfs_abort_transaction(trans, ret);
-				goto out;
-			}
-		}
-
-		ret = add_to_free_space_tree(trans, bytenr, num_bytes);
-		if (ret) {
-			btrfs_abort_transaction(trans, ret);
-			goto out;
-		}
-
-		ret = btrfs_update_block_group(trans, bytenr, num_bytes, 0);
-		if (ret) {
-			btrfs_abort_transaction(trans, ret);
-			goto out;
-		}
+		ret = do_free_extent_accounting(trans, bytenr, num_bytes, is_data);
 	}
 	btrfs_release_path(path);
 
 out:
 	btrfs_free_path(path);
 	return ret;
+err_dump:
+	/*
+	 * Leaf dump can take up a lot of log buffer, so we only do full leaf
+	 * dump for debug build.
+	 */
+	if (IS_ENABLED(CONFIG_BTRFS_DEBUG)) {
+		btrfs_crit(info, "path->slots[0]=%d extent_slot=%d",
+			   path->slots[0], extent_slot);
+		btrfs_print_leaf(path->nodes[0]);
+	}
+
+	btrfs_free_path(path);
+	return -EUCLEAN;
 }
 
 /*
@@ -3264,44 +3255,67 @@ out_delayed_unlock:
 }
 
 void btrfs_free_tree_block(struct btrfs_trans_handle *trans,
-			   struct btrfs_root *root,
+			   u64 root_id,
 			   struct extent_buffer *buf,
 			   u64 parent, int last_ref)
 {
-	struct btrfs_fs_info *fs_info = root->fs_info;
+	struct btrfs_fs_info *fs_info = trans->fs_info;
 	struct btrfs_ref generic_ref = { 0 };
-	int pin = 1;
 	int ret;
 
 	btrfs_init_generic_ref(&generic_ref, BTRFS_DROP_DELAYED_REF,
 			       buf->start, buf->len, parent);
 	btrfs_init_tree_ref(&generic_ref, btrfs_header_level(buf),
-			    root->root_key.objectid);
-
-	if (root->root_key.objectid != BTRFS_TREE_LOG_OBJECTID) {
-		int old_ref_mod, new_ref_mod;
+			    root_id, 0, false);
 
+	if (root_id != BTRFS_TREE_LOG_OBJECTID) {
 		btrfs_ref_tree_mod(fs_info, &generic_ref);
-		ret = btrfs_add_delayed_tree_ref(trans, &generic_ref, NULL,
-						 &old_ref_mod, &new_ref_mod);
+		ret = btrfs_add_delayed_tree_ref(trans, &generic_ref, NULL);
 		BUG_ON(ret); /* -ENOMEM */
-		pin = old_ref_mod >= 0 && new_ref_mod < 0;
 	}
 
 	if (last_ref && btrfs_header_generation(buf) == trans->transid) {
 		struct btrfs_block_group *cache;
+		bool must_pin = false;
 
-		if (root->root_key.objectid != BTRFS_TREE_LOG_OBJECTID) {
+		if (root_id != BTRFS_TREE_LOG_OBJECTID) {
 			ret = check_ref_cleanup(trans, buf->start);
-			if (!ret)
+			if (!ret) {
+				btrfs_redirty_list_add(trans->transaction, buf);
 				goto out;
+			}
 		}
 
-		pin = 0;
 		cache = btrfs_lookup_block_group(fs_info, buf->start);
 
 		if (btrfs_header_flag(buf, BTRFS_HEADER_FLAG_WRITTEN)) {
-			pin_down_extent(cache, buf->start, buf->len, 1);
+			pin_down_extent(trans, cache, buf->start, buf->len, 1);
+			btrfs_put_block_group(cache);
+			goto out;
+		}
+
+		/*
+		 * If there are tree mod log users we may have recorded mod log
+		 * operations for this node.  If we re-allocate this node we
+		 * could replay operations on this node that happened when it
+		 * existed in a completely different root.  For example if it
+		 * was part of root A, then was reallocated to root B, and we
+		 * are doing a btrfs_old_search_slot(root b), we could replay
+		 * operations that happened when the block was part of root A,
+		 * giving us an inconsistent view of the btree.
+		 *
+		 * We are safe from races here because at this point no other
+		 * node or root points to this extent buffer, so if after this
+		 * check a new tree mod log user joins we will not have an
+		 * existing log of operations on this node that we have to
+		 * contend with.
+		 */
+		if (test_bit(BTRFS_FS_TREE_MOD_LOG_USERS, &fs_info->flags))
+			must_pin = true;
+
+		if (must_pin || btrfs_is_zoned(fs_info)) {
+			btrfs_redirty_list_add(trans->transaction, buf);
+			pin_down_extent(trans, cache, buf->start, buf->len, 1);
 			btrfs_put_block_group(cache);
 			goto out;
 		}
@@ -3314,9 +3328,6 @@ void btrfs_free_tree_block(struct btrfs_trans_handle *trans,
 		trace_btrfs_reserved_extent_free(fs_info, buf->start, buf->len);
 	}
 out:
-	if (pin)
-		add_pinned_bytes(fs_info, &generic_ref);
-
 	if (last_ref) {
 		/*
 		 * Deleting the buffer, clear the corrupt flag since it doesn't
@@ -3330,7 +3341,6 @@ out:
 int btrfs_free_extent(struct btrfs_trans_handle *trans, struct btrfs_ref *ref)
 {
 	struct btrfs_fs_info *fs_info = trans->fs_info;
-	int old_ref_mod, new_ref_mod;
 	int ret;
 
 	if (btrfs_is_testing(fs_info))
@@ -3341,30 +3351,24 @@ int btrfs_free_extent(struct btrfs_trans_handle *trans, struct btrfs_ref *ref)
 	 * tree, just update pinning info and exit early.
 	 */
 	if ((ref->type == BTRFS_REF_METADATA &&
-	     ref->tree_ref.root == BTRFS_TREE_LOG_OBJECTID) ||
+	     ref->tree_ref.owning_root == BTRFS_TREE_LOG_OBJECTID) ||
 	    (ref->type == BTRFS_REF_DATA &&
-	     ref->data_ref.ref_root == BTRFS_TREE_LOG_OBJECTID)) {
+	     ref->data_ref.owning_root == BTRFS_TREE_LOG_OBJECTID)) {
 		/* unlocks the pinned mutex */
-		btrfs_pin_extent(fs_info, ref->bytenr, ref->len, 1);
-		old_ref_mod = new_ref_mod = 0;
+		btrfs_pin_extent(trans, ref->bytenr, ref->len, 1);
 		ret = 0;
 	} else if (ref->type == BTRFS_REF_METADATA) {
-		ret = btrfs_add_delayed_tree_ref(trans, ref, NULL,
-						 &old_ref_mod, &new_ref_mod);
+		ret = btrfs_add_delayed_tree_ref(trans, ref, NULL);
 	} else {
-		ret = btrfs_add_delayed_data_ref(trans, ref, 0,
-						 &old_ref_mod, &new_ref_mod);
+		ret = btrfs_add_delayed_data_ref(trans, ref, 0);
 	}
 
 	if (!((ref->type == BTRFS_REF_METADATA &&
-	       ref->tree_ref.root == BTRFS_TREE_LOG_OBJECTID) ||
+	       ref->tree_ref.owning_root == BTRFS_TREE_LOG_OBJECTID) ||
 	      (ref->type == BTRFS_REF_DATA &&
-	       ref->data_ref.ref_root == BTRFS_TREE_LOG_OBJECTID)))
+	       ref->data_ref.owning_root == BTRFS_TREE_LOG_OBJECTID)))
 		btrfs_ref_tree_mod(fs_info, ref);
 
-	if (ret == 0 && old_ref_mod >= 0 && new_ref_mod < 0)
-		add_pinned_bytes(fs_info, ref);
-
 	return ret;
 }
 
@@ -3395,6 +3399,7 @@ static struct btrfs_block_group *btrfs_lock_cluster(
 		   struct btrfs_block_group *block_group,
 		   struct btrfs_free_cluster *cluster,
 		   int delalloc)
+	__acquires(&cluster->refill_lock)
 {
 	struct btrfs_block_group *used_bg = NULL;
 
@@ -3438,13 +3443,20 @@ btrfs_release_block_group(struct btrfs_block_group *cache,
 	btrfs_put_block_group(cache);
 }
 
+enum btrfs_extent_allocation_policy {
+	BTRFS_EXTENT_ALLOC_CLUSTERED,
+	BTRFS_EXTENT_ALLOC_ZONED,
+};
+
 /*
  * Structure used internally for find_free_extent() function.  Wraps needed
  * parameters.
  */
 struct find_free_extent_ctl {
 	/* Basic allocation info */
+	u64 ram_bytes;
 	u64 num_bytes;
+	u64 min_alloc_size;
 	u64 empty_size;
 	u64 flags;
 	int delalloc;
@@ -3454,10 +3466,18 @@ struct find_free_extent_ctl {
 
 	/* For clustered allocation */
 	u64 empty_cluster;
+	struct btrfs_free_cluster *last_ptr;
+	bool use_cluster;
 
 	bool have_caching_bg;
 	bool orig_have_caching_bg;
 
+	/* Allocation is called for tree-log */
+	bool for_treelog;
+
+	/* Allocation is called for data relocation */
+	bool for_data_reloc;
+
 	/* RAID index, converted from flags */
 	int index;
 
@@ -3489,6 +3509,12 @@ struct find_free_extent_ctl {
 
 	/* Found result */
 	u64 found_offset;
+
+	/* Hint where to start looking for an empty space */
+	u64 hint_byte;
+
+	/* Allocation policy */
+	enum btrfs_extent_allocation_policy policy;
 };
 
 
@@ -3501,11 +3527,11 @@ struct find_free_extent_ctl {
  * Return 0 means we have found a location and set ffe_ctl->found_offset.
  */
 static int find_free_extent_clustered(struct btrfs_block_group *bg,
-		struct btrfs_free_cluster *last_ptr,
-		struct find_free_extent_ctl *ffe_ctl,
-		struct btrfs_block_group **cluster_bg_ret)
+				      struct find_free_extent_ctl *ffe_ctl,
+				      struct btrfs_block_group **cluster_bg_ret)
 {
 	struct btrfs_block_group *cluster_bg;
+	struct btrfs_free_cluster *last_ptr = ffe_ctl->last_ptr;
 	u64 aligned_cluster;
 	u64 offset;
 	int ret;
@@ -3605,9 +3631,9 @@ refill_cluster:
  * Return -EAGAIN to inform caller that we need to re-search this block group
  */
 static int find_free_extent_unclustered(struct btrfs_block_group *bg,
-		struct btrfs_free_cluster *last_ptr,
-		struct find_free_extent_ctl *ffe_ctl)
+					struct find_free_extent_ctl *ffe_ctl)
 {
+	struct btrfs_free_cluster *last_ptr = ffe_ctl->last_ptr;
 	u64 offset;
 
 	/*
@@ -3663,40 +3689,382 @@ static int find_free_extent_unclustered(struct btrfs_block_group *bg,
 	return 0;
 }
 
+static int do_allocation_clustered(struct btrfs_block_group *block_group,
+				   struct find_free_extent_ctl *ffe_ctl,
+				   struct btrfs_block_group **bg_ret)
+{
+	int ret;
+
+	/* We want to try and use the cluster allocator, so lets look there */
+	if (ffe_ctl->last_ptr && ffe_ctl->use_cluster) {
+		ret = find_free_extent_clustered(block_group, ffe_ctl, bg_ret);
+		if (ret >= 0 || ret == -EAGAIN)
+			return ret;
+		/* ret == -ENOENT case falls through */
+	}
+
+	return find_free_extent_unclustered(block_group, ffe_ctl);
+}
+
+/*
+ * Tree-log block group locking
+ * ============================
+ *
+ * fs_info::treelog_bg_lock protects the fs_info::treelog_bg which
+ * indicates the starting address of a block group, which is reserved only
+ * for tree-log metadata.
+ *
+ * Lock nesting
+ * ============
+ *
+ * space_info::lock
+ *   block_group::lock
+ *     fs_info::treelog_bg_lock
+ */
+
+/*
+ * Simple allocator for sequential-only block group. It only allows sequential
+ * allocation. No need to play with trees. This function also reserves the
+ * bytes as in btrfs_add_reserved_bytes.
+ */
+static int do_allocation_zoned(struct btrfs_block_group *block_group,
+			       struct find_free_extent_ctl *ffe_ctl,
+			       struct btrfs_block_group **bg_ret)
+{
+	struct btrfs_fs_info *fs_info = block_group->fs_info;
+	struct btrfs_space_info *space_info = block_group->space_info;
+	struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
+	u64 start = block_group->start;
+	u64 num_bytes = ffe_ctl->num_bytes;
+	u64 avail;
+	u64 bytenr = block_group->start;
+	u64 log_bytenr;
+	u64 data_reloc_bytenr;
+	int ret = 0;
+	bool skip = false;
+
+	ASSERT(btrfs_is_zoned(block_group->fs_info));
+
+	/*
+	 * Do not allow non-tree-log blocks in the dedicated tree-log block
+	 * group, and vice versa.
+	 */
+	spin_lock(&fs_info->treelog_bg_lock);
+	log_bytenr = fs_info->treelog_bg;
+	if (log_bytenr && ((ffe_ctl->for_treelog && bytenr != log_bytenr) ||
+			   (!ffe_ctl->for_treelog && bytenr == log_bytenr)))
+		skip = true;
+	spin_unlock(&fs_info->treelog_bg_lock);
+	if (skip)
+		return 1;
+
+	/*
+	 * Do not allow non-relocation blocks in the dedicated relocation block
+	 * group, and vice versa.
+	 */
+	spin_lock(&fs_info->relocation_bg_lock);
+	data_reloc_bytenr = fs_info->data_reloc_bg;
+	if (data_reloc_bytenr &&
+	    ((ffe_ctl->for_data_reloc && bytenr != data_reloc_bytenr) ||
+	     (!ffe_ctl->for_data_reloc && bytenr == data_reloc_bytenr)))
+		skip = true;
+	spin_unlock(&fs_info->relocation_bg_lock);
+	if (skip)
+		return 1;
+
+	/* Check RO and no space case before trying to activate it */
+	spin_lock(&block_group->lock);
+	if (block_group->ro || btrfs_zoned_bg_is_full(block_group)) {
+		ret = 1;
+		/*
+		 * May need to clear fs_info->{treelog,data_reloc}_bg.
+		 * Return the error after taking the locks.
+		 */
+	}
+	spin_unlock(&block_group->lock);
+
+	if (!ret && !btrfs_zone_activate(block_group)) {
+		ret = 1;
+		/*
+		 * May need to clear fs_info->{treelog,data_reloc}_bg.
+		 * Return the error after taking the locks.
+		 */
+	}
+
+	spin_lock(&space_info->lock);
+	spin_lock(&block_group->lock);
+	spin_lock(&fs_info->treelog_bg_lock);
+	spin_lock(&fs_info->relocation_bg_lock);
+
+	if (ret)
+		goto out;
+
+	ASSERT(!ffe_ctl->for_treelog ||
+	       block_group->start == fs_info->treelog_bg ||
+	       fs_info->treelog_bg == 0);
+	ASSERT(!ffe_ctl->for_data_reloc ||
+	       block_group->start == fs_info->data_reloc_bg ||
+	       fs_info->data_reloc_bg == 0);
+
+	if (block_group->ro ||
+	    test_bit(BLOCK_GROUP_FLAG_ZONED_DATA_RELOC, &block_group->runtime_flags)) {
+		ret = 1;
+		goto out;
+	}
+
+	/*
+	 * Do not allow currently using block group to be tree-log dedicated
+	 * block group.
+	 */
+	if (ffe_ctl->for_treelog && !fs_info->treelog_bg &&
+	    (block_group->used || block_group->reserved)) {
+		ret = 1;
+		goto out;
+	}
+
+	/*
+	 * Do not allow currently used block group to be the data relocation
+	 * dedicated block group.
+	 */
+	if (ffe_ctl->for_data_reloc && !fs_info->data_reloc_bg &&
+	    (block_group->used || block_group->reserved)) {
+		ret = 1;
+		goto out;
+	}
+
+	WARN_ON_ONCE(block_group->alloc_offset > block_group->zone_capacity);
+	avail = block_group->zone_capacity - block_group->alloc_offset;
+	if (avail < num_bytes) {
+		if (ffe_ctl->max_extent_size < avail) {
+			/*
+			 * With sequential allocator, free space is always
+			 * contiguous
+			 */
+			ffe_ctl->max_extent_size = avail;
+			ffe_ctl->total_free_space = avail;
+		}
+		ret = 1;
+		goto out;
+	}
+
+	if (ffe_ctl->for_treelog && !fs_info->treelog_bg)
+		fs_info->treelog_bg = block_group->start;
+
+	if (ffe_ctl->for_data_reloc && !fs_info->data_reloc_bg)
+		fs_info->data_reloc_bg = block_group->start;
+
+	ffe_ctl->found_offset = start + block_group->alloc_offset;
+	block_group->alloc_offset += num_bytes;
+	spin_lock(&ctl->tree_lock);
+	ctl->free_space -= num_bytes;
+	spin_unlock(&ctl->tree_lock);
+
+	/*
+	 * We do not check if found_offset is aligned to stripesize. The
+	 * address is anyway rewritten when using zone append writing.
+	 */
+
+	ffe_ctl->search_start = ffe_ctl->found_offset;
+
+out:
+	if (ret && ffe_ctl->for_treelog)
+		fs_info->treelog_bg = 0;
+	if (ret && ffe_ctl->for_data_reloc &&
+	    fs_info->data_reloc_bg == block_group->start) {
+		/*
+		 * Do not allow further allocations from this block group.
+		 * Compared to increasing the ->ro, setting the
+		 * ->zoned_data_reloc_ongoing flag still allows nocow
+		 *  writers to come in. See btrfs_inc_nocow_writers().
+		 *
+		 * We need to disable an allocation to avoid an allocation of
+		 * regular (non-relocation data) extent. With mix of relocation
+		 * extents and regular extents, we can dispatch WRITE commands
+		 * (for relocation extents) and ZONE APPEND commands (for
+		 * regular extents) at the same time to the same zone, which
+		 * easily break the write pointer.
+		 */
+		set_bit(BLOCK_GROUP_FLAG_ZONED_DATA_RELOC, &block_group->runtime_flags);
+		fs_info->data_reloc_bg = 0;
+	}
+	spin_unlock(&fs_info->relocation_bg_lock);
+	spin_unlock(&fs_info->treelog_bg_lock);
+	spin_unlock(&block_group->lock);
+	spin_unlock(&space_info->lock);
+	return ret;
+}
+
+static int do_allocation(struct btrfs_block_group *block_group,
+			 struct find_free_extent_ctl *ffe_ctl,
+			 struct btrfs_block_group **bg_ret)
+{
+	switch (ffe_ctl->policy) {
+	case BTRFS_EXTENT_ALLOC_CLUSTERED:
+		return do_allocation_clustered(block_group, ffe_ctl, bg_ret);
+	case BTRFS_EXTENT_ALLOC_ZONED:
+		return do_allocation_zoned(block_group, ffe_ctl, bg_ret);
+	default:
+		BUG();
+	}
+}
+
+static void release_block_group(struct btrfs_block_group *block_group,
+				struct find_free_extent_ctl *ffe_ctl,
+				int delalloc)
+{
+	switch (ffe_ctl->policy) {
+	case BTRFS_EXTENT_ALLOC_CLUSTERED:
+		ffe_ctl->retry_clustered = false;
+		ffe_ctl->retry_unclustered = false;
+		break;
+	case BTRFS_EXTENT_ALLOC_ZONED:
+		/* Nothing to do */
+		break;
+	default:
+		BUG();
+	}
+
+	BUG_ON(btrfs_bg_flags_to_raid_index(block_group->flags) !=
+	       ffe_ctl->index);
+	btrfs_release_block_group(block_group, delalloc);
+}
+
+static void found_extent_clustered(struct find_free_extent_ctl *ffe_ctl,
+				   struct btrfs_key *ins)
+{
+	struct btrfs_free_cluster *last_ptr = ffe_ctl->last_ptr;
+
+	if (!ffe_ctl->use_cluster && last_ptr) {
+		spin_lock(&last_ptr->lock);
+		last_ptr->window_start = ins->objectid;
+		spin_unlock(&last_ptr->lock);
+	}
+}
+
+static void found_extent(struct find_free_extent_ctl *ffe_ctl,
+			 struct btrfs_key *ins)
+{
+	switch (ffe_ctl->policy) {
+	case BTRFS_EXTENT_ALLOC_CLUSTERED:
+		found_extent_clustered(ffe_ctl, ins);
+		break;
+	case BTRFS_EXTENT_ALLOC_ZONED:
+		/* Nothing to do */
+		break;
+	default:
+		BUG();
+	}
+}
+
+static int can_allocate_chunk_zoned(struct btrfs_fs_info *fs_info,
+				    struct find_free_extent_ctl *ffe_ctl)
+{
+	/* If we can activate new zone, just allocate a chunk and use it */
+	if (btrfs_can_activate_zone(fs_info->fs_devices, ffe_ctl->flags))
+		return 0;
+
+	/*
+	 * We already reached the max active zones. Try to finish one block
+	 * group to make a room for a new block group. This is only possible
+	 * for a data block group because btrfs_zone_finish() may need to wait
+	 * for a running transaction which can cause a deadlock for metadata
+	 * allocation.
+	 */
+	if (ffe_ctl->flags & BTRFS_BLOCK_GROUP_DATA) {
+		int ret = btrfs_zone_finish_one_bg(fs_info);
+
+		if (ret == 1)
+			return 0;
+		else if (ret < 0)
+			return ret;
+	}
+
+	/*
+	 * If we have enough free space left in an already active block group
+	 * and we can't activate any other zone now, do not allow allocating a
+	 * new chunk and let find_free_extent() retry with a smaller size.
+	 */
+	if (ffe_ctl->max_extent_size >= ffe_ctl->min_alloc_size)
+		return -ENOSPC;
+
+	/*
+	 * Even min_alloc_size is not left in any block groups. Since we cannot
+	 * activate a new block group, allocating it may not help. Let's tell a
+	 * caller to try again and hope it progress something by writing some
+	 * parts of the region. That is only possible for data block groups,
+	 * where a part of the region can be written.
+	 */
+	if (ffe_ctl->flags & BTRFS_BLOCK_GROUP_DATA)
+		return -EAGAIN;
+
+	/*
+	 * We cannot activate a new block group and no enough space left in any
+	 * block groups. So, allocating a new block group may not help. But,
+	 * there is nothing to do anyway, so let's go with it.
+	 */
+	return 0;
+}
+
+static int can_allocate_chunk(struct btrfs_fs_info *fs_info,
+			      struct find_free_extent_ctl *ffe_ctl)
+{
+	switch (ffe_ctl->policy) {
+	case BTRFS_EXTENT_ALLOC_CLUSTERED:
+		return 0;
+	case BTRFS_EXTENT_ALLOC_ZONED:
+		return can_allocate_chunk_zoned(fs_info, ffe_ctl);
+	default:
+		BUG();
+	}
+}
+
+static int chunk_allocation_failed(struct find_free_extent_ctl *ffe_ctl)
+{
+	switch (ffe_ctl->policy) {
+	case BTRFS_EXTENT_ALLOC_CLUSTERED:
+		/*
+		 * If we can't allocate a new chunk we've already looped through
+		 * at least once, move on to the NO_EMPTY_SIZE case.
+		 */
+		ffe_ctl->loop = LOOP_NO_EMPTY_SIZE;
+		return 0;
+	case BTRFS_EXTENT_ALLOC_ZONED:
+		/* Give up here */
+		return -ENOSPC;
+	default:
+		BUG();
+	}
+}
+
 /*
  * Return >0 means caller needs to re-search for free extent
  * Return 0 means we have the needed free extent.
  * Return <0 means we failed to locate any free extent.
  */
 static int find_free_extent_update_loop(struct btrfs_fs_info *fs_info,
-					struct btrfs_free_cluster *last_ptr,
 					struct btrfs_key *ins,
 					struct find_free_extent_ctl *ffe_ctl,
-					int full_search, bool use_cluster)
+					bool full_search)
 {
-	struct btrfs_root *root = fs_info->extent_root;
+	struct btrfs_root *root = fs_info->chunk_root;
 	int ret;
 
 	if ((ffe_ctl->loop == LOOP_CACHING_NOWAIT) &&
 	    ffe_ctl->have_caching_bg && !ffe_ctl->orig_have_caching_bg)
 		ffe_ctl->orig_have_caching_bg = true;
 
-	if (!ins->objectid && ffe_ctl->loop >= LOOP_CACHING_WAIT &&
-	    ffe_ctl->have_caching_bg)
-		return 1;
-
-	if (!ins->objectid && ++(ffe_ctl->index) < BTRFS_NR_RAID_TYPES)
-		return 1;
-
 	if (ins->objectid) {
-		if (!use_cluster && last_ptr) {
-			spin_lock(&last_ptr->lock);
-			last_ptr->window_start = ins->objectid;
-			spin_unlock(&last_ptr->lock);
-		}
+		found_extent(ffe_ctl, ins);
 		return 0;
 	}
 
+	if (ffe_ctl->loop >= LOOP_CACHING_WAIT && ffe_ctl->have_caching_bg)
+		return 1;
+
+	ffe_ctl->index++;
+	if (ffe_ctl->index < BTRFS_NR_RAID_TYPES)
+		return 1;
+
 	/*
 	 * LOOP_CACHING_NOWAIT, search partially cached block groups, kicking
 	 *			caching kthreads as we move along
@@ -3725,6 +4093,11 @@ static int find_free_extent_update_loop(struct btrfs_fs_info *fs_info,
 			struct btrfs_trans_handle *trans;
 			int exist = 0;
 
+			/*Check if allocation policy allows to create a new chunk */
+			ret = can_allocate_chunk(fs_info, ffe_ctl);
+			if (ret)
+				return ret;
+
 			trans = current->journal_info;
 			if (trans)
 				exist = 1;
@@ -3737,18 +4110,12 @@ static int find_free_extent_update_loop(struct btrfs_fs_info *fs_info,
 			}
 
 			ret = btrfs_chunk_alloc(trans, ffe_ctl->flags,
-						CHUNK_ALLOC_FORCE);
-
-			/*
-			 * If we can't allocate a new chunk we've already looped
-			 * through at least once, move on to the NO_EMPTY_SIZE
-			 * case.
-			 */
-			if (ret == -ENOSPC)
-				ffe_ctl->loop = LOOP_NO_EMPTY_SIZE;
+						CHUNK_ALLOC_FORCE_FOR_EXTENT);
 
 			/* Do not bail out on ENOSPC since we can do more. */
-			if (ret < 0 && ret != -ENOSPC)
+			if (ret == -ENOSPC)
+				ret = chunk_allocation_failed(ffe_ctl);
+			else if (ret < 0)
 				btrfs_abort_transaction(trans, ret);
 			else
 				ret = 0;
@@ -3759,6 +4126,9 @@ static int find_free_extent_update_loop(struct btrfs_fs_info *fs_info,
 		}
 
 		if (ffe_ctl->loop == LOOP_NO_EMPTY_SIZE) {
+			if (ffe_ctl->policy != BTRFS_EXTENT_ALLOC_CLUSTERED)
+				return -ENOSPC;
+
 			/*
 			 * Don't loop again if we already have no empty_size and
 			 * no empty_cluster.
@@ -3774,6 +4144,85 @@ static int find_free_extent_update_loop(struct btrfs_fs_info *fs_info,
 	return -ENOSPC;
 }
 
+static int prepare_allocation_clustered(struct btrfs_fs_info *fs_info,
+					struct find_free_extent_ctl *ffe_ctl,
+					struct btrfs_space_info *space_info,
+					struct btrfs_key *ins)
+{
+	/*
+	 * If our free space is heavily fragmented we may not be able to make
+	 * big contiguous allocations, so instead of doing the expensive search
+	 * for free space, simply return ENOSPC with our max_extent_size so we
+	 * can go ahead and search for a more manageable chunk.
+	 *
+	 * If our max_extent_size is large enough for our allocation simply
+	 * disable clustering since we will likely not be able to find enough
+	 * space to create a cluster and induce latency trying.
+	 */
+	if (space_info->max_extent_size) {
+		spin_lock(&space_info->lock);
+		if (space_info->max_extent_size &&
+		    ffe_ctl->num_bytes > space_info->max_extent_size) {
+			ins->offset = space_info->max_extent_size;
+			spin_unlock(&space_info->lock);
+			return -ENOSPC;
+		} else if (space_info->max_extent_size) {
+			ffe_ctl->use_cluster = false;
+		}
+		spin_unlock(&space_info->lock);
+	}
+
+	ffe_ctl->last_ptr = fetch_cluster_info(fs_info, space_info,
+					       &ffe_ctl->empty_cluster);
+	if (ffe_ctl->last_ptr) {
+		struct btrfs_free_cluster *last_ptr = ffe_ctl->last_ptr;
+
+		spin_lock(&last_ptr->lock);
+		if (last_ptr->block_group)
+			ffe_ctl->hint_byte = last_ptr->window_start;
+		if (last_ptr->fragmented) {
+			/*
+			 * We still set window_start so we can keep track of the
+			 * last place we found an allocation to try and save
+			 * some time.
+			 */
+			ffe_ctl->hint_byte = last_ptr->window_start;
+			ffe_ctl->use_cluster = false;
+		}
+		spin_unlock(&last_ptr->lock);
+	}
+
+	return 0;
+}
+
+static int prepare_allocation(struct btrfs_fs_info *fs_info,
+			      struct find_free_extent_ctl *ffe_ctl,
+			      struct btrfs_space_info *space_info,
+			      struct btrfs_key *ins)
+{
+	switch (ffe_ctl->policy) {
+	case BTRFS_EXTENT_ALLOC_CLUSTERED:
+		return prepare_allocation_clustered(fs_info, ffe_ctl,
+						    space_info, ins);
+	case BTRFS_EXTENT_ALLOC_ZONED:
+		if (ffe_ctl->for_treelog) {
+			spin_lock(&fs_info->treelog_bg_lock);
+			if (fs_info->treelog_bg)
+				ffe_ctl->hint_byte = fs_info->treelog_bg;
+			spin_unlock(&fs_info->treelog_bg_lock);
+		}
+		if (ffe_ctl->for_data_reloc) {
+			spin_lock(&fs_info->relocation_bg_lock);
+			if (fs_info->data_reloc_bg)
+				ffe_ctl->hint_byte = fs_info->data_reloc_bg;
+			spin_unlock(&fs_info->relocation_bg_lock);
+		}
+		return 0;
+	default:
+		BUG();
+	}
+}
+
 /*
  * walks the btree of allocated extents and find a hole of a given size.
  * The key ins is changed to record the hole:
@@ -3799,93 +4248,63 @@ static int find_free_extent_update_loop(struct btrfs_fs_info *fs_info,
  * |- Push harder to find free extents
  *    |- If not found, re-iterate all block groups
  */
-static noinline int find_free_extent(struct btrfs_fs_info *fs_info,
-				u64 ram_bytes, u64 num_bytes, u64 empty_size,
-				u64 hint_byte, struct btrfs_key *ins,
-				u64 flags, int delalloc)
+static noinline int find_free_extent(struct btrfs_root *root,
+				     struct btrfs_key *ins,
+				     struct find_free_extent_ctl *ffe_ctl)
 {
+	struct btrfs_fs_info *fs_info = root->fs_info;
 	int ret = 0;
 	int cache_block_group_error = 0;
-	struct btrfs_free_cluster *last_ptr = NULL;
 	struct btrfs_block_group *block_group = NULL;
-	struct find_free_extent_ctl ffe_ctl = {0};
 	struct btrfs_space_info *space_info;
-	bool use_cluster = true;
 	bool full_search = false;
 
-	WARN_ON(num_bytes < fs_info->sectorsize);
+	WARN_ON(ffe_ctl->num_bytes < fs_info->sectorsize);
 
-	ffe_ctl.num_bytes = num_bytes;
-	ffe_ctl.empty_size = empty_size;
-	ffe_ctl.flags = flags;
-	ffe_ctl.search_start = 0;
-	ffe_ctl.retry_clustered = false;
-	ffe_ctl.retry_unclustered = false;
-	ffe_ctl.delalloc = delalloc;
-	ffe_ctl.index = btrfs_bg_flags_to_raid_index(flags);
-	ffe_ctl.have_caching_bg = false;
-	ffe_ctl.orig_have_caching_bg = false;
-	ffe_ctl.found_offset = 0;
+	ffe_ctl->search_start = 0;
+	/* For clustered allocation */
+	ffe_ctl->empty_cluster = 0;
+	ffe_ctl->last_ptr = NULL;
+	ffe_ctl->use_cluster = true;
+	ffe_ctl->have_caching_bg = false;
+	ffe_ctl->orig_have_caching_bg = false;
+	ffe_ctl->index = btrfs_bg_flags_to_raid_index(ffe_ctl->flags);
+	ffe_ctl->loop = 0;
+	/* For clustered allocation */
+	ffe_ctl->retry_clustered = false;
+	ffe_ctl->retry_unclustered = false;
+	ffe_ctl->cached = 0;
+	ffe_ctl->max_extent_size = 0;
+	ffe_ctl->total_free_space = 0;
+	ffe_ctl->found_offset = 0;
+	ffe_ctl->policy = BTRFS_EXTENT_ALLOC_CLUSTERED;
+
+	if (btrfs_is_zoned(fs_info))
+		ffe_ctl->policy = BTRFS_EXTENT_ALLOC_ZONED;
 
 	ins->type = BTRFS_EXTENT_ITEM_KEY;
 	ins->objectid = 0;
 	ins->offset = 0;
 
-	trace_find_free_extent(fs_info, num_bytes, empty_size, flags);
+	trace_find_free_extent(root, ffe_ctl->num_bytes, ffe_ctl->empty_size,
+			       ffe_ctl->flags);
 
-	space_info = btrfs_find_space_info(fs_info, flags);
+	space_info = btrfs_find_space_info(fs_info, ffe_ctl->flags);
 	if (!space_info) {
-		btrfs_err(fs_info, "No space info for %llu", flags);
+		btrfs_err(fs_info, "No space info for %llu", ffe_ctl->flags);
 		return -ENOSPC;
 	}
 
-	/*
-	 * If our free space is heavily fragmented we may not be able to make
-	 * big contiguous allocations, so instead of doing the expensive search
-	 * for free space, simply return ENOSPC with our max_extent_size so we
-	 * can go ahead and search for a more manageable chunk.
-	 *
-	 * If our max_extent_size is large enough for our allocation simply
-	 * disable clustering since we will likely not be able to find enough
-	 * space to create a cluster and induce latency trying.
-	 */
-	if (unlikely(space_info->max_extent_size)) {
-		spin_lock(&space_info->lock);
-		if (space_info->max_extent_size &&
-		    num_bytes > space_info->max_extent_size) {
-			ins->offset = space_info->max_extent_size;
-			spin_unlock(&space_info->lock);
-			return -ENOSPC;
-		} else if (space_info->max_extent_size) {
-			use_cluster = false;
-		}
-		spin_unlock(&space_info->lock);
-	}
-
-	last_ptr = fetch_cluster_info(fs_info, space_info,
-				      &ffe_ctl.empty_cluster);
-	if (last_ptr) {
-		spin_lock(&last_ptr->lock);
-		if (last_ptr->block_group)
-			hint_byte = last_ptr->window_start;
-		if (last_ptr->fragmented) {
-			/*
-			 * We still set window_start so we can keep track of the
-			 * last place we found an allocation to try and save
-			 * some time.
-			 */
-			hint_byte = last_ptr->window_start;
-			use_cluster = false;
-		}
-		spin_unlock(&last_ptr->lock);
-	}
+	ret = prepare_allocation(fs_info, ffe_ctl, space_info, ins);
+	if (ret < 0)
+		return ret;
 
-	ffe_ctl.search_start = max(ffe_ctl.search_start,
-				   first_logical_byte(fs_info, 0));
-	ffe_ctl.search_start = max(ffe_ctl.search_start, hint_byte);
-	if (ffe_ctl.search_start == hint_byte) {
+	ffe_ctl->search_start = max(ffe_ctl->search_start,
+				    first_logical_byte(fs_info));
+	ffe_ctl->search_start = max(ffe_ctl->search_start, ffe_ctl->hint_byte);
+	if (ffe_ctl->search_start == ffe_ctl->hint_byte) {
 		block_group = btrfs_lookup_block_group(fs_info,
-						       ffe_ctl.search_start);
+						       ffe_ctl->search_start);
 		/*
 		 * we don't want to use the block group if it doesn't match our
 		 * allocation bits, or if its not cached.
@@ -3893,7 +4312,7 @@ static noinline int find_free_extent(struct btrfs_fs_info *fs_info,
 		 * However if we are re-searching with an ideal block group
 		 * picked out then we don't care that the block group is cached.
 		 */
-		if (block_group && block_group_bits(block_group, flags) &&
+		if (block_group && block_group_bits(block_group, ffe_ctl->flags) &&
 		    block_group->cached != BTRFS_CACHE_NO) {
 			down_read(&space_info->groups_sem);
 			if (list_empty(&block_group->list) ||
@@ -3907,9 +4326,10 @@ static noinline int find_free_extent(struct btrfs_fs_info *fs_info,
 				btrfs_put_block_group(block_group);
 				up_read(&space_info->groups_sem);
 			} else {
-				ffe_ctl.index = btrfs_bg_flags_to_raid_index(
-						block_group->flags);
-				btrfs_lock_block_group(block_group, delalloc);
+				ffe_ctl->index = btrfs_bg_flags_to_raid_index(
+							block_group->flags);
+				btrfs_lock_block_group(block_group,
+						       ffe_ctl->delalloc);
 				goto have_block_group;
 			}
 		} else if (block_group) {
@@ -3917,26 +4337,33 @@ static noinline int find_free_extent(struct btrfs_fs_info *fs_info,
 		}
 	}
 search:
-	ffe_ctl.have_caching_bg = false;
-	if (ffe_ctl.index == btrfs_bg_flags_to_raid_index(flags) ||
-	    ffe_ctl.index == 0)
+	ffe_ctl->have_caching_bg = false;
+	if (ffe_ctl->index == btrfs_bg_flags_to_raid_index(ffe_ctl->flags) ||
+	    ffe_ctl->index == 0)
 		full_search = true;
 	down_read(&space_info->groups_sem);
 	list_for_each_entry(block_group,
-			    &space_info->block_groups[ffe_ctl.index], list) {
+			    &space_info->block_groups[ffe_ctl->index], list) {
+		struct btrfs_block_group *bg_ret;
+
 		/* If the block group is read-only, we can skip it entirely. */
-		if (unlikely(block_group->ro))
+		if (unlikely(block_group->ro)) {
+			if (ffe_ctl->for_treelog)
+				btrfs_clear_treelog_bg(block_group);
+			if (ffe_ctl->for_data_reloc)
+				btrfs_clear_data_reloc_bg(block_group);
 			continue;
+		}
 
-		btrfs_grab_block_group(block_group, delalloc);
-		ffe_ctl.search_start = block_group->start;
+		btrfs_grab_block_group(block_group, ffe_ctl->delalloc);
+		ffe_ctl->search_start = block_group->start;
 
 		/*
 		 * this can happen if we end up cycling through all the
 		 * raid types, but we want to make sure we only allocate
 		 * for the proper type.
 		 */
-		if (!block_group_bits(block_group, flags)) {
+		if (!block_group_bits(block_group, ffe_ctl->flags)) {
 			u64 extra = BTRFS_BLOCK_GROUP_DUP |
 				BTRFS_BLOCK_GROUP_RAID1_MASK |
 				BTRFS_BLOCK_GROUP_RAID56_MASK |
@@ -3947,7 +4374,7 @@ search:
 			 * doesn't provide them, bail.  This does allow us to
 			 * fill raid0 from raid1.
 			 */
-			if ((flags & extra) && !(block_group->flags & extra))
+			if ((ffe_ctl->flags & extra) && !(block_group->flags & extra))
 				goto loop;
 
 			/*
@@ -3955,15 +4382,15 @@ search:
 			 * It's possible that we have MIXED_GROUP flag but no
 			 * block group is mixed.  Just skip such block group.
 			 */
-			btrfs_release_block_group(block_group, delalloc);
+			btrfs_release_block_group(block_group, ffe_ctl->delalloc);
 			continue;
 		}
 
 have_block_group:
-		ffe_ctl.cached = btrfs_block_group_done(block_group);
-		if (unlikely(!ffe_ctl.cached)) {
-			ffe_ctl.have_caching_bg = true;
-			ret = btrfs_cache_block_group(block_group, 0);
+		ffe_ctl->cached = btrfs_block_group_done(block_group);
+		if (unlikely(!ffe_ctl->cached)) {
+			ffe_ctl->have_caching_bg = true;
+			ret = btrfs_cache_block_group(block_group, false);
 
 			/*
 			 * If we get ENOMEM here or something else we want to
@@ -3984,83 +4411,64 @@ have_block_group:
 		if (unlikely(block_group->cached == BTRFS_CACHE_ERROR))
 			goto loop;
 
-		/*
-		 * Ok we want to try and use the cluster allocator, so
-		 * lets look there
-		 */
-		if (last_ptr && use_cluster) {
-			struct btrfs_block_group *cluster_bg = NULL;
-
-			ret = find_free_extent_clustered(block_group, last_ptr,
-							 &ffe_ctl, &cluster_bg);
-
-			if (ret == 0) {
-				if (cluster_bg && cluster_bg != block_group) {
-					btrfs_release_block_group(block_group,
-								  delalloc);
-					block_group = cluster_bg;
-				}
-				goto checks;
-			} else if (ret == -EAGAIN) {
-				goto have_block_group;
-			} else if (ret > 0) {
-				goto loop;
+		bg_ret = NULL;
+		ret = do_allocation(block_group, ffe_ctl, &bg_ret);
+		if (ret == 0) {
+			if (bg_ret && bg_ret != block_group) {
+				btrfs_release_block_group(block_group,
+							  ffe_ctl->delalloc);
+				block_group = bg_ret;
 			}
-			/* ret == -ENOENT case falls through */
-		}
-
-		ret = find_free_extent_unclustered(block_group, last_ptr,
-						   &ffe_ctl);
-		if (ret == -EAGAIN)
+		} else if (ret == -EAGAIN) {
 			goto have_block_group;
-		else if (ret > 0)
+		} else if (ret > 0) {
 			goto loop;
-		/* ret == 0 case falls through */
-checks:
-		ffe_ctl.search_start = round_up(ffe_ctl.found_offset,
-					     fs_info->stripesize);
+		}
+
+		/* Checks */
+		ffe_ctl->search_start = round_up(ffe_ctl->found_offset,
+						 fs_info->stripesize);
 
 		/* move on to the next group */
-		if (ffe_ctl.search_start + num_bytes >
+		if (ffe_ctl->search_start + ffe_ctl->num_bytes >
 		    block_group->start + block_group->length) {
-			btrfs_add_free_space(block_group, ffe_ctl.found_offset,
-					     num_bytes);
+			btrfs_add_free_space_unused(block_group,
+					    ffe_ctl->found_offset,
+					    ffe_ctl->num_bytes);
 			goto loop;
 		}
 
-		if (ffe_ctl.found_offset < ffe_ctl.search_start)
-			btrfs_add_free_space(block_group, ffe_ctl.found_offset,
-				ffe_ctl.search_start - ffe_ctl.found_offset);
+		if (ffe_ctl->found_offset < ffe_ctl->search_start)
+			btrfs_add_free_space_unused(block_group,
+					ffe_ctl->found_offset,
+					ffe_ctl->search_start - ffe_ctl->found_offset);
 
-		ret = btrfs_add_reserved_bytes(block_group, ram_bytes,
-				num_bytes, delalloc);
+		ret = btrfs_add_reserved_bytes(block_group, ffe_ctl->ram_bytes,
+					       ffe_ctl->num_bytes,
+					       ffe_ctl->delalloc);
 		if (ret == -EAGAIN) {
-			btrfs_add_free_space(block_group, ffe_ctl.found_offset,
-					     num_bytes);
+			btrfs_add_free_space_unused(block_group,
+					ffe_ctl->found_offset,
+					ffe_ctl->num_bytes);
 			goto loop;
 		}
 		btrfs_inc_block_group_reservations(block_group);
 
 		/* we are all good, lets return */
-		ins->objectid = ffe_ctl.search_start;
-		ins->offset = num_bytes;
+		ins->objectid = ffe_ctl->search_start;
+		ins->offset = ffe_ctl->num_bytes;
 
-		trace_btrfs_reserve_extent(block_group, ffe_ctl.search_start,
-					   num_bytes);
-		btrfs_release_block_group(block_group, delalloc);
+		trace_btrfs_reserve_extent(block_group, ffe_ctl->search_start,
+					   ffe_ctl->num_bytes);
+		btrfs_release_block_group(block_group, ffe_ctl->delalloc);
 		break;
 loop:
-		ffe_ctl.retry_clustered = false;
-		ffe_ctl.retry_unclustered = false;
-		BUG_ON(btrfs_bg_flags_to_raid_index(block_group->flags) !=
-		       ffe_ctl.index);
-		btrfs_release_block_group(block_group, delalloc);
+		release_block_group(block_group, ffe_ctl, ffe_ctl->delalloc);
 		cond_resched();
 	}
 	up_read(&space_info->groups_sem);
 
-	ret = find_free_extent_update_loop(fs_info, last_ptr, ins, &ffe_ctl,
-					   full_search, use_cluster);
+	ret = find_free_extent_update_loop(fs_info, ins, ffe_ctl, full_search);
 	if (ret > 0)
 		goto search;
 
@@ -4069,12 +4477,12 @@ loop:
 		 * Use ffe_ctl->total_free_space as fallback if we can't find
 		 * any contiguous hole.
 		 */
-		if (!ffe_ctl.max_extent_size)
-			ffe_ctl.max_extent_size = ffe_ctl.total_free_space;
+		if (!ffe_ctl->max_extent_size)
+			ffe_ctl->max_extent_size = ffe_ctl->total_free_space;
 		spin_lock(&space_info->lock);
-		space_info->max_extent_size = ffe_ctl.max_extent_size;
+		space_info->max_extent_size = ffe_ctl->max_extent_size;
 		spin_unlock(&space_info->lock);
-		ins->offset = ffe_ctl.max_extent_size;
+		ins->offset = ffe_ctl->max_extent_size;
 	} else if (ret == -ENOSPC) {
 		ret = cache_block_group_error;
 	}
@@ -4132,15 +4540,28 @@ int btrfs_reserve_extent(struct btrfs_root *root, u64 ram_bytes,
 			 struct btrfs_key *ins, int is_data, int delalloc)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
+	struct find_free_extent_ctl ffe_ctl = {};
 	bool final_tried = num_bytes == min_alloc_size;
 	u64 flags;
 	int ret;
+	bool for_treelog = (root->root_key.objectid == BTRFS_TREE_LOG_OBJECTID);
+	bool for_data_reloc = (btrfs_is_data_reloc_root(root) && is_data);
 
 	flags = get_alloc_profile_by_root(root, is_data);
 again:
 	WARN_ON(num_bytes < fs_info->sectorsize);
-	ret = find_free_extent(fs_info, ram_bytes, num_bytes, empty_size,
-			       hint_byte, ins, flags, delalloc);
+
+	ffe_ctl.ram_bytes = ram_bytes;
+	ffe_ctl.num_bytes = num_bytes;
+	ffe_ctl.min_alloc_size = min_alloc_size;
+	ffe_ctl.empty_size = empty_size;
+	ffe_ctl.flags = flags;
+	ffe_ctl.delalloc = delalloc;
+	ffe_ctl.hint_byte = hint_byte;
+	ffe_ctl.for_treelog = for_treelog;
+	ffe_ctl.for_data_reloc = for_data_reloc;
+
+	ret = find_free_extent(root, ins, &ffe_ctl);
 	if (!ret && !is_data) {
 		btrfs_dec_block_group_reservations(fs_info, ins->objectid);
 	} else if (ret == -ENOSPC) {
@@ -4158,8 +4579,8 @@ again:
 
 			sinfo = btrfs_find_space_info(fs_info, flags);
 			btrfs_err(fs_info,
-				  "allocation failed flags %llu, wanted %llu",
-				  flags, num_bytes);
+	"allocation failed flags %llu, wanted %llu tree-log %d, relocation: %d",
+				  flags, num_bytes, for_treelog, for_data_reloc);
 			if (sinfo)
 				btrfs_dump_space_info(fs_info, sinfo,
 						      num_bytes, 1);
@@ -4189,28 +4610,53 @@ int btrfs_free_reserved_extent(struct btrfs_fs_info *fs_info,
 	return 0;
 }
 
-int btrfs_pin_reserved_extent(struct btrfs_fs_info *fs_info, u64 start, u64 len)
+int btrfs_pin_reserved_extent(struct btrfs_trans_handle *trans, u64 start,
+			      u64 len)
 {
 	struct btrfs_block_group *cache;
 	int ret = 0;
 
-	cache = btrfs_lookup_block_group(fs_info, start);
+	cache = btrfs_lookup_block_group(trans->fs_info, start);
 	if (!cache) {
-		btrfs_err(fs_info, "unable to find block group for %llu", start);
+		btrfs_err(trans->fs_info, "unable to find block group for %llu",
+			  start);
 		return -ENOSPC;
 	}
 
-	ret = pin_down_extent(cache, start, len, 1);
+	ret = pin_down_extent(trans, cache, start, len, 1);
 	btrfs_put_block_group(cache);
 	return ret;
 }
 
+static int alloc_reserved_extent(struct btrfs_trans_handle *trans, u64 bytenr,
+				 u64 num_bytes)
+{
+	struct btrfs_fs_info *fs_info = trans->fs_info;
+	int ret;
+
+	ret = remove_from_free_space_tree(trans, bytenr, num_bytes);
+	if (ret)
+		return ret;
+
+	ret = btrfs_update_block_group(trans, bytenr, num_bytes, true);
+	if (ret) {
+		ASSERT(!ret);
+		btrfs_err(fs_info, "update block group failed for %llu %llu",
+			  bytenr, num_bytes);
+		return ret;
+	}
+
+	trace_btrfs_reserved_extent_alloc(fs_info, bytenr, num_bytes);
+	return 0;
+}
+
 static int alloc_reserved_file_extent(struct btrfs_trans_handle *trans,
 				      u64 parent, u64 root_objectid,
 				      u64 flags, u64 owner, u64 offset,
 				      struct btrfs_key *ins, int ref_mod)
 {
 	struct btrfs_fs_info *fs_info = trans->fs_info;
+	struct btrfs_root *extent_root;
 	int ret;
 	struct btrfs_extent_item *extent_item;
 	struct btrfs_extent_inline_ref *iref;
@@ -4230,9 +4676,8 @@ static int alloc_reserved_file_extent(struct btrfs_trans_handle *trans,
 	if (!path)
 		return -ENOMEM;
 
-	path->leave_spinning = 1;
-	ret = btrfs_insert_empty_item(trans, fs_info->extent_root, path,
-				      ins, size);
+	extent_root = btrfs_extent_root(fs_info, ins->objectid);
+	ret = btrfs_insert_empty_item(trans, extent_root, path, ins, size);
 	if (ret) {
 		btrfs_free_path(path);
 		return ret;
@@ -4265,18 +4710,7 @@ static int alloc_reserved_file_extent(struct btrfs_trans_handle *trans,
 	btrfs_mark_buffer_dirty(path->nodes[0]);
 	btrfs_free_path(path);
 
-	ret = remove_from_free_space_tree(trans, ins->objectid, ins->offset);
-	if (ret)
-		return ret;
-
-	ret = btrfs_update_block_group(trans, ins->objectid, ins->offset, 1);
-	if (ret) { /* -ENOENT, logic error */
-		btrfs_err(fs_info, "update block group failed for %llu %llu",
-			ins->objectid, ins->offset);
-		BUG();
-	}
-	trace_btrfs_reserved_extent_alloc(fs_info, ins->objectid, ins->offset);
-	return ret;
+	return alloc_reserved_extent(trans, ins->objectid, ins->offset);
 }
 
 static int alloc_reserved_tree_block(struct btrfs_trans_handle *trans,
@@ -4284,6 +4718,7 @@ static int alloc_reserved_tree_block(struct btrfs_trans_handle *trans,
 				     struct btrfs_delayed_extent_op *extent_op)
 {
 	struct btrfs_fs_info *fs_info = trans->fs_info;
+	struct btrfs_root *extent_root;
 	int ret;
 	struct btrfs_extent_item *extent_item;
 	struct btrfs_key extent_key;
@@ -4293,7 +4728,6 @@ static int alloc_reserved_tree_block(struct btrfs_trans_handle *trans,
 	struct extent_buffer *leaf;
 	struct btrfs_delayed_tree_ref *ref;
 	u32 size = sizeof(*extent_item) + sizeof(*iref);
-	u64 num_bytes;
 	u64 flags = extent_op->flags_to_set;
 	bool skinny_metadata = btrfs_fs_incompat(fs_info, SKINNY_METADATA);
 
@@ -4303,21 +4737,19 @@ static int alloc_reserved_tree_block(struct btrfs_trans_handle *trans,
 	if (skinny_metadata) {
 		extent_key.offset = ref->level;
 		extent_key.type = BTRFS_METADATA_ITEM_KEY;
-		num_bytes = fs_info->nodesize;
 	} else {
 		extent_key.offset = node->num_bytes;
 		extent_key.type = BTRFS_EXTENT_ITEM_KEY;
 		size += sizeof(*block_info);
-		num_bytes = node->num_bytes;
 	}
 
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
 
-	path->leave_spinning = 1;
-	ret = btrfs_insert_empty_item(trans, fs_info->extent_root, path,
-				      &extent_key, size);
+	extent_root = btrfs_extent_root(fs_info, extent_key.objectid);
+	ret = btrfs_insert_empty_item(trans, extent_root, path, &extent_key,
+				      size);
 	if (ret) {
 		btrfs_free_path(path);
 		return ret;
@@ -4341,7 +4773,6 @@ static int alloc_reserved_tree_block(struct btrfs_trans_handle *trans,
 	}
 
 	if (node->type == BTRFS_SHARED_BLOCK_REF_KEY) {
-		BUG_ON(!(flags & BTRFS_BLOCK_FLAG_FULL_BACKREF));
 		btrfs_set_extent_inline_ref_type(leaf, iref,
 						 BTRFS_SHARED_BLOCK_REF_KEY);
 		btrfs_set_extent_inline_ref_offset(leaf, iref, ref->parent);
@@ -4354,22 +4785,7 @@ static int alloc_reserved_tree_block(struct btrfs_trans_handle *trans,
 	btrfs_mark_buffer_dirty(leaf);
 	btrfs_free_path(path);
 
-	ret = remove_from_free_space_tree(trans, extent_key.objectid,
-					  num_bytes);
-	if (ret)
-		return ret;
-
-	ret = btrfs_update_block_group(trans, extent_key.objectid,
-				       fs_info->nodesize, 1);
-	if (ret) { /* -ENOENT, logic error */
-		btrfs_err(fs_info, "update block group failed for %llu %llu",
-			extent_key.objectid, extent_key.offset);
-		BUG();
-	}
-
-	trace_btrfs_reserved_extent_alloc(fs_info, extent_key.objectid,
-					  fs_info->nodesize);
-	return ret;
+	return alloc_reserved_extent(trans, node->bytenr, fs_info->nodesize);
 }
 
 int btrfs_alloc_reserved_file_extent(struct btrfs_trans_handle *trans,
@@ -4378,17 +4794,16 @@ int btrfs_alloc_reserved_file_extent(struct btrfs_trans_handle *trans,
 				     struct btrfs_key *ins)
 {
 	struct btrfs_ref generic_ref = { 0 };
-	int ret;
 
 	BUG_ON(root->root_key.objectid == BTRFS_TREE_LOG_OBJECTID);
 
 	btrfs_init_generic_ref(&generic_ref, BTRFS_ADD_DELAYED_EXTENT,
 			       ins->objectid, ins->offset, 0);
-	btrfs_init_data_ref(&generic_ref, root->root_key.objectid, owner, offset);
+	btrfs_init_data_ref(&generic_ref, root->root_key.objectid, owner,
+			    offset, 0, false);
 	btrfs_ref_tree_mod(root->fs_info, &generic_ref);
-	ret = btrfs_add_delayed_data_ref(trans, &generic_ref,
-					 ram_bytes, NULL, NULL);
-	return ret;
+
+	return btrfs_add_delayed_data_ref(trans, &generic_ref, ram_bytes);
 }
 
 /*
@@ -4431,19 +4846,21 @@ int btrfs_alloc_logged_file_extent(struct btrfs_trans_handle *trans,
 	ret = alloc_reserved_file_extent(trans, 0, root_objectid, 0, owner,
 					 offset, ins, 1);
 	if (ret)
-		btrfs_pin_extent(fs_info, ins->objectid, ins->offset, 1);
+		btrfs_pin_extent(trans, ins->objectid, ins->offset, 1);
 	btrfs_put_block_group(block_group);
 	return ret;
 }
 
 static struct extent_buffer *
 btrfs_init_new_buffer(struct btrfs_trans_handle *trans, struct btrfs_root *root,
-		      u64 bytenr, int level, u64 owner)
+		      u64 bytenr, int level, u64 owner,
+		      enum btrfs_lock_nesting nest)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct extent_buffer *buf;
+	u64 lockdep_owner = owner;
 
-	buf = btrfs_find_create_tree_block(fs_info, bytenr);
+	buf = btrfs_find_create_tree_block(fs_info, bytenr, owner, level);
 	if (IS_ERR(buf))
 		return buf;
 
@@ -4460,12 +4877,35 @@ btrfs_init_new_buffer(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 		return ERR_PTR(-EUCLEAN);
 	}
 
-	btrfs_set_buffer_lockdep_class(root->root_key.objectid, buf, level);
-	btrfs_tree_lock(buf);
+	/*
+	 * The reloc trees are just snapshots, so we need them to appear to be
+	 * just like any other fs tree WRT lockdep.
+	 *
+	 * The exception however is in replace_path() in relocation, where we
+	 * hold the lock on the original fs root and then search for the reloc
+	 * root.  At that point we need to make sure any reloc root buffers are
+	 * set to the BTRFS_TREE_RELOC_OBJECTID lockdep class in order to make
+	 * lockdep happy.
+	 */
+	if (lockdep_owner == BTRFS_TREE_RELOC_OBJECTID &&
+	    !test_bit(BTRFS_ROOT_RESET_LOCKDEP_CLASS, &root->state))
+		lockdep_owner = BTRFS_FS_TREE_OBJECTID;
+
+	/* btrfs_clean_tree_block() accesses generation field. */
+	btrfs_set_header_generation(buf, trans->transid);
+
+	/*
+	 * This needs to stay, because we could allocate a freed block from an
+	 * old tree into a new tree, so we need to make sure this new block is
+	 * set to the appropriate level and owner.
+	 */
+	btrfs_set_buffer_lockdep_class(lockdep_owner, buf, level);
+
+	__btrfs_tree_lock(buf, nest);
 	btrfs_clean_tree_block(buf);
 	clear_bit(EXTENT_BUFFER_STALE, &buf->bflags);
+	clear_bit(EXTENT_BUFFER_NO_CHECK, &buf->bflags);
 
-	btrfs_set_lock_blocking_write(buf);
 	set_extent_buffer_uptodate(buf);
 
 	memzero_extent_buffer(buf, 0, sizeof(struct btrfs_header));
@@ -4493,7 +4933,6 @@ btrfs_init_new_buffer(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 		set_extent_dirty(&trans->transaction->dirty_pages, buf->start,
 			 buf->start + buf->len - 1, GFP_NOFS);
 	}
-	trans->dirty = true;
 	/* this returns a buffer locked for blocking */
 	return buf;
 }
@@ -4507,7 +4946,8 @@ struct extent_buffer *btrfs_alloc_tree_block(struct btrfs_trans_handle *trans,
 					     u64 parent, u64 root_objectid,
 					     const struct btrfs_disk_key *key,
 					     int level, u64 hint,
-					     u64 empty_size)
+					     u64 empty_size,
+					     enum btrfs_lock_nesting nest)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct btrfs_key ins;
@@ -4523,7 +4963,7 @@ struct extent_buffer *btrfs_alloc_tree_block(struct btrfs_trans_handle *trans,
 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
 	if (btrfs_is_testing(fs_info)) {
 		buf = btrfs_init_new_buffer(trans, root, root->alloc_bytenr,
-					    level, root_objectid);
+					    level, root_objectid, nest);
 		if (!IS_ERR(buf))
 			root->alloc_bytenr += blocksize;
 		return buf;
@@ -4540,7 +4980,7 @@ struct extent_buffer *btrfs_alloc_tree_block(struct btrfs_trans_handle *trans,
 		goto out_unuse;
 
 	buf = btrfs_init_new_buffer(trans, root, ins.objectid, level,
-				    root_objectid);
+				    root_objectid, nest);
 	if (IS_ERR(buf)) {
 		ret = PTR_ERR(buf);
 		goto out_free_reserved;
@@ -4566,16 +5006,14 @@ struct extent_buffer *btrfs_alloc_tree_block(struct btrfs_trans_handle *trans,
 		extent_op->flags_to_set = flags;
 		extent_op->update_key = skinny_metadata ? false : true;
 		extent_op->update_flags = true;
-		extent_op->is_data = false;
 		extent_op->level = level;
 
 		btrfs_init_generic_ref(&generic_ref, BTRFS_ADD_DELAYED_EXTENT,
 				       ins.objectid, ins.offset, parent);
-		generic_ref.real_root = root->root_key.objectid;
-		btrfs_init_tree_ref(&generic_ref, level, root_objectid);
+		btrfs_init_tree_ref(&generic_ref, level, root_objectid,
+				    root->root_key.objectid, false);
 		btrfs_ref_tree_mod(fs_info, &generic_ref);
-		ret = btrfs_add_delayed_tree_ref(trans, &generic_ref,
-						 extent_op, NULL, NULL);
+		ret = btrfs_add_delayed_tree_ref(trans, &generic_ref, extent_op);
 		if (ret)
 			goto out_free_delayed;
 	}
@@ -4584,6 +5022,7 @@ struct extent_buffer *btrfs_alloc_tree_block(struct btrfs_trans_handle *trans,
 out_free_delayed:
 	btrfs_free_delayed_extent_op(extent_op);
 out_free_buf:
+	btrfs_tree_unlock(buf);
 	free_extent_buffer(buf);
 out_free_reserved:
 	btrfs_free_reserved_extent(fs_info, ins.objectid, ins.offset, 0);
@@ -4685,7 +5124,7 @@ static noinline void reada_walk_down(struct btrfs_trans_handle *trans,
 				continue;
 		}
 reada:
-		readahead_tree_block(fs_info, bytenr);
+		btrfs_readahead_node_child(eb, slot);
 		nread++;
 	}
 	wc->reada_slot = slot;
@@ -4750,9 +5189,8 @@ static noinline int walk_down_proc(struct btrfs_trans_handle *trans,
 		BUG_ON(ret); /* -ENOMEM */
 		ret = btrfs_dec_ref(trans, root, eb, 0);
 		BUG_ON(ret); /* -ENOMEM */
-		ret = btrfs_set_disk_extent_flags(trans, eb->start,
-						  eb->len, flag,
-						  btrfs_header_level(eb), 0);
+		ret = btrfs_set_disk_extent_flags(trans, eb, flag,
+						  btrfs_header_level(eb));
 		BUG_ON(ret); /* -ENOMEM */
 		wc->flags[level] |= flag;
 	}
@@ -4845,16 +5283,13 @@ static noinline int do_walk_down(struct btrfs_trans_handle *trans,
 
 	next = find_extent_buffer(fs_info, bytenr);
 	if (!next) {
-		next = btrfs_find_create_tree_block(fs_info, bytenr);
+		next = btrfs_find_create_tree_block(fs_info, bytenr,
+				root->root_key.objectid, level - 1);
 		if (IS_ERR(next))
 			return PTR_ERR(next);
-
-		btrfs_set_buffer_lockdep_class(root->root_key.objectid, next,
-					       level - 1);
 		reada = 1;
 	}
 	btrfs_tree_lock(next);
-	btrfs_set_lock_blocking_write(next);
 
 	ret = btrfs_lookup_extent_info(trans, fs_info, bytenr, level - 1, 1,
 				       &wc->refs[level - 1],
@@ -4905,8 +5340,8 @@ static noinline int do_walk_down(struct btrfs_trans_handle *trans,
 	if (!next) {
 		if (reada && level == 1)
 			reada_walk_down(trans, root, wc, path);
-		next = read_tree_block(fs_info, bytenr, generation, level - 1,
-				       &first_key);
+		next = read_tree_block(fs_info, bytenr, root->root_key.objectid,
+				       generation, level - 1, &first_key);
 		if (IS_ERR(next)) {
 			return PTR_ERR(next);
 		} else if (!extent_buffer_uptodate(next)) {
@@ -4914,7 +5349,6 @@ static noinline int do_walk_down(struct btrfs_trans_handle *trans,
 			return -EIO;
 		}
 		btrfs_tree_lock(next);
-		btrfs_set_lock_blocking_write(next);
 	}
 
 	level--;
@@ -4926,7 +5360,7 @@ static noinline int do_walk_down(struct btrfs_trans_handle *trans,
 	}
 	path->nodes[level] = next;
 	path->slots[level] = 0;
-	path->locks[level] = BTRFS_WRITE_LOCK_BLOCKING;
+	path->locks[level] = BTRFS_WRITE_LOCK;
 	wc->level = level;
 	if (wc->level == 1)
 		wc->reada_slot = 0;
@@ -4994,7 +5428,8 @@ skip:
 
 		btrfs_init_generic_ref(&ref, BTRFS_DROP_DELAYED_REF, bytenr,
 				       fs_info->nodesize, parent);
-		btrfs_init_tree_ref(&ref, level - 1, root->root_key.objectid);
+		btrfs_init_tree_ref(&ref, level - 1, root->root_key.objectid,
+				    0, false);
 		ret = btrfs_free_extent(trans, &ref);
 		if (ret)
 			goto out_unlock;
@@ -5054,8 +5489,7 @@ static noinline int walk_up_proc(struct btrfs_trans_handle *trans,
 		if (!path->locks[level]) {
 			BUG_ON(level == 0);
 			btrfs_tree_lock(eb);
-			btrfs_set_lock_blocking_write(eb);
-			path->locks[level] = BTRFS_WRITE_LOCK_BLOCKING;
+			path->locks[level] = BTRFS_WRITE_LOCK;
 
 			ret = btrfs_lookup_extent_info(trans, fs_info,
 						       eb->start, level, 1,
@@ -5098,8 +5532,7 @@ static noinline int walk_up_proc(struct btrfs_trans_handle *trans,
 		if (!path->locks[level] &&
 		    btrfs_header_generation(eb) == trans->transid) {
 			btrfs_tree_lock(eb);
-			btrfs_set_lock_blocking_write(eb);
-			path->locks[level] = BTRFS_WRITE_LOCK_BLOCKING;
+			path->locks[level] = BTRFS_WRITE_LOCK;
 		}
 		btrfs_clean_tree_block(eb);
 	}
@@ -5117,7 +5550,8 @@ static noinline int walk_up_proc(struct btrfs_trans_handle *trans,
 			goto owner_mismatch;
 	}
 
-	btrfs_free_tree_block(trans, root, eb, parent, wc->refs[level] == 1);
+	btrfs_free_tree_block(trans, btrfs_root_id(root), eb, parent,
+			      wc->refs[level] == 1);
 out:
 	wc->refs[level] = 0;
 	wc->flags[level] = 0;
@@ -5209,10 +5643,10 @@ static noinline int walk_up_tree(struct btrfs_trans_handle *trans,
  *
  * If called with for_reloc == 0, may exit early with -EAGAIN
  */
-int btrfs_drop_snapshot(struct btrfs_root *root,
-			 struct btrfs_block_rsv *block_rsv, int update_ref,
-			 int for_reloc)
+int btrfs_drop_snapshot(struct btrfs_root *root, int update_ref, int for_reloc)
 {
+	const bool is_reloc_root = (root->root_key.objectid ==
+				    BTRFS_TREE_RELOC_OBJECTID);
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct btrfs_path *path;
 	struct btrfs_trans_handle *trans;
@@ -5224,6 +5658,7 @@ int btrfs_drop_snapshot(struct btrfs_root *root,
 	int ret;
 	int level;
 	bool root_dropped = false;
+	bool unfinished_drop = false;
 
 	btrfs_debug(fs_info, "Drop subvolume %llu", root->root_key.objectid);
 
@@ -5240,7 +5675,14 @@ int btrfs_drop_snapshot(struct btrfs_root *root,
 		goto out;
 	}
 
-	trans = btrfs_start_transaction(tree_root, 0);
+	/*
+	 * Use join to avoid potential EINTR from transaction start. See
+	 * wait_reserve_ticket and the whole reservation callchain.
+	 */
+	if (for_reloc)
+		trans = btrfs_join_transaction(tree_root);
+	else
+		trans = btrfs_start_transaction(tree_root, 0);
 	if (IS_ERR(trans)) {
 		err = PTR_ERR(trans);
 		goto out_free;
@@ -5250,9 +5692,6 @@ int btrfs_drop_snapshot(struct btrfs_root *root,
 	if (err)
 		goto out_end_trans;
 
-	if (block_rsv)
-		trans->block_rsv = block_rsv;
-
 	/*
 	 * This will help us catch people modifying the fs tree while we're
 	 * dropping it.  It is unsafe to mess with the fs tree while it's being
@@ -5262,12 +5701,13 @@ int btrfs_drop_snapshot(struct btrfs_root *root,
 	 * already dropped.
 	 */
 	set_bit(BTRFS_ROOT_DELETING, &root->state);
+	unfinished_drop = test_bit(BTRFS_ROOT_UNFINISHED_DROP, &root->state);
+
 	if (btrfs_disk_key_objectid(&root_item->drop_progress) == 0) {
 		level = btrfs_header_level(root->node);
 		path->nodes[level] = btrfs_lock_root_node(root);
-		btrfs_set_lock_blocking_write(path->nodes[level]);
 		path->slots[level] = 0;
-		path->locks[level] = BTRFS_WRITE_LOCK_BLOCKING;
+		path->locks[level] = BTRFS_WRITE_LOCK;
 		memset(&wc->update_progress, 0,
 		       sizeof(wc->update_progress));
 	} else {
@@ -5275,7 +5715,7 @@ int btrfs_drop_snapshot(struct btrfs_root *root,
 		memcpy(&wc->update_progress, &key,
 		       sizeof(wc->update_progress));
 
-		level = root_item->drop_level;
+		level = btrfs_root_drop_level(root_item);
 		BUG_ON(level == 0);
 		path->lowest_level = level;
 		ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
@@ -5295,8 +5735,7 @@ int btrfs_drop_snapshot(struct btrfs_root *root,
 		level = btrfs_header_level(root->node);
 		while (1) {
 			btrfs_tree_lock(path->nodes[level]);
-			btrfs_set_lock_blocking_write(path->nodes[level]);
-			path->locks[level] = BTRFS_WRITE_LOCK_BLOCKING;
+			path->locks[level] = BTRFS_WRITE_LOCK;
 
 			ret = btrfs_lookup_extent_info(trans, fs_info,
 						path->nodes[level]->start,
@@ -5308,7 +5747,7 @@ int btrfs_drop_snapshot(struct btrfs_root *root,
 			}
 			BUG_ON(wc->refs[level] == 0);
 
-			if (level == root_item->drop_level)
+			if (level == btrfs_root_drop_level(root_item))
 				break;
 
 			btrfs_tree_unlock(path->nodes[level]);
@@ -5353,7 +5792,7 @@ int btrfs_drop_snapshot(struct btrfs_root *root,
 		}
 		btrfs_cpu_key_to_disk(&root_item->drop_progress,
 				      &wc->drop_progress);
-		root_item->drop_level = wc->drop_level;
+		btrfs_set_root_drop_level(root_item, wc->drop_level);
 
 		BUG_ON(wc->level == 0);
 		if (btrfs_should_end_transaction(trans) ||
@@ -5367,6 +5806,9 @@ int btrfs_drop_snapshot(struct btrfs_root *root,
 				goto out_end_trans;
 			}
 
+			if (!is_reloc_root)
+				btrfs_set_last_root_drop_gen(fs_info, trans->transid);
+
 			btrfs_end_transaction_throttle(trans);
 			if (!for_reloc && btrfs_need_cleaner_sleep(fs_info)) {
 				btrfs_debug(fs_info,
@@ -5375,13 +5817,19 @@ int btrfs_drop_snapshot(struct btrfs_root *root,
 				goto out_free;
 			}
 
-			trans = btrfs_start_transaction(tree_root, 0);
+		       /*
+			* Use join to avoid potential EINTR from transaction
+			* start. See wait_reserve_ticket and the whole
+			* reservation callchain.
+			*/
+			if (for_reloc)
+				trans = btrfs_join_transaction(tree_root);
+			else
+				trans = btrfs_start_transaction(tree_root, 0);
 			if (IS_ERR(trans)) {
 				err = PTR_ERR(trans);
 				goto out_free;
 			}
-			if (block_rsv)
-				trans->block_rsv = block_rsv;
 		}
 	}
 	btrfs_release_path(path);
@@ -5395,7 +5843,7 @@ int btrfs_drop_snapshot(struct btrfs_root *root,
 		goto out_end_trans;
 	}
 
-	if (root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID) {
+	if (!is_reloc_root) {
 		ret = btrfs_find_root(tree_root, &root->root_key, path,
 				      NULL, NULL);
 		if (ret < 0) {
@@ -5413,21 +5861,36 @@ int btrfs_drop_snapshot(struct btrfs_root *root,
 		}
 	}
 
-	if (test_bit(BTRFS_ROOT_IN_RADIX, &root->state)) {
+	/*
+	 * This subvolume is going to be completely dropped, and won't be
+	 * recorded as dirty roots, thus pertrans meta rsv will not be freed at
+	 * commit transaction time.  So free it here manually.
+	 */
+	btrfs_qgroup_convert_reserved_meta(root, INT_MAX);
+	btrfs_qgroup_free_meta_all_pertrans(root);
+
+	if (test_bit(BTRFS_ROOT_IN_RADIX, &root->state))
 		btrfs_add_dropped_root(trans, root);
-	} else {
-		free_extent_buffer(root->node);
-		free_extent_buffer(root->commit_root);
-		btrfs_put_fs_root(root);
-	}
+	else
+		btrfs_put_root(root);
 	root_dropped = true;
 out_end_trans:
+	if (!is_reloc_root)
+		btrfs_set_last_root_drop_gen(fs_info, trans->transid);
+
 	btrfs_end_transaction_throttle(trans);
 out_free:
 	kfree(wc);
 	btrfs_free_path(path);
 out:
 	/*
+	 * We were an unfinished drop root, check to see if there are any
+	 * pending, and if not clear and wake up any waiters.
+	 */
+	if (!err && unfinished_drop)
+		btrfs_maybe_wake_unfinished_drop(fs_info);
+
+	/*
 	 * So if we need to stop dropping the snapshot for whatever reason we
 	 * need to make sure to add it back to the dead root list so that we
 	 * keep trying to do the work later.  This also cleans up roots if we
@@ -5436,8 +5899,6 @@ out:
 	 */
 	if (!for_reloc && !root_dropped)
 		btrfs_add_dead_root(root);
-	if (err && err != -EAGAIN)
-		btrfs_handle_fs_error(fs_info, err, NULL);
 	return err;
 }
 
@@ -5472,17 +5933,17 @@ int btrfs_drop_subtree(struct btrfs_trans_handle *trans,
 		return -ENOMEM;
 	}
 
-	btrfs_assert_tree_locked(parent);
+	btrfs_assert_tree_write_locked(parent);
 	parent_level = btrfs_header_level(parent);
 	atomic_inc(&parent->refs);
 	path->nodes[parent_level] = parent;
 	path->slots[parent_level] = btrfs_header_nritems(parent);
 
-	btrfs_assert_tree_locked(node);
+	btrfs_assert_tree_write_locked(node);
 	level = btrfs_header_level(node);
 	path->nodes[level] = node;
 	path->slots[level] = 0;
-	path->locks[level] = BTRFS_WRITE_LOCK_BLOCKING;
+	path->locks[level] = BTRFS_WRITE_LOCK;
 
 	wc->refs[parent_level] = 1;
 	wc->flags[parent_level] = BTRFS_BLOCK_FLAG_FULL_BACKREF;
@@ -5574,13 +6035,13 @@ int btrfs_error_unpin_extent_range(struct btrfs_fs_info *fs_info,
  */
 static int btrfs_trim_free_extents(struct btrfs_device *device, u64 *trimmed)
 {
-	u64 start = SZ_1M, len = 0, end = 0;
+	u64 start = BTRFS_DEVICE_RANGE_RESERVED, len = 0, end = 0;
 	int ret;
 
 	*trimmed = 0;
 
 	/* Discard not supported = nothing to do. */
-	if (!blk_queue_discard(bdev_get_queue(device->bdev)))
+	if (!bdev_max_discard_sectors(device->bdev))
 		return 0;
 
 	/* Not writable = nothing to do. */
@@ -5605,8 +6066,21 @@ static int btrfs_trim_free_extents(struct btrfs_device *device, u64 *trimmed)
 					    &start, &end,
 					    CHUNK_TRIMMED | CHUNK_ALLOCATED);
 
-		/* Ensure we skip the reserved area in the first 1M */
-		start = max_t(u64, start, SZ_1M);
+		/* Check if there are any CHUNK_* bits left */
+		if (start > device->total_bytes) {
+			WARN_ON(IS_ENABLED(CONFIG_BTRFS_DEBUG));
+			btrfs_warn_in_rcu(fs_info,
+"ignoring attempt to trim beyond device size: offset %llu length %llu device %s device size %llu",
+					  start, end - start + 1,
+					  rcu_str_deref(device->name),
+					  device->total_bytes);
+			mutex_unlock(&fs_info->chunk_mutex);
+			ret = 0;
+			break;
+		}
+
+		/* Ensure we skip the reserved space on each device. */
+		start = max_t(u64, start, BTRFS_DEVICE_RANGE_RESERVED);
 
 		/*
 		 * If find_first_clear_extent_bit find a range that spans the
@@ -5660,9 +6134,9 @@ static int btrfs_trim_free_extents(struct btrfs_device *device, u64 *trimmed)
  */
 int btrfs_trim_fs(struct btrfs_fs_info *fs_info, struct fstrim_range *range)
 {
+	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
 	struct btrfs_block_group *cache = NULL;
 	struct btrfs_device *device;
-	struct list_head *devices;
 	u64 group_trimmed;
 	u64 range_end = U64_MAX;
 	u64 start;
@@ -5674,6 +6148,9 @@ int btrfs_trim_fs(struct btrfs_fs_info *fs_info, struct fstrim_range *range)
 	int dev_ret = 0;
 	int ret = 0;
 
+	if (range->start == U64_MAX)
+		return -EINVAL;
+
 	/*
 	 * Check range overflow if range->len is set.
 	 * The default range->len is U64_MAX.
@@ -5694,13 +6171,7 @@ int btrfs_trim_fs(struct btrfs_fs_info *fs_info, struct fstrim_range *range)
 
 		if (end - start >= range->minlen) {
 			if (!btrfs_block_group_done(cache)) {
-				ret = btrfs_cache_block_group(cache, 0);
-				if (ret) {
-					bg_failed++;
-					bg_ret = ret;
-					continue;
-				}
-				ret = btrfs_wait_block_group_cache_done(cache);
+				ret = btrfs_cache_block_group(cache, true);
 				if (ret) {
 					bg_failed++;
 					bg_ret = ret;
@@ -5726,9 +6197,12 @@ int btrfs_trim_fs(struct btrfs_fs_info *fs_info, struct fstrim_range *range)
 		btrfs_warn(fs_info,
 			"failed to trim %llu block group(s), last error %d",
 			bg_failed, bg_ret);
-	mutex_lock(&fs_info->fs_devices->device_list_mutex);
-	devices = &fs_info->fs_devices->devices;
-	list_for_each_entry(device, devices, dev_list) {
+
+	mutex_lock(&fs_devices->device_list_mutex);
+	list_for_each_entry(device, &fs_devices->devices, dev_list) {
+		if (test_bit(BTRFS_DEV_STATE_MISSING, &device->dev_state))
+			continue;
+
 		ret = btrfs_trim_free_extents(device, &group_trimmed);
 		if (ret) {
 			dev_failed++;
@@ -5738,7 +6212,7 @@ int btrfs_trim_fs(struct btrfs_fs_info *fs_info, struct fstrim_range *range)
 
 		trimmed += group_trimmed;
 	}
-	mutex_unlock(&fs_info->fs_devices->device_list_mutex);
+	mutex_unlock(&fs_devices->device_list_mutex);
 
 	if (dev_failed)
 		btrfs_warn(fs_info,
@@ -5749,47 +6223,3 @@ int btrfs_trim_fs(struct btrfs_fs_info *fs_info, struct fstrim_range *range)
 		return bg_ret;
 	return dev_ret;
 }
-
-/*
- * btrfs_{start,end}_write_no_snapshotting() are similar to
- * mnt_{want,drop}_write(), they are used to prevent some tasks from writing
- * data into the page cache through nocow before the subvolume is snapshoted,
- * but flush the data into disk after the snapshot creation, or to prevent
- * operations while snapshotting is ongoing and that cause the snapshot to be
- * inconsistent (writes followed by expanding truncates for example).
- */
-void btrfs_end_write_no_snapshotting(struct btrfs_root *root)
-{
-	percpu_counter_dec(&root->subv_writers->counter);
-	cond_wake_up(&root->subv_writers->wait);
-}
-
-int btrfs_start_write_no_snapshotting(struct btrfs_root *root)
-{
-	if (atomic_read(&root->will_be_snapshotted))
-		return 0;
-
-	percpu_counter_inc(&root->subv_writers->counter);
-	/*
-	 * Make sure counter is updated before we check for snapshot creation.
-	 */
-	smp_mb();
-	if (atomic_read(&root->will_be_snapshotted)) {
-		btrfs_end_write_no_snapshotting(root);
-		return 0;
-	}
-	return 1;
-}
-
-void btrfs_wait_for_snapshot_creation(struct btrfs_root *root)
-{
-	while (true) {
-		int ret;
-
-		ret = btrfs_start_write_no_snapshotting(root);
-		if (ret)
-			break;
-		wait_var_event(&root->will_be_snapshotted,
-			       !atomic_read(&root->will_be_snapshotted));
-	}
-}
diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c
index c0f202741e09..4dcf22e051ff 100644
--- a/fs/btrfs/extent_io.c
+++ b/fs/btrfs/extent_io.c
@@ -6,13 +6,15 @@
 #include <linux/mm.h>
 #include <linux/pagemap.h>
 #include <linux/page-flags.h>
+#include <linux/sched/mm.h>
 #include <linux/spinlock.h>
 #include <linux/blkdev.h>
 #include <linux/swap.h>
 #include <linux/writeback.h>
 #include <linux/pagevec.h>
 #include <linux/prefetch.h>
-#include <linux/cleancache.h>
+#include <linux/fsverity.h>
+#include "misc.h"
 #include "extent_io.h"
 #include "extent-io-tree.h"
 #include "extent_map.h"
@@ -24,105 +26,80 @@
 #include "rcu-string.h"
 #include "backref.h"
 #include "disk-io.h"
+#include "subpage.h"
+#include "zoned.h"
+#include "block-group.h"
+#include "compression.h"
 
-static struct kmem_cache *extent_state_cache;
 static struct kmem_cache *extent_buffer_cache;
-static struct bio_set btrfs_bioset;
-
-static inline bool extent_state_in_tree(const struct extent_state *state)
-{
-	return !RB_EMPTY_NODE(&state->rb_node);
-}
 
 #ifdef CONFIG_BTRFS_DEBUG
-static LIST_HEAD(buffers);
-static LIST_HEAD(states);
-
-static DEFINE_SPINLOCK(leak_lock);
-
-static inline
-void btrfs_leak_debug_add(struct list_head *new, struct list_head *head)
+static inline void btrfs_leak_debug_add_eb(struct extent_buffer *eb)
 {
+	struct btrfs_fs_info *fs_info = eb->fs_info;
 	unsigned long flags;
 
-	spin_lock_irqsave(&leak_lock, flags);
-	list_add(new, head);
-	spin_unlock_irqrestore(&leak_lock, flags);
+	spin_lock_irqsave(&fs_info->eb_leak_lock, flags);
+	list_add(&eb->leak_list, &fs_info->allocated_ebs);
+	spin_unlock_irqrestore(&fs_info->eb_leak_lock, flags);
 }
 
-static inline
-void btrfs_leak_debug_del(struct list_head *entry)
+static inline void btrfs_leak_debug_del_eb(struct extent_buffer *eb)
 {
+	struct btrfs_fs_info *fs_info = eb->fs_info;
 	unsigned long flags;
 
-	spin_lock_irqsave(&leak_lock, flags);
-	list_del(entry);
-	spin_unlock_irqrestore(&leak_lock, flags);
+	spin_lock_irqsave(&fs_info->eb_leak_lock, flags);
+	list_del(&eb->leak_list);
+	spin_unlock_irqrestore(&fs_info->eb_leak_lock, flags);
 }
 
-static inline void btrfs_extent_buffer_leak_debug_check(void)
+void btrfs_extent_buffer_leak_debug_check(struct btrfs_fs_info *fs_info)
 {
 	struct extent_buffer *eb;
+	unsigned long flags;
 
-	while (!list_empty(&buffers)) {
-		eb = list_entry(buffers.next, struct extent_buffer, leak_list);
-		pr_err("BTRFS: buffer leak start %llu len %lu refs %d bflags %lu\n",
-		       eb->start, eb->len, atomic_read(&eb->refs), eb->bflags);
-		list_del(&eb->leak_list);
-		kmem_cache_free(extent_buffer_cache, eb);
-	}
-}
-
-static inline void btrfs_extent_state_leak_debug_check(void)
-{
-	struct extent_state *state;
-
-	while (!list_empty(&states)) {
-		state = list_entry(states.next, struct extent_state, leak_list);
-		pr_err("BTRFS: state leak: start %llu end %llu state %u in tree %d refs %d\n",
-		       state->start, state->end, state->state,
-		       extent_state_in_tree(state),
-		       refcount_read(&state->refs));
-		list_del(&state->leak_list);
-		kmem_cache_free(extent_state_cache, state);
-	}
-}
-
-#define btrfs_debug_check_extent_io_range(tree, start, end)		\
-	__btrfs_debug_check_extent_io_range(__func__, (tree), (start), (end))
-static inline void __btrfs_debug_check_extent_io_range(const char *caller,
-		struct extent_io_tree *tree, u64 start, u64 end)
-{
-	struct inode *inode = tree->private_data;
-	u64 isize;
-
-	if (!inode || !is_data_inode(inode))
+	/*
+	 * If we didn't get into open_ctree our allocated_ebs will not be
+	 * initialized, so just skip this.
+	 */
+	if (!fs_info->allocated_ebs.next)
 		return;
 
-	isize = i_size_read(inode);
-	if (end >= PAGE_SIZE && (end % 2) == 0 && end != isize - 1) {
-		btrfs_debug_rl(BTRFS_I(inode)->root->fs_info,
-		    "%s: ino %llu isize %llu odd range [%llu,%llu]",
-			caller, btrfs_ino(BTRFS_I(inode)), isize, start, end);
+	WARN_ON(!list_empty(&fs_info->allocated_ebs));
+	spin_lock_irqsave(&fs_info->eb_leak_lock, flags);
+	while (!list_empty(&fs_info->allocated_ebs)) {
+		eb = list_first_entry(&fs_info->allocated_ebs,
+				      struct extent_buffer, leak_list);
+		pr_err(
+	"BTRFS: buffer leak start %llu len %lu refs %d bflags %lu owner %llu\n",
+		       eb->start, eb->len, atomic_read(&eb->refs), eb->bflags,
+		       btrfs_header_owner(eb));
+		list_del(&eb->leak_list);
+		kmem_cache_free(extent_buffer_cache, eb);
 	}
+	spin_unlock_irqrestore(&fs_info->eb_leak_lock, flags);
 }
 #else
-#define btrfs_leak_debug_add(new, head)	do {} while (0)
-#define btrfs_leak_debug_del(entry)	do {} while (0)
-#define btrfs_extent_buffer_leak_debug_check()	do {} while (0)
-#define btrfs_extent_state_leak_debug_check()	do {} while (0)
-#define btrfs_debug_check_extent_io_range(c, s, e)	do {} while (0)
+#define btrfs_leak_debug_add_eb(eb)			do {} while (0)
+#define btrfs_leak_debug_del_eb(eb)			do {} while (0)
 #endif
 
-struct tree_entry {
-	u64 start;
-	u64 end;
-	struct rb_node rb_node;
+/*
+ * Structure to record info about the bio being assembled, and other info like
+ * how many bytes are there before stripe/ordered extent boundary.
+ */
+struct btrfs_bio_ctrl {
+	struct bio *bio;
+	int mirror_num;
+	enum btrfs_compression_type compress_type;
+	u32 len_to_stripe_boundary;
+	u32 len_to_oe_boundary;
+	btrfs_bio_end_io_t end_io_func;
 };
 
 struct extent_page_data {
-	struct bio *bio;
-	struct extent_io_tree *tree;
+	struct btrfs_bio_ctrl bio_ctrl;
 	/* tells writepage not to lock the state bits for this range
 	 * it still does the unlocking
 	 */
@@ -132,86 +109,59 @@ struct extent_page_data {
 	unsigned int sync_io:1;
 };
 
-static int add_extent_changeset(struct extent_state *state, unsigned bits,
-				 struct extent_changeset *changeset,
-				 int set)
+static void submit_one_bio(struct btrfs_bio_ctrl *bio_ctrl)
 {
-	int ret;
+	struct bio *bio;
+	struct bio_vec *bv;
+	struct inode *inode;
+	int mirror_num;
 
-	if (!changeset)
-		return 0;
-	if (set && (state->state & bits) == bits)
-		return 0;
-	if (!set && (state->state & bits) == 0)
-		return 0;
-	changeset->bytes_changed += state->end - state->start + 1;
-	ret = ulist_add(&changeset->range_changed, state->start, state->end,
-			GFP_ATOMIC);
-	return ret;
-}
+	if (!bio_ctrl->bio)
+		return;
 
-static int __must_check submit_one_bio(struct bio *bio, int mirror_num,
-				       unsigned long bio_flags)
-{
-	blk_status_t ret = 0;
-	struct extent_io_tree *tree = bio->bi_private;
+	bio = bio_ctrl->bio;
+	bv = bio_first_bvec_all(bio);
+	inode = bv->bv_page->mapping->host;
+	mirror_num = bio_ctrl->mirror_num;
+
+	/* Caller should ensure the bio has at least some range added */
+	ASSERT(bio->bi_iter.bi_size);
 
-	bio->bi_private = NULL;
+	btrfs_bio(bio)->file_offset = page_offset(bv->bv_page) + bv->bv_offset;
 
-	if (tree->ops)
-		ret = tree->ops->submit_bio_hook(tree->private_data, bio,
-						 mirror_num, bio_flags);
+	if (!is_data_inode(inode))
+		btrfs_submit_metadata_bio(inode, bio, mirror_num);
+	else if (btrfs_op(bio) == BTRFS_MAP_WRITE)
+		btrfs_submit_data_write_bio(inode, bio, mirror_num);
 	else
-		btrfsic_submit_bio(bio);
+		btrfs_submit_data_read_bio(inode, bio, mirror_num,
+					   bio_ctrl->compress_type);
 
-	return blk_status_to_errno(ret);
-}
-
-/* Cleanup unsubmitted bios */
-static void end_write_bio(struct extent_page_data *epd, int ret)
-{
-	if (epd->bio) {
-		epd->bio->bi_status = errno_to_blk_status(ret);
-		bio_endio(epd->bio);
-		epd->bio = NULL;
-	}
+	/* The bio is owned by the end_io handler now */
+	bio_ctrl->bio = NULL;
 }
 
 /*
- * Submit bio from extent page data via submit_one_bio
- *
- * Return 0 if everything is OK.
- * Return <0 for error.
+ * Submit or fail the current bio in an extent_page_data structure.
  */
-static int __must_check flush_write_bio(struct extent_page_data *epd)
+static void submit_write_bio(struct extent_page_data *epd, int ret)
 {
-	int ret = 0;
+	struct bio *bio = epd->bio_ctrl.bio;
 
-	if (epd->bio) {
-		ret = submit_one_bio(epd->bio, 0, 0);
-		/*
-		 * Clean up of epd->bio is handled by its endio function.
-		 * And endio is either triggered by successful bio execution
-		 * or the error handler of submit bio hook.
-		 * So at this point, no matter what happened, we don't need
-		 * to clean up epd->bio.
-		 */
-		epd->bio = NULL;
-	}
-	return ret;
-}
+	if (!bio)
+		return;
 
-int __init extent_state_cache_init(void)
-{
-	extent_state_cache = kmem_cache_create("btrfs_extent_state",
-			sizeof(struct extent_state), 0,
-			SLAB_MEM_SPREAD, NULL);
-	if (!extent_state_cache)
-		return -ENOMEM;
-	return 0;
+	if (ret) {
+		ASSERT(ret < 0);
+		btrfs_bio_end_io(btrfs_bio(bio), errno_to_blk_status(ret));
+		/* The bio is owned by the end_io handler now */
+		epd->bio_ctrl.bio = NULL;
+	} else {
+		submit_one_bio(&epd->bio_ctrl);
+	}
 }
 
-int __init extent_io_init(void)
+int __init extent_buffer_init_cachep(void)
 {
 	extent_buffer_cache = kmem_cache_create("btrfs_extent_buffer",
 			sizeof(struct extent_buffer), 0,
@@ -219,1246 +169,17 @@ int __init extent_io_init(void)
 	if (!extent_buffer_cache)
 		return -ENOMEM;
 
-	if (bioset_init(&btrfs_bioset, BIO_POOL_SIZE,
-			offsetof(struct btrfs_io_bio, bio),
-			BIOSET_NEED_BVECS))
-		goto free_buffer_cache;
-
-	if (bioset_integrity_create(&btrfs_bioset, BIO_POOL_SIZE))
-		goto free_bioset;
-
 	return 0;
-
-free_bioset:
-	bioset_exit(&btrfs_bioset);
-
-free_buffer_cache:
-	kmem_cache_destroy(extent_buffer_cache);
-	extent_buffer_cache = NULL;
-	return -ENOMEM;
-}
-
-void __cold extent_state_cache_exit(void)
-{
-	btrfs_extent_state_leak_debug_check();
-	kmem_cache_destroy(extent_state_cache);
 }
 
-void __cold extent_io_exit(void)
+void __cold extent_buffer_free_cachep(void)
 {
-	btrfs_extent_buffer_leak_debug_check();
-
 	/*
 	 * Make sure all delayed rcu free are flushed before we
 	 * destroy caches.
 	 */
 	rcu_barrier();
 	kmem_cache_destroy(extent_buffer_cache);
-	bioset_exit(&btrfs_bioset);
-}
-
-void extent_io_tree_init(struct btrfs_fs_info *fs_info,
-			 struct extent_io_tree *tree, unsigned int owner,
-			 void *private_data)
-{
-	tree->fs_info = fs_info;
-	tree->state = RB_ROOT;
-	tree->ops = NULL;
-	tree->dirty_bytes = 0;
-	spin_lock_init(&tree->lock);
-	tree->private_data = private_data;
-	tree->owner = owner;
-}
-
-void extent_io_tree_release(struct extent_io_tree *tree)
-{
-	spin_lock(&tree->lock);
-	/*
-	 * Do a single barrier for the waitqueue_active check here, the state
-	 * of the waitqueue should not change once extent_io_tree_release is
-	 * called.
-	 */
-	smp_mb();
-	while (!RB_EMPTY_ROOT(&tree->state)) {
-		struct rb_node *node;
-		struct extent_state *state;
-
-		node = rb_first(&tree->state);
-		state = rb_entry(node, struct extent_state, rb_node);
-		rb_erase(&state->rb_node, &tree->state);
-		RB_CLEAR_NODE(&state->rb_node);
-		/*
-		 * btree io trees aren't supposed to have tasks waiting for
-		 * changes in the flags of extent states ever.
-		 */
-		ASSERT(!waitqueue_active(&state->wq));
-		free_extent_state(state);
-
-		cond_resched_lock(&tree->lock);
-	}
-	spin_unlock(&tree->lock);
-}
-
-static struct extent_state *alloc_extent_state(gfp_t mask)
-{
-	struct extent_state *state;
-
-	/*
-	 * The given mask might be not appropriate for the slab allocator,
-	 * drop the unsupported bits
-	 */
-	mask &= ~(__GFP_DMA32|__GFP_HIGHMEM);
-	state = kmem_cache_alloc(extent_state_cache, mask);
-	if (!state)
-		return state;
-	state->state = 0;
-	state->failrec = NULL;
-	RB_CLEAR_NODE(&state->rb_node);
-	btrfs_leak_debug_add(&state->leak_list, &states);
-	refcount_set(&state->refs, 1);
-	init_waitqueue_head(&state->wq);
-	trace_alloc_extent_state(state, mask, _RET_IP_);
-	return state;
-}
-
-void free_extent_state(struct extent_state *state)
-{
-	if (!state)
-		return;
-	if (refcount_dec_and_test(&state->refs)) {
-		WARN_ON(extent_state_in_tree(state));
-		btrfs_leak_debug_del(&state->leak_list);
-		trace_free_extent_state(state, _RET_IP_);
-		kmem_cache_free(extent_state_cache, state);
-	}
-}
-
-static struct rb_node *tree_insert(struct rb_root *root,
-				   struct rb_node *search_start,
-				   u64 offset,
-				   struct rb_node *node,
-				   struct rb_node ***p_in,
-				   struct rb_node **parent_in)
-{
-	struct rb_node **p;
-	struct rb_node *parent = NULL;
-	struct tree_entry *entry;
-
-	if (p_in && parent_in) {
-		p = *p_in;
-		parent = *parent_in;
-		goto do_insert;
-	}
-
-	p = search_start ? &search_start : &root->rb_node;
-	while (*p) {
-		parent = *p;
-		entry = rb_entry(parent, struct tree_entry, rb_node);
-
-		if (offset < entry->start)
-			p = &(*p)->rb_left;
-		else if (offset > entry->end)
-			p = &(*p)->rb_right;
-		else
-			return parent;
-	}
-
-do_insert:
-	rb_link_node(node, parent, p);
-	rb_insert_color(node, root);
-	return NULL;
-}
-
-/**
- * __etree_search - searche @tree for an entry that contains @offset. Such
- * entry would have entry->start <= offset && entry->end >= offset.
- *
- * @tree - the tree to search
- * @offset - offset that should fall within an entry in @tree
- * @next_ret - pointer to the first entry whose range ends after @offset
- * @prev - pointer to the first entry whose range begins before @offset
- * @p_ret - pointer where new node should be anchored (used when inserting an
- *	    entry in the tree)
- * @parent_ret - points to entry which would have been the parent of the entry,
- *               containing @offset
- *
- * This function returns a pointer to the entry that contains @offset byte
- * address. If no such entry exists, then NULL is returned and the other
- * pointer arguments to the function are filled, otherwise the found entry is
- * returned and other pointers are left untouched.
- */
-static struct rb_node *__etree_search(struct extent_io_tree *tree, u64 offset,
-				      struct rb_node **next_ret,
-				      struct rb_node **prev_ret,
-				      struct rb_node ***p_ret,
-				      struct rb_node **parent_ret)
-{
-	struct rb_root *root = &tree->state;
-	struct rb_node **n = &root->rb_node;
-	struct rb_node *prev = NULL;
-	struct rb_node *orig_prev = NULL;
-	struct tree_entry *entry;
-	struct tree_entry *prev_entry = NULL;
-
-	while (*n) {
-		prev = *n;
-		entry = rb_entry(prev, struct tree_entry, rb_node);
-		prev_entry = entry;
-
-		if (offset < entry->start)
-			n = &(*n)->rb_left;
-		else if (offset > entry->end)
-			n = &(*n)->rb_right;
-		else
-			return *n;
-	}
-
-	if (p_ret)
-		*p_ret = n;
-	if (parent_ret)
-		*parent_ret = prev;
-
-	if (next_ret) {
-		orig_prev = prev;
-		while (prev && offset > prev_entry->end) {
-			prev = rb_next(prev);
-			prev_entry = rb_entry(prev, struct tree_entry, rb_node);
-		}
-		*next_ret = prev;
-		prev = orig_prev;
-	}
-
-	if (prev_ret) {
-		prev_entry = rb_entry(prev, struct tree_entry, rb_node);
-		while (prev && offset < prev_entry->start) {
-			prev = rb_prev(prev);
-			prev_entry = rb_entry(prev, struct tree_entry, rb_node);
-		}
-		*prev_ret = prev;
-	}
-	return NULL;
-}
-
-static inline struct rb_node *
-tree_search_for_insert(struct extent_io_tree *tree,
-		       u64 offset,
-		       struct rb_node ***p_ret,
-		       struct rb_node **parent_ret)
-{
-	struct rb_node *next= NULL;
-	struct rb_node *ret;
-
-	ret = __etree_search(tree, offset, &next, NULL, p_ret, parent_ret);
-	if (!ret)
-		return next;
-	return ret;
-}
-
-static inline struct rb_node *tree_search(struct extent_io_tree *tree,
-					  u64 offset)
-{
-	return tree_search_for_insert(tree, offset, NULL, NULL);
-}
-
-/*
- * utility function to look for merge candidates inside a given range.
- * Any extents with matching state are merged together into a single
- * extent in the tree.  Extents with EXTENT_IO in their state field
- * are not merged because the end_io handlers need to be able to do
- * operations on them without sleeping (or doing allocations/splits).
- *
- * This should be called with the tree lock held.
- */
-static void merge_state(struct extent_io_tree *tree,
-		        struct extent_state *state)
-{
-	struct extent_state *other;
-	struct rb_node *other_node;
-
-	if (state->state & (EXTENT_LOCKED | EXTENT_BOUNDARY))
-		return;
-
-	other_node = rb_prev(&state->rb_node);
-	if (other_node) {
-		other = rb_entry(other_node, struct extent_state, rb_node);
-		if (other->end == state->start - 1 &&
-		    other->state == state->state) {
-			if (tree->private_data &&
-			    is_data_inode(tree->private_data))
-				btrfs_merge_delalloc_extent(tree->private_data,
-							    state, other);
-			state->start = other->start;
-			rb_erase(&other->rb_node, &tree->state);
-			RB_CLEAR_NODE(&other->rb_node);
-			free_extent_state(other);
-		}
-	}
-	other_node = rb_next(&state->rb_node);
-	if (other_node) {
-		other = rb_entry(other_node, struct extent_state, rb_node);
-		if (other->start == state->end + 1 &&
-		    other->state == state->state) {
-			if (tree->private_data &&
-			    is_data_inode(tree->private_data))
-				btrfs_merge_delalloc_extent(tree->private_data,
-							    state, other);
-			state->end = other->end;
-			rb_erase(&other->rb_node, &tree->state);
-			RB_CLEAR_NODE(&other->rb_node);
-			free_extent_state(other);
-		}
-	}
-}
-
-static void set_state_bits(struct extent_io_tree *tree,
-			   struct extent_state *state, unsigned *bits,
-			   struct extent_changeset *changeset);
-
-/*
- * insert an extent_state struct into the tree.  'bits' are set on the
- * struct before it is inserted.
- *
- * This may return -EEXIST if the extent is already there, in which case the
- * state struct is freed.
- *
- * The tree lock is not taken internally.  This is a utility function and
- * probably isn't what you want to call (see set/clear_extent_bit).
- */
-static int insert_state(struct extent_io_tree *tree,
-			struct extent_state *state, u64 start, u64 end,
-			struct rb_node ***p,
-			struct rb_node **parent,
-			unsigned *bits, struct extent_changeset *changeset)
-{
-	struct rb_node *node;
-
-	if (end < start) {
-		btrfs_err(tree->fs_info,
-			"insert state: end < start %llu %llu", end, start);
-		WARN_ON(1);
-	}
-	state->start = start;
-	state->end = end;
-
-	set_state_bits(tree, state, bits, changeset);
-
-	node = tree_insert(&tree->state, NULL, end, &state->rb_node, p, parent);
-	if (node) {
-		struct extent_state *found;
-		found = rb_entry(node, struct extent_state, rb_node);
-		btrfs_err(tree->fs_info,
-		       "found node %llu %llu on insert of %llu %llu",
-		       found->start, found->end, start, end);
-		return -EEXIST;
-	}
-	merge_state(tree, state);
-	return 0;
-}
-
-/*
- * split a given extent state struct in two, inserting the preallocated
- * struct 'prealloc' as the newly created second half.  'split' indicates an
- * offset inside 'orig' where it should be split.
- *
- * Before calling,
- * the tree has 'orig' at [orig->start, orig->end].  After calling, there
- * are two extent state structs in the tree:
- * prealloc: [orig->start, split - 1]
- * orig: [ split, orig->end ]
- *
- * The tree locks are not taken by this function. They need to be held
- * by the caller.
- */
-static int split_state(struct extent_io_tree *tree, struct extent_state *orig,
-		       struct extent_state *prealloc, u64 split)
-{
-	struct rb_node *node;
-
-	if (tree->private_data && is_data_inode(tree->private_data))
-		btrfs_split_delalloc_extent(tree->private_data, orig, split);
-
-	prealloc->start = orig->start;
-	prealloc->end = split - 1;
-	prealloc->state = orig->state;
-	orig->start = split;
-
-	node = tree_insert(&tree->state, &orig->rb_node, prealloc->end,
-			   &prealloc->rb_node, NULL, NULL);
-	if (node) {
-		free_extent_state(prealloc);
-		return -EEXIST;
-	}
-	return 0;
-}
-
-static struct extent_state *next_state(struct extent_state *state)
-{
-	struct rb_node *next = rb_next(&state->rb_node);
-	if (next)
-		return rb_entry(next, struct extent_state, rb_node);
-	else
-		return NULL;
-}
-
-/*
- * utility function to clear some bits in an extent state struct.
- * it will optionally wake up anyone waiting on this state (wake == 1).
- *
- * If no bits are set on the state struct after clearing things, the
- * struct is freed and removed from the tree
- */
-static struct extent_state *clear_state_bit(struct extent_io_tree *tree,
-					    struct extent_state *state,
-					    unsigned *bits, int wake,
-					    struct extent_changeset *changeset)
-{
-	struct extent_state *next;
-	unsigned bits_to_clear = *bits & ~EXTENT_CTLBITS;
-	int ret;
-
-	if ((bits_to_clear & EXTENT_DIRTY) && (state->state & EXTENT_DIRTY)) {
-		u64 range = state->end - state->start + 1;
-		WARN_ON(range > tree->dirty_bytes);
-		tree->dirty_bytes -= range;
-	}
-
-	if (tree->private_data && is_data_inode(tree->private_data))
-		btrfs_clear_delalloc_extent(tree->private_data, state, bits);
-
-	ret = add_extent_changeset(state, bits_to_clear, changeset, 0);
-	BUG_ON(ret < 0);
-	state->state &= ~bits_to_clear;
-	if (wake)
-		wake_up(&state->wq);
-	if (state->state == 0) {
-		next = next_state(state);
-		if (extent_state_in_tree(state)) {
-			rb_erase(&state->rb_node, &tree->state);
-			RB_CLEAR_NODE(&state->rb_node);
-			free_extent_state(state);
-		} else {
-			WARN_ON(1);
-		}
-	} else {
-		merge_state(tree, state);
-		next = next_state(state);
-	}
-	return next;
-}
-
-static struct extent_state *
-alloc_extent_state_atomic(struct extent_state *prealloc)
-{
-	if (!prealloc)
-		prealloc = alloc_extent_state(GFP_ATOMIC);
-
-	return prealloc;
-}
-
-static void extent_io_tree_panic(struct extent_io_tree *tree, int err)
-{
-	struct inode *inode = tree->private_data;
-
-	btrfs_panic(btrfs_sb(inode->i_sb), err,
-	"locking error: extent tree was modified by another thread while locked");
-}
-
-/*
- * clear some bits on a range in the tree.  This may require splitting
- * or inserting elements in the tree, so the gfp mask is used to
- * indicate which allocations or sleeping are allowed.
- *
- * pass 'wake' == 1 to kick any sleepers, and 'delete' == 1 to remove
- * the given range from the tree regardless of state (ie for truncate).
- *
- * the range [start, end] is inclusive.
- *
- * This takes the tree lock, and returns 0 on success and < 0 on error.
- */
-int __clear_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
-			      unsigned bits, int wake, int delete,
-			      struct extent_state **cached_state,
-			      gfp_t mask, struct extent_changeset *changeset)
-{
-	struct extent_state *state;
-	struct extent_state *cached;
-	struct extent_state *prealloc = NULL;
-	struct rb_node *node;
-	u64 last_end;
-	int err;
-	int clear = 0;
-
-	btrfs_debug_check_extent_io_range(tree, start, end);
-	trace_btrfs_clear_extent_bit(tree, start, end - start + 1, bits);
-
-	if (bits & EXTENT_DELALLOC)
-		bits |= EXTENT_NORESERVE;
-
-	if (delete)
-		bits |= ~EXTENT_CTLBITS;
-
-	if (bits & (EXTENT_LOCKED | EXTENT_BOUNDARY))
-		clear = 1;
-again:
-	if (!prealloc && gfpflags_allow_blocking(mask)) {
-		/*
-		 * Don't care for allocation failure here because we might end
-		 * up not needing the pre-allocated extent state at all, which
-		 * is the case if we only have in the tree extent states that
-		 * cover our input range and don't cover too any other range.
-		 * If we end up needing a new extent state we allocate it later.
-		 */
-		prealloc = alloc_extent_state(mask);
-	}
-
-	spin_lock(&tree->lock);
-	if (cached_state) {
-		cached = *cached_state;
-
-		if (clear) {
-			*cached_state = NULL;
-			cached_state = NULL;
-		}
-
-		if (cached && extent_state_in_tree(cached) &&
-		    cached->start <= start && cached->end > start) {
-			if (clear)
-				refcount_dec(&cached->refs);
-			state = cached;
-			goto hit_next;
-		}
-		if (clear)
-			free_extent_state(cached);
-	}
-	/*
-	 * this search will find the extents that end after
-	 * our range starts
-	 */
-	node = tree_search(tree, start);
-	if (!node)
-		goto out;
-	state = rb_entry(node, struct extent_state, rb_node);
-hit_next:
-	if (state->start > end)
-		goto out;
-	WARN_ON(state->end < start);
-	last_end = state->end;
-
-	/* the state doesn't have the wanted bits, go ahead */
-	if (!(state->state & bits)) {
-		state = next_state(state);
-		goto next;
-	}
-
-	/*
-	 *     | ---- desired range ---- |
-	 *  | state | or
-	 *  | ------------- state -------------- |
-	 *
-	 * We need to split the extent we found, and may flip
-	 * bits on second half.
-	 *
-	 * If the extent we found extends past our range, we
-	 * just split and search again.  It'll get split again
-	 * the next time though.
-	 *
-	 * If the extent we found is inside our range, we clear
-	 * the desired bit on it.
-	 */
-
-	if (state->start < start) {
-		prealloc = alloc_extent_state_atomic(prealloc);
-		BUG_ON(!prealloc);
-		err = split_state(tree, state, prealloc, start);
-		if (err)
-			extent_io_tree_panic(tree, err);
-
-		prealloc = NULL;
-		if (err)
-			goto out;
-		if (state->end <= end) {
-			state = clear_state_bit(tree, state, &bits, wake,
-						changeset);
-			goto next;
-		}
-		goto search_again;
-	}
-	/*
-	 * | ---- desired range ---- |
-	 *                        | state |
-	 * We need to split the extent, and clear the bit
-	 * on the first half
-	 */
-	if (state->start <= end && state->end > end) {
-		prealloc = alloc_extent_state_atomic(prealloc);
-		BUG_ON(!prealloc);
-		err = split_state(tree, state, prealloc, end + 1);
-		if (err)
-			extent_io_tree_panic(tree, err);
-
-		if (wake)
-			wake_up(&state->wq);
-
-		clear_state_bit(tree, prealloc, &bits, wake, changeset);
-
-		prealloc = NULL;
-		goto out;
-	}
-
-	state = clear_state_bit(tree, state, &bits, wake, changeset);
-next:
-	if (last_end == (u64)-1)
-		goto out;
-	start = last_end + 1;
-	if (start <= end && state && !need_resched())
-		goto hit_next;
-
-search_again:
-	if (start > end)
-		goto out;
-	spin_unlock(&tree->lock);
-	if (gfpflags_allow_blocking(mask))
-		cond_resched();
-	goto again;
-
-out:
-	spin_unlock(&tree->lock);
-	if (prealloc)
-		free_extent_state(prealloc);
-
-	return 0;
-
-}
-
-static void wait_on_state(struct extent_io_tree *tree,
-			  struct extent_state *state)
-		__releases(tree->lock)
-		__acquires(tree->lock)
-{
-	DEFINE_WAIT(wait);
-	prepare_to_wait(&state->wq, &wait, TASK_UNINTERRUPTIBLE);
-	spin_unlock(&tree->lock);
-	schedule();
-	spin_lock(&tree->lock);
-	finish_wait(&state->wq, &wait);
-}
-
-/*
- * waits for one or more bits to clear on a range in the state tree.
- * The range [start, end] is inclusive.
- * The tree lock is taken by this function
- */
-static void wait_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
-			    unsigned long bits)
-{
-	struct extent_state *state;
-	struct rb_node *node;
-
-	btrfs_debug_check_extent_io_range(tree, start, end);
-
-	spin_lock(&tree->lock);
-again:
-	while (1) {
-		/*
-		 * this search will find all the extents that end after
-		 * our range starts
-		 */
-		node = tree_search(tree, start);
-process_node:
-		if (!node)
-			break;
-
-		state = rb_entry(node, struct extent_state, rb_node);
-
-		if (state->start > end)
-			goto out;
-
-		if (state->state & bits) {
-			start = state->start;
-			refcount_inc(&state->refs);
-			wait_on_state(tree, state);
-			free_extent_state(state);
-			goto again;
-		}
-		start = state->end + 1;
-
-		if (start > end)
-			break;
-
-		if (!cond_resched_lock(&tree->lock)) {
-			node = rb_next(node);
-			goto process_node;
-		}
-	}
-out:
-	spin_unlock(&tree->lock);
-}
-
-static void set_state_bits(struct extent_io_tree *tree,
-			   struct extent_state *state,
-			   unsigned *bits, struct extent_changeset *changeset)
-{
-	unsigned bits_to_set = *bits & ~EXTENT_CTLBITS;
-	int ret;
-
-	if (tree->private_data && is_data_inode(tree->private_data))
-		btrfs_set_delalloc_extent(tree->private_data, state, bits);
-
-	if ((bits_to_set & EXTENT_DIRTY) && !(state->state & EXTENT_DIRTY)) {
-		u64 range = state->end - state->start + 1;
-		tree->dirty_bytes += range;
-	}
-	ret = add_extent_changeset(state, bits_to_set, changeset, 1);
-	BUG_ON(ret < 0);
-	state->state |= bits_to_set;
-}
-
-static void cache_state_if_flags(struct extent_state *state,
-				 struct extent_state **cached_ptr,
-				 unsigned flags)
-{
-	if (cached_ptr && !(*cached_ptr)) {
-		if (!flags || (state->state & flags)) {
-			*cached_ptr = state;
-			refcount_inc(&state->refs);
-		}
-	}
-}
-
-static void cache_state(struct extent_state *state,
-			struct extent_state **cached_ptr)
-{
-	return cache_state_if_flags(state, cached_ptr,
-				    EXTENT_LOCKED | EXTENT_BOUNDARY);
-}
-
-/*
- * set some bits on a range in the tree.  This may require allocations or
- * sleeping, so the gfp mask is used to indicate what is allowed.
- *
- * If any of the exclusive bits are set, this will fail with -EEXIST if some
- * part of the range already has the desired bits set.  The start of the
- * existing range is returned in failed_start in this case.
- *
- * [start, end] is inclusive This takes the tree lock.
- */
-
-static int __must_check
-__set_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
-		 unsigned bits, unsigned exclusive_bits,
-		 u64 *failed_start, struct extent_state **cached_state,
-		 gfp_t mask, struct extent_changeset *changeset)
-{
-	struct extent_state *state;
-	struct extent_state *prealloc = NULL;
-	struct rb_node *node;
-	struct rb_node **p;
-	struct rb_node *parent;
-	int err = 0;
-	u64 last_start;
-	u64 last_end;
-
-	btrfs_debug_check_extent_io_range(tree, start, end);
-	trace_btrfs_set_extent_bit(tree, start, end - start + 1, bits);
-
-again:
-	if (!prealloc && gfpflags_allow_blocking(mask)) {
-		/*
-		 * Don't care for allocation failure here because we might end
-		 * up not needing the pre-allocated extent state at all, which
-		 * is the case if we only have in the tree extent states that
-		 * cover our input range and don't cover too any other range.
-		 * If we end up needing a new extent state we allocate it later.
-		 */
-		prealloc = alloc_extent_state(mask);
-	}
-
-	spin_lock(&tree->lock);
-	if (cached_state && *cached_state) {
-		state = *cached_state;
-		if (state->start <= start && state->end > start &&
-		    extent_state_in_tree(state)) {
-			node = &state->rb_node;
-			goto hit_next;
-		}
-	}
-	/*
-	 * this search will find all the extents that end after
-	 * our range starts.
-	 */
-	node = tree_search_for_insert(tree, start, &p, &parent);
-	if (!node) {
-		prealloc = alloc_extent_state_atomic(prealloc);
-		BUG_ON(!prealloc);
-		err = insert_state(tree, prealloc, start, end,
-				   &p, &parent, &bits, changeset);
-		if (err)
-			extent_io_tree_panic(tree, err);
-
-		cache_state(prealloc, cached_state);
-		prealloc = NULL;
-		goto out;
-	}
-	state = rb_entry(node, struct extent_state, rb_node);
-hit_next:
-	last_start = state->start;
-	last_end = state->end;
-
-	/*
-	 * | ---- desired range ---- |
-	 * | state |
-	 *
-	 * Just lock what we found and keep going
-	 */
-	if (state->start == start && state->end <= end) {
-		if (state->state & exclusive_bits) {
-			*failed_start = state->start;
-			err = -EEXIST;
-			goto out;
-		}
-
-		set_state_bits(tree, state, &bits, changeset);
-		cache_state(state, cached_state);
-		merge_state(tree, state);
-		if (last_end == (u64)-1)
-			goto out;
-		start = last_end + 1;
-		state = next_state(state);
-		if (start < end && state && state->start == start &&
-		    !need_resched())
-			goto hit_next;
-		goto search_again;
-	}
-
-	/*
-	 *     | ---- desired range ---- |
-	 * | state |
-	 *   or
-	 * | ------------- state -------------- |
-	 *
-	 * We need to split the extent we found, and may flip bits on
-	 * second half.
-	 *
-	 * If the extent we found extends past our
-	 * range, we just split and search again.  It'll get split
-	 * again the next time though.
-	 *
-	 * If the extent we found is inside our range, we set the
-	 * desired bit on it.
-	 */
-	if (state->start < start) {
-		if (state->state & exclusive_bits) {
-			*failed_start = start;
-			err = -EEXIST;
-			goto out;
-		}
-
-		prealloc = alloc_extent_state_atomic(prealloc);
-		BUG_ON(!prealloc);
-		err = split_state(tree, state, prealloc, start);
-		if (err)
-			extent_io_tree_panic(tree, err);
-
-		prealloc = NULL;
-		if (err)
-			goto out;
-		if (state->end <= end) {
-			set_state_bits(tree, state, &bits, changeset);
-			cache_state(state, cached_state);
-			merge_state(tree, state);
-			if (last_end == (u64)-1)
-				goto out;
-			start = last_end + 1;
-			state = next_state(state);
-			if (start < end && state && state->start == start &&
-			    !need_resched())
-				goto hit_next;
-		}
-		goto search_again;
-	}
-	/*
-	 * | ---- desired range ---- |
-	 *     | state | or               | state |
-	 *
-	 * There's a hole, we need to insert something in it and
-	 * ignore the extent we found.
-	 */
-	if (state->start > start) {
-		u64 this_end;
-		if (end < last_start)
-			this_end = end;
-		else
-			this_end = last_start - 1;
-
-		prealloc = alloc_extent_state_atomic(prealloc);
-		BUG_ON(!prealloc);
-
-		/*
-		 * Avoid to free 'prealloc' if it can be merged with
-		 * the later extent.
-		 */
-		err = insert_state(tree, prealloc, start, this_end,
-				   NULL, NULL, &bits, changeset);
-		if (err)
-			extent_io_tree_panic(tree, err);
-
-		cache_state(prealloc, cached_state);
-		prealloc = NULL;
-		start = this_end + 1;
-		goto search_again;
-	}
-	/*
-	 * | ---- desired range ---- |
-	 *                        | state |
-	 * We need to split the extent, and set the bit
-	 * on the first half
-	 */
-	if (state->start <= end && state->end > end) {
-		if (state->state & exclusive_bits) {
-			*failed_start = start;
-			err = -EEXIST;
-			goto out;
-		}
-
-		prealloc = alloc_extent_state_atomic(prealloc);
-		BUG_ON(!prealloc);
-		err = split_state(tree, state, prealloc, end + 1);
-		if (err)
-			extent_io_tree_panic(tree, err);
-
-		set_state_bits(tree, prealloc, &bits, changeset);
-		cache_state(prealloc, cached_state);
-		merge_state(tree, prealloc);
-		prealloc = NULL;
-		goto out;
-	}
-
-search_again:
-	if (start > end)
-		goto out;
-	spin_unlock(&tree->lock);
-	if (gfpflags_allow_blocking(mask))
-		cond_resched();
-	goto again;
-
-out:
-	spin_unlock(&tree->lock);
-	if (prealloc)
-		free_extent_state(prealloc);
-
-	return err;
-
-}
-
-int set_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
-		   unsigned bits, u64 * failed_start,
-		   struct extent_state **cached_state, gfp_t mask)
-{
-	return __set_extent_bit(tree, start, end, bits, 0, failed_start,
-				cached_state, mask, NULL);
-}
-
-
-/**
- * convert_extent_bit - convert all bits in a given range from one bit to
- * 			another
- * @tree:	the io tree to search
- * @start:	the start offset in bytes
- * @end:	the end offset in bytes (inclusive)
- * @bits:	the bits to set in this range
- * @clear_bits:	the bits to clear in this range
- * @cached_state:	state that we're going to cache
- *
- * This will go through and set bits for the given range.  If any states exist
- * already in this range they are set with the given bit and cleared of the
- * clear_bits.  This is only meant to be used by things that are mergeable, ie
- * converting from say DELALLOC to DIRTY.  This is not meant to be used with
- * boundary bits like LOCK.
- *
- * All allocations are done with GFP_NOFS.
- */
-int convert_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
-		       unsigned bits, unsigned clear_bits,
-		       struct extent_state **cached_state)
-{
-	struct extent_state *state;
-	struct extent_state *prealloc = NULL;
-	struct rb_node *node;
-	struct rb_node **p;
-	struct rb_node *parent;
-	int err = 0;
-	u64 last_start;
-	u64 last_end;
-	bool first_iteration = true;
-
-	btrfs_debug_check_extent_io_range(tree, start, end);
-	trace_btrfs_convert_extent_bit(tree, start, end - start + 1, bits,
-				       clear_bits);
-
-again:
-	if (!prealloc) {
-		/*
-		 * Best effort, don't worry if extent state allocation fails
-		 * here for the first iteration. We might have a cached state
-		 * that matches exactly the target range, in which case no
-		 * extent state allocations are needed. We'll only know this
-		 * after locking the tree.
-		 */
-		prealloc = alloc_extent_state(GFP_NOFS);
-		if (!prealloc && !first_iteration)
-			return -ENOMEM;
-	}
-
-	spin_lock(&tree->lock);
-	if (cached_state && *cached_state) {
-		state = *cached_state;
-		if (state->start <= start && state->end > start &&
-		    extent_state_in_tree(state)) {
-			node = &state->rb_node;
-			goto hit_next;
-		}
-	}
-
-	/*
-	 * this search will find all the extents that end after
-	 * our range starts.
-	 */
-	node = tree_search_for_insert(tree, start, &p, &parent);
-	if (!node) {
-		prealloc = alloc_extent_state_atomic(prealloc);
-		if (!prealloc) {
-			err = -ENOMEM;
-			goto out;
-		}
-		err = insert_state(tree, prealloc, start, end,
-				   &p, &parent, &bits, NULL);
-		if (err)
-			extent_io_tree_panic(tree, err);
-		cache_state(prealloc, cached_state);
-		prealloc = NULL;
-		goto out;
-	}
-	state = rb_entry(node, struct extent_state, rb_node);
-hit_next:
-	last_start = state->start;
-	last_end = state->end;
-
-	/*
-	 * | ---- desired range ---- |
-	 * | state |
-	 *
-	 * Just lock what we found and keep going
-	 */
-	if (state->start == start && state->end <= end) {
-		set_state_bits(tree, state, &bits, NULL);
-		cache_state(state, cached_state);
-		state = clear_state_bit(tree, state, &clear_bits, 0, NULL);
-		if (last_end == (u64)-1)
-			goto out;
-		start = last_end + 1;
-		if (start < end && state && state->start == start &&
-		    !need_resched())
-			goto hit_next;
-		goto search_again;
-	}
-
-	/*
-	 *     | ---- desired range ---- |
-	 * | state |
-	 *   or
-	 * | ------------- state -------------- |
-	 *
-	 * We need to split the extent we found, and may flip bits on
-	 * second half.
-	 *
-	 * If the extent we found extends past our
-	 * range, we just split and search again.  It'll get split
-	 * again the next time though.
-	 *
-	 * If the extent we found is inside our range, we set the
-	 * desired bit on it.
-	 */
-	if (state->start < start) {
-		prealloc = alloc_extent_state_atomic(prealloc);
-		if (!prealloc) {
-			err = -ENOMEM;
-			goto out;
-		}
-		err = split_state(tree, state, prealloc, start);
-		if (err)
-			extent_io_tree_panic(tree, err);
-		prealloc = NULL;
-		if (err)
-			goto out;
-		if (state->end <= end) {
-			set_state_bits(tree, state, &bits, NULL);
-			cache_state(state, cached_state);
-			state = clear_state_bit(tree, state, &clear_bits, 0,
-						NULL);
-			if (last_end == (u64)-1)
-				goto out;
-			start = last_end + 1;
-			if (start < end && state && state->start == start &&
-			    !need_resched())
-				goto hit_next;
-		}
-		goto search_again;
-	}
-	/*
-	 * | ---- desired range ---- |
-	 *     | state | or               | state |
-	 *
-	 * There's a hole, we need to insert something in it and
-	 * ignore the extent we found.
-	 */
-	if (state->start > start) {
-		u64 this_end;
-		if (end < last_start)
-			this_end = end;
-		else
-			this_end = last_start - 1;
-
-		prealloc = alloc_extent_state_atomic(prealloc);
-		if (!prealloc) {
-			err = -ENOMEM;
-			goto out;
-		}
-
-		/*
-		 * Avoid to free 'prealloc' if it can be merged with
-		 * the later extent.
-		 */
-		err = insert_state(tree, prealloc, start, this_end,
-				   NULL, NULL, &bits, NULL);
-		if (err)
-			extent_io_tree_panic(tree, err);
-		cache_state(prealloc, cached_state);
-		prealloc = NULL;
-		start = this_end + 1;
-		goto search_again;
-	}
-	/*
-	 * | ---- desired range ---- |
-	 *                        | state |
-	 * We need to split the extent, and set the bit
-	 * on the first half
-	 */
-	if (state->start <= end && state->end > end) {
-		prealloc = alloc_extent_state_atomic(prealloc);
-		if (!prealloc) {
-			err = -ENOMEM;
-			goto out;
-		}
-
-		err = split_state(tree, state, prealloc, end + 1);
-		if (err)
-			extent_io_tree_panic(tree, err);
-
-		set_state_bits(tree, prealloc, &bits, NULL);
-		cache_state(prealloc, cached_state);
-		clear_state_bit(tree, prealloc, &clear_bits, 0, NULL);
-		prealloc = NULL;
-		goto out;
-	}
-
-search_again:
-	if (start > end)
-		goto out;
-	spin_unlock(&tree->lock);
-	cond_resched();
-	first_iteration = false;
-	goto again;
-
-out:
-	spin_unlock(&tree->lock);
-	if (prealloc)
-		free_extent_state(prealloc);
-
-	return err;
-}
-
-/* wrappers around set/clear extent bit */
-int set_record_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
-			   unsigned bits, struct extent_changeset *changeset)
-{
-	/*
-	 * We don't support EXTENT_LOCKED yet, as current changeset will
-	 * record any bits changed, so for EXTENT_LOCKED case, it will
-	 * either fail with -EEXIST or changeset will record the whole
-	 * range.
-	 */
-	BUG_ON(bits & EXTENT_LOCKED);
-
-	return __set_extent_bit(tree, start, end, bits, 0, NULL, NULL, GFP_NOFS,
-				changeset);
-}
-
-int set_extent_bits_nowait(struct extent_io_tree *tree, u64 start, u64 end,
-			   unsigned bits)
-{
-	return __set_extent_bit(tree, start, end, bits, 0, NULL, NULL,
-				GFP_NOWAIT, NULL);
-}
-
-int clear_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
-		     unsigned bits, int wake, int delete,
-		     struct extent_state **cached)
-{
-	return __clear_extent_bit(tree, start, end, bits, wake, delete,
-				  cached, GFP_NOFS, NULL);
-}
-
-int clear_record_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
-		unsigned bits, struct extent_changeset *changeset)
-{
-	/*
-	 * Don't support EXTENT_LOCKED case, same reason as
-	 * set_record_extent_bits().
-	 */
-	BUG_ON(bits & EXTENT_LOCKED);
-
-	return __clear_extent_bit(tree, start, end, bits, 0, 0, NULL, GFP_NOFS,
-				  changeset);
-}
-
-/*
- * either insert or lock state struct between start and end use mask to tell
- * us if waiting is desired.
- */
-int lock_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
-		     struct extent_state **cached_state)
-{
-	int err;
-	u64 failed_start;
-
-	while (1) {
-		err = __set_extent_bit(tree, start, end, EXTENT_LOCKED,
-				       EXTENT_LOCKED, &failed_start,
-				       cached_state, GFP_NOFS, NULL);
-		if (err == -EEXIST) {
-			wait_extent_bit(tree, failed_start, end, EXTENT_LOCKED);
-			start = failed_start;
-		} else
-			break;
-		WARN_ON(start > end);
-	}
-	return err;
-}
-
-int try_lock_extent(struct extent_io_tree *tree, u64 start, u64 end)
-{
-	int err;
-	u64 failed_start;
-
-	err = __set_extent_bit(tree, start, end, EXTENT_LOCKED, EXTENT_LOCKED,
-			       &failed_start, NULL, GFP_NOFS, NULL);
-	if (err == -EEXIST) {
-		if (failed_start > start)
-			clear_extent_bit(tree, start, failed_start - 1,
-					 EXTENT_LOCKED, 1, 0, NULL);
-		return 0;
-	}
-	return 1;
 }
 
 void extent_range_clear_dirty_for_io(struct inode *inode, u64 start, u64 end)
@@ -1478,276 +199,142 @@ void extent_range_clear_dirty_for_io(struct inode *inode, u64 start, u64 end)
 
 void extent_range_redirty_for_io(struct inode *inode, u64 start, u64 end)
 {
+	struct address_space *mapping = inode->i_mapping;
 	unsigned long index = start >> PAGE_SHIFT;
 	unsigned long end_index = end >> PAGE_SHIFT;
-	struct page *page;
+	struct folio *folio;
 
 	while (index <= end_index) {
-		page = find_get_page(inode->i_mapping, index);
-		BUG_ON(!page); /* Pages should be in the extent_io_tree */
-		__set_page_dirty_nobuffers(page);
-		account_page_redirty(page);
-		put_page(page);
-		index++;
+		folio = filemap_get_folio(mapping, index);
+		filemap_dirty_folio(mapping, folio);
+		folio_account_redirty(folio);
+		index += folio_nr_pages(folio);
+		folio_put(folio);
 	}
 }
 
-/* find the first state struct with 'bits' set after 'start', and
- * return it.  tree->lock must be held.  NULL will returned if
- * nothing was found after 'start'
+/*
+ * Process one page for __process_pages_contig().
+ *
+ * Return >0 if we hit @page == @locked_page.
+ * Return 0 if we updated the page status.
+ * Return -EGAIN if the we need to try again.
+ * (For PAGE_LOCK case but got dirty page or page not belong to mapping)
  */
-static struct extent_state *
-find_first_extent_bit_state(struct extent_io_tree *tree,
-			    u64 start, unsigned bits)
+static int process_one_page(struct btrfs_fs_info *fs_info,
+			    struct address_space *mapping,
+			    struct page *page, struct page *locked_page,
+			    unsigned long page_ops, u64 start, u64 end)
 {
-	struct rb_node *node;
-	struct extent_state *state;
+	u32 len;
 
-	/*
-	 * this search will find all the extents that end after
-	 * our range starts.
-	 */
-	node = tree_search(tree, start);
-	if (!node)
-		goto out;
+	ASSERT(end + 1 - start != 0 && end + 1 - start < U32_MAX);
+	len = end + 1 - start;
 
-	while (1) {
-		state = rb_entry(node, struct extent_state, rb_node);
-		if (state->end >= start && (state->state & bits))
-			return state;
+	if (page_ops & PAGE_SET_ORDERED)
+		btrfs_page_clamp_set_ordered(fs_info, page, start, len);
+	if (page_ops & PAGE_SET_ERROR)
+		btrfs_page_clamp_set_error(fs_info, page, start, len);
+	if (page_ops & PAGE_START_WRITEBACK) {
+		btrfs_page_clamp_clear_dirty(fs_info, page, start, len);
+		btrfs_page_clamp_set_writeback(fs_info, page, start, len);
+	}
+	if (page_ops & PAGE_END_WRITEBACK)
+		btrfs_page_clamp_clear_writeback(fs_info, page, start, len);
 
-		node = rb_next(node);
-		if (!node)
-			break;
+	if (page == locked_page)
+		return 1;
+
+	if (page_ops & PAGE_LOCK) {
+		int ret;
+
+		ret = btrfs_page_start_writer_lock(fs_info, page, start, len);
+		if (ret)
+			return ret;
+		if (!PageDirty(page) || page->mapping != mapping) {
+			btrfs_page_end_writer_lock(fs_info, page, start, len);
+			return -EAGAIN;
+		}
 	}
-out:
-	return NULL;
+	if (page_ops & PAGE_UNLOCK)
+		btrfs_page_end_writer_lock(fs_info, page, start, len);
+	return 0;
 }
 
-/*
- * find the first offset in the io tree with 'bits' set. zero is
- * returned if we find something, and *start_ret and *end_ret are
- * set to reflect the state struct that was found.
- *
- * If nothing was found, 1 is returned. If found something, return 0.
- */
-int find_first_extent_bit(struct extent_io_tree *tree, u64 start,
-			  u64 *start_ret, u64 *end_ret, unsigned bits,
-			  struct extent_state **cached_state)
+static int __process_pages_contig(struct address_space *mapping,
+				  struct page *locked_page,
+				  u64 start, u64 end, unsigned long page_ops,
+				  u64 *processed_end)
 {
-	struct extent_state *state;
-	int ret = 1;
+	struct btrfs_fs_info *fs_info = btrfs_sb(mapping->host->i_sb);
+	pgoff_t start_index = start >> PAGE_SHIFT;
+	pgoff_t end_index = end >> PAGE_SHIFT;
+	pgoff_t index = start_index;
+	unsigned long pages_processed = 0;
+	struct folio_batch fbatch;
+	int err = 0;
+	int i;
 
-	spin_lock(&tree->lock);
-	if (cached_state && *cached_state) {
-		state = *cached_state;
-		if (state->end == start - 1 && extent_state_in_tree(state)) {
-			while ((state = next_state(state)) != NULL) {
-				if (state->state & bits)
-					goto got_it;
-			}
-			free_extent_state(*cached_state);
-			*cached_state = NULL;
-			goto out;
-		}
-		free_extent_state(*cached_state);
-		*cached_state = NULL;
+	if (page_ops & PAGE_LOCK) {
+		ASSERT(page_ops == PAGE_LOCK);
+		ASSERT(processed_end && *processed_end == start);
 	}
 
-	state = find_first_extent_bit_state(tree, start, bits);
-got_it:
-	if (state) {
-		cache_state_if_flags(state, cached_state, 0);
-		*start_ret = state->start;
-		*end_ret = state->end;
-		ret = 0;
-	}
-out:
-	spin_unlock(&tree->lock);
-	return ret;
-}
+	if ((page_ops & PAGE_SET_ERROR) && start_index <= end_index)
+		mapping_set_error(mapping, -EIO);
 
-/**
- * find_first_clear_extent_bit - find the first range that has @bits not set.
- * This range could start before @start.
- *
- * @tree - the tree to search
- * @start - the offset at/after which the found extent should start
- * @start_ret - records the beginning of the range
- * @end_ret - records the end of the range (inclusive)
- * @bits - the set of bits which must be unset
- *
- * Since unallocated range is also considered one which doesn't have the bits
- * set it's possible that @end_ret contains -1, this happens in case the range
- * spans (last_range_end, end of device]. In this case it's up to the caller to
- * trim @end_ret to the appropriate size.
- */
-void find_first_clear_extent_bit(struct extent_io_tree *tree, u64 start,
-				 u64 *start_ret, u64 *end_ret, unsigned bits)
-{
-	struct extent_state *state;
-	struct rb_node *node, *prev = NULL, *next;
+	folio_batch_init(&fbatch);
+	while (index <= end_index) {
+		int found_folios;
 
-	spin_lock(&tree->lock);
+		found_folios = filemap_get_folios_contig(mapping, &index,
+				end_index, &fbatch);
 
-	/* Find first extent with bits cleared */
-	while (1) {
-		node = __etree_search(tree, start, &next, &prev, NULL, NULL);
-		if (!node && !next && !prev) {
-			/*
-			 * Tree is completely empty, send full range and let
-			 * caller deal with it
-			 */
-			*start_ret = 0;
-			*end_ret = -1;
-			goto out;
-		} else if (!node && !next) {
+		if (found_folios == 0) {
 			/*
-			 * We are past the last allocated chunk, set start at
-			 * the end of the last extent.
+			 * Only if we're going to lock these pages, we can find
+			 * nothing at @index.
 			 */
-			state = rb_entry(prev, struct extent_state, rb_node);
-			*start_ret = state->end + 1;
-			*end_ret = -1;
+			ASSERT(page_ops & PAGE_LOCK);
+			err = -EAGAIN;
 			goto out;
-		} else if (!node) {
-			node = next;
 		}
-		/*
-		 * At this point 'node' either contains 'start' or start is
-		 * before 'node'
-		 */
-		state = rb_entry(node, struct extent_state, rb_node);
 
-		if (in_range(start, state->start, state->end - state->start + 1)) {
-			if (state->state & bits) {
-				/*
-				 * |--range with bits sets--|
-				 *    |
-				 *    start
-				 */
-				start = state->end + 1;
-			} else {
-				/*
-				 * 'start' falls within a range that doesn't
-				 * have the bits set, so take its start as
-				 * the beginning of the desired range
-				 *
-				 * |--range with bits cleared----|
-				 *      |
-				 *      start
-				 */
-				*start_ret = state->start;
-				break;
-			}
-		} else {
-			/*
-			 * |---prev range---|---hole/unset---|---node range---|
-			 *                          |
-			 *                        start
-			 *
-			 *                        or
-			 *
-			 * |---hole/unset--||--first node--|
-			 * 0   |
-			 *    start
-			 */
-			if (prev) {
-				state = rb_entry(prev, struct extent_state,
-						 rb_node);
-				*start_ret = state->end + 1;
-			} else {
-				*start_ret = 0;
+		for (i = 0; i < found_folios; i++) {
+			int process_ret;
+			struct folio *folio = fbatch.folios[i];
+			process_ret = process_one_page(fs_info, mapping,
+					&folio->page, locked_page, page_ops,
+					start, end);
+			if (process_ret < 0) {
+				err = -EAGAIN;
+				folio_batch_release(&fbatch);
+				goto out;
 			}
-			break;
+			pages_processed += folio_nr_pages(folio);
 		}
-	}
-
-	/*
-	 * Find the longest stretch from start until an entry which has the
-	 * bits set
-	 */
-	while (1) {
-		state = rb_entry(node, struct extent_state, rb_node);
-		if (state->end >= start && !(state->state & bits)) {
-			*end_ret = state->end;
-		} else {
-			*end_ret = state->start - 1;
-			break;
-		}
-
-		node = rb_next(node);
-		if (!node)
-			break;
+		folio_batch_release(&fbatch);
+		cond_resched();
 	}
 out:
-	spin_unlock(&tree->lock);
-}
-
-/*
- * find a contiguous range of bytes in the file marked as delalloc, not
- * more than 'max_bytes'.  start and end are used to return the range,
- *
- * true is returned if we find something, false if nothing was in the tree
- */
-bool btrfs_find_delalloc_range(struct extent_io_tree *tree, u64 *start,
-			       u64 *end, u64 max_bytes,
-			       struct extent_state **cached_state)
-{
-	struct rb_node *node;
-	struct extent_state *state;
-	u64 cur_start = *start;
-	bool found = false;
-	u64 total_bytes = 0;
-
-	spin_lock(&tree->lock);
-
-	/*
-	 * this search will find all the extents that end after
-	 * our range starts.
-	 */
-	node = tree_search(tree, cur_start);
-	if (!node) {
-		*end = (u64)-1;
-		goto out;
-	}
-
-	while (1) {
-		state = rb_entry(node, struct extent_state, rb_node);
-		if (found && (state->start != cur_start ||
-			      (state->state & EXTENT_BOUNDARY))) {
-			goto out;
-		}
-		if (!(state->state & EXTENT_DELALLOC)) {
-			if (!found)
-				*end = state->end;
-			goto out;
-		}
-		if (!found) {
-			*start = state->start;
-			*cached_state = state;
-			refcount_inc(&state->refs);
-		}
-		found = true;
-		*end = state->end;
-		cur_start = state->end + 1;
-		node = rb_next(node);
-		total_bytes += state->end - state->start + 1;
-		if (total_bytes >= max_bytes)
-			break;
-		if (!node)
-			break;
+	if (err && processed_end) {
+		/*
+		 * Update @processed_end. I know this is awful since it has
+		 * two different return value patterns (inclusive vs exclusive).
+		 *
+		 * But the exclusive pattern is necessary if @start is 0, or we
+		 * underflow and check against processed_end won't work as
+		 * expected.
+		 */
+		if (pages_processed)
+			*processed_end = min(end,
+			((u64)(start_index + pages_processed) << PAGE_SHIFT) - 1);
+		else
+			*processed_end = start;
 	}
-out:
-	spin_unlock(&tree->lock);
-	return found;
+	return err;
 }
 
-static int __process_pages_contig(struct address_space *mapping,
-				  struct page *locked_page,
-				  pgoff_t start_index, pgoff_t end_index,
-				  unsigned long page_ops, pgoff_t *index_ret);
-
 static noinline void __unlock_for_delalloc(struct inode *inode,
 					   struct page *locked_page,
 					   u64 start, u64 end)
@@ -1759,7 +346,7 @@ static noinline void __unlock_for_delalloc(struct inode *inode,
 	if (index == locked_page->index && end_index == index)
 		return;
 
-	__process_pages_contig(inode->i_mapping, locked_page, index, end_index,
+	__process_pages_contig(inode->i_mapping, locked_page, start, end,
 			       PAGE_UNLOCK, NULL);
 }
 
@@ -1769,36 +356,48 @@ static noinline int lock_delalloc_pages(struct inode *inode,
 					u64 delalloc_end)
 {
 	unsigned long index = delalloc_start >> PAGE_SHIFT;
-	unsigned long index_ret = index;
 	unsigned long end_index = delalloc_end >> PAGE_SHIFT;
+	u64 processed_end = delalloc_start;
 	int ret;
 
 	ASSERT(locked_page);
 	if (index == locked_page->index && index == end_index)
 		return 0;
 
-	ret = __process_pages_contig(inode->i_mapping, locked_page, index,
-				     end_index, PAGE_LOCK, &index_ret);
-	if (ret == -EAGAIN)
+	ret = __process_pages_contig(inode->i_mapping, locked_page, delalloc_start,
+				     delalloc_end, PAGE_LOCK, &processed_end);
+	if (ret == -EAGAIN && processed_end > delalloc_start)
 		__unlock_for_delalloc(inode, locked_page, delalloc_start,
-				      (u64)index_ret << PAGE_SHIFT);
+				      processed_end);
 	return ret;
 }
 
 /*
  * Find and lock a contiguous range of bytes in the file marked as delalloc, no
- * more than @max_bytes.  @Start and @end are used to return the range,
+ * more than @max_bytes.
+ *
+ * @start:	The original start bytenr to search.
+ *		Will store the extent range start bytenr.
+ * @end:	The original end bytenr of the search range
+ *		Will store the extent range end bytenr.
  *
- * Return: true if we find something
- *         false if nothing was in the tree
+ * Return true if we find a delalloc range which starts inside the original
+ * range, and @start/@end will store the delalloc range start/end.
+ *
+ * Return false if we can't find any delalloc range which starts inside the
+ * original range, and @start/@end will be the non-delalloc range start/end.
  */
 EXPORT_FOR_TESTS
 noinline_for_stack bool find_lock_delalloc_range(struct inode *inode,
 				    struct page *locked_page, u64 *start,
 				    u64 *end)
 {
+	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	struct extent_io_tree *tree = &BTRFS_I(inode)->io_tree;
-	u64 max_bytes = BTRFS_MAX_EXTENT_SIZE;
+	const u64 orig_start = *start;
+	const u64 orig_end = *end;
+	/* The sanity tests may not set a valid fs_info. */
+	u64 max_bytes = fs_info ? fs_info->max_extent_size : BTRFS_MAX_EXTENT_SIZE;
 	u64 delalloc_start;
 	u64 delalloc_end;
 	bool found;
@@ -1806,15 +405,23 @@ noinline_for_stack bool find_lock_delalloc_range(struct inode *inode,
 	int ret;
 	int loops = 0;
 
+	/* Caller should pass a valid @end to indicate the search range end */
+	ASSERT(orig_end > orig_start);
+
+	/* The range should at least cover part of the page */
+	ASSERT(!(orig_start >= page_offset(locked_page) + PAGE_SIZE ||
+		 orig_end <= page_offset(locked_page)));
 again:
 	/* step one, find a bunch of delalloc bytes starting at start */
 	delalloc_start = *start;
 	delalloc_end = 0;
 	found = btrfs_find_delalloc_range(tree, &delalloc_start, &delalloc_end,
 					  max_bytes, &cached_state);
-	if (!found || delalloc_end <= *start) {
+	if (!found || delalloc_end <= *start || delalloc_start > orig_end) {
 		*start = delalloc_start;
-		*end = delalloc_end;
+
+		/* @delalloc_end can be -1, never go beyond @orig_end */
+		*end = min(delalloc_end, orig_end);
 		free_extent_state(cached_state);
 		return false;
 	}
@@ -1854,14 +461,14 @@ again:
 	}
 
 	/* step three, lock the state bits for the whole range */
-	lock_extent_bits(tree, delalloc_start, delalloc_end, &cached_state);
+	lock_extent(tree, delalloc_start, delalloc_end, &cached_state);
 
 	/* then test to make sure it is all still delalloc */
 	ret = test_range_bit(tree, delalloc_start, delalloc_end,
 			     EXTENT_DELALLOC, 1, cached_state);
 	if (!ret) {
-		unlock_extent_cached(tree, delalloc_start, delalloc_end,
-				     &cached_state);
+		unlock_extent(tree, delalloc_start, delalloc_end,
+			      &cached_state);
 		__unlock_for_delalloc(inode, locked_page,
 			      delalloc_start, delalloc_end);
 		cond_resched();
@@ -1874,306 +481,50 @@ out_failed:
 	return found;
 }
 
-static int __process_pages_contig(struct address_space *mapping,
+void extent_clear_unlock_delalloc(struct btrfs_inode *inode, u64 start, u64 end,
 				  struct page *locked_page,
-				  pgoff_t start_index, pgoff_t end_index,
-				  unsigned long page_ops, pgoff_t *index_ret)
+				  u32 clear_bits, unsigned long page_ops)
 {
-	unsigned long nr_pages = end_index - start_index + 1;
-	unsigned long pages_locked = 0;
-	pgoff_t index = start_index;
-	struct page *pages[16];
-	unsigned ret;
-	int err = 0;
-	int i;
-
-	if (page_ops & PAGE_LOCK) {
-		ASSERT(page_ops == PAGE_LOCK);
-		ASSERT(index_ret && *index_ret == start_index);
-	}
-
-	if ((page_ops & PAGE_SET_ERROR) && nr_pages > 0)
-		mapping_set_error(mapping, -EIO);
-
-	while (nr_pages > 0) {
-		ret = find_get_pages_contig(mapping, index,
-				     min_t(unsigned long,
-				     nr_pages, ARRAY_SIZE(pages)), pages);
-		if (ret == 0) {
-			/*
-			 * Only if we're going to lock these pages,
-			 * can we find nothing at @index.
-			 */
-			ASSERT(page_ops & PAGE_LOCK);
-			err = -EAGAIN;
-			goto out;
-		}
-
-		for (i = 0; i < ret; i++) {
-			if (page_ops & PAGE_SET_PRIVATE2)
-				SetPagePrivate2(pages[i]);
+	clear_extent_bit(&inode->io_tree, start, end, clear_bits, NULL);
 
-			if (locked_page && pages[i] == locked_page) {
-				put_page(pages[i]);
-				pages_locked++;
-				continue;
-			}
-			if (page_ops & PAGE_CLEAR_DIRTY)
-				clear_page_dirty_for_io(pages[i]);
-			if (page_ops & PAGE_SET_WRITEBACK)
-				set_page_writeback(pages[i]);
-			if (page_ops & PAGE_SET_ERROR)
-				SetPageError(pages[i]);
-			if (page_ops & PAGE_END_WRITEBACK)
-				end_page_writeback(pages[i]);
-			if (page_ops & PAGE_UNLOCK)
-				unlock_page(pages[i]);
-			if (page_ops & PAGE_LOCK) {
-				lock_page(pages[i]);
-				if (!PageDirty(pages[i]) ||
-				    pages[i]->mapping != mapping) {
-					unlock_page(pages[i]);
-					put_page(pages[i]);
-					err = -EAGAIN;
-					goto out;
-				}
-			}
-			put_page(pages[i]);
-			pages_locked++;
-		}
-		nr_pages -= ret;
-		index += ret;
-		cond_resched();
-	}
-out:
-	if (err && index_ret)
-		*index_ret = start_index + pages_locked - 1;
-	return err;
+	__process_pages_contig(inode->vfs_inode.i_mapping, locked_page,
+			       start, end, page_ops, NULL);
 }
 
-void extent_clear_unlock_delalloc(struct inode *inode, u64 start, u64 end,
-				  struct page *locked_page,
-				  unsigned clear_bits,
-				  unsigned long page_ops)
+static int insert_failrec(struct btrfs_inode *inode,
+			  struct io_failure_record *failrec)
 {
-	clear_extent_bit(&BTRFS_I(inode)->io_tree, start, end, clear_bits, 1, 0,
-			 NULL);
-
-	__process_pages_contig(inode->i_mapping, locked_page,
-			       start >> PAGE_SHIFT, end >> PAGE_SHIFT,
-			       page_ops, NULL);
-}
-
-/*
- * count the number of bytes in the tree that have a given bit(s)
- * set.  This can be fairly slow, except for EXTENT_DIRTY which is
- * cached.  The total number found is returned.
- */
-u64 count_range_bits(struct extent_io_tree *tree,
-		     u64 *start, u64 search_end, u64 max_bytes,
-		     unsigned bits, int contig)
-{
-	struct rb_node *node;
-	struct extent_state *state;
-	u64 cur_start = *start;
-	u64 total_bytes = 0;
-	u64 last = 0;
-	int found = 0;
-
-	if (WARN_ON(search_end <= cur_start))
-		return 0;
-
-	spin_lock(&tree->lock);
-	if (cur_start == 0 && bits == EXTENT_DIRTY) {
-		total_bytes = tree->dirty_bytes;
-		goto out;
-	}
-	/*
-	 * this search will find all the extents that end after
-	 * our range starts.
-	 */
-	node = tree_search(tree, cur_start);
-	if (!node)
-		goto out;
-
-	while (1) {
-		state = rb_entry(node, struct extent_state, rb_node);
-		if (state->start > search_end)
-			break;
-		if (contig && found && state->start > last + 1)
-			break;
-		if (state->end >= cur_start && (state->state & bits) == bits) {
-			total_bytes += min(search_end, state->end) + 1 -
-				       max(cur_start, state->start);
-			if (total_bytes >= max_bytes)
-				break;
-			if (!found) {
-				*start = max(cur_start, state->start);
-				found = 1;
-			}
-			last = state->end;
-		} else if (contig && found) {
-			break;
-		}
-		node = rb_next(node);
-		if (!node)
-			break;
-	}
-out:
-	spin_unlock(&tree->lock);
-	return total_bytes;
-}
-
-/*
- * set the private field for a given byte offset in the tree.  If there isn't
- * an extent_state there already, this does nothing.
- */
-int set_state_failrec(struct extent_io_tree *tree, u64 start,
-		      struct io_failure_record *failrec)
-{
-	struct rb_node *node;
-	struct extent_state *state;
-	int ret = 0;
-
-	spin_lock(&tree->lock);
-	/*
-	 * this search will find all the extents that end after
-	 * our range starts.
-	 */
-	node = tree_search(tree, start);
-	if (!node) {
-		ret = -ENOENT;
-		goto out;
-	}
-	state = rb_entry(node, struct extent_state, rb_node);
-	if (state->start != start) {
-		ret = -ENOENT;
-		goto out;
-	}
-	state->failrec = failrec;
-out:
-	spin_unlock(&tree->lock);
-	return ret;
-}
+	struct rb_node *exist;
 
-int get_state_failrec(struct extent_io_tree *tree, u64 start,
-		      struct io_failure_record **failrec)
-{
-	struct rb_node *node;
-	struct extent_state *state;
-	int ret = 0;
+	spin_lock(&inode->io_failure_lock);
+	exist = rb_simple_insert(&inode->io_failure_tree, failrec->bytenr,
+				 &failrec->rb_node);
+	spin_unlock(&inode->io_failure_lock);
 
-	spin_lock(&tree->lock);
-	/*
-	 * this search will find all the extents that end after
-	 * our range starts.
-	 */
-	node = tree_search(tree, start);
-	if (!node) {
-		ret = -ENOENT;
-		goto out;
-	}
-	state = rb_entry(node, struct extent_state, rb_node);
-	if (state->start != start) {
-		ret = -ENOENT;
-		goto out;
-	}
-	*failrec = state->failrec;
-out:
-	spin_unlock(&tree->lock);
-	return ret;
+	return (exist == NULL) ? 0 : -EEXIST;
 }
 
-/*
- * searches a range in the state tree for a given mask.
- * If 'filled' == 1, this returns 1 only if every extent in the tree
- * has the bits set.  Otherwise, 1 is returned if any bit in the
- * range is found set.
- */
-int test_range_bit(struct extent_io_tree *tree, u64 start, u64 end,
-		   unsigned bits, int filled, struct extent_state *cached)
+static struct io_failure_record *get_failrec(struct btrfs_inode *inode, u64 start)
 {
-	struct extent_state *state = NULL;
 	struct rb_node *node;
-	int bitset = 0;
-
-	spin_lock(&tree->lock);
-	if (cached && extent_state_in_tree(cached) && cached->start <= start &&
-	    cached->end > start)
-		node = &cached->rb_node;
-	else
-		node = tree_search(tree, start);
-	while (node && start <= end) {
-		state = rb_entry(node, struct extent_state, rb_node);
+	struct io_failure_record *failrec = ERR_PTR(-ENOENT);
 
-		if (filled && state->start > start) {
-			bitset = 0;
-			break;
-		}
-
-		if (state->start > end)
-			break;
-
-		if (state->state & bits) {
-			bitset = 1;
-			if (!filled)
-				break;
-		} else if (filled) {
-			bitset = 0;
-			break;
-		}
-
-		if (state->end == (u64)-1)
-			break;
-
-		start = state->end + 1;
-		if (start > end)
-			break;
-		node = rb_next(node);
-		if (!node) {
-			if (filled)
-				bitset = 0;
-			break;
-		}
-	}
-	spin_unlock(&tree->lock);
-	return bitset;
+	spin_lock(&inode->io_failure_lock);
+	node = rb_simple_search(&inode->io_failure_tree, start);
+	if (node)
+		failrec = rb_entry(node, struct io_failure_record, rb_node);
+	spin_unlock(&inode->io_failure_lock);
+	return failrec;
 }
 
-/*
- * helper function to set a given page up to date if all the
- * extents in the tree for that page are up to date
- */
-static void check_page_uptodate(struct extent_io_tree *tree, struct page *page)
+static void free_io_failure(struct btrfs_inode *inode,
+			    struct io_failure_record *rec)
 {
-	u64 start = page_offset(page);
-	u64 end = start + PAGE_SIZE - 1;
-	if (test_range_bit(tree, start, end, EXTENT_UPTODATE, 1, NULL))
-		SetPageUptodate(page);
-}
-
-int free_io_failure(struct extent_io_tree *failure_tree,
-		    struct extent_io_tree *io_tree,
-		    struct io_failure_record *rec)
-{
-	int ret;
-	int err = 0;
-
-	set_state_failrec(failure_tree, rec->start, NULL);
-	ret = clear_extent_bits(failure_tree, rec->start,
-				rec->start + rec->len - 1,
-				EXTENT_LOCKED | EXTENT_DIRTY);
-	if (ret)
-		err = ret;
-
-	ret = clear_extent_bits(io_tree, rec->start,
-				rec->start + rec->len - 1,
-				EXTENT_DAMAGED);
-	if (ret && !err)
-		err = ret;
+	spin_lock(&inode->io_failure_lock);
+	rb_erase(&rec->rb_node, &inode->io_failure_tree);
+	spin_unlock(&inode->io_failure_lock);
 
 	kfree(rec);
-	return err;
 }
 
 /*
@@ -2186,26 +537,28 @@ int free_io_failure(struct extent_io_tree *failure_tree,
  * currently, there can be no more than two copies of every data bit. thus,
  * exactly one rewrite is required.
  */
-int repair_io_failure(struct btrfs_fs_info *fs_info, u64 ino, u64 start,
-		      u64 length, u64 logical, struct page *page,
-		      unsigned int pg_offset, int mirror_num)
+static int repair_io_failure(struct btrfs_fs_info *fs_info, u64 ino, u64 start,
+			     u64 length, u64 logical, struct page *page,
+			     unsigned int pg_offset, int mirror_num)
 {
-	struct bio *bio;
 	struct btrfs_device *dev;
+	struct bio_vec bvec;
+	struct bio bio;
 	u64 map_length = 0;
 	u64 sector;
-	struct btrfs_bio *bbio = NULL;
-	int ret;
+	struct btrfs_io_context *bioc = NULL;
+	int ret = 0;
 
 	ASSERT(!(fs_info->sb->s_flags & SB_RDONLY));
 	BUG_ON(!mirror_num);
 
-	bio = btrfs_io_bio_alloc(1);
-	bio->bi_iter.bi_size = 0;
+	if (btrfs_repair_one_zone(fs_info, logical))
+		return 0;
+
 	map_length = length;
 
 	/*
-	 * Avoid races with device replace and make sure our bbio has devices
+	 * Avoid races with device replace and make sure our bioc has devices
 	 * associated to its stripes that don't go away while we are doing the
 	 * read repair operation.
 	 */
@@ -2218,56 +571,54 @@ int repair_io_failure(struct btrfs_fs_info *fs_info, u64 ino, u64 start,
 		 * stripe's dev and sector.
 		 */
 		ret = btrfs_map_block(fs_info, BTRFS_MAP_READ, logical,
-				      &map_length, &bbio, 0);
-		if (ret) {
-			btrfs_bio_counter_dec(fs_info);
-			bio_put(bio);
-			return -EIO;
-		}
-		ASSERT(bbio->mirror_num == 1);
+				      &map_length, &bioc, 0);
+		if (ret)
+			goto out_counter_dec;
+		ASSERT(bioc->mirror_num == 1);
 	} else {
 		ret = btrfs_map_block(fs_info, BTRFS_MAP_WRITE, logical,
-				      &map_length, &bbio, mirror_num);
-		if (ret) {
-			btrfs_bio_counter_dec(fs_info);
-			bio_put(bio);
-			return -EIO;
-		}
-		BUG_ON(mirror_num != bbio->mirror_num);
+				      &map_length, &bioc, mirror_num);
+		if (ret)
+			goto out_counter_dec;
+		BUG_ON(mirror_num != bioc->mirror_num);
 	}
 
-	sector = bbio->stripes[bbio->mirror_num - 1].physical >> 9;
-	bio->bi_iter.bi_sector = sector;
-	dev = bbio->stripes[bbio->mirror_num - 1].dev;
-	btrfs_put_bbio(bbio);
+	sector = bioc->stripes[bioc->mirror_num - 1].physical >> 9;
+	dev = bioc->stripes[bioc->mirror_num - 1].dev;
+	btrfs_put_bioc(bioc);
+
 	if (!dev || !dev->bdev ||
 	    !test_bit(BTRFS_DEV_STATE_WRITEABLE, &dev->dev_state)) {
-		btrfs_bio_counter_dec(fs_info);
-		bio_put(bio);
-		return -EIO;
+		ret = -EIO;
+		goto out_counter_dec;
 	}
-	bio_set_dev(bio, dev->bdev);
-	bio->bi_opf = REQ_OP_WRITE | REQ_SYNC;
-	bio_add_page(bio, page, length, pg_offset);
 
-	if (btrfsic_submit_bio_wait(bio)) {
+	bio_init(&bio, dev->bdev, &bvec, 1, REQ_OP_WRITE | REQ_SYNC);
+	bio.bi_iter.bi_sector = sector;
+	__bio_add_page(&bio, page, length, pg_offset);
+
+	btrfsic_check_bio(&bio);
+	ret = submit_bio_wait(&bio);
+	if (ret) {
 		/* try to remap that extent elsewhere? */
-		btrfs_bio_counter_dec(fs_info);
-		bio_put(bio);
 		btrfs_dev_stat_inc_and_print(dev, BTRFS_DEV_STAT_WRITE_ERRS);
-		return -EIO;
+		goto out_bio_uninit;
 	}
 
 	btrfs_info_rl_in_rcu(fs_info,
 		"read error corrected: ino %llu off %llu (dev %s sector %llu)",
 				  ino, start,
 				  rcu_str_deref(dev->name), sector);
+	ret = 0;
+
+out_bio_uninit:
+	bio_uninit(&bio);
+out_counter_dec:
 	btrfs_bio_counter_dec(fs_info);
-	bio_put(bio);
-	return 0;
+	return ret;
 }
 
-int btrfs_repair_eb_io_failure(struct extent_buffer *eb, int mirror_num)
+int btrfs_repair_eb_io_failure(const struct extent_buffer *eb, int mirror_num)
 {
 	struct btrfs_fs_info *fs_info = eb->fs_info;
 	u64 start = eb->start;
@@ -2290,63 +641,59 @@ int btrfs_repair_eb_io_failure(struct extent_buffer *eb, int mirror_num)
 	return ret;
 }
 
+static int next_mirror(const struct io_failure_record *failrec, int cur_mirror)
+{
+	if (cur_mirror == failrec->num_copies)
+		return cur_mirror + 1 - failrec->num_copies;
+	return cur_mirror + 1;
+}
+
+static int prev_mirror(const struct io_failure_record *failrec, int cur_mirror)
+{
+	if (cur_mirror == 1)
+		return failrec->num_copies;
+	return cur_mirror - 1;
+}
+
 /*
  * each time an IO finishes, we do a fast check in the IO failure tree
  * to see if we need to process or clean up an io_failure_record
  */
-int clean_io_failure(struct btrfs_fs_info *fs_info,
-		     struct extent_io_tree *failure_tree,
-		     struct extent_io_tree *io_tree, u64 start,
-		     struct page *page, u64 ino, unsigned int pg_offset)
+int btrfs_clean_io_failure(struct btrfs_inode *inode, u64 start,
+			   struct page *page, unsigned int pg_offset)
 {
-	u64 private;
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
+	struct extent_io_tree *io_tree = &inode->io_tree;
+	u64 ino = btrfs_ino(inode);
+	u64 locked_start, locked_end;
 	struct io_failure_record *failrec;
-	struct extent_state *state;
-	int num_copies;
+	int mirror;
 	int ret;
 
-	private = 0;
-	ret = count_range_bits(failure_tree, &private, (u64)-1, 1,
-			       EXTENT_DIRTY, 0);
-	if (!ret)
-		return 0;
-
-	ret = get_state_failrec(failure_tree, start, &failrec);
-	if (ret)
+	failrec = get_failrec(inode, start);
+	if (IS_ERR(failrec))
 		return 0;
 
 	BUG_ON(!failrec->this_mirror);
 
-	if (failrec->in_validation) {
-		/* there was no real error, just free the record */
-		btrfs_debug(fs_info,
-			"clean_io_failure: freeing dummy error at %llu",
-			failrec->start);
-		goto out;
-	}
 	if (sb_rdonly(fs_info->sb))
 		goto out;
 
-	spin_lock(&io_tree->lock);
-	state = find_first_extent_bit_state(io_tree,
-					    failrec->start,
-					    EXTENT_LOCKED);
-	spin_unlock(&io_tree->lock);
-
-	if (state && state->start <= failrec->start &&
-	    state->end >= failrec->start + failrec->len - 1) {
-		num_copies = btrfs_num_copies(fs_info, failrec->logical,
-					      failrec->len);
-		if (num_copies > 1)  {
-			repair_io_failure(fs_info, ino, start, failrec->len,
-					  failrec->logical, page, pg_offset,
-					  failrec->failed_mirror);
-		}
-	}
+	ret = find_first_extent_bit(io_tree, failrec->bytenr, &locked_start,
+				    &locked_end, EXTENT_LOCKED, NULL);
+	if (ret || locked_start > failrec->bytenr ||
+	    locked_end < failrec->bytenr + failrec->len - 1)
+		goto out;
 
-out:
-	free_io_failure(failure_tree, io_tree, failrec);
+	mirror = failrec->this_mirror;
+	do {
+		mirror = prev_mirror(failrec, mirror);
+		repair_io_failure(fs_info, ino, start, failrec->len,
+				  failrec->logical, page, pg_offset, mirror);
+	} while (mirror != failrec->failed_mirror);
 
+out:
+	free_io_failure(inode, failrec);
 	return 0;
 }
 
@@ -2358,284 +705,288 @@ out:
  */
 void btrfs_free_io_failure_record(struct btrfs_inode *inode, u64 start, u64 end)
 {
-	struct extent_io_tree *failure_tree = &inode->io_failure_tree;
 	struct io_failure_record *failrec;
-	struct extent_state *state, *next;
+	struct rb_node *node, *next;
 
-	if (RB_EMPTY_ROOT(&failure_tree->state))
+	if (RB_EMPTY_ROOT(&inode->io_failure_tree))
 		return;
 
-	spin_lock(&failure_tree->lock);
-	state = find_first_extent_bit_state(failure_tree, start, EXTENT_DIRTY);
-	while (state) {
-		if (state->start > end)
+	spin_lock(&inode->io_failure_lock);
+	node = rb_simple_search_first(&inode->io_failure_tree, start);
+	while (node) {
+		failrec = rb_entry(node, struct io_failure_record, rb_node);
+		if (failrec->bytenr > end)
 			break;
 
-		ASSERT(state->end <= end);
-
-		next = next_state(state);
-
-		failrec = state->failrec;
-		free_extent_state(state);
+		next = rb_next(node);
+		rb_erase(&failrec->rb_node, &inode->io_failure_tree);
 		kfree(failrec);
 
-		state = next;
+		node = next;
 	}
-	spin_unlock(&failure_tree->lock);
+	spin_unlock(&inode->io_failure_lock);
 }
 
-int btrfs_get_io_failure_record(struct inode *inode, u64 start, u64 end,
-		struct io_failure_record **failrec_ret)
+static struct io_failure_record *btrfs_get_io_failure_record(struct inode *inode,
+							     struct btrfs_bio *bbio,
+							     unsigned int bio_offset)
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+	u64 start = bbio->file_offset + bio_offset;
 	struct io_failure_record *failrec;
-	struct extent_map *em;
-	struct extent_io_tree *failure_tree = &BTRFS_I(inode)->io_failure_tree;
-	struct extent_io_tree *tree = &BTRFS_I(inode)->io_tree;
-	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
+	const u32 sectorsize = fs_info->sectorsize;
 	int ret;
-	u64 logical;
-
-	ret = get_state_failrec(failure_tree, start, &failrec);
-	if (ret) {
-		failrec = kzalloc(sizeof(*failrec), GFP_NOFS);
-		if (!failrec)
-			return -ENOMEM;
-
-		failrec->start = start;
-		failrec->len = end - start + 1;
-		failrec->this_mirror = 0;
-		failrec->bio_flags = 0;
-		failrec->in_validation = 0;
-
-		read_lock(&em_tree->lock);
-		em = lookup_extent_mapping(em_tree, start, failrec->len);
-		if (!em) {
-			read_unlock(&em_tree->lock);
-			kfree(failrec);
-			return -EIO;
-		}
-
-		if (em->start > start || em->start + em->len <= start) {
-			free_extent_map(em);
-			em = NULL;
-		}
-		read_unlock(&em_tree->lock);
-		if (!em) {
-			kfree(failrec);
-			return -EIO;
-		}
-
-		logical = start - em->start;
-		logical = em->block_start + logical;
-		if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags)) {
-			logical = em->block_start;
-			failrec->bio_flags = EXTENT_BIO_COMPRESSED;
-			extent_set_compress_type(&failrec->bio_flags,
-						 em->compress_type);
-		}
 
+	failrec = get_failrec(BTRFS_I(inode), start);
+	if (!IS_ERR(failrec)) {
 		btrfs_debug(fs_info,
-			"Get IO Failure Record: (new) logical=%llu, start=%llu, len=%llu",
-			logical, start, failrec->len);
-
-		failrec->logical = logical;
-		free_extent_map(em);
-
-		/* set the bits in the private failure tree */
-		ret = set_extent_bits(failure_tree, start, end,
-					EXTENT_LOCKED | EXTENT_DIRTY);
-		if (ret >= 0)
-			ret = set_state_failrec(failure_tree, start, failrec);
-		/* set the bits in the inode's tree */
-		if (ret >= 0)
-			ret = set_extent_bits(tree, start, end, EXTENT_DAMAGED);
-		if (ret < 0) {
-			kfree(failrec);
-			return ret;
-		}
-	} else {
-		btrfs_debug(fs_info,
-			"Get IO Failure Record: (found) logical=%llu, start=%llu, len=%llu, validation=%d",
-			failrec->logical, failrec->start, failrec->len,
-			failrec->in_validation);
+	"Get IO Failure Record: (found) logical=%llu, start=%llu, len=%llu",
+			failrec->logical, failrec->bytenr, failrec->len);
 		/*
 		 * when data can be on disk more than twice, add to failrec here
 		 * (e.g. with a list for failed_mirror) to make
 		 * clean_io_failure() clean all those errors at once.
 		 */
+		ASSERT(failrec->this_mirror == bbio->mirror_num);
+		ASSERT(failrec->len == fs_info->sectorsize);
+		return failrec;
 	}
 
-	*failrec_ret = failrec;
+	failrec = kzalloc(sizeof(*failrec), GFP_NOFS);
+	if (!failrec)
+		return ERR_PTR(-ENOMEM);
 
-	return 0;
-}
+	RB_CLEAR_NODE(&failrec->rb_node);
+	failrec->bytenr = start;
+	failrec->len = sectorsize;
+	failrec->failed_mirror = bbio->mirror_num;
+	failrec->this_mirror = bbio->mirror_num;
+	failrec->logical = (bbio->iter.bi_sector << SECTOR_SHIFT) + bio_offset;
 
-bool btrfs_check_repairable(struct inode *inode, unsigned failed_bio_pages,
-			   struct io_failure_record *failrec, int failed_mirror)
-{
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	int num_copies;
+	btrfs_debug(fs_info,
+		    "new io failure record logical %llu start %llu",
+		    failrec->logical, start);
 
-	num_copies = btrfs_num_copies(fs_info, failrec->logical, failrec->len);
-	if (num_copies == 1) {
+	failrec->num_copies = btrfs_num_copies(fs_info, failrec->logical, sectorsize);
+	if (failrec->num_copies == 1) {
 		/*
-		 * we only have a single copy of the data, so don't bother with
-		 * all the retry and error correction code that follows. no
+		 * We only have a single copy of the data, so don't bother with
+		 * all the retry and error correction code that follows. No
 		 * matter what the error is, it is very likely to persist.
 		 */
 		btrfs_debug(fs_info,
-			"Check Repairable: cannot repair, num_copies=%d, next_mirror %d, failed_mirror %d",
-			num_copies, failrec->this_mirror, failed_mirror);
-		return false;
+			"cannot repair logical %llu num_copies %d",
+			failrec->logical, failrec->num_copies);
+		kfree(failrec);
+		return ERR_PTR(-EIO);
+	}
+
+	/* Set the bits in the private failure tree */
+	ret = insert_failrec(BTRFS_I(inode), failrec);
+	if (ret) {
+		kfree(failrec);
+		return ERR_PTR(ret);
 	}
 
+	return failrec;
+}
+
+int btrfs_repair_one_sector(struct inode *inode, struct btrfs_bio *failed_bbio,
+			    u32 bio_offset, struct page *page, unsigned int pgoff,
+			    submit_bio_hook_t *submit_bio_hook)
+{
+	u64 start = failed_bbio->file_offset + bio_offset;
+	struct io_failure_record *failrec;
+	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+	struct bio *failed_bio = &failed_bbio->bio;
+	const int icsum = bio_offset >> fs_info->sectorsize_bits;
+	struct bio *repair_bio;
+	struct btrfs_bio *repair_bbio;
+
+	btrfs_debug(fs_info,
+		   "repair read error: read error at %llu", start);
+
+	BUG_ON(bio_op(failed_bio) == REQ_OP_WRITE);
+
+	failrec = btrfs_get_io_failure_record(inode, failed_bbio, bio_offset);
+	if (IS_ERR(failrec))
+		return PTR_ERR(failrec);
+
 	/*
-	 * there are two premises:
-	 *	a) deliver good data to the caller
-	 *	b) correct the bad sectors on disk
+	 * There are two premises:
+	 * a) deliver good data to the caller
+	 * b) correct the bad sectors on disk
+	 *
+	 * Since we're only doing repair for one sector, we only need to get
+	 * a good copy of the failed sector and if we succeed, we have setup
+	 * everything for repair_io_failure to do the rest for us.
 	 */
-	if (failed_bio_pages > 1) {
-		/*
-		 * to fulfill b), we need to know the exact failing sectors, as
-		 * we don't want to rewrite any more than the failed ones. thus,
-		 * we need separate read requests for the failed bio
-		 *
-		 * if the following BUG_ON triggers, our validation request got
-		 * merged. we need separate requests for our algorithm to work.
-		 */
-		BUG_ON(failrec->in_validation);
-		failrec->in_validation = 1;
-		failrec->this_mirror = failed_mirror;
-	} else {
-		/*
-		 * we're ready to fulfill a) and b) alongside. get a good copy
-		 * of the failed sector and if we succeed, we have setup
-		 * everything for repair_io_failure to do the rest for us.
-		 */
-		if (failrec->in_validation) {
-			BUG_ON(failrec->this_mirror != failed_mirror);
-			failrec->in_validation = 0;
-			failrec->this_mirror = 0;
-		}
-		failrec->failed_mirror = failed_mirror;
-		failrec->this_mirror++;
-		if (failrec->this_mirror == failed_mirror)
-			failrec->this_mirror++;
+	failrec->this_mirror = next_mirror(failrec, failrec->this_mirror);
+	if (failrec->this_mirror == failrec->failed_mirror) {
+		btrfs_debug(fs_info,
+			"failed to repair num_copies %d this_mirror %d failed_mirror %d",
+			failrec->num_copies, failrec->this_mirror, failrec->failed_mirror);
+		free_io_failure(BTRFS_I(inode), failrec);
+		return -EIO;
 	}
 
-	if (failrec->this_mirror > num_copies) {
-		btrfs_debug(fs_info,
-			"Check Repairable: (fail) num_copies=%d, next_mirror %d, failed_mirror %d",
-			num_copies, failrec->this_mirror, failed_mirror);
-		return false;
+	repair_bio = btrfs_bio_alloc(1, REQ_OP_READ, failed_bbio->end_io,
+				     failed_bbio->private);
+	repair_bbio = btrfs_bio(repair_bio);
+	repair_bbio->file_offset = start;
+	repair_bio->bi_iter.bi_sector = failrec->logical >> 9;
+
+	if (failed_bbio->csum) {
+		const u32 csum_size = fs_info->csum_size;
+
+		repair_bbio->csum = repair_bbio->csum_inline;
+		memcpy(repair_bbio->csum,
+		       failed_bbio->csum + csum_size * icsum, csum_size);
 	}
 
-	return true;
-}
+	bio_add_page(repair_bio, page, failrec->len, pgoff);
+	repair_bbio->iter = repair_bio->bi_iter;
+
+	btrfs_debug(btrfs_sb(inode->i_sb),
+		    "repair read error: submitting new read to mirror %d",
+		    failrec->this_mirror);
 
+	/*
+	 * At this point we have a bio, so any errors from submit_bio_hook()
+	 * will be handled by the endio on the repair_bio, so we can't return an
+	 * error here.
+	 */
+	submit_bio_hook(inode, repair_bio, failrec->this_mirror, 0);
+	return BLK_STS_OK;
+}
 
-struct bio *btrfs_create_repair_bio(struct inode *inode, struct bio *failed_bio,
-				    struct io_failure_record *failrec,
-				    struct page *page, int pg_offset, int icsum,
-				    bio_end_io_t *endio_func, void *data)
+static void end_page_read(struct page *page, bool uptodate, u64 start, u32 len)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	struct bio *bio;
-	struct btrfs_io_bio *btrfs_failed_bio;
-	struct btrfs_io_bio *btrfs_bio;
-
-	bio = btrfs_io_bio_alloc(1);
-	bio->bi_end_io = endio_func;
-	bio->bi_iter.bi_sector = failrec->logical >> 9;
-	bio->bi_iter.bi_size = 0;
-	bio->bi_private = data;
+	struct btrfs_fs_info *fs_info = btrfs_sb(page->mapping->host->i_sb);
 
-	btrfs_failed_bio = btrfs_io_bio(failed_bio);
-	if (btrfs_failed_bio->csum) {
-		u16 csum_size = btrfs_super_csum_size(fs_info->super_copy);
+	ASSERT(page_offset(page) <= start &&
+	       start + len <= page_offset(page) + PAGE_SIZE);
 
-		btrfs_bio = btrfs_io_bio(bio);
-		btrfs_bio->csum = btrfs_bio->csum_inline;
-		icsum *= csum_size;
-		memcpy(btrfs_bio->csum, btrfs_failed_bio->csum + icsum,
-		       csum_size);
+	if (uptodate) {
+		if (fsverity_active(page->mapping->host) &&
+		    !PageError(page) &&
+		    !PageUptodate(page) &&
+		    start < i_size_read(page->mapping->host) &&
+		    !fsverity_verify_page(page)) {
+			btrfs_page_set_error(fs_info, page, start, len);
+		} else {
+			btrfs_page_set_uptodate(fs_info, page, start, len);
+		}
+	} else {
+		btrfs_page_clear_uptodate(fs_info, page, start, len);
+		btrfs_page_set_error(fs_info, page, start, len);
 	}
 
-	bio_add_page(bio, page, failrec->len, pg_offset);
+	if (!btrfs_is_subpage(fs_info, page))
+		unlock_page(page);
+	else
+		btrfs_subpage_end_reader(fs_info, page, start, len);
+}
+
+static void end_sector_io(struct page *page, u64 offset, bool uptodate)
+{
+	struct btrfs_inode *inode = BTRFS_I(page->mapping->host);
+	const u32 sectorsize = inode->root->fs_info->sectorsize;
+	struct extent_state *cached = NULL;
 
-	return bio;
+	end_page_read(page, uptodate, offset, sectorsize);
+	if (uptodate)
+		set_extent_uptodate(&inode->io_tree, offset,
+				    offset + sectorsize - 1, &cached, GFP_ATOMIC);
+	unlock_extent_atomic(&inode->io_tree, offset, offset + sectorsize - 1,
+			     &cached);
 }
 
-/*
- * This is a generic handler for readpage errors. If other copies exist, read
- * those and write back good data to the failed position. Does not investigate
- * in remapping the failed extent elsewhere, hoping the device will be smart
- * enough to do this as needed
- */
-static int bio_readpage_error(struct bio *failed_bio, u64 phy_offset,
-			      struct page *page, u64 start, u64 end,
-			      int failed_mirror)
+static void submit_data_read_repair(struct inode *inode,
+				    struct btrfs_bio *failed_bbio,
+				    u32 bio_offset, const struct bio_vec *bvec,
+				    unsigned int error_bitmap)
 {
-	struct io_failure_record *failrec;
-	struct inode *inode = page->mapping->host;
-	struct extent_io_tree *tree = &BTRFS_I(inode)->io_tree;
-	struct extent_io_tree *failure_tree = &BTRFS_I(inode)->io_failure_tree;
-	struct bio *bio;
-	int read_mode = 0;
-	blk_status_t status;
-	int ret;
-	unsigned failed_bio_pages = failed_bio->bi_iter.bi_size >> PAGE_SHIFT;
+	const unsigned int pgoff = bvec->bv_offset;
+	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+	struct page *page = bvec->bv_page;
+	const u64 start = page_offset(bvec->bv_page) + bvec->bv_offset;
+	const u64 end = start + bvec->bv_len - 1;
+	const u32 sectorsize = fs_info->sectorsize;
+	const int nr_bits = (end + 1 - start) >> fs_info->sectorsize_bits;
+	int i;
 
-	BUG_ON(bio_op(failed_bio) == REQ_OP_WRITE);
+	BUG_ON(bio_op(&failed_bbio->bio) == REQ_OP_WRITE);
 
-	ret = btrfs_get_io_failure_record(inode, start, end, &failrec);
-	if (ret)
-		return ret;
+	/* This repair is only for data */
+	ASSERT(is_data_inode(inode));
 
-	if (!btrfs_check_repairable(inode, failed_bio_pages, failrec,
-				    failed_mirror)) {
-		free_io_failure(failure_tree, tree, failrec);
-		return -EIO;
-	}
+	/* We're here because we had some read errors or csum mismatch */
+	ASSERT(error_bitmap);
 
-	if (failed_bio_pages > 1)
-		read_mode |= REQ_FAILFAST_DEV;
+	/*
+	 * We only get called on buffered IO, thus page must be mapped and bio
+	 * must not be cloned.
+	 */
+	ASSERT(page->mapping && !bio_flagged(&failed_bbio->bio, BIO_CLONED));
 
-	phy_offset >>= inode->i_sb->s_blocksize_bits;
-	bio = btrfs_create_repair_bio(inode, failed_bio, failrec, page,
-				      start - page_offset(page),
-				      (int)phy_offset, failed_bio->bi_end_io,
-				      NULL);
-	bio->bi_opf = REQ_OP_READ | read_mode;
+	/* Iterate through all the sectors in the range */
+	for (i = 0; i < nr_bits; i++) {
+		const unsigned int offset = i * sectorsize;
+		bool uptodate = false;
+		int ret;
 
-	btrfs_debug(btrfs_sb(inode->i_sb),
-		"Repair Read Error: submitting new read[%#x] to this_mirror=%d, in_validation=%d",
-		read_mode, failrec->this_mirror, failrec->in_validation);
+		if (!(error_bitmap & (1U << i))) {
+			/*
+			 * This sector has no error, just end the page read
+			 * and unlock the range.
+			 */
+			uptodate = true;
+			goto next;
+		}
 
-	status = tree->ops->submit_bio_hook(tree->private_data, bio, failrec->this_mirror,
-					 failrec->bio_flags);
-	if (status) {
-		free_io_failure(failure_tree, tree, failrec);
-		bio_put(bio);
-		ret = blk_status_to_errno(status);
+		ret = btrfs_repair_one_sector(inode, failed_bbio,
+				bio_offset + offset, page, pgoff + offset,
+				btrfs_submit_data_read_bio);
+		if (!ret) {
+			/*
+			 * We have submitted the read repair, the page release
+			 * will be handled by the endio function of the
+			 * submitted repair bio.
+			 * Thus we don't need to do any thing here.
+			 */
+			continue;
+		}
+		/*
+		 * Continue on failed repair, otherwise the remaining sectors
+		 * will not be properly unlocked.
+		 */
+next:
+		end_sector_io(page, start + offset, uptodate);
 	}
-
-	return ret;
 }
 
 /* lots and lots of room for performance fixes in the end_bio funcs */
 
 void end_extent_writepage(struct page *page, int err, u64 start, u64 end)
 {
-	int uptodate = (err == 0);
+	struct btrfs_inode *inode;
+	const bool uptodate = (err == 0);
 	int ret = 0;
 
-	btrfs_writepage_endio_finish_ordered(page, start, end, uptodate);
+	ASSERT(page && page->mapping);
+	inode = BTRFS_I(page->mapping->host);
+	btrfs_writepage_endio_finish_ordered(inode, page, start, end, uptodate);
 
 	if (!uptodate) {
-		ClearPageUptodate(page);
-		SetPageError(page);
+		const struct btrfs_fs_info *fs_info = inode->root->fs_info;
+		u32 len;
+
+		ASSERT(end + 1 - start <= U32_MAX);
+		len = end + 1 - start;
+
+		btrfs_page_clear_uptodate(fs_info, page, start, len);
+		btrfs_page_set_error(fs_info, page, start, len);
 		ret = err < 0 ? err : -EIO;
 		mapping_set_error(page->mapping, ret);
 	}
@@ -2650,56 +1001,155 @@ void end_extent_writepage(struct page *page, int err, u64 start, u64 end)
  * Scheduling is not allowed, so the extent state tree is expected
  * to have one and only one object corresponding to this IO.
  */
-static void end_bio_extent_writepage(struct bio *bio)
+static void end_bio_extent_writepage(struct btrfs_bio *bbio)
 {
+	struct bio *bio = &bbio->bio;
 	int error = blk_status_to_errno(bio->bi_status);
 	struct bio_vec *bvec;
 	u64 start;
 	u64 end;
 	struct bvec_iter_all iter_all;
+	bool first_bvec = true;
 
 	ASSERT(!bio_flagged(bio, BIO_CLONED));
 	bio_for_each_segment_all(bvec, bio, iter_all) {
 		struct page *page = bvec->bv_page;
 		struct inode *inode = page->mapping->host;
 		struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-
-		/* We always issue full-page reads, but if some block
-		 * in a page fails to read, blk_update_request() will
-		 * advance bv_offset and adjust bv_len to compensate.
-		 * Print a warning for nonzero offsets, and an error
-		 * if they don't add up to a full page.  */
-		if (bvec->bv_offset || bvec->bv_len != PAGE_SIZE) {
-			if (bvec->bv_offset + bvec->bv_len != PAGE_SIZE)
-				btrfs_err(fs_info,
-				   "partial page write in btrfs with offset %u and length %u",
-					bvec->bv_offset, bvec->bv_len);
-			else
-				btrfs_info(fs_info,
-				   "incomplete page write in btrfs with offset %u and length %u",
-					bvec->bv_offset, bvec->bv_len);
+		const u32 sectorsize = fs_info->sectorsize;
+
+		/* Our read/write should always be sector aligned. */
+		if (!IS_ALIGNED(bvec->bv_offset, sectorsize))
+			btrfs_err(fs_info,
+		"partial page write in btrfs with offset %u and length %u",
+				  bvec->bv_offset, bvec->bv_len);
+		else if (!IS_ALIGNED(bvec->bv_len, sectorsize))
+			btrfs_info(fs_info,
+		"incomplete page write with offset %u and length %u",
+				   bvec->bv_offset, bvec->bv_len);
+
+		start = page_offset(page) + bvec->bv_offset;
+		end = start + bvec->bv_len - 1;
+
+		if (first_bvec) {
+			btrfs_record_physical_zoned(inode, start, bio);
+			first_bvec = false;
 		}
 
-		start = page_offset(page);
-		end = start + bvec->bv_offset + bvec->bv_len - 1;
-
 		end_extent_writepage(page, error, start, end);
-		end_page_writeback(page);
+
+		btrfs_page_clear_writeback(fs_info, page, start, bvec->bv_len);
 	}
 
 	bio_put(bio);
 }
 
-static void
-endio_readpage_release_extent(struct extent_io_tree *tree, u64 start, u64 len,
-			      int uptodate)
+/*
+ * Record previously processed extent range
+ *
+ * For endio_readpage_release_extent() to handle a full extent range, reducing
+ * the extent io operations.
+ */
+struct processed_extent {
+	struct btrfs_inode *inode;
+	/* Start of the range in @inode */
+	u64 start;
+	/* End of the range in @inode */
+	u64 end;
+	bool uptodate;
+};
+
+/*
+ * Try to release processed extent range
+ *
+ * May not release the extent range right now if the current range is
+ * contiguous to processed extent.
+ *
+ * Will release processed extent when any of @inode, @uptodate, the range is
+ * no longer contiguous to the processed range.
+ *
+ * Passing @inode == NULL will force processed extent to be released.
+ */
+static void endio_readpage_release_extent(struct processed_extent *processed,
+			      struct btrfs_inode *inode, u64 start, u64 end,
+			      bool uptodate)
 {
 	struct extent_state *cached = NULL;
-	u64 end = start + len - 1;
+	struct extent_io_tree *tree;
 
-	if (uptodate && tree->track_uptodate)
-		set_extent_uptodate(tree, start, end, &cached, GFP_ATOMIC);
-	unlock_extent_cached_atomic(tree, start, end, &cached);
+	/* The first extent, initialize @processed */
+	if (!processed->inode)
+		goto update;
+
+	/*
+	 * Contiguous to processed extent, just uptodate the end.
+	 *
+	 * Several things to notice:
+	 *
+	 * - bio can be merged as long as on-disk bytenr is contiguous
+	 *   This means we can have page belonging to other inodes, thus need to
+	 *   check if the inode still matches.
+	 * - bvec can contain range beyond current page for multi-page bvec
+	 *   Thus we need to do processed->end + 1 >= start check
+	 */
+	if (processed->inode == inode && processed->uptodate == uptodate &&
+	    processed->end + 1 >= start && end >= processed->end) {
+		processed->end = end;
+		return;
+	}
+
+	tree = &processed->inode->io_tree;
+	/*
+	 * Now we don't have range contiguous to the processed range, release
+	 * the processed range now.
+	 */
+	unlock_extent_atomic(tree, processed->start, processed->end, &cached);
+
+update:
+	/* Update processed to current range */
+	processed->inode = inode;
+	processed->start = start;
+	processed->end = end;
+	processed->uptodate = uptodate;
+}
+
+static void begin_page_read(struct btrfs_fs_info *fs_info, struct page *page)
+{
+	ASSERT(PageLocked(page));
+	if (!btrfs_is_subpage(fs_info, page))
+		return;
+
+	ASSERT(PagePrivate(page));
+	btrfs_subpage_start_reader(fs_info, page, page_offset(page), PAGE_SIZE);
+}
+
+/*
+ * Find extent buffer for a givne bytenr.
+ *
+ * This is for end_bio_extent_readpage(), thus we can't do any unsafe locking
+ * in endio context.
+ */
+static struct extent_buffer *find_extent_buffer_readpage(
+		struct btrfs_fs_info *fs_info, struct page *page, u64 bytenr)
+{
+	struct extent_buffer *eb;
+
+	/*
+	 * For regular sectorsize, we can use page->private to grab extent
+	 * buffer
+	 */
+	if (fs_info->nodesize >= PAGE_SIZE) {
+		ASSERT(PagePrivate(page) && page->private);
+		return (struct extent_buffer *)page->private;
+	}
+
+	/* For subpage case, we need to lookup buffer radix tree */
+	rcu_read_lock();
+	eb = radix_tree_lookup(&fs_info->buffer_radix,
+			       bytenr >> fs_info->sectorsize_bits);
+	rcu_read_unlock();
+	ASSERT(eb);
+	return eb;
 }
 
 /*
@@ -2713,325 +1163,541 @@ endio_readpage_release_extent(struct extent_io_tree *tree, u64 start, u64 len,
  * Scheduling is not allowed, so the extent state tree is expected
  * to have one and only one object corresponding to this IO.
  */
-static void end_bio_extent_readpage(struct bio *bio)
+static void end_bio_extent_readpage(struct btrfs_bio *bbio)
 {
+	struct bio *bio = &bbio->bio;
 	struct bio_vec *bvec;
-	int uptodate = !bio->bi_status;
-	struct btrfs_io_bio *io_bio = btrfs_io_bio(bio);
-	struct extent_io_tree *tree, *failure_tree;
-	u64 offset = 0;
-	u64 start;
-	u64 end;
-	u64 len;
-	u64 extent_start = 0;
-	u64 extent_len = 0;
+	struct processed_extent processed = { 0 };
+	/*
+	 * The offset to the beginning of a bio, since one bio can never be
+	 * larger than UINT_MAX, u32 here is enough.
+	 */
+	u32 bio_offset = 0;
 	int mirror;
-	int ret;
 	struct bvec_iter_all iter_all;
 
 	ASSERT(!bio_flagged(bio, BIO_CLONED));
 	bio_for_each_segment_all(bvec, bio, iter_all) {
+		bool uptodate = !bio->bi_status;
 		struct page *page = bvec->bv_page;
 		struct inode *inode = page->mapping->host;
 		struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-		bool data_inode = btrfs_ino(BTRFS_I(inode))
-			!= BTRFS_BTREE_INODE_OBJECTID;
+		const u32 sectorsize = fs_info->sectorsize;
+		unsigned int error_bitmap = (unsigned int)-1;
+		bool repair = false;
+		u64 start;
+		u64 end;
+		u32 len;
 
 		btrfs_debug(fs_info,
 			"end_bio_extent_readpage: bi_sector=%llu, err=%d, mirror=%u",
-			(u64)bio->bi_iter.bi_sector, bio->bi_status,
-			io_bio->mirror_num);
-		tree = &BTRFS_I(inode)->io_tree;
-		failure_tree = &BTRFS_I(inode)->io_failure_tree;
-
-		/* We always issue full-page reads, but if some block
-		 * in a page fails to read, blk_update_request() will
-		 * advance bv_offset and adjust bv_len to compensate.
-		 * Print a warning for nonzero offsets, and an error
-		 * if they don't add up to a full page.  */
-		if (bvec->bv_offset || bvec->bv_len != PAGE_SIZE) {
-			if (bvec->bv_offset + bvec->bv_len != PAGE_SIZE)
-				btrfs_err(fs_info,
-					"partial page read in btrfs with offset %u and length %u",
-					bvec->bv_offset, bvec->bv_len);
-			else
-				btrfs_info(fs_info,
-					"incomplete page read in btrfs with offset %u and length %u",
-					bvec->bv_offset, bvec->bv_len);
-		}
+			bio->bi_iter.bi_sector, bio->bi_status,
+			bbio->mirror_num);
 
-		start = page_offset(page);
-		end = start + bvec->bv_offset + bvec->bv_len - 1;
+		/*
+		 * We always issue full-sector reads, but if some block in a
+		 * page fails to read, blk_update_request() will advance
+		 * bv_offset and adjust bv_len to compensate.  Print a warning
+		 * for unaligned offsets, and an error if they don't add up to
+		 * a full sector.
+		 */
+		if (!IS_ALIGNED(bvec->bv_offset, sectorsize))
+			btrfs_err(fs_info,
+		"partial page read in btrfs with offset %u and length %u",
+				  bvec->bv_offset, bvec->bv_len);
+		else if (!IS_ALIGNED(bvec->bv_offset + bvec->bv_len,
+				     sectorsize))
+			btrfs_info(fs_info,
+		"incomplete page read with offset %u and length %u",
+				   bvec->bv_offset, bvec->bv_len);
+
+		start = page_offset(page) + bvec->bv_offset;
+		end = start + bvec->bv_len - 1;
 		len = bvec->bv_len;
 
-		mirror = io_bio->mirror_num;
+		mirror = bbio->mirror_num;
 		if (likely(uptodate)) {
-			ret = tree->ops->readpage_end_io_hook(io_bio, offset,
-							      page, start, end,
-							      mirror);
-			if (ret)
-				uptodate = 0;
-			else
-				clean_io_failure(BTRFS_I(inode)->root->fs_info,
-						 failure_tree, tree, start,
-						 page,
-						 btrfs_ino(BTRFS_I(inode)), 0);
+			if (is_data_inode(inode)) {
+				error_bitmap = btrfs_verify_data_csum(bbio,
+						bio_offset, page, start, end);
+				if (error_bitmap)
+					uptodate = false;
+			} else {
+				if (btrfs_validate_metadata_buffer(bbio,
+						page, start, end, mirror))
+					uptodate = false;
+			}
 		}
 
-		if (likely(uptodate))
-			goto readpage_ok;
+		if (likely(uptodate)) {
+			loff_t i_size = i_size_read(inode);
+			pgoff_t end_index = i_size >> PAGE_SHIFT;
 
-		if (data_inode) {
+			btrfs_clean_io_failure(BTRFS_I(inode), start, page, 0);
 
 			/*
-			 * The generic bio_readpage_error handles errors the
-			 * following way: If possible, new read requests are
-			 * created and submitted and will end up in
-			 * end_bio_extent_readpage as well (if we're lucky,
-			 * not in the !uptodate case). In that case it returns
-			 * 0 and we just go on with the next page in our bio.
-			 * If it can't handle the error it will return -EIO and
-			 * we remain responsible for that page.
+			 * Zero out the remaining part if this range straddles
+			 * i_size.
+			 *
+			 * Here we should only zero the range inside the bvec,
+			 * not touch anything else.
+			 *
+			 * NOTE: i_size is exclusive while end is inclusive.
 			 */
-			ret = bio_readpage_error(bio, offset, page, start, end,
-						 mirror);
-			if (ret == 0) {
-				uptodate = !bio->bi_status;
-				offset += len;
-				continue;
+			if (page->index == end_index && i_size <= end) {
+				u32 zero_start = max(offset_in_page(i_size),
+						     offset_in_page(start));
+
+				zero_user_segment(page, zero_start,
+						  offset_in_page(end) + 1);
 			}
+		} else if (is_data_inode(inode)) {
+			/*
+			 * Only try to repair bios that actually made it to a
+			 * device.  If the bio failed to be submitted mirror
+			 * is 0 and we need to fail it without retrying.
+			 *
+			 * This also includes the high level bios for compressed
+			 * extents - these never make it to a device and repair
+			 * is already handled on the lower compressed bio.
+			 */
+			if (mirror > 0)
+				repair = true;
 		} else {
 			struct extent_buffer *eb;
 
-			eb = (struct extent_buffer *)page->private;
+			eb = find_extent_buffer_readpage(fs_info, page, start);
 			set_bit(EXTENT_BUFFER_READ_ERR, &eb->bflags);
 			eb->read_mirror = mirror;
 			atomic_dec(&eb->io_pages);
-			if (test_and_clear_bit(EXTENT_BUFFER_READAHEAD,
-					       &eb->bflags))
-				btree_readahead_hook(eb, -EIO);
 		}
-readpage_ok:
-		if (likely(uptodate)) {
-			loff_t i_size = i_size_read(inode);
-			pgoff_t end_index = i_size >> PAGE_SHIFT;
-			unsigned off;
 
-			/* Zero out the end if this page straddles i_size */
-			off = offset_in_page(i_size);
-			if (page->index == end_index && off)
-				zero_user_segment(page, off, PAGE_SIZE);
-			SetPageUptodate(page);
-		} else {
-			ClearPageUptodate(page);
-			SetPageError(page);
-		}
-		unlock_page(page);
-		offset += len;
-
-		if (unlikely(!uptodate)) {
-			if (extent_len) {
-				endio_readpage_release_extent(tree,
-							      extent_start,
-							      extent_len, 1);
-				extent_start = 0;
-				extent_len = 0;
-			}
-			endio_readpage_release_extent(tree, start,
-						      end - start + 1, 0);
-		} else if (!extent_len) {
-			extent_start = start;
-			extent_len = end + 1 - start;
-		} else if (extent_start + extent_len == start) {
-			extent_len += end + 1 - start;
+		if (repair) {
+			/*
+			 * submit_data_read_repair() will handle all the good
+			 * and bad sectors, we just continue to the next bvec.
+			 */
+			submit_data_read_repair(inode, bbio, bio_offset, bvec,
+						error_bitmap);
 		} else {
-			endio_readpage_release_extent(tree, extent_start,
-						      extent_len, uptodate);
-			extent_start = start;
-			extent_len = end + 1 - start;
+			/* Update page status and unlock */
+			end_page_read(page, uptodate, start, len);
+			endio_readpage_release_extent(&processed, BTRFS_I(inode),
+					start, end, PageUptodate(page));
 		}
-	}
 
-	if (extent_len)
-		endio_readpage_release_extent(tree, extent_start, extent_len,
-					      uptodate);
-	btrfs_io_bio_free_csum(io_bio);
+		ASSERT(bio_offset + len > bio_offset);
+		bio_offset += len;
+
+	}
+	/* Release the last extent */
+	endio_readpage_release_extent(&processed, NULL, 0, 0, false);
+	btrfs_bio_free_csum(bbio);
 	bio_put(bio);
 }
 
-/*
- * Initialize the members up to but not including 'bio'. Use after allocating a
- * new bio by bio_alloc_bioset as it does not initialize the bytes outside of
- * 'bio' because use of __GFP_ZERO is not supported.
+/**
+ * Populate every free slot in a provided array with pages.
+ *
+ * @nr_pages:   number of pages to allocate
+ * @page_array: the array to fill with pages; any existing non-null entries in
+ * 		the array will be skipped
+ *
+ * Return: 0        if all pages were able to be allocated;
+ *         -ENOMEM  otherwise, and the caller is responsible for freeing all
+ *                  non-null page pointers in the array.
  */
-static inline void btrfs_io_bio_init(struct btrfs_io_bio *btrfs_bio)
+int btrfs_alloc_page_array(unsigned int nr_pages, struct page **page_array)
 {
-	memset(btrfs_bio, 0, offsetof(struct btrfs_io_bio, bio));
+	unsigned int allocated;
+
+	for (allocated = 0; allocated < nr_pages;) {
+		unsigned int last = allocated;
+
+		allocated = alloc_pages_bulk_array(GFP_NOFS, nr_pages, page_array);
+
+		if (allocated == nr_pages)
+			return 0;
+
+		/*
+		 * During this iteration, no page could be allocated, even
+		 * though alloc_pages_bulk_array() falls back to alloc_page()
+		 * if  it could not bulk-allocate. So we must be out of memory.
+		 */
+		if (allocated == last)
+			return -ENOMEM;
+
+		memalloc_retry_wait(GFP_NOFS);
+	}
+	return 0;
 }
 
-/*
- * The following helpers allocate a bio. As it's backed by a bioset, it'll
- * never fail.  We're returning a bio right now but you can call btrfs_io_bio
- * for the appropriate container_of magic
+/**
+ * Attempt to add a page to bio
+ *
+ * @bio_ctrl:	record both the bio, and its bio_flags
+ * @page:	page to add to the bio
+ * @disk_bytenr:  offset of the new bio or to check whether we are adding
+ *                a contiguous page to the previous one
+ * @size:	portion of page that we want to write
+ * @pg_offset:	starting offset in the page
+ * @compress_type:   compression type of the current bio to see if we can merge them
+ *
+ * Attempt to add a page to bio considering stripe alignment etc.
+ *
+ * Return >= 0 for the number of bytes added to the bio.
+ * Can return 0 if the current bio is already at stripe/zone boundary.
+ * Return <0 for error.
  */
-struct bio *btrfs_bio_alloc(u64 first_byte)
-{
-	struct bio *bio;
+static int btrfs_bio_add_page(struct btrfs_bio_ctrl *bio_ctrl,
+			      struct page *page,
+			      u64 disk_bytenr, unsigned int size,
+			      unsigned int pg_offset,
+			      enum btrfs_compression_type compress_type)
+{
+	struct bio *bio = bio_ctrl->bio;
+	u32 bio_size = bio->bi_iter.bi_size;
+	u32 real_size;
+	const sector_t sector = disk_bytenr >> SECTOR_SHIFT;
+	bool contig = false;
+	int ret;
 
-	bio = bio_alloc_bioset(GFP_NOFS, BIO_MAX_PAGES, &btrfs_bioset);
-	bio->bi_iter.bi_sector = first_byte >> 9;
-	btrfs_io_bio_init(btrfs_io_bio(bio));
-	return bio;
-}
+	ASSERT(bio);
+	/* The limit should be calculated when bio_ctrl->bio is allocated */
+	ASSERT(bio_ctrl->len_to_oe_boundary && bio_ctrl->len_to_stripe_boundary);
+	if (bio_ctrl->compress_type != compress_type)
+		return 0;
 
-struct bio *btrfs_bio_clone(struct bio *bio)
-{
-	struct btrfs_io_bio *btrfs_bio;
-	struct bio *new;
 
-	/* Bio allocation backed by a bioset does not fail */
-	new = bio_clone_fast(bio, GFP_NOFS, &btrfs_bioset);
-	btrfs_bio = btrfs_io_bio(new);
-	btrfs_io_bio_init(btrfs_bio);
-	btrfs_bio->iter = bio->bi_iter;
-	return new;
+	if (bio->bi_iter.bi_size == 0) {
+		/* We can always add a page into an empty bio. */
+		contig = true;
+	} else if (bio_ctrl->compress_type == BTRFS_COMPRESS_NONE) {
+		struct bio_vec *bvec = bio_last_bvec_all(bio);
+
+		/*
+		 * The contig check requires the following conditions to be met:
+		 * 1) The pages are belonging to the same inode
+		 *    This is implied by the call chain.
+		 *
+		 * 2) The range has adjacent logical bytenr
+		 *
+		 * 3) The range has adjacent file offset
+		 *    This is required for the usage of btrfs_bio->file_offset.
+		 */
+		if (bio_end_sector(bio) == sector &&
+		    page_offset(bvec->bv_page) + bvec->bv_offset +
+		    bvec->bv_len == page_offset(page) + pg_offset)
+			contig = true;
+	} else {
+		/*
+		 * For compression, all IO should have its logical bytenr
+		 * set to the starting bytenr of the compressed extent.
+		 */
+		contig = bio->bi_iter.bi_sector == sector;
+	}
+
+	if (!contig)
+		return 0;
+
+	real_size = min(bio_ctrl->len_to_oe_boundary,
+			bio_ctrl->len_to_stripe_boundary) - bio_size;
+	real_size = min(real_size, size);
+
+	/*
+	 * If real_size is 0, never call bio_add_*_page(), as even size is 0,
+	 * bio will still execute its endio function on the page!
+	 */
+	if (real_size == 0)
+		return 0;
+
+	if (bio_op(bio) == REQ_OP_ZONE_APPEND)
+		ret = bio_add_zone_append_page(bio, page, real_size, pg_offset);
+	else
+		ret = bio_add_page(bio, page, real_size, pg_offset);
+
+	return ret;
 }
 
-struct bio *btrfs_io_bio_alloc(unsigned int nr_iovecs)
+static int calc_bio_boundaries(struct btrfs_bio_ctrl *bio_ctrl,
+			       struct btrfs_inode *inode, u64 file_offset)
 {
-	struct bio *bio;
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
+	struct btrfs_io_geometry geom;
+	struct btrfs_ordered_extent *ordered;
+	struct extent_map *em;
+	u64 logical = (bio_ctrl->bio->bi_iter.bi_sector << SECTOR_SHIFT);
+	int ret;
+
+	/*
+	 * Pages for compressed extent are never submitted to disk directly,
+	 * thus it has no real boundary, just set them to U32_MAX.
+	 *
+	 * The split happens for real compressed bio, which happens in
+	 * btrfs_submit_compressed_read/write().
+	 */
+	if (bio_ctrl->compress_type != BTRFS_COMPRESS_NONE) {
+		bio_ctrl->len_to_oe_boundary = U32_MAX;
+		bio_ctrl->len_to_stripe_boundary = U32_MAX;
+		return 0;
+	}
+	em = btrfs_get_chunk_map(fs_info, logical, fs_info->sectorsize);
+	if (IS_ERR(em))
+		return PTR_ERR(em);
+	ret = btrfs_get_io_geometry(fs_info, em, btrfs_op(bio_ctrl->bio),
+				    logical, &geom);
+	free_extent_map(em);
+	if (ret < 0) {
+		return ret;
+	}
+	if (geom.len > U32_MAX)
+		bio_ctrl->len_to_stripe_boundary = U32_MAX;
+	else
+		bio_ctrl->len_to_stripe_boundary = (u32)geom.len;
 
-	/* Bio allocation backed by a bioset does not fail */
-	bio = bio_alloc_bioset(GFP_NOFS, nr_iovecs, &btrfs_bioset);
-	btrfs_io_bio_init(btrfs_io_bio(bio));
-	return bio;
+	if (bio_op(bio_ctrl->bio) != REQ_OP_ZONE_APPEND) {
+		bio_ctrl->len_to_oe_boundary = U32_MAX;
+		return 0;
+	}
+
+	/* Ordered extent not yet created, so we're good */
+	ordered = btrfs_lookup_ordered_extent(inode, file_offset);
+	if (!ordered) {
+		bio_ctrl->len_to_oe_boundary = U32_MAX;
+		return 0;
+	}
+
+	bio_ctrl->len_to_oe_boundary = min_t(u32, U32_MAX,
+		ordered->disk_bytenr + ordered->disk_num_bytes - logical);
+	btrfs_put_ordered_extent(ordered);
+	return 0;
 }
 
-struct bio *btrfs_bio_clone_partial(struct bio *orig, int offset, int size)
+static int alloc_new_bio(struct btrfs_inode *inode,
+			 struct btrfs_bio_ctrl *bio_ctrl,
+			 struct writeback_control *wbc,
+			 blk_opf_t opf,
+			 u64 disk_bytenr, u32 offset, u64 file_offset,
+			 enum btrfs_compression_type compress_type)
 {
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
 	struct bio *bio;
-	struct btrfs_io_bio *btrfs_bio;
+	int ret;
 
-	/* this will never fail when it's backed by a bioset */
-	bio = bio_clone_fast(orig, GFP_NOFS, &btrfs_bioset);
-	ASSERT(bio);
+	ASSERT(bio_ctrl->end_io_func);
 
-	btrfs_bio = btrfs_io_bio(bio);
-	btrfs_io_bio_init(btrfs_bio);
+	bio = btrfs_bio_alloc(BIO_MAX_VECS, opf, bio_ctrl->end_io_func, NULL);
+	/*
+	 * For compressed page range, its disk_bytenr is always @disk_bytenr
+	 * passed in, no matter if we have added any range into previous bio.
+	 */
+	if (compress_type != BTRFS_COMPRESS_NONE)
+		bio->bi_iter.bi_sector = disk_bytenr >> SECTOR_SHIFT;
+	else
+		bio->bi_iter.bi_sector = (disk_bytenr + offset) >> SECTOR_SHIFT;
+	bio_ctrl->bio = bio;
+	bio_ctrl->compress_type = compress_type;
+	ret = calc_bio_boundaries(bio_ctrl, inode, file_offset);
+	if (ret < 0)
+		goto error;
 
-	bio_trim(bio, offset >> 9, size >> 9);
-	btrfs_bio->iter = bio->bi_iter;
-	return bio;
+	if (wbc) {
+		/*
+		 * For Zone append we need the correct block_device that we are
+		 * going to write to set in the bio to be able to respect the
+		 * hardware limitation.  Look it up here:
+		 */
+		if (bio_op(bio) == REQ_OP_ZONE_APPEND) {
+			struct btrfs_device *dev;
+
+			dev = btrfs_zoned_get_device(fs_info, disk_bytenr,
+						     fs_info->sectorsize);
+			if (IS_ERR(dev)) {
+				ret = PTR_ERR(dev);
+				goto error;
+			}
+
+			bio_set_dev(bio, dev->bdev);
+		} else {
+			/*
+			 * Otherwise pick the last added device to support
+			 * cgroup writeback.  For multi-device file systems this
+			 * means blk-cgroup policies have to always be set on the
+			 * last added/replaced device.  This is a bit odd but has
+			 * been like that for a long time.
+			 */
+			bio_set_dev(bio, fs_info->fs_devices->latest_dev->bdev);
+		}
+		wbc_init_bio(wbc, bio);
+	} else {
+		ASSERT(bio_op(bio) != REQ_OP_ZONE_APPEND);
+	}
+	return 0;
+error:
+	bio_ctrl->bio = NULL;
+	btrfs_bio_end_io(btrfs_bio(bio), errno_to_blk_status(ret));
+	return ret;
 }
 
 /*
  * @opf:	bio REQ_OP_* and REQ_* flags as one value
- * @tree:	tree so we can call our merge_bio hook
  * @wbc:	optional writeback control for io accounting
+ * @disk_bytenr: logical bytenr where the write will be
  * @page:	page to add to the bio
+ * @size:	portion of page that we want to write to
  * @pg_offset:	offset of the new bio or to check whether we are adding
  *              a contiguous page to the previous one
- * @size:	portion of page that we want to write
- * @offset:	starting offset in the page
- * @bio_ret:	must be valid pointer, newly allocated bio will be stored there
- * @end_io_func:     end_io callback for new bio
- * @mirror_num:	     desired mirror to read/write
- * @prev_bio_flags:  flags of previous bio to see if we can merge the current one
- * @bio_flags:	flags of the current bio to see if we can merge them
+ * @compress_type:   compress type for current bio
+ *
+ * The will either add the page into the existing @bio_ctrl->bio, or allocate a
+ * new one in @bio_ctrl->bio.
+ * The mirror number for this IO should already be initizlied in
+ * @bio_ctrl->mirror_num.
  */
-static int submit_extent_page(unsigned int opf, struct extent_io_tree *tree,
+static int submit_extent_page(blk_opf_t opf,
 			      struct writeback_control *wbc,
-			      struct page *page, u64 offset,
+			      struct btrfs_bio_ctrl *bio_ctrl,
+			      u64 disk_bytenr, struct page *page,
 			      size_t size, unsigned long pg_offset,
-			      struct bio **bio_ret,
-			      bio_end_io_t end_io_func,
-			      int mirror_num,
-			      unsigned long prev_bio_flags,
-			      unsigned long bio_flags,
+			      enum btrfs_compression_type compress_type,
 			      bool force_bio_submit)
 {
 	int ret = 0;
-	struct bio *bio;
-	size_t page_size = min_t(size_t, size, PAGE_SIZE);
-	sector_t sector = offset >> 9;
+	struct btrfs_inode *inode = BTRFS_I(page->mapping->host);
+	unsigned int cur = pg_offset;
 
-	ASSERT(bio_ret);
+	ASSERT(bio_ctrl);
 
-	if (*bio_ret) {
-		bool contig;
-		bool can_merge = true;
+	ASSERT(pg_offset < PAGE_SIZE && size <= PAGE_SIZE &&
+	       pg_offset + size <= PAGE_SIZE);
 
-		bio = *bio_ret;
-		if (prev_bio_flags & EXTENT_BIO_COMPRESSED)
-			contig = bio->bi_iter.bi_sector == sector;
-		else
-			contig = bio_end_sector(bio) == sector;
+	ASSERT(bio_ctrl->end_io_func);
 
-		ASSERT(tree->ops);
-		if (btrfs_bio_fits_in_stripe(page, page_size, bio, bio_flags))
-			can_merge = false;
+	if (force_bio_submit)
+		submit_one_bio(bio_ctrl);
 
-		if (prev_bio_flags != bio_flags || !contig || !can_merge ||
-		    force_bio_submit ||
-		    bio_add_page(bio, page, page_size, pg_offset) < page_size) {
-			ret = submit_one_bio(bio, mirror_num, prev_bio_flags);
-			if (ret < 0) {
-				*bio_ret = NULL;
+	while (cur < pg_offset + size) {
+		u32 offset = cur - pg_offset;
+		int added;
+
+		/* Allocate new bio if needed */
+		if (!bio_ctrl->bio) {
+			ret = alloc_new_bio(inode, bio_ctrl, wbc, opf,
+					    disk_bytenr, offset,
+					    page_offset(page) + cur,
+					    compress_type);
+			if (ret < 0)
 				return ret;
-			}
-			bio = NULL;
-		} else {
-			if (wbc)
-				wbc_account_cgroup_owner(wbc, page, page_size);
-			return 0;
 		}
+		/*
+		 * We must go through btrfs_bio_add_page() to ensure each
+		 * page range won't cross various boundaries.
+		 */
+		if (compress_type != BTRFS_COMPRESS_NONE)
+			added = btrfs_bio_add_page(bio_ctrl, page, disk_bytenr,
+					size - offset, pg_offset + offset,
+					compress_type);
+		else
+			added = btrfs_bio_add_page(bio_ctrl, page,
+					disk_bytenr + offset, size - offset,
+					pg_offset + offset, compress_type);
+
+		/* Metadata page range should never be split */
+		if (!is_data_inode(&inode->vfs_inode))
+			ASSERT(added == 0 || added == size - offset);
+
+		/* At least we added some page, update the account */
+		if (wbc && added)
+			wbc_account_cgroup_owner(wbc, page, added);
+
+		/* We have reached boundary, submit right now */
+		if (added < size - offset) {
+			/* The bio should contain some page(s) */
+			ASSERT(bio_ctrl->bio->bi_iter.bi_size);
+			submit_one_bio(bio_ctrl);
+		}
+		cur += added;
 	}
+	return 0;
+}
 
-	bio = btrfs_bio_alloc(offset);
-	bio_add_page(bio, page, page_size, pg_offset);
-	bio->bi_end_io = end_io_func;
-	bio->bi_private = tree;
-	bio->bi_write_hint = page->mapping->host->i_write_hint;
-	bio->bi_opf = opf;
-	if (wbc) {
-		struct block_device *bdev;
+static int attach_extent_buffer_page(struct extent_buffer *eb,
+				     struct page *page,
+				     struct btrfs_subpage *prealloc)
+{
+	struct btrfs_fs_info *fs_info = eb->fs_info;
+	int ret = 0;
 
-		bdev = BTRFS_I(page->mapping->host)->root->fs_info->fs_devices->latest_bdev;
-		bio_set_dev(bio, bdev);
-		wbc_init_bio(wbc, bio);
-		wbc_account_cgroup_owner(wbc, page, page_size);
+	/*
+	 * If the page is mapped to btree inode, we should hold the private
+	 * lock to prevent race.
+	 * For cloned or dummy extent buffers, their pages are not mapped and
+	 * will not race with any other ebs.
+	 */
+	if (page->mapping)
+		lockdep_assert_held(&page->mapping->private_lock);
+
+	if (fs_info->nodesize >= PAGE_SIZE) {
+		if (!PagePrivate(page))
+			attach_page_private(page, eb);
+		else
+			WARN_ON(page->private != (unsigned long)eb);
+		return 0;
 	}
 
-	*bio_ret = bio;
+	/* Already mapped, just free prealloc */
+	if (PagePrivate(page)) {
+		btrfs_free_subpage(prealloc);
+		return 0;
+	}
 
+	if (prealloc)
+		/* Has preallocated memory for subpage */
+		attach_page_private(page, prealloc);
+	else
+		/* Do new allocation to attach subpage */
+		ret = btrfs_attach_subpage(fs_info, page,
+					   BTRFS_SUBPAGE_METADATA);
 	return ret;
 }
 
-static void attach_extent_buffer_page(struct extent_buffer *eb,
-				      struct page *page)
+int set_page_extent_mapped(struct page *page)
 {
-	if (!PagePrivate(page)) {
-		SetPagePrivate(page);
-		get_page(page);
-		set_page_private(page, (unsigned long)eb);
-	} else {
-		WARN_ON(page->private != (unsigned long)eb);
-	}
+	struct btrfs_fs_info *fs_info;
+
+	ASSERT(page->mapping);
+
+	if (PagePrivate(page))
+		return 0;
+
+	fs_info = btrfs_sb(page->mapping->host->i_sb);
+
+	if (btrfs_is_subpage(fs_info, page))
+		return btrfs_attach_subpage(fs_info, page, BTRFS_SUBPAGE_DATA);
+
+	attach_page_private(page, (void *)EXTENT_PAGE_PRIVATE);
+	return 0;
 }
 
-void set_page_extent_mapped(struct page *page)
+void clear_page_extent_mapped(struct page *page)
 {
-	if (!PagePrivate(page)) {
-		SetPagePrivate(page);
-		get_page(page);
-		set_page_private(page, EXTENT_PAGE_PRIVATE);
-	}
+	struct btrfs_fs_info *fs_info;
+
+	ASSERT(page->mapping);
+
+	if (!PagePrivate(page))
+		return;
+
+	fs_info = btrfs_sb(page->mapping->host->i_sb);
+	if (btrfs_is_subpage(fs_info, page))
+		return btrfs_detach_subpage(fs_info, page);
+
+	detach_page_private(page);
 }
 
 static struct extent_map *
 __get_extent_map(struct inode *inode, struct page *page, size_t pg_offset,
-		 u64 start, u64 len, get_extent_t *get_extent,
-		 struct extent_map **em_cached)
+		 u64 start, u64 len, struct extent_map **em_cached)
 {
 	struct extent_map *em;
 
@@ -3047,8 +1713,8 @@ __get_extent_map(struct inode *inode, struct page *page, size_t pg_offset,
 		*em_cached = NULL;
 	}
 
-	em = get_extent(BTRFS_I(inode), page, pg_offset, start, len);
-	if (em_cached && !IS_ERR_OR_NULL(em)) {
+	em = btrfs_get_extent(BTRFS_I(inode), page, pg_offset, start, len);
+	if (em_cached && !IS_ERR(em)) {
 		BUG_ON(*em_cached);
 		refcount_inc(&em->refs);
 		*em_cached = em;
@@ -3062,106 +1728,88 @@ __get_extent_map(struct inode *inode, struct page *page, size_t pg_offset,
  * XXX JDM: This needs looking at to ensure proper page locking
  * return 0 on success, otherwise return error
  */
-static int __do_readpage(struct extent_io_tree *tree,
-			 struct page *page,
-			 get_extent_t *get_extent,
-			 struct extent_map **em_cached,
-			 struct bio **bio, int mirror_num,
-			 unsigned long *bio_flags, unsigned int read_flags,
-			 u64 *prev_em_start)
+static int btrfs_do_readpage(struct page *page, struct extent_map **em_cached,
+		      struct btrfs_bio_ctrl *bio_ctrl,
+		      blk_opf_t read_flags, u64 *prev_em_start)
 {
 	struct inode *inode = page->mapping->host;
+	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	u64 start = page_offset(page);
 	const u64 end = start + PAGE_SIZE - 1;
 	u64 cur = start;
 	u64 extent_offset;
 	u64 last_byte = i_size_read(inode);
 	u64 block_start;
-	u64 cur_end;
 	struct extent_map *em;
 	int ret = 0;
-	int nr = 0;
 	size_t pg_offset = 0;
 	size_t iosize;
-	size_t disk_io_size;
 	size_t blocksize = inode->i_sb->s_blocksize;
-	unsigned long this_bio_flag = 0;
-
-	set_page_extent_mapped(page);
+	struct extent_io_tree *tree = &BTRFS_I(inode)->io_tree;
 
-	if (!PageUptodate(page)) {
-		if (cleancache_get_page(page) == 0) {
-			BUG_ON(blocksize != PAGE_SIZE);
-			unlock_extent(tree, start, end);
-			goto out;
-		}
+	ret = set_page_extent_mapped(page);
+	if (ret < 0) {
+		unlock_extent(tree, start, end, NULL);
+		btrfs_page_set_error(fs_info, page, start, PAGE_SIZE);
+		unlock_page(page);
+		goto out;
 	}
 
 	if (page->index == last_byte >> PAGE_SHIFT) {
-		char *userpage;
 		size_t zero_offset = offset_in_page(last_byte);
 
 		if (zero_offset) {
 			iosize = PAGE_SIZE - zero_offset;
-			userpage = kmap_atomic(page);
-			memset(userpage + zero_offset, 0, iosize);
-			flush_dcache_page(page);
-			kunmap_atomic(userpage);
+			memzero_page(page, zero_offset, iosize);
 		}
 	}
+	bio_ctrl->end_io_func = end_bio_extent_readpage;
+	begin_page_read(fs_info, page);
 	while (cur <= end) {
+		unsigned long this_bio_flag = 0;
 		bool force_bio_submit = false;
-		u64 offset;
+		u64 disk_bytenr;
 
+		ASSERT(IS_ALIGNED(cur, fs_info->sectorsize));
 		if (cur >= last_byte) {
-			char *userpage;
 			struct extent_state *cached = NULL;
 
 			iosize = PAGE_SIZE - pg_offset;
-			userpage = kmap_atomic(page);
-			memset(userpage + pg_offset, 0, iosize);
-			flush_dcache_page(page);
-			kunmap_atomic(userpage);
+			memzero_page(page, pg_offset, iosize);
 			set_extent_uptodate(tree, cur, cur + iosize - 1,
 					    &cached, GFP_NOFS);
-			unlock_extent_cached(tree, cur,
-					     cur + iosize - 1, &cached);
+			unlock_extent(tree, cur, cur + iosize - 1, &cached);
+			end_page_read(page, true, cur, iosize);
 			break;
 		}
 		em = __get_extent_map(inode, page, pg_offset, cur,
-				      end - cur + 1, get_extent, em_cached);
-		if (IS_ERR_OR_NULL(em)) {
-			SetPageError(page);
-			unlock_extent(tree, cur, end);
+				      end - cur + 1, em_cached);
+		if (IS_ERR(em)) {
+			unlock_extent(tree, cur, end, NULL);
+			end_page_read(page, false, cur, end + 1 - cur);
+			ret = PTR_ERR(em);
 			break;
 		}
 		extent_offset = cur - em->start;
 		BUG_ON(extent_map_end(em) <= cur);
 		BUG_ON(end < cur);
 
-		if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags)) {
-			this_bio_flag |= EXTENT_BIO_COMPRESSED;
-			extent_set_compress_type(&this_bio_flag,
-						 em->compress_type);
-		}
+		if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags))
+			this_bio_flag = em->compress_type;
 
 		iosize = min(extent_map_end(em) - cur, end - cur + 1);
-		cur_end = min(extent_map_end(em) - 1, end);
 		iosize = ALIGN(iosize, blocksize);
-		if (this_bio_flag & EXTENT_BIO_COMPRESSED) {
-			disk_io_size = em->block_len;
-			offset = em->block_start;
-		} else {
-			offset = em->block_start + extent_offset;
-			disk_io_size = iosize;
-		}
+		if (this_bio_flag != BTRFS_COMPRESS_NONE)
+			disk_bytenr = em->block_start;
+		else
+			disk_bytenr = em->block_start + extent_offset;
 		block_start = em->block_start;
 		if (test_bit(EXTENT_FLAG_PREALLOC, &em->flags))
 			block_start = EXTENT_MAP_HOLE;
 
 		/*
 		 * If we have a file range that points to a compressed extent
-		 * and it's followed by a consecutive file range that points to
+		 * and it's followed by a consecutive file range that points
 		 * to the same compressed extent (possibly with a different
 		 * offset and/or length, so it either points to the whole extent
 		 * or only part of it), we must make sure we do not submit a
@@ -3206,126 +1854,83 @@ static int __do_readpage(struct extent_io_tree *tree,
 
 		/* we've found a hole, just zero and go on */
 		if (block_start == EXTENT_MAP_HOLE) {
-			char *userpage;
 			struct extent_state *cached = NULL;
 
-			userpage = kmap_atomic(page);
-			memset(userpage + pg_offset, 0, iosize);
-			flush_dcache_page(page);
-			kunmap_atomic(userpage);
+			memzero_page(page, pg_offset, iosize);
 
 			set_extent_uptodate(tree, cur, cur + iosize - 1,
 					    &cached, GFP_NOFS);
-			unlock_extent_cached(tree, cur,
-					     cur + iosize - 1, &cached);
+			unlock_extent(tree, cur, cur + iosize - 1, &cached);
+			end_page_read(page, true, cur, iosize);
 			cur = cur + iosize;
 			pg_offset += iosize;
 			continue;
 		}
 		/* the get_extent function already copied into the page */
-		if (test_range_bit(tree, cur, cur_end,
-				   EXTENT_UPTODATE, 1, NULL)) {
-			check_page_uptodate(tree, page);
-			unlock_extent(tree, cur, cur + iosize - 1);
-			cur = cur + iosize;
-			pg_offset += iosize;
-			continue;
-		}
-		/* we have an inline extent but it didn't get marked up
-		 * to date.  Error out
-		 */
 		if (block_start == EXTENT_MAP_INLINE) {
-			SetPageError(page);
-			unlock_extent(tree, cur, cur + iosize - 1);
+			unlock_extent(tree, cur, cur + iosize - 1, NULL);
+			end_page_read(page, true, cur, iosize);
 			cur = cur + iosize;
 			pg_offset += iosize;
 			continue;
 		}
 
-		ret = submit_extent_page(REQ_OP_READ | read_flags, tree, NULL,
-					 page, offset, disk_io_size,
-					 pg_offset, bio,
-					 end_bio_extent_readpage, mirror_num,
-					 *bio_flags,
-					 this_bio_flag,
+		ret = submit_extent_page(REQ_OP_READ | read_flags, NULL,
+					 bio_ctrl, disk_bytenr, page, iosize,
+					 pg_offset, this_bio_flag,
 					 force_bio_submit);
-		if (!ret) {
-			nr++;
-			*bio_flags = this_bio_flag;
-		} else {
-			SetPageError(page);
-			unlock_extent(tree, cur, cur + iosize - 1);
+		if (ret) {
+			/*
+			 * We have to unlock the remaining range, or the page
+			 * will never be unlocked.
+			 */
+			unlock_extent(tree, cur, end, NULL);
+			end_page_read(page, false, cur, end + 1 - cur);
 			goto out;
 		}
 		cur = cur + iosize;
 		pg_offset += iosize;
 	}
 out:
-	if (!nr) {
-		if (!PageError(page))
-			SetPageUptodate(page);
-		unlock_page(page);
-	}
 	return ret;
 }
 
-static inline void contiguous_readpages(struct extent_io_tree *tree,
-					     struct page *pages[], int nr_pages,
-					     u64 start, u64 end,
-					     struct extent_map **em_cached,
-					     struct bio **bio,
-					     unsigned long *bio_flags,
-					     u64 *prev_em_start)
-{
-	struct btrfs_inode *inode = BTRFS_I(pages[0]->mapping->host);
-	int index;
-
-	btrfs_lock_and_flush_ordered_range(tree, inode, start, end, NULL);
-
-	for (index = 0; index < nr_pages; index++) {
-		__do_readpage(tree, pages[index], btrfs_get_extent, em_cached,
-				bio, 0, bio_flags, REQ_RAHEAD, prev_em_start);
-		put_page(pages[index]);
-	}
-}
-
-static int __extent_read_full_page(struct extent_io_tree *tree,
-				   struct page *page,
-				   get_extent_t *get_extent,
-				   struct bio **bio, int mirror_num,
-				   unsigned long *bio_flags,
-				   unsigned int read_flags)
+int btrfs_read_folio(struct file *file, struct folio *folio)
 {
+	struct page *page = &folio->page;
 	struct btrfs_inode *inode = BTRFS_I(page->mapping->host);
 	u64 start = page_offset(page);
 	u64 end = start + PAGE_SIZE - 1;
+	struct btrfs_bio_ctrl bio_ctrl = { 0 };
 	int ret;
 
-	btrfs_lock_and_flush_ordered_range(tree, inode, start, end, NULL);
+	btrfs_lock_and_flush_ordered_range(inode, start, end, NULL);
 
-	ret = __do_readpage(tree, page, get_extent, NULL, bio, mirror_num,
-			    bio_flags, read_flags, NULL);
+	ret = btrfs_do_readpage(page, NULL, &bio_ctrl, 0, NULL);
+	/*
+	 * If btrfs_do_readpage() failed we will want to submit the assembled
+	 * bio to do the cleanup.
+	 */
+	submit_one_bio(&bio_ctrl);
 	return ret;
 }
 
-int extent_read_full_page(struct extent_io_tree *tree, struct page *page,
-			    get_extent_t *get_extent, int mirror_num)
+static inline void contiguous_readpages(struct page *pages[], int nr_pages,
+					u64 start, u64 end,
+					struct extent_map **em_cached,
+					struct btrfs_bio_ctrl *bio_ctrl,
+					u64 *prev_em_start)
 {
-	struct bio *bio = NULL;
-	unsigned long bio_flags = 0;
-	int ret;
+	struct btrfs_inode *inode = BTRFS_I(pages[0]->mapping->host);
+	int index;
 
-	ret = __extent_read_full_page(tree, page, get_extent, &bio, mirror_num,
-				      &bio_flags, 0);
-	if (bio)
-		ret = submit_one_bio(bio, mirror_num, bio_flags);
-	return ret;
-}
+	btrfs_lock_and_flush_ordered_range(inode, start, end, NULL);
 
-static void update_nr_written(struct writeback_control *wbc,
-			      unsigned long nr_written)
-{
-	wbc->nr_to_write -= nr_written;
+	for (index = 0; index < nr_pages; index++) {
+		btrfs_do_readpage(pages[index], em_cached, bio_ctrl,
+				  REQ_RAHEAD, prev_em_start);
+		put_page(pages[index]);
+	}
 }
 
 /*
@@ -3338,20 +1943,22 @@ static void update_nr_written(struct writeback_control *wbc,
  * This returns 0 if all went well (page still locked)
  * This returns < 0 if there were errors (page still locked)
  */
-static noinline_for_stack int writepage_delalloc(struct inode *inode,
-		struct page *page, struct writeback_control *wbc,
-		u64 delalloc_start, unsigned long *nr_written)
+static noinline_for_stack int writepage_delalloc(struct btrfs_inode *inode,
+		struct page *page, struct writeback_control *wbc)
 {
-	u64 page_end = delalloc_start + PAGE_SIZE - 1;
-	bool found;
+	const u64 page_end = page_offset(page) + PAGE_SIZE - 1;
+	u64 delalloc_start = page_offset(page);
 	u64 delalloc_to_write = 0;
-	u64 delalloc_end = 0;
+	/* How many pages are started by btrfs_run_delalloc_range() */
+	unsigned long nr_written = 0;
 	int ret;
 	int page_started = 0;
 
+	while (delalloc_start < page_end) {
+		u64 delalloc_end = page_end;
+		bool found;
 
-	while (delalloc_end < page_end) {
-		found = find_lock_delalloc_range(inode, page,
+		found = find_lock_delalloc_range(&inode->vfs_inode, page,
 					       &delalloc_start,
 					       &delalloc_end);
 		if (!found) {
@@ -3359,17 +1966,11 @@ static noinline_for_stack int writepage_delalloc(struct inode *inode,
 			continue;
 		}
 		ret = btrfs_run_delalloc_range(inode, page, delalloc_start,
-				delalloc_end, &page_started, nr_written, wbc);
+				delalloc_end, &page_started, &nr_written, wbc);
 		if (ret) {
-			SetPageError(page);
-			/*
-			 * btrfs_run_delalloc_range should return < 0 for error
-			 * but just in case, we use > 0 here meaning the IO is
-			 * started, so we don't want to return > 0 unless
-			 * things are going well.
-			 */
-			ret = ret < 0 ? ret : -EIO;
-			goto done;
+			btrfs_page_set_error(inode->root->fs_info, page,
+					     page_offset(page), PAGE_SIZE);
+			return ret;
 		}
 		/*
 		 * delalloc_end is already one less than the total length, so
@@ -3388,23 +1989,69 @@ static noinline_for_stack int writepage_delalloc(struct inode *inode,
 					 thresh);
 	}
 
-	/* did the fill delalloc function already unlock and start
-	 * the IO?
-	 */
+	/* Did btrfs_run_dealloc_range() already unlock and start the IO? */
 	if (page_started) {
 		/*
-		 * we've unlocked the page, so we can't update
-		 * the mapping's writeback index, just update
-		 * nr_to_write.
+		 * We've unlocked the page, so we can't update the mapping's
+		 * writeback index, just update nr_to_write.
 		 */
-		wbc->nr_to_write -= *nr_written;
+		wbc->nr_to_write -= nr_written;
 		return 1;
 	}
 
-	ret = 0;
+	return 0;
+}
 
-done:
-	return ret;
+/*
+ * Find the first byte we need to write.
+ *
+ * For subpage, one page can contain several sectors, and
+ * __extent_writepage_io() will just grab all extent maps in the page
+ * range and try to submit all non-inline/non-compressed extents.
+ *
+ * This is a big problem for subpage, we shouldn't re-submit already written
+ * data at all.
+ * This function will lookup subpage dirty bit to find which range we really
+ * need to submit.
+ *
+ * Return the next dirty range in [@start, @end).
+ * If no dirty range is found, @start will be page_offset(page) + PAGE_SIZE.
+ */
+static void find_next_dirty_byte(struct btrfs_fs_info *fs_info,
+				 struct page *page, u64 *start, u64 *end)
+{
+	struct btrfs_subpage *subpage = (struct btrfs_subpage *)page->private;
+	struct btrfs_subpage_info *spi = fs_info->subpage_info;
+	u64 orig_start = *start;
+	/* Declare as unsigned long so we can use bitmap ops */
+	unsigned long flags;
+	int range_start_bit;
+	int range_end_bit;
+
+	/*
+	 * For regular sector size == page size case, since one page only
+	 * contains one sector, we return the page offset directly.
+	 */
+	if (!btrfs_is_subpage(fs_info, page)) {
+		*start = page_offset(page);
+		*end = page_offset(page) + PAGE_SIZE;
+		return;
+	}
+
+	range_start_bit = spi->dirty_offset +
+			  (offset_in_page(orig_start) >> fs_info->sectorsize_bits);
+
+	/* We should have the page locked, but just in case */
+	spin_lock_irqsave(&subpage->lock, flags);
+	bitmap_next_set_region(subpage->bitmaps, &range_start_bit, &range_end_bit,
+			       spi->dirty_offset + spi->bitmap_nr_bits);
+	spin_unlock_irqrestore(&subpage->lock, flags);
+
+	range_start_bit -= spi->dirty_offset;
+	range_end_bit -= spi->dirty_offset;
+
+	*start = page_offset(page) + range_start_bit * fs_info->sectorsize;
+	*end = page_offset(page) + range_end_bit * fs_info->sectorsize;
 }
 
 /*
@@ -3415,35 +2062,31 @@ done:
  * 0 if all went well (page still locked)
  * < 0 if there were errors (page still locked)
  */
-static noinline_for_stack int __extent_writepage_io(struct inode *inode,
+static noinline_for_stack int __extent_writepage_io(struct btrfs_inode *inode,
 				 struct page *page,
 				 struct writeback_control *wbc,
 				 struct extent_page_data *epd,
 				 loff_t i_size,
-				 unsigned long nr_written,
 				 int *nr_ret)
 {
-	struct extent_io_tree *tree = epd->tree;
-	u64 start = page_offset(page);
-	u64 page_end = start + PAGE_SIZE - 1;
-	u64 end;
-	u64 cur = start;
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
+	u64 cur = page_offset(page);
+	u64 end = cur + PAGE_SIZE - 1;
 	u64 extent_offset;
 	u64 block_start;
-	u64 iosize;
 	struct extent_map *em;
-	size_t pg_offset = 0;
-	size_t blocksize;
+	int saved_ret = 0;
 	int ret = 0;
 	int nr = 0;
-	const unsigned int write_flags = wbc_to_write_flags(wbc);
+	enum req_op op = REQ_OP_WRITE;
+	const blk_opf_t write_flags = wbc_to_write_flags(wbc);
+	bool has_error = false;
 	bool compressed;
 
-	ret = btrfs_writepage_cow_fixup(page, start, page_end);
+	ret = btrfs_writepage_cow_fixup(page);
 	if (ret) {
 		/* Fixup worker will requeue */
 		redirty_page_for_writepage(wbc, page);
-		update_nr_written(wbc, nr_written);
 		unlock_page(page);
 		return 1;
 	}
@@ -3452,37 +2095,67 @@ static noinline_for_stack int __extent_writepage_io(struct inode *inode,
 	 * we don't want to touch the inode after unlocking the page,
 	 * so we update the mapping writeback index now
 	 */
-	update_nr_written(wbc, nr_written + 1);
-
-	end = page_end;
-	blocksize = inode->i_sb->s_blocksize;
+	wbc->nr_to_write--;
 
+	epd->bio_ctrl.end_io_func = end_bio_extent_writepage;
 	while (cur <= end) {
+		u64 disk_bytenr;
 		u64 em_end;
-		u64 offset;
+		u64 dirty_range_start = cur;
+		u64 dirty_range_end;
+		u32 iosize;
 
 		if (cur >= i_size) {
-			btrfs_writepage_endio_finish_ordered(page, cur,
-							     page_end, 1);
+			btrfs_writepage_endio_finish_ordered(inode, page, cur,
+							     end, true);
+			/*
+			 * This range is beyond i_size, thus we don't need to
+			 * bother writing back.
+			 * But we still need to clear the dirty subpage bit, or
+			 * the next time the page gets dirtied, we will try to
+			 * writeback the sectors with subpage dirty bits,
+			 * causing writeback without ordered extent.
+			 */
+			btrfs_page_clear_dirty(fs_info, page, cur, end + 1 - cur);
 			break;
 		}
-		em = btrfs_get_extent(BTRFS_I(inode), NULL, 0, cur,
-				      end - cur + 1);
-		if (IS_ERR_OR_NULL(em)) {
-			SetPageError(page);
+
+		find_next_dirty_byte(fs_info, page, &dirty_range_start,
+				     &dirty_range_end);
+		if (cur < dirty_range_start) {
+			cur = dirty_range_start;
+			continue;
+		}
+
+		em = btrfs_get_extent(inode, NULL, 0, cur, end - cur + 1);
+		if (IS_ERR(em)) {
+			btrfs_page_set_error(fs_info, page, cur, end - cur + 1);
 			ret = PTR_ERR_OR_ZERO(em);
+			has_error = true;
+			if (!saved_ret)
+				saved_ret = ret;
 			break;
 		}
 
 		extent_offset = cur - em->start;
 		em_end = extent_map_end(em);
-		BUG_ON(em_end <= cur);
-		BUG_ON(end < cur);
-		iosize = min(em_end - cur, end - cur + 1);
-		iosize = ALIGN(iosize, blocksize);
-		offset = em->block_start + extent_offset;
+		ASSERT(cur <= em_end);
+		ASSERT(cur < end);
+		ASSERT(IS_ALIGNED(em->start, fs_info->sectorsize));
+		ASSERT(IS_ALIGNED(em->len, fs_info->sectorsize));
 		block_start = em->block_start;
 		compressed = test_bit(EXTENT_FLAG_COMPRESSED, &em->flags);
+		disk_bytenr = em->block_start + extent_offset;
+
+		/*
+		 * Note that em_end from extent_map_end() and dirty_range_end from
+		 * find_next_dirty_byte() are all exclusive
+		 */
+		iosize = min(min(em_end, end + 1), dirty_range_end) - cur;
+
+		if (btrfs_use_zone_append(inode, em->block_start))
+			op = REQ_OP_ZONE_APPEND;
+
 		free_extent_map(em);
 		em = NULL;
 
@@ -3495,35 +2168,55 @@ static noinline_for_stack int __extent_writepage_io(struct inode *inode,
 			if (compressed)
 				nr++;
 			else
-				btrfs_writepage_endio_finish_ordered(page, cur,
-							cur + iosize - 1, 1);
+				btrfs_writepage_endio_finish_ordered(inode,
+						page, cur, cur + iosize - 1, true);
+			btrfs_page_clear_dirty(fs_info, page, cur, iosize);
 			cur += iosize;
-			pg_offset += iosize;
 			continue;
 		}
 
-		btrfs_set_range_writeback(tree, cur, cur + iosize - 1);
+		btrfs_set_range_writeback(inode, cur, cur + iosize - 1);
 		if (!PageWriteback(page)) {
-			btrfs_err(BTRFS_I(inode)->root->fs_info,
+			btrfs_err(inode->root->fs_info,
 				   "page %lu not writeback, cur %llu end %llu",
 			       page->index, cur, end);
 		}
 
-		ret = submit_extent_page(REQ_OP_WRITE | write_flags, tree, wbc,
-					 page, offset, iosize, pg_offset,
-					 &epd->bio,
-					 end_bio_extent_writepage,
-					 0, 0, 0, false);
+		/*
+		 * Although the PageDirty bit is cleared before entering this
+		 * function, subpage dirty bit is not cleared.
+		 * So clear subpage dirty bit here so next time we won't submit
+		 * page for range already written to disk.
+		 */
+		btrfs_page_clear_dirty(fs_info, page, cur, iosize);
+
+		ret = submit_extent_page(op | write_flags, wbc,
+					 &epd->bio_ctrl, disk_bytenr,
+					 page, iosize,
+					 cur - page_offset(page),
+					 0, false);
 		if (ret) {
-			SetPageError(page);
+			has_error = true;
+			if (!saved_ret)
+				saved_ret = ret;
+
+			btrfs_page_set_error(fs_info, page, cur, iosize);
 			if (PageWriteback(page))
-				end_page_writeback(page);
+				btrfs_page_clear_writeback(fs_info, page, cur,
+							   iosize);
 		}
 
-		cur = cur + iosize;
-		pg_offset += iosize;
+		cur += iosize;
 		nr++;
 	}
+	/*
+	 * If we finish without problem, we should not only clear page dirty,
+	 * but also empty subpage dirty bits
+	 */
+	if (!has_error)
+		btrfs_page_assert_not_dirty(fs_info, page);
+	else
+		ret = saved_ret;
 	*nr_ret = nr;
 	return ret;
 }
@@ -3540,52 +2233,51 @@ static noinline_for_stack int __extent_writepage_io(struct inode *inode,
 static int __extent_writepage(struct page *page, struct writeback_control *wbc,
 			      struct extent_page_data *epd)
 {
+	struct folio *folio = page_folio(page);
 	struct inode *inode = page->mapping->host;
-	u64 start = page_offset(page);
-	u64 page_end = start + PAGE_SIZE - 1;
+	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+	const u64 page_start = page_offset(page);
+	const u64 page_end = page_start + PAGE_SIZE - 1;
 	int ret;
 	int nr = 0;
 	size_t pg_offset;
 	loff_t i_size = i_size_read(inode);
 	unsigned long end_index = i_size >> PAGE_SHIFT;
-	unsigned long nr_written = 0;
 
 	trace___extent_writepage(page, inode, wbc);
 
 	WARN_ON(!PageLocked(page));
 
-	ClearPageError(page);
+	btrfs_page_clear_error(btrfs_sb(inode->i_sb), page,
+			       page_offset(page), PAGE_SIZE);
 
 	pg_offset = offset_in_page(i_size);
 	if (page->index > end_index ||
 	   (page->index == end_index && !pg_offset)) {
-		page->mapping->a_ops->invalidatepage(page, 0, PAGE_SIZE);
-		unlock_page(page);
+		folio_invalidate(folio, 0, folio_size(folio));
+		folio_unlock(folio);
 		return 0;
 	}
 
-	if (page->index == end_index) {
-		char *userpage;
+	if (page->index == end_index)
+		memzero_page(page, pg_offset, PAGE_SIZE - pg_offset);
 
-		userpage = kmap_atomic(page);
-		memset(userpage + pg_offset, 0,
-		       PAGE_SIZE - pg_offset);
-		kunmap_atomic(userpage);
-		flush_dcache_page(page);
+	ret = set_page_extent_mapped(page);
+	if (ret < 0) {
+		SetPageError(page);
+		goto done;
 	}
 
-	set_page_extent_mapped(page);
-
 	if (!epd->extent_locked) {
-		ret = writepage_delalloc(inode, page, wbc, start, &nr_written);
+		ret = writepage_delalloc(BTRFS_I(inode), page, wbc);
 		if (ret == 1)
 			return 0;
 		if (ret)
 			goto done;
 	}
 
-	ret = __extent_writepage_io(inode, page, wbc, epd,
-				    i_size, nr_written, &nr);
+	ret = __extent_writepage_io(BTRFS_I(inode), page, wbc, epd, i_size,
+				    &nr);
 	if (ret == 1)
 		return 0;
 
@@ -3595,11 +2287,52 @@ done:
 		set_page_writeback(page);
 		end_page_writeback(page);
 	}
-	if (PageError(page)) {
-		ret = ret < 0 ? ret : -EIO;
-		end_extent_writepage(page, ret, start, page_end);
+	/*
+	 * Here we used to have a check for PageError() and then set @ret and
+	 * call end_extent_writepage().
+	 *
+	 * But in fact setting @ret here will cause different error paths
+	 * between subpage and regular sectorsize.
+	 *
+	 * For regular page size, we never submit current page, but only add
+	 * current page to current bio.
+	 * The bio submission can only happen in next page.
+	 * Thus if we hit the PageError() branch, @ret is already set to
+	 * non-zero value and will not get updated for regular sectorsize.
+	 *
+	 * But for subpage case, it's possible we submit part of current page,
+	 * thus can get PageError() set by submitted bio of the same page,
+	 * while our @ret is still 0.
+	 *
+	 * So here we unify the behavior and don't set @ret.
+	 * Error can still be properly passed to higher layer as page will
+	 * be set error, here we just don't handle the IO failure.
+	 *
+	 * NOTE: This is just a hotfix for subpage.
+	 * The root fix will be properly ending ordered extent when we hit
+	 * an error during writeback.
+	 *
+	 * But that needs a bigger refactoring, as we not only need to grab the
+	 * submitted OE, but also need to know exactly at which bytenr we hit
+	 * the error.
+	 * Currently the full page based __extent_writepage_io() is not
+	 * capable of that.
+	 */
+	if (PageError(page))
+		end_extent_writepage(page, ret, page_start, page_end);
+	if (epd->extent_locked) {
+		/*
+		 * If epd->extent_locked, it's from extent_write_locked_range(),
+		 * the page can either be locked by lock_page() or
+		 * process_one_page().
+		 * Let btrfs_page_unlock_writer() handle both cases.
+		 */
+		ASSERT(wbc);
+		btrfs_page_unlock_writer(fs_info, page, wbc->range_start,
+					 wbc->range_end + 1 - wbc->range_start);
+	} else {
+		unlock_page(page);
 	}
-	unlock_page(page);
 	ASSERT(ret <= 0);
 	return ret;
 }
@@ -3618,24 +2351,25 @@ static void end_extent_buffer_writeback(struct extent_buffer *eb)
 }
 
 /*
- * Lock eb pages and flush the bio if we can't the locks
+ * Lock extent buffer status and pages for writeback.
+ *
+ * May try to flush write bio if we can't get the lock.
  *
- * Return  0 if nothing went wrong
- * Return >0 is same as 0, except bio is not submitted
- * Return <0 if something went wrong, no page is locked
+ * Return  0 if the extent buffer doesn't need to be submitted.
+ *           (E.g. the extent buffer is not dirty)
+ * Return >0 is the extent buffer is submitted to bio.
+ * Return <0 if something went wrong, no page is locked.
  */
 static noinline_for_stack int lock_extent_buffer_for_io(struct extent_buffer *eb,
 			  struct extent_page_data *epd)
 {
 	struct btrfs_fs_info *fs_info = eb->fs_info;
-	int i, num_pages, failed_page_nr;
+	int i, num_pages;
 	int flush = 0;
 	int ret = 0;
 
 	if (!btrfs_try_tree_write_lock(eb)) {
-		ret = flush_write_bio(epd);
-		if (ret < 0)
-			return ret;
+		submit_write_bio(epd, 0);
 		flush = 1;
 		btrfs_tree_lock(eb);
 	}
@@ -3645,9 +2379,7 @@ static noinline_for_stack int lock_extent_buffer_for_io(struct extent_buffer *eb
 		if (!epd->sync_io)
 			return 0;
 		if (!flush) {
-			ret = flush_write_bio(epd);
-			if (ret < 0)
-				return ret;
+			submit_write_bio(epd, 0);
 			flush = 1;
 		}
 		while (1) {
@@ -3679,7 +2411,13 @@ static noinline_for_stack int lock_extent_buffer_for_io(struct extent_buffer *eb
 
 	btrfs_tree_unlock(eb);
 
-	if (!ret)
+	/*
+	 * Either we don't need to submit any tree block, or we're submitting
+	 * subpage eb.
+	 * Subpage metadata doesn't use page locking at all, so we can skip
+	 * the page locking.
+	 */
+	if (!ret || fs_info->nodesize < PAGE_SIZE)
 		return ret;
 
 	num_pages = num_extent_pages(eb);
@@ -3688,14 +2426,7 @@ static noinline_for_stack int lock_extent_buffer_for_io(struct extent_buffer *eb
 
 		if (!trylock_page(p)) {
 			if (!flush) {
-				int err;
-
-				err = flush_write_bio(epd);
-				if (err < 0) {
-					ret = err;
-					failed_page_nr = i;
-					goto err_unlock;
-				}
+				submit_write_bio(epd, 0);
 				flush = 1;
 			}
 			lock_page(p);
@@ -3703,41 +2434,34 @@ static noinline_for_stack int lock_extent_buffer_for_io(struct extent_buffer *eb
 	}
 
 	return ret;
-err_unlock:
-	/* Unlock already locked pages */
-	for (i = 0; i < failed_page_nr; i++)
-		unlock_page(eb->pages[i]);
-	/*
-	 * Clear EXTENT_BUFFER_WRITEBACK and wake up anyone waiting on it.
-	 * Also set back EXTENT_BUFFER_DIRTY so future attempts to this eb can
-	 * be made and undo everything done before.
-	 */
-	btrfs_tree_lock(eb);
-	spin_lock(&eb->refs_lock);
-	set_bit(EXTENT_BUFFER_DIRTY, &eb->bflags);
-	end_extent_buffer_writeback(eb);
-	spin_unlock(&eb->refs_lock);
-	percpu_counter_add_batch(&fs_info->dirty_metadata_bytes, eb->len,
-				 fs_info->dirty_metadata_batch);
-	btrfs_clear_header_flag(eb, BTRFS_HEADER_FLAG_WRITTEN);
-	btrfs_tree_unlock(eb);
-	return ret;
 }
 
-static void set_btree_ioerr(struct page *page)
+static void set_btree_ioerr(struct page *page, struct extent_buffer *eb)
 {
-	struct extent_buffer *eb = (struct extent_buffer *)page->private;
-	struct btrfs_fs_info *fs_info;
+	struct btrfs_fs_info *fs_info = eb->fs_info;
 
-	SetPageError(page);
+	btrfs_page_set_error(fs_info, page, eb->start, eb->len);
 	if (test_and_set_bit(EXTENT_BUFFER_WRITE_ERR, &eb->bflags))
 		return;
 
 	/*
+	 * A read may stumble upon this buffer later, make sure that it gets an
+	 * error and knows there was an error.
+	 */
+	clear_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
+
+	/*
+	 * We need to set the mapping with the io error as well because a write
+	 * error will flip the file system readonly, and then syncfs() will
+	 * return a 0 because we are readonly if we don't modify the err seq for
+	 * the superblock.
+	 */
+	mapping_set_error(page->mapping, -EIO);
+
+	/*
 	 * If we error out, we should add back the dirty_metadata_bytes
 	 * to make it consistent.
 	 */
-	fs_info = eb->fs_info;
 	percpu_counter_add_batch(&fs_info->dirty_metadata_bytes,
 				 eb->len, fs_info->dirty_metadata_batch);
 
@@ -3781,21 +2505,106 @@ static void set_btree_ioerr(struct page *page)
 	 */
 	switch (eb->log_index) {
 	case -1:
-		set_bit(BTRFS_FS_BTREE_ERR, &eb->fs_info->flags);
+		set_bit(BTRFS_FS_BTREE_ERR, &fs_info->flags);
 		break;
 	case 0:
-		set_bit(BTRFS_FS_LOG1_ERR, &eb->fs_info->flags);
+		set_bit(BTRFS_FS_LOG1_ERR, &fs_info->flags);
 		break;
 	case 1:
-		set_bit(BTRFS_FS_LOG2_ERR, &eb->fs_info->flags);
+		set_bit(BTRFS_FS_LOG2_ERR, &fs_info->flags);
 		break;
 	default:
 		BUG(); /* unexpected, logic error */
 	}
 }
 
-static void end_bio_extent_buffer_writepage(struct bio *bio)
+/*
+ * The endio specific version which won't touch any unsafe spinlock in endio
+ * context.
+ */
+static struct extent_buffer *find_extent_buffer_nolock(
+		struct btrfs_fs_info *fs_info, u64 start)
+{
+	struct extent_buffer *eb;
+
+	rcu_read_lock();
+	eb = radix_tree_lookup(&fs_info->buffer_radix,
+			       start >> fs_info->sectorsize_bits);
+	if (eb && atomic_inc_not_zero(&eb->refs)) {
+		rcu_read_unlock();
+		return eb;
+	}
+	rcu_read_unlock();
+	return NULL;
+}
+
+/*
+ * The endio function for subpage extent buffer write.
+ *
+ * Unlike end_bio_extent_buffer_writepage(), we only call end_page_writeback()
+ * after all extent buffers in the page has finished their writeback.
+ */
+static void end_bio_subpage_eb_writepage(struct btrfs_bio *bbio)
+{
+	struct bio *bio = &bbio->bio;
+	struct btrfs_fs_info *fs_info;
+	struct bio_vec *bvec;
+	struct bvec_iter_all iter_all;
+
+	fs_info = btrfs_sb(bio_first_page_all(bio)->mapping->host->i_sb);
+	ASSERT(fs_info->nodesize < PAGE_SIZE);
+
+	ASSERT(!bio_flagged(bio, BIO_CLONED));
+	bio_for_each_segment_all(bvec, bio, iter_all) {
+		struct page *page = bvec->bv_page;
+		u64 bvec_start = page_offset(page) + bvec->bv_offset;
+		u64 bvec_end = bvec_start + bvec->bv_len - 1;
+		u64 cur_bytenr = bvec_start;
+
+		ASSERT(IS_ALIGNED(bvec->bv_len, fs_info->nodesize));
+
+		/* Iterate through all extent buffers in the range */
+		while (cur_bytenr <= bvec_end) {
+			struct extent_buffer *eb;
+			int done;
+
+			/*
+			 * Here we can't use find_extent_buffer(), as it may
+			 * try to lock eb->refs_lock, which is not safe in endio
+			 * context.
+			 */
+			eb = find_extent_buffer_nolock(fs_info, cur_bytenr);
+			ASSERT(eb);
+
+			cur_bytenr = eb->start + eb->len;
+
+			ASSERT(test_bit(EXTENT_BUFFER_WRITEBACK, &eb->bflags));
+			done = atomic_dec_and_test(&eb->io_pages);
+			ASSERT(done);
+
+			if (bio->bi_status ||
+			    test_bit(EXTENT_BUFFER_WRITE_ERR, &eb->bflags)) {
+				ClearPageUptodate(page);
+				set_btree_ioerr(page, eb);
+			}
+
+			btrfs_subpage_clear_writeback(fs_info, page, eb->start,
+						      eb->len);
+			end_extent_buffer_writeback(eb);
+			/*
+			 * free_extent_buffer() will grab spinlock which is not
+			 * safe in endio context. Thus here we manually dec
+			 * the ref.
+			 */
+			atomic_dec(&eb->refs);
+		}
+	}
+	bio_put(bio);
+}
+
+static void end_bio_extent_buffer_writepage(struct btrfs_bio *bbio)
 {
+	struct bio *bio = &bbio->bio;
 	struct bio_vec *bvec;
 	struct extent_buffer *eb;
 	int done;
@@ -3812,7 +2621,7 @@ static void end_bio_extent_buffer_writepage(struct bio *bio)
 		if (bio->bi_status ||
 		    test_bit(EXTENT_BUFFER_WRITE_ERR, &eb->bflags)) {
 			ClearPageUptodate(page);
-			set_btree_ioerr(page);
+			set_btree_ioerr(page, eb);
 		}
 
 		end_page_writeback(page);
@@ -3826,51 +2635,105 @@ static void end_bio_extent_buffer_writepage(struct bio *bio)
 	bio_put(bio);
 }
 
-static noinline_for_stack int write_one_eb(struct extent_buffer *eb,
-			struct writeback_control *wbc,
-			struct extent_page_data *epd)
+static void prepare_eb_write(struct extent_buffer *eb)
 {
-	struct btrfs_fs_info *fs_info = eb->fs_info;
-	struct extent_io_tree *tree = &BTRFS_I(fs_info->btree_inode)->io_tree;
-	u64 offset = eb->start;
 	u32 nritems;
-	int i, num_pages;
-	unsigned long start, end;
-	unsigned int write_flags = wbc_to_write_flags(wbc) | REQ_META;
-	int ret = 0;
+	unsigned long start;
+	unsigned long end;
 
 	clear_bit(EXTENT_BUFFER_WRITE_ERR, &eb->bflags);
-	num_pages = num_extent_pages(eb);
-	atomic_set(&eb->io_pages, num_pages);
+	atomic_set(&eb->io_pages, num_extent_pages(eb));
 
-	/* set btree blocks beyond nritems with 0 to avoid stale content. */
+	/* Set btree blocks beyond nritems with 0 to avoid stale content */
 	nritems = btrfs_header_nritems(eb);
 	if (btrfs_header_level(eb) > 0) {
 		end = btrfs_node_key_ptr_offset(nritems);
-
 		memzero_extent_buffer(eb, end, eb->len - end);
 	} else {
 		/*
-		 * leaf:
+		 * Leaf:
 		 * header 0 1 2 .. N ... data_N .. data_2 data_1 data_0
 		 */
 		start = btrfs_item_nr_offset(nritems);
 		end = BTRFS_LEAF_DATA_OFFSET + leaf_data_end(eb);
 		memzero_extent_buffer(eb, start, end - start);
 	}
+}
+
+/*
+ * Unlike the work in write_one_eb(), we rely completely on extent locking.
+ * Page locking is only utilized at minimum to keep the VMM code happy.
+ */
+static int write_one_subpage_eb(struct extent_buffer *eb,
+				struct writeback_control *wbc,
+				struct extent_page_data *epd)
+{
+	struct btrfs_fs_info *fs_info = eb->fs_info;
+	struct page *page = eb->pages[0];
+	blk_opf_t write_flags = wbc_to_write_flags(wbc);
+	bool no_dirty_ebs = false;
+	int ret;
+
+	prepare_eb_write(eb);
+
+	/* clear_page_dirty_for_io() in subpage helper needs page locked */
+	lock_page(page);
+	btrfs_subpage_set_writeback(fs_info, page, eb->start, eb->len);
 
+	/* Check if this is the last dirty bit to update nr_written */
+	no_dirty_ebs = btrfs_subpage_clear_and_test_dirty(fs_info, page,
+							  eb->start, eb->len);
+	if (no_dirty_ebs)
+		clear_page_dirty_for_io(page);
+
+	epd->bio_ctrl.end_io_func = end_bio_subpage_eb_writepage;
+
+	ret = submit_extent_page(REQ_OP_WRITE | write_flags, wbc,
+			&epd->bio_ctrl, eb->start, page, eb->len,
+			eb->start - page_offset(page), 0, false);
+	if (ret) {
+		btrfs_subpage_clear_writeback(fs_info, page, eb->start, eb->len);
+		set_btree_ioerr(page, eb);
+		unlock_page(page);
+
+		if (atomic_dec_and_test(&eb->io_pages))
+			end_extent_buffer_writeback(eb);
+		return -EIO;
+	}
+	unlock_page(page);
+	/*
+	 * Submission finished without problem, if no range of the page is
+	 * dirty anymore, we have submitted a page.  Update nr_written in wbc.
+	 */
+	if (no_dirty_ebs)
+		wbc->nr_to_write--;
+	return ret;
+}
+
+static noinline_for_stack int write_one_eb(struct extent_buffer *eb,
+			struct writeback_control *wbc,
+			struct extent_page_data *epd)
+{
+	u64 disk_bytenr = eb->start;
+	int i, num_pages;
+	blk_opf_t write_flags = wbc_to_write_flags(wbc);
+	int ret = 0;
+
+	prepare_eb_write(eb);
+
+	epd->bio_ctrl.end_io_func = end_bio_extent_buffer_writepage;
+
+	num_pages = num_extent_pages(eb);
 	for (i = 0; i < num_pages; i++) {
 		struct page *p = eb->pages[i];
 
 		clear_page_dirty_for_io(p);
 		set_page_writeback(p);
-		ret = submit_extent_page(REQ_OP_WRITE | write_flags, tree, wbc,
-					 p, offset, PAGE_SIZE, 0,
-					 &epd->bio,
-					 end_bio_extent_buffer_writepage,
-					 0, 0, 0, false);
+		ret = submit_extent_page(REQ_OP_WRITE | write_flags, wbc,
+					 &epd->bio_ctrl, disk_bytenr, p,
+					 PAGE_SIZE, 0, 0, false);
 		if (ret) {
-			set_btree_ioerr(p);
+			set_btree_ioerr(p, eb);
 			if (PageWriteback(p))
 				end_page_writeback(p);
 			if (atomic_sub_and_test(num_pages - i, &eb->io_pages))
@@ -3878,8 +2741,8 @@ static noinline_for_stack int write_one_eb(struct extent_buffer *eb,
 			ret = -EIO;
 			break;
 		}
-		offset += PAGE_SIZE;
-		update_nr_written(wbc, 1);
+		disk_bytenr += PAGE_SIZE;
+		wbc->nr_to_write--;
 		unlock_page(p);
 	}
 
@@ -3894,17 +2757,206 @@ static noinline_for_stack int write_one_eb(struct extent_buffer *eb,
 	return ret;
 }
 
+/*
+ * Submit one subpage btree page.
+ *
+ * The main difference to submit_eb_page() is:
+ * - Page locking
+ *   For subpage, we don't rely on page locking at all.
+ *
+ * - Flush write bio
+ *   We only flush bio if we may be unable to fit current extent buffers into
+ *   current bio.
+ *
+ * Return >=0 for the number of submitted extent buffers.
+ * Return <0 for fatal error.
+ */
+static int submit_eb_subpage(struct page *page,
+			     struct writeback_control *wbc,
+			     struct extent_page_data *epd)
+{
+	struct btrfs_fs_info *fs_info = btrfs_sb(page->mapping->host->i_sb);
+	int submitted = 0;
+	u64 page_start = page_offset(page);
+	int bit_start = 0;
+	int sectors_per_node = fs_info->nodesize >> fs_info->sectorsize_bits;
+	int ret;
+
+	/* Lock and write each dirty extent buffers in the range */
+	while (bit_start < fs_info->subpage_info->bitmap_nr_bits) {
+		struct btrfs_subpage *subpage = (struct btrfs_subpage *)page->private;
+		struct extent_buffer *eb;
+		unsigned long flags;
+		u64 start;
+
+		/*
+		 * Take private lock to ensure the subpage won't be detached
+		 * in the meantime.
+		 */
+		spin_lock(&page->mapping->private_lock);
+		if (!PagePrivate(page)) {
+			spin_unlock(&page->mapping->private_lock);
+			break;
+		}
+		spin_lock_irqsave(&subpage->lock, flags);
+		if (!test_bit(bit_start + fs_info->subpage_info->dirty_offset,
+			      subpage->bitmaps)) {
+			spin_unlock_irqrestore(&subpage->lock, flags);
+			spin_unlock(&page->mapping->private_lock);
+			bit_start++;
+			continue;
+		}
+
+		start = page_start + bit_start * fs_info->sectorsize;
+		bit_start += sectors_per_node;
+
+		/*
+		 * Here we just want to grab the eb without touching extra
+		 * spin locks, so call find_extent_buffer_nolock().
+		 */
+		eb = find_extent_buffer_nolock(fs_info, start);
+		spin_unlock_irqrestore(&subpage->lock, flags);
+		spin_unlock(&page->mapping->private_lock);
+
+		/*
+		 * The eb has already reached 0 refs thus find_extent_buffer()
+		 * doesn't return it. We don't need to write back such eb
+		 * anyway.
+		 */
+		if (!eb)
+			continue;
+
+		ret = lock_extent_buffer_for_io(eb, epd);
+		if (ret == 0) {
+			free_extent_buffer(eb);
+			continue;
+		}
+		if (ret < 0) {
+			free_extent_buffer(eb);
+			goto cleanup;
+		}
+		ret = write_one_subpage_eb(eb, wbc, epd);
+		free_extent_buffer(eb);
+		if (ret < 0)
+			goto cleanup;
+		submitted++;
+	}
+	return submitted;
+
+cleanup:
+	/* We hit error, end bio for the submitted extent buffers */
+	submit_write_bio(epd, ret);
+	return ret;
+}
+
+/*
+ * Submit all page(s) of one extent buffer.
+ *
+ * @page:	the page of one extent buffer
+ * @eb_context:	to determine if we need to submit this page, if current page
+ *		belongs to this eb, we don't need to submit
+ *
+ * The caller should pass each page in their bytenr order, and here we use
+ * @eb_context to determine if we have submitted pages of one extent buffer.
+ *
+ * If we have, we just skip until we hit a new page that doesn't belong to
+ * current @eb_context.
+ *
+ * If not, we submit all the page(s) of the extent buffer.
+ *
+ * Return >0 if we have submitted the extent buffer successfully.
+ * Return 0 if we don't need to submit the page, as it's already submitted by
+ * previous call.
+ * Return <0 for fatal error.
+ */
+static int submit_eb_page(struct page *page, struct writeback_control *wbc,
+			  struct extent_page_data *epd,
+			  struct extent_buffer **eb_context)
+{
+	struct address_space *mapping = page->mapping;
+	struct btrfs_block_group *cache = NULL;
+	struct extent_buffer *eb;
+	int ret;
+
+	if (!PagePrivate(page))
+		return 0;
+
+	if (btrfs_sb(page->mapping->host->i_sb)->nodesize < PAGE_SIZE)
+		return submit_eb_subpage(page, wbc, epd);
+
+	spin_lock(&mapping->private_lock);
+	if (!PagePrivate(page)) {
+		spin_unlock(&mapping->private_lock);
+		return 0;
+	}
+
+	eb = (struct extent_buffer *)page->private;
+
+	/*
+	 * Shouldn't happen and normally this would be a BUG_ON but no point
+	 * crashing the machine for something we can survive anyway.
+	 */
+	if (WARN_ON(!eb)) {
+		spin_unlock(&mapping->private_lock);
+		return 0;
+	}
+
+	if (eb == *eb_context) {
+		spin_unlock(&mapping->private_lock);
+		return 0;
+	}
+	ret = atomic_inc_not_zero(&eb->refs);
+	spin_unlock(&mapping->private_lock);
+	if (!ret)
+		return 0;
+
+	if (!btrfs_check_meta_write_pointer(eb->fs_info, eb, &cache)) {
+		/*
+		 * If for_sync, this hole will be filled with
+		 * trasnsaction commit.
+		 */
+		if (wbc->sync_mode == WB_SYNC_ALL && !wbc->for_sync)
+			ret = -EAGAIN;
+		else
+			ret = 0;
+		free_extent_buffer(eb);
+		return ret;
+	}
+
+	*eb_context = eb;
+
+	ret = lock_extent_buffer_for_io(eb, epd);
+	if (ret <= 0) {
+		btrfs_revert_meta_write_pointer(cache, eb);
+		if (cache)
+			btrfs_put_block_group(cache);
+		free_extent_buffer(eb);
+		return ret;
+	}
+	if (cache) {
+		/*
+		 * Implies write in zoned mode. Mark the last eb in a block group.
+		 */
+		btrfs_schedule_zone_finish_bg(cache, eb);
+		btrfs_put_block_group(cache);
+	}
+	ret = write_one_eb(eb, wbc, epd);
+	free_extent_buffer(eb);
+	if (ret < 0)
+		return ret;
+	return 1;
+}
+
 int btree_write_cache_pages(struct address_space *mapping,
 				   struct writeback_control *wbc)
 {
-	struct extent_io_tree *tree = &BTRFS_I(mapping->host)->io_tree;
-	struct extent_buffer *eb, *prev_eb = NULL;
+	struct extent_buffer *eb_context = NULL;
 	struct extent_page_data epd = {
-		.bio = NULL,
-		.tree = tree,
+		.bio_ctrl = { 0 },
 		.extent_locked = 0,
 		.sync_io = wbc->sync_mode == WB_SYNC_ALL,
 	};
+	struct btrfs_fs_info *fs_info = BTRFS_I(mapping->host)->root->fs_info;
 	int ret = 0;
 	int done = 0;
 	int nr_to_write_done = 0;
@@ -3933,6 +2985,7 @@ int btree_write_cache_pages(struct address_space *mapping,
 		tag = PAGECACHE_TAG_TOWRITE;
 	else
 		tag = PAGECACHE_TAG_DIRTY;
+	btrfs_zoned_meta_io_lock(fs_info);
 retry:
 	if (wbc->sync_mode == WB_SYNC_ALL)
 		tag_pages_for_writeback(mapping, index, end);
@@ -3944,55 +2997,13 @@ retry:
 		for (i = 0; i < nr_pages; i++) {
 			struct page *page = pvec.pages[i];
 
-			if (!PagePrivate(page))
+			ret = submit_eb_page(page, wbc, &epd, &eb_context);
+			if (ret == 0)
 				continue;
-
-			spin_lock(&mapping->private_lock);
-			if (!PagePrivate(page)) {
-				spin_unlock(&mapping->private_lock);
-				continue;
-			}
-
-			eb = (struct extent_buffer *)page->private;
-
-			/*
-			 * Shouldn't happen and normally this would be a BUG_ON
-			 * but no sense in crashing the users box for something
-			 * we can survive anyway.
-			 */
-			if (WARN_ON(!eb)) {
-				spin_unlock(&mapping->private_lock);
-				continue;
-			}
-
-			if (eb == prev_eb) {
-				spin_unlock(&mapping->private_lock);
-				continue;
-			}
-
-			ret = atomic_inc_not_zero(&eb->refs);
-			spin_unlock(&mapping->private_lock);
-			if (!ret)
-				continue;
-
-			prev_eb = eb;
-			ret = lock_extent_buffer_for_io(eb, &epd);
-			if (!ret) {
-				free_extent_buffer(eb);
-				continue;
-			} else if (ret < 0) {
-				done = 1;
-				free_extent_buffer(eb);
-				break;
-			}
-
-			ret = write_one_eb(eb, wbc, &epd);
-			if (ret) {
+			if (ret < 0) {
 				done = 1;
-				free_extent_buffer(eb);
 				break;
 			}
-			free_extent_buffer(eb);
 
 			/*
 			 * the filesystem may choose to bump up nr_to_write.
@@ -4013,20 +3024,52 @@ retry:
 		index = 0;
 		goto retry;
 	}
-	ASSERT(ret <= 0);
-	if (ret < 0) {
-		end_write_bio(&epd, ret);
-		return ret;
-	}
-	ret = flush_write_bio(&epd);
+	/*
+	 * If something went wrong, don't allow any metadata write bio to be
+	 * submitted.
+	 *
+	 * This would prevent use-after-free if we had dirty pages not
+	 * cleaned up, which can still happen by fuzzed images.
+	 *
+	 * - Bad extent tree
+	 *   Allowing existing tree block to be allocated for other trees.
+	 *
+	 * - Log tree operations
+	 *   Exiting tree blocks get allocated to log tree, bumps its
+	 *   generation, then get cleaned in tree re-balance.
+	 *   Such tree block will not be written back, since it's clean,
+	 *   thus no WRITTEN flag set.
+	 *   And after log writes back, this tree block is not traced by
+	 *   any dirty extent_io_tree.
+	 *
+	 * - Offending tree block gets re-dirtied from its original owner
+	 *   Since it has bumped generation, no WRITTEN flag, it can be
+	 *   reused without COWing. This tree block will not be traced
+	 *   by btrfs_transaction::dirty_pages.
+	 *
+	 *   Now such dirty tree block will not be cleaned by any dirty
+	 *   extent io tree. Thus we don't want to submit such wild eb
+	 *   if the fs already has error.
+	 *
+	 * We can get ret > 0 from submit_extent_page() indicating how many ebs
+	 * were submitted. Reset it to 0 to avoid false alerts for the caller.
+	 */
+	if (ret > 0)
+		ret = 0;
+	if (!ret && BTRFS_FS_ERROR(fs_info))
+		ret = -EROFS;
+	submit_write_bio(&epd, ret);
+
+	btrfs_zoned_meta_io_unlock(fs_info);
 	return ret;
 }
 
 /**
- * write_cache_pages - walk the list of dirty pages of the given address space and write all of them.
+ * Walk the list of dirty pages of the given address space and write all of them.
+ *
  * @mapping: address space structure to write
- * @wbc: subtract the number of written pages from *@wbc->nr_to_write
- * @data: data passed to __extent_writepage function
+ * @wbc:     subtract the number of written pages from *@wbc->nr_to_write
+ * @epd:     holds context for the write, namely the bio
  *
  * If a page is already under I/O, write_cache_pages() skips it, even
  * if it's dirty.  This is desirable behaviour for memory-cleaning writeback,
@@ -4119,8 +3162,7 @@ retry:
 			 * tmpfs file mapping
 			 */
 			if (!trylock_page(page)) {
-				ret = flush_write_bio(epd);
-				BUG_ON(ret < 0);
+				submit_write_bio(epd, 0);
 				lock_page(page);
 			}
 
@@ -4130,10 +3172,8 @@ retry:
 			}
 
 			if (wbc->sync_mode != WB_SYNC_NONE) {
-				if (PageWriteback(page)) {
-					ret = flush_write_bio(epd);
-					BUG_ON(ret < 0);
-				}
+				if (PageWriteback(page))
+					submit_write_bio(epd, 0);
 				wait_on_page_writeback(page);
 			}
 
@@ -4173,9 +3213,8 @@ retry:
 		 * page in our current bio, and thus deadlock, so flush the
 		 * write bio here.
 		 */
-		ret = flush_write_bio(epd);
-		if (!ret)
-			goto retry;
+		submit_write_bio(epd, 0);
+		goto retry;
 	}
 
 	if (wbc->range_cyclic || (wbc->nr_to_write > 0 && range_whole))
@@ -4185,47 +3224,28 @@ retry:
 	return ret;
 }
 
-int extent_write_full_page(struct page *page, struct writeback_control *wbc)
-{
-	int ret;
-	struct extent_page_data epd = {
-		.bio = NULL,
-		.tree = &BTRFS_I(page->mapping->host)->io_tree,
-		.extent_locked = 0,
-		.sync_io = wbc->sync_mode == WB_SYNC_ALL,
-	};
-
-	ret = __extent_writepage(page, wbc, &epd);
-	ASSERT(ret <= 0);
-	if (ret < 0) {
-		end_write_bio(&epd, ret);
-		return ret;
-	}
-
-	ret = flush_write_bio(&epd);
-	ASSERT(ret <= 0);
-	return ret;
-}
-
-int extent_write_locked_range(struct inode *inode, u64 start, u64 end,
-			      int mode)
+/*
+ * Submit the pages in the range to bio for call sites which delalloc range has
+ * already been ran (aka, ordered extent inserted) and all pages are still
+ * locked.
+ */
+int extent_write_locked_range(struct inode *inode, u64 start, u64 end)
 {
+	bool found_error = false;
+	int first_error = 0;
 	int ret = 0;
 	struct address_space *mapping = inode->i_mapping;
-	struct extent_io_tree *tree = &BTRFS_I(inode)->io_tree;
 	struct page *page;
-	unsigned long nr_pages = (end - start + PAGE_SIZE) >>
-		PAGE_SHIFT;
-
+	u64 cur = start;
+	unsigned long nr_pages;
+	const u32 sectorsize = btrfs_sb(inode->i_sb)->sectorsize;
 	struct extent_page_data epd = {
-		.bio = NULL,
-		.tree = tree,
+		.bio_ctrl = { 0 },
 		.extent_locked = 1,
-		.sync_io = mode == WB_SYNC_ALL,
+		.sync_io = 1,
 	};
 	struct writeback_control wbc_writepages = {
-		.sync_mode	= mode,
-		.nr_to_write	= nr_pages * 2,
+		.sync_mode	= WB_SYNC_ALL,
 		.range_start	= start,
 		.range_end	= end + 1,
 		/* We're called from an async helper function */
@@ -4233,125 +3253,119 @@ int extent_write_locked_range(struct inode *inode, u64 start, u64 end,
 		.no_cgroup_owner = 1,
 	};
 
+	ASSERT(IS_ALIGNED(start, sectorsize) && IS_ALIGNED(end + 1, sectorsize));
+	nr_pages = (round_up(end, PAGE_SIZE) - round_down(start, PAGE_SIZE)) >>
+		   PAGE_SHIFT;
+	wbc_writepages.nr_to_write = nr_pages * 2;
+
 	wbc_attach_fdatawrite_inode(&wbc_writepages, inode);
-	while (start <= end) {
-		page = find_get_page(mapping, start >> PAGE_SHIFT);
-		if (clear_page_dirty_for_io(page))
-			ret = __extent_writepage(page, &wbc_writepages, &epd);
-		else {
-			btrfs_writepage_endio_finish_ordered(page, start,
-						    start + PAGE_SIZE - 1, 1);
-			unlock_page(page);
+	while (cur <= end) {
+		u64 cur_end = min(round_down(cur, PAGE_SIZE) + PAGE_SIZE - 1, end);
+
+		page = find_get_page(mapping, cur >> PAGE_SHIFT);
+		/*
+		 * All pages in the range are locked since
+		 * btrfs_run_delalloc_range(), thus there is no way to clear
+		 * the page dirty flag.
+		 */
+		ASSERT(PageLocked(page));
+		ASSERT(PageDirty(page));
+		clear_page_dirty_for_io(page);
+		ret = __extent_writepage(page, &wbc_writepages, &epd);
+		ASSERT(ret <= 0);
+		if (ret < 0) {
+			found_error = true;
+			first_error = ret;
 		}
 		put_page(page);
-		start += PAGE_SIZE;
+		cur = cur_end + 1;
 	}
 
-	ASSERT(ret <= 0);
-	if (ret == 0)
-		ret = flush_write_bio(&epd);
-	else
-		end_write_bio(&epd, ret);
+	submit_write_bio(&epd, found_error ? ret : 0);
 
 	wbc_detach_inode(&wbc_writepages);
+	if (found_error)
+		return first_error;
 	return ret;
 }
 
 int extent_writepages(struct address_space *mapping,
 		      struct writeback_control *wbc)
 {
+	struct inode *inode = mapping->host;
 	int ret = 0;
 	struct extent_page_data epd = {
-		.bio = NULL,
-		.tree = &BTRFS_I(mapping->host)->io_tree,
+		.bio_ctrl = { 0 },
 		.extent_locked = 0,
 		.sync_io = wbc->sync_mode == WB_SYNC_ALL,
 	};
 
+	/*
+	 * Allow only a single thread to do the reloc work in zoned mode to
+	 * protect the write pointer updates.
+	 */
+	btrfs_zoned_data_reloc_lock(BTRFS_I(inode));
 	ret = extent_write_cache_pages(mapping, wbc, &epd);
-	ASSERT(ret <= 0);
-	if (ret < 0) {
-		end_write_bio(&epd, ret);
-		return ret;
-	}
-	ret = flush_write_bio(&epd);
+	submit_write_bio(&epd, ret);
+	btrfs_zoned_data_reloc_unlock(BTRFS_I(inode));
 	return ret;
 }
 
-int extent_readpages(struct address_space *mapping, struct list_head *pages,
-		     unsigned nr_pages)
+void extent_readahead(struct readahead_control *rac)
 {
-	struct bio *bio = NULL;
-	unsigned long bio_flags = 0;
+	struct btrfs_bio_ctrl bio_ctrl = { 0 };
 	struct page *pagepool[16];
 	struct extent_map *em_cached = NULL;
-	struct extent_io_tree *tree = &BTRFS_I(mapping->host)->io_tree;
-	int nr = 0;
 	u64 prev_em_start = (u64)-1;
+	int nr;
 
-	while (!list_empty(pages)) {
-		u64 contig_end = 0;
-
-		for (nr = 0; nr < ARRAY_SIZE(pagepool) && !list_empty(pages);) {
-			struct page *page = lru_to_page(pages);
-
-			prefetchw(&page->flags);
-			list_del(&page->lru);
-			if (add_to_page_cache_lru(page, mapping, page->index,
-						readahead_gfp_mask(mapping))) {
-				put_page(page);
-				break;
-			}
+	while ((nr = readahead_page_batch(rac, pagepool))) {
+		u64 contig_start = readahead_pos(rac);
+		u64 contig_end = contig_start + readahead_batch_length(rac) - 1;
 
-			pagepool[nr++] = page;
-			contig_end = page_offset(page) + PAGE_SIZE - 1;
-		}
-
-		if (nr) {
-			u64 contig_start = page_offset(pagepool[0]);
-
-			ASSERT(contig_start + nr * PAGE_SIZE - 1 == contig_end);
-
-			contiguous_readpages(tree, pagepool, nr, contig_start,
-				     contig_end, &em_cached, &bio, &bio_flags,
-				     &prev_em_start);
-		}
+		contiguous_readpages(pagepool, nr, contig_start, contig_end,
+				&em_cached, &bio_ctrl, &prev_em_start);
 	}
 
 	if (em_cached)
 		free_extent_map(em_cached);
-
-	if (bio)
-		return submit_one_bio(bio, 0, bio_flags);
-	return 0;
+	submit_one_bio(&bio_ctrl);
 }
 
 /*
- * basic invalidatepage code, this waits on any locked or writeback
- * ranges corresponding to the page, and then deletes any extent state
+ * basic invalidate_folio code, this waits on any locked or writeback
+ * ranges corresponding to the folio, and then deletes any extent state
  * records from the tree
  */
-int extent_invalidatepage(struct extent_io_tree *tree,
-			  struct page *page, unsigned long offset)
+int extent_invalidate_folio(struct extent_io_tree *tree,
+			  struct folio *folio, size_t offset)
 {
 	struct extent_state *cached_state = NULL;
-	u64 start = page_offset(page);
-	u64 end = start + PAGE_SIZE - 1;
-	size_t blocksize = page->mapping->host->i_sb->s_blocksize;
+	u64 start = folio_pos(folio);
+	u64 end = start + folio_size(folio) - 1;
+	size_t blocksize = folio->mapping->host->i_sb->s_blocksize;
+
+	/* This function is only called for the btree inode */
+	ASSERT(tree->owner == IO_TREE_BTREE_INODE_IO);
 
 	start += ALIGN(offset, blocksize);
 	if (start > end)
 		return 0;
 
-	lock_extent_bits(tree, start, end, &cached_state);
-	wait_on_page_writeback(page);
-	clear_extent_bit(tree, start, end, EXTENT_LOCKED | EXTENT_DELALLOC |
-			 EXTENT_DO_ACCOUNTING, 1, 1, &cached_state);
+	lock_extent(tree, start, end, &cached_state);
+	folio_wait_writeback(folio);
+
+	/*
+	 * Currently for btree io tree, only EXTENT_LOCKED is utilized,
+	 * so here we only need to unlock the extent range to free any
+	 * existing extent state.
+	 */
+	unlock_extent(tree, start, end, &cached_state);
 	return 0;
 }
 
 /*
- * a helper for releasepage, this tests for areas of the page that
+ * a helper for release_folio, this tests for areas of the page that
  * are locked or under IO and drops the related state bits if it is safe
  * to drop the page.
  */
@@ -4365,13 +3379,17 @@ static int try_release_extent_state(struct extent_io_tree *tree,
 	if (test_range_bit(tree, start, end, EXTENT_LOCKED, 0, NULL)) {
 		ret = 0;
 	} else {
+		u32 clear_bits = ~(EXTENT_LOCKED | EXTENT_NODATASUM |
+				   EXTENT_DELALLOC_NEW | EXTENT_CTLBITS);
+
 		/*
-		 * at this point we can safely clear everything except the
-		 * locked bit and the nodatasum bit
+		 * At this point we can safely clear everything except the
+		 * locked bit, the nodatasum bit and the delalloc new bit.
+		 * The delalloc new bit will be cleared by ordered extent
+		 * completion.
 		 */
-		ret = __clear_extent_bit(tree, start, end,
-				 ~(EXTENT_LOCKED | EXTENT_NODATASUM),
-				 0, 0, NULL, mask, NULL);
+		ret = __clear_extent_bit(tree, start, end, clear_bits, NULL,
+					 mask, NULL);
 
 		/* if clear_extent_bit failed for enomem reasons,
 		 * we can't allow the release to continue.
@@ -4385,7 +3403,7 @@ static int try_release_extent_state(struct extent_io_tree *tree,
 }
 
 /*
- * a helper for releasepage.  As long as there are no locked extents
+ * a helper for release_folio.  As long as there are no locked extents
  * in the range corresponding to the page, both state records and extent
  * map records are removed
  */
@@ -4402,6 +3420,9 @@ int try_release_extent_mapping(struct page *page, gfp_t mask)
 	    page->mapping->host->i_size > SZ_16M) {
 		u64 len;
 		while (start <= end) {
+			struct btrfs_fs_info *fs_info;
+			u64 cur_gen;
+
 			len = end - start + 1;
 			write_lock(&map->lock);
 			em = lookup_extent_mapping(map, start, len);
@@ -4415,62 +3436,58 @@ int try_release_extent_mapping(struct page *page, gfp_t mask)
 				free_extent_map(em);
 				break;
 			}
-			if (!test_range_bit(tree, em->start,
-					    extent_map_end(em) - 1,
-					    EXTENT_LOCKED, 0, NULL)) {
-				set_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
-					&btrfs_inode->runtime_flags);
-				remove_extent_mapping(map, em);
-				/* once for the rb tree */
-				free_extent_map(em);
-			}
+			if (test_range_bit(tree, em->start,
+					   extent_map_end(em) - 1,
+					   EXTENT_LOCKED, 0, NULL))
+				goto next;
+			/*
+			 * If it's not in the list of modified extents, used
+			 * by a fast fsync, we can remove it. If it's being
+			 * logged we can safely remove it since fsync took an
+			 * extra reference on the em.
+			 */
+			if (list_empty(&em->list) ||
+			    test_bit(EXTENT_FLAG_LOGGING, &em->flags))
+				goto remove_em;
+			/*
+			 * If it's in the list of modified extents, remove it
+			 * only if its generation is older then the current one,
+			 * in which case we don't need it for a fast fsync.
+			 * Otherwise don't remove it, we could be racing with an
+			 * ongoing fast fsync that could miss the new extent.
+			 */
+			fs_info = btrfs_inode->root->fs_info;
+			spin_lock(&fs_info->trans_lock);
+			cur_gen = fs_info->generation;
+			spin_unlock(&fs_info->trans_lock);
+			if (em->generation >= cur_gen)
+				goto next;
+remove_em:
+			/*
+			 * We only remove extent maps that are not in the list of
+			 * modified extents or that are in the list but with a
+			 * generation lower then the current generation, so there
+			 * is no need to set the full fsync flag on the inode (it
+			 * hurts the fsync performance for workloads with a data
+			 * size that exceeds or is close to the system's memory).
+			 */
+			remove_extent_mapping(map, em);
+			/* once for the rb tree */
+			free_extent_map(em);
+next:
 			start = extent_map_end(em);
 			write_unlock(&map->lock);
 
 			/* once for us */
 			free_extent_map(em);
+
+			cond_resched(); /* Allow large-extent preemption. */
 		}
 	}
 	return try_release_extent_state(tree, page, mask);
 }
 
 /*
- * helper function for fiemap, which doesn't want to see any holes.
- * This maps until we find something past 'last'
- */
-static struct extent_map *get_extent_skip_holes(struct inode *inode,
-						u64 offset, u64 last)
-{
-	u64 sectorsize = btrfs_inode_sectorsize(inode);
-	struct extent_map *em;
-	u64 len;
-
-	if (offset >= last)
-		return NULL;
-
-	while (1) {
-		len = last - offset;
-		if (len == 0)
-			break;
-		len = ALIGN(len, sectorsize);
-		em = btrfs_get_extent_fiemap(BTRFS_I(inode), offset, len);
-		if (IS_ERR_OR_NULL(em))
-			return em;
-
-		/* if this isn't a hole return it */
-		if (em->block_start != EXTENT_MAP_HOLE)
-			return em;
-
-		/* this is a hole, advance to the next extent */
-		offset = extent_map_end(em);
-		free_extent_map(em);
-		if (offset >= last)
-			break;
-	}
-	return NULL;
-}
-
-/*
  * To cache previous fiemap extent
  *
  * Will be used for merging fiemap extent
@@ -4499,6 +3516,9 @@ static int emit_fiemap_extent(struct fiemap_extent_info *fieinfo,
 {
 	int ret = 0;
 
+	/* Set at the end of extent_fiemap(). */
+	ASSERT((flags & FIEMAP_EXTENT_LAST) == 0);
+
 	if (!cache->cached)
 		goto assign;
 
@@ -4522,16 +3542,13 @@ static int emit_fiemap_extent(struct fiemap_extent_info *fieinfo,
 	 *    So truly compressed (physical size smaller than logical size)
 	 *    extents won't get merged with each other
 	 *
-	 * 3) Share same flags except FIEMAP_EXTENT_LAST
-	 *    So regular extent won't get merged with prealloc extent
+	 * 3) Share same flags
 	 */
 	if (cache->offset + cache->len  == offset &&
 	    cache->phys + cache->len == phys  &&
-	    (cache->flags & ~FIEMAP_EXTENT_LAST) ==
-			(flags & ~FIEMAP_EXTENT_LAST)) {
+	    cache->flags == flags) {
 		cache->len += len;
-		cache->flags |= flags;
-		goto try_submit_last;
+		return 0;
 	}
 
 	/* Not mergeable, need to submit cached one */
@@ -4546,13 +3563,8 @@ assign:
 	cache->phys = phys;
 	cache->len = len;
 	cache->flags = flags;
-try_submit_last:
-	if (cache->flags & FIEMAP_EXTENT_LAST) {
-		ret = fiemap_fill_next_extent(fieinfo, cache->offset,
-				cache->phys, cache->len, cache->flags);
-		cache->cached = false;
-	}
-	return ret;
+
+	return 0;
 }
 
 /*
@@ -4582,212 +3594,534 @@ static int emit_last_fiemap_cache(struct fiemap_extent_info *fieinfo,
 	return ret;
 }
 
-int extent_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
-		__u64 start, __u64 len)
+static int fiemap_next_leaf_item(struct btrfs_inode *inode, struct btrfs_path *path)
 {
-	int ret = 0;
-	u64 off = start;
-	u64 max = start + len;
-	u32 flags = 0;
-	u32 found_type;
-	u64 last;
-	u64 last_for_get_extent = 0;
-	u64 disko = 0;
-	u64 isize = i_size_read(inode);
-	struct btrfs_key found_key;
-	struct extent_map *em = NULL;
-	struct extent_state *cached_state = NULL;
-	struct btrfs_path *path;
-	struct btrfs_root *root = BTRFS_I(inode)->root;
-	struct fiemap_cache cache = { 0 };
-	struct ulist *roots;
-	struct ulist *tmp_ulist;
-	int end = 0;
-	u64 em_start = 0;
-	u64 em_len = 0;
-	u64 em_end = 0;
+	struct extent_buffer *clone;
+	struct btrfs_key key;
+	int slot;
+	int ret;
 
-	if (len == 0)
-		return -EINVAL;
+	path->slots[0]++;
+	if (path->slots[0] < btrfs_header_nritems(path->nodes[0]))
+		return 0;
 
-	path = btrfs_alloc_path();
-	if (!path)
+	ret = btrfs_next_leaf(inode->root, path);
+	if (ret != 0)
+		return ret;
+
+	/*
+	 * Don't bother with cloning if there are no more file extent items for
+	 * our inode.
+	 */
+	btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
+	if (key.objectid != btrfs_ino(inode) || key.type != BTRFS_EXTENT_DATA_KEY)
+		return 1;
+
+	/* See the comment at fiemap_search_slot() about why we clone. */
+	clone = btrfs_clone_extent_buffer(path->nodes[0]);
+	if (!clone)
 		return -ENOMEM;
-	path->leave_spinning = 1;
 
-	roots = ulist_alloc(GFP_KERNEL);
-	tmp_ulist = ulist_alloc(GFP_KERNEL);
-	if (!roots || !tmp_ulist) {
-		ret = -ENOMEM;
-		goto out_free_ulist;
+	slot = path->slots[0];
+	btrfs_release_path(path);
+	path->nodes[0] = clone;
+	path->slots[0] = slot;
+
+	return 0;
+}
+
+/*
+ * Search for the first file extent item that starts at a given file offset or
+ * the one that starts immediately before that offset.
+ * Returns: 0 on success, < 0 on error, 1 if not found.
+ */
+static int fiemap_search_slot(struct btrfs_inode *inode, struct btrfs_path *path,
+			      u64 file_offset)
+{
+	const u64 ino = btrfs_ino(inode);
+	struct btrfs_root *root = inode->root;
+	struct extent_buffer *clone;
+	struct btrfs_key key;
+	int slot;
+	int ret;
+
+	key.objectid = ino;
+	key.type = BTRFS_EXTENT_DATA_KEY;
+	key.offset = file_offset;
+
+	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
+	if (ret < 0)
+		return ret;
+
+	if (ret > 0 && path->slots[0] > 0) {
+		btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0] - 1);
+		if (key.objectid == ino && key.type == BTRFS_EXTENT_DATA_KEY)
+			path->slots[0]--;
 	}
 
-	start = round_down(start, btrfs_inode_sectorsize(inode));
-	len = round_up(max, btrfs_inode_sectorsize(inode)) - start;
+	if (path->slots[0] >= btrfs_header_nritems(path->nodes[0])) {
+		ret = btrfs_next_leaf(root, path);
+		if (ret != 0)
+			return ret;
+
+		btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
+		if (key.objectid != ino || key.type != BTRFS_EXTENT_DATA_KEY)
+			return 1;
+	}
 
 	/*
-	 * lookup the last file extent.  We're not using i_size here
-	 * because there might be preallocation past i_size
+	 * We clone the leaf and use it during fiemap. This is because while
+	 * using the leaf we do expensive things like checking if an extent is
+	 * shared, which can take a long time. In order to prevent blocking
+	 * other tasks for too long, we use a clone of the leaf. We have locked
+	 * the file range in the inode's io tree, so we know none of our file
+	 * extent items can change. This way we avoid blocking other tasks that
+	 * want to insert items for other inodes in the same leaf or b+tree
+	 * rebalance operations (triggered for example when someone is trying
+	 * to push items into this leaf when trying to insert an item in a
+	 * neighbour leaf).
+	 * We also need the private clone because holding a read lock on an
+	 * extent buffer of the subvolume's b+tree will make lockdep unhappy
+	 * when we call fiemap_fill_next_extent(), because that may cause a page
+	 * fault when filling the user space buffer with fiemap data.
 	 */
-	ret = btrfs_lookup_file_extent(NULL, root, path,
-			btrfs_ino(BTRFS_I(inode)), -1, 0);
-	if (ret < 0) {
-		goto out_free_ulist;
-	} else {
-		WARN_ON(!ret);
-		if (ret == 1)
-			ret = 0;
-	}
+	clone = btrfs_clone_extent_buffer(path->nodes[0]);
+	if (!clone)
+		return -ENOMEM;
+
+	slot = path->slots[0];
+	btrfs_release_path(path);
+	path->nodes[0] = clone;
+	path->slots[0] = slot;
+
+	return 0;
+}
+
+/*
+ * Process a range which is a hole or a prealloc extent in the inode's subvolume
+ * btree. If @disk_bytenr is 0, we are dealing with a hole, otherwise a prealloc
+ * extent. The end offset (@end) is inclusive.
+ */
+static int fiemap_process_hole(struct btrfs_inode *inode,
+			       struct fiemap_extent_info *fieinfo,
+			       struct fiemap_cache *cache,
+			       struct btrfs_backref_shared_cache *backref_cache,
+			       u64 disk_bytenr, u64 extent_offset,
+			       u64 extent_gen,
+			       struct ulist *roots, struct ulist *tmp_ulist,
+			       u64 start, u64 end)
+{
+	const u64 i_size = i_size_read(&inode->vfs_inode);
+	const u64 ino = btrfs_ino(inode);
+	u64 cur_offset = start;
+	u64 last_delalloc_end = 0;
+	u32 prealloc_flags = FIEMAP_EXTENT_UNWRITTEN;
+	bool checked_extent_shared = false;
+	int ret;
+
+	/*
+	 * There can be no delalloc past i_size, so don't waste time looking for
+	 * it beyond i_size.
+	 */
+	while (cur_offset < end && cur_offset < i_size) {
+		u64 delalloc_start;
+		u64 delalloc_end;
+		u64 prealloc_start;
+		u64 prealloc_len = 0;
+		bool delalloc;
+
+		delalloc = btrfs_find_delalloc_in_range(inode, cur_offset, end,
+							&delalloc_start,
+							&delalloc_end);
+		if (!delalloc)
+			break;
 
-	path->slots[0]--;
-	btrfs_item_key_to_cpu(path->nodes[0], &found_key, path->slots[0]);
-	found_type = found_key.type;
-
-	/* No extents, but there might be delalloc bits */
-	if (found_key.objectid != btrfs_ino(BTRFS_I(inode)) ||
-	    found_type != BTRFS_EXTENT_DATA_KEY) {
-		/* have to trust i_size as the end */
-		last = (u64)-1;
-		last_for_get_extent = isize;
-	} else {
 		/*
-		 * remember the start of the last extent.  There are a
-		 * bunch of different factors that go into the length of the
-		 * extent, so its much less complex to remember where it started
+		 * If this is a prealloc extent we have to report every section
+		 * of it that has no delalloc.
 		 */
-		last = found_key.offset;
-		last_for_get_extent = last + 1;
+		if (disk_bytenr != 0) {
+			if (last_delalloc_end == 0) {
+				prealloc_start = start;
+				prealloc_len = delalloc_start - start;
+			} else {
+				prealloc_start = last_delalloc_end + 1;
+				prealloc_len = delalloc_start - prealloc_start;
+			}
+		}
+
+		if (prealloc_len > 0) {
+			if (!checked_extent_shared && fieinfo->fi_extents_max) {
+				ret = btrfs_is_data_extent_shared(inode->root,
+							  ino, disk_bytenr,
+							  extent_gen, roots,
+							  tmp_ulist,
+							  backref_cache);
+				if (ret < 0)
+					return ret;
+				else if (ret > 0)
+					prealloc_flags |= FIEMAP_EXTENT_SHARED;
+
+				checked_extent_shared = true;
+			}
+			ret = emit_fiemap_extent(fieinfo, cache, prealloc_start,
+						 disk_bytenr + extent_offset,
+						 prealloc_len, prealloc_flags);
+			if (ret)
+				return ret;
+			extent_offset += prealloc_len;
+		}
+
+		ret = emit_fiemap_extent(fieinfo, cache, delalloc_start, 0,
+					 delalloc_end + 1 - delalloc_start,
+					 FIEMAP_EXTENT_DELALLOC |
+					 FIEMAP_EXTENT_UNKNOWN);
+		if (ret)
+			return ret;
+
+		last_delalloc_end = delalloc_end;
+		cur_offset = delalloc_end + 1;
+		extent_offset += cur_offset - delalloc_start;
+		cond_resched();
 	}
-	btrfs_release_path(path);
 
 	/*
-	 * we might have some extents allocated but more delalloc past those
-	 * extents.  so, we trust isize unless the start of the last extent is
-	 * beyond isize
+	 * Either we found no delalloc for the whole prealloc extent or we have
+	 * a prealloc extent that spans i_size or starts at or after i_size.
 	 */
-	if (last < isize) {
-		last = (u64)-1;
-		last_for_get_extent = isize;
+	if (disk_bytenr != 0 && last_delalloc_end < end) {
+		u64 prealloc_start;
+		u64 prealloc_len;
+
+		if (last_delalloc_end == 0) {
+			prealloc_start = start;
+			prealloc_len = end + 1 - start;
+		} else {
+			prealloc_start = last_delalloc_end + 1;
+			prealloc_len = end + 1 - prealloc_start;
+		}
+
+		if (!checked_extent_shared && fieinfo->fi_extents_max) {
+			ret = btrfs_is_data_extent_shared(inode->root,
+							  ino, disk_bytenr,
+							  extent_gen, roots,
+							  tmp_ulist,
+							  backref_cache);
+			if (ret < 0)
+				return ret;
+			else if (ret > 0)
+				prealloc_flags |= FIEMAP_EXTENT_SHARED;
+		}
+		ret = emit_fiemap_extent(fieinfo, cache, prealloc_start,
+					 disk_bytenr + extent_offset,
+					 prealloc_len, prealloc_flags);
+		if (ret)
+			return ret;
 	}
 
-	lock_extent_bits(&BTRFS_I(inode)->io_tree, start, start + len - 1,
-			 &cached_state);
+	return 0;
+}
 
-	em = get_extent_skip_holes(inode, start, last_for_get_extent);
-	if (!em)
-		goto out;
-	if (IS_ERR(em)) {
-		ret = PTR_ERR(em);
+static int fiemap_find_last_extent_offset(struct btrfs_inode *inode,
+					  struct btrfs_path *path,
+					  u64 *last_extent_end_ret)
+{
+	const u64 ino = btrfs_ino(inode);
+	struct btrfs_root *root = inode->root;
+	struct extent_buffer *leaf;
+	struct btrfs_file_extent_item *ei;
+	struct btrfs_key key;
+	u64 disk_bytenr;
+	int ret;
+
+	/*
+	 * Lookup the last file extent. We're not using i_size here because
+	 * there might be preallocation past i_size.
+	 */
+	ret = btrfs_lookup_file_extent(NULL, root, path, ino, (u64)-1, 0);
+	/* There can't be a file extent item at offset (u64)-1 */
+	ASSERT(ret != 0);
+	if (ret < 0)
+		return ret;
+
+	/*
+	 * For a non-existing key, btrfs_search_slot() always leaves us at a
+	 * slot > 0, except if the btree is empty, which is impossible because
+	 * at least it has the inode item for this inode and all the items for
+	 * the root inode 256.
+	 */
+	ASSERT(path->slots[0] > 0);
+	path->slots[0]--;
+	leaf = path->nodes[0];
+	btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
+	if (key.objectid != ino || key.type != BTRFS_EXTENT_DATA_KEY) {
+		/* No file extent items in the subvolume tree. */
+		*last_extent_end_ret = 0;
+		return 0;
+	}
+
+	/*
+	 * For an inline extent, the disk_bytenr is where inline data starts at,
+	 * so first check if we have an inline extent item before checking if we
+	 * have an implicit hole (disk_bytenr == 0).
+	 */
+	ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_file_extent_item);
+	if (btrfs_file_extent_type(leaf, ei) == BTRFS_FILE_EXTENT_INLINE) {
+		*last_extent_end_ret = btrfs_file_extent_end(path);
+		return 0;
+	}
+
+	/*
+	 * Find the last file extent item that is not a hole (when NO_HOLES is
+	 * not enabled). This should take at most 2 iterations in the worst
+	 * case: we have one hole file extent item at slot 0 of a leaf and
+	 * another hole file extent item as the last item in the previous leaf.
+	 * This is because we merge file extent items that represent holes.
+	 */
+	disk_bytenr = btrfs_file_extent_disk_bytenr(leaf, ei);
+	while (disk_bytenr == 0) {
+		ret = btrfs_previous_item(root, path, ino, BTRFS_EXTENT_DATA_KEY);
+		if (ret < 0) {
+			return ret;
+		} else if (ret > 0) {
+			/* No file extent items that are not holes. */
+			*last_extent_end_ret = 0;
+			return 0;
+		}
+		leaf = path->nodes[0];
+		ei = btrfs_item_ptr(leaf, path->slots[0],
+				    struct btrfs_file_extent_item);
+		disk_bytenr = btrfs_file_extent_disk_bytenr(leaf, ei);
+	}
+
+	*last_extent_end_ret = btrfs_file_extent_end(path);
+	return 0;
+}
+
+int extent_fiemap(struct btrfs_inode *inode, struct fiemap_extent_info *fieinfo,
+		  u64 start, u64 len)
+{
+	const u64 ino = btrfs_ino(inode);
+	struct extent_state *cached_state = NULL;
+	struct btrfs_path *path;
+	struct btrfs_root *root = inode->root;
+	struct fiemap_cache cache = { 0 };
+	struct btrfs_backref_shared_cache *backref_cache;
+	struct ulist *roots;
+	struct ulist *tmp_ulist;
+	u64 last_extent_end;
+	u64 prev_extent_end;
+	u64 lockstart;
+	u64 lockend;
+	bool stopped = false;
+	int ret;
+
+	backref_cache = kzalloc(sizeof(*backref_cache), GFP_KERNEL);
+	path = btrfs_alloc_path();
+	roots = ulist_alloc(GFP_KERNEL);
+	tmp_ulist = ulist_alloc(GFP_KERNEL);
+	if (!backref_cache || !path || !roots || !tmp_ulist) {
+		ret = -ENOMEM;
 		goto out;
 	}
 
-	while (!end) {
-		u64 offset_in_extent = 0;
+	lockstart = round_down(start, root->fs_info->sectorsize);
+	lockend = round_up(start + len, root->fs_info->sectorsize);
+	prev_extent_end = lockstart;
 
-		/* break if the extent we found is outside the range */
-		if (em->start >= max || extent_map_end(em) < off)
-			break;
+	lock_extent(&inode->io_tree, lockstart, lockend, &cached_state);
 
-		/*
-		 * get_extent may return an extent that starts before our
-		 * requested range.  We have to make sure the ranges
-		 * we return to fiemap always move forward and don't
-		 * overlap, so adjust the offsets here
-		 */
-		em_start = max(em->start, off);
+	ret = fiemap_find_last_extent_offset(inode, path, &last_extent_end);
+	if (ret < 0)
+		goto out_unlock;
+	btrfs_release_path(path);
 
+	path->reada = READA_FORWARD;
+	ret = fiemap_search_slot(inode, path, lockstart);
+	if (ret < 0) {
+		goto out_unlock;
+	} else if (ret > 0) {
 		/*
-		 * record the offset from the start of the extent
-		 * for adjusting the disk offset below.  Only do this if the
-		 * extent isn't compressed since our in ram offset may be past
-		 * what we have actually allocated on disk.
+		 * No file extent item found, but we may have delalloc between
+		 * the current offset and i_size. So check for that.
 		 */
-		if (!test_bit(EXTENT_FLAG_COMPRESSED, &em->flags))
-			offset_in_extent = em_start - em->start;
-		em_end = extent_map_end(em);
-		em_len = em_end - em_start;
-		flags = 0;
-		if (em->block_start < EXTENT_MAP_LAST_BYTE)
-			disko = em->block_start + offset_in_extent;
-		else
-			disko = 0;
+		ret = 0;
+		goto check_eof_delalloc;
+	}
+
+	while (prev_extent_end < lockend) {
+		struct extent_buffer *leaf = path->nodes[0];
+		struct btrfs_file_extent_item *ei;
+		struct btrfs_key key;
+		u64 extent_end;
+		u64 extent_len;
+		u64 extent_offset = 0;
+		u64 extent_gen;
+		u64 disk_bytenr = 0;
+		u64 flags = 0;
+		int extent_type;
+		u8 compression;
+
+		btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
+		if (key.objectid != ino || key.type != BTRFS_EXTENT_DATA_KEY)
+			break;
+
+		extent_end = btrfs_file_extent_end(path);
 
 		/*
-		 * bump off for our next call to get_extent
+		 * The first iteration can leave us at an extent item that ends
+		 * before our range's start. Move to the next item.
 		 */
-		off = extent_map_end(em);
-		if (off >= max)
-			end = 1;
-
-		if (em->block_start == EXTENT_MAP_LAST_BYTE) {
-			end = 1;
-			flags |= FIEMAP_EXTENT_LAST;
-		} else if (em->block_start == EXTENT_MAP_INLINE) {
-			flags |= (FIEMAP_EXTENT_DATA_INLINE |
-				  FIEMAP_EXTENT_NOT_ALIGNED);
-		} else if (em->block_start == EXTENT_MAP_DELALLOC) {
-			flags |= (FIEMAP_EXTENT_DELALLOC |
-				  FIEMAP_EXTENT_UNKNOWN);
-		} else if (fieinfo->fi_extents_max) {
-			u64 bytenr = em->block_start -
-				(em->start - em->orig_start);
+		if (extent_end <= lockstart)
+			goto next_item;
 
-			/*
-			 * As btrfs supports shared space, this information
-			 * can be exported to userspace tools via
-			 * flag FIEMAP_EXTENT_SHARED.  If fi_extents_max == 0
-			 * then we're just getting a count and we can skip the
-			 * lookup stuff.
-			 */
-			ret = btrfs_check_shared(root,
-						 btrfs_ino(BTRFS_I(inode)),
-						 bytenr, roots, tmp_ulist);
-			if (ret < 0)
-				goto out_free;
-			if (ret)
-				flags |= FIEMAP_EXTENT_SHARED;
-			ret = 0;
+		/* We have in implicit hole (NO_HOLES feature enabled). */
+		if (prev_extent_end < key.offset) {
+			const u64 range_end = min(key.offset, lockend) - 1;
+
+			ret = fiemap_process_hole(inode, fieinfo, &cache,
+						  backref_cache, 0, 0, 0,
+						  roots, tmp_ulist,
+						  prev_extent_end, range_end);
+			if (ret < 0) {
+				goto out_unlock;
+			} else if (ret > 0) {
+				/* fiemap_fill_next_extent() told us to stop. */
+				stopped = true;
+				break;
+			}
+
+			/* We've reached the end of the fiemap range, stop. */
+			if (key.offset >= lockend) {
+				stopped = true;
+				break;
+			}
 		}
-		if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags))
+
+		extent_len = extent_end - key.offset;
+		ei = btrfs_item_ptr(leaf, path->slots[0],
+				    struct btrfs_file_extent_item);
+		compression = btrfs_file_extent_compression(leaf, ei);
+		extent_type = btrfs_file_extent_type(leaf, ei);
+		extent_gen = btrfs_file_extent_generation(leaf, ei);
+
+		if (extent_type != BTRFS_FILE_EXTENT_INLINE) {
+			disk_bytenr = btrfs_file_extent_disk_bytenr(leaf, ei);
+			if (compression == BTRFS_COMPRESS_NONE)
+				extent_offset = btrfs_file_extent_offset(leaf, ei);
+		}
+
+		if (compression != BTRFS_COMPRESS_NONE)
 			flags |= FIEMAP_EXTENT_ENCODED;
-		if (test_bit(EXTENT_FLAG_PREALLOC, &em->flags))
-			flags |= FIEMAP_EXTENT_UNWRITTEN;
 
-		free_extent_map(em);
-		em = NULL;
-		if ((em_start >= last) || em_len == (u64)-1 ||
-		   (last == (u64)-1 && isize <= em_end)) {
-			flags |= FIEMAP_EXTENT_LAST;
-			end = 1;
+		if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
+			flags |= FIEMAP_EXTENT_DATA_INLINE;
+			flags |= FIEMAP_EXTENT_NOT_ALIGNED;
+			ret = emit_fiemap_extent(fieinfo, &cache, key.offset, 0,
+						 extent_len, flags);
+		} else if (extent_type == BTRFS_FILE_EXTENT_PREALLOC) {
+			ret = fiemap_process_hole(inode, fieinfo, &cache,
+						  backref_cache,
+						  disk_bytenr, extent_offset,
+						  extent_gen, roots, tmp_ulist,
+						  key.offset, extent_end - 1);
+		} else if (disk_bytenr == 0) {
+			/* We have an explicit hole. */
+			ret = fiemap_process_hole(inode, fieinfo, &cache,
+						  backref_cache, 0, 0, 0,
+						  roots, tmp_ulist,
+						  key.offset, extent_end - 1);
+		} else {
+			/* We have a regular extent. */
+			if (fieinfo->fi_extents_max) {
+				ret = btrfs_is_data_extent_shared(root, ino,
+								  disk_bytenr,
+								  extent_gen,
+								  roots,
+								  tmp_ulist,
+								  backref_cache);
+				if (ret < 0)
+					goto out_unlock;
+				else if (ret > 0)
+					flags |= FIEMAP_EXTENT_SHARED;
+			}
+
+			ret = emit_fiemap_extent(fieinfo, &cache, key.offset,
+						 disk_bytenr + extent_offset,
+						 extent_len, flags);
 		}
 
-		/* now scan forward to see if this is really the last extent. */
-		em = get_extent_skip_holes(inode, off, last_for_get_extent);
-		if (IS_ERR(em)) {
-			ret = PTR_ERR(em);
-			goto out;
+		if (ret < 0) {
+			goto out_unlock;
+		} else if (ret > 0) {
+			/* fiemap_fill_next_extent() told us to stop. */
+			stopped = true;
+			break;
 		}
-		if (!em) {
-			flags |= FIEMAP_EXTENT_LAST;
-			end = 1;
+
+		prev_extent_end = extent_end;
+next_item:
+		if (fatal_signal_pending(current)) {
+			ret = -EINTR;
+			goto out_unlock;
 		}
-		ret = emit_fiemap_extent(fieinfo, &cache, em_start, disko,
-					   em_len, flags);
-		if (ret) {
-			if (ret == 1)
-				ret = 0;
-			goto out_free;
+
+		ret = fiemap_next_leaf_item(inode, path);
+		if (ret < 0) {
+			goto out_unlock;
+		} else if (ret > 0) {
+			/* No more file extent items for this inode. */
+			break;
 		}
+		cond_resched();
 	}
-out_free:
-	if (!ret)
-		ret = emit_last_fiemap_cache(fieinfo, &cache);
-	free_extent_map(em);
-out:
-	unlock_extent_cached(&BTRFS_I(inode)->io_tree, start, start + len - 1,
-			     &cached_state);
 
-out_free_ulist:
+check_eof_delalloc:
+	/*
+	 * Release (and free) the path before emitting any final entries to
+	 * fiemap_fill_next_extent() to keep lockdep happy. This is because
+	 * once we find no more file extent items exist, we may have a
+	 * non-cloned leaf, and fiemap_fill_next_extent() can trigger page
+	 * faults when copying data to the user space buffer.
+	 */
+	btrfs_free_path(path);
+	path = NULL;
+
+	if (!stopped && prev_extent_end < lockend) {
+		ret = fiemap_process_hole(inode, fieinfo, &cache, backref_cache,
+					  0, 0, 0, roots, tmp_ulist,
+					  prev_extent_end, lockend - 1);
+		if (ret < 0)
+			goto out_unlock;
+		prev_extent_end = lockend;
+	}
+
+	if (cache.cached && cache.offset + cache.len >= last_extent_end) {
+		const u64 i_size = i_size_read(&inode->vfs_inode);
+
+		if (prev_extent_end < i_size) {
+			u64 delalloc_start;
+			u64 delalloc_end;
+			bool delalloc;
+
+			delalloc = btrfs_find_delalloc_in_range(inode,
+								prev_extent_end,
+								i_size - 1,
+								&delalloc_start,
+								&delalloc_end);
+			if (!delalloc)
+				cache.flags |= FIEMAP_EXTENT_LAST;
+		} else {
+			cache.flags |= FIEMAP_EXTENT_LAST;
+		}
+	}
+
+	ret = emit_last_fiemap_cache(fieinfo, &cache);
+
+out_unlock:
+	unlock_extent(&inode->io_tree, lockstart, lockend, &cached_state);
+out:
+	kfree(backref_cache);
 	btrfs_free_path(path);
 	ulist_free(roots);
 	ulist_free(tmp_ulist);
@@ -4796,36 +4130,55 @@ out_free_ulist:
 
 static void __free_extent_buffer(struct extent_buffer *eb)
 {
-	btrfs_leak_debug_del(&eb->leak_list);
 	kmem_cache_free(extent_buffer_cache, eb);
 }
 
-int extent_buffer_under_io(struct extent_buffer *eb)
+int extent_buffer_under_io(const struct extent_buffer *eb)
 {
 	return (atomic_read(&eb->io_pages) ||
 		test_bit(EXTENT_BUFFER_WRITEBACK, &eb->bflags) ||
 		test_bit(EXTENT_BUFFER_DIRTY, &eb->bflags));
 }
 
-/*
- * Release all pages attached to the extent buffer.
- */
-static void btrfs_release_extent_buffer_pages(struct extent_buffer *eb)
+static bool page_range_has_eb(struct btrfs_fs_info *fs_info, struct page *page)
 {
-	int i;
-	int num_pages;
-	int mapped = !test_bit(EXTENT_BUFFER_UNMAPPED, &eb->bflags);
+	struct btrfs_subpage *subpage;
 
-	BUG_ON(extent_buffer_under_io(eb));
+	lockdep_assert_held(&page->mapping->private_lock);
 
-	num_pages = num_extent_pages(eb);
-	for (i = 0; i < num_pages; i++) {
-		struct page *page = eb->pages[i];
+	if (PagePrivate(page)) {
+		subpage = (struct btrfs_subpage *)page->private;
+		if (atomic_read(&subpage->eb_refs))
+			return true;
+		/*
+		 * Even there is no eb refs here, we may still have
+		 * end_page_read() call relying on page::private.
+		 */
+		if (atomic_read(&subpage->readers))
+			return true;
+	}
+	return false;
+}
 
-		if (!page)
-			continue;
+static void detach_extent_buffer_page(struct extent_buffer *eb, struct page *page)
+{
+	struct btrfs_fs_info *fs_info = eb->fs_info;
+	const bool mapped = !test_bit(EXTENT_BUFFER_UNMAPPED, &eb->bflags);
+
+	/*
+	 * For mapped eb, we're going to change the page private, which should
+	 * be done under the private_lock.
+	 */
+	if (mapped)
+		spin_lock(&page->mapping->private_lock);
+
+	if (!PagePrivate(page)) {
 		if (mapped)
-			spin_lock(&page->mapping->private_lock);
+			spin_unlock(&page->mapping->private_lock);
+		return;
+	}
+
+	if (fs_info->nodesize >= PAGE_SIZE) {
 		/*
 		 * We do this since we'll remove the pages after we've
 		 * removed the eb from the radix tree, so we could race
@@ -4842,14 +4195,51 @@ static void btrfs_release_extent_buffer_pages(struct extent_buffer *eb)
 			 * We need to make sure we haven't be attached
 			 * to a new eb.
 			 */
-			ClearPagePrivate(page);
-			set_page_private(page, 0);
-			/* One for the page private */
-			put_page(page);
+			detach_page_private(page);
 		}
-
 		if (mapped)
 			spin_unlock(&page->mapping->private_lock);
+		return;
+	}
+
+	/*
+	 * For subpage, we can have dummy eb with page private.  In this case,
+	 * we can directly detach the private as such page is only attached to
+	 * one dummy eb, no sharing.
+	 */
+	if (!mapped) {
+		btrfs_detach_subpage(fs_info, page);
+		return;
+	}
+
+	btrfs_page_dec_eb_refs(fs_info, page);
+
+	/*
+	 * We can only detach the page private if there are no other ebs in the
+	 * page range and no unfinished IO.
+	 */
+	if (!page_range_has_eb(fs_info, page))
+		btrfs_detach_subpage(fs_info, page);
+
+	spin_unlock(&page->mapping->private_lock);
+}
+
+/* Release all pages attached to the extent buffer */
+static void btrfs_release_extent_buffer_pages(struct extent_buffer *eb)
+{
+	int i;
+	int num_pages;
+
+	ASSERT(!extent_buffer_under_io(eb));
+
+	num_pages = num_extent_pages(eb);
+	for (i = 0; i < num_pages; i++) {
+		struct page *page = eb->pages[i];
+
+		if (!page)
+			continue;
+
+		detach_extent_buffer_page(eb, page);
 
 		/* One for when we allocated the page */
 		put_page(page);
@@ -4862,6 +4252,7 @@ static void btrfs_release_extent_buffer_pages(struct extent_buffer *eb)
 static inline void btrfs_release_extent_buffer(struct extent_buffer *eb)
 {
 	btrfs_release_extent_buffer_pages(eb);
+	btrfs_leak_debug_del_eb(eb);
 	__free_extent_buffer(eb);
 }
 
@@ -4876,62 +4267,58 @@ __alloc_extent_buffer(struct btrfs_fs_info *fs_info, u64 start,
 	eb->len = len;
 	eb->fs_info = fs_info;
 	eb->bflags = 0;
-	rwlock_init(&eb->lock);
-	atomic_set(&eb->blocking_readers, 0);
-	eb->blocking_writers = 0;
-	eb->lock_nested = false;
-	init_waitqueue_head(&eb->write_lock_wq);
-	init_waitqueue_head(&eb->read_lock_wq);
+	init_rwsem(&eb->lock);
 
-	btrfs_leak_debug_add(&eb->leak_list, &buffers);
+	btrfs_leak_debug_add_eb(eb);
+	INIT_LIST_HEAD(&eb->release_list);
 
 	spin_lock_init(&eb->refs_lock);
 	atomic_set(&eb->refs, 1);
 	atomic_set(&eb->io_pages, 0);
 
-	/*
-	 * Sanity checks, currently the maximum is 64k covered by 16x 4k pages
-	 */
-	BUILD_BUG_ON(BTRFS_MAX_METADATA_BLOCKSIZE
-		> MAX_INLINE_EXTENT_BUFFER_SIZE);
-	BUG_ON(len > MAX_INLINE_EXTENT_BUFFER_SIZE);
-
-#ifdef CONFIG_BTRFS_DEBUG
-	eb->spinning_writers = 0;
-	atomic_set(&eb->spinning_readers, 0);
-	atomic_set(&eb->read_locks, 0);
-	eb->write_locks = 0;
-#endif
+	ASSERT(len <= BTRFS_MAX_METADATA_BLOCKSIZE);
 
 	return eb;
 }
 
-struct extent_buffer *btrfs_clone_extent_buffer(struct extent_buffer *src)
+struct extent_buffer *btrfs_clone_extent_buffer(const struct extent_buffer *src)
 {
 	int i;
-	struct page *p;
 	struct extent_buffer *new;
 	int num_pages = num_extent_pages(src);
+	int ret;
 
 	new = __alloc_extent_buffer(src->fs_info, src->start, src->len);
 	if (new == NULL)
 		return NULL;
 
+	/*
+	 * Set UNMAPPED before calling btrfs_release_extent_buffer(), as
+	 * btrfs_release_extent_buffer() have different behavior for
+	 * UNMAPPED subpage extent buffer.
+	 */
+	set_bit(EXTENT_BUFFER_UNMAPPED, &new->bflags);
+
+	memset(new->pages, 0, sizeof(*new->pages) * num_pages);
+	ret = btrfs_alloc_page_array(num_pages, new->pages);
+	if (ret) {
+		btrfs_release_extent_buffer(new);
+		return NULL;
+	}
+
 	for (i = 0; i < num_pages; i++) {
-		p = alloc_page(GFP_NOFS);
-		if (!p) {
+		int ret;
+		struct page *p = new->pages[i];
+
+		ret = attach_extent_buffer_page(new, p, NULL);
+		if (ret < 0) {
 			btrfs_release_extent_buffer(new);
 			return NULL;
 		}
-		attach_extent_buffer_page(new, p);
 		WARN_ON(PageDirty(p));
-		SetPageUptodate(p);
-		new->pages[i] = p;
 		copy_page(page_address(p), page_address(src->pages[i]));
 	}
-
-	set_bit(EXTENT_BUFFER_UPTODATE, &new->bflags);
-	set_bit(EXTENT_BUFFER_UNMAPPED, &new->bflags);
+	set_extent_buffer_uptodate(new);
 
 	return new;
 }
@@ -4942,25 +4329,37 @@ struct extent_buffer *__alloc_dummy_extent_buffer(struct btrfs_fs_info *fs_info,
 	struct extent_buffer *eb;
 	int num_pages;
 	int i;
+	int ret;
 
 	eb = __alloc_extent_buffer(fs_info, start, len);
 	if (!eb)
 		return NULL;
 
 	num_pages = num_extent_pages(eb);
+	ret = btrfs_alloc_page_array(num_pages, eb->pages);
+	if (ret)
+		goto err;
+
 	for (i = 0; i < num_pages; i++) {
-		eb->pages[i] = alloc_page(GFP_NOFS);
-		if (!eb->pages[i])
+		struct page *p = eb->pages[i];
+
+		ret = attach_extent_buffer_page(eb, p, NULL);
+		if (ret < 0)
 			goto err;
 	}
+
 	set_extent_buffer_uptodate(eb);
 	btrfs_set_header_nritems(eb, 0);
 	set_bit(EXTENT_BUFFER_UNMAPPED, &eb->bflags);
 
 	return eb;
 err:
-	for (; i > 0; i--)
-		__free_page(eb->pages[i - 1]);
+	for (i = 0; i < num_pages; i++) {
+		if (eb->pages[i]) {
+			detach_extent_buffer_page(eb, eb->pages[i]);
+			__free_page(eb->pages[i]);
+		}
+	}
 	__free_extent_buffer(eb);
 	return NULL;
 }
@@ -4974,25 +4373,28 @@ struct extent_buffer *alloc_dummy_extent_buffer(struct btrfs_fs_info *fs_info,
 static void check_buffer_tree_ref(struct extent_buffer *eb)
 {
 	int refs;
-	/* the ref bit is tricky.  We have to make sure it is set
-	 * if we have the buffer dirty.   Otherwise the
-	 * code to free a buffer can end up dropping a dirty
-	 * page
+	/*
+	 * The TREE_REF bit is first set when the extent_buffer is added
+	 * to the radix tree. It is also reset, if unset, when a new reference
+	 * is created by find_extent_buffer.
 	 *
-	 * Once the ref bit is set, it won't go away while the
-	 * buffer is dirty or in writeback, and it also won't
-	 * go away while we have the reference count on the
-	 * eb bumped.
+	 * It is only cleared in two cases: freeing the last non-tree
+	 * reference to the extent_buffer when its STALE bit is set or
+	 * calling release_folio when the tree reference is the only reference.
 	 *
-	 * We can't just set the ref bit without bumping the
-	 * ref on the eb because free_extent_buffer might
-	 * see the ref bit and try to clear it.  If this happens
-	 * free_extent_buffer might end up dropping our original
-	 * ref by mistake and freeing the page before we are able
-	 * to add one more ref.
+	 * In both cases, care is taken to ensure that the extent_buffer's
+	 * pages are not under io. However, release_folio can be concurrently
+	 * called with creating new references, which is prone to race
+	 * conditions between the calls to check_buffer_tree_ref in those
+	 * codepaths and clearing TREE_REF in try_release_extent_buffer.
 	 *
-	 * So bump the ref count first, then set the bit.  If someone
-	 * beat us to it, drop the ref we added.
+	 * The actual lifetime of the extent_buffer in the radix tree is
+	 * adequately protected by the refcount, but the TREE_REF bit and
+	 * its corresponding reference are not. To protect against this
+	 * class of races, we call check_buffer_tree_ref from the codepaths
+	 * which trigger io after they set eb->io_pages. Note that once io is
+	 * initiated, TREE_REF can no longer be cleared, so that is the
+	 * moment at which any such race is best fixed.
 	 */
 	refs = atomic_read(&eb->refs);
 	if (refs >= 2 && test_bit(EXTENT_BUFFER_TREE_REF, &eb->bflags))
@@ -5025,36 +4427,28 @@ struct extent_buffer *find_extent_buffer(struct btrfs_fs_info *fs_info,
 {
 	struct extent_buffer *eb;
 
-	rcu_read_lock();
-	eb = radix_tree_lookup(&fs_info->buffer_radix,
-			       start >> PAGE_SHIFT);
-	if (eb && atomic_inc_not_zero(&eb->refs)) {
-		rcu_read_unlock();
-		/*
-		 * Lock our eb's refs_lock to avoid races with
-		 * free_extent_buffer. When we get our eb it might be flagged
-		 * with EXTENT_BUFFER_STALE and another task running
-		 * free_extent_buffer might have seen that flag set,
-		 * eb->refs == 2, that the buffer isn't under IO (dirty and
-		 * writeback flags not set) and it's still in the tree (flag
-		 * EXTENT_BUFFER_TREE_REF set), therefore being in the process
-		 * of decrementing the extent buffer's reference count twice.
-		 * So here we could race and increment the eb's reference count,
-		 * clear its stale flag, mark it as dirty and drop our reference
-		 * before the other task finishes executing free_extent_buffer,
-		 * which would later result in an attempt to free an extent
-		 * buffer that is dirty.
-		 */
-		if (test_bit(EXTENT_BUFFER_STALE, &eb->bflags)) {
-			spin_lock(&eb->refs_lock);
-			spin_unlock(&eb->refs_lock);
-		}
-		mark_extent_buffer_accessed(eb, NULL);
-		return eb;
+	eb = find_extent_buffer_nolock(fs_info, start);
+	if (!eb)
+		return NULL;
+	/*
+	 * Lock our eb's refs_lock to avoid races with free_extent_buffer().
+	 * When we get our eb it might be flagged with EXTENT_BUFFER_STALE and
+	 * another task running free_extent_buffer() might have seen that flag
+	 * set, eb->refs == 2, that the buffer isn't under IO (dirty and
+	 * writeback flags not set) and it's still in the tree (flag
+	 * EXTENT_BUFFER_TREE_REF set), therefore being in the process of
+	 * decrementing the extent buffer's reference count twice.  So here we
+	 * could race and increment the eb's reference count, clear its stale
+	 * flag, mark it as dirty and drop our reference before the other task
+	 * finishes executing free_extent_buffer, which would later result in
+	 * an attempt to free an extent buffer that is dirty.
+	 */
+	if (test_bit(EXTENT_BUFFER_STALE, &eb->bflags)) {
+		spin_lock(&eb->refs_lock);
+		spin_unlock(&eb->refs_lock);
 	}
-	rcu_read_unlock();
-
-	return NULL;
+	mark_extent_buffer_accessed(eb, NULL);
+	return eb;
 }
 
 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
@@ -5079,7 +4473,7 @@ again:
 	}
 	spin_lock(&fs_info->buffer_lock);
 	ret = radix_tree_insert(&fs_info->buffer_radix,
-				start >> PAGE_SHIFT, eb);
+				start >> fs_info->sectorsize_bits, eb);
 	spin_unlock(&fs_info->buffer_lock);
 	radix_tree_preload_end();
 	if (ret == -EEXIST) {
@@ -5099,8 +4493,64 @@ free_eb:
 }
 #endif
 
+static struct extent_buffer *grab_extent_buffer(
+		struct btrfs_fs_info *fs_info, struct page *page)
+{
+	struct extent_buffer *exists;
+
+	/*
+	 * For subpage case, we completely rely on radix tree to ensure we
+	 * don't try to insert two ebs for the same bytenr.  So here we always
+	 * return NULL and just continue.
+	 */
+	if (fs_info->nodesize < PAGE_SIZE)
+		return NULL;
+
+	/* Page not yet attached to an extent buffer */
+	if (!PagePrivate(page))
+		return NULL;
+
+	/*
+	 * We could have already allocated an eb for this page and attached one
+	 * so lets see if we can get a ref on the existing eb, and if we can we
+	 * know it's good and we can just return that one, else we know we can
+	 * just overwrite page->private.
+	 */
+	exists = (struct extent_buffer *)page->private;
+	if (atomic_inc_not_zero(&exists->refs))
+		return exists;
+
+	WARN_ON(PageDirty(page));
+	detach_page_private(page);
+	return NULL;
+}
+
+static int check_eb_alignment(struct btrfs_fs_info *fs_info, u64 start)
+{
+	if (!IS_ALIGNED(start, fs_info->sectorsize)) {
+		btrfs_err(fs_info, "bad tree block start %llu", start);
+		return -EINVAL;
+	}
+
+	if (fs_info->nodesize < PAGE_SIZE &&
+	    offset_in_page(start) + fs_info->nodesize > PAGE_SIZE) {
+		btrfs_err(fs_info,
+		"tree block crosses page boundary, start %llu nodesize %u",
+			  start, fs_info->nodesize);
+		return -EINVAL;
+	}
+	if (fs_info->nodesize >= PAGE_SIZE &&
+	    !PAGE_ALIGNED(start)) {
+		btrfs_err(fs_info,
+		"tree block is not page aligned, start %llu nodesize %u",
+			  start, fs_info->nodesize);
+		return -EINVAL;
+	}
+	return 0;
+}
+
 struct extent_buffer *alloc_extent_buffer(struct btrfs_fs_info *fs_info,
-					  u64 start)
+					  u64 start, u64 owner_root, int level)
 {
 	unsigned long len = fs_info->nodesize;
 	int num_pages;
@@ -5110,13 +4560,23 @@ struct extent_buffer *alloc_extent_buffer(struct btrfs_fs_info *fs_info,
 	struct extent_buffer *exists = NULL;
 	struct page *p;
 	struct address_space *mapping = fs_info->btree_inode->i_mapping;
+	u64 lockdep_owner = owner_root;
 	int uptodate = 1;
 	int ret;
 
-	if (!IS_ALIGNED(start, fs_info->sectorsize)) {
-		btrfs_err(fs_info, "bad tree block start %llu", start);
+	if (check_eb_alignment(fs_info, start))
 		return ERR_PTR(-EINVAL);
+
+#if BITS_PER_LONG == 32
+	if (start >= MAX_LFS_FILESIZE) {
+		btrfs_err_rl(fs_info,
+		"extent buffer %llu is beyond 32bit page cache limit", start);
+		btrfs_err_32bit_limit(fs_info);
+		return ERR_PTR(-EOVERFLOW);
 	}
+	if (start >= BTRFS_32BIT_EARLY_WARN_THRESHOLD)
+		btrfs_warn_32bit_limit(fs_info);
+#endif
 
 	eb = find_extent_buffer(fs_info, start);
 	if (eb)
@@ -5126,44 +4586,72 @@ struct extent_buffer *alloc_extent_buffer(struct btrfs_fs_info *fs_info,
 	if (!eb)
 		return ERR_PTR(-ENOMEM);
 
+	/*
+	 * The reloc trees are just snapshots, so we need them to appear to be
+	 * just like any other fs tree WRT lockdep.
+	 */
+	if (lockdep_owner == BTRFS_TREE_RELOC_OBJECTID)
+		lockdep_owner = BTRFS_FS_TREE_OBJECTID;
+
+	btrfs_set_buffer_lockdep_class(lockdep_owner, eb, level);
+
 	num_pages = num_extent_pages(eb);
 	for (i = 0; i < num_pages; i++, index++) {
+		struct btrfs_subpage *prealloc = NULL;
+
 		p = find_or_create_page(mapping, index, GFP_NOFS|__GFP_NOFAIL);
 		if (!p) {
 			exists = ERR_PTR(-ENOMEM);
 			goto free_eb;
 		}
 
-		spin_lock(&mapping->private_lock);
-		if (PagePrivate(p)) {
-			/*
-			 * We could have already allocated an eb for this page
-			 * and attached one so lets see if we can get a ref on
-			 * the existing eb, and if we can we know it's good and
-			 * we can just return that one, else we know we can just
-			 * overwrite page->private.
-			 */
-			exists = (struct extent_buffer *)p->private;
-			if (atomic_inc_not_zero(&exists->refs)) {
-				spin_unlock(&mapping->private_lock);
+		/*
+		 * Preallocate page->private for subpage case, so that we won't
+		 * allocate memory with private_lock hold.  The memory will be
+		 * freed by attach_extent_buffer_page() or freed manually if
+		 * we exit earlier.
+		 *
+		 * Although we have ensured one subpage eb can only have one
+		 * page, but it may change in the future for 16K page size
+		 * support, so we still preallocate the memory in the loop.
+		 */
+		if (fs_info->nodesize < PAGE_SIZE) {
+			prealloc = btrfs_alloc_subpage(fs_info, BTRFS_SUBPAGE_METADATA);
+			if (IS_ERR(prealloc)) {
+				ret = PTR_ERR(prealloc);
 				unlock_page(p);
 				put_page(p);
-				mark_extent_buffer_accessed(exists, p);
+				exists = ERR_PTR(ret);
 				goto free_eb;
 			}
-			exists = NULL;
+		}
 
-			/*
-			 * Do this so attach doesn't complain and we need to
-			 * drop the ref the old guy had.
-			 */
-			ClearPagePrivate(p);
-			WARN_ON(PageDirty(p));
+		spin_lock(&mapping->private_lock);
+		exists = grab_extent_buffer(fs_info, p);
+		if (exists) {
+			spin_unlock(&mapping->private_lock);
+			unlock_page(p);
 			put_page(p);
+			mark_extent_buffer_accessed(exists, p);
+			btrfs_free_subpage(prealloc);
+			goto free_eb;
 		}
-		attach_extent_buffer_page(eb, p);
+		/* Should not fail, as we have preallocated the memory */
+		ret = attach_extent_buffer_page(eb, p, prealloc);
+		ASSERT(!ret);
+		/*
+		 * To inform we have extra eb under allocation, so that
+		 * detach_extent_buffer_page() won't release the page private
+		 * when the eb hasn't yet been inserted into radix tree.
+		 *
+		 * The ref will be decreased when the eb released the page, in
+		 * detach_extent_buffer_page().
+		 * Thus needs no special handling in error path.
+		 */
+		btrfs_page_inc_eb_refs(fs_info, p);
 		spin_unlock(&mapping->private_lock);
-		WARN_ON(PageDirty(p));
+
+		WARN_ON(btrfs_page_test_dirty(fs_info, p, eb->start, eb->len));
 		eb->pages[i] = p;
 		if (!PageUptodate(p))
 			uptodate = 0;
@@ -5171,7 +4659,7 @@ struct extent_buffer *alloc_extent_buffer(struct btrfs_fs_info *fs_info,
 		/*
 		 * We can't unlock the pages just yet since the extent buffer
 		 * hasn't been properly inserted in the radix tree, this
-		 * opens a race with btree_releasepage which can free a page
+		 * opens a race with btree_release_folio which can free a page
 		 * while we are still filling in all pages for the buffer and
 		 * we could crash.
 		 */
@@ -5187,7 +4675,7 @@ again:
 
 	spin_lock(&fs_info->buffer_lock);
 	ret = radix_tree_insert(&fs_info->buffer_radix,
-				start >> PAGE_SHIFT, eb);
+				start >> fs_info->sectorsize_bits, eb);
 	spin_unlock(&fs_info->buffer_lock);
 	radix_tree_preload_end();
 	if (ret == -EEXIST) {
@@ -5203,7 +4691,7 @@ again:
 
 	/*
 	 * Now it's safe to unlock the pages because any calls to
-	 * btree_releasepage will correctly detect that a page belongs to a
+	 * btree_release_folio will correctly detect that a page belongs to a
 	 * live buffer and won't free them prematurely.
 	 */
 	for (i = 0; i < num_pages; i++)
@@ -5230,6 +4718,7 @@ static inline void btrfs_release_extent_buffer_rcu(struct rcu_head *head)
 }
 
 static int release_extent_buffer(struct extent_buffer *eb)
+	__releases(&eb->refs_lock)
 {
 	lockdep_assert_held(&eb->refs_lock);
 
@@ -5242,12 +4731,13 @@ static int release_extent_buffer(struct extent_buffer *eb)
 
 			spin_lock(&fs_info->buffer_lock);
 			radix_tree_delete(&fs_info->buffer_radix,
-					  eb->start >> PAGE_SHIFT);
+					  eb->start >> fs_info->sectorsize_bits);
 			spin_unlock(&fs_info->buffer_lock);
 		} else {
 			spin_unlock(&eb->refs_lock);
 		}
 
+		btrfs_leak_debug_del_eb(eb);
 		/* Should be safe to release our pages at this point */
 		btrfs_release_extent_buffer_pages(eb);
 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
@@ -5267,18 +4757,16 @@ static int release_extent_buffer(struct extent_buffer *eb)
 void free_extent_buffer(struct extent_buffer *eb)
 {
 	int refs;
-	int old;
 	if (!eb)
 		return;
 
+	refs = atomic_read(&eb->refs);
 	while (1) {
-		refs = atomic_read(&eb->refs);
 		if ((!test_bit(EXTENT_BUFFER_UNMAPPED, &eb->bflags) && refs <= 3)
 		    || (test_bit(EXTENT_BUFFER_UNMAPPED, &eb->bflags) &&
 			refs == 1))
 			break;
-		old = atomic_cmpxchg(&eb->refs, refs, refs - 1);
-		if (old == refs)
+		if (atomic_try_cmpxchg(&eb->refs, &refs, refs - 1))
 			return;
 	}
 
@@ -5310,28 +4798,51 @@ void free_extent_buffer_stale(struct extent_buffer *eb)
 	release_extent_buffer(eb);
 }
 
-void clear_extent_buffer_dirty(struct extent_buffer *eb)
+static void btree_clear_page_dirty(struct page *page)
+{
+	ASSERT(PageDirty(page));
+	ASSERT(PageLocked(page));
+	clear_page_dirty_for_io(page);
+	xa_lock_irq(&page->mapping->i_pages);
+	if (!PageDirty(page))
+		__xa_clear_mark(&page->mapping->i_pages,
+				page_index(page), PAGECACHE_TAG_DIRTY);
+	xa_unlock_irq(&page->mapping->i_pages);
+}
+
+static void clear_subpage_extent_buffer_dirty(const struct extent_buffer *eb)
+{
+	struct btrfs_fs_info *fs_info = eb->fs_info;
+	struct page *page = eb->pages[0];
+	bool last;
+
+	/* btree_clear_page_dirty() needs page locked */
+	lock_page(page);
+	last = btrfs_subpage_clear_and_test_dirty(fs_info, page, eb->start,
+						  eb->len);
+	if (last)
+		btree_clear_page_dirty(page);
+	unlock_page(page);
+	WARN_ON(atomic_read(&eb->refs) == 0);
+}
+
+void clear_extent_buffer_dirty(const struct extent_buffer *eb)
 {
 	int i;
 	int num_pages;
 	struct page *page;
 
+	if (eb->fs_info->nodesize < PAGE_SIZE)
+		return clear_subpage_extent_buffer_dirty(eb);
+
 	num_pages = num_extent_pages(eb);
 
 	for (i = 0; i < num_pages; i++) {
 		page = eb->pages[i];
 		if (!PageDirty(page))
 			continue;
-
 		lock_page(page);
-		WARN_ON(!PagePrivate(page));
-
-		clear_page_dirty_for_io(page);
-		xa_lock_irq(&page->mapping->i_pages);
-		if (!PageDirty(page))
-			__xa_clear_mark(&page->mapping->i_pages,
-					page_index(page), PAGECACHE_TAG_DIRTY);
-		xa_unlock_irq(&page->mapping->i_pages);
+		btree_clear_page_dirty(page);
 		ClearPageError(page);
 		unlock_page(page);
 	}
@@ -5352,10 +4863,28 @@ bool set_extent_buffer_dirty(struct extent_buffer *eb)
 	WARN_ON(atomic_read(&eb->refs) == 0);
 	WARN_ON(!test_bit(EXTENT_BUFFER_TREE_REF, &eb->bflags));
 
-	if (!was_dirty)
-		for (i = 0; i < num_pages; i++)
-			set_page_dirty(eb->pages[i]);
+	if (!was_dirty) {
+		bool subpage = eb->fs_info->nodesize < PAGE_SIZE;
 
+		/*
+		 * For subpage case, we can have other extent buffers in the
+		 * same page, and in clear_subpage_extent_buffer_dirty() we
+		 * have to clear page dirty without subpage lock held.
+		 * This can cause race where our page gets dirty cleared after
+		 * we just set it.
+		 *
+		 * Thankfully, clear_subpage_extent_buffer_dirty() has locked
+		 * its page for other reasons, we can use page lock to prevent
+		 * the above race.
+		 */
+		if (subpage)
+			lock_page(eb->pages[0]);
+		for (i = 0; i < num_pages; i++)
+			btrfs_page_set_dirty(eb->fs_info, eb->pages[i],
+					     eb->start, eb->len);
+		if (subpage)
+			unlock_page(eb->pages[0]);
+	}
 #ifdef CONFIG_BTRFS_DEBUG
 	for (i = 0; i < num_pages; i++)
 		ASSERT(PageDirty(eb->pages[i]));
@@ -5366,33 +4895,117 @@ bool set_extent_buffer_dirty(struct extent_buffer *eb)
 
 void clear_extent_buffer_uptodate(struct extent_buffer *eb)
 {
-	int i;
+	struct btrfs_fs_info *fs_info = eb->fs_info;
 	struct page *page;
 	int num_pages;
+	int i;
 
 	clear_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
 	num_pages = num_extent_pages(eb);
 	for (i = 0; i < num_pages; i++) {
 		page = eb->pages[i];
-		if (page)
+		if (!page)
+			continue;
+
+		/*
+		 * This is special handling for metadata subpage, as regular
+		 * btrfs_is_subpage() can not handle cloned/dummy metadata.
+		 */
+		if (fs_info->nodesize >= PAGE_SIZE)
 			ClearPageUptodate(page);
+		else
+			btrfs_subpage_clear_uptodate(fs_info, page, eb->start,
+						     eb->len);
 	}
 }
 
 void set_extent_buffer_uptodate(struct extent_buffer *eb)
 {
-	int i;
+	struct btrfs_fs_info *fs_info = eb->fs_info;
 	struct page *page;
 	int num_pages;
+	int i;
 
 	set_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
 	num_pages = num_extent_pages(eb);
 	for (i = 0; i < num_pages; i++) {
 		page = eb->pages[i];
-		SetPageUptodate(page);
+
+		/*
+		 * This is special handling for metadata subpage, as regular
+		 * btrfs_is_subpage() can not handle cloned/dummy metadata.
+		 */
+		if (fs_info->nodesize >= PAGE_SIZE)
+			SetPageUptodate(page);
+		else
+			btrfs_subpage_set_uptodate(fs_info, page, eb->start,
+						   eb->len);
 	}
 }
 
+static int read_extent_buffer_subpage(struct extent_buffer *eb, int wait,
+				      int mirror_num)
+{
+	struct btrfs_fs_info *fs_info = eb->fs_info;
+	struct extent_io_tree *io_tree;
+	struct page *page = eb->pages[0];
+	struct btrfs_bio_ctrl bio_ctrl = {
+		.mirror_num = mirror_num,
+	};
+	int ret = 0;
+
+	ASSERT(!test_bit(EXTENT_BUFFER_UNMAPPED, &eb->bflags));
+	ASSERT(PagePrivate(page));
+	io_tree = &BTRFS_I(fs_info->btree_inode)->io_tree;
+
+	if (wait == WAIT_NONE) {
+		if (!try_lock_extent(io_tree, eb->start, eb->start + eb->len - 1))
+			return -EAGAIN;
+	} else {
+		ret = lock_extent(io_tree, eb->start, eb->start + eb->len - 1, NULL);
+		if (ret < 0)
+			return ret;
+	}
+
+	ret = 0;
+	if (test_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags) ||
+	    PageUptodate(page) ||
+	    btrfs_subpage_test_uptodate(fs_info, page, eb->start, eb->len)) {
+		set_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
+		unlock_extent(io_tree, eb->start, eb->start + eb->len - 1, NULL);
+		return ret;
+	}
+
+	clear_bit(EXTENT_BUFFER_READ_ERR, &eb->bflags);
+	eb->read_mirror = 0;
+	atomic_set(&eb->io_pages, 1);
+	check_buffer_tree_ref(eb);
+	bio_ctrl.end_io_func = end_bio_extent_readpage;
+
+	btrfs_subpage_clear_error(fs_info, page, eb->start, eb->len);
+
+	btrfs_subpage_start_reader(fs_info, page, eb->start, eb->len);
+	ret = submit_extent_page(REQ_OP_READ, NULL, &bio_ctrl,
+				 eb->start, page, eb->len,
+				 eb->start - page_offset(page), 0, true);
+	if (ret) {
+		/*
+		 * In the endio function, if we hit something wrong we will
+		 * increase the io_pages, so here we need to decrease it for
+		 * error path.
+		 */
+		atomic_dec(&eb->io_pages);
+	}
+	submit_one_bio(&bio_ctrl);
+	if (ret || wait != WAIT_COMPLETE)
+		return ret;
+
+	wait_extent_bit(io_tree, eb->start, eb->start + eb->len - 1, EXTENT_LOCKED);
+	if (!test_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags))
+		ret = -EIO;
+	return ret;
+}
+
 int read_extent_buffer_pages(struct extent_buffer *eb, int wait, int mirror_num)
 {
 	int i;
@@ -5403,17 +5016,35 @@ int read_extent_buffer_pages(struct extent_buffer *eb, int wait, int mirror_num)
 	int all_uptodate = 1;
 	int num_pages;
 	unsigned long num_reads = 0;
-	struct bio *bio = NULL;
-	unsigned long bio_flags = 0;
-	struct extent_io_tree *tree = &BTRFS_I(eb->fs_info->btree_inode)->io_tree;
+	struct btrfs_bio_ctrl bio_ctrl = {
+		.mirror_num = mirror_num,
+	};
 
 	if (test_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags))
 		return 0;
 
+	/*
+	 * We could have had EXTENT_BUFFER_UPTODATE cleared by the write
+	 * operation, which could potentially still be in flight.  In this case
+	 * we simply want to return an error.
+	 */
+	if (unlikely(test_bit(EXTENT_BUFFER_WRITE_ERR, &eb->bflags)))
+		return -EIO;
+
+	if (eb->fs_info->nodesize < PAGE_SIZE)
+		return read_extent_buffer_subpage(eb, wait, mirror_num);
+
 	num_pages = num_extent_pages(eb);
 	for (i = 0; i < num_pages; i++) {
 		page = eb->pages[i];
 		if (wait == WAIT_NONE) {
+			/*
+			 * WAIT_NONE is only utilized by readahead. If we can't
+			 * acquire the lock atomically it means either the eb
+			 * is being read out or under modification.
+			 * Either way the eb will be or has been cached,
+			 * readahead can exit safely.
+			 */
 			if (!trylock_page(page))
 				goto unlock_exit;
 		} else {
@@ -5442,6 +5073,12 @@ int read_extent_buffer_pages(struct extent_buffer *eb, int wait, int mirror_num)
 	clear_bit(EXTENT_BUFFER_READ_ERR, &eb->bflags);
 	eb->read_mirror = 0;
 	atomic_set(&eb->io_pages, num_reads);
+	/*
+	 * It is possible for release_folio to clear the TREE_REF bit before we
+	 * set io_pages. See check_buffer_tree_ref for a more detailed comment.
+	 */
+	check_buffer_tree_ref(eb);
+	bio_ctrl.end_io_func = end_bio_extent_readpage;
 	for (i = 0; i < num_pages; i++) {
 		page = eb->pages[i];
 
@@ -5453,20 +5090,18 @@ int read_extent_buffer_pages(struct extent_buffer *eb, int wait, int mirror_num)
 			}
 
 			ClearPageError(page);
-			err = __extent_read_full_page(tree, page,
-						      btree_get_extent, &bio,
-						      mirror_num, &bio_flags,
-						      REQ_META);
+			err = submit_extent_page(REQ_OP_READ, NULL,
+					 &bio_ctrl, page_offset(page), page,
+					 PAGE_SIZE, 0, 0, false);
 			if (err) {
-				ret = err;
 				/*
-				 * We use &bio in above __extent_read_full_page,
-				 * so we ensure that if it returns error, the
-				 * current page fails to add itself to bio and
-				 * it's been unlocked.
-				 *
-				 * We must dec io_pages by ourselves.
+				 * We failed to submit the bio so it's the
+				 * caller's responsibility to perform cleanup
+				 * i.e unlock page/set error bit.
 				 */
+				ret = err;
+				SetPageError(page);
+				unlock_page(page);
 				atomic_dec(&eb->io_pages);
 			}
 		} else {
@@ -5474,11 +5109,7 @@ int read_extent_buffer_pages(struct extent_buffer *eb, int wait, int mirror_num)
 		}
 	}
 
-	if (bio) {
-		err = submit_one_bio(bio, mirror_num, bio_flags);
-		if (err)
-			return err;
-	}
+	submit_one_bio(&bio_ctrl);
 
 	if (ret || wait != WAIT_COMPLETE)
 		return ret;
@@ -5501,6 +5132,36 @@ unlock_exit:
 	return ret;
 }
 
+static bool report_eb_range(const struct extent_buffer *eb, unsigned long start,
+			    unsigned long len)
+{
+	btrfs_warn(eb->fs_info,
+		"access to eb bytenr %llu len %lu out of range start %lu len %lu",
+		eb->start, eb->len, start, len);
+	WARN_ON(IS_ENABLED(CONFIG_BTRFS_DEBUG));
+
+	return true;
+}
+
+/*
+ * Check if the [start, start + len) range is valid before reading/writing
+ * the eb.
+ * NOTE: @start and @len are offset inside the eb, not logical address.
+ *
+ * Caller should not touch the dst/src memory if this function returns error.
+ */
+static inline int check_eb_range(const struct extent_buffer *eb,
+				 unsigned long start, unsigned long len)
+{
+	unsigned long offset;
+
+	/* start, start + len should not go beyond eb->len nor overflow */
+	if (unlikely(check_add_overflow(start, len, &offset) || offset > eb->len))
+		return report_eb_range(eb, start, len);
+
+	return false;
+}
+
 void read_extent_buffer(const struct extent_buffer *eb, void *dstv,
 			unsigned long start, unsigned long len)
 {
@@ -5509,17 +5170,12 @@ void read_extent_buffer(const struct extent_buffer *eb, void *dstv,
 	struct page *page;
 	char *kaddr;
 	char *dst = (char *)dstv;
-	size_t start_offset = offset_in_page(eb->start);
-	unsigned long i = (start_offset + start) >> PAGE_SHIFT;
+	unsigned long i = get_eb_page_index(start);
 
-	if (start + len > eb->len) {
-		WARN(1, KERN_ERR "btrfs bad mapping eb start %llu len %lu, wanted %lu %lu\n",
-		     eb->start, eb->len, start, len);
-		memset(dst, 0, len);
+	if (check_eb_range(eb, start, len))
 		return;
-	}
 
-	offset = offset_in_page(start_offset + start);
+	offset = get_eb_offset_in_page(eb, start);
 
 	while (len > 0) {
 		page = eb->pages[i];
@@ -5535,30 +5191,29 @@ void read_extent_buffer(const struct extent_buffer *eb, void *dstv,
 	}
 }
 
-int read_extent_buffer_to_user(const struct extent_buffer *eb,
-			       void __user *dstv,
-			       unsigned long start, unsigned long len)
+int read_extent_buffer_to_user_nofault(const struct extent_buffer *eb,
+				       void __user *dstv,
+				       unsigned long start, unsigned long len)
 {
 	size_t cur;
 	size_t offset;
 	struct page *page;
 	char *kaddr;
 	char __user *dst = (char __user *)dstv;
-	size_t start_offset = offset_in_page(eb->start);
-	unsigned long i = (start_offset + start) >> PAGE_SHIFT;
+	unsigned long i = get_eb_page_index(start);
 	int ret = 0;
 
 	WARN_ON(start > eb->len);
 	WARN_ON(start + len > eb->start + eb->len);
 
-	offset = offset_in_page(start_offset + start);
+	offset = get_eb_offset_in_page(eb, start);
 
 	while (len > 0) {
 		page = eb->pages[i];
 
 		cur = min(len, (PAGE_SIZE - offset));
 		kaddr = page_address(page);
-		if (copy_to_user(dst, kaddr + offset, cur)) {
+		if (copy_to_user_nofault(dst, kaddr + offset, cur)) {
 			ret = -EFAULT;
 			break;
 		}
@@ -5572,48 +5227,6 @@ int read_extent_buffer_to_user(const struct extent_buffer *eb,
 	return ret;
 }
 
-/*
- * return 0 if the item is found within a page.
- * return 1 if the item spans two pages.
- * return -EINVAL otherwise.
- */
-int map_private_extent_buffer(const struct extent_buffer *eb,
-			      unsigned long start, unsigned long min_len,
-			      char **map, unsigned long *map_start,
-			      unsigned long *map_len)
-{
-	size_t offset;
-	char *kaddr;
-	struct page *p;
-	size_t start_offset = offset_in_page(eb->start);
-	unsigned long i = (start_offset + start) >> PAGE_SHIFT;
-	unsigned long end_i = (start_offset + start + min_len - 1) >>
-		PAGE_SHIFT;
-
-	if (start + min_len > eb->len) {
-		WARN(1, KERN_ERR "btrfs bad mapping eb start %llu len %lu, wanted %lu %lu\n",
-		       eb->start, eb->len, start, min_len);
-		return -EINVAL;
-	}
-
-	if (i != end_i)
-		return 1;
-
-	if (i == 0) {
-		offset = start_offset;
-		*map_start = 0;
-	} else {
-		offset = 0;
-		*map_start = ((u64)i << PAGE_SHIFT) - start_offset;
-	}
-
-	p = eb->pages[i];
-	kaddr = page_address(p);
-	*map = kaddr + offset;
-	*map_len = PAGE_SIZE - offset;
-	return 0;
-}
-
 int memcmp_extent_buffer(const struct extent_buffer *eb, const void *ptrv,
 			 unsigned long start, unsigned long len)
 {
@@ -5622,14 +5235,13 @@ int memcmp_extent_buffer(const struct extent_buffer *eb, const void *ptrv,
 	struct page *page;
 	char *kaddr;
 	char *ptr = (char *)ptrv;
-	size_t start_offset = offset_in_page(eb->start);
-	unsigned long i = (start_offset + start) >> PAGE_SHIFT;
+	unsigned long i = get_eb_page_index(start);
 	int ret = 0;
 
-	WARN_ON(start > eb->len);
-	WARN_ON(start + len > eb->start + eb->len);
+	if (check_eb_range(eb, start, len))
+		return -EINVAL;
 
-	offset = offset_in_page(start_offset + start);
+	offset = get_eb_offset_in_page(eb, start);
 
 	while (len > 0) {
 		page = eb->pages[i];
@@ -5649,28 +5261,61 @@ int memcmp_extent_buffer(const struct extent_buffer *eb, const void *ptrv,
 	return ret;
 }
 
-void write_extent_buffer_chunk_tree_uuid(struct extent_buffer *eb,
+/*
+ * Check that the extent buffer is uptodate.
+ *
+ * For regular sector size == PAGE_SIZE case, check if @page is uptodate.
+ * For subpage case, check if the range covered by the eb has EXTENT_UPTODATE.
+ */
+static void assert_eb_page_uptodate(const struct extent_buffer *eb,
+				    struct page *page)
+{
+	struct btrfs_fs_info *fs_info = eb->fs_info;
+
+	/*
+	 * If we are using the commit root we could potentially clear a page
+	 * Uptodate while we're using the extent buffer that we've previously
+	 * looked up.  We don't want to complain in this case, as the page was
+	 * valid before, we just didn't write it out.  Instead we want to catch
+	 * the case where we didn't actually read the block properly, which
+	 * would have !PageUptodate && !PageError, as we clear PageError before
+	 * reading.
+	 */
+	if (fs_info->nodesize < PAGE_SIZE) {
+		bool uptodate, error;
+
+		uptodate = btrfs_subpage_test_uptodate(fs_info, page,
+						       eb->start, eb->len);
+		error = btrfs_subpage_test_error(fs_info, page, eb->start, eb->len);
+		WARN_ON(!uptodate && !error);
+	} else {
+		WARN_ON(!PageUptodate(page) && !PageError(page));
+	}
+}
+
+void write_extent_buffer_chunk_tree_uuid(const struct extent_buffer *eb,
 		const void *srcv)
 {
 	char *kaddr;
 
-	WARN_ON(!PageUptodate(eb->pages[0]));
-	kaddr = page_address(eb->pages[0]);
-	memcpy(kaddr + offsetof(struct btrfs_header, chunk_tree_uuid), srcv,
-			BTRFS_FSID_SIZE);
+	assert_eb_page_uptodate(eb, eb->pages[0]);
+	kaddr = page_address(eb->pages[0]) +
+		get_eb_offset_in_page(eb, offsetof(struct btrfs_header,
+						   chunk_tree_uuid));
+	memcpy(kaddr, srcv, BTRFS_FSID_SIZE);
 }
 
-void write_extent_buffer_fsid(struct extent_buffer *eb, const void *srcv)
+void write_extent_buffer_fsid(const struct extent_buffer *eb, const void *srcv)
 {
 	char *kaddr;
 
-	WARN_ON(!PageUptodate(eb->pages[0]));
-	kaddr = page_address(eb->pages[0]);
-	memcpy(kaddr + offsetof(struct btrfs_header, fsid), srcv,
-			BTRFS_FSID_SIZE);
+	assert_eb_page_uptodate(eb, eb->pages[0]);
+	kaddr = page_address(eb->pages[0]) +
+		get_eb_offset_in_page(eb, offsetof(struct btrfs_header, fsid));
+	memcpy(kaddr, srcv, BTRFS_FSID_SIZE);
 }
 
-void write_extent_buffer(struct extent_buffer *eb, const void *srcv,
+void write_extent_buffer(const struct extent_buffer *eb, const void *srcv,
 			 unsigned long start, unsigned long len)
 {
 	size_t cur;
@@ -5678,17 +5323,18 @@ void write_extent_buffer(struct extent_buffer *eb, const void *srcv,
 	struct page *page;
 	char *kaddr;
 	char *src = (char *)srcv;
-	size_t start_offset = offset_in_page(eb->start);
-	unsigned long i = (start_offset + start) >> PAGE_SHIFT;
+	unsigned long i = get_eb_page_index(start);
 
-	WARN_ON(start > eb->len);
-	WARN_ON(start + len > eb->start + eb->len);
+	WARN_ON(test_bit(EXTENT_BUFFER_NO_CHECK, &eb->bflags));
 
-	offset = offset_in_page(start_offset + start);
+	if (check_eb_range(eb, start, len))
+		return;
+
+	offset = get_eb_offset_in_page(eb, start);
 
 	while (len > 0) {
 		page = eb->pages[i];
-		WARN_ON(!PageUptodate(page));
+		assert_eb_page_uptodate(eb, page);
 
 		cur = min(len, PAGE_SIZE - offset);
 		kaddr = page_address(page);
@@ -5701,24 +5347,23 @@ void write_extent_buffer(struct extent_buffer *eb, const void *srcv,
 	}
 }
 
-void memzero_extent_buffer(struct extent_buffer *eb, unsigned long start,
+void memzero_extent_buffer(const struct extent_buffer *eb, unsigned long start,
 		unsigned long len)
 {
 	size_t cur;
 	size_t offset;
 	struct page *page;
 	char *kaddr;
-	size_t start_offset = offset_in_page(eb->start);
-	unsigned long i = (start_offset + start) >> PAGE_SHIFT;
+	unsigned long i = get_eb_page_index(start);
 
-	WARN_ON(start > eb->len);
-	WARN_ON(start + len > eb->start + eb->len);
+	if (check_eb_range(eb, start, len))
+		return;
 
-	offset = offset_in_page(start_offset + start);
+	offset = get_eb_offset_in_page(eb, start);
 
 	while (len > 0) {
 		page = eb->pages[i];
-		WARN_ON(!PageUptodate(page));
+		assert_eb_page_uptodate(eb, page);
 
 		cur = min(len, PAGE_SIZE - offset);
 		kaddr = page_address(page);
@@ -5730,21 +5375,32 @@ void memzero_extent_buffer(struct extent_buffer *eb, unsigned long start,
 	}
 }
 
-void copy_extent_buffer_full(struct extent_buffer *dst,
-			     struct extent_buffer *src)
+void copy_extent_buffer_full(const struct extent_buffer *dst,
+			     const struct extent_buffer *src)
 {
 	int i;
 	int num_pages;
 
 	ASSERT(dst->len == src->len);
 
-	num_pages = num_extent_pages(dst);
-	for (i = 0; i < num_pages; i++)
-		copy_page(page_address(dst->pages[i]),
-				page_address(src->pages[i]));
+	if (dst->fs_info->nodesize >= PAGE_SIZE) {
+		num_pages = num_extent_pages(dst);
+		for (i = 0; i < num_pages; i++)
+			copy_page(page_address(dst->pages[i]),
+				  page_address(src->pages[i]));
+	} else {
+		size_t src_offset = get_eb_offset_in_page(src, 0);
+		size_t dst_offset = get_eb_offset_in_page(dst, 0);
+
+		ASSERT(src->fs_info->nodesize < PAGE_SIZE);
+		memcpy(page_address(dst->pages[0]) + dst_offset,
+		       page_address(src->pages[0]) + src_offset,
+		       src->len);
+	}
 }
 
-void copy_extent_buffer(struct extent_buffer *dst, struct extent_buffer *src,
+void copy_extent_buffer(const struct extent_buffer *dst,
+			const struct extent_buffer *src,
 			unsigned long dst_offset, unsigned long src_offset,
 			unsigned long len)
 {
@@ -5753,16 +5409,19 @@ void copy_extent_buffer(struct extent_buffer *dst, struct extent_buffer *src,
 	size_t offset;
 	struct page *page;
 	char *kaddr;
-	size_t start_offset = offset_in_page(dst->start);
-	unsigned long i = (start_offset + dst_offset) >> PAGE_SHIFT;
+	unsigned long i = get_eb_page_index(dst_offset);
+
+	if (check_eb_range(dst, dst_offset, len) ||
+	    check_eb_range(src, src_offset, len))
+		return;
 
 	WARN_ON(src->len != dst_len);
 
-	offset = offset_in_page(start_offset + dst_offset);
+	offset = get_eb_offset_in_page(dst, dst_offset);
 
 	while (len > 0) {
 		page = dst->pages[i];
-		WARN_ON(!PageUptodate(page));
+		assert_eb_page_uptodate(dst, page);
 
 		cur = min(len, (unsigned long)(PAGE_SIZE - offset));
 
@@ -5789,12 +5448,11 @@ void copy_extent_buffer(struct extent_buffer *dst, struct extent_buffer *src,
  * This helper hides the ugliness of finding the byte in an extent buffer which
  * contains a given bit.
  */
-static inline void eb_bitmap_offset(struct extent_buffer *eb,
+static inline void eb_bitmap_offset(const struct extent_buffer *eb,
 				    unsigned long start, unsigned long nr,
 				    unsigned long *page_index,
 				    size_t *page_offset)
 {
-	size_t start_offset = offset_in_page(eb->start);
 	size_t byte_offset = BIT_BYTE(nr);
 	size_t offset;
 
@@ -5803,7 +5461,7 @@ static inline void eb_bitmap_offset(struct extent_buffer *eb,
 	 * the bitmap item in the extent buffer + the offset of the byte in the
 	 * bitmap item.
 	 */
-	offset = start_offset + start + byte_offset;
+	offset = start + offset_in_page(eb->start) + byte_offset;
 
 	*page_index = offset >> PAGE_SHIFT;
 	*page_offset = offset_in_page(offset);
@@ -5815,7 +5473,7 @@ static inline void eb_bitmap_offset(struct extent_buffer *eb,
  * @start: offset of the bitmap item in the extent buffer
  * @nr: bit number to test
  */
-int extent_buffer_test_bit(struct extent_buffer *eb, unsigned long start,
+int extent_buffer_test_bit(const struct extent_buffer *eb, unsigned long start,
 			   unsigned long nr)
 {
 	u8 *kaddr;
@@ -5825,7 +5483,7 @@ int extent_buffer_test_bit(struct extent_buffer *eb, unsigned long start,
 
 	eb_bitmap_offset(eb, start, nr, &i, &offset);
 	page = eb->pages[i];
-	WARN_ON(!PageUptodate(page));
+	assert_eb_page_uptodate(eb, page);
 	kaddr = page_address(page);
 	return 1U & (kaddr[offset] >> (nr & (BITS_PER_BYTE - 1)));
 }
@@ -5837,7 +5495,7 @@ int extent_buffer_test_bit(struct extent_buffer *eb, unsigned long start,
  * @pos: bit number of the first bit
  * @len: number of bits to set
  */
-void extent_buffer_bitmap_set(struct extent_buffer *eb, unsigned long start,
+void extent_buffer_bitmap_set(const struct extent_buffer *eb, unsigned long start,
 			      unsigned long pos, unsigned long len)
 {
 	u8 *kaddr;
@@ -5850,7 +5508,7 @@ void extent_buffer_bitmap_set(struct extent_buffer *eb, unsigned long start,
 
 	eb_bitmap_offset(eb, start, pos, &i, &offset);
 	page = eb->pages[i];
-	WARN_ON(!PageUptodate(page));
+	assert_eb_page_uptodate(eb, page);
 	kaddr = page_address(page);
 
 	while (len >= bits_to_set) {
@@ -5861,7 +5519,7 @@ void extent_buffer_bitmap_set(struct extent_buffer *eb, unsigned long start,
 		if (++offset >= PAGE_SIZE && len > 0) {
 			offset = 0;
 			page = eb->pages[++i];
-			WARN_ON(!PageUptodate(page));
+			assert_eb_page_uptodate(eb, page);
 			kaddr = page_address(page);
 		}
 	}
@@ -5879,8 +5537,9 @@ void extent_buffer_bitmap_set(struct extent_buffer *eb, unsigned long start,
  * @pos: bit number of the first bit
  * @len: number of bits to clear
  */
-void extent_buffer_bitmap_clear(struct extent_buffer *eb, unsigned long start,
-				unsigned long pos, unsigned long len)
+void extent_buffer_bitmap_clear(const struct extent_buffer *eb,
+				unsigned long start, unsigned long pos,
+				unsigned long len)
 {
 	u8 *kaddr;
 	struct page *page;
@@ -5892,7 +5551,7 @@ void extent_buffer_bitmap_clear(struct extent_buffer *eb, unsigned long start,
 
 	eb_bitmap_offset(eb, start, pos, &i, &offset);
 	page = eb->pages[i];
-	WARN_ON(!PageUptodate(page));
+	assert_eb_page_uptodate(eb, page);
 	kaddr = page_address(page);
 
 	while (len >= bits_to_clear) {
@@ -5903,7 +5562,7 @@ void extent_buffer_bitmap_clear(struct extent_buffer *eb, unsigned long start,
 		if (++offset >= PAGE_SIZE && len > 0) {
 			offset = 0;
 			page = eb->pages[++i];
-			WARN_ON(!PageUptodate(page));
+			assert_eb_page_uptodate(eb, page);
 			kaddr = page_address(page);
 		}
 	}
@@ -5941,36 +5600,26 @@ static void copy_pages(struct page *dst_page, struct page *src_page,
 		memcpy(dst_kaddr + dst_off, src_kaddr + src_off, len);
 }
 
-void memcpy_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset,
-			   unsigned long src_offset, unsigned long len)
+void memcpy_extent_buffer(const struct extent_buffer *dst,
+			  unsigned long dst_offset, unsigned long src_offset,
+			  unsigned long len)
 {
-	struct btrfs_fs_info *fs_info = dst->fs_info;
 	size_t cur;
 	size_t dst_off_in_page;
 	size_t src_off_in_page;
-	size_t start_offset = offset_in_page(dst->start);
 	unsigned long dst_i;
 	unsigned long src_i;
 
-	if (src_offset + len > dst->len) {
-		btrfs_err(fs_info,
-			"memmove bogus src_offset %lu move len %lu dst len %lu",
-			 src_offset, len, dst->len);
-		BUG();
-	}
-	if (dst_offset + len > dst->len) {
-		btrfs_err(fs_info,
-			"memmove bogus dst_offset %lu move len %lu dst len %lu",
-			 dst_offset, len, dst->len);
-		BUG();
-	}
+	if (check_eb_range(dst, dst_offset, len) ||
+	    check_eb_range(dst, src_offset, len))
+		return;
 
 	while (len > 0) {
-		dst_off_in_page = offset_in_page(start_offset + dst_offset);
-		src_off_in_page = offset_in_page(start_offset + src_offset);
+		dst_off_in_page = get_eb_offset_in_page(dst, dst_offset);
+		src_off_in_page = get_eb_offset_in_page(dst, src_offset);
 
-		dst_i = (start_offset + dst_offset) >> PAGE_SHIFT;
-		src_i = (start_offset + src_offset) >> PAGE_SHIFT;
+		dst_i = get_eb_page_index(dst_offset);
+		src_i = get_eb_page_index(src_offset);
 
 		cur = min(len, (unsigned long)(PAGE_SIZE -
 					       src_off_in_page));
@@ -5986,41 +5635,31 @@ void memcpy_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset,
 	}
 }
 
-void memmove_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset,
-			   unsigned long src_offset, unsigned long len)
+void memmove_extent_buffer(const struct extent_buffer *dst,
+			   unsigned long dst_offset, unsigned long src_offset,
+			   unsigned long len)
 {
-	struct btrfs_fs_info *fs_info = dst->fs_info;
 	size_t cur;
 	size_t dst_off_in_page;
 	size_t src_off_in_page;
 	unsigned long dst_end = dst_offset + len - 1;
 	unsigned long src_end = src_offset + len - 1;
-	size_t start_offset = offset_in_page(dst->start);
 	unsigned long dst_i;
 	unsigned long src_i;
 
-	if (src_offset + len > dst->len) {
-		btrfs_err(fs_info,
-			  "memmove bogus src_offset %lu move len %lu len %lu",
-			  src_offset, len, dst->len);
-		BUG();
-	}
-	if (dst_offset + len > dst->len) {
-		btrfs_err(fs_info,
-			  "memmove bogus dst_offset %lu move len %lu len %lu",
-			  dst_offset, len, dst->len);
-		BUG();
-	}
+	if (check_eb_range(dst, dst_offset, len) ||
+	    check_eb_range(dst, src_offset, len))
+		return;
 	if (dst_offset < src_offset) {
 		memcpy_extent_buffer(dst, dst_offset, src_offset, len);
 		return;
 	}
 	while (len > 0) {
-		dst_i = (start_offset + dst_end) >> PAGE_SHIFT;
-		src_i = (start_offset + src_end) >> PAGE_SHIFT;
+		dst_i = get_eb_page_index(dst_end);
+		src_i = get_eb_page_index(src_end);
 
-		dst_off_in_page = offset_in_page(start_offset + dst_end);
-		src_off_in_page = offset_in_page(start_offset + src_end);
+		dst_off_in_page = get_eb_offset_in_page(dst, dst_end);
+		src_off_in_page = get_eb_offset_in_page(dst, src_end);
 
 		cur = min_t(unsigned long, len, src_off_in_page + 1);
 		cur = min(cur, dst_off_in_page + 1);
@@ -6034,13 +5673,124 @@ void memmove_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset,
 	}
 }
 
+#define GANG_LOOKUP_SIZE	16
+static struct extent_buffer *get_next_extent_buffer(
+		struct btrfs_fs_info *fs_info, struct page *page, u64 bytenr)
+{
+	struct extent_buffer *gang[GANG_LOOKUP_SIZE];
+	struct extent_buffer *found = NULL;
+	u64 page_start = page_offset(page);
+	u64 cur = page_start;
+
+	ASSERT(in_range(bytenr, page_start, PAGE_SIZE));
+	lockdep_assert_held(&fs_info->buffer_lock);
+
+	while (cur < page_start + PAGE_SIZE) {
+		int ret;
+		int i;
+
+		ret = radix_tree_gang_lookup(&fs_info->buffer_radix,
+				(void **)gang, cur >> fs_info->sectorsize_bits,
+				min_t(unsigned int, GANG_LOOKUP_SIZE,
+				      PAGE_SIZE / fs_info->nodesize));
+		if (ret == 0)
+			goto out;
+		for (i = 0; i < ret; i++) {
+			/* Already beyond page end */
+			if (gang[i]->start >= page_start + PAGE_SIZE)
+				goto out;
+			/* Found one */
+			if (gang[i]->start >= bytenr) {
+				found = gang[i];
+				goto out;
+			}
+		}
+		cur = gang[ret - 1]->start + gang[ret - 1]->len;
+	}
+out:
+	return found;
+}
+
+static int try_release_subpage_extent_buffer(struct page *page)
+{
+	struct btrfs_fs_info *fs_info = btrfs_sb(page->mapping->host->i_sb);
+	u64 cur = page_offset(page);
+	const u64 end = page_offset(page) + PAGE_SIZE;
+	int ret;
+
+	while (cur < end) {
+		struct extent_buffer *eb = NULL;
+
+		/*
+		 * Unlike try_release_extent_buffer() which uses page->private
+		 * to grab buffer, for subpage case we rely on radix tree, thus
+		 * we need to ensure radix tree consistency.
+		 *
+		 * We also want an atomic snapshot of the radix tree, thus go
+		 * with spinlock rather than RCU.
+		 */
+		spin_lock(&fs_info->buffer_lock);
+		eb = get_next_extent_buffer(fs_info, page, cur);
+		if (!eb) {
+			/* No more eb in the page range after or at cur */
+			spin_unlock(&fs_info->buffer_lock);
+			break;
+		}
+		cur = eb->start + eb->len;
+
+		/*
+		 * The same as try_release_extent_buffer(), to ensure the eb
+		 * won't disappear out from under us.
+		 */
+		spin_lock(&eb->refs_lock);
+		if (atomic_read(&eb->refs) != 1 || extent_buffer_under_io(eb)) {
+			spin_unlock(&eb->refs_lock);
+			spin_unlock(&fs_info->buffer_lock);
+			break;
+		}
+		spin_unlock(&fs_info->buffer_lock);
+
+		/*
+		 * If tree ref isn't set then we know the ref on this eb is a
+		 * real ref, so just return, this eb will likely be freed soon
+		 * anyway.
+		 */
+		if (!test_and_clear_bit(EXTENT_BUFFER_TREE_REF, &eb->bflags)) {
+			spin_unlock(&eb->refs_lock);
+			break;
+		}
+
+		/*
+		 * Here we don't care about the return value, we will always
+		 * check the page private at the end.  And
+		 * release_extent_buffer() will release the refs_lock.
+		 */
+		release_extent_buffer(eb);
+	}
+	/*
+	 * Finally to check if we have cleared page private, as if we have
+	 * released all ebs in the page, the page private should be cleared now.
+	 */
+	spin_lock(&page->mapping->private_lock);
+	if (!PagePrivate(page))
+		ret = 1;
+	else
+		ret = 0;
+	spin_unlock(&page->mapping->private_lock);
+	return ret;
+
+}
+
 int try_release_extent_buffer(struct page *page)
 {
 	struct extent_buffer *eb;
 
+	if (btrfs_sb(page->mapping->host->i_sb)->nodesize < PAGE_SIZE)
+		return try_release_subpage_extent_buffer(page);
+
 	/*
-	 * We need to make sure nobody is attaching this page to an eb right
-	 * now.
+	 * We need to make sure nobody is changing page->private, as we rely on
+	 * page->private as the pointer to extent buffer.
 	 */
 	spin_lock(&page->mapping->private_lock);
 	if (!PagePrivate(page)) {
@@ -6075,3 +5825,54 @@ int try_release_extent_buffer(struct page *page)
 
 	return release_extent_buffer(eb);
 }
+
+/*
+ * btrfs_readahead_tree_block - attempt to readahead a child block
+ * @fs_info:	the fs_info
+ * @bytenr:	bytenr to read
+ * @owner_root: objectid of the root that owns this eb
+ * @gen:	generation for the uptodate check, can be 0
+ * @level:	level for the eb
+ *
+ * Attempt to readahead a tree block at @bytenr.  If @gen is 0 then we do a
+ * normal uptodate check of the eb, without checking the generation.  If we have
+ * to read the block we will not block on anything.
+ */
+void btrfs_readahead_tree_block(struct btrfs_fs_info *fs_info,
+				u64 bytenr, u64 owner_root, u64 gen, int level)
+{
+	struct extent_buffer *eb;
+	int ret;
+
+	eb = btrfs_find_create_tree_block(fs_info, bytenr, owner_root, level);
+	if (IS_ERR(eb))
+		return;
+
+	if (btrfs_buffer_uptodate(eb, gen, 1)) {
+		free_extent_buffer(eb);
+		return;
+	}
+
+	ret = read_extent_buffer_pages(eb, WAIT_NONE, 0);
+	if (ret < 0)
+		free_extent_buffer_stale(eb);
+	else
+		free_extent_buffer(eb);
+}
+
+/*
+ * btrfs_readahead_node_child - readahead a node's child block
+ * @node:	parent node we're reading from
+ * @slot:	slot in the parent node for the child we want to read
+ *
+ * A helper for btrfs_readahead_tree_block, we simply read the bytenr pointed at
+ * the slot in the node provided.
+ */
+void btrfs_readahead_node_child(struct extent_buffer *node, int slot)
+{
+	btrfs_readahead_tree_block(node->fs_info,
+				   btrfs_node_blockptr(node, slot),
+				   btrfs_header_owner(node),
+				   btrfs_node_ptr_generation(node, slot),
+				   btrfs_header_level(node) - 1);
+}
diff --git a/fs/btrfs/extent_io.h b/fs/btrfs/extent_io.h
index 5d205bbaafdc..7929f054dda3 100644
--- a/fs/btrfs/extent_io.h
+++ b/fs/btrfs/extent_io.h
@@ -5,15 +5,11 @@
 
 #include <linux/rbtree.h>
 #include <linux/refcount.h>
+#include <linux/fiemap.h>
+#include <linux/btrfs_tree.h>
+#include "compression.h"
 #include "ulist.h"
 
-/*
- * flags for bio submission. The high bits indicate the compression
- * type for this bio
- */
-#define EXTENT_BIO_COMPRESSED 1
-#define EXTENT_BIO_FLAG_SHIFT 16
-
 enum {
 	EXTENT_BUFFER_UPTODATE,
 	EXTENT_BUFFER_DIRTY,
@@ -29,16 +25,17 @@ enum {
 	EXTENT_BUFFER_IN_TREE,
 	/* write IO error */
 	EXTENT_BUFFER_WRITE_ERR,
+	EXTENT_BUFFER_NO_CHECK,
 };
 
 /* these are flags for __process_pages_contig */
 #define PAGE_UNLOCK		(1 << 0)
-#define PAGE_CLEAR_DIRTY	(1 << 1)
-#define PAGE_SET_WRITEBACK	(1 << 2)
-#define PAGE_END_WRITEBACK	(1 << 3)
-#define PAGE_SET_PRIVATE2	(1 << 4)
-#define PAGE_SET_ERROR		(1 << 5)
-#define PAGE_LOCK		(1 << 6)
+/* Page starts writeback, clear dirty bit and set writeback bit */
+#define PAGE_START_WRITEBACK	(1 << 1)
+#define PAGE_END_WRITEBACK	(1 << 2)
+#define PAGE_SET_ORDERED	(1 << 3)
+#define PAGE_SET_ERROR		(1 << 4)
+#define PAGE_LOCK		(1 << 5)
 
 /*
  * page->private values.  Every page that is controlled by the extent
@@ -60,30 +57,24 @@ enum {
 #define BITMAP_LAST_BYTE_MASK(nbits) \
 	(BYTE_MASK >> (-(nbits) & (BITS_PER_BYTE - 1)))
 
+struct btrfs_bio;
 struct btrfs_root;
 struct btrfs_inode;
-struct btrfs_io_bio;
+struct btrfs_fs_info;
 struct io_failure_record;
 struct extent_io_tree;
 
-typedef blk_status_t (extent_submit_bio_start_t)(void *private_data,
-		struct bio *bio, u64 bio_offset);
+int __init extent_buffer_init_cachep(void);
+void __cold extent_buffer_free_cachep(void);
 
-struct extent_io_ops {
-	/*
-	 * The following callbacks must be always defined, the function
-	 * pointer will be called unconditionally.
-	 */
-	blk_status_t (*submit_bio_hook)(struct inode *inode, struct bio *bio,
-					int mirror_num, unsigned long bio_flags);
-	int (*readpage_end_io_hook)(struct btrfs_io_bio *io_bio, u64 phy_offset,
-				    struct page *page, u64 start, u64 end,
-				    int mirror);
-};
+typedef void (submit_bio_hook_t)(struct inode *inode, struct bio *bio,
+					 int mirror_num,
+					 enum btrfs_compression_type compress_type);
 
+typedef blk_status_t (extent_submit_bio_start_t)(struct inode *inode,
+		struct bio *bio, u64 dio_file_offset);
 
-#define INLINE_EXTENT_BUFFER_PAGES 16
-#define MAX_INLINE_EXTENT_BUFFER_SIZE (INLINE_EXTENT_BUFFER_PAGES * PAGE_SIZE)
+#define INLINE_EXTENT_BUFFER_PAGES     (BTRFS_MAX_METADATA_BLOCKSIZE / PAGE_SIZE)
 struct extent_buffer {
 	u64 start;
 	unsigned long len;
@@ -95,31 +86,14 @@ struct extent_buffer {
 	int read_mirror;
 	struct rcu_head rcu_head;
 	pid_t lock_owner;
-
-	int blocking_writers;
-	atomic_t blocking_readers;
-	bool lock_nested;
 	/* >= 0 if eb belongs to a log tree, -1 otherwise */
-	short log_index;
-
-	/* protects write locks */
-	rwlock_t lock;
+	s8 log_index;
 
-	/* readers use lock_wq while they wait for the write
-	 * lock holders to unlock
-	 */
-	wait_queue_head_t write_lock_wq;
+	struct rw_semaphore lock;
 
-	/* writers use read_lock_wq while they wait for readers
-	 * to unlock
-	 */
-	wait_queue_head_t read_lock_wq;
 	struct page *pages[INLINE_EXTENT_BUFFER_PAGES];
+	struct list_head release_list;
 #ifdef CONFIG_BTRFS_DEBUG
-	int spinning_writers;
-	atomic_t spinning_readers;
-	atomic_t read_locks;
-	int write_locks;
 	struct list_head leak_list;
 #endif
 };
@@ -129,7 +103,7 @@ struct extent_buffer {
  */
 struct extent_changeset {
 	/* How many bytes are set/cleared in this operation */
-	unsigned int bytes_changed;
+	u64 bytes_changed;
 
 	/* Changed ranges */
 	struct ulist range_changed;
@@ -169,48 +143,30 @@ static inline void extent_changeset_free(struct extent_changeset *changeset)
 	kfree(changeset);
 }
 
-static inline void extent_set_compress_type(unsigned long *bio_flags,
-					    int compress_type)
-{
-	*bio_flags |= compress_type << EXTENT_BIO_FLAG_SHIFT;
-}
-
-static inline int extent_compress_type(unsigned long bio_flags)
-{
-	return bio_flags >> EXTENT_BIO_FLAG_SHIFT;
-}
-
 struct extent_map_tree;
 
-typedef struct extent_map *(get_extent_t)(struct btrfs_inode *inode,
-					  struct page *page, size_t pg_offset,
-					  u64 start, u64 len);
-
 int try_release_extent_mapping(struct page *page, gfp_t mask);
 int try_release_extent_buffer(struct page *page);
 
-int extent_read_full_page(struct extent_io_tree *tree, struct page *page,
-			  get_extent_t *get_extent, int mirror_num);
-int extent_write_full_page(struct page *page, struct writeback_control *wbc);
-int extent_write_locked_range(struct inode *inode, u64 start, u64 end,
-			      int mode);
+int btrfs_read_folio(struct file *file, struct folio *folio);
+int extent_write_locked_range(struct inode *inode, u64 start, u64 end);
 int extent_writepages(struct address_space *mapping,
 		      struct writeback_control *wbc);
 int btree_write_cache_pages(struct address_space *mapping,
 			    struct writeback_control *wbc);
-int extent_readpages(struct address_space *mapping, struct list_head *pages,
-		     unsigned nr_pages);
-int extent_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
-		__u64 start, __u64 len);
-void set_page_extent_mapped(struct page *page);
+void extent_readahead(struct readahead_control *rac);
+int extent_fiemap(struct btrfs_inode *inode, struct fiemap_extent_info *fieinfo,
+		  u64 start, u64 len);
+int set_page_extent_mapped(struct page *page);
+void clear_page_extent_mapped(struct page *page);
 
 struct extent_buffer *alloc_extent_buffer(struct btrfs_fs_info *fs_info,
-					  u64 start);
+					  u64 start, u64 owner_root, int level);
 struct extent_buffer *__alloc_dummy_extent_buffer(struct btrfs_fs_info *fs_info,
 						  u64 start, unsigned long len);
 struct extent_buffer *alloc_dummy_extent_buffer(struct btrfs_fs_info *fs_info,
 						u64 start);
-struct extent_buffer *btrfs_clone_extent_buffer(struct extent_buffer *src);
+struct extent_buffer *btrfs_clone_extent_buffer(const struct extent_buffer *src);
 struct extent_buffer *find_extent_buffer(struct btrfs_fs_info *fs_info,
 					 u64 start);
 void free_extent_buffer(struct extent_buffer *eb);
@@ -221,14 +177,23 @@ void free_extent_buffer_stale(struct extent_buffer *eb);
 int read_extent_buffer_pages(struct extent_buffer *eb, int wait,
 			     int mirror_num);
 void wait_on_extent_buffer_writeback(struct extent_buffer *eb);
+void btrfs_readahead_tree_block(struct btrfs_fs_info *fs_info,
+				u64 bytenr, u64 owner_root, u64 gen, int level);
+void btrfs_readahead_node_child(struct extent_buffer *node, int slot);
 
 static inline int num_extent_pages(const struct extent_buffer *eb)
 {
-	return (round_up(eb->start + eb->len, PAGE_SIZE) >> PAGE_SHIFT) -
-	       (eb->start >> PAGE_SHIFT);
+	/*
+	 * For sectorsize == PAGE_SIZE case, since nodesize is always aligned to
+	 * sectorsize, it's just eb->len >> PAGE_SHIFT.
+	 *
+	 * For sectorsize < PAGE_SIZE case, we could have nodesize < PAGE_SIZE,
+	 * thus have to ensure we get at least one page.
+	 */
+	return (eb->len >> PAGE_SHIFT) ?: 1;
 }
 
-static inline int extent_buffer_uptodate(struct extent_buffer *eb)
+static inline int extent_buffer_uptodate(const struct extent_buffer *eb)
 {
 	return test_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
 }
@@ -238,86 +203,82 @@ int memcmp_extent_buffer(const struct extent_buffer *eb, const void *ptrv,
 void read_extent_buffer(const struct extent_buffer *eb, void *dst,
 			unsigned long start,
 			unsigned long len);
-int read_extent_buffer_to_user(const struct extent_buffer *eb,
-			       void __user *dst, unsigned long start,
-			       unsigned long len);
-void write_extent_buffer_fsid(struct extent_buffer *eb, const void *src);
-void write_extent_buffer_chunk_tree_uuid(struct extent_buffer *eb,
+int read_extent_buffer_to_user_nofault(const struct extent_buffer *eb,
+				       void __user *dst, unsigned long start,
+				       unsigned long len);
+void write_extent_buffer_fsid(const struct extent_buffer *eb, const void *src);
+void write_extent_buffer_chunk_tree_uuid(const struct extent_buffer *eb,
 		const void *src);
-void write_extent_buffer(struct extent_buffer *eb, const void *src,
+void write_extent_buffer(const struct extent_buffer *eb, const void *src,
 			 unsigned long start, unsigned long len);
-void copy_extent_buffer_full(struct extent_buffer *dst,
-			     struct extent_buffer *src);
-void copy_extent_buffer(struct extent_buffer *dst, struct extent_buffer *src,
+void copy_extent_buffer_full(const struct extent_buffer *dst,
+			     const struct extent_buffer *src);
+void copy_extent_buffer(const struct extent_buffer *dst,
+			const struct extent_buffer *src,
 			unsigned long dst_offset, unsigned long src_offset,
 			unsigned long len);
-void memcpy_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset,
-			   unsigned long src_offset, unsigned long len);
-void memmove_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset,
-			   unsigned long src_offset, unsigned long len);
-void memzero_extent_buffer(struct extent_buffer *eb, unsigned long start,
+void memcpy_extent_buffer(const struct extent_buffer *dst,
+			  unsigned long dst_offset, unsigned long src_offset,
+			  unsigned long len);
+void memmove_extent_buffer(const struct extent_buffer *dst,
+			   unsigned long dst_offset, unsigned long src_offset,
+			   unsigned long len);
+void memzero_extent_buffer(const struct extent_buffer *eb, unsigned long start,
 			   unsigned long len);
-int extent_buffer_test_bit(struct extent_buffer *eb, unsigned long start,
+int extent_buffer_test_bit(const struct extent_buffer *eb, unsigned long start,
 			   unsigned long pos);
-void extent_buffer_bitmap_set(struct extent_buffer *eb, unsigned long start,
+void extent_buffer_bitmap_set(const struct extent_buffer *eb, unsigned long start,
 			      unsigned long pos, unsigned long len);
-void extent_buffer_bitmap_clear(struct extent_buffer *eb, unsigned long start,
-				unsigned long pos, unsigned long len);
-void clear_extent_buffer_dirty(struct extent_buffer *eb);
+void extent_buffer_bitmap_clear(const struct extent_buffer *eb,
+				unsigned long start, unsigned long pos,
+				unsigned long len);
+void clear_extent_buffer_dirty(const struct extent_buffer *eb);
 bool set_extent_buffer_dirty(struct extent_buffer *eb);
 void set_extent_buffer_uptodate(struct extent_buffer *eb);
 void clear_extent_buffer_uptodate(struct extent_buffer *eb);
-int extent_buffer_under_io(struct extent_buffer *eb);
-int map_private_extent_buffer(const struct extent_buffer *eb,
-			      unsigned long offset, unsigned long min_len,
-			      char **map, unsigned long *map_start,
-			      unsigned long *map_len);
+int extent_buffer_under_io(const struct extent_buffer *eb);
 void extent_range_clear_dirty_for_io(struct inode *inode, u64 start, u64 end);
 void extent_range_redirty_for_io(struct inode *inode, u64 start, u64 end);
-void extent_clear_unlock_delalloc(struct inode *inode, u64 start, u64 end,
+void extent_clear_unlock_delalloc(struct btrfs_inode *inode, u64 start, u64 end,
 				  struct page *locked_page,
-				  unsigned bits_to_clear,
-				  unsigned long page_ops);
-struct bio *btrfs_bio_alloc(u64 first_byte);
-struct bio *btrfs_io_bio_alloc(unsigned int nr_iovecs);
-struct bio *btrfs_bio_clone(struct bio *bio);
-struct bio *btrfs_bio_clone_partial(struct bio *orig, int offset, int size);
+				  u32 bits_to_clear, unsigned long page_ops);
+int extent_invalidate_folio(struct extent_io_tree *tree,
+			    struct folio *folio, size_t offset);
 
-struct btrfs_fs_info;
-struct btrfs_inode;
+int btrfs_alloc_page_array(unsigned int nr_pages, struct page **page_array);
 
-int repair_io_failure(struct btrfs_fs_info *fs_info, u64 ino, u64 start,
-		      u64 length, u64 logical, struct page *page,
-		      unsigned int pg_offset, int mirror_num);
 void end_extent_writepage(struct page *page, int err, u64 start, u64 end);
-int btrfs_repair_eb_io_failure(struct extent_buffer *eb, int mirror_num);
+int btrfs_repair_eb_io_failure(const struct extent_buffer *eb, int mirror_num);
 
 /*
  * When IO fails, either with EIO or csum verification fails, we
  * try other mirrors that might have a good copy of the data.  This
  * io_failure_record is used to record state as we go through all the
- * mirrors.  If another mirror has good data, the page is set up to date
+ * mirrors.  If another mirror has good data, the sector is set up to date
  * and things continue.  If a good mirror can't be found, the original
  * bio end_io callback is called to indicate things have failed.
  */
 struct io_failure_record {
+	/* Use rb_simple_node for search/insert */
+	struct {
+		struct rb_node rb_node;
+		u64 bytenr;
+	};
 	struct page *page;
-	u64 start;
 	u64 len;
 	u64 logical;
-	unsigned long bio_flags;
 	int this_mirror;
 	int failed_mirror;
-	int in_validation;
+	int num_copies;
 };
 
+int btrfs_repair_one_sector(struct inode *inode, struct btrfs_bio *failed_bbio,
+			    u32 bio_offset, struct page *page, unsigned int pgoff,
+			    submit_bio_hook_t *submit_bio_hook);
+void btrfs_free_io_failure_record(struct btrfs_inode *inode, u64 start, u64 end);
+int btrfs_clean_io_failure(struct btrfs_inode *inode, u64 start,
+			   struct page *page, unsigned int pg_offset);
 
-bool btrfs_check_repairable(struct inode *inode, unsigned failed_bio_pages,
-			    struct io_failure_record *failrec, int fail_mirror);
-struct bio *btrfs_create_repair_bio(struct inode *inode, struct bio *failed_bio,
-				    struct io_failure_record *failrec,
-				    struct page *page, int pg_offset, int icsum,
-				    bio_end_io_t *endio_func, void *data);
 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
 bool find_lock_delalloc_range(struct inode *inode,
 			     struct page *locked_page, u64 *start,
@@ -325,4 +286,11 @@ bool find_lock_delalloc_range(struct inode *inode,
 #endif
 struct extent_buffer *alloc_test_extent_buffer(struct btrfs_fs_info *fs_info,
 					       u64 start);
+
+#ifdef CONFIG_BTRFS_DEBUG
+void btrfs_extent_buffer_leak_debug_check(struct btrfs_fs_info *fs_info);
+#else
+#define btrfs_extent_buffer_leak_debug_check(fs_info)	do {} while (0)
+#endif
+
 #endif
diff --git a/fs/btrfs/extent_map.c b/fs/btrfs/extent_map.c
index bd6229fb2b6f..6092a4eedc92 100644
--- a/fs/btrfs/extent_map.c
+++ b/fs/btrfs/extent_map.c
@@ -7,6 +7,7 @@
 #include "volumes.h"
 #include "extent_map.h"
 #include "compression.h"
+#include "btrfs_inode.h"
 
 
 static struct kmem_cache *extent_map_cache;
@@ -54,9 +55,7 @@ struct extent_map *alloc_extent_map(void)
 	if (!em)
 		return NULL;
 	RB_CLEAR_NODE(&em->rb_node);
-	em->flags = 0;
 	em->compress_type = BTRFS_COMPRESS_NONE;
-	em->generation = 0;
 	refcount_set(&em->refs, 1);
 	INIT_LIST_HEAD(&em->list);
 	return em;
@@ -73,7 +72,6 @@ void free_extent_map(struct extent_map *em)
 {
 	if (!em)
 		return;
-	WARN_ON(refcount_read(&em->refs) == 0);
 	if (refcount_dec_and_test(&em->refs)) {
 		WARN_ON(extent_map_in_tree(em));
 		WARN_ON(!list_empty(&em->list));
@@ -143,8 +141,7 @@ static int tree_insert(struct rb_root_cached *root, struct extent_map *em)
  * it can't be found, try to find some neighboring extents
  */
 static struct rb_node *__tree_search(struct rb_root *root, u64 offset,
-				     struct rb_node **prev_ret,
-				     struct rb_node **next_ret)
+				     struct rb_node **prev_or_next_ret)
 {
 	struct rb_node *n = root->rb_node;
 	struct rb_node *prev = NULL;
@@ -152,6 +149,8 @@ static struct rb_node *__tree_search(struct rb_root *root, u64 offset,
 	struct extent_map *entry;
 	struct extent_map *prev_entry = NULL;
 
+	ASSERT(prev_or_next_ret);
+
 	while (n) {
 		entry = rb_entry(n, struct extent_map, rb_node);
 		prev = n;
@@ -165,24 +164,29 @@ static struct rb_node *__tree_search(struct rb_root *root, u64 offset,
 			return n;
 	}
 
-	if (prev_ret) {
-		orig_prev = prev;
-		while (prev && offset >= extent_map_end(prev_entry)) {
-			prev = rb_next(prev);
-			prev_entry = rb_entry(prev, struct extent_map, rb_node);
-		}
-		*prev_ret = prev;
-		prev = orig_prev;
+	orig_prev = prev;
+	while (prev && offset >= extent_map_end(prev_entry)) {
+		prev = rb_next(prev);
+		prev_entry = rb_entry(prev, struct extent_map, rb_node);
 	}
 
-	if (next_ret) {
+	/*
+	 * Previous extent map found, return as in this case the caller does not
+	 * care about the next one.
+	 */
+	if (prev) {
+		*prev_or_next_ret = prev;
+		return NULL;
+	}
+
+	prev = orig_prev;
+	prev_entry = rb_entry(prev, struct extent_map, rb_node);
+	while (prev && offset < prev_entry->start) {
+		prev = rb_prev(prev);
 		prev_entry = rb_entry(prev, struct extent_map, rb_node);
-		while (prev && offset < prev_entry->start) {
-			prev = rb_prev(prev);
-			prev_entry = rb_entry(prev, struct extent_map, rb_node);
-		}
-		*next_ret = prev;
 	}
+	*prev_or_next_ret = prev;
+
 	return NULL;
 }
 
@@ -261,6 +265,7 @@ static void try_merge_map(struct extent_map_tree *tree, struct extent_map *em)
 			em->mod_len = (em->mod_len + em->mod_start) - merge->mod_start;
 			em->mod_start = merge->mod_start;
 			em->generation = max(em->generation, merge->generation);
+			set_bit(EXTENT_FLAG_MERGED, &em->flags);
 
 			rb_erase_cached(&merge->rb_node, &tree->map);
 			RB_CLEAR_NODE(&merge->rb_node);
@@ -278,6 +283,7 @@ static void try_merge_map(struct extent_map_tree *tree, struct extent_map *em)
 		RB_CLEAR_NODE(&merge->rb_node);
 		em->mod_len = (merge->mod_start + merge->mod_len) - em->mod_start;
 		em->generation = max(em->generation, merge->generation);
+		set_bit(EXTENT_FLAG_MERGED, &em->flags);
 		free_extent_map(merge);
 	}
 }
@@ -334,6 +340,8 @@ out:
 
 void clear_em_logging(struct extent_map_tree *tree, struct extent_map *em)
 {
+	lockdep_assert_held_write(&tree->lock);
+
 	clear_bit(EXTENT_FLAG_LOGGING, &em->flags);
 	if (extent_map_in_tree(em))
 		try_merge_map(tree, em);
@@ -360,7 +368,7 @@ static void extent_map_device_set_bits(struct extent_map *em, unsigned bits)
 	int i;
 
 	for (i = 0; i < map->num_stripes; i++) {
-		struct btrfs_bio_stripe *stripe = &map->stripes[i];
+		struct btrfs_io_stripe *stripe = &map->stripes[i];
 		struct btrfs_device *device = stripe->dev;
 
 		set_extent_bits_nowait(&device->alloc_state, stripe->physical,
@@ -375,19 +383,22 @@ static void extent_map_device_clear_bits(struct extent_map *em, unsigned bits)
 	int i;
 
 	for (i = 0; i < map->num_stripes; i++) {
-		struct btrfs_bio_stripe *stripe = &map->stripes[i];
+		struct btrfs_io_stripe *stripe = &map->stripes[i];
 		struct btrfs_device *device = stripe->dev;
 
 		__clear_extent_bit(&device->alloc_state, stripe->physical,
 				   stripe->physical + stripe_size - 1, bits,
-				   0, 0, NULL, GFP_NOWAIT, NULL);
+				   NULL, GFP_NOWAIT, NULL);
 	}
 }
 
 /**
- * add_extent_mapping - add new extent map to the extent tree
+ * Add new extent map to the extent tree
+ *
  * @tree:	tree to insert new map in
  * @em:		map to insert
+ * @modified:	indicate whether the given @em should be added to the
+ *	        modified list, which indicates the extent needs to be logged
  *
  * Insert @em into @tree or perform a simple forward/backward merge with
  * existing mappings.  The extent_map struct passed in will be inserted
@@ -420,16 +431,13 @@ __lookup_extent_mapping(struct extent_map_tree *tree,
 {
 	struct extent_map *em;
 	struct rb_node *rb_node;
-	struct rb_node *prev = NULL;
-	struct rb_node *next = NULL;
+	struct rb_node *prev_or_next = NULL;
 	u64 end = range_end(start, len);
 
-	rb_node = __tree_search(&tree->map.rb_root, start, &prev, &next);
+	rb_node = __tree_search(&tree->map.rb_root, start, &prev_or_next);
 	if (!rb_node) {
-		if (prev)
-			rb_node = prev;
-		else if (next)
-			rb_node = next;
+		if (prev_or_next)
+			rb_node = prev_or_next;
 		else
 			return NULL;
 	}
@@ -487,6 +495,8 @@ struct extent_map *search_extent_mapping(struct extent_map_tree *tree,
  */
 void remove_extent_mapping(struct extent_map_tree *tree, struct extent_map *em)
 {
+	lockdep_assert_held_write(&tree->lock);
+
 	WARN_ON(test_bit(EXTENT_FLAG_PINNED, &em->flags));
 	rb_erase_cached(&em->rb_node, &tree->map);
 	if (!test_bit(EXTENT_FLAG_LOGGING, &em->flags))
@@ -501,6 +511,8 @@ void replace_extent_mapping(struct extent_map_tree *tree,
 			    struct extent_map *new,
 			    int modified)
 {
+	lockdep_assert_held_write(&tree->lock);
+
 	WARN_ON(test_bit(EXTENT_FLAG_PINNED, &cur->flags));
 	ASSERT(extent_map_in_tree(cur));
 	if (!test_bit(EXTENT_FLAG_LOGGING, &cur->flags))
@@ -574,12 +586,13 @@ static noinline int merge_extent_mapping(struct extent_map_tree *em_tree,
 }
 
 /**
- * btrfs_add_extent_mapping - add extent mapping into em_tree
- * @fs_info - used for tracepoint
- * @em_tree - the extent tree into which we want to insert the extent mapping
- * @em_in   - extent we are inserting
- * @start   - start of the logical range btrfs_get_extent() is requesting
- * @len     - length of the logical range btrfs_get_extent() is requesting
+ * Add extent mapping into em_tree
+ *
+ * @fs_info:  the filesystem
+ * @em_tree:  extent tree into which we want to insert the extent mapping
+ * @em_in:    extent we are inserting
+ * @start:    start of the logical range btrfs_get_extent() is requesting
+ * @len:      length of the logical range btrfs_get_extent() is requesting
  *
  * Note that @em_in's range may be different from [start, start+len),
  * but they must be overlapped.
@@ -648,3 +661,293 @@ int btrfs_add_extent_mapping(struct btrfs_fs_info *fs_info,
 	ASSERT(ret == 0 || ret == -EEXIST);
 	return ret;
 }
+
+/*
+ * Drop all extent maps from a tree in the fastest possible way, rescheduling
+ * if needed. This avoids searching the tree, from the root down to the first
+ * extent map, before each deletion.
+ */
+static void drop_all_extent_maps_fast(struct extent_map_tree *tree)
+{
+	write_lock(&tree->lock);
+	while (!RB_EMPTY_ROOT(&tree->map.rb_root)) {
+		struct extent_map *em;
+		struct rb_node *node;
+
+		node = rb_first_cached(&tree->map);
+		em = rb_entry(node, struct extent_map, rb_node);
+		clear_bit(EXTENT_FLAG_PINNED, &em->flags);
+		clear_bit(EXTENT_FLAG_LOGGING, &em->flags);
+		remove_extent_mapping(tree, em);
+		free_extent_map(em);
+		cond_resched_rwlock_write(&tree->lock);
+	}
+	write_unlock(&tree->lock);
+}
+
+/*
+ * Drop all extent maps in a given range.
+ *
+ * @inode:       The target inode.
+ * @start:       Start offset of the range.
+ * @end:         End offset of the range (inclusive value).
+ * @skip_pinned: Indicate if pinned extent maps should be ignored or not.
+ *
+ * This drops all the extent maps that intersect the given range [@start, @end].
+ * Extent maps that partially overlap the range and extend behind or beyond it,
+ * are split.
+ * The caller should have locked an appropriate file range in the inode's io
+ * tree before calling this function.
+ */
+void btrfs_drop_extent_map_range(struct btrfs_inode *inode, u64 start, u64 end,
+				 bool skip_pinned)
+{
+	struct extent_map *split;
+	struct extent_map *split2;
+	struct extent_map *em;
+	struct extent_map_tree *em_tree = &inode->extent_tree;
+	u64 len = end - start + 1;
+
+	WARN_ON(end < start);
+	if (end == (u64)-1) {
+		if (start == 0 && !skip_pinned) {
+			drop_all_extent_maps_fast(em_tree);
+			return;
+		}
+		len = (u64)-1;
+	} else {
+		/* Make end offset exclusive for use in the loop below. */
+		end++;
+	}
+
+	/*
+	 * It's ok if we fail to allocate the extent maps, see the comment near
+	 * the bottom of the loop below. We only need two spare extent maps in
+	 * the worst case, where the first extent map that intersects our range
+	 * starts before the range and the last extent map that intersects our
+	 * range ends after our range (and they might be the same extent map),
+	 * because we need to split those two extent maps at the boundaries.
+	 */
+	split = alloc_extent_map();
+	split2 = alloc_extent_map();
+
+	write_lock(&em_tree->lock);
+	em = lookup_extent_mapping(em_tree, start, len);
+
+	while (em) {
+		/* extent_map_end() returns exclusive value (last byte + 1). */
+		const u64 em_end = extent_map_end(em);
+		struct extent_map *next_em = NULL;
+		u64 gen;
+		unsigned long flags;
+		bool modified;
+		bool compressed;
+
+		if (em_end < end) {
+			next_em = next_extent_map(em);
+			if (next_em) {
+				if (next_em->start < end)
+					refcount_inc(&next_em->refs);
+				else
+					next_em = NULL;
+			}
+		}
+
+		if (skip_pinned && test_bit(EXTENT_FLAG_PINNED, &em->flags)) {
+			start = em_end;
+			if (end != (u64)-1)
+				len = start + len - em_end;
+			goto next;
+		}
+
+		clear_bit(EXTENT_FLAG_PINNED, &em->flags);
+		clear_bit(EXTENT_FLAG_LOGGING, &flags);
+		modified = !list_empty(&em->list);
+
+		/*
+		 * The extent map does not cross our target range, so no need to
+		 * split it, we can remove it directly.
+		 */
+		if (em->start >= start && em_end <= end)
+			goto remove_em;
+
+		flags = em->flags;
+		gen = em->generation;
+		compressed = test_bit(EXTENT_FLAG_COMPRESSED, &em->flags);
+
+		if (em->start < start) {
+			if (!split) {
+				split = split2;
+				split2 = NULL;
+				if (!split)
+					goto remove_em;
+			}
+			split->start = em->start;
+			split->len = start - em->start;
+
+			if (em->block_start < EXTENT_MAP_LAST_BYTE) {
+				split->orig_start = em->orig_start;
+				split->block_start = em->block_start;
+
+				if (compressed)
+					split->block_len = em->block_len;
+				else
+					split->block_len = split->len;
+				split->orig_block_len = max(split->block_len,
+						em->orig_block_len);
+				split->ram_bytes = em->ram_bytes;
+			} else {
+				split->orig_start = split->start;
+				split->block_len = 0;
+				split->block_start = em->block_start;
+				split->orig_block_len = 0;
+				split->ram_bytes = split->len;
+			}
+
+			split->generation = gen;
+			split->flags = flags;
+			split->compress_type = em->compress_type;
+			replace_extent_mapping(em_tree, em, split, modified);
+			free_extent_map(split);
+			split = split2;
+			split2 = NULL;
+		}
+		if (em_end > end) {
+			if (!split) {
+				split = split2;
+				split2 = NULL;
+				if (!split)
+					goto remove_em;
+			}
+			split->start = start + len;
+			split->len = em_end - (start + len);
+			split->block_start = em->block_start;
+			split->flags = flags;
+			split->compress_type = em->compress_type;
+			split->generation = gen;
+
+			if (em->block_start < EXTENT_MAP_LAST_BYTE) {
+				split->orig_block_len = max(em->block_len,
+						    em->orig_block_len);
+
+				split->ram_bytes = em->ram_bytes;
+				if (compressed) {
+					split->block_len = em->block_len;
+					split->orig_start = em->orig_start;
+				} else {
+					const u64 diff = start + len - em->start;
+
+					split->block_len = split->len;
+					split->block_start += diff;
+					split->orig_start = em->orig_start;
+				}
+			} else {
+				split->ram_bytes = split->len;
+				split->orig_start = split->start;
+				split->block_len = 0;
+				split->orig_block_len = 0;
+			}
+
+			if (extent_map_in_tree(em)) {
+				replace_extent_mapping(em_tree, em, split,
+						       modified);
+			} else {
+				int ret;
+
+				ret = add_extent_mapping(em_tree, split,
+							 modified);
+				/* Logic error, shouldn't happen. */
+				ASSERT(ret == 0);
+				if (WARN_ON(ret != 0) && modified)
+					btrfs_set_inode_full_sync(inode);
+			}
+			free_extent_map(split);
+			split = NULL;
+		}
+remove_em:
+		if (extent_map_in_tree(em)) {
+			/*
+			 * If the extent map is still in the tree it means that
+			 * either of the following is true:
+			 *
+			 * 1) It fits entirely in our range (doesn't end beyond
+			 *    it or starts before it);
+			 *
+			 * 2) It starts before our range and/or ends after our
+			 *    range, and we were not able to allocate the extent
+			 *    maps for split operations, @split and @split2.
+			 *
+			 * If we are at case 2) then we just remove the entire
+			 * extent map - this is fine since if anyone needs it to
+			 * access the subranges outside our range, will just
+			 * load it again from the subvolume tree's file extent
+			 * item. However if the extent map was in the list of
+			 * modified extents, then we must mark the inode for a
+			 * full fsync, otherwise a fast fsync will miss this
+			 * extent if it's new and needs to be logged.
+			 */
+			if ((em->start < start || em_end > end) && modified) {
+				ASSERT(!split);
+				btrfs_set_inode_full_sync(inode);
+			}
+			remove_extent_mapping(em_tree, em);
+		}
+
+		/*
+		 * Once for the tree reference (we replaced or removed the
+		 * extent map from the tree).
+		 */
+		free_extent_map(em);
+next:
+		/* Once for us (for our lookup reference). */
+		free_extent_map(em);
+
+		em = next_em;
+	}
+
+	write_unlock(&em_tree->lock);
+
+	free_extent_map(split);
+	free_extent_map(split2);
+}
+
+/*
+ * Replace a range in the inode's extent map tree with a new extent map.
+ *
+ * @inode:      The target inode.
+ * @new_em:     The new extent map to add to the inode's extent map tree.
+ * @modified:   Indicate if the new extent map should be added to the list of
+ *              modified extents (for fast fsync tracking).
+ *
+ * Drops all the extent maps in the inode's extent map tree that intersect the
+ * range of the new extent map and adds the new extent map to the tree.
+ * The caller should have locked an appropriate file range in the inode's io
+ * tree before calling this function.
+ */
+int btrfs_replace_extent_map_range(struct btrfs_inode *inode,
+				   struct extent_map *new_em,
+				   bool modified)
+{
+	const u64 end = new_em->start + new_em->len - 1;
+	struct extent_map_tree *tree = &inode->extent_tree;
+	int ret;
+
+	ASSERT(!extent_map_in_tree(new_em));
+
+	/*
+	 * The caller has locked an appropriate file range in the inode's io
+	 * tree, but getting -EEXIST when adding the new extent map can still
+	 * happen in case there are extents that partially cover the range, and
+	 * this is due to two tasks operating on different parts of the extent.
+	 * See commit 18e83ac75bfe67 ("Btrfs: fix unexpected EEXIST from
+	 * btrfs_get_extent") for an example and details.
+	 */
+	do {
+		btrfs_drop_extent_map_range(inode, new_em->start, end, false);
+		write_lock(&tree->lock);
+		ret = add_extent_mapping(tree, new_em, modified);
+		write_unlock(&tree->lock);
+	} while (ret == -EEXIST);
+
+	return ret;
+}
diff --git a/fs/btrfs/extent_map.h b/fs/btrfs/extent_map.h
index 8e217337dff9..ad311864272a 100644
--- a/fs/btrfs/extent_map.h
+++ b/fs/btrfs/extent_map.h
@@ -25,6 +25,8 @@ enum {
 	EXTENT_FLAG_FILLING,
 	/* filesystem extent mapping type */
 	EXTENT_FLAG_FS_MAPPING,
+	/* This em is merged from two or more physically adjacent ems */
+	EXTENT_FLAG_MERGED,
 };
 
 struct extent_map {
@@ -40,6 +42,12 @@ struct extent_map {
 	u64 ram_bytes;
 	u64 block_start;
 	u64 block_len;
+
+	/*
+	 * Generation of the extent map, for merged em it's the highest
+	 * generation of all merged ems.
+	 * For non-merged extents, it's from btrfs_file_extent_item::generation.
+	 */
 	u64 generation;
 	unsigned long flags;
 	/* Used for chunk mappings, flag EXTENT_FLAG_FS_MAPPING must be set */
@@ -55,6 +63,8 @@ struct extent_map_tree {
 	rwlock_t lock;
 };
 
+struct btrfs_inode;
+
 static inline int extent_map_in_tree(const struct extent_map *em)
 {
 	return !RB_EMPTY_NODE(&em->rb_node);
@@ -96,5 +106,11 @@ struct extent_map *search_extent_mapping(struct extent_map_tree *tree,
 int btrfs_add_extent_mapping(struct btrfs_fs_info *fs_info,
 			     struct extent_map_tree *em_tree,
 			     struct extent_map **em_in, u64 start, u64 len);
+void btrfs_drop_extent_map_range(struct btrfs_inode *inode,
+				 u64 start, u64 end,
+				 bool skip_pinned);
+int btrfs_replace_extent_map_range(struct btrfs_inode *inode,
+				   struct extent_map *new_em,
+				   bool modified);
 
 #endif
diff --git a/fs/btrfs/file-item.c b/fs/btrfs/file-item.c
index c2f365662d55..6bb9fa961a6a 100644
--- a/fs/btrfs/file-item.c
+++ b/fs/btrfs/file-item.c
@@ -9,6 +9,7 @@
 #include <linux/highmem.h>
 #include <linux/sched/mm.h>
 #include <crypto/hash.h>
+#include "misc.h"
 #include "ctree.h"
 #include "disk-io.h"
 #include "transaction.h"
@@ -23,6 +24,103 @@
 #define MAX_CSUM_ITEMS(r, size) (min_t(u32, __MAX_CSUM_ITEMS(r, size), \
 				       PAGE_SIZE))
 
+/**
+ * Set inode's size according to filesystem options
+ *
+ * @inode:      inode we want to update the disk_i_size for
+ * @new_i_size: i_size we want to set to, 0 if we use i_size
+ *
+ * With NO_HOLES set this simply sets the disk_is_size to whatever i_size_read()
+ * returns as it is perfectly fine with a file that has holes without hole file
+ * extent items.
+ *
+ * However without NO_HOLES we need to only return the area that is contiguous
+ * from the 0 offset of the file.  Otherwise we could end up adjust i_size up
+ * to an extent that has a gap in between.
+ *
+ * Finally new_i_size should only be set in the case of truncate where we're not
+ * ready to use i_size_read() as the limiter yet.
+ */
+void btrfs_inode_safe_disk_i_size_write(struct btrfs_inode *inode, u64 new_i_size)
+{
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
+	u64 start, end, i_size;
+	int ret;
+
+	i_size = new_i_size ?: i_size_read(&inode->vfs_inode);
+	if (btrfs_fs_incompat(fs_info, NO_HOLES)) {
+		inode->disk_i_size = i_size;
+		return;
+	}
+
+	spin_lock(&inode->lock);
+	ret = find_contiguous_extent_bit(&inode->file_extent_tree, 0, &start,
+					 &end, EXTENT_DIRTY);
+	if (!ret && start == 0)
+		i_size = min(i_size, end + 1);
+	else
+		i_size = 0;
+	inode->disk_i_size = i_size;
+	spin_unlock(&inode->lock);
+}
+
+/**
+ * Mark range within a file as having a new extent inserted
+ *
+ * @inode: inode being modified
+ * @start: start file offset of the file extent we've inserted
+ * @len:   logical length of the file extent item
+ *
+ * Call when we are inserting a new file extent where there was none before.
+ * Does not need to call this in the case where we're replacing an existing file
+ * extent, however if not sure it's fine to call this multiple times.
+ *
+ * The start and len must match the file extent item, so thus must be sectorsize
+ * aligned.
+ */
+int btrfs_inode_set_file_extent_range(struct btrfs_inode *inode, u64 start,
+				      u64 len)
+{
+	if (len == 0)
+		return 0;
+
+	ASSERT(IS_ALIGNED(start + len, inode->root->fs_info->sectorsize));
+
+	if (btrfs_fs_incompat(inode->root->fs_info, NO_HOLES))
+		return 0;
+	return set_extent_bits(&inode->file_extent_tree, start, start + len - 1,
+			       EXTENT_DIRTY);
+}
+
+/**
+ * Marks an inode range as not having a backing extent
+ *
+ * @inode: inode being modified
+ * @start: start file offset of the file extent we've inserted
+ * @len:   logical length of the file extent item
+ *
+ * Called when we drop a file extent, for example when we truncate.  Doesn't
+ * need to be called for cases where we're replacing a file extent, like when
+ * we've COWed a file extent.
+ *
+ * The start and len must match the file extent item, so thus must be sectorsize
+ * aligned.
+ */
+int btrfs_inode_clear_file_extent_range(struct btrfs_inode *inode, u64 start,
+					u64 len)
+{
+	if (len == 0)
+		return 0;
+
+	ASSERT(IS_ALIGNED(start + len, inode->root->fs_info->sectorsize) ||
+	       len == (u64)-1);
+
+	if (btrfs_fs_incompat(inode->root->fs_info, NO_HOLES))
+		return 0;
+	return clear_extent_bit(&inode->file_extent_tree, start,
+				start + len - 1, EXTENT_DIRTY, NULL);
+}
+
 static inline u32 max_ordered_sum_bytes(struct btrfs_fs_info *fs_info,
 					u16 csum_size)
 {
@@ -31,12 +129,20 @@ static inline u32 max_ordered_sum_bytes(struct btrfs_fs_info *fs_info,
 	return ncsums * fs_info->sectorsize;
 }
 
-int btrfs_insert_file_extent(struct btrfs_trans_handle *trans,
+/*
+ * Calculate the total size needed to allocate for an ordered sum structure
+ * spanning @bytes in the file.
+ */
+static int btrfs_ordered_sum_size(struct btrfs_fs_info *fs_info, unsigned long bytes)
+{
+	int num_sectors = (int)DIV_ROUND_UP(bytes, fs_info->sectorsize);
+
+	return sizeof(struct btrfs_ordered_sum) + num_sectors * fs_info->csum_size;
+}
+
+int btrfs_insert_hole_extent(struct btrfs_trans_handle *trans,
 			     struct btrfs_root *root,
-			     u64 objectid, u64 pos,
-			     u64 disk_offset, u64 disk_num_bytes,
-			     u64 num_bytes, u64 offset, u64 ram_bytes,
-			     u8 compression, u8 encryption, u16 other_encoding)
+			     u64 objectid, u64 pos, u64 num_bytes)
 {
 	int ret = 0;
 	struct btrfs_file_extent_item *item;
@@ -51,7 +157,6 @@ int btrfs_insert_file_extent(struct btrfs_trans_handle *trans,
 	file_key.offset = pos;
 	file_key.type = BTRFS_EXTENT_DATA_KEY;
 
-	path->leave_spinning = 1;
 	ret = btrfs_insert_empty_item(trans, root, path, &file_key,
 				      sizeof(*item));
 	if (ret < 0)
@@ -60,16 +165,16 @@ int btrfs_insert_file_extent(struct btrfs_trans_handle *trans,
 	leaf = path->nodes[0];
 	item = btrfs_item_ptr(leaf, path->slots[0],
 			      struct btrfs_file_extent_item);
-	btrfs_set_file_extent_disk_bytenr(leaf, item, disk_offset);
-	btrfs_set_file_extent_disk_num_bytes(leaf, item, disk_num_bytes);
-	btrfs_set_file_extent_offset(leaf, item, offset);
+	btrfs_set_file_extent_disk_bytenr(leaf, item, 0);
+	btrfs_set_file_extent_disk_num_bytes(leaf, item, 0);
+	btrfs_set_file_extent_offset(leaf, item, 0);
 	btrfs_set_file_extent_num_bytes(leaf, item, num_bytes);
-	btrfs_set_file_extent_ram_bytes(leaf, item, ram_bytes);
+	btrfs_set_file_extent_ram_bytes(leaf, item, num_bytes);
 	btrfs_set_file_extent_generation(leaf, item, trans->transid);
 	btrfs_set_file_extent_type(leaf, item, BTRFS_FILE_EXTENT_REG);
-	btrfs_set_file_extent_compression(leaf, item, compression);
-	btrfs_set_file_extent_encryption(leaf, item, encryption);
-	btrfs_set_file_extent_other_encoding(leaf, item, other_encoding);
+	btrfs_set_file_extent_compression(leaf, item, 0);
+	btrfs_set_file_extent_encryption(leaf, item, 0);
+	btrfs_set_file_extent_other_encoding(leaf, item, 0);
 
 	btrfs_mark_buffer_dirty(leaf);
 out:
@@ -90,7 +195,7 @@ btrfs_lookup_csum(struct btrfs_trans_handle *trans,
 	struct btrfs_csum_item *item;
 	struct extent_buffer *leaf;
 	u64 csum_offset = 0;
-	u16 csum_size = btrfs_super_csum_size(fs_info->super_copy);
+	const u32 csum_size = fs_info->csum_size;
 	int csums_in_item;
 
 	file_key.objectid = BTRFS_EXTENT_CSUM_OBJECTID;
@@ -110,8 +215,8 @@ btrfs_lookup_csum(struct btrfs_trans_handle *trans,
 			goto fail;
 
 		csum_offset = (bytenr - found_key.offset) >>
-				fs_info->sb->s_blocksize_bits;
-		csums_in_item = btrfs_item_size_nr(leaf, path->slots[0]);
+				fs_info->sectorsize_bits;
+		csums_in_item = btrfs_item_size(leaf, path->slots[0]);
 		csums_in_item /= csum_size;
 
 		if (csum_offset == csums_in_item) {
@@ -136,7 +241,6 @@ int btrfs_lookup_file_extent(struct btrfs_trans_handle *trans,
 			     struct btrfs_path *path, u64 objectid,
 			     u64 offset, int mod)
 {
-	int ret;
 	struct btrfs_key file_key;
 	int ins_len = mod < 0 ? -1 : 0;
 	int cow = mod != 0;
@@ -144,64 +248,189 @@ int btrfs_lookup_file_extent(struct btrfs_trans_handle *trans,
 	file_key.objectid = objectid;
 	file_key.offset = offset;
 	file_key.type = BTRFS_EXTENT_DATA_KEY;
-	ret = btrfs_search_slot(trans, root, &file_key, path, ins_len, cow);
+
+	return btrfs_search_slot(trans, root, &file_key, path, ins_len, cow);
+}
+
+/*
+ * Find checksums for logical bytenr range [disk_bytenr, disk_bytenr + len) and
+ * estore the result to @dst.
+ *
+ * Return >0 for the number of sectors we found.
+ * Return 0 for the range [disk_bytenr, disk_bytenr + sectorsize) has no csum
+ * for it. Caller may want to try next sector until one range is hit.
+ * Return <0 for fatal error.
+ */
+static int search_csum_tree(struct btrfs_fs_info *fs_info,
+			    struct btrfs_path *path, u64 disk_bytenr,
+			    u64 len, u8 *dst)
+{
+	struct btrfs_root *csum_root;
+	struct btrfs_csum_item *item = NULL;
+	struct btrfs_key key;
+	const u32 sectorsize = fs_info->sectorsize;
+	const u32 csum_size = fs_info->csum_size;
+	u32 itemsize;
+	int ret;
+	u64 csum_start;
+	u64 csum_len;
+
+	ASSERT(IS_ALIGNED(disk_bytenr, sectorsize) &&
+	       IS_ALIGNED(len, sectorsize));
+
+	/* Check if the current csum item covers disk_bytenr */
+	if (path->nodes[0]) {
+		item = btrfs_item_ptr(path->nodes[0], path->slots[0],
+				      struct btrfs_csum_item);
+		btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
+		itemsize = btrfs_item_size(path->nodes[0], path->slots[0]);
+
+		csum_start = key.offset;
+		csum_len = (itemsize / csum_size) * sectorsize;
+
+		if (in_range(disk_bytenr, csum_start, csum_len))
+			goto found;
+	}
+
+	/* Current item doesn't contain the desired range, search again */
+	btrfs_release_path(path);
+	csum_root = btrfs_csum_root(fs_info, disk_bytenr);
+	item = btrfs_lookup_csum(NULL, csum_root, path, disk_bytenr, 0);
+	if (IS_ERR(item)) {
+		ret = PTR_ERR(item);
+		goto out;
+	}
+	btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
+	itemsize = btrfs_item_size(path->nodes[0], path->slots[0]);
+
+	csum_start = key.offset;
+	csum_len = (itemsize / csum_size) * sectorsize;
+	ASSERT(in_range(disk_bytenr, csum_start, csum_len));
+
+found:
+	ret = (min(csum_start + csum_len, disk_bytenr + len) -
+		   disk_bytenr) >> fs_info->sectorsize_bits;
+	read_extent_buffer(path->nodes[0], dst, (unsigned long)item,
+			ret * csum_size);
+out:
+	if (ret == -ENOENT || ret == -EFBIG)
+		ret = 0;
+	return ret;
+}
+
+/*
+ * Locate the file_offset of @cur_disk_bytenr of a @bio.
+ *
+ * Bio of btrfs represents read range of
+ * [bi_sector << 9, bi_sector << 9 + bi_size).
+ * Knowing this, we can iterate through each bvec to locate the page belong to
+ * @cur_disk_bytenr and get the file offset.
+ *
+ * @inode is used to determine if the bvec page really belongs to @inode.
+ *
+ * Return 0 if we can't find the file offset
+ * Return >0 if we find the file offset and restore it to @file_offset_ret
+ */
+static int search_file_offset_in_bio(struct bio *bio, struct inode *inode,
+				     u64 disk_bytenr, u64 *file_offset_ret)
+{
+	struct bvec_iter iter;
+	struct bio_vec bvec;
+	u64 cur = bio->bi_iter.bi_sector << SECTOR_SHIFT;
+	int ret = 0;
+
+	bio_for_each_segment(bvec, bio, iter) {
+		struct page *page = bvec.bv_page;
+
+		if (cur > disk_bytenr)
+			break;
+		if (cur + bvec.bv_len <= disk_bytenr) {
+			cur += bvec.bv_len;
+			continue;
+		}
+		ASSERT(in_range(disk_bytenr, cur, bvec.bv_len));
+		if (page->mapping && page->mapping->host &&
+		    page->mapping->host == inode) {
+			ret = 1;
+			*file_offset_ret = page_offset(page) + bvec.bv_offset +
+					   disk_bytenr - cur;
+			break;
+		}
+	}
 	return ret;
 }
 
 /**
- * btrfs_lookup_bio_sums - Look up checksums for a bio.
+ * Lookup the checksum for the read bio in csum tree.
+ *
  * @inode: inode that the bio is for.
- * @bio: bio embedded in btrfs_io_bio.
- * @offset: Unless (u64)-1, look up checksums for this offset in the file.
- *          If (u64)-1, use the page offsets from the bio instead.
- * @dst: Buffer of size btrfs_super_csum_size() used to return checksum. If
- *       NULL, the checksum is returned in btrfs_io_bio(bio)->csum instead.
+ * @bio: bio to look up.
+ * @dst: Buffer of size nblocks * btrfs_super_csum_size() used to return
+ *       checksum (nblocks = bio->bi_iter.bi_size / fs_info->sectorsize). If
+ *       NULL, the checksum buffer is allocated and returned in
+ *       btrfs_bio(bio)->csum instead.
  *
  * Return: BLK_STS_RESOURCE if allocating memory fails, BLK_STS_OK otherwise.
  */
-blk_status_t btrfs_lookup_bio_sums(struct inode *inode, struct bio *bio,
-				   u64 offset, u8 *dst)
+blk_status_t btrfs_lookup_bio_sums(struct inode *inode, struct bio *bio, u8 *dst)
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	struct bio_vec bvec;
-	struct bvec_iter iter;
-	struct btrfs_io_bio *btrfs_bio = btrfs_io_bio(bio);
-	struct btrfs_csum_item *item = NULL;
 	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
+	struct btrfs_bio *bbio = NULL;
 	struct btrfs_path *path;
-	const bool page_offsets = (offset == (u64)-1);
+	const u32 sectorsize = fs_info->sectorsize;
+	const u32 csum_size = fs_info->csum_size;
+	u32 orig_len = bio->bi_iter.bi_size;
+	u64 orig_disk_bytenr = bio->bi_iter.bi_sector << SECTOR_SHIFT;
+	u64 cur_disk_bytenr;
 	u8 *csum;
-	u64 item_start_offset = 0;
-	u64 item_last_offset = 0;
-	u64 disk_bytenr;
-	u64 page_bytes_left;
-	u32 diff;
-	int nblocks;
+	const unsigned int nblocks = orig_len >> fs_info->sectorsize_bits;
 	int count = 0;
-	u16 csum_size = btrfs_super_csum_size(fs_info->super_copy);
+	blk_status_t ret = BLK_STS_OK;
+
+	if ((BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM) ||
+	    test_bit(BTRFS_FS_STATE_NO_CSUMS, &fs_info->fs_state))
+		return BLK_STS_OK;
 
+	/*
+	 * This function is only called for read bio.
+	 *
+	 * This means two things:
+	 * - All our csums should only be in csum tree
+	 *   No ordered extents csums, as ordered extents are only for write
+	 *   path.
+	 * - No need to bother any other info from bvec
+	 *   Since we're looking up csums, the only important info is the
+	 *   disk_bytenr and the length, which can be extracted from bi_iter
+	 *   directly.
+	 */
+	ASSERT(bio_op(bio) == REQ_OP_READ);
 	path = btrfs_alloc_path();
 	if (!path)
 		return BLK_STS_RESOURCE;
 
-	nblocks = bio->bi_iter.bi_size >> inode->i_sb->s_blocksize_bits;
 	if (!dst) {
+		bbio = btrfs_bio(bio);
+
 		if (nblocks * csum_size > BTRFS_BIO_INLINE_CSUM_SIZE) {
-			btrfs_bio->csum = kmalloc_array(nblocks, csum_size,
-							GFP_NOFS);
-			if (!btrfs_bio->csum) {
+			bbio->csum = kmalloc_array(nblocks, csum_size, GFP_NOFS);
+			if (!bbio->csum) {
 				btrfs_free_path(path);
 				return BLK_STS_RESOURCE;
 			}
 		} else {
-			btrfs_bio->csum = btrfs_bio->csum_inline;
+			bbio->csum = bbio->csum_inline;
 		}
-		csum = btrfs_bio->csum;
+		csum = bbio->csum;
 	} else {
 		csum = dst;
 	}
 
-	if (bio->bi_iter.bi_size > PAGE_SIZE * 8)
+	/*
+	 * If requested number of sectors is larger than one leaf can contain,
+	 * kick the readahead for csum tree.
+	 */
+	if (nblocks > fs_info->csums_per_leaf)
 		path->reada = READA_FORWARD;
 
 	/*
@@ -215,91 +444,75 @@ blk_status_t btrfs_lookup_bio_sums(struct inode *inode, struct bio *bio,
 		path->skip_locking = 1;
 	}
 
-	disk_bytenr = (u64)bio->bi_iter.bi_sector << 9;
-
-	bio_for_each_segment(bvec, bio, iter) {
-		page_bytes_left = bvec.bv_len;
-		if (count)
-			goto next;
-
-		if (page_offsets)
-			offset = page_offset(bvec.bv_page) + bvec.bv_offset;
-		count = btrfs_find_ordered_sum(inode, offset, disk_bytenr,
-					       csum, nblocks);
-		if (count)
-			goto found;
+	for (cur_disk_bytenr = orig_disk_bytenr;
+	     cur_disk_bytenr < orig_disk_bytenr + orig_len;
+	     cur_disk_bytenr += (count * sectorsize)) {
+		u64 search_len = orig_disk_bytenr + orig_len - cur_disk_bytenr;
+		unsigned int sector_offset;
+		u8 *csum_dst;
 
-		if (!item || disk_bytenr < item_start_offset ||
-		    disk_bytenr >= item_last_offset) {
-			struct btrfs_key found_key;
-			u32 item_size;
-
-			if (item)
-				btrfs_release_path(path);
-			item = btrfs_lookup_csum(NULL, fs_info->csum_root,
-						 path, disk_bytenr, 0);
-			if (IS_ERR(item)) {
-				count = 1;
-				memset(csum, 0, csum_size);
-				if (BTRFS_I(inode)->root->root_key.objectid ==
-				    BTRFS_DATA_RELOC_TREE_OBJECTID) {
-					set_extent_bits(io_tree, offset,
-						offset + fs_info->sectorsize - 1,
-						EXTENT_NODATASUM);
-				} else {
-					btrfs_info_rl(fs_info,
-						   "no csum found for inode %llu start %llu",
-					       btrfs_ino(BTRFS_I(inode)), offset);
-				}
-				item = NULL;
-				btrfs_release_path(path);
-				goto found;
-			}
-			btrfs_item_key_to_cpu(path->nodes[0], &found_key,
-					      path->slots[0]);
-
-			item_start_offset = found_key.offset;
-			item_size = btrfs_item_size_nr(path->nodes[0],
-						       path->slots[0]);
-			item_last_offset = item_start_offset +
-				(item_size / csum_size) *
-				fs_info->sectorsize;
-			item = btrfs_item_ptr(path->nodes[0], path->slots[0],
-					      struct btrfs_csum_item);
+		/*
+		 * Although both cur_disk_bytenr and orig_disk_bytenr is u64,
+		 * we're calculating the offset to the bio start.
+		 *
+		 * Bio size is limited to UINT_MAX, thus unsigned int is large
+		 * enough to contain the raw result, not to mention the right
+		 * shifted result.
+		 */
+		ASSERT(cur_disk_bytenr - orig_disk_bytenr < UINT_MAX);
+		sector_offset = (cur_disk_bytenr - orig_disk_bytenr) >>
+				fs_info->sectorsize_bits;
+		csum_dst = csum + sector_offset * csum_size;
+
+		count = search_csum_tree(fs_info, path, cur_disk_bytenr,
+					 search_len, csum_dst);
+		if (count < 0) {
+			ret = errno_to_blk_status(count);
+			if (bbio)
+				btrfs_bio_free_csum(bbio);
+			break;
 		}
+
 		/*
-		 * this byte range must be able to fit inside
-		 * a single leaf so it will also fit inside a u32
+		 * We didn't find a csum for this range.  We need to make sure
+		 * we complain loudly about this, because we are not NODATASUM.
+		 *
+		 * However for the DATA_RELOC inode we could potentially be
+		 * relocating data extents for a NODATASUM inode, so the inode
+		 * itself won't be marked with NODATASUM, but the extent we're
+		 * copying is in fact NODATASUM.  If we don't find a csum we
+		 * assume this is the case.
 		 */
-		diff = disk_bytenr - item_start_offset;
-		diff = diff / fs_info->sectorsize;
-		diff = diff * csum_size;
-		count = min_t(int, nblocks, (item_last_offset - disk_bytenr) >>
-					    inode->i_sb->s_blocksize_bits);
-		read_extent_buffer(path->nodes[0], csum,
-				   ((unsigned long)item) + diff,
-				   csum_size * count);
-found:
-		csum += count * csum_size;
-		nblocks -= count;
-next:
-		while (count > 0) {
-			count--;
-			disk_bytenr += fs_info->sectorsize;
-			offset += fs_info->sectorsize;
-			page_bytes_left -= fs_info->sectorsize;
-			if (!page_bytes_left)
-				break; /* move to next bio */
+		if (count == 0) {
+			memset(csum_dst, 0, csum_size);
+			count = 1;
+
+			if (BTRFS_I(inode)->root->root_key.objectid ==
+			    BTRFS_DATA_RELOC_TREE_OBJECTID) {
+				u64 file_offset;
+				int ret;
+
+				ret = search_file_offset_in_bio(bio, inode,
+						cur_disk_bytenr, &file_offset);
+				if (ret)
+					set_extent_bits(io_tree, file_offset,
+						file_offset + sectorsize - 1,
+						EXTENT_NODATASUM);
+			} else {
+				btrfs_warn_rl(fs_info,
+			"csum hole found for disk bytenr range [%llu, %llu)",
+				cur_disk_bytenr, cur_disk_bytenr + sectorsize);
+			}
 		}
 	}
 
-	WARN_ON_ONCE(count);
 	btrfs_free_path(path);
-	return BLK_STS_OK;
+	return ret;
 }
 
 int btrfs_lookup_csums_range(struct btrfs_root *root, u64 start, u64 end,
-			     struct list_head *list, int search_commit)
+			     struct list_head *list, int search_commit,
+			     bool nowait)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct btrfs_key key;
@@ -312,7 +525,7 @@ int btrfs_lookup_csums_range(struct btrfs_root *root, u64 start, u64 end,
 	int ret;
 	size_t size;
 	u64 csum_end;
-	u16 csum_size = btrfs_super_csum_size(fs_info->super_copy);
+	const u32 csum_size = fs_info->csum_size;
 
 	ASSERT(IS_ALIGNED(start, fs_info->sectorsize) &&
 	       IS_ALIGNED(end + 1, fs_info->sectorsize));
@@ -321,6 +534,7 @@ int btrfs_lookup_csums_range(struct btrfs_root *root, u64 start, u64 end,
 	if (!path)
 		return -ENOMEM;
 
+	path->nowait = nowait;
 	if (search_commit) {
 		path->skip_locking = 1;
 		path->reada = READA_FORWARD;
@@ -339,10 +553,9 @@ int btrfs_lookup_csums_range(struct btrfs_root *root, u64 start, u64 end,
 		btrfs_item_key_to_cpu(leaf, &key, path->slots[0] - 1);
 		if (key.objectid == BTRFS_EXTENT_CSUM_OBJECTID &&
 		    key.type == BTRFS_EXTENT_CSUM_KEY) {
-			offset = (start - key.offset) >>
-				 fs_info->sb->s_blocksize_bits;
+			offset = (start - key.offset) >> fs_info->sectorsize_bits;
 			if (offset * csum_size <
-			    btrfs_item_size_nr(leaf, path->slots[0] - 1))
+			    btrfs_item_size(leaf, path->slots[0] - 1))
 				path->slots[0]--;
 		}
 	}
@@ -367,7 +580,7 @@ int btrfs_lookup_csums_range(struct btrfs_root *root, u64 start, u64 end,
 		if (key.offset > start)
 			start = key.offset;
 
-		size = btrfs_item_size_nr(leaf, path->slots[0]);
+		size = btrfs_item_size(leaf, path->slots[0]);
 		csum_end = key.offset + (size / csum_size) * fs_info->sectorsize;
 		if (csum_end <= start) {
 			path->slots[0]++;
@@ -390,10 +603,9 @@ int btrfs_lookup_csums_range(struct btrfs_root *root, u64 start, u64 end,
 			sums->bytenr = start;
 			sums->len = (int)size;
 
-			offset = (start - key.offset) >>
-				fs_info->sb->s_blocksize_bits;
+			offset = (start - key.offset) >> fs_info->sectorsize_bits;
 			offset *= csum_size;
-			size >>= fs_info->sb->s_blocksize_bits;
+			size >>= fs_info->sectorsize_bits;
 
 			read_extent_buffer(path->nodes[0],
 					   sums->sums,
@@ -418,34 +630,34 @@ fail:
 	return ret;
 }
 
-/*
- * btrfs_csum_one_bio - Calculates checksums of the data contained inside a bio
+/**
+ * Calculate checksums of the data contained inside a bio
+ *
  * @inode:	 Owner of the data inside the bio
  * @bio:	 Contains the data to be checksummed
- * @file_start:  offset in file this bio begins to describe
- * @contig:	 Boolean. If true/1 means all bio vecs in this bio are
- *		 contiguous and they begin at @file_start in the file. False/0
- *		 means this bio can contains potentially discontigous bio vecs
- *		 so the logical offset of each should be calculated separately.
+ * @offset:      If (u64)-1, @bio may contain discontiguous bio vecs, so the
+ *               file offsets are determined from the page offsets in the bio.
+ *               Otherwise, this is the starting file offset of the bio vecs in
+ *               @bio, which must be contiguous.
+ * @one_ordered: If true, @bio only refers to one ordered extent.
  */
-blk_status_t btrfs_csum_one_bio(struct inode *inode, struct bio *bio,
-		       u64 file_start, int contig)
+blk_status_t btrfs_csum_one_bio(struct btrfs_inode *inode, struct bio *bio,
+				u64 offset, bool one_ordered)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
 	SHASH_DESC_ON_STACK(shash, fs_info->csum_shash);
 	struct btrfs_ordered_sum *sums;
 	struct btrfs_ordered_extent *ordered = NULL;
+	const bool use_page_offsets = (offset == (u64)-1);
 	char *data;
 	struct bvec_iter iter;
 	struct bio_vec bvec;
 	int index;
-	int nr_sectors;
+	unsigned int blockcount;
 	unsigned long total_bytes = 0;
 	unsigned long this_sum_bytes = 0;
 	int i;
-	u64 offset;
 	unsigned nofs_flag;
-	const u16 csum_size = btrfs_super_csum_size(fs_info->super_copy);
 
 	nofs_flag = memalloc_nofs_save();
 	sums = kvzalloc(btrfs_ordered_sum_size(fs_info, bio->bi_iter.bi_size),
@@ -458,32 +670,39 @@ blk_status_t btrfs_csum_one_bio(struct inode *inode, struct bio *bio,
 	sums->len = bio->bi_iter.bi_size;
 	INIT_LIST_HEAD(&sums->list);
 
-	if (contig)
-		offset = file_start;
-	else
-		offset = 0; /* shut up gcc */
-
-	sums->bytenr = (u64)bio->bi_iter.bi_sector << 9;
+	sums->bytenr = bio->bi_iter.bi_sector << 9;
 	index = 0;
 
 	shash->tfm = fs_info->csum_shash;
 
 	bio_for_each_segment(bvec, bio, iter) {
-		if (!contig)
+		if (use_page_offsets)
 			offset = page_offset(bvec.bv_page) + bvec.bv_offset;
 
 		if (!ordered) {
 			ordered = btrfs_lookup_ordered_extent(inode, offset);
-			BUG_ON(!ordered); /* Logic error */
+			/*
+			 * The bio range is not covered by any ordered extent,
+			 * must be a code logic error.
+			 */
+			if (unlikely(!ordered)) {
+				WARN(1, KERN_WARNING
+			"no ordered extent for root %llu ino %llu offset %llu\n",
+				     inode->root->root_key.objectid,
+				     btrfs_ino(inode), offset);
+				kvfree(sums);
+				return BLK_STS_IOERR;
+			}
 		}
 
-		nr_sectors = BTRFS_BYTES_TO_BLKS(fs_info,
+		blockcount = BTRFS_BYTES_TO_BLKS(fs_info,
 						 bvec.bv_len + fs_info->sectorsize
 						 - 1);
 
-		for (i = 0; i < nr_sectors; i++) {
-			if (offset >= ordered->file_offset + ordered->num_bytes ||
-			    offset < ordered->file_offset) {
+		for (i = 0; i < blockcount; i++) {
+			if (!one_ordered &&
+			    !in_range(offset, ordered->file_offset,
+				      ordered->num_bytes)) {
 				unsigned long bytes_left;
 
 				sums->len = this_sum_bytes;
@@ -502,19 +721,18 @@ blk_status_t btrfs_csum_one_bio(struct inode *inode, struct bio *bio,
 				ordered = btrfs_lookup_ordered_extent(inode,
 								offset);
 				ASSERT(ordered); /* Logic error */
-				sums->bytenr = ((u64)bio->bi_iter.bi_sector << 9)
+				sums->bytenr = (bio->bi_iter.bi_sector << 9)
 					+ total_bytes;
 				index = 0;
 			}
 
-			crypto_shash_init(shash);
-			data = kmap_atomic(bvec.bv_page);
-			crypto_shash_update(shash, data + bvec.bv_offset
-					    + (i * fs_info->sectorsize),
-					    fs_info->sectorsize);
-			kunmap_atomic(data);
-			crypto_shash_final(shash, (char *)(sums->sums + index));
-			index += csum_size;
+			data = bvec_kmap_local(&bvec);
+			crypto_shash_digest(shash,
+					    data + (i * fs_info->sectorsize),
+					    fs_info->sectorsize,
+					    sums->sums + index);
+			kunmap_local(data);
+			index += fs_info->csum_size;
 			offset += fs_info->sectorsize;
 			this_sum_bytes += fs_info->sectorsize;
 			total_bytes += fs_info->sectorsize;
@@ -544,14 +762,14 @@ static noinline void truncate_one_csum(struct btrfs_fs_info *fs_info,
 				       u64 bytenr, u64 len)
 {
 	struct extent_buffer *leaf;
-	u16 csum_size = btrfs_super_csum_size(fs_info->super_copy);
+	const u32 csum_size = fs_info->csum_size;
 	u64 csum_end;
 	u64 end_byte = bytenr + len;
-	u32 blocksize_bits = fs_info->sb->s_blocksize_bits;
+	u32 blocksize_bits = fs_info->sectorsize_bits;
 
 	leaf = path->nodes[0];
-	csum_end = btrfs_item_size_nr(leaf, path->slots[0]) / csum_size;
-	csum_end <<= fs_info->sb->s_blocksize_bits;
+	csum_end = btrfs_item_size(leaf, path->slots[0]) / csum_size;
+	csum_end <<= blocksize_bits;
 	csum_end += key->offset;
 
 	if (key->offset < bytenr && csum_end <= end_byte) {
@@ -597,11 +815,11 @@ int btrfs_del_csums(struct btrfs_trans_handle *trans,
 	u64 end_byte = bytenr + len;
 	u64 csum_end;
 	struct extent_buffer *leaf;
-	int ret;
-	u16 csum_size = btrfs_super_csum_size(fs_info->super_copy);
-	int blocksize_bits = fs_info->sb->s_blocksize_bits;
+	int ret = 0;
+	const u32 csum_size = fs_info->csum_size;
+	u32 blocksize_bits = fs_info->sectorsize_bits;
 
-	ASSERT(root == fs_info->csum_root ||
+	ASSERT(root->root_key.objectid == BTRFS_CSUM_TREE_OBJECTID ||
 	       root->root_key.objectid == BTRFS_TREE_LOG_OBJECTID);
 
 	path = btrfs_alloc_path();
@@ -613,9 +831,9 @@ int btrfs_del_csums(struct btrfs_trans_handle *trans,
 		key.offset = end_byte - 1;
 		key.type = BTRFS_EXTENT_CSUM_KEY;
 
-		path->leave_spinning = 1;
 		ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
 		if (ret > 0) {
+			ret = 0;
 			if (path->slots[0] == 0)
 				break;
 			path->slots[0]--;
@@ -634,7 +852,7 @@ int btrfs_del_csums(struct btrfs_trans_handle *trans,
 		if (key.offset >= end_byte)
 			break;
 
-		csum_end = btrfs_item_size_nr(leaf, path->slots[0]) / csum_size;
+		csum_end = btrfs_item_size(leaf, path->slots[0]) / csum_size;
 		csum_end <<= blocksize_bits;
 		csum_end += key.offset;
 
@@ -672,7 +890,7 @@ int btrfs_del_csums(struct btrfs_trans_handle *trans,
 			ret = btrfs_del_items(trans, root, path,
 					      path->slots[0], del_nr);
 			if (ret)
-				goto out;
+				break;
 			if (key.offset == bytenr)
 				break;
 		} else if (key.offset < bytenr && csum_end > end_byte) {
@@ -716,8 +934,9 @@ int btrfs_del_csums(struct btrfs_trans_handle *trans,
 			ret = btrfs_split_item(trans, root, path, &key, offset);
 			if (ret && ret != -EAGAIN) {
 				btrfs_abort_transaction(trans, ret);
-				goto out;
+				break;
 			}
+			ret = 0;
 
 			key.offset = end_byte - 1;
 		} else {
@@ -727,12 +946,41 @@ int btrfs_del_csums(struct btrfs_trans_handle *trans,
 		}
 		btrfs_release_path(path);
 	}
-	ret = 0;
-out:
 	btrfs_free_path(path);
 	return ret;
 }
 
+static int find_next_csum_offset(struct btrfs_root *root,
+				 struct btrfs_path *path,
+				 u64 *next_offset)
+{
+	const u32 nritems = btrfs_header_nritems(path->nodes[0]);
+	struct btrfs_key found_key;
+	int slot = path->slots[0] + 1;
+	int ret;
+
+	if (nritems == 0 || slot >= nritems) {
+		ret = btrfs_next_leaf(root, path);
+		if (ret < 0) {
+			return ret;
+		} else if (ret > 0) {
+			*next_offset = (u64)-1;
+			return 0;
+		}
+		slot = path->slots[0];
+	}
+
+	btrfs_item_key_to_cpu(path->nodes[0], &found_key, slot);
+
+	if (found_key.objectid != BTRFS_EXTENT_CSUM_OBJECTID ||
+	    found_key.type != BTRFS_EXTENT_CSUM_KEY)
+		*next_offset = (u64)-1;
+	else
+		*next_offset = found_key.offset;
+
+	return 0;
+}
+
 int btrfs_csum_file_blocks(struct btrfs_trans_handle *trans,
 			   struct btrfs_root *root,
 			   struct btrfs_ordered_sum *sums)
@@ -748,12 +996,11 @@ int btrfs_csum_file_blocks(struct btrfs_trans_handle *trans,
 	u64 total_bytes = 0;
 	u64 csum_offset;
 	u64 bytenr;
-	u32 nritems;
 	u32 ins_size;
 	int index = 0;
 	int found_next;
 	int ret;
-	u16 csum_size = btrfs_super_csum_size(fs_info->super_copy);
+	const u32 csum_size = fs_info->csum_size;
 
 	path = btrfs_alloc_path();
 	if (!path)
@@ -773,55 +1020,56 @@ again:
 		item_end = btrfs_item_ptr(leaf, path->slots[0],
 					  struct btrfs_csum_item);
 		item_end = (struct btrfs_csum_item *)((char *)item_end +
-			   btrfs_item_size_nr(leaf, path->slots[0]));
+			   btrfs_item_size(leaf, path->slots[0]));
 		goto found;
 	}
 	ret = PTR_ERR(item);
 	if (ret != -EFBIG && ret != -ENOENT)
-		goto fail_unlock;
+		goto out;
 
 	if (ret == -EFBIG) {
 		u32 item_size;
 		/* we found one, but it isn't big enough yet */
 		leaf = path->nodes[0];
-		item_size = btrfs_item_size_nr(leaf, path->slots[0]);
+		item_size = btrfs_item_size(leaf, path->slots[0]);
 		if ((item_size / csum_size) >=
 		    MAX_CSUM_ITEMS(fs_info, csum_size)) {
 			/* already at max size, make a new one */
 			goto insert;
 		}
 	} else {
-		int slot = path->slots[0] + 1;
-		/* we didn't find a csum item, insert one */
-		nritems = btrfs_header_nritems(path->nodes[0]);
-		if (!nritems || (path->slots[0] >= nritems - 1)) {
-			ret = btrfs_next_leaf(root, path);
-			if (ret == 1)
-				found_next = 1;
-			if (ret != 0)
-				goto insert;
-			slot = path->slots[0];
-		}
-		btrfs_item_key_to_cpu(path->nodes[0], &found_key, slot);
-		if (found_key.objectid != BTRFS_EXTENT_CSUM_OBJECTID ||
-		    found_key.type != BTRFS_EXTENT_CSUM_KEY) {
-			found_next = 1;
-			goto insert;
-		}
-		next_offset = found_key.offset;
+		/* We didn't find a csum item, insert one. */
+		ret = find_next_csum_offset(root, path, &next_offset);
+		if (ret < 0)
+			goto out;
 		found_next = 1;
 		goto insert;
 	}
 
 	/*
-	 * at this point, we know the tree has an item, but it isn't big
-	 * enough yet to put our csum in.  Grow it
+	 * At this point, we know the tree has a checksum item that ends at an
+	 * offset matching the start of the checksum range we want to insert.
+	 * We try to extend that item as much as possible and then add as many
+	 * checksums to it as they fit.
+	 *
+	 * First check if the leaf has enough free space for at least one
+	 * checksum. If it has go directly to the item extension code, otherwise
+	 * release the path and do a search for insertion before the extension.
 	 */
+	if (btrfs_leaf_free_space(leaf) >= csum_size) {
+		btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
+		csum_offset = (bytenr - found_key.offset) >>
+			fs_info->sectorsize_bits;
+		goto extend_csum;
+	}
+
 	btrfs_release_path(path);
+	path->search_for_extension = 1;
 	ret = btrfs_search_slot(trans, root, &file_key, path,
 				csum_size, 1);
+	path->search_for_extension = 0;
 	if (ret < 0)
-		goto fail_unlock;
+		goto out;
 
 	if (ret > 0) {
 		if (path->slots[0] == 0)
@@ -831,8 +1079,7 @@ again:
 
 	leaf = path->nodes[0];
 	btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
-	csum_offset = (bytenr - found_key.offset) >>
-			fs_info->sb->s_blocksize_bits;
+	csum_offset = (bytenr - found_key.offset) >> fs_info->sectorsize_bits;
 
 	if (found_key.type != BTRFS_EXTENT_CSUM_KEY ||
 	    found_key.objectid != BTRFS_EXTENT_CSUM_OBJECTID ||
@@ -840,30 +1087,64 @@ again:
 		goto insert;
 	}
 
-	if (csum_offset == btrfs_item_size_nr(leaf, path->slots[0]) /
+extend_csum:
+	if (csum_offset == btrfs_item_size(leaf, path->slots[0]) /
 	    csum_size) {
 		int extend_nr;
 		u64 tmp;
 		u32 diff;
-		u32 free_space;
-
-		if (btrfs_leaf_free_space(leaf) <
-				 sizeof(struct btrfs_item) + csum_size * 2)
-			goto insert;
 
-		free_space = btrfs_leaf_free_space(leaf) -
-					 sizeof(struct btrfs_item) - csum_size;
 		tmp = sums->len - total_bytes;
-		tmp >>= fs_info->sb->s_blocksize_bits;
+		tmp >>= fs_info->sectorsize_bits;
 		WARN_ON(tmp < 1);
+		extend_nr = max_t(int, 1, tmp);
+
+		/*
+		 * A log tree can already have checksum items with a subset of
+		 * the checksums we are trying to log. This can happen after
+		 * doing a sequence of partial writes into prealloc extents and
+		 * fsyncs in between, with a full fsync logging a larger subrange
+		 * of an extent for which a previous fast fsync logged a smaller
+		 * subrange. And this happens in particular due to merging file
+		 * extent items when we complete an ordered extent for a range
+		 * covered by a prealloc extent - this is done at
+		 * btrfs_mark_extent_written().
+		 *
+		 * So if we try to extend the previous checksum item, which has
+		 * a range that ends at the start of the range we want to insert,
+		 * make sure we don't extend beyond the start offset of the next
+		 * checksum item. If we are at the last item in the leaf, then
+		 * forget the optimization of extending and add a new checksum
+		 * item - it is not worth the complexity of releasing the path,
+		 * getting the first key for the next leaf, repeat the btree
+		 * search, etc, because log trees are temporary anyway and it
+		 * would only save a few bytes of leaf space.
+		 */
+		if (root->root_key.objectid == BTRFS_TREE_LOG_OBJECTID) {
+			if (path->slots[0] + 1 >=
+			    btrfs_header_nritems(path->nodes[0])) {
+				ret = find_next_csum_offset(root, path, &next_offset);
+				if (ret < 0)
+					goto out;
+				found_next = 1;
+				goto insert;
+			}
+
+			ret = find_next_csum_offset(root, path, &next_offset);
+			if (ret < 0)
+				goto out;
+
+			tmp = (next_offset - bytenr) >> fs_info->sectorsize_bits;
+			if (tmp <= INT_MAX)
+				extend_nr = min_t(int, extend_nr, tmp);
+		}
 
-		extend_nr = max_t(int, 1, (int)tmp);
 		diff = (csum_offset + extend_nr) * csum_size;
 		diff = min(diff,
 			   MAX_CSUM_ITEMS(fs_info, csum_size) * csum_size);
 
-		diff = diff - btrfs_item_size_nr(leaf, path->slots[0]);
-		diff = min(free_space, diff);
+		diff = diff - btrfs_item_size(leaf, path->slots[0]);
+		diff = min_t(u32, btrfs_leaf_free_space(leaf), diff);
 		diff /= csum_size;
 		diff *= csum_size;
 
@@ -879,9 +1160,9 @@ insert:
 		u64 tmp;
 
 		tmp = sums->len - total_bytes;
-		tmp >>= fs_info->sb->s_blocksize_bits;
+		tmp >>= fs_info->sectorsize_bits;
 		tmp = min(tmp, (next_offset - file_key.offset) >>
-					 fs_info->sb->s_blocksize_bits);
+					 fs_info->sectorsize_bits);
 
 		tmp = max_t(u64, 1, tmp);
 		tmp = min_t(u64, tmp, MAX_CSUM_ITEMS(fs_info, csum_size));
@@ -889,24 +1170,21 @@ insert:
 	} else {
 		ins_size = csum_size;
 	}
-	path->leave_spinning = 1;
 	ret = btrfs_insert_empty_item(trans, root, path, &file_key,
 				      ins_size);
-	path->leave_spinning = 0;
 	if (ret < 0)
-		goto fail_unlock;
+		goto out;
 	if (WARN_ON(ret != 0))
-		goto fail_unlock;
+		goto out;
 	leaf = path->nodes[0];
 csum:
 	item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_csum_item);
 	item_end = (struct btrfs_csum_item *)((unsigned char *)item +
-				      btrfs_item_size_nr(leaf, path->slots[0]));
+				      btrfs_item_size(leaf, path->slots[0]));
 	item = (struct btrfs_csum_item *)((unsigned char *)item +
 					  csum_offset * csum_size);
 found:
-	ins_size = (u32)(sums->len - total_bytes) >>
-		   fs_info->sb->s_blocksize_bits;
+	ins_size = (u32)(sums->len - total_bytes) >> fs_info->sectorsize_bits;
 	ins_size *= csum_size;
 	ins_size = min_t(u32, (unsigned long)item_end - (unsigned long)item,
 			      ins_size);
@@ -926,9 +1204,6 @@ found:
 out:
 	btrfs_free_path(path);
 	return ret;
-
-fail_unlock:
-	goto out;
 }
 
 void btrfs_extent_item_to_extent_map(struct btrfs_inode *inode,
@@ -949,19 +1224,9 @@ void btrfs_extent_item_to_extent_map(struct btrfs_inode *inode,
 
 	btrfs_item_key_to_cpu(leaf, &key, slot);
 	extent_start = key.offset;
-
-	if (type == BTRFS_FILE_EXTENT_REG ||
-	    type == BTRFS_FILE_EXTENT_PREALLOC) {
-		extent_end = extent_start +
-			btrfs_file_extent_num_bytes(leaf, fi);
-	} else if (type == BTRFS_FILE_EXTENT_INLINE) {
-		size_t size;
-		size = btrfs_file_extent_ram_bytes(leaf, fi);
-		extent_end = ALIGN(extent_start + size,
-				   fs_info->sectorsize);
-	}
-
+	extent_end = btrfs_file_extent_end(path);
 	em->ram_bytes = btrfs_file_extent_ram_bytes(leaf, fi);
+	em->generation = btrfs_file_extent_generation(leaf, fi);
 	if (type == BTRFS_FILE_EXTENT_REG ||
 	    type == BTRFS_FILE_EXTENT_PREALLOC) {
 		em->start = extent_start;
@@ -1007,3 +1272,30 @@ void btrfs_extent_item_to_extent_map(struct btrfs_inode *inode,
 			  root->root_key.objectid);
 	}
 }
+
+/*
+ * Returns the end offset (non inclusive) of the file extent item the given path
+ * points to. If it points to an inline extent, the returned offset is rounded
+ * up to the sector size.
+ */
+u64 btrfs_file_extent_end(const struct btrfs_path *path)
+{
+	const struct extent_buffer *leaf = path->nodes[0];
+	const int slot = path->slots[0];
+	struct btrfs_file_extent_item *fi;
+	struct btrfs_key key;
+	u64 end;
+
+	btrfs_item_key_to_cpu(leaf, &key, slot);
+	ASSERT(key.type == BTRFS_EXTENT_DATA_KEY);
+	fi = btrfs_item_ptr(leaf, slot, struct btrfs_file_extent_item);
+
+	if (btrfs_file_extent_type(leaf, fi) == BTRFS_FILE_EXTENT_INLINE) {
+		end = btrfs_file_extent_ram_bytes(leaf, fi);
+		end = ALIGN(key.offset + end, leaf->fs_info->sectorsize);
+	} else {
+		end = key.offset + btrfs_file_extent_num_bytes(leaf, fi);
+	}
+
+	return end;
+}
diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index a16da274c9aa..d01631d47806 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -16,6 +16,7 @@
 #include <linux/btrfs.h>
 #include <linux/uio.h>
 #include <linux/iversion.h>
+#include <linux/fsverity.h>
 #include "ctree.h"
 #include "disk-io.h"
 #include "transaction.h"
@@ -27,6 +28,8 @@
 #include "qgroup.h"
 #include "compression.h"
 #include "delalloc-space.h"
+#include "reflink.h"
+#include "subpage.h"
 
 static struct kmem_cache *btrfs_inode_defrag_cachep;
 /*
@@ -47,11 +50,14 @@ struct inode_defrag {
 	/* root objectid */
 	u64 root;
 
-	/* last offset we were able to defrag */
-	u64 last_offset;
-
-	/* if we've wrapped around back to zero once already */
-	int cycled;
+	/*
+	 * The extent size threshold for autodefrag.
+	 *
+	 * This value is different for compressed/non-compressed extents,
+	 * thus needs to be passed from higher layer.
+	 * (aka, inode_should_defrag())
+	 */
+	u32 extent_thresh;
 };
 
 static int __compare_inode_defrag(struct inode_defrag *defrag1,
@@ -104,8 +110,8 @@ static int __btrfs_add_inode_defrag(struct btrfs_inode *inode,
 			 */
 			if (defrag->transid < entry->transid)
 				entry->transid = defrag->transid;
-			if (defrag->last_offset > entry->last_offset)
-				entry->last_offset = defrag->last_offset;
+			entry->extent_thresh = min(defrag->extent_thresh,
+						   entry->extent_thresh);
 			return -EEXIST;
 		}
 	}
@@ -131,7 +137,7 @@ static inline int __need_auto_defrag(struct btrfs_fs_info *fs_info)
  * enabled
  */
 int btrfs_add_inode_defrag(struct btrfs_trans_handle *trans,
-			   struct btrfs_inode *inode)
+			   struct btrfs_inode *inode, u32 extent_thresh)
 {
 	struct btrfs_root *root = inode->root;
 	struct btrfs_fs_info *fs_info = root->fs_info;
@@ -157,6 +163,7 @@ int btrfs_add_inode_defrag(struct btrfs_trans_handle *trans,
 	defrag->ino = btrfs_ino(inode);
 	defrag->transid = transid;
 	defrag->root = root->root_key.objectid;
+	defrag->extent_thresh = extent_thresh;
 
 	spin_lock(&fs_info->defrag_inodes_lock);
 	if (!test_bit(BTRFS_INODE_IN_DEFRAG, &inode->runtime_flags)) {
@@ -176,34 +183,6 @@ int btrfs_add_inode_defrag(struct btrfs_trans_handle *trans,
 }
 
 /*
- * Requeue the defrag object. If there is a defrag object that points to
- * the same inode in the tree, we will merge them together (by
- * __btrfs_add_inode_defrag()) and free the one that we want to requeue.
- */
-static void btrfs_requeue_inode_defrag(struct btrfs_inode *inode,
-				       struct inode_defrag *defrag)
-{
-	struct btrfs_fs_info *fs_info = inode->root->fs_info;
-	int ret;
-
-	if (!__need_auto_defrag(fs_info))
-		goto out;
-
-	/*
-	 * Here we don't check the IN_DEFRAG flag, because we need merge
-	 * them together.
-	 */
-	spin_lock(&fs_info->defrag_inodes_lock);
-	ret = __btrfs_add_inode_defrag(inode, defrag);
-	spin_unlock(&fs_info->defrag_inodes_lock);
-	if (ret)
-		goto out;
-	return;
-out:
-	kmem_cache_free(btrfs_inode_defrag_cachep, defrag);
-}
-
-/*
  * pick the defragable inode that we want, if it doesn't exist, we will get
  * the next one.
  */
@@ -274,70 +253,55 @@ static int __btrfs_run_defrag_inode(struct btrfs_fs_info *fs_info,
 {
 	struct btrfs_root *inode_root;
 	struct inode *inode;
-	struct btrfs_key key;
 	struct btrfs_ioctl_defrag_range_args range;
-	int num_defrag;
-	int index;
-	int ret;
-
-	/* get the inode */
-	key.objectid = defrag->root;
-	key.type = BTRFS_ROOT_ITEM_KEY;
-	key.offset = (u64)-1;
+	int ret = 0;
+	u64 cur = 0;
 
-	index = srcu_read_lock(&fs_info->subvol_srcu);
+again:
+	if (test_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state))
+		goto cleanup;
+	if (!__need_auto_defrag(fs_info))
+		goto cleanup;
 
-	inode_root = btrfs_read_fs_root_no_name(fs_info, &key);
+	/* get the inode */
+	inode_root = btrfs_get_fs_root(fs_info, defrag->root, true);
 	if (IS_ERR(inode_root)) {
 		ret = PTR_ERR(inode_root);
 		goto cleanup;
 	}
 
-	key.objectid = defrag->ino;
-	key.type = BTRFS_INODE_ITEM_KEY;
-	key.offset = 0;
-	inode = btrfs_iget(fs_info->sb, &key, inode_root);
+	inode = btrfs_iget(fs_info->sb, defrag->ino, inode_root);
+	btrfs_put_root(inode_root);
 	if (IS_ERR(inode)) {
 		ret = PTR_ERR(inode);
 		goto cleanup;
 	}
-	srcu_read_unlock(&fs_info->subvol_srcu, index);
+
+	if (cur >= i_size_read(inode)) {
+		iput(inode);
+		goto cleanup;
+	}
 
 	/* do a chunk of defrag */
 	clear_bit(BTRFS_INODE_IN_DEFRAG, &BTRFS_I(inode)->runtime_flags);
 	memset(&range, 0, sizeof(range));
 	range.len = (u64)-1;
-	range.start = defrag->last_offset;
+	range.start = cur;
+	range.extent_thresh = defrag->extent_thresh;
 
 	sb_start_write(fs_info->sb);
-	num_defrag = btrfs_defrag_file(inode, NULL, &range, defrag->transid,
+	ret = btrfs_defrag_file(inode, NULL, &range, defrag->transid,
 				       BTRFS_DEFRAG_BATCH);
 	sb_end_write(fs_info->sb);
-	/*
-	 * if we filled the whole defrag batch, there
-	 * must be more work to do.  Queue this defrag
-	 * again
-	 */
-	if (num_defrag == BTRFS_DEFRAG_BATCH) {
-		defrag->last_offset = range.start;
-		btrfs_requeue_inode_defrag(BTRFS_I(inode), defrag);
-	} else if (defrag->last_offset && !defrag->cycled) {
-		/*
-		 * we didn't fill our defrag batch, but
-		 * we didn't start at zero.  Make sure we loop
-		 * around to the start of the file.
-		 */
-		defrag->last_offset = 0;
-		defrag->cycled = 1;
-		btrfs_requeue_inode_defrag(BTRFS_I(inode), defrag);
-	} else {
-		kmem_cache_free(btrfs_inode_defrag_cachep, defrag);
-	}
-
 	iput(inode);
-	return 0;
+
+	if (ret < 0)
+		goto cleanup;
+
+	cur = max(cur + fs_info->sectorsize, range.start);
+	goto again;
+
 cleanup:
-	srcu_read_unlock(&fs_info->subvol_srcu, index);
 	kmem_cache_free(btrfs_inode_defrag_cachep, defrag);
 	return ret;
 }
@@ -409,7 +373,7 @@ static noinline int btrfs_copy_from_user(loff_t pos, size_t write_bytes,
 		/*
 		 * Copy data from userspace to the current page
 		 */
-		copied = iov_iter_copy_from_user_atomic(page, i, offset, count);
+		copied = copy_page_from_iter_atomic(page, offset, count, i);
 
 		/* Flush processor's dcache for this page */
 		flush_dcache_page(page);
@@ -423,20 +387,19 @@ static noinline int btrfs_copy_from_user(loff_t pos, size_t write_bytes,
 		 * The rest of the btrfs_file_write code will fall
 		 * back to page at a time copies after we return 0.
 		 */
-		if (!PageUptodate(page) && copied < count)
-			copied = 0;
+		if (unlikely(copied < count)) {
+			if (!PageUptodate(page)) {
+				iov_iter_revert(i, copied);
+				copied = 0;
+			}
+			if (!copied)
+				break;
+		}
 
-		iov_iter_advance(i, copied);
 		write_bytes -= copied;
 		total_copied += copied;
-
-		/* Return to btrfs_file_write_iter to fault page */
-		if (unlikely(copied == 0))
-			break;
-
-		if (copied < PAGE_SIZE - offset) {
-			offset += copied;
-		} else {
+		offset += copied;
+		if (offset == PAGE_SIZE) {
 			pg++;
 			offset = 0;
 		}
@@ -447,9 +410,15 @@ static noinline int btrfs_copy_from_user(loff_t pos, size_t write_bytes,
 /*
  * unlocks pages after btrfs_file_write is done with them
  */
-static void btrfs_drop_pages(struct page **pages, size_t num_pages)
+static void btrfs_drop_pages(struct btrfs_fs_info *fs_info,
+			     struct page **pages, size_t num_pages,
+			     u64 pos, u64 copied)
 {
 	size_t i;
+	u64 block_start = round_down(pos, fs_info->sectorsize);
+	u64 block_len = round_up(pos + copied, fs_info->sectorsize) - block_start;
+
+	ASSERT(block_len <= U32_MAX);
 	for (i = 0; i < num_pages; i++) {
 		/* page checked is some magic around finding pages that
 		 * have been modified without going through btrfs_set_page_dirty
@@ -457,77 +426,44 @@ static void btrfs_drop_pages(struct page **pages, size_t num_pages)
 		 * accessed as prepare_pages should have marked them accessed
 		 * in prepare_pages via find_or_create_page()
 		 */
-		ClearPageChecked(pages[i]);
+		btrfs_page_clamp_clear_checked(fs_info, pages[i], block_start,
+					       block_len);
 		unlock_page(pages[i]);
 		put_page(pages[i]);
 	}
 }
 
-static int btrfs_find_new_delalloc_bytes(struct btrfs_inode *inode,
-					 const u64 start,
-					 const u64 len,
-					 struct extent_state **cached_state)
-{
-	u64 search_start = start;
-	const u64 end = start + len - 1;
-
-	while (search_start < end) {
-		const u64 search_len = end - search_start + 1;
-		struct extent_map *em;
-		u64 em_len;
-		int ret = 0;
-
-		em = btrfs_get_extent(inode, NULL, 0, search_start, search_len);
-		if (IS_ERR(em))
-			return PTR_ERR(em);
-
-		if (em->block_start != EXTENT_MAP_HOLE)
-			goto next;
-
-		em_len = em->len;
-		if (em->start < search_start)
-			em_len -= search_start - em->start;
-		if (em_len > search_len)
-			em_len = search_len;
-
-		ret = set_extent_bit(&inode->io_tree, search_start,
-				     search_start + em_len - 1,
-				     EXTENT_DELALLOC_NEW,
-				     NULL, cached_state, GFP_NOFS);
-next:
-		search_start = extent_map_end(em);
-		free_extent_map(em);
-		if (ret)
-			return ret;
-	}
-	return 0;
-}
-
 /*
- * after copy_from_user, pages need to be dirtied and we need to make
- * sure holes are created between the current EOF and the start of
- * any next extents (if required).
- *
- * this also makes the decision about creating an inline extent vs
- * doing real data extents, marking pages dirty and delalloc as required.
+ * After btrfs_copy_from_user(), update the following things for delalloc:
+ * - Mark newly dirtied pages as DELALLOC in the io tree.
+ *   Used to advise which range is to be written back.
+ * - Mark modified pages as Uptodate/Dirty and not needing COW fixup
+ * - Update inode size for past EOF write
  */
-int btrfs_dirty_pages(struct inode *inode, struct page **pages,
+int btrfs_dirty_pages(struct btrfs_inode *inode, struct page **pages,
 		      size_t num_pages, loff_t pos, size_t write_bytes,
-		      struct extent_state **cached)
+		      struct extent_state **cached, bool noreserve)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
 	int err = 0;
 	int i;
 	u64 num_bytes;
 	u64 start_pos;
 	u64 end_of_last_block;
 	u64 end_pos = pos + write_bytes;
-	loff_t isize = i_size_read(inode);
+	loff_t isize = i_size_read(&inode->vfs_inode);
 	unsigned int extra_bits = 0;
 
-	start_pos = pos & ~((u64) fs_info->sectorsize - 1);
+	if (write_bytes == 0)
+		return 0;
+
+	if (noreserve)
+		extra_bits |= EXTENT_NORESERVE;
+
+	start_pos = round_down(pos, fs_info->sectorsize);
 	num_bytes = round_up(write_bytes + pos - start_pos,
 			     fs_info->sectorsize);
+	ASSERT(num_bytes <= U32_MAX);
 
 	end_of_last_block = start_pos + num_bytes - 1;
 
@@ -535,27 +471,9 @@ int btrfs_dirty_pages(struct inode *inode, struct page **pages,
 	 * The pages may have already been dirty, clear out old accounting so
 	 * we can set things up properly
 	 */
-	clear_extent_bit(&BTRFS_I(inode)->io_tree, start_pos, end_of_last_block,
+	clear_extent_bit(&inode->io_tree, start_pos, end_of_last_block,
 			 EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG,
-			 0, 0, cached);
-
-	if (!btrfs_is_free_space_inode(BTRFS_I(inode))) {
-		if (start_pos >= isize &&
-		    !(BTRFS_I(inode)->flags & BTRFS_INODE_PREALLOC)) {
-			/*
-			 * There can't be any extents following eof in this case
-			 * so just set the delalloc new bit for the range
-			 * directly.
-			 */
-			extra_bits |= EXTENT_DELALLOC_NEW;
-		} else {
-			err = btrfs_find_new_delalloc_bytes(BTRFS_I(inode),
-							    start_pos,
-							    num_bytes, cached);
-			if (err)
-				return err;
-		}
-	}
+			 cached);
 
 	err = btrfs_set_extent_delalloc(inode, start_pos, end_of_last_block,
 					extra_bits, cached);
@@ -564,9 +482,10 @@ int btrfs_dirty_pages(struct inode *inode, struct page **pages,
 
 	for (i = 0; i < num_pages; i++) {
 		struct page *p = pages[i];
-		SetPageUptodate(p);
-		ClearPageChecked(p);
-		set_page_dirty(p);
+
+		btrfs_page_clamp_set_uptodate(fs_info, p, start_pos, num_bytes);
+		btrfs_page_clamp_clear_checked(fs_info, p, start_pos, num_bytes);
+		btrfs_page_clamp_set_dirty(fs_info, p, start_pos, num_bytes);
 	}
 
 	/*
@@ -575,164 +494,11 @@ int btrfs_dirty_pages(struct inode *inode, struct page **pages,
 	 * at this time.
 	 */
 	if (end_pos > isize)
-		i_size_write(inode, end_pos);
+		i_size_write(&inode->vfs_inode, end_pos);
 	return 0;
 }
 
 /*
- * this drops all the extents in the cache that intersect the range
- * [start, end].  Existing extents are split as required.
- */
-void btrfs_drop_extent_cache(struct btrfs_inode *inode, u64 start, u64 end,
-			     int skip_pinned)
-{
-	struct extent_map *em;
-	struct extent_map *split = NULL;
-	struct extent_map *split2 = NULL;
-	struct extent_map_tree *em_tree = &inode->extent_tree;
-	u64 len = end - start + 1;
-	u64 gen;
-	int ret;
-	int testend = 1;
-	unsigned long flags;
-	int compressed = 0;
-	bool modified;
-
-	WARN_ON(end < start);
-	if (end == (u64)-1) {
-		len = (u64)-1;
-		testend = 0;
-	}
-	while (1) {
-		int no_splits = 0;
-
-		modified = false;
-		if (!split)
-			split = alloc_extent_map();
-		if (!split2)
-			split2 = alloc_extent_map();
-		if (!split || !split2)
-			no_splits = 1;
-
-		write_lock(&em_tree->lock);
-		em = lookup_extent_mapping(em_tree, start, len);
-		if (!em) {
-			write_unlock(&em_tree->lock);
-			break;
-		}
-		flags = em->flags;
-		gen = em->generation;
-		if (skip_pinned && test_bit(EXTENT_FLAG_PINNED, &em->flags)) {
-			if (testend && em->start + em->len >= start + len) {
-				free_extent_map(em);
-				write_unlock(&em_tree->lock);
-				break;
-			}
-			start = em->start + em->len;
-			if (testend)
-				len = start + len - (em->start + em->len);
-			free_extent_map(em);
-			write_unlock(&em_tree->lock);
-			continue;
-		}
-		compressed = test_bit(EXTENT_FLAG_COMPRESSED, &em->flags);
-		clear_bit(EXTENT_FLAG_PINNED, &em->flags);
-		clear_bit(EXTENT_FLAG_LOGGING, &flags);
-		modified = !list_empty(&em->list);
-		if (no_splits)
-			goto next;
-
-		if (em->start < start) {
-			split->start = em->start;
-			split->len = start - em->start;
-
-			if (em->block_start < EXTENT_MAP_LAST_BYTE) {
-				split->orig_start = em->orig_start;
-				split->block_start = em->block_start;
-
-				if (compressed)
-					split->block_len = em->block_len;
-				else
-					split->block_len = split->len;
-				split->orig_block_len = max(split->block_len,
-						em->orig_block_len);
-				split->ram_bytes = em->ram_bytes;
-			} else {
-				split->orig_start = split->start;
-				split->block_len = 0;
-				split->block_start = em->block_start;
-				split->orig_block_len = 0;
-				split->ram_bytes = split->len;
-			}
-
-			split->generation = gen;
-			split->flags = flags;
-			split->compress_type = em->compress_type;
-			replace_extent_mapping(em_tree, em, split, modified);
-			free_extent_map(split);
-			split = split2;
-			split2 = NULL;
-		}
-		if (testend && em->start + em->len > start + len) {
-			u64 diff = start + len - em->start;
-
-			split->start = start + len;
-			split->len = em->start + em->len - (start + len);
-			split->flags = flags;
-			split->compress_type = em->compress_type;
-			split->generation = gen;
-
-			if (em->block_start < EXTENT_MAP_LAST_BYTE) {
-				split->orig_block_len = max(em->block_len,
-						    em->orig_block_len);
-
-				split->ram_bytes = em->ram_bytes;
-				if (compressed) {
-					split->block_len = em->block_len;
-					split->block_start = em->block_start;
-					split->orig_start = em->orig_start;
-				} else {
-					split->block_len = split->len;
-					split->block_start = em->block_start
-						+ diff;
-					split->orig_start = em->orig_start;
-				}
-			} else {
-				split->ram_bytes = split->len;
-				split->orig_start = split->start;
-				split->block_len = 0;
-				split->block_start = em->block_start;
-				split->orig_block_len = 0;
-			}
-
-			if (extent_map_in_tree(em)) {
-				replace_extent_mapping(em_tree, em, split,
-						       modified);
-			} else {
-				ret = add_extent_mapping(em_tree, split,
-							 modified);
-				ASSERT(ret == 0); /* Logic error */
-			}
-			free_extent_map(split);
-			split = NULL;
-		}
-next:
-		if (extent_map_in_tree(em))
-			remove_extent_mapping(em_tree, em);
-		write_unlock(&em_tree->lock);
-
-		/* once for us */
-		free_extent_map(em);
-		/* once for the tree*/
-		free_extent_map(em);
-	}
-	if (split)
-		free_extent_map(split);
-	if (split2)
-		free_extent_map(split2);
-}
-
-/*
  * this is very complex, but the basic idea is to drop all extents
  * in the range start - end.  hint_block is filled in with a block number
  * that would be a good hint to the block allocator for this file.
@@ -740,14 +506,16 @@ next:
  * If an extent intersects the range but is not entirely inside the range
  * it is either truncated or split.  Anything entirely inside the range
  * is deleted from the tree.
+ *
+ * Note: the VFS' inode number of bytes is not updated, it's up to the caller
+ * to deal with that. We set the field 'bytes_found' of the arguments structure
+ * with the number of allocated bytes found in the target range, so that the
+ * caller can update the inode's number of bytes in an atomic way when
+ * replacing extents in a range to avoid races with stat(2).
  */
-int __btrfs_drop_extents(struct btrfs_trans_handle *trans,
-			 struct btrfs_root *root, struct inode *inode,
-			 struct btrfs_path *path, u64 start, u64 end,
-			 u64 *drop_end, int drop_cache,
-			 int replace_extent,
-			 u32 extent_item_size,
-			 int *key_inserted)
+int btrfs_drop_extents(struct btrfs_trans_handle *trans,
+		       struct btrfs_root *root, struct btrfs_inode *inode,
+		       struct btrfs_drop_extents_args *args)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct extent_buffer *leaf;
@@ -755,13 +523,13 @@ int __btrfs_drop_extents(struct btrfs_trans_handle *trans,
 	struct btrfs_ref ref = { 0 };
 	struct btrfs_key key;
 	struct btrfs_key new_key;
-	u64 ino = btrfs_ino(BTRFS_I(inode));
-	u64 search_start = start;
+	u64 ino = btrfs_ino(inode);
+	u64 search_start = args->start;
 	u64 disk_bytenr = 0;
 	u64 num_bytes = 0;
 	u64 extent_offset = 0;
 	u64 extent_end = 0;
-	u64 last_end = start;
+	u64 last_end = args->start;
 	int del_nr = 0;
 	int del_slot = 0;
 	int extent_type;
@@ -770,23 +538,36 @@ int __btrfs_drop_extents(struct btrfs_trans_handle *trans,
 	int modify_tree = -1;
 	int update_refs;
 	int found = 0;
-	int leafs_visited = 0;
+	struct btrfs_path *path = args->path;
 
-	if (drop_cache)
-		btrfs_drop_extent_cache(BTRFS_I(inode), start, end - 1, 0);
+	args->bytes_found = 0;
+	args->extent_inserted = false;
 
-	if (start >= BTRFS_I(inode)->disk_i_size && !replace_extent)
+	/* Must always have a path if ->replace_extent is true */
+	ASSERT(!(args->replace_extent && !args->path));
+
+	if (!path) {
+		path = btrfs_alloc_path();
+		if (!path) {
+			ret = -ENOMEM;
+			goto out;
+		}
+	}
+
+	if (args->drop_cache)
+		btrfs_drop_extent_map_range(inode, args->start, args->end - 1, false);
+
+	if (args->start >= inode->disk_i_size && !args->replace_extent)
 		modify_tree = 0;
 
-	update_refs = (test_bit(BTRFS_ROOT_REF_COWS, &root->state) ||
-		       root == fs_info->tree_root);
+	update_refs = (root->root_key.objectid != BTRFS_TREE_LOG_OBJECTID);
 	while (1) {
 		recow = 0;
 		ret = btrfs_lookup_file_extent(trans, root, path, ino,
 					       search_start, modify_tree);
 		if (ret < 0)
 			break;
-		if (ret > 0 && path->slots[0] > 0 && search_start == start) {
+		if (ret > 0 && path->slots[0] > 0 && search_start == args->start) {
 			leaf = path->nodes[0];
 			btrfs_item_key_to_cpu(leaf, &key, path->slots[0] - 1);
 			if (key.objectid == ino &&
@@ -794,7 +575,6 @@ int __btrfs_drop_extents(struct btrfs_trans_handle *trans,
 				path->slots[0]--;
 		}
 		ret = 0;
-		leafs_visited++;
 next_slot:
 		leaf = path->nodes[0];
 		if (path->slots[0] >= btrfs_header_nritems(leaf)) {
@@ -806,7 +586,6 @@ next_slot:
 				ret = 0;
 				break;
 			}
-			leafs_visited++;
 			leaf = path->nodes[0];
 			recow = 1;
 		}
@@ -821,7 +600,7 @@ next_slot:
 			path->slots[0]++;
 			goto next_slot;
 		}
-		if (key.type > BTRFS_EXTENT_DATA_KEY || key.offset >= end)
+		if (key.type > BTRFS_EXTENT_DATA_KEY || key.offset >= args->end)
 			break;
 
 		fi = btrfs_item_ptr(leaf, path->slots[0],
@@ -863,7 +642,7 @@ next_slot:
 		}
 
 		found = 1;
-		search_start = max(key.offset, start);
+		search_start = max(key.offset, args->start);
 		if (recow || !modify_tree) {
 			modify_tree = -1;
 			btrfs_release_path(path);
@@ -874,7 +653,7 @@ next_slot:
 		 *     | - range to drop - |
 		 *  | -------- extent -------- |
 		 */
-		if (start > key.offset && end < extent_end) {
+		if (args->start > key.offset && args->end < extent_end) {
 			BUG_ON(del_nr > 0);
 			if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
 				ret = -EOPNOTSUPP;
@@ -882,7 +661,7 @@ next_slot:
 			}
 
 			memcpy(&new_key, &key, sizeof(new_key));
-			new_key.offset = start;
+			new_key.offset = args->start;
 			ret = btrfs_duplicate_item(trans, root, path,
 						   &new_key);
 			if (ret == -EAGAIN) {
@@ -896,15 +675,15 @@ next_slot:
 			fi = btrfs_item_ptr(leaf, path->slots[0] - 1,
 					    struct btrfs_file_extent_item);
 			btrfs_set_file_extent_num_bytes(leaf, fi,
-							start - key.offset);
+							args->start - key.offset);
 
 			fi = btrfs_item_ptr(leaf, path->slots[0],
 					    struct btrfs_file_extent_item);
 
-			extent_offset += start - key.offset;
+			extent_offset += args->start - key.offset;
 			btrfs_set_file_extent_offset(leaf, fi, extent_offset);
 			btrfs_set_file_extent_num_bytes(leaf, fi,
-							extent_end - start);
+							extent_end - args->start);
 			btrfs_mark_buffer_dirty(leaf);
 
 			if (update_refs && disk_bytenr > 0) {
@@ -914,11 +693,12 @@ next_slot:
 				btrfs_init_data_ref(&ref,
 						root->root_key.objectid,
 						new_key.objectid,
-						start - extent_offset);
+						args->start - extent_offset,
+						0, false);
 				ret = btrfs_inc_extent_ref(trans, &ref);
 				BUG_ON(ret); /* -ENOMEM */
 			}
-			key.offset = start;
+			key.offset = args->start;
 		}
 		/*
 		 * From here on out we will have actually dropped something, so
@@ -930,23 +710,23 @@ next_slot:
 		 *  | ---- range to drop ----- |
 		 *      | -------- extent -------- |
 		 */
-		if (start <= key.offset && end < extent_end) {
+		if (args->start <= key.offset && args->end < extent_end) {
 			if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
 				ret = -EOPNOTSUPP;
 				break;
 			}
 
 			memcpy(&new_key, &key, sizeof(new_key));
-			new_key.offset = end;
+			new_key.offset = args->end;
 			btrfs_set_item_key_safe(fs_info, path, &new_key);
 
-			extent_offset += end - key.offset;
+			extent_offset += args->end - key.offset;
 			btrfs_set_file_extent_offset(leaf, fi, extent_offset);
 			btrfs_set_file_extent_num_bytes(leaf, fi,
-							extent_end - end);
+							extent_end - args->end);
 			btrfs_mark_buffer_dirty(leaf);
 			if (update_refs && disk_bytenr > 0)
-				inode_sub_bytes(inode, end - key.offset);
+				args->bytes_found += args->end - key.offset;
 			break;
 		}
 
@@ -955,7 +735,7 @@ next_slot:
 		 *       | ---- range to drop ----- |
 		 *  | -------- extent -------- |
 		 */
-		if (start > key.offset && end >= extent_end) {
+		if (args->start > key.offset && args->end >= extent_end) {
 			BUG_ON(del_nr > 0);
 			if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
 				ret = -EOPNOTSUPP;
@@ -963,11 +743,11 @@ next_slot:
 			}
 
 			btrfs_set_file_extent_num_bytes(leaf, fi,
-							start - key.offset);
+							args->start - key.offset);
 			btrfs_mark_buffer_dirty(leaf);
 			if (update_refs && disk_bytenr > 0)
-				inode_sub_bytes(inode, extent_end - start);
-			if (end == extent_end)
+				args->bytes_found += extent_end - args->start;
+			if (args->end == extent_end)
 				break;
 
 			path->slots[0]++;
@@ -978,7 +758,7 @@ next_slot:
 		 *  | ---- range to drop ----- |
 		 *    | ------ extent ------ |
 		 */
-		if (start <= key.offset && end >= extent_end) {
+		if (args->start <= key.offset && args->end >= extent_end) {
 delete_extent_item:
 			if (del_nr == 0) {
 				del_slot = path->slots[0];
@@ -990,8 +770,7 @@ delete_extent_item:
 
 			if (update_refs &&
 			    extent_type == BTRFS_FILE_EXTENT_INLINE) {
-				inode_sub_bytes(inode,
-						extent_end - key.offset);
+				args->bytes_found += extent_end - key.offset;
 				extent_end = ALIGN(extent_end,
 						   fs_info->sectorsize);
 			} else if (update_refs && disk_bytenr > 0) {
@@ -1001,14 +780,14 @@ delete_extent_item:
 				btrfs_init_data_ref(&ref,
 						root->root_key.objectid,
 						key.objectid,
-						key.offset - extent_offset);
+						key.offset - extent_offset, 0,
+						false);
 				ret = btrfs_free_extent(trans, &ref);
 				BUG_ON(ret); /* -ENOMEM */
-				inode_sub_bytes(inode,
-						extent_end - key.offset);
+				args->bytes_found += extent_end - key.offset;
 			}
 
-			if (end == extent_end)
+			if (args->end == extent_end)
 				break;
 
 			if (path->slots[0] + 1 < btrfs_header_nritems(leaf)) {
@@ -1038,7 +817,7 @@ delete_extent_item:
 		 * Set path->slots[0] to first slot, so that after the delete
 		 * if items are move off from our leaf to its immediate left or
 		 * right neighbor leafs, we end up with a correct and adjusted
-		 * path->slots[0] for our insertion (if replace_extent != 0).
+		 * path->slots[0] for our insertion (if args->replace_extent).
 		 */
 		path->slots[0] = del_slot;
 		ret = btrfs_del_items(trans, root, path, del_slot, del_nr);
@@ -1052,15 +831,14 @@ delete_extent_item:
 	 * which case it unlocked our path, so check path->locks[0] matches a
 	 * write lock.
 	 */
-	if (!ret && replace_extent && leafs_visited == 1 &&
-	    (path->locks[0] == BTRFS_WRITE_LOCK_BLOCKING ||
-	     path->locks[0] == BTRFS_WRITE_LOCK) &&
+	if (!ret && args->replace_extent &&
+	    path->locks[0] == BTRFS_WRITE_LOCK &&
 	    btrfs_leaf_free_space(leaf) >=
-	    sizeof(struct btrfs_item) + extent_item_size) {
+	    sizeof(struct btrfs_item) + args->extent_item_size) {
 
 		key.objectid = ino;
 		key.type = BTRFS_EXTENT_DATA_KEY;
-		key.offset = start;
+		key.offset = args->start;
 		if (!del_nr && path->slots[0] < btrfs_header_nritems(leaf)) {
 			struct btrfs_key slot_key;
 
@@ -1068,34 +846,17 @@ delete_extent_item:
 			if (btrfs_comp_cpu_keys(&key, &slot_key) > 0)
 				path->slots[0]++;
 		}
-		setup_items_for_insert(root, path, &key,
-				       &extent_item_size,
-				       extent_item_size,
-				       sizeof(struct btrfs_item) +
-				       extent_item_size, 1);
-		*key_inserted = 1;
+		btrfs_setup_item_for_insert(root, path, &key, args->extent_item_size);
+		args->extent_inserted = true;
 	}
 
-	if (!replace_extent || !(*key_inserted))
+	if (!args->path)
+		btrfs_free_path(path);
+	else if (!args->extent_inserted)
 		btrfs_release_path(path);
-	if (drop_end)
-		*drop_end = found ? min(end, last_end) : end;
-	return ret;
-}
-
-int btrfs_drop_extents(struct btrfs_trans_handle *trans,
-		       struct btrfs_root *root, struct inode *inode, u64 start,
-		       u64 end, int drop_cache)
-{
-	struct btrfs_path *path;
-	int ret;
+out:
+	args->drop_end = found ? min(args->end, last_end) : args->end;
 
-	path = btrfs_alloc_path();
-	if (!path)
-		return -ENOMEM;
-	ret = __btrfs_drop_extents(trans, root, inode, path, start, end, NULL,
-				   drop_cache, 0, 0, NULL);
-	btrfs_free_path(path);
 	return ret;
 }
 
@@ -1160,7 +921,7 @@ int btrfs_mark_extent_written(struct btrfs_trans_handle *trans,
 	int del_nr = 0;
 	int del_slot = 0;
 	int recow;
-	int ret;
+	int ret = 0;
 	u64 ino = btrfs_ino(inode);
 
 	path = btrfs_alloc_path();
@@ -1296,7 +1057,7 @@ again:
 		btrfs_init_generic_ref(&ref, BTRFS_ADD_DELAYED_REF, bytenr,
 				       num_bytes, 0);
 		btrfs_init_data_ref(&ref, root->root_key.objectid, ino,
-				    orig_offset);
+				    orig_offset, 0, false);
 		ret = btrfs_inc_extent_ref(trans, &ref);
 		if (ret) {
 			btrfs_abort_transaction(trans, ret);
@@ -1321,7 +1082,8 @@ again:
 	other_end = 0;
 	btrfs_init_generic_ref(&ref, BTRFS_DROP_DELAYED_REF, bytenr,
 			       num_bytes, 0);
-	btrfs_init_data_ref(&ref, root->root_key.objectid, ino, orig_offset);
+	btrfs_init_data_ref(&ref, root->root_key.objectid, ino, orig_offset,
+			    0, false);
 	if (extent_mergeable(leaf, path->slots[0] + 1,
 			     ino, bytenr, orig_offset,
 			     &other_start, &other_end)) {
@@ -1381,7 +1143,7 @@ again:
 	}
 out:
 	btrfs_free_path(path);
-	return 0;
+	return ret;
 }
 
 /*
@@ -1392,11 +1154,12 @@ static int prepare_uptodate_page(struct inode *inode,
 				 struct page *page, u64 pos,
 				 bool force_uptodate)
 {
+	struct folio *folio = page_folio(page);
 	int ret = 0;
 
 	if (((pos & (PAGE_SIZE - 1)) || force_uptodate) &&
 	    !PageUptodate(page)) {
-		ret = btrfs_readpage(NULL, page);
+		ret = btrfs_read_folio(NULL, folio);
 		if (ret)
 			return ret;
 		lock_page(page);
@@ -1404,7 +1167,18 @@ static int prepare_uptodate_page(struct inode *inode,
 			unlock_page(page);
 			return -EIO;
 		}
-		if (page->mapping != inode->i_mapping) {
+
+		/*
+		 * Since btrfs_read_folio() will unlock the folio before it
+		 * returns, there is a window where btrfs_release_folio() can be
+		 * called to release the page.  Here we check both inode
+		 * mapping and PagePrivate() to make sure the page was not
+		 * released.
+		 *
+		 * The private flag check is essential for subpage as we need
+		 * to store extra bitmap using page->private.
+		 */
+		if (page->mapping != inode->i_mapping || !PagePrivate(page)) {
 			unlock_page(page);
 			return -EAGAIN;
 		}
@@ -1412,26 +1186,60 @@ static int prepare_uptodate_page(struct inode *inode,
 	return 0;
 }
 
+static unsigned int get_prepare_fgp_flags(bool nowait)
+{
+	unsigned int fgp_flags = FGP_LOCK | FGP_ACCESSED | FGP_CREAT;
+
+	if (nowait)
+		fgp_flags |= FGP_NOWAIT;
+
+	return fgp_flags;
+}
+
+static gfp_t get_prepare_gfp_flags(struct inode *inode, bool nowait)
+{
+	gfp_t gfp;
+
+	gfp = btrfs_alloc_write_mask(inode->i_mapping);
+	if (nowait) {
+		gfp &= ~__GFP_DIRECT_RECLAIM;
+		gfp |= GFP_NOWAIT;
+	}
+
+	return gfp;
+}
+
 /*
  * this just gets pages into the page cache and locks them down.
  */
 static noinline int prepare_pages(struct inode *inode, struct page **pages,
 				  size_t num_pages, loff_t pos,
-				  size_t write_bytes, bool force_uptodate)
+				  size_t write_bytes, bool force_uptodate,
+				  bool nowait)
 {
 	int i;
 	unsigned long index = pos >> PAGE_SHIFT;
-	gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping);
+	gfp_t mask = get_prepare_gfp_flags(inode, nowait);
+	unsigned int fgp_flags = get_prepare_fgp_flags(nowait);
 	int err = 0;
 	int faili;
 
 	for (i = 0; i < num_pages; i++) {
 again:
-		pages[i] = find_or_create_page(inode->i_mapping, index + i,
-					       mask | __GFP_WRITE);
+		pages[i] = pagecache_get_page(inode->i_mapping, index + i,
+					      fgp_flags, mask | __GFP_WRITE);
 		if (!pages[i]) {
 			faili = i - 1;
-			err = -ENOMEM;
+			if (nowait)
+				err = -EAGAIN;
+			else
+				err = -ENOMEM;
+			goto fail;
+		}
+
+		err = set_page_extent_mapped(pages[i]);
+		if (err < 0) {
+			faili = i;
 			goto fail;
 		}
 
@@ -1443,7 +1251,7 @@ again:
 						    pos + write_bytes, false);
 		if (err) {
 			put_page(pages[i]);
-			if (err == -EAGAIN) {
+			if (!nowait && err == -EAGAIN) {
 				err = 0;
 				goto again;
 			}
@@ -1478,7 +1286,7 @@ static noinline int
 lock_and_cleanup_extent_if_need(struct btrfs_inode *inode, struct page **pages,
 				size_t num_pages, loff_t pos,
 				size_t write_bytes,
-				u64 *lockstart, u64 *lockend,
+				u64 *lockstart, u64 *lockend, bool nowait,
 				struct extent_state **cached_state)
 {
 	struct btrfs_fs_info *fs_info = inode->root->fs_info;
@@ -1488,28 +1296,37 @@ lock_and_cleanup_extent_if_need(struct btrfs_inode *inode, struct page **pages,
 	int ret = 0;
 
 	start_pos = round_down(pos, fs_info->sectorsize);
-	last_pos = start_pos
-		+ round_up(pos + write_bytes - start_pos,
-			   fs_info->sectorsize) - 1;
+	last_pos = round_up(pos + write_bytes, fs_info->sectorsize) - 1;
 
 	if (start_pos < inode->vfs_inode.i_size) {
 		struct btrfs_ordered_extent *ordered;
 
-		lock_extent_bits(&inode->io_tree, start_pos, last_pos,
-				cached_state);
+		if (nowait) {
+			if (!try_lock_extent(&inode->io_tree, start_pos, last_pos)) {
+				for (i = 0; i < num_pages; i++) {
+					unlock_page(pages[i]);
+					put_page(pages[i]);
+					pages[i] = NULL;
+				}
+
+				return -EAGAIN;
+			}
+		} else {
+			lock_extent(&inode->io_tree, start_pos, last_pos, cached_state);
+		}
+
 		ordered = btrfs_lookup_ordered_range(inode, start_pos,
 						     last_pos - start_pos + 1);
 		if (ordered &&
 		    ordered->file_offset + ordered->num_bytes > start_pos &&
 		    ordered->file_offset <= last_pos) {
-			unlock_extent_cached(&inode->io_tree, start_pos,
-					last_pos, cached_state);
+			unlock_extent(&inode->io_tree, start_pos, last_pos,
+				      cached_state);
 			for (i = 0; i < num_pages; i++) {
 				unlock_page(pages[i]);
 				put_page(pages[i]);
 			}
-			btrfs_start_ordered_extent(&inode->vfs_inode,
-					ordered, 1);
+			btrfs_start_ordered_extent(ordered, 1);
 			btrfs_put_ordered_extent(ordered);
 			return -EAGAIN;
 		}
@@ -1522,29 +1339,36 @@ lock_and_cleanup_extent_if_need(struct btrfs_inode *inode, struct page **pages,
 	}
 
 	/*
-	 * It's possible the pages are dirty right now, but we don't want
-	 * to clean them yet because copy_from_user may catch a page fault
-	 * and we might have to fall back to one page at a time.  If that
-	 * happens, we'll unlock these pages and we'd have a window where
-	 * reclaim could sneak in and drop the once-dirty page on the floor
-	 * without writing it.
-	 *
-	 * We have the pages locked and the extent range locked, so there's
-	 * no way someone can start IO on any dirty pages in this range.
-	 *
-	 * We'll call btrfs_dirty_pages() later on, and that will flip around
-	 * delalloc bits and dirty the pages as required.
+	 * We should be called after prepare_pages() which should have locked
+	 * all pages in the range.
 	 */
-	for (i = 0; i < num_pages; i++) {
-		set_page_extent_mapped(pages[i]);
+	for (i = 0; i < num_pages; i++)
 		WARN_ON(!PageLocked(pages[i]));
-	}
 
 	return ret;
 }
 
-static noinline int check_can_nocow(struct btrfs_inode *inode, loff_t pos,
-				    size_t *write_bytes)
+/*
+ * Check if we can do nocow write into the range [@pos, @pos + @write_bytes)
+ *
+ * @pos:         File offset.
+ * @write_bytes: The length to write, will be updated to the nocow writeable
+ *               range.
+ *
+ * This function will flush ordered extents in the range to ensure proper
+ * nocow checks.
+ *
+ * Return:
+ * > 0          If we can nocow, and updates @write_bytes.
+ *  0           If we can't do a nocow write.
+ * -EAGAIN      If we can't do a nocow write because snapshoting of the inode's
+ *              root is in progress.
+ * < 0          If an error happened.
+ *
+ * NOTE: Callers need to call btrfs_check_nocow_unlock() if we return > 0.
+ */
+int btrfs_check_nocow_lock(struct btrfs_inode *inode, loff_t pos,
+			   size_t *write_bytes, bool nowait)
 {
 	struct btrfs_fs_info *fs_info = inode->root->fs_info;
 	struct btrfs_root *root = inode->root;
@@ -1552,41 +1376,117 @@ static noinline int check_can_nocow(struct btrfs_inode *inode, loff_t pos,
 	u64 num_bytes;
 	int ret;
 
-	ret = btrfs_start_write_no_snapshotting(root);
-	if (!ret)
+	if (!(inode->flags & (BTRFS_INODE_NODATACOW | BTRFS_INODE_PREALLOC)))
+		return 0;
+
+	if (!btrfs_drew_try_write_lock(&root->snapshot_lock))
 		return -EAGAIN;
 
 	lockstart = round_down(pos, fs_info->sectorsize);
 	lockend = round_up(pos + *write_bytes,
 			   fs_info->sectorsize) - 1;
-
-	btrfs_lock_and_flush_ordered_range(&inode->io_tree, inode, lockstart,
-					   lockend, NULL);
-
 	num_bytes = lockend - lockstart + 1;
-	ret = can_nocow_extent(&inode->vfs_inode, lockstart, &num_bytes,
-			NULL, NULL, NULL);
-	if (ret <= 0) {
-		ret = 0;
-		btrfs_end_write_no_snapshotting(root);
+
+	if (nowait) {
+		if (!btrfs_try_lock_ordered_range(inode, lockstart, lockend)) {
+			btrfs_drew_write_unlock(&root->snapshot_lock);
+			return -EAGAIN;
+		}
 	} else {
+		btrfs_lock_and_flush_ordered_range(inode, lockstart, lockend, NULL);
+	}
+	ret = can_nocow_extent(&inode->vfs_inode, lockstart, &num_bytes,
+			NULL, NULL, NULL, nowait, false);
+	if (ret <= 0)
+		btrfs_drew_write_unlock(&root->snapshot_lock);
+	else
 		*write_bytes = min_t(size_t, *write_bytes ,
 				     num_bytes - pos + lockstart);
-	}
-
-	unlock_extent(&inode->io_tree, lockstart, lockend);
+	unlock_extent(&inode->io_tree, lockstart, lockend, NULL);
 
 	return ret;
 }
 
+void btrfs_check_nocow_unlock(struct btrfs_inode *inode)
+{
+	btrfs_drew_write_unlock(&inode->root->snapshot_lock);
+}
+
+static void update_time_for_write(struct inode *inode)
+{
+	struct timespec64 now;
+
+	if (IS_NOCMTIME(inode))
+		return;
+
+	now = current_time(inode);
+	if (!timespec64_equal(&inode->i_mtime, &now))
+		inode->i_mtime = now;
+
+	if (!timespec64_equal(&inode->i_ctime, &now))
+		inode->i_ctime = now;
+
+	if (IS_I_VERSION(inode))
+		inode_inc_iversion(inode);
+}
+
+static int btrfs_write_check(struct kiocb *iocb, struct iov_iter *from,
+			     size_t count)
+{
+	struct file *file = iocb->ki_filp;
+	struct inode *inode = file_inode(file);
+	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+	loff_t pos = iocb->ki_pos;
+	int ret;
+	loff_t oldsize;
+	loff_t start_pos;
+
+	/*
+	 * Quickly bail out on NOWAIT writes if we don't have the nodatacow or
+	 * prealloc flags, as without those flags we always have to COW. We will
+	 * later check if we can really COW into the target range (using
+	 * can_nocow_extent() at btrfs_get_blocks_direct_write()).
+	 */
+	if ((iocb->ki_flags & IOCB_NOWAIT) &&
+	    !(BTRFS_I(inode)->flags & (BTRFS_INODE_NODATACOW | BTRFS_INODE_PREALLOC)))
+		return -EAGAIN;
+
+	current->backing_dev_info = inode_to_bdi(inode);
+	ret = file_remove_privs(file);
+	if (ret)
+		return ret;
+
+	/*
+	 * We reserve space for updating the inode when we reserve space for the
+	 * extent we are going to write, so we will enospc out there.  We don't
+	 * need to start yet another transaction to update the inode as we will
+	 * update the inode when we finish writing whatever data we write.
+	 */
+	update_time_for_write(inode);
+
+	start_pos = round_down(pos, fs_info->sectorsize);
+	oldsize = i_size_read(inode);
+	if (start_pos > oldsize) {
+		/* Expand hole size to cover write data, preventing empty gap */
+		loff_t end_pos = round_up(pos + count, fs_info->sectorsize);
+
+		ret = btrfs_cont_expand(BTRFS_I(inode), oldsize, end_pos);
+		if (ret) {
+			current->backing_dev_info = NULL;
+			return ret;
+		}
+	}
+
+	return 0;
+}
+
 static noinline ssize_t btrfs_buffered_write(struct kiocb *iocb,
 					       struct iov_iter *i)
 {
 	struct file *file = iocb->ki_filp;
-	loff_t pos = iocb->ki_pos;
+	loff_t pos;
 	struct inode *inode = file_inode(file);
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	struct btrfs_root *root = BTRFS_I(inode)->root;
 	struct page **pages = NULL;
 	struct extent_changeset *data_reserved = NULL;
 	u64 release_bytes = 0;
@@ -1594,17 +1494,39 @@ static noinline ssize_t btrfs_buffered_write(struct kiocb *iocb,
 	u64 lockend;
 	size_t num_written = 0;
 	int nrptrs;
-	int ret = 0;
+	ssize_t ret;
 	bool only_release_metadata = false;
 	bool force_page_uptodate = false;
+	loff_t old_isize = i_size_read(inode);
+	unsigned int ilock_flags = 0;
+	const bool nowait = (iocb->ki_flags & IOCB_NOWAIT);
+	unsigned int bdp_flags = (nowait ? BDP_ASYNC : 0);
+
+	if (nowait)
+		ilock_flags |= BTRFS_ILOCK_TRY;
+
+	ret = btrfs_inode_lock(inode, ilock_flags);
+	if (ret < 0)
+		return ret;
+
+	ret = generic_write_checks(iocb, i);
+	if (ret <= 0)
+		goto out;
 
+	ret = btrfs_write_check(iocb, i, ret);
+	if (ret < 0)
+		goto out;
+
+	pos = iocb->ki_pos;
 	nrptrs = min(DIV_ROUND_UP(iov_iter_count(i), PAGE_SIZE),
 			PAGE_SIZE / (sizeof(struct page *)));
 	nrptrs = min(nrptrs, current->nr_dirtied_pause - current->nr_dirtied);
 	nrptrs = max(nrptrs, 8);
 	pages = kmalloc_array(nrptrs, sizeof(struct page *), GFP_KERNEL);
-	if (!pages)
-		return -ENOMEM;
+	if (!pages) {
+		ret = -ENOMEM;
+		goto out;
+	}
 
 	while (iov_iter_count(i) > 0) {
 		struct extent_state *cached_state = NULL;
@@ -1613,8 +1535,7 @@ static noinline ssize_t btrfs_buffered_write(struct kiocb *iocb,
 		size_t write_bytes = min(iov_iter_count(i),
 					 nrptrs * (size_t)PAGE_SIZE -
 					 offset);
-		size_t num_pages = DIV_ROUND_UP(write_bytes + offset,
-						PAGE_SIZE);
+		size_t num_pages;
 		size_t reserve_bytes;
 		size_t dirty_pages;
 		size_t copied;
@@ -1622,72 +1543,82 @@ static noinline ssize_t btrfs_buffered_write(struct kiocb *iocb,
 		size_t num_sectors;
 		int extents_locked;
 
-		WARN_ON(num_pages > nrptrs);
-
 		/*
 		 * Fault pages before locking them in prepare_pages
 		 * to avoid recursive lock
 		 */
-		if (unlikely(iov_iter_fault_in_readable(i, write_bytes))) {
+		if (unlikely(fault_in_iov_iter_readable(i, write_bytes))) {
 			ret = -EFAULT;
 			break;
 		}
 
 		only_release_metadata = false;
 		sector_offset = pos & (fs_info->sectorsize - 1);
-		reserve_bytes = round_up(write_bytes + sector_offset,
-				fs_info->sectorsize);
 
 		extent_changeset_release(data_reserved);
-		ret = btrfs_check_data_free_space(inode, &data_reserved, pos,
-						  write_bytes);
+		ret = btrfs_check_data_free_space(BTRFS_I(inode),
+						  &data_reserved, pos,
+						  write_bytes, nowait);
 		if (ret < 0) {
-			if ((BTRFS_I(inode)->flags & (BTRFS_INODE_NODATACOW |
-						      BTRFS_INODE_PREALLOC)) &&
-			    check_can_nocow(BTRFS_I(inode), pos,
-					&write_bytes) > 0) {
-				/*
-				 * For nodata cow case, no need to reserve
-				 * data space.
-				 */
-				only_release_metadata = true;
-				/*
-				 * our prealloc extent may be smaller than
-				 * write_bytes, so scale down.
-				 */
-				num_pages = DIV_ROUND_UP(write_bytes + offset,
-							 PAGE_SIZE);
-				reserve_bytes = round_up(write_bytes +
-							 sector_offset,
-							 fs_info->sectorsize);
-			} else {
+			int can_nocow;
+
+			if (nowait && (ret == -ENOSPC || ret == -EAGAIN)) {
+				ret = -EAGAIN;
 				break;
 			}
+
+			/*
+			 * If we don't have to COW at the offset, reserve
+			 * metadata only. write_bytes may get smaller than
+			 * requested here.
+			 */
+			can_nocow = btrfs_check_nocow_lock(BTRFS_I(inode), pos,
+							   &write_bytes, nowait);
+			if (can_nocow < 0)
+				ret = can_nocow;
+			if (can_nocow > 0)
+				ret = 0;
+			if (ret)
+				break;
+			only_release_metadata = true;
 		}
 
+		num_pages = DIV_ROUND_UP(write_bytes + offset, PAGE_SIZE);
+		WARN_ON(num_pages > nrptrs);
+		reserve_bytes = round_up(write_bytes + sector_offset,
+					 fs_info->sectorsize);
 		WARN_ON(reserve_bytes == 0);
 		ret = btrfs_delalloc_reserve_metadata(BTRFS_I(inode),
-				reserve_bytes);
+						      reserve_bytes,
+						      reserve_bytes, nowait);
 		if (ret) {
 			if (!only_release_metadata)
-				btrfs_free_reserved_data_space(inode,
+				btrfs_free_reserved_data_space(BTRFS_I(inode),
 						data_reserved, pos,
 						write_bytes);
 			else
-				btrfs_end_write_no_snapshotting(root);
+				btrfs_check_nocow_unlock(BTRFS_I(inode));
+
+			if (nowait && ret == -ENOSPC)
+				ret = -EAGAIN;
 			break;
 		}
 
 		release_bytes = reserve_bytes;
 again:
+		ret = balance_dirty_pages_ratelimited_flags(inode->i_mapping, bdp_flags);
+		if (ret) {
+			btrfs_delalloc_release_extents(BTRFS_I(inode), reserve_bytes);
+			break;
+		}
+
 		/*
 		 * This is going to setup the pages array with the number of
 		 * pages we want, so we don't really need to worry about the
 		 * contents of pages from loop to loop
 		 */
 		ret = prepare_pages(inode, pages, num_pages,
-				    pos, write_bytes,
-				    force_page_uptodate);
+				    pos, write_bytes, force_page_uptodate, false);
 		if (ret) {
 			btrfs_delalloc_release_extents(BTRFS_I(inode),
 						       reserve_bytes);
@@ -1697,10 +1628,11 @@ again:
 		extents_locked = lock_and_cleanup_extent_if_need(
 				BTRFS_I(inode), pages,
 				num_pages, pos, write_bytes, &lockstart,
-				&lockend, &cached_state);
+				&lockend, nowait, &cached_state);
 		if (extents_locked < 0) {
-			if (extents_locked == -EAGAIN)
+			if (!nowait && extents_locked == -EAGAIN)
 				goto again;
+
 			btrfs_delalloc_release_extents(BTRFS_I(inode),
 						       reserve_bytes);
 			ret = extents_locked;
@@ -1733,8 +1665,7 @@ again:
 
 		if (num_sectors > dirty_sectors) {
 			/* release everything except the sectors we dirtied */
-			release_bytes -= dirty_sectors <<
-						fs_info->sb->s_blocksize_bits;
+			release_bytes -= dirty_sectors << fs_info->sectorsize_bits;
 			if (only_release_metadata) {
 				btrfs_delalloc_release_metadata(BTRFS_I(inode),
 							release_bytes, true);
@@ -1744,7 +1675,7 @@ again:
 				__pos = round_down(pos,
 						   fs_info->sectorsize) +
 					(dirty_pages << PAGE_SHIFT);
-				btrfs_delalloc_release_space(inode,
+				btrfs_delalloc_release_space(BTRFS_I(inode),
 						data_reserved, __pos,
 						release_bytes, true);
 			}
@@ -1753,9 +1684,9 @@ again:
 		release_bytes = round_up(copied + sector_offset,
 					fs_info->sectorsize);
 
-		if (copied > 0)
-			ret = btrfs_dirty_pages(inode, pages, dirty_pages,
-						pos, copied, &cached_state);
+		ret = btrfs_dirty_pages(BTRFS_I(inode), pages,
+					dirty_pages, pos, copied,
+					&cached_state, only_release_metadata);
 
 		/*
 		 * If we have not locked the extent range, because the range's
@@ -1765,40 +1696,25 @@ again:
 		 * possible cached extent state to avoid a memory leak.
 		 */
 		if (extents_locked)
-			unlock_extent_cached(&BTRFS_I(inode)->io_tree,
-					     lockstart, lockend, &cached_state);
+			unlock_extent(&BTRFS_I(inode)->io_tree, lockstart,
+				      lockend, &cached_state);
 		else
 			free_extent_state(cached_state);
 
 		btrfs_delalloc_release_extents(BTRFS_I(inode), reserve_bytes);
 		if (ret) {
-			btrfs_drop_pages(pages, num_pages);
+			btrfs_drop_pages(fs_info, pages, num_pages, pos, copied);
 			break;
 		}
 
 		release_bytes = 0;
 		if (only_release_metadata)
-			btrfs_end_write_no_snapshotting(root);
-
-		if (only_release_metadata && copied > 0) {
-			lockstart = round_down(pos,
-					       fs_info->sectorsize);
-			lockend = round_up(pos + copied,
-					   fs_info->sectorsize) - 1;
-
-			set_extent_bit(&BTRFS_I(inode)->io_tree, lockstart,
-				       lockend, EXTENT_NORESERVE, NULL,
-				       NULL, GFP_NOFS);
-		}
+			btrfs_check_nocow_unlock(BTRFS_I(inode));
 
-		btrfs_drop_pages(pages, num_pages);
+		btrfs_drop_pages(fs_info, pages, num_pages, pos, copied);
 
 		cond_resched();
 
-		balance_dirty_pages_ratelimited(inode->i_mapping);
-		if (dirty_pages < (fs_info->nodesize >> PAGE_SHIFT) + 1)
-			btrfs_btree_balance_dirty(fs_info);
-
 		pos += copied;
 		num_written += copied;
 	}
@@ -1807,34 +1723,177 @@ again:
 
 	if (release_bytes) {
 		if (only_release_metadata) {
-			btrfs_end_write_no_snapshotting(root);
+			btrfs_check_nocow_unlock(BTRFS_I(inode));
 			btrfs_delalloc_release_metadata(BTRFS_I(inode),
 					release_bytes, true);
 		} else {
-			btrfs_delalloc_release_space(inode, data_reserved,
+			btrfs_delalloc_release_space(BTRFS_I(inode),
+					data_reserved,
 					round_down(pos, fs_info->sectorsize),
 					release_bytes, true);
 		}
 	}
 
 	extent_changeset_free(data_reserved);
+	if (num_written > 0) {
+		pagecache_isize_extended(inode, old_isize, iocb->ki_pos);
+		iocb->ki_pos += num_written;
+	}
+out:
+	btrfs_inode_unlock(inode, ilock_flags);
 	return num_written ? num_written : ret;
 }
 
-static ssize_t __btrfs_direct_write(struct kiocb *iocb, struct iov_iter *from)
+static ssize_t check_direct_IO(struct btrfs_fs_info *fs_info,
+			       const struct iov_iter *iter, loff_t offset)
+{
+	const u32 blocksize_mask = fs_info->sectorsize - 1;
+
+	if (offset & blocksize_mask)
+		return -EINVAL;
+
+	if (iov_iter_alignment(iter) & blocksize_mask)
+		return -EINVAL;
+
+	return 0;
+}
+
+static ssize_t btrfs_direct_write(struct kiocb *iocb, struct iov_iter *from)
 {
 	struct file *file = iocb->ki_filp;
 	struct inode *inode = file_inode(file);
+	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	loff_t pos;
-	ssize_t written;
+	ssize_t written = 0;
 	ssize_t written_buffered;
+	size_t prev_left = 0;
 	loff_t endbyte;
-	int err;
+	ssize_t err;
+	unsigned int ilock_flags = 0;
+	struct iomap_dio *dio;
+
+	if (iocb->ki_flags & IOCB_NOWAIT)
+		ilock_flags |= BTRFS_ILOCK_TRY;
 
-	written = generic_file_direct_write(iocb, from);
+	/* If the write DIO is within EOF, use a shared lock */
+	if (iocb->ki_pos + iov_iter_count(from) <= i_size_read(inode))
+		ilock_flags |= BTRFS_ILOCK_SHARED;
 
-	if (written < 0 || !iov_iter_count(from))
-		return written;
+relock:
+	err = btrfs_inode_lock(inode, ilock_flags);
+	if (err < 0)
+		return err;
+
+	err = generic_write_checks(iocb, from);
+	if (err <= 0) {
+		btrfs_inode_unlock(inode, ilock_flags);
+		return err;
+	}
+
+	err = btrfs_write_check(iocb, from, err);
+	if (err < 0) {
+		btrfs_inode_unlock(inode, ilock_flags);
+		goto out;
+	}
+
+	pos = iocb->ki_pos;
+	/*
+	 * Re-check since file size may have changed just before taking the
+	 * lock or pos may have changed because of O_APPEND in generic_write_check()
+	 */
+	if ((ilock_flags & BTRFS_ILOCK_SHARED) &&
+	    pos + iov_iter_count(from) > i_size_read(inode)) {
+		btrfs_inode_unlock(inode, ilock_flags);
+		ilock_flags &= ~BTRFS_ILOCK_SHARED;
+		goto relock;
+	}
+
+	if (check_direct_IO(fs_info, from, pos)) {
+		btrfs_inode_unlock(inode, ilock_flags);
+		goto buffered;
+	}
+
+	/*
+	 * The iov_iter can be mapped to the same file range we are writing to.
+	 * If that's the case, then we will deadlock in the iomap code, because
+	 * it first calls our callback btrfs_dio_iomap_begin(), which will create
+	 * an ordered extent, and after that it will fault in the pages that the
+	 * iov_iter refers to. During the fault in we end up in the readahead
+	 * pages code (starting at btrfs_readahead()), which will lock the range,
+	 * find that ordered extent and then wait for it to complete (at
+	 * btrfs_lock_and_flush_ordered_range()), resulting in a deadlock since
+	 * obviously the ordered extent can never complete as we didn't submit
+	 * yet the respective bio(s). This always happens when the buffer is
+	 * memory mapped to the same file range, since the iomap DIO code always
+	 * invalidates pages in the target file range (after starting and waiting
+	 * for any writeback).
+	 *
+	 * So here we disable page faults in the iov_iter and then retry if we
+	 * got -EFAULT, faulting in the pages before the retry.
+	 */
+	from->nofault = true;
+	dio = btrfs_dio_write(iocb, from, written);
+	from->nofault = false;
+
+	/*
+	 * iomap_dio_complete() will call btrfs_sync_file() if we have a dsync
+	 * iocb, and that needs to lock the inode. So unlock it before calling
+	 * iomap_dio_complete() to avoid a deadlock.
+	 */
+	btrfs_inode_unlock(inode, ilock_flags);
+
+	if (IS_ERR_OR_NULL(dio))
+		err = PTR_ERR_OR_ZERO(dio);
+	else
+		err = iomap_dio_complete(dio);
+
+	/* No increment (+=) because iomap returns a cumulative value. */
+	if (err > 0)
+		written = err;
+
+	if (iov_iter_count(from) > 0 && (err == -EFAULT || err > 0)) {
+		const size_t left = iov_iter_count(from);
+		/*
+		 * We have more data left to write. Try to fault in as many as
+		 * possible of the remainder pages and retry. We do this without
+		 * releasing and locking again the inode, to prevent races with
+		 * truncate.
+		 *
+		 * Also, in case the iov refers to pages in the file range of the
+		 * file we want to write to (due to a mmap), we could enter an
+		 * infinite loop if we retry after faulting the pages in, since
+		 * iomap will invalidate any pages in the range early on, before
+		 * it tries to fault in the pages of the iov. So we keep track of
+		 * how much was left of iov in the previous EFAULT and fallback
+		 * to buffered IO in case we haven't made any progress.
+		 */
+		if (left == prev_left) {
+			err = -ENOTBLK;
+		} else {
+			fault_in_iov_iter_readable(from, left);
+			prev_left = left;
+			goto relock;
+		}
+	}
+
+	/*
+	 * If 'err' is -ENOTBLK or we have not written all data, then it means
+	 * we must fallback to buffered IO.
+	 */
+	if ((err < 0 && err != -ENOTBLK) || !iov_iter_count(from))
+		goto out;
+
+buffered:
+	/*
+	 * If we are in a NOWAIT context, then return -EAGAIN to signal the caller
+	 * it must retry the operation in a context where blocking is acceptable,
+	 * since we currently don't have NOWAIT semantics support for buffered IO
+	 * and may block there for many reasons (reserving space for example).
+	 */
+	if (iocb->ki_flags & IOCB_NOWAIT) {
+		err = -EAGAIN;
+		goto out;
+	}
 
 	pos = iocb->ki_pos;
 	written_buffered = btrfs_buffered_write(iocb, from);
@@ -1858,150 +1917,91 @@ static ssize_t __btrfs_direct_write(struct kiocb *iocb, struct iov_iter *from)
 	invalidate_mapping_pages(file->f_mapping, pos >> PAGE_SHIFT,
 				 endbyte >> PAGE_SHIFT);
 out:
-	return written ? written : err;
-}
-
-static void update_time_for_write(struct inode *inode)
-{
-	struct timespec64 now;
-
-	if (IS_NOCMTIME(inode))
-		return;
-
-	now = current_time(inode);
-	if (!timespec64_equal(&inode->i_mtime, &now))
-		inode->i_mtime = now;
-
-	if (!timespec64_equal(&inode->i_ctime, &now))
-		inode->i_ctime = now;
-
-	if (IS_I_VERSION(inode))
-		inode_inc_iversion(inode);
+	return err < 0 ? err : written;
 }
 
-static ssize_t btrfs_file_write_iter(struct kiocb *iocb,
-				    struct iov_iter *from)
+static ssize_t btrfs_encoded_write(struct kiocb *iocb, struct iov_iter *from,
+			const struct btrfs_ioctl_encoded_io_args *encoded)
 {
 	struct file *file = iocb->ki_filp;
 	struct inode *inode = file_inode(file);
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	struct btrfs_root *root = BTRFS_I(inode)->root;
-	u64 start_pos;
-	u64 end_pos;
-	ssize_t num_written = 0;
-	const bool sync = iocb->ki_flags & IOCB_DSYNC;
-	ssize_t err;
-	loff_t pos;
-	size_t count;
-	loff_t oldsize;
-	int clean_page = 0;
+	loff_t count;
+	ssize_t ret;
 
-	if (!(iocb->ki_flags & IOCB_DIRECT) &&
-	    (iocb->ki_flags & IOCB_NOWAIT))
-		return -EOPNOTSUPP;
-
-	if (iocb->ki_flags & IOCB_NOWAIT) {
-		if (!inode_trylock(inode))
-			return -EAGAIN;
-	} else {
-		inode_lock(inode);
-	}
-
-	err = generic_write_checks(iocb, from);
-	if (err <= 0) {
-		inode_unlock(inode);
-		return err;
-	}
-
-	pos = iocb->ki_pos;
-	count = iov_iter_count(from);
-	if (iocb->ki_flags & IOCB_NOWAIT) {
+	btrfs_inode_lock(inode, 0);
+	count = encoded->len;
+	ret = generic_write_checks_count(iocb, &count);
+	if (ret == 0 && count != encoded->len) {
 		/*
-		 * We will allocate space in case nodatacow is not set,
-		 * so bail
+		 * The write got truncated by generic_write_checks_count(). We
+		 * can't do a partial encoded write.
 		 */
-		if (!(BTRFS_I(inode)->flags & (BTRFS_INODE_NODATACOW |
-					      BTRFS_INODE_PREALLOC)) ||
-		    check_can_nocow(BTRFS_I(inode), pos, &count) <= 0) {
-			inode_unlock(inode);
-			return -EAGAIN;
-		}
+		ret = -EFBIG;
 	}
-
-	current->backing_dev_info = inode_to_bdi(inode);
-	err = file_remove_privs(file);
-	if (err) {
-		inode_unlock(inode);
+	if (ret || encoded->len == 0)
 		goto out;
-	}
 
-	/*
-	 * If BTRFS flips readonly due to some impossible error
-	 * (fs_info->fs_state now has BTRFS_SUPER_FLAG_ERROR),
-	 * although we have opened a file as writable, we have
-	 * to stop this write operation to ensure FS consistency.
-	 */
-	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
-		inode_unlock(inode);
-		err = -EROFS;
+	ret = btrfs_write_check(iocb, from, encoded->len);
+	if (ret < 0)
 		goto out;
-	}
+
+	ret = btrfs_do_encoded_write(iocb, from, encoded);
+out:
+	btrfs_inode_unlock(inode, 0);
+	return ret;
+}
+
+ssize_t btrfs_do_write_iter(struct kiocb *iocb, struct iov_iter *from,
+			    const struct btrfs_ioctl_encoded_io_args *encoded)
+{
+	struct file *file = iocb->ki_filp;
+	struct btrfs_inode *inode = BTRFS_I(file_inode(file));
+	ssize_t num_written, num_sync;
+	const bool sync = iocb_is_dsync(iocb);
 
 	/*
-	 * We reserve space for updating the inode when we reserve space for the
-	 * extent we are going to write, so we will enospc out there.  We don't
-	 * need to start yet another transaction to update the inode as we will
-	 * update the inode when we finish writing whatever data we write.
+	 * If the fs flips readonly due to some impossible error, although we
+	 * have opened a file as writable, we have to stop this write operation
+	 * to ensure consistency.
 	 */
-	update_time_for_write(inode);
+	if (BTRFS_FS_ERROR(inode->root->fs_info))
+		return -EROFS;
 
-	start_pos = round_down(pos, fs_info->sectorsize);
-	oldsize = i_size_read(inode);
-	if (start_pos > oldsize) {
-		/* Expand hole size to cover write data, preventing empty gap */
-		end_pos = round_up(pos + count,
-				   fs_info->sectorsize);
-		err = btrfs_cont_expand(inode, oldsize, end_pos);
-		if (err) {
-			inode_unlock(inode);
-			goto out;
-		}
-		if (start_pos > round_up(oldsize, fs_info->sectorsize))
-			clean_page = 1;
-	}
+	if (encoded && (iocb->ki_flags & IOCB_NOWAIT))
+		return -EOPNOTSUPP;
 
 	if (sync)
-		atomic_inc(&BTRFS_I(inode)->sync_writers);
-
-	if (iocb->ki_flags & IOCB_DIRECT) {
-		num_written = __btrfs_direct_write(iocb, from);
+		atomic_inc(&inode->sync_writers);
+
+	if (encoded) {
+		num_written = btrfs_encoded_write(iocb, from, encoded);
+		num_sync = encoded->len;
+	} else if (iocb->ki_flags & IOCB_DIRECT) {
+		num_written = btrfs_direct_write(iocb, from);
+		num_sync = num_written;
 	} else {
 		num_written = btrfs_buffered_write(iocb, from);
-		if (num_written > 0)
-			iocb->ki_pos = pos + num_written;
-		if (clean_page)
-			pagecache_isize_extended(inode, oldsize,
-						i_size_read(inode));
+		num_sync = num_written;
 	}
 
-	inode_unlock(inode);
+	btrfs_set_inode_last_sub_trans(inode);
 
-	/*
-	 * We also have to set last_sub_trans to the current log transid,
-	 * otherwise subsequent syncs to a file that's been synced in this
-	 * transaction will appear to have already occurred.
-	 */
-	spin_lock(&BTRFS_I(inode)->lock);
-	BTRFS_I(inode)->last_sub_trans = root->log_transid;
-	spin_unlock(&BTRFS_I(inode)->lock);
-	if (num_written > 0)
-		num_written = generic_write_sync(iocb, num_written);
+	if (num_sync > 0) {
+		num_sync = generic_write_sync(iocb, num_sync);
+		if (num_sync < 0)
+			num_written = num_sync;
+	}
 
 	if (sync)
-		atomic_dec(&BTRFS_I(inode)->sync_writers);
-out:
+		atomic_dec(&inode->sync_writers);
+
 	current->backing_dev_info = NULL;
-	return num_written ? num_written : err;
+	return num_written;
+}
+
+static ssize_t btrfs_file_write_iter(struct kiocb *iocb, struct iov_iter *from)
+{
+	return btrfs_do_write_iter(iocb, from, NULL);
 }
 
 int btrfs_release_file(struct inode *inode, struct file *filp)
@@ -2014,12 +2014,12 @@ int btrfs_release_file(struct inode *inode, struct file *filp)
 	filp->private_data = NULL;
 
 	/*
-	 * ordered_data_close is set by setattr when we are about to truncate
-	 * a file from a non-zero size to a zero size.  This tries to
-	 * flush down new bytes that may have been written if the
-	 * application were using truncate to replace a file in place.
+	 * Set by setattr when we are about to truncate a file from a non-zero
+	 * size to a zero size.  This tries to flush down new bytes that may
+	 * have been written if the application were using truncate to replace
+	 * a file in place.
 	 */
-	if (test_and_clear_bit(BTRFS_INODE_ORDERED_DATA_CLOSE,
+	if (test_and_clear_bit(BTRFS_INODE_FLUSH_ON_CLOSE,
 			       &BTRFS_I(inode)->runtime_flags))
 			filemap_flush(inode->i_mapping);
 	return 0;
@@ -2045,6 +2045,30 @@ static int start_ordered_ops(struct inode *inode, loff_t start, loff_t end)
 	return ret;
 }
 
+static inline bool skip_inode_logging(const struct btrfs_log_ctx *ctx)
+{
+	struct btrfs_inode *inode = BTRFS_I(ctx->inode);
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
+
+	if (btrfs_inode_in_log(inode, fs_info->generation) &&
+	    list_empty(&ctx->ordered_extents))
+		return true;
+
+	/*
+	 * If we are doing a fast fsync we can not bail out if the inode's
+	 * last_trans is <= then the last committed transaction, because we only
+	 * update the last_trans of the inode during ordered extent completion,
+	 * and for a fast fsync we don't wait for that, we only wait for the
+	 * writeback to complete.
+	 */
+	if (inode->last_trans <= fs_info->last_trans_committed &&
+	    (test_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &inode->runtime_flags) ||
+	     list_empty(&ctx->ordered_extents)))
+		return true;
+
+	return false;
+}
+
 /*
  * fsync call for both files and directories.  This logs the inode into
  * the tree log instead of forcing full commits whenever possible.
@@ -2065,12 +2089,26 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
 	struct btrfs_trans_handle *trans;
 	struct btrfs_log_ctx ctx;
 	int ret = 0, err;
+	u64 len;
+	bool full_sync;
 
 	trace_btrfs_sync_file(file, datasync);
 
 	btrfs_init_log_ctx(&ctx, inode);
 
 	/*
+	 * Always set the range to a full range, otherwise we can get into
+	 * several problems, from missing file extent items to represent holes
+	 * when not using the NO_HOLES feature, to log tree corruption due to
+	 * races between hole detection during logging and completion of ordered
+	 * extents outside the range, to missing checksums due to ordered extents
+	 * for which we flushed only a subset of their pages.
+	 */
+	start = 0;
+	end = LLONG_MAX;
+	len = (u64)LLONG_MAX + 1;
+
+	/*
 	 * We write the dirty pages in the range and wait until they complete
 	 * out of the ->i_mutex. If so, we can flush the dirty pages by
 	 * multi-task, and make the performance up.  See
@@ -2080,36 +2118,16 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
 	if (ret)
 		goto out;
 
-	inode_lock(inode);
-
-	/*
-	 * We take the dio_sem here because the tree log stuff can race with
-	 * lockless dio writes and get an extent map logged for an extent we
-	 * never waited on.  We need it this high up for lockdep reasons.
-	 */
-	down_write(&BTRFS_I(inode)->dio_sem);
+	btrfs_inode_lock(inode, BTRFS_ILOCK_MMAP);
 
 	atomic_inc(&root->log_batch);
 
 	/*
-	 * If the inode needs a full sync, make sure we use a full range to
-	 * avoid log tree corruption, due to hole detection racing with ordered
-	 * extent completion for adjacent ranges, and assertion failures during
-	 * hole detection. Do this while holding the inode lock, to avoid races
-	 * with other tasks.
-	 */
-	if (test_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
-		     &BTRFS_I(inode)->runtime_flags)) {
-		start = 0;
-		end = LLONG_MAX;
-	}
-
-	/*
-	 * Before we acquired the inode's lock, someone may have dirtied more
-	 * pages in the target range. We need to make sure that writeback for
-	 * any such pages does not start while we are logging the inode, because
-	 * if it does, any of the following might happen when we are not doing a
-	 * full inode sync:
+	 * Before we acquired the inode's lock and the mmap lock, someone may
+	 * have dirtied more pages in the target range. We need to make sure
+	 * that writeback for any such pages does not start while we are logging
+	 * the inode, because if it does, any of the following might happen when
+	 * we are not doing a full inode sync:
 	 *
 	 * 1) We log an extent after its writeback finishes but before its
 	 *    checksums are added to the csum tree, leading to -EIO errors
@@ -2124,28 +2142,56 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
 	 */
 	ret = start_ordered_ops(inode, start, end);
 	if (ret) {
-		inode_unlock(inode);
+		btrfs_inode_unlock(inode, BTRFS_ILOCK_MMAP);
 		goto out;
 	}
 
 	/*
+	 * Always check for the full sync flag while holding the inode's lock,
+	 * to avoid races with other tasks. The flag must be either set all the
+	 * time during logging or always off all the time while logging.
+	 * We check the flag here after starting delalloc above, because when
+	 * running delalloc the full sync flag may be set if we need to drop
+	 * extra extent map ranges due to temporary memory allocation failures.
+	 */
+	full_sync = test_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
+			     &BTRFS_I(inode)->runtime_flags);
+
+	/*
 	 * We have to do this here to avoid the priority inversion of waiting on
 	 * IO of a lower priority task while holding a transaction open.
 	 *
-	 * Also, the range length can be represented by u64, we have to do the
-	 * typecasts to avoid signed overflow if it's [0, LLONG_MAX].
+	 * For a full fsync we wait for the ordered extents to complete while
+	 * for a fast fsync we wait just for writeback to complete, and then
+	 * attach the ordered extents to the transaction so that a transaction
+	 * commit waits for their completion, to avoid data loss if we fsync,
+	 * the current transaction commits before the ordered extents complete
+	 * and a power failure happens right after that.
+	 *
+	 * For zoned filesystem, if a write IO uses a ZONE_APPEND command, the
+	 * logical address recorded in the ordered extent may change. We need
+	 * to wait for the IO to stabilize the logical address.
 	 */
-	ret = btrfs_wait_ordered_range(inode, start, (u64)end - (u64)start + 1);
-	if (ret) {
-		up_write(&BTRFS_I(inode)->dio_sem);
-		inode_unlock(inode);
-		goto out;
+	if (full_sync || btrfs_is_zoned(fs_info)) {
+		ret = btrfs_wait_ordered_range(inode, start, len);
+	} else {
+		/*
+		 * Get our ordered extents as soon as possible to avoid doing
+		 * checksum lookups in the csum tree, and use instead the
+		 * checksums attached to the ordered extents.
+		 */
+		btrfs_get_ordered_extents_for_logging(BTRFS_I(inode),
+						      &ctx.ordered_extents);
+		ret = filemap_fdatawait_range(inode->i_mapping, start, end);
 	}
+
+	if (ret)
+		goto out_release_extents;
+
 	atomic_inc(&root->log_batch);
 
 	smp_mb();
-	if (btrfs_inode_in_log(BTRFS_I(inode), fs_info->generation) ||
-	    BTRFS_I(inode)->last_trans <= fs_info->last_trans_committed) {
+	if (skip_inode_logging(&ctx)) {
 		/*
 		 * We've had everything committed since the last time we were
 		 * modified so clear this flag in case it was set for whatever
@@ -2161,9 +2207,7 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
 		 * checked called fsync.
 		 */
 		ret = filemap_check_wb_err(inode->i_mapping, file->f_wb_err);
-		up_write(&BTRFS_I(inode)->dio_sem);
-		inode_unlock(inode);
-		goto out;
+		goto out_release_extents;
 	}
 
 	/*
@@ -2180,15 +2224,15 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
 	trans = btrfs_start_transaction(root, 0);
 	if (IS_ERR(trans)) {
 		ret = PTR_ERR(trans);
-		up_write(&BTRFS_I(inode)->dio_sem);
-		inode_unlock(inode);
-		goto out;
+		goto out_release_extents;
 	}
+	trans->in_fsync = true;
 
-	ret = btrfs_log_dentry_safe(trans, dentry, start, end, &ctx);
+	ret = btrfs_log_dentry_safe(trans, dentry, &ctx);
+	btrfs_release_log_ctx_extents(&ctx);
 	if (ret < 0) {
 		/* Fallthrough and commit/free transaction. */
-		ret = 1;
+		ret = BTRFS_LOG_FORCE_COMMIT;
 	}
 
 	/* we've logged all the items and now have a consistent
@@ -2201,27 +2245,76 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
 	 * file again, but that will end up using the synchronization
 	 * inside btrfs_sync_log to keep things safe.
 	 */
-	up_write(&BTRFS_I(inode)->dio_sem);
-	inode_unlock(inode);
+	btrfs_inode_unlock(inode, BTRFS_ILOCK_MMAP);
 
-	if (ret != BTRFS_NO_LOG_SYNC) {
+	if (ret == BTRFS_NO_LOG_SYNC) {
+		ret = btrfs_end_transaction(trans);
+		goto out;
+	}
+
+	/* We successfully logged the inode, attempt to sync the log. */
+	if (!ret) {
+		ret = btrfs_sync_log(trans, root, &ctx);
 		if (!ret) {
-			ret = btrfs_sync_log(trans, root, &ctx);
-			if (!ret) {
-				ret = btrfs_end_transaction(trans);
-				goto out;
-			}
+			ret = btrfs_end_transaction(trans);
+			goto out;
 		}
-		ret = btrfs_commit_transaction(trans);
-	} else {
+	}
+
+	/*
+	 * At this point we need to commit the transaction because we had
+	 * btrfs_need_log_full_commit() or some other error.
+	 *
+	 * If we didn't do a full sync we have to stop the trans handle, wait on
+	 * the ordered extents, start it again and commit the transaction.  If
+	 * we attempt to wait on the ordered extents here we could deadlock with
+	 * something like fallocate() that is holding the extent lock trying to
+	 * start a transaction while some other thread is trying to commit the
+	 * transaction while we (fsync) are currently holding the transaction
+	 * open.
+	 */
+	if (!full_sync) {
 		ret = btrfs_end_transaction(trans);
+		if (ret)
+			goto out;
+		ret = btrfs_wait_ordered_range(inode, start, len);
+		if (ret)
+			goto out;
+
+		/*
+		 * This is safe to use here because we're only interested in
+		 * making sure the transaction that had the ordered extents is
+		 * committed.  We aren't waiting on anything past this point,
+		 * we're purely getting the transaction and committing it.
+		 */
+		trans = btrfs_attach_transaction_barrier(root);
+		if (IS_ERR(trans)) {
+			ret = PTR_ERR(trans);
+
+			/*
+			 * We committed the transaction and there's no currently
+			 * running transaction, this means everything we care
+			 * about made it to disk and we are done.
+			 */
+			if (ret == -ENOENT)
+				ret = 0;
+			goto out;
+		}
 	}
+
+	ret = btrfs_commit_transaction(trans);
 out:
 	ASSERT(list_empty(&ctx.list));
+	ASSERT(list_empty(&ctx.conflict_inodes));
 	err = file_check_and_advance_wb_err(file);
 	if (!ret)
 		ret = err;
 	return ret > 0 ? -EIO : ret;
+
+out_release_extents:
+	btrfs_release_log_ctx_extents(&ctx);
+	btrfs_inode_unlock(inode, BTRFS_ILOCK_MMAP);
+	goto out;
 }
 
 static const struct vm_operations_struct btrfs_file_vm_ops = {
@@ -2234,7 +2327,7 @@ static int btrfs_file_mmap(struct file	*filp, struct vm_area_struct *vma)
 {
 	struct address_space *mapping = filp->f_mapping;
 
-	if (!mapping->a_ops->readpage)
+	if (!mapping->a_ops->read_folio)
 		return -ENOEXEC;
 
 	file_accessed(filp);
@@ -2281,7 +2374,6 @@ static int fill_holes(struct btrfs_trans_handle *trans,
 	struct extent_buffer *leaf;
 	struct btrfs_file_extent_item *fi;
 	struct extent_map *hole_em;
-	struct extent_map_tree *em_tree = &inode->extent_tree;
 	struct btrfs_key key;
 	int ret;
 
@@ -2315,6 +2407,7 @@ static int fill_holes(struct btrfs_trans_handle *trans,
 		btrfs_set_file_extent_num_bytes(leaf, fi, num_bytes);
 		btrfs_set_file_extent_ram_bytes(leaf, fi, num_bytes);
 		btrfs_set_file_extent_offset(leaf, fi, 0);
+		btrfs_set_file_extent_generation(leaf, fi, trans->transid);
 		btrfs_mark_buffer_dirty(leaf);
 		goto out;
 	}
@@ -2331,13 +2424,14 @@ static int fill_holes(struct btrfs_trans_handle *trans,
 		btrfs_set_file_extent_num_bytes(leaf, fi, num_bytes);
 		btrfs_set_file_extent_ram_bytes(leaf, fi, num_bytes);
 		btrfs_set_file_extent_offset(leaf, fi, 0);
+		btrfs_set_file_extent_generation(leaf, fi, trans->transid);
 		btrfs_mark_buffer_dirty(leaf);
 		goto out;
 	}
 	btrfs_release_path(path);
 
-	ret = btrfs_insert_file_extent(trans, root, btrfs_ino(inode),
-			offset, 0, 0, end - offset, 0, end - offset, 0, 0, 0);
+	ret = btrfs_insert_hole_extent(trans, root, btrfs_ino(inode), offset,
+				       end - offset);
 	if (ret)
 		return ret;
 
@@ -2346,8 +2440,8 @@ out:
 
 	hole_em = alloc_extent_map();
 	if (!hole_em) {
-		btrfs_drop_extent_cache(inode, offset, end - 1, 0);
-		set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &inode->runtime_flags);
+		btrfs_drop_extent_map_range(inode, offset, end - 1, false);
+		btrfs_set_inode_full_sync(inode);
 	} else {
 		hole_em->start = offset;
 		hole_em->len = end - offset;
@@ -2360,16 +2454,10 @@ out:
 		hole_em->compress_type = BTRFS_COMPRESS_NONE;
 		hole_em->generation = trans->transid;
 
-		do {
-			btrfs_drop_extent_cache(inode, offset, end - 1, 0);
-			write_lock(&em_tree->lock);
-			ret = add_extent_mapping(em_tree, hole_em, 1);
-			write_unlock(&em_tree->lock);
-		} while (ret == -EEXIST);
+		ret = btrfs_replace_extent_map_range(inode, hole_em, true);
 		free_extent_map(hole_em);
 		if (ret)
-			set_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
-					&inode->runtime_flags);
+			btrfs_set_inode_full_sync(inode);
 	}
 
 	return 0;
@@ -2381,13 +2469,13 @@ out:
  *	   em->start + em->len > start)
  * When a hole extent is found, return 1 and modify start/len.
  */
-static int find_first_non_hole(struct inode *inode, u64 *start, u64 *len)
+static int find_first_non_hole(struct btrfs_inode *inode, u64 *start, u64 *len)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
 	struct extent_map *em;
 	int ret = 0;
 
-	em = btrfs_get_extent(BTRFS_I(inode), NULL, 0,
+	em = btrfs_get_extent(inode, NULL, 0,
 			      round_down(*start, fs_info->sectorsize),
 			      round_up(*len, fs_info->sectorsize));
 	if (IS_ERR(em))
@@ -2404,100 +2492,129 @@ static int find_first_non_hole(struct inode *inode, u64 *start, u64 *len)
 	return ret;
 }
 
-static int btrfs_punch_hole_lock_range(struct inode *inode,
-				       const u64 lockstart,
-				       const u64 lockend,
-				       struct extent_state **cached_state)
+static void btrfs_punch_hole_lock_range(struct inode *inode,
+					const u64 lockstart,
+					const u64 lockend,
+					struct extent_state **cached_state)
 {
-	while (1) {
-		struct btrfs_ordered_extent *ordered;
-		int ret;
+	/*
+	 * For subpage case, if the range is not at page boundary, we could
+	 * have pages at the leading/tailing part of the range.
+	 * This could lead to dead loop since filemap_range_has_page()
+	 * will always return true.
+	 * So here we need to do extra page alignment for
+	 * filemap_range_has_page().
+	 */
+	const u64 page_lockstart = round_up(lockstart, PAGE_SIZE);
+	const u64 page_lockend = round_down(lockend + 1, PAGE_SIZE) - 1;
 
+	while (1) {
 		truncate_pagecache_range(inode, lockstart, lockend);
 
-		lock_extent_bits(&BTRFS_I(inode)->io_tree, lockstart, lockend,
-				 cached_state);
-		ordered = btrfs_lookup_first_ordered_extent(inode, lockend);
-
+		lock_extent(&BTRFS_I(inode)->io_tree, lockstart, lockend,
+			    cached_state);
 		/*
-		 * We need to make sure we have no ordered extents in this range
-		 * and nobody raced in and read a page in this range, if we did
-		 * we need to try again.
+		 * We can't have ordered extents in the range, nor dirty/writeback
+		 * pages, because we have locked the inode's VFS lock in exclusive
+		 * mode, we have locked the inode's i_mmap_lock in exclusive mode,
+		 * we have flushed all delalloc in the range and we have waited
+		 * for any ordered extents in the range to complete.
+		 * We can race with anyone reading pages from this range, so after
+		 * locking the range check if we have pages in the range, and if
+		 * we do, unlock the range and retry.
 		 */
-		if ((!ordered ||
-		    (ordered->file_offset + ordered->num_bytes <= lockstart ||
-		     ordered->file_offset > lockend)) &&
-		     !filemap_range_has_page(inode->i_mapping,
-					     lockstart, lockend)) {
-			if (ordered)
-				btrfs_put_ordered_extent(ordered);
+		if (!filemap_range_has_page(inode->i_mapping, page_lockstart,
+					    page_lockend))
 			break;
-		}
-		if (ordered)
-			btrfs_put_ordered_extent(ordered);
-		unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart,
-				     lockend, cached_state);
-		ret = btrfs_wait_ordered_range(inode, lockstart,
-					       lockend - lockstart + 1);
-		if (ret)
-			return ret;
+
+		unlock_extent(&BTRFS_I(inode)->io_tree, lockstart, lockend,
+			      cached_state);
 	}
-	return 0;
+
+	btrfs_assert_inode_range_clean(BTRFS_I(inode), lockstart, lockend);
 }
 
-static int btrfs_insert_clone_extent(struct btrfs_trans_handle *trans,
-				     struct inode *inode,
+static int btrfs_insert_replace_extent(struct btrfs_trans_handle *trans,
+				     struct btrfs_inode *inode,
 				     struct btrfs_path *path,
-				     struct btrfs_clone_extent_info *clone_info,
-				     const u64 clone_len)
+				     struct btrfs_replace_extent_info *extent_info,
+				     const u64 replace_len,
+				     const u64 bytes_to_drop)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	struct btrfs_root *root = BTRFS_I(inode)->root;
+	struct btrfs_fs_info *fs_info = trans->fs_info;
+	struct btrfs_root *root = inode->root;
 	struct btrfs_file_extent_item *extent;
 	struct extent_buffer *leaf;
 	struct btrfs_key key;
 	int slot;
 	struct btrfs_ref ref = { 0 };
-	u64 ref_offset;
 	int ret;
 
-	if (clone_len == 0)
+	if (replace_len == 0)
 		return 0;
 
-	if (clone_info->disk_offset == 0 &&
-	    btrfs_fs_incompat(fs_info, NO_HOLES))
+	if (extent_info->disk_offset == 0 &&
+	    btrfs_fs_incompat(fs_info, NO_HOLES)) {
+		btrfs_update_inode_bytes(inode, 0, bytes_to_drop);
 		return 0;
+	}
 
-	key.objectid = btrfs_ino(BTRFS_I(inode));
+	key.objectid = btrfs_ino(inode);
 	key.type = BTRFS_EXTENT_DATA_KEY;
-	key.offset = clone_info->file_offset;
+	key.offset = extent_info->file_offset;
 	ret = btrfs_insert_empty_item(trans, root, path, &key,
-				      clone_info->item_size);
+				      sizeof(struct btrfs_file_extent_item));
 	if (ret)
 		return ret;
 	leaf = path->nodes[0];
 	slot = path->slots[0];
-	write_extent_buffer(leaf, clone_info->extent_buf,
+	write_extent_buffer(leaf, extent_info->extent_buf,
 			    btrfs_item_ptr_offset(leaf, slot),
-			    clone_info->item_size);
+			    sizeof(struct btrfs_file_extent_item));
 	extent = btrfs_item_ptr(leaf, slot, struct btrfs_file_extent_item);
-	btrfs_set_file_extent_offset(leaf, extent, clone_info->data_offset);
-	btrfs_set_file_extent_num_bytes(leaf, extent, clone_len);
+	ASSERT(btrfs_file_extent_type(leaf, extent) != BTRFS_FILE_EXTENT_INLINE);
+	btrfs_set_file_extent_offset(leaf, extent, extent_info->data_offset);
+	btrfs_set_file_extent_num_bytes(leaf, extent, replace_len);
+	if (extent_info->is_new_extent)
+		btrfs_set_file_extent_generation(leaf, extent, trans->transid);
 	btrfs_mark_buffer_dirty(leaf);
 	btrfs_release_path(path);
 
+	ret = btrfs_inode_set_file_extent_range(inode, extent_info->file_offset,
+						replace_len);
+	if (ret)
+		return ret;
+
 	/* If it's a hole, nothing more needs to be done. */
-	if (clone_info->disk_offset == 0)
+	if (extent_info->disk_offset == 0) {
+		btrfs_update_inode_bytes(inode, 0, bytes_to_drop);
 		return 0;
+	}
 
-	inode_add_bytes(inode, clone_len);
-	btrfs_init_generic_ref(&ref, BTRFS_ADD_DELAYED_REF,
-			       clone_info->disk_offset,
-			       clone_info->disk_len, 0);
-	ref_offset = clone_info->file_offset - clone_info->data_offset;
-	btrfs_init_data_ref(&ref, root->root_key.objectid,
-			    btrfs_ino(BTRFS_I(inode)), ref_offset);
-	ret = btrfs_inc_extent_ref(trans, &ref);
+	btrfs_update_inode_bytes(inode, replace_len, bytes_to_drop);
+
+	if (extent_info->is_new_extent && extent_info->insertions == 0) {
+		key.objectid = extent_info->disk_offset;
+		key.type = BTRFS_EXTENT_ITEM_KEY;
+		key.offset = extent_info->disk_len;
+		ret = btrfs_alloc_reserved_file_extent(trans, root,
+						       btrfs_ino(inode),
+						       extent_info->file_offset,
+						       extent_info->qgroup_reserved,
+						       &key);
+	} else {
+		u64 ref_offset;
+
+		btrfs_init_generic_ref(&ref, BTRFS_ADD_DELAYED_REF,
+				       extent_info->disk_offset,
+				       extent_info->disk_len, 0);
+		ref_offset = extent_info->file_offset - extent_info->data_offset;
+		btrfs_init_data_ref(&ref, root->root_key.objectid,
+				    btrfs_ino(inode), ref_offset, 0, false);
+		ret = btrfs_inc_extent_ref(trans, &ref);
+	}
+
+	extent_info->insertions++;
 
 	return ret;
 }
@@ -2505,26 +2622,27 @@ static int btrfs_insert_clone_extent(struct btrfs_trans_handle *trans,
 /*
  * The respective range must have been previously locked, as well as the inode.
  * The end offset is inclusive (last byte of the range).
- * @clone_info is NULL for fallocate's hole punching and non-NULL for extent
- * cloning.
- * When cloning, we don't want to end up in a state where we dropped extents
- * without inserting a new one, so we must abort the transaction to avoid a
- * corruption.
+ * @extent_info is NULL for fallocate's hole punching and non-NULL when replacing
+ * the file range with an extent.
+ * When not punching a hole, we don't want to end up in a state where we dropped
+ * extents without inserting a new one, so we must abort the transaction to avoid
+ * a corruption.
  */
-int btrfs_punch_hole_range(struct inode *inode, struct btrfs_path *path,
-			   const u64 start, const u64 end,
-			   struct btrfs_clone_extent_info *clone_info,
-			   struct btrfs_trans_handle **trans_out)
+int btrfs_replace_file_extents(struct btrfs_inode *inode,
+			       struct btrfs_path *path, const u64 start,
+			       const u64 end,
+			       struct btrfs_replace_extent_info *extent_info,
+			       struct btrfs_trans_handle **trans_out)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+	struct btrfs_drop_extents_args drop_args = { 0 };
+	struct btrfs_root *root = inode->root;
+	struct btrfs_fs_info *fs_info = root->fs_info;
 	u64 min_size = btrfs_calc_insert_metadata_size(fs_info, 1);
-	u64 ino_size = round_up(inode->i_size, fs_info->sectorsize);
-	struct btrfs_root *root = BTRFS_I(inode)->root;
+	u64 ino_size = round_up(inode->vfs_inode.i_size, fs_info->sectorsize);
 	struct btrfs_trans_handle *trans = NULL;
 	struct btrfs_block_rsv *rsv;
 	unsigned int rsv_count;
 	u64 cur_offset;
-	u64 drop_end;
 	u64 len = end - start;
 	int ret = 0;
 
@@ -2537,15 +2655,15 @@ int btrfs_punch_hole_range(struct inode *inode, struct btrfs_path *path,
 		goto out;
 	}
 	rsv->size = btrfs_calc_insert_metadata_size(fs_info, 1);
-	rsv->failfast = 1;
+	rsv->failfast = true;
 
 	/*
 	 * 1 - update the inode
 	 * 1 - removing the extents in the range
-	 * 1 - adding the hole extent if no_holes isn't set or if we are cloning
-	 *     an extent
+	 * 1 - adding the hole extent if no_holes isn't set or if we are
+	 *     replacing the range with a new extent
 	 */
-	if (!btrfs_fs_incompat(fs_info, NO_HOLES) || clone_info)
+	if (!btrfs_fs_incompat(fs_info, NO_HOLES) || extent_info)
 		rsv_count = 3;
 	else
 		rsv_count = 2;
@@ -2559,33 +2677,43 @@ int btrfs_punch_hole_range(struct inode *inode, struct btrfs_path *path,
 
 	ret = btrfs_block_rsv_migrate(&fs_info->trans_block_rsv, rsv,
 				      min_size, false);
-	BUG_ON(ret);
+	if (WARN_ON(ret))
+		goto out_trans;
 	trans->block_rsv = rsv;
 
 	cur_offset = start;
+	drop_args.path = path;
+	drop_args.end = end + 1;
+	drop_args.drop_cache = true;
 	while (cur_offset < end) {
-		ret = __btrfs_drop_extents(trans, root, inode, path,
-					   cur_offset, end + 1, &drop_end,
-					   1, 0, 0, NULL);
+		drop_args.start = cur_offset;
+		ret = btrfs_drop_extents(trans, root, inode, &drop_args);
+		/* If we are punching a hole decrement the inode's byte count */
+		if (!extent_info)
+			btrfs_update_inode_bytes(inode, 0,
+						 drop_args.bytes_found);
 		if (ret != -ENOSPC) {
 			/*
-			 * When cloning we want to avoid transaction aborts when
-			 * nothing was done and we are attempting to clone parts
-			 * of inline extents, in such cases -EOPNOTSUPP is
-			 * returned by __btrfs_drop_extents() without having
-			 * changed anything in the file.
+			 * The only time we don't want to abort is if we are
+			 * attempting to clone a partial inline extent, in which
+			 * case we'll get EOPNOTSUPP.  However if we aren't
+			 * clone we need to abort no matter what, because if we
+			 * got EOPNOTSUPP via prealloc then we messed up and
+			 * need to abort.
 			 */
-			if (clone_info && ret && ret != -EOPNOTSUPP)
+			if (ret &&
+			    (ret != -EOPNOTSUPP ||
+			     (extent_info && extent_info->is_new_extent)))
 				btrfs_abort_transaction(trans, ret);
 			break;
 		}
 
 		trans->block_rsv = &fs_info->trans_block_rsv;
 
-		if (!clone_info && cur_offset < drop_end &&
+		if (!extent_info && cur_offset < drop_args.drop_end &&
 		    cur_offset < ino_size) {
-			ret = fill_holes(trans, BTRFS_I(inode), path,
-					cur_offset, drop_end);
+			ret = fill_holes(trans, inode, path, cur_offset,
+					 drop_args.drop_end);
 			if (ret) {
 				/*
 				 * If we failed then we didn't insert our hole
@@ -2596,23 +2724,62 @@ int btrfs_punch_hole_range(struct inode *inode, struct btrfs_path *path,
 				btrfs_abort_transaction(trans, ret);
 				break;
 			}
+		} else if (!extent_info && cur_offset < drop_args.drop_end) {
+			/*
+			 * We are past the i_size here, but since we didn't
+			 * insert holes we need to clear the mapped area so we
+			 * know to not set disk_i_size in this area until a new
+			 * file extent is inserted here.
+			 */
+			ret = btrfs_inode_clear_file_extent_range(inode,
+					cur_offset,
+					drop_args.drop_end - cur_offset);
+			if (ret) {
+				/*
+				 * We couldn't clear our area, so we could
+				 * presumably adjust up and corrupt the fs, so
+				 * we need to abort.
+				 */
+				btrfs_abort_transaction(trans, ret);
+				break;
+			}
 		}
 
-		if (clone_info && drop_end > clone_info->file_offset) {
-			u64 clone_len = drop_end - clone_info->file_offset;
+		if (extent_info &&
+		    drop_args.drop_end > extent_info->file_offset) {
+			u64 replace_len = drop_args.drop_end -
+					  extent_info->file_offset;
 
-			ret = btrfs_insert_clone_extent(trans, inode, path,
-							clone_info, clone_len);
+			ret = btrfs_insert_replace_extent(trans, inode,	path,
+					extent_info, replace_len,
+					drop_args.bytes_found);
 			if (ret) {
 				btrfs_abort_transaction(trans, ret);
 				break;
 			}
-			clone_info->data_len -= clone_len;
-			clone_info->data_offset += clone_len;
-			clone_info->file_offset += clone_len;
+			extent_info->data_len -= replace_len;
+			extent_info->data_offset += replace_len;
+			extent_info->file_offset += replace_len;
 		}
 
-		cur_offset = drop_end;
+		/*
+		 * We are releasing our handle on the transaction, balance the
+		 * dirty pages of the btree inode and flush delayed items, and
+		 * then get a new transaction handle, which may now point to a
+		 * new transaction in case someone else may have committed the
+		 * transaction we used to replace/drop file extent items. So
+		 * bump the inode's iversion and update mtime and ctime except
+		 * if we are called from a dedupe context. This is because a
+		 * power failure/crash may happen after the transaction is
+		 * committed and before we finish replacing/dropping all the
+		 * file extent items we need.
+		 */
+		inode_inc_iversion(&inode->vfs_inode);
+
+		if (!extent_info || extent_info->update_times) {
+			inode->vfs_inode.i_mtime = current_time(&inode->vfs_inode);
+			inode->vfs_inode.i_ctime = inode->vfs_inode.i_mtime;
+		}
 
 		ret = btrfs_update_inode(trans, root, inode);
 		if (ret)
@@ -2630,10 +2797,13 @@ int btrfs_punch_hole_range(struct inode *inode, struct btrfs_path *path,
 
 		ret = btrfs_block_rsv_migrate(&fs_info->trans_block_rsv,
 					      rsv, min_size, false);
-		BUG_ON(ret);	/* shouldn't happen */
+		if (WARN_ON(ret))
+			break;
 		trans->block_rsv = rsv;
 
-		if (!clone_info) {
+		cur_offset = drop_args.drop_end;
+		len = end - cur_offset;
+		if (!extent_info && len) {
 			ret = find_first_non_hole(inode, &cur_offset, &len);
 			if (unlikely(ret < 0))
 				break;
@@ -2647,14 +2817,11 @@ int btrfs_punch_hole_range(struct inode *inode, struct btrfs_path *path,
 	/*
 	 * If we were cloning, force the next fsync to be a full one since we
 	 * we replaced (or just dropped in the case of cloning holes when
-	 * NO_HOLES is enabled) extents and extent maps.
-	 * This is for the sake of simplicity, and cloning into files larger
-	 * than 16Mb would force the full fsync any way (when
-	 * try_release_extent_mapping() is invoked during page cache truncation.
+	 * NO_HOLES is enabled) file extent items and did not setup new extent
+	 * maps for the replacement extents (or holes).
 	 */
-	if (clone_info)
-		set_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
-			&BTRFS_I(inode)->runtime_flags);
+	if (extent_info && !extent_info->is_new_extent)
+		btrfs_set_inode_full_sync(inode);
 
 	if (ret)
 		goto out_trans;
@@ -2671,25 +2838,36 @@ int btrfs_punch_hole_range(struct inode *inode, struct btrfs_path *path,
 	 * will not record the existence of the hole region
 	 * [existing_hole_start, lockend].
 	 */
-	if (drop_end <= end)
-		drop_end = end + 1;
+	if (drop_args.drop_end <= end)
+		drop_args.drop_end = end + 1;
 	/*
 	 * Don't insert file hole extent item if it's for a range beyond eof
 	 * (because it's useless) or if it represents a 0 bytes range (when
 	 * cur_offset == drop_end).
 	 */
-	if (!clone_info && cur_offset < ino_size && cur_offset < drop_end) {
-		ret = fill_holes(trans, BTRFS_I(inode), path,
-				cur_offset, drop_end);
+	if (!extent_info && cur_offset < ino_size &&
+	    cur_offset < drop_args.drop_end) {
+		ret = fill_holes(trans, inode, path, cur_offset,
+				 drop_args.drop_end);
 		if (ret) {
 			/* Same comment as above. */
 			btrfs_abort_transaction(trans, ret);
 			goto out_trans;
 		}
+	} else if (!extent_info && cur_offset < drop_args.drop_end) {
+		/* See the comment in the loop above for the reasoning here. */
+		ret = btrfs_inode_clear_file_extent_range(inode, cur_offset,
+					drop_args.drop_end - cur_offset);
+		if (ret) {
+			btrfs_abort_transaction(trans, ret);
+			goto out_trans;
+		}
+
 	}
-	if (clone_info) {
-		ret = btrfs_insert_clone_extent(trans, inode, path, clone_info,
-						clone_info->data_len);
+	if (extent_info) {
+		ret = btrfs_insert_replace_extent(trans, inode, path,
+				extent_info, extent_info->data_len,
+				drop_args.bytes_found);
 		if (ret) {
 			btrfs_abort_transaction(trans, ret);
 			goto out_trans;
@@ -2711,8 +2889,9 @@ out:
 	return ret;
 }
 
-static int btrfs_punch_hole(struct inode *inode, loff_t offset, loff_t len)
+static int btrfs_punch_hole(struct file *file, loff_t offset, loff_t len)
 {
+	struct inode *inode = file_inode(file);
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	struct btrfs_root *root = BTRFS_I(inode)->root;
 	struct extent_state *cached_state = NULL;
@@ -2729,13 +2908,14 @@ static int btrfs_punch_hole(struct inode *inode, loff_t offset, loff_t len)
 	bool truncated_block = false;
 	bool updated_inode = false;
 
+	btrfs_inode_lock(inode, BTRFS_ILOCK_MMAP);
+
 	ret = btrfs_wait_ordered_range(inode, offset, len);
 	if (ret)
-		return ret;
+		goto out_only_mutex;
 
-	inode_lock(inode);
 	ino_size = round_up(inode->i_size, fs_info->sectorsize);
-	ret = find_first_non_hole(inode, &offset, &len);
+	ret = find_first_non_hole(BTRFS_I(inode), &offset, &len);
 	if (ret < 0)
 		goto out_only_mutex;
 	if (ret && !len) {
@@ -2744,9 +2924,12 @@ static int btrfs_punch_hole(struct inode *inode, loff_t offset, loff_t len)
 		goto out_only_mutex;
 	}
 
-	lockstart = round_up(offset, btrfs_inode_sectorsize(inode));
-	lockend = round_down(offset + len,
-			     btrfs_inode_sectorsize(inode)) - 1;
+	ret = file_modified(file);
+	if (ret)
+		goto out_only_mutex;
+
+	lockstart = round_up(offset, fs_info->sectorsize);
+	lockend = round_down(offset + len, fs_info->sectorsize) - 1;
 	same_block = (BTRFS_BYTES_TO_BLKS(fs_info, offset))
 		== (BTRFS_BYTES_TO_BLKS(fs_info, offset + len - 1));
 	/*
@@ -2760,7 +2943,8 @@ static int btrfs_punch_hole(struct inode *inode, loff_t offset, loff_t len)
 	if (same_block && len < fs_info->sectorsize) {
 		if (offset < ino_size) {
 			truncated_block = true;
-			ret = btrfs_truncate_block(inode, offset, len, 0);
+			ret = btrfs_truncate_block(BTRFS_I(inode), offset, len,
+						   0);
 		} else {
 			ret = 0;
 		}
@@ -2770,9 +2954,9 @@ static int btrfs_punch_hole(struct inode *inode, loff_t offset, loff_t len)
 	/* zero back part of the first block */
 	if (offset < ino_size) {
 		truncated_block = true;
-		ret = btrfs_truncate_block(inode, offset, 0, 0);
+		ret = btrfs_truncate_block(BTRFS_I(inode), offset, 0, 0);
 		if (ret) {
-			inode_unlock(inode);
+			btrfs_inode_unlock(inode, BTRFS_ILOCK_MMAP);
 			return ret;
 		}
 	}
@@ -2785,7 +2969,7 @@ static int btrfs_punch_hole(struct inode *inode, loff_t offset, loff_t len)
 		/* after truncate page, check hole again */
 		len = offset + len - lockstart;
 		offset = lockstart;
-		ret = find_first_non_hole(inode, &offset, &len);
+		ret = find_first_non_hole(BTRFS_I(inode), &offset, &len);
 		if (ret < 0)
 			goto out_only_mutex;
 		if (ret && !len) {
@@ -2799,14 +2983,14 @@ static int btrfs_punch_hole(struct inode *inode, loff_t offset, loff_t len)
 	tail_start = lockend + 1;
 	tail_len = offset + len - tail_start;
 	if (tail_len) {
-		ret = find_first_non_hole(inode, &tail_start, &tail_len);
+		ret = find_first_non_hole(BTRFS_I(inode), &tail_start, &tail_len);
 		if (unlikely(ret < 0))
 			goto out_only_mutex;
 		if (!ret) {
 			/* zero the front end of the last page */
 			if (tail_start + tail_len < ino_size) {
 				truncated_block = true;
-				ret = btrfs_truncate_block(inode,
+				ret = btrfs_truncate_block(BTRFS_I(inode),
 							tail_start + tail_len,
 							0, 1);
 				if (ret)
@@ -2820,10 +3004,7 @@ static int btrfs_punch_hole(struct inode *inode, loff_t offset, loff_t len)
 		goto out_only_mutex;
 	}
 
-	ret = btrfs_punch_hole_lock_range(inode, lockstart, lockend,
-					  &cached_state);
-	if (ret)
-		goto out_only_mutex;
+	btrfs_punch_hole_lock_range(inode, lockstart, lockend, &cached_state);
 
 	path = btrfs_alloc_path();
 	if (!path) {
@@ -2831,22 +3012,23 @@ static int btrfs_punch_hole(struct inode *inode, loff_t offset, loff_t len)
 		goto out;
 	}
 
-	ret = btrfs_punch_hole_range(inode, path, lockstart, lockend, NULL,
-				     &trans);
+	ret = btrfs_replace_file_extents(BTRFS_I(inode), path, lockstart,
+					 lockend, NULL, &trans);
 	btrfs_free_path(path);
 	if (ret)
 		goto out;
 
 	ASSERT(trans != NULL);
 	inode_inc_iversion(inode);
-	inode->i_mtime = inode->i_ctime = current_time(inode);
-	ret = btrfs_update_inode(trans, root, inode);
+	inode->i_mtime = current_time(inode);
+	inode->i_ctime = inode->i_mtime;
+	ret = btrfs_update_inode(trans, root, BTRFS_I(inode));
 	updated_inode = true;
 	btrfs_end_transaction(trans);
 	btrfs_btree_balance_dirty(fs_info);
 out:
-	unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart, lockend,
-			     &cached_state);
+	unlock_extent(&BTRFS_I(inode)->io_tree, lockstart, lockend,
+		      &cached_state);
 out_only_mutex:
 	if (!updated_inode && truncated_block && !ret) {
 		/*
@@ -2867,13 +3049,13 @@ out_only_mutex:
 		} else {
 			int ret2;
 
-			ret = btrfs_update_inode(trans, root, inode);
+			ret = btrfs_update_inode(trans, root, BTRFS_I(inode));
 			ret2 = btrfs_end_transaction(trans);
 			if (!ret)
 				ret = ret2;
 		}
 	}
-	inode_unlock(inode);
+	btrfs_inode_unlock(inode, BTRFS_ILOCK_MMAP);
 	return ret;
 }
 
@@ -2892,22 +3074,20 @@ struct falloc_range {
  */
 static int add_falloc_range(struct list_head *head, u64 start, u64 len)
 {
-	struct falloc_range *prev = NULL;
 	struct falloc_range *range = NULL;
 
-	if (list_empty(head))
-		goto insert;
-
-	/*
-	 * As fallocate iterate by bytenr order, we only need to check
-	 * the last range.
-	 */
-	prev = list_entry(head->prev, struct falloc_range, list);
-	if (prev->start + prev->len == start) {
-		prev->len += len;
-		return 0;
+	if (!list_empty(head)) {
+		/*
+		 * As fallocate iterates by bytenr order, we only need to check
+		 * the last range.
+		 */
+		range = list_last_entry(head, struct falloc_range, list);
+		if (range->start + range->len == start) {
+			range->len += len;
+			return 0;
+		}
 	}
-insert:
+
 	range = kmalloc(sizeof(*range), GFP_KERNEL);
 	if (!range)
 		return -ENOMEM;
@@ -2935,8 +3115,8 @@ static int btrfs_fallocate_update_isize(struct inode *inode,
 
 	inode->i_ctime = current_time(inode);
 	i_size_write(inode, end);
-	btrfs_ordered_update_i_size(inode, end, NULL);
-	ret = btrfs_update_inode(trans, root, inode);
+	btrfs_inode_safe_disk_i_size_write(BTRFS_I(inode), 0);
+	ret = btrfs_update_inode(trans, root, BTRFS_I(inode));
 	ret2 = btrfs_end_transaction(trans);
 
 	return ret ? ret : ret2;
@@ -2948,15 +3128,15 @@ enum {
 	RANGE_BOUNDARY_HOLE,
 };
 
-static int btrfs_zero_range_check_range_boundary(struct inode *inode,
+static int btrfs_zero_range_check_range_boundary(struct btrfs_inode *inode,
 						 u64 offset)
 {
-	const u64 sectorsize = btrfs_inode_sectorsize(inode);
+	const u64 sectorsize = inode->root->fs_info->sectorsize;
 	struct extent_map *em;
 	int ret;
 
 	offset = round_down(offset, sectorsize);
-	em = btrfs_get_extent(BTRFS_I(inode), NULL, 0, offset, sectorsize);
+	em = btrfs_get_extent(inode, NULL, 0, offset, sectorsize);
 	if (IS_ERR(em))
 		return PTR_ERR(em);
 
@@ -2981,14 +3161,12 @@ static int btrfs_zero_range(struct inode *inode,
 	struct extent_changeset *data_reserved = NULL;
 	int ret;
 	u64 alloc_hint = 0;
-	const u64 sectorsize = btrfs_inode_sectorsize(inode);
+	const u64 sectorsize = fs_info->sectorsize;
 	u64 alloc_start = round_down(offset, sectorsize);
 	u64 alloc_end = round_up(offset + len, sectorsize);
 	u64 bytes_to_reserve = 0;
 	bool space_reserved = false;
 
-	inode_dio_wait(inode);
-
 	em = btrfs_get_extent(BTRFS_I(inode), NULL, 0, alloc_start,
 			      alloc_end - alloc_start);
 	if (IS_ERR(em)) {
@@ -3048,7 +3226,8 @@ static int btrfs_zero_range(struct inode *inode,
 		}
 		if (len < sectorsize && em->block_start != EXTENT_MAP_HOLE) {
 			free_extent_map(em);
-			ret = btrfs_truncate_block(inode, offset, len, 0);
+			ret = btrfs_truncate_block(BTRFS_I(inode), offset, len,
+						   0);
 			if (!ret)
 				ret = btrfs_fallocate_update_isize(inode,
 								   offset + len,
@@ -3071,14 +3250,15 @@ static int btrfs_zero_range(struct inode *inode,
 	 * to cover them.
 	 */
 	if (!IS_ALIGNED(offset, sectorsize)) {
-		ret = btrfs_zero_range_check_range_boundary(inode, offset);
+		ret = btrfs_zero_range_check_range_boundary(BTRFS_I(inode),
+							    offset);
 		if (ret < 0)
 			goto out;
 		if (ret == RANGE_BOUNDARY_HOLE) {
 			alloc_start = round_down(offset, sectorsize);
 			ret = 0;
 		} else if (ret == RANGE_BOUNDARY_WRITTEN_EXTENT) {
-			ret = btrfs_truncate_block(inode, offset, 0, 0);
+			ret = btrfs_truncate_block(BTRFS_I(inode), offset, 0, 0);
 			if (ret)
 				goto out;
 		} else {
@@ -3087,7 +3267,7 @@ static int btrfs_zero_range(struct inode *inode,
 	}
 
 	if (!IS_ALIGNED(offset + len, sectorsize)) {
-		ret = btrfs_zero_range_check_range_boundary(inode,
+		ret = btrfs_zero_range_check_range_boundary(BTRFS_I(inode),
 							    offset + len);
 		if (ret < 0)
 			goto out;
@@ -3095,7 +3275,8 @@ static int btrfs_zero_range(struct inode *inode,
 			alloc_end = round_up(offset + len, sectorsize);
 			ret = 0;
 		} else if (ret == RANGE_BOUNDARY_WRITTEN_EXTENT) {
-			ret = btrfs_truncate_block(inode, offset + len, 0, 1);
+			ret = btrfs_truncate_block(BTRFS_I(inode), offset + len,
+						   0, 1);
 			if (ret)
 				goto out;
 		} else {
@@ -3115,20 +3296,21 @@ reserve_space:
 		if (ret < 0)
 			goto out;
 		space_reserved = true;
-		ret = btrfs_qgroup_reserve_data(inode, &data_reserved,
+		btrfs_punch_hole_lock_range(inode, lockstart, lockend,
+					    &cached_state);
+		ret = btrfs_qgroup_reserve_data(BTRFS_I(inode), &data_reserved,
 						alloc_start, bytes_to_reserve);
-		if (ret)
-			goto out;
-		ret = btrfs_punch_hole_lock_range(inode, lockstart, lockend,
-						  &cached_state);
-		if (ret)
+		if (ret) {
+			unlock_extent(&BTRFS_I(inode)->io_tree, lockstart,
+				      lockend, &cached_state);
 			goto out;
+		}
 		ret = btrfs_prealloc_file_range(inode, mode, alloc_start,
 						alloc_end - alloc_start,
 						i_blocksize(inode),
 						offset + len, &alloc_hint);
-		unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart,
-				     lockend, &cached_state);
+		unlock_extent(&BTRFS_I(inode)->io_tree, lockstart, lockend,
+			      &cached_state);
 		/* btrfs_prealloc_file_range releases reserved space on error */
 		if (ret) {
 			space_reserved = false;
@@ -3138,7 +3320,7 @@ reserve_space:
 	ret = btrfs_fallocate_update_isize(inode, offset + len, mode);
  out:
 	if (ret && space_reserved)
-		btrfs_free_reserved_data_space(inode, data_reserved,
+		btrfs_free_reserved_data_space(BTRFS_I(inode), data_reserved,
 					       alloc_start, bytes_to_reserve);
 	extent_changeset_free(data_reserved);
 
@@ -3161,10 +3343,17 @@ static long btrfs_fallocate(struct file *file, int mode,
 	u64 alloc_hint = 0;
 	u64 locked_end;
 	u64 actual_end = 0;
+	u64 data_space_needed = 0;
+	u64 data_space_reserved = 0;
+	u64 qgroup_reserved = 0;
 	struct extent_map *em;
-	int blocksize = btrfs_inode_sectorsize(inode);
+	int blocksize = BTRFS_I(inode)->root->fs_info->sectorsize;
 	int ret;
 
+	/* Do not allow fallocate in ZONED mode */
+	if (btrfs_is_zoned(btrfs_sb(inode->i_sb)))
+		return -EOPNOTSUPP;
+
 	alloc_start = round_down(offset, blocksize);
 	alloc_end = round_up(offset + len, blocksize);
 	cur_offset = alloc_start;
@@ -3175,21 +3364,9 @@ static long btrfs_fallocate(struct file *file, int mode,
 		return -EOPNOTSUPP;
 
 	if (mode & FALLOC_FL_PUNCH_HOLE)
-		return btrfs_punch_hole(inode, offset, len);
-
-	/*
-	 * Only trigger disk allocation, don't trigger qgroup reserve
-	 *
-	 * For qgroup space, it will be checked later.
-	 */
-	if (!(mode & FALLOC_FL_ZERO_RANGE)) {
-		ret = btrfs_alloc_data_chunk_ondemand(BTRFS_I(inode),
-						      alloc_end - alloc_start);
-		if (ret < 0)
-			return ret;
-	}
+		return btrfs_punch_hole(file, offset, len);
 
-	inode_lock(inode);
+	btrfs_inode_lock(inode, BTRFS_ILOCK_MMAP);
 
 	if (!(mode & FALLOC_FL_KEEP_SIZE) && offset + len > inode->i_size) {
 		ret = inode_newsize_ok(inode, offset + len);
@@ -3197,6 +3374,10 @@ static long btrfs_fallocate(struct file *file, int mode,
 			goto out;
 	}
 
+	ret = file_modified(file);
+	if (ret)
+		goto out;
+
 	/*
 	 * TODO: Move these two operations after we have checked
 	 * accurate reserved space, or fallocate can still fail but
@@ -3205,7 +3386,7 @@ static long btrfs_fallocate(struct file *file, int mode,
 	 * But that's a minor problem and won't do much harm BTW.
 	 */
 	if (alloc_start > inode->i_size) {
-		ret = btrfs_cont_expand(inode, i_size_read(inode),
+		ret = btrfs_cont_expand(BTRFS_I(inode), i_size_read(inode),
 					alloc_start);
 		if (ret)
 			goto out;
@@ -3215,14 +3396,18 @@ static long btrfs_fallocate(struct file *file, int mode,
 		 * need to zero out the end of the block if i_size lands in the
 		 * middle of a block.
 		 */
-		ret = btrfs_truncate_block(inode, inode->i_size, 0, 0);
+		ret = btrfs_truncate_block(BTRFS_I(inode), inode->i_size, 0, 0);
 		if (ret)
 			goto out;
 	}
 
 	/*
-	 * wait for ordered IO before we have any locks.  We'll loop again
-	 * below with the locks held.
+	 * We have locked the inode at the VFS level (in exclusive mode) and we
+	 * have locked the i_mmap_lock lock (in exclusive mode). Now before
+	 * locking the file range, flush all dealloc in the range and wait for
+	 * all ordered extents in the range to complete. After this we can lock
+	 * the file range and, due to the previous locking we did, we know there
+	 * can't be more delalloc or ordered extents in the range.
 	 */
 	ret = btrfs_wait_ordered_range(inode, alloc_start,
 				       alloc_end - alloc_start);
@@ -3231,42 +3416,15 @@ static long btrfs_fallocate(struct file *file, int mode,
 
 	if (mode & FALLOC_FL_ZERO_RANGE) {
 		ret = btrfs_zero_range(inode, offset, len, mode);
-		inode_unlock(inode);
+		btrfs_inode_unlock(inode, BTRFS_ILOCK_MMAP);
 		return ret;
 	}
 
 	locked_end = alloc_end - 1;
-	while (1) {
-		struct btrfs_ordered_extent *ordered;
+	lock_extent(&BTRFS_I(inode)->io_tree, alloc_start, locked_end,
+		    &cached_state);
 
-		/* the extent lock is ordered inside the running
-		 * transaction
-		 */
-		lock_extent_bits(&BTRFS_I(inode)->io_tree, alloc_start,
-				 locked_end, &cached_state);
-		ordered = btrfs_lookup_first_ordered_extent(inode, locked_end);
-
-		if (ordered &&
-		    ordered->file_offset + ordered->num_bytes > alloc_start &&
-		    ordered->file_offset < alloc_end) {
-			btrfs_put_ordered_extent(ordered);
-			unlock_extent_cached(&BTRFS_I(inode)->io_tree,
-					     alloc_start, locked_end,
-					     &cached_state);
-			/*
-			 * we can't wait on the range with the transaction
-			 * running or with the extent lock held
-			 */
-			ret = btrfs_wait_ordered_range(inode, alloc_start,
-						       alloc_end - alloc_start);
-			if (ret)
-				goto out;
-		} else {
-			if (ordered)
-				btrfs_put_ordered_extent(ordered);
-			break;
-		}
-	}
+	btrfs_assert_inode_range_clean(BTRFS_I(inode), alloc_start, locked_end);
 
 	/* First, check if we exceed the qgroup limit */
 	INIT_LIST_HEAD(&reserve_list);
@@ -3283,46 +3441,64 @@ static long btrfs_fallocate(struct file *file, int mode,
 		if (em->block_start == EXTENT_MAP_HOLE ||
 		    (cur_offset >= inode->i_size &&
 		     !test_bit(EXTENT_FLAG_PREALLOC, &em->flags))) {
-			ret = add_falloc_range(&reserve_list, cur_offset,
-					       last_byte - cur_offset);
+			const u64 range_len = last_byte - cur_offset;
+
+			ret = add_falloc_range(&reserve_list, cur_offset, range_len);
 			if (ret < 0) {
 				free_extent_map(em);
 				break;
 			}
-			ret = btrfs_qgroup_reserve_data(inode, &data_reserved,
-					cur_offset, last_byte - cur_offset);
+			ret = btrfs_qgroup_reserve_data(BTRFS_I(inode),
+					&data_reserved, cur_offset, range_len);
 			if (ret < 0) {
-				cur_offset = last_byte;
 				free_extent_map(em);
 				break;
 			}
-		} else {
-			/*
-			 * Do not need to reserve unwritten extent for this
-			 * range, free reserved data space first, otherwise
-			 * it'll result in false ENOSPC error.
-			 */
-			btrfs_free_reserved_data_space(inode, data_reserved,
-					cur_offset, last_byte - cur_offset);
+			qgroup_reserved += range_len;
+			data_space_needed += range_len;
 		}
 		free_extent_map(em);
 		cur_offset = last_byte;
 	}
 
+	if (!ret && data_space_needed > 0) {
+		/*
+		 * We are safe to reserve space here as we can't have delalloc
+		 * in the range, see above.
+		 */
+		ret = btrfs_alloc_data_chunk_ondemand(BTRFS_I(inode),
+						      data_space_needed);
+		if (!ret)
+			data_space_reserved = data_space_needed;
+	}
+
 	/*
 	 * If ret is still 0, means we're OK to fallocate.
 	 * Or just cleanup the list and exit.
 	 */
 	list_for_each_entry_safe(range, tmp, &reserve_list, list) {
-		if (!ret)
+		if (!ret) {
 			ret = btrfs_prealloc_file_range(inode, mode,
 					range->start,
 					range->len, i_blocksize(inode),
 					offset + len, &alloc_hint);
-		else
-			btrfs_free_reserved_data_space(inode,
-					data_reserved, range->start,
-					range->len);
+			/*
+			 * btrfs_prealloc_file_range() releases space even
+			 * if it returns an error.
+			 */
+			data_space_reserved -= range->len;
+			qgroup_reserved -= range->len;
+		} else if (data_space_reserved > 0) {
+			btrfs_free_reserved_data_space(BTRFS_I(inode),
+					       data_reserved, range->start,
+					       range->len);
+			data_space_reserved -= range->len;
+			qgroup_reserved -= range->len;
+		} else if (qgroup_reserved > 0) {
+			btrfs_qgroup_free_data(BTRFS_I(inode), data_reserved,
+					       range->start, range->len);
+			qgroup_reserved -= range->len;
+		}
 		list_del(&range->list);
 		kfree(range);
 	}
@@ -3335,35 +3511,290 @@ static long btrfs_fallocate(struct file *file, int mode,
 	 */
 	ret = btrfs_fallocate_update_isize(inode, actual_end, mode);
 out_unlock:
-	unlock_extent_cached(&BTRFS_I(inode)->io_tree, alloc_start, locked_end,
-			     &cached_state);
+	unlock_extent(&BTRFS_I(inode)->io_tree, alloc_start, locked_end,
+		      &cached_state);
 out:
-	inode_unlock(inode);
-	/* Let go of our reservation. */
-	if (ret != 0 && !(mode & FALLOC_FL_ZERO_RANGE))
-		btrfs_free_reserved_data_space(inode, data_reserved,
-				cur_offset, alloc_end - cur_offset);
+	btrfs_inode_unlock(inode, BTRFS_ILOCK_MMAP);
 	extent_changeset_free(data_reserved);
 	return ret;
 }
 
-static loff_t find_desired_extent(struct inode *inode, loff_t offset,
+/*
+ * Helper for btrfs_find_delalloc_in_range(). Find a subrange in a given range
+ * that has unflushed and/or flushing delalloc. There might be other adjacent
+ * subranges after the one it found, so btrfs_find_delalloc_in_range() keeps
+ * looping while it gets adjacent subranges, and merging them together.
+ */
+static bool find_delalloc_subrange(struct btrfs_inode *inode, u64 start, u64 end,
+				   u64 *delalloc_start_ret, u64 *delalloc_end_ret)
+{
+	const u64 len = end + 1 - start;
+	struct extent_map_tree *em_tree = &inode->extent_tree;
+	struct extent_map *em;
+	u64 em_end;
+	u64 delalloc_len;
+
+	/*
+	 * Search the io tree first for EXTENT_DELALLOC. If we find any, it
+	 * means we have delalloc (dirty pages) for which writeback has not
+	 * started yet.
+	 */
+	*delalloc_start_ret = start;
+	delalloc_len = count_range_bits(&inode->io_tree, delalloc_start_ret, end,
+					len, EXTENT_DELALLOC, 1);
+	/*
+	 * If delalloc was found then *delalloc_start_ret has a sector size
+	 * aligned value (rounded down).
+	 */
+	if (delalloc_len > 0)
+		*delalloc_end_ret = *delalloc_start_ret + delalloc_len - 1;
+
+	/*
+	 * Now also check if there's any extent map in the range that does not
+	 * map to a hole or prealloc extent. We do this because:
+	 *
+	 * 1) When delalloc is flushed, the file range is locked, we clear the
+	 *    EXTENT_DELALLOC bit from the io tree and create an extent map for
+	 *    an allocated extent. So we might just have been called after
+	 *    delalloc is flushed and before the ordered extent completes and
+	 *    inserts the new file extent item in the subvolume's btree;
+	 *
+	 * 2) We may have an extent map created by flushing delalloc for a
+	 *    subrange that starts before the subrange we found marked with
+	 *    EXTENT_DELALLOC in the io tree.
+	 */
+	read_lock(&em_tree->lock);
+	em = lookup_extent_mapping(em_tree, start, len);
+	read_unlock(&em_tree->lock);
+
+	/* extent_map_end() returns a non-inclusive end offset. */
+	em_end = em ? extent_map_end(em) : 0;
+
+	/*
+	 * If we have a hole/prealloc extent map, check the next one if this one
+	 * ends before our range's end.
+	 */
+	if (em && (em->block_start == EXTENT_MAP_HOLE ||
+		   test_bit(EXTENT_FLAG_PREALLOC, &em->flags)) && em_end < end) {
+		struct extent_map *next_em;
+
+		read_lock(&em_tree->lock);
+		next_em = lookup_extent_mapping(em_tree, em_end, len - em_end);
+		read_unlock(&em_tree->lock);
+
+		free_extent_map(em);
+		em_end = next_em ? extent_map_end(next_em) : 0;
+		em = next_em;
+	}
+
+	if (em && (em->block_start == EXTENT_MAP_HOLE ||
+		   test_bit(EXTENT_FLAG_PREALLOC, &em->flags))) {
+		free_extent_map(em);
+		em = NULL;
+	}
+
+	/*
+	 * No extent map or one for a hole or prealloc extent. Use the delalloc
+	 * range we found in the io tree if we have one.
+	 */
+	if (!em)
+		return (delalloc_len > 0);
+
+	/*
+	 * We don't have any range as EXTENT_DELALLOC in the io tree, so the
+	 * extent map is the only subrange representing delalloc.
+	 */
+	if (delalloc_len == 0) {
+		*delalloc_start_ret = em->start;
+		*delalloc_end_ret = min(end, em_end - 1);
+		free_extent_map(em);
+		return true;
+	}
+
+	/*
+	 * The extent map represents a delalloc range that starts before the
+	 * delalloc range we found in the io tree.
+	 */
+	if (em->start < *delalloc_start_ret) {
+		*delalloc_start_ret = em->start;
+		/*
+		 * If the ranges are adjacent, return a combined range.
+		 * Otherwise return the extent map's range.
+		 */
+		if (em_end < *delalloc_start_ret)
+			*delalloc_end_ret = min(end, em_end - 1);
+
+		free_extent_map(em);
+		return true;
+	}
+
+	/*
+	 * The extent map starts after the delalloc range we found in the io
+	 * tree. If it's adjacent, return a combined range, otherwise return
+	 * the range found in the io tree.
+	 */
+	if (*delalloc_end_ret + 1 == em->start)
+		*delalloc_end_ret = min(end, em_end - 1);
+
+	free_extent_map(em);
+	return true;
+}
+
+/*
+ * Check if there's delalloc in a given range.
+ *
+ * @inode:               The inode.
+ * @start:               The start offset of the range. It does not need to be
+ *                       sector size aligned.
+ * @end:                 The end offset (inclusive value) of the search range.
+ *                       It does not need to be sector size aligned.
+ * @delalloc_start_ret:  Output argument, set to the start offset of the
+ *                       subrange found with delalloc (may not be sector size
+ *                       aligned).
+ * @delalloc_end_ret:    Output argument, set to he end offset (inclusive value)
+ *                       of the subrange found with delalloc.
+ *
+ * Returns true if a subrange with delalloc is found within the given range, and
+ * if so it sets @delalloc_start_ret and @delalloc_end_ret with the start and
+ * end offsets of the subrange.
+ */
+bool btrfs_find_delalloc_in_range(struct btrfs_inode *inode, u64 start, u64 end,
+				  u64 *delalloc_start_ret, u64 *delalloc_end_ret)
+{
+	u64 cur_offset = round_down(start, inode->root->fs_info->sectorsize);
+	u64 prev_delalloc_end = 0;
+	bool ret = false;
+
+	while (cur_offset < end) {
+		u64 delalloc_start;
+		u64 delalloc_end;
+		bool delalloc;
+
+		delalloc = find_delalloc_subrange(inode, cur_offset, end,
+						  &delalloc_start,
+						  &delalloc_end);
+		if (!delalloc)
+			break;
+
+		if (prev_delalloc_end == 0) {
+			/* First subrange found. */
+			*delalloc_start_ret = max(delalloc_start, start);
+			*delalloc_end_ret = delalloc_end;
+			ret = true;
+		} else if (delalloc_start == prev_delalloc_end + 1) {
+			/* Subrange adjacent to the previous one, merge them. */
+			*delalloc_end_ret = delalloc_end;
+		} else {
+			/* Subrange not adjacent to the previous one, exit. */
+			break;
+		}
+
+		prev_delalloc_end = delalloc_end;
+		cur_offset = delalloc_end + 1;
+		cond_resched();
+	}
+
+	return ret;
+}
+
+/*
+ * Check if there's a hole or delalloc range in a range representing a hole (or
+ * prealloc extent) found in the inode's subvolume btree.
+ *
+ * @inode:      The inode.
+ * @whence:     Seek mode (SEEK_DATA or SEEK_HOLE).
+ * @start:      Start offset of the hole region. It does not need to be sector
+ *              size aligned.
+ * @end:        End offset (inclusive value) of the hole region. It does not
+ *              need to be sector size aligned.
+ * @start_ret:  Return parameter, used to set the start of the subrange in the
+ *              hole that matches the search criteria (seek mode), if such
+ *              subrange is found (return value of the function is true).
+ *              The value returned here may not be sector size aligned.
+ *
+ * Returns true if a subrange matching the given seek mode is found, and if one
+ * is found, it updates @start_ret with the start of the subrange.
+ */
+static bool find_desired_extent_in_hole(struct btrfs_inode *inode, int whence,
+					u64 start, u64 end, u64 *start_ret)
+{
+	u64 delalloc_start;
+	u64 delalloc_end;
+	bool delalloc;
+
+	delalloc = btrfs_find_delalloc_in_range(inode, start, end,
+						&delalloc_start, &delalloc_end);
+	if (delalloc && whence == SEEK_DATA) {
+		*start_ret = delalloc_start;
+		return true;
+	}
+
+	if (delalloc && whence == SEEK_HOLE) {
+		/*
+		 * We found delalloc but it starts after out start offset. So we
+		 * have a hole between our start offset and the delalloc start.
+		 */
+		if (start < delalloc_start) {
+			*start_ret = start;
+			return true;
+		}
+		/*
+		 * Delalloc range starts at our start offset.
+		 * If the delalloc range's length is smaller than our range,
+		 * then it means we have a hole that starts where the delalloc
+		 * subrange ends.
+		 */
+		if (delalloc_end < end) {
+			*start_ret = delalloc_end + 1;
+			return true;
+		}
+
+		/* There's delalloc for the whole range. */
+		return false;
+	}
+
+	if (!delalloc && whence == SEEK_HOLE) {
+		*start_ret = start;
+		return true;
+	}
+
+	/*
+	 * No delalloc in the range and we are seeking for data. The caller has
+	 * to iterate to the next extent item in the subvolume btree.
+	 */
+	return false;
+}
+
+static loff_t find_desired_extent(struct btrfs_inode *inode, loff_t offset,
 				  int whence)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	struct extent_map *em = NULL;
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
 	struct extent_state *cached_state = NULL;
-	loff_t i_size = inode->i_size;
+	const loff_t i_size = i_size_read(&inode->vfs_inode);
+	const u64 ino = btrfs_ino(inode);
+	struct btrfs_root *root = inode->root;
+	struct btrfs_path *path;
+	struct btrfs_key key;
+	u64 last_extent_end;
 	u64 lockstart;
 	u64 lockend;
 	u64 start;
-	u64 len;
-	int ret = 0;
+	int ret;
+	bool found = false;
 
 	if (i_size == 0 || offset >= i_size)
 		return -ENXIO;
 
 	/*
+	 * Quick path. If the inode has no prealloc extents and its number of
+	 * bytes used matches its i_size, then it can not have holes.
+	 */
+	if (whence == SEEK_HOLE &&
+	    !(inode->flags & BTRFS_INODE_PREALLOC) &&
+	    inode_get_bytes(&inode->vfs_inode) == i_size)
+		return i_size;
+
+	/*
 	 * offset can be negative, in this case we start finding DATA/HOLE from
 	 * the very start of the file.
 	 */
@@ -3374,46 +3805,164 @@ static loff_t find_desired_extent(struct inode *inode, loff_t offset,
 	if (lockend <= lockstart)
 		lockend = lockstart + fs_info->sectorsize;
 	lockend--;
-	len = lockend - lockstart + 1;
 
-	lock_extent_bits(&BTRFS_I(inode)->io_tree, lockstart, lockend,
-			 &cached_state);
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+	path->reada = READA_FORWARD;
+
+	key.objectid = ino;
+	key.type = BTRFS_EXTENT_DATA_KEY;
+	key.offset = start;
+
+	last_extent_end = lockstart;
+
+	lock_extent(&inode->io_tree, lockstart, lockend, &cached_state);
+
+	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
+	if (ret < 0) {
+		goto out;
+	} else if (ret > 0 && path->slots[0] > 0) {
+		btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0] - 1);
+		if (key.objectid == ino && key.type == BTRFS_EXTENT_DATA_KEY)
+			path->slots[0]--;
+	}
 
 	while (start < i_size) {
-		em = btrfs_get_extent_fiemap(BTRFS_I(inode), start, len);
-		if (IS_ERR(em)) {
-			ret = PTR_ERR(em);
-			em = NULL;
-			break;
+		struct extent_buffer *leaf = path->nodes[0];
+		struct btrfs_file_extent_item *extent;
+		u64 extent_end;
+
+		if (path->slots[0] >= btrfs_header_nritems(leaf)) {
+			ret = btrfs_next_leaf(root, path);
+			if (ret < 0)
+				goto out;
+			else if (ret > 0)
+				break;
+
+			leaf = path->nodes[0];
 		}
 
-		if (whence == SEEK_HOLE &&
-		    (em->block_start == EXTENT_MAP_HOLE ||
-		     test_bit(EXTENT_FLAG_PREALLOC, &em->flags)))
-			break;
-		else if (whence == SEEK_DATA &&
-			   (em->block_start != EXTENT_MAP_HOLE &&
-			    !test_bit(EXTENT_FLAG_PREALLOC, &em->flags)))
+		btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
+		if (key.objectid != ino || key.type != BTRFS_EXTENT_DATA_KEY)
 			break;
 
-		start = em->start + em->len;
-		free_extent_map(em);
-		em = NULL;
+		extent_end = btrfs_file_extent_end(path);
+
+		/*
+		 * In the first iteration we may have a slot that points to an
+		 * extent that ends before our start offset, so skip it.
+		 */
+		if (extent_end <= start) {
+			path->slots[0]++;
+			continue;
+		}
+
+		/* We have an implicit hole, NO_HOLES feature is likely set. */
+		if (last_extent_end < key.offset) {
+			u64 search_start = last_extent_end;
+			u64 found_start;
+
+			/*
+			 * First iteration, @start matches @offset and it's
+			 * within the hole.
+			 */
+			if (start == offset)
+				search_start = offset;
+
+			found = find_desired_extent_in_hole(inode, whence,
+							    search_start,
+							    key.offset - 1,
+							    &found_start);
+			if (found) {
+				start = found_start;
+				break;
+			}
+			/*
+			 * Didn't find data or a hole (due to delalloc) in the
+			 * implicit hole range, so need to analyze the extent.
+			 */
+		}
+
+		extent = btrfs_item_ptr(leaf, path->slots[0],
+					struct btrfs_file_extent_item);
+
+		if (btrfs_file_extent_disk_bytenr(leaf, extent) == 0 ||
+		    btrfs_file_extent_type(leaf, extent) ==
+		    BTRFS_FILE_EXTENT_PREALLOC) {
+			/*
+			 * Explicit hole or prealloc extent, search for delalloc.
+			 * A prealloc extent is treated like a hole.
+			 */
+			u64 search_start = key.offset;
+			u64 found_start;
+
+			/*
+			 * First iteration, @start matches @offset and it's
+			 * within the hole.
+			 */
+			if (start == offset)
+				search_start = offset;
+
+			found = find_desired_extent_in_hole(inode, whence,
+							    search_start,
+							    extent_end - 1,
+							    &found_start);
+			if (found) {
+				start = found_start;
+				break;
+			}
+			/*
+			 * Didn't find data or a hole (due to delalloc) in the
+			 * implicit hole range, so need to analyze the next
+			 * extent item.
+			 */
+		} else {
+			/*
+			 * Found a regular or inline extent.
+			 * If we are seeking for data, adjust the start offset
+			 * and stop, we're done.
+			 */
+			if (whence == SEEK_DATA) {
+				start = max_t(u64, key.offset, offset);
+				found = true;
+				break;
+			}
+			/*
+			 * Else, we are seeking for a hole, check the next file
+			 * extent item.
+			 */
+		}
+
+		start = extent_end;
+		last_extent_end = extent_end;
+		path->slots[0]++;
+		if (fatal_signal_pending(current)) {
+			ret = -EINTR;
+			goto out;
+		}
 		cond_resched();
 	}
-	free_extent_map(em);
-	unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart, lockend,
-			     &cached_state);
-	if (ret) {
-		offset = ret;
-	} else {
-		if (whence == SEEK_DATA && start >= i_size)
-			offset = -ENXIO;
-		else
-			offset = min_t(loff_t, start, i_size);
+
+	/* We have an implicit hole from the last extent found up to i_size. */
+	if (!found && start < i_size) {
+		found = find_desired_extent_in_hole(inode, whence, start,
+						    i_size - 1, &start);
+		if (!found)
+			start = i_size;
 	}
 
-	return offset;
+out:
+	unlock_extent(&inode->io_tree, lockstart, lockend, &cached_state);
+	btrfs_free_path(path);
+
+	if (ret < 0)
+		return ret;
+
+	if (whence == SEEK_DATA && start >= i_size)
+		return -ENXIO;
+
+	return min_t(loff_t, start, i_size);
 }
 
 static loff_t btrfs_file_llseek(struct file *file, loff_t offset, int whence)
@@ -3425,9 +3974,9 @@ static loff_t btrfs_file_llseek(struct file *file, loff_t offset, int whence)
 		return generic_file_llseek(file, offset, whence);
 	case SEEK_DATA:
 	case SEEK_HOLE:
-		inode_lock_shared(inode);
-		offset = find_desired_extent(inode, offset, whence);
-		inode_unlock_shared(inode);
+		btrfs_inode_lock(inode, BTRFS_ILOCK_SHARED);
+		offset = find_desired_extent(BTRFS_I(inode), offset, whence);
+		btrfs_inode_unlock(inode, BTRFS_ILOCK_SHARED);
 		break;
 	}
 
@@ -3439,18 +3988,126 @@ static loff_t btrfs_file_llseek(struct file *file, loff_t offset, int whence)
 
 static int btrfs_file_open(struct inode *inode, struct file *filp)
 {
-	filp->f_mode |= FMODE_NOWAIT;
+	int ret;
+
+	filp->f_mode |= FMODE_NOWAIT | FMODE_BUF_RASYNC | FMODE_BUF_WASYNC;
+
+	ret = fsverity_file_open(inode, filp);
+	if (ret)
+		return ret;
 	return generic_file_open(inode, filp);
 }
 
+static int check_direct_read(struct btrfs_fs_info *fs_info,
+			     const struct iov_iter *iter, loff_t offset)
+{
+	int ret;
+	int i, seg;
+
+	ret = check_direct_IO(fs_info, iter, offset);
+	if (ret < 0)
+		return ret;
+
+	if (!iter_is_iovec(iter))
+		return 0;
+
+	for (seg = 0; seg < iter->nr_segs; seg++)
+		for (i = seg + 1; i < iter->nr_segs; i++)
+			if (iter->iov[seg].iov_base == iter->iov[i].iov_base)
+				return -EINVAL;
+	return 0;
+}
+
+static ssize_t btrfs_direct_read(struct kiocb *iocb, struct iov_iter *to)
+{
+	struct inode *inode = file_inode(iocb->ki_filp);
+	size_t prev_left = 0;
+	ssize_t read = 0;
+	ssize_t ret;
+
+	if (fsverity_active(inode))
+		return 0;
+
+	if (check_direct_read(btrfs_sb(inode->i_sb), to, iocb->ki_pos))
+		return 0;
+
+	btrfs_inode_lock(inode, BTRFS_ILOCK_SHARED);
+again:
+	/*
+	 * This is similar to what we do for direct IO writes, see the comment
+	 * at btrfs_direct_write(), but we also disable page faults in addition
+	 * to disabling them only at the iov_iter level. This is because when
+	 * reading from a hole or prealloc extent, iomap calls iov_iter_zero(),
+	 * which can still trigger page fault ins despite having set ->nofault
+	 * to true of our 'to' iov_iter.
+	 *
+	 * The difference to direct IO writes is that we deadlock when trying
+	 * to lock the extent range in the inode's tree during he page reads
+	 * triggered by the fault in (while for writes it is due to waiting for
+	 * our own ordered extent). This is because for direct IO reads,
+	 * btrfs_dio_iomap_begin() returns with the extent range locked, which
+	 * is only unlocked in the endio callback (end_bio_extent_readpage()).
+	 */
+	pagefault_disable();
+	to->nofault = true;
+	ret = btrfs_dio_read(iocb, to, read);
+	to->nofault = false;
+	pagefault_enable();
+
+	/* No increment (+=) because iomap returns a cumulative value. */
+	if (ret > 0)
+		read = ret;
+
+	if (iov_iter_count(to) > 0 && (ret == -EFAULT || ret > 0)) {
+		const size_t left = iov_iter_count(to);
+
+		if (left == prev_left) {
+			/*
+			 * We didn't make any progress since the last attempt,
+			 * fallback to a buffered read for the remainder of the
+			 * range. This is just to avoid any possibility of looping
+			 * for too long.
+			 */
+			ret = read;
+		} else {
+			/*
+			 * We made some progress since the last retry or this is
+			 * the first time we are retrying. Fault in as many pages
+			 * as possible and retry.
+			 */
+			fault_in_iov_iter_writeable(to, left);
+			prev_left = left;
+			goto again;
+		}
+	}
+	btrfs_inode_unlock(inode, BTRFS_ILOCK_SHARED);
+	return ret < 0 ? ret : read;
+}
+
+static ssize_t btrfs_file_read_iter(struct kiocb *iocb, struct iov_iter *to)
+{
+	ssize_t ret = 0;
+
+	if (iocb->ki_flags & IOCB_DIRECT) {
+		ret = btrfs_direct_read(iocb, to);
+		if (ret < 0 || !iov_iter_count(to) ||
+		    iocb->ki_pos >= i_size_read(file_inode(iocb->ki_filp)))
+			return ret;
+	}
+
+	return filemap_read(iocb, to, ret);
+}
+
 const struct file_operations btrfs_file_operations = {
 	.llseek		= btrfs_file_llseek,
-	.read_iter      = generic_file_read_iter,
+	.read_iter      = btrfs_file_read_iter,
 	.splice_read	= generic_file_splice_read,
 	.write_iter	= btrfs_file_write_iter,
+	.splice_write	= iter_file_splice_write,
 	.mmap		= btrfs_file_mmap,
 	.open		= btrfs_file_open,
 	.release	= btrfs_release_file,
+	.get_unmapped_area = thp_get_unmapped_area,
 	.fsync		= btrfs_sync_file,
 	.fallocate	= btrfs_fallocate,
 	.unlocked_ioctl	= btrfs_ioctl,
diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c
index 0598fd3c6e3f..f4023651dd68 100644
--- a/fs/btrfs/free-space-cache.c
+++ b/fs/btrfs/free-space-cache.c
@@ -11,17 +11,19 @@
 #include <linux/ratelimit.h>
 #include <linux/error-injection.h>
 #include <linux/sched/mm.h>
+#include "misc.h"
 #include "ctree.h"
 #include "free-space-cache.h"
 #include "transaction.h"
 #include "disk-io.h"
 #include "extent_io.h"
-#include "inode-map.h"
 #include "volumes.h"
 #include "space-info.h"
 #include "delalloc-space.h"
 #include "block-group.h"
 #include "discard.h"
+#include "subpage.h"
+#include "inode-item.h"
 
 #define BITS_PER_BITMAP		(PAGE_SIZE * 8UL)
 #define MAX_CACHE_BYTES_PER_GIG	SZ_64K
@@ -33,16 +35,36 @@ struct btrfs_trim_range {
 	struct list_head list;
 };
 
-static int count_bitmap_extents(struct btrfs_free_space_ctl *ctl,
-				struct btrfs_free_space *bitmap_info);
 static int link_free_space(struct btrfs_free_space_ctl *ctl,
 			   struct btrfs_free_space *info);
 static void unlink_free_space(struct btrfs_free_space_ctl *ctl,
-			      struct btrfs_free_space *info);
-static int btrfs_wait_cache_io_root(struct btrfs_root *root,
-			     struct btrfs_trans_handle *trans,
-			     struct btrfs_io_ctl *io_ctl,
-			     struct btrfs_path *path);
+			      struct btrfs_free_space *info, bool update_stat);
+static int search_bitmap(struct btrfs_free_space_ctl *ctl,
+			 struct btrfs_free_space *bitmap_info, u64 *offset,
+			 u64 *bytes, bool for_alloc);
+static void free_bitmap(struct btrfs_free_space_ctl *ctl,
+			struct btrfs_free_space *bitmap_info);
+static void bitmap_clear_bits(struct btrfs_free_space_ctl *ctl,
+			      struct btrfs_free_space *info, u64 offset,
+			      u64 bytes, bool update_stats);
+
+static void __btrfs_remove_free_space_cache(struct btrfs_free_space_ctl *ctl)
+{
+	struct btrfs_free_space *info;
+	struct rb_node *node;
+
+	while ((node = rb_last(&ctl->free_space_offset)) != NULL) {
+		info = rb_entry(node, struct btrfs_free_space, offset_index);
+		if (!info->bitmap) {
+			unlink_free_space(ctl, info, true);
+			kmem_cache_free(btrfs_free_space_cachep, info);
+		} else {
+			free_bitmap(ctl, info);
+		}
+
+		cond_resched_lock(&ctl->tree_lock);
+	}
+}
 
 static struct inode *__lookup_free_space_inode(struct btrfs_root *root,
 					       struct btrfs_path *path,
@@ -82,7 +104,7 @@ static struct inode *__lookup_free_space_inode(struct btrfs_root *root,
 	 * sure NOFS is set to keep us from deadlocking.
 	 */
 	nofs_flag = memalloc_nofs_save();
-	inode = btrfs_iget_path(fs_info->sb, &location, root, path);
+	inode = btrfs_iget_path(fs_info->sb, location.objectid, root, path);
 	btrfs_release_path(path);
 	memalloc_nofs_restore(nofs_flag);
 	if (IS_ERR(inode))
@@ -122,10 +144,8 @@ struct inode *lookup_free_space_inode(struct btrfs_block_group *block_group,
 		block_group->disk_cache_state = BTRFS_DC_CLEAR;
 	}
 
-	if (!block_group->iref) {
+	if (!test_and_set_bit(BLOCK_GROUP_FLAG_IREF, &block_group->runtime_flags))
 		block_group->inode = igrab(inode);
-		block_group->iref = 1;
-	}
 	spin_unlock(&block_group->lock);
 
 	return inode;
@@ -141,17 +161,15 @@ static int __create_free_space_inode(struct btrfs_root *root,
 	struct btrfs_free_space_header *header;
 	struct btrfs_inode_item *inode_item;
 	struct extent_buffer *leaf;
-	u64 flags = BTRFS_INODE_NOCOMPRESS | BTRFS_INODE_PREALLOC;
+	/* We inline CRCs for the free disk space cache */
+	const u64 flags = BTRFS_INODE_NOCOMPRESS | BTRFS_INODE_PREALLOC |
+			  BTRFS_INODE_NODATASUM | BTRFS_INODE_NODATACOW;
 	int ret;
 
 	ret = btrfs_insert_empty_inode(trans, root, path, ino);
 	if (ret)
 		return ret;
 
-	/* We inline crc's for the free disk space cache */
-	if (ino != BTRFS_FREE_INO_OBJECTID)
-		flags |= BTRFS_INODE_NODATASUM | BTRFS_INODE_NODATACOW;
-
 	leaf = path->nodes[0];
 	inode_item = btrfs_item_ptr(leaf, path->slots[0],
 				    struct btrfs_inode_item);
@@ -199,7 +217,7 @@ int create_free_space_inode(struct btrfs_trans_handle *trans,
 	int ret;
 	u64 ino;
 
-	ret = btrfs_find_free_objectid(trans->fs_info->tree_root, &ino);
+	ret = btrfs_get_free_objectid(trans->fs_info->tree_root, &ino);
 	if (ret < 0)
 		return ret;
 
@@ -207,6 +225,64 @@ int create_free_space_inode(struct btrfs_trans_handle *trans,
 					 ino, block_group->start);
 }
 
+/*
+ * inode is an optional sink: if it is NULL, btrfs_remove_free_space_inode
+ * handles lookup, otherwise it takes ownership and iputs the inode.
+ * Don't reuse an inode pointer after passing it into this function.
+ */
+int btrfs_remove_free_space_inode(struct btrfs_trans_handle *trans,
+				  struct inode *inode,
+				  struct btrfs_block_group *block_group)
+{
+	struct btrfs_path *path;
+	struct btrfs_key key;
+	int ret = 0;
+
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+
+	if (!inode)
+		inode = lookup_free_space_inode(block_group, path);
+	if (IS_ERR(inode)) {
+		if (PTR_ERR(inode) != -ENOENT)
+			ret = PTR_ERR(inode);
+		goto out;
+	}
+	ret = btrfs_orphan_add(trans, BTRFS_I(inode));
+	if (ret) {
+		btrfs_add_delayed_iput(inode);
+		goto out;
+	}
+	clear_nlink(inode);
+	/* One for the block groups ref */
+	spin_lock(&block_group->lock);
+	if (test_and_clear_bit(BLOCK_GROUP_FLAG_IREF, &block_group->runtime_flags)) {
+		block_group->inode = NULL;
+		spin_unlock(&block_group->lock);
+		iput(inode);
+	} else {
+		spin_unlock(&block_group->lock);
+	}
+	/* One for the lookup ref */
+	btrfs_add_delayed_iput(inode);
+
+	key.objectid = BTRFS_FREE_SPACE_OBJECTID;
+	key.type = 0;
+	key.offset = block_group->start;
+	ret = btrfs_search_slot(trans, trans->fs_info->tree_root, &key, path,
+				-1, 1);
+	if (ret) {
+		if (ret > 0)
+			ret = 0;
+		goto out;
+	}
+	ret = btrfs_del_item(trans, trans->fs_info->tree_root, path);
+out:
+	btrfs_free_path(path);
+	return ret;
+}
+
 int btrfs_check_trunc_cache_free_space(struct btrfs_fs_info *fs_info,
 				       struct btrfs_block_rsv *rsv)
 {
@@ -228,9 +304,18 @@ int btrfs_check_trunc_cache_free_space(struct btrfs_fs_info *fs_info,
 
 int btrfs_truncate_free_space_cache(struct btrfs_trans_handle *trans,
 				    struct btrfs_block_group *block_group,
-				    struct inode *inode)
+				    struct inode *vfs_inode)
 {
-	struct btrfs_root *root = BTRFS_I(inode)->root;
+	struct btrfs_truncate_control control = {
+		.inode = BTRFS_I(vfs_inode),
+		.new_size = 0,
+		.ino = btrfs_ino(BTRFS_I(vfs_inode)),
+		.min_type = BTRFS_EXTENT_DATA_KEY,
+		.clear_extent_range = true,
+	};
+	struct btrfs_inode *inode = BTRFS_I(vfs_inode);
+	struct btrfs_root *root = inode->root;
+	struct extent_state *cached_state = NULL;
 	int ret = 0;
 	bool locked = false;
 
@@ -260,15 +345,22 @@ int btrfs_truncate_free_space_cache(struct btrfs_trans_handle *trans,
 		btrfs_free_path(path);
 	}
 
-	btrfs_i_size_write(BTRFS_I(inode), 0);
-	truncate_pagecache(inode, 0);
+	btrfs_i_size_write(inode, 0);
+	truncate_pagecache(vfs_inode, 0);
+
+	lock_extent(&inode->io_tree, 0, (u64)-1, &cached_state);
+	btrfs_drop_extent_map_range(inode, 0, (u64)-1, false);
 
 	/*
 	 * We skip the throttling logic for free space cache inodes, so we don't
 	 * need to check for -EAGAIN.
 	 */
-	ret = btrfs_truncate_inode_items(trans, root, inode,
-					 0, BTRFS_EXTENT_DATA_KEY);
+	ret = btrfs_truncate_inode_items(trans, root, &control);
+
+	inode_sub_bytes(&inode->vfs_inode, control.sub_bytes);
+	btrfs_inode_safe_disk_i_size_write(inode, control.last_size);
+
+	unlock_extent(&inode->io_tree, 0, (u64)-1, &cached_state);
 	if (ret)
 		goto fail;
 
@@ -285,35 +377,24 @@ fail:
 
 static void readahead_cache(struct inode *inode)
 {
-	struct file_ra_state *ra;
+	struct file_ra_state ra;
 	unsigned long last_index;
 
-	ra = kzalloc(sizeof(*ra), GFP_NOFS);
-	if (!ra)
-		return;
-
-	file_ra_state_init(ra, inode->i_mapping);
+	file_ra_state_init(&ra, inode->i_mapping);
 	last_index = (i_size_read(inode) - 1) >> PAGE_SHIFT;
 
-	page_cache_sync_readahead(inode->i_mapping, ra, NULL, 0, last_index);
-
-	kfree(ra);
+	page_cache_sync_readahead(inode->i_mapping, &ra, NULL, 0, last_index);
 }
 
 static int io_ctl_init(struct btrfs_io_ctl *io_ctl, struct inode *inode,
 		       int write)
 {
 	int num_pages;
-	int check_crcs = 0;
 
 	num_pages = DIV_ROUND_UP(i_size_read(inode), PAGE_SIZE);
 
-	if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FREE_INO_OBJECTID)
-		check_crcs = 1;
-
 	/* Make sure we can fit our crcs and generation into the first page */
-	if (write && check_crcs &&
-	    (num_pages * sizeof(u32) + sizeof(u64)) > PAGE_SIZE)
+	if (write && (num_pages * sizeof(u32) + sizeof(u64)) > PAGE_SIZE)
 		return -ENOSPC;
 
 	memset(io_ctl, 0, sizeof(struct btrfs_io_ctl));
@@ -324,7 +405,6 @@ static int io_ctl_init(struct btrfs_io_ctl *io_ctl, struct inode *inode,
 
 	io_ctl->num_pages = num_pages;
 	io_ctl->fs_info = btrfs_sb(inode->i_sb);
-	io_ctl->check_crcs = check_crcs;
 	io_ctl->inode = inode;
 
 	return 0;
@@ -364,29 +444,43 @@ static void io_ctl_drop_pages(struct btrfs_io_ctl *io_ctl)
 
 	for (i = 0; i < io_ctl->num_pages; i++) {
 		if (io_ctl->pages[i]) {
-			ClearPageChecked(io_ctl->pages[i]);
+			btrfs_page_clear_checked(io_ctl->fs_info,
+					io_ctl->pages[i],
+					page_offset(io_ctl->pages[i]),
+					PAGE_SIZE);
 			unlock_page(io_ctl->pages[i]);
 			put_page(io_ctl->pages[i]);
 		}
 	}
 }
 
-static int io_ctl_prepare_pages(struct btrfs_io_ctl *io_ctl, struct inode *inode,
-				int uptodate)
+static int io_ctl_prepare_pages(struct btrfs_io_ctl *io_ctl, bool uptodate)
 {
 	struct page *page;
+	struct inode *inode = io_ctl->inode;
 	gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping);
 	int i;
 
 	for (i = 0; i < io_ctl->num_pages; i++) {
+		int ret;
+
 		page = find_or_create_page(inode->i_mapping, i, mask);
 		if (!page) {
 			io_ctl_drop_pages(io_ctl);
 			return -ENOMEM;
 		}
+
+		ret = set_page_extent_mapped(page);
+		if (ret < 0) {
+			unlock_page(page);
+			put_page(page);
+			io_ctl_drop_pages(io_ctl);
+			return ret;
+		}
+
 		io_ctl->pages[i] = page;
 		if (uptodate && !PageUptodate(page)) {
-			btrfs_readpage(NULL, page);
+			btrfs_read_folio(NULL, page_folio(page));
 			lock_page(page);
 			if (page->mapping != inode->i_mapping) {
 				btrfs_err(BTRFS_I(inode)->root->fs_info,
@@ -403,59 +497,43 @@ static int io_ctl_prepare_pages(struct btrfs_io_ctl *io_ctl, struct inode *inode
 		}
 	}
 
-	for (i = 0; i < io_ctl->num_pages; i++) {
+	for (i = 0; i < io_ctl->num_pages; i++)
 		clear_page_dirty_for_io(io_ctl->pages[i]);
-		set_page_extent_mapped(io_ctl->pages[i]);
-	}
 
 	return 0;
 }
 
 static void io_ctl_set_generation(struct btrfs_io_ctl *io_ctl, u64 generation)
 {
-	__le64 *val;
-
 	io_ctl_map_page(io_ctl, 1);
 
 	/*
 	 * Skip the csum areas.  If we don't check crcs then we just have a
 	 * 64bit chunk at the front of the first page.
 	 */
-	if (io_ctl->check_crcs) {
-		io_ctl->cur += (sizeof(u32) * io_ctl->num_pages);
-		io_ctl->size -= sizeof(u64) + (sizeof(u32) * io_ctl->num_pages);
-	} else {
-		io_ctl->cur += sizeof(u64);
-		io_ctl->size -= sizeof(u64) * 2;
-	}
+	io_ctl->cur += (sizeof(u32) * io_ctl->num_pages);
+	io_ctl->size -= sizeof(u64) + (sizeof(u32) * io_ctl->num_pages);
 
-	val = io_ctl->cur;
-	*val = cpu_to_le64(generation);
+	put_unaligned_le64(generation, io_ctl->cur);
 	io_ctl->cur += sizeof(u64);
 }
 
 static int io_ctl_check_generation(struct btrfs_io_ctl *io_ctl, u64 generation)
 {
-	__le64 *gen;
+	u64 cache_gen;
 
 	/*
 	 * Skip the crc area.  If we don't check crcs then we just have a 64bit
 	 * chunk at the front of the first page.
 	 */
-	if (io_ctl->check_crcs) {
-		io_ctl->cur += sizeof(u32) * io_ctl->num_pages;
-		io_ctl->size -= sizeof(u64) +
-			(sizeof(u32) * io_ctl->num_pages);
-	} else {
-		io_ctl->cur += sizeof(u64);
-		io_ctl->size -= sizeof(u64) * 2;
-	}
+	io_ctl->cur += sizeof(u32) * io_ctl->num_pages;
+	io_ctl->size -= sizeof(u64) + (sizeof(u32) * io_ctl->num_pages);
 
-	gen = io_ctl->cur;
-	if (le64_to_cpu(*gen) != generation) {
+	cache_gen = get_unaligned_le64(io_ctl->cur);
+	if (cache_gen != generation) {
 		btrfs_err_rl(io_ctl->fs_info,
 			"space cache generation (%llu) does not match inode (%llu)",
-				*gen, generation);
+				cache_gen, generation);
 		io_ctl_unmap_page(io_ctl);
 		return -EIO;
 	}
@@ -469,11 +547,6 @@ static void io_ctl_set_crc(struct btrfs_io_ctl *io_ctl, int index)
 	u32 crc = ~(u32)0;
 	unsigned offset = 0;
 
-	if (!io_ctl->check_crcs) {
-		io_ctl_unmap_page(io_ctl);
-		return;
-	}
-
 	if (index == 0)
 		offset = sizeof(u32) * io_ctl->num_pages;
 
@@ -491,11 +564,6 @@ static int io_ctl_check_crc(struct btrfs_io_ctl *io_ctl, int index)
 	u32 crc = ~(u32)0;
 	unsigned offset = 0;
 
-	if (!io_ctl->check_crcs) {
-		io_ctl_map_page(io_ctl, 0);
-		return 0;
-	}
-
 	if (index == 0)
 		offset = sizeof(u32) * io_ctl->num_pages;
 
@@ -525,8 +593,8 @@ static int io_ctl_add_entry(struct btrfs_io_ctl *io_ctl, u64 offset, u64 bytes,
 		return -ENOSPC;
 
 	entry = io_ctl->cur;
-	entry->offset = cpu_to_le64(offset);
-	entry->bytes = cpu_to_le64(bytes);
+	put_unaligned_le64(offset, &entry->offset);
+	put_unaligned_le64(bytes, &entry->bytes);
 	entry->type = (bitmap) ? BTRFS_FREE_SPACE_BITMAP :
 		BTRFS_FREE_SPACE_EXTENT;
 	io_ctl->cur += sizeof(struct btrfs_free_space_entry);
@@ -599,8 +667,8 @@ static int io_ctl_read_entry(struct btrfs_io_ctl *io_ctl,
 	}
 
 	e = io_ctl->cur;
-	entry->offset = le64_to_cpu(e->offset);
-	entry->bytes = le64_to_cpu(e->bytes);
+	entry->offset = get_unaligned_le64(&e->offset);
+	entry->bytes = get_unaligned_le64(&e->bytes);
 	*type = e->type;
 	io_ctl->cur += sizeof(struct btrfs_free_space_entry);
 	io_ctl->size -= sizeof(struct btrfs_free_space_entry);
@@ -628,42 +696,48 @@ static int io_ctl_read_bitmap(struct btrfs_io_ctl *io_ctl,
 	return 0;
 }
 
-/*
- * Since we attach pinned extents after the fact we can have contiguous sections
- * of free space that are split up in entries.  This poses a problem with the
- * tree logging stuff since it could have allocated across what appears to be 2
- * entries since we would have merged the entries when adding the pinned extents
- * back to the free space cache.  So run through the space cache that we just
- * loaded and merge contiguous entries.  This will make the log replay stuff not
- * blow up and it will make for nicer allocator behavior.
- */
-static void merge_space_tree(struct btrfs_free_space_ctl *ctl)
+static void recalculate_thresholds(struct btrfs_free_space_ctl *ctl)
 {
-	struct btrfs_free_space *e, *prev = NULL;
-	struct rb_node *n;
+	struct btrfs_block_group *block_group = ctl->block_group;
+	u64 max_bytes;
+	u64 bitmap_bytes;
+	u64 extent_bytes;
+	u64 size = block_group->length;
+	u64 bytes_per_bg = BITS_PER_BITMAP * ctl->unit;
+	u64 max_bitmaps = div64_u64(size + bytes_per_bg - 1, bytes_per_bg);
 
-again:
-	spin_lock(&ctl->tree_lock);
-	for (n = rb_first(&ctl->free_space_offset); n; n = rb_next(n)) {
-		e = rb_entry(n, struct btrfs_free_space, offset_index);
-		if (!prev)
-			goto next;
-		if (e->bitmap || prev->bitmap)
-			goto next;
-		if (prev->offset + prev->bytes == e->offset) {
-			unlink_free_space(ctl, prev);
-			unlink_free_space(ctl, e);
-			prev->bytes += e->bytes;
-			kmem_cache_free(btrfs_free_space_cachep, e);
-			link_free_space(ctl, prev);
-			prev = NULL;
-			spin_unlock(&ctl->tree_lock);
-			goto again;
-		}
-next:
-		prev = e;
-	}
-	spin_unlock(&ctl->tree_lock);
+	max_bitmaps = max_t(u64, max_bitmaps, 1);
+
+	if (ctl->total_bitmaps > max_bitmaps)
+		btrfs_err(block_group->fs_info,
+"invalid free space control: bg start=%llu len=%llu total_bitmaps=%u unit=%u max_bitmaps=%llu bytes_per_bg=%llu",
+			  block_group->start, block_group->length,
+			  ctl->total_bitmaps, ctl->unit, max_bitmaps,
+			  bytes_per_bg);
+	ASSERT(ctl->total_bitmaps <= max_bitmaps);
+
+	/*
+	 * We are trying to keep the total amount of memory used per 1GiB of
+	 * space to be MAX_CACHE_BYTES_PER_GIG.  However, with a reclamation
+	 * mechanism of pulling extents >= FORCE_EXTENT_THRESHOLD out of
+	 * bitmaps, we may end up using more memory than this.
+	 */
+	if (size < SZ_1G)
+		max_bytes = MAX_CACHE_BYTES_PER_GIG;
+	else
+		max_bytes = MAX_CACHE_BYTES_PER_GIG * div_u64(size, SZ_1G);
+
+	bitmap_bytes = ctl->total_bitmaps * ctl->unit;
+
+	/*
+	 * we want the extent entry threshold to always be at most 1/2 the max
+	 * bytes we can have, or whatever is less than that.
+	 */
+	extent_bytes = max_bytes - bitmap_bytes;
+	extent_bytes = min_t(u64, extent_bytes, max_bytes >> 1);
+
+	ctl->extents_thresh =
+		div_u64(extent_bytes, sizeof(struct btrfs_free_space));
 }
 
 static int __load_free_space_cache(struct btrfs_root *root, struct inode *inode,
@@ -732,7 +806,7 @@ static int __load_free_space_cache(struct btrfs_root *root, struct inode *inode,
 
 	readahead_cache(inode);
 
-	ret = io_ctl_prepare_pages(&io_ctl, inode, 1);
+	ret = io_ctl_prepare_pages(&io_ctl, true);
 	if (ret)
 		goto out;
 
@@ -747,8 +821,10 @@ static int __load_free_space_cache(struct btrfs_root *root, struct inode *inode,
 	while (num_entries) {
 		e = kmem_cache_zalloc(btrfs_free_space_cachep,
 				      GFP_NOFS);
-		if (!e)
+		if (!e) {
+			ret = -ENOMEM;
 			goto free_cache;
+		}
 
 		ret = io_ctl_read_entry(&io_ctl, e, &type);
 		if (ret) {
@@ -756,17 +832,8 @@ static int __load_free_space_cache(struct btrfs_root *root, struct inode *inode,
 			goto free_cache;
 		}
 
-		/*
-		 * Sync discard ensures that the free space cache is always
-		 * trimmed.  So when reading this in, the state should reflect
-		 * that.  We also do this for async as a stop gap for lack of
-		 * persistence.
-		 */
-		if (btrfs_test_opt(fs_info, DISCARD_SYNC) ||
-		    btrfs_test_opt(fs_info, DISCARD_ASYNC))
-			e->trim_state = BTRFS_TRIM_STATE_TRIMMED;
-
 		if (!e->bytes) {
+			ret = -1;
 			kmem_cache_free(btrfs_free_space_cachep, e);
 			goto free_cache;
 		}
@@ -787,6 +854,7 @@ static int __load_free_space_cache(struct btrfs_root *root, struct inode *inode,
 			e->bitmap = kmem_cache_zalloc(
 					btrfs_free_space_bitmap_cachep, GFP_NOFS);
 			if (!e->bitmap) {
+				ret = -ENOMEM;
 				kmem_cache_free(
 					btrfs_free_space_cachep, e);
 				goto free_cache;
@@ -794,7 +862,7 @@ static int __load_free_space_cache(struct btrfs_root *root, struct inode *inode,
 			spin_lock(&ctl->tree_lock);
 			ret = link_free_space(ctl, e);
 			ctl->total_bitmaps++;
-			ctl->op->recalc_thresholds(ctl);
+			recalculate_thresholds(ctl);
 			spin_unlock(&ctl->tree_lock);
 			if (ret) {
 				btrfs_err(fs_info,
@@ -819,31 +887,64 @@ static int __load_free_space_cache(struct btrfs_root *root, struct inode *inode,
 		ret = io_ctl_read_bitmap(&io_ctl, e);
 		if (ret)
 			goto free_cache;
-		e->bitmap_extents = count_bitmap_extents(ctl, e);
-		if (!btrfs_free_space_trimmed(e)) {
-			ctl->discardable_extents[BTRFS_STAT_CURR] +=
-				e->bitmap_extents;
-			ctl->discardable_bytes[BTRFS_STAT_CURR] += e->bytes;
-		}
 	}
 
 	io_ctl_drop_pages(&io_ctl);
-	merge_space_tree(ctl);
 	ret = 1;
 out:
-	btrfs_discard_update_discardable(ctl->private, ctl);
 	io_ctl_free(&io_ctl);
 	return ret;
 free_cache:
 	io_ctl_drop_pages(&io_ctl);
+
+	spin_lock(&ctl->tree_lock);
 	__btrfs_remove_free_space_cache(ctl);
+	spin_unlock(&ctl->tree_lock);
 	goto out;
 }
 
+static int copy_free_space_cache(struct btrfs_block_group *block_group,
+				 struct btrfs_free_space_ctl *ctl)
+{
+	struct btrfs_free_space *info;
+	struct rb_node *n;
+	int ret = 0;
+
+	while (!ret && (n = rb_first(&ctl->free_space_offset)) != NULL) {
+		info = rb_entry(n, struct btrfs_free_space, offset_index);
+		if (!info->bitmap) {
+			unlink_free_space(ctl, info, true);
+			ret = btrfs_add_free_space(block_group, info->offset,
+						   info->bytes);
+			kmem_cache_free(btrfs_free_space_cachep, info);
+		} else {
+			u64 offset = info->offset;
+			u64 bytes = ctl->unit;
+
+			while (search_bitmap(ctl, info, &offset, &bytes,
+					     false) == 0) {
+				ret = btrfs_add_free_space(block_group, offset,
+							   bytes);
+				if (ret)
+					break;
+				bitmap_clear_bits(ctl, info, offset, bytes, true);
+				offset = info->offset;
+				bytes = ctl->unit;
+			}
+			free_bitmap(ctl, info);
+		}
+		cond_resched();
+	}
+	return ret;
+}
+
+static struct lock_class_key btrfs_free_space_inode_key;
+
 int load_free_space_cache(struct btrfs_block_group *block_group)
 {
 	struct btrfs_fs_info *fs_info = block_group->fs_info;
 	struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
+	struct btrfs_free_space_ctl tmp_ctl = {};
 	struct inode *inode;
 	struct btrfs_path *path;
 	int ret = 0;
@@ -851,6 +952,13 @@ int load_free_space_cache(struct btrfs_block_group *block_group)
 	u64 used = block_group->used;
 
 	/*
+	 * Because we could potentially discard our loaded free space, we want
+	 * to load everything into a temporary structure first, and then if it's
+	 * valid copy it all into the actual free space ctl.
+	 */
+	btrfs_init_free_space_ctl(block_group, &tmp_ctl);
+
+	/*
 	 * If this block group has been marked to be cleared for one reason or
 	 * another then we can't trust the on disk cache, so just return.
 	 */
@@ -901,19 +1009,39 @@ int load_free_space_cache(struct btrfs_block_group *block_group)
 	}
 	spin_unlock(&block_group->lock);
 
-	ret = __load_free_space_cache(fs_info->tree_root, inode, ctl,
+	/*
+	 * Reinitialize the class of struct inode's mapping->invalidate_lock for
+	 * free space inodes to prevent false positives related to locks for normal
+	 * inodes.
+	 */
+	lockdep_set_class(&(&inode->i_data)->invalidate_lock,
+			  &btrfs_free_space_inode_key);
+
+	ret = __load_free_space_cache(fs_info->tree_root, inode, &tmp_ctl,
 				      path, block_group->start);
 	btrfs_free_path(path);
 	if (ret <= 0)
 		goto out;
 
-	spin_lock(&ctl->tree_lock);
-	matched = (ctl->free_space == (block_group->length - used -
-				       block_group->bytes_super));
-	spin_unlock(&ctl->tree_lock);
+	matched = (tmp_ctl.free_space == (block_group->length - used -
+					  block_group->bytes_super));
 
-	if (!matched) {
-		__btrfs_remove_free_space_cache(ctl);
+	if (matched) {
+		ret = copy_free_space_cache(block_group, &tmp_ctl);
+		/*
+		 * ret == 1 means we successfully loaded the free space cache,
+		 * so we need to re-set it here.
+		 */
+		if (ret == 0)
+			ret = 1;
+	} else {
+		/*
+		 * We need to call the _locked variant so we don't try to update
+		 * the discard counters.
+		 */
+		spin_lock(&tmp_ctl.tree_lock);
+		__btrfs_remove_free_space_cache(&tmp_ctl);
+		spin_unlock(&tmp_ctl.tree_lock);
 		btrfs_warn(fs_info,
 			   "block group %llu has wrong amount of free space",
 			   block_group->start);
@@ -932,6 +1060,9 @@ out:
 			   block_group->start);
 	}
 
+	spin_lock(&ctl->tree_lock);
+	btrfs_discard_update_discardable(block_group);
+	spin_unlock(&ctl->tree_lock);
 	iput(inode);
 	return ret;
 }
@@ -1032,7 +1163,7 @@ update_cache_item(struct btrfs_trans_handle *trans,
 	ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
 	if (ret < 0) {
 		clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, inode->i_size - 1,
-				 EXTENT_DELALLOC, 0, 0, NULL);
+				 EXTENT_DELALLOC, NULL);
 		goto fail;
 	}
 	leaf = path->nodes[0];
@@ -1044,8 +1175,8 @@ update_cache_item(struct btrfs_trans_handle *trans,
 		if (found_key.objectid != BTRFS_FREE_SPACE_OBJECTID ||
 		    found_key.offset != offset) {
 			clear_extent_bit(&BTRFS_I(inode)->io_tree, 0,
-					 inode->i_size - 1, EXTENT_DELALLOC, 0,
-					 0, NULL);
+					 inode->i_size - 1, EXTENT_DELALLOC,
+					 NULL);
 			btrfs_release_path(path);
 			goto fail;
 		}
@@ -1067,6 +1198,7 @@ fail:
 }
 
 static noinline_for_stack int write_pinned_extent_entries(
+			    struct btrfs_trans_handle *trans,
 			    struct btrfs_block_group *block_group,
 			    struct btrfs_io_ctl *io_ctl,
 			    int *entries)
@@ -1085,7 +1217,7 @@ static noinline_for_stack int write_pinned_extent_entries(
 	 * We shouldn't have switched the pinned extents yet so this is the
 	 * right one
 	 */
-	unpin = block_group->fs_info->pinned_extents;
+	unpin = &trans->transaction->pinned_extents;
 
 	start = block_group->start;
 
@@ -1140,7 +1272,7 @@ static int flush_dirty_cache(struct inode *inode)
 	ret = btrfs_wait_ordered_range(inode, 0, (u64)-1);
 	if (ret)
 		clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, inode->i_size - 1,
-				 EXTENT_DELALLOC, 0, 0, NULL);
+				 EXTENT_DELALLOC, NULL);
 
 	return ret;
 }
@@ -1160,8 +1292,8 @@ cleanup_write_cache_enospc(struct inode *inode,
 			   struct extent_state **cached_state)
 {
 	io_ctl_drop_pages(io_ctl);
-	unlock_extent_cached(&BTRFS_I(inode)->io_tree, 0,
-			     i_size_read(inode) - 1, cached_state);
+	unlock_extent(&BTRFS_I(inode)->io_tree, 0, i_size_read(inode) - 1,
+		      cached_state);
 }
 
 static int __btrfs_wait_cache_io(struct btrfs_root *root,
@@ -1185,19 +1317,15 @@ static int __btrfs_wait_cache_io(struct btrfs_root *root,
 	ret = update_cache_item(trans, root, inode, path, offset,
 				io_ctl->entries, io_ctl->bitmaps);
 out:
-	io_ctl_free(io_ctl);
 	if (ret) {
 		invalidate_inode_pages2(inode->i_mapping);
 		BTRFS_I(inode)->generation = 0;
-		if (block_group) {
-#ifdef DEBUG
-			btrfs_err(root->fs_info,
-				  "failed to write free space cache for block group %llu",
-				  block_group->start);
-#endif
-		}
+		if (block_group)
+			btrfs_debug(root->fs_info,
+	  "failed to write free space cache for block group %llu error %d",
+				  block_group->start, ret);
 	}
-	btrfs_update_inode(trans, root, inode);
+	btrfs_update_inode(trans, root, BTRFS_I(inode));
 
 	if (block_group) {
 		/* the dirty list is protected by the dirty_bgs_lock */
@@ -1226,14 +1354,6 @@ out:
 
 }
 
-static int btrfs_wait_cache_io_root(struct btrfs_root *root,
-				    struct btrfs_trans_handle *trans,
-				    struct btrfs_io_ctl *io_ctl,
-				    struct btrfs_path *path)
-{
-	return __btrfs_wait_cache_io(root, trans, NULL, io_ctl, path, 0);
-}
-
 int btrfs_wait_cache_io(struct btrfs_trans_handle *trans,
 			struct btrfs_block_group *block_group,
 			struct btrfs_path *path)
@@ -1244,11 +1364,14 @@ int btrfs_wait_cache_io(struct btrfs_trans_handle *trans,
 }
 
 /**
- * __btrfs_write_out_cache - write out cached info to an inode
- * @root - the root the inode belongs to
- * @ctl - the free space cache we are going to write out
- * @block_group - the block_group for this cache if it belongs to a block_group
- * @trans - the trans handle
+ * Write out cached info to an inode
+ *
+ * @root:        root the inode belongs to
+ * @inode:       freespace inode we are writing out
+ * @ctl:         free space cache we are going to write out
+ * @block_group: block_group for this cache if it belongs to a block_group
+ * @io_ctl:      holds context for the io
+ * @trans:       the trans handle
  *
  * This function writes out a free space cache struct to disk for quick recovery
  * on mount.  This will return 0 if it was successful in writing the cache out,
@@ -1291,12 +1414,12 @@ static int __btrfs_write_out_cache(struct btrfs_root *root, struct inode *inode,
 	}
 
 	/* Lock all pages first so we can lock the extent safely. */
-	ret = io_ctl_prepare_pages(io_ctl, inode, 0);
+	ret = io_ctl_prepare_pages(io_ctl, false);
 	if (ret)
 		goto out_unlock;
 
-	lock_extent_bits(&BTRFS_I(inode)->io_tree, 0, i_size_read(inode) - 1,
-			 &cached_state);
+	lock_extent(&BTRFS_I(inode)->io_tree, 0, i_size_read(inode) - 1,
+		    &cached_state);
 
 	io_ctl_set_generation(io_ctl, trans->transid);
 
@@ -1317,7 +1440,7 @@ static int __btrfs_write_out_cache(struct btrfs_root *root, struct inode *inode,
 	 * If this changes while we are working we'll get added back to
 	 * the dirty list and redo it.  No locking needed
 	 */
-	ret = write_pinned_extent_entries(block_group, io_ctl, &entries);
+	ret = write_pinned_extent_entries(trans, block_group, io_ctl, &entries);
 	if (ret)
 		goto out_nospc_locked;
 
@@ -1336,8 +1459,9 @@ static int __btrfs_write_out_cache(struct btrfs_root *root, struct inode *inode,
 	io_ctl_zero_remaining_pages(io_ctl);
 
 	/* Everything is written out, now we dirty the pages in the file. */
-	ret = btrfs_dirty_pages(inode, io_ctl->pages, io_ctl->num_pages, 0,
-				i_size_read(inode), &cached_state);
+	ret = btrfs_dirty_pages(BTRFS_I(inode), io_ctl->pages,
+				io_ctl->num_pages, 0, i_size_read(inode),
+				&cached_state, false);
 	if (ret)
 		goto out_nospc;
 
@@ -1348,13 +1472,14 @@ static int __btrfs_write_out_cache(struct btrfs_root *root, struct inode *inode,
 	 * them out later
 	 */
 	io_ctl_drop_pages(io_ctl);
+	io_ctl_free(io_ctl);
 
-	unlock_extent_cached(&BTRFS_I(inode)->io_tree, 0,
-			     i_size_read(inode) - 1, &cached_state);
+	unlock_extent(&BTRFS_I(inode)->io_tree, 0, i_size_read(inode) - 1,
+		      &cached_state);
 
 	/*
 	 * at this point the pages are under IO and we're happy,
-	 * The caller is responsible for waiting on them and updating the
+	 * The caller is responsible for waiting on them and updating
 	 * the cache and the inode
 	 */
 	io_ctl->entries = entries;
@@ -1366,18 +1491,6 @@ static int __btrfs_write_out_cache(struct btrfs_root *root, struct inode *inode,
 
 	return 0;
 
-out:
-	io_ctl->inode = NULL;
-	io_ctl_free(io_ctl);
-	if (ret) {
-		invalidate_inode_pages2(inode->i_mapping);
-		BTRFS_I(inode)->generation = 0;
-	}
-	btrfs_update_inode(trans, root, inode);
-	if (must_iput)
-		iput(inode);
-	return ret;
-
 out_nospc_locked:
 	cleanup_bitmap_list(&bitmap_list);
 	spin_unlock(&ctl->tree_lock);
@@ -1390,7 +1503,17 @@ out_unlock:
 	if (block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA))
 		up_write(&block_group->data_rwsem);
 
-	goto out;
+out:
+	io_ctl->inode = NULL;
+	io_ctl_free(io_ctl);
+	if (ret) {
+		invalidate_inode_pages2(inode->i_mapping);
+		BTRFS_I(inode)->generation = 0;
+	}
+	btrfs_update_inode(trans, root, BTRFS_I(inode));
+	if (must_iput)
+		iput(inode);
+	return ret;
 }
 
 int btrfs_write_out_cache(struct btrfs_trans_handle *trans,
@@ -1416,11 +1539,9 @@ int btrfs_write_out_cache(struct btrfs_trans_handle *trans,
 	ret = __btrfs_write_out_cache(fs_info->tree_root, inode, ctl,
 				block_group, &block_group->io_ctl, trans);
 	if (ret) {
-#ifdef DEBUG
-		btrfs_err(fs_info,
-			  "failed to write free space cache for block group %llu",
-			  block_group->start);
-#endif
+		btrfs_debug(fs_info,
+	  "failed to write free space cache for block group %llu error %d",
+			  block_group->start, ret);
 		spin_lock(&block_group->lock);
 		block_group->disk_cache_state = BTRFS_DC_ERROR;
 		spin_unlock(&block_group->lock);
@@ -1517,6 +1638,50 @@ static int tree_insert_offset(struct rb_root *root, u64 offset,
 }
 
 /*
+ * This is a little subtle.  We *only* have ->max_extent_size set if we actually
+ * searched through the bitmap and figured out the largest ->max_extent_size,
+ * otherwise it's 0.  In the case that it's 0 we don't want to tell the
+ * allocator the wrong thing, we want to use the actual real max_extent_size
+ * we've found already if it's larger, or we want to use ->bytes.
+ *
+ * This matters because find_free_space() will skip entries who's ->bytes is
+ * less than the required bytes.  So if we didn't search down this bitmap, we
+ * may pick some previous entry that has a smaller ->max_extent_size than we
+ * have.  For example, assume we have two entries, one that has
+ * ->max_extent_size set to 4K and ->bytes set to 1M.  A second entry hasn't set
+ * ->max_extent_size yet, has ->bytes set to 8K and it's contiguous.  We will
+ *  call into find_free_space(), and return with max_extent_size == 4K, because
+ *  that first bitmap entry had ->max_extent_size set, but the second one did
+ *  not.  If instead we returned 8K we'd come in searching for 8K, and find the
+ *  8K contiguous range.
+ *
+ *  Consider the other case, we have 2 8K chunks in that second entry and still
+ *  don't have ->max_extent_size set.  We'll return 16K, and the next time the
+ *  allocator comes in it'll fully search our second bitmap, and this time it'll
+ *  get an uptodate value of 8K as the maximum chunk size.  Then we'll get the
+ *  right allocation the next loop through.
+ */
+static inline u64 get_max_extent_size(const struct btrfs_free_space *entry)
+{
+	if (entry->bitmap && entry->max_extent_size)
+		return entry->max_extent_size;
+	return entry->bytes;
+}
+
+/*
+ * We want the largest entry to be leftmost, so this is inverted from what you'd
+ * normally expect.
+ */
+static bool entry_less(struct rb_node *node, const struct rb_node *parent)
+{
+	const struct btrfs_free_space *entry, *exist;
+
+	entry = rb_entry(node, struct btrfs_free_space, bytes_index);
+	exist = rb_entry(parent, struct btrfs_free_space, bytes_index);
+	return get_max_extent_size(exist) < get_max_extent_size(entry);
+}
+
+/*
  * searches the tree for the given offset.
  *
  * fuzzy - If this is set, then we are trying to make an allocation, and we just
@@ -1528,15 +1693,10 @@ tree_search_offset(struct btrfs_free_space_ctl *ctl,
 		   u64 offset, int bitmap_only, int fuzzy)
 {
 	struct rb_node *n = ctl->free_space_offset.rb_node;
-	struct btrfs_free_space *entry, *prev = NULL;
+	struct btrfs_free_space *entry = NULL, *prev = NULL;
 
 	/* find entry that is closest to the 'offset' */
-	while (1) {
-		if (!n) {
-			entry = NULL;
-			break;
-		}
-
+	while (n) {
 		entry = rb_entry(n, struct btrfs_free_space, offset_index);
 		prev = entry;
 
@@ -1546,6 +1706,8 @@ tree_search_offset(struct btrfs_free_space_ctl *ctl,
 			n = n->rb_right;
 		else
 			break;
+
+		entry = NULL;
 	}
 
 	if (bitmap_only) {
@@ -1622,6 +1784,10 @@ tree_search_offset(struct btrfs_free_space_ctl *ctl,
 		return NULL;
 
 	while (1) {
+		n = rb_next(&entry->offset_index);
+		if (!n)
+			return NULL;
+		entry = rb_entry(n, struct btrfs_free_space, offset_index);
 		if (entry->bitmap) {
 			if (entry->offset + BITS_PER_BITMAP *
 			    ctl->unit > offset)
@@ -1630,33 +1796,25 @@ tree_search_offset(struct btrfs_free_space_ctl *ctl,
 			if (entry->offset + entry->bytes > offset)
 				break;
 		}
-
-		n = rb_next(&entry->offset_index);
-		if (!n)
-			return NULL;
-		entry = rb_entry(n, struct btrfs_free_space, offset_index);
 	}
 	return entry;
 }
 
-static inline void
-__unlink_free_space(struct btrfs_free_space_ctl *ctl,
-		    struct btrfs_free_space *info)
+static inline void unlink_free_space(struct btrfs_free_space_ctl *ctl,
+				     struct btrfs_free_space *info,
+				     bool update_stat)
 {
 	rb_erase(&info->offset_index, &ctl->free_space_offset);
+	rb_erase_cached(&info->bytes_index, &ctl->free_space_bytes);
 	ctl->free_extents--;
 
 	if (!info->bitmap && !btrfs_free_space_trimmed(info)) {
 		ctl->discardable_extents[BTRFS_STAT_CURR]--;
 		ctl->discardable_bytes[BTRFS_STAT_CURR] -= info->bytes;
 	}
-}
 
-static void unlink_free_space(struct btrfs_free_space_ctl *ctl,
-			      struct btrfs_free_space *info)
-{
-	__unlink_free_space(ctl, info);
-	ctl->free_space -= info->bytes;
+	if (update_stat)
+		ctl->free_space -= info->bytes;
 }
 
 static int link_free_space(struct btrfs_free_space_ctl *ctl,
@@ -1670,6 +1828,8 @@ static int link_free_space(struct btrfs_free_space_ctl *ctl,
 	if (ret)
 		return ret;
 
+	rb_add_cached(&info->bytes_index, &ctl->free_space_bytes, entry_less);
+
 	if (!info->bitmap && !btrfs_free_space_trimmed(info)) {
 		ctl->discardable_extents[BTRFS_STAT_CURR]++;
 		ctl->discardable_bytes[BTRFS_STAT_CURR] += info->bytes;
@@ -1680,47 +1840,25 @@ static int link_free_space(struct btrfs_free_space_ctl *ctl,
 	return ret;
 }
 
-static void recalculate_thresholds(struct btrfs_free_space_ctl *ctl)
+static void relink_bitmap_entry(struct btrfs_free_space_ctl *ctl,
+				struct btrfs_free_space *info)
 {
-	struct btrfs_block_group *block_group = ctl->private;
-	u64 max_bytes;
-	u64 bitmap_bytes;
-	u64 extent_bytes;
-	u64 size = block_group->length;
-	u64 bytes_per_bg = BITS_PER_BITMAP * ctl->unit;
-	u64 max_bitmaps = div64_u64(size + bytes_per_bg - 1, bytes_per_bg);
-
-	max_bitmaps = max_t(u64, max_bitmaps, 1);
-
-	ASSERT(ctl->total_bitmaps <= max_bitmaps);
-
-	/*
-	 * We are trying to keep the total amount of memory used per 1GiB of
-	 * space to be MAX_CACHE_BYTES_PER_GIG.  However, with a reclamation
-	 * mechanism of pulling extents >= FORCE_EXTENT_THRESHOLD out of
-	 * bitmaps, we may end up using more memory than this.
-	 */
-	if (size < SZ_1G)
-		max_bytes = MAX_CACHE_BYTES_PER_GIG;
-	else
-		max_bytes = MAX_CACHE_BYTES_PER_GIG * div_u64(size, SZ_1G);
-
-	bitmap_bytes = ctl->total_bitmaps * ctl->unit;
+	ASSERT(info->bitmap);
 
 	/*
-	 * we want the extent entry threshold to always be at most 1/2 the max
-	 * bytes we can have, or whatever is less than that.
+	 * If our entry is empty it's because we're on a cluster and we don't
+	 * want to re-link it into our ctl bytes index.
 	 */
-	extent_bytes = max_bytes - bitmap_bytes;
-	extent_bytes = min_t(u64, extent_bytes, max_bytes >> 1);
+	if (RB_EMPTY_NODE(&info->bytes_index))
+		return;
 
-	ctl->extents_thresh =
-		div_u64(extent_bytes, sizeof(struct btrfs_free_space));
+	rb_erase_cached(&info->bytes_index, &ctl->free_space_bytes);
+	rb_add_cached(&info->bytes_index, &ctl->free_space_bytes, entry_less);
 }
 
-static inline void __bitmap_clear_bits(struct btrfs_free_space_ctl *ctl,
-				       struct btrfs_free_space *info,
-				       u64 offset, u64 bytes)
+static inline void bitmap_clear_bits(struct btrfs_free_space_ctl *ctl,
+				     struct btrfs_free_space *info,
+				     u64 offset, u64 bytes, bool update_stat)
 {
 	unsigned long start, count, end;
 	int extent_delta = -1;
@@ -1736,6 +1874,8 @@ static inline void __bitmap_clear_bits(struct btrfs_free_space_ctl *ctl,
 	if (info->max_extent_size > ctl->unit)
 		info->max_extent_size = 0;
 
+	relink_bitmap_entry(ctl, info);
+
 	if (start && test_bit(start - 1, info->bitmap))
 		extent_delta++;
 
@@ -1747,14 +1887,9 @@ static inline void __bitmap_clear_bits(struct btrfs_free_space_ctl *ctl,
 		ctl->discardable_extents[BTRFS_STAT_CURR] += extent_delta;
 		ctl->discardable_bytes[BTRFS_STAT_CURR] -= bytes;
 	}
-}
 
-static void bitmap_clear_bits(struct btrfs_free_space_ctl *ctl,
-			      struct btrfs_free_space *info, u64 offset,
-			      u64 bytes)
-{
-	__bitmap_clear_bits(ctl, info, offset, bytes);
-	ctl->free_space -= bytes;
+	if (update_stat)
+		ctl->free_space -= bytes;
 }
 
 static void bitmap_set_bits(struct btrfs_free_space_ctl *ctl,
@@ -1771,9 +1906,16 @@ static void bitmap_set_bits(struct btrfs_free_space_ctl *ctl,
 
 	bitmap_set(info->bitmap, start, count);
 
+	/*
+	 * We set some bytes, we have no idea what the max extent size is
+	 * anymore.
+	 */
+	info->max_extent_size = 0;
 	info->bytes += bytes;
 	ctl->free_space += bytes;
 
+	relink_bitmap_entry(ctl, info);
+
 	if (start && test_bit(start - 1, info->bitmap))
 		extent_delta--;
 
@@ -1841,20 +1983,14 @@ static int search_bitmap(struct btrfs_free_space_ctl *ctl,
 
 	*bytes = (u64)(max_bits) * ctl->unit;
 	bitmap_info->max_extent_size = *bytes;
+	relink_bitmap_entry(ctl, bitmap_info);
 	return -1;
 }
 
-static inline u64 get_max_extent_size(struct btrfs_free_space *entry)
-{
-	if (entry->bitmap)
-		return entry->max_extent_size;
-	return entry->bytes;
-}
-
 /* Cache the size of the max extent in bytes */
 static struct btrfs_free_space *
 find_free_space(struct btrfs_free_space_ctl *ctl, u64 *offset, u64 *bytes,
-		unsigned long align, u64 *max_extent_size)
+		unsigned long align, u64 *max_extent_size, bool use_bytes_index)
 {
 	struct btrfs_free_space *entry;
 	struct rb_node *node;
@@ -1864,16 +2000,38 @@ find_free_space(struct btrfs_free_space_ctl *ctl, u64 *offset, u64 *bytes,
 
 	if (!ctl->free_space_offset.rb_node)
 		goto out;
+again:
+	if (use_bytes_index) {
+		node = rb_first_cached(&ctl->free_space_bytes);
+	} else {
+		entry = tree_search_offset(ctl, offset_to_bitmap(ctl, *offset),
+					   0, 1);
+		if (!entry)
+			goto out;
+		node = &entry->offset_index;
+	}
 
-	entry = tree_search_offset(ctl, offset_to_bitmap(ctl, *offset), 0, 1);
-	if (!entry)
-		goto out;
+	for (; node; node = rb_next(node)) {
+		if (use_bytes_index)
+			entry = rb_entry(node, struct btrfs_free_space,
+					 bytes_index);
+		else
+			entry = rb_entry(node, struct btrfs_free_space,
+					 offset_index);
 
-	for (node = &entry->offset_index; node; node = rb_next(node)) {
-		entry = rb_entry(node, struct btrfs_free_space, offset_index);
+		/*
+		 * If we are using the bytes index then all subsequent entries
+		 * in this tree are going to be < bytes, so simply set the max
+		 * extent size and exit the loop.
+		 *
+		 * If we're using the offset index then we need to keep going
+		 * through the rest of the tree.
+		 */
 		if (entry->bytes < *bytes) {
 			*max_extent_size = max(get_max_extent_size(entry),
 					       *max_extent_size);
+			if (use_bytes_index)
+				break;
 			continue;
 		}
 
@@ -1890,6 +2048,13 @@ find_free_space(struct btrfs_free_space_ctl *ctl, u64 *offset, u64 *bytes,
 			tmp = entry->offset;
 		}
 
+		/*
+		 * We don't break here if we're using the bytes index because we
+		 * may have another entry that has the correct alignment that is
+		 * the right size, so we don't want to miss that possibility.
+		 * At worst this adds another loop through the logic, but if we
+		 * broke here we could prematurely ENOSPC.
+		 */
 		if (entry->bytes < *bytes + align_off) {
 			*max_extent_size = max(get_max_extent_size(entry),
 					       *max_extent_size);
@@ -1897,6 +2062,7 @@ find_free_space(struct btrfs_free_space_ctl *ctl, u64 *offset, u64 *bytes,
 		}
 
 		if (entry->bitmap) {
+			struct rb_node *old_next = rb_next(node);
 			u64 size = *bytes;
 
 			ret = search_bitmap(ctl, entry, &tmp, &size, true);
@@ -1909,6 +2075,15 @@ find_free_space(struct btrfs_free_space_ctl *ctl, u64 *offset, u64 *bytes,
 					max(get_max_extent_size(entry),
 					    *max_extent_size);
 			}
+
+			/*
+			 * The bitmap may have gotten re-arranged in the space
+			 * index here because the max_extent_size may have been
+			 * updated.  Start from the beginning again if this
+			 * happened.
+			 */
+			if (use_bytes_index && old_next != rb_next(node))
+				goto again;
 			continue;
 		}
 
@@ -1920,29 +2095,6 @@ out:
 	return NULL;
 }
 
-static int count_bitmap_extents(struct btrfs_free_space_ctl *ctl,
-				struct btrfs_free_space *bitmap_info)
-{
-	struct btrfs_block_group *block_group = ctl->private;
-	u64 bytes = bitmap_info->bytes;
-	unsigned int rs, re;
-	int count = 0;
-
-	if (!block_group || !bytes)
-		return count;
-
-	bitmap_for_each_set_region(bitmap_info->bitmap, rs, re, 0,
-				   BITS_PER_BITMAP) {
-		bytes -= (rs - re) * ctl->unit;
-		count++;
-
-		if (!bytes)
-			break;
-	}
-
-	return count;
-}
-
 static void add_new_bitmap(struct btrfs_free_space_ctl *ctl,
 			   struct btrfs_free_space *info, u64 offset)
 {
@@ -1952,8 +2104,7 @@ static void add_new_bitmap(struct btrfs_free_space_ctl *ctl,
 	INIT_LIST_HEAD(&info->list);
 	link_free_space(ctl, info);
 	ctl->total_bitmaps++;
-
-	ctl->op->recalc_thresholds(ctl);
+	recalculate_thresholds(ctl);
 }
 
 static void free_bitmap(struct btrfs_free_space_ctl *ctl,
@@ -1971,11 +2122,11 @@ static void free_bitmap(struct btrfs_free_space_ctl *ctl,
 		ctl->discardable_bytes[BTRFS_STAT_CURR] -= bitmap_info->bytes;
 
 	}
-	unlink_free_space(ctl, bitmap_info);
+	unlink_free_space(ctl, bitmap_info, true);
 	kmem_cache_free(btrfs_free_space_bitmap_cachep, bitmap_info->bitmap);
 	kmem_cache_free(btrfs_free_space_cachep, bitmap_info);
 	ctl->total_bitmaps--;
-	ctl->op->recalc_thresholds(ctl);
+	recalculate_thresholds(ctl);
 }
 
 static noinline int remove_from_bitmap(struct btrfs_free_space_ctl *ctl,
@@ -2009,7 +2160,7 @@ again:
 	/* Cannot clear past the end of the bitmap */
 	search_bytes = min(search_bytes, end - search_start + 1);
 
-	bitmap_clear_bits(ctl, bitmap_info, search_start, search_bytes);
+	bitmap_clear_bits(ctl, bitmap_info, search_start, search_bytes, true);
 	*offset += search_bytes;
 	*bytes -= search_bytes;
 
@@ -2081,12 +2232,6 @@ static u64 add_bytes_to_bitmap(struct btrfs_free_space_ctl *ctl,
 
 	bitmap_set_bits(ctl, info, offset, bytes_to_set);
 
-	/*
-	 * We set some bytes, we have no idea what the max extent size is
-	 * anymore.
-	 */
-	info->max_extent_size = 0;
-
 	return bytes_to_set;
 
 }
@@ -2094,7 +2239,7 @@ static u64 add_bytes_to_bitmap(struct btrfs_free_space_ctl *ctl,
 static bool use_bitmap(struct btrfs_free_space_ctl *ctl,
 		      struct btrfs_free_space *info)
 {
-	struct btrfs_block_group *block_group = ctl->private;
+	struct btrfs_block_group *block_group = ctl->block_group;
 	struct btrfs_fs_info *fs_info = block_group->fs_info;
 	bool forced = false;
 
@@ -2142,7 +2287,6 @@ static bool use_bitmap(struct btrfs_free_space_ctl *ctl,
 }
 
 static const struct btrfs_free_space_op free_space_op = {
-	.recalc_thresholds	= recalculate_thresholds,
 	.use_bitmap		= use_bitmap,
 };
 
@@ -2164,7 +2308,7 @@ static int insert_into_bitmap(struct btrfs_free_space_ctl *ctl,
 		return 0;
 
 	if (ctl->op == &free_space_op)
-		block_group = ctl->private;
+		block_group = ctl->block_group;
 again:
 	/*
 	 * Since we link bitmaps right into the cluster we need to see if we
@@ -2287,7 +2431,7 @@ out:
 static bool try_merge_free_space(struct btrfs_free_space_ctl *ctl,
 			  struct btrfs_free_space *info, bool update_stat)
 {
-	struct btrfs_free_space *left_info;
+	struct btrfs_free_space *left_info = NULL;
 	struct btrfs_free_space *right_info;
 	bool merged = false;
 	u64 offset = info->offset;
@@ -2303,16 +2447,13 @@ static bool try_merge_free_space(struct btrfs_free_space_ctl *ctl,
 	if (right_info && rb_prev(&right_info->offset_index))
 		left_info = rb_entry(rb_prev(&right_info->offset_index),
 				     struct btrfs_free_space, offset_index);
-	else
+	else if (!right_info)
 		left_info = tree_search_offset(ctl, offset - 1, 0, 0);
 
 	/* See try_merge_free_space() comment. */
 	if (right_info && !right_info->bitmap &&
 	    (!is_trimmed || btrfs_free_space_trimmed(right_info))) {
-		if (update_stat)
-			unlink_free_space(ctl, right_info);
-		else
-			__unlink_free_space(ctl, right_info);
+		unlink_free_space(ctl, right_info, update_stat);
 		info->bytes += right_info->bytes;
 		kmem_cache_free(btrfs_free_space_cachep, right_info);
 		merged = true;
@@ -2322,10 +2463,7 @@ static bool try_merge_free_space(struct btrfs_free_space_ctl *ctl,
 	if (left_info && !left_info->bitmap &&
 	    left_info->offset + left_info->bytes == offset &&
 	    (!is_trimmed || btrfs_free_space_trimmed(left_info))) {
-		if (update_stat)
-			unlink_free_space(ctl, left_info);
-		else
-			__unlink_free_space(ctl, left_info);
+		unlink_free_space(ctl, left_info, update_stat);
 		info->offset = left_info->offset;
 		info->bytes += left_info->bytes;
 		kmem_cache_free(btrfs_free_space_cachep, left_info);
@@ -2361,10 +2499,7 @@ static bool steal_from_bitmap_to_end(struct btrfs_free_space_ctl *ctl,
 	if (!btrfs_free_space_trimmed(bitmap))
 		info->trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
 
-	if (update_stat)
-		bitmap_clear_bits(ctl, bitmap, end, bytes);
-	else
-		__bitmap_clear_bits(ctl, bitmap, end, bytes);
+	bitmap_clear_bits(ctl, bitmap, end, bytes, update_stat);
 
 	if (!bitmap->bytes)
 		free_bitmap(ctl, bitmap);
@@ -2418,10 +2553,7 @@ static bool steal_from_bitmap_to_front(struct btrfs_free_space_ctl *ctl,
 	if (!btrfs_free_space_trimmed(bitmap))
 		info->trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
 
-	if (update_stat)
-		bitmap_clear_bits(ctl, bitmap, info->offset, bytes);
-	else
-		__bitmap_clear_bits(ctl, bitmap, info->offset, bytes);
+	bitmap_clear_bits(ctl, bitmap, info->offset, bytes, update_stat);
 
 	if (!bitmap->bytes)
 		free_bitmap(ctl, bitmap);
@@ -2465,16 +2597,18 @@ static void steal_from_bitmap(struct btrfs_free_space_ctl *ctl,
 	}
 }
 
-int __btrfs_add_free_space(struct btrfs_fs_info *fs_info,
-			   struct btrfs_free_space_ctl *ctl,
+int __btrfs_add_free_space(struct btrfs_block_group *block_group,
 			   u64 offset, u64 bytes,
 			   enum btrfs_trim_state trim_state)
 {
-	struct btrfs_block_group *block_group = ctl->private;
+	struct btrfs_fs_info *fs_info = block_group->fs_info;
+	struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
 	struct btrfs_free_space *info;
 	int ret = 0;
 	u64 filter_bytes = bytes;
 
+	ASSERT(!btrfs_is_zoned(fs_info));
+
 	info = kmem_cache_zalloc(btrfs_free_space_cachep, GFP_NOFS);
 	if (!info)
 		return -ENOMEM;
@@ -2483,6 +2617,7 @@ int __btrfs_add_free_space(struct btrfs_fs_info *fs_info,
 	info->bytes = bytes;
 	info->trim_state = trim_state;
 	RB_CLEAR_NODE(&info->offset_index);
+	RB_CLEAR_NODE(&info->bytes_index);
 
 	spin_lock(&ctl->tree_lock);
 
@@ -2516,7 +2651,7 @@ link:
 	if (ret)
 		kmem_cache_free(btrfs_free_space_cachep, info);
 out:
-	btrfs_discard_update_discardable(block_group, ctl);
+	btrfs_discard_update_discardable(block_group);
 	spin_unlock(&ctl->tree_lock);
 
 	if (ret) {
@@ -2532,17 +2667,87 @@ out:
 	return ret;
 }
 
+static int __btrfs_add_free_space_zoned(struct btrfs_block_group *block_group,
+					u64 bytenr, u64 size, bool used)
+{
+	struct btrfs_space_info *sinfo = block_group->space_info;
+	struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
+	u64 offset = bytenr - block_group->start;
+	u64 to_free, to_unusable;
+	int bg_reclaim_threshold = 0;
+	bool initial = (size == block_group->length);
+	u64 reclaimable_unusable;
+
+	WARN_ON(!initial && offset + size > block_group->zone_capacity);
+
+	if (!initial)
+		bg_reclaim_threshold = READ_ONCE(sinfo->bg_reclaim_threshold);
+
+	spin_lock(&ctl->tree_lock);
+	if (!used)
+		to_free = size;
+	else if (initial)
+		to_free = block_group->zone_capacity;
+	else if (offset >= block_group->alloc_offset)
+		to_free = size;
+	else if (offset + size <= block_group->alloc_offset)
+		to_free = 0;
+	else
+		to_free = offset + size - block_group->alloc_offset;
+	to_unusable = size - to_free;
+
+	ctl->free_space += to_free;
+	/*
+	 * If the block group is read-only, we should account freed space into
+	 * bytes_readonly.
+	 */
+	if (!block_group->ro)
+		block_group->zone_unusable += to_unusable;
+	spin_unlock(&ctl->tree_lock);
+	if (!used) {
+		spin_lock(&block_group->lock);
+		block_group->alloc_offset -= size;
+		spin_unlock(&block_group->lock);
+	}
+
+	reclaimable_unusable = block_group->zone_unusable -
+			       (block_group->length - block_group->zone_capacity);
+	/* All the region is now unusable. Mark it as unused and reclaim */
+	if (block_group->zone_unusable == block_group->length) {
+		btrfs_mark_bg_unused(block_group);
+	} else if (bg_reclaim_threshold &&
+		   reclaimable_unusable >=
+		   div_factor_fine(block_group->zone_capacity,
+				   bg_reclaim_threshold)) {
+		btrfs_mark_bg_to_reclaim(block_group);
+	}
+
+	return 0;
+}
+
 int btrfs_add_free_space(struct btrfs_block_group *block_group,
 			 u64 bytenr, u64 size)
 {
 	enum btrfs_trim_state trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
 
+	if (btrfs_is_zoned(block_group->fs_info))
+		return __btrfs_add_free_space_zoned(block_group, bytenr, size,
+						    true);
+
 	if (btrfs_test_opt(block_group->fs_info, DISCARD_SYNC))
 		trim_state = BTRFS_TRIM_STATE_TRIMMED;
 
-	return __btrfs_add_free_space(block_group->fs_info,
-				      block_group->free_space_ctl,
-				      bytenr, size, trim_state);
+	return __btrfs_add_free_space(block_group, bytenr, size, trim_state);
+}
+
+int btrfs_add_free_space_unused(struct btrfs_block_group *block_group,
+				u64 bytenr, u64 size)
+{
+	if (btrfs_is_zoned(block_group->fs_info))
+		return __btrfs_add_free_space_zoned(block_group, bytenr, size,
+						    false);
+
+	return btrfs_add_free_space(block_group, bytenr, size);
 }
 
 /*
@@ -2555,13 +2760,15 @@ int btrfs_add_free_space_async_trimmed(struct btrfs_block_group *block_group,
 {
 	enum btrfs_trim_state trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
 
+	if (btrfs_is_zoned(block_group->fs_info))
+		return __btrfs_add_free_space_zoned(block_group, bytenr, size,
+						    true);
+
 	if (btrfs_test_opt(block_group->fs_info, DISCARD_SYNC) ||
 	    btrfs_test_opt(block_group->fs_info, DISCARD_ASYNC))
 		trim_state = BTRFS_TRIM_STATE_TRIMMED;
 
-	return __btrfs_add_free_space(block_group->fs_info,
-				      block_group->free_space_ctl,
-				      bytenr, size, trim_state);
+	return __btrfs_add_free_space(block_group, bytenr, size, trim_state);
 }
 
 int btrfs_remove_free_space(struct btrfs_block_group *block_group,
@@ -2572,6 +2779,26 @@ int btrfs_remove_free_space(struct btrfs_block_group *block_group,
 	int ret;
 	bool re_search = false;
 
+	if (btrfs_is_zoned(block_group->fs_info)) {
+		/*
+		 * This can happen with conventional zones when replaying log.
+		 * Since the allocation info of tree-log nodes are not recorded
+		 * to the extent-tree, calculate_alloc_pointer() failed to
+		 * advance the allocation pointer after last allocated tree log
+		 * node blocks.
+		 *
+		 * This function is called from
+		 * btrfs_pin_extent_for_log_replay() when replaying the log.
+		 * Advance the pointer not to overwrite the tree-log nodes.
+		 */
+		if (block_group->start + block_group->alloc_offset <
+		    offset + bytes) {
+			block_group->alloc_offset =
+				offset + bytes - block_group->start;
+		}
+		return 0;
+	}
+
 	spin_lock(&ctl->tree_lock);
 
 again:
@@ -2600,7 +2827,7 @@ again:
 
 	re_search = false;
 	if (!info->bitmap) {
-		unlink_free_space(ctl, info);
+		unlink_free_space(ctl, info, true);
 		if (offset == info->offset) {
 			u64 to_free = min(bytes, info->bytes);
 
@@ -2636,7 +2863,7 @@ again:
 			}
 			spin_unlock(&ctl->tree_lock);
 
-			ret = __btrfs_add_free_space(block_group->fs_info, ctl,
+			ret = __btrfs_add_free_space(block_group,
 						     offset + bytes,
 						     old_end - (offset + bytes),
 						     info->trim_state);
@@ -2651,7 +2878,7 @@ again:
 		goto again;
 	}
 out_lock:
-	btrfs_discard_update_discardable(block_group, ctl);
+	btrfs_discard_update_discardable(block_group);
 	spin_unlock(&ctl->tree_lock);
 out:
 	return ret;
@@ -2666,6 +2893,18 @@ void btrfs_dump_free_space(struct btrfs_block_group *block_group,
 	struct rb_node *n;
 	int count = 0;
 
+	/*
+	 * Zoned btrfs does not use free space tree and cluster. Just print
+	 * out the free space after the allocation offset.
+	 */
+	if (btrfs_is_zoned(fs_info)) {
+		btrfs_info(fs_info, "free space %llu active %d",
+			   block_group->zone_capacity - block_group->alloc_offset,
+			   test_bit(BLOCK_GROUP_FLAG_ZONE_IS_ACTIVE,
+				    &block_group->runtime_flags));
+		return;
+	}
+
 	spin_lock(&ctl->tree_lock);
 	for (n = rb_first(&ctl->free_space_offset); n; n = rb_next(n)) {
 		info = rb_entry(n, struct btrfs_free_space, offset_index);
@@ -2682,16 +2921,17 @@ void btrfs_dump_free_space(struct btrfs_block_group *block_group,
 		   "%d blocks of free space at or bigger than bytes is", count);
 }
 
-void btrfs_init_free_space_ctl(struct btrfs_block_group *block_group)
+void btrfs_init_free_space_ctl(struct btrfs_block_group *block_group,
+			       struct btrfs_free_space_ctl *ctl)
 {
 	struct btrfs_fs_info *fs_info = block_group->fs_info;
-	struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
 
 	spin_lock_init(&ctl->tree_lock);
 	ctl->unit = fs_info->sectorsize;
 	ctl->start = block_group->start;
-	ctl->private = block_group;
+	ctl->block_group = block_group;
 	ctl->op = &free_space_op;
+	ctl->free_space_bytes = RB_ROOT_CACHED;
 	INIT_LIST_HEAD(&ctl->trimming_ranges);
 	mutex_init(&ctl->cache_writeout_mutex);
 
@@ -2709,8 +2949,7 @@ void btrfs_init_free_space_ctl(struct btrfs_block_group *block_group)
  * pointed to by the cluster, someone else raced in and freed the
  * cluster already.  In that case, we just return without changing anything
  */
-static int
-__btrfs_return_cluster_to_free_space(
+static void __btrfs_return_cluster_to_free_space(
 			     struct btrfs_block_group *block_group,
 			     struct btrfs_free_cluster *cluster)
 {
@@ -2719,8 +2958,10 @@ __btrfs_return_cluster_to_free_space(
 	struct rb_node *node;
 
 	spin_lock(&cluster->lock);
-	if (cluster->block_group != block_group)
-		goto out;
+	if (cluster->block_group != block_group) {
+		spin_unlock(&cluster->lock);
+		return;
+	}
 
 	cluster->block_group = NULL;
 	cluster->window_start = 0;
@@ -2756,41 +2997,12 @@ __btrfs_return_cluster_to_free_space(
 		}
 		tree_insert_offset(&ctl->free_space_offset,
 				   entry->offset, &entry->offset_index, bitmap);
+		rb_add_cached(&entry->bytes_index, &ctl->free_space_bytes,
+			      entry_less);
 	}
 	cluster->root = RB_ROOT;
-
-out:
 	spin_unlock(&cluster->lock);
 	btrfs_put_block_group(block_group);
-	return 0;
-}
-
-static void __btrfs_remove_free_space_cache_locked(
-				struct btrfs_free_space_ctl *ctl)
-{
-	struct btrfs_free_space *info;
-	struct rb_node *node;
-
-	while ((node = rb_last(&ctl->free_space_offset)) != NULL) {
-		info = rb_entry(node, struct btrfs_free_space, offset_index);
-		if (!info->bitmap) {
-			unlink_free_space(ctl, info);
-			kmem_cache_free(btrfs_free_space_cachep, info);
-		} else {
-			free_bitmap(ctl, info);
-		}
-
-		cond_resched_lock(&ctl->tree_lock);
-	}
-}
-
-void __btrfs_remove_free_space_cache(struct btrfs_free_space_ctl *ctl)
-{
-	spin_lock(&ctl->tree_lock);
-	__btrfs_remove_free_space_cache_locked(ctl);
-	if (ctl->private)
-		btrfs_discard_update_discardable(ctl->private, ctl);
-	spin_unlock(&ctl->tree_lock);
 }
 
 void btrfs_remove_free_space_cache(struct btrfs_block_group *block_group)
@@ -2810,8 +3022,8 @@ void btrfs_remove_free_space_cache(struct btrfs_block_group *block_group)
 
 		cond_resched_lock(&ctl->tree_lock);
 	}
-	__btrfs_remove_free_space_cache_locked(ctl);
-	btrfs_discard_update_discardable(block_group, ctl);
+	__btrfs_remove_free_space_cache(ctl);
+	btrfs_discard_update_discardable(block_group);
 	spin_unlock(&ctl->tree_lock);
 
 }
@@ -2860,16 +3072,20 @@ u64 btrfs_find_space_for_alloc(struct btrfs_block_group *block_group,
 	u64 align_gap = 0;
 	u64 align_gap_len = 0;
 	enum btrfs_trim_state align_gap_trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
+	bool use_bytes_index = (offset == block_group->start);
+
+	ASSERT(!btrfs_is_zoned(block_group->fs_info));
 
 	spin_lock(&ctl->tree_lock);
 	entry = find_free_space(ctl, &offset, &bytes_search,
-				block_group->full_stripe_len, max_extent_size);
+				block_group->full_stripe_len, max_extent_size,
+				use_bytes_index);
 	if (!entry)
 		goto out;
 
 	ret = offset;
 	if (entry->bitmap) {
-		bitmap_clear_bits(ctl, entry, offset, bytes);
+		bitmap_clear_bits(ctl, entry, offset, bytes, true);
 
 		if (!btrfs_free_space_trimmed(entry))
 			atomic64_add(bytes, &discard_ctl->discard_bytes_saved);
@@ -2877,7 +3093,7 @@ u64 btrfs_find_space_for_alloc(struct btrfs_block_group *block_group,
 		if (!entry->bytes)
 			free_bitmap(ctl, entry);
 	} else {
-		unlink_free_space(ctl, entry);
+		unlink_free_space(ctl, entry, true);
 		align_gap_len = offset - entry->offset;
 		align_gap = entry->offset;
 		align_gap_trim_state = entry->trim_state;
@@ -2895,12 +3111,11 @@ u64 btrfs_find_space_for_alloc(struct btrfs_block_group *block_group,
 			link_free_space(ctl, entry);
 	}
 out:
-	btrfs_discard_update_discardable(block_group, ctl);
+	btrfs_discard_update_discardable(block_group);
 	spin_unlock(&ctl->tree_lock);
 
 	if (align_gap_len)
-		__btrfs_add_free_space(block_group->fs_info, ctl,
-				       align_gap, align_gap_len,
+		__btrfs_add_free_space(block_group, align_gap, align_gap_len,
 				       align_gap_trim_state);
 	return ret;
 }
@@ -2913,12 +3128,11 @@ out:
  * Otherwise, it'll get a reference on the block group pointed to by the
  * cluster and remove the cluster from it.
  */
-int btrfs_return_cluster_to_free_space(
+void btrfs_return_cluster_to_free_space(
 			       struct btrfs_block_group *block_group,
 			       struct btrfs_free_cluster *cluster)
 {
 	struct btrfs_free_space_ctl *ctl;
-	int ret;
 
 	/* first, get a safe pointer to the block group */
 	spin_lock(&cluster->lock);
@@ -2926,28 +3140,27 @@ int btrfs_return_cluster_to_free_space(
 		block_group = cluster->block_group;
 		if (!block_group) {
 			spin_unlock(&cluster->lock);
-			return 0;
+			return;
 		}
 	} else if (cluster->block_group != block_group) {
 		/* someone else has already freed it don't redo their work */
 		spin_unlock(&cluster->lock);
-		return 0;
+		return;
 	}
-	atomic_inc(&block_group->count);
+	btrfs_get_block_group(block_group);
 	spin_unlock(&cluster->lock);
 
 	ctl = block_group->free_space_ctl;
 
 	/* now return any extents the cluster had on it */
 	spin_lock(&ctl->tree_lock);
-	ret = __btrfs_return_cluster_to_free_space(block_group, cluster);
+	__btrfs_return_cluster_to_free_space(block_group, cluster);
 	spin_unlock(&ctl->tree_lock);
 
 	btrfs_discard_queue_work(&block_group->fs_info->discard_ctl, block_group);
 
 	/* finally drop our ref */
 	btrfs_put_block_group(block_group);
-	return ret;
 }
 
 static u64 btrfs_alloc_from_bitmap(struct btrfs_block_group *block_group,
@@ -2973,7 +3186,7 @@ static u64 btrfs_alloc_from_bitmap(struct btrfs_block_group *block_group,
 	}
 
 	ret = search_start;
-	__bitmap_clear_bits(ctl, entry, ret, bytes);
+	bitmap_clear_bits(ctl, entry, ret, bytes, false);
 
 	return ret;
 }
@@ -2994,6 +3207,8 @@ u64 btrfs_alloc_from_cluster(struct btrfs_block_group *block_group,
 	struct rb_node *node;
 	u64 ret = 0;
 
+	ASSERT(!btrfs_is_zoned(block_group->fs_info));
+
 	spin_lock(&cluster->lock);
 	if (bytes > cluster->max_size)
 		goto out;
@@ -3042,8 +3257,6 @@ u64 btrfs_alloc_from_cluster(struct btrfs_block_group *block_group,
 			entry->bytes -= bytes;
 		}
 
-		if (entry->bytes == 0)
-			rb_erase(&entry->offset_index, &cluster->root);
 		break;
 	}
 out:
@@ -3060,19 +3273,23 @@ out:
 	ctl->free_space -= bytes;
 	if (!entry->bitmap && !btrfs_free_space_trimmed(entry))
 		ctl->discardable_bytes[BTRFS_STAT_CURR] -= bytes;
+
+	spin_lock(&cluster->lock);
 	if (entry->bytes == 0) {
+		rb_erase(&entry->offset_index, &cluster->root);
 		ctl->free_extents--;
 		if (entry->bitmap) {
 			kmem_cache_free(btrfs_free_space_bitmap_cachep,
 					entry->bitmap);
 			ctl->total_bitmaps--;
-			ctl->op->recalc_thresholds(ctl);
+			recalculate_thresholds(ctl);
 		} else if (!btrfs_free_space_trimmed(entry)) {
 			ctl->discardable_extents[BTRFS_STAT_CURR]--;
 		}
 		kmem_cache_free(btrfs_free_space_cachep, entry);
 	}
 
+	spin_unlock(&cluster->lock);
 	spin_unlock(&ctl->tree_lock);
 
 	return ret;
@@ -3145,6 +3362,17 @@ again:
 
 	cluster->window_start = start * ctl->unit + entry->offset;
 	rb_erase(&entry->offset_index, &ctl->free_space_offset);
+	rb_erase_cached(&entry->bytes_index, &ctl->free_space_bytes);
+
+	/*
+	 * We need to know if we're currently on the normal space index when we
+	 * manipulate the bitmap so that we know we need to remove and re-insert
+	 * it into the space_index tree.  Clear the bytes_index node here so the
+	 * bitmap manipulation helpers know not to mess with the space_index
+	 * until this bitmap entry is added back into the normal cache.
+	 */
+	RB_CLEAR_NODE(&entry->bytes_index);
+
 	ret = tree_insert_offset(&cluster->root, entry->offset,
 				 &entry->offset_index, 1);
 	ASSERT(!ret); /* -EEXIST; Logic error */
@@ -3235,6 +3463,7 @@ setup_cluster_no_bitmap(struct btrfs_block_group *block_group,
 			continue;
 
 		rb_erase(&entry->offset_index, &ctl->free_space_offset);
+		rb_erase_cached(&entry->bytes_index, &ctl->free_space_bytes);
 		ret = tree_insert_offset(&cluster->root, entry->offset,
 					 &entry->offset_index, 0);
 		total_size += entry->bytes;
@@ -3320,7 +3549,8 @@ int btrfs_find_space_cluster(struct btrfs_block_group *block_group,
 	 * data, keep it dense.
 	 */
 	if (btrfs_test_opt(fs_info, SSD_SPREAD)) {
-		cont1_bytes = min_bytes = bytes + empty_size;
+		cont1_bytes = bytes + empty_size;
+		min_bytes = cont1_bytes;
 	} else if (block_group->flags & BTRFS_BLOCK_GROUP_METADATA) {
 		cont1_bytes = bytes;
 		min_bytes = fs_info->sectorsize;
@@ -3364,7 +3594,7 @@ int btrfs_find_space_cluster(struct btrfs_block_group *block_group,
 		list_del_init(&entry->list);
 
 	if (!ret) {
-		atomic_inc(&block_group->count);
+		btrfs_get_block_group(block_group);
 		list_add_tail(&cluster->block_group_list,
 			      &block_group->cluster_list);
 		cluster->block_group = block_group;
@@ -3426,13 +3656,13 @@ static int do_trimming(struct btrfs_block_group *block_group,
 
 	mutex_lock(&ctl->cache_writeout_mutex);
 	if (reserved_start < start)
-		__btrfs_add_free_space(fs_info, ctl, reserved_start,
+		__btrfs_add_free_space(block_group, reserved_start,
 				       start - reserved_start,
 				       reserved_trim_state);
 	if (start + bytes < reserved_start + reserved_bytes)
-		__btrfs_add_free_space(fs_info, ctl, end, reserved_end - end,
+		__btrfs_add_free_space(block_group, end, reserved_end - end,
 				       reserved_trim_state);
-	__btrfs_add_free_space(fs_info, ctl, start, bytes, trim_state);
+	__btrfs_add_free_space(block_group, start, bytes, trim_state);
 	list_del(&trim_entry->list);
 	mutex_unlock(&ctl->cache_writeout_mutex);
 
@@ -3506,7 +3736,7 @@ static int trim_no_bitmap(struct btrfs_block_group *block_group,
 				mutex_unlock(&ctl->cache_writeout_mutex);
 				goto next;
 			}
-			unlink_free_space(ctl, entry);
+			unlink_free_space(ctl, entry, true);
 			/*
 			 * Let bytes = BTRFS_MAX_DISCARD_SIZE + X.
 			 * If X < BTRFS_ASYNC_DISCARD_MIN_FILTER, we won't trim
@@ -3532,7 +3762,7 @@ static int trim_no_bitmap(struct btrfs_block_group *block_group,
 				goto next;
 			}
 
-			unlink_free_space(ctl, entry);
+			unlink_free_space(ctl, entry, true);
 			kmem_cache_free(btrfs_free_space_cachep, entry);
 		}
 
@@ -3719,7 +3949,7 @@ static int trim_bitmaps(struct btrfs_block_group *block_group,
 		    bytes > (max_discard_size + minlen))
 			bytes = max_discard_size;
 
-		bitmap_clear_bits(ctl, entry, start, bytes);
+		bitmap_clear_bits(ctl, entry, start, bytes, true);
 		if (entry->bytes == 0)
 			free_bitmap(ctl, entry);
 
@@ -3763,46 +3993,6 @@ out:
 	return ret;
 }
 
-void btrfs_get_block_group_trimming(struct btrfs_block_group *cache)
-{
-	atomic_inc(&cache->trimming);
-}
-
-void btrfs_put_block_group_trimming(struct btrfs_block_group *block_group)
-{
-	struct btrfs_fs_info *fs_info = block_group->fs_info;
-	struct extent_map_tree *em_tree;
-	struct extent_map *em;
-	bool cleanup;
-
-	spin_lock(&block_group->lock);
-	cleanup = (atomic_dec_and_test(&block_group->trimming) &&
-		   block_group->removed);
-	spin_unlock(&block_group->lock);
-
-	if (cleanup) {
-		mutex_lock(&fs_info->chunk_mutex);
-		em_tree = &fs_info->mapping_tree;
-		write_lock(&em_tree->lock);
-		em = lookup_extent_mapping(em_tree, block_group->start,
-					   1);
-		BUG_ON(!em); /* logic error, can't happen */
-		remove_extent_mapping(em_tree, em);
-		write_unlock(&em_tree->lock);
-		mutex_unlock(&fs_info->chunk_mutex);
-
-		/* once for us and once for the tree */
-		free_extent_map(em);
-		free_extent_map(em);
-
-		/*
-		 * We've left one free space entry and other tasks trimming
-		 * this block group have left 1 entry each one. Free them.
-		 */
-		__btrfs_remove_free_space_cache(block_group->free_space_ctl);
-	}
-}
-
 int btrfs_trim_block_group(struct btrfs_block_group *block_group,
 			   u64 *trimmed, u64 start, u64 end, u64 minlen)
 {
@@ -3810,14 +4000,16 @@ int btrfs_trim_block_group(struct btrfs_block_group *block_group,
 	int ret;
 	u64 rem = 0;
 
+	ASSERT(!btrfs_is_zoned(block_group->fs_info));
+
 	*trimmed = 0;
 
 	spin_lock(&block_group->lock);
-	if (block_group->removed) {
+	if (test_bit(BLOCK_GROUP_FLAG_REMOVED, &block_group->runtime_flags)) {
 		spin_unlock(&block_group->lock);
 		return 0;
 	}
-	btrfs_get_block_group_trimming(block_group);
+	btrfs_freeze_block_group(block_group);
 	spin_unlock(&block_group->lock);
 
 	ret = trim_no_bitmap(block_group, trimmed, start, end, minlen, false);
@@ -3830,7 +4022,7 @@ int btrfs_trim_block_group(struct btrfs_block_group *block_group,
 	if (rem)
 		reset_trimming_bitmap(ctl, offset_to_bitmap(ctl, end));
 out:
-	btrfs_put_block_group_trimming(block_group);
+	btrfs_unfreeze_block_group(block_group);
 	return ret;
 }
 
@@ -3843,15 +4035,15 @@ int btrfs_trim_block_group_extents(struct btrfs_block_group *block_group,
 	*trimmed = 0;
 
 	spin_lock(&block_group->lock);
-	if (block_group->removed) {
+	if (test_bit(BLOCK_GROUP_FLAG_REMOVED, &block_group->runtime_flags)) {
 		spin_unlock(&block_group->lock);
 		return 0;
 	}
-	btrfs_get_block_group_trimming(block_group);
+	btrfs_freeze_block_group(block_group);
 	spin_unlock(&block_group->lock);
 
 	ret = trim_no_bitmap(block_group, trimmed, start, end, minlen, async);
-	btrfs_put_block_group_trimming(block_group);
+	btrfs_unfreeze_block_group(block_group);
 
 	return ret;
 }
@@ -3865,183 +4057,77 @@ int btrfs_trim_block_group_bitmaps(struct btrfs_block_group *block_group,
 	*trimmed = 0;
 
 	spin_lock(&block_group->lock);
-	if (block_group->removed) {
+	if (test_bit(BLOCK_GROUP_FLAG_REMOVED, &block_group->runtime_flags)) {
 		spin_unlock(&block_group->lock);
 		return 0;
 	}
-	btrfs_get_block_group_trimming(block_group);
+	btrfs_freeze_block_group(block_group);
 	spin_unlock(&block_group->lock);
 
 	ret = trim_bitmaps(block_group, trimmed, start, end, minlen, maxlen,
 			   async);
 
-	btrfs_put_block_group_trimming(block_group);
+	btrfs_unfreeze_block_group(block_group);
 
 	return ret;
 }
 
-/*
- * Find the left-most item in the cache tree, and then return the
- * smallest inode number in the item.
- *
- * Note: the returned inode number may not be the smallest one in
- * the tree, if the left-most item is a bitmap.
- */
-u64 btrfs_find_ino_for_alloc(struct btrfs_root *fs_root)
-{
-	struct btrfs_free_space_ctl *ctl = fs_root->free_ino_ctl;
-	struct btrfs_free_space *entry = NULL;
-	u64 ino = 0;
-
-	spin_lock(&ctl->tree_lock);
-
-	if (RB_EMPTY_ROOT(&ctl->free_space_offset))
-		goto out;
-
-	entry = rb_entry(rb_first(&ctl->free_space_offset),
-			 struct btrfs_free_space, offset_index);
-
-	if (!entry->bitmap) {
-		ino = entry->offset;
-
-		unlink_free_space(ctl, entry);
-		entry->offset++;
-		entry->bytes--;
-		if (!entry->bytes)
-			kmem_cache_free(btrfs_free_space_cachep, entry);
-		else
-			link_free_space(ctl, entry);
-	} else {
-		u64 offset = 0;
-		u64 count = 1;
-		int ret;
-
-		ret = search_bitmap(ctl, entry, &offset, &count, true);
-		/* Logic error; Should be empty if it can't find anything */
-		ASSERT(!ret);
-
-		ino = offset;
-		bitmap_clear_bits(ctl, entry, offset, 1);
-		if (entry->bytes == 0)
-			free_bitmap(ctl, entry);
-	}
-out:
-	spin_unlock(&ctl->tree_lock);
-
-	return ino;
-}
-
-struct inode *lookup_free_ino_inode(struct btrfs_root *root,
-				    struct btrfs_path *path)
-{
-	struct inode *inode = NULL;
-
-	spin_lock(&root->ino_cache_lock);
-	if (root->ino_cache_inode)
-		inode = igrab(root->ino_cache_inode);
-	spin_unlock(&root->ino_cache_lock);
-	if (inode)
-		return inode;
-
-	inode = __lookup_free_space_inode(root, path, 0);
-	if (IS_ERR(inode))
-		return inode;
-
-	spin_lock(&root->ino_cache_lock);
-	if (!btrfs_fs_closing(root->fs_info))
-		root->ino_cache_inode = igrab(inode);
-	spin_unlock(&root->ino_cache_lock);
-
-	return inode;
-}
-
-int create_free_ino_inode(struct btrfs_root *root,
-			  struct btrfs_trans_handle *trans,
-			  struct btrfs_path *path)
+bool btrfs_free_space_cache_v1_active(struct btrfs_fs_info *fs_info)
 {
-	return __create_free_space_inode(root, trans, path,
-					 BTRFS_FREE_INO_OBJECTID, 0);
+	return btrfs_super_cache_generation(fs_info->super_copy);
 }
 
-int load_free_ino_cache(struct btrfs_fs_info *fs_info, struct btrfs_root *root)
+static int cleanup_free_space_cache_v1(struct btrfs_fs_info *fs_info,
+				       struct btrfs_trans_handle *trans)
 {
-	struct btrfs_free_space_ctl *ctl = root->free_ino_ctl;
-	struct btrfs_path *path;
-	struct inode *inode;
+	struct btrfs_block_group *block_group;
+	struct rb_node *node;
 	int ret = 0;
-	u64 root_gen = btrfs_root_generation(&root->root_item);
-
-	if (!btrfs_test_opt(fs_info, INODE_MAP_CACHE))
-		return 0;
 
-	/*
-	 * If we're unmounting then just return, since this does a search on the
-	 * normal root and not the commit root and we could deadlock.
-	 */
-	if (btrfs_fs_closing(fs_info))
-		return 0;
+	btrfs_info(fs_info, "cleaning free space cache v1");
 
-	path = btrfs_alloc_path();
-	if (!path)
-		return 0;
-
-	inode = lookup_free_ino_inode(root, path);
-	if (IS_ERR(inode))
-		goto out;
-
-	if (root_gen != BTRFS_I(inode)->generation)
-		goto out_put;
-
-	ret = __load_free_space_cache(root, inode, ctl, path, 0);
-
-	if (ret < 0)
-		btrfs_err(fs_info,
-			"failed to load free ino cache for root %llu",
-			root->root_key.objectid);
-out_put:
-	iput(inode);
+	node = rb_first_cached(&fs_info->block_group_cache_tree);
+	while (node) {
+		block_group = rb_entry(node, struct btrfs_block_group, cache_node);
+		ret = btrfs_remove_free_space_inode(trans, NULL, block_group);
+		if (ret)
+			goto out;
+		node = rb_next(node);
+	}
 out:
-	btrfs_free_path(path);
 	return ret;
 }
 
-int btrfs_write_out_ino_cache(struct btrfs_root *root,
-			      struct btrfs_trans_handle *trans,
-			      struct btrfs_path *path,
-			      struct inode *inode)
+int btrfs_set_free_space_cache_v1_active(struct btrfs_fs_info *fs_info, bool active)
 {
-	struct btrfs_fs_info *fs_info = root->fs_info;
-	struct btrfs_free_space_ctl *ctl = root->free_ino_ctl;
+	struct btrfs_trans_handle *trans;
 	int ret;
-	struct btrfs_io_ctl io_ctl;
-	bool release_metadata = true;
 
-	if (!btrfs_test_opt(fs_info, INODE_MAP_CACHE))
-		return 0;
+	/*
+	 * update_super_roots will appropriately set or unset
+	 * super_copy->cache_generation based on SPACE_CACHE and
+	 * BTRFS_FS_CLEANUP_SPACE_CACHE_V1. For this reason, we need a
+	 * transaction commit whether we are enabling space cache v1 and don't
+	 * have any other work to do, or are disabling it and removing free
+	 * space inodes.
+	 */
+	trans = btrfs_start_transaction(fs_info->tree_root, 0);
+	if (IS_ERR(trans))
+		return PTR_ERR(trans);
 
-	memset(&io_ctl, 0, sizeof(io_ctl));
-	ret = __btrfs_write_out_cache(root, inode, ctl, NULL, &io_ctl, trans);
-	if (!ret) {
-		/*
-		 * At this point writepages() didn't error out, so our metadata
-		 * reservation is released when the writeback finishes, at
-		 * inode.c:btrfs_finish_ordered_io(), regardless of it finishing
-		 * with or without an error.
-		 */
-		release_metadata = false;
-		ret = btrfs_wait_cache_io_root(root, trans, &io_ctl, path);
+	if (!active) {
+		set_bit(BTRFS_FS_CLEANUP_SPACE_CACHE_V1, &fs_info->flags);
+		ret = cleanup_free_space_cache_v1(fs_info, trans);
+		if (ret) {
+			btrfs_abort_transaction(trans, ret);
+			btrfs_end_transaction(trans);
+			goto out;
+		}
 	}
 
-	if (ret) {
-		if (release_metadata)
-			btrfs_delalloc_release_metadata(BTRFS_I(inode),
-					inode->i_size, true);
-#ifdef DEBUG
-		btrfs_err(fs_info,
-			  "failed to write free ino cache for root %llu",
-			  root->root_key.objectid);
-#endif
-	}
+	ret = btrfs_commit_transaction(trans);
+out:
+	clear_bit(BTRFS_FS_CLEANUP_SPACE_CACHE_V1, &fs_info->flags);
 
 	return ret;
 }
diff --git a/fs/btrfs/free-space-cache.h b/fs/btrfs/free-space-cache.h
index 2e0a8077aa74..6d419ba53e95 100644
--- a/fs/btrfs/free-space-cache.h
+++ b/fs/btrfs/free-space-cache.h
@@ -22,6 +22,7 @@ enum btrfs_trim_state {
 
 struct btrfs_free_space {
 	struct rb_node offset_index;
+	struct rb_node bytes_index;
 	u64 offset;
 	u64 bytes;
 	u64 max_extent_size;
@@ -45,6 +46,7 @@ static inline bool btrfs_free_space_trimming_bitmap(
 struct btrfs_free_space_ctl {
 	spinlock_t tree_lock;
 	struct rb_root free_space_offset;
+	struct rb_root_cached free_space_bytes;
 	u64 free_space;
 	int extents_thresh;
 	int free_extents;
@@ -54,13 +56,12 @@ struct btrfs_free_space_ctl {
 	s32 discardable_extents[BTRFS_STAT_NR_ENTRIES];
 	s64 discardable_bytes[BTRFS_STAT_NR_ENTRIES];
 	const struct btrfs_free_space_op *op;
-	void *private;
+	struct btrfs_block_group *block_group;
 	struct mutex cache_writeout_mutex;
 	struct list_head trimming_ranges;
 };
 
 struct btrfs_free_space_op {
-	void (*recalc_thresholds)(struct btrfs_free_space_ctl *ctl);
 	bool (*use_bitmap)(struct btrfs_free_space_ctl *ctl,
 			   struct btrfs_free_space *info);
 };
@@ -76,7 +77,6 @@ struct btrfs_io_ctl {
 	int num_pages;
 	int entries;
 	int bitmaps;
-	unsigned check_crcs:1;
 };
 
 struct inode *lookup_free_space_inode(struct btrfs_block_group *block_group,
@@ -84,6 +84,9 @@ struct inode *lookup_free_space_inode(struct btrfs_block_group *block_group,
 int create_free_space_inode(struct btrfs_trans_handle *trans,
 			    struct btrfs_block_group *block_group,
 			    struct btrfs_path *path);
+int btrfs_remove_free_space_inode(struct btrfs_trans_handle *trans,
+				  struct inode *inode,
+				  struct btrfs_block_group *block_group);
 
 int btrfs_check_trunc_cache_free_space(struct btrfs_fs_info *fs_info,
 				       struct btrfs_block_rsv *rsv);
@@ -97,36 +100,24 @@ int btrfs_wait_cache_io(struct btrfs_trans_handle *trans,
 int btrfs_write_out_cache(struct btrfs_trans_handle *trans,
 			  struct btrfs_block_group *block_group,
 			  struct btrfs_path *path);
-struct inode *lookup_free_ino_inode(struct btrfs_root *root,
-				    struct btrfs_path *path);
-int create_free_ino_inode(struct btrfs_root *root,
-			  struct btrfs_trans_handle *trans,
-			  struct btrfs_path *path);
-int load_free_ino_cache(struct btrfs_fs_info *fs_info,
-			struct btrfs_root *root);
-int btrfs_write_out_ino_cache(struct btrfs_root *root,
-			      struct btrfs_trans_handle *trans,
-			      struct btrfs_path *path,
-			      struct inode *inode);
 
-void btrfs_init_free_space_ctl(struct btrfs_block_group *block_group);
-int __btrfs_add_free_space(struct btrfs_fs_info *fs_info,
-			   struct btrfs_free_space_ctl *ctl,
-			   u64 bytenr, u64 size,
-			   enum btrfs_trim_state trim_state);
+void btrfs_init_free_space_ctl(struct btrfs_block_group *block_group,
+			       struct btrfs_free_space_ctl *ctl);
+int __btrfs_add_free_space(struct btrfs_block_group *block_group, u64 bytenr,
+			   u64 size, enum btrfs_trim_state trim_state);
 int btrfs_add_free_space(struct btrfs_block_group *block_group,
 			 u64 bytenr, u64 size);
+int btrfs_add_free_space_unused(struct btrfs_block_group *block_group,
+				u64 bytenr, u64 size);
 int btrfs_add_free_space_async_trimmed(struct btrfs_block_group *block_group,
 				       u64 bytenr, u64 size);
 int btrfs_remove_free_space(struct btrfs_block_group *block_group,
 			    u64 bytenr, u64 size);
-void __btrfs_remove_free_space_cache(struct btrfs_free_space_ctl *ctl);
 void btrfs_remove_free_space_cache(struct btrfs_block_group *block_group);
 bool btrfs_is_free_space_trimmed(struct btrfs_block_group *block_group);
 u64 btrfs_find_space_for_alloc(struct btrfs_block_group *block_group,
 			       u64 offset, u64 bytes, u64 empty_size,
 			       u64 *max_extent_size);
-u64 btrfs_find_ino_for_alloc(struct btrfs_root *fs_root);
 void btrfs_dump_free_space(struct btrfs_block_group *block_group,
 			   u64 bytes);
 int btrfs_find_space_cluster(struct btrfs_block_group *block_group,
@@ -136,7 +127,7 @@ void btrfs_init_free_cluster(struct btrfs_free_cluster *cluster);
 u64 btrfs_alloc_from_cluster(struct btrfs_block_group *block_group,
 			     struct btrfs_free_cluster *cluster, u64 bytes,
 			     u64 min_start, u64 *max_extent_size);
-int btrfs_return_cluster_to_free_space(
+void btrfs_return_cluster_to_free_space(
 			       struct btrfs_block_group *block_group,
 			       struct btrfs_free_cluster *cluster);
 int btrfs_trim_block_group(struct btrfs_block_group *block_group,
@@ -148,6 +139,8 @@ int btrfs_trim_block_group_bitmaps(struct btrfs_block_group *block_group,
 				   u64 *trimmed, u64 start, u64 end, u64 minlen,
 				   u64 maxlen, bool async);
 
+bool btrfs_free_space_cache_v1_active(struct btrfs_fs_info *fs_info);
+int btrfs_set_free_space_cache_v1_active(struct btrfs_fs_info *fs_info, bool active);
 /* Support functions for running our sanity tests */
 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
 int test_add_free_space_entry(struct btrfs_block_group *cache,
diff --git a/fs/btrfs/free-space-tree.c b/fs/btrfs/free-space-tree.c
index 258cb3fae17a..367bcfcf68f5 100644
--- a/fs/btrfs/free-space-tree.c
+++ b/fs/btrfs/free-space-tree.c
@@ -16,12 +16,30 @@ static int __add_block_group_free_space(struct btrfs_trans_handle *trans,
 					struct btrfs_block_group *block_group,
 					struct btrfs_path *path);
 
+static struct btrfs_root *btrfs_free_space_root(
+				struct btrfs_block_group *block_group)
+{
+	struct btrfs_key key = {
+		.objectid = BTRFS_FREE_SPACE_TREE_OBJECTID,
+		.type = BTRFS_ROOT_ITEM_KEY,
+		.offset = 0,
+	};
+
+	if (btrfs_fs_incompat(block_group->fs_info, EXTENT_TREE_V2))
+		key.offset = block_group->global_root_id;
+	return btrfs_global_root(block_group->fs_info, &key);
+}
+
 void set_free_space_tree_thresholds(struct btrfs_block_group *cache)
 {
 	u32 bitmap_range;
 	size_t bitmap_size;
 	u64 num_bitmaps, total_bitmap_size;
 
+	if (WARN_ON(cache->length == 0))
+		btrfs_warn(cache->fs_info, "block group %llu length is zero",
+			   cache->start);
+
 	/*
 	 * We convert to bitmaps when the disk space required for using extents
 	 * exceeds that required for using bitmaps.
@@ -47,7 +65,7 @@ static int add_new_free_space_info(struct btrfs_trans_handle *trans,
 				   struct btrfs_block_group *block_group,
 				   struct btrfs_path *path)
 {
-	struct btrfs_root *root = trans->fs_info->free_space_root;
+	struct btrfs_root *root = btrfs_free_space_root(block_group);
 	struct btrfs_free_space_info *info;
 	struct btrfs_key key;
 	struct extent_buffer *leaf;
@@ -81,7 +99,7 @@ struct btrfs_free_space_info *search_free_space_info(
 		struct btrfs_path *path, int cow)
 {
 	struct btrfs_fs_info *fs_info = block_group->fs_info;
-	struct btrfs_root *root = fs_info->free_space_root;
+	struct btrfs_root *root = btrfs_free_space_root(block_group);
 	struct btrfs_key key;
 	int ret;
 
@@ -132,9 +150,10 @@ static int btrfs_search_prev_slot(struct btrfs_trans_handle *trans,
 	return 0;
 }
 
-static inline u32 free_space_bitmap_size(u64 size, u32 sectorsize)
+static inline u32 free_space_bitmap_size(const struct btrfs_fs_info *fs_info,
+					 u64 size)
 {
-	return DIV_ROUND_UP((u32)div_u64(size, sectorsize), BITS_PER_BYTE);
+	return DIV_ROUND_UP(size >> fs_info->sectorsize_bits, BITS_PER_BYTE);
 }
 
 static unsigned long *alloc_bitmap(u32 bitmap_size)
@@ -183,7 +202,7 @@ int convert_free_space_to_bitmaps(struct btrfs_trans_handle *trans,
 				  struct btrfs_path *path)
 {
 	struct btrfs_fs_info *fs_info = trans->fs_info;
-	struct btrfs_root *root = fs_info->free_space_root;
+	struct btrfs_root *root = btrfs_free_space_root(block_group);
 	struct btrfs_free_space_info *info;
 	struct btrfs_key key, found_key;
 	struct extent_buffer *leaf;
@@ -196,8 +215,7 @@ int convert_free_space_to_bitmaps(struct btrfs_trans_handle *trans,
 	int done = 0, nr;
 	int ret;
 
-	bitmap_size = free_space_bitmap_size(block_group->length,
-					     fs_info->sectorsize);
+	bitmap_size = free_space_bitmap_size(fs_info, block_group->length);
 	bitmap = alloc_bitmap(bitmap_size);
 	if (!bitmap) {
 		ret = -ENOMEM;
@@ -286,8 +304,7 @@ int convert_free_space_to_bitmaps(struct btrfs_trans_handle *trans,
 		u32 data_size;
 
 		extent_size = min(end - i, bitmap_range);
-		data_size = free_space_bitmap_size(extent_size,
-						   fs_info->sectorsize);
+		data_size = free_space_bitmap_size(fs_info, extent_size);
 
 		key.objectid = i;
 		key.type = BTRFS_FREE_SPACE_BITMAP_KEY;
@@ -323,7 +340,7 @@ int convert_free_space_to_extents(struct btrfs_trans_handle *trans,
 				  struct btrfs_path *path)
 {
 	struct btrfs_fs_info *fs_info = trans->fs_info;
-	struct btrfs_root *root = fs_info->free_space_root;
+	struct btrfs_root *root = btrfs_free_space_root(block_group);
 	struct btrfs_free_space_info *info;
 	struct btrfs_key key, found_key;
 	struct extent_buffer *leaf;
@@ -335,8 +352,7 @@ int convert_free_space_to_extents(struct btrfs_trans_handle *trans,
 	int done = 0, nr;
 	int ret;
 
-	bitmap_size = free_space_bitmap_size(block_group->length,
-					     fs_info->sectorsize);
+	bitmap_size = free_space_bitmap_size(fs_info, block_group->length);
 	bitmap = alloc_bitmap(bitmap_size);
 	if (!bitmap) {
 		ret = -ENOMEM;
@@ -379,8 +395,8 @@ int convert_free_space_to_extents(struct btrfs_trans_handle *trans,
 						     fs_info->sectorsize *
 						     BITS_PER_BYTE);
 				bitmap_cursor = ((char *)bitmap) + bitmap_pos;
-				data_size = free_space_bitmap_size(found_key.offset,
-								   fs_info->sectorsize);
+				data_size = free_space_bitmap_size(fs_info,
+								found_key.offset);
 
 				ptr = btrfs_item_ptr_offset(leaf, path->slots[0] - 1);
 				read_extent_buffer(leaf, bitmap_cursor, ptr,
@@ -412,7 +428,7 @@ int convert_free_space_to_extents(struct btrfs_trans_handle *trans,
 	btrfs_mark_buffer_dirty(leaf);
 	btrfs_release_path(path);
 
-	nrbits = div_u64(block_group->length, block_group->fs_info->sectorsize);
+	nrbits = block_group->length >> block_group->fs_info->sectorsize_bits;
 	start_bit = find_next_bit_le(bitmap, nrbits, 0);
 
 	while (start_bit < nrbits) {
@@ -536,8 +552,8 @@ static void free_space_set_bits(struct btrfs_block_group *block_group,
 		end = found_end;
 
 	ptr = btrfs_item_ptr_offset(leaf, path->slots[0]);
-	first = div_u64(*start - found_start, fs_info->sectorsize);
-	last = div_u64(end - found_start, fs_info->sectorsize);
+	first = (*start - found_start) >> fs_info->sectorsize_bits;
+	last = (end - found_start) >> fs_info->sectorsize_bits;
 	if (bit)
 		extent_buffer_bitmap_set(leaf, ptr, first, last - first);
 	else
@@ -584,7 +600,7 @@ static int modify_free_space_bitmap(struct btrfs_trans_handle *trans,
 				    struct btrfs_path *path,
 				    u64 start, u64 size, int remove)
 {
-	struct btrfs_root *root = block_group->fs_info->free_space_root;
+	struct btrfs_root *root = btrfs_free_space_root(block_group);
 	struct btrfs_key key;
 	u64 end = start + size;
 	u64 cur_start, cur_size;
@@ -697,7 +713,7 @@ static int remove_free_space_extent(struct btrfs_trans_handle *trans,
 				    struct btrfs_path *path,
 				    u64 start, u64 size)
 {
-	struct btrfs_root *root = trans->fs_info->free_space_root;
+	struct btrfs_root *root = btrfs_free_space_root(block_group);
 	struct btrfs_key key;
 	u64 found_start, found_end;
 	u64 end = start + size;
@@ -849,7 +865,7 @@ static int add_free_space_extent(struct btrfs_trans_handle *trans,
 				 struct btrfs_path *path,
 				 u64 start, u64 size)
 {
-	struct btrfs_root *root = trans->fs_info->free_space_root;
+	struct btrfs_root *root = btrfs_free_space_root(block_group);
 	struct btrfs_key key, new_key;
 	u64 found_start, found_end;
 	u64 end = start + size;
@@ -1044,7 +1060,7 @@ out:
 static int populate_free_space_tree(struct btrfs_trans_handle *trans,
 				    struct btrfs_block_group *block_group)
 {
-	struct btrfs_root *extent_root = trans->fs_info->extent_root;
+	struct btrfs_root *extent_root;
 	struct btrfs_path *path, *path2;
 	struct btrfs_key key;
 	u64 start, end;
@@ -1078,6 +1094,7 @@ static int populate_free_space_tree(struct btrfs_trans_handle *trans,
 	key.type = BTRFS_EXTENT_ITEM_KEY;
 	key.offset = 0;
 
+	extent_root = btrfs_extent_root(trans->fs_info, key.objectid);
 	ret = btrfs_search_slot_for_read(extent_root, &key, path, 1, 0);
 	if (ret < 0)
 		goto out_locked;
@@ -1148,15 +1165,20 @@ int btrfs_create_free_space_tree(struct btrfs_fs_info *fs_info)
 		return PTR_ERR(trans);
 
 	set_bit(BTRFS_FS_CREATING_FREE_SPACE_TREE, &fs_info->flags);
+	set_bit(BTRFS_FS_FREE_SPACE_TREE_UNTRUSTED, &fs_info->flags);
 	free_space_root = btrfs_create_tree(trans,
 					    BTRFS_FREE_SPACE_TREE_OBJECTID);
 	if (IS_ERR(free_space_root)) {
 		ret = PTR_ERR(free_space_root);
 		goto abort;
 	}
-	fs_info->free_space_root = free_space_root;
+	ret = btrfs_global_root_insert(free_space_root);
+	if (ret) {
+		btrfs_put_root(free_space_root);
+		goto abort;
+	}
 
-	node = rb_first(&fs_info->block_group_cache_tree);
+	node = rb_first_cached(&fs_info->block_group_cache_tree);
 	while (node) {
 		block_group = rb_entry(node, struct btrfs_block_group,
 				       cache_node);
@@ -1169,11 +1191,18 @@ int btrfs_create_free_space_tree(struct btrfs_fs_info *fs_info)
 	btrfs_set_fs_compat_ro(fs_info, FREE_SPACE_TREE);
 	btrfs_set_fs_compat_ro(fs_info, FREE_SPACE_TREE_VALID);
 	clear_bit(BTRFS_FS_CREATING_FREE_SPACE_TREE, &fs_info->flags);
+	ret = btrfs_commit_transaction(trans);
 
-	return btrfs_commit_transaction(trans);
+	/*
+	 * Now that we've committed the transaction any reading of our commit
+	 * root will be safe, so we can cache from the free space tree now.
+	 */
+	clear_bit(BTRFS_FS_FREE_SPACE_TREE_UNTRUSTED, &fs_info->flags);
+	return ret;
 
 abort:
 	clear_bit(BTRFS_FS_CREATING_FREE_SPACE_TREE, &fs_info->flags);
+	clear_bit(BTRFS_FS_FREE_SPACE_TREE_UNTRUSTED, &fs_info->flags);
 	btrfs_abort_transaction(trans, ret);
 	btrfs_end_transaction(trans);
 	return ret;
@@ -1191,8 +1220,6 @@ static int clear_free_space_tree(struct btrfs_trans_handle *trans,
 	if (!path)
 		return -ENOMEM;
 
-	path->leave_spinning = 1;
-
 	key.objectid = 0;
 	key.type = 0;
 	key.offset = 0;
@@ -1224,7 +1251,12 @@ int btrfs_clear_free_space_tree(struct btrfs_fs_info *fs_info)
 {
 	struct btrfs_trans_handle *trans;
 	struct btrfs_root *tree_root = fs_info->tree_root;
-	struct btrfs_root *free_space_root = fs_info->free_space_root;
+	struct btrfs_key key = {
+		.objectid = BTRFS_FREE_SPACE_TREE_OBJECTID,
+		.type = BTRFS_ROOT_ITEM_KEY,
+		.offset = 0,
+	};
+	struct btrfs_root *free_space_root = btrfs_global_root(fs_info, &key);
 	int ret;
 
 	trans = btrfs_start_transaction(tree_root, 0);
@@ -1233,7 +1265,6 @@ int btrfs_clear_free_space_tree(struct btrfs_fs_info *fs_info)
 
 	btrfs_clear_fs_compat_ro(fs_info, FREE_SPACE_TREE);
 	btrfs_clear_fs_compat_ro(fs_info, FREE_SPACE_TREE_VALID);
-	fs_info->free_space_root = NULL;
 
 	ret = clear_free_space_tree(trans, free_space_root);
 	if (ret)
@@ -1243,17 +1274,16 @@ int btrfs_clear_free_space_tree(struct btrfs_fs_info *fs_info)
 	if (ret)
 		goto abort;
 
+	btrfs_global_root_delete(free_space_root);
 	list_del(&free_space_root->dirty_list);
 
 	btrfs_tree_lock(free_space_root->node);
 	btrfs_clean_tree_block(free_space_root->node);
 	btrfs_tree_unlock(free_space_root->node);
-	btrfs_free_tree_block(trans, free_space_root, free_space_root->node,
-			      0, 1);
+	btrfs_free_tree_block(trans, btrfs_root_id(free_space_root),
+			      free_space_root->node, 0, 1);
 
-	free_extent_buffer(free_space_root->node);
-	free_extent_buffer(free_space_root->commit_root);
-	kfree(free_space_root);
+	btrfs_put_root(free_space_root);
 
 	return btrfs_commit_transaction(trans);
 
@@ -1313,7 +1343,7 @@ out:
 int remove_block_group_free_space(struct btrfs_trans_handle *trans,
 				  struct btrfs_block_group *block_group)
 {
-	struct btrfs_root *root = trans->fs_info->free_space_root;
+	struct btrfs_root *root = btrfs_free_space_root(block_group);
 	struct btrfs_path *path;
 	struct btrfs_key key, found_key;
 	struct extent_buffer *leaf;
@@ -1404,7 +1434,7 @@ static int load_free_space_bitmaps(struct btrfs_caching_control *caching_ctl,
 
 	block_group = caching_ctl->block_group;
 	fs_info = block_group->fs_info;
-	root = fs_info->free_space_root;
+	root = btrfs_free_space_root(block_group);
 
 	end = block_group->start + block_group->length;
 
@@ -1423,8 +1453,6 @@ static int load_free_space_bitmaps(struct btrfs_caching_control *caching_ctl,
 		ASSERT(key.type == BTRFS_FREE_SPACE_BITMAP_KEY);
 		ASSERT(key.objectid < end && key.objectid + key.offset <= end);
 
-		caching_ctl->progress = key.objectid;
-
 		offset = key.objectid;
 		while (offset < key.objectid + key.offset) {
 			bit = free_space_test_bit(block_group, path, offset);
@@ -1460,8 +1488,6 @@ static int load_free_space_bitmaps(struct btrfs_caching_control *caching_ctl,
 		goto out;
 	}
 
-	caching_ctl->progress = (u64)-1;
-
 	ret = 0;
 out:
 	return ret;
@@ -1482,7 +1508,7 @@ static int load_free_space_extents(struct btrfs_caching_control *caching_ctl,
 
 	block_group = caching_ctl->block_group;
 	fs_info = block_group->fs_info;
-	root = fs_info->free_space_root;
+	root = btrfs_free_space_root(block_group);
 
 	end = block_group->start + block_group->length;
 
@@ -1501,8 +1527,6 @@ static int load_free_space_extents(struct btrfs_caching_control *caching_ctl,
 		ASSERT(key.type == BTRFS_FREE_SPACE_EXTENT_KEY);
 		ASSERT(key.objectid < end && key.objectid + key.offset <= end);
 
-		caching_ctl->progress = key.objectid;
-
 		total_found += add_new_free_space(block_group, key.objectid,
 						  key.objectid + key.offset);
 		if (total_found > CACHING_CTL_WAKE_UP) {
@@ -1522,8 +1546,6 @@ static int load_free_space_extents(struct btrfs_caching_control *caching_ctl,
 		goto out;
 	}
 
-	caching_ctl->progress = (u64)-1;
-
 	ret = 0;
 out:
 	return ret;
diff --git a/fs/btrfs/inode-item.c b/fs/btrfs/inode-item.c
index 668701832845..0eeb5ea87894 100644
--- a/fs/btrfs/inode-item.c
+++ b/fs/btrfs/inode-item.c
@@ -4,6 +4,7 @@
  */
 
 #include "ctree.h"
+#include "inode-item.h"
 #include "disk-io.h"
 #include "transaction.h"
 #include "print-tree.h"
@@ -19,7 +20,7 @@ struct btrfs_inode_ref *btrfs_find_name_in_backref(struct extent_buffer *leaf,
 	u32 cur_offset = 0;
 	int len;
 
-	item_size = btrfs_item_size_nr(leaf, slot);
+	item_size = btrfs_item_size(leaf, slot);
 	ptr = btrfs_item_ptr_offset(leaf, slot);
 	while (cur_offset < item_size) {
 		ref = (struct btrfs_inode_ref *)(ptr + cur_offset);
@@ -45,7 +46,7 @@ struct btrfs_inode_extref *btrfs_find_name_in_ext_backref(
 	u32 cur_offset = 0;
 	int ref_name_len;
 
-	item_size = btrfs_item_size_nr(leaf, slot);
+	item_size = btrfs_item_size(leaf, slot);
 	ptr = btrfs_item_ptr_offset(leaf, slot);
 
 	/*
@@ -119,8 +120,6 @@ static int btrfs_del_inode_extref(struct btrfs_trans_handle *trans,
 	if (!path)
 		return -ENOMEM;
 
-	path->leave_spinning = 1;
-
 	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
 	if (ret > 0)
 		ret = -ENOENT;
@@ -141,7 +140,7 @@ static int btrfs_del_inode_extref(struct btrfs_trans_handle *trans,
 	}
 
 	leaf = path->nodes[0];
-	item_size = btrfs_item_size_nr(leaf, path->slots[0]);
+	item_size = btrfs_item_size(leaf, path->slots[0]);
 	if (index)
 		*index = btrfs_inode_extref_index(leaf, extref);
 
@@ -193,8 +192,6 @@ int btrfs_del_inode_ref(struct btrfs_trans_handle *trans,
 	if (!path)
 		return -ENOMEM;
 
-	path->leave_spinning = 1;
-
 	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
 	if (ret > 0) {
 		ret = -ENOENT;
@@ -212,7 +209,7 @@ int btrfs_del_inode_ref(struct btrfs_trans_handle *trans,
 		goto out;
 	}
 	leaf = path->nodes[0];
-	item_size = btrfs_item_size_nr(leaf, path->slots[0]);
+	item_size = btrfs_item_size(leaf, path->slots[0]);
 
 	if (index)
 		*index = btrfs_inode_ref_index(leaf, ref);
@@ -260,7 +257,6 @@ static int btrfs_insert_inode_extref(struct btrfs_trans_handle *trans,
 	struct btrfs_path *path;
 	struct btrfs_key key;
 	struct extent_buffer *leaf;
-	struct btrfs_item *item;
 
 	key.objectid = inode_objectid;
 	key.type = BTRFS_INODE_EXTREF_KEY;
@@ -270,7 +266,6 @@ static int btrfs_insert_inode_extref(struct btrfs_trans_handle *trans,
 	if (!path)
 		return -ENOMEM;
 
-	path->leave_spinning = 1;
 	ret = btrfs_insert_empty_item(trans, root, path, &key,
 				      ins_len);
 	if (ret == -EEXIST) {
@@ -287,9 +282,8 @@ static int btrfs_insert_inode_extref(struct btrfs_trans_handle *trans,
 		goto out;
 
 	leaf = path->nodes[0];
-	item = btrfs_item_nr(path->slots[0]);
 	ptr = (unsigned long)btrfs_item_ptr(leaf, path->slots[0], char);
-	ptr += btrfs_item_size(leaf, item) - ins_len;
+	ptr += btrfs_item_size(leaf, path->slots[0]) - ins_len;
 	extref = (struct btrfs_inode_extref *)ptr;
 
 	btrfs_set_inode_extref_name_len(path->nodes[0], extref, name_len);
@@ -327,7 +321,6 @@ int btrfs_insert_inode_ref(struct btrfs_trans_handle *trans,
 	if (!path)
 		return -ENOMEM;
 
-	path->leave_spinning = 1;
 	path->skip_release_on_error = 1;
 	ret = btrfs_insert_empty_item(trans, root, path, &key,
 				      ins_len);
@@ -338,7 +331,7 @@ int btrfs_insert_inode_ref(struct btrfs_trans_handle *trans,
 		if (ref)
 			goto out;
 
-		old_size = btrfs_item_size_nr(path->nodes[0], path->slots[0]);
+		old_size = btrfs_item_size(path->nodes[0], path->slots[0]);
 		btrfs_extend_item(path, ins_len);
 		ref = btrfs_item_ptr(path->nodes[0], path->slots[0],
 				     struct btrfs_inode_ref);
@@ -425,3 +418,332 @@ int btrfs_lookup_inode(struct btrfs_trans_handle *trans, struct btrfs_root
 	}
 	return ret;
 }
+
+static inline void btrfs_trace_truncate(struct btrfs_inode *inode,
+					struct extent_buffer *leaf,
+					struct btrfs_file_extent_item *fi,
+					u64 offset, int extent_type, int slot)
+{
+	if (!inode)
+		return;
+	if (extent_type == BTRFS_FILE_EXTENT_INLINE)
+		trace_btrfs_truncate_show_fi_inline(inode, leaf, fi, slot,
+						    offset);
+	else
+		trace_btrfs_truncate_show_fi_regular(inode, leaf, fi, offset);
+}
+
+/*
+ * Remove inode items from a given root.
+ *
+ * @trans:		A transaction handle.
+ * @root:		The root from which to remove items.
+ * @inode:		The inode whose items we want to remove.
+ * @control:		The btrfs_truncate_control to control how and what we
+ *			are truncating.
+ *
+ * Remove all keys associated with the inode from the given root that have a key
+ * with a type greater than or equals to @min_type. When @min_type has a value of
+ * BTRFS_EXTENT_DATA_KEY, only remove file extent items that have an offset value
+ * greater than or equals to @new_size. If a file extent item that starts before
+ * @new_size and ends after it is found, its length is adjusted.
+ *
+ * Returns: 0 on success, < 0 on error and NEED_TRUNCATE_BLOCK when @min_type is
+ * BTRFS_EXTENT_DATA_KEY and the caller must truncate the last block.
+ */
+int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans,
+			       struct btrfs_root *root,
+			       struct btrfs_truncate_control *control)
+{
+	struct btrfs_fs_info *fs_info = root->fs_info;
+	struct btrfs_path *path;
+	struct extent_buffer *leaf;
+	struct btrfs_file_extent_item *fi;
+	struct btrfs_key key;
+	struct btrfs_key found_key;
+	u64 new_size = control->new_size;
+	u64 extent_num_bytes = 0;
+	u64 extent_offset = 0;
+	u64 item_end = 0;
+	u32 found_type = (u8)-1;
+	int del_item;
+	int pending_del_nr = 0;
+	int pending_del_slot = 0;
+	int extent_type = -1;
+	int ret;
+	u64 bytes_deleted = 0;
+	bool be_nice = false;
+
+	ASSERT(control->inode || !control->clear_extent_range);
+	ASSERT(new_size == 0 || control->min_type == BTRFS_EXTENT_DATA_KEY);
+
+	control->last_size = new_size;
+	control->sub_bytes = 0;
+
+	/*
+	 * For shareable roots we want to back off from time to time, this turns
+	 * out to be subvolume roots, reloc roots, and data reloc roots.
+	 */
+	if (test_bit(BTRFS_ROOT_SHAREABLE, &root->state))
+		be_nice = true;
+
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+	path->reada = READA_BACK;
+
+	key.objectid = control->ino;
+	key.offset = (u64)-1;
+	key.type = (u8)-1;
+
+search_again:
+	/*
+	 * With a 16K leaf size and 128MiB extents, you can actually queue up a
+	 * huge file in a single leaf.  Most of the time that bytes_deleted is
+	 * > 0, it will be huge by the time we get here
+	 */
+	if (be_nice && bytes_deleted > SZ_32M &&
+	    btrfs_should_end_transaction(trans)) {
+		ret = -EAGAIN;
+		goto out;
+	}
+
+	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
+	if (ret < 0)
+		goto out;
+
+	if (ret > 0) {
+		ret = 0;
+		/* There are no items in the tree for us to truncate, we're done */
+		if (path->slots[0] == 0)
+			goto out;
+		path->slots[0]--;
+	}
+
+	while (1) {
+		u64 clear_start = 0, clear_len = 0, extent_start = 0;
+		bool should_throttle = false;
+
+		fi = NULL;
+		leaf = path->nodes[0];
+		btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
+		found_type = found_key.type;
+
+		if (found_key.objectid != control->ino)
+			break;
+
+		if (found_type < control->min_type)
+			break;
+
+		item_end = found_key.offset;
+		if (found_type == BTRFS_EXTENT_DATA_KEY) {
+			fi = btrfs_item_ptr(leaf, path->slots[0],
+					    struct btrfs_file_extent_item);
+			extent_type = btrfs_file_extent_type(leaf, fi);
+			if (extent_type != BTRFS_FILE_EXTENT_INLINE)
+				item_end +=
+				    btrfs_file_extent_num_bytes(leaf, fi);
+			else if (extent_type == BTRFS_FILE_EXTENT_INLINE)
+				item_end += btrfs_file_extent_ram_bytes(leaf, fi);
+
+			btrfs_trace_truncate(control->inode, leaf, fi,
+					     found_key.offset, extent_type,
+					     path->slots[0]);
+			item_end--;
+		}
+		if (found_type > control->min_type) {
+			del_item = 1;
+		} else {
+			if (item_end < new_size)
+				break;
+			if (found_key.offset >= new_size)
+				del_item = 1;
+			else
+				del_item = 0;
+		}
+
+		/* FIXME, shrink the extent if the ref count is only 1 */
+		if (found_type != BTRFS_EXTENT_DATA_KEY)
+			goto delete;
+
+		control->extents_found++;
+
+		if (extent_type != BTRFS_FILE_EXTENT_INLINE) {
+			u64 num_dec;
+
+			clear_start = found_key.offset;
+			extent_start = btrfs_file_extent_disk_bytenr(leaf, fi);
+			if (!del_item) {
+				u64 orig_num_bytes =
+					btrfs_file_extent_num_bytes(leaf, fi);
+				extent_num_bytes = ALIGN(new_size -
+						found_key.offset,
+						fs_info->sectorsize);
+				clear_start = ALIGN(new_size, fs_info->sectorsize);
+
+				btrfs_set_file_extent_num_bytes(leaf, fi,
+							 extent_num_bytes);
+				num_dec = (orig_num_bytes - extent_num_bytes);
+				if (extent_start != 0)
+					control->sub_bytes += num_dec;
+				btrfs_mark_buffer_dirty(leaf);
+			} else {
+				extent_num_bytes =
+					btrfs_file_extent_disk_num_bytes(leaf, fi);
+				extent_offset = found_key.offset -
+					btrfs_file_extent_offset(leaf, fi);
+
+				/* FIXME blocksize != 4096 */
+				num_dec = btrfs_file_extent_num_bytes(leaf, fi);
+				if (extent_start != 0)
+					control->sub_bytes += num_dec;
+			}
+			clear_len = num_dec;
+		} else if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
+			/*
+			 * We can't truncate inline items that have had
+			 * special encodings
+			 */
+			if (!del_item &&
+			    btrfs_file_extent_encryption(leaf, fi) == 0 &&
+			    btrfs_file_extent_other_encoding(leaf, fi) == 0 &&
+			    btrfs_file_extent_compression(leaf, fi) == 0) {
+				u32 size = (u32)(new_size - found_key.offset);
+
+				btrfs_set_file_extent_ram_bytes(leaf, fi, size);
+				size = btrfs_file_extent_calc_inline_size(size);
+				btrfs_truncate_item(path, size, 1);
+			} else if (!del_item) {
+				/*
+				 * We have to bail so the last_size is set to
+				 * just before this extent.
+				 */
+				ret = BTRFS_NEED_TRUNCATE_BLOCK;
+				break;
+			} else {
+				/*
+				 * Inline extents are special, we just treat
+				 * them as a full sector worth in the file
+				 * extent tree just for simplicity sake.
+				 */
+				clear_len = fs_info->sectorsize;
+			}
+
+			control->sub_bytes += item_end + 1 - new_size;
+		}
+delete:
+		/*
+		 * We only want to clear the file extent range if we're
+		 * modifying the actual inode's mapping, which is just the
+		 * normal truncate path.
+		 */
+		if (control->clear_extent_range) {
+			ret = btrfs_inode_clear_file_extent_range(control->inode,
+						  clear_start, clear_len);
+			if (ret) {
+				btrfs_abort_transaction(trans, ret);
+				break;
+			}
+		}
+
+		if (del_item) {
+			ASSERT(!pending_del_nr ||
+			       ((path->slots[0] + 1) == pending_del_slot));
+
+			control->last_size = found_key.offset;
+			if (!pending_del_nr) {
+				/* No pending yet, add ourselves */
+				pending_del_slot = path->slots[0];
+				pending_del_nr = 1;
+			} else if (pending_del_nr &&
+				   path->slots[0] + 1 == pending_del_slot) {
+				/* Hop on the pending chunk */
+				pending_del_nr++;
+				pending_del_slot = path->slots[0];
+			}
+		} else {
+			control->last_size = new_size;
+			break;
+		}
+
+		if (del_item && extent_start != 0 && !control->skip_ref_updates) {
+			struct btrfs_ref ref = { 0 };
+
+			bytes_deleted += extent_num_bytes;
+
+			btrfs_init_generic_ref(&ref, BTRFS_DROP_DELAYED_REF,
+					extent_start, extent_num_bytes, 0);
+			btrfs_init_data_ref(&ref, btrfs_header_owner(leaf),
+					control->ino, extent_offset,
+					root->root_key.objectid, false);
+			ret = btrfs_free_extent(trans, &ref);
+			if (ret) {
+				btrfs_abort_transaction(trans, ret);
+				break;
+			}
+			if (be_nice) {
+				if (btrfs_should_throttle_delayed_refs(trans))
+					should_throttle = true;
+			}
+		}
+
+		if (found_type == BTRFS_INODE_ITEM_KEY)
+			break;
+
+		if (path->slots[0] == 0 ||
+		    path->slots[0] != pending_del_slot ||
+		    should_throttle) {
+			if (pending_del_nr) {
+				ret = btrfs_del_items(trans, root, path,
+						pending_del_slot,
+						pending_del_nr);
+				if (ret) {
+					btrfs_abort_transaction(trans, ret);
+					break;
+				}
+				pending_del_nr = 0;
+			}
+			btrfs_release_path(path);
+
+			/*
+			 * We can generate a lot of delayed refs, so we need to
+			 * throttle every once and a while and make sure we're
+			 * adding enough space to keep up with the work we are
+			 * generating.  Since we hold a transaction here we
+			 * can't flush, and we don't want to FLUSH_LIMIT because
+			 * we could have generated too many delayed refs to
+			 * actually allocate, so just bail if we're short and
+			 * let the normal reservation dance happen higher up.
+			 */
+			if (should_throttle) {
+				ret = btrfs_delayed_refs_rsv_refill(fs_info,
+							BTRFS_RESERVE_NO_FLUSH);
+				if (ret) {
+					ret = -EAGAIN;
+					break;
+				}
+			}
+			goto search_again;
+		} else {
+			path->slots[0]--;
+		}
+	}
+out:
+	if (ret >= 0 && pending_del_nr) {
+		int err;
+
+		err = btrfs_del_items(trans, root, path, pending_del_slot,
+				      pending_del_nr);
+		if (err) {
+			btrfs_abort_transaction(trans, err);
+			ret = err;
+		}
+	}
+
+	ASSERT(control->last_size >= new_size);
+	if (!ret && control->last_size > new_size)
+		control->last_size = new_size;
+
+	btrfs_free_path(path);
+	return ret;
+}
diff --git a/fs/btrfs/inode-item.h b/fs/btrfs/inode-item.h
new file mode 100644
index 000000000000..a8fc16d0147f
--- /dev/null
+++ b/fs/btrfs/inode-item.h
@@ -0,0 +1,96 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+
+#ifndef BTRFS_INODE_ITEM_H
+#define BTRFS_INODE_ITEM_H
+
+#include <linux/types.h>
+
+struct btrfs_trans_handle;
+struct btrfs_root;
+struct btrfs_path;
+struct btrfs_key;
+struct btrfs_inode_extref;
+struct btrfs_inode;
+struct extent_buffer;
+
+/*
+ * Return this if we need to call truncate_block for the last bit of the
+ * truncate.
+ */
+#define BTRFS_NEED_TRUNCATE_BLOCK		1
+
+struct btrfs_truncate_control {
+	/*
+	 * IN: the inode we're operating on, this can be NULL if
+	 * ->clear_extent_range is false.
+	 */
+	struct btrfs_inode *inode;
+
+	/* IN: the size we're truncating to. */
+	u64 new_size;
+
+	/* OUT: the number of extents truncated. */
+	u64 extents_found;
+
+	/* OUT: the last size we truncated this inode to. */
+	u64 last_size;
+
+	/* OUT: the number of bytes to sub from this inode. */
+	u64 sub_bytes;
+
+	/* IN: the ino we are truncating. */
+	u64 ino;
+
+	/*
+	 * IN: minimum key type to remove.  All key types with this type are
+	 * removed only if their offset >= new_size.
+	 */
+	u32 min_type;
+
+	/*
+	 * IN: true if we don't want to do extent reference updates for any file
+	 * extents we drop.
+	 */
+	bool skip_ref_updates;
+
+	/*
+	 * IN: true if we need to clear the file extent range for the inode as
+	 * we drop the file extent items.
+	 */
+	bool clear_extent_range;
+};
+
+int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans,
+			       struct btrfs_root *root,
+			       struct btrfs_truncate_control *control);
+int btrfs_insert_inode_ref(struct btrfs_trans_handle *trans,
+			   struct btrfs_root *root,
+			   const char *name, int name_len,
+			   u64 inode_objectid, u64 ref_objectid, u64 index);
+int btrfs_del_inode_ref(struct btrfs_trans_handle *trans,
+			   struct btrfs_root *root,
+			   const char *name, int name_len,
+			   u64 inode_objectid, u64 ref_objectid, u64 *index);
+int btrfs_insert_empty_inode(struct btrfs_trans_handle *trans,
+			     struct btrfs_root *root,
+			     struct btrfs_path *path, u64 objectid);
+int btrfs_lookup_inode(struct btrfs_trans_handle *trans, struct btrfs_root
+		       *root, struct btrfs_path *path,
+		       struct btrfs_key *location, int mod);
+
+struct btrfs_inode_extref *btrfs_lookup_inode_extref(
+			  struct btrfs_trans_handle *trans,
+			  struct btrfs_root *root,
+			  struct btrfs_path *path,
+			  const char *name, int name_len,
+			  u64 inode_objectid, u64 ref_objectid, int ins_len,
+			  int cow);
+
+struct btrfs_inode_ref *btrfs_find_name_in_backref(struct extent_buffer *leaf,
+						   int slot, const char *name,
+						   int name_len);
+struct btrfs_inode_extref *btrfs_find_name_in_ext_backref(
+		struct extent_buffer *leaf, int slot, u64 ref_objectid,
+		const char *name, int name_len);
+
+#endif
diff --git a/fs/btrfs/inode-map.c b/fs/btrfs/inode-map.c
deleted file mode 100644
index d5c9c69d8263..000000000000
--- a/fs/btrfs/inode-map.c
+++ /dev/null
@@ -1,581 +0,0 @@
-// SPDX-License-Identifier: GPL-2.0
-/*
- * Copyright (C) 2007 Oracle.  All rights reserved.
- */
-
-#include <linux/kthread.h>
-#include <linux/pagemap.h>
-
-#include "ctree.h"
-#include "disk-io.h"
-#include "free-space-cache.h"
-#include "inode-map.h"
-#include "transaction.h"
-#include "delalloc-space.h"
-
-static void fail_caching_thread(struct btrfs_root *root)
-{
-	struct btrfs_fs_info *fs_info = root->fs_info;
-
-	btrfs_warn(fs_info, "failed to start inode caching task");
-	btrfs_clear_pending_and_info(fs_info, INODE_MAP_CACHE,
-				     "disabling inode map caching");
-	spin_lock(&root->ino_cache_lock);
-	root->ino_cache_state = BTRFS_CACHE_ERROR;
-	spin_unlock(&root->ino_cache_lock);
-	wake_up(&root->ino_cache_wait);
-}
-
-static int caching_kthread(void *data)
-{
-	struct btrfs_root *root = data;
-	struct btrfs_fs_info *fs_info = root->fs_info;
-	struct btrfs_free_space_ctl *ctl = root->free_ino_ctl;
-	struct btrfs_key key;
-	struct btrfs_path *path;
-	struct extent_buffer *leaf;
-	u64 last = (u64)-1;
-	int slot;
-	int ret;
-
-	if (!btrfs_test_opt(fs_info, INODE_MAP_CACHE))
-		return 0;
-
-	path = btrfs_alloc_path();
-	if (!path) {
-		fail_caching_thread(root);
-		return -ENOMEM;
-	}
-
-	/* Since the commit root is read-only, we can safely skip locking. */
-	path->skip_locking = 1;
-	path->search_commit_root = 1;
-	path->reada = READA_FORWARD;
-
-	key.objectid = BTRFS_FIRST_FREE_OBJECTID;
-	key.offset = 0;
-	key.type = BTRFS_INODE_ITEM_KEY;
-again:
-	/* need to make sure the commit_root doesn't disappear */
-	down_read(&fs_info->commit_root_sem);
-
-	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
-	if (ret < 0)
-		goto out;
-
-	while (1) {
-		if (btrfs_fs_closing(fs_info))
-			goto out;
-
-		leaf = path->nodes[0];
-		slot = path->slots[0];
-		if (slot >= btrfs_header_nritems(leaf)) {
-			ret = btrfs_next_leaf(root, path);
-			if (ret < 0)
-				goto out;
-			else if (ret > 0)
-				break;
-
-			if (need_resched() ||
-			    btrfs_transaction_in_commit(fs_info)) {
-				leaf = path->nodes[0];
-
-				if (WARN_ON(btrfs_header_nritems(leaf) == 0))
-					break;
-
-				/*
-				 * Save the key so we can advances forward
-				 * in the next search.
-				 */
-				btrfs_item_key_to_cpu(leaf, &key, 0);
-				btrfs_release_path(path);
-				root->ino_cache_progress = last;
-				up_read(&fs_info->commit_root_sem);
-				schedule_timeout(1);
-				goto again;
-			} else
-				continue;
-		}
-
-		btrfs_item_key_to_cpu(leaf, &key, slot);
-
-		if (key.type != BTRFS_INODE_ITEM_KEY)
-			goto next;
-
-		if (key.objectid >= root->highest_objectid)
-			break;
-
-		if (last != (u64)-1 && last + 1 != key.objectid) {
-			__btrfs_add_free_space(fs_info, ctl, last + 1,
-					       key.objectid - last - 1, 0);
-			wake_up(&root->ino_cache_wait);
-		}
-
-		last = key.objectid;
-next:
-		path->slots[0]++;
-	}
-
-	if (last < root->highest_objectid - 1) {
-		__btrfs_add_free_space(fs_info, ctl, last + 1,
-				       root->highest_objectid - last - 1, 0);
-	}
-
-	spin_lock(&root->ino_cache_lock);
-	root->ino_cache_state = BTRFS_CACHE_FINISHED;
-	spin_unlock(&root->ino_cache_lock);
-
-	root->ino_cache_progress = (u64)-1;
-	btrfs_unpin_free_ino(root);
-out:
-	wake_up(&root->ino_cache_wait);
-	up_read(&fs_info->commit_root_sem);
-
-	btrfs_free_path(path);
-
-	return ret;
-}
-
-static void start_caching(struct btrfs_root *root)
-{
-	struct btrfs_fs_info *fs_info = root->fs_info;
-	struct btrfs_free_space_ctl *ctl = root->free_ino_ctl;
-	struct task_struct *tsk;
-	int ret;
-	u64 objectid;
-
-	if (!btrfs_test_opt(fs_info, INODE_MAP_CACHE))
-		return;
-
-	spin_lock(&root->ino_cache_lock);
-	if (root->ino_cache_state != BTRFS_CACHE_NO) {
-		spin_unlock(&root->ino_cache_lock);
-		return;
-	}
-
-	root->ino_cache_state = BTRFS_CACHE_STARTED;
-	spin_unlock(&root->ino_cache_lock);
-
-	ret = load_free_ino_cache(fs_info, root);
-	if (ret == 1) {
-		spin_lock(&root->ino_cache_lock);
-		root->ino_cache_state = BTRFS_CACHE_FINISHED;
-		spin_unlock(&root->ino_cache_lock);
-		wake_up(&root->ino_cache_wait);
-		return;
-	}
-
-	/*
-	 * It can be quite time-consuming to fill the cache by searching
-	 * through the extent tree, and this can keep ino allocation path
-	 * waiting. Therefore at start we quickly find out the highest
-	 * inode number and we know we can use inode numbers which fall in
-	 * [highest_ino + 1, BTRFS_LAST_FREE_OBJECTID].
-	 */
-	ret = btrfs_find_free_objectid(root, &objectid);
-	if (!ret && objectid <= BTRFS_LAST_FREE_OBJECTID) {
-		__btrfs_add_free_space(fs_info, ctl, objectid,
-				       BTRFS_LAST_FREE_OBJECTID - objectid + 1,
-				       0);
-		wake_up(&root->ino_cache_wait);
-	}
-
-	tsk = kthread_run(caching_kthread, root, "btrfs-ino-cache-%llu",
-			  root->root_key.objectid);
-	if (IS_ERR(tsk))
-		fail_caching_thread(root);
-}
-
-int btrfs_find_free_ino(struct btrfs_root *root, u64 *objectid)
-{
-	if (!btrfs_test_opt(root->fs_info, INODE_MAP_CACHE))
-		return btrfs_find_free_objectid(root, objectid);
-
-again:
-	*objectid = btrfs_find_ino_for_alloc(root);
-
-	if (*objectid != 0)
-		return 0;
-
-	start_caching(root);
-
-	wait_event(root->ino_cache_wait,
-		   root->ino_cache_state == BTRFS_CACHE_FINISHED ||
-		   root->ino_cache_state == BTRFS_CACHE_ERROR ||
-		   root->free_ino_ctl->free_space > 0);
-
-	if (root->ino_cache_state == BTRFS_CACHE_FINISHED &&
-	    root->free_ino_ctl->free_space == 0)
-		return -ENOSPC;
-	else if (root->ino_cache_state == BTRFS_CACHE_ERROR)
-		return btrfs_find_free_objectid(root, objectid);
-	else
-		goto again;
-}
-
-void btrfs_return_ino(struct btrfs_root *root, u64 objectid)
-{
-	struct btrfs_fs_info *fs_info = root->fs_info;
-	struct btrfs_free_space_ctl *pinned = root->free_ino_pinned;
-
-	if (!btrfs_test_opt(fs_info, INODE_MAP_CACHE))
-		return;
-again:
-	if (root->ino_cache_state == BTRFS_CACHE_FINISHED) {
-		__btrfs_add_free_space(fs_info, pinned, objectid, 1, 0);
-	} else {
-		down_write(&fs_info->commit_root_sem);
-		spin_lock(&root->ino_cache_lock);
-		if (root->ino_cache_state == BTRFS_CACHE_FINISHED) {
-			spin_unlock(&root->ino_cache_lock);
-			up_write(&fs_info->commit_root_sem);
-			goto again;
-		}
-		spin_unlock(&root->ino_cache_lock);
-
-		start_caching(root);
-
-		__btrfs_add_free_space(fs_info, pinned, objectid, 1, 0);
-
-		up_write(&fs_info->commit_root_sem);
-	}
-}
-
-/*
- * When a transaction is committed, we'll move those inode numbers which are
- * smaller than root->ino_cache_progress from pinned tree to free_ino tree, and
- * others will just be dropped, because the commit root we were searching has
- * changed.
- *
- * Must be called with root->fs_info->commit_root_sem held
- */
-void btrfs_unpin_free_ino(struct btrfs_root *root)
-{
-	struct btrfs_free_space_ctl *ctl = root->free_ino_ctl;
-	struct rb_root *rbroot = &root->free_ino_pinned->free_space_offset;
-	spinlock_t *rbroot_lock = &root->free_ino_pinned->tree_lock;
-	struct btrfs_free_space *info;
-	struct rb_node *n;
-	u64 count;
-
-	if (!btrfs_test_opt(root->fs_info, INODE_MAP_CACHE))
-		return;
-
-	while (1) {
-		spin_lock(rbroot_lock);
-		n = rb_first(rbroot);
-		if (!n) {
-			spin_unlock(rbroot_lock);
-			break;
-		}
-
-		info = rb_entry(n, struct btrfs_free_space, offset_index);
-		BUG_ON(info->bitmap); /* Logic error */
-
-		if (info->offset > root->ino_cache_progress)
-			count = 0;
-		else
-			count = min(root->ino_cache_progress - info->offset + 1,
-				    info->bytes);
-
-		rb_erase(&info->offset_index, rbroot);
-		spin_unlock(rbroot_lock);
-		if (count)
-			__btrfs_add_free_space(root->fs_info, ctl,
-					       info->offset, count, 0);
-		kmem_cache_free(btrfs_free_space_cachep, info);
-	}
-}
-
-#define INIT_THRESHOLD	((SZ_32K / 2) / sizeof(struct btrfs_free_space))
-#define INODES_PER_BITMAP (PAGE_SIZE * 8)
-
-/*
- * The goal is to keep the memory used by the free_ino tree won't
- * exceed the memory if we use bitmaps only.
- */
-static void recalculate_thresholds(struct btrfs_free_space_ctl *ctl)
-{
-	struct btrfs_free_space *info;
-	struct rb_node *n;
-	int max_ino;
-	int max_bitmaps;
-
-	n = rb_last(&ctl->free_space_offset);
-	if (!n) {
-		ctl->extents_thresh = INIT_THRESHOLD;
-		return;
-	}
-	info = rb_entry(n, struct btrfs_free_space, offset_index);
-
-	/*
-	 * Find the maximum inode number in the filesystem. Note we
-	 * ignore the fact that this can be a bitmap, because we are
-	 * not doing precise calculation.
-	 */
-	max_ino = info->bytes - 1;
-
-	max_bitmaps = ALIGN(max_ino, INODES_PER_BITMAP) / INODES_PER_BITMAP;
-	if (max_bitmaps <= ctl->total_bitmaps) {
-		ctl->extents_thresh = 0;
-		return;
-	}
-
-	ctl->extents_thresh = (max_bitmaps - ctl->total_bitmaps) *
-				PAGE_SIZE / sizeof(*info);
-}
-
-/*
- * We don't fall back to bitmap, if we are below the extents threshold
- * or this chunk of inode numbers is a big one.
- */
-static bool use_bitmap(struct btrfs_free_space_ctl *ctl,
-		       struct btrfs_free_space *info)
-{
-	if (ctl->free_extents < ctl->extents_thresh ||
-	    info->bytes > INODES_PER_BITMAP / 10)
-		return false;
-
-	return true;
-}
-
-static const struct btrfs_free_space_op free_ino_op = {
-	.recalc_thresholds	= recalculate_thresholds,
-	.use_bitmap		= use_bitmap,
-};
-
-static void pinned_recalc_thresholds(struct btrfs_free_space_ctl *ctl)
-{
-}
-
-static bool pinned_use_bitmap(struct btrfs_free_space_ctl *ctl,
-			      struct btrfs_free_space *info)
-{
-	/*
-	 * We always use extents for two reasons:
-	 *
-	 * - The pinned tree is only used during the process of caching
-	 *   work.
-	 * - Make code simpler. See btrfs_unpin_free_ino().
-	 */
-	return false;
-}
-
-static const struct btrfs_free_space_op pinned_free_ino_op = {
-	.recalc_thresholds	= pinned_recalc_thresholds,
-	.use_bitmap		= pinned_use_bitmap,
-};
-
-void btrfs_init_free_ino_ctl(struct btrfs_root *root)
-{
-	struct btrfs_free_space_ctl *ctl = root->free_ino_ctl;
-	struct btrfs_free_space_ctl *pinned = root->free_ino_pinned;
-
-	spin_lock_init(&ctl->tree_lock);
-	ctl->unit = 1;
-	ctl->start = 0;
-	ctl->private = NULL;
-	ctl->op = &free_ino_op;
-	INIT_LIST_HEAD(&ctl->trimming_ranges);
-	mutex_init(&ctl->cache_writeout_mutex);
-
-	/*
-	 * Initially we allow to use 16K of ram to cache chunks of
-	 * inode numbers before we resort to bitmaps. This is somewhat
-	 * arbitrary, but it will be adjusted in runtime.
-	 */
-	ctl->extents_thresh = INIT_THRESHOLD;
-
-	spin_lock_init(&pinned->tree_lock);
-	pinned->unit = 1;
-	pinned->start = 0;
-	pinned->private = NULL;
-	pinned->extents_thresh = 0;
-	pinned->op = &pinned_free_ino_op;
-}
-
-int btrfs_save_ino_cache(struct btrfs_root *root,
-			 struct btrfs_trans_handle *trans)
-{
-	struct btrfs_fs_info *fs_info = root->fs_info;
-	struct btrfs_free_space_ctl *ctl = root->free_ino_ctl;
-	struct btrfs_path *path;
-	struct inode *inode;
-	struct btrfs_block_rsv *rsv;
-	struct extent_changeset *data_reserved = NULL;
-	u64 num_bytes;
-	u64 alloc_hint = 0;
-	int ret;
-	int prealloc;
-	bool retry = false;
-
-	/* only fs tree and subvol/snap needs ino cache */
-	if (root->root_key.objectid != BTRFS_FS_TREE_OBJECTID &&
-	    (root->root_key.objectid < BTRFS_FIRST_FREE_OBJECTID ||
-	     root->root_key.objectid > BTRFS_LAST_FREE_OBJECTID))
-		return 0;
-
-	/* Don't save inode cache if we are deleting this root */
-	if (btrfs_root_refs(&root->root_item) == 0)
-		return 0;
-
-	if (!btrfs_test_opt(fs_info, INODE_MAP_CACHE))
-		return 0;
-
-	path = btrfs_alloc_path();
-	if (!path)
-		return -ENOMEM;
-
-	rsv = trans->block_rsv;
-	trans->block_rsv = &fs_info->trans_block_rsv;
-
-	num_bytes = trans->bytes_reserved;
-	/*
-	 * 1 item for inode item insertion if need
-	 * 4 items for inode item update (in the worst case)
-	 * 1 items for slack space if we need do truncation
-	 * 1 item for free space object
-	 * 3 items for pre-allocation
-	 */
-	trans->bytes_reserved = btrfs_calc_insert_metadata_size(fs_info, 10);
-	ret = btrfs_block_rsv_add(root, trans->block_rsv,
-				  trans->bytes_reserved,
-				  BTRFS_RESERVE_NO_FLUSH);
-	if (ret)
-		goto out;
-	trace_btrfs_space_reservation(fs_info, "ino_cache", trans->transid,
-				      trans->bytes_reserved, 1);
-again:
-	inode = lookup_free_ino_inode(root, path);
-	if (IS_ERR(inode) && (PTR_ERR(inode) != -ENOENT || retry)) {
-		ret = PTR_ERR(inode);
-		goto out_release;
-	}
-
-	if (IS_ERR(inode)) {
-		BUG_ON(retry); /* Logic error */
-		retry = true;
-
-		ret = create_free_ino_inode(root, trans, path);
-		if (ret)
-			goto out_release;
-		goto again;
-	}
-
-	BTRFS_I(inode)->generation = 0;
-	ret = btrfs_update_inode(trans, root, inode);
-	if (ret) {
-		btrfs_abort_transaction(trans, ret);
-		goto out_put;
-	}
-
-	if (i_size_read(inode) > 0) {
-		ret = btrfs_truncate_free_space_cache(trans, NULL, inode);
-		if (ret) {
-			if (ret != -ENOSPC)
-				btrfs_abort_transaction(trans, ret);
-			goto out_put;
-		}
-	}
-
-	spin_lock(&root->ino_cache_lock);
-	if (root->ino_cache_state != BTRFS_CACHE_FINISHED) {
-		ret = -1;
-		spin_unlock(&root->ino_cache_lock);
-		goto out_put;
-	}
-	spin_unlock(&root->ino_cache_lock);
-
-	spin_lock(&ctl->tree_lock);
-	prealloc = sizeof(struct btrfs_free_space) * ctl->free_extents;
-	prealloc = ALIGN(prealloc, PAGE_SIZE);
-	prealloc += ctl->total_bitmaps * PAGE_SIZE;
-	spin_unlock(&ctl->tree_lock);
-
-	/* Just to make sure we have enough space */
-	prealloc += 8 * PAGE_SIZE;
-
-	ret = btrfs_delalloc_reserve_space(inode, &data_reserved, 0, prealloc);
-	if (ret)
-		goto out_put;
-
-	ret = btrfs_prealloc_file_range_trans(inode, trans, 0, 0, prealloc,
-					      prealloc, prealloc, &alloc_hint);
-	if (ret) {
-		btrfs_delalloc_release_extents(BTRFS_I(inode), prealloc);
-		btrfs_delalloc_release_metadata(BTRFS_I(inode), prealloc, true);
-		goto out_put;
-	}
-
-	ret = btrfs_write_out_ino_cache(root, trans, path, inode);
-	btrfs_delalloc_release_extents(BTRFS_I(inode), prealloc);
-out_put:
-	iput(inode);
-out_release:
-	trace_btrfs_space_reservation(fs_info, "ino_cache", trans->transid,
-				      trans->bytes_reserved, 0);
-	btrfs_block_rsv_release(fs_info, trans->block_rsv,
-				trans->bytes_reserved);
-out:
-	trans->block_rsv = rsv;
-	trans->bytes_reserved = num_bytes;
-
-	btrfs_free_path(path);
-	extent_changeset_free(data_reserved);
-	return ret;
-}
-
-int btrfs_find_highest_objectid(struct btrfs_root *root, u64 *objectid)
-{
-	struct btrfs_path *path;
-	int ret;
-	struct extent_buffer *l;
-	struct btrfs_key search_key;
-	struct btrfs_key found_key;
-	int slot;
-
-	path = btrfs_alloc_path();
-	if (!path)
-		return -ENOMEM;
-
-	search_key.objectid = BTRFS_LAST_FREE_OBJECTID;
-	search_key.type = -1;
-	search_key.offset = (u64)-1;
-	ret = btrfs_search_slot(NULL, root, &search_key, path, 0, 0);
-	if (ret < 0)
-		goto error;
-	BUG_ON(ret == 0); /* Corruption */
-	if (path->slots[0] > 0) {
-		slot = path->slots[0] - 1;
-		l = path->nodes[0];
-		btrfs_item_key_to_cpu(l, &found_key, slot);
-		*objectid = max_t(u64, found_key.objectid,
-				  BTRFS_FIRST_FREE_OBJECTID - 1);
-	} else {
-		*objectid = BTRFS_FIRST_FREE_OBJECTID - 1;
-	}
-	ret = 0;
-error:
-	btrfs_free_path(path);
-	return ret;
-}
-
-int btrfs_find_free_objectid(struct btrfs_root *root, u64 *objectid)
-{
-	int ret;
-	mutex_lock(&root->objectid_mutex);
-
-	if (unlikely(root->highest_objectid >= BTRFS_LAST_FREE_OBJECTID)) {
-		btrfs_warn(root->fs_info,
-			   "the objectid of root %llu reaches its highest value",
-			   root->root_key.objectid);
-		ret = -ENOSPC;
-		goto out;
-	}
-
-	*objectid = ++root->highest_objectid;
-	ret = 0;
-out:
-	mutex_unlock(&root->objectid_mutex);
-	return ret;
-}
diff --git a/fs/btrfs/inode-map.h b/fs/btrfs/inode-map.h
deleted file mode 100644
index 7a962811dffe..000000000000
--- a/fs/btrfs/inode-map.h
+++ /dev/null
@@ -1,16 +0,0 @@
-/* SPDX-License-Identifier: GPL-2.0 */
-
-#ifndef BTRFS_INODE_MAP_H
-#define BTRFS_INODE_MAP_H
-
-void btrfs_init_free_ino_ctl(struct btrfs_root *root);
-void btrfs_unpin_free_ino(struct btrfs_root *root);
-void btrfs_return_ino(struct btrfs_root *root, u64 objectid);
-int btrfs_find_free_ino(struct btrfs_root *root, u64 *objectid);
-int btrfs_save_ino_cache(struct btrfs_root *root,
-			 struct btrfs_trans_handle *trans);
-
-int btrfs_find_free_objectid(struct btrfs_root *root, u64 *objectid);
-int btrfs_find_highest_objectid(struct btrfs_root *root, u64 *objectid);
-
-#endif
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index d267eb5caa7b..0e516aefbf51 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -3,9 +3,10 @@
  * Copyright (C) 2007 Oracle.  All rights reserved.
  */
 
+#include <crypto/hash.h>
 #include <linux/kernel.h>
 #include <linux/bio.h>
-#include <linux/buffer_head.h>
+#include <linux/blk-cgroup.h>
 #include <linux/file.h>
 #include <linux/fs.h>
 #include <linux/pagemap.h>
@@ -28,8 +29,11 @@
 #include <linux/magic.h>
 #include <linux/iversion.h>
 #include <linux/swap.h>
+#include <linux/migrate.h>
 #include <linux/sched/mm.h>
+#include <linux/iomap.h>
 #include <asm/unaligned.h>
+#include <linux/fsverity.h>
 #include "misc.h"
 #include "ctree.h"
 #include "disk-io.h"
@@ -43,22 +47,56 @@
 #include "compression.h"
 #include "locking.h"
 #include "free-space-cache.h"
-#include "inode-map.h"
 #include "props.h"
 #include "qgroup.h"
 #include "delalloc-space.h"
 #include "block-group.h"
+#include "space-info.h"
+#include "zoned.h"
+#include "subpage.h"
+#include "inode-item.h"
 
 struct btrfs_iget_args {
-	struct btrfs_key *location;
+	u64 ino;
 	struct btrfs_root *root;
 };
 
 struct btrfs_dio_data {
-	u64 reserve;
-	u64 unsubmitted_oe_range_start;
-	u64 unsubmitted_oe_range_end;
-	int overwrite;
+	ssize_t submitted;
+	struct extent_changeset *data_reserved;
+	bool data_space_reserved;
+	bool nocow_done;
+};
+
+struct btrfs_dio_private {
+	struct inode *inode;
+
+	/*
+	 * Since DIO can use anonymous page, we cannot use page_offset() to
+	 * grab the file offset, thus need a dedicated member for file offset.
+	 */
+	u64 file_offset;
+	/* Used for bio::bi_size */
+	u32 bytes;
+
+	/*
+	 * References to this structure. There is one reference per in-flight
+	 * bio plus one while we're still setting up.
+	 */
+	refcount_t refs;
+
+	/* Array of checksums */
+	u8 *csums;
+
+	/* This must be last */
+	struct bio bio;
+};
+
+static struct bio_set btrfs_dio_bioset;
+
+struct btrfs_rename_ctx {
+	/* Output field. Stores the index number of the old directory entry. */
+	u64 index;
 };
 
 static const struct inode_operations btrfs_dir_inode_operations;
@@ -67,7 +105,6 @@ static const struct inode_operations btrfs_special_inode_operations;
 static const struct inode_operations btrfs_file_inode_operations;
 static const struct address_space_operations btrfs_aops;
 static const struct file_operations btrfs_dir_file_operations;
-static const struct extent_io_ops btrfs_extent_io_ops;
 
 static struct kmem_cache *btrfs_inode_cachep;
 struct kmem_cache *btrfs_trans_handle_cachep;
@@ -77,20 +114,66 @@ struct kmem_cache *btrfs_free_space_bitmap_cachep;
 
 static int btrfs_setsize(struct inode *inode, struct iattr *attr);
 static int btrfs_truncate(struct inode *inode, bool skip_writeback);
-static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent);
-static noinline int cow_file_range(struct inode *inode,
+static noinline int cow_file_range(struct btrfs_inode *inode,
 				   struct page *locked_page,
 				   u64 start, u64 end, int *page_started,
-				   unsigned long *nr_written, int unlock);
-static struct extent_map *create_io_em(struct inode *inode, u64 start, u64 len,
-				       u64 orig_start, u64 block_start,
+				   unsigned long *nr_written, int unlock,
+				   u64 *done_offset);
+static struct extent_map *create_io_em(struct btrfs_inode *inode, u64 start,
+				       u64 len, u64 orig_start, u64 block_start,
 				       u64 block_len, u64 orig_block_len,
 				       u64 ram_bytes, int compress_type,
 				       int type);
 
-static void __endio_write_update_ordered(struct inode *inode,
-					 const u64 offset, const u64 bytes,
-					 const bool uptodate);
+/*
+ * btrfs_inode_lock - lock inode i_rwsem based on arguments passed
+ *
+ * ilock_flags can have the following bit set:
+ *
+ * BTRFS_ILOCK_SHARED - acquire a shared lock on the inode
+ * BTRFS_ILOCK_TRY - try to acquire the lock, if fails on first attempt
+ *		     return -EAGAIN
+ * BTRFS_ILOCK_MMAP - acquire a write lock on the i_mmap_lock
+ */
+int btrfs_inode_lock(struct inode *inode, unsigned int ilock_flags)
+{
+	if (ilock_flags & BTRFS_ILOCK_SHARED) {
+		if (ilock_flags & BTRFS_ILOCK_TRY) {
+			if (!inode_trylock_shared(inode))
+				return -EAGAIN;
+			else
+				return 0;
+		}
+		inode_lock_shared(inode);
+	} else {
+		if (ilock_flags & BTRFS_ILOCK_TRY) {
+			if (!inode_trylock(inode))
+				return -EAGAIN;
+			else
+				return 0;
+		}
+		inode_lock(inode);
+	}
+	if (ilock_flags & BTRFS_ILOCK_MMAP)
+		down_write(&BTRFS_I(inode)->i_mmap_lock);
+	return 0;
+}
+
+/*
+ * btrfs_inode_unlock - unock inode i_rwsem
+ *
+ * ilock_flags should contain the same bits set as passed to btrfs_inode_lock()
+ * to decide whether the lock acquired is shared or exclusive.
+ */
+void btrfs_inode_unlock(struct inode *inode, unsigned int ilock_flags)
+{
+	if (ilock_flags & BTRFS_ILOCK_MMAP)
+		up_write(&BTRFS_I(inode)->i_mmap_lock);
+	if (ilock_flags & BTRFS_ILOCK_SHARED)
+		inode_unlock_shared(inode);
+	else
+		inode_unlock(inode);
+}
 
 /*
  * Cleanup all submitted ordered extents in specified range to handle errors
@@ -102,58 +185,92 @@ static void __endio_write_update_ordered(struct inode *inode,
  * to be released, which we want to happen only when finishing the ordered
  * extent (btrfs_finish_ordered_io()).
  */
-static inline void btrfs_cleanup_ordered_extents(struct inode *inode,
+static inline void btrfs_cleanup_ordered_extents(struct btrfs_inode *inode,
 						 struct page *locked_page,
 						 u64 offset, u64 bytes)
 {
 	unsigned long index = offset >> PAGE_SHIFT;
 	unsigned long end_index = (offset + bytes - 1) >> PAGE_SHIFT;
-	u64 page_start = page_offset(locked_page);
-	u64 page_end = page_start + PAGE_SIZE - 1;
-
+	u64 page_start, page_end;
 	struct page *page;
 
+	if (locked_page) {
+		page_start = page_offset(locked_page);
+		page_end = page_start + PAGE_SIZE - 1;
+	}
+
 	while (index <= end_index) {
-		page = find_get_page(inode->i_mapping, index);
+		/*
+		 * For locked page, we will call end_extent_writepage() on it
+		 * in run_delalloc_range() for the error handling.  That
+		 * end_extent_writepage() function will call
+		 * btrfs_mark_ordered_io_finished() to clear page Ordered and
+		 * run the ordered extent accounting.
+		 *
+		 * Here we can't just clear the Ordered bit, or
+		 * btrfs_mark_ordered_io_finished() would skip the accounting
+		 * for the page range, and the ordered extent will never finish.
+		 */
+		if (locked_page && index == (page_start >> PAGE_SHIFT)) {
+			index++;
+			continue;
+		}
+		page = find_get_page(inode->vfs_inode.i_mapping, index);
 		index++;
 		if (!page)
 			continue;
-		ClearPagePrivate2(page);
+
+		/*
+		 * Here we just clear all Ordered bits for every page in the
+		 * range, then btrfs_mark_ordered_io_finished() will handle
+		 * the ordered extent accounting for the range.
+		 */
+		btrfs_page_clamp_clear_ordered(inode->root->fs_info, page,
+					       offset, bytes);
 		put_page(page);
 	}
 
-	/*
-	 * In case this page belongs to the delalloc range being instantiated
-	 * then skip it, since the first page of a range is going to be
-	 * properly cleaned up by the caller of run_delalloc_range
-	 */
-	if (page_start >= offset && page_end <= (offset + bytes - 1)) {
-		offset += PAGE_SIZE;
-		bytes -= PAGE_SIZE;
+	if (locked_page) {
+		/* The locked page covers the full range, nothing needs to be done */
+		if (bytes + offset <= page_start + PAGE_SIZE)
+			return;
+		/*
+		 * In case this page belongs to the delalloc range being
+		 * instantiated then skip it, since the first page of a range is
+		 * going to be properly cleaned up by the caller of
+		 * run_delalloc_range
+		 */
+		if (page_start >= offset && page_end <= (offset + bytes - 1)) {
+			bytes = offset + bytes - page_offset(locked_page) - PAGE_SIZE;
+			offset = page_offset(locked_page) + PAGE_SIZE;
+		}
 	}
 
-	return __endio_write_update_ordered(inode, offset, bytes, false);
+	return btrfs_mark_ordered_io_finished(inode, NULL, offset, bytes, false);
 }
 
 static int btrfs_dirty_inode(struct inode *inode);
 
-#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
-void btrfs_test_inode_set_ops(struct inode *inode)
-{
-	BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops;
-}
-#endif
-
 static int btrfs_init_inode_security(struct btrfs_trans_handle *trans,
-				     struct inode *inode,  struct inode *dir,
-				     const struct qstr *qstr)
+				     struct btrfs_new_inode_args *args)
 {
 	int err;
 
-	err = btrfs_init_acl(trans, inode, dir);
-	if (!err)
-		err = btrfs_xattr_security_init(trans, inode, dir, qstr);
-	return err;
+	if (args->default_acl) {
+		err = __btrfs_set_acl(trans, args->inode, args->default_acl,
+				      ACL_TYPE_DEFAULT);
+		if (err)
+			return err;
+	}
+	if (args->acl) {
+		err = __btrfs_set_acl(trans, args->inode, args->acl, ACL_TYPE_ACCESS);
+		if (err)
+			return err;
+	}
+	if (!args->default_acl && !args->acl)
+		cache_no_acl(args->inode);
+	return btrfs_xattr_security_init(trans, args->inode, args->dir,
+					 &args->dentry->d_name);
 }
 
 /*
@@ -162,12 +279,14 @@ static int btrfs_init_inode_security(struct btrfs_trans_handle *trans,
  * no overlapping inline items exist in the btree
  */
 static int insert_inline_extent(struct btrfs_trans_handle *trans,
-				struct btrfs_path *path, int extent_inserted,
-				struct btrfs_root *root, struct inode *inode,
-				u64 start, size_t size, size_t compressed_size,
+				struct btrfs_path *path,
+				struct btrfs_inode *inode, bool extent_inserted,
+				size_t size, size_t compressed_size,
 				int compress_type,
-				struct page **compressed_pages)
+				struct page **compressed_pages,
+				bool update_i_size)
 {
+	struct btrfs_root *root = inode->root;
 	struct extent_buffer *leaf;
 	struct page *page = NULL;
 	char *kaddr;
@@ -175,7 +294,7 @@ static int insert_inline_extent(struct btrfs_trans_handle *trans,
 	struct btrfs_file_extent_item *ei;
 	int ret;
 	size_t cur_size = size;
-	unsigned long offset;
+	u64 i_size;
 
 	ASSERT((compressed_size > 0 && compressed_pages) ||
 	       (compressed_size == 0 && !compressed_pages));
@@ -183,18 +302,15 @@ static int insert_inline_extent(struct btrfs_trans_handle *trans,
 	if (compressed_size && compressed_pages)
 		cur_size = compressed_size;
 
-	inode_add_bytes(inode, size);
-
 	if (!extent_inserted) {
 		struct btrfs_key key;
 		size_t datasize;
 
-		key.objectid = btrfs_ino(BTRFS_I(inode));
-		key.offset = start;
+		key.objectid = btrfs_ino(inode);
+		key.offset = 0;
 		key.type = BTRFS_EXTENT_DATA_KEY;
 
 		datasize = btrfs_file_extent_calc_inline_size(cur_size);
-		path->leave_spinning = 1;
 		ret = btrfs_insert_empty_item(trans, root, path, &key,
 					      datasize);
 		if (ret)
@@ -218,9 +334,9 @@ static int insert_inline_extent(struct btrfs_trans_handle *trans,
 			cur_size = min_t(unsigned long, compressed_size,
 				       PAGE_SIZE);
 
-			kaddr = kmap_atomic(cpage);
+			kaddr = kmap_local_page(cpage);
 			write_extent_buffer(leaf, kaddr, ptr, cur_size);
-			kunmap_atomic(kaddr);
+			kunmap_local(kaddr);
 
 			i++;
 			ptr += cur_size;
@@ -229,29 +345,38 @@ static int insert_inline_extent(struct btrfs_trans_handle *trans,
 		btrfs_set_file_extent_compression(leaf, ei,
 						  compress_type);
 	} else {
-		page = find_get_page(inode->i_mapping,
-				     start >> PAGE_SHIFT);
+		page = find_get_page(inode->vfs_inode.i_mapping, 0);
 		btrfs_set_file_extent_compression(leaf, ei, 0);
-		kaddr = kmap_atomic(page);
-		offset = offset_in_page(start);
-		write_extent_buffer(leaf, kaddr + offset, ptr, size);
-		kunmap_atomic(kaddr);
+		kaddr = kmap_local_page(page);
+		write_extent_buffer(leaf, kaddr, ptr, size);
+		kunmap_local(kaddr);
 		put_page(page);
 	}
 	btrfs_mark_buffer_dirty(leaf);
 	btrfs_release_path(path);
 
 	/*
-	 * we're an inline extent, so nobody can
-	 * extend the file past i_size without locking
-	 * a page we already have locked.
+	 * We align size to sectorsize for inline extents just for simplicity
+	 * sake.
+	 */
+	ret = btrfs_inode_set_file_extent_range(inode, 0,
+					ALIGN(size, root->fs_info->sectorsize));
+	if (ret)
+		goto fail;
+
+	/*
+	 * We're an inline extent, so nobody can extend the file past i_size
+	 * without locking a page we already have locked.
 	 *
-	 * We must do any isize and inode updates
-	 * before we unlock the pages.  Otherwise we
-	 * could end up racing with unlink.
+	 * We must do any i_size and inode updates before we unlock the pages.
+	 * Otherwise we could end up racing with unlink.
 	 */
-	BTRFS_I(inode)->disk_i_size = inode->i_size;
-	ret = btrfs_update_inode(trans, root, inode);
+	i_size = i_size_read(&inode->vfs_inode);
+	if (update_i_size && size > i_size) {
+		i_size_write(&inode->vfs_inode, size);
+		i_size = size;
+	}
+	inode->disk_i_size = i_size;
 
 fail:
 	return ret;
@@ -263,36 +388,31 @@ fail:
  * does the checks required to make sure the data is small enough
  * to fit as an inline extent.
  */
-static noinline int cow_file_range_inline(struct inode *inode, u64 start,
-					  u64 end, size_t compressed_size,
+static noinline int cow_file_range_inline(struct btrfs_inode *inode, u64 size,
+					  size_t compressed_size,
 					  int compress_type,
-					  struct page **compressed_pages)
+					  struct page **compressed_pages,
+					  bool update_i_size)
 {
-	struct btrfs_root *root = BTRFS_I(inode)->root;
+	struct btrfs_drop_extents_args drop_args = { 0 };
+	struct btrfs_root *root = inode->root;
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct btrfs_trans_handle *trans;
-	u64 isize = i_size_read(inode);
-	u64 actual_end = min(end + 1, isize);
-	u64 inline_len = actual_end - start;
-	u64 aligned_end = ALIGN(end, fs_info->sectorsize);
-	u64 data_len = inline_len;
+	u64 data_len = (compressed_size ?: size);
 	int ret;
 	struct btrfs_path *path;
-	int extent_inserted = 0;
-	u32 extent_item_size;
-
-	if (compressed_size)
-		data_len = compressed_size;
 
-	if (start > 0 ||
-	    actual_end > fs_info->sectorsize ||
+	/*
+	 * We can create an inline extent if it ends at or beyond the current
+	 * i_size, is no larger than a sector (decompressed), and the (possibly
+	 * compressed) data fits in a leaf and the configured maximum inline
+	 * size.
+	 */
+	if (size < i_size_read(&inode->vfs_inode) ||
+	    size > fs_info->sectorsize ||
 	    data_len > BTRFS_MAX_INLINE_DATA_SIZE(fs_info) ||
-	    (!compressed_size &&
-	    (actual_end & (fs_info->sectorsize - 1)) == 0) ||
-	    end + 1 < isize ||
-	    data_len > fs_info->max_inline) {
+	    data_len > fs_info->max_inline)
 		return 1;
-	}
 
 	path = btrfs_alloc_path();
 	if (!path)
@@ -303,29 +423,23 @@ static noinline int cow_file_range_inline(struct inode *inode, u64 start,
 		btrfs_free_path(path);
 		return PTR_ERR(trans);
 	}
-	trans->block_rsv = &BTRFS_I(inode)->block_rsv;
-
-	if (compressed_size && compressed_pages)
-		extent_item_size = btrfs_file_extent_calc_inline_size(
-		   compressed_size);
-	else
-		extent_item_size = btrfs_file_extent_calc_inline_size(
-		    inline_len);
+	trans->block_rsv = &inode->block_rsv;
 
-	ret = __btrfs_drop_extents(trans, root, inode, path,
-				   start, aligned_end, NULL,
-				   1, 1, extent_item_size, &extent_inserted);
+	drop_args.path = path;
+	drop_args.start = 0;
+	drop_args.end = fs_info->sectorsize;
+	drop_args.drop_cache = true;
+	drop_args.replace_extent = true;
+	drop_args.extent_item_size = btrfs_file_extent_calc_inline_size(data_len);
+	ret = btrfs_drop_extents(trans, root, inode, &drop_args);
 	if (ret) {
 		btrfs_abort_transaction(trans, ret);
 		goto out;
 	}
 
-	if (isize > actual_end)
-		inline_len = min_t(u64, isize, actual_end);
-	ret = insert_inline_extent(trans, path, extent_inserted,
-				   root, inode, start,
-				   inline_len, compressed_size,
-				   compress_type, compressed_pages);
+	ret = insert_inline_extent(trans, path, inode, drop_args.extent_inserted,
+				   size, compressed_size, compress_type,
+				   compressed_pages, update_i_size);
 	if (ret && ret != -ENOSPC) {
 		btrfs_abort_transaction(trans, ret);
 		goto out;
@@ -334,8 +448,17 @@ static noinline int cow_file_range_inline(struct inode *inode, u64 start,
 		goto out;
 	}
 
-	set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &BTRFS_I(inode)->runtime_flags);
-	btrfs_drop_extent_cache(BTRFS_I(inode), start, aligned_end - 1, 0);
+	btrfs_update_inode_bytes(inode, size, drop_args.bytes_found);
+	ret = btrfs_update_inode(trans, root, inode);
+	if (ret && ret != -ENOSPC) {
+		btrfs_abort_transaction(trans, ret);
+		goto out;
+	} else if (ret == -ENOSPC) {
+		ret = 1;
+		goto out;
+	}
+
+	btrfs_set_inode_full_sync(inode);
 out:
 	/*
 	 * Don't forget to free the reserved space, as for inlined extent
@@ -364,15 +487,14 @@ struct async_chunk {
 	struct page *locked_page;
 	u64 start;
 	u64 end;
-	unsigned int write_flags;
+	blk_opf_t write_flags;
 	struct list_head extents;
 	struct cgroup_subsys_state *blkcg_css;
 	struct btrfs_work work;
-	atomic_t *pending;
+	struct async_cow *async_cow;
 };
 
 struct async_cow {
-	/* Number of chunks in flight; must be first in the structure */
 	atomic_t num_chunks;
 	struct async_chunk chunks[];
 };
@@ -399,53 +521,75 @@ static noinline int add_async_extent(struct async_chunk *cow,
 }
 
 /*
- * Check if the inode has flags compatible with compression
- */
-static inline bool inode_can_compress(struct inode *inode)
-{
-	if (BTRFS_I(inode)->flags & BTRFS_INODE_NODATACOW ||
-	    BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM)
-		return false;
-	return true;
-}
-
-/*
  * Check if the inode needs to be submitted to compression, based on mount
  * options, defragmentation, properties or heuristics.
  */
-static inline int inode_need_compress(struct inode *inode, u64 start, u64 end)
+static inline int inode_need_compress(struct btrfs_inode *inode, u64 start,
+				      u64 end)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
 
-	if (!inode_can_compress(inode)) {
+	if (!btrfs_inode_can_compress(inode)) {
 		WARN(IS_ENABLED(CONFIG_BTRFS_DEBUG),
 			KERN_ERR "BTRFS: unexpected compression for ino %llu\n",
-			btrfs_ino(BTRFS_I(inode)));
+			btrfs_ino(inode));
 		return 0;
 	}
+	/*
+	 * Special check for subpage.
+	 *
+	 * We lock the full page then run each delalloc range in the page, thus
+	 * for the following case, we will hit some subpage specific corner case:
+	 *
+	 * 0		32K		64K
+	 * |	|///////|	|///////|
+	 *		\- A		\- B
+	 *
+	 * In above case, both range A and range B will try to unlock the full
+	 * page [0, 64K), causing the one finished later will have page
+	 * unlocked already, triggering various page lock requirement BUG_ON()s.
+	 *
+	 * So here we add an artificial limit that subpage compression can only
+	 * if the range is fully page aligned.
+	 *
+	 * In theory we only need to ensure the first page is fully covered, but
+	 * the tailing partial page will be locked until the full compression
+	 * finishes, delaying the write of other range.
+	 *
+	 * TODO: Make btrfs_run_delalloc_range() to lock all delalloc range
+	 * first to prevent any submitted async extent to unlock the full page.
+	 * By this, we can ensure for subpage case that only the last async_cow
+	 * will unlock the full page.
+	 */
+	if (fs_info->sectorsize < PAGE_SIZE) {
+		if (!PAGE_ALIGNED(start) ||
+		    !PAGE_ALIGNED(end + 1))
+			return 0;
+	}
+
 	/* force compress */
 	if (btrfs_test_opt(fs_info, FORCE_COMPRESS))
 		return 1;
 	/* defrag ioctl */
-	if (BTRFS_I(inode)->defrag_compress)
+	if (inode->defrag_compress)
 		return 1;
 	/* bad compression ratios */
-	if (BTRFS_I(inode)->flags & BTRFS_INODE_NOCOMPRESS)
+	if (inode->flags & BTRFS_INODE_NOCOMPRESS)
 		return 0;
 	if (btrfs_test_opt(fs_info, COMPRESS) ||
-	    BTRFS_I(inode)->flags & BTRFS_INODE_COMPRESS ||
-	    BTRFS_I(inode)->prop_compress)
-		return btrfs_compress_heuristic(inode, start, end);
+	    inode->flags & BTRFS_INODE_COMPRESS ||
+	    inode->prop_compress)
+		return btrfs_compress_heuristic(&inode->vfs_inode, start, end);
 	return 0;
 }
 
 static inline void inode_should_defrag(struct btrfs_inode *inode,
-		u64 start, u64 end, u64 num_bytes, u64 small_write)
+		u64 start, u64 end, u64 num_bytes, u32 small_write)
 {
 	/* If this is a small write inside eof, kick off a defrag */
 	if (num_bytes < small_write &&
 	    (start > 0 || end + 1 < inode->disk_i_size))
-		btrfs_add_inode_defrag(NULL, inode);
+		btrfs_add_inode_defrag(NULL, inode, small_write);
 }
 
 /*
@@ -504,7 +648,6 @@ static noinline int compress_file_range(struct async_chunk *async_chunk)
 again:
 	will_compress = 0;
 	nr_pages = (end >> PAGE_SHIFT) - (start >> PAGE_SHIFT) + 1;
-	BUILD_BUG_ON((BTRFS_MAX_COMPRESSED % PAGE_SIZE) != 0);
 	nr_pages = min_t(unsigned long, nr_pages,
 			BTRFS_MAX_COMPRESSED / PAGE_SIZE);
 
@@ -524,13 +667,24 @@ again:
 	total_compressed = actual_end - start;
 
 	/*
-	 * skip compression for a small file range(<=blocksize) that
+	 * Skip compression for a small file range(<=blocksize) that
 	 * isn't an inline extent, since it doesn't save disk space at all.
 	 */
 	if (total_compressed <= blocksize &&
 	   (start > 0 || end + 1 < BTRFS_I(inode)->disk_i_size))
 		goto cleanup_and_bail_uncompressed;
 
+	/*
+	 * For subpage case, we require full page alignment for the sector
+	 * aligned range.
+	 * Thus we must also check against @actual_end, not just @end.
+	 */
+	if (blocksize < PAGE_SIZE) {
+		if (!PAGE_ALIGNED(start) ||
+		    !PAGE_ALIGNED(round_up(actual_end, blocksize)))
+			goto cleanup_and_bail_uncompressed;
+	}
+
 	total_compressed = min_t(unsigned long, total_compressed,
 			BTRFS_MAX_UNCOMPRESSED);
 	total_in = 0;
@@ -541,7 +695,7 @@ again:
 	 * inode has not been flagged as nocompress.  This flag can
 	 * change at any time if we discover bad compression ratios.
 	 */
-	if (inode_need_compress(inode, start, end)) {
+	if (inode_need_compress(BTRFS_I(inode), start, end)) {
 		WARN_ON(pages);
 		pages = kcalloc(nr_pages, sizeof(struct page *), GFP_NOFS);
 		if (!pages) {
@@ -584,34 +738,35 @@ again:
 		if (!ret) {
 			unsigned long offset = offset_in_page(total_compressed);
 			struct page *page = pages[nr_pages - 1];
-			char *kaddr;
 
 			/* zero the tail end of the last page, we might be
 			 * sending it down to disk
 			 */
-			if (offset) {
-				kaddr = kmap_atomic(page);
-				memset(kaddr + offset, 0,
-				       PAGE_SIZE - offset);
-				kunmap_atomic(kaddr);
-			}
+			if (offset)
+				memzero_page(page, offset, PAGE_SIZE - offset);
 			will_compress = 1;
 		}
 	}
 cont:
-	if (start == 0) {
+	/*
+	 * Check cow_file_range() for why we don't even try to create inline
+	 * extent for subpage case.
+	 */
+	if (start == 0 && fs_info->sectorsize == PAGE_SIZE) {
 		/* lets try to make an inline extent */
 		if (ret || total_in < actual_end) {
 			/* we didn't compress the entire range, try
 			 * to make an uncompressed inline extent.
 			 */
-			ret = cow_file_range_inline(inode, start, end, 0,
-						    BTRFS_COMPRESS_NONE, NULL);
+			ret = cow_file_range_inline(BTRFS_I(inode), actual_end,
+						    0, BTRFS_COMPRESS_NONE,
+						    NULL, false);
 		} else {
 			/* try making a compressed inline extent */
-			ret = cow_file_range_inline(inode, start, end,
+			ret = cow_file_range_inline(BTRFS_I(inode), actual_end,
 						    total_compressed,
-						    compress_type, pages);
+						    compress_type, pages,
+						    false);
 		}
 		if (ret <= 0) {
 			unsigned long clear_flags = EXTENT_DELALLOC |
@@ -631,20 +786,26 @@ cont:
 			 * our outstanding extent for clearing delalloc for this
 			 * range.
 			 */
-			extent_clear_unlock_delalloc(inode, start, end, NULL,
+			extent_clear_unlock_delalloc(BTRFS_I(inode), start, end,
+						     NULL,
 						     clear_flags,
 						     PAGE_UNLOCK |
-						     PAGE_CLEAR_DIRTY |
-						     PAGE_SET_WRITEBACK |
+						     PAGE_START_WRITEBACK |
 						     page_error_op |
 						     PAGE_END_WRITEBACK);
 
-			for (i = 0; i < nr_pages; i++) {
-				WARN_ON(pages[i]->mapping);
-				put_page(pages[i]);
+			/*
+			 * Ensure we only free the compressed pages if we have
+			 * them allocated, as we can still reach here with
+			 * inode_need_compress() == false.
+			 */
+			if (pages) {
+				for (i = 0; i < nr_pages; i++) {
+					WARN_ON(pages[i]->mapping);
+					put_page(pages[i]);
+				}
+				kfree(pages);
 			}
-			kfree(pages);
-
 			return 0;
 		}
 	}
@@ -662,7 +823,7 @@ cont:
 		 * win, compare the page count read with the blocks on disk,
 		 * compression must free at least one sector size
 		 */
-		total_in = ALIGN(total_in, PAGE_SIZE);
+		total_in = round_up(total_in, fs_info->sectorsize);
 		if (total_compressed + blocksize <= total_in) {
 			compressed_extents++;
 
@@ -743,185 +904,207 @@ static void free_async_extent_pages(struct async_extent *async_extent)
 	async_extent->pages = NULL;
 }
 
-/*
- * phase two of compressed writeback.  This is the ordered portion
- * of the code, which only gets called in the order the work was
- * queued.  We walk all the async extents created by compress_file_range
- * and send them down to the disk.
- */
-static noinline void submit_compressed_extents(struct async_chunk *async_chunk)
+static int submit_uncompressed_range(struct btrfs_inode *inode,
+				     struct async_extent *async_extent,
+				     struct page *locked_page)
 {
-	struct inode *inode = async_chunk->inode;
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	struct async_extent *async_extent;
-	u64 alloc_hint = 0;
-	struct btrfs_key ins;
-	struct extent_map *em;
-	struct btrfs_root *root = BTRFS_I(inode)->root;
-	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
-	int ret = 0;
+	u64 start = async_extent->start;
+	u64 end = async_extent->start + async_extent->ram_size - 1;
+	unsigned long nr_written = 0;
+	int page_started = 0;
+	int ret;
 
-again:
-	while (!list_empty(&async_chunk->extents)) {
-		async_extent = list_entry(async_chunk->extents.next,
-					  struct async_extent, list);
-		list_del(&async_extent->list);
+	/*
+	 * Call cow_file_range() to run the delalloc range directly, since we
+	 * won't go to NOCOW or async path again.
+	 *
+	 * Also we call cow_file_range() with @unlock_page == 0, so that we
+	 * can directly submit them without interruption.
+	 */
+	ret = cow_file_range(inode, locked_page, start, end, &page_started,
+			     &nr_written, 0, NULL);
+	/* Inline extent inserted, page gets unlocked and everything is done */
+	if (page_started) {
+		ret = 0;
+		goto out;
+	}
+	if (ret < 0) {
+		btrfs_cleanup_ordered_extents(inode, locked_page, start, end - start + 1);
+		if (locked_page) {
+			const u64 page_start = page_offset(locked_page);
+			const u64 page_end = page_start + PAGE_SIZE - 1;
 
-retry:
-		lock_extent(io_tree, async_extent->start,
-			    async_extent->start + async_extent->ram_size - 1);
-		/* did the compression code fall back to uncompressed IO? */
-		if (!async_extent->pages) {
-			int page_started = 0;
-			unsigned long nr_written = 0;
+			btrfs_page_set_error(inode->root->fs_info, locked_page,
+					     page_start, PAGE_SIZE);
+			set_page_writeback(locked_page);
+			end_page_writeback(locked_page);
+			end_extent_writepage(locked_page, ret, page_start, page_end);
+			unlock_page(locked_page);
+		}
+		goto out;
+	}
 
-			/* allocate blocks */
-			ret = cow_file_range(inode, async_chunk->locked_page,
-					     async_extent->start,
-					     async_extent->start +
-					     async_extent->ram_size - 1,
-					     &page_started, &nr_written, 0);
+	ret = extent_write_locked_range(&inode->vfs_inode, start, end);
+	/* All pages will be unlocked, including @locked_page */
+out:
+	kfree(async_extent);
+	return ret;
+}
 
-			/* JDM XXX */
+static int submit_one_async_extent(struct btrfs_inode *inode,
+				   struct async_chunk *async_chunk,
+				   struct async_extent *async_extent,
+				   u64 *alloc_hint)
+{
+	struct extent_io_tree *io_tree = &inode->io_tree;
+	struct btrfs_root *root = inode->root;
+	struct btrfs_fs_info *fs_info = root->fs_info;
+	struct btrfs_key ins;
+	struct page *locked_page = NULL;
+	struct extent_map *em;
+	int ret = 0;
+	u64 start = async_extent->start;
+	u64 end = async_extent->start + async_extent->ram_size - 1;
 
-			/*
-			 * if page_started, cow_file_range inserted an
-			 * inline extent and took care of all the unlocking
-			 * and IO for us.  Otherwise, we need to submit
-			 * all those pages down to the drive.
-			 */
-			if (!page_started && !ret)
-				extent_write_locked_range(inode,
-						  async_extent->start,
-						  async_extent->start +
-						  async_extent->ram_size - 1,
-						  WB_SYNC_ALL);
-			else if (ret && async_chunk->locked_page)
-				unlock_page(async_chunk->locked_page);
-			kfree(async_extent);
-			cond_resched();
-			continue;
-		}
+	/*
+	 * If async_chunk->locked_page is in the async_extent range, we need to
+	 * handle it.
+	 */
+	if (async_chunk->locked_page) {
+		u64 locked_page_start = page_offset(async_chunk->locked_page);
+		u64 locked_page_end = locked_page_start + PAGE_SIZE - 1;
 
-		ret = btrfs_reserve_extent(root, async_extent->ram_size,
-					   async_extent->compressed_size,
-					   async_extent->compressed_size,
-					   0, alloc_hint, &ins, 1, 1);
-		if (ret) {
-			free_async_extent_pages(async_extent);
-
-			if (ret == -ENOSPC) {
-				unlock_extent(io_tree, async_extent->start,
-					      async_extent->start +
-					      async_extent->ram_size - 1);
-
-				/*
-				 * we need to redirty the pages if we decide to
-				 * fallback to uncompressed IO, otherwise we
-				 * will not submit these pages down to lower
-				 * layers.
-				 */
-				extent_range_redirty_for_io(inode,
-						async_extent->start,
-						async_extent->start +
-						async_extent->ram_size - 1);
-
-				goto retry;
-			}
-			goto out_free;
-		}
-		/*
-		 * here we're doing allocation and writeback of the
-		 * compressed pages
-		 */
-		em = create_io_em(inode, async_extent->start,
-				  async_extent->ram_size, /* len */
-				  async_extent->start, /* orig_start */
-				  ins.objectid, /* block_start */
-				  ins.offset, /* block_len */
-				  ins.offset, /* orig_block_len */
-				  async_extent->ram_size, /* ram_bytes */
-				  async_extent->compress_type,
-				  BTRFS_ORDERED_COMPRESSED);
-		if (IS_ERR(em))
-			/* ret value is not necessary due to void function */
-			goto out_free_reserve;
-		free_extent_map(em);
+		if (!(start >= locked_page_end || end <= locked_page_start))
+			locked_page = async_chunk->locked_page;
+	}
+	lock_extent(io_tree, start, end, NULL);
 
-		ret = btrfs_add_ordered_extent_compress(inode,
-						async_extent->start,
-						ins.objectid,
-						async_extent->ram_size,
-						ins.offset,
-						BTRFS_ORDERED_COMPRESSED,
-						async_extent->compress_type);
-		if (ret) {
-			btrfs_drop_extent_cache(BTRFS_I(inode),
-						async_extent->start,
-						async_extent->start +
-						async_extent->ram_size - 1, 0);
-			goto out_free_reserve;
-		}
-		btrfs_dec_block_group_reservations(fs_info, ins.objectid);
+	/* We have fall back to uncompressed write */
+	if (!async_extent->pages)
+		return submit_uncompressed_range(inode, async_extent, locked_page);
 
+	ret = btrfs_reserve_extent(root, async_extent->ram_size,
+				   async_extent->compressed_size,
+				   async_extent->compressed_size,
+				   0, *alloc_hint, &ins, 1, 1);
+	if (ret) {
+		free_async_extent_pages(async_extent);
 		/*
-		 * clear dirty, set writeback and unlock the pages.
+		 * Here we used to try again by going back to non-compressed
+		 * path for ENOSPC.  But we can't reserve space even for
+		 * compressed size, how could it work for uncompressed size
+		 * which requires larger size?  So here we directly go error
+		 * path.
 		 */
-		extent_clear_unlock_delalloc(inode, async_extent->start,
-				async_extent->start +
-				async_extent->ram_size - 1,
-				NULL, EXTENT_LOCKED | EXTENT_DELALLOC,
-				PAGE_UNLOCK | PAGE_CLEAR_DIRTY |
-				PAGE_SET_WRITEBACK);
-		if (btrfs_submit_compressed_write(inode,
-				    async_extent->start,
-				    async_extent->ram_size,
-				    ins.objectid,
-				    ins.offset, async_extent->pages,
-				    async_extent->nr_pages,
-				    async_chunk->write_flags,
-				    async_chunk->blkcg_css)) {
-			struct page *p = async_extent->pages[0];
-			const u64 start = async_extent->start;
-			const u64 end = start + async_extent->ram_size - 1;
-
-			p->mapping = inode->i_mapping;
-			btrfs_writepage_endio_finish_ordered(p, start, end, 0);
-
-			p->mapping = NULL;
-			extent_clear_unlock_delalloc(inode, start, end,
-						     NULL, 0,
-						     PAGE_END_WRITEBACK |
-						     PAGE_SET_ERROR);
-			free_async_extent_pages(async_extent);
-		}
-		alloc_hint = ins.objectid + ins.offset;
-		kfree(async_extent);
-		cond_resched();
+		goto out_free;
+	}
+
+	/* Here we're doing allocation and writeback of the compressed pages */
+	em = create_io_em(inode, start,
+			  async_extent->ram_size,	/* len */
+			  start,			/* orig_start */
+			  ins.objectid,			/* block_start */
+			  ins.offset,			/* block_len */
+			  ins.offset,			/* orig_block_len */
+			  async_extent->ram_size,	/* ram_bytes */
+			  async_extent->compress_type,
+			  BTRFS_ORDERED_COMPRESSED);
+	if (IS_ERR(em)) {
+		ret = PTR_ERR(em);
+		goto out_free_reserve;
 	}
-	return;
+	free_extent_map(em);
+
+	ret = btrfs_add_ordered_extent(inode, start,		/* file_offset */
+				       async_extent->ram_size,	/* num_bytes */
+				       async_extent->ram_size,	/* ram_bytes */
+				       ins.objectid,		/* disk_bytenr */
+				       ins.offset,		/* disk_num_bytes */
+				       0,			/* offset */
+				       1 << BTRFS_ORDERED_COMPRESSED,
+				       async_extent->compress_type);
+	if (ret) {
+		btrfs_drop_extent_map_range(inode, start, end, false);
+		goto out_free_reserve;
+	}
+	btrfs_dec_block_group_reservations(fs_info, ins.objectid);
+
+	/* Clear dirty, set writeback and unlock the pages. */
+	extent_clear_unlock_delalloc(inode, start, end,
+			NULL, EXTENT_LOCKED | EXTENT_DELALLOC,
+			PAGE_UNLOCK | PAGE_START_WRITEBACK);
+	if (btrfs_submit_compressed_write(inode, start,	/* file_offset */
+			    async_extent->ram_size,	/* num_bytes */
+			    ins.objectid,		/* disk_bytenr */
+			    ins.offset,			/* compressed_len */
+			    async_extent->pages,	/* compressed_pages */
+			    async_extent->nr_pages,
+			    async_chunk->write_flags,
+			    async_chunk->blkcg_css, true)) {
+		const u64 start = async_extent->start;
+		const u64 end = start + async_extent->ram_size - 1;
+
+		btrfs_writepage_endio_finish_ordered(inode, NULL, start, end, 0);
+
+		extent_clear_unlock_delalloc(inode, start, end, NULL, 0,
+					     PAGE_END_WRITEBACK | PAGE_SET_ERROR);
+		free_async_extent_pages(async_extent);
+	}
+	*alloc_hint = ins.objectid + ins.offset;
+	kfree(async_extent);
+	return ret;
+
 out_free_reserve:
 	btrfs_dec_block_group_reservations(fs_info, ins.objectid);
 	btrfs_free_reserved_extent(fs_info, ins.objectid, ins.offset, 1);
 out_free:
-	extent_clear_unlock_delalloc(inode, async_extent->start,
-				     async_extent->start +
-				     async_extent->ram_size - 1,
+	extent_clear_unlock_delalloc(inode, start, end,
 				     NULL, EXTENT_LOCKED | EXTENT_DELALLOC |
 				     EXTENT_DELALLOC_NEW |
 				     EXTENT_DEFRAG | EXTENT_DO_ACCOUNTING,
-				     PAGE_UNLOCK | PAGE_CLEAR_DIRTY |
-				     PAGE_SET_WRITEBACK | PAGE_END_WRITEBACK |
-				     PAGE_SET_ERROR);
+				     PAGE_UNLOCK | PAGE_START_WRITEBACK |
+				     PAGE_END_WRITEBACK | PAGE_SET_ERROR);
 	free_async_extent_pages(async_extent);
 	kfree(async_extent);
-	goto again;
+	return ret;
+}
+
+/*
+ * Phase two of compressed writeback.  This is the ordered portion of the code,
+ * which only gets called in the order the work was queued.  We walk all the
+ * async extents created by compress_file_range and send them down to the disk.
+ */
+static noinline void submit_compressed_extents(struct async_chunk *async_chunk)
+{
+	struct btrfs_inode *inode = BTRFS_I(async_chunk->inode);
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
+	struct async_extent *async_extent;
+	u64 alloc_hint = 0;
+	int ret = 0;
+
+	while (!list_empty(&async_chunk->extents)) {
+		u64 extent_start;
+		u64 ram_size;
+
+		async_extent = list_entry(async_chunk->extents.next,
+					  struct async_extent, list);
+		list_del(&async_extent->list);
+		extent_start = async_extent->start;
+		ram_size = async_extent->ram_size;
+
+		ret = submit_one_async_extent(inode, async_chunk, async_extent,
+					      &alloc_hint);
+		btrfs_debug(fs_info,
+"async extent submission failed root=%lld inode=%llu start=%llu len=%llu ret=%d",
+			    inode->root->root_key.objectid,
+			    btrfs_ino(inode), extent_start, ram_size, ret);
+	}
 }
 
-static u64 get_extent_allocation_hint(struct inode *inode, u64 start,
+static u64 get_extent_allocation_hint(struct btrfs_inode *inode, u64 start,
 				      u64 num_bytes)
 {
-	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
+	struct extent_map_tree *em_tree = &inode->extent_tree;
 	struct extent_map *em;
 	u64 alloc_hint = 0;
 
@@ -962,18 +1145,43 @@ static u64 get_extent_allocation_hint(struct inode *inode, u64 start,
  * *page_started is set to one if we unlock locked_page and do everything
  * required to start IO on it.  It may be clean and already done with
  * IO when we return.
+ *
+ * When unlock == 1, we unlock the pages in successfully allocated regions.
+ * When unlock == 0, we leave them locked for writing them out.
+ *
+ * However, we unlock all the pages except @locked_page in case of failure.
+ *
+ * In summary, page locking state will be as follow:
+ *
+ * - page_started == 1 (return value)
+ *     - All the pages are unlocked. IO is started.
+ *     - Note that this can happen only on success
+ * - unlock == 1
+ *     - All the pages except @locked_page are unlocked in any case
+ * - unlock == 0
+ *     - On success, all the pages are locked for writing out them
+ *     - On failure, all the pages except @locked_page are unlocked
+ *
+ * When a failure happens in the second or later iteration of the
+ * while-loop, the ordered extents created in previous iterations are kept
+ * intact. So, the caller must clean them up by calling
+ * btrfs_cleanup_ordered_extents(). See btrfs_run_delalloc_range() for
+ * example.
  */
-static noinline int cow_file_range(struct inode *inode,
+static noinline int cow_file_range(struct btrfs_inode *inode,
 				   struct page *locked_page,
 				   u64 start, u64 end, int *page_started,
-				   unsigned long *nr_written, int unlock)
+				   unsigned long *nr_written, int unlock,
+				   u64 *done_offset)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	struct btrfs_root *root = BTRFS_I(inode)->root;
+	struct btrfs_root *root = inode->root;
+	struct btrfs_fs_info *fs_info = root->fs_info;
 	u64 alloc_hint = 0;
+	u64 orig_start = start;
 	u64 num_bytes;
 	unsigned long ram_size;
 	u64 cur_alloc_size = 0;
+	u64 min_alloc_size;
 	u64 blocksize = fs_info->sectorsize;
 	struct btrfs_key ins;
 	struct extent_map *em;
@@ -982,8 +1190,7 @@ static noinline int cow_file_range(struct inode *inode,
 	bool extent_reserved = false;
 	int ret = 0;
 
-	if (btrfs_is_free_space_inode(BTRFS_I(inode))) {
-		WARN_ON_ONCE(1);
+	if (btrfs_is_free_space_inode(inode)) {
 		ret = -EINVAL;
 		goto out_unlock;
 	}
@@ -992,12 +1199,25 @@ static noinline int cow_file_range(struct inode *inode,
 	num_bytes = max(blocksize,  num_bytes);
 	ASSERT(num_bytes <= btrfs_super_total_bytes(fs_info->super_copy));
 
-	inode_should_defrag(BTRFS_I(inode), start, end, num_bytes, SZ_64K);
+	inode_should_defrag(inode, start, end, num_bytes, SZ_64K);
+
+	/*
+	 * Due to the page size limit, for subpage we can only trigger the
+	 * writeback for the dirty sectors of page, that means data writeback
+	 * is doing more writeback than what we want.
+	 *
+	 * This is especially unexpected for some call sites like fallocate,
+	 * where we only increase i_size after everything is done.
+	 * This means we can trigger inline extent even if we didn't want to.
+	 * So here we skip inline extent creation completely.
+	 */
+	if (start == 0 && fs_info->sectorsize == PAGE_SIZE) {
+		u64 actual_end = min_t(u64, i_size_read(&inode->vfs_inode),
+				       end + 1);
 
-	if (start == 0) {
 		/* lets try to make an inline extent */
-		ret = cow_file_range_inline(inode, start, end, 0,
-					    BTRFS_COMPRESS_NONE, NULL);
+		ret = cow_file_range_inline(inode, actual_end, 0,
+					    BTRFS_COMPRESS_NONE, NULL, false);
 		if (ret == 0) {
 			/*
 			 * We use DO_ACCOUNTING here because we need the
@@ -1005,15 +1225,28 @@ static noinline int cow_file_range(struct inode *inode,
 			 * our outstanding extent for clearing delalloc for this
 			 * range.
 			 */
-			extent_clear_unlock_delalloc(inode, start, end, NULL,
+			extent_clear_unlock_delalloc(inode, start, end,
+				     locked_page,
 				     EXTENT_LOCKED | EXTENT_DELALLOC |
 				     EXTENT_DELALLOC_NEW | EXTENT_DEFRAG |
 				     EXTENT_DO_ACCOUNTING, PAGE_UNLOCK |
-				     PAGE_CLEAR_DIRTY | PAGE_SET_WRITEBACK |
-				     PAGE_END_WRITEBACK);
+				     PAGE_START_WRITEBACK | PAGE_END_WRITEBACK);
 			*nr_written = *nr_written +
 			     (end - start + PAGE_SIZE) / PAGE_SIZE;
 			*page_started = 1;
+			/*
+			 * locked_page is locked by the caller of
+			 * writepage_delalloc(), not locked by
+			 * __process_pages_contig().
+			 *
+			 * We can't let __process_pages_contig() to unlock it,
+			 * as it doesn't have any subpage::writers recorded.
+			 *
+			 * Here we manually unlock the page, since the caller
+			 * can't use page_started to determine if it's an
+			 * inline extent or a compressed extent.
+			 */
+			unlock_page(locked_page);
 			goto out;
 		} else if (ret < 0) {
 			goto out_unlock;
@@ -1021,13 +1254,27 @@ static noinline int cow_file_range(struct inode *inode,
 	}
 
 	alloc_hint = get_extent_allocation_hint(inode, start, num_bytes);
-	btrfs_drop_extent_cache(BTRFS_I(inode), start,
-			start + num_bytes - 1, 0);
+
+	/*
+	 * Relocation relies on the relocated extents to have exactly the same
+	 * size as the original extents. Normally writeback for relocation data
+	 * extents follows a NOCOW path because relocation preallocates the
+	 * extents. However, due to an operation such as scrub turning a block
+	 * group to RO mode, it may fallback to COW mode, so we must make sure
+	 * an extent allocated during COW has exactly the requested size and can
+	 * not be split into smaller extents, otherwise relocation breaks and
+	 * fails during the stage where it updates the bytenr of file extent
+	 * items.
+	 */
+	if (btrfs_is_data_reloc_root(root))
+		min_alloc_size = num_bytes;
+	else
+		min_alloc_size = fs_info->sectorsize;
 
 	while (num_bytes > 0) {
 		cur_alloc_size = num_bytes;
 		ret = btrfs_reserve_extent(root, cur_alloc_size, cur_alloc_size,
-					   fs_info->sectorsize, 0, alloc_hint,
+					   min_alloc_size, 0, alloc_hint,
 					   &ins, 1, 1);
 		if (ret < 0)
 			goto out_unlock;
@@ -1049,13 +1296,14 @@ static noinline int cow_file_range(struct inode *inode,
 		}
 		free_extent_map(em);
 
-		ret = btrfs_add_ordered_extent(inode, start, ins.objectid,
-					       ram_size, cur_alloc_size, 0);
+		ret = btrfs_add_ordered_extent(inode, start, ram_size, ram_size,
+					       ins.objectid, cur_alloc_size, 0,
+					       1 << BTRFS_ORDERED_REGULAR,
+					       BTRFS_COMPRESS_NONE);
 		if (ret)
 			goto out_drop_extent_cache;
 
-		if (root->root_key.objectid ==
-		    BTRFS_DATA_RELOC_TREE_OBJECTID) {
+		if (btrfs_is_data_reloc_root(root)) {
 			ret = btrfs_reloc_clone_csums(inode, start,
 						      cur_alloc_size);
 			/*
@@ -1070,24 +1318,25 @@ static noinline int cow_file_range(struct inode *inode,
 			 * skip current ordered extent.
 			 */
 			if (ret)
-				btrfs_drop_extent_cache(BTRFS_I(inode), start,
-						start + ram_size - 1, 0);
+				btrfs_drop_extent_map_range(inode, start,
+							    start + ram_size - 1,
+							    false);
 		}
 
 		btrfs_dec_block_group_reservations(fs_info, ins.objectid);
 
-		/* we're not doing compressed IO, don't unlock the first
-		 * page (which the caller expects to stay locked), don't
-		 * clear any dirty bits and don't set any writeback bits
+		/*
+		 * We're not doing compressed IO, don't unlock the first page
+		 * (which the caller expects to stay locked), don't clear any
+		 * dirty bits and don't set any writeback bits
 		 *
-		 * Do set the Private2 bit so we know this page was properly
-		 * setup for writepage
+		 * Do set the Ordered (Private2) bit so we know this page was
+		 * properly setup for writepage.
 		 */
 		page_ops = unlock ? PAGE_UNLOCK : 0;
-		page_ops |= PAGE_SET_PRIVATE2;
+		page_ops |= PAGE_SET_ORDERED;
 
-		extent_clear_unlock_delalloc(inode, start,
-					     start + ram_size - 1,
+		extent_clear_unlock_delalloc(inode, start, start + ram_size - 1,
 					     locked_page,
 					     EXTENT_LOCKED | EXTENT_DELALLOC,
 					     page_ops);
@@ -1111,39 +1360,89 @@ out:
 	return ret;
 
 out_drop_extent_cache:
-	btrfs_drop_extent_cache(BTRFS_I(inode), start, start + ram_size - 1, 0);
+	btrfs_drop_extent_map_range(inode, start, start + ram_size - 1, false);
 out_reserve:
 	btrfs_dec_block_group_reservations(fs_info, ins.objectid);
 	btrfs_free_reserved_extent(fs_info, ins.objectid, ins.offset, 1);
 out_unlock:
+	/*
+	 * If done_offset is non-NULL and ret == -EAGAIN, we expect the
+	 * caller to write out the successfully allocated region and retry.
+	 */
+	if (done_offset && ret == -EAGAIN) {
+		if (orig_start < start)
+			*done_offset = start - 1;
+		else
+			*done_offset = start;
+		return ret;
+	} else if (ret == -EAGAIN) {
+		/* Convert to -ENOSPC since the caller cannot retry. */
+		ret = -ENOSPC;
+	}
+
+	/*
+	 * Now, we have three regions to clean up:
+	 *
+	 * |-------(1)----|---(2)---|-------------(3)----------|
+	 * `- orig_start  `- start  `- start + cur_alloc_size  `- end
+	 *
+	 * We process each region below.
+	 */
+
 	clear_bits = EXTENT_LOCKED | EXTENT_DELALLOC | EXTENT_DELALLOC_NEW |
 		EXTENT_DEFRAG | EXTENT_CLEAR_META_RESV;
-	page_ops = PAGE_UNLOCK | PAGE_CLEAR_DIRTY | PAGE_SET_WRITEBACK |
-		PAGE_END_WRITEBACK;
-	/*
-	 * If we reserved an extent for our delalloc range (or a subrange) and
-	 * failed to create the respective ordered extent, then it means that
-	 * when we reserved the extent we decremented the extent's size from
-	 * the data space_info's bytes_may_use counter and incremented the
-	 * space_info's bytes_reserved counter by the same amount. We must make
-	 * sure extent_clear_unlock_delalloc() does not try to decrement again
-	 * the data space_info's bytes_may_use counter, therefore we do not pass
-	 * it the flag EXTENT_CLEAR_DATA_RESV.
+	page_ops = PAGE_UNLOCK | PAGE_START_WRITEBACK | PAGE_END_WRITEBACK;
+
+	/*
+	 * For the range (1). We have already instantiated the ordered extents
+	 * for this region. They are cleaned up by
+	 * btrfs_cleanup_ordered_extents() in e.g,
+	 * btrfs_run_delalloc_range(). EXTENT_LOCKED | EXTENT_DELALLOC are
+	 * already cleared in the above loop. And, EXTENT_DELALLOC_NEW |
+	 * EXTENT_DEFRAG | EXTENT_CLEAR_META_RESV are handled by the cleanup
+	 * function.
+	 *
+	 * However, in case of unlock == 0, we still need to unlock the pages
+	 * (except @locked_page) to ensure all the pages are unlocked.
+	 */
+	if (!unlock && orig_start < start) {
+		if (!locked_page)
+			mapping_set_error(inode->vfs_inode.i_mapping, ret);
+		extent_clear_unlock_delalloc(inode, orig_start, start - 1,
+					     locked_page, 0, page_ops);
+	}
+
+	/*
+	 * For the range (2). If we reserved an extent for our delalloc range
+	 * (or a subrange) and failed to create the respective ordered extent,
+	 * then it means that when we reserved the extent we decremented the
+	 * extent's size from the data space_info's bytes_may_use counter and
+	 * incremented the space_info's bytes_reserved counter by the same
+	 * amount. We must make sure extent_clear_unlock_delalloc() does not try
+	 * to decrement again the data space_info's bytes_may_use counter,
+	 * therefore we do not pass it the flag EXTENT_CLEAR_DATA_RESV.
 	 */
 	if (extent_reserved) {
 		extent_clear_unlock_delalloc(inode, start,
-					     start + cur_alloc_size,
+					     start + cur_alloc_size - 1,
 					     locked_page,
 					     clear_bits,
 					     page_ops);
 		start += cur_alloc_size;
 		if (start >= end)
-			goto out;
+			return ret;
 	}
+
+	/*
+	 * For the range (3). We never touched the region. In addition to the
+	 * clear_bits above, we add EXTENT_CLEAR_DATA_RESV to release the data
+	 * space_info's bytes_may_use counter, reserved in
+	 * btrfs_check_data_free_space().
+	 */
 	extent_clear_unlock_delalloc(inode, start, end, locked_page,
 				     clear_bits | EXTENT_CLEAR_DATA_RESV,
 				     page_ops);
-	goto out;
+	return ret;
 }
 
 /*
@@ -1176,11 +1475,6 @@ static noinline void async_cow_submit(struct btrfs_work *work)
 	nr_pages = (async_chunk->end - async_chunk->start + PAGE_SIZE) >>
 		PAGE_SHIFT;
 
-	/* atomic_sub_return implies a barrier */
-	if (atomic_sub_return(nr_pages, &fs_info->async_delalloc_pages) <
-	    5 * SZ_1M)
-		cond_wake_up_nomb(&fs_info->async_submit_wait);
-
 	/*
 	 * ->inode could be NULL if async_chunk_start has failed to compress,
 	 * in which case we don't have anything to submit, yet we need to
@@ -1189,32 +1483,36 @@ static noinline void async_cow_submit(struct btrfs_work *work)
 	 */
 	if (async_chunk->inode)
 		submit_compressed_extents(async_chunk);
+
+	/* atomic_sub_return implies a barrier */
+	if (atomic_sub_return(nr_pages, &fs_info->async_delalloc_pages) <
+	    5 * SZ_1M)
+		cond_wake_up_nomb(&fs_info->async_submit_wait);
 }
 
 static noinline void async_cow_free(struct btrfs_work *work)
 {
 	struct async_chunk *async_chunk;
+	struct async_cow *async_cow;
 
 	async_chunk = container_of(work, struct async_chunk, work);
 	if (async_chunk->inode)
 		btrfs_add_delayed_iput(async_chunk->inode);
 	if (async_chunk->blkcg_css)
 		css_put(async_chunk->blkcg_css);
-	/*
-	 * Since the pointer to 'pending' is at the beginning of the array of
-	 * async_chunk's, freeing it ensures the whole array has been freed.
-	 */
-	if (atomic_dec_and_test(async_chunk->pending))
-		kvfree(async_chunk->pending);
+
+	async_cow = async_chunk->async_cow;
+	if (atomic_dec_and_test(&async_cow->num_chunks))
+		kvfree(async_cow);
 }
 
-static int cow_file_range_async(struct inode *inode,
+static int cow_file_range_async(struct btrfs_inode *inode,
 				struct writeback_control *wbc,
 				struct page *locked_page,
 				u64 start, u64 end, int *page_started,
 				unsigned long *nr_written)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
 	struct cgroup_subsys_state *blkcg_css = wbc_blkcg_css(wbc);
 	struct async_cow *ctx;
 	struct async_chunk *async_chunk;
@@ -1224,11 +1522,11 @@ static int cow_file_range_async(struct inode *inode,
 	int i;
 	bool should_compress;
 	unsigned nofs_flag;
-	const unsigned int write_flags = wbc_to_write_flags(wbc);
+	const blk_opf_t write_flags = wbc_to_write_flags(wbc);
 
-	unlock_extent(&BTRFS_I(inode)->io_tree, start, end);
+	unlock_extent(&inode->io_tree, start, end, NULL);
 
-	if (BTRFS_I(inode)->flags & BTRFS_INODE_NOCOMPRESS &&
+	if (inode->flags & BTRFS_INODE_NOCOMPRESS &&
 	    !btrfs_test_opt(fs_info, FORCE_COMPRESS)) {
 		num_chunks = 1;
 		should_compress = false;
@@ -1244,9 +1542,8 @@ static int cow_file_range_async(struct inode *inode,
 		unsigned clear_bits = EXTENT_LOCKED | EXTENT_DELALLOC |
 			EXTENT_DELALLOC_NEW | EXTENT_DEFRAG |
 			EXTENT_DO_ACCOUNTING;
-		unsigned long page_ops = PAGE_UNLOCK | PAGE_CLEAR_DIRTY |
-			PAGE_SET_WRITEBACK | PAGE_END_WRITEBACK |
-			PAGE_SET_ERROR;
+		unsigned long page_ops = PAGE_UNLOCK | PAGE_START_WRITEBACK |
+					 PAGE_END_WRITEBACK | PAGE_SET_ERROR;
 
 		extent_clear_unlock_delalloc(inode, start, end, locked_page,
 					     clear_bits, page_ops);
@@ -1266,9 +1563,9 @@ static int cow_file_range_async(struct inode *inode,
 		 * igrab is called higher up in the call chain, take only the
 		 * lightweight reference for the callback lifetime
 		 */
-		ihold(inode);
-		async_chunk[i].pending = &ctx->num_chunks;
-		async_chunk[i].inode = inode;
+		ihold(&inode->vfs_inode);
+		async_chunk[i].async_cow = ctx;
+		async_chunk[i].inode = &inode->vfs_inode;
 		async_chunk[i].start = start;
 		async_chunk[i].end = cur_end;
 		async_chunk[i].write_flags = write_flags;
@@ -1323,15 +1620,62 @@ static int cow_file_range_async(struct inode *inode,
 	return 0;
 }
 
-static noinline int csum_exist_in_range(struct btrfs_fs_info *fs_info,
-					u64 bytenr, u64 num_bytes)
+static noinline int run_delalloc_zoned(struct btrfs_inode *inode,
+				       struct page *locked_page, u64 start,
+				       u64 end, int *page_started,
+				       unsigned long *nr_written)
 {
+	u64 done_offset = end;
 	int ret;
+	bool locked_page_done = false;
+
+	while (start <= end) {
+		ret = cow_file_range(inode, locked_page, start, end, page_started,
+				     nr_written, 0, &done_offset);
+		if (ret && ret != -EAGAIN)
+			return ret;
+
+		if (*page_started) {
+			ASSERT(ret == 0);
+			return 0;
+		}
+
+		if (ret == 0)
+			done_offset = end;
+
+		if (done_offset == start) {
+			wait_on_bit_io(&inode->root->fs_info->flags,
+				       BTRFS_FS_NEED_ZONE_FINISH,
+				       TASK_UNINTERRUPTIBLE);
+			continue;
+		}
+
+		if (!locked_page_done) {
+			__set_page_dirty_nobuffers(locked_page);
+			account_page_redirty(locked_page);
+		}
+		locked_page_done = true;
+		extent_write_locked_range(&inode->vfs_inode, start, done_offset);
+
+		start = done_offset + 1;
+	}
+
+	*page_started = 1;
+
+	return 0;
+}
+
+static noinline int csum_exist_in_range(struct btrfs_fs_info *fs_info,
+					u64 bytenr, u64 num_bytes, bool nowait)
+{
+	struct btrfs_root *csum_root = btrfs_csum_root(fs_info, bytenr);
 	struct btrfs_ordered_sum *sums;
+	int ret;
 	LIST_HEAD(list);
 
-	ret = btrfs_lookup_csums_range(fs_info->csum_root, bytenr,
-				       bytenr + num_bytes - 1, &list, 0);
+	ret = btrfs_lookup_csums_range(csum_root, bytenr,
+				       bytenr + num_bytes - 1, &list, 0,
+				       nowait);
 	if (ret == 0 && list_empty(&list))
 		return 0;
 
@@ -1345,6 +1689,209 @@ static noinline int csum_exist_in_range(struct btrfs_fs_info *fs_info,
 	return 1;
 }
 
+static int fallback_to_cow(struct btrfs_inode *inode, struct page *locked_page,
+			   const u64 start, const u64 end,
+			   int *page_started, unsigned long *nr_written)
+{
+	const bool is_space_ino = btrfs_is_free_space_inode(inode);
+	const bool is_reloc_ino = btrfs_is_data_reloc_root(inode->root);
+	const u64 range_bytes = end + 1 - start;
+	struct extent_io_tree *io_tree = &inode->io_tree;
+	u64 range_start = start;
+	u64 count;
+
+	/*
+	 * If EXTENT_NORESERVE is set it means that when the buffered write was
+	 * made we had not enough available data space and therefore we did not
+	 * reserve data space for it, since we though we could do NOCOW for the
+	 * respective file range (either there is prealloc extent or the inode
+	 * has the NOCOW bit set).
+	 *
+	 * However when we need to fallback to COW mode (because for example the
+	 * block group for the corresponding extent was turned to RO mode by a
+	 * scrub or relocation) we need to do the following:
+	 *
+	 * 1) We increment the bytes_may_use counter of the data space info.
+	 *    If COW succeeds, it allocates a new data extent and after doing
+	 *    that it decrements the space info's bytes_may_use counter and
+	 *    increments its bytes_reserved counter by the same amount (we do
+	 *    this at btrfs_add_reserved_bytes()). So we need to increment the
+	 *    bytes_may_use counter to compensate (when space is reserved at
+	 *    buffered write time, the bytes_may_use counter is incremented);
+	 *
+	 * 2) We clear the EXTENT_NORESERVE bit from the range. We do this so
+	 *    that if the COW path fails for any reason, it decrements (through
+	 *    extent_clear_unlock_delalloc()) the bytes_may_use counter of the
+	 *    data space info, which we incremented in the step above.
+	 *
+	 * If we need to fallback to cow and the inode corresponds to a free
+	 * space cache inode or an inode of the data relocation tree, we must
+	 * also increment bytes_may_use of the data space_info for the same
+	 * reason. Space caches and relocated data extents always get a prealloc
+	 * extent for them, however scrub or balance may have set the block
+	 * group that contains that extent to RO mode and therefore force COW
+	 * when starting writeback.
+	 */
+	count = count_range_bits(io_tree, &range_start, end, range_bytes,
+				 EXTENT_NORESERVE, 0);
+	if (count > 0 || is_space_ino || is_reloc_ino) {
+		u64 bytes = count;
+		struct btrfs_fs_info *fs_info = inode->root->fs_info;
+		struct btrfs_space_info *sinfo = fs_info->data_sinfo;
+
+		if (is_space_ino || is_reloc_ino)
+			bytes = range_bytes;
+
+		spin_lock(&sinfo->lock);
+		btrfs_space_info_update_bytes_may_use(fs_info, sinfo, bytes);
+		spin_unlock(&sinfo->lock);
+
+		if (count > 0)
+			clear_extent_bit(io_tree, start, end, EXTENT_NORESERVE,
+					 NULL);
+	}
+
+	return cow_file_range(inode, locked_page, start, end, page_started,
+			      nr_written, 1, NULL);
+}
+
+struct can_nocow_file_extent_args {
+	/* Input fields. */
+
+	/* Start file offset of the range we want to NOCOW. */
+	u64 start;
+	/* End file offset (inclusive) of the range we want to NOCOW. */
+	u64 end;
+	bool writeback_path;
+	bool strict;
+	/*
+	 * Free the path passed to can_nocow_file_extent() once it's not needed
+	 * anymore.
+	 */
+	bool free_path;
+
+	/* Output fields. Only set when can_nocow_file_extent() returns 1. */
+
+	u64 disk_bytenr;
+	u64 disk_num_bytes;
+	u64 extent_offset;
+	/* Number of bytes that can be written to in NOCOW mode. */
+	u64 num_bytes;
+};
+
+/*
+ * Check if we can NOCOW the file extent that the path points to.
+ * This function may return with the path released, so the caller should check
+ * if path->nodes[0] is NULL or not if it needs to use the path afterwards.
+ *
+ * Returns: < 0 on error
+ *            0 if we can not NOCOW
+ *            1 if we can NOCOW
+ */
+static int can_nocow_file_extent(struct btrfs_path *path,
+				 struct btrfs_key *key,
+				 struct btrfs_inode *inode,
+				 struct can_nocow_file_extent_args *args)
+{
+	const bool is_freespace_inode = btrfs_is_free_space_inode(inode);
+	struct extent_buffer *leaf = path->nodes[0];
+	struct btrfs_root *root = inode->root;
+	struct btrfs_file_extent_item *fi;
+	u64 extent_end;
+	u8 extent_type;
+	int can_nocow = 0;
+	int ret = 0;
+	bool nowait = path->nowait;
+
+	fi = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_file_extent_item);
+	extent_type = btrfs_file_extent_type(leaf, fi);
+
+	if (extent_type == BTRFS_FILE_EXTENT_INLINE)
+		goto out;
+
+	/* Can't access these fields unless we know it's not an inline extent. */
+	args->disk_bytenr = btrfs_file_extent_disk_bytenr(leaf, fi);
+	args->disk_num_bytes = btrfs_file_extent_disk_num_bytes(leaf, fi);
+	args->extent_offset = btrfs_file_extent_offset(leaf, fi);
+
+	if (!(inode->flags & BTRFS_INODE_NODATACOW) &&
+	    extent_type == BTRFS_FILE_EXTENT_REG)
+		goto out;
+
+	/*
+	 * If the extent was created before the generation where the last snapshot
+	 * for its subvolume was created, then this implies the extent is shared,
+	 * hence we must COW.
+	 */
+	if (!args->strict &&
+	    btrfs_file_extent_generation(leaf, fi) <=
+	    btrfs_root_last_snapshot(&root->root_item))
+		goto out;
+
+	/* An explicit hole, must COW. */
+	if (args->disk_bytenr == 0)
+		goto out;
+
+	/* Compressed/encrypted/encoded extents must be COWed. */
+	if (btrfs_file_extent_compression(leaf, fi) ||
+	    btrfs_file_extent_encryption(leaf, fi) ||
+	    btrfs_file_extent_other_encoding(leaf, fi))
+		goto out;
+
+	extent_end = btrfs_file_extent_end(path);
+
+	/*
+	 * The following checks can be expensive, as they need to take other
+	 * locks and do btree or rbtree searches, so release the path to avoid
+	 * blocking other tasks for too long.
+	 */
+	btrfs_release_path(path);
+
+	ret = btrfs_cross_ref_exist(root, btrfs_ino(inode),
+				    key->offset - args->extent_offset,
+				    args->disk_bytenr, false, path);
+	WARN_ON_ONCE(ret > 0 && is_freespace_inode);
+	if (ret != 0)
+		goto out;
+
+	if (args->free_path) {
+		/*
+		 * We don't need the path anymore, plus through the
+		 * csum_exist_in_range() call below we will end up allocating
+		 * another path. So free the path to avoid unnecessary extra
+		 * memory usage.
+		 */
+		btrfs_free_path(path);
+		path = NULL;
+	}
+
+	/* If there are pending snapshots for this root, we must COW. */
+	if (args->writeback_path && !is_freespace_inode &&
+	    atomic_read(&root->snapshot_force_cow))
+		goto out;
+
+	args->disk_bytenr += args->extent_offset;
+	args->disk_bytenr += args->start - key->offset;
+	args->num_bytes = min(args->end + 1, extent_end) - args->start;
+
+	/*
+	 * Force COW if csums exist in the range. This ensures that csums for a
+	 * given extent are either valid or do not exist.
+	 */
+	ret = csum_exist_in_range(root->fs_info, args->disk_bytenr, args->num_bytes,
+				  nowait);
+	WARN_ON_ONCE(ret > 0 && is_freespace_inode);
+	if (ret != 0)
+		goto out;
+
+	can_nocow = 1;
+ out:
+	if (args->free_path && path)
+		btrfs_free_path(path);
+
+	return ret < 0 ? ret : can_nocow;
+}
+
 /*
  * when nowcow writeback call back.  This checks for snapshots or COW copies
  * of the extents that exist in the file, and COWs the file as required.
@@ -1352,23 +1899,23 @@ static noinline int csum_exist_in_range(struct btrfs_fs_info *fs_info,
  * If no cow copies or snapshots exist, we write directly to the existing
  * blocks on disk
  */
-static noinline int run_delalloc_nocow(struct inode *inode,
+static noinline int run_delalloc_nocow(struct btrfs_inode *inode,
 				       struct page *locked_page,
 				       const u64 start, const u64 end,
-				       int *page_started, int force,
+				       int *page_started,
 				       unsigned long *nr_written)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	struct btrfs_root *root = BTRFS_I(inode)->root;
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
+	struct btrfs_root *root = inode->root;
 	struct btrfs_path *path;
 	u64 cow_start = (u64)-1;
 	u64 cur_offset = start;
 	int ret;
 	bool check_prev = true;
-	const bool freespace_inode = btrfs_is_free_space_inode(BTRFS_I(inode));
-	u64 ino = btrfs_ino(BTRFS_I(inode));
+	u64 ino = btrfs_ino(inode);
+	struct btrfs_block_group *bg;
 	bool nocow = false;
-	u64 disk_bytenr = 0;
+	struct can_nocow_file_extent_args nocow_args = { 0 };
 
 	path = btrfs_alloc_path();
 	if (!path) {
@@ -1376,21 +1923,21 @@ static noinline int run_delalloc_nocow(struct inode *inode,
 					     EXTENT_LOCKED | EXTENT_DELALLOC |
 					     EXTENT_DO_ACCOUNTING |
 					     EXTENT_DEFRAG, PAGE_UNLOCK |
-					     PAGE_CLEAR_DIRTY |
-					     PAGE_SET_WRITEBACK |
+					     PAGE_START_WRITEBACK |
 					     PAGE_END_WRITEBACK);
 		return -ENOMEM;
 	}
 
+	nocow_args.end = end;
+	nocow_args.writeback_path = true;
+
 	while (1) {
 		struct btrfs_key found_key;
 		struct btrfs_file_extent_item *fi;
 		struct extent_buffer *leaf;
 		u64 extent_end;
-		u64 extent_offset;
-		u64 num_bytes = 0;
-		u64 disk_num_bytes;
 		u64 ram_bytes;
+		u64 nocow_end;
 		int extent_type;
 
 		nocow = false;
@@ -1466,109 +2013,38 @@ next_slot:
 		fi = btrfs_item_ptr(leaf, path->slots[0],
 				    struct btrfs_file_extent_item);
 		extent_type = btrfs_file_extent_type(leaf, fi);
-
+		/* If this is triggered then we have a memory corruption. */
+		ASSERT(extent_type < BTRFS_NR_FILE_EXTENT_TYPES);
+		if (WARN_ON(extent_type >= BTRFS_NR_FILE_EXTENT_TYPES)) {
+			ret = -EUCLEAN;
+			goto error;
+		}
 		ram_bytes = btrfs_file_extent_ram_bytes(leaf, fi);
-		if (extent_type == BTRFS_FILE_EXTENT_REG ||
-		    extent_type == BTRFS_FILE_EXTENT_PREALLOC) {
-			disk_bytenr = btrfs_file_extent_disk_bytenr(leaf, fi);
-			extent_offset = btrfs_file_extent_offset(leaf, fi);
-			extent_end = found_key.offset +
-				btrfs_file_extent_num_bytes(leaf, fi);
-			disk_num_bytes =
-				btrfs_file_extent_disk_num_bytes(leaf, fi);
-			/*
-			 * If the extent we got ends before our current offset,
-			 * skip to the next extent.
-			 */
-			if (extent_end <= cur_offset) {
-				path->slots[0]++;
-				goto next_slot;
-			}
-			/* Skip holes */
-			if (disk_bytenr == 0)
-				goto out_check;
-			/* Skip compressed/encrypted/encoded extents */
-			if (btrfs_file_extent_compression(leaf, fi) ||
-			    btrfs_file_extent_encryption(leaf, fi) ||
-			    btrfs_file_extent_other_encoding(leaf, fi))
-				goto out_check;
-			/*
-			 * If extent is created before the last volume's snapshot
-			 * this implies the extent is shared, hence we can't do
-			 * nocow. This is the same check as in
-			 * btrfs_cross_ref_exist but without calling
-			 * btrfs_search_slot.
-			 */
-			if (!freespace_inode &&
-			    btrfs_file_extent_generation(leaf, fi) <=
-			    btrfs_root_last_snapshot(&root->root_item))
-				goto out_check;
-			if (extent_type == BTRFS_FILE_EXTENT_REG && !force)
-				goto out_check;
-			/* If extent is RO, we must COW it */
-			if (btrfs_extent_readonly(fs_info, disk_bytenr))
-				goto out_check;
-			ret = btrfs_cross_ref_exist(root, ino,
-						    found_key.offset -
-						    extent_offset, disk_bytenr);
-			if (ret) {
-				/*
-				 * ret could be -EIO if the above fails to read
-				 * metadata.
-				 */
-				if (ret < 0) {
-					if (cow_start != (u64)-1)
-						cur_offset = cow_start;
-					goto error;
-				}
+		extent_end = btrfs_file_extent_end(path);
 
-				WARN_ON_ONCE(freespace_inode);
-				goto out_check;
-			}
-			disk_bytenr += extent_offset;
-			disk_bytenr += cur_offset - found_key.offset;
-			num_bytes = min(end + 1, extent_end) - cur_offset;
-			/*
-			 * If there are pending snapshots for this root, we
-			 * fall into common COW way
-			 */
-			if (!freespace_inode && atomic_read(&root->snapshot_force_cow))
-				goto out_check;
-			/*
-			 * force cow if csum exists in the range.
-			 * this ensure that csum for a given extent are
-			 * either valid or do not exist.
-			 */
-			ret = csum_exist_in_range(fs_info, disk_bytenr,
-						  num_bytes);
-			if (ret) {
-				/*
-				 * ret could be -EIO if the above fails to read
-				 * metadata.
-				 */
-				if (ret < 0) {
-					if (cow_start != (u64)-1)
-						cur_offset = cow_start;
-					goto error;
-				}
-				WARN_ON_ONCE(freespace_inode);
-				goto out_check;
-			}
-			if (!btrfs_inc_nocow_writers(fs_info, disk_bytenr))
-				goto out_check;
-			nocow = true;
-		} else if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
-			extent_end = found_key.offset + ram_bytes;
-			extent_end = ALIGN(extent_end, fs_info->sectorsize);
-			/* Skip extents outside of our requested range */
-			if (extent_end <= start) {
-				path->slots[0]++;
-				goto next_slot;
-			}
-		} else {
-			/* If this triggers then we have a memory corruption */
-			BUG();
+		/*
+		 * If the extent we got ends before our current offset, skip to
+		 * the next extent.
+		 */
+		if (extent_end <= cur_offset) {
+			path->slots[0]++;
+			goto next_slot;
+		}
+
+		nocow_args.start = cur_offset;
+		ret = can_nocow_file_extent(path, &found_key, inode, &nocow_args);
+		if (ret < 0) {
+			if (cow_start != (u64)-1)
+				cur_offset = cow_start;
+			goto error;
+		} else if (ret == 0) {
+			goto out_check;
 		}
+
+		ret = 0;
+		bg = btrfs_inc_nocow_writers(fs_info, nocow_args.disk_bytenr);
+		if (bg)
+			nocow = true;
 out_check:
 		/*
 		 * If nocow is false then record the beginning of the range
@@ -1580,89 +2056,88 @@ out_check:
 			cur_offset = extent_end;
 			if (cur_offset > end)
 				break;
+			if (!path->nodes[0])
+				continue;
 			path->slots[0]++;
 			goto next_slot;
 		}
 
-		btrfs_release_path(path);
-
 		/*
 		 * COW range from cow_start to found_key.offset - 1. As the key
 		 * will contain the beginning of the first extent that can be
 		 * NOCOW, following one which needs to be COW'ed
 		 */
 		if (cow_start != (u64)-1) {
-			ret = cow_file_range(inode, locked_page,
-					     cow_start, found_key.offset - 1,
-					     page_started, nr_written, 1);
-			if (ret) {
-				if (nocow)
-					btrfs_dec_nocow_writers(fs_info,
-								disk_bytenr);
+			ret = fallback_to_cow(inode, locked_page,
+					      cow_start, found_key.offset - 1,
+					      page_started, nr_written);
+			if (ret)
 				goto error;
-			}
 			cow_start = (u64)-1;
 		}
 
+		nocow_end = cur_offset + nocow_args.num_bytes - 1;
+
 		if (extent_type == BTRFS_FILE_EXTENT_PREALLOC) {
-			u64 orig_start = found_key.offset - extent_offset;
+			u64 orig_start = found_key.offset - nocow_args.extent_offset;
 			struct extent_map *em;
 
-			em = create_io_em(inode, cur_offset, num_bytes,
+			em = create_io_em(inode, cur_offset, nocow_args.num_bytes,
 					  orig_start,
-					  disk_bytenr, /* block_start */
-					  num_bytes, /* block_len */
-					  disk_num_bytes, /* orig_block_len */
+					  nocow_args.disk_bytenr, /* block_start */
+					  nocow_args.num_bytes, /* block_len */
+					  nocow_args.disk_num_bytes, /* orig_block_len */
 					  ram_bytes, BTRFS_COMPRESS_NONE,
 					  BTRFS_ORDERED_PREALLOC);
 			if (IS_ERR(em)) {
-				if (nocow)
-					btrfs_dec_nocow_writers(fs_info,
-								disk_bytenr);
 				ret = PTR_ERR(em);
 				goto error;
 			}
 			free_extent_map(em);
-			ret = btrfs_add_ordered_extent(inode, cur_offset,
-						       disk_bytenr, num_bytes,
-						       num_bytes,
-						       BTRFS_ORDERED_PREALLOC);
+			ret = btrfs_add_ordered_extent(inode,
+					cur_offset, nocow_args.num_bytes,
+					nocow_args.num_bytes,
+					nocow_args.disk_bytenr,
+					nocow_args.num_bytes, 0,
+					1 << BTRFS_ORDERED_PREALLOC,
+					BTRFS_COMPRESS_NONE);
 			if (ret) {
-				btrfs_drop_extent_cache(BTRFS_I(inode),
-							cur_offset,
-							cur_offset + num_bytes - 1,
-							0);
+				btrfs_drop_extent_map_range(inode, cur_offset,
+							    nocow_end, false);
 				goto error;
 			}
 		} else {
 			ret = btrfs_add_ordered_extent(inode, cur_offset,
-						       disk_bytenr, num_bytes,
-						       num_bytes,
-						       BTRFS_ORDERED_NOCOW);
+						       nocow_args.num_bytes,
+						       nocow_args.num_bytes,
+						       nocow_args.disk_bytenr,
+						       nocow_args.num_bytes,
+						       0,
+						       1 << BTRFS_ORDERED_NOCOW,
+						       BTRFS_COMPRESS_NONE);
 			if (ret)
 				goto error;
 		}
 
-		if (nocow)
-			btrfs_dec_nocow_writers(fs_info, disk_bytenr);
-		nocow = false;
+		if (nocow) {
+			btrfs_dec_nocow_writers(bg);
+			nocow = false;
+		}
 
-		if (root->root_key.objectid ==
-		    BTRFS_DATA_RELOC_TREE_OBJECTID)
+		if (btrfs_is_data_reloc_root(root))
 			/*
 			 * Error handled later, as we must prevent
 			 * extent_clear_unlock_delalloc() in error handler
 			 * from freeing metadata of created ordered extent.
 			 */
 			ret = btrfs_reloc_clone_csums(inode, cur_offset,
-						      num_bytes);
+						      nocow_args.num_bytes);
 
-		extent_clear_unlock_delalloc(inode, cur_offset,
-					     cur_offset + num_bytes - 1,
+		extent_clear_unlock_delalloc(inode, cur_offset, nocow_end,
 					     locked_page, EXTENT_LOCKED |
 					     EXTENT_DELALLOC |
 					     EXTENT_CLEAR_DATA_RESV,
-					     PAGE_UNLOCK | PAGE_SET_PRIVATE2);
+					     PAGE_UNLOCK | PAGE_SET_ORDERED);
 
 		cur_offset = extent_end;
 
@@ -1683,75 +2158,82 @@ out_check:
 
 	if (cow_start != (u64)-1) {
 		cur_offset = end;
-		ret = cow_file_range(inode, locked_page, cow_start, end,
-				     page_started, nr_written, 1);
+		ret = fallback_to_cow(inode, locked_page, cow_start, end,
+				      page_started, nr_written);
 		if (ret)
 			goto error;
 	}
 
 error:
 	if (nocow)
-		btrfs_dec_nocow_writers(fs_info, disk_bytenr);
+		btrfs_dec_nocow_writers(bg);
 
 	if (ret && cur_offset < end)
 		extent_clear_unlock_delalloc(inode, cur_offset, end,
 					     locked_page, EXTENT_LOCKED |
 					     EXTENT_DELALLOC | EXTENT_DEFRAG |
 					     EXTENT_DO_ACCOUNTING, PAGE_UNLOCK |
-					     PAGE_CLEAR_DIRTY |
-					     PAGE_SET_WRITEBACK |
+					     PAGE_START_WRITEBACK |
 					     PAGE_END_WRITEBACK);
 	btrfs_free_path(path);
 	return ret;
 }
 
-static inline int need_force_cow(struct inode *inode, u64 start, u64 end)
+static bool should_nocow(struct btrfs_inode *inode, u64 start, u64 end)
 {
-
-	if (!(BTRFS_I(inode)->flags & BTRFS_INODE_NODATACOW) &&
-	    !(BTRFS_I(inode)->flags & BTRFS_INODE_PREALLOC))
-		return 0;
-
-	/*
-	 * @defrag_bytes is a hint value, no spinlock held here,
-	 * if is not zero, it means the file is defragging.
-	 * Force cow if given extent needs to be defragged.
-	 */
-	if (BTRFS_I(inode)->defrag_bytes &&
-	    test_range_bit(&BTRFS_I(inode)->io_tree, start, end,
-			   EXTENT_DEFRAG, 0, NULL))
-		return 1;
-
-	return 0;
+	if (inode->flags & (BTRFS_INODE_NODATACOW | BTRFS_INODE_PREALLOC)) {
+		if (inode->defrag_bytes &&
+		    test_range_bit(&inode->io_tree, start, end, EXTENT_DEFRAG,
+				   0, NULL))
+			return false;
+		return true;
+	}
+	return false;
 }
 
 /*
  * Function to process delayed allocation (create CoW) for ranges which are
  * being touched for the first time.
  */
-int btrfs_run_delalloc_range(struct inode *inode, struct page *locked_page,
+int btrfs_run_delalloc_range(struct btrfs_inode *inode, struct page *locked_page,
 		u64 start, u64 end, int *page_started, unsigned long *nr_written,
 		struct writeback_control *wbc)
 {
 	int ret;
-	int force_cow = need_force_cow(inode, start, end);
+	const bool zoned = btrfs_is_zoned(inode->root->fs_info);
 
-	if (BTRFS_I(inode)->flags & BTRFS_INODE_NODATACOW && !force_cow) {
-		ret = run_delalloc_nocow(inode, locked_page, start, end,
-					 page_started, 1, nr_written);
-	} else if (BTRFS_I(inode)->flags & BTRFS_INODE_PREALLOC && !force_cow) {
+	/*
+	 * The range must cover part of the @locked_page, or the returned
+	 * @page_started can confuse the caller.
+	 */
+	ASSERT(!(end <= page_offset(locked_page) ||
+		 start >= page_offset(locked_page) + PAGE_SIZE));
+
+	if (should_nocow(inode, start, end)) {
+		/*
+		 * Normally on a zoned device we're only doing COW writes, but
+		 * in case of relocation on a zoned filesystem we have taken
+		 * precaution, that we're only writing sequentially. It's safe
+		 * to use run_delalloc_nocow() here, like for  regular
+		 * preallocated inodes.
+		 */
+		ASSERT(!zoned || btrfs_is_data_reloc_root(inode->root));
 		ret = run_delalloc_nocow(inode, locked_page, start, end,
-					 page_started, 0, nr_written);
-	} else if (!inode_can_compress(inode) ||
+					 page_started, nr_written);
+	} else if (!btrfs_inode_can_compress(inode) ||
 		   !inode_need_compress(inode, start, end)) {
-		ret = cow_file_range(inode, locked_page, start, end,
-				      page_started, nr_written, 1);
+		if (zoned)
+			ret = run_delalloc_zoned(inode, locked_page, start, end,
+						 page_started, nr_written);
+		else
+			ret = cow_file_range(inode, locked_page, start, end,
+					     page_started, nr_written, 1, NULL);
 	} else {
-		set_bit(BTRFS_INODE_HAS_ASYNC_EXTENT,
-			&BTRFS_I(inode)->runtime_flags);
+		set_bit(BTRFS_INODE_HAS_ASYNC_EXTENT, &inode->runtime_flags);
 		ret = cow_file_range_async(inode, wbc, locked_page, start, end,
 					   page_started, nr_written);
 	}
+	ASSERT(ret <= 0);
 	if (ret)
 		btrfs_cleanup_ordered_extents(inode, locked_page, start,
 					      end - start + 1);
@@ -1761,6 +2243,7 @@ int btrfs_run_delalloc_range(struct inode *inode, struct page *locked_page,
 void btrfs_split_delalloc_extent(struct inode *inode,
 				 struct extent_state *orig, u64 split)
 {
+	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	u64 size;
 
 	/* not delalloc, ignore it */
@@ -1768,7 +2251,7 @@ void btrfs_split_delalloc_extent(struct inode *inode,
 		return;
 
 	size = orig->end - orig->start + 1;
-	if (size > BTRFS_MAX_EXTENT_SIZE) {
+	if (size > fs_info->max_extent_size) {
 		u32 num_extents;
 		u64 new_size;
 
@@ -1777,10 +2260,10 @@ void btrfs_split_delalloc_extent(struct inode *inode,
 		 * applies here, just in reverse.
 		 */
 		new_size = orig->end - split + 1;
-		num_extents = count_max_extents(new_size);
+		num_extents = count_max_extents(fs_info, new_size);
 		new_size = split - orig->start;
-		num_extents += count_max_extents(new_size);
-		if (count_max_extents(size) >= num_extents)
+		num_extents += count_max_extents(fs_info, new_size);
+		if (count_max_extents(fs_info, size) >= num_extents)
 			return;
 	}
 
@@ -1797,6 +2280,7 @@ void btrfs_split_delalloc_extent(struct inode *inode,
 void btrfs_merge_delalloc_extent(struct inode *inode, struct extent_state *new,
 				 struct extent_state *other)
 {
+	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	u64 new_size, old_size;
 	u32 num_extents;
 
@@ -1810,7 +2294,7 @@ void btrfs_merge_delalloc_extent(struct inode *inode, struct extent_state *new,
 		new_size = other->end - new->start + 1;
 
 	/* we're not bigger than the max, unreserve the space and go */
-	if (new_size <= BTRFS_MAX_EXTENT_SIZE) {
+	if (new_size <= fs_info->max_extent_size) {
 		spin_lock(&BTRFS_I(inode)->lock);
 		btrfs_mod_outstanding_extents(BTRFS_I(inode), -1);
 		spin_unlock(&BTRFS_I(inode)->lock);
@@ -1836,10 +2320,10 @@ void btrfs_merge_delalloc_extent(struct inode *inode, struct extent_state *new,
 	 * this case.
 	 */
 	old_size = other->end - other->start + 1;
-	num_extents = count_max_extents(old_size);
+	num_extents = count_max_extents(fs_info, old_size);
 	old_size = new->end - new->start + 1;
-	num_extents += count_max_extents(old_size);
-	if (count_max_extents(new_size) >= num_extents)
+	num_extents += count_max_extents(fs_info, old_size);
+	if (count_max_extents(fs_info, new_size) >= num_extents)
 		return;
 
 	spin_lock(&BTRFS_I(inode)->lock);
@@ -1904,21 +2388,21 @@ static void btrfs_del_delalloc_inode(struct btrfs_root *root,
  * list of inodes that have pending delalloc work to be done.
  */
 void btrfs_set_delalloc_extent(struct inode *inode, struct extent_state *state,
-			       unsigned *bits)
+			       u32 bits)
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 
-	if ((*bits & EXTENT_DEFRAG) && !(*bits & EXTENT_DELALLOC))
+	if ((bits & EXTENT_DEFRAG) && !(bits & EXTENT_DELALLOC))
 		WARN_ON(1);
 	/*
 	 * set_bit and clear bit hooks normally require _irqsave/restore
 	 * but in this case, we are only testing for the DELALLOC
 	 * bit, which is only set or cleared with irqs on
 	 */
-	if (!(state->state & EXTENT_DELALLOC) && (*bits & EXTENT_DELALLOC)) {
+	if (!(state->state & EXTENT_DELALLOC) && (bits & EXTENT_DELALLOC)) {
 		struct btrfs_root *root = BTRFS_I(inode)->root;
 		u64 len = state->end + 1 - state->start;
-		u32 num_extents = count_max_extents(len);
+		u32 num_extents = count_max_extents(fs_info, len);
 		bool do_list = !btrfs_is_free_space_inode(BTRFS_I(inode));
 
 		spin_lock(&BTRFS_I(inode)->lock);
@@ -1933,7 +2417,7 @@ void btrfs_set_delalloc_extent(struct inode *inode, struct extent_state *state,
 					 fs_info->delalloc_batch);
 		spin_lock(&BTRFS_I(inode)->lock);
 		BTRFS_I(inode)->delalloc_bytes += len;
-		if (*bits & EXTENT_DEFRAG)
+		if (bits & EXTENT_DEFRAG)
 			BTRFS_I(inode)->defrag_bytes += len;
 		if (do_list && !test_bit(BTRFS_INODE_IN_DELALLOC_LIST,
 					 &BTRFS_I(inode)->runtime_flags))
@@ -1942,7 +2426,7 @@ void btrfs_set_delalloc_extent(struct inode *inode, struct extent_state *state,
 	}
 
 	if (!(state->state & EXTENT_DELALLOC_NEW) &&
-	    (*bits & EXTENT_DELALLOC_NEW)) {
+	    (bits & EXTENT_DELALLOC_NEW)) {
 		spin_lock(&BTRFS_I(inode)->lock);
 		BTRFS_I(inode)->new_delalloc_bytes += state->end + 1 -
 			state->start;
@@ -1955,14 +2439,14 @@ void btrfs_set_delalloc_extent(struct inode *inode, struct extent_state *state,
  * accounting happens.
  */
 void btrfs_clear_delalloc_extent(struct inode *vfs_inode,
-				 struct extent_state *state, unsigned *bits)
+				 struct extent_state *state, u32 bits)
 {
 	struct btrfs_inode *inode = BTRFS_I(vfs_inode);
 	struct btrfs_fs_info *fs_info = btrfs_sb(vfs_inode->i_sb);
 	u64 len = state->end + 1 - state->start;
-	u32 num_extents = count_max_extents(len);
+	u32 num_extents = count_max_extents(fs_info, len);
 
-	if ((state->state & EXTENT_DEFRAG) && (*bits & EXTENT_DEFRAG)) {
+	if ((state->state & EXTENT_DEFRAG) && (bits & EXTENT_DEFRAG)) {
 		spin_lock(&inode->lock);
 		inode->defrag_bytes -= len;
 		spin_unlock(&inode->lock);
@@ -1973,7 +2457,7 @@ void btrfs_clear_delalloc_extent(struct inode *vfs_inode,
 	 * but in this case, we are only testing for the DELALLOC
 	 * bit, which is only set or cleared with irqs on
 	 */
-	if ((state->state & EXTENT_DELALLOC) && (*bits & EXTENT_DELALLOC)) {
+	if ((state->state & EXTENT_DELALLOC) && (bits & EXTENT_DELALLOC)) {
 		struct btrfs_root *root = inode->root;
 		bool do_list = !btrfs_is_free_space_inode(inode);
 
@@ -1986,7 +2470,7 @@ void btrfs_clear_delalloc_extent(struct inode *vfs_inode,
 		 * don't need to call delalloc_release_metadata if there is an
 		 * error.
 		 */
-		if (*bits & EXTENT_CLEAR_META_RESV &&
+		if (bits & EXTENT_CLEAR_META_RESV &&
 		    root != fs_info->tree_root)
 			btrfs_delalloc_release_metadata(inode, len, false);
 
@@ -1994,12 +2478,10 @@ void btrfs_clear_delalloc_extent(struct inode *vfs_inode,
 		if (btrfs_is_testing(fs_info))
 			return;
 
-		if (root->root_key.objectid != BTRFS_DATA_RELOC_TREE_OBJECTID &&
+		if (!btrfs_is_data_reloc_root(root) &&
 		    do_list && !(state->state & EXTENT_NORESERVE) &&
-		    (*bits & EXTENT_CLEAR_DATA_RESV))
-			btrfs_free_reserved_data_space_noquota(
-					&inode->vfs_inode,
-					state->start, len);
+		    (bits & EXTENT_CLEAR_DATA_RESV))
+			btrfs_free_reserved_data_space_noquota(fs_info, len);
 
 		percpu_counter_add_batch(&fs_info->delalloc_bytes, -len,
 					 fs_info->delalloc_batch);
@@ -2013,55 +2495,17 @@ void btrfs_clear_delalloc_extent(struct inode *vfs_inode,
 	}
 
 	if ((state->state & EXTENT_DELALLOC_NEW) &&
-	    (*bits & EXTENT_DELALLOC_NEW)) {
+	    (bits & EXTENT_DELALLOC_NEW)) {
 		spin_lock(&inode->lock);
 		ASSERT(inode->new_delalloc_bytes >= len);
 		inode->new_delalloc_bytes -= len;
+		if (bits & EXTENT_ADD_INODE_BYTES)
+			inode_add_bytes(&inode->vfs_inode, len);
 		spin_unlock(&inode->lock);
 	}
 }
 
 /*
- * btrfs_bio_fits_in_stripe - Checks whether the size of the given bio will fit
- * in a chunk's stripe. This function ensures that bios do not span a
- * stripe/chunk
- *
- * @page - The page we are about to add to the bio
- * @size - size we want to add to the bio
- * @bio - bio we want to ensure is smaller than a stripe
- * @bio_flags - flags of the bio
- *
- * return 1 if page cannot be added to the bio
- * return 0 if page can be added to the bio
- * return error otherwise
- */
-int btrfs_bio_fits_in_stripe(struct page *page, size_t size, struct bio *bio,
-			     unsigned long bio_flags)
-{
-	struct inode *inode = page->mapping->host;
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	u64 logical = (u64)bio->bi_iter.bi_sector << 9;
-	u64 length = 0;
-	u64 map_length;
-	int ret;
-	struct btrfs_io_geometry geom;
-
-	if (bio_flags & EXTENT_BIO_COMPRESSED)
-		return 0;
-
-	length = bio->bi_iter.bi_size;
-	map_length = length;
-	ret = btrfs_get_io_geometry(fs_info, btrfs_op(bio), logical, map_length,
-				    &geom);
-	if (ret < 0)
-		return ret;
-
-	if (geom.len < length + size)
-		return 1;
-	return 0;
-}
-
-/*
  * in order to insert checksums into the metadata in large chunks,
  * we wait until bio submission time.   All the pages in the bio are
  * checksummed and sums are attached onto the ordered extent record.
@@ -2069,107 +2513,275 @@ int btrfs_bio_fits_in_stripe(struct page *page, size_t size, struct bio *bio,
  * At IO completion time the cums attached on the ordered extent record
  * are inserted into the btree
  */
-static blk_status_t btrfs_submit_bio_start(void *private_data, struct bio *bio,
-				    u64 bio_offset)
+static blk_status_t btrfs_submit_bio_start(struct inode *inode, struct bio *bio,
+					   u64 dio_file_offset)
 {
-	struct inode *inode = private_data;
-	blk_status_t ret = 0;
-
-	ret = btrfs_csum_one_bio(inode, bio, 0, 0);
-	BUG_ON(ret); /* -ENOMEM */
-	return 0;
+	return btrfs_csum_one_bio(BTRFS_I(inode), bio, (u64)-1, false);
 }
 
 /*
- * extent_io.c submission hook. This does the right thing for csum calculation
- * on write, or reading the csums from the tree before a read.
- *
- * Rules about async/sync submit,
- * a) read:				sync submit
- *
- * b) write without checksum:		sync submit
+ * Split an extent_map at [start, start + len]
  *
- * c) write with checksum:
- *    c-1) if bio is issued by fsync:	sync submit
- *         (sync_writers != 0)
- *
- *    c-2) if root is reloc root:	sync submit
- *         (only in case of buffered IO)
- *
- *    c-3) otherwise:			async submit
+ * This function is intended to be used only for extract_ordered_extent().
  */
-static blk_status_t btrfs_submit_bio_hook(struct inode *inode, struct bio *bio,
-					  int mirror_num,
-					  unsigned long bio_flags)
+static int split_zoned_em(struct btrfs_inode *inode, u64 start, u64 len,
+			  u64 pre, u64 post)
+{
+	struct extent_map_tree *em_tree = &inode->extent_tree;
+	struct extent_map *em;
+	struct extent_map *split_pre = NULL;
+	struct extent_map *split_mid = NULL;
+	struct extent_map *split_post = NULL;
+	int ret = 0;
+	unsigned long flags;
+
+	/* Sanity check */
+	if (pre == 0 && post == 0)
+		return 0;
+
+	split_pre = alloc_extent_map();
+	if (pre)
+		split_mid = alloc_extent_map();
+	if (post)
+		split_post = alloc_extent_map();
+	if (!split_pre || (pre && !split_mid) || (post && !split_post)) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	ASSERT(pre + post < len);
+
+	lock_extent(&inode->io_tree, start, start + len - 1, NULL);
+	write_lock(&em_tree->lock);
+	em = lookup_extent_mapping(em_tree, start, len);
+	if (!em) {
+		ret = -EIO;
+		goto out_unlock;
+	}
+
+	ASSERT(em->len == len);
+	ASSERT(!test_bit(EXTENT_FLAG_COMPRESSED, &em->flags));
+	ASSERT(em->block_start < EXTENT_MAP_LAST_BYTE);
+	ASSERT(test_bit(EXTENT_FLAG_PINNED, &em->flags));
+	ASSERT(!test_bit(EXTENT_FLAG_LOGGING, &em->flags));
+	ASSERT(!list_empty(&em->list));
+
+	flags = em->flags;
+	clear_bit(EXTENT_FLAG_PINNED, &em->flags);
+
+	/* First, replace the em with a new extent_map starting from * em->start */
+	split_pre->start = em->start;
+	split_pre->len = (pre ? pre : em->len - post);
+	split_pre->orig_start = split_pre->start;
+	split_pre->block_start = em->block_start;
+	split_pre->block_len = split_pre->len;
+	split_pre->orig_block_len = split_pre->block_len;
+	split_pre->ram_bytes = split_pre->len;
+	split_pre->flags = flags;
+	split_pre->compress_type = em->compress_type;
+	split_pre->generation = em->generation;
+
+	replace_extent_mapping(em_tree, em, split_pre, 1);
+
+	/*
+	 * Now we only have an extent_map at:
+	 *     [em->start, em->start + pre] if pre != 0
+	 *     [em->start, em->start + em->len - post] if pre == 0
+	 */
+
+	if (pre) {
+		/* Insert the middle extent_map */
+		split_mid->start = em->start + pre;
+		split_mid->len = em->len - pre - post;
+		split_mid->orig_start = split_mid->start;
+		split_mid->block_start = em->block_start + pre;
+		split_mid->block_len = split_mid->len;
+		split_mid->orig_block_len = split_mid->block_len;
+		split_mid->ram_bytes = split_mid->len;
+		split_mid->flags = flags;
+		split_mid->compress_type = em->compress_type;
+		split_mid->generation = em->generation;
+		add_extent_mapping(em_tree, split_mid, 1);
+	}
+
+	if (post) {
+		split_post->start = em->start + em->len - post;
+		split_post->len = post;
+		split_post->orig_start = split_post->start;
+		split_post->block_start = em->block_start + em->len - post;
+		split_post->block_len = split_post->len;
+		split_post->orig_block_len = split_post->block_len;
+		split_post->ram_bytes = split_post->len;
+		split_post->flags = flags;
+		split_post->compress_type = em->compress_type;
+		split_post->generation = em->generation;
+		add_extent_mapping(em_tree, split_post, 1);
+	}
+
+	/* Once for us */
+	free_extent_map(em);
+	/* Once for the tree */
+	free_extent_map(em);
+
+out_unlock:
+	write_unlock(&em_tree->lock);
+	unlock_extent(&inode->io_tree, start, start + len - 1, NULL);
+out:
+	free_extent_map(split_pre);
+	free_extent_map(split_mid);
+	free_extent_map(split_post);
+
+	return ret;
+}
 
+static blk_status_t extract_ordered_extent(struct btrfs_inode *inode,
+					   struct bio *bio, loff_t file_offset)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	struct btrfs_root *root = BTRFS_I(inode)->root;
-	enum btrfs_wq_endio_type metadata = BTRFS_WQ_ENDIO_DATA;
-	blk_status_t ret = 0;
-	int skip_sum;
-	int async = !atomic_read(&BTRFS_I(inode)->sync_writers);
+	struct btrfs_ordered_extent *ordered;
+	u64 start = (u64)bio->bi_iter.bi_sector << SECTOR_SHIFT;
+	u64 file_len;
+	u64 len = bio->bi_iter.bi_size;
+	u64 end = start + len;
+	u64 ordered_end;
+	u64 pre, post;
+	int ret = 0;
 
-	skip_sum = BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM;
+	ordered = btrfs_lookup_ordered_extent(inode, file_offset);
+	if (WARN_ON_ONCE(!ordered))
+		return BLK_STS_IOERR;
 
-	if (btrfs_is_free_space_inode(BTRFS_I(inode)))
-		metadata = BTRFS_WQ_ENDIO_FREE_SPACE;
+	/* No need to split */
+	if (ordered->disk_num_bytes == len)
+		goto out;
 
-	if (bio_op(bio) != REQ_OP_WRITE) {
-		ret = btrfs_bio_wq_end_io(fs_info, bio, metadata);
-		if (ret)
-			goto out;
+	/* We cannot split once end_bio'd ordered extent */
+	if (WARN_ON_ONCE(ordered->bytes_left != ordered->disk_num_bytes)) {
+		ret = -EINVAL;
+		goto out;
+	}
 
-		if (bio_flags & EXTENT_BIO_COMPRESSED) {
-			ret = btrfs_submit_compressed_read(inode, bio,
-							   mirror_num,
-							   bio_flags);
-			goto out;
-		} else if (!skip_sum) {
-			ret = btrfs_lookup_bio_sums(inode, bio, (u64)-1, NULL);
-			if (ret)
-				goto out;
-		}
-		goto mapit;
-	} else if (async && !skip_sum) {
-		/* csum items have already been cloned */
-		if (root->root_key.objectid == BTRFS_DATA_RELOC_TREE_OBJECTID)
-			goto mapit;
-		/* we're doing a write, do the async checksumming */
-		ret = btrfs_wq_submit_bio(fs_info, bio, mirror_num, bio_flags,
-					  0, inode, btrfs_submit_bio_start);
+	/* We cannot split a compressed ordered extent */
+	if (WARN_ON_ONCE(ordered->disk_num_bytes != ordered->num_bytes)) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	ordered_end = ordered->disk_bytenr + ordered->disk_num_bytes;
+	/* bio must be in one ordered extent */
+	if (WARN_ON_ONCE(start < ordered->disk_bytenr || end > ordered_end)) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	/* Checksum list should be empty */
+	if (WARN_ON_ONCE(!list_empty(&ordered->list))) {
+		ret = -EINVAL;
 		goto out;
-	} else if (!skip_sum) {
-		ret = btrfs_csum_one_bio(inode, bio, 0, 0);
-		if (ret)
-			goto out;
 	}
 
-mapit:
-	ret = btrfs_map_bio(fs_info, bio, mirror_num);
+	file_len = ordered->num_bytes;
+	pre = start - ordered->disk_bytenr;
+	post = ordered_end - end;
+
+	ret = btrfs_split_ordered_extent(ordered, pre, post);
+	if (ret)
+		goto out;
+	ret = split_zoned_em(inode, file_offset, file_len, pre, post);
 
 out:
+	btrfs_put_ordered_extent(ordered);
+
+	return errno_to_blk_status(ret);
+}
+
+void btrfs_submit_data_write_bio(struct inode *inode, struct bio *bio, int mirror_num)
+{
+	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+	struct btrfs_inode *bi = BTRFS_I(inode);
+	blk_status_t ret;
+
+	if (bio_op(bio) == REQ_OP_ZONE_APPEND) {
+		ret = extract_ordered_extent(bi, bio,
+				page_offset(bio_first_bvec_all(bio)->bv_page));
+		if (ret) {
+			btrfs_bio_end_io(btrfs_bio(bio), ret);
+			return;
+		}
+	}
+
+	/*
+	 * If we need to checksum, and the I/O is not issued by fsync and
+	 * friends, that is ->sync_writers != 0, defer the submission to a
+	 * workqueue to parallelize it.
+	 *
+	 * Csum items for reloc roots have already been cloned at this point,
+	 * so they are handled as part of the no-checksum case.
+	 */
+	if (!(bi->flags & BTRFS_INODE_NODATASUM) &&
+	    !test_bit(BTRFS_FS_STATE_NO_CSUMS, &fs_info->fs_state) &&
+	    !btrfs_is_data_reloc_root(bi->root)) {
+		if (!atomic_read(&bi->sync_writers) &&
+		    btrfs_wq_submit_bio(inode, bio, mirror_num, 0,
+					btrfs_submit_bio_start))
+			return;
+
+		ret = btrfs_csum_one_bio(bi, bio, (u64)-1, false);
+		if (ret) {
+			btrfs_bio_end_io(btrfs_bio(bio), ret);
+			return;
+		}
+	}
+	btrfs_submit_bio(fs_info, bio, mirror_num);
+}
+
+void btrfs_submit_data_read_bio(struct inode *inode, struct bio *bio,
+			int mirror_num, enum btrfs_compression_type compress_type)
+{
+	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+	blk_status_t ret;
+
+	if (compress_type != BTRFS_COMPRESS_NONE) {
+		/*
+		 * btrfs_submit_compressed_read will handle completing the bio
+		 * if there were any errors, so just return here.
+		 */
+		btrfs_submit_compressed_read(inode, bio, mirror_num);
+		return;
+	}
+
+	/* Save the original iter for read repair */
+	btrfs_bio(bio)->iter = bio->bi_iter;
+
+	/*
+	 * Lookup bio sums does extra checks around whether we need to csum or
+	 * not, which is why we ignore skip_sum here.
+	 */
+	ret = btrfs_lookup_bio_sums(inode, bio, NULL);
 	if (ret) {
-		bio->bi_status = ret;
-		bio_endio(bio);
+		btrfs_bio_end_io(btrfs_bio(bio), ret);
+		return;
 	}
-	return ret;
+
+	btrfs_submit_bio(fs_info, bio, mirror_num);
 }
 
 /*
  * given a list of ordered sums record them in the inode.  This happens
  * at IO completion time based on sums calculated at bio submission time.
  */
-static noinline int add_pending_csums(struct btrfs_trans_handle *trans,
-			     struct inode *inode, struct list_head *list)
+static int add_pending_csums(struct btrfs_trans_handle *trans,
+			     struct list_head *list)
 {
 	struct btrfs_ordered_sum *sum;
+	struct btrfs_root *csum_root = NULL;
 	int ret;
 
 	list_for_each_entry(sum, list, list) {
 		trans->adding_csums = true;
-		ret = btrfs_csum_file_blocks(trans,
-		       BTRFS_I(inode)->root->fs_info->csum_root, sum);
+		if (!csum_root)
+			csum_root = btrfs_csum_root(trans->fs_info,
+						    sum->bytenr);
+		ret = btrfs_csum_file_blocks(trans, csum_root, sum);
 		trans->adding_csums = false;
 		if (ret)
 			return ret;
@@ -2177,13 +2789,71 @@ static noinline int add_pending_csums(struct btrfs_trans_handle *trans,
 	return 0;
 }
 
-int btrfs_set_extent_delalloc(struct inode *inode, u64 start, u64 end,
+static int btrfs_find_new_delalloc_bytes(struct btrfs_inode *inode,
+					 const u64 start,
+					 const u64 len,
+					 struct extent_state **cached_state)
+{
+	u64 search_start = start;
+	const u64 end = start + len - 1;
+
+	while (search_start < end) {
+		const u64 search_len = end - search_start + 1;
+		struct extent_map *em;
+		u64 em_len;
+		int ret = 0;
+
+		em = btrfs_get_extent(inode, NULL, 0, search_start, search_len);
+		if (IS_ERR(em))
+			return PTR_ERR(em);
+
+		if (em->block_start != EXTENT_MAP_HOLE)
+			goto next;
+
+		em_len = em->len;
+		if (em->start < search_start)
+			em_len -= search_start - em->start;
+		if (em_len > search_len)
+			em_len = search_len;
+
+		ret = set_extent_bit(&inode->io_tree, search_start,
+				     search_start + em_len - 1,
+				     EXTENT_DELALLOC_NEW, cached_state,
+				     GFP_NOFS);
+next:
+		search_start = extent_map_end(em);
+		free_extent_map(em);
+		if (ret)
+			return ret;
+	}
+	return 0;
+}
+
+int btrfs_set_extent_delalloc(struct btrfs_inode *inode, u64 start, u64 end,
 			      unsigned int extra_bits,
 			      struct extent_state **cached_state)
 {
 	WARN_ON(PAGE_ALIGNED(end));
-	return set_extent_delalloc(&BTRFS_I(inode)->io_tree, start, end,
-				   extra_bits, cached_state);
+
+	if (start >= i_size_read(&inode->vfs_inode) &&
+	    !(inode->flags & BTRFS_INODE_PREALLOC)) {
+		/*
+		 * There can't be any extents following eof in this case so just
+		 * set the delalloc new bit for the range directly.
+		 */
+		extra_bits |= EXTENT_DELALLOC_NEW;
+	} else {
+		int ret;
+
+		ret = btrfs_find_new_delalloc_bytes(inode, start,
+						    end + 1 - start,
+						    cached_state);
+		if (ret)
+			return ret;
+	}
+
+	return set_extent_delalloc(&inode->io_tree, start, end, extra_bits,
+				   cached_state);
 }
 
 /* see btrfs_writepage_start_hook for details on why this is required */
@@ -2200,7 +2870,7 @@ static void btrfs_writepage_fixup_worker(struct btrfs_work *work)
 	struct extent_state *cached_state = NULL;
 	struct extent_changeset *data_reserved = NULL;
 	struct page *page;
-	struct inode *inode;
+	struct btrfs_inode *inode;
 	u64 page_start;
 	u64 page_end;
 	int ret = 0;
@@ -2208,7 +2878,7 @@ static void btrfs_writepage_fixup_worker(struct btrfs_work *work)
 
 	fixup = container_of(work, struct btrfs_writepage_fixup, work);
 	page = fixup->page;
-	inode = fixup->inode;
+	inode = BTRFS_I(fixup->inode);
 	page_start = page_offset(page);
 	page_end = page_offset(page) + PAGE_SIZE - 1;
 
@@ -2245,8 +2915,7 @@ again:
 		 *    when the page was already properly dealt with.
 		 */
 		if (!ret) {
-			btrfs_delalloc_release_extents(BTRFS_I(inode),
-						       PAGE_SIZE);
+			btrfs_delalloc_release_extents(inode, PAGE_SIZE);
 			btrfs_delalloc_release_space(inode, data_reserved,
 						     page_start, PAGE_SIZE,
 						     true);
@@ -2262,20 +2931,18 @@ again:
 	if (ret)
 		goto out_page;
 
-	lock_extent_bits(&BTRFS_I(inode)->io_tree, page_start, page_end,
-			 &cached_state);
+	lock_extent(&inode->io_tree, page_start, page_end, &cached_state);
 
 	/* already ordered? We're done */
-	if (PagePrivate2(page))
+	if (PageOrdered(page))
 		goto out_reserved;
 
-	ordered = btrfs_lookup_ordered_range(BTRFS_I(inode), page_start,
-					PAGE_SIZE);
+	ordered = btrfs_lookup_ordered_range(inode, page_start, PAGE_SIZE);
 	if (ordered) {
-		unlock_extent_cached(&BTRFS_I(inode)->io_tree, page_start,
-				     page_end, &cached_state);
+		unlock_extent(&inode->io_tree, page_start, page_end,
+			      &cached_state);
 		unlock_page(page);
-		btrfs_start_ordered_extent(inode, ordered, 1);
+		btrfs_start_ordered_extent(ordered, 1);
 		btrfs_put_ordered_extent(ordered);
 		goto again;
 	}
@@ -2295,12 +2962,11 @@ again:
 	BUG_ON(!PageDirty(page));
 	free_delalloc_space = false;
 out_reserved:
-	btrfs_delalloc_release_extents(BTRFS_I(inode), PAGE_SIZE);
+	btrfs_delalloc_release_extents(inode, PAGE_SIZE);
 	if (free_delalloc_space)
 		btrfs_delalloc_release_space(inode, data_reserved, page_start,
 					     PAGE_SIZE, true);
-	unlock_extent_cached(&BTRFS_I(inode)->io_tree, page_start, page_end,
-			     &cached_state);
+	unlock_extent(&inode->io_tree, page_start, page_end, &cached_state);
 out_page:
 	if (ret) {
 		/*
@@ -2312,7 +2978,7 @@ out_page:
 		clear_page_dirty_for_io(page);
 		SetPageError(page);
 	}
-	ClearPageChecked(page);
+	btrfs_page_clear_checked(inode->root->fs_info, page, page_start, PAGE_SIZE);
 	unlock_page(page);
 	put_page(page);
 	kfree(fixup);
@@ -2322,7 +2988,7 @@ out_page:
 	 * that could need flushing space. Recursing back to fixup worker would
 	 * deadlock.
 	 */
-	btrfs_add_delayed_iput(inode);
+	btrfs_add_delayed_iput(&inode->vfs_inode);
 }
 
 /*
@@ -2336,14 +3002,14 @@ out_page:
  * to fix it up.  The async helper will wait for ordered extents, set
  * the delalloc bit and make it safe to write the page.
  */
-int btrfs_writepage_cow_fixup(struct page *page, u64 start, u64 end)
+int btrfs_writepage_cow_fixup(struct page *page)
 {
 	struct inode *inode = page->mapping->host;
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	struct btrfs_writepage_fixup *fixup;
 
-	/* this page is properly in the ordered list */
-	if (TestClearPagePrivate2(page))
+	/* This page has ordered extent covering it already */
+	if (PageOrdered(page))
 		return 0;
 
 	/*
@@ -2367,7 +3033,7 @@ int btrfs_writepage_cow_fixup(struct page *page, u64 start, u64 end)
 	 * page->mapping outside of the page lock.
 	 */
 	ihold(inode);
-	SetPageChecked(page);
+	btrfs_page_set_checked(fs_info, page, page_offset(page), PAGE_SIZE);
 	get_page(page);
 	btrfs_init_work(&fixup->work, btrfs_writepage_fixup_worker, NULL, NULL);
 	fixup->page = page;
@@ -2378,19 +3044,22 @@ int btrfs_writepage_cow_fixup(struct page *page, u64 start, u64 end)
 }
 
 static int insert_reserved_file_extent(struct btrfs_trans_handle *trans,
-				       struct inode *inode, u64 file_pos,
-				       u64 disk_bytenr, u64 disk_num_bytes,
-				       u64 num_bytes, u64 ram_bytes,
-				       u8 compression, u8 encryption,
-				       u16 other_encoding, int extent_type)
+				       struct btrfs_inode *inode, u64 file_pos,
+				       struct btrfs_file_extent_item *stack_fi,
+				       const bool update_inode_bytes,
+				       u64 qgroup_reserved)
 {
-	struct btrfs_root *root = BTRFS_I(inode)->root;
-	struct btrfs_file_extent_item *fi;
+	struct btrfs_root *root = inode->root;
+	const u64 sectorsize = root->fs_info->sectorsize;
 	struct btrfs_path *path;
 	struct extent_buffer *leaf;
 	struct btrfs_key ins;
-	u64 qg_released;
-	int extent_inserted = 0;
+	u64 disk_num_bytes = btrfs_stack_file_extent_disk_num_bytes(stack_fi);
+	u64 disk_bytenr = btrfs_stack_file_extent_disk_bytenr(stack_fi);
+	u64 offset = btrfs_stack_file_extent_offset(stack_fi);
+	u64 num_bytes = btrfs_stack_file_extent_num_bytes(stack_fi);
+	u64 ram_bytes = btrfs_stack_file_extent_ram_bytes(stack_fi);
+	struct btrfs_drop_extents_args drop_args = { 0 };
 	int ret;
 
 	path = btrfs_alloc_path();
@@ -2406,57 +3075,64 @@ static int insert_reserved_file_extent(struct btrfs_trans_handle *trans,
 	 * the caller is expected to unpin it and allow it to be merged
 	 * with the others.
 	 */
-	ret = __btrfs_drop_extents(trans, root, inode, path, file_pos,
-				   file_pos + num_bytes, NULL, 0,
-				   1, sizeof(*fi), &extent_inserted);
+	drop_args.path = path;
+	drop_args.start = file_pos;
+	drop_args.end = file_pos + num_bytes;
+	drop_args.replace_extent = true;
+	drop_args.extent_item_size = sizeof(*stack_fi);
+	ret = btrfs_drop_extents(trans, root, inode, &drop_args);
 	if (ret)
 		goto out;
 
-	if (!extent_inserted) {
-		ins.objectid = btrfs_ino(BTRFS_I(inode));
+	if (!drop_args.extent_inserted) {
+		ins.objectid = btrfs_ino(inode);
 		ins.offset = file_pos;
 		ins.type = BTRFS_EXTENT_DATA_KEY;
 
-		path->leave_spinning = 1;
 		ret = btrfs_insert_empty_item(trans, root, path, &ins,
-					      sizeof(*fi));
+					      sizeof(*stack_fi));
 		if (ret)
 			goto out;
 	}
 	leaf = path->nodes[0];
-	fi = btrfs_item_ptr(leaf, path->slots[0],
-			    struct btrfs_file_extent_item);
-	btrfs_set_file_extent_generation(leaf, fi, trans->transid);
-	btrfs_set_file_extent_type(leaf, fi, extent_type);
-	btrfs_set_file_extent_disk_bytenr(leaf, fi, disk_bytenr);
-	btrfs_set_file_extent_disk_num_bytes(leaf, fi, disk_num_bytes);
-	btrfs_set_file_extent_offset(leaf, fi, 0);
-	btrfs_set_file_extent_num_bytes(leaf, fi, num_bytes);
-	btrfs_set_file_extent_ram_bytes(leaf, fi, ram_bytes);
-	btrfs_set_file_extent_compression(leaf, fi, compression);
-	btrfs_set_file_extent_encryption(leaf, fi, encryption);
-	btrfs_set_file_extent_other_encoding(leaf, fi, other_encoding);
+	btrfs_set_stack_file_extent_generation(stack_fi, trans->transid);
+	write_extent_buffer(leaf, stack_fi,
+			btrfs_item_ptr_offset(leaf, path->slots[0]),
+			sizeof(struct btrfs_file_extent_item));
 
 	btrfs_mark_buffer_dirty(leaf);
 	btrfs_release_path(path);
 
-	inode_add_bytes(inode, num_bytes);
+	/*
+	 * If we dropped an inline extent here, we know the range where it is
+	 * was not marked with the EXTENT_DELALLOC_NEW bit, so we update the
+	 * number of bytes only for that range containing the inline extent.
+	 * The remaining of the range will be processed when clearning the
+	 * EXTENT_DELALLOC_BIT bit through the ordered extent completion.
+	 */
+	if (file_pos == 0 && !IS_ALIGNED(drop_args.bytes_found, sectorsize)) {
+		u64 inline_size = round_down(drop_args.bytes_found, sectorsize);
+
+		inline_size = drop_args.bytes_found - inline_size;
+		btrfs_update_inode_bytes(inode, sectorsize, inline_size);
+		drop_args.bytes_found -= inline_size;
+		num_bytes -= sectorsize;
+	}
+
+	if (update_inode_bytes)
+		btrfs_update_inode_bytes(inode, num_bytes, drop_args.bytes_found);
 
 	ins.objectid = disk_bytenr;
 	ins.offset = disk_num_bytes;
 	ins.type = BTRFS_EXTENT_ITEM_KEY;
 
-	/*
-	 * Release the reserved range from inode dirty range map, as it is
-	 * already moved into delayed_ref_head
-	 */
-	ret = btrfs_qgroup_release_data(inode, file_pos, ram_bytes);
-	if (ret < 0)
+	ret = btrfs_inode_set_file_extent_range(inode, file_pos, ram_bytes);
+	if (ret)
 		goto out;
-	qg_released = ret;
-	ret = btrfs_alloc_reserved_file_extent(trans, root,
-					       btrfs_ino(BTRFS_I(inode)),
-					       file_pos, qg_released, &ins);
+
+	ret = btrfs_alloc_reserved_file_extent(trans, root, btrfs_ino(inode),
+					       file_pos - offset,
+					       qgroup_reserved, &ins);
 out:
 	btrfs_free_path(path);
 
@@ -2478,17 +3154,56 @@ static void btrfs_release_delalloc_bytes(struct btrfs_fs_info *fs_info,
 	btrfs_put_block_group(cache);
 }
 
-/* as ordered data IO finishes, this gets called so we can finish
+static int insert_ordered_extent_file_extent(struct btrfs_trans_handle *trans,
+					     struct btrfs_ordered_extent *oe)
+{
+	struct btrfs_file_extent_item stack_fi;
+	bool update_inode_bytes;
+	u64 num_bytes = oe->num_bytes;
+	u64 ram_bytes = oe->ram_bytes;
+
+	memset(&stack_fi, 0, sizeof(stack_fi));
+	btrfs_set_stack_file_extent_type(&stack_fi, BTRFS_FILE_EXTENT_REG);
+	btrfs_set_stack_file_extent_disk_bytenr(&stack_fi, oe->disk_bytenr);
+	btrfs_set_stack_file_extent_disk_num_bytes(&stack_fi,
+						   oe->disk_num_bytes);
+	btrfs_set_stack_file_extent_offset(&stack_fi, oe->offset);
+	if (test_bit(BTRFS_ORDERED_TRUNCATED, &oe->flags)) {
+		num_bytes = oe->truncated_len;
+		ram_bytes = num_bytes;
+	}
+	btrfs_set_stack_file_extent_num_bytes(&stack_fi, num_bytes);
+	btrfs_set_stack_file_extent_ram_bytes(&stack_fi, ram_bytes);
+	btrfs_set_stack_file_extent_compression(&stack_fi, oe->compress_type);
+	/* Encryption and other encoding is reserved and all 0 */
+
+	/*
+	 * For delalloc, when completing an ordered extent we update the inode's
+	 * bytes when clearing the range in the inode's io tree, so pass false
+	 * as the argument 'update_inode_bytes' to insert_reserved_file_extent(),
+	 * except if the ordered extent was truncated.
+	 */
+	update_inode_bytes = test_bit(BTRFS_ORDERED_DIRECT, &oe->flags) ||
+			     test_bit(BTRFS_ORDERED_ENCODED, &oe->flags) ||
+			     test_bit(BTRFS_ORDERED_TRUNCATED, &oe->flags);
+
+	return insert_reserved_file_extent(trans, BTRFS_I(oe->inode),
+					   oe->file_offset, &stack_fi,
+					   update_inode_bytes, oe->qgroup_rsv);
+}
+
+/*
+ * As ordered data IO finishes, this gets called so we can finish
  * an ordered extent if the range of bytes in the file it covers are
  * fully written.
  */
-static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent)
+int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent)
 {
-	struct inode *inode = ordered_extent->inode;
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	struct btrfs_root *root = BTRFS_I(inode)->root;
+	struct btrfs_inode *inode = BTRFS_I(ordered_extent->inode);
+	struct btrfs_root *root = inode->root;
+	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct btrfs_trans_handle *trans = NULL;
-	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
+	struct extent_io_tree *io_tree = &inode->io_tree;
 	struct extent_state *cached_state = NULL;
 	u64 start, end;
 	int compress_type = 0;
@@ -2496,27 +3211,35 @@ static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent)
 	u64 logical_len = ordered_extent->num_bytes;
 	bool freespace_inode;
 	bool truncated = false;
-	bool range_locked = false;
-	bool clear_new_delalloc_bytes = false;
 	bool clear_reserved_extent = true;
-	unsigned int clear_bits;
+	unsigned int clear_bits = EXTENT_DEFRAG;
 
 	start = ordered_extent->file_offset;
 	end = start + ordered_extent->num_bytes - 1;
 
 	if (!test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags) &&
 	    !test_bit(BTRFS_ORDERED_PREALLOC, &ordered_extent->flags) &&
-	    !test_bit(BTRFS_ORDERED_DIRECT, &ordered_extent->flags))
-		clear_new_delalloc_bytes = true;
+	    !test_bit(BTRFS_ORDERED_DIRECT, &ordered_extent->flags) &&
+	    !test_bit(BTRFS_ORDERED_ENCODED, &ordered_extent->flags))
+		clear_bits |= EXTENT_DELALLOC_NEW;
 
-	freespace_inode = btrfs_is_free_space_inode(BTRFS_I(inode));
+	freespace_inode = btrfs_is_free_space_inode(inode);
+	if (!freespace_inode)
+		btrfs_lockdep_acquire(fs_info, btrfs_ordered_extent);
 
 	if (test_bit(BTRFS_ORDERED_IOERR, &ordered_extent->flags)) {
 		ret = -EIO;
 		goto out;
 	}
 
-	btrfs_free_io_failure_record(BTRFS_I(inode), start, end);
+	/* A valid bdev implies a write on a sequential zone */
+	if (ordered_extent->bdev) {
+		btrfs_rewrite_logical_zoned(ordered_extent);
+		btrfs_zone_finish_endio(fs_info, ordered_extent->disk_bytenr,
+					ordered_extent->disk_num_bytes);
+	}
+
+	btrfs_free_io_failure_record(inode, start, end);
 
 	if (test_bit(BTRFS_ORDERED_TRUNCATED, &ordered_extent->flags)) {
 		truncated = true;
@@ -2529,14 +3252,7 @@ static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent)
 	if (test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags)) {
 		BUG_ON(!list_empty(&ordered_extent->list)); /* Logic error */
 
-		/*
-		 * For mwrite(mmap + memset to write) case, we still reserve
-		 * space for NOCOW range.
-		 * As NOCOW won't cause a new delayed ref, just free the space
-		 */
-		btrfs_qgroup_free_data(inode, NULL, start,
-				       ordered_extent->num_bytes);
-		btrfs_ordered_update_i_size(inode, 0, ordered_extent);
+		btrfs_inode_safe_disk_i_size_write(inode, 0);
 		if (freespace_inode)
 			trans = btrfs_join_transaction_spacecache(root);
 		else
@@ -2546,15 +3262,15 @@ static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent)
 			trans = NULL;
 			goto out;
 		}
-		trans->block_rsv = &BTRFS_I(inode)->block_rsv;
+		trans->block_rsv = &inode->block_rsv;
 		ret = btrfs_update_inode_fallback(trans, root, inode);
 		if (ret) /* -ENOMEM or corruption */
 			btrfs_abort_transaction(trans, ret);
 		goto out;
 	}
 
-	range_locked = true;
-	lock_extent_bits(io_tree, start, end, &cached_state);
+	clear_bits |= EXTENT_LOCKED;
+	lock_extent(io_tree, start, end, &cached_state);
 
 	if (freespace_inode)
 		trans = btrfs_join_transaction_spacecache(root);
@@ -2566,26 +3282,21 @@ static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent)
 		goto out;
 	}
 
-	trans->block_rsv = &BTRFS_I(inode)->block_rsv;
+	trans->block_rsv = &inode->block_rsv;
 
 	if (test_bit(BTRFS_ORDERED_COMPRESSED, &ordered_extent->flags))
 		compress_type = ordered_extent->compress_type;
 	if (test_bit(BTRFS_ORDERED_PREALLOC, &ordered_extent->flags)) {
 		BUG_ON(compress_type);
-		btrfs_qgroup_free_data(inode, NULL, start,
-				       ordered_extent->num_bytes);
-		ret = btrfs_mark_extent_written(trans, BTRFS_I(inode),
+		ret = btrfs_mark_extent_written(trans, inode,
 						ordered_extent->file_offset,
 						ordered_extent->file_offset +
 						logical_len);
+		btrfs_zoned_release_data_reloc_bg(fs_info, ordered_extent->disk_bytenr,
+						  ordered_extent->disk_num_bytes);
 	} else {
 		BUG_ON(root == fs_info->tree_root);
-		ret = insert_reserved_file_extent(trans, inode, start,
-						ordered_extent->disk_bytenr,
-						ordered_extent->disk_num_bytes,
-						logical_len, logical_len,
-						compress_type, 0, 0,
-						BTRFS_FILE_EXTENT_REG);
+		ret = insert_ordered_extent_file_extent(trans, ordered_extent);
 		if (!ret) {
 			clear_reserved_extent = false;
 			btrfs_release_delalloc_bytes(fs_info,
@@ -2593,21 +3304,31 @@ static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent)
 						ordered_extent->disk_num_bytes);
 		}
 	}
-	unpin_extent_cache(&BTRFS_I(inode)->extent_tree,
-			   ordered_extent->file_offset,
+	unpin_extent_cache(&inode->extent_tree, ordered_extent->file_offset,
 			   ordered_extent->num_bytes, trans->transid);
 	if (ret < 0) {
 		btrfs_abort_transaction(trans, ret);
 		goto out;
 	}
 
-	ret = add_pending_csums(trans, inode, &ordered_extent->list);
+	ret = add_pending_csums(trans, &ordered_extent->list);
 	if (ret) {
 		btrfs_abort_transaction(trans, ret);
 		goto out;
 	}
 
-	btrfs_ordered_update_i_size(inode, 0, ordered_extent);
+	/*
+	 * If this is a new delalloc range, clear its new delalloc flag to
+	 * update the inode's number of bytes. This needs to be done first
+	 * before updating the inode item.
+	 */
+	if ((clear_bits & EXTENT_DELALLOC_NEW) &&
+	    !test_bit(BTRFS_ORDERED_TRUNCATED, &ordered_extent->flags))
+		clear_extent_bit(&inode->io_tree, start, end,
+				 EXTENT_DELALLOC_NEW | EXTENT_ADD_INODE_BYTES,
+				 &cached_state);
+
+	btrfs_inode_safe_disk_i_size_write(inode, 0);
 	ret = btrfs_update_inode_fallback(trans, root, inode);
 	if (ret) { /* -ENOMEM or corruption */
 		btrfs_abort_transaction(trans, ret);
@@ -2615,13 +3336,7 @@ static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent)
 	}
 	ret = 0;
 out:
-	clear_bits = EXTENT_DEFRAG;
-	if (range_locked)
-		clear_bits |= EXTENT_LOCKED;
-	if (clear_new_delalloc_bytes)
-		clear_bits |= EXTENT_DELALLOC_NEW;
-	clear_extent_bit(&BTRFS_I(inode)->io_tree, start, end, clear_bits,
-			 (clear_bits & EXTENT_LOCKED) ? 1 : 0, 0,
+	clear_extent_bit(&inode->io_tree, start, end, clear_bits,
 			 &cached_state);
 
 	if (trans)
@@ -2630,12 +3345,24 @@ out:
 	if (ret || truncated) {
 		u64 unwritten_start = start;
 
+		/*
+		 * If we failed to finish this ordered extent for any reason we
+		 * need to make sure BTRFS_ORDERED_IOERR is set on the ordered
+		 * extent, and mark the inode with the error if it wasn't
+		 * already set.  Any error during writeback would have already
+		 * set the mapping error, so we need to set it if we're the ones
+		 * marking this ordered extent as failed.
+		 */
+		if (ret && !test_and_set_bit(BTRFS_ORDERED_IOERR,
+					     &ordered_extent->flags))
+			mapping_set_error(ordered_extent->inode->i_mapping, -EIO);
+
 		if (truncated)
 			unwritten_start += logical_len;
 		clear_extent_uptodate(io_tree, unwritten_start, end, NULL);
 
-		/* Drop the cache for the part of the extent we didn't write. */
-		btrfs_drop_extent_cache(BTRFS_I(inode), unwritten_start, end, 0);
+		/* Drop extent maps for the part of the extent we didn't write. */
+		btrfs_drop_extent_map_range(inode, unwritten_start, end, false);
 
 		/*
 		 * If the ordered extent had an IOERR or something else went
@@ -2680,103 +3407,150 @@ out:
 	return ret;
 }
 
-static void finish_ordered_fn(struct btrfs_work *work)
+void btrfs_writepage_endio_finish_ordered(struct btrfs_inode *inode,
+					  struct page *page, u64 start,
+					  u64 end, bool uptodate)
 {
-	struct btrfs_ordered_extent *ordered_extent;
-	ordered_extent = container_of(work, struct btrfs_ordered_extent, work);
-	btrfs_finish_ordered_io(ordered_extent);
+	trace_btrfs_writepage_end_io_hook(inode, start, end, uptodate);
+
+	btrfs_mark_ordered_io_finished(inode, page, start, end + 1 - start, uptodate);
 }
 
-void btrfs_writepage_endio_finish_ordered(struct page *page, u64 start,
-					  u64 end, int uptodate)
+/*
+ * Verify the checksum for a single sector without any extra action that depend
+ * on the type of I/O.
+ */
+int btrfs_check_sector_csum(struct btrfs_fs_info *fs_info, struct page *page,
+			    u32 pgoff, u8 *csum, const u8 * const csum_expected)
 {
-	struct inode *inode = page->mapping->host;
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	struct btrfs_ordered_extent *ordered_extent = NULL;
-	struct btrfs_workqueue *wq;
+	SHASH_DESC_ON_STACK(shash, fs_info->csum_shash);
+	char *kaddr;
 
-	trace_btrfs_writepage_end_io_hook(page, start, end, uptodate);
+	ASSERT(pgoff + fs_info->sectorsize <= PAGE_SIZE);
 
-	ClearPagePrivate2(page);
-	if (!btrfs_dec_test_ordered_pending(inode, &ordered_extent, start,
-					    end - start + 1, uptodate))
-		return;
+	shash->tfm = fs_info->csum_shash;
 
-	if (btrfs_is_free_space_inode(BTRFS_I(inode)))
-		wq = fs_info->endio_freespace_worker;
-	else
-		wq = fs_info->endio_write_workers;
+	kaddr = kmap_local_page(page) + pgoff;
+	crypto_shash_digest(shash, kaddr, fs_info->sectorsize, csum);
+	kunmap_local(kaddr);
+
+	if (memcmp(csum, csum_expected, fs_info->csum_size))
+		return -EIO;
+	return 0;
+}
+
+static u8 *btrfs_csum_ptr(const struct btrfs_fs_info *fs_info, u8 *csums, u64 offset)
+{
+	u64 offset_in_sectors = offset >> fs_info->sectorsize_bits;
 
-	btrfs_init_work(&ordered_extent->work, finish_ordered_fn, NULL, NULL);
-	btrfs_queue_work(wq, &ordered_extent->work);
+	return csums + offset_in_sectors * fs_info->csum_size;
 }
 
-static int __readpage_endio_check(struct inode *inode,
-				  struct btrfs_io_bio *io_bio,
-				  int icsum, struct page *page,
-				  int pgoff, u64 start, size_t len)
+/*
+ * check_data_csum - verify checksum of one sector of uncompressed data
+ * @inode:	inode
+ * @bbio:	btrfs_bio which contains the csum
+ * @bio_offset:	offset to the beginning of the bio (in bytes)
+ * @page:	page where is the data to be verified
+ * @pgoff:	offset inside the page
+ *
+ * The length of such check is always one sector size.
+ *
+ * When csum mismatch is detected, we will also report the error and fill the
+ * corrupted range with zero. (Thus it needs the extra parameters)
+ */
+int btrfs_check_data_csum(struct inode *inode, struct btrfs_bio *bbio,
+			  u32 bio_offset, struct page *page, u32 pgoff)
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	SHASH_DESC_ON_STACK(shash, fs_info->csum_shash);
-	char *kaddr;
-	u16 csum_size = btrfs_super_csum_size(fs_info->super_copy);
+	u32 len = fs_info->sectorsize;
 	u8 *csum_expected;
 	u8 csum[BTRFS_CSUM_SIZE];
 
-	csum_expected = ((u8 *)io_bio->csum) + icsum * csum_size;
+	ASSERT(pgoff + len <= PAGE_SIZE);
 
-	kaddr = kmap_atomic(page);
-	shash->tfm = fs_info->csum_shash;
-
-	crypto_shash_init(shash);
-	crypto_shash_update(shash, kaddr + pgoff, len);
-	crypto_shash_final(shash, csum);
+	csum_expected = btrfs_csum_ptr(fs_info, bbio->csum, bio_offset);
 
-	if (memcmp(csum, csum_expected, csum_size))
+	if (btrfs_check_sector_csum(fs_info, page, pgoff, csum, csum_expected))
 		goto zeroit;
-
-	kunmap_atomic(kaddr);
 	return 0;
+
 zeroit:
-	btrfs_print_data_csum_error(BTRFS_I(inode), start, csum, csum_expected,
-				    io_bio->mirror_num);
-	memset(kaddr + pgoff, 1, len);
-	flush_dcache_page(page);
-	kunmap_atomic(kaddr);
+	btrfs_print_data_csum_error(BTRFS_I(inode),
+				    bbio->file_offset + bio_offset,
+				    csum, csum_expected, bbio->mirror_num);
+	if (bbio->device)
+		btrfs_dev_stat_inc_and_print(bbio->device,
+					     BTRFS_DEV_STAT_CORRUPTION_ERRS);
+	memzero_page(page, pgoff, len);
 	return -EIO;
 }
 
 /*
- * when reads are done, we need to check csums to verify the data is correct
+ * When reads are done, we need to check csums to verify the data is correct.
  * if there's a match, we allow the bio to finish.  If not, the code in
  * extent_io.c will try to find good copies for us.
+ *
+ * @bio_offset:	offset to the beginning of the bio (in bytes)
+ * @start:	file offset of the range start
+ * @end:	file offset of the range end (inclusive)
+ *
+ * Return a bitmap where bit set means a csum mismatch, and bit not set means
+ * csum match.
  */
-static int btrfs_readpage_end_io_hook(struct btrfs_io_bio *io_bio,
-				      u64 phy_offset, struct page *page,
-				      u64 start, u64 end, int mirror)
+unsigned int btrfs_verify_data_csum(struct btrfs_bio *bbio,
+				    u32 bio_offset, struct page *page,
+				    u64 start, u64 end)
 {
-	size_t offset = start - page_offset(page);
 	struct inode *inode = page->mapping->host;
+	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
 	struct btrfs_root *root = BTRFS_I(inode)->root;
+	const u32 sectorsize = root->fs_info->sectorsize;
+	u32 pg_off;
+	unsigned int result = 0;
 
-	if (PageChecked(page)) {
-		ClearPageChecked(page);
+	/*
+	 * This only happens for NODATASUM or compressed read.
+	 * Normally this should be covered by above check for compressed read
+	 * or the next check for NODATASUM.  Just do a quicker exit here.
+	 */
+	if (bbio->csum == NULL)
 		return 0;
-	}
 
 	if (BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM)
 		return 0;
 
-	if (root->root_key.objectid == BTRFS_DATA_RELOC_TREE_OBJECTID &&
-	    test_range_bit(io_tree, start, end, EXTENT_NODATASUM, 1, NULL)) {
-		clear_extent_bits(io_tree, start, end, EXTENT_NODATASUM);
+	if (unlikely(test_bit(BTRFS_FS_STATE_NO_CSUMS, &fs_info->fs_state)))
 		return 0;
-	}
 
-	phy_offset >>= inode->i_sb->s_blocksize_bits;
-	return __readpage_endio_check(inode, io_bio, phy_offset, page, offset,
-				      start, (size_t)(end - start + 1));
+	ASSERT(page_offset(page) <= start &&
+	       end <= page_offset(page) + PAGE_SIZE - 1);
+	for (pg_off = offset_in_page(start);
+	     pg_off < offset_in_page(end);
+	     pg_off += sectorsize, bio_offset += sectorsize) {
+		u64 file_offset = pg_off + page_offset(page);
+		int ret;
+
+		if (btrfs_is_data_reloc_root(root) &&
+		    test_range_bit(io_tree, file_offset,
+				   file_offset + sectorsize - 1,
+				   EXTENT_NODATASUM, 1, NULL)) {
+			/* Skip the range without csum for data reloc inode */
+			clear_extent_bits(io_tree, file_offset,
+					  file_offset + sectorsize - 1,
+					  EXTENT_NODATASUM);
+			continue;
+		}
+		ret = btrfs_check_data_csum(inode, bbio, bio_offset, page, pg_off);
+		if (ret < 0) {
+			const int nr_bit = (pg_off - offset_in_page(start)) >>
+				     root->fs_info->sectorsize_bits;
+
+			result |= (1U << nr_bit);
+		}
+	}
+	return result;
 }
 
 /*
@@ -2838,19 +3612,22 @@ void btrfs_run_delayed_iputs(struct btrfs_fs_info *fs_info)
 		inode = list_first_entry(&fs_info->delayed_iputs,
 				struct btrfs_inode, delayed_iput);
 		run_delayed_iput_locked(fs_info, inode);
+		cond_resched_lock(&fs_info->delayed_iput_lock);
 	}
 	spin_unlock(&fs_info->delayed_iput_lock);
 }
 
 /**
- * btrfs_wait_on_delayed_iputs - wait on the delayed iputs to be done running
- * @fs_info - the fs_info for this fs
- * @return - EINTR if we were killed, 0 if nothing's pending
+ * Wait for flushing all delayed iputs
+ *
+ * @fs_info:  the filesystem
  *
  * This will wait on any delayed iputs that are currently running with KILLABLE
  * set.  Once they are all done running we will return, unless we are killed in
  * which case we return EINTR. This helps in user operations like fallocate etc
  * that might get blocked on the iputs.
+ *
+ * Return EINTR if we were killed, 0 if nothing's pending
  */
 int btrfs_wait_on_delayed_iputs(struct btrfs_fs_info *fs_info)
 {
@@ -2904,7 +3681,7 @@ int btrfs_orphan_cleanup(struct btrfs_root *root)
 	u64 last_objectid = 0;
 	int ret = 0, nr_unlink = 0;
 
-	if (cmpxchg(&root->orphan_cleanup_state, 0, ORPHAN_CLEANUP_STARTED))
+	if (test_and_set_bit(BTRFS_ROOT_ORPHAN_CLEANUP, &root->state))
 		return 0;
 
 	path = btrfs_alloc_path();
@@ -2966,37 +3743,38 @@ int btrfs_orphan_cleanup(struct btrfs_root *root)
 		found_key.objectid = found_key.offset;
 		found_key.type = BTRFS_INODE_ITEM_KEY;
 		found_key.offset = 0;
-		inode = btrfs_iget(fs_info->sb, &found_key, root);
+		inode = btrfs_iget(fs_info->sb, last_objectid, root);
 		ret = PTR_ERR_OR_ZERO(inode);
 		if (ret && ret != -ENOENT)
 			goto out;
 
 		if (ret == -ENOENT && root == fs_info->tree_root) {
 			struct btrfs_root *dead_root;
-			struct btrfs_fs_info *fs_info = root->fs_info;
 			int is_dead_root = 0;
 
 			/*
-			 * this is an orphan in the tree root. Currently these
+			 * This is an orphan in the tree root. Currently these
 			 * could come from 2 sources:
-			 *  a) a snapshot deletion in progress
+			 *  a) a root (snapshot/subvolume) deletion in progress
 			 *  b) a free space cache inode
-			 * We need to distinguish those two, as the snapshot
-			 * orphan must not get deleted.
-			 * find_dead_roots already ran before us, so if this
-			 * is a snapshot deletion, we should find the root
-			 * in the dead_roots list
+			 * We need to distinguish those two, as the orphan item
+			 * for a root must not get deleted before the deletion
+			 * of the snapshot/subvolume's tree completes.
+			 *
+			 * btrfs_find_orphan_roots() ran before us, which has
+			 * found all deleted roots and loaded them into
+			 * fs_info->fs_roots_radix. So here we can find if an
+			 * orphan item corresponds to a deleted root by looking
+			 * up the root from that radix tree.
 			 */
-			spin_lock(&fs_info->trans_lock);
-			list_for_each_entry(dead_root, &fs_info->dead_roots,
-					    root_list) {
-				if (dead_root->root_key.objectid ==
-				    found_key.objectid) {
-					is_dead_root = 1;
-					break;
-				}
-			}
-			spin_unlock(&fs_info->trans_lock);
+
+			spin_lock(&fs_info->fs_roots_radix_lock);
+			dead_root = radix_tree_lookup(&fs_info->fs_roots_radix,
+							 (unsigned long)found_key.objectid);
+			if (dead_root && btrfs_root_refs(&dead_root->root_item) == 0)
+				is_dead_root = 1;
+			spin_unlock(&fs_info->fs_roots_radix_lock);
+
 			if (is_dead_root) {
 				/* prevent this orphan from being found again */
 				key.offset = found_key.objectid - 1;
@@ -3007,7 +3785,14 @@ int btrfs_orphan_cleanup(struct btrfs_root *root)
 
 		/*
 		 * If we have an inode with links, there are a couple of
-		 * possibilities. Old kernels (before v3.12) used to create an
+		 * possibilities:
+		 *
+		 * 1. We were halfway through creating fsverity metadata for the
+		 * file. In that case, the orphan item represents incomplete
+		 * fsverity metadata which must be cleaned up with
+		 * btrfs_drop_verity_items and deleting the orphan item.
+
+		 * 2. Old kernels (before v3.12) used to create an
 		 * orphan item for truncate indicating that there were possibly
 		 * extent items past i_size that needed to be deleted. In v3.12,
 		 * truncate was changed to update i_size in sync with the extent
@@ -3025,8 +3810,12 @@ int btrfs_orphan_cleanup(struct btrfs_root *root)
 		 * but either way, we can delete the orphan item.
 		 */
 		if (ret == -ENOENT || inode->i_nlink) {
-			if (!ret)
+			if (!ret) {
+				ret = btrfs_drop_verity_items(BTRFS_I(inode));
 				iput(inode);
+				if (ret)
+					goto out;
+			}
 			trans = btrfs_start_transaction(root, 1);
 			if (IS_ERR(trans)) {
 				ret = PTR_ERR(trans);
@@ -3050,8 +3839,6 @@ int btrfs_orphan_cleanup(struct btrfs_root *root)
 	/* release the path since we're done with it */
 	btrfs_release_path(path);
 
-	root->orphan_cleanup_state = ORPHAN_CLEANUP_DONE;
-
 	if (test_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED, &root->state)) {
 		trans = btrfs_join_transaction(root);
 		if (!IS_ERR(trans))
@@ -3187,6 +3974,8 @@ static int btrfs_read_locked_inode(struct inode *inode,
 	i_uid_write(inode, btrfs_inode_uid(leaf, inode_item));
 	i_gid_write(inode, btrfs_inode_gid(leaf, inode_item));
 	btrfs_i_size_write(BTRFS_I(inode), btrfs_inode_size(leaf, inode_item));
+	btrfs_inode_set_file_extent_range(BTRFS_I(inode), 0,
+			round_up(i_size_read(inode), fs_info->sectorsize));
 
 	inode->i_atime.tv_sec = btrfs_timespec_sec(leaf, &inode_item->atime);
 	inode->i_atime.tv_nsec = btrfs_timespec_nsec(leaf, &inode_item->atime);
@@ -3213,7 +4002,8 @@ static int btrfs_read_locked_inode(struct inode *inode,
 	rdev = btrfs_inode_rdev(leaf, inode_item);
 
 	BTRFS_I(inode)->index_cnt = (u64)-1;
-	BTRFS_I(inode)->flags = btrfs_inode_flags(leaf, inode_item);
+	btrfs_inode_split_flags(btrfs_inode_flags(leaf, inode_item),
+				&BTRFS_I(inode)->flags, &BTRFS_I(inode)->ro_flags);
 
 cache_index:
 	/*
@@ -3258,6 +4048,14 @@ cache_index:
 	 */
 	BTRFS_I(inode)->last_unlink_trans = BTRFS_I(inode)->last_trans;
 
+	/*
+	 * Same logic as for last_unlink_trans. We don't persist the generation
+	 * of the last transaction where this inode was used for a reflink
+	 * operation, so after eviction and reloading the inode we must be
+	 * pessimistic and assume the last transaction that modified the inode.
+	 */
+	BTRFS_I(inode)->last_reflink_trans = BTRFS_I(inode)->last_trans;
+
 	path->slots[0]++;
 	if (inode->i_nlink != 1 ||
 	    path->slots[0] >= btrfs_header_nritems(leaf))
@@ -3305,7 +4103,6 @@ cache_acl:
 	switch (inode->i_mode & S_IFMT) {
 	case S_IFREG:
 		inode->i_mapping->a_ops = &btrfs_aops;
-		BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops;
 		inode->i_fop = &btrfs_file_operations;
 		inode->i_op = &btrfs_file_inode_operations;
 		break;
@@ -3337,53 +4134,54 @@ static void fill_inode_item(struct btrfs_trans_handle *trans,
 			    struct inode *inode)
 {
 	struct btrfs_map_token token;
+	u64 flags;
 
 	btrfs_init_map_token(&token, leaf);
 
-	btrfs_set_token_inode_uid(leaf, item, i_uid_read(inode), &token);
-	btrfs_set_token_inode_gid(leaf, item, i_gid_read(inode), &token);
-	btrfs_set_token_inode_size(leaf, item, BTRFS_I(inode)->disk_i_size,
-				   &token);
-	btrfs_set_token_inode_mode(leaf, item, inode->i_mode, &token);
-	btrfs_set_token_inode_nlink(leaf, item, inode->i_nlink, &token);
-
-	btrfs_set_token_timespec_sec(leaf, &item->atime,
-				     inode->i_atime.tv_sec, &token);
-	btrfs_set_token_timespec_nsec(leaf, &item->atime,
-				      inode->i_atime.tv_nsec, &token);
-
-	btrfs_set_token_timespec_sec(leaf, &item->mtime,
-				     inode->i_mtime.tv_sec, &token);
-	btrfs_set_token_timespec_nsec(leaf, &item->mtime,
-				      inode->i_mtime.tv_nsec, &token);
-
-	btrfs_set_token_timespec_sec(leaf, &item->ctime,
-				     inode->i_ctime.tv_sec, &token);
-	btrfs_set_token_timespec_nsec(leaf, &item->ctime,
-				      inode->i_ctime.tv_nsec, &token);
-
-	btrfs_set_token_timespec_sec(leaf, &item->otime,
-				     BTRFS_I(inode)->i_otime.tv_sec, &token);
-	btrfs_set_token_timespec_nsec(leaf, &item->otime,
-				      BTRFS_I(inode)->i_otime.tv_nsec, &token);
-
-	btrfs_set_token_inode_nbytes(leaf, item, inode_get_bytes(inode),
-				     &token);
-	btrfs_set_token_inode_generation(leaf, item, BTRFS_I(inode)->generation,
-					 &token);
-	btrfs_set_token_inode_sequence(leaf, item, inode_peek_iversion(inode),
-				       &token);
-	btrfs_set_token_inode_transid(leaf, item, trans->transid, &token);
-	btrfs_set_token_inode_rdev(leaf, item, inode->i_rdev, &token);
-	btrfs_set_token_inode_flags(leaf, item, BTRFS_I(inode)->flags, &token);
-	btrfs_set_token_inode_block_group(leaf, item, 0, &token);
+	btrfs_set_token_inode_uid(&token, item, i_uid_read(inode));
+	btrfs_set_token_inode_gid(&token, item, i_gid_read(inode));
+	btrfs_set_token_inode_size(&token, item, BTRFS_I(inode)->disk_i_size);
+	btrfs_set_token_inode_mode(&token, item, inode->i_mode);
+	btrfs_set_token_inode_nlink(&token, item, inode->i_nlink);
+
+	btrfs_set_token_timespec_sec(&token, &item->atime,
+				     inode->i_atime.tv_sec);
+	btrfs_set_token_timespec_nsec(&token, &item->atime,
+				      inode->i_atime.tv_nsec);
+
+	btrfs_set_token_timespec_sec(&token, &item->mtime,
+				     inode->i_mtime.tv_sec);
+	btrfs_set_token_timespec_nsec(&token, &item->mtime,
+				      inode->i_mtime.tv_nsec);
+
+	btrfs_set_token_timespec_sec(&token, &item->ctime,
+				     inode->i_ctime.tv_sec);
+	btrfs_set_token_timespec_nsec(&token, &item->ctime,
+				      inode->i_ctime.tv_nsec);
+
+	btrfs_set_token_timespec_sec(&token, &item->otime,
+				     BTRFS_I(inode)->i_otime.tv_sec);
+	btrfs_set_token_timespec_nsec(&token, &item->otime,
+				      BTRFS_I(inode)->i_otime.tv_nsec);
+
+	btrfs_set_token_inode_nbytes(&token, item, inode_get_bytes(inode));
+	btrfs_set_token_inode_generation(&token, item,
+					 BTRFS_I(inode)->generation);
+	btrfs_set_token_inode_sequence(&token, item, inode_peek_iversion(inode));
+	btrfs_set_token_inode_transid(&token, item, trans->transid);
+	btrfs_set_token_inode_rdev(&token, item, inode->i_rdev);
+	flags = btrfs_inode_combine_flags(BTRFS_I(inode)->flags,
+					  BTRFS_I(inode)->ro_flags);
+	btrfs_set_token_inode_flags(&token, item, flags);
+	btrfs_set_token_inode_block_group(&token, item, 0);
 }
 
 /*
  * copy everything in the in-memory inode into the btree.
  */
 static noinline int btrfs_update_inode_item(struct btrfs_trans_handle *trans,
-				struct btrfs_root *root, struct inode *inode)
+				struct btrfs_root *root,
+				struct btrfs_inode *inode)
 {
 	struct btrfs_inode_item *inode_item;
 	struct btrfs_path *path;
@@ -3394,9 +4192,7 @@ static noinline int btrfs_update_inode_item(struct btrfs_trans_handle *trans,
 	if (!path)
 		return -ENOMEM;
 
-	path->leave_spinning = 1;
-	ret = btrfs_lookup_inode(trans, root, path, &BTRFS_I(inode)->location,
-				 1);
+	ret = btrfs_lookup_inode(trans, root, path, &inode->location, 1);
 	if (ret) {
 		if (ret > 0)
 			ret = -ENOENT;
@@ -3407,7 +4203,7 @@ static noinline int btrfs_update_inode_item(struct btrfs_trans_handle *trans,
 	inode_item = btrfs_item_ptr(leaf, path->slots[0],
 				    struct btrfs_inode_item);
 
-	fill_inode_item(trans, leaf, inode_item, inode);
+	fill_inode_item(trans, leaf, inode_item, &inode->vfs_inode);
 	btrfs_mark_buffer_dirty(leaf);
 	btrfs_set_inode_last_trans(trans, inode);
 	ret = 0;
@@ -3420,7 +4216,8 @@ failed:
  * copy everything in the in-memory inode into the btree.
  */
 noinline int btrfs_update_inode(struct btrfs_trans_handle *trans,
-				struct btrfs_root *root, struct inode *inode)
+				struct btrfs_root *root,
+				struct btrfs_inode *inode)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	int ret;
@@ -3432,8 +4229,8 @@ noinline int btrfs_update_inode(struct btrfs_trans_handle *trans,
 	 * The data relocation inode should also be directly updated
 	 * without delay
 	 */
-	if (!btrfs_is_free_space_inode(BTRFS_I(inode))
-	    && root->root_key.objectid != BTRFS_DATA_RELOC_TREE_OBJECTID
+	if (!btrfs_is_free_space_inode(inode)
+	    && !btrfs_is_data_reloc_root(root)
 	    && !test_bit(BTRFS_FS_LOG_RECOVERING, &fs_info->flags)) {
 		btrfs_update_root_times(trans, root);
 
@@ -3446,9 +4243,8 @@ noinline int btrfs_update_inode(struct btrfs_trans_handle *trans,
 	return btrfs_update_inode_item(trans, root, inode);
 }
 
-noinline int btrfs_update_inode_fallback(struct btrfs_trans_handle *trans,
-					 struct btrfs_root *root,
-					 struct inode *inode)
+int btrfs_update_inode_fallback(struct btrfs_trans_handle *trans,
+				struct btrfs_root *root, struct btrfs_inode *inode)
 {
 	int ret;
 
@@ -3464,11 +4260,12 @@ noinline int btrfs_update_inode_fallback(struct btrfs_trans_handle *trans,
  * also drops the back refs in the inode to the directory
  */
 static int __btrfs_unlink_inode(struct btrfs_trans_handle *trans,
-				struct btrfs_root *root,
 				struct btrfs_inode *dir,
 				struct btrfs_inode *inode,
-				const char *name, int name_len)
+				const char *name, int name_len,
+				struct btrfs_rename_ctx *rename_ctx)
 {
+	struct btrfs_root *root = dir->root;
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct btrfs_path *path;
 	int ret = 0;
@@ -3483,7 +4280,6 @@ static int __btrfs_unlink_inode(struct btrfs_trans_handle *trans,
 		goto out;
 	}
 
-	path->leave_spinning = 1;
 	di = btrfs_lookup_dir_item(trans, root, path, dir_ino,
 				    name, name_len, -1);
 	if (IS_ERR_OR_NULL(di)) {
@@ -3523,26 +4319,28 @@ static int __btrfs_unlink_inode(struct btrfs_trans_handle *trans,
 		goto err;
 	}
 skip_backref:
+	if (rename_ctx)
+		rename_ctx->index = index;
+
 	ret = btrfs_delete_delayed_dir_index(trans, dir, index);
 	if (ret) {
 		btrfs_abort_transaction(trans, ret);
 		goto err;
 	}
 
-	ret = btrfs_del_inode_ref_in_log(trans, root, name, name_len, inode,
-			dir_ino);
-	if (ret != 0 && ret != -ENOENT) {
-		btrfs_abort_transaction(trans, ret);
-		goto err;
+	/*
+	 * If we are in a rename context, we don't need to update anything in the
+	 * log. That will be done later during the rename by btrfs_log_new_name().
+	 * Besides that, doing it here would only cause extra unnecessary btree
+	 * operations on the log tree, increasing latency for applications.
+	 */
+	if (!rename_ctx) {
+		btrfs_del_inode_ref_in_log(trans, root, name, name_len, inode,
+					   dir_ino);
+		btrfs_del_dir_entries_in_log(trans, root, name, name_len, dir,
+					     index);
 	}
 
-	ret = btrfs_del_dir_entries_in_log(trans, root, name, name_len, dir,
-			index);
-	if (ret == -ENOENT)
-		ret = 0;
-	else if (ret)
-		btrfs_abort_transaction(trans, ret);
-
 	/*
 	 * If we have a pending delayed iput we could end up with the final iput
 	 * being run in btrfs-cleaner context.  If we have enough of these built
@@ -3561,23 +4359,23 @@ err:
 	btrfs_i_size_write(dir, dir->vfs_inode.i_size - name_len * 2);
 	inode_inc_iversion(&inode->vfs_inode);
 	inode_inc_iversion(&dir->vfs_inode);
-	inode->vfs_inode.i_ctime = dir->vfs_inode.i_mtime =
-		dir->vfs_inode.i_ctime = current_time(&inode->vfs_inode);
-	ret = btrfs_update_inode(trans, root, &dir->vfs_inode);
+	inode->vfs_inode.i_ctime = current_time(&inode->vfs_inode);
+	dir->vfs_inode.i_mtime = inode->vfs_inode.i_ctime;
+	dir->vfs_inode.i_ctime = inode->vfs_inode.i_ctime;
+	ret = btrfs_update_inode(trans, root, dir);
 out:
 	return ret;
 }
 
 int btrfs_unlink_inode(struct btrfs_trans_handle *trans,
-		       struct btrfs_root *root,
 		       struct btrfs_inode *dir, struct btrfs_inode *inode,
 		       const char *name, int name_len)
 {
 	int ret;
-	ret = __btrfs_unlink_inode(trans, root, dir, inode, name, name_len);
+	ret = __btrfs_unlink_inode(trans, dir, inode, name, name_len, NULL);
 	if (!ret) {
 		drop_nlink(&inode->vfs_inode);
-		ret = btrfs_update_inode(trans, root, &inode->vfs_inode);
+		ret = btrfs_update_inode(trans, inode->root, inode);
 	}
 	return ret;
 }
@@ -3600,13 +4398,13 @@ static struct btrfs_trans_handle *__unlink_start_trans(struct inode *dir)
 	 * 1 for the dir index
 	 * 1 for the inode ref
 	 * 1 for the inode
+	 * 1 for the parent inode
 	 */
-	return btrfs_start_transaction_fallback_global_rsv(root, 5, 5);
+	return btrfs_start_transaction_fallback_global_rsv(root, 6);
 }
 
 static int btrfs_unlink(struct inode *dir, struct dentry *dentry)
 {
-	struct btrfs_root *root = BTRFS_I(dir)->root;
 	struct btrfs_trans_handle *trans;
 	struct inode *inode = d_inode(dentry);
 	int ret;
@@ -3618,7 +4416,7 @@ static int btrfs_unlink(struct inode *dir, struct dentry *dentry)
 	btrfs_record_unlink_dir(trans, BTRFS_I(dir), BTRFS_I(d_inode(dentry)),
 			0);
 
-	ret = btrfs_unlink_inode(trans, root, BTRFS_I(dir),
+	ret = btrfs_unlink_inode(trans, BTRFS_I(dir),
 			BTRFS_I(d_inode(dentry)), dentry->d_name.name,
 			dentry->d_name.len);
 	if (ret)
@@ -3632,7 +4430,7 @@ static int btrfs_unlink(struct inode *dir, struct dentry *dentry)
 
 out:
 	btrfs_end_transaction(trans);
-	btrfs_btree_balance_dirty(root->fs_info);
+	btrfs_btree_balance_dirty(BTRFS_I(dir)->root->fs_info);
 	return ret;
 }
 
@@ -3686,7 +4484,7 @@ static int btrfs_unlink_subvol(struct btrfs_trans_handle *trans,
 	 * This is a placeholder inode for a subvolume we didn't have a
 	 * reference to at the time of the snapshot creation.  In the meantime
 	 * we could have renamed the real subvol link into our snapshot, so
-	 * depending on btrfs_del_root_ref to return -ENOENT here is incorret.
+	 * depending on btrfs_del_root_ref to return -ENOENT here is incorrect.
 	 * Instead simply lookup the dir_index_item for this entry so we can
 	 * remove it.  Otherwise we know we have a ref to the root and we can
 	 * call btrfs_del_root_ref, and it _shouldn't_ fail.
@@ -3725,8 +4523,9 @@ static int btrfs_unlink_subvol(struct btrfs_trans_handle *trans,
 
 	btrfs_i_size_write(BTRFS_I(dir), dir->i_size - name_len * 2);
 	inode_inc_iversion(dir);
-	dir->i_mtime = dir->i_ctime = current_time(dir);
-	ret = btrfs_update_inode_fallback(trans, root, dir);
+	dir->i_mtime = current_time(dir);
+	dir->i_ctime = dir->i_mtime;
+	ret = btrfs_update_inode_fallback(trans, root, BTRFS_I(dir));
 	if (ret)
 		btrfs_abort_transaction(trans, ret);
 out:
@@ -3799,7 +4598,7 @@ static void btrfs_prune_dentries(struct btrfs_root *root)
 	struct inode *inode;
 	u64 objectid = 0;
 
-	if (!test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state))
+	if (!BTRFS_FS_ERROR(fs_info))
 		WARN_ON(btrfs_root_refs(&root->root_item) != 0);
 
 	spin_lock(&root->inode_lock);
@@ -3863,7 +4662,6 @@ int btrfs_delete_subvolume(struct inode *dir, struct dentry *dentry)
 	struct btrfs_block_rsv block_rsv;
 	u64 root_flags;
 	int ret;
-	int err;
 
 	/*
 	 * Don't allow to delete a subvolume with send in progress. This is
@@ -3878,6 +4676,13 @@ int btrfs_delete_subvolume(struct inode *dir, struct dentry *dentry)
 			   dest->root_key.objectid);
 		return -EPERM;
 	}
+	if (atomic_read(&dest->nr_swapfiles)) {
+		spin_unlock(&dest->root_item_lock);
+		btrfs_warn(fs_info,
+			   "attempt to delete subvolume %llu with active swapfile",
+			   root->root_key.objectid);
+		return -EPERM;
+	}
 	root_flags = btrfs_root_flags(&dest->root_item);
 	btrfs_set_root_flags(&dest->root_item,
 			     root_flags | BTRFS_ROOT_SUBVOL_DEAD);
@@ -3885,8 +4690,8 @@ int btrfs_delete_subvolume(struct inode *dir, struct dentry *dentry)
 
 	down_write(&fs_info->subvol_sem);
 
-	err = may_destroy_subvol(dest);
-	if (err)
+	ret = may_destroy_subvol(dest);
+	if (ret)
 		goto out_up_write;
 
 	btrfs_init_block_rsv(&block_rsv, BTRFS_BLOCK_RSV_TEMP);
@@ -3895,13 +4700,13 @@ int btrfs_delete_subvolume(struct inode *dir, struct dentry *dentry)
 	 * two for dir entries,
 	 * two for root ref/backref.
 	 */
-	err = btrfs_subvolume_reserve_metadata(root, &block_rsv, 5, true);
-	if (err)
+	ret = btrfs_subvolume_reserve_metadata(root, &block_rsv, 5, true);
+	if (ret)
 		goto out_up_write;
 
 	trans = btrfs_start_transaction(root, 0);
 	if (IS_ERR(trans)) {
-		err = PTR_ERR(trans);
+		ret = PTR_ERR(trans);
 		goto out_release;
 	}
 	trans->block_rsv = &block_rsv;
@@ -3911,16 +4716,19 @@ int btrfs_delete_subvolume(struct inode *dir, struct dentry *dentry)
 
 	ret = btrfs_unlink_subvol(trans, dir, dentry);
 	if (ret) {
-		err = ret;
 		btrfs_abort_transaction(trans, ret);
 		goto out_end_trans;
 	}
 
-	btrfs_record_root_in_trans(trans, dest);
+	ret = btrfs_record_root_in_trans(trans, dest);
+	if (ret) {
+		btrfs_abort_transaction(trans, ret);
+		goto out_end_trans;
+	}
 
 	memset(&dest->root_item.drop_progress, 0,
 		sizeof(dest->root_item.drop_progress));
-	dest->root_item.drop_level = 0;
+	btrfs_set_root_drop_level(&dest->root_item, 0);
 	btrfs_set_root_refs(&dest->root_item, 0);
 
 	if (!test_and_set_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED, &dest->state)) {
@@ -3929,7 +4737,6 @@ int btrfs_delete_subvolume(struct inode *dir, struct dentry *dentry)
 					dest->root_key.objectid);
 		if (ret) {
 			btrfs_abort_transaction(trans, ret);
-			err = ret;
 			goto out_end_trans;
 		}
 	}
@@ -3939,7 +4746,6 @@ int btrfs_delete_subvolume(struct inode *dir, struct dentry *dentry)
 				  dest->root_key.objectid);
 	if (ret && ret != -ENOENT) {
 		btrfs_abort_transaction(trans, ret);
-		err = ret;
 		goto out_end_trans;
 	}
 	if (!btrfs_is_empty_uuid(dest->root_item.received_uuid)) {
@@ -3949,23 +4755,22 @@ int btrfs_delete_subvolume(struct inode *dir, struct dentry *dentry)
 					  dest->root_key.objectid);
 		if (ret && ret != -ENOENT) {
 			btrfs_abort_transaction(trans, ret);
-			err = ret;
 			goto out_end_trans;
 		}
 	}
 
+	free_anon_bdev(dest->anon_dev);
+	dest->anon_dev = 0;
 out_end_trans:
 	trans->block_rsv = NULL;
 	trans->bytes_reserved = 0;
 	ret = btrfs_end_transaction(trans);
-	if (ret && !err)
-		err = ret;
 	inode->i_flags |= S_DEAD;
 out_release:
-	btrfs_subvolume_release_metadata(fs_info, &block_rsv);
+	btrfs_subvolume_release_metadata(root, &block_rsv);
 out_up_write:
 	up_write(&fs_info->subvol_sem);
-	if (err) {
+	if (ret) {
 		spin_lock(&dest->root_item_lock);
 		root_flags = btrfs_root_flags(&dest->root_item);
 		btrfs_set_root_flags(&dest->root_item,
@@ -3975,29 +4780,29 @@ out_up_write:
 		d_invalidate(dentry);
 		btrfs_prune_dentries(dest);
 		ASSERT(dest->send_in_progress == 0);
-
-		/* the last ref */
-		if (dest->ino_cache_inode) {
-			iput(dest->ino_cache_inode);
-			dest->ino_cache_inode = NULL;
-		}
 	}
 
-	return err;
+	return ret;
 }
 
 static int btrfs_rmdir(struct inode *dir, struct dentry *dentry)
 {
 	struct inode *inode = d_inode(dentry);
+	struct btrfs_fs_info *fs_info = BTRFS_I(inode)->root->fs_info;
 	int err = 0;
-	struct btrfs_root *root = BTRFS_I(dir)->root;
 	struct btrfs_trans_handle *trans;
 	u64 last_unlink_trans;
 
 	if (inode->i_size > BTRFS_EMPTY_DIR_SIZE)
 		return -ENOTEMPTY;
-	if (btrfs_ino(BTRFS_I(inode)) == BTRFS_FIRST_FREE_OBJECTID)
+	if (btrfs_ino(BTRFS_I(inode)) == BTRFS_FIRST_FREE_OBJECTID) {
+		if (unlikely(btrfs_fs_incompat(fs_info, EXTENT_TREE_V2))) {
+			btrfs_err(fs_info,
+			"extent tree v2 doesn't support snapshot deletion yet");
+			return -EOPNOTSUPP;
+		}
 		return btrfs_delete_subvolume(dir, dentry);
+	}
 
 	trans = __unlink_start_trans(dir);
 	if (IS_ERR(trans))
@@ -4015,7 +4820,7 @@ static int btrfs_rmdir(struct inode *dir, struct dentry *dentry)
 	last_unlink_trans = BTRFS_I(inode)->last_unlink_trans;
 
 	/* now the directory is empty */
-	err = btrfs_unlink_inode(trans, root, BTRFS_I(dir),
+	err = btrfs_unlink_inode(trans, BTRFS_I(dir),
 			BTRFS_I(d_inode(dentry)), dentry->d_name.name,
 			dentry->d_name.len);
 	if (!err) {
@@ -4036,348 +4841,12 @@ static int btrfs_rmdir(struct inode *dir, struct dentry *dentry)
 	}
 out:
 	btrfs_end_transaction(trans);
-	btrfs_btree_balance_dirty(root->fs_info);
+	btrfs_btree_balance_dirty(fs_info);
 
 	return err;
 }
 
 /*
- * Return this if we need to call truncate_block for the last bit of the
- * truncate.
- */
-#define NEED_TRUNCATE_BLOCK 1
-
-/*
- * this can truncate away extent items, csum items and directory items.
- * It starts at a high offset and removes keys until it can't find
- * any higher than new_size
- *
- * csum items that cross the new i_size are truncated to the new size
- * as well.
- *
- * min_type is the minimum key type to truncate down to.  If set to 0, this
- * will kill all the items on this inode, including the INODE_ITEM_KEY.
- */
-int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans,
-			       struct btrfs_root *root,
-			       struct inode *inode,
-			       u64 new_size, u32 min_type)
-{
-	struct btrfs_fs_info *fs_info = root->fs_info;
-	struct btrfs_path *path;
-	struct extent_buffer *leaf;
-	struct btrfs_file_extent_item *fi;
-	struct btrfs_key key;
-	struct btrfs_key found_key;
-	u64 extent_start = 0;
-	u64 extent_num_bytes = 0;
-	u64 extent_offset = 0;
-	u64 item_end = 0;
-	u64 last_size = new_size;
-	u32 found_type = (u8)-1;
-	int found_extent;
-	int del_item;
-	int pending_del_nr = 0;
-	int pending_del_slot = 0;
-	int extent_type = -1;
-	int ret;
-	u64 ino = btrfs_ino(BTRFS_I(inode));
-	u64 bytes_deleted = 0;
-	bool be_nice = false;
-	bool should_throttle = false;
-	const u64 lock_start = ALIGN_DOWN(new_size, fs_info->sectorsize);
-	struct extent_state *cached_state = NULL;
-
-	BUG_ON(new_size > 0 && min_type != BTRFS_EXTENT_DATA_KEY);
-
-	/*
-	 * for non-free space inodes and ref cows, we want to back off from
-	 * time to time
-	 */
-	if (!btrfs_is_free_space_inode(BTRFS_I(inode)) &&
-	    test_bit(BTRFS_ROOT_REF_COWS, &root->state))
-		be_nice = true;
-
-	path = btrfs_alloc_path();
-	if (!path)
-		return -ENOMEM;
-	path->reada = READA_BACK;
-
-	if (root->root_key.objectid != BTRFS_TREE_LOG_OBJECTID)
-		lock_extent_bits(&BTRFS_I(inode)->io_tree, lock_start, (u64)-1,
-				 &cached_state);
-
-	/*
-	 * We want to drop from the next block forward in case this new size is
-	 * not block aligned since we will be keeping the last block of the
-	 * extent just the way it is.
-	 */
-	if (test_bit(BTRFS_ROOT_REF_COWS, &root->state) ||
-	    root == fs_info->tree_root)
-		btrfs_drop_extent_cache(BTRFS_I(inode), ALIGN(new_size,
-					fs_info->sectorsize),
-					(u64)-1, 0);
-
-	/*
-	 * This function is also used to drop the items in the log tree before
-	 * we relog the inode, so if root != BTRFS_I(inode)->root, it means
-	 * it is used to drop the logged items. So we shouldn't kill the delayed
-	 * items.
-	 */
-	if (min_type == 0 && root == BTRFS_I(inode)->root)
-		btrfs_kill_delayed_inode_items(BTRFS_I(inode));
-
-	key.objectid = ino;
-	key.offset = (u64)-1;
-	key.type = (u8)-1;
-
-search_again:
-	/*
-	 * with a 16K leaf size and 128MB extents, you can actually queue
-	 * up a huge file in a single leaf.  Most of the time that
-	 * bytes_deleted is > 0, it will be huge by the time we get here
-	 */
-	if (be_nice && bytes_deleted > SZ_32M &&
-	    btrfs_should_end_transaction(trans)) {
-		ret = -EAGAIN;
-		goto out;
-	}
-
-	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
-	if (ret < 0)
-		goto out;
-
-	if (ret > 0) {
-		ret = 0;
-		/* there are no items in the tree for us to truncate, we're
-		 * done
-		 */
-		if (path->slots[0] == 0)
-			goto out;
-		path->slots[0]--;
-	}
-
-	while (1) {
-		fi = NULL;
-		leaf = path->nodes[0];
-		btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
-		found_type = found_key.type;
-
-		if (found_key.objectid != ino)
-			break;
-
-		if (found_type < min_type)
-			break;
-
-		item_end = found_key.offset;
-		if (found_type == BTRFS_EXTENT_DATA_KEY) {
-			fi = btrfs_item_ptr(leaf, path->slots[0],
-					    struct btrfs_file_extent_item);
-			extent_type = btrfs_file_extent_type(leaf, fi);
-			if (extent_type != BTRFS_FILE_EXTENT_INLINE) {
-				item_end +=
-				    btrfs_file_extent_num_bytes(leaf, fi);
-
-				trace_btrfs_truncate_show_fi_regular(
-					BTRFS_I(inode), leaf, fi,
-					found_key.offset);
-			} else if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
-				item_end += btrfs_file_extent_ram_bytes(leaf,
-									fi);
-
-				trace_btrfs_truncate_show_fi_inline(
-					BTRFS_I(inode), leaf, fi, path->slots[0],
-					found_key.offset);
-			}
-			item_end--;
-		}
-		if (found_type > min_type) {
-			del_item = 1;
-		} else {
-			if (item_end < new_size)
-				break;
-			if (found_key.offset >= new_size)
-				del_item = 1;
-			else
-				del_item = 0;
-		}
-		found_extent = 0;
-		/* FIXME, shrink the extent if the ref count is only 1 */
-		if (found_type != BTRFS_EXTENT_DATA_KEY)
-			goto delete;
-
-		if (extent_type != BTRFS_FILE_EXTENT_INLINE) {
-			u64 num_dec;
-			extent_start = btrfs_file_extent_disk_bytenr(leaf, fi);
-			if (!del_item) {
-				u64 orig_num_bytes =
-					btrfs_file_extent_num_bytes(leaf, fi);
-				extent_num_bytes = ALIGN(new_size -
-						found_key.offset,
-						fs_info->sectorsize);
-				btrfs_set_file_extent_num_bytes(leaf, fi,
-							 extent_num_bytes);
-				num_dec = (orig_num_bytes -
-					   extent_num_bytes);
-				if (test_bit(BTRFS_ROOT_REF_COWS,
-					     &root->state) &&
-				    extent_start != 0)
-					inode_sub_bytes(inode, num_dec);
-				btrfs_mark_buffer_dirty(leaf);
-			} else {
-				extent_num_bytes =
-					btrfs_file_extent_disk_num_bytes(leaf,
-									 fi);
-				extent_offset = found_key.offset -
-					btrfs_file_extent_offset(leaf, fi);
-
-				/* FIXME blocksize != 4096 */
-				num_dec = btrfs_file_extent_num_bytes(leaf, fi);
-				if (extent_start != 0) {
-					found_extent = 1;
-					if (test_bit(BTRFS_ROOT_REF_COWS,
-						     &root->state))
-						inode_sub_bytes(inode, num_dec);
-				}
-			}
-		} else if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
-			/*
-			 * we can't truncate inline items that have had
-			 * special encodings
-			 */
-			if (!del_item &&
-			    btrfs_file_extent_encryption(leaf, fi) == 0 &&
-			    btrfs_file_extent_other_encoding(leaf, fi) == 0 &&
-			    btrfs_file_extent_compression(leaf, fi) == 0) {
-				u32 size = (u32)(new_size - found_key.offset);
-
-				btrfs_set_file_extent_ram_bytes(leaf, fi, size);
-				size = btrfs_file_extent_calc_inline_size(size);
-				btrfs_truncate_item(path, size, 1);
-			} else if (!del_item) {
-				/*
-				 * We have to bail so the last_size is set to
-				 * just before this extent.
-				 */
-				ret = NEED_TRUNCATE_BLOCK;
-				break;
-			}
-
-			if (test_bit(BTRFS_ROOT_REF_COWS, &root->state))
-				inode_sub_bytes(inode, item_end + 1 - new_size);
-		}
-delete:
-		if (del_item)
-			last_size = found_key.offset;
-		else
-			last_size = new_size;
-		if (del_item) {
-			if (!pending_del_nr) {
-				/* no pending yet, add ourselves */
-				pending_del_slot = path->slots[0];
-				pending_del_nr = 1;
-			} else if (pending_del_nr &&
-				   path->slots[0] + 1 == pending_del_slot) {
-				/* hop on the pending chunk */
-				pending_del_nr++;
-				pending_del_slot = path->slots[0];
-			} else {
-				BUG();
-			}
-		} else {
-			break;
-		}
-		should_throttle = false;
-
-		if (found_extent &&
-		    (test_bit(BTRFS_ROOT_REF_COWS, &root->state) ||
-		     root == fs_info->tree_root)) {
-			struct btrfs_ref ref = { 0 };
-
-			bytes_deleted += extent_num_bytes;
-
-			btrfs_init_generic_ref(&ref, BTRFS_DROP_DELAYED_REF,
-					extent_start, extent_num_bytes, 0);
-			ref.real_root = root->root_key.objectid;
-			btrfs_init_data_ref(&ref, btrfs_header_owner(leaf),
-					ino, extent_offset);
-			ret = btrfs_free_extent(trans, &ref);
-			if (ret) {
-				btrfs_abort_transaction(trans, ret);
-				break;
-			}
-			if (be_nice) {
-				if (btrfs_should_throttle_delayed_refs(trans))
-					should_throttle = true;
-			}
-		}
-
-		if (found_type == BTRFS_INODE_ITEM_KEY)
-			break;
-
-		if (path->slots[0] == 0 ||
-		    path->slots[0] != pending_del_slot ||
-		    should_throttle) {
-			if (pending_del_nr) {
-				ret = btrfs_del_items(trans, root, path,
-						pending_del_slot,
-						pending_del_nr);
-				if (ret) {
-					btrfs_abort_transaction(trans, ret);
-					break;
-				}
-				pending_del_nr = 0;
-			}
-			btrfs_release_path(path);
-
-			/*
-			 * We can generate a lot of delayed refs, so we need to
-			 * throttle every once and a while and make sure we're
-			 * adding enough space to keep up with the work we are
-			 * generating.  Since we hold a transaction here we
-			 * can't flush, and we don't want to FLUSH_LIMIT because
-			 * we could have generated too many delayed refs to
-			 * actually allocate, so just bail if we're short and
-			 * let the normal reservation dance happen higher up.
-			 */
-			if (should_throttle) {
-				ret = btrfs_delayed_refs_rsv_refill(fs_info,
-							BTRFS_RESERVE_NO_FLUSH);
-				if (ret) {
-					ret = -EAGAIN;
-					break;
-				}
-			}
-			goto search_again;
-		} else {
-			path->slots[0]--;
-		}
-	}
-out:
-	if (ret >= 0 && pending_del_nr) {
-		int err;
-
-		err = btrfs_del_items(trans, root, path, pending_del_slot,
-				      pending_del_nr);
-		if (err) {
-			btrfs_abort_transaction(trans, err);
-			ret = err;
-		}
-	}
-	if (root->root_key.objectid != BTRFS_TREE_LOG_OBJECTID) {
-		ASSERT(last_size >= new_size);
-		if (!ret && last_size > new_size)
-			last_size = new_size;
-		btrfs_ordered_update_i_size(inode, last_size, NULL);
-		unlock_extent_cached(&BTRFS_I(inode)->io_tree, lock_start,
-				     (u64)-1, &cached_state);
-	}
-
-	btrfs_free_path(path);
-	return ret;
-}
-
-/*
  * btrfs_truncate_block - read, zero a chunk and write a block
  * @inode - inode that we're zeroing
  * @from - the offset to start zeroing
@@ -4388,21 +4857,22 @@ out:
  * This will find the block for the "from" offset and cow the block and zero the
  * part we want to zero.  This is used with truncate and hole punching.
  */
-int btrfs_truncate_block(struct inode *inode, loff_t from, loff_t len,
-			int front)
+int btrfs_truncate_block(struct btrfs_inode *inode, loff_t from, loff_t len,
+			 int front)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	struct address_space *mapping = inode->i_mapping;
-	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
+	struct address_space *mapping = inode->vfs_inode.i_mapping;
+	struct extent_io_tree *io_tree = &inode->io_tree;
 	struct btrfs_ordered_extent *ordered;
 	struct extent_state *cached_state = NULL;
 	struct extent_changeset *data_reserved = NULL;
-	char *kaddr;
+	bool only_release_metadata = false;
 	u32 blocksize = fs_info->sectorsize;
 	pgoff_t index = from >> PAGE_SHIFT;
 	unsigned offset = from & (blocksize - 1);
 	struct page *page;
 	gfp_t mask = btrfs_alloc_write_mask(mapping);
+	size_t write_bytes = blocksize;
 	int ret = 0;
 	u64 block_start;
 	u64 block_end;
@@ -4414,23 +4884,38 @@ int btrfs_truncate_block(struct inode *inode, loff_t from, loff_t len,
 	block_start = round_down(from, blocksize);
 	block_end = block_start + blocksize - 1;
 
-	ret = btrfs_delalloc_reserve_space(inode, &data_reserved,
-					   block_start, blocksize);
-	if (ret)
+	ret = btrfs_check_data_free_space(inode, &data_reserved, block_start,
+					  blocksize, false);
+	if (ret < 0) {
+		if (btrfs_check_nocow_lock(inode, block_start, &write_bytes, false) > 0) {
+			/* For nocow case, no need to reserve data space */
+			only_release_metadata = true;
+		} else {
+			goto out;
+		}
+	}
+	ret = btrfs_delalloc_reserve_metadata(inode, blocksize, blocksize, false);
+	if (ret < 0) {
+		if (!only_release_metadata)
+			btrfs_free_reserved_data_space(inode, data_reserved,
+						       block_start, blocksize);
 		goto out;
-
+	}
 again:
 	page = find_or_create_page(mapping, index, mask);
 	if (!page) {
-		btrfs_delalloc_release_space(inode, data_reserved,
-					     block_start, blocksize, true);
-		btrfs_delalloc_release_extents(BTRFS_I(inode), blocksize);
+		btrfs_delalloc_release_space(inode, data_reserved, block_start,
+					     blocksize, true);
+		btrfs_delalloc_release_extents(inode, blocksize);
 		ret = -ENOMEM;
 		goto out;
 	}
+	ret = set_page_extent_mapped(page);
+	if (ret < 0)
+		goto out_unlock;
 
 	if (!PageUptodate(page)) {
-		ret = btrfs_readpage(NULL, page);
+		ret = btrfs_read_folio(NULL, page_folio(page));
 		lock_page(page);
 		if (page->mapping != mapping) {
 			unlock_page(page);
@@ -4444,78 +4929,82 @@ again:
 	}
 	wait_on_page_writeback(page);
 
-	lock_extent_bits(io_tree, block_start, block_end, &cached_state);
-	set_page_extent_mapped(page);
+	lock_extent(io_tree, block_start, block_end, &cached_state);
 
 	ordered = btrfs_lookup_ordered_extent(inode, block_start);
 	if (ordered) {
-		unlock_extent_cached(io_tree, block_start, block_end,
-				     &cached_state);
+		unlock_extent(io_tree, block_start, block_end, &cached_state);
 		unlock_page(page);
 		put_page(page);
-		btrfs_start_ordered_extent(inode, ordered, 1);
+		btrfs_start_ordered_extent(ordered, 1);
 		btrfs_put_ordered_extent(ordered);
 		goto again;
 	}
 
-	clear_extent_bit(&BTRFS_I(inode)->io_tree, block_start, block_end,
+	clear_extent_bit(&inode->io_tree, block_start, block_end,
 			 EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG,
-			 0, 0, &cached_state);
+			 &cached_state);
 
 	ret = btrfs_set_extent_delalloc(inode, block_start, block_end, 0,
 					&cached_state);
 	if (ret) {
-		unlock_extent_cached(io_tree, block_start, block_end,
-				     &cached_state);
+		unlock_extent(io_tree, block_start, block_end, &cached_state);
 		goto out_unlock;
 	}
 
 	if (offset != blocksize) {
 		if (!len)
 			len = blocksize - offset;
-		kaddr = kmap(page);
 		if (front)
-			memset(kaddr + (block_start - page_offset(page)),
-				0, offset);
+			memzero_page(page, (block_start - page_offset(page)),
+				     offset);
 		else
-			memset(kaddr + (block_start - page_offset(page)) +  offset,
-				0, len);
-		flush_dcache_page(page);
-		kunmap(page);
+			memzero_page(page, (block_start - page_offset(page)) + offset,
+				     len);
 	}
-	ClearPageChecked(page);
-	set_page_dirty(page);
-	unlock_extent_cached(io_tree, block_start, block_end, &cached_state);
+	btrfs_page_clear_checked(fs_info, page, block_start,
+				 block_end + 1 - block_start);
+	btrfs_page_set_dirty(fs_info, page, block_start, block_end + 1 - block_start);
+	unlock_extent(io_tree, block_start, block_end, &cached_state);
+
+	if (only_release_metadata)
+		set_extent_bit(&inode->io_tree, block_start, block_end,
+			       EXTENT_NORESERVE, NULL, GFP_NOFS);
 
 out_unlock:
-	if (ret)
-		btrfs_delalloc_release_space(inode, data_reserved, block_start,
-					     blocksize, true);
-	btrfs_delalloc_release_extents(BTRFS_I(inode), blocksize);
+	if (ret) {
+		if (only_release_metadata)
+			btrfs_delalloc_release_metadata(inode, blocksize, true);
+		else
+			btrfs_delalloc_release_space(inode, data_reserved,
+					block_start, blocksize, true);
+	}
+	btrfs_delalloc_release_extents(inode, blocksize);
 	unlock_page(page);
 	put_page(page);
 out:
+	if (only_release_metadata)
+		btrfs_check_nocow_unlock(inode);
 	extent_changeset_free(data_reserved);
 	return ret;
 }
 
-static int maybe_insert_hole(struct btrfs_root *root, struct inode *inode,
+static int maybe_insert_hole(struct btrfs_root *root, struct btrfs_inode *inode,
 			     u64 offset, u64 len)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct btrfs_trans_handle *trans;
+	struct btrfs_drop_extents_args drop_args = { 0 };
 	int ret;
 
 	/*
-	 * Still need to make sure the inode looks like it's been updated so
-	 * that any holes get logged if we fsync.
+	 * If NO_HOLES is enabled, we don't need to do anything.
+	 * Later, up in the call chain, either btrfs_set_inode_last_sub_trans()
+	 * or btrfs_update_inode() will be called, which guarantee that the next
+	 * fsync will know this inode was changed and needs to be logged.
 	 */
-	if (btrfs_fs_incompat(fs_info, NO_HOLES)) {
-		BTRFS_I(inode)->last_trans = fs_info->generation;
-		BTRFS_I(inode)->last_sub_trans = root->log_transid;
-		BTRFS_I(inode)->last_log_commit = root->last_log_commit;
+	if (btrfs_fs_incompat(fs_info, NO_HOLES))
 		return 0;
-	}
 
 	/*
 	 * 1 - for the one we're dropping
@@ -4526,19 +5015,24 @@ static int maybe_insert_hole(struct btrfs_root *root, struct inode *inode,
 	if (IS_ERR(trans))
 		return PTR_ERR(trans);
 
-	ret = btrfs_drop_extents(trans, root, inode, offset, offset + len, 1);
+	drop_args.start = offset;
+	drop_args.end = offset + len;
+	drop_args.drop_cache = true;
+
+	ret = btrfs_drop_extents(trans, root, inode, &drop_args);
 	if (ret) {
 		btrfs_abort_transaction(trans, ret);
 		btrfs_end_transaction(trans);
 		return ret;
 	}
 
-	ret = btrfs_insert_file_extent(trans, root, btrfs_ino(BTRFS_I(inode)),
-			offset, 0, 0, len, 0, len, 0, 0, 0);
-	if (ret)
+	ret = btrfs_insert_hole_extent(trans, root, btrfs_ino(inode), offset, len);
+	if (ret) {
 		btrfs_abort_transaction(trans, ret);
-	else
+	} else {
+		btrfs_update_inode_bytes(inode, 0, drop_args.bytes_found);
 		btrfs_update_inode(trans, root, inode);
+	}
 	btrfs_end_transaction(trans);
 	return ret;
 }
@@ -4549,14 +5043,13 @@ static int maybe_insert_hole(struct btrfs_root *root, struct inode *inode,
  * these file extents so that btrfs_get_extent will return a EXTENT_MAP_HOLE for
  * the range between oldsize and size
  */
-int btrfs_cont_expand(struct inode *inode, loff_t oldsize, loff_t size)
+int btrfs_cont_expand(struct btrfs_inode *inode, loff_t oldsize, loff_t size)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	struct btrfs_root *root = BTRFS_I(inode)->root;
-	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
+	struct btrfs_root *root = inode->root;
+	struct btrfs_fs_info *fs_info = root->fs_info;
+	struct extent_io_tree *io_tree = &inode->io_tree;
 	struct extent_map *em = NULL;
 	struct extent_state *cached_state = NULL;
-	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
 	u64 hole_start = ALIGN(oldsize, fs_info->sectorsize);
 	u64 block_end = ALIGN(size, fs_info->sectorsize);
 	u64 last_byte;
@@ -4576,11 +5069,11 @@ int btrfs_cont_expand(struct inode *inode, loff_t oldsize, loff_t size)
 	if (size <= hole_start)
 		return 0;
 
-	btrfs_lock_and_flush_ordered_range(io_tree, BTRFS_I(inode), hole_start,
-					   block_end - 1, &cached_state);
+	btrfs_lock_and_flush_ordered_range(inode, hole_start, block_end - 1,
+					   &cached_state);
 	cur_offset = hole_start;
 	while (1) {
-		em = btrfs_get_extent(BTRFS_I(inode), NULL, 0, cur_offset,
+		em = btrfs_get_extent(inode, NULL, 0, cur_offset,
 				      block_end - cur_offset);
 		if (IS_ERR(em)) {
 			err = PTR_ERR(em);
@@ -4589,20 +5082,27 @@ int btrfs_cont_expand(struct inode *inode, loff_t oldsize, loff_t size)
 		}
 		last_byte = min(extent_map_end(em), block_end);
 		last_byte = ALIGN(last_byte, fs_info->sectorsize);
+		hole_size = last_byte - cur_offset;
+
 		if (!test_bit(EXTENT_FLAG_PREALLOC, &em->flags)) {
 			struct extent_map *hole_em;
-			hole_size = last_byte - cur_offset;
 
 			err = maybe_insert_hole(root, inode, cur_offset,
 						hole_size);
 			if (err)
 				break;
-			btrfs_drop_extent_cache(BTRFS_I(inode), cur_offset,
-						cur_offset + hole_size - 1, 0);
+
+			err = btrfs_inode_set_file_extent_range(inode,
+							cur_offset, hole_size);
+			if (err)
+				break;
+
 			hole_em = alloc_extent_map();
 			if (!hole_em) {
-				set_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
-					&BTRFS_I(inode)->runtime_flags);
+				btrfs_drop_extent_map_range(inode, cur_offset,
+						    cur_offset + hole_size - 1,
+						    false);
+				btrfs_set_inode_full_sync(inode);
 				goto next;
 			}
 			hole_em->start = cur_offset;
@@ -4616,18 +5116,13 @@ int btrfs_cont_expand(struct inode *inode, loff_t oldsize, loff_t size)
 			hole_em->compress_type = BTRFS_COMPRESS_NONE;
 			hole_em->generation = fs_info->generation;
 
-			while (1) {
-				write_lock(&em_tree->lock);
-				err = add_extent_mapping(em_tree, hole_em, 1);
-				write_unlock(&em_tree->lock);
-				if (err != -EEXIST)
-					break;
-				btrfs_drop_extent_cache(BTRFS_I(inode),
-							cur_offset,
-							cur_offset +
-							hole_size - 1, 0);
-			}
+			err = btrfs_replace_extent_map_range(inode, hole_em, true);
 			free_extent_map(hole_em);
+		} else {
+			err = btrfs_inode_set_file_extent_range(inode,
+							cur_offset, hole_size);
+			if (err)
+				break;
 		}
 next:
 		free_extent_map(em);
@@ -4637,7 +5132,7 @@ next:
 			break;
 	}
 	free_extent_map(em);
-	unlock_extent_cached(io_tree, hole_start, block_end - 1, &cached_state);
+	unlock_extent(io_tree, hole_start, block_end - 1, &cached_state);
 	return err;
 }
 
@@ -4658,9 +5153,10 @@ static int btrfs_setsize(struct inode *inode, struct iattr *attr)
 	 */
 	if (newsize != oldsize) {
 		inode_inc_iversion(inode);
-		if (!(mask & (ATTR_CTIME | ATTR_MTIME)))
-			inode->i_ctime = inode->i_mtime =
-				current_time(inode);
+		if (!(mask & (ATTR_CTIME | ATTR_MTIME))) {
+			inode->i_mtime = current_time(inode);
+			inode->i_ctime = inode->i_mtime;
+		}
 	}
 
 	if (newsize > oldsize) {
@@ -4671,42 +5167,48 @@ static int btrfs_setsize(struct inode *inode, struct iattr *attr)
 		 * truncation, it must capture all writes that happened before
 		 * this truncation.
 		 */
-		btrfs_wait_for_snapshot_creation(root);
-		ret = btrfs_cont_expand(inode, oldsize, newsize);
+		btrfs_drew_write_lock(&root->snapshot_lock);
+		ret = btrfs_cont_expand(BTRFS_I(inode), oldsize, newsize);
 		if (ret) {
-			btrfs_end_write_no_snapshotting(root);
+			btrfs_drew_write_unlock(&root->snapshot_lock);
 			return ret;
 		}
 
 		trans = btrfs_start_transaction(root, 1);
 		if (IS_ERR(trans)) {
-			btrfs_end_write_no_snapshotting(root);
+			btrfs_drew_write_unlock(&root->snapshot_lock);
 			return PTR_ERR(trans);
 		}
 
 		i_size_write(inode, newsize);
-		btrfs_ordered_update_i_size(inode, i_size_read(inode), NULL);
+		btrfs_inode_safe_disk_i_size_write(BTRFS_I(inode), 0);
 		pagecache_isize_extended(inode, oldsize, newsize);
-		ret = btrfs_update_inode(trans, root, inode);
-		btrfs_end_write_no_snapshotting(root);
+		ret = btrfs_update_inode(trans, root, BTRFS_I(inode));
+		btrfs_drew_write_unlock(&root->snapshot_lock);
 		btrfs_end_transaction(trans);
 	} else {
+		struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+
+		if (btrfs_is_zoned(fs_info)) {
+			ret = btrfs_wait_ordered_range(inode,
+					ALIGN(newsize, fs_info->sectorsize),
+					(u64)-1);
+			if (ret)
+				return ret;
+		}
 
 		/*
 		 * We're truncating a file that used to have good data down to
-		 * zero. Make sure it gets into the ordered flush list so that
-		 * any new writes get down to disk quickly.
+		 * zero. Make sure any new writes to the file get on disk
+		 * on close.
 		 */
 		if (newsize == 0)
-			set_bit(BTRFS_INODE_ORDERED_DATA_CLOSE,
+			set_bit(BTRFS_INODE_FLUSH_ON_CLOSE,
 				&BTRFS_I(inode)->runtime_flags);
 
 		truncate_setsize(inode, newsize);
 
-		/* Disable nonlocked read DIO to avoid the endless truncate */
-		btrfs_inode_block_unlocked_dio(BTRFS_I(inode));
 		inode_dio_wait(inode);
-		btrfs_inode_resume_unlocked_dio(BTRFS_I(inode));
 
 		ret = btrfs_truncate(inode, newsize == oldsize);
 		if (ret && inode->i_nlink) {
@@ -4728,7 +5230,8 @@ static int btrfs_setsize(struct inode *inode, struct iattr *attr)
 	return ret;
 }
 
-static int btrfs_setattr(struct dentry *dentry, struct iattr *attr)
+static int btrfs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry,
+			 struct iattr *attr)
 {
 	struct inode *inode = d_inode(dentry);
 	struct btrfs_root *root = BTRFS_I(inode)->root;
@@ -4737,7 +5240,7 @@ static int btrfs_setattr(struct dentry *dentry, struct iattr *attr)
 	if (btrfs_root_readonly(root))
 		return -EROFS;
 
-	err = setattr_prepare(dentry, attr);
+	err = setattr_prepare(mnt_userns, dentry, attr);
 	if (err)
 		return err;
 
@@ -4748,60 +5251,44 @@ static int btrfs_setattr(struct dentry *dentry, struct iattr *attr)
 	}
 
 	if (attr->ia_valid) {
-		setattr_copy(inode, attr);
+		setattr_copy(mnt_userns, inode, attr);
 		inode_inc_iversion(inode);
 		err = btrfs_dirty_inode(inode);
 
 		if (!err && attr->ia_valid & ATTR_MODE)
-			err = posix_acl_chmod(inode, inode->i_mode);
+			err = posix_acl_chmod(mnt_userns, inode, inode->i_mode);
 	}
 
 	return err;
 }
 
 /*
- * While truncating the inode pages during eviction, we get the VFS calling
- * btrfs_invalidatepage() against each page of the inode. This is slow because
- * the calls to btrfs_invalidatepage() result in a huge amount of calls to
- * lock_extent_bits() and clear_extent_bit(), which keep merging and splitting
- * extent_state structures over and over, wasting lots of time.
+ * While truncating the inode pages during eviction, we get the VFS
+ * calling btrfs_invalidate_folio() against each folio of the inode. This
+ * is slow because the calls to btrfs_invalidate_folio() result in a
+ * huge amount of calls to lock_extent() and clear_extent_bit(),
+ * which keep merging and splitting extent_state structures over and over,
+ * wasting lots of time.
  *
- * Therefore if the inode is being evicted, let btrfs_invalidatepage() skip all
- * those expensive operations on a per page basis and do only the ordered io
- * finishing, while we release here the extent_map and extent_state structures,
- * without the excessive merging and splitting.
+ * Therefore if the inode is being evicted, let btrfs_invalidate_folio()
+ * skip all those expensive operations on a per folio basis and do only
+ * the ordered io finishing, while we release here the extent_map and
+ * extent_state structures, without the excessive merging and splitting.
  */
 static void evict_inode_truncate_pages(struct inode *inode)
 {
 	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
-	struct extent_map_tree *map_tree = &BTRFS_I(inode)->extent_tree;
 	struct rb_node *node;
 
 	ASSERT(inode->i_state & I_FREEING);
 	truncate_inode_pages_final(&inode->i_data);
 
-	write_lock(&map_tree->lock);
-	while (!RB_EMPTY_ROOT(&map_tree->map.rb_root)) {
-		struct extent_map *em;
-
-		node = rb_first_cached(&map_tree->map);
-		em = rb_entry(node, struct extent_map, rb_node);
-		clear_bit(EXTENT_FLAG_PINNED, &em->flags);
-		clear_bit(EXTENT_FLAG_LOGGING, &em->flags);
-		remove_extent_mapping(map_tree, em);
-		free_extent_map(em);
-		if (need_resched()) {
-			write_unlock(&map_tree->lock);
-			cond_resched();
-			write_lock(&map_tree->lock);
-		}
-	}
-	write_unlock(&map_tree->lock);
+	btrfs_drop_extent_map_range(BTRFS_I(inode), 0, (u64)-1, false);
 
 	/*
 	 * Keep looping until we have no more ranges in the io tree.
-	 * We can have ongoing bios started by readpages (called from readahead)
-	 * that have their endio callback (extent_io.c:end_bio_extent_readpage)
+	 * We can have ongoing bios started by readahead that have
+	 * their endio callback (extent_io.c:end_bio_extent_readpage)
 	 * still in progress (unlocked the pages in the bio but did not yet
 	 * unlocked the ranges in the io tree). Therefore this means some
 	 * ranges can still be locked and eviction started because before
@@ -4829,22 +5316,22 @@ static void evict_inode_truncate_pages(struct inode *inode)
 		state_flags = state->state;
 		spin_unlock(&io_tree->lock);
 
-		lock_extent_bits(io_tree, start, end, &cached_state);
+		lock_extent(io_tree, start, end, &cached_state);
 
 		/*
 		 * If still has DELALLOC flag, the extent didn't reach disk,
 		 * and its reserved space won't be freed by delayed_ref.
 		 * So we need to free its reserved space here.
-		 * (Refer to comment in btrfs_invalidatepage, case 2)
+		 * (Refer to comment in btrfs_invalidate_folio, case 2)
 		 *
 		 * Note, end is the bytenr of last byte, so we need + 1 here.
 		 */
 		if (state_flags & EXTENT_DELALLOC)
-			btrfs_qgroup_free_data(inode, NULL, start, end - start + 1);
+			btrfs_qgroup_free_data(BTRFS_I(inode), NULL, start,
+					       end - start + 1);
 
 		clear_extent_bit(io_tree, start, end,
-				 EXTENT_LOCKED | EXTENT_DELALLOC |
-				 EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG, 1, 1,
+				 EXTENT_CLEAR_ALL_BITS | EXTENT_DO_ACCOUNTING,
 				 &cached_state);
 
 		cond_resched();
@@ -4857,7 +5344,6 @@ static struct btrfs_trans_handle *evict_refill_and_join(struct btrfs_root *root,
 							struct btrfs_block_rsv *rsv)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
-	struct btrfs_block_rsv *global_rsv = &fs_info->global_block_rsv;
 	struct btrfs_trans_handle *trans;
 	u64 delayed_refs_extra = btrfs_calc_insert_metadata_size(fs_info, 1);
 	int ret;
@@ -4872,18 +5358,16 @@ static struct btrfs_trans_handle *evict_refill_and_join(struct btrfs_root *root,
 	 * above.  We reserve our extra bit here because we generate a ton of
 	 * delayed refs activity by truncating.
 	 *
-	 * If we cannot make our reservation we'll attempt to steal from the
-	 * global reserve, because we really want to be able to free up space.
+	 * BTRFS_RESERVE_FLUSH_EVICT will steal from the global_rsv if it can,
+	 * if we fail to make this reservation we can re-try without the
+	 * delayed_refs_extra so we can make some forward progress.
 	 */
-	ret = btrfs_block_rsv_refill(root, rsv, rsv->size + delayed_refs_extra,
+	ret = btrfs_block_rsv_refill(fs_info, rsv, rsv->size + delayed_refs_extra,
 				     BTRFS_RESERVE_FLUSH_EVICT);
 	if (ret) {
-		/*
-		 * Try to steal from the global reserve if there is space for
-		 * it.
-		 */
-		if (btrfs_check_space_for_delayed_refs(fs_info) ||
-		    btrfs_block_rsv_migrate(global_rsv, rsv, rsv->size, 0)) {
+		ret = btrfs_block_rsv_refill(fs_info, rsv, rsv->size,
+					     BTRFS_RESERVE_FLUSH_EVICT);
+		if (ret) {
 			btrfs_warn(fs_info,
 				   "could not allocate space for delete; will truncate on mount");
 			return ERR_PTR(-ENOSPC);
@@ -4915,6 +5399,7 @@ void btrfs_evict_inode(struct inode *inode)
 	trace_btrfs_inode_evict(inode);
 
 	if (!root) {
+		fsverity_cleanup_inode(inode);
 		clear_inode(inode);
 		return;
 	}
@@ -4941,26 +5426,45 @@ void btrfs_evict_inode(struct inode *inode)
 		goto no_delete;
 	}
 
+	/*
+	 * This makes sure the inode item in tree is uptodate and the space for
+	 * the inode update is released.
+	 */
 	ret = btrfs_commit_inode_delayed_inode(BTRFS_I(inode));
 	if (ret)
 		goto no_delete;
 
+	/*
+	 * This drops any pending insert or delete operations we have for this
+	 * inode.  We could have a delayed dir index deletion queued up, but
+	 * we're removing the inode completely so that'll be taken care of in
+	 * the truncate.
+	 */
+	btrfs_kill_delayed_inode_items(BTRFS_I(inode));
+
 	rsv = btrfs_alloc_block_rsv(fs_info, BTRFS_BLOCK_RSV_TEMP);
 	if (!rsv)
 		goto no_delete;
 	rsv->size = btrfs_calc_metadata_size(fs_info, 1);
-	rsv->failfast = 1;
+	rsv->failfast = true;
 
 	btrfs_i_size_write(BTRFS_I(inode), 0);
 
 	while (1) {
+		struct btrfs_truncate_control control = {
+			.inode = BTRFS_I(inode),
+			.ino = btrfs_ino(BTRFS_I(inode)),
+			.new_size = 0,
+			.min_type = 0,
+		};
+
 		trans = evict_refill_and_join(root, rsv);
 		if (IS_ERR(trans))
 			goto free_rsv;
 
 		trans->block_rsv = rsv;
 
-		ret = btrfs_truncate_inode_items(trans, root, inode, 0, 0);
+		ret = btrfs_truncate_inode_items(trans, root, &control);
 		trans->block_rsv = &fs_info->trans_block_rsv;
 		btrfs_end_transaction(trans);
 		btrfs_btree_balance_dirty(fs_info);
@@ -4987,10 +5491,6 @@ void btrfs_evict_inode(struct inode *inode)
 		btrfs_end_transaction(trans);
 	}
 
-	if (!(root == fs_info->tree_root ||
-	      root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID))
-		btrfs_return_ino(root, btrfs_ino(BTRFS_I(inode)));
-
 free_rsv:
 	btrfs_free_block_rsv(fs_info, rsv);
 no_delete:
@@ -5000,6 +5500,7 @@ no_delete:
 	 * to retry these periodically in the future.
 	 */
 	btrfs_remove_delayed_node(BTRFS_I(inode));
+	fsverity_cleanup_inode(inode);
 	clear_inode(inode);
 }
 
@@ -5098,7 +5599,7 @@ static int fixup_tree_root_location(struct btrfs_fs_info *fs_info,
 
 	btrfs_release_path(path);
 
-	new_root = btrfs_read_fs_root_no_name(fs_info, location);
+	new_root = btrfs_get_fs_root(fs_info, location->objectid, true);
 	if (IS_ERR(new_root)) {
 		err = PTR_ERR(new_root);
 		goto out;
@@ -5150,15 +5651,15 @@ static void inode_tree_add(struct inode *inode)
 	spin_unlock(&root->inode_lock);
 }
 
-static void inode_tree_del(struct inode *inode)
+static void inode_tree_del(struct btrfs_inode *inode)
 {
-	struct btrfs_root *root = BTRFS_I(inode)->root;
+	struct btrfs_root *root = inode->root;
 	int empty = 0;
 
 	spin_lock(&root->inode_lock);
-	if (!RB_EMPTY_NODE(&BTRFS_I(inode)->rb_node)) {
-		rb_erase(&BTRFS_I(inode)->rb_node, &root->inode_tree);
-		RB_CLEAR_NODE(&BTRFS_I(inode)->rb_node);
+	if (!RB_EMPTY_NODE(&inode->rb_node)) {
+		rb_erase(&inode->rb_node, &root->inode_tree);
+		RB_CLEAR_NODE(&inode->rb_node);
 		empty = RB_EMPTY_ROOT(&root->inode_tree);
 	}
 	spin_unlock(&root->inode_lock);
@@ -5176,29 +5677,37 @@ static void inode_tree_del(struct inode *inode)
 static int btrfs_init_locked_inode(struct inode *inode, void *p)
 {
 	struct btrfs_iget_args *args = p;
-	inode->i_ino = args->location->objectid;
-	memcpy(&BTRFS_I(inode)->location, args->location,
-	       sizeof(*args->location));
-	BTRFS_I(inode)->root = args->root;
+
+	inode->i_ino = args->ino;
+	BTRFS_I(inode)->location.objectid = args->ino;
+	BTRFS_I(inode)->location.type = BTRFS_INODE_ITEM_KEY;
+	BTRFS_I(inode)->location.offset = 0;
+	BTRFS_I(inode)->root = btrfs_grab_root(args->root);
+	BUG_ON(args->root && !BTRFS_I(inode)->root);
+
+	if (args->root && args->root == args->root->fs_info->tree_root &&
+	    args->ino != BTRFS_BTREE_INODE_OBJECTID)
+		set_bit(BTRFS_INODE_FREE_SPACE_INODE,
+			&BTRFS_I(inode)->runtime_flags);
 	return 0;
 }
 
 static int btrfs_find_actor(struct inode *inode, void *opaque)
 {
 	struct btrfs_iget_args *args = opaque;
-	return args->location->objectid == BTRFS_I(inode)->location.objectid &&
+
+	return args->ino == BTRFS_I(inode)->location.objectid &&
 		args->root == BTRFS_I(inode)->root;
 }
 
-static struct inode *btrfs_iget_locked(struct super_block *s,
-				       struct btrfs_key *location,
+static struct inode *btrfs_iget_locked(struct super_block *s, u64 ino,
 				       struct btrfs_root *root)
 {
 	struct inode *inode;
 	struct btrfs_iget_args args;
-	unsigned long hashval = btrfs_inode_hash(location->objectid, root);
+	unsigned long hashval = btrfs_inode_hash(ino, root);
 
-	args.location = location;
+	args.ino = ino;
 	args.root = root;
 
 	inode = iget5_locked(s, hashval, btrfs_find_actor,
@@ -5208,17 +5717,17 @@ static struct inode *btrfs_iget_locked(struct super_block *s,
 }
 
 /*
- * Get an inode object given its location and corresponding root.
+ * Get an inode object given its inode number and corresponding root.
  * Path can be preallocated to prevent recursing back to iget through
  * allocator. NULL is also valid but may require an additional allocation
  * later.
  */
-struct inode *btrfs_iget_path(struct super_block *s, struct btrfs_key *location,
+struct inode *btrfs_iget_path(struct super_block *s, u64 ino,
 			      struct btrfs_root *root, struct btrfs_path *path)
 {
 	struct inode *inode;
 
-	inode = btrfs_iget_locked(s, location, root);
+	inode = btrfs_iget_locked(s, ino, root);
 	if (!inode)
 		return ERR_PTR(-ENOMEM);
 
@@ -5245,10 +5754,9 @@ struct inode *btrfs_iget_path(struct super_block *s, struct btrfs_key *location,
 	return inode;
 }
 
-struct inode *btrfs_iget(struct super_block *s, struct btrfs_key *location,
-			 struct btrfs_root *root)
+struct inode *btrfs_iget(struct super_block *s, u64 ino, struct btrfs_root *root)
 {
-	return btrfs_iget_path(s, location, root, NULL);
+	return btrfs_iget_path(s, ino, root, NULL);
 }
 
 static struct inode *new_simple_dir(struct super_block *s,
@@ -5260,7 +5768,7 @@ static struct inode *new_simple_dir(struct super_block *s,
 	if (!inode)
 		return ERR_PTR(-ENOMEM);
 
-	BTRFS_I(inode)->root = root;
+	BTRFS_I(inode)->root = btrfs_grab_root(root);
 	memcpy(&BTRFS_I(inode)->location, key, sizeof(*key));
 	set_bit(BTRFS_INODE_DUMMY, &BTRFS_I(inode)->runtime_flags);
 
@@ -5281,21 +5789,17 @@ static struct inode *new_simple_dir(struct super_block *s,
 	return inode;
 }
 
+static_assert(BTRFS_FT_UNKNOWN == FT_UNKNOWN);
+static_assert(BTRFS_FT_REG_FILE == FT_REG_FILE);
+static_assert(BTRFS_FT_DIR == FT_DIR);
+static_assert(BTRFS_FT_CHRDEV == FT_CHRDEV);
+static_assert(BTRFS_FT_BLKDEV == FT_BLKDEV);
+static_assert(BTRFS_FT_FIFO == FT_FIFO);
+static_assert(BTRFS_FT_SOCK == FT_SOCK);
+static_assert(BTRFS_FT_SYMLINK == FT_SYMLINK);
+
 static inline u8 btrfs_inode_type(struct inode *inode)
 {
-	/*
-	 * Compile-time asserts that generic FT_* types still match
-	 * BTRFS_FT_* types
-	 */
-	BUILD_BUG_ON(BTRFS_FT_UNKNOWN != FT_UNKNOWN);
-	BUILD_BUG_ON(BTRFS_FT_REG_FILE != FT_REG_FILE);
-	BUILD_BUG_ON(BTRFS_FT_DIR != FT_DIR);
-	BUILD_BUG_ON(BTRFS_FT_CHRDEV != FT_CHRDEV);
-	BUILD_BUG_ON(BTRFS_FT_BLKDEV != FT_BLKDEV);
-	BUILD_BUG_ON(BTRFS_FT_FIFO != FT_FIFO);
-	BUILD_BUG_ON(BTRFS_FT_SOCK != FT_SOCK);
-	BUILD_BUG_ON(BTRFS_FT_SYMLINK != FT_SYMLINK);
-
 	return fs_umode_to_ftype(inode->i_mode);
 }
 
@@ -5307,7 +5811,6 @@ struct inode *btrfs_lookup_dentry(struct inode *dir, struct dentry *dentry)
 	struct btrfs_root *sub_root = root;
 	struct btrfs_key location;
 	u8 di_type = 0;
-	int index;
 	int ret = 0;
 
 	if (dentry->d_name.len > BTRFS_NAME_LEN)
@@ -5318,7 +5821,7 @@ struct inode *btrfs_lookup_dentry(struct inode *dir, struct dentry *dentry)
 		return ERR_PTR(ret);
 
 	if (location.type == BTRFS_INODE_ITEM_KEY) {
-		inode = btrfs_iget(dir->i_sb, &location, root);
+		inode = btrfs_iget(dir->i_sb, location.objectid, root);
 		if (IS_ERR(inode))
 			return inode;
 
@@ -5334,20 +5837,20 @@ struct inode *btrfs_lookup_dentry(struct inode *dir, struct dentry *dentry)
 		return inode;
 	}
 
-	index = srcu_read_lock(&fs_info->subvol_srcu);
 	ret = fixup_tree_root_location(fs_info, dir, dentry,
 				       &location, &sub_root);
 	if (ret < 0) {
 		if (ret != -ENOENT)
 			inode = ERR_PTR(ret);
 		else
-			inode = new_simple_dir(dir->i_sb, &location, sub_root);
+			inode = new_simple_dir(dir->i_sb, &location, root);
 	} else {
-		inode = btrfs_iget(dir->i_sb, &location, sub_root);
-	}
-	srcu_read_unlock(&fs_info->subvol_srcu, index);
+		inode = btrfs_iget(dir->i_sb, location.objectid, sub_root);
+		btrfs_put_root(sub_root);
+
+		if (IS_ERR(inode))
+			return inode;
 
-	if (!IS_ERR(inode) && root != sub_root) {
 		down_read(&fs_info->cleanup_work_sem);
 		if (!sb_rdonly(inode->i_sb))
 			ret = btrfs_orphan_cleanup(sub_root);
@@ -5453,8 +5956,6 @@ static int btrfs_real_readdir(struct file *file, struct dir_context *ctx)
 	struct list_head ins_list;
 	struct list_head del_list;
 	int ret;
-	struct extent_buffer *leaf;
-	int slot;
 	char *name_ptr;
 	int name_len;
 	int entries = 0;
@@ -5481,35 +5982,19 @@ again:
 	key.offset = ctx->pos;
 	key.objectid = btrfs_ino(BTRFS_I(inode));
 
-	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
-	if (ret < 0)
-		goto err;
-
-	while (1) {
+	btrfs_for_each_slot(root, &key, &found_key, path, ret) {
 		struct dir_entry *entry;
-
-		leaf = path->nodes[0];
-		slot = path->slots[0];
-		if (slot >= btrfs_header_nritems(leaf)) {
-			ret = btrfs_next_leaf(root, path);
-			if (ret < 0)
-				goto err;
-			else if (ret > 0)
-				break;
-			continue;
-		}
-
-		btrfs_item_key_to_cpu(leaf, &found_key, slot);
+		struct extent_buffer *leaf = path->nodes[0];
 
 		if (found_key.objectid != key.objectid)
 			break;
 		if (found_key.type != BTRFS_DIR_INDEX_KEY)
 			break;
 		if (found_key.offset < ctx->pos)
-			goto next;
+			continue;
 		if (btrfs_should_delete_dir_index(&del_list, found_key.offset))
-			goto next;
-		di = btrfs_item_ptr(leaf, slot, struct btrfs_dir_item);
+			continue;
+		di = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_dir_item);
 		name_len = btrfs_dir_name_len(leaf, di);
 		if ((total_len + sizeof(struct dir_entry) + name_len) >=
 		    PAGE_SIZE) {
@@ -5536,9 +6021,11 @@ again:
 		entries++;
 		addr += sizeof(struct dir_entry) + name_len;
 		total_len += sizeof(struct dir_entry) + name_len;
-next:
-		path->slots[0]++;
 	}
+	/* Catch error encountered during iteration */
+	if (ret < 0)
+		goto err;
+
 	btrfs_release_path(path);
 
 	ret = btrfs_filldir(private->filldir_buf, entries, ctx);
@@ -5599,15 +6086,15 @@ static int btrfs_dirty_inode(struct inode *inode)
 	if (IS_ERR(trans))
 		return PTR_ERR(trans);
 
-	ret = btrfs_update_inode(trans, root, inode);
-	if (ret && ret == -ENOSPC) {
+	ret = btrfs_update_inode(trans, root, BTRFS_I(inode));
+	if (ret && (ret == -ENOSPC || ret == -EDQUOT)) {
 		/* whoops, lets try again with the full transaction */
 		btrfs_end_transaction(trans);
 		trans = btrfs_start_transaction(root, 1);
 		if (IS_ERR(trans))
 			return PTR_ERR(trans);
 
-		ret = btrfs_update_inode(trans, root, inode);
+		ret = btrfs_update_inode(trans, root, BTRFS_I(inode));
 	}
 	btrfs_end_transaction(trans);
 	if (BTRFS_I(inode)->delayed_node)
@@ -5669,14 +6156,8 @@ static int btrfs_set_inode_index_count(struct btrfs_inode *inode)
 		goto out;
 	ret = 0;
 
-	/*
-	 * MAGIC NUMBER EXPLANATION:
-	 * since we search a directory based on f_pos we have to start at 2
-	 * since '.' and '..' have f_pos of 0 and 1 respectively, so everybody
-	 * else has to start at 2
-	 */
 	if (path->slots[0] == 0) {
-		inode->index_cnt = 2;
+		inode->index_cnt = BTRFS_DIR_START_INDEX;
 		goto out;
 	}
 
@@ -5687,7 +6168,7 @@ static int btrfs_set_inode_index_count(struct btrfs_inode *inode)
 
 	if (found_key.objectid != btrfs_ino(inode) ||
 	    found_key.type != BTRFS_DIR_INDEX_KEY) {
-		inode->index_cnt = 2;
+		inode->index_cnt = BTRFS_DIR_START_INDEX;
 		goto out;
 	}
 
@@ -5723,7 +6204,8 @@ int btrfs_set_inode_index(struct btrfs_inode *dir, u64 *index)
 static int btrfs_insert_inode_locked(struct inode *inode)
 {
 	struct btrfs_iget_args args;
-	args.location = &BTRFS_I(inode)->location;
+
+	args.ino = BTRFS_I(inode)->location.objectid;
 	args.root = BTRFS_I(inode)->root;
 
 	return insert_inode_locked4(inode,
@@ -5731,6 +6213,57 @@ static int btrfs_insert_inode_locked(struct inode *inode)
 		   btrfs_find_actor, &args);
 }
 
+int btrfs_new_inode_prepare(struct btrfs_new_inode_args *args,
+			    unsigned int *trans_num_items)
+{
+	struct inode *dir = args->dir;
+	struct inode *inode = args->inode;
+	int ret;
+
+	ret = posix_acl_create(dir, &inode->i_mode, &args->default_acl, &args->acl);
+	if (ret)
+		return ret;
+
+	/* 1 to add inode item */
+	*trans_num_items = 1;
+	/* 1 to add compression property */
+	if (BTRFS_I(dir)->prop_compress)
+		(*trans_num_items)++;
+	/* 1 to add default ACL xattr */
+	if (args->default_acl)
+		(*trans_num_items)++;
+	/* 1 to add access ACL xattr */
+	if (args->acl)
+		(*trans_num_items)++;
+#ifdef CONFIG_SECURITY
+	/* 1 to add LSM xattr */
+	if (dir->i_security)
+		(*trans_num_items)++;
+#endif
+	if (args->orphan) {
+		/* 1 to add orphan item */
+		(*trans_num_items)++;
+	} else {
+		/*
+		 * 1 to add dir item
+		 * 1 to add dir index
+		 * 1 to update parent inode item
+		 *
+		 * No need for 1 unit for the inode ref item because it is
+		 * inserted in a batch together with the inode item at
+		 * btrfs_create_new_inode().
+		 */
+		*trans_num_items += 3;
+	}
+	return 0;
+}
+
+void btrfs_new_inode_args_destroy(struct btrfs_new_inode_args *args)
+{
+	posix_acl_release(args->acl);
+	posix_acl_release(args->default_acl);
+}
+
 /*
  * Inherit flags from the parent inode.
  *
@@ -5740,9 +6273,6 @@ static void btrfs_inherit_iflags(struct inode *inode, struct inode *dir)
 {
 	unsigned int flags;
 
-	if (!dir)
-		return;
-
 	flags = BTRFS_I(dir)->flags;
 
 	if (flags & BTRFS_INODE_NOCOMPRESS) {
@@ -5762,81 +6292,92 @@ static void btrfs_inherit_iflags(struct inode *inode, struct inode *dir)
 	btrfs_sync_inode_flags_to_i_flags(inode);
 }
 
-static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans,
-				     struct btrfs_root *root,
-				     struct inode *dir,
-				     const char *name, int name_len,
-				     u64 ref_objectid, u64 objectid,
-				     umode_t mode, u64 *index)
+int btrfs_create_new_inode(struct btrfs_trans_handle *trans,
+			   struct btrfs_new_inode_args *args)
 {
-	struct btrfs_fs_info *fs_info = root->fs_info;
-	struct inode *inode;
+	struct inode *dir = args->dir;
+	struct inode *inode = args->inode;
+	const char *name = args->orphan ? NULL : args->dentry->d_name.name;
+	int name_len = args->orphan ? 0 : args->dentry->d_name.len;
+	struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb);
+	struct btrfs_root *root;
 	struct btrfs_inode_item *inode_item;
 	struct btrfs_key *location;
 	struct btrfs_path *path;
+	u64 objectid;
 	struct btrfs_inode_ref *ref;
 	struct btrfs_key key[2];
 	u32 sizes[2];
-	int nitems = name ? 2 : 1;
+	struct btrfs_item_batch batch;
 	unsigned long ptr;
-	unsigned int nofs_flag;
 	int ret;
 
 	path = btrfs_alloc_path();
 	if (!path)
-		return ERR_PTR(-ENOMEM);
-
-	nofs_flag = memalloc_nofs_save();
-	inode = new_inode(fs_info->sb);
-	memalloc_nofs_restore(nofs_flag);
-	if (!inode) {
-		btrfs_free_path(path);
-		return ERR_PTR(-ENOMEM);
-	}
+		return -ENOMEM;
 
-	/*
-	 * O_TMPFILE, set link count to 0, so that after this point,
-	 * we fill in an inode item with the correct link count.
-	 */
-	if (!name)
-		set_nlink(inode, 0);
+	if (!args->subvol)
+		BTRFS_I(inode)->root = btrfs_grab_root(BTRFS_I(dir)->root);
+	root = BTRFS_I(inode)->root;
 
-	/*
-	 * we have to initialize this early, so we can reclaim the inode
-	 * number if we fail afterwards in this function.
-	 */
+	ret = btrfs_get_free_objectid(root, &objectid);
+	if (ret)
+		goto out;
 	inode->i_ino = objectid;
 
-	if (dir && name) {
+	if (args->orphan) {
+		/*
+		 * O_TMPFILE, set link count to 0, so that after this point, we
+		 * fill in an inode item with the correct link count.
+		 */
+		set_nlink(inode, 0);
+	} else {
 		trace_btrfs_inode_request(dir);
 
-		ret = btrfs_set_inode_index(BTRFS_I(dir), index);
-		if (ret) {
-			btrfs_free_path(path);
-			iput(inode);
-			return ERR_PTR(ret);
-		}
-	} else if (dir) {
-		*index = 0;
+		ret = btrfs_set_inode_index(BTRFS_I(dir), &BTRFS_I(inode)->dir_index);
+		if (ret)
+			goto out;
 	}
-	/*
-	 * index_cnt is ignored for everything but a dir,
-	 * btrfs_set_inode_index_count has an explanation for the magic
-	 * number
-	 */
-	BTRFS_I(inode)->index_cnt = 2;
-	BTRFS_I(inode)->dir_index = *index;
-	BTRFS_I(inode)->root = root;
+	/* index_cnt is ignored for everything but a dir. */
+	BTRFS_I(inode)->index_cnt = BTRFS_DIR_START_INDEX;
 	BTRFS_I(inode)->generation = trans->transid;
 	inode->i_generation = BTRFS_I(inode)->generation;
 
 	/*
+	 * Subvolumes don't inherit flags from their parent directory.
+	 * Originally this was probably by accident, but we probably can't
+	 * change it now without compatibility issues.
+	 */
+	if (!args->subvol)
+		btrfs_inherit_iflags(inode, dir);
+
+	if (S_ISREG(inode->i_mode)) {
+		if (btrfs_test_opt(fs_info, NODATASUM))
+			BTRFS_I(inode)->flags |= BTRFS_INODE_NODATASUM;
+		if (btrfs_test_opt(fs_info, NODATACOW))
+			BTRFS_I(inode)->flags |= BTRFS_INODE_NODATACOW |
+				BTRFS_INODE_NODATASUM;
+	}
+
+	location = &BTRFS_I(inode)->location;
+	location->objectid = objectid;
+	location->offset = 0;
+	location->type = BTRFS_INODE_ITEM_KEY;
+
+	ret = btrfs_insert_inode_locked(inode);
+	if (ret < 0) {
+		if (!args->orphan)
+			BTRFS_I(dir)->index_cnt--;
+		goto out;
+	}
+
+	/*
 	 * We could have gotten an inode number from somebody who was fsynced
 	 * and then removed in this same transaction, so let's just set full
 	 * sync since it will be a full sync anyway and this will blow away the
 	 * old info in the log.
 	 */
-	set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &BTRFS_I(inode)->runtime_flags);
+	btrfs_set_inode_full_sync(BTRFS_I(inode));
 
 	key[0].objectid = objectid;
 	key[0].type = BTRFS_INODE_ITEM_KEY;
@@ -5844,7 +6385,7 @@ static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans,
 
 	sizes[0] = sizeof(struct btrfs_inode_item);
 
-	if (name) {
+	if (!args->orphan) {
 		/*
 		 * Start new inodes with an inode_ref. This is slightly more
 		 * efficient for small numbers of hard links since they will
@@ -5853,85 +6394,133 @@ static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans,
 		 */
 		key[1].objectid = objectid;
 		key[1].type = BTRFS_INODE_REF_KEY;
-		key[1].offset = ref_objectid;
-
-		sizes[1] = name_len + sizeof(*ref);
+		if (args->subvol) {
+			key[1].offset = objectid;
+			sizes[1] = 2 + sizeof(*ref);
+		} else {
+			key[1].offset = btrfs_ino(BTRFS_I(dir));
+			sizes[1] = name_len + sizeof(*ref);
+		}
 	}
 
-	location = &BTRFS_I(inode)->location;
-	location->objectid = objectid;
-	location->offset = 0;
-	location->type = BTRFS_INODE_ITEM_KEY;
-
-	ret = btrfs_insert_inode_locked(inode);
-	if (ret < 0) {
-		iput(inode);
-		goto fail;
+	batch.keys = &key[0];
+	batch.data_sizes = &sizes[0];
+	batch.total_data_size = sizes[0] + (args->orphan ? 0 : sizes[1]);
+	batch.nr = args->orphan ? 1 : 2;
+	ret = btrfs_insert_empty_items(trans, root, path, &batch);
+	if (ret != 0) {
+		btrfs_abort_transaction(trans, ret);
+		goto discard;
 	}
 
-	path->leave_spinning = 1;
-	ret = btrfs_insert_empty_items(trans, root, path, key, sizes, nitems);
-	if (ret != 0)
-		goto fail_unlock;
-
-	inode_init_owner(inode, dir, mode);
-	inode_set_bytes(inode, 0);
-
 	inode->i_mtime = current_time(inode);
 	inode->i_atime = inode->i_mtime;
 	inode->i_ctime = inode->i_mtime;
 	BTRFS_I(inode)->i_otime = inode->i_mtime;
 
+	/*
+	 * We're going to fill the inode item now, so at this point the inode
+	 * must be fully initialized.
+	 */
+
 	inode_item = btrfs_item_ptr(path->nodes[0], path->slots[0],
 				  struct btrfs_inode_item);
 	memzero_extent_buffer(path->nodes[0], (unsigned long)inode_item,
 			     sizeof(*inode_item));
 	fill_inode_item(trans, path->nodes[0], inode_item, inode);
 
-	if (name) {
+	if (!args->orphan) {
 		ref = btrfs_item_ptr(path->nodes[0], path->slots[0] + 1,
 				     struct btrfs_inode_ref);
-		btrfs_set_inode_ref_name_len(path->nodes[0], ref, name_len);
-		btrfs_set_inode_ref_index(path->nodes[0], ref, *index);
 		ptr = (unsigned long)(ref + 1);
-		write_extent_buffer(path->nodes[0], name, ptr, name_len);
+		if (args->subvol) {
+			btrfs_set_inode_ref_name_len(path->nodes[0], ref, 2);
+			btrfs_set_inode_ref_index(path->nodes[0], ref, 0);
+			write_extent_buffer(path->nodes[0], "..", ptr, 2);
+		} else {
+			btrfs_set_inode_ref_name_len(path->nodes[0], ref, name_len);
+			btrfs_set_inode_ref_index(path->nodes[0], ref,
+						  BTRFS_I(inode)->dir_index);
+			write_extent_buffer(path->nodes[0], name, ptr, name_len);
+		}
 	}
 
 	btrfs_mark_buffer_dirty(path->nodes[0]);
+	/*
+	 * We don't need the path anymore, plus inheriting properties, adding
+	 * ACLs, security xattrs, orphan item or adding the link, will result in
+	 * allocating yet another path. So just free our path.
+	 */
 	btrfs_free_path(path);
+	path = NULL;
 
-	btrfs_inherit_iflags(inode, dir);
+	if (args->subvol) {
+		struct inode *parent;
 
-	if (S_ISREG(mode)) {
-		if (btrfs_test_opt(fs_info, NODATASUM))
-			BTRFS_I(inode)->flags |= BTRFS_INODE_NODATASUM;
-		if (btrfs_test_opt(fs_info, NODATACOW))
-			BTRFS_I(inode)->flags |= BTRFS_INODE_NODATACOW |
-				BTRFS_INODE_NODATASUM;
+		/*
+		 * Subvolumes inherit properties from their parent subvolume,
+		 * not the directory they were created in.
+		 */
+		parent = btrfs_iget(fs_info->sb, BTRFS_FIRST_FREE_OBJECTID,
+				    BTRFS_I(dir)->root);
+		if (IS_ERR(parent)) {
+			ret = PTR_ERR(parent);
+		} else {
+			ret = btrfs_inode_inherit_props(trans, inode, parent);
+			iput(parent);
+		}
+	} else {
+		ret = btrfs_inode_inherit_props(trans, inode, dir);
+	}
+	if (ret) {
+		btrfs_err(fs_info,
+			  "error inheriting props for ino %llu (root %llu): %d",
+			  btrfs_ino(BTRFS_I(inode)), root->root_key.objectid,
+			  ret);
+	}
+
+	/*
+	 * Subvolumes don't inherit ACLs or get passed to the LSM. This is
+	 * probably a bug.
+	 */
+	if (!args->subvol) {
+		ret = btrfs_init_inode_security(trans, args);
+		if (ret) {
+			btrfs_abort_transaction(trans, ret);
+			goto discard;
+		}
 	}
 
 	inode_tree_add(inode);
 
 	trace_btrfs_inode_new(inode);
-	btrfs_set_inode_last_trans(trans, inode);
+	btrfs_set_inode_last_trans(trans, BTRFS_I(inode));
 
 	btrfs_update_root_times(trans, root);
 
-	ret = btrfs_inode_inherit_props(trans, inode, dir);
-	if (ret)
-		btrfs_err(fs_info,
-			  "error inheriting props for ino %llu (root %llu): %d",
-			btrfs_ino(BTRFS_I(inode)), root->root_key.objectid, ret);
+	if (args->orphan) {
+		ret = btrfs_orphan_add(trans, BTRFS_I(inode));
+	} else {
+		ret = btrfs_add_link(trans, BTRFS_I(dir), BTRFS_I(inode), name,
+				     name_len, 0, BTRFS_I(inode)->dir_index);
+	}
+	if (ret) {
+		btrfs_abort_transaction(trans, ret);
+		goto discard;
+	}
 
-	return inode;
+	return 0;
 
-fail_unlock:
+discard:
+	/*
+	 * discard_new_inode() calls iput(), but the caller owns the reference
+	 * to the inode.
+	 */
+	ihold(inode);
 	discard_new_inode(inode);
-fail:
-	if (dir && name)
-		BTRFS_I(dir)->index_cnt--;
+out:
 	btrfs_free_path(path);
-	return ERR_PTR(ret);
+	return ret;
 }
 
 /*
@@ -5995,7 +6584,7 @@ int btrfs_add_link(struct btrfs_trans_handle *trans,
 		parent_inode->vfs_inode.i_mtime = now;
 		parent_inode->vfs_inode.i_ctime = now;
 	}
-	ret = btrfs_update_inode(trans, root, &parent_inode->vfs_inode);
+	ret = btrfs_update_inode(trans, root, parent_inode);
 	if (ret)
 		btrfs_abort_transaction(trans, ret);
 	return ret;
@@ -6023,148 +6612,71 @@ fail_dir_item:
 	return ret;
 }
 
-static int btrfs_add_nondir(struct btrfs_trans_handle *trans,
-			    struct btrfs_inode *dir, struct dentry *dentry,
-			    struct btrfs_inode *inode, int backref, u64 index)
-{
-	int err = btrfs_add_link(trans, dir, inode,
-				 dentry->d_name.name, dentry->d_name.len,
-				 backref, index);
-	if (err > 0)
-		err = -EEXIST;
-	return err;
-}
-
-static int btrfs_mknod(struct inode *dir, struct dentry *dentry,
-			umode_t mode, dev_t rdev)
+static int btrfs_create_common(struct inode *dir, struct dentry *dentry,
+			       struct inode *inode)
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb);
-	struct btrfs_trans_handle *trans;
 	struct btrfs_root *root = BTRFS_I(dir)->root;
-	struct inode *inode = NULL;
+	struct btrfs_new_inode_args new_inode_args = {
+		.dir = dir,
+		.dentry = dentry,
+		.inode = inode,
+	};
+	unsigned int trans_num_items;
+	struct btrfs_trans_handle *trans;
 	int err;
-	u64 objectid;
-	u64 index = 0;
-
-	/*
-	 * 2 for inode item and ref
-	 * 2 for dir items
-	 * 1 for xattr if selinux is on
-	 */
-	trans = btrfs_start_transaction(root, 5);
-	if (IS_ERR(trans))
-		return PTR_ERR(trans);
 
-	err = btrfs_find_free_ino(root, &objectid);
+	err = btrfs_new_inode_prepare(&new_inode_args, &trans_num_items);
 	if (err)
-		goto out_unlock;
+		goto out_inode;
 
-	inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name,
-			dentry->d_name.len, btrfs_ino(BTRFS_I(dir)), objectid,
-			mode, &index);
-	if (IS_ERR(inode)) {
-		err = PTR_ERR(inode);
-		inode = NULL;
-		goto out_unlock;
+	trans = btrfs_start_transaction(root, trans_num_items);
+	if (IS_ERR(trans)) {
+		err = PTR_ERR(trans);
+		goto out_new_inode_args;
 	}
 
-	/*
-	* If the active LSM wants to access the inode during
-	* d_instantiate it needs these. Smack checks to see
-	* if the filesystem supports xattrs by looking at the
-	* ops vector.
-	*/
-	inode->i_op = &btrfs_special_inode_operations;
-	init_special_inode(inode, inode->i_mode, rdev);
-
-	err = btrfs_init_inode_security(trans, inode, dir, &dentry->d_name);
-	if (err)
-		goto out_unlock;
-
-	err = btrfs_add_nondir(trans, BTRFS_I(dir), dentry, BTRFS_I(inode),
-			0, index);
-	if (err)
-		goto out_unlock;
-
-	btrfs_update_inode(trans, root, inode);
-	d_instantiate_new(dentry, inode);
+	err = btrfs_create_new_inode(trans, &new_inode_args);
+	if (!err)
+		d_instantiate_new(dentry, inode);
 
-out_unlock:
 	btrfs_end_transaction(trans);
 	btrfs_btree_balance_dirty(fs_info);
-	if (err && inode) {
-		inode_dec_link_count(inode);
-		discard_new_inode(inode);
-	}
+out_new_inode_args:
+	btrfs_new_inode_args_destroy(&new_inode_args);
+out_inode:
+	if (err)
+		iput(inode);
 	return err;
 }
 
-static int btrfs_create(struct inode *dir, struct dentry *dentry,
-			umode_t mode, bool excl)
+static int btrfs_mknod(struct user_namespace *mnt_userns, struct inode *dir,
+		       struct dentry *dentry, umode_t mode, dev_t rdev)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb);
-	struct btrfs_trans_handle *trans;
-	struct btrfs_root *root = BTRFS_I(dir)->root;
-	struct inode *inode = NULL;
-	int err;
-	u64 objectid;
-	u64 index = 0;
+	struct inode *inode;
 
-	/*
-	 * 2 for inode item and ref
-	 * 2 for dir items
-	 * 1 for xattr if selinux is on
-	 */
-	trans = btrfs_start_transaction(root, 5);
-	if (IS_ERR(trans))
-		return PTR_ERR(trans);
+	inode = new_inode(dir->i_sb);
+	if (!inode)
+		return -ENOMEM;
+	inode_init_owner(mnt_userns, inode, dir, mode);
+	inode->i_op = &btrfs_special_inode_operations;
+	init_special_inode(inode, inode->i_mode, rdev);
+	return btrfs_create_common(dir, dentry, inode);
+}
 
-	err = btrfs_find_free_ino(root, &objectid);
-	if (err)
-		goto out_unlock;
+static int btrfs_create(struct user_namespace *mnt_userns, struct inode *dir,
+			struct dentry *dentry, umode_t mode, bool excl)
+{
+	struct inode *inode;
 
-	inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name,
-			dentry->d_name.len, btrfs_ino(BTRFS_I(dir)), objectid,
-			mode, &index);
-	if (IS_ERR(inode)) {
-		err = PTR_ERR(inode);
-		inode = NULL;
-		goto out_unlock;
-	}
-	/*
-	* If the active LSM wants to access the inode during
-	* d_instantiate it needs these. Smack checks to see
-	* if the filesystem supports xattrs by looking at the
-	* ops vector.
-	*/
+	inode = new_inode(dir->i_sb);
+	if (!inode)
+		return -ENOMEM;
+	inode_init_owner(mnt_userns, inode, dir, mode);
 	inode->i_fop = &btrfs_file_operations;
 	inode->i_op = &btrfs_file_inode_operations;
 	inode->i_mapping->a_ops = &btrfs_aops;
-
-	err = btrfs_init_inode_security(trans, inode, dir, &dentry->d_name);
-	if (err)
-		goto out_unlock;
-
-	err = btrfs_update_inode(trans, root, inode);
-	if (err)
-		goto out_unlock;
-
-	err = btrfs_add_nondir(trans, BTRFS_I(dir), dentry, BTRFS_I(inode),
-			0, index);
-	if (err)
-		goto out_unlock;
-
-	BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops;
-	d_instantiate_new(dentry, inode);
-
-out_unlock:
-	btrfs_end_transaction(trans);
-	if (err && inode) {
-		inode_dec_link_count(inode);
-		discard_new_inode(inode);
-	}
-	btrfs_btree_balance_dirty(fs_info);
-	return err;
+	return btrfs_create_common(dir, dentry, inode);
 }
 
 static int btrfs_link(struct dentry *old_dentry, struct inode *dir,
@@ -6210,16 +6722,15 @@ static int btrfs_link(struct dentry *old_dentry, struct inode *dir,
 	ihold(inode);
 	set_bit(BTRFS_INODE_COPY_EVERYTHING, &BTRFS_I(inode)->runtime_flags);
 
-	err = btrfs_add_nondir(trans, BTRFS_I(dir), dentry, BTRFS_I(inode),
-			1, index);
+	err = btrfs_add_link(trans, BTRFS_I(dir), BTRFS_I(inode),
+			     dentry->d_name.name, dentry->d_name.len, 1, index);
 
 	if (err) {
 		drop_inode = 1;
 	} else {
 		struct dentry *parent = dentry->d_parent;
-		int ret;
 
-		err = btrfs_update_inode(trans, root, inode);
+		err = btrfs_update_inode(trans, root, BTRFS_I(inode));
 		if (err)
 			goto fail;
 		if (inode->i_nlink == 1) {
@@ -6232,12 +6743,7 @@ static int btrfs_link(struct dentry *old_dentry, struct inode *dir,
 				goto fail;
 		}
 		d_instantiate(dentry, inode);
-		ret = btrfs_log_new_name(trans, BTRFS_I(inode), NULL, parent,
-					 true, NULL);
-		if (ret == BTRFS_NEED_TRANS_COMMIT) {
-			err = btrfs_commit_transaction(trans);
-			trans = NULL;
-		}
+		btrfs_log_new_name(trans, old_dentry, NULL, 0, parent);
 	}
 
 fail:
@@ -6251,67 +6757,18 @@ fail:
 	return err;
 }
 
-static int btrfs_mkdir(struct inode *dir, struct dentry *dentry, umode_t mode)
+static int btrfs_mkdir(struct user_namespace *mnt_userns, struct inode *dir,
+		       struct dentry *dentry, umode_t mode)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb);
-	struct inode *inode = NULL;
-	struct btrfs_trans_handle *trans;
-	struct btrfs_root *root = BTRFS_I(dir)->root;
-	int err = 0;
-	u64 objectid = 0;
-	u64 index = 0;
-
-	/*
-	 * 2 items for inode and ref
-	 * 2 items for dir items
-	 * 1 for xattr if selinux is on
-	 */
-	trans = btrfs_start_transaction(root, 5);
-	if (IS_ERR(trans))
-		return PTR_ERR(trans);
-
-	err = btrfs_find_free_ino(root, &objectid);
-	if (err)
-		goto out_fail;
-
-	inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name,
-			dentry->d_name.len, btrfs_ino(BTRFS_I(dir)), objectid,
-			S_IFDIR | mode, &index);
-	if (IS_ERR(inode)) {
-		err = PTR_ERR(inode);
-		inode = NULL;
-		goto out_fail;
-	}
+	struct inode *inode;
 
-	/* these must be set before we unlock the inode */
+	inode = new_inode(dir->i_sb);
+	if (!inode)
+		return -ENOMEM;
+	inode_init_owner(mnt_userns, inode, dir, S_IFDIR | mode);
 	inode->i_op = &btrfs_dir_inode_operations;
 	inode->i_fop = &btrfs_dir_file_operations;
-
-	err = btrfs_init_inode_security(trans, inode, dir, &dentry->d_name);
-	if (err)
-		goto out_fail;
-
-	btrfs_i_size_write(BTRFS_I(inode), 0);
-	err = btrfs_update_inode(trans, root, inode);
-	if (err)
-		goto out_fail;
-
-	err = btrfs_add_link(trans, BTRFS_I(dir), BTRFS_I(inode),
-			dentry->d_name.name,
-			dentry->d_name.len, 0, index);
-	if (err)
-		goto out_fail;
-
-	d_instantiate_new(dentry, inode);
-
-out_fail:
-	btrfs_end_transaction(trans);
-	if (err && inode) {
-		inode_dec_link_count(inode);
-		discard_new_inode(inode);
-	}
-	btrfs_btree_balance_dirty(fs_info);
-	return err;
+	return btrfs_create_common(dir, dentry, inode);
 }
 
 static noinline int uncompress_inline(struct btrfs_path *path,
@@ -6330,8 +6787,7 @@ static noinline int uncompress_inline(struct btrfs_path *path,
 	WARN_ON(pg_offset != 0);
 	compress_type = btrfs_file_extent_compression(leaf, item);
 	max_size = btrfs_file_extent_ram_bytes(leaf, item);
-	inline_size = btrfs_file_extent_inline_item_len(leaf,
-					btrfs_item_nr(path->slots[0]));
+	inline_size = btrfs_file_extent_inline_item_len(leaf, path->slots[0]);
 	tmp = kmalloc(inline_size, GFP_NOFS);
 	if (!tmp)
 		return -ENOMEM;
@@ -6351,11 +6807,9 @@ static noinline int uncompress_inline(struct btrfs_path *path,
 	 * cover that region here.
 	 */
 
-	if (max_size + pg_offset < PAGE_SIZE) {
-		char *map = kmap(page);
-		memset(map + pg_offset + max_size, 0, PAGE_SIZE - max_size - pg_offset);
-		kunmap(page);
-	}
+	if (max_size + pg_offset < PAGE_SIZE)
+		memzero_page(page,  pg_offset + max_size,
+			     PAGE_SIZE - max_size - pg_offset);
 	kfree(tmp);
 	return ret;
 }
@@ -6383,8 +6837,7 @@ struct extent_map *btrfs_get_extent(struct btrfs_inode *inode,
 				    u64 start, u64 len)
 {
 	struct btrfs_fs_info *fs_info = inode->root->fs_info;
-	int ret;
-	int err = 0;
+	int ret = 0;
 	u64 extent_start = 0;
 	u64 extent_end = 0;
 	u64 objectid = btrfs_ino(inode);
@@ -6396,7 +6849,6 @@ struct extent_map *btrfs_get_extent(struct btrfs_inode *inode,
 	struct btrfs_key found_key;
 	struct extent_map *em = NULL;
 	struct extent_map_tree *em_tree = &inode->extent_tree;
-	struct extent_io_tree *io_tree = &inode->io_tree;
 
 	read_lock(&em_tree->lock);
 	em = lookup_extent_mapping(em_tree, start, len);
@@ -6412,7 +6864,7 @@ struct extent_map *btrfs_get_extent(struct btrfs_inode *inode,
 	}
 	em = alloc_extent_map();
 	if (!em) {
-		err = -ENOMEM;
+		ret = -ENOMEM;
 		goto out;
 	}
 	em->start = EXTENT_MAP_HOLE;
@@ -6422,7 +6874,7 @@ struct extent_map *btrfs_get_extent(struct btrfs_inode *inode,
 
 	path = btrfs_alloc_path();
 	if (!path) {
-		err = -ENOMEM;
+		ret = -ENOMEM;
 		goto out;
 	}
 
@@ -6430,19 +6882,23 @@ struct extent_map *btrfs_get_extent(struct btrfs_inode *inode,
 	path->reada = READA_FORWARD;
 
 	/*
-	 * Unless we're going to uncompress the inline extent, no sleep would
-	 * happen.
+	 * The same explanation in load_free_space_cache applies here as well,
+	 * we only read when we're loading the free space cache, and at that
+	 * point the commit_root has everything we need.
 	 */
-	path->leave_spinning = 1;
+	if (btrfs_is_free_space_inode(inode)) {
+		path->search_commit_root = 1;
+		path->skip_locking = 1;
+	}
 
 	ret = btrfs_lookup_file_extent(NULL, root, path, objectid, start, 0);
 	if (ret < 0) {
-		err = ret;
 		goto out;
 	} else if (ret > 0) {
 		if (path->slots[0] == 0)
 			goto not_found;
 		path->slots[0]--;
+		ret = 0;
 	}
 
 	leaf = path->nodes[0];
@@ -6463,6 +6919,7 @@ struct extent_map *btrfs_get_extent(struct btrfs_inode *inode,
 
 	extent_type = btrfs_file_extent_type(leaf, item);
 	extent_start = found_key.offset;
+	extent_end = btrfs_file_extent_end(path);
 	if (extent_type == BTRFS_FILE_EXTENT_REG ||
 	    extent_type == BTRFS_FILE_EXTENT_PREALLOC) {
 		/* Only regular file could have regular/prealloc extent */
@@ -6473,18 +6930,9 @@ struct extent_map *btrfs_get_extent(struct btrfs_inode *inode,
 				   btrfs_ino(inode));
 			goto out;
 		}
-		extent_end = extent_start +
-		       btrfs_file_extent_num_bytes(leaf, item);
-
 		trace_btrfs_get_extent_show_fi_regular(inode, leaf, item,
 						       extent_start);
 	} else if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
-		size_t size;
-
-		size = btrfs_file_extent_ram_bytes(leaf, item);
-		extent_end = ALIGN(extent_start + size,
-				   fs_info->sectorsize);
-
 		trace_btrfs_get_extent_show_fi_inline(inode, leaf, item,
 						      path->slots[0],
 						      extent_start);
@@ -6494,12 +6942,11 @@ next:
 		path->slots[0]++;
 		if (path->slots[0] >= btrfs_header_nritems(leaf)) {
 			ret = btrfs_next_leaf(root, path);
-			if (ret < 0) {
-				err = ret;
+			if (ret < 0)
 				goto out;
-			} else if (ret > 0) {
+			else if (ret > 0)
 				goto not_found;
-			}
+
 			leaf = path->nodes[0];
 		}
 		btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
@@ -6544,18 +6991,15 @@ next:
 		em->orig_start = em->start;
 		ptr = btrfs_file_extent_inline_start(item) + extent_offset;
 
-		btrfs_set_path_blocking(path);
 		if (!PageUptodate(page)) {
 			if (btrfs_file_extent_compression(leaf, item) !=
 			    BTRFS_COMPRESS_NONE) {
 				ret = uncompress_inline(path, page, pg_offset,
 							extent_offset, item);
-				if (ret) {
-					err = ret;
+				if (ret)
 					goto out;
-				}
 			} else {
-				map = kmap(page);
+				map = kmap_local_page(page);
 				read_extent_buffer(leaf, map + pg_offset, ptr,
 						   copy_size);
 				if (pg_offset + copy_size < PAGE_SIZE) {
@@ -6563,12 +7007,10 @@ next:
 					       PAGE_SIZE - pg_offset -
 					       copy_size);
 				}
-				kunmap(page);
+				kunmap_local(map);
 			}
 			flush_dcache_page(page);
 		}
-		set_extent_uptodate(io_tree, em->start,
-				    extent_map_end(em) - 1, NULL, GFP_NOFS);
 		goto insert;
 	}
 not_found:
@@ -6577,160 +7019,32 @@ not_found:
 	em->len = len;
 	em->block_start = EXTENT_MAP_HOLE;
 insert:
+	ret = 0;
 	btrfs_release_path(path);
 	if (em->start > start || extent_map_end(em) <= start) {
 		btrfs_err(fs_info,
 			  "bad extent! em: [%llu %llu] passed [%llu %llu]",
 			  em->start, em->len, start, len);
-		err = -EIO;
+		ret = -EIO;
 		goto out;
 	}
 
-	err = 0;
 	write_lock(&em_tree->lock);
-	err = btrfs_add_extent_mapping(fs_info, em_tree, &em, start, len);
+	ret = btrfs_add_extent_mapping(fs_info, em_tree, &em, start, len);
 	write_unlock(&em_tree->lock);
 out:
 	btrfs_free_path(path);
 
 	trace_btrfs_get_extent(root, inode, em);
 
-	if (err) {
-		free_extent_map(em);
-		return ERR_PTR(err);
-	}
-	BUG_ON(!em); /* Error is always set */
-	return em;
-}
-
-struct extent_map *btrfs_get_extent_fiemap(struct btrfs_inode *inode,
-					   u64 start, u64 len)
-{
-	struct extent_map *em;
-	struct extent_map *hole_em = NULL;
-	u64 delalloc_start = start;
-	u64 end;
-	u64 delalloc_len;
-	u64 delalloc_end;
-	int err = 0;
-
-	em = btrfs_get_extent(inode, NULL, 0, start, len);
-	if (IS_ERR(em))
-		return em;
-	/*
-	 * If our em maps to:
-	 * - a hole or
-	 * - a pre-alloc extent,
-	 * there might actually be delalloc bytes behind it.
-	 */
-	if (em->block_start != EXTENT_MAP_HOLE &&
-	    !test_bit(EXTENT_FLAG_PREALLOC, &em->flags))
-		return em;
-	else
-		hole_em = em;
-
-	/* check to see if we've wrapped (len == -1 or similar) */
-	end = start + len;
-	if (end < start)
-		end = (u64)-1;
-	else
-		end -= 1;
-
-	em = NULL;
-
-	/* ok, we didn't find anything, lets look for delalloc */
-	delalloc_len = count_range_bits(&inode->io_tree, &delalloc_start,
-				 end, len, EXTENT_DELALLOC, 1);
-	delalloc_end = delalloc_start + delalloc_len;
-	if (delalloc_end < delalloc_start)
-		delalloc_end = (u64)-1;
-
-	/*
-	 * We didn't find anything useful, return the original results from
-	 * get_extent()
-	 */
-	if (delalloc_start > end || delalloc_end <= start) {
-		em = hole_em;
-		hole_em = NULL;
-		goto out;
-	}
-
-	/*
-	 * Adjust the delalloc_start to make sure it doesn't go backwards from
-	 * the start they passed in
-	 */
-	delalloc_start = max(start, delalloc_start);
-	delalloc_len = delalloc_end - delalloc_start;
-
-	if (delalloc_len > 0) {
-		u64 hole_start;
-		u64 hole_len;
-		const u64 hole_end = extent_map_end(hole_em);
-
-		em = alloc_extent_map();
-		if (!em) {
-			err = -ENOMEM;
-			goto out;
-		}
-
-		ASSERT(hole_em);
-		/*
-		 * When btrfs_get_extent can't find anything it returns one
-		 * huge hole
-		 *
-		 * Make sure what it found really fits our range, and adjust to
-		 * make sure it is based on the start from the caller
-		 */
-		if (hole_end <= start || hole_em->start > end) {
-		       free_extent_map(hole_em);
-		       hole_em = NULL;
-		} else {
-		       hole_start = max(hole_em->start, start);
-		       hole_len = hole_end - hole_start;
-		}
-
-		if (hole_em && delalloc_start > hole_start) {
-			/*
-			 * Our hole starts before our delalloc, so we have to
-			 * return just the parts of the hole that go until the
-			 * delalloc starts
-			 */
-			em->len = min(hole_len, delalloc_start - hole_start);
-			em->start = hole_start;
-			em->orig_start = hole_start;
-			/*
-			 * Don't adjust block start at all, it is fixed at
-			 * EXTENT_MAP_HOLE
-			 */
-			em->block_start = hole_em->block_start;
-			em->block_len = hole_len;
-			if (test_bit(EXTENT_FLAG_PREALLOC, &hole_em->flags))
-				set_bit(EXTENT_FLAG_PREALLOC, &em->flags);
-		} else {
-			/*
-			 * Hole is out of passed range or it starts after
-			 * delalloc range
-			 */
-			em->start = delalloc_start;
-			em->len = delalloc_len;
-			em->orig_start = delalloc_start;
-			em->block_start = EXTENT_MAP_DELALLOC;
-			em->block_len = delalloc_len;
-		}
-	} else {
-		return hole_em;
-	}
-out:
-
-	free_extent_map(hole_em);
-	if (err) {
+	if (ret) {
 		free_extent_map(em);
-		return ERR_PTR(err);
+		return ERR_PTR(ret);
 	}
 	return em;
 }
 
-static struct extent_map *btrfs_create_dio_extent(struct inode *inode,
+static struct extent_map *btrfs_create_dio_extent(struct btrfs_inode *inode,
 						  const u64 start,
 						  const u64 len,
 						  const u64 orig_start,
@@ -6744,21 +7058,23 @@ static struct extent_map *btrfs_create_dio_extent(struct inode *inode,
 	int ret;
 
 	if (type != BTRFS_ORDERED_NOCOW) {
-		em = create_io_em(inode, start, len, orig_start,
-				  block_start, block_len, orig_block_len,
-				  ram_bytes,
+		em = create_io_em(inode, start, len, orig_start, block_start,
+				  block_len, orig_block_len, ram_bytes,
 				  BTRFS_COMPRESS_NONE, /* compress_type */
 				  type);
 		if (IS_ERR(em))
 			goto out;
 	}
-	ret = btrfs_add_ordered_extent_dio(inode, start, block_start,
-					   len, block_len, type);
+	ret = btrfs_add_ordered_extent(inode, start, len, len, block_start,
+				       block_len, 0,
+				       (1 << type) |
+				       (1 << BTRFS_ORDERED_DIRECT),
+				       BTRFS_COMPRESS_NONE);
 	if (ret) {
 		if (em) {
 			free_extent_map(em);
-			btrfs_drop_extent_cache(BTRFS_I(inode), start,
-						start + len - 1, 0);
+			btrfs_drop_extent_map_range(inode, start,
+						    start + len - 1, false);
 		}
 		em = ERR_PTR(ret);
 	}
@@ -6767,11 +7083,11 @@ static struct extent_map *btrfs_create_dio_extent(struct inode *inode,
 	return em;
 }
 
-static struct extent_map *btrfs_new_extent_direct(struct inode *inode,
+static struct extent_map *btrfs_new_extent_direct(struct btrfs_inode *inode,
 						  u64 start, u64 len)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	struct btrfs_root *root = BTRFS_I(inode)->root;
+	struct btrfs_root *root = inode->root;
+	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct extent_map *em;
 	struct btrfs_key ins;
 	u64 alloc_hint;
@@ -6788,21 +7104,51 @@ static struct extent_map *btrfs_new_extent_direct(struct inode *inode,
 				     ins.offset, BTRFS_ORDERED_REGULAR);
 	btrfs_dec_block_group_reservations(fs_info, ins.objectid);
 	if (IS_ERR(em))
-		btrfs_free_reserved_extent(fs_info, ins.objectid,
-					   ins.offset, 1);
+		btrfs_free_reserved_extent(fs_info, ins.objectid, ins.offset,
+					   1);
 
 	return em;
 }
 
+static bool btrfs_extent_readonly(struct btrfs_fs_info *fs_info, u64 bytenr)
+{
+	struct btrfs_block_group *block_group;
+	bool readonly = false;
+
+	block_group = btrfs_lookup_block_group(fs_info, bytenr);
+	if (!block_group || block_group->ro)
+		readonly = true;
+	if (block_group)
+		btrfs_put_block_group(block_group);
+	return readonly;
+}
+
 /*
- * returns 1 when the nocow is safe, < 1 on error, 0 if the
- * block must be cow'd
+ * Check if we can do nocow write into the range [@offset, @offset + @len)
+ *
+ * @offset:	File offset
+ * @len:	The length to write, will be updated to the nocow writeable
+ *		range
+ * @orig_start:	(optional) Return the original file offset of the file extent
+ * @orig_len:	(optional) Return the original on-disk length of the file extent
+ * @ram_bytes:	(optional) Return the ram_bytes of the file extent
+ * @strict:	if true, omit optimizations that might force us into unnecessary
+ *		cow. e.g., don't trust generation number.
+ *
+ * Return:
+ * >0	and update @len if we can do nocow write
+ *  0	if we can't do nocow write
+ * <0	if error happened
+ *
+ * NOTE: This only checks the file extents, caller is responsible to wait for
+ *	 any ordered extents.
  */
 noinline int can_nocow_extent(struct inode *inode, u64 offset, u64 *len,
 			      u64 *orig_start, u64 *orig_block_len,
-			      u64 *ram_bytes)
+			      u64 *ram_bytes, bool nowait, bool strict)
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+	struct can_nocow_file_extent_args nocow_args = { 0 };
 	struct btrfs_path *path;
 	int ret;
 	struct extent_buffer *leaf;
@@ -6810,35 +7156,29 @@ noinline int can_nocow_extent(struct inode *inode, u64 offset, u64 *len,
 	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
 	struct btrfs_file_extent_item *fi;
 	struct btrfs_key key;
-	u64 disk_bytenr;
-	u64 backref_offset;
-	u64 extent_end;
-	u64 num_bytes;
-	int slot;
 	int found_type;
-	bool nocow = (BTRFS_I(inode)->flags & BTRFS_INODE_NODATACOW);
 
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
+	path->nowait = nowait;
 
 	ret = btrfs_lookup_file_extent(NULL, root, path,
 			btrfs_ino(BTRFS_I(inode)), offset, 0);
 	if (ret < 0)
 		goto out;
 
-	slot = path->slots[0];
 	if (ret == 1) {
-		if (slot == 0) {
+		if (path->slots[0] == 0) {
 			/* can't find the item, must cow */
 			ret = 0;
 			goto out;
 		}
-		slot--;
+		path->slots[0]--;
 	}
 	ret = 0;
 	leaf = path->nodes[0];
-	btrfs_item_key_to_cpu(leaf, &key, slot);
+	btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
 	if (key.objectid != btrfs_ino(BTRFS_I(inode)) ||
 	    key.type != BTRFS_EXTENT_DATA_KEY) {
 		/* not our file or wrong item type, must cow */
@@ -6850,54 +7190,38 @@ noinline int can_nocow_extent(struct inode *inode, u64 offset, u64 *len,
 		goto out;
 	}
 
-	fi = btrfs_item_ptr(leaf, slot, struct btrfs_file_extent_item);
-	found_type = btrfs_file_extent_type(leaf, fi);
-	if (found_type != BTRFS_FILE_EXTENT_REG &&
-	    found_type != BTRFS_FILE_EXTENT_PREALLOC) {
-		/* not a regular extent, must cow */
-		goto out;
-	}
-
-	if (!nocow && found_type == BTRFS_FILE_EXTENT_REG)
+	if (btrfs_file_extent_end(path) <= offset)
 		goto out;
 
-	extent_end = key.offset + btrfs_file_extent_num_bytes(leaf, fi);
-	if (extent_end <= offset)
-		goto out;
+	fi = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_file_extent_item);
+	found_type = btrfs_file_extent_type(leaf, fi);
+	if (ram_bytes)
+		*ram_bytes = btrfs_file_extent_ram_bytes(leaf, fi);
 
-	disk_bytenr = btrfs_file_extent_disk_bytenr(leaf, fi);
-	if (disk_bytenr == 0)
-		goto out;
+	nocow_args.start = offset;
+	nocow_args.end = offset + *len - 1;
+	nocow_args.strict = strict;
+	nocow_args.free_path = true;
 
-	if (btrfs_file_extent_compression(leaf, fi) ||
-	    btrfs_file_extent_encryption(leaf, fi) ||
-	    btrfs_file_extent_other_encoding(leaf, fi))
-		goto out;
+	ret = can_nocow_file_extent(path, &key, BTRFS_I(inode), &nocow_args);
+	/* can_nocow_file_extent() has freed the path. */
+	path = NULL;
 
-	/*
-	 * Do the same check as in btrfs_cross_ref_exist but without the
-	 * unnecessary search.
-	 */
-	if (btrfs_file_extent_generation(leaf, fi) <=
-	    btrfs_root_last_snapshot(&root->root_item))
+	if (ret != 1) {
+		/* Treat errors as not being able to NOCOW. */
+		ret = 0;
 		goto out;
-
-	backref_offset = btrfs_file_extent_offset(leaf, fi);
-
-	if (orig_start) {
-		*orig_start = key.offset - backref_offset;
-		*orig_block_len = btrfs_file_extent_disk_num_bytes(leaf, fi);
-		*ram_bytes = btrfs_file_extent_ram_bytes(leaf, fi);
 	}
 
-	if (btrfs_extent_readonly(fs_info, disk_bytenr))
+	ret = 0;
+	if (btrfs_extent_readonly(fs_info, nocow_args.disk_bytenr))
 		goto out;
 
-	num_bytes = min(offset + *len, extent_end) - offset;
-	if (!nocow && found_type == BTRFS_FILE_EXTENT_PREALLOC) {
+	if (!(BTRFS_I(inode)->flags & BTRFS_INODE_NODATACOW) &&
+	    found_type == BTRFS_FILE_EXTENT_PREALLOC) {
 		u64 range_end;
 
-		range_end = round_up(offset + num_bytes,
+		range_end = round_up(offset + nocow_args.num_bytes,
 				     root->fs_info->sectorsize) - 1;
 		ret = test_range_bit(io_tree, offset, range_end,
 				     EXTENT_DELALLOC, 0, NULL);
@@ -6907,35 +7231,12 @@ noinline int can_nocow_extent(struct inode *inode, u64 offset, u64 *len,
 		}
 	}
 
-	btrfs_release_path(path);
-
-	/*
-	 * look for other files referencing this extent, if we
-	 * find any we must cow
-	 */
-
-	ret = btrfs_cross_ref_exist(root, btrfs_ino(BTRFS_I(inode)),
-				    key.offset - backref_offset, disk_bytenr);
-	if (ret) {
-		ret = 0;
-		goto out;
-	}
+	if (orig_start)
+		*orig_start = key.offset - nocow_args.extent_offset;
+	if (orig_block_len)
+		*orig_block_len = nocow_args.disk_num_bytes;
 
-	/*
-	 * adjust disk_bytenr and num_bytes to cover just the bytes
-	 * in this extent we are about to write.  If there
-	 * are any csums in that range we have to cow in order
-	 * to keep the csums correct
-	 */
-	disk_bytenr += backref_offset;
-	disk_bytenr += offset - key.offset;
-	if (csum_exist_in_range(fs_info, disk_bytenr, num_bytes))
-		goto out;
-	/*
-	 * all of the above have passed, it is safe to overwrite this extent
-	 * without cow
-	 */
-	*len = num_bytes;
+	*len = nocow_args.num_bytes;
 	ret = 1;
 out:
 	btrfs_free_path(path);
@@ -6943,14 +7244,22 @@ out:
 }
 
 static int lock_extent_direct(struct inode *inode, u64 lockstart, u64 lockend,
-			      struct extent_state **cached_state, int writing)
+			      struct extent_state **cached_state,
+			      unsigned int iomap_flags)
 {
+	const bool writing = (iomap_flags & IOMAP_WRITE);
+	const bool nowait = (iomap_flags & IOMAP_NOWAIT);
+	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
 	struct btrfs_ordered_extent *ordered;
 	int ret = 0;
 
 	while (1) {
-		lock_extent_bits(&BTRFS_I(inode)->io_tree, lockstart, lockend,
-				 cached_state);
+		if (nowait) {
+			if (!try_lock_extent(io_tree, lockstart, lockend))
+				return -EAGAIN;
+		} else {
+			lock_extent(io_tree, lockstart, lockend, cached_state);
+		}
 		/*
 		 * We're concerned with the entire range that we're going to be
 		 * doing DIO to, so we need to make sure there's no ordered
@@ -6971,10 +7280,14 @@ static int lock_extent_direct(struct inode *inode, u64 lockstart, u64 lockend,
 							 lockstart, lockend)))
 			break;
 
-		unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart, lockend,
-				     cached_state);
+		unlock_extent(io_tree, lockstart, lockend, cached_state);
 
 		if (ordered) {
+			if (nowait) {
+				btrfs_put_ordered_extent(ordered);
+				ret = -EAGAIN;
+				break;
+			}
 			/*
 			 * If we are doing a DIO read and the ordered extent we
 			 * found is for a buffered write, we can not wait for it
@@ -6992,9 +7305,9 @@ static int lock_extent_direct(struct inode *inode, u64 lockstart, u64 lockend,
 			 */
 			if (writing ||
 			    test_bit(BTRFS_ORDERED_DIRECT, &ordered->flags))
-				btrfs_start_ordered_extent(inode, ordered, 1);
+				btrfs_start_ordered_extent(ordered, 1);
 			else
-				ret = -ENOTBLK;
+				ret = nowait ? -EAGAIN : -ENOTBLK;
 			btrfs_put_ordered_extent(ordered);
 		} else {
 			/*
@@ -7002,15 +7315,15 @@ static int lock_extent_direct(struct inode *inode, u64 lockstart, u64 lockend,
 			 * for it to complete) and then invalidate the pages for
 			 * this range (through invalidate_inode_pages2_range()),
 			 * but that can lead us to a deadlock with a concurrent
-			 * call to readpages() (a buffered read or a defrag call
+			 * call to readahead (a buffered read or a defrag call
 			 * triggered a readahead) on a page lock due to an
 			 * ordered dio extent we created before but did not have
 			 * yet a corresponding bio submitted (whence it can not
-			 * complete), which makes readpages() wait for that
+			 * complete), which makes readahead wait for that
 			 * ordered extent to complete while holding a lock on
 			 * that page.
 			 */
-			ret = -ENOTBLK;
+			ret = nowait ? -EAGAIN : -ENOTBLK;
 		}
 
 		if (ret)
@@ -7023,13 +7336,12 @@ static int lock_extent_direct(struct inode *inode, u64 lockstart, u64 lockend,
 }
 
 /* The callers of this must take lock_extent() */
-static struct extent_map *create_io_em(struct inode *inode, u64 start, u64 len,
-				       u64 orig_start, u64 block_start,
+static struct extent_map *create_io_em(struct btrfs_inode *inode, u64 start,
+				       u64 len, u64 orig_start, u64 block_start,
 				       u64 block_len, u64 orig_block_len,
 				       u64 ram_bytes, int compress_type,
 				       int type)
 {
-	struct extent_map_tree *em_tree;
 	struct extent_map *em;
 	int ret;
 
@@ -7038,7 +7350,6 @@ static struct extent_map *create_io_em(struct inode *inode, u64 start, u64 len,
 	       type == BTRFS_ORDERED_NOCOW ||
 	       type == BTRFS_ORDERED_REGULAR);
 
-	em_tree = &BTRFS_I(inode)->extent_tree;
 	em = alloc_extent_map();
 	if (!em)
 		return ERR_PTR(-ENOMEM);
@@ -7059,18 +7370,7 @@ static struct extent_map *create_io_em(struct inode *inode, u64 start, u64 len,
 		em->compress_type = compress_type;
 	}
 
-	do {
-		btrfs_drop_extent_cache(BTRFS_I(inode), em->start,
-				em->start + em->len - 1, 0);
-		write_lock(&em_tree->lock);
-		ret = add_extent_mapping(em_tree, em, 1);
-		write_unlock(&em_tree->lock);
-		/*
-		 * The caller has taken lock_extent(), who could race with us
-		 * to add em?
-		 */
-	} while (ret == -EEXIST);
-
+	ret = btrfs_replace_extent_map_range(inode, em, true);
 	if (ret) {
 		free_extent_map(em);
 		return ERR_PTR(ret);
@@ -7081,36 +7381,21 @@ static struct extent_map *create_io_em(struct inode *inode, u64 start, u64 len,
 }
 
 
-static int btrfs_get_blocks_direct_read(struct extent_map *em,
-					struct buffer_head *bh_result,
-					struct inode *inode,
-					u64 start, u64 len)
-{
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-
-	if (em->block_start == EXTENT_MAP_HOLE ||
-			test_bit(EXTENT_FLAG_PREALLOC, &em->flags))
-		return -ENOENT;
-
-	len = min(len, em->len - (start - em->start));
-
-	bh_result->b_blocknr = (em->block_start + (start - em->start)) >>
-		inode->i_blkbits;
-	bh_result->b_size = len;
-	bh_result->b_bdev = fs_info->fs_devices->latest_bdev;
-	set_buffer_mapped(bh_result);
-
-	return 0;
-}
-
 static int btrfs_get_blocks_direct_write(struct extent_map **map,
-					 struct buffer_head *bh_result,
 					 struct inode *inode,
 					 struct btrfs_dio_data *dio_data,
-					 u64 start, u64 len)
+					 u64 start, u64 len,
+					 unsigned int iomap_flags)
 {
+	const bool nowait = (iomap_flags & IOMAP_NOWAIT);
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	struct extent_map *em = *map;
+	int type;
+	u64 block_start, orig_start, orig_block_len, ram_bytes;
+	struct btrfs_block_group *bg;
+	bool can_nocow = false;
+	bool space_reserved = false;
+	u64 prev_len;
 	int ret = 0;
 
 	/*
@@ -7125,9 +7410,6 @@ static int btrfs_get_blocks_direct_write(struct extent_map **map,
 	if (test_bit(EXTENT_FLAG_PREALLOC, &em->flags) ||
 	    ((BTRFS_I(inode)->flags & BTRFS_INODE_NODATACOW) &&
 	     em->block_start != EXTENT_MAP_HOLE)) {
-		int type;
-		u64 block_start, orig_start, orig_block_len, ram_bytes;
-
 		if (test_bit(EXTENT_FLAG_PREALLOC, &em->flags))
 			type = BTRFS_ORDERED_PREALLOC;
 		else
@@ -7136,108 +7418,207 @@ static int btrfs_get_blocks_direct_write(struct extent_map **map,
 		block_start = em->block_start + (start - em->start);
 
 		if (can_nocow_extent(inode, start, &len, &orig_start,
-				     &orig_block_len, &ram_bytes) == 1 &&
-		    btrfs_inc_nocow_writers(fs_info, block_start)) {
-			struct extent_map *em2;
-
-			em2 = btrfs_create_dio_extent(inode, start, len,
-						      orig_start, block_start,
-						      len, orig_block_len,
-						      ram_bytes, type);
-			btrfs_dec_nocow_writers(fs_info, block_start);
-			if (type == BTRFS_ORDERED_PREALLOC) {
-				free_extent_map(em);
-				*map = em = em2;
-			}
-
-			if (em2 && IS_ERR(em2)) {
-				ret = PTR_ERR(em2);
-				goto out;
-			}
-			/*
-			 * For inode marked NODATACOW or extent marked PREALLOC,
-			 * use the existing or preallocated extent, so does not
-			 * need to adjust btrfs_space_info's bytes_may_use.
-			 */
-			btrfs_free_reserved_data_space_noquota(inode, start,
-							       len);
-			goto skip_cow;
+				     &orig_block_len, &ram_bytes, false, false) == 1) {
+			bg = btrfs_inc_nocow_writers(fs_info, block_start);
+			if (bg)
+				can_nocow = true;
 		}
 	}
 
-	/* this will cow the extent */
-	len = bh_result->b_size;
-	free_extent_map(em);
-	*map = em = btrfs_new_extent_direct(inode, start, len);
-	if (IS_ERR(em)) {
-		ret = PTR_ERR(em);
-		goto out;
-	}
+	prev_len = len;
+	if (can_nocow) {
+		struct extent_map *em2;
 
-	len = min(len, em->len - (start - em->start));
+		/* We can NOCOW, so only need to reserve metadata space. */
+		ret = btrfs_delalloc_reserve_metadata(BTRFS_I(inode), len, len,
+						      nowait);
+		if (ret < 0) {
+			/* Our caller expects us to free the input extent map. */
+			free_extent_map(em);
+			*map = NULL;
+			btrfs_dec_nocow_writers(bg);
+			if (nowait && (ret == -ENOSPC || ret == -EDQUOT))
+				ret = -EAGAIN;
+			goto out;
+		}
+		space_reserved = true;
 
-skip_cow:
-	bh_result->b_blocknr = (em->block_start + (start - em->start)) >>
-		inode->i_blkbits;
-	bh_result->b_size = len;
-	bh_result->b_bdev = fs_info->fs_devices->latest_bdev;
-	set_buffer_mapped(bh_result);
+		em2 = btrfs_create_dio_extent(BTRFS_I(inode), start, len,
+					      orig_start, block_start,
+					      len, orig_block_len,
+					      ram_bytes, type);
+		btrfs_dec_nocow_writers(bg);
+		if (type == BTRFS_ORDERED_PREALLOC) {
+			free_extent_map(em);
+			*map = em2;
+			em = em2;
+		}
 
-	if (!test_bit(EXTENT_FLAG_PREALLOC, &em->flags))
-		set_buffer_new(bh_result);
+		if (IS_ERR(em2)) {
+			ret = PTR_ERR(em2);
+			goto out;
+		}
+
+		dio_data->nocow_done = true;
+	} else {
+		/* Our caller expects us to free the input extent map. */
+		free_extent_map(em);
+		*map = NULL;
+
+		if (nowait)
+			return -EAGAIN;
+
+		/*
+		 * If we could not allocate data space before locking the file
+		 * range and we can't do a NOCOW write, then we have to fail.
+		 */
+		if (!dio_data->data_space_reserved)
+			return -ENOSPC;
+
+		/*
+		 * We have to COW and we have already reserved data space before,
+		 * so now we reserve only metadata.
+		 */
+		ret = btrfs_delalloc_reserve_metadata(BTRFS_I(inode), len, len,
+						      false);
+		if (ret < 0)
+			goto out;
+		space_reserved = true;
+
+		em = btrfs_new_extent_direct(BTRFS_I(inode), start, len);
+		if (IS_ERR(em)) {
+			ret = PTR_ERR(em);
+			goto out;
+		}
+		*map = em;
+		len = min(len, em->len - (start - em->start));
+		if (len < prev_len)
+			btrfs_delalloc_release_metadata(BTRFS_I(inode),
+							prev_len - len, true);
+	}
+
+	/*
+	 * We have created our ordered extent, so we can now release our reservation
+	 * for an outstanding extent.
+	 */
+	btrfs_delalloc_release_extents(BTRFS_I(inode), prev_len);
 
 	/*
 	 * Need to update the i_size under the extent lock so buffered
 	 * readers will get the updated i_size when we unlock.
 	 */
-	if (!dio_data->overwrite && start + len > i_size_read(inode))
+	if (start + len > i_size_read(inode))
 		i_size_write(inode, start + len);
-
-	WARN_ON(dio_data->reserve < len);
-	dio_data->reserve -= len;
-	dio_data->unsubmitted_oe_range_end = start + len;
-	current->journal_info = dio_data;
 out:
+	if (ret && space_reserved) {
+		btrfs_delalloc_release_extents(BTRFS_I(inode), len);
+		btrfs_delalloc_release_metadata(BTRFS_I(inode), len, true);
+	}
 	return ret;
 }
 
-static int btrfs_get_blocks_direct(struct inode *inode, sector_t iblock,
-				   struct buffer_head *bh_result, int create)
+static int btrfs_dio_iomap_begin(struct inode *inode, loff_t start,
+		loff_t length, unsigned int flags, struct iomap *iomap,
+		struct iomap *srcmap)
 {
+	struct iomap_iter *iter = container_of(iomap, struct iomap_iter, iomap);
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	struct extent_map *em;
 	struct extent_state *cached_state = NULL;
-	struct btrfs_dio_data *dio_data = NULL;
-	u64 start = iblock << inode->i_blkbits;
+	struct btrfs_dio_data *dio_data = iter->private;
 	u64 lockstart, lockend;
-	u64 len = bh_result->b_size;
+	const bool write = !!(flags & IOMAP_WRITE);
 	int ret = 0;
+	u64 len = length;
+	const u64 data_alloc_len = length;
+	bool unlock_extents = false;
 
-	if (!create)
-		len = min_t(u64, len, fs_info->sectorsize);
+	/*
+	 * We could potentially fault if we have a buffer > PAGE_SIZE, and if
+	 * we're NOWAIT we may submit a bio for a partial range and return
+	 * EIOCBQUEUED, which would result in an errant short read.
+	 *
+	 * The best way to handle this would be to allow for partial completions
+	 * of iocb's, so we could submit the partial bio, return and fault in
+	 * the rest of the pages, and then submit the io for the rest of the
+	 * range.  However we don't have that currently, so simply return
+	 * -EAGAIN at this point so that the normal path is used.
+	 */
+	if (!write && (flags & IOMAP_NOWAIT) && length > PAGE_SIZE)
+		return -EAGAIN;
+
+	/*
+	 * Cap the size of reads to that usually seen in buffered I/O as we need
+	 * to allocate a contiguous array for the checksums.
+	 */
+	if (!write)
+		len = min_t(u64, len, fs_info->sectorsize * BTRFS_MAX_BIO_SECTORS);
 
 	lockstart = start;
 	lockend = start + len - 1;
 
-	if (current->journal_info) {
-		/*
-		 * Need to pull our outstanding extents and set journal_info to NULL so
-		 * that anything that needs to check if there's a transaction doesn't get
-		 * confused.
-		 */
-		dio_data = current->journal_info;
-		current->journal_info = NULL;
+	/*
+	 * iomap_dio_rw() only does filemap_write_and_wait_range(), which isn't
+	 * enough if we've written compressed pages to this area, so we need to
+	 * flush the dirty pages again to make absolutely sure that any
+	 * outstanding dirty pages are on disk - the first flush only starts
+	 * compression on the data, while keeping the pages locked, so by the
+	 * time the second flush returns we know bios for the compressed pages
+	 * were submitted and finished, and the pages no longer under writeback.
+	 *
+	 * If we have a NOWAIT request and we have any pages in the range that
+	 * are locked, likely due to compression still in progress, we don't want
+	 * to block on page locks. We also don't want to block on pages marked as
+	 * dirty or under writeback (same as for the non-compression case).
+	 * iomap_dio_rw() did the same check, but after that and before we got
+	 * here, mmap'ed writes may have happened or buffered reads started
+	 * (readpage() and readahead(), which lock pages), as we haven't locked
+	 * the file range yet.
+	 */
+	if (test_bit(BTRFS_INODE_HAS_ASYNC_EXTENT,
+		     &BTRFS_I(inode)->runtime_flags)) {
+		if (flags & IOMAP_NOWAIT) {
+			if (filemap_range_needs_writeback(inode->i_mapping,
+							  lockstart, lockend))
+				return -EAGAIN;
+		} else {
+			ret = filemap_fdatawrite_range(inode->i_mapping, start,
+						       start + length - 1);
+			if (ret)
+				return ret;
+		}
+	}
+
+	memset(dio_data, 0, sizeof(*dio_data));
+
+	/*
+	 * We always try to allocate data space and must do it before locking
+	 * the file range, to avoid deadlocks with concurrent writes to the same
+	 * range if the range has several extents and the writes don't expand the
+	 * current i_size (the inode lock is taken in shared mode). If we fail to
+	 * allocate data space here we continue and later, after locking the
+	 * file range, we fail with ENOSPC only if we figure out we can not do a
+	 * NOCOW write.
+	 */
+	if (write && !(flags & IOMAP_NOWAIT)) {
+		ret = btrfs_check_data_free_space(BTRFS_I(inode),
+						  &dio_data->data_reserved,
+						  start, data_alloc_len, false);
+		if (!ret)
+			dio_data->data_space_reserved = true;
+		else if (ret && !(BTRFS_I(inode)->flags &
+				  (BTRFS_INODE_NODATACOW | BTRFS_INODE_PREALLOC)))
+			goto err;
 	}
 
 	/*
 	 * If this errors out it's because we couldn't invalidate pagecache for
-	 * this range and we need to fallback to buffered.
+	 * this range and we need to fallback to buffered IO, or we are doing a
+	 * NOWAIT read/write and we need to block.
 	 */
-	if (lock_extent_direct(inode, lockstart, lockend, &cached_state,
-			       create)) {
-		ret = -ENOTBLK;
+	ret = lock_extent_direct(inode, lockstart, lockend, &cached_state, flags);
+	if (ret < 0)
 		goto err;
-	}
 
 	em = btrfs_get_extent(BTRFS_I(inode), NULL, 0, start, len);
 	if (IS_ERR(em)) {
@@ -7262,1128 +7643,737 @@ static int btrfs_get_blocks_direct(struct inode *inode, sector_t iblock,
 	if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags) ||
 	    em->block_start == EXTENT_MAP_INLINE) {
 		free_extent_map(em);
-		ret = -ENOTBLK;
+		/*
+		 * If we are in a NOWAIT context, return -EAGAIN in order to
+		 * fallback to buffered IO. This is not only because we can
+		 * block with buffered IO (no support for NOWAIT semantics at
+		 * the moment) but also to avoid returning short reads to user
+		 * space - this happens if we were able to read some data from
+		 * previous non-compressed extents and then when we fallback to
+		 * buffered IO, at btrfs_file_read_iter() by calling
+		 * filemap_read(), we fail to fault in pages for the read buffer,
+		 * in which case filemap_read() returns a short read (the number
+		 * of bytes previously read is > 0, so it does not return -EFAULT).
+		 */
+		ret = (flags & IOMAP_NOWAIT) ? -EAGAIN : -ENOTBLK;
 		goto unlock_err;
 	}
 
-	if (create) {
-		ret = btrfs_get_blocks_direct_write(&em, bh_result, inode,
-						    dio_data, start, len);
+	len = min(len, em->len - (start - em->start));
+
+	/*
+	 * If we have a NOWAIT request and the range contains multiple extents
+	 * (or a mix of extents and holes), then we return -EAGAIN to make the
+	 * caller fallback to a context where it can do a blocking (without
+	 * NOWAIT) request. This way we avoid doing partial IO and returning
+	 * success to the caller, which is not optimal for writes and for reads
+	 * it can result in unexpected behaviour for an application.
+	 *
+	 * When doing a read, because we use IOMAP_DIO_PARTIAL when calling
+	 * iomap_dio_rw(), we can end up returning less data then what the caller
+	 * asked for, resulting in an unexpected, and incorrect, short read.
+	 * That is, the caller asked to read N bytes and we return less than that,
+	 * which is wrong unless we are crossing EOF. This happens if we get a
+	 * page fault error when trying to fault in pages for the buffer that is
+	 * associated to the struct iov_iter passed to iomap_dio_rw(), and we
+	 * have previously submitted bios for other extents in the range, in
+	 * which case iomap_dio_rw() may return us EIOCBQUEUED if not all of
+	 * those bios have completed by the time we get the page fault error,
+	 * which we return back to our caller - we should only return EIOCBQUEUED
+	 * after we have submitted bios for all the extents in the range.
+	 */
+	if ((flags & IOMAP_NOWAIT) && len < length) {
+		free_extent_map(em);
+		ret = -EAGAIN;
+		goto unlock_err;
+	}
+
+	if (write) {
+		ret = btrfs_get_blocks_direct_write(&em, inode, dio_data,
+						    start, len, flags);
 		if (ret < 0)
 			goto unlock_err;
+		unlock_extents = true;
+		/* Recalc len in case the new em is smaller than requested */
+		len = min(len, em->len - (start - em->start));
+		if (dio_data->data_space_reserved) {
+			u64 release_offset;
+			u64 release_len = 0;
+
+			if (dio_data->nocow_done) {
+				release_offset = start;
+				release_len = data_alloc_len;
+			} else if (len < data_alloc_len) {
+				release_offset = start + len;
+				release_len = data_alloc_len - len;
+			}
 
-		unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart,
-				     lockend, &cached_state);
-	} else {
-		ret = btrfs_get_blocks_direct_read(em, bh_result, inode,
-						   start, len);
-		/* Can be negative only if we read from a hole */
-		if (ret < 0) {
-			ret = 0;
-			free_extent_map(em);
-			goto unlock_err;
+			if (release_len > 0)
+				btrfs_free_reserved_data_space(BTRFS_I(inode),
+							       dio_data->data_reserved,
+							       release_offset,
+							       release_len);
 		}
+	} else {
 		/*
 		 * We need to unlock only the end area that we aren't using.
 		 * The rest is going to be unlocked by the endio routine.
 		 */
-		lockstart = start + bh_result->b_size;
-		if (lockstart < lockend) {
-			unlock_extent_cached(&BTRFS_I(inode)->io_tree,
-					     lockstart, lockend, &cached_state);
-		} else {
-			free_extent_state(cached_state);
-		}
+		lockstart = start + len;
+		if (lockstart < lockend)
+			unlock_extents = true;
 	}
 
+	if (unlock_extents)
+		unlock_extent(&BTRFS_I(inode)->io_tree, lockstart, lockend,
+			      &cached_state);
+	else
+		free_extent_state(cached_state);
+
+	/*
+	 * Translate extent map information to iomap.
+	 * We trim the extents (and move the addr) even though iomap code does
+	 * that, since we have locked only the parts we are performing I/O in.
+	 */
+	if ((em->block_start == EXTENT_MAP_HOLE) ||
+	    (test_bit(EXTENT_FLAG_PREALLOC, &em->flags) && !write)) {
+		iomap->addr = IOMAP_NULL_ADDR;
+		iomap->type = IOMAP_HOLE;
+	} else {
+		iomap->addr = em->block_start + (start - em->start);
+		iomap->type = IOMAP_MAPPED;
+	}
+	iomap->offset = start;
+	iomap->bdev = fs_info->fs_devices->latest_dev->bdev;
+	iomap->length = len;
+
+	if (write && btrfs_use_zone_append(BTRFS_I(inode), em->block_start))
+		iomap->flags |= IOMAP_F_ZONE_APPEND;
+
 	free_extent_map(em);
 
 	return 0;
 
 unlock_err:
-	unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart, lockend,
-			     &cached_state);
+	unlock_extent(&BTRFS_I(inode)->io_tree, lockstart, lockend,
+		      &cached_state);
 err:
-	if (dio_data)
-		current->journal_info = dio_data;
-	return ret;
-}
-
-static inline blk_status_t submit_dio_repair_bio(struct inode *inode,
-						 struct bio *bio,
-						 int mirror_num)
-{
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	blk_status_t ret;
-
-	BUG_ON(bio_op(bio) == REQ_OP_WRITE);
-
-	ret = btrfs_bio_wq_end_io(fs_info, bio, BTRFS_WQ_ENDIO_DIO_REPAIR);
-	if (ret)
-		return ret;
-
-	ret = btrfs_map_bio(fs_info, bio, mirror_num);
+	if (dio_data->data_space_reserved) {
+		btrfs_free_reserved_data_space(BTRFS_I(inode),
+					       dio_data->data_reserved,
+					       start, data_alloc_len);
+		extent_changeset_free(dio_data->data_reserved);
+	}
 
 	return ret;
 }
 
-static int btrfs_check_dio_repairable(struct inode *inode,
-				      struct bio *failed_bio,
-				      struct io_failure_record *failrec,
-				      int failed_mirror)
+static int btrfs_dio_iomap_end(struct inode *inode, loff_t pos, loff_t length,
+		ssize_t written, unsigned int flags, struct iomap *iomap)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	int num_copies;
-
-	num_copies = btrfs_num_copies(fs_info, failrec->logical, failrec->len);
-	if (num_copies == 1) {
-		/*
-		 * we only have a single copy of the data, so don't bother with
-		 * all the retry and error correction code that follows. no
-		 * matter what the error is, it is very likely to persist.
-		 */
-		btrfs_debug(fs_info,
-			"Check DIO Repairable: cannot repair, num_copies=%d, next_mirror %d, failed_mirror %d",
-			num_copies, failrec->this_mirror, failed_mirror);
-		return 0;
-	}
-
-	failrec->failed_mirror = failed_mirror;
-	failrec->this_mirror++;
-	if (failrec->this_mirror == failed_mirror)
-		failrec->this_mirror++;
+	struct iomap_iter *iter = container_of(iomap, struct iomap_iter, iomap);
+	struct btrfs_dio_data *dio_data = iter->private;
+	size_t submitted = dio_data->submitted;
+	const bool write = !!(flags & IOMAP_WRITE);
+	int ret = 0;
 
-	if (failrec->this_mirror > num_copies) {
-		btrfs_debug(fs_info,
-			"Check DIO Repairable: (fail) num_copies=%d, next_mirror %d, failed_mirror %d",
-			num_copies, failrec->this_mirror, failed_mirror);
+	if (!write && (iomap->type == IOMAP_HOLE)) {
+		/* If reading from a hole, unlock and return */
+		unlock_extent(&BTRFS_I(inode)->io_tree, pos, pos + length - 1,
+			      NULL);
 		return 0;
 	}
 
-	return 1;
-}
-
-static blk_status_t dio_read_error(struct inode *inode, struct bio *failed_bio,
-				   struct page *page, unsigned int pgoff,
-				   u64 start, u64 end, int failed_mirror,
-				   bio_end_io_t *repair_endio, void *repair_arg)
-{
-	struct io_failure_record *failrec;
-	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
-	struct extent_io_tree *failure_tree = &BTRFS_I(inode)->io_failure_tree;
-	struct bio *bio;
-	int isector;
-	unsigned int read_mode = 0;
-	int segs;
-	int ret;
-	blk_status_t status;
-	struct bio_vec bvec;
-
-	BUG_ON(bio_op(failed_bio) == REQ_OP_WRITE);
-
-	ret = btrfs_get_io_failure_record(inode, start, end, &failrec);
-	if (ret)
-		return errno_to_blk_status(ret);
-
-	ret = btrfs_check_dio_repairable(inode, failed_bio, failrec,
-					 failed_mirror);
-	if (!ret) {
-		free_io_failure(failure_tree, io_tree, failrec);
-		return BLK_STS_IOERR;
-	}
-
-	segs = bio_segments(failed_bio);
-	bio_get_first_bvec(failed_bio, &bvec);
-	if (segs > 1 ||
-	    (bvec.bv_len > btrfs_inode_sectorsize(inode)))
-		read_mode |= REQ_FAILFAST_DEV;
-
-	isector = start - btrfs_io_bio(failed_bio)->logical;
-	isector >>= inode->i_sb->s_blocksize_bits;
-	bio = btrfs_create_repair_bio(inode, failed_bio, failrec, page,
-				pgoff, isector, repair_endio, repair_arg);
-	bio->bi_opf = REQ_OP_READ | read_mode;
-
-	btrfs_debug(BTRFS_I(inode)->root->fs_info,
-		    "repair DIO read error: submitting new dio read[%#x] to this_mirror=%d, in_validation=%d",
-		    read_mode, failrec->this_mirror, failrec->in_validation);
-
-	status = submit_dio_repair_bio(inode, bio, failrec->this_mirror);
-	if (status) {
-		free_io_failure(failure_tree, io_tree, failrec);
-		bio_put(bio);
+	if (submitted < length) {
+		pos += submitted;
+		length -= submitted;
+		if (write)
+			btrfs_mark_ordered_io_finished(BTRFS_I(inode), NULL,
+						       pos, length, false);
+		else
+			unlock_extent(&BTRFS_I(inode)->io_tree, pos,
+				      pos + length - 1, NULL);
+		ret = -ENOTBLK;
 	}
 
-	return status;
-}
-
-struct btrfs_retry_complete {
-	struct completion done;
-	struct inode *inode;
-	u64 start;
-	int uptodate;
-};
-
-static void btrfs_retry_endio_nocsum(struct bio *bio)
-{
-	struct btrfs_retry_complete *done = bio->bi_private;
-	struct inode *inode = done->inode;
-	struct bio_vec *bvec;
-	struct extent_io_tree *io_tree, *failure_tree;
-	struct bvec_iter_all iter_all;
-
-	if (bio->bi_status)
-		goto end;
-
-	ASSERT(bio->bi_vcnt == 1);
-	io_tree = &BTRFS_I(inode)->io_tree;
-	failure_tree = &BTRFS_I(inode)->io_failure_tree;
-	ASSERT(bio_first_bvec_all(bio)->bv_len == btrfs_inode_sectorsize(inode));
-
-	done->uptodate = 1;
-	ASSERT(!bio_flagged(bio, BIO_CLONED));
-	bio_for_each_segment_all(bvec, bio, iter_all)
-		clean_io_failure(BTRFS_I(inode)->root->fs_info, failure_tree,
-				 io_tree, done->start, bvec->bv_page,
-				 btrfs_ino(BTRFS_I(inode)), 0);
-end:
-	complete(&done->done);
-	bio_put(bio);
+	if (write)
+		extent_changeset_free(dio_data->data_reserved);
+	return ret;
 }
 
-static blk_status_t __btrfs_correct_data_nocsum(struct inode *inode,
-						struct btrfs_io_bio *io_bio)
+static void btrfs_dio_private_put(struct btrfs_dio_private *dip)
 {
-	struct btrfs_fs_info *fs_info;
-	struct bio_vec bvec;
-	struct bvec_iter iter;
-	struct btrfs_retry_complete done;
-	u64 start;
-	unsigned int pgoff;
-	u32 sectorsize;
-	int nr_sectors;
-	blk_status_t ret;
-	blk_status_t err = BLK_STS_OK;
-
-	fs_info = BTRFS_I(inode)->root->fs_info;
-	sectorsize = fs_info->sectorsize;
-
-	start = io_bio->logical;
-	done.inode = inode;
-	io_bio->bio.bi_iter = io_bio->iter;
-
-	bio_for_each_segment(bvec, &io_bio->bio, iter) {
-		nr_sectors = BTRFS_BYTES_TO_BLKS(fs_info, bvec.bv_len);
-		pgoff = bvec.bv_offset;
-
-next_block_or_try_again:
-		done.uptodate = 0;
-		done.start = start;
-		init_completion(&done.done);
-
-		ret = dio_read_error(inode, &io_bio->bio, bvec.bv_page,
-				pgoff, start, start + sectorsize - 1,
-				io_bio->mirror_num,
-				btrfs_retry_endio_nocsum, &done);
-		if (ret) {
-			err = ret;
-			goto next;
-		}
-
-		wait_for_completion_io(&done.done);
-
-		if (!done.uptodate) {
-			/* We might have another mirror, so try again */
-			goto next_block_or_try_again;
-		}
-
-next:
-		start += sectorsize;
+	/*
+	 * This implies a barrier so that stores to dio_bio->bi_status before
+	 * this and loads of dio_bio->bi_status after this are fully ordered.
+	 */
+	if (!refcount_dec_and_test(&dip->refs))
+		return;
 
-		nr_sectors--;
-		if (nr_sectors) {
-			pgoff += sectorsize;
-			ASSERT(pgoff < PAGE_SIZE);
-			goto next_block_or_try_again;
-		}
+	if (btrfs_op(&dip->bio) == BTRFS_MAP_WRITE) {
+		btrfs_mark_ordered_io_finished(BTRFS_I(dip->inode), NULL,
+					       dip->file_offset, dip->bytes,
+					       !dip->bio.bi_status);
+	} else {
+		unlock_extent(&BTRFS_I(dip->inode)->io_tree,
+			      dip->file_offset,
+			      dip->file_offset + dip->bytes - 1, NULL);
 	}
 
-	return err;
+	kfree(dip->csums);
+	bio_endio(&dip->bio);
 }
 
-static void btrfs_retry_endio(struct bio *bio)
+static void submit_dio_repair_bio(struct inode *inode, struct bio *bio,
+				  int mirror_num,
+				  enum btrfs_compression_type compress_type)
 {
-	struct btrfs_retry_complete *done = bio->bi_private;
-	struct btrfs_io_bio *io_bio = btrfs_io_bio(bio);
-	struct extent_io_tree *io_tree, *failure_tree;
-	struct inode *inode = done->inode;
-	struct bio_vec *bvec;
-	int uptodate;
-	int ret;
-	int i = 0;
-	struct bvec_iter_all iter_all;
-
-	if (bio->bi_status)
-		goto end;
-
-	uptodate = 1;
-
-	ASSERT(bio->bi_vcnt == 1);
-	ASSERT(bio_first_bvec_all(bio)->bv_len == btrfs_inode_sectorsize(done->inode));
-
-	io_tree = &BTRFS_I(inode)->io_tree;
-	failure_tree = &BTRFS_I(inode)->io_failure_tree;
+	struct btrfs_dio_private *dip = btrfs_bio(bio)->private;
+	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 
-	ASSERT(!bio_flagged(bio, BIO_CLONED));
-	bio_for_each_segment_all(bvec, bio, iter_all) {
-		ret = __readpage_endio_check(inode, io_bio, i, bvec->bv_page,
-					     bvec->bv_offset, done->start,
-					     bvec->bv_len);
-		if (!ret)
-			clean_io_failure(BTRFS_I(inode)->root->fs_info,
-					 failure_tree, io_tree, done->start,
-					 bvec->bv_page,
-					 btrfs_ino(BTRFS_I(inode)),
-					 bvec->bv_offset);
-		else
-			uptodate = 0;
-		i++;
-	}
+	BUG_ON(bio_op(bio) == REQ_OP_WRITE);
 
-	done->uptodate = uptodate;
-end:
-	complete(&done->done);
-	bio_put(bio);
+	refcount_inc(&dip->refs);
+	btrfs_submit_bio(fs_info, bio, mirror_num);
 }
 
-static blk_status_t __btrfs_subio_endio_read(struct inode *inode,
-		struct btrfs_io_bio *io_bio, blk_status_t err)
+static blk_status_t btrfs_check_read_dio_bio(struct btrfs_dio_private *dip,
+					     struct btrfs_bio *bbio,
+					     const bool uptodate)
 {
-	struct btrfs_fs_info *fs_info;
-	struct bio_vec bvec;
+	struct inode *inode = dip->inode;
+	struct btrfs_fs_info *fs_info = BTRFS_I(inode)->root->fs_info;
+	const bool csum = !(BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM);
+	blk_status_t err = BLK_STS_OK;
 	struct bvec_iter iter;
-	struct btrfs_retry_complete done;
-	u64 start;
-	u64 offset = 0;
-	u32 sectorsize;
-	int nr_sectors;
-	unsigned int pgoff;
-	int csum_pos;
-	bool uptodate = (err == 0);
-	int ret;
-	blk_status_t status;
-
-	fs_info = BTRFS_I(inode)->root->fs_info;
-	sectorsize = fs_info->sectorsize;
-
-	err = BLK_STS_OK;
-	start = io_bio->logical;
-	done.inode = inode;
-	io_bio->bio.bi_iter = io_bio->iter;
+	struct bio_vec bv;
+	u32 offset;
 
-	bio_for_each_segment(bvec, &io_bio->bio, iter) {
-		nr_sectors = BTRFS_BYTES_TO_BLKS(fs_info, bvec.bv_len);
+	btrfs_bio_for_each_sector(fs_info, bv, bbio, iter, offset) {
+		u64 start = bbio->file_offset + offset;
 
-		pgoff = bvec.bv_offset;
-next_block:
-		if (uptodate) {
-			csum_pos = BTRFS_BYTES_TO_BLKS(fs_info, offset);
-			ret = __readpage_endio_check(inode, io_bio, csum_pos,
-					bvec.bv_page, pgoff, start, sectorsize);
-			if (likely(!ret))
-				goto next;
-		}
-try_again:
-		done.uptodate = 0;
-		done.start = start;
-		init_completion(&done.done);
-
-		status = dio_read_error(inode, &io_bio->bio, bvec.bv_page,
-					pgoff, start, start + sectorsize - 1,
-					io_bio->mirror_num, btrfs_retry_endio,
-					&done);
-		if (status) {
-			err = status;
-			goto next;
-		}
-
-		wait_for_completion_io(&done.done);
-
-		if (!done.uptodate) {
-			/* We might have another mirror, so try again */
-			goto try_again;
-		}
-next:
-		offset += sectorsize;
-		start += sectorsize;
-
-		ASSERT(nr_sectors);
+		if (uptodate &&
+		    (!csum || !btrfs_check_data_csum(inode, bbio, offset, bv.bv_page,
+					       bv.bv_offset))) {
+			btrfs_clean_io_failure(BTRFS_I(inode), start,
+					       bv.bv_page, bv.bv_offset);
+		} else {
+			int ret;
 
-		nr_sectors--;
-		if (nr_sectors) {
-			pgoff += sectorsize;
-			ASSERT(pgoff < PAGE_SIZE);
-			goto next_block;
+			ret = btrfs_repair_one_sector(inode, bbio, offset,
+					bv.bv_page, bv.bv_offset,
+					submit_dio_repair_bio);
+			if (ret)
+				err = errno_to_blk_status(ret);
 		}
 	}
 
 	return err;
 }
 
-static blk_status_t btrfs_subio_endio_read(struct inode *inode,
-		struct btrfs_io_bio *io_bio, blk_status_t err)
-{
-	bool skip_csum = BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM;
-
-	if (skip_csum) {
-		if (unlikely(err))
-			return __btrfs_correct_data_nocsum(inode, io_bio);
-		else
-			return BLK_STS_OK;
-	} else {
-		return __btrfs_subio_endio_read(inode, io_bio, err);
-	}
-}
-
-static void btrfs_endio_direct_read(struct bio *bio)
-{
-	struct btrfs_dio_private *dip = bio->bi_private;
-	struct inode *inode = dip->inode;
-	struct bio *dio_bio;
-	struct btrfs_io_bio *io_bio = btrfs_io_bio(bio);
-	blk_status_t err = bio->bi_status;
-
-	if (dip->flags & BTRFS_DIO_ORIG_BIO_SUBMITTED)
-		err = btrfs_subio_endio_read(inode, io_bio, err);
-
-	unlock_extent(&BTRFS_I(inode)->io_tree, dip->logical_offset,
-		      dip->logical_offset + dip->bytes - 1);
-	dio_bio = dip->dio_bio;
-
-	kfree(dip);
-
-	dio_bio->bi_status = err;
-	dio_end_io(dio_bio);
-	btrfs_io_bio_free_csum(io_bio);
-	bio_put(bio);
-}
-
-static void __endio_write_update_ordered(struct inode *inode,
-					 const u64 offset, const u64 bytes,
-					 const bool uptodate)
-{
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	struct btrfs_ordered_extent *ordered = NULL;
-	struct btrfs_workqueue *wq;
-	u64 ordered_offset = offset;
-	u64 ordered_bytes = bytes;
-	u64 last_offset;
-
-	if (btrfs_is_free_space_inode(BTRFS_I(inode)))
-		wq = fs_info->endio_freespace_worker;
-	else
-		wq = fs_info->endio_write_workers;
-
-	while (ordered_offset < offset + bytes) {
-		last_offset = ordered_offset;
-		if (btrfs_dec_test_first_ordered_pending(inode, &ordered,
-							   &ordered_offset,
-							   ordered_bytes,
-							   uptodate)) {
-			btrfs_init_work(&ordered->work, finish_ordered_fn, NULL,
-					NULL);
-			btrfs_queue_work(wq, &ordered->work);
-		}
-		/*
-		 * If btrfs_dec_test_ordered_pending does not find any ordered
-		 * extent in the range, we can exit.
-		 */
-		if (ordered_offset == last_offset)
-			return;
-		/*
-		 * Our bio might span multiple ordered extents. In this case
-		 * we keep going until we have accounted the whole dio.
-		 */
-		if (ordered_offset < offset + bytes) {
-			ordered_bytes = offset + bytes - ordered_offset;
-			ordered = NULL;
-		}
-	}
-}
-
-static void btrfs_endio_direct_write(struct bio *bio)
+static blk_status_t btrfs_submit_bio_start_direct_io(struct inode *inode,
+						     struct bio *bio,
+						     u64 dio_file_offset)
 {
-	struct btrfs_dio_private *dip = bio->bi_private;
-	struct bio *dio_bio = dip->dio_bio;
-
-	__endio_write_update_ordered(dip->inode, dip->logical_offset,
-				     dip->bytes, !bio->bi_status);
-
-	kfree(dip);
-
-	dio_bio->bi_status = bio->bi_status;
-	dio_end_io(dio_bio);
-	bio_put(bio);
-}
-
-static blk_status_t btrfs_submit_bio_start_direct_io(void *private_data,
-				    struct bio *bio, u64 offset)
-{
-	struct inode *inode = private_data;
-	blk_status_t ret;
-	ret = btrfs_csum_one_bio(inode, bio, offset, 1);
-	BUG_ON(ret); /* -ENOMEM */
-	return 0;
+	return btrfs_csum_one_bio(BTRFS_I(inode), bio, dio_file_offset, false);
 }
 
-static void btrfs_end_dio_bio(struct bio *bio)
+static void btrfs_end_dio_bio(struct btrfs_bio *bbio)
 {
-	struct btrfs_dio_private *dip = bio->bi_private;
+	struct btrfs_dio_private *dip = bbio->private;
+	struct bio *bio = &bbio->bio;
 	blk_status_t err = bio->bi_status;
 
 	if (err)
 		btrfs_warn(BTRFS_I(dip->inode)->root->fs_info,
 			   "direct IO failed ino %llu rw %d,%u sector %#Lx len %u err no %d",
 			   btrfs_ino(BTRFS_I(dip->inode)), bio_op(bio),
-			   bio->bi_opf,
-			   (unsigned long long)bio->bi_iter.bi_sector,
+			   bio->bi_opf, bio->bi_iter.bi_sector,
 			   bio->bi_iter.bi_size, err);
 
-	if (dip->subio_endio)
-		err = dip->subio_endio(dip->inode, btrfs_io_bio(bio), err);
+	if (bio_op(bio) == REQ_OP_READ)
+		err = btrfs_check_read_dio_bio(dip, bbio, !err);
 
-	if (err) {
-		/*
-		 * We want to perceive the errors flag being set before
-		 * decrementing the reference count. We don't need a barrier
-		 * since atomic operations with a return value are fully
-		 * ordered as per atomic_t.txt
-		 */
-		dip->errors = 1;
-	}
+	if (err)
+		dip->bio.bi_status = err;
 
-	/* if there are more bios still pending for this dio, just exit */
-	if (!atomic_dec_and_test(&dip->pending_bios))
-		goto out;
+	btrfs_record_physical_zoned(dip->inode, bbio->file_offset, bio);
 
-	if (dip->errors) {
-		bio_io_error(dip->orig_bio);
-	} else {
-		dip->dio_bio->bi_status = BLK_STS_OK;
-		bio_endio(dip->orig_bio);
-	}
-out:
 	bio_put(bio);
+	btrfs_dio_private_put(dip);
 }
 
-static inline blk_status_t btrfs_lookup_and_bind_dio_csum(struct inode *inode,
-						 struct btrfs_dio_private *dip,
-						 struct bio *bio,
-						 u64 file_offset)
-{
-	struct btrfs_io_bio *io_bio = btrfs_io_bio(bio);
-	struct btrfs_io_bio *orig_io_bio = btrfs_io_bio(dip->orig_bio);
-	u16 csum_size;
-	blk_status_t ret;
-
-	/*
-	 * We load all the csum data we need when we submit
-	 * the first bio to reduce the csum tree search and
-	 * contention.
-	 */
-	if (dip->logical_offset == file_offset) {
-		ret = btrfs_lookup_bio_sums(inode, dip->orig_bio, file_offset,
-					    NULL);
-		if (ret)
-			return ret;
-	}
-
-	if (bio == dip->orig_bio)
-		return 0;
-
-	file_offset -= dip->logical_offset;
-	file_offset >>= inode->i_sb->s_blocksize_bits;
-	csum_size = btrfs_super_csum_size(btrfs_sb(inode->i_sb)->super_copy);
-	io_bio->csum = orig_io_bio->csum + csum_size * file_offset;
-
-	return 0;
-}
-
-static inline blk_status_t btrfs_submit_dio_bio(struct bio *bio,
-		struct inode *inode, u64 file_offset, int async_submit)
+static void btrfs_submit_dio_bio(struct bio *bio, struct inode *inode,
+				 u64 file_offset, int async_submit)
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	struct btrfs_dio_private *dip = bio->bi_private;
-	bool write = bio_op(bio) == REQ_OP_WRITE;
+	struct btrfs_dio_private *dip = btrfs_bio(bio)->private;
 	blk_status_t ret;
 
-	/* Check btrfs_submit_bio_hook() for rules about async submit. */
-	if (async_submit)
-		async_submit = !atomic_read(&BTRFS_I(inode)->sync_writers);
-
-	if (!write) {
-		ret = btrfs_bio_wq_end_io(fs_info, bio, BTRFS_WQ_ENDIO_DATA);
-		if (ret)
-			goto err;
-	}
+	/* Save the original iter for read repair */
+	if (btrfs_op(bio) == BTRFS_MAP_READ)
+		btrfs_bio(bio)->iter = bio->bi_iter;
 
 	if (BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM)
 		goto map;
 
-	if (write && async_submit) {
-		ret = btrfs_wq_submit_bio(fs_info, bio, 0, 0,
-					  file_offset, inode,
-					  btrfs_submit_bio_start_direct_io);
-		goto err;
-	} else if (write) {
+	if (btrfs_op(bio) == BTRFS_MAP_WRITE) {
+		/* Check btrfs_submit_data_write_bio() for async submit rules */
+		if (async_submit && !atomic_read(&BTRFS_I(inode)->sync_writers) &&
+		    btrfs_wq_submit_bio(inode, bio, 0, file_offset,
+					btrfs_submit_bio_start_direct_io))
+			return;
+
 		/*
 		 * If we aren't doing async submit, calculate the csum of the
 		 * bio now.
 		 */
-		ret = btrfs_csum_one_bio(inode, bio, file_offset, 1);
-		if (ret)
-			goto err;
+		ret = btrfs_csum_one_bio(BTRFS_I(inode), bio, file_offset, false);
+		if (ret) {
+			btrfs_bio_end_io(btrfs_bio(bio), ret);
+			return;
+		}
 	} else {
-		ret = btrfs_lookup_and_bind_dio_csum(inode, dip, bio,
-						     file_offset);
-		if (ret)
-			goto err;
+		btrfs_bio(bio)->csum = btrfs_csum_ptr(fs_info, dip->csums,
+						      file_offset - dip->file_offset);
 	}
 map:
-	ret = btrfs_map_bio(fs_info, bio, 0);
-err:
-	return ret;
+	btrfs_submit_bio(fs_info, bio, 0);
 }
 
-static int btrfs_submit_direct_hook(struct btrfs_dio_private *dip)
+static void btrfs_submit_direct(const struct iomap_iter *iter,
+		struct bio *dio_bio, loff_t file_offset)
 {
-	struct inode *inode = dip->inode;
+	struct btrfs_dio_private *dip =
+		container_of(dio_bio, struct btrfs_dio_private, bio);
+	struct inode *inode = iter->inode;
+	const bool write = (btrfs_op(dio_bio) == BTRFS_MAP_WRITE);
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+	const bool raid56 = (btrfs_data_alloc_profile(fs_info) &
+			     BTRFS_BLOCK_GROUP_RAID56_MASK);
 	struct bio *bio;
-	struct bio *orig_bio = dip->orig_bio;
-	u64 start_sector = orig_bio->bi_iter.bi_sector;
-	u64 file_offset = dip->logical_offset;
+	u64 start_sector;
 	int async_submit = 0;
 	u64 submit_len;
-	int clone_offset = 0;
-	int clone_len;
+	u64 clone_offset = 0;
+	u64 clone_len;
+	u64 logical;
 	int ret;
 	blk_status_t status;
 	struct btrfs_io_geometry geom;
+	struct btrfs_dio_data *dio_data = iter->private;
+	struct extent_map *em = NULL;
 
-	submit_len = orig_bio->bi_iter.bi_size;
-	ret = btrfs_get_io_geometry(fs_info, btrfs_op(orig_bio),
-				    start_sector << 9, submit_len, &geom);
-	if (ret)
-		return -EIO;
+	dip->inode = inode;
+	dip->file_offset = file_offset;
+	dip->bytes = dio_bio->bi_iter.bi_size;
+	refcount_set(&dip->refs, 1);
+	dip->csums = NULL;
 
-	if (geom.len >= submit_len) {
-		bio = orig_bio;
-		dip->flags |= BTRFS_DIO_ORIG_BIO_SUBMITTED;
-		goto submit;
+	if (!write && !(BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM)) {
+		unsigned int nr_sectors =
+			(dio_bio->bi_iter.bi_size >> fs_info->sectorsize_bits);
+
+		/*
+		 * Load the csums up front to reduce csum tree searches and
+		 * contention when submitting bios.
+		 */
+		status = BLK_STS_RESOURCE;
+		dip->csums = kcalloc(nr_sectors, fs_info->csum_size, GFP_NOFS);
+		if (!dip->csums)
+			goto out_err;
+
+		status = btrfs_lookup_bio_sums(inode, dio_bio, dip->csums);
+		if (status != BLK_STS_OK)
+			goto out_err;
 	}
 
-	/* async crcs make it difficult to collect full stripe writes. */
-	if (btrfs_data_alloc_profile(fs_info) & BTRFS_BLOCK_GROUP_RAID56_MASK)
-		async_submit = 0;
-	else
-		async_submit = 1;
+	start_sector = dio_bio->bi_iter.bi_sector;
+	submit_len = dio_bio->bi_iter.bi_size;
 
-	/* bio split */
-	ASSERT(geom.len <= INT_MAX);
-	atomic_inc(&dip->pending_bios);
 	do {
-		clone_len = min_t(int, submit_len, geom.len);
+		logical = start_sector << 9;
+		em = btrfs_get_chunk_map(fs_info, logical, submit_len);
+		if (IS_ERR(em)) {
+			status = errno_to_blk_status(PTR_ERR(em));
+			em = NULL;
+			goto out_err_em;
+		}
+		ret = btrfs_get_io_geometry(fs_info, em, btrfs_op(dio_bio),
+					    logical, &geom);
+		if (ret) {
+			status = errno_to_blk_status(ret);
+			goto out_err_em;
+		}
+
+		clone_len = min(submit_len, geom.len);
+		ASSERT(clone_len <= UINT_MAX);
 
 		/*
 		 * This will never fail as it's passing GPF_NOFS and
 		 * the allocation is backed by btrfs_bioset.
 		 */
-		bio = btrfs_bio_clone_partial(orig_bio, clone_offset,
-					      clone_len);
-		bio->bi_private = dip;
-		bio->bi_end_io = btrfs_end_dio_bio;
-		btrfs_io_bio(bio)->logical = file_offset;
+		bio = btrfs_bio_clone_partial(dio_bio, clone_offset, clone_len,
+					      btrfs_end_dio_bio, dip);
+		btrfs_bio(bio)->file_offset = file_offset;
+
+		if (bio_op(bio) == REQ_OP_ZONE_APPEND) {
+			status = extract_ordered_extent(BTRFS_I(inode), bio,
+							file_offset);
+			if (status) {
+				bio_put(bio);
+				goto out_err;
+			}
+		}
 
 		ASSERT(submit_len >= clone_len);
 		submit_len -= clone_len;
-		if (submit_len == 0)
-			break;
 
 		/*
 		 * Increase the count before we submit the bio so we know
 		 * the end IO handler won't happen before we increase the
 		 * count. Otherwise, the dip might get freed before we're
 		 * done setting it up.
+		 *
+		 * We transfer the initial reference to the last bio, so we
+		 * don't need to increment the reference count for the last one.
 		 */
-		atomic_inc(&dip->pending_bios);
-
-		status = btrfs_submit_dio_bio(bio, inode, file_offset,
-						async_submit);
-		if (status) {
-			bio_put(bio);
-			atomic_dec(&dip->pending_bios);
-			goto out_err;
+		if (submit_len > 0) {
+			refcount_inc(&dip->refs);
+			/*
+			 * If we are submitting more than one bio, submit them
+			 * all asynchronously. The exception is RAID 5 or 6, as
+			 * asynchronous checksums make it difficult to collect
+			 * full stripe writes.
+			 */
+			if (!raid56)
+				async_submit = 1;
 		}
 
+		btrfs_submit_dio_bio(bio, inode, file_offset, async_submit);
+
+		dio_data->submitted += clone_len;
 		clone_offset += clone_len;
 		start_sector += clone_len >> 9;
 		file_offset += clone_len;
 
-		ret = btrfs_get_io_geometry(fs_info, btrfs_op(orig_bio),
-				      start_sector << 9, submit_len, &geom);
-		if (ret)
-			goto out_err;
+		free_extent_map(em);
 	} while (submit_len > 0);
+	return;
 
-submit:
-	status = btrfs_submit_dio_bio(bio, inode, file_offset, async_submit);
-	if (!status)
-		return 0;
-
-	bio_put(bio);
+out_err_em:
+	free_extent_map(em);
 out_err:
-	dip->errors = 1;
-	/*
-	 * Before atomic variable goto zero, we must  make sure dip->errors is
-	 * perceived to be set. This ordering is ensured by the fact that an
-	 * atomic operations with a return value are fully ordered as per
-	 * atomic_t.txt
-	 */
-	if (atomic_dec_and_test(&dip->pending_bios))
-		bio_io_error(dip->orig_bio);
-
-	/* bio_end_io() will handle error, so we needn't return it */
-	return 0;
+	dio_bio->bi_status = status;
+	btrfs_dio_private_put(dip);
 }
 
-static void btrfs_submit_direct(struct bio *dio_bio, struct inode *inode,
-				loff_t file_offset)
-{
-	struct btrfs_dio_private *dip = NULL;
-	struct bio *bio = NULL;
-	struct btrfs_io_bio *io_bio;
-	bool write = (bio_op(dio_bio) == REQ_OP_WRITE);
-	int ret = 0;
-
-	bio = btrfs_bio_clone(dio_bio);
-
-	dip = kzalloc(sizeof(*dip), GFP_NOFS);
-	if (!dip) {
-		ret = -ENOMEM;
-		goto free_ordered;
-	}
-
-	dip->private = dio_bio->bi_private;
-	dip->inode = inode;
-	dip->logical_offset = file_offset;
-	dip->bytes = dio_bio->bi_iter.bi_size;
-	dip->disk_bytenr = (u64)dio_bio->bi_iter.bi_sector << 9;
-	bio->bi_private = dip;
-	dip->orig_bio = bio;
-	dip->dio_bio = dio_bio;
-	atomic_set(&dip->pending_bios, 0);
-	io_bio = btrfs_io_bio(bio);
-	io_bio->logical = file_offset;
-
-	if (write) {
-		bio->bi_end_io = btrfs_endio_direct_write;
-	} else {
-		bio->bi_end_io = btrfs_endio_direct_read;
-		dip->subio_endio = btrfs_subio_endio_read;
-	}
-
-	/*
-	 * Reset the range for unsubmitted ordered extents (to a 0 length range)
-	 * even if we fail to submit a bio, because in such case we do the
-	 * corresponding error handling below and it must not be done a second
-	 * time by btrfs_direct_IO().
-	 */
-	if (write) {
-		struct btrfs_dio_data *dio_data = current->journal_info;
-
-		dio_data->unsubmitted_oe_range_end = dip->logical_offset +
-			dip->bytes;
-		dio_data->unsubmitted_oe_range_start =
-			dio_data->unsubmitted_oe_range_end;
-	}
-
-	ret = btrfs_submit_direct_hook(dip);
-	if (!ret)
-		return;
+static const struct iomap_ops btrfs_dio_iomap_ops = {
+	.iomap_begin            = btrfs_dio_iomap_begin,
+	.iomap_end              = btrfs_dio_iomap_end,
+};
 
-	btrfs_io_bio_free_csum(io_bio);
+static const struct iomap_dio_ops btrfs_dio_ops = {
+	.submit_io		= btrfs_submit_direct,
+	.bio_set		= &btrfs_dio_bioset,
+};
 
-free_ordered:
-	/*
-	 * If we arrived here it means either we failed to submit the dip
-	 * or we either failed to clone the dio_bio or failed to allocate the
-	 * dip. If we cloned the dio_bio and allocated the dip, we can just
-	 * call bio_endio against our io_bio so that we get proper resource
-	 * cleanup if we fail to submit the dip, otherwise, we must do the
-	 * same as btrfs_endio_direct_[write|read] because we can't call these
-	 * callbacks - they require an allocated dip and a clone of dio_bio.
-	 */
-	if (bio && dip) {
-		bio_io_error(bio);
-		/*
-		 * The end io callbacks free our dip, do the final put on bio
-		 * and all the cleanup and final put for dio_bio (through
-		 * dio_end_io()).
-		 */
-		dip = NULL;
-		bio = NULL;
-	} else {
-		if (write)
-			__endio_write_update_ordered(inode,
-						file_offset,
-						dio_bio->bi_iter.bi_size,
-						false);
-		else
-			unlock_extent(&BTRFS_I(inode)->io_tree, file_offset,
-			      file_offset + dio_bio->bi_iter.bi_size - 1);
+ssize_t btrfs_dio_read(struct kiocb *iocb, struct iov_iter *iter, size_t done_before)
+{
+	struct btrfs_dio_data data;
 
-		dio_bio->bi_status = BLK_STS_IOERR;
-		/*
-		 * Releases and cleans up our dio_bio, no need to bio_put()
-		 * nor bio_endio()/bio_io_error() against dio_bio.
-		 */
-		dio_end_io(dio_bio);
-	}
-	if (bio)
-		bio_put(bio);
-	kfree(dip);
+	return iomap_dio_rw(iocb, iter, &btrfs_dio_iomap_ops, &btrfs_dio_ops,
+			    IOMAP_DIO_PARTIAL, &data, done_before);
 }
 
-static ssize_t check_direct_IO(struct btrfs_fs_info *fs_info,
-			       const struct iov_iter *iter, loff_t offset)
+struct iomap_dio *btrfs_dio_write(struct kiocb *iocb, struct iov_iter *iter,
+				  size_t done_before)
 {
-	int seg;
-	int i;
-	unsigned int blocksize_mask = fs_info->sectorsize - 1;
-	ssize_t retval = -EINVAL;
+	struct btrfs_dio_data data;
 
-	if (offset & blocksize_mask)
-		goto out;
-
-	if (iov_iter_alignment(iter) & blocksize_mask)
-		goto out;
-
-	/* If this is a write we don't need to check anymore */
-	if (iov_iter_rw(iter) != READ || !iter_is_iovec(iter))
-		return 0;
-	/*
-	 * Check to make sure we don't have duplicate iov_base's in this
-	 * iovec, if so return EINVAL, otherwise we'll get csum errors
-	 * when reading back.
-	 */
-	for (seg = 0; seg < iter->nr_segs; seg++) {
-		for (i = seg + 1; i < iter->nr_segs; i++) {
-			if (iter->iov[seg].iov_base == iter->iov[i].iov_base)
-				goto out;
-		}
-	}
-	retval = 0;
-out:
-	return retval;
+	return __iomap_dio_rw(iocb, iter, &btrfs_dio_iomap_ops, &btrfs_dio_ops,
+			    IOMAP_DIO_PARTIAL, &data, done_before);
 }
 
-static ssize_t btrfs_direct_IO(struct kiocb *iocb, struct iov_iter *iter)
+static int btrfs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
+			u64 start, u64 len)
 {
-	struct file *file = iocb->ki_filp;
-	struct inode *inode = file->f_mapping->host;
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	struct btrfs_dio_data dio_data = { 0 };
-	struct extent_changeset *data_reserved = NULL;
-	loff_t offset = iocb->ki_pos;
-	size_t count = 0;
-	int flags = 0;
-	bool wakeup = true;
-	bool relock = false;
-	ssize_t ret;
-
-	if (check_direct_IO(fs_info, iter, offset))
-		return 0;
+	int	ret;
 
-	inode_dio_begin(inode);
+	ret = fiemap_prep(inode, fieinfo, start, &len, 0);
+	if (ret)
+		return ret;
 
 	/*
-	 * The generic stuff only does filemap_write_and_wait_range, which
-	 * isn't enough if we've written compressed pages to this area, so
-	 * we need to flush the dirty pages again to make absolutely sure
-	 * that any outstanding dirty pages are on disk.
+	 * fiemap_prep() called filemap_write_and_wait() for the whole possible
+	 * file range (0 to LLONG_MAX), but that is not enough if we have
+	 * compression enabled. The first filemap_fdatawrite_range() only kicks
+	 * in the compression of data (in an async thread) and will return
+	 * before the compression is done and writeback is started. A second
+	 * filemap_fdatawrite_range() is needed to wait for the compression to
+	 * complete and writeback to start. We also need to wait for ordered
+	 * extents to complete, because our fiemap implementation uses mainly
+	 * file extent items to list the extents, searching for extent maps
+	 * only for file ranges with holes or prealloc extents to figure out
+	 * if we have delalloc in those ranges.
 	 */
-	count = iov_iter_count(iter);
-	if (test_bit(BTRFS_INODE_HAS_ASYNC_EXTENT,
-		     &BTRFS_I(inode)->runtime_flags))
-		filemap_fdatawrite_range(inode->i_mapping, offset,
-					 offset + count - 1);
-
-	if (iov_iter_rw(iter) == WRITE) {
-		/*
-		 * If the write DIO is beyond the EOF, we need update
-		 * the isize, but it is protected by i_mutex. So we can
-		 * not unlock the i_mutex at this case.
-		 */
-		if (offset + count <= inode->i_size) {
-			dio_data.overwrite = 1;
-			inode_unlock(inode);
-			relock = true;
-		} else if (iocb->ki_flags & IOCB_NOWAIT) {
-			ret = -EAGAIN;
-			goto out;
-		}
-		ret = btrfs_delalloc_reserve_space(inode, &data_reserved,
-						   offset, count);
+	if (fieinfo->fi_flags & FIEMAP_FLAG_SYNC) {
+		ret = btrfs_wait_ordered_range(inode, 0, LLONG_MAX);
 		if (ret)
-			goto out;
-
-		/*
-		 * We need to know how many extents we reserved so that we can
-		 * do the accounting properly if we go over the number we
-		 * originally calculated.  Abuse current->journal_info for this.
-		 */
-		dio_data.reserve = round_up(count,
-					    fs_info->sectorsize);
-		dio_data.unsubmitted_oe_range_start = (u64)offset;
-		dio_data.unsubmitted_oe_range_end = (u64)offset;
-		current->journal_info = &dio_data;
-		down_read(&BTRFS_I(inode)->dio_sem);
-	} else if (test_bit(BTRFS_INODE_READDIO_NEED_LOCK,
-				     &BTRFS_I(inode)->runtime_flags)) {
-		inode_dio_end(inode);
-		flags = DIO_LOCKING | DIO_SKIP_HOLES;
-		wakeup = false;
-	}
-
-	ret = __blockdev_direct_IO(iocb, inode,
-				   fs_info->fs_devices->latest_bdev,
-				   iter, btrfs_get_blocks_direct, NULL,
-				   btrfs_submit_direct, flags);
-	if (iov_iter_rw(iter) == WRITE) {
-		up_read(&BTRFS_I(inode)->dio_sem);
-		current->journal_info = NULL;
-		if (ret < 0 && ret != -EIOCBQUEUED) {
-			if (dio_data.reserve)
-				btrfs_delalloc_release_space(inode, data_reserved,
-					offset, dio_data.reserve, true);
-			/*
-			 * On error we might have left some ordered extents
-			 * without submitting corresponding bios for them, so
-			 * cleanup them up to avoid other tasks getting them
-			 * and waiting for them to complete forever.
-			 */
-			if (dio_data.unsubmitted_oe_range_start <
-			    dio_data.unsubmitted_oe_range_end)
-				__endio_write_update_ordered(inode,
-					dio_data.unsubmitted_oe_range_start,
-					dio_data.unsubmitted_oe_range_end -
-					dio_data.unsubmitted_oe_range_start,
-					false);
-		} else if (ret >= 0 && (size_t)ret < count)
-			btrfs_delalloc_release_space(inode, data_reserved,
-					offset, count - (size_t)ret, true);
-		btrfs_delalloc_release_extents(BTRFS_I(inode), count);
+			return ret;
 	}
-out:
-	if (wakeup)
-		inode_dio_end(inode);
-	if (relock)
-		inode_lock(inode);
 
-	extent_changeset_free(data_reserved);
-	return ret;
+	return extent_fiemap(BTRFS_I(inode), fieinfo, start, len);
 }
 
-#define BTRFS_FIEMAP_FLAGS	(FIEMAP_FLAG_SYNC)
-
-static int btrfs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
-		__u64 start, __u64 len)
+static int btrfs_writepages(struct address_space *mapping,
+			    struct writeback_control *wbc)
 {
-	int	ret;
-
-	ret = fiemap_check_flags(fieinfo, BTRFS_FIEMAP_FLAGS);
-	if (ret)
-		return ret;
-
-	return extent_fiemap(inode, fieinfo, start, len);
+	return extent_writepages(mapping, wbc);
 }
 
-int btrfs_readpage(struct file *file, struct page *page)
+static void btrfs_readahead(struct readahead_control *rac)
 {
-	struct extent_io_tree *tree;
-	tree = &BTRFS_I(page->mapping->host)->io_tree;
-	return extent_read_full_page(tree, page, btrfs_get_extent, 0);
+	extent_readahead(rac);
 }
 
-static int btrfs_writepage(struct page *page, struct writeback_control *wbc)
+/*
+ * For release_folio() and invalidate_folio() we have a race window where
+ * folio_end_writeback() is called but the subpage spinlock is not yet released.
+ * If we continue to release/invalidate the page, we could cause use-after-free
+ * for subpage spinlock.  So this function is to spin and wait for subpage
+ * spinlock.
+ */
+static void wait_subpage_spinlock(struct page *page)
 {
-	struct inode *inode = page->mapping->host;
-	int ret;
+	struct btrfs_fs_info *fs_info = btrfs_sb(page->mapping->host->i_sb);
+	struct btrfs_subpage *subpage;
 
-	if (current->flags & PF_MEMALLOC) {
-		redirty_page_for_writepage(wbc, page);
-		unlock_page(page);
-		return 0;
-	}
+	if (!btrfs_is_subpage(fs_info, page))
+		return;
+
+	ASSERT(PagePrivate(page) && page->private);
+	subpage = (struct btrfs_subpage *)page->private;
 
 	/*
-	 * If we are under memory pressure we will call this directly from the
-	 * VM, we need to make sure we have the inode referenced for the ordered
-	 * extent.  If not just return like we didn't do anything.
+	 * This may look insane as we just acquire the spinlock and release it,
+	 * without doing anything.  But we just want to make sure no one is
+	 * still holding the subpage spinlock.
+	 * And since the page is not dirty nor writeback, and we have page
+	 * locked, the only possible way to hold a spinlock is from the endio
+	 * function to clear page writeback.
+	 *
+	 * Here we just acquire the spinlock so that all existing callers
+	 * should exit and we're safe to release/invalidate the page.
 	 */
-	if (!igrab(inode)) {
-		redirty_page_for_writepage(wbc, page);
-		return AOP_WRITEPAGE_ACTIVATE;
-	}
-	ret = extent_write_full_page(page, wbc);
-	btrfs_add_delayed_iput(inode);
-	return ret;
+	spin_lock_irq(&subpage->lock);
+	spin_unlock_irq(&subpage->lock);
 }
 
-static int btrfs_writepages(struct address_space *mapping,
-			    struct writeback_control *wbc)
+static bool __btrfs_release_folio(struct folio *folio, gfp_t gfp_flags)
 {
-	return extent_writepages(mapping, wbc);
+	int ret = try_release_extent_mapping(&folio->page, gfp_flags);
+
+	if (ret == 1) {
+		wait_subpage_spinlock(&folio->page);
+		clear_page_extent_mapped(&folio->page);
+	}
+	return ret;
 }
 
-static int
-btrfs_readpages(struct file *file, struct address_space *mapping,
-		struct list_head *pages, unsigned nr_pages)
+static bool btrfs_release_folio(struct folio *folio, gfp_t gfp_flags)
 {
-	return extent_readpages(mapping, pages, nr_pages);
+	if (folio_test_writeback(folio) || folio_test_dirty(folio))
+		return false;
+	return __btrfs_release_folio(folio, gfp_flags);
 }
 
-static int __btrfs_releasepage(struct page *page, gfp_t gfp_flags)
+#ifdef CONFIG_MIGRATION
+static int btrfs_migrate_folio(struct address_space *mapping,
+			     struct folio *dst, struct folio *src,
+			     enum migrate_mode mode)
 {
-	int ret = try_release_extent_mapping(page, gfp_flags);
-	if (ret == 1) {
-		ClearPagePrivate(page);
-		set_page_private(page, 0);
-		put_page(page);
+	int ret = filemap_migrate_folio(mapping, dst, src, mode);
+
+	if (ret != MIGRATEPAGE_SUCCESS)
+		return ret;
+
+	if (folio_test_ordered(src)) {
+		folio_clear_ordered(src);
+		folio_set_ordered(dst);
 	}
-	return ret;
-}
 
-static int btrfs_releasepage(struct page *page, gfp_t gfp_flags)
-{
-	if (PageWriteback(page) || PageDirty(page))
-		return 0;
-	return __btrfs_releasepage(page, gfp_flags);
+	return MIGRATEPAGE_SUCCESS;
 }
+#else
+#define btrfs_migrate_folio NULL
+#endif
 
-static void btrfs_invalidatepage(struct page *page, unsigned int offset,
-				 unsigned int length)
+static void btrfs_invalidate_folio(struct folio *folio, size_t offset,
+				 size_t length)
 {
-	struct inode *inode = page->mapping->host;
-	struct extent_io_tree *tree;
-	struct btrfs_ordered_extent *ordered;
+	struct btrfs_inode *inode = BTRFS_I(folio->mapping->host);
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
+	struct extent_io_tree *tree = &inode->io_tree;
 	struct extent_state *cached_state = NULL;
-	u64 page_start = page_offset(page);
-	u64 page_end = page_start + PAGE_SIZE - 1;
-	u64 start;
-	u64 end;
-	int inode_evicting = inode->i_state & I_FREEING;
+	u64 page_start = folio_pos(folio);
+	u64 page_end = page_start + folio_size(folio) - 1;
+	u64 cur;
+	int inode_evicting = inode->vfs_inode.i_state & I_FREEING;
 
 	/*
-	 * we have the page locked, so new writeback can't start,
-	 * and the dirty bit won't be cleared while we are here.
+	 * We have folio locked so no new ordered extent can be created on this
+	 * page, nor bio can be submitted for this folio.
+	 *
+	 * But already submitted bio can still be finished on this folio.
+	 * Furthermore, endio function won't skip folio which has Ordered
+	 * (Private2) already cleared, so it's possible for endio and
+	 * invalidate_folio to do the same ordered extent accounting twice
+	 * on one folio.
 	 *
-	 * Wait for IO on this page so that we can safely clear
-	 * the PagePrivate2 bit and do ordered accounting
+	 * So here we wait for any submitted bios to finish, so that we won't
+	 * do double ordered extent accounting on the same folio.
 	 */
-	wait_on_page_writeback(page);
+	folio_wait_writeback(folio);
+	wait_subpage_spinlock(&folio->page);
 
-	tree = &BTRFS_I(inode)->io_tree;
-	if (offset) {
-		btrfs_releasepage(page, GFP_NOFS);
+	/*
+	 * For subpage case, we have call sites like
+	 * btrfs_punch_hole_lock_range() which passes range not aligned to
+	 * sectorsize.
+	 * If the range doesn't cover the full folio, we don't need to and
+	 * shouldn't clear page extent mapped, as folio->private can still
+	 * record subpage dirty bits for other part of the range.
+	 *
+	 * For cases that invalidate the full folio even the range doesn't
+	 * cover the full folio, like invalidating the last folio, we're
+	 * still safe to wait for ordered extent to finish.
+	 */
+	if (!(offset == 0 && length == folio_size(folio))) {
+		btrfs_release_folio(folio, GFP_NOFS);
 		return;
 	}
 
 	if (!inode_evicting)
-		lock_extent_bits(tree, page_start, page_end, &cached_state);
-again:
-	start = page_start;
-	ordered = btrfs_lookup_ordered_range(BTRFS_I(inode), start,
-					page_end - start + 1);
-	if (ordered) {
-		end = min(page_end,
-			  ordered->file_offset + ordered->num_bytes - 1);
+		lock_extent(tree, page_start, page_end, &cached_state);
+
+	cur = page_start;
+	while (cur < page_end) {
+		struct btrfs_ordered_extent *ordered;
+		u64 range_end;
+		u32 range_len;
+		u32 extra_flags = 0;
+
+		ordered = btrfs_lookup_first_ordered_range(inode, cur,
+							   page_end + 1 - cur);
+		if (!ordered) {
+			range_end = page_end;
+			/*
+			 * No ordered extent covering this range, we are safe
+			 * to delete all extent states in the range.
+			 */
+			extra_flags = EXTENT_CLEAR_ALL_BITS;
+			goto next;
+		}
+		if (ordered->file_offset > cur) {
+			/*
+			 * There is a range between [cur, oe->file_offset) not
+			 * covered by any ordered extent.
+			 * We are safe to delete all extent states, and handle
+			 * the ordered extent in the next iteration.
+			 */
+			range_end = ordered->file_offset - 1;
+			extra_flags = EXTENT_CLEAR_ALL_BITS;
+			goto next;
+		}
+
+		range_end = min(ordered->file_offset + ordered->num_bytes - 1,
+				page_end);
+		ASSERT(range_end + 1 - cur < U32_MAX);
+		range_len = range_end + 1 - cur;
+		if (!btrfs_page_test_ordered(fs_info, &folio->page, cur, range_len)) {
+			/*
+			 * If Ordered (Private2) is cleared, it means endio has
+			 * already been executed for the range.
+			 * We can't delete the extent states as
+			 * btrfs_finish_ordered_io() may still use some of them.
+			 */
+			goto next;
+		}
+		btrfs_page_clear_ordered(fs_info, &folio->page, cur, range_len);
+
 		/*
-		 * IO on this page will never be started, so we need
-		 * to account for any ordered extents now
+		 * IO on this page will never be started, so we need to account
+		 * for any ordered extents now. Don't clear EXTENT_DELALLOC_NEW
+		 * here, must leave that up for the ordered extent completion.
+		 *
+		 * This will also unlock the range for incoming
+		 * btrfs_finish_ordered_io().
 		 */
 		if (!inode_evicting)
-			clear_extent_bit(tree, start, end,
-					 EXTENT_DELALLOC | EXTENT_DELALLOC_NEW |
+			clear_extent_bit(tree, cur, range_end,
+					 EXTENT_DELALLOC |
 					 EXTENT_LOCKED | EXTENT_DO_ACCOUNTING |
-					 EXTENT_DEFRAG, 1, 0, &cached_state);
-		/*
-		 * whoever cleared the private bit is responsible
-		 * for the finish_ordered_io
-		 */
-		if (TestClearPagePrivate2(page)) {
-			struct btrfs_ordered_inode_tree *tree;
-			u64 new_len;
+					 EXTENT_DEFRAG, &cached_state);
 
-			tree = &BTRFS_I(inode)->ordered_tree;
+		spin_lock_irq(&inode->ordered_tree.lock);
+		set_bit(BTRFS_ORDERED_TRUNCATED, &ordered->flags);
+		ordered->truncated_len = min(ordered->truncated_len,
+					     cur - ordered->file_offset);
+		spin_unlock_irq(&inode->ordered_tree.lock);
 
-			spin_lock_irq(&tree->lock);
-			set_bit(BTRFS_ORDERED_TRUNCATED, &ordered->flags);
-			new_len = start - ordered->file_offset;
-			if (new_len < ordered->truncated_len)
-				ordered->truncated_len = new_len;
-			spin_unlock_irq(&tree->lock);
-
-			if (btrfs_dec_test_ordered_pending(inode, &ordered,
-							   start,
-							   end - start + 1, 1))
-				btrfs_finish_ordered_io(ordered);
+		/*
+		 * If the ordered extent has finished, we're safe to delete all
+		 * the extent states of the range, otherwise
+		 * btrfs_finish_ordered_io() will get executed by endio for
+		 * other pages, so we can't delete extent states.
+		 */
+		if (btrfs_dec_test_ordered_pending(inode, &ordered,
+						   cur, range_end + 1 - cur)) {
+			btrfs_finish_ordered_io(ordered);
+			/*
+			 * The ordered extent has finished, now we're again
+			 * safe to delete all extent states of the range.
+			 */
+			extra_flags = EXTENT_CLEAR_ALL_BITS;
 		}
-		btrfs_put_ordered_extent(ordered);
+next:
+		if (ordered)
+			btrfs_put_ordered_extent(ordered);
+		/*
+		 * Qgroup reserved space handler
+		 * Sector(s) here will be either:
+		 *
+		 * 1) Already written to disk or bio already finished
+		 *    Then its QGROUP_RESERVED bit in io_tree is already cleared.
+		 *    Qgroup will be handled by its qgroup_record then.
+		 *    btrfs_qgroup_free_data() call will do nothing here.
+		 *
+		 * 2) Not written to disk yet
+		 *    Then btrfs_qgroup_free_data() call will clear the
+		 *    QGROUP_RESERVED bit of its io_tree, and free the qgroup
+		 *    reserved data space.
+		 *    Since the IO will never happen for this page.
+		 */
+		btrfs_qgroup_free_data(inode, NULL, cur, range_end + 1 - cur);
 		if (!inode_evicting) {
-			cached_state = NULL;
-			lock_extent_bits(tree, start, end,
-					 &cached_state);
+			clear_extent_bit(tree, cur, range_end, EXTENT_LOCKED |
+				 EXTENT_DELALLOC | EXTENT_UPTODATE |
+				 EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG |
+				 extra_flags, &cached_state);
 		}
-
-		start = end + 1;
-		if (start < page_end)
-			goto again;
+		cur = range_end + 1;
 	}
-
 	/*
-	 * Qgroup reserved space handler
-	 * Page here will be either
-	 * 1) Already written to disk
-	 *    In this case, its reserved space is released from data rsv map
-	 *    and will be freed by delayed_ref handler finally.
-	 *    So even we call qgroup_free_data(), it won't decrease reserved
-	 *    space.
-	 * 2) Not written to disk
-	 *    This means the reserved space should be freed here. However,
-	 *    if a truncate invalidates the page (by clearing PageDirty)
-	 *    and the page is accounted for while allocating extent
-	 *    in btrfs_check_data_free_space() we let delayed_ref to
-	 *    free the entire extent.
-	 */
-	if (PageDirty(page))
-		btrfs_qgroup_free_data(inode, NULL, page_start, PAGE_SIZE);
-	if (!inode_evicting) {
-		clear_extent_bit(tree, page_start, page_end, EXTENT_LOCKED |
-				 EXTENT_DELALLOC | EXTENT_DELALLOC_NEW |
-				 EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG, 1, 1,
-				 &cached_state);
-
-		__btrfs_releasepage(page, GFP_NOFS);
-	}
-
-	ClearPageChecked(page);
-	if (PagePrivate(page)) {
-		ClearPagePrivate(page);
-		set_page_private(page, 0);
-		put_page(page);
-	}
+	 * We have iterated through all ordered extents of the page, the page
+	 * should not have Ordered (Private2) anymore, or the above iteration
+	 * did something wrong.
+	 */
+	ASSERT(!folio_test_ordered(folio));
+	btrfs_page_clear_checked(fs_info, &folio->page, folio_pos(folio), folio_size(folio));
+	if (!inode_evicting)
+		__btrfs_release_folio(folio, GFP_NOFS);
+	clear_page_extent_mapped(&folio->page);
 }
 
 /*
@@ -8410,7 +8400,6 @@ vm_fault_t btrfs_page_mkwrite(struct vm_fault *vmf)
 	struct btrfs_ordered_extent *ordered;
 	struct extent_state *cached_state = NULL;
 	struct extent_changeset *data_reserved = NULL;
-	char *kaddr;
 	unsigned long zero_start;
 	loff_t size;
 	vm_fault_t ret;
@@ -8432,12 +8421,12 @@ vm_fault_t btrfs_page_mkwrite(struct vm_fault *vmf)
 	 * Reserving delalloc space after obtaining the page lock can lead to
 	 * deadlock. For example, if a dirty page is locked by this function
 	 * and the call to btrfs_delalloc_reserve_space() ends up triggering
-	 * dirty page write out, then the btrfs_writepage() function could
+	 * dirty page write out, then the btrfs_writepages() function could
 	 * end up waiting indefinitely to get a lock on the page currently
 	 * being processed by btrfs_page_mkwrite() function.
 	 */
-	ret2 = btrfs_delalloc_reserve_space(inode, &data_reserved, page_start,
-					   reserved_space);
+	ret2 = btrfs_delalloc_reserve_space(BTRFS_I(inode), &data_reserved,
+					    page_start, reserved_space);
 	if (!ret2) {
 		ret2 = file_update_time(vmf->vma->vm_file);
 		reserved = 1;
@@ -8451,6 +8440,7 @@ vm_fault_t btrfs_page_mkwrite(struct vm_fault *vmf)
 
 	ret = VM_FAULT_NOPAGE; /* make the VM retry the fault */
 again:
+	down_read(&BTRFS_I(inode)->i_mmap_lock);
 	lock_page(page);
 	size = i_size_read(inode);
 
@@ -8461,8 +8451,13 @@ again:
 	}
 	wait_on_page_writeback(page);
 
-	lock_extent_bits(io_tree, page_start, page_end, &cached_state);
-	set_page_extent_mapped(page);
+	lock_extent(io_tree, page_start, page_end, &cached_state);
+	ret2 = set_page_extent_mapped(page);
+	if (ret2 < 0) {
+		ret = vmf_error(ret2);
+		unlock_extent(io_tree, page_start, page_end, &cached_state);
+		goto out_unlock;
+	}
 
 	/*
 	 * we can't set the delalloc bits if there are pending ordered
@@ -8471,10 +8466,10 @@ again:
 	ordered = btrfs_lookup_ordered_range(BTRFS_I(inode), page_start,
 			PAGE_SIZE);
 	if (ordered) {
-		unlock_extent_cached(io_tree, page_start, page_end,
-				     &cached_state);
+		unlock_extent(io_tree, page_start, page_end, &cached_state);
 		unlock_page(page);
-		btrfs_start_ordered_extent(inode, ordered, 1);
+		up_read(&BTRFS_I(inode)->i_mmap_lock);
+		btrfs_start_ordered_extent(ordered, 1);
 		btrfs_put_ordered_extent(ordered);
 		goto again;
 	}
@@ -8484,9 +8479,9 @@ again:
 					  fs_info->sectorsize);
 		if (reserved_space < PAGE_SIZE) {
 			end = page_start + reserved_space - 1;
-			btrfs_delalloc_release_space(inode, data_reserved,
-					page_start, PAGE_SIZE - reserved_space,
-					true);
+			btrfs_delalloc_release_space(BTRFS_I(inode),
+					data_reserved, page_start,
+					PAGE_SIZE - reserved_space, true);
 		}
 	}
 
@@ -8499,13 +8494,12 @@ again:
 	 */
 	clear_extent_bit(&BTRFS_I(inode)->io_tree, page_start, end,
 			  EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING |
-			  EXTENT_DEFRAG, 0, 0, &cached_state);
+			  EXTENT_DEFRAG, &cached_state);
 
-	ret2 = btrfs_set_extent_delalloc(inode, page_start, end, 0,
+	ret2 = btrfs_set_extent_delalloc(BTRFS_I(inode), page_start, end, 0,
 					&cached_state);
 	if (ret2) {
-		unlock_extent_cached(io_tree, page_start, page_end,
-				     &cached_state);
+		unlock_extent(io_tree, page_start, page_end, &cached_state);
 		ret = VM_FAULT_SIGBUS;
 		goto out_unlock;
 	}
@@ -8516,21 +8510,17 @@ again:
 	else
 		zero_start = PAGE_SIZE;
 
-	if (zero_start != PAGE_SIZE) {
-		kaddr = kmap(page);
-		memset(kaddr + zero_start, 0, PAGE_SIZE - zero_start);
-		flush_dcache_page(page);
-		kunmap(page);
-	}
-	ClearPageChecked(page);
-	set_page_dirty(page);
-	SetPageUptodate(page);
+	if (zero_start != PAGE_SIZE)
+		memzero_page(page, zero_start, PAGE_SIZE - zero_start);
 
-	BTRFS_I(inode)->last_trans = fs_info->generation;
-	BTRFS_I(inode)->last_sub_trans = BTRFS_I(inode)->root->log_transid;
-	BTRFS_I(inode)->last_log_commit = BTRFS_I(inode)->root->last_log_commit;
+	btrfs_page_clear_checked(fs_info, page, page_start, PAGE_SIZE);
+	btrfs_page_set_dirty(fs_info, page, page_start, end + 1 - page_start);
+	btrfs_page_set_uptodate(fs_info, page, page_start, end + 1 - page_start);
 
-	unlock_extent_cached(io_tree, page_start, page_end, &cached_state);
+	btrfs_set_inode_last_sub_trans(BTRFS_I(inode));
+
+	unlock_extent(io_tree, page_start, page_end, &cached_state);
+	up_read(&BTRFS_I(inode)->i_mmap_lock);
 
 	btrfs_delalloc_release_extents(BTRFS_I(inode), PAGE_SIZE);
 	sb_end_pagefault(inode->i_sb);
@@ -8539,9 +8529,10 @@ again:
 
 out_unlock:
 	unlock_page(page);
+	up_read(&BTRFS_I(inode)->i_mmap_lock);
 out:
 	btrfs_delalloc_release_extents(BTRFS_I(inode), PAGE_SIZE);
-	btrfs_delalloc_release_space(inode, data_reserved, page_start,
+	btrfs_delalloc_release_space(BTRFS_I(inode), data_reserved, page_start,
 				     reserved_space, (ret != 0));
 out_noreserve:
 	sb_end_pagefault(inode->i_sb);
@@ -8551,6 +8542,12 @@ out_noreserve:
 
 static int btrfs_truncate(struct inode *inode, bool skip_writeback)
 {
+	struct btrfs_truncate_control control = {
+		.inode = BTRFS_I(inode),
+		.ino = btrfs_ino(BTRFS_I(inode)),
+		.min_type = BTRFS_EXTENT_DATA_KEY,
+		.clear_extent_range = true,
+	};
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	struct btrfs_root *root = BTRFS_I(inode)->root;
 	struct btrfs_block_rsv *rsv;
@@ -8598,7 +8595,7 @@ static int btrfs_truncate(struct inode *inode, bool skip_writeback)
 	if (!rsv)
 		return -ENOMEM;
 	rsv->size = min_size;
-	rsv->failfast = 1;
+	rsv->failfast = true;
 
 	/*
 	 * 1 for the truncate slack space
@@ -8615,25 +8612,38 @@ static int btrfs_truncate(struct inode *inode, bool skip_writeback)
 				      min_size, false);
 	BUG_ON(ret);
 
-	/*
-	 * So if we truncate and then write and fsync we normally would just
-	 * write the extents that changed, which is a problem if we need to
-	 * first truncate that entire inode.  So set this flag so we write out
-	 * all of the extents in the inode to the sync log so we're completely
-	 * safe.
-	 */
-	set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &BTRFS_I(inode)->runtime_flags);
 	trans->block_rsv = rsv;
 
 	while (1) {
-		ret = btrfs_truncate_inode_items(trans, root, inode,
-						 inode->i_size,
-						 BTRFS_EXTENT_DATA_KEY);
+		struct extent_state *cached_state = NULL;
+		const u64 new_size = inode->i_size;
+		const u64 lock_start = ALIGN_DOWN(new_size, fs_info->sectorsize);
+
+		control.new_size = new_size;
+		lock_extent(&BTRFS_I(inode)->io_tree, lock_start, (u64)-1,
+				 &cached_state);
+		/*
+		 * We want to drop from the next block forward in case this new
+		 * size is not block aligned since we will be keeping the last
+		 * block of the extent just the way it is.
+		 */
+		btrfs_drop_extent_map_range(BTRFS_I(inode),
+					    ALIGN(new_size, fs_info->sectorsize),
+					    (u64)-1, false);
+
+		ret = btrfs_truncate_inode_items(trans, root, &control);
+
+		inode_sub_bytes(inode, control.sub_bytes);
+		btrfs_inode_safe_disk_i_size_write(BTRFS_I(inode), control.last_size);
+
+		unlock_extent(&BTRFS_I(inode)->io_tree, lock_start, (u64)-1,
+			      &cached_state);
+
 		trans->block_rsv = &fs_info->trans_block_rsv;
 		if (ret != -ENOSPC && ret != -EAGAIN)
 			break;
 
-		ret = btrfs_update_inode(trans, root, inode);
+		ret = btrfs_update_inode(trans, root, BTRFS_I(inode));
 		if (ret)
 			break;
 
@@ -8647,7 +8657,7 @@ static int btrfs_truncate(struct inode *inode, bool skip_writeback)
 			break;
 		}
 
-		btrfs_block_rsv_release(fs_info, rsv, -1);
+		btrfs_block_rsv_release(fs_info, rsv, -1, NULL);
 		ret = btrfs_block_rsv_migrate(&fs_info->trans_block_rsv,
 					      rsv, min_size, false);
 		BUG_ON(ret);	/* shouldn't happen */
@@ -8656,15 +8666,15 @@ static int btrfs_truncate(struct inode *inode, bool skip_writeback)
 
 	/*
 	 * We can't call btrfs_truncate_block inside a trans handle as we could
-	 * deadlock with freeze, if we got NEED_TRUNCATE_BLOCK then we know
-	 * we've truncated everything except the last little bit, and can do
-	 * btrfs_truncate_block and then update the disk_i_size.
+	 * deadlock with freeze, if we got BTRFS_NEED_TRUNCATE_BLOCK then we
+	 * know we've truncated everything except the last little bit, and can
+	 * do btrfs_truncate_block and then update the disk_i_size.
 	 */
-	if (ret == NEED_TRUNCATE_BLOCK) {
+	if (ret == BTRFS_NEED_TRUNCATE_BLOCK) {
 		btrfs_end_transaction(trans);
 		btrfs_btree_balance_dirty(fs_info);
 
-		ret = btrfs_truncate_block(inode, inode->i_size, 0, 0);
+		ret = btrfs_truncate_block(BTRFS_I(inode), inode->i_size, 0, 0);
 		if (ret)
 			goto out;
 		trans = btrfs_start_transaction(root, 1);
@@ -8672,14 +8682,14 @@ static int btrfs_truncate(struct inode *inode, bool skip_writeback)
 			ret = PTR_ERR(trans);
 			goto out;
 		}
-		btrfs_ordered_update_i_size(inode, inode->i_size, NULL);
+		btrfs_inode_safe_disk_i_size_write(BTRFS_I(inode), 0);
 	}
 
 	if (trans) {
 		int ret2;
 
 		trans->block_rsv = &fs_info->trans_block_rsv;
-		ret2 = btrfs_update_inode(trans, root, inode);
+		ret2 = btrfs_update_inode(trans, root, BTRFS_I(inode));
 		if (ret2 && !ret)
 			ret = ret2;
 
@@ -8690,45 +8700,43 @@ static int btrfs_truncate(struct inode *inode, bool skip_writeback)
 	}
 out:
 	btrfs_free_block_rsv(fs_info, rsv);
+	/*
+	 * So if we truncate and then write and fsync we normally would just
+	 * write the extents that changed, which is a problem if we need to
+	 * first truncate that entire inode.  So set this flag so we write out
+	 * all of the extents in the inode to the sync log so we're completely
+	 * safe.
+	 *
+	 * If no extents were dropped or trimmed we don't need to force the next
+	 * fsync to truncate all the inode's items from the log and re-log them
+	 * all. This means the truncate operation did not change the file size,
+	 * or changed it to a smaller size but there was only an implicit hole
+	 * between the old i_size and the new i_size, and there were no prealloc
+	 * extents beyond i_size to drop.
+	 */
+	if (control.extents_found > 0)
+		btrfs_set_inode_full_sync(BTRFS_I(inode));
 
 	return ret;
 }
 
-/*
- * create a new subvolume directory/inode (helper for the ioctl).
- */
-int btrfs_create_subvol_root(struct btrfs_trans_handle *trans,
-			     struct btrfs_root *new_root,
-			     struct btrfs_root *parent_root,
-			     u64 new_dirid)
+struct inode *btrfs_new_subvol_inode(struct user_namespace *mnt_userns,
+				     struct inode *dir)
 {
 	struct inode *inode;
-	int err;
-	u64 index = 0;
-
-	inode = btrfs_new_inode(trans, new_root, NULL, "..", 2,
-				new_dirid, new_dirid,
-				S_IFDIR | (~current_umask() & S_IRWXUGO),
-				&index);
-	if (IS_ERR(inode))
-		return PTR_ERR(inode);
-	inode->i_op = &btrfs_dir_inode_operations;
-	inode->i_fop = &btrfs_dir_file_operations;
-
-	set_nlink(inode, 1);
-	btrfs_i_size_write(BTRFS_I(inode), 0);
-	unlock_new_inode(inode);
 
-	err = btrfs_subvol_inherit_props(trans, new_root, parent_root);
-	if (err)
-		btrfs_err(new_root->fs_info,
-			  "error inheriting subvolume %llu properties: %d",
-			  new_root->root_key.objectid, err);
-
-	err = btrfs_update_inode(trans, new_root, inode);
-
-	iput(inode);
-	return err;
+	inode = new_inode(dir->i_sb);
+	if (inode) {
+		/*
+		 * Subvolumes don't inherit the sgid bit or the parent's gid if
+		 * the parent's sgid bit is set. This is probably a bug.
+		 */
+		inode_init_owner(mnt_userns, inode, NULL,
+				 S_IFDIR | (~current_umask() & S_IRWXUGO));
+		inode->i_op = &btrfs_dir_inode_operations;
+		inode->i_fop = &btrfs_dir_file_operations;
+	}
+	return inode;
 }
 
 struct inode *btrfs_alloc_inode(struct super_block *sb)
@@ -8737,7 +8745,7 @@ struct inode *btrfs_alloc_inode(struct super_block *sb)
 	struct btrfs_inode *ei;
 	struct inode *inode;
 
-	ei = kmem_cache_alloc(btrfs_inode_cachep, GFP_KERNEL);
+	ei = alloc_inode_sb(sb, btrfs_inode_cachep, GFP_KERNEL);
 	if (!ei)
 		return NULL;
 
@@ -8751,13 +8759,16 @@ struct inode *btrfs_alloc_inode(struct super_block *sb)
 	ei->defrag_bytes = 0;
 	ei->disk_i_size = 0;
 	ei->flags = 0;
+	ei->ro_flags = 0;
 	ei->csum_bytes = 0;
 	ei->index_cnt = (u64)-1;
 	ei->dir_index = 0;
 	ei->last_unlink_trans = 0;
+	ei->last_reflink_trans = 0;
 	ei->last_log_commit = 0;
 
 	spin_lock_init(&ei->lock);
+	spin_lock_init(&ei->io_failure_lock);
 	ei->outstanding_extents = 0;
 	if (sb->s_magic != BTRFS_TEST_MAGIC)
 		btrfs_init_metadata_block_rsv(fs_info, &ei->block_rsv,
@@ -8774,17 +8785,16 @@ struct inode *btrfs_alloc_inode(struct super_block *sb)
 	inode = &ei->vfs_inode;
 	extent_map_tree_init(&ei->extent_tree);
 	extent_io_tree_init(fs_info, &ei->io_tree, IO_TREE_INODE_IO, inode);
-	extent_io_tree_init(fs_info, &ei->io_failure_tree,
-			    IO_TREE_INODE_IO_FAILURE, inode);
-	ei->io_tree.track_uptodate = true;
-	ei->io_failure_tree.track_uptodate = true;
+	extent_io_tree_init(fs_info, &ei->file_extent_tree,
+			    IO_TREE_INODE_FILE_EXTENT, NULL);
+	ei->io_failure_tree = RB_ROOT;
 	atomic_set(&ei->sync_writers, 0);
 	mutex_init(&ei->log_mutex);
 	btrfs_ordered_inode_tree_init(&ei->ordered_tree);
 	INIT_LIST_HEAD(&ei->delalloc_inodes);
 	INIT_LIST_HEAD(&ei->delayed_iput);
 	RB_CLEAR_NODE(&ei->rb_node);
-	init_rwsem(&ei->dio_sem);
+	init_rwsem(&ei->i_mmap_lock);
 
 	return inode;
 }
@@ -8792,7 +8802,7 @@ struct inode *btrfs_alloc_inode(struct super_block *sb)
 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
 void btrfs_test_destroy_inode(struct inode *inode)
 {
-	btrfs_drop_extent_cache(BTRFS_I(inode), 0, (u64)-1, 0);
+	btrfs_drop_extent_map_range(BTRFS_I(inode), 0, (u64)-1, false);
 	kmem_cache_free(btrfs_inode_cachep, BTRFS_I(inode));
 }
 #endif
@@ -8802,21 +8812,24 @@ void btrfs_free_inode(struct inode *inode)
 	kmem_cache_free(btrfs_inode_cachep, BTRFS_I(inode));
 }
 
-void btrfs_destroy_inode(struct inode *inode)
+void btrfs_destroy_inode(struct inode *vfs_inode)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	struct btrfs_ordered_extent *ordered;
-	struct btrfs_root *root = BTRFS_I(inode)->root;
+	struct btrfs_inode *inode = BTRFS_I(vfs_inode);
+	struct btrfs_root *root = inode->root;
+	bool freespace_inode;
 
-	WARN_ON(!hlist_empty(&inode->i_dentry));
-	WARN_ON(inode->i_data.nrpages);
-	WARN_ON(BTRFS_I(inode)->block_rsv.reserved);
-	WARN_ON(BTRFS_I(inode)->block_rsv.size);
-	WARN_ON(BTRFS_I(inode)->outstanding_extents);
-	WARN_ON(BTRFS_I(inode)->delalloc_bytes);
-	WARN_ON(BTRFS_I(inode)->new_delalloc_bytes);
-	WARN_ON(BTRFS_I(inode)->csum_bytes);
-	WARN_ON(BTRFS_I(inode)->defrag_bytes);
+	WARN_ON(!hlist_empty(&vfs_inode->i_dentry));
+	WARN_ON(vfs_inode->i_data.nrpages);
+	WARN_ON(inode->block_rsv.reserved);
+	WARN_ON(inode->block_rsv.size);
+	WARN_ON(inode->outstanding_extents);
+	if (!S_ISDIR(vfs_inode->i_mode)) {
+		WARN_ON(inode->delalloc_bytes);
+		WARN_ON(inode->new_delalloc_bytes);
+	}
+	WARN_ON(inode->csum_bytes);
+	WARN_ON(inode->defrag_bytes);
 
 	/*
 	 * This can happen where we create an inode, but somebody else also
@@ -8826,14 +8839,24 @@ void btrfs_destroy_inode(struct inode *inode)
 	if (!root)
 		return;
 
+	/*
+	 * If this is a free space inode do not take the ordered extents lockdep
+	 * map.
+	 */
+	freespace_inode = btrfs_is_free_space_inode(inode);
+
 	while (1) {
 		ordered = btrfs_lookup_first_ordered_extent(inode, (u64)-1);
 		if (!ordered)
 			break;
 		else {
-			btrfs_err(fs_info,
+			btrfs_err(root->fs_info,
 				  "found ordered extent %llu %llu on inode cleanup",
 				  ordered->file_offset, ordered->num_bytes);
+
+			if (!freespace_inode)
+				btrfs_lockdep_acquire(root->fs_info, btrfs_ordered_extent);
+
 			btrfs_remove_ordered_extent(inode, ordered);
 			btrfs_put_ordered_extent(ordered);
 			btrfs_put_ordered_extent(ordered);
@@ -8841,7 +8864,9 @@ void btrfs_destroy_inode(struct inode *inode)
 	}
 	btrfs_qgroup_check_reserved_leak(inode);
 	inode_tree_del(inode);
-	btrfs_drop_extent_cache(BTRFS_I(inode), 0, (u64)-1, 0);
+	btrfs_drop_extent_map_range(inode, 0, (u64)-1, false);
+	btrfs_inode_clear_file_extent_range(inode, 0, (u64)-1);
+	btrfs_put_root(inode->root);
 }
 
 int btrfs_drop_inode(struct inode *inode)
@@ -8860,7 +8885,7 @@ int btrfs_drop_inode(struct inode *inode)
 
 static void init_once(void *foo)
 {
-	struct btrfs_inode *ei = (struct btrfs_inode *) foo;
+	struct btrfs_inode *ei = foo;
 
 	inode_init_once(&ei->vfs_inode);
 }
@@ -8872,6 +8897,7 @@ void __cold btrfs_destroy_cachep(void)
 	 * destroy cache.
 	 */
 	rcu_barrier();
+	bioset_exit(&btrfs_dio_bioset);
 	kmem_cache_destroy(btrfs_inode_cachep);
 	kmem_cache_destroy(btrfs_trans_handle_cachep);
 	kmem_cache_destroy(btrfs_path_cachep);
@@ -8908,23 +8934,31 @@ int __init btrfs_init_cachep(void)
 
 	btrfs_free_space_bitmap_cachep = kmem_cache_create("btrfs_free_space_bitmap",
 							PAGE_SIZE, PAGE_SIZE,
-							SLAB_RED_ZONE, NULL);
+							SLAB_MEM_SPREAD, NULL);
 	if (!btrfs_free_space_bitmap_cachep)
 		goto fail;
 
+	if (bioset_init(&btrfs_dio_bioset, BIO_POOL_SIZE,
+			offsetof(struct btrfs_dio_private, bio),
+			BIOSET_NEED_BVECS))
+		goto fail;
+
 	return 0;
 fail:
 	btrfs_destroy_cachep();
 	return -ENOMEM;
 }
 
-static int btrfs_getattr(const struct path *path, struct kstat *stat,
+static int btrfs_getattr(struct user_namespace *mnt_userns,
+			 const struct path *path, struct kstat *stat,
 			 u32 request_mask, unsigned int flags)
 {
 	u64 delalloc_bytes;
+	u64 inode_bytes;
 	struct inode *inode = d_inode(path->dentry);
 	u32 blocksize = inode->i_sb->s_blocksize;
 	u32 bi_flags = BTRFS_I(inode)->flags;
+	u32 bi_ro_flags = BTRFS_I(inode)->ro_flags;
 
 	stat->result_mask |= STATX_BTIME;
 	stat->btime.tv_sec = BTRFS_I(inode)->i_otime.tv_sec;
@@ -8937,19 +8971,22 @@ static int btrfs_getattr(const struct path *path, struct kstat *stat,
 		stat->attributes |= STATX_ATTR_IMMUTABLE;
 	if (bi_flags & BTRFS_INODE_NODUMP)
 		stat->attributes |= STATX_ATTR_NODUMP;
+	if (bi_ro_flags & BTRFS_INODE_RO_VERITY)
+		stat->attributes |= STATX_ATTR_VERITY;
 
 	stat->attributes_mask |= (STATX_ATTR_APPEND |
 				  STATX_ATTR_COMPRESSED |
 				  STATX_ATTR_IMMUTABLE |
 				  STATX_ATTR_NODUMP);
 
-	generic_fillattr(inode, stat);
+	generic_fillattr(mnt_userns, inode, stat);
 	stat->dev = BTRFS_I(inode)->root->anon_dev;
 
 	spin_lock(&BTRFS_I(inode)->lock);
 	delalloc_bytes = BTRFS_I(inode)->new_delalloc_bytes;
+	inode_bytes = inode_get_bytes(inode);
 	spin_unlock(&BTRFS_I(inode)->lock);
-	stat->blocks = (ALIGN(inode_get_bytes(inode), blocksize) +
+	stat->blocks = (ALIGN(inode_bytes, blocksize) +
 			ALIGN(delalloc_bytes, blocksize)) >> 9;
 	return 0;
 }
@@ -8961,53 +8998,79 @@ static int btrfs_rename_exchange(struct inode *old_dir,
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(old_dir->i_sb);
 	struct btrfs_trans_handle *trans;
+	unsigned int trans_num_items;
 	struct btrfs_root *root = BTRFS_I(old_dir)->root;
 	struct btrfs_root *dest = BTRFS_I(new_dir)->root;
 	struct inode *new_inode = new_dentry->d_inode;
 	struct inode *old_inode = old_dentry->d_inode;
 	struct timespec64 ctime = current_time(old_inode);
-	struct dentry *parent;
+	struct btrfs_rename_ctx old_rename_ctx;
+	struct btrfs_rename_ctx new_rename_ctx;
 	u64 old_ino = btrfs_ino(BTRFS_I(old_inode));
 	u64 new_ino = btrfs_ino(BTRFS_I(new_inode));
 	u64 old_idx = 0;
 	u64 new_idx = 0;
 	int ret;
-	bool root_log_pinned = false;
-	bool dest_log_pinned = false;
-	struct btrfs_log_ctx ctx_root;
-	struct btrfs_log_ctx ctx_dest;
-	bool sync_log_root = false;
-	bool sync_log_dest = false;
-	bool commit_transaction = false;
+	int ret2;
+	bool need_abort = false;
 
-	/* we only allow rename subvolume link between subvolumes */
-	if (old_ino != BTRFS_FIRST_FREE_OBJECTID && root != dest)
+	/*
+	 * For non-subvolumes allow exchange only within one subvolume, in the
+	 * same inode namespace. Two subvolumes (represented as directory) can
+	 * be exchanged as they're a logical link and have a fixed inode number.
+	 */
+	if (root != dest &&
+	    (old_ino != BTRFS_FIRST_FREE_OBJECTID ||
+	     new_ino != BTRFS_FIRST_FREE_OBJECTID))
 		return -EXDEV;
 
-	btrfs_init_log_ctx(&ctx_root, old_inode);
-	btrfs_init_log_ctx(&ctx_dest, new_inode);
-
 	/* close the race window with snapshot create/destroy ioctl */
 	if (old_ino == BTRFS_FIRST_FREE_OBJECTID ||
 	    new_ino == BTRFS_FIRST_FREE_OBJECTID)
 		down_read(&fs_info->subvol_sem);
 
 	/*
-	 * We want to reserve the absolute worst case amount of items.  So if
-	 * both inodes are subvols and we need to unlink them then that would
-	 * require 4 item modifications, but if they are both normal inodes it
-	 * would require 5 item modifications, so we'll assume their normal
-	 * inodes.  So 5 * 2 is 10, plus 2 for the new links, so 12 total items
-	 * should cover the worst case number of items we'll modify.
+	 * For each inode:
+	 * 1 to remove old dir item
+	 * 1 to remove old dir index
+	 * 1 to add new dir item
+	 * 1 to add new dir index
+	 * 1 to update parent inode
+	 *
+	 * If the parents are the same, we only need to account for one
 	 */
-	trans = btrfs_start_transaction(root, 12);
+	trans_num_items = (old_dir == new_dir ? 9 : 10);
+	if (old_ino == BTRFS_FIRST_FREE_OBJECTID) {
+		/*
+		 * 1 to remove old root ref
+		 * 1 to remove old root backref
+		 * 1 to add new root ref
+		 * 1 to add new root backref
+		 */
+		trans_num_items += 4;
+	} else {
+		/*
+		 * 1 to update inode item
+		 * 1 to remove old inode ref
+		 * 1 to add new inode ref
+		 */
+		trans_num_items += 3;
+	}
+	if (new_ino == BTRFS_FIRST_FREE_OBJECTID)
+		trans_num_items += 4;
+	else
+		trans_num_items += 3;
+	trans = btrfs_start_transaction(root, trans_num_items);
 	if (IS_ERR(trans)) {
 		ret = PTR_ERR(trans);
 		goto out_notrans;
 	}
 
-	if (dest != root)
-		btrfs_record_root_in_trans(trans, dest);
+	if (dest != root) {
+		ret = btrfs_record_root_in_trans(trans, dest);
+		if (ret)
+			goto out_fail;
+	}
 
 	/*
 	 * We need to find a free sequence number both in the source and
@@ -9028,8 +9091,6 @@ static int btrfs_rename_exchange(struct inode *old_dir,
 		/* force full log commit if subvolume involved. */
 		btrfs_set_log_full_commit(trans);
 	} else {
-		btrfs_pin_log_trans(root);
-		root_log_pinned = true;
 		ret = btrfs_insert_inode_ref(trans, dest,
 					     new_dentry->d_name.name,
 					     new_dentry->d_name.len,
@@ -9038,6 +9099,7 @@ static int btrfs_rename_exchange(struct inode *old_dir,
 					     old_idx);
 		if (ret)
 			goto out_fail;
+		need_abort = true;
 	}
 
 	/* And now for the dest. */
@@ -9045,16 +9107,17 @@ static int btrfs_rename_exchange(struct inode *old_dir,
 		/* force full log commit if subvolume involved. */
 		btrfs_set_log_full_commit(trans);
 	} else {
-		btrfs_pin_log_trans(dest);
-		dest_log_pinned = true;
 		ret = btrfs_insert_inode_ref(trans, root,
 					     old_dentry->d_name.name,
 					     old_dentry->d_name.len,
 					     new_ino,
 					     btrfs_ino(BTRFS_I(old_dir)),
 					     new_idx);
-		if (ret)
+		if (ret) {
+			if (need_abort)
+				btrfs_abort_transaction(trans, ret);
 			goto out_fail;
+		}
 	}
 
 	/* Update inode version and ctime/mtime. */
@@ -9062,8 +9125,10 @@ static int btrfs_rename_exchange(struct inode *old_dir,
 	inode_inc_iversion(new_dir);
 	inode_inc_iversion(old_inode);
 	inode_inc_iversion(new_inode);
-	old_dir->i_ctime = old_dir->i_mtime = ctime;
-	new_dir->i_ctime = new_dir->i_mtime = ctime;
+	old_dir->i_mtime = ctime;
+	old_dir->i_ctime = ctime;
+	new_dir->i_mtime = ctime;
+	new_dir->i_ctime = ctime;
 	old_inode->i_ctime = ctime;
 	new_inode->i_ctime = ctime;
 
@@ -9078,12 +9143,13 @@ static int btrfs_rename_exchange(struct inode *old_dir,
 	if (old_ino == BTRFS_FIRST_FREE_OBJECTID) {
 		ret = btrfs_unlink_subvol(trans, old_dir, old_dentry);
 	} else { /* src is an inode */
-		ret = __btrfs_unlink_inode(trans, root, BTRFS_I(old_dir),
+		ret = __btrfs_unlink_inode(trans, BTRFS_I(old_dir),
 					   BTRFS_I(old_dentry->d_inode),
 					   old_dentry->d_name.name,
-					   old_dentry->d_name.len);
+					   old_dentry->d_name.len,
+					   &old_rename_ctx);
 		if (!ret)
-			ret = btrfs_update_inode(trans, root, old_inode);
+			ret = btrfs_update_inode(trans, root, BTRFS_I(old_inode));
 	}
 	if (ret) {
 		btrfs_abort_transaction(trans, ret);
@@ -9094,12 +9160,13 @@ static int btrfs_rename_exchange(struct inode *old_dir,
 	if (new_ino == BTRFS_FIRST_FREE_OBJECTID) {
 		ret = btrfs_unlink_subvol(trans, new_dir, new_dentry);
 	} else { /* dest is an inode */
-		ret = __btrfs_unlink_inode(trans, dest, BTRFS_I(new_dir),
+		ret = __btrfs_unlink_inode(trans, BTRFS_I(new_dir),
 					   BTRFS_I(new_dentry->d_inode),
 					   new_dentry->d_name.name,
-					   new_dentry->d_name.len);
+					   new_dentry->d_name.len,
+					   &new_rename_ctx);
 		if (!ret)
-			ret = btrfs_update_inode(trans, dest, new_inode);
+			ret = btrfs_update_inode(trans, dest, BTRFS_I(new_inode));
 	}
 	if (ret) {
 		btrfs_abort_transaction(trans, ret);
@@ -9127,175 +9194,77 @@ static int btrfs_rename_exchange(struct inode *old_dir,
 	if (new_inode->i_nlink == 1)
 		BTRFS_I(new_inode)->dir_index = new_idx;
 
-	if (root_log_pinned) {
-		parent = new_dentry->d_parent;
-		ret = btrfs_log_new_name(trans, BTRFS_I(old_inode),
-					 BTRFS_I(old_dir), parent,
-					 false, &ctx_root);
-		if (ret == BTRFS_NEED_LOG_SYNC)
-			sync_log_root = true;
-		else if (ret == BTRFS_NEED_TRANS_COMMIT)
-			commit_transaction = true;
-		ret = 0;
+	/*
+	 * Now pin the logs of the roots. We do it to ensure that no other task
+	 * can sync the logs while we are in progress with the rename, because
+	 * that could result in an inconsistency in case any of the inodes that
+	 * are part of this rename operation were logged before.
+	 */
+	if (old_ino != BTRFS_FIRST_FREE_OBJECTID)
+		btrfs_pin_log_trans(root);
+	if (new_ino != BTRFS_FIRST_FREE_OBJECTID)
+		btrfs_pin_log_trans(dest);
+
+	/* Do the log updates for all inodes. */
+	if (old_ino != BTRFS_FIRST_FREE_OBJECTID)
+		btrfs_log_new_name(trans, old_dentry, BTRFS_I(old_dir),
+				   old_rename_ctx.index, new_dentry->d_parent);
+	if (new_ino != BTRFS_FIRST_FREE_OBJECTID)
+		btrfs_log_new_name(trans, new_dentry, BTRFS_I(new_dir),
+				   new_rename_ctx.index, old_dentry->d_parent);
+
+	/* Now unpin the logs. */
+	if (old_ino != BTRFS_FIRST_FREE_OBJECTID)
 		btrfs_end_log_trans(root);
-		root_log_pinned = false;
-	}
-	if (dest_log_pinned) {
-		if (!commit_transaction) {
-			parent = old_dentry->d_parent;
-			ret = btrfs_log_new_name(trans, BTRFS_I(new_inode),
-						 BTRFS_I(new_dir), parent,
-						 false, &ctx_dest);
-			if (ret == BTRFS_NEED_LOG_SYNC)
-				sync_log_dest = true;
-			else if (ret == BTRFS_NEED_TRANS_COMMIT)
-				commit_transaction = true;
-			ret = 0;
-		}
+	if (new_ino != BTRFS_FIRST_FREE_OBJECTID)
 		btrfs_end_log_trans(dest);
-		dest_log_pinned = false;
-	}
 out_fail:
-	/*
-	 * If we have pinned a log and an error happened, we unpin tasks
-	 * trying to sync the log and force them to fallback to a transaction
-	 * commit if the log currently contains any of the inodes involved in
-	 * this rename operation (to ensure we do not persist a log with an
-	 * inconsistent state for any of these inodes or leading to any
-	 * inconsistencies when replayed). If the transaction was aborted, the
-	 * abortion reason is propagated to userspace when attempting to commit
-	 * the transaction. If the log does not contain any of these inodes, we
-	 * allow the tasks to sync it.
-	 */
-	if (ret && (root_log_pinned || dest_log_pinned)) {
-		if (btrfs_inode_in_log(BTRFS_I(old_dir), fs_info->generation) ||
-		    btrfs_inode_in_log(BTRFS_I(new_dir), fs_info->generation) ||
-		    btrfs_inode_in_log(BTRFS_I(old_inode), fs_info->generation) ||
-		    (new_inode &&
-		     btrfs_inode_in_log(BTRFS_I(new_inode), fs_info->generation)))
-			btrfs_set_log_full_commit(trans);
-
-		if (root_log_pinned) {
-			btrfs_end_log_trans(root);
-			root_log_pinned = false;
-		}
-		if (dest_log_pinned) {
-			btrfs_end_log_trans(dest);
-			dest_log_pinned = false;
-		}
-	}
-	if (!ret && sync_log_root && !commit_transaction) {
-		ret = btrfs_sync_log(trans, BTRFS_I(old_inode)->root,
-				     &ctx_root);
-		if (ret)
-			commit_transaction = true;
-	}
-	if (!ret && sync_log_dest && !commit_transaction) {
-		ret = btrfs_sync_log(trans, BTRFS_I(new_inode)->root,
-				     &ctx_dest);
-		if (ret)
-			commit_transaction = true;
-	}
-	if (commit_transaction) {
-		/*
-		 * We may have set commit_transaction when logging the new name
-		 * in the destination root, in which case we left the source
-		 * root context in the list of log contextes. So make sure we
-		 * remove it to avoid invalid memory accesses, since the context
-		 * was allocated in our stack frame.
-		 */
-		if (sync_log_root) {
-			mutex_lock(&root->log_mutex);
-			list_del_init(&ctx_root.list);
-			mutex_unlock(&root->log_mutex);
-		}
-		ret = btrfs_commit_transaction(trans);
-	} else {
-		int ret2;
-
-		ret2 = btrfs_end_transaction(trans);
-		ret = ret ? ret : ret2;
-	}
+	ret2 = btrfs_end_transaction(trans);
+	ret = ret ? ret : ret2;
 out_notrans:
 	if (new_ino == BTRFS_FIRST_FREE_OBJECTID ||
 	    old_ino == BTRFS_FIRST_FREE_OBJECTID)
 		up_read(&fs_info->subvol_sem);
 
-	ASSERT(list_empty(&ctx_root.list));
-	ASSERT(list_empty(&ctx_dest.list));
-
 	return ret;
 }
 
-static int btrfs_whiteout_for_rename(struct btrfs_trans_handle *trans,
-				     struct btrfs_root *root,
-				     struct inode *dir,
-				     struct dentry *dentry)
+static struct inode *new_whiteout_inode(struct user_namespace *mnt_userns,
+					struct inode *dir)
 {
-	int ret;
 	struct inode *inode;
-	u64 objectid;
-	u64 index;
-
-	ret = btrfs_find_free_ino(root, &objectid);
-	if (ret)
-		return ret;
-
-	inode = btrfs_new_inode(trans, root, dir,
-				dentry->d_name.name,
-				dentry->d_name.len,
-				btrfs_ino(BTRFS_I(dir)),
-				objectid,
-				S_IFCHR | WHITEOUT_MODE,
-				&index);
 
-	if (IS_ERR(inode)) {
-		ret = PTR_ERR(inode);
-		return ret;
+	inode = new_inode(dir->i_sb);
+	if (inode) {
+		inode_init_owner(mnt_userns, inode, dir,
+				 S_IFCHR | WHITEOUT_MODE);
+		inode->i_op = &btrfs_special_inode_operations;
+		init_special_inode(inode, inode->i_mode, WHITEOUT_DEV);
 	}
-
-	inode->i_op = &btrfs_special_inode_operations;
-	init_special_inode(inode, inode->i_mode,
-		WHITEOUT_DEV);
-
-	ret = btrfs_init_inode_security(trans, inode, dir,
-				&dentry->d_name);
-	if (ret)
-		goto out;
-
-	ret = btrfs_add_nondir(trans, BTRFS_I(dir), dentry,
-				BTRFS_I(inode), 0, index);
-	if (ret)
-		goto out;
-
-	ret = btrfs_update_inode(trans, root, inode);
-out:
-	unlock_new_inode(inode);
-	if (ret)
-		inode_dec_link_count(inode);
-	iput(inode);
-
-	return ret;
+	return inode;
 }
 
-static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry,
-			   struct inode *new_dir, struct dentry *new_dentry,
-			   unsigned int flags)
+static int btrfs_rename(struct user_namespace *mnt_userns,
+			struct inode *old_dir, struct dentry *old_dentry,
+			struct inode *new_dir, struct dentry *new_dentry,
+			unsigned int flags)
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(old_dir->i_sb);
+	struct btrfs_new_inode_args whiteout_args = {
+		.dir = old_dir,
+		.dentry = old_dentry,
+	};
 	struct btrfs_trans_handle *trans;
 	unsigned int trans_num_items;
 	struct btrfs_root *root = BTRFS_I(old_dir)->root;
 	struct btrfs_root *dest = BTRFS_I(new_dir)->root;
 	struct inode *new_inode = d_inode(new_dentry);
 	struct inode *old_inode = d_inode(old_dentry);
+	struct btrfs_rename_ctx rename_ctx;
 	u64 index = 0;
 	int ret;
+	int ret2;
 	u64 old_ino = btrfs_ino(BTRFS_I(old_inode));
-	bool log_pinned = false;
-	struct btrfs_log_ctx ctx;
-	bool sync_log = false;
-	bool commit_transaction = false;
 
 	if (btrfs_ino(BTRFS_I(new_dir)) == BTRFS_EMPTY_SUBVOL_DIR_OBJECTID)
 		return -EPERM;
@@ -9339,31 +9308,67 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry,
 	if (new_inode && S_ISREG(old_inode->i_mode) && new_inode->i_size)
 		filemap_flush(old_inode->i_mapping);
 
-	/* close the racy window with snapshot create/destroy ioctl */
-	if (old_ino == BTRFS_FIRST_FREE_OBJECTID)
+	if (flags & RENAME_WHITEOUT) {
+		whiteout_args.inode = new_whiteout_inode(mnt_userns, old_dir);
+		if (!whiteout_args.inode)
+			return -ENOMEM;
+		ret = btrfs_new_inode_prepare(&whiteout_args, &trans_num_items);
+		if (ret)
+			goto out_whiteout_inode;
+	} else {
+		/* 1 to update the old parent inode. */
+		trans_num_items = 1;
+	}
+
+	if (old_ino == BTRFS_FIRST_FREE_OBJECTID) {
+		/* Close the race window with snapshot create/destroy ioctl */
 		down_read(&fs_info->subvol_sem);
+		/*
+		 * 1 to remove old root ref
+		 * 1 to remove old root backref
+		 * 1 to add new root ref
+		 * 1 to add new root backref
+		 */
+		trans_num_items += 4;
+	} else {
+		/*
+		 * 1 to update inode
+		 * 1 to remove old inode ref
+		 * 1 to add new inode ref
+		 */
+		trans_num_items += 3;
+	}
 	/*
-	 * We want to reserve the absolute worst case amount of items.  So if
-	 * both inodes are subvols and we need to unlink them then that would
-	 * require 4 item modifications, but if they are both normal inodes it
-	 * would require 5 item modifications, so we'll assume they are normal
-	 * inodes.  So 5 * 2 is 10, plus 1 for the new link, so 11 total items
-	 * should cover the worst case number of items we'll modify.
-	 * If our rename has the whiteout flag, we need more 5 units for the
-	 * new inode (1 inode item, 1 inode ref, 2 dir items and 1 xattr item
-	 * when selinux is enabled).
-	 */
-	trans_num_items = 11;
-	if (flags & RENAME_WHITEOUT)
+	 * 1 to remove old dir item
+	 * 1 to remove old dir index
+	 * 1 to add new dir item
+	 * 1 to add new dir index
+	 */
+	trans_num_items += 4;
+	/* 1 to update new parent inode if it's not the same as the old parent */
+	if (new_dir != old_dir)
+		trans_num_items++;
+	if (new_inode) {
+		/*
+		 * 1 to update inode
+		 * 1 to remove inode ref
+		 * 1 to remove dir item
+		 * 1 to remove dir index
+		 * 1 to possibly add orphan item
+		 */
 		trans_num_items += 5;
+	}
 	trans = btrfs_start_transaction(root, trans_num_items);
 	if (IS_ERR(trans)) {
 		ret = PTR_ERR(trans);
 		goto out_notrans;
 	}
 
-	if (dest != root)
-		btrfs_record_root_in_trans(trans, dest);
+	if (dest != root) {
+		ret = btrfs_record_root_in_trans(trans, dest);
+		if (ret)
+			goto out_fail;
+	}
 
 	ret = btrfs_set_inode_index(BTRFS_I(new_dir), &index);
 	if (ret)
@@ -9374,8 +9379,6 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry,
 		/* force full log commit if subvolume involved. */
 		btrfs_set_log_full_commit(trans);
 	} else {
-		btrfs_pin_log_trans(root);
-		log_pinned = true;
 		ret = btrfs_insert_inode_ref(trans, dest,
 					     new_dentry->d_name.name,
 					     new_dentry->d_name.len,
@@ -9388,9 +9391,11 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry,
 	inode_inc_iversion(old_dir);
 	inode_inc_iversion(new_dir);
 	inode_inc_iversion(old_inode);
-	old_dir->i_ctime = old_dir->i_mtime =
-	new_dir->i_ctime = new_dir->i_mtime =
-	old_inode->i_ctime = current_time(old_dir);
+	old_dir->i_mtime = current_time(old_dir);
+	old_dir->i_ctime = old_dir->i_mtime;
+	new_dir->i_mtime = old_dir->i_mtime;
+	new_dir->i_ctime = old_dir->i_mtime;
+	old_inode->i_ctime = old_dir->i_mtime;
 
 	if (old_dentry->d_parent != new_dentry->d_parent)
 		btrfs_record_unlink_dir(trans, BTRFS_I(old_dir),
@@ -9399,12 +9404,13 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry,
 	if (unlikely(old_ino == BTRFS_FIRST_FREE_OBJECTID)) {
 		ret = btrfs_unlink_subvol(trans, old_dir, old_dentry);
 	} else {
-		ret = __btrfs_unlink_inode(trans, root, BTRFS_I(old_dir),
+		ret = __btrfs_unlink_inode(trans, BTRFS_I(old_dir),
 					BTRFS_I(d_inode(old_dentry)),
 					old_dentry->d_name.name,
-					old_dentry->d_name.len);
+					old_dentry->d_name.len,
+					&rename_ctx);
 		if (!ret)
-			ret = btrfs_update_inode(trans, root, old_inode);
+			ret = btrfs_update_inode(trans, root, BTRFS_I(old_inode));
 	}
 	if (ret) {
 		btrfs_abort_transaction(trans, ret);
@@ -9419,7 +9425,7 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry,
 			ret = btrfs_unlink_subvol(trans, new_dir, new_dentry);
 			BUG_ON(new_inode->i_nlink == 0);
 		} else {
-			ret = btrfs_unlink_inode(trans, dest, BTRFS_I(new_dir),
+			ret = btrfs_unlink_inode(trans, BTRFS_I(new_dir),
 						 BTRFS_I(d_inode(new_dentry)),
 						 new_dentry->d_name.name,
 						 new_dentry->d_name.len);
@@ -9444,90 +9450,54 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry,
 	if (old_inode->i_nlink == 1)
 		BTRFS_I(old_inode)->dir_index = index;
 
-	if (log_pinned) {
-		struct dentry *parent = new_dentry->d_parent;
-
-		btrfs_init_log_ctx(&ctx, old_inode);
-		ret = btrfs_log_new_name(trans, BTRFS_I(old_inode),
-					 BTRFS_I(old_dir), parent,
-					 false, &ctx);
-		if (ret == BTRFS_NEED_LOG_SYNC)
-			sync_log = true;
-		else if (ret == BTRFS_NEED_TRANS_COMMIT)
-			commit_transaction = true;
-		ret = 0;
-		btrfs_end_log_trans(root);
-		log_pinned = false;
-	}
+	if (old_ino != BTRFS_FIRST_FREE_OBJECTID)
+		btrfs_log_new_name(trans, old_dentry, BTRFS_I(old_dir),
+				   rename_ctx.index, new_dentry->d_parent);
 
 	if (flags & RENAME_WHITEOUT) {
-		ret = btrfs_whiteout_for_rename(trans, root, old_dir,
-						old_dentry);
-
+		ret = btrfs_create_new_inode(trans, &whiteout_args);
 		if (ret) {
 			btrfs_abort_transaction(trans, ret);
 			goto out_fail;
+		} else {
+			unlock_new_inode(whiteout_args.inode);
+			iput(whiteout_args.inode);
+			whiteout_args.inode = NULL;
 		}
 	}
 out_fail:
-	/*
-	 * If we have pinned the log and an error happened, we unpin tasks
-	 * trying to sync the log and force them to fallback to a transaction
-	 * commit if the log currently contains any of the inodes involved in
-	 * this rename operation (to ensure we do not persist a log with an
-	 * inconsistent state for any of these inodes or leading to any
-	 * inconsistencies when replayed). If the transaction was aborted, the
-	 * abortion reason is propagated to userspace when attempting to commit
-	 * the transaction. If the log does not contain any of these inodes, we
-	 * allow the tasks to sync it.
-	 */
-	if (ret && log_pinned) {
-		if (btrfs_inode_in_log(BTRFS_I(old_dir), fs_info->generation) ||
-		    btrfs_inode_in_log(BTRFS_I(new_dir), fs_info->generation) ||
-		    btrfs_inode_in_log(BTRFS_I(old_inode), fs_info->generation) ||
-		    (new_inode &&
-		     btrfs_inode_in_log(BTRFS_I(new_inode), fs_info->generation)))
-			btrfs_set_log_full_commit(trans);
-
-		btrfs_end_log_trans(root);
-		log_pinned = false;
-	}
-	if (!ret && sync_log) {
-		ret = btrfs_sync_log(trans, BTRFS_I(old_inode)->root, &ctx);
-		if (ret)
-			commit_transaction = true;
-	} else if (sync_log) {
-		mutex_lock(&root->log_mutex);
-		list_del(&ctx.list);
-		mutex_unlock(&root->log_mutex);
-	}
-	if (commit_transaction) {
-		ret = btrfs_commit_transaction(trans);
-	} else {
-		int ret2;
-
-		ret2 = btrfs_end_transaction(trans);
-		ret = ret ? ret : ret2;
-	}
+	ret2 = btrfs_end_transaction(trans);
+	ret = ret ? ret : ret2;
 out_notrans:
 	if (old_ino == BTRFS_FIRST_FREE_OBJECTID)
 		up_read(&fs_info->subvol_sem);
-
+	if (flags & RENAME_WHITEOUT)
+		btrfs_new_inode_args_destroy(&whiteout_args);
+out_whiteout_inode:
+	if (flags & RENAME_WHITEOUT)
+		iput(whiteout_args.inode);
 	return ret;
 }
 
-static int btrfs_rename2(struct inode *old_dir, struct dentry *old_dentry,
-			 struct inode *new_dir, struct dentry *new_dentry,
-			 unsigned int flags)
+static int btrfs_rename2(struct user_namespace *mnt_userns, struct inode *old_dir,
+			 struct dentry *old_dentry, struct inode *new_dir,
+			 struct dentry *new_dentry, unsigned int flags)
 {
+	int ret;
+
 	if (flags & ~(RENAME_NOREPLACE | RENAME_EXCHANGE | RENAME_WHITEOUT))
 		return -EINVAL;
 
 	if (flags & RENAME_EXCHANGE)
-		return btrfs_rename_exchange(old_dir, old_dentry, new_dir,
-					  new_dentry);
+		ret = btrfs_rename_exchange(old_dir, old_dentry, new_dir,
+					    new_dentry);
+	else
+		ret = btrfs_rename(mnt_userns, old_dir, old_dentry, new_dir,
+				   new_dentry, flags);
+
+	btrfs_btree_balance_dirty(BTRFS_I(new_dir)->root->fs_info);
 
-	return btrfs_rename(old_dir, old_dentry, new_dir, new_dentry, flags);
+	return ret;
 }
 
 struct btrfs_delalloc_work {
@@ -9574,7 +9544,9 @@ static struct btrfs_delalloc_work *btrfs_alloc_delalloc_work(struct inode *inode
  * some fairly slow code that needs optimization. This walks the list
  * of all the inodes with pending delalloc and forces them to disk.
  */
-static int start_delalloc_inodes(struct btrfs_root *root, int nr, bool snapshot)
+static int start_delalloc_inodes(struct btrfs_root *root,
+				 struct writeback_control *wbc, bool snapshot,
+				 bool in_reclaim_context)
 {
 	struct btrfs_inode *binode;
 	struct inode *inode;
@@ -9582,6 +9554,7 @@ static int start_delalloc_inodes(struct btrfs_root *root, int nr, bool snapshot)
 	struct list_head works;
 	struct list_head splice;
 	int ret = 0;
+	bool full_flush = wbc->nr_to_write == LONG_MAX;
 
 	INIT_LIST_HEAD(&works);
 	INIT_LIST_HEAD(&splice);
@@ -9595,6 +9568,11 @@ static int start_delalloc_inodes(struct btrfs_root *root, int nr, bool snapshot)
 
 		list_move_tail(&binode->delalloc_inodes,
 			       &root->delalloc_inodes);
+
+		if (in_reclaim_context &&
+		    test_bit(BTRFS_INODE_NO_DELALLOC_FLUSH, &binode->runtime_flags))
+			continue;
+
 		inode = igrab(&binode->vfs_inode);
 		if (!inode) {
 			cond_resched_lock(&root->delalloc_lock);
@@ -9605,18 +9583,22 @@ static int start_delalloc_inodes(struct btrfs_root *root, int nr, bool snapshot)
 		if (snapshot)
 			set_bit(BTRFS_INODE_SNAPSHOT_FLUSH,
 				&binode->runtime_flags);
-		work = btrfs_alloc_delalloc_work(inode);
-		if (!work) {
-			iput(inode);
-			ret = -ENOMEM;
-			goto out;
+		if (full_flush) {
+			work = btrfs_alloc_delalloc_work(inode);
+			if (!work) {
+				iput(inode);
+				ret = -ENOMEM;
+				goto out;
+			}
+			list_add_tail(&work->list, &works);
+			btrfs_queue_work(root->fs_info->flush_workers,
+					 &work->work);
+		} else {
+			ret = filemap_fdatawrite_wbc(inode->i_mapping, wbc);
+			btrfs_add_delayed_iput(inode);
+			if (ret || wbc->nr_to_write <= 0)
+				goto out;
 		}
-		list_add_tail(&work->list, &works);
-		btrfs_queue_work(root->fs_info->flush_workers,
-				 &work->work);
-		ret++;
-		if (nr != -1 && ret >= nr)
-			goto out;
 		cond_resched();
 		spin_lock(&root->delalloc_lock);
 	}
@@ -9638,27 +9620,36 @@ out:
 	return ret;
 }
 
-int btrfs_start_delalloc_snapshot(struct btrfs_root *root)
+int btrfs_start_delalloc_snapshot(struct btrfs_root *root, bool in_reclaim_context)
 {
+	struct writeback_control wbc = {
+		.nr_to_write = LONG_MAX,
+		.sync_mode = WB_SYNC_NONE,
+		.range_start = 0,
+		.range_end = LLONG_MAX,
+	};
 	struct btrfs_fs_info *fs_info = root->fs_info;
-	int ret;
 
-	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state))
+	if (BTRFS_FS_ERROR(fs_info))
 		return -EROFS;
 
-	ret = start_delalloc_inodes(root, -1, true);
-	if (ret > 0)
-		ret = 0;
-	return ret;
+	return start_delalloc_inodes(root, &wbc, true, in_reclaim_context);
 }
 
-int btrfs_start_delalloc_roots(struct btrfs_fs_info *fs_info, int nr)
+int btrfs_start_delalloc_roots(struct btrfs_fs_info *fs_info, long nr,
+			       bool in_reclaim_context)
 {
+	struct writeback_control wbc = {
+		.nr_to_write = nr,
+		.sync_mode = WB_SYNC_NONE,
+		.range_start = 0,
+		.range_end = LLONG_MAX,
+	};
 	struct btrfs_root *root;
 	struct list_head splice;
 	int ret;
 
-	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state))
+	if (BTRFS_FS_ERROR(fs_info))
 		return -EROFS;
 
 	INIT_LIST_HEAD(&splice);
@@ -9666,24 +9657,26 @@ int btrfs_start_delalloc_roots(struct btrfs_fs_info *fs_info, int nr)
 	mutex_lock(&fs_info->delalloc_root_mutex);
 	spin_lock(&fs_info->delalloc_root_lock);
 	list_splice_init(&fs_info->delalloc_roots, &splice);
-	while (!list_empty(&splice) && nr) {
+	while (!list_empty(&splice)) {
+		/*
+		 * Reset nr_to_write here so we know that we're doing a full
+		 * flush.
+		 */
+		if (nr == LONG_MAX)
+			wbc.nr_to_write = LONG_MAX;
+
 		root = list_first_entry(&splice, struct btrfs_root,
 					delalloc_root);
-		root = btrfs_grab_fs_root(root);
+		root = btrfs_grab_root(root);
 		BUG_ON(!root);
 		list_move_tail(&root->delalloc_root,
 			       &fs_info->delalloc_roots);
 		spin_unlock(&fs_info->delalloc_root_lock);
 
-		ret = start_delalloc_inodes(root, nr, false);
-		btrfs_put_fs_root(root);
-		if (ret < 0)
+		ret = start_delalloc_inodes(root, &wbc, false, in_reclaim_context);
+		btrfs_put_root(root);
+		if (ret < 0 || wbc.nr_to_write <= 0)
 			goto out;
-
-		if (nr != -1) {
-			nr -= ret;
-			WARN_ON(nr < 0);
-		}
 		spin_lock(&fs_info->delalloc_root_lock);
 	}
 	spin_unlock(&fs_info->delalloc_root_lock);
@@ -9699,18 +9692,21 @@ out:
 	return ret;
 }
 
-static int btrfs_symlink(struct inode *dir, struct dentry *dentry,
-			 const char *symname)
+static int btrfs_symlink(struct user_namespace *mnt_userns, struct inode *dir,
+			 struct dentry *dentry, const char *symname)
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb);
 	struct btrfs_trans_handle *trans;
 	struct btrfs_root *root = BTRFS_I(dir)->root;
 	struct btrfs_path *path;
 	struct btrfs_key key;
-	struct inode *inode = NULL;
+	struct inode *inode;
+	struct btrfs_new_inode_args new_inode_args = {
+		.dir = dir,
+		.dentry = dentry,
+	};
+	unsigned int trans_num_items;
 	int err;
-	u64 objectid;
-	u64 index = 0;
 	int name_len;
 	int datasize;
 	unsigned long ptr;
@@ -9721,49 +9717,40 @@ static int btrfs_symlink(struct inode *dir, struct dentry *dentry,
 	if (name_len > BTRFS_MAX_INLINE_DATA_SIZE(fs_info))
 		return -ENAMETOOLONG;
 
-	/*
-	 * 2 items for inode item and ref
-	 * 2 items for dir items
-	 * 1 item for updating parent inode item
-	 * 1 item for the inline extent item
-	 * 1 item for xattr if selinux is on
-	 */
-	trans = btrfs_start_transaction(root, 7);
-	if (IS_ERR(trans))
-		return PTR_ERR(trans);
+	inode = new_inode(dir->i_sb);
+	if (!inode)
+		return -ENOMEM;
+	inode_init_owner(mnt_userns, inode, dir, S_IFLNK | S_IRWXUGO);
+	inode->i_op = &btrfs_symlink_inode_operations;
+	inode_nohighmem(inode);
+	inode->i_mapping->a_ops = &btrfs_aops;
+	btrfs_i_size_write(BTRFS_I(inode), name_len);
+	inode_set_bytes(inode, name_len);
 
-	err = btrfs_find_free_ino(root, &objectid);
+	new_inode_args.inode = inode;
+	err = btrfs_new_inode_prepare(&new_inode_args, &trans_num_items);
 	if (err)
-		goto out_unlock;
+		goto out_inode;
+	/* 1 additional item for the inline extent */
+	trans_num_items++;
 
-	inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name,
-				dentry->d_name.len, btrfs_ino(BTRFS_I(dir)),
-				objectid, S_IFLNK|S_IRWXUGO, &index);
-	if (IS_ERR(inode)) {
-		err = PTR_ERR(inode);
-		inode = NULL;
-		goto out_unlock;
+	trans = btrfs_start_transaction(root, trans_num_items);
+	if (IS_ERR(trans)) {
+		err = PTR_ERR(trans);
+		goto out_new_inode_args;
 	}
 
-	/*
-	* If the active LSM wants to access the inode during
-	* d_instantiate it needs these. Smack checks to see
-	* if the filesystem supports xattrs by looking at the
-	* ops vector.
-	*/
-	inode->i_fop = &btrfs_file_operations;
-	inode->i_op = &btrfs_file_inode_operations;
-	inode->i_mapping->a_ops = &btrfs_aops;
-	BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops;
-
-	err = btrfs_init_inode_security(trans, inode, dir, &dentry->d_name);
+	err = btrfs_create_new_inode(trans, &new_inode_args);
 	if (err)
-		goto out_unlock;
+		goto out;
 
 	path = btrfs_alloc_path();
 	if (!path) {
 		err = -ENOMEM;
-		goto out_unlock;
+		btrfs_abort_transaction(trans, err);
+		discard_new_inode(inode);
+		inode = NULL;
+		goto out;
 	}
 	key.objectid = btrfs_ino(BTRFS_I(inode));
 	key.offset = 0;
@@ -9772,8 +9759,11 @@ static int btrfs_symlink(struct inode *dir, struct dentry *dentry,
 	err = btrfs_insert_empty_item(trans, root, path, &key,
 				      datasize);
 	if (err) {
+		btrfs_abort_transaction(trans, err);
 		btrfs_free_path(path);
-		goto out_unlock;
+		discard_new_inode(inode);
+		inode = NULL;
+		goto out;
 	}
 	leaf = path->nodes[0];
 	ei = btrfs_item_ptr(leaf, path->slots[0],
@@ -9791,41 +9781,102 @@ static int btrfs_symlink(struct inode *dir, struct dentry *dentry,
 	btrfs_mark_buffer_dirty(leaf);
 	btrfs_free_path(path);
 
-	inode->i_op = &btrfs_symlink_inode_operations;
-	inode_nohighmem(inode);
-	inode_set_bytes(inode, name_len);
-	btrfs_i_size_write(BTRFS_I(inode), name_len);
-	err = btrfs_update_inode(trans, root, inode);
-	/*
-	 * Last step, add directory indexes for our symlink inode. This is the
-	 * last step to avoid extra cleanup of these indexes if an error happens
-	 * elsewhere above.
-	 */
-	if (!err)
-		err = btrfs_add_nondir(trans, BTRFS_I(dir), dentry,
-				BTRFS_I(inode), 0, index);
-	if (err)
-		goto out_unlock;
-
 	d_instantiate_new(dentry, inode);
-
-out_unlock:
+	err = 0;
+out:
 	btrfs_end_transaction(trans);
-	if (err && inode) {
-		inode_dec_link_count(inode);
-		discard_new_inode(inode);
-	}
 	btrfs_btree_balance_dirty(fs_info);
+out_new_inode_args:
+	btrfs_new_inode_args_destroy(&new_inode_args);
+out_inode:
+	if (err)
+		iput(inode);
 	return err;
 }
 
+static struct btrfs_trans_handle *insert_prealloc_file_extent(
+				       struct btrfs_trans_handle *trans_in,
+				       struct btrfs_inode *inode,
+				       struct btrfs_key *ins,
+				       u64 file_offset)
+{
+	struct btrfs_file_extent_item stack_fi;
+	struct btrfs_replace_extent_info extent_info;
+	struct btrfs_trans_handle *trans = trans_in;
+	struct btrfs_path *path;
+	u64 start = ins->objectid;
+	u64 len = ins->offset;
+	int qgroup_released;
+	int ret;
+
+	memset(&stack_fi, 0, sizeof(stack_fi));
+
+	btrfs_set_stack_file_extent_type(&stack_fi, BTRFS_FILE_EXTENT_PREALLOC);
+	btrfs_set_stack_file_extent_disk_bytenr(&stack_fi, start);
+	btrfs_set_stack_file_extent_disk_num_bytes(&stack_fi, len);
+	btrfs_set_stack_file_extent_num_bytes(&stack_fi, len);
+	btrfs_set_stack_file_extent_ram_bytes(&stack_fi, len);
+	btrfs_set_stack_file_extent_compression(&stack_fi, BTRFS_COMPRESS_NONE);
+	/* Encryption and other encoding is reserved and all 0 */
+
+	qgroup_released = btrfs_qgroup_release_data(inode, file_offset, len);
+	if (qgroup_released < 0)
+		return ERR_PTR(qgroup_released);
+
+	if (trans) {
+		ret = insert_reserved_file_extent(trans, inode,
+						  file_offset, &stack_fi,
+						  true, qgroup_released);
+		if (ret)
+			goto free_qgroup;
+		return trans;
+	}
+
+	extent_info.disk_offset = start;
+	extent_info.disk_len = len;
+	extent_info.data_offset = 0;
+	extent_info.data_len = len;
+	extent_info.file_offset = file_offset;
+	extent_info.extent_buf = (char *)&stack_fi;
+	extent_info.is_new_extent = true;
+	extent_info.update_times = true;
+	extent_info.qgroup_reserved = qgroup_released;
+	extent_info.insertions = 0;
+
+	path = btrfs_alloc_path();
+	if (!path) {
+		ret = -ENOMEM;
+		goto free_qgroup;
+	}
+
+	ret = btrfs_replace_file_extents(inode, path, file_offset,
+				     file_offset + len - 1, &extent_info,
+				     &trans);
+	btrfs_free_path(path);
+	if (ret)
+		goto free_qgroup;
+	return trans;
+
+free_qgroup:
+	/*
+	 * We have released qgroup data range at the beginning of the function,
+	 * and normally qgroup_released bytes will be freed when committing
+	 * transaction.
+	 * But if we error out early, we have to free what we have released
+	 * or we leak qgroup data reservation.
+	 */
+	btrfs_qgroup_free_refroot(inode->root->fs_info,
+			inode->root->root_key.objectid, qgroup_released,
+			BTRFS_QGROUP_RSV_DATA);
+	return ERR_PTR(ret);
+}
+
 static int __btrfs_prealloc_file_range(struct inode *inode, int mode,
 				       u64 start, u64 num_bytes, u64 min_size,
 				       loff_t actual_len, u64 *alloc_hint,
 				       struct btrfs_trans_handle *trans)
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
 	struct extent_map *em;
 	struct btrfs_root *root = BTRFS_I(inode)->root;
 	struct btrfs_key ins;
@@ -9841,14 +9892,6 @@ static int __btrfs_prealloc_file_range(struct inode *inode, int mode,
 	if (trans)
 		own_trans = false;
 	while (num_bytes > 0) {
-		if (own_trans) {
-			trans = btrfs_start_transaction(root, 3);
-			if (IS_ERR(trans)) {
-				ret = PTR_ERR(trans);
-				break;
-			}
-		}
-
 		cur_bytes = min_t(u64, num_bytes, SZ_256M);
 		cur_bytes = max(cur_bytes, min_size);
 		/*
@@ -9860,11 +9903,8 @@ static int __btrfs_prealloc_file_range(struct inode *inode, int mode,
 		cur_bytes = min(cur_bytes, last_alloc);
 		ret = btrfs_reserve_extent(root, cur_bytes, cur_bytes,
 				min_size, 0, *alloc_hint, &ins, 1, 0);
-		if (ret) {
-			if (own_trans)
-				btrfs_end_transaction(trans);
+		if (ret)
 			break;
-		}
 
 		/*
 		 * We've reserved this space, and thus converted it from
@@ -9874,30 +9914,29 @@ static int __btrfs_prealloc_file_range(struct inode *inode, int mode,
 		 * clear_offset by our extent size.
 		 */
 		clear_offset += ins.offset;
-		btrfs_dec_block_group_reservations(fs_info, ins.objectid);
 
 		last_alloc = ins.offset;
-		ret = insert_reserved_file_extent(trans, inode,
-						  cur_offset, ins.objectid,
-						  ins.offset, ins.offset,
-						  ins.offset, 0, 0, 0,
-						  BTRFS_FILE_EXTENT_PREALLOC);
-		if (ret) {
+		trans = insert_prealloc_file_extent(trans, BTRFS_I(inode),
+						    &ins, cur_offset);
+		/*
+		 * Now that we inserted the prealloc extent we can finally
+		 * decrement the number of reservations in the block group.
+		 * If we did it before, we could race with relocation and have
+		 * relocation miss the reserved extent, making it fail later.
+		 */
+		btrfs_dec_block_group_reservations(fs_info, ins.objectid);
+		if (IS_ERR(trans)) {
+			ret = PTR_ERR(trans);
 			btrfs_free_reserved_extent(fs_info, ins.objectid,
 						   ins.offset, 0);
-			btrfs_abort_transaction(trans, ret);
-			if (own_trans)
-				btrfs_end_transaction(trans);
 			break;
 		}
 
-		btrfs_drop_extent_cache(BTRFS_I(inode), cur_offset,
-					cur_offset + ins.offset -1, 0);
-
 		em = alloc_extent_map();
 		if (!em) {
-			set_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
-				&BTRFS_I(inode)->runtime_flags);
+			btrfs_drop_extent_map_range(BTRFS_I(inode), cur_offset,
+					    cur_offset + ins.offset - 1, false);
+			btrfs_set_inode_full_sync(BTRFS_I(inode));
 			goto next;
 		}
 
@@ -9911,16 +9950,7 @@ static int __btrfs_prealloc_file_range(struct inode *inode, int mode,
 		set_bit(EXTENT_FLAG_PREALLOC, &em->flags);
 		em->generation = trans->transid;
 
-		while (1) {
-			write_lock(&em_tree->lock);
-			ret = add_extent_mapping(em_tree, em, 1);
-			write_unlock(&em_tree->lock);
-			if (ret != -EEXIST)
-				break;
-			btrfs_drop_extent_cache(BTRFS_I(inode), cur_offset,
-						cur_offset + ins.offset - 1,
-						0);
-		}
+		ret = btrfs_replace_extent_map_range(BTRFS_I(inode), em, true);
 		free_extent_map(em);
 next:
 		num_bytes -= ins.offset;
@@ -9938,10 +9968,10 @@ next:
 			else
 				i_size = cur_offset;
 			i_size_write(inode, i_size);
-			btrfs_ordered_update_i_size(inode, i_size, NULL);
+			btrfs_inode_safe_disk_i_size_write(BTRFS_I(inode), 0);
 		}
 
-		ret = btrfs_update_inode(trans, root, inode);
+		ret = btrfs_update_inode(trans, root, BTRFS_I(inode));
 
 		if (ret) {
 			btrfs_abort_transaction(trans, ret);
@@ -9950,11 +9980,13 @@ next:
 			break;
 		}
 
-		if (own_trans)
+		if (own_trans) {
 			btrfs_end_transaction(trans);
+			trans = NULL;
+		}
 	}
 	if (clear_offset < end)
-		btrfs_free_reserved_data_space(inode, NULL, clear_offset,
+		btrfs_free_reserved_data_space(BTRFS_I(inode), NULL, clear_offset,
 			end - clear_offset + 1);
 	return ret;
 }
@@ -9977,12 +10009,8 @@ int btrfs_prealloc_file_range_trans(struct inode *inode,
 					   min_size, actual_len, alloc_hint, trans);
 }
 
-static int btrfs_set_page_dirty(struct page *page)
-{
-	return __set_page_dirty_nobuffers(page);
-}
-
-static int btrfs_permission(struct inode *inode, int mask)
+static int btrfs_permission(struct user_namespace *mnt_userns,
+			    struct inode *inode, int mask)
 {
 	struct btrfs_root *root = BTRFS_I(inode)->root;
 	umode_t mode = inode->i_mode;
@@ -9994,90 +10022,813 @@ static int btrfs_permission(struct inode *inode, int mask)
 		if (BTRFS_I(inode)->flags & BTRFS_INODE_READONLY)
 			return -EACCES;
 	}
-	return generic_permission(inode, mask);
+	return generic_permission(mnt_userns, inode, mask);
 }
 
-static int btrfs_tmpfile(struct inode *dir, struct dentry *dentry, umode_t mode)
+static int btrfs_tmpfile(struct user_namespace *mnt_userns, struct inode *dir,
+			 struct file *file, umode_t mode)
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb);
 	struct btrfs_trans_handle *trans;
 	struct btrfs_root *root = BTRFS_I(dir)->root;
-	struct inode *inode = NULL;
-	u64 objectid;
-	u64 index;
-	int ret = 0;
-
-	/*
-	 * 5 units required for adding orphan entry
-	 */
-	trans = btrfs_start_transaction(root, 5);
-	if (IS_ERR(trans))
-		return PTR_ERR(trans);
-
-	ret = btrfs_find_free_ino(root, &objectid);
-	if (ret)
-		goto out;
-
-	inode = btrfs_new_inode(trans, root, dir, NULL, 0,
-			btrfs_ino(BTRFS_I(dir)), objectid, mode, &index);
-	if (IS_ERR(inode)) {
-		ret = PTR_ERR(inode);
-		inode = NULL;
-		goto out;
-	}
+	struct inode *inode;
+	struct btrfs_new_inode_args new_inode_args = {
+		.dir = dir,
+		.dentry = file->f_path.dentry,
+		.orphan = true,
+	};
+	unsigned int trans_num_items;
+	int ret;
 
+	inode = new_inode(dir->i_sb);
+	if (!inode)
+		return -ENOMEM;
+	inode_init_owner(mnt_userns, inode, dir, mode);
 	inode->i_fop = &btrfs_file_operations;
 	inode->i_op = &btrfs_file_inode_operations;
-
 	inode->i_mapping->a_ops = &btrfs_aops;
-	BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops;
 
-	ret = btrfs_init_inode_security(trans, inode, dir, NULL);
+	new_inode_args.inode = inode;
+	ret = btrfs_new_inode_prepare(&new_inode_args, &trans_num_items);
 	if (ret)
-		goto out;
+		goto out_inode;
 
-	ret = btrfs_update_inode(trans, root, inode);
-	if (ret)
-		goto out;
-	ret = btrfs_orphan_add(trans, BTRFS_I(inode));
-	if (ret)
-		goto out;
+	trans = btrfs_start_transaction(root, trans_num_items);
+	if (IS_ERR(trans)) {
+		ret = PTR_ERR(trans);
+		goto out_new_inode_args;
+	}
+
+	ret = btrfs_create_new_inode(trans, &new_inode_args);
 
 	/*
-	 * We set number of links to 0 in btrfs_new_inode(), and here we set
-	 * it to 1 because d_tmpfile() will issue a warning if the count is 0,
-	 * through:
+	 * We set number of links to 0 in btrfs_create_new_inode(), and here we
+	 * set it to 1 because d_tmpfile() will issue a warning if the count is
+	 * 0, through:
 	 *
 	 *    d_tmpfile() -> inode_dec_link_count() -> drop_nlink()
 	 */
 	set_nlink(inode, 1);
-	d_tmpfile(dentry, inode);
-	unlock_new_inode(inode);
-	mark_inode_dirty(inode);
-out:
+
+	if (!ret) {
+		d_tmpfile(file, inode);
+		unlock_new_inode(inode);
+		mark_inode_dirty(inode);
+	}
+
 	btrfs_end_transaction(trans);
-	if (ret && inode)
-		discard_new_inode(inode);
 	btrfs_btree_balance_dirty(fs_info);
-	return ret;
+out_new_inode_args:
+	btrfs_new_inode_args_destroy(&new_inode_args);
+out_inode:
+	if (ret)
+		iput(inode);
+	return finish_open_simple(file, ret);
 }
 
-void btrfs_set_range_writeback(struct extent_io_tree *tree, u64 start, u64 end)
+void btrfs_set_range_writeback(struct btrfs_inode *inode, u64 start, u64 end)
 {
-	struct inode *inode = tree->private_data;
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
 	unsigned long index = start >> PAGE_SHIFT;
 	unsigned long end_index = end >> PAGE_SHIFT;
 	struct page *page;
+	u32 len;
 
+	ASSERT(end + 1 - start <= U32_MAX);
+	len = end + 1 - start;
 	while (index <= end_index) {
-		page = find_get_page(inode->i_mapping, index);
+		page = find_get_page(inode->vfs_inode.i_mapping, index);
 		ASSERT(page); /* Pages should be in the extent_io_tree */
-		set_page_writeback(page);
+
+		btrfs_page_set_writeback(fs_info, page, start, len);
 		put_page(page);
 		index++;
 	}
 }
 
+int btrfs_encoded_io_compression_from_extent(struct btrfs_fs_info *fs_info,
+					     int compress_type)
+{
+	switch (compress_type) {
+	case BTRFS_COMPRESS_NONE:
+		return BTRFS_ENCODED_IO_COMPRESSION_NONE;
+	case BTRFS_COMPRESS_ZLIB:
+		return BTRFS_ENCODED_IO_COMPRESSION_ZLIB;
+	case BTRFS_COMPRESS_LZO:
+		/*
+		 * The LZO format depends on the sector size. 64K is the maximum
+		 * sector size that we support.
+		 */
+		if (fs_info->sectorsize < SZ_4K || fs_info->sectorsize > SZ_64K)
+			return -EINVAL;
+		return BTRFS_ENCODED_IO_COMPRESSION_LZO_4K +
+		       (fs_info->sectorsize_bits - 12);
+	case BTRFS_COMPRESS_ZSTD:
+		return BTRFS_ENCODED_IO_COMPRESSION_ZSTD;
+	default:
+		return -EUCLEAN;
+	}
+}
+
+static ssize_t btrfs_encoded_read_inline(
+				struct kiocb *iocb,
+				struct iov_iter *iter, u64 start,
+				u64 lockend,
+				struct extent_state **cached_state,
+				u64 extent_start, size_t count,
+				struct btrfs_ioctl_encoded_io_args *encoded,
+				bool *unlocked)
+{
+	struct btrfs_inode *inode = BTRFS_I(file_inode(iocb->ki_filp));
+	struct btrfs_root *root = inode->root;
+	struct btrfs_fs_info *fs_info = root->fs_info;
+	struct extent_io_tree *io_tree = &inode->io_tree;
+	struct btrfs_path *path;
+	struct extent_buffer *leaf;
+	struct btrfs_file_extent_item *item;
+	u64 ram_bytes;
+	unsigned long ptr;
+	void *tmp;
+	ssize_t ret;
+
+	path = btrfs_alloc_path();
+	if (!path) {
+		ret = -ENOMEM;
+		goto out;
+	}
+	ret = btrfs_lookup_file_extent(NULL, root, path, btrfs_ino(inode),
+				       extent_start, 0);
+	if (ret) {
+		if (ret > 0) {
+			/* The extent item disappeared? */
+			ret = -EIO;
+		}
+		goto out;
+	}
+	leaf = path->nodes[0];
+	item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_file_extent_item);
+
+	ram_bytes = btrfs_file_extent_ram_bytes(leaf, item);
+	ptr = btrfs_file_extent_inline_start(item);
+
+	encoded->len = min_t(u64, extent_start + ram_bytes,
+			     inode->vfs_inode.i_size) - iocb->ki_pos;
+	ret = btrfs_encoded_io_compression_from_extent(fs_info,
+				 btrfs_file_extent_compression(leaf, item));
+	if (ret < 0)
+		goto out;
+	encoded->compression = ret;
+	if (encoded->compression) {
+		size_t inline_size;
+
+		inline_size = btrfs_file_extent_inline_item_len(leaf,
+								path->slots[0]);
+		if (inline_size > count) {
+			ret = -ENOBUFS;
+			goto out;
+		}
+		count = inline_size;
+		encoded->unencoded_len = ram_bytes;
+		encoded->unencoded_offset = iocb->ki_pos - extent_start;
+	} else {
+		count = min_t(u64, count, encoded->len);
+		encoded->len = count;
+		encoded->unencoded_len = count;
+		ptr += iocb->ki_pos - extent_start;
+	}
+
+	tmp = kmalloc(count, GFP_NOFS);
+	if (!tmp) {
+		ret = -ENOMEM;
+		goto out;
+	}
+	read_extent_buffer(leaf, tmp, ptr, count);
+	btrfs_release_path(path);
+	unlock_extent(io_tree, start, lockend, cached_state);
+	btrfs_inode_unlock(&inode->vfs_inode, BTRFS_ILOCK_SHARED);
+	*unlocked = true;
+
+	ret = copy_to_iter(tmp, count, iter);
+	if (ret != count)
+		ret = -EFAULT;
+	kfree(tmp);
+out:
+	btrfs_free_path(path);
+	return ret;
+}
+
+struct btrfs_encoded_read_private {
+	struct btrfs_inode *inode;
+	u64 file_offset;
+	wait_queue_head_t wait;
+	atomic_t pending;
+	blk_status_t status;
+	bool skip_csum;
+};
+
+static blk_status_t submit_encoded_read_bio(struct btrfs_inode *inode,
+					    struct bio *bio, int mirror_num)
+{
+	struct btrfs_encoded_read_private *priv = btrfs_bio(bio)->private;
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
+	blk_status_t ret;
+
+	if (!priv->skip_csum) {
+		ret = btrfs_lookup_bio_sums(&inode->vfs_inode, bio, NULL);
+		if (ret)
+			return ret;
+	}
+
+	atomic_inc(&priv->pending);
+	btrfs_submit_bio(fs_info, bio, mirror_num);
+	return BLK_STS_OK;
+}
+
+static blk_status_t btrfs_encoded_read_verify_csum(struct btrfs_bio *bbio)
+{
+	const bool uptodate = (bbio->bio.bi_status == BLK_STS_OK);
+	struct btrfs_encoded_read_private *priv = bbio->private;
+	struct btrfs_inode *inode = priv->inode;
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
+	u32 sectorsize = fs_info->sectorsize;
+	struct bio_vec *bvec;
+	struct bvec_iter_all iter_all;
+	u32 bio_offset = 0;
+
+	if (priv->skip_csum || !uptodate)
+		return bbio->bio.bi_status;
+
+	bio_for_each_segment_all(bvec, &bbio->bio, iter_all) {
+		unsigned int i, nr_sectors, pgoff;
+
+		nr_sectors = BTRFS_BYTES_TO_BLKS(fs_info, bvec->bv_len);
+		pgoff = bvec->bv_offset;
+		for (i = 0; i < nr_sectors; i++) {
+			ASSERT(pgoff < PAGE_SIZE);
+			if (btrfs_check_data_csum(&inode->vfs_inode, bbio, bio_offset,
+					    bvec->bv_page, pgoff))
+				return BLK_STS_IOERR;
+			bio_offset += sectorsize;
+			pgoff += sectorsize;
+		}
+	}
+	return BLK_STS_OK;
+}
+
+static void btrfs_encoded_read_endio(struct btrfs_bio *bbio)
+{
+	struct btrfs_encoded_read_private *priv = bbio->private;
+	blk_status_t status;
+
+	status = btrfs_encoded_read_verify_csum(bbio);
+	if (status) {
+		/*
+		 * The memory barrier implied by the atomic_dec_return() here
+		 * pairs with the memory barrier implied by the
+		 * atomic_dec_return() or io_wait_event() in
+		 * btrfs_encoded_read_regular_fill_pages() to ensure that this
+		 * write is observed before the load of status in
+		 * btrfs_encoded_read_regular_fill_pages().
+		 */
+		WRITE_ONCE(priv->status, status);
+	}
+	if (!atomic_dec_return(&priv->pending))
+		wake_up(&priv->wait);
+	btrfs_bio_free_csum(bbio);
+	bio_put(&bbio->bio);
+}
+
+int btrfs_encoded_read_regular_fill_pages(struct btrfs_inode *inode,
+					  u64 file_offset, u64 disk_bytenr,
+					  u64 disk_io_size, struct page **pages)
+{
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
+	struct btrfs_encoded_read_private priv = {
+		.inode = inode,
+		.file_offset = file_offset,
+		.pending = ATOMIC_INIT(1),
+		.skip_csum = (inode->flags & BTRFS_INODE_NODATASUM),
+	};
+	unsigned long i = 0;
+	u64 cur = 0;
+	int ret;
+
+	init_waitqueue_head(&priv.wait);
+	/*
+	 * Submit bios for the extent, splitting due to bio or stripe limits as
+	 * necessary.
+	 */
+	while (cur < disk_io_size) {
+		struct extent_map *em;
+		struct btrfs_io_geometry geom;
+		struct bio *bio = NULL;
+		u64 remaining;
+
+		em = btrfs_get_chunk_map(fs_info, disk_bytenr + cur,
+					 disk_io_size - cur);
+		if (IS_ERR(em)) {
+			ret = PTR_ERR(em);
+		} else {
+			ret = btrfs_get_io_geometry(fs_info, em, BTRFS_MAP_READ,
+						    disk_bytenr + cur, &geom);
+			free_extent_map(em);
+		}
+		if (ret) {
+			WRITE_ONCE(priv.status, errno_to_blk_status(ret));
+			break;
+		}
+		remaining = min(geom.len, disk_io_size - cur);
+		while (bio || remaining) {
+			size_t bytes = min_t(u64, remaining, PAGE_SIZE);
+
+			if (!bio) {
+				bio = btrfs_bio_alloc(BIO_MAX_VECS, REQ_OP_READ,
+						      btrfs_encoded_read_endio,
+						      &priv);
+				bio->bi_iter.bi_sector =
+					(disk_bytenr + cur) >> SECTOR_SHIFT;
+			}
+
+			if (!bytes ||
+			    bio_add_page(bio, pages[i], bytes, 0) < bytes) {
+				blk_status_t status;
+
+				status = submit_encoded_read_bio(inode, bio, 0);
+				if (status) {
+					WRITE_ONCE(priv.status, status);
+					bio_put(bio);
+					goto out;
+				}
+				bio = NULL;
+				continue;
+			}
+
+			i++;
+			cur += bytes;
+			remaining -= bytes;
+		}
+	}
+
+out:
+	if (atomic_dec_return(&priv.pending))
+		io_wait_event(priv.wait, !atomic_read(&priv.pending));
+	/* See btrfs_encoded_read_endio() for ordering. */
+	return blk_status_to_errno(READ_ONCE(priv.status));
+}
+
+static ssize_t btrfs_encoded_read_regular(struct kiocb *iocb,
+					  struct iov_iter *iter,
+					  u64 start, u64 lockend,
+					  struct extent_state **cached_state,
+					  u64 disk_bytenr, u64 disk_io_size,
+					  size_t count, bool compressed,
+					  bool *unlocked)
+{
+	struct btrfs_inode *inode = BTRFS_I(file_inode(iocb->ki_filp));
+	struct extent_io_tree *io_tree = &inode->io_tree;
+	struct page **pages;
+	unsigned long nr_pages, i;
+	u64 cur;
+	size_t page_offset;
+	ssize_t ret;
+
+	nr_pages = DIV_ROUND_UP(disk_io_size, PAGE_SIZE);
+	pages = kcalloc(nr_pages, sizeof(struct page *), GFP_NOFS);
+	if (!pages)
+		return -ENOMEM;
+	ret = btrfs_alloc_page_array(nr_pages, pages);
+	if (ret) {
+		ret = -ENOMEM;
+		goto out;
+		}
+
+	ret = btrfs_encoded_read_regular_fill_pages(inode, start, disk_bytenr,
+						    disk_io_size, pages);
+	if (ret)
+		goto out;
+
+	unlock_extent(io_tree, start, lockend, cached_state);
+	btrfs_inode_unlock(&inode->vfs_inode, BTRFS_ILOCK_SHARED);
+	*unlocked = true;
+
+	if (compressed) {
+		i = 0;
+		page_offset = 0;
+	} else {
+		i = (iocb->ki_pos - start) >> PAGE_SHIFT;
+		page_offset = (iocb->ki_pos - start) & (PAGE_SIZE - 1);
+	}
+	cur = 0;
+	while (cur < count) {
+		size_t bytes = min_t(size_t, count - cur,
+				     PAGE_SIZE - page_offset);
+
+		if (copy_page_to_iter(pages[i], page_offset, bytes,
+				      iter) != bytes) {
+			ret = -EFAULT;
+			goto out;
+		}
+		i++;
+		cur += bytes;
+		page_offset = 0;
+	}
+	ret = count;
+out:
+	for (i = 0; i < nr_pages; i++) {
+		if (pages[i])
+			__free_page(pages[i]);
+	}
+	kfree(pages);
+	return ret;
+}
+
+ssize_t btrfs_encoded_read(struct kiocb *iocb, struct iov_iter *iter,
+			   struct btrfs_ioctl_encoded_io_args *encoded)
+{
+	struct btrfs_inode *inode = BTRFS_I(file_inode(iocb->ki_filp));
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
+	struct extent_io_tree *io_tree = &inode->io_tree;
+	ssize_t ret;
+	size_t count = iov_iter_count(iter);
+	u64 start, lockend, disk_bytenr, disk_io_size;
+	struct extent_state *cached_state = NULL;
+	struct extent_map *em;
+	bool unlocked = false;
+
+	file_accessed(iocb->ki_filp);
+
+	btrfs_inode_lock(&inode->vfs_inode, BTRFS_ILOCK_SHARED);
+
+	if (iocb->ki_pos >= inode->vfs_inode.i_size) {
+		btrfs_inode_unlock(&inode->vfs_inode, BTRFS_ILOCK_SHARED);
+		return 0;
+	}
+	start = ALIGN_DOWN(iocb->ki_pos, fs_info->sectorsize);
+	/*
+	 * We don't know how long the extent containing iocb->ki_pos is, but if
+	 * it's compressed we know that it won't be longer than this.
+	 */
+	lockend = start + BTRFS_MAX_UNCOMPRESSED - 1;
+
+	for (;;) {
+		struct btrfs_ordered_extent *ordered;
+
+		ret = btrfs_wait_ordered_range(&inode->vfs_inode, start,
+					       lockend - start + 1);
+		if (ret)
+			goto out_unlock_inode;
+		lock_extent(io_tree, start, lockend, &cached_state);
+		ordered = btrfs_lookup_ordered_range(inode, start,
+						     lockend - start + 1);
+		if (!ordered)
+			break;
+		btrfs_put_ordered_extent(ordered);
+		unlock_extent(io_tree, start, lockend, &cached_state);
+		cond_resched();
+	}
+
+	em = btrfs_get_extent(inode, NULL, 0, start, lockend - start + 1);
+	if (IS_ERR(em)) {
+		ret = PTR_ERR(em);
+		goto out_unlock_extent;
+	}
+
+	if (em->block_start == EXTENT_MAP_INLINE) {
+		u64 extent_start = em->start;
+
+		/*
+		 * For inline extents we get everything we need out of the
+		 * extent item.
+		 */
+		free_extent_map(em);
+		em = NULL;
+		ret = btrfs_encoded_read_inline(iocb, iter, start, lockend,
+						&cached_state, extent_start,
+						count, encoded, &unlocked);
+		goto out;
+	}
+
+	/*
+	 * We only want to return up to EOF even if the extent extends beyond
+	 * that.
+	 */
+	encoded->len = min_t(u64, extent_map_end(em),
+			     inode->vfs_inode.i_size) - iocb->ki_pos;
+	if (em->block_start == EXTENT_MAP_HOLE ||
+	    test_bit(EXTENT_FLAG_PREALLOC, &em->flags)) {
+		disk_bytenr = EXTENT_MAP_HOLE;
+		count = min_t(u64, count, encoded->len);
+		encoded->len = count;
+		encoded->unencoded_len = count;
+	} else if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags)) {
+		disk_bytenr = em->block_start;
+		/*
+		 * Bail if the buffer isn't large enough to return the whole
+		 * compressed extent.
+		 */
+		if (em->block_len > count) {
+			ret = -ENOBUFS;
+			goto out_em;
+		}
+		disk_io_size = em->block_len;
+		count = em->block_len;
+		encoded->unencoded_len = em->ram_bytes;
+		encoded->unencoded_offset = iocb->ki_pos - em->orig_start;
+		ret = btrfs_encoded_io_compression_from_extent(fs_info,
+							     em->compress_type);
+		if (ret < 0)
+			goto out_em;
+		encoded->compression = ret;
+	} else {
+		disk_bytenr = em->block_start + (start - em->start);
+		if (encoded->len > count)
+			encoded->len = count;
+		/*
+		 * Don't read beyond what we locked. This also limits the page
+		 * allocations that we'll do.
+		 */
+		disk_io_size = min(lockend + 1, iocb->ki_pos + encoded->len) - start;
+		count = start + disk_io_size - iocb->ki_pos;
+		encoded->len = count;
+		encoded->unencoded_len = count;
+		disk_io_size = ALIGN(disk_io_size, fs_info->sectorsize);
+	}
+	free_extent_map(em);
+	em = NULL;
+
+	if (disk_bytenr == EXTENT_MAP_HOLE) {
+		unlock_extent(io_tree, start, lockend, &cached_state);
+		btrfs_inode_unlock(&inode->vfs_inode, BTRFS_ILOCK_SHARED);
+		unlocked = true;
+		ret = iov_iter_zero(count, iter);
+		if (ret != count)
+			ret = -EFAULT;
+	} else {
+		ret = btrfs_encoded_read_regular(iocb, iter, start, lockend,
+						 &cached_state, disk_bytenr,
+						 disk_io_size, count,
+						 encoded->compression,
+						 &unlocked);
+	}
+
+out:
+	if (ret >= 0)
+		iocb->ki_pos += encoded->len;
+out_em:
+	free_extent_map(em);
+out_unlock_extent:
+	if (!unlocked)
+		unlock_extent(io_tree, start, lockend, &cached_state);
+out_unlock_inode:
+	if (!unlocked)
+		btrfs_inode_unlock(&inode->vfs_inode, BTRFS_ILOCK_SHARED);
+	return ret;
+}
+
+ssize_t btrfs_do_encoded_write(struct kiocb *iocb, struct iov_iter *from,
+			       const struct btrfs_ioctl_encoded_io_args *encoded)
+{
+	struct btrfs_inode *inode = BTRFS_I(file_inode(iocb->ki_filp));
+	struct btrfs_root *root = inode->root;
+	struct btrfs_fs_info *fs_info = root->fs_info;
+	struct extent_io_tree *io_tree = &inode->io_tree;
+	struct extent_changeset *data_reserved = NULL;
+	struct extent_state *cached_state = NULL;
+	int compression;
+	size_t orig_count;
+	u64 start, end;
+	u64 num_bytes, ram_bytes, disk_num_bytes;
+	unsigned long nr_pages, i;
+	struct page **pages;
+	struct btrfs_key ins;
+	bool extent_reserved = false;
+	struct extent_map *em;
+	ssize_t ret;
+
+	switch (encoded->compression) {
+	case BTRFS_ENCODED_IO_COMPRESSION_ZLIB:
+		compression = BTRFS_COMPRESS_ZLIB;
+		break;
+	case BTRFS_ENCODED_IO_COMPRESSION_ZSTD:
+		compression = BTRFS_COMPRESS_ZSTD;
+		break;
+	case BTRFS_ENCODED_IO_COMPRESSION_LZO_4K:
+	case BTRFS_ENCODED_IO_COMPRESSION_LZO_8K:
+	case BTRFS_ENCODED_IO_COMPRESSION_LZO_16K:
+	case BTRFS_ENCODED_IO_COMPRESSION_LZO_32K:
+	case BTRFS_ENCODED_IO_COMPRESSION_LZO_64K:
+		/* The sector size must match for LZO. */
+		if (encoded->compression -
+		    BTRFS_ENCODED_IO_COMPRESSION_LZO_4K + 12 !=
+		    fs_info->sectorsize_bits)
+			return -EINVAL;
+		compression = BTRFS_COMPRESS_LZO;
+		break;
+	default:
+		return -EINVAL;
+	}
+	if (encoded->encryption != BTRFS_ENCODED_IO_ENCRYPTION_NONE)
+		return -EINVAL;
+
+	orig_count = iov_iter_count(from);
+
+	/* The extent size must be sane. */
+	if (encoded->unencoded_len > BTRFS_MAX_UNCOMPRESSED ||
+	    orig_count > BTRFS_MAX_COMPRESSED || orig_count == 0)
+		return -EINVAL;
+
+	/*
+	 * The compressed data must be smaller than the decompressed data.
+	 *
+	 * It's of course possible for data to compress to larger or the same
+	 * size, but the buffered I/O path falls back to no compression for such
+	 * data, and we don't want to break any assumptions by creating these
+	 * extents.
+	 *
+	 * Note that this is less strict than the current check we have that the
+	 * compressed data must be at least one sector smaller than the
+	 * decompressed data. We only want to enforce the weaker requirement
+	 * from old kernels that it is at least one byte smaller.
+	 */
+	if (orig_count >= encoded->unencoded_len)
+		return -EINVAL;
+
+	/* The extent must start on a sector boundary. */
+	start = iocb->ki_pos;
+	if (!IS_ALIGNED(start, fs_info->sectorsize))
+		return -EINVAL;
+
+	/*
+	 * The extent must end on a sector boundary. However, we allow a write
+	 * which ends at or extends i_size to have an unaligned length; we round
+	 * up the extent size and set i_size to the unaligned end.
+	 */
+	if (start + encoded->len < inode->vfs_inode.i_size &&
+	    !IS_ALIGNED(start + encoded->len, fs_info->sectorsize))
+		return -EINVAL;
+
+	/* Finally, the offset in the unencoded data must be sector-aligned. */
+	if (!IS_ALIGNED(encoded->unencoded_offset, fs_info->sectorsize))
+		return -EINVAL;
+
+	num_bytes = ALIGN(encoded->len, fs_info->sectorsize);
+	ram_bytes = ALIGN(encoded->unencoded_len, fs_info->sectorsize);
+	end = start + num_bytes - 1;
+
+	/*
+	 * If the extent cannot be inline, the compressed data on disk must be
+	 * sector-aligned. For convenience, we extend it with zeroes if it
+	 * isn't.
+	 */
+	disk_num_bytes = ALIGN(orig_count, fs_info->sectorsize);
+	nr_pages = DIV_ROUND_UP(disk_num_bytes, PAGE_SIZE);
+	pages = kvcalloc(nr_pages, sizeof(struct page *), GFP_KERNEL_ACCOUNT);
+	if (!pages)
+		return -ENOMEM;
+	for (i = 0; i < nr_pages; i++) {
+		size_t bytes = min_t(size_t, PAGE_SIZE, iov_iter_count(from));
+		char *kaddr;
+
+		pages[i] = alloc_page(GFP_KERNEL_ACCOUNT);
+		if (!pages[i]) {
+			ret = -ENOMEM;
+			goto out_pages;
+		}
+		kaddr = kmap_local_page(pages[i]);
+		if (copy_from_iter(kaddr, bytes, from) != bytes) {
+			kunmap_local(kaddr);
+			ret = -EFAULT;
+			goto out_pages;
+		}
+		if (bytes < PAGE_SIZE)
+			memset(kaddr + bytes, 0, PAGE_SIZE - bytes);
+		kunmap_local(kaddr);
+	}
+
+	for (;;) {
+		struct btrfs_ordered_extent *ordered;
+
+		ret = btrfs_wait_ordered_range(&inode->vfs_inode, start, num_bytes);
+		if (ret)
+			goto out_pages;
+		ret = invalidate_inode_pages2_range(inode->vfs_inode.i_mapping,
+						    start >> PAGE_SHIFT,
+						    end >> PAGE_SHIFT);
+		if (ret)
+			goto out_pages;
+		lock_extent(io_tree, start, end, &cached_state);
+		ordered = btrfs_lookup_ordered_range(inode, start, num_bytes);
+		if (!ordered &&
+		    !filemap_range_has_page(inode->vfs_inode.i_mapping, start, end))
+			break;
+		if (ordered)
+			btrfs_put_ordered_extent(ordered);
+		unlock_extent(io_tree, start, end, &cached_state);
+		cond_resched();
+	}
+
+	/*
+	 * We don't use the higher-level delalloc space functions because our
+	 * num_bytes and disk_num_bytes are different.
+	 */
+	ret = btrfs_alloc_data_chunk_ondemand(inode, disk_num_bytes);
+	if (ret)
+		goto out_unlock;
+	ret = btrfs_qgroup_reserve_data(inode, &data_reserved, start, num_bytes);
+	if (ret)
+		goto out_free_data_space;
+	ret = btrfs_delalloc_reserve_metadata(inode, num_bytes, disk_num_bytes,
+					      false);
+	if (ret)
+		goto out_qgroup_free_data;
+
+	/* Try an inline extent first. */
+	if (start == 0 && encoded->unencoded_len == encoded->len &&
+	    encoded->unencoded_offset == 0) {
+		ret = cow_file_range_inline(inode, encoded->len, orig_count,
+					    compression, pages, true);
+		if (ret <= 0) {
+			if (ret == 0)
+				ret = orig_count;
+			goto out_delalloc_release;
+		}
+	}
+
+	ret = btrfs_reserve_extent(root, disk_num_bytes, disk_num_bytes,
+				   disk_num_bytes, 0, 0, &ins, 1, 1);
+	if (ret)
+		goto out_delalloc_release;
+	extent_reserved = true;
+
+	em = create_io_em(inode, start, num_bytes,
+			  start - encoded->unencoded_offset, ins.objectid,
+			  ins.offset, ins.offset, ram_bytes, compression,
+			  BTRFS_ORDERED_COMPRESSED);
+	if (IS_ERR(em)) {
+		ret = PTR_ERR(em);
+		goto out_free_reserved;
+	}
+	free_extent_map(em);
+
+	ret = btrfs_add_ordered_extent(inode, start, num_bytes, ram_bytes,
+				       ins.objectid, ins.offset,
+				       encoded->unencoded_offset,
+				       (1 << BTRFS_ORDERED_ENCODED) |
+				       (1 << BTRFS_ORDERED_COMPRESSED),
+				       compression);
+	if (ret) {
+		btrfs_drop_extent_map_range(inode, start, end, false);
+		goto out_free_reserved;
+	}
+	btrfs_dec_block_group_reservations(fs_info, ins.objectid);
+
+	if (start + encoded->len > inode->vfs_inode.i_size)
+		i_size_write(&inode->vfs_inode, start + encoded->len);
+
+	unlock_extent(io_tree, start, end, &cached_state);
+
+	btrfs_delalloc_release_extents(inode, num_bytes);
+
+	if (btrfs_submit_compressed_write(inode, start, num_bytes, ins.objectid,
+					  ins.offset, pages, nr_pages, 0, NULL,
+					  false)) {
+		btrfs_writepage_endio_finish_ordered(inode, pages[0], start, end, 0);
+		ret = -EIO;
+		goto out_pages;
+	}
+	ret = orig_count;
+	goto out;
+
+out_free_reserved:
+	btrfs_dec_block_group_reservations(fs_info, ins.objectid);
+	btrfs_free_reserved_extent(fs_info, ins.objectid, ins.offset, 1);
+out_delalloc_release:
+	btrfs_delalloc_release_extents(inode, num_bytes);
+	btrfs_delalloc_release_metadata(inode, disk_num_bytes, ret < 0);
+out_qgroup_free_data:
+	if (ret < 0)
+		btrfs_qgroup_free_data(inode, data_reserved, start, num_bytes);
+out_free_data_space:
+	/*
+	 * If btrfs_reserve_extent() succeeded, then we already decremented
+	 * bytes_may_use.
+	 */
+	if (!extent_reserved)
+		btrfs_free_reserved_data_space_noquota(fs_info, disk_num_bytes);
+out_unlock:
+	unlock_extent(io_tree, start, end, &cached_state);
+out_pages:
+	for (i = 0; i < nr_pages; i++) {
+		if (pages[i])
+			__free_page(pages[i]);
+	}
+	kvfree(pages);
+out:
+	if (ret >= 0)
+		iocb->ki_pos += encoded->len;
+	return ret;
+}
+
 #ifdef CONFIG_SWAP
 /*
  * Add an entry indicating a block group or device which is pinned by a
@@ -10098,6 +10849,7 @@ static int btrfs_add_swapfile_pin(struct inode *inode, void *ptr,
 	sp->ptr = ptr;
 	sp->inode = inode;
 	sp->is_block_group = is_block_group;
+	sp->bg_extent_count = 1;
 
 	spin_lock(&fs_info->swapfile_pins_lock);
 	p = &fs_info->swapfile_pins.rb_node;
@@ -10111,6 +10863,8 @@ static int btrfs_add_swapfile_pin(struct inode *inode, void *ptr,
 			   (sp->ptr == entry->ptr && sp->inode > entry->inode)) {
 			p = &(*p)->rb_right;
 		} else {
+			if (is_block_group)
+				entry->bg_extent_count++;
 			spin_unlock(&fs_info->swapfile_pins_lock);
 			kfree(sp);
 			return 1;
@@ -10136,8 +10890,11 @@ static void btrfs_free_swapfile_pins(struct inode *inode)
 		sp = rb_entry(node, struct btrfs_swapfile_pin, node);
 		if (sp->inode == inode) {
 			rb_erase(&sp->node, &fs_info->swapfile_pins);
-			if (sp->is_block_group)
+			if (sp->is_block_group) {
+				btrfs_dec_block_group_swap_extents(sp->ptr,
+							   sp->bg_extent_count);
 				btrfs_put_block_group(sp->ptr);
+			}
 			kfree(sp);
 		}
 		node = next;
@@ -10159,9 +10916,19 @@ static int btrfs_add_swap_extent(struct swap_info_struct *sis,
 				 struct btrfs_swap_info *bsi)
 {
 	unsigned long nr_pages;
+	unsigned long max_pages;
 	u64 first_ppage, first_ppage_reported, next_ppage;
 	int ret;
 
+	/*
+	 * Our swapfile may have had its size extended after the swap header was
+	 * written. In that case activating the swapfile should not go beyond
+	 * the max size set in the swap header.
+	 */
+	if (bsi->nr_pages >= sis->max)
+		return 0;
+
+	max_pages = sis->max - bsi->nr_pages;
 	first_ppage = ALIGN(bsi->block_start, PAGE_SIZE) >> PAGE_SHIFT;
 	next_ppage = ALIGN_DOWN(bsi->block_start + bsi->block_len,
 				PAGE_SIZE) >> PAGE_SHIFT;
@@ -10169,6 +10936,7 @@ static int btrfs_add_swap_extent(struct swap_info_struct *sis,
 	if (first_ppage >= next_ppage)
 		return 0;
 	nr_pages = next_ppage - first_ppage;
+	nr_pages = min(nr_pages, max_pages);
 
 	first_ppage_reported = first_ppage;
 	if (bsi->start == 0)
@@ -10198,7 +10966,8 @@ static int btrfs_swap_activate(struct swap_info_struct *sis, struct file *file,
 			       sector_t *span)
 {
 	struct inode *inode = file_inode(file);
-	struct btrfs_fs_info *fs_info = BTRFS_I(inode)->root->fs_info;
+	struct btrfs_root *root = BTRFS_I(inode)->root;
+	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
 	struct extent_state *cached_state = NULL;
 	struct extent_map *em = NULL;
@@ -10237,29 +11006,58 @@ static int btrfs_swap_activate(struct swap_info_struct *sis, struct file *file,
 
 	/*
 	 * Balance or device remove/replace/resize can move stuff around from
-	 * under us. The EXCL_OP flag makes sure they aren't running/won't run
-	 * concurrently while we are mapping the swap extents, and
-	 * fs_info->swapfile_pins prevents them from running while the swap file
-	 * is active and moving the extents. Note that this also prevents a
-	 * concurrent device add which isn't actually necessary, but it's not
+	 * under us. The exclop protection makes sure they aren't running/won't
+	 * run concurrently while we are mapping the swap extents, and
+	 * fs_info->swapfile_pins prevents them from running while the swap
+	 * file is active and moving the extents. Note that this also prevents
+	 * a concurrent device add which isn't actually necessary, but it's not
 	 * really worth the trouble to allow it.
 	 */
-	if (test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags)) {
+	if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_SWAP_ACTIVATE)) {
 		btrfs_warn(fs_info,
 	   "cannot activate swapfile while exclusive operation is running");
 		return -EBUSY;
 	}
+
+	/*
+	 * Prevent snapshot creation while we are activating the swap file.
+	 * We do not want to race with snapshot creation. If snapshot creation
+	 * already started before we bumped nr_swapfiles from 0 to 1 and
+	 * completes before the first write into the swap file after it is
+	 * activated, than that write would fallback to COW.
+	 */
+	if (!btrfs_drew_try_write_lock(&root->snapshot_lock)) {
+		btrfs_exclop_finish(fs_info);
+		btrfs_warn(fs_info,
+	   "cannot activate swapfile because snapshot creation is in progress");
+		return -EINVAL;
+	}
 	/*
 	 * Snapshots can create extents which require COW even if NODATACOW is
 	 * set. We use this counter to prevent snapshots. We must increment it
 	 * before walking the extents because we don't want a concurrent
 	 * snapshot to run after we've already checked the extents.
+	 *
+	 * It is possible that subvolume is marked for deletion but still not
+	 * removed yet. To prevent this race, we check the root status before
+	 * activating the swapfile.
 	 */
-	atomic_inc(&BTRFS_I(inode)->root->nr_swapfiles);
+	spin_lock(&root->root_item_lock);
+	if (btrfs_root_dead(root)) {
+		spin_unlock(&root->root_item_lock);
+
+		btrfs_exclop_finish(fs_info);
+		btrfs_warn(fs_info,
+		"cannot activate swapfile because subvolume %llu is being deleted",
+			root->root_key.objectid);
+		return -EPERM;
+	}
+	atomic_inc(&root->nr_swapfiles);
+	spin_unlock(&root->root_item_lock);
 
 	isize = ALIGN_DOWN(inode->i_size, fs_info->sectorsize);
 
-	lock_extent_bits(io_tree, 0, isize - 1, &cached_state);
+	lock_extent(io_tree, 0, isize - 1, &cached_state);
 	start = 0;
 	while (start < isize) {
 		u64 logical_block_start, physical_block_start;
@@ -10300,7 +11098,7 @@ static int btrfs_swap_activate(struct swap_info_struct *sis, struct file *file,
 		free_extent_map(em);
 		em = NULL;
 
-		ret = can_nocow_extent(inode, start, &len, NULL, NULL, NULL);
+		ret = can_nocow_extent(inode, start, &len, NULL, NULL, NULL, false, true);
 		if (ret < 0) {
 			goto out;
 		} else if (ret) {
@@ -10352,6 +11150,17 @@ static int btrfs_swap_activate(struct swap_info_struct *sis, struct file *file,
 			goto out;
 		}
 
+		if (!btrfs_inc_block_group_swap_extents(bg)) {
+			btrfs_warn(fs_info,
+			   "block group for swapfile at %llu is read-only%s",
+			   bg->start,
+			   atomic_read(&fs_info->scrubs_running) ?
+				       " (scrub running)" : "");
+			btrfs_put_block_group(bg);
+			ret = -EINVAL;
+			goto out;
+		}
+
 		ret = btrfs_add_swapfile_pin(inode, bg, true);
 		if (ret) {
 			btrfs_put_block_group(bg);
@@ -10385,12 +11194,14 @@ out:
 	if (!IS_ERR_OR_NULL(em))
 		free_extent_map(em);
 
-	unlock_extent_cached(io_tree, 0, isize - 1, &cached_state);
+	unlock_extent(io_tree, 0, isize - 1, &cached_state);
 
 	if (ret)
 		btrfs_swap_deactivate(file);
 
-	clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
+	btrfs_drew_write_unlock(&root->snapshot_lock);
+
+	btrfs_exclop_finish(fs_info);
 
 	if (ret)
 		return ret;
@@ -10415,6 +11226,62 @@ static int btrfs_swap_activate(struct swap_info_struct *sis, struct file *file,
 }
 #endif
 
+/*
+ * Update the number of bytes used in the VFS' inode. When we replace extents in
+ * a range (clone, dedupe, fallocate's zero range), we must update the number of
+ * bytes used by the inode in an atomic manner, so that concurrent stat(2) calls
+ * always get a correct value.
+ */
+void btrfs_update_inode_bytes(struct btrfs_inode *inode,
+			      const u64 add_bytes,
+			      const u64 del_bytes)
+{
+	if (add_bytes == del_bytes)
+		return;
+
+	spin_lock(&inode->lock);
+	if (del_bytes > 0)
+		inode_sub_bytes(&inode->vfs_inode, del_bytes);
+	if (add_bytes > 0)
+		inode_add_bytes(&inode->vfs_inode, add_bytes);
+	spin_unlock(&inode->lock);
+}
+
+/**
+ * Verify that there are no ordered extents for a given file range.
+ *
+ * @inode:   The target inode.
+ * @start:   Start offset of the file range, should be sector size aligned.
+ * @end:     End offset (inclusive) of the file range, its value +1 should be
+ *           sector size aligned.
+ *
+ * This should typically be used for cases where we locked an inode's VFS lock in
+ * exclusive mode, we have also locked the inode's i_mmap_lock in exclusive mode,
+ * we have flushed all delalloc in the range, we have waited for all ordered
+ * extents in the range to complete and finally we have locked the file range in
+ * the inode's io_tree.
+ */
+void btrfs_assert_inode_range_clean(struct btrfs_inode *inode, u64 start, u64 end)
+{
+	struct btrfs_root *root = inode->root;
+	struct btrfs_ordered_extent *ordered;
+
+	if (!IS_ENABLED(CONFIG_BTRFS_ASSERT))
+		return;
+
+	ordered = btrfs_lookup_first_ordered_range(inode, start, end + 1 - start);
+	if (ordered) {
+		btrfs_err(root->fs_info,
+"found unexpected ordered extent in file range [%llu, %llu] for inode %llu root %llu (ordered range [%llu, %llu])",
+			  start, end, btrfs_ino(inode), root->root_key.objectid,
+			  ordered->file_offset,
+			  ordered->file_offset + ordered->num_bytes - 1);
+		btrfs_put_ordered_extent(ordered);
+	}
+
+	ASSERT(ordered == NULL);
+}
+
 static const struct inode_operations btrfs_dir_inode_operations = {
 	.getattr	= btrfs_getattr,
 	.lookup		= btrfs_lookup,
@@ -10433,6 +11300,8 @@ static const struct inode_operations btrfs_dir_inode_operations = {
 	.set_acl	= btrfs_set_acl,
 	.update_time	= btrfs_update_time,
 	.tmpfile        = btrfs_tmpfile,
+	.fileattr_get	= btrfs_fileattr_get,
+	.fileattr_set	= btrfs_fileattr_set,
 };
 
 static const struct file_operations btrfs_dir_file_operations = {
@@ -10448,12 +11317,6 @@ static const struct file_operations btrfs_dir_file_operations = {
 	.fsync		= btrfs_sync_file,
 };
 
-static const struct extent_io_ops btrfs_extent_io_ops = {
-	/* mandatory callbacks */
-	.submit_bio_hook = btrfs_submit_bio_hook,
-	.readpage_end_io_hook = btrfs_readpage_end_io_hook,
-};
-
 /*
  * btrfs doesn't support the bmap operation because swapfiles
  * use bmap to make a mapping of extents in the file.  They assume
@@ -10467,14 +11330,14 @@ static const struct extent_io_ops btrfs_extent_io_ops = {
  * For now we're avoiding this by dropping bmap.
  */
 static const struct address_space_operations btrfs_aops = {
-	.readpage	= btrfs_readpage,
-	.writepage	= btrfs_writepage,
+	.read_folio	= btrfs_read_folio,
 	.writepages	= btrfs_writepages,
-	.readpages	= btrfs_readpages,
-	.direct_IO	= btrfs_direct_IO,
-	.invalidatepage = btrfs_invalidatepage,
-	.releasepage	= btrfs_releasepage,
-	.set_page_dirty	= btrfs_set_page_dirty,
+	.readahead	= btrfs_readahead,
+	.direct_IO	= noop_direct_IO,
+	.invalidate_folio = btrfs_invalidate_folio,
+	.release_folio	= btrfs_release_folio,
+	.migrate_folio	= btrfs_migrate_folio,
+	.dirty_folio	= filemap_dirty_folio,
 	.error_remove_page = generic_error_remove_page,
 	.swap_activate	= btrfs_swap_activate,
 	.swap_deactivate = btrfs_swap_deactivate,
@@ -10489,6 +11352,8 @@ static const struct inode_operations btrfs_file_inode_operations = {
 	.get_acl	= btrfs_get_acl,
 	.set_acl	= btrfs_set_acl,
 	.update_time	= btrfs_update_time,
+	.fileattr_get	= btrfs_fileattr_get,
+	.fileattr_set	= btrfs_fileattr_set,
 };
 static const struct inode_operations btrfs_special_inode_operations = {
 	.getattr	= btrfs_getattr,
diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
index 4f4b13830b25..d5dd8bed1488 100644
--- a/fs/btrfs/ioctl.c
+++ b/fs/btrfs/ioctl.c
@@ -26,14 +26,17 @@
 #include <linux/btrfs.h>
 #include <linux/uaccess.h>
 #include <linux/iversion.h>
+#include <linux/fileattr.h>
+#include <linux/fsverity.h>
+#include <linux/sched/xacct.h>
 #include "ctree.h"
 #include "disk-io.h"
+#include "export.h"
 #include "transaction.h"
 #include "btrfs_inode.h"
 #include "print-tree.h"
 #include "volumes.h"
 #include "locking.h"
-#include "inode-map.h"
 #include "backref.h"
 #include "rcu-string.h"
 #include "send.h"
@@ -46,6 +49,7 @@
 #include "space-info.h"
 #include "delalloc-space.h"
 #include "block-group.h"
+#include "subpage.h"
 
 #ifdef CONFIG_64BIT
 /* If we have a 32-bit userspace and 64-bit kernel, then the UAPI
@@ -79,16 +83,31 @@ struct btrfs_ioctl_send_args_32 {
 	compat_uptr_t clone_sources;	/* in */
 	__u64 parent_root;		/* in */
 	__u64 flags;			/* in */
-	__u64 reserved[4];		/* in */
+	__u32 version;			/* in */
+	__u8  reserved[28];		/* in */
 } __attribute__ ((__packed__));
 
 #define BTRFS_IOC_SEND_32 _IOW(BTRFS_IOCTL_MAGIC, 38, \
 			       struct btrfs_ioctl_send_args_32)
-#endif
 
-static int btrfs_clone(struct inode *src, struct inode *inode,
-		       u64 off, u64 olen, u64 olen_aligned, u64 destoff,
-		       int no_time_update);
+struct btrfs_ioctl_encoded_io_args_32 {
+	compat_uptr_t iov;
+	compat_ulong_t iovcnt;
+	__s64 offset;
+	__u64 flags;
+	__u64 len;
+	__u64 unencoded_len;
+	__u64 unencoded_offset;
+	__u32 compression;
+	__u32 encryption;
+	__u8 reserved[64];
+};
+
+#define BTRFS_IOC_ENCODED_READ_32 _IOR(BTRFS_IOCTL_MAGIC, 64, \
+				       struct btrfs_ioctl_encoded_io_args_32)
+#define BTRFS_IOC_ENCODED_WRITE_32 _IOW(BTRFS_IOCTL_MAGIC, 64, \
+					struct btrfs_ioctl_encoded_io_args_32)
+#endif
 
 /* Mask out flags that are inappropriate for the given type of inode. */
 static unsigned int btrfs_mask_fsflags_for_type(struct inode *inode,
@@ -106,9 +125,11 @@ static unsigned int btrfs_mask_fsflags_for_type(struct inode *inode,
  * Export internal inode flags to the format expected by the FS_IOC_GETFLAGS
  * ioctl.
  */
-static unsigned int btrfs_inode_flags_to_fsflags(unsigned int flags)
+static unsigned int btrfs_inode_flags_to_fsflags(struct btrfs_inode *binode)
 {
 	unsigned int iflags = 0;
+	u32 flags = binode->flags;
+	u32 ro_flags = binode->ro_flags;
 
 	if (flags & BTRFS_INODE_SYNC)
 		iflags |= FS_SYNC_FL;
@@ -124,6 +145,8 @@ static unsigned int btrfs_inode_flags_to_fsflags(unsigned int flags)
 		iflags |= FS_DIRSYNC_FL;
 	if (flags & BTRFS_INODE_NODATACOW)
 		iflags |= FS_NOCOW_FL;
+	if (ro_flags & BTRFS_INODE_RO_VERITY)
+		iflags |= FS_VERITY_FL;
 
 	if (flags & BTRFS_INODE_NOCOMPRESS)
 		iflags |= FS_NOCOMP_FL;
@@ -151,24 +174,19 @@ void btrfs_sync_inode_flags_to_i_flags(struct inode *inode)
 		new_fl |= S_NOATIME;
 	if (binode->flags & BTRFS_INODE_DIRSYNC)
 		new_fl |= S_DIRSYNC;
+	if (binode->ro_flags & BTRFS_INODE_RO_VERITY)
+		new_fl |= S_VERITY;
 
 	set_mask_bits(&inode->i_flags,
-		      S_SYNC | S_APPEND | S_IMMUTABLE | S_NOATIME | S_DIRSYNC,
-		      new_fl);
+		      S_SYNC | S_APPEND | S_IMMUTABLE | S_NOATIME | S_DIRSYNC |
+		      S_VERITY, new_fl);
 }
 
-static int btrfs_ioctl_getflags(struct file *file, void __user *arg)
-{
-	struct btrfs_inode *binode = BTRFS_I(file_inode(file));
-	unsigned int flags = btrfs_inode_flags_to_fsflags(binode->flags);
-
-	if (copy_to_user(arg, &flags, sizeof(flags)))
-		return -EFAULT;
-	return 0;
-}
-
-/* Check if @flags are a supported and valid set of FS_*_FL flags */
-static int check_fsflags(unsigned int flags)
+/*
+ * Check if @flags are a supported and valid set of FS_*_FL flags and that
+ * the old and new flags are not conflicting
+ */
+static int check_fsflags(unsigned int old_flags, unsigned int flags)
 {
 	if (flags & ~(FS_IMMUTABLE_FL | FS_APPEND_FL | \
 		      FS_NOATIME_FL | FS_NODUMP_FL | \
@@ -177,15 +195,47 @@ static int check_fsflags(unsigned int flags)
 		      FS_NOCOW_FL))
 		return -EOPNOTSUPP;
 
+	/* COMPR and NOCOMP on new/old are valid */
 	if ((flags & FS_NOCOMP_FL) && (flags & FS_COMPR_FL))
 		return -EINVAL;
 
+	if ((flags & FS_COMPR_FL) && (flags & FS_NOCOW_FL))
+		return -EINVAL;
+
+	/* NOCOW and compression options are mutually exclusive */
+	if ((old_flags & FS_NOCOW_FL) && (flags & (FS_COMPR_FL | FS_NOCOMP_FL)))
+		return -EINVAL;
+	if ((flags & FS_NOCOW_FL) && (old_flags & (FS_COMPR_FL | FS_NOCOMP_FL)))
+		return -EINVAL;
+
 	return 0;
 }
 
-static int btrfs_ioctl_setflags(struct file *file, void __user *arg)
+static int check_fsflags_compatible(struct btrfs_fs_info *fs_info,
+				    unsigned int flags)
 {
-	struct inode *inode = file_inode(file);
+	if (btrfs_is_zoned(fs_info) && (flags & FS_NOCOW_FL))
+		return -EPERM;
+
+	return 0;
+}
+
+/*
+ * Set flags/xflags from the internal inode flags. The remaining items of
+ * fsxattr are zeroed.
+ */
+int btrfs_fileattr_get(struct dentry *dentry, struct fileattr *fa)
+{
+	struct btrfs_inode *binode = BTRFS_I(d_inode(dentry));
+
+	fileattr_fill_flags(fa, btrfs_inode_flags_to_fsflags(binode));
+	return 0;
+}
+
+int btrfs_fileattr_set(struct user_namespace *mnt_userns,
+		       struct dentry *dentry, struct fileattr *fa)
+{
+	struct inode *inode = d_inode(dentry);
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	struct btrfs_inode *binode = BTRFS_I(inode);
 	struct btrfs_root *root = binode->root;
@@ -193,33 +243,25 @@ static int btrfs_ioctl_setflags(struct file *file, void __user *arg)
 	unsigned int fsflags, old_fsflags;
 	int ret;
 	const char *comp = NULL;
-	u32 binode_flags = binode->flags;
-
-	if (!inode_owner_or_capable(inode))
-		return -EPERM;
+	u32 binode_flags;
 
 	if (btrfs_root_readonly(root))
 		return -EROFS;
 
-	if (copy_from_user(&fsflags, arg, sizeof(fsflags)))
-		return -EFAULT;
+	if (fileattr_has_fsx(fa))
+		return -EOPNOTSUPP;
 
-	ret = check_fsflags(fsflags);
+	fsflags = btrfs_mask_fsflags_for_type(inode, fa->flags);
+	old_fsflags = btrfs_inode_flags_to_fsflags(binode);
+	ret = check_fsflags(old_fsflags, fsflags);
 	if (ret)
 		return ret;
 
-	ret = mnt_want_write_file(file);
+	ret = check_fsflags_compatible(fs_info, fsflags);
 	if (ret)
 		return ret;
 
-	inode_lock(inode);
-
-	fsflags = btrfs_mask_fsflags_for_type(inode, fsflags);
-	old_fsflags = btrfs_inode_flags_to_fsflags(binode->flags);
-	ret = vfs_ioc_setflags_prepare(inode, old_fsflags, fsflags);
-	if (ret)
-		goto out_unlock;
-
+	binode_flags = binode->flags;
 	if (fsflags & FS_SYNC_FL)
 		binode_flags |= BTRFS_INODE_SYNC;
 	else
@@ -240,6 +282,16 @@ static int btrfs_ioctl_setflags(struct file *file, void __user *arg)
 		binode_flags |= BTRFS_INODE_NOATIME;
 	else
 		binode_flags &= ~BTRFS_INODE_NOATIME;
+
+	/* If coming from FS_IOC_FSSETXATTR then skip unconverted flags */
+	if (!fa->flags_valid) {
+		/* 1 item for the inode */
+		trans = btrfs_start_transaction(root, 1);
+		if (IS_ERR(trans))
+			return PTR_ERR(trans);
+		goto update_flags;
+	}
+
 	if (fsflags & FS_DIRSYNC_FL)
 		binode_flags |= BTRFS_INODE_DIRSYNC;
 	else
@@ -280,10 +332,8 @@ static int btrfs_ioctl_setflags(struct file *file, void __user *arg)
 		binode_flags |= BTRFS_INODE_NOCOMPRESS;
 	} else if (fsflags & FS_COMPR_FL) {
 
-		if (IS_SWAPFILE(inode)) {
-			ret = -ETXTBSY;
-			goto out_unlock;
-		}
+		if (IS_SWAPFILE(inode))
+			return -ETXTBSY;
 
 		binode_flags |= BTRFS_INODE_COMPRESS;
 		binode_flags &= ~BTRFS_INODE_NOCOMPRESS;
@@ -300,10 +350,8 @@ static int btrfs_ioctl_setflags(struct file *file, void __user *arg)
 	 * 2 for properties
 	 */
 	trans = btrfs_start_transaction(root, 3);
-	if (IS_ERR(trans)) {
-		ret = PTR_ERR(trans);
-		goto out_unlock;
-	}
+	if (IS_ERR(trans))
+		return PTR_ERR(trans);
 
 	if (comp) {
 		ret = btrfs_set_prop(trans, inode, "btrfs.compression", comp,
@@ -321,161 +369,98 @@ static int btrfs_ioctl_setflags(struct file *file, void __user *arg)
 		}
 	}
 
+update_flags:
 	binode->flags = binode_flags;
 	btrfs_sync_inode_flags_to_i_flags(inode);
 	inode_inc_iversion(inode);
 	inode->i_ctime = current_time(inode);
-	ret = btrfs_update_inode(trans, root, inode);
+	ret = btrfs_update_inode(trans, root, BTRFS_I(inode));
 
  out_end_trans:
 	btrfs_end_transaction(trans);
- out_unlock:
-	inode_unlock(inode);
-	mnt_drop_write_file(file);
 	return ret;
 }
 
 /*
- * Translate btrfs internal inode flags to xflags as expected by the
- * FS_IOC_FSGETXATT ioctl. Filter only the supported ones, unknown flags are
- * silently dropped.
+ * Start exclusive operation @type, return true on success
  */
-static unsigned int btrfs_inode_flags_to_xflags(unsigned int flags)
+bool btrfs_exclop_start(struct btrfs_fs_info *fs_info,
+			enum btrfs_exclusive_operation type)
 {
-	unsigned int xflags = 0;
+	bool ret = false;
 
-	if (flags & BTRFS_INODE_APPEND)
-		xflags |= FS_XFLAG_APPEND;
-	if (flags & BTRFS_INODE_IMMUTABLE)
-		xflags |= FS_XFLAG_IMMUTABLE;
-	if (flags & BTRFS_INODE_NOATIME)
-		xflags |= FS_XFLAG_NOATIME;
-	if (flags & BTRFS_INODE_NODUMP)
-		xflags |= FS_XFLAG_NODUMP;
-	if (flags & BTRFS_INODE_SYNC)
-		xflags |= FS_XFLAG_SYNC;
-
-	return xflags;
-}
+	spin_lock(&fs_info->super_lock);
+	if (fs_info->exclusive_operation == BTRFS_EXCLOP_NONE) {
+		fs_info->exclusive_operation = type;
+		ret = true;
+	}
+	spin_unlock(&fs_info->super_lock);
 
-/* Check if @flags are a supported and valid set of FS_XFLAGS_* flags */
-static int check_xflags(unsigned int flags)
-{
-	if (flags & ~(FS_XFLAG_APPEND | FS_XFLAG_IMMUTABLE | FS_XFLAG_NOATIME |
-		      FS_XFLAG_NODUMP | FS_XFLAG_SYNC))
-		return -EOPNOTSUPP;
-	return 0;
+	return ret;
 }
 
 /*
- * Set the xflags from the internal inode flags. The remaining items of fsxattr
- * are zeroed.
+ * Conditionally allow to enter the exclusive operation in case it's compatible
+ * with the running one.  This must be paired with btrfs_exclop_start_unlock and
+ * btrfs_exclop_finish.
+ *
+ * Compatibility:
+ * - the same type is already running
+ * - when trying to add a device and balance has been paused
+ * - not BTRFS_EXCLOP_NONE - this is intentionally incompatible and the caller
+ *   must check the condition first that would allow none -> @type
  */
-static int btrfs_ioctl_fsgetxattr(struct file *file, void __user *arg)
+bool btrfs_exclop_start_try_lock(struct btrfs_fs_info *fs_info,
+				 enum btrfs_exclusive_operation type)
 {
-	struct btrfs_inode *binode = BTRFS_I(file_inode(file));
-	struct fsxattr fa;
-
-	simple_fill_fsxattr(&fa, btrfs_inode_flags_to_xflags(binode->flags));
-	if (copy_to_user(arg, &fa, sizeof(fa)))
-		return -EFAULT;
+	spin_lock(&fs_info->super_lock);
+	if (fs_info->exclusive_operation == type ||
+	    (fs_info->exclusive_operation == BTRFS_EXCLOP_BALANCE_PAUSED &&
+	     type == BTRFS_EXCLOP_DEV_ADD))
+		return true;
 
-	return 0;
+	spin_unlock(&fs_info->super_lock);
+	return false;
 }
 
-static int btrfs_ioctl_fssetxattr(struct file *file, void __user *arg)
+void btrfs_exclop_start_unlock(struct btrfs_fs_info *fs_info)
 {
-	struct inode *inode = file_inode(file);
-	struct btrfs_inode *binode = BTRFS_I(inode);
-	struct btrfs_root *root = binode->root;
-	struct btrfs_trans_handle *trans;
-	struct fsxattr fa, old_fa;
-	unsigned old_flags;
-	unsigned old_i_flags;
-	int ret = 0;
-
-	if (!inode_owner_or_capable(inode))
-		return -EPERM;
-
-	if (btrfs_root_readonly(root))
-		return -EROFS;
-
-	if (copy_from_user(&fa, arg, sizeof(fa)))
-		return -EFAULT;
-
-	ret = check_xflags(fa.fsx_xflags);
-	if (ret)
-		return ret;
-
-	if (fa.fsx_extsize != 0 || fa.fsx_projid != 0 || fa.fsx_cowextsize != 0)
-		return -EOPNOTSUPP;
-
-	ret = mnt_want_write_file(file);
-	if (ret)
-		return ret;
-
-	inode_lock(inode);
-
-	old_flags = binode->flags;
-	old_i_flags = inode->i_flags;
-
-	simple_fill_fsxattr(&old_fa,
-			    btrfs_inode_flags_to_xflags(binode->flags));
-	ret = vfs_ioc_fssetxattr_check(inode, &old_fa, &fa);
-	if (ret)
-		goto out_unlock;
-
-	if (fa.fsx_xflags & FS_XFLAG_SYNC)
-		binode->flags |= BTRFS_INODE_SYNC;
-	else
-		binode->flags &= ~BTRFS_INODE_SYNC;
-	if (fa.fsx_xflags & FS_XFLAG_IMMUTABLE)
-		binode->flags |= BTRFS_INODE_IMMUTABLE;
-	else
-		binode->flags &= ~BTRFS_INODE_IMMUTABLE;
-	if (fa.fsx_xflags & FS_XFLAG_APPEND)
-		binode->flags |= BTRFS_INODE_APPEND;
-	else
-		binode->flags &= ~BTRFS_INODE_APPEND;
-	if (fa.fsx_xflags & FS_XFLAG_NODUMP)
-		binode->flags |= BTRFS_INODE_NODUMP;
-	else
-		binode->flags &= ~BTRFS_INODE_NODUMP;
-	if (fa.fsx_xflags & FS_XFLAG_NOATIME)
-		binode->flags |= BTRFS_INODE_NOATIME;
-	else
-		binode->flags &= ~BTRFS_INODE_NOATIME;
-
-	/* 1 item for the inode */
-	trans = btrfs_start_transaction(root, 1);
-	if (IS_ERR(trans)) {
-		ret = PTR_ERR(trans);
-		goto out_unlock;
-	}
-
-	btrfs_sync_inode_flags_to_i_flags(inode);
-	inode_inc_iversion(inode);
-	inode->i_ctime = current_time(inode);
-	ret = btrfs_update_inode(trans, root, inode);
+	spin_unlock(&fs_info->super_lock);
+}
 
-	btrfs_end_transaction(trans);
+void btrfs_exclop_finish(struct btrfs_fs_info *fs_info)
+{
+	spin_lock(&fs_info->super_lock);
+	WRITE_ONCE(fs_info->exclusive_operation, BTRFS_EXCLOP_NONE);
+	spin_unlock(&fs_info->super_lock);
+	sysfs_notify(&fs_info->fs_devices->fsid_kobj, NULL, "exclusive_operation");
+}
 
-out_unlock:
-	if (ret) {
-		binode->flags = old_flags;
-		inode->i_flags = old_i_flags;
+void btrfs_exclop_balance(struct btrfs_fs_info *fs_info,
+			  enum btrfs_exclusive_operation op)
+{
+	switch (op) {
+	case BTRFS_EXCLOP_BALANCE_PAUSED:
+		spin_lock(&fs_info->super_lock);
+		ASSERT(fs_info->exclusive_operation == BTRFS_EXCLOP_BALANCE ||
+		       fs_info->exclusive_operation == BTRFS_EXCLOP_DEV_ADD);
+		fs_info->exclusive_operation = BTRFS_EXCLOP_BALANCE_PAUSED;
+		spin_unlock(&fs_info->super_lock);
+		break;
+	case BTRFS_EXCLOP_BALANCE:
+		spin_lock(&fs_info->super_lock);
+		ASSERT(fs_info->exclusive_operation == BTRFS_EXCLOP_BALANCE_PAUSED);
+		fs_info->exclusive_operation = BTRFS_EXCLOP_BALANCE;
+		spin_unlock(&fs_info->super_lock);
+		break;
+	default:
+		btrfs_warn(fs_info,
+			"invalid exclop balance operation %d requested", op);
 	}
-
-	inode_unlock(inode);
-	mnt_drop_write_file(file);
-
-	return ret;
 }
 
-static int btrfs_ioctl_getversion(struct file *file, int __user *arg)
+static int btrfs_ioctl_getversion(struct inode *inode, int __user *arg)
 {
-	struct inode *inode = file_inode(file);
-
 	return put_user(inode->i_generation, arg);
 }
 
@@ -483,7 +468,6 @@ static noinline int btrfs_ioctl_fitrim(struct btrfs_fs_info *fs_info,
 					void __user *arg)
 {
 	struct btrfs_device *device;
-	struct request_queue *q;
 	struct fstrim_range range;
 	u64 minlen = ULLONG_MAX;
 	u64 num_devices = 0;
@@ -493,6 +477,14 @@ static noinline int btrfs_ioctl_fitrim(struct btrfs_fs_info *fs_info,
 		return -EPERM;
 
 	/*
+	 * btrfs_trim_block_group() depends on space cache, which is not
+	 * available in zoned filesystem. So, disallow fitrim on a zoned
+	 * filesystem for now.
+	 */
+	if (btrfs_is_zoned(fs_info))
+		return -EOPNOTSUPP;
+
+	/*
 	 * If the fs is mounted with nologreplay, which requires it to be
 	 * mounted in RO mode as well, we can not allow discard on free space
 	 * inside block groups, because log trees refer to extents that are not
@@ -505,14 +497,11 @@ static noinline int btrfs_ioctl_fitrim(struct btrfs_fs_info *fs_info,
 	rcu_read_lock();
 	list_for_each_entry_rcu(device, &fs_info->fs_devices->devices,
 				dev_list) {
-		if (!device->bdev)
+		if (!device->bdev || !bdev_max_discard_sectors(device->bdev))
 			continue;
-		q = bdev_get_queue(device->bdev);
-		if (blk_queue_discard(q)) {
-			num_devices++;
-			minlen = min_t(u64, q->limits.discard_granularity,
-				     minlen);
-		}
+		num_devices++;
+		minlen = min_t(u64, bdev_discard_granularity(device->bdev),
+				    minlen);
 	}
 	rcu_read_unlock();
 
@@ -551,10 +540,35 @@ int __pure btrfs_is_empty_uuid(u8 *uuid)
 	return 1;
 }
 
-static noinline int create_subvol(struct inode *dir,
-				  struct dentry *dentry,
-				  const char *name, int namelen,
-				  u64 *async_transid,
+/*
+ * Calculate the number of transaction items to reserve for creating a subvolume
+ * or snapshot, not including the inode, directory entries, or parent directory.
+ */
+static unsigned int create_subvol_num_items(struct btrfs_qgroup_inherit *inherit)
+{
+	/*
+	 * 1 to add root block
+	 * 1 to add root item
+	 * 1 to add root ref
+	 * 1 to add root backref
+	 * 1 to add UUID item
+	 * 1 to add qgroup info
+	 * 1 to add qgroup limit
+	 *
+	 * Ideally the last two would only be accounted if qgroups are enabled,
+	 * but that can change between now and the time we would insert them.
+	 */
+	unsigned int num_items = 7;
+
+	if (inherit) {
+		/* 2 to add qgroup relations for each inherited qgroup */
+		num_items += 2 * inherit->num_qgroups;
+	}
+	return num_items;
+}
+
+static noinline int create_subvol(struct user_namespace *mnt_userns,
+				  struct inode *dir, struct dentry *dentry,
 				  struct btrfs_qgroup_inherit *inherit)
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb);
@@ -567,21 +581,23 @@ static noinline int create_subvol(struct inode *dir,
 	struct btrfs_root *new_root;
 	struct btrfs_block_rsv block_rsv;
 	struct timespec64 cur_time = current_time(dir);
-	struct inode *inode;
+	struct btrfs_new_inode_args new_inode_args = {
+		.dir = dir,
+		.dentry = dentry,
+		.subvol = true,
+	};
+	unsigned int trans_num_items;
 	int ret;
-	int err;
+	dev_t anon_dev;
 	u64 objectid;
-	u64 new_dirid = BTRFS_FIRST_FREE_OBJECTID;
-	u64 index = 0;
-	uuid_le new_uuid;
 
 	root_item = kzalloc(sizeof(*root_item), GFP_KERNEL);
 	if (!root_item)
 		return -ENOMEM;
 
-	ret = btrfs_find_free_objectid(fs_info->tree_root, &objectid);
+	ret = btrfs_get_free_objectid(fs_info->tree_root, &objectid);
 	if (ret)
-		goto fail_free;
+		goto out_root_item;
 
 	/*
 	 * Don't create subvolume whose level is not zero. Or qgroup will be
@@ -589,35 +605,47 @@ static noinline int create_subvol(struct inode *dir,
 	 */
 	if (btrfs_qgroup_level(objectid)) {
 		ret = -ENOSPC;
-		goto fail_free;
+		goto out_root_item;
 	}
 
+	ret = get_anon_bdev(&anon_dev);
+	if (ret < 0)
+		goto out_root_item;
+
+	new_inode_args.inode = btrfs_new_subvol_inode(mnt_userns, dir);
+	if (!new_inode_args.inode) {
+		ret = -ENOMEM;
+		goto out_anon_dev;
+	}
+	ret = btrfs_new_inode_prepare(&new_inode_args, &trans_num_items);
+	if (ret)
+		goto out_inode;
+	trans_num_items += create_subvol_num_items(inherit);
+
 	btrfs_init_block_rsv(&block_rsv, BTRFS_BLOCK_RSV_TEMP);
-	/*
-	 * The same as the snapshot creation, please see the comment
-	 * of create_snapshot().
-	 */
-	ret = btrfs_subvolume_reserve_metadata(root, &block_rsv, 8, false);
+	ret = btrfs_subvolume_reserve_metadata(root, &block_rsv,
+					       trans_num_items, false);
 	if (ret)
-		goto fail_free;
+		goto out_new_inode_args;
 
 	trans = btrfs_start_transaction(root, 0);
 	if (IS_ERR(trans)) {
 		ret = PTR_ERR(trans);
-		btrfs_subvolume_release_metadata(fs_info, &block_rsv);
-		goto fail_free;
+		btrfs_subvolume_release_metadata(root, &block_rsv);
+		goto out_new_inode_args;
 	}
 	trans->block_rsv = &block_rsv;
 	trans->bytes_reserved = block_rsv.size;
 
 	ret = btrfs_qgroup_inherit(trans, 0, objectid, inherit);
 	if (ret)
-		goto fail;
+		goto out;
 
-	leaf = btrfs_alloc_tree_block(trans, root, 0, objectid, NULL, 0, 0, 0);
+	leaf = btrfs_alloc_tree_block(trans, root, 0, objectid, NULL, 0, 0, 0,
+				      BTRFS_NESTING_NORMAL);
 	if (IS_ERR(leaf)) {
 		ret = PTR_ERR(leaf);
-		goto fail;
+		goto out;
 	}
 
 	btrfs_mark_buffer_dirty(leaf);
@@ -643,8 +671,7 @@ static noinline int create_subvol(struct inode *dir,
 
 	btrfs_set_root_generation_v2(root_item,
 			btrfs_root_generation(root_item));
-	uuid_le_gen(&new_uuid);
-	memcpy(root_item->uuid, new_uuid.b, BTRFS_UUID_SIZE);
+	generate_random_guid(root_item->uuid);
 	btrfs_set_stack_timespec_sec(&root_item->otime, cur_time.tv_sec);
 	btrfs_set_stack_timespec_nsec(&root_item->otime, cur_time.tv_nsec);
 	root_item->ctime = root_item->otime;
@@ -652,118 +679,106 @@ static noinline int create_subvol(struct inode *dir,
 	btrfs_set_root_otransid(root_item, trans->transid);
 
 	btrfs_tree_unlock(leaf);
-	free_extent_buffer(leaf);
-	leaf = NULL;
 
-	btrfs_set_root_dirid(root_item, new_dirid);
+	btrfs_set_root_dirid(root_item, BTRFS_FIRST_FREE_OBJECTID);
 
 	key.objectid = objectid;
 	key.offset = 0;
 	key.type = BTRFS_ROOT_ITEM_KEY;
 	ret = btrfs_insert_root(trans, fs_info->tree_root, &key,
 				root_item);
-	if (ret)
-		goto fail;
-
-	key.offset = (u64)-1;
-	new_root = btrfs_read_fs_root_no_name(fs_info, &key);
-	if (IS_ERR(new_root)) {
-		ret = PTR_ERR(new_root);
-		btrfs_abort_transaction(trans, ret);
-		goto fail;
-	}
-
-	btrfs_record_root_in_trans(trans, new_root);
-
-	ret = btrfs_create_subvol_root(trans, new_root, root, new_dirid);
 	if (ret) {
-		/* We potentially lose an unused inode item here */
-		btrfs_abort_transaction(trans, ret);
-		goto fail;
+		/*
+		 * Since we don't abort the transaction in this case, free the
+		 * tree block so that we don't leak space and leave the
+		 * filesystem in an inconsistent state (an extent item in the
+		 * extent tree with a backreference for a root that does not
+		 * exists).
+		 */
+		btrfs_tree_lock(leaf);
+		btrfs_clean_tree_block(leaf);
+		btrfs_tree_unlock(leaf);
+		btrfs_free_tree_block(trans, objectid, leaf, 0, 1);
+		free_extent_buffer(leaf);
+		goto out;
 	}
 
-	mutex_lock(&new_root->objectid_mutex);
-	new_root->highest_objectid = new_dirid;
-	mutex_unlock(&new_root->objectid_mutex);
+	free_extent_buffer(leaf);
+	leaf = NULL;
 
-	/*
-	 * insert the directory item
-	 */
-	ret = btrfs_set_inode_index(BTRFS_I(dir), &index);
-	if (ret) {
+	new_root = btrfs_get_new_fs_root(fs_info, objectid, anon_dev);
+	if (IS_ERR(new_root)) {
+		ret = PTR_ERR(new_root);
 		btrfs_abort_transaction(trans, ret);
-		goto fail;
+		goto out;
 	}
+	/* anon_dev is owned by new_root now. */
+	anon_dev = 0;
+	BTRFS_I(new_inode_args.inode)->root = new_root;
+	/* ... and new_root is owned by new_inode_args.inode now. */
 
-	ret = btrfs_insert_dir_item(trans, name, namelen, BTRFS_I(dir), &key,
-				    BTRFS_FT_DIR, index);
+	ret = btrfs_record_root_in_trans(trans, new_root);
 	if (ret) {
 		btrfs_abort_transaction(trans, ret);
-		goto fail;
+		goto out;
 	}
 
-	btrfs_i_size_write(BTRFS_I(dir), dir->i_size + namelen * 2);
-	ret = btrfs_update_inode(trans, root, dir);
+	ret = btrfs_uuid_tree_add(trans, root_item->uuid,
+				  BTRFS_UUID_KEY_SUBVOL, objectid);
 	if (ret) {
 		btrfs_abort_transaction(trans, ret);
-		goto fail;
+		goto out;
 	}
 
-	ret = btrfs_add_root_ref(trans, objectid, root->root_key.objectid,
-				 btrfs_ino(BTRFS_I(dir)), index, name, namelen);
+	ret = btrfs_create_new_inode(trans, &new_inode_args);
 	if (ret) {
 		btrfs_abort_transaction(trans, ret);
-		goto fail;
+		goto out;
 	}
 
-	ret = btrfs_uuid_tree_add(trans, root_item->uuid,
-				  BTRFS_UUID_KEY_SUBVOL, objectid);
-	if (ret)
-		btrfs_abort_transaction(trans, ret);
+	d_instantiate_new(dentry, new_inode_args.inode);
+	new_inode_args.inode = NULL;
 
-fail:
-	kfree(root_item);
+out:
 	trans->block_rsv = NULL;
 	trans->bytes_reserved = 0;
-	btrfs_subvolume_release_metadata(fs_info, &block_rsv);
-
-	if (async_transid) {
-		*async_transid = trans->transid;
-		err = btrfs_commit_transaction_async(trans, 1);
-		if (err)
-			err = btrfs_commit_transaction(trans);
-	} else {
-		err = btrfs_commit_transaction(trans);
-	}
-	if (err && !ret)
-		ret = err;
+	btrfs_subvolume_release_metadata(root, &block_rsv);
 
-	if (!ret) {
-		inode = btrfs_lookup_dentry(dir, dentry);
-		if (IS_ERR(inode))
-			return PTR_ERR(inode);
-		d_instantiate(dentry, inode);
-	}
-	return ret;
-
-fail_free:
+	if (ret)
+		btrfs_end_transaction(trans);
+	else
+		ret = btrfs_commit_transaction(trans);
+out_new_inode_args:
+	btrfs_new_inode_args_destroy(&new_inode_args);
+out_inode:
+	iput(new_inode_args.inode);
+out_anon_dev:
+	if (anon_dev)
+		free_anon_bdev(anon_dev);
+out_root_item:
 	kfree(root_item);
 	return ret;
 }
 
 static int create_snapshot(struct btrfs_root *root, struct inode *dir,
-			   struct dentry *dentry,
-			   u64 *async_transid, bool readonly,
+			   struct dentry *dentry, bool readonly,
 			   struct btrfs_qgroup_inherit *inherit)
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb);
 	struct inode *inode;
 	struct btrfs_pending_snapshot *pending_snapshot;
+	unsigned int trans_num_items;
 	struct btrfs_trans_handle *trans;
 	int ret;
-	bool snapshot_force_cow = false;
 
-	if (!test_bit(BTRFS_ROOT_REF_COWS, &root->state))
+	/* We do not support snapshotting right now. */
+	if (btrfs_fs_incompat(fs_info, EXTENT_TREE_V2)) {
+		btrfs_warn(fs_info,
+			   "extent tree v2 doesn't support snapshotting yet");
+		return -EOPNOTSUPP;
+	}
+
+	if (!test_bit(BTRFS_ROOT_SHAREABLE, &root->state))
 		return -EINVAL;
 
 	if (atomic_read(&root->nr_swapfiles)) {
@@ -776,6 +791,9 @@ static int create_snapshot(struct btrfs_root *root, struct inode *dir,
 	if (!pending_snapshot)
 		return -ENOMEM;
 
+	ret = get_anon_bdev(&pending_snapshot->anon_dev);
+	if (ret < 0)
+		goto free_pending;
 	pending_snapshot->root_item = kzalloc(sizeof(struct btrfs_root_item),
 			GFP_KERNEL);
 	pending_snapshot->path = btrfs_alloc_path();
@@ -784,46 +802,19 @@ static int create_snapshot(struct btrfs_root *root, struct inode *dir,
 		goto free_pending;
 	}
 
-	/*
-	 * Force new buffered writes to reserve space even when NOCOW is
-	 * possible. This is to avoid later writeback (running dealloc) to
-	 * fallback to COW mode and unexpectedly fail with ENOSPC.
-	 */
-	atomic_inc(&root->will_be_snapshotted);
-	smp_mb__after_atomic();
-	/* wait for no snapshot writes */
-	wait_event(root->subv_writers->wait,
-		   percpu_counter_sum(&root->subv_writers->counter) == 0);
-
-	ret = btrfs_start_delalloc_snapshot(root);
-	if (ret)
-		goto dec_and_free;
-
-	/*
-	 * All previous writes have started writeback in NOCOW mode, so now
-	 * we force future writes to fallback to COW mode during snapshot
-	 * creation.
-	 */
-	atomic_inc(&root->snapshot_force_cow);
-	snapshot_force_cow = true;
-
-	btrfs_wait_ordered_extents(root, U64_MAX, 0, (u64)-1);
-
 	btrfs_init_block_rsv(&pending_snapshot->block_rsv,
 			     BTRFS_BLOCK_RSV_TEMP);
 	/*
-	 * 1 - parent dir inode
-	 * 2 - dir entries
-	 * 1 - root item
-	 * 2 - root ref/backref
-	 * 1 - root of snapshot
-	 * 1 - UUID item
+	 * 1 to add dir item
+	 * 1 to add dir index
+	 * 1 to update parent inode item
 	 */
+	trans_num_items = create_subvol_num_items(inherit) + 3;
 	ret = btrfs_subvolume_reserve_metadata(BTRFS_I(dir)->root,
-					&pending_snapshot->block_rsv, 8,
-					false);
+					       &pending_snapshot->block_rsv,
+					       trans_num_items, false);
 	if (ret)
-		goto dec_and_free;
+		goto free_pending;
 
 	pending_snapshot->dentry = dentry;
 	pending_snapshot->root = root;
@@ -837,18 +828,9 @@ static int create_snapshot(struct btrfs_root *root, struct inode *dir,
 		goto fail;
 	}
 
-	spin_lock(&fs_info->trans_lock);
-	list_add(&pending_snapshot->list,
-		 &trans->transaction->pending_snapshots);
-	spin_unlock(&fs_info->trans_lock);
-	if (async_transid) {
-		*async_transid = trans->transid;
-		ret = btrfs_commit_transaction_async(trans, 1);
-		if (ret)
-			ret = btrfs_commit_transaction(trans);
-	} else {
-		ret = btrfs_commit_transaction(trans);
-	}
+	trans->pending_snapshot = pending_snapshot;
+
+	ret = btrfs_commit_transaction(trans);
 	if (ret)
 		goto fail;
 
@@ -868,14 +850,16 @@ static int create_snapshot(struct btrfs_root *root, struct inode *dir,
 
 	d_instantiate(dentry, inode);
 	ret = 0;
+	pending_snapshot->anon_dev = 0;
 fail:
-	btrfs_subvolume_release_metadata(fs_info, &pending_snapshot->block_rsv);
-dec_and_free:
-	if (snapshot_force_cow)
-		atomic_dec(&root->snapshot_force_cow);
-	if (atomic_dec_and_test(&root->will_be_snapshotted))
-		wake_up_var(&root->will_be_snapshotted);
+	/* Prevent double freeing of anon_dev */
+	if (ret && pending_snapshot->snap)
+		pending_snapshot->snap->anon_dev = 0;
+	btrfs_put_root(pending_snapshot->snap);
+	btrfs_subvolume_release_metadata(root, &pending_snapshot->block_rsv);
 free_pending:
+	if (pending_snapshot->anon_dev)
+		free_anon_bdev(pending_snapshot->anon_dev);
 	kfree(pending_snapshot->root_item);
 	btrfs_free_path(pending_snapshot->path);
 	kfree(pending_snapshot);
@@ -903,7 +887,8 @@ free_pending:
  *     nfs_async_unlink().
  */
 
-static int btrfs_may_delete(struct inode *dir, struct dentry *victim, int isdir)
+static int btrfs_may_delete(struct user_namespace *mnt_userns,
+			    struct inode *dir, struct dentry *victim, int isdir)
 {
 	int error;
 
@@ -913,13 +898,14 @@ static int btrfs_may_delete(struct inode *dir, struct dentry *victim, int isdir)
 	BUG_ON(d_inode(victim->d_parent) != dir);
 	audit_inode_child(dir, victim, AUDIT_TYPE_CHILD_DELETE);
 
-	error = inode_permission(dir, MAY_WRITE | MAY_EXEC);
+	error = inode_permission(mnt_userns, dir, MAY_WRITE | MAY_EXEC);
 	if (error)
 		return error;
 	if (IS_APPEND(dir))
 		return -EPERM;
-	if (check_sticky(dir, d_inode(victim)) || IS_APPEND(d_inode(victim)) ||
-	    IS_IMMUTABLE(d_inode(victim)) || IS_SWAPFILE(d_inode(victim)))
+	if (check_sticky(mnt_userns, dir, d_inode(victim)) ||
+	    IS_APPEND(d_inode(victim)) || IS_IMMUTABLE(d_inode(victim)) ||
+	    IS_SWAPFILE(d_inode(victim)))
 		return -EPERM;
 	if (isdir) {
 		if (!d_is_dir(victim))
@@ -936,13 +922,16 @@ static int btrfs_may_delete(struct inode *dir, struct dentry *victim, int isdir)
 }
 
 /* copy of may_create in fs/namei.c() */
-static inline int btrfs_may_create(struct inode *dir, struct dentry *child)
+static inline int btrfs_may_create(struct user_namespace *mnt_userns,
+				   struct inode *dir, struct dentry *child)
 {
 	if (d_really_is_positive(child))
 		return -EEXIST;
 	if (IS_DEADDIR(dir))
 		return -ENOENT;
-	return inode_permission(dir, MAY_WRITE | MAY_EXEC);
+	if (!fsuidgid_has_mapping(dir->i_sb, mnt_userns))
+		return -EOVERFLOW;
+	return inode_permission(mnt_userns, dir, MAY_WRITE | MAY_EXEC);
 }
 
 /*
@@ -951,9 +940,10 @@ static inline int btrfs_may_create(struct inode *dir, struct dentry *child)
  * inside this filesystem so it's quite a bit simpler.
  */
 static noinline int btrfs_mksubvol(const struct path *parent,
+				   struct user_namespace *mnt_userns,
 				   const char *name, int namelen,
 				   struct btrfs_root *snap_src,
-				   u64 *async_transid, bool readonly,
+				   bool readonly,
 				   struct btrfs_qgroup_inherit *inherit)
 {
 	struct inode *dir = d_inode(parent->dentry);
@@ -965,12 +955,12 @@ static noinline int btrfs_mksubvol(const struct path *parent,
 	if (error == -EINTR)
 		return error;
 
-	dentry = lookup_one_len(name, parent->dentry, namelen);
+	dentry = lookup_one(mnt_userns, name, parent->dentry, namelen);
 	error = PTR_ERR(dentry);
 	if (IS_ERR(dentry))
 		goto out_unlock;
 
-	error = btrfs_may_create(dir, dentry);
+	error = btrfs_may_create(mnt_userns, dir, dentry);
 	if (error)
 		goto out_dput;
 
@@ -989,13 +979,11 @@ static noinline int btrfs_mksubvol(const struct path *parent,
 	if (btrfs_root_refs(&BTRFS_I(dir)->root->root_item) == 0)
 		goto out_up_read;
 
-	if (snap_src) {
-		error = create_snapshot(snap_src, dir, dentry,
-					async_transid, readonly, inherit);
-	} else {
-		error = create_subvol(dir, dentry, name, namelen,
-				      async_transid, inherit);
-	}
+	if (snap_src)
+		error = create_snapshot(snap_src, dir, dentry, readonly, inherit);
+	else
+		error = create_subvol(mnt_userns, dir, dentry, inherit);
+
 	if (!error)
 		fsnotify_mkdir(dir, dentry);
 out_up_read:
@@ -1003,133 +991,237 @@ out_up_read:
 out_dput:
 	dput(dentry);
 out_unlock:
-	inode_unlock(dir);
+	btrfs_inode_unlock(dir, 0);
 	return error;
 }
 
-/*
- * When we're defragging a range, we don't want to kick it off again
- * if it is really just waiting for delalloc to send it down.
- * If we find a nice big extent or delalloc range for the bytes in the
- * file you want to defrag, we return 0 to let you know to skip this
- * part of the file
- */
-static int check_defrag_in_cache(struct inode *inode, u64 offset, u32 thresh)
+static noinline int btrfs_mksnapshot(const struct path *parent,
+				   struct user_namespace *mnt_userns,
+				   const char *name, int namelen,
+				   struct btrfs_root *root,
+				   bool readonly,
+				   struct btrfs_qgroup_inherit *inherit)
 {
-	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
-	struct extent_map *em = NULL;
-	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
-	u64 end;
+	int ret;
+	bool snapshot_force_cow = false;
 
-	read_lock(&em_tree->lock);
-	em = lookup_extent_mapping(em_tree, offset, PAGE_SIZE);
-	read_unlock(&em_tree->lock);
+	/*
+	 * Force new buffered writes to reserve space even when NOCOW is
+	 * possible. This is to avoid later writeback (running dealloc) to
+	 * fallback to COW mode and unexpectedly fail with ENOSPC.
+	 */
+	btrfs_drew_read_lock(&root->snapshot_lock);
 
-	if (em) {
-		end = extent_map_end(em);
-		free_extent_map(em);
-		if (end - offset > thresh)
-			return 0;
-	}
-	/* if we already have a nice delalloc here, just stop */
-	thresh /= 2;
-	end = count_range_bits(io_tree, &offset, offset + thresh,
-			       thresh, EXTENT_DELALLOC, 1);
-	if (end >= thresh)
-		return 0;
-	return 1;
+	ret = btrfs_start_delalloc_snapshot(root, false);
+	if (ret)
+		goto out;
+
+	/*
+	 * All previous writes have started writeback in NOCOW mode, so now
+	 * we force future writes to fallback to COW mode during snapshot
+	 * creation.
+	 */
+	atomic_inc(&root->snapshot_force_cow);
+	snapshot_force_cow = true;
+
+	btrfs_wait_ordered_extents(root, U64_MAX, 0, (u64)-1);
+
+	ret = btrfs_mksubvol(parent, mnt_userns, name, namelen,
+			     root, readonly, inherit);
+out:
+	if (snapshot_force_cow)
+		atomic_dec(&root->snapshot_force_cow);
+	btrfs_drew_read_unlock(&root->snapshot_lock);
+	return ret;
 }
 
 /*
- * helper function to walk through a file and find extents
- * newer than a specific transid, and smaller than thresh.
+ * Defrag specific helper to get an extent map.
+ *
+ * Differences between this and btrfs_get_extent() are:
+ *
+ * - No extent_map will be added to inode->extent_tree
+ *   To reduce memory usage in the long run.
  *
- * This is used by the defragging code to find new and small
- * extents
+ * - Extra optimization to skip file extents older than @newer_than
+ *   By using btrfs_search_forward() we can skip entire file ranges that
+ *   have extents created in past transactions, because btrfs_search_forward()
+ *   will not visit leaves and nodes with a generation smaller than given
+ *   minimal generation threshold (@newer_than).
+ *
+ * Return valid em if we find a file extent matching the requirement.
+ * Return NULL if we can not find a file extent matching the requirement.
+ *
+ * Return ERR_PTR() for error.
  */
-static int find_new_extents(struct btrfs_root *root,
-			    struct inode *inode, u64 newer_than,
-			    u64 *off, u32 thresh)
+static struct extent_map *defrag_get_extent(struct btrfs_inode *inode,
+					    u64 start, u64 newer_than)
 {
-	struct btrfs_path *path;
-	struct btrfs_key min_key;
-	struct extent_buffer *leaf;
-	struct btrfs_file_extent_item *extent;
-	int type;
+	struct btrfs_root *root = inode->root;
+	struct btrfs_file_extent_item *fi;
+	struct btrfs_path path = { 0 };
+	struct extent_map *em;
+	struct btrfs_key key;
+	u64 ino = btrfs_ino(inode);
 	int ret;
-	u64 ino = btrfs_ino(BTRFS_I(inode));
 
-	path = btrfs_alloc_path();
-	if (!path)
-		return -ENOMEM;
+	em = alloc_extent_map();
+	if (!em) {
+		ret = -ENOMEM;
+		goto err;
+	}
 
-	min_key.objectid = ino;
-	min_key.type = BTRFS_EXTENT_DATA_KEY;
-	min_key.offset = *off;
+	key.objectid = ino;
+	key.type = BTRFS_EXTENT_DATA_KEY;
+	key.offset = start;
 
-	while (1) {
-		ret = btrfs_search_forward(root, &min_key, path, newer_than);
-		if (ret != 0)
-			goto none;
-process_slot:
-		if (min_key.objectid != ino)
-			goto none;
-		if (min_key.type != BTRFS_EXTENT_DATA_KEY)
-			goto none;
+	if (newer_than) {
+		ret = btrfs_search_forward(root, &key, &path, newer_than);
+		if (ret < 0)
+			goto err;
+		/* Can't find anything newer */
+		if (ret > 0)
+			goto not_found;
+	} else {
+		ret = btrfs_search_slot(NULL, root, &key, &path, 0, 0);
+		if (ret < 0)
+			goto err;
+	}
+	if (path.slots[0] >= btrfs_header_nritems(path.nodes[0])) {
+		/*
+		 * If btrfs_search_slot() makes path to point beyond nritems,
+		 * we should not have an empty leaf, as this inode must at
+		 * least have its INODE_ITEM.
+		 */
+		ASSERT(btrfs_header_nritems(path.nodes[0]));
+		path.slots[0] = btrfs_header_nritems(path.nodes[0]) - 1;
+	}
+	btrfs_item_key_to_cpu(path.nodes[0], &key, path.slots[0]);
+	/* Perfect match, no need to go one slot back */
+	if (key.objectid == ino && key.type == BTRFS_EXTENT_DATA_KEY &&
+	    key.offset == start)
+		goto iterate;
 
-		leaf = path->nodes[0];
-		extent = btrfs_item_ptr(leaf, path->slots[0],
-					struct btrfs_file_extent_item);
-
-		type = btrfs_file_extent_type(leaf, extent);
-		if (type == BTRFS_FILE_EXTENT_REG &&
-		    btrfs_file_extent_num_bytes(leaf, extent) < thresh &&
-		    check_defrag_in_cache(inode, min_key.offset, thresh)) {
-			*off = min_key.offset;
-			btrfs_free_path(path);
-			return 0;
-		}
+	/* We didn't find a perfect match, needs to go one slot back */
+	if (path.slots[0] > 0) {
+		btrfs_item_key_to_cpu(path.nodes[0], &key, path.slots[0]);
+		if (key.objectid == ino && key.type == BTRFS_EXTENT_DATA_KEY)
+			path.slots[0]--;
+	}
+
+iterate:
+	/* Iterate through the path to find a file extent covering @start */
+	while (true) {
+		u64 extent_end;
+
+		if (path.slots[0] >= btrfs_header_nritems(path.nodes[0]))
+			goto next;
+
+		btrfs_item_key_to_cpu(path.nodes[0], &key, path.slots[0]);
+
+		/*
+		 * We may go one slot back to INODE_REF/XATTR item, then
+		 * need to go forward until we reach an EXTENT_DATA.
+		 * But we should still has the correct ino as key.objectid.
+		 */
+		if (WARN_ON(key.objectid < ino) || key.type < BTRFS_EXTENT_DATA_KEY)
+			goto next;
+
+		/* It's beyond our target range, definitely not extent found */
+		if (key.objectid > ino || key.type > BTRFS_EXTENT_DATA_KEY)
+			goto not_found;
 
-		path->slots[0]++;
-		if (path->slots[0] < btrfs_header_nritems(leaf)) {
-			btrfs_item_key_to_cpu(leaf, &min_key, path->slots[0]);
-			goto process_slot;
+		/*
+		 *	|	|<- File extent ->|
+		 *	\- start
+		 *
+		 * This means there is a hole between start and key.offset.
+		 */
+		if (key.offset > start) {
+			em->start = start;
+			em->orig_start = start;
+			em->block_start = EXTENT_MAP_HOLE;
+			em->len = key.offset - start;
+			break;
 		}
 
-		if (min_key.offset == (u64)-1)
-			goto none;
+		fi = btrfs_item_ptr(path.nodes[0], path.slots[0],
+				    struct btrfs_file_extent_item);
+		extent_end = btrfs_file_extent_end(&path);
 
-		min_key.offset++;
-		btrfs_release_path(path);
+		/*
+		 *	|<- file extent ->|	|
+		 *				\- start
+		 *
+		 * We haven't reached start, search next slot.
+		 */
+		if (extent_end <= start)
+			goto next;
+
+		/* Now this extent covers @start, convert it to em */
+		btrfs_extent_item_to_extent_map(inode, &path, fi, false, em);
+		break;
+next:
+		ret = btrfs_next_item(root, &path);
+		if (ret < 0)
+			goto err;
+		if (ret > 0)
+			goto not_found;
 	}
-none:
-	btrfs_free_path(path);
-	return -ENOENT;
+	btrfs_release_path(&path);
+	return em;
+
+not_found:
+	btrfs_release_path(&path);
+	free_extent_map(em);
+	return NULL;
+
+err:
+	btrfs_release_path(&path);
+	free_extent_map(em);
+	return ERR_PTR(ret);
 }
 
-static struct extent_map *defrag_lookup_extent(struct inode *inode, u64 start)
+static struct extent_map *defrag_lookup_extent(struct inode *inode, u64 start,
+					       u64 newer_than, bool locked)
 {
 	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
 	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
 	struct extent_map *em;
-	u64 len = PAGE_SIZE;
+	const u32 sectorsize = BTRFS_I(inode)->root->fs_info->sectorsize;
 
 	/*
 	 * hopefully we have this extent in the tree already, try without
 	 * the full extent lock
 	 */
 	read_lock(&em_tree->lock);
-	em = lookup_extent_mapping(em_tree, start, len);
+	em = lookup_extent_mapping(em_tree, start, sectorsize);
 	read_unlock(&em_tree->lock);
 
+	/*
+	 * We can get a merged extent, in that case, we need to re-search
+	 * tree to get the original em for defrag.
+	 *
+	 * If @newer_than is 0 or em::generation < newer_than, we can trust
+	 * this em, as either we don't care about the generation, or the
+	 * merged extent map will be rejected anyway.
+	 */
+	if (em && test_bit(EXTENT_FLAG_MERGED, &em->flags) &&
+	    newer_than && em->generation >= newer_than) {
+		free_extent_map(em);
+		em = NULL;
+	}
+
 	if (!em) {
 		struct extent_state *cached = NULL;
-		u64 end = start + len - 1;
+		u64 end = start + sectorsize - 1;
 
 		/* get the big lock and read metadata off disk */
-		lock_extent_bits(io_tree, start, end, &cached);
-		em = btrfs_get_extent(BTRFS_I(inode), NULL, 0, start, len);
-		unlock_extent_cached(io_tree, start, end, &cached);
+		if (!locked)
+			lock_extent(io_tree, start, end, &cached);
+		em = defrag_get_extent(BTRFS_I(inode), start, newer_than);
+		if (!locked)
+			unlock_extent(io_tree, start, end, &cached);
 
 		if (IS_ERR(em))
 			return NULL;
@@ -1138,274 +1230,586 @@ static struct extent_map *defrag_lookup_extent(struct inode *inode, u64 start)
 	return em;
 }
 
-static bool defrag_check_next_extent(struct inode *inode, struct extent_map *em)
+static u32 get_extent_max_capacity(const struct btrfs_fs_info *fs_info,
+				   const struct extent_map *em)
+{
+	if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags))
+		return BTRFS_MAX_COMPRESSED;
+	return fs_info->max_extent_size;
+}
+
+static bool defrag_check_next_extent(struct inode *inode, struct extent_map *em,
+				     u32 extent_thresh, u64 newer_than, bool locked)
 {
+	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	struct extent_map *next;
-	bool ret = true;
+	bool ret = false;
 
 	/* this is the last extent */
 	if (em->start + em->len >= i_size_read(inode))
 		return false;
 
-	next = defrag_lookup_extent(inode, em->start + em->len);
+	/*
+	 * Here we need to pass @newer_then when checking the next extent, or
+	 * we will hit a case we mark current extent for defrag, but the next
+	 * one will not be a target.
+	 * This will just cause extra IO without really reducing the fragments.
+	 */
+	next = defrag_lookup_extent(inode, em->start + em->len, newer_than, locked);
+	/* No more em or hole */
 	if (!next || next->block_start >= EXTENT_MAP_LAST_BYTE)
-		ret = false;
-	else if ((em->block_start + em->block_len == next->block_start) &&
-		 (em->block_len > SZ_128K && next->block_len > SZ_128K))
-		ret = false;
+		goto out;
+	if (test_bit(EXTENT_FLAG_PREALLOC, &next->flags))
+		goto out;
+	/*
+	 * If the next extent is at its max capacity, defragging current extent
+	 * makes no sense, as the total number of extents won't change.
+	 */
+	if (next->len >= get_extent_max_capacity(fs_info, em))
+		goto out;
+	/* Skip older extent */
+	if (next->generation < newer_than)
+		goto out;
+	/* Also check extent size */
+	if (next->len >= extent_thresh)
+		goto out;
 
+	ret = true;
+out:
 	free_extent_map(next);
 	return ret;
 }
 
-static int should_defrag_range(struct inode *inode, u64 start, u32 thresh,
-			       u64 *last_len, u64 *skip, u64 *defrag_end,
-			       int compress)
+/*
+ * Prepare one page to be defragged.
+ *
+ * This will ensure:
+ *
+ * - Returned page is locked and has been set up properly.
+ * - No ordered extent exists in the page.
+ * - The page is uptodate.
+ *
+ * NOTE: Caller should also wait for page writeback after the cluster is
+ * prepared, here we don't do writeback wait for each page.
+ */
+static struct page *defrag_prepare_one_page(struct btrfs_inode *inode,
+					    pgoff_t index)
 {
-	struct extent_map *em;
-	int ret = 1;
-	bool next_mergeable = true;
-	bool prev_mergeable = true;
+	struct address_space *mapping = inode->vfs_inode.i_mapping;
+	gfp_t mask = btrfs_alloc_write_mask(mapping);
+	u64 page_start = (u64)index << PAGE_SHIFT;
+	u64 page_end = page_start + PAGE_SIZE - 1;
+	struct extent_state *cached_state = NULL;
+	struct page *page;
+	int ret;
+
+again:
+	page = find_or_create_page(mapping, index, mask);
+	if (!page)
+		return ERR_PTR(-ENOMEM);
 
 	/*
-	 * make sure that once we start defragging an extent, we keep on
-	 * defragging it
+	 * Since we can defragment files opened read-only, we can encounter
+	 * transparent huge pages here (see CONFIG_READ_ONLY_THP_FOR_FS). We
+	 * can't do I/O using huge pages yet, so return an error for now.
+	 * Filesystem transparent huge pages are typically only used for
+	 * executables that explicitly enable them, so this isn't very
+	 * restrictive.
 	 */
-	if (start < *defrag_end)
-		return 1;
+	if (PageCompound(page)) {
+		unlock_page(page);
+		put_page(page);
+		return ERR_PTR(-ETXTBSY);
+	}
 
-	*skip = 0;
+	ret = set_page_extent_mapped(page);
+	if (ret < 0) {
+		unlock_page(page);
+		put_page(page);
+		return ERR_PTR(ret);
+	}
 
-	em = defrag_lookup_extent(inode, start);
-	if (!em)
-		return 0;
+	/* Wait for any existing ordered extent in the range */
+	while (1) {
+		struct btrfs_ordered_extent *ordered;
 
-	/* this will cover holes, and inline extents */
-	if (em->block_start >= EXTENT_MAP_LAST_BYTE) {
-		ret = 0;
-		goto out;
-	}
+		lock_extent(&inode->io_tree, page_start, page_end, &cached_state);
+		ordered = btrfs_lookup_ordered_range(inode, page_start, PAGE_SIZE);
+		unlock_extent(&inode->io_tree, page_start, page_end,
+			      &cached_state);
+		if (!ordered)
+			break;
 
-	if (!*defrag_end)
-		prev_mergeable = false;
+		unlock_page(page);
+		btrfs_start_ordered_extent(ordered, 1);
+		btrfs_put_ordered_extent(ordered);
+		lock_page(page);
+		/*
+		 * We unlocked the page above, so we need check if it was
+		 * released or not.
+		 */
+		if (page->mapping != mapping || !PagePrivate(page)) {
+			unlock_page(page);
+			put_page(page);
+			goto again;
+		}
+	}
 
-	next_mergeable = defrag_check_next_extent(inode, em);
-	/*
-	 * we hit a real extent, if it is big or the next extent is not a
-	 * real extent, don't bother defragging it
-	 */
-	if (!compress && (*last_len == 0 || *last_len >= thresh) &&
-	    (em->len >= thresh || (!next_mergeable && !prev_mergeable)))
-		ret = 0;
-out:
 	/*
-	 * last_len ends up being a counter of how many bytes we've defragged.
-	 * every time we choose not to defrag an extent, we reset *last_len
-	 * so that the next tiny extent will force a defrag.
-	 *
-	 * The end result of this is that tiny extents before a single big
-	 * extent will force at least part of that big extent to be defragged.
+	 * Now the page range has no ordered extent any more.  Read the page to
+	 * make it uptodate.
 	 */
-	if (ret) {
-		*defrag_end = extent_map_end(em);
-	} else {
-		*last_len = 0;
-		*skip = extent_map_end(em);
-		*defrag_end = 0;
+	if (!PageUptodate(page)) {
+		btrfs_read_folio(NULL, page_folio(page));
+		lock_page(page);
+		if (page->mapping != mapping || !PagePrivate(page)) {
+			unlock_page(page);
+			put_page(page);
+			goto again;
+		}
+		if (!PageUptodate(page)) {
+			unlock_page(page);
+			put_page(page);
+			return ERR_PTR(-EIO);
+		}
 	}
-
-	free_extent_map(em);
-	return ret;
+	return page;
 }
 
+struct defrag_target_range {
+	struct list_head list;
+	u64 start;
+	u64 len;
+};
+
 /*
- * it doesn't do much good to defrag one or two pages
- * at a time.  This pulls in a nice chunk of pages
- * to COW and defrag.
- *
- * It also makes sure the delalloc code has enough
- * dirty data to avoid making new small extents as part
- * of the defrag
+ * Collect all valid target extents.
  *
- * It's a good idea to start RA on this range
- * before calling this.
+ * @start:	   file offset to lookup
+ * @len:	   length to lookup
+ * @extent_thresh: file extent size threshold, any extent size >= this value
+ *		   will be ignored
+ * @newer_than:    only defrag extents newer than this value
+ * @do_compress:   whether the defrag is doing compression
+ *		   if true, @extent_thresh will be ignored and all regular
+ *		   file extents meeting @newer_than will be targets.
+ * @locked:	   if the range has already held extent lock
+ * @target_list:   list of targets file extents
  */
-static int cluster_pages_for_defrag(struct inode *inode,
-				    struct page **pages,
-				    unsigned long start_index,
-				    unsigned long num_pages)
+static int defrag_collect_targets(struct btrfs_inode *inode,
+				  u64 start, u64 len, u32 extent_thresh,
+				  u64 newer_than, bool do_compress,
+				  bool locked, struct list_head *target_list,
+				  u64 *last_scanned_ret)
 {
-	unsigned long file_end;
-	u64 isize = i_size_read(inode);
-	u64 page_start;
-	u64 page_end;
-	u64 page_cnt;
-	int ret;
-	int i;
-	int i_done;
-	struct btrfs_ordered_extent *ordered;
-	struct extent_state *cached_state = NULL;
-	struct extent_io_tree *tree;
-	struct extent_changeset *data_reserved = NULL;
-	gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping);
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
+	bool last_is_target = false;
+	u64 cur = start;
+	int ret = 0;
 
-	file_end = (isize - 1) >> PAGE_SHIFT;
-	if (!isize || start_index > file_end)
-		return 0;
+	while (cur < start + len) {
+		struct extent_map *em;
+		struct defrag_target_range *new;
+		bool next_mergeable = true;
+		u64 range_len;
 
-	page_cnt = min_t(u64, (u64)num_pages, (u64)file_end - start_index + 1);
+		last_is_target = false;
+		em = defrag_lookup_extent(&inode->vfs_inode, cur,
+					  newer_than, locked);
+		if (!em)
+			break;
 
-	ret = btrfs_delalloc_reserve_space(inode, &data_reserved,
-			start_index << PAGE_SHIFT,
-			page_cnt << PAGE_SHIFT);
-	if (ret)
-		return ret;
-	i_done = 0;
-	tree = &BTRFS_I(inode)->io_tree;
+		/*
+		 * If the file extent is an inlined one, we may still want to
+		 * defrag it (fallthrough) if it will cause a regular extent.
+		 * This is for users who want to convert inline extents to
+		 * regular ones through max_inline= mount option.
+		 */
+		if (em->block_start == EXTENT_MAP_INLINE &&
+		    em->len <= inode->root->fs_info->max_inline)
+			goto next;
 
-	/* step one, lock all the pages */
-	for (i = 0; i < page_cnt; i++) {
-		struct page *page;
-again:
-		page = find_or_create_page(inode->i_mapping,
-					   start_index + i, mask);
-		if (!page)
-			break;
+		/* Skip hole/delalloc/preallocated extents */
+		if (em->block_start == EXTENT_MAP_HOLE ||
+		    em->block_start == EXTENT_MAP_DELALLOC ||
+		    test_bit(EXTENT_FLAG_PREALLOC, &em->flags))
+			goto next;
 
-		page_start = page_offset(page);
-		page_end = page_start + PAGE_SIZE - 1;
-		while (1) {
-			lock_extent_bits(tree, page_start, page_end,
-					 &cached_state);
-			ordered = btrfs_lookup_ordered_extent(inode,
-							      page_start);
-			unlock_extent_cached(tree, page_start, page_end,
-					     &cached_state);
-			if (!ordered)
-				break;
+		/* Skip older extent */
+		if (em->generation < newer_than)
+			goto next;
 
-			unlock_page(page);
-			btrfs_start_ordered_extent(inode, ordered, 1);
-			btrfs_put_ordered_extent(ordered);
-			lock_page(page);
-			/*
-			 * we unlocked the page above, so we need check if
-			 * it was released or not.
-			 */
-			if (page->mapping != inode->i_mapping) {
-				unlock_page(page);
-				put_page(page);
-				goto again;
-			}
+		/* This em is under writeback, no need to defrag */
+		if (em->generation == (u64)-1)
+			goto next;
+
+		/*
+		 * Our start offset might be in the middle of an existing extent
+		 * map, so take that into account.
+		 */
+		range_len = em->len - (cur - em->start);
+		/*
+		 * If this range of the extent map is already flagged for delalloc,
+		 * skip it, because:
+		 *
+		 * 1) We could deadlock later, when trying to reserve space for
+		 *    delalloc, because in case we can't immediately reserve space
+		 *    the flusher can start delalloc and wait for the respective
+		 *    ordered extents to complete. The deadlock would happen
+		 *    because we do the space reservation while holding the range
+		 *    locked, and starting writeback, or finishing an ordered
+		 *    extent, requires locking the range;
+		 *
+		 * 2) If there's delalloc there, it means there's dirty pages for
+		 *    which writeback has not started yet (we clean the delalloc
+		 *    flag when starting writeback and after creating an ordered
+		 *    extent). If we mark pages in an adjacent range for defrag,
+		 *    then we will have a larger contiguous range for delalloc,
+		 *    very likely resulting in a larger extent after writeback is
+		 *    triggered (except in a case of free space fragmentation).
+		 */
+		if (test_range_bit(&inode->io_tree, cur, cur + range_len - 1,
+				   EXTENT_DELALLOC, 0, NULL))
+			goto next;
+
+		/*
+		 * For do_compress case, we want to compress all valid file
+		 * extents, thus no @extent_thresh or mergeable check.
+		 */
+		if (do_compress)
+			goto add;
+
+		/* Skip too large extent */
+		if (range_len >= extent_thresh)
+			goto next;
+
+		/*
+		 * Skip extents already at its max capacity, this is mostly for
+		 * compressed extents, which max cap is only 128K.
+		 */
+		if (em->len >= get_extent_max_capacity(fs_info, em))
+			goto next;
+
+		/*
+		 * Normally there are no more extents after an inline one, thus
+		 * @next_mergeable will normally be false and not defragged.
+		 * So if an inline extent passed all above checks, just add it
+		 * for defrag, and be converted to regular extents.
+		 */
+		if (em->block_start == EXTENT_MAP_INLINE)
+			goto add;
+
+		next_mergeable = defrag_check_next_extent(&inode->vfs_inode, em,
+						extent_thresh, newer_than, locked);
+		if (!next_mergeable) {
+			struct defrag_target_range *last;
+
+			/* Empty target list, no way to merge with last entry */
+			if (list_empty(target_list))
+				goto next;
+			last = list_entry(target_list->prev,
+					  struct defrag_target_range, list);
+			/* Not mergeable with last entry */
+			if (last->start + last->len != cur)
+				goto next;
+
+			/* Mergeable, fall through to add it to @target_list. */
 		}
 
-		if (!PageUptodate(page)) {
-			btrfs_readpage(NULL, page);
-			lock_page(page);
-			if (!PageUptodate(page)) {
-				unlock_page(page);
-				put_page(page);
-				ret = -EIO;
-				break;
+add:
+		last_is_target = true;
+		range_len = min(extent_map_end(em), start + len) - cur;
+		/*
+		 * This one is a good target, check if it can be merged into
+		 * last range of the target list.
+		 */
+		if (!list_empty(target_list)) {
+			struct defrag_target_range *last;
+
+			last = list_entry(target_list->prev,
+					  struct defrag_target_range, list);
+			ASSERT(last->start + last->len <= cur);
+			if (last->start + last->len == cur) {
+				/* Mergeable, enlarge the last entry */
+				last->len += range_len;
+				goto next;
 			}
+			/* Fall through to allocate a new entry */
 		}
 
-		if (page->mapping != inode->i_mapping) {
-			unlock_page(page);
-			put_page(page);
-			goto again;
+		/* Allocate new defrag_target_range */
+		new = kmalloc(sizeof(*new), GFP_NOFS);
+		if (!new) {
+			free_extent_map(em);
+			ret = -ENOMEM;
+			break;
 		}
+		new->start = cur;
+		new->len = range_len;
+		list_add_tail(&new->list, target_list);
 
-		pages[i] = page;
-		i_done++;
+next:
+		cur = extent_map_end(em);
+		free_extent_map(em);
 	}
-	if (!i_done || ret)
-		goto out;
+	if (ret < 0) {
+		struct defrag_target_range *entry;
+		struct defrag_target_range *tmp;
 
-	if (!(inode->i_sb->s_flags & SB_ACTIVE))
-		goto out;
+		list_for_each_entry_safe(entry, tmp, target_list, list) {
+			list_del_init(&entry->list);
+			kfree(entry);
+		}
+	}
+	if (!ret && last_scanned_ret) {
+		/*
+		 * If the last extent is not a target, the caller can skip to
+		 * the end of that extent.
+		 * Otherwise, we can only go the end of the specified range.
+		 */
+		if (!last_is_target)
+			*last_scanned_ret = max(cur, *last_scanned_ret);
+		else
+			*last_scanned_ret = max(start + len, *last_scanned_ret);
+	}
+	return ret;
+}
 
-	/*
-	 * so now we have a nice long stream of locked
-	 * and up to date pages, lets wait on them
-	 */
-	for (i = 0; i < i_done; i++)
+#define CLUSTER_SIZE	(SZ_256K)
+static_assert(IS_ALIGNED(CLUSTER_SIZE, PAGE_SIZE));
+
+/*
+ * Defrag one contiguous target range.
+ *
+ * @inode:	target inode
+ * @target:	target range to defrag
+ * @pages:	locked pages covering the defrag range
+ * @nr_pages:	number of locked pages
+ *
+ * Caller should ensure:
+ *
+ * - Pages are prepared
+ *   Pages should be locked, no ordered extent in the pages range,
+ *   no writeback.
+ *
+ * - Extent bits are locked
+ */
+static int defrag_one_locked_target(struct btrfs_inode *inode,
+				    struct defrag_target_range *target,
+				    struct page **pages, int nr_pages,
+				    struct extent_state **cached_state)
+{
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
+	struct extent_changeset *data_reserved = NULL;
+	const u64 start = target->start;
+	const u64 len = target->len;
+	unsigned long last_index = (start + len - 1) >> PAGE_SHIFT;
+	unsigned long start_index = start >> PAGE_SHIFT;
+	unsigned long first_index = page_index(pages[0]);
+	int ret = 0;
+	int i;
+
+	ASSERT(last_index - first_index + 1 <= nr_pages);
+
+	ret = btrfs_delalloc_reserve_space(inode, &data_reserved, start, len);
+	if (ret < 0)
+		return ret;
+	clear_extent_bit(&inode->io_tree, start, start + len - 1,
+			 EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING |
+			 EXTENT_DEFRAG, cached_state);
+	set_extent_defrag(&inode->io_tree, start, start + len - 1, cached_state);
+
+	/* Update the page status */
+	for (i = start_index - first_index; i <= last_index - first_index; i++) {
+		ClearPageChecked(pages[i]);
+		btrfs_page_clamp_set_dirty(fs_info, pages[i], start, len);
+	}
+	btrfs_delalloc_release_extents(inode, len);
+	extent_changeset_free(data_reserved);
+
+	return ret;
+}
+
+static int defrag_one_range(struct btrfs_inode *inode, u64 start, u32 len,
+			    u32 extent_thresh, u64 newer_than, bool do_compress,
+			    u64 *last_scanned_ret)
+{
+	struct extent_state *cached_state = NULL;
+	struct defrag_target_range *entry;
+	struct defrag_target_range *tmp;
+	LIST_HEAD(target_list);
+	struct page **pages;
+	const u32 sectorsize = inode->root->fs_info->sectorsize;
+	u64 last_index = (start + len - 1) >> PAGE_SHIFT;
+	u64 start_index = start >> PAGE_SHIFT;
+	unsigned int nr_pages = last_index - start_index + 1;
+	int ret = 0;
+	int i;
+
+	ASSERT(nr_pages <= CLUSTER_SIZE / PAGE_SIZE);
+	ASSERT(IS_ALIGNED(start, sectorsize) && IS_ALIGNED(len, sectorsize));
+
+	pages = kcalloc(nr_pages, sizeof(struct page *), GFP_NOFS);
+	if (!pages)
+		return -ENOMEM;
+
+	/* Prepare all pages */
+	for (i = 0; i < nr_pages; i++) {
+		pages[i] = defrag_prepare_one_page(inode, start_index + i);
+		if (IS_ERR(pages[i])) {
+			ret = PTR_ERR(pages[i]);
+			pages[i] = NULL;
+			goto free_pages;
+		}
+	}
+	for (i = 0; i < nr_pages; i++)
 		wait_on_page_writeback(pages[i]);
 
-	page_start = page_offset(pages[0]);
-	page_end = page_offset(pages[i_done - 1]) + PAGE_SIZE;
+	/* Lock the pages range */
+	lock_extent(&inode->io_tree, start_index << PAGE_SHIFT,
+		    (last_index << PAGE_SHIFT) + PAGE_SIZE - 1,
+		    &cached_state);
+	/*
+	 * Now we have a consistent view about the extent map, re-check
+	 * which range really needs to be defragged.
+	 *
+	 * And this time we have extent locked already, pass @locked = true
+	 * so that we won't relock the extent range and cause deadlock.
+	 */
+	ret = defrag_collect_targets(inode, start, len, extent_thresh,
+				     newer_than, do_compress, true,
+				     &target_list, last_scanned_ret);
+	if (ret < 0)
+		goto unlock_extent;
 
-	lock_extent_bits(&BTRFS_I(inode)->io_tree,
-			 page_start, page_end - 1, &cached_state);
-	clear_extent_bit(&BTRFS_I(inode)->io_tree, page_start,
-			  page_end - 1, EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING |
-			  EXTENT_DEFRAG, 0, 0, &cached_state);
+	list_for_each_entry(entry, &target_list, list) {
+		ret = defrag_one_locked_target(inode, entry, pages, nr_pages,
+					       &cached_state);
+		if (ret < 0)
+			break;
+	}
 
-	if (i_done != page_cnt) {
-		spin_lock(&BTRFS_I(inode)->lock);
-		btrfs_mod_outstanding_extents(BTRFS_I(inode), 1);
-		spin_unlock(&BTRFS_I(inode)->lock);
-		btrfs_delalloc_release_space(inode, data_reserved,
-				start_index << PAGE_SHIFT,
-				(page_cnt - i_done) << PAGE_SHIFT, true);
+	list_for_each_entry_safe(entry, tmp, &target_list, list) {
+		list_del_init(&entry->list);
+		kfree(entry);
 	}
+unlock_extent:
+	unlock_extent(&inode->io_tree, start_index << PAGE_SHIFT,
+		      (last_index << PAGE_SHIFT) + PAGE_SIZE - 1,
+		      &cached_state);
+free_pages:
+	for (i = 0; i < nr_pages; i++) {
+		if (pages[i]) {
+			unlock_page(pages[i]);
+			put_page(pages[i]);
+		}
+	}
+	kfree(pages);
+	return ret;
+}
 
+static int defrag_one_cluster(struct btrfs_inode *inode,
+			      struct file_ra_state *ra,
+			      u64 start, u32 len, u32 extent_thresh,
+			      u64 newer_than, bool do_compress,
+			      unsigned long *sectors_defragged,
+			      unsigned long max_sectors,
+			      u64 *last_scanned_ret)
+{
+	const u32 sectorsize = inode->root->fs_info->sectorsize;
+	struct defrag_target_range *entry;
+	struct defrag_target_range *tmp;
+	LIST_HEAD(target_list);
+	int ret;
 
-	set_extent_defrag(&BTRFS_I(inode)->io_tree, page_start, page_end - 1,
-			  &cached_state);
+	ret = defrag_collect_targets(inode, start, len, extent_thresh,
+				     newer_than, do_compress, false,
+				     &target_list, NULL);
+	if (ret < 0)
+		goto out;
 
-	unlock_extent_cached(&BTRFS_I(inode)->io_tree,
-			     page_start, page_end - 1, &cached_state);
+	list_for_each_entry(entry, &target_list, list) {
+		u32 range_len = entry->len;
 
-	for (i = 0; i < i_done; i++) {
-		clear_page_dirty_for_io(pages[i]);
-		ClearPageChecked(pages[i]);
-		set_page_extent_mapped(pages[i]);
-		set_page_dirty(pages[i]);
-		unlock_page(pages[i]);
-		put_page(pages[i]);
+		/* Reached or beyond the limit */
+		if (max_sectors && *sectors_defragged >= max_sectors) {
+			ret = 1;
+			break;
+		}
+
+		if (max_sectors)
+			range_len = min_t(u32, range_len,
+				(max_sectors - *sectors_defragged) * sectorsize);
+
+		/*
+		 * If defrag_one_range() has updated last_scanned_ret,
+		 * our range may already be invalid (e.g. hole punched).
+		 * Skip if our range is before last_scanned_ret, as there is
+		 * no need to defrag the range anymore.
+		 */
+		if (entry->start + range_len <= *last_scanned_ret)
+			continue;
+
+		if (ra)
+			page_cache_sync_readahead(inode->vfs_inode.i_mapping,
+				ra, NULL, entry->start >> PAGE_SHIFT,
+				((entry->start + range_len - 1) >> PAGE_SHIFT) -
+				(entry->start >> PAGE_SHIFT) + 1);
+		/*
+		 * Here we may not defrag any range if holes are punched before
+		 * we locked the pages.
+		 * But that's fine, it only affects the @sectors_defragged
+		 * accounting.
+		 */
+		ret = defrag_one_range(inode, entry->start, range_len,
+				       extent_thresh, newer_than, do_compress,
+				       last_scanned_ret);
+		if (ret < 0)
+			break;
+		*sectors_defragged += range_len >>
+				      inode->root->fs_info->sectorsize_bits;
 	}
-	btrfs_delalloc_release_extents(BTRFS_I(inode), page_cnt << PAGE_SHIFT);
-	extent_changeset_free(data_reserved);
-	return i_done;
 out:
-	for (i = 0; i < i_done; i++) {
-		unlock_page(pages[i]);
-		put_page(pages[i]);
-	}
-	btrfs_delalloc_release_space(inode, data_reserved,
-			start_index << PAGE_SHIFT,
-			page_cnt << PAGE_SHIFT, true);
-	btrfs_delalloc_release_extents(BTRFS_I(inode), page_cnt << PAGE_SHIFT);
-	extent_changeset_free(data_reserved);
+	list_for_each_entry_safe(entry, tmp, &target_list, list) {
+		list_del_init(&entry->list);
+		kfree(entry);
+	}
+	if (ret >= 0)
+		*last_scanned_ret = max(*last_scanned_ret, start + len);
 	return ret;
-
 }
 
-int btrfs_defrag_file(struct inode *inode, struct file *file,
+/*
+ * Entry point to file defragmentation.
+ *
+ * @inode:	   inode to be defragged
+ * @ra:		   readahead state (can be NUL)
+ * @range:	   defrag options including range and flags
+ * @newer_than:	   minimum transid to defrag
+ * @max_to_defrag: max number of sectors to be defragged, if 0, the whole inode
+ *		   will be defragged.
+ *
+ * Return <0 for error.
+ * Return >=0 for the number of sectors defragged, and range->start will be updated
+ * to indicate the file offset where next defrag should be started at.
+ * (Mostly for autodefrag, which sets @max_to_defrag thus we may exit early without
+ *  defragging all the range).
+ */
+int btrfs_defrag_file(struct inode *inode, struct file_ra_state *ra,
 		      struct btrfs_ioctl_defrag_range_args *range,
 		      u64 newer_than, unsigned long max_to_defrag)
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	struct btrfs_root *root = BTRFS_I(inode)->root;
-	struct file_ra_state *ra = NULL;
-	unsigned long last_index;
+	unsigned long sectors_defragged = 0;
 	u64 isize = i_size_read(inode);
-	u64 last_len = 0;
-	u64 skip = 0;
-	u64 defrag_end = 0;
-	u64 newer_off = range->start;
-	unsigned long i;
-	unsigned long ra_index = 0;
-	int ret;
-	int defrag_count = 0;
+	u64 cur;
+	u64 last_byte;
+	bool do_compress = range->flags & BTRFS_DEFRAG_RANGE_COMPRESS;
+	bool ra_allocated = false;
 	int compress_type = BTRFS_COMPRESS_ZLIB;
+	int ret = 0;
 	u32 extent_thresh = range->extent_thresh;
-	unsigned long max_cluster = SZ_256K >> PAGE_SHIFT;
-	unsigned long cluster = max_cluster;
-	u64 new_align = ~((u64)SZ_128K - 1);
-	struct page **pages = NULL;
-	bool do_compress = range->flags & BTRFS_DEFRAG_RANGE_COMPRESS;
+	pgoff_t start_index;
 
 	if (isize == 0)
 		return 0;
@@ -1423,178 +1827,162 @@ int btrfs_defrag_file(struct inode *inode, struct file *file,
 	if (extent_thresh == 0)
 		extent_thresh = SZ_256K;
 
+	if (range->start + range->len > range->start) {
+		/* Got a specific range */
+		last_byte = min(isize, range->start + range->len);
+	} else {
+		/* Defrag until file end */
+		last_byte = isize;
+	}
+
+	/* Align the range */
+	cur = round_down(range->start, fs_info->sectorsize);
+	last_byte = round_up(last_byte, fs_info->sectorsize) - 1;
+
 	/*
-	 * If we were not given a file, allocate a readahead context. As
+	 * If we were not given a ra, allocate a readahead context. As
 	 * readahead is just an optimization, defrag will work without it so
 	 * we don't error out.
 	 */
-	if (!file) {
+	if (!ra) {
+		ra_allocated = true;
 		ra = kzalloc(sizeof(*ra), GFP_KERNEL);
 		if (ra)
 			file_ra_state_init(ra, inode->i_mapping);
-	} else {
-		ra = &file->f_ra;
-	}
-
-	pages = kmalloc_array(max_cluster, sizeof(struct page *), GFP_KERNEL);
-	if (!pages) {
-		ret = -ENOMEM;
-		goto out_ra;
-	}
-
-	/* find the last page to defrag */
-	if (range->start + range->len > range->start) {
-		last_index = min_t(u64, isize - 1,
-			 range->start + range->len - 1) >> PAGE_SHIFT;
-	} else {
-		last_index = (isize - 1) >> PAGE_SHIFT;
 	}
 
-	if (newer_than) {
-		ret = find_new_extents(root, inode, newer_than,
-				       &newer_off, SZ_64K);
-		if (!ret) {
-			range->start = newer_off;
-			/*
-			 * we always align our defrag to help keep
-			 * the extents in the file evenly spaced
-			 */
-			i = (newer_off & new_align) >> PAGE_SHIFT;
-		} else
-			goto out_ra;
-	} else {
-		i = range->start >> PAGE_SHIFT;
-	}
-	if (!max_to_defrag)
-		max_to_defrag = last_index - i + 1;
-
 	/*
-	 * make writeback starts from i, so the defrag range can be
-	 * written sequentially.
+	 * Make writeback start from the beginning of the range, so that the
+	 * defrag range can be written sequentially.
 	 */
-	if (i < inode->i_mapping->writeback_index)
-		inode->i_mapping->writeback_index = i;
+	start_index = cur >> PAGE_SHIFT;
+	if (start_index < inode->i_mapping->writeback_index)
+		inode->i_mapping->writeback_index = start_index;
 
-	while (i <= last_index && defrag_count < max_to_defrag &&
-	       (i < DIV_ROUND_UP(i_size_read(inode), PAGE_SIZE))) {
-		/*
-		 * make sure we stop running if someone unmounts
-		 * the FS
-		 */
-		if (!(inode->i_sb->s_flags & SB_ACTIVE))
-			break;
+	while (cur < last_byte) {
+		const unsigned long prev_sectors_defragged = sectors_defragged;
+		u64 last_scanned = cur;
+		u64 cluster_end;
 
 		if (btrfs_defrag_cancelled(fs_info)) {
-			btrfs_debug(fs_info, "defrag_file cancelled");
 			ret = -EAGAIN;
 			break;
 		}
 
-		if (!should_defrag_range(inode, (u64)i << PAGE_SHIFT,
-					 extent_thresh, &last_len, &skip,
-					 &defrag_end, do_compress)){
-			unsigned long next;
-			/*
-			 * the should_defrag function tells us how much to skip
-			 * bump our counter by the suggested amount
-			 */
-			next = DIV_ROUND_UP(skip, PAGE_SIZE);
-			i = max(i + 1, next);
-			continue;
-		}
-
-		if (!newer_than) {
-			cluster = (PAGE_ALIGN(defrag_end) >>
-				   PAGE_SHIFT) - i;
-			cluster = min(cluster, max_cluster);
-		} else {
-			cluster = max_cluster;
-		}
-
-		if (i + cluster > ra_index) {
-			ra_index = max(i, ra_index);
-			if (ra)
-				page_cache_sync_readahead(inode->i_mapping, ra,
-						file, ra_index, cluster);
-			ra_index += cluster;
-		}
+		/* We want the cluster end at page boundary when possible */
+		cluster_end = (((cur >> PAGE_SHIFT) +
+			       (SZ_256K >> PAGE_SHIFT)) << PAGE_SHIFT) - 1;
+		cluster_end = min(cluster_end, last_byte);
 
-		inode_lock(inode);
+		btrfs_inode_lock(inode, 0);
 		if (IS_SWAPFILE(inode)) {
 			ret = -ETXTBSY;
-		} else {
-			if (do_compress)
-				BTRFS_I(inode)->defrag_compress = compress_type;
-			ret = cluster_pages_for_defrag(inode, pages, i, cluster);
+			btrfs_inode_unlock(inode, 0);
+			break;
 		}
-		if (ret < 0) {
-			inode_unlock(inode);
-			goto out_ra;
+		if (!(inode->i_sb->s_flags & SB_ACTIVE)) {
+			btrfs_inode_unlock(inode, 0);
+			break;
 		}
+		if (do_compress)
+			BTRFS_I(inode)->defrag_compress = compress_type;
+		ret = defrag_one_cluster(BTRFS_I(inode), ra, cur,
+				cluster_end + 1 - cur, extent_thresh,
+				newer_than, do_compress, &sectors_defragged,
+				max_to_defrag, &last_scanned);
 
-		defrag_count += ret;
-		balance_dirty_pages_ratelimited(inode->i_mapping);
-		inode_unlock(inode);
-
-		if (newer_than) {
-			if (newer_off == (u64)-1)
-				break;
+		if (sectors_defragged > prev_sectors_defragged)
+			balance_dirty_pages_ratelimited(inode->i_mapping);
 
-			if (ret > 0)
-				i += ret;
-
-			newer_off = max(newer_off + 1,
-					(u64)i << PAGE_SHIFT);
-
-			ret = find_new_extents(root, inode, newer_than,
-					       &newer_off, SZ_64K);
-			if (!ret) {
-				range->start = newer_off;
-				i = (newer_off & new_align) >> PAGE_SHIFT;
-			} else {
-				break;
-			}
-		} else {
-			if (ret > 0) {
-				i += ret;
-				last_len += ret << PAGE_SHIFT;
-			} else {
-				i++;
-				last_len = 0;
-			}
+		btrfs_inode_unlock(inode, 0);
+		if (ret < 0)
+			break;
+		cur = max(cluster_end + 1, last_scanned);
+		if (ret > 0) {
+			ret = 0;
+			break;
 		}
+		cond_resched();
 	}
 
-	if ((range->flags & BTRFS_DEFRAG_RANGE_START_IO)) {
-		filemap_flush(inode->i_mapping);
-		if (test_bit(BTRFS_INODE_HAS_ASYNC_EXTENT,
-			     &BTRFS_I(inode)->runtime_flags))
+	if (ra_allocated)
+		kfree(ra);
+	/*
+	 * Update range.start for autodefrag, this will indicate where to start
+	 * in next run.
+	 */
+	range->start = cur;
+	if (sectors_defragged) {
+		/*
+		 * We have defragged some sectors, for compression case they
+		 * need to be written back immediately.
+		 */
+		if (range->flags & BTRFS_DEFRAG_RANGE_START_IO) {
 			filemap_flush(inode->i_mapping);
+			if (test_bit(BTRFS_INODE_HAS_ASYNC_EXTENT,
+				     &BTRFS_I(inode)->runtime_flags))
+				filemap_flush(inode->i_mapping);
+		}
+		if (range->compress_type == BTRFS_COMPRESS_LZO)
+			btrfs_set_fs_incompat(fs_info, COMPRESS_LZO);
+		else if (range->compress_type == BTRFS_COMPRESS_ZSTD)
+			btrfs_set_fs_incompat(fs_info, COMPRESS_ZSTD);
+		ret = sectors_defragged;
 	}
+	if (do_compress) {
+		btrfs_inode_lock(inode, 0);
+		BTRFS_I(inode)->defrag_compress = BTRFS_COMPRESS_NONE;
+		btrfs_inode_unlock(inode, 0);
+	}
+	return ret;
+}
 
-	if (range->compress_type == BTRFS_COMPRESS_LZO) {
-		btrfs_set_fs_incompat(fs_info, COMPRESS_LZO);
-	} else if (range->compress_type == BTRFS_COMPRESS_ZSTD) {
-		btrfs_set_fs_incompat(fs_info, COMPRESS_ZSTD);
+/*
+ * Try to start exclusive operation @type or cancel it if it's running.
+ *
+ * Return:
+ *   0        - normal mode, newly claimed op started
+ *  >0        - normal mode, something else is running,
+ *              return BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS to user space
+ * ECANCELED  - cancel mode, successful cancel
+ * ENOTCONN   - cancel mode, operation not running anymore
+ */
+static int exclop_start_or_cancel_reloc(struct btrfs_fs_info *fs_info,
+			enum btrfs_exclusive_operation type, bool cancel)
+{
+	if (!cancel) {
+		/* Start normal op */
+		if (!btrfs_exclop_start(fs_info, type))
+			return BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
+		/* Exclusive operation is now claimed */
+		return 0;
 	}
 
-	ret = defrag_count;
+	/* Cancel running op */
+	if (btrfs_exclop_start_try_lock(fs_info, type)) {
+		/*
+		 * This blocks any exclop finish from setting it to NONE, so we
+		 * request cancellation. Either it runs and we will wait for it,
+		 * or it has finished and no waiting will happen.
+		 */
+		atomic_inc(&fs_info->reloc_cancel_req);
+		btrfs_exclop_start_unlock(fs_info);
+
+		if (test_bit(BTRFS_FS_RELOC_RUNNING, &fs_info->flags))
+			wait_on_bit(&fs_info->flags, BTRFS_FS_RELOC_RUNNING,
+				    TASK_INTERRUPTIBLE);
 
-out_ra:
-	if (do_compress) {
-		inode_lock(inode);
-		BTRFS_I(inode)->defrag_compress = BTRFS_COMPRESS_NONE;
-		inode_unlock(inode);
+		return -ECANCELED;
 	}
-	if (!file)
-		kfree(ra);
-	kfree(pages);
-	return ret;
+
+	/* Something else is running or none */
+	return -ENOTCONN;
 }
 
 static noinline int btrfs_ioctl_resize(struct file *file,
 					void __user *arg)
 {
+	BTRFS_DEV_LOOKUP_ARGS(args);
 	struct inode *inode = file_inode(file);
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	u64 new_size;
@@ -1609,6 +1997,7 @@ static noinline int btrfs_ioctl_resize(struct file *file,
 	char *devstr = NULL;
 	int ret = 0;
 	int mod = 0;
+	bool cancel;
 
 	if (!capable(CAP_SYS_ADMIN))
 		return -EPERM;
@@ -1617,20 +2006,23 @@ static noinline int btrfs_ioctl_resize(struct file *file,
 	if (ret)
 		return ret;
 
-	if (test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags)) {
-		mnt_drop_write_file(file);
-		return BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
-	}
-
+	/*
+	 * Read the arguments before checking exclusivity to be able to
+	 * distinguish regular resize and cancel
+	 */
 	vol_args = memdup_user(arg, sizeof(*vol_args));
 	if (IS_ERR(vol_args)) {
 		ret = PTR_ERR(vol_args);
-		goto out;
+		goto out_drop;
 	}
-
 	vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
-
 	sizestr = vol_args->name;
+	cancel = (strcmp("cancel", sizestr) == 0);
+	ret = exclop_start_or_cancel_reloc(fs_info, BTRFS_EXCLOP_RESIZE, cancel);
+	if (ret)
+		goto out_free;
+	/* Exclusive operation is now claimed */
+
 	devstr = strchr(sizestr, ':');
 	if (devstr) {
 		sizestr = devstr + 1;
@@ -1638,20 +2030,21 @@ static noinline int btrfs_ioctl_resize(struct file *file,
 		devstr = vol_args->name;
 		ret = kstrtoull(devstr, 10, &devid);
 		if (ret)
-			goto out_free;
+			goto out_finish;
 		if (!devid) {
 			ret = -EINVAL;
-			goto out_free;
+			goto out_finish;
 		}
 		btrfs_info(fs_info, "resizing devid %llu", devid);
 	}
 
-	device = btrfs_find_device(fs_info->fs_devices, devid, NULL, NULL, true);
+	args.devid = devid;
+	device = btrfs_find_device(fs_info->fs_devices, &args);
 	if (!device) {
 		btrfs_info(fs_info, "resizer unable to find device %llu",
 			   devid);
 		ret = -ENODEV;
-		goto out_free;
+		goto out_finish;
 	}
 
 	if (!test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state)) {
@@ -1659,11 +2052,11 @@ static noinline int btrfs_ioctl_resize(struct file *file,
 			   "resizer unable to apply on readonly device %llu",
 		       devid);
 		ret = -EPERM;
-		goto out_free;
+		goto out_finish;
 	}
 
 	if (!strcmp(sizestr, "max"))
-		new_size = device->bdev->bd_inode->i_size;
+		new_size = bdev_nr_bytes(device->bdev);
 	else {
 		if (sizestr[0] == '-') {
 			mod = -1;
@@ -1675,13 +2068,13 @@ static noinline int btrfs_ioctl_resize(struct file *file,
 		new_size = memparse(sizestr, &retptr);
 		if (*retptr != '\0' || new_size == 0) {
 			ret = -EINVAL;
-			goto out_free;
+			goto out_finish;
 		}
 	}
 
 	if (test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state)) {
 		ret = -EPERM;
-		goto out_free;
+		goto out_finish;
 	}
 
 	old_size = btrfs_device_get_total_bytes(device);
@@ -1689,36 +2082,33 @@ static noinline int btrfs_ioctl_resize(struct file *file,
 	if (mod < 0) {
 		if (new_size > old_size) {
 			ret = -EINVAL;
-			goto out_free;
+			goto out_finish;
 		}
 		new_size = old_size - new_size;
 	} else if (mod > 0) {
 		if (new_size > ULLONG_MAX - old_size) {
 			ret = -ERANGE;
-			goto out_free;
+			goto out_finish;
 		}
 		new_size = old_size + new_size;
 	}
 
 	if (new_size < SZ_256M) {
 		ret = -EINVAL;
-		goto out_free;
+		goto out_finish;
 	}
-	if (new_size > device->bdev->bd_inode->i_size) {
+	if (new_size > bdev_nr_bytes(device->bdev)) {
 		ret = -EFBIG;
-		goto out_free;
+		goto out_finish;
 	}
 
 	new_size = round_down(new_size, fs_info->sectorsize);
 
-	btrfs_info_in_rcu(fs_info, "new size for %s is %llu",
-			  rcu_str_deref(device->name), new_size);
-
 	if (new_size > old_size) {
 		trans = btrfs_start_transaction(root, 0);
 		if (IS_ERR(trans)) {
 			ret = PTR_ERR(trans);
-			goto out_free;
+			goto out_finish;
 		}
 		ret = btrfs_grow_device(trans, device, new_size);
 		btrfs_commit_transaction(trans);
@@ -1726,17 +2116,24 @@ static noinline int btrfs_ioctl_resize(struct file *file,
 		ret = btrfs_shrink_device(device, new_size);
 	} /* equal, nothing need to do */
 
+	if (ret == 0 && new_size != old_size)
+		btrfs_info_in_rcu(fs_info,
+			"resize device %s (devid %llu) from %llu to %llu",
+			rcu_str_deref(device->name), device->devid,
+			old_size, new_size);
+out_finish:
+	btrfs_exclop_finish(fs_info);
 out_free:
 	kfree(vol_args);
-out:
-	clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
+out_drop:
 	mnt_drop_write_file(file);
 	return ret;
 }
 
-static noinline int btrfs_ioctl_snap_create_transid(struct file *file,
+static noinline int __btrfs_ioctl_snap_create(struct file *file,
+				struct user_namespace *mnt_userns,
 				const char *name, unsigned long fd, int subvol,
-				u64 *transid, bool readonly,
+				bool readonly,
 				struct btrfs_qgroup_inherit *inherit)
 {
 	int namelen;
@@ -1762,8 +2159,8 @@ static noinline int btrfs_ioctl_snap_create_transid(struct file *file,
 	}
 
 	if (subvol) {
-		ret = btrfs_mksubvol(&file->f_path, name, namelen,
-				     NULL, transid, readonly, inherit);
+		ret = btrfs_mksubvol(&file->f_path, mnt_userns, name,
+				     namelen, NULL, readonly, inherit);
 	} else {
 		struct fd src = fdget(fd);
 		struct inode *src_inode;
@@ -1777,16 +2174,17 @@ static noinline int btrfs_ioctl_snap_create_transid(struct file *file,
 			btrfs_info(BTRFS_I(file_inode(file))->root->fs_info,
 				   "Snapshot src from another FS");
 			ret = -EXDEV;
-		} else if (!inode_owner_or_capable(src_inode)) {
+		} else if (!inode_owner_or_capable(mnt_userns, src_inode)) {
 			/*
 			 * Subvolume creation is not restricted, but snapshots
 			 * are limited to own subvolumes only
 			 */
 			ret = -EPERM;
 		} else {
-			ret = btrfs_mksubvol(&file->f_path, name, namelen,
-					     BTRFS_I(src_inode)->root,
-					     transid, readonly, inherit);
+			ret = btrfs_mksnapshot(&file->f_path, mnt_userns,
+					       name, namelen,
+					       BTRFS_I(src_inode)->root,
+					       readonly, inherit);
 		}
 		fdput(src);
 	}
@@ -1810,9 +2208,9 @@ static noinline int btrfs_ioctl_snap_create(struct file *file,
 		return PTR_ERR(vol_args);
 	vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
 
-	ret = btrfs_ioctl_snap_create_transid(file, vol_args->name,
-					      vol_args->fd, subvol,
-					      NULL, false, NULL);
+	ret = __btrfs_ioctl_snap_create(file, file_mnt_user_ns(file),
+					vol_args->name, vol_args->fd, subvol,
+					false, NULL);
 
 	kfree(vol_args);
 	return ret;
@@ -1823,8 +2221,6 @@ static noinline int btrfs_ioctl_snap_create_v2(struct file *file,
 {
 	struct btrfs_ioctl_vol_args_v2 *vol_args;
 	int ret;
-	u64 transid = 0;
-	u64 *ptr = NULL;
 	bool readonly = false;
 	struct btrfs_qgroup_inherit *inherit = NULL;
 
@@ -1836,26 +2232,18 @@ static noinline int btrfs_ioctl_snap_create_v2(struct file *file,
 		return PTR_ERR(vol_args);
 	vol_args->name[BTRFS_SUBVOL_NAME_MAX] = '\0';
 
-	if (vol_args->flags &
-	    ~(BTRFS_SUBVOL_CREATE_ASYNC | BTRFS_SUBVOL_RDONLY |
-	      BTRFS_SUBVOL_QGROUP_INHERIT)) {
+	if (vol_args->flags & ~BTRFS_SUBVOL_CREATE_ARGS_MASK) {
 		ret = -EOPNOTSUPP;
 		goto free_args;
 	}
 
-	if (vol_args->flags & BTRFS_SUBVOL_CREATE_ASYNC) {
-		struct inode *inode = file_inode(file);
-		struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-
-		btrfs_warn(fs_info,
-"SNAP_CREATE_V2 ioctl with CREATE_ASYNC is deprecated and will be removed in kernel 5.7");
-
-		ptr = &transid;
-	}
 	if (vol_args->flags & BTRFS_SUBVOL_RDONLY)
 		readonly = true;
 	if (vol_args->flags & BTRFS_SUBVOL_QGROUP_INHERIT) {
-		if (vol_args->size > PAGE_SIZE) {
+		u64 nums;
+
+		if (vol_args->size < sizeof(*inherit) ||
+		    vol_args->size > PAGE_SIZE) {
 			ret = -EINVAL;
 			goto free_args;
 		}
@@ -1864,20 +2252,27 @@ static noinline int btrfs_ioctl_snap_create_v2(struct file *file,
 			ret = PTR_ERR(inherit);
 			goto free_args;
 		}
+
+		if (inherit->num_qgroups > PAGE_SIZE ||
+		    inherit->num_ref_copies > PAGE_SIZE ||
+		    inherit->num_excl_copies > PAGE_SIZE) {
+			ret = -EINVAL;
+			goto free_inherit;
+		}
+
+		nums = inherit->num_qgroups + 2 * inherit->num_ref_copies +
+		       2 * inherit->num_excl_copies;
+		if (vol_args->size != struct_size(inherit, qgroups, nums)) {
+			ret = -EINVAL;
+			goto free_inherit;
+		}
 	}
 
-	ret = btrfs_ioctl_snap_create_transid(file, vol_args->name,
-					      vol_args->fd, subvol, ptr,
-					      readonly, inherit);
+	ret = __btrfs_ioctl_snap_create(file, file_mnt_user_ns(file),
+					vol_args->name, vol_args->fd, subvol,
+					readonly, inherit);
 	if (ret)
 		goto free_inherit;
-
-	if (ptr && copy_to_user(arg +
-				offsetof(struct btrfs_ioctl_vol_args_v2,
-					transid),
-				ptr, sizeof(*ptr)))
-		ret = -EFAULT;
-
 free_inherit:
 	kfree(inherit);
 free_args:
@@ -1885,10 +2280,9 @@ free_args:
 	return ret;
 }
 
-static noinline int btrfs_ioctl_subvol_getflags(struct file *file,
+static noinline int btrfs_ioctl_subvol_getflags(struct inode *inode,
 						void __user *arg)
 {
-	struct inode *inode = file_inode(file);
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	struct btrfs_root *root = BTRFS_I(inode)->root;
 	int ret = 0;
@@ -1919,7 +2313,7 @@ static noinline int btrfs_ioctl_subvol_setflags(struct file *file,
 	u64 flags;
 	int ret = 0;
 
-	if (!inode_owner_or_capable(inode))
+	if (!inode_owner_or_capable(file_mnt_user_ns(file), inode))
 		return -EPERM;
 
 	ret = mnt_want_write_file(file);
@@ -1936,11 +2330,6 @@ static noinline int btrfs_ioctl_subvol_setflags(struct file *file,
 		goto out_drop_write;
 	}
 
-	if (flags & BTRFS_SUBVOL_CREATE_ASYNC) {
-		ret = -EINVAL;
-		goto out_drop_write;
-	}
-
 	if (flags & ~BTRFS_SUBVOL_RDONLY) {
 		ret = -EOPNOTSUPP;
 		goto out_drop_write;
@@ -2057,7 +2446,7 @@ static noinline int copy_to_sk(struct btrfs_path *path,
 
 	for (i = slot; i < nritems; i++) {
 		item_off = btrfs_item_ptr_offset(leaf, i);
-		item_len = btrfs_item_size_nr(leaf, i);
+		item_len = btrfs_item_size(leaf, i);
 
 		btrfs_item_key_to_cpu(leaf, key, i);
 		if (!key_in_sk(key, sk))
@@ -2090,9 +2479,14 @@ static noinline int copy_to_sk(struct btrfs_path *path,
 		sh.len = item_len;
 		sh.transid = found_transid;
 
-		/* copy search result header */
-		if (copy_to_user(ubuf + *sk_offset, &sh, sizeof(sh))) {
-			ret = -EFAULT;
+		/*
+		 * Copy search result header. If we fault then loop again so we
+		 * can fault in the pages and -EFAULT there if there's a
+		 * problem. Otherwise we'll fault and then copy the buffer in
+		 * properly this next time through
+		 */
+		if (copy_to_user_nofault(ubuf + *sk_offset, &sh, sizeof(sh))) {
+			ret = 0;
 			goto out;
 		}
 
@@ -2100,10 +2494,14 @@ static noinline int copy_to_sk(struct btrfs_path *path,
 
 		if (item_len) {
 			char __user *up = ubuf + *sk_offset;
-			/* copy the item */
-			if (read_extent_buffer_to_user(leaf, up,
-						       item_off, item_len)) {
-				ret = -EFAULT;
+			/*
+			 * Copy the item, same behavior as above, but reset the
+			 * * sk_offset so we copy the full thing again.
+			 */
+			if (read_extent_buffer_to_user_nofault(leaf, up,
+						item_off, item_len)) {
+				ret = 0;
+				*sk_offset -= sizeof(sh);
 				goto out;
 			}
 
@@ -2174,12 +2572,9 @@ static noinline int search_ioctl(struct inode *inode,
 
 	if (sk->tree_id == 0) {
 		/* search the root of the inode that was passed */
-		root = BTRFS_I(inode)->root;
+		root = btrfs_grab_root(BTRFS_I(inode)->root);
 	} else {
-		key.objectid = sk->tree_id;
-		key.type = BTRFS_ROOT_ITEM_KEY;
-		key.offset = (u64)-1;
-		root = btrfs_read_fs_root_no_name(info, &key);
+		root = btrfs_get_fs_root(info, sk->tree_id, true);
 		if (IS_ERR(root)) {
 			btrfs_free_path(path);
 			return PTR_ERR(root);
@@ -2191,6 +2586,15 @@ static noinline int search_ioctl(struct inode *inode,
 	key.offset = sk->min_offset;
 
 	while (1) {
+		ret = -EFAULT;
+		/*
+		 * Ensure that the whole user buffer is faulted in at sub-page
+		 * granularity, otherwise the loop may live-lock.
+		 */
+		if (fault_in_subpage_writeable(ubuf + sk_offset,
+					       *buf_size - sk_offset))
+			break;
+
 		ret = btrfs_search_forward(root, &key, path, sk->min_transid);
 		if (ret != 0) {
 			if (ret > 0)
@@ -2208,30 +2612,27 @@ static noinline int search_ioctl(struct inode *inode,
 		ret = 0;
 err:
 	sk->nr_items = num_found;
+	btrfs_put_root(root);
 	btrfs_free_path(path);
 	return ret;
 }
 
-static noinline int btrfs_ioctl_tree_search(struct file *file,
-					   void __user *argp)
+static noinline int btrfs_ioctl_tree_search(struct inode *inode,
+					    void __user *argp)
 {
-	struct btrfs_ioctl_search_args __user *uargs;
+	struct btrfs_ioctl_search_args __user *uargs = argp;
 	struct btrfs_ioctl_search_key sk;
-	struct inode *inode;
 	int ret;
 	size_t buf_size;
 
 	if (!capable(CAP_SYS_ADMIN))
 		return -EPERM;
 
-	uargs = (struct btrfs_ioctl_search_args __user *)argp;
-
 	if (copy_from_user(&sk, &uargs->key, sizeof(sk)))
 		return -EFAULT;
 
 	buf_size = sizeof(uargs->buf);
 
-	inode = file_inode(file);
 	ret = search_ioctl(inode, &sk, &buf_size, uargs->buf);
 
 	/*
@@ -2246,12 +2647,11 @@ static noinline int btrfs_ioctl_tree_search(struct file *file,
 	return ret;
 }
 
-static noinline int btrfs_ioctl_tree_search_v2(struct file *file,
+static noinline int btrfs_ioctl_tree_search_v2(struct inode *inode,
 					       void __user *argp)
 {
-	struct btrfs_ioctl_search_args_v2 __user *uarg;
+	struct btrfs_ioctl_search_args_v2 __user *uarg = argp;
 	struct btrfs_ioctl_search_args_v2 args;
-	struct inode *inode;
 	int ret;
 	size_t buf_size;
 	const size_t buf_limit = SZ_16M;
@@ -2260,7 +2660,6 @@ static noinline int btrfs_ioctl_tree_search_v2(struct file *file,
 		return -EPERM;
 
 	/* copy search header and buffer size */
-	uarg = (struct btrfs_ioctl_search_args_v2 __user *)argp;
 	if (copy_from_user(&args, uarg, sizeof(args)))
 		return -EFAULT;
 
@@ -2270,7 +2669,6 @@ static noinline int btrfs_ioctl_tree_search_v2(struct file *file,
 	if (buf_size > buf_limit)
 		buf_size = buf_limit;
 
-	inode = file_inode(file);
 	ret = search_ioctl(inode, &args.key, &buf_size,
 			   (char __user *)(&uarg->buf[0]));
 	if (ret == 0 && copy_to_user(&uarg->key, &args.key, sizeof(args.key)))
@@ -2311,12 +2709,10 @@ static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info,
 
 	ptr = &name[BTRFS_INO_LOOKUP_PATH_MAX - 1];
 
-	key.objectid = tree_id;
-	key.type = BTRFS_ROOT_ITEM_KEY;
-	key.offset = (u64)-1;
-	root = btrfs_read_fs_root_no_name(info, &key);
+	root = btrfs_get_fs_root(info, tree_id, true);
 	if (IS_ERR(root)) {
 		ret = PTR_ERR(root);
+		root = NULL;
 		goto out;
 	}
 
@@ -2325,23 +2721,16 @@ static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info,
 	key.offset = (u64)-1;
 
 	while (1) {
-		ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
+		ret = btrfs_search_backwards(root, &key, path);
 		if (ret < 0)
 			goto out;
 		else if (ret > 0) {
-			ret = btrfs_previous_item(root, path, dirid,
-						  BTRFS_INODE_REF_KEY);
-			if (ret < 0)
-				goto out;
-			else if (ret > 0) {
-				ret = -ENOENT;
-				goto out;
-			}
+			ret = -ENOENT;
+			goto out;
 		}
 
 		l = path->nodes[0];
 		slot = path->slots[0];
-		btrfs_item_key_to_cpu(l, &key, slot);
 
 		iref = btrfs_item_ptr(l, slot, struct btrfs_inode_ref);
 		len = btrfs_inode_ref_name_len(l, iref);
@@ -2367,11 +2756,13 @@ static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info,
 	name[total_len] = '\0';
 	ret = 0;
 out:
+	btrfs_put_root(root);
 	btrfs_free_path(path);
 	return ret;
 }
 
-static int btrfs_search_path_in_tree_user(struct inode *inode,
+static int btrfs_search_path_in_tree_user(struct user_namespace *mnt_userns,
+				struct inode *inode,
 				struct btrfs_ioctl_ino_lookup_user_args *args)
 {
 	struct btrfs_fs_info *fs_info = BTRFS_I(inode)->root->fs_info;
@@ -2383,7 +2774,7 @@ static int btrfs_search_path_in_tree_user(struct inode *inode,
 	unsigned long item_len;
 	struct btrfs_inode_ref *iref;
 	struct btrfs_root_ref *rref;
-	struct btrfs_root *root;
+	struct btrfs_root *root = NULL;
 	struct btrfs_path *path;
 	struct btrfs_key key, key2;
 	struct extent_buffer *leaf;
@@ -2405,10 +2796,7 @@ static int btrfs_search_path_in_tree_user(struct inode *inode,
 	if (dirid != upper_limit.objectid) {
 		ptr = &args->path[BTRFS_INO_LOOKUP_USER_PATH_MAX - 1];
 
-		key.objectid = treeid;
-		key.type = BTRFS_ROOT_ITEM_KEY;
-		key.offset = (u64)-1;
-		root = btrfs_read_fs_root_no_name(fs_info, &key);
+		root = btrfs_get_fs_root(fs_info, treeid, true);
 		if (IS_ERR(root)) {
 			ret = PTR_ERR(root);
 			goto out;
@@ -2418,23 +2806,16 @@ static int btrfs_search_path_in_tree_user(struct inode *inode,
 		key.type = BTRFS_INODE_REF_KEY;
 		key.offset = (u64)-1;
 		while (1) {
-			ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
-			if (ret < 0) {
-				goto out;
-			} else if (ret > 0) {
-				ret = btrfs_previous_item(root, path, dirid,
-							  BTRFS_INODE_REF_KEY);
-				if (ret < 0) {
-					goto out;
-				} else if (ret > 0) {
-					ret = -ENOENT;
-					goto out;
-				}
+			ret = btrfs_search_backwards(root, &key, path);
+			if (ret < 0)
+				goto out_put;
+			else if (ret > 0) {
+				ret = -ENOENT;
+				goto out_put;
 			}
 
 			leaf = path->nodes[0];
 			slot = path->slots[0];
-			btrfs_item_key_to_cpu(leaf, &key, slot);
 
 			iref = btrfs_item_ptr(leaf, slot, struct btrfs_inode_ref);
 			len = btrfs_inode_ref_name_len(leaf, iref);
@@ -2442,7 +2823,7 @@ static int btrfs_search_path_in_tree_user(struct inode *inode,
 			total_len += len + 1;
 			if (ptr < args->path) {
 				ret = -ENAMETOOLONG;
-				goto out;
+				goto out_put;
 			}
 
 			*(ptr + len) = '/';
@@ -2453,10 +2834,10 @@ static int btrfs_search_path_in_tree_user(struct inode *inode,
 			ret = btrfs_previous_item(root, path, dirid,
 						  BTRFS_INODE_ITEM_KEY);
 			if (ret < 0) {
-				goto out;
+				goto out_put;
 			} else if (ret > 0) {
 				ret = -ENOENT;
-				goto out;
+				goto out_put;
 			}
 
 			leaf = path->nodes[0];
@@ -2464,26 +2845,27 @@ static int btrfs_search_path_in_tree_user(struct inode *inode,
 			btrfs_item_key_to_cpu(leaf, &key2, slot);
 			if (key2.objectid != dirid) {
 				ret = -ENOENT;
-				goto out;
+				goto out_put;
 			}
 
-			temp_inode = btrfs_iget(sb, &key2, root);
+			temp_inode = btrfs_iget(sb, key2.objectid, root);
 			if (IS_ERR(temp_inode)) {
 				ret = PTR_ERR(temp_inode);
-				goto out;
+				goto out_put;
 			}
-			ret = inode_permission(temp_inode, MAY_READ | MAY_EXEC);
+			ret = inode_permission(mnt_userns, temp_inode,
+					       MAY_READ | MAY_EXEC);
 			iput(temp_inode);
 			if (ret) {
 				ret = -EACCES;
-				goto out;
+				goto out_put;
 			}
 
 			if (key.offset == upper_limit.objectid)
 				break;
 			if (key.objectid == BTRFS_FIRST_FREE_OBJECTID) {
 				ret = -EACCES;
-				goto out;
+				goto out_put;
 			}
 
 			btrfs_release_path(path);
@@ -2494,15 +2876,16 @@ static int btrfs_search_path_in_tree_user(struct inode *inode,
 
 		memmove(args->path, ptr, total_len);
 		args->path[total_len] = '\0';
+		btrfs_put_root(root);
+		root = NULL;
 		btrfs_release_path(path);
 	}
 
 	/* Get the bottom subvolume's name from ROOT_REF */
-	root = fs_info->tree_root;
 	key.objectid = treeid;
 	key.type = BTRFS_ROOT_REF_KEY;
 	key.offset = args->treeid;
-	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
+	ret = btrfs_search_slot(NULL, fs_info->tree_root, &key, path, 0, 0);
 	if (ret < 0) {
 		goto out;
 	} else if (ret > 0) {
@@ -2515,7 +2898,7 @@ static int btrfs_search_path_in_tree_user(struct inode *inode,
 	btrfs_item_key_to_cpu(leaf, &key, slot);
 
 	item_off = btrfs_item_ptr_offset(leaf, slot);
-	item_len = btrfs_item_size_nr(leaf, slot);
+	item_len = btrfs_item_size(leaf, slot);
 	/* Check if dirid in ROOT_REF corresponds to passed dirid */
 	rref = btrfs_item_ptr(leaf, slot, struct btrfs_root_ref);
 	if (args->dirid != btrfs_root_ref_dirid(leaf, rref)) {
@@ -2529,30 +2912,29 @@ static int btrfs_search_path_in_tree_user(struct inode *inode,
 	read_extent_buffer(leaf, args->name, item_off, item_len);
 	args->name[item_len] = 0;
 
+out_put:
+	btrfs_put_root(root);
 out:
 	btrfs_free_path(path);
 	return ret;
 }
 
-static noinline int btrfs_ioctl_ino_lookup(struct file *file,
+static noinline int btrfs_ioctl_ino_lookup(struct btrfs_root *root,
 					   void __user *argp)
 {
 	struct btrfs_ioctl_ino_lookup_args *args;
-	struct inode *inode;
 	int ret = 0;
 
 	args = memdup_user(argp, sizeof(*args));
 	if (IS_ERR(args))
 		return PTR_ERR(args);
 
-	inode = file_inode(file);
-
 	/*
 	 * Unprivileged query to obtain the containing subvolume root id. The
 	 * path is reset so it's consistent with btrfs_search_path_in_tree.
 	 */
 	if (args->treeid == 0)
-		args->treeid = BTRFS_I(inode)->root->root_key.objectid;
+		args->treeid = root->root_key.objectid;
 
 	if (args->objectid == BTRFS_FIRST_FREE_OBJECTID) {
 		args->name[0] = 0;
@@ -2564,7 +2946,7 @@ static noinline int btrfs_ioctl_ino_lookup(struct file *file,
 		goto out;
 	}
 
-	ret = btrfs_search_path_in_tree(BTRFS_I(inode)->root->fs_info,
+	ret = btrfs_search_path_in_tree(root->fs_info,
 					args->treeid, args->objectid,
 					args->name);
 
@@ -2610,7 +2992,7 @@ static int btrfs_ioctl_ino_lookup_user(struct file *file, void __user *argp)
 		return -EACCES;
 	}
 
-	ret = btrfs_search_path_in_tree_user(inode, args);
+	ret = btrfs_search_path_in_tree_user(file_mnt_user_ns(file), inode, args);
 
 	if (ret == 0 && copy_to_user(argp, args, sizeof(*args)))
 		ret = -EFAULT;
@@ -2620,7 +3002,7 @@ static int btrfs_ioctl_ino_lookup_user(struct file *file, void __user *argp)
 }
 
 /* Get the subvolume information in BTRFS_ROOT_ITEM and BTRFS_ROOT_BACKREF */
-static int btrfs_ioctl_get_subvol_info(struct file *file, void __user *argp)
+static int btrfs_ioctl_get_subvol_info(struct inode *inode, void __user *argp)
 {
 	struct btrfs_ioctl_get_subvol_info_args *subvol_info;
 	struct btrfs_fs_info *fs_info;
@@ -2632,7 +3014,6 @@ static int btrfs_ioctl_get_subvol_info(struct file *file, void __user *argp)
 	struct extent_buffer *leaf;
 	unsigned long item_off;
 	unsigned long item_len;
-	struct inode *inode;
 	int slot;
 	int ret = 0;
 
@@ -2646,17 +3027,14 @@ static int btrfs_ioctl_get_subvol_info(struct file *file, void __user *argp)
 		return -ENOMEM;
 	}
 
-	inode = file_inode(file);
 	fs_info = BTRFS_I(inode)->root->fs_info;
 
 	/* Get root_item of inode's subvolume */
 	key.objectid = BTRFS_I(inode)->root->root_key.objectid;
-	key.type = BTRFS_ROOT_ITEM_KEY;
-	key.offset = (u64)-1;
-	root = btrfs_read_fs_root_no_name(fs_info, &key);
+	root = btrfs_get_fs_root(fs_info, key.objectid, true);
 	if (IS_ERR(root)) {
 		ret = PTR_ERR(root);
-		goto out;
+		goto out_free;
 	}
 	root_item = &root->root_item;
 
@@ -2689,16 +3067,14 @@ static int btrfs_ioctl_get_subvol_info(struct file *file, void __user *argp)
 
 	if (key.objectid != BTRFS_FS_TREE_OBJECTID) {
 		/* Search root tree for ROOT_BACKREF of this subvolume */
-		root = fs_info->tree_root;
-
 		key.type = BTRFS_ROOT_BACKREF_KEY;
 		key.offset = 0;
-		ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
+		ret = btrfs_search_slot(NULL, fs_info->tree_root, &key, path, 0, 0);
 		if (ret < 0) {
 			goto out;
 		} else if (path->slots[0] >=
 			   btrfs_header_nritems(path->nodes[0])) {
-			ret = btrfs_next_leaf(root, path);
+			ret = btrfs_next_leaf(fs_info->tree_root, path);
 			if (ret < 0) {
 				goto out;
 			} else if (ret > 0) {
@@ -2719,7 +3095,7 @@ static int btrfs_ioctl_get_subvol_info(struct file *file, void __user *argp)
 
 			item_off = btrfs_item_ptr_offset(leaf, slot)
 					+ sizeof(struct btrfs_root_ref);
-			item_len = btrfs_item_size_nr(leaf, slot)
+			item_len = btrfs_item_size(leaf, slot)
 					- sizeof(struct btrfs_root_ref);
 			read_extent_buffer(leaf, subvol_info->name,
 					   item_off, item_len);
@@ -2733,8 +3109,10 @@ static int btrfs_ioctl_get_subvol_info(struct file *file, void __user *argp)
 		ret = -EFAULT;
 
 out:
+	btrfs_put_root(root);
+out_free:
 	btrfs_free_path(path);
-	kzfree(subvol_info);
+	kfree(subvol_info);
 	return ret;
 }
 
@@ -2742,15 +3120,14 @@ out:
  * Return ROOT_REF information of the subvolume containing this inode
  * except the subvolume name.
  */
-static int btrfs_ioctl_get_subvol_rootref(struct file *file, void __user *argp)
+static int btrfs_ioctl_get_subvol_rootref(struct btrfs_root *root,
+					  void __user *argp)
 {
 	struct btrfs_ioctl_get_subvol_rootref_args *rootrefs;
 	struct btrfs_root_ref *rref;
-	struct btrfs_root *root;
 	struct btrfs_path *path;
 	struct btrfs_key key;
 	struct extent_buffer *leaf;
-	struct inode *inode;
 	u64 objectid;
 	int slot;
 	int ret;
@@ -2766,15 +3143,13 @@ static int btrfs_ioctl_get_subvol_rootref(struct file *file, void __user *argp)
 		return PTR_ERR(rootrefs);
 	}
 
-	inode = file_inode(file);
-	root = BTRFS_I(inode)->root->fs_info->tree_root;
-	objectid = BTRFS_I(inode)->root->root_key.objectid;
-
+	objectid = root->root_key.objectid;
 	key.objectid = objectid;
 	key.type = BTRFS_ROOT_REF_KEY;
 	key.offset = rootrefs->min_treeid;
 	found = 0;
 
+	root = root->fs_info->tree_root;
 	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
 	if (ret < 0) {
 		goto out;
@@ -2836,7 +3211,8 @@ out:
 }
 
 static noinline int btrfs_ioctl_snap_destroy(struct file *file,
-					     void __user *arg)
+					     void __user *arg,
+					     bool destroy_v2)
 {
 	struct dentry *parent = file->f_path.dentry;
 	struct btrfs_fs_info *fs_info = btrfs_sb(parent->d_sb);
@@ -2845,34 +3221,145 @@ static noinline int btrfs_ioctl_snap_destroy(struct file *file,
 	struct inode *inode;
 	struct btrfs_root *root = BTRFS_I(dir)->root;
 	struct btrfs_root *dest = NULL;
-	struct btrfs_ioctl_vol_args *vol_args;
-	int namelen;
+	struct btrfs_ioctl_vol_args *vol_args = NULL;
+	struct btrfs_ioctl_vol_args_v2 *vol_args2 = NULL;
+	struct user_namespace *mnt_userns = file_mnt_user_ns(file);
+	char *subvol_name, *subvol_name_ptr = NULL;
+	int subvol_namelen;
 	int err = 0;
+	bool destroy_parent = false;
 
-	if (!S_ISDIR(dir->i_mode))
-		return -ENOTDIR;
+	/* We don't support snapshots with extent tree v2 yet. */
+	if (btrfs_fs_incompat(fs_info, EXTENT_TREE_V2)) {
+		btrfs_err(fs_info,
+			  "extent tree v2 doesn't support snapshot deletion yet");
+		return -EOPNOTSUPP;
+	}
 
-	vol_args = memdup_user(arg, sizeof(*vol_args));
-	if (IS_ERR(vol_args))
-		return PTR_ERR(vol_args);
+	if (destroy_v2) {
+		vol_args2 = memdup_user(arg, sizeof(*vol_args2));
+		if (IS_ERR(vol_args2))
+			return PTR_ERR(vol_args2);
 
-	vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
-	namelen = strlen(vol_args->name);
-	if (strchr(vol_args->name, '/') ||
-	    strncmp(vol_args->name, "..", namelen) == 0) {
-		err = -EINVAL;
-		goto out;
+		if (vol_args2->flags & ~BTRFS_SUBVOL_DELETE_ARGS_MASK) {
+			err = -EOPNOTSUPP;
+			goto out;
+		}
+
+		/*
+		 * If SPEC_BY_ID is not set, we are looking for the subvolume by
+		 * name, same as v1 currently does.
+		 */
+		if (!(vol_args2->flags & BTRFS_SUBVOL_SPEC_BY_ID)) {
+			vol_args2->name[BTRFS_SUBVOL_NAME_MAX] = 0;
+			subvol_name = vol_args2->name;
+
+			err = mnt_want_write_file(file);
+			if (err)
+				goto out;
+		} else {
+			struct inode *old_dir;
+
+			if (vol_args2->subvolid < BTRFS_FIRST_FREE_OBJECTID) {
+				err = -EINVAL;
+				goto out;
+			}
+
+			err = mnt_want_write_file(file);
+			if (err)
+				goto out;
+
+			dentry = btrfs_get_dentry(fs_info->sb,
+					BTRFS_FIRST_FREE_OBJECTID,
+					vol_args2->subvolid, 0, 0);
+			if (IS_ERR(dentry)) {
+				err = PTR_ERR(dentry);
+				goto out_drop_write;
+			}
+
+			/*
+			 * Change the default parent since the subvolume being
+			 * deleted can be outside of the current mount point.
+			 */
+			parent = btrfs_get_parent(dentry);
+
+			/*
+			 * At this point dentry->d_name can point to '/' if the
+			 * subvolume we want to destroy is outsite of the
+			 * current mount point, so we need to release the
+			 * current dentry and execute the lookup to return a new
+			 * one with ->d_name pointing to the
+			 * <mount point>/subvol_name.
+			 */
+			dput(dentry);
+			if (IS_ERR(parent)) {
+				err = PTR_ERR(parent);
+				goto out_drop_write;
+			}
+			old_dir = dir;
+			dir = d_inode(parent);
+
+			/*
+			 * If v2 was used with SPEC_BY_ID, a new parent was
+			 * allocated since the subvolume can be outside of the
+			 * current mount point. Later on we need to release this
+			 * new parent dentry.
+			 */
+			destroy_parent = true;
+
+			/*
+			 * On idmapped mounts, deletion via subvolid is
+			 * restricted to subvolumes that are immediate
+			 * ancestors of the inode referenced by the file
+			 * descriptor in the ioctl. Otherwise the idmapping
+			 * could potentially be abused to delete subvolumes
+			 * anywhere in the filesystem the user wouldn't be able
+			 * to delete without an idmapped mount.
+			 */
+			if (old_dir != dir && mnt_userns != &init_user_ns) {
+				err = -EOPNOTSUPP;
+				goto free_parent;
+			}
+
+			subvol_name_ptr = btrfs_get_subvol_name_from_objectid(
+						fs_info, vol_args2->subvolid);
+			if (IS_ERR(subvol_name_ptr)) {
+				err = PTR_ERR(subvol_name_ptr);
+				goto free_parent;
+			}
+			/* subvol_name_ptr is already nul terminated */
+			subvol_name = (char *)kbasename(subvol_name_ptr);
+		}
+	} else {
+		vol_args = memdup_user(arg, sizeof(*vol_args));
+		if (IS_ERR(vol_args))
+			return PTR_ERR(vol_args);
+
+		vol_args->name[BTRFS_PATH_NAME_MAX] = 0;
+		subvol_name = vol_args->name;
+
+		err = mnt_want_write_file(file);
+		if (err)
+			goto out;
 	}
 
-	err = mnt_want_write_file(file);
-	if (err)
-		goto out;
+	subvol_namelen = strlen(subvol_name);
 
+	if (strchr(subvol_name, '/') ||
+	    strncmp(subvol_name, "..", subvol_namelen) == 0) {
+		err = -EINVAL;
+		goto free_subvol_name;
+	}
+
+	if (!S_ISDIR(dir->i_mode)) {
+		err = -ENOTDIR;
+		goto free_subvol_name;
+	}
 
 	err = down_write_killable_nested(&dir->i_rwsem, I_MUTEX_PARENT);
 	if (err == -EINTR)
-		goto out_drop_write;
-	dentry = lookup_one_len(vol_args->name, parent, namelen);
+		goto free_subvol_name;
+	dentry = lookup_one(mnt_userns, subvol_name, parent, subvol_namelen);
 	if (IS_ERR(dentry)) {
 		err = PTR_ERR(dentry);
 		goto out_unlock_dir;
@@ -2914,13 +3401,13 @@ static noinline int btrfs_ioctl_snap_destroy(struct file *file,
 		if (root == dest)
 			goto out_dput;
 
-		err = inode_permission(inode, MAY_WRITE | MAY_EXEC);
+		err = inode_permission(mnt_userns, inode, MAY_WRITE | MAY_EXEC);
 		if (err)
 			goto out_dput;
 	}
 
 	/* check if subvolume may be deleted by a user */
-	err = btrfs_may_delete(dir, dentry, 1);
+	err = btrfs_may_delete(mnt_userns, dir, dentry, 1);
 	if (err)
 		goto out_dput;
 
@@ -2929,21 +3416,25 @@ static noinline int btrfs_ioctl_snap_destroy(struct file *file,
 		goto out_dput;
 	}
 
-	inode_lock(inode);
+	btrfs_inode_lock(inode, 0);
 	err = btrfs_delete_subvolume(dir, dentry);
-	inode_unlock(inode);
-	if (!err) {
-		fsnotify_rmdir(dir, dentry);
-		d_delete(dentry);
-	}
+	btrfs_inode_unlock(inode, 0);
+	if (!err)
+		d_delete_notify(dir, dentry);
 
 out_dput:
 	dput(dentry);
 out_unlock_dir:
-	inode_unlock(dir);
+	btrfs_inode_unlock(dir, 0);
+free_subvol_name:
+	kfree(subvol_name_ptr);
+free_parent:
+	if (destroy_parent)
+		dput(parent);
 out_drop_write:
 	mnt_drop_write_file(file);
 out:
+	kfree(vol_args2);
 	kfree(vol_args);
 	return err;
 }
@@ -2952,7 +3443,7 @@ static int btrfs_ioctl_defrag(struct file *file, void __user *argp)
 {
 	struct inode *inode = file_inode(file);
 	struct btrfs_root *root = BTRFS_I(inode)->root;
-	struct btrfs_ioctl_defrag_range_args *range;
+	struct btrfs_ioctl_defrag_range_args range = {0};
 	int ret;
 
 	ret = mnt_want_write_file(file);
@@ -2979,38 +3470,29 @@ static int btrfs_ioctl_defrag(struct file *file, void __user *argp)
 		 * running and allows defrag on files open in read-only mode.
 		 */
 		if (!capable(CAP_SYS_ADMIN) &&
-		    inode_permission(inode, MAY_WRITE)) {
+		    inode_permission(&init_user_ns, inode, MAY_WRITE)) {
 			ret = -EPERM;
 			goto out;
 		}
 
-		range = kzalloc(sizeof(*range), GFP_KERNEL);
-		if (!range) {
-			ret = -ENOMEM;
-			goto out;
-		}
-
 		if (argp) {
-			if (copy_from_user(range, argp,
-					   sizeof(*range))) {
+			if (copy_from_user(&range, argp, sizeof(range))) {
 				ret = -EFAULT;
-				kfree(range);
 				goto out;
 			}
 			/* compression requires us to start the IO */
-			if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) {
-				range->flags |= BTRFS_DEFRAG_RANGE_START_IO;
-				range->extent_thresh = (u32)-1;
+			if ((range.flags & BTRFS_DEFRAG_RANGE_COMPRESS)) {
+				range.flags |= BTRFS_DEFRAG_RANGE_START_IO;
+				range.extent_thresh = (u32)-1;
 			}
 		} else {
 			/* the rest are all set to zero by kzalloc */
-			range->len = (u64)-1;
+			range.len = (u64)-1;
 		}
-		ret = btrfs_defrag_file(file_inode(file), file,
-					range, BTRFS_OLDEST_GENERATION, 0);
+		ret = btrfs_defrag_file(file_inode(file), &file->f_ra,
+					&range, BTRFS_OLDEST_GENERATION, 0);
 		if (ret > 0)
 			ret = 0;
-		kfree(range);
 		break;
 	default:
 		ret = -EINVAL;
@@ -3023,13 +3505,30 @@ out:
 static long btrfs_ioctl_add_dev(struct btrfs_fs_info *fs_info, void __user *arg)
 {
 	struct btrfs_ioctl_vol_args *vol_args;
+	bool restore_op = false;
 	int ret;
 
 	if (!capable(CAP_SYS_ADMIN))
 		return -EPERM;
 
-	if (test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags))
-		return BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
+	if (btrfs_fs_incompat(fs_info, EXTENT_TREE_V2)) {
+		btrfs_err(fs_info, "device add not supported on extent tree v2 yet");
+		return -EINVAL;
+	}
+
+	if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_DEV_ADD)) {
+		if (!btrfs_exclop_start_try_lock(fs_info, BTRFS_EXCLOP_DEV_ADD))
+			return BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
+
+		/*
+		 * We can do the device add because we have a paused balanced,
+		 * change the exclusive op type and remember we should bring
+		 * back the paused balance
+		 */
+		fs_info->exclusive_operation = BTRFS_EXCLOP_DEV_ADD;
+		btrfs_exclop_start_unlock(fs_info);
+		restore_op = true;
+	}
 
 	vol_args = memdup_user(arg, sizeof(*vol_args));
 	if (IS_ERR(vol_args)) {
@@ -3045,48 +3544,60 @@ static long btrfs_ioctl_add_dev(struct btrfs_fs_info *fs_info, void __user *arg)
 
 	kfree(vol_args);
 out:
-	clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
+	if (restore_op)
+		btrfs_exclop_balance(fs_info, BTRFS_EXCLOP_BALANCE_PAUSED);
+	else
+		btrfs_exclop_finish(fs_info);
 	return ret;
 }
 
 static long btrfs_ioctl_rm_dev_v2(struct file *file, void __user *arg)
 {
+	BTRFS_DEV_LOOKUP_ARGS(args);
 	struct inode *inode = file_inode(file);
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	struct btrfs_ioctl_vol_args_v2 *vol_args;
+	struct block_device *bdev = NULL;
+	fmode_t mode;
 	int ret;
+	bool cancel = false;
 
 	if (!capable(CAP_SYS_ADMIN))
 		return -EPERM;
 
-	ret = mnt_want_write_file(file);
-	if (ret)
-		return ret;
-
 	vol_args = memdup_user(arg, sizeof(*vol_args));
-	if (IS_ERR(vol_args)) {
-		ret = PTR_ERR(vol_args);
-		goto err_drop;
-	}
+	if (IS_ERR(vol_args))
+		return PTR_ERR(vol_args);
 
-	/* Check for compatibility reject unknown flags */
-	if (vol_args->flags & ~BTRFS_VOL_ARG_V2_FLAGS_SUPPORTED) {
+	if (vol_args->flags & ~BTRFS_DEVICE_REMOVE_ARGS_MASK) {
 		ret = -EOPNOTSUPP;
 		goto out;
 	}
 
-	if (test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags)) {
-		ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
-		goto out;
-	}
-
+	vol_args->name[BTRFS_SUBVOL_NAME_MAX] = '\0';
 	if (vol_args->flags & BTRFS_DEVICE_SPEC_BY_ID) {
-		ret = btrfs_rm_device(fs_info, NULL, vol_args->devid);
+		args.devid = vol_args->devid;
+	} else if (!strcmp("cancel", vol_args->name)) {
+		cancel = true;
 	} else {
-		vol_args->name[BTRFS_SUBVOL_NAME_MAX] = '\0';
-		ret = btrfs_rm_device(fs_info, vol_args->name, 0);
+		ret = btrfs_get_dev_args_from_path(fs_info, &args, vol_args->name);
+		if (ret)
+			goto out;
 	}
-	clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
+
+	ret = mnt_want_write_file(file);
+	if (ret)
+		goto out;
+
+	ret = exclop_start_or_cancel_reloc(fs_info, BTRFS_EXCLOP_DEV_REMOVE,
+					   cancel);
+	if (ret)
+		goto err_drop;
+
+	/* Exclusive operation is now claimed */
+	ret = btrfs_rm_device(fs_info, &args, &bdev, &mode);
+
+	btrfs_exclop_finish(fs_info);
 
 	if (!ret) {
 		if (vol_args->flags & BTRFS_DEVICE_SPEC_BY_ID)
@@ -3096,49 +3607,62 @@ static long btrfs_ioctl_rm_dev_v2(struct file *file, void __user *arg)
 			btrfs_info(fs_info, "device deleted: %s",
 					vol_args->name);
 	}
-out:
-	kfree(vol_args);
 err_drop:
 	mnt_drop_write_file(file);
+	if (bdev)
+		blkdev_put(bdev, mode);
+out:
+	btrfs_put_dev_args_from_path(&args);
+	kfree(vol_args);
 	return ret;
 }
 
 static long btrfs_ioctl_rm_dev(struct file *file, void __user *arg)
 {
+	BTRFS_DEV_LOOKUP_ARGS(args);
 	struct inode *inode = file_inode(file);
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	struct btrfs_ioctl_vol_args *vol_args;
+	struct block_device *bdev = NULL;
+	fmode_t mode;
 	int ret;
+	bool cancel = false;
 
 	if (!capable(CAP_SYS_ADMIN))
 		return -EPERM;
 
-	ret = mnt_want_write_file(file);
-	if (ret)
-		return ret;
+	vol_args = memdup_user(arg, sizeof(*vol_args));
+	if (IS_ERR(vol_args))
+		return PTR_ERR(vol_args);
 
-	if (test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags)) {
-		ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
-		goto out_drop_write;
+	vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
+	if (!strcmp("cancel", vol_args->name)) {
+		cancel = true;
+	} else {
+		ret = btrfs_get_dev_args_from_path(fs_info, &args, vol_args->name);
+		if (ret)
+			goto out;
 	}
 
-	vol_args = memdup_user(arg, sizeof(*vol_args));
-	if (IS_ERR(vol_args)) {
-		ret = PTR_ERR(vol_args);
+	ret = mnt_want_write_file(file);
+	if (ret)
 		goto out;
-	}
 
-	vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
-	ret = btrfs_rm_device(fs_info, vol_args->name, 0);
+	ret = exclop_start_or_cancel_reloc(fs_info, BTRFS_EXCLOP_DEV_REMOVE,
+					   cancel);
+	if (ret == 0) {
+		ret = btrfs_rm_device(fs_info, &args, &bdev, &mode);
+		if (!ret)
+			btrfs_info(fs_info, "disk deleted %s", vol_args->name);
+		btrfs_exclop_finish(fs_info);
+	}
 
-	if (!ret)
-		btrfs_info(fs_info, "disk deleted %s", vol_args->name);
-	kfree(vol_args);
-out:
-	clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
-out_drop_write:
 	mnt_drop_write_file(file);
-
+	if (bdev)
+		blkdev_put(bdev, mode);
+out:
+	btrfs_put_dev_args_from_path(&args);
+	kfree(vol_args);
 	return ret;
 }
 
@@ -3148,11 +3672,15 @@ static long btrfs_ioctl_fs_info(struct btrfs_fs_info *fs_info,
 	struct btrfs_ioctl_fs_info_args *fi_args;
 	struct btrfs_device *device;
 	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
+	u64 flags_in;
 	int ret = 0;
 
-	fi_args = kzalloc(sizeof(*fi_args), GFP_KERNEL);
-	if (!fi_args)
-		return -ENOMEM;
+	fi_args = memdup_user(arg, sizeof(*fi_args));
+	if (IS_ERR(fi_args))
+		return PTR_ERR(fi_args);
+
+	flags_in = fi_args->flags;
+	memset(fi_args, 0, sizeof(*fi_args));
 
 	rcu_read_lock();
 	fi_args->num_devices = fs_devices->num_devices;
@@ -3168,6 +3696,23 @@ static long btrfs_ioctl_fs_info(struct btrfs_fs_info *fs_info,
 	fi_args->sectorsize = fs_info->sectorsize;
 	fi_args->clone_alignment = fs_info->sectorsize;
 
+	if (flags_in & BTRFS_FS_INFO_FLAG_CSUM_INFO) {
+		fi_args->csum_type = btrfs_super_csum_type(fs_info->super_copy);
+		fi_args->csum_size = btrfs_super_csum_size(fs_info->super_copy);
+		fi_args->flags |= BTRFS_FS_INFO_FLAG_CSUM_INFO;
+	}
+
+	if (flags_in & BTRFS_FS_INFO_FLAG_GENERATION) {
+		fi_args->generation = fs_info->generation;
+		fi_args->flags |= BTRFS_FS_INFO_FLAG_GENERATION;
+	}
+
+	if (flags_in & BTRFS_FS_INFO_FLAG_METADATA_UUID) {
+		memcpy(&fi_args->metadata_uuid, fs_devices->metadata_uuid,
+		       sizeof(fi_args->metadata_uuid));
+		fi_args->flags |= BTRFS_FS_INFO_FLAG_METADATA_UUID;
+	}
+
 	if (copy_to_user(arg, fi_args, sizeof(*fi_args)))
 		ret = -EFAULT;
 
@@ -3178,22 +3723,21 @@ static long btrfs_ioctl_fs_info(struct btrfs_fs_info *fs_info,
 static long btrfs_ioctl_dev_info(struct btrfs_fs_info *fs_info,
 				 void __user *arg)
 {
+	BTRFS_DEV_LOOKUP_ARGS(args);
 	struct btrfs_ioctl_dev_info_args *di_args;
 	struct btrfs_device *dev;
 	int ret = 0;
-	char *s_uuid = NULL;
 
 	di_args = memdup_user(arg, sizeof(*di_args));
 	if (IS_ERR(di_args))
 		return PTR_ERR(di_args);
 
+	args.devid = di_args->devid;
 	if (!btrfs_is_empty_uuid(di_args->uuid))
-		s_uuid = di_args->uuid;
+		args.uuid = di_args->uuid;
 
 	rcu_read_lock();
-	dev = btrfs_find_device(fs_info->fs_devices, di_args->devid, s_uuid,
-				NULL, true);
-
+	dev = btrfs_find_device(fs_info->fs_devices, &args);
 	if (!dev) {
 		ret = -ENODEV;
 		goto out;
@@ -3220,733 +3764,6 @@ out:
 	return ret;
 }
 
-static void btrfs_double_extent_unlock(struct inode *inode1, u64 loff1,
-				       struct inode *inode2, u64 loff2, u64 len)
-{
-	unlock_extent(&BTRFS_I(inode1)->io_tree, loff1, loff1 + len - 1);
-	unlock_extent(&BTRFS_I(inode2)->io_tree, loff2, loff2 + len - 1);
-}
-
-static void btrfs_double_extent_lock(struct inode *inode1, u64 loff1,
-				     struct inode *inode2, u64 loff2, u64 len)
-{
-	if (inode1 < inode2) {
-		swap(inode1, inode2);
-		swap(loff1, loff2);
-	} else if (inode1 == inode2 && loff2 < loff1) {
-		swap(loff1, loff2);
-	}
-	lock_extent(&BTRFS_I(inode1)->io_tree, loff1, loff1 + len - 1);
-	lock_extent(&BTRFS_I(inode2)->io_tree, loff2, loff2 + len - 1);
-}
-
-static int btrfs_extent_same_range(struct inode *src, u64 loff, u64 len,
-				   struct inode *dst, u64 dst_loff)
-{
-	const u64 bs = BTRFS_I(src)->root->fs_info->sb->s_blocksize;
-	int ret;
-
-	/*
-	 * Lock destination range to serialize with concurrent readpages() and
-	 * source range to serialize with relocation.
-	 */
-	btrfs_double_extent_lock(src, loff, dst, dst_loff, len);
-	ret = btrfs_clone(src, dst, loff, len, ALIGN(len, bs), dst_loff, 1);
-	btrfs_double_extent_unlock(src, loff, dst, dst_loff, len);
-
-	return ret;
-}
-
-#define BTRFS_MAX_DEDUPE_LEN	SZ_16M
-
-static int btrfs_extent_same(struct inode *src, u64 loff, u64 olen,
-			     struct inode *dst, u64 dst_loff)
-{
-	int ret;
-	u64 i, tail_len, chunk_count;
-	struct btrfs_root *root_dst = BTRFS_I(dst)->root;
-
-	spin_lock(&root_dst->root_item_lock);
-	if (root_dst->send_in_progress) {
-		btrfs_warn_rl(root_dst->fs_info,
-"cannot deduplicate to root %llu while send operations are using it (%d in progress)",
-			      root_dst->root_key.objectid,
-			      root_dst->send_in_progress);
-		spin_unlock(&root_dst->root_item_lock);
-		return -EAGAIN;
-	}
-	root_dst->dedupe_in_progress++;
-	spin_unlock(&root_dst->root_item_lock);
-
-	tail_len = olen % BTRFS_MAX_DEDUPE_LEN;
-	chunk_count = div_u64(olen, BTRFS_MAX_DEDUPE_LEN);
-
-	for (i = 0; i < chunk_count; i++) {
-		ret = btrfs_extent_same_range(src, loff, BTRFS_MAX_DEDUPE_LEN,
-					      dst, dst_loff);
-		if (ret)
-			goto out;
-
-		loff += BTRFS_MAX_DEDUPE_LEN;
-		dst_loff += BTRFS_MAX_DEDUPE_LEN;
-	}
-
-	if (tail_len > 0)
-		ret = btrfs_extent_same_range(src, loff, tail_len, dst,
-					      dst_loff);
-out:
-	spin_lock(&root_dst->root_item_lock);
-	root_dst->dedupe_in_progress--;
-	spin_unlock(&root_dst->root_item_lock);
-
-	return ret;
-}
-
-static int clone_finish_inode_update(struct btrfs_trans_handle *trans,
-				     struct inode *inode,
-				     u64 endoff,
-				     const u64 destoff,
-				     const u64 olen,
-				     int no_time_update)
-{
-	struct btrfs_root *root = BTRFS_I(inode)->root;
-	int ret;
-
-	inode_inc_iversion(inode);
-	if (!no_time_update)
-		inode->i_mtime = inode->i_ctime = current_time(inode);
-	/*
-	 * We round up to the block size at eof when determining which
-	 * extents to clone above, but shouldn't round up the file size.
-	 */
-	if (endoff > destoff + olen)
-		endoff = destoff + olen;
-	if (endoff > inode->i_size)
-		btrfs_i_size_write(BTRFS_I(inode), endoff);
-
-	ret = btrfs_update_inode(trans, root, inode);
-	if (ret) {
-		btrfs_abort_transaction(trans, ret);
-		btrfs_end_transaction(trans);
-		goto out;
-	}
-	ret = btrfs_end_transaction(trans);
-out:
-	return ret;
-}
-
-/*
- * Make sure we do not end up inserting an inline extent into a file that has
- * already other (non-inline) extents. If a file has an inline extent it can
- * not have any other extents and the (single) inline extent must start at the
- * file offset 0. Failing to respect these rules will lead to file corruption,
- * resulting in EIO errors on read/write operations, hitting BUG_ON's in mm, etc
- *
- * We can have extents that have been already written to disk or we can have
- * dirty ranges still in delalloc, in which case the extent maps and items are
- * created only when we run delalloc, and the delalloc ranges might fall outside
- * the range we are currently locking in the inode's io tree. So we check the
- * inode's i_size because of that (i_size updates are done while holding the
- * i_mutex, which we are holding here).
- * We also check to see if the inode has a size not greater than "datal" but has
- * extents beyond it, due to an fallocate with FALLOC_FL_KEEP_SIZE (and we are
- * protected against such concurrent fallocate calls by the i_mutex).
- *
- * If the file has no extents but a size greater than datal, do not allow the
- * copy because we would need turn the inline extent into a non-inline one (even
- * with NO_HOLES enabled). If we find our destination inode only has one inline
- * extent, just overwrite it with the source inline extent if its size is less
- * than the source extent's size, or we could copy the source inline extent's
- * data into the destination inode's inline extent if the later is greater then
- * the former.
- */
-static int clone_copy_inline_extent(struct inode *dst,
-				    struct btrfs_trans_handle *trans,
-				    struct btrfs_path *path,
-				    struct btrfs_key *new_key,
-				    const u64 drop_start,
-				    const u64 datal,
-				    const u64 skip,
-				    const u64 size,
-				    char *inline_data)
-{
-	struct btrfs_fs_info *fs_info = btrfs_sb(dst->i_sb);
-	struct btrfs_root *root = BTRFS_I(dst)->root;
-	const u64 aligned_end = ALIGN(new_key->offset + datal,
-				      fs_info->sectorsize);
-	int ret;
-	struct btrfs_key key;
-
-	if (new_key->offset > 0)
-		return -EOPNOTSUPP;
-
-	key.objectid = btrfs_ino(BTRFS_I(dst));
-	key.type = BTRFS_EXTENT_DATA_KEY;
-	key.offset = 0;
-	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
-	if (ret < 0) {
-		return ret;
-	} else if (ret > 0) {
-		if (path->slots[0] >= btrfs_header_nritems(path->nodes[0])) {
-			ret = btrfs_next_leaf(root, path);
-			if (ret < 0)
-				return ret;
-			else if (ret > 0)
-				goto copy_inline_extent;
-		}
-		btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
-		if (key.objectid == btrfs_ino(BTRFS_I(dst)) &&
-		    key.type == BTRFS_EXTENT_DATA_KEY) {
-			ASSERT(key.offset > 0);
-			return -EOPNOTSUPP;
-		}
-	} else if (i_size_read(dst) <= datal) {
-		struct btrfs_file_extent_item *ei;
-		u64 ext_len;
-
-		/*
-		 * If the file size is <= datal, make sure there are no other
-		 * extents following (can happen do to an fallocate call with
-		 * the flag FALLOC_FL_KEEP_SIZE).
-		 */
-		ei = btrfs_item_ptr(path->nodes[0], path->slots[0],
-				    struct btrfs_file_extent_item);
-		/*
-		 * If it's an inline extent, it can not have other extents
-		 * following it.
-		 */
-		if (btrfs_file_extent_type(path->nodes[0], ei) ==
-		    BTRFS_FILE_EXTENT_INLINE)
-			goto copy_inline_extent;
-
-		ext_len = btrfs_file_extent_num_bytes(path->nodes[0], ei);
-		if (ext_len > aligned_end)
-			return -EOPNOTSUPP;
-
-		ret = btrfs_next_item(root, path);
-		if (ret < 0) {
-			return ret;
-		} else if (ret == 0) {
-			btrfs_item_key_to_cpu(path->nodes[0], &key,
-					      path->slots[0]);
-			if (key.objectid == btrfs_ino(BTRFS_I(dst)) &&
-			    key.type == BTRFS_EXTENT_DATA_KEY)
-				return -EOPNOTSUPP;
-		}
-	}
-
-copy_inline_extent:
-	/*
-	 * We have no extent items, or we have an extent at offset 0 which may
-	 * or may not be inlined. All these cases are dealt the same way.
-	 */
-	if (i_size_read(dst) > datal) {
-		/*
-		 * If the destination inode has an inline extent...
-		 * This would require copying the data from the source inline
-		 * extent into the beginning of the destination's inline extent.
-		 * But this is really complex, both extents can be compressed
-		 * or just one of them, which would require decompressing and
-		 * re-compressing data (which could increase the new compressed
-		 * size, not allowing the compressed data to fit anymore in an
-		 * inline extent).
-		 * So just don't support this case for now (it should be rare,
-		 * we are not really saving space when cloning inline extents).
-		 */
-		return -EOPNOTSUPP;
-	}
-
-	btrfs_release_path(path);
-	ret = btrfs_drop_extents(trans, root, dst, drop_start, aligned_end, 1);
-	if (ret)
-		return ret;
-	ret = btrfs_insert_empty_item(trans, root, path, new_key, size);
-	if (ret)
-		return ret;
-
-	if (skip) {
-		const u32 start = btrfs_file_extent_calc_inline_size(0);
-
-		memmove(inline_data + start, inline_data + start + skip, datal);
-	}
-
-	write_extent_buffer(path->nodes[0], inline_data,
-			    btrfs_item_ptr_offset(path->nodes[0],
-						  path->slots[0]),
-			    size);
-	inode_add_bytes(dst, datal);
-	set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &BTRFS_I(dst)->runtime_flags);
-
-	return 0;
-}
-
-/**
- * btrfs_clone() - clone a range from inode file to another
- *
- * @src: Inode to clone from
- * @inode: Inode to clone to
- * @off: Offset within source to start clone from
- * @olen: Original length, passed by user, of range to clone
- * @olen_aligned: Block-aligned value of olen
- * @destoff: Offset within @inode to start clone
- * @no_time_update: Whether to update mtime/ctime on the target inode
- */
-static int btrfs_clone(struct inode *src, struct inode *inode,
-		       const u64 off, const u64 olen, const u64 olen_aligned,
-		       const u64 destoff, int no_time_update)
-{
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	struct btrfs_root *root = BTRFS_I(inode)->root;
-	struct btrfs_path *path = NULL;
-	struct extent_buffer *leaf;
-	struct btrfs_trans_handle *trans;
-	char *buf = NULL;
-	struct btrfs_key key;
-	u32 nritems;
-	int slot;
-	int ret;
-	const u64 len = olen_aligned;
-	u64 last_dest_end = destoff;
-
-	ret = -ENOMEM;
-	buf = kvmalloc(fs_info->nodesize, GFP_KERNEL);
-	if (!buf)
-		return ret;
-
-	path = btrfs_alloc_path();
-	if (!path) {
-		kvfree(buf);
-		return ret;
-	}
-
-	path->reada = READA_FORWARD;
-	/* clone data */
-	key.objectid = btrfs_ino(BTRFS_I(src));
-	key.type = BTRFS_EXTENT_DATA_KEY;
-	key.offset = off;
-
-	while (1) {
-		u64 next_key_min_offset = key.offset + 1;
-		struct btrfs_file_extent_item *extent;
-		int type;
-		u32 size;
-		struct btrfs_key new_key;
-		u64 disko = 0, diskl = 0;
-		u64 datao = 0, datal = 0;
-		u8 comp;
-		u64 drop_start;
-
-		/*
-		 * note the key will change type as we walk through the
-		 * tree.
-		 */
-		path->leave_spinning = 1;
-		ret = btrfs_search_slot(NULL, BTRFS_I(src)->root, &key, path,
-				0, 0);
-		if (ret < 0)
-			goto out;
-		/*
-		 * First search, if no extent item that starts at offset off was
-		 * found but the previous item is an extent item, it's possible
-		 * it might overlap our target range, therefore process it.
-		 */
-		if (key.offset == off && ret > 0 && path->slots[0] > 0) {
-			btrfs_item_key_to_cpu(path->nodes[0], &key,
-					      path->slots[0] - 1);
-			if (key.type == BTRFS_EXTENT_DATA_KEY)
-				path->slots[0]--;
-		}
-
-		nritems = btrfs_header_nritems(path->nodes[0]);
-process_slot:
-		if (path->slots[0] >= nritems) {
-			ret = btrfs_next_leaf(BTRFS_I(src)->root, path);
-			if (ret < 0)
-				goto out;
-			if (ret > 0)
-				break;
-			nritems = btrfs_header_nritems(path->nodes[0]);
-		}
-		leaf = path->nodes[0];
-		slot = path->slots[0];
-
-		btrfs_item_key_to_cpu(leaf, &key, slot);
-		if (key.type > BTRFS_EXTENT_DATA_KEY ||
-		    key.objectid != btrfs_ino(BTRFS_I(src)))
-			break;
-
-		ASSERT(key.type == BTRFS_EXTENT_DATA_KEY);
-
-		extent = btrfs_item_ptr(leaf, slot,
-					struct btrfs_file_extent_item);
-		comp = btrfs_file_extent_compression(leaf, extent);
-		type = btrfs_file_extent_type(leaf, extent);
-		if (type == BTRFS_FILE_EXTENT_REG ||
-		    type == BTRFS_FILE_EXTENT_PREALLOC) {
-			disko = btrfs_file_extent_disk_bytenr(leaf, extent);
-			diskl = btrfs_file_extent_disk_num_bytes(leaf, extent);
-			datao = btrfs_file_extent_offset(leaf, extent);
-			datal = btrfs_file_extent_num_bytes(leaf, extent);
-		} else if (type == BTRFS_FILE_EXTENT_INLINE) {
-			/* Take upper bound, may be compressed */
-			datal = btrfs_file_extent_ram_bytes(leaf, extent);
-		}
-
-		/*
-		 * The first search might have left us at an extent item that
-		 * ends before our target range's start, can happen if we have
-		 * holes and NO_HOLES feature enabled.
-		 */
-		if (key.offset + datal <= off) {
-			path->slots[0]++;
-			goto process_slot;
-		} else if (key.offset >= off + len) {
-			break;
-		}
-		next_key_min_offset = key.offset + datal;
-		size = btrfs_item_size_nr(leaf, slot);
-		read_extent_buffer(leaf, buf, btrfs_item_ptr_offset(leaf, slot),
-				   size);
-
-		btrfs_release_path(path);
-		path->leave_spinning = 0;
-
-		memcpy(&new_key, &key, sizeof(new_key));
-		new_key.objectid = btrfs_ino(BTRFS_I(inode));
-		if (off <= key.offset)
-			new_key.offset = key.offset + destoff - off;
-		else
-			new_key.offset = destoff;
-
-		/*
-		 * Deal with a hole that doesn't have an extent item that
-		 * represents it (NO_HOLES feature enabled).
-		 * This hole is either in the middle of the cloning range or at
-		 * the beginning (fully overlaps it or partially overlaps it).
-		 */
-		if (new_key.offset != last_dest_end)
-			drop_start = last_dest_end;
-		else
-			drop_start = new_key.offset;
-
-		if (type == BTRFS_FILE_EXTENT_REG ||
-		    type == BTRFS_FILE_EXTENT_PREALLOC) {
-			struct btrfs_clone_extent_info clone_info;
-
-			/*
-			 *    a  | --- range to clone ---|  b
-			 * | ------------- extent ------------- |
-			 */
-
-			/* Subtract range b */
-			if (key.offset + datal > off + len)
-				datal = off + len - key.offset;
-
-			/* Subtract range a */
-			if (off > key.offset) {
-				datao += off - key.offset;
-				datal -= off - key.offset;
-			}
-
-			clone_info.disk_offset = disko;
-			clone_info.disk_len = diskl;
-			clone_info.data_offset = datao;
-			clone_info.data_len = datal;
-			clone_info.file_offset = new_key.offset;
-			clone_info.extent_buf = buf;
-			clone_info.item_size = size;
-			ret = btrfs_punch_hole_range(inode, path,
-						     drop_start,
-						     new_key.offset + datal - 1,
-						     &clone_info, &trans);
-			if (ret)
-				goto out;
-		} else if (type == BTRFS_FILE_EXTENT_INLINE) {
-			u64 skip = 0;
-			u64 trim = 0;
-
-			if (off > key.offset) {
-				skip = off - key.offset;
-				new_key.offset += skip;
-			}
-
-			if (key.offset + datal > off + len)
-				trim = key.offset + datal - (off + len);
-
-			if (comp && (skip || trim)) {
-				ret = -EINVAL;
-				goto out;
-			}
-			size -= skip + trim;
-			datal -= skip + trim;
-
-			/*
-			 * If our extent is inline, we know we will drop or
-			 * adjust at most 1 extent item in the destination root.
-			 *
-			 * 1 - adjusting old extent (we may have to split it)
-			 * 1 - add new extent
-			 * 1 - inode update
-			 */
-			trans = btrfs_start_transaction(root, 3);
-			if (IS_ERR(trans)) {
-				ret = PTR_ERR(trans);
-				goto out;
-			}
-
-			ret = clone_copy_inline_extent(inode, trans, path,
-						       &new_key, drop_start,
-						       datal, skip, size, buf);
-			if (ret) {
-				if (ret != -EOPNOTSUPP)
-					btrfs_abort_transaction(trans, ret);
-				btrfs_end_transaction(trans);
-				goto out;
-			}
-		}
-
-		btrfs_release_path(path);
-
-		last_dest_end = ALIGN(new_key.offset + datal,
-				      fs_info->sectorsize);
-		ret = clone_finish_inode_update(trans, inode, last_dest_end,
-						destoff, olen, no_time_update);
-		if (ret)
-			goto out;
-		if (new_key.offset + datal >= destoff + len)
-			break;
-
-		btrfs_release_path(path);
-		key.offset = next_key_min_offset;
-
-		if (fatal_signal_pending(current)) {
-			ret = -EINTR;
-			goto out;
-		}
-	}
-	ret = 0;
-
-	if (last_dest_end < destoff + len) {
-		/*
-		 * We have an implicit hole that fully or partially overlaps our
-		 * cloning range at its end. This means that we either have the
-		 * NO_HOLES feature enabled or the implicit hole happened due to
-		 * mixing buffered and direct IO writes against this file.
-		 */
-		btrfs_release_path(path);
-		path->leave_spinning = 0;
-
-		ret = btrfs_punch_hole_range(inode, path,
-					     last_dest_end, destoff + len - 1,
-					     NULL, &trans);
-		if (ret)
-			goto out;
-
-		ret = clone_finish_inode_update(trans, inode, destoff + len,
-						destoff, olen, no_time_update);
-	}
-
-out:
-	btrfs_free_path(path);
-	kvfree(buf);
-	return ret;
-}
-
-static noinline int btrfs_clone_files(struct file *file, struct file *file_src,
-					u64 off, u64 olen, u64 destoff)
-{
-	struct inode *inode = file_inode(file);
-	struct inode *src = file_inode(file_src);
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	int ret;
-	u64 len = olen;
-	u64 bs = fs_info->sb->s_blocksize;
-
-	/*
-	 * TODO:
-	 * - split compressed inline extents.  annoying: we need to
-	 *   decompress into destination's address_space (the file offset
-	 *   may change, so source mapping won't do), then recompress (or
-	 *   otherwise reinsert) a subrange.
-	 *
-	 * - split destination inode's inline extents.  The inline extents can
-	 *   be either compressed or non-compressed.
-	 */
-
-	/*
-	 * VFS's generic_remap_file_range_prep() protects us from cloning the
-	 * eof block into the middle of a file, which would result in corruption
-	 * if the file size is not blocksize aligned. So we don't need to check
-	 * for that case here.
-	 */
-	if (off + len == src->i_size)
-		len = ALIGN(src->i_size, bs) - off;
-
-	if (destoff > inode->i_size) {
-		const u64 wb_start = ALIGN_DOWN(inode->i_size, bs);
-
-		ret = btrfs_cont_expand(inode, inode->i_size, destoff);
-		if (ret)
-			return ret;
-		/*
-		 * We may have truncated the last block if the inode's size is
-		 * not sector size aligned, so we need to wait for writeback to
-		 * complete before proceeding further, otherwise we can race
-		 * with cloning and attempt to increment a reference to an
-		 * extent that no longer exists (writeback completed right after
-		 * we found the previous extent covering eof and before we
-		 * attempted to increment its reference count).
-		 */
-		ret = btrfs_wait_ordered_range(inode, wb_start,
-					       destoff - wb_start);
-		if (ret)
-			return ret;
-	}
-
-	/*
-	 * Lock destination range to serialize with concurrent readpages() and
-	 * source range to serialize with relocation.
-	 */
-	btrfs_double_extent_lock(src, off, inode, destoff, len);
-	ret = btrfs_clone(src, inode, off, olen, len, destoff, 0);
-	btrfs_double_extent_unlock(src, off, inode, destoff, len);
-	/*
-	 * Truncate page cache pages so that future reads will see the cloned
-	 * data immediately and not the previous data.
-	 */
-	truncate_inode_pages_range(&inode->i_data,
-				round_down(destoff, PAGE_SIZE),
-				round_up(destoff + len, PAGE_SIZE) - 1);
-
-	return ret;
-}
-
-static int btrfs_remap_file_range_prep(struct file *file_in, loff_t pos_in,
-				       struct file *file_out, loff_t pos_out,
-				       loff_t *len, unsigned int remap_flags)
-{
-	struct inode *inode_in = file_inode(file_in);
-	struct inode *inode_out = file_inode(file_out);
-	u64 bs = BTRFS_I(inode_out)->root->fs_info->sb->s_blocksize;
-	bool same_inode = inode_out == inode_in;
-	u64 wb_len;
-	int ret;
-
-	if (!(remap_flags & REMAP_FILE_DEDUP)) {
-		struct btrfs_root *root_out = BTRFS_I(inode_out)->root;
-
-		if (btrfs_root_readonly(root_out))
-			return -EROFS;
-
-		if (file_in->f_path.mnt != file_out->f_path.mnt ||
-		    inode_in->i_sb != inode_out->i_sb)
-			return -EXDEV;
-	}
-
-	/* don't make the dst file partly checksummed */
-	if ((BTRFS_I(inode_in)->flags & BTRFS_INODE_NODATASUM) !=
-	    (BTRFS_I(inode_out)->flags & BTRFS_INODE_NODATASUM)) {
-		return -EINVAL;
-	}
-
-	/*
-	 * Now that the inodes are locked, we need to start writeback ourselves
-	 * and can not rely on the writeback from the VFS's generic helper
-	 * generic_remap_file_range_prep() because:
-	 *
-	 * 1) For compression we must call filemap_fdatawrite_range() range
-	 *    twice (btrfs_fdatawrite_range() does it for us), and the generic
-	 *    helper only calls it once;
-	 *
-	 * 2) filemap_fdatawrite_range(), called by the generic helper only
-	 *    waits for the writeback to complete, i.e. for IO to be done, and
-	 *    not for the ordered extents to complete. We need to wait for them
-	 *    to complete so that new file extent items are in the fs tree.
-	 */
-	if (*len == 0 && !(remap_flags & REMAP_FILE_DEDUP))
-		wb_len = ALIGN(inode_in->i_size, bs) - ALIGN_DOWN(pos_in, bs);
-	else
-		wb_len = ALIGN(*len, bs);
-
-	/*
-	 * Since we don't lock ranges, wait for ongoing lockless dio writes (as
-	 * any in progress could create its ordered extents after we wait for
-	 * existing ordered extents below).
-	 */
-	inode_dio_wait(inode_in);
-	if (!same_inode)
-		inode_dio_wait(inode_out);
-
-	/*
-	 * Workaround to make sure NOCOW buffered write reach disk as NOCOW.
-	 *
-	 * Btrfs' back references do not have a block level granularity, they
-	 * work at the whole extent level.
-	 * NOCOW buffered write without data space reserved may not be able
-	 * to fall back to CoW due to lack of data space, thus could cause
-	 * data loss.
-	 *
-	 * Here we take a shortcut by flushing the whole inode, so that all
-	 * nocow write should reach disk as nocow before we increase the
-	 * reference of the extent. We could do better by only flushing NOCOW
-	 * data, but that needs extra accounting.
-	 *
-	 * Also we don't need to check ASYNC_EXTENT, as async extent will be
-	 * CoWed anyway, not affecting nocow part.
-	 */
-	ret = filemap_flush(inode_in->i_mapping);
-	if (ret < 0)
-		return ret;
-
-	ret = btrfs_wait_ordered_range(inode_in, ALIGN_DOWN(pos_in, bs),
-				       wb_len);
-	if (ret < 0)
-		return ret;
-	ret = btrfs_wait_ordered_range(inode_out, ALIGN_DOWN(pos_out, bs),
-				       wb_len);
-	if (ret < 0)
-		return ret;
-
-	return generic_remap_file_range_prep(file_in, pos_in, file_out, pos_out,
-					    len, remap_flags);
-}
-
-loff_t btrfs_remap_file_range(struct file *src_file, loff_t off,
-		struct file *dst_file, loff_t destoff, loff_t len,
-		unsigned int remap_flags)
-{
-	struct inode *src_inode = file_inode(src_file);
-	struct inode *dst_inode = file_inode(dst_file);
-	bool same_inode = dst_inode == src_inode;
-	int ret;
-
-	if (remap_flags & ~(REMAP_FILE_DEDUP | REMAP_FILE_ADVISORY))
-		return -EINVAL;
-
-	if (same_inode)
-		inode_lock(src_inode);
-	else
-		lock_two_nondirectories(src_inode, dst_inode);
-
-	ret = btrfs_remap_file_range_prep(src_file, off, dst_file, destoff,
-					  &len, remap_flags);
-	if (ret < 0 || len == 0)
-		goto out_unlock;
-
-	if (remap_flags & REMAP_FILE_DEDUP)
-		ret = btrfs_extent_same(src_inode, off, len, dst_inode, destoff);
-	else
-		ret = btrfs_clone_files(dst_file, src_file, off, len, destoff);
-
-out_unlock:
-	if (same_inode)
-		inode_unlock(src_inode);
-	else
-		unlock_two_nondirectories(src_inode, dst_inode);
-
-	return ret < 0 ? ret : len;
-}
-
 static long btrfs_ioctl_default_subvol(struct file *file, void __user *argp)
 {
 	struct inode *inode = file_inode(file);
@@ -3955,8 +3772,7 @@ static long btrfs_ioctl_default_subvol(struct file *file, void __user *argp)
 	struct btrfs_root *new_root;
 	struct btrfs_dir_item *di;
 	struct btrfs_trans_handle *trans;
-	struct btrfs_path *path;
-	struct btrfs_key location;
+	struct btrfs_path *path = NULL;
 	struct btrfs_disk_key disk_key;
 	u64 objectid = 0;
 	u64 dir_id;
@@ -3977,53 +3793,50 @@ static long btrfs_ioctl_default_subvol(struct file *file, void __user *argp)
 	if (!objectid)
 		objectid = BTRFS_FS_TREE_OBJECTID;
 
-	location.objectid = objectid;
-	location.type = BTRFS_ROOT_ITEM_KEY;
-	location.offset = (u64)-1;
-
-	new_root = btrfs_read_fs_root_no_name(fs_info, &location);
+	new_root = btrfs_get_fs_root(fs_info, objectid, true);
 	if (IS_ERR(new_root)) {
 		ret = PTR_ERR(new_root);
 		goto out;
 	}
 	if (!is_fstree(new_root->root_key.objectid)) {
 		ret = -ENOENT;
-		goto out;
+		goto out_free;
 	}
 
 	path = btrfs_alloc_path();
 	if (!path) {
 		ret = -ENOMEM;
-		goto out;
+		goto out_free;
 	}
-	path->leave_spinning = 1;
 
 	trans = btrfs_start_transaction(root, 1);
 	if (IS_ERR(trans)) {
-		btrfs_free_path(path);
 		ret = PTR_ERR(trans);
-		goto out;
+		goto out_free;
 	}
 
 	dir_id = btrfs_super_root_dir(fs_info->super_copy);
 	di = btrfs_lookup_dir_item(trans, fs_info->tree_root, path,
 				   dir_id, "default", 7, 1);
 	if (IS_ERR_OR_NULL(di)) {
-		btrfs_free_path(path);
+		btrfs_release_path(path);
 		btrfs_end_transaction(trans);
 		btrfs_err(fs_info,
 			  "Umm, you don't have the default diritem, this isn't going to work");
 		ret = -ENOENT;
-		goto out;
+		goto out_free;
 	}
 
 	btrfs_cpu_key_to_disk(&disk_key, &new_root->root_key);
 	btrfs_set_dir_item_key(path->nodes[0], di, &disk_key);
 	btrfs_mark_buffer_dirty(path->nodes[0]);
-	btrfs_free_path(path);
+	btrfs_release_path(path);
 
 	btrfs_set_fs_incompat(fs_info, DEFAULT_SUBVOL);
 	btrfs_end_transaction(trans);
+out_free:
+	btrfs_put_root(new_root);
+	btrfs_free_path(path);
 out:
 	mnt_drop_write_file(file);
 	return ret;
@@ -4074,15 +3887,12 @@ static long btrfs_ioctl_space_info(struct btrfs_fs_info *fs_info,
 		struct btrfs_space_info *tmp;
 
 		info = NULL;
-		rcu_read_lock();
-		list_for_each_entry_rcu(tmp, &fs_info->space_info,
-					list) {
+		list_for_each_entry(tmp, &fs_info->space_info, list) {
 			if (tmp->flags == types[i]) {
 				info = tmp;
 				break;
 			}
 		}
-		rcu_read_unlock();
 
 		if (!info)
 			continue;
@@ -4130,15 +3940,12 @@ static long btrfs_ioctl_space_info(struct btrfs_fs_info *fs_info,
 			break;
 
 		info = NULL;
-		rcu_read_lock();
-		list_for_each_entry_rcu(tmp, &fs_info->space_info,
-					list) {
+		list_for_each_entry(tmp, &fs_info->space_info, list) {
 			if (tmp->flags == types[i]) {
 				info = tmp;
 				break;
 			}
 		}
-		rcu_read_unlock();
 
 		if (!info)
 			continue;
@@ -4192,7 +3999,6 @@ static noinline long btrfs_ioctl_start_sync(struct btrfs_root *root,
 {
 	struct btrfs_trans_handle *trans;
 	u64 transid;
-	int ret;
 
 	trans = btrfs_attach_transaction_barrier(root);
 	if (IS_ERR(trans)) {
@@ -4204,11 +4010,7 @@ static noinline long btrfs_ioctl_start_sync(struct btrfs_root *root,
 		goto out;
 	}
 	transid = trans->transid;
-	ret = btrfs_commit_transaction_async(trans, 0);
-	if (ret) {
-		btrfs_end_transaction(trans);
-		return ret;
-	}
+	btrfs_commit_transaction_async(trans);
 out:
 	if (argp)
 		if (copy_to_user(argp, &transid, sizeof(transid)))
@@ -4239,6 +4041,11 @@ static long btrfs_ioctl_scrub(struct file *file, void __user *arg)
 	if (!capable(CAP_SYS_ADMIN))
 		return -EPERM;
 
+	if (btrfs_fs_incompat(fs_info, EXTENT_TREE_V2)) {
+		btrfs_err(fs_info, "scrub is not supported on extent tree v2 yet");
+		return -EINVAL;
+	}
+
 	sa = memdup_user(arg, sizeof(*sa));
 	if (IS_ERR(sa))
 		return PTR_ERR(sa);
@@ -4338,6 +4145,11 @@ static long btrfs_ioctl_dev_replace(struct btrfs_fs_info *fs_info,
 	if (!capable(CAP_SYS_ADMIN))
 		return -EPERM;
 
+	if (btrfs_fs_incompat(fs_info, EXTENT_TREE_V2)) {
+		btrfs_err(fs_info, "device replace not supported on extent tree v2 yet");
+		return -EINVAL;
+	}
+
 	p = memdup_user(arg, sizeof(*p));
 	if (IS_ERR(p))
 		return PTR_ERR(p);
@@ -4348,11 +4160,11 @@ static long btrfs_ioctl_dev_replace(struct btrfs_fs_info *fs_info,
 			ret = -EROFS;
 			goto out;
 		}
-		if (test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags)) {
+		if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_DEV_REPLACE)) {
 			ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
 		} else {
 			ret = btrfs_dev_replace_by_ioctl(fs_info, p);
-			clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
+			btrfs_exclop_finish(fs_info);
 		}
 		break;
 	case BTRFS_IOCTL_DEV_REPLACE_CMD_STATUS:
@@ -4434,26 +4246,6 @@ out:
 	return ret;
 }
 
-static int build_ino_list(u64 inum, u64 offset, u64 root, void *ctx)
-{
-	struct btrfs_data_container *inodes = ctx;
-	const size_t c = 3 * sizeof(u64);
-
-	if (inodes->bytes_left >= c) {
-		inodes->bytes_left -= c;
-		inodes->val[inodes->elem_cnt] = inum;
-		inodes->val[inodes->elem_cnt + 1] = offset;
-		inodes->val[inodes->elem_cnt + 2] = root;
-		inodes->elem_cnt += 3;
-	} else {
-		inodes->bytes_missing += c - inodes->bytes_left;
-		inodes->bytes_left = 0;
-		inodes->elem_missed += 3;
-	}
-
-	return 0;
-}
-
 static long btrfs_ioctl_logical_to_ino(struct btrfs_fs_info *fs_info,
 					void __user *arg, int version)
 {
@@ -4503,7 +4295,7 @@ static long btrfs_ioctl_logical_to_ino(struct btrfs_fs_info *fs_info,
 	}
 
 	ret = iterate_inodes_from_logical(loi->logical, fs_info, path,
-					  build_ino_list, inodes, ignore_offset);
+					  inodes, ignore_offset);
 	if (ret == -EINVAL)
 		ret = -ENOENT;
 	if (ret < 0)
@@ -4546,13 +4338,79 @@ void btrfs_update_ioctl_balance_args(struct btrfs_fs_info *fs_info,
 	spin_unlock(&fs_info->balance_lock);
 }
 
+/**
+ * Try to acquire fs_info::balance_mutex as well as set BTRFS_EXLCOP_BALANCE as
+ * required.
+ *
+ * @fs_info:       the filesystem
+ * @excl_acquired: ptr to boolean value which is set to false in case balance
+ *                 is being resumed
+ *
+ * Return 0 on success in which case both fs_info::balance is acquired as well
+ * as exclusive ops are blocked. In case of failure return an error code.
+ */
+static int btrfs_try_lock_balance(struct btrfs_fs_info *fs_info, bool *excl_acquired)
+{
+	int ret;
+
+	/*
+	 * Exclusive operation is locked. Three possibilities:
+	 *   (1) some other op is running
+	 *   (2) balance is running
+	 *   (3) balance is paused -- special case (think resume)
+	 */
+	while (1) {
+		if (btrfs_exclop_start(fs_info, BTRFS_EXCLOP_BALANCE)) {
+			*excl_acquired = true;
+			mutex_lock(&fs_info->balance_mutex);
+			return 0;
+		}
+
+		mutex_lock(&fs_info->balance_mutex);
+		if (fs_info->balance_ctl) {
+			/* This is either (2) or (3) */
+			if (test_bit(BTRFS_FS_BALANCE_RUNNING, &fs_info->flags)) {
+				/* This is (2) */
+				ret = -EINPROGRESS;
+				goto out_failure;
+
+			} else {
+				mutex_unlock(&fs_info->balance_mutex);
+				/*
+				 * Lock released to allow other waiters to
+				 * continue, we'll reexamine the status again.
+				 */
+				mutex_lock(&fs_info->balance_mutex);
+
+				if (fs_info->balance_ctl &&
+				    !test_bit(BTRFS_FS_BALANCE_RUNNING, &fs_info->flags)) {
+					/* This is (3) */
+					*excl_acquired = false;
+					return 0;
+				}
+			}
+		} else {
+			/* This is (1) */
+			ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
+			goto out_failure;
+		}
+
+		mutex_unlock(&fs_info->balance_mutex);
+	}
+
+out_failure:
+	mutex_unlock(&fs_info->balance_mutex);
+	*excl_acquired = false;
+	return ret;
+}
+
 static long btrfs_ioctl_balance(struct file *file, void __user *arg)
 {
 	struct btrfs_root *root = BTRFS_I(file_inode(file))->root;
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct btrfs_ioctl_balance_args *bargs;
 	struct btrfs_balance_control *bctl;
-	bool need_unlock; /* for mut. excl. ops lock */
+	bool need_unlock = true;
 	int ret;
 
 	if (!capable(CAP_SYS_ADMIN))
@@ -4562,133 +4420,80 @@ static long btrfs_ioctl_balance(struct file *file, void __user *arg)
 	if (ret)
 		return ret;
 
-again:
-	if (!test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags)) {
-		mutex_lock(&fs_info->balance_mutex);
-		need_unlock = true;
-		goto locked;
+	bargs = memdup_user(arg, sizeof(*bargs));
+	if (IS_ERR(bargs)) {
+		ret = PTR_ERR(bargs);
+		bargs = NULL;
+		goto out;
 	}
 
-	/*
-	 * mut. excl. ops lock is locked.  Three possibilities:
-	 *   (1) some other op is running
-	 *   (2) balance is running
-	 *   (3) balance is paused -- special case (think resume)
-	 */
-	mutex_lock(&fs_info->balance_mutex);
-	if (fs_info->balance_ctl) {
-		/* this is either (2) or (3) */
-		if (!test_bit(BTRFS_FS_BALANCE_RUNNING, &fs_info->flags)) {
-			mutex_unlock(&fs_info->balance_mutex);
-			/*
-			 * Lock released to allow other waiters to continue,
-			 * we'll reexamine the status again.
-			 */
-			mutex_lock(&fs_info->balance_mutex);
-
-			if (fs_info->balance_ctl &&
-			    !test_bit(BTRFS_FS_BALANCE_RUNNING, &fs_info->flags)) {
-				/* this is (3) */
-				need_unlock = false;
-				goto locked;
-			}
-
-			mutex_unlock(&fs_info->balance_mutex);
-			goto again;
-		} else {
-			/* this is (2) */
-			mutex_unlock(&fs_info->balance_mutex);
-			ret = -EINPROGRESS;
-			goto out;
-		}
-	} else {
-		/* this is (1) */
-		mutex_unlock(&fs_info->balance_mutex);
-		ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
+	ret = btrfs_try_lock_balance(fs_info, &need_unlock);
+	if (ret)
 		goto out;
-	}
 
-locked:
-	BUG_ON(!test_bit(BTRFS_FS_EXCL_OP, &fs_info->flags));
+	lockdep_assert_held(&fs_info->balance_mutex);
 
-	if (arg) {
-		bargs = memdup_user(arg, sizeof(*bargs));
-		if (IS_ERR(bargs)) {
-			ret = PTR_ERR(bargs);
+	if (bargs->flags & BTRFS_BALANCE_RESUME) {
+		if (!fs_info->balance_ctl) {
+			ret = -ENOTCONN;
 			goto out_unlock;
 		}
 
-		if (bargs->flags & BTRFS_BALANCE_RESUME) {
-			if (!fs_info->balance_ctl) {
-				ret = -ENOTCONN;
-				goto out_bargs;
-			}
+		bctl = fs_info->balance_ctl;
+		spin_lock(&fs_info->balance_lock);
+		bctl->flags |= BTRFS_BALANCE_RESUME;
+		spin_unlock(&fs_info->balance_lock);
+		btrfs_exclop_balance(fs_info, BTRFS_EXCLOP_BALANCE);
 
-			bctl = fs_info->balance_ctl;
-			spin_lock(&fs_info->balance_lock);
-			bctl->flags |= BTRFS_BALANCE_RESUME;
-			spin_unlock(&fs_info->balance_lock);
+		goto do_balance;
+	}
 
-			goto do_balance;
-		}
-	} else {
-		bargs = NULL;
+	if (bargs->flags & ~(BTRFS_BALANCE_ARGS_MASK | BTRFS_BALANCE_TYPE_MASK)) {
+		ret = -EINVAL;
+		goto out_unlock;
 	}
 
 	if (fs_info->balance_ctl) {
 		ret = -EINPROGRESS;
-		goto out_bargs;
+		goto out_unlock;
 	}
 
 	bctl = kzalloc(sizeof(*bctl), GFP_KERNEL);
 	if (!bctl) {
 		ret = -ENOMEM;
-		goto out_bargs;
-	}
-
-	if (arg) {
-		memcpy(&bctl->data, &bargs->data, sizeof(bctl->data));
-		memcpy(&bctl->meta, &bargs->meta, sizeof(bctl->meta));
-		memcpy(&bctl->sys, &bargs->sys, sizeof(bctl->sys));
-
-		bctl->flags = bargs->flags;
-	} else {
-		/* balance everything - no filters */
-		bctl->flags |= BTRFS_BALANCE_TYPE_MASK;
+		goto out_unlock;
 	}
 
-	if (bctl->flags & ~(BTRFS_BALANCE_ARGS_MASK | BTRFS_BALANCE_TYPE_MASK)) {
-		ret = -EINVAL;
-		goto out_bctl;
-	}
+	memcpy(&bctl->data, &bargs->data, sizeof(bctl->data));
+	memcpy(&bctl->meta, &bargs->meta, sizeof(bctl->meta));
+	memcpy(&bctl->sys, &bargs->sys, sizeof(bctl->sys));
 
+	bctl->flags = bargs->flags;
 do_balance:
 	/*
-	 * Ownership of bctl and filesystem flag BTRFS_FS_EXCL_OP goes to
-	 * btrfs_balance.  bctl is freed in reset_balance_state, or, if
-	 * restriper was paused all the way until unmount, in free_fs_info.
-	 * The flag should be cleared after reset_balance_state.
+	 * Ownership of bctl and exclusive operation goes to btrfs_balance.
+	 * bctl is freed in reset_balance_state, or, if restriper was paused
+	 * all the way until unmount, in free_fs_info.  The flag should be
+	 * cleared after reset_balance_state.
 	 */
 	need_unlock = false;
 
 	ret = btrfs_balance(fs_info, bctl, bargs);
 	bctl = NULL;
 
-	if ((ret == 0 || ret == -ECANCELED) && arg) {
+	if (ret == 0 || ret == -ECANCELED) {
 		if (copy_to_user(arg, bargs, sizeof(*bargs)))
 			ret = -EFAULT;
 	}
 
-out_bctl:
 	kfree(bctl);
-out_bargs:
-	kfree(bargs);
 out_unlock:
 	mutex_unlock(&fs_info->balance_mutex);
 	if (need_unlock)
-		clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
+		btrfs_exclop_finish(fs_info);
 out:
 	mnt_drop_write_file(file);
+	kfree(bargs);
 	return ret;
 }
 
@@ -4966,26 +4771,20 @@ drop_write:
 static long btrfs_ioctl_quota_rescan_status(struct btrfs_fs_info *fs_info,
 						void __user *arg)
 {
-	struct btrfs_ioctl_quota_rescan_args *qsa;
-	int ret = 0;
+	struct btrfs_ioctl_quota_rescan_args qsa = {0};
 
 	if (!capable(CAP_SYS_ADMIN))
 		return -EPERM;
 
-	qsa = kzalloc(sizeof(*qsa), GFP_KERNEL);
-	if (!qsa)
-		return -ENOMEM;
-
 	if (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN) {
-		qsa->flags = 1;
-		qsa->progress = fs_info->qgroup_rescan_progress.objectid;
+		qsa.flags = 1;
+		qsa.progress = fs_info->qgroup_rescan_progress.objectid;
 	}
 
-	if (copy_to_user(arg, qsa, sizeof(*qsa)))
-		ret = -EFAULT;
+	if (copy_to_user(arg, &qsa, sizeof(qsa)))
+		return -EFAULT;
 
-	kfree(qsa);
-	return ret;
+	return 0;
 }
 
 static long btrfs_ioctl_quota_rescan_wait(struct btrfs_fs_info *fs_info,
@@ -4998,6 +4797,7 @@ static long btrfs_ioctl_quota_rescan_wait(struct btrfs_fs_info *fs_info,
 }
 
 static long _btrfs_ioctl_set_received_subvol(struct file *file,
+					    struct user_namespace *mnt_userns,
 					    struct btrfs_ioctl_received_subvol_args *sa)
 {
 	struct inode *inode = file_inode(file);
@@ -5009,7 +4809,7 @@ static long _btrfs_ioctl_set_received_subvol(struct file *file,
 	int ret = 0;
 	int received_uuid_changed;
 
-	if (!inode_owner_or_capable(inode))
+	if (!inode_owner_or_capable(mnt_userns, inode))
 		return -EPERM;
 
 	ret = mnt_want_write_file(file);
@@ -5114,7 +4914,7 @@ static long btrfs_ioctl_set_received_subvol_32(struct file *file,
 	args64->rtime.nsec = args32->rtime.nsec;
 	args64->flags = args32->flags;
 
-	ret = _btrfs_ioctl_set_received_subvol(file, args64);
+	ret = _btrfs_ioctl_set_received_subvol(file, file_mnt_user_ns(file), args64);
 	if (ret)
 		goto out;
 
@@ -5148,7 +4948,7 @@ static long btrfs_ioctl_set_received_subvol(struct file *file,
 	if (IS_ERR(sa))
 		return PTR_ERR(sa);
 
-	ret = _btrfs_ioctl_set_received_subvol(file, sa);
+	ret = _btrfs_ioctl_set_received_subvol(file, file_mnt_user_ns(file), sa);
 
 	if (ret)
 		goto out;
@@ -5400,7 +5200,7 @@ out_drop_write:
 	return ret;
 }
 
-static int _btrfs_ioctl_send(struct file *file, void __user *argp, bool compat)
+static int _btrfs_ioctl_send(struct inode *inode, void __user *argp, bool compat)
 {
 	struct btrfs_ioctl_send_args *arg;
 	int ret;
@@ -5430,11 +5230,194 @@ static int _btrfs_ioctl_send(struct file *file, void __user *argp, bool compat)
 		if (IS_ERR(arg))
 			return PTR_ERR(arg);
 	}
-	ret = btrfs_ioctl_send(file, arg);
+	ret = btrfs_ioctl_send(inode, arg);
 	kfree(arg);
 	return ret;
 }
 
+static int btrfs_ioctl_encoded_read(struct file *file, void __user *argp,
+				    bool compat)
+{
+	struct btrfs_ioctl_encoded_io_args args = { 0 };
+	size_t copy_end_kernel = offsetofend(struct btrfs_ioctl_encoded_io_args,
+					     flags);
+	size_t copy_end;
+	struct iovec iovstack[UIO_FASTIOV];
+	struct iovec *iov = iovstack;
+	struct iov_iter iter;
+	loff_t pos;
+	struct kiocb kiocb;
+	ssize_t ret;
+
+	if (!capable(CAP_SYS_ADMIN)) {
+		ret = -EPERM;
+		goto out_acct;
+	}
+
+	if (compat) {
+#if defined(CONFIG_64BIT) && defined(CONFIG_COMPAT)
+		struct btrfs_ioctl_encoded_io_args_32 args32;
+
+		copy_end = offsetofend(struct btrfs_ioctl_encoded_io_args_32,
+				       flags);
+		if (copy_from_user(&args32, argp, copy_end)) {
+			ret = -EFAULT;
+			goto out_acct;
+		}
+		args.iov = compat_ptr(args32.iov);
+		args.iovcnt = args32.iovcnt;
+		args.offset = args32.offset;
+		args.flags = args32.flags;
+#else
+		return -ENOTTY;
+#endif
+	} else {
+		copy_end = copy_end_kernel;
+		if (copy_from_user(&args, argp, copy_end)) {
+			ret = -EFAULT;
+			goto out_acct;
+		}
+	}
+	if (args.flags != 0) {
+		ret = -EINVAL;
+		goto out_acct;
+	}
+
+	ret = import_iovec(READ, args.iov, args.iovcnt, ARRAY_SIZE(iovstack),
+			   &iov, &iter);
+	if (ret < 0)
+		goto out_acct;
+
+	if (iov_iter_count(&iter) == 0) {
+		ret = 0;
+		goto out_iov;
+	}
+	pos = args.offset;
+	ret = rw_verify_area(READ, file, &pos, args.len);
+	if (ret < 0)
+		goto out_iov;
+
+	init_sync_kiocb(&kiocb, file);
+	kiocb.ki_pos = pos;
+
+	ret = btrfs_encoded_read(&kiocb, &iter, &args);
+	if (ret >= 0) {
+		fsnotify_access(file);
+		if (copy_to_user(argp + copy_end,
+				 (char *)&args + copy_end_kernel,
+				 sizeof(args) - copy_end_kernel))
+			ret = -EFAULT;
+	}
+
+out_iov:
+	kfree(iov);
+out_acct:
+	if (ret > 0)
+		add_rchar(current, ret);
+	inc_syscr(current);
+	return ret;
+}
+
+static int btrfs_ioctl_encoded_write(struct file *file, void __user *argp, bool compat)
+{
+	struct btrfs_ioctl_encoded_io_args args;
+	struct iovec iovstack[UIO_FASTIOV];
+	struct iovec *iov = iovstack;
+	struct iov_iter iter;
+	loff_t pos;
+	struct kiocb kiocb;
+	ssize_t ret;
+
+	if (!capable(CAP_SYS_ADMIN)) {
+		ret = -EPERM;
+		goto out_acct;
+	}
+
+	if (!(file->f_mode & FMODE_WRITE)) {
+		ret = -EBADF;
+		goto out_acct;
+	}
+
+	if (compat) {
+#if defined(CONFIG_64BIT) && defined(CONFIG_COMPAT)
+		struct btrfs_ioctl_encoded_io_args_32 args32;
+
+		if (copy_from_user(&args32, argp, sizeof(args32))) {
+			ret = -EFAULT;
+			goto out_acct;
+		}
+		args.iov = compat_ptr(args32.iov);
+		args.iovcnt = args32.iovcnt;
+		args.offset = args32.offset;
+		args.flags = args32.flags;
+		args.len = args32.len;
+		args.unencoded_len = args32.unencoded_len;
+		args.unencoded_offset = args32.unencoded_offset;
+		args.compression = args32.compression;
+		args.encryption = args32.encryption;
+		memcpy(args.reserved, args32.reserved, sizeof(args.reserved));
+#else
+		return -ENOTTY;
+#endif
+	} else {
+		if (copy_from_user(&args, argp, sizeof(args))) {
+			ret = -EFAULT;
+			goto out_acct;
+		}
+	}
+
+	ret = -EINVAL;
+	if (args.flags != 0)
+		goto out_acct;
+	if (memchr_inv(args.reserved, 0, sizeof(args.reserved)))
+		goto out_acct;
+	if (args.compression == BTRFS_ENCODED_IO_COMPRESSION_NONE &&
+	    args.encryption == BTRFS_ENCODED_IO_ENCRYPTION_NONE)
+		goto out_acct;
+	if (args.compression >= BTRFS_ENCODED_IO_COMPRESSION_TYPES ||
+	    args.encryption >= BTRFS_ENCODED_IO_ENCRYPTION_TYPES)
+		goto out_acct;
+	if (args.unencoded_offset > args.unencoded_len)
+		goto out_acct;
+	if (args.len > args.unencoded_len - args.unencoded_offset)
+		goto out_acct;
+
+	ret = import_iovec(WRITE, args.iov, args.iovcnt, ARRAY_SIZE(iovstack),
+			   &iov, &iter);
+	if (ret < 0)
+		goto out_acct;
+
+	file_start_write(file);
+
+	if (iov_iter_count(&iter) == 0) {
+		ret = 0;
+		goto out_end_write;
+	}
+	pos = args.offset;
+	ret = rw_verify_area(WRITE, file, &pos, args.len);
+	if (ret < 0)
+		goto out_end_write;
+
+	init_sync_kiocb(&kiocb, file);
+	ret = kiocb_set_rw_flags(&kiocb, 0);
+	if (ret)
+		goto out_end_write;
+	kiocb.ki_pos = pos;
+
+	ret = btrfs_do_write_iter(&kiocb, &iter, &args);
+	if (ret > 0)
+		fsnotify_modify(file);
+
+out_end_write:
+	file_end_write(file);
+	kfree(iov);
+out_acct:
+	if (ret > 0)
+		add_wchar(current, ret);
+	inc_syscw(current);
+	return ret;
+}
+
 long btrfs_ioctl(struct file *file, unsigned int
 		cmd, unsigned long arg)
 {
@@ -5444,12 +5427,8 @@ long btrfs_ioctl(struct file *file, unsigned int
 	void __user *argp = (void __user *)arg;
 
 	switch (cmd) {
-	case FS_IOC_GETFLAGS:
-		return btrfs_ioctl_getflags(file, argp);
-	case FS_IOC_SETFLAGS:
-		return btrfs_ioctl_setflags(file, argp);
 	case FS_IOC_GETVERSION:
-		return btrfs_ioctl_getversion(file, argp);
+		return btrfs_ioctl_getversion(inode, argp);
 	case FS_IOC_GETFSLABEL:
 		return btrfs_ioctl_get_fslabel(fs_info, argp);
 	case FS_IOC_SETFSLABEL:
@@ -5465,9 +5444,11 @@ long btrfs_ioctl(struct file *file, unsigned int
 	case BTRFS_IOC_SUBVOL_CREATE_V2:
 		return btrfs_ioctl_snap_create_v2(file, argp, 1);
 	case BTRFS_IOC_SNAP_DESTROY:
-		return btrfs_ioctl_snap_destroy(file, argp);
+		return btrfs_ioctl_snap_destroy(file, argp, false);
+	case BTRFS_IOC_SNAP_DESTROY_V2:
+		return btrfs_ioctl_snap_destroy(file, argp, true);
 	case BTRFS_IOC_SUBVOL_GETFLAGS:
-		return btrfs_ioctl_subvol_getflags(file, argp);
+		return btrfs_ioctl_subvol_getflags(inode, argp);
 	case BTRFS_IOC_SUBVOL_SETFLAGS:
 		return btrfs_ioctl_subvol_setflags(file, argp);
 	case BTRFS_IOC_DEFAULT_SUBVOL:
@@ -5488,14 +5469,12 @@ long btrfs_ioctl(struct file *file, unsigned int
 		return btrfs_ioctl_fs_info(fs_info, argp);
 	case BTRFS_IOC_DEV_INFO:
 		return btrfs_ioctl_dev_info(fs_info, argp);
-	case BTRFS_IOC_BALANCE:
-		return btrfs_ioctl_balance(file, NULL);
 	case BTRFS_IOC_TREE_SEARCH:
-		return btrfs_ioctl_tree_search(file, argp);
+		return btrfs_ioctl_tree_search(inode, argp);
 	case BTRFS_IOC_TREE_SEARCH_V2:
-		return btrfs_ioctl_tree_search_v2(file, argp);
+		return btrfs_ioctl_tree_search_v2(inode, argp);
 	case BTRFS_IOC_INO_LOOKUP:
-		return btrfs_ioctl_ino_lookup(file, argp);
+		return btrfs_ioctl_ino_lookup(root, argp);
 	case BTRFS_IOC_INO_PATHS:
 		return btrfs_ioctl_ino_to_path(root, argp);
 	case BTRFS_IOC_LOGICAL_INO:
@@ -5507,7 +5486,7 @@ long btrfs_ioctl(struct file *file, unsigned int
 	case BTRFS_IOC_SYNC: {
 		int ret;
 
-		ret = btrfs_start_delalloc_roots(fs_info, -1);
+		ret = btrfs_start_delalloc_roots(fs_info, LONG_MAX, false);
 		if (ret)
 			return ret;
 		ret = btrfs_sync_fs(inode->i_sb, 1);
@@ -5542,10 +5521,10 @@ long btrfs_ioctl(struct file *file, unsigned int
 		return btrfs_ioctl_set_received_subvol_32(file, argp);
 #endif
 	case BTRFS_IOC_SEND:
-		return _btrfs_ioctl_send(file, argp, false);
+		return _btrfs_ioctl_send(inode, argp, false);
 #if defined(CONFIG_64BIT) && defined(CONFIG_COMPAT)
 	case BTRFS_IOC_SEND_32:
-		return _btrfs_ioctl_send(file, argp, true);
+		return _btrfs_ioctl_send(inode, argp, true);
 #endif
 	case BTRFS_IOC_GET_DEV_STATS:
 		return btrfs_ioctl_get_dev_stats(fs_info, argp);
@@ -5571,16 +5550,26 @@ long btrfs_ioctl(struct file *file, unsigned int
 		return btrfs_ioctl_get_features(fs_info, argp);
 	case BTRFS_IOC_SET_FEATURES:
 		return btrfs_ioctl_set_features(file, argp);
-	case FS_IOC_FSGETXATTR:
-		return btrfs_ioctl_fsgetxattr(file, argp);
-	case FS_IOC_FSSETXATTR:
-		return btrfs_ioctl_fssetxattr(file, argp);
 	case BTRFS_IOC_GET_SUBVOL_INFO:
-		return btrfs_ioctl_get_subvol_info(file, argp);
+		return btrfs_ioctl_get_subvol_info(inode, argp);
 	case BTRFS_IOC_GET_SUBVOL_ROOTREF:
-		return btrfs_ioctl_get_subvol_rootref(file, argp);
+		return btrfs_ioctl_get_subvol_rootref(root, argp);
 	case BTRFS_IOC_INO_LOOKUP_USER:
 		return btrfs_ioctl_ino_lookup_user(file, argp);
+	case FS_IOC_ENABLE_VERITY:
+		return fsverity_ioctl_enable(file, (const void __user *)argp);
+	case FS_IOC_MEASURE_VERITY:
+		return fsverity_ioctl_measure(file, argp);
+	case BTRFS_IOC_ENCODED_READ:
+		return btrfs_ioctl_encoded_read(file, argp, false);
+	case BTRFS_IOC_ENCODED_WRITE:
+		return btrfs_ioctl_encoded_write(file, argp, false);
+#if defined(CONFIG_64BIT) && defined(CONFIG_COMPAT)
+	case BTRFS_IOC_ENCODED_READ_32:
+		return btrfs_ioctl_encoded_read(file, argp, true);
+	case BTRFS_IOC_ENCODED_WRITE_32:
+		return btrfs_ioctl_encoded_write(file, argp, true);
+#endif
 	}
 
 	return -ENOTTY;
@@ -5594,12 +5583,6 @@ long btrfs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
 	 * handling is necessary.
 	 */
 	switch (cmd) {
-	case FS_IOC32_GETFLAGS:
-		cmd = FS_IOC_GETFLAGS;
-		break;
-	case FS_IOC32_SETFLAGS:
-		cmd = FS_IOC_SETFLAGS;
-		break;
 	case FS_IOC32_GETVERSION:
 		cmd = FS_IOC_GETVERSION;
 		break;
diff --git a/fs/btrfs/locking.c b/fs/btrfs/locking.c
index 571c4826c428..0eab3cb274a1 100644
--- a/fs/btrfs/locking.c
+++ b/fs/btrfs/locking.c
@@ -14,488 +14,203 @@
 #include "locking.h"
 
 /*
- * Extent buffer locking
- * =====================
- *
- * The locks use a custom scheme that allows to do more operations than are
- * available fromt current locking primitives. The building blocks are still
- * rwlock and wait queues.
- *
- * Required semantics:
- *
- * - reader/writer exclusion
- * - writer/writer exclusion
- * - reader/reader sharing
- * - spinning lock semantics
- * - blocking lock semantics
- * - try-lock semantics for readers and writers
- * - one level nesting, allowing read lock to be taken by the same thread that
- *   already has write lock
- *
- * The extent buffer locks (also called tree locks) manage access to eb data
- * related to the storage in the b-tree (keys, items, but not the individual
- * members of eb).
- * We want concurrency of many readers and safe updates. The underlying locking
- * is done by read-write spinlock and the blocking part is implemented using
- * counters and wait queues.
- *
- * spinning semantics - the low-level rwlock is held so all other threads that
- *                      want to take it are spinning on it.
- *
- * blocking semantics - the low-level rwlock is not held but the counter
- *                      denotes how many times the blocking lock was held;
- *                      sleeping is possible
- *
- * Write lock always allows only one thread to access the data.
- *
- *
- * Debugging
- * ---------
- *
- * There are additional state counters that are asserted in various contexts,
- * removed from non-debug build to reduce extent_buffer size and for
- * performance reasons.
- *
- *
- * Lock nesting
- * ------------
- *
- * A write operation on a tree might indirectly start a look up on the same
- * tree.  This can happen when btrfs_cow_block locks the tree and needs to
- * lookup free extents.
- *
- * btrfs_cow_block
- *   ..
- *   alloc_tree_block_no_bg_flush
- *     btrfs_alloc_tree_block
- *       btrfs_reserve_extent
- *         ..
- *         load_free_space_cache
- *           ..
- *           btrfs_lookup_file_extent
- *             btrfs_search_slot
- *
- *
- * Locking pattern - spinning
- * --------------------------
- *
- * The simple locking scenario, the +--+ denotes the spinning section.
- *
- * +- btrfs_tree_lock
- * | - extent_buffer::rwlock is held
- * | - no heavy operations should happen, eg. IO, memory allocations, large
- * |   structure traversals
- * +- btrfs_tree_unock
-*
-*
- * Locking pattern - blocking
- * --------------------------
- *
- * The blocking write uses the following scheme.  The +--+ denotes the spinning
- * section.
- *
- * +- btrfs_tree_lock
- * |
- * +- btrfs_set_lock_blocking_write
- *
- *   - allowed: IO, memory allocations, etc.
- *
- * -- btrfs_tree_unlock - note, no explicit unblocking necessary
- *
- *
- * Blocking read is similar.
- *
- * +- btrfs_tree_read_lock
- * |
- * +- btrfs_set_lock_blocking_read
- *
- *  - heavy operations allowed
- *
- * +- btrfs_tree_read_unlock_blocking
- * |
- * +- btrfs_tree_read_unlock
- *
+ * Lockdep class keys for extent_buffer->lock's in this root.  For a given
+ * eb, the lockdep key is determined by the btrfs_root it belongs to and
+ * the level the eb occupies in the tree.
+ *
+ * Different roots are used for different purposes and may nest inside each
+ * other and they require separate keysets.  As lockdep keys should be
+ * static, assign keysets according to the purpose of the root as indicated
+ * by btrfs_root->root_key.objectid.  This ensures that all special purpose
+ * roots have separate keysets.
+ *
+ * Lock-nesting across peer nodes is always done with the immediate parent
+ * node locked thus preventing deadlock.  As lockdep doesn't know this, use
+ * subclass to avoid triggering lockdep warning in such cases.
+ *
+ * The key is set by the readpage_end_io_hook after the buffer has passed
+ * csum validation but before the pages are unlocked.  It is also set by
+ * btrfs_init_new_buffer on freshly allocated blocks.
+ *
+ * We also add a check to make sure the highest level of the tree is the
+ * same as our lockdep setup here.  If BTRFS_MAX_LEVEL changes, this code
+ * needs update as well.
  */
+#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#if BTRFS_MAX_LEVEL != 8
+#error
+#endif
 
-#ifdef CONFIG_BTRFS_DEBUG
-static inline void btrfs_assert_spinning_writers_get(struct extent_buffer *eb)
-{
-	WARN_ON(eb->spinning_writers);
-	eb->spinning_writers++;
-}
-
-static inline void btrfs_assert_spinning_writers_put(struct extent_buffer *eb)
-{
-	WARN_ON(eb->spinning_writers != 1);
-	eb->spinning_writers--;
-}
-
-static inline void btrfs_assert_no_spinning_writers(struct extent_buffer *eb)
-{
-	WARN_ON(eb->spinning_writers);
-}
-
-static inline void btrfs_assert_spinning_readers_get(struct extent_buffer *eb)
-{
-	atomic_inc(&eb->spinning_readers);
-}
-
-static inline void btrfs_assert_spinning_readers_put(struct extent_buffer *eb)
-{
-	WARN_ON(atomic_read(&eb->spinning_readers) == 0);
-	atomic_dec(&eb->spinning_readers);
-}
-
-static inline void btrfs_assert_tree_read_locks_get(struct extent_buffer *eb)
+#define DEFINE_LEVEL(stem, level)					\
+	.names[level] = "btrfs-" stem "-0" #level,
+
+#define DEFINE_NAME(stem)						\
+	DEFINE_LEVEL(stem, 0)						\
+	DEFINE_LEVEL(stem, 1)						\
+	DEFINE_LEVEL(stem, 2)						\
+	DEFINE_LEVEL(stem, 3)						\
+	DEFINE_LEVEL(stem, 4)						\
+	DEFINE_LEVEL(stem, 5)						\
+	DEFINE_LEVEL(stem, 6)						\
+	DEFINE_LEVEL(stem, 7)
+
+static struct btrfs_lockdep_keyset {
+	u64			id;		/* root objectid */
+	/* Longest entry: btrfs-free-space-00 */
+	char			names[BTRFS_MAX_LEVEL][20];
+	struct lock_class_key	keys[BTRFS_MAX_LEVEL];
+} btrfs_lockdep_keysets[] = {
+	{ .id = BTRFS_ROOT_TREE_OBJECTID,	DEFINE_NAME("root")	},
+	{ .id = BTRFS_EXTENT_TREE_OBJECTID,	DEFINE_NAME("extent")	},
+	{ .id = BTRFS_CHUNK_TREE_OBJECTID,	DEFINE_NAME("chunk")	},
+	{ .id = BTRFS_DEV_TREE_OBJECTID,	DEFINE_NAME("dev")	},
+	{ .id = BTRFS_CSUM_TREE_OBJECTID,	DEFINE_NAME("csum")	},
+	{ .id = BTRFS_QUOTA_TREE_OBJECTID,	DEFINE_NAME("quota")	},
+	{ .id = BTRFS_TREE_LOG_OBJECTID,	DEFINE_NAME("log")	},
+	{ .id = BTRFS_TREE_RELOC_OBJECTID,	DEFINE_NAME("treloc")	},
+	{ .id = BTRFS_DATA_RELOC_TREE_OBJECTID,	DEFINE_NAME("dreloc")	},
+	{ .id = BTRFS_UUID_TREE_OBJECTID,	DEFINE_NAME("uuid")	},
+	{ .id = BTRFS_FREE_SPACE_TREE_OBJECTID,	DEFINE_NAME("free-space") },
+	{ .id = 0,				DEFINE_NAME("tree")	},
+};
+
+#undef DEFINE_LEVEL
+#undef DEFINE_NAME
+
+void btrfs_set_buffer_lockdep_class(u64 objectid, struct extent_buffer *eb, int level)
 {
-	atomic_inc(&eb->read_locks);
-}
+	struct btrfs_lockdep_keyset *ks;
 
-static inline void btrfs_assert_tree_read_locks_put(struct extent_buffer *eb)
-{
-	atomic_dec(&eb->read_locks);
-}
+	BUG_ON(level >= ARRAY_SIZE(ks->keys));
 
-static inline void btrfs_assert_tree_read_locked(struct extent_buffer *eb)
-{
-	BUG_ON(!atomic_read(&eb->read_locks));
-}
+	/* Find the matching keyset, id 0 is the default entry */
+	for (ks = btrfs_lockdep_keysets; ks->id; ks++)
+		if (ks->id == objectid)
+			break;
 
-static inline void btrfs_assert_tree_write_locks_get(struct extent_buffer *eb)
-{
-	eb->write_locks++;
+	lockdep_set_class_and_name(&eb->lock, &ks->keys[level], ks->names[level]);
 }
 
-static inline void btrfs_assert_tree_write_locks_put(struct extent_buffer *eb)
+void btrfs_maybe_reset_lockdep_class(struct btrfs_root *root, struct extent_buffer *eb)
 {
-	eb->write_locks--;
+	if (test_bit(BTRFS_ROOT_RESET_LOCKDEP_CLASS, &root->state))
+		btrfs_set_buffer_lockdep_class(root->root_key.objectid,
+					       eb, btrfs_header_level(eb));
 }
 
-#else
-static void btrfs_assert_spinning_writers_get(struct extent_buffer *eb) { }
-static void btrfs_assert_spinning_writers_put(struct extent_buffer *eb) { }
-static void btrfs_assert_no_spinning_writers(struct extent_buffer *eb) { }
-static void btrfs_assert_spinning_readers_put(struct extent_buffer *eb) { }
-static void btrfs_assert_spinning_readers_get(struct extent_buffer *eb) { }
-static void btrfs_assert_tree_read_locked(struct extent_buffer *eb) { }
-static void btrfs_assert_tree_read_locks_get(struct extent_buffer *eb) { }
-static void btrfs_assert_tree_read_locks_put(struct extent_buffer *eb) { }
-static void btrfs_assert_tree_write_locks_get(struct extent_buffer *eb) { }
-static void btrfs_assert_tree_write_locks_put(struct extent_buffer *eb) { }
 #endif
 
 /*
- * Mark already held read lock as blocking. Can be nested in write lock by the
- * same thread.
+ * Extent buffer locking
+ * =====================
  *
- * Use when there are potentially long operations ahead so other thread waiting
- * on the lock will not actively spin but sleep instead.
+ * We use a rw_semaphore for tree locking, and the semantics are exactly the
+ * same:
  *
- * The rwlock is released and blocking reader counter is increased.
- */
-void btrfs_set_lock_blocking_read(struct extent_buffer *eb)
-{
-	trace_btrfs_set_lock_blocking_read(eb);
-	/*
-	 * No lock is required.  The lock owner may change if we have a read
-	 * lock, but it won't change to or away from us.  If we have the write
-	 * lock, we are the owner and it'll never change.
-	 */
-	if (eb->lock_nested && current->pid == eb->lock_owner)
-		return;
-	btrfs_assert_tree_read_locked(eb);
-	atomic_inc(&eb->blocking_readers);
-	btrfs_assert_spinning_readers_put(eb);
-	read_unlock(&eb->lock);
-}
-
-/*
- * Mark already held write lock as blocking.
- *
- * Use when there are potentially long operations ahead so other threads
- * waiting on the lock will not actively spin but sleep instead.
+ * - reader/writer exclusion
+ * - writer/writer exclusion
+ * - reader/reader sharing
+ * - try-lock semantics for readers and writers
  *
- * The rwlock is released and blocking writers is set.
+ * The rwsem implementation does opportunistic spinning which reduces number of
+ * times the locking task needs to sleep.
  */
-void btrfs_set_lock_blocking_write(struct extent_buffer *eb)
-{
-	trace_btrfs_set_lock_blocking_write(eb);
-	/*
-	 * No lock is required.  The lock owner may change if we have a read
-	 * lock, but it won't change to or away from us.  If we have the write
-	 * lock, we are the owner and it'll never change.
-	 */
-	if (eb->lock_nested && current->pid == eb->lock_owner)
-		return;
-	if (eb->blocking_writers == 0) {
-		btrfs_assert_spinning_writers_put(eb);
-		btrfs_assert_tree_locked(eb);
-		WRITE_ONCE(eb->blocking_writers, 1);
-		write_unlock(&eb->lock);
-	}
-}
 
 /*
- * Lock the extent buffer for read. Wait for any writers (spinning or blocking).
- * Can be nested in write lock by the same thread.
+ * __btrfs_tree_read_lock - lock extent buffer for read
+ * @eb:		the eb to be locked
+ * @nest:	the nesting level to be used for lockdep
  *
- * Use when the locked section does only lightweight actions and busy waiting
- * would be cheaper than making other threads do the wait/wake loop.
- *
- * The rwlock is held upon exit.
+ * This takes the read lock on the extent buffer, using the specified nesting
+ * level for lockdep purposes.
  */
-void btrfs_tree_read_lock(struct extent_buffer *eb)
+void __btrfs_tree_read_lock(struct extent_buffer *eb, enum btrfs_lock_nesting nest)
 {
 	u64 start_ns = 0;
 
 	if (trace_btrfs_tree_read_lock_enabled())
 		start_ns = ktime_get_ns();
-again:
-	read_lock(&eb->lock);
-	BUG_ON(eb->blocking_writers == 0 &&
-	       current->pid == eb->lock_owner);
-	if (eb->blocking_writers) {
-		if (current->pid == eb->lock_owner) {
-			/*
-			 * This extent is already write-locked by our thread.
-			 * We allow an additional read lock to be added because
-			 * it's for the same thread. btrfs_find_all_roots()
-			 * depends on this as it may be called on a partly
-			 * (write-)locked tree.
-			 */
-			BUG_ON(eb->lock_nested);
-			eb->lock_nested = true;
-			read_unlock(&eb->lock);
-			trace_btrfs_tree_read_lock(eb, start_ns);
-			return;
-		}
-		read_unlock(&eb->lock);
-		wait_event(eb->write_lock_wq,
-			   READ_ONCE(eb->blocking_writers) == 0);
-		goto again;
-	}
-	btrfs_assert_tree_read_locks_get(eb);
-	btrfs_assert_spinning_readers_get(eb);
+
+	down_read_nested(&eb->lock, nest);
 	trace_btrfs_tree_read_lock(eb, start_ns);
 }
 
-/*
- * Lock extent buffer for read, optimistically expecting that there are no
- * contending blocking writers. If there are, don't wait.
- *
- * Return 1 if the rwlock has been taken, 0 otherwise
- */
-int btrfs_tree_read_lock_atomic(struct extent_buffer *eb)
+void btrfs_tree_read_lock(struct extent_buffer *eb)
 {
-	if (READ_ONCE(eb->blocking_writers))
-		return 0;
-
-	read_lock(&eb->lock);
-	/* Refetch value after lock */
-	if (READ_ONCE(eb->blocking_writers)) {
-		read_unlock(&eb->lock);
-		return 0;
-	}
-	btrfs_assert_tree_read_locks_get(eb);
-	btrfs_assert_spinning_readers_get(eb);
-	trace_btrfs_tree_read_lock_atomic(eb);
-	return 1;
+	__btrfs_tree_read_lock(eb, BTRFS_NESTING_NORMAL);
 }
 
 /*
- * Try-lock for read. Don't block or wait for contending writers.
+ * Try-lock for read.
  *
- * Retrun 1 if the rwlock has been taken, 0 otherwise
+ * Return 1 if the rwlock has been taken, 0 otherwise
  */
 int btrfs_try_tree_read_lock(struct extent_buffer *eb)
 {
-	if (READ_ONCE(eb->blocking_writers))
-		return 0;
-
-	if (!read_trylock(&eb->lock))
-		return 0;
-
-	/* Refetch value after lock */
-	if (READ_ONCE(eb->blocking_writers)) {
-		read_unlock(&eb->lock);
-		return 0;
+	if (down_read_trylock(&eb->lock)) {
+		trace_btrfs_try_tree_read_lock(eb);
+		return 1;
 	}
-	btrfs_assert_tree_read_locks_get(eb);
-	btrfs_assert_spinning_readers_get(eb);
-	trace_btrfs_try_tree_read_lock(eb);
-	return 1;
+	return 0;
 }
 
 /*
- * Try-lock for write. May block until the lock is uncontended, but does not
- * wait until it is free.
+ * Try-lock for write.
  *
- * Retrun 1 if the rwlock has been taken, 0 otherwise
+ * Return 1 if the rwlock has been taken, 0 otherwise
  */
 int btrfs_try_tree_write_lock(struct extent_buffer *eb)
 {
-	if (READ_ONCE(eb->blocking_writers) || atomic_read(&eb->blocking_readers))
-		return 0;
-
-	write_lock(&eb->lock);
-	/* Refetch value after lock */
-	if (READ_ONCE(eb->blocking_writers) || atomic_read(&eb->blocking_readers)) {
-		write_unlock(&eb->lock);
-		return 0;
+	if (down_write_trylock(&eb->lock)) {
+		eb->lock_owner = current->pid;
+		trace_btrfs_try_tree_write_lock(eb);
+		return 1;
 	}
-	btrfs_assert_tree_write_locks_get(eb);
-	btrfs_assert_spinning_writers_get(eb);
-	eb->lock_owner = current->pid;
-	trace_btrfs_try_tree_write_lock(eb);
-	return 1;
+	return 0;
 }
 
 /*
- * Release read lock. Must be used only if the lock is in spinning mode.  If
- * the read lock is nested, must pair with read lock before the write unlock.
- *
- * The rwlock is not held upon exit.
+ * Release read lock.
  */
 void btrfs_tree_read_unlock(struct extent_buffer *eb)
 {
 	trace_btrfs_tree_read_unlock(eb);
-	/*
-	 * if we're nested, we have the write lock.  No new locking
-	 * is needed as long as we are the lock owner.
-	 * The write unlock will do a barrier for us, and the lock_nested
-	 * field only matters to the lock owner.
-	 */
-	if (eb->lock_nested && current->pid == eb->lock_owner) {
-		eb->lock_nested = false;
-		return;
-	}
-	btrfs_assert_tree_read_locked(eb);
-	btrfs_assert_spinning_readers_put(eb);
-	btrfs_assert_tree_read_locks_put(eb);
-	read_unlock(&eb->lock);
+	up_read(&eb->lock);
 }
 
 /*
- * Release read lock, previously set to blocking by a pairing call to
- * btrfs_set_lock_blocking_read(). Can be nested in write lock by the same
- * thread.
+ * __btrfs_tree_lock - lock eb for write
+ * @eb:		the eb to lock
+ * @nest:	the nesting to use for the lock
  *
- * State of rwlock is unchanged, last reader wakes waiting threads.
+ * Returns with the eb->lock write locked.
  */
-void btrfs_tree_read_unlock_blocking(struct extent_buffer *eb)
-{
-	trace_btrfs_tree_read_unlock_blocking(eb);
-	/*
-	 * if we're nested, we have the write lock.  No new locking
-	 * is needed as long as we are the lock owner.
-	 * The write unlock will do a barrier for us, and the lock_nested
-	 * field only matters to the lock owner.
-	 */
-	if (eb->lock_nested && current->pid == eb->lock_owner) {
-		eb->lock_nested = false;
-		return;
-	}
-	btrfs_assert_tree_read_locked(eb);
-	WARN_ON(atomic_read(&eb->blocking_readers) == 0);
-	/* atomic_dec_and_test implies a barrier */
-	if (atomic_dec_and_test(&eb->blocking_readers))
-		cond_wake_up_nomb(&eb->read_lock_wq);
-	btrfs_assert_tree_read_locks_put(eb);
-}
-
-/*
- * Lock for write. Wait for all blocking and spinning readers and writers. This
- * starts context where reader lock could be nested by the same thread.
- *
- * The rwlock is held for write upon exit.
- */
-void btrfs_tree_lock(struct extent_buffer *eb)
+void __btrfs_tree_lock(struct extent_buffer *eb, enum btrfs_lock_nesting nest)
+	__acquires(&eb->lock)
 {
 	u64 start_ns = 0;
 
 	if (trace_btrfs_tree_lock_enabled())
 		start_ns = ktime_get_ns();
 
-	WARN_ON(eb->lock_owner == current->pid);
-again:
-	wait_event(eb->read_lock_wq, atomic_read(&eb->blocking_readers) == 0);
-	wait_event(eb->write_lock_wq, READ_ONCE(eb->blocking_writers) == 0);
-	write_lock(&eb->lock);
-	/* Refetch value after lock */
-	if (atomic_read(&eb->blocking_readers) ||
-	    READ_ONCE(eb->blocking_writers)) {
-		write_unlock(&eb->lock);
-		goto again;
-	}
-	btrfs_assert_spinning_writers_get(eb);
-	btrfs_assert_tree_write_locks_get(eb);
+	down_write_nested(&eb->lock, nest);
 	eb->lock_owner = current->pid;
 	trace_btrfs_tree_lock(eb, start_ns);
 }
 
-/*
- * Release the write lock, either blocking or spinning (ie. there's no need
- * for an explicit blocking unlock, like btrfs_tree_read_unlock_blocking).
- * This also ends the context for nesting, the read lock must have been
- * released already.
- *
- * Tasks blocked and waiting are woken, rwlock is not held upon exit.
- */
-void btrfs_tree_unlock(struct extent_buffer *eb)
+void btrfs_tree_lock(struct extent_buffer *eb)
 {
-	/*
-	 * This is read both locked and unlocked but always by the same thread
-	 * that already owns the lock so we don't need to use READ_ONCE
-	 */
-	int blockers = eb->blocking_writers;
-
-	BUG_ON(blockers > 1);
-
-	btrfs_assert_tree_locked(eb);
-	trace_btrfs_tree_unlock(eb);
-	eb->lock_owner = 0;
-	btrfs_assert_tree_write_locks_put(eb);
-
-	if (blockers) {
-		btrfs_assert_no_spinning_writers(eb);
-		/* Unlocked write */
-		WRITE_ONCE(eb->blocking_writers, 0);
-		/*
-		 * We need to order modifying blocking_writers above with
-		 * actually waking up the sleepers to ensure they see the
-		 * updated value of blocking_writers
-		 */
-		cond_wake_up(&eb->write_lock_wq);
-	} else {
-		btrfs_assert_spinning_writers_put(eb);
-		write_unlock(&eb->lock);
-	}
+	__btrfs_tree_lock(eb, BTRFS_NESTING_NORMAL);
 }
 
 /*
- * Set all locked nodes in the path to blocking locks.  This should be done
- * before scheduling
+ * Release the write lock.
  */
-void btrfs_set_path_blocking(struct btrfs_path *p)
+void btrfs_tree_unlock(struct extent_buffer *eb)
 {
-	int i;
-
-	for (i = 0; i < BTRFS_MAX_LEVEL; i++) {
-		if (!p->nodes[i] || !p->locks[i])
-			continue;
-		/*
-		 * If we currently have a spinning reader or writer lock this
-		 * will bump the count of blocking holders and drop the
-		 * spinlock.
-		 */
-		if (p->locks[i] == BTRFS_READ_LOCK) {
-			btrfs_set_lock_blocking_read(p->nodes[i]);
-			p->locks[i] = BTRFS_READ_LOCK_BLOCKING;
-		} else if (p->locks[i] == BTRFS_WRITE_LOCK) {
-			btrfs_set_lock_blocking_write(p->nodes[i]);
-			p->locks[i] = BTRFS_WRITE_LOCK_BLOCKING;
-		}
-	}
+	trace_btrfs_tree_unlock(eb);
+	eb->lock_owner = 0;
+	up_write(&eb->lock);
 }
 
 /*
@@ -523,3 +238,167 @@ void btrfs_unlock_up_safe(struct btrfs_path *path, int level)
 		path->locks[i] = 0;
 	}
 }
+
+/*
+ * Loop around taking references on and locking the root node of the tree until
+ * we end up with a lock on the root node.
+ *
+ * Return: root extent buffer with write lock held
+ */
+struct extent_buffer *btrfs_lock_root_node(struct btrfs_root *root)
+{
+	struct extent_buffer *eb;
+
+	while (1) {
+		eb = btrfs_root_node(root);
+
+		btrfs_maybe_reset_lockdep_class(root, eb);
+		btrfs_tree_lock(eb);
+		if (eb == root->node)
+			break;
+		btrfs_tree_unlock(eb);
+		free_extent_buffer(eb);
+	}
+	return eb;
+}
+
+/*
+ * Loop around taking references on and locking the root node of the tree until
+ * we end up with a lock on the root node.
+ *
+ * Return: root extent buffer with read lock held
+ */
+struct extent_buffer *btrfs_read_lock_root_node(struct btrfs_root *root)
+{
+	struct extent_buffer *eb;
+
+	while (1) {
+		eb = btrfs_root_node(root);
+
+		btrfs_maybe_reset_lockdep_class(root, eb);
+		btrfs_tree_read_lock(eb);
+		if (eb == root->node)
+			break;
+		btrfs_tree_read_unlock(eb);
+		free_extent_buffer(eb);
+	}
+	return eb;
+}
+
+/*
+ * Loop around taking references on and locking the root node of the tree in
+ * nowait mode until we end up with a lock on the root node or returning to
+ * avoid blocking.
+ *
+ * Return: root extent buffer with read lock held or -EAGAIN.
+ */
+struct extent_buffer *btrfs_try_read_lock_root_node(struct btrfs_root *root)
+{
+	struct extent_buffer *eb;
+
+	while (1) {
+		eb = btrfs_root_node(root);
+		if (!btrfs_try_tree_read_lock(eb)) {
+			free_extent_buffer(eb);
+			return ERR_PTR(-EAGAIN);
+		}
+		if (eb == root->node)
+			break;
+		btrfs_tree_read_unlock(eb);
+		free_extent_buffer(eb);
+	}
+	return eb;
+}
+
+/*
+ * DREW locks
+ * ==========
+ *
+ * DREW stands for double-reader-writer-exclusion lock. It's used in situation
+ * where you want to provide A-B exclusion but not AA or BB.
+ *
+ * Currently implementation gives more priority to reader. If a reader and a
+ * writer both race to acquire their respective sides of the lock the writer
+ * would yield its lock as soon as it detects a concurrent reader. Additionally
+ * if there are pending readers no new writers would be allowed to come in and
+ * acquire the lock.
+ */
+
+int btrfs_drew_lock_init(struct btrfs_drew_lock *lock)
+{
+	int ret;
+
+	ret = percpu_counter_init(&lock->writers, 0, GFP_KERNEL);
+	if (ret)
+		return ret;
+
+	atomic_set(&lock->readers, 0);
+	init_waitqueue_head(&lock->pending_readers);
+	init_waitqueue_head(&lock->pending_writers);
+
+	return 0;
+}
+
+void btrfs_drew_lock_destroy(struct btrfs_drew_lock *lock)
+{
+	percpu_counter_destroy(&lock->writers);
+}
+
+/* Return true if acquisition is successful, false otherwise */
+bool btrfs_drew_try_write_lock(struct btrfs_drew_lock *lock)
+{
+	if (atomic_read(&lock->readers))
+		return false;
+
+	percpu_counter_inc(&lock->writers);
+
+	/* Ensure writers count is updated before we check for pending readers */
+	smp_mb();
+	if (atomic_read(&lock->readers)) {
+		btrfs_drew_write_unlock(lock);
+		return false;
+	}
+
+	return true;
+}
+
+void btrfs_drew_write_lock(struct btrfs_drew_lock *lock)
+{
+	while (true) {
+		if (btrfs_drew_try_write_lock(lock))
+			return;
+		wait_event(lock->pending_writers, !atomic_read(&lock->readers));
+	}
+}
+
+void btrfs_drew_write_unlock(struct btrfs_drew_lock *lock)
+{
+	percpu_counter_dec(&lock->writers);
+	cond_wake_up(&lock->pending_readers);
+}
+
+void btrfs_drew_read_lock(struct btrfs_drew_lock *lock)
+{
+	atomic_inc(&lock->readers);
+
+	/*
+	 * Ensure the pending reader count is perceieved BEFORE this reader
+	 * goes to sleep in case of active writers. This guarantees new writers
+	 * won't be allowed and that the current reader will be woken up when
+	 * the last active writer finishes its jobs.
+	 */
+	smp_mb__after_atomic();
+
+	wait_event(lock->pending_readers,
+		   percpu_counter_sum(&lock->writers) == 0);
+}
+
+void btrfs_drew_read_unlock(struct btrfs_drew_lock *lock)
+{
+	/*
+	 * atomic_dec_and_test implies a full barrier, so woken up writers
+	 * are guaranteed to see the decrement
+	 */
+	if (atomic_dec_and_test(&lock->readers))
+		wake_up(&lock->pending_writers);
+}
diff --git a/fs/btrfs/locking.h b/fs/btrfs/locking.h
index 21a285883e89..490c7a79e995 100644
--- a/fs/btrfs/locking.h
+++ b/fs/btrfs/locking.h
@@ -6,46 +6,144 @@
 #ifndef BTRFS_LOCKING_H
 #define BTRFS_LOCKING_H
 
+#include <linux/atomic.h>
+#include <linux/wait.h>
+#include <linux/percpu_counter.h>
 #include "extent_io.h"
 
 #define BTRFS_WRITE_LOCK 1
 #define BTRFS_READ_LOCK 2
-#define BTRFS_WRITE_LOCK_BLOCKING 3
-#define BTRFS_READ_LOCK_BLOCKING 4
 
+/*
+ * We are limited in number of subclasses by MAX_LOCKDEP_SUBCLASSES, which at
+ * the time of this patch is 8, which is how many we use.  Keep this in mind if
+ * you decide you want to add another subclass.
+ */
+enum btrfs_lock_nesting {
+	BTRFS_NESTING_NORMAL,
+
+	/*
+	 * When we COW a block we are holding the lock on the original block,
+	 * and since our lockdep maps are rootid+level, this confuses lockdep
+	 * when we lock the newly allocated COW'd block.  Handle this by having
+	 * a subclass for COW'ed blocks so that lockdep doesn't complain.
+	 */
+	BTRFS_NESTING_COW,
+
+	/*
+	 * Oftentimes we need to lock adjacent nodes on the same level while
+	 * still holding the lock on the original node we searched to, such as
+	 * for searching forward or for split/balance.
+	 *
+	 * Because of this we need to indicate to lockdep that this is
+	 * acceptable by having a different subclass for each of these
+	 * operations.
+	 */
+	BTRFS_NESTING_LEFT,
+	BTRFS_NESTING_RIGHT,
+
+	/*
+	 * When splitting we will be holding a lock on the left/right node when
+	 * we need to cow that node, thus we need a new set of subclasses for
+	 * these two operations.
+	 */
+	BTRFS_NESTING_LEFT_COW,
+	BTRFS_NESTING_RIGHT_COW,
+
+	/*
+	 * When splitting we may push nodes to the left or right, but still use
+	 * the subsequent nodes in our path, keeping our locks on those adjacent
+	 * blocks.  Thus when we go to allocate a new split block we've already
+	 * used up all of our available subclasses, so this subclass exists to
+	 * handle this case where we need to allocate a new split block.
+	 */
+	BTRFS_NESTING_SPLIT,
+
+	/*
+	 * When promoting a new block to a root we need to have a special
+	 * subclass so we don't confuse lockdep, as it will appear that we are
+	 * locking a higher level node before a lower level one.  Copying also
+	 * has this problem as it appears we're locking the same block again
+	 * when we make a snapshot of an existing root.
+	 */
+	BTRFS_NESTING_NEW_ROOT,
+
+	/*
+	 * We are limited to MAX_LOCKDEP_SUBLCLASSES number of subclasses, so
+	 * add this in here and add a static_assert to keep us from going over
+	 * the limit.  As of this writing we're limited to 8, and we're
+	 * definitely using 8, hence this check to keep us from messing up in
+	 * the future.
+	 */
+	BTRFS_NESTING_MAX,
+};
+
+static_assert(BTRFS_NESTING_MAX <= MAX_LOCKDEP_SUBCLASSES,
+	      "too many lock subclasses defined");
+
+struct btrfs_path;
+
+void __btrfs_tree_lock(struct extent_buffer *eb, enum btrfs_lock_nesting nest);
 void btrfs_tree_lock(struct extent_buffer *eb);
 void btrfs_tree_unlock(struct extent_buffer *eb);
 
+void __btrfs_tree_read_lock(struct extent_buffer *eb, enum btrfs_lock_nesting nest);
 void btrfs_tree_read_lock(struct extent_buffer *eb);
 void btrfs_tree_read_unlock(struct extent_buffer *eb);
-void btrfs_tree_read_unlock_blocking(struct extent_buffer *eb);
-void btrfs_set_lock_blocking_read(struct extent_buffer *eb);
-void btrfs_set_lock_blocking_write(struct extent_buffer *eb);
 int btrfs_try_tree_read_lock(struct extent_buffer *eb);
 int btrfs_try_tree_write_lock(struct extent_buffer *eb);
-int btrfs_tree_read_lock_atomic(struct extent_buffer *eb);
+struct extent_buffer *btrfs_lock_root_node(struct btrfs_root *root);
+struct extent_buffer *btrfs_read_lock_root_node(struct btrfs_root *root);
+struct extent_buffer *btrfs_try_read_lock_root_node(struct btrfs_root *root);
 
 #ifdef CONFIG_BTRFS_DEBUG
-static inline void btrfs_assert_tree_locked(struct extent_buffer *eb) {
-	BUG_ON(!eb->write_locks);
+static inline void btrfs_assert_tree_write_locked(struct extent_buffer *eb)
+{
+	lockdep_assert_held_write(&eb->lock);
 }
 #else
-static inline void btrfs_assert_tree_locked(struct extent_buffer *eb) { }
+static inline void btrfs_assert_tree_write_locked(struct extent_buffer *eb) { }
 #endif
 
-void btrfs_set_path_blocking(struct btrfs_path *p);
 void btrfs_unlock_up_safe(struct btrfs_path *path, int level);
 
 static inline void btrfs_tree_unlock_rw(struct extent_buffer *eb, int rw)
 {
-	if (rw == BTRFS_WRITE_LOCK || rw == BTRFS_WRITE_LOCK_BLOCKING)
+	if (rw == BTRFS_WRITE_LOCK)
 		btrfs_tree_unlock(eb);
-	else if (rw == BTRFS_READ_LOCK_BLOCKING)
-		btrfs_tree_read_unlock_blocking(eb);
 	else if (rw == BTRFS_READ_LOCK)
 		btrfs_tree_read_unlock(eb);
 	else
 		BUG();
 }
 
+struct btrfs_drew_lock {
+	atomic_t readers;
+	struct percpu_counter writers;
+	wait_queue_head_t pending_writers;
+	wait_queue_head_t pending_readers;
+};
+
+int btrfs_drew_lock_init(struct btrfs_drew_lock *lock);
+void btrfs_drew_lock_destroy(struct btrfs_drew_lock *lock);
+void btrfs_drew_write_lock(struct btrfs_drew_lock *lock);
+bool btrfs_drew_try_write_lock(struct btrfs_drew_lock *lock);
+void btrfs_drew_write_unlock(struct btrfs_drew_lock *lock);
+void btrfs_drew_read_lock(struct btrfs_drew_lock *lock);
+void btrfs_drew_read_unlock(struct btrfs_drew_lock *lock);
+
+#ifdef CONFIG_DEBUG_LOCK_ALLOC
+void btrfs_set_buffer_lockdep_class(u64 objectid, struct extent_buffer *eb, int level);
+void btrfs_maybe_reset_lockdep_class(struct btrfs_root *root, struct extent_buffer *eb);
+#else
+static inline void btrfs_set_buffer_lockdep_class(u64 objectid,
+					struct extent_buffer *eb, int level)
+{
+}
+static inline void btrfs_maybe_reset_lockdep_class(struct btrfs_root *root,
+						   struct extent_buffer *eb)
+{
+}
+#endif
+
 #endif
diff --git a/fs/btrfs/lzo.c b/fs/btrfs/lzo.c
index aa9cd11f4b78..89bc5f825e0a 100644
--- a/fs/btrfs/lzo.c
+++ b/fs/btrfs/lzo.c
@@ -14,6 +14,7 @@
 #include <linux/lzo.h>
 #include <linux/refcount.h>
 #include "compression.h"
+#include "ctree.h"
 
 #define LZO_LEN	4
 
@@ -31,19 +32,19 @@
  *     payload.
  *     One regular LZO compressed extent can have one or more segments.
  *     For inlined LZO compressed extent, only one segment is allowed.
- *     One segment represents at most one page of uncompressed data.
+ *     One segment represents at most one sector of uncompressed data.
  *
  * 2.1 Segment header
  *     Fixed size. LZO_LEN (4) bytes long, LE32.
  *     Records the total size of the segment (not including the header).
- *     Segment header never crosses page boundary, thus it's possible to
- *     have at most 3 padding zeros at the end of the page.
+ *     Segment header never crosses sector boundary, thus it's possible to
+ *     have at most 3 padding zeros at the end of the sector.
  *
  * 2.2 Data Payload
- *     Variable size. Size up limit should be lzo1x_worst_compress(PAGE_SIZE)
- *     which is 4419 for a 4KiB page.
+ *     Variable size. Size up limit should be lzo1x_worst_compress(sectorsize)
+ *     which is 4419 for a 4KiB sectorsize.
  *
- * Example:
+ * Example with 4K sectorsize:
  * Page 1:
  *          0     0x2   0x4   0x6   0x8   0xa   0xc   0xe     0x10
  * 0x0000   |  Header   | SegHdr 01 | Data payload 01 ...     |
@@ -54,6 +55,9 @@
  * 0x1000   | SegHdr N+1| Data payload N+1 ...                |
  */
 
+#define WORKSPACE_BUF_LENGTH	(lzo1x_worst_compress(PAGE_SIZE))
+#define WORKSPACE_CBUF_LENGTH	(lzo1x_worst_compress(PAGE_SIZE))
+
 struct workspace {
 	void *mem;
 	void *buf;	/* where decompressed data goes */
@@ -82,8 +86,8 @@ struct list_head *lzo_alloc_workspace(unsigned int level)
 		return ERR_PTR(-ENOMEM);
 
 	workspace->mem = kvmalloc(LZO1X_MEM_COMPRESS, GFP_KERNEL);
-	workspace->buf = kvmalloc(lzo1x_worst_compress(PAGE_SIZE), GFP_KERNEL);
-	workspace->cbuf = kvmalloc(lzo1x_worst_compress(PAGE_SIZE), GFP_KERNEL);
+	workspace->buf = kvmalloc(WORKSPACE_BUF_LENGTH, GFP_KERNEL);
+	workspace->cbuf = kvmalloc(WORKSPACE_CBUF_LENGTH, GFP_KERNEL);
 	if (!workspace->mem || !workspace->buf || !workspace->cbuf)
 		goto fail;
 
@@ -111,312 +115,313 @@ static inline size_t read_compress_length(const char *buf)
 	return le32_to_cpu(dlen);
 }
 
+/*
+ * Will do:
+ *
+ * - Write a segment header into the destination
+ * - Copy the compressed buffer into the destination
+ * - Make sure we have enough space in the last sector to fit a segment header
+ *   If not, we will pad at most (LZO_LEN (4)) - 1 bytes of zeros.
+ *
+ * Will allocate new pages when needed.
+ */
+static int copy_compressed_data_to_page(char *compressed_data,
+					size_t compressed_size,
+					struct page **out_pages,
+					unsigned long max_nr_page,
+					u32 *cur_out,
+					const u32 sectorsize)
+{
+	u32 sector_bytes_left;
+	u32 orig_out;
+	struct page *cur_page;
+	char *kaddr;
+
+	if ((*cur_out / PAGE_SIZE) >= max_nr_page)
+		return -E2BIG;
+
+	/*
+	 * We never allow a segment header crossing sector boundary, previous
+	 * run should ensure we have enough space left inside the sector.
+	 */
+	ASSERT((*cur_out / sectorsize) == (*cur_out + LZO_LEN - 1) / sectorsize);
+
+	cur_page = out_pages[*cur_out / PAGE_SIZE];
+	/* Allocate a new page */
+	if (!cur_page) {
+		cur_page = alloc_page(GFP_NOFS);
+		if (!cur_page)
+			return -ENOMEM;
+		out_pages[*cur_out / PAGE_SIZE] = cur_page;
+	}
+
+	kaddr = kmap_local_page(cur_page);
+	write_compress_length(kaddr + offset_in_page(*cur_out),
+			      compressed_size);
+	*cur_out += LZO_LEN;
+
+	orig_out = *cur_out;
+
+	/* Copy compressed data */
+	while (*cur_out - orig_out < compressed_size) {
+		u32 copy_len = min_t(u32, sectorsize - *cur_out % sectorsize,
+				     orig_out + compressed_size - *cur_out);
+
+		kunmap_local(kaddr);
+
+		if ((*cur_out / PAGE_SIZE) >= max_nr_page)
+			return -E2BIG;
+
+		cur_page = out_pages[*cur_out / PAGE_SIZE];
+		/* Allocate a new page */
+		if (!cur_page) {
+			cur_page = alloc_page(GFP_NOFS);
+			if (!cur_page)
+				return -ENOMEM;
+			out_pages[*cur_out / PAGE_SIZE] = cur_page;
+		}
+		kaddr = kmap_local_page(cur_page);
+
+		memcpy(kaddr + offset_in_page(*cur_out),
+		       compressed_data + *cur_out - orig_out, copy_len);
+
+		*cur_out += copy_len;
+	}
+
+	/*
+	 * Check if we can fit the next segment header into the remaining space
+	 * of the sector.
+	 */
+	sector_bytes_left = round_up(*cur_out, sectorsize) - *cur_out;
+	if (sector_bytes_left >= LZO_LEN || sector_bytes_left == 0)
+		goto out;
+
+	/* The remaining size is not enough, pad it with zeros */
+	memset(kaddr + offset_in_page(*cur_out), 0,
+	       sector_bytes_left);
+	*cur_out += sector_bytes_left;
+
+out:
+	kunmap_local(kaddr);
+	return 0;
+}
+
 int lzo_compress_pages(struct list_head *ws, struct address_space *mapping,
 		u64 start, struct page **pages, unsigned long *out_pages,
 		unsigned long *total_in, unsigned long *total_out)
 {
 	struct workspace *workspace = list_entry(ws, struct workspace, list);
+	const u32 sectorsize = btrfs_sb(mapping->host->i_sb)->sectorsize;
+	struct page *page_in = NULL;
+	char *sizes_ptr;
+	const unsigned long max_nr_page = *out_pages;
 	int ret = 0;
-	char *data_in;
-	char *cpage_out;
-	int nr_pages = 0;
-	struct page *in_page = NULL;
-	struct page *out_page = NULL;
-	unsigned long bytes_left;
-	unsigned long len = *total_out;
-	unsigned long nr_dest_pages = *out_pages;
-	const unsigned long max_out = nr_dest_pages * PAGE_SIZE;
-	size_t in_len;
-	size_t out_len;
-	char *buf;
-	unsigned long tot_in = 0;
-	unsigned long tot_out = 0;
-	unsigned long pg_bytes_left;
-	unsigned long out_offset;
-	unsigned long bytes;
+	/* Points to the file offset of input data */
+	u64 cur_in = start;
+	/* Points to the current output byte */
+	u32 cur_out = 0;
+	u32 len = *total_out;
 
+	ASSERT(max_nr_page > 0);
 	*out_pages = 0;
 	*total_out = 0;
 	*total_in = 0;
 
-	in_page = find_get_page(mapping, start >> PAGE_SHIFT);
-	data_in = kmap(in_page);
-
 	/*
-	 * store the size of all chunks of compressed data in
-	 * the first 4 bytes
+	 * Skip the header for now, we will later come back and write the total
+	 * compressed size
 	 */
-	out_page = alloc_page(GFP_NOFS | __GFP_HIGHMEM);
-	if (out_page == NULL) {
-		ret = -ENOMEM;
-		goto out;
-	}
-	cpage_out = kmap(out_page);
-	out_offset = LZO_LEN;
-	tot_out = LZO_LEN;
-	pages[0] = out_page;
-	nr_pages = 1;
-	pg_bytes_left = PAGE_SIZE - LZO_LEN;
-
-	/* compress at most one page of data each time */
-	in_len = min(len, PAGE_SIZE);
-	while (tot_in < len) {
-		ret = lzo1x_1_compress(data_in, in_len, workspace->cbuf,
-				       &out_len, workspace->mem);
-		if (ret != LZO_E_OK) {
-			pr_debug("BTRFS: lzo in loop returned %d\n",
-			       ret);
+	cur_out += LZO_LEN;
+	while (cur_in < start + len) {
+		char *data_in;
+		const u32 sectorsize_mask = sectorsize - 1;
+		u32 sector_off = (cur_in - start) & sectorsize_mask;
+		u32 in_len;
+		size_t out_len;
+
+		/* Get the input page first */
+		if (!page_in) {
+			page_in = find_get_page(mapping, cur_in >> PAGE_SHIFT);
+			ASSERT(page_in);
+		}
+
+		/* Compress at most one sector of data each time */
+		in_len = min_t(u32, start + len - cur_in, sectorsize - sector_off);
+		ASSERT(in_len);
+		data_in = kmap_local_page(page_in);
+		ret = lzo1x_1_compress(data_in +
+				       offset_in_page(cur_in), in_len,
+				       workspace->cbuf, &out_len,
+				       workspace->mem);
+		kunmap_local(data_in);
+		if (ret < 0) {
+			pr_debug("BTRFS: lzo in loop returned %d\n", ret);
 			ret = -EIO;
 			goto out;
 		}
 
-		/* store the size of this chunk of compressed data */
-		write_compress_length(cpage_out + out_offset, out_len);
-		tot_out += LZO_LEN;
-		out_offset += LZO_LEN;
-		pg_bytes_left -= LZO_LEN;
-
-		tot_in += in_len;
-		tot_out += out_len;
-
-		/* copy bytes from the working buffer into the pages */
-		buf = workspace->cbuf;
-		while (out_len) {
-			bytes = min_t(unsigned long, pg_bytes_left, out_len);
-
-			memcpy(cpage_out + out_offset, buf, bytes);
-
-			out_len -= bytes;
-			pg_bytes_left -= bytes;
-			buf += bytes;
-			out_offset += bytes;
+		ret = copy_compressed_data_to_page(workspace->cbuf, out_len,
+						   pages, max_nr_page,
+						   &cur_out, sectorsize);
+		if (ret < 0)
+			goto out;
 
-			/*
-			 * we need another page for writing out.
-			 *
-			 * Note if there's less than 4 bytes left, we just
-			 * skip to a new page.
-			 */
-			if ((out_len == 0 && pg_bytes_left < LZO_LEN) ||
-			    pg_bytes_left == 0) {
-				if (pg_bytes_left) {
-					memset(cpage_out + out_offset, 0,
-					       pg_bytes_left);
-					tot_out += pg_bytes_left;
-				}
-
-				/* we're done, don't allocate new page */
-				if (out_len == 0 && tot_in >= len)
-					break;
-
-				kunmap(out_page);
-				if (nr_pages == nr_dest_pages) {
-					out_page = NULL;
-					ret = -E2BIG;
-					goto out;
-				}
-
-				out_page = alloc_page(GFP_NOFS | __GFP_HIGHMEM);
-				if (out_page == NULL) {
-					ret = -ENOMEM;
-					goto out;
-				}
-				cpage_out = kmap(out_page);
-				pages[nr_pages++] = out_page;
-
-				pg_bytes_left = PAGE_SIZE;
-				out_offset = 0;
-			}
-		}
+		cur_in += in_len;
 
-		/* we're making it bigger, give up */
-		if (tot_in > 8192 && tot_in < tot_out) {
+		/*
+		 * Check if we're making it bigger after two sectors.  And if
+		 * it is so, give up.
+		 */
+		if (cur_in - start > sectorsize * 2 && cur_in - start < cur_out) {
 			ret = -E2BIG;
 			goto out;
 		}
 
-		/* we're all done */
-		if (tot_in >= len)
-			break;
-
-		if (tot_out > max_out)
-			break;
+		/* Check if we have reached page boundary */
+		if (IS_ALIGNED(cur_in, PAGE_SIZE)) {
+			put_page(page_in);
+			page_in = NULL;
+		}
+	}
 
-		bytes_left = len - tot_in;
-		kunmap(in_page);
-		put_page(in_page);
+	/* Store the size of all chunks of compressed data */
+	sizes_ptr = kmap_local_page(pages[0]);
+	write_compress_length(sizes_ptr, cur_out);
+	kunmap_local(sizes_ptr);
 
-		start += PAGE_SIZE;
-		in_page = find_get_page(mapping, start >> PAGE_SHIFT);
-		data_in = kmap(in_page);
-		in_len = min(bytes_left, PAGE_SIZE);
-	}
+	ret = 0;
+	*total_out = cur_out;
+	*total_in = cur_in - start;
+out:
+	if (page_in)
+		put_page(page_in);
+	*out_pages = DIV_ROUND_UP(cur_out, PAGE_SIZE);
+	return ret;
+}
 
-	if (tot_out >= tot_in) {
-		ret = -E2BIG;
-		goto out;
-	}
+/*
+ * Copy the compressed segment payload into @dest.
+ *
+ * For the payload there will be no padding, just need to do page switching.
+ */
+static void copy_compressed_segment(struct compressed_bio *cb,
+				    char *dest, u32 len, u32 *cur_in)
+{
+	u32 orig_in = *cur_in;
 
-	/* store the size of all chunks of compressed data */
-	cpage_out = kmap(pages[0]);
-	write_compress_length(cpage_out, tot_out);
+	while (*cur_in < orig_in + len) {
+		struct page *cur_page;
+		u32 copy_len = min_t(u32, PAGE_SIZE - offset_in_page(*cur_in),
+					  orig_in + len - *cur_in);
 
-	kunmap(pages[0]);
+		ASSERT(copy_len);
+		cur_page = cb->compressed_pages[*cur_in / PAGE_SIZE];
 
-	ret = 0;
-	*total_out = tot_out;
-	*total_in = tot_in;
-out:
-	*out_pages = nr_pages;
-	if (out_page)
-		kunmap(out_page);
+		memcpy_from_page(dest + *cur_in - orig_in, cur_page,
+				 offset_in_page(*cur_in), copy_len);
 
-	if (in_page) {
-		kunmap(in_page);
-		put_page(in_page);
+		*cur_in += copy_len;
 	}
-
-	return ret;
 }
 
 int lzo_decompress_bio(struct list_head *ws, struct compressed_bio *cb)
 {
 	struct workspace *workspace = list_entry(ws, struct workspace, list);
-	int ret = 0, ret2;
-	char *data_in;
-	unsigned long page_in_index = 0;
-	size_t srclen = cb->compressed_len;
-	unsigned long total_pages_in = DIV_ROUND_UP(srclen, PAGE_SIZE);
-	unsigned long buf_start;
-	unsigned long buf_offset = 0;
-	unsigned long bytes;
-	unsigned long working_bytes;
-	size_t in_len;
-	size_t out_len;
-	const size_t max_segment_len = lzo1x_worst_compress(PAGE_SIZE);
-	unsigned long in_offset;
-	unsigned long in_page_bytes_left;
-	unsigned long tot_in;
-	unsigned long tot_out;
-	unsigned long tot_len;
-	char *buf;
-	bool may_late_unmap, need_unmap;
-	struct page **pages_in = cb->compressed_pages;
-	u64 disk_start = cb->start;
-	struct bio *orig_bio = cb->orig_bio;
-
-	data_in = kmap(pages_in[0]);
-	tot_len = read_compress_length(data_in);
+	const struct btrfs_fs_info *fs_info = btrfs_sb(cb->inode->i_sb);
+	const u32 sectorsize = fs_info->sectorsize;
+	char *kaddr;
+	int ret;
+	/* Compressed data length, can be unaligned */
+	u32 len_in;
+	/* Offset inside the compressed data */
+	u32 cur_in = 0;
+	/* Bytes decompressed so far */
+	u32 cur_out = 0;
+
+	kaddr = kmap_local_page(cb->compressed_pages[0]);
+	len_in = read_compress_length(kaddr);
+	kunmap_local(kaddr);
+	cur_in += LZO_LEN;
+
 	/*
-	 * Compressed data header check.
+	 * LZO header length check
 	 *
-	 * The real compressed size can't exceed the maximum extent length, and
-	 * all pages should be used (whole unused page with just the segment
-	 * header is not possible).  If this happens it means the compressed
-	 * extent is corrupted.
+	 * The total length should not exceed the maximum extent length,
+	 * and all sectors should be used.
+	 * If this happens, it means the compressed extent is corrupted.
 	 */
-	if (tot_len > min_t(size_t, BTRFS_MAX_COMPRESSED, srclen) ||
-	    tot_len < srclen - PAGE_SIZE) {
-		ret = -EUCLEAN;
-		goto done;
+	if (len_in > min_t(size_t, BTRFS_MAX_COMPRESSED, cb->compressed_len) ||
+	    round_up(len_in, sectorsize) < cb->compressed_len) {
+		btrfs_err(fs_info,
+			"invalid lzo header, lzo len %u compressed len %u",
+			len_in, cb->compressed_len);
+		return -EUCLEAN;
 	}
 
-	tot_in = LZO_LEN;
-	in_offset = LZO_LEN;
-	in_page_bytes_left = PAGE_SIZE - LZO_LEN;
-
-	tot_out = 0;
-
-	while (tot_in < tot_len) {
-		in_len = read_compress_length(data_in + in_offset);
-		in_page_bytes_left -= LZO_LEN;
-		in_offset += LZO_LEN;
-		tot_in += LZO_LEN;
+	/* Go through each lzo segment */
+	while (cur_in < len_in) {
+		struct page *cur_page;
+		/* Length of the compressed segment */
+		u32 seg_len;
+		u32 sector_bytes_left;
+		size_t out_len = lzo1x_worst_compress(sectorsize);
 
 		/*
-		 * Segment header check.
-		 *
-		 * The segment length must not exceed the maximum LZO
-		 * compression size, nor the total compressed size.
+		 * We should always have enough space for one segment header
+		 * inside current sector.
 		 */
-		if (in_len > max_segment_len || tot_in + in_len > tot_len) {
-			ret = -EUCLEAN;
-			goto done;
-		}
-
-		tot_in += in_len;
-		working_bytes = in_len;
-		may_late_unmap = need_unmap = false;
-
-		/* fast path: avoid using the working buffer */
-		if (in_page_bytes_left >= in_len) {
-			buf = data_in + in_offset;
-			bytes = in_len;
-			may_late_unmap = true;
-			goto cont;
+		ASSERT(cur_in / sectorsize ==
+		       (cur_in + LZO_LEN - 1) / sectorsize);
+		cur_page = cb->compressed_pages[cur_in / PAGE_SIZE];
+		ASSERT(cur_page);
+		kaddr = kmap_local_page(cur_page);
+		seg_len = read_compress_length(kaddr + offset_in_page(cur_in));
+		kunmap_local(kaddr);
+		cur_in += LZO_LEN;
+
+		if (seg_len > WORKSPACE_CBUF_LENGTH) {
+			/*
+			 * seg_len shouldn't be larger than we have allocated
+			 * for workspace->cbuf
+			 */
+			btrfs_err(fs_info, "unexpectedly large lzo segment len %u",
+					seg_len);
+			ret = -EIO;
+			goto out;
 		}
 
-		/* copy bytes from the pages into the working buffer */
-		buf = workspace->cbuf;
-		buf_offset = 0;
-		while (working_bytes) {
-			bytes = min(working_bytes, in_page_bytes_left);
-
-			memcpy(buf + buf_offset, data_in + in_offset, bytes);
-			buf_offset += bytes;
-cont:
-			working_bytes -= bytes;
-			in_page_bytes_left -= bytes;
-			in_offset += bytes;
-
-			/* check if we need to pick another page */
-			if ((working_bytes == 0 && in_page_bytes_left < LZO_LEN)
-			    || in_page_bytes_left == 0) {
-				tot_in += in_page_bytes_left;
-
-				if (working_bytes == 0 && tot_in >= tot_len)
-					break;
-
-				if (page_in_index + 1 >= total_pages_in) {
-					ret = -EIO;
-					goto done;
-				}
-
-				if (may_late_unmap)
-					need_unmap = true;
-				else
-					kunmap(pages_in[page_in_index]);
-
-				data_in = kmap(pages_in[++page_in_index]);
-
-				in_page_bytes_left = PAGE_SIZE;
-				in_offset = 0;
-			}
-		}
+		/* Copy the compressed segment payload into workspace */
+		copy_compressed_segment(cb, workspace->cbuf, seg_len, &cur_in);
 
-		out_len = max_segment_len;
-		ret = lzo1x_decompress_safe(buf, in_len, workspace->buf,
-					    &out_len);
-		if (need_unmap)
-			kunmap(pages_in[page_in_index - 1]);
+		/* Decompress the data */
+		ret = lzo1x_decompress_safe(workspace->cbuf, seg_len,
+					    workspace->buf, &out_len);
 		if (ret != LZO_E_OK) {
-			pr_warn("BTRFS: decompress failed\n");
+			btrfs_err(fs_info, "failed to decompress");
 			ret = -EIO;
-			break;
+			goto out;
 		}
 
-		buf_start = tot_out;
-		tot_out += out_len;
+		/* Copy the data into inode pages */
+		ret = btrfs_decompress_buf2page(workspace->buf, out_len, cb, cur_out);
+		cur_out += out_len;
+
+		/* All data read, exit */
+		if (ret == 0)
+			goto out;
+		ret = 0;
 
-		ret2 = btrfs_decompress_buf2page(workspace->buf, buf_start,
-						 tot_out, disk_start, orig_bio);
-		if (ret2 == 0)
-			break;
+		/* Check if the sector has enough space for a segment header */
+		sector_bytes_left = sectorsize - (cur_in % sectorsize);
+		if (sector_bytes_left >= LZO_LEN)
+			continue;
+
+		/* Skip the padding zeros */
+		cur_in += sector_bytes_left;
 	}
-done:
-	kunmap(pages_in[page_in_index]);
+out:
 	if (!ret)
-		zero_fill_bio(orig_bio);
+		zero_fill_bio(cb->orig_bio);
 	return ret;
 }
 
@@ -427,7 +432,7 @@ int lzo_decompress(struct list_head *ws, unsigned char *data_in,
 	struct workspace *workspace = list_entry(ws, struct workspace, list);
 	size_t in_len;
 	size_t out_len;
-	size_t max_segment_len = lzo1x_worst_compress(PAGE_SIZE);
+	size_t max_segment_len = WORKSPACE_BUF_LENGTH;
 	int ret = 0;
 	char *kaddr;
 	unsigned long bytes;
@@ -467,7 +472,7 @@ int lzo_decompress(struct list_head *ws, unsigned char *data_in,
 	destlen = min_t(unsigned long, destlen, PAGE_SIZE);
 	bytes = min_t(unsigned long, destlen, out_len - start_byte);
 
-	kaddr = kmap_atomic(dest_page);
+	kaddr = kmap_local_page(dest_page);
 	memcpy(kaddr, workspace->buf + start_byte, bytes);
 
 	/*
@@ -477,7 +482,7 @@ int lzo_decompress(struct list_head *ws, unsigned char *data_in,
 	 */
 	if (bytes < destlen)
 		memset(kaddr+bytes, 0, destlen-bytes);
-	kunmap_atomic(kaddr);
+	kunmap_local(kaddr);
 out:
 	return ret;
 }
diff --git a/fs/btrfs/misc.h b/fs/btrfs/misc.h
index 72bab64ecf60..f9850edfd726 100644
--- a/fs/btrfs/misc.h
+++ b/fs/btrfs/misc.h
@@ -5,7 +5,8 @@
 
 #include <linux/sched.h>
 #include <linux/wait.h>
-#include <asm/div64.h>
+#include <linux/math64.h>
+#include <linux/rbtree.h>
 
 #define in_range(b, first, len) ((b) >= (first) && (b) < (first) + (len))
 
@@ -58,4 +59,92 @@ static inline bool has_single_bit_set(u64 n)
 	return is_power_of_two_u64(n);
 }
 
+/*
+ * Simple bytenr based rb_tree relate structures
+ *
+ * Any structure wants to use bytenr as single search index should have their
+ * structure start with these members.
+ */
+struct rb_simple_node {
+	struct rb_node rb_node;
+	u64 bytenr;
+};
+
+static inline struct rb_node *rb_simple_search(struct rb_root *root, u64 bytenr)
+{
+	struct rb_node *node = root->rb_node;
+	struct rb_simple_node *entry;
+
+	while (node) {
+		entry = rb_entry(node, struct rb_simple_node, rb_node);
+
+		if (bytenr < entry->bytenr)
+			node = node->rb_left;
+		else if (bytenr > entry->bytenr)
+			node = node->rb_right;
+		else
+			return node;
+	}
+	return NULL;
+}
+
+/*
+ * Search @root from an entry that starts or comes after @bytenr.
+ *
+ * @root:	the root to search.
+ * @bytenr:	bytenr to search from.
+ *
+ * Return the rb_node that start at or after @bytenr.  If there is no entry at
+ * or after @bytner return NULL.
+ */
+static inline struct rb_node *rb_simple_search_first(struct rb_root *root,
+						     u64 bytenr)
+{
+	struct rb_node *node = root->rb_node, *ret = NULL;
+	struct rb_simple_node *entry, *ret_entry = NULL;
+
+	while (node) {
+		entry = rb_entry(node, struct rb_simple_node, rb_node);
+
+		if (bytenr < entry->bytenr) {
+			if (!ret || entry->bytenr < ret_entry->bytenr) {
+				ret = node;
+				ret_entry = entry;
+			}
+
+			node = node->rb_left;
+		} else if (bytenr > entry->bytenr) {
+			node = node->rb_right;
+		} else {
+			return node;
+		}
+	}
+
+	return ret;
+}
+
+static inline struct rb_node *rb_simple_insert(struct rb_root *root, u64 bytenr,
+					       struct rb_node *node)
+{
+	struct rb_node **p = &root->rb_node;
+	struct rb_node *parent = NULL;
+	struct rb_simple_node *entry;
+
+	while (*p) {
+		parent = *p;
+		entry = rb_entry(parent, struct rb_simple_node, rb_node);
+
+		if (bytenr < entry->bytenr)
+			p = &(*p)->rb_left;
+		else if (bytenr > entry->bytenr)
+			p = &(*p)->rb_right;
+		else
+			return parent;
+	}
+
+	rb_link_node(node, parent, p);
+	rb_insert_color(node, root);
+	return NULL;
+}
+
 #endif
diff --git a/fs/btrfs/ordered-data.c b/fs/btrfs/ordered-data.c
index a65f189a5b94..e54f8280031f 100644
--- a/fs/btrfs/ordered-data.c
+++ b/fs/btrfs/ordered-data.c
@@ -15,6 +15,8 @@
 #include "disk-io.h"
 #include "compression.h"
 #include "delalloc-space.h"
+#include "qgroup.h"
+#include "subpage.h"
 
 static struct kmem_cache *btrfs_ordered_extent_cache;
 
@@ -106,17 +108,6 @@ static struct rb_node *__tree_search(struct rb_root *root, u64 file_offset,
 	return NULL;
 }
 
-/*
- * helper to check if a given offset is inside a given entry
- */
-static int offset_in_entry(struct btrfs_ordered_extent *entry, u64 file_offset)
-{
-	if (file_offset < entry->file_offset ||
-	    entry->file_offset + entry->num_bytes <= file_offset)
-		return 0;
-	return 1;
-}
-
 static int range_overlaps(struct btrfs_ordered_extent *entry, u64 file_offset,
 			  u64 len)
 {
@@ -141,7 +132,7 @@ static inline struct rb_node *tree_search(struct btrfs_ordered_inode_tree *tree,
 	if (tree->last) {
 		entry = rb_entry(tree->last, struct btrfs_ordered_extent,
 				 rb_node);
-		if (offset_in_entry(entry, file_offset))
+		if (in_range(file_offset, entry->file_offset, entry->num_bytes))
 			return tree->last;
 	}
 	ret = __tree_search(root, file_offset, &prev);
@@ -152,53 +143,83 @@ static inline struct rb_node *tree_search(struct btrfs_ordered_inode_tree *tree,
 	return ret;
 }
 
-/* allocate and add a new ordered_extent into the per-inode tree.
+/**
+ * Add an ordered extent to the per-inode tree.
  *
- * The tree is given a single reference on the ordered extent that was
- * inserted.
+ * @inode:           Inode that this extent is for.
+ * @file_offset:     Logical offset in file where the extent starts.
+ * @num_bytes:       Logical length of extent in file.
+ * @ram_bytes:       Full length of unencoded data.
+ * @disk_bytenr:     Offset of extent on disk.
+ * @disk_num_bytes:  Size of extent on disk.
+ * @offset:          Offset into unencoded data where file data starts.
+ * @flags:           Flags specifying type of extent (1 << BTRFS_ORDERED_*).
+ * @compress_type:   Compression algorithm used for data.
+ *
+ * Most of these parameters correspond to &struct btrfs_file_extent_item. The
+ * tree is given a single reference on the ordered extent that was inserted.
+ *
+ * Return: 0 or -ENOMEM.
  */
-static int __btrfs_add_ordered_extent(struct inode *inode, u64 file_offset,
-				      u64 disk_bytenr, u64 num_bytes,
-				      u64 disk_num_bytes, int type, int dio,
-				      int compress_type)
+int btrfs_add_ordered_extent(struct btrfs_inode *inode, u64 file_offset,
+			     u64 num_bytes, u64 ram_bytes, u64 disk_bytenr,
+			     u64 disk_num_bytes, u64 offset, unsigned flags,
+			     int compress_type)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	struct btrfs_root *root = BTRFS_I(inode)->root;
-	struct btrfs_ordered_inode_tree *tree;
+	struct btrfs_root *root = inode->root;
+	struct btrfs_fs_info *fs_info = root->fs_info;
+	struct btrfs_ordered_inode_tree *tree = &inode->ordered_tree;
 	struct rb_node *node;
 	struct btrfs_ordered_extent *entry;
+	int ret;
 
-	tree = &BTRFS_I(inode)->ordered_tree;
+	if (flags &
+	    ((1 << BTRFS_ORDERED_NOCOW) | (1 << BTRFS_ORDERED_PREALLOC))) {
+		/* For nocow write, we can release the qgroup rsv right now */
+		ret = btrfs_qgroup_free_data(inode, NULL, file_offset, num_bytes);
+		if (ret < 0)
+			return ret;
+		ret = 0;
+	} else {
+		/*
+		 * The ordered extent has reserved qgroup space, release now
+		 * and pass the reserved number for qgroup_record to free.
+		 */
+		ret = btrfs_qgroup_release_data(inode, file_offset, num_bytes);
+		if (ret < 0)
+			return ret;
+	}
 	entry = kmem_cache_zalloc(btrfs_ordered_extent_cache, GFP_NOFS);
 	if (!entry)
 		return -ENOMEM;
 
 	entry->file_offset = file_offset;
-	entry->disk_bytenr = disk_bytenr;
 	entry->num_bytes = num_bytes;
+	entry->ram_bytes = ram_bytes;
+	entry->disk_bytenr = disk_bytenr;
 	entry->disk_num_bytes = disk_num_bytes;
+	entry->offset = offset;
 	entry->bytes_left = num_bytes;
-	entry->inode = igrab(inode);
+	entry->inode = igrab(&inode->vfs_inode);
 	entry->compress_type = compress_type;
 	entry->truncated_len = (u64)-1;
-	if (type != BTRFS_ORDERED_IO_DONE && type != BTRFS_ORDERED_COMPLETE)
-		set_bit(type, &entry->flags);
+	entry->qgroup_rsv = ret;
+	entry->physical = (u64)-1;
 
-	if (dio) {
-		percpu_counter_add_batch(&fs_info->dio_bytes, num_bytes,
-					 fs_info->delalloc_batch);
-		set_bit(BTRFS_ORDERED_DIRECT, &entry->flags);
-	}
+	ASSERT((flags & ~BTRFS_ORDERED_TYPE_FLAGS) == 0);
+	entry->flags = flags;
+
+	percpu_counter_add_batch(&fs_info->ordered_bytes, num_bytes,
+				 fs_info->delalloc_batch);
 
 	/* one ref for the tree */
 	refcount_set(&entry->refs, 1);
 	init_waitqueue_head(&entry->wait);
 	INIT_LIST_HEAD(&entry->list);
+	INIT_LIST_HEAD(&entry->log_list);
 	INIT_LIST_HEAD(&entry->root_extent_list);
 	INIT_LIST_HEAD(&entry->work_list);
 	init_completion(&entry->completion);
-	INIT_LIST_HEAD(&entry->log_list);
-	INIT_LIST_HEAD(&entry->trans_list);
 
 	trace_btrfs_ordered_extent_add(inode, entry);
 
@@ -228,41 +249,13 @@ static int __btrfs_add_ordered_extent(struct inode *inode, u64 file_offset,
 	 * that work has been done at higher layers, so this is truly the
 	 * smallest the extent is going to get.
 	 */
-	spin_lock(&BTRFS_I(inode)->lock);
-	btrfs_mod_outstanding_extents(BTRFS_I(inode), 1);
-	spin_unlock(&BTRFS_I(inode)->lock);
+	spin_lock(&inode->lock);
+	btrfs_mod_outstanding_extents(inode, 1);
+	spin_unlock(&inode->lock);
 
 	return 0;
 }
 
-int btrfs_add_ordered_extent(struct inode *inode, u64 file_offset,
-			     u64 disk_bytenr, u64 num_bytes, u64 disk_num_bytes,
-			     int type)
-{
-	return __btrfs_add_ordered_extent(inode, file_offset, disk_bytenr,
-					  num_bytes, disk_num_bytes, type, 0,
-					  BTRFS_COMPRESS_NONE);
-}
-
-int btrfs_add_ordered_extent_dio(struct inode *inode, u64 file_offset,
-				 u64 disk_bytenr, u64 num_bytes,
-				 u64 disk_num_bytes, int type)
-{
-	return __btrfs_add_ordered_extent(inode, file_offset, disk_bytenr,
-					  num_bytes, disk_num_bytes, type, 1,
-					  BTRFS_COMPRESS_NONE);
-}
-
-int btrfs_add_ordered_extent_compress(struct inode *inode, u64 file_offset,
-				      u64 disk_bytenr, u64 num_bytes,
-				      u64 disk_num_bytes, int type,
-				      int compress_type)
-{
-	return __btrfs_add_ordered_extent(inode, file_offset, disk_bytenr,
-					  num_bytes, disk_num_bytes, type, 0,
-					  compress_type);
-}
-
 /*
  * Add a struct btrfs_ordered_sum into the list of checksums to be inserted
  * when an ordered extent is finished.  If the list covers more than one
@@ -279,100 +272,178 @@ void btrfs_add_ordered_sum(struct btrfs_ordered_extent *entry,
 	spin_unlock_irq(&tree->lock);
 }
 
+static void finish_ordered_fn(struct btrfs_work *work)
+{
+	struct btrfs_ordered_extent *ordered_extent;
+
+	ordered_extent = container_of(work, struct btrfs_ordered_extent, work);
+	btrfs_finish_ordered_io(ordered_extent);
+}
+
 /*
- * this is used to account for finished IO across a given range
- * of the file.  The IO may span ordered extents.  If
- * a given ordered_extent is completely done, 1 is returned, otherwise
- * 0.
+ * Mark all ordered extents io inside the specified range finished.
  *
- * test_and_set_bit on a flag in the struct btrfs_ordered_extent is used
- * to make sure this function only returns 1 once for a given ordered extent.
+ * @page:	 The involved page for the operation.
+ *		 For uncompressed buffered IO, the page status also needs to be
+ *		 updated to indicate whether the pending ordered io is finished.
+ *		 Can be NULL for direct IO and compressed write.
+ *		 For these cases, callers are ensured they won't execute the
+ *		 endio function twice.
  *
- * file_offset is updated to one byte past the range that is recorded as
- * complete.  This allows you to walk forward in the file.
+ * This function is called for endio, thus the range must have ordered
+ * extent(s) covering it.
  */
-int btrfs_dec_test_first_ordered_pending(struct inode *inode,
-				   struct btrfs_ordered_extent **cached,
-				   u64 *file_offset, u64 io_size, int uptodate)
+void btrfs_mark_ordered_io_finished(struct btrfs_inode *inode,
+				    struct page *page, u64 file_offset,
+				    u64 num_bytes, bool uptodate)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	struct btrfs_ordered_inode_tree *tree;
+	struct btrfs_ordered_inode_tree *tree = &inode->ordered_tree;
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
+	struct btrfs_workqueue *wq;
 	struct rb_node *node;
 	struct btrfs_ordered_extent *entry = NULL;
-	int ret;
 	unsigned long flags;
-	u64 dec_end;
-	u64 dec_start;
-	u64 to_dec;
+	u64 cur = file_offset;
+
+	if (btrfs_is_free_space_inode(inode))
+		wq = fs_info->endio_freespace_worker;
+	else
+		wq = fs_info->endio_write_workers;
+
+	if (page)
+		ASSERT(page->mapping && page_offset(page) <= file_offset &&
+		       file_offset + num_bytes <= page_offset(page) + PAGE_SIZE);
 
-	tree = &BTRFS_I(inode)->ordered_tree;
 	spin_lock_irqsave(&tree->lock, flags);
-	node = tree_search(tree, *file_offset);
-	if (!node) {
-		ret = 1;
-		goto out;
-	}
+	while (cur < file_offset + num_bytes) {
+		u64 entry_end;
+		u64 end;
+		u32 len;
 
-	entry = rb_entry(node, struct btrfs_ordered_extent, rb_node);
-	if (!offset_in_entry(entry, *file_offset)) {
-		ret = 1;
-		goto out;
-	}
+		node = tree_search(tree, cur);
+		/* No ordered extents at all */
+		if (!node)
+			break;
 
-	dec_start = max(*file_offset, entry->file_offset);
-	dec_end = min(*file_offset + io_size,
-		      entry->file_offset + entry->num_bytes);
-	*file_offset = dec_end;
-	if (dec_start > dec_end) {
-		btrfs_crit(fs_info, "bad ordering dec_start %llu end %llu",
-			   dec_start, dec_end);
-	}
-	to_dec = dec_end - dec_start;
-	if (to_dec > entry->bytes_left) {
-		btrfs_crit(fs_info,
-			   "bad ordered accounting left %llu size %llu",
-			   entry->bytes_left, to_dec);
-	}
-	entry->bytes_left -= to_dec;
-	if (!uptodate)
-		set_bit(BTRFS_ORDERED_IOERR, &entry->flags);
+		entry = rb_entry(node, struct btrfs_ordered_extent, rb_node);
+		entry_end = entry->file_offset + entry->num_bytes;
+		/*
+		 * |<-- OE --->|  |
+		 *		  cur
+		 * Go to next OE.
+		 */
+		if (cur >= entry_end) {
+			node = rb_next(node);
+			/* No more ordered extents, exit */
+			if (!node)
+				break;
+			entry = rb_entry(node, struct btrfs_ordered_extent,
+					 rb_node);
+
+			/* Go to next ordered extent and continue */
+			cur = entry->file_offset;
+			continue;
+		}
+		/*
+		 * |	|<--- OE --->|
+		 * cur
+		 * Go to the start of OE.
+		 */
+		if (cur < entry->file_offset) {
+			cur = entry->file_offset;
+			continue;
+		}
 
-	if (entry->bytes_left == 0) {
-		ret = test_and_set_bit(BTRFS_ORDERED_IO_DONE, &entry->flags);
-		/* test_and_set_bit implies a barrier */
-		cond_wake_up_nomb(&entry->wait);
-	} else {
-		ret = 1;
-	}
-out:
-	if (!ret && cached && entry) {
-		*cached = entry;
-		refcount_inc(&entry->refs);
+		/*
+		 * Now we are definitely inside one ordered extent.
+		 *
+		 * |<--- OE --->|
+		 *	|
+		 *	cur
+		 */
+		end = min(entry->file_offset + entry->num_bytes,
+			  file_offset + num_bytes) - 1;
+		ASSERT(end + 1 - cur < U32_MAX);
+		len = end + 1 - cur;
+
+		if (page) {
+			/*
+			 * Ordered (Private2) bit indicates whether we still
+			 * have pending io unfinished for the ordered extent.
+			 *
+			 * If there's no such bit, we need to skip to next range.
+			 */
+			if (!btrfs_page_test_ordered(fs_info, page, cur, len)) {
+				cur += len;
+				continue;
+			}
+			btrfs_page_clear_ordered(fs_info, page, cur, len);
+		}
+
+		/* Now we're fine to update the accounting */
+		if (unlikely(len > entry->bytes_left)) {
+			WARN_ON(1);
+			btrfs_crit(fs_info,
+"bad ordered extent accounting, root=%llu ino=%llu OE offset=%llu OE len=%llu to_dec=%u left=%llu",
+				   inode->root->root_key.objectid,
+				   btrfs_ino(inode),
+				   entry->file_offset,
+				   entry->num_bytes,
+				   len, entry->bytes_left);
+			entry->bytes_left = 0;
+		} else {
+			entry->bytes_left -= len;
+		}
+
+		if (!uptodate)
+			set_bit(BTRFS_ORDERED_IOERR, &entry->flags);
+
+		/*
+		 * All the IO of the ordered extent is finished, we need to queue
+		 * the finish_func to be executed.
+		 */
+		if (entry->bytes_left == 0) {
+			set_bit(BTRFS_ORDERED_IO_DONE, &entry->flags);
+			cond_wake_up(&entry->wait);
+			refcount_inc(&entry->refs);
+			trace_btrfs_ordered_extent_mark_finished(inode, entry);
+			spin_unlock_irqrestore(&tree->lock, flags);
+			btrfs_init_work(&entry->work, finish_ordered_fn, NULL, NULL);
+			btrfs_queue_work(wq, &entry->work);
+			spin_lock_irqsave(&tree->lock, flags);
+		}
+		cur += len;
 	}
 	spin_unlock_irqrestore(&tree->lock, flags);
-	return ret == 0;
 }
 
 /*
- * this is used to account for finished IO across a given range
- * of the file.  The IO should not span ordered extents.  If
- * a given ordered_extent is completely done, 1 is returned, otherwise
- * 0.
+ * Finish IO for one ordered extent across a given range.  The range can only
+ * contain one ordered extent.
  *
- * test_and_set_bit on a flag in the struct btrfs_ordered_extent is used
- * to make sure this function only returns 1 once for a given ordered extent.
+ * @cached:	 The cached ordered extent. If not NULL, we can skip the tree
+ *               search and use the ordered extent directly.
+ * 		 Will be also used to store the finished ordered extent.
+ * @file_offset: File offset for the finished IO
+ * @io_size:	 Length of the finish IO range
+ *
+ * Return true if the ordered extent is finished in the range, and update
+ * @cached.
+ * Return false otherwise.
+ *
+ * NOTE: The range can NOT cross multiple ordered extents.
+ * Thus caller should ensure the range doesn't cross ordered extents.
  */
-int btrfs_dec_test_ordered_pending(struct inode *inode,
-				   struct btrfs_ordered_extent **cached,
-				   u64 file_offset, u64 io_size, int uptodate)
+bool btrfs_dec_test_ordered_pending(struct btrfs_inode *inode,
+				    struct btrfs_ordered_extent **cached,
+				    u64 file_offset, u64 io_size)
 {
-	struct btrfs_ordered_inode_tree *tree;
+	struct btrfs_ordered_inode_tree *tree = &inode->ordered_tree;
 	struct rb_node *node;
 	struct btrfs_ordered_extent *entry = NULL;
 	unsigned long flags;
-	int ret;
+	bool finished = false;
 
-	tree = &BTRFS_I(inode)->ordered_tree;
 	spin_lock_irqsave(&tree->lock, flags);
 	if (cached && *cached) {
 		entry = *cached;
@@ -380,41 +451,38 @@ int btrfs_dec_test_ordered_pending(struct inode *inode,
 	}
 
 	node = tree_search(tree, file_offset);
-	if (!node) {
-		ret = 1;
+	if (!node)
 		goto out;
-	}
 
 	entry = rb_entry(node, struct btrfs_ordered_extent, rb_node);
 have_entry:
-	if (!offset_in_entry(entry, file_offset)) {
-		ret = 1;
+	if (!in_range(file_offset, entry->file_offset, entry->num_bytes))
 		goto out;
-	}
 
-	if (io_size > entry->bytes_left) {
-		btrfs_crit(BTRFS_I(inode)->root->fs_info,
+	if (io_size > entry->bytes_left)
+		btrfs_crit(inode->root->fs_info,
 			   "bad ordered accounting left %llu size %llu",
 		       entry->bytes_left, io_size);
-	}
+
 	entry->bytes_left -= io_size;
-	if (!uptodate)
-		set_bit(BTRFS_ORDERED_IOERR, &entry->flags);
 
 	if (entry->bytes_left == 0) {
-		ret = test_and_set_bit(BTRFS_ORDERED_IO_DONE, &entry->flags);
+		/*
+		 * Ensure only one caller can set the flag and finished_ret
+		 * accordingly
+		 */
+		finished = !test_and_set_bit(BTRFS_ORDERED_IO_DONE, &entry->flags);
 		/* test_and_set_bit implies a barrier */
 		cond_wake_up_nomb(&entry->wait);
-	} else {
-		ret = 1;
 	}
 out:
-	if (!ret && cached && entry) {
+	if (finished && cached && entry) {
 		*cached = entry;
 		refcount_inc(&entry->refs);
+		trace_btrfs_ordered_extent_dec_test_pending(inode, entry);
 	}
 	spin_unlock_irqrestore(&tree->lock, flags);
-	return ret == 0;
+	return finished;
 }
 
 /*
@@ -426,12 +494,11 @@ void btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry)
 	struct list_head *cur;
 	struct btrfs_ordered_sum *sum;
 
-	trace_btrfs_ordered_extent_put(entry->inode, entry);
+	trace_btrfs_ordered_extent_put(BTRFS_I(entry->inode), entry);
 
 	if (refcount_dec_and_test(&entry->refs)) {
-		ASSERT(list_empty(&entry->log_list));
-		ASSERT(list_empty(&entry->trans_list));
 		ASSERT(list_empty(&entry->root_extent_list));
+		ASSERT(list_empty(&entry->log_list));
 		ASSERT(RB_EMPTY_NODE(&entry->rb_node));
 		if (entry->inode)
 			btrfs_add_delayed_iput(entry->inode);
@@ -449,26 +516,39 @@ void btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry)
  * remove an ordered extent from the tree.  No references are dropped
  * and waiters are woken up.
  */
-void btrfs_remove_ordered_extent(struct inode *inode,
+void btrfs_remove_ordered_extent(struct btrfs_inode *btrfs_inode,
 				 struct btrfs_ordered_extent *entry)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	struct btrfs_ordered_inode_tree *tree;
-	struct btrfs_inode *btrfs_inode = BTRFS_I(inode);
 	struct btrfs_root *root = btrfs_inode->root;
+	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct rb_node *node;
+	bool pending;
+	bool freespace_inode;
+
+	/*
+	 * If this is a free space inode the thread has not acquired the ordered
+	 * extents lockdep map.
+	 */
+	freespace_inode = btrfs_is_free_space_inode(btrfs_inode);
 
+	btrfs_lockdep_acquire(fs_info, btrfs_trans_pending_ordered);
 	/* This is paired with btrfs_add_ordered_extent. */
 	spin_lock(&btrfs_inode->lock);
 	btrfs_mod_outstanding_extents(btrfs_inode, -1);
 	spin_unlock(&btrfs_inode->lock);
-	if (root != fs_info->tree_root)
-		btrfs_delalloc_release_metadata(btrfs_inode, entry->num_bytes,
-						false);
+	if (root != fs_info->tree_root) {
+		u64 release;
 
-	if (test_bit(BTRFS_ORDERED_DIRECT, &entry->flags))
-		percpu_counter_add_batch(&fs_info->dio_bytes, -entry->num_bytes,
-					 fs_info->delalloc_batch);
+		if (test_bit(BTRFS_ORDERED_ENCODED, &entry->flags))
+			release = entry->disk_num_bytes;
+		else
+			release = entry->num_bytes;
+		btrfs_delalloc_release_metadata(btrfs_inode, release, false);
+	}
+
+	percpu_counter_add_batch(&fs_info->ordered_bytes, -entry->num_bytes,
+				 fs_info->delalloc_batch);
 
 	tree = &btrfs_inode->ordered_tree;
 	spin_lock_irq(&tree->lock);
@@ -478,13 +558,43 @@ void btrfs_remove_ordered_extent(struct inode *inode,
 	if (tree->last == node)
 		tree->last = NULL;
 	set_bit(BTRFS_ORDERED_COMPLETE, &entry->flags);
+	pending = test_and_clear_bit(BTRFS_ORDERED_PENDING, &entry->flags);
 	spin_unlock_irq(&tree->lock);
 
+	/*
+	 * The current running transaction is waiting on us, we need to let it
+	 * know that we're complete and wake it up.
+	 */
+	if (pending) {
+		struct btrfs_transaction *trans;
+
+		/*
+		 * The checks for trans are just a formality, it should be set,
+		 * but if it isn't we don't want to deref/assert under the spin
+		 * lock, so be nice and check if trans is set, but ASSERT() so
+		 * if it isn't set a developer will notice.
+		 */
+		spin_lock(&fs_info->trans_lock);
+		trans = fs_info->running_transaction;
+		if (trans)
+			refcount_inc(&trans->use_count);
+		spin_unlock(&fs_info->trans_lock);
+
+		ASSERT(trans);
+		if (trans) {
+			if (atomic_dec_and_test(&trans->pending_ordered))
+				wake_up(&trans->pending_wait);
+			btrfs_put_transaction(trans);
+		}
+	}
+
+	btrfs_lockdep_release(fs_info, btrfs_trans_pending_ordered);
+
 	spin_lock(&root->ordered_extent_lock);
 	list_del_init(&entry->root_extent_list);
 	root->nr_ordered_extents--;
 
-	trace_btrfs_ordered_extent_remove(inode, entry);
+	trace_btrfs_ordered_extent_remove(btrfs_inode, entry);
 
 	if (!root->nr_ordered_extents) {
 		spin_lock(&fs_info->ordered_root_lock);
@@ -494,6 +604,8 @@ void btrfs_remove_ordered_extent(struct inode *inode,
 	}
 	spin_unlock(&root->ordered_extent_lock);
 	wake_up(&entry->wait);
+	if (!freespace_inode)
+		btrfs_lockdep_release(fs_info, btrfs_ordered_extent);
 }
 
 static void btrfs_run_ordered_extent_work(struct btrfs_work *work)
@@ -501,7 +613,7 @@ static void btrfs_run_ordered_extent_work(struct btrfs_work *work)
 	struct btrfs_ordered_extent *ordered;
 
 	ordered = container_of(work, struct btrfs_ordered_extent, flush_work);
-	btrfs_start_ordered_extent(ordered->inode, ordered, 1);
+	btrfs_start_ordered_extent(ordered, 1);
 	complete(&ordered->completion);
 }
 
@@ -580,7 +692,7 @@ void btrfs_wait_ordered_roots(struct btrfs_fs_info *fs_info, u64 nr,
 	while (!list_empty(&splice) && nr) {
 		root = list_first_entry(&splice, struct btrfs_root,
 					ordered_root);
-		root = btrfs_grab_fs_root(root);
+		root = btrfs_grab_root(root);
 		BUG_ON(!root);
 		list_move_tail(&root->ordered_root,
 			       &fs_info->ordered_roots);
@@ -588,7 +700,7 @@ void btrfs_wait_ordered_roots(struct btrfs_fs_info *fs_info, u64 nr,
 
 		done = btrfs_wait_ordered_extents(root, nr,
 						  range_start, range_len);
-		btrfs_put_fs_root(root);
+		btrfs_put_root(root);
 
 		spin_lock(&fs_info->ordered_root_lock);
 		if (nr != U64_MAX) {
@@ -607,23 +719,31 @@ void btrfs_wait_ordered_roots(struct btrfs_fs_info *fs_info, u64 nr,
  * in the extent, and it waits on the io completion code to insert
  * metadata into the btree corresponding to the extent
  */
-void btrfs_start_ordered_extent(struct inode *inode,
-				       struct btrfs_ordered_extent *entry,
-				       int wait)
+void btrfs_start_ordered_extent(struct btrfs_ordered_extent *entry, int wait)
 {
 	u64 start = entry->file_offset;
 	u64 end = start + entry->num_bytes - 1;
+	struct btrfs_inode *inode = BTRFS_I(entry->inode);
+	bool freespace_inode;
 
 	trace_btrfs_ordered_extent_start(inode, entry);
 
 	/*
+	 * If this is a free space inode do not take the ordered extents lockdep
+	 * map.
+	 */
+	freespace_inode = btrfs_is_free_space_inode(inode);
+
+	/*
 	 * pages in the range can be dirty, clean or writeback.  We
 	 * start IO on any dirty ones so the wait doesn't stall waiting
 	 * for the flusher thread to find them
 	 */
 	if (!test_bit(BTRFS_ORDERED_DIRECT, &entry->flags))
-		filemap_fdatawrite_range(inode->i_mapping, start, end);
+		filemap_fdatawrite_range(inode->vfs_inode.i_mapping, start, end);
 	if (wait) {
+		if (!freespace_inode)
+			btrfs_might_wait_for_event(inode->root->fs_info, btrfs_ordered_extent);
 		wait_event(entry->wait, test_bit(BTRFS_ORDERED_COMPLETE,
 						 &entry->flags));
 	}
@@ -666,7 +786,7 @@ int btrfs_wait_ordered_range(struct inode *inode, u64 start, u64 len)
 
 	end = orig_end;
 	while (1) {
-		ordered = btrfs_lookup_first_ordered_extent(inode, end);
+		ordered = btrfs_lookup_first_ordered_extent(BTRFS_I(inode), end);
 		if (!ordered)
 			break;
 		if (ordered->file_offset > orig_end) {
@@ -677,7 +797,7 @@ int btrfs_wait_ordered_range(struct inode *inode, u64 start, u64 len)
 			btrfs_put_ordered_extent(ordered);
 			break;
 		}
-		btrfs_start_ordered_extent(inode, ordered, 1);
+		btrfs_start_ordered_extent(ordered, 1);
 		end = ordered->file_offset;
 		/*
 		 * If the ordered extent had an error save the error but don't
@@ -698,26 +818,29 @@ int btrfs_wait_ordered_range(struct inode *inode, u64 start, u64 len)
  * find an ordered extent corresponding to file_offset.  return NULL if
  * nothing is found, otherwise take a reference on the extent and return it
  */
-struct btrfs_ordered_extent *btrfs_lookup_ordered_extent(struct inode *inode,
+struct btrfs_ordered_extent *btrfs_lookup_ordered_extent(struct btrfs_inode *inode,
 							 u64 file_offset)
 {
 	struct btrfs_ordered_inode_tree *tree;
 	struct rb_node *node;
 	struct btrfs_ordered_extent *entry = NULL;
+	unsigned long flags;
 
-	tree = &BTRFS_I(inode)->ordered_tree;
-	spin_lock_irq(&tree->lock);
+	tree = &inode->ordered_tree;
+	spin_lock_irqsave(&tree->lock, flags);
 	node = tree_search(tree, file_offset);
 	if (!node)
 		goto out;
 
 	entry = rb_entry(node, struct btrfs_ordered_extent, rb_node);
-	if (!offset_in_entry(entry, file_offset))
+	if (!in_range(file_offset, entry->file_offset, entry->num_bytes))
 		entry = NULL;
-	if (entry)
+	if (entry) {
 		refcount_inc(&entry->refs);
+		trace_btrfs_ordered_extent_lookup(inode, entry);
+	}
 out:
-	spin_unlock_irq(&tree->lock);
+	spin_unlock_irqrestore(&tree->lock, flags);
 	return entry;
 }
 
@@ -755,24 +878,55 @@ struct btrfs_ordered_extent *btrfs_lookup_ordered_range(
 			break;
 	}
 out:
-	if (entry)
+	if (entry) {
 		refcount_inc(&entry->refs);
+		trace_btrfs_ordered_extent_lookup_range(inode, entry);
+	}
 	spin_unlock_irq(&tree->lock);
 	return entry;
 }
 
 /*
+ * Adds all ordered extents to the given list. The list ends up sorted by the
+ * file_offset of the ordered extents.
+ */
+void btrfs_get_ordered_extents_for_logging(struct btrfs_inode *inode,
+					   struct list_head *list)
+{
+	struct btrfs_ordered_inode_tree *tree = &inode->ordered_tree;
+	struct rb_node *n;
+
+	ASSERT(inode_is_locked(&inode->vfs_inode));
+
+	spin_lock_irq(&tree->lock);
+	for (n = rb_first(&tree->tree); n; n = rb_next(n)) {
+		struct btrfs_ordered_extent *ordered;
+
+		ordered = rb_entry(n, struct btrfs_ordered_extent, rb_node);
+
+		if (test_bit(BTRFS_ORDERED_LOGGED, &ordered->flags))
+			continue;
+
+		ASSERT(list_empty(&ordered->log_list));
+		list_add_tail(&ordered->log_list, list);
+		refcount_inc(&ordered->refs);
+		trace_btrfs_ordered_extent_lookup_for_logging(inode, ordered);
+	}
+	spin_unlock_irq(&tree->lock);
+}
+
+/*
  * lookup and return any extent before 'file_offset'.  NULL is returned
  * if none is found
  */
 struct btrfs_ordered_extent *
-btrfs_lookup_first_ordered_extent(struct inode *inode, u64 file_offset)
+btrfs_lookup_first_ordered_extent(struct btrfs_inode *inode, u64 file_offset)
 {
 	struct btrfs_ordered_inode_tree *tree;
 	struct rb_node *node;
 	struct btrfs_ordered_extent *entry = NULL;
 
-	tree = &BTRFS_I(inode)->ordered_tree;
+	tree = &inode->ordered_tree;
 	spin_lock_irq(&tree->lock);
 	node = tree_search(tree, file_offset);
 	if (!node)
@@ -780,190 +934,94 @@ btrfs_lookup_first_ordered_extent(struct inode *inode, u64 file_offset)
 
 	entry = rb_entry(node, struct btrfs_ordered_extent, rb_node);
 	refcount_inc(&entry->refs);
+	trace_btrfs_ordered_extent_lookup_first(inode, entry);
 out:
 	spin_unlock_irq(&tree->lock);
 	return entry;
 }
 
 /*
- * After an extent is done, call this to conditionally update the on disk
- * i_size.  i_size is updated to cover any fully written part of the file.
+ * Lookup the first ordered extent that overlaps the range
+ * [@file_offset, @file_offset + @len).
+ *
+ * The difference between this and btrfs_lookup_first_ordered_extent() is
+ * that this one won't return any ordered extent that does not overlap the range.
+ * And the difference against btrfs_lookup_ordered_extent() is, this function
+ * ensures the first ordered extent gets returned.
  */
-int btrfs_ordered_update_i_size(struct inode *inode, u64 offset,
-				struct btrfs_ordered_extent *ordered)
+struct btrfs_ordered_extent *btrfs_lookup_first_ordered_range(
+			struct btrfs_inode *inode, u64 file_offset, u64 len)
 {
-	struct btrfs_ordered_inode_tree *tree = &BTRFS_I(inode)->ordered_tree;
-	u64 disk_i_size;
-	u64 new_i_size;
-	u64 i_size = i_size_read(inode);
+	struct btrfs_ordered_inode_tree *tree = &inode->ordered_tree;
 	struct rb_node *node;
-	struct rb_node *prev = NULL;
-	struct btrfs_ordered_extent *test;
-	int ret = 1;
-	u64 orig_offset = offset;
+	struct rb_node *cur;
+	struct rb_node *prev;
+	struct rb_node *next;
+	struct btrfs_ordered_extent *entry = NULL;
 
 	spin_lock_irq(&tree->lock);
-	if (ordered) {
-		offset = entry_end(ordered);
-		if (test_bit(BTRFS_ORDERED_TRUNCATED, &ordered->flags))
-			offset = min(offset,
-				     ordered->file_offset +
-				     ordered->truncated_len);
-	} else {
-		offset = ALIGN(offset, btrfs_inode_sectorsize(inode));
-	}
-	disk_i_size = BTRFS_I(inode)->disk_i_size;
-
-	/*
-	 * truncate file.
-	 * If ordered is not NULL, then this is called from endio and
-	 * disk_i_size will be updated by either truncate itself or any
-	 * in-flight IOs which are inside the disk_i_size.
-	 *
-	 * Because btrfs_setsize() may set i_size with disk_i_size if truncate
-	 * fails somehow, we need to make sure we have a precise disk_i_size by
-	 * updating it as usual.
-	 *
-	 */
-	if (!ordered && disk_i_size > i_size) {
-		BTRFS_I(inode)->disk_i_size = orig_offset;
-		ret = 0;
-		goto out;
-	}
-
-	/*
-	 * if the disk i_size is already at the inode->i_size, or
-	 * this ordered extent is inside the disk i_size, we're done
-	 */
-	if (disk_i_size == i_size)
-		goto out;
-
+	node = tree->tree.rb_node;
 	/*
-	 * We still need to update disk_i_size if outstanding_isize is greater
-	 * than disk_i_size.
+	 * Here we don't want to use tree_search() which will use tree->last
+	 * and screw up the search order.
+	 * And __tree_search() can't return the adjacent ordered extents
+	 * either, thus here we do our own search.
 	 */
-	if (offset <= disk_i_size &&
-	    (!ordered || ordered->outstanding_isize <= disk_i_size))
-		goto out;
+	while (node) {
+		entry = rb_entry(node, struct btrfs_ordered_extent, rb_node);
 
-	/*
-	 * walk backward from this ordered extent to disk_i_size.
-	 * if we find an ordered extent then we can't update disk i_size
-	 * yet
-	 */
-	if (ordered) {
-		node = rb_prev(&ordered->rb_node);
-	} else {
-		prev = tree_search(tree, offset);
-		/*
-		 * we insert file extents without involving ordered struct,
-		 * so there should be no ordered struct cover this offset
-		 */
-		if (prev) {
-			test = rb_entry(prev, struct btrfs_ordered_extent,
-					rb_node);
-			BUG_ON(offset_in_entry(test, offset));
+		if (file_offset < entry->file_offset) {
+			node = node->rb_left;
+		} else if (file_offset >= entry_end(entry)) {
+			node = node->rb_right;
+		} else {
+			/*
+			 * Direct hit, got an ordered extent that starts at
+			 * @file_offset
+			 */
+			goto out;
 		}
-		node = prev;
 	}
-	for (; node; node = rb_prev(node)) {
-		test = rb_entry(node, struct btrfs_ordered_extent, rb_node);
-
-		/* We treat this entry as if it doesn't exist */
-		if (test_bit(BTRFS_ORDERED_UPDATED_ISIZE, &test->flags))
-			continue;
-
-		if (entry_end(test) <= disk_i_size)
-			break;
-		if (test->file_offset >= i_size)
-			break;
-
-		/*
-		 * We don't update disk_i_size now, so record this undealt
-		 * i_size. Or we will not know the real i_size.
-		 */
-		if (test->outstanding_isize < offset)
-			test->outstanding_isize = offset;
-		if (ordered &&
-		    ordered->outstanding_isize > test->outstanding_isize)
-			test->outstanding_isize = ordered->outstanding_isize;
+	if (!entry) {
+		/* Empty tree */
 		goto out;
 	}
-	new_i_size = min_t(u64, offset, i_size);
-
-	/*
-	 * Some ordered extents may completed before the current one, and
-	 * we hold the real i_size in ->outstanding_isize.
-	 */
-	if (ordered && ordered->outstanding_isize > new_i_size)
-		new_i_size = min_t(u64, ordered->outstanding_isize, i_size);
-	BTRFS_I(inode)->disk_i_size = new_i_size;
-	ret = 0;
-out:
-	/*
-	 * We need to do this because we can't remove ordered extents until
-	 * after the i_disk_size has been updated and then the inode has been
-	 * updated to reflect the change, so we need to tell anybody who finds
-	 * this ordered extent that we've already done all the real work, we
-	 * just haven't completed all the other work.
-	 */
-	if (ordered)
-		set_bit(BTRFS_ORDERED_UPDATED_ISIZE, &ordered->flags);
-	spin_unlock_irq(&tree->lock);
-	return ret;
-}
 
-/*
- * search the ordered extents for one corresponding to 'offset' and
- * try to find a checksum.  This is used because we allow pages to
- * be reclaimed before their checksum is actually put into the btree
- */
-int btrfs_find_ordered_sum(struct inode *inode, u64 offset, u64 disk_bytenr,
-			   u8 *sum, int len)
-{
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	struct btrfs_ordered_sum *ordered_sum;
-	struct btrfs_ordered_extent *ordered;
-	struct btrfs_ordered_inode_tree *tree = &BTRFS_I(inode)->ordered_tree;
-	unsigned long num_sectors;
-	unsigned long i;
-	u32 sectorsize = btrfs_inode_sectorsize(inode);
-	const u16 csum_size = btrfs_super_csum_size(fs_info->super_copy);
-	int index = 0;
-
-	ordered = btrfs_lookup_ordered_extent(inode, offset);
-	if (!ordered)
-		return 0;
-
-	spin_lock_irq(&tree->lock);
-	list_for_each_entry_reverse(ordered_sum, &ordered->list, list) {
-		if (disk_bytenr >= ordered_sum->bytenr &&
-		    disk_bytenr < ordered_sum->bytenr + ordered_sum->len) {
-			i = (disk_bytenr - ordered_sum->bytenr) >>
-			    inode->i_sb->s_blocksize_bits;
-			num_sectors = ordered_sum->len >>
-				      inode->i_sb->s_blocksize_bits;
-			num_sectors = min_t(int, len - index, num_sectors - i);
-			memcpy(sum + index, ordered_sum->sums + i * csum_size,
-			       num_sectors * csum_size);
-
-			index += (int)num_sectors * csum_size;
-			if (index == len)
-				goto out;
-			disk_bytenr += num_sectors * sectorsize;
-		}
+	cur = &entry->rb_node;
+	/* We got an entry around @file_offset, check adjacent entries */
+	if (entry->file_offset < file_offset) {
+		prev = cur;
+		next = rb_next(cur);
+	} else {
+		prev = rb_prev(cur);
+		next = cur;
+	}
+	if (prev) {
+		entry = rb_entry(prev, struct btrfs_ordered_extent, rb_node);
+		if (range_overlaps(entry, file_offset, len))
+			goto out;
 	}
+	if (next) {
+		entry = rb_entry(next, struct btrfs_ordered_extent, rb_node);
+		if (range_overlaps(entry, file_offset, len))
+			goto out;
+	}
+	/* No ordered extent in the range */
+	entry = NULL;
 out:
+	if (entry) {
+		refcount_inc(&entry->refs);
+		trace_btrfs_ordered_extent_lookup_first_range(inode, entry);
+	}
+
 	spin_unlock_irq(&tree->lock);
-	btrfs_put_ordered_extent(ordered);
-	return index;
+	return entry;
 }
 
 /*
  * btrfs_flush_ordered_range - Lock the passed range and ensures all pending
  * ordered extents in it are run to completion.
  *
- * @tree:         IO tree used for locking out other users of the range
  * @inode:        Inode whose ordered tree is to be searched
  * @start:        Beginning of range to flush
  * @end:          Last byte of range to lock
@@ -973,8 +1031,7 @@ out:
  * This function always returns with the given range locked, ensuring after it's
  * called no order extent can be pending.
  */
-void btrfs_lock_and_flush_ordered_range(struct extent_io_tree *tree,
-					struct btrfs_inode *inode, u64 start,
+void btrfs_lock_and_flush_ordered_range(struct btrfs_inode *inode, u64 start,
 					u64 end,
 					struct extent_state **cached_state)
 {
@@ -986,7 +1043,7 @@ void btrfs_lock_and_flush_ordered_range(struct extent_io_tree *tree,
 		cachedp = cached_state;
 
 	while (1) {
-		lock_extent_bits(tree, start, end, cachedp);
+		lock_extent(&inode->io_tree, start, end, cachedp);
 		ordered = btrfs_lookup_ordered_range(inode, start,
 						     end - start + 1);
 		if (!ordered) {
@@ -999,12 +1056,101 @@ void btrfs_lock_and_flush_ordered_range(struct extent_io_tree *tree,
 				refcount_dec(&cache->refs);
 			break;
 		}
-		unlock_extent_cached(tree, start, end, cachedp);
-		btrfs_start_ordered_extent(&inode->vfs_inode, ordered, 1);
+		unlock_extent(&inode->io_tree, start, end, cachedp);
+		btrfs_start_ordered_extent(ordered, 1);
 		btrfs_put_ordered_extent(ordered);
 	}
 }
 
+/*
+ * Lock the passed range and ensure all pending ordered extents in it are run
+ * to completion in nowait mode.
+ *
+ * Return true if btrfs_lock_ordered_range does not return any extents,
+ * otherwise false.
+ */
+bool btrfs_try_lock_ordered_range(struct btrfs_inode *inode, u64 start, u64 end)
+{
+	struct btrfs_ordered_extent *ordered;
+
+	if (!try_lock_extent(&inode->io_tree, start, end))
+		return false;
+
+	ordered = btrfs_lookup_ordered_range(inode, start, end - start + 1);
+	if (!ordered)
+		return true;
+
+	btrfs_put_ordered_extent(ordered);
+	unlock_extent(&inode->io_tree, start, end, NULL);
+
+	return false;
+}
+
+
+static int clone_ordered_extent(struct btrfs_ordered_extent *ordered, u64 pos,
+				u64 len)
+{
+	struct inode *inode = ordered->inode;
+	struct btrfs_fs_info *fs_info = BTRFS_I(inode)->root->fs_info;
+	u64 file_offset = ordered->file_offset + pos;
+	u64 disk_bytenr = ordered->disk_bytenr + pos;
+	unsigned long flags = ordered->flags & BTRFS_ORDERED_TYPE_FLAGS;
+
+	/*
+	 * The splitting extent is already counted and will be added again in
+	 * btrfs_add_ordered_extent_*(). Subtract len to avoid double counting.
+	 */
+	percpu_counter_add_batch(&fs_info->ordered_bytes, -len,
+				 fs_info->delalloc_batch);
+	WARN_ON_ONCE(flags & (1 << BTRFS_ORDERED_COMPRESSED));
+	return btrfs_add_ordered_extent(BTRFS_I(inode), file_offset, len, len,
+					disk_bytenr, len, 0, flags,
+					ordered->compress_type);
+}
+
+int btrfs_split_ordered_extent(struct btrfs_ordered_extent *ordered, u64 pre,
+				u64 post)
+{
+	struct inode *inode = ordered->inode;
+	struct btrfs_ordered_inode_tree *tree = &BTRFS_I(inode)->ordered_tree;
+	struct rb_node *node;
+	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+	int ret = 0;
+
+	trace_btrfs_ordered_extent_split(BTRFS_I(inode), ordered);
+
+	spin_lock_irq(&tree->lock);
+	/* Remove from tree once */
+	node = &ordered->rb_node;
+	rb_erase(node, &tree->tree);
+	RB_CLEAR_NODE(node);
+	if (tree->last == node)
+		tree->last = NULL;
+
+	ordered->file_offset += pre;
+	ordered->disk_bytenr += pre;
+	ordered->num_bytes -= (pre + post);
+	ordered->disk_num_bytes -= (pre + post);
+	ordered->bytes_left -= (pre + post);
+
+	/* Re-insert the node */
+	node = tree_insert(&tree->tree, ordered->file_offset, &ordered->rb_node);
+	if (node)
+		btrfs_panic(fs_info, -EEXIST,
+			"zoned: inconsistency in ordered tree at offset %llu",
+			    ordered->file_offset);
+
+	spin_unlock_irq(&tree->lock);
+
+	if (pre)
+		ret = clone_ordered_extent(ordered, 0, pre);
+	if (ret == 0 && post)
+		ret = clone_ordered_extent(ordered, pre + ordered->disk_num_bytes,
+					   post);
+
+	return ret;
+}
+
 int __init ordered_data_init(void)
 {
 	btrfs_ordered_extent_cache = kmem_cache_create("btrfs_ordered_extent",
diff --git a/fs/btrfs/ordered-data.h b/fs/btrfs/ordered-data.h
index 3beb4da4ab41..f59f2dbdb25e 100644
--- a/fs/btrfs/ordered-data.h
+++ b/fs/btrfs/ordered-data.h
@@ -27,7 +27,7 @@ struct btrfs_ordered_sum {
 };
 
 /*
- * bits for the flags field:
+ * Bits for btrfs_ordered_extent::flags.
  *
  * BTRFS_ORDERED_IO_DONE is set when all of the blocks are written.
  * It is used to make sure metadata is inserted into the tree only once
@@ -38,31 +38,54 @@ struct btrfs_ordered_sum {
  * IO is done and any metadata is inserted into the tree.
  */
 enum {
+	/*
+	 * Different types for ordered extents, one and only one of the 4 types
+	 * need to be set when creating ordered extent.
+	 *
+	 * REGULAR:	For regular non-compressed COW write
+	 * NOCOW:	For NOCOW write into existing non-hole extent
+	 * PREALLOC:	For NOCOW write into preallocated extent
+	 * COMPRESSED:	For compressed COW write
+	 */
+	BTRFS_ORDERED_REGULAR,
+	BTRFS_ORDERED_NOCOW,
+	BTRFS_ORDERED_PREALLOC,
+	BTRFS_ORDERED_COMPRESSED,
+
+	/*
+	 * Extra bit for direct io, can only be set for
+	 * REGULAR/NOCOW/PREALLOC. No direct io for compressed extent.
+	 */
+	BTRFS_ORDERED_DIRECT,
+
+	/* Extra status bits for ordered extents */
+
 	/* set when all the pages are written */
 	BTRFS_ORDERED_IO_DONE,
 	/* set when removed from the tree */
 	BTRFS_ORDERED_COMPLETE,
-	/* set when we want to write in place */
-	BTRFS_ORDERED_NOCOW,
-	/* writing a zlib compressed extent */
-	BTRFS_ORDERED_COMPRESSED,
-	/* set when writing to preallocated extent */
-	BTRFS_ORDERED_PREALLOC,
-	/* set when we're doing DIO with this extent */
-	BTRFS_ORDERED_DIRECT,
 	/* We had an io error when writing this out */
 	BTRFS_ORDERED_IOERR,
-	/*
-	 * indicates whether this ordered extent has done its due diligence in
-	 * updating the isize
-	 */
-	BTRFS_ORDERED_UPDATED_ISIZE,
 	/* Set when we have to truncate an extent */
 	BTRFS_ORDERED_TRUNCATED,
-	/* Regular IO for COW */
-	BTRFS_ORDERED_REGULAR,
+	/* Used during fsync to track already logged extents */
+	BTRFS_ORDERED_LOGGED,
+	/* We have already logged all the csums of the ordered extent */
+	BTRFS_ORDERED_LOGGED_CSUM,
+	/* We wait for this extent to complete in the current transaction */
+	BTRFS_ORDERED_PENDING,
+	/* BTRFS_IOC_ENCODED_WRITE */
+	BTRFS_ORDERED_ENCODED,
 };
 
+/* BTRFS_ORDERED_* flags that specify the type of the extent. */
+#define BTRFS_ORDERED_TYPE_FLAGS ((1UL << BTRFS_ORDERED_REGULAR) |	\
+				  (1UL << BTRFS_ORDERED_NOCOW) |	\
+				  (1UL << BTRFS_ORDERED_PREALLOC) |	\
+				  (1UL << BTRFS_ORDERED_COMPRESSED) |	\
+				  (1UL << BTRFS_ORDERED_DIRECT) |	\
+				  (1UL << BTRFS_ORDERED_ENCODED))
+
 struct btrfs_ordered_extent {
 	/* logical offset in the file */
 	u64 file_offset;
@@ -71,9 +94,11 @@ struct btrfs_ordered_extent {
 	 * These fields directly correspond to the same fields in
 	 * btrfs_file_extent_item.
 	 */
-	u64 disk_bytenr;
 	u64 num_bytes;
+	u64 ram_bytes;
+	u64 disk_bytenr;
 	u64 disk_num_bytes;
+	u64 offset;
 
 	/* number of bytes that still need writing */
 	u64 bytes_left;
@@ -97,6 +122,9 @@ struct btrfs_ordered_extent {
 	/* compression algorithm */
 	int compress_type;
 
+	/* Qgroup reserved space */
+	int qgroup_rsv;
+
 	/* reference count */
 	refcount_t refs;
 
@@ -106,12 +134,9 @@ struct btrfs_ordered_extent {
 	/* list of checksums for insertion when the extent io is done */
 	struct list_head list;
 
-	/* If we need to wait on this to be done */
+	/* used for fast fsyncs */
 	struct list_head log_list;
 
-	/* If the transaction needs to wait on this ordered extent */
-	struct list_head trans_list;
-
 	/* used to wait for the BTRFS_ORDERED_COMPLETE bit */
 	wait_queue_head_t wait;
 
@@ -126,20 +151,14 @@ struct btrfs_ordered_extent {
 	struct completion completion;
 	struct btrfs_work flush_work;
 	struct list_head work_list;
-};
 
-/*
- * calculates the total size you need to allocate for an ordered sum
- * structure spanning 'bytes' in the file
- */
-static inline int btrfs_ordered_sum_size(struct btrfs_fs_info *fs_info,
-					 unsigned long bytes)
-{
-	int num_sectors = (int)DIV_ROUND_UP(bytes, fs_info->sectorsize);
-	int csum_size = btrfs_super_csum_size(fs_info->super_copy);
-
-	return sizeof(struct btrfs_ordered_sum) + num_sectors * csum_size;
-}
+	/*
+	 * Used to reverse-map physical address returned from ZONE_APPEND write
+	 * command in a workqueue context
+	 */
+	u64 physical;
+	struct block_device *bdev;
+};
 
 static inline void
 btrfs_ordered_inode_tree_init(struct btrfs_ordered_inode_tree *t)
@@ -149,51 +168,47 @@ btrfs_ordered_inode_tree_init(struct btrfs_ordered_inode_tree *t)
 	t->last = NULL;
 }
 
+int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent);
+
 void btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry);
-void btrfs_remove_ordered_extent(struct inode *inode,
+void btrfs_remove_ordered_extent(struct btrfs_inode *btrfs_inode,
 				struct btrfs_ordered_extent *entry);
-int btrfs_dec_test_ordered_pending(struct inode *inode,
-				   struct btrfs_ordered_extent **cached,
-				   u64 file_offset, u64 io_size, int uptodate);
-int btrfs_dec_test_first_ordered_pending(struct inode *inode,
-				   struct btrfs_ordered_extent **cached,
-				   u64 *file_offset, u64 io_size,
-				   int uptodate);
-int btrfs_add_ordered_extent(struct inode *inode, u64 file_offset,
-			     u64 disk_bytenr, u64 num_bytes, u64 disk_num_bytes,
-			     int type);
-int btrfs_add_ordered_extent_dio(struct inode *inode, u64 file_offset,
-				 u64 disk_bytenr, u64 num_bytes,
-				 u64 disk_num_bytes, int type);
-int btrfs_add_ordered_extent_compress(struct inode *inode, u64 file_offset,
-				      u64 disk_bytenr, u64 num_bytes,
-				      u64 disk_num_bytes, int type,
-				      int compress_type);
+void btrfs_mark_ordered_io_finished(struct btrfs_inode *inode,
+				struct page *page, u64 file_offset,
+				u64 num_bytes, bool uptodate);
+bool btrfs_dec_test_ordered_pending(struct btrfs_inode *inode,
+				    struct btrfs_ordered_extent **cached,
+				    u64 file_offset, u64 io_size);
+int btrfs_add_ordered_extent(struct btrfs_inode *inode, u64 file_offset,
+			     u64 num_bytes, u64 ram_bytes, u64 disk_bytenr,
+			     u64 disk_num_bytes, u64 offset, unsigned flags,
+			     int compress_type);
 void btrfs_add_ordered_sum(struct btrfs_ordered_extent *entry,
 			   struct btrfs_ordered_sum *sum);
-struct btrfs_ordered_extent *btrfs_lookup_ordered_extent(struct inode *inode,
+struct btrfs_ordered_extent *btrfs_lookup_ordered_extent(struct btrfs_inode *inode,
 							 u64 file_offset);
-void btrfs_start_ordered_extent(struct inode *inode,
-				struct btrfs_ordered_extent *entry, int wait);
+void btrfs_start_ordered_extent(struct btrfs_ordered_extent *entry, int wait);
 int btrfs_wait_ordered_range(struct inode *inode, u64 start, u64 len);
 struct btrfs_ordered_extent *
-btrfs_lookup_first_ordered_extent(struct inode * inode, u64 file_offset);
+btrfs_lookup_first_ordered_extent(struct btrfs_inode *inode, u64 file_offset);
+struct btrfs_ordered_extent *btrfs_lookup_first_ordered_range(
+			struct btrfs_inode *inode, u64 file_offset, u64 len);
 struct btrfs_ordered_extent *btrfs_lookup_ordered_range(
 		struct btrfs_inode *inode,
 		u64 file_offset,
 		u64 len);
-int btrfs_ordered_update_i_size(struct inode *inode, u64 offset,
-				struct btrfs_ordered_extent *ordered);
-int btrfs_find_ordered_sum(struct inode *inode, u64 offset, u64 disk_bytenr,
-			   u8 *sum, int len);
+void btrfs_get_ordered_extents_for_logging(struct btrfs_inode *inode,
+					   struct list_head *list);
 u64 btrfs_wait_ordered_extents(struct btrfs_root *root, u64 nr,
 			       const u64 range_start, const u64 range_len);
 void btrfs_wait_ordered_roots(struct btrfs_fs_info *fs_info, u64 nr,
 			      const u64 range_start, const u64 range_len);
-void btrfs_lock_and_flush_ordered_range(struct extent_io_tree *tree,
-					struct btrfs_inode *inode, u64 start,
+void btrfs_lock_and_flush_ordered_range(struct btrfs_inode *inode, u64 start,
 					u64 end,
 					struct extent_state **cached_state);
+bool btrfs_try_lock_ordered_range(struct btrfs_inode *inode, u64 start, u64 end);
+int btrfs_split_ordered_extent(struct btrfs_ordered_extent *ordered, u64 pre,
+			       u64 post);
 int __init ordered_data_init(void);
 void __cold ordered_data_exit(void);
 
diff --git a/fs/btrfs/print-tree.c b/fs/btrfs/print-tree.c
index 61f44e78e3c9..dd8777872143 100644
--- a/fs/btrfs/print-tree.c
+++ b/fs/btrfs/print-tree.c
@@ -7,6 +7,45 @@
 #include "disk-io.h"
 #include "print-tree.h"
 
+struct root_name_map {
+	u64 id;
+	char name[16];
+};
+
+static const struct root_name_map root_map[] = {
+	{ BTRFS_ROOT_TREE_OBJECTID,		"ROOT_TREE"		},
+	{ BTRFS_EXTENT_TREE_OBJECTID,		"EXTENT_TREE"		},
+	{ BTRFS_CHUNK_TREE_OBJECTID,		"CHUNK_TREE"		},
+	{ BTRFS_DEV_TREE_OBJECTID,		"DEV_TREE"		},
+	{ BTRFS_FS_TREE_OBJECTID,		"FS_TREE"		},
+	{ BTRFS_CSUM_TREE_OBJECTID,		"CSUM_TREE"		},
+	{ BTRFS_TREE_LOG_OBJECTID,		"TREE_LOG"		},
+	{ BTRFS_QUOTA_TREE_OBJECTID,		"QUOTA_TREE"		},
+	{ BTRFS_UUID_TREE_OBJECTID,		"UUID_TREE"		},
+	{ BTRFS_FREE_SPACE_TREE_OBJECTID,	"FREE_SPACE_TREE"	},
+	{ BTRFS_BLOCK_GROUP_TREE_OBJECTID,	"BLOCK_GROUP_TREE"	},
+	{ BTRFS_DATA_RELOC_TREE_OBJECTID,	"DATA_RELOC_TREE"	},
+};
+
+const char *btrfs_root_name(const struct btrfs_key *key, char *buf)
+{
+	int i;
+
+	if (key->objectid == BTRFS_TREE_RELOC_OBJECTID) {
+		snprintf(buf, BTRFS_ROOT_NAME_BUF_LEN,
+			 "TREE_RELOC offset=%llu", key->offset);
+		return buf;
+	}
+
+	for (i = 0; i < ARRAY_SIZE(root_map); i++) {
+		if (root_map[i].id == key->objectid)
+			return root_map[i].name;
+	}
+
+	snprintf(buf, BTRFS_ROOT_NAME_BUF_LEN, "%llu", key->objectid);
+	return buf;
+}
+
 static void print_chunk(struct extent_buffer *eb, struct btrfs_chunk *chunk)
 {
 	int num_stripes = btrfs_chunk_num_stripes(eb, chunk);
@@ -47,7 +86,7 @@ static void print_extent_item(struct extent_buffer *eb, int slot, int type)
 	struct btrfs_disk_key key;
 	unsigned long end;
 	unsigned long ptr;
-	u32 item_size = btrfs_item_size_nr(eb, slot);
+	u32 item_size = btrfs_item_size(eb, slot);
 	u64 flags;
 	u64 offset;
 	int ref_index = 0;
@@ -95,9 +134,10 @@ static void print_extent_item(struct extent_buffer *eb, int slot, int type)
 			 * offset is supposed to be a tree block which
 			 * must be aligned to nodesize.
 			 */
-			if (!IS_ALIGNED(offset, eb->fs_info->nodesize))
-				pr_info("\t\t\t(parent %llu is NOT ALIGNED to nodesize %llu)\n",
-					offset, (unsigned long long)eb->fs_info->nodesize);
+			if (!IS_ALIGNED(offset, eb->fs_info->sectorsize))
+				pr_info(
+			"\t\t\t(parent %llu not aligned to sectorsize %u)\n",
+					offset, eb->fs_info->sectorsize);
 			break;
 		case BTRFS_EXTENT_DATA_REF_KEY:
 			dref = (struct btrfs_extent_data_ref *)(&iref->offset);
@@ -112,8 +152,9 @@ static void print_extent_item(struct extent_buffer *eb, int slot, int type)
 			 * must be aligned to nodesize.
 			 */
 			if (!IS_ALIGNED(offset, eb->fs_info->nodesize))
-				pr_info("\t\t\t(parent %llu is NOT ALIGNED to nodesize %llu)\n",
-				     offset, (unsigned long long)eb->fs_info->nodesize);
+				pr_info(
+			"\t\t\t(parent %llu not aligned to sectorsize %u)\n",
+				     offset, eb->fs_info->sectorsize);
 			break;
 		default:
 			pr_cont("(extent %llu has INVALID ref type %d)\n",
@@ -137,8 +178,7 @@ static void print_uuid_item(struct extent_buffer *l, unsigned long offset,
 		__le64 subvol_id;
 
 		read_extent_buffer(l, &subvol_id, offset, sizeof(subvol_id));
-		pr_info("\t\tsubvol_id %llu\n",
-		       (unsigned long long)le64_to_cpu(subvol_id));
+		pr_info("\t\tsubvol_id %llu\n", le64_to_cpu(subvol_id));
 		item_size -= sizeof(u64);
 		offset += sizeof(u64);
 	}
@@ -151,15 +191,8 @@ static void print_uuid_item(struct extent_buffer *l, unsigned long offset,
 static void print_eb_refs_lock(struct extent_buffer *eb)
 {
 #ifdef CONFIG_BTRFS_DEBUG
-	btrfs_info(eb->fs_info,
-"refs %u lock (w:%d r:%d bw:%d br:%d sw:%d sr:%d) lock_owner %u current %u",
-		   atomic_read(&eb->refs), eb->write_locks,
-		   atomic_read(&eb->read_locks),
-		   eb->blocking_writers,
-		   atomic_read(&eb->blocking_readers),
-		   eb->spinning_writers,
-		   atomic_read(&eb->spinning_readers),
-		   eb->lock_owner, current->pid);
+	btrfs_info(eb->fs_info, "refs %u lock_owner %u current %u",
+		   atomic_read(&eb->refs), eb->lock_owner, current->pid);
 #endif
 }
 
@@ -168,7 +201,6 @@ void btrfs_print_leaf(struct extent_buffer *l)
 	struct btrfs_fs_info *fs_info;
 	int i;
 	u32 type, nr;
-	struct btrfs_item *item;
 	struct btrfs_root_item *ri;
 	struct btrfs_dir_item *di;
 	struct btrfs_inode_item *ii;
@@ -192,12 +224,11 @@ void btrfs_print_leaf(struct extent_buffer *l)
 		   btrfs_leaf_free_space(l), btrfs_header_owner(l));
 	print_eb_refs_lock(l);
 	for (i = 0 ; i < nr ; i++) {
-		item = btrfs_item_nr(i);
 		btrfs_item_key_to_cpu(l, &key, i);
 		type = key.type;
 		pr_info("\titem %d key (%llu %u %llu) itemoff %d itemsize %d\n",
 			i, key.objectid, type, key.offset,
-			btrfs_item_offset(l, item), btrfs_item_size(l, item));
+			btrfs_item_offset(l, i), btrfs_item_size(l, i));
 		switch (type) {
 		case BTRFS_INODE_ITEM_KEY:
 			ii = btrfs_item_ptr(l, i, struct btrfs_inode_item);
@@ -315,7 +346,7 @@ void btrfs_print_leaf(struct extent_buffer *l)
 		case BTRFS_UUID_KEY_SUBVOL:
 		case BTRFS_UUID_KEY_RECEIVED_SUBVOL:
 			print_uuid_item(l, btrfs_item_ptr_offset(l, i),
-					btrfs_item_size_nr(l, i));
+					btrfs_item_size(l, i));
 			break;
 		}
 	}
@@ -358,11 +389,12 @@ void btrfs_print_tree(struct extent_buffer *c, bool follow)
 
 		btrfs_node_key_to_cpu(c, &first_key, i);
 		next = read_tree_block(fs_info, btrfs_node_blockptr(c, i),
+				       btrfs_header_owner(c),
 				       btrfs_node_ptr_generation(c, i),
 				       level - 1, &first_key);
-		if (IS_ERR(next)) {
+		if (IS_ERR(next))
 			continue;
-		} else if (!extent_buffer_uptodate(next)) {
+		if (!extent_buffer_uptodate(next)) {
 			free_extent_buffer(next);
 			continue;
 		}
diff --git a/fs/btrfs/print-tree.h b/fs/btrfs/print-tree.h
index e6bb38fd75ad..8c3e9319ec4e 100644
--- a/fs/btrfs/print-tree.h
+++ b/fs/btrfs/print-tree.h
@@ -6,7 +6,11 @@
 #ifndef BTRFS_PRINT_TREE_H
 #define BTRFS_PRINT_TREE_H
 
+/* Buffer size to contain tree name and possibly additional data (offset) */
+#define BTRFS_ROOT_NAME_BUF_LEN				48
+
 void btrfs_print_leaf(struct extent_buffer *l);
 void btrfs_print_tree(struct extent_buffer *c, bool follow);
+const char *btrfs_root_name(const struct btrfs_key *key, char *buf);
 
 #endif
diff --git a/fs/btrfs/props.c b/fs/btrfs/props.c
index deb59e7cfcac..055a631276ce 100644
--- a/fs/btrfs/props.c
+++ b/fs/btrfs/props.c
@@ -17,9 +17,11 @@ static DEFINE_HASHTABLE(prop_handlers_ht, BTRFS_PROP_HANDLERS_HT_BITS);
 struct prop_handler {
 	struct hlist_node node;
 	const char *xattr_name;
-	int (*validate)(const char *value, size_t len);
+	int (*validate)(const struct btrfs_inode *inode, const char *value,
+			size_t len);
 	int (*apply)(struct inode *inode, const char *value, size_t len);
 	const char *(*extract)(struct inode *inode);
+	bool (*ignore)(const struct btrfs_inode *inode);
 	int inheritable;
 };
 
@@ -55,7 +57,8 @@ find_prop_handler(const char *name,
 	return NULL;
 }
 
-int btrfs_validate_prop(const char *name, const char *value, size_t value_len)
+int btrfs_validate_prop(const struct btrfs_inode *inode, const char *name,
+			const char *value, size_t value_len)
 {
 	const struct prop_handler *handler;
 
@@ -69,7 +72,29 @@ int btrfs_validate_prop(const char *name, const char *value, size_t value_len)
 	if (value_len == 0)
 		return 0;
 
-	return handler->validate(value, value_len);
+	return handler->validate(inode, value, value_len);
+}
+
+/*
+ * Check if a property should be ignored (not set) for an inode.
+ *
+ * @inode:     The target inode.
+ * @name:      The property's name.
+ *
+ * The caller must be sure the given property name is valid, for example by
+ * having previously called btrfs_validate_prop().
+ *
+ * Returns:    true if the property should be ignored for the given inode
+ *             false if the property must not be ignored for the given inode
+ */
+bool btrfs_ignore_prop(const struct btrfs_inode *inode, const char *name)
+{
+	const struct prop_handler *handler;
+
+	handler = find_prop_handler(name, NULL);
+	ASSERT(handler != NULL);
+
+	return handler->ignore(inode);
 }
 
 int btrfs_set_prop(struct btrfs_trans_handle *trans, struct inode *inode,
@@ -158,7 +183,7 @@ static int iterate_object_props(struct btrfs_root *root,
 
 		di = btrfs_item_ptr(leaf, slot, struct btrfs_dir_item);
 		cur = 0;
-		total_len = btrfs_item_size_nr(leaf, slot);
+		total_len = btrfs_item_size(leaf, slot);
 
 		while (cur < total_len) {
 			u32 name_len = btrfs_dir_name_len(leaf, di);
@@ -245,21 +270,26 @@ int btrfs_load_inode_props(struct inode *inode, struct btrfs_path *path)
 {
 	struct btrfs_root *root = BTRFS_I(inode)->root;
 	u64 ino = btrfs_ino(BTRFS_I(inode));
-	int ret;
-
-	ret = iterate_object_props(root, path, ino, inode_prop_iterator, inode);
 
-	return ret;
+	return iterate_object_props(root, path, ino, inode_prop_iterator, inode);
 }
 
-static int prop_compression_validate(const char *value, size_t len)
+static int prop_compression_validate(const struct btrfs_inode *inode,
+				     const char *value, size_t len)
 {
+	if (!btrfs_inode_can_compress(inode))
+		return -EINVAL;
+
 	if (!value)
 		return 0;
 
 	if (btrfs_compress_is_valid_type(value, len))
 		return 0;
 
+	if ((len == 2 && strncmp("no", value, 2) == 0) ||
+	    (len == 4 && strncmp("none", value, 4) == 0))
+		return 0;
+
 	return -EINVAL;
 }
 
@@ -269,7 +299,17 @@ static int prop_compression_apply(struct inode *inode, const char *value,
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	int type;
 
+	/* Reset to defaults */
 	if (len == 0) {
+		BTRFS_I(inode)->flags &= ~BTRFS_INODE_COMPRESS;
+		BTRFS_I(inode)->flags &= ~BTRFS_INODE_NOCOMPRESS;
+		BTRFS_I(inode)->prop_compress = BTRFS_COMPRESS_NONE;
+		return 0;
+	}
+
+	/* Set NOCOMPRESS flag */
+	if ((len == 2 && strncmp("no", value, 2) == 0) ||
+	    (len == 4 && strncmp("none", value, 4) == 0)) {
 		BTRFS_I(inode)->flags |= BTRFS_INODE_NOCOMPRESS;
 		BTRFS_I(inode)->flags &= ~BTRFS_INODE_COMPRESS;
 		BTRFS_I(inode)->prop_compress = BTRFS_COMPRESS_NONE;
@@ -296,6 +336,22 @@ static int prop_compression_apply(struct inode *inode, const char *value,
 	return 0;
 }
 
+static bool prop_compression_ignore(const struct btrfs_inode *inode)
+{
+	/*
+	 * Compression only has effect for regular files, and for directories
+	 * we set it just to propagate it to new files created inside them.
+	 * Everything else (symlinks, devices, sockets, fifos) is pointless as
+	 * it will do nothing, so don't waste metadata space on a compression
+	 * xattr for anything that is neither a file nor a directory.
+	 */
+	if (!S_ISREG(inode->vfs_inode.i_mode) &&
+	    !S_ISDIR(inode->vfs_inode.i_mode))
+		return true;
+
+	return false;
+}
+
 static const char *prop_compression_extract(struct inode *inode)
 {
 	switch (BTRFS_I(inode)->prop_compress) {
@@ -316,13 +372,13 @@ static struct prop_handler prop_handlers[] = {
 		.validate = prop_compression_validate,
 		.apply = prop_compression_apply,
 		.extract = prop_compression_extract,
+		.ignore = prop_compression_ignore,
 		.inheritable = 1
 	},
 };
 
-static int inherit_props(struct btrfs_trans_handle *trans,
-			 struct inode *inode,
-			 struct inode *parent)
+int btrfs_inode_inherit_props(struct btrfs_trans_handle *trans,
+			      struct inode *inode, struct inode *parent)
 {
 	struct btrfs_root *root = BTRFS_I(inode)->root;
 	struct btrfs_fs_info *fs_info = root->fs_info;
@@ -342,15 +398,18 @@ static int inherit_props(struct btrfs_trans_handle *trans,
 		if (!h->inheritable)
 			continue;
 
+		if (h->ignore(BTRFS_I(inode)))
+			continue;
+
 		value = h->extract(parent);
 		if (!value)
 			continue;
 
 		/*
 		 * This is not strictly necessary as the property should be
-		 * valid, but in case it isn't, don't propagate it futher.
+		 * valid, but in case it isn't, don't propagate it further.
 		 */
-		ret = h->validate(value, strlen(value));
+		ret = h->validate(BTRFS_I(inode), value, strlen(value));
 		if (ret)
 			continue;
 
@@ -363,8 +422,9 @@ static int inherit_props(struct btrfs_trans_handle *trans,
 		 */
 		if (need_reserve) {
 			num_bytes = btrfs_calc_insert_metadata_size(fs_info, 1);
-			ret = btrfs_block_rsv_add(root, trans->block_rsv,
-					num_bytes, BTRFS_RESERVE_NO_FLUSH);
+			ret = btrfs_block_rsv_add(fs_info, trans->block_rsv,
+						  num_bytes,
+						  BTRFS_RESERVE_NO_FLUSH);
 			if (ret)
 				return ret;
 		}
@@ -383,7 +443,7 @@ static int inherit_props(struct btrfs_trans_handle *trans,
 
 		if (need_reserve) {
 			btrfs_block_rsv_release(fs_info, trans->block_rsv,
-					num_bytes);
+					num_bytes, NULL);
 			if (ret)
 				return ret;
 		}
@@ -393,46 +453,6 @@ static int inherit_props(struct btrfs_trans_handle *trans,
 	return 0;
 }
 
-int btrfs_inode_inherit_props(struct btrfs_trans_handle *trans,
-			      struct inode *inode,
-			      struct inode *dir)
-{
-	if (!dir)
-		return 0;
-
-	return inherit_props(trans, inode, dir);
-}
-
-int btrfs_subvol_inherit_props(struct btrfs_trans_handle *trans,
-			       struct btrfs_root *root,
-			       struct btrfs_root *parent_root)
-{
-	struct super_block *sb = root->fs_info->sb;
-	struct btrfs_key key;
-	struct inode *parent_inode, *child_inode;
-	int ret;
-
-	key.objectid = BTRFS_FIRST_FREE_OBJECTID;
-	key.type = BTRFS_INODE_ITEM_KEY;
-	key.offset = 0;
-
-	parent_inode = btrfs_iget(sb, &key, parent_root);
-	if (IS_ERR(parent_inode))
-		return PTR_ERR(parent_inode);
-
-	child_inode = btrfs_iget(sb, &key, root);
-	if (IS_ERR(child_inode)) {
-		iput(parent_inode);
-		return PTR_ERR(child_inode);
-	}
-
-	ret = inherit_props(trans, child_inode, parent_inode);
-	iput(child_inode);
-	iput(parent_inode);
-
-	return ret;
-}
-
 void __init btrfs_props_init(void)
 {
 	int i;
diff --git a/fs/btrfs/props.h b/fs/btrfs/props.h
index 40b2c65b518c..ca9dd3df129b 100644
--- a/fs/btrfs/props.h
+++ b/fs/btrfs/props.h
@@ -13,7 +13,9 @@ void __init btrfs_props_init(void);
 int btrfs_set_prop(struct btrfs_trans_handle *trans, struct inode *inode,
 		   const char *name, const char *value, size_t value_len,
 		   int flags);
-int btrfs_validate_prop(const char *name, const char *value, size_t value_len);
+int btrfs_validate_prop(const struct btrfs_inode *inode, const char *name,
+			const char *value, size_t value_len);
+bool btrfs_ignore_prop(const struct btrfs_inode *inode, const char *name);
 
 int btrfs_load_inode_props(struct inode *inode, struct btrfs_path *path);
 
@@ -21,8 +23,4 @@ int btrfs_inode_inherit_props(struct btrfs_trans_handle *trans,
 			      struct inode *inode,
 			      struct inode *dir);
 
-int btrfs_subvol_inherit_props(struct btrfs_trans_handle *trans,
-			       struct btrfs_root *root,
-			       struct btrfs_root *parent_root);
-
 #endif
diff --git a/fs/btrfs/qgroup.c b/fs/btrfs/qgroup.c
index ff1870ff3474..9334c3157c22 100644
--- a/fs/btrfs/qgroup.c
+++ b/fs/btrfs/qgroup.c
@@ -11,7 +11,7 @@
 #include <linux/slab.h>
 #include <linux/workqueue.h>
 #include <linux/btrfs.h>
-#include <linux/sizes.h>
+#include <linux/sched/mm.h>
 
 #include "ctree.h"
 #include "transaction.h"
@@ -22,18 +22,8 @@
 #include "extent_io.h"
 #include "qgroup.h"
 #include "block-group.h"
-
-/* TODO XXX FIXME
- *  - subvol delete -> delete when ref goes to 0? delete limits also?
- *  - reorganize keys
- *  - compressed
- *  - sync
- *  - copy also limits on subvol creation
- *  - limit
- *  - caches for ulists
- *  - performance benchmarks
- *  - check all ioctl parameters
- */
+#include "sysfs.h"
+#include "tree-mod-log.h"
 
 /*
  * Helpers to access qgroup reservation
@@ -220,7 +210,8 @@ static struct btrfs_qgroup *add_qgroup_rb(struct btrfs_fs_info *fs_info,
 	return qgroup;
 }
 
-static void __del_qgroup_rb(struct btrfs_qgroup *qgroup)
+static void __del_qgroup_rb(struct btrfs_fs_info *fs_info,
+			    struct btrfs_qgroup *qgroup)
 {
 	struct btrfs_qgroup_list *list;
 
@@ -240,7 +231,6 @@ static void __del_qgroup_rb(struct btrfs_qgroup *qgroup)
 		list_del(&list->next_member);
 		kfree(list);
 	}
-	kfree(qgroup);
 }
 
 /* must be called with qgroup_lock held */
@@ -252,20 +242,23 @@ static int del_qgroup_rb(struct btrfs_fs_info *fs_info, u64 qgroupid)
 		return -ENOENT;
 
 	rb_erase(&qgroup->node, &fs_info->qgroup_tree);
-	__del_qgroup_rb(qgroup);
+	__del_qgroup_rb(fs_info, qgroup);
 	return 0;
 }
 
-/* must be called with qgroup_lock held */
-static int add_relation_rb(struct btrfs_fs_info *fs_info,
-			   u64 memberid, u64 parentid)
+/*
+ * Add relation specified by two qgroups.
+ *
+ * Must be called with qgroup_lock held.
+ *
+ * Return: 0        on success
+ *         -ENOENT  if one of the qgroups is NULL
+ *         <0       other errors
+ */
+static int __add_relation_rb(struct btrfs_qgroup *member, struct btrfs_qgroup *parent)
 {
-	struct btrfs_qgroup *member;
-	struct btrfs_qgroup *parent;
 	struct btrfs_qgroup_list *list;
 
-	member = find_qgroup_rb(fs_info, memberid);
-	parent = find_qgroup_rb(fs_info, parentid);
 	if (!member || !parent)
 		return -ENOENT;
 
@@ -281,7 +274,27 @@ static int add_relation_rb(struct btrfs_fs_info *fs_info,
 	return 0;
 }
 
-/* must be called with qgroup_lock held */
+/*
+ * Add relation specified by two qgroup ids.
+ *
+ * Must be called with qgroup_lock held.
+ *
+ * Return: 0        on success
+ *         -ENOENT  if one of the ids does not exist
+ *         <0       other errors
+ */
+static int add_relation_rb(struct btrfs_fs_info *fs_info, u64 memberid, u64 parentid)
+{
+	struct btrfs_qgroup *member;
+	struct btrfs_qgroup *parent;
+
+	member = find_qgroup_rb(fs_info, memberid);
+	parent = find_qgroup_rb(fs_info, parentid);
+
+	return __add_relation_rb(member, parent);
+}
+
+/* Must be called with qgroup_lock held */
 static int del_relation_rb(struct btrfs_fs_info *fs_info,
 			   u64 memberid, u64 parentid)
 {
@@ -320,6 +333,13 @@ int btrfs_verify_qgroup_counts(struct btrfs_fs_info *fs_info, u64 qgroupid,
 }
 #endif
 
+static void qgroup_mark_inconsistent(struct btrfs_fs_info *fs_info)
+{
+	fs_info->qgroup_flags |= (BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT |
+				  BTRFS_QGROUP_RUNTIME_FLAG_CANCEL_RESCAN |
+				  BTRFS_QGROUP_RUNTIME_FLAG_NO_ACCOUNTING);
+}
+
 /*
  * The full config is read in one go, only called from open_ctree()
  * It doesn't use any locking, as at this point we're still single-threaded
@@ -351,6 +371,9 @@ int btrfs_read_qgroup_config(struct btrfs_fs_info *fs_info)
 		goto out;
 	}
 
+	ret = btrfs_sysfs_add_qgroups(fs_info);
+	if (ret < 0)
+		goto out;
 	/* default this to quota off, in case no status key is found */
 	fs_info->qgroup_flags = 0;
 
@@ -385,7 +408,7 @@ int btrfs_read_qgroup_config(struct btrfs_fs_info *fs_info)
 			}
 			if (btrfs_qgroup_status_generation(l, ptr) !=
 			    fs_info->generation) {
-				flags |= BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
+				qgroup_mark_inconsistent(fs_info);
 				btrfs_err(fs_info,
 					"qgroup generation mismatch, marked as inconsistent");
 			}
@@ -403,7 +426,7 @@ int btrfs_read_qgroup_config(struct btrfs_fs_info *fs_info)
 		if ((qgroup && found_key.type == BTRFS_QGROUP_INFO_KEY) ||
 		    (!qgroup && found_key.type == BTRFS_QGROUP_LIMIT_KEY)) {
 			btrfs_err(fs_info, "inconsistent qgroup config");
-			flags |= BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
+			qgroup_mark_inconsistent(fs_info);
 		}
 		if (!qgroup) {
 			qgroup = add_qgroup_rb(fs_info, found_key.offset);
@@ -412,6 +435,10 @@ int btrfs_read_qgroup_config(struct btrfs_fs_info *fs_info)
 				goto out;
 			}
 		}
+		ret = btrfs_sysfs_add_one_qgroup(fs_info, qgroup);
+		if (ret < 0)
+			goto out;
+
 		switch (found_key.type) {
 		case BTRFS_QGROUP_INFO_KEY: {
 			struct btrfs_qgroup_info_item *ptr;
@@ -488,24 +515,63 @@ next2:
 			break;
 	}
 out:
+	btrfs_free_path(path);
 	fs_info->qgroup_flags |= flags;
 	if (!(fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_ON))
 		clear_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags);
 	else if (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN &&
 		 ret >= 0)
 		ret = qgroup_rescan_init(fs_info, rescan_progress, 0);
-	btrfs_free_path(path);
 
 	if (ret < 0) {
 		ulist_free(fs_info->qgroup_ulist);
 		fs_info->qgroup_ulist = NULL;
 		fs_info->qgroup_flags &= ~BTRFS_QGROUP_STATUS_FLAG_RESCAN;
+		btrfs_sysfs_del_qgroups(fs_info);
 	}
 
 	return ret < 0 ? ret : 0;
 }
 
 /*
+ * Called in close_ctree() when quota is still enabled.  This verifies we don't
+ * leak some reserved space.
+ *
+ * Return false if no reserved space is left.
+ * Return true if some reserved space is leaked.
+ */
+bool btrfs_check_quota_leak(struct btrfs_fs_info *fs_info)
+{
+	struct rb_node *node;
+	bool ret = false;
+
+	if (!test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags))
+		return ret;
+	/*
+	 * Since we're unmounting, there is no race and no need to grab qgroup
+	 * lock.  And here we don't go post-order to provide a more user
+	 * friendly sorted result.
+	 */
+	for (node = rb_first(&fs_info->qgroup_tree); node; node = rb_next(node)) {
+		struct btrfs_qgroup *qgroup;
+		int i;
+
+		qgroup = rb_entry(node, struct btrfs_qgroup, node);
+		for (i = 0; i < BTRFS_QGROUP_RSV_LAST; i++) {
+			if (qgroup->rsv.values[i]) {
+				ret = true;
+				btrfs_warn(fs_info,
+		"qgroup %hu/%llu has unreleased space, type %d rsv %llu",
+				   btrfs_qgroup_level(qgroup->qgroupid),
+				   btrfs_qgroup_subvolid(qgroup->qgroupid),
+				   i, qgroup->rsv.values[i]);
+			}
+		}
+	}
+	return ret;
+}
+
+/*
  * This is called from close_ctree() or open_ctree() or btrfs_quota_disable(),
  * first two are in single-threaded paths.And for the third one, we have set
  * quota_root to be null with qgroup_lock held before, so it is safe to clean
@@ -519,7 +585,9 @@ void btrfs_free_qgroup_config(struct btrfs_fs_info *fs_info)
 	while ((n = rb_first(&fs_info->qgroup_tree))) {
 		qgroup = rb_entry(n, struct btrfs_qgroup, node);
 		rb_erase(n, &fs_info->qgroup_tree);
-		__del_qgroup_rb(qgroup);
+		__del_qgroup_rb(fs_info, qgroup);
+		btrfs_sysfs_del_one_qgroup(fs_info, qgroup);
+		kfree(qgroup);
 	}
 	/*
 	 * We call btrfs_free_qgroup_config() when unmounting
@@ -528,6 +596,7 @@ void btrfs_free_qgroup_config(struct btrfs_fs_info *fs_info)
 	 */
 	ulist_free(fs_info->qgroup_ulist);
 	fs_info->qgroup_ulist = NULL;
+	btrfs_sysfs_del_qgroups(fs_info);
 }
 
 static int add_qgroup_relation_item(struct btrfs_trans_handle *trans, u64 src,
@@ -816,7 +885,8 @@ static int update_qgroup_status_item(struct btrfs_trans_handle *trans)
 	l = path->nodes[0];
 	slot = path->slots[0];
 	ptr = btrfs_item_ptr(l, slot, struct btrfs_qgroup_status_item);
-	btrfs_set_qgroup_status_flags(l, ptr, fs_info->qgroup_flags);
+	btrfs_set_qgroup_status_flags(l, ptr, fs_info->qgroup_flags &
+				      BTRFS_QGROUP_STATUS_FLAGS_MASK);
 	btrfs_set_qgroup_status_generation(l, ptr, trans->transid);
 	btrfs_set_qgroup_status_rescan(l, ptr,
 				fs_info->qgroup_rescan_progress.objectid);
@@ -844,8 +914,6 @@ static int btrfs_clean_quota_tree(struct btrfs_trans_handle *trans,
 	if (!path)
 		return -ENOMEM;
 
-	path->leave_spinning = 1;
-
 	key.objectid = 0;
 	key.offset = 0;
 	key.type = 0;
@@ -887,19 +955,53 @@ int btrfs_quota_enable(struct btrfs_fs_info *fs_info)
 	struct btrfs_key found_key;
 	struct btrfs_qgroup *qgroup = NULL;
 	struct btrfs_trans_handle *trans = NULL;
+	struct ulist *ulist = NULL;
 	int ret = 0;
 	int slot;
 
+	/*
+	 * We need to have subvol_sem write locked, to prevent races between
+	 * concurrent tasks trying to enable quotas, because we will unlock
+	 * and relock qgroup_ioctl_lock before setting fs_info->quota_root
+	 * and before setting BTRFS_FS_QUOTA_ENABLED.
+	 */
+	lockdep_assert_held_write(&fs_info->subvol_sem);
+
+	if (btrfs_fs_incompat(fs_info, EXTENT_TREE_V2)) {
+		btrfs_err(fs_info,
+			  "qgroups are currently unsupported in extent tree v2");
+		return -EINVAL;
+	}
+
 	mutex_lock(&fs_info->qgroup_ioctl_lock);
 	if (fs_info->quota_root)
 		goto out;
 
-	fs_info->qgroup_ulist = ulist_alloc(GFP_KERNEL);
-	if (!fs_info->qgroup_ulist) {
+	ulist = ulist_alloc(GFP_KERNEL);
+	if (!ulist) {
 		ret = -ENOMEM;
 		goto out;
 	}
 
+	ret = btrfs_sysfs_add_qgroups(fs_info);
+	if (ret < 0)
+		goto out;
+
+	/*
+	 * Unlock qgroup_ioctl_lock before starting the transaction. This is to
+	 * avoid lock acquisition inversion problems (reported by lockdep) between
+	 * qgroup_ioctl_lock and the vfs freeze semaphores, acquired when we
+	 * start a transaction.
+	 * After we started the transaction lock qgroup_ioctl_lock again and
+	 * check if someone else created the quota root in the meanwhile. If so,
+	 * just return success and release the transaction handle.
+	 *
+	 * Also we don't need to worry about someone else calling
+	 * btrfs_sysfs_add_qgroups() after we unlock and getting an error because
+	 * that function returns 0 (success) when the sysfs entries already exist.
+	 */
+	mutex_unlock(&fs_info->qgroup_ioctl_lock);
+
 	/*
 	 * 1 for quota root item
 	 * 1 for BTRFS_QGROUP_STATUS item
@@ -909,12 +1011,20 @@ int btrfs_quota_enable(struct btrfs_fs_info *fs_info)
 	 * would be a lot of overkill.
 	 */
 	trans = btrfs_start_transaction(tree_root, 2);
+
+	mutex_lock(&fs_info->qgroup_ioctl_lock);
 	if (IS_ERR(trans)) {
 		ret = PTR_ERR(trans);
 		trans = NULL;
 		goto out;
 	}
 
+	if (fs_info->quota_root)
+		goto out;
+
+	fs_info->qgroup_ulist = ulist;
+	ulist = NULL;
+
 	/*
 	 * initially create the quota tree
 	 */
@@ -950,7 +1060,8 @@ int btrfs_quota_enable(struct btrfs_fs_info *fs_info)
 	btrfs_set_qgroup_status_version(leaf, ptr, BTRFS_QGROUP_STATUS_VERSION);
 	fs_info->qgroup_flags = BTRFS_QGROUP_STATUS_FLAG_ON |
 				BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
-	btrfs_set_qgroup_status_flags(leaf, ptr, fs_info->qgroup_flags);
+	btrfs_set_qgroup_status_flags(leaf, ptr, fs_info->qgroup_flags &
+				      BTRFS_QGROUP_STATUS_FLAGS_MASK);
 	btrfs_set_qgroup_status_rescan(leaf, ptr, 0);
 
 	btrfs_mark_buffer_dirty(leaf);
@@ -974,6 +1085,10 @@ int btrfs_quota_enable(struct btrfs_fs_info *fs_info)
 		btrfs_item_key_to_cpu(leaf, &found_key, slot);
 
 		if (found_key.type == BTRFS_ROOT_REF_KEY) {
+
+			/* Release locks on tree_root before we access quota_root */
+			btrfs_release_path(path);
+
 			ret = add_qgroup_item(trans, quota_root,
 					      found_key.offset);
 			if (ret) {
@@ -987,6 +1102,25 @@ int btrfs_quota_enable(struct btrfs_fs_info *fs_info)
 				btrfs_abort_transaction(trans, ret);
 				goto out_free_path;
 			}
+			ret = btrfs_sysfs_add_one_qgroup(fs_info, qgroup);
+			if (ret < 0) {
+				btrfs_abort_transaction(trans, ret);
+				goto out_free_path;
+			}
+			ret = btrfs_search_slot_for_read(tree_root, &found_key,
+							 path, 1, 0);
+			if (ret < 0) {
+				btrfs_abort_transaction(trans, ret);
+				goto out_free_path;
+			}
+			if (ret > 0) {
+				/*
+				 * Shouldn't happen, but in case it does we
+				 * don't need to do the btrfs_next_item, just
+				 * continue.
+				 */
+				continue;
+			}
 		}
 		ret = btrfs_next_item(tree_root, path);
 		if (ret < 0) {
@@ -1011,9 +1145,25 @@ out_add_root:
 		btrfs_abort_transaction(trans, ret);
 		goto out_free_path;
 	}
+	ret = btrfs_sysfs_add_one_qgroup(fs_info, qgroup);
+	if (ret < 0) {
+		btrfs_abort_transaction(trans, ret);
+		goto out_free_path;
+	}
 
+	mutex_unlock(&fs_info->qgroup_ioctl_lock);
+	/*
+	 * Commit the transaction while not holding qgroup_ioctl_lock, to avoid
+	 * a deadlock with tasks concurrently doing other qgroup operations, such
+	 * adding/removing qgroups or adding/deleting qgroup relations for example,
+	 * because all qgroup operations first start or join a transaction and then
+	 * lock the qgroup_ioctl_lock mutex.
+	 * We are safe from a concurrent task trying to enable quotas, by calling
+	 * this function, since we are serialized by fs_info->subvol_sem.
+	 */
 	ret = btrfs_commit_transaction(trans);
 	trans = NULL;
+	mutex_lock(&fs_info->qgroup_ioctl_lock);
 	if (ret)
 		goto out_free_path;
 
@@ -1030,26 +1180,43 @@ out_add_root:
 	ret = qgroup_rescan_init(fs_info, 0, 1);
 	if (!ret) {
 	        qgroup_rescan_zero_tracking(fs_info);
+		fs_info->qgroup_rescan_running = true;
 	        btrfs_queue_work(fs_info->qgroup_rescan_workers,
 	                         &fs_info->qgroup_rescan_work);
+	} else {
+		/*
+		 * We have set both BTRFS_FS_QUOTA_ENABLED and
+		 * BTRFS_QGROUP_STATUS_FLAG_ON, so we can only fail with
+		 * -EINPROGRESS. That can happen because someone started the
+		 * rescan worker by calling quota rescan ioctl before we
+		 * attempted to initialize the rescan worker. Failure due to
+		 * quotas disabled in the meanwhile is not possible, because
+		 * we are holding a write lock on fs_info->subvol_sem, which
+		 * is also acquired when disabling quotas.
+		 * Ignore such error, and any other error would need to undo
+		 * everything we did in the transaction we just committed.
+		 */
+		ASSERT(ret == -EINPROGRESS);
+		ret = 0;
 	}
 
 out_free_path:
 	btrfs_free_path(path);
 out_free_root:
-	if (ret) {
-		free_extent_buffer(quota_root->node);
-		free_extent_buffer(quota_root->commit_root);
-		kfree(quota_root);
-	}
+	if (ret)
+		btrfs_put_root(quota_root);
 out:
 	if (ret) {
 		ulist_free(fs_info->qgroup_ulist);
 		fs_info->qgroup_ulist = NULL;
-		if (trans)
-			btrfs_end_transaction(trans);
+		btrfs_sysfs_del_qgroups(fs_info);
 	}
 	mutex_unlock(&fs_info->qgroup_ioctl_lock);
+	if (ret && trans)
+		btrfs_end_transaction(trans);
+	else if (trans)
+		ret = btrfs_end_transaction(trans);
+	ulist_free(ulist);
 	return ret;
 }
 
@@ -1059,28 +1226,60 @@ int btrfs_quota_disable(struct btrfs_fs_info *fs_info)
 	struct btrfs_trans_handle *trans = NULL;
 	int ret = 0;
 
+	/*
+	 * We need to have subvol_sem write locked, to prevent races between
+	 * concurrent tasks trying to disable quotas, because we will unlock
+	 * and relock qgroup_ioctl_lock across BTRFS_FS_QUOTA_ENABLED changes.
+	 */
+	lockdep_assert_held_write(&fs_info->subvol_sem);
+
 	mutex_lock(&fs_info->qgroup_ioctl_lock);
 	if (!fs_info->quota_root)
 		goto out;
 
 	/*
+	 * Unlock the qgroup_ioctl_lock mutex before waiting for the rescan worker to
+	 * complete. Otherwise we can deadlock because btrfs_remove_qgroup() needs
+	 * to lock that mutex while holding a transaction handle and the rescan
+	 * worker needs to commit a transaction.
+	 */
+	mutex_unlock(&fs_info->qgroup_ioctl_lock);
+
+	/*
+	 * Request qgroup rescan worker to complete and wait for it. This wait
+	 * must be done before transaction start for quota disable since it may
+	 * deadlock with transaction by the qgroup rescan worker.
+	 */
+	clear_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags);
+	btrfs_qgroup_wait_for_completion(fs_info, false);
+
+	/*
 	 * 1 For the root item
 	 *
 	 * We should also reserve enough items for the quota tree deletion in
 	 * btrfs_clean_quota_tree but this is not done.
+	 *
+	 * Also, we must always start a transaction without holding the mutex
+	 * qgroup_ioctl_lock, see btrfs_quota_enable().
 	 */
 	trans = btrfs_start_transaction(fs_info->tree_root, 1);
+
+	mutex_lock(&fs_info->qgroup_ioctl_lock);
 	if (IS_ERR(trans)) {
 		ret = PTR_ERR(trans);
+		trans = NULL;
+		set_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags);
 		goto out;
 	}
 
-	clear_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags);
-	btrfs_qgroup_wait_for_completion(fs_info, false);
+	if (!fs_info->quota_root)
+		goto out;
+
 	spin_lock(&fs_info->qgroup_lock);
 	quota_root = fs_info->quota_root;
 	fs_info->quota_root = NULL;
 	fs_info->qgroup_flags &= ~BTRFS_QGROUP_STATUS_FLAG_ON;
+	fs_info->qgroup_drop_subtree_thres = BTRFS_MAX_LEVEL;
 	spin_unlock(&fs_info->qgroup_lock);
 
 	btrfs_free_qgroup_config(fs_info);
@@ -1088,13 +1287,13 @@ int btrfs_quota_disable(struct btrfs_fs_info *fs_info)
 	ret = btrfs_clean_quota_tree(trans, quota_root);
 	if (ret) {
 		btrfs_abort_transaction(trans, ret);
-		goto end_trans;
+		goto out;
 	}
 
 	ret = btrfs_del_root(trans, &quota_root->root_key);
 	if (ret) {
 		btrfs_abort_transaction(trans, ret);
-		goto end_trans;
+		goto out;
 	}
 
 	list_del(&quota_root->dirty_list);
@@ -1102,16 +1301,18 @@ int btrfs_quota_disable(struct btrfs_fs_info *fs_info)
 	btrfs_tree_lock(quota_root->node);
 	btrfs_clean_tree_block(quota_root->node);
 	btrfs_tree_unlock(quota_root->node);
-	btrfs_free_tree_block(trans, quota_root, quota_root->node, 0, 1);
+	btrfs_free_tree_block(trans, btrfs_root_id(quota_root),
+			      quota_root->node, 0, 1);
 
-	free_extent_buffer(quota_root->node);
-	free_extent_buffer(quota_root->commit_root);
-	kfree(quota_root);
+	btrfs_put_root(quota_root);
 
-end_trans:
-	ret = btrfs_end_transaction(trans);
 out:
 	mutex_unlock(&fs_info->qgroup_ioctl_lock);
+	if (ret && trans)
+		btrfs_end_transaction(trans);
+	else if (trans)
+		ret = btrfs_end_transaction(trans);
+
 	return ret;
 }
 
@@ -1247,13 +1448,17 @@ int btrfs_add_qgroup_relation(struct btrfs_trans_handle *trans, u64 src,
 	struct btrfs_qgroup *member;
 	struct btrfs_qgroup_list *list;
 	struct ulist *tmp;
+	unsigned int nofs_flag;
 	int ret = 0;
 
 	/* Check the level of src and dst first */
 	if (btrfs_qgroup_level(src) >= btrfs_qgroup_level(dst))
 		return -EINVAL;
 
+	/* We hold a transaction handle open, must do a NOFS allocation. */
+	nofs_flag = memalloc_nofs_save();
 	tmp = ulist_alloc(GFP_KERNEL);
+	memalloc_nofs_restore(nofs_flag);
 	if (!tmp)
 		return -ENOMEM;
 
@@ -1288,7 +1493,7 @@ int btrfs_add_qgroup_relation(struct btrfs_trans_handle *trans, u64 src,
 	}
 
 	spin_lock(&fs_info->qgroup_lock);
-	ret = add_relation_rb(fs_info, src, dst);
+	ret = __add_relation_rb(member, parent);
 	if (ret < 0) {
 		spin_unlock(&fs_info->qgroup_lock);
 		goto out;
@@ -1310,10 +1515,14 @@ static int __del_qgroup_relation(struct btrfs_trans_handle *trans, u64 src,
 	struct btrfs_qgroup_list *list;
 	struct ulist *tmp;
 	bool found = false;
+	unsigned int nofs_flag;
 	int ret = 0;
 	int ret2;
 
+	/* We hold a transaction handle open, must do a NOFS allocation. */
+	nofs_flag = memalloc_nofs_save();
 	tmp = ulist_alloc(GFP_KERNEL);
+	memalloc_nofs_restore(nofs_flag);
 	if (!tmp)
 		return -ENOMEM;
 
@@ -1402,8 +1611,11 @@ int btrfs_create_qgroup(struct btrfs_trans_handle *trans, u64 qgroupid)
 	qgroup = add_qgroup_rb(fs_info, qgroupid);
 	spin_unlock(&fs_info->qgroup_lock);
 
-	if (IS_ERR(qgroup))
+	if (IS_ERR(qgroup)) {
 		ret = PTR_ERR(qgroup);
+		goto out;
+	}
+	ret = btrfs_sysfs_add_one_qgroup(fs_info, qgroup);
 out:
 	mutex_unlock(&fs_info->qgroup_ioctl_lock);
 	return ret;
@@ -1450,6 +1662,14 @@ int btrfs_remove_qgroup(struct btrfs_trans_handle *trans, u64 qgroupid)
 	spin_lock(&fs_info->qgroup_lock);
 	del_qgroup_rb(fs_info, qgroupid);
 	spin_unlock(&fs_info->qgroup_lock);
+
+	/*
+	 * Remove the qgroup from sysfs now without holding the qgroup_lock
+	 * spinlock, since the sysfs_remove_group() function needs to take
+	 * the mutex kernfs_mutex through kernfs_remove_by_name_ns().
+	 */
+	btrfs_sysfs_del_one_qgroup(fs_info, qgroup);
+	kfree(qgroup);
 out:
 	mutex_unlock(&fs_info->qgroup_ioctl_lock);
 	return ret;
@@ -1522,7 +1742,7 @@ int btrfs_limit_qgroup(struct btrfs_trans_handle *trans, u64 qgroupid,
 
 	ret = update_qgroup_limit_item(trans, qgroup);
 	if (ret) {
-		fs_info->qgroup_flags |= BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
+		qgroup_mark_inconsistent(fs_info);
 		btrfs_info(fs_info, "unable to update quota limit for %llu",
 		       qgroupid);
 	}
@@ -1567,17 +1787,42 @@ int btrfs_qgroup_trace_extent_nolock(struct btrfs_fs_info *fs_info,
 	return 0;
 }
 
-int btrfs_qgroup_trace_extent_post(struct btrfs_fs_info *fs_info,
+int btrfs_qgroup_trace_extent_post(struct btrfs_trans_handle *trans,
 				   struct btrfs_qgroup_extent_record *qrecord)
 {
 	struct ulist *old_root;
 	u64 bytenr = qrecord->bytenr;
 	int ret;
 
-	ret = btrfs_find_all_roots(NULL, fs_info, bytenr, 0, &old_root, false);
+	/*
+	 * We are always called in a context where we are already holding a
+	 * transaction handle. Often we are called when adding a data delayed
+	 * reference from btrfs_truncate_inode_items() (truncating or unlinking),
+	 * in which case we will be holding a write lock on extent buffer from a
+	 * subvolume tree. In this case we can't allow btrfs_find_all_roots() to
+	 * acquire fs_info->commit_root_sem, because that is a higher level lock
+	 * that must be acquired before locking any extent buffers.
+	 *
+	 * So we want btrfs_find_all_roots() to not acquire the commit_root_sem
+	 * but we can't pass it a non-NULL transaction handle, because otherwise
+	 * it would not use commit roots and would lock extent buffers, causing
+	 * a deadlock if it ends up trying to read lock the same extent buffer
+	 * that was previously write locked at btrfs_truncate_inode_items().
+	 *
+	 * So pass a NULL transaction handle to btrfs_find_all_roots() and
+	 * explicitly tell it to not acquire the commit_root_sem - if we are
+	 * holding a transaction handle we don't need its protection.
+	 */
+	ASSERT(trans != NULL);
+
+	if (trans->fs_info->qgroup_flags & BTRFS_QGROUP_RUNTIME_FLAG_NO_ACCOUNTING)
+		return 0;
+
+	ret = btrfs_find_all_roots(NULL, trans->fs_info, bytenr, 0, &old_root,
+				   true);
 	if (ret < 0) {
-		fs_info->qgroup_flags |= BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
-		btrfs_warn(fs_info,
+		qgroup_mark_inconsistent(trans->fs_info);
+		btrfs_warn(trans->fs_info,
 "error accounting new delayed refs extent (err code: %d), quota inconsistent",
 			ret);
 		return 0;
@@ -1621,7 +1866,7 @@ int btrfs_qgroup_trace_extent(struct btrfs_trans_handle *trans, u64 bytenr,
 		kfree(record);
 		return 0;
 	}
-	return btrfs_qgroup_trace_extent_post(fs_info, record);
+	return btrfs_qgroup_trace_extent_post(trans, record);
 }
 
 int btrfs_qgroup_trace_leaf_items(struct btrfs_trans_handle *trans,
@@ -1814,34 +2059,22 @@ static int qgroup_trace_extent_swap(struct btrfs_trans_handle* trans,
 		struct btrfs_key dst_key;
 
 		if (src_path->nodes[cur_level] == NULL) {
-			struct btrfs_key first_key;
 			struct extent_buffer *eb;
 			int parent_slot;
-			u64 child_gen;
-			u64 child_bytenr;
 
 			eb = src_path->nodes[cur_level + 1];
 			parent_slot = src_path->slots[cur_level + 1];
-			child_bytenr = btrfs_node_blockptr(eb, parent_slot);
-			child_gen = btrfs_node_ptr_generation(eb, parent_slot);
-			btrfs_node_key_to_cpu(eb, &first_key, parent_slot);
 
-			eb = read_tree_block(fs_info, child_bytenr, child_gen,
-					     cur_level, &first_key);
+			eb = btrfs_read_node_slot(eb, parent_slot);
 			if (IS_ERR(eb)) {
 				ret = PTR_ERR(eb);
 				goto out;
-			} else if (!extent_buffer_uptodate(eb)) {
-				free_extent_buffer(eb);
-				ret = -EIO;
-				goto out;
 			}
 
 			src_path->nodes[cur_level] = eb;
 
 			btrfs_tree_read_lock(eb);
-			btrfs_set_lock_blocking_read(eb);
-			src_path->locks[cur_level] = BTRFS_READ_LOCK_BLOCKING;
+			src_path->locks[cur_level] = BTRFS_READ_LOCK;
 		}
 
 		src_path->slots[cur_level] = dst_path->slots[cur_level];
@@ -1936,10 +2169,8 @@ static int qgroup_trace_new_subtree_blocks(struct btrfs_trans_handle* trans,
 
 	/* Read the tree block if needed */
 	if (dst_path->nodes[cur_level] == NULL) {
-		struct btrfs_key first_key;
 		int parent_slot;
 		u64 child_gen;
-		u64 child_bytenr;
 
 		/*
 		 * dst_path->nodes[root_level] must be initialized before
@@ -1958,31 +2189,23 @@ static int qgroup_trace_new_subtree_blocks(struct btrfs_trans_handle* trans,
 		  */
 		eb = dst_path->nodes[cur_level + 1];
 		parent_slot = dst_path->slots[cur_level + 1];
-		child_bytenr = btrfs_node_blockptr(eb, parent_slot);
 		child_gen = btrfs_node_ptr_generation(eb, parent_slot);
-		btrfs_node_key_to_cpu(eb, &first_key, parent_slot);
 
 		/* This node is old, no need to trace */
 		if (child_gen < last_snapshot)
 			goto out;
 
-		eb = read_tree_block(fs_info, child_bytenr, child_gen,
-				     cur_level, &first_key);
+		eb = btrfs_read_node_slot(eb, parent_slot);
 		if (IS_ERR(eb)) {
 			ret = PTR_ERR(eb);
 			goto out;
-		} else if (!extent_buffer_uptodate(eb)) {
-			free_extent_buffer(eb);
-			ret = -EIO;
-			goto out;
 		}
 
 		dst_path->nodes[cur_level] = eb;
 		dst_path->slots[cur_level] = 0;
 
 		btrfs_tree_read_lock(eb);
-		btrfs_set_lock_blocking_read(eb);
-		dst_path->locks[cur_level] = BTRFS_READ_LOCK_BLOCKING;
+		dst_path->locks[cur_level] = BTRFS_READ_LOCK;
 		need_cleanup = true;
 	}
 
@@ -2074,7 +2297,7 @@ static int qgroup_trace_subtree_swap(struct btrfs_trans_handle *trans,
 out:
 	btrfs_free_path(dst_path);
 	if (ret < 0)
-		fs_info->qgroup_flags |= BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
+		qgroup_mark_inconsistent(fs_info);
 	return ret;
 }
 
@@ -2085,6 +2308,7 @@ int btrfs_qgroup_trace_subtree(struct btrfs_trans_handle *trans,
 	struct btrfs_fs_info *fs_info = trans->fs_info;
 	int ret = 0;
 	int level;
+	u8 drop_subptree_thres;
 	struct extent_buffer *eb = root_eb;
 	struct btrfs_path *path = NULL;
 
@@ -2094,8 +2318,25 @@ int btrfs_qgroup_trace_subtree(struct btrfs_trans_handle *trans,
 	if (!test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags))
 		return 0;
 
+	spin_lock(&fs_info->qgroup_lock);
+	drop_subptree_thres = fs_info->qgroup_drop_subtree_thres;
+	spin_unlock(&fs_info->qgroup_lock);
+
+	/*
+	 * This function only gets called for snapshot drop, if we hit a high
+	 * node here, it means we are going to change ownership for quite a lot
+	 * of extents, which will greatly slow down btrfs_commit_transaction().
+	 *
+	 * So here if we find a high tree here, we just skip the accounting and
+	 * mark qgroup inconsistent.
+	 */
+	if (root_level >= drop_subptree_thres) {
+		qgroup_mark_inconsistent(fs_info);
+		return 0;
+	}
+
 	if (!extent_buffer_uptodate(root_eb)) {
-		ret = btrfs_read_buffer(root_eb, root_gen, root_level, NULL);
+		ret = btrfs_read_extent_buffer(root_eb, root_gen, root_level, NULL);
 		if (ret)
 			goto out;
 	}
@@ -2126,38 +2367,28 @@ walk_down:
 	level = root_level;
 	while (level >= 0) {
 		if (path->nodes[level] == NULL) {
-			struct btrfs_key first_key;
 			int parent_slot;
-			u64 child_gen;
 			u64 child_bytenr;
 
 			/*
-			 * We need to get child blockptr/gen from parent before
-			 * we can read it.
+			 * We need to get child blockptr from parent before we
+			 * can read it.
 			  */
 			eb = path->nodes[level + 1];
 			parent_slot = path->slots[level + 1];
 			child_bytenr = btrfs_node_blockptr(eb, parent_slot);
-			child_gen = btrfs_node_ptr_generation(eb, parent_slot);
-			btrfs_node_key_to_cpu(eb, &first_key, parent_slot);
 
-			eb = read_tree_block(fs_info, child_bytenr, child_gen,
-					     level, &first_key);
+			eb = btrfs_read_node_slot(eb, parent_slot);
 			if (IS_ERR(eb)) {
 				ret = PTR_ERR(eb);
 				goto out;
-			} else if (!extent_buffer_uptodate(eb)) {
-				free_extent_buffer(eb);
-				ret = -EIO;
-				goto out;
 			}
 
 			path->nodes[level] = eb;
 			path->slots[level] = 0;
 
 			btrfs_tree_read_lock(eb);
-			btrfs_set_lock_blocking_read(eb);
-			path->locks[level] = BTRFS_READ_LOCK_BLOCKING;
+			path->locks[level] = BTRFS_READ_LOCK;
 
 			ret = btrfs_qgroup_trace_extent(trans, child_bytenr,
 							fs_info->nodesize,
@@ -2253,7 +2484,7 @@ static int qgroup_update_refcnt(struct btrfs_fs_info *fs_info,
  * Update qgroup rfer/excl counters.
  * Rfer update is easy, codes can explain themselves.
  *
- * Excl update is tricky, the update is split into 2 part.
+ * Excl update is tricky, the update is split into 2 parts.
  * Part 1: Possible exclusive <-> sharing detect:
  *	|	A	|	!A	|
  *  -------------------------------------
@@ -2416,10 +2647,11 @@ int btrfs_qgroup_account_extent(struct btrfs_trans_handle *trans, u64 bytenr,
 	int ret = 0;
 
 	/*
-	 * If quotas get disabled meanwhile, the resouces need to be freed and
+	 * If quotas get disabled meanwhile, the resources need to be freed and
 	 * we can't just exit here.
 	 */
-	if (!test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags))
+	if (!test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags) ||
+	    fs_info->qgroup_flags & BTRFS_QGROUP_RUNTIME_FLAG_NO_ACCOUNTING)
 		goto out_free;
 
 	if (new_roots) {
@@ -2515,7 +2747,8 @@ int btrfs_qgroup_account_extents(struct btrfs_trans_handle *trans)
 		num_dirty_extents++;
 		trace_btrfs_qgroup_account_extents(fs_info, record);
 
-		if (!ret) {
+		if (!ret && !(fs_info->qgroup_flags &
+			      BTRFS_QGROUP_RUNTIME_FLAG_NO_ACCOUNTING)) {
 			/*
 			 * Old roots should be searched when inserting qgroup
 			 * extent record
@@ -2535,12 +2768,12 @@ int btrfs_qgroup_account_extents(struct btrfs_trans_handle *trans)
 					record->data_rsv,
 					BTRFS_QGROUP_RSV_DATA);
 			/*
-			 * Use SEQ_LAST as time_seq to do special search, which
-			 * doesn't lock tree or delayed_refs and search current
-			 * root. It's safe inside commit_transaction().
+			 * Use BTRFS_SEQ_LAST as time_seq to do special search,
+			 * which doesn't lock tree or delayed_refs and search
+			 * current root. It's safe inside commit_transaction().
 			 */
 			ret = btrfs_find_all_roots(trans, fs_info,
-				record->bytenr, SEQ_LAST, &new_roots, false);
+			   record->bytenr, BTRFS_SEQ_LAST, &new_roots, false);
 			if (ret < 0)
 				goto cleanup;
 			if (qgroup_to_skip) {
@@ -2588,12 +2821,10 @@ int btrfs_run_qgroups(struct btrfs_trans_handle *trans)
 		spin_unlock(&fs_info->qgroup_lock);
 		ret = update_qgroup_info_item(trans, qgroup);
 		if (ret)
-			fs_info->qgroup_flags |=
-					BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
+			qgroup_mark_inconsistent(fs_info);
 		ret = update_qgroup_limit_item(trans, qgroup);
 		if (ret)
-			fs_info->qgroup_flags |=
-					BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
+			qgroup_mark_inconsistent(fs_info);
 		spin_lock(&fs_info->qgroup_lock);
 	}
 	if (test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags))
@@ -2604,7 +2835,7 @@ int btrfs_run_qgroups(struct btrfs_trans_handle *trans)
 
 	ret = update_qgroup_status_item(trans);
 	if (ret)
-		fs_info->qgroup_flags |= BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
+		qgroup_mark_inconsistent(fs_info);
 
 	return ret;
 }
@@ -2626,6 +2857,7 @@ int btrfs_qgroup_inherit(struct btrfs_trans_handle *trans, u64 srcid,
 	struct btrfs_root *quota_root;
 	struct btrfs_qgroup *srcgroup;
 	struct btrfs_qgroup *dstgroup;
+	bool need_rescan = false;
 	u32 level_size = 0;
 	u64 nums;
 
@@ -2721,7 +2953,7 @@ int btrfs_qgroup_inherit(struct btrfs_trans_handle *trans, u64 srcid,
 
 		ret = update_qgroup_limit_item(trans, dstgroup);
 		if (ret) {
-			fs_info->qgroup_flags |= BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
+			qgroup_mark_inconsistent(fs_info);
 			btrfs_info(fs_info,
 				   "unable to update quota limit for %llu",
 				   dstgroup->qgroupid);
@@ -2769,6 +3001,13 @@ int btrfs_qgroup_inherit(struct btrfs_trans_handle *trans, u64 srcid,
 				goto unlock;
 		}
 		++i_qgroups;
+
+		/*
+		 * If we're doing a snapshot, and adding the snapshot to a new
+		 * qgroup, the numbers are guaranteed to be incorrect.
+		 */
+		if (srcid)
+			need_rescan = true;
 	}
 
 	for (i = 0; i <  inherit->num_ref_copies; ++i, i_qgroups += 2) {
@@ -2788,6 +3027,9 @@ int btrfs_qgroup_inherit(struct btrfs_trans_handle *trans, u64 srcid,
 
 		dst->rfer = src->rfer - level_size;
 		dst->rfer_cmpr = src->rfer_cmpr - level_size;
+
+		/* Manually tweaking numbers certainly needs a rescan */
+		need_rescan = true;
 	}
 	for (i = 0; i <  inherit->num_excl_copies; ++i, i_qgroups += 2) {
 		struct btrfs_qgroup *src;
@@ -2806,30 +3048,23 @@ int btrfs_qgroup_inherit(struct btrfs_trans_handle *trans, u64 srcid,
 
 		dst->excl = src->excl + level_size;
 		dst->excl_cmpr = src->excl_cmpr + level_size;
+		need_rescan = true;
 	}
 
 unlock:
 	spin_unlock(&fs_info->qgroup_lock);
+	if (!ret)
+		ret = btrfs_sysfs_add_one_qgroup(fs_info, dstgroup);
 out:
 	if (!committing)
 		mutex_unlock(&fs_info->qgroup_ioctl_lock);
+	if (need_rescan)
+		qgroup_mark_inconsistent(fs_info);
 	return ret;
 }
 
-/*
- * Two limits to commit transaction in advance.
- *
- * For RATIO, it will be 1/RATIO of the remaining limit as threshold.
- * For SIZE, it will be in byte unit as threshold.
- */
-#define QGROUP_FREE_RATIO		32
-#define QGROUP_FREE_SIZE		SZ_32M
-static bool qgroup_check_limits(struct btrfs_fs_info *fs_info,
-				const struct btrfs_qgroup *qg, u64 num_bytes)
+static bool qgroup_check_limits(const struct btrfs_qgroup *qg, u64 num_bytes)
 {
-	u64 free;
-	u64 threshold;
-
 	if ((qg->lim_flags & BTRFS_QGROUP_LIMIT_MAX_RFER) &&
 	    qgroup_rsv_total(qg) + (s64)qg->rfer + num_bytes > qg->max_rfer)
 		return false;
@@ -2838,32 +3073,6 @@ static bool qgroup_check_limits(struct btrfs_fs_info *fs_info,
 	    qgroup_rsv_total(qg) + (s64)qg->excl + num_bytes > qg->max_excl)
 		return false;
 
-	/*
-	 * Even if we passed the check, it's better to check if reservation
-	 * for meta_pertrans is pushing us near limit.
-	 * If there is too much pertrans reservation or it's near the limit,
-	 * let's try commit transaction to free some, using transaction_kthread
-	 */
-	if ((qg->lim_flags & (BTRFS_QGROUP_LIMIT_MAX_RFER |
-			      BTRFS_QGROUP_LIMIT_MAX_EXCL))) {
-		if (qg->lim_flags & BTRFS_QGROUP_LIMIT_MAX_EXCL) {
-			free = qg->max_excl - qgroup_rsv_total(qg) - qg->excl;
-			threshold = min_t(u64, qg->max_excl / QGROUP_FREE_RATIO,
-					  QGROUP_FREE_SIZE);
-		} else {
-			free = qg->max_rfer - qgroup_rsv_total(qg) - qg->rfer;
-			threshold = min_t(u64, qg->max_rfer / QGROUP_FREE_RATIO,
-					  QGROUP_FREE_SIZE);
-		}
-
-		/*
-		 * Use transaction_kthread to commit transaction, so we no
-		 * longer need to bother nested transaction nor lock context.
-		 */
-		if (free < threshold)
-			btrfs_commit_transaction_locksafe(fs_info);
-	}
-
 	return true;
 }
 
@@ -2911,7 +3120,7 @@ static int qgroup_reserve(struct btrfs_root *root, u64 num_bytes, bool enforce,
 
 		qg = unode_aux_to_qgroup(unode);
 
-		if (enforce && !qgroup_check_limits(fs_info, qg, num_bytes)) {
+		if (enforce && !qgroup_check_limits(qg, num_bytes)) {
 			ret = -EDQUOT;
 			goto out;
 		}
@@ -3036,6 +3245,7 @@ static int qgroup_rescan_leaf(struct btrfs_trans_handle *trans,
 			      struct btrfs_path *path)
 {
 	struct btrfs_fs_info *fs_info = trans->fs_info;
+	struct btrfs_root *extent_root;
 	struct btrfs_key found;
 	struct extent_buffer *scratch_leaf = NULL;
 	struct ulist *roots = NULL;
@@ -3045,7 +3255,9 @@ static int qgroup_rescan_leaf(struct btrfs_trans_handle *trans,
 	int ret;
 
 	mutex_lock(&fs_info->qgroup_rescan_lock);
-	ret = btrfs_search_slot_for_read(fs_info->extent_root,
+	extent_root = btrfs_extent_root(fs_info,
+				fs_info->qgroup_rescan_progress.objectid);
+	ret = btrfs_search_slot_for_read(extent_root,
 					 &fs_info->qgroup_rescan_progress,
 					 path, 1, 0);
 
@@ -3116,6 +3328,14 @@ out:
 	return ret;
 }
 
+static bool rescan_should_stop(struct btrfs_fs_info *fs_info)
+{
+	return btrfs_fs_closing(fs_info) ||
+		test_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state) ||
+		!test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags) ||
+			  fs_info->qgroup_flags & BTRFS_QGROUP_RUNTIME_FLAG_CANCEL_RESCAN;
+}
+
 static void btrfs_qgroup_rescan_worker(struct btrfs_work *work)
 {
 	struct btrfs_fs_info *fs_info = container_of(work, struct btrfs_fs_info,
@@ -3124,6 +3344,7 @@ static void btrfs_qgroup_rescan_worker(struct btrfs_work *work)
 	struct btrfs_trans_handle *trans = NULL;
 	int err = -ENOMEM;
 	int ret = 0;
+	bool stopped = false;
 
 	path = btrfs_alloc_path();
 	if (!path)
@@ -3136,17 +3357,15 @@ static void btrfs_qgroup_rescan_worker(struct btrfs_work *work)
 	path->skip_locking = 1;
 
 	err = 0;
-	while (!err && !btrfs_fs_closing(fs_info)) {
+	while (!err && !(stopped = rescan_should_stop(fs_info))) {
 		trans = btrfs_start_transaction(fs_info->fs_root, 0);
 		if (IS_ERR(trans)) {
 			err = PTR_ERR(trans);
 			break;
 		}
-		if (!test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags)) {
-			err = -EINTR;
-		} else {
-			err = qgroup_rescan_leaf(trans, path);
-		}
+
+		err = qgroup_rescan_leaf(trans, path);
+
 		if (err > 0)
 			btrfs_commit_transaction(trans);
 		else
@@ -3160,7 +3379,7 @@ out:
 	if (err > 0 &&
 	    fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT) {
 		fs_info->qgroup_flags &= ~BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
-	} else if (err < 0) {
+	} else if (err < 0 || stopped) {
 		fs_info->qgroup_flags |= BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
 	}
 	mutex_unlock(&fs_info->qgroup_rescan_lock);
@@ -3179,7 +3398,8 @@ out:
 	}
 
 	mutex_lock(&fs_info->qgroup_rescan_lock);
-	if (!btrfs_fs_closing(fs_info))
+	if (!stopped ||
+	    fs_info->qgroup_flags & BTRFS_QGROUP_RUNTIME_FLAG_CANCEL_RESCAN)
 		fs_info->qgroup_flags &= ~BTRFS_QGROUP_STATUS_FLAG_RESCAN;
 	if (trans) {
 		ret = update_qgroup_status_item(trans);
@@ -3190,6 +3410,7 @@ out:
 		}
 	}
 	fs_info->qgroup_rescan_running = false;
+	fs_info->qgroup_flags &= ~BTRFS_QGROUP_RUNTIME_FLAG_CANCEL_RESCAN;
 	complete_all(&fs_info->qgroup_rescan_completion);
 	mutex_unlock(&fs_info->qgroup_rescan_lock);
 
@@ -3198,8 +3419,10 @@ out:
 
 	btrfs_end_transaction(trans);
 
-	if (btrfs_fs_closing(fs_info)) {
+	if (stopped) {
 		btrfs_info(fs_info, "qgroup scan paused");
+	} else if (fs_info->qgroup_flags & BTRFS_QGROUP_RUNTIME_FLAG_CANCEL_RESCAN) {
+		btrfs_info(fs_info, "qgroup scan cancelled");
 	} else if (err >= 0) {
 		btrfs_info(fs_info, "qgroup scan completed%s",
 			err > 0 ? " (inconsistency flag cleared)" : "");
@@ -3237,7 +3460,6 @@ qgroup_rescan_init(struct btrfs_fs_info *fs_info, u64 progress_objectid,
 	}
 
 	mutex_lock(&fs_info->qgroup_rescan_lock);
-	spin_lock(&fs_info->qgroup_lock);
 
 	if (init_flags) {
 		if (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN) {
@@ -3249,10 +3471,12 @@ qgroup_rescan_init(struct btrfs_fs_info *fs_info, u64 progress_objectid,
 			btrfs_warn(fs_info,
 			"qgroup rescan init failed, qgroup is not enabled");
 			ret = -EINVAL;
+		} else if (!test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags)) {
+			/* Quota disable is in progress */
+			ret = -EBUSY;
 		}
 
 		if (ret) {
-			spin_unlock(&fs_info->qgroup_lock);
 			mutex_unlock(&fs_info->qgroup_rescan_lock);
 			return ret;
 		}
@@ -3261,11 +3485,10 @@ qgroup_rescan_init(struct btrfs_fs_info *fs_info, u64 progress_objectid,
 
 	memset(&fs_info->qgroup_rescan_progress, 0,
 		sizeof(fs_info->qgroup_rescan_progress));
+	fs_info->qgroup_flags &= ~(BTRFS_QGROUP_RUNTIME_FLAG_CANCEL_RESCAN |
+				   BTRFS_QGROUP_RUNTIME_FLAG_NO_ACCOUNTING);
 	fs_info->qgroup_rescan_progress.objectid = progress_objectid;
 	init_completion(&fs_info->qgroup_rescan_completion);
-	fs_info->qgroup_rescan_running = true;
-
-	spin_unlock(&fs_info->qgroup_lock);
 	mutex_unlock(&fs_info->qgroup_rescan_lock);
 
 	btrfs_init_work(&fs_info->qgroup_rescan_work,
@@ -3326,8 +3549,11 @@ btrfs_qgroup_rescan(struct btrfs_fs_info *fs_info)
 
 	qgroup_rescan_zero_tracking(fs_info);
 
+	mutex_lock(&fs_info->qgroup_rescan_lock);
+	fs_info->qgroup_rescan_running = true;
 	btrfs_queue_work(fs_info->qgroup_rescan_workers,
 			 &fs_info->qgroup_rescan_work);
+	mutex_unlock(&fs_info->qgroup_rescan_lock);
 
 	return 0;
 }
@@ -3339,9 +3565,7 @@ int btrfs_qgroup_wait_for_completion(struct btrfs_fs_info *fs_info,
 	int ret = 0;
 
 	mutex_lock(&fs_info->qgroup_rescan_lock);
-	spin_lock(&fs_info->qgroup_lock);
 	running = fs_info->qgroup_rescan_running;
-	spin_unlock(&fs_info->qgroup_lock);
 	mutex_unlock(&fs_info->qgroup_rescan_lock);
 
 	if (!running)
@@ -3363,33 +3587,142 @@ int btrfs_qgroup_wait_for_completion(struct btrfs_fs_info *fs_info,
 void
 btrfs_qgroup_rescan_resume(struct btrfs_fs_info *fs_info)
 {
-	if (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN)
+	if (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN) {
+		mutex_lock(&fs_info->qgroup_rescan_lock);
+		fs_info->qgroup_rescan_running = true;
 		btrfs_queue_work(fs_info->qgroup_rescan_workers,
 				 &fs_info->qgroup_rescan_work);
+		mutex_unlock(&fs_info->qgroup_rescan_lock);
+	}
+}
+
+#define rbtree_iterate_from_safe(node, next, start)				\
+       for (node = start; node && ({ next = rb_next(node); 1;}); node = next)
+
+static int qgroup_unreserve_range(struct btrfs_inode *inode,
+				  struct extent_changeset *reserved, u64 start,
+				  u64 len)
+{
+	struct rb_node *node;
+	struct rb_node *next;
+	struct ulist_node *entry;
+	int ret = 0;
+
+	node = reserved->range_changed.root.rb_node;
+	if (!node)
+		return 0;
+	while (node) {
+		entry = rb_entry(node, struct ulist_node, rb_node);
+		if (entry->val < start)
+			node = node->rb_right;
+		else
+			node = node->rb_left;
+	}
+
+	if (entry->val > start && rb_prev(&entry->rb_node))
+		entry = rb_entry(rb_prev(&entry->rb_node), struct ulist_node,
+				 rb_node);
+
+	rbtree_iterate_from_safe(node, next, &entry->rb_node) {
+		u64 entry_start;
+		u64 entry_end;
+		u64 entry_len;
+		int clear_ret;
+
+		entry = rb_entry(node, struct ulist_node, rb_node);
+		entry_start = entry->val;
+		entry_end = entry->aux;
+		entry_len = entry_end - entry_start + 1;
+
+		if (entry_start >= start + len)
+			break;
+		if (entry_start + entry_len <= start)
+			continue;
+		/*
+		 * Now the entry is in [start, start + len), revert the
+		 * EXTENT_QGROUP_RESERVED bit.
+		 */
+		clear_ret = clear_extent_bits(&inode->io_tree, entry_start,
+					      entry_end, EXTENT_QGROUP_RESERVED);
+		if (!ret && clear_ret < 0)
+			ret = clear_ret;
+
+		ulist_del(&reserved->range_changed, entry->val, entry->aux);
+		if (likely(reserved->bytes_changed >= entry_len)) {
+			reserved->bytes_changed -= entry_len;
+		} else {
+			WARN_ON(1);
+			reserved->bytes_changed = 0;
+		}
+	}
+
+	return ret;
 }
 
 /*
- * Reserve qgroup space for range [start, start + len).
+ * Try to free some space for qgroup.
  *
- * This function will either reserve space from related qgroups or doing
- * nothing if the range is already reserved.
+ * For qgroup, there are only 3 ways to free qgroup space:
+ * - Flush nodatacow write
+ *   Any nodatacow write will free its reserved data space at run_delalloc_range().
+ *   In theory, we should only flush nodatacow inodes, but it's not yet
+ *   possible, so we need to flush the whole root.
  *
- * Return 0 for successful reserve
- * Return <0 for error (including -EQUOT)
+ * - Wait for ordered extents
+ *   When ordered extents are finished, their reserved metadata is finally
+ *   converted to per_trans status, which can be freed by later commit
+ *   transaction.
  *
- * NOTE: this function may sleep for memory allocation.
- *       if btrfs_qgroup_reserve_data() is called multiple times with
- *       same @reserved, caller must ensure when error happens it's OK
- *       to free *ALL* reserved space.
+ * - Commit transaction
+ *   This would free the meta_per_trans space.
+ *   In theory this shouldn't provide much space, but any more qgroup space
+ *   is needed.
  */
-int btrfs_qgroup_reserve_data(struct inode *inode,
+static int try_flush_qgroup(struct btrfs_root *root)
+{
+	struct btrfs_trans_handle *trans;
+	int ret;
+
+	/* Can't hold an open transaction or we run the risk of deadlocking. */
+	ASSERT(current->journal_info == NULL);
+	if (WARN_ON(current->journal_info))
+		return 0;
+
+	/*
+	 * We don't want to run flush again and again, so if there is a running
+	 * one, we won't try to start a new flush, but exit directly.
+	 */
+	if (test_and_set_bit(BTRFS_ROOT_QGROUP_FLUSHING, &root->state)) {
+		wait_event(root->qgroup_flush_wait,
+			!test_bit(BTRFS_ROOT_QGROUP_FLUSHING, &root->state));
+		return 0;
+	}
+
+	ret = btrfs_start_delalloc_snapshot(root, true);
+	if (ret < 0)
+		goto out;
+	btrfs_wait_ordered_extents(root, U64_MAX, 0, (u64)-1);
+
+	trans = btrfs_join_transaction(root);
+	if (IS_ERR(trans)) {
+		ret = PTR_ERR(trans);
+		goto out;
+	}
+
+	ret = btrfs_commit_transaction(trans);
+out:
+	clear_bit(BTRFS_ROOT_QGROUP_FLUSHING, &root->state);
+	wake_up(&root->qgroup_flush_wait);
+	return ret;
+}
+
+static int qgroup_reserve_data(struct btrfs_inode *inode,
 			struct extent_changeset **reserved_ret, u64 start,
 			u64 len)
 {
-	struct btrfs_root *root = BTRFS_I(inode)->root;
-	struct ulist_node *unode;
-	struct ulist_iterator uiter;
+	struct btrfs_root *root = inode->root;
 	struct extent_changeset *reserved;
+	bool new_reserved = false;
 	u64 orig_reserved;
 	u64 to_reserve;
 	int ret;
@@ -3402,6 +3735,7 @@ int btrfs_qgroup_reserve_data(struct inode *inode,
 	if (WARN_ON(!reserved_ret))
 		return -EINVAL;
 	if (!*reserved_ret) {
+		new_reserved = true;
 		*reserved_ret = extent_changeset_alloc();
 		if (!*reserved_ret)
 			return -ENOMEM;
@@ -3409,15 +3743,15 @@ int btrfs_qgroup_reserve_data(struct inode *inode,
 	reserved = *reserved_ret;
 	/* Record already reserved space */
 	orig_reserved = reserved->bytes_changed;
-	ret = set_record_extent_bits(&BTRFS_I(inode)->io_tree, start,
+	ret = set_record_extent_bits(&inode->io_tree, start,
 			start + len -1, EXTENT_QGROUP_RESERVED, reserved);
 
 	/* Newly reserved space */
 	to_reserve = reserved->bytes_changed - orig_reserved;
-	trace_btrfs_qgroup_reserve_data(inode, start, len,
+	trace_btrfs_qgroup_reserve_data(&inode->vfs_inode, start, len,
 					to_reserve, QGROUP_RESERVE);
 	if (ret < 0)
-		goto cleanup;
+		goto out;
 	ret = qgroup_reserve(root, to_reserve, true, BTRFS_QGROUP_RSV_DATA);
 	if (ret < 0)
 		goto cleanup;
@@ -3425,23 +3759,48 @@ int btrfs_qgroup_reserve_data(struct inode *inode,
 	return ret;
 
 cleanup:
-	/* cleanup *ALL* already reserved ranges */
-	ULIST_ITER_INIT(&uiter);
-	while ((unode = ulist_next(&reserved->range_changed, &uiter)))
-		clear_extent_bit(&BTRFS_I(inode)->io_tree, unode->val,
-				 unode->aux, EXTENT_QGROUP_RESERVED, 0, 0, NULL);
-	/* Also free data bytes of already reserved one */
-	btrfs_qgroup_free_refroot(root->fs_info, root->root_key.objectid,
-				  orig_reserved, BTRFS_QGROUP_RSV_DATA);
-	extent_changeset_release(reserved);
+	qgroup_unreserve_range(inode, reserved, start, len);
+out:
+	if (new_reserved) {
+		extent_changeset_free(reserved);
+		*reserved_ret = NULL;
+	}
 	return ret;
 }
 
+/*
+ * Reserve qgroup space for range [start, start + len).
+ *
+ * This function will either reserve space from related qgroups or do nothing
+ * if the range is already reserved.
+ *
+ * Return 0 for successful reservation
+ * Return <0 for error (including -EQUOT)
+ *
+ * NOTE: This function may sleep for memory allocation, dirty page flushing and
+ *	 commit transaction. So caller should not hold any dirty page locked.
+ */
+int btrfs_qgroup_reserve_data(struct btrfs_inode *inode,
+			struct extent_changeset **reserved_ret, u64 start,
+			u64 len)
+{
+	int ret;
+
+	ret = qgroup_reserve_data(inode, reserved_ret, start, len);
+	if (ret <= 0 && ret != -EDQUOT)
+		return ret;
+
+	ret = try_flush_qgroup(inode->root);
+	if (ret < 0)
+		return ret;
+	return qgroup_reserve_data(inode, reserved_ret, start, len);
+}
+
 /* Free ranges specified by @reserved, normally in error path */
-static int qgroup_free_reserved_data(struct inode *inode,
+static int qgroup_free_reserved_data(struct btrfs_inode *inode,
 			struct extent_changeset *reserved, u64 start, u64 len)
 {
-	struct btrfs_root *root = BTRFS_I(inode)->root;
+	struct btrfs_root *root = inode->root;
 	struct ulist_node *unode;
 	struct ulist_iterator uiter;
 	struct extent_changeset changeset;
@@ -3477,8 +3836,8 @@ static int qgroup_free_reserved_data(struct inode *inode,
 		 * EXTENT_QGROUP_RESERVED, we won't double free.
 		 * So not need to rush.
 		 */
-		ret = clear_record_extent_bits(&BTRFS_I(inode)->io_tree,
-				free_start, free_start + free_len - 1,
+		ret = clear_record_extent_bits(&inode->io_tree, free_start,
+				free_start + free_len - 1,
 				EXTENT_QGROUP_RESERVED, &changeset);
 		if (ret < 0)
 			goto out;
@@ -3492,7 +3851,7 @@ out:
 	return ret;
 }
 
-static int __btrfs_qgroup_release_data(struct inode *inode,
+static int __btrfs_qgroup_release_data(struct btrfs_inode *inode,
 			struct extent_changeset *reserved, u64 start, u64 len,
 			int free)
 {
@@ -3500,8 +3859,7 @@ static int __btrfs_qgroup_release_data(struct inode *inode,
 	int trace_op = QGROUP_RELEASE;
 	int ret;
 
-	if (!test_bit(BTRFS_FS_QUOTA_ENABLED,
-		      &BTRFS_I(inode)->root->fs_info->flags))
+	if (!test_bit(BTRFS_FS_QUOTA_ENABLED, &inode->root->fs_info->flags))
 		return 0;
 
 	/* In release case, we shouldn't have @reserved */
@@ -3509,18 +3867,18 @@ static int __btrfs_qgroup_release_data(struct inode *inode,
 	if (free && reserved)
 		return qgroup_free_reserved_data(inode, reserved, start, len);
 	extent_changeset_init(&changeset);
-	ret = clear_record_extent_bits(&BTRFS_I(inode)->io_tree, start, 
-			start + len -1, EXTENT_QGROUP_RESERVED, &changeset);
+	ret = clear_record_extent_bits(&inode->io_tree, start, start + len -1,
+				       EXTENT_QGROUP_RESERVED, &changeset);
 	if (ret < 0)
 		goto out;
 
 	if (free)
 		trace_op = QGROUP_FREE;
-	trace_btrfs_qgroup_release_data(inode, start, len,
+	trace_btrfs_qgroup_release_data(&inode->vfs_inode, start, len,
 					changeset.bytes_changed, trace_op);
 	if (free)
-		btrfs_qgroup_free_refroot(BTRFS_I(inode)->root->fs_info,
-				BTRFS_I(inode)->root->root_key.objectid,
+		btrfs_qgroup_free_refroot(inode->root->fs_info,
+				inode->root->root_key.objectid,
 				changeset.bytes_changed, BTRFS_QGROUP_RSV_DATA);
 	ret = changeset.bytes_changed;
 out:
@@ -3540,7 +3898,7 @@ out:
  *
  * NOTE: This function may sleep for memory allocation.
  */
-int btrfs_qgroup_free_data(struct inode *inode,
+int btrfs_qgroup_free_data(struct btrfs_inode *inode,
 			struct extent_changeset *reserved, u64 start, u64 len)
 {
 	return __btrfs_qgroup_release_data(inode, reserved, start, len, 1);
@@ -3561,7 +3919,7 @@ int btrfs_qgroup_free_data(struct inode *inode,
  *
  * NOTE: This function may sleep for memory allocation.
  */
-int btrfs_qgroup_release_data(struct inode *inode, u64 start, u64 len)
+int btrfs_qgroup_release_data(struct btrfs_inode *inode, u64 start, u64 len)
 {
 	return __btrfs_qgroup_release_data(inode, NULL, start, len, 0);
 }
@@ -3606,8 +3964,8 @@ static int sub_root_meta_rsv(struct btrfs_root *root, int num_bytes,
 	return num_bytes;
 }
 
-int __btrfs_qgroup_reserve_meta(struct btrfs_root *root, int num_bytes,
-				enum btrfs_qgroup_rsv_type type, bool enforce)
+int btrfs_qgroup_reserve_meta(struct btrfs_root *root, int num_bytes,
+			      enum btrfs_qgroup_rsv_type type, bool enforce)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	int ret;
@@ -3633,6 +3991,22 @@ int __btrfs_qgroup_reserve_meta(struct btrfs_root *root, int num_bytes,
 	return ret;
 }
 
+int __btrfs_qgroup_reserve_meta(struct btrfs_root *root, int num_bytes,
+				enum btrfs_qgroup_rsv_type type, bool enforce,
+				bool noflush)
+{
+	int ret;
+
+	ret = btrfs_qgroup_reserve_meta(root, num_bytes, type, enforce);
+	if ((ret <= 0 && ret != -EDQUOT) || noflush)
+		return ret;
+
+	ret = try_flush_qgroup(root);
+	if (ret < 0)
+		return ret;
+	return btrfs_qgroup_reserve_meta(root, num_bytes, type, enforce);
+}
+
 void btrfs_qgroup_free_meta_all_pertrans(struct btrfs_root *root)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
@@ -3732,7 +4106,7 @@ void btrfs_qgroup_convert_reserved_meta(struct btrfs_root *root, int num_bytes)
  * Check qgroup reserved space leaking, normally at destroy inode
  * time
  */
-void btrfs_qgroup_check_reserved_leak(struct inode *inode)
+void btrfs_qgroup_check_reserved_leak(struct btrfs_inode *inode)
 {
 	struct extent_changeset changeset;
 	struct ulist_node *unode;
@@ -3740,19 +4114,19 @@ void btrfs_qgroup_check_reserved_leak(struct inode *inode)
 	int ret;
 
 	extent_changeset_init(&changeset);
-	ret = clear_record_extent_bits(&BTRFS_I(inode)->io_tree, 0, (u64)-1,
+	ret = clear_record_extent_bits(&inode->io_tree, 0, (u64)-1,
 			EXTENT_QGROUP_RESERVED, &changeset);
 
 	WARN_ON(ret < 0);
 	if (WARN_ON(changeset.bytes_changed)) {
 		ULIST_ITER_INIT(&iter);
 		while ((unode = ulist_next(&changeset.range_changed, &iter))) {
-			btrfs_warn(BTRFS_I(inode)->root->fs_info,
-				"leaking qgroup reserved space, ino: %lu, start: %llu, end: %llu",
-				inode->i_ino, unode->val, unode->aux);
+			btrfs_warn(inode->root->fs_info,
+		"leaking qgroup reserved space, ino: %llu, start: %llu, end: %llu",
+				btrfs_ino(inode), unode->val, unode->aux);
 		}
-		btrfs_qgroup_free_refroot(BTRFS_I(inode)->root->fs_info,
-				BTRFS_I(inode)->root->root_key.objectid,
+		btrfs_qgroup_free_refroot(inode->root->fs_info,
+				inode->root->root_key.objectid,
 				changeset.bytes_changed, BTRFS_QGROUP_RSV_DATA);
 
 	}
@@ -3910,8 +4284,7 @@ out_unlock:
 	spin_unlock(&blocks->lock);
 out:
 	if (ret < 0)
-		fs_info->qgroup_flags |=
-			BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
+		qgroup_mark_inconsistent(fs_info);
 	return ret;
 }
 
@@ -3975,7 +4348,7 @@ int btrfs_qgroup_trace_subtree_after_cow(struct btrfs_trans_handle *trans,
 	spin_unlock(&blocks->lock);
 
 	/* Read out reloc subtree root */
-	reloc_eb = read_tree_block(fs_info, block->reloc_bytenr,
+	reloc_eb = read_tree_block(fs_info, block->reloc_bytenr, 0,
 				   block->reloc_generation, block->level,
 				   &block->first_key);
 	if (IS_ERR(reloc_eb)) {
@@ -3998,7 +4371,7 @@ out:
 		btrfs_err_rl(fs_info,
 			     "failed to account subtree at bytenr %llu: %d",
 			     subvol_eb->start, ret);
-		fs_info->qgroup_flags |= BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
+		qgroup_mark_inconsistent(fs_info);
 	}
 	return ret;
 }
diff --git a/fs/btrfs/qgroup.h b/fs/btrfs/qgroup.h
index 1bc654459469..578c77e94200 100644
--- a/fs/btrfs/qgroup.h
+++ b/fs/btrfs/qgroup.h
@@ -8,6 +8,7 @@
 
 #include <linux/spinlock.h>
 #include <linux/rbtree.h>
+#include <linux/kobject.h>
 #include "ulist.h"
 #include "delayed-ref.h"
 
@@ -99,6 +100,9 @@
  *     subtree rescan for them.
  */
 
+#define BTRFS_QGROUP_RUNTIME_FLAG_CANCEL_RESCAN		(1UL << 3)
+#define BTRFS_QGROUP_RUNTIME_FLAG_NO_ACCOUNTING		(1UL << 4)
+
 /*
  * Record a dirty extent, and info qgroup to update quota on it
  * TODO: Use kmem cache to alloc it.
@@ -223,8 +227,18 @@ struct btrfs_qgroup {
 	 */
 	u64 old_refcnt;
 	u64 new_refcnt;
+
+	/*
+	 * Sysfs kobjectid
+	 */
+	struct kobject kobj;
 };
 
+static inline u64 btrfs_qgroup_subvolid(u64 qgroupid)
+{
+	return (qgroupid & ((1ULL << BTRFS_QGROUP_LEVEL_SHIFT) - 1));
+}
+
 /*
  * For qgroup event trace points only
  */
@@ -287,7 +301,7 @@ int btrfs_qgroup_trace_extent_nolock(
  * using current root, then we can move all expensive backref walk out of
  * transaction committing, but not now as qgroup accounting will be wrong again.
  */
-int btrfs_qgroup_trace_extent_post(struct btrfs_fs_info *fs_info,
+int btrfs_qgroup_trace_extent_post(struct btrfs_trans_handle *trans,
 				   struct btrfs_qgroup_extent_record *qrecord);
 
 /*
@@ -344,26 +358,32 @@ int btrfs_verify_qgroup_counts(struct btrfs_fs_info *fs_info, u64 qgroupid,
 #endif
 
 /* New io_tree based accurate qgroup reserve API */
-int btrfs_qgroup_reserve_data(struct inode *inode,
+int btrfs_qgroup_reserve_data(struct btrfs_inode *inode,
 			struct extent_changeset **reserved, u64 start, u64 len);
-int btrfs_qgroup_release_data(struct inode *inode, u64 start, u64 len);
-int btrfs_qgroup_free_data(struct inode *inode,
-			struct extent_changeset *reserved, u64 start, u64 len);
-
+int btrfs_qgroup_release_data(struct btrfs_inode *inode, u64 start, u64 len);
+int btrfs_qgroup_free_data(struct btrfs_inode *inode,
+			   struct extent_changeset *reserved, u64 start,
+			   u64 len);
+int btrfs_qgroup_reserve_meta(struct btrfs_root *root, int num_bytes,
+			      enum btrfs_qgroup_rsv_type type, bool enforce);
 int __btrfs_qgroup_reserve_meta(struct btrfs_root *root, int num_bytes,
-				enum btrfs_qgroup_rsv_type type, bool enforce);
+				enum btrfs_qgroup_rsv_type type, bool enforce,
+				bool noflush);
 /* Reserve metadata space for pertrans and prealloc type */
 static inline int btrfs_qgroup_reserve_meta_pertrans(struct btrfs_root *root,
 				int num_bytes, bool enforce)
 {
 	return __btrfs_qgroup_reserve_meta(root, num_bytes,
-			BTRFS_QGROUP_RSV_META_PERTRANS, enforce);
+					   BTRFS_QGROUP_RSV_META_PERTRANS,
+					   enforce, false);
 }
 static inline int btrfs_qgroup_reserve_meta_prealloc(struct btrfs_root *root,
-				int num_bytes, bool enforce)
+						     int num_bytes, bool enforce,
+						     bool noflush)
 {
 	return __btrfs_qgroup_reserve_meta(root, num_bytes,
-			BTRFS_QGROUP_RSV_META_PREALLOC, enforce);
+					   BTRFS_QGROUP_RSV_META_PREALLOC,
+					   enforce, noflush);
 }
 
 void __btrfs_qgroup_free_meta(struct btrfs_root *root, int num_bytes,
@@ -399,7 +419,7 @@ void btrfs_qgroup_free_meta_all_pertrans(struct btrfs_root *root);
  */
 void btrfs_qgroup_convert_reserved_meta(struct btrfs_root *root, int num_bytes);
 
-void btrfs_qgroup_check_reserved_leak(struct inode *inode);
+void btrfs_qgroup_check_reserved_leak(struct btrfs_inode *inode);
 
 /* btrfs_qgroup_swapped_blocks related functions */
 void btrfs_qgroup_init_swapped_blocks(
@@ -415,5 +435,6 @@ int btrfs_qgroup_add_swapped_blocks(struct btrfs_trans_handle *trans,
 int btrfs_qgroup_trace_subtree_after_cow(struct btrfs_trans_handle *trans,
 		struct btrfs_root *root, struct extent_buffer *eb);
 void btrfs_qgroup_destroy_extent_records(struct btrfs_transaction *trans);
+bool btrfs_check_quota_leak(struct btrfs_fs_info *fs_info);
 
 #endif
diff --git a/fs/btrfs/raid56.c b/fs/btrfs/raid56.c
index a8e53c8e7b01..82c8e991300e 100644
--- a/fs/btrfs/raid56.c
+++ b/fs/btrfs/raid56.c
@@ -13,6 +13,7 @@
 #include <linux/list_sort.h>
 #include <linux/raid/xor.h>
 #include <linux/mm.h>
+#include "misc.h"
 #include "ctree.h"
 #include "disk-io.h"
 #include "volumes.h"
@@ -51,133 +52,21 @@ struct btrfs_stripe_hash_table {
 	struct btrfs_stripe_hash table[];
 };
 
-enum btrfs_rbio_ops {
-	BTRFS_RBIO_WRITE,
-	BTRFS_RBIO_READ_REBUILD,
-	BTRFS_RBIO_PARITY_SCRUB,
-	BTRFS_RBIO_REBUILD_MISSING,
-};
-
-struct btrfs_raid_bio {
-	struct btrfs_fs_info *fs_info;
-	struct btrfs_bio *bbio;
-
-	/* while we're doing rmw on a stripe
-	 * we put it into a hash table so we can
-	 * lock the stripe and merge more rbios
-	 * into it.
-	 */
-	struct list_head hash_list;
-
-	/*
-	 * LRU list for the stripe cache
-	 */
-	struct list_head stripe_cache;
-
-	/*
-	 * for scheduling work in the helper threads
-	 */
-	struct btrfs_work work;
-
-	/*
-	 * bio list and bio_list_lock are used
-	 * to add more bios into the stripe
-	 * in hopes of avoiding the full rmw
-	 */
-	struct bio_list bio_list;
-	spinlock_t bio_list_lock;
-
-	/* also protected by the bio_list_lock, the
-	 * plug list is used by the plugging code
-	 * to collect partial bios while plugged.  The
-	 * stripe locking code also uses it to hand off
-	 * the stripe lock to the next pending IO
-	 */
-	struct list_head plug_list;
-
-	/*
-	 * flags that tell us if it is safe to
-	 * merge with this bio
-	 */
-	unsigned long flags;
-
-	/* size of each individual stripe on disk */
-	int stripe_len;
-
-	/* number of data stripes (no p/q) */
-	int nr_data;
-
-	int real_stripes;
-
-	int stripe_npages;
-	/*
-	 * set if we're doing a parity rebuild
-	 * for a read from higher up, which is handled
-	 * differently from a parity rebuild as part of
-	 * rmw
-	 */
-	enum btrfs_rbio_ops operation;
-
-	/* first bad stripe */
-	int faila;
-
-	/* second bad stripe (for raid6 use) */
-	int failb;
-
-	int scrubp;
-	/*
-	 * number of pages needed to represent the full
-	 * stripe
-	 */
-	int nr_pages;
-
-	/*
-	 * size of all the bios in the bio_list.  This
-	 * helps us decide if the rbio maps to a full
-	 * stripe or not
-	 */
-	int bio_list_bytes;
-
-	int generic_bio_cnt;
-
-	refcount_t refs;
-
-	atomic_t stripes_pending;
-
-	atomic_t error;
-	/*
-	 * these are two arrays of pointers.  We allocate the
-	 * rbio big enough to hold them both and setup their
-	 * locations when the rbio is allocated
-	 */
-
-	/* pointers to pages that we allocated for
-	 * reading/writing stripes directly from the disk (including P/Q)
-	 */
-	struct page **stripe_pages;
-
-	/*
-	 * pointers to the pages in the bio_list.  Stored
-	 * here for faster lookup
-	 */
-	struct page **bio_pages;
-
-	/*
-	 * bitmap to record which horizontal stripe has data
-	 */
-	unsigned long *dbitmap;
-
-	/* allocated with real_stripes-many pointers for finish_*() calls */
-	void **finish_pointers;
-
-	/* allocated with stripe_npages-many bits for finish_*() calls */
-	unsigned long *finish_pbitmap;
+/*
+ * A bvec like structure to present a sector inside a page.
+ *
+ * Unlike bvec we don't need bvlen, as it's fixed to sectorsize.
+ */
+struct sector_ptr {
+	struct page *page;
+	unsigned int pgoff:24;
+	unsigned int uptodate:8;
 };
 
 static int __raid56_parity_recover(struct btrfs_raid_bio *rbio);
 static noinline void finish_rmw(struct btrfs_raid_bio *rbio);
-static void rmw_work(struct btrfs_work *work);
-static void read_rebuild_work(struct btrfs_work *work);
+static void rmw_work(struct work_struct *work);
+static void read_rebuild_work(struct work_struct *work);
 static int fail_bio_stripe(struct btrfs_raid_bio *rbio, struct bio *bio);
 static int fail_rbio_index(struct btrfs_raid_bio *rbio, int failed);
 static void __free_raid_bio(struct btrfs_raid_bio *rbio);
@@ -186,12 +75,12 @@ static int alloc_rbio_pages(struct btrfs_raid_bio *rbio);
 
 static noinline void finish_parity_scrub(struct btrfs_raid_bio *rbio,
 					 int need_check);
-static void scrub_parity_work(struct btrfs_work *work);
+static void scrub_parity_work(struct work_struct *work);
 
-static void start_async_work(struct btrfs_raid_bio *rbio, btrfs_func_t work_func)
+static void start_async_work(struct btrfs_raid_bio *rbio, work_func_t work_func)
 {
-	btrfs_init_work(&rbio->work, work_func, NULL, NULL);
-	btrfs_queue_work(rbio->fs_info->rmw_workers, &rbio->work);
+	INIT_WORK(&rbio->work, work_func);
+	queue_work(rbio->bioc->fs_info->rmw_workers, &rbio->work);
 }
 
 /*
@@ -206,7 +95,6 @@ int btrfs_alloc_stripe_hash_table(struct btrfs_fs_info *info)
 	struct btrfs_stripe_hash *h;
 	int num_entries = 1 << BTRFS_STRIPE_HASH_TABLE_BITS;
 	int i;
-	int table_size;
 
 	if (info->stripe_hash_table)
 		return 0;
@@ -218,8 +106,7 @@ int btrfs_alloc_stripe_hash_table(struct btrfs_fs_info *info)
 	 * Try harder to allocate and fallback to vmalloc to lower the chance
 	 * of a failing mount.
 	 */
-	table_size = sizeof(*table) + sizeof(*h) * num_entries;
-	table = kvzalloc(table_size, GFP_KERNEL);
+	table = kvzalloc(struct_size(table, table, num_entries), GFP_KERNEL);
 	if (!table)
 		return -ENOMEM;
 
@@ -235,14 +122,13 @@ int btrfs_alloc_stripe_hash_table(struct btrfs_fs_info *info)
 	}
 
 	x = cmpxchg(&info->stripe_hash_table, NULL, table);
-	if (x)
-		kvfree(x);
+	kvfree(x);
 	return 0;
 }
 
 /*
  * caching an rbio means to copy anything from the
- * bio_pages array into the stripe_pages array.  We
+ * bio_sectors array into the stripe_pages array.  We
  * use the page uptodate bit in the stripe cache array
  * to indicate if it has valid data
  *
@@ -252,26 +138,24 @@ int btrfs_alloc_stripe_hash_table(struct btrfs_fs_info *info)
 static void cache_rbio_pages(struct btrfs_raid_bio *rbio)
 {
 	int i;
-	char *s;
-	char *d;
 	int ret;
 
 	ret = alloc_rbio_pages(rbio);
 	if (ret)
 		return;
 
-	for (i = 0; i < rbio->nr_pages; i++) {
-		if (!rbio->bio_pages[i])
+	for (i = 0; i < rbio->nr_sectors; i++) {
+		/* Some range not covered by bio (partial write), skip it */
+		if (!rbio->bio_sectors[i].page)
 			continue;
 
-		s = kmap(rbio->bio_pages[i]);
-		d = kmap(rbio->stripe_pages[i]);
-
-		copy_page(d, s);
-
-		kunmap(rbio->bio_pages[i]);
-		kunmap(rbio->stripe_pages[i]);
-		SetPageUptodate(rbio->stripe_pages[i]);
+		ASSERT(rbio->stripe_sectors[i].page);
+		memcpy_page(rbio->stripe_sectors[i].page,
+			    rbio->stripe_sectors[i].pgoff,
+			    rbio->bio_sectors[i].page,
+			    rbio->bio_sectors[i].pgoff,
+			    rbio->bioc->fs_info->sectorsize);
+		rbio->stripe_sectors[i].uptodate = 1;
 	}
 	set_bit(RBIO_CACHE_READY_BIT, &rbio->flags);
 }
@@ -281,7 +165,7 @@ static void cache_rbio_pages(struct btrfs_raid_bio *rbio)
  */
 static int rbio_bucket(struct btrfs_raid_bio *rbio)
 {
-	u64 num = rbio->bbio->raid_map[0];
+	u64 num = rbio->bioc->raid_map[0];
 
 	/*
 	 * we shift down quite a bit.  We're using byte
@@ -294,32 +178,86 @@ static int rbio_bucket(struct btrfs_raid_bio *rbio)
 	return hash_64(num >> 16, BTRFS_STRIPE_HASH_TABLE_BITS);
 }
 
+static bool full_page_sectors_uptodate(struct btrfs_raid_bio *rbio,
+				       unsigned int page_nr)
+{
+	const u32 sectorsize = rbio->bioc->fs_info->sectorsize;
+	const u32 sectors_per_page = PAGE_SIZE / sectorsize;
+	int i;
+
+	ASSERT(page_nr < rbio->nr_pages);
+
+	for (i = sectors_per_page * page_nr;
+	     i < sectors_per_page * page_nr + sectors_per_page;
+	     i++) {
+		if (!rbio->stripe_sectors[i].uptodate)
+			return false;
+	}
+	return true;
+}
+
 /*
- * stealing an rbio means taking all the uptodate pages from the stripe
- * array in the source rbio and putting them into the destination rbio
+ * Update the stripe_sectors[] array to use correct page and pgoff
+ *
+ * Should be called every time any page pointer in stripes_pages[] got modified.
+ */
+static void index_stripe_sectors(struct btrfs_raid_bio *rbio)
+{
+	const u32 sectorsize = rbio->bioc->fs_info->sectorsize;
+	u32 offset;
+	int i;
+
+	for (i = 0, offset = 0; i < rbio->nr_sectors; i++, offset += sectorsize) {
+		int page_index = offset >> PAGE_SHIFT;
+
+		ASSERT(page_index < rbio->nr_pages);
+		rbio->stripe_sectors[i].page = rbio->stripe_pages[page_index];
+		rbio->stripe_sectors[i].pgoff = offset_in_page(offset);
+	}
+}
+
+static void steal_rbio_page(struct btrfs_raid_bio *src,
+			    struct btrfs_raid_bio *dest, int page_nr)
+{
+	const u32 sectorsize = src->bioc->fs_info->sectorsize;
+	const u32 sectors_per_page = PAGE_SIZE / sectorsize;
+	int i;
+
+	if (dest->stripe_pages[page_nr])
+		__free_page(dest->stripe_pages[page_nr]);
+	dest->stripe_pages[page_nr] = src->stripe_pages[page_nr];
+	src->stripe_pages[page_nr] = NULL;
+
+	/* Also update the sector->uptodate bits. */
+	for (i = sectors_per_page * page_nr;
+	     i < sectors_per_page * page_nr + sectors_per_page; i++)
+		dest->stripe_sectors[i].uptodate = true;
+}
+
+/*
+ * Stealing an rbio means taking all the uptodate pages from the stripe array
+ * in the source rbio and putting them into the destination rbio.
+ *
+ * This will also update the involved stripe_sectors[] which are referring to
+ * the old pages.
  */
 static void steal_rbio(struct btrfs_raid_bio *src, struct btrfs_raid_bio *dest)
 {
 	int i;
 	struct page *s;
-	struct page *d;
 
 	if (!test_bit(RBIO_CACHE_READY_BIT, &src->flags))
 		return;
 
 	for (i = 0; i < dest->nr_pages; i++) {
 		s = src->stripe_pages[i];
-		if (!s || !PageUptodate(s)) {
+		if (!s || !full_page_sectors_uptodate(src, i))
 			continue;
-		}
 
-		d = dest->stripe_pages[i];
-		if (d)
-			__free_page(d);
-
-		dest->stripe_pages[i] = s;
-		src->stripe_pages[i] = NULL;
+		steal_rbio_page(src, dest, i);
 	}
+	index_stripe_sectors(dest);
+	index_stripe_sectors(src);
 }
 
 /*
@@ -334,7 +272,9 @@ static void merge_rbio(struct btrfs_raid_bio *dest,
 {
 	bio_list_merge(&dest->bio_list, &victim->bio_list);
 	dest->bio_list_bytes += victim->bio_list_bytes;
-	dest->generic_bio_cnt += victim->generic_bio_cnt;
+	/* Also inherit the bitmaps from @victim. */
+	bitmap_or(&dest->dbitmap, &victim->dbitmap, &dest->dbitmap,
+		  dest->stripe_nsectors);
 	bio_list_init(&victim->bio_list);
 }
 
@@ -355,7 +295,7 @@ static void __remove_rbio_from_cache(struct btrfs_raid_bio *rbio)
 	if (!test_bit(RBIO_CACHE_BIT, &rbio->flags))
 		return;
 
-	table = rbio->fs_info->stripe_hash_table;
+	table = rbio->bioc->fs_info->stripe_hash_table;
 	h = table->table + bucket;
 
 	/* hold the lock for the bucket because we may be
@@ -410,7 +350,7 @@ static void remove_rbio_from_cache(struct btrfs_raid_bio *rbio)
 	if (!test_bit(RBIO_CACHE_BIT, &rbio->flags))
 		return;
 
-	table = rbio->fs_info->stripe_hash_table;
+	table = rbio->bioc->fs_info->stripe_hash_table;
 
 	spin_lock_irqsave(&table->cache_lock, flags);
 	__remove_rbio_from_cache(rbio);
@@ -470,7 +410,7 @@ static void cache_rbio(struct btrfs_raid_bio *rbio)
 	if (!test_bit(RBIO_CACHE_READY_BIT, &rbio->flags))
 		return;
 
-	table = rbio->fs_info->stripe_hash_table;
+	table = rbio->bioc->fs_info->stripe_hash_table;
 
 	spin_lock_irqsave(&table->cache_lock, flags);
 	spin_lock(&rbio->bio_list_lock);
@@ -533,9 +473,9 @@ static int rbio_is_full(struct btrfs_raid_bio *rbio)
 	int ret = 1;
 
 	spin_lock_irqsave(&rbio->bio_list_lock, flags);
-	if (size != rbio->nr_data * rbio->stripe_len)
+	if (size != rbio->nr_data * BTRFS_STRIPE_LEN)
 		ret = 0;
-	BUG_ON(size > rbio->nr_data * rbio->stripe_len);
+	BUG_ON(size > rbio->nr_data * BTRFS_STRIPE_LEN);
 	spin_unlock_irqrestore(&rbio->bio_list_lock, flags);
 
 	return ret;
@@ -569,8 +509,7 @@ static int rbio_can_merge(struct btrfs_raid_bio *last,
 	    test_bit(RBIO_CACHE_BIT, &cur->flags))
 		return 0;
 
-	if (last->bbio->raid_map[0] !=
-	    cur->bbio->raid_map[0])
+	if (last->bioc->raid_map[0] != cur->bioc->raid_map[0])
 		return 0;
 
 	/* we can't merge with different operations */
@@ -612,39 +551,39 @@ static int rbio_can_merge(struct btrfs_raid_bio *last,
 	return 1;
 }
 
-static int rbio_stripe_page_index(struct btrfs_raid_bio *rbio, int stripe,
-				  int index)
+static unsigned int rbio_stripe_sector_index(const struct btrfs_raid_bio *rbio,
+					     unsigned int stripe_nr,
+					     unsigned int sector_nr)
 {
-	return stripe * rbio->stripe_npages + index;
+	ASSERT(stripe_nr < rbio->real_stripes);
+	ASSERT(sector_nr < rbio->stripe_nsectors);
+
+	return stripe_nr * rbio->stripe_nsectors + sector_nr;
 }
 
-/*
- * these are just the pages from the rbio array, not from anything
- * the FS sent down to us
- */
-static struct page *rbio_stripe_page(struct btrfs_raid_bio *rbio, int stripe,
-				     int index)
+/* Return a sector from rbio->stripe_sectors, not from the bio list */
+static struct sector_ptr *rbio_stripe_sector(const struct btrfs_raid_bio *rbio,
+					     unsigned int stripe_nr,
+					     unsigned int sector_nr)
 {
-	return rbio->stripe_pages[rbio_stripe_page_index(rbio, stripe, index)];
+	return &rbio->stripe_sectors[rbio_stripe_sector_index(rbio, stripe_nr,
+							      sector_nr)];
 }
 
-/*
- * helper to index into the pstripe
- */
-static struct page *rbio_pstripe_page(struct btrfs_raid_bio *rbio, int index)
+/* Grab a sector inside P stripe */
+static struct sector_ptr *rbio_pstripe_sector(const struct btrfs_raid_bio *rbio,
+					      unsigned int sector_nr)
 {
-	return rbio_stripe_page(rbio, rbio->nr_data, index);
+	return rbio_stripe_sector(rbio, rbio->nr_data, sector_nr);
 }
 
-/*
- * helper to index into the qstripe, returns null
- * if there is no qstripe
- */
-static struct page *rbio_qstripe_page(struct btrfs_raid_bio *rbio, int index)
+/* Grab a sector inside Q stripe, return NULL if not RAID6 */
+static struct sector_ptr *rbio_qstripe_sector(const struct btrfs_raid_bio *rbio,
+					      unsigned int sector_nr)
 {
 	if (rbio->nr_data + 1 == rbio->real_stripes)
 		return NULL;
-	return rbio_stripe_page(rbio, rbio->nr_data + 1, index);
+	return rbio_stripe_sector(rbio, rbio->nr_data + 1, sector_nr);
 }
 
 /*
@@ -679,11 +618,11 @@ static noinline int lock_stripe_add(struct btrfs_raid_bio *rbio)
 	struct btrfs_raid_bio *cache_drop = NULL;
 	int ret = 0;
 
-	h = rbio->fs_info->stripe_hash_table->table + rbio_bucket(rbio);
+	h = rbio->bioc->fs_info->stripe_hash_table->table + rbio_bucket(rbio);
 
 	spin_lock_irqsave(&h->lock, flags);
 	list_for_each_entry(cur, &h->hash_list, hash_list) {
-		if (cur->bbio->raid_map[0] != rbio->bbio->raid_map[0])
+		if (cur->bioc->raid_map[0] != rbio->bioc->raid_map[0])
 			continue;
 
 		spin_lock(&cur->bio_list_lock);
@@ -761,7 +700,7 @@ static noinline void unlock_stripe(struct btrfs_raid_bio *rbio)
 	int keep_cache = 0;
 
 	bucket = rbio_bucket(rbio);
-	h = rbio->fs_info->stripe_hash_table->table + bucket;
+	h = rbio->bioc->fs_info->stripe_hash_table->table + bucket;
 
 	if (list_empty(&rbio->plug_list))
 		cache_rbio(rbio);
@@ -848,7 +787,7 @@ static void __free_raid_bio(struct btrfs_raid_bio *rbio)
 		}
 	}
 
-	btrfs_put_bbio(rbio->bbio);
+	btrfs_put_bioc(rbio->bioc);
 	kfree(rbio);
 }
 
@@ -874,8 +813,12 @@ static void rbio_orig_end_io(struct btrfs_raid_bio *rbio, blk_status_t err)
 	struct bio *cur = bio_list_get(&rbio->bio_list);
 	struct bio *extra;
 
-	if (rbio->generic_bio_cnt)
-		btrfs_bio_counter_sub(rbio->fs_info, rbio->generic_bio_cnt);
+	/*
+	 * Clear the data bitmap, as the rbio may be cached for later usage.
+	 * do this before before unlock_stripe() so there will be no new bio
+	 * for this bio.
+	 */
+	bitmap_clear(&rbio->dbitmap, 0, rbio->stripe_nsectors);
 
 	/*
 	 * At this moment, rbio->bio_list is empty, however since rbio does not
@@ -916,54 +859,50 @@ static void raid_write_end_io(struct bio *bio)
 
 	/* OK, we have read all the stripes we need to. */
 	max_errors = (rbio->operation == BTRFS_RBIO_PARITY_SCRUB) ?
-		     0 : rbio->bbio->max_errors;
+		     0 : rbio->bioc->max_errors;
 	if (atomic_read(&rbio->error) > max_errors)
 		err = BLK_STS_IOERR;
 
 	rbio_orig_end_io(rbio, err);
 }
 
-/*
- * the read/modify/write code wants to use the original bio for
- * any pages it included, and then use the rbio for everything
- * else.  This function decides if a given index (stripe number)
- * and page number in that stripe fall inside the original bio
- * or the rbio.
+/**
+ * Get a sector pointer specified by its @stripe_nr and @sector_nr
  *
- * if you set bio_list_only, you'll get a NULL back for any ranges
- * that are outside the bio_list
+ * @rbio:               The raid bio
+ * @stripe_nr:          Stripe number, valid range [0, real_stripe)
+ * @sector_nr:		Sector number inside the stripe,
+ *			valid range [0, stripe_nsectors)
+ * @bio_list_only:      Whether to use sectors inside the bio list only.
  *
- * This doesn't take any refs on anything, you get a bare page pointer
- * and the caller must bump refs as required.
- *
- * You must call index_rbio_pages once before you can trust
- * the answers from this function.
+ * The read/modify/write code wants to reuse the original bio page as much
+ * as possible, and only use stripe_sectors as fallback.
  */
-static struct page *page_in_rbio(struct btrfs_raid_bio *rbio,
-				 int index, int pagenr, int bio_list_only)
+static struct sector_ptr *sector_in_rbio(struct btrfs_raid_bio *rbio,
+					 int stripe_nr, int sector_nr,
+					 bool bio_list_only)
 {
-	int chunk_page;
-	struct page *p = NULL;
+	struct sector_ptr *sector;
+	int index;
 
-	chunk_page = index * (rbio->stripe_len >> PAGE_SHIFT) + pagenr;
+	ASSERT(stripe_nr >= 0 && stripe_nr < rbio->real_stripes);
+	ASSERT(sector_nr >= 0 && sector_nr < rbio->stripe_nsectors);
+
+	index = stripe_nr * rbio->stripe_nsectors + sector_nr;
+	ASSERT(index >= 0 && index < rbio->nr_sectors);
 
 	spin_lock_irq(&rbio->bio_list_lock);
-	p = rbio->bio_pages[chunk_page];
+	sector = &rbio->bio_sectors[index];
+	if (sector->page || bio_list_only) {
+		/* Don't return sector without a valid page pointer */
+		if (!sector->page)
+			sector = NULL;
+		spin_unlock_irq(&rbio->bio_list_lock);
+		return sector;
+	}
 	spin_unlock_irq(&rbio->bio_list_lock);
 
-	if (p || bio_list_only)
-		return p;
-
-	return rbio->stripe_pages[chunk_page];
-}
-
-/*
- * number of pages we need for the entire stripe across all the
- * drives
- */
-static unsigned long rbio_nr_pages(unsigned long stripe_len, int nr_stripes)
-{
-	return DIV_ROUND_UP(stripe_len, PAGE_SIZE) * nr_stripes;
+	return &rbio->stripe_sectors[index];
 }
 
 /*
@@ -971,23 +910,30 @@ static unsigned long rbio_nr_pages(unsigned long stripe_len, int nr_stripes)
  * this does not allocate any pages for rbio->pages.
  */
 static struct btrfs_raid_bio *alloc_rbio(struct btrfs_fs_info *fs_info,
-					 struct btrfs_bio *bbio,
-					 u64 stripe_len)
-{
+					 struct btrfs_io_context *bioc)
+{
+	const unsigned int real_stripes = bioc->num_stripes - bioc->num_tgtdevs;
+	const unsigned int stripe_npages = BTRFS_STRIPE_LEN >> PAGE_SHIFT;
+	const unsigned int num_pages = stripe_npages * real_stripes;
+	const unsigned int stripe_nsectors =
+		BTRFS_STRIPE_LEN >> fs_info->sectorsize_bits;
+	const unsigned int num_sectors = stripe_nsectors * real_stripes;
 	struct btrfs_raid_bio *rbio;
-	int nr_data = 0;
-	int real_stripes = bbio->num_stripes - bbio->num_tgtdevs;
-	int num_pages = rbio_nr_pages(stripe_len, real_stripes);
-	int stripe_npages = DIV_ROUND_UP(stripe_len, PAGE_SIZE);
 	void *p;
 
+	/* PAGE_SIZE must also be aligned to sectorsize for subpage support */
+	ASSERT(IS_ALIGNED(PAGE_SIZE, fs_info->sectorsize));
+	/*
+	 * Our current stripe len should be fixed to 64k thus stripe_nsectors
+	 * (at most 16) should be no larger than BITS_PER_LONG.
+	 */
+	ASSERT(stripe_nsectors <= BITS_PER_LONG);
+
 	rbio = kzalloc(sizeof(*rbio) +
 		       sizeof(*rbio->stripe_pages) * num_pages +
-		       sizeof(*rbio->bio_pages) * num_pages +
-		       sizeof(*rbio->finish_pointers) * real_stripes +
-		       sizeof(*rbio->dbitmap) * BITS_TO_LONGS(stripe_npages) +
-		       sizeof(*rbio->finish_pbitmap) *
-				BITS_TO_LONGS(stripe_npages),
+		       sizeof(*rbio->bio_sectors) * num_sectors +
+		       sizeof(*rbio->stripe_sectors) * num_sectors +
+		       sizeof(*rbio->finish_pointers) * real_stripes,
 		       GFP_NOFS);
 	if (!rbio)
 		return ERR_PTR(-ENOMEM);
@@ -997,12 +943,13 @@ static struct btrfs_raid_bio *alloc_rbio(struct btrfs_fs_info *fs_info,
 	spin_lock_init(&rbio->bio_list_lock);
 	INIT_LIST_HEAD(&rbio->stripe_cache);
 	INIT_LIST_HEAD(&rbio->hash_list);
-	rbio->bbio = bbio;
-	rbio->fs_info = fs_info;
-	rbio->stripe_len = stripe_len;
+	btrfs_get_bioc(bioc);
+	rbio->bioc = bioc;
 	rbio->nr_pages = num_pages;
+	rbio->nr_sectors = num_sectors;
 	rbio->real_stripes = real_stripes;
 	rbio->stripe_npages = stripe_npages;
+	rbio->stripe_nsectors = stripe_nsectors;
 	rbio->faila = -1;
 	rbio->failb = -1;
 	refcount_set(&rbio->refs, 1);
@@ -1010,8 +957,8 @@ static struct btrfs_raid_bio *alloc_rbio(struct btrfs_fs_info *fs_info,
 	atomic_set(&rbio->stripes_pending, 0);
 
 	/*
-	 * the stripe_pages, bio_pages, etc arrays point to the extra
-	 * memory we allocated past the end of the rbio
+	 * The stripe_pages, bio_sectors, etc arrays point to the extra memory
+	 * we allocated past the end of the rbio.
 	 */
 	p = rbio + 1;
 #define CONSUME_ALLOC(ptr, count)	do {				\
@@ -1019,80 +966,76 @@ static struct btrfs_raid_bio *alloc_rbio(struct btrfs_fs_info *fs_info,
 		p = (unsigned char *)p + sizeof(*(ptr)) * (count);	\
 	} while (0)
 	CONSUME_ALLOC(rbio->stripe_pages, num_pages);
-	CONSUME_ALLOC(rbio->bio_pages, num_pages);
+	CONSUME_ALLOC(rbio->bio_sectors, num_sectors);
+	CONSUME_ALLOC(rbio->stripe_sectors, num_sectors);
 	CONSUME_ALLOC(rbio->finish_pointers, real_stripes);
-	CONSUME_ALLOC(rbio->dbitmap, BITS_TO_LONGS(stripe_npages));
-	CONSUME_ALLOC(rbio->finish_pbitmap, BITS_TO_LONGS(stripe_npages));
 #undef  CONSUME_ALLOC
 
-	if (bbio->map_type & BTRFS_BLOCK_GROUP_RAID5)
-		nr_data = real_stripes - 1;
-	else if (bbio->map_type & BTRFS_BLOCK_GROUP_RAID6)
-		nr_data = real_stripes - 2;
-	else
-		BUG();
+	ASSERT(btrfs_nr_parity_stripes(bioc->map_type));
+	rbio->nr_data = real_stripes - btrfs_nr_parity_stripes(bioc->map_type);
 
-	rbio->nr_data = nr_data;
 	return rbio;
 }
 
 /* allocate pages for all the stripes in the bio, including parity */
 static int alloc_rbio_pages(struct btrfs_raid_bio *rbio)
 {
-	int i;
-	struct page *page;
+	int ret;
 
-	for (i = 0; i < rbio->nr_pages; i++) {
-		if (rbio->stripe_pages[i])
-			continue;
-		page = alloc_page(GFP_NOFS | __GFP_HIGHMEM);
-		if (!page)
-			return -ENOMEM;
-		rbio->stripe_pages[i] = page;
-	}
+	ret = btrfs_alloc_page_array(rbio->nr_pages, rbio->stripe_pages);
+	if (ret < 0)
+		return ret;
+	/* Mapping all sectors */
+	index_stripe_sectors(rbio);
 	return 0;
 }
 
 /* only allocate pages for p/q stripes */
 static int alloc_rbio_parity_pages(struct btrfs_raid_bio *rbio)
 {
-	int i;
-	struct page *page;
+	const int data_pages = rbio->nr_data * rbio->stripe_npages;
+	int ret;
 
-	i = rbio_stripe_page_index(rbio, rbio->nr_data, 0);
+	ret = btrfs_alloc_page_array(rbio->nr_pages - data_pages,
+				     rbio->stripe_pages + data_pages);
+	if (ret < 0)
+		return ret;
 
-	for (; i < rbio->nr_pages; i++) {
-		if (rbio->stripe_pages[i])
-			continue;
-		page = alloc_page(GFP_NOFS | __GFP_HIGHMEM);
-		if (!page)
-			return -ENOMEM;
-		rbio->stripe_pages[i] = page;
-	}
+	index_stripe_sectors(rbio);
 	return 0;
 }
 
 /*
- * add a single page from a specific stripe into our list of bios for IO
- * this will try to merge into existing bios if possible, and returns
- * zero if all went well.
+ * Add a single sector @sector into our list of bios for IO.
+ *
+ * Return 0 if everything went well.
+ * Return <0 for error.
  */
-static int rbio_add_io_page(struct btrfs_raid_bio *rbio,
-			    struct bio_list *bio_list,
-			    struct page *page,
-			    int stripe_nr,
-			    unsigned long page_index,
-			    unsigned long bio_max_len)
-{
+static int rbio_add_io_sector(struct btrfs_raid_bio *rbio,
+			      struct bio_list *bio_list,
+			      struct sector_ptr *sector,
+			      unsigned int stripe_nr,
+			      unsigned int sector_nr,
+			      enum req_op op)
+{
+	const u32 sectorsize = rbio->bioc->fs_info->sectorsize;
 	struct bio *last = bio_list->tail;
-	u64 last_end = 0;
 	int ret;
 	struct bio *bio;
-	struct btrfs_bio_stripe *stripe;
+	struct btrfs_io_stripe *stripe;
 	u64 disk_start;
 
-	stripe = &rbio->bbio->stripes[stripe_nr];
-	disk_start = stripe->physical + (page_index << PAGE_SHIFT);
+	/*
+	 * Note: here stripe_nr has taken device replace into consideration,
+	 * thus it can be larger than rbio->real_stripe.
+	 * So here we check against bioc->num_stripes, not rbio->real_stripes.
+	 */
+	ASSERT(stripe_nr >= 0 && stripe_nr < rbio->bioc->num_stripes);
+	ASSERT(sector_nr >= 0 && sector_nr < rbio->stripe_nsectors);
+	ASSERT(sector->page);
+
+	stripe = &rbio->bioc->stripes[stripe_nr];
+	disk_start = stripe->physical + sector_nr * sectorsize;
 
 	/* if the device is missing, just fail this stripe */
 	if (!stripe->dev->bdev)
@@ -1100,30 +1043,30 @@ static int rbio_add_io_page(struct btrfs_raid_bio *rbio,
 
 	/* see if we can add this page onto our existing bio */
 	if (last) {
-		last_end = (u64)last->bi_iter.bi_sector << 9;
+		u64 last_end = last->bi_iter.bi_sector << 9;
 		last_end += last->bi_iter.bi_size;
 
 		/*
 		 * we can't merge these if they are from different
 		 * devices or if they are not contiguous
 		 */
-		if (last_end == disk_start && stripe->dev->bdev &&
-		    !last->bi_status &&
-		    last->bi_disk == stripe->dev->bdev->bd_disk &&
-		    last->bi_partno == stripe->dev->bdev->bd_partno) {
-			ret = bio_add_page(last, page, PAGE_SIZE, 0);
-			if (ret == PAGE_SIZE)
+		if (last_end == disk_start && !last->bi_status &&
+		    last->bi_bdev == stripe->dev->bdev) {
+			ret = bio_add_page(last, sector->page, sectorsize,
+					   sector->pgoff);
+			if (ret == sectorsize)
 				return 0;
 		}
 	}
 
 	/* put a new bio on the list */
-	bio = btrfs_io_bio_alloc(bio_max_len >> PAGE_SHIFT ?: 1);
-	bio->bi_iter.bi_size = 0;
-	bio_set_dev(bio, stripe->dev->bdev);
+	bio = bio_alloc(stripe->dev->bdev,
+			max(BTRFS_STRIPE_LEN >> PAGE_SHIFT, 1),
+			op, GFP_NOFS);
 	bio->bi_iter.bi_sector = disk_start >> 9;
+	bio->bi_private = rbio;
 
-	bio_add_page(bio, page, PAGE_SIZE, 0);
+	bio_add_page(bio, sector->page, sectorsize, sector->pgoff);
 	bio_list_add(bio_list, bio);
 	return 0;
 }
@@ -1145,6 +1088,29 @@ static void validate_rbio_for_rmw(struct btrfs_raid_bio *rbio)
 	}
 }
 
+static void index_one_bio(struct btrfs_raid_bio *rbio, struct bio *bio)
+{
+	const u32 sectorsize = rbio->bioc->fs_info->sectorsize;
+	struct bio_vec bvec;
+	struct bvec_iter iter;
+	u32 offset = (bio->bi_iter.bi_sector << SECTOR_SHIFT) -
+		     rbio->bioc->raid_map[0];
+
+	bio_for_each_segment(bvec, bio, iter) {
+		u32 bvec_offset;
+
+		for (bvec_offset = 0; bvec_offset < bvec.bv_len;
+		     bvec_offset += sectorsize, offset += sectorsize) {
+			int index = offset / sectorsize;
+			struct sector_ptr *sector = &rbio->bio_sectors[index];
+
+			sector->page = bvec.bv_page;
+			sector->pgoff = bvec.bv_offset + bvec_offset;
+			ASSERT(sector->pgoff < PAGE_SIZE);
+		}
+	}
+}
+
 /*
  * helper function to walk our bio list and populate the bio_pages array with
  * the result.  This seems expensive, but it is faster than constantly
@@ -1156,29 +1122,40 @@ static void validate_rbio_for_rmw(struct btrfs_raid_bio *rbio)
 static void index_rbio_pages(struct btrfs_raid_bio *rbio)
 {
 	struct bio *bio;
-	u64 start;
-	unsigned long stripe_offset;
-	unsigned long page_index;
 
 	spin_lock_irq(&rbio->bio_list_lock);
-	bio_list_for_each(bio, &rbio->bio_list) {
-		struct bio_vec bvec;
-		struct bvec_iter iter;
-		int i = 0;
+	bio_list_for_each(bio, &rbio->bio_list)
+		index_one_bio(rbio, bio);
+
+	spin_unlock_irq(&rbio->bio_list_lock);
+}
+
+static void bio_get_trace_info(struct btrfs_raid_bio *rbio, struct bio *bio,
+			       struct raid56_bio_trace_info *trace_info)
+{
+	const struct btrfs_io_context *bioc = rbio->bioc;
+	int i;
 
-		start = (u64)bio->bi_iter.bi_sector << 9;
-		stripe_offset = start - rbio->bbio->raid_map[0];
-		page_index = stripe_offset >> PAGE_SHIFT;
+	ASSERT(bioc);
 
-		if (bio_flagged(bio, BIO_CLONED))
-			bio->bi_iter = btrfs_io_bio(bio)->iter;
+	/* We rely on bio->bi_bdev to find the stripe number. */
+	if (!bio->bi_bdev)
+		goto not_found;
 
-		bio_for_each_segment(bvec, bio, iter) {
-			rbio->bio_pages[page_index + i] = bvec.bv_page;
-			i++;
-		}
+	for (i = 0; i < bioc->num_stripes; i++) {
+		if (bio->bi_bdev != bioc->stripes[i].dev->bdev)
+			continue;
+		trace_info->stripe_nr = i;
+		trace_info->devid = bioc->stripes[i].dev->devid;
+		trace_info->offset = (bio->bi_iter.bi_sector << SECTOR_SHIFT) -
+				     bioc->stripes[i].physical;
+		return;
 	}
-	spin_unlock_irq(&rbio->bio_list_lock);
+
+not_found:
+	trace_info->devid = -1;
+	trace_info->offset = -1;
+	trace_info->stripe_nr = -1;
 }
 
 /*
@@ -1191,27 +1168,31 @@ static void index_rbio_pages(struct btrfs_raid_bio *rbio)
  */
 static noinline void finish_rmw(struct btrfs_raid_bio *rbio)
 {
-	struct btrfs_bio *bbio = rbio->bbio;
+	struct btrfs_io_context *bioc = rbio->bioc;
+	const u32 sectorsize = bioc->fs_info->sectorsize;
 	void **pointers = rbio->finish_pointers;
 	int nr_data = rbio->nr_data;
+	/* The total sector number inside the full stripe. */
+	int total_sector_nr;
 	int stripe;
-	int pagenr;
-	int p_stripe = -1;
-	int q_stripe = -1;
+	/* Sector number inside a stripe. */
+	int sectornr;
+	bool has_qstripe;
 	struct bio_list bio_list;
 	struct bio *bio;
 	int ret;
 
 	bio_list_init(&bio_list);
 
-	if (rbio->real_stripes - rbio->nr_data == 1) {
-		p_stripe = rbio->real_stripes - 1;
-	} else if (rbio->real_stripes - rbio->nr_data == 2) {
-		p_stripe = rbio->real_stripes - 2;
-		q_stripe = rbio->real_stripes - 1;
-	} else {
+	if (rbio->real_stripes - rbio->nr_data == 1)
+		has_qstripe = false;
+	else if (rbio->real_stripes - rbio->nr_data == 2)
+		has_qstripe = true;
+	else
 		BUG();
-	}
+
+	/* We should have at least one data sector. */
+	ASSERT(bitmap_weight(&rbio->dbitmap, rbio->stripe_nsectors));
 
 	/* at this point we either have a full stripe,
 	 * or we've read the full stripe from the drive.
@@ -1242,103 +1223,123 @@ static noinline void finish_rmw(struct btrfs_raid_bio *rbio)
 	else
 		clear_bit(RBIO_CACHE_READY_BIT, &rbio->flags);
 
-	for (pagenr = 0; pagenr < rbio->stripe_npages; pagenr++) {
-		struct page *p;
-		/* first collect one page from each data stripe */
+	for (sectornr = 0; sectornr < rbio->stripe_nsectors; sectornr++) {
+		struct sector_ptr *sector;
+
+		/* First collect one sector from each data stripe */
 		for (stripe = 0; stripe < nr_data; stripe++) {
-			p = page_in_rbio(rbio, stripe, pagenr, 0);
-			pointers[stripe] = kmap(p);
+			sector = sector_in_rbio(rbio, stripe, sectornr, 0);
+			pointers[stripe] = kmap_local_page(sector->page) +
+					   sector->pgoff;
 		}
 
-		/* then add the parity stripe */
-		p = rbio_pstripe_page(rbio, pagenr);
-		SetPageUptodate(p);
-		pointers[stripe++] = kmap(p);
-
-		if (q_stripe != -1) {
+		/* Then add the parity stripe */
+		sector = rbio_pstripe_sector(rbio, sectornr);
+		sector->uptodate = 1;
+		pointers[stripe++] = kmap_local_page(sector->page) + sector->pgoff;
 
+		if (has_qstripe) {
 			/*
-			 * raid6, add the qstripe and call the
-			 * library function to fill in our p/q
+			 * RAID6, add the qstripe and call the library function
+			 * to fill in our p/q
 			 */
-			p = rbio_qstripe_page(rbio, pagenr);
-			SetPageUptodate(p);
-			pointers[stripe++] = kmap(p);
+			sector = rbio_qstripe_sector(rbio, sectornr);
+			sector->uptodate = 1;
+			pointers[stripe++] = kmap_local_page(sector->page) +
+					     sector->pgoff;
 
-			raid6_call.gen_syndrome(rbio->real_stripes, PAGE_SIZE,
+			raid6_call.gen_syndrome(rbio->real_stripes, sectorsize,
 						pointers);
 		} else {
 			/* raid5 */
-			copy_page(pointers[nr_data], pointers[0]);
-			run_xor(pointers + 1, nr_data - 1, PAGE_SIZE);
+			memcpy(pointers[nr_data], pointers[0], sectorsize);
+			run_xor(pointers + 1, nr_data - 1, sectorsize);
 		}
-
-
-		for (stripe = 0; stripe < rbio->real_stripes; stripe++)
-			kunmap(page_in_rbio(rbio, stripe, pagenr, 0));
+		for (stripe = stripe - 1; stripe >= 0; stripe--)
+			kunmap_local(pointers[stripe]);
 	}
 
 	/*
-	 * time to start writing.  Make bios for everything from the
-	 * higher layers (the bio_list in our rbio) and our p/q.  Ignore
-	 * everything else.
+	 * Start writing.  Make bios for everything from the higher layers (the
+	 * bio_list in our rbio) and our P/Q.  Ignore everything else.
 	 */
-	for (stripe = 0; stripe < rbio->real_stripes; stripe++) {
-		for (pagenr = 0; pagenr < rbio->stripe_npages; pagenr++) {
-			struct page *page;
-			if (stripe < rbio->nr_data) {
-				page = page_in_rbio(rbio, stripe, pagenr, 1);
-				if (!page)
-					continue;
-			} else {
-			       page = rbio_stripe_page(rbio, stripe, pagenr);
-			}
+	for (total_sector_nr = 0; total_sector_nr < rbio->nr_sectors;
+	     total_sector_nr++) {
+		struct sector_ptr *sector;
+
+		stripe = total_sector_nr / rbio->stripe_nsectors;
+		sectornr = total_sector_nr % rbio->stripe_nsectors;
+
+		/* This vertical stripe has no data, skip it. */
+		if (!test_bit(sectornr, &rbio->dbitmap))
+			continue;
 
-			ret = rbio_add_io_page(rbio, &bio_list,
-				       page, stripe, pagenr, rbio->stripe_len);
-			if (ret)
-				goto cleanup;
+		if (stripe < rbio->nr_data) {
+			sector = sector_in_rbio(rbio, stripe, sectornr, 1);
+			if (!sector)
+				continue;
+		} else {
+			sector = rbio_stripe_sector(rbio, stripe, sectornr);
 		}
+
+		ret = rbio_add_io_sector(rbio, &bio_list, sector, stripe,
+					 sectornr, REQ_OP_WRITE);
+		if (ret)
+			goto cleanup;
 	}
 
-	if (likely(!bbio->num_tgtdevs))
+	if (likely(!bioc->num_tgtdevs))
 		goto write_data;
 
-	for (stripe = 0; stripe < rbio->real_stripes; stripe++) {
-		if (!bbio->tgtdev_map[stripe])
+	for (total_sector_nr = 0; total_sector_nr < rbio->nr_sectors;
+	     total_sector_nr++) {
+		struct sector_ptr *sector;
+
+		stripe = total_sector_nr / rbio->stripe_nsectors;
+		sectornr = total_sector_nr % rbio->stripe_nsectors;
+
+		if (!bioc->tgtdev_map[stripe]) {
+			/*
+			 * We can skip the whole stripe completely, note
+			 * total_sector_nr will be increased by one anyway.
+			 */
+			ASSERT(sectornr == 0);
+			total_sector_nr += rbio->stripe_nsectors - 1;
 			continue;
+		}
 
-		for (pagenr = 0; pagenr < rbio->stripe_npages; pagenr++) {
-			struct page *page;
-			if (stripe < rbio->nr_data) {
-				page = page_in_rbio(rbio, stripe, pagenr, 1);
-				if (!page)
-					continue;
-			} else {
-			       page = rbio_stripe_page(rbio, stripe, pagenr);
-			}
+		/* This vertical stripe has no data, skip it. */
+		if (!test_bit(sectornr, &rbio->dbitmap))
+			continue;
 
-			ret = rbio_add_io_page(rbio, &bio_list, page,
-					       rbio->bbio->tgtdev_map[stripe],
-					       pagenr, rbio->stripe_len);
-			if (ret)
-				goto cleanup;
+		if (stripe < rbio->nr_data) {
+			sector = sector_in_rbio(rbio, stripe, sectornr, 1);
+			if (!sector)
+				continue;
+		} else {
+			sector = rbio_stripe_sector(rbio, stripe, sectornr);
 		}
+
+		ret = rbio_add_io_sector(rbio, &bio_list, sector,
+					 rbio->bioc->tgtdev_map[stripe],
+					 sectornr, REQ_OP_WRITE);
+		if (ret)
+			goto cleanup;
 	}
 
 write_data:
 	atomic_set(&rbio->stripes_pending, bio_list_size(&bio_list));
 	BUG_ON(atomic_read(&rbio->stripes_pending) == 0);
 
-	while (1) {
-		bio = bio_list_pop(&bio_list);
-		if (!bio)
-			break;
-
-		bio->bi_private = rbio;
+	while ((bio = bio_list_pop(&bio_list))) {
 		bio->bi_end_io = raid_write_end_io;
-		bio->bi_opf = REQ_OP_WRITE;
 
+		if (trace_raid56_write_stripe_enabled()) {
+			struct raid56_bio_trace_info trace_info = { 0 };
+
+			bio_get_trace_info(rbio, bio, &trace_info);
+			trace_raid56_write_stripe(rbio, bio, &trace_info);
+		}
 		submit_bio(bio);
 	}
 	return;
@@ -1359,20 +1360,15 @@ static int find_bio_stripe(struct btrfs_raid_bio *rbio,
 			   struct bio *bio)
 {
 	u64 physical = bio->bi_iter.bi_sector;
-	u64 stripe_start;
 	int i;
-	struct btrfs_bio_stripe *stripe;
+	struct btrfs_io_stripe *stripe;
 
 	physical <<= 9;
 
-	for (i = 0; i < rbio->bbio->num_stripes; i++) {
-		stripe = &rbio->bbio->stripes[i];
-		stripe_start = stripe->physical;
-		if (physical >= stripe_start &&
-		    physical < stripe_start + rbio->stripe_len &&
-		    stripe->dev->bdev &&
-		    bio->bi_disk == stripe->dev->bdev->bd_disk &&
-		    bio->bi_partno == stripe->dev->bdev->bd_partno) {
+	for (i = 0; i < rbio->bioc->num_stripes; i++) {
+		stripe = &rbio->bioc->stripes[i];
+		if (in_range(physical, stripe->physical, BTRFS_STRIPE_LEN) &&
+		    stripe->dev->bdev && bio->bi_bdev == stripe->dev->bdev) {
 			return i;
 		}
 	}
@@ -1387,18 +1383,14 @@ static int find_bio_stripe(struct btrfs_raid_bio *rbio,
 static int find_logical_bio_stripe(struct btrfs_raid_bio *rbio,
 				   struct bio *bio)
 {
-	u64 logical = bio->bi_iter.bi_sector;
-	u64 stripe_start;
+	u64 logical = bio->bi_iter.bi_sector << 9;
 	int i;
 
-	logical <<= 9;
-
 	for (i = 0; i < rbio->nr_data; i++) {
-		stripe_start = rbio->bbio->raid_map[i];
-		if (logical >= stripe_start &&
-		    logical < stripe_start + rbio->stripe_len) {
+		u64 stripe_start = rbio->bioc->raid_map[i];
+
+		if (in_range(logical, stripe_start, BTRFS_STRIPE_LEN))
 			return i;
-		}
 	}
 	return -1;
 }
@@ -1450,56 +1442,89 @@ static int fail_bio_stripe(struct btrfs_raid_bio *rbio,
 }
 
 /*
+ * For subpage case, we can no longer set page Uptodate directly for
+ * stripe_pages[], thus we need to locate the sector.
+ */
+static struct sector_ptr *find_stripe_sector(struct btrfs_raid_bio *rbio,
+					     struct page *page,
+					     unsigned int pgoff)
+{
+	int i;
+
+	for (i = 0; i < rbio->nr_sectors; i++) {
+		struct sector_ptr *sector = &rbio->stripe_sectors[i];
+
+		if (sector->page == page && sector->pgoff == pgoff)
+			return sector;
+	}
+	return NULL;
+}
+
+/*
  * this sets each page in the bio uptodate.  It should only be used on private
  * rbio pages, nothing that comes in from the higher layers
  */
-static void set_bio_pages_uptodate(struct bio *bio)
+static void set_bio_pages_uptodate(struct btrfs_raid_bio *rbio, struct bio *bio)
 {
+	const u32 sectorsize = rbio->bioc->fs_info->sectorsize;
 	struct bio_vec *bvec;
 	struct bvec_iter_all iter_all;
 
 	ASSERT(!bio_flagged(bio, BIO_CLONED));
 
-	bio_for_each_segment_all(bvec, bio, iter_all)
-		SetPageUptodate(bvec->bv_page);
+	bio_for_each_segment_all(bvec, bio, iter_all) {
+		struct sector_ptr *sector;
+		int pgoff;
+
+		for (pgoff = bvec->bv_offset; pgoff - bvec->bv_offset < bvec->bv_len;
+		     pgoff += sectorsize) {
+			sector = find_stripe_sector(rbio, bvec->bv_page, pgoff);
+			ASSERT(sector);
+			if (sector)
+				sector->uptodate = 1;
+		}
+	}
 }
 
-/*
- * end io for the read phase of the rmw cycle.  All the bios here are physical
- * stripe bios we've read from the disk so we can recalculate the parity of the
- * stripe.
- *
- * This will usually kick off finish_rmw once all the bios are read in, but it
- * may trigger parity reconstruction if we had any errors along the way
- */
-static void raid_rmw_end_io(struct bio *bio)
+static void raid56_bio_end_io(struct bio *bio)
 {
 	struct btrfs_raid_bio *rbio = bio->bi_private;
 
 	if (bio->bi_status)
 		fail_bio_stripe(rbio, bio);
 	else
-		set_bio_pages_uptodate(bio);
+		set_bio_pages_uptodate(rbio, bio);
 
 	bio_put(bio);
 
-	if (!atomic_dec_and_test(&rbio->stripes_pending))
-		return;
+	if (atomic_dec_and_test(&rbio->stripes_pending))
+		queue_work(rbio->bioc->fs_info->endio_raid56_workers,
+			   &rbio->end_io_work);
+}
 
-	if (atomic_read(&rbio->error) > rbio->bbio->max_errors)
-		goto cleanup;
+/*
+ * End io handler for the read phase of the RMW cycle.  All the bios here are
+ * physical stripe bios we've read from the disk so we can recalculate the
+ * parity of the stripe.
+ *
+ * This will usually kick off finish_rmw once all the bios are read in, but it
+ * may trigger parity reconstruction if we had any errors along the way
+ */
+static void raid56_rmw_end_io_work(struct work_struct *work)
+{
+	struct btrfs_raid_bio *rbio =
+		container_of(work, struct btrfs_raid_bio, end_io_work);
+
+	if (atomic_read(&rbio->error) > rbio->bioc->max_errors) {
+		rbio_orig_end_io(rbio, BLK_STS_IOERR);
+		return;
+	}
 
 	/*
-	 * this will normally call finish_rmw to start our write
-	 * but if there are any failed stripes we'll reconstruct
-	 * from parity first
+	 * This will normally call finish_rmw to start our write but if there
+	 * are any failed stripes we'll reconstruct from parity first.
 	 */
 	validate_rbio_for_rmw(rbio);
-	return;
-
-cleanup:
-
-	rbio_orig_end_io(rbio, BLK_STS_IOERR);
 }
 
 /*
@@ -1510,9 +1535,9 @@ static int raid56_rmw_stripe(struct btrfs_raid_bio *rbio)
 {
 	int bios_to_read = 0;
 	struct bio_list bio_list;
+	const int nr_data_sectors = rbio->stripe_nsectors * rbio->nr_data;
 	int ret;
-	int pagenr;
-	int stripe;
+	int total_sector_nr;
 	struct bio *bio;
 
 	bio_list_init(&bio_list);
@@ -1524,36 +1549,34 @@ static int raid56_rmw_stripe(struct btrfs_raid_bio *rbio)
 	index_rbio_pages(rbio);
 
 	atomic_set(&rbio->error, 0);
-	/*
-	 * build a list of bios to read all the missing parts of this
-	 * stripe
-	 */
-	for (stripe = 0; stripe < rbio->nr_data; stripe++) {
-		for (pagenr = 0; pagenr < rbio->stripe_npages; pagenr++) {
-			struct page *page;
-			/*
-			 * we want to find all the pages missing from
-			 * the rbio and read them from the disk.  If
-			 * page_in_rbio finds a page in the bio list
-			 * we don't need to read it off the stripe.
-			 */
-			page = page_in_rbio(rbio, stripe, pagenr, 1);
-			if (page)
-				continue;
+	/* Build a list of bios to read all the missing data sectors. */
+	for (total_sector_nr = 0; total_sector_nr < nr_data_sectors;
+	     total_sector_nr++) {
+		struct sector_ptr *sector;
+		int stripe = total_sector_nr / rbio->stripe_nsectors;
+		int sectornr = total_sector_nr % rbio->stripe_nsectors;
 
-			page = rbio_stripe_page(rbio, stripe, pagenr);
-			/*
-			 * the bio cache may have handed us an uptodate
-			 * page.  If so, be happy and use it
-			 */
-			if (PageUptodate(page))
-				continue;
+		/*
+		 * We want to find all the sectors missing from the rbio and
+		 * read them from the disk.  If sector_in_rbio() finds a page
+		 * in the bio list we don't need to read it off the stripe.
+		 */
+		sector = sector_in_rbio(rbio, stripe, sectornr, 1);
+		if (sector)
+			continue;
 
-			ret = rbio_add_io_page(rbio, &bio_list, page,
-				       stripe, pagenr, rbio->stripe_len);
-			if (ret)
-				goto cleanup;
-		}
+		sector = rbio_stripe_sector(rbio, stripe, sectornr);
+		/*
+		 * The bio cache may have handed us an uptodate page.  If so,
+		 * use it.
+		 */
+		if (sector->uptodate)
+			continue;
+
+		ret = rbio_add_io_sector(rbio, &bio_list, sector,
+			       stripe, sectornr, REQ_OP_READ);
+		if (ret)
+			goto cleanup;
 	}
 
 	bios_to_read = bio_list_size(&bio_list);
@@ -1568,21 +1591,20 @@ static int raid56_rmw_stripe(struct btrfs_raid_bio *rbio)
 	}
 
 	/*
-	 * the bbio may be freed once we submit the last bio.  Make sure
-	 * not to touch it after that
+	 * The bioc may be freed once we submit the last bio. Make sure not to
+	 * touch it after that.
 	 */
 	atomic_set(&rbio->stripes_pending, bios_to_read);
-	while (1) {
-		bio = bio_list_pop(&bio_list);
-		if (!bio)
-			break;
+	INIT_WORK(&rbio->end_io_work, raid56_rmw_end_io_work);
+	while ((bio = bio_list_pop(&bio_list))) {
+		bio->bi_end_io = raid56_bio_end_io;
 
-		bio->bi_private = rbio;
-		bio->bi_end_io = raid_rmw_end_io;
-		bio->bi_opf = REQ_OP_READ;
-
-		btrfs_bio_wq_end_io(rbio->fs_info, bio, BTRFS_WQ_ENDIO_RAID56);
+		if (trace_raid56_read_partial_enabled()) {
+			struct raid56_bio_trace_info trace_info = { 0 };
 
+			bio_get_trace_info(rbio, bio, &trace_info);
+			trace_raid56_read_partial(rbio, bio, &trace_info);
+		}
 		submit_bio(bio);
 	}
 	/* the actual write will happen once the reads are done */
@@ -1610,10 +1632,8 @@ static int full_stripe_write(struct btrfs_raid_bio *rbio)
 	int ret;
 
 	ret = alloc_rbio_parity_pages(rbio);
-	if (ret) {
-		__free_raid_bio(rbio);
+	if (ret)
 		return ret;
-	}
 
 	ret = lock_stripe_add(rbio);
 	if (ret == 0)
@@ -1661,18 +1681,19 @@ struct btrfs_plug_cb {
 	struct blk_plug_cb cb;
 	struct btrfs_fs_info *info;
 	struct list_head rbio_list;
-	struct btrfs_work work;
+	struct work_struct work;
 };
 
 /*
  * rbios on the plug list are sorted for easier merging.
  */
-static int plug_cmp(void *priv, struct list_head *a, struct list_head *b)
+static int plug_cmp(void *priv, const struct list_head *a,
+		    const struct list_head *b)
 {
-	struct btrfs_raid_bio *ra = container_of(a, struct btrfs_raid_bio,
-						 plug_list);
-	struct btrfs_raid_bio *rb = container_of(b, struct btrfs_raid_bio,
-						 plug_list);
+	const struct btrfs_raid_bio *ra = container_of(a, struct btrfs_raid_bio,
+						       plug_list);
+	const struct btrfs_raid_bio *rb = container_of(b, struct btrfs_raid_bio,
+						       plug_list);
 	u64 a_sector = ra->bio_list.head->bi_iter.bi_sector;
 	u64 b_sector = rb->bio_list.head->bi_iter.bi_sector;
 
@@ -1728,7 +1749,7 @@ static void run_plug(struct btrfs_plug_cb *plug)
  * if the unplug comes from schedule, we have to push the
  * work off to a helper thread
  */
-static void unplug_work(struct btrfs_work *work)
+static void unplug_work(struct work_struct *work)
 {
 	struct btrfs_plug_cb *plug;
 	plug = container_of(work, struct btrfs_plug_cb, work);
@@ -1741,36 +1762,58 @@ static void btrfs_raid_unplug(struct blk_plug_cb *cb, bool from_schedule)
 	plug = container_of(cb, struct btrfs_plug_cb, cb);
 
 	if (from_schedule) {
-		btrfs_init_work(&plug->work, unplug_work, NULL, NULL);
-		btrfs_queue_work(plug->info->rmw_workers,
-				 &plug->work);
+		INIT_WORK(&plug->work, unplug_work);
+		queue_work(plug->info->rmw_workers, &plug->work);
 		return;
 	}
 	run_plug(plug);
 }
 
+/* Add the original bio into rbio->bio_list, and update rbio::dbitmap. */
+static void rbio_add_bio(struct btrfs_raid_bio *rbio, struct bio *orig_bio)
+{
+	const struct btrfs_fs_info *fs_info = rbio->bioc->fs_info;
+	const u64 orig_logical = orig_bio->bi_iter.bi_sector << SECTOR_SHIFT;
+	const u64 full_stripe_start = rbio->bioc->raid_map[0];
+	const u32 orig_len = orig_bio->bi_iter.bi_size;
+	const u32 sectorsize = fs_info->sectorsize;
+	u64 cur_logical;
+
+	ASSERT(orig_logical >= full_stripe_start &&
+	       orig_logical + orig_len <= full_stripe_start +
+	       rbio->nr_data * BTRFS_STRIPE_LEN);
+
+	bio_list_add(&rbio->bio_list, orig_bio);
+	rbio->bio_list_bytes += orig_bio->bi_iter.bi_size;
+
+	/* Update the dbitmap. */
+	for (cur_logical = orig_logical; cur_logical < orig_logical + orig_len;
+	     cur_logical += sectorsize) {
+		int bit = ((u32)(cur_logical - full_stripe_start) >>
+			   fs_info->sectorsize_bits) % rbio->stripe_nsectors;
+
+		set_bit(bit, &rbio->dbitmap);
+	}
+}
+
 /*
  * our main entry point for writes from the rest of the FS.
  */
-int raid56_parity_write(struct btrfs_fs_info *fs_info, struct bio *bio,
-			struct btrfs_bio *bbio, u64 stripe_len)
+void raid56_parity_write(struct bio *bio, struct btrfs_io_context *bioc)
 {
+	struct btrfs_fs_info *fs_info = bioc->fs_info;
 	struct btrfs_raid_bio *rbio;
 	struct btrfs_plug_cb *plug = NULL;
 	struct blk_plug_cb *cb;
-	int ret;
+	int ret = 0;
 
-	rbio = alloc_rbio(fs_info, bbio, stripe_len);
+	rbio = alloc_rbio(fs_info, bioc);
 	if (IS_ERR(rbio)) {
-		btrfs_put_bbio(bbio);
-		return PTR_ERR(rbio);
+		ret = PTR_ERR(rbio);
+		goto fail;
 	}
-	bio_list_add(&rbio->bio_list, bio);
-	rbio->bio_list_bytes = bio->bi_iter.bi_size;
 	rbio->operation = BTRFS_RBIO_WRITE;
-
-	btrfs_bio_counter_inc_noblocked(fs_info);
-	rbio->generic_bio_cnt = 1;
+	rbio_add_bio(rbio, bio);
 
 	/*
 	 * don't plug on full rbios, just get them out the door
@@ -1778,9 +1821,11 @@ int raid56_parity_write(struct btrfs_fs_info *fs_info, struct bio *bio,
 	 */
 	if (rbio_is_full(rbio)) {
 		ret = full_stripe_write(rbio);
-		if (ret)
-			btrfs_bio_counter_dec(fs_info);
-		return ret;
+		if (ret) {
+			__free_raid_bio(rbio);
+			goto fail;
+		}
+		return;
 	}
 
 	cb = blk_check_plugged(btrfs_raid_unplug, fs_info, sizeof(*plug));
@@ -1791,13 +1836,19 @@ int raid56_parity_write(struct btrfs_fs_info *fs_info, struct bio *bio,
 			INIT_LIST_HEAD(&plug->rbio_list);
 		}
 		list_add_tail(&rbio->plug_list, &plug->rbio_list);
-		ret = 0;
 	} else {
 		ret = __raid56_parity_write(rbio);
-		if (ret)
-			btrfs_bio_counter_dec(fs_info);
+		if (ret) {
+			__free_raid_bio(rbio);
+			goto fail;
+		}
 	}
-	return ret;
+
+	return;
+
+fail:
+	bio->bi_status = errno_to_blk_status(ret);
+	bio_endio(bio);
 }
 
 /*
@@ -1807,19 +1858,34 @@ int raid56_parity_write(struct btrfs_fs_info *fs_info, struct bio *bio,
  */
 static void __raid_recover_end_io(struct btrfs_raid_bio *rbio)
 {
-	int pagenr, stripe;
+	const u32 sectorsize = rbio->bioc->fs_info->sectorsize;
+	int sectornr, stripe;
 	void **pointers;
+	void **unmap_array;
 	int faila = -1, failb = -1;
-	struct page *page;
 	blk_status_t err;
 	int i;
 
+	/*
+	 * This array stores the pointer for each sector, thus it has the extra
+	 * pgoff value added from each sector
+	 */
 	pointers = kcalloc(rbio->real_stripes, sizeof(void *), GFP_NOFS);
 	if (!pointers) {
 		err = BLK_STS_RESOURCE;
 		goto cleanup_io;
 	}
 
+	/*
+	 * Store copy of pointers that does not get reordered during
+	 * reconstruction so that kunmap_local works.
+	 */
+	unmap_array = kcalloc(rbio->real_stripes, sizeof(void *), GFP_NOFS);
+	if (!unmap_array) {
+		err = BLK_STS_RESOURCE;
+		goto cleanup_pointers;
+	}
+
 	faila = rbio->faila;
 	failb = rbio->failb;
 
@@ -1832,39 +1898,44 @@ static void __raid_recover_end_io(struct btrfs_raid_bio *rbio)
 
 	index_rbio_pages(rbio);
 
-	for (pagenr = 0; pagenr < rbio->stripe_npages; pagenr++) {
+	for (sectornr = 0; sectornr < rbio->stripe_nsectors; sectornr++) {
+		struct sector_ptr *sector;
+
 		/*
 		 * Now we just use bitmap to mark the horizontal stripes in
 		 * which we have data when doing parity scrub.
 		 */
 		if (rbio->operation == BTRFS_RBIO_PARITY_SCRUB &&
-		    !test_bit(pagenr, rbio->dbitmap))
+		    !test_bit(sectornr, &rbio->dbitmap))
 			continue;
 
-		/* setup our array of pointers with pages
-		 * from each stripe
+		/*
+		 * Setup our array of pointers with sectors from each stripe
+		 *
+		 * NOTE: store a duplicate array of pointers to preserve the
+		 * pointer order
 		 */
 		for (stripe = 0; stripe < rbio->real_stripes; stripe++) {
 			/*
-			 * if we're rebuilding a read, we have to use
+			 * If we're rebuilding a read, we have to use
 			 * pages from the bio list
 			 */
 			if ((rbio->operation == BTRFS_RBIO_READ_REBUILD ||
 			     rbio->operation == BTRFS_RBIO_REBUILD_MISSING) &&
 			    (stripe == faila || stripe == failb)) {
-				page = page_in_rbio(rbio, stripe, pagenr, 0);
+				sector = sector_in_rbio(rbio, stripe, sectornr, 0);
 			} else {
-				page = rbio_stripe_page(rbio, stripe, pagenr);
+				sector = rbio_stripe_sector(rbio, stripe, sectornr);
 			}
-			pointers[stripe] = kmap(page);
+			ASSERT(sector->page);
+			pointers[stripe] = kmap_local_page(sector->page) +
+					   sector->pgoff;
+			unmap_array[stripe] = pointers[stripe];
 		}
 
-		/* all raid6 handling here */
-		if (rbio->bbio->map_type & BTRFS_BLOCK_GROUP_RAID6) {
-			/*
-			 * single failure, rebuild from parity raid5
-			 * style
-			 */
+		/* All raid6 handling here */
+		if (rbio->bioc->map_type & BTRFS_BLOCK_GROUP_RAID6) {
+			/* Single failure, rebuild from parity raid5 style */
 			if (failb < 0) {
 				if (faila == rbio->nr_data) {
 					/*
@@ -1883,11 +1954,8 @@ static void __raid_recover_end_io(struct btrfs_raid_bio *rbio)
 			}
 
 			/* make sure our ps and qs are in order */
-			if (faila > failb) {
-				int tmp = failb;
-				failb = faila;
-				faila = tmp;
-			}
+			if (faila > failb)
+				swap(faila, failb);
 
 			/* if the q stripe is failed, do a pstripe reconstruction
 			 * from the xors.
@@ -1895,8 +1963,8 @@ static void __raid_recover_end_io(struct btrfs_raid_bio *rbio)
 			 * here due to a crc mismatch and we can't give them the
 			 * data they want
 			 */
-			if (rbio->bbio->raid_map[failb] == RAID6_Q_STRIPE) {
-				if (rbio->bbio->raid_map[faila] ==
+			if (rbio->bioc->raid_map[failb] == RAID6_Q_STRIPE) {
+				if (rbio->bioc->raid_map[faila] ==
 				    RAID5_P_STRIPE) {
 					err = BLK_STS_IOERR;
 					goto cleanup;
@@ -1908,12 +1976,12 @@ static void __raid_recover_end_io(struct btrfs_raid_bio *rbio)
 				goto pstripe;
 			}
 
-			if (rbio->bbio->raid_map[failb] == RAID5_P_STRIPE) {
+			if (rbio->bioc->raid_map[failb] == RAID5_P_STRIPE) {
 				raid6_datap_recov(rbio->real_stripes,
-						  PAGE_SIZE, faila, pointers);
+						  sectorsize, faila, pointers);
 			} else {
 				raid6_2data_recov(rbio->real_stripes,
-						  PAGE_SIZE, faila, failb,
+						  sectorsize, faila, failb,
 						  pointers);
 			}
 		} else {
@@ -1923,7 +1991,7 @@ static void __raid_recover_end_io(struct btrfs_raid_bio *rbio)
 			BUG_ON(failb != -1);
 pstripe:
 			/* Copy parity block into failed block to start with */
-			copy_page(pointers[faila], pointers[rbio->nr_data]);
+			memcpy(pointers[faila], pointers[rbio->nr_data], sectorsize);
 
 			/* rearrange the pointer array */
 			p = pointers[faila];
@@ -1932,7 +2000,7 @@ pstripe:
 			pointers[rbio->nr_data - 1] = p;
 
 			/* xor in the rest */
-			run_xor(pointers, rbio->nr_data - 1, PAGE_SIZE);
+			run_xor(pointers, rbio->nr_data - 1, sectorsize);
 		}
 		/* if we're doing this rebuild as part of an rmw, go through
 		 * and set all of our private rbio pages in the
@@ -1941,35 +2009,25 @@ pstripe:
 		 * other endio functions will fiddle the uptodate bits
 		 */
 		if (rbio->operation == BTRFS_RBIO_WRITE) {
-			for (i = 0;  i < rbio->stripe_npages; i++) {
+			for (i = 0;  i < rbio->stripe_nsectors; i++) {
 				if (faila != -1) {
-					page = rbio_stripe_page(rbio, faila, i);
-					SetPageUptodate(page);
+					sector = rbio_stripe_sector(rbio, faila, i);
+					sector->uptodate = 1;
 				}
 				if (failb != -1) {
-					page = rbio_stripe_page(rbio, failb, i);
-					SetPageUptodate(page);
+					sector = rbio_stripe_sector(rbio, failb, i);
+					sector->uptodate = 1;
 				}
 			}
 		}
-		for (stripe = 0; stripe < rbio->real_stripes; stripe++) {
-			/*
-			 * if we're rebuilding a read, we have to use
-			 * pages from the bio list
-			 */
-			if ((rbio->operation == BTRFS_RBIO_READ_REBUILD ||
-			     rbio->operation == BTRFS_RBIO_REBUILD_MISSING) &&
-			    (stripe == faila || stripe == failb)) {
-				page = page_in_rbio(rbio, stripe, pagenr, 0);
-			} else {
-				page = rbio_stripe_page(rbio, stripe, pagenr);
-			}
-			kunmap(page);
-		}
+		for (stripe = rbio->real_stripes - 1; stripe >= 0; stripe--)
+			kunmap_local(unmap_array[stripe]);
 	}
 
 	err = BLK_STS_OK;
 cleanup:
+	kfree(unmap_array);
+cleanup_pointers:
 	kfree(pointers);
 
 cleanup_io:
@@ -2017,27 +2075,15 @@ cleanup_io:
 }
 
 /*
- * This is called only for stripes we've read from disk to
- * reconstruct the parity.
+ * This is called only for stripes we've read from disk to reconstruct the
+ * parity.
  */
-static void raid_recover_end_io(struct bio *bio)
+static void raid_recover_end_io_work(struct work_struct *work)
 {
-	struct btrfs_raid_bio *rbio = bio->bi_private;
+	struct btrfs_raid_bio *rbio =
+		container_of(work, struct btrfs_raid_bio, end_io_work);
 
-	/*
-	 * we only read stripe pages off the disk, set them
-	 * up to date if there were no errors
-	 */
-	if (bio->bi_status)
-		fail_bio_stripe(rbio, bio);
-	else
-		set_bio_pages_uptodate(bio);
-	bio_put(bio);
-
-	if (!atomic_dec_and_test(&rbio->stripes_pending))
-		return;
-
-	if (atomic_read(&rbio->error) > rbio->bbio->max_errors)
+	if (atomic_read(&rbio->error) > rbio->bioc->max_errors)
 		rbio_orig_end_io(rbio, BLK_STS_IOERR);
 	else
 		__raid_recover_end_io(rbio);
@@ -2056,8 +2102,7 @@ static int __raid56_parity_recover(struct btrfs_raid_bio *rbio)
 	int bios_to_read = 0;
 	struct bio_list bio_list;
 	int ret;
-	int pagenr;
-	int stripe;
+	int total_sector_nr;
 	struct bio *bio;
 
 	bio_list_init(&bio_list);
@@ -2069,33 +2114,31 @@ static int __raid56_parity_recover(struct btrfs_raid_bio *rbio)
 	atomic_set(&rbio->error, 0);
 
 	/*
-	 * read everything that hasn't failed.  Thanks to the
-	 * stripe cache, it is possible that some or all of these
-	 * pages are going to be uptodate.
+	 * Read everything that hasn't failed. However this time we will
+	 * not trust any cached sector.
+	 * As we may read out some stale data but higher layer is not reading
+	 * that stale part.
+	 *
+	 * So here we always re-read everything in recovery path.
 	 */
-	for (stripe = 0; stripe < rbio->real_stripes; stripe++) {
+	for (total_sector_nr = 0; total_sector_nr < rbio->nr_sectors;
+	     total_sector_nr++) {
+		int stripe = total_sector_nr / rbio->stripe_nsectors;
+		int sectornr = total_sector_nr % rbio->stripe_nsectors;
+		struct sector_ptr *sector;
+
 		if (rbio->faila == stripe || rbio->failb == stripe) {
 			atomic_inc(&rbio->error);
+			/* Skip the current stripe. */
+			ASSERT(sectornr == 0);
+			total_sector_nr += rbio->stripe_nsectors - 1;
 			continue;
 		}
-
-		for (pagenr = 0; pagenr < rbio->stripe_npages; pagenr++) {
-			struct page *p;
-
-			/*
-			 * the rmw code may have already read this
-			 * page in
-			 */
-			p = rbio_stripe_page(rbio, stripe, pagenr);
-			if (PageUptodate(p))
-				continue;
-
-			ret = rbio_add_io_page(rbio, &bio_list,
-				       rbio_stripe_page(rbio, stripe, pagenr),
-				       stripe, pagenr, rbio->stripe_len);
-			if (ret < 0)
-				goto cleanup;
-		}
+		sector = rbio_stripe_sector(rbio, stripe, sectornr);
+		ret = rbio_add_io_sector(rbio, &bio_list, sector, stripe,
+					 sectornr, REQ_OP_READ);
+		if (ret < 0)
+			goto cleanup;
 	}
 
 	bios_to_read = bio_list_size(&bio_list);
@@ -2105,33 +2148,32 @@ static int __raid56_parity_recover(struct btrfs_raid_bio *rbio)
 		 * were up to date, or we might have no bios to read because
 		 * the devices were gone.
 		 */
-		if (atomic_read(&rbio->error) <= rbio->bbio->max_errors) {
+		if (atomic_read(&rbio->error) <= rbio->bioc->max_errors) {
 			__raid_recover_end_io(rbio);
-			goto out;
+			return 0;
 		} else {
 			goto cleanup;
 		}
 	}
 
 	/*
-	 * the bbio may be freed once we submit the last bio.  Make sure
-	 * not to touch it after that
+	 * The bioc may be freed once we submit the last bio. Make sure not to
+	 * touch it after that.
 	 */
 	atomic_set(&rbio->stripes_pending, bios_to_read);
-	while (1) {
-		bio = bio_list_pop(&bio_list);
-		if (!bio)
-			break;
-
-		bio->bi_private = rbio;
-		bio->bi_end_io = raid_recover_end_io;
-		bio->bi_opf = REQ_OP_READ;
+	INIT_WORK(&rbio->end_io_work, raid_recover_end_io_work);
+	while ((bio = bio_list_pop(&bio_list))) {
+		bio->bi_end_io = raid56_bio_end_io;
 
-		btrfs_bio_wq_end_io(rbio->fs_info, bio, BTRFS_WQ_ENDIO_RAID56);
+		if (trace_raid56_scrub_read_recover_enabled()) {
+			struct raid56_bio_trace_info trace_info = { 0 };
 
+			bio_get_trace_info(rbio, bio, &trace_info);
+			trace_raid56_scrub_read_recover(rbio, bio, &trace_info);
+		}
 		submit_bio(bio);
 	}
-out:
+
 	return 0;
 
 cleanup:
@@ -2151,46 +2193,30 @@ cleanup:
  * so we assume the bio they send down corresponds to a failed part
  * of the drive.
  */
-int raid56_parity_recover(struct btrfs_fs_info *fs_info, struct bio *bio,
-			  struct btrfs_bio *bbio, u64 stripe_len,
-			  int mirror_num, int generic_io)
+void raid56_parity_recover(struct bio *bio, struct btrfs_io_context *bioc,
+			   int mirror_num)
 {
+	struct btrfs_fs_info *fs_info = bioc->fs_info;
 	struct btrfs_raid_bio *rbio;
-	int ret;
 
-	if (generic_io) {
-		ASSERT(bbio->mirror_num == mirror_num);
-		btrfs_io_bio(bio)->mirror_num = mirror_num;
-	}
-
-	rbio = alloc_rbio(fs_info, bbio, stripe_len);
+	rbio = alloc_rbio(fs_info, bioc);
 	if (IS_ERR(rbio)) {
-		if (generic_io)
-			btrfs_put_bbio(bbio);
-		return PTR_ERR(rbio);
+		bio->bi_status = errno_to_blk_status(PTR_ERR(rbio));
+		goto out_end_bio;
 	}
 
 	rbio->operation = BTRFS_RBIO_READ_REBUILD;
-	bio_list_add(&rbio->bio_list, bio);
-	rbio->bio_list_bytes = bio->bi_iter.bi_size;
+	rbio_add_bio(rbio, bio);
 
 	rbio->faila = find_logical_bio_stripe(rbio, bio);
 	if (rbio->faila == -1) {
 		btrfs_warn(fs_info,
-	"%s could not find the bad stripe in raid56 so that we cannot recover any more (bio has logical %llu len %llu, bbio has map_type %llu)",
-			   __func__, (u64)bio->bi_iter.bi_sector << 9,
-			   (u64)bio->bi_iter.bi_size, bbio->map_type);
-		if (generic_io)
-			btrfs_put_bbio(bbio);
-		kfree(rbio);
-		return -EIO;
-	}
-
-	if (generic_io) {
-		btrfs_bio_counter_inc_noblocked(fs_info);
-		rbio->generic_bio_cnt = 1;
-	} else {
-		btrfs_get_bbio(bbio);
+"%s could not find the bad stripe in raid56 so that we cannot recover any more (bio has logical %llu len %llu, bioc has map_type %llu)",
+			   __func__, bio->bi_iter.bi_sector << 9,
+			   (u64)bio->bi_iter.bi_size, bioc->map_type);
+		__free_raid_bio(rbio);
+		bio->bi_status = BLK_STS_IOERR;
+		goto out_end_bio;
 	}
 
 	/*
@@ -2210,27 +2236,21 @@ int raid56_parity_recover(struct btrfs_fs_info *fs_info, struct bio *bio,
 			rbio->failb--;
 	}
 
-	ret = lock_stripe_add(rbio);
+	if (lock_stripe_add(rbio))
+		return;
 
 	/*
-	 * __raid56_parity_recover will end the bio with
-	 * any errors it hits.  We don't want to return
-	 * its error value up the stack because our caller
-	 * will end up calling bio_endio with any nonzero
-	 * return
+	 * This adds our rbio to the list of rbios that will be handled after
+	 * the current lock owner is done.
 	 */
-	if (ret == 0)
-		__raid56_parity_recover(rbio);
-	/*
-	 * our rbio has been added to the list of
-	 * rbios that will be handled after the
-	 * currently lock owner is done
-	 */
-	return 0;
+	__raid56_parity_recover(rbio);
+	return;
 
+out_end_bio:
+	bio_endio(bio);
 }
 
-static void rmw_work(struct btrfs_work *work)
+static void rmw_work(struct work_struct *work)
 {
 	struct btrfs_raid_bio *rbio;
 
@@ -2238,7 +2258,7 @@ static void rmw_work(struct btrfs_work *work)
 	raid56_rmw_stripe(rbio);
 }
 
-static void read_rebuild_work(struct btrfs_work *work)
+static void read_rebuild_work(struct work_struct *work)
 {
 	struct btrfs_raid_bio *rbio;
 
@@ -2249,23 +2269,23 @@ static void read_rebuild_work(struct btrfs_work *work)
 /*
  * The following code is used to scrub/replace the parity stripe
  *
- * Caller must have already increased bio_counter for getting @bbio.
+ * Caller must have already increased bio_counter for getting @bioc.
  *
  * Note: We need make sure all the pages that add into the scrub/replace
  * raid bio are correct and not be changed during the scrub/replace. That
  * is those pages just hold metadata or file data with checksum.
  */
 
-struct btrfs_raid_bio *
-raid56_parity_alloc_scrub_rbio(struct btrfs_fs_info *fs_info, struct bio *bio,
-			       struct btrfs_bio *bbio, u64 stripe_len,
-			       struct btrfs_device *scrub_dev,
-			       unsigned long *dbitmap, int stripe_nsectors)
+struct btrfs_raid_bio *raid56_parity_alloc_scrub_rbio(struct bio *bio,
+				struct btrfs_io_context *bioc,
+				struct btrfs_device *scrub_dev,
+				unsigned long *dbitmap, int stripe_nsectors)
 {
+	struct btrfs_fs_info *fs_info = bioc->fs_info;
 	struct btrfs_raid_bio *rbio;
 	int i;
 
-	rbio = alloc_rbio(fs_info, bbio, stripe_len);
+	rbio = alloc_rbio(fs_info, bioc);
 	if (IS_ERR(rbio))
 		return NULL;
 	bio_list_add(&rbio->bio_list, bio);
@@ -2277,45 +2297,37 @@ raid56_parity_alloc_scrub_rbio(struct btrfs_fs_info *fs_info, struct bio *bio,
 	rbio->operation = BTRFS_RBIO_PARITY_SCRUB;
 
 	/*
-	 * After mapping bbio with BTRFS_MAP_WRITE, parities have been sorted
+	 * After mapping bioc with BTRFS_MAP_WRITE, parities have been sorted
 	 * to the end position, so this search can start from the first parity
 	 * stripe.
 	 */
 	for (i = rbio->nr_data; i < rbio->real_stripes; i++) {
-		if (bbio->stripes[i].dev == scrub_dev) {
+		if (bioc->stripes[i].dev == scrub_dev) {
 			rbio->scrubp = i;
 			break;
 		}
 	}
 	ASSERT(i < rbio->real_stripes);
 
-	/* Now we just support the sectorsize equals to page size */
-	ASSERT(fs_info->sectorsize == PAGE_SIZE);
-	ASSERT(rbio->stripe_npages == stripe_nsectors);
-	bitmap_copy(rbio->dbitmap, dbitmap, stripe_nsectors);
-
-	/*
-	 * We have already increased bio_counter when getting bbio, record it
-	 * so we can free it at rbio_orig_end_io().
-	 */
-	rbio->generic_bio_cnt = 1;
-
+	bitmap_copy(&rbio->dbitmap, dbitmap, stripe_nsectors);
 	return rbio;
 }
 
 /* Used for both parity scrub and missing. */
 void raid56_add_scrub_pages(struct btrfs_raid_bio *rbio, struct page *page,
-			    u64 logical)
+			    unsigned int pgoff, u64 logical)
 {
+	const u32 sectorsize = rbio->bioc->fs_info->sectorsize;
 	int stripe_offset;
 	int index;
 
-	ASSERT(logical >= rbio->bbio->raid_map[0]);
-	ASSERT(logical + PAGE_SIZE <= rbio->bbio->raid_map[0] +
-				rbio->stripe_len * rbio->nr_data);
-	stripe_offset = (int)(logical - rbio->bbio->raid_map[0]);
-	index = stripe_offset >> PAGE_SHIFT;
-	rbio->bio_pages[index] = page;
+	ASSERT(logical >= rbio->bioc->raid_map[0]);
+	ASSERT(logical + sectorsize <= rbio->bioc->raid_map[0] +
+				       BTRFS_STRIPE_LEN * rbio->nr_data);
+	stripe_offset = (int)(logical - rbio->bioc->raid_map[0]);
+	index = stripe_offset / sectorsize;
+	rbio->bio_sectors[index].page = page;
+	rbio->bio_sectors[index].pgoff = pgoff;
 }
 
 /*
@@ -2324,39 +2336,41 @@ void raid56_add_scrub_pages(struct btrfs_raid_bio *rbio, struct page *page,
  */
 static int alloc_rbio_essential_pages(struct btrfs_raid_bio *rbio)
 {
-	int i;
-	int bit;
-	int index;
-	struct page *page;
+	const u32 sectorsize = rbio->bioc->fs_info->sectorsize;
+	int total_sector_nr;
 
-	for_each_set_bit(bit, rbio->dbitmap, rbio->stripe_npages) {
-		for (i = 0; i < rbio->real_stripes; i++) {
-			index = i * rbio->stripe_npages + bit;
-			if (rbio->stripe_pages[index])
-				continue;
+	for (total_sector_nr = 0; total_sector_nr < rbio->nr_sectors;
+	     total_sector_nr++) {
+		struct page *page;
+		int sectornr = total_sector_nr % rbio->stripe_nsectors;
+		int index = (total_sector_nr * sectorsize) >> PAGE_SHIFT;
 
-			page = alloc_page(GFP_NOFS | __GFP_HIGHMEM);
-			if (!page)
-				return -ENOMEM;
-			rbio->stripe_pages[index] = page;
-		}
+		if (!test_bit(sectornr, &rbio->dbitmap))
+			continue;
+		if (rbio->stripe_pages[index])
+			continue;
+		page = alloc_page(GFP_NOFS);
+		if (!page)
+			return -ENOMEM;
+		rbio->stripe_pages[index] = page;
 	}
+	index_stripe_sectors(rbio);
 	return 0;
 }
 
 static noinline void finish_parity_scrub(struct btrfs_raid_bio *rbio,
 					 int need_check)
 {
-	struct btrfs_bio *bbio = rbio->bbio;
+	struct btrfs_io_context *bioc = rbio->bioc;
+	const u32 sectorsize = bioc->fs_info->sectorsize;
 	void **pointers = rbio->finish_pointers;
-	unsigned long *pbitmap = rbio->finish_pbitmap;
+	unsigned long *pbitmap = &rbio->finish_pbitmap;
 	int nr_data = rbio->nr_data;
 	int stripe;
-	int pagenr;
-	int p_stripe = -1;
-	int q_stripe = -1;
-	struct page *p_page = NULL;
-	struct page *q_page = NULL;
+	int sectornr;
+	bool has_qstripe;
+	struct sector_ptr p_sector = { 0 };
+	struct sector_ptr q_sector = { 0 };
 	struct bio_list bio_list;
 	struct bio *bio;
 	int is_replace = 0;
@@ -2364,18 +2378,16 @@ static noinline void finish_parity_scrub(struct btrfs_raid_bio *rbio,
 
 	bio_list_init(&bio_list);
 
-	if (rbio->real_stripes - rbio->nr_data == 1) {
-		p_stripe = rbio->real_stripes - 1;
-	} else if (rbio->real_stripes - rbio->nr_data == 2) {
-		p_stripe = rbio->real_stripes - 2;
-		q_stripe = rbio->real_stripes - 1;
-	} else {
+	if (rbio->real_stripes - rbio->nr_data == 1)
+		has_qstripe = false;
+	else if (rbio->real_stripes - rbio->nr_data == 2)
+		has_qstripe = true;
+	else
 		BUG();
-	}
 
-	if (bbio->num_tgtdevs && bbio->tgtdev_map[rbio->scrubp]) {
+	if (bioc->num_tgtdevs && bioc->tgtdev_map[rbio->scrubp]) {
 		is_replace = 1;
-		bitmap_copy(pbitmap, rbio->dbitmap, rbio->stripe_npages);
+		bitmap_copy(pbitmap, &rbio->dbitmap, rbio->stripe_nsectors);
 	}
 
 	/*
@@ -2388,68 +2400,73 @@ static noinline void finish_parity_scrub(struct btrfs_raid_bio *rbio,
 	if (!need_check)
 		goto writeback;
 
-	p_page = alloc_page(GFP_NOFS | __GFP_HIGHMEM);
-	if (!p_page)
+	p_sector.page = alloc_page(GFP_NOFS);
+	if (!p_sector.page)
 		goto cleanup;
-	SetPageUptodate(p_page);
-
-	if (q_stripe != -1) {
-		q_page = alloc_page(GFP_NOFS | __GFP_HIGHMEM);
-		if (!q_page) {
-			__free_page(p_page);
+	p_sector.pgoff = 0;
+	p_sector.uptodate = 1;
+
+	if (has_qstripe) {
+		/* RAID6, allocate and map temp space for the Q stripe */
+		q_sector.page = alloc_page(GFP_NOFS);
+		if (!q_sector.page) {
+			__free_page(p_sector.page);
+			p_sector.page = NULL;
 			goto cleanup;
 		}
-		SetPageUptodate(q_page);
+		q_sector.pgoff = 0;
+		q_sector.uptodate = 1;
+		pointers[rbio->real_stripes - 1] = kmap_local_page(q_sector.page);
 	}
 
 	atomic_set(&rbio->error, 0);
 
-	for_each_set_bit(pagenr, rbio->dbitmap, rbio->stripe_npages) {
-		struct page *p;
+	/* Map the parity stripe just once */
+	pointers[nr_data] = kmap_local_page(p_sector.page);
+
+	for_each_set_bit(sectornr, &rbio->dbitmap, rbio->stripe_nsectors) {
+		struct sector_ptr *sector;
 		void *parity;
+
 		/* first collect one page from each data stripe */
 		for (stripe = 0; stripe < nr_data; stripe++) {
-			p = page_in_rbio(rbio, stripe, pagenr, 0);
-			pointers[stripe] = kmap(p);
+			sector = sector_in_rbio(rbio, stripe, sectornr, 0);
+			pointers[stripe] = kmap_local_page(sector->page) +
+					   sector->pgoff;
 		}
 
-		/* then add the parity stripe */
-		pointers[stripe++] = kmap(p_page);
-
-		if (q_stripe != -1) {
-
-			/*
-			 * raid6, add the qstripe and call the
-			 * library function to fill in our p/q
-			 */
-			pointers[stripe++] = kmap(q_page);
-
-			raid6_call.gen_syndrome(rbio->real_stripes, PAGE_SIZE,
+		if (has_qstripe) {
+			/* RAID6, call the library function to fill in our P/Q */
+			raid6_call.gen_syndrome(rbio->real_stripes, sectorsize,
 						pointers);
 		} else {
 			/* raid5 */
-			copy_page(pointers[nr_data], pointers[0]);
-			run_xor(pointers + 1, nr_data - 1, PAGE_SIZE);
+			memcpy(pointers[nr_data], pointers[0], sectorsize);
+			run_xor(pointers + 1, nr_data - 1, sectorsize);
 		}
 
 		/* Check scrubbing parity and repair it */
-		p = rbio_stripe_page(rbio, rbio->scrubp, pagenr);
-		parity = kmap(p);
-		if (memcmp(parity, pointers[rbio->scrubp], PAGE_SIZE))
-			copy_page(parity, pointers[rbio->scrubp]);
+		sector = rbio_stripe_sector(rbio, rbio->scrubp, sectornr);
+		parity = kmap_local_page(sector->page) + sector->pgoff;
+		if (memcmp(parity, pointers[rbio->scrubp], sectorsize) != 0)
+			memcpy(parity, pointers[rbio->scrubp], sectorsize);
 		else
 			/* Parity is right, needn't writeback */
-			bitmap_clear(rbio->dbitmap, pagenr, 1);
-		kunmap(p);
+			bitmap_clear(&rbio->dbitmap, sectornr, 1);
+		kunmap_local(parity);
 
-		for (stripe = 0; stripe < nr_data; stripe++)
-			kunmap(page_in_rbio(rbio, stripe, pagenr, 0));
-		kunmap(p_page);
+		for (stripe = nr_data - 1; stripe >= 0; stripe--)
+			kunmap_local(pointers[stripe]);
 	}
 
-	__free_page(p_page);
-	if (q_page)
-		__free_page(q_page);
+	kunmap_local(pointers[nr_data]);
+	__free_page(p_sector.page);
+	p_sector.page = NULL;
+	if (q_sector.page) {
+		kunmap_local(pointers[rbio->real_stripes - 1]);
+		__free_page(q_sector.page);
+		q_sector.page = NULL;
+	}
 
 writeback:
 	/*
@@ -2457,12 +2474,12 @@ writeback:
 	 * higher layers (the bio_list in our rbio) and our p/q.  Ignore
 	 * everything else.
 	 */
-	for_each_set_bit(pagenr, rbio->dbitmap, rbio->stripe_npages) {
-		struct page *page;
+	for_each_set_bit(sectornr, &rbio->dbitmap, rbio->stripe_nsectors) {
+		struct sector_ptr *sector;
 
-		page = rbio_stripe_page(rbio, rbio->scrubp, pagenr);
-		ret = rbio_add_io_page(rbio, &bio_list,
-			       page, rbio->scrubp, pagenr, rbio->stripe_len);
+		sector = rbio_stripe_sector(rbio, rbio->scrubp, sectornr);
+		ret = rbio_add_io_sector(rbio, &bio_list, sector, rbio->scrubp,
+					 sectornr, REQ_OP_WRITE);
 		if (ret)
 			goto cleanup;
 	}
@@ -2470,13 +2487,13 @@ writeback:
 	if (!is_replace)
 		goto submit_write;
 
-	for_each_set_bit(pagenr, pbitmap, rbio->stripe_npages) {
-		struct page *page;
+	for_each_set_bit(sectornr, pbitmap, rbio->stripe_nsectors) {
+		struct sector_ptr *sector;
 
-		page = rbio_stripe_page(rbio, rbio->scrubp, pagenr);
-		ret = rbio_add_io_page(rbio, &bio_list, page,
-				       bbio->tgtdev_map[rbio->scrubp],
-				       pagenr, rbio->stripe_len);
+		sector = rbio_stripe_sector(rbio, rbio->scrubp, sectornr);
+		ret = rbio_add_io_sector(rbio, &bio_list, sector,
+				       bioc->tgtdev_map[rbio->scrubp],
+				       sectornr, REQ_OP_WRITE);
 		if (ret)
 			goto cleanup;
 	}
@@ -2491,15 +2508,15 @@ submit_write:
 
 	atomic_set(&rbio->stripes_pending, nr_data);
 
-	while (1) {
-		bio = bio_list_pop(&bio_list);
-		if (!bio)
-			break;
-
-		bio->bi_private = rbio;
+	while ((bio = bio_list_pop(&bio_list))) {
 		bio->bi_end_io = raid_write_end_io;
-		bio->bi_opf = REQ_OP_WRITE;
 
+		if (trace_raid56_scrub_write_stripe_enabled()) {
+			struct raid56_bio_trace_info trace_info = { 0 };
+
+			bio_get_trace_info(rbio, bio, &trace_info);
+			trace_raid56_scrub_write_stripe(rbio, bio, &trace_info);
+		}
 		submit_bio(bio);
 	}
 	return;
@@ -2527,7 +2544,7 @@ static inline int is_data_stripe(struct btrfs_raid_bio *rbio, int stripe)
  */
 static void validate_rbio_for_parity_scrub(struct btrfs_raid_bio *rbio)
 {
-	if (atomic_read(&rbio->error) > rbio->bbio->max_errors)
+	if (atomic_read(&rbio->error) > rbio->bioc->max_errors)
 		goto cleanup;
 
 	if (rbio->faila >= 0 || rbio->failb >= 0) {
@@ -2548,7 +2565,7 @@ static void validate_rbio_for_parity_scrub(struct btrfs_raid_bio *rbio)
 		 * the data, so the capability of the repair is declined.
 		 * (In the case of RAID5, we can not repair anything)
 		 */
-		if (dfail > rbio->bbio->max_errors - 1)
+		if (dfail > rbio->bioc->max_errors - 1)
 			goto cleanup;
 
 		/*
@@ -2587,24 +2604,14 @@ cleanup:
  * This will usually kick off finish_rmw once all the bios are read in, but it
  * may trigger parity reconstruction if we had any errors along the way
  */
-static void raid56_parity_scrub_end_io(struct bio *bio)
+static void raid56_parity_scrub_end_io_work(struct work_struct *work)
 {
-	struct btrfs_raid_bio *rbio = bio->bi_private;
-
-	if (bio->bi_status)
-		fail_bio_stripe(rbio, bio);
-	else
-		set_bio_pages_uptodate(bio);
-
-	bio_put(bio);
-
-	if (!atomic_dec_and_test(&rbio->stripes_pending))
-		return;
+	struct btrfs_raid_bio *rbio =
+		container_of(work, struct btrfs_raid_bio, end_io_work);
 
 	/*
-	 * this will normally call finish_rmw to start our write
-	 * but if there are any failed stripes we'll reconstruct
-	 * from parity first
+	 * This will normally call finish_rmw to start our write, but if there
+	 * are any failed stripes we'll reconstruct from parity first
 	 */
 	validate_rbio_for_parity_scrub(rbio);
 }
@@ -2614,8 +2621,7 @@ static void raid56_parity_scrub_stripe(struct btrfs_raid_bio *rbio)
 	int bios_to_read = 0;
 	struct bio_list bio_list;
 	int ret;
-	int pagenr;
-	int stripe;
+	int total_sector_nr;
 	struct bio *bio;
 
 	bio_list_init(&bio_list);
@@ -2625,36 +2631,38 @@ static void raid56_parity_scrub_stripe(struct btrfs_raid_bio *rbio)
 		goto cleanup;
 
 	atomic_set(&rbio->error, 0);
-	/*
-	 * build a list of bios to read all the missing parts of this
-	 * stripe
-	 */
-	for (stripe = 0; stripe < rbio->real_stripes; stripe++) {
-		for_each_set_bit(pagenr, rbio->dbitmap, rbio->stripe_npages) {
-			struct page *page;
-			/*
-			 * we want to find all the pages missing from
-			 * the rbio and read them from the disk.  If
-			 * page_in_rbio finds a page in the bio list
-			 * we don't need to read it off the stripe.
-			 */
-			page = page_in_rbio(rbio, stripe, pagenr, 1);
-			if (page)
-				continue;
+	/* Build a list of bios to read all the missing parts. */
+	for (total_sector_nr = 0; total_sector_nr < rbio->nr_sectors;
+	     total_sector_nr++) {
+		int sectornr = total_sector_nr % rbio->stripe_nsectors;
+		int stripe = total_sector_nr / rbio->stripe_nsectors;
+		struct sector_ptr *sector;
+
+		/* No data in the vertical stripe, no need to read. */
+		if (!test_bit(sectornr, &rbio->dbitmap))
+			continue;
 
-			page = rbio_stripe_page(rbio, stripe, pagenr);
-			/*
-			 * the bio cache may have handed us an uptodate
-			 * page.  If so, be happy and use it
-			 */
-			if (PageUptodate(page))
-				continue;
+		/*
+		 * We want to find all the sectors missing from the rbio and
+		 * read them from the disk. If sector_in_rbio() finds a sector
+		 * in the bio list we don't need to read it off the stripe.
+		 */
+		sector = sector_in_rbio(rbio, stripe, sectornr, 1);
+		if (sector)
+			continue;
 
-			ret = rbio_add_io_page(rbio, &bio_list, page,
-				       stripe, pagenr, rbio->stripe_len);
-			if (ret)
-				goto cleanup;
-		}
+		sector = rbio_stripe_sector(rbio, stripe, sectornr);
+		/*
+		 * The bio cache may have handed us an uptodate sector.  If so,
+		 * use it.
+		 */
+		if (sector->uptodate)
+			continue;
+
+		ret = rbio_add_io_sector(rbio, &bio_list, sector, stripe,
+					 sectornr, REQ_OP_READ);
+		if (ret)
+			goto cleanup;
 	}
 
 	bios_to_read = bio_list_size(&bio_list);
@@ -2669,21 +2677,20 @@ static void raid56_parity_scrub_stripe(struct btrfs_raid_bio *rbio)
 	}
 
 	/*
-	 * the bbio may be freed once we submit the last bio.  Make sure
-	 * not to touch it after that
+	 * The bioc may be freed once we submit the last bio. Make sure not to
+	 * touch it after that.
 	 */
 	atomic_set(&rbio->stripes_pending, bios_to_read);
-	while (1) {
-		bio = bio_list_pop(&bio_list);
-		if (!bio)
-			break;
-
-		bio->bi_private = rbio;
-		bio->bi_end_io = raid56_parity_scrub_end_io;
-		bio->bi_opf = REQ_OP_READ;
+	INIT_WORK(&rbio->end_io_work, raid56_parity_scrub_end_io_work);
+	while ((bio = bio_list_pop(&bio_list))) {
+		bio->bi_end_io = raid56_bio_end_io;
 
-		btrfs_bio_wq_end_io(rbio->fs_info, bio, BTRFS_WQ_ENDIO_RAID56);
+		if (trace_raid56_scrub_read_enabled()) {
+			struct raid56_bio_trace_info trace_info = { 0 };
 
+			bio_get_trace_info(rbio, bio, &trace_info);
+			trace_raid56_scrub_read(rbio, bio, &trace_info);
+		}
 		submit_bio(bio);
 	}
 	/* the actual write will happen once the reads are done */
@@ -2701,7 +2708,7 @@ finish:
 	validate_rbio_for_parity_scrub(rbio);
 }
 
-static void scrub_parity_work(struct btrfs_work *work)
+static void scrub_parity_work(struct work_struct *work)
 {
 	struct btrfs_raid_bio *rbio;
 
@@ -2718,12 +2725,12 @@ void raid56_parity_submit_scrub_rbio(struct btrfs_raid_bio *rbio)
 /* The following code is used for dev replace of a missing RAID 5/6 device. */
 
 struct btrfs_raid_bio *
-raid56_alloc_missing_rbio(struct btrfs_fs_info *fs_info, struct bio *bio,
-			  struct btrfs_bio *bbio, u64 length)
+raid56_alloc_missing_rbio(struct bio *bio, struct btrfs_io_context *bioc)
 {
+	struct btrfs_fs_info *fs_info = bioc->fs_info;
 	struct btrfs_raid_bio *rbio;
 
-	rbio = alloc_rbio(fs_info, bbio, length);
+	rbio = alloc_rbio(fs_info, bioc);
 	if (IS_ERR(rbio))
 		return NULL;
 
@@ -2737,17 +2744,13 @@ raid56_alloc_missing_rbio(struct btrfs_fs_info *fs_info, struct bio *bio,
 
 	rbio->faila = find_logical_bio_stripe(rbio, bio);
 	if (rbio->faila == -1) {
-		BUG();
-		kfree(rbio);
+		btrfs_warn_rl(fs_info,
+	"can not determine the failed stripe number for full stripe %llu",
+			      bioc->raid_map[0]);
+		__free_raid_bio(rbio);
 		return NULL;
 	}
 
-	/*
-	 * When we get bbio, we have already increased bio_counter, record it
-	 * so we can free it at rbio_orig_end_io()
-	 */
-	rbio->generic_bio_cnt = 1;
-
 	return rbio;
 }
 
diff --git a/fs/btrfs/raid56.h b/fs/btrfs/raid56.h
index 2503485db859..91d5c0adad15 100644
--- a/fs/btrfs/raid56.h
+++ b/fs/btrfs/raid56.h
@@ -7,48 +7,177 @@
 #ifndef BTRFS_RAID56_H
 #define BTRFS_RAID56_H
 
-static inline int nr_parity_stripes(const struct map_lookup *map)
-{
-	if (map->type & BTRFS_BLOCK_GROUP_RAID5)
-		return 1;
-	else if (map->type & BTRFS_BLOCK_GROUP_RAID6)
-		return 2;
-	else
-		return 0;
-}
+#include <linux/workqueue.h>
+#include "volumes.h"
+
+enum btrfs_rbio_ops {
+	BTRFS_RBIO_WRITE,
+	BTRFS_RBIO_READ_REBUILD,
+	BTRFS_RBIO_PARITY_SCRUB,
+	BTRFS_RBIO_REBUILD_MISSING,
+};
+
+struct btrfs_raid_bio {
+	struct btrfs_io_context *bioc;
+
+	/*
+	 * While we're doing RMW on a stripe we put it into a hash table so we
+	 * can lock the stripe and merge more rbios into it.
+	 */
+	struct list_head hash_list;
+
+	/* LRU list for the stripe cache */
+	struct list_head stripe_cache;
+
+	/* For scheduling work in the helper threads */
+	struct work_struct work;
+
+	/*
+	 * bio_list and bio_list_lock are used to add more bios into the stripe
+	 * in hopes of avoiding the full RMW
+	 */
+	struct bio_list bio_list;
+	spinlock_t bio_list_lock;
+
+	/*
+	 * Also protected by the bio_list_lock, the plug list is used by the
+	 * plugging code to collect partial bios while plugged.  The stripe
+	 * locking code also uses it to hand off the stripe lock to the next
+	 * pending IO.
+	 */
+	struct list_head plug_list;
+
+	/* Flags that tell us if it is safe to merge with this bio. */
+	unsigned long flags;
+
+	/*
+	 * Set if we're doing a parity rebuild for a read from higher up, which
+	 * is handled differently from a parity rebuild as part of RMW.
+	 */
+	enum btrfs_rbio_ops operation;
+
+	/* How many pages there are for the full stripe including P/Q */
+	u16 nr_pages;
+
+	/* How many sectors there are for the full stripe including P/Q */
+	u16 nr_sectors;
+
+	/* Number of data stripes (no p/q) */
+	u8 nr_data;
+
+	/* Numer of all stripes (including P/Q) */
+	u8 real_stripes;
+
+	/* How many pages there are for each stripe */
+	u8 stripe_npages;
+
+	/* How many sectors there are for each stripe */
+	u8 stripe_nsectors;
+
+	/* First bad stripe, -1 means no corruption */
+	s8 faila;
+
+	/* Second bad stripe (for RAID6 use) */
+	s8 failb;
+
+	/* Stripe number that we're scrubbing  */
+	u8 scrubp;
+
+	/*
+	 * Size of all the bios in the bio_list.  This helps us decide if the
+	 * rbio maps to a full stripe or not.
+	 */
+	int bio_list_bytes;
+
+	refcount_t refs;
+
+	atomic_t stripes_pending;
+
+	atomic_t error;
+
+	struct work_struct end_io_work;
+
+	/* Bitmap to record which horizontal stripe has data */
+	unsigned long dbitmap;
+
+	/* Allocated with stripe_nsectors-many bits for finish_*() calls */
+	unsigned long finish_pbitmap;
+
+	/*
+	 * These are two arrays of pointers.  We allocate the rbio big enough
+	 * to hold them both and setup their locations when the rbio is
+	 * allocated.
+	 */
+
+	/*
+	 * Pointers to pages that we allocated for reading/writing stripes
+	 * directly from the disk (including P/Q).
+	 */
+	struct page **stripe_pages;
+
+	/* Pointers to the sectors in the bio_list, for faster lookup */
+	struct sector_ptr *bio_sectors;
+
+	/*
+	 * For subpage support, we need to map each sector to above
+	 * stripe_pages.
+	 */
+	struct sector_ptr *stripe_sectors;
+
+	/* Allocated with real_stripes-many pointers for finish_*() calls */
+	void **finish_pointers;
+};
+
+/*
+ * For trace event usage only. Records useful debug info for each bio submitted
+ * by RAID56 to each physical device.
+ *
+ * No matter signed or not, (-1) is always the one indicating we can not grab
+ * the proper stripe number.
+ */
+struct raid56_bio_trace_info {
+	u64 devid;
+
+	/* The offset inside the stripe. (<= STRIPE_LEN) */
+	u32 offset;
+
+	/*
+	 * Stripe number.
+	 * 0 is the first data stripe, and nr_data for P stripe,
+	 * nr_data + 1 for Q stripe.
+	 * >= real_stripes for
+	 */
+	u8 stripe_nr;
+};
 
 static inline int nr_data_stripes(const struct map_lookup *map)
 {
-	return map->num_stripes - nr_parity_stripes(map);
+	return map->num_stripes - btrfs_nr_parity_stripes(map->type);
 }
+
 #define RAID5_P_STRIPE ((u64)-2)
 #define RAID6_Q_STRIPE ((u64)-1)
 
 #define is_parity_stripe(x) (((x) == RAID5_P_STRIPE) ||		\
 			     ((x) == RAID6_Q_STRIPE))
 
-struct btrfs_raid_bio;
 struct btrfs_device;
 
-int raid56_parity_recover(struct btrfs_fs_info *fs_info, struct bio *bio,
-			  struct btrfs_bio *bbio, u64 stripe_len,
-			  int mirror_num, int generic_io);
-int raid56_parity_write(struct btrfs_fs_info *fs_info, struct bio *bio,
-			       struct btrfs_bio *bbio, u64 stripe_len);
+void raid56_parity_recover(struct bio *bio, struct btrfs_io_context *bioc,
+			   int mirror_num);
+void raid56_parity_write(struct bio *bio, struct btrfs_io_context *bioc);
 
 void raid56_add_scrub_pages(struct btrfs_raid_bio *rbio, struct page *page,
-			    u64 logical);
+			    unsigned int pgoff, u64 logical);
 
-struct btrfs_raid_bio *
-raid56_parity_alloc_scrub_rbio(struct btrfs_fs_info *fs_info, struct bio *bio,
-			       struct btrfs_bio *bbio, u64 stripe_len,
-			       struct btrfs_device *scrub_dev,
-			       unsigned long *dbitmap, int stripe_nsectors);
+struct btrfs_raid_bio *raid56_parity_alloc_scrub_rbio(struct bio *bio,
+				struct btrfs_io_context *bioc,
+				struct btrfs_device *scrub_dev,
+				unsigned long *dbitmap, int stripe_nsectors);
 void raid56_parity_submit_scrub_rbio(struct btrfs_raid_bio *rbio);
 
 struct btrfs_raid_bio *
-raid56_alloc_missing_rbio(struct btrfs_fs_info *fs_info, struct bio *bio,
-			  struct btrfs_bio *bbio, u64 length);
+raid56_alloc_missing_rbio(struct bio *bio, struct btrfs_io_context *bioc);
 void raid56_submit_missing_rbio(struct btrfs_raid_bio *rbio);
 
 int btrfs_alloc_stripe_hash_table(struct btrfs_fs_info *info);
diff --git a/fs/btrfs/rcu-string.h b/fs/btrfs/rcu-string.h
index a97dc74a4d3d..5c1a617eb25d 100644
--- a/fs/btrfs/rcu-string.h
+++ b/fs/btrfs/rcu-string.h
@@ -8,7 +8,7 @@
 
 struct rcu_string {
 	struct rcu_head rcu;
-	char str[0];
+	char str[];
 };
 
 static inline struct rcu_string *rcu_string_strdup(const char *src, gfp_t mask)
diff --git a/fs/btrfs/reada.c b/fs/btrfs/reada.c
deleted file mode 100644
index 243a2e44526e..000000000000
--- a/fs/btrfs/reada.c
+++ /dev/null
@@ -1,1014 +0,0 @@
-// SPDX-License-Identifier: GPL-2.0
-/*
- * Copyright (C) 2011 STRATO.  All rights reserved.
- */
-
-#include <linux/sched.h>
-#include <linux/pagemap.h>
-#include <linux/writeback.h>
-#include <linux/blkdev.h>
-#include <linux/slab.h>
-#include <linux/workqueue.h>
-#include "ctree.h"
-#include "volumes.h"
-#include "disk-io.h"
-#include "transaction.h"
-#include "dev-replace.h"
-#include "block-group.h"
-
-#undef DEBUG
-
-/*
- * This is the implementation for the generic read ahead framework.
- *
- * To trigger a readahead, btrfs_reada_add must be called. It will start
- * a read ahead for the given range [start, end) on tree root. The returned
- * handle can either be used to wait on the readahead to finish
- * (btrfs_reada_wait), or to send it to the background (btrfs_reada_detach).
- *
- * The read ahead works as follows:
- * On btrfs_reada_add, the root of the tree is inserted into a radix_tree.
- * reada_start_machine will then search for extents to prefetch and trigger
- * some reads. When a read finishes for a node, all contained node/leaf
- * pointers that lie in the given range will also be enqueued. The reads will
- * be triggered in sequential order, thus giving a big win over a naive
- * enumeration. It will also make use of multi-device layouts. Each disk
- * will have its on read pointer and all disks will by utilized in parallel.
- * Also will no two disks read both sides of a mirror simultaneously, as this
- * would waste seeking capacity. Instead both disks will read different parts
- * of the filesystem.
- * Any number of readaheads can be started in parallel. The read order will be
- * determined globally, i.e. 2 parallel readaheads will normally finish faster
- * than the 2 started one after another.
- */
-
-#define MAX_IN_FLIGHT 6
-
-struct reada_extctl {
-	struct list_head	list;
-	struct reada_control	*rc;
-	u64			generation;
-};
-
-struct reada_extent {
-	u64			logical;
-	struct btrfs_key	top;
-	struct list_head	extctl;
-	int 			refcnt;
-	spinlock_t		lock;
-	struct reada_zone	*zones[BTRFS_MAX_MIRRORS];
-	int			nzones;
-	int			scheduled;
-};
-
-struct reada_zone {
-	u64			start;
-	u64			end;
-	u64			elems;
-	struct list_head	list;
-	spinlock_t		lock;
-	int			locked;
-	struct btrfs_device	*device;
-	struct btrfs_device	*devs[BTRFS_MAX_MIRRORS]; /* full list, incl
-							   * self */
-	int			ndevs;
-	struct kref		refcnt;
-};
-
-struct reada_machine_work {
-	struct btrfs_work	work;
-	struct btrfs_fs_info	*fs_info;
-};
-
-static void reada_extent_put(struct btrfs_fs_info *, struct reada_extent *);
-static void reada_control_release(struct kref *kref);
-static void reada_zone_release(struct kref *kref);
-static void reada_start_machine(struct btrfs_fs_info *fs_info);
-static void __reada_start_machine(struct btrfs_fs_info *fs_info);
-
-static int reada_add_block(struct reada_control *rc, u64 logical,
-			   struct btrfs_key *top, u64 generation);
-
-/* recurses */
-/* in case of err, eb might be NULL */
-static void __readahead_hook(struct btrfs_fs_info *fs_info,
-			     struct reada_extent *re, struct extent_buffer *eb,
-			     int err)
-{
-	int nritems;
-	int i;
-	u64 bytenr;
-	u64 generation;
-	struct list_head list;
-
-	spin_lock(&re->lock);
-	/*
-	 * just take the full list from the extent. afterwards we
-	 * don't need the lock anymore
-	 */
-	list_replace_init(&re->extctl, &list);
-	re->scheduled = 0;
-	spin_unlock(&re->lock);
-
-	/*
-	 * this is the error case, the extent buffer has not been
-	 * read correctly. We won't access anything from it and
-	 * just cleanup our data structures. Effectively this will
-	 * cut the branch below this node from read ahead.
-	 */
-	if (err)
-		goto cleanup;
-
-	/*
-	 * FIXME: currently we just set nritems to 0 if this is a leaf,
-	 * effectively ignoring the content. In a next step we could
-	 * trigger more readahead depending from the content, e.g.
-	 * fetch the checksums for the extents in the leaf.
-	 */
-	if (!btrfs_header_level(eb))
-		goto cleanup;
-
-	nritems = btrfs_header_nritems(eb);
-	generation = btrfs_header_generation(eb);
-	for (i = 0; i < nritems; i++) {
-		struct reada_extctl *rec;
-		u64 n_gen;
-		struct btrfs_key key;
-		struct btrfs_key next_key;
-
-		btrfs_node_key_to_cpu(eb, &key, i);
-		if (i + 1 < nritems)
-			btrfs_node_key_to_cpu(eb, &next_key, i + 1);
-		else
-			next_key = re->top;
-		bytenr = btrfs_node_blockptr(eb, i);
-		n_gen = btrfs_node_ptr_generation(eb, i);
-
-		list_for_each_entry(rec, &list, list) {
-			struct reada_control *rc = rec->rc;
-
-			/*
-			 * if the generation doesn't match, just ignore this
-			 * extctl. This will probably cut off a branch from
-			 * prefetch. Alternatively one could start a new (sub-)
-			 * prefetch for this branch, starting again from root.
-			 * FIXME: move the generation check out of this loop
-			 */
-#ifdef DEBUG
-			if (rec->generation != generation) {
-				btrfs_debug(fs_info,
-					    "generation mismatch for (%llu,%d,%llu) %llu != %llu",
-					    key.objectid, key.type, key.offset,
-					    rec->generation, generation);
-			}
-#endif
-			if (rec->generation == generation &&
-			    btrfs_comp_cpu_keys(&key, &rc->key_end) < 0 &&
-			    btrfs_comp_cpu_keys(&next_key, &rc->key_start) > 0)
-				reada_add_block(rc, bytenr, &next_key, n_gen);
-		}
-	}
-
-cleanup:
-	/*
-	 * free extctl records
-	 */
-	while (!list_empty(&list)) {
-		struct reada_control *rc;
-		struct reada_extctl *rec;
-
-		rec = list_first_entry(&list, struct reada_extctl, list);
-		list_del(&rec->list);
-		rc = rec->rc;
-		kfree(rec);
-
-		kref_get(&rc->refcnt);
-		if (atomic_dec_and_test(&rc->elems)) {
-			kref_put(&rc->refcnt, reada_control_release);
-			wake_up(&rc->wait);
-		}
-		kref_put(&rc->refcnt, reada_control_release);
-
-		reada_extent_put(fs_info, re);	/* one ref for each entry */
-	}
-
-	return;
-}
-
-int btree_readahead_hook(struct extent_buffer *eb, int err)
-{
-	struct btrfs_fs_info *fs_info = eb->fs_info;
-	int ret = 0;
-	struct reada_extent *re;
-
-	/* find extent */
-	spin_lock(&fs_info->reada_lock);
-	re = radix_tree_lookup(&fs_info->reada_tree,
-			       eb->start >> PAGE_SHIFT);
-	if (re)
-		re->refcnt++;
-	spin_unlock(&fs_info->reada_lock);
-	if (!re) {
-		ret = -1;
-		goto start_machine;
-	}
-
-	__readahead_hook(fs_info, re, eb, err);
-	reada_extent_put(fs_info, re);	/* our ref */
-
-start_machine:
-	reada_start_machine(fs_info);
-	return ret;
-}
-
-static struct reada_zone *reada_find_zone(struct btrfs_device *dev, u64 logical,
-					  struct btrfs_bio *bbio)
-{
-	struct btrfs_fs_info *fs_info = dev->fs_info;
-	int ret;
-	struct reada_zone *zone;
-	struct btrfs_block_group *cache = NULL;
-	u64 start;
-	u64 end;
-	int i;
-
-	zone = NULL;
-	spin_lock(&fs_info->reada_lock);
-	ret = radix_tree_gang_lookup(&dev->reada_zones, (void **)&zone,
-				     logical >> PAGE_SHIFT, 1);
-	if (ret == 1 && logical >= zone->start && logical <= zone->end) {
-		kref_get(&zone->refcnt);
-		spin_unlock(&fs_info->reada_lock);
-		return zone;
-	}
-
-	spin_unlock(&fs_info->reada_lock);
-
-	cache = btrfs_lookup_block_group(fs_info, logical);
-	if (!cache)
-		return NULL;
-
-	start = cache->start;
-	end = start + cache->length - 1;
-	btrfs_put_block_group(cache);
-
-	zone = kzalloc(sizeof(*zone), GFP_KERNEL);
-	if (!zone)
-		return NULL;
-
-	ret = radix_tree_preload(GFP_KERNEL);
-	if (ret) {
-		kfree(zone);
-		return NULL;
-	}
-
-	zone->start = start;
-	zone->end = end;
-	INIT_LIST_HEAD(&zone->list);
-	spin_lock_init(&zone->lock);
-	zone->locked = 0;
-	kref_init(&zone->refcnt);
-	zone->elems = 0;
-	zone->device = dev; /* our device always sits at index 0 */
-	for (i = 0; i < bbio->num_stripes; ++i) {
-		/* bounds have already been checked */
-		zone->devs[i] = bbio->stripes[i].dev;
-	}
-	zone->ndevs = bbio->num_stripes;
-
-	spin_lock(&fs_info->reada_lock);
-	ret = radix_tree_insert(&dev->reada_zones,
-				(unsigned long)(zone->end >> PAGE_SHIFT),
-				zone);
-
-	if (ret == -EEXIST) {
-		kfree(zone);
-		ret = radix_tree_gang_lookup(&dev->reada_zones, (void **)&zone,
-					     logical >> PAGE_SHIFT, 1);
-		if (ret == 1 && logical >= zone->start && logical <= zone->end)
-			kref_get(&zone->refcnt);
-		else
-			zone = NULL;
-	}
-	spin_unlock(&fs_info->reada_lock);
-	radix_tree_preload_end();
-
-	return zone;
-}
-
-static struct reada_extent *reada_find_extent(struct btrfs_fs_info *fs_info,
-					      u64 logical,
-					      struct btrfs_key *top)
-{
-	int ret;
-	struct reada_extent *re = NULL;
-	struct reada_extent *re_exist = NULL;
-	struct btrfs_bio *bbio = NULL;
-	struct btrfs_device *dev;
-	struct btrfs_device *prev_dev;
-	u64 length;
-	int real_stripes;
-	int nzones = 0;
-	unsigned long index = logical >> PAGE_SHIFT;
-	int dev_replace_is_ongoing;
-	int have_zone = 0;
-
-	spin_lock(&fs_info->reada_lock);
-	re = radix_tree_lookup(&fs_info->reada_tree, index);
-	if (re)
-		re->refcnt++;
-	spin_unlock(&fs_info->reada_lock);
-
-	if (re)
-		return re;
-
-	re = kzalloc(sizeof(*re), GFP_KERNEL);
-	if (!re)
-		return NULL;
-
-	re->logical = logical;
-	re->top = *top;
-	INIT_LIST_HEAD(&re->extctl);
-	spin_lock_init(&re->lock);
-	re->refcnt = 1;
-
-	/*
-	 * map block
-	 */
-	length = fs_info->nodesize;
-	ret = btrfs_map_block(fs_info, BTRFS_MAP_GET_READ_MIRRORS, logical,
-			&length, &bbio, 0);
-	if (ret || !bbio || length < fs_info->nodesize)
-		goto error;
-
-	if (bbio->num_stripes > BTRFS_MAX_MIRRORS) {
-		btrfs_err(fs_info,
-			   "readahead: more than %d copies not supported",
-			   BTRFS_MAX_MIRRORS);
-		goto error;
-	}
-
-	real_stripes = bbio->num_stripes - bbio->num_tgtdevs;
-	for (nzones = 0; nzones < real_stripes; ++nzones) {
-		struct reada_zone *zone;
-
-		dev = bbio->stripes[nzones].dev;
-
-		/* cannot read ahead on missing device. */
-		if (!dev->bdev)
-			continue;
-
-		zone = reada_find_zone(dev, logical, bbio);
-		if (!zone)
-			continue;
-
-		re->zones[re->nzones++] = zone;
-		spin_lock(&zone->lock);
-		if (!zone->elems)
-			kref_get(&zone->refcnt);
-		++zone->elems;
-		spin_unlock(&zone->lock);
-		spin_lock(&fs_info->reada_lock);
-		kref_put(&zone->refcnt, reada_zone_release);
-		spin_unlock(&fs_info->reada_lock);
-	}
-	if (re->nzones == 0) {
-		/* not a single zone found, error and out */
-		goto error;
-	}
-
-	/* Insert extent in reada tree + all per-device trees, all or nothing */
-	down_read(&fs_info->dev_replace.rwsem);
-	ret = radix_tree_preload(GFP_KERNEL);
-	if (ret) {
-		up_read(&fs_info->dev_replace.rwsem);
-		goto error;
-	}
-
-	spin_lock(&fs_info->reada_lock);
-	ret = radix_tree_insert(&fs_info->reada_tree, index, re);
-	if (ret == -EEXIST) {
-		re_exist = radix_tree_lookup(&fs_info->reada_tree, index);
-		re_exist->refcnt++;
-		spin_unlock(&fs_info->reada_lock);
-		radix_tree_preload_end();
-		up_read(&fs_info->dev_replace.rwsem);
-		goto error;
-	}
-	if (ret) {
-		spin_unlock(&fs_info->reada_lock);
-		radix_tree_preload_end();
-		up_read(&fs_info->dev_replace.rwsem);
-		goto error;
-	}
-	radix_tree_preload_end();
-	prev_dev = NULL;
-	dev_replace_is_ongoing = btrfs_dev_replace_is_ongoing(
-			&fs_info->dev_replace);
-	for (nzones = 0; nzones < re->nzones; ++nzones) {
-		dev = re->zones[nzones]->device;
-
-		if (dev == prev_dev) {
-			/*
-			 * in case of DUP, just add the first zone. As both
-			 * are on the same device, there's nothing to gain
-			 * from adding both.
-			 * Also, it wouldn't work, as the tree is per device
-			 * and adding would fail with EEXIST
-			 */
-			continue;
-		}
-		if (!dev->bdev)
-			continue;
-
-		if (dev_replace_is_ongoing &&
-		    dev == fs_info->dev_replace.tgtdev) {
-			/*
-			 * as this device is selected for reading only as
-			 * a last resort, skip it for read ahead.
-			 */
-			continue;
-		}
-		prev_dev = dev;
-		ret = radix_tree_insert(&dev->reada_extents, index, re);
-		if (ret) {
-			while (--nzones >= 0) {
-				dev = re->zones[nzones]->device;
-				BUG_ON(dev == NULL);
-				/* ignore whether the entry was inserted */
-				radix_tree_delete(&dev->reada_extents, index);
-			}
-			radix_tree_delete(&fs_info->reada_tree, index);
-			spin_unlock(&fs_info->reada_lock);
-			up_read(&fs_info->dev_replace.rwsem);
-			goto error;
-		}
-		have_zone = 1;
-	}
-	spin_unlock(&fs_info->reada_lock);
-	up_read(&fs_info->dev_replace.rwsem);
-
-	if (!have_zone)
-		goto error;
-
-	btrfs_put_bbio(bbio);
-	return re;
-
-error:
-	for (nzones = 0; nzones < re->nzones; ++nzones) {
-		struct reada_zone *zone;
-
-		zone = re->zones[nzones];
-		kref_get(&zone->refcnt);
-		spin_lock(&zone->lock);
-		--zone->elems;
-		if (zone->elems == 0) {
-			/*
-			 * no fs_info->reada_lock needed, as this can't be
-			 * the last ref
-			 */
-			kref_put(&zone->refcnt, reada_zone_release);
-		}
-		spin_unlock(&zone->lock);
-
-		spin_lock(&fs_info->reada_lock);
-		kref_put(&zone->refcnt, reada_zone_release);
-		spin_unlock(&fs_info->reada_lock);
-	}
-	btrfs_put_bbio(bbio);
-	kfree(re);
-	return re_exist;
-}
-
-static void reada_extent_put(struct btrfs_fs_info *fs_info,
-			     struct reada_extent *re)
-{
-	int i;
-	unsigned long index = re->logical >> PAGE_SHIFT;
-
-	spin_lock(&fs_info->reada_lock);
-	if (--re->refcnt) {
-		spin_unlock(&fs_info->reada_lock);
-		return;
-	}
-
-	radix_tree_delete(&fs_info->reada_tree, index);
-	for (i = 0; i < re->nzones; ++i) {
-		struct reada_zone *zone = re->zones[i];
-
-		radix_tree_delete(&zone->device->reada_extents, index);
-	}
-
-	spin_unlock(&fs_info->reada_lock);
-
-	for (i = 0; i < re->nzones; ++i) {
-		struct reada_zone *zone = re->zones[i];
-
-		kref_get(&zone->refcnt);
-		spin_lock(&zone->lock);
-		--zone->elems;
-		if (zone->elems == 0) {
-			/* no fs_info->reada_lock needed, as this can't be
-			 * the last ref */
-			kref_put(&zone->refcnt, reada_zone_release);
-		}
-		spin_unlock(&zone->lock);
-
-		spin_lock(&fs_info->reada_lock);
-		kref_put(&zone->refcnt, reada_zone_release);
-		spin_unlock(&fs_info->reada_lock);
-	}
-
-	kfree(re);
-}
-
-static void reada_zone_release(struct kref *kref)
-{
-	struct reada_zone *zone = container_of(kref, struct reada_zone, refcnt);
-
-	radix_tree_delete(&zone->device->reada_zones,
-			  zone->end >> PAGE_SHIFT);
-
-	kfree(zone);
-}
-
-static void reada_control_release(struct kref *kref)
-{
-	struct reada_control *rc = container_of(kref, struct reada_control,
-						refcnt);
-
-	kfree(rc);
-}
-
-static int reada_add_block(struct reada_control *rc, u64 logical,
-			   struct btrfs_key *top, u64 generation)
-{
-	struct btrfs_fs_info *fs_info = rc->fs_info;
-	struct reada_extent *re;
-	struct reada_extctl *rec;
-
-	/* takes one ref */
-	re = reada_find_extent(fs_info, logical, top);
-	if (!re)
-		return -1;
-
-	rec = kzalloc(sizeof(*rec), GFP_KERNEL);
-	if (!rec) {
-		reada_extent_put(fs_info, re);
-		return -ENOMEM;
-	}
-
-	rec->rc = rc;
-	rec->generation = generation;
-	atomic_inc(&rc->elems);
-
-	spin_lock(&re->lock);
-	list_add_tail(&rec->list, &re->extctl);
-	spin_unlock(&re->lock);
-
-	/* leave the ref on the extent */
-
-	return 0;
-}
-
-/*
- * called with fs_info->reada_lock held
- */
-static void reada_peer_zones_set_lock(struct reada_zone *zone, int lock)
-{
-	int i;
-	unsigned long index = zone->end >> PAGE_SHIFT;
-
-	for (i = 0; i < zone->ndevs; ++i) {
-		struct reada_zone *peer;
-		peer = radix_tree_lookup(&zone->devs[i]->reada_zones, index);
-		if (peer && peer->device != zone->device)
-			peer->locked = lock;
-	}
-}
-
-/*
- * called with fs_info->reada_lock held
- */
-static int reada_pick_zone(struct btrfs_device *dev)
-{
-	struct reada_zone *top_zone = NULL;
-	struct reada_zone *top_locked_zone = NULL;
-	u64 top_elems = 0;
-	u64 top_locked_elems = 0;
-	unsigned long index = 0;
-	int ret;
-
-	if (dev->reada_curr_zone) {
-		reada_peer_zones_set_lock(dev->reada_curr_zone, 0);
-		kref_put(&dev->reada_curr_zone->refcnt, reada_zone_release);
-		dev->reada_curr_zone = NULL;
-	}
-	/* pick the zone with the most elements */
-	while (1) {
-		struct reada_zone *zone;
-
-		ret = radix_tree_gang_lookup(&dev->reada_zones,
-					     (void **)&zone, index, 1);
-		if (ret == 0)
-			break;
-		index = (zone->end >> PAGE_SHIFT) + 1;
-		if (zone->locked) {
-			if (zone->elems > top_locked_elems) {
-				top_locked_elems = zone->elems;
-				top_locked_zone = zone;
-			}
-		} else {
-			if (zone->elems > top_elems) {
-				top_elems = zone->elems;
-				top_zone = zone;
-			}
-		}
-	}
-	if (top_zone)
-		dev->reada_curr_zone = top_zone;
-	else if (top_locked_zone)
-		dev->reada_curr_zone = top_locked_zone;
-	else
-		return 0;
-
-	dev->reada_next = dev->reada_curr_zone->start;
-	kref_get(&dev->reada_curr_zone->refcnt);
-	reada_peer_zones_set_lock(dev->reada_curr_zone, 1);
-
-	return 1;
-}
-
-static int reada_tree_block_flagged(struct btrfs_fs_info *fs_info, u64 bytenr,
-				    int mirror_num, struct extent_buffer **eb)
-{
-	struct extent_buffer *buf = NULL;
-	int ret;
-
-	buf = btrfs_find_create_tree_block(fs_info, bytenr);
-	if (IS_ERR(buf))
-		return 0;
-
-	set_bit(EXTENT_BUFFER_READAHEAD, &buf->bflags);
-
-	ret = read_extent_buffer_pages(buf, WAIT_PAGE_LOCK, mirror_num);
-	if (ret) {
-		free_extent_buffer_stale(buf);
-		return ret;
-	}
-
-	if (test_bit(EXTENT_BUFFER_CORRUPT, &buf->bflags)) {
-		free_extent_buffer_stale(buf);
-		return -EIO;
-	} else if (extent_buffer_uptodate(buf)) {
-		*eb = buf;
-	} else {
-		free_extent_buffer(buf);
-	}
-	return 0;
-}
-
-static int reada_start_machine_dev(struct btrfs_device *dev)
-{
-	struct btrfs_fs_info *fs_info = dev->fs_info;
-	struct reada_extent *re = NULL;
-	int mirror_num = 0;
-	struct extent_buffer *eb = NULL;
-	u64 logical;
-	int ret;
-	int i;
-
-	spin_lock(&fs_info->reada_lock);
-	if (dev->reada_curr_zone == NULL) {
-		ret = reada_pick_zone(dev);
-		if (!ret) {
-			spin_unlock(&fs_info->reada_lock);
-			return 0;
-		}
-	}
-	/*
-	 * FIXME currently we issue the reads one extent at a time. If we have
-	 * a contiguous block of extents, we could also coagulate them or use
-	 * plugging to speed things up
-	 */
-	ret = radix_tree_gang_lookup(&dev->reada_extents, (void **)&re,
-				     dev->reada_next >> PAGE_SHIFT, 1);
-	if (ret == 0 || re->logical > dev->reada_curr_zone->end) {
-		ret = reada_pick_zone(dev);
-		if (!ret) {
-			spin_unlock(&fs_info->reada_lock);
-			return 0;
-		}
-		re = NULL;
-		ret = radix_tree_gang_lookup(&dev->reada_extents, (void **)&re,
-					dev->reada_next >> PAGE_SHIFT, 1);
-	}
-	if (ret == 0) {
-		spin_unlock(&fs_info->reada_lock);
-		return 0;
-	}
-	dev->reada_next = re->logical + fs_info->nodesize;
-	re->refcnt++;
-
-	spin_unlock(&fs_info->reada_lock);
-
-	spin_lock(&re->lock);
-	if (re->scheduled || list_empty(&re->extctl)) {
-		spin_unlock(&re->lock);
-		reada_extent_put(fs_info, re);
-		return 0;
-	}
-	re->scheduled = 1;
-	spin_unlock(&re->lock);
-
-	/*
-	 * find mirror num
-	 */
-	for (i = 0; i < re->nzones; ++i) {
-		if (re->zones[i]->device == dev) {
-			mirror_num = i + 1;
-			break;
-		}
-	}
-	logical = re->logical;
-
-	atomic_inc(&dev->reada_in_flight);
-	ret = reada_tree_block_flagged(fs_info, logical, mirror_num, &eb);
-	if (ret)
-		__readahead_hook(fs_info, re, NULL, ret);
-	else if (eb)
-		__readahead_hook(fs_info, re, eb, ret);
-
-	if (eb)
-		free_extent_buffer(eb);
-
-	atomic_dec(&dev->reada_in_flight);
-	reada_extent_put(fs_info, re);
-
-	return 1;
-
-}
-
-static void reada_start_machine_worker(struct btrfs_work *work)
-{
-	struct reada_machine_work *rmw;
-	int old_ioprio;
-
-	rmw = container_of(work, struct reada_machine_work, work);
-
-	old_ioprio = IOPRIO_PRIO_VALUE(task_nice_ioclass(current),
-				       task_nice_ioprio(current));
-	set_task_ioprio(current, BTRFS_IOPRIO_READA);
-	__reada_start_machine(rmw->fs_info);
-	set_task_ioprio(current, old_ioprio);
-
-	atomic_dec(&rmw->fs_info->reada_works_cnt);
-
-	kfree(rmw);
-}
-
-static void __reada_start_machine(struct btrfs_fs_info *fs_info)
-{
-	struct btrfs_device *device;
-	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
-	u64 enqueued;
-	u64 total = 0;
-	int i;
-
-again:
-	do {
-		enqueued = 0;
-		mutex_lock(&fs_devices->device_list_mutex);
-		list_for_each_entry(device, &fs_devices->devices, dev_list) {
-			if (atomic_read(&device->reada_in_flight) <
-			    MAX_IN_FLIGHT)
-				enqueued += reada_start_machine_dev(device);
-		}
-		mutex_unlock(&fs_devices->device_list_mutex);
-		total += enqueued;
-	} while (enqueued && total < 10000);
-	if (fs_devices->seed) {
-		fs_devices = fs_devices->seed;
-		goto again;
-	}
-
-	if (enqueued == 0)
-		return;
-
-	/*
-	 * If everything is already in the cache, this is effectively single
-	 * threaded. To a) not hold the caller for too long and b) to utilize
-	 * more cores, we broke the loop above after 10000 iterations and now
-	 * enqueue to workers to finish it. This will distribute the load to
-	 * the cores.
-	 */
-	for (i = 0; i < 2; ++i) {
-		reada_start_machine(fs_info);
-		if (atomic_read(&fs_info->reada_works_cnt) >
-		    BTRFS_MAX_MIRRORS * 2)
-			break;
-	}
-}
-
-static void reada_start_machine(struct btrfs_fs_info *fs_info)
-{
-	struct reada_machine_work *rmw;
-
-	rmw = kzalloc(sizeof(*rmw), GFP_KERNEL);
-	if (!rmw) {
-		/* FIXME we cannot handle this properly right now */
-		BUG();
-	}
-	btrfs_init_work(&rmw->work, reada_start_machine_worker, NULL, NULL);
-	rmw->fs_info = fs_info;
-
-	btrfs_queue_work(fs_info->readahead_workers, &rmw->work);
-	atomic_inc(&fs_info->reada_works_cnt);
-}
-
-#ifdef DEBUG
-static void dump_devs(struct btrfs_fs_info *fs_info, int all)
-{
-	struct btrfs_device *device;
-	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
-	unsigned long index;
-	int ret;
-	int i;
-	int j;
-	int cnt;
-
-	spin_lock(&fs_info->reada_lock);
-	list_for_each_entry(device, &fs_devices->devices, dev_list) {
-		btrfs_debug(fs_info, "dev %lld has %d in flight", device->devid,
-			atomic_read(&device->reada_in_flight));
-		index = 0;
-		while (1) {
-			struct reada_zone *zone;
-			ret = radix_tree_gang_lookup(&device->reada_zones,
-						     (void **)&zone, index, 1);
-			if (ret == 0)
-				break;
-			pr_debug("  zone %llu-%llu elems %llu locked %d devs",
-				    zone->start, zone->end, zone->elems,
-				    zone->locked);
-			for (j = 0; j < zone->ndevs; ++j) {
-				pr_cont(" %lld",
-					zone->devs[j]->devid);
-			}
-			if (device->reada_curr_zone == zone)
-				pr_cont(" curr off %llu",
-					device->reada_next - zone->start);
-			pr_cont("\n");
-			index = (zone->end >> PAGE_SHIFT) + 1;
-		}
-		cnt = 0;
-		index = 0;
-		while (all) {
-			struct reada_extent *re = NULL;
-
-			ret = radix_tree_gang_lookup(&device->reada_extents,
-						     (void **)&re, index, 1);
-			if (ret == 0)
-				break;
-			pr_debug("  re: logical %llu size %u empty %d scheduled %d",
-				re->logical, fs_info->nodesize,
-				list_empty(&re->extctl), re->scheduled);
-
-			for (i = 0; i < re->nzones; ++i) {
-				pr_cont(" zone %llu-%llu devs",
-					re->zones[i]->start,
-					re->zones[i]->end);
-				for (j = 0; j < re->zones[i]->ndevs; ++j) {
-					pr_cont(" %lld",
-						re->zones[i]->devs[j]->devid);
-				}
-			}
-			pr_cont("\n");
-			index = (re->logical >> PAGE_SHIFT) + 1;
-			if (++cnt > 15)
-				break;
-		}
-	}
-
-	index = 0;
-	cnt = 0;
-	while (all) {
-		struct reada_extent *re = NULL;
-
-		ret = radix_tree_gang_lookup(&fs_info->reada_tree, (void **)&re,
-					     index, 1);
-		if (ret == 0)
-			break;
-		if (!re->scheduled) {
-			index = (re->logical >> PAGE_SHIFT) + 1;
-			continue;
-		}
-		pr_debug("re: logical %llu size %u list empty %d scheduled %d",
-			re->logical, fs_info->nodesize,
-			list_empty(&re->extctl), re->scheduled);
-		for (i = 0; i < re->nzones; ++i) {
-			pr_cont(" zone %llu-%llu devs",
-				re->zones[i]->start,
-				re->zones[i]->end);
-			for (j = 0; j < re->zones[i]->ndevs; ++j) {
-				pr_cont(" %lld",
-				       re->zones[i]->devs[j]->devid);
-			}
-		}
-		pr_cont("\n");
-		index = (re->logical >> PAGE_SHIFT) + 1;
-	}
-	spin_unlock(&fs_info->reada_lock);
-}
-#endif
-
-/*
- * interface
- */
-struct reada_control *btrfs_reada_add(struct btrfs_root *root,
-			struct btrfs_key *key_start, struct btrfs_key *key_end)
-{
-	struct reada_control *rc;
-	u64 start;
-	u64 generation;
-	int ret;
-	struct extent_buffer *node;
-	static struct btrfs_key max_key = {
-		.objectid = (u64)-1,
-		.type = (u8)-1,
-		.offset = (u64)-1
-	};
-
-	rc = kzalloc(sizeof(*rc), GFP_KERNEL);
-	if (!rc)
-		return ERR_PTR(-ENOMEM);
-
-	rc->fs_info = root->fs_info;
-	rc->key_start = *key_start;
-	rc->key_end = *key_end;
-	atomic_set(&rc->elems, 0);
-	init_waitqueue_head(&rc->wait);
-	kref_init(&rc->refcnt);
-	kref_get(&rc->refcnt); /* one ref for having elements */
-
-	node = btrfs_root_node(root);
-	start = node->start;
-	generation = btrfs_header_generation(node);
-	free_extent_buffer(node);
-
-	ret = reada_add_block(rc, start, &max_key, generation);
-	if (ret) {
-		kfree(rc);
-		return ERR_PTR(ret);
-	}
-
-	reada_start_machine(root->fs_info);
-
-	return rc;
-}
-
-#ifdef DEBUG
-int btrfs_reada_wait(void *handle)
-{
-	struct reada_control *rc = handle;
-	struct btrfs_fs_info *fs_info = rc->fs_info;
-
-	while (atomic_read(&rc->elems)) {
-		if (!atomic_read(&fs_info->reada_works_cnt))
-			reada_start_machine(fs_info);
-		wait_event_timeout(rc->wait, atomic_read(&rc->elems) == 0,
-				   5 * HZ);
-		dump_devs(fs_info, atomic_read(&rc->elems) < 10 ? 1 : 0);
-	}
-
-	dump_devs(fs_info, atomic_read(&rc->elems) < 10 ? 1 : 0);
-
-	kref_put(&rc->refcnt, reada_control_release);
-
-	return 0;
-}
-#else
-int btrfs_reada_wait(void *handle)
-{
-	struct reada_control *rc = handle;
-	struct btrfs_fs_info *fs_info = rc->fs_info;
-
-	while (atomic_read(&rc->elems)) {
-		if (!atomic_read(&fs_info->reada_works_cnt))
-			reada_start_machine(fs_info);
-		wait_event_timeout(rc->wait, atomic_read(&rc->elems) == 0,
-				   (HZ + 9) / 10);
-	}
-
-	kref_put(&rc->refcnt, reada_control_release);
-
-	return 0;
-}
-#endif
-
-void btrfs_reada_detach(void *handle)
-{
-	struct reada_control *rc = handle;
-
-	kref_put(&rc->refcnt, reada_control_release);
-}
diff --git a/fs/btrfs/ref-verify.c b/fs/btrfs/ref-verify.c
index 454a1015d026..a248f46cfe72 100644
--- a/fs/btrfs/ref-verify.c
+++ b/fs/btrfs/ref-verify.c
@@ -218,11 +218,11 @@ static void __print_stack_trace(struct btrfs_fs_info *fs_info,
 	stack_trace_print(ra->trace, ra->trace_len, 2);
 }
 #else
-static void inline __save_stack_trace(struct ref_action *ra)
+static inline void __save_stack_trace(struct ref_action *ra)
 {
 }
 
-static void inline __print_stack_trace(struct btrfs_fs_info *fs_info,
+static inline void __print_stack_trace(struct btrfs_fs_info *fs_info,
 				       struct ref_action *ra)
 {
 	btrfs_err(fs_info, "  ref-verify: no stacktrace support");
@@ -264,8 +264,8 @@ static struct block_entry *add_block_entry(struct btrfs_fs_info *fs_info,
 	struct block_entry *be = NULL, *exist;
 	struct root_entry *re = NULL;
 
-	re = kzalloc(sizeof(struct root_entry), GFP_KERNEL);
-	be = kzalloc(sizeof(struct block_entry), GFP_KERNEL);
+	re = kzalloc(sizeof(struct root_entry), GFP_NOFS);
+	be = kzalloc(sizeof(struct block_entry), GFP_NOFS);
 	if (!be || !re) {
 		kfree(re);
 		kfree(be);
@@ -286,6 +286,8 @@ static struct block_entry *add_block_entry(struct btrfs_fs_info *fs_info,
 			exist_re = insert_root_entry(&exist->roots, re);
 			if (exist_re)
 				kfree(re);
+		} else {
+			kfree(re);
 		}
 		kfree(be);
 		return exist;
@@ -311,7 +313,7 @@ static int add_tree_block(struct btrfs_fs_info *fs_info, u64 ref_root,
 	struct root_entry *re;
 	struct ref_entry *ref = NULL, *exist;
 
-	ref = kmalloc(sizeof(struct ref_entry), GFP_KERNEL);
+	ref = kmalloc(sizeof(struct ref_entry), GFP_NOFS);
 	if (!ref)
 		return -ENOMEM;
 
@@ -356,7 +358,7 @@ static int add_shared_data_ref(struct btrfs_fs_info *fs_info,
 	struct block_entry *be;
 	struct ref_entry *ref;
 
-	ref = kzalloc(sizeof(struct ref_entry), GFP_KERNEL);
+	ref = kzalloc(sizeof(struct ref_entry), GFP_NOFS);
 	if (!ref)
 		return -ENOMEM;
 	be = add_block_entry(fs_info, bytenr, num_bytes, 0);
@@ -391,7 +393,7 @@ static int add_extent_data_ref(struct btrfs_fs_info *fs_info,
 	u64 offset = btrfs_extent_data_ref_offset(leaf, dref);
 	u32 num_refs = btrfs_extent_data_ref_count(leaf, dref);
 
-	ref = kzalloc(sizeof(struct ref_entry), GFP_KERNEL);
+	ref = kzalloc(sizeof(struct ref_entry), GFP_NOFS);
 	if (!ref)
 		return -ENOMEM;
 	be = add_block_entry(fs_info, bytenr, num_bytes, ref_root);
@@ -433,7 +435,7 @@ static int process_extent_item(struct btrfs_fs_info *fs_info,
 	struct btrfs_extent_data_ref *dref;
 	struct btrfs_shared_data_ref *sref;
 	struct extent_buffer *leaf = path->nodes[0];
-	u32 item_size = btrfs_item_size_nr(leaf, slot);
+	u32 item_size = btrfs_item_size(leaf, slot);
 	unsigned long end, ptr;
 	u64 offset, flags, count;
 	int type, ret;
@@ -493,14 +495,15 @@ static int process_extent_item(struct btrfs_fs_info *fs_info,
 }
 
 static int process_leaf(struct btrfs_root *root,
-			struct btrfs_path *path, u64 *bytenr, u64 *num_bytes)
+			struct btrfs_path *path, u64 *bytenr, u64 *num_bytes,
+			int *tree_block_level)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct extent_buffer *leaf = path->nodes[0];
 	struct btrfs_extent_data_ref *dref;
 	struct btrfs_shared_data_ref *sref;
 	u32 count;
-	int i = 0, tree_block_level = 0, ret = 0;
+	int i = 0, ret = 0;
 	struct btrfs_key key;
 	int nritems = btrfs_header_nritems(leaf);
 
@@ -509,19 +512,19 @@ static int process_leaf(struct btrfs_root *root,
 		switch (key.type) {
 		case BTRFS_EXTENT_ITEM_KEY:
 			*num_bytes = key.offset;
-			/* fall through */
+			fallthrough;
 		case BTRFS_METADATA_ITEM_KEY:
 			*bytenr = key.objectid;
 			ret = process_extent_item(fs_info, path, &key, i,
-						  &tree_block_level);
+						  tree_block_level);
 			break;
 		case BTRFS_TREE_BLOCK_REF_KEY:
 			ret = add_tree_block(fs_info, key.offset, 0,
-					     key.objectid, tree_block_level);
+					     key.objectid, *tree_block_level);
 			break;
 		case BTRFS_SHARED_BLOCK_REF_KEY:
 			ret = add_tree_block(fs_info, 0, key.offset,
-					     key.objectid, tree_block_level);
+					     key.objectid, *tree_block_level);
 			break;
 		case BTRFS_EXTENT_DATA_REF_KEY:
 			dref = btrfs_item_ptr(leaf, i,
@@ -547,38 +550,25 @@ static int process_leaf(struct btrfs_root *root,
 
 /* Walk down to the leaf from the given level */
 static int walk_down_tree(struct btrfs_root *root, struct btrfs_path *path,
-			  int level, u64 *bytenr, u64 *num_bytes)
+			  int level, u64 *bytenr, u64 *num_bytes,
+			  int *tree_block_level)
 {
-	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct extent_buffer *eb;
-	u64 block_bytenr, gen;
 	int ret = 0;
 
 	while (level >= 0) {
 		if (level) {
-			struct btrfs_key first_key;
-
-			block_bytenr = btrfs_node_blockptr(path->nodes[level],
-							   path->slots[level]);
-			gen = btrfs_node_ptr_generation(path->nodes[level],
-							path->slots[level]);
-			btrfs_node_key_to_cpu(path->nodes[level], &first_key,
-					      path->slots[level]);
-			eb = read_tree_block(fs_info, block_bytenr, gen,
-					     level - 1, &first_key);
+			eb = btrfs_read_node_slot(path->nodes[level],
+						  path->slots[level]);
 			if (IS_ERR(eb))
 				return PTR_ERR(eb);
-			if (!extent_buffer_uptodate(eb)) {
-				free_extent_buffer(eb);
-				return -EIO;
-			}
 			btrfs_tree_read_lock(eb);
-			btrfs_set_lock_blocking_read(eb);
 			path->nodes[level-1] = eb;
 			path->slots[level-1] = 0;
-			path->locks[level-1] = BTRFS_READ_LOCK_BLOCKING;
+			path->locks[level-1] = BTRFS_READ_LOCK;
 		} else {
-			ret = process_leaf(root, path, bytenr, num_bytes);
+			ret = process_leaf(root, path, bytenr, num_bytes,
+					   tree_block_level);
 			if (ret)
 				break;
 		}
@@ -679,19 +669,19 @@ int btrfs_ref_tree_mod(struct btrfs_fs_info *fs_info,
 	u64 bytenr = generic_ref->bytenr;
 	u64 num_bytes = generic_ref->len;
 	u64 parent = generic_ref->parent;
-	u64 ref_root;
-	u64 owner;
-	u64 offset;
+	u64 ref_root = 0;
+	u64 owner = 0;
+	u64 offset = 0;
 
 	if (!btrfs_test_opt(fs_info, REF_VERIFY))
 		return 0;
 
 	if (generic_ref->type == BTRFS_REF_METADATA) {
-		ref_root = generic_ref->tree_ref.root;
+		if (!parent)
+			ref_root = generic_ref->tree_ref.owning_root;
 		owner = generic_ref->tree_ref.level;
-		offset = 0;
-	} else {
-		ref_root = generic_ref->data_ref.ref_root;
+	} else if (!parent) {
+		ref_root = generic_ref->data_ref.owning_root;
 		owner = generic_ref->data_ref.ino;
 		offset = generic_ref->data_ref.offset;
 	}
@@ -706,13 +696,10 @@ int btrfs_ref_tree_mod(struct btrfs_fs_info *fs_info,
 		goto out;
 	}
 
-	if (parent) {
-		ref->parent = parent;
-	} else {
-		ref->root_objectid = ref_root;
-		ref->owner = owner;
-		ref->offset = offset;
-	}
+	ref->parent = parent;
+	ref->owner = owner;
+	ref->root_objectid = ref_root;
+	ref->offset = offset;
 	ref->num_refs = (action == BTRFS_DROP_DELAYED_REF) ? -1 : 1;
 
 	memcpy(&ra->ref, ref, sizeof(struct ref_entry));
@@ -797,8 +784,16 @@ int btrfs_ref_tree_mod(struct btrfs_fs_info *fs_info,
 		if (!be) {
 			btrfs_err(fs_info,
 "trying to do action %d to bytenr %llu num_bytes %llu but there is no existing entry!",
-				  action, (unsigned long long)bytenr,
-				  (unsigned long long)num_bytes);
+				  action, bytenr, num_bytes);
+			dump_ref_action(fs_info, ra);
+			kfree(ref);
+			kfree(ra);
+			goto out_unlock;
+		} else if (be->num_refs == 0) {
+			btrfs_err(fs_info,
+		"trying to do action %d for a bytenr that has 0 total references",
+				action);
+			dump_block_entry(fs_info, be);
 			dump_ref_action(fs_info, ra);
 			kfree(ref);
 			kfree(ra);
@@ -849,6 +844,7 @@ int btrfs_ref_tree_mod(struct btrfs_fs_info *fs_info,
 "dropping a ref for a root that doesn't have a ref on the block");
 			dump_block_entry(fs_info, be);
 			dump_ref_action(fs_info, ra);
+			kfree(ref);
 			kfree(ra);
 			goto out_unlock;
 		}
@@ -976,8 +972,10 @@ void btrfs_free_ref_tree_range(struct btrfs_fs_info *fs_info, u64 start,
 /* Walk down all roots and build the ref tree, meant to be called at mount */
 int btrfs_build_ref_tree(struct btrfs_fs_info *fs_info)
 {
+	struct btrfs_root *extent_root;
 	struct btrfs_path *path;
 	struct extent_buffer *eb;
+	int tree_block_level = 0;
 	u64 bytenr = 0, num_bytes = 0;
 	int ret, level;
 
@@ -988,12 +986,12 @@ int btrfs_build_ref_tree(struct btrfs_fs_info *fs_info)
 	if (!path)
 		return -ENOMEM;
 
-	eb = btrfs_read_lock_root_node(fs_info->extent_root);
-	btrfs_set_lock_blocking_read(eb);
+	extent_root = btrfs_extent_root(fs_info, 0);
+	eb = btrfs_read_lock_root_node(extent_root);
 	level = btrfs_header_level(eb);
 	path->nodes[level] = eb;
 	path->slots[level] = 0;
-	path->locks[level] = BTRFS_READ_LOCK_BLOCKING;
+	path->locks[level] = BTRFS_READ_LOCK;
 
 	while (1) {
 		/*
@@ -1002,8 +1000,8 @@ int btrfs_build_ref_tree(struct btrfs_fs_info *fs_info)
 		 * would have had to added a ref key item which may appear on a
 		 * different leaf from the original extent item.
 		 */
-		ret = walk_down_tree(fs_info->extent_root, path, level,
-				     &bytenr, &num_bytes);
+		ret = walk_down_tree(extent_root, path, level,
+				     &bytenr, &num_bytes, &tree_block_level);
 		if (ret)
 			break;
 		ret = walk_up_tree(path, &level);
diff --git a/fs/btrfs/reflink.c b/fs/btrfs/reflink.c
new file mode 100644
index 000000000000..f50586ff85c8
--- /dev/null
+++ b/fs/btrfs/reflink.c
@@ -0,0 +1,930 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <linux/blkdev.h>
+#include <linux/iversion.h>
+#include "compression.h"
+#include "ctree.h"
+#include "delalloc-space.h"
+#include "disk-io.h"
+#include "reflink.h"
+#include "transaction.h"
+#include "subpage.h"
+
+#define BTRFS_MAX_DEDUPE_LEN	SZ_16M
+
+static int clone_finish_inode_update(struct btrfs_trans_handle *trans,
+				     struct inode *inode,
+				     u64 endoff,
+				     const u64 destoff,
+				     const u64 olen,
+				     int no_time_update)
+{
+	struct btrfs_root *root = BTRFS_I(inode)->root;
+	int ret;
+
+	inode_inc_iversion(inode);
+	if (!no_time_update) {
+		inode->i_mtime = current_time(inode);
+		inode->i_ctime = inode->i_mtime;
+	}
+	/*
+	 * We round up to the block size at eof when determining which
+	 * extents to clone above, but shouldn't round up the file size.
+	 */
+	if (endoff > destoff + olen)
+		endoff = destoff + olen;
+	if (endoff > inode->i_size) {
+		i_size_write(inode, endoff);
+		btrfs_inode_safe_disk_i_size_write(BTRFS_I(inode), 0);
+	}
+
+	ret = btrfs_update_inode(trans, root, BTRFS_I(inode));
+	if (ret) {
+		btrfs_abort_transaction(trans, ret);
+		btrfs_end_transaction(trans);
+		goto out;
+	}
+	ret = btrfs_end_transaction(trans);
+out:
+	return ret;
+}
+
+static int copy_inline_to_page(struct btrfs_inode *inode,
+			       const u64 file_offset,
+			       char *inline_data,
+			       const u64 size,
+			       const u64 datal,
+			       const u8 comp_type)
+{
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
+	const u32 block_size = fs_info->sectorsize;
+	const u64 range_end = file_offset + block_size - 1;
+	const size_t inline_size = size - btrfs_file_extent_calc_inline_size(0);
+	char *data_start = inline_data + btrfs_file_extent_calc_inline_size(0);
+	struct extent_changeset *data_reserved = NULL;
+	struct page *page = NULL;
+	struct address_space *mapping = inode->vfs_inode.i_mapping;
+	int ret;
+
+	ASSERT(IS_ALIGNED(file_offset, block_size));
+
+	/*
+	 * We have flushed and locked the ranges of the source and destination
+	 * inodes, we also have locked the inodes, so we are safe to do a
+	 * reservation here. Also we must not do the reservation while holding
+	 * a transaction open, otherwise we would deadlock.
+	 */
+	ret = btrfs_delalloc_reserve_space(inode, &data_reserved, file_offset,
+					   block_size);
+	if (ret)
+		goto out;
+
+	page = find_or_create_page(mapping, file_offset >> PAGE_SHIFT,
+				   btrfs_alloc_write_mask(mapping));
+	if (!page) {
+		ret = -ENOMEM;
+		goto out_unlock;
+	}
+
+	ret = set_page_extent_mapped(page);
+	if (ret < 0)
+		goto out_unlock;
+
+	clear_extent_bit(&inode->io_tree, file_offset, range_end,
+			 EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG,
+			 NULL);
+	ret = btrfs_set_extent_delalloc(inode, file_offset, range_end, 0, NULL);
+	if (ret)
+		goto out_unlock;
+
+	/*
+	 * After dirtying the page our caller will need to start a transaction,
+	 * and if we are low on metadata free space, that can cause flushing of
+	 * delalloc for all inodes in order to get metadata space released.
+	 * However we are holding the range locked for the whole duration of
+	 * the clone/dedupe operation, so we may deadlock if that happens and no
+	 * other task releases enough space. So mark this inode as not being
+	 * possible to flush to avoid such deadlock. We will clear that flag
+	 * when we finish cloning all extents, since a transaction is started
+	 * after finding each extent to clone.
+	 */
+	set_bit(BTRFS_INODE_NO_DELALLOC_FLUSH, &inode->runtime_flags);
+
+	if (comp_type == BTRFS_COMPRESS_NONE) {
+		memcpy_to_page(page, offset_in_page(file_offset), data_start,
+			       datal);
+	} else {
+		ret = btrfs_decompress(comp_type, data_start, page,
+				       offset_in_page(file_offset),
+				       inline_size, datal);
+		if (ret)
+			goto out_unlock;
+		flush_dcache_page(page);
+	}
+
+	/*
+	 * If our inline data is smaller then the block/page size, then the
+	 * remaining of the block/page is equivalent to zeroes. We had something
+	 * like the following done:
+	 *
+	 * $ xfs_io -f -c "pwrite -S 0xab 0 500" file
+	 * $ sync  # (or fsync)
+	 * $ xfs_io -c "falloc 0 4K" file
+	 * $ xfs_io -c "pwrite -S 0xcd 4K 4K"
+	 *
+	 * So what's in the range [500, 4095] corresponds to zeroes.
+	 */
+	if (datal < block_size)
+		memzero_page(page, datal, block_size - datal);
+
+	btrfs_page_set_uptodate(fs_info, page, file_offset, block_size);
+	btrfs_page_clear_checked(fs_info, page, file_offset, block_size);
+	btrfs_page_set_dirty(fs_info, page, file_offset, block_size);
+out_unlock:
+	if (page) {
+		unlock_page(page);
+		put_page(page);
+	}
+	if (ret)
+		btrfs_delalloc_release_space(inode, data_reserved, file_offset,
+					     block_size, true);
+	btrfs_delalloc_release_extents(inode, block_size);
+out:
+	extent_changeset_free(data_reserved);
+
+	return ret;
+}
+
+/*
+ * Deal with cloning of inline extents. We try to copy the inline extent from
+ * the source inode to destination inode when possible. When not possible we
+ * copy the inline extent's data into the respective page of the inode.
+ */
+static int clone_copy_inline_extent(struct inode *dst,
+				    struct btrfs_path *path,
+				    struct btrfs_key *new_key,
+				    const u64 drop_start,
+				    const u64 datal,
+				    const u64 size,
+				    const u8 comp_type,
+				    char *inline_data,
+				    struct btrfs_trans_handle **trans_out)
+{
+	struct btrfs_fs_info *fs_info = btrfs_sb(dst->i_sb);
+	struct btrfs_root *root = BTRFS_I(dst)->root;
+	const u64 aligned_end = ALIGN(new_key->offset + datal,
+				      fs_info->sectorsize);
+	struct btrfs_trans_handle *trans = NULL;
+	struct btrfs_drop_extents_args drop_args = { 0 };
+	int ret;
+	struct btrfs_key key;
+
+	if (new_key->offset > 0) {
+		ret = copy_inline_to_page(BTRFS_I(dst), new_key->offset,
+					  inline_data, size, datal, comp_type);
+		goto out;
+	}
+
+	key.objectid = btrfs_ino(BTRFS_I(dst));
+	key.type = BTRFS_EXTENT_DATA_KEY;
+	key.offset = 0;
+	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
+	if (ret < 0) {
+		return ret;
+	} else if (ret > 0) {
+		if (path->slots[0] >= btrfs_header_nritems(path->nodes[0])) {
+			ret = btrfs_next_leaf(root, path);
+			if (ret < 0)
+				return ret;
+			else if (ret > 0)
+				goto copy_inline_extent;
+		}
+		btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
+		if (key.objectid == btrfs_ino(BTRFS_I(dst)) &&
+		    key.type == BTRFS_EXTENT_DATA_KEY) {
+			/*
+			 * There's an implicit hole at file offset 0, copy the
+			 * inline extent's data to the page.
+			 */
+			ASSERT(key.offset > 0);
+			goto copy_to_page;
+		}
+	} else if (i_size_read(dst) <= datal) {
+		struct btrfs_file_extent_item *ei;
+
+		ei = btrfs_item_ptr(path->nodes[0], path->slots[0],
+				    struct btrfs_file_extent_item);
+		/*
+		 * If it's an inline extent replace it with the source inline
+		 * extent, otherwise copy the source inline extent data into
+		 * the respective page at the destination inode.
+		 */
+		if (btrfs_file_extent_type(path->nodes[0], ei) ==
+		    BTRFS_FILE_EXTENT_INLINE)
+			goto copy_inline_extent;
+
+		goto copy_to_page;
+	}
+
+copy_inline_extent:
+	/*
+	 * We have no extent items, or we have an extent at offset 0 which may
+	 * or may not be inlined. All these cases are dealt the same way.
+	 */
+	if (i_size_read(dst) > datal) {
+		/*
+		 * At the destination offset 0 we have either a hole, a regular
+		 * extent or an inline extent larger then the one we want to
+		 * clone. Deal with all these cases by copying the inline extent
+		 * data into the respective page at the destination inode.
+		 */
+		goto copy_to_page;
+	}
+
+	/*
+	 * Release path before starting a new transaction so we don't hold locks
+	 * that would confuse lockdep.
+	 */
+	btrfs_release_path(path);
+	/*
+	 * If we end up here it means were copy the inline extent into a leaf
+	 * of the destination inode. We know we will drop or adjust at most one
+	 * extent item in the destination root.
+	 *
+	 * 1 unit - adjusting old extent (we may have to split it)
+	 * 1 unit - add new extent
+	 * 1 unit - inode update
+	 */
+	trans = btrfs_start_transaction(root, 3);
+	if (IS_ERR(trans)) {
+		ret = PTR_ERR(trans);
+		trans = NULL;
+		goto out;
+	}
+	drop_args.path = path;
+	drop_args.start = drop_start;
+	drop_args.end = aligned_end;
+	drop_args.drop_cache = true;
+	ret = btrfs_drop_extents(trans, root, BTRFS_I(dst), &drop_args);
+	if (ret)
+		goto out;
+	ret = btrfs_insert_empty_item(trans, root, path, new_key, size);
+	if (ret)
+		goto out;
+
+	write_extent_buffer(path->nodes[0], inline_data,
+			    btrfs_item_ptr_offset(path->nodes[0],
+						  path->slots[0]),
+			    size);
+	btrfs_update_inode_bytes(BTRFS_I(dst), datal, drop_args.bytes_found);
+	btrfs_set_inode_full_sync(BTRFS_I(dst));
+	ret = btrfs_inode_set_file_extent_range(BTRFS_I(dst), 0, aligned_end);
+out:
+	if (!ret && !trans) {
+		/*
+		 * No transaction here means we copied the inline extent into a
+		 * page of the destination inode.
+		 *
+		 * 1 unit to update inode item
+		 */
+		trans = btrfs_start_transaction(root, 1);
+		if (IS_ERR(trans)) {
+			ret = PTR_ERR(trans);
+			trans = NULL;
+		}
+	}
+	if (ret && trans) {
+		btrfs_abort_transaction(trans, ret);
+		btrfs_end_transaction(trans);
+	}
+	if (!ret)
+		*trans_out = trans;
+
+	return ret;
+
+copy_to_page:
+	/*
+	 * Release our path because we don't need it anymore and also because
+	 * copy_inline_to_page() needs to reserve data and metadata, which may
+	 * need to flush delalloc when we are low on available space and
+	 * therefore cause a deadlock if writeback of an inline extent needs to
+	 * write to the same leaf or an ordered extent completion needs to write
+	 * to the same leaf.
+	 */
+	btrfs_release_path(path);
+
+	ret = copy_inline_to_page(BTRFS_I(dst), new_key->offset,
+				  inline_data, size, datal, comp_type);
+	goto out;
+}
+
+/**
+ * btrfs_clone() - clone a range from inode file to another
+ *
+ * @src: Inode to clone from
+ * @inode: Inode to clone to
+ * @off: Offset within source to start clone from
+ * @olen: Original length, passed by user, of range to clone
+ * @olen_aligned: Block-aligned value of olen
+ * @destoff: Offset within @inode to start clone
+ * @no_time_update: Whether to update mtime/ctime on the target inode
+ */
+static int btrfs_clone(struct inode *src, struct inode *inode,
+		       const u64 off, const u64 olen, const u64 olen_aligned,
+		       const u64 destoff, int no_time_update)
+{
+	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+	struct btrfs_path *path = NULL;
+	struct extent_buffer *leaf;
+	struct btrfs_trans_handle *trans;
+	char *buf = NULL;
+	struct btrfs_key key;
+	u32 nritems;
+	int slot;
+	int ret;
+	const u64 len = olen_aligned;
+	u64 last_dest_end = destoff;
+	u64 prev_extent_end = off;
+
+	ret = -ENOMEM;
+	buf = kvmalloc(fs_info->nodesize, GFP_KERNEL);
+	if (!buf)
+		return ret;
+
+	path = btrfs_alloc_path();
+	if (!path) {
+		kvfree(buf);
+		return ret;
+	}
+
+	path->reada = READA_FORWARD;
+	/* Clone data */
+	key.objectid = btrfs_ino(BTRFS_I(src));
+	key.type = BTRFS_EXTENT_DATA_KEY;
+	key.offset = off;
+
+	while (1) {
+		struct btrfs_file_extent_item *extent;
+		u64 extent_gen;
+		int type;
+		u32 size;
+		struct btrfs_key new_key;
+		u64 disko = 0, diskl = 0;
+		u64 datao = 0, datal = 0;
+		u8 comp;
+		u64 drop_start;
+
+		/* Note the key will change type as we walk through the tree */
+		ret = btrfs_search_slot(NULL, BTRFS_I(src)->root, &key, path,
+				0, 0);
+		if (ret < 0)
+			goto out;
+		/*
+		 * First search, if no extent item that starts at offset off was
+		 * found but the previous item is an extent item, it's possible
+		 * it might overlap our target range, therefore process it.
+		 */
+		if (key.offset == off && ret > 0 && path->slots[0] > 0) {
+			btrfs_item_key_to_cpu(path->nodes[0], &key,
+					      path->slots[0] - 1);
+			if (key.type == BTRFS_EXTENT_DATA_KEY)
+				path->slots[0]--;
+		}
+
+		nritems = btrfs_header_nritems(path->nodes[0]);
+process_slot:
+		if (path->slots[0] >= nritems) {
+			ret = btrfs_next_leaf(BTRFS_I(src)->root, path);
+			if (ret < 0)
+				goto out;
+			if (ret > 0)
+				break;
+			nritems = btrfs_header_nritems(path->nodes[0]);
+		}
+		leaf = path->nodes[0];
+		slot = path->slots[0];
+
+		btrfs_item_key_to_cpu(leaf, &key, slot);
+		if (key.type > BTRFS_EXTENT_DATA_KEY ||
+		    key.objectid != btrfs_ino(BTRFS_I(src)))
+			break;
+
+		ASSERT(key.type == BTRFS_EXTENT_DATA_KEY);
+
+		extent = btrfs_item_ptr(leaf, slot,
+					struct btrfs_file_extent_item);
+		extent_gen = btrfs_file_extent_generation(leaf, extent);
+		comp = btrfs_file_extent_compression(leaf, extent);
+		type = btrfs_file_extent_type(leaf, extent);
+		if (type == BTRFS_FILE_EXTENT_REG ||
+		    type == BTRFS_FILE_EXTENT_PREALLOC) {
+			disko = btrfs_file_extent_disk_bytenr(leaf, extent);
+			diskl = btrfs_file_extent_disk_num_bytes(leaf, extent);
+			datao = btrfs_file_extent_offset(leaf, extent);
+			datal = btrfs_file_extent_num_bytes(leaf, extent);
+		} else if (type == BTRFS_FILE_EXTENT_INLINE) {
+			/* Take upper bound, may be compressed */
+			datal = btrfs_file_extent_ram_bytes(leaf, extent);
+		}
+
+		/*
+		 * The first search might have left us at an extent item that
+		 * ends before our target range's start, can happen if we have
+		 * holes and NO_HOLES feature enabled.
+		 *
+		 * Subsequent searches may leave us on a file range we have
+		 * processed before - this happens due to a race with ordered
+		 * extent completion for a file range that is outside our source
+		 * range, but that range was part of a file extent item that
+		 * also covered a leading part of our source range.
+		 */
+		if (key.offset + datal <= prev_extent_end) {
+			path->slots[0]++;
+			goto process_slot;
+		} else if (key.offset >= off + len) {
+			break;
+		}
+
+		prev_extent_end = key.offset + datal;
+		size = btrfs_item_size(leaf, slot);
+		read_extent_buffer(leaf, buf, btrfs_item_ptr_offset(leaf, slot),
+				   size);
+
+		btrfs_release_path(path);
+
+		memcpy(&new_key, &key, sizeof(new_key));
+		new_key.objectid = btrfs_ino(BTRFS_I(inode));
+		if (off <= key.offset)
+			new_key.offset = key.offset + destoff - off;
+		else
+			new_key.offset = destoff;
+
+		/*
+		 * Deal with a hole that doesn't have an extent item that
+		 * represents it (NO_HOLES feature enabled).
+		 * This hole is either in the middle of the cloning range or at
+		 * the beginning (fully overlaps it or partially overlaps it).
+		 */
+		if (new_key.offset != last_dest_end)
+			drop_start = last_dest_end;
+		else
+			drop_start = new_key.offset;
+
+		if (type == BTRFS_FILE_EXTENT_REG ||
+		    type == BTRFS_FILE_EXTENT_PREALLOC) {
+			struct btrfs_replace_extent_info clone_info;
+
+			/*
+			 *    a  | --- range to clone ---|  b
+			 * | ------------- extent ------------- |
+			 */
+
+			/* Subtract range b */
+			if (key.offset + datal > off + len)
+				datal = off + len - key.offset;
+
+			/* Subtract range a */
+			if (off > key.offset) {
+				datao += off - key.offset;
+				datal -= off - key.offset;
+			}
+
+			clone_info.disk_offset = disko;
+			clone_info.disk_len = diskl;
+			clone_info.data_offset = datao;
+			clone_info.data_len = datal;
+			clone_info.file_offset = new_key.offset;
+			clone_info.extent_buf = buf;
+			clone_info.is_new_extent = false;
+			clone_info.update_times = !no_time_update;
+			ret = btrfs_replace_file_extents(BTRFS_I(inode), path,
+					drop_start, new_key.offset + datal - 1,
+					&clone_info, &trans);
+			if (ret)
+				goto out;
+		} else {
+			ASSERT(type == BTRFS_FILE_EXTENT_INLINE);
+			/*
+			 * Inline extents always have to start at file offset 0
+			 * and can never be bigger then the sector size. We can
+			 * never clone only parts of an inline extent, since all
+			 * reflink operations must start at a sector size aligned
+			 * offset, and the length must be aligned too or end at
+			 * the i_size (which implies the whole inlined data).
+			 */
+			ASSERT(key.offset == 0);
+			ASSERT(datal <= fs_info->sectorsize);
+			if (WARN_ON(type != BTRFS_FILE_EXTENT_INLINE) ||
+			    WARN_ON(key.offset != 0) ||
+			    WARN_ON(datal > fs_info->sectorsize)) {
+				ret = -EUCLEAN;
+				goto out;
+			}
+
+			ret = clone_copy_inline_extent(inode, path, &new_key,
+						       drop_start, datal, size,
+						       comp, buf, &trans);
+			if (ret)
+				goto out;
+		}
+
+		btrfs_release_path(path);
+
+		/*
+		 * Whenever we share an extent we update the last_reflink_trans
+		 * of each inode to the current transaction. This is needed to
+		 * make sure fsync does not log multiple checksum items with
+		 * overlapping ranges (because some extent items might refer
+		 * only to sections of the original extent). For the destination
+		 * inode we do this regardless of the generation of the extents
+		 * or even if they are inline extents or explicit holes, to make
+		 * sure a full fsync does not skip them. For the source inode,
+		 * we only need to update last_reflink_trans in case it's a new
+		 * extent that is not a hole or an inline extent, to deal with
+		 * the checksums problem on fsync.
+		 */
+		if (extent_gen == trans->transid && disko > 0)
+			BTRFS_I(src)->last_reflink_trans = trans->transid;
+
+		BTRFS_I(inode)->last_reflink_trans = trans->transid;
+
+		last_dest_end = ALIGN(new_key.offset + datal,
+				      fs_info->sectorsize);
+		ret = clone_finish_inode_update(trans, inode, last_dest_end,
+						destoff, olen, no_time_update);
+		if (ret)
+			goto out;
+		if (new_key.offset + datal >= destoff + len)
+			break;
+
+		btrfs_release_path(path);
+		key.offset = prev_extent_end;
+
+		if (fatal_signal_pending(current)) {
+			ret = -EINTR;
+			goto out;
+		}
+
+		cond_resched();
+	}
+	ret = 0;
+
+	if (last_dest_end < destoff + len) {
+		/*
+		 * We have an implicit hole that fully or partially overlaps our
+		 * cloning range at its end. This means that we either have the
+		 * NO_HOLES feature enabled or the implicit hole happened due to
+		 * mixing buffered and direct IO writes against this file.
+		 */
+		btrfs_release_path(path);
+
+		/*
+		 * When using NO_HOLES and we are cloning a range that covers
+		 * only a hole (no extents) into a range beyond the current
+		 * i_size, punching a hole in the target range will not create
+		 * an extent map defining a hole, because the range starts at or
+		 * beyond current i_size. If the file previously had an i_size
+		 * greater than the new i_size set by this clone operation, we
+		 * need to make sure the next fsync is a full fsync, so that it
+		 * detects and logs a hole covering a range from the current
+		 * i_size to the new i_size. If the clone range covers extents,
+		 * besides a hole, then we know the full sync flag was already
+		 * set by previous calls to btrfs_replace_file_extents() that
+		 * replaced file extent items.
+		 */
+		if (last_dest_end >= i_size_read(inode))
+			btrfs_set_inode_full_sync(BTRFS_I(inode));
+
+		ret = btrfs_replace_file_extents(BTRFS_I(inode), path,
+				last_dest_end, destoff + len - 1, NULL, &trans);
+		if (ret)
+			goto out;
+
+		ret = clone_finish_inode_update(trans, inode, destoff + len,
+						destoff, olen, no_time_update);
+	}
+
+out:
+	btrfs_free_path(path);
+	kvfree(buf);
+	clear_bit(BTRFS_INODE_NO_DELALLOC_FLUSH, &BTRFS_I(inode)->runtime_flags);
+
+	return ret;
+}
+
+static void btrfs_double_extent_unlock(struct inode *inode1, u64 loff1,
+				       struct inode *inode2, u64 loff2, u64 len)
+{
+	unlock_extent(&BTRFS_I(inode1)->io_tree, loff1, loff1 + len - 1, NULL);
+	unlock_extent(&BTRFS_I(inode2)->io_tree, loff2, loff2 + len - 1, NULL);
+}
+
+static void btrfs_double_extent_lock(struct inode *inode1, u64 loff1,
+				     struct inode *inode2, u64 loff2, u64 len)
+{
+	u64 range1_end = loff1 + len - 1;
+	u64 range2_end = loff2 + len - 1;
+
+	if (inode1 < inode2) {
+		swap(inode1, inode2);
+		swap(loff1, loff2);
+		swap(range1_end, range2_end);
+	} else if (inode1 == inode2 && loff2 < loff1) {
+		swap(loff1, loff2);
+		swap(range1_end, range2_end);
+	}
+
+	lock_extent(&BTRFS_I(inode1)->io_tree, loff1, range1_end, NULL);
+	lock_extent(&BTRFS_I(inode2)->io_tree, loff2, range2_end, NULL);
+
+	btrfs_assert_inode_range_clean(BTRFS_I(inode1), loff1, range1_end);
+	btrfs_assert_inode_range_clean(BTRFS_I(inode2), loff2, range2_end);
+}
+
+static void btrfs_double_mmap_lock(struct inode *inode1, struct inode *inode2)
+{
+	if (inode1 < inode2)
+		swap(inode1, inode2);
+	down_write(&BTRFS_I(inode1)->i_mmap_lock);
+	down_write_nested(&BTRFS_I(inode2)->i_mmap_lock, SINGLE_DEPTH_NESTING);
+}
+
+static void btrfs_double_mmap_unlock(struct inode *inode1, struct inode *inode2)
+{
+	up_write(&BTRFS_I(inode1)->i_mmap_lock);
+	up_write(&BTRFS_I(inode2)->i_mmap_lock);
+}
+
+static int btrfs_extent_same_range(struct inode *src, u64 loff, u64 len,
+				   struct inode *dst, u64 dst_loff)
+{
+	struct btrfs_fs_info *fs_info = BTRFS_I(src)->root->fs_info;
+	const u64 bs = fs_info->sb->s_blocksize;
+	int ret;
+
+	/*
+	 * Lock destination range to serialize with concurrent readahead() and
+	 * source range to serialize with relocation.
+	 */
+	btrfs_double_extent_lock(src, loff, dst, dst_loff, len);
+	ret = btrfs_clone(src, dst, loff, len, ALIGN(len, bs), dst_loff, 1);
+	btrfs_double_extent_unlock(src, loff, dst, dst_loff, len);
+
+	btrfs_btree_balance_dirty(fs_info);
+
+	return ret;
+}
+
+static int btrfs_extent_same(struct inode *src, u64 loff, u64 olen,
+			     struct inode *dst, u64 dst_loff)
+{
+	int ret = 0;
+	u64 i, tail_len, chunk_count;
+	struct btrfs_root *root_dst = BTRFS_I(dst)->root;
+
+	spin_lock(&root_dst->root_item_lock);
+	if (root_dst->send_in_progress) {
+		btrfs_warn_rl(root_dst->fs_info,
+"cannot deduplicate to root %llu while send operations are using it (%d in progress)",
+			      root_dst->root_key.objectid,
+			      root_dst->send_in_progress);
+		spin_unlock(&root_dst->root_item_lock);
+		return -EAGAIN;
+	}
+	root_dst->dedupe_in_progress++;
+	spin_unlock(&root_dst->root_item_lock);
+
+	tail_len = olen % BTRFS_MAX_DEDUPE_LEN;
+	chunk_count = div_u64(olen, BTRFS_MAX_DEDUPE_LEN);
+
+	for (i = 0; i < chunk_count; i++) {
+		ret = btrfs_extent_same_range(src, loff, BTRFS_MAX_DEDUPE_LEN,
+					      dst, dst_loff);
+		if (ret)
+			goto out;
+
+		loff += BTRFS_MAX_DEDUPE_LEN;
+		dst_loff += BTRFS_MAX_DEDUPE_LEN;
+	}
+
+	if (tail_len > 0)
+		ret = btrfs_extent_same_range(src, loff, tail_len, dst, dst_loff);
+out:
+	spin_lock(&root_dst->root_item_lock);
+	root_dst->dedupe_in_progress--;
+	spin_unlock(&root_dst->root_item_lock);
+
+	return ret;
+}
+
+static noinline int btrfs_clone_files(struct file *file, struct file *file_src,
+					u64 off, u64 olen, u64 destoff)
+{
+	struct inode *inode = file_inode(file);
+	struct inode *src = file_inode(file_src);
+	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+	int ret;
+	int wb_ret;
+	u64 len = olen;
+	u64 bs = fs_info->sb->s_blocksize;
+
+	/*
+	 * VFS's generic_remap_file_range_prep() protects us from cloning the
+	 * eof block into the middle of a file, which would result in corruption
+	 * if the file size is not blocksize aligned. So we don't need to check
+	 * for that case here.
+	 */
+	if (off + len == src->i_size)
+		len = ALIGN(src->i_size, bs) - off;
+
+	if (destoff > inode->i_size) {
+		const u64 wb_start = ALIGN_DOWN(inode->i_size, bs);
+
+		ret = btrfs_cont_expand(BTRFS_I(inode), inode->i_size, destoff);
+		if (ret)
+			return ret;
+		/*
+		 * We may have truncated the last block if the inode's size is
+		 * not sector size aligned, so we need to wait for writeback to
+		 * complete before proceeding further, otherwise we can race
+		 * with cloning and attempt to increment a reference to an
+		 * extent that no longer exists (writeback completed right after
+		 * we found the previous extent covering eof and before we
+		 * attempted to increment its reference count).
+		 */
+		ret = btrfs_wait_ordered_range(inode, wb_start,
+					       destoff - wb_start);
+		if (ret)
+			return ret;
+	}
+
+	/*
+	 * Lock destination range to serialize with concurrent readahead() and
+	 * source range to serialize with relocation.
+	 */
+	btrfs_double_extent_lock(src, off, inode, destoff, len);
+	ret = btrfs_clone(src, inode, off, olen, len, destoff, 0);
+	btrfs_double_extent_unlock(src, off, inode, destoff, len);
+
+	/*
+	 * We may have copied an inline extent into a page of the destination
+	 * range, so wait for writeback to complete before truncating pages
+	 * from the page cache. This is a rare case.
+	 */
+	wb_ret = btrfs_wait_ordered_range(inode, destoff, len);
+	ret = ret ? ret : wb_ret;
+	/*
+	 * Truncate page cache pages so that future reads will see the cloned
+	 * data immediately and not the previous data.
+	 */
+	truncate_inode_pages_range(&inode->i_data,
+				round_down(destoff, PAGE_SIZE),
+				round_up(destoff + len, PAGE_SIZE) - 1);
+
+	btrfs_btree_balance_dirty(fs_info);
+
+	return ret;
+}
+
+static int btrfs_remap_file_range_prep(struct file *file_in, loff_t pos_in,
+				       struct file *file_out, loff_t pos_out,
+				       loff_t *len, unsigned int remap_flags)
+{
+	struct inode *inode_in = file_inode(file_in);
+	struct inode *inode_out = file_inode(file_out);
+	u64 bs = BTRFS_I(inode_out)->root->fs_info->sb->s_blocksize;
+	u64 wb_len;
+	int ret;
+
+	if (!(remap_flags & REMAP_FILE_DEDUP)) {
+		struct btrfs_root *root_out = BTRFS_I(inode_out)->root;
+
+		if (btrfs_root_readonly(root_out))
+			return -EROFS;
+
+		ASSERT(inode_in->i_sb == inode_out->i_sb);
+	}
+
+	/* Don't make the dst file partly checksummed */
+	if ((BTRFS_I(inode_in)->flags & BTRFS_INODE_NODATASUM) !=
+	    (BTRFS_I(inode_out)->flags & BTRFS_INODE_NODATASUM)) {
+		return -EINVAL;
+	}
+
+	/*
+	 * Now that the inodes are locked, we need to start writeback ourselves
+	 * and can not rely on the writeback from the VFS's generic helper
+	 * generic_remap_file_range_prep() because:
+	 *
+	 * 1) For compression we must call filemap_fdatawrite_range() range
+	 *    twice (btrfs_fdatawrite_range() does it for us), and the generic
+	 *    helper only calls it once;
+	 *
+	 * 2) filemap_fdatawrite_range(), called by the generic helper only
+	 *    waits for the writeback to complete, i.e. for IO to be done, and
+	 *    not for the ordered extents to complete. We need to wait for them
+	 *    to complete so that new file extent items are in the fs tree.
+	 */
+	if (*len == 0 && !(remap_flags & REMAP_FILE_DEDUP))
+		wb_len = ALIGN(inode_in->i_size, bs) - ALIGN_DOWN(pos_in, bs);
+	else
+		wb_len = ALIGN(*len, bs);
+
+	/*
+	 * Workaround to make sure NOCOW buffered write reach disk as NOCOW.
+	 *
+	 * Btrfs' back references do not have a block level granularity, they
+	 * work at the whole extent level.
+	 * NOCOW buffered write without data space reserved may not be able
+	 * to fall back to CoW due to lack of data space, thus could cause
+	 * data loss.
+	 *
+	 * Here we take a shortcut by flushing the whole inode, so that all
+	 * nocow write should reach disk as nocow before we increase the
+	 * reference of the extent. We could do better by only flushing NOCOW
+	 * data, but that needs extra accounting.
+	 *
+	 * Also we don't need to check ASYNC_EXTENT, as async extent will be
+	 * CoWed anyway, not affecting nocow part.
+	 */
+	ret = filemap_flush(inode_in->i_mapping);
+	if (ret < 0)
+		return ret;
+
+	ret = btrfs_wait_ordered_range(inode_in, ALIGN_DOWN(pos_in, bs),
+				       wb_len);
+	if (ret < 0)
+		return ret;
+	ret = btrfs_wait_ordered_range(inode_out, ALIGN_DOWN(pos_out, bs),
+				       wb_len);
+	if (ret < 0)
+		return ret;
+
+	return generic_remap_file_range_prep(file_in, pos_in, file_out, pos_out,
+					    len, remap_flags);
+}
+
+static bool file_sync_write(const struct file *file)
+{
+	if (file->f_flags & (__O_SYNC | O_DSYNC))
+		return true;
+	if (IS_SYNC(file_inode(file)))
+		return true;
+
+	return false;
+}
+
+loff_t btrfs_remap_file_range(struct file *src_file, loff_t off,
+		struct file *dst_file, loff_t destoff, loff_t len,
+		unsigned int remap_flags)
+{
+	struct inode *src_inode = file_inode(src_file);
+	struct inode *dst_inode = file_inode(dst_file);
+	bool same_inode = dst_inode == src_inode;
+	int ret;
+
+	if (remap_flags & ~(REMAP_FILE_DEDUP | REMAP_FILE_ADVISORY))
+		return -EINVAL;
+
+	if (same_inode) {
+		btrfs_inode_lock(src_inode, BTRFS_ILOCK_MMAP);
+	} else {
+		lock_two_nondirectories(src_inode, dst_inode);
+		btrfs_double_mmap_lock(src_inode, dst_inode);
+	}
+
+	ret = btrfs_remap_file_range_prep(src_file, off, dst_file, destoff,
+					  &len, remap_flags);
+	if (ret < 0 || len == 0)
+		goto out_unlock;
+
+	if (remap_flags & REMAP_FILE_DEDUP)
+		ret = btrfs_extent_same(src_inode, off, len, dst_inode, destoff);
+	else
+		ret = btrfs_clone_files(dst_file, src_file, off, len, destoff);
+
+out_unlock:
+	if (same_inode) {
+		btrfs_inode_unlock(src_inode, BTRFS_ILOCK_MMAP);
+	} else {
+		btrfs_double_mmap_unlock(src_inode, dst_inode);
+		unlock_two_nondirectories(src_inode, dst_inode);
+	}
+
+	/*
+	 * If either the source or the destination file was opened with O_SYNC,
+	 * O_DSYNC or has the S_SYNC attribute, fsync both the destination and
+	 * source files/ranges, so that after a successful return (0) followed
+	 * by a power failure results in the reflinked data to be readable from
+	 * both files/ranges.
+	 */
+	if (ret == 0 && len > 0 &&
+	    (file_sync_write(src_file) || file_sync_write(dst_file))) {
+		ret = btrfs_sync_file(src_file, off, off + len - 1, 0);
+		if (ret == 0)
+			ret = btrfs_sync_file(dst_file, destoff,
+					      destoff + len - 1, 0);
+	}
+
+	return ret < 0 ? ret : len;
+}
diff --git a/fs/btrfs/reflink.h b/fs/btrfs/reflink.h
new file mode 100644
index 000000000000..ecb309b4dad0
--- /dev/null
+++ b/fs/btrfs/reflink.h
@@ -0,0 +1,12 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+
+#ifndef BTRFS_REFLINK_H
+#define BTRFS_REFLINK_H
+
+#include <linux/fs.h>
+
+loff_t btrfs_remap_file_range(struct file *file_in, loff_t pos_in,
+			      struct file *file_out, loff_t pos_out,
+			      loff_t len, unsigned int remap_flags);
+
+#endif /* BTRFS_REFLINK_H */
diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c
index 995d4b8b1cfd..666a37a0ee89 100644
--- a/fs/btrfs/relocation.c
+++ b/fs/btrfs/relocation.c
@@ -9,6 +9,7 @@
 #include <linux/blkdev.h>
 #include <linux/rbtree.h>
 #include <linux/slab.h>
+#include <linux/error-injection.h>
 #include "ctree.h"
 #include "disk-io.h"
 #include "transaction.h"
@@ -17,106 +18,72 @@
 #include "btrfs_inode.h"
 #include "async-thread.h"
 #include "free-space-cache.h"
-#include "inode-map.h"
 #include "qgroup.h"
 #include "print-tree.h"
 #include "delalloc-space.h"
 #include "block-group.h"
+#include "backref.h"
+#include "misc.h"
+#include "subpage.h"
+#include "zoned.h"
+#include "inode-item.h"
 
 /*
- * backref_node, mapping_node and tree_block start with this
- */
-struct tree_entry {
-	struct rb_node rb_node;
-	u64 bytenr;
-};
-
-/*
- * present a tree block in the backref cache
- */
-struct backref_node {
-	struct rb_node rb_node;
-	u64 bytenr;
-
-	u64 new_bytenr;
-	/* objectid of tree block owner, can be not uptodate */
-	u64 owner;
-	/* link to pending, changed or detached list */
-	struct list_head list;
-	/* list of upper level blocks reference this block */
-	struct list_head upper;
-	/* list of child blocks in the cache */
-	struct list_head lower;
-	/* NULL if this node is not tree root */
-	struct btrfs_root *root;
-	/* extent buffer got by COW the block */
-	struct extent_buffer *eb;
-	/* level of tree block */
-	unsigned int level:8;
-	/* is the block in non-reference counted tree */
-	unsigned int cowonly:1;
-	/* 1 if no child node in the cache */
-	unsigned int lowest:1;
-	/* is the extent buffer locked */
-	unsigned int locked:1;
-	/* has the block been processed */
-	unsigned int processed:1;
-	/* have backrefs of this block been checked */
-	unsigned int checked:1;
-	/*
-	 * 1 if corresponding block has been cowed but some upper
-	 * level block pointers may not point to the new location
-	 */
-	unsigned int pending:1;
-	/*
-	 * 1 if the backref node isn't connected to any other
-	 * backref node.
-	 */
-	unsigned int detached:1;
-};
-
-/*
- * present a block pointer in the backref cache
+ * Relocation overview
+ *
+ * [What does relocation do]
+ *
+ * The objective of relocation is to relocate all extents of the target block
+ * group to other block groups.
+ * This is utilized by resize (shrink only), profile converting, compacting
+ * space, or balance routine to spread chunks over devices.
+ *
+ * 		Before		|		After
+ * ------------------------------------------------------------------
+ *  BG A: 10 data extents	| BG A: deleted
+ *  BG B:  2 data extents	| BG B: 10 data extents (2 old + 8 relocated)
+ *  BG C:  1 extents		| BG C:  3 data extents (1 old + 2 relocated)
+ *
+ * [How does relocation work]
+ *
+ * 1.   Mark the target block group read-only
+ *      New extents won't be allocated from the target block group.
+ *
+ * 2.1  Record each extent in the target block group
+ *      To build a proper map of extents to be relocated.
+ *
+ * 2.2  Build data reloc tree and reloc trees
+ *      Data reloc tree will contain an inode, recording all newly relocated
+ *      data extents.
+ *      There will be only one data reloc tree for one data block group.
+ *
+ *      Reloc tree will be a special snapshot of its source tree, containing
+ *      relocated tree blocks.
+ *      Each tree referring to a tree block in target block group will get its
+ *      reloc tree built.
+ *
+ * 2.3  Swap source tree with its corresponding reloc tree
+ *      Each involved tree only refers to new extents after swap.
+ *
+ * 3.   Cleanup reloc trees and data reloc tree.
+ *      As old extents in the target block group are still referenced by reloc
+ *      trees, we need to clean them up before really freeing the target block
+ *      group.
+ *
+ * The main complexity is in steps 2.2 and 2.3.
+ *
+ * The entry point of relocation is relocate_block_group() function.
  */
-struct backref_edge {
-	struct list_head list[2];
-	struct backref_node *node[2];
-};
 
-#define LOWER	0
-#define UPPER	1
 #define RELOCATION_RESERVED_NODES	256
-
-struct backref_cache {
-	/* red black tree of all backref nodes in the cache */
-	struct rb_root rb_root;
-	/* for passing backref nodes to btrfs_reloc_cow_block */
-	struct backref_node *path[BTRFS_MAX_LEVEL];
-	/*
-	 * list of blocks that have been cowed but some block
-	 * pointers in upper level blocks may not reflect the
-	 * new location
-	 */
-	struct list_head pending[BTRFS_MAX_LEVEL];
-	/* list of backref nodes with no child node */
-	struct list_head leaves;
-	/* list of blocks that have been cowed in current transaction */
-	struct list_head changed;
-	/* list of detached backref node. */
-	struct list_head detached;
-
-	u64 last_trans;
-
-	int nr_nodes;
-	int nr_edges;
-};
-
 /*
  * map address of tree root to tree
  */
 struct mapping_node {
-	struct rb_node rb_node;
-	u64 bytenr;
+	struct {
+		struct rb_node rb_node;
+		u64 bytenr;
+	}; /* Use rb_simle_node for search/insert */
 	void *data;
 };
 
@@ -129,8 +96,11 @@ struct mapping_tree {
  * present a tree block to process
  */
 struct tree_block {
-	struct rb_node rb_node;
-	u64 bytenr;
+	struct {
+		struct rb_node rb_node;
+		u64 bytenr;
+	}; /* Use rb_simple_node for search/insert */
+	u64 owner;
 	struct btrfs_key key;
 	unsigned int level:8;
 	unsigned int key_ready:1;
@@ -155,7 +125,7 @@ struct reloc_control {
 
 	struct btrfs_block_rsv *block_rsv;
 
-	struct backref_cache backref_cache;
+	struct btrfs_backref_cache backref_cache;
 
 	struct file_extent_cluster cluster;
 	/* tree blocks have been processed */
@@ -186,167 +156,41 @@ struct reloc_control {
 #define MOVE_DATA_EXTENTS	0
 #define UPDATE_DATA_PTRS	1
 
-static void remove_backref_node(struct backref_cache *cache,
-				struct backref_node *node);
-static void __mark_block_processed(struct reloc_control *rc,
-				   struct backref_node *node);
-
-static void mapping_tree_init(struct mapping_tree *tree)
-{
-	tree->rb_root = RB_ROOT;
-	spin_lock_init(&tree->lock);
-}
-
-static void backref_cache_init(struct backref_cache *cache)
-{
-	int i;
-	cache->rb_root = RB_ROOT;
-	for (i = 0; i < BTRFS_MAX_LEVEL; i++)
-		INIT_LIST_HEAD(&cache->pending[i]);
-	INIT_LIST_HEAD(&cache->changed);
-	INIT_LIST_HEAD(&cache->detached);
-	INIT_LIST_HEAD(&cache->leaves);
-}
-
-static void backref_cache_cleanup(struct backref_cache *cache)
-{
-	struct backref_node *node;
-	int i;
-
-	while (!list_empty(&cache->detached)) {
-		node = list_entry(cache->detached.next,
-				  struct backref_node, list);
-		remove_backref_node(cache, node);
-	}
-
-	while (!list_empty(&cache->leaves)) {
-		node = list_entry(cache->leaves.next,
-				  struct backref_node, lower);
-		remove_backref_node(cache, node);
-	}
-
-	cache->last_trans = 0;
-
-	for (i = 0; i < BTRFS_MAX_LEVEL; i++)
-		ASSERT(list_empty(&cache->pending[i]));
-	ASSERT(list_empty(&cache->changed));
-	ASSERT(list_empty(&cache->detached));
-	ASSERT(RB_EMPTY_ROOT(&cache->rb_root));
-	ASSERT(!cache->nr_nodes);
-	ASSERT(!cache->nr_edges);
-}
-
-static struct backref_node *alloc_backref_node(struct backref_cache *cache)
-{
-	struct backref_node *node;
-
-	node = kzalloc(sizeof(*node), GFP_NOFS);
-	if (node) {
-		INIT_LIST_HEAD(&node->list);
-		INIT_LIST_HEAD(&node->upper);
-		INIT_LIST_HEAD(&node->lower);
-		RB_CLEAR_NODE(&node->rb_node);
-		cache->nr_nodes++;
-	}
-	return node;
-}
-
-static void free_backref_node(struct backref_cache *cache,
-			      struct backref_node *node)
-{
-	if (node) {
-		cache->nr_nodes--;
-		kfree(node);
-	}
-}
-
-static struct backref_edge *alloc_backref_edge(struct backref_cache *cache)
-{
-	struct backref_edge *edge;
-
-	edge = kzalloc(sizeof(*edge), GFP_NOFS);
-	if (edge)
-		cache->nr_edges++;
-	return edge;
-}
-
-static void free_backref_edge(struct backref_cache *cache,
-			      struct backref_edge *edge)
+static void mark_block_processed(struct reloc_control *rc,
+				 struct btrfs_backref_node *node)
 {
-	if (edge) {
-		cache->nr_edges--;
-		kfree(edge);
-	}
-}
+	u32 blocksize;
 
-static struct rb_node *tree_insert(struct rb_root *root, u64 bytenr,
-				   struct rb_node *node)
-{
-	struct rb_node **p = &root->rb_node;
-	struct rb_node *parent = NULL;
-	struct tree_entry *entry;
-
-	while (*p) {
-		parent = *p;
-		entry = rb_entry(parent, struct tree_entry, rb_node);
-
-		if (bytenr < entry->bytenr)
-			p = &(*p)->rb_left;
-		else if (bytenr > entry->bytenr)
-			p = &(*p)->rb_right;
-		else
-			return parent;
+	if (node->level == 0 ||
+	    in_range(node->bytenr, rc->block_group->start,
+		     rc->block_group->length)) {
+		blocksize = rc->extent_root->fs_info->nodesize;
+		set_extent_bits(&rc->processed_blocks, node->bytenr,
+				node->bytenr + blocksize - 1, EXTENT_DIRTY);
 	}
-
-	rb_link_node(node, parent, p);
-	rb_insert_color(node, root);
-	return NULL;
+	node->processed = 1;
 }
 
-static struct rb_node *tree_search(struct rb_root *root, u64 bytenr)
-{
-	struct rb_node *n = root->rb_node;
-	struct tree_entry *entry;
-
-	while (n) {
-		entry = rb_entry(n, struct tree_entry, rb_node);
-
-		if (bytenr < entry->bytenr)
-			n = n->rb_left;
-		else if (bytenr > entry->bytenr)
-			n = n->rb_right;
-		else
-			return n;
-	}
-	return NULL;
-}
 
-static void backref_tree_panic(struct rb_node *rb_node, int errno, u64 bytenr)
+static void mapping_tree_init(struct mapping_tree *tree)
 {
-
-	struct btrfs_fs_info *fs_info = NULL;
-	struct backref_node *bnode = rb_entry(rb_node, struct backref_node,
-					      rb_node);
-	if (bnode->root)
-		fs_info = bnode->root->fs_info;
-	btrfs_panic(fs_info, errno,
-		    "Inconsistency in backref cache found at offset %llu",
-		    bytenr);
+	tree->rb_root = RB_ROOT;
+	spin_lock_init(&tree->lock);
 }
 
 /*
  * walk up backref nodes until reach node presents tree root
  */
-static struct backref_node *walk_up_backref(struct backref_node *node,
-					    struct backref_edge *edges[],
-					    int *index)
+static struct btrfs_backref_node *walk_up_backref(
+		struct btrfs_backref_node *node,
+		struct btrfs_backref_edge *edges[], int *index)
 {
-	struct backref_edge *edge;
+	struct btrfs_backref_edge *edge;
 	int idx = *index;
 
 	while (!list_empty(&node->upper)) {
 		edge = list_entry(node->upper.next,
-				  struct backref_edge, list[LOWER]);
+				  struct btrfs_backref_edge, list[LOWER]);
 		edges[idx++] = edge;
 		node = edge->node[UPPER];
 	}
@@ -358,11 +202,11 @@ static struct backref_node *walk_up_backref(struct backref_node *node,
 /*
  * walk down backref nodes to find start of next reference path
  */
-static struct backref_node *walk_down_backref(struct backref_edge *edges[],
-					      int *index)
+static struct btrfs_backref_node *walk_down_backref(
+		struct btrfs_backref_edge *edges[], int *index)
 {
-	struct backref_edge *edge;
-	struct backref_node *lower;
+	struct btrfs_backref_edge *edge;
+	struct btrfs_backref_node *lower;
 	int idx = *index;
 
 	while (idx > 0) {
@@ -373,7 +217,7 @@ static struct backref_node *walk_down_backref(struct backref_edge *edges[],
 			continue;
 		}
 		edge = list_entry(edge->list[LOWER].next,
-				  struct backref_edge, list[LOWER]);
+				  struct btrfs_backref_edge, list[LOWER]);
 		edges[idx - 1] = edge;
 		*index = idx;
 		return edge->node[UPPER];
@@ -382,95 +226,24 @@ static struct backref_node *walk_down_backref(struct backref_edge *edges[],
 	return NULL;
 }
 
-static void unlock_node_buffer(struct backref_node *node)
-{
-	if (node->locked) {
-		btrfs_tree_unlock(node->eb);
-		node->locked = 0;
-	}
-}
-
-static void drop_node_buffer(struct backref_node *node)
-{
-	if (node->eb) {
-		unlock_node_buffer(node);
-		free_extent_buffer(node->eb);
-		node->eb = NULL;
-	}
-}
-
-static void drop_backref_node(struct backref_cache *tree,
-			      struct backref_node *node)
-{
-	BUG_ON(!list_empty(&node->upper));
-
-	drop_node_buffer(node);
-	list_del(&node->list);
-	list_del(&node->lower);
-	if (!RB_EMPTY_NODE(&node->rb_node))
-		rb_erase(&node->rb_node, &tree->rb_root);
-	free_backref_node(tree, node);
-}
-
-/*
- * remove a backref node from the backref cache
- */
-static void remove_backref_node(struct backref_cache *cache,
-				struct backref_node *node)
-{
-	struct backref_node *upper;
-	struct backref_edge *edge;
-
-	if (!node)
-		return;
-
-	BUG_ON(!node->lowest && !node->detached);
-	while (!list_empty(&node->upper)) {
-		edge = list_entry(node->upper.next, struct backref_edge,
-				  list[LOWER]);
-		upper = edge->node[UPPER];
-		list_del(&edge->list[LOWER]);
-		list_del(&edge->list[UPPER]);
-		free_backref_edge(cache, edge);
-
-		if (RB_EMPTY_NODE(&upper->rb_node)) {
-			BUG_ON(!list_empty(&node->upper));
-			drop_backref_node(cache, node);
-			node = upper;
-			node->lowest = 1;
-			continue;
-		}
-		/*
-		 * add the node to leaf node list if no other
-		 * child block cached.
-		 */
-		if (list_empty(&upper->lower)) {
-			list_add_tail(&upper->lower, &cache->leaves);
-			upper->lowest = 1;
-		}
-	}
-
-	drop_backref_node(cache, node);
-}
-
-static void update_backref_node(struct backref_cache *cache,
-				struct backref_node *node, u64 bytenr)
+static void update_backref_node(struct btrfs_backref_cache *cache,
+				struct btrfs_backref_node *node, u64 bytenr)
 {
 	struct rb_node *rb_node;
 	rb_erase(&node->rb_node, &cache->rb_root);
 	node->bytenr = bytenr;
-	rb_node = tree_insert(&cache->rb_root, node->bytenr, &node->rb_node);
+	rb_node = rb_simple_insert(&cache->rb_root, node->bytenr, &node->rb_node);
 	if (rb_node)
-		backref_tree_panic(rb_node, -EEXIST, bytenr);
+		btrfs_backref_panic(cache->fs_info, bytenr, -EEXIST);
 }
 
 /*
  * update backref cache after a transaction commit
  */
 static int update_backref_cache(struct btrfs_trans_handle *trans,
-				struct backref_cache *cache)
+				struct btrfs_backref_cache *cache)
 {
-	struct backref_node *node;
+	struct btrfs_backref_node *node;
 	int level = 0;
 
 	if (cache->last_trans == 0) {
@@ -488,13 +261,13 @@ static int update_backref_cache(struct btrfs_trans_handle *trans,
 	 */
 	while (!list_empty(&cache->detached)) {
 		node = list_entry(cache->detached.next,
-				  struct backref_node, list);
-		remove_backref_node(cache, node);
+				  struct btrfs_backref_node, list);
+		btrfs_backref_cleanup_node(cache, node);
 	}
 
 	while (!list_empty(&cache->changed)) {
 		node = list_entry(cache->changed.next,
-				  struct backref_node, list);
+				  struct btrfs_backref_node, list);
 		list_del_init(&node->list);
 		BUG_ON(node->pending);
 		update_backref_node(cache, node, node->new_bytenr);
@@ -535,7 +308,8 @@ static bool reloc_root_is_dead(struct btrfs_root *root)
  *
  * Reloc tree after swap is considered dead, thus not considered as valid.
  * This is enough for most callers, as they don't distinguish dead reloc root
- * from no reloc root.  But should_ignore_root() below is a special case.
+ * from no reloc root.  But btrfs_should_ignore_reloc_root() below is a
+ * special case.
  */
 static bool have_reloc_root(struct btrfs_root *root)
 {
@@ -546,11 +320,11 @@ static bool have_reloc_root(struct btrfs_root *root)
 	return true;
 }
 
-static int should_ignore_root(struct btrfs_root *root)
+int btrfs_should_ignore_reloc_root(struct btrfs_root *root)
 {
 	struct btrfs_root *reloc_root;
 
-	if (!test_bit(BTRFS_ROOT_REF_COWS, &root->state))
+	if (!test_bit(BTRFS_ROOT_SHAREABLE, &root->state))
 		return 0;
 
 	/* This root has been merged with its reloc tree, we can ignore it */
@@ -561,8 +335,8 @@ static int should_ignore_root(struct btrfs_root *root)
 	if (!reloc_root)
 		return 0;
 
-	if (btrfs_root_last_snapshot(&reloc_root->root_item) ==
-	    root->fs_info->running_transaction->transid - 1)
+	if (btrfs_header_generation(reloc_root->commit_root) ==
+	    root->fs_info->running_transaction->transid)
 		return 0;
 	/*
 	 * if there is reloc tree and it was created in previous
@@ -572,624 +346,187 @@ static int should_ignore_root(struct btrfs_root *root)
 	 */
 	return 1;
 }
+
 /*
  * find reloc tree by address of tree root
  */
-static struct btrfs_root *find_reloc_root(struct reloc_control *rc,
-					  u64 bytenr)
+struct btrfs_root *find_reloc_root(struct btrfs_fs_info *fs_info, u64 bytenr)
 {
+	struct reloc_control *rc = fs_info->reloc_ctl;
 	struct rb_node *rb_node;
 	struct mapping_node *node;
 	struct btrfs_root *root = NULL;
 
+	ASSERT(rc);
 	spin_lock(&rc->reloc_root_tree.lock);
-	rb_node = tree_search(&rc->reloc_root_tree.rb_root, bytenr);
+	rb_node = rb_simple_search(&rc->reloc_root_tree.rb_root, bytenr);
 	if (rb_node) {
 		node = rb_entry(rb_node, struct mapping_node, rb_node);
-		root = (struct btrfs_root *)node->data;
+		root = node->data;
 	}
 	spin_unlock(&rc->reloc_root_tree.lock);
-	return root;
+	return btrfs_grab_root(root);
 }
 
-static int is_cowonly_root(u64 root_objectid)
+/*
+ * For useless nodes, do two major clean ups:
+ *
+ * - Cleanup the children edges and nodes
+ *   If child node is also orphan (no parent) during cleanup, then the child
+ *   node will also be cleaned up.
+ *
+ * - Freeing up leaves (level 0), keeps nodes detached
+ *   For nodes, the node is still cached as "detached"
+ *
+ * Return false if @node is not in the @useless_nodes list.
+ * Return true if @node is in the @useless_nodes list.
+ */
+static bool handle_useless_nodes(struct reloc_control *rc,
+				 struct btrfs_backref_node *node)
 {
-	if (root_objectid == BTRFS_ROOT_TREE_OBJECTID ||
-	    root_objectid == BTRFS_EXTENT_TREE_OBJECTID ||
-	    root_objectid == BTRFS_CHUNK_TREE_OBJECTID ||
-	    root_objectid == BTRFS_DEV_TREE_OBJECTID ||
-	    root_objectid == BTRFS_TREE_LOG_OBJECTID ||
-	    root_objectid == BTRFS_CSUM_TREE_OBJECTID ||
-	    root_objectid == BTRFS_UUID_TREE_OBJECTID ||
-	    root_objectid == BTRFS_QUOTA_TREE_OBJECTID ||
-	    root_objectid == BTRFS_FREE_SPACE_TREE_OBJECTID)
-		return 1;
-	return 0;
-}
+	struct btrfs_backref_cache *cache = &rc->backref_cache;
+	struct list_head *useless_node = &cache->useless_node;
+	bool ret = false;
 
-static struct btrfs_root *read_fs_root(struct btrfs_fs_info *fs_info,
-					u64 root_objectid)
-{
-	struct btrfs_key key;
+	while (!list_empty(useless_node)) {
+		struct btrfs_backref_node *cur;
 
-	key.objectid = root_objectid;
-	key.type = BTRFS_ROOT_ITEM_KEY;
-	if (is_cowonly_root(root_objectid))
-		key.offset = 0;
-	else
-		key.offset = (u64)-1;
+		cur = list_first_entry(useless_node, struct btrfs_backref_node,
+				 list);
+		list_del_init(&cur->list);
 
-	return btrfs_get_fs_root(fs_info, &key, false);
-}
+		/* Only tree root nodes can be added to @useless_nodes */
+		ASSERT(list_empty(&cur->upper));
 
-static noinline_for_stack
-int find_inline_backref(struct extent_buffer *leaf, int slot,
-			unsigned long *ptr, unsigned long *end)
-{
-	struct btrfs_key key;
-	struct btrfs_extent_item *ei;
-	struct btrfs_tree_block_info *bi;
-	u32 item_size;
+		if (cur == node)
+			ret = true;
 
-	btrfs_item_key_to_cpu(leaf, &key, slot);
+		/* The node is the lowest node */
+		if (cur->lowest) {
+			list_del_init(&cur->lower);
+			cur->lowest = 0;
+		}
 
-	item_size = btrfs_item_size_nr(leaf, slot);
-	if (item_size < sizeof(*ei)) {
-		btrfs_print_v0_err(leaf->fs_info);
-		btrfs_handle_fs_error(leaf->fs_info, -EINVAL, NULL);
-		return 1;
-	}
-	ei = btrfs_item_ptr(leaf, slot, struct btrfs_extent_item);
-	WARN_ON(!(btrfs_extent_flags(leaf, ei) &
-		  BTRFS_EXTENT_FLAG_TREE_BLOCK));
+		/* Cleanup the lower edges */
+		while (!list_empty(&cur->lower)) {
+			struct btrfs_backref_edge *edge;
+			struct btrfs_backref_node *lower;
 
-	if (key.type == BTRFS_EXTENT_ITEM_KEY &&
-	    item_size <= sizeof(*ei) + sizeof(*bi)) {
-		WARN_ON(item_size < sizeof(*ei) + sizeof(*bi));
-		return 1;
-	}
-	if (key.type == BTRFS_METADATA_ITEM_KEY &&
-	    item_size <= sizeof(*ei)) {
-		WARN_ON(item_size < sizeof(*ei));
-		return 1;
-	}
+			edge = list_entry(cur->lower.next,
+					struct btrfs_backref_edge, list[UPPER]);
+			list_del(&edge->list[UPPER]);
+			list_del(&edge->list[LOWER]);
+			lower = edge->node[LOWER];
+			btrfs_backref_free_edge(cache, edge);
 
-	if (key.type == BTRFS_EXTENT_ITEM_KEY) {
-		bi = (struct btrfs_tree_block_info *)(ei + 1);
-		*ptr = (unsigned long)(bi + 1);
-	} else {
-		*ptr = (unsigned long)(ei + 1);
+			/* Child node is also orphan, queue for cleanup */
+			if (list_empty(&lower->upper))
+				list_add(&lower->list, useless_node);
+		}
+		/* Mark this block processed for relocation */
+		mark_block_processed(rc, cur);
+
+		/*
+		 * Backref nodes for tree leaves are deleted from the cache.
+		 * Backref nodes for upper level tree blocks are left in the
+		 * cache to avoid unnecessary backref lookup.
+		 */
+		if (cur->level > 0) {
+			list_add(&cur->list, &cache->detached);
+			cur->detached = 1;
+		} else {
+			rb_erase(&cur->rb_node, &cache->rb_root);
+			btrfs_backref_free_node(cache, cur);
+		}
 	}
-	*end = (unsigned long)ei + item_size;
-	return 0;
+	return ret;
 }
 
 /*
- * build backref tree for a given tree block. root of the backref tree
- * corresponds the tree block, leaves of the backref tree correspond
- * roots of b-trees that reference the tree block.
+ * Build backref tree for a given tree block. Root of the backref tree
+ * corresponds the tree block, leaves of the backref tree correspond roots of
+ * b-trees that reference the tree block.
  *
- * the basic idea of this function is check backrefs of a given block
- * to find upper level blocks that reference the block, and then check
- * backrefs of these upper level blocks recursively. the recursion stop
- * when tree root is reached or backrefs for the block is cached.
+ * The basic idea of this function is check backrefs of a given block to find
+ * upper level blocks that reference the block, and then check backrefs of
+ * these upper level blocks recursively. The recursion stops when tree root is
+ * reached or backrefs for the block is cached.
  *
- * NOTE: if we find backrefs for a block are cached, we know backrefs
- * for all upper level blocks that directly/indirectly reference the
- * block are also cached.
+ * NOTE: if we find that backrefs for a block are cached, we know backrefs for
+ * all upper level blocks that directly/indirectly reference the block are also
+ * cached.
  */
-static noinline_for_stack
-struct backref_node *build_backref_tree(struct reloc_control *rc,
-					struct btrfs_key *node_key,
-					int level, u64 bytenr)
+static noinline_for_stack struct btrfs_backref_node *build_backref_tree(
+			struct reloc_control *rc, struct btrfs_key *node_key,
+			int level, u64 bytenr)
 {
-	struct backref_cache *cache = &rc->backref_cache;
-	struct btrfs_path *path1; /* For searching extent root */
-	struct btrfs_path *path2; /* For searching parent of TREE_BLOCK_REF */
-	struct extent_buffer *eb;
-	struct btrfs_root *root;
-	struct backref_node *cur;
-	struct backref_node *upper;
-	struct backref_node *lower;
-	struct backref_node *node = NULL;
-	struct backref_node *exist = NULL;
-	struct backref_edge *edge;
-	struct rb_node *rb_node;
-	struct btrfs_key key;
-	unsigned long end;
-	unsigned long ptr;
-	LIST_HEAD(list); /* Pending edge list, upper node needs to be checked */
-	LIST_HEAD(useless);
-	int cowonly;
+	struct btrfs_backref_iter *iter;
+	struct btrfs_backref_cache *cache = &rc->backref_cache;
+	/* For searching parent of TREE_BLOCK_REF */
+	struct btrfs_path *path;
+	struct btrfs_backref_node *cur;
+	struct btrfs_backref_node *node = NULL;
+	struct btrfs_backref_edge *edge;
 	int ret;
 	int err = 0;
-	bool need_check = true;
 
-	path1 = btrfs_alloc_path();
-	path2 = btrfs_alloc_path();
-	if (!path1 || !path2) {
+	iter = btrfs_backref_iter_alloc(rc->extent_root->fs_info, GFP_NOFS);
+	if (!iter)
+		return ERR_PTR(-ENOMEM);
+	path = btrfs_alloc_path();
+	if (!path) {
 		err = -ENOMEM;
 		goto out;
 	}
-	path1->reada = READA_FORWARD;
-	path2->reada = READA_FORWARD;
 
-	node = alloc_backref_node(cache);
+	node = btrfs_backref_alloc_node(cache, bytenr, level);
 	if (!node) {
 		err = -ENOMEM;
 		goto out;
 	}
 
-	node->bytenr = bytenr;
-	node->level = level;
 	node->lowest = 1;
 	cur = node;
-again:
-	end = 0;
-	ptr = 0;
-	key.objectid = cur->bytenr;
-	key.type = BTRFS_METADATA_ITEM_KEY;
-	key.offset = (u64)-1;
-
-	path1->search_commit_root = 1;
-	path1->skip_locking = 1;
-	ret = btrfs_search_slot(NULL, rc->extent_root, &key, path1,
-				0, 0);
-	if (ret < 0) {
-		err = ret;
-		goto out;
-	}
-	ASSERT(ret);
-	ASSERT(path1->slots[0]);
-
-	path1->slots[0]--;
-
-	WARN_ON(cur->checked);
-	if (!list_empty(&cur->upper)) {
-		/*
-		 * the backref was added previously when processing
-		 * backref of type BTRFS_TREE_BLOCK_REF_KEY
-		 */
-		ASSERT(list_is_singular(&cur->upper));
-		edge = list_entry(cur->upper.next, struct backref_edge,
-				  list[LOWER]);
-		ASSERT(list_empty(&edge->list[UPPER]));
-		exist = edge->node[UPPER];
-		/*
-		 * add the upper level block to pending list if we need
-		 * check its backrefs
-		 */
-		if (!exist->checked)
-			list_add_tail(&edge->list[UPPER], &list);
-	} else {
-		exist = NULL;
-	}
-
-	while (1) {
-		cond_resched();
-		eb = path1->nodes[0];
-
-		if (ptr >= end) {
-			if (path1->slots[0] >= btrfs_header_nritems(eb)) {
-				ret = btrfs_next_leaf(rc->extent_root, path1);
-				if (ret < 0) {
-					err = ret;
-					goto out;
-				}
-				if (ret > 0)
-					break;
-				eb = path1->nodes[0];
-			}
-
-			btrfs_item_key_to_cpu(eb, &key, path1->slots[0]);
-			if (key.objectid != cur->bytenr) {
-				WARN_ON(exist);
-				break;
-			}
-
-			if (key.type == BTRFS_EXTENT_ITEM_KEY ||
-			    key.type == BTRFS_METADATA_ITEM_KEY) {
-				ret = find_inline_backref(eb, path1->slots[0],
-							  &ptr, &end);
-				if (ret)
-					goto next;
-			}
-		}
-
-		if (ptr < end) {
-			/* update key for inline back ref */
-			struct btrfs_extent_inline_ref *iref;
-			int type;
-			iref = (struct btrfs_extent_inline_ref *)ptr;
-			type = btrfs_get_extent_inline_ref_type(eb, iref,
-							BTRFS_REF_TYPE_BLOCK);
-			if (type == BTRFS_REF_TYPE_INVALID) {
-				err = -EUCLEAN;
-				goto out;
-			}
-			key.type = type;
-			key.offset = btrfs_extent_inline_ref_offset(eb, iref);
-
-			WARN_ON(key.type != BTRFS_TREE_BLOCK_REF_KEY &&
-				key.type != BTRFS_SHARED_BLOCK_REF_KEY);
-		}
-
-		/*
-		 * Parent node found and matches current inline ref, no need to
-		 * rebuild this node for this inline ref.
-		 */
-		if (exist &&
-		    ((key.type == BTRFS_TREE_BLOCK_REF_KEY &&
-		      exist->owner == key.offset) ||
-		     (key.type == BTRFS_SHARED_BLOCK_REF_KEY &&
-		      exist->bytenr == key.offset))) {
-			exist = NULL;
-			goto next;
-		}
-
-		/* SHARED_BLOCK_REF means key.offset is the parent bytenr */
-		if (key.type == BTRFS_SHARED_BLOCK_REF_KEY) {
-			if (key.objectid == key.offset) {
-				/*
-				 * Only root blocks of reloc trees use backref
-				 * pointing to itself.
-				 */
-				root = find_reloc_root(rc, cur->bytenr);
-				ASSERT(root);
-				cur->root = root;
-				break;
-			}
-
-			edge = alloc_backref_edge(cache);
-			if (!edge) {
-				err = -ENOMEM;
-				goto out;
-			}
-			rb_node = tree_search(&cache->rb_root, key.offset);
-			if (!rb_node) {
-				upper = alloc_backref_node(cache);
-				if (!upper) {
-					free_backref_edge(cache, edge);
-					err = -ENOMEM;
-					goto out;
-				}
-				upper->bytenr = key.offset;
-				upper->level = cur->level + 1;
-				/*
-				 *  backrefs for the upper level block isn't
-				 *  cached, add the block to pending list
-				 */
-				list_add_tail(&edge->list[UPPER], &list);
-			} else {
-				upper = rb_entry(rb_node, struct backref_node,
-						 rb_node);
-				ASSERT(upper->checked);
-				INIT_LIST_HEAD(&edge->list[UPPER]);
-			}
-			list_add_tail(&edge->list[LOWER], &cur->upper);
-			edge->node[LOWER] = cur;
-			edge->node[UPPER] = upper;
-
-			goto next;
-		} else if (unlikely(key.type == BTRFS_EXTENT_REF_V0_KEY)) {
-			err = -EINVAL;
-			btrfs_print_v0_err(rc->extent_root->fs_info);
-			btrfs_handle_fs_error(rc->extent_root->fs_info, err,
-					      NULL);
-			goto out;
-		} else if (key.type != BTRFS_TREE_BLOCK_REF_KEY) {
-			goto next;
-		}
-
-		/*
-		 * key.type == BTRFS_TREE_BLOCK_REF_KEY, inline ref offset
-		 * means the root objectid. We need to search the tree to get
-		 * its parent bytenr.
-		 */
-		root = read_fs_root(rc->extent_root->fs_info, key.offset);
-		if (IS_ERR(root)) {
-			err = PTR_ERR(root);
-			goto out;
-		}
-
-		if (!test_bit(BTRFS_ROOT_REF_COWS, &root->state))
-			cur->cowonly = 1;
-
-		if (btrfs_root_level(&root->root_item) == cur->level) {
-			/* tree root */
-			ASSERT(btrfs_root_bytenr(&root->root_item) ==
-			       cur->bytenr);
-			if (should_ignore_root(root))
-				list_add(&cur->list, &useless);
-			else
-				cur->root = root;
-			break;
-		}
-
-		level = cur->level + 1;
 
-		/* Search the tree to find parent blocks referring the block. */
-		path2->search_commit_root = 1;
-		path2->skip_locking = 1;
-		path2->lowest_level = level;
-		ret = btrfs_search_slot(NULL, root, node_key, path2, 0, 0);
-		path2->lowest_level = 0;
+	/* Breadth-first search to build backref cache */
+	do {
+		ret = btrfs_backref_add_tree_node(cache, path, iter, node_key,
+						  cur);
 		if (ret < 0) {
 			err = ret;
 			goto out;
 		}
-		if (ret > 0 && path2->slots[level] > 0)
-			path2->slots[level]--;
-
-		eb = path2->nodes[level];
-		if (btrfs_node_blockptr(eb, path2->slots[level]) !=
-		    cur->bytenr) {
-			btrfs_err(root->fs_info,
-	"couldn't find block (%llu) (level %d) in tree (%llu) with key (%llu %u %llu)",
-				  cur->bytenr, level - 1,
-				  root->root_key.objectid,
-				  node_key->objectid, node_key->type,
-				  node_key->offset);
-			err = -ENOENT;
-			goto out;
-		}
-		lower = cur;
-		need_check = true;
-
-		/* Add all nodes and edges in the path */
-		for (; level < BTRFS_MAX_LEVEL; level++) {
-			if (!path2->nodes[level]) {
-				ASSERT(btrfs_root_bytenr(&root->root_item) ==
-				       lower->bytenr);
-				if (should_ignore_root(root))
-					list_add(&lower->list, &useless);
-				else
-					lower->root = root;
-				break;
-			}
-
-			edge = alloc_backref_edge(cache);
-			if (!edge) {
-				err = -ENOMEM;
-				goto out;
-			}
-
-			eb = path2->nodes[level];
-			rb_node = tree_search(&cache->rb_root, eb->start);
-			if (!rb_node) {
-				upper = alloc_backref_node(cache);
-				if (!upper) {
-					free_backref_edge(cache, edge);
-					err = -ENOMEM;
-					goto out;
-				}
-				upper->bytenr = eb->start;
-				upper->owner = btrfs_header_owner(eb);
-				upper->level = lower->level + 1;
-				if (!test_bit(BTRFS_ROOT_REF_COWS,
-					      &root->state))
-					upper->cowonly = 1;
-
-				/*
-				 * if we know the block isn't shared
-				 * we can void checking its backrefs.
-				 */
-				if (btrfs_block_can_be_shared(root, eb))
-					upper->checked = 0;
-				else
-					upper->checked = 1;
-
-				/*
-				 * add the block to pending list if we
-				 * need check its backrefs, we only do this once
-				 * while walking up a tree as we will catch
-				 * anything else later on.
-				 */
-				if (!upper->checked && need_check) {
-					need_check = false;
-					list_add_tail(&edge->list[UPPER],
-						      &list);
-				} else {
-					if (upper->checked)
-						need_check = true;
-					INIT_LIST_HEAD(&edge->list[UPPER]);
-				}
-			} else {
-				upper = rb_entry(rb_node, struct backref_node,
-						 rb_node);
-				ASSERT(upper->checked);
-				INIT_LIST_HEAD(&edge->list[UPPER]);
-				if (!upper->owner)
-					upper->owner = btrfs_header_owner(eb);
-			}
-			list_add_tail(&edge->list[LOWER], &lower->upper);
-			edge->node[LOWER] = lower;
-			edge->node[UPPER] = upper;
-
-			if (rb_node)
-				break;
-			lower = upper;
-			upper = NULL;
-		}
-		btrfs_release_path(path2);
-next:
-		if (ptr < end) {
-			ptr += btrfs_extent_inline_ref_size(key.type);
-			if (ptr >= end) {
-				WARN_ON(ptr > end);
-				ptr = 0;
-				end = 0;
-			}
+		edge = list_first_entry_or_null(&cache->pending_edge,
+				struct btrfs_backref_edge, list[UPPER]);
+		/*
+		 * The pending list isn't empty, take the first block to
+		 * process
+		 */
+		if (edge) {
+			list_del_init(&edge->list[UPPER]);
+			cur = edge->node[UPPER];
 		}
-		if (ptr >= end)
-			path1->slots[0]++;
-	}
-	btrfs_release_path(path1);
-
-	cur->checked = 1;
-	WARN_ON(exist);
-
-	/* the pending list isn't empty, take the first block to process */
-	if (!list_empty(&list)) {
-		edge = list_entry(list.next, struct backref_edge, list[UPPER]);
-		list_del_init(&edge->list[UPPER]);
-		cur = edge->node[UPPER];
-		goto again;
-	}
+	} while (edge);
 
-	/*
-	 * everything goes well, connect backref nodes and insert backref nodes
-	 * into the cache.
-	 */
-	ASSERT(node->checked);
-	cowonly = node->cowonly;
-	if (!cowonly) {
-		rb_node = tree_insert(&cache->rb_root, node->bytenr,
-				      &node->rb_node);
-		if (rb_node)
-			backref_tree_panic(rb_node, -EEXIST, node->bytenr);
-		list_add_tail(&node->lower, &cache->leaves);
+	/* Finish the upper linkage of newly added edges/nodes */
+	ret = btrfs_backref_finish_upper_links(cache, node);
+	if (ret < 0) {
+		err = ret;
+		goto out;
 	}
 
-	list_for_each_entry(edge, &node->upper, list[LOWER])
-		list_add_tail(&edge->list[UPPER], &list);
-
-	while (!list_empty(&list)) {
-		edge = list_entry(list.next, struct backref_edge, list[UPPER]);
-		list_del_init(&edge->list[UPPER]);
-		upper = edge->node[UPPER];
-		if (upper->detached) {
-			list_del(&edge->list[LOWER]);
-			lower = edge->node[LOWER];
-			free_backref_edge(cache, edge);
-			if (list_empty(&lower->upper))
-				list_add(&lower->list, &useless);
-			continue;
-		}
-
-		if (!RB_EMPTY_NODE(&upper->rb_node)) {
-			if (upper->lowest) {
-				list_del_init(&upper->lower);
-				upper->lowest = 0;
-			}
-
-			list_add_tail(&edge->list[UPPER], &upper->lower);
-			continue;
-		}
-
-		if (!upper->checked) {
-			/*
-			 * Still want to blow up for developers since this is a
-			 * logic bug.
-			 */
-			ASSERT(0);
-			err = -EINVAL;
-			goto out;
-		}
-		if (cowonly != upper->cowonly) {
-			ASSERT(0);
-			err = -EINVAL;
-			goto out;
-		}
-
-		if (!cowonly) {
-			rb_node = tree_insert(&cache->rb_root, upper->bytenr,
-					      &upper->rb_node);
-			if (rb_node)
-				backref_tree_panic(rb_node, -EEXIST,
-						   upper->bytenr);
-		}
-
-		list_add_tail(&edge->list[UPPER], &upper->lower);
-
-		list_for_each_entry(edge, &upper->upper, list[LOWER])
-			list_add_tail(&edge->list[UPPER], &list);
-	}
-	/*
-	 * process useless backref nodes. backref nodes for tree leaves
-	 * are deleted from the cache. backref nodes for upper level
-	 * tree blocks are left in the cache to avoid unnecessary backref
-	 * lookup.
-	 */
-	while (!list_empty(&useless)) {
-		upper = list_entry(useless.next, struct backref_node, list);
-		list_del_init(&upper->list);
-		ASSERT(list_empty(&upper->upper));
-		if (upper == node)
-			node = NULL;
-		if (upper->lowest) {
-			list_del_init(&upper->lower);
-			upper->lowest = 0;
-		}
-		while (!list_empty(&upper->lower)) {
-			edge = list_entry(upper->lower.next,
-					  struct backref_edge, list[UPPER]);
-			list_del(&edge->list[UPPER]);
-			list_del(&edge->list[LOWER]);
-			lower = edge->node[LOWER];
-			free_backref_edge(cache, edge);
-
-			if (list_empty(&lower->upper))
-				list_add(&lower->list, &useless);
-		}
-		__mark_block_processed(rc, upper);
-		if (upper->level > 0) {
-			list_add(&upper->list, &cache->detached);
-			upper->detached = 1;
-		} else {
-			rb_erase(&upper->rb_node, &cache->rb_root);
-			free_backref_node(cache, upper);
-		}
-	}
+	if (handle_useless_nodes(rc, node))
+		node = NULL;
 out:
-	btrfs_free_path(path1);
-	btrfs_free_path(path2);
+	btrfs_backref_iter_free(iter);
+	btrfs_free_path(path);
 	if (err) {
-		while (!list_empty(&useless)) {
-			lower = list_entry(useless.next,
-					   struct backref_node, list);
-			list_del_init(&lower->list);
-		}
-		while (!list_empty(&list)) {
-			edge = list_first_entry(&list, struct backref_edge,
-						list[UPPER]);
-			list_del(&edge->list[UPPER]);
-			list_del(&edge->list[LOWER]);
-			lower = edge->node[LOWER];
-			upper = edge->node[UPPER];
-			free_backref_edge(cache, edge);
-
-			/*
-			 * Lower is no longer linked to any upper backref nodes
-			 * and isn't in the cache, we can free it ourselves.
-			 */
-			if (list_empty(&lower->upper) &&
-			    RB_EMPTY_NODE(&lower->rb_node))
-				list_add(&lower->list, &useless);
-
-			if (!RB_EMPTY_NODE(&upper->rb_node))
-				continue;
-
-			/* Add this guy's upper edges to the list to process */
-			list_for_each_entry(edge, &upper->upper, list[LOWER])
-				list_add_tail(&edge->list[UPPER], &list);
-			if (list_empty(&upper->upper))
-				list_add(&upper->list, &useless);
-		}
-
-		while (!list_empty(&useless)) {
-			lower = list_entry(useless.next,
-					   struct backref_node, list);
-			list_del_init(&lower->list);
-			if (lower == node)
-				node = NULL;
-			free_backref_node(cache, lower);
-		}
-
-		free_backref_node(cache, node);
+		btrfs_backref_error_cleanup(cache, node);
 		return ERR_PTR(err);
 	}
 	ASSERT(!node || !node->detached);
+	ASSERT(list_empty(&cache->useless_node) &&
+	       list_empty(&cache->pending_edge));
 	return node;
 }
 
@@ -1204,19 +541,19 @@ static int clone_backref_node(struct btrfs_trans_handle *trans,
 			      struct btrfs_root *dest)
 {
 	struct btrfs_root *reloc_root = src->reloc_root;
-	struct backref_cache *cache = &rc->backref_cache;
-	struct backref_node *node = NULL;
-	struct backref_node *new_node;
-	struct backref_edge *edge;
-	struct backref_edge *new_edge;
+	struct btrfs_backref_cache *cache = &rc->backref_cache;
+	struct btrfs_backref_node *node = NULL;
+	struct btrfs_backref_node *new_node;
+	struct btrfs_backref_edge *edge;
+	struct btrfs_backref_edge *new_edge;
 	struct rb_node *rb_node;
 
 	if (cache->last_trans > 0)
 		update_backref_cache(trans, cache);
 
-	rb_node = tree_search(&cache->rb_root, src->commit_root->start);
+	rb_node = rb_simple_search(&cache->rb_root, src->commit_root->start);
 	if (rb_node) {
-		node = rb_entry(rb_node, struct backref_node, rb_node);
+		node = rb_entry(rb_node, struct btrfs_backref_node, rb_node);
 		if (node->detached)
 			node = NULL;
 		else
@@ -1224,10 +561,10 @@ static int clone_backref_node(struct btrfs_trans_handle *trans,
 	}
 
 	if (!node) {
-		rb_node = tree_search(&cache->rb_root,
-				      reloc_root->commit_root->start);
+		rb_node = rb_simple_search(&cache->rb_root,
+					   reloc_root->commit_root->start);
 		if (rb_node) {
-			node = rb_entry(rb_node, struct backref_node,
+			node = rb_entry(rb_node, struct btrfs_backref_node,
 					rb_node);
 			BUG_ON(node->detached);
 		}
@@ -1236,35 +573,33 @@ static int clone_backref_node(struct btrfs_trans_handle *trans,
 	if (!node)
 		return 0;
 
-	new_node = alloc_backref_node(cache);
+	new_node = btrfs_backref_alloc_node(cache, dest->node->start,
+					    node->level);
 	if (!new_node)
 		return -ENOMEM;
 
-	new_node->bytenr = dest->node->start;
-	new_node->level = node->level;
 	new_node->lowest = node->lowest;
 	new_node->checked = 1;
-	new_node->root = dest;
+	new_node->root = btrfs_grab_root(dest);
+	ASSERT(new_node->root);
 
 	if (!node->lowest) {
 		list_for_each_entry(edge, &node->lower, list[UPPER]) {
-			new_edge = alloc_backref_edge(cache);
+			new_edge = btrfs_backref_alloc_edge(cache);
 			if (!new_edge)
 				goto fail;
 
-			new_edge->node[UPPER] = new_node;
-			new_edge->node[LOWER] = edge->node[LOWER];
-			list_add_tail(&new_edge->list[UPPER],
-				      &new_node->lower);
+			btrfs_backref_link_edge(new_edge, edge->node[LOWER],
+						new_node, LINK_UPPER);
 		}
 	} else {
 		list_add_tail(&new_node->lower, &cache->leaves);
 	}
 
-	rb_node = tree_insert(&cache->rb_root, new_node->bytenr,
-			      &new_node->rb_node);
+	rb_node = rb_simple_insert(&cache->rb_root, new_node->bytenr,
+				   &new_node->rb_node);
 	if (rb_node)
-		backref_tree_panic(rb_node, -EEXIST, new_node->bytenr);
+		btrfs_backref_panic(trans->fs_info, new_node->bytenr, -EEXIST);
 
 	if (!new_node->lowest) {
 		list_for_each_entry(new_edge, &new_node->lower, list[UPPER]) {
@@ -1276,11 +611,11 @@ static int clone_backref_node(struct btrfs_trans_handle *trans,
 fail:
 	while (!list_empty(&new_node->lower)) {
 		new_edge = list_entry(new_node->lower.next,
-				      struct backref_edge, list[UPPER]);
+				      struct btrfs_backref_edge, list[UPPER]);
 		list_del(&new_edge->list[UPPER]);
-		free_backref_edge(cache, new_edge);
+		btrfs_backref_free_edge(cache, new_edge);
 	}
-	free_backref_node(cache, new_node);
+	btrfs_backref_free_node(cache, new_node);
 	return -ENOMEM;
 }
 
@@ -1298,17 +633,18 @@ static int __must_check __add_reloc_root(struct btrfs_root *root)
 	if (!node)
 		return -ENOMEM;
 
-	node->bytenr = root->node->start;
+	node->bytenr = root->commit_root->start;
 	node->data = root;
 
 	spin_lock(&rc->reloc_root_tree.lock);
-	rb_node = tree_insert(&rc->reloc_root_tree.rb_root,
-			      node->bytenr, &node->rb_node);
+	rb_node = rb_simple_insert(&rc->reloc_root_tree.rb_root,
+				   node->bytenr, &node->rb_node);
 	spin_unlock(&rc->reloc_root_tree.lock);
 	if (rb_node) {
-		btrfs_panic(fs_info, -EEXIST,
+		btrfs_err(fs_info,
 			    "Duplicate root found for start=%llu while inserting into relocation tree",
 			    node->bytenr);
+		return -EEXIST;
 	}
 
 	list_add_tail(&root->root_list, &rc->reloc_roots);
@@ -1325,24 +661,37 @@ static void __del_reloc_root(struct btrfs_root *root)
 	struct rb_node *rb_node;
 	struct mapping_node *node = NULL;
 	struct reloc_control *rc = fs_info->reloc_ctl;
+	bool put_ref = false;
 
 	if (rc && root->node) {
 		spin_lock(&rc->reloc_root_tree.lock);
-		rb_node = tree_search(&rc->reloc_root_tree.rb_root,
-				      root->node->start);
+		rb_node = rb_simple_search(&rc->reloc_root_tree.rb_root,
+					   root->commit_root->start);
 		if (rb_node) {
 			node = rb_entry(rb_node, struct mapping_node, rb_node);
 			rb_erase(&node->rb_node, &rc->reloc_root_tree.rb_root);
+			RB_CLEAR_NODE(&node->rb_node);
 		}
 		spin_unlock(&rc->reloc_root_tree.lock);
-		if (!node)
-			return;
-		BUG_ON((struct btrfs_root *)node->data != root);
+		ASSERT(!node || (struct btrfs_root *)node->data == root);
 	}
 
+	/*
+	 * We only put the reloc root here if it's on the list.  There's a lot
+	 * of places where the pattern is to splice the rc->reloc_roots, process
+	 * the reloc roots, and then add the reloc root back onto
+	 * rc->reloc_roots.  If we call __del_reloc_root while it's off of the
+	 * list we don't want the reference being dropped, because the guy
+	 * messing with the list is in charge of the reference.
+	 */
 	spin_lock(&fs_info->trans_lock);
-	list_del_init(&root->root_list);
+	if (!list_empty(&root->root_list)) {
+		put_ref = true;
+		list_del_init(&root->root_list);
+	}
 	spin_unlock(&fs_info->trans_lock);
+	if (put_ref)
+		btrfs_put_root(root);
 	kfree(node);
 }
 
@@ -1350,7 +699,7 @@ static void __del_reloc_root(struct btrfs_root *root)
  * helper to update the 'address of tree root -> reloc tree'
  * mapping
  */
-static int __update_reloc_root(struct btrfs_root *root, u64 new_bytenr)
+static int __update_reloc_root(struct btrfs_root *root)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct rb_node *rb_node;
@@ -1358,8 +707,8 @@ static int __update_reloc_root(struct btrfs_root *root, u64 new_bytenr)
 	struct reloc_control *rc = fs_info->reloc_ctl;
 
 	spin_lock(&rc->reloc_root_tree.lock);
-	rb_node = tree_search(&rc->reloc_root_tree.rb_root,
-			      root->node->start);
+	rb_node = rb_simple_search(&rc->reloc_root_tree.rb_root,
+				   root->commit_root->start);
 	if (rb_node) {
 		node = rb_entry(rb_node, struct mapping_node, rb_node);
 		rb_erase(&node->rb_node, &rc->reloc_root_tree.rb_root);
@@ -1371,12 +720,12 @@ static int __update_reloc_root(struct btrfs_root *root, u64 new_bytenr)
 	BUG_ON((struct btrfs_root *)node->data != root);
 
 	spin_lock(&rc->reloc_root_tree.lock);
-	node->bytenr = new_bytenr;
-	rb_node = tree_insert(&rc->reloc_root_tree.rb_root,
-			      node->bytenr, &node->rb_node);
+	node->bytenr = root->node->start;
+	rb_node = rb_simple_insert(&rc->reloc_root_tree.rb_root,
+				   node->bytenr, &node->rb_node);
 	spin_unlock(&rc->reloc_root_tree.lock);
 	if (rb_node)
-		backref_tree_panic(rb_node, -EEXIST, node->bytenr);
+		btrfs_backref_panic(fs_info, node->bytenr, -EEXIST);
 	return 0;
 }
 
@@ -1388,10 +737,12 @@ static struct btrfs_root *create_reloc_root(struct btrfs_trans_handle *trans,
 	struct extent_buffer *eb;
 	struct btrfs_root_item *root_item;
 	struct btrfs_key root_key;
-	int ret;
+	int ret = 0;
+	bool must_abort = false;
 
 	root_item = kmalloc(sizeof(*root_item), GFP_NOFS);
-	BUG_ON(!root_item);
+	if (!root_item)
+		return ERR_PTR(-ENOMEM);
 
 	root_key.objectid = BTRFS_TREE_RELOC_OBJECTID;
 	root_key.type = BTRFS_ROOT_ITEM_KEY;
@@ -1403,7 +754,9 @@ static struct btrfs_root *create_reloc_root(struct btrfs_trans_handle *trans,
 		/* called by btrfs_init_reloc_root */
 		ret = btrfs_copy_root(trans, root, root->commit_root, &eb,
 				      BTRFS_TREE_RELOC_OBJECTID);
-		BUG_ON(ret);
+		if (ret)
+			goto fail;
+
 		/*
 		 * Set the last_snapshot field to the generation of the commit
 		 * root - like this ctree.c:btrfs_block_can_be_shared() behaves
@@ -1424,9 +777,16 @@ static struct btrfs_root *create_reloc_root(struct btrfs_trans_handle *trans,
 		 */
 		ret = btrfs_copy_root(trans, root, root->node, &eb,
 				      BTRFS_TREE_RELOC_OBJECTID);
-		BUG_ON(ret);
+		if (ret)
+			goto fail;
 	}
 
+	/*
+	 * We have changed references at this point, we must abort the
+	 * transaction if anything fails.
+	 */
+	must_abort = true;
+
 	memcpy(root_item, &root->root_item, sizeof(*root_item));
 	btrfs_set_root_bytenr(root_item, eb->start);
 	btrfs_set_root_level(root_item, btrfs_header_level(eb));
@@ -1436,7 +796,7 @@ static struct btrfs_root *create_reloc_root(struct btrfs_trans_handle *trans,
 		btrfs_set_root_refs(root_item, 0);
 		memset(&root_item->drop_progress, 0,
 		       sizeof(struct btrfs_disk_key));
-		root_item->drop_level = 0;
+		btrfs_set_root_drop_level(root_item, 0);
 	}
 
 	btrfs_tree_unlock(eb);
@@ -1444,18 +804,33 @@ static struct btrfs_root *create_reloc_root(struct btrfs_trans_handle *trans,
 
 	ret = btrfs_insert_root(trans, fs_info->tree_root,
 				&root_key, root_item);
-	BUG_ON(ret);
+	if (ret)
+		goto fail;
+
 	kfree(root_item);
 
-	reloc_root = btrfs_read_fs_root(fs_info->tree_root, &root_key);
-	BUG_ON(IS_ERR(reloc_root));
+	reloc_root = btrfs_read_tree_root(fs_info->tree_root, &root_key);
+	if (IS_ERR(reloc_root)) {
+		ret = PTR_ERR(reloc_root);
+		goto abort;
+	}
+	set_bit(BTRFS_ROOT_SHAREABLE, &reloc_root->state);
 	reloc_root->last_trans = trans->transid;
 	return reloc_root;
+fail:
+	kfree(root_item);
+abort:
+	if (must_abort)
+		btrfs_abort_transaction(trans, ret);
+	return ERR_PTR(ret);
 }
 
 /*
  * create reloc tree for a given fs tree. reloc tree is just a
  * snapshot of the fs tree with special root objectid.
+ *
+ * The reloc_root comes out of here with two references, one for
+ * root->reloc_root, and another for being on the rc->reloc_roots list.
  */
 int btrfs_init_reloc_root(struct btrfs_trans_handle *trans,
 			  struct btrfs_root *root)
@@ -1467,6 +842,9 @@ int btrfs_init_reloc_root(struct btrfs_trans_handle *trans,
 	int clear_rsv = 0;
 	int ret;
 
+	if (!rc)
+		return 0;
+
 	/*
 	 * The subvolume has reloc tree but the swap is finished, no need to
 	 * create/update the dead reloc tree
@@ -1474,13 +852,25 @@ int btrfs_init_reloc_root(struct btrfs_trans_handle *trans,
 	if (reloc_root_is_dead(root))
 		return 0;
 
+	/*
+	 * This is subtle but important.  We do not do
+	 * record_root_in_transaction for reloc roots, instead we record their
+	 * corresponding fs root, and then here we update the last trans for the
+	 * reloc root.  This means that we have to do this for the entire life
+	 * of the reloc root, regardless of which stage of the relocation we are
+	 * in.
+	 */
 	if (root->reloc_root) {
 		reloc_root = root->reloc_root;
 		reloc_root->last_trans = trans->transid;
 		return 0;
 	}
 
-	if (!rc || !rc->create_reloc_tree ||
+	/*
+	 * We are merging reloc roots, we do not need new reloc trees.  Also
+	 * reloc trees never need their own reloc tree.
+	 */
+	if (!rc->create_reloc_tree ||
 	    root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID)
 		return 0;
 
@@ -1492,10 +882,17 @@ int btrfs_init_reloc_root(struct btrfs_trans_handle *trans,
 	reloc_root = create_reloc_root(trans, root, root->root_key.objectid);
 	if (clear_rsv)
 		trans->block_rsv = rsv;
+	if (IS_ERR(reloc_root))
+		return PTR_ERR(reloc_root);
 
 	ret = __add_reloc_root(reloc_root);
-	BUG_ON(ret < 0);
-	root->reloc_root = reloc_root;
+	ASSERT(ret != -EEXIST);
+	if (ret) {
+		/* Pairs with create_reloc_root */
+		btrfs_put_root(reloc_root);
+		return ret;
+	}
+	root->reloc_root = btrfs_grab_root(reloc_root);
 	return 0;
 }
 
@@ -1511,11 +908,18 @@ int btrfs_update_reloc_root(struct btrfs_trans_handle *trans,
 	int ret;
 
 	if (!have_reloc_root(root))
-		goto out;
+		return 0;
 
 	reloc_root = root->reloc_root;
 	root_item = &reloc_root->root_item;
 
+	/*
+	 * We are probably ok here, but __del_reloc_root() will drop its ref of
+	 * the root.  We have the ref for root->reloc_root, but just in case
+	 * hold it while we update the reloc root.
+	 */
+	btrfs_grab_root(reloc_root);
+
 	/* root->reloc_root will stay until current relocation finished */
 	if (fs_info->reloc_ctl->merge_reloc_tree &&
 	    btrfs_root_refs(root_item) == 0) {
@@ -1529,6 +933,7 @@ int btrfs_update_reloc_root(struct btrfs_trans_handle *trans,
 	}
 
 	if (reloc_root->commit_root != reloc_root->node) {
+		__update_reloc_root(reloc_root);
 		btrfs_set_root_node(root_item, reloc_root->node);
 		free_extent_buffer(reloc_root->commit_root);
 		reloc_root->commit_root = btrfs_root_node(reloc_root);
@@ -1536,10 +941,8 @@ int btrfs_update_reloc_root(struct btrfs_trans_handle *trans,
 
 	ret = btrfs_update_root(trans, fs_info->tree_root,
 				&reloc_root->root_key, root_item);
-	BUG_ON(ret);
-
-out:
-	return 0;
+	btrfs_put_root(reloc_root);
+	return ret;
 }
 
 /*
@@ -1596,14 +999,6 @@ again:
 	return NULL;
 }
 
-static int in_block_group(u64 bytenr, struct btrfs_block_group *block_group)
-{
-	if (bytenr >= block_group->start &&
-	    bytenr < block_group->start + block_group->length)
-		return 1;
-	return 0;
-}
-
 /*
  * get new location of data
  */
@@ -1701,11 +1096,12 @@ int replace_file_extents(struct btrfs_trans_handle *trans,
 		num_bytes = btrfs_file_extent_disk_num_bytes(leaf, fi);
 		if (bytenr == 0)
 			continue;
-		if (!in_block_group(bytenr, rc->block_group))
+		if (!in_range(bytenr, rc->block_group->start,
+			      rc->block_group->length))
 			continue;
 
 		/*
-		 * if we are modifying block in fs tree, wait for readpage
+		 * if we are modifying block in fs tree, wait for read_folio
 		 * to complete and drop the extent cache
 		 */
 		if (root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID) {
@@ -1728,10 +1124,10 @@ int replace_file_extents(struct btrfs_trans_handle *trans,
 				if (!ret)
 					continue;
 
-				btrfs_drop_extent_cache(BTRFS_I(inode),
-						key.offset,	end, 1);
+				btrfs_drop_extent_map_range(BTRFS_I(inode),
+							    key.offset, end, true);
 				unlock_extent(&BTRFS_I(inode)->io_tree,
-					      key.offset, end);
+					      key.offset, end, NULL);
 			}
 		}
 
@@ -1751,9 +1147,9 @@ int replace_file_extents(struct btrfs_trans_handle *trans,
 		key.offset -= btrfs_file_extent_offset(leaf, fi);
 		btrfs_init_generic_ref(&ref, BTRFS_ADD_DELAYED_REF, new_bytenr,
 				       num_bytes, parent);
-		ref.real_root = root->root_key.objectid;
 		btrfs_init_data_ref(&ref, btrfs_header_owner(leaf),
-				    key.objectid, key.offset);
+				    key.objectid, key.offset,
+				    root->root_key.objectid, false);
 		ret = btrfs_inc_extent_ref(trans, &ref);
 		if (ret) {
 			btrfs_abort_transaction(trans, ret);
@@ -1762,9 +1158,9 @@ int replace_file_extents(struct btrfs_trans_handle *trans,
 
 		btrfs_init_generic_ref(&ref, BTRFS_DROP_DELAYED_REF, bytenr,
 				       num_bytes, parent);
-		ref.real_root = root->root_key.objectid;
 		btrfs_init_data_ref(&ref, btrfs_header_owner(leaf),
-				    key.objectid, key.offset);
+				    key.objectid, key.offset,
+				    root->root_key.objectid, false);
 		ret = btrfs_free_extent(trans, &ref);
 		if (ret) {
 			btrfs_abort_transaction(trans, ret);
@@ -1820,8 +1216,8 @@ int replace_path(struct btrfs_trans_handle *trans, struct reloc_control *rc,
 	int ret;
 	int slot;
 
-	BUG_ON(src->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID);
-	BUG_ON(dest->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID);
+	ASSERT(src->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID);
+	ASSERT(dest->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID);
 
 	last_snapshot = btrfs_root_last_snapshot(&src->root_item);
 again:
@@ -1829,7 +1225,6 @@ again:
 	btrfs_node_key_to_cpu(path->nodes[lowest_level], &key, slot);
 
 	eb = btrfs_lock_root_node(dest);
-	btrfs_set_lock_blocking_write(eb);
 	level = btrfs_header_level(eb);
 
 	if (level < lowest_level) {
@@ -1839,10 +1234,14 @@ again:
 	}
 
 	if (cow) {
-		ret = btrfs_cow_block(trans, dest, eb, NULL, 0, &eb);
-		BUG_ON(ret);
+		ret = btrfs_cow_block(trans, dest, eb, NULL, 0, &eb,
+				      BTRFS_NESTING_COW);
+		if (ret) {
+			btrfs_tree_unlock(eb);
+			free_extent_buffer(eb);
+			return ret;
+		}
 	}
-	btrfs_set_lock_blocking_write(eb);
 
 	if (next_key) {
 		next_key->objectid = (u64)-1;
@@ -1852,12 +1251,10 @@ again:
 
 	parent = eb;
 	while (1) {
-		struct btrfs_key first_key;
-
 		level = btrfs_header_level(parent);
-		BUG_ON(level < lowest_level);
+		ASSERT(level >= lowest_level);
 
-		ret = btrfs_bin_search(parent, &key, level, &slot);
+		ret = btrfs_bin_search(parent, &key, &slot);
 		if (ret < 0)
 			break;
 		if (ret && slot > 0)
@@ -1869,7 +1266,6 @@ again:
 		old_bytenr = btrfs_node_blockptr(parent, slot);
 		blocksize = fs_info->nodesize;
 		old_ptr_gen = btrfs_node_ptr_generation(parent, slot);
-		btrfs_node_key_to_cpu(parent, &first_key, slot);
 
 		if (level <= max_level) {
 			eb = path->nodes[level];
@@ -1894,23 +1290,22 @@ again:
 				break;
 			}
 
-			eb = read_tree_block(fs_info, old_bytenr, old_ptr_gen,
-					     level - 1, &first_key);
+			eb = btrfs_read_node_slot(parent, slot);
 			if (IS_ERR(eb)) {
 				ret = PTR_ERR(eb);
 				break;
-			} else if (!extent_buffer_uptodate(eb)) {
-				ret = -EIO;
-				free_extent_buffer(eb);
-				break;
 			}
 			btrfs_tree_lock(eb);
 			if (cow) {
 				ret = btrfs_cow_block(trans, dest, eb, parent,
-						      slot, &eb);
-				BUG_ON(ret);
+						      slot, &eb,
+						      BTRFS_NESTING_COW);
+				if (ret) {
+					btrfs_tree_unlock(eb);
+					free_extent_buffer(eb);
+					break;
+				}
 			}
-			btrfs_set_lock_blocking_write(eb);
 
 			btrfs_tree_unlock(parent);
 			free_extent_buffer(parent);
@@ -1931,9 +1326,15 @@ again:
 		btrfs_release_path(path);
 
 		path->lowest_level = level;
+		set_bit(BTRFS_ROOT_RESET_LOCKDEP_CLASS, &src->state);
 		ret = btrfs_search_slot(trans, src, &key, path, 0, 1);
+		clear_bit(BTRFS_ROOT_RESET_LOCKDEP_CLASS, &src->state);
 		path->lowest_level = 0;
-		BUG_ON(ret);
+		if (ret) {
+			if (ret > 0)
+				ret = -ENOENT;
+			break;
+		}
 
 		/*
 		 * Info qgroup to trace both subtrees.
@@ -1970,30 +1371,42 @@ again:
 
 		btrfs_init_generic_ref(&ref, BTRFS_ADD_DELAYED_REF, old_bytenr,
 				       blocksize, path->nodes[level]->start);
-		ref.skip_qgroup = true;
-		btrfs_init_tree_ref(&ref, level - 1, src->root_key.objectid);
+		btrfs_init_tree_ref(&ref, level - 1, src->root_key.objectid,
+				    0, true);
 		ret = btrfs_inc_extent_ref(trans, &ref);
-		BUG_ON(ret);
+		if (ret) {
+			btrfs_abort_transaction(trans, ret);
+			break;
+		}
 		btrfs_init_generic_ref(&ref, BTRFS_ADD_DELAYED_REF, new_bytenr,
 				       blocksize, 0);
-		ref.skip_qgroup = true;
-		btrfs_init_tree_ref(&ref, level - 1, dest->root_key.objectid);
+		btrfs_init_tree_ref(&ref, level - 1, dest->root_key.objectid, 0,
+				    true);
 		ret = btrfs_inc_extent_ref(trans, &ref);
-		BUG_ON(ret);
+		if (ret) {
+			btrfs_abort_transaction(trans, ret);
+			break;
+		}
 
 		btrfs_init_generic_ref(&ref, BTRFS_DROP_DELAYED_REF, new_bytenr,
 				       blocksize, path->nodes[level]->start);
-		btrfs_init_tree_ref(&ref, level - 1, src->root_key.objectid);
-		ref.skip_qgroup = true;
+		btrfs_init_tree_ref(&ref, level - 1, src->root_key.objectid,
+				    0, true);
 		ret = btrfs_free_extent(trans, &ref);
-		BUG_ON(ret);
+		if (ret) {
+			btrfs_abort_transaction(trans, ret);
+			break;
+		}
 
 		btrfs_init_generic_ref(&ref, BTRFS_DROP_DELAYED_REF, old_bytenr,
 				       blocksize, 0);
-		btrfs_init_tree_ref(&ref, level - 1, dest->root_key.objectid);
-		ref.skip_qgroup = true;
+		btrfs_init_tree_ref(&ref, level - 1, dest->root_key.objectid,
+				    0, true);
 		ret = btrfs_free_extent(trans, &ref);
-		BUG_ON(ret);
+		if (ret) {
+			btrfs_abort_transaction(trans, ret);
+			break;
+		}
 
 		btrfs_unlock_up_safe(path, 0);
 
@@ -2049,10 +1462,8 @@ static noinline_for_stack
 int walk_down_reloc_tree(struct btrfs_root *root, struct btrfs_path *path,
 			 int *level)
 {
-	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct extent_buffer *eb = NULL;
 	int i;
-	u64 bytenr;
 	u64 ptr_gen = 0;
 	u64 last_snapshot;
 	u32 nritems;
@@ -2060,8 +1471,6 @@ int walk_down_reloc_tree(struct btrfs_root *root, struct btrfs_path *path,
 	last_snapshot = btrfs_root_last_snapshot(&root->root_item);
 
 	for (i = *level; i > 0; i--) {
-		struct btrfs_key first_key;
-
 		eb = path->nodes[i];
 		nritems = btrfs_header_nritems(eb);
 		while (path->slots[i] < nritems) {
@@ -2081,16 +1490,9 @@ int walk_down_reloc_tree(struct btrfs_root *root, struct btrfs_path *path,
 			return 0;
 		}
 
-		bytenr = btrfs_node_blockptr(eb, path->slots[i]);
-		btrfs_node_key_to_cpu(eb, &first_key, path->slots[i]);
-		eb = read_tree_block(fs_info, bytenr, ptr_gen, i - 1,
-				     &first_key);
-		if (IS_ERR(eb)) {
+		eb = btrfs_read_node_slot(eb, path->slots[i]);
+		if (IS_ERR(eb))
 			return PTR_ERR(eb);
-		} else if (!extent_buffer_uptodate(eb)) {
-			free_extent_buffer(eb);
-			return -EIO;
-		}
 		BUG_ON(btrfs_header_level(eb) != i - 1);
 		path->nodes[i - 1] = eb;
 		path->slots[i - 1] = 0;
@@ -2163,10 +1565,10 @@ static int invalidate_extent_cache(struct btrfs_root *root,
 			end = (u64)-1;
 		}
 
-		/* the lock_extent waits for readpage to complete */
-		lock_extent(&BTRFS_I(inode)->io_tree, start, end);
-		btrfs_drop_extent_cache(BTRFS_I(inode), start, end, 1);
-		unlock_extent(&BTRFS_I(inode)->io_tree, start, end);
+		/* the lock_extent waits for read_folio to complete */
+		lock_extent(&BTRFS_I(inode)->io_tree, start, end, NULL);
+		btrfs_drop_extent_map_range(BTRFS_I(inode), start, end, true);
+		unlock_extent(&BTRFS_I(inode)->io_tree, start, end, NULL);
 	}
 	return 0;
 }
@@ -2192,12 +1594,13 @@ static int find_next_key(struct btrfs_path *path, int level,
 /*
  * Insert current subvolume into reloc_control::dirty_subvol_roots
  */
-static void insert_dirty_subvol(struct btrfs_trans_handle *trans,
-				struct reloc_control *rc,
-				struct btrfs_root *root)
+static int insert_dirty_subvol(struct btrfs_trans_handle *trans,
+			       struct reloc_control *rc,
+			       struct btrfs_root *root)
 {
 	struct btrfs_root *reloc_root = root->reloc_root;
 	struct btrfs_root_item *reloc_root_item;
+	int ret;
 
 	/* @root must be a subvolume tree root with a valid reloc tree */
 	ASSERT(root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID);
@@ -2206,14 +1609,18 @@ static void insert_dirty_subvol(struct btrfs_trans_handle *trans,
 	reloc_root_item = &reloc_root->root_item;
 	memset(&reloc_root_item->drop_progress, 0,
 		sizeof(reloc_root_item->drop_progress));
-	reloc_root_item->drop_level = 0;
+	btrfs_set_root_drop_level(reloc_root_item, 0);
 	btrfs_set_root_refs(reloc_root_item, 0);
-	btrfs_update_reloc_root(trans, root);
+	ret = btrfs_update_reloc_root(trans, root);
+	if (ret)
+		return ret;
 
 	if (list_empty(&root->reloc_dirty_list)) {
-		btrfs_grab_fs_root(root);
+		btrfs_grab_root(root);
 		list_add_tail(&root->reloc_dirty_list, &rc->dirty_subvol_roots);
 	}
+
+	return 0;
 }
 
 static int clean_dirty_subvols(struct reloc_control *rc)
@@ -2231,24 +1638,34 @@ static int clean_dirty_subvols(struct reloc_control *rc)
 
 			list_del_init(&root->reloc_dirty_list);
 			root->reloc_root = NULL;
-			if (reloc_root) {
-
-				ret2 = btrfs_drop_snapshot(reloc_root, NULL, 0, 1);
-				if (ret2 < 0 && !ret)
-					ret = ret2;
-			}
 			/*
 			 * Need barrier to ensure clear_bit() only happens after
 			 * root->reloc_root = NULL. Pairs with have_reloc_root.
 			 */
 			smp_wmb();
 			clear_bit(BTRFS_ROOT_DEAD_RELOC_TREE, &root->state);
-			btrfs_put_fs_root(root);
+			if (reloc_root) {
+				/*
+				 * btrfs_drop_snapshot drops our ref we hold for
+				 * ->reloc_root.  If it fails however we must
+				 * drop the ref ourselves.
+				 */
+				ret2 = btrfs_drop_snapshot(reloc_root, 0, 1);
+				if (ret2 < 0) {
+					btrfs_put_root(reloc_root);
+					if (!ret)
+						ret = ret2;
+				}
+			}
+			btrfs_put_root(root);
 		} else {
 			/* Orphan reloc tree, just clean it up */
-			ret2 = btrfs_drop_snapshot(root, NULL, 0, 1);
-			if (ret2 < 0 && !ret)
-				ret = ret2;
+			ret2 = btrfs_drop_snapshot(root, 0, 1);
+			if (ret2 < 0) {
+				btrfs_put_root(root);
+				if (!ret)
+					ret = ret2;
+			}
 		}
 	}
 	return ret;
@@ -2269,11 +1686,11 @@ static noinline_for_stack int merge_reloc_root(struct reloc_control *rc,
 	struct btrfs_root_item *root_item;
 	struct btrfs_path *path;
 	struct extent_buffer *leaf;
+	int reserve_level;
 	int level;
 	int max_level;
 	int replaced = 0;
-	int ret;
-	int err = 0;
+	int ret = 0;
 	u32 min_reserved;
 
 	path = btrfs_alloc_path();
@@ -2292,7 +1709,7 @@ static noinline_for_stack int merge_reloc_root(struct reloc_control *rc,
 	} else {
 		btrfs_disk_key_to_cpu(&key, &root_item->drop_progress);
 
-		level = root_item->drop_level;
+		level = btrfs_root_drop_level(root_item);
 		BUG_ON(level == 0);
 		path->lowest_level = level;
 		ret = btrfs_search_slot(NULL, reloc_root, &key, path, 0, 0);
@@ -2309,32 +1726,50 @@ static noinline_for_stack int merge_reloc_root(struct reloc_control *rc,
 		btrfs_unlock_up_safe(path, 0);
 	}
 
-	min_reserved = fs_info->nodesize * (BTRFS_MAX_LEVEL - 1) * 2;
+	/*
+	 * In merge_reloc_root(), we modify the upper level pointer to swap the
+	 * tree blocks between reloc tree and subvolume tree.  Thus for tree
+	 * block COW, we COW at most from level 1 to root level for each tree.
+	 *
+	 * Thus the needed metadata size is at most root_level * nodesize,
+	 * and * 2 since we have two trees to COW.
+	 */
+	reserve_level = max_t(int, 1, btrfs_root_level(root_item));
+	min_reserved = fs_info->nodesize * reserve_level * 2;
 	memset(&next_key, 0, sizeof(next_key));
 
 	while (1) {
-		ret = btrfs_block_rsv_refill(root, rc->block_rsv, min_reserved,
-					     BTRFS_RESERVE_FLUSH_ALL);
-		if (ret) {
-			err = ret;
+		ret = btrfs_block_rsv_refill(fs_info, rc->block_rsv,
+					     min_reserved,
+					     BTRFS_RESERVE_FLUSH_LIMIT);
+		if (ret)
 			goto out;
-		}
 		trans = btrfs_start_transaction(root, 0);
 		if (IS_ERR(trans)) {
-			err = PTR_ERR(trans);
+			ret = PTR_ERR(trans);
 			trans = NULL;
 			goto out;
 		}
+
+		/*
+		 * At this point we no longer have a reloc_control, so we can't
+		 * depend on btrfs_init_reloc_root to update our last_trans.
+		 *
+		 * But that's ok, we started the trans handle on our
+		 * corresponding fs_root, which means it's been added to the
+		 * dirty list.  At commit time we'll still call
+		 * btrfs_update_reloc_root() and update our root item
+		 * appropriately.
+		 */
+		reloc_root->last_trans = trans->transid;
 		trans->block_rsv = rc->block_rsv;
 
 		replaced = 0;
 		max_level = level;
 
 		ret = walk_down_reloc_tree(reloc_root, path, &level);
-		if (ret < 0) {
-			err = ret;
+		if (ret < 0)
 			goto out;
-		}
 		if (ret > 0)
 			break;
 
@@ -2345,11 +1780,8 @@ static noinline_for_stack int merge_reloc_root(struct reloc_control *rc,
 			ret = replace_path(trans, rc, root, reloc_root, path,
 					   &next_key, level, max_level);
 		}
-		if (ret < 0) {
-			err = ret;
+		if (ret < 0)
 			goto out;
-		}
-
 		if (ret > 0) {
 			level = ret;
 			btrfs_node_key_to_cpu(path->nodes[level], &key,
@@ -2368,7 +1800,7 @@ static noinline_for_stack int merge_reloc_root(struct reloc_control *rc,
 		 */
 		btrfs_node_key(path->nodes[level], &root_item->drop_progress,
 			       path->slots[level]);
-		root_item->drop_level = level;
+		btrfs_set_root_drop_level(root_item, level);
 
 		btrfs_end_transaction_throttle(trans);
 		trans = NULL;
@@ -2384,16 +1816,18 @@ static noinline_for_stack int merge_reloc_root(struct reloc_control *rc,
 	 * relocated and the block is tree root.
 	 */
 	leaf = btrfs_lock_root_node(root);
-	ret = btrfs_cow_block(trans, root, leaf, NULL, 0, &leaf);
+	ret = btrfs_cow_block(trans, root, leaf, NULL, 0, &leaf,
+			      BTRFS_NESTING_COW);
 	btrfs_tree_unlock(leaf);
 	free_extent_buffer(leaf);
-	if (ret < 0)
-		err = ret;
 out:
 	btrfs_free_path(path);
 
-	if (err == 0)
-		insert_dirty_subvol(trans, rc, root);
+	if (ret == 0) {
+		ret = insert_dirty_subvol(trans, rc, root);
+		if (ret)
+			btrfs_abort_transaction(trans, ret);
+	}
 
 	if (trans)
 		btrfs_end_transaction_throttle(trans);
@@ -2403,7 +1837,7 @@ out:
 	if (replaced && rc->stage == UPDATE_DATA_PTRS)
 		invalidate_extent_cache(root, &key, &next_key);
 
-	return err;
+	return ret;
 }
 
 static noinline_for_stack
@@ -2425,7 +1859,7 @@ int prepare_to_merge(struct reloc_control *rc, int err)
 again:
 	if (!err) {
 		num_bytes = rc->merging_rsv_size;
-		ret = btrfs_block_rsv_add(root, rc->block_rsv, num_bytes,
+		ret = btrfs_block_rsv_add(fs_info, rc->block_rsv, num_bytes,
 					  BTRFS_RESERVE_FLUSH_ALL);
 		if (ret)
 			err = ret;
@@ -2435,7 +1869,7 @@ again:
 	if (IS_ERR(trans)) {
 		if (!err)
 			btrfs_block_rsv_release(fs_info, rc->block_rsv,
-						num_bytes);
+						num_bytes, NULL);
 		return PTR_ERR(trans);
 	}
 
@@ -2443,7 +1877,7 @@ again:
 		if (num_bytes != rc->merging_rsv_size) {
 			btrfs_end_transaction(trans);
 			btrfs_block_rsv_release(fs_info, rc->block_rsv,
-						num_bytes);
+						num_bytes, NULL);
 			goto again;
 		}
 	}
@@ -2455,9 +1889,20 @@ again:
 					struct btrfs_root, root_list);
 		list_del_init(&reloc_root->root_list);
 
-		root = read_fs_root(fs_info, reloc_root->root_key.offset);
-		BUG_ON(IS_ERR(root));
-		BUG_ON(root->reloc_root != reloc_root);
+		root = btrfs_get_fs_root(fs_info, reloc_root->root_key.offset,
+				false);
+		if (IS_ERR(root)) {
+			/*
+			 * Even if we have an error we need this reloc root
+			 * back on our list so we can clean up properly.
+			 */
+			list_add(&reloc_root->root_list, &reloc_roots);
+			btrfs_abort_transaction(trans, (int)PTR_ERR(root));
+			if (!err)
+				err = PTR_ERR(root);
+			break;
+		}
+		ASSERT(root->reloc_root == reloc_root);
 
 		/*
 		 * set reference count to 1, so btrfs_recover_relocation
@@ -2465,15 +1910,27 @@ again:
 		 */
 		if (!err)
 			btrfs_set_root_refs(&reloc_root->root_item, 1);
-		btrfs_update_reloc_root(trans, root);
+		ret = btrfs_update_reloc_root(trans, root);
 
+		/*
+		 * Even if we have an error we need this reloc root back on our
+		 * list so we can clean up properly.
+		 */
 		list_add(&reloc_root->root_list, &reloc_roots);
+		btrfs_put_root(root);
+
+		if (ret) {
+			btrfs_abort_transaction(trans, ret);
+			if (!err)
+				err = ret;
+			break;
+		}
 	}
 
 	list_splice(&reloc_roots, &rc->reloc_roots);
 
 	if (!err)
-		btrfs_commit_transaction(trans);
+		err = btrfs_commit_transaction(trans);
 	else
 		btrfs_end_transaction(trans);
 	return err;
@@ -2482,17 +1939,10 @@ again:
 static noinline_for_stack
 void free_reloc_roots(struct list_head *list)
 {
-	struct btrfs_root *reloc_root;
+	struct btrfs_root *reloc_root, *tmp;
 
-	while (!list_empty(list)) {
-		reloc_root = list_entry(list->next, struct btrfs_root,
-					root_list);
+	list_for_each_entry_safe(reloc_root, tmp, list, root_list)
 		__del_reloc_root(reloc_root);
-		free_extent_buffer(reloc_root->node);
-		free_extent_buffer(reloc_root->commit_root);
-		reloc_root->node = NULL;
-		reloc_root->commit_root = NULL;
-	}
 }
 
 static noinline_for_stack
@@ -2522,13 +1972,34 @@ again:
 		reloc_root = list_entry(reloc_roots.next,
 					struct btrfs_root, root_list);
 
+		root = btrfs_get_fs_root(fs_info, reloc_root->root_key.offset,
+					 false);
 		if (btrfs_root_refs(&reloc_root->root_item) > 0) {
-			root = read_fs_root(fs_info,
-					    reloc_root->root_key.offset);
-			BUG_ON(IS_ERR(root));
-			BUG_ON(root->reloc_root != reloc_root);
-
+			if (IS_ERR(root)) {
+				/*
+				 * For recovery we read the fs roots on mount,
+				 * and if we didn't find the root then we marked
+				 * the reloc root as a garbage root.  For normal
+				 * relocation obviously the root should exist in
+				 * memory.  However there's no reason we can't
+				 * handle the error properly here just in case.
+				 */
+				ASSERT(0);
+				ret = PTR_ERR(root);
+				goto out;
+			}
+			if (root->reloc_root != reloc_root) {
+				/*
+				 * This is actually impossible without something
+				 * going really wrong (like weird race condition
+				 * or cosmic rays).
+				 */
+				ASSERT(0);
+				ret = -EINVAL;
+				goto out;
+			}
 			ret = merge_reloc_root(rc, root);
+			btrfs_put_root(root);
 			if (ret) {
 				if (list_empty(&reloc_root->root_list))
 					list_add_tail(&reloc_root->root_list,
@@ -2536,6 +2007,16 @@ again:
 				goto out;
 			}
 		} else {
+			if (!IS_ERR(root)) {
+				if (root->reloc_root == reloc_root) {
+					root->reloc_root = NULL;
+					btrfs_put_root(reloc_root);
+				}
+				clear_bit(BTRFS_ROOT_DEAD_RELOC_TREE,
+					  &root->state);
+				btrfs_put_root(root);
+			}
+
 			list_del_init(&reloc_root->root_list);
 			/* Don't forget to queue this reloc root for cleanup */
 			list_add_tail(&reloc_root->reloc_dirty_list,
@@ -2550,18 +2031,30 @@ again:
 out:
 	if (ret) {
 		btrfs_handle_fs_error(fs_info, ret, NULL);
-		if (!list_empty(&reloc_roots))
-			free_reloc_roots(&reloc_roots);
+		free_reloc_roots(&reloc_roots);
 
 		/* new reloc root may be added */
 		mutex_lock(&fs_info->reloc_mutex);
 		list_splice_init(&rc->reloc_roots, &reloc_roots);
 		mutex_unlock(&fs_info->reloc_mutex);
-		if (!list_empty(&reloc_roots))
-			free_reloc_roots(&reloc_roots);
+		free_reloc_roots(&reloc_roots);
 	}
 
-	BUG_ON(!RB_EMPTY_ROOT(&rc->reloc_root_tree.rb_root));
+	/*
+	 * We used to have
+	 *
+	 * BUG_ON(!RB_EMPTY_ROOT(&rc->reloc_root_tree.rb_root));
+	 *
+	 * here, but it's wrong.  If we fail to start the transaction in
+	 * prepare_to_merge() we will have only 0 ref reloc roots, none of which
+	 * have actually been removed from the reloc_root_tree rb tree.  This is
+	 * fine because we're bailing here, and we hold a reference on the root
+	 * for the list that holds it, so these roots will be cleaned up when we
+	 * do the reloc_dirty_list afterwards.  Meanwhile the root->reloc_root
+	 * will be cleaned up on unmount.
+	 *
+	 * The remaining nodes will be cleaned up by free_reloc_control.
+	 */
 }
 
 static void free_block_list(struct rb_root *blocks)
@@ -2580,51 +2073,126 @@ static int record_reloc_root_in_trans(struct btrfs_trans_handle *trans,
 {
 	struct btrfs_fs_info *fs_info = reloc_root->fs_info;
 	struct btrfs_root *root;
+	int ret;
 
 	if (reloc_root->last_trans == trans->transid)
 		return 0;
 
-	root = read_fs_root(fs_info, reloc_root->root_key.offset);
-	BUG_ON(IS_ERR(root));
-	BUG_ON(root->reloc_root != reloc_root);
+	root = btrfs_get_fs_root(fs_info, reloc_root->root_key.offset, false);
 
-	return btrfs_record_root_in_trans(trans, root);
+	/*
+	 * This should succeed, since we can't have a reloc root without having
+	 * already looked up the actual root and created the reloc root for this
+	 * root.
+	 *
+	 * However if there's some sort of corruption where we have a ref to a
+	 * reloc root without a corresponding root this could return ENOENT.
+	 */
+	if (IS_ERR(root)) {
+		ASSERT(0);
+		return PTR_ERR(root);
+	}
+	if (root->reloc_root != reloc_root) {
+		ASSERT(0);
+		btrfs_err(fs_info,
+			  "root %llu has two reloc roots associated with it",
+			  reloc_root->root_key.offset);
+		btrfs_put_root(root);
+		return -EUCLEAN;
+	}
+	ret = btrfs_record_root_in_trans(trans, root);
+	btrfs_put_root(root);
+
+	return ret;
 }
 
 static noinline_for_stack
 struct btrfs_root *select_reloc_root(struct btrfs_trans_handle *trans,
 				     struct reloc_control *rc,
-				     struct backref_node *node,
-				     struct backref_edge *edges[])
+				     struct btrfs_backref_node *node,
+				     struct btrfs_backref_edge *edges[])
 {
-	struct backref_node *next;
+	struct btrfs_backref_node *next;
 	struct btrfs_root *root;
 	int index = 0;
+	int ret;
 
 	next = node;
 	while (1) {
 		cond_resched();
 		next = walk_up_backref(next, edges, &index);
 		root = next->root;
-		BUG_ON(!root);
-		BUG_ON(!test_bit(BTRFS_ROOT_REF_COWS, &root->state));
+
+		/*
+		 * If there is no root, then our references for this block are
+		 * incomplete, as we should be able to walk all the way up to a
+		 * block that is owned by a root.
+		 *
+		 * This path is only for SHAREABLE roots, so if we come upon a
+		 * non-SHAREABLE root then we have backrefs that resolve
+		 * improperly.
+		 *
+		 * Both of these cases indicate file system corruption, or a bug
+		 * in the backref walking code.
+		 */
+		if (!root) {
+			ASSERT(0);
+			btrfs_err(trans->fs_info,
+		"bytenr %llu doesn't have a backref path ending in a root",
+				  node->bytenr);
+			return ERR_PTR(-EUCLEAN);
+		}
+		if (!test_bit(BTRFS_ROOT_SHAREABLE, &root->state)) {
+			ASSERT(0);
+			btrfs_err(trans->fs_info,
+	"bytenr %llu has multiple refs with one ending in a non-shareable root",
+				  node->bytenr);
+			return ERR_PTR(-EUCLEAN);
+		}
 
 		if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID) {
-			record_reloc_root_in_trans(trans, root);
+			ret = record_reloc_root_in_trans(trans, root);
+			if (ret)
+				return ERR_PTR(ret);
 			break;
 		}
 
-		btrfs_record_root_in_trans(trans, root);
+		ret = btrfs_record_root_in_trans(trans, root);
+		if (ret)
+			return ERR_PTR(ret);
 		root = root->reloc_root;
 
+		/*
+		 * We could have raced with another thread which failed, so
+		 * root->reloc_root may not be set, return ENOENT in this case.
+		 */
+		if (!root)
+			return ERR_PTR(-ENOENT);
+
 		if (next->new_bytenr != root->node->start) {
-			BUG_ON(next->new_bytenr);
-			BUG_ON(!list_empty(&next->list));
+			/*
+			 * We just created the reloc root, so we shouldn't have
+			 * ->new_bytenr set and this shouldn't be in the changed
+			 *  list.  If it is then we have multiple roots pointing
+			 *  at the same bytenr which indicates corruption, or
+			 *  we've made a mistake in the backref walking code.
+			 */
+			ASSERT(next->new_bytenr == 0);
+			ASSERT(list_empty(&next->list));
+			if (next->new_bytenr || !list_empty(&next->list)) {
+				btrfs_err(trans->fs_info,
+	"bytenr %llu possibly has multiple roots pointing at the same bytenr %llu",
+					  node->bytenr, next->bytenr);
+				return ERR_PTR(-EUCLEAN);
+			}
+
 			next->new_bytenr = root->node->start;
-			next->root = root;
+			btrfs_put_root(next->root);
+			next->root = btrfs_grab_root(root);
+			ASSERT(next->root);
 			list_add_tail(&next->list,
 				      &rc->backref_cache.changed);
-			__mark_block_processed(rc, next);
+			mark_block_processed(rc, next);
 			break;
 		}
 
@@ -2634,8 +2202,14 @@ struct btrfs_root *select_reloc_root(struct btrfs_trans_handle *trans,
 		if (!next || next->level <= node->level)
 			break;
 	}
-	if (!root)
-		return NULL;
+	if (!root) {
+		/*
+		 * This can happen if there's fs corruption or if there's a bug
+		 * in the backref lookup code.
+		 */
+		ASSERT(0);
+		return ERR_PTR(-ENOENT);
+	}
 
 	next = node;
 	/* setup backref node path for btrfs_reloc_cow_block */
@@ -2649,18 +2223,21 @@ struct btrfs_root *select_reloc_root(struct btrfs_trans_handle *trans,
 }
 
 /*
- * select a tree root for relocation. return NULL if the block
- * is reference counted. we should use do_relocation() in this
- * case. return a tree root pointer if the block isn't reference
- * counted. return -ENOENT if the block is root of reloc tree.
+ * Select a tree root for relocation.
+ *
+ * Return NULL if the block is not shareable. We should use do_relocation() in
+ * this case.
+ *
+ * Return a tree root pointer if the block is shareable.
+ * Return -ENOENT if the block is root of reloc tree.
  */
 static noinline_for_stack
-struct btrfs_root *select_one_root(struct backref_node *node)
+struct btrfs_root *select_one_root(struct btrfs_backref_node *node)
 {
-	struct backref_node *next;
+	struct btrfs_backref_node *next;
 	struct btrfs_root *root;
 	struct btrfs_root *fs_root = NULL;
-	struct backref_edge *edges[BTRFS_MAX_LEVEL - 1];
+	struct btrfs_backref_edge *edges[BTRFS_MAX_LEVEL - 1];
 	int index = 0;
 
 	next = node;
@@ -2668,10 +2245,16 @@ struct btrfs_root *select_one_root(struct backref_node *node)
 		cond_resched();
 		next = walk_up_backref(next, edges, &index);
 		root = next->root;
-		BUG_ON(!root);
 
-		/* no other choice for non-references counted tree */
-		if (!test_bit(BTRFS_ROOT_REF_COWS, &root->state))
+		/*
+		 * This can occur if we have incomplete extent refs leading all
+		 * the way up a particular path, in this case return -EUCLEAN.
+		 */
+		if (!root)
+			return ERR_PTR(-EUCLEAN);
+
+		/* No other choice for non-shareable tree */
+		if (!test_bit(BTRFS_ROOT_SHAREABLE, &root->state))
 			return root;
 
 		if (root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID)
@@ -2692,12 +2275,12 @@ struct btrfs_root *select_one_root(struct backref_node *node)
 
 static noinline_for_stack
 u64 calcu_metadata_size(struct reloc_control *rc,
-			struct backref_node *node, int reserve)
+			struct btrfs_backref_node *node, int reserve)
 {
 	struct btrfs_fs_info *fs_info = rc->extent_root->fs_info;
-	struct backref_node *next = node;
-	struct backref_edge *edge;
-	struct backref_edge *edges[BTRFS_MAX_LEVEL - 1];
+	struct btrfs_backref_node *next = node;
+	struct btrfs_backref_edge *edge;
+	struct btrfs_backref_edge *edges[BTRFS_MAX_LEVEL - 1];
 	u64 num_bytes = 0;
 	int index = 0;
 
@@ -2715,7 +2298,7 @@ u64 calcu_metadata_size(struct reloc_control *rc,
 				break;
 
 			edge = list_entry(next->upper.next,
-					  struct backref_edge, list[LOWER]);
+					struct btrfs_backref_edge, list[LOWER]);
 			edges[index++] = edge;
 			next = edge->node[UPPER];
 		}
@@ -2726,7 +2309,7 @@ u64 calcu_metadata_size(struct reloc_control *rc,
 
 static int reserve_metadata_space(struct btrfs_trans_handle *trans,
 				  struct reloc_control *rc,
-				  struct backref_node *node)
+				  struct btrfs_backref_node *node)
 {
 	struct btrfs_root *root = rc->extent_root;
 	struct btrfs_fs_info *fs_info = root->fs_info;
@@ -2744,8 +2327,8 @@ static int reserve_metadata_space(struct btrfs_trans_handle *trans,
 	 * If we get an enospc just kick back -EAGAIN so we know to drop the
 	 * transaction and try to refill when we can flush all the things.
 	 */
-	ret = btrfs_block_rsv_refill(root, rc->block_rsv, num_bytes,
-				BTRFS_RESERVE_FLUSH_LIMIT);
+	ret = btrfs_block_rsv_refill(fs_info, rc->block_rsv, num_bytes,
+				     BTRFS_RESERVE_FLUSH_LIMIT);
 	if (ret) {
 		tmp = fs_info->nodesize * RELOCATION_RESERVED_NODES;
 		while (tmp <= rc->reserved_bytes)
@@ -2774,60 +2357,58 @@ static int reserve_metadata_space(struct btrfs_trans_handle *trans,
  */
 static int do_relocation(struct btrfs_trans_handle *trans,
 			 struct reloc_control *rc,
-			 struct backref_node *node,
+			 struct btrfs_backref_node *node,
 			 struct btrfs_key *key,
 			 struct btrfs_path *path, int lowest)
 {
-	struct btrfs_fs_info *fs_info = rc->extent_root->fs_info;
-	struct backref_node *upper;
-	struct backref_edge *edge;
-	struct backref_edge *edges[BTRFS_MAX_LEVEL - 1];
+	struct btrfs_backref_node *upper;
+	struct btrfs_backref_edge *edge;
+	struct btrfs_backref_edge *edges[BTRFS_MAX_LEVEL - 1];
 	struct btrfs_root *root;
 	struct extent_buffer *eb;
 	u32 blocksize;
 	u64 bytenr;
-	u64 generation;
 	int slot;
-	int ret;
-	int err = 0;
+	int ret = 0;
 
-	BUG_ON(lowest && node->eb);
+	/*
+	 * If we are lowest then this is the first time we're processing this
+	 * block, and thus shouldn't have an eb associated with it yet.
+	 */
+	ASSERT(!lowest || !node->eb);
 
 	path->lowest_level = node->level + 1;
 	rc->backref_cache.path[node->level] = node;
 	list_for_each_entry(edge, &node->upper, list[LOWER]) {
-		struct btrfs_key first_key;
 		struct btrfs_ref ref = { 0 };
 
 		cond_resched();
 
 		upper = edge->node[UPPER];
 		root = select_reloc_root(trans, rc, upper, edges);
-		BUG_ON(!root);
+		if (IS_ERR(root)) {
+			ret = PTR_ERR(root);
+			goto next;
+		}
 
 		if (upper->eb && !upper->locked) {
 			if (!lowest) {
-				ret = btrfs_bin_search(upper->eb, key,
-						       upper->level, &slot);
-				if (ret < 0) {
-					err = ret;
+				ret = btrfs_bin_search(upper->eb, key, &slot);
+				if (ret < 0)
 					goto next;
-				}
 				BUG_ON(ret);
 				bytenr = btrfs_node_blockptr(upper->eb, slot);
 				if (node->eb->start == bytenr)
 					goto next;
 			}
-			drop_node_buffer(upper);
+			btrfs_backref_drop_node_buffer(upper);
 		}
 
 		if (!upper->eb) {
 			ret = btrfs_search_slot(trans, root, key, path, 0, 1);
 			if (ret) {
-				if (ret < 0)
-					err = ret;
-				else
-					err = -ENOENT;
+				if (ret > 0)
+					ret = -ENOENT;
 
 				btrfs_release_path(path);
 				break;
@@ -2846,12 +2427,9 @@ static int do_relocation(struct btrfs_trans_handle *trans,
 			slot = path->slots[upper->level];
 			btrfs_release_path(path);
 		} else {
-			ret = btrfs_bin_search(upper->eb, key, upper->level,
-					       &slot);
-			if (ret < 0) {
-				err = ret;
+			ret = btrfs_bin_search(upper->eb, key, &slot);
+			if (ret < 0)
 				goto next;
-			}
 			BUG_ON(ret);
 		}
 
@@ -2862,7 +2440,7 @@ static int do_relocation(struct btrfs_trans_handle *trans,
 		"lowest leaf/node mismatch: bytenr %llu node->bytenr %llu slot %d upper %llu",
 					  bytenr, node->bytenr, slot,
 					  upper->eb->start);
-				err = -EIO;
+				ret = -EIO;
 				goto next;
 			}
 		} else {
@@ -2871,31 +2449,25 @@ static int do_relocation(struct btrfs_trans_handle *trans,
 		}
 
 		blocksize = root->fs_info->nodesize;
-		generation = btrfs_node_ptr_generation(upper->eb, slot);
-		btrfs_node_key_to_cpu(upper->eb, &first_key, slot);
-		eb = read_tree_block(fs_info, bytenr, generation,
-				     upper->level - 1, &first_key);
+		eb = btrfs_read_node_slot(upper->eb, slot);
 		if (IS_ERR(eb)) {
-			err = PTR_ERR(eb);
-			goto next;
-		} else if (!extent_buffer_uptodate(eb)) {
-			free_extent_buffer(eb);
-			err = -EIO;
+			ret = PTR_ERR(eb);
 			goto next;
 		}
 		btrfs_tree_lock(eb);
-		btrfs_set_lock_blocking_write(eb);
 
 		if (!node->eb) {
 			ret = btrfs_cow_block(trans, root, eb, upper->eb,
-					      slot, &eb);
+					      slot, &eb, BTRFS_NESTING_COW);
 			btrfs_tree_unlock(eb);
 			free_extent_buffer(eb);
-			if (ret < 0) {
-				err = ret;
+			if (ret < 0)
 				goto next;
-			}
-			BUG_ON(node->eb != eb);
+			/*
+			 * We've just COWed this block, it should have updated
+			 * the correct backref node entry.
+			 */
+			ASSERT(node->eb == eb);
 		} else {
 			btrfs_set_node_blockptr(upper->eb, slot,
 						node->eb->start);
@@ -2906,38 +2478,44 @@ static int do_relocation(struct btrfs_trans_handle *trans,
 			btrfs_init_generic_ref(&ref, BTRFS_ADD_DELAYED_REF,
 					       node->eb->start, blocksize,
 					       upper->eb->start);
-			ref.real_root = root->root_key.objectid;
 			btrfs_init_tree_ref(&ref, node->level,
-					    btrfs_header_owner(upper->eb));
+					    btrfs_header_owner(upper->eb),
+					    root->root_key.objectid, false);
 			ret = btrfs_inc_extent_ref(trans, &ref);
-			BUG_ON(ret);
-
-			ret = btrfs_drop_subtree(trans, root, eb, upper->eb);
-			BUG_ON(ret);
+			if (!ret)
+				ret = btrfs_drop_subtree(trans, root, eb,
+							 upper->eb);
+			if (ret)
+				btrfs_abort_transaction(trans, ret);
 		}
 next:
 		if (!upper->pending)
-			drop_node_buffer(upper);
+			btrfs_backref_drop_node_buffer(upper);
 		else
-			unlock_node_buffer(upper);
-		if (err)
+			btrfs_backref_unlock_node_buffer(upper);
+		if (ret)
 			break;
 	}
 
-	if (!err && node->pending) {
-		drop_node_buffer(node);
+	if (!ret && node->pending) {
+		btrfs_backref_drop_node_buffer(node);
 		list_move_tail(&node->list, &rc->backref_cache.changed);
 		node->pending = 0;
 	}
 
 	path->lowest_level = 0;
-	BUG_ON(err == -ENOSPC);
-	return err;
+
+	/*
+	 * We should have allocated all of our space in the block rsv and thus
+	 * shouldn't ENOSPC.
+	 */
+	ASSERT(ret != -ENOSPC);
+	return ret;
 }
 
 static int link_to_upper(struct btrfs_trans_handle *trans,
 			 struct reloc_control *rc,
-			 struct backref_node *node,
+			 struct btrfs_backref_node *node,
 			 struct btrfs_path *path)
 {
 	struct btrfs_key key;
@@ -2951,15 +2529,15 @@ static int finish_pending_nodes(struct btrfs_trans_handle *trans,
 				struct btrfs_path *path, int err)
 {
 	LIST_HEAD(list);
-	struct backref_cache *cache = &rc->backref_cache;
-	struct backref_node *node;
+	struct btrfs_backref_cache *cache = &rc->backref_cache;
+	struct btrfs_backref_node *node;
 	int level;
 	int ret;
 
 	for (level = 0; level < BTRFS_MAX_LEVEL; level++) {
 		while (!list_empty(&cache->pending[level])) {
 			node = list_entry(cache->pending[level].next,
-					  struct backref_node, list);
+					  struct btrfs_backref_node, list);
 			list_move_tail(&node->list, &list);
 			BUG_ON(!node->pending);
 
@@ -2974,35 +2552,16 @@ static int finish_pending_nodes(struct btrfs_trans_handle *trans,
 	return err;
 }
 
-static void mark_block_processed(struct reloc_control *rc,
-				 u64 bytenr, u32 blocksize)
-{
-	set_extent_bits(&rc->processed_blocks, bytenr, bytenr + blocksize - 1,
-			EXTENT_DIRTY);
-}
-
-static void __mark_block_processed(struct reloc_control *rc,
-				   struct backref_node *node)
-{
-	u32 blocksize;
-	if (node->level == 0 ||
-	    in_block_group(node->bytenr, rc->block_group)) {
-		blocksize = rc->extent_root->fs_info->nodesize;
-		mark_block_processed(rc, node->bytenr, blocksize);
-	}
-	node->processed = 1;
-}
-
 /*
  * mark a block and all blocks directly/indirectly reference the block
  * as processed.
  */
 static void update_processed_blocks(struct reloc_control *rc,
-				    struct backref_node *node)
+				    struct btrfs_backref_node *node)
 {
-	struct backref_node *next = node;
-	struct backref_edge *edge;
-	struct backref_edge *edges[BTRFS_MAX_LEVEL - 1];
+	struct btrfs_backref_node *next = node;
+	struct btrfs_backref_edge *edge;
+	struct btrfs_backref_edge *edges[BTRFS_MAX_LEVEL - 1];
 	int index = 0;
 
 	while (next) {
@@ -3011,13 +2570,13 @@ static void update_processed_blocks(struct reloc_control *rc,
 			if (next->processed)
 				break;
 
-			__mark_block_processed(rc, next);
+			mark_block_processed(rc, next);
 
 			if (list_empty(&next->upper))
 				break;
 
 			edge = list_entry(next->upper.next,
-					  struct backref_edge, list[LOWER]);
+					struct btrfs_backref_edge, list[LOWER]);
 			edges[index++] = edge;
 			next = edge->node[UPPER];
 		}
@@ -3040,12 +2599,11 @@ static int get_tree_block_key(struct btrfs_fs_info *fs_info,
 {
 	struct extent_buffer *eb;
 
-	BUG_ON(block->key_ready);
-	eb = read_tree_block(fs_info, block->bytenr, block->key.offset,
-			     block->level, NULL);
-	if (IS_ERR(eb)) {
+	eb = read_tree_block(fs_info, block->bytenr, block->owner,
+			     block->key.offset, block->level, NULL);
+	if (IS_ERR(eb))
 		return PTR_ERR(eb);
-	} else if (!extent_buffer_uptodate(eb)) {
+	if (!extent_buffer_uptodate(eb)) {
 		free_extent_buffer(eb);
 		return -EIO;
 	}
@@ -3063,7 +2621,7 @@ static int get_tree_block_key(struct btrfs_fs_info *fs_info,
  */
 static int relocate_tree_block(struct btrfs_trans_handle *trans,
 				struct reloc_control *rc,
-				struct backref_node *node,
+				struct btrfs_backref_node *node,
 				struct btrfs_key *key,
 				struct btrfs_path *path)
 {
@@ -3073,32 +2631,77 @@ static int relocate_tree_block(struct btrfs_trans_handle *trans,
 	if (!node)
 		return 0;
 
+	/*
+	 * If we fail here we want to drop our backref_node because we are going
+	 * to start over and regenerate the tree for it.
+	 */
+	ret = reserve_metadata_space(trans, rc, node);
+	if (ret)
+		goto out;
+
 	BUG_ON(node->processed);
 	root = select_one_root(node);
-	if (root == ERR_PTR(-ENOENT)) {
-		update_processed_blocks(rc, node);
-		goto out;
-	}
+	if (IS_ERR(root)) {
+		ret = PTR_ERR(root);
 
-	if (!root || test_bit(BTRFS_ROOT_REF_COWS, &root->state)) {
-		ret = reserve_metadata_space(trans, rc, node);
-		if (ret)
-			goto out;
+		/* See explanation in select_one_root for the -EUCLEAN case. */
+		ASSERT(ret == -ENOENT);
+		if (ret == -ENOENT) {
+			ret = 0;
+			update_processed_blocks(rc, node);
+		}
+		goto out;
 	}
 
 	if (root) {
-		if (test_bit(BTRFS_ROOT_REF_COWS, &root->state)) {
-			BUG_ON(node->new_bytenr);
-			BUG_ON(!list_empty(&node->list));
-			btrfs_record_root_in_trans(trans, root);
+		if (test_bit(BTRFS_ROOT_SHAREABLE, &root->state)) {
+			/*
+			 * This block was the root block of a root, and this is
+			 * the first time we're processing the block and thus it
+			 * should not have had the ->new_bytenr modified and
+			 * should have not been included on the changed list.
+			 *
+			 * However in the case of corruption we could have
+			 * multiple refs pointing to the same block improperly,
+			 * and thus we would trip over these checks.  ASSERT()
+			 * for the developer case, because it could indicate a
+			 * bug in the backref code, however error out for a
+			 * normal user in the case of corruption.
+			 */
+			ASSERT(node->new_bytenr == 0);
+			ASSERT(list_empty(&node->list));
+			if (node->new_bytenr || !list_empty(&node->list)) {
+				btrfs_err(root->fs_info,
+				  "bytenr %llu has improper references to it",
+					  node->bytenr);
+				ret = -EUCLEAN;
+				goto out;
+			}
+			ret = btrfs_record_root_in_trans(trans, root);
+			if (ret)
+				goto out;
+			/*
+			 * Another thread could have failed, need to check if we
+			 * have reloc_root actually set.
+			 */
+			if (!root->reloc_root) {
+				ret = -ENOENT;
+				goto out;
+			}
 			root = root->reloc_root;
 			node->new_bytenr = root->node->start;
-			node->root = root;
+			btrfs_put_root(node->root);
+			node->root = btrfs_grab_root(root);
+			ASSERT(node->root);
 			list_add_tail(&node->list, &rc->backref_cache.changed);
 		} else {
 			path->lowest_level = node->level;
+			if (root == root->fs_info->chunk_root)
+				btrfs_reserve_chunk_metadata(trans, false);
 			ret = btrfs_search_slot(trans, root, key, path, 0, 1);
 			btrfs_release_path(path);
+			if (root == root->fs_info->chunk_root)
+				btrfs_trans_release_chunk_metadata(trans);
 			if (ret > 0)
 				ret = 0;
 		}
@@ -3109,7 +2712,7 @@ static int relocate_tree_block(struct btrfs_trans_handle *trans,
 	}
 out:
 	if (ret || node->level == 0 || node->cowonly)
-		remove_backref_node(&rc->backref_cache, node);
+		btrfs_backref_cleanup_node(&rc->backref_cache, node);
 	return ret;
 }
 
@@ -3121,7 +2724,7 @@ int relocate_tree_blocks(struct btrfs_trans_handle *trans,
 			 struct reloc_control *rc, struct rb_root *blocks)
 {
 	struct btrfs_fs_info *fs_info = rc->extent_root->fs_info;
-	struct backref_node *node;
+	struct btrfs_backref_node *node;
 	struct btrfs_path *path;
 	struct tree_block *block;
 	struct tree_block *next;
@@ -3137,7 +2740,9 @@ int relocate_tree_blocks(struct btrfs_trans_handle *trans,
 	/* Kick in readahead for tree blocks with missing keys */
 	rbtree_postorder_for_each_entry_safe(block, next, blocks, rb_node) {
 		if (!block->key_ready)
-			readahead_tree_block(fs_info, block->bytenr);
+			btrfs_readahead_tree_block(fs_info, block->bytenr,
+						   block->owner, 0,
+						   block->level);
 	}
 
 	/* Get first keys */
@@ -3161,9 +2766,8 @@ int relocate_tree_blocks(struct btrfs_trans_handle *trans,
 		ret = relocate_tree_block(trans, rc, node, &block->key,
 					  path);
 		if (ret < 0) {
-			if (ret != -EAGAIN || &block->rb_node == rb_first(blocks))
-				err = ret;
-			goto out;
+			err = ret;
+			break;
 		}
 	}
 out:
@@ -3176,66 +2780,116 @@ out_free_blocks:
 	return err;
 }
 
-static noinline_for_stack
-int prealloc_file_extent_cluster(struct inode *inode,
-				 struct file_extent_cluster *cluster)
+static noinline_for_stack int prealloc_file_extent_cluster(
+				struct btrfs_inode *inode,
+				struct file_extent_cluster *cluster)
 {
 	u64 alloc_hint = 0;
 	u64 start;
 	u64 end;
-	u64 offset = BTRFS_I(inode)->index_cnt;
+	u64 offset = inode->index_cnt;
 	u64 num_bytes;
-	int nr = 0;
+	int nr;
 	int ret = 0;
+	u64 i_size = i_size_read(&inode->vfs_inode);
 	u64 prealloc_start = cluster->start - offset;
 	u64 prealloc_end = cluster->end - offset;
-	u64 cur_offset;
-	struct extent_changeset *data_reserved = NULL;
+	u64 cur_offset = prealloc_start;
 
-	BUG_ON(cluster->start != cluster->boundary[0]);
-	inode_lock(inode);
+	/*
+	 * For subpage case, previous i_size may not be aligned to PAGE_SIZE.
+	 * This means the range [i_size, PAGE_END + 1) is filled with zeros by
+	 * btrfs_do_readpage() call of previously relocated file cluster.
+	 *
+	 * If the current cluster starts in the above range, btrfs_do_readpage()
+	 * will skip the read, and relocate_one_page() will later writeback
+	 * the padding zeros as new data, causing data corruption.
+	 *
+	 * Here we have to manually invalidate the range (i_size, PAGE_END + 1).
+	 */
+	if (!IS_ALIGNED(i_size, PAGE_SIZE)) {
+		struct address_space *mapping = inode->vfs_inode.i_mapping;
+		struct btrfs_fs_info *fs_info = inode->root->fs_info;
+		const u32 sectorsize = fs_info->sectorsize;
+		struct page *page;
+
+		ASSERT(sectorsize < PAGE_SIZE);
+		ASSERT(IS_ALIGNED(i_size, sectorsize));
+
+		/*
+		 * Subpage can't handle page with DIRTY but without UPTODATE
+		 * bit as it can lead to the following deadlock:
+		 *
+		 * btrfs_read_folio()
+		 * | Page already *locked*
+		 * |- btrfs_lock_and_flush_ordered_range()
+		 *    |- btrfs_start_ordered_extent()
+		 *       |- extent_write_cache_pages()
+		 *          |- lock_page()
+		 *             We try to lock the page we already hold.
+		 *
+		 * Here we just writeback the whole data reloc inode, so that
+		 * we will be ensured to have no dirty range in the page, and
+		 * are safe to clear the uptodate bits.
+		 *
+		 * This shouldn't cause too much overhead, as we need to write
+		 * the data back anyway.
+		 */
+		ret = filemap_write_and_wait(mapping);
+		if (ret < 0)
+			return ret;
 
-	ret = btrfs_check_data_free_space(inode, &data_reserved, prealloc_start,
-					  prealloc_end + 1 - prealloc_start);
+		clear_extent_bits(&inode->io_tree, i_size,
+				  round_up(i_size, PAGE_SIZE) - 1,
+				  EXTENT_UPTODATE);
+		page = find_lock_page(mapping, i_size >> PAGE_SHIFT);
+		/*
+		 * If page is freed we don't need to do anything then, as we
+		 * will re-read the whole page anyway.
+		 */
+		if (page) {
+			btrfs_subpage_clear_uptodate(fs_info, page, i_size,
+					round_up(i_size, PAGE_SIZE) - i_size);
+			unlock_page(page);
+			put_page(page);
+		}
+	}
+
+	BUG_ON(cluster->start != cluster->boundary[0]);
+	ret = btrfs_alloc_data_chunk_ondemand(inode,
+					      prealloc_end + 1 - prealloc_start);
 	if (ret)
-		goto out;
+		return ret;
 
-	cur_offset = prealloc_start;
-	while (nr < cluster->nr) {
+	btrfs_inode_lock(&inode->vfs_inode, 0);
+	for (nr = 0; nr < cluster->nr; nr++) {
 		start = cluster->boundary[nr] - offset;
 		if (nr + 1 < cluster->nr)
 			end = cluster->boundary[nr + 1] - 1 - offset;
 		else
 			end = cluster->end - offset;
 
-		lock_extent(&BTRFS_I(inode)->io_tree, start, end);
+		lock_extent(&inode->io_tree, start, end, NULL);
 		num_bytes = end + 1 - start;
-		if (cur_offset < start)
-			btrfs_free_reserved_data_space(inode, data_reserved,
-					cur_offset, start - cur_offset);
-		ret = btrfs_prealloc_file_range(inode, 0, start,
+		ret = btrfs_prealloc_file_range(&inode->vfs_inode, 0, start,
 						num_bytes, num_bytes,
 						end + 1, &alloc_hint);
 		cur_offset = end + 1;
-		unlock_extent(&BTRFS_I(inode)->io_tree, start, end);
+		unlock_extent(&inode->io_tree, start, end, NULL);
 		if (ret)
 			break;
-		nr++;
 	}
+	btrfs_inode_unlock(&inode->vfs_inode, 0);
+
 	if (cur_offset < prealloc_end)
-		btrfs_free_reserved_data_space(inode, data_reserved,
-				cur_offset, prealloc_end + 1 - cur_offset);
-out:
-	inode_unlock(inode);
-	extent_changeset_free(data_reserved);
+		btrfs_free_reserved_data_space_noquota(inode->root->fs_info,
+					       prealloc_end + 1 - cur_offset);
 	return ret;
 }
 
-static noinline_for_stack
-int setup_extent_mapping(struct inode *inode, u64 start, u64 end,
-			 u64 block_start)
+static noinline_for_stack int setup_relocation_extent_mapping(struct inode *inode,
+				u64 start, u64 end, u64 block_start)
 {
-	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
 	struct extent_map *em;
 	int ret = 0;
 
@@ -3249,34 +2903,169 @@ int setup_extent_mapping(struct inode *inode, u64 start, u64 end,
 	em->block_start = block_start;
 	set_bit(EXTENT_FLAG_PINNED, &em->flags);
 
-	lock_extent(&BTRFS_I(inode)->io_tree, start, end);
-	while (1) {
-		write_lock(&em_tree->lock);
-		ret = add_extent_mapping(em_tree, em, 0);
-		write_unlock(&em_tree->lock);
-		if (ret != -EEXIST) {
-			free_extent_map(em);
-			break;
+	lock_extent(&BTRFS_I(inode)->io_tree, start, end, NULL);
+	ret = btrfs_replace_extent_map_range(BTRFS_I(inode), em, false);
+	unlock_extent(&BTRFS_I(inode)->io_tree, start, end, NULL);
+	free_extent_map(em);
+
+	return ret;
+}
+
+/*
+ * Allow error injection to test balance/relocation cancellation
+ */
+noinline int btrfs_should_cancel_balance(struct btrfs_fs_info *fs_info)
+{
+	return atomic_read(&fs_info->balance_cancel_req) ||
+		atomic_read(&fs_info->reloc_cancel_req) ||
+		fatal_signal_pending(current);
+}
+ALLOW_ERROR_INJECTION(btrfs_should_cancel_balance, TRUE);
+
+static u64 get_cluster_boundary_end(struct file_extent_cluster *cluster,
+				    int cluster_nr)
+{
+	/* Last extent, use cluster end directly */
+	if (cluster_nr >= cluster->nr - 1)
+		return cluster->end;
+
+	/* Use next boundary start*/
+	return cluster->boundary[cluster_nr + 1] - 1;
+}
+
+static int relocate_one_page(struct inode *inode, struct file_ra_state *ra,
+			     struct file_extent_cluster *cluster,
+			     int *cluster_nr, unsigned long page_index)
+{
+	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+	u64 offset = BTRFS_I(inode)->index_cnt;
+	const unsigned long last_index = (cluster->end - offset) >> PAGE_SHIFT;
+	gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping);
+	struct page *page;
+	u64 page_start;
+	u64 page_end;
+	u64 cur;
+	int ret;
+
+	ASSERT(page_index <= last_index);
+	page = find_lock_page(inode->i_mapping, page_index);
+	if (!page) {
+		page_cache_sync_readahead(inode->i_mapping, ra, NULL,
+				page_index, last_index + 1 - page_index);
+		page = find_or_create_page(inode->i_mapping, page_index, mask);
+		if (!page)
+			return -ENOMEM;
+	}
+	ret = set_page_extent_mapped(page);
+	if (ret < 0)
+		goto release_page;
+
+	if (PageReadahead(page))
+		page_cache_async_readahead(inode->i_mapping, ra, NULL,
+				page_folio(page), page_index,
+				last_index + 1 - page_index);
+
+	if (!PageUptodate(page)) {
+		btrfs_read_folio(NULL, page_folio(page));
+		lock_page(page);
+		if (!PageUptodate(page)) {
+			ret = -EIO;
+			goto release_page;
+		}
+	}
+
+	page_start = page_offset(page);
+	page_end = page_start + PAGE_SIZE - 1;
+
+	/*
+	 * Start from the cluster, as for subpage case, the cluster can start
+	 * inside the page.
+	 */
+	cur = max(page_start, cluster->boundary[*cluster_nr] - offset);
+	while (cur <= page_end) {
+		u64 extent_start = cluster->boundary[*cluster_nr] - offset;
+		u64 extent_end = get_cluster_boundary_end(cluster,
+						*cluster_nr) - offset;
+		u64 clamped_start = max(page_start, extent_start);
+		u64 clamped_end = min(page_end, extent_end);
+		u32 clamped_len = clamped_end + 1 - clamped_start;
+
+		/* Reserve metadata for this range */
+		ret = btrfs_delalloc_reserve_metadata(BTRFS_I(inode),
+						      clamped_len, clamped_len,
+						      false);
+		if (ret)
+			goto release_page;
+
+		/* Mark the range delalloc and dirty for later writeback */
+		lock_extent(&BTRFS_I(inode)->io_tree, clamped_start, clamped_end, NULL);
+		ret = btrfs_set_extent_delalloc(BTRFS_I(inode), clamped_start,
+						clamped_end, 0, NULL);
+		if (ret) {
+			clear_extent_bits(&BTRFS_I(inode)->io_tree,
+					clamped_start, clamped_end,
+					EXTENT_LOCKED | EXTENT_BOUNDARY);
+			btrfs_delalloc_release_metadata(BTRFS_I(inode),
+							clamped_len, true);
+			btrfs_delalloc_release_extents(BTRFS_I(inode),
+						       clamped_len);
+			goto release_page;
+		}
+		btrfs_page_set_dirty(fs_info, page, clamped_start, clamped_len);
+
+		/*
+		 * Set the boundary if it's inside the page.
+		 * Data relocation requires the destination extents to have the
+		 * same size as the source.
+		 * EXTENT_BOUNDARY bit prevents current extent from being merged
+		 * with previous extent.
+		 */
+		if (in_range(cluster->boundary[*cluster_nr] - offset,
+			     page_start, PAGE_SIZE)) {
+			u64 boundary_start = cluster->boundary[*cluster_nr] -
+						offset;
+			u64 boundary_end = boundary_start +
+					   fs_info->sectorsize - 1;
+
+			set_extent_bits(&BTRFS_I(inode)->io_tree,
+					boundary_start, boundary_end,
+					EXTENT_BOUNDARY);
+		}
+		unlock_extent(&BTRFS_I(inode)->io_tree, clamped_start, clamped_end, NULL);
+		btrfs_delalloc_release_extents(BTRFS_I(inode), clamped_len);
+		cur += clamped_len;
+
+		/* Crossed extent end, go to next extent */
+		if (cur >= extent_end) {
+			(*cluster_nr)++;
+			/* Just finished the last extent of the cluster, exit. */
+			if (*cluster_nr >= cluster->nr)
+				break;
 		}
-		btrfs_drop_extent_cache(BTRFS_I(inode), start, end, 0);
 	}
-	unlock_extent(&BTRFS_I(inode)->io_tree, start, end);
+	unlock_page(page);
+	put_page(page);
+
+	balance_dirty_pages_ratelimited(inode->i_mapping);
+	btrfs_throttle(fs_info);
+	if (btrfs_should_cancel_balance(fs_info))
+		ret = -ECANCELED;
+	return ret;
+
+release_page:
+	unlock_page(page);
+	put_page(page);
 	return ret;
 }
 
 static int relocate_file_extent_cluster(struct inode *inode,
 					struct file_extent_cluster *cluster)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
-	u64 page_start;
-	u64 page_end;
 	u64 offset = BTRFS_I(inode)->index_cnt;
 	unsigned long index;
 	unsigned long last_index;
-	struct page *page;
 	struct file_ra_state *ra;
-	gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping);
-	int nr = 0;
+	int cluster_nr = 0;
 	int ret = 0;
 
 	if (!cluster->nr)
@@ -3286,107 +3075,23 @@ static int relocate_file_extent_cluster(struct inode *inode,
 	if (!ra)
 		return -ENOMEM;
 
-	ret = prealloc_file_extent_cluster(inode, cluster);
+	ret = prealloc_file_extent_cluster(BTRFS_I(inode), cluster);
 	if (ret)
 		goto out;
 
 	file_ra_state_init(ra, inode->i_mapping);
 
-	ret = setup_extent_mapping(inode, cluster->start - offset,
+	ret = setup_relocation_extent_mapping(inode, cluster->start - offset,
 				   cluster->end - offset, cluster->start);
 	if (ret)
 		goto out;
 
-	index = (cluster->start - offset) >> PAGE_SHIFT;
 	last_index = (cluster->end - offset) >> PAGE_SHIFT;
-	while (index <= last_index) {
-		ret = btrfs_delalloc_reserve_metadata(BTRFS_I(inode),
-				PAGE_SIZE);
-		if (ret)
-			goto out;
-
-		page = find_lock_page(inode->i_mapping, index);
-		if (!page) {
-			page_cache_sync_readahead(inode->i_mapping,
-						  ra, NULL, index,
-						  last_index + 1 - index);
-			page = find_or_create_page(inode->i_mapping, index,
-						   mask);
-			if (!page) {
-				btrfs_delalloc_release_metadata(BTRFS_I(inode),
-							PAGE_SIZE, true);
-				btrfs_delalloc_release_extents(BTRFS_I(inode),
-							PAGE_SIZE);
-				ret = -ENOMEM;
-				goto out;
-			}
-		}
-
-		if (PageReadahead(page)) {
-			page_cache_async_readahead(inode->i_mapping,
-						   ra, NULL, page, index,
-						   last_index + 1 - index);
-		}
-
-		if (!PageUptodate(page)) {
-			btrfs_readpage(NULL, page);
-			lock_page(page);
-			if (!PageUptodate(page)) {
-				unlock_page(page);
-				put_page(page);
-				btrfs_delalloc_release_metadata(BTRFS_I(inode),
-							PAGE_SIZE, true);
-				btrfs_delalloc_release_extents(BTRFS_I(inode),
-							       PAGE_SIZE);
-				ret = -EIO;
-				goto out;
-			}
-		}
-
-		page_start = page_offset(page);
-		page_end = page_start + PAGE_SIZE - 1;
-
-		lock_extent(&BTRFS_I(inode)->io_tree, page_start, page_end);
-
-		set_page_extent_mapped(page);
-
-		if (nr < cluster->nr &&
-		    page_start + offset == cluster->boundary[nr]) {
-			set_extent_bits(&BTRFS_I(inode)->io_tree,
-					page_start, page_end,
-					EXTENT_BOUNDARY);
-			nr++;
-		}
-
-		ret = btrfs_set_extent_delalloc(inode, page_start, page_end, 0,
-						NULL);
-		if (ret) {
-			unlock_page(page);
-			put_page(page);
-			btrfs_delalloc_release_metadata(BTRFS_I(inode),
-							 PAGE_SIZE, true);
-			btrfs_delalloc_release_extents(BTRFS_I(inode),
-			                               PAGE_SIZE);
-
-			clear_extent_bits(&BTRFS_I(inode)->io_tree,
-					  page_start, page_end,
-					  EXTENT_LOCKED | EXTENT_BOUNDARY);
-			goto out;
-
-		}
-		set_page_dirty(page);
-
-		unlock_extent(&BTRFS_I(inode)->io_tree,
-			      page_start, page_end);
-		unlock_page(page);
-		put_page(page);
-
-		index++;
-		btrfs_delalloc_release_extents(BTRFS_I(inode), PAGE_SIZE);
-		balance_dirty_pages_ratelimited(inode->i_mapping);
-		btrfs_throttle(fs_info);
-	}
-	WARN_ON(nr != cluster->nr);
+	for (index = (cluster->start - offset) >> PAGE_SHIFT;
+	     index <= last_index && !ret; index++)
+		ret = relocate_one_page(inode, ra, cluster, &cluster_nr, index);
+	if (ret == 0)
+		WARN_ON(cluster_nr != cluster->nr);
 out:
 	kfree(ra);
 	return ret;
@@ -3439,21 +3144,58 @@ static int add_tree_block(struct reloc_control *rc,
 	u32 item_size;
 	int level = -1;
 	u64 generation;
+	u64 owner = 0;
 
 	eb =  path->nodes[0];
-	item_size = btrfs_item_size_nr(eb, path->slots[0]);
+	item_size = btrfs_item_size(eb, path->slots[0]);
 
 	if (extent_key->type == BTRFS_METADATA_ITEM_KEY ||
 	    item_size >= sizeof(*ei) + sizeof(*bi)) {
+		unsigned long ptr = 0, end;
+
 		ei = btrfs_item_ptr(eb, path->slots[0],
 				struct btrfs_extent_item);
+		end = (unsigned long)ei + item_size;
 		if (extent_key->type == BTRFS_EXTENT_ITEM_KEY) {
 			bi = (struct btrfs_tree_block_info *)(ei + 1);
 			level = btrfs_tree_block_level(eb, bi);
+			ptr = (unsigned long)(bi + 1);
 		} else {
 			level = (int)extent_key->offset;
+			ptr = (unsigned long)(ei + 1);
 		}
 		generation = btrfs_extent_generation(eb, ei);
+
+		/*
+		 * We're reading random blocks without knowing their owner ahead
+		 * of time.  This is ok most of the time, as all reloc roots and
+		 * fs roots have the same lock type.  However normal trees do
+		 * not, and the only way to know ahead of time is to read the
+		 * inline ref offset.  We know it's an fs root if
+		 *
+		 * 1. There's more than one ref.
+		 * 2. There's a SHARED_DATA_REF_KEY set.
+		 * 3. FULL_BACKREF is set on the flags.
+		 *
+		 * Otherwise it's safe to assume that the ref offset == the
+		 * owner of this block, so we can use that when calling
+		 * read_tree_block.
+		 */
+		if (btrfs_extent_refs(eb, ei) == 1 &&
+		    !(btrfs_extent_flags(eb, ei) &
+		      BTRFS_BLOCK_FLAG_FULL_BACKREF) &&
+		    ptr < end) {
+			struct btrfs_extent_inline_ref *iref;
+			int type;
+
+			iref = (struct btrfs_extent_inline_ref *)ptr;
+			type = btrfs_get_extent_inline_ref_type(eb, iref,
+							BTRFS_REF_TYPE_BLOCK);
+			if (type == BTRFS_REF_TYPE_INVALID)
+				return -EINVAL;
+			if (type == BTRFS_TREE_BLOCK_REF_KEY)
+				owner = btrfs_extent_inline_ref_offset(eb, iref);
+		}
 	} else if (unlikely(item_size == sizeof(struct btrfs_extent_item_v0))) {
 		btrfs_print_v0_err(eb->fs_info);
 		btrfs_handle_fs_error(eb->fs_info, -EINVAL, NULL);
@@ -3475,10 +3217,12 @@ static int add_tree_block(struct reloc_control *rc,
 	block->key.offset = generation;
 	block->level = level;
 	block->key_ready = 0;
+	block->owner = owner;
 
-	rb_node = tree_insert(blocks, block->bytenr, &block->rb_node);
+	rb_node = rb_simple_insert(blocks, block->bytenr, &block->rb_node);
 	if (rb_node)
-		backref_tree_panic(rb_node, -EEXIST, block->bytenr);
+		btrfs_backref_panic(rc->extent_root->fs_info, block->bytenr,
+				    -EEXIST);
 
 	return 0;
 }
@@ -3499,7 +3243,7 @@ static int __add_tree_block(struct reloc_control *rc,
 	if (tree_block_processed(bytenr, rc))
 		return 0;
 
-	if (tree_search(blocks, bytenr))
+	if (rb_simple_search(blocks, bytenr))
 		return 0;
 
 	path = btrfs_alloc_path();
@@ -3556,37 +3300,11 @@ out:
 	return ret;
 }
 
-/*
- * helper to check if the block use full backrefs for pointers in it
- */
-static int block_use_full_backref(struct reloc_control *rc,
-				  struct extent_buffer *eb)
-{
-	u64 flags;
-	int ret;
-
-	if (btrfs_header_flag(eb, BTRFS_HEADER_FLAG_RELOC) ||
-	    btrfs_header_backref_rev(eb) < BTRFS_MIXED_BACKREF_REV)
-		return 1;
-
-	ret = btrfs_lookup_extent_info(NULL, rc->extent_root->fs_info,
-				       eb->start, btrfs_header_level(eb), 1,
-				       NULL, &flags);
-	BUG_ON(ret);
-
-	if (flags & BTRFS_BLOCK_FLAG_FULL_BACKREF)
-		ret = 1;
-	else
-		ret = 0;
-	return ret;
-}
-
 static int delete_block_group_cache(struct btrfs_fs_info *fs_info,
 				    struct btrfs_block_group *block_group,
 				    struct inode *inode,
 				    u64 ino)
 {
-	struct btrfs_key key;
 	struct btrfs_root *root = fs_info->tree_root;
 	struct btrfs_trans_handle *trans;
 	int ret = 0;
@@ -3594,11 +3312,7 @@ static int delete_block_group_cache(struct btrfs_fs_info *fs_info,
 	if (inode)
 		goto truncate;
 
-	key.objectid = ino;
-	key.type = BTRFS_INODE_ITEM_KEY;
-	key.offset = 0;
-
-	inode = btrfs_iget(fs_info->sb, &key, root);
+	inode = btrfs_iget(fs_info->sb, ino, root);
 	if (IS_ERR(inode))
 		return -ENOENT;
 
@@ -3624,172 +3338,45 @@ out:
 }
 
 /*
- * helper to add tree blocks for backref of type BTRFS_EXTENT_DATA_REF_KEY
- * this function scans fs tree to find blocks reference the data extent
+ * Locate the free space cache EXTENT_DATA in root tree leaf and delete the
+ * cache inode, to avoid free space cache data extent blocking data relocation.
  */
-static int find_data_references(struct reloc_control *rc,
-				struct btrfs_key *extent_key,
-				struct extent_buffer *leaf,
-				struct btrfs_extent_data_ref *ref,
-				struct rb_root *blocks)
+static int delete_v1_space_cache(struct extent_buffer *leaf,
+				 struct btrfs_block_group *block_group,
+				 u64 data_bytenr)
 {
-	struct btrfs_fs_info *fs_info = rc->extent_root->fs_info;
-	struct btrfs_path *path;
-	struct tree_block *block;
-	struct btrfs_root *root;
-	struct btrfs_file_extent_item *fi;
-	struct rb_node *rb_node;
+	u64 space_cache_ino;
+	struct btrfs_file_extent_item *ei;
 	struct btrfs_key key;
-	u64 ref_root;
-	u64 ref_objectid;
-	u64 ref_offset;
-	u32 ref_count;
-	u32 nritems;
-	int err = 0;
-	int added = 0;
-	int counted;
+	bool found = false;
+	int i;
 	int ret;
 
-	ref_root = btrfs_extent_data_ref_root(leaf, ref);
-	ref_objectid = btrfs_extent_data_ref_objectid(leaf, ref);
-	ref_offset = btrfs_extent_data_ref_offset(leaf, ref);
-	ref_count = btrfs_extent_data_ref_count(leaf, ref);
-
-	/*
-	 * This is an extent belonging to the free space cache, lets just delete
-	 * it and redo the search.
-	 */
-	if (ref_root == BTRFS_ROOT_TREE_OBJECTID) {
-		ret = delete_block_group_cache(fs_info, rc->block_group,
-					       NULL, ref_objectid);
-		if (ret != -ENOENT)
-			return ret;
-		ret = 0;
-	}
-
-	path = btrfs_alloc_path();
-	if (!path)
-		return -ENOMEM;
-	path->reada = READA_FORWARD;
-
-	root = read_fs_root(fs_info, ref_root);
-	if (IS_ERR(root)) {
-		err = PTR_ERR(root);
-		goto out;
-	}
-
-	key.objectid = ref_objectid;
-	key.type = BTRFS_EXTENT_DATA_KEY;
-	if (ref_offset > ((u64)-1 << 32))
-		key.offset = 0;
-	else
-		key.offset = ref_offset;
-
-	path->search_commit_root = 1;
-	path->skip_locking = 1;
-	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
-	if (ret < 0) {
-		err = ret;
-		goto out;
-	}
-
-	leaf = path->nodes[0];
-	nritems = btrfs_header_nritems(leaf);
-	/*
-	 * the references in tree blocks that use full backrefs
-	 * are not counted in
-	 */
-	if (block_use_full_backref(rc, leaf))
-		counted = 0;
-	else
-		counted = 1;
-	rb_node = tree_search(blocks, leaf->start);
-	if (rb_node) {
-		if (counted)
-			added = 1;
-		else
-			path->slots[0] = nritems;
-	}
-
-	while (ref_count > 0) {
-		while (path->slots[0] >= nritems) {
-			ret = btrfs_next_leaf(root, path);
-			if (ret < 0) {
-				err = ret;
-				goto out;
-			}
-			if (WARN_ON(ret > 0))
-				goto out;
-
-			leaf = path->nodes[0];
-			nritems = btrfs_header_nritems(leaf);
-			added = 0;
+	if (btrfs_header_owner(leaf) != BTRFS_ROOT_TREE_OBJECTID)
+		return 0;
 
-			if (block_use_full_backref(rc, leaf))
-				counted = 0;
-			else
-				counted = 1;
-			rb_node = tree_search(blocks, leaf->start);
-			if (rb_node) {
-				if (counted)
-					added = 1;
-				else
-					path->slots[0] = nritems;
-			}
-		}
+	for (i = 0; i < btrfs_header_nritems(leaf); i++) {
+		u8 type;
 
-		btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
-		if (WARN_ON(key.objectid != ref_objectid ||
-		    key.type != BTRFS_EXTENT_DATA_KEY))
+		btrfs_item_key_to_cpu(leaf, &key, i);
+		if (key.type != BTRFS_EXTENT_DATA_KEY)
+			continue;
+		ei = btrfs_item_ptr(leaf, i, struct btrfs_file_extent_item);
+		type = btrfs_file_extent_type(leaf, ei);
+
+		if ((type == BTRFS_FILE_EXTENT_REG ||
+		     type == BTRFS_FILE_EXTENT_PREALLOC) &&
+		    btrfs_file_extent_disk_bytenr(leaf, ei) == data_bytenr) {
+			found = true;
+			space_cache_ino = key.objectid;
 			break;
-
-		fi = btrfs_item_ptr(leaf, path->slots[0],
-				    struct btrfs_file_extent_item);
-
-		if (btrfs_file_extent_type(leaf, fi) ==
-		    BTRFS_FILE_EXTENT_INLINE)
-			goto next;
-
-		if (btrfs_file_extent_disk_bytenr(leaf, fi) !=
-		    extent_key->objectid)
-			goto next;
-
-		key.offset -= btrfs_file_extent_offset(leaf, fi);
-		if (key.offset != ref_offset)
-			goto next;
-
-		if (counted)
-			ref_count--;
-		if (added)
-			goto next;
-
-		if (!tree_block_processed(leaf->start, rc)) {
-			block = kmalloc(sizeof(*block), GFP_NOFS);
-			if (!block) {
-				err = -ENOMEM;
-				break;
-			}
-			block->bytenr = leaf->start;
-			btrfs_item_key_to_cpu(leaf, &block->key, 0);
-			block->level = 0;
-			block->key_ready = 1;
-			rb_node = tree_insert(blocks, block->bytenr,
-					      &block->rb_node);
-			if (rb_node)
-				backref_tree_panic(rb_node, -EEXIST,
-						   block->bytenr);
 		}
-		if (counted)
-			added = 1;
-		else
-			path->slots[0] = nritems;
-next:
-		path->slots[0]++;
-
 	}
-out:
-	btrfs_free_path(path);
-	return err;
+	if (!found)
+		return -ENOENT;
+	ret = delete_block_group_cache(leaf->fs_info, block_group, NULL,
+					space_cache_ino);
+	return ret;
 }
 
 /*
@@ -3801,91 +3388,41 @@ int add_data_references(struct reloc_control *rc,
 			struct btrfs_path *path,
 			struct rb_root *blocks)
 {
-	struct btrfs_key key;
-	struct extent_buffer *eb;
-	struct btrfs_extent_data_ref *dref;
-	struct btrfs_extent_inline_ref *iref;
-	unsigned long ptr;
-	unsigned long end;
-	u32 blocksize = rc->extent_root->fs_info->nodesize;
+	struct btrfs_fs_info *fs_info = rc->extent_root->fs_info;
+	struct ulist *leaves = NULL;
+	struct ulist_iterator leaf_uiter;
+	struct ulist_node *ref_node = NULL;
+	const u32 blocksize = fs_info->nodesize;
 	int ret = 0;
-	int err = 0;
 
-	eb = path->nodes[0];
-	ptr = btrfs_item_ptr_offset(eb, path->slots[0]);
-	end = ptr + btrfs_item_size_nr(eb, path->slots[0]);
-	ptr += sizeof(struct btrfs_extent_item);
-
-	while (ptr < end) {
-		iref = (struct btrfs_extent_inline_ref *)ptr;
-		key.type = btrfs_get_extent_inline_ref_type(eb, iref,
-							BTRFS_REF_TYPE_DATA);
-		if (key.type == BTRFS_SHARED_DATA_REF_KEY) {
-			key.offset = btrfs_extent_inline_ref_offset(eb, iref);
-			ret = __add_tree_block(rc, key.offset, blocksize,
-					       blocks);
-		} else if (key.type == BTRFS_EXTENT_DATA_REF_KEY) {
-			dref = (struct btrfs_extent_data_ref *)(&iref->offset);
-			ret = find_data_references(rc, extent_key,
-						   eb, dref, blocks);
-		} else {
-			ret = -EUCLEAN;
-			btrfs_err(rc->extent_root->fs_info,
-		     "extent %llu slot %d has an invalid inline ref type",
-			     eb->start, path->slots[0]);
-		}
-		if (ret) {
-			err = ret;
-			goto out;
-		}
-		ptr += btrfs_extent_inline_ref_size(key.type);
-	}
-	WARN_ON(ptr > end);
+	btrfs_release_path(path);
+	ret = btrfs_find_all_leafs(NULL, fs_info, extent_key->objectid,
+				   0, &leaves, NULL, true);
+	if (ret < 0)
+		return ret;
 
-	while (1) {
-		cond_resched();
-		eb = path->nodes[0];
-		if (path->slots[0] >= btrfs_header_nritems(eb)) {
-			ret = btrfs_next_leaf(rc->extent_root, path);
-			if (ret < 0) {
-				err = ret;
-				break;
-			}
-			if (ret > 0)
-				break;
-			eb = path->nodes[0];
-		}
+	ULIST_ITER_INIT(&leaf_uiter);
+	while ((ref_node = ulist_next(leaves, &leaf_uiter))) {
+		struct extent_buffer *eb;
 
-		btrfs_item_key_to_cpu(eb, &key, path->slots[0]);
-		if (key.objectid != extent_key->objectid)
+		eb = read_tree_block(fs_info, ref_node->val, 0, 0, 0, NULL);
+		if (IS_ERR(eb)) {
+			ret = PTR_ERR(eb);
 			break;
-
-		if (key.type == BTRFS_SHARED_DATA_REF_KEY) {
-			ret = __add_tree_block(rc, key.offset, blocksize,
-					       blocks);
-		} else if (key.type == BTRFS_EXTENT_DATA_REF_KEY) {
-			dref = btrfs_item_ptr(eb, path->slots[0],
-					      struct btrfs_extent_data_ref);
-			ret = find_data_references(rc, extent_key,
-						   eb, dref, blocks);
-		} else if (unlikely(key.type == BTRFS_EXTENT_REF_V0_KEY)) {
-			btrfs_print_v0_err(eb->fs_info);
-			btrfs_handle_fs_error(eb->fs_info, -EINVAL, NULL);
-			ret = -EINVAL;
-		} else {
-			ret = 0;
 		}
-		if (ret) {
-			err = ret;
+		ret = delete_v1_space_cache(eb, rc->block_group,
+					    extent_key->objectid);
+		free_extent_buffer(eb);
+		if (ret < 0)
+			break;
+		ret = __add_tree_block(rc, ref_node->val, blocksize, blocks);
+		if (ret < 0)
 			break;
-		}
-		path->slots[0]++;
 	}
-out:
-	btrfs_release_path(path);
-	if (err)
+	if (ret < 0)
 		free_block_list(blocks);
-	return err;
+	ulist_free(leaves);
+	return ret;
 }
 
 /*
@@ -3992,20 +3529,6 @@ static void unset_reloc_control(struct reloc_control *rc)
 	mutex_unlock(&fs_info->reloc_mutex);
 }
 
-static int check_extent_flags(u64 flags)
-{
-	if ((flags & BTRFS_EXTENT_FLAG_DATA) &&
-	    (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK))
-		return 1;
-	if (!(flags & BTRFS_EXTENT_FLAG_DATA) &&
-	    !(flags & BTRFS_EXTENT_FLAG_TREE_BLOCK))
-		return 1;
-	if ((flags & BTRFS_EXTENT_FLAG_DATA) &&
-	    (flags & BTRFS_BLOCK_FLAG_FULL_BACKREF))
-		return 1;
-	return 0;
-}
-
 static noinline_for_stack
 int prepare_to_relocate(struct reloc_control *rc)
 {
@@ -4025,7 +3548,7 @@ int prepare_to_relocate(struct reloc_control *rc)
 	rc->reserved_bytes = 0;
 	rc->block_rsv->size = rc->extent_root->fs_info->nodesize *
 			      RELOCATION_RESERVED_NODES;
-	ret = btrfs_block_rsv_refill(rc->extent_root,
+	ret = btrfs_block_rsv_refill(rc->extent_root->fs_info,
 				     rc->block_rsv, rc->block_rsv->size,
 				     BTRFS_RESERVE_FLUSH_ALL);
 	if (ret)
@@ -4044,8 +3567,12 @@ int prepare_to_relocate(struct reloc_control *rc)
 		 */
 		return PTR_ERR(trans);
 	}
-	btrfs_commit_transaction(trans);
-	return 0;
+
+	ret = btrfs_commit_transaction(trans);
+	if (ret)
+		unset_reloc_control(rc);
+
+	return ret;
 }
 
 static noinline_for_stack int relocate_block_group(struct reloc_control *rc)
@@ -4057,7 +3584,6 @@ static noinline_for_stack int relocate_block_group(struct reloc_control *rc)
 	struct btrfs_path *path;
 	struct btrfs_extent_item *ei;
 	u64 flags;
-	u32 item_size;
 	int ret;
 	int err = 0;
 	int progress = 0;
@@ -4075,9 +3601,9 @@ static noinline_for_stack int relocate_block_group(struct reloc_control *rc)
 
 	while (1) {
 		rc->reserved_bytes = 0;
-		ret = btrfs_block_rsv_refill(rc->extent_root,
-					rc->block_rsv, rc->block_rsv->size,
-					BTRFS_RESERVE_FLUSH_ALL);
+		ret = btrfs_block_rsv_refill(fs_info, rc->block_rsv,
+					     rc->block_rsv->size,
+					     BTRFS_RESERVE_FLUSH_ALL);
 		if (ret) {
 			err = ret;
 			break;
@@ -4106,19 +3632,7 @@ restart:
 
 		ei = btrfs_item_ptr(path->nodes[0], path->slots[0],
 				    struct btrfs_extent_item);
-		item_size = btrfs_item_size_nr(path->nodes[0], path->slots[0]);
-		if (item_size >= sizeof(*ei)) {
-			flags = btrfs_extent_flags(path->nodes[0], ei);
-			ret = check_extent_flags(flags);
-			BUG_ON(ret);
-		} else if (unlikely(item_size == sizeof(struct btrfs_extent_item_v0))) {
-			err = -EINVAL;
-			btrfs_print_v0_err(trans->fs_info);
-			btrfs_abort_transaction(trans, err);
-			break;
-		} else {
-			BUG();
-		}
+		flags = btrfs_extent_flags(path->nodes[0], ei);
 
 		if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
 			ret = add_tree_block(rc, &key, path, &blocks);
@@ -4137,12 +3651,6 @@ restart:
 		if (!RB_EMPTY_ROOT(&blocks)) {
 			ret = relocate_tree_blocks(trans, rc, &blocks);
 			if (ret < 0) {
-				/*
-				 * if we fail to relocate tree blocks, force to update
-				 * backref cache when committing transaction.
-				 */
-				rc->backref_cache.last_trans = trans->transid - 1;
-
 				if (ret != -EAGAIN) {
 					err = ret;
 					break;
@@ -4166,6 +3674,10 @@ restart:
 				break;
 			}
 		}
+		if (btrfs_should_cancel_balance(fs_info)) {
+			err = -ECANCELED;
+			break;
+		}
 	}
 	if (trans && progress && err == -ENOSPC) {
 		ret = btrfs_force_chunk_alloc(trans, rc->block_group->flags);
@@ -4194,16 +3706,24 @@ restart:
 	rc->create_reloc_tree = 0;
 	set_reloc_control(rc);
 
-	backref_cache_cleanup(&rc->backref_cache);
-	btrfs_block_rsv_release(fs_info, rc->block_rsv, (u64)-1);
+	btrfs_backref_release_cache(&rc->backref_cache);
+	btrfs_block_rsv_release(fs_info, rc->block_rsv, (u64)-1, NULL);
 
+	/*
+	 * Even in the case when the relocation is cancelled, we should all go
+	 * through prepare_to_merge() and merge_reloc_roots().
+	 *
+	 * For error (including cancelled balance), prepare_to_merge() will
+	 * mark all reloc trees orphan, then queue them for cleanup in
+	 * merge_reloc_roots()
+	 */
 	err = prepare_to_merge(rc, err);
 
 	merge_reloc_roots(rc);
 
 	rc->merge_reloc_tree = 0;
 	unset_reloc_control(rc);
-	btrfs_block_rsv_release(fs_info, rc->block_rsv, (u64)-1);
+	btrfs_block_rsv_release(fs_info, rc->block_rsv, (u64)-1, NULL);
 
 	/* get rid of pinned extents */
 	trans = btrfs_join_transaction(rc->extent_root);
@@ -4211,11 +3731,13 @@ restart:
 		err = PTR_ERR(trans);
 		goto out_free;
 	}
-	btrfs_commit_transaction(trans);
+	ret = btrfs_commit_transaction(trans);
+	if (ret && !err)
+		err = ret;
+out_free:
 	ret = clean_dirty_subvols(rc);
 	if (ret < 0 && !err)
 		err = ret;
-out_free:
 	btrfs_free_block_rsv(fs_info, rc->block_rsv);
 	btrfs_free_path(path);
 	return err;
@@ -4251,6 +3773,35 @@ out:
 	return ret;
 }
 
+static void delete_orphan_inode(struct btrfs_trans_handle *trans,
+				struct btrfs_root *root, u64 objectid)
+{
+	struct btrfs_path *path;
+	struct btrfs_key key;
+	int ret = 0;
+
+	path = btrfs_alloc_path();
+	if (!path) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	key.objectid = objectid;
+	key.type = BTRFS_INODE_ITEM_KEY;
+	key.offset = 0;
+	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
+	if (ret) {
+		if (ret > 0)
+			ret = -ENOENT;
+		goto out;
+	}
+	ret = btrfs_del_item(trans, root, path);
+out:
+	if (ret)
+		btrfs_abort_transaction(trans, ret);
+	btrfs_free_path(path);
+}
+
 /*
  * helper to create inode for data relocation.
  * the inode is in data relocation tree and its link count is 0
@@ -4262,44 +3813,86 @@ struct inode *create_reloc_inode(struct btrfs_fs_info *fs_info,
 	struct inode *inode = NULL;
 	struct btrfs_trans_handle *trans;
 	struct btrfs_root *root;
-	struct btrfs_key key;
 	u64 objectid;
 	int err = 0;
 
-	root = read_fs_root(fs_info, BTRFS_DATA_RELOC_TREE_OBJECTID);
-	if (IS_ERR(root))
-		return ERR_CAST(root);
-
+	root = btrfs_grab_root(fs_info->data_reloc_root);
 	trans = btrfs_start_transaction(root, 6);
-	if (IS_ERR(trans))
+	if (IS_ERR(trans)) {
+		btrfs_put_root(root);
 		return ERR_CAST(trans);
+	}
 
-	err = btrfs_find_free_objectid(root, &objectid);
+	err = btrfs_get_free_objectid(root, &objectid);
 	if (err)
 		goto out;
 
 	err = __insert_orphan_inode(trans, root, objectid);
-	BUG_ON(err);
+	if (err)
+		goto out;
 
-	key.objectid = objectid;
-	key.type = BTRFS_INODE_ITEM_KEY;
-	key.offset = 0;
-	inode = btrfs_iget(fs_info->sb, &key, root);
-	BUG_ON(IS_ERR(inode));
+	inode = btrfs_iget(fs_info->sb, objectid, root);
+	if (IS_ERR(inode)) {
+		delete_orphan_inode(trans, root, objectid);
+		err = PTR_ERR(inode);
+		inode = NULL;
+		goto out;
+	}
 	BTRFS_I(inode)->index_cnt = group->start;
 
 	err = btrfs_orphan_add(trans, BTRFS_I(inode));
 out:
+	btrfs_put_root(root);
 	btrfs_end_transaction(trans);
 	btrfs_btree_balance_dirty(fs_info);
 	if (err) {
-		if (inode)
-			iput(inode);
+		iput(inode);
 		inode = ERR_PTR(err);
 	}
 	return inode;
 }
 
+/*
+ * Mark start of chunk relocation that is cancellable. Check if the cancellation
+ * has been requested meanwhile and don't start in that case.
+ *
+ * Return:
+ *   0             success
+ *   -EINPROGRESS  operation is already in progress, that's probably a bug
+ *   -ECANCELED    cancellation request was set before the operation started
+ */
+static int reloc_chunk_start(struct btrfs_fs_info *fs_info)
+{
+	if (test_and_set_bit(BTRFS_FS_RELOC_RUNNING, &fs_info->flags)) {
+		/* This should not happen */
+		btrfs_err(fs_info, "reloc already running, cannot start");
+		return -EINPROGRESS;
+	}
+
+	if (atomic_read(&fs_info->reloc_cancel_req) > 0) {
+		btrfs_info(fs_info, "chunk relocation canceled on start");
+		/*
+		 * On cancel, clear all requests but let the caller mark
+		 * the end after cleanup operations.
+		 */
+		atomic_set(&fs_info->reloc_cancel_req, 0);
+		return -ECANCELED;
+	}
+	return 0;
+}
+
+/*
+ * Mark end of chunk relocation that is cancellable and wake any waiters.
+ */
+static void reloc_chunk_end(struct btrfs_fs_info *fs_info)
+{
+	/* Requested after start, clear bit first so any waiters can continue */
+	if (atomic_read(&fs_info->reloc_cancel_req) > 0)
+		btrfs_info(fs_info, "chunk relocation canceled during operation");
+	clear_and_wake_up_bit(BTRFS_FS_RELOC_RUNNING, &fs_info->flags);
+	atomic_set(&fs_info->reloc_cancel_req, 0);
+}
+
 static struct reloc_control *alloc_reloc_control(struct btrfs_fs_info *fs_info)
 {
 	struct reloc_control *rc;
@@ -4310,13 +3903,25 @@ static struct reloc_control *alloc_reloc_control(struct btrfs_fs_info *fs_info)
 
 	INIT_LIST_HEAD(&rc->reloc_roots);
 	INIT_LIST_HEAD(&rc->dirty_subvol_roots);
-	backref_cache_init(&rc->backref_cache);
+	btrfs_backref_init_cache(fs_info, &rc->backref_cache, 1);
 	mapping_tree_init(&rc->reloc_root_tree);
 	extent_io_tree_init(fs_info, &rc->processed_blocks,
 			    IO_TREE_RELOC_BLOCKS, NULL);
 	return rc;
 }
 
+static void free_reloc_control(struct reloc_control *rc)
+{
+	struct mapping_node *node, *tmp;
+
+	free_reloc_roots(&rc->reloc_roots);
+	rbtree_postorder_for_each_entry_safe(node, tmp,
+			&rc->reloc_root_tree.rb_root, rb_node)
+		kfree(node);
+
+	kfree(rc);
+}
+
 /*
  * Print the block group being relocated
  */
@@ -4347,7 +3952,7 @@ static const char *stage_to_string(int stage)
 int btrfs_relocate_block_group(struct btrfs_fs_info *fs_info, u64 group_start)
 {
 	struct btrfs_block_group *bg;
-	struct btrfs_root *extent_root = fs_info->extent_root;
+	struct btrfs_root *extent_root = btrfs_extent_root(fs_info, group_start);
 	struct reloc_control *rc;
 	struct inode *inode;
 	struct btrfs_path *path;
@@ -4355,10 +3960,34 @@ int btrfs_relocate_block_group(struct btrfs_fs_info *fs_info, u64 group_start)
 	int rw = 0;
 	int err = 0;
 
+	/*
+	 * This only gets set if we had a half-deleted snapshot on mount.  We
+	 * cannot allow relocation to start while we're still trying to clean up
+	 * these pending deletions.
+	 */
+	ret = wait_on_bit(&fs_info->flags, BTRFS_FS_UNFINISHED_DROPS, TASK_INTERRUPTIBLE);
+	if (ret)
+		return ret;
+
+	/* We may have been woken up by close_ctree, so bail if we're closing. */
+	if (btrfs_fs_closing(fs_info))
+		return -EINTR;
+
 	bg = btrfs_lookup_block_group(fs_info, group_start);
 	if (!bg)
 		return -ENOENT;
 
+	/*
+	 * Relocation of a data block group creates ordered extents.  Without
+	 * sb_start_write(), we can freeze the filesystem while unfinished
+	 * ordered extents are left. Such ordered extents can cause a deadlock
+	 * e.g. when syncfs() is waiting for their completion but they can't
+	 * finish because they block when joining a transaction, due to the
+	 * fact that the freeze locks are being held in write mode.
+	 */
+	if (bg->flags & BTRFS_BLOCK_GROUP_DATA)
+		ASSERT(sb_write_started(fs_info->sb));
+
 	if (btrfs_pinned_by_swapfile(fs_info, bg)) {
 		btrfs_put_block_group(bg);
 		return -ETXTBSY;
@@ -4370,6 +3999,12 @@ int btrfs_relocate_block_group(struct btrfs_fs_info *fs_info, u64 group_start)
 		return -ENOMEM;
 	}
 
+	ret = reloc_chunk_start(fs_info);
+	if (ret < 0) {
+		err = ret;
+		goto out_put_bg;
+	}
+
 	rc->extent_root = extent_root;
 	rc->block_group = bg;
 
@@ -4414,6 +4049,9 @@ int btrfs_relocate_block_group(struct btrfs_fs_info *fs_info, u64 group_start)
 				 rc->block_group->start,
 				 rc->block_group->length);
 
+	ret = btrfs_zone_finish(rc->block_group);
+	WARN_ON(ret && ret != -EAGAIN);
+
 	while (1) {
 		int finishes_stage;
 
@@ -4460,8 +4098,10 @@ out:
 	if (err && rw)
 		btrfs_dec_block_group_ro(rc->block_group);
 	iput(rc->data_inode);
-	btrfs_put_block_group(rc->block_group);
-	kfree(rc);
+out_put_bg:
+	btrfs_put_block_group(bg);
+	reloc_chunk_end(fs_info);
+	free_reloc_control(rc);
 	return err;
 }
 
@@ -4477,7 +4117,7 @@ static noinline_for_stack int mark_garbage_root(struct btrfs_root *root)
 
 	memset(&root->root_item.drop_progress, 0,
 		sizeof(root->root_item.drop_progress));
-	root->root_item.drop_level = 0;
+	btrfs_set_root_drop_level(&root->root_item, 0);
 	btrfs_set_root_refs(&root->root_item, 0);
 	ret = btrfs_update_root(trans, fs_info->tree_root,
 				&root->root_key, &root->root_item);
@@ -4494,9 +4134,8 @@ static noinline_for_stack int mark_garbage_root(struct btrfs_root *root)
  * this function resumes merging reloc trees with corresponding fs trees.
  * this is important for keeping the sharing of tree blocks
  */
-int btrfs_recover_relocation(struct btrfs_root *root)
+int btrfs_recover_relocation(struct btrfs_fs_info *fs_info)
 {
-	struct btrfs_fs_info *fs_info = root->fs_info;
 	LIST_HEAD(reloc_roots);
 	struct btrfs_key key;
 	struct btrfs_root *fs_root;
@@ -4537,17 +4176,18 @@ int btrfs_recover_relocation(struct btrfs_root *root)
 		    key.type != BTRFS_ROOT_ITEM_KEY)
 			break;
 
-		reloc_root = btrfs_read_fs_root(root, &key);
+		reloc_root = btrfs_read_tree_root(fs_info->tree_root, &key);
 		if (IS_ERR(reloc_root)) {
 			err = PTR_ERR(reloc_root);
 			goto out;
 		}
 
+		set_bit(BTRFS_ROOT_SHAREABLE, &reloc_root->state);
 		list_add(&reloc_root->root_list, &reloc_roots);
 
 		if (btrfs_root_refs(&reloc_root->root_item) > 0) {
-			fs_root = read_fs_root(fs_info,
-					       reloc_root->root_key.offset);
+			fs_root = btrfs_get_fs_root(fs_info,
+					reloc_root->root_key.offset, false);
 			if (IS_ERR(fs_root)) {
 				ret = PTR_ERR(fs_root);
 				if (ret != -ENOENT) {
@@ -4559,6 +4199,8 @@ int btrfs_recover_relocation(struct btrfs_root *root)
 					err = ret;
 					goto out;
 				}
+			} else {
+				btrfs_put_root(fs_root);
 			}
 		}
 
@@ -4578,15 +4220,20 @@ int btrfs_recover_relocation(struct btrfs_root *root)
 		goto out;
 	}
 
-	rc->extent_root = fs_info->extent_root;
+	ret = reloc_chunk_start(fs_info);
+	if (ret < 0) {
+		err = ret;
+		goto out_end;
+	}
+
+	rc->extent_root = btrfs_extent_root(fs_info, 0);
 
 	set_reloc_control(rc);
 
 	trans = btrfs_join_transaction(rc->extent_root);
 	if (IS_ERR(trans)) {
-		unset_reloc_control(rc);
 		err = PTR_ERR(trans);
-		goto out_free;
+		goto out_unset;
 	}
 
 	rc->merge_reloc_tree = 1;
@@ -4602,21 +4249,30 @@ int btrfs_recover_relocation(struct btrfs_root *root)
 			continue;
 		}
 
-		fs_root = read_fs_root(fs_info, reloc_root->root_key.offset);
+		fs_root = btrfs_get_fs_root(fs_info, reloc_root->root_key.offset,
+					    false);
 		if (IS_ERR(fs_root)) {
 			err = PTR_ERR(fs_root);
 			list_add_tail(&reloc_root->root_list, &reloc_roots);
-			goto out_free;
+			btrfs_end_transaction(trans);
+			goto out_unset;
 		}
 
 		err = __add_reloc_root(reloc_root);
-		BUG_ON(err < 0); /* -ENOMEM or logic error */
-		fs_root->reloc_root = reloc_root;
+		ASSERT(err != -EEXIST);
+		if (err) {
+			list_add_tail(&reloc_root->root_list, &reloc_roots);
+			btrfs_put_root(fs_root);
+			btrfs_end_transaction(trans);
+			goto out_unset;
+		}
+		fs_root->reloc_root = btrfs_grab_root(reloc_root);
+		btrfs_put_root(fs_root);
 	}
 
 	err = btrfs_commit_transaction(trans);
 	if (err)
-		goto out_free;
+		goto out_unset;
 
 	merge_reloc_roots(rc);
 
@@ -4625,28 +4281,29 @@ int btrfs_recover_relocation(struct btrfs_root *root)
 	trans = btrfs_join_transaction(rc->extent_root);
 	if (IS_ERR(trans)) {
 		err = PTR_ERR(trans);
-		goto out_free;
+		goto out_clean;
 	}
 	err = btrfs_commit_transaction(trans);
-
+out_clean:
 	ret = clean_dirty_subvols(rc);
 	if (ret < 0 && !err)
 		err = ret;
-out_free:
-	kfree(rc);
+out_unset:
+	unset_reloc_control(rc);
+out_end:
+	reloc_chunk_end(fs_info);
+	free_reloc_control(rc);
 out:
-	if (!list_empty(&reloc_roots))
-		free_reloc_roots(&reloc_roots);
+	free_reloc_roots(&reloc_roots);
 
 	btrfs_free_path(path);
 
 	if (err == 0) {
 		/* cleanup orphan inode in data relocation tree */
-		fs_root = read_fs_root(fs_info, BTRFS_DATA_RELOC_TREE_OBJECTID);
-		if (IS_ERR(fs_root))
-			err = PTR_ERR(fs_root);
-		else
-			err = btrfs_orphan_cleanup(fs_root);
+		fs_root = btrfs_grab_root(fs_info->data_reloc_root);
+		ASSERT(fs_root);
+		err = btrfs_orphan_cleanup(fs_root);
+		btrfs_put_root(fs_root);
 	}
 	return err;
 }
@@ -4657,9 +4314,10 @@ out:
  * cloning checksum properly handles the nodatasum extents.
  * it also saves CPU time to re-calculate the checksum.
  */
-int btrfs_reloc_clone_csums(struct inode *inode, u64 file_pos, u64 len)
+int btrfs_reloc_clone_csums(struct btrfs_inode *inode, u64 file_pos, u64 len)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
+	struct btrfs_root *csum_root;
 	struct btrfs_ordered_sum *sums;
 	struct btrfs_ordered_extent *ordered;
 	int ret;
@@ -4670,9 +4328,10 @@ int btrfs_reloc_clone_csums(struct inode *inode, u64 file_pos, u64 len)
 	ordered = btrfs_lookup_ordered_extent(inode, file_pos);
 	BUG_ON(ordered->file_offset != file_pos || ordered->num_bytes != len);
 
-	disk_bytenr = file_pos + BTRFS_I(inode)->index_cnt;
-	ret = btrfs_lookup_csums_range(fs_info->csum_root, disk_bytenr,
-				       disk_bytenr + len - 1, &list, 0);
+	disk_bytenr = file_pos + inode->index_cnt;
+	csum_root = btrfs_csum_root(fs_info, disk_bytenr);
+	ret = btrfs_lookup_csums_range(csum_root, disk_bytenr,
+				       disk_bytenr + len - 1, &list, 0, false);
 	if (ret)
 		goto out;
 
@@ -4708,7 +4367,7 @@ int btrfs_reloc_cow_block(struct btrfs_trans_handle *trans,
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct reloc_control *rc;
-	struct backref_node *node;
+	struct btrfs_backref_node *node;
 	int first_cow = 0;
 	int level;
 	int ret = 0;
@@ -4717,13 +4376,7 @@ int btrfs_reloc_cow_block(struct btrfs_trans_handle *trans,
 	if (!rc)
 		return 0;
 
-	BUG_ON(rc->stage == UPDATE_DATA_PTRS &&
-	       root->root_key.objectid == BTRFS_DATA_RELOC_TREE_OBJECTID);
-
-	if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID) {
-		if (buf == root->node)
-			__update_reloc_root(root, cow->start);
-	}
+	BUG_ON(rc->stage == UPDATE_DATA_PTRS && btrfs_is_data_reloc_root(root));
 
 	level = btrfs_header_level(buf);
 	if (btrfs_header_generation(buf) <=
@@ -4738,7 +4391,7 @@ int btrfs_reloc_cow_block(struct btrfs_trans_handle *trans,
 		BUG_ON(node->bytenr != buf->start &&
 		       node->new_bytenr != buf->start);
 
-		drop_node_buffer(node);
+		btrfs_backref_drop_node_buffer(node);
 		atomic_inc(&cow->refs);
 		node->eb = cow;
 		node->new_bytenr = cow->start;
@@ -4750,7 +4403,7 @@ int btrfs_reloc_cow_block(struct btrfs_trans_handle *trans,
 		}
 
 		if (first_cow)
-			__mark_block_processed(rc, node);
+			mark_block_processed(rc, node);
 
 		if (first_cow && level > 0)
 			rc->nodes_relocated += buf->len;
@@ -4795,6 +4448,10 @@ void btrfs_reloc_pre_snapshot(struct btrfs_pending_snapshot *pending,
 /*
  * called after snapshot is created. migrate block reservation
  * and create reloc root for the newly created snapshot
+ *
+ * This is similar to btrfs_init_reloc_root(), we come out of here with two
+ * references held on the reloc_root, one for root->reloc_root and one for
+ * rc->reloc_roots.
  */
 int btrfs_reloc_post_snapshot(struct btrfs_trans_handle *trans,
 			       struct btrfs_pending_snapshot *pending)
@@ -4826,8 +4483,13 @@ int btrfs_reloc_post_snapshot(struct btrfs_trans_handle *trans,
 		return PTR_ERR(reloc_root);
 
 	ret = __add_reloc_root(reloc_root);
-	BUG_ON(ret < 0);
-	new_root->reloc_root = reloc_root;
+	ASSERT(ret != -EEXIST);
+	if (ret) {
+		/* Pairs with create_reloc_root */
+		btrfs_put_root(reloc_root);
+		return ret;
+	}
+	new_root->reloc_root = btrfs_grab_root(reloc_root);
 
 	if (rc->create_reloc_tree)
 		ret = clone_backref_node(trans, rc, root, reloc_root);
diff --git a/fs/btrfs/root-tree.c b/fs/btrfs/root-tree.c
index 612411c74550..e1f599d7a916 100644
--- a/fs/btrfs/root-tree.c
+++ b/fs/btrfs/root-tree.c
@@ -22,11 +22,10 @@
 static void btrfs_read_root_item(struct extent_buffer *eb, int slot,
 				struct btrfs_root_item *item)
 {
-	uuid_le uuid;
 	u32 len;
 	int need_reset = 0;
 
-	len = btrfs_item_size_nr(eb, slot);
+	len = btrfs_item_size(eb, slot);
 	read_extent_buffer(eb, item, btrfs_item_ptr_offset(eb, slot),
 			   min_t(u32, len, sizeof(*item)));
 	if (len < sizeof(*item))
@@ -40,12 +39,9 @@ static void btrfs_read_root_item(struct extent_buffer *eb, int slot,
 		need_reset = 1;
 	}
 	if (need_reset) {
-		memset(&item->generation_v2, 0,
-			sizeof(*item) - offsetof(struct btrfs_root_item,
-					generation_v2));
-
-		uuid_le_gen(&uuid);
-		memcpy(item->uuid, uuid.b, BTRFS_UUID_SIZE);
+		/* Clear all members from generation_v2 onwards. */
+		memset_startat(item, 0, generation_v2);
+		generate_random_guid(item->uuid);
 	}
 }
 
@@ -150,7 +146,7 @@ int btrfs_update_root(struct btrfs_trans_handle *trans, struct btrfs_root
 	l = path->nodes[0];
 	slot = path->slots[0];
 	ptr = btrfs_item_ptr_offset(l, slot);
-	old_len = btrfs_item_size_nr(l, slot);
+	old_len = btrfs_item_size(l, slot);
 
 	/*
 	 * If this is the first time we update the root item which originated
@@ -212,7 +208,6 @@ int btrfs_find_orphan_roots(struct btrfs_fs_info *fs_info)
 	struct extent_buffer *leaf;
 	struct btrfs_path *path;
 	struct btrfs_key key;
-	struct btrfs_key root_key;
 	struct btrfs_root *root;
 	int err = 0;
 	int ret;
@@ -225,10 +220,9 @@ int btrfs_find_orphan_roots(struct btrfs_fs_info *fs_info)
 	key.type = BTRFS_ORPHAN_ITEM_KEY;
 	key.offset = 0;
 
-	root_key.type = BTRFS_ROOT_ITEM_KEY;
-	root_key.offset = (u64)-1;
-
 	while (1) {
+		u64 root_objectid;
+
 		ret = btrfs_search_slot(NULL, tree_root, &key, path, 0, 0);
 		if (ret < 0) {
 			err = ret;
@@ -252,28 +246,10 @@ int btrfs_find_orphan_roots(struct btrfs_fs_info *fs_info)
 		    key.type != BTRFS_ORPHAN_ITEM_KEY)
 			break;
 
-		root_key.objectid = key.offset;
+		root_objectid = key.offset;
 		key.offset++;
 
-		/*
-		 * The root might have been inserted already, as before we look
-		 * for orphan roots, log replay might have happened, which
-		 * triggers a transaction commit and qgroup accounting, which
-		 * in turn reads and inserts fs roots while doing backref
-		 * walking.
-		 */
-		root = btrfs_lookup_fs_root(fs_info, root_key.objectid);
-		if (root) {
-			WARN_ON(!test_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED,
-					  &root->state));
-			if (btrfs_root_refs(&root->root_item) == 0) {
-				set_bit(BTRFS_ROOT_DEAD_TREE, &root->state);
-				btrfs_add_dead_root(root);
-			}
-			continue;
-		}
-
-		root = btrfs_read_fs_root(tree_root, &root_key);
+		root = btrfs_get_fs_root(fs_info, root_objectid, false);
 		err = PTR_ERR_OR_ZERO(root);
 		if (err && err != -ENOENT) {
 			break;
@@ -290,7 +266,7 @@ int btrfs_find_orphan_roots(struct btrfs_fs_info *fs_info)
 				break;
 			}
 			err = btrfs_del_orphan_item(trans, tree_root,
-						    root_key.objectid);
+						    root_objectid);
 			btrfs_end_transaction(trans);
 			if (err) {
 				btrfs_handle_fs_error(fs_info, err,
@@ -300,25 +276,27 @@ int btrfs_find_orphan_roots(struct btrfs_fs_info *fs_info)
 			continue;
 		}
 
-		err = btrfs_init_fs_root(root);
-		if (err) {
-			btrfs_free_fs_root(root);
-			break;
-		}
-
-		set_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED, &root->state);
-
-		err = btrfs_insert_fs_root(fs_info, root);
-		if (err) {
-			BUG_ON(err == -EEXIST);
-			btrfs_free_fs_root(root);
-			break;
-		}
-
+		WARN_ON(!test_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED, &root->state));
 		if (btrfs_root_refs(&root->root_item) == 0) {
+			struct btrfs_key drop_key;
+
+			btrfs_disk_key_to_cpu(&drop_key, &root->root_item.drop_progress);
+			/*
+			 * If we have a non-zero drop_progress then we know we
+			 * made it partly through deleting this snapshot, and
+			 * thus we need to make sure we block any balance from
+			 * happening until this snapshot is completely dropped.
+			 */
+			if (drop_key.objectid != 0 || drop_key.type != 0 ||
+			    drop_key.offset != 0) {
+				set_bit(BTRFS_FS_UNFINISHED_DROPS, &fs_info->flags);
+				set_bit(BTRFS_ROOT_UNFINISHED_DROP, &root->state);
+			}
+
 			set_bit(BTRFS_ROOT_DEAD_TREE, &root->state);
 			btrfs_add_dead_root(root);
 		}
+		btrfs_put_root(root);
 	}
 
 	btrfs_free_path(path);
@@ -359,7 +337,6 @@ int btrfs_del_root_ref(struct btrfs_trans_handle *trans, u64 root_id,
 	struct extent_buffer *leaf;
 	struct btrfs_key key;
 	unsigned long ptr;
-	int err = 0;
 	int ret;
 
 	path = btrfs_alloc_path();
@@ -371,8 +348,9 @@ int btrfs_del_root_ref(struct btrfs_trans_handle *trans, u64 root_id,
 	key.offset = ref_id;
 again:
 	ret = btrfs_search_slot(trans, tree_root, &key, path, -1, 1);
-	BUG_ON(ret < 0);
-	if (ret == 0) {
+	if (ret < 0) {
+		goto out;
+	} else if (ret == 0) {
 		leaf = path->nodes[0];
 		ref = btrfs_item_ptr(leaf, path->slots[0],
 				     struct btrfs_root_ref);
@@ -380,18 +358,18 @@ again:
 		if ((btrfs_root_ref_dirid(leaf, ref) != dirid) ||
 		    (btrfs_root_ref_name_len(leaf, ref) != name_len) ||
 		    memcmp_extent_buffer(leaf, name, ptr, name_len)) {
-			err = -ENOENT;
+			ret = -ENOENT;
 			goto out;
 		}
 		*sequence = btrfs_root_ref_sequence(leaf, ref);
 
 		ret = btrfs_del_item(trans, tree_root, path);
-		if (ret) {
-			err = ret;
+		if (ret)
 			goto out;
-		}
-	} else
-		err = -ENOENT;
+	} else {
+		ret = -ENOENT;
+		goto out;
+	}
 
 	if (key.type == BTRFS_ROOT_BACKREF_KEY) {
 		btrfs_release_path(path);
@@ -403,7 +381,7 @@ again:
 
 out:
 	btrfs_free_path(path);
-	return err;
+	return ret;
 }
 
 /*
@@ -530,7 +508,8 @@ int btrfs_subvolume_reserve_metadata(struct btrfs_root *root,
 		/* One for parent inode, two for dir entries */
 		qgroup_num_bytes = 3 * fs_info->nodesize;
 		ret = btrfs_qgroup_reserve_meta_prealloc(root,
-				qgroup_num_bytes, true);
+							 qgroup_num_bytes, true,
+							 false);
 		if (ret)
 			return ret;
 	}
@@ -538,7 +517,7 @@ int btrfs_subvolume_reserve_metadata(struct btrfs_root *root,
 	num_bytes = btrfs_calc_insert_metadata_size(fs_info, items);
 	rsv->space_info = btrfs_find_space_info(fs_info,
 					    BTRFS_BLOCK_GROUP_METADATA);
-	ret = btrfs_block_rsv_add(root, rsv, num_bytes,
+	ret = btrfs_block_rsv_add(fs_info, rsv, num_bytes,
 				  BTRFS_RESERVE_FLUSH_ALL);
 
 	if (ret == -ENOSPC && use_global_rsv)
@@ -547,11 +526,20 @@ int btrfs_subvolume_reserve_metadata(struct btrfs_root *root,
 	if (ret && qgroup_num_bytes)
 		btrfs_qgroup_free_meta_prealloc(root, qgroup_num_bytes);
 
+	if (!ret) {
+		spin_lock(&rsv->lock);
+		rsv->qgroup_rsv_reserved += qgroup_num_bytes;
+		spin_unlock(&rsv->lock);
+	}
 	return ret;
 }
 
-void btrfs_subvolume_release_metadata(struct btrfs_fs_info *fs_info,
+void btrfs_subvolume_release_metadata(struct btrfs_root *root,
 				      struct btrfs_block_rsv *rsv)
 {
-	btrfs_block_rsv_release(fs_info, rsv, (u64)-1);
+	struct btrfs_fs_info *fs_info = root->fs_info;
+	u64 qgroup_to_release;
+
+	btrfs_block_rsv_release(fs_info, rsv, (u64)-1, &qgroup_to_release);
+	btrfs_qgroup_convert_reserved_meta(root, qgroup_to_release);
 }
diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c
index 61b37c56a7fb..196c4c6ed1ed 100644
--- a/fs/btrfs/scrub.c
+++ b/fs/btrfs/scrub.c
@@ -20,6 +20,7 @@
 #include "rcu-string.h"
 #include "raid56.h"
 #include "block-group.h"
+#include "zoned.h"
 
 /*
  * This is only the first step towards a full-features scrub. It reads all
@@ -38,44 +39,39 @@ struct scrub_block;
 struct scrub_ctx;
 
 /*
- * the following three values only influence the performance.
+ * The following three values only influence the performance.
+ *
  * The last one configures the number of parallel and outstanding I/O
- * operations. The first two values configure an upper limit for the number
+ * operations. The first one configures an upper limit for the number
  * of (dynamically allocated) pages that are added to a bio.
  */
-#define SCRUB_PAGES_PER_RD_BIO	32	/* 128k per bio */
-#define SCRUB_PAGES_PER_WR_BIO	32	/* 128k per bio */
-#define SCRUB_BIOS_PER_SCTX	64	/* 8MB per device in flight */
+#define SCRUB_SECTORS_PER_BIO	32	/* 128KiB per bio for 4KiB pages */
+#define SCRUB_BIOS_PER_SCTX	64	/* 8MiB per device in flight for 4KiB pages */
 
 /*
- * the following value times PAGE_SIZE needs to be large enough to match the
+ * The following value times PAGE_SIZE needs to be large enough to match the
  * largest node/leaf/sector size that shall be supported.
- * Values larger than BTRFS_STRIPE_LEN are not supported.
  */
-#define SCRUB_MAX_PAGES_PER_BLOCK	16	/* 64k per node/leaf/sector */
+#define SCRUB_MAX_SECTORS_PER_BLOCK	(BTRFS_MAX_METADATA_BLOCKSIZE / SZ_4K)
+
+#define SCRUB_MAX_PAGES			(DIV_ROUND_UP(BTRFS_MAX_METADATA_BLOCKSIZE, PAGE_SIZE))
 
 struct scrub_recover {
 	refcount_t		refs;
-	struct btrfs_bio	*bbio;
+	struct btrfs_io_context	*bioc;
 	u64			map_length;
 };
 
-struct scrub_page {
+struct scrub_sector {
 	struct scrub_block	*sblock;
-	struct page		*page;
-	struct btrfs_device	*dev;
 	struct list_head	list;
 	u64			flags;  /* extent flags */
 	u64			generation;
-	u64			logical;
-	u64			physical;
-	u64			physical_for_dev_replace;
+	/* Offset in bytes to @sblock. */
+	u32			offset;
 	atomic_t		refs;
-	struct {
-		unsigned int	mirror_num:8;
-		unsigned int	have_csum:1;
-		unsigned int	io_error:1;
-	};
+	unsigned int		have_csum:1;
+	unsigned int		io_error:1;
 	u8			csum[BTRFS_CSUM_SIZE];
 
 	struct scrub_recover	*recover;
@@ -89,20 +85,30 @@ struct scrub_bio {
 	blk_status_t		status;
 	u64			logical;
 	u64			physical;
-#if SCRUB_PAGES_PER_WR_BIO >= SCRUB_PAGES_PER_RD_BIO
-	struct scrub_page	*pagev[SCRUB_PAGES_PER_WR_BIO];
-#else
-	struct scrub_page	*pagev[SCRUB_PAGES_PER_RD_BIO];
-#endif
-	int			page_count;
+	struct scrub_sector	*sectors[SCRUB_SECTORS_PER_BIO];
+	int			sector_count;
 	int			next_free;
-	struct btrfs_work	work;
+	struct work_struct	work;
 };
 
 struct scrub_block {
-	struct scrub_page	*pagev[SCRUB_MAX_PAGES_PER_BLOCK];
-	int			page_count;
-	atomic_t		outstanding_pages;
+	/*
+	 * Each page will have its page::private used to record the logical
+	 * bytenr.
+	 */
+	struct page		*pages[SCRUB_MAX_PAGES];
+	struct scrub_sector	*sectors[SCRUB_MAX_SECTORS_PER_BLOCK];
+	struct btrfs_device	*dev;
+	/* Logical bytenr of the sblock */
+	u64			logical;
+	u64			physical;
+	u64			physical_for_dev_replace;
+	/* Length of sblock in bytes */
+	u32			len;
+	int			sector_count;
+	int			mirror_num;
+
+	atomic_t		outstanding_sectors;
 	refcount_t		refs; /* free mem on transition to zero */
 	struct scrub_ctx	*sctx;
 	struct scrub_parity	*sparity;
@@ -116,7 +122,7 @@ struct scrub_block {
 		/* It is for the data with checksum */
 		unsigned int	data_corrected:1;
 	};
-	struct btrfs_work	work;
+	struct work_struct	work;
 };
 
 /* Used for the chunks with parity stripe such RAID5/6 */
@@ -131,25 +137,23 @@ struct scrub_parity {
 
 	int			nsectors;
 
-	u64			stripe_len;
+	u32			stripe_len;
 
 	refcount_t		refs;
 
-	struct list_head	spages;
+	struct list_head	sectors_list;
 
 	/* Work of parity check and repair */
-	struct btrfs_work	work;
+	struct work_struct	work;
 
 	/* Mark the parity blocks which have data */
-	unsigned long		*dbitmap;
+	unsigned long		dbitmap;
 
 	/*
 	 * Mark the parity blocks which have data, but errors happen when
 	 * read data or check data
 	 */
-	unsigned long		*ebitmap;
-
-	unsigned long		bitmap[0];
+	unsigned long		ebitmap;
 };
 
 struct scrub_ctx {
@@ -161,17 +165,20 @@ struct scrub_ctx {
 	atomic_t		workers_pending;
 	spinlock_t		list_lock;
 	wait_queue_head_t	list_wait;
-	u16			csum_size;
 	struct list_head	csum_list;
 	atomic_t		cancel_req;
 	int			readonly;
-	int			pages_per_rd_bio;
+	int			sectors_per_bio;
+
+	/* State of IO submission throttling affecting the associated device */
+	ktime_t			throttle_deadline;
+	u64			throttle_sent;
 
 	int			is_dev_replace;
+	u64			write_pointer;
 
 	struct scrub_bio        *wr_curr_bio;
 	struct mutex            wr_lock;
-	int                     pages_per_wr_bio; /* <= SCRUB_PAGES_PER_WR_BIO */
 	struct btrfs_device     *wr_tgtdev;
 	bool                    flush_all_writes;
 
@@ -207,59 +214,217 @@ struct full_stripe_lock {
 	struct mutex mutex;
 };
 
-static void scrub_pending_bio_inc(struct scrub_ctx *sctx);
-static void scrub_pending_bio_dec(struct scrub_ctx *sctx);
-static int scrub_handle_errored_block(struct scrub_block *sblock_to_check);
+#ifndef CONFIG_64BIT
+/* This structure is for archtectures whose (void *) is smaller than u64 */
+struct scrub_page_private {
+	u64 logical;
+};
+#endif
+
+static int attach_scrub_page_private(struct page *page, u64 logical)
+{
+#ifdef CONFIG_64BIT
+	attach_page_private(page, (void *)logical);
+	return 0;
+#else
+	struct scrub_page_private *spp;
+
+	spp = kmalloc(sizeof(*spp), GFP_KERNEL);
+	if (!spp)
+		return -ENOMEM;
+	spp->logical = logical;
+	attach_page_private(page, (void *)spp);
+	return 0;
+#endif
+}
+
+static void detach_scrub_page_private(struct page *page)
+{
+#ifdef CONFIG_64BIT
+	detach_page_private(page);
+	return;
+#else
+	struct scrub_page_private *spp;
+
+	spp = detach_page_private(page);
+	kfree(spp);
+	return;
+#endif
+}
+
+static struct scrub_block *alloc_scrub_block(struct scrub_ctx *sctx,
+					     struct btrfs_device *dev,
+					     u64 logical, u64 physical,
+					     u64 physical_for_dev_replace,
+					     int mirror_num)
+{
+	struct scrub_block *sblock;
+
+	sblock = kzalloc(sizeof(*sblock), GFP_KERNEL);
+	if (!sblock)
+		return NULL;
+	refcount_set(&sblock->refs, 1);
+	sblock->sctx = sctx;
+	sblock->logical = logical;
+	sblock->physical = physical;
+	sblock->physical_for_dev_replace = physical_for_dev_replace;
+	sblock->dev = dev;
+	sblock->mirror_num = mirror_num;
+	sblock->no_io_error_seen = 1;
+	/*
+	 * Scrub_block::pages will be allocated at alloc_scrub_sector() when
+	 * the corresponding page is not allocated.
+	 */
+	return sblock;
+}
+
+/*
+ * Allocate a new scrub sector and attach it to @sblock.
+ *
+ * Will also allocate new pages for @sblock if needed.
+ */
+static struct scrub_sector *alloc_scrub_sector(struct scrub_block *sblock,
+					       u64 logical, gfp_t gfp)
+{
+	const pgoff_t page_index = (logical - sblock->logical) >> PAGE_SHIFT;
+	struct scrub_sector *ssector;
+
+	/* We must never have scrub_block exceed U32_MAX in size. */
+	ASSERT(logical - sblock->logical < U32_MAX);
+
+	ssector = kzalloc(sizeof(*ssector), gfp);
+	if (!ssector)
+		return NULL;
+
+	/* Allocate a new page if the slot is not allocated */
+	if (!sblock->pages[page_index]) {
+		int ret;
+
+		sblock->pages[page_index] = alloc_page(gfp);
+		if (!sblock->pages[page_index]) {
+			kfree(ssector);
+			return NULL;
+		}
+		ret = attach_scrub_page_private(sblock->pages[page_index],
+				sblock->logical + (page_index << PAGE_SHIFT));
+		if (ret < 0) {
+			kfree(ssector);
+			__free_page(sblock->pages[page_index]);
+			sblock->pages[page_index] = NULL;
+			return NULL;
+		}
+	}
+
+	atomic_set(&ssector->refs, 1);
+	ssector->sblock = sblock;
+	/* The sector to be added should not be used */
+	ASSERT(sblock->sectors[sblock->sector_count] == NULL);
+	ssector->offset = logical - sblock->logical;
+
+	/* The sector count must be smaller than the limit */
+	ASSERT(sblock->sector_count < SCRUB_MAX_SECTORS_PER_BLOCK);
+
+	sblock->sectors[sblock->sector_count] = ssector;
+	sblock->sector_count++;
+	sblock->len += sblock->sctx->fs_info->sectorsize;
+
+	return ssector;
+}
+
+static struct page *scrub_sector_get_page(struct scrub_sector *ssector)
+{
+	struct scrub_block *sblock = ssector->sblock;
+	pgoff_t index;
+	/*
+	 * When calling this function, ssector must be alreaday attached to the
+	 * parent sblock.
+	 */
+	ASSERT(sblock);
+
+	/* The range should be inside the sblock range */
+	ASSERT(ssector->offset < sblock->len);
+
+	index = ssector->offset >> PAGE_SHIFT;
+	ASSERT(index < SCRUB_MAX_PAGES);
+	ASSERT(sblock->pages[index]);
+	ASSERT(PagePrivate(sblock->pages[index]));
+	return sblock->pages[index];
+}
+
+static unsigned int scrub_sector_get_page_offset(struct scrub_sector *ssector)
+{
+	struct scrub_block *sblock = ssector->sblock;
+
+	/*
+	 * When calling this function, ssector must be already attached to the
+	 * parent sblock.
+	 */
+	ASSERT(sblock);
+
+	/* The range should be inside the sblock range */
+	ASSERT(ssector->offset < sblock->len);
+
+	return offset_in_page(ssector->offset);
+}
+
+static char *scrub_sector_get_kaddr(struct scrub_sector *ssector)
+{
+	return page_address(scrub_sector_get_page(ssector)) +
+	       scrub_sector_get_page_offset(ssector);
+}
+
+static int bio_add_scrub_sector(struct bio *bio, struct scrub_sector *ssector,
+				unsigned int len)
+{
+	return bio_add_page(bio, scrub_sector_get_page(ssector), len,
+			    scrub_sector_get_page_offset(ssector));
+}
+
 static int scrub_setup_recheck_block(struct scrub_block *original_sblock,
-				     struct scrub_block *sblocks_for_recheck);
+				     struct scrub_block *sblocks_for_recheck[]);
 static void scrub_recheck_block(struct btrfs_fs_info *fs_info,
 				struct scrub_block *sblock,
 				int retry_failed_mirror);
 static void scrub_recheck_block_checksum(struct scrub_block *sblock);
 static int scrub_repair_block_from_good_copy(struct scrub_block *sblock_bad,
 					     struct scrub_block *sblock_good);
-static int scrub_repair_page_from_good_copy(struct scrub_block *sblock_bad,
+static int scrub_repair_sector_from_good_copy(struct scrub_block *sblock_bad,
 					    struct scrub_block *sblock_good,
-					    int page_num, int force_write);
+					    int sector_num, int force_write);
 static void scrub_write_block_to_dev_replace(struct scrub_block *sblock);
-static int scrub_write_page_to_dev_replace(struct scrub_block *sblock,
-					   int page_num);
+static int scrub_write_sector_to_dev_replace(struct scrub_block *sblock,
+					     int sector_num);
 static int scrub_checksum_data(struct scrub_block *sblock);
 static int scrub_checksum_tree_block(struct scrub_block *sblock);
 static int scrub_checksum_super(struct scrub_block *sblock);
-static void scrub_block_get(struct scrub_block *sblock);
 static void scrub_block_put(struct scrub_block *sblock);
-static void scrub_page_get(struct scrub_page *spage);
-static void scrub_page_put(struct scrub_page *spage);
+static void scrub_sector_get(struct scrub_sector *sector);
+static void scrub_sector_put(struct scrub_sector *sector);
 static void scrub_parity_get(struct scrub_parity *sparity);
 static void scrub_parity_put(struct scrub_parity *sparity);
-static int scrub_add_page_to_rd_bio(struct scrub_ctx *sctx,
-				    struct scrub_page *spage);
-static int scrub_pages(struct scrub_ctx *sctx, u64 logical, u64 len,
-		       u64 physical, struct btrfs_device *dev, u64 flags,
-		       u64 gen, int mirror_num, u8 *csum, int force,
-		       u64 physical_for_dev_replace);
+static int scrub_sectors(struct scrub_ctx *sctx, u64 logical, u32 len,
+			 u64 physical, struct btrfs_device *dev, u64 flags,
+			 u64 gen, int mirror_num, u8 *csum,
+			 u64 physical_for_dev_replace);
 static void scrub_bio_end_io(struct bio *bio);
-static void scrub_bio_end_io_worker(struct btrfs_work *work);
+static void scrub_bio_end_io_worker(struct work_struct *work);
 static void scrub_block_complete(struct scrub_block *sblock);
-static void scrub_remap_extent(struct btrfs_fs_info *fs_info,
-			       u64 extent_logical, u64 extent_len,
-			       u64 *extent_physical,
-			       struct btrfs_device **extent_dev,
-			       int *extent_mirror_num);
-static int scrub_add_page_to_wr_bio(struct scrub_ctx *sctx,
-				    struct scrub_page *spage);
+static void scrub_find_good_copy(struct btrfs_fs_info *fs_info,
+				 u64 extent_logical, u32 extent_len,
+				 u64 *extent_physical,
+				 struct btrfs_device **extent_dev,
+				 int *extent_mirror_num);
+static int scrub_add_sector_to_wr_bio(struct scrub_ctx *sctx,
+				      struct scrub_sector *sector);
 static void scrub_wr_submit(struct scrub_ctx *sctx);
 static void scrub_wr_bio_end_io(struct bio *bio);
-static void scrub_wr_bio_end_io_worker(struct btrfs_work *work);
-static void __scrub_blocked_if_needed(struct btrfs_fs_info *fs_info);
-static void scrub_blocked_if_needed(struct btrfs_fs_info *fs_info);
+static void scrub_wr_bio_end_io_worker(struct work_struct *work);
 static void scrub_put_ctx(struct scrub_ctx *sctx);
 
-static inline int scrub_is_page_on_raid56(struct scrub_page *page)
+static inline int scrub_is_page_on_raid56(struct scrub_sector *sector)
 {
-	return page->recover &&
-	       (page->recover->bbio->map_type & BTRFS_BLOCK_GROUP_RAID56_MASK);
+	return sector->recover &&
+	       (sector->recover->bioc->map_type & BTRFS_BLOCK_GROUP_RAID56_MASK);
 }
 
 static void scrub_pending_bio_inc(struct scrub_ctx *sctx)
@@ -546,10 +711,8 @@ static noinline_for_stack void scrub_free_ctx(struct scrub_ctx *sctx)
 	if (sctx->curr != -1) {
 		struct scrub_bio *sbio = sctx->bios[sctx->curr];
 
-		for (i = 0; i < sbio->page_count; i++) {
-			WARN_ON(!sbio->pagev[i]->page);
-			scrub_block_put(sbio->pagev[i]->sblock);
-		}
+		for (i = 0; i < sbio->sector_count; i++)
+			scrub_block_put(sbio->sectors[i]->sblock);
 		bio_put(sbio->bio);
 	}
 
@@ -583,7 +746,7 @@ static noinline_for_stack struct scrub_ctx *scrub_setup_ctx(
 		goto nomem;
 	refcount_set(&sctx->refs, 1);
 	sctx->is_dev_replace = is_dev_replace;
-	sctx->pages_per_rd_bio = SCRUB_PAGES_PER_RD_BIO;
+	sctx->sectors_per_bio = SCRUB_SECTORS_PER_BIO;
 	sctx->curr = -1;
 	sctx->fs_info = fs_info;
 	INIT_LIST_HEAD(&sctx->csum_list);
@@ -597,9 +760,8 @@ static noinline_for_stack struct scrub_ctx *scrub_setup_ctx(
 
 		sbio->index = i;
 		sbio->sctx = sctx;
-		sbio->page_count = 0;
-		btrfs_init_work(&sbio->work, scrub_bio_end_io_worker, NULL,
-				NULL);
+		sbio->sector_count = 0;
+		INIT_WORK(&sbio->work, scrub_bio_end_io_worker);
 
 		if (i != SCRUB_BIOS_PER_SCTX - 1)
 			sctx->bios[i]->next_free = i + 1;
@@ -610,18 +772,17 @@ static noinline_for_stack struct scrub_ctx *scrub_setup_ctx(
 	atomic_set(&sctx->bios_in_flight, 0);
 	atomic_set(&sctx->workers_pending, 0);
 	atomic_set(&sctx->cancel_req, 0);
-	sctx->csum_size = btrfs_super_csum_size(fs_info->super_copy);
 
 	spin_lock_init(&sctx->list_lock);
 	spin_lock_init(&sctx->stat_lock);
 	init_waitqueue_head(&sctx->list_wait);
+	sctx->throttle_deadline = 0;
 
 	WARN_ON(sctx->wr_curr_bio != NULL);
 	mutex_init(&sctx->wr_lock);
 	sctx->wr_curr_bio = NULL;
 	if (is_dev_replace) {
 		WARN_ON(!fs_info->dev_replace.tgtdev);
-		sctx->pages_per_wr_bio = SCRUB_PAGES_PER_WR_BIO;
 		sctx->wr_tgtdev = fs_info->dev_replace.tgtdev;
 		sctx->flush_all_writes = false;
 	}
@@ -636,7 +797,6 @@ nomem:
 static int scrub_print_warning_inode(u64 inum, u64 offset, u64 root,
 				     void *warn_ctx)
 {
-	u64 isize;
 	u32 nlink;
 	int ret;
 	int i;
@@ -647,13 +807,9 @@ static int scrub_print_warning_inode(u64 inum, u64 offset, u64 root,
 	struct btrfs_fs_info *fs_info = swarn->dev->fs_info;
 	struct inode_fs_paths *ipath = NULL;
 	struct btrfs_root *local_root;
-	struct btrfs_key root_key;
 	struct btrfs_key key;
 
-	root_key.objectid = root;
-	root_key.type = BTRFS_ROOT_ITEM_KEY;
-	root_key.offset = (u64)-1;
-	local_root = btrfs_read_fs_root_no_name(fs_info, &root_key);
+	local_root = btrfs_get_fs_root(fs_info, root, true);
 	if (IS_ERR(local_root)) {
 		ret = PTR_ERR(local_root);
 		goto err;
@@ -668,6 +824,7 @@ static int scrub_print_warning_inode(u64 inum, u64 offset, u64 root,
 
 	ret = btrfs_search_slot(NULL, local_root, &key, swarn->path, 0, 0);
 	if (ret) {
+		btrfs_put_root(local_root);
 		btrfs_release_path(swarn->path);
 		goto err;
 	}
@@ -675,7 +832,6 @@ static int scrub_print_warning_inode(u64 inum, u64 offset, u64 root,
 	eb = swarn->path->nodes[0];
 	inode_item = btrfs_item_ptr(eb, swarn->path->slots[0],
 					struct btrfs_inode_item);
-	isize = btrfs_inode_size(eb, inode_item);
 	nlink = btrfs_inode_nlink(eb, inode_item);
 	btrfs_release_path(swarn->path);
 
@@ -688,6 +844,7 @@ static int scrub_print_warning_inode(u64 inum, u64 offset, u64 root,
 	ipath = init_ipath(4096, local_root, swarn->path);
 	memalloc_nofs_restore(nofs_flag);
 	if (IS_ERR(ipath)) {
+		btrfs_put_root(local_root);
 		ret = PTR_ERR(ipath);
 		ipath = NULL;
 		goto err;
@@ -703,14 +860,15 @@ static int scrub_print_warning_inode(u64 inum, u64 offset, u64 root,
 	 */
 	for (i = 0; i < ipath->fspath->elem_cnt; ++i)
 		btrfs_warn_in_rcu(fs_info,
-"%s at logical %llu on dev %s, physical %llu, root %llu, inode %llu, offset %llu, length %llu, links %u (path: %s)",
+"%s at logical %llu on dev %s, physical %llu, root %llu, inode %llu, offset %llu, length %u, links %u (path: %s)",
 				  swarn->errstr, swarn->logical,
 				  rcu_str_deref(swarn->dev->name),
 				  swarn->physical,
 				  root, inum, offset,
-				  min(isize - offset, (u64)PAGE_SIZE), nlink,
+				  fs_info->sectorsize, nlink,
 				  (char *)(unsigned long)ipath->fspath->val[i]);
 
+	btrfs_put_root(local_root);
 	free_ipath(ipath);
 	return 0;
 
@@ -743,16 +901,23 @@ static void scrub_print_warning(const char *errstr, struct scrub_block *sblock)
 	u8 ref_level = 0;
 	int ret;
 
-	WARN_ON(sblock->page_count < 1);
-	dev = sblock->pagev[0]->dev;
+	WARN_ON(sblock->sector_count < 1);
+	dev = sblock->dev;
 	fs_info = sblock->sctx->fs_info;
 
+	/* Super block error, no need to search extent tree. */
+	if (sblock->sectors[0]->flags & BTRFS_EXTENT_FLAG_SUPER) {
+		btrfs_warn_in_rcu(fs_info, "%s on device %s, physical %llu",
+			errstr, rcu_str_deref(dev->name),
+			sblock->physical);
+		return;
+	}
 	path = btrfs_alloc_path();
 	if (!path)
 		return;
 
-	swarn.physical = sblock->pagev[0]->physical;
-	swarn.logical = sblock->pagev[0]->logical;
+	swarn.physical = sblock->physical;
+	swarn.logical = sblock->logical;
 	swarn.errstr = errstr;
 	swarn.dev = NULL;
 
@@ -766,7 +931,7 @@ static void scrub_print_warning(const char *errstr, struct scrub_block *sblock)
 
 	eb = path->nodes[0];
 	ei = btrfs_item_ptr(eb, path->slots[0], struct btrfs_extent_item);
-	item_size = btrfs_item_size_nr(eb, path->slots[0]);
+	item_size = btrfs_item_size(eb, path->slots[0]);
 
 	if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
 		do {
@@ -806,15 +971,15 @@ static inline void scrub_put_recover(struct btrfs_fs_info *fs_info,
 {
 	if (refcount_dec_and_test(&recover->refs)) {
 		btrfs_bio_counter_dec(fs_info);
-		btrfs_put_bbio(recover->bbio);
+		btrfs_put_bioc(recover->bioc);
 		kfree(recover);
 	}
 }
 
 /*
  * scrub_handle_errored_block gets called when either verification of the
- * pages failed or the bio failed to read, e.g. with EIO. In the latter
- * case, this function handles all pages in the bio, even though only one
+ * sectors failed or the bio failed to read, e.g. with EIO. In the latter
+ * case, this function handles all sectors in the bio, even though only one
  * may be bad.
  * The goal of this function is to repair the errored block by using the
  * contents of one of the mirrors.
@@ -822,43 +987,48 @@ static inline void scrub_put_recover(struct btrfs_fs_info *fs_info,
 static int scrub_handle_errored_block(struct scrub_block *sblock_to_check)
 {
 	struct scrub_ctx *sctx = sblock_to_check->sctx;
-	struct btrfs_device *dev;
+	struct btrfs_device *dev = sblock_to_check->dev;
 	struct btrfs_fs_info *fs_info;
 	u64 logical;
 	unsigned int failed_mirror_index;
 	unsigned int is_metadata;
 	unsigned int have_csum;
-	struct scrub_block *sblocks_for_recheck; /* holds one for each mirror */
+	/* One scrub_block for each mirror */
+	struct scrub_block *sblocks_for_recheck[BTRFS_MAX_MIRRORS] = { 0 };
 	struct scrub_block *sblock_bad;
 	int ret;
 	int mirror_index;
-	int page_num;
+	int sector_num;
 	int success;
 	bool full_stripe_locked;
 	unsigned int nofs_flag;
-	static DEFINE_RATELIMIT_STATE(_rs, DEFAULT_RATELIMIT_INTERVAL,
+	static DEFINE_RATELIMIT_STATE(rs, DEFAULT_RATELIMIT_INTERVAL,
 				      DEFAULT_RATELIMIT_BURST);
 
-	BUG_ON(sblock_to_check->page_count < 1);
+	BUG_ON(sblock_to_check->sector_count < 1);
 	fs_info = sctx->fs_info;
-	if (sblock_to_check->pagev[0]->flags & BTRFS_EXTENT_FLAG_SUPER) {
+	if (sblock_to_check->sectors[0]->flags & BTRFS_EXTENT_FLAG_SUPER) {
 		/*
-		 * if we find an error in a super block, we just report it.
+		 * If we find an error in a super block, we just report it.
 		 * They will get written with the next transaction commit
 		 * anyway
 		 */
+		scrub_print_warning("super block error", sblock_to_check);
 		spin_lock(&sctx->stat_lock);
 		++sctx->stat.super_errors;
 		spin_unlock(&sctx->stat_lock);
+		btrfs_dev_stat_inc_and_print(dev, BTRFS_DEV_STAT_CORRUPTION_ERRS);
 		return 0;
 	}
-	logical = sblock_to_check->pagev[0]->logical;
-	BUG_ON(sblock_to_check->pagev[0]->mirror_num < 1);
-	failed_mirror_index = sblock_to_check->pagev[0]->mirror_num - 1;
-	is_metadata = !(sblock_to_check->pagev[0]->flags &
+	logical = sblock_to_check->logical;
+	ASSERT(sblock_to_check->mirror_num);
+	failed_mirror_index = sblock_to_check->mirror_num - 1;
+	is_metadata = !(sblock_to_check->sectors[0]->flags &
 			BTRFS_EXTENT_FLAG_DATA);
-	have_csum = sblock_to_check->pagev[0]->have_csum;
-	dev = sblock_to_check->pagev[0]->dev;
+	have_csum = sblock_to_check->sectors[0]->have_csum;
+
+	if (!sctx->is_dev_replace && btrfs_repair_one_zone(fs_info, logical))
+		return 0;
 
 	/*
 	 * We must use GFP_NOFS because the scrub task might be waiting for a
@@ -866,7 +1036,7 @@ static int scrub_handle_errored_block(struct scrub_block *sblock_to_check)
 	 * might be waiting the scrub task to pause (which needs to wait for all
 	 * the worker tasks to complete before pausing).
 	 * We do allocations in the workers through insert_full_stripe_lock()
-	 * and scrub_add_page_to_wr_bio(), which happens down the call chain of
+	 * and scrub_add_sector_to_wr_bio(), which happens down the call chain of
 	 * this function.
 	 */
 	nofs_flag = memalloc_nofs_save();
@@ -893,44 +1063,55 @@ static int scrub_handle_errored_block(struct scrub_block *sblock_to_check)
 	 * read all mirrors one after the other. This includes to
 	 * re-read the extent or metadata block that failed (that was
 	 * the cause that this fixup code is called) another time,
-	 * page by page this time in order to know which pages
+	 * sector by sector this time in order to know which sectors
 	 * caused I/O errors and which ones are good (for all mirrors).
 	 * It is the goal to handle the situation when more than one
 	 * mirror contains I/O errors, but the errors do not
 	 * overlap, i.e. the data can be repaired by selecting the
-	 * pages from those mirrors without I/O error on the
-	 * particular pages. One example (with blocks >= 2 * PAGE_SIZE)
-	 * would be that mirror #1 has an I/O error on the first page,
-	 * the second page is good, and mirror #2 has an I/O error on
-	 * the second page, but the first page is good.
-	 * Then the first page of the first mirror can be repaired by
-	 * taking the first page of the second mirror, and the
-	 * second page of the second mirror can be repaired by
-	 * copying the contents of the 2nd page of the 1st mirror.
-	 * One more note: if the pages of one mirror contain I/O
+	 * sectors from those mirrors without I/O error on the
+	 * particular sectors. One example (with blocks >= 2 * sectorsize)
+	 * would be that mirror #1 has an I/O error on the first sector,
+	 * the second sector is good, and mirror #2 has an I/O error on
+	 * the second sector, but the first sector is good.
+	 * Then the first sector of the first mirror can be repaired by
+	 * taking the first sector of the second mirror, and the
+	 * second sector of the second mirror can be repaired by
+	 * copying the contents of the 2nd sector of the 1st mirror.
+	 * One more note: if the sectors of one mirror contain I/O
 	 * errors, the checksum cannot be verified. In order to get
 	 * the best data for repairing, the first attempt is to find
 	 * a mirror without I/O errors and with a validated checksum.
-	 * Only if this is not possible, the pages are picked from
+	 * Only if this is not possible, the sectors are picked from
 	 * mirrors with I/O errors without considering the checksum.
 	 * If the latter is the case, at the end, the checksum of the
 	 * repaired area is verified in order to correctly maintain
 	 * the statistics.
 	 */
-
-	sblocks_for_recheck = kcalloc(BTRFS_MAX_MIRRORS,
-				      sizeof(*sblocks_for_recheck), GFP_KERNEL);
-	if (!sblocks_for_recheck) {
-		spin_lock(&sctx->stat_lock);
-		sctx->stat.malloc_errors++;
-		sctx->stat.read_errors++;
-		sctx->stat.uncorrectable_errors++;
-		spin_unlock(&sctx->stat_lock);
-		btrfs_dev_stat_inc_and_print(dev, BTRFS_DEV_STAT_READ_ERRS);
-		goto out;
+	for (mirror_index = 0; mirror_index < BTRFS_MAX_MIRRORS; mirror_index++) {
+		/*
+		 * Note: the two members refs and outstanding_sectors are not
+		 * used in the blocks that are used for the recheck procedure.
+		 *
+		 * But alloc_scrub_block() will initialize sblock::ref anyway,
+		 * so we can use scrub_block_put() to clean them up.
+		 *
+		 * And here we don't setup the physical/dev for the sblock yet,
+		 * they will be correctly initialized in scrub_setup_recheck_block().
+		 */
+		sblocks_for_recheck[mirror_index] = alloc_scrub_block(sctx, NULL,
+							logical, 0, 0, mirror_index);
+		if (!sblocks_for_recheck[mirror_index]) {
+			spin_lock(&sctx->stat_lock);
+			sctx->stat.malloc_errors++;
+			sctx->stat.read_errors++;
+			sctx->stat.uncorrectable_errors++;
+			spin_unlock(&sctx->stat_lock);
+			btrfs_dev_stat_inc_and_print(dev, BTRFS_DEV_STAT_READ_ERRS);
+			goto out;
+		}
 	}
 
-	/* setup the context, map the logical blocks and alloc the pages */
+	/* Setup the context, map the logical blocks and alloc the sectors */
 	ret = scrub_setup_recheck_block(sblock_to_check, sblocks_for_recheck);
 	if (ret) {
 		spin_lock(&sctx->stat_lock);
@@ -941,7 +1122,7 @@ static int scrub_handle_errored_block(struct scrub_block *sblock_to_check)
 		goto out;
 	}
 	BUG_ON(failed_mirror_index >= BTRFS_MAX_MIRRORS);
-	sblock_bad = sblocks_for_recheck + failed_mirror_index;
+	sblock_bad = sblocks_for_recheck[failed_mirror_index];
 
 	/* build and submit the bios for the failed mirror, check checksums */
 	scrub_recheck_block(fs_info, sblock_bad, 1);
@@ -949,7 +1130,7 @@ static int scrub_handle_errored_block(struct scrub_block *sblock_to_check)
 	if (!sblock_bad->header_error && !sblock_bad->checksum_error &&
 	    sblock_bad->no_io_error_seen) {
 		/*
-		 * the error disappeared after reading page by page, or
+		 * The error disappeared after reading sector by sector, or
 		 * the area was part of a huge bio and other parts of the
 		 * bio caused I/O errors, or the block layer merged several
 		 * read requests into one and the error is caused by a
@@ -970,14 +1151,14 @@ static int scrub_handle_errored_block(struct scrub_block *sblock_to_check)
 		spin_lock(&sctx->stat_lock);
 		sctx->stat.read_errors++;
 		spin_unlock(&sctx->stat_lock);
-		if (__ratelimit(&_rs))
+		if (__ratelimit(&rs))
 			scrub_print_warning("i/o error", sblock_to_check);
 		btrfs_dev_stat_inc_and_print(dev, BTRFS_DEV_STAT_READ_ERRS);
 	} else if (sblock_bad->checksum_error) {
 		spin_lock(&sctx->stat_lock);
 		sctx->stat.csum_errors++;
 		spin_unlock(&sctx->stat_lock);
-		if (__ratelimit(&_rs))
+		if (__ratelimit(&rs))
 			scrub_print_warning("checksum error", sblock_to_check);
 		btrfs_dev_stat_inc_and_print(dev,
 					     BTRFS_DEV_STAT_CORRUPTION_ERRS);
@@ -985,7 +1166,7 @@ static int scrub_handle_errored_block(struct scrub_block *sblock_to_check)
 		spin_lock(&sctx->stat_lock);
 		sctx->stat.verify_errors++;
 		spin_unlock(&sctx->stat_lock);
-		if (__ratelimit(&_rs))
+		if (__ratelimit(&rs))
 			scrub_print_warning("checksum/header error",
 					    sblock_to_check);
 		if (sblock_bad->generation_error)
@@ -1010,10 +1191,10 @@ static int scrub_handle_errored_block(struct scrub_block *sblock_to_check)
 	 * that is known to contain an error is rewritten. Afterwards
 	 * the block is known to be corrected.
 	 * If a mirror is found which is completely correct, and no
-	 * checksum is present, only those pages are rewritten that had
+	 * checksum is present, only those sectors are rewritten that had
 	 * an I/O error in the block to be repaired, since it cannot be
-	 * determined, which copy of the other pages is better (and it
-	 * could happen otherwise that a correct page would be
+	 * determined, which copy of the other sectors is better (and it
+	 * could happen otherwise that a correct sector would be
 	 * overwritten by a bad one).
 	 */
 	for (mirror_index = 0; ;mirror_index++) {
@@ -1023,26 +1204,25 @@ static int scrub_handle_errored_block(struct scrub_block *sblock_to_check)
 			continue;
 
 		/* raid56's mirror can be more than BTRFS_MAX_MIRRORS */
-		if (!scrub_is_page_on_raid56(sblock_bad->pagev[0])) {
+		if (!scrub_is_page_on_raid56(sblock_bad->sectors[0])) {
 			if (mirror_index >= BTRFS_MAX_MIRRORS)
 				break;
-			if (!sblocks_for_recheck[mirror_index].page_count)
+			if (!sblocks_for_recheck[mirror_index]->sector_count)
 				break;
 
-			sblock_other = sblocks_for_recheck + mirror_index;
+			sblock_other = sblocks_for_recheck[mirror_index];
 		} else {
-			struct scrub_recover *r = sblock_bad->pagev[0]->recover;
-			int max_allowed = r->bbio->num_stripes -
-						r->bbio->num_tgtdevs;
+			struct scrub_recover *r = sblock_bad->sectors[0]->recover;
+			int max_allowed = r->bioc->num_stripes - r->bioc->num_tgtdevs;
 
 			if (mirror_index >= max_allowed)
 				break;
-			if (!sblocks_for_recheck[1].page_count)
+			if (!sblocks_for_recheck[1]->sector_count)
 				break;
 
 			ASSERT(failed_mirror_index == 0);
-			sblock_other = sblocks_for_recheck + 1;
-			sblock_other->pagev[0]->mirror_num = 1 + mirror_index;
+			sblock_other = sblocks_for_recheck[1];
+			sblock_other->mirror_num = 1 + mirror_index;
 		}
 
 		/* build and submit the bios, check checksums */
@@ -1068,39 +1248,39 @@ static int scrub_handle_errored_block(struct scrub_block *sblock_to_check)
 
 	/*
 	 * In case of I/O errors in the area that is supposed to be
-	 * repaired, continue by picking good copies of those pages.
-	 * Select the good pages from mirrors to rewrite bad pages from
+	 * repaired, continue by picking good copies of those sectors.
+	 * Select the good sectors from mirrors to rewrite bad sectors from
 	 * the area to fix. Afterwards verify the checksum of the block
 	 * that is supposed to be repaired. This verification step is
 	 * only done for the purpose of statistic counting and for the
 	 * final scrub report, whether errors remain.
 	 * A perfect algorithm could make use of the checksum and try
-	 * all possible combinations of pages from the different mirrors
+	 * all possible combinations of sectors from the different mirrors
 	 * until the checksum verification succeeds. For example, when
-	 * the 2nd page of mirror #1 faces I/O errors, and the 2nd page
+	 * the 2nd sector of mirror #1 faces I/O errors, and the 2nd sector
 	 * of mirror #2 is readable but the final checksum test fails,
-	 * then the 2nd page of mirror #3 could be tried, whether now
+	 * then the 2nd sector of mirror #3 could be tried, whether now
 	 * the final checksum succeeds. But this would be a rare
 	 * exception and is therefore not implemented. At least it is
 	 * avoided that the good copy is overwritten.
 	 * A more useful improvement would be to pick the sectors
 	 * without I/O error based on sector sizes (512 bytes on legacy
-	 * disks) instead of on PAGE_SIZE. Then maybe 512 byte of one
+	 * disks) instead of on sectorsize. Then maybe 512 byte of one
 	 * mirror could be repaired by taking 512 byte of a different
-	 * mirror, even if other 512 byte sectors in the same PAGE_SIZE
+	 * mirror, even if other 512 byte sectors in the same sectorsize
 	 * area are unreadable.
 	 */
 	success = 1;
-	for (page_num = 0; page_num < sblock_bad->page_count;
-	     page_num++) {
-		struct scrub_page *page_bad = sblock_bad->pagev[page_num];
+	for (sector_num = 0; sector_num < sblock_bad->sector_count;
+	     sector_num++) {
+		struct scrub_sector *sector_bad = sblock_bad->sectors[sector_num];
 		struct scrub_block *sblock_other = NULL;
 
-		/* skip no-io-error page in scrub */
-		if (!page_bad->io_error && !sctx->is_dev_replace)
+		/* Skip no-io-error sectors in scrub */
+		if (!sector_bad->io_error && !sctx->is_dev_replace)
 			continue;
 
-		if (scrub_is_page_on_raid56(sblock_bad->pagev[0])) {
+		if (scrub_is_page_on_raid56(sblock_bad->sectors[0])) {
 			/*
 			 * In case of dev replace, if raid56 rebuild process
 			 * didn't work out correct data, then copy the content
@@ -1109,16 +1289,15 @@ static int scrub_handle_errored_block(struct scrub_block *sblock_to_check)
 			 * sblock_for_recheck array to target device.
 			 */
 			sblock_other = NULL;
-		} else if (page_bad->io_error) {
-			/* try to find no-io-error page in mirrors */
+		} else if (sector_bad->io_error) {
+			/* Try to find no-io-error sector in mirrors */
 			for (mirror_index = 0;
 			     mirror_index < BTRFS_MAX_MIRRORS &&
-			     sblocks_for_recheck[mirror_index].page_count > 0;
+			     sblocks_for_recheck[mirror_index]->sector_count > 0;
 			     mirror_index++) {
-				if (!sblocks_for_recheck[mirror_index].
-				    pagev[page_num]->io_error) {
-					sblock_other = sblocks_for_recheck +
-						       mirror_index;
+				if (!sblocks_for_recheck[mirror_index]->
+				    sectors[sector_num]->io_error) {
+					sblock_other = sblocks_for_recheck[mirror_index];
 					break;
 				}
 			}
@@ -1128,27 +1307,26 @@ static int scrub_handle_errored_block(struct scrub_block *sblock_to_check)
 
 		if (sctx->is_dev_replace) {
 			/*
-			 * did not find a mirror to fetch the page
-			 * from. scrub_write_page_to_dev_replace()
-			 * handles this case (page->io_error), by
-			 * filling the block with zeros before
-			 * submitting the write request
+			 * Did not find a mirror to fetch the sector from.
+			 * scrub_write_sector_to_dev_replace() handles this
+			 * case (sector->io_error), by filling the block with
+			 * zeros before submitting the write request
 			 */
 			if (!sblock_other)
 				sblock_other = sblock_bad;
 
-			if (scrub_write_page_to_dev_replace(sblock_other,
-							    page_num) != 0) {
+			if (scrub_write_sector_to_dev_replace(sblock_other,
+							      sector_num) != 0) {
 				atomic64_inc(
 					&fs_info->dev_replace.num_write_errors);
 				success = 0;
 			}
 		} else if (sblock_other) {
-			ret = scrub_repair_page_from_good_copy(sblock_bad,
-							       sblock_other,
-							       page_num, 0);
+			ret = scrub_repair_sector_from_good_copy(sblock_bad,
+								 sblock_other,
+								 sector_num, 0);
 			if (0 == ret)
-				page_bad->io_error = 0;
+				sector_bad->io_error = 0;
 			else
 				success = 0;
 		}
@@ -1193,27 +1371,28 @@ did_not_correct_error:
 	}
 
 out:
-	if (sblocks_for_recheck) {
-		for (mirror_index = 0; mirror_index < BTRFS_MAX_MIRRORS;
-		     mirror_index++) {
-			struct scrub_block *sblock = sblocks_for_recheck +
-						     mirror_index;
-			struct scrub_recover *recover;
-			int page_index;
-
-			for (page_index = 0; page_index < sblock->page_count;
-			     page_index++) {
-				sblock->pagev[page_index]->sblock = NULL;
-				recover = sblock->pagev[page_index]->recover;
-				if (recover) {
-					scrub_put_recover(fs_info, recover);
-					sblock->pagev[page_index]->recover =
-									NULL;
-				}
-				scrub_page_put(sblock->pagev[page_index]);
+	for (mirror_index = 0; mirror_index < BTRFS_MAX_MIRRORS; mirror_index++) {
+		struct scrub_block *sblock = sblocks_for_recheck[mirror_index];
+		struct scrub_recover *recover;
+		int sector_index;
+
+		/* Not allocated, continue checking the next mirror */
+		if (!sblock)
+			continue;
+
+		for (sector_index = 0; sector_index < sblock->sector_count;
+		     sector_index++) {
+			/*
+			 * Here we just cleanup the recover, each sector will be
+			 * properly cleaned up by later scrub_block_put()
+			 */
+			recover = sblock->sectors[sector_index]->recover;
+			if (recover) {
+				scrub_put_recover(fs_info, recover);
+				sblock->sectors[sector_index]->recover = NULL;
 			}
 		}
-		kfree(sblocks_for_recheck);
+		scrub_block_put(sblock);
 	}
 
 	ret = unlock_full_stripe(fs_info, logical, full_stripe_locked);
@@ -1223,19 +1402,18 @@ out:
 	return 0;
 }
 
-static inline int scrub_nr_raid_mirrors(struct btrfs_bio *bbio)
+static inline int scrub_nr_raid_mirrors(struct btrfs_io_context *bioc)
 {
-	if (bbio->map_type & BTRFS_BLOCK_GROUP_RAID5)
+	if (bioc->map_type & BTRFS_BLOCK_GROUP_RAID5)
 		return 2;
-	else if (bbio->map_type & BTRFS_BLOCK_GROUP_RAID6)
+	else if (bioc->map_type & BTRFS_BLOCK_GROUP_RAID6)
 		return 3;
 	else
-		return (int)bbio->num_stripes;
+		return (int)bioc->num_stripes;
 }
 
 static inline void scrub_stripe_index_and_offset(u64 logical, u64 map_type,
 						 u64 *raid_map,
-						 u64 mapped_length,
 						 int nstripes, int mirror,
 						 int *stripe_index,
 						 u64 *stripe_offset)
@@ -1250,7 +1428,7 @@ static inline void scrub_stripe_index_and_offset(u64 logical, u64 map_type,
 				continue;
 
 			if (logical >= raid_map[i] &&
-			    logical < raid_map[i] + mapped_length)
+			    logical < raid_map[i] + BTRFS_STRIPE_LEN)
 				break;
 		}
 
@@ -1264,125 +1442,112 @@ static inline void scrub_stripe_index_and_offset(u64 logical, u64 map_type,
 }
 
 static int scrub_setup_recheck_block(struct scrub_block *original_sblock,
-				     struct scrub_block *sblocks_for_recheck)
+				     struct scrub_block *sblocks_for_recheck[])
 {
 	struct scrub_ctx *sctx = original_sblock->sctx;
 	struct btrfs_fs_info *fs_info = sctx->fs_info;
-	u64 length = original_sblock->page_count * PAGE_SIZE;
-	u64 logical = original_sblock->pagev[0]->logical;
-	u64 generation = original_sblock->pagev[0]->generation;
-	u64 flags = original_sblock->pagev[0]->flags;
-	u64 have_csum = original_sblock->pagev[0]->have_csum;
+	u64 logical = original_sblock->logical;
+	u64 length = original_sblock->sector_count << fs_info->sectorsize_bits;
+	u64 generation = original_sblock->sectors[0]->generation;
+	u64 flags = original_sblock->sectors[0]->flags;
+	u64 have_csum = original_sblock->sectors[0]->have_csum;
 	struct scrub_recover *recover;
-	struct btrfs_bio *bbio;
+	struct btrfs_io_context *bioc;
 	u64 sublen;
 	u64 mapped_length;
 	u64 stripe_offset;
 	int stripe_index;
-	int page_index = 0;
+	int sector_index = 0;
 	int mirror_index;
 	int nmirrors;
 	int ret;
 
-	/*
-	 * note: the two members refs and outstanding_pages
-	 * are not used (and not set) in the blocks that are used for
-	 * the recheck procedure
-	 */
-
 	while (length > 0) {
-		sublen = min_t(u64, length, PAGE_SIZE);
+		sublen = min_t(u64, length, fs_info->sectorsize);
 		mapped_length = sublen;
-		bbio = NULL;
+		bioc = NULL;
 
 		/*
-		 * with a length of PAGE_SIZE, each returned stripe
-		 * represents one mirror
+		 * With a length of sectorsize, each returned stripe represents
+		 * one mirror
 		 */
 		btrfs_bio_counter_inc_blocked(fs_info);
 		ret = btrfs_map_sblock(fs_info, BTRFS_MAP_GET_READ_MIRRORS,
-				logical, &mapped_length, &bbio);
-		if (ret || !bbio || mapped_length < sublen) {
-			btrfs_put_bbio(bbio);
+				       logical, &mapped_length, &bioc);
+		if (ret || !bioc || mapped_length < sublen) {
+			btrfs_put_bioc(bioc);
 			btrfs_bio_counter_dec(fs_info);
 			return -EIO;
 		}
 
 		recover = kzalloc(sizeof(struct scrub_recover), GFP_NOFS);
 		if (!recover) {
-			btrfs_put_bbio(bbio);
+			btrfs_put_bioc(bioc);
 			btrfs_bio_counter_dec(fs_info);
 			return -ENOMEM;
 		}
 
 		refcount_set(&recover->refs, 1);
-		recover->bbio = bbio;
+		recover->bioc = bioc;
 		recover->map_length = mapped_length;
 
-		BUG_ON(page_index >= SCRUB_MAX_PAGES_PER_BLOCK);
+		ASSERT(sector_index < SCRUB_MAX_SECTORS_PER_BLOCK);
 
-		nmirrors = min(scrub_nr_raid_mirrors(bbio), BTRFS_MAX_MIRRORS);
+		nmirrors = min(scrub_nr_raid_mirrors(bioc), BTRFS_MAX_MIRRORS);
 
 		for (mirror_index = 0; mirror_index < nmirrors;
 		     mirror_index++) {
 			struct scrub_block *sblock;
-			struct scrub_page *page;
+			struct scrub_sector *sector;
 
-			sblock = sblocks_for_recheck + mirror_index;
+			sblock = sblocks_for_recheck[mirror_index];
 			sblock->sctx = sctx;
 
-			page = kzalloc(sizeof(*page), GFP_NOFS);
-			if (!page) {
-leave_nomem:
+			sector = alloc_scrub_sector(sblock, logical, GFP_NOFS);
+			if (!sector) {
 				spin_lock(&sctx->stat_lock);
 				sctx->stat.malloc_errors++;
 				spin_unlock(&sctx->stat_lock);
 				scrub_put_recover(fs_info, recover);
 				return -ENOMEM;
 			}
-			scrub_page_get(page);
-			sblock->pagev[page_index] = page;
-			page->sblock = sblock;
-			page->flags = flags;
-			page->generation = generation;
-			page->logical = logical;
-			page->have_csum = have_csum;
+			sector->flags = flags;
+			sector->generation = generation;
+			sector->have_csum = have_csum;
 			if (have_csum)
-				memcpy(page->csum,
-				       original_sblock->pagev[0]->csum,
-				       sctx->csum_size);
+				memcpy(sector->csum,
+				       original_sblock->sectors[0]->csum,
+				       sctx->fs_info->csum_size);
 
 			scrub_stripe_index_and_offset(logical,
-						      bbio->map_type,
-						      bbio->raid_map,
-						      mapped_length,
-						      bbio->num_stripes -
-						      bbio->num_tgtdevs,
+						      bioc->map_type,
+						      bioc->raid_map,
+						      bioc->num_stripes -
+						      bioc->num_tgtdevs,
 						      mirror_index,
 						      &stripe_index,
 						      &stripe_offset);
-			page->physical = bbio->stripes[stripe_index].physical +
-					 stripe_offset;
-			page->dev = bbio->stripes[stripe_index].dev;
-
-			BUG_ON(page_index >= original_sblock->page_count);
-			page->physical_for_dev_replace =
-				original_sblock->pagev[page_index]->
-				physical_for_dev_replace;
-			/* for missing devices, dev->bdev is NULL */
-			page->mirror_num = mirror_index + 1;
-			sblock->page_count++;
-			page->page = alloc_page(GFP_NOFS);
-			if (!page->page)
-				goto leave_nomem;
+			/*
+			 * We're at the first sector, also populate @sblock
+			 * physical and dev.
+			 */
+			if (sector_index == 0) {
+				sblock->physical =
+					bioc->stripes[stripe_index].physical +
+					stripe_offset;
+				sblock->dev = bioc->stripes[stripe_index].dev;
+				sblock->physical_for_dev_replace =
+					original_sblock->physical_for_dev_replace;
+			}
 
+			BUG_ON(sector_index >= original_sblock->sector_count);
 			scrub_get_recover(recover);
-			page->recover = recover;
+			sector->recover = recover;
 		}
 		scrub_put_recover(fs_info, recover);
 		length -= sublen;
 		logical += sublen;
-		page_index++;
+		sector_index++;
 	}
 
 	return 0;
@@ -1395,22 +1560,15 @@ static void scrub_bio_wait_endio(struct bio *bio)
 
 static int scrub_submit_raid56_bio_wait(struct btrfs_fs_info *fs_info,
 					struct bio *bio,
-					struct scrub_page *page)
+					struct scrub_sector *sector)
 {
 	DECLARE_COMPLETION_ONSTACK(done);
-	int ret;
-	int mirror_num;
 
-	bio->bi_iter.bi_sector = page->logical >> 9;
+	bio->bi_iter.bi_sector = (sector->offset + sector->sblock->logical) >>
+				 SECTOR_SHIFT;
 	bio->bi_private = &done;
 	bio->bi_end_io = scrub_bio_wait_endio;
-
-	mirror_num = page->sblock->pagev[0]->mirror_num;
-	ret = raid56_parity_recover(fs_info, bio, page->recover->bbio,
-				    page->recover->map_length,
-				    mirror_num, 0);
-	if (ret)
-		return ret;
+	raid56_parity_recover(bio, sector->recover->bioc, sector->sblock->mirror_num);
 
 	wait_for_completion_io(&done);
 	return blk_status_to_errno(bio->bi_status);
@@ -1419,26 +1577,24 @@ static int scrub_submit_raid56_bio_wait(struct btrfs_fs_info *fs_info,
 static void scrub_recheck_block_on_raid56(struct btrfs_fs_info *fs_info,
 					  struct scrub_block *sblock)
 {
-	struct scrub_page *first_page = sblock->pagev[0];
+	struct scrub_sector *first_sector = sblock->sectors[0];
 	struct bio *bio;
-	int page_num;
+	int i;
 
-	/* All pages in sblock belong to the same stripe on the same device. */
-	ASSERT(first_page->dev);
-	if (!first_page->dev->bdev)
+	/* All sectors in sblock belong to the same stripe on the same device. */
+	ASSERT(sblock->dev);
+	if (!sblock->dev->bdev)
 		goto out;
 
-	bio = btrfs_io_bio_alloc(BIO_MAX_PAGES);
-	bio_set_dev(bio, first_page->dev->bdev);
+	bio = bio_alloc(sblock->dev->bdev, BIO_MAX_VECS, REQ_OP_READ, GFP_NOFS);
 
-	for (page_num = 0; page_num < sblock->page_count; page_num++) {
-		struct scrub_page *page = sblock->pagev[page_num];
+	for (i = 0; i < sblock->sector_count; i++) {
+		struct scrub_sector *sector = sblock->sectors[i];
 
-		WARN_ON(!page->page);
-		bio_add_page(bio, page->page, PAGE_SIZE, 0);
+		bio_add_scrub_sector(bio, sector, fs_info->sectorsize);
 	}
 
-	if (scrub_submit_raid56_bio_wait(fs_info, bio, first_page)) {
+	if (scrub_submit_raid56_bio_wait(fs_info, bio, first_sector)) {
 		bio_put(bio);
 		goto out;
 	}
@@ -1449,65 +1605,63 @@ static void scrub_recheck_block_on_raid56(struct btrfs_fs_info *fs_info,
 
 	return;
 out:
-	for (page_num = 0; page_num < sblock->page_count; page_num++)
-		sblock->pagev[page_num]->io_error = 1;
+	for (i = 0; i < sblock->sector_count; i++)
+		sblock->sectors[i]->io_error = 1;
 
 	sblock->no_io_error_seen = 0;
 }
 
 /*
- * this function will check the on disk data for checksum errors, header
- * errors and read I/O errors. If any I/O errors happen, the exact pages
- * which are errored are marked as being bad. The goal is to enable scrub
- * to take those pages that are not errored from all the mirrors so that
- * the pages that are errored in the just handled mirror can be repaired.
+ * This function will check the on disk data for checksum errors, header errors
+ * and read I/O errors. If any I/O errors happen, the exact sectors which are
+ * errored are marked as being bad. The goal is to enable scrub to take those
+ * sectors that are not errored from all the mirrors so that the sectors that
+ * are errored in the just handled mirror can be repaired.
  */
 static void scrub_recheck_block(struct btrfs_fs_info *fs_info,
 				struct scrub_block *sblock,
 				int retry_failed_mirror)
 {
-	int page_num;
+	int i;
 
 	sblock->no_io_error_seen = 1;
 
 	/* short cut for raid56 */
-	if (!retry_failed_mirror && scrub_is_page_on_raid56(sblock->pagev[0]))
+	if (!retry_failed_mirror && scrub_is_page_on_raid56(sblock->sectors[0]))
 		return scrub_recheck_block_on_raid56(fs_info, sblock);
 
-	for (page_num = 0; page_num < sblock->page_count; page_num++) {
-		struct bio *bio;
-		struct scrub_page *page = sblock->pagev[page_num];
+	for (i = 0; i < sblock->sector_count; i++) {
+		struct scrub_sector *sector = sblock->sectors[i];
+		struct bio bio;
+		struct bio_vec bvec;
 
-		if (page->dev->bdev == NULL) {
-			page->io_error = 1;
+		if (sblock->dev->bdev == NULL) {
+			sector->io_error = 1;
 			sblock->no_io_error_seen = 0;
 			continue;
 		}
 
-		WARN_ON(!page->page);
-		bio = btrfs_io_bio_alloc(1);
-		bio_set_dev(bio, page->dev->bdev);
+		bio_init(&bio, sblock->dev->bdev, &bvec, 1, REQ_OP_READ);
+		bio_add_scrub_sector(&bio, sector, fs_info->sectorsize);
+		bio.bi_iter.bi_sector = (sblock->physical + sector->offset) >>
+					SECTOR_SHIFT;
 
-		bio_add_page(bio, page->page, PAGE_SIZE, 0);
-		bio->bi_iter.bi_sector = page->physical >> 9;
-		bio->bi_opf = REQ_OP_READ;
-
-		if (btrfsic_submit_bio_wait(bio)) {
-			page->io_error = 1;
+		btrfsic_check_bio(&bio);
+		if (submit_bio_wait(&bio)) {
+			sector->io_error = 1;
 			sblock->no_io_error_seen = 0;
 		}
 
-		bio_put(bio);
+		bio_uninit(&bio);
 	}
 
 	if (sblock->no_io_error_seen)
 		scrub_recheck_block_checksum(sblock);
 }
 
-static inline int scrub_check_fsid(u8 fsid[],
-				   struct scrub_page *spage)
+static inline int scrub_check_fsid(u8 fsid[], struct scrub_sector *sector)
 {
-	struct btrfs_fs_devices *fs_devices = spage->dev->fs_devices;
+	struct btrfs_fs_devices *fs_devices = sector->sblock->dev->fs_devices;
 	int ret;
 
 	ret = memcmp(fsid, fs_devices->fsid, BTRFS_FSID_SIZE);
@@ -1520,7 +1674,7 @@ static void scrub_recheck_block_checksum(struct scrub_block *sblock)
 	sblock->checksum_error = 0;
 	sblock->generation_error = 0;
 
-	if (sblock->pagev[0]->flags & BTRFS_EXTENT_FLAG_DATA)
+	if (sblock->sectors[0]->flags & BTRFS_EXTENT_FLAG_DATA)
 		scrub_checksum_data(sblock);
 	else
 		scrub_checksum_tree_block(sblock);
@@ -1529,15 +1683,14 @@ static void scrub_recheck_block_checksum(struct scrub_block *sblock)
 static int scrub_repair_block_from_good_copy(struct scrub_block *sblock_bad,
 					     struct scrub_block *sblock_good)
 {
-	int page_num;
+	int i;
 	int ret = 0;
 
-	for (page_num = 0; page_num < sblock_bad->page_count; page_num++) {
+	for (i = 0; i < sblock_bad->sector_count; i++) {
 		int ret_sub;
 
-		ret_sub = scrub_repair_page_from_good_copy(sblock_bad,
-							   sblock_good,
-							   page_num, 1);
+		ret_sub = scrub_repair_sector_from_good_copy(sblock_bad,
+							     sblock_good, i, 1);
 		if (ret_sub)
 			ret = ret_sub;
 	}
@@ -1545,46 +1698,42 @@ static int scrub_repair_block_from_good_copy(struct scrub_block *sblock_bad,
 	return ret;
 }
 
-static int scrub_repair_page_from_good_copy(struct scrub_block *sblock_bad,
-					    struct scrub_block *sblock_good,
-					    int page_num, int force_write)
+static int scrub_repair_sector_from_good_copy(struct scrub_block *sblock_bad,
+					      struct scrub_block *sblock_good,
+					      int sector_num, int force_write)
 {
-	struct scrub_page *page_bad = sblock_bad->pagev[page_num];
-	struct scrub_page *page_good = sblock_good->pagev[page_num];
+	struct scrub_sector *sector_bad = sblock_bad->sectors[sector_num];
+	struct scrub_sector *sector_good = sblock_good->sectors[sector_num];
 	struct btrfs_fs_info *fs_info = sblock_bad->sctx->fs_info;
+	const u32 sectorsize = fs_info->sectorsize;
 
-	BUG_ON(page_bad->page == NULL);
-	BUG_ON(page_good->page == NULL);
 	if (force_write || sblock_bad->header_error ||
-	    sblock_bad->checksum_error || page_bad->io_error) {
-		struct bio *bio;
+	    sblock_bad->checksum_error || sector_bad->io_error) {
+		struct bio bio;
+		struct bio_vec bvec;
 		int ret;
 
-		if (!page_bad->dev->bdev) {
+		if (!sblock_bad->dev->bdev) {
 			btrfs_warn_rl(fs_info,
 				"scrub_repair_page_from_good_copy(bdev == NULL) is unexpected");
 			return -EIO;
 		}
 
-		bio = btrfs_io_bio_alloc(1);
-		bio_set_dev(bio, page_bad->dev->bdev);
-		bio->bi_iter.bi_sector = page_bad->physical >> 9;
-		bio->bi_opf = REQ_OP_WRITE;
+		bio_init(&bio, sblock_bad->dev->bdev, &bvec, 1, REQ_OP_WRITE);
+		bio.bi_iter.bi_sector = (sblock_bad->physical +
+					 sector_bad->offset) >> SECTOR_SHIFT;
+		ret = bio_add_scrub_sector(&bio, sector_good, sectorsize);
 
-		ret = bio_add_page(bio, page_good->page, PAGE_SIZE, 0);
-		if (PAGE_SIZE != ret) {
-			bio_put(bio);
-			return -EIO;
-		}
+		btrfsic_check_bio(&bio);
+		ret = submit_bio_wait(&bio);
+		bio_uninit(&bio);
 
-		if (btrfsic_submit_bio_wait(bio)) {
-			btrfs_dev_stat_inc_and_print(page_bad->dev,
+		if (ret) {
+			btrfs_dev_stat_inc_and_print(sblock_bad->dev,
 				BTRFS_DEV_STAT_WRITE_ERRS);
 			atomic64_inc(&fs_info->dev_replace.num_write_errors);
-			bio_put(bio);
 			return -EIO;
 		}
-		bio_put(bio);
 	}
 
 	return 0;
@@ -1593,7 +1742,7 @@ static int scrub_repair_page_from_good_copy(struct scrub_block *sblock_bad,
 static void scrub_write_block_to_dev_replace(struct scrub_block *sblock)
 {
 	struct btrfs_fs_info *fs_info = sblock->sctx->fs_info;
-	int page_num;
+	int i;
 
 	/*
 	 * This block is used for the check of the parity on the source device,
@@ -1602,36 +1751,60 @@ static void scrub_write_block_to_dev_replace(struct scrub_block *sblock)
 	if (sblock->sparity)
 		return;
 
-	for (page_num = 0; page_num < sblock->page_count; page_num++) {
+	for (i = 0; i < sblock->sector_count; i++) {
 		int ret;
 
-		ret = scrub_write_page_to_dev_replace(sblock, page_num);
+		ret = scrub_write_sector_to_dev_replace(sblock, i);
 		if (ret)
 			atomic64_inc(&fs_info->dev_replace.num_write_errors);
 	}
 }
 
-static int scrub_write_page_to_dev_replace(struct scrub_block *sblock,
-					   int page_num)
+static int scrub_write_sector_to_dev_replace(struct scrub_block *sblock, int sector_num)
+{
+	const u32 sectorsize = sblock->sctx->fs_info->sectorsize;
+	struct scrub_sector *sector = sblock->sectors[sector_num];
+
+	if (sector->io_error)
+		memset(scrub_sector_get_kaddr(sector), 0, sectorsize);
+
+	return scrub_add_sector_to_wr_bio(sblock->sctx, sector);
+}
+
+static int fill_writer_pointer_gap(struct scrub_ctx *sctx, u64 physical)
 {
-	struct scrub_page *spage = sblock->pagev[page_num];
+	int ret = 0;
+	u64 length;
 
-	BUG_ON(spage->page == NULL);
-	if (spage->io_error) {
-		void *mapped_buffer = kmap_atomic(spage->page);
+	if (!btrfs_is_zoned(sctx->fs_info))
+		return 0;
+
+	if (!btrfs_dev_is_sequential(sctx->wr_tgtdev, physical))
+		return 0;
 
-		clear_page(mapped_buffer);
-		flush_dcache_page(spage->page);
-		kunmap_atomic(mapped_buffer);
+	if (sctx->write_pointer < physical) {
+		length = physical - sctx->write_pointer;
+
+		ret = btrfs_zoned_issue_zeroout(sctx->wr_tgtdev,
+						sctx->write_pointer, length);
+		if (!ret)
+			sctx->write_pointer = physical;
 	}
-	return scrub_add_page_to_wr_bio(sblock->sctx, spage);
+	return ret;
 }
 
-static int scrub_add_page_to_wr_bio(struct scrub_ctx *sctx,
-				    struct scrub_page *spage)
+static void scrub_block_get(struct scrub_block *sblock)
 {
+	refcount_inc(&sblock->refs);
+}
+
+static int scrub_add_sector_to_wr_bio(struct scrub_ctx *sctx,
+				      struct scrub_sector *sector)
+{
+	struct scrub_block *sblock = sector->sblock;
 	struct scrub_bio *sbio;
 	int ret;
+	const u32 sectorsize = sctx->fs_info->sectorsize;
 
 	mutex_lock(&sctx->wr_lock);
 again:
@@ -1643,38 +1816,39 @@ again:
 			return -ENOMEM;
 		}
 		sctx->wr_curr_bio->sctx = sctx;
-		sctx->wr_curr_bio->page_count = 0;
+		sctx->wr_curr_bio->sector_count = 0;
 	}
 	sbio = sctx->wr_curr_bio;
-	if (sbio->page_count == 0) {
-		struct bio *bio;
+	if (sbio->sector_count == 0) {
+		ret = fill_writer_pointer_gap(sctx, sector->offset +
+					      sblock->physical_for_dev_replace);
+		if (ret) {
+			mutex_unlock(&sctx->wr_lock);
+			return ret;
+		}
 
-		sbio->physical = spage->physical_for_dev_replace;
-		sbio->logical = spage->logical;
+		sbio->physical = sblock->physical_for_dev_replace + sector->offset;
+		sbio->logical = sblock->logical + sector->offset;
 		sbio->dev = sctx->wr_tgtdev;
-		bio = sbio->bio;
-		if (!bio) {
-			bio = btrfs_io_bio_alloc(sctx->pages_per_wr_bio);
-			sbio->bio = bio;
+		if (!sbio->bio) {
+			sbio->bio = bio_alloc(sbio->dev->bdev, sctx->sectors_per_bio,
+					      REQ_OP_WRITE, GFP_NOFS);
 		}
-
-		bio->bi_private = sbio;
-		bio->bi_end_io = scrub_wr_bio_end_io;
-		bio_set_dev(bio, sbio->dev->bdev);
-		bio->bi_iter.bi_sector = sbio->physical >> 9;
-		bio->bi_opf = REQ_OP_WRITE;
+		sbio->bio->bi_private = sbio;
+		sbio->bio->bi_end_io = scrub_wr_bio_end_io;
+		sbio->bio->bi_iter.bi_sector = sbio->physical >> 9;
 		sbio->status = 0;
-	} else if (sbio->physical + sbio->page_count * PAGE_SIZE !=
-		   spage->physical_for_dev_replace ||
-		   sbio->logical + sbio->page_count * PAGE_SIZE !=
-		   spage->logical) {
+	} else if (sbio->physical + sbio->sector_count * sectorsize !=
+		   sblock->physical_for_dev_replace + sector->offset ||
+		   sbio->logical + sbio->sector_count * sectorsize !=
+		   sblock->logical + sector->offset) {
 		scrub_wr_submit(sctx);
 		goto again;
 	}
 
-	ret = bio_add_page(sbio->bio, spage->page, PAGE_SIZE, 0);
-	if (ret != PAGE_SIZE) {
-		if (sbio->page_count < 1) {
+	ret = bio_add_scrub_sector(sbio->bio, sector, sectorsize);
+	if (ret != sectorsize) {
+		if (sbio->sector_count < 1) {
 			bio_put(sbio->bio);
 			sbio->bio = NULL;
 			mutex_unlock(&sctx->wr_lock);
@@ -1684,10 +1858,17 @@ again:
 		goto again;
 	}
 
-	sbio->pagev[sbio->page_count] = spage;
-	scrub_page_get(spage);
-	sbio->page_count++;
-	if (sbio->page_count == sctx->pages_per_wr_bio)
+	sbio->sectors[sbio->sector_count] = sector;
+	scrub_sector_get(sector);
+	/*
+	 * Since ssector no longer holds a page, but uses sblock::pages, we
+	 * have to ensure the sblock had not been freed before our write bio
+	 * finished.
+	 */
+	scrub_block_get(sector->sblock);
+
+	sbio->sector_count++;
+	if (sbio->sector_count == sctx->sectors_per_bio)
 		scrub_wr_submit(sctx);
 	mutex_unlock(&sctx->wr_lock);
 
@@ -1703,13 +1884,17 @@ static void scrub_wr_submit(struct scrub_ctx *sctx)
 
 	sbio = sctx->wr_curr_bio;
 	sctx->wr_curr_bio = NULL;
-	WARN_ON(!sbio->bio->bi_disk);
 	scrub_pending_bio_inc(sctx);
 	/* process all writes in a single worker thread. Then the block layer
 	 * orders the requests before sending them to the driver which
 	 * doubled the write performance on spinning disks when measured
 	 * with Linux 3.5 */
-	btrfsic_submit_bio(sbio->bio);
+	btrfsic_check_bio(sbio->bio);
+	submit_bio(sbio->bio);
+
+	if (btrfs_is_zoned(sctx->fs_info))
+		sctx->write_pointer = sbio->physical + sbio->sector_count *
+			sctx->fs_info->sectorsize;
 }
 
 static void scrub_wr_bio_end_io(struct bio *bio)
@@ -1720,31 +1905,37 @@ static void scrub_wr_bio_end_io(struct bio *bio)
 	sbio->status = bio->bi_status;
 	sbio->bio = bio;
 
-	btrfs_init_work(&sbio->work, scrub_wr_bio_end_io_worker, NULL, NULL);
-	btrfs_queue_work(fs_info->scrub_wr_completion_workers, &sbio->work);
+	INIT_WORK(&sbio->work, scrub_wr_bio_end_io_worker);
+	queue_work(fs_info->scrub_wr_completion_workers, &sbio->work);
 }
 
-static void scrub_wr_bio_end_io_worker(struct btrfs_work *work)
+static void scrub_wr_bio_end_io_worker(struct work_struct *work)
 {
 	struct scrub_bio *sbio = container_of(work, struct scrub_bio, work);
 	struct scrub_ctx *sctx = sbio->sctx;
 	int i;
 
-	WARN_ON(sbio->page_count > SCRUB_PAGES_PER_WR_BIO);
+	ASSERT(sbio->sector_count <= SCRUB_SECTORS_PER_BIO);
 	if (sbio->status) {
 		struct btrfs_dev_replace *dev_replace =
 			&sbio->sctx->fs_info->dev_replace;
 
-		for (i = 0; i < sbio->page_count; i++) {
-			struct scrub_page *spage = sbio->pagev[i];
+		for (i = 0; i < sbio->sector_count; i++) {
+			struct scrub_sector *sector = sbio->sectors[i];
 
-			spage->io_error = 1;
+			sector->io_error = 1;
 			atomic64_inc(&dev_replace->num_write_errors);
 		}
 	}
 
-	for (i = 0; i < sbio->page_count; i++)
-		scrub_page_put(sbio->pagev[i]);
+	/*
+	 * In scrub_add_sector_to_wr_bio() we grab extra ref for sblock, now in
+	 * endio we should put the sblock.
+	 */
+	for (i = 0; i < sbio->sector_count; i++) {
+		scrub_block_put(sbio->sectors[i]->sblock);
+		scrub_sector_put(sbio->sectors[i]);
+	}
 
 	bio_put(sbio->bio);
 	kfree(sbio);
@@ -1768,15 +1959,15 @@ static int scrub_checksum(struct scrub_block *sblock)
 	sblock->generation_error = 0;
 	sblock->checksum_error = 0;
 
-	WARN_ON(sblock->page_count < 1);
-	flags = sblock->pagev[0]->flags;
+	WARN_ON(sblock->sector_count < 1);
+	flags = sblock->sectors[0]->flags;
 	ret = 0;
 	if (flags & BTRFS_EXTENT_FLAG_DATA)
 		ret = scrub_checksum_data(sblock);
 	else if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK)
 		ret = scrub_checksum_tree_block(sblock);
 	else if (flags & BTRFS_EXTENT_FLAG_SUPER)
-		(void)scrub_checksum_super(sblock);
+		ret = scrub_checksum_super(sblock);
 	else
 		WARN_ON(1);
 	if (ret)
@@ -1791,44 +1982,23 @@ static int scrub_checksum_data(struct scrub_block *sblock)
 	struct btrfs_fs_info *fs_info = sctx->fs_info;
 	SHASH_DESC_ON_STACK(shash, fs_info->csum_shash);
 	u8 csum[BTRFS_CSUM_SIZE];
-	u8 *on_disk_csum;
-	struct page *page;
-	void *buffer;
-	u64 len;
-	int index;
+	struct scrub_sector *sector;
+	char *kaddr;
 
-	BUG_ON(sblock->page_count < 1);
-	if (!sblock->pagev[0]->have_csum)
+	BUG_ON(sblock->sector_count < 1);
+	sector = sblock->sectors[0];
+	if (!sector->have_csum)
 		return 0;
 
+	kaddr = scrub_sector_get_kaddr(sector);
+
 	shash->tfm = fs_info->csum_shash;
 	crypto_shash_init(shash);
 
-	on_disk_csum = sblock->pagev[0]->csum;
-	page = sblock->pagev[0]->page;
-	buffer = kmap_atomic(page);
-
-	len = sctx->fs_info->sectorsize;
-	index = 0;
-	for (;;) {
-		u64 l = min_t(u64, len, PAGE_SIZE);
-
-		crypto_shash_update(shash, buffer, l);
-		kunmap_atomic(buffer);
-		len -= l;
-		if (len == 0)
-			break;
-		index++;
-		BUG_ON(index >= sblock->page_count);
-		BUG_ON(!sblock->pagev[index]->page);
-		page = sblock->pagev[index]->page;
-		buffer = kmap_atomic(page);
-	}
+	crypto_shash_digest(shash, kaddr, fs_info->sectorsize, csum);
 
-	crypto_shash_final(shash, csum);
-	if (memcmp(csum, on_disk_csum, sctx->csum_size))
+	if (memcmp(csum, sector->csum, fs_info->csum_size))
 		sblock->checksum_error = 1;
-
 	return sblock->checksum_error;
 }
 
@@ -1840,65 +2010,59 @@ static int scrub_checksum_tree_block(struct scrub_block *sblock)
 	SHASH_DESC_ON_STACK(shash, fs_info->csum_shash);
 	u8 calculated_csum[BTRFS_CSUM_SIZE];
 	u8 on_disk_csum[BTRFS_CSUM_SIZE];
-	struct page *page;
-	void *mapped_buffer;
-	u64 mapped_size;
-	void *p;
-	u64 len;
-	int index;
+	/*
+	 * This is done in sectorsize steps even for metadata as there's a
+	 * constraint for nodesize to be aligned to sectorsize. This will need
+	 * to change so we don't misuse data and metadata units like that.
+	 */
+	const u32 sectorsize = sctx->fs_info->sectorsize;
+	const int num_sectors = fs_info->nodesize >> fs_info->sectorsize_bits;
+	int i;
+	struct scrub_sector *sector;
+	char *kaddr;
 
-	shash->tfm = fs_info->csum_shash;
-	crypto_shash_init(shash);
+	BUG_ON(sblock->sector_count < 1);
+
+	/* Each member in sectors is just one sector */
+	ASSERT(sblock->sector_count == num_sectors);
 
-	BUG_ON(sblock->page_count < 1);
-	page = sblock->pagev[0]->page;
-	mapped_buffer = kmap_atomic(page);
-	h = (struct btrfs_header *)mapped_buffer;
-	memcpy(on_disk_csum, h->csum, sctx->csum_size);
+	sector = sblock->sectors[0];
+	kaddr = scrub_sector_get_kaddr(sector);
+	h = (struct btrfs_header *)kaddr;
+	memcpy(on_disk_csum, h->csum, sctx->fs_info->csum_size);
 
 	/*
 	 * we don't use the getter functions here, as we
 	 * a) don't have an extent buffer and
 	 * b) the page is already kmapped
 	 */
-	if (sblock->pagev[0]->logical != btrfs_stack_header_bytenr(h))
+	if (sblock->logical != btrfs_stack_header_bytenr(h))
 		sblock->header_error = 1;
 
-	if (sblock->pagev[0]->generation != btrfs_stack_header_generation(h)) {
+	if (sector->generation != btrfs_stack_header_generation(h)) {
 		sblock->header_error = 1;
 		sblock->generation_error = 1;
 	}
 
-	if (!scrub_check_fsid(h->fsid, sblock->pagev[0]))
+	if (!scrub_check_fsid(h->fsid, sector))
 		sblock->header_error = 1;
 
 	if (memcmp(h->chunk_tree_uuid, fs_info->chunk_tree_uuid,
 		   BTRFS_UUID_SIZE))
 		sblock->header_error = 1;
 
-	len = sctx->fs_info->nodesize - BTRFS_CSUM_SIZE;
-	mapped_size = PAGE_SIZE - BTRFS_CSUM_SIZE;
-	p = ((u8 *)mapped_buffer) + BTRFS_CSUM_SIZE;
-	index = 0;
-	for (;;) {
-		u64 l = min_t(u64, len, mapped_size);
+	shash->tfm = fs_info->csum_shash;
+	crypto_shash_init(shash);
+	crypto_shash_update(shash, kaddr + BTRFS_CSUM_SIZE,
+			    sectorsize - BTRFS_CSUM_SIZE);
 
-		crypto_shash_update(shash, p, l);
-		kunmap_atomic(mapped_buffer);
-		len -= l;
-		if (len == 0)
-			break;
-		index++;
-		BUG_ON(index >= sblock->page_count);
-		BUG_ON(!sblock->pagev[index]->page);
-		page = sblock->pagev[index]->page;
-		mapped_buffer = kmap_atomic(page);
-		mapped_size = PAGE_SIZE;
-		p = mapped_buffer;
+	for (i = 1; i < num_sectors; i++) {
+		kaddr = scrub_sector_get_kaddr(sblock->sectors[i]);
+		crypto_shash_update(shash, kaddr, sectorsize);
 	}
 
 	crypto_shash_final(shash, calculated_csum);
-	if (memcmp(calculated_csum, on_disk_csum, sctx->csum_size))
+	if (memcmp(calculated_csum, on_disk_csum, sctx->fs_info->csum_size))
 		sblock->checksum_error = 1;
 
 	return sblock->header_error || sblock->checksum_error;
@@ -1911,84 +2075,36 @@ static int scrub_checksum_super(struct scrub_block *sblock)
 	struct btrfs_fs_info *fs_info = sctx->fs_info;
 	SHASH_DESC_ON_STACK(shash, fs_info->csum_shash);
 	u8 calculated_csum[BTRFS_CSUM_SIZE];
-	u8 on_disk_csum[BTRFS_CSUM_SIZE];
-	struct page *page;
-	void *mapped_buffer;
-	u64 mapped_size;
-	void *p;
+	struct scrub_sector *sector;
+	char *kaddr;
 	int fail_gen = 0;
 	int fail_cor = 0;
-	u64 len;
-	int index;
-
-	shash->tfm = fs_info->csum_shash;
-	crypto_shash_init(shash);
 
-	BUG_ON(sblock->page_count < 1);
-	page = sblock->pagev[0]->page;
-	mapped_buffer = kmap_atomic(page);
-	s = (struct btrfs_super_block *)mapped_buffer;
-	memcpy(on_disk_csum, s->csum, sctx->csum_size);
+	BUG_ON(sblock->sector_count < 1);
+	sector = sblock->sectors[0];
+	kaddr = scrub_sector_get_kaddr(sector);
+	s = (struct btrfs_super_block *)kaddr;
 
-	if (sblock->pagev[0]->logical != btrfs_super_bytenr(s))
+	if (sblock->logical != btrfs_super_bytenr(s))
 		++fail_cor;
 
-	if (sblock->pagev[0]->generation != btrfs_super_generation(s))
+	if (sector->generation != btrfs_super_generation(s))
 		++fail_gen;
 
-	if (!scrub_check_fsid(s->fsid, sblock->pagev[0]))
+	if (!scrub_check_fsid(s->fsid, sector))
 		++fail_cor;
 
-	len = BTRFS_SUPER_INFO_SIZE - BTRFS_CSUM_SIZE;
-	mapped_size = PAGE_SIZE - BTRFS_CSUM_SIZE;
-	p = ((u8 *)mapped_buffer) + BTRFS_CSUM_SIZE;
-	index = 0;
-	for (;;) {
-		u64 l = min_t(u64, len, mapped_size);
-
-		crypto_shash_update(shash, p, l);
-		kunmap_atomic(mapped_buffer);
-		len -= l;
-		if (len == 0)
-			break;
-		index++;
-		BUG_ON(index >= sblock->page_count);
-		BUG_ON(!sblock->pagev[index]->page);
-		page = sblock->pagev[index]->page;
-		mapped_buffer = kmap_atomic(page);
-		mapped_size = PAGE_SIZE;
-		p = mapped_buffer;
-	}
+	shash->tfm = fs_info->csum_shash;
+	crypto_shash_init(shash);
+	crypto_shash_digest(shash, kaddr + BTRFS_CSUM_SIZE,
+			BTRFS_SUPER_INFO_SIZE - BTRFS_CSUM_SIZE, calculated_csum);
 
-	crypto_shash_final(shash, calculated_csum);
-	if (memcmp(calculated_csum, on_disk_csum, sctx->csum_size))
+	if (memcmp(calculated_csum, s->csum, sctx->fs_info->csum_size))
 		++fail_cor;
 
-	if (fail_cor + fail_gen) {
-		/*
-		 * if we find an error in a super block, we just report it.
-		 * They will get written with the next transaction commit
-		 * anyway
-		 */
-		spin_lock(&sctx->stat_lock);
-		++sctx->stat.super_errors;
-		spin_unlock(&sctx->stat_lock);
-		if (fail_cor)
-			btrfs_dev_stat_inc_and_print(sblock->pagev[0]->dev,
-				BTRFS_DEV_STAT_CORRUPTION_ERRS);
-		else
-			btrfs_dev_stat_inc_and_print(sblock->pagev[0]->dev,
-				BTRFS_DEV_STAT_GENERATION_ERRS);
-	}
-
 	return fail_cor + fail_gen;
 }
 
-static void scrub_block_get(struct scrub_block *sblock)
-{
-	refcount_inc(&sblock->refs);
-}
-
 static void scrub_block_put(struct scrub_block *sblock)
 {
 	if (refcount_dec_and_test(&sblock->refs)) {
@@ -1997,24 +2113,86 @@ static void scrub_block_put(struct scrub_block *sblock)
 		if (sblock->sparity)
 			scrub_parity_put(sblock->sparity);
 
-		for (i = 0; i < sblock->page_count; i++)
-			scrub_page_put(sblock->pagev[i]);
+		for (i = 0; i < sblock->sector_count; i++)
+			scrub_sector_put(sblock->sectors[i]);
+		for (i = 0; i < DIV_ROUND_UP(sblock->len, PAGE_SIZE); i++) {
+			if (sblock->pages[i]) {
+				detach_scrub_page_private(sblock->pages[i]);
+				__free_page(sblock->pages[i]);
+			}
+		}
 		kfree(sblock);
 	}
 }
 
-static void scrub_page_get(struct scrub_page *spage)
+static void scrub_sector_get(struct scrub_sector *sector)
+{
+	atomic_inc(&sector->refs);
+}
+
+static void scrub_sector_put(struct scrub_sector *sector)
 {
-	atomic_inc(&spage->refs);
+	if (atomic_dec_and_test(&sector->refs))
+		kfree(sector);
 }
 
-static void scrub_page_put(struct scrub_page *spage)
+/*
+ * Throttling of IO submission, bandwidth-limit based, the timeslice is 1
+ * second.  Limit can be set via /sys/fs/UUID/devinfo/devid/scrub_speed_max.
+ */
+static void scrub_throttle(struct scrub_ctx *sctx)
 {
-	if (atomic_dec_and_test(&spage->refs)) {
-		if (spage->page)
-			__free_page(spage->page);
-		kfree(spage);
+	const int time_slice = 1000;
+	struct scrub_bio *sbio;
+	struct btrfs_device *device;
+	s64 delta;
+	ktime_t now;
+	u32 div;
+	u64 bwlimit;
+
+	sbio = sctx->bios[sctx->curr];
+	device = sbio->dev;
+	bwlimit = READ_ONCE(device->scrub_speed_max);
+	if (bwlimit == 0)
+		return;
+
+	/*
+	 * Slice is divided into intervals when the IO is submitted, adjust by
+	 * bwlimit and maximum of 64 intervals.
+	 */
+	div = max_t(u32, 1, (u32)(bwlimit / (16 * 1024 * 1024)));
+	div = min_t(u32, 64, div);
+
+	/* Start new epoch, set deadline */
+	now = ktime_get();
+	if (sctx->throttle_deadline == 0) {
+		sctx->throttle_deadline = ktime_add_ms(now, time_slice / div);
+		sctx->throttle_sent = 0;
+	}
+
+	/* Still in the time to send? */
+	if (ktime_before(now, sctx->throttle_deadline)) {
+		/* If current bio is within the limit, send it */
+		sctx->throttle_sent += sbio->bio->bi_iter.bi_size;
+		if (sctx->throttle_sent <= div_u64(bwlimit, div))
+			return;
+
+		/* We're over the limit, sleep until the rest of the slice */
+		delta = ktime_ms_delta(sctx->throttle_deadline, now);
+	} else {
+		/* New request after deadline, start new epoch */
+		delta = 0;
+	}
+
+	if (delta) {
+		long timeout;
+
+		timeout = div_u64(delta * HZ, 1000);
+		schedule_timeout_interruptible(timeout);
 	}
+
+	/* Next call will start the deadline period */
+	sctx->throttle_deadline = 0;
 }
 
 static void scrub_submit(struct scrub_ctx *sctx)
@@ -2024,17 +2202,21 @@ static void scrub_submit(struct scrub_ctx *sctx)
 	if (sctx->curr == -1)
 		return;
 
+	scrub_throttle(sctx);
+
 	sbio = sctx->bios[sctx->curr];
 	sctx->curr = -1;
 	scrub_pending_bio_inc(sctx);
-	btrfsic_submit_bio(sbio->bio);
+	btrfsic_check_bio(sbio->bio);
+	submit_bio(sbio->bio);
 }
 
-static int scrub_add_page_to_rd_bio(struct scrub_ctx *sctx,
-				    struct scrub_page *spage)
+static int scrub_add_sector_to_rd_bio(struct scrub_ctx *sctx,
+				      struct scrub_sector *sector)
 {
-	struct scrub_block *sblock = spage->sblock;
+	struct scrub_block *sblock = sector->sblock;
 	struct scrub_bio *sbio;
+	const u32 sectorsize = sctx->fs_info->sectorsize;
 	int ret;
 
 again:
@@ -2047,7 +2229,7 @@ again:
 		if (sctx->curr != -1) {
 			sctx->first_free = sctx->bios[sctx->curr]->next_free;
 			sctx->bios[sctx->curr]->next_free = -1;
-			sctx->bios[sctx->curr]->page_count = 0;
+			sctx->bios[sctx->curr]->sector_count = 0;
 			spin_unlock(&sctx->list_lock);
 		} else {
 			spin_unlock(&sctx->list_lock);
@@ -2055,37 +2237,31 @@ again:
 		}
 	}
 	sbio = sctx->bios[sctx->curr];
-	if (sbio->page_count == 0) {
-		struct bio *bio;
-
-		sbio->physical = spage->physical;
-		sbio->logical = spage->logical;
-		sbio->dev = spage->dev;
-		bio = sbio->bio;
-		if (!bio) {
-			bio = btrfs_io_bio_alloc(sctx->pages_per_rd_bio);
-			sbio->bio = bio;
+	if (sbio->sector_count == 0) {
+		sbio->physical = sblock->physical + sector->offset;
+		sbio->logical = sblock->logical + sector->offset;
+		sbio->dev = sblock->dev;
+		if (!sbio->bio) {
+			sbio->bio = bio_alloc(sbio->dev->bdev, sctx->sectors_per_bio,
+					      REQ_OP_READ, GFP_NOFS);
 		}
-
-		bio->bi_private = sbio;
-		bio->bi_end_io = scrub_bio_end_io;
-		bio_set_dev(bio, sbio->dev->bdev);
-		bio->bi_iter.bi_sector = sbio->physical >> 9;
-		bio->bi_opf = REQ_OP_READ;
+		sbio->bio->bi_private = sbio;
+		sbio->bio->bi_end_io = scrub_bio_end_io;
+		sbio->bio->bi_iter.bi_sector = sbio->physical >> 9;
 		sbio->status = 0;
-	} else if (sbio->physical + sbio->page_count * PAGE_SIZE !=
-		   spage->physical ||
-		   sbio->logical + sbio->page_count * PAGE_SIZE !=
-		   spage->logical ||
-		   sbio->dev != spage->dev) {
+	} else if (sbio->physical + sbio->sector_count * sectorsize !=
+		   sblock->physical + sector->offset ||
+		   sbio->logical + sbio->sector_count * sectorsize !=
+		   sblock->logical + sector->offset ||
+		   sbio->dev != sblock->dev) {
 		scrub_submit(sctx);
 		goto again;
 	}
 
-	sbio->pagev[sbio->page_count] = spage;
-	ret = bio_add_page(sbio->bio, spage->page, PAGE_SIZE, 0);
-	if (ret != PAGE_SIZE) {
-		if (sbio->page_count < 1) {
+	sbio->sectors[sbio->sector_count] = sector;
+	ret = bio_add_scrub_sector(sbio->bio, sector, sectorsize);
+	if (ret != sectorsize) {
+		if (sbio->sector_count < 1) {
 			bio_put(sbio->bio);
 			sbio->bio = NULL;
 			return -EIO;
@@ -2095,9 +2271,9 @@ again:
 	}
 
 	scrub_block_get(sblock); /* one for the page added to the bio */
-	atomic_inc(&sblock->outstanding_pages);
-	sbio->page_count++;
-	if (sbio->page_count == sctx->pages_per_rd_bio)
+	atomic_inc(&sblock->outstanding_sectors);
+	sbio->sector_count++;
+	if (sbio->sector_count == sctx->sectors_per_bio)
 		scrub_submit(sctx);
 
 	return 0;
@@ -2108,15 +2284,16 @@ static void scrub_missing_raid56_end_io(struct bio *bio)
 	struct scrub_block *sblock = bio->bi_private;
 	struct btrfs_fs_info *fs_info = sblock->sctx->fs_info;
 
+	btrfs_bio_counter_dec(fs_info);
 	if (bio->bi_status)
 		sblock->no_io_error_seen = 0;
 
 	bio_put(bio);
 
-	btrfs_queue_work(fs_info->scrub_workers, &sblock->work);
+	queue_work(fs_info->scrub_workers, &sblock->work);
 }
 
-static void scrub_missing_raid56_worker(struct btrfs_work *work)
+static void scrub_missing_raid56_worker(struct work_struct *work)
 {
 	struct scrub_block *sblock = container_of(work, struct scrub_block, work);
 	struct scrub_ctx *sctx = sblock->sctx;
@@ -2124,8 +2301,8 @@ static void scrub_missing_raid56_worker(struct btrfs_work *work)
 	u64 logical;
 	struct btrfs_device *dev;
 
-	logical = sblock->pagev[0]->logical;
-	dev = sblock->pagev[0]->dev;
+	logical = sblock->logical;
+	dev = sblock->dev;
 
 	if (sblock->no_io_error_seen)
 		scrub_recheck_block_checksum(sblock);
@@ -2162,9 +2339,9 @@ static void scrub_missing_raid56_pages(struct scrub_block *sblock)
 {
 	struct scrub_ctx *sctx = sblock->sctx;
 	struct btrfs_fs_info *fs_info = sctx->fs_info;
-	u64 length = sblock->page_count * PAGE_SIZE;
-	u64 logical = sblock->pagev[0]->logical;
-	struct btrfs_bio *bbio = NULL;
+	u64 length = sblock->sector_count << fs_info->sectorsize_bits;
+	u64 logical = sblock->logical;
+	struct btrfs_io_context *bioc = NULL;
 	struct bio *bio;
 	struct btrfs_raid_bio *rbio;
 	int ret;
@@ -2172,61 +2349,66 @@ static void scrub_missing_raid56_pages(struct scrub_block *sblock)
 
 	btrfs_bio_counter_inc_blocked(fs_info);
 	ret = btrfs_map_sblock(fs_info, BTRFS_MAP_GET_READ_MIRRORS, logical,
-			&length, &bbio);
-	if (ret || !bbio || !bbio->raid_map)
-		goto bbio_out;
+			       &length, &bioc);
+	if (ret || !bioc || !bioc->raid_map)
+		goto bioc_out;
 
 	if (WARN_ON(!sctx->is_dev_replace ||
-		    !(bbio->map_type & BTRFS_BLOCK_GROUP_RAID56_MASK))) {
+		    !(bioc->map_type & BTRFS_BLOCK_GROUP_RAID56_MASK))) {
 		/*
 		 * We shouldn't be scrubbing a missing device. Even for dev
 		 * replace, we should only get here for RAID 5/6. We either
 		 * managed to mount something with no mirrors remaining or
-		 * there's a bug in scrub_remap_extent()/btrfs_map_block().
+		 * there's a bug in scrub_find_good_copy()/btrfs_map_block().
 		 */
-		goto bbio_out;
+		goto bioc_out;
 	}
 
-	bio = btrfs_io_bio_alloc(0);
+	bio = bio_alloc(NULL, BIO_MAX_VECS, REQ_OP_READ, GFP_NOFS);
 	bio->bi_iter.bi_sector = logical >> 9;
 	bio->bi_private = sblock;
 	bio->bi_end_io = scrub_missing_raid56_end_io;
 
-	rbio = raid56_alloc_missing_rbio(fs_info, bio, bbio, length);
+	rbio = raid56_alloc_missing_rbio(bio, bioc);
 	if (!rbio)
 		goto rbio_out;
 
-	for (i = 0; i < sblock->page_count; i++) {
-		struct scrub_page *spage = sblock->pagev[i];
+	for (i = 0; i < sblock->sector_count; i++) {
+		struct scrub_sector *sector = sblock->sectors[i];
 
-		raid56_add_scrub_pages(rbio, spage->page, spage->logical);
+		raid56_add_scrub_pages(rbio, scrub_sector_get_page(sector),
+				       scrub_sector_get_page_offset(sector),
+				       sector->offset + sector->sblock->logical);
 	}
 
-	btrfs_init_work(&sblock->work, scrub_missing_raid56_worker, NULL, NULL);
+	INIT_WORK(&sblock->work, scrub_missing_raid56_worker);
 	scrub_block_get(sblock);
 	scrub_pending_bio_inc(sctx);
 	raid56_submit_missing_rbio(rbio);
+	btrfs_put_bioc(bioc);
 	return;
 
 rbio_out:
 	bio_put(bio);
-bbio_out:
+bioc_out:
 	btrfs_bio_counter_dec(fs_info);
-	btrfs_put_bbio(bbio);
+	btrfs_put_bioc(bioc);
 	spin_lock(&sctx->stat_lock);
 	sctx->stat.malloc_errors++;
 	spin_unlock(&sctx->stat_lock);
 }
 
-static int scrub_pages(struct scrub_ctx *sctx, u64 logical, u64 len,
+static int scrub_sectors(struct scrub_ctx *sctx, u64 logical, u32 len,
 		       u64 physical, struct btrfs_device *dev, u64 flags,
-		       u64 gen, int mirror_num, u8 *csum, int force,
+		       u64 gen, int mirror_num, u8 *csum,
 		       u64 physical_for_dev_replace)
 {
 	struct scrub_block *sblock;
+	const u32 sectorsize = sctx->fs_info->sectorsize;
 	int index;
 
-	sblock = kzalloc(sizeof(*sblock), GFP_KERNEL);
+	sblock = alloc_scrub_block(sctx, dev, logical, physical,
+				   physical_for_dev_replace, mirror_num);
 	if (!sblock) {
 		spin_lock(&sctx->stat_lock);
 		sctx->stat.malloc_errors++;
@@ -2234,53 +2416,38 @@ static int scrub_pages(struct scrub_ctx *sctx, u64 logical, u64 len,
 		return -ENOMEM;
 	}
 
-	/* one ref inside this function, plus one for each page added to
-	 * a bio later on */
-	refcount_set(&sblock->refs, 1);
-	sblock->sctx = sctx;
-	sblock->no_io_error_seen = 1;
-
 	for (index = 0; len > 0; index++) {
-		struct scrub_page *spage;
-		u64 l = min_t(u64, len, PAGE_SIZE);
+		struct scrub_sector *sector;
+		/*
+		 * Here we will allocate one page for one sector to scrub.
+		 * This is fine if PAGE_SIZE == sectorsize, but will cost
+		 * more memory for PAGE_SIZE > sectorsize case.
+		 */
+		u32 l = min(sectorsize, len);
 
-		spage = kzalloc(sizeof(*spage), GFP_KERNEL);
-		if (!spage) {
-leave_nomem:
+		sector = alloc_scrub_sector(sblock, logical, GFP_KERNEL);
+		if (!sector) {
 			spin_lock(&sctx->stat_lock);
 			sctx->stat.malloc_errors++;
 			spin_unlock(&sctx->stat_lock);
 			scrub_block_put(sblock);
 			return -ENOMEM;
 		}
-		BUG_ON(index >= SCRUB_MAX_PAGES_PER_BLOCK);
-		scrub_page_get(spage);
-		sblock->pagev[index] = spage;
-		spage->sblock = sblock;
-		spage->dev = dev;
-		spage->flags = flags;
-		spage->generation = gen;
-		spage->logical = logical;
-		spage->physical = physical;
-		spage->physical_for_dev_replace = physical_for_dev_replace;
-		spage->mirror_num = mirror_num;
+		sector->flags = flags;
+		sector->generation = gen;
 		if (csum) {
-			spage->have_csum = 1;
-			memcpy(spage->csum, csum, sctx->csum_size);
+			sector->have_csum = 1;
+			memcpy(sector->csum, csum, sctx->fs_info->csum_size);
 		} else {
-			spage->have_csum = 0;
+			sector->have_csum = 0;
 		}
-		sblock->page_count++;
-		spage->page = alloc_page(GFP_KERNEL);
-		if (!spage->page)
-			goto leave_nomem;
 		len -= l;
 		logical += l;
 		physical += l;
 		physical_for_dev_replace += l;
 	}
 
-	WARN_ON(sblock->page_count == 0);
+	WARN_ON(sblock->sector_count == 0);
 	if (test_bit(BTRFS_DEV_STATE_MISSING, &dev->dev_state)) {
 		/*
 		 * This case should only be hit for RAID 5/6 device replace. See
@@ -2288,18 +2455,18 @@ leave_nomem:
 		 */
 		scrub_missing_raid56_pages(sblock);
 	} else {
-		for (index = 0; index < sblock->page_count; index++) {
-			struct scrub_page *spage = sblock->pagev[index];
+		for (index = 0; index < sblock->sector_count; index++) {
+			struct scrub_sector *sector = sblock->sectors[index];
 			int ret;
 
-			ret = scrub_add_page_to_rd_bio(sctx, spage);
+			ret = scrub_add_sector_to_rd_bio(sctx, sector);
 			if (ret) {
 				scrub_block_put(sblock);
 				return ret;
 			}
 		}
 
-		if (force)
+		if (flags & BTRFS_EXTENT_FLAG_SUPER)
 			scrub_submit(sctx);
 	}
 
@@ -2316,31 +2483,31 @@ static void scrub_bio_end_io(struct bio *bio)
 	sbio->status = bio->bi_status;
 	sbio->bio = bio;
 
-	btrfs_queue_work(fs_info->scrub_workers, &sbio->work);
+	queue_work(fs_info->scrub_workers, &sbio->work);
 }
 
-static void scrub_bio_end_io_worker(struct btrfs_work *work)
+static void scrub_bio_end_io_worker(struct work_struct *work)
 {
 	struct scrub_bio *sbio = container_of(work, struct scrub_bio, work);
 	struct scrub_ctx *sctx = sbio->sctx;
 	int i;
 
-	BUG_ON(sbio->page_count > SCRUB_PAGES_PER_RD_BIO);
+	ASSERT(sbio->sector_count <= SCRUB_SECTORS_PER_BIO);
 	if (sbio->status) {
-		for (i = 0; i < sbio->page_count; i++) {
-			struct scrub_page *spage = sbio->pagev[i];
+		for (i = 0; i < sbio->sector_count; i++) {
+			struct scrub_sector *sector = sbio->sectors[i];
 
-			spage->io_error = 1;
-			spage->sblock->no_io_error_seen = 0;
+			sector->io_error = 1;
+			sector->sblock->no_io_error_seen = 0;
 		}
 	}
 
-	/* now complete the scrub_block items that have all pages completed */
-	for (i = 0; i < sbio->page_count; i++) {
-		struct scrub_page *spage = sbio->pagev[i];
-		struct scrub_block *sblock = spage->sblock;
+	/* Now complete the scrub_block items that have all pages completed */
+	for (i = 0; i < sbio->sector_count; i++) {
+		struct scrub_sector *sector = sbio->sectors[i];
+		struct scrub_block *sblock = sector->sblock;
 
-		if (atomic_dec_and_test(&sblock->outstanding_pages))
+		if (atomic_dec_and_test(&sblock->outstanding_sectors))
 			scrub_block_complete(sblock);
 		scrub_block_put(sblock);
 	}
@@ -2363,12 +2530,11 @@ static void scrub_bio_end_io_worker(struct btrfs_work *work)
 
 static inline void __scrub_mark_bitmap(struct scrub_parity *sparity,
 				       unsigned long *bitmap,
-				       u64 start, u64 len)
+				       u64 start, u32 len)
 {
 	u64 offset;
-	u64 nsectors64;
 	u32 nsectors;
-	int sectorsize = sparity->sctx->fs_info->sectorsize;
+	u32 sectorsize_bits = sparity->sctx->fs_info->sectorsize_bits;
 
 	if (len >= sparity->stripe_len) {
 		bitmap_set(bitmap, 0, sparity->nsectors);
@@ -2377,11 +2543,8 @@ static inline void __scrub_mark_bitmap(struct scrub_parity *sparity,
 
 	start -= sparity->logic_start;
 	start = div64_u64_rem(start, sparity->stripe_len, &offset);
-	offset = div_u64(offset, sectorsize);
-	nsectors64 = div_u64(len, sectorsize);
-
-	ASSERT(nsectors64 < UINT_MAX);
-	nsectors = (u32)nsectors64;
+	offset = offset >> sectorsize_bits;
+	nsectors = len >> sectorsize_bits;
 
 	if (offset + nsectors <= sparity->nsectors) {
 		bitmap_set(bitmap, offset, nsectors);
@@ -2393,15 +2556,15 @@ static inline void __scrub_mark_bitmap(struct scrub_parity *sparity,
 }
 
 static inline void scrub_parity_mark_sectors_error(struct scrub_parity *sparity,
-						   u64 start, u64 len)
+						   u64 start, u32 len)
 {
-	__scrub_mark_bitmap(sparity, sparity->ebitmap, start, len);
+	__scrub_mark_bitmap(sparity, &sparity->ebitmap, start, len);
 }
 
 static inline void scrub_parity_mark_sectors_data(struct scrub_parity *sparity,
-						  u64 start, u64 len)
+						  u64 start, u32 len)
 {
-	__scrub_mark_bitmap(sparity, sparity->dbitmap, start, len);
+	__scrub_mark_bitmap(sparity, &sparity->dbitmap, start, len);
 }
 
 static void scrub_block_complete(struct scrub_block *sblock)
@@ -2423,55 +2586,88 @@ static void scrub_block_complete(struct scrub_block *sblock)
 	}
 
 	if (sblock->sparity && corrupted && !sblock->data_corrected) {
-		u64 start = sblock->pagev[0]->logical;
-		u64 end = sblock->pagev[sblock->page_count - 1]->logical +
-			  PAGE_SIZE;
+		u64 start = sblock->logical;
+		u64 end = sblock->logical +
+			  sblock->sectors[sblock->sector_count - 1]->offset +
+			  sblock->sctx->fs_info->sectorsize;
 
+		ASSERT(end - start <= U32_MAX);
 		scrub_parity_mark_sectors_error(sblock->sparity,
 						start, end - start);
 	}
 }
 
+static void drop_csum_range(struct scrub_ctx *sctx, struct btrfs_ordered_sum *sum)
+{
+	sctx->stat.csum_discards += sum->len >> sctx->fs_info->sectorsize_bits;
+	list_del(&sum->list);
+	kfree(sum);
+}
+
+/*
+ * Find the desired csum for range [logical, logical + sectorsize), and store
+ * the csum into @csum.
+ *
+ * The search source is sctx->csum_list, which is a pre-populated list
+ * storing bytenr ordered csum ranges.  We're responsible to cleanup any range
+ * that is before @logical.
+ *
+ * Return 0 if there is no csum for the range.
+ * Return 1 if there is csum for the range and copied to @csum.
+ */
 static int scrub_find_csum(struct scrub_ctx *sctx, u64 logical, u8 *csum)
 {
-	struct btrfs_ordered_sum *sum = NULL;
-	unsigned long index;
-	unsigned long num_sectors;
+	bool found = false;
 
 	while (!list_empty(&sctx->csum_list)) {
+		struct btrfs_ordered_sum *sum = NULL;
+		unsigned long index;
+		unsigned long num_sectors;
+
 		sum = list_first_entry(&sctx->csum_list,
 				       struct btrfs_ordered_sum, list);
+		/* The current csum range is beyond our range, no csum found */
 		if (sum->bytenr > logical)
-			return 0;
-		if (sum->bytenr + sum->len > logical)
 			break;
 
-		++sctx->stat.csum_discards;
-		list_del(&sum->list);
-		kfree(sum);
-		sum = NULL;
-	}
-	if (!sum)
-		return 0;
+		/*
+		 * The current sum is before our bytenr, since scrub is always
+		 * done in bytenr order, the csum will never be used anymore,
+		 * clean it up so that later calls won't bother with the range,
+		 * and continue search the next range.
+		 */
+		if (sum->bytenr + sum->len <= logical) {
+			drop_csum_range(sctx, sum);
+			continue;
+		}
 
-	index = div_u64(logical - sum->bytenr, sctx->fs_info->sectorsize);
-	ASSERT(index < UINT_MAX);
+		/* Now the csum range covers our bytenr, copy the csum */
+		found = true;
+		index = (logical - sum->bytenr) >> sctx->fs_info->sectorsize_bits;
+		num_sectors = sum->len >> sctx->fs_info->sectorsize_bits;
 
-	num_sectors = sum->len / sctx->fs_info->sectorsize;
-	memcpy(csum, sum->sums + index * sctx->csum_size, sctx->csum_size);
-	if (index == num_sectors - 1) {
-		list_del(&sum->list);
-		kfree(sum);
+		memcpy(csum, sum->sums + index * sctx->fs_info->csum_size,
+		       sctx->fs_info->csum_size);
+
+		/* Cleanup the range if we're at the end of the csum range */
+		if (index == num_sectors - 1)
+			drop_csum_range(sctx, sum);
+		break;
 	}
+	if (!found)
+		return 0;
 	return 1;
 }
 
 /* scrub extent tries to collect up to 64 kB for each bio */
 static int scrub_extent(struct scrub_ctx *sctx, struct map_lookup *map,
-			u64 logical, u64 len,
+			u64 logical, u32 len,
 			u64 physical, struct btrfs_device *dev, u64 flags,
-			u64 gen, int mirror_num, u64 physical_for_dev_replace)
+			u64 gen, int mirror_num)
 {
+	struct btrfs_device *src_dev = dev;
+	u64 src_physical = physical;
+	int src_mirror = mirror_num;
 	int ret;
 	u8 csum[BTRFS_CSUM_SIZE];
 	u32 blocksize;
@@ -2499,8 +2695,20 @@ static int scrub_extent(struct scrub_ctx *sctx, struct map_lookup *map,
 		WARN_ON(1);
 	}
 
+	/*
+	 * For dev-replace case, we can have @dev being a missing device.
+	 * Regular scrub will avoid its execution on missing device at all,
+	 * as that would trigger tons of read error.
+	 *
+	 * Reading from missing device will cause read error counts to
+	 * increase unnecessarily.
+	 * So here we change the read source to a good mirror.
+	 */
+	if (sctx->is_dev_replace && !dev->bdev)
+		scrub_find_good_copy(sctx->fs_info, logical, len, &src_physical,
+				     &src_dev, &src_mirror);
 	while (len) {
-		u64 l = min_t(u64, len, blocksize);
+		u32 l = min(len, blocksize);
 		int have_csum = 0;
 
 		if (flags & BTRFS_EXTENT_FLAG_DATA) {
@@ -2509,29 +2717,32 @@ static int scrub_extent(struct scrub_ctx *sctx, struct map_lookup *map,
 			if (have_csum == 0)
 				++sctx->stat.no_csum;
 		}
-		ret = scrub_pages(sctx, logical, l, physical, dev, flags, gen,
-				  mirror_num, have_csum ? csum : NULL, 0,
-				  physical_for_dev_replace);
+		ret = scrub_sectors(sctx, logical, l, src_physical, src_dev,
+				    flags, gen, src_mirror,
+				    have_csum ? csum : NULL, physical);
 		if (ret)
 			return ret;
 		len -= l;
 		logical += l;
 		physical += l;
-		physical_for_dev_replace += l;
+		src_physical += l;
 	}
 	return 0;
 }
 
-static int scrub_pages_for_parity(struct scrub_parity *sparity,
-				  u64 logical, u64 len,
+static int scrub_sectors_for_parity(struct scrub_parity *sparity,
+				  u64 logical, u32 len,
 				  u64 physical, struct btrfs_device *dev,
 				  u64 flags, u64 gen, int mirror_num, u8 *csum)
 {
 	struct scrub_ctx *sctx = sparity->sctx;
 	struct scrub_block *sblock;
+	const u32 sectorsize = sctx->fs_info->sectorsize;
 	int index;
 
-	sblock = kzalloc(sizeof(*sblock), GFP_KERNEL);
+	ASSERT(IS_ALIGNED(len, sectorsize));
+
+	sblock = alloc_scrub_block(sctx, dev, logical, physical, physical, mirror_num);
 	if (!sblock) {
 		spin_lock(&sctx->stat_lock);
 		sctx->stat.malloc_errors++;
@@ -2539,75 +2750,58 @@ static int scrub_pages_for_parity(struct scrub_parity *sparity,
 		return -ENOMEM;
 	}
 
-	/* one ref inside this function, plus one for each page added to
-	 * a bio later on */
-	refcount_set(&sblock->refs, 1);
-	sblock->sctx = sctx;
-	sblock->no_io_error_seen = 1;
 	sblock->sparity = sparity;
 	scrub_parity_get(sparity);
 
 	for (index = 0; len > 0; index++) {
-		struct scrub_page *spage;
-		u64 l = min_t(u64, len, PAGE_SIZE);
+		struct scrub_sector *sector;
 
-		spage = kzalloc(sizeof(*spage), GFP_KERNEL);
-		if (!spage) {
-leave_nomem:
+		sector = alloc_scrub_sector(sblock, logical, GFP_KERNEL);
+		if (!sector) {
 			spin_lock(&sctx->stat_lock);
 			sctx->stat.malloc_errors++;
 			spin_unlock(&sctx->stat_lock);
 			scrub_block_put(sblock);
 			return -ENOMEM;
 		}
-		BUG_ON(index >= SCRUB_MAX_PAGES_PER_BLOCK);
-		/* For scrub block */
-		scrub_page_get(spage);
-		sblock->pagev[index] = spage;
+		sblock->sectors[index] = sector;
 		/* For scrub parity */
-		scrub_page_get(spage);
-		list_add_tail(&spage->list, &sparity->spages);
-		spage->sblock = sblock;
-		spage->dev = dev;
-		spage->flags = flags;
-		spage->generation = gen;
-		spage->logical = logical;
-		spage->physical = physical;
-		spage->mirror_num = mirror_num;
+		scrub_sector_get(sector);
+		list_add_tail(&sector->list, &sparity->sectors_list);
+		sector->flags = flags;
+		sector->generation = gen;
 		if (csum) {
-			spage->have_csum = 1;
-			memcpy(spage->csum, csum, sctx->csum_size);
+			sector->have_csum = 1;
+			memcpy(sector->csum, csum, sctx->fs_info->csum_size);
 		} else {
-			spage->have_csum = 0;
+			sector->have_csum = 0;
 		}
-		sblock->page_count++;
-		spage->page = alloc_page(GFP_KERNEL);
-		if (!spage->page)
-			goto leave_nomem;
-		len -= l;
-		logical += l;
-		physical += l;
+
+		/* Iterate over the stripe range in sectorsize steps */
+		len -= sectorsize;
+		logical += sectorsize;
+		physical += sectorsize;
 	}
 
-	WARN_ON(sblock->page_count == 0);
-	for (index = 0; index < sblock->page_count; index++) {
-		struct scrub_page *spage = sblock->pagev[index];
+	WARN_ON(sblock->sector_count == 0);
+	for (index = 0; index < sblock->sector_count; index++) {
+		struct scrub_sector *sector = sblock->sectors[index];
 		int ret;
 
-		ret = scrub_add_page_to_rd_bio(sctx, spage);
+		ret = scrub_add_sector_to_rd_bio(sctx, sector);
 		if (ret) {
 			scrub_block_put(sblock);
 			return ret;
 		}
 	}
 
-	/* last one frees, either here or in bio completion for last page */
+	/* Last one frees, either here or in bio completion for last sector */
 	scrub_block_put(sblock);
 	return 0;
 }
 
 static int scrub_extent_for_parity(struct scrub_parity *sparity,
-				   u64 logical, u64 len,
+				   u64 logical, u32 len,
 				   u64 physical, struct btrfs_device *dev,
 				   u64 flags, u64 gen, int mirror_num)
 {
@@ -2631,7 +2825,7 @@ static int scrub_extent_for_parity(struct scrub_parity *sparity,
 	}
 
 	while (len) {
-		u64 l = min_t(u64, len, blocksize);
+		u32 l = min(len, blocksize);
 		int have_csum = 0;
 
 		if (flags & BTRFS_EXTENT_FLAG_DATA) {
@@ -2640,7 +2834,7 @@ static int scrub_extent_for_parity(struct scrub_parity *sparity,
 			if (have_csum == 0)
 				goto skip;
 		}
-		ret = scrub_pages_for_parity(sparity, logical, l, physical, dev,
+		ret = scrub_sectors_for_parity(sparity, logical, l, physical, dev,
 					     flags, gen, mirror_num,
 					     have_csum ? csum : NULL);
 		if (ret)
@@ -2700,10 +2894,10 @@ static int get_raid56_logic_offset(u64 physical, int num,
 static void scrub_free_parity(struct scrub_parity *sparity)
 {
 	struct scrub_ctx *sctx = sparity->sctx;
-	struct scrub_page *curr, *next;
+	struct scrub_sector *curr, *next;
 	int nbits;
 
-	nbits = bitmap_weight(sparity->ebitmap, sparity->nsectors);
+	nbits = bitmap_weight(&sparity->ebitmap, sparity->nsectors);
 	if (nbits) {
 		spin_lock(&sctx->stat_lock);
 		sctx->stat.read_errors += nbits;
@@ -2711,38 +2905,38 @@ static void scrub_free_parity(struct scrub_parity *sparity)
 		spin_unlock(&sctx->stat_lock);
 	}
 
-	list_for_each_entry_safe(curr, next, &sparity->spages, list) {
+	list_for_each_entry_safe(curr, next, &sparity->sectors_list, list) {
 		list_del_init(&curr->list);
-		scrub_page_put(curr);
+		scrub_sector_put(curr);
 	}
 
 	kfree(sparity);
 }
 
-static void scrub_parity_bio_endio_worker(struct btrfs_work *work)
+static void scrub_parity_bio_endio_worker(struct work_struct *work)
 {
 	struct scrub_parity *sparity = container_of(work, struct scrub_parity,
 						    work);
 	struct scrub_ctx *sctx = sparity->sctx;
 
+	btrfs_bio_counter_dec(sctx->fs_info);
 	scrub_free_parity(sparity);
 	scrub_pending_bio_dec(sctx);
 }
 
 static void scrub_parity_bio_endio(struct bio *bio)
 {
-	struct scrub_parity *sparity = (struct scrub_parity *)bio->bi_private;
+	struct scrub_parity *sparity = bio->bi_private;
 	struct btrfs_fs_info *fs_info = sparity->sctx->fs_info;
 
 	if (bio->bi_status)
-		bitmap_or(sparity->ebitmap, sparity->ebitmap, sparity->dbitmap,
-			  sparity->nsectors);
+		bitmap_or(&sparity->ebitmap, &sparity->ebitmap,
+			  &sparity->dbitmap, sparity->nsectors);
 
 	bio_put(bio);
 
-	btrfs_init_work(&sparity->work, scrub_parity_bio_endio_worker, NULL,
-			NULL);
-	btrfs_queue_work(fs_info->scrub_parity_workers, &sparity->work);
+	INIT_WORK(&sparity->work, scrub_parity_bio_endio_worker);
+	queue_work(fs_info->scrub_parity_workers, &sparity->work);
 }
 
 static void scrub_parity_check_and_repair(struct scrub_parity *sparity)
@@ -2751,31 +2945,32 @@ static void scrub_parity_check_and_repair(struct scrub_parity *sparity)
 	struct btrfs_fs_info *fs_info = sctx->fs_info;
 	struct bio *bio;
 	struct btrfs_raid_bio *rbio;
-	struct btrfs_bio *bbio = NULL;
+	struct btrfs_io_context *bioc = NULL;
 	u64 length;
 	int ret;
 
-	if (!bitmap_andnot(sparity->dbitmap, sparity->dbitmap, sparity->ebitmap,
-			   sparity->nsectors))
+	if (!bitmap_andnot(&sparity->dbitmap, &sparity->dbitmap,
+			   &sparity->ebitmap, sparity->nsectors))
 		goto out;
 
 	length = sparity->logic_end - sparity->logic_start;
 
 	btrfs_bio_counter_inc_blocked(fs_info);
 	ret = btrfs_map_sblock(fs_info, BTRFS_MAP_WRITE, sparity->logic_start,
-			       &length, &bbio);
-	if (ret || !bbio || !bbio->raid_map)
-		goto bbio_out;
+			       &length, &bioc);
+	if (ret || !bioc || !bioc->raid_map)
+		goto bioc_out;
 
-	bio = btrfs_io_bio_alloc(0);
+	bio = bio_alloc(NULL, BIO_MAX_VECS, REQ_OP_READ, GFP_NOFS);
 	bio->bi_iter.bi_sector = sparity->logic_start >> 9;
 	bio->bi_private = sparity;
 	bio->bi_end_io = scrub_parity_bio_endio;
 
-	rbio = raid56_parity_alloc_scrub_rbio(fs_info, bio, bbio,
-					      length, sparity->scrub_dev,
-					      sparity->dbitmap,
+	rbio = raid56_parity_alloc_scrub_rbio(bio, bioc,
+					      sparity->scrub_dev,
+					      &sparity->dbitmap,
 					      sparity->nsectors);
+	btrfs_put_bioc(bioc);
 	if (!rbio)
 		goto rbio_out;
 
@@ -2785,10 +2980,9 @@ static void scrub_parity_check_and_repair(struct scrub_parity *sparity)
 
 rbio_out:
 	bio_put(bio);
-bbio_out:
+bioc_out:
 	btrfs_bio_counter_dec(fs_info);
-	btrfs_put_bbio(bbio);
-	bitmap_or(sparity->ebitmap, sparity->ebitmap, sparity->dbitmap,
+	bitmap_or(&sparity->ebitmap, &sparity->ebitmap, &sparity->dbitmap,
 		  sparity->nsectors);
 	spin_lock(&sctx->stat_lock);
 	sctx->stat.malloc_errors++;
@@ -2797,11 +2991,6 @@ out:
 	scrub_free_parity(sparity);
 }
 
-static inline int scrub_calc_parity_bitmap_len(int nsectors)
-{
-	return DIV_ROUND_UP(nsectors, BITS_PER_LONG) * sizeof(long);
-}
-
 static void scrub_parity_get(struct scrub_parity *sparity)
 {
 	refcount_inc(&sparity->refs);
@@ -2815,46 +3004,287 @@ static void scrub_parity_put(struct scrub_parity *sparity)
 	scrub_parity_check_and_repair(sparity);
 }
 
+/*
+ * Return 0 if the extent item range covers any byte of the range.
+ * Return <0 if the extent item is before @search_start.
+ * Return >0 if the extent item is after @start_start + @search_len.
+ */
+static int compare_extent_item_range(struct btrfs_path *path,
+				     u64 search_start, u64 search_len)
+{
+	struct btrfs_fs_info *fs_info = path->nodes[0]->fs_info;
+	u64 len;
+	struct btrfs_key key;
+
+	btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
+	ASSERT(key.type == BTRFS_EXTENT_ITEM_KEY ||
+	       key.type == BTRFS_METADATA_ITEM_KEY);
+	if (key.type == BTRFS_METADATA_ITEM_KEY)
+		len = fs_info->nodesize;
+	else
+		len = key.offset;
+
+	if (key.objectid + len <= search_start)
+		return -1;
+	if (key.objectid >= search_start + search_len)
+		return 1;
+	return 0;
+}
+
+/*
+ * Locate one extent item which covers any byte in range
+ * [@search_start, @search_start + @search_length)
+ *
+ * If the path is not initialized, we will initialize the search by doing
+ * a btrfs_search_slot().
+ * If the path is already initialized, we will use the path as the initial
+ * slot, to avoid duplicated btrfs_search_slot() calls.
+ *
+ * NOTE: If an extent item starts before @search_start, we will still
+ * return the extent item. This is for data extent crossing stripe boundary.
+ *
+ * Return 0 if we found such extent item, and @path will point to the extent item.
+ * Return >0 if no such extent item can be found, and @path will be released.
+ * Return <0 if hit fatal error, and @path will be released.
+ */
+static int find_first_extent_item(struct btrfs_root *extent_root,
+				  struct btrfs_path *path,
+				  u64 search_start, u64 search_len)
+{
+	struct btrfs_fs_info *fs_info = extent_root->fs_info;
+	struct btrfs_key key;
+	int ret;
+
+	/* Continue using the existing path */
+	if (path->nodes[0])
+		goto search_forward;
+
+	if (btrfs_fs_incompat(fs_info, SKINNY_METADATA))
+		key.type = BTRFS_METADATA_ITEM_KEY;
+	else
+		key.type = BTRFS_EXTENT_ITEM_KEY;
+	key.objectid = search_start;
+	key.offset = (u64)-1;
+
+	ret = btrfs_search_slot(NULL, extent_root, &key, path, 0, 0);
+	if (ret < 0)
+		return ret;
+
+	ASSERT(ret > 0);
+	/*
+	 * Here we intentionally pass 0 as @min_objectid, as there could be
+	 * an extent item starting before @search_start.
+	 */
+	ret = btrfs_previous_extent_item(extent_root, path, 0);
+	if (ret < 0)
+		return ret;
+	/*
+	 * No matter whether we have found an extent item, the next loop will
+	 * properly do every check on the key.
+	 */
+search_forward:
+	while (true) {
+		btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
+		if (key.objectid >= search_start + search_len)
+			break;
+		if (key.type != BTRFS_METADATA_ITEM_KEY &&
+		    key.type != BTRFS_EXTENT_ITEM_KEY)
+			goto next;
+
+		ret = compare_extent_item_range(path, search_start, search_len);
+		if (ret == 0)
+			return ret;
+		if (ret > 0)
+			break;
+next:
+		path->slots[0]++;
+		if (path->slots[0] >= btrfs_header_nritems(path->nodes[0])) {
+			ret = btrfs_next_leaf(extent_root, path);
+			if (ret) {
+				/* Either no more item or fatal error */
+				btrfs_release_path(path);
+				return ret;
+			}
+		}
+	}
+	btrfs_release_path(path);
+	return 1;
+}
+
+static void get_extent_info(struct btrfs_path *path, u64 *extent_start_ret,
+			    u64 *size_ret, u64 *flags_ret, u64 *generation_ret)
+{
+	struct btrfs_key key;
+	struct btrfs_extent_item *ei;
+
+	btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
+	ASSERT(key.type == BTRFS_METADATA_ITEM_KEY ||
+	       key.type == BTRFS_EXTENT_ITEM_KEY);
+	*extent_start_ret = key.objectid;
+	if (key.type == BTRFS_METADATA_ITEM_KEY)
+		*size_ret = path->nodes[0]->fs_info->nodesize;
+	else
+		*size_ret = key.offset;
+	ei = btrfs_item_ptr(path->nodes[0], path->slots[0], struct btrfs_extent_item);
+	*flags_ret = btrfs_extent_flags(path->nodes[0], ei);
+	*generation_ret = btrfs_extent_generation(path->nodes[0], ei);
+}
+
+static bool does_range_cross_boundary(u64 extent_start, u64 extent_len,
+				      u64 boundary_start, u64 boudary_len)
+{
+	return (extent_start < boundary_start &&
+		extent_start + extent_len > boundary_start) ||
+	       (extent_start < boundary_start + boudary_len &&
+		extent_start + extent_len > boundary_start + boudary_len);
+}
+
+static int scrub_raid56_data_stripe_for_parity(struct scrub_ctx *sctx,
+					       struct scrub_parity *sparity,
+					       struct map_lookup *map,
+					       struct btrfs_device *sdev,
+					       struct btrfs_path *path,
+					       u64 logical)
+{
+	struct btrfs_fs_info *fs_info = sctx->fs_info;
+	struct btrfs_root *extent_root = btrfs_extent_root(fs_info, logical);
+	struct btrfs_root *csum_root = btrfs_csum_root(fs_info, logical);
+	u64 cur_logical = logical;
+	int ret;
+
+	ASSERT(map->type & BTRFS_BLOCK_GROUP_RAID56_MASK);
+
+	/* Path must not be populated */
+	ASSERT(!path->nodes[0]);
+
+	while (cur_logical < logical + map->stripe_len) {
+		struct btrfs_io_context *bioc = NULL;
+		struct btrfs_device *extent_dev;
+		u64 extent_start;
+		u64 extent_size;
+		u64 mapped_length;
+		u64 extent_flags;
+		u64 extent_gen;
+		u64 extent_physical;
+		u64 extent_mirror_num;
+
+		ret = find_first_extent_item(extent_root, path, cur_logical,
+					     logical + map->stripe_len - cur_logical);
+		/* No more extent item in this data stripe */
+		if (ret > 0) {
+			ret = 0;
+			break;
+		}
+		if (ret < 0)
+			break;
+		get_extent_info(path, &extent_start, &extent_size, &extent_flags,
+				&extent_gen);
+
+		/* Metadata should not cross stripe boundaries */
+		if ((extent_flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) &&
+		    does_range_cross_boundary(extent_start, extent_size,
+					      logical, map->stripe_len)) {
+			btrfs_err(fs_info,
+	"scrub: tree block %llu spanning stripes, ignored. logical=%llu",
+				  extent_start, logical);
+			spin_lock(&sctx->stat_lock);
+			sctx->stat.uncorrectable_errors++;
+			spin_unlock(&sctx->stat_lock);
+			cur_logical += extent_size;
+			continue;
+		}
+
+		/* Skip hole range which doesn't have any extent */
+		cur_logical = max(extent_start, cur_logical);
+
+		/* Truncate the range inside this data stripe */
+		extent_size = min(extent_start + extent_size,
+				  logical + map->stripe_len) - cur_logical;
+		extent_start = cur_logical;
+		ASSERT(extent_size <= U32_MAX);
+
+		scrub_parity_mark_sectors_data(sparity, extent_start, extent_size);
+
+		mapped_length = extent_size;
+		ret = btrfs_map_block(fs_info, BTRFS_MAP_READ, extent_start,
+				      &mapped_length, &bioc, 0);
+		if (!ret && (!bioc || mapped_length < extent_size))
+			ret = -EIO;
+		if (ret) {
+			btrfs_put_bioc(bioc);
+			scrub_parity_mark_sectors_error(sparity, extent_start,
+							extent_size);
+			break;
+		}
+		extent_physical = bioc->stripes[0].physical;
+		extent_mirror_num = bioc->mirror_num;
+		extent_dev = bioc->stripes[0].dev;
+		btrfs_put_bioc(bioc);
+
+		ret = btrfs_lookup_csums_range(csum_root, extent_start,
+					       extent_start + extent_size - 1,
+					       &sctx->csum_list, 1, false);
+		if (ret) {
+			scrub_parity_mark_sectors_error(sparity, extent_start,
+							extent_size);
+			break;
+		}
+
+		ret = scrub_extent_for_parity(sparity, extent_start,
+					      extent_size, extent_physical,
+					      extent_dev, extent_flags,
+					      extent_gen, extent_mirror_num);
+		scrub_free_csums(sctx);
+
+		if (ret) {
+			scrub_parity_mark_sectors_error(sparity, extent_start,
+							extent_size);
+			break;
+		}
+
+		cond_resched();
+		cur_logical += extent_size;
+	}
+	btrfs_release_path(path);
+	return ret;
+}
+
 static noinline_for_stack int scrub_raid56_parity(struct scrub_ctx *sctx,
 						  struct map_lookup *map,
 						  struct btrfs_device *sdev,
-						  struct btrfs_path *path,
 						  u64 logic_start,
 						  u64 logic_end)
 {
 	struct btrfs_fs_info *fs_info = sctx->fs_info;
-	struct btrfs_root *root = fs_info->extent_root;
-	struct btrfs_root *csum_root = fs_info->csum_root;
-	struct btrfs_extent_item *extent;
-	struct btrfs_bio *bbio = NULL;
-	u64 flags;
+	struct btrfs_path *path;
+	u64 cur_logical;
 	int ret;
-	int slot;
-	struct extent_buffer *l;
-	struct btrfs_key key;
-	u64 generation;
-	u64 extent_logical;
-	u64 extent_physical;
-	u64 extent_len;
-	u64 mapped_length;
-	struct btrfs_device *extent_dev;
 	struct scrub_parity *sparity;
 	int nsectors;
-	int bitmap_len;
-	int extent_mirror_num;
-	int stop_loop = 0;
 
-	nsectors = div_u64(map->stripe_len, fs_info->sectorsize);
-	bitmap_len = scrub_calc_parity_bitmap_len(nsectors);
-	sparity = kzalloc(sizeof(struct scrub_parity) + 2 * bitmap_len,
-			  GFP_NOFS);
+	path = btrfs_alloc_path();
+	if (!path) {
+		spin_lock(&sctx->stat_lock);
+		sctx->stat.malloc_errors++;
+		spin_unlock(&sctx->stat_lock);
+		return -ENOMEM;
+	}
+	path->search_commit_root = 1;
+	path->skip_locking = 1;
+
+	ASSERT(map->stripe_len <= U32_MAX);
+	nsectors = map->stripe_len >> fs_info->sectorsize_bits;
+	ASSERT(nsectors <= BITS_PER_LONG);
+	sparity = kzalloc(sizeof(struct scrub_parity), GFP_NOFS);
 	if (!sparity) {
 		spin_lock(&sctx->stat_lock);
 		sctx->stat.malloc_errors++;
 		spin_unlock(&sctx->stat_lock);
+		btrfs_free_path(path);
 		return -ENOMEM;
 	}
 
+	ASSERT(map->stripe_len <= U32_MAX);
 	sparity->stripe_len = map->stripe_len;
 	sparity->nsectors = nsectors;
 	sparity->sctx = sctx;
@@ -2862,258 +3292,299 @@ static noinline_for_stack int scrub_raid56_parity(struct scrub_ctx *sctx,
 	sparity->logic_start = logic_start;
 	sparity->logic_end = logic_end;
 	refcount_set(&sparity->refs, 1);
-	INIT_LIST_HEAD(&sparity->spages);
-	sparity->dbitmap = sparity->bitmap;
-	sparity->ebitmap = (void *)sparity->bitmap + bitmap_len;
+	INIT_LIST_HEAD(&sparity->sectors_list);
 
 	ret = 0;
-	while (logic_start < logic_end) {
-		if (btrfs_fs_incompat(fs_info, SKINNY_METADATA))
-			key.type = BTRFS_METADATA_ITEM_KEY;
-		else
-			key.type = BTRFS_EXTENT_ITEM_KEY;
-		key.objectid = logic_start;
-		key.offset = (u64)-1;
-
-		ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
+	for (cur_logical = logic_start; cur_logical < logic_end;
+	     cur_logical += map->stripe_len) {
+		ret = scrub_raid56_data_stripe_for_parity(sctx, sparity, map,
+							  sdev, path, cur_logical);
 		if (ret < 0)
-			goto out;
+			break;
+	}
 
-		if (ret > 0) {
-			ret = btrfs_previous_extent_item(root, path, 0);
-			if (ret < 0)
-				goto out;
-			if (ret > 0) {
-				btrfs_release_path(path);
-				ret = btrfs_search_slot(NULL, root, &key,
-							path, 0, 0);
-				if (ret < 0)
-					goto out;
-			}
-		}
+	scrub_parity_put(sparity);
+	scrub_submit(sctx);
+	mutex_lock(&sctx->wr_lock);
+	scrub_wr_submit(sctx);
+	mutex_unlock(&sctx->wr_lock);
 
-		stop_loop = 0;
-		while (1) {
-			u64 bytes;
+	btrfs_free_path(path);
+	return ret < 0 ? ret : 0;
+}
 
-			l = path->nodes[0];
-			slot = path->slots[0];
-			if (slot >= btrfs_header_nritems(l)) {
-				ret = btrfs_next_leaf(root, path);
-				if (ret == 0)
-					continue;
-				if (ret < 0)
-					goto out;
+static void sync_replace_for_zoned(struct scrub_ctx *sctx)
+{
+	if (!btrfs_is_zoned(sctx->fs_info))
+		return;
 
-				stop_loop = 1;
-				break;
-			}
-			btrfs_item_key_to_cpu(l, &key, slot);
+	sctx->flush_all_writes = true;
+	scrub_submit(sctx);
+	mutex_lock(&sctx->wr_lock);
+	scrub_wr_submit(sctx);
+	mutex_unlock(&sctx->wr_lock);
+
+	wait_event(sctx->list_wait, atomic_read(&sctx->bios_in_flight) == 0);
+}
 
-			if (key.type != BTRFS_EXTENT_ITEM_KEY &&
-			    key.type != BTRFS_METADATA_ITEM_KEY)
-				goto next;
+static int sync_write_pointer_for_zoned(struct scrub_ctx *sctx, u64 logical,
+					u64 physical, u64 physical_end)
+{
+	struct btrfs_fs_info *fs_info = sctx->fs_info;
+	int ret = 0;
 
-			if (key.type == BTRFS_METADATA_ITEM_KEY)
-				bytes = fs_info->nodesize;
-			else
-				bytes = key.offset;
+	if (!btrfs_is_zoned(fs_info))
+		return 0;
 
-			if (key.objectid + bytes <= logic_start)
-				goto next;
+	wait_event(sctx->list_wait, atomic_read(&sctx->bios_in_flight) == 0);
 
-			if (key.objectid >= logic_end) {
-				stop_loop = 1;
-				break;
-			}
+	mutex_lock(&sctx->wr_lock);
+	if (sctx->write_pointer < physical_end) {
+		ret = btrfs_sync_zone_write_pointer(sctx->wr_tgtdev, logical,
+						    physical,
+						    sctx->write_pointer);
+		if (ret)
+			btrfs_err(fs_info,
+				  "zoned: failed to recover write pointer");
+	}
+	mutex_unlock(&sctx->wr_lock);
+	btrfs_dev_clear_zone_empty(sctx->wr_tgtdev, physical);
 
-			while (key.objectid >= logic_start + map->stripe_len)
-				logic_start += map->stripe_len;
-
-			extent = btrfs_item_ptr(l, slot,
-						struct btrfs_extent_item);
-			flags = btrfs_extent_flags(l, extent);
-			generation = btrfs_extent_generation(l, extent);
-
-			if ((flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) &&
-			    (key.objectid < logic_start ||
-			     key.objectid + bytes >
-			     logic_start + map->stripe_len)) {
-				btrfs_err(fs_info,
-					  "scrub: tree block %llu spanning stripes, ignored. logical=%llu",
-					  key.objectid, logic_start);
-				spin_lock(&sctx->stat_lock);
-				sctx->stat.uncorrectable_errors++;
-				spin_unlock(&sctx->stat_lock);
-				goto next;
-			}
-again:
-			extent_logical = key.objectid;
-			extent_len = bytes;
+	return ret;
+}
 
-			if (extent_logical < logic_start) {
-				extent_len -= logic_start - extent_logical;
-				extent_logical = logic_start;
-			}
+/*
+ * Scrub one range which can only has simple mirror based profile.
+ * (Including all range in SINGLE/DUP/RAID1/RAID1C*, and each stripe in
+ *  RAID0/RAID10).
+ *
+ * Since we may need to handle a subset of block group, we need @logical_start
+ * and @logical_length parameter.
+ */
+static int scrub_simple_mirror(struct scrub_ctx *sctx,
+			       struct btrfs_root *extent_root,
+			       struct btrfs_root *csum_root,
+			       struct btrfs_block_group *bg,
+			       struct map_lookup *map,
+			       u64 logical_start, u64 logical_length,
+			       struct btrfs_device *device,
+			       u64 physical, int mirror_num)
+{
+	struct btrfs_fs_info *fs_info = sctx->fs_info;
+	const u64 logical_end = logical_start + logical_length;
+	/* An artificial limit, inherit from old scrub behavior */
+	const u32 max_length = SZ_64K;
+	struct btrfs_path path = { 0 };
+	u64 cur_logical = logical_start;
+	int ret;
 
-			if (extent_logical + extent_len >
-			    logic_start + map->stripe_len)
-				extent_len = logic_start + map->stripe_len -
-					     extent_logical;
-
-			scrub_parity_mark_sectors_data(sparity, extent_logical,
-						       extent_len);
-
-			mapped_length = extent_len;
-			bbio = NULL;
-			ret = btrfs_map_block(fs_info, BTRFS_MAP_READ,
-					extent_logical, &mapped_length, &bbio,
-					0);
-			if (!ret) {
-				if (!bbio || mapped_length < extent_len)
-					ret = -EIO;
-			}
-			if (ret) {
-				btrfs_put_bbio(bbio);
-				goto out;
-			}
-			extent_physical = bbio->stripes[0].physical;
-			extent_mirror_num = bbio->mirror_num;
-			extent_dev = bbio->stripes[0].dev;
-			btrfs_put_bbio(bbio);
-
-			ret = btrfs_lookup_csums_range(csum_root,
-						extent_logical,
-						extent_logical + extent_len - 1,
-						&sctx->csum_list, 1);
-			if (ret)
-				goto out;
+	/* The range must be inside the bg */
+	ASSERT(logical_start >= bg->start && logical_end <= bg->start + bg->length);
+
+	path.search_commit_root = 1;
+	path.skip_locking = 1;
+	/* Go through each extent items inside the logical range */
+	while (cur_logical < logical_end) {
+		u64 extent_start;
+		u64 extent_len;
+		u64 extent_flags;
+		u64 extent_gen;
+		u64 scrub_len;
 
-			ret = scrub_extent_for_parity(sparity, extent_logical,
-						      extent_len,
-						      extent_physical,
-						      extent_dev, flags,
-						      generation,
-						      extent_mirror_num);
+		/* Canceled? */
+		if (atomic_read(&fs_info->scrub_cancel_req) ||
+		    atomic_read(&sctx->cancel_req)) {
+			ret = -ECANCELED;
+			break;
+		}
+		/* Paused? */
+		if (atomic_read(&fs_info->scrub_pause_req)) {
+			/* Push queued extents */
+			sctx->flush_all_writes = true;
+			scrub_submit(sctx);
+			mutex_lock(&sctx->wr_lock);
+			scrub_wr_submit(sctx);
+			mutex_unlock(&sctx->wr_lock);
+			wait_event(sctx->list_wait,
+				   atomic_read(&sctx->bios_in_flight) == 0);
+			sctx->flush_all_writes = false;
+			scrub_blocked_if_needed(fs_info);
+		}
+		/* Block group removed? */
+		spin_lock(&bg->lock);
+		if (test_bit(BLOCK_GROUP_FLAG_REMOVED, &bg->runtime_flags)) {
+			spin_unlock(&bg->lock);
+			ret = 0;
+			break;
+		}
+		spin_unlock(&bg->lock);
 
-			scrub_free_csums(sctx);
+		ret = find_first_extent_item(extent_root, &path, cur_logical,
+					     logical_end - cur_logical);
+		if (ret > 0) {
+			/* No more extent, just update the accounting */
+			sctx->stat.last_physical = physical + logical_length;
+			ret = 0;
+			break;
+		}
+		if (ret < 0)
+			break;
+		get_extent_info(&path, &extent_start, &extent_len,
+				&extent_flags, &extent_gen);
+		/* Skip hole range which doesn't have any extent */
+		cur_logical = max(extent_start, cur_logical);
 
+		/*
+		 * Scrub len has three limits:
+		 * - Extent size limit
+		 * - Scrub range limit
+		 *   This is especially imporatant for RAID0/RAID10 to reuse
+		 *   this function
+		 * - Max scrub size limit
+		 */
+		scrub_len = min(min(extent_start + extent_len,
+				    logical_end), cur_logical + max_length) -
+			    cur_logical;
+
+		if (extent_flags & BTRFS_EXTENT_FLAG_DATA) {
+			ret = btrfs_lookup_csums_range(csum_root, cur_logical,
+					cur_logical + scrub_len - 1,
+					&sctx->csum_list, 1, false);
 			if (ret)
-				goto out;
+				break;
+		}
+		if ((extent_flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) &&
+		    does_range_cross_boundary(extent_start, extent_len,
+					      logical_start, logical_length)) {
+			btrfs_err(fs_info,
+"scrub: tree block %llu spanning boundaries, ignored. boundary=[%llu, %llu)",
+				  extent_start, logical_start, logical_end);
+			spin_lock(&sctx->stat_lock);
+			sctx->stat.uncorrectable_errors++;
+			spin_unlock(&sctx->stat_lock);
+			cur_logical += scrub_len;
+			continue;
+		}
+		ret = scrub_extent(sctx, map, cur_logical, scrub_len,
+				   cur_logical - logical_start + physical,
+				   device, extent_flags, extent_gen,
+				   mirror_num);
+		scrub_free_csums(sctx);
+		if (ret)
+			break;
+		if (sctx->is_dev_replace)
+			sync_replace_for_zoned(sctx);
+		cur_logical += scrub_len;
+		/* Don't hold CPU for too long time */
+		cond_resched();
+	}
+	btrfs_release_path(&path);
+	return ret;
+}
 
-			if (extent_logical + extent_len <
-			    key.objectid + bytes) {
-				logic_start += map->stripe_len;
+/* Calculate the full stripe length for simple stripe based profiles */
+static u64 simple_stripe_full_stripe_len(const struct map_lookup *map)
+{
+	ASSERT(map->type & (BTRFS_BLOCK_GROUP_RAID0 |
+			    BTRFS_BLOCK_GROUP_RAID10));
 
-				if (logic_start >= logic_end) {
-					stop_loop = 1;
-					break;
-				}
+	return map->num_stripes / map->sub_stripes * map->stripe_len;
+}
 
-				if (logic_start < key.objectid + bytes) {
-					cond_resched();
-					goto again;
-				}
-			}
-next:
-			path->slots[0]++;
-		}
+/* Get the logical bytenr for the stripe */
+static u64 simple_stripe_get_logical(struct map_lookup *map,
+				     struct btrfs_block_group *bg,
+				     int stripe_index)
+{
+	ASSERT(map->type & (BTRFS_BLOCK_GROUP_RAID0 |
+			    BTRFS_BLOCK_GROUP_RAID10));
+	ASSERT(stripe_index < map->num_stripes);
 
-		btrfs_release_path(path);
+	/*
+	 * (stripe_index / sub_stripes) gives how many data stripes we need to
+	 * skip.
+	 */
+	return (stripe_index / map->sub_stripes) * map->stripe_len + bg->start;
+}
 
-		if (stop_loop)
-			break;
+/* Get the mirror number for the stripe */
+static int simple_stripe_mirror_num(struct map_lookup *map, int stripe_index)
+{
+	ASSERT(map->type & (BTRFS_BLOCK_GROUP_RAID0 |
+			    BTRFS_BLOCK_GROUP_RAID10));
+	ASSERT(stripe_index < map->num_stripes);
 
-		logic_start += map->stripe_len;
-	}
-out:
-	if (ret < 0)
-		scrub_parity_mark_sectors_error(sparity, logic_start,
-						logic_end - logic_start);
-	scrub_parity_put(sparity);
-	scrub_submit(sctx);
-	mutex_lock(&sctx->wr_lock);
-	scrub_wr_submit(sctx);
-	mutex_unlock(&sctx->wr_lock);
+	/* For RAID0, it's fixed to 1, for RAID10 it's 0,1,0,1... */
+	return stripe_index % map->sub_stripes + 1;
+}
 
-	btrfs_release_path(path);
-	return ret < 0 ? ret : 0;
+static int scrub_simple_stripe(struct scrub_ctx *sctx,
+			       struct btrfs_root *extent_root,
+			       struct btrfs_root *csum_root,
+			       struct btrfs_block_group *bg,
+			       struct map_lookup *map,
+			       struct btrfs_device *device,
+			       int stripe_index)
+{
+	const u64 logical_increment = simple_stripe_full_stripe_len(map);
+	const u64 orig_logical = simple_stripe_get_logical(map, bg, stripe_index);
+	const u64 orig_physical = map->stripes[stripe_index].physical;
+	const int mirror_num = simple_stripe_mirror_num(map, stripe_index);
+	u64 cur_logical = orig_logical;
+	u64 cur_physical = orig_physical;
+	int ret = 0;
+
+	while (cur_logical < bg->start + bg->length) {
+		/*
+		 * Inside each stripe, RAID0 is just SINGLE, and RAID10 is
+		 * just RAID1, so we can reuse scrub_simple_mirror() to scrub
+		 * this stripe.
+		 */
+		ret = scrub_simple_mirror(sctx, extent_root, csum_root, bg, map,
+					  cur_logical, map->stripe_len, device,
+					  cur_physical, mirror_num);
+		if (ret)
+			return ret;
+		/* Skip to next stripe which belongs to the target device */
+		cur_logical += logical_increment;
+		/* For physical offset, we just go to next stripe */
+		cur_physical += map->stripe_len;
+	}
+	return ret;
 }
 
 static noinline_for_stack int scrub_stripe(struct scrub_ctx *sctx,
-					   struct map_lookup *map,
+					   struct btrfs_block_group *bg,
+					   struct extent_map *em,
 					   struct btrfs_device *scrub_dev,
-					   int num, u64 base, u64 length)
+					   int stripe_index)
 {
-	struct btrfs_path *path, *ppath;
+	struct btrfs_path *path;
 	struct btrfs_fs_info *fs_info = sctx->fs_info;
-	struct btrfs_root *root = fs_info->extent_root;
-	struct btrfs_root *csum_root = fs_info->csum_root;
-	struct btrfs_extent_item *extent;
+	struct btrfs_root *root;
+	struct btrfs_root *csum_root;
 	struct blk_plug plug;
-	u64 flags;
+	struct map_lookup *map = em->map_lookup;
+	const u64 profile = map->type & BTRFS_BLOCK_GROUP_PROFILE_MASK;
+	const u64 chunk_logical = bg->start;
 	int ret;
-	int slot;
-	u64 nstripes;
-	struct extent_buffer *l;
-	u64 physical;
+	u64 physical = map->stripes[stripe_index].physical;
+	const u64 dev_stripe_len = btrfs_calc_stripe_length(em);
+	const u64 physical_end = physical + dev_stripe_len;
 	u64 logical;
 	u64 logic_end;
-	u64 physical_end;
-	u64 generation;
-	int mirror_num;
-	struct reada_control *reada1;
-	struct reada_control *reada2;
-	struct btrfs_key key;
-	struct btrfs_key key_end;
-	u64 increment = map->stripe_len;
+	/* The logical increment after finishing one stripe */
+	u64 increment;
+	/* Offset inside the chunk */
 	u64 offset;
-	u64 extent_logical;
-	u64 extent_physical;
-	u64 extent_len;
 	u64 stripe_logical;
 	u64 stripe_end;
-	struct btrfs_device *extent_dev;
-	int extent_mirror_num;
 	int stop_loop = 0;
 
-	physical = map->stripes[num].physical;
-	offset = 0;
-	nstripes = div64_u64(length, map->stripe_len);
-	if (map->type & BTRFS_BLOCK_GROUP_RAID0) {
-		offset = map->stripe_len * num;
-		increment = map->stripe_len * map->num_stripes;
-		mirror_num = 1;
-	} else if (map->type & BTRFS_BLOCK_GROUP_RAID10) {
-		int factor = map->num_stripes / map->sub_stripes;
-		offset = map->stripe_len * (num / map->sub_stripes);
-		increment = map->stripe_len * factor;
-		mirror_num = num % map->sub_stripes + 1;
-	} else if (map->type & BTRFS_BLOCK_GROUP_RAID1_MASK) {
-		increment = map->stripe_len;
-		mirror_num = num % map->num_stripes + 1;
-	} else if (map->type & BTRFS_BLOCK_GROUP_DUP) {
-		increment = map->stripe_len;
-		mirror_num = num % map->num_stripes + 1;
-	} else if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) {
-		get_raid56_logic_offset(physical, num, map, &offset, NULL);
-		increment = map->stripe_len * nr_data_stripes(map);
-		mirror_num = 1;
-	} else {
-		increment = map->stripe_len;
-		mirror_num = 1;
-	}
-
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
 
-	ppath = btrfs_alloc_path();
-	if (!ppath) {
-		btrfs_free_path(path);
-		return -ENOMEM;
-	}
-
 	/*
 	 * work on commit root. The related disk blocks are static as
 	 * long as COW is applied. This means, it is save to rewrite
@@ -3121,49 +3592,14 @@ static noinline_for_stack int scrub_stripe(struct scrub_ctx *sctx,
 	 */
 	path->search_commit_root = 1;
 	path->skip_locking = 1;
+	path->reada = READA_FORWARD;
 
-	ppath->search_commit_root = 1;
-	ppath->skip_locking = 1;
-	/*
-	 * trigger the readahead for extent tree csum tree and wait for
-	 * completion. During readahead, the scrub is officially paused
-	 * to not hold off transaction commits
-	 */
-	logical = base + offset;
-	physical_end = physical + nstripes * map->stripe_len;
-	if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) {
-		get_raid56_logic_offset(physical_end, num,
-					map, &logic_end, NULL);
-		logic_end += base;
-	} else {
-		logic_end = logical + increment * nstripes;
-	}
 	wait_event(sctx->list_wait,
 		   atomic_read(&sctx->bios_in_flight) == 0);
 	scrub_blocked_if_needed(fs_info);
 
-	/* FIXME it might be better to start readahead at commit root */
-	key.objectid = logical;
-	key.type = BTRFS_EXTENT_ITEM_KEY;
-	key.offset = (u64)0;
-	key_end.objectid = logic_end;
-	key_end.type = BTRFS_METADATA_ITEM_KEY;
-	key_end.offset = (u64)-1;
-	reada1 = btrfs_reada_add(root, &key, &key_end);
-
-	key.objectid = BTRFS_EXTENT_CSUM_OBJECTID;
-	key.type = BTRFS_EXTENT_CSUM_KEY;
-	key.offset = logical;
-	key_end.objectid = BTRFS_EXTENT_CSUM_OBJECTID;
-	key_end.type = BTRFS_EXTENT_CSUM_KEY;
-	key_end.offset = logic_end;
-	reada2 = btrfs_reada_add(csum_root, &key, &key_end);
-
-	if (!IS_ERR(reada1))
-		btrfs_reada_wait(reada1);
-	if (!IS_ERR(reada2))
-		btrfs_reada_wait(reada2);
-
+	root = btrfs_extent_root(fs_info, bg->start);
+	csum_root = btrfs_csum_root(fs_info, bg->start);
 
 	/*
 	 * collect all data csums for the stripe to avoid seeking during
@@ -3171,229 +3607,98 @@ static noinline_for_stack int scrub_stripe(struct scrub_ctx *sctx,
 	 */
 	blk_start_plug(&plug);
 
+	if (sctx->is_dev_replace &&
+	    btrfs_dev_is_sequential(sctx->wr_tgtdev, physical)) {
+		mutex_lock(&sctx->wr_lock);
+		sctx->write_pointer = physical;
+		mutex_unlock(&sctx->wr_lock);
+		sctx->flush_all_writes = true;
+	}
+
 	/*
-	 * now find all extents for each stripe and scrub them
+	 * There used to be a big double loop to handle all profiles using the
+	 * same routine, which grows larger and more gross over time.
+	 *
+	 * So here we handle each profile differently, so simpler profiles
+	 * have simpler scrubbing function.
 	 */
-	ret = 0;
-	while (physical < physical_end) {
-		/*
-		 * canceled?
-		 */
-		if (atomic_read(&fs_info->scrub_cancel_req) ||
-		    atomic_read(&sctx->cancel_req)) {
-			ret = -ECANCELED;
-			goto out;
-		}
+	if (!(profile & (BTRFS_BLOCK_GROUP_RAID0 | BTRFS_BLOCK_GROUP_RAID10 |
+			 BTRFS_BLOCK_GROUP_RAID56_MASK))) {
 		/*
-		 * check to see if we have to pause
+		 * Above check rules out all complex profile, the remaining
+		 * profiles are SINGLE|DUP|RAID1|RAID1C*, which is simple
+		 * mirrored duplication without stripe.
+		 *
+		 * Only @physical and @mirror_num needs to calculated using
+		 * @stripe_index.
 		 */
-		if (atomic_read(&fs_info->scrub_pause_req)) {
-			/* push queued extents */
-			sctx->flush_all_writes = true;
-			scrub_submit(sctx);
-			mutex_lock(&sctx->wr_lock);
-			scrub_wr_submit(sctx);
-			mutex_unlock(&sctx->wr_lock);
-			wait_event(sctx->list_wait,
-				   atomic_read(&sctx->bios_in_flight) == 0);
-			sctx->flush_all_writes = false;
-			scrub_blocked_if_needed(fs_info);
-		}
-
-		if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) {
-			ret = get_raid56_logic_offset(physical, num, map,
-						      &logical,
-						      &stripe_logical);
-			logical += base;
-			if (ret) {
-				/* it is parity strip */
-				stripe_logical += base;
-				stripe_end = stripe_logical + increment;
-				ret = scrub_raid56_parity(sctx, map, scrub_dev,
-							  ppath, stripe_logical,
-							  stripe_end);
-				if (ret)
-					goto out;
-				goto skip;
-			}
-		}
-
-		if (btrfs_fs_incompat(fs_info, SKINNY_METADATA))
-			key.type = BTRFS_METADATA_ITEM_KEY;
-		else
-			key.type = BTRFS_EXTENT_ITEM_KEY;
-		key.objectid = logical;
-		key.offset = (u64)-1;
-
-		ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
-		if (ret < 0)
-			goto out;
-
-		if (ret > 0) {
-			ret = btrfs_previous_extent_item(root, path, 0);
-			if (ret < 0)
-				goto out;
-			if (ret > 0) {
-				/* there's no smaller item, so stick with the
-				 * larger one */
-				btrfs_release_path(path);
-				ret = btrfs_search_slot(NULL, root, &key,
-							path, 0, 0);
-				if (ret < 0)
-					goto out;
-			}
-		}
-
-		stop_loop = 0;
-		while (1) {
-			u64 bytes;
-
-			l = path->nodes[0];
-			slot = path->slots[0];
-			if (slot >= btrfs_header_nritems(l)) {
-				ret = btrfs_next_leaf(root, path);
-				if (ret == 0)
-					continue;
-				if (ret < 0)
-					goto out;
-
-				stop_loop = 1;
-				break;
-			}
-			btrfs_item_key_to_cpu(l, &key, slot);
-
-			if (key.type != BTRFS_EXTENT_ITEM_KEY &&
-			    key.type != BTRFS_METADATA_ITEM_KEY)
-				goto next;
-
-			if (key.type == BTRFS_METADATA_ITEM_KEY)
-				bytes = fs_info->nodesize;
-			else
-				bytes = key.offset;
-
-			if (key.objectid + bytes <= logical)
-				goto next;
-
-			if (key.objectid >= logical + map->stripe_len) {
-				/* out of this device extent */
-				if (key.objectid >= logic_end)
-					stop_loop = 1;
-				break;
-			}
-
-			extent = btrfs_item_ptr(l, slot,
-						struct btrfs_extent_item);
-			flags = btrfs_extent_flags(l, extent);
-			generation = btrfs_extent_generation(l, extent);
-
-			if ((flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) &&
-			    (key.objectid < logical ||
-			     key.objectid + bytes >
-			     logical + map->stripe_len)) {
-				btrfs_err(fs_info,
-					   "scrub: tree block %llu spanning stripes, ignored. logical=%llu",
-				       key.objectid, logical);
-				spin_lock(&sctx->stat_lock);
-				sctx->stat.uncorrectable_errors++;
-				spin_unlock(&sctx->stat_lock);
-				goto next;
-			}
-
-again:
-			extent_logical = key.objectid;
-			extent_len = bytes;
-
-			/*
-			 * trim extent to this stripe
-			 */
-			if (extent_logical < logical) {
-				extent_len -= logical - extent_logical;
-				extent_logical = logical;
-			}
-			if (extent_logical + extent_len >
-			    logical + map->stripe_len) {
-				extent_len = logical + map->stripe_len -
-					     extent_logical;
-			}
+		ret = scrub_simple_mirror(sctx, root, csum_root, bg, map,
+				bg->start, bg->length, scrub_dev,
+				map->stripes[stripe_index].physical,
+				stripe_index + 1);
+		offset = 0;
+		goto out;
+	}
+	if (profile & (BTRFS_BLOCK_GROUP_RAID0 | BTRFS_BLOCK_GROUP_RAID10)) {
+		ret = scrub_simple_stripe(sctx, root, csum_root, bg, map,
+					  scrub_dev, stripe_index);
+		offset = map->stripe_len * (stripe_index / map->sub_stripes);
+		goto out;
+	}
 
-			extent_physical = extent_logical - logical + physical;
-			extent_dev = scrub_dev;
-			extent_mirror_num = mirror_num;
-			if (sctx->is_dev_replace)
-				scrub_remap_extent(fs_info, extent_logical,
-						   extent_len, &extent_physical,
-						   &extent_dev,
-						   &extent_mirror_num);
-
-			ret = btrfs_lookup_csums_range(csum_root,
-						       extent_logical,
-						       extent_logical +
-						       extent_len - 1,
-						       &sctx->csum_list, 1);
-			if (ret)
-				goto out;
+	/* Only RAID56 goes through the old code */
+	ASSERT(map->type & BTRFS_BLOCK_GROUP_RAID56_MASK);
+	ret = 0;
 
-			ret = scrub_extent(sctx, map, extent_logical, extent_len,
-					   extent_physical, extent_dev, flags,
-					   generation, extent_mirror_num,
-					   extent_logical - logical + physical);
+	/* Calculate the logical end of the stripe */
+	get_raid56_logic_offset(physical_end, stripe_index,
+				map, &logic_end, NULL);
+	logic_end += chunk_logical;
 
-			scrub_free_csums(sctx);
+	/* Initialize @offset in case we need to go to out: label */
+	get_raid56_logic_offset(physical, stripe_index, map, &offset, NULL);
+	increment = map->stripe_len * nr_data_stripes(map);
 
+	/*
+	 * Due to the rotation, for RAID56 it's better to iterate each stripe
+	 * using their physical offset.
+	 */
+	while (physical < physical_end) {
+		ret = get_raid56_logic_offset(physical, stripe_index, map,
+					      &logical, &stripe_logical);
+		logical += chunk_logical;
+		if (ret) {
+			/* it is parity strip */
+			stripe_logical += chunk_logical;
+			stripe_end = stripe_logical + increment;
+			ret = scrub_raid56_parity(sctx, map, scrub_dev,
+						  stripe_logical,
+						  stripe_end);
 			if (ret)
 				goto out;
+			goto next;
+		}
 
-			if (extent_logical + extent_len <
-			    key.objectid + bytes) {
-				if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) {
-					/*
-					 * loop until we find next data stripe
-					 * or we have finished all stripes.
-					 */
-loop:
-					physical += map->stripe_len;
-					ret = get_raid56_logic_offset(physical,
-							num, map, &logical,
-							&stripe_logical);
-					logical += base;
-
-					if (ret && physical < physical_end) {
-						stripe_logical += base;
-						stripe_end = stripe_logical +
-								increment;
-						ret = scrub_raid56_parity(sctx,
-							map, scrub_dev, ppath,
-							stripe_logical,
-							stripe_end);
-						if (ret)
-							goto out;
-						goto loop;
-					}
-				} else {
-					physical += map->stripe_len;
-					logical += increment;
-				}
-				if (logical < key.objectid + bytes) {
-					cond_resched();
-					goto again;
-				}
-
-				if (physical >= physical_end) {
-					stop_loop = 1;
-					break;
-				}
-			}
+		/*
+		 * Now we're at a data stripe, scrub each extents in the range.
+		 *
+		 * At this stage, if we ignore the repair part, inside each data
+		 * stripe it is no different than SINGLE profile.
+		 * We can reuse scrub_simple_mirror() here, as the repair part
+		 * is still based on @mirror_num.
+		 */
+		ret = scrub_simple_mirror(sctx, root, csum_root, bg, map,
+					  logical, map->stripe_len,
+					  scrub_dev, physical, 1);
+		if (ret < 0)
+			goto out;
 next:
-			path->slots[0]++;
-		}
-		btrfs_release_path(path);
-skip:
 		logical += increment;
 		physical += map->stripe_len;
 		spin_lock(&sctx->stat_lock);
 		if (stop_loop)
-			sctx->stat.last_physical = map->stripes[num].physical +
-						   length;
+			sctx->stat.last_physical =
+				map->stripes[stripe_index].physical + dev_stripe_len;
 		else
 			sctx->stat.last_physical = physical;
 		spin_unlock(&sctx->stat_lock);
@@ -3409,15 +3714,26 @@ out:
 
 	blk_finish_plug(&plug);
 	btrfs_free_path(path);
-	btrfs_free_path(ppath);
+
+	if (sctx->is_dev_replace && ret >= 0) {
+		int ret2;
+
+		ret2 = sync_write_pointer_for_zoned(sctx,
+				chunk_logical + offset,
+				map->stripes[stripe_index].physical,
+				physical_end);
+		if (ret2)
+			ret = ret2;
+	}
+
 	return ret < 0 ? ret : 0;
 }
 
 static noinline_for_stack int scrub_chunk(struct scrub_ctx *sctx,
+					  struct btrfs_block_group *bg,
 					  struct btrfs_device *scrub_dev,
-					  u64 chunk_offset, u64 length,
 					  u64 dev_offset,
-					  struct btrfs_block_group *cache)
+					  u64 dev_extent_len)
 {
 	struct btrfs_fs_info *fs_info = sctx->fs_info;
 	struct extent_map_tree *map_tree = &fs_info->mapping_tree;
@@ -3427,7 +3743,7 @@ static noinline_for_stack int scrub_chunk(struct scrub_ctx *sctx,
 	int ret = 0;
 
 	read_lock(&map_tree->lock);
-	em = lookup_extent_mapping(map_tree, chunk_offset, 1);
+	em = lookup_extent_mapping(map_tree, bg->start, bg->length);
 	read_unlock(&map_tree->lock);
 
 	if (!em) {
@@ -3435,26 +3751,23 @@ static noinline_for_stack int scrub_chunk(struct scrub_ctx *sctx,
 		 * Might have been an unused block group deleted by the cleaner
 		 * kthread or relocation.
 		 */
-		spin_lock(&cache->lock);
-		if (!cache->removed)
+		spin_lock(&bg->lock);
+		if (!test_bit(BLOCK_GROUP_FLAG_REMOVED, &bg->runtime_flags))
 			ret = -EINVAL;
-		spin_unlock(&cache->lock);
+		spin_unlock(&bg->lock);
 
 		return ret;
 	}
-
-	map = em->map_lookup;
-	if (em->start != chunk_offset)
+	if (em->start != bg->start)
 		goto out;
-
-	if (em->len < length)
+	if (em->len < dev_extent_len)
 		goto out;
 
+	map = em->map_lookup;
 	for (i = 0; i < map->num_stripes; ++i) {
 		if (map->stripes[i].dev->bdev == scrub_dev->bdev &&
 		    map->stripes[i].physical == dev_offset) {
-			ret = scrub_stripe(sctx, map, scrub_dev, i,
-					   chunk_offset, length);
+			ret = scrub_stripe(sctx, bg, em, scrub_dev, i);
 			if (ret)
 				goto out;
 		}
@@ -3465,6 +3778,25 @@ out:
 	return ret;
 }
 
+static int finish_extent_writes_for_zoned(struct btrfs_root *root,
+					  struct btrfs_block_group *cache)
+{
+	struct btrfs_fs_info *fs_info = cache->fs_info;
+	struct btrfs_trans_handle *trans;
+
+	if (!btrfs_is_zoned(fs_info))
+		return 0;
+
+	btrfs_wait_block_group_reservations(cache);
+	btrfs_wait_nocow_writers(cache);
+	btrfs_wait_ordered_roots(fs_info, U64_MAX, cache->start, cache->length);
+
+	trans = btrfs_join_transaction(root);
+	if (IS_ERR(trans))
+		return PTR_ERR(trans);
+	return btrfs_commit_transaction(trans);
+}
+
 static noinline_for_stack
 int scrub_enumerate_chunks(struct scrub_ctx *sctx,
 			   struct btrfs_device *scrub_dev, u64 start, u64 end)
@@ -3473,7 +3805,6 @@ int scrub_enumerate_chunks(struct scrub_ctx *sctx,
 	struct btrfs_path *path;
 	struct btrfs_fs_info *fs_info = sctx->fs_info;
 	struct btrfs_root *root = fs_info->dev_root;
-	u64 length;
 	u64 chunk_offset;
 	int ret = 0;
 	int ro_set;
@@ -3497,6 +3828,8 @@ int scrub_enumerate_chunks(struct scrub_ctx *sctx,
 	key.type = BTRFS_DEV_EXTENT_KEY;
 
 	while (1) {
+		u64 dev_extent_len;
+
 		ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
 		if (ret < 0)
 			break;
@@ -3533,9 +3866,9 @@ int scrub_enumerate_chunks(struct scrub_ctx *sctx,
 			break;
 
 		dev_extent = btrfs_item_ptr(l, slot, struct btrfs_dev_extent);
-		length = btrfs_dev_extent_length(l, dev_extent);
+		dev_extent_len = btrfs_dev_extent_length(l, dev_extent);
 
-		if (found_key.offset + length <= start)
+		if (found_key.offset + dev_extent_len <= start)
 			goto skip;
 
 		chunk_offset = btrfs_dev_extent_chunk_offset(l, dev_extent);
@@ -3551,6 +3884,55 @@ int scrub_enumerate_chunks(struct scrub_ctx *sctx,
 		if (!cache)
 			goto skip;
 
+		ASSERT(cache->start <= chunk_offset);
+		/*
+		 * We are using the commit root to search for device extents, so
+		 * that means we could have found a device extent item from a
+		 * block group that was deleted in the current transaction. The
+		 * logical start offset of the deleted block group, stored at
+		 * @chunk_offset, might be part of the logical address range of
+		 * a new block group (which uses different physical extents).
+		 * In this case btrfs_lookup_block_group() has returned the new
+		 * block group, and its start address is less than @chunk_offset.
+		 *
+		 * We skip such new block groups, because it's pointless to
+		 * process them, as we won't find their extents because we search
+		 * for them using the commit root of the extent tree. For a device
+		 * replace it's also fine to skip it, we won't miss copying them
+		 * to the target device because we have the write duplication
+		 * setup through the regular write path (by btrfs_map_block()),
+		 * and we have committed a transaction when we started the device
+		 * replace, right after setting up the device replace state.
+		 */
+		if (cache->start < chunk_offset) {
+			btrfs_put_block_group(cache);
+			goto skip;
+		}
+
+		if (sctx->is_dev_replace && btrfs_is_zoned(fs_info)) {
+			if (!test_bit(BLOCK_GROUP_FLAG_TO_COPY, &cache->runtime_flags)) {
+				btrfs_put_block_group(cache);
+				goto skip;
+			}
+		}
+
+		/*
+		 * Make sure that while we are scrubbing the corresponding block
+		 * group doesn't get its logical address and its device extents
+		 * reused for another block group, which can possibly be of a
+		 * different type and different profile. We do this to prevent
+		 * false error detections and crashes due to bogus attempts to
+		 * repair extents.
+		 */
+		spin_lock(&cache->lock);
+		if (test_bit(BLOCK_GROUP_FLAG_REMOVED, &cache->runtime_flags)) {
+			spin_unlock(&cache->lock);
+			btrfs_put_block_group(cache);
+			goto skip;
+		}
+		btrfs_freeze_block_group(cache);
+		spin_unlock(&cache->lock);
+
 		/*
 		 * we need call btrfs_inc_block_group_ro() with scrubs_paused,
 		 * to avoid deadlock caused by:
@@ -3592,6 +3974,16 @@ int scrub_enumerate_chunks(struct scrub_ctx *sctx,
 		 * group is not RO.
 		 */
 		ret = btrfs_inc_block_group_ro(cache, sctx->is_dev_replace);
+		if (!ret && sctx->is_dev_replace) {
+			ret = finish_extent_writes_for_zoned(root, cache);
+			if (ret) {
+				btrfs_dec_block_group_ro(cache);
+				scrub_pause_off(fs_info);
+				btrfs_put_block_group(cache);
+				break;
+			}
+		}
+
 		if (ret == 0) {
 			ro_set = 1;
 		} else if (ret == -ENOSPC && !sctx->is_dev_replace) {
@@ -3603,9 +3995,17 @@ int scrub_enumerate_chunks(struct scrub_ctx *sctx,
 			 * commit_transactions.
 			 */
 			ro_set = 0;
+		} else if (ret == -ETXTBSY) {
+			btrfs_warn(fs_info,
+		   "skipping scrub of block group %llu due to active swapfile",
+				   cache->start);
+			scrub_pause_off(fs_info);
+			ret = 0;
+			goto skip_unfreeze;
 		} else {
 			btrfs_warn(fs_info,
 				   "failed setting block group ro: %d", ret);
+			btrfs_unfreeze_block_group(cache);
 			btrfs_put_block_group(cache);
 			scrub_pause_off(fs_info);
 			break;
@@ -3624,13 +4024,13 @@ int scrub_enumerate_chunks(struct scrub_ctx *sctx,
 
 		scrub_pause_off(fs_info);
 		down_write(&dev_replace->rwsem);
-		dev_replace->cursor_right = found_key.offset + length;
+		dev_replace->cursor_right = found_key.offset + dev_extent_len;
 		dev_replace->cursor_left = found_key.offset;
 		dev_replace->item_needs_writeback = 1;
 		up_write(&dev_replace->rwsem);
 
-		ret = scrub_chunk(sctx, scrub_dev, chunk_offset, length,
-				  found_key.offset, cache);
+		ret = scrub_chunk(sctx, cache, scrub_dev, found_key.offset,
+				  dev_extent_len);
 
 		/*
 		 * flush, submit all pending read and write bios, afterwards
@@ -3664,6 +4064,11 @@ int scrub_enumerate_chunks(struct scrub_ctx *sctx,
 
 		scrub_pause_off(fs_info);
 
+		if (sctx->is_dev_replace &&
+		    !btrfs_finish_block_group_to_copy(dev_replace->srcdev,
+						      cache, found_key.offset))
+			ro_set = 0;
+
 		down_write(&dev_replace->rwsem);
 		dev_replace->cursor_left = dev_replace->cursor_right;
 		dev_replace->item_needs_writeback = 1;
@@ -3680,8 +4085,8 @@ int scrub_enumerate_chunks(struct scrub_ctx *sctx,
 		 * balance is triggered or it becomes used and unused again.
 		 */
 		spin_lock(&cache->lock);
-		if (!cache->removed && !cache->ro && cache->reserved == 0 &&
-		    cache->used == 0) {
+		if (!test_bit(BLOCK_GROUP_FLAG_REMOVED, &cache->runtime_flags) &&
+		    !cache->ro && cache->reserved == 0 && cache->used == 0) {
 			spin_unlock(&cache->lock);
 			if (btrfs_test_opt(fs_info, DISCARD_ASYNC))
 				btrfs_discard_queue_work(&fs_info->discard_ctl,
@@ -3691,7 +4096,8 @@ int scrub_enumerate_chunks(struct scrub_ctx *sctx,
 		} else {
 			spin_unlock(&cache->lock);
 		}
-
+skip_unfreeze:
+		btrfs_unfreeze_block_group(cache);
 		btrfs_put_block_group(cache);
 		if (ret)
 			break;
@@ -3705,7 +4111,7 @@ int scrub_enumerate_chunks(struct scrub_ctx *sctx,
 			break;
 		}
 skip:
-		key.offset = found_key.offset + length;
+		key.offset = found_key.offset + dev_extent_len;
 		btrfs_release_path(path);
 	}
 
@@ -3723,8 +4129,8 @@ static noinline_for_stack int scrub_supers(struct scrub_ctx *sctx,
 	int	ret;
 	struct btrfs_fs_info *fs_info = sctx->fs_info;
 
-	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state))
-		return -EIO;
+	if (BTRFS_FS_ERROR(fs_info))
+		return -EROFS;
 
 	/* Seed devices of a new filesystem has their own generation. */
 	if (scrub_dev->fs_devices != fs_info->fs_devices)
@@ -3737,10 +4143,12 @@ static noinline_for_stack int scrub_supers(struct scrub_ctx *sctx,
 		if (bytenr + BTRFS_SUPER_INFO_SIZE >
 		    scrub_dev->commit_total_bytes)
 			break;
+		if (!btrfs_check_super_location(scrub_dev, bytenr))
+			continue;
 
-		ret = scrub_pages(sctx, bytenr, BTRFS_SUPER_INFO_SIZE, bytenr,
-				  scrub_dev, BTRFS_EXTENT_FLAG_SUPER, gen, i,
-				  NULL, 1, bytenr);
+		ret = scrub_sectors(sctx, bytenr, BTRFS_SUPER_INFO_SIZE, bytenr,
+				    scrub_dev, BTRFS_EXTENT_FLAG_SUPER, gen, i,
+				    NULL, bytenr);
 		if (ret)
 			return ret;
 	}
@@ -3749,125 +4157,136 @@ static noinline_for_stack int scrub_supers(struct scrub_ctx *sctx,
 	return 0;
 }
 
+static void scrub_workers_put(struct btrfs_fs_info *fs_info)
+{
+	if (refcount_dec_and_mutex_lock(&fs_info->scrub_workers_refcnt,
+					&fs_info->scrub_lock)) {
+		struct workqueue_struct *scrub_workers = fs_info->scrub_workers;
+		struct workqueue_struct *scrub_wr_comp =
+						fs_info->scrub_wr_completion_workers;
+		struct workqueue_struct *scrub_parity =
+						fs_info->scrub_parity_workers;
+
+		fs_info->scrub_workers = NULL;
+		fs_info->scrub_wr_completion_workers = NULL;
+		fs_info->scrub_parity_workers = NULL;
+		mutex_unlock(&fs_info->scrub_lock);
+
+		if (scrub_workers)
+			destroy_workqueue(scrub_workers);
+		if (scrub_wr_comp)
+			destroy_workqueue(scrub_wr_comp);
+		if (scrub_parity)
+			destroy_workqueue(scrub_parity);
+	}
+}
+
 /*
  * get a reference count on fs_info->scrub_workers. start worker if necessary
  */
 static noinline_for_stack int scrub_workers_get(struct btrfs_fs_info *fs_info,
 						int is_dev_replace)
 {
+	struct workqueue_struct *scrub_workers = NULL;
+	struct workqueue_struct *scrub_wr_comp = NULL;
+	struct workqueue_struct *scrub_parity = NULL;
 	unsigned int flags = WQ_FREEZABLE | WQ_UNBOUND;
 	int max_active = fs_info->thread_pool_size;
+	int ret = -ENOMEM;
 
-	lockdep_assert_held(&fs_info->scrub_lock);
+	if (refcount_inc_not_zero(&fs_info->scrub_workers_refcnt))
+		return 0;
 
-	if (refcount_read(&fs_info->scrub_workers_refcnt) == 0) {
-		ASSERT(fs_info->scrub_workers == NULL);
-		fs_info->scrub_workers = btrfs_alloc_workqueue(fs_info, "scrub",
-				flags, is_dev_replace ? 1 : max_active, 4);
-		if (!fs_info->scrub_workers)
-			goto fail_scrub_workers;
-
-		ASSERT(fs_info->scrub_wr_completion_workers == NULL);
-		fs_info->scrub_wr_completion_workers =
-			btrfs_alloc_workqueue(fs_info, "scrubwrc", flags,
-					      max_active, 2);
-		if (!fs_info->scrub_wr_completion_workers)
-			goto fail_scrub_wr_completion_workers;
-
-		ASSERT(fs_info->scrub_parity_workers == NULL);
-		fs_info->scrub_parity_workers =
-			btrfs_alloc_workqueue(fs_info, "scrubparity", flags,
-					      max_active, 2);
-		if (!fs_info->scrub_parity_workers)
-			goto fail_scrub_parity_workers;
+	scrub_workers = alloc_workqueue("btrfs-scrub", flags,
+					is_dev_replace ? 1 : max_active);
+	if (!scrub_workers)
+		goto fail_scrub_workers;
+
+	scrub_wr_comp = alloc_workqueue("btrfs-scrubwrc", flags, max_active);
+	if (!scrub_wr_comp)
+		goto fail_scrub_wr_completion_workers;
 
+	scrub_parity = alloc_workqueue("btrfs-scrubparity", flags, max_active);
+	if (!scrub_parity)
+		goto fail_scrub_parity_workers;
+
+	mutex_lock(&fs_info->scrub_lock);
+	if (refcount_read(&fs_info->scrub_workers_refcnt) == 0) {
+		ASSERT(fs_info->scrub_workers == NULL &&
+		       fs_info->scrub_wr_completion_workers == NULL &&
+		       fs_info->scrub_parity_workers == NULL);
+		fs_info->scrub_workers = scrub_workers;
+		fs_info->scrub_wr_completion_workers = scrub_wr_comp;
+		fs_info->scrub_parity_workers = scrub_parity;
 		refcount_set(&fs_info->scrub_workers_refcnt, 1);
-	} else {
-		refcount_inc(&fs_info->scrub_workers_refcnt);
+		mutex_unlock(&fs_info->scrub_lock);
+		return 0;
 	}
-	return 0;
+	/* Other thread raced in and created the workers for us */
+	refcount_inc(&fs_info->scrub_workers_refcnt);
+	mutex_unlock(&fs_info->scrub_lock);
 
+	ret = 0;
+	destroy_workqueue(scrub_parity);
 fail_scrub_parity_workers:
-	btrfs_destroy_workqueue(fs_info->scrub_wr_completion_workers);
+	destroy_workqueue(scrub_wr_comp);
 fail_scrub_wr_completion_workers:
-	btrfs_destroy_workqueue(fs_info->scrub_workers);
+	destroy_workqueue(scrub_workers);
 fail_scrub_workers:
-	return -ENOMEM;
+	return ret;
 }
 
 int btrfs_scrub_dev(struct btrfs_fs_info *fs_info, u64 devid, u64 start,
 		    u64 end, struct btrfs_scrub_progress *progress,
 		    int readonly, int is_dev_replace)
 {
+	struct btrfs_dev_lookup_args args = { .devid = devid };
 	struct scrub_ctx *sctx;
 	int ret;
 	struct btrfs_device *dev;
 	unsigned int nofs_flag;
-	struct btrfs_workqueue *scrub_workers = NULL;
-	struct btrfs_workqueue *scrub_wr_comp = NULL;
-	struct btrfs_workqueue *scrub_parity = NULL;
+	bool need_commit = false;
 
 	if (btrfs_fs_closing(fs_info))
 		return -EAGAIN;
 
-	if (fs_info->nodesize > BTRFS_STRIPE_LEN) {
-		/*
-		 * in this case scrub is unable to calculate the checksum
-		 * the way scrub is implemented. Do not handle this
-		 * situation at all because it won't ever happen.
-		 */
-		btrfs_err(fs_info,
-			   "scrub: size assumption nodesize <= BTRFS_STRIPE_LEN (%d <= %d) fails",
-		       fs_info->nodesize,
-		       BTRFS_STRIPE_LEN);
-		return -EINVAL;
-	}
-
-	if (fs_info->sectorsize != PAGE_SIZE) {
-		/* not supported for data w/o checksums */
-		btrfs_err_rl(fs_info,
-			   "scrub: size assumption sectorsize != PAGE_SIZE (%d != %lu) fails",
-		       fs_info->sectorsize, PAGE_SIZE);
-		return -EINVAL;
-	}
+	/* At mount time we have ensured nodesize is in the range of [4K, 64K]. */
+	ASSERT(fs_info->nodesize <= BTRFS_STRIPE_LEN);
 
-	if (fs_info->nodesize >
-	    PAGE_SIZE * SCRUB_MAX_PAGES_PER_BLOCK ||
-	    fs_info->sectorsize > PAGE_SIZE * SCRUB_MAX_PAGES_PER_BLOCK) {
-		/*
-		 * would exhaust the array bounds of pagev member in
-		 * struct scrub_block
-		 */
-		btrfs_err(fs_info,
-			  "scrub: size assumption nodesize and sectorsize <= SCRUB_MAX_PAGES_PER_BLOCK (%d <= %d && %d <= %d) fails",
-		       fs_info->nodesize,
-		       SCRUB_MAX_PAGES_PER_BLOCK,
-		       fs_info->sectorsize,
-		       SCRUB_MAX_PAGES_PER_BLOCK);
-		return -EINVAL;
-	}
+	/*
+	 * SCRUB_MAX_SECTORS_PER_BLOCK is calculated using the largest possible
+	 * value (max nodesize / min sectorsize), thus nodesize should always
+	 * be fine.
+	 */
+	ASSERT(fs_info->nodesize <=
+	       SCRUB_MAX_SECTORS_PER_BLOCK << fs_info->sectorsize_bits);
 
 	/* Allocate outside of device_list_mutex */
 	sctx = scrub_setup_ctx(fs_info, is_dev_replace);
 	if (IS_ERR(sctx))
 		return PTR_ERR(sctx);
 
+	ret = scrub_workers_get(fs_info, is_dev_replace);
+	if (ret)
+		goto out_free_ctx;
+
 	mutex_lock(&fs_info->fs_devices->device_list_mutex);
-	dev = btrfs_find_device(fs_info->fs_devices, devid, NULL, NULL, true);
+	dev = btrfs_find_device(fs_info->fs_devices, &args);
 	if (!dev || (test_bit(BTRFS_DEV_STATE_MISSING, &dev->dev_state) &&
 		     !is_dev_replace)) {
 		mutex_unlock(&fs_info->fs_devices->device_list_mutex);
 		ret = -ENODEV;
-		goto out_free_ctx;
+		goto out;
 	}
 
 	if (!is_dev_replace && !readonly &&
 	    !test_bit(BTRFS_DEV_STATE_WRITEABLE, &dev->dev_state)) {
 		mutex_unlock(&fs_info->fs_devices->device_list_mutex);
-		btrfs_err_in_rcu(fs_info, "scrub: device %s is not writable",
-				rcu_str_deref(dev->name));
+		btrfs_err_in_rcu(fs_info,
+			"scrub on devid %llu: filesystem on %s is not writable",
+				 devid, rcu_str_deref(dev->name));
 		ret = -EROFS;
-		goto out_free_ctx;
+		goto out;
 	}
 
 	mutex_lock(&fs_info->scrub_lock);
@@ -3876,7 +4295,7 @@ int btrfs_scrub_dev(struct btrfs_fs_info *fs_info, u64 devid, u64 start,
 		mutex_unlock(&fs_info->scrub_lock);
 		mutex_unlock(&fs_info->fs_devices->device_list_mutex);
 		ret = -EIO;
-		goto out_free_ctx;
+		goto out;
 	}
 
 	down_read(&fs_info->dev_replace.rwsem);
@@ -3887,17 +4306,10 @@ int btrfs_scrub_dev(struct btrfs_fs_info *fs_info, u64 devid, u64 start,
 		mutex_unlock(&fs_info->scrub_lock);
 		mutex_unlock(&fs_info->fs_devices->device_list_mutex);
 		ret = -EINPROGRESS;
-		goto out_free_ctx;
+		goto out;
 	}
 	up_read(&fs_info->dev_replace.rwsem);
 
-	ret = scrub_workers_get(fs_info, is_dev_replace);
-	if (ret) {
-		mutex_unlock(&fs_info->scrub_lock);
-		mutex_unlock(&fs_info->fs_devices->device_list_mutex);
-		goto out_free_ctx;
-	}
-
 	sctx->readonly = readonly;
 	dev->scrub_ctx = sctx;
 	mutex_unlock(&fs_info->fs_devices->device_list_mutex);
@@ -3913,7 +4325,7 @@ int btrfs_scrub_dev(struct btrfs_fs_info *fs_info, u64 devid, u64 start,
 	/*
 	 * In order to avoid deadlock with reclaim when there is a transaction
 	 * trying to pause scrub, make sure we use GFP_NOFS for all the
-	 * allocations done at btrfs_scrub_pages() and scrub_pages_for_parity()
+	 * allocations done at btrfs_scrub_sectors() and scrub_sectors_for_parity()
 	 * invoked by our callees. The pausing request is done when the
 	 * transaction commit starts, and it blocks the transaction until scrub
 	 * is paused (done at specific points at scrub_stripe() or right above
@@ -3921,6 +4333,12 @@ int btrfs_scrub_dev(struct btrfs_fs_info *fs_info, u64 devid, u64 start,
 	 */
 	nofs_flag = memalloc_nofs_save();
 	if (!is_dev_replace) {
+		u64 old_super_errors;
+
+		spin_lock(&sctx->stat_lock);
+		old_super_errors = sctx->stat.super_errors;
+		spin_unlock(&sctx->stat_lock);
+
 		btrfs_info(fs_info, "scrub: started on devid %llu", devid);
 		/*
 		 * by holding device list mutex, we can
@@ -3929,6 +4347,16 @@ int btrfs_scrub_dev(struct btrfs_fs_info *fs_info, u64 devid, u64 start,
 		mutex_lock(&fs_info->fs_devices->device_list_mutex);
 		ret = scrub_supers(sctx, dev);
 		mutex_unlock(&fs_info->fs_devices->device_list_mutex);
+
+		spin_lock(&sctx->stat_lock);
+		/*
+		 * Super block errors found, but we can not commit transaction
+		 * at current context, since btrfs_commit_transaction() needs
+		 * to pause the current running scrub (hold by ourselves).
+		 */
+		if (sctx->stat.super_errors > old_super_errors && !sctx->readonly)
+			need_commit = true;
+		spin_unlock(&sctx->stat_lock);
 	}
 
 	if (!ret)
@@ -3950,24 +4378,33 @@ int btrfs_scrub_dev(struct btrfs_fs_info *fs_info, u64 devid, u64 start,
 
 	mutex_lock(&fs_info->scrub_lock);
 	dev->scrub_ctx = NULL;
-	if (refcount_dec_and_test(&fs_info->scrub_workers_refcnt)) {
-		scrub_workers = fs_info->scrub_workers;
-		scrub_wr_comp = fs_info->scrub_wr_completion_workers;
-		scrub_parity = fs_info->scrub_parity_workers;
-
-		fs_info->scrub_workers = NULL;
-		fs_info->scrub_wr_completion_workers = NULL;
-		fs_info->scrub_parity_workers = NULL;
-	}
 	mutex_unlock(&fs_info->scrub_lock);
 
-	btrfs_destroy_workqueue(scrub_workers);
-	btrfs_destroy_workqueue(scrub_wr_comp);
-	btrfs_destroy_workqueue(scrub_parity);
+	scrub_workers_put(fs_info);
 	scrub_put_ctx(sctx);
 
+	/*
+	 * We found some super block errors before, now try to force a
+	 * transaction commit, as scrub has finished.
+	 */
+	if (need_commit) {
+		struct btrfs_trans_handle *trans;
+
+		trans = btrfs_start_transaction(fs_info->tree_root, 0);
+		if (IS_ERR(trans)) {
+			ret = PTR_ERR(trans);
+			btrfs_err(fs_info,
+	"scrub: failed to start transaction to fix super block errors: %d", ret);
+			return ret;
+		}
+		ret = btrfs_commit_transaction(trans);
+		if (ret < 0)
+			btrfs_err(fs_info,
+	"scrub: failed to commit transaction to fix super block errors: %d", ret);
+	}
 	return ret;
-
+out:
+	scrub_workers_put(fs_info);
 out_free_ctx:
 	scrub_free_ctx(sctx);
 
@@ -4042,11 +4479,12 @@ int btrfs_scrub_cancel_dev(struct btrfs_device *dev)
 int btrfs_scrub_progress(struct btrfs_fs_info *fs_info, u64 devid,
 			 struct btrfs_scrub_progress *progress)
 {
+	struct btrfs_dev_lookup_args args = { .devid = devid };
 	struct btrfs_device *dev;
 	struct scrub_ctx *sctx = NULL;
 
 	mutex_lock(&fs_info->fs_devices->device_list_mutex);
-	dev = btrfs_find_device(fs_info->fs_devices, devid, NULL, NULL, true);
+	dev = btrfs_find_device(fs_info->fs_devices, &args);
 	if (dev)
 		sctx = dev->scrub_ctx;
 	if (sctx)
@@ -4056,27 +4494,27 @@ int btrfs_scrub_progress(struct btrfs_fs_info *fs_info, u64 devid,
 	return dev ? (sctx ? 0 : -ENOTCONN) : -ENODEV;
 }
 
-static void scrub_remap_extent(struct btrfs_fs_info *fs_info,
-			       u64 extent_logical, u64 extent_len,
-			       u64 *extent_physical,
-			       struct btrfs_device **extent_dev,
-			       int *extent_mirror_num)
+static void scrub_find_good_copy(struct btrfs_fs_info *fs_info,
+				 u64 extent_logical, u32 extent_len,
+				 u64 *extent_physical,
+				 struct btrfs_device **extent_dev,
+				 int *extent_mirror_num)
 {
 	u64 mapped_length;
-	struct btrfs_bio *bbio = NULL;
+	struct btrfs_io_context *bioc = NULL;
 	int ret;
 
 	mapped_length = extent_len;
 	ret = btrfs_map_block(fs_info, BTRFS_MAP_READ, extent_logical,
-			      &mapped_length, &bbio, 0);
-	if (ret || !bbio || mapped_length < extent_len ||
-	    !bbio->stripes[0].dev->bdev) {
-		btrfs_put_bbio(bbio);
+			      &mapped_length, &bioc, 0);
+	if (ret || !bioc || mapped_length < extent_len ||
+	    !bioc->stripes[0].dev->bdev) {
+		btrfs_put_bioc(bioc);
 		return;
 	}
 
-	*extent_physical = bbio->stripes[0].physical;
-	*extent_mirror_num = bbio->mirror_num;
-	*extent_dev = bbio->stripes[0].dev;
-	btrfs_put_bbio(bbio);
+	*extent_physical = bioc->stripes[0].physical;
+	*extent_mirror_num = bioc->mirror_num;
+	*extent_dev = bioc->stripes[0].dev;
+	btrfs_put_bioc(bioc);
 }
diff --git a/fs/btrfs/send.c b/fs/btrfs/send.c
index a055b657cb85..145c84b44fd0 100644
--- a/fs/btrfs/send.c
+++ b/fs/btrfs/send.c
@@ -15,14 +15,18 @@
 #include <linux/string.h>
 #include <linux/compat.h>
 #include <linux/crc32c.h>
+#include <linux/fsverity.h>
 
 #include "send.h"
+#include "ctree.h"
 #include "backref.h"
 #include "locking.h"
 #include "disk-io.h"
 #include "btrfs_inode.h"
 #include "transaction.h"
 #include "compression.h"
+#include "xattr.h"
+#include "print-tree.h"
 
 /*
  * Maximum number of references an extent can have in order for us to attempt to
@@ -80,9 +84,15 @@ struct send_ctx {
 	char *send_buf;
 	u32 send_size;
 	u32 send_max_size;
-	u64 total_send_size;
-	u64 cmd_send_size[BTRFS_SEND_C_MAX + 1];
+	/*
+	 * Whether BTRFS_SEND_A_DATA attribute was already added to current
+	 * command (since protocol v2, data must be the last attribute).
+	 */
+	bool put_data;
+	struct page **send_buf_pages;
 	u64 flags;	/* 'flags' member of btrfs_ioctl_send_args is u64 */
+	/* Protocol version compatibility requested */
+	u32 proto;
 
 	struct btrfs_root *send_root;
 	struct btrfs_root *parent_root;
@@ -95,20 +105,31 @@ struct send_ctx {
 	struct btrfs_key *cmp_key;
 
 	/*
+	 * Keep track of the generation of the last transaction that was used
+	 * for relocating a block group. This is periodically checked in order
+	 * to detect if a relocation happened since the last check, so that we
+	 * don't operate on stale extent buffers for nodes (level >= 1) or on
+	 * stale disk_bytenr values of file extent items.
+	 */
+	u64 last_reloc_trans;
+
+	/*
 	 * infos of the currently processed inode. In case of deleted inodes,
 	 * these are the values from the deleted inode.
 	 */
 	u64 cur_ino;
 	u64 cur_inode_gen;
-	int cur_inode_new;
-	int cur_inode_new_gen;
-	int cur_inode_deleted;
 	u64 cur_inode_size;
 	u64 cur_inode_mode;
 	u64 cur_inode_rdev;
 	u64 cur_inode_last_extent;
 	u64 cur_inode_next_write_offset;
+	bool cur_inode_new;
+	bool cur_inode_new_gen;
+	bool cur_inode_deleted;
 	bool ignore_cur_inode;
+	bool cur_inode_needs_verity;
+	void *verity_descriptor;
 
 	u64 send_progress;
 
@@ -119,9 +140,14 @@ struct send_ctx {
 	struct list_head name_cache_list;
 	int name_cache_size;
 
+	/*
+	 * The inode we are currently processing. It's not NULL only when we
+	 * need to issue write commands for data extents from this inode.
+	 */
+	struct inode *cur_inode;
 	struct file_ra_state ra;
-
-	char *read_buf;
+	u64 page_cache_clear_start;
+	bool clean_page_cache;
 
 	/*
 	 * We process inodes by their increasing order, so if before an
@@ -217,6 +243,9 @@ struct send_ctx {
 	 * Indexed by the inode number of the directory to be deleted.
 	 */
 	struct rb_root orphan_dirs;
+
+	struct rb_root rbtree_new_refs;
+	struct rb_root rbtree_deleted_refs;
 };
 
 struct pending_dir_move {
@@ -237,6 +266,7 @@ struct waiting_dir_move {
 	 * after this directory is moved, we can try to rmdir the ino rmdir_ino.
 	 */
 	u64 rmdir_ino;
+	u64 rmdir_gen;
 	bool orphanized;
 };
 
@@ -277,11 +307,6 @@ enum btrfs_compare_tree_result {
 	BTRFS_COMPARE_TREE_CHANGED,
 	BTRFS_COMPARE_TREE_SAME,
 };
-typedef int (*btrfs_changed_cb_t)(struct btrfs_path *left_path,
-				  struct btrfs_path *right_path,
-				  struct btrfs_key *key,
-				  enum btrfs_compare_tree_result result,
-				  void *ctx);
 
 __cold
 static void inconsistent_snapshot_error(struct send_ctx *sctx,
@@ -317,12 +342,23 @@ static void inconsistent_snapshot_error(struct send_ctx *sctx,
 		   sctx->parent_root->root_key.objectid : 0));
 }
 
+__maybe_unused
+static bool proto_cmd_ok(const struct send_ctx *sctx, int cmd)
+{
+	switch (sctx->proto) {
+	case 1:	 return cmd <= BTRFS_SEND_C_MAX_V1;
+	case 2:	 return cmd <= BTRFS_SEND_C_MAX_V2;
+	case 3:	 return cmd <= BTRFS_SEND_C_MAX_V3;
+	default: return false;
+	}
+}
+
 static int is_waiting_for_move(struct send_ctx *sctx, u64 ino);
 
 static struct waiting_dir_move *
 get_waiting_dir_move(struct send_ctx *sctx, u64 ino);
 
-static int is_waiting_for_rm(struct send_ctx *sctx, u64 dir_ino);
+static int is_waiting_for_rm(struct send_ctx *sctx, u64 dir_ino, u64 gen);
 
 static int need_send_hole(struct send_ctx *sctx)
 {
@@ -511,17 +547,12 @@ out:
 
 static int fs_path_copy(struct fs_path *p, struct fs_path *from)
 {
-	int ret;
-
 	p->reversed = from->reversed;
 	fs_path_reset(p);
 
-	ret = fs_path_add_path(p, from);
-
-	return ret;
+	return fs_path_add_path(p, from);
 }
 
-
 static void fs_path_unreverse(struct fs_path *p)
 {
 	char *tmp;
@@ -558,15 +589,10 @@ static int write_buf(struct file *filp, const void *buf, u32 len, loff_t *off)
 
 	while (pos < len) {
 		ret = kernel_write(filp, buf + pos, len - pos, off);
-		/* TODO handle that correctly */
-		/*if (ret == -ERESTARTSYS) {
-			continue;
-		}*/
 		if (ret < 0)
 			return ret;
-		if (ret == 0) {
+		if (ret == 0)
 			return -EIO;
-		}
 		pos += ret;
 	}
 
@@ -579,12 +605,15 @@ static int tlv_put(struct send_ctx *sctx, u16 attr, const void *data, int len)
 	int total_len = sizeof(*hdr) + len;
 	int left = sctx->send_max_size - sctx->send_size;
 
+	if (WARN_ON_ONCE(sctx->put_data))
+		return -EINVAL;
+
 	if (unlikely(left < total_len))
 		return -EOVERFLOW;
 
 	hdr = (struct btrfs_tlv_header *) (sctx->send_buf + sctx->send_size);
-	hdr->tlv_type = cpu_to_le16(attr);
-	hdr->tlv_len = cpu_to_le16(len);
+	put_unaligned_le16(attr, &hdr->tlv_type);
+	put_unaligned_le16(len, &hdr->tlv_len);
 	memcpy(hdr + 1, data, len);
 	sctx->send_size += total_len;
 
@@ -599,6 +628,8 @@ static int tlv_put(struct send_ctx *sctx, u16 attr, const void *data, int len)
 		return tlv_put(sctx, attr, &__tmp, sizeof(__tmp));	\
 	}
 
+TLV_PUT_DEFINE_INT(8)
+TLV_PUT_DEFINE_INT(32)
 TLV_PUT_DEFINE_INT(64)
 
 static int tlv_put_string(struct send_ctx *sctx, u16 attr,
@@ -674,8 +705,7 @@ static int send_header(struct send_ctx *sctx)
 	struct btrfs_stream_header hdr;
 
 	strcpy(hdr.magic, BTRFS_SEND_STREAM_MAGIC);
-	hdr.version = cpu_to_le32(BTRFS_SEND_STREAM_VERSION);
-
+	hdr.version = cpu_to_le32(sctx->proto);
 	return write_buf(sctx->send_filp, &hdr, sizeof(hdr),
 					&sctx->send_off);
 }
@@ -694,7 +724,7 @@ static int begin_cmd(struct send_ctx *sctx, int cmd)
 
 	sctx->send_size += sizeof(*hdr);
 	hdr = (struct btrfs_cmd_header *)sctx->send_buf;
-	hdr->cmd = cpu_to_le16(cmd);
+	put_unaligned_le16(cmd, &hdr->cmd);
 
 	return 0;
 }
@@ -706,18 +736,17 @@ static int send_cmd(struct send_ctx *sctx)
 	u32 crc;
 
 	hdr = (struct btrfs_cmd_header *)sctx->send_buf;
-	hdr->len = cpu_to_le32(sctx->send_size - sizeof(*hdr));
-	hdr->crc = 0;
+	put_unaligned_le32(sctx->send_size - sizeof(*hdr), &hdr->len);
+	put_unaligned_le32(0, &hdr->crc);
 
 	crc = btrfs_crc32c(0, (unsigned char *)sctx->send_buf, sctx->send_size);
-	hdr->crc = cpu_to_le32(crc);
+	put_unaligned_le32(crc, &hdr->crc);
 
 	ret = write_buf(sctx->send_filp, sctx->send_buf, sctx->send_size,
 					&sctx->send_off);
 
-	sctx->total_send_size += sctx->send_size;
-	sctx->cmd_send_size[le16_to_cpu(hdr->cmd)] += sctx->send_size;
 	sctx->send_size = 0;
+	sctx->put_data = false;
 
 	return ret;
 }
@@ -818,17 +847,32 @@ out:
 	return ret;
 }
 
+struct btrfs_inode_info {
+	u64 size;
+	u64 gen;
+	u64 mode;
+	u64 uid;
+	u64 gid;
+	u64 rdev;
+	u64 fileattr;
+	u64 nlink;
+};
+
 /*
  * Helper function to retrieve some fields from an inode item.
  */
-static int __get_inode_info(struct btrfs_root *root, struct btrfs_path *path,
-			  u64 ino, u64 *size, u64 *gen, u64 *mode, u64 *uid,
-			  u64 *gid, u64 *rdev)
+static int get_inode_info(struct btrfs_root *root, u64 ino,
+			  struct btrfs_inode_info *info)
 {
 	int ret;
+	struct btrfs_path *path;
 	struct btrfs_inode_item *ii;
 	struct btrfs_key key;
 
+	path = alloc_path_for_send();
+	if (!path)
+		return -ENOMEM;
+
 	key.objectid = ino;
 	key.type = BTRFS_INODE_ITEM_KEY;
 	key.offset = 0;
@@ -836,41 +880,43 @@ static int __get_inode_info(struct btrfs_root *root, struct btrfs_path *path,
 	if (ret) {
 		if (ret > 0)
 			ret = -ENOENT;
-		return ret;
+		goto out;
 	}
 
+	if (!info)
+		goto out;
+
 	ii = btrfs_item_ptr(path->nodes[0], path->slots[0],
 			struct btrfs_inode_item);
-	if (size)
-		*size = btrfs_inode_size(path->nodes[0], ii);
-	if (gen)
-		*gen = btrfs_inode_generation(path->nodes[0], ii);
-	if (mode)
-		*mode = btrfs_inode_mode(path->nodes[0], ii);
-	if (uid)
-		*uid = btrfs_inode_uid(path->nodes[0], ii);
-	if (gid)
-		*gid = btrfs_inode_gid(path->nodes[0], ii);
-	if (rdev)
-		*rdev = btrfs_inode_rdev(path->nodes[0], ii);
+	info->size = btrfs_inode_size(path->nodes[0], ii);
+	info->gen = btrfs_inode_generation(path->nodes[0], ii);
+	info->mode = btrfs_inode_mode(path->nodes[0], ii);
+	info->uid = btrfs_inode_uid(path->nodes[0], ii);
+	info->gid = btrfs_inode_gid(path->nodes[0], ii);
+	info->rdev = btrfs_inode_rdev(path->nodes[0], ii);
+	info->nlink = btrfs_inode_nlink(path->nodes[0], ii);
+	/*
+	 * Transfer the unchanged u64 value of btrfs_inode_item::flags, that's
+	 * otherwise logically split to 32/32 parts.
+	 */
+	info->fileattr = btrfs_inode_flags(path->nodes[0], ii);
 
+out:
+	btrfs_free_path(path);
 	return ret;
 }
 
-static int get_inode_info(struct btrfs_root *root,
-			  u64 ino, u64 *size, u64 *gen,
-			  u64 *mode, u64 *uid, u64 *gid,
-			  u64 *rdev)
+static int get_inode_gen(struct btrfs_root *root, u64 ino, u64 *gen)
 {
-	struct btrfs_path *path;
 	int ret;
+	struct btrfs_inode_info info;
 
-	path = alloc_path_for_send();
-	if (!path)
-		return -ENOMEM;
-	ret = __get_inode_info(root, path, ino, size, gen, mode, uid, gid,
-			       rdev);
-	btrfs_free_path(path);
+	if (!gen)
+		return -EPERM;
+
+	ret = get_inode_info(root, ino, &info);
+	if (!ret)
+		*gen = info.gen;
 	return ret;
 }
 
@@ -891,7 +937,6 @@ static int iterate_inode_ref(struct btrfs_root *root, struct btrfs_path *path,
 			     iterate_inode_ref_t iterate, void *ctx)
 {
 	struct extent_buffer *eb = path->nodes[0];
-	struct btrfs_item *item;
 	struct btrfs_inode_ref *iref;
 	struct btrfs_inode_extref *extref;
 	struct btrfs_path *tmp_path;
@@ -923,12 +968,11 @@ static int iterate_inode_ref(struct btrfs_root *root, struct btrfs_path *path,
 	if (found_key->type == BTRFS_INODE_REF_KEY) {
 		ptr = (unsigned long)btrfs_item_ptr(eb, slot,
 						    struct btrfs_inode_ref);
-		item = btrfs_item_nr(slot);
-		total = btrfs_item_size(eb, item);
+		total = btrfs_item_size(eb, slot);
 		elem_size = sizeof(*iref);
 	} else {
 		ptr = btrfs_item_ptr_offset(eb, slot);
-		total = btrfs_item_size_nr(eb, slot);
+		total = btrfs_item_size(eb, slot);
 		elem_size = sizeof(*extref);
 	}
 
@@ -997,7 +1041,7 @@ out:
 typedef int (*iterate_dir_item_t)(int num, struct btrfs_key *di_key,
 				  const char *name, int name_len,
 				  const char *data, int data_len,
-				  u8 type, void *ctx);
+				  void *ctx);
 
 /*
  * Helper function to iterate the entries in ONE btrfs_dir_item.
@@ -1011,7 +1055,6 @@ static int iterate_dir_item(struct btrfs_root *root, struct btrfs_path *path,
 {
 	int ret = 0;
 	struct extent_buffer *eb;
-	struct btrfs_item *item;
 	struct btrfs_dir_item *di;
 	struct btrfs_key di_key;
 	char *buf = NULL;
@@ -1023,7 +1066,6 @@ static int iterate_dir_item(struct btrfs_root *root, struct btrfs_path *path,
 	u32 total;
 	int slot;
 	int num;
-	u8 type;
 
 	/*
 	 * Start with a small buffer (1 page). If later we end up needing more
@@ -1040,20 +1082,18 @@ static int iterate_dir_item(struct btrfs_root *root, struct btrfs_path *path,
 
 	eb = path->nodes[0];
 	slot = path->slots[0];
-	item = btrfs_item_nr(slot);
 	di = btrfs_item_ptr(eb, slot, struct btrfs_dir_item);
 	cur = 0;
 	len = 0;
-	total = btrfs_item_size(eb, item);
+	total = btrfs_item_size(eb, slot);
 
 	num = 0;
 	while (cur < total) {
 		name_len = btrfs_dir_name_len(eb, di);
 		data_len = btrfs_dir_data_len(eb, di);
-		type = btrfs_dir_type(eb, di);
 		btrfs_dir_item_key_to_cpu(eb, di, &di_key);
 
-		if (type == BTRFS_FT_XATTR) {
+		if (btrfs_dir_type(eb, di) == BTRFS_FT_XATTR) {
 			if (name_len > XATTR_NAME_MAX) {
 				ret = -ENAMETOOLONG;
 				goto out;
@@ -1103,7 +1143,7 @@ static int iterate_dir_item(struct btrfs_root *root, struct btrfs_path *path,
 		cur += len;
 
 		ret = iterate(num, &di_key, buf, name_len, buf + name_len,
-				data_len, type, ctx);
+			      data_len, ctx);
 		if (ret < 0)
 			goto out;
 		if (ret) {
@@ -1196,9 +1236,6 @@ struct backref_ctx {
 	/* may be truncated in case it's the last extent in a file */
 	u64 extent_len;
 
-	/* data offset in the file extent item */
-	u64 data_offset;
-
 	/* Just to check for bugs in backref resolving */
 	int found_itself;
 };
@@ -1206,7 +1243,7 @@ struct backref_ctx {
 static int __clone_root_cmp_bsearch(const void *key, const void *elt)
 {
 	u64 root = (u64)(uintptr_t)key;
-	struct clone_root *cr = (struct clone_root *)elt;
+	const struct clone_root *cr = elt;
 
 	if (root < cr->root->root_key.objectid)
 		return -1;
@@ -1217,8 +1254,8 @@ static int __clone_root_cmp_bsearch(const void *key, const void *elt)
 
 static int __clone_root_cmp_sort(const void *e1, const void *e2)
 {
-	struct clone_root *cr1 = (struct clone_root *)e1;
-	struct clone_root *cr2 = (struct clone_root *)e2;
+	const struct clone_root *cr1 = e1;
+	const struct clone_root *cr2 = e2;
 
 	if (cr1->root->root_key.objectid < cr2->root->root_key.objectid)
 		return -1;
@@ -1315,7 +1352,7 @@ static int find_extent_clone(struct send_ctx *sctx,
 	u64 flags = 0;
 	struct btrfs_file_extent_item *fi;
 	struct extent_buffer *eb = path->nodes[0];
-	struct backref_ctx *backref_ctx = NULL;
+	struct backref_ctx backref_ctx = {0};
 	struct clone_root *cur_clone_root;
 	struct btrfs_key found_key;
 	struct btrfs_path *tmp_path;
@@ -1330,12 +1367,6 @@ static int find_extent_clone(struct send_ctx *sctx,
 	/* We only use this path under the commit sem */
 	tmp_path->need_commit_sem = 0;
 
-	backref_ctx = kmalloc(sizeof(*backref_ctx), GFP_KERNEL);
-	if (!backref_ctx) {
-		ret = -ENOMEM;
-		goto out;
-	}
-
 	if (data_offset >= ino_size) {
 		/*
 		 * There may be extents that lie behind the file's size.
@@ -1400,25 +1431,12 @@ static int find_extent_clone(struct send_ctx *sctx,
 		cur_clone_root->found_refs = 0;
 	}
 
-	backref_ctx->sctx = sctx;
-	backref_ctx->found = 0;
-	backref_ctx->cur_objectid = ino;
-	backref_ctx->cur_offset = data_offset;
-	backref_ctx->found_itself = 0;
-	backref_ctx->extent_len = num_bytes;
-	/*
-	 * For non-compressed extents iterate_extent_inodes() gives us extent
-	 * offsets that already take into account the data offset, but not for
-	 * compressed extents, since the offset is logical and not relative to
-	 * the physical extent locations. We must take this into account to
-	 * avoid sending clone offsets that go beyond the source file's size,
-	 * which would result in the clone ioctl failing with -EINVAL on the
-	 * receiving end.
-	 */
-	if (compressed == BTRFS_COMPRESS_NONE)
-		backref_ctx->data_offset = 0;
-	else
-		backref_ctx->data_offset = btrfs_file_extent_offset(eb, fi);
+	backref_ctx.sctx = sctx;
+	backref_ctx.found = 0;
+	backref_ctx.cur_objectid = ino;
+	backref_ctx.cur_offset = data_offset;
+	backref_ctx.found_itself = 0;
+	backref_ctx.extent_len = num_bytes;
 
 	/*
 	 * The last extent of a file may be too large due to page alignment.
@@ -1426,7 +1444,7 @@ static int find_extent_clone(struct send_ctx *sctx,
 	 * __iterate_backrefs work.
 	 */
 	if (data_offset + num_bytes >= ino_size)
-		backref_ctx->extent_len = ino_size - data_offset;
+		backref_ctx.extent_len = ino_size - data_offset;
 
 	/*
 	 * Now collect all backrefs.
@@ -1437,12 +1455,32 @@ static int find_extent_clone(struct send_ctx *sctx,
 		extent_item_pos = 0;
 	ret = iterate_extent_inodes(fs_info, found_key.objectid,
 				    extent_item_pos, 1, __iterate_backrefs,
-				    backref_ctx, false);
+				    &backref_ctx, false);
 
 	if (ret < 0)
 		goto out;
 
-	if (!backref_ctx->found_itself) {
+	down_read(&fs_info->commit_root_sem);
+	if (fs_info->last_reloc_trans > sctx->last_reloc_trans) {
+		/*
+		 * A transaction commit for a transaction in which block group
+		 * relocation was done just happened.
+		 * The disk_bytenr of the file extent item we processed is
+		 * possibly stale, referring to the extent's location before
+		 * relocation. So act as if we haven't found any clone sources
+		 * and fallback to write commands, which will read the correct
+		 * data from the new extent location. Otherwise we will fail
+		 * below because we haven't found our own back reference or we
+		 * could be getting incorrect sources in case the old extent
+		 * was already reallocated after the relocation.
+		 */
+		up_read(&fs_info->commit_root_sem);
+		ret = -ENOENT;
+		goto out;
+	}
+	up_read(&fs_info->commit_root_sem);
+
+	if (!backref_ctx.found_itself) {
 		/* found a bug in backref code? */
 		ret = -EIO;
 		btrfs_err(fs_info,
@@ -1455,7 +1493,7 @@ static int find_extent_clone(struct send_ctx *sctx,
 		    "find_extent_clone: data_offset=%llu, ino=%llu, num_bytes=%llu, logical=%llu",
 		    data_offset, ino, num_bytes, logical);
 
-	if (!backref_ctx->found)
+	if (!backref_ctx.found)
 		btrfs_debug(fs_info, "no clones found");
 
 	cur_clone_root = NULL;
@@ -1479,7 +1517,6 @@ static int find_extent_clone(struct send_ctx *sctx,
 
 out:
 	btrfs_free_path(tmp_path);
-	kfree(backref_ctx);
 	return ret;
 }
 
@@ -1622,21 +1659,22 @@ static int get_cur_inode_state(struct send_ctx *sctx, u64 ino, u64 gen)
 	int right_ret;
 	u64 left_gen;
 	u64 right_gen;
+	struct btrfs_inode_info info;
 
-	ret = get_inode_info(sctx->send_root, ino, NULL, &left_gen, NULL, NULL,
-			NULL, NULL);
+	ret = get_inode_info(sctx->send_root, ino, &info);
 	if (ret < 0 && ret != -ENOENT)
 		goto out;
-	left_ret = ret;
+	left_ret = (info.nlink == 0) ? -ENOENT : ret;
+	left_gen = info.gen;
 
 	if (!sctx->parent_root) {
 		right_ret = -ENOENT;
 	} else {
-		ret = get_inode_info(sctx->parent_root, ino, NULL, &right_gen,
-				NULL, NULL, NULL, NULL);
+		ret = get_inode_info(sctx->parent_root, ino, &info);
 		if (ret < 0 && ret != -ENOENT)
 			goto out;
-		right_ret = ret;
+		right_ret = (info.nlink == 0) ? -ENOENT : ret;
+		right_gen = info.gen;
 	}
 
 	if (!left_ret && !right_ret) {
@@ -1708,8 +1746,7 @@ out:
  */
 static int lookup_dir_item_inode(struct btrfs_root *root,
 				 u64 dir, const char *name, int name_len,
-				 u64 *found_inode,
-				 u8 *found_type)
+				 u64 *found_inode)
 {
 	int ret = 0;
 	struct btrfs_dir_item *di;
@@ -1732,7 +1769,6 @@ static int lookup_dir_item_inode(struct btrfs_root *root,
 		goto out;
 	}
 	*found_inode = key.objectid;
-	*found_type = btrfs_dir_type(path->nodes[0], di);
 
 out:
 	btrfs_free_path(path);
@@ -1797,8 +1833,7 @@ static int get_first_ref(struct btrfs_root *root, u64 ino,
 	btrfs_release_path(path);
 
 	if (dir_gen) {
-		ret = get_inode_info(root, parent_dir, NULL, dir_gen, NULL,
-				     NULL, NULL, NULL);
+		ret = get_inode_gen(root, parent_dir, dir_gen);
 		if (ret < 0)
 			goto out;
 	}
@@ -1855,7 +1890,7 @@ static int will_overwrite_ref(struct send_ctx *sctx, u64 dir, u64 dir_gen,
 	int ret = 0;
 	u64 gen;
 	u64 other_inode = 0;
-	u8 other_type = 0;
+	struct btrfs_inode_info info;
 
 	if (!sctx->parent_root)
 		goto out;
@@ -1870,8 +1905,7 @@ static int will_overwrite_ref(struct send_ctx *sctx, u64 dir, u64 dir_gen,
 	 * and we can just unlink this entry.
 	 */
 	if (sctx->parent_root && dir != BTRFS_FIRST_FREE_OBJECTID) {
-		ret = get_inode_info(sctx->parent_root, dir, NULL, &gen, NULL,
-				     NULL, NULL, NULL);
+		ret = get_inode_gen(sctx->parent_root, dir, &gen);
 		if (ret < 0 && ret != -ENOENT)
 			goto out;
 		if (ret) {
@@ -1883,7 +1917,7 @@ static int will_overwrite_ref(struct send_ctx *sctx, u64 dir, u64 dir_gen,
 	}
 
 	ret = lookup_dir_item_inode(sctx->parent_root, dir, name, name_len,
-			&other_inode, &other_type);
+				    &other_inode);
 	if (ret < 0 && ret != -ENOENT)
 		goto out;
 	if (ret) {
@@ -1898,13 +1932,14 @@ static int will_overwrite_ref(struct send_ctx *sctx, u64 dir, u64 dir_gen,
 	 */
 	if (other_inode > sctx->send_progress ||
 	    is_waiting_for_move(sctx, other_inode)) {
-		ret = get_inode_info(sctx->parent_root, other_inode, NULL,
-				who_gen, who_mode, NULL, NULL, NULL);
+		ret = get_inode_info(sctx->parent_root, other_inode, &info);
 		if (ret < 0)
 			goto out;
 
 		ret = 1;
 		*who_ino = other_inode;
+		*who_gen = info.gen;
+		*who_mode = info.mode;
 	} else {
 		ret = 0;
 	}
@@ -1928,7 +1963,6 @@ static int did_overwrite_ref(struct send_ctx *sctx,
 	int ret = 0;
 	u64 gen;
 	u64 ow_inode;
-	u8 other_type;
 
 	if (!sctx->parent_root)
 		goto out;
@@ -1938,8 +1972,7 @@ static int did_overwrite_ref(struct send_ctx *sctx,
 		goto out;
 
 	if (dir != BTRFS_FIRST_FREE_OBJECTID) {
-		ret = get_inode_info(sctx->send_root, dir, NULL, &gen, NULL,
-				     NULL, NULL, NULL);
+		ret = get_inode_gen(sctx->send_root, dir, &gen);
 		if (ret < 0 && ret != -ENOENT)
 			goto out;
 		if (ret) {
@@ -1952,7 +1985,7 @@ static int did_overwrite_ref(struct send_ctx *sctx,
 
 	/* check if the ref was overwritten by another ref */
 	ret = lookup_dir_item_inode(sctx->send_root, dir, name, name_len,
-			&ow_inode, &other_type);
+				    &ow_inode);
 	if (ret < 0 && ret != -ENOENT)
 		goto out;
 	if (ret) {
@@ -1961,8 +1994,7 @@ static int did_overwrite_ref(struct send_ctx *sctx,
 		goto out;
 	}
 
-	ret = get_inode_info(sctx->send_root, ow_inode, NULL, &gen, NULL, NULL,
-			NULL, NULL);
+	ret = get_inode_gen(sctx->send_root, ow_inode, &gen);
 	if (ret < 0)
 		goto out;
 
@@ -2099,16 +2131,6 @@ static struct name_cache_entry *name_cache_search(struct send_ctx *sctx,
 }
 
 /*
- * Removes the entry from the list and adds it back to the end. This marks the
- * entry as recently used so that name_cache_clean_unused does not remove it.
- */
-static void name_cache_used(struct send_ctx *sctx, struct name_cache_entry *nce)
-{
-	list_del(&nce->list);
-	list_add_tail(&nce->list, &sctx->name_cache_list);
-}
-
-/*
  * Remove some entries from the beginning of name_cache_list.
  */
 static void name_cache_clean_unused(struct send_ctx *sctx)
@@ -2168,7 +2190,13 @@ static int __get_cur_name_and_parent(struct send_ctx *sctx,
 			kfree(nce);
 			nce = NULL;
 		} else {
-			name_cache_used(sctx, nce);
+			/*
+			 * Removes the entry from the list and adds it back to
+			 * the end.  This marks the entry as recently used so
+			 * that name_cache_clean_unused does not remove it.
+			 */
+			list_move_tail(&nce->list, &sctx->name_cache_list);
+
 			*parent_ino = nce->parent_ino;
 			*parent_gen = nce->parent_gen;
 			ret = fs_path_add(dest, nce->name, nce->name_len);
@@ -2182,7 +2210,7 @@ static int __get_cur_name_and_parent(struct send_ctx *sctx,
 	/*
 	 * If the inode is not existent yet, add the orphan name and return 1.
 	 * This should only happen for the parent dir that we determine in
-	 * __record_new_ref
+	 * record_new_ref_if_needed().
 	 */
 	ret = is_inode_existent(sctx, ino, gen);
 	if (ret < 0)
@@ -2305,7 +2333,7 @@ static int get_cur_path(struct send_ctx *sctx, u64 ino, u64 gen,
 
 		fs_path_reset(name);
 
-		if (is_waiting_for_rm(sctx, ino)) {
+		if (is_waiting_for_rm(sctx, ino, gen)) {
 			ret = gen_unique_name(sctx, ino, gen, name);
 			if (ret < 0)
 				goto out;
@@ -2416,7 +2444,7 @@ static int send_subvol_begin(struct send_ctx *sctx)
 			    sctx->send_root->root_item.uuid);
 
 	TLV_PUT_U64(sctx, BTRFS_SEND_A_CTRANSID,
-		    le64_to_cpu(sctx->send_root->root_item.ctransid));
+		    btrfs_root_ctransid(&sctx->send_root->root_item));
 	if (parent_root) {
 		if (!btrfs_is_empty_uuid(parent_root->root_item.received_uuid))
 			TLV_PUT_UUID(sctx, BTRFS_SEND_A_CLONE_UUID,
@@ -2425,7 +2453,7 @@ static int send_subvol_begin(struct send_ctx *sctx)
 			TLV_PUT_UUID(sctx, BTRFS_SEND_A_CLONE_UUID,
 				     parent_root->root_item.uuid);
 		TLV_PUT_U64(sctx, BTRFS_SEND_A_CLONE_CTRANSID,
-			    le64_to_cpu(sctx->parent_root->root_item.ctransid));
+			    btrfs_root_ctransid(&sctx->parent_root->root_item));
 	}
 
 	ret = send_cmd(sctx);
@@ -2497,6 +2525,39 @@ out:
 	return ret;
 }
 
+static int send_fileattr(struct send_ctx *sctx, u64 ino, u64 gen, u64 fileattr)
+{
+	struct btrfs_fs_info *fs_info = sctx->send_root->fs_info;
+	int ret = 0;
+	struct fs_path *p;
+
+	if (sctx->proto < 2)
+		return 0;
+
+	btrfs_debug(fs_info, "send_fileattr %llu fileattr=%llu", ino, fileattr);
+
+	p = fs_path_alloc();
+	if (!p)
+		return -ENOMEM;
+
+	ret = begin_cmd(sctx, BTRFS_SEND_C_FILEATTR);
+	if (ret < 0)
+		goto out;
+
+	ret = get_cur_path(sctx, ino, gen, p);
+	if (ret < 0)
+		goto out;
+	TLV_PUT_PATH(sctx, BTRFS_SEND_A_PATH, p);
+	TLV_PUT_U64(sctx, BTRFS_SEND_A_FILEATTR, fileattr);
+
+	ret = send_cmd(sctx);
+
+tlv_put_failure:
+out:
+	fs_path_free(p);
+	return ret;
+}
+
 static int send_chown(struct send_ctx *sctx, u64 ino, u64 gen, u64 uid, u64 gid)
 {
 	struct btrfs_fs_info *fs_info = sctx->send_root->fs_info;
@@ -2576,7 +2637,8 @@ static int send_utimes(struct send_ctx *sctx, u64 ino, u64 gen)
 	TLV_PUT_BTRFS_TIMESPEC(sctx, BTRFS_SEND_A_ATIME, eb, &ii->atime);
 	TLV_PUT_BTRFS_TIMESPEC(sctx, BTRFS_SEND_A_MTIME, eb, &ii->mtime);
 	TLV_PUT_BTRFS_TIMESPEC(sctx, BTRFS_SEND_A_CTIME, eb, &ii->ctime);
-	/* TODO Add otime support when the otime patches get into upstream */
+	if (sctx->proto >= 2)
+		TLV_PUT_BTRFS_TIMESPEC(sctx, BTRFS_SEND_A_OTIME, eb, &ii->otime);
 
 	ret = send_cmd(sctx);
 
@@ -2598,6 +2660,7 @@ static int send_create_inode(struct send_ctx *sctx, u64 ino)
 	int ret = 0;
 	struct fs_path *p;
 	int cmd;
+	struct btrfs_inode_info info;
 	u64 gen;
 	u64 mode;
 	u64 rdev;
@@ -2609,10 +2672,12 @@ static int send_create_inode(struct send_ctx *sctx, u64 ino)
 		return -ENOMEM;
 
 	if (ino != sctx->cur_ino) {
-		ret = get_inode_info(sctx->send_root, ino, NULL, &gen, &mode,
-				     NULL, NULL, &rdev);
+		ret = get_inode_info(sctx->send_root, ino, &info);
 		if (ret < 0)
 			goto out;
+		gen = info.gen;
+		mode = info.mode;
+		rdev = info.rdev;
 	} else {
 		gen = sctx->cur_inode_gen;
 		mode = sctx->cur_inode_mode;
@@ -2680,61 +2745,43 @@ out:
 static int did_create_dir(struct send_ctx *sctx, u64 dir)
 {
 	int ret = 0;
+	int iter_ret = 0;
 	struct btrfs_path *path = NULL;
 	struct btrfs_key key;
 	struct btrfs_key found_key;
 	struct btrfs_key di_key;
-	struct extent_buffer *eb;
 	struct btrfs_dir_item *di;
-	int slot;
 
 	path = alloc_path_for_send();
-	if (!path) {
-		ret = -ENOMEM;
-		goto out;
-	}
+	if (!path)
+		return -ENOMEM;
 
 	key.objectid = dir;
 	key.type = BTRFS_DIR_INDEX_KEY;
 	key.offset = 0;
-	ret = btrfs_search_slot(NULL, sctx->send_root, &key, path, 0, 0);
-	if (ret < 0)
-		goto out;
 
-	while (1) {
-		eb = path->nodes[0];
-		slot = path->slots[0];
-		if (slot >= btrfs_header_nritems(eb)) {
-			ret = btrfs_next_leaf(sctx->send_root, path);
-			if (ret < 0) {
-				goto out;
-			} else if (ret > 0) {
-				ret = 0;
-				break;
-			}
-			continue;
-		}
+	btrfs_for_each_slot(sctx->send_root, &key, &found_key, path, iter_ret) {
+		struct extent_buffer *eb = path->nodes[0];
 
-		btrfs_item_key_to_cpu(eb, &found_key, slot);
 		if (found_key.objectid != key.objectid ||
 		    found_key.type != key.type) {
 			ret = 0;
-			goto out;
+			break;
 		}
 
-		di = btrfs_item_ptr(eb, slot, struct btrfs_dir_item);
+		di = btrfs_item_ptr(eb, path->slots[0], struct btrfs_dir_item);
 		btrfs_dir_item_key_to_cpu(eb, di, &di_key);
 
 		if (di_key.type != BTRFS_ROOT_ITEM_KEY &&
 		    di_key.objectid < sctx->send_progress) {
 			ret = 1;
-			goto out;
+			break;
 		}
-
-		path->slots[0]++;
 	}
+	/* Catch error found during iteration */
+	if (iter_ret < 0)
+		ret = iter_ret;
 
-out:
 	btrfs_free_path(path);
 	return ret;
 }
@@ -2752,19 +2799,12 @@ static int send_create_inode_if_needed(struct send_ctx *sctx)
 	if (S_ISDIR(sctx->cur_inode_mode)) {
 		ret = did_create_dir(sctx, sctx->cur_ino);
 		if (ret < 0)
-			goto out;
-		if (ret) {
-			ret = 0;
-			goto out;
-		}
+			return ret;
+		else if (ret > 0)
+			return 0;
 	}
 
-	ret = send_create_inode(sctx, sctx->cur_ino);
-	if (ret < 0)
-		goto out;
-
-out:
-	return ret;
+	return send_create_inode(sctx, sctx->cur_ino);
 }
 
 struct recorded_ref {
@@ -2774,48 +2814,50 @@ struct recorded_ref {
 	u64 dir;
 	u64 dir_gen;
 	int name_len;
+	struct rb_node node;
+	struct rb_root *root;
 };
 
-static void set_ref_path(struct recorded_ref *ref, struct fs_path *path)
+static struct recorded_ref *recorded_ref_alloc(void)
 {
-	ref->full_path = path;
-	ref->name = (char *)kbasename(ref->full_path->start);
-	ref->name_len = ref->full_path->end - ref->name;
+	struct recorded_ref *ref;
+
+	ref = kzalloc(sizeof(*ref), GFP_KERNEL);
+	if (!ref)
+		return NULL;
+	RB_CLEAR_NODE(&ref->node);
+	INIT_LIST_HEAD(&ref->list);
+	return ref;
 }
 
-/*
- * We need to process new refs before deleted refs, but compare_tree gives us
- * everything mixed. So we first record all refs and later process them.
- * This function is a helper to record one ref.
- */
-static int __record_ref(struct list_head *head, u64 dir,
-		      u64 dir_gen, struct fs_path *path)
+static void recorded_ref_free(struct recorded_ref *ref)
 {
-	struct recorded_ref *ref;
-
-	ref = kmalloc(sizeof(*ref), GFP_KERNEL);
 	if (!ref)
-		return -ENOMEM;
+		return;
+	if (!RB_EMPTY_NODE(&ref->node))
+		rb_erase(&ref->node, ref->root);
+	list_del(&ref->list);
+	fs_path_free(ref->full_path);
+	kfree(ref);
+}
 
-	ref->dir = dir;
-	ref->dir_gen = dir_gen;
-	set_ref_path(ref, path);
-	list_add_tail(&ref->list, head);
-	return 0;
+static void set_ref_path(struct recorded_ref *ref, struct fs_path *path)
+{
+	ref->full_path = path;
+	ref->name = (char *)kbasename(ref->full_path->start);
+	ref->name_len = ref->full_path->end - ref->name;
 }
 
 static int dup_ref(struct recorded_ref *ref, struct list_head *list)
 {
 	struct recorded_ref *new;
 
-	new = kmalloc(sizeof(*ref), GFP_KERNEL);
+	new = recorded_ref_alloc();
 	if (!new)
 		return -ENOMEM;
 
 	new->dir = ref->dir;
 	new->dir_gen = ref->dir_gen;
-	new->full_path = NULL;
-	INIT_LIST_HEAD(&new->list);
 	list_add_tail(&new->list, list);
 	return 0;
 }
@@ -2826,9 +2868,7 @@ static void __free_recorded_refs(struct list_head *head)
 
 	while (!list_empty(head)) {
 		cur = list_entry(head->next, struct recorded_ref, list);
-		fs_path_free(cur->full_path);
-		list_del(&cur->list);
-		kfree(cur);
+		recorded_ref_free(cur);
 	}
 }
 
@@ -2864,8 +2904,8 @@ out:
 	return ret;
 }
 
-static struct orphan_dir_info *
-add_orphan_dir_info(struct send_ctx *sctx, u64 dir_ino)
+static struct orphan_dir_info *add_orphan_dir_info(struct send_ctx *sctx,
+						   u64 dir_ino, u64 dir_gen)
 {
 	struct rb_node **p = &sctx->orphan_dirs.rb_node;
 	struct rb_node *parent = NULL;
@@ -2874,20 +2914,23 @@ add_orphan_dir_info(struct send_ctx *sctx, u64 dir_ino)
 	while (*p) {
 		parent = *p;
 		entry = rb_entry(parent, struct orphan_dir_info, node);
-		if (dir_ino < entry->ino) {
+		if (dir_ino < entry->ino)
 			p = &(*p)->rb_left;
-		} else if (dir_ino > entry->ino) {
+		else if (dir_ino > entry->ino)
 			p = &(*p)->rb_right;
-		} else {
+		else if (dir_gen < entry->gen)
+			p = &(*p)->rb_left;
+		else if (dir_gen > entry->gen)
+			p = &(*p)->rb_right;
+		else
 			return entry;
-		}
 	}
 
 	odi = kmalloc(sizeof(*odi), GFP_KERNEL);
 	if (!odi)
 		return ERR_PTR(-ENOMEM);
 	odi->ino = dir_ino;
-	odi->gen = 0;
+	odi->gen = dir_gen;
 	odi->last_dir_index_offset = 0;
 
 	rb_link_node(&odi->node, parent, p);
@@ -2895,8 +2938,8 @@ add_orphan_dir_info(struct send_ctx *sctx, u64 dir_ino)
 	return odi;
 }
 
-static struct orphan_dir_info *
-get_orphan_dir_info(struct send_ctx *sctx, u64 dir_ino)
+static struct orphan_dir_info *get_orphan_dir_info(struct send_ctx *sctx,
+						   u64 dir_ino, u64 gen)
 {
 	struct rb_node *n = sctx->orphan_dirs.rb_node;
 	struct orphan_dir_info *entry;
@@ -2907,15 +2950,19 @@ get_orphan_dir_info(struct send_ctx *sctx, u64 dir_ino)
 			n = n->rb_left;
 		else if (dir_ino > entry->ino)
 			n = n->rb_right;
+		else if (gen < entry->gen)
+			n = n->rb_left;
+		else if (gen > entry->gen)
+			n = n->rb_right;
 		else
 			return entry;
 	}
 	return NULL;
 }
 
-static int is_waiting_for_rm(struct send_ctx *sctx, u64 dir_ino)
+static int is_waiting_for_rm(struct send_ctx *sctx, u64 dir_ino, u64 gen)
 {
-	struct orphan_dir_info *odi = get_orphan_dir_info(sctx, dir_ino);
+	struct orphan_dir_info *odi = get_orphan_dir_info(sctx, dir_ino, gen);
 
 	return odi != NULL;
 }
@@ -2938,6 +2985,7 @@ static int can_rmdir(struct send_ctx *sctx, u64 dir, u64 dir_gen,
 		     u64 send_progress)
 {
 	int ret = 0;
+	int iter_ret = 0;
 	struct btrfs_root *root = sctx->parent_root;
 	struct btrfs_path *path;
 	struct btrfs_key key;
@@ -2960,27 +3008,13 @@ static int can_rmdir(struct send_ctx *sctx, u64 dir, u64 dir_gen,
 	key.type = BTRFS_DIR_INDEX_KEY;
 	key.offset = 0;
 
-	odi = get_orphan_dir_info(sctx, dir);
+	odi = get_orphan_dir_info(sctx, dir, dir_gen);
 	if (odi)
 		key.offset = odi->last_dir_index_offset;
 
-	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
-	if (ret < 0)
-		goto out;
-
-	while (1) {
+	btrfs_for_each_slot(root, &key, &found_key, path, iter_ret) {
 		struct waiting_dir_move *dm;
 
-		if (path->slots[0] >= btrfs_header_nritems(path->nodes[0])) {
-			ret = btrfs_next_leaf(root, path);
-			if (ret < 0)
-				goto out;
-			else if (ret > 0)
-				break;
-			continue;
-		}
-		btrfs_item_key_to_cpu(path->nodes[0], &found_key,
-				      path->slots[0]);
 		if (found_key.objectid != key.objectid ||
 		    found_key.type != key.type)
 			break;
@@ -2991,7 +3025,7 @@ static int can_rmdir(struct send_ctx *sctx, u64 dir, u64 dir_gen,
 
 		dm = get_waiting_dir_move(sctx, loc.objectid);
 		if (dm) {
-			odi = add_orphan_dir_info(sctx, dir);
+			odi = add_orphan_dir_info(sctx, dir, dir_gen);
 			if (IS_ERR(odi)) {
 				ret = PTR_ERR(odi);
 				goto out;
@@ -2999,12 +3033,13 @@ static int can_rmdir(struct send_ctx *sctx, u64 dir, u64 dir_gen,
 			odi->gen = dir_gen;
 			odi->last_dir_index_offset = found_key.offset;
 			dm->rmdir_ino = dir;
+			dm->rmdir_gen = dir_gen;
 			ret = 0;
 			goto out;
 		}
 
 		if (loc.objectid > send_progress) {
-			odi = add_orphan_dir_info(sctx, dir);
+			odi = add_orphan_dir_info(sctx, dir, dir_gen);
 			if (IS_ERR(odi)) {
 				ret = PTR_ERR(odi);
 				goto out;
@@ -3014,8 +3049,10 @@ static int can_rmdir(struct send_ctx *sctx, u64 dir, u64 dir_gen,
 			ret = 0;
 			goto out;
 		}
-
-		path->slots[0]++;
+	}
+	if (iter_ret < 0) {
+		ret = iter_ret;
+		goto out;
 	}
 	free_orphan_dir_info(sctx, odi);
 
@@ -3044,6 +3081,7 @@ static int add_waiting_dir_move(struct send_ctx *sctx, u64 ino, bool orphanized)
 		return -ENOMEM;
 	dm->ino = ino;
 	dm->rmdir_ino = 0;
+	dm->rmdir_gen = 0;
 	dm->orphanized = orphanized;
 
 	while (*p) {
@@ -3189,7 +3227,7 @@ static int path_loop(struct send_ctx *sctx, struct fs_path *name,
 	while (ino != BTRFS_FIRST_FREE_OBJECTID) {
 		fs_path_reset(name);
 
-		if (is_waiting_for_rm(sctx, ino))
+		if (is_waiting_for_rm(sctx, ino, gen))
 			break;
 		if (is_waiting_for_move(sctx, ino)) {
 			if (*ancestor_ino == 0)
@@ -3229,6 +3267,7 @@ static int apply_dir_move(struct send_ctx *sctx, struct pending_dir_move *pm)
 	u64 parent_ino, parent_gen;
 	struct waiting_dir_move *dm = NULL;
 	u64 rmdir_ino = 0;
+	u64 rmdir_gen;
 	u64 ancestor;
 	bool is_orphan;
 	int ret;
@@ -3243,6 +3282,7 @@ static int apply_dir_move(struct send_ctx *sctx, struct pending_dir_move *pm)
 	dm = get_waiting_dir_move(sctx, pm->ino);
 	ASSERT(dm);
 	rmdir_ino = dm->rmdir_ino;
+	rmdir_gen = dm->rmdir_gen;
 	is_orphan = dm->orphanized;
 	free_waiting_dir_move(sctx, dm);
 
@@ -3279,6 +3319,7 @@ static int apply_dir_move(struct send_ctx *sctx, struct pending_dir_move *pm)
 			dm = get_waiting_dir_move(sctx, pm->ino);
 			ASSERT(dm);
 			dm->rmdir_ino = rmdir_ino;
+			dm->rmdir_gen = rmdir_gen;
 		}
 		goto out;
 	}
@@ -3297,7 +3338,7 @@ static int apply_dir_move(struct send_ctx *sctx, struct pending_dir_move *pm)
 		struct orphan_dir_info *odi;
 		u64 gen;
 
-		odi = get_orphan_dir_info(sctx, rmdir_ino);
+		odi = get_orphan_dir_info(sctx, rmdir_ino, rmdir_gen);
 		if (!odi) {
 			/* already deleted */
 			goto finish;
@@ -3336,8 +3377,7 @@ finish:
 		/*
 		 * The parent inode might have been deleted in the send snapshot
 		 */
-		ret = get_inode_info(sctx->send_root, cur->dir, NULL,
-				     NULL, NULL, NULL, NULL, NULL);
+		ret = get_inode_info(sctx->send_root, cur->dir, NULL);
 		if (ret == -ENOENT) {
 			ret = 0;
 			continue;
@@ -3511,12 +3551,10 @@ static int wait_for_dest_dir_move(struct send_ctx *sctx,
 		goto out;
 	}
 
-	ret = get_inode_info(sctx->parent_root, di_key.objectid, NULL,
-			     &left_gen, NULL, NULL, NULL, NULL);
+	ret = get_inode_gen(sctx->parent_root, di_key.objectid, &left_gen);
 	if (ret < 0)
 		goto out;
-	ret = get_inode_info(sctx->send_root, di_key.objectid, NULL,
-			     &right_gen, NULL, NULL, NULL, NULL);
+	ret = get_inode_gen(sctx->send_root, di_key.objectid, &right_gen);
 	if (ret < 0) {
 		if (ret == -ENOENT)
 			ret = 0;
@@ -3579,7 +3617,7 @@ static int check_ino_in_path(struct btrfs_root *root,
 }
 
 /*
- * Check if ino ino1 is an ancestor of inode ino2 in the given root for any
+ * Check if inode ino1 is an ancestor of inode ino2 in the given root for any
  * possible path (in case ino2 is not a directory and has multiple hard links).
  * Return 1 if true, 0 if false and < 0 on error.
  */
@@ -3591,6 +3629,7 @@ static int is_ancestor(struct btrfs_root *root,
 {
 	bool free_fs_path = false;
 	int ret = 0;
+	int iter_ret = 0;
 	struct btrfs_path *path = NULL;
 	struct btrfs_key key;
 
@@ -3611,33 +3650,19 @@ static int is_ancestor(struct btrfs_root *root,
 	key.type = BTRFS_INODE_REF_KEY;
 	key.offset = 0;
 
-	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
-	if (ret < 0)
-		goto out;
-
-	while (true) {
+	btrfs_for_each_slot(root, &key, &key, path, iter_ret) {
 		struct extent_buffer *leaf = path->nodes[0];
 		int slot = path->slots[0];
 		u32 cur_offset = 0;
 		u32 item_size;
 
-		if (slot >= btrfs_header_nritems(leaf)) {
-			ret = btrfs_next_leaf(root, path);
-			if (ret < 0)
-				goto out;
-			if (ret > 0)
-				break;
-			continue;
-		}
-
-		btrfs_item_key_to_cpu(leaf, &key, slot);
 		if (key.objectid != ino2)
 			break;
 		if (key.type != BTRFS_INODE_REF_KEY &&
 		    key.type != BTRFS_INODE_EXTREF_KEY)
 			break;
 
-		item_size = btrfs_item_size_nr(leaf, slot);
+		item_size = btrfs_item_size(leaf, slot);
 		while (cur_offset < item_size) {
 			u64 parent;
 			u64 parent_gen;
@@ -3659,8 +3684,7 @@ static int is_ancestor(struct btrfs_root *root,
 				cur_offset = item_size;
 			}
 
-			ret = get_inode_info(root, parent, NULL, &parent_gen,
-					     NULL, NULL, NULL, NULL);
+			ret = get_inode_gen(root, parent, &parent_gen);
 			if (ret < 0)
 				goto out;
 			ret = check_ino_in_path(root, ino1, ino1_gen,
@@ -3668,10 +3692,12 @@ static int is_ancestor(struct btrfs_root *root,
 			if (ret)
 				goto out;
 		}
-		path->slots[0]++;
 	}
 	ret = 0;
- out:
+	if (iter_ret < 0)
+		ret = iter_ret;
+
+out:
 	btrfs_free_path(path);
 	if (free_fs_path)
 		fs_path_free(fs_path);
@@ -3748,9 +3774,7 @@ static int wait_for_parent_move(struct send_ctx *sctx,
 		     memcmp(path_before->start, path_after->start, len1))) {
 			u64 parent_ino_gen;
 
-			ret = get_inode_info(sctx->parent_root, ino, NULL,
-					     &parent_ino_gen, NULL, NULL, NULL,
-					     NULL);
+			ret = get_inode_gen(sctx->parent_root, ino, &parent_ino_gen);
 			if (ret < 0)
 				goto out;
 			if (ino_gen == parent_ino_gen) {
@@ -3812,6 +3836,72 @@ static int update_ref_path(struct send_ctx *sctx, struct recorded_ref *ref)
 }
 
 /*
+ * When processing the new references for an inode we may orphanize an existing
+ * directory inode because its old name conflicts with one of the new references
+ * of the current inode. Later, when processing another new reference of our
+ * inode, we might need to orphanize another inode, but the path we have in the
+ * reference reflects the pre-orphanization name of the directory we previously
+ * orphanized. For example:
+ *
+ * parent snapshot looks like:
+ *
+ * .                                     (ino 256)
+ * |----- f1                             (ino 257)
+ * |----- f2                             (ino 258)
+ * |----- d1/                            (ino 259)
+ *        |----- d2/                     (ino 260)
+ *
+ * send snapshot looks like:
+ *
+ * .                                     (ino 256)
+ * |----- d1                             (ino 258)
+ * |----- f2/                            (ino 259)
+ *        |----- f2_link/                (ino 260)
+ *        |       |----- f1              (ino 257)
+ *        |
+ *        |----- d2                      (ino 258)
+ *
+ * When processing inode 257 we compute the name for inode 259 as "d1", and we
+ * cache it in the name cache. Later when we start processing inode 258, when
+ * collecting all its new references we set a full path of "d1/d2" for its new
+ * reference with name "d2". When we start processing the new references we
+ * start by processing the new reference with name "d1", and this results in
+ * orphanizing inode 259, since its old reference causes a conflict. Then we
+ * move on the next new reference, with name "d2", and we find out we must
+ * orphanize inode 260, as its old reference conflicts with ours - but for the
+ * orphanization we use a source path corresponding to the path we stored in the
+ * new reference, which is "d1/d2" and not "o259-6-0/d2" - this makes the
+ * receiver fail since the path component "d1/" no longer exists, it was renamed
+ * to "o259-6-0/" when processing the previous new reference. So in this case we
+ * must recompute the path in the new reference and use it for the new
+ * orphanization operation.
+ */
+static int refresh_ref_path(struct send_ctx *sctx, struct recorded_ref *ref)
+{
+	char *name;
+	int ret;
+
+	name = kmemdup(ref->name, ref->name_len, GFP_KERNEL);
+	if (!name)
+		return -ENOMEM;
+
+	fs_path_reset(ref->full_path);
+	ret = get_cur_path(sctx, ref->dir, ref->dir_gen, ref->full_path);
+	if (ret < 0)
+		goto out;
+
+	ret = fs_path_add(ref->full_path, name, ref->name_len);
+	if (ret < 0)
+		goto out;
+
+	/* Update the reference's base name pointer. */
+	set_ref_path(ref, ref->full_path);
+out:
+	kfree(name);
+	return ret;
+}
+
+/*
  * This does all the move/link/unlink/rmdir magic.
  */
 static int process_recorded_refs(struct send_ctx *sctx, int *pending_move)
@@ -3879,52 +3969,56 @@ static int process_recorded_refs(struct send_ctx *sctx, int *pending_move)
 			goto out;
 	}
 
+	/*
+	 * Before doing any rename and link operations, do a first pass on the
+	 * new references to orphanize any unprocessed inodes that may have a
+	 * reference that conflicts with one of the new references of the current
+	 * inode. This needs to happen first because a new reference may conflict
+	 * with the old reference of a parent directory, so we must make sure
+	 * that the path used for link and rename commands don't use an
+	 * orphanized name when an ancestor was not yet orphanized.
+	 *
+	 * Example:
+	 *
+	 * Parent snapshot:
+	 *
+	 * .                                                      (ino 256)
+	 * |----- testdir/                                        (ino 259)
+	 * |          |----- a                                    (ino 257)
+	 * |
+	 * |----- b                                               (ino 258)
+	 *
+	 * Send snapshot:
+	 *
+	 * .                                                      (ino 256)
+	 * |----- testdir_2/                                      (ino 259)
+	 * |          |----- a                                    (ino 260)
+	 * |
+	 * |----- testdir                                         (ino 257)
+	 * |----- b                                               (ino 257)
+	 * |----- b2                                              (ino 258)
+	 *
+	 * Processing the new reference for inode 257 with name "b" may happen
+	 * before processing the new reference with name "testdir". If so, we
+	 * must make sure that by the time we send a link command to create the
+	 * hard link "b", inode 259 was already orphanized, since the generated
+	 * path in "valid_path" already contains the orphanized name for 259.
+	 * We are processing inode 257, so only later when processing 259 we do
+	 * the rename operation to change its temporary (orphanized) name to
+	 * "testdir_2".
+	 */
 	list_for_each_entry(cur, &sctx->new_refs, list) {
-		/*
-		 * We may have refs where the parent directory does not exist
-		 * yet. This happens if the parent directories inum is higher
-		 * than the current inum. To handle this case, we create the
-		 * parent directory out of order. But we need to check if this
-		 * did already happen before due to other refs in the same dir.
-		 */
 		ret = get_cur_inode_state(sctx, cur->dir, cur->dir_gen);
 		if (ret < 0)
 			goto out;
-		if (ret == inode_state_will_create) {
-			ret = 0;
-			/*
-			 * First check if any of the current inodes refs did
-			 * already create the dir.
-			 */
-			list_for_each_entry(cur2, &sctx->new_refs, list) {
-				if (cur == cur2)
-					break;
-				if (cur2->dir == cur->dir) {
-					ret = 1;
-					break;
-				}
-			}
-
-			/*
-			 * If that did not happen, check if a previous inode
-			 * did already create the dir.
-			 */
-			if (!ret)
-				ret = did_create_dir(sctx, cur->dir);
-			if (ret < 0)
-				goto out;
-			if (!ret) {
-				ret = send_create_inode(sctx, cur->dir);
-				if (ret < 0)
-					goto out;
-			}
-		}
+		if (ret == inode_state_will_create)
+			continue;
 
 		/*
-		 * Check if this new ref would overwrite the first ref of
-		 * another unprocessed inode. If yes, orphanize the
-		 * overwritten inode. If we find an overwritten ref that is
-		 * not the first ref, simply unlink it.
+		 * Check if this new ref would overwrite the first ref of another
+		 * unprocessed inode. If yes, orphanize the overwritten inode.
+		 * If we find an overwritten ref that is not the first ref,
+		 * simply unlink it.
 		 */
 		ret = will_overwrite_ref(sctx, cur->dir, cur->dir_gen,
 				cur->name, cur->name_len,
@@ -3941,6 +4035,12 @@ static int process_recorded_refs(struct send_ctx *sctx, int *pending_move)
 				struct name_cache_entry *nce;
 				struct waiting_dir_move *wdm;
 
+				if (orphanized_dir) {
+					ret = refresh_ref_path(sctx, cur);
+					if (ret < 0)
+						goto out;
+				}
+
 				ret = orphanize_inode(sctx, ow_inode, ow_gen,
 						cur->full_path);
 				if (ret < 0)
@@ -3997,12 +4097,66 @@ static int process_recorded_refs(struct send_ctx *sctx, int *pending_move)
 				if (ret < 0)
 					goto out;
 			} else {
+				/*
+				 * If we previously orphanized a directory that
+				 * collided with a new reference that we already
+				 * processed, recompute the current path because
+				 * that directory may be part of the path.
+				 */
+				if (orphanized_dir) {
+					ret = refresh_ref_path(sctx, cur);
+					if (ret < 0)
+						goto out;
+				}
 				ret = send_unlink(sctx, cur->full_path);
 				if (ret < 0)
 					goto out;
 			}
 		}
 
+	}
+
+	list_for_each_entry(cur, &sctx->new_refs, list) {
+		/*
+		 * We may have refs where the parent directory does not exist
+		 * yet. This happens if the parent directories inum is higher
+		 * than the current inum. To handle this case, we create the
+		 * parent directory out of order. But we need to check if this
+		 * did already happen before due to other refs in the same dir.
+		 */
+		ret = get_cur_inode_state(sctx, cur->dir, cur->dir_gen);
+		if (ret < 0)
+			goto out;
+		if (ret == inode_state_will_create) {
+			ret = 0;
+			/*
+			 * First check if any of the current inodes refs did
+			 * already create the dir.
+			 */
+			list_for_each_entry(cur2, &sctx->new_refs, list) {
+				if (cur == cur2)
+					break;
+				if (cur2->dir == cur->dir) {
+					ret = 1;
+					break;
+				}
+			}
+
+			/*
+			 * If that did not happen, check if a previous inode
+			 * did already create the dir.
+			 */
+			if (!ret)
+				ret = did_create_dir(sctx, cur->dir);
+			if (ret < 0)
+				goto out;
+			if (!ret) {
+				ret = send_create_inode(sctx, cur->dir);
+				if (ret < 0)
+					goto out;
+			}
+		}
+
 		if (S_ISDIR(sctx->cur_inode_mode) && sctx->parent_root) {
 			ret = wait_for_dest_dir_move(sctx, cur, is_orphan);
 			if (ret < 0)
@@ -4214,185 +4368,167 @@ out:
 	return ret;
 }
 
-static int record_ref(struct btrfs_root *root, u64 dir, struct fs_path *name,
-		      void *ctx, struct list_head *refs)
+static int rbtree_ref_comp(const void *k, const struct rb_node *node)
+{
+	const struct recorded_ref *data = k;
+	const struct recorded_ref *ref = rb_entry(node, struct recorded_ref, node);
+	int result;
+
+	if (data->dir > ref->dir)
+		return 1;
+	if (data->dir < ref->dir)
+		return -1;
+	if (data->dir_gen > ref->dir_gen)
+		return 1;
+	if (data->dir_gen < ref->dir_gen)
+		return -1;
+	if (data->name_len > ref->name_len)
+		return 1;
+	if (data->name_len < ref->name_len)
+		return -1;
+	result = strcmp(data->name, ref->name);
+	if (result > 0)
+		return 1;
+	if (result < 0)
+		return -1;
+	return 0;
+}
+
+static bool rbtree_ref_less(struct rb_node *node, const struct rb_node *parent)
+{
+	const struct recorded_ref *entry = rb_entry(node, struct recorded_ref, node);
+
+	return rbtree_ref_comp(entry, parent) < 0;
+}
+
+static int record_ref_in_tree(struct rb_root *root, struct list_head *refs,
+			      struct fs_path *name, u64 dir, u64 dir_gen,
+			      struct send_ctx *sctx)
 {
 	int ret = 0;
-	struct send_ctx *sctx = ctx;
-	struct fs_path *p;
-	u64 gen;
+	struct fs_path *path = NULL;
+	struct recorded_ref *ref = NULL;
 
-	p = fs_path_alloc();
-	if (!p)
-		return -ENOMEM;
+	path = fs_path_alloc();
+	if (!path) {
+		ret = -ENOMEM;
+		goto out;
+	}
 
-	ret = get_inode_info(root, dir, NULL, &gen, NULL, NULL,
-			NULL, NULL);
-	if (ret < 0)
+	ref = recorded_ref_alloc();
+	if (!ref) {
+		ret = -ENOMEM;
 		goto out;
+	}
 
-	ret = get_cur_path(sctx, dir, gen, p);
+	ret = get_cur_path(sctx, dir, dir_gen, path);
 	if (ret < 0)
 		goto out;
-	ret = fs_path_add_path(p, name);
+	ret = fs_path_add_path(path, name);
 	if (ret < 0)
 		goto out;
 
-	ret = __record_ref(refs, dir, gen, p);
-
+	ref->dir = dir;
+	ref->dir_gen = dir_gen;
+	set_ref_path(ref, path);
+	list_add_tail(&ref->list, refs);
+	rb_add(&ref->node, root, rbtree_ref_less);
+	ref->root = root;
 out:
-	if (ret)
-		fs_path_free(p);
+	if (ret) {
+		if (path && (!ref || !ref->full_path))
+			fs_path_free(path);
+		recorded_ref_free(ref);
+	}
 	return ret;
 }
 
-static int __record_new_ref(int num, u64 dir, int index,
-			    struct fs_path *name,
-			    void *ctx)
-{
-	struct send_ctx *sctx = ctx;
-	return record_ref(sctx->send_root, dir, name, ctx, &sctx->new_refs);
-}
-
-
-static int __record_deleted_ref(int num, u64 dir, int index,
-				struct fs_path *name,
-				void *ctx)
+static int record_new_ref_if_needed(int num, u64 dir, int index,
+				    struct fs_path *name, void *ctx)
 {
+	int ret = 0;
 	struct send_ctx *sctx = ctx;
-	return record_ref(sctx->parent_root, dir, name, ctx,
-			  &sctx->deleted_refs);
-}
-
-static int record_new_ref(struct send_ctx *sctx)
-{
-	int ret;
+	struct rb_node *node = NULL;
+	struct recorded_ref data;
+	struct recorded_ref *ref;
+	u64 dir_gen;
 
-	ret = iterate_inode_ref(sctx->send_root, sctx->left_path,
-				sctx->cmp_key, 0, __record_new_ref, sctx);
+	ret = get_inode_gen(sctx->send_root, dir, &dir_gen);
 	if (ret < 0)
 		goto out;
-	ret = 0;
 
+	data.dir = dir;
+	data.dir_gen = dir_gen;
+	set_ref_path(&data, name);
+	node = rb_find(&data, &sctx->rbtree_deleted_refs, rbtree_ref_comp);
+	if (node) {
+		ref = rb_entry(node, struct recorded_ref, node);
+		recorded_ref_free(ref);
+	} else {
+		ret = record_ref_in_tree(&sctx->rbtree_new_refs,
+					 &sctx->new_refs, name, dir, dir_gen,
+					 sctx);
+	}
 out:
 	return ret;
 }
 
-static int record_deleted_ref(struct send_ctx *sctx)
+static int record_deleted_ref_if_needed(int num, u64 dir, int index,
+					struct fs_path *name, void *ctx)
 {
-	int ret;
+	int ret = 0;
+	struct send_ctx *sctx = ctx;
+	struct rb_node *node = NULL;
+	struct recorded_ref data;
+	struct recorded_ref *ref;
+	u64 dir_gen;
 
-	ret = iterate_inode_ref(sctx->parent_root, sctx->right_path,
-				sctx->cmp_key, 0, __record_deleted_ref, sctx);
+	ret = get_inode_gen(sctx->parent_root, dir, &dir_gen);
 	if (ret < 0)
 		goto out;
-	ret = 0;
 
+	data.dir = dir;
+	data.dir_gen = dir_gen;
+	set_ref_path(&data, name);
+	node = rb_find(&data, &sctx->rbtree_new_refs, rbtree_ref_comp);
+	if (node) {
+		ref = rb_entry(node, struct recorded_ref, node);
+		recorded_ref_free(ref);
+	} else {
+		ret = record_ref_in_tree(&sctx->rbtree_deleted_refs,
+					 &sctx->deleted_refs, name, dir,
+					 dir_gen, sctx);
+	}
 out:
 	return ret;
 }
 
-struct find_ref_ctx {
-	u64 dir;
-	u64 dir_gen;
-	struct btrfs_root *root;
-	struct fs_path *name;
-	int found_idx;
-};
-
-static int __find_iref(int num, u64 dir, int index,
-		       struct fs_path *name,
-		       void *ctx_)
-{
-	struct find_ref_ctx *ctx = ctx_;
-	u64 dir_gen;
-	int ret;
-
-	if (dir == ctx->dir && fs_path_len(name) == fs_path_len(ctx->name) &&
-	    strncmp(name->start, ctx->name->start, fs_path_len(name)) == 0) {
-		/*
-		 * To avoid doing extra lookups we'll only do this if everything
-		 * else matches.
-		 */
-		ret = get_inode_info(ctx->root, dir, NULL, &dir_gen, NULL,
-				     NULL, NULL, NULL);
-		if (ret)
-			return ret;
-		if (dir_gen != ctx->dir_gen)
-			return 0;
-		ctx->found_idx = num;
-		return 1;
-	}
-	return 0;
-}
-
-static int find_iref(struct btrfs_root *root,
-		     struct btrfs_path *path,
-		     struct btrfs_key *key,
-		     u64 dir, u64 dir_gen, struct fs_path *name)
+static int record_new_ref(struct send_ctx *sctx)
 {
 	int ret;
-	struct find_ref_ctx ctx;
-
-	ctx.dir = dir;
-	ctx.name = name;
-	ctx.dir_gen = dir_gen;
-	ctx.found_idx = -1;
-	ctx.root = root;
 
-	ret = iterate_inode_ref(root, path, key, 0, __find_iref, &ctx);
+	ret = iterate_inode_ref(sctx->send_root, sctx->left_path,
+				sctx->cmp_key, 0, record_new_ref_if_needed, sctx);
 	if (ret < 0)
-		return ret;
-
-	if (ctx.found_idx == -1)
-		return -ENOENT;
-
-	return ctx.found_idx;
-}
-
-static int __record_changed_new_ref(int num, u64 dir, int index,
-				    struct fs_path *name,
-				    void *ctx)
-{
-	u64 dir_gen;
-	int ret;
-	struct send_ctx *sctx = ctx;
-
-	ret = get_inode_info(sctx->send_root, dir, NULL, &dir_gen, NULL,
-			     NULL, NULL, NULL);
-	if (ret)
-		return ret;
-
-	ret = find_iref(sctx->parent_root, sctx->right_path,
-			sctx->cmp_key, dir, dir_gen, name);
-	if (ret == -ENOENT)
-		ret = __record_new_ref(num, dir, index, name, sctx);
-	else if (ret > 0)
-		ret = 0;
+		goto out;
+	ret = 0;
 
+out:
 	return ret;
 }
 
-static int __record_changed_deleted_ref(int num, u64 dir, int index,
-					struct fs_path *name,
-					void *ctx)
+static int record_deleted_ref(struct send_ctx *sctx)
 {
-	u64 dir_gen;
 	int ret;
-	struct send_ctx *sctx = ctx;
 
-	ret = get_inode_info(sctx->parent_root, dir, NULL, &dir_gen, NULL,
-			     NULL, NULL, NULL);
-	if (ret)
-		return ret;
-
-	ret = find_iref(sctx->send_root, sctx->left_path, sctx->cmp_key,
-			dir, dir_gen, name);
-	if (ret == -ENOENT)
-		ret = __record_deleted_ref(num, dir, index, name, sctx);
-	else if (ret > 0)
-		ret = 0;
+	ret = iterate_inode_ref(sctx->parent_root, sctx->right_path,
+				sctx->cmp_key, 0, record_deleted_ref_if_needed,
+				sctx);
+	if (ret < 0)
+		goto out;
+	ret = 0;
 
+out:
 	return ret;
 }
 
@@ -4401,11 +4537,11 @@ static int record_changed_ref(struct send_ctx *sctx)
 	int ret = 0;
 
 	ret = iterate_inode_ref(sctx->send_root, sctx->left_path,
-			sctx->cmp_key, 0, __record_changed_new_ref, sctx);
+			sctx->cmp_key, 0, record_new_ref_if_needed, sctx);
 	if (ret < 0)
 		goto out;
 	ret = iterate_inode_ref(sctx->parent_root, sctx->right_path,
-			sctx->cmp_key, 0, __record_changed_deleted_ref, sctx);
+			sctx->cmp_key, 0, record_deleted_ref_if_needed, sctx);
 	if (ret < 0)
 		goto out;
 	ret = 0;
@@ -4421,13 +4557,12 @@ out:
 static int process_all_refs(struct send_ctx *sctx,
 			    enum btrfs_compare_tree_result cmd)
 {
-	int ret;
+	int ret = 0;
+	int iter_ret = 0;
 	struct btrfs_root *root;
 	struct btrfs_path *path;
 	struct btrfs_key key;
 	struct btrfs_key found_key;
-	struct extent_buffer *eb;
-	int slot;
 	iterate_inode_ref_t cb;
 	int pending_move = 0;
 
@@ -4437,10 +4572,10 @@ static int process_all_refs(struct send_ctx *sctx,
 
 	if (cmd == BTRFS_COMPARE_TREE_NEW) {
 		root = sctx->send_root;
-		cb = __record_new_ref;
+		cb = record_new_ref_if_needed;
 	} else if (cmd == BTRFS_COMPARE_TREE_DELETED) {
 		root = sctx->parent_root;
-		cb = __record_deleted_ref;
+		cb = record_deleted_ref_if_needed;
 	} else {
 		btrfs_err(sctx->send_root->fs_info,
 				"Wrong command %d in process_all_refs", cmd);
@@ -4451,24 +4586,7 @@ static int process_all_refs(struct send_ctx *sctx,
 	key.objectid = sctx->cmp_key->objectid;
 	key.type = BTRFS_INODE_REF_KEY;
 	key.offset = 0;
-	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
-	if (ret < 0)
-		goto out;
-
-	while (1) {
-		eb = path->nodes[0];
-		slot = path->slots[0];
-		if (slot >= btrfs_header_nritems(eb)) {
-			ret = btrfs_next_leaf(root, path);
-			if (ret < 0)
-				goto out;
-			else if (ret > 0)
-				break;
-			continue;
-		}
-
-		btrfs_item_key_to_cpu(eb, &found_key, slot);
-
+	btrfs_for_each_slot(root, &key, &found_key, path, iter_ret) {
 		if (found_key.objectid != key.objectid ||
 		    (found_key.type != BTRFS_INODE_REF_KEY &&
 		     found_key.type != BTRFS_INODE_EXTREF_KEY))
@@ -4477,8 +4595,11 @@ static int process_all_refs(struct send_ctx *sctx,
 		ret = iterate_inode_ref(root, path, &found_key, 0, cb, sctx);
 		if (ret < 0)
 			goto out;
-
-		path->slots[0]++;
+	}
+	/* Catch error found during iteration */
+	if (iter_ret < 0) {
+		ret = iter_ret;
+		goto out;
 	}
 	btrfs_release_path(path);
 
@@ -4536,15 +4657,18 @@ out:
 }
 
 static int __process_new_xattr(int num, struct btrfs_key *di_key,
-			       const char *name, int name_len,
-			       const char *data, int data_len,
-			       u8 type, void *ctx)
+			       const char *name, int name_len, const char *data,
+			       int data_len, void *ctx)
 {
 	int ret;
 	struct send_ctx *sctx = ctx;
 	struct fs_path *p;
 	struct posix_acl_xattr_header dummy_acl;
 
+	/* Capabilities are emitted by finish_inode_if_needed */
+	if (!strncmp(name, XATTR_NAME_CAPS, name_len))
+		return 0;
+
 	p = fs_path_alloc();
 	if (!p)
 		return -ENOMEM;
@@ -4578,8 +4702,7 @@ out:
 
 static int __process_deleted_xattr(int num, struct btrfs_key *di_key,
 				   const char *name, int name_len,
-				   const char *data, int data_len,
-				   u8 type, void *ctx)
+				   const char *data, int data_len, void *ctx)
 {
 	int ret;
 	struct send_ctx *sctx = ctx;
@@ -4624,10 +4747,8 @@ struct find_xattr_ctx {
 	int found_data_len;
 };
 
-static int __find_xattr(int num, struct btrfs_key *di_key,
-			const char *name, int name_len,
-			const char *data, int data_len,
-			u8 type, void *vctx)
+static int __find_xattr(int num, struct btrfs_key *di_key, const char *name,
+			int name_len, const char *data, int data_len, void *vctx)
 {
 	struct find_xattr_ctx *ctx = vctx;
 
@@ -4677,7 +4798,7 @@ static int find_xattr(struct btrfs_root *root,
 static int __process_changed_new_xattr(int num, struct btrfs_key *di_key,
 				       const char *name, int name_len,
 				       const char *data, int data_len,
-				       u8 type, void *ctx)
+				       void *ctx)
 {
 	int ret;
 	struct send_ctx *sctx = ctx;
@@ -4689,12 +4810,12 @@ static int __process_changed_new_xattr(int num, struct btrfs_key *di_key,
 			 &found_data_len);
 	if (ret == -ENOENT) {
 		ret = __process_new_xattr(num, di_key, name, name_len, data,
-				data_len, type, ctx);
+					  data_len, ctx);
 	} else if (ret >= 0) {
 		if (data_len != found_data_len ||
 		    memcmp(data, found_data, data_len)) {
 			ret = __process_new_xattr(num, di_key, name, name_len,
-					data, data_len, type, ctx);
+						  data, data_len, ctx);
 		} else {
 			ret = 0;
 		}
@@ -4707,7 +4828,7 @@ static int __process_changed_new_xattr(int num, struct btrfs_key *di_key,
 static int __process_changed_deleted_xattr(int num, struct btrfs_key *di_key,
 					   const char *name, int name_len,
 					   const char *data, int data_len,
-					   u8 type, void *ctx)
+					   void *ctx)
 {
 	int ret;
 	struct send_ctx *sctx = ctx;
@@ -4716,7 +4837,7 @@ static int __process_changed_deleted_xattr(int num, struct btrfs_key *di_key,
 			 name, name_len, NULL, NULL);
 	if (ret == -ENOENT)
 		ret = __process_deleted_xattr(num, di_key, name, name_len, data,
-				data_len, type, ctx);
+					      data_len, ctx);
 	else if (ret >= 0)
 		ret = 0;
 
@@ -4740,13 +4861,12 @@ out:
 
 static int process_all_new_xattrs(struct send_ctx *sctx)
 {
-	int ret;
+	int ret = 0;
+	int iter_ret = 0;
 	struct btrfs_root *root;
 	struct btrfs_path *path;
 	struct btrfs_key key;
 	struct btrfs_key found_key;
-	struct extent_buffer *eb;
-	int slot;
 
 	path = alloc_path_for_send();
 	if (!path)
@@ -4757,124 +4877,199 @@ static int process_all_new_xattrs(struct send_ctx *sctx)
 	key.objectid = sctx->cmp_key->objectid;
 	key.type = BTRFS_XATTR_ITEM_KEY;
 	key.offset = 0;
-	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
-	if (ret < 0)
-		goto out;
-
-	while (1) {
-		eb = path->nodes[0];
-		slot = path->slots[0];
-		if (slot >= btrfs_header_nritems(eb)) {
-			ret = btrfs_next_leaf(root, path);
-			if (ret < 0) {
-				goto out;
-			} else if (ret > 0) {
-				ret = 0;
-				break;
-			}
-			continue;
-		}
-
-		btrfs_item_key_to_cpu(eb, &found_key, slot);
+	btrfs_for_each_slot(root, &key, &found_key, path, iter_ret) {
 		if (found_key.objectid != key.objectid ||
 		    found_key.type != key.type) {
 			ret = 0;
-			goto out;
+			break;
 		}
 
 		ret = iterate_dir_item(root, path, __process_new_xattr, sctx);
 		if (ret < 0)
-			goto out;
-
-		path->slots[0]++;
+			break;
 	}
+	/* Catch error found during iteration */
+	if (iter_ret < 0)
+		ret = iter_ret;
 
-out:
 	btrfs_free_path(path);
 	return ret;
 }
 
-static ssize_t fill_read_buf(struct send_ctx *sctx, u64 offset, u32 len)
+static int send_verity(struct send_ctx *sctx, struct fs_path *path,
+		       struct fsverity_descriptor *desc)
+{
+	int ret;
+
+	ret = begin_cmd(sctx, BTRFS_SEND_C_ENABLE_VERITY);
+	if (ret < 0)
+		goto out;
+
+	TLV_PUT_PATH(sctx, BTRFS_SEND_A_PATH, path);
+	TLV_PUT_U8(sctx, BTRFS_SEND_A_VERITY_ALGORITHM,
+			le8_to_cpu(desc->hash_algorithm));
+	TLV_PUT_U32(sctx, BTRFS_SEND_A_VERITY_BLOCK_SIZE,
+			1U << le8_to_cpu(desc->log_blocksize));
+	TLV_PUT(sctx, BTRFS_SEND_A_VERITY_SALT_DATA, desc->salt,
+			le8_to_cpu(desc->salt_size));
+	TLV_PUT(sctx, BTRFS_SEND_A_VERITY_SIG_DATA, desc->signature,
+			le32_to_cpu(desc->sig_size));
+
+	ret = send_cmd(sctx);
+
+tlv_put_failure:
+out:
+	return ret;
+}
+
+static int process_verity(struct send_ctx *sctx)
+{
+	int ret = 0;
+	struct btrfs_fs_info *fs_info = sctx->send_root->fs_info;
+	struct inode *inode;
+	struct fs_path *p;
+
+	inode = btrfs_iget(fs_info->sb, sctx->cur_ino, sctx->send_root);
+	if (IS_ERR(inode))
+		return PTR_ERR(inode);
+
+	ret = btrfs_get_verity_descriptor(inode, NULL, 0);
+	if (ret < 0)
+		goto iput;
+
+	if (ret > FS_VERITY_MAX_DESCRIPTOR_SIZE) {
+		ret = -EMSGSIZE;
+		goto iput;
+	}
+	if (!sctx->verity_descriptor) {
+		sctx->verity_descriptor = kvmalloc(FS_VERITY_MAX_DESCRIPTOR_SIZE,
+						   GFP_KERNEL);
+		if (!sctx->verity_descriptor) {
+			ret = -ENOMEM;
+			goto iput;
+		}
+	}
+
+	ret = btrfs_get_verity_descriptor(inode, sctx->verity_descriptor, ret);
+	if (ret < 0)
+		goto iput;
+
+	p = fs_path_alloc();
+	if (!p) {
+		ret = -ENOMEM;
+		goto iput;
+	}
+	ret = get_cur_path(sctx, sctx->cur_ino, sctx->cur_inode_gen, p);
+	if (ret < 0)
+		goto free_path;
+
+	ret = send_verity(sctx, p, sctx->verity_descriptor);
+	if (ret < 0)
+		goto free_path;
+
+free_path:
+	fs_path_free(p);
+iput:
+	iput(inode);
+	return ret;
+}
+
+static inline u64 max_send_read_size(const struct send_ctx *sctx)
+{
+	return sctx->send_max_size - SZ_16K;
+}
+
+static int put_data_header(struct send_ctx *sctx, u32 len)
+{
+	if (WARN_ON_ONCE(sctx->put_data))
+		return -EINVAL;
+	sctx->put_data = true;
+	if (sctx->proto >= 2) {
+		/*
+		 * Since v2, the data attribute header doesn't include a length,
+		 * it is implicitly to the end of the command.
+		 */
+		if (sctx->send_max_size - sctx->send_size < sizeof(__le16) + len)
+			return -EOVERFLOW;
+		put_unaligned_le16(BTRFS_SEND_A_DATA, sctx->send_buf + sctx->send_size);
+		sctx->send_size += sizeof(__le16);
+	} else {
+		struct btrfs_tlv_header *hdr;
+
+		if (sctx->send_max_size - sctx->send_size < sizeof(*hdr) + len)
+			return -EOVERFLOW;
+		hdr = (struct btrfs_tlv_header *)(sctx->send_buf + sctx->send_size);
+		put_unaligned_le16(BTRFS_SEND_A_DATA, &hdr->tlv_type);
+		put_unaligned_le16(len, &hdr->tlv_len);
+		sctx->send_size += sizeof(*hdr);
+	}
+	return 0;
+}
+
+static int put_file_data(struct send_ctx *sctx, u64 offset, u32 len)
 {
 	struct btrfs_root *root = sctx->send_root;
 	struct btrfs_fs_info *fs_info = root->fs_info;
-	struct inode *inode;
 	struct page *page;
-	char *addr;
-	struct btrfs_key key;
 	pgoff_t index = offset >> PAGE_SHIFT;
 	pgoff_t last_index;
 	unsigned pg_offset = offset_in_page(offset);
-	ssize_t ret = 0;
-
-	key.objectid = sctx->cur_ino;
-	key.type = BTRFS_INODE_ITEM_KEY;
-	key.offset = 0;
-
-	inode = btrfs_iget(fs_info->sb, &key, root);
-	if (IS_ERR(inode))
-		return PTR_ERR(inode);
+	int ret;
 
-	if (offset + len > i_size_read(inode)) {
-		if (offset > i_size_read(inode))
-			len = 0;
-		else
-			len = offset - i_size_read(inode);
-	}
-	if (len == 0)
-		goto out;
+	ret = put_data_header(sctx, len);
+	if (ret)
+		return ret;
 
 	last_index = (offset + len - 1) >> PAGE_SHIFT;
 
-	/* initial readahead */
-	memset(&sctx->ra, 0, sizeof(struct file_ra_state));
-	file_ra_state_init(&sctx->ra, inode->i_mapping);
-
 	while (index <= last_index) {
 		unsigned cur_len = min_t(unsigned, len,
 					 PAGE_SIZE - pg_offset);
 
-		page = find_lock_page(inode->i_mapping, index);
+		page = find_lock_page(sctx->cur_inode->i_mapping, index);
 		if (!page) {
-			page_cache_sync_readahead(inode->i_mapping, &sctx->ra,
-				NULL, index, last_index + 1 - index);
+			page_cache_sync_readahead(sctx->cur_inode->i_mapping,
+						  &sctx->ra, NULL, index,
+						  last_index + 1 - index);
 
-			page = find_or_create_page(inode->i_mapping, index,
-					GFP_KERNEL);
+			page = find_or_create_page(sctx->cur_inode->i_mapping,
+						   index, GFP_KERNEL);
 			if (!page) {
 				ret = -ENOMEM;
 				break;
 			}
 		}
 
-		if (PageReadahead(page)) {
-			page_cache_async_readahead(inode->i_mapping, &sctx->ra,
-				NULL, page, index, last_index + 1 - index);
-		}
+		if (PageReadahead(page))
+			page_cache_async_readahead(sctx->cur_inode->i_mapping,
+						   &sctx->ra, NULL, page_folio(page),
+						   index, last_index + 1 - index);
 
 		if (!PageUptodate(page)) {
-			btrfs_readpage(NULL, page);
+			btrfs_read_folio(NULL, page_folio(page));
 			lock_page(page);
 			if (!PageUptodate(page)) {
 				unlock_page(page);
+				btrfs_err(fs_info,
+			"send: IO error at offset %llu for inode %llu root %llu",
+					page_offset(page), sctx->cur_ino,
+					sctx->send_root->root_key.objectid);
 				put_page(page);
 				ret = -EIO;
 				break;
 			}
 		}
 
-		addr = kmap(page);
-		memcpy(sctx->read_buf + ret, addr + pg_offset, cur_len);
-		kunmap(page);
+		memcpy_from_page(sctx->send_buf + sctx->send_size, page,
+				 pg_offset, cur_len);
 		unlock_page(page);
 		put_page(page);
 		index++;
 		pg_offset = 0;
 		len -= cur_len;
-		ret += cur_len;
+		sctx->send_size += cur_len;
 	}
-out:
-	iput(inode);
+
 	return ret;
 }
 
@@ -4887,7 +5082,6 @@ static int send_write(struct send_ctx *sctx, u64 offset, u32 len)
 	struct btrfs_fs_info *fs_info = sctx->send_root->fs_info;
 	int ret = 0;
 	struct fs_path *p;
-	ssize_t num_read = 0;
 
 	p = fs_path_alloc();
 	if (!p)
@@ -4895,13 +5089,6 @@ static int send_write(struct send_ctx *sctx, u64 offset, u32 len)
 
 	btrfs_debug(fs_info, "send_write offset=%llu, len=%d", offset, len);
 
-	num_read = fill_read_buf(sctx, offset, len);
-	if (num_read <= 0) {
-		if (num_read < 0)
-			ret = num_read;
-		goto out;
-	}
-
 	ret = begin_cmd(sctx, BTRFS_SEND_C_WRITE);
 	if (ret < 0)
 		goto out;
@@ -4912,16 +5099,16 @@ static int send_write(struct send_ctx *sctx, u64 offset, u32 len)
 
 	TLV_PUT_PATH(sctx, BTRFS_SEND_A_PATH, p);
 	TLV_PUT_U64(sctx, BTRFS_SEND_A_FILE_OFFSET, offset);
-	TLV_PUT(sctx, BTRFS_SEND_A_DATA, sctx->read_buf, num_read);
+	ret = put_file_data(sctx, offset, len);
+	if (ret < 0)
+		goto out;
 
 	ret = send_cmd(sctx);
 
 tlv_put_failure:
 out:
 	fs_path_free(p);
-	if (ret < 0)
-		return ret;
-	return num_read;
+	return ret;
 }
 
 /*
@@ -4957,8 +5144,7 @@ static int send_clone(struct send_ctx *sctx,
 	TLV_PUT_PATH(sctx, BTRFS_SEND_A_PATH, p);
 
 	if (clone_root->root == sctx->send_root) {
-		ret = get_inode_info(sctx->send_root, clone_root->ino, NULL,
-				&gen, NULL, NULL, NULL, NULL);
+		ret = get_inode_gen(sctx->send_root, clone_root->ino, &gen);
 		if (ret < 0)
 			goto out;
 		ret = get_cur_path(sctx, clone_root->ino, gen, p);
@@ -4984,7 +5170,7 @@ static int send_clone(struct send_ctx *sctx,
 		TLV_PUT_UUID(sctx, BTRFS_SEND_A_CLONE_UUID,
 			     clone_root->root->root_item.uuid);
 	TLV_PUT_U64(sctx, BTRFS_SEND_A_CLONE_CTRANSID,
-		    le64_to_cpu(clone_root->root->root_item.ctransid));
+		    btrfs_root_ctransid(&clone_root->root->root_item));
 	TLV_PUT_PATH(sctx, BTRFS_SEND_A_CLONE_PATH, p);
 	TLV_PUT_U64(sctx, BTRFS_SEND_A_CLONE_OFFSET,
 			clone_root->offset);
@@ -5033,8 +5219,8 @@ out:
 static int send_hole(struct send_ctx *sctx, u64 end)
 {
 	struct fs_path *p = NULL;
+	u64 read_size = max_send_read_size(sctx);
 	u64 offset = sctx->cur_inode_last_extent;
-	u64 len;
 	int ret = 0;
 
 	/*
@@ -5061,16 +5247,19 @@ static int send_hole(struct send_ctx *sctx, u64 end)
 	ret = get_cur_path(sctx, sctx->cur_ino, sctx->cur_inode_gen, p);
 	if (ret < 0)
 		goto tlv_put_failure;
-	memset(sctx->read_buf, 0, BTRFS_SEND_READ_SIZE);
 	while (offset < end) {
-		len = min_t(u64, end - offset, BTRFS_SEND_READ_SIZE);
+		u64 len = min(end - offset, read_size);
 
 		ret = begin_cmd(sctx, BTRFS_SEND_C_WRITE);
 		if (ret < 0)
 			break;
 		TLV_PUT_PATH(sctx, BTRFS_SEND_A_PATH, p);
 		TLV_PUT_U64(sctx, BTRFS_SEND_A_FILE_OFFSET, offset);
-		TLV_PUT(sctx, BTRFS_SEND_A_DATA, sctx->read_buf, len);
+		ret = put_data_header(sctx, len);
+		if (ret < 0)
+			break;
+		memset(sctx->send_buf + sctx->send_size, 0, len);
+		sctx->send_size += len;
 		ret = send_cmd(sctx);
 		if (ret < 0)
 			break;
@@ -5082,41 +5271,359 @@ tlv_put_failure:
 	return ret;
 }
 
-static int send_extent_data(struct send_ctx *sctx,
-			    const u64 offset,
-			    const u64 len)
+static int send_encoded_inline_extent(struct send_ctx *sctx,
+				      struct btrfs_path *path, u64 offset,
+				      u64 len)
 {
+	struct btrfs_root *root = sctx->send_root;
+	struct btrfs_fs_info *fs_info = root->fs_info;
+	struct inode *inode;
+	struct fs_path *fspath;
+	struct extent_buffer *leaf = path->nodes[0];
+	struct btrfs_key key;
+	struct btrfs_file_extent_item *ei;
+	u64 ram_bytes;
+	size_t inline_size;
+	int ret;
+
+	inode = btrfs_iget(fs_info->sb, sctx->cur_ino, root);
+	if (IS_ERR(inode))
+		return PTR_ERR(inode);
+
+	fspath = fs_path_alloc();
+	if (!fspath) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	ret = begin_cmd(sctx, BTRFS_SEND_C_ENCODED_WRITE);
+	if (ret < 0)
+		goto out;
+
+	ret = get_cur_path(sctx, sctx->cur_ino, sctx->cur_inode_gen, fspath);
+	if (ret < 0)
+		goto out;
+
+	btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
+	ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_file_extent_item);
+	ram_bytes = btrfs_file_extent_ram_bytes(leaf, ei);
+	inline_size = btrfs_file_extent_inline_item_len(leaf, path->slots[0]);
+
+	TLV_PUT_PATH(sctx, BTRFS_SEND_A_PATH, fspath);
+	TLV_PUT_U64(sctx, BTRFS_SEND_A_FILE_OFFSET, offset);
+	TLV_PUT_U64(sctx, BTRFS_SEND_A_UNENCODED_FILE_LEN,
+		    min(key.offset + ram_bytes - offset, len));
+	TLV_PUT_U64(sctx, BTRFS_SEND_A_UNENCODED_LEN, ram_bytes);
+	TLV_PUT_U64(sctx, BTRFS_SEND_A_UNENCODED_OFFSET, offset - key.offset);
+	ret = btrfs_encoded_io_compression_from_extent(fs_info,
+				btrfs_file_extent_compression(leaf, ei));
+	if (ret < 0)
+		goto out;
+	TLV_PUT_U32(sctx, BTRFS_SEND_A_COMPRESSION, ret);
+
+	ret = put_data_header(sctx, inline_size);
+	if (ret < 0)
+		goto out;
+	read_extent_buffer(leaf, sctx->send_buf + sctx->send_size,
+			   btrfs_file_extent_inline_start(ei), inline_size);
+	sctx->send_size += inline_size;
+
+	ret = send_cmd(sctx);
+
+tlv_put_failure:
+out:
+	fs_path_free(fspath);
+	iput(inode);
+	return ret;
+}
+
+static int send_encoded_extent(struct send_ctx *sctx, struct btrfs_path *path,
+			       u64 offset, u64 len)
+{
+	struct btrfs_root *root = sctx->send_root;
+	struct btrfs_fs_info *fs_info = root->fs_info;
+	struct inode *inode;
+	struct fs_path *fspath;
+	struct extent_buffer *leaf = path->nodes[0];
+	struct btrfs_key key;
+	struct btrfs_file_extent_item *ei;
+	u64 disk_bytenr, disk_num_bytes;
+	u32 data_offset;
+	struct btrfs_cmd_header *hdr;
+	u32 crc;
+	int ret;
+
+	inode = btrfs_iget(fs_info->sb, sctx->cur_ino, root);
+	if (IS_ERR(inode))
+		return PTR_ERR(inode);
+
+	fspath = fs_path_alloc();
+	if (!fspath) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	ret = begin_cmd(sctx, BTRFS_SEND_C_ENCODED_WRITE);
+	if (ret < 0)
+		goto out;
+
+	ret = get_cur_path(sctx, sctx->cur_ino, sctx->cur_inode_gen, fspath);
+	if (ret < 0)
+		goto out;
+
+	btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
+	ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_file_extent_item);
+	disk_bytenr = btrfs_file_extent_disk_bytenr(leaf, ei);
+	disk_num_bytes = btrfs_file_extent_disk_num_bytes(leaf, ei);
+
+	TLV_PUT_PATH(sctx, BTRFS_SEND_A_PATH, fspath);
+	TLV_PUT_U64(sctx, BTRFS_SEND_A_FILE_OFFSET, offset);
+	TLV_PUT_U64(sctx, BTRFS_SEND_A_UNENCODED_FILE_LEN,
+		    min(key.offset + btrfs_file_extent_num_bytes(leaf, ei) - offset,
+			len));
+	TLV_PUT_U64(sctx, BTRFS_SEND_A_UNENCODED_LEN,
+		    btrfs_file_extent_ram_bytes(leaf, ei));
+	TLV_PUT_U64(sctx, BTRFS_SEND_A_UNENCODED_OFFSET,
+		    offset - key.offset + btrfs_file_extent_offset(leaf, ei));
+	ret = btrfs_encoded_io_compression_from_extent(fs_info,
+				btrfs_file_extent_compression(leaf, ei));
+	if (ret < 0)
+		goto out;
+	TLV_PUT_U32(sctx, BTRFS_SEND_A_COMPRESSION, ret);
+	TLV_PUT_U32(sctx, BTRFS_SEND_A_ENCRYPTION, 0);
+
+	ret = put_data_header(sctx, disk_num_bytes);
+	if (ret < 0)
+		goto out;
+
+	/*
+	 * We want to do I/O directly into the send buffer, so get the next page
+	 * boundary in the send buffer. This means that there may be a gap
+	 * between the beginning of the command and the file data.
+	 */
+	data_offset = ALIGN(sctx->send_size, PAGE_SIZE);
+	if (data_offset > sctx->send_max_size ||
+	    sctx->send_max_size - data_offset < disk_num_bytes) {
+		ret = -EOVERFLOW;
+		goto out;
+	}
+
+	/*
+	 * Note that send_buf is a mapping of send_buf_pages, so this is really
+	 * reading into send_buf.
+	 */
+	ret = btrfs_encoded_read_regular_fill_pages(BTRFS_I(inode), offset,
+						    disk_bytenr, disk_num_bytes,
+						    sctx->send_buf_pages +
+						    (data_offset >> PAGE_SHIFT));
+	if (ret)
+		goto out;
+
+	hdr = (struct btrfs_cmd_header *)sctx->send_buf;
+	hdr->len = cpu_to_le32(sctx->send_size + disk_num_bytes - sizeof(*hdr));
+	hdr->crc = 0;
+	crc = btrfs_crc32c(0, sctx->send_buf, sctx->send_size);
+	crc = btrfs_crc32c(crc, sctx->send_buf + data_offset, disk_num_bytes);
+	hdr->crc = cpu_to_le32(crc);
+
+	ret = write_buf(sctx->send_filp, sctx->send_buf, sctx->send_size,
+			&sctx->send_off);
+	if (!ret) {
+		ret = write_buf(sctx->send_filp, sctx->send_buf + data_offset,
+				disk_num_bytes, &sctx->send_off);
+	}
+	sctx->send_size = 0;
+	sctx->put_data = false;
+
+tlv_put_failure:
+out:
+	fs_path_free(fspath);
+	iput(inode);
+	return ret;
+}
+
+static int send_extent_data(struct send_ctx *sctx, struct btrfs_path *path,
+			    const u64 offset, const u64 len)
+{
+	const u64 end = offset + len;
+	struct extent_buffer *leaf = path->nodes[0];
+	struct btrfs_file_extent_item *ei;
+	u64 read_size = max_send_read_size(sctx);
 	u64 sent = 0;
 
 	if (sctx->flags & BTRFS_SEND_FLAG_NO_FILE_DATA)
 		return send_update_extent(sctx, offset, len);
 
+	ei = btrfs_item_ptr(leaf, path->slots[0],
+			    struct btrfs_file_extent_item);
+	if ((sctx->flags & BTRFS_SEND_FLAG_COMPRESSED) &&
+	    btrfs_file_extent_compression(leaf, ei) != BTRFS_COMPRESS_NONE) {
+		bool is_inline = (btrfs_file_extent_type(leaf, ei) ==
+				  BTRFS_FILE_EXTENT_INLINE);
+
+		/*
+		 * Send the compressed extent unless the compressed data is
+		 * larger than the decompressed data. This can happen if we're
+		 * not sending the entire extent, either because it has been
+		 * partially overwritten/truncated or because this is a part of
+		 * the extent that we couldn't clone in clone_range().
+		 */
+		if (is_inline &&
+		    btrfs_file_extent_inline_item_len(leaf,
+						      path->slots[0]) <= len) {
+			return send_encoded_inline_extent(sctx, path, offset,
+							  len);
+		} else if (!is_inline &&
+			   btrfs_file_extent_disk_num_bytes(leaf, ei) <= len) {
+			return send_encoded_extent(sctx, path, offset, len);
+		}
+	}
+
+	if (sctx->cur_inode == NULL) {
+		struct btrfs_root *root = sctx->send_root;
+
+		sctx->cur_inode = btrfs_iget(root->fs_info->sb, sctx->cur_ino, root);
+		if (IS_ERR(sctx->cur_inode)) {
+			int err = PTR_ERR(sctx->cur_inode);
+
+			sctx->cur_inode = NULL;
+			return err;
+		}
+		memset(&sctx->ra, 0, sizeof(struct file_ra_state));
+		file_ra_state_init(&sctx->ra, sctx->cur_inode->i_mapping);
+
+		/*
+		 * It's very likely there are no pages from this inode in the page
+		 * cache, so after reading extents and sending their data, we clean
+		 * the page cache to avoid trashing the page cache (adding pressure
+		 * to the page cache and forcing eviction of other data more useful
+		 * for applications).
+		 *
+		 * We decide if we should clean the page cache simply by checking
+		 * if the inode's mapping nrpages is 0 when we first open it, and
+		 * not by using something like filemap_range_has_page() before
+		 * reading an extent because when we ask the readahead code to
+		 * read a given file range, it may (and almost always does) read
+		 * pages from beyond that range (see the documentation for
+		 * page_cache_sync_readahead()), so it would not be reliable,
+		 * because after reading the first extent future calls to
+		 * filemap_range_has_page() would return true because the readahead
+		 * on the previous extent resulted in reading pages of the current
+		 * extent as well.
+		 */
+		sctx->clean_page_cache = (sctx->cur_inode->i_mapping->nrpages == 0);
+		sctx->page_cache_clear_start = round_down(offset, PAGE_SIZE);
+	}
+
 	while (sent < len) {
-		u64 size = len - sent;
+		u64 size = min(len - sent, read_size);
 		int ret;
 
-		if (size > BTRFS_SEND_READ_SIZE)
-			size = BTRFS_SEND_READ_SIZE;
 		ret = send_write(sctx, offset + sent, size);
 		if (ret < 0)
 			return ret;
-		if (!ret)
-			break;
-		sent += ret;
+		sent += size;
 	}
+
+	if (sctx->clean_page_cache && IS_ALIGNED(end, PAGE_SIZE)) {
+		/*
+		 * Always operate only on ranges that are a multiple of the page
+		 * size. This is not only to prevent zeroing parts of a page in
+		 * the case of subpage sector size, but also to guarantee we evict
+		 * pages, as passing a range that is smaller than page size does
+		 * not evict the respective page (only zeroes part of its content).
+		 *
+		 * Always start from the end offset of the last range cleared.
+		 * This is because the readahead code may (and very often does)
+		 * reads pages beyond the range we request for readahead. So if
+		 * we have an extent layout like this:
+		 *
+		 *            [ extent A ] [ extent B ] [ extent C ]
+		 *
+		 * When we ask page_cache_sync_readahead() to read extent A, it
+		 * may also trigger reads for pages of extent B. If we are doing
+		 * an incremental send and extent B has not changed between the
+		 * parent and send snapshots, some or all of its pages may end
+		 * up being read and placed in the page cache. So when truncating
+		 * the page cache we always start from the end offset of the
+		 * previously processed extent up to the end of the current
+		 * extent.
+		 */
+		truncate_inode_pages_range(&sctx->cur_inode->i_data,
+					   sctx->page_cache_clear_start,
+					   end - 1);
+		sctx->page_cache_clear_start = end;
+	}
+
 	return 0;
 }
 
-static int clone_range(struct send_ctx *sctx,
-		       struct clone_root *clone_root,
-		       const u64 disk_byte,
-		       u64 data_offset,
-		       u64 offset,
-		       u64 len)
+/*
+ * Search for a capability xattr related to sctx->cur_ino. If the capability is
+ * found, call send_set_xattr function to emit it.
+ *
+ * Return 0 if there isn't a capability, or when the capability was emitted
+ * successfully, or < 0 if an error occurred.
+ */
+static int send_capabilities(struct send_ctx *sctx)
+{
+	struct fs_path *fspath = NULL;
+	struct btrfs_path *path;
+	struct btrfs_dir_item *di;
+	struct extent_buffer *leaf;
+	unsigned long data_ptr;
+	char *buf = NULL;
+	int buf_len;
+	int ret = 0;
+
+	path = alloc_path_for_send();
+	if (!path)
+		return -ENOMEM;
+
+	di = btrfs_lookup_xattr(NULL, sctx->send_root, path, sctx->cur_ino,
+				XATTR_NAME_CAPS, strlen(XATTR_NAME_CAPS), 0);
+	if (!di) {
+		/* There is no xattr for this inode */
+		goto out;
+	} else if (IS_ERR(di)) {
+		ret = PTR_ERR(di);
+		goto out;
+	}
+
+	leaf = path->nodes[0];
+	buf_len = btrfs_dir_data_len(leaf, di);
+
+	fspath = fs_path_alloc();
+	buf = kmalloc(buf_len, GFP_KERNEL);
+	if (!fspath || !buf) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	ret = get_cur_path(sctx, sctx->cur_ino, sctx->cur_inode_gen, fspath);
+	if (ret < 0)
+		goto out;
+
+	data_ptr = (unsigned long)(di + 1) + btrfs_dir_name_len(leaf, di);
+	read_extent_buffer(leaf, buf, data_ptr, buf_len);
+
+	ret = send_set_xattr(sctx, fspath, XATTR_NAME_CAPS,
+			strlen(XATTR_NAME_CAPS), buf, buf_len);
+out:
+	kfree(buf);
+	fs_path_free(fspath);
+	btrfs_free_path(path);
+	return ret;
+}
+
+static int clone_range(struct send_ctx *sctx, struct btrfs_path *dst_path,
+		       struct clone_root *clone_root, const u64 disk_byte,
+		       u64 data_offset, u64 offset, u64 len)
 {
 	struct btrfs_path *path;
 	struct btrfs_key key;
 	int ret;
+	struct btrfs_inode_info info;
 	u64 clone_src_i_size = 0;
 
 	/*
@@ -5136,7 +5643,7 @@ static int clone_range(struct send_ctx *sctx,
 	 */
 	if (clone_root->offset == 0 &&
 	    len == sctx->send_root->fs_info->sectorsize)
-		return send_extent_data(sctx, offset, len);
+		return send_extent_data(sctx, dst_path, offset, len);
 
 	path = alloc_path_for_send();
 	if (!path)
@@ -5146,11 +5653,11 @@ static int clone_range(struct send_ctx *sctx,
 	 * There are inodes that have extents that lie behind its i_size. Don't
 	 * accept clones from these extents.
 	 */
-	ret = __get_inode_info(clone_root->root, path, clone_root->ino,
-			       &clone_src_i_size, NULL, NULL, NULL, NULL, NULL);
+	ret = get_inode_info(clone_root->root, clone_root->ino, &info);
 	btrfs_release_path(path);
 	if (ret < 0)
 		goto out;
+	clone_src_i_size = info.size;
 
 	/*
 	 * We can't send a clone operation for the entire range if we find
@@ -5233,7 +5740,8 @@ static int clone_range(struct send_ctx *sctx,
 
 			if (hole_len > len)
 				hole_len = len;
-			ret = send_extent_data(sctx, offset, hole_len);
+			ret = send_extent_data(sctx, dst_path, offset,
+					       hole_len);
 			if (ret < 0)
 				goto out;
 
@@ -5306,14 +5814,16 @@ static int clone_range(struct send_ctx *sctx,
 					if (ret < 0)
 						goto out;
 				}
-				ret = send_extent_data(sctx, offset + slen,
+				ret = send_extent_data(sctx, dst_path,
+						       offset + slen,
 						       clone_len - slen);
 			} else {
 				ret = send_clone(sctx, offset, clone_len,
 						 clone_root);
 			}
 		} else {
-			ret = send_extent_data(sctx, offset, clone_len);
+			ret = send_extent_data(sctx, dst_path, offset,
+					       clone_len);
 		}
 
 		if (ret < 0)
@@ -5324,13 +5834,28 @@ static int clone_range(struct send_ctx *sctx,
 			break;
 		offset += clone_len;
 		clone_root->offset += clone_len;
+
+		/*
+		 * If we are cloning from the file we are currently processing,
+		 * and using the send root as the clone root, we must stop once
+		 * the current clone offset reaches the current eof of the file
+		 * at the receiver, otherwise we would issue an invalid clone
+		 * operation (source range going beyond eof) and cause the
+		 * receiver to fail. So if we reach the current eof, bail out
+		 * and fallback to a regular write.
+		 */
+		if (clone_root->root == sctx->send_root &&
+		    clone_root->ino == sctx->cur_ino &&
+		    clone_root->offset >= sctx->cur_inode_next_write_offset)
+			break;
+
 		data_offset += clone_len;
 next:
 		path->slots[0]++;
 	}
 
 	if (len > 0)
-		ret = send_extent_data(sctx, offset, len);
+		ret = send_extent_data(sctx, dst_path, offset, len);
 	else
 		ret = 0;
 out:
@@ -5344,51 +5869,29 @@ static int send_write_or_clone(struct send_ctx *sctx,
 			       struct clone_root *clone_root)
 {
 	int ret = 0;
-	struct btrfs_file_extent_item *ei;
 	u64 offset = key->offset;
-	u64 len;
-	u8 type;
+	u64 end;
 	u64 bs = sctx->send_root->fs_info->sb->s_blocksize;
 
-	ei = btrfs_item_ptr(path->nodes[0], path->slots[0],
-			struct btrfs_file_extent_item);
-	type = btrfs_file_extent_type(path->nodes[0], ei);
-	if (type == BTRFS_FILE_EXTENT_INLINE) {
-		len = btrfs_file_extent_ram_bytes(path->nodes[0], ei);
-		/*
-		 * it is possible the inline item won't cover the whole page,
-		 * but there may be items after this page.  Make
-		 * sure to send the whole thing
-		 */
-		len = PAGE_ALIGN(len);
-	} else {
-		len = btrfs_file_extent_num_bytes(path->nodes[0], ei);
-	}
-
-	if (offset >= sctx->cur_inode_size) {
-		ret = 0;
-		goto out;
-	}
-	if (offset + len > sctx->cur_inode_size)
-		len = sctx->cur_inode_size - offset;
-	if (len == 0) {
-		ret = 0;
-		goto out;
-	}
+	end = min_t(u64, btrfs_file_extent_end(path), sctx->cur_inode_size);
+	if (offset >= end)
+		return 0;
 
-	if (clone_root && IS_ALIGNED(offset + len, bs)) {
+	if (clone_root && IS_ALIGNED(end, bs)) {
+		struct btrfs_file_extent_item *ei;
 		u64 disk_byte;
 		u64 data_offset;
 
+		ei = btrfs_item_ptr(path->nodes[0], path->slots[0],
+				    struct btrfs_file_extent_item);
 		disk_byte = btrfs_file_extent_disk_bytenr(path->nodes[0], ei);
 		data_offset = btrfs_file_extent_offset(path->nodes[0], ei);
-		ret = clone_range(sctx, clone_root, disk_byte, data_offset,
-				  offset, len);
+		ret = clone_range(sctx, path, clone_root, disk_byte,
+				  data_offset, offset, end - offset);
 	} else {
-		ret = send_extent_data(sctx, offset, len);
+		ret = send_extent_data(sctx, path, offset, end - offset);
 	}
-	sctx->cur_inode_next_write_offset = offset + len;
-out:
+	sctx->cur_inode_next_write_offset = end;
 	return ret;
 }
 
@@ -5586,10 +6089,7 @@ static int get_last_extent(struct send_ctx *sctx, u64 offset)
 {
 	struct btrfs_path *path;
 	struct btrfs_root *root = sctx->send_root;
-	struct btrfs_file_extent_item *fi;
 	struct btrfs_key key;
-	u64 extent_end;
-	u8 type;
 	int ret;
 
 	path = alloc_path_for_send();
@@ -5609,18 +6109,7 @@ static int get_last_extent(struct send_ctx *sctx, u64 offset)
 	if (key.objectid != sctx->cur_ino || key.type != BTRFS_EXTENT_DATA_KEY)
 		goto out;
 
-	fi = btrfs_item_ptr(path->nodes[0], path->slots[0],
-			    struct btrfs_file_extent_item);
-	type = btrfs_file_extent_type(path->nodes[0], fi);
-	if (type == BTRFS_FILE_EXTENT_INLINE) {
-		u64 size = btrfs_file_extent_ram_bytes(path->nodes[0], fi);
-		extent_end = ALIGN(key.offset + size,
-				   sctx->send_root->fs_info->sectorsize);
-	} else {
-		extent_end = key.offset +
-			btrfs_file_extent_num_bytes(path->nodes[0], fi);
-	}
-	sctx->cur_inode_last_extent = extent_end;
+	sctx->cur_inode_last_extent = btrfs_file_extent_end(path);
 out:
 	btrfs_free_path(path);
 	return ret;
@@ -5674,16 +6163,7 @@ static int range_is_hole_in_parent(struct send_ctx *sctx,
 			break;
 
 		fi = btrfs_item_ptr(leaf, slot, struct btrfs_file_extent_item);
-		if (btrfs_file_extent_type(leaf, fi) ==
-		    BTRFS_FILE_EXTENT_INLINE) {
-			u64 size = btrfs_file_extent_ram_bytes(leaf, fi);
-
-			extent_end = ALIGN(key.offset + size,
-					   root->fs_info->sectorsize);
-		} else {
-			extent_end = key.offset +
-				btrfs_file_extent_num_bytes(leaf, fi);
-		}
+		extent_end = btrfs_file_extent_end(path);
 		if (extent_end <= start)
 			goto next;
 		if (btrfs_file_extent_disk_bytenr(leaf, fi) == 0) {
@@ -5704,9 +6184,6 @@ out:
 static int maybe_send_hole(struct send_ctx *sctx, struct btrfs_path *path,
 			   struct btrfs_key *key)
 {
-	struct btrfs_file_extent_item *fi;
-	u64 extent_end;
-	u8 type;
 	int ret = 0;
 
 	if (sctx->cur_ino != key->objectid || !need_send_hole(sctx))
@@ -5718,18 +6195,6 @@ static int maybe_send_hole(struct send_ctx *sctx, struct btrfs_path *path,
 			return ret;
 	}
 
-	fi = btrfs_item_ptr(path->nodes[0], path->slots[0],
-			    struct btrfs_file_extent_item);
-	type = btrfs_file_extent_type(path->nodes[0], fi);
-	if (type == BTRFS_FILE_EXTENT_INLINE) {
-		u64 size = btrfs_file_extent_ram_bytes(path->nodes[0], fi);
-		extent_end = ALIGN(key->offset + size,
-				   sctx->send_root->fs_info->sectorsize);
-	} else {
-		extent_end = key->offset +
-			btrfs_file_extent_num_bytes(path->nodes[0], fi);
-	}
-
 	if (path->slots[0] == 0 &&
 	    sctx->cur_inode_last_extent < key->offset) {
 		/*
@@ -5755,7 +6220,7 @@ static int maybe_send_hole(struct send_ctx *sctx, struct btrfs_path *path,
 		else
 			ret = 0;
 	}
-	sctx->cur_inode_last_extent = extent_end;
+	sctx->cur_inode_last_extent = btrfs_file_extent_end(path);
 	return ret;
 }
 
@@ -5821,13 +6286,12 @@ out:
 
 static int process_all_extents(struct send_ctx *sctx)
 {
-	int ret;
+	int ret = 0;
+	int iter_ret = 0;
 	struct btrfs_root *root;
 	struct btrfs_path *path;
 	struct btrfs_key key;
 	struct btrfs_key found_key;
-	struct extent_buffer *eb;
-	int slot;
 
 	root = sctx->send_root;
 	path = alloc_path_for_send();
@@ -5837,41 +6301,21 @@ static int process_all_extents(struct send_ctx *sctx)
 	key.objectid = sctx->cmp_key->objectid;
 	key.type = BTRFS_EXTENT_DATA_KEY;
 	key.offset = 0;
-	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
-	if (ret < 0)
-		goto out;
-
-	while (1) {
-		eb = path->nodes[0];
-		slot = path->slots[0];
-
-		if (slot >= btrfs_header_nritems(eb)) {
-			ret = btrfs_next_leaf(root, path);
-			if (ret < 0) {
-				goto out;
-			} else if (ret > 0) {
-				ret = 0;
-				break;
-			}
-			continue;
-		}
-
-		btrfs_item_key_to_cpu(eb, &found_key, slot);
-
+	btrfs_for_each_slot(root, &key, &found_key, path, iter_ret) {
 		if (found_key.objectid != key.objectid ||
 		    found_key.type != key.type) {
 			ret = 0;
-			goto out;
+			break;
 		}
 
 		ret = process_extent(sctx, path, &found_key);
 		if (ret < 0)
-			goto out;
-
-		path->slots[0]++;
+			break;
 	}
+	/* Catch error found during iteration */
+	if (iter_ret < 0)
+		ret = iter_ret;
 
-out:
 	btrfs_free_path(path);
 	return ret;
 }
@@ -5902,14 +6346,18 @@ out:
 static int finish_inode_if_needed(struct send_ctx *sctx, int at_end)
 {
 	int ret = 0;
+	struct btrfs_inode_info info;
 	u64 left_mode;
 	u64 left_uid;
 	u64 left_gid;
+	u64 left_fileattr;
 	u64 right_mode;
 	u64 right_uid;
 	u64 right_gid;
+	u64 right_fileattr;
 	int need_chmod = 0;
 	int need_chown = 0;
+	bool need_fileattr = false;
 	int need_truncate = 1;
 	int pending_move = 0;
 	int refs_processed = 0;
@@ -5941,11 +6389,13 @@ static int finish_inode_if_needed(struct send_ctx *sctx, int at_end)
 		goto out;
 	if (!at_end && sctx->cmp_key->objectid == sctx->cur_ino)
 		goto out;
-
-	ret = get_inode_info(sctx->send_root, sctx->cur_ino, NULL, NULL,
-			&left_mode, &left_uid, &left_gid, NULL);
+	ret = get_inode_info(sctx->send_root, sctx->cur_ino, &info);
 	if (ret < 0)
 		goto out;
+	left_mode = info.mode;
+	left_uid = info.uid;
+	left_gid = info.gid;
+	left_fileattr = info.fileattr;
 
 	if (!sctx->parent_root || sctx->cur_inode_new) {
 		need_chown = 1;
@@ -5956,16 +6406,21 @@ static int finish_inode_if_needed(struct send_ctx *sctx, int at_end)
 	} else {
 		u64 old_size;
 
-		ret = get_inode_info(sctx->parent_root, sctx->cur_ino,
-				&old_size, NULL, &right_mode, &right_uid,
-				&right_gid, NULL);
+		ret = get_inode_info(sctx->parent_root, sctx->cur_ino, &info);
 		if (ret < 0)
 			goto out;
+		old_size = info.size;
+		right_mode = info.mode;
+		right_uid = info.uid;
+		right_gid = info.gid;
+		right_fileattr = info.fileattr;
 
 		if (left_uid != right_uid || left_gid != right_gid)
 			need_chown = 1;
 		if (!S_ISLNK(sctx->cur_inode_mode) && left_mode != right_mode)
 			need_chmod = 1;
+		if (!S_ISLNK(sctx->cur_inode_mode) && left_fileattr != right_fileattr)
+			need_fileattr = true;
 		if ((old_size == sctx->cur_inode_size) ||
 		    (sctx->cur_inode_size > old_size &&
 		     sctx->cur_inode_next_write_offset == sctx->cur_inode_size))
@@ -6009,6 +6464,23 @@ static int finish_inode_if_needed(struct send_ctx *sctx, int at_end)
 		if (ret < 0)
 			goto out;
 	}
+	if (need_fileattr) {
+		ret = send_fileattr(sctx, sctx->cur_ino, sctx->cur_inode_gen,
+				    left_fileattr);
+		if (ret < 0)
+			goto out;
+	}
+
+	if (proto_cmd_ok(sctx, BTRFS_SEND_C_ENABLE_VERITY)
+	    && sctx->cur_inode_needs_verity) {
+		ret = process_verity(sctx);
+		if (ret < 0)
+			goto out;
+	}
+
+	ret = send_capabilities(sctx);
+	if (ret < 0)
+		goto out;
 
 	/*
 	 * If other directory inodes depended on our current directory
@@ -6035,91 +6507,28 @@ out:
 	return ret;
 }
 
-struct parent_paths_ctx {
-	struct list_head *refs;
-	struct send_ctx *sctx;
-};
-
-static int record_parent_ref(int num, u64 dir, int index, struct fs_path *name,
-			     void *ctx)
-{
-	struct parent_paths_ctx *ppctx = ctx;
-
-	return record_ref(ppctx->sctx->parent_root, dir, name, ppctx->sctx,
-			  ppctx->refs);
-}
-
-/*
- * Issue unlink operations for all paths of the current inode found in the
- * parent snapshot.
- */
-static int btrfs_unlink_all_paths(struct send_ctx *sctx)
+static void close_current_inode(struct send_ctx *sctx)
 {
-	LIST_HEAD(deleted_refs);
-	struct btrfs_path *path;
-	struct btrfs_key key;
-	struct parent_paths_ctx ctx;
-	int ret;
-
-	path = alloc_path_for_send();
-	if (!path)
-		return -ENOMEM;
-
-	key.objectid = sctx->cur_ino;
-	key.type = BTRFS_INODE_REF_KEY;
-	key.offset = 0;
-	ret = btrfs_search_slot(NULL, sctx->parent_root, &key, path, 0, 0);
-	if (ret < 0)
-		goto out;
-
-	ctx.refs = &deleted_refs;
-	ctx.sctx = sctx;
-
-	while (true) {
-		struct extent_buffer *eb = path->nodes[0];
-		int slot = path->slots[0];
+	u64 i_size;
 
-		if (slot >= btrfs_header_nritems(eb)) {
-			ret = btrfs_next_leaf(sctx->parent_root, path);
-			if (ret < 0)
-				goto out;
-			else if (ret > 0)
-				break;
-			continue;
-		}
-
-		btrfs_item_key_to_cpu(eb, &key, slot);
-		if (key.objectid != sctx->cur_ino)
-			break;
-		if (key.type != BTRFS_INODE_REF_KEY &&
-		    key.type != BTRFS_INODE_EXTREF_KEY)
-			break;
-
-		ret = iterate_inode_ref(sctx->parent_root, path, &key, 1,
-					record_parent_ref, &ctx);
-		if (ret < 0)
-			goto out;
+	if (sctx->cur_inode == NULL)
+		return;
 
-		path->slots[0]++;
-	}
+	i_size = i_size_read(sctx->cur_inode);
 
-	while (!list_empty(&deleted_refs)) {
-		struct recorded_ref *ref;
+	/*
+	 * If we are doing an incremental send, we may have extents between the
+	 * last processed extent and the i_size that have not been processed
+	 * because they haven't changed but we may have read some of their pages
+	 * through readahead, see the comments at send_extent_data().
+	 */
+	if (sctx->clean_page_cache && sctx->page_cache_clear_start < i_size)
+		truncate_inode_pages_range(&sctx->cur_inode->i_data,
+					   sctx->page_cache_clear_start,
+					   round_up(i_size, PAGE_SIZE) - 1);
 
-		ref = list_first_entry(&deleted_refs, struct recorded_ref, list);
-		ret = send_unlink(sctx, ref->full_path);
-		if (ret < 0)
-			goto out;
-		fs_path_free(ref->full_path);
-		list_del(&ref->list);
-		kfree(ref);
-	}
-	ret = 0;
-out:
-	btrfs_free_path(path);
-	if (ret)
-		__free_recorded_refs(&deleted_refs);
-	return ret;
+	iput(sctx->cur_inode);
+	sctx->cur_inode = NULL;
 }
 
 static int changed_inode(struct send_ctx *sctx,
@@ -6132,8 +6541,10 @@ static int changed_inode(struct send_ctx *sctx,
 	u64 left_gen = 0;
 	u64 right_gen = 0;
 
+	close_current_inode(sctx);
+
 	sctx->cur_ino = key->objectid;
-	sctx->cur_inode_new_gen = 0;
+	sctx->cur_inode_new_gen = false;
 	sctx->cur_inode_last_extent = (u64)-1;
 	sctx->cur_inode_next_write_offset = 0;
 	sctx->ignore_cur_inode = false;
@@ -6174,7 +6585,7 @@ static int changed_inode(struct send_ctx *sctx,
 		 */
 		if (left_gen != right_gen &&
 		    sctx->cur_ino != BTRFS_FIRST_FREE_OBJECTID)
-			sctx->cur_inode_new_gen = 1;
+			sctx->cur_inode_new_gen = true;
 	}
 
 	/*
@@ -6186,28 +6597,39 @@ static int changed_inode(struct send_ctx *sctx,
 	 * file descriptor against it or turning a RO snapshot into RW mode,
 	 * keep an open file descriptor against a file, delete it and then
 	 * turn the snapshot back to RO mode before using it for a send
-	 * operation. So if we find such cases, ignore the inode and all its
-	 * items completely if it's a new inode, or if it's a changed inode
-	 * make sure all its previous paths (from the parent snapshot) are all
-	 * unlinked and all other the inode items are ignored.
+	 * operation. The former is what the receiver operation does.
+	 * Therefore, if we want to send these snapshots soon after they're
+	 * received, we need to handle orphan inodes as well. Moreover, orphans
+	 * can appear not only in the send snapshot but also in the parent
+	 * snapshot. Here are several cases:
+	 *
+	 * Case 1: BTRFS_COMPARE_TREE_NEW
+	 *       |  send snapshot  | action
+	 * --------------------------------
+	 * nlink |        0        | ignore
+	 *
+	 * Case 2: BTRFS_COMPARE_TREE_DELETED
+	 *       | parent snapshot | action
+	 * ----------------------------------
+	 * nlink |        0        | as usual
+	 * Note: No unlinks will be sent because there're no paths for it.
+	 *
+	 * Case 3: BTRFS_COMPARE_TREE_CHANGED
+	 *           |       | parent snapshot | send snapshot | action
+	 * -----------------------------------------------------------------------
+	 * subcase 1 | nlink |        0        |       0       | ignore
+	 * subcase 2 | nlink |       >0        |       0       | new_gen(deletion)
+	 * subcase 3 | nlink |        0        |      >0       | new_gen(creation)
+	 *
 	 */
-	if (result == BTRFS_COMPARE_TREE_NEW ||
-	    result == BTRFS_COMPARE_TREE_CHANGED) {
-		u32 nlinks;
-
-		nlinks = btrfs_inode_nlink(sctx->left_path->nodes[0], left_ii);
-		if (nlinks == 0) {
+	if (result == BTRFS_COMPARE_TREE_NEW) {
+		if (btrfs_inode_nlink(sctx->left_path->nodes[0], left_ii) == 0) {
 			sctx->ignore_cur_inode = true;
-			if (result == BTRFS_COMPARE_TREE_CHANGED)
-				ret = btrfs_unlink_all_paths(sctx);
 			goto out;
 		}
-	}
-
-	if (result == BTRFS_COMPARE_TREE_NEW) {
 		sctx->cur_inode_gen = left_gen;
-		sctx->cur_inode_new = 1;
-		sctx->cur_inode_deleted = 0;
+		sctx->cur_inode_new = true;
+		sctx->cur_inode_deleted = false;
 		sctx->cur_inode_size = btrfs_inode_size(
 				sctx->left_path->nodes[0], left_ii);
 		sctx->cur_inode_mode = btrfs_inode_mode(
@@ -6218,13 +6640,23 @@ static int changed_inode(struct send_ctx *sctx,
 			ret = send_create_inode_if_needed(sctx);
 	} else if (result == BTRFS_COMPARE_TREE_DELETED) {
 		sctx->cur_inode_gen = right_gen;
-		sctx->cur_inode_new = 0;
-		sctx->cur_inode_deleted = 1;
+		sctx->cur_inode_new = false;
+		sctx->cur_inode_deleted = true;
 		sctx->cur_inode_size = btrfs_inode_size(
 				sctx->right_path->nodes[0], right_ii);
 		sctx->cur_inode_mode = btrfs_inode_mode(
 				sctx->right_path->nodes[0], right_ii);
 	} else if (result == BTRFS_COMPARE_TREE_CHANGED) {
+		u32 new_nlinks, old_nlinks;
+
+		new_nlinks = btrfs_inode_nlink(sctx->left_path->nodes[0], left_ii);
+		old_nlinks = btrfs_inode_nlink(sctx->right_path->nodes[0], right_ii);
+		if (new_nlinks == 0 && old_nlinks == 0) {
+			sctx->ignore_cur_inode = true;
+			goto out;
+		} else if (new_nlinks == 0 || old_nlinks == 0) {
+			sctx->cur_inode_new_gen = 1;
+		}
 		/*
 		 * We need to do some special handling in case the inode was
 		 * reported as changed with a changed generation number. This
@@ -6236,58 +6668,66 @@ static int changed_inode(struct send_ctx *sctx,
 			/*
 			 * First, process the inode as if it was deleted.
 			 */
-			sctx->cur_inode_gen = right_gen;
-			sctx->cur_inode_new = 0;
-			sctx->cur_inode_deleted = 1;
-			sctx->cur_inode_size = btrfs_inode_size(
-					sctx->right_path->nodes[0], right_ii);
-			sctx->cur_inode_mode = btrfs_inode_mode(
-					sctx->right_path->nodes[0], right_ii);
-			ret = process_all_refs(sctx,
-					BTRFS_COMPARE_TREE_DELETED);
-			if (ret < 0)
-				goto out;
+			if (old_nlinks > 0) {
+				sctx->cur_inode_gen = right_gen;
+				sctx->cur_inode_new = false;
+				sctx->cur_inode_deleted = true;
+				sctx->cur_inode_size = btrfs_inode_size(
+						sctx->right_path->nodes[0], right_ii);
+				sctx->cur_inode_mode = btrfs_inode_mode(
+						sctx->right_path->nodes[0], right_ii);
+				ret = process_all_refs(sctx,
+						BTRFS_COMPARE_TREE_DELETED);
+				if (ret < 0)
+					goto out;
+			}
 
 			/*
 			 * Now process the inode as if it was new.
 			 */
-			sctx->cur_inode_gen = left_gen;
-			sctx->cur_inode_new = 1;
-			sctx->cur_inode_deleted = 0;
-			sctx->cur_inode_size = btrfs_inode_size(
-					sctx->left_path->nodes[0], left_ii);
-			sctx->cur_inode_mode = btrfs_inode_mode(
-					sctx->left_path->nodes[0], left_ii);
-			sctx->cur_inode_rdev = btrfs_inode_rdev(
-					sctx->left_path->nodes[0], left_ii);
-			ret = send_create_inode_if_needed(sctx);
-			if (ret < 0)
-				goto out;
+			if (new_nlinks > 0) {
+				sctx->cur_inode_gen = left_gen;
+				sctx->cur_inode_new = true;
+				sctx->cur_inode_deleted = false;
+				sctx->cur_inode_size = btrfs_inode_size(
+						sctx->left_path->nodes[0],
+						left_ii);
+				sctx->cur_inode_mode = btrfs_inode_mode(
+						sctx->left_path->nodes[0],
+						left_ii);
+				sctx->cur_inode_rdev = btrfs_inode_rdev(
+						sctx->left_path->nodes[0],
+						left_ii);
+				ret = send_create_inode_if_needed(sctx);
+				if (ret < 0)
+					goto out;
 
-			ret = process_all_refs(sctx, BTRFS_COMPARE_TREE_NEW);
-			if (ret < 0)
-				goto out;
-			/*
-			 * Advance send_progress now as we did not get into
-			 * process_recorded_refs_if_needed in the new_gen case.
-			 */
-			sctx->send_progress = sctx->cur_ino + 1;
+				ret = process_all_refs(sctx, BTRFS_COMPARE_TREE_NEW);
+				if (ret < 0)
+					goto out;
+				/*
+				 * Advance send_progress now as we did not get
+				 * into process_recorded_refs_if_needed in the
+				 * new_gen case.
+				 */
+				sctx->send_progress = sctx->cur_ino + 1;
 
-			/*
-			 * Now process all extents and xattrs of the inode as if
-			 * they were all new.
-			 */
-			ret = process_all_extents(sctx);
-			if (ret < 0)
-				goto out;
-			ret = process_all_new_xattrs(sctx);
-			if (ret < 0)
-				goto out;
+				/*
+				 * Now process all extents and xattrs of the
+				 * inode as if they were all new.
+				 */
+				ret = process_all_extents(sctx);
+				if (ret < 0)
+					goto out;
+				ret = process_all_new_xattrs(sctx);
+				if (ret < 0)
+					goto out;
+			}
 		} else {
 			sctx->cur_inode_gen = left_gen;
-			sctx->cur_inode_new = 0;
-			sctx->cur_inode_new_gen = 0;
-			sctx->cur_inode_deleted = 0;
+			sctx->cur_inode_new = false;
+			sctx->cur_inode_new_gen = false;
+			sctx->cur_inode_deleted = false;
 			sctx->cur_inode_size = btrfs_inode_size(
 					sctx->left_path->nodes[0], left_ii);
 			sctx->cur_inode_mode = btrfs_inode_mode(
@@ -6379,7 +6819,7 @@ static int changed_extent(struct send_ctx *sctx,
 	 * updates the inode item, but it only changes the iversion (sequence
 	 * field in the inode item) of the inode, so if a file is deduplicated
 	 * the same amount of times in both the parent and send snapshots, its
-	 * iversion becames the same in both snapshots, whence the inode item is
+	 * iversion becomes the same in both snapshots, whence the inode item is
 	 * the same on both snapshots.
 	 */
 	if (sctx->cur_ino != sctx->cmp_key->objectid)
@@ -6394,18 +6834,27 @@ static int changed_extent(struct send_ctx *sctx,
 	return ret;
 }
 
+static int changed_verity(struct send_ctx *sctx, enum btrfs_compare_tree_result result)
+{
+	int ret = 0;
+
+	if (!sctx->cur_inode_new_gen && !sctx->cur_inode_deleted) {
+		if (result == BTRFS_COMPARE_TREE_NEW)
+			sctx->cur_inode_needs_verity = true;
+	}
+	return ret;
+}
+
 static int dir_changed(struct send_ctx *sctx, u64 dir)
 {
 	u64 orig_gen, new_gen;
 	int ret;
 
-	ret = get_inode_info(sctx->send_root, dir, NULL, &new_gen, NULL, NULL,
-			     NULL, NULL);
+	ret = get_inode_gen(sctx->send_root, dir, &new_gen);
 	if (ret)
 		return ret;
 
-	ret = get_inode_info(sctx->parent_root, dir, NULL, &orig_gen, NULL,
-			     NULL, NULL, NULL);
+	ret = get_inode_gen(sctx->parent_root, dir, &orig_gen);
 	if (ret)
 		return ret;
 
@@ -6433,7 +6882,7 @@ static int compare_refs(struct send_ctx *sctx, struct btrfs_path *path,
 	}
 
 	leaf = path->nodes[0];
-	item_size = btrfs_item_size_nr(leaf, path->slots[0]);
+	item_size = btrfs_item_size(leaf, path->slots[0]);
 	ptr = btrfs_item_ptr_offset(leaf, path->slots[0]);
 	while (cur_offset < item_size) {
 		extref = (struct btrfs_inode_extref *)(ptr +
@@ -6460,10 +6909,53 @@ static int changed_cb(struct btrfs_path *left_path,
 		      struct btrfs_path *right_path,
 		      struct btrfs_key *key,
 		      enum btrfs_compare_tree_result result,
-		      void *ctx)
+		      struct send_ctx *sctx)
 {
 	int ret = 0;
-	struct send_ctx *sctx = ctx;
+
+	/*
+	 * We can not hold the commit root semaphore here. This is because in
+	 * the case of sending and receiving to the same filesystem, using a
+	 * pipe, could result in a deadlock:
+	 *
+	 * 1) The task running send blocks on the pipe because it's full;
+	 *
+	 * 2) The task running receive, which is the only consumer of the pipe,
+	 *    is waiting for a transaction commit (for example due to a space
+	 *    reservation when doing a write or triggering a transaction commit
+	 *    when creating a subvolume);
+	 *
+	 * 3) The transaction is waiting to write lock the commit root semaphore,
+	 *    but can not acquire it since it's being held at 1).
+	 *
+	 * Down this call chain we write to the pipe through kernel_write().
+	 * The same type of problem can also happen when sending to a file that
+	 * is stored in the same filesystem - when reserving space for a write
+	 * into the file, we can trigger a transaction commit.
+	 *
+	 * Our caller has supplied us with clones of leaves from the send and
+	 * parent roots, so we're safe here from a concurrent relocation and
+	 * further reallocation of metadata extents while we are here. Below we
+	 * also assert that the leaves are clones.
+	 */
+	lockdep_assert_not_held(&sctx->send_root->fs_info->commit_root_sem);
+
+	/*
+	 * We always have a send root, so left_path is never NULL. We will not
+	 * have a leaf when we have reached the end of the send root but have
+	 * not yet reached the end of the parent root.
+	 */
+	if (left_path->nodes[0])
+		ASSERT(test_bit(EXTENT_BUFFER_UNMAPPED,
+				&left_path->nodes[0]->bflags));
+	/*
+	 * When doing a full send we don't have a parent root, so right_path is
+	 * NULL. When doing an incremental send, we may have reached the end of
+	 * the parent root already, so we don't have a leaf at right_path.
+	 */
+	if (right_path && right_path->nodes[0])
+		ASSERT(test_bit(EXTENT_BUFFER_UNMAPPED,
+				&right_path->nodes[0]->bflags));
 
 	if (result == BTRFS_COMPARE_TREE_SAME) {
 		if (key->type == BTRFS_INODE_REF_KEY ||
@@ -6505,29 +6997,69 @@ static int changed_cb(struct btrfs_path *left_path,
 			ret = changed_xattr(sctx, result);
 		else if (key->type == BTRFS_EXTENT_DATA_KEY)
 			ret = changed_extent(sctx, result);
+		else if (key->type == BTRFS_VERITY_DESC_ITEM_KEY &&
+			 key->offset == 0)
+			ret = changed_verity(sctx, result);
 	}
 
 out:
 	return ret;
 }
 
+static int search_key_again(const struct send_ctx *sctx,
+			    struct btrfs_root *root,
+			    struct btrfs_path *path,
+			    const struct btrfs_key *key)
+{
+	int ret;
+
+	if (!path->need_commit_sem)
+		lockdep_assert_held_read(&root->fs_info->commit_root_sem);
+
+	/*
+	 * Roots used for send operations are readonly and no one can add,
+	 * update or remove keys from them, so we should be able to find our
+	 * key again. The only exception is deduplication, which can operate on
+	 * readonly roots and add, update or remove keys to/from them - but at
+	 * the moment we don't allow it to run in parallel with send.
+	 */
+	ret = btrfs_search_slot(NULL, root, key, path, 0, 0);
+	ASSERT(ret <= 0);
+	if (ret > 0) {
+		btrfs_print_tree(path->nodes[path->lowest_level], false);
+		btrfs_err(root->fs_info,
+"send: key (%llu %u %llu) not found in %s root %llu, lowest_level %d, slot %d",
+			  key->objectid, key->type, key->offset,
+			  (root == sctx->parent_root ? "parent" : "send"),
+			  root->root_key.objectid, path->lowest_level,
+			  path->slots[path->lowest_level]);
+		return -EUCLEAN;
+	}
+
+	return ret;
+}
+
 static int full_send_tree(struct send_ctx *sctx)
 {
 	int ret;
 	struct btrfs_root *send_root = sctx->send_root;
 	struct btrfs_key key;
+	struct btrfs_fs_info *fs_info = send_root->fs_info;
 	struct btrfs_path *path;
-	struct extent_buffer *eb;
-	int slot;
 
 	path = alloc_path_for_send();
 	if (!path)
 		return -ENOMEM;
+	path->reada = READA_FORWARD_ALWAYS;
 
 	key.objectid = BTRFS_FIRST_FREE_OBJECTID;
 	key.type = BTRFS_INODE_ITEM_KEY;
 	key.offset = 0;
 
+	down_read(&fs_info->commit_root_sem);
+	sctx->last_reloc_trans = fs_info->last_reloc_trans;
+	up_read(&fs_info->commit_root_sem);
+
 	ret = btrfs_search_slot_for_read(send_root, &key, path, 1, 0);
 	if (ret < 0)
 		goto out;
@@ -6535,15 +7067,35 @@ static int full_send_tree(struct send_ctx *sctx)
 		goto out_finish;
 
 	while (1) {
-		eb = path->nodes[0];
-		slot = path->slots[0];
-		btrfs_item_key_to_cpu(eb, &key, slot);
+		btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
 
 		ret = changed_cb(path, NULL, &key,
 				 BTRFS_COMPARE_TREE_NEW, sctx);
 		if (ret < 0)
 			goto out;
 
+		down_read(&fs_info->commit_root_sem);
+		if (fs_info->last_reloc_trans > sctx->last_reloc_trans) {
+			sctx->last_reloc_trans = fs_info->last_reloc_trans;
+			up_read(&fs_info->commit_root_sem);
+			/*
+			 * A transaction used for relocating a block group was
+			 * committed or is about to finish its commit. Release
+			 * our path (leaf) and restart the search, so that we
+			 * avoid operating on any file extent items that are
+			 * stale, with a disk_bytenr that reflects a pre
+			 * relocation value. This way we avoid as much as
+			 * possible to fallback to regular writes when checking
+			 * if we can clone file ranges.
+			 */
+			btrfs_release_path(path);
+			ret = search_key_again(sctx, send_root, path, &key);
+			if (ret < 0)
+				goto out;
+		} else {
+			up_read(&fs_info->commit_root_sem);
+		}
+
 		ret = btrfs_next_item(send_root, path);
 		if (ret < 0)
 			goto out;
@@ -6561,18 +7113,58 @@ out:
 	return ret;
 }
 
-static int tree_move_down(struct btrfs_path *path, int *level)
+static int replace_node_with_clone(struct btrfs_path *path, int level)
+{
+	struct extent_buffer *clone;
+
+	clone = btrfs_clone_extent_buffer(path->nodes[level]);
+	if (!clone)
+		return -ENOMEM;
+
+	free_extent_buffer(path->nodes[level]);
+	path->nodes[level] = clone;
+
+	return 0;
+}
+
+static int tree_move_down(struct btrfs_path *path, int *level, u64 reada_min_gen)
 {
 	struct extent_buffer *eb;
+	struct extent_buffer *parent = path->nodes[*level];
+	int slot = path->slots[*level];
+	const int nritems = btrfs_header_nritems(parent);
+	u64 reada_max;
+	u64 reada_done = 0;
+
+	lockdep_assert_held_read(&parent->fs_info->commit_root_sem);
 
 	BUG_ON(*level == 0);
-	eb = btrfs_read_node_slot(path->nodes[*level], path->slots[*level]);
+	eb = btrfs_read_node_slot(parent, slot);
 	if (IS_ERR(eb))
 		return PTR_ERR(eb);
 
+	/*
+	 * Trigger readahead for the next leaves we will process, so that it is
+	 * very likely that when we need them they are already in memory and we
+	 * will not block on disk IO. For nodes we only do readahead for one,
+	 * since the time window between processing nodes is typically larger.
+	 */
+	reada_max = (*level == 1 ? SZ_128K : eb->fs_info->nodesize);
+
+	for (slot++; slot < nritems && reada_done < reada_max; slot++) {
+		if (btrfs_node_ptr_generation(parent, slot) > reada_min_gen) {
+			btrfs_readahead_node_child(parent, slot);
+			reada_done += eb->fs_info->nodesize;
+		}
+	}
+
 	path->nodes[*level - 1] = eb;
 	path->slots[*level - 1] = 0;
 	(*level)--;
+
+	if (*level == 0)
+		return replace_node_with_clone(path, 0);
+
 	return 0;
 }
 
@@ -6586,8 +7178,10 @@ static int tree_move_next_or_upnext(struct btrfs_path *path,
 	path->slots[*level]++;
 
 	while (path->slots[*level] >= nritems) {
-		if (*level == root_level)
+		if (*level == root_level) {
+			path->slots[*level] = nritems - 1;
 			return -1;
+		}
 
 		/* move upnext */
 		path->slots[*level] = 0;
@@ -6609,23 +7203,30 @@ static int tree_move_next_or_upnext(struct btrfs_path *path,
 static int tree_advance(struct btrfs_path *path,
 			int *level, int root_level,
 			int allow_down,
-			struct btrfs_key *key)
+			struct btrfs_key *key,
+			u64 reada_min_gen)
 {
 	int ret;
 
 	if (*level == 0 || !allow_down) {
 		ret = tree_move_next_or_upnext(path, level, root_level);
 	} else {
-		ret = tree_move_down(path, level);
-	}
-	if (ret >= 0) {
-		if (*level == 0)
-			btrfs_item_key_to_cpu(path->nodes[*level], key,
-					path->slots[*level]);
-		else
-			btrfs_node_key_to_cpu(path->nodes[*level], key,
-					path->slots[*level]);
+		ret = tree_move_down(path, level, reada_min_gen);
 	}
+
+	/*
+	 * Even if we have reached the end of a tree, ret is -1, update the key
+	 * anyway, so that in case we need to restart due to a block group
+	 * relocation, we can assert that the last key of the root node still
+	 * exists in the tree.
+	 */
+	if (*level == 0)
+		btrfs_item_key_to_cpu(path->nodes[*level], key,
+				      path->slots[*level]);
+	else
+		btrfs_node_key_to_cpu(path->nodes[*level], key,
+				      path->slots[*level]);
+
 	return ret;
 }
 
@@ -6637,8 +7238,8 @@ static int tree_compare_item(struct btrfs_path *left_path,
 	int len1, len2;
 	unsigned long off1, off2;
 
-	len1 = btrfs_item_size_nr(left_path->nodes[0], left_path->slots[0]);
-	len2 = btrfs_item_size_nr(right_path->nodes[0], right_path->slots[0]);
+	len1 = btrfs_item_size(left_path->nodes[0], left_path->slots[0]);
+	len2 = btrfs_item_size(right_path->nodes[0], right_path->slots[0]);
 	if (len1 != len2)
 		return 1;
 
@@ -6655,6 +7256,97 @@ static int tree_compare_item(struct btrfs_path *left_path,
 }
 
 /*
+ * A transaction used for relocating a block group was committed or is about to
+ * finish its commit. Release our paths and restart the search, so that we are
+ * not using stale extent buffers:
+ *
+ * 1) For levels > 0, we are only holding references of extent buffers, without
+ *    any locks on them, which does not prevent them from having been relocated
+ *    and reallocated after the last time we released the commit root semaphore.
+ *    The exception are the root nodes, for which we always have a clone, see
+ *    the comment at btrfs_compare_trees();
+ *
+ * 2) For leaves, level 0, we are holding copies (clones) of extent buffers, so
+ *    we are safe from the concurrent relocation and reallocation. However they
+ *    can have file extent items with a pre relocation disk_bytenr value, so we
+ *    restart the start from the current commit roots and clone the new leaves so
+ *    that we get the post relocation disk_bytenr values. Not doing so, could
+ *    make us clone the wrong data in case there are new extents using the old
+ *    disk_bytenr that happen to be shared.
+ */
+static int restart_after_relocation(struct btrfs_path *left_path,
+				    struct btrfs_path *right_path,
+				    const struct btrfs_key *left_key,
+				    const struct btrfs_key *right_key,
+				    int left_level,
+				    int right_level,
+				    const struct send_ctx *sctx)
+{
+	int root_level;
+	int ret;
+
+	lockdep_assert_held_read(&sctx->send_root->fs_info->commit_root_sem);
+
+	btrfs_release_path(left_path);
+	btrfs_release_path(right_path);
+
+	/*
+	 * Since keys can not be added or removed to/from our roots because they
+	 * are readonly and we do not allow deduplication to run in parallel
+	 * (which can add, remove or change keys), the layout of the trees should
+	 * not change.
+	 */
+	left_path->lowest_level = left_level;
+	ret = search_key_again(sctx, sctx->send_root, left_path, left_key);
+	if (ret < 0)
+		return ret;
+
+	right_path->lowest_level = right_level;
+	ret = search_key_again(sctx, sctx->parent_root, right_path, right_key);
+	if (ret < 0)
+		return ret;
+
+	/*
+	 * If the lowest level nodes are leaves, clone them so that they can be
+	 * safely used by changed_cb() while not under the protection of the
+	 * commit root semaphore, even if relocation and reallocation happens in
+	 * parallel.
+	 */
+	if (left_level == 0) {
+		ret = replace_node_with_clone(left_path, 0);
+		if (ret < 0)
+			return ret;
+	}
+
+	if (right_level == 0) {
+		ret = replace_node_with_clone(right_path, 0);
+		if (ret < 0)
+			return ret;
+	}
+
+	/*
+	 * Now clone the root nodes (unless they happen to be the leaves we have
+	 * already cloned). This is to protect against concurrent snapshotting of
+	 * the send and parent roots (see the comment at btrfs_compare_trees()).
+	 */
+	root_level = btrfs_header_level(sctx->send_root->commit_root);
+	if (root_level > 0) {
+		ret = replace_node_with_clone(left_path, root_level);
+		if (ret < 0)
+			return ret;
+	}
+
+	root_level = btrfs_header_level(sctx->parent_root->commit_root);
+	if (root_level > 0) {
+		ret = replace_node_with_clone(right_path, root_level);
+		if (ret < 0)
+			return ret;
+	}
+
+	return 0;
+}
+
+/*
  * This function compares two trees and calls the provided callback for
  * every changed/new/deleted item it finds.
  * If shared tree blocks are encountered, whole subtrees are skipped, making
@@ -6668,8 +7360,7 @@ static int tree_compare_item(struct btrfs_path *left_path,
  * If it detects a change, it aborts immediately.
  */
 static int btrfs_compare_trees(struct btrfs_root *left_root,
-			struct btrfs_root *right_root,
-			btrfs_changed_cb_t changed_cb, void *ctx)
+			struct btrfs_root *right_root, struct send_ctx *sctx)
 {
 	struct btrfs_fs_info *fs_info = left_root->fs_info;
 	int ret;
@@ -6683,14 +7374,15 @@ static int btrfs_compare_trees(struct btrfs_root *left_root,
 	int right_root_level;
 	int left_level;
 	int right_level;
-	int left_end_reached;
-	int right_end_reached;
-	int advance_left;
-	int advance_right;
+	int left_end_reached = 0;
+	int right_end_reached = 0;
+	int advance_left = 0;
+	int advance_right = 0;
 	u64 left_blockptr;
 	u64 right_blockptr;
 	u64 left_gen;
 	u64 right_gen;
+	u64 reada_min_gen;
 
 	left_path = btrfs_alloc_path();
 	if (!left_path) {
@@ -6753,12 +7445,18 @@ static int btrfs_compare_trees(struct btrfs_root *left_root,
 	down_read(&fs_info->commit_root_sem);
 	left_level = btrfs_header_level(left_root->commit_root);
 	left_root_level = left_level;
+	/*
+	 * We clone the root node of the send and parent roots to prevent races
+	 * with snapshot creation of these roots. Snapshot creation COWs the
+	 * root node of a tree, so after the transaction is committed the old
+	 * extent can be reallocated while this send operation is still ongoing.
+	 * So we clone them, under the commit root semaphore, to be race free.
+	 */
 	left_path->nodes[left_level] =
 			btrfs_clone_extent_buffer(left_root->commit_root);
 	if (!left_path->nodes[left_level]) {
-		up_read(&fs_info->commit_root_sem);
 		ret = -ENOMEM;
-		goto out;
+		goto out_unlock;
 	}
 
 	right_level = btrfs_header_level(right_root->commit_root);
@@ -6766,11 +7464,17 @@ static int btrfs_compare_trees(struct btrfs_root *left_root,
 	right_path->nodes[right_level] =
 			btrfs_clone_extent_buffer(right_root->commit_root);
 	if (!right_path->nodes[right_level]) {
-		up_read(&fs_info->commit_root_sem);
 		ret = -ENOMEM;
-		goto out;
+		goto out_unlock;
 	}
-	up_read(&fs_info->commit_root_sem);
+	/*
+	 * Our right root is the parent root, while the left root is the "send"
+	 * root. We know that all new nodes/leaves in the left root must have
+	 * a generation greater than the right root's generation, so we trigger
+	 * readahead for those nodes and leaves of the left root, as we know we
+	 * will need to read them at some point.
+	 */
+	reada_min_gen = btrfs_header_generation(right_root->commit_root);
 
 	if (left_level == 0)
 		btrfs_item_key_to_cpu(left_path->nodes[left_level],
@@ -6785,78 +7489,94 @@ static int btrfs_compare_trees(struct btrfs_root *left_root,
 		btrfs_node_key_to_cpu(right_path->nodes[right_level],
 				&right_key, right_path->slots[right_level]);
 
-	left_end_reached = right_end_reached = 0;
-	advance_left = advance_right = 0;
+	sctx->last_reloc_trans = fs_info->last_reloc_trans;
 
 	while (1) {
-		cond_resched();
+		if (need_resched() ||
+		    rwsem_is_contended(&fs_info->commit_root_sem)) {
+			up_read(&fs_info->commit_root_sem);
+			cond_resched();
+			down_read(&fs_info->commit_root_sem);
+		}
+
+		if (fs_info->last_reloc_trans > sctx->last_reloc_trans) {
+			ret = restart_after_relocation(left_path, right_path,
+						       &left_key, &right_key,
+						       left_level, right_level,
+						       sctx);
+			if (ret < 0)
+				goto out_unlock;
+			sctx->last_reloc_trans = fs_info->last_reloc_trans;
+		}
+
 		if (advance_left && !left_end_reached) {
 			ret = tree_advance(left_path, &left_level,
 					left_root_level,
 					advance_left != ADVANCE_ONLY_NEXT,
-					&left_key);
+					&left_key, reada_min_gen);
 			if (ret == -1)
 				left_end_reached = ADVANCE;
 			else if (ret < 0)
-				goto out;
+				goto out_unlock;
 			advance_left = 0;
 		}
 		if (advance_right && !right_end_reached) {
 			ret = tree_advance(right_path, &right_level,
 					right_root_level,
 					advance_right != ADVANCE_ONLY_NEXT,
-					&right_key);
+					&right_key, reada_min_gen);
 			if (ret == -1)
 				right_end_reached = ADVANCE;
 			else if (ret < 0)
-				goto out;
+				goto out_unlock;
 			advance_right = 0;
 		}
 
 		if (left_end_reached && right_end_reached) {
 			ret = 0;
-			goto out;
+			goto out_unlock;
 		} else if (left_end_reached) {
 			if (right_level == 0) {
+				up_read(&fs_info->commit_root_sem);
 				ret = changed_cb(left_path, right_path,
 						&right_key,
 						BTRFS_COMPARE_TREE_DELETED,
-						ctx);
+						sctx);
 				if (ret < 0)
 					goto out;
+				down_read(&fs_info->commit_root_sem);
 			}
 			advance_right = ADVANCE;
 			continue;
 		} else if (right_end_reached) {
 			if (left_level == 0) {
+				up_read(&fs_info->commit_root_sem);
 				ret = changed_cb(left_path, right_path,
 						&left_key,
 						BTRFS_COMPARE_TREE_NEW,
-						ctx);
+						sctx);
 				if (ret < 0)
 					goto out;
+				down_read(&fs_info->commit_root_sem);
 			}
 			advance_left = ADVANCE;
 			continue;
 		}
 
 		if (left_level == 0 && right_level == 0) {
+			up_read(&fs_info->commit_root_sem);
 			cmp = btrfs_comp_cpu_keys(&left_key, &right_key);
 			if (cmp < 0) {
 				ret = changed_cb(left_path, right_path,
 						&left_key,
 						BTRFS_COMPARE_TREE_NEW,
-						ctx);
-				if (ret < 0)
-					goto out;
+						sctx);
 				advance_left = ADVANCE;
 			} else if (cmp > 0) {
 				ret = changed_cb(left_path, right_path,
 						&right_key,
 						BTRFS_COMPARE_TREE_DELETED,
-						ctx);
-				if (ret < 0)
-					goto out;
+						sctx);
 				advance_right = ADVANCE;
 			} else {
 				enum btrfs_compare_tree_result result;
@@ -6869,12 +7589,14 @@ static int btrfs_compare_trees(struct btrfs_root *left_root,
 				else
 					result = BTRFS_COMPARE_TREE_SAME;
 				ret = changed_cb(left_path, right_path,
-						 &left_key, result, ctx);
-				if (ret < 0)
-					goto out;
+						 &left_key, result, sctx);
 				advance_left = ADVANCE;
 				advance_right = ADVANCE;
 			}
+
+			if (ret < 0)
+				goto out;
+			down_read(&fs_info->commit_root_sem);
 		} else if (left_level == right_level) {
 			cmp = btrfs_comp_cpu_keys(&left_key, &right_key);
 			if (cmp < 0) {
@@ -6914,6 +7636,8 @@ static int btrfs_compare_trees(struct btrfs_root *left_root,
 		}
 	}
 
+out_unlock:
+	up_read(&fs_info->commit_root_sem);
 out:
 	btrfs_free_path(left_path);
 	btrfs_free_path(right_path);
@@ -6936,8 +7660,7 @@ static int send_subvol(struct send_ctx *sctx)
 		goto out;
 
 	if (sctx->parent_root) {
-		ret = btrfs_compare_trees(sctx->send_root, sctx->parent_root,
-				changed_cb, sctx);
+		ret = btrfs_compare_trees(sctx->send_root, sctx->parent_root, sctx);
 		if (ret < 0)
 			goto out;
 		ret = finish_inode_if_needed(sctx, 1);
@@ -7014,7 +7737,7 @@ static int flush_delalloc_roots(struct send_ctx *sctx)
 	int i;
 
 	if (root) {
-		ret = btrfs_start_delalloc_snapshot(root);
+		ret = btrfs_start_delalloc_snapshot(root, false);
 		if (ret)
 			return ret;
 		btrfs_wait_ordered_extents(root, U64_MAX, 0, U64_MAX);
@@ -7022,7 +7745,7 @@ static int flush_delalloc_roots(struct send_ctx *sctx)
 
 	for (i = 0; i < sctx->clone_roots_cnt; i++) {
 		root = sctx->clone_roots[i].root;
-		ret = btrfs_start_delalloc_snapshot(root);
+		ret = btrfs_start_delalloc_snapshot(root, false);
 		if (ret)
 			return ret;
 		btrfs_wait_ordered_extents(root, U64_MAX, 0, U64_MAX);
@@ -7053,20 +7776,18 @@ static void dedupe_in_progress_warn(const struct btrfs_root *root)
 		      root->root_key.objectid, root->dedupe_in_progress);
 }
 
-long btrfs_ioctl_send(struct file *mnt_file, struct btrfs_ioctl_send_args *arg)
+long btrfs_ioctl_send(struct inode *inode, struct btrfs_ioctl_send_args *arg)
 {
 	int ret = 0;
-	struct btrfs_root *send_root = BTRFS_I(file_inode(mnt_file))->root;
+	struct btrfs_root *send_root = BTRFS_I(inode)->root;
 	struct btrfs_fs_info *fs_info = send_root->fs_info;
 	struct btrfs_root *clone_root;
-	struct btrfs_key key;
 	struct send_ctx *sctx = NULL;
 	u32 i;
 	u64 *clone_sources_tmp = NULL;
 	int clone_sources_to_rollback = 0;
-	unsigned alloc_size;
+	size_t alloc_size;
 	int sort_clone_roots = 0;
-	int index;
 
 	if (!capable(CAP_SYS_ADMIN))
 		return -EPERM;
@@ -7104,13 +7825,6 @@ long btrfs_ioctl_send(struct file *mnt_file, struct btrfs_ioctl_send_args *arg)
 		goto out;
 	}
 
-	if (!access_ok(arg->clone_sources,
-			sizeof(*arg->clone_sources) *
-			arg->clone_sources_count)) {
-		ret = -EFAULT;
-		goto out;
-	}
-
 	if (arg->flags & ~BTRFS_SEND_FLAG_MASK) {
 		ret = -EINVAL;
 		goto out;
@@ -7129,6 +7843,21 @@ long btrfs_ioctl_send(struct file *mnt_file, struct btrfs_ioctl_send_args *arg)
 
 	sctx->flags = arg->flags;
 
+	if (arg->flags & BTRFS_SEND_FLAG_VERSION) {
+		if (arg->version > BTRFS_SEND_STREAM_VERSION) {
+			ret = -EPROTO;
+			goto out;
+		}
+		/* Zero means "use the highest version" */
+		sctx->proto = arg->version ?: BTRFS_SEND_STREAM_VERSION;
+	} else {
+		sctx->proto = 1;
+	}
+	if ((arg->flags & BTRFS_SEND_FLAG_COMPRESSED) && sctx->proto < 2) {
+		ret = -EINVAL;
+		goto out;
+	}
+
 	sctx->send_filp = fget(arg->send_fd);
 	if (!sctx->send_filp) {
 		ret = -EBADF;
@@ -7147,15 +7876,32 @@ long btrfs_ioctl_send(struct file *mnt_file, struct btrfs_ioctl_send_args *arg)
 
 	sctx->clone_roots_cnt = arg->clone_sources_count;
 
-	sctx->send_max_size = BTRFS_SEND_BUF_SIZE;
-	sctx->send_buf = kvmalloc(sctx->send_max_size, GFP_KERNEL);
-	if (!sctx->send_buf) {
-		ret = -ENOMEM;
-		goto out;
-	}
+	if (sctx->proto >= 2) {
+		u32 send_buf_num_pages;
 
-	sctx->read_buf = kvmalloc(BTRFS_SEND_READ_SIZE, GFP_KERNEL);
-	if (!sctx->read_buf) {
+		sctx->send_max_size = ALIGN(SZ_16K + BTRFS_MAX_COMPRESSED, PAGE_SIZE);
+		sctx->send_buf = vmalloc(sctx->send_max_size);
+		if (!sctx->send_buf) {
+			ret = -ENOMEM;
+			goto out;
+		}
+		send_buf_num_pages = sctx->send_max_size >> PAGE_SHIFT;
+		sctx->send_buf_pages = kcalloc(send_buf_num_pages,
+					       sizeof(*sctx->send_buf_pages),
+					       GFP_KERNEL);
+		if (!sctx->send_buf_pages) {
+			ret = -ENOMEM;
+			goto out;
+		}
+		for (i = 0; i < send_buf_num_pages; i++) {
+			sctx->send_buf_pages[i] =
+				vmalloc_to_page(sctx->send_buf + (i << PAGE_SHIFT));
+		}
+	} else {
+		sctx->send_max_size = BTRFS_SEND_BUF_SIZE_V1;
+		sctx->send_buf = kvmalloc(sctx->send_max_size, GFP_KERNEL);
+	}
+	if (!sctx->send_buf) {
 		ret = -ENOMEM;
 		goto out;
 	}
@@ -7163,16 +7909,19 @@ long btrfs_ioctl_send(struct file *mnt_file, struct btrfs_ioctl_send_args *arg)
 	sctx->pending_dir_moves = RB_ROOT;
 	sctx->waiting_dir_moves = RB_ROOT;
 	sctx->orphan_dirs = RB_ROOT;
+	sctx->rbtree_new_refs = RB_ROOT;
+	sctx->rbtree_deleted_refs = RB_ROOT;
 
-	alloc_size = sizeof(struct clone_root) * (arg->clone_sources_count + 1);
-
-	sctx->clone_roots = kzalloc(alloc_size, GFP_KERNEL);
+	sctx->clone_roots = kvcalloc(sizeof(*sctx->clone_roots),
+				     arg->clone_sources_count + 1,
+				     GFP_KERNEL);
 	if (!sctx->clone_roots) {
 		ret = -ENOMEM;
 		goto out;
 	}
 
-	alloc_size = arg->clone_sources_count * sizeof(*arg->clone_sources);
+	alloc_size = array_size(sizeof(*arg->clone_sources),
+				arg->clone_sources_count);
 
 	if (arg->clone_sources_count) {
 		clone_sources_tmp = kvmalloc(alloc_size, GFP_KERNEL);
@@ -7189,15 +7938,9 @@ long btrfs_ioctl_send(struct file *mnt_file, struct btrfs_ioctl_send_args *arg)
 		}
 
 		for (i = 0; i < arg->clone_sources_count; i++) {
-			key.objectid = clone_sources_tmp[i];
-			key.type = BTRFS_ROOT_ITEM_KEY;
-			key.offset = (u64)-1;
-
-			index = srcu_read_lock(&fs_info->subvol_srcu);
-
-			clone_root = btrfs_read_fs_root_no_name(fs_info, &key);
+			clone_root = btrfs_get_fs_root(fs_info,
+						clone_sources_tmp[i], true);
 			if (IS_ERR(clone_root)) {
-				srcu_read_unlock(&fs_info->subvol_srcu, index);
 				ret = PTR_ERR(clone_root);
 				goto out;
 			}
@@ -7205,20 +7948,19 @@ long btrfs_ioctl_send(struct file *mnt_file, struct btrfs_ioctl_send_args *arg)
 			if (!btrfs_root_readonly(clone_root) ||
 			    btrfs_root_dead(clone_root)) {
 				spin_unlock(&clone_root->root_item_lock);
-				srcu_read_unlock(&fs_info->subvol_srcu, index);
+				btrfs_put_root(clone_root);
 				ret = -EPERM;
 				goto out;
 			}
 			if (clone_root->dedupe_in_progress) {
 				dedupe_in_progress_warn(clone_root);
 				spin_unlock(&clone_root->root_item_lock);
-				srcu_read_unlock(&fs_info->subvol_srcu, index);
+				btrfs_put_root(clone_root);
 				ret = -EAGAIN;
 				goto out;
 			}
 			clone_root->send_in_progress++;
 			spin_unlock(&clone_root->root_item_lock);
-			srcu_read_unlock(&fs_info->subvol_srcu, index);
 
 			sctx->clone_roots[i].root = clone_root;
 			clone_sources_to_rollback = i + 1;
@@ -7228,15 +7970,9 @@ long btrfs_ioctl_send(struct file *mnt_file, struct btrfs_ioctl_send_args *arg)
 	}
 
 	if (arg->parent_root) {
-		key.objectid = arg->parent_root;
-		key.type = BTRFS_ROOT_ITEM_KEY;
-		key.offset = (u64)-1;
-
-		index = srcu_read_lock(&fs_info->subvol_srcu);
-
-		sctx->parent_root = btrfs_read_fs_root_no_name(fs_info, &key);
+		sctx->parent_root = btrfs_get_fs_root(fs_info, arg->parent_root,
+						      true);
 		if (IS_ERR(sctx->parent_root)) {
-			srcu_read_unlock(&fs_info->subvol_srcu, index);
 			ret = PTR_ERR(sctx->parent_root);
 			goto out;
 		}
@@ -7246,20 +7982,16 @@ long btrfs_ioctl_send(struct file *mnt_file, struct btrfs_ioctl_send_args *arg)
 		if (!btrfs_root_readonly(sctx->parent_root) ||
 				btrfs_root_dead(sctx->parent_root)) {
 			spin_unlock(&sctx->parent_root->root_item_lock);
-			srcu_read_unlock(&fs_info->subvol_srcu, index);
 			ret = -EPERM;
 			goto out;
 		}
 		if (sctx->parent_root->dedupe_in_progress) {
 			dedupe_in_progress_warn(sctx->parent_root);
 			spin_unlock(&sctx->parent_root->root_item_lock);
-			srcu_read_unlock(&fs_info->subvol_srcu, index);
 			ret = -EAGAIN;
 			goto out;
 		}
 		spin_unlock(&sctx->parent_root->root_item_lock);
-
-		srcu_read_unlock(&fs_info->subvol_srcu, index);
 	}
 
 	/*
@@ -7267,7 +7999,8 @@ long btrfs_ioctl_send(struct file *mnt_file, struct btrfs_ioctl_send_args *arg)
 	 * is behind the current send position. This is checked while searching
 	 * for possible clone sources.
 	 */
-	sctx->clone_roots[sctx->clone_roots_cnt++].root = sctx->send_root;
+	sctx->clone_roots[sctx->clone_roots_cnt++].root =
+		btrfs_grab_root(sctx->send_root);
 
 	/* We do a bsearch later */
 	sort(sctx->clone_roots, sctx->clone_roots_cnt,
@@ -7283,23 +8016,7 @@ long btrfs_ioctl_send(struct file *mnt_file, struct btrfs_ioctl_send_args *arg)
 	if (ret)
 		goto out;
 
-	mutex_lock(&fs_info->balance_mutex);
-	if (test_bit(BTRFS_FS_BALANCE_RUNNING, &fs_info->flags)) {
-		mutex_unlock(&fs_info->balance_mutex);
-		btrfs_warn_rl(fs_info,
-		"cannot run send because a balance operation is in progress");
-		ret = -EAGAIN;
-		goto out;
-	}
-	fs_info->send_in_progress++;
-	mutex_unlock(&fs_info->balance_mutex);
-
-	current->journal_info = BTRFS_SEND_TRANS_STUB;
 	ret = send_subvol(sctx);
-	current->journal_info = NULL;
-	mutex_lock(&fs_info->balance_mutex);
-	fs_info->send_in_progress--;
-	mutex_unlock(&fs_info->balance_mutex);
 	if (ret < 0)
 		goto out;
 
@@ -7352,18 +8069,24 @@ out:
 	}
 
 	if (sort_clone_roots) {
-		for (i = 0; i < sctx->clone_roots_cnt; i++)
+		for (i = 0; i < sctx->clone_roots_cnt; i++) {
 			btrfs_root_dec_send_in_progress(
 					sctx->clone_roots[i].root);
+			btrfs_put_root(sctx->clone_roots[i].root);
+		}
 	} else {
-		for (i = 0; sctx && i < clone_sources_to_rollback; i++)
+		for (i = 0; sctx && i < clone_sources_to_rollback; i++) {
 			btrfs_root_dec_send_in_progress(
 					sctx->clone_roots[i].root);
+			btrfs_put_root(sctx->clone_roots[i].root);
+		}
 
 		btrfs_root_dec_send_in_progress(send_root);
 	}
-	if (sctx && !IS_ERR_OR_NULL(sctx->parent_root))
+	if (sctx && !IS_ERR_OR_NULL(sctx->parent_root)) {
 		btrfs_root_dec_send_in_progress(sctx->parent_root);
+		btrfs_put_root(sctx->parent_root);
+	}
 
 	kvfree(clone_sources_tmp);
 
@@ -7372,11 +8095,14 @@ out:
 			fput(sctx->send_filp);
 
 		kvfree(sctx->clone_roots);
+		kfree(sctx->send_buf_pages);
 		kvfree(sctx->send_buf);
-		kvfree(sctx->read_buf);
+		kvfree(sctx->verity_descriptor);
 
 		name_cache_free(sctx);
 
+		close_current_inode(sctx);
+
 		kfree(sctx);
 	}
 
diff --git a/fs/btrfs/send.h b/fs/btrfs/send.h
index ead397f7034f..f7585cfa7e52 100644
--- a/fs/btrfs/send.h
+++ b/fs/btrfs/send.h
@@ -7,13 +7,24 @@
 #ifndef BTRFS_SEND_H
 #define BTRFS_SEND_H
 
-#include "ctree.h"
+#include <linux/types.h>
 
 #define BTRFS_SEND_STREAM_MAGIC "btrfs-stream"
-#define BTRFS_SEND_STREAM_VERSION 1
+/* Conditional support for the upcoming protocol version. */
+#ifdef CONFIG_BTRFS_DEBUG
+#define BTRFS_SEND_STREAM_VERSION 3
+#else
+#define BTRFS_SEND_STREAM_VERSION 2
+#endif
+
+/*
+ * In send stream v1, no command is larger than 64K. In send stream v2, no limit
+ * should be assumed.
+ */
+#define BTRFS_SEND_BUF_SIZE_V1				SZ_64K
 
-#define BTRFS_SEND_BUF_SIZE SZ_64K
-#define BTRFS_SEND_READ_SIZE (48 * SZ_1K)
+struct inode;
+struct btrfs_ioctl_send_args;
 
 enum btrfs_tlv_type {
 	BTRFS_TLV_U8,
@@ -47,80 +58,126 @@ struct btrfs_tlv_header {
 
 /* commands */
 enum btrfs_send_cmd {
-	BTRFS_SEND_C_UNSPEC,
-
-	BTRFS_SEND_C_SUBVOL,
-	BTRFS_SEND_C_SNAPSHOT,
-
-	BTRFS_SEND_C_MKFILE,
-	BTRFS_SEND_C_MKDIR,
-	BTRFS_SEND_C_MKNOD,
-	BTRFS_SEND_C_MKFIFO,
-	BTRFS_SEND_C_MKSOCK,
-	BTRFS_SEND_C_SYMLINK,
-
-	BTRFS_SEND_C_RENAME,
-	BTRFS_SEND_C_LINK,
-	BTRFS_SEND_C_UNLINK,
-	BTRFS_SEND_C_RMDIR,
-
-	BTRFS_SEND_C_SET_XATTR,
-	BTRFS_SEND_C_REMOVE_XATTR,
-
-	BTRFS_SEND_C_WRITE,
-	BTRFS_SEND_C_CLONE,
-
-	BTRFS_SEND_C_TRUNCATE,
-	BTRFS_SEND_C_CHMOD,
-	BTRFS_SEND_C_CHOWN,
-	BTRFS_SEND_C_UTIMES,
-
-	BTRFS_SEND_C_END,
-	BTRFS_SEND_C_UPDATE_EXTENT,
-	__BTRFS_SEND_C_MAX,
+	BTRFS_SEND_C_UNSPEC		= 0,
+
+	/* Version 1 */
+	BTRFS_SEND_C_SUBVOL		= 1,
+	BTRFS_SEND_C_SNAPSHOT		= 2,
+
+	BTRFS_SEND_C_MKFILE		= 3,
+	BTRFS_SEND_C_MKDIR		= 4,
+	BTRFS_SEND_C_MKNOD		= 5,
+	BTRFS_SEND_C_MKFIFO		= 6,
+	BTRFS_SEND_C_MKSOCK		= 7,
+	BTRFS_SEND_C_SYMLINK		= 8,
+
+	BTRFS_SEND_C_RENAME		= 9,
+	BTRFS_SEND_C_LINK		= 10,
+	BTRFS_SEND_C_UNLINK		= 11,
+	BTRFS_SEND_C_RMDIR		= 12,
+
+	BTRFS_SEND_C_SET_XATTR		= 13,
+	BTRFS_SEND_C_REMOVE_XATTR	= 14,
+
+	BTRFS_SEND_C_WRITE		= 15,
+	BTRFS_SEND_C_CLONE		= 16,
+
+	BTRFS_SEND_C_TRUNCATE		= 17,
+	BTRFS_SEND_C_CHMOD		= 18,
+	BTRFS_SEND_C_CHOWN		= 19,
+	BTRFS_SEND_C_UTIMES		= 20,
+
+	BTRFS_SEND_C_END		= 21,
+	BTRFS_SEND_C_UPDATE_EXTENT	= 22,
+	BTRFS_SEND_C_MAX_V1		= 22,
+
+	/* Version 2 */
+	BTRFS_SEND_C_FALLOCATE		= 23,
+	BTRFS_SEND_C_FILEATTR		= 24,
+	BTRFS_SEND_C_ENCODED_WRITE	= 25,
+	BTRFS_SEND_C_MAX_V2		= 25,
+
+	/* Version 3 */
+	BTRFS_SEND_C_ENABLE_VERITY	= 26,
+	BTRFS_SEND_C_MAX_V3		= 26,
+	/* End */
+	BTRFS_SEND_C_MAX		= 26,
 };
-#define BTRFS_SEND_C_MAX (__BTRFS_SEND_C_MAX - 1)
 
 /* attributes in send stream */
 enum {
-	BTRFS_SEND_A_UNSPEC,
-
-	BTRFS_SEND_A_UUID,
-	BTRFS_SEND_A_CTRANSID,
-
-	BTRFS_SEND_A_INO,
-	BTRFS_SEND_A_SIZE,
-	BTRFS_SEND_A_MODE,
-	BTRFS_SEND_A_UID,
-	BTRFS_SEND_A_GID,
-	BTRFS_SEND_A_RDEV,
-	BTRFS_SEND_A_CTIME,
-	BTRFS_SEND_A_MTIME,
-	BTRFS_SEND_A_ATIME,
-	BTRFS_SEND_A_OTIME,
-
-	BTRFS_SEND_A_XATTR_NAME,
-	BTRFS_SEND_A_XATTR_DATA,
-
-	BTRFS_SEND_A_PATH,
-	BTRFS_SEND_A_PATH_TO,
-	BTRFS_SEND_A_PATH_LINK,
-
-	BTRFS_SEND_A_FILE_OFFSET,
-	BTRFS_SEND_A_DATA,
-
-	BTRFS_SEND_A_CLONE_UUID,
-	BTRFS_SEND_A_CLONE_CTRANSID,
-	BTRFS_SEND_A_CLONE_PATH,
-	BTRFS_SEND_A_CLONE_OFFSET,
-	BTRFS_SEND_A_CLONE_LEN,
-
-	__BTRFS_SEND_A_MAX,
+	BTRFS_SEND_A_UNSPEC		= 0,
+
+	/* Version 1 */
+	BTRFS_SEND_A_UUID		= 1,
+	BTRFS_SEND_A_CTRANSID		= 2,
+
+	BTRFS_SEND_A_INO		= 3,
+	BTRFS_SEND_A_SIZE		= 4,
+	BTRFS_SEND_A_MODE		= 5,
+	BTRFS_SEND_A_UID		= 6,
+	BTRFS_SEND_A_GID		= 7,
+	BTRFS_SEND_A_RDEV		= 8,
+	BTRFS_SEND_A_CTIME		= 9,
+	BTRFS_SEND_A_MTIME		= 10,
+	BTRFS_SEND_A_ATIME		= 11,
+	BTRFS_SEND_A_OTIME		= 12,
+
+	BTRFS_SEND_A_XATTR_NAME		= 13,
+	BTRFS_SEND_A_XATTR_DATA		= 14,
+
+	BTRFS_SEND_A_PATH		= 15,
+	BTRFS_SEND_A_PATH_TO		= 16,
+	BTRFS_SEND_A_PATH_LINK		= 17,
+
+	BTRFS_SEND_A_FILE_OFFSET	= 18,
+	/*
+	 * As of send stream v2, this attribute is special: it must be the last
+	 * attribute in a command, its header contains only the type, and its
+	 * length is implicitly the remaining length of the command.
+	 */
+	BTRFS_SEND_A_DATA		= 19,
+
+	BTRFS_SEND_A_CLONE_UUID		= 20,
+	BTRFS_SEND_A_CLONE_CTRANSID	= 21,
+	BTRFS_SEND_A_CLONE_PATH		= 22,
+	BTRFS_SEND_A_CLONE_OFFSET	= 23,
+	BTRFS_SEND_A_CLONE_LEN		= 24,
+
+	BTRFS_SEND_A_MAX_V1		= 24,
+
+	/* Version 2 */
+	BTRFS_SEND_A_FALLOCATE_MODE	= 25,
+
+	/*
+	 * File attributes from the FS_*_FL namespace (i_flags, xflags),
+	 * translated to BTRFS_INODE_* bits (BTRFS_INODE_FLAG_MASK) and stored
+	 * in btrfs_inode_item::flags (represented by btrfs_inode::flags and
+	 * btrfs_inode::ro_flags).
+	 */
+	BTRFS_SEND_A_FILEATTR		= 26,
+
+	BTRFS_SEND_A_UNENCODED_FILE_LEN	= 27,
+	BTRFS_SEND_A_UNENCODED_LEN	= 28,
+	BTRFS_SEND_A_UNENCODED_OFFSET	= 29,
+	/*
+	 * COMPRESSION and ENCRYPTION default to NONE (0) if omitted from
+	 * BTRFS_SEND_C_ENCODED_WRITE.
+	 */
+	BTRFS_SEND_A_COMPRESSION	= 30,
+	BTRFS_SEND_A_ENCRYPTION		= 31,
+	BTRFS_SEND_A_MAX_V2		= 31,
+
+	/* Version 3 */
+	BTRFS_SEND_A_VERITY_ALGORITHM	= 32,
+	BTRFS_SEND_A_VERITY_BLOCK_SIZE	= 33,
+	BTRFS_SEND_A_VERITY_SALT_DATA	= 34,
+	BTRFS_SEND_A_VERITY_SIG_DATA	= 35,
+	BTRFS_SEND_A_MAX_V3		= 35,
+
+	__BTRFS_SEND_A_MAX		= 35,
 };
-#define BTRFS_SEND_A_MAX (__BTRFS_SEND_A_MAX - 1)
 
-#ifdef __KERNEL__
-long btrfs_ioctl_send(struct file *mnt_file, struct btrfs_ioctl_send_args *arg);
-#endif
+long btrfs_ioctl_send(struct inode *inode, struct btrfs_ioctl_send_args *arg);
 
 #endif
diff --git a/fs/btrfs/space-info.c b/fs/btrfs/space-info.c
index 01297c5b2666..f171bf875633 100644
--- a/fs/btrfs/space-info.c
+++ b/fs/btrfs/space-info.c
@@ -9,6 +9,155 @@
 #include "ordered-data.h"
 #include "transaction.h"
 #include "block-group.h"
+#include "zoned.h"
+
+/*
+ * HOW DOES SPACE RESERVATION WORK
+ *
+ * If you want to know about delalloc specifically, there is a separate comment
+ * for that with the delalloc code.  This comment is about how the whole system
+ * works generally.
+ *
+ * BASIC CONCEPTS
+ *
+ *   1) space_info.  This is the ultimate arbiter of how much space we can use.
+ *   There's a description of the bytes_ fields with the struct declaration,
+ *   refer to that for specifics on each field.  Suffice it to say that for
+ *   reservations we care about total_bytes - SUM(space_info->bytes_) when
+ *   determining if there is space to make an allocation.  There is a space_info
+ *   for METADATA, SYSTEM, and DATA areas.
+ *
+ *   2) block_rsv's.  These are basically buckets for every different type of
+ *   metadata reservation we have.  You can see the comment in the block_rsv
+ *   code on the rules for each type, but generally block_rsv->reserved is how
+ *   much space is accounted for in space_info->bytes_may_use.
+ *
+ *   3) btrfs_calc*_size.  These are the worst case calculations we used based
+ *   on the number of items we will want to modify.  We have one for changing
+ *   items, and one for inserting new items.  Generally we use these helpers to
+ *   determine the size of the block reserves, and then use the actual bytes
+ *   values to adjust the space_info counters.
+ *
+ * MAKING RESERVATIONS, THE NORMAL CASE
+ *
+ *   We call into either btrfs_reserve_data_bytes() or
+ *   btrfs_reserve_metadata_bytes(), depending on which we're looking for, with
+ *   num_bytes we want to reserve.
+ *
+ *   ->reserve
+ *     space_info->bytes_may_reserve += num_bytes
+ *
+ *   ->extent allocation
+ *     Call btrfs_add_reserved_bytes() which does
+ *     space_info->bytes_may_reserve -= num_bytes
+ *     space_info->bytes_reserved += extent_bytes
+ *
+ *   ->insert reference
+ *     Call btrfs_update_block_group() which does
+ *     space_info->bytes_reserved -= extent_bytes
+ *     space_info->bytes_used += extent_bytes
+ *
+ * MAKING RESERVATIONS, FLUSHING NORMALLY (non-priority)
+ *
+ *   Assume we are unable to simply make the reservation because we do not have
+ *   enough space
+ *
+ *   -> __reserve_bytes
+ *     create a reserve_ticket with ->bytes set to our reservation, add it to
+ *     the tail of space_info->tickets, kick async flush thread
+ *
+ *   ->handle_reserve_ticket
+ *     wait on ticket->wait for ->bytes to be reduced to 0, or ->error to be set
+ *     on the ticket.
+ *
+ *   -> btrfs_async_reclaim_metadata_space/btrfs_async_reclaim_data_space
+ *     Flushes various things attempting to free up space.
+ *
+ *   -> btrfs_try_granting_tickets()
+ *     This is called by anything that either subtracts space from
+ *     space_info->bytes_may_use, ->bytes_pinned, etc, or adds to the
+ *     space_info->total_bytes.  This loops through the ->priority_tickets and
+ *     then the ->tickets list checking to see if the reservation can be
+ *     completed.  If it can the space is added to space_info->bytes_may_use and
+ *     the ticket is woken up.
+ *
+ *   -> ticket wakeup
+ *     Check if ->bytes == 0, if it does we got our reservation and we can carry
+ *     on, if not return the appropriate error (ENOSPC, but can be EINTR if we
+ *     were interrupted.)
+ *
+ * MAKING RESERVATIONS, FLUSHING HIGH PRIORITY
+ *
+ *   Same as the above, except we add ourselves to the
+ *   space_info->priority_tickets, and we do not use ticket->wait, we simply
+ *   call flush_space() ourselves for the states that are safe for us to call
+ *   without deadlocking and hope for the best.
+ *
+ * THE FLUSHING STATES
+ *
+ *   Generally speaking we will have two cases for each state, a "nice" state
+ *   and a "ALL THE THINGS" state.  In btrfs we delay a lot of work in order to
+ *   reduce the locking over head on the various trees, and even to keep from
+ *   doing any work at all in the case of delayed refs.  Each of these delayed
+ *   things however hold reservations, and so letting them run allows us to
+ *   reclaim space so we can make new reservations.
+ *
+ *   FLUSH_DELAYED_ITEMS
+ *     Every inode has a delayed item to update the inode.  Take a simple write
+ *     for example, we would update the inode item at write time to update the
+ *     mtime, and then again at finish_ordered_io() time in order to update the
+ *     isize or bytes.  We keep these delayed items to coalesce these operations
+ *     into a single operation done on demand.  These are an easy way to reclaim
+ *     metadata space.
+ *
+ *   FLUSH_DELALLOC
+ *     Look at the delalloc comment to get an idea of how much space is reserved
+ *     for delayed allocation.  We can reclaim some of this space simply by
+ *     running delalloc, but usually we need to wait for ordered extents to
+ *     reclaim the bulk of this space.
+ *
+ *   FLUSH_DELAYED_REFS
+ *     We have a block reserve for the outstanding delayed refs space, and every
+ *     delayed ref operation holds a reservation.  Running these is a quick way
+ *     to reclaim space, but we want to hold this until the end because COW can
+ *     churn a lot and we can avoid making some extent tree modifications if we
+ *     are able to delay for as long as possible.
+ *
+ *   ALLOC_CHUNK
+ *     We will skip this the first time through space reservation, because of
+ *     overcommit and we don't want to have a lot of useless metadata space when
+ *     our worst case reservations will likely never come true.
+ *
+ *   RUN_DELAYED_IPUTS
+ *     If we're freeing inodes we're likely freeing checksums, file extent
+ *     items, and extent tree items.  Loads of space could be freed up by these
+ *     operations, however they won't be usable until the transaction commits.
+ *
+ *   COMMIT_TRANS
+ *     This will commit the transaction.  Historically we had a lot of logic
+ *     surrounding whether or not we'd commit the transaction, but this waits born
+ *     out of a pre-tickets era where we could end up committing the transaction
+ *     thousands of times in a row without making progress.  Now thanks to our
+ *     ticketing system we know if we're not making progress and can error
+ *     everybody out after a few commits rather than burning the disk hoping for
+ *     a different answer.
+ *
+ * OVERCOMMIT
+ *
+ *   Because we hold so many reservations for metadata we will allow you to
+ *   reserve more space than is currently free in the currently allocate
+ *   metadata space.  This only happens with metadata, data does not allow
+ *   overcommitting.
+ *
+ *   You can see the current logic for when we allow overcommit in
+ *   btrfs_can_overcommit(), but it only applies to unallocated space.  If there
+ *   is no unallocated space to be had, all reservations are kept within the
+ *   free space in the allocated metadata chunks.
+ *
+ *   Because of overcommitting, you generally want to use the
+ *   btrfs_can_overcommit() logic for metadata allocations, as it does the right
+ *   thing with or without extra unallocated space.
+ */
 
 u64 __pure btrfs_space_info_used(struct btrfs_space_info *s_info,
 			  bool may_use_included)
@@ -16,6 +165,7 @@ u64 __pure btrfs_space_info_used(struct btrfs_space_info *s_info,
 	ASSERT(s_info);
 	return s_info->bytes_used + s_info->bytes_reserved +
 		s_info->bytes_pinned + s_info->bytes_readonly +
+		s_info->bytes_zone_unusable +
 		(may_use_included ? s_info->bytes_may_use : 0);
 }
 
@@ -28,10 +178,45 @@ void btrfs_clear_space_info_full(struct btrfs_fs_info *info)
 	struct list_head *head = &info->space_info;
 	struct btrfs_space_info *found;
 
-	rcu_read_lock();
-	list_for_each_entry_rcu(found, head, list)
+	list_for_each_entry(found, head, list)
 		found->full = 0;
-	rcu_read_unlock();
+}
+
+/*
+ * Block groups with more than this value (percents) of unusable space will be
+ * scheduled for background reclaim.
+ */
+#define BTRFS_DEFAULT_ZONED_RECLAIM_THRESH			(75)
+
+/*
+ * Calculate chunk size depending on volume type (regular or zoned).
+ */
+static u64 calc_chunk_size(const struct btrfs_fs_info *fs_info, u64 flags)
+{
+	if (btrfs_is_zoned(fs_info))
+		return fs_info->zone_size;
+
+	ASSERT(flags & BTRFS_BLOCK_GROUP_TYPE_MASK);
+
+	if (flags & BTRFS_BLOCK_GROUP_DATA)
+		return BTRFS_MAX_DATA_CHUNK_SIZE;
+	else if (flags & BTRFS_BLOCK_GROUP_SYSTEM)
+		return SZ_32M;
+
+	/* Handle BTRFS_BLOCK_GROUP_METADATA */
+	if (fs_info->fs_devices->total_rw_bytes > 50ULL * SZ_1G)
+		return SZ_1G;
+
+	return SZ_256M;
+}
+
+/*
+ * Update default chunk size.
+ */
+void btrfs_update_space_info_chunk_size(struct btrfs_space_info *space_info,
+					u64 chunk_size)
+{
+	WRITE_ONCE(space_info->chunk_size, chunk_size);
 }
 
 static int create_space_info(struct btrfs_fs_info *info, u64 flags)
@@ -45,13 +230,6 @@ static int create_space_info(struct btrfs_fs_info *info, u64 flags)
 	if (!space_info)
 		return -ENOMEM;
 
-	ret = percpu_counter_init(&space_info->total_bytes_pinned, 0,
-				 GFP_KERNEL);
-	if (ret) {
-		kfree(space_info);
-		return ret;
-	}
-
 	for (i = 0; i < BTRFS_NR_RAID_TYPES; i++)
 		INIT_LIST_HEAD(&space_info->block_groups[i]);
 	init_rwsem(&space_info->groups_sem);
@@ -61,12 +239,17 @@ static int create_space_info(struct btrfs_fs_info *info, u64 flags)
 	INIT_LIST_HEAD(&space_info->ro_bgs);
 	INIT_LIST_HEAD(&space_info->tickets);
 	INIT_LIST_HEAD(&space_info->priority_tickets);
+	space_info->clamp = 1;
+	btrfs_update_space_info_chunk_size(space_info, calc_chunk_size(info, flags));
+
+	if (btrfs_is_zoned(info))
+		space_info->bg_reclaim_threshold = BTRFS_DEFAULT_ZONED_RECLAIM_THRESH;
 
 	ret = btrfs_sysfs_add_space_info_type(info, space_info);
 	if (ret)
 		return ret;
 
-	list_add_rcu(&space_info->list, &info->space_info);
+	list_add(&space_info->list, &info->space_info);
 	if (flags & BTRFS_BLOCK_GROUP_DATA)
 		info->data_sinfo = space_info;
 
@@ -110,29 +293,36 @@ out:
 	return ret;
 }
 
-void btrfs_update_space_info(struct btrfs_fs_info *info, u64 flags,
-			     u64 total_bytes, u64 bytes_used,
-			     u64 bytes_readonly,
-			     struct btrfs_space_info **space_info)
+void btrfs_add_bg_to_space_info(struct btrfs_fs_info *info,
+				struct btrfs_block_group *block_group)
 {
 	struct btrfs_space_info *found;
-	int factor;
+	int factor, index;
 
-	factor = btrfs_bg_type_to_factor(flags);
+	factor = btrfs_bg_type_to_factor(block_group->flags);
 
-	found = btrfs_find_space_info(info, flags);
+	found = btrfs_find_space_info(info, block_group->flags);
 	ASSERT(found);
 	spin_lock(&found->lock);
-	found->total_bytes += total_bytes;
-	found->disk_total += total_bytes * factor;
-	found->bytes_used += bytes_used;
-	found->disk_used += bytes_used * factor;
-	found->bytes_readonly += bytes_readonly;
-	if (total_bytes > 0)
+	found->total_bytes += block_group->length;
+	if (test_bit(BLOCK_GROUP_FLAG_ZONE_IS_ACTIVE, &block_group->runtime_flags))
+		found->active_total_bytes += block_group->length;
+	found->disk_total += block_group->length * factor;
+	found->bytes_used += block_group->used;
+	found->disk_used += block_group->used * factor;
+	found->bytes_readonly += block_group->bytes_super;
+	found->bytes_zone_unusable += block_group->zone_unusable;
+	if (block_group->length > 0)
 		found->full = 0;
 	btrfs_try_granting_tickets(info, found);
 	spin_unlock(&found->lock);
-	*space_info = found;
+
+	block_group->space_info = found;
+
+	index = btrfs_bg_flags_to_raid_index(block_group->flags);
+	down_write(&found->groups_sem);
+	list_add_tail(&block_group->list, &found->block_groups[index]);
+	up_write(&found->groups_sem);
 }
 
 struct btrfs_space_info *btrfs_find_space_info(struct btrfs_fs_info *info,
@@ -143,41 +333,26 @@ struct btrfs_space_info *btrfs_find_space_info(struct btrfs_fs_info *info,
 
 	flags &= BTRFS_BLOCK_GROUP_TYPE_MASK;
 
-	rcu_read_lock();
-	list_for_each_entry_rcu(found, head, list) {
-		if (found->flags & flags) {
-			rcu_read_unlock();
+	list_for_each_entry(found, head, list) {
+		if (found->flags & flags)
 			return found;
-		}
 	}
-	rcu_read_unlock();
 	return NULL;
 }
 
-static inline u64 calc_global_rsv_need_space(struct btrfs_block_rsv *global)
-{
-	return (global->size << 1);
-}
-
-int btrfs_can_overcommit(struct btrfs_fs_info *fs_info,
-			 struct btrfs_space_info *space_info, u64 bytes,
-			 enum btrfs_reserve_flush_enum flush)
+static u64 calc_available_free_space(struct btrfs_fs_info *fs_info,
+			  struct btrfs_space_info *space_info,
+			  enum btrfs_reserve_flush_enum flush)
 {
 	u64 profile;
 	u64 avail;
-	u64 used;
 	int factor;
 
-	/* Don't overcommit when in mixed mode. */
-	if (space_info->flags & BTRFS_BLOCK_GROUP_DATA)
-		return 0;
-
 	if (space_info->flags & BTRFS_BLOCK_GROUP_SYSTEM)
 		profile = btrfs_system_alloc_profile(fs_info);
 	else
 		profile = btrfs_metadata_alloc_profile(fs_info);
 
-	used = btrfs_space_info_used(space_info, true);
 	avail = atomic64_read(&fs_info->free_chunk_space);
 
 	/*
@@ -198,12 +373,57 @@ int btrfs_can_overcommit(struct btrfs_fs_info *fs_info,
 		avail >>= 3;
 	else
 		avail >>= 1;
+	return avail;
+}
 
-	if (used + bytes < space_info->total_bytes + avail)
+static inline u64 writable_total_bytes(struct btrfs_fs_info *fs_info,
+				       struct btrfs_space_info *space_info)
+{
+	/*
+	 * On regular filesystem, all total_bytes are always writable. On zoned
+	 * filesystem, there may be a limitation imposed by max_active_zones.
+	 * For metadata allocation, we cannot finish an existing active block
+	 * group to avoid a deadlock. Thus, we need to consider only the active
+	 * groups to be writable for metadata space.
+	 */
+	if (!btrfs_is_zoned(fs_info) || (space_info->flags & BTRFS_BLOCK_GROUP_DATA))
+		return space_info->total_bytes;
+
+	return space_info->active_total_bytes;
+}
+
+int btrfs_can_overcommit(struct btrfs_fs_info *fs_info,
+			 struct btrfs_space_info *space_info, u64 bytes,
+			 enum btrfs_reserve_flush_enum flush)
+{
+	u64 avail;
+	u64 used;
+
+	/* Don't overcommit when in mixed mode */
+	if (space_info->flags & BTRFS_BLOCK_GROUP_DATA)
+		return 0;
+
+	used = btrfs_space_info_used(space_info, true);
+	if (btrfs_is_zoned(fs_info) && (space_info->flags & BTRFS_BLOCK_GROUP_METADATA))
+		avail = 0;
+	else
+		avail = calc_available_free_space(fs_info, space_info, flush);
+
+	if (used + bytes < writable_total_bytes(fs_info, space_info) + avail)
 		return 1;
 	return 0;
 }
 
+static void remove_ticket(struct btrfs_space_info *space_info,
+			  struct reserve_ticket *ticket)
+{
+	if (!list_empty(&ticket->list)) {
+		list_del_init(&ticket->list);
+		ASSERT(space_info->reclaim_size >= ticket->bytes);
+		space_info->reclaim_size -= ticket->bytes;
+	}
+}
+
 /*
  * This is for space we already have accounted in space_info->bytes_may_use, so
  * basically when we're returning space from block_rsv's.
@@ -224,14 +444,14 @@ again:
 
 		ticket = list_first_entry(head, struct reserve_ticket, list);
 
-		/* Check and see if our ticket can be satisified now. */
-		if ((used + ticket->bytes <= space_info->total_bytes) ||
+		/* Check and see if our ticket can be satisfied now. */
+		if ((used + ticket->bytes <= writable_total_bytes(fs_info, space_info)) ||
 		    btrfs_can_overcommit(fs_info, space_info, ticket->bytes,
 					 flush)) {
 			btrfs_space_info_update_bytes_may_use(fs_info,
 							      space_info,
 							      ticket->bytes);
-			list_del_init(&ticket->list);
+			remove_ticket(space_info, ticket);
 			ticket->bytes = 0;
 			space_info->tickets_id++;
 			wake_up(&ticket->wait);
@@ -256,27 +476,47 @@ do {									\
 	spin_unlock(&__rsv->lock);					\
 } while (0)
 
+static const char *space_info_flag_to_str(const struct btrfs_space_info *space_info)
+{
+	switch (space_info->flags) {
+	case BTRFS_BLOCK_GROUP_SYSTEM:
+		return "SYSTEM";
+	case BTRFS_BLOCK_GROUP_METADATA | BTRFS_BLOCK_GROUP_DATA:
+		return "DATA+METADATA";
+	case BTRFS_BLOCK_GROUP_DATA:
+		return "DATA";
+	case BTRFS_BLOCK_GROUP_METADATA:
+		return "METADATA";
+	default:
+		return "UNKNOWN";
+	}
+}
+
+static void dump_global_block_rsv(struct btrfs_fs_info *fs_info)
+{
+	DUMP_BLOCK_RSV(fs_info, global_block_rsv);
+	DUMP_BLOCK_RSV(fs_info, trans_block_rsv);
+	DUMP_BLOCK_RSV(fs_info, chunk_block_rsv);
+	DUMP_BLOCK_RSV(fs_info, delayed_block_rsv);
+	DUMP_BLOCK_RSV(fs_info, delayed_refs_rsv);
+}
+
 static void __btrfs_dump_space_info(struct btrfs_fs_info *fs_info,
 				    struct btrfs_space_info *info)
 {
+	const char *flag_str = space_info_flag_to_str(info);
 	lockdep_assert_held(&info->lock);
 
-	btrfs_info(fs_info, "space_info %llu has %llu free, is %sfull",
-		   info->flags,
-		   info->total_bytes - btrfs_space_info_used(info, true),
+	/* The free space could be negative in case of overcommit */
+	btrfs_info(fs_info, "space_info %s has %lld free, is %sfull",
+		   flag_str,
+		   (s64)(info->total_bytes - btrfs_space_info_used(info, true)),
 		   info->full ? "" : "not ");
 	btrfs_info(fs_info,
-		"space_info total=%llu, used=%llu, pinned=%llu, reserved=%llu, may_use=%llu, readonly=%llu",
+"space_info total=%llu, used=%llu, pinned=%llu, reserved=%llu, may_use=%llu, readonly=%llu zone_unusable=%llu",
 		info->total_bytes, info->bytes_used, info->bytes_pinned,
 		info->bytes_reserved, info->bytes_may_use,
-		info->bytes_readonly);
-
-	DUMP_BLOCK_RSV(fs_info, global_block_rsv);
-	DUMP_BLOCK_RSV(fs_info, trans_block_rsv);
-	DUMP_BLOCK_RSV(fs_info, chunk_block_rsv);
-	DUMP_BLOCK_RSV(fs_info, delayed_block_rsv);
-	DUMP_BLOCK_RSV(fs_info, delayed_refs_rsv);
-
+		info->bytes_readonly, info->bytes_zone_unusable);
 }
 
 void btrfs_dump_space_info(struct btrfs_fs_info *fs_info,
@@ -288,6 +528,7 @@ void btrfs_dump_space_info(struct btrfs_fs_info *fs_info,
 
 	spin_lock(&info->lock);
 	__btrfs_dump_space_info(fs_info, info);
+	dump_global_block_rsv(fs_info);
 	spin_unlock(&info->lock);
 
 	if (!dump_block_groups)
@@ -298,39 +539,18 @@ again:
 	list_for_each_entry(cache, &info->block_groups[index], list) {
 		spin_lock(&cache->lock);
 		btrfs_info(fs_info,
-			"block group %llu has %llu bytes, %llu used %llu pinned %llu reserved %s",
+			"block group %llu has %llu bytes, %llu used %llu pinned %llu reserved %llu zone_unusable %s",
 			cache->start, cache->length, cache->used, cache->pinned,
-			cache->reserved, cache->ro ? "[readonly]" : "");
-		btrfs_dump_free_space(cache, bytes);
+			cache->reserved, cache->zone_unusable,
+			cache->ro ? "[readonly]" : "");
 		spin_unlock(&cache->lock);
+		btrfs_dump_free_space(cache, bytes);
 	}
 	if (++index < BTRFS_NR_RAID_TYPES)
 		goto again;
 	up_read(&info->groups_sem);
 }
 
-static void btrfs_writeback_inodes_sb_nr(struct btrfs_fs_info *fs_info,
-					 unsigned long nr_pages, int nr_items)
-{
-	struct super_block *sb = fs_info->sb;
-
-	if (down_read_trylock(&sb->s_umount)) {
-		writeback_inodes_sb_nr(sb, nr_pages, WB_REASON_FS_FREE_SPACE);
-		up_read(&sb->s_umount);
-	} else {
-		/*
-		 * We needn't worry the filesystem going from r/w to r/o though
-		 * we don't acquire ->s_umount mutex, because the filesystem
-		 * should guarantee the delalloc inodes list be empty after
-		 * the filesystem is readonly(all dirty pages are written to
-		 * the disk).
-		 */
-		btrfs_start_delalloc_roots(fs_info, nr_items);
-		if (!current->journal_info)
-			btrfs_wait_ordered_roots(fs_info, nr_items, 0, (u64)-1);
-	}
-}
-
 static inline u64 calc_reclaim_items_nr(struct btrfs_fs_info *fs_info,
 					u64 to_reclaim)
 {
@@ -349,86 +569,100 @@ static inline u64 calc_reclaim_items_nr(struct btrfs_fs_info *fs_info,
 /*
  * shrink metadata reservation for delalloc
  */
-static void shrink_delalloc(struct btrfs_fs_info *fs_info, u64 to_reclaim,
-			    u64 orig, bool wait_ordered)
+static void shrink_delalloc(struct btrfs_fs_info *fs_info,
+			    struct btrfs_space_info *space_info,
+			    u64 to_reclaim, bool wait_ordered,
+			    bool for_preempt)
 {
-	struct btrfs_space_info *space_info;
 	struct btrfs_trans_handle *trans;
 	u64 delalloc_bytes;
-	u64 dio_bytes;
-	u64 async_pages;
+	u64 ordered_bytes;
 	u64 items;
 	long time_left;
-	unsigned long nr_pages;
 	int loops;
 
-	/* Calc the number of the pages we need flush for space reservation */
-	items = calc_reclaim_items_nr(fs_info, to_reclaim);
-	to_reclaim = items * EXTENT_SIZE_PER_ITEM;
-
-	trans = (struct btrfs_trans_handle *)current->journal_info;
-	space_info = btrfs_find_space_info(fs_info, BTRFS_BLOCK_GROUP_METADATA);
-
-	delalloc_bytes = percpu_counter_sum_positive(
-						&fs_info->delalloc_bytes);
-	dio_bytes = percpu_counter_sum_positive(&fs_info->dio_bytes);
-	if (delalloc_bytes == 0 && dio_bytes == 0) {
-		if (trans)
-			return;
-		if (wait_ordered)
-			btrfs_wait_ordered_roots(fs_info, items, 0, (u64)-1);
+	delalloc_bytes = percpu_counter_sum_positive(&fs_info->delalloc_bytes);
+	ordered_bytes = percpu_counter_sum_positive(&fs_info->ordered_bytes);
+	if (delalloc_bytes == 0 && ordered_bytes == 0)
 		return;
+
+	/* Calc the number of the pages we need flush for space reservation */
+	if (to_reclaim == U64_MAX) {
+		items = U64_MAX;
+	} else {
+		/*
+		 * to_reclaim is set to however much metadata we need to
+		 * reclaim, but reclaiming that much data doesn't really track
+		 * exactly.  What we really want to do is reclaim full inode's
+		 * worth of reservations, however that's not available to us
+		 * here.  We will take a fraction of the delalloc bytes for our
+		 * flushing loops and hope for the best.  Delalloc will expand
+		 * the amount we write to cover an entire dirty extent, which
+		 * will reclaim the metadata reservation for that range.  If
+		 * it's not enough subsequent flush stages will be more
+		 * aggressive.
+		 */
+		to_reclaim = max(to_reclaim, delalloc_bytes >> 3);
+		items = calc_reclaim_items_nr(fs_info, to_reclaim) * 2;
 	}
 
+	trans = current->journal_info;
+
 	/*
 	 * If we are doing more ordered than delalloc we need to just wait on
 	 * ordered extents, otherwise we'll waste time trying to flush delalloc
 	 * that likely won't give us the space back we need.
 	 */
-	if (dio_bytes > delalloc_bytes)
+	if (ordered_bytes > delalloc_bytes && !for_preempt)
 		wait_ordered = true;
 
 	loops = 0;
-	while ((delalloc_bytes || dio_bytes) && loops < 3) {
-		nr_pages = min(delalloc_bytes, to_reclaim) >> PAGE_SHIFT;
+	while ((delalloc_bytes || ordered_bytes) && loops < 3) {
+		u64 temp = min(delalloc_bytes, to_reclaim) >> PAGE_SHIFT;
+		long nr_pages = min_t(u64, temp, LONG_MAX);
+		int async_pages;
 
-		/*
-		 * Triggers inode writeback for up to nr_pages. This will invoke
-		 * ->writepages callback and trigger delalloc filling
-		 *  (btrfs_run_delalloc_range()).
-		 */
-		btrfs_writeback_inodes_sb_nr(fs_info, nr_pages, items);
+		btrfs_start_delalloc_roots(fs_info, nr_pages, true);
 
 		/*
-		 * We need to wait for the compressed pages to start before
-		 * we continue.
+		 * We need to make sure any outstanding async pages are now
+		 * processed before we continue.  This is because things like
+		 * sync_inode() try to be smart and skip writing if the inode is
+		 * marked clean.  We don't use filemap_fwrite for flushing
+		 * because we want to control how many pages we write out at a
+		 * time, thus this is the only safe way to make sure we've
+		 * waited for outstanding compressed workers to have started
+		 * their jobs and thus have ordered extents set up properly.
+		 *
+		 * This exists because we do not want to wait for each
+		 * individual inode to finish its async work, we simply want to
+		 * start the IO on everybody, and then come back here and wait
+		 * for all of the async work to catch up.  Once we're done with
+		 * that we know we'll have ordered extents for everything and we
+		 * can decide if we wait for that or not.
+		 *
+		 * If we choose to replace this in the future, make absolutely
+		 * sure that the proper waiting is being done in the async case,
+		 * as there have been bugs in that area before.
 		 */
 		async_pages = atomic_read(&fs_info->async_delalloc_pages);
 		if (!async_pages)
 			goto skip_async;
 
 		/*
-		 * Calculate how many compressed pages we want to be written
-		 * before we continue. I.e if there are more async pages than we
-		 * require wait_event will wait until nr_pages are written.
+		 * We don't want to wait forever, if we wrote less pages in this
+		 * loop than we have outstanding, only wait for that number of
+		 * pages, otherwise we can wait for all async pages to finish
+		 * before continuing.
 		 */
-		if (async_pages <= nr_pages)
-			async_pages = 0;
-		else
+		if (async_pages > nr_pages)
 			async_pages -= nr_pages;
-
+		else
+			async_pages = 0;
 		wait_event(fs_info->async_submit_wait,
 			   atomic_read(&fs_info->async_delalloc_pages) <=
-			   (int)async_pages);
+			   async_pages);
 skip_async:
-		spin_lock(&space_info->lock);
-		if (list_empty(&space_info->tickets) &&
-		    list_empty(&space_info->priority_tickets)) {
-			spin_unlock(&space_info->lock);
-			break;
-		}
-		spin_unlock(&space_info->lock);
-
 		loops++;
 		if (wait_ordered && !trans) {
 			btrfs_wait_ordered_roots(fs_info, items, 0, (u64)-1);
@@ -437,104 +671,28 @@ skip_async:
 			if (time_left)
 				break;
 		}
-		delalloc_bytes = percpu_counter_sum_positive(
-						&fs_info->delalloc_bytes);
-		dio_bytes = percpu_counter_sum_positive(&fs_info->dio_bytes);
-	}
-}
-
-/**
- * maybe_commit_transaction - possibly commit the transaction if its ok to
- * @root - the root we're allocating for
- * @bytes - the number of bytes we want to reserve
- * @force - force the commit
- *
- * This will check to make sure that committing the transaction will actually
- * get us somewhere and then commit the transaction if it does.  Otherwise it
- * will return -ENOSPC.
- */
-static int may_commit_transaction(struct btrfs_fs_info *fs_info,
-				  struct btrfs_space_info *space_info)
-{
-	struct reserve_ticket *ticket = NULL;
-	struct btrfs_block_rsv *delayed_rsv = &fs_info->delayed_block_rsv;
-	struct btrfs_block_rsv *delayed_refs_rsv = &fs_info->delayed_refs_rsv;
-	struct btrfs_trans_handle *trans;
-	u64 bytes_needed;
-	u64 reclaim_bytes = 0;
-	u64 cur_free_bytes = 0;
-
-	trans = (struct btrfs_trans_handle *)current->journal_info;
-	if (trans)
-		return -EAGAIN;
-
-	spin_lock(&space_info->lock);
-	cur_free_bytes = btrfs_space_info_used(space_info, true);
-	if (cur_free_bytes < space_info->total_bytes)
-		cur_free_bytes = space_info->total_bytes - cur_free_bytes;
-	else
-		cur_free_bytes = 0;
-
-	if (!list_empty(&space_info->priority_tickets))
-		ticket = list_first_entry(&space_info->priority_tickets,
-					  struct reserve_ticket, list);
-	else if (!list_empty(&space_info->tickets))
-		ticket = list_first_entry(&space_info->tickets,
-					  struct reserve_ticket, list);
-	bytes_needed = (ticket) ? ticket->bytes : 0;
-
-	if (bytes_needed > cur_free_bytes)
-		bytes_needed -= cur_free_bytes;
-	else
-		bytes_needed = 0;
-	spin_unlock(&space_info->lock);
-
-	if (!bytes_needed)
-		return 0;
 
-	trans = btrfs_join_transaction(fs_info->extent_root);
-	if (IS_ERR(trans))
-		return PTR_ERR(trans);
+		/*
+		 * If we are for preemption we just want a one-shot of delalloc
+		 * flushing so we can stop flushing if we decide we don't need
+		 * to anymore.
+		 */
+		if (for_preempt)
+			break;
 
-	/*
-	 * See if there is enough pinned space to make this reservation, or if
-	 * we have block groups that are going to be freed, allowing us to
-	 * possibly do a chunk allocation the next loop through.
-	 */
-	if (test_bit(BTRFS_TRANS_HAVE_FREE_BGS, &trans->transaction->flags) ||
-	    __percpu_counter_compare(&space_info->total_bytes_pinned,
-				     bytes_needed,
-				     BTRFS_TOTAL_BYTES_PINNED_BATCH) >= 0)
-		goto commit;
+		spin_lock(&space_info->lock);
+		if (list_empty(&space_info->tickets) &&
+		    list_empty(&space_info->priority_tickets)) {
+			spin_unlock(&space_info->lock);
+			break;
+		}
+		spin_unlock(&space_info->lock);
 
-	/*
-	 * See if there is some space in the delayed insertion reservation for
-	 * this reservation.
-	 */
-	if (space_info != delayed_rsv->space_info)
-		goto enospc;
-
-	spin_lock(&delayed_rsv->lock);
-	reclaim_bytes += delayed_rsv->reserved;
-	spin_unlock(&delayed_rsv->lock);
-
-	spin_lock(&delayed_refs_rsv->lock);
-	reclaim_bytes += delayed_refs_rsv->reserved;
-	spin_unlock(&delayed_refs_rsv->lock);
-	if (reclaim_bytes >= bytes_needed)
-		goto commit;
-	bytes_needed -= reclaim_bytes;
-
-	if (__percpu_counter_compare(&space_info->total_bytes_pinned,
-				   bytes_needed,
-				   BTRFS_TOTAL_BYTES_PINNED_BATCH) < 0)
-		goto enospc;
-
-commit:
-	return btrfs_commit_transaction(trans);
-enospc:
-	btrfs_end_transaction(trans);
-	return -ENOSPC;
+		delalloc_bytes = percpu_counter_sum_positive(
+						&fs_info->delalloc_bytes);
+		ordered_bytes = percpu_counter_sum_positive(
+						&fs_info->ordered_bytes);
+	}
 }
 
 /*
@@ -544,9 +702,9 @@ enospc:
  */
 static void flush_space(struct btrfs_fs_info *fs_info,
 		       struct btrfs_space_info *space_info, u64 num_bytes,
-		       int state)
+		       enum btrfs_flush_state state, bool for_preempt)
 {
-	struct btrfs_root *root = fs_info->extent_root;
+	struct btrfs_root *root = fs_info->tree_root;
 	struct btrfs_trans_handle *trans;
 	int nr;
 	int ret = 0;
@@ -569,8 +727,11 @@ static void flush_space(struct btrfs_fs_info *fs_info,
 		break;
 	case FLUSH_DELALLOC:
 	case FLUSH_DELALLOC_WAIT:
-		shrink_delalloc(fs_info, num_bytes * 2, num_bytes,
-				state == FLUSH_DELALLOC_WAIT);
+	case FLUSH_DELALLOC_FULL:
+		if (state == FLUSH_DELALLOC_FULL)
+			num_bytes = U64_MAX;
+		shrink_delalloc(fs_info, space_info, num_bytes,
+				state != FLUSH_DELALLOC, for_preempt);
 		break;
 	case FLUSH_DELAYED_REFS_NR:
 	case FLUSH_DELAYED_REFS:
@@ -588,16 +749,45 @@ static void flush_space(struct btrfs_fs_info *fs_info,
 		break;
 	case ALLOC_CHUNK:
 	case ALLOC_CHUNK_FORCE:
+		/*
+		 * For metadata space on zoned filesystem, reaching here means we
+		 * don't have enough space left in active_total_bytes. Try to
+		 * activate a block group first, because we may have inactive
+		 * block group already allocated.
+		 */
+		ret = btrfs_zoned_activate_one_bg(fs_info, space_info, false);
+		if (ret < 0)
+			break;
+		else if (ret == 1)
+			break;
+
 		trans = btrfs_join_transaction(root);
 		if (IS_ERR(trans)) {
 			ret = PTR_ERR(trans);
 			break;
 		}
 		ret = btrfs_chunk_alloc(trans,
-				btrfs_metadata_alloc_profile(fs_info),
+				btrfs_get_alloc_profile(fs_info, space_info->flags),
 				(state == ALLOC_CHUNK) ? CHUNK_ALLOC_NO_FORCE :
 					CHUNK_ALLOC_FORCE);
 		btrfs_end_transaction(trans);
+
+		/*
+		 * For metadata space on zoned filesystem, allocating a new chunk
+		 * is not enough. We still need to activate the block * group.
+		 * Active the newly allocated block group by (maybe) finishing
+		 * a block group.
+		 */
+		if (ret == 1) {
+			ret = btrfs_zoned_activate_one_bg(fs_info, space_info, true);
+			/*
+			 * Revert to the original ret regardless we could finish
+			 * one block group or not.
+			 */
+			if (ret >= 0)
+				ret = 1;
+		}
+
 		if (ret > 0 || ret == -ENOSPC)
 			ret = 0;
 		break;
@@ -611,7 +801,13 @@ static void flush_space(struct btrfs_fs_info *fs_info,
 		btrfs_wait_on_delayed_iputs(fs_info);
 		break;
 	case COMMIT_TRANS:
-		ret = may_commit_transaction(fs_info, space_info);
+		ASSERT(current->journal_info == NULL);
+		trans = btrfs_join_transaction(root);
+		if (IS_ERR(trans)) {
+			ret = PTR_ERR(trans);
+			break;
+		}
+		ret = btrfs_commit_transaction(trans);
 		break;
 	default:
 		ret = -ENOSPC;
@@ -619,7 +815,7 @@ static void flush_space(struct btrfs_fs_info *fs_info,
 	}
 
 	trace_btrfs_flush_space(fs_info, space_info->flags, num_bytes, state,
-				ret);
+				ret, for_preempt);
 	return;
 }
 
@@ -627,57 +823,174 @@ static inline u64
 btrfs_calc_reclaim_metadata_size(struct btrfs_fs_info *fs_info,
 				 struct btrfs_space_info *space_info)
 {
-	struct reserve_ticket *ticket;
 	u64 used;
-	u64 expected;
-	u64 to_reclaim = 0;
-
-	list_for_each_entry(ticket, &space_info->tickets, list)
-		to_reclaim += ticket->bytes;
-	list_for_each_entry(ticket, &space_info->priority_tickets, list)
-		to_reclaim += ticket->bytes;
-	if (to_reclaim)
-		return to_reclaim;
-
-	to_reclaim = min_t(u64, num_online_cpus() * SZ_1M, SZ_16M);
-	if (btrfs_can_overcommit(fs_info, space_info, to_reclaim,
-				 BTRFS_RESERVE_FLUSH_ALL))
-		return 0;
+	u64 avail;
+	u64 total;
+	u64 to_reclaim = space_info->reclaim_size;
+
+	lockdep_assert_held(&space_info->lock);
 
+	avail = calc_available_free_space(fs_info, space_info,
+					  BTRFS_RESERVE_FLUSH_ALL);
 	used = btrfs_space_info_used(space_info, true);
 
-	if (btrfs_can_overcommit(fs_info, space_info, SZ_1M,
-				 BTRFS_RESERVE_FLUSH_ALL))
-		expected = div_factor_fine(space_info->total_bytes, 95);
-	else
-		expected = div_factor_fine(space_info->total_bytes, 90);
+	/*
+	 * We may be flushing because suddenly we have less space than we had
+	 * before, and now we're well over-committed based on our current free
+	 * space.  If that's the case add in our overage so we make sure to put
+	 * appropriate pressure on the flushing state machine.
+	 */
+	total = writable_total_bytes(fs_info, space_info);
+	if (total + avail < used)
+		to_reclaim += used - (total + avail);
 
-	if (used > expected)
-		to_reclaim = used - expected;
-	else
-		to_reclaim = 0;
-	to_reclaim = min(to_reclaim, space_info->bytes_may_use +
-				     space_info->bytes_reserved);
 	return to_reclaim;
 }
 
-static inline int need_do_async_reclaim(struct btrfs_fs_info *fs_info,
-					struct btrfs_space_info *space_info,
-					u64 used)
+static bool need_preemptive_reclaim(struct btrfs_fs_info *fs_info,
+				    struct btrfs_space_info *space_info)
 {
-	u64 thresh = div_factor_fine(space_info->total_bytes, 98);
+	u64 global_rsv_size = fs_info->global_block_rsv.reserved;
+	u64 ordered, delalloc;
+	u64 total = writable_total_bytes(fs_info, space_info);
+	u64 thresh;
+	u64 used;
+
+	thresh = div_factor_fine(total, 90);
+
+	lockdep_assert_held(&space_info->lock);
 
 	/* If we're just plain full then async reclaim just slows us down. */
-	if ((space_info->bytes_used + space_info->bytes_reserved) >= thresh)
-		return 0;
+	if ((space_info->bytes_used + space_info->bytes_reserved +
+	     global_rsv_size) >= thresh)
+		return false;
 
-	if (!btrfs_calc_reclaim_metadata_size(fs_info, space_info))
-		return 0;
+	used = space_info->bytes_may_use + space_info->bytes_pinned;
+
+	/* The total flushable belongs to the global rsv, don't flush. */
+	if (global_rsv_size >= used)
+		return false;
+
+	/*
+	 * 128MiB is 1/4 of the maximum global rsv size.  If we have less than
+	 * that devoted to other reservations then there's no sense in flushing,
+	 * we don't have a lot of things that need flushing.
+	 */
+	if (used - global_rsv_size <= SZ_128M)
+		return false;
+
+	/*
+	 * We have tickets queued, bail so we don't compete with the async
+	 * flushers.
+	 */
+	if (space_info->reclaim_size)
+		return false;
+
+	/*
+	 * If we have over half of the free space occupied by reservations or
+	 * pinned then we want to start flushing.
+	 *
+	 * We do not do the traditional thing here, which is to say
+	 *
+	 *   if (used >= ((total_bytes + avail) / 2))
+	 *     return 1;
+	 *
+	 * because this doesn't quite work how we want.  If we had more than 50%
+	 * of the space_info used by bytes_used and we had 0 available we'd just
+	 * constantly run the background flusher.  Instead we want it to kick in
+	 * if our reclaimable space exceeds our clamped free space.
+	 *
+	 * Our clamping range is 2^1 -> 2^8.  Practically speaking that means
+	 * the following:
+	 *
+	 * Amount of RAM        Minimum threshold       Maximum threshold
+	 *
+	 *        256GiB                     1GiB                  128GiB
+	 *        128GiB                   512MiB                   64GiB
+	 *         64GiB                   256MiB                   32GiB
+	 *         32GiB                   128MiB                   16GiB
+	 *         16GiB                    64MiB                    8GiB
+	 *
+	 * These are the range our thresholds will fall in, corresponding to how
+	 * much delalloc we need for the background flusher to kick in.
+	 */
+
+	thresh = calc_available_free_space(fs_info, space_info,
+					   BTRFS_RESERVE_FLUSH_ALL);
+	used = space_info->bytes_used + space_info->bytes_reserved +
+	       space_info->bytes_readonly + global_rsv_size;
+	if (used < total)
+		thresh += total - used;
+	thresh >>= space_info->clamp;
+
+	used = space_info->bytes_pinned;
+
+	/*
+	 * If we have more ordered bytes than delalloc bytes then we're either
+	 * doing a lot of DIO, or we simply don't have a lot of delalloc waiting
+	 * around.  Preemptive flushing is only useful in that it can free up
+	 * space before tickets need to wait for things to finish.  In the case
+	 * of ordered extents, preemptively waiting on ordered extents gets us
+	 * nothing, if our reservations are tied up in ordered extents we'll
+	 * simply have to slow down writers by forcing them to wait on ordered
+	 * extents.
+	 *
+	 * In the case that ordered is larger than delalloc, only include the
+	 * block reserves that we would actually be able to directly reclaim
+	 * from.  In this case if we're heavy on metadata operations this will
+	 * clearly be heavy enough to warrant preemptive flushing.  In the case
+	 * of heavy DIO or ordered reservations, preemptive flushing will just
+	 * waste time and cause us to slow down.
+	 *
+	 * We want to make sure we truly are maxed out on ordered however, so
+	 * cut ordered in half, and if it's still higher than delalloc then we
+	 * can keep flushing.  This is to avoid the case where we start
+	 * flushing, and now delalloc == ordered and we stop preemptively
+	 * flushing when we could still have several gigs of delalloc to flush.
+	 */
+	ordered = percpu_counter_read_positive(&fs_info->ordered_bytes) >> 1;
+	delalloc = percpu_counter_read_positive(&fs_info->delalloc_bytes);
+	if (ordered >= delalloc)
+		used += fs_info->delayed_refs_rsv.reserved +
+			fs_info->delayed_block_rsv.reserved;
+	else
+		used += space_info->bytes_may_use - global_rsv_size;
 
 	return (used >= thresh && !btrfs_fs_closing(fs_info) &&
 		!test_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state));
 }
 
+static bool steal_from_global_rsv(struct btrfs_fs_info *fs_info,
+				  struct btrfs_space_info *space_info,
+				  struct reserve_ticket *ticket)
+{
+	struct btrfs_block_rsv *global_rsv = &fs_info->global_block_rsv;
+	u64 min_bytes;
+
+	if (!ticket->steal)
+		return false;
+
+	if (global_rsv->space_info != space_info)
+		return false;
+
+	spin_lock(&global_rsv->lock);
+	min_bytes = div_factor(global_rsv->size, 1);
+	if (global_rsv->reserved < min_bytes + ticket->bytes) {
+		spin_unlock(&global_rsv->lock);
+		return false;
+	}
+	global_rsv->reserved -= ticket->bytes;
+	remove_ticket(space_info, ticket);
+	ticket->bytes = 0;
+	wake_up(&ticket->wait);
+	space_info->tickets_id++;
+	if (global_rsv->reserved < global_rsv->size)
+		global_rsv->full = 0;
+	spin_unlock(&global_rsv->lock);
+
+	return true;
+}
+
 /*
  * maybe_fail_all_tickets - we've exhausted our flushing, start failing tickets
  * @fs_info - fs_info for this fs
@@ -698,7 +1011,9 @@ static bool maybe_fail_all_tickets(struct btrfs_fs_info *fs_info,
 {
 	struct reserve_ticket *ticket;
 	u64 tickets_id = space_info->tickets_id;
-	u64 first_ticket_bytes = 0;
+	const bool aborted = BTRFS_FS_ERROR(fs_info);
+
+	trace_btrfs_fail_all_tickets(fs_info, space_info);
 
 	if (btrfs_test_opt(fs_info, ENOSPC_DEBUG)) {
 		btrfs_info(fs_info, "cannot satisfy tickets, dumping space info");
@@ -710,27 +1025,18 @@ static bool maybe_fail_all_tickets(struct btrfs_fs_info *fs_info,
 		ticket = list_first_entry(&space_info->tickets,
 					  struct reserve_ticket, list);
 
-		/*
-		 * may_commit_transaction will avoid committing the transaction
-		 * if it doesn't feel like the space reclaimed by the commit
-		 * would result in the ticket succeeding.  However if we have a
-		 * smaller ticket in the queue it may be small enough to be
-		 * satisified by committing the transaction, so if any
-		 * subsequent ticket is smaller than the first ticket go ahead
-		 * and send us back for another loop through the enospc flushing
-		 * code.
-		 */
-		if (first_ticket_bytes == 0)
-			first_ticket_bytes = ticket->bytes;
-		else if (first_ticket_bytes > ticket->bytes)
+		if (!aborted && steal_from_global_rsv(fs_info, space_info, ticket))
 			return true;
 
-		if (btrfs_test_opt(fs_info, ENOSPC_DEBUG))
+		if (!aborted && btrfs_test_opt(fs_info, ENOSPC_DEBUG))
 			btrfs_info(fs_info, "failing ticket with %llu bytes",
 				   ticket->bytes);
 
-		list_del_init(&ticket->list);
-		ticket->error = -ENOSPC;
+		remove_ticket(space_info, ticket);
+		if (aborted)
+			ticket->error = -EIO;
+		else
+			ticket->error = -ENOSPC;
 		wake_up(&ticket->wait);
 
 		/*
@@ -739,7 +1045,8 @@ static bool maybe_fail_all_tickets(struct btrfs_fs_info *fs_info,
 		 * here to see if we can make progress with the next ticket in
 		 * the list.
 		 */
-		btrfs_try_granting_tickets(fs_info, space_info);
+		if (!aborted)
+			btrfs_try_granting_tickets(fs_info, space_info);
 	}
 	return (tickets_id != space_info->tickets_id);
 }
@@ -754,7 +1061,7 @@ static void btrfs_async_reclaim_metadata_space(struct work_struct *work)
 	struct btrfs_fs_info *fs_info;
 	struct btrfs_space_info *space_info;
 	u64 to_reclaim;
-	int flush_state;
+	enum btrfs_flush_state flush_state;
 	int commit_cycles = 0;
 	u64 last_tickets_id;
 
@@ -773,7 +1080,7 @@ static void btrfs_async_reclaim_metadata_space(struct work_struct *work)
 
 	flush_state = FLUSH_DELAYED_ITEMS_NR;
 	do {
-		flush_space(fs_info, space_info, to_reclaim, flush_state);
+		flush_space(fs_info, space_info, to_reclaim, flush_state, false);
 		spin_lock(&space_info->lock);
 		if (list_empty(&space_info->tickets)) {
 			space_info->flush = 0;
@@ -792,6 +1099,14 @@ static void btrfs_async_reclaim_metadata_space(struct work_struct *work)
 		}
 
 		/*
+		 * We do not want to empty the system of delalloc unless we're
+		 * under heavy pressure, so allow one trip through the flushing
+		 * logic before we start doing a FLUSH_DELALLOC_FULL.
+		 */
+		if (flush_state == FLUSH_DELALLOC_FULL && !commit_cycles)
+			flush_state++;
+
+		/*
 		 * We don't want to force a chunk allocation until we've tried
 		 * pretty hard to reclaim space.  Think of the case where we
 		 * freed up a bunch of space and so have a lot of pinned space
@@ -821,9 +1136,229 @@ static void btrfs_async_reclaim_metadata_space(struct work_struct *work)
 	} while (flush_state <= COMMIT_TRANS);
 }
 
-void btrfs_init_async_reclaim_work(struct work_struct *work)
+/*
+ * This handles pre-flushing of metadata space before we get to the point that
+ * we need to start blocking threads on tickets.  The logic here is different
+ * from the other flush paths because it doesn't rely on tickets to tell us how
+ * much we need to flush, instead it attempts to keep us below the 80% full
+ * watermark of space by flushing whichever reservation pool is currently the
+ * largest.
+ */
+static void btrfs_preempt_reclaim_metadata_space(struct work_struct *work)
+{
+	struct btrfs_fs_info *fs_info;
+	struct btrfs_space_info *space_info;
+	struct btrfs_block_rsv *delayed_block_rsv;
+	struct btrfs_block_rsv *delayed_refs_rsv;
+	struct btrfs_block_rsv *global_rsv;
+	struct btrfs_block_rsv *trans_rsv;
+	int loops = 0;
+
+	fs_info = container_of(work, struct btrfs_fs_info,
+			       preempt_reclaim_work);
+	space_info = btrfs_find_space_info(fs_info, BTRFS_BLOCK_GROUP_METADATA);
+	delayed_block_rsv = &fs_info->delayed_block_rsv;
+	delayed_refs_rsv = &fs_info->delayed_refs_rsv;
+	global_rsv = &fs_info->global_block_rsv;
+	trans_rsv = &fs_info->trans_block_rsv;
+
+	spin_lock(&space_info->lock);
+	while (need_preemptive_reclaim(fs_info, space_info)) {
+		enum btrfs_flush_state flush;
+		u64 delalloc_size = 0;
+		u64 to_reclaim, block_rsv_size;
+		u64 global_rsv_size = global_rsv->reserved;
+
+		loops++;
+
+		/*
+		 * We don't have a precise counter for the metadata being
+		 * reserved for delalloc, so we'll approximate it by subtracting
+		 * out the block rsv's space from the bytes_may_use.  If that
+		 * amount is higher than the individual reserves, then we can
+		 * assume it's tied up in delalloc reservations.
+		 */
+		block_rsv_size = global_rsv_size +
+			delayed_block_rsv->reserved +
+			delayed_refs_rsv->reserved +
+			trans_rsv->reserved;
+		if (block_rsv_size < space_info->bytes_may_use)
+			delalloc_size = space_info->bytes_may_use - block_rsv_size;
+
+		/*
+		 * We don't want to include the global_rsv in our calculation,
+		 * because that's space we can't touch.  Subtract it from the
+		 * block_rsv_size for the next checks.
+		 */
+		block_rsv_size -= global_rsv_size;
+
+		/*
+		 * We really want to avoid flushing delalloc too much, as it
+		 * could result in poor allocation patterns, so only flush it if
+		 * it's larger than the rest of the pools combined.
+		 */
+		if (delalloc_size > block_rsv_size) {
+			to_reclaim = delalloc_size;
+			flush = FLUSH_DELALLOC;
+		} else if (space_info->bytes_pinned >
+			   (delayed_block_rsv->reserved +
+			    delayed_refs_rsv->reserved)) {
+			to_reclaim = space_info->bytes_pinned;
+			flush = COMMIT_TRANS;
+		} else if (delayed_block_rsv->reserved >
+			   delayed_refs_rsv->reserved) {
+			to_reclaim = delayed_block_rsv->reserved;
+			flush = FLUSH_DELAYED_ITEMS_NR;
+		} else {
+			to_reclaim = delayed_refs_rsv->reserved;
+			flush = FLUSH_DELAYED_REFS_NR;
+		}
+
+		spin_unlock(&space_info->lock);
+
+		/*
+		 * We don't want to reclaim everything, just a portion, so scale
+		 * down the to_reclaim by 1/4.  If it takes us down to 0,
+		 * reclaim 1 items worth.
+		 */
+		to_reclaim >>= 2;
+		if (!to_reclaim)
+			to_reclaim = btrfs_calc_insert_metadata_size(fs_info, 1);
+		flush_space(fs_info, space_info, to_reclaim, flush, true);
+		cond_resched();
+		spin_lock(&space_info->lock);
+	}
+
+	/* We only went through once, back off our clamping. */
+	if (loops == 1 && !space_info->reclaim_size)
+		space_info->clamp = max(1, space_info->clamp - 1);
+	trace_btrfs_done_preemptive_reclaim(fs_info, space_info);
+	spin_unlock(&space_info->lock);
+}
+
+/*
+ * FLUSH_DELALLOC_WAIT:
+ *   Space is freed from flushing delalloc in one of two ways.
+ *
+ *   1) compression is on and we allocate less space than we reserved
+ *   2) we are overwriting existing space
+ *
+ *   For #1 that extra space is reclaimed as soon as the delalloc pages are
+ *   COWed, by way of btrfs_add_reserved_bytes() which adds the actual extent
+ *   length to ->bytes_reserved, and subtracts the reserved space from
+ *   ->bytes_may_use.
+ *
+ *   For #2 this is trickier.  Once the ordered extent runs we will drop the
+ *   extent in the range we are overwriting, which creates a delayed ref for
+ *   that freed extent.  This however is not reclaimed until the transaction
+ *   commits, thus the next stages.
+ *
+ * RUN_DELAYED_IPUTS
+ *   If we are freeing inodes, we want to make sure all delayed iputs have
+ *   completed, because they could have been on an inode with i_nlink == 0, and
+ *   thus have been truncated and freed up space.  But again this space is not
+ *   immediately re-usable, it comes in the form of a delayed ref, which must be
+ *   run and then the transaction must be committed.
+ *
+ * COMMIT_TRANS
+ *   This is where we reclaim all of the pinned space generated by running the
+ *   iputs
+ *
+ * ALLOC_CHUNK_FORCE
+ *   For data we start with alloc chunk force, however we could have been full
+ *   before, and then the transaction commit could have freed new block groups,
+ *   so if we now have space to allocate do the force chunk allocation.
+ */
+static const enum btrfs_flush_state data_flush_states[] = {
+	FLUSH_DELALLOC_FULL,
+	RUN_DELAYED_IPUTS,
+	COMMIT_TRANS,
+	ALLOC_CHUNK_FORCE,
+};
+
+static void btrfs_async_reclaim_data_space(struct work_struct *work)
+{
+	struct btrfs_fs_info *fs_info;
+	struct btrfs_space_info *space_info;
+	u64 last_tickets_id;
+	enum btrfs_flush_state flush_state = 0;
+
+	fs_info = container_of(work, struct btrfs_fs_info, async_data_reclaim_work);
+	space_info = fs_info->data_sinfo;
+
+	spin_lock(&space_info->lock);
+	if (list_empty(&space_info->tickets)) {
+		space_info->flush = 0;
+		spin_unlock(&space_info->lock);
+		return;
+	}
+	last_tickets_id = space_info->tickets_id;
+	spin_unlock(&space_info->lock);
+
+	while (!space_info->full) {
+		flush_space(fs_info, space_info, U64_MAX, ALLOC_CHUNK_FORCE, false);
+		spin_lock(&space_info->lock);
+		if (list_empty(&space_info->tickets)) {
+			space_info->flush = 0;
+			spin_unlock(&space_info->lock);
+			return;
+		}
+
+		/* Something happened, fail everything and bail. */
+		if (BTRFS_FS_ERROR(fs_info))
+			goto aborted_fs;
+		last_tickets_id = space_info->tickets_id;
+		spin_unlock(&space_info->lock);
+	}
+
+	while (flush_state < ARRAY_SIZE(data_flush_states)) {
+		flush_space(fs_info, space_info, U64_MAX,
+			    data_flush_states[flush_state], false);
+		spin_lock(&space_info->lock);
+		if (list_empty(&space_info->tickets)) {
+			space_info->flush = 0;
+			spin_unlock(&space_info->lock);
+			return;
+		}
+
+		if (last_tickets_id == space_info->tickets_id) {
+			flush_state++;
+		} else {
+			last_tickets_id = space_info->tickets_id;
+			flush_state = 0;
+		}
+
+		if (flush_state >= ARRAY_SIZE(data_flush_states)) {
+			if (space_info->full) {
+				if (maybe_fail_all_tickets(fs_info, space_info))
+					flush_state = 0;
+				else
+					space_info->flush = 0;
+			} else {
+				flush_state = 0;
+			}
+
+			/* Something happened, fail everything and bail. */
+			if (BTRFS_FS_ERROR(fs_info))
+				goto aborted_fs;
+
+		}
+		spin_unlock(&space_info->lock);
+	}
+	return;
+
+aborted_fs:
+	maybe_fail_all_tickets(fs_info, space_info);
+	space_info->flush = 0;
+	spin_unlock(&space_info->lock);
+}
+
+void btrfs_init_async_reclaim_work(struct btrfs_fs_info *fs_info)
 {
-	INIT_WORK(work, btrfs_async_reclaim_metadata_space);
+	INIT_WORK(&fs_info->async_reclaim_work, btrfs_async_reclaim_metadata_space);
+	INIT_WORK(&fs_info->async_data_reclaim_work, btrfs_async_reclaim_data_space);
+	INIT_WORK(&fs_info->preempt_reclaim_work,
+		  btrfs_preempt_reclaim_metadata_space);
 }
 
 static const enum btrfs_flush_state priority_flush_states[] = {
@@ -839,6 +1374,7 @@ static const enum btrfs_flush_state evict_flush_states[] = {
 	FLUSH_DELAYED_REFS,
 	FLUSH_DELALLOC,
 	FLUSH_DELALLOC_WAIT,
+	FLUSH_DELALLOC_FULL,
 	ALLOC_CHUNK,
 	COMMIT_TRANS,
 };
@@ -850,27 +1386,74 @@ static void priority_reclaim_metadata_space(struct btrfs_fs_info *fs_info,
 				int states_nr)
 {
 	u64 to_reclaim;
-	int flush_state;
+	int flush_state = 0;
 
 	spin_lock(&space_info->lock);
 	to_reclaim = btrfs_calc_reclaim_metadata_size(fs_info, space_info);
-	if (!to_reclaim) {
+	/*
+	 * This is the priority reclaim path, so to_reclaim could be >0 still
+	 * because we may have only satisfied the priority tickets and still
+	 * left non priority tickets on the list.  We would then have
+	 * to_reclaim but ->bytes == 0.
+	 */
+	if (ticket->bytes == 0) {
 		spin_unlock(&space_info->lock);
 		return;
 	}
-	spin_unlock(&space_info->lock);
 
-	flush_state = 0;
-	do {
-		flush_space(fs_info, space_info, to_reclaim, states[flush_state]);
+	while (flush_state < states_nr) {
+		spin_unlock(&space_info->lock);
+		flush_space(fs_info, space_info, to_reclaim, states[flush_state],
+			    false);
 		flush_state++;
 		spin_lock(&space_info->lock);
 		if (ticket->bytes == 0) {
 			spin_unlock(&space_info->lock);
 			return;
 		}
+	}
+
+	/* Attempt to steal from the global rsv if we can. */
+	if (!steal_from_global_rsv(fs_info, space_info, ticket)) {
+		ticket->error = -ENOSPC;
+		remove_ticket(space_info, ticket);
+	}
+
+	/*
+	 * We must run try_granting_tickets here because we could be a large
+	 * ticket in front of a smaller ticket that can now be satisfied with
+	 * the available space.
+	 */
+	btrfs_try_granting_tickets(fs_info, space_info);
+	spin_unlock(&space_info->lock);
+}
+
+static void priority_reclaim_data_space(struct btrfs_fs_info *fs_info,
+					struct btrfs_space_info *space_info,
+					struct reserve_ticket *ticket)
+{
+	spin_lock(&space_info->lock);
+
+	/* We could have been granted before we got here. */
+	if (ticket->bytes == 0) {
 		spin_unlock(&space_info->lock);
-	} while (flush_state < states_nr);
+		return;
+	}
+
+	while (!space_info->full) {
+		spin_unlock(&space_info->lock);
+		flush_space(fs_info, space_info, U64_MAX, ALLOC_CHUNK_FORCE, false);
+		spin_lock(&space_info->lock);
+		if (ticket->bytes == 0) {
+			spin_unlock(&space_info->lock);
+			return;
+		}
+	}
+
+	ticket->error = -ENOSPC;
+	remove_ticket(space_info, ticket);
+	btrfs_try_granting_tickets(fs_info, space_info);
+	spin_unlock(&space_info->lock);
 }
 
 static void wait_reserve_ticket(struct btrfs_fs_info *fs_info,
@@ -893,7 +1476,7 @@ static void wait_reserve_ticket(struct btrfs_fs_info *fs_info,
 			 * despite getting an error, resulting in a space leak
 			 * (bytes_may_use counter of our space_info).
 			 */
-			list_del_init(&ticket->list);
+			remove_ticket(space_info, ticket);
 			ticket->error = -EINTR;
 			break;
 		}
@@ -908,11 +1491,14 @@ static void wait_reserve_ticket(struct btrfs_fs_info *fs_info,
 }
 
 /**
- * handle_reserve_ticket - do the appropriate flushing and waiting for a ticket
- * @fs_info - the fs
- * @space_info - the space_info for the reservation
- * @ticket - the ticket for the reservation
- * @flush - how much we can flush
+ * Do the appropriate flushing and waiting for a ticket
+ *
+ * @fs_info:    the filesystem
+ * @space_info: space info for the reservation
+ * @ticket:     ticket for the reservation
+ * @start_ns:   timestamp when the reservation started
+ * @orig_bytes: amount of bytes originally reserved
+ * @flush:      how much we can flush
  *
  * This does the work of figuring out how to flush for the ticket, waiting for
  * the reservation, and returning the appropriate error if there is one.
@@ -920,12 +1506,15 @@ static void wait_reserve_ticket(struct btrfs_fs_info *fs_info,
 static int handle_reserve_ticket(struct btrfs_fs_info *fs_info,
 				 struct btrfs_space_info *space_info,
 				 struct reserve_ticket *ticket,
+				 u64 start_ns, u64 orig_bytes,
 				 enum btrfs_reserve_flush_enum flush)
 {
 	int ret;
 
 	switch (flush) {
+	case BTRFS_RESERVE_FLUSH_DATA:
 	case BTRFS_RESERVE_FLUSH_ALL:
+	case BTRFS_RESERVE_FLUSH_ALL_STEAL:
 		wait_reserve_ticket(fs_info, space_info, ticket);
 		break;
 	case BTRFS_RESERVE_FLUSH_LIMIT:
@@ -938,24 +1527,15 @@ static int handle_reserve_ticket(struct btrfs_fs_info *fs_info,
 						evict_flush_states,
 						ARRAY_SIZE(evict_flush_states));
 		break;
+	case BTRFS_RESERVE_FLUSH_FREE_SPACE_INODE:
+		priority_reclaim_data_space(fs_info, space_info, ticket);
+		break;
 	default:
 		ASSERT(0);
 		break;
 	}
 
-	spin_lock(&space_info->lock);
 	ret = ticket->error;
-	if (ticket->bytes || ticket->error) {
-		/*
-		 * Need to delete here for priority tickets. For regular tickets
-		 * either the async reclaim job deletes the ticket from the list
-		 * or we delete it ourselves at wait_reserve_ticket().
-		 */
-		list_del_init(&ticket->list);
-		if (!ret)
-			ret = -ENOSPC;
-	}
-	spin_unlock(&space_info->lock);
 	ASSERT(list_empty(&ticket->list));
 	/*
 	 * Check that we can't have an error set if the reservation succeeded,
@@ -964,15 +1544,52 @@ static int handle_reserve_ticket(struct btrfs_fs_info *fs_info,
 	 * space wasn't reserved at all).
 	 */
 	ASSERT(!(ticket->bytes == 0 && ticket->error));
+	trace_btrfs_reserve_ticket(fs_info, space_info->flags, orig_bytes,
+				   start_ns, flush, ticket->error);
 	return ret;
 }
 
+/*
+ * This returns true if this flush state will go through the ordinary flushing
+ * code.
+ */
+static inline bool is_normal_flushing(enum btrfs_reserve_flush_enum flush)
+{
+	return	(flush == BTRFS_RESERVE_FLUSH_ALL) ||
+		(flush == BTRFS_RESERVE_FLUSH_ALL_STEAL);
+}
+
+static inline void maybe_clamp_preempt(struct btrfs_fs_info *fs_info,
+				       struct btrfs_space_info *space_info)
+{
+	u64 ordered = percpu_counter_sum_positive(&fs_info->ordered_bytes);
+	u64 delalloc = percpu_counter_sum_positive(&fs_info->delalloc_bytes);
+
+	/*
+	 * If we're heavy on ordered operations then clamping won't help us.  We
+	 * need to clamp specifically to keep up with dirty'ing buffered
+	 * writers, because there's not a 1:1 correlation of writing delalloc
+	 * and freeing space, like there is with flushing delayed refs or
+	 * delayed nodes.  If we're already more ordered than delalloc then
+	 * we're keeping up, otherwise we aren't and should probably clamp.
+	 */
+	if (ordered < delalloc)
+		space_info->clamp = min(space_info->clamp + 1, 8);
+}
+
+static inline bool can_steal(enum btrfs_reserve_flush_enum flush)
+{
+	return (flush == BTRFS_RESERVE_FLUSH_ALL_STEAL ||
+		flush == BTRFS_RESERVE_FLUSH_EVICT);
+}
+
 /**
- * reserve_metadata_bytes - try to reserve bytes from the block_rsv's space
- * @root - the root we're allocating for
- * @space_info - the space info we want to allocate from
- * @orig_bytes - the number of bytes we want
- * @flush - whether or not we can flush to make our reservation
+ * Try to reserve bytes from the block_rsv's space
+ *
+ * @fs_info:    the filesystem
+ * @space_info: space info we want to allocate from
+ * @orig_bytes: number of bytes we want
+ * @flush:      whether or not we can flush to make our reservation
  *
  * This will reserve orig_bytes number of bytes from the space info associated
  * with the block_rsv.  If there is not enough space it will make an attempt to
@@ -981,12 +1598,13 @@ static int handle_reserve_ticket(struct btrfs_fs_info *fs_info,
  * regain reservations will be made and this will fail if there is not enough
  * space already.
  */
-static int __reserve_metadata_bytes(struct btrfs_fs_info *fs_info,
-				    struct btrfs_space_info *space_info,
-				    u64 orig_bytes,
-				    enum btrfs_reserve_flush_enum flush)
+static int __reserve_bytes(struct btrfs_fs_info *fs_info,
+			   struct btrfs_space_info *space_info, u64 orig_bytes,
+			   enum btrfs_reserve_flush_enum flush)
 {
+	struct work_struct *async_work;
 	struct reserve_ticket ticket;
+	u64 start_ns = 0;
 	u64 used;
 	int ret = 0;
 	bool pending_tickets;
@@ -994,18 +1612,32 @@ static int __reserve_metadata_bytes(struct btrfs_fs_info *fs_info,
 	ASSERT(orig_bytes);
 	ASSERT(!current->journal_info || flush != BTRFS_RESERVE_FLUSH_ALL);
 
+	if (flush == BTRFS_RESERVE_FLUSH_DATA)
+		async_work = &fs_info->async_data_reclaim_work;
+	else
+		async_work = &fs_info->async_reclaim_work;
+
 	spin_lock(&space_info->lock);
 	ret = -ENOSPC;
 	used = btrfs_space_info_used(space_info, true);
-	pending_tickets = !list_empty(&space_info->tickets) ||
-		!list_empty(&space_info->priority_tickets);
+
+	/*
+	 * We don't want NO_FLUSH allocations to jump everybody, they can
+	 * generally handle ENOSPC in a different way, so treat them the same as
+	 * normal flushers when it comes to skipping pending tickets.
+	 */
+	if (is_normal_flushing(flush) || (flush == BTRFS_RESERVE_NO_FLUSH))
+		pending_tickets = !list_empty(&space_info->tickets) ||
+			!list_empty(&space_info->priority_tickets);
+	else
+		pending_tickets = !list_empty(&space_info->priority_tickets);
 
 	/*
 	 * Carry on if we have enough space (short-circuit) OR call
 	 * can_overcommit() to ensure we can overcommit to continue.
 	 */
 	if (!pending_tickets &&
-	    ((used + orig_bytes <= space_info->total_bytes) ||
+	    ((used + orig_bytes <= writable_total_bytes(fs_info, space_info)) ||
 	     btrfs_can_overcommit(fs_info, space_info, orig_bytes, flush))) {
 		btrfs_space_info_update_bytes_may_use(fs_info, space_info,
 						      orig_bytes);
@@ -1022,51 +1654,67 @@ static int __reserve_metadata_bytes(struct btrfs_fs_info *fs_info,
 	if (ret && flush != BTRFS_RESERVE_NO_FLUSH) {
 		ticket.bytes = orig_bytes;
 		ticket.error = 0;
+		space_info->reclaim_size += ticket.bytes;
 		init_waitqueue_head(&ticket.wait);
-		if (flush == BTRFS_RESERVE_FLUSH_ALL) {
+		ticket.steal = can_steal(flush);
+		if (trace_btrfs_reserve_ticket_enabled())
+			start_ns = ktime_get_ns();
+
+		if (flush == BTRFS_RESERVE_FLUSH_ALL ||
+		    flush == BTRFS_RESERVE_FLUSH_ALL_STEAL ||
+		    flush == BTRFS_RESERVE_FLUSH_DATA) {
 			list_add_tail(&ticket.list, &space_info->tickets);
 			if (!space_info->flush) {
+				/*
+				 * We were forced to add a reserve ticket, so
+				 * our preemptive flushing is unable to keep
+				 * up.  Clamp down on the threshold for the
+				 * preemptive flushing in order to keep up with
+				 * the workload.
+				 */
+				maybe_clamp_preempt(fs_info, space_info);
+
 				space_info->flush = 1;
 				trace_btrfs_trigger_flush(fs_info,
 							  space_info->flags,
 							  orig_bytes, flush,
 							  "enospc");
-				queue_work(system_unbound_wq,
-					   &fs_info->async_reclaim_work);
+				queue_work(system_unbound_wq, async_work);
 			}
 		} else {
 			list_add_tail(&ticket.list,
 				      &space_info->priority_tickets);
 		}
 	} else if (!ret && space_info->flags & BTRFS_BLOCK_GROUP_METADATA) {
-		used += orig_bytes;
 		/*
 		 * We will do the space reservation dance during log replay,
 		 * which means we won't have fs_info->fs_root set, so don't do
 		 * the async reclaim as we will panic.
 		 */
 		if (!test_bit(BTRFS_FS_LOG_RECOVERING, &fs_info->flags) &&
-		    need_do_async_reclaim(fs_info, space_info, used) &&
-		    !work_busy(&fs_info->async_reclaim_work)) {
+		    !work_busy(&fs_info->preempt_reclaim_work) &&
+		    need_preemptive_reclaim(fs_info, space_info)) {
 			trace_btrfs_trigger_flush(fs_info, space_info->flags,
 						  orig_bytes, flush, "preempt");
 			queue_work(system_unbound_wq,
-				   &fs_info->async_reclaim_work);
+				   &fs_info->preempt_reclaim_work);
 		}
 	}
 	spin_unlock(&space_info->lock);
 	if (!ret || flush == BTRFS_RESERVE_NO_FLUSH)
 		return ret;
 
-	return handle_reserve_ticket(fs_info, space_info, &ticket, flush);
+	return handle_reserve_ticket(fs_info, space_info, &ticket, start_ns,
+				     orig_bytes, flush);
 }
 
 /**
- * reserve_metadata_bytes - try to reserve bytes from the block_rsv's space
- * @root - the root we're allocating for
- * @block_rsv - the block_rsv we're allocating for
- * @orig_bytes - the number of bytes we want
- * @flush - whether or not we can flush to make our reservation
+ * Trye to reserve metadata bytes from the block_rsv's space
+ *
+ * @fs_info:    the filesystem
+ * @block_rsv:  block_rsv we're allocating for
+ * @orig_bytes: number of bytes we want
+ * @flush:      whether or not we can flush to make our reservation
  *
  * This will reserve orig_bytes number of bytes from the space info associated
  * with the block_rsv.  If there is not enough space it will make an attempt to
@@ -1075,23 +1723,14 @@ static int __reserve_metadata_bytes(struct btrfs_fs_info *fs_info,
  * regain reservations will be made and this will fail if there is not enough
  * space already.
  */
-int btrfs_reserve_metadata_bytes(struct btrfs_root *root,
+int btrfs_reserve_metadata_bytes(struct btrfs_fs_info *fs_info,
 				 struct btrfs_block_rsv *block_rsv,
 				 u64 orig_bytes,
 				 enum btrfs_reserve_flush_enum flush)
 {
-	struct btrfs_fs_info *fs_info = root->fs_info;
-	struct btrfs_block_rsv *global_rsv = &fs_info->global_block_rsv;
 	int ret;
 
-	ret = __reserve_metadata_bytes(fs_info, block_rsv->space_info,
-				       orig_bytes, flush);
-	if (ret == -ENOSPC &&
-	    unlikely(root->orphan_cleanup_state == ORPHAN_CLEANUP_STARTED)) {
-		if (block_rsv != global_rsv &&
-		    !btrfs_block_rsv_use_bytes(global_rsv, orig_bytes))
-			ret = 0;
-	}
+	ret = __reserve_bytes(fs_info, block_rsv->space_info, orig_bytes, flush);
 	if (ret == -ENOSPC) {
 		trace_btrfs_space_reservation(fs_info, "space_info:enospc",
 					      block_rsv->space_info->flags,
@@ -1103,3 +1742,48 @@ int btrfs_reserve_metadata_bytes(struct btrfs_root *root,
 	}
 	return ret;
 }
+
+/**
+ * Try to reserve data bytes for an allocation
+ *
+ * @fs_info: the filesystem
+ * @bytes:   number of bytes we need
+ * @flush:   how we are allowed to flush
+ *
+ * This will reserve bytes from the data space info.  If there is not enough
+ * space then we will attempt to flush space as specified by flush.
+ */
+int btrfs_reserve_data_bytes(struct btrfs_fs_info *fs_info, u64 bytes,
+			     enum btrfs_reserve_flush_enum flush)
+{
+	struct btrfs_space_info *data_sinfo = fs_info->data_sinfo;
+	int ret;
+
+	ASSERT(flush == BTRFS_RESERVE_FLUSH_DATA ||
+	       flush == BTRFS_RESERVE_FLUSH_FREE_SPACE_INODE ||
+	       flush == BTRFS_RESERVE_NO_FLUSH);
+	ASSERT(!current->journal_info || flush != BTRFS_RESERVE_FLUSH_DATA);
+
+	ret = __reserve_bytes(fs_info, data_sinfo, bytes, flush);
+	if (ret == -ENOSPC) {
+		trace_btrfs_space_reservation(fs_info, "space_info:enospc",
+					      data_sinfo->flags, bytes, 1);
+		if (btrfs_test_opt(fs_info, ENOSPC_DEBUG))
+			btrfs_dump_space_info(fs_info, data_sinfo, bytes, 0);
+	}
+	return ret;
+}
+
+/* Dump all the space infos when we abort a transaction due to ENOSPC. */
+__cold void btrfs_dump_space_info_for_trans_abort(struct btrfs_fs_info *fs_info)
+{
+	struct btrfs_space_info *space_info;
+
+	btrfs_info(fs_info, "dumping space info:");
+	list_for_each_entry(space_info, &fs_info->space_info, list) {
+		spin_lock(&space_info->lock);
+		__btrfs_dump_space_info(fs_info, space_info);
+		spin_unlock(&space_info->lock);
+	}
+	dump_global_block_rsv(fs_info);
+}
diff --git a/fs/btrfs/space-info.h b/fs/btrfs/space-info.h
index 24514cd2c6c1..ce66023a9eb8 100644
--- a/fs/btrfs/space-info.h
+++ b/fs/btrfs/space-info.h
@@ -3,6 +3,8 @@
 #ifndef BTRFS_SPACE_INFO_H
 #define BTRFS_SPACE_INFO_H
 
+#include "volumes.h"
+
 struct btrfs_space_info {
 	spinlock_t lock;
 
@@ -17,10 +19,26 @@ struct btrfs_space_info {
 	u64 bytes_may_use;	/* number of bytes that may be used for
 				   delalloc/allocations */
 	u64 bytes_readonly;	/* total bytes that are read only */
+	/* Total bytes in the space, but only accounts active block groups. */
+	u64 active_total_bytes;
+	u64 bytes_zone_unusable;	/* total bytes that are unusable until
+					   resetting the device zone */
 
 	u64 max_extent_size;	/* This will hold the maximum extent size of
 				   the space info if we had an ENOSPC in the
 				   allocator. */
+	/* Chunk size in bytes */
+	u64 chunk_size;
+
+	/*
+	 * Once a block group drops below this threshold (percents) we'll
+	 * schedule it for reclaim.
+	 */
+	int bg_reclaim_threshold;
+
+	int clamp;		/* Used to scale our threshold for preemptive
+				   flushing. The value is >> clamp, so turns
+				   out to be a 2^clamp divisor. */
 
 	unsigned int full:1;	/* indicates that we cannot allocate any more
 				   chunks for this space */
@@ -37,23 +55,18 @@ struct btrfs_space_info {
 
 	u64 flags;
 
-	/*
-	 * bytes_pinned is kept in line with what is actually pinned, as in
-	 * we've called update_block_group and dropped the bytes_used counter
-	 * and increased the bytes_pinned counter.  However this means that
-	 * bytes_pinned does not reflect the bytes that will be pinned once the
-	 * delayed refs are flushed, so this counter is inc'ed every time we
-	 * call btrfs_free_extent so it is a realtime count of what will be
-	 * freed once the transaction is committed.  It will be zeroed every
-	 * time the transaction commits.
-	 */
-	struct percpu_counter total_bytes_pinned;
-
 	struct list_head list;
 	/* Protected by the spinlock 'lock'. */
 	struct list_head ro_bgs;
 	struct list_head priority_tickets;
 	struct list_head tickets;
+
+	/*
+	 * Size of space that needs to be reclaimed in order to satisfy pending
+	 * tickets
+	 */
+	u64 reclaim_size;
+
 	/*
 	 * tickets_id just indicates the next ticket will be handled, so note
 	 * it's not stored per ticket.
@@ -71,6 +84,7 @@ struct btrfs_space_info {
 struct reserve_ticket {
 	u64 bytes;
 	int error;
+	bool steal;
 	struct list_head list;
 	wait_queue_head_t wait;
 };
@@ -109,10 +123,10 @@ DECLARE_SPACE_INFO_UPDATE(bytes_may_use, "space_info");
 DECLARE_SPACE_INFO_UPDATE(bytes_pinned, "pinned");
 
 int btrfs_init_space_info(struct btrfs_fs_info *fs_info);
-void btrfs_update_space_info(struct btrfs_fs_info *info, u64 flags,
-			     u64 total_bytes, u64 bytes_used,
-			     u64 bytes_readonly,
-			     struct btrfs_space_info **space_info);
+void btrfs_add_bg_to_space_info(struct btrfs_fs_info *info,
+				struct btrfs_block_group *block_group);
+void btrfs_update_space_info_chunk_size(struct btrfs_space_info *space_info,
+					u64 chunk_size);
 struct btrfs_space_info *btrfs_find_space_info(struct btrfs_fs_info *info,
 					       u64 flags);
 u64 __pure btrfs_space_info_used(struct btrfs_space_info *s_info,
@@ -121,7 +135,7 @@ void btrfs_clear_space_info_full(struct btrfs_fs_info *info);
 void btrfs_dump_space_info(struct btrfs_fs_info *fs_info,
 			   struct btrfs_space_info *info, u64 bytes,
 			   int dump_block_groups);
-int btrfs_reserve_metadata_bytes(struct btrfs_root *root,
+int btrfs_reserve_metadata_bytes(struct btrfs_fs_info *fs_info,
 				 struct btrfs_block_rsv *block_rsv,
 				 u64 orig_bytes,
 				 enum btrfs_reserve_flush_enum flush);
@@ -141,5 +155,9 @@ static inline void btrfs_space_info_free_bytes_may_use(
 	btrfs_try_granting_tickets(fs_info, space_info);
 	spin_unlock(&space_info->lock);
 }
+int btrfs_reserve_data_bytes(struct btrfs_fs_info *fs_info, u64 bytes,
+			     enum btrfs_reserve_flush_enum flush);
+void btrfs_dump_space_info_for_trans_abort(struct btrfs_fs_info *fs_info);
+void btrfs_init_async_reclaim_work(struct btrfs_fs_info *fs_info);
 
 #endif /* BTRFS_SPACE_INFO_H */
diff --git a/fs/btrfs/struct-funcs.c b/fs/btrfs/struct-funcs.c
index 73f7987143df..12455b2b41de 100644
--- a/fs/btrfs/struct-funcs.c
+++ b/fs/btrfs/struct-funcs.c
@@ -7,161 +7,149 @@
 
 #include "ctree.h"
 
-static inline u8 get_unaligned_le8(const void *p)
+static bool check_setget_bounds(const struct extent_buffer *eb,
+				const void *ptr, unsigned off, int size)
 {
-       return *(u8 *)p;
-}
+	const unsigned long member_offset = (unsigned long)ptr + off;
 
-static inline void put_unaligned_le8(u8 val, void *p)
-{
-       *(u8 *)p = val;
+	if (unlikely(member_offset + size > eb->len)) {
+		btrfs_warn(eb->fs_info,
+		"bad eb member %s: ptr 0x%lx start %llu member offset %lu size %d",
+			(member_offset > eb->len ? "start" : "end"),
+			(unsigned long)ptr, eb->start, member_offset, size);
+		return false;
+	}
+
+	return true;
 }
 
 /*
- * this is some deeply nasty code.
+ * Macro templates that define helpers to read/write extent buffer data of a
+ * given size, that are also used via ctree.h for access to item members by
+ * specialized helpers.
  *
- * The end result is that anyone who #includes ctree.h gets a
- * declaration for the btrfs_set_foo functions and btrfs_foo functions,
- * which are wrappers of btrfs_set_token_#bits functions and
- * btrfs_get_token_#bits functions, which are defined in this file.
+ * Generic helpers:
+ * - btrfs_set_8 (for 8/16/32/64)
+ * - btrfs_get_8 (for 8/16/32/64)
  *
- * These setget functions do all the extent_buffer related mapping
- * required to efficiently read and write specific fields in the extent
- * buffers.  Every pointer to metadata items in btrfs is really just
- * an unsigned long offset into the extent buffer which has been
- * cast to a specific type.  This gives us all the gcc type checking.
+ * Generic helpers with a token (cached address of the most recently accessed
+ * page):
+ * - btrfs_set_token_8 (for 8/16/32/64)
+ * - btrfs_get_token_8 (for 8/16/32/64)
  *
- * The extent buffer api is used to do the page spanning work required to
- * have a metadata blocksize different from the page size.
+ * The set/get functions handle data spanning two pages transparently, in case
+ * metadata block size is larger than page.  Every pointer to metadata items is
+ * an offset into the extent buffer page array, cast to a specific type.  This
+ * gives us all the type checking.
  *
- * There are 2 variants defined, one with a token pointer and one without.
+ * The extent buffer pages stored in the array pages do not form a contiguous
+ * phyusical range, but the API functions assume the linear offset to the range
+ * from 0 to metadata node size.
  */
 
 #define DEFINE_BTRFS_SETGET_BITS(bits)					\
-u##bits btrfs_get_token_##bits(const struct extent_buffer *eb,		\
-			       const void *ptr, unsigned long off,	\
-			       struct btrfs_map_token *token)		\
+u##bits btrfs_get_token_##bits(struct btrfs_map_token *token,		\
+			       const void *ptr, unsigned long off)	\
 {									\
-	unsigned long part_offset = (unsigned long)ptr;			\
-	unsigned long offset = part_offset + off;			\
-	void *p;							\
-	int err;							\
-	char *kaddr;							\
-	unsigned long map_start;					\
-	unsigned long map_len;						\
-	int size = sizeof(u##bits);					\
-	u##bits res;							\
+	const unsigned long member_offset = (unsigned long)ptr + off;	\
+	const unsigned long idx = get_eb_page_index(member_offset);	\
+	const unsigned long oip = get_eb_offset_in_page(token->eb,	\
+							member_offset);	\
+	const int size = sizeof(u##bits);				\
+	u8 lebytes[sizeof(u##bits)];					\
+	const int part = PAGE_SIZE - oip;				\
 									\
 	ASSERT(token);							\
-	ASSERT(token->eb == eb);					\
-									\
-	if (token->kaddr && token->offset <= offset &&			\
-	   (token->offset + PAGE_SIZE >= offset + size)) {	\
-		kaddr = token->kaddr;					\
-		p = kaddr + part_offset - token->offset;		\
-		res = get_unaligned_le##bits(p + off);			\
-		return res;						\
+	ASSERT(token->kaddr);						\
+	ASSERT(check_setget_bounds(token->eb, ptr, off, size));		\
+	if (token->offset <= member_offset &&				\
+	    member_offset + size <= token->offset + PAGE_SIZE) {	\
+		return get_unaligned_le##bits(token->kaddr + oip);	\
 	}								\
-	err = map_private_extent_buffer(eb, offset, size,		\
-					&kaddr, &map_start, &map_len);	\
-	if (err) {							\
-		__le##bits leres;					\
+	token->kaddr = page_address(token->eb->pages[idx]);		\
+	token->offset = idx << PAGE_SHIFT;				\
+	if (INLINE_EXTENT_BUFFER_PAGES == 1 || oip + size <= PAGE_SIZE ) \
+		return get_unaligned_le##bits(token->kaddr + oip);	\
 									\
-		read_extent_buffer(eb, &leres, offset, size);		\
-		return le##bits##_to_cpu(leres);			\
-	}								\
-	p = kaddr + part_offset - map_start;				\
-	res = get_unaligned_le##bits(p + off);				\
-	token->kaddr = kaddr;						\
-	token->offset = map_start;					\
-	return res;							\
+	memcpy(lebytes, token->kaddr + oip, part);			\
+	token->kaddr = page_address(token->eb->pages[idx + 1]);		\
+	token->offset = (idx + 1) << PAGE_SHIFT;			\
+	memcpy(lebytes + part, token->kaddr, size - part);		\
+	return get_unaligned_le##bits(lebytes);				\
 }									\
 u##bits btrfs_get_##bits(const struct extent_buffer *eb,		\
 			 const void *ptr, unsigned long off)		\
 {									\
-	unsigned long part_offset = (unsigned long)ptr;			\
-	unsigned long offset = part_offset + off;			\
-	void *p;							\
-	int err;							\
-	char *kaddr;							\
-	unsigned long map_start;					\
-	unsigned long map_len;						\
-	int size = sizeof(u##bits);					\
-	u##bits res;							\
+	const unsigned long member_offset = (unsigned long)ptr + off;	\
+	const unsigned long oip = get_eb_offset_in_page(eb, member_offset); \
+	const unsigned long idx = get_eb_page_index(member_offset);	\
+	char *kaddr = page_address(eb->pages[idx]);			\
+	const int size = sizeof(u##bits);				\
+	const int part = PAGE_SIZE - oip;				\
+	u8 lebytes[sizeof(u##bits)];					\
 									\
-	err = map_private_extent_buffer(eb, offset, size,		\
-					&kaddr, &map_start, &map_len);	\
-	if (err) {							\
-		__le##bits leres;					\
+	ASSERT(check_setget_bounds(eb, ptr, off, size));		\
+	if (INLINE_EXTENT_BUFFER_PAGES == 1 || oip + size <= PAGE_SIZE)	\
+		return get_unaligned_le##bits(kaddr + oip);		\
 									\
-		read_extent_buffer(eb, &leres, offset, size);		\
-		return le##bits##_to_cpu(leres);			\
-	}								\
-	p = kaddr + part_offset - map_start;				\
-	res = get_unaligned_le##bits(p + off);				\
-	return res;							\
+	memcpy(lebytes, kaddr + oip, part);				\
+	kaddr = page_address(eb->pages[idx + 1]);			\
+	memcpy(lebytes + part, kaddr, size - part);			\
+	return get_unaligned_le##bits(lebytes);				\
 }									\
-void btrfs_set_token_##bits(struct extent_buffer *eb,			\
+void btrfs_set_token_##bits(struct btrfs_map_token *token,		\
 			    const void *ptr, unsigned long off,		\
-			    u##bits val,				\
-			    struct btrfs_map_token *token)		\
+			    u##bits val)				\
 {									\
-	unsigned long part_offset = (unsigned long)ptr;			\
-	unsigned long offset = part_offset + off;			\
-	void *p;							\
-	int err;							\
-	char *kaddr;							\
-	unsigned long map_start;					\
-	unsigned long map_len;						\
-	int size = sizeof(u##bits);					\
+	const unsigned long member_offset = (unsigned long)ptr + off;	\
+	const unsigned long idx = get_eb_page_index(member_offset);	\
+	const unsigned long oip = get_eb_offset_in_page(token->eb,	\
+							member_offset);	\
+	const int size = sizeof(u##bits);				\
+	u8 lebytes[sizeof(u##bits)];					\
+	const int part = PAGE_SIZE - oip;				\
 									\
 	ASSERT(token);							\
-	ASSERT(token->eb == eb);					\
-									\
-	if (token->kaddr && token->offset <= offset &&			\
-	   (token->offset + PAGE_SIZE >= offset + size)) {	\
-		kaddr = token->kaddr;					\
-		p = kaddr + part_offset - token->offset;		\
-		put_unaligned_le##bits(val, p + off);			\
+	ASSERT(token->kaddr);						\
+	ASSERT(check_setget_bounds(token->eb, ptr, off, size));		\
+	if (token->offset <= member_offset &&				\
+	    member_offset + size <= token->offset + PAGE_SIZE) {	\
+		put_unaligned_le##bits(val, token->kaddr + oip);	\
 		return;							\
 	}								\
-	err = map_private_extent_buffer(eb, offset, size,		\
-			&kaddr, &map_start, &map_len);			\
-	if (err) {							\
-		__le##bits val2;					\
-									\
-		val2 = cpu_to_le##bits(val);				\
-		write_extent_buffer(eb, &val2, offset, size);		\
+	token->kaddr = page_address(token->eb->pages[idx]);		\
+	token->offset = idx << PAGE_SHIFT;				\
+	if (INLINE_EXTENT_BUFFER_PAGES == 1 || oip + size <= PAGE_SIZE) { \
+		put_unaligned_le##bits(val, token->kaddr + oip);	\
 		return;							\
 	}								\
-	p = kaddr + part_offset - map_start;				\
-	put_unaligned_le##bits(val, p + off);				\
-	token->kaddr = kaddr;						\
-	token->offset = map_start;					\
+	put_unaligned_le##bits(val, lebytes);				\
+	memcpy(token->kaddr + oip, lebytes, part);			\
+	token->kaddr = page_address(token->eb->pages[idx + 1]);		\
+	token->offset = (idx + 1) << PAGE_SHIFT;			\
+	memcpy(token->kaddr, lebytes + part, size - part);		\
 }									\
-void btrfs_set_##bits(struct extent_buffer *eb, void *ptr,		\
+void btrfs_set_##bits(const struct extent_buffer *eb, void *ptr,	\
 		      unsigned long off, u##bits val)			\
 {									\
-	unsigned long part_offset = (unsigned long)ptr;			\
-	unsigned long offset = part_offset + off;			\
-	void *p;							\
-	int err;							\
-	char *kaddr;							\
-	unsigned long map_start;					\
-	unsigned long map_len;						\
-	int size = sizeof(u##bits);					\
+	const unsigned long member_offset = (unsigned long)ptr + off;	\
+	const unsigned long oip = get_eb_offset_in_page(eb, member_offset); \
+	const unsigned long idx = get_eb_page_index(member_offset);	\
+	char *kaddr = page_address(eb->pages[idx]);			\
+	const int size = sizeof(u##bits);				\
+	const int part = PAGE_SIZE - oip;				\
+	u8 lebytes[sizeof(u##bits)];					\
 									\
-	err = map_private_extent_buffer(eb, offset, size,		\
-			&kaddr, &map_start, &map_len);			\
-	if (err) {							\
-		__le##bits val2;					\
-									\
-		val2 = cpu_to_le##bits(val);				\
-		write_extent_buffer(eb, &val2, offset, size);		\
+	ASSERT(check_setget_bounds(eb, ptr, off, size));		\
+	if (INLINE_EXTENT_BUFFER_PAGES == 1 || oip + size <= PAGE_SIZE) { \
+		put_unaligned_le##bits(val, kaddr + oip);		\
 		return;							\
 	}								\
-	p = kaddr + part_offset - map_start;				\
-	put_unaligned_le##bits(val, p + off);				\
+									\
+	put_unaligned_le##bits(val, lebytes);				\
+	memcpy(kaddr + oip, lebytes, part);				\
+	kaddr = page_address(eb->pages[idx + 1]);			\
+	memcpy(kaddr, lebytes + part, size - part);			\
 }
 
 DEFINE_BTRFS_SETGET_BITS(8)
diff --git a/fs/btrfs/subpage.c b/fs/btrfs/subpage.c
new file mode 100644
index 000000000000..9a176af847d7
--- /dev/null
+++ b/fs/btrfs/subpage.c
@@ -0,0 +1,768 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <linux/slab.h>
+#include "ctree.h"
+#include "subpage.h"
+#include "btrfs_inode.h"
+
+/*
+ * Subpage (sectorsize < PAGE_SIZE) support overview:
+ *
+ * Limitations:
+ *
+ * - Only support 64K page size for now
+ *   This is to make metadata handling easier, as 64K page would ensure
+ *   all nodesize would fit inside one page, thus we don't need to handle
+ *   cases where a tree block crosses several pages.
+ *
+ * - Only metadata read-write for now
+ *   The data read-write part is in development.
+ *
+ * - Metadata can't cross 64K page boundary
+ *   btrfs-progs and kernel have done that for a while, thus only ancient
+ *   filesystems could have such problem.  For such case, do a graceful
+ *   rejection.
+ *
+ * Special behavior:
+ *
+ * - Metadata
+ *   Metadata read is fully supported.
+ *   Meaning when reading one tree block will only trigger the read for the
+ *   needed range, other unrelated range in the same page will not be touched.
+ *
+ *   Metadata write support is partial.
+ *   The writeback is still for the full page, but we will only submit
+ *   the dirty extent buffers in the page.
+ *
+ *   This means, if we have a metadata page like this:
+ *
+ *   Page offset
+ *   0         16K         32K         48K        64K
+ *   |/////////|           |///////////|
+ *        \- Tree block A        \- Tree block B
+ *
+ *   Even if we just want to writeback tree block A, we will also writeback
+ *   tree block B if it's also dirty.
+ *
+ *   This may cause extra metadata writeback which results more COW.
+ *
+ * Implementation:
+ *
+ * - Common
+ *   Both metadata and data will use a new structure, btrfs_subpage, to
+ *   record the status of each sector inside a page.  This provides the extra
+ *   granularity needed.
+ *
+ * - Metadata
+ *   Since we have multiple tree blocks inside one page, we can't rely on page
+ *   locking anymore, or we will have greatly reduced concurrency or even
+ *   deadlocks (hold one tree lock while trying to lock another tree lock in
+ *   the same page).
+ *
+ *   Thus for metadata locking, subpage support relies on io_tree locking only.
+ *   This means a slightly higher tree locking latency.
+ */
+
+bool btrfs_is_subpage(const struct btrfs_fs_info *fs_info, struct page *page)
+{
+	if (fs_info->sectorsize >= PAGE_SIZE)
+		return false;
+
+	/*
+	 * Only data pages (either through DIO or compression) can have no
+	 * mapping. And if page->mapping->host is data inode, it's subpage.
+	 * As we have ruled our sectorsize >= PAGE_SIZE case already.
+	 */
+	if (!page->mapping || !page->mapping->host ||
+	    is_data_inode(page->mapping->host))
+		return true;
+
+	/*
+	 * Now the only remaining case is metadata, which we only go subpage
+	 * routine if nodesize < PAGE_SIZE.
+	 */
+	if (fs_info->nodesize < PAGE_SIZE)
+		return true;
+	return false;
+}
+
+void btrfs_init_subpage_info(struct btrfs_subpage_info *subpage_info, u32 sectorsize)
+{
+	unsigned int cur = 0;
+	unsigned int nr_bits;
+
+	ASSERT(IS_ALIGNED(PAGE_SIZE, sectorsize));
+
+	nr_bits = PAGE_SIZE / sectorsize;
+	subpage_info->bitmap_nr_bits = nr_bits;
+
+	subpage_info->uptodate_offset = cur;
+	cur += nr_bits;
+
+	subpage_info->error_offset = cur;
+	cur += nr_bits;
+
+	subpage_info->dirty_offset = cur;
+	cur += nr_bits;
+
+	subpage_info->writeback_offset = cur;
+	cur += nr_bits;
+
+	subpage_info->ordered_offset = cur;
+	cur += nr_bits;
+
+	subpage_info->checked_offset = cur;
+	cur += nr_bits;
+
+	subpage_info->total_nr_bits = cur;
+}
+
+int btrfs_attach_subpage(const struct btrfs_fs_info *fs_info,
+			 struct page *page, enum btrfs_subpage_type type)
+{
+	struct btrfs_subpage *subpage;
+
+	/*
+	 * We have cases like a dummy extent buffer page, which is not mapped
+	 * and doesn't need to be locked.
+	 */
+	if (page->mapping)
+		ASSERT(PageLocked(page));
+
+	/* Either not subpage, or the page already has private attached */
+	if (!btrfs_is_subpage(fs_info, page) || PagePrivate(page))
+		return 0;
+
+	subpage = btrfs_alloc_subpage(fs_info, type);
+	if (IS_ERR(subpage))
+		return  PTR_ERR(subpage);
+
+	attach_page_private(page, subpage);
+	return 0;
+}
+
+void btrfs_detach_subpage(const struct btrfs_fs_info *fs_info,
+			  struct page *page)
+{
+	struct btrfs_subpage *subpage;
+
+	/* Either not subpage, or already detached */
+	if (!btrfs_is_subpage(fs_info, page) || !PagePrivate(page))
+		return;
+
+	subpage = detach_page_private(page);
+	ASSERT(subpage);
+	btrfs_free_subpage(subpage);
+}
+
+struct btrfs_subpage *btrfs_alloc_subpage(const struct btrfs_fs_info *fs_info,
+					  enum btrfs_subpage_type type)
+{
+	struct btrfs_subpage *ret;
+	unsigned int real_size;
+
+	ASSERT(fs_info->sectorsize < PAGE_SIZE);
+
+	real_size = struct_size(ret, bitmaps,
+			BITS_TO_LONGS(fs_info->subpage_info->total_nr_bits));
+	ret = kzalloc(real_size, GFP_NOFS);
+	if (!ret)
+		return ERR_PTR(-ENOMEM);
+
+	spin_lock_init(&ret->lock);
+	if (type == BTRFS_SUBPAGE_METADATA) {
+		atomic_set(&ret->eb_refs, 0);
+	} else {
+		atomic_set(&ret->readers, 0);
+		atomic_set(&ret->writers, 0);
+	}
+	return ret;
+}
+
+void btrfs_free_subpage(struct btrfs_subpage *subpage)
+{
+	kfree(subpage);
+}
+
+/*
+ * Increase the eb_refs of current subpage.
+ *
+ * This is important for eb allocation, to prevent race with last eb freeing
+ * of the same page.
+ * With the eb_refs increased before the eb inserted into radix tree,
+ * detach_extent_buffer_page() won't detach the page private while we're still
+ * allocating the extent buffer.
+ */
+void btrfs_page_inc_eb_refs(const struct btrfs_fs_info *fs_info,
+			    struct page *page)
+{
+	struct btrfs_subpage *subpage;
+
+	if (!btrfs_is_subpage(fs_info, page))
+		return;
+
+	ASSERT(PagePrivate(page) && page->mapping);
+	lockdep_assert_held(&page->mapping->private_lock);
+
+	subpage = (struct btrfs_subpage *)page->private;
+	atomic_inc(&subpage->eb_refs);
+}
+
+void btrfs_page_dec_eb_refs(const struct btrfs_fs_info *fs_info,
+			    struct page *page)
+{
+	struct btrfs_subpage *subpage;
+
+	if (!btrfs_is_subpage(fs_info, page))
+		return;
+
+	ASSERT(PagePrivate(page) && page->mapping);
+	lockdep_assert_held(&page->mapping->private_lock);
+
+	subpage = (struct btrfs_subpage *)page->private;
+	ASSERT(atomic_read(&subpage->eb_refs));
+	atomic_dec(&subpage->eb_refs);
+}
+
+static void btrfs_subpage_assert(const struct btrfs_fs_info *fs_info,
+		struct page *page, u64 start, u32 len)
+{
+	/* Basic checks */
+	ASSERT(PagePrivate(page) && page->private);
+	ASSERT(IS_ALIGNED(start, fs_info->sectorsize) &&
+	       IS_ALIGNED(len, fs_info->sectorsize));
+	/*
+	 * The range check only works for mapped page, we can still have
+	 * unmapped page like dummy extent buffer pages.
+	 */
+	if (page->mapping)
+		ASSERT(page_offset(page) <= start &&
+		       start + len <= page_offset(page) + PAGE_SIZE);
+}
+
+void btrfs_subpage_start_reader(const struct btrfs_fs_info *fs_info,
+		struct page *page, u64 start, u32 len)
+{
+	struct btrfs_subpage *subpage = (struct btrfs_subpage *)page->private;
+	const int nbits = len >> fs_info->sectorsize_bits;
+
+	btrfs_subpage_assert(fs_info, page, start, len);
+
+	atomic_add(nbits, &subpage->readers);
+}
+
+void btrfs_subpage_end_reader(const struct btrfs_fs_info *fs_info,
+		struct page *page, u64 start, u32 len)
+{
+	struct btrfs_subpage *subpage = (struct btrfs_subpage *)page->private;
+	const int nbits = len >> fs_info->sectorsize_bits;
+	bool is_data;
+	bool last;
+
+	btrfs_subpage_assert(fs_info, page, start, len);
+	is_data = is_data_inode(page->mapping->host);
+	ASSERT(atomic_read(&subpage->readers) >= nbits);
+	last = atomic_sub_and_test(nbits, &subpage->readers);
+
+	/*
+	 * For data we need to unlock the page if the last read has finished.
+	 *
+	 * And please don't replace @last with atomic_sub_and_test() call
+	 * inside if () condition.
+	 * As we want the atomic_sub_and_test() to be always executed.
+	 */
+	if (is_data && last)
+		unlock_page(page);
+}
+
+static void btrfs_subpage_clamp_range(struct page *page, u64 *start, u32 *len)
+{
+	u64 orig_start = *start;
+	u32 orig_len = *len;
+
+	*start = max_t(u64, page_offset(page), orig_start);
+	/*
+	 * For certain call sites like btrfs_drop_pages(), we may have pages
+	 * beyond the target range. In that case, just set @len to 0, subpage
+	 * helpers can handle @len == 0 without any problem.
+	 */
+	if (page_offset(page) >= orig_start + orig_len)
+		*len = 0;
+	else
+		*len = min_t(u64, page_offset(page) + PAGE_SIZE,
+			     orig_start + orig_len) - *start;
+}
+
+void btrfs_subpage_start_writer(const struct btrfs_fs_info *fs_info,
+		struct page *page, u64 start, u32 len)
+{
+	struct btrfs_subpage *subpage = (struct btrfs_subpage *)page->private;
+	const int nbits = (len >> fs_info->sectorsize_bits);
+	int ret;
+
+	btrfs_subpage_assert(fs_info, page, start, len);
+
+	ASSERT(atomic_read(&subpage->readers) == 0);
+	ret = atomic_add_return(nbits, &subpage->writers);
+	ASSERT(ret == nbits);
+}
+
+bool btrfs_subpage_end_and_test_writer(const struct btrfs_fs_info *fs_info,
+		struct page *page, u64 start, u32 len)
+{
+	struct btrfs_subpage *subpage = (struct btrfs_subpage *)page->private;
+	const int nbits = (len >> fs_info->sectorsize_bits);
+
+	btrfs_subpage_assert(fs_info, page, start, len);
+
+	/*
+	 * We have call sites passing @lock_page into
+	 * extent_clear_unlock_delalloc() for compression path.
+	 *
+	 * This @locked_page is locked by plain lock_page(), thus its
+	 * subpage::writers is 0.  Handle them in a special way.
+	 */
+	if (atomic_read(&subpage->writers) == 0)
+		return true;
+
+	ASSERT(atomic_read(&subpage->writers) >= nbits);
+	return atomic_sub_and_test(nbits, &subpage->writers);
+}
+
+/*
+ * Lock a page for delalloc page writeback.
+ *
+ * Return -EAGAIN if the page is not properly initialized.
+ * Return 0 with the page locked, and writer counter updated.
+ *
+ * Even with 0 returned, the page still need extra check to make sure
+ * it's really the correct page, as the caller is using
+ * filemap_get_folios_contig(), which can race with page invalidating.
+ */
+int btrfs_page_start_writer_lock(const struct btrfs_fs_info *fs_info,
+		struct page *page, u64 start, u32 len)
+{
+	if (unlikely(!fs_info) || !btrfs_is_subpage(fs_info, page)) {
+		lock_page(page);
+		return 0;
+	}
+	lock_page(page);
+	if (!PagePrivate(page) || !page->private) {
+		unlock_page(page);
+		return -EAGAIN;
+	}
+	btrfs_subpage_clamp_range(page, &start, &len);
+	btrfs_subpage_start_writer(fs_info, page, start, len);
+	return 0;
+}
+
+void btrfs_page_end_writer_lock(const struct btrfs_fs_info *fs_info,
+		struct page *page, u64 start, u32 len)
+{
+	if (unlikely(!fs_info) || !btrfs_is_subpage(fs_info, page))
+		return unlock_page(page);
+	btrfs_subpage_clamp_range(page, &start, &len);
+	if (btrfs_subpage_end_and_test_writer(fs_info, page, start, len))
+		unlock_page(page);
+}
+
+static bool bitmap_test_range_all_set(unsigned long *addr, unsigned int start,
+				      unsigned int nbits)
+{
+	unsigned int found_zero;
+
+	found_zero = find_next_zero_bit(addr, start + nbits, start);
+	if (found_zero == start + nbits)
+		return true;
+	return false;
+}
+
+static bool bitmap_test_range_all_zero(unsigned long *addr, unsigned int start,
+				       unsigned int nbits)
+{
+	unsigned int found_set;
+
+	found_set = find_next_bit(addr, start + nbits, start);
+	if (found_set == start + nbits)
+		return true;
+	return false;
+}
+
+#define subpage_calc_start_bit(fs_info, page, name, start, len)		\
+({									\
+	unsigned int start_bit;						\
+									\
+	btrfs_subpage_assert(fs_info, page, start, len);		\
+	start_bit = offset_in_page(start) >> fs_info->sectorsize_bits;	\
+	start_bit += fs_info->subpage_info->name##_offset;		\
+	start_bit;							\
+})
+
+#define subpage_test_bitmap_all_set(fs_info, subpage, name)		\
+	bitmap_test_range_all_set(subpage->bitmaps,			\
+			fs_info->subpage_info->name##_offset,		\
+			fs_info->subpage_info->bitmap_nr_bits)
+
+#define subpage_test_bitmap_all_zero(fs_info, subpage, name)		\
+	bitmap_test_range_all_zero(subpage->bitmaps,			\
+			fs_info->subpage_info->name##_offset,		\
+			fs_info->subpage_info->bitmap_nr_bits)
+
+void btrfs_subpage_set_uptodate(const struct btrfs_fs_info *fs_info,
+		struct page *page, u64 start, u32 len)
+{
+	struct btrfs_subpage *subpage = (struct btrfs_subpage *)page->private;
+	unsigned int start_bit = subpage_calc_start_bit(fs_info, page,
+							uptodate, start, len);
+	unsigned long flags;
+
+	spin_lock_irqsave(&subpage->lock, flags);
+	bitmap_set(subpage->bitmaps, start_bit, len >> fs_info->sectorsize_bits);
+	if (subpage_test_bitmap_all_set(fs_info, subpage, uptodate))
+		SetPageUptodate(page);
+	spin_unlock_irqrestore(&subpage->lock, flags);
+}
+
+void btrfs_subpage_clear_uptodate(const struct btrfs_fs_info *fs_info,
+		struct page *page, u64 start, u32 len)
+{
+	struct btrfs_subpage *subpage = (struct btrfs_subpage *)page->private;
+	unsigned int start_bit = subpage_calc_start_bit(fs_info, page,
+							uptodate, start, len);
+	unsigned long flags;
+
+	spin_lock_irqsave(&subpage->lock, flags);
+	bitmap_clear(subpage->bitmaps, start_bit, len >> fs_info->sectorsize_bits);
+	ClearPageUptodate(page);
+	spin_unlock_irqrestore(&subpage->lock, flags);
+}
+
+void btrfs_subpage_set_error(const struct btrfs_fs_info *fs_info,
+		struct page *page, u64 start, u32 len)
+{
+	struct btrfs_subpage *subpage = (struct btrfs_subpage *)page->private;
+	unsigned int start_bit = subpage_calc_start_bit(fs_info, page,
+							error, start, len);
+	unsigned long flags;
+
+	spin_lock_irqsave(&subpage->lock, flags);
+	bitmap_set(subpage->bitmaps, start_bit, len >> fs_info->sectorsize_bits);
+	SetPageError(page);
+	spin_unlock_irqrestore(&subpage->lock, flags);
+}
+
+void btrfs_subpage_clear_error(const struct btrfs_fs_info *fs_info,
+		struct page *page, u64 start, u32 len)
+{
+	struct btrfs_subpage *subpage = (struct btrfs_subpage *)page->private;
+	unsigned int start_bit = subpage_calc_start_bit(fs_info, page,
+							error, start, len);
+	unsigned long flags;
+
+	spin_lock_irqsave(&subpage->lock, flags);
+	bitmap_clear(subpage->bitmaps, start_bit, len >> fs_info->sectorsize_bits);
+	if (subpage_test_bitmap_all_zero(fs_info, subpage, error))
+		ClearPageError(page);
+	spin_unlock_irqrestore(&subpage->lock, flags);
+}
+
+void btrfs_subpage_set_dirty(const struct btrfs_fs_info *fs_info,
+		struct page *page, u64 start, u32 len)
+{
+	struct btrfs_subpage *subpage = (struct btrfs_subpage *)page->private;
+	unsigned int start_bit = subpage_calc_start_bit(fs_info, page,
+							dirty, start, len);
+	unsigned long flags;
+
+	spin_lock_irqsave(&subpage->lock, flags);
+	bitmap_set(subpage->bitmaps, start_bit, len >> fs_info->sectorsize_bits);
+	spin_unlock_irqrestore(&subpage->lock, flags);
+	set_page_dirty(page);
+}
+
+/*
+ * Extra clear_and_test function for subpage dirty bitmap.
+ *
+ * Return true if we're the last bits in the dirty_bitmap and clear the
+ * dirty_bitmap.
+ * Return false otherwise.
+ *
+ * NOTE: Callers should manually clear page dirty for true case, as we have
+ * extra handling for tree blocks.
+ */
+bool btrfs_subpage_clear_and_test_dirty(const struct btrfs_fs_info *fs_info,
+		struct page *page, u64 start, u32 len)
+{
+	struct btrfs_subpage *subpage = (struct btrfs_subpage *)page->private;
+	unsigned int start_bit = subpage_calc_start_bit(fs_info, page,
+							dirty, start, len);
+	unsigned long flags;
+	bool last = false;
+
+	spin_lock_irqsave(&subpage->lock, flags);
+	bitmap_clear(subpage->bitmaps, start_bit, len >> fs_info->sectorsize_bits);
+	if (subpage_test_bitmap_all_zero(fs_info, subpage, dirty))
+		last = true;
+	spin_unlock_irqrestore(&subpage->lock, flags);
+	return last;
+}
+
+void btrfs_subpage_clear_dirty(const struct btrfs_fs_info *fs_info,
+		struct page *page, u64 start, u32 len)
+{
+	bool last;
+
+	last = btrfs_subpage_clear_and_test_dirty(fs_info, page, start, len);
+	if (last)
+		clear_page_dirty_for_io(page);
+}
+
+void btrfs_subpage_set_writeback(const struct btrfs_fs_info *fs_info,
+		struct page *page, u64 start, u32 len)
+{
+	struct btrfs_subpage *subpage = (struct btrfs_subpage *)page->private;
+	unsigned int start_bit = subpage_calc_start_bit(fs_info, page,
+							writeback, start, len);
+	unsigned long flags;
+
+	spin_lock_irqsave(&subpage->lock, flags);
+	bitmap_set(subpage->bitmaps, start_bit, len >> fs_info->sectorsize_bits);
+	set_page_writeback(page);
+	spin_unlock_irqrestore(&subpage->lock, flags);
+}
+
+void btrfs_subpage_clear_writeback(const struct btrfs_fs_info *fs_info,
+		struct page *page, u64 start, u32 len)
+{
+	struct btrfs_subpage *subpage = (struct btrfs_subpage *)page->private;
+	unsigned int start_bit = subpage_calc_start_bit(fs_info, page,
+							writeback, start, len);
+	unsigned long flags;
+
+	spin_lock_irqsave(&subpage->lock, flags);
+	bitmap_clear(subpage->bitmaps, start_bit, len >> fs_info->sectorsize_bits);
+	if (subpage_test_bitmap_all_zero(fs_info, subpage, writeback)) {
+		ASSERT(PageWriteback(page));
+		end_page_writeback(page);
+	}
+	spin_unlock_irqrestore(&subpage->lock, flags);
+}
+
+void btrfs_subpage_set_ordered(const struct btrfs_fs_info *fs_info,
+		struct page *page, u64 start, u32 len)
+{
+	struct btrfs_subpage *subpage = (struct btrfs_subpage *)page->private;
+	unsigned int start_bit = subpage_calc_start_bit(fs_info, page,
+							ordered, start, len);
+	unsigned long flags;
+
+	spin_lock_irqsave(&subpage->lock, flags);
+	bitmap_set(subpage->bitmaps, start_bit, len >> fs_info->sectorsize_bits);
+	SetPageOrdered(page);
+	spin_unlock_irqrestore(&subpage->lock, flags);
+}
+
+void btrfs_subpage_clear_ordered(const struct btrfs_fs_info *fs_info,
+		struct page *page, u64 start, u32 len)
+{
+	struct btrfs_subpage *subpage = (struct btrfs_subpage *)page->private;
+	unsigned int start_bit = subpage_calc_start_bit(fs_info, page,
+							ordered, start, len);
+	unsigned long flags;
+
+	spin_lock_irqsave(&subpage->lock, flags);
+	bitmap_clear(subpage->bitmaps, start_bit, len >> fs_info->sectorsize_bits);
+	if (subpage_test_bitmap_all_zero(fs_info, subpage, ordered))
+		ClearPageOrdered(page);
+	spin_unlock_irqrestore(&subpage->lock, flags);
+}
+
+void btrfs_subpage_set_checked(const struct btrfs_fs_info *fs_info,
+			       struct page *page, u64 start, u32 len)
+{
+	struct btrfs_subpage *subpage = (struct btrfs_subpage *)page->private;
+	unsigned int start_bit = subpage_calc_start_bit(fs_info, page,
+							checked, start, len);
+	unsigned long flags;
+
+	spin_lock_irqsave(&subpage->lock, flags);
+	bitmap_set(subpage->bitmaps, start_bit, len >> fs_info->sectorsize_bits);
+	if (subpage_test_bitmap_all_set(fs_info, subpage, checked))
+		SetPageChecked(page);
+	spin_unlock_irqrestore(&subpage->lock, flags);
+}
+
+void btrfs_subpage_clear_checked(const struct btrfs_fs_info *fs_info,
+				 struct page *page, u64 start, u32 len)
+{
+	struct btrfs_subpage *subpage = (struct btrfs_subpage *)page->private;
+	unsigned int start_bit = subpage_calc_start_bit(fs_info, page,
+							checked, start, len);
+	unsigned long flags;
+
+	spin_lock_irqsave(&subpage->lock, flags);
+	bitmap_clear(subpage->bitmaps, start_bit, len >> fs_info->sectorsize_bits);
+	ClearPageChecked(page);
+	spin_unlock_irqrestore(&subpage->lock, flags);
+}
+
+/*
+ * Unlike set/clear which is dependent on each page status, for test all bits
+ * are tested in the same way.
+ */
+#define IMPLEMENT_BTRFS_SUBPAGE_TEST_OP(name)				\
+bool btrfs_subpage_test_##name(const struct btrfs_fs_info *fs_info,	\
+		struct page *page, u64 start, u32 len)			\
+{									\
+	struct btrfs_subpage *subpage = (struct btrfs_subpage *)page->private; \
+	unsigned int start_bit = subpage_calc_start_bit(fs_info, page,	\
+						name, start, len);	\
+	unsigned long flags;						\
+	bool ret;							\
+									\
+	spin_lock_irqsave(&subpage->lock, flags);			\
+	ret = bitmap_test_range_all_set(subpage->bitmaps, start_bit,	\
+				len >> fs_info->sectorsize_bits);	\
+	spin_unlock_irqrestore(&subpage->lock, flags);			\
+	return ret;							\
+}
+IMPLEMENT_BTRFS_SUBPAGE_TEST_OP(uptodate);
+IMPLEMENT_BTRFS_SUBPAGE_TEST_OP(error);
+IMPLEMENT_BTRFS_SUBPAGE_TEST_OP(dirty);
+IMPLEMENT_BTRFS_SUBPAGE_TEST_OP(writeback);
+IMPLEMENT_BTRFS_SUBPAGE_TEST_OP(ordered);
+IMPLEMENT_BTRFS_SUBPAGE_TEST_OP(checked);
+
+/*
+ * Note that, in selftests (extent-io-tests), we can have empty fs_info passed
+ * in.  We only test sectorsize == PAGE_SIZE cases so far, thus we can fall
+ * back to regular sectorsize branch.
+ */
+#define IMPLEMENT_BTRFS_PAGE_OPS(name, set_page_func, clear_page_func,	\
+			       test_page_func)				\
+void btrfs_page_set_##name(const struct btrfs_fs_info *fs_info,		\
+		struct page *page, u64 start, u32 len)			\
+{									\
+	if (unlikely(!fs_info) || !btrfs_is_subpage(fs_info, page)) {	\
+		set_page_func(page);					\
+		return;							\
+	}								\
+	btrfs_subpage_set_##name(fs_info, page, start, len);		\
+}									\
+void btrfs_page_clear_##name(const struct btrfs_fs_info *fs_info,	\
+		struct page *page, u64 start, u32 len)			\
+{									\
+	if (unlikely(!fs_info) || !btrfs_is_subpage(fs_info, page)) {	\
+		clear_page_func(page);					\
+		return;							\
+	}								\
+	btrfs_subpage_clear_##name(fs_info, page, start, len);		\
+}									\
+bool btrfs_page_test_##name(const struct btrfs_fs_info *fs_info,	\
+		struct page *page, u64 start, u32 len)			\
+{									\
+	if (unlikely(!fs_info) || !btrfs_is_subpage(fs_info, page))	\
+		return test_page_func(page);				\
+	return btrfs_subpage_test_##name(fs_info, page, start, len);	\
+}									\
+void btrfs_page_clamp_set_##name(const struct btrfs_fs_info *fs_info,	\
+		struct page *page, u64 start, u32 len)			\
+{									\
+	if (unlikely(!fs_info) || !btrfs_is_subpage(fs_info, page)) {	\
+		set_page_func(page);					\
+		return;							\
+	}								\
+	btrfs_subpage_clamp_range(page, &start, &len);			\
+	btrfs_subpage_set_##name(fs_info, page, start, len);		\
+}									\
+void btrfs_page_clamp_clear_##name(const struct btrfs_fs_info *fs_info, \
+		struct page *page, u64 start, u32 len)			\
+{									\
+	if (unlikely(!fs_info) || !btrfs_is_subpage(fs_info, page)) {	\
+		clear_page_func(page);					\
+		return;							\
+	}								\
+	btrfs_subpage_clamp_range(page, &start, &len);			\
+	btrfs_subpage_clear_##name(fs_info, page, start, len);		\
+}									\
+bool btrfs_page_clamp_test_##name(const struct btrfs_fs_info *fs_info,	\
+		struct page *page, u64 start, u32 len)			\
+{									\
+	if (unlikely(!fs_info) || !btrfs_is_subpage(fs_info, page))	\
+		return test_page_func(page);				\
+	btrfs_subpage_clamp_range(page, &start, &len);			\
+	return btrfs_subpage_test_##name(fs_info, page, start, len);	\
+}
+IMPLEMENT_BTRFS_PAGE_OPS(uptodate, SetPageUptodate, ClearPageUptodate,
+			 PageUptodate);
+IMPLEMENT_BTRFS_PAGE_OPS(error, SetPageError, ClearPageError, PageError);
+IMPLEMENT_BTRFS_PAGE_OPS(dirty, set_page_dirty, clear_page_dirty_for_io,
+			 PageDirty);
+IMPLEMENT_BTRFS_PAGE_OPS(writeback, set_page_writeback, end_page_writeback,
+			 PageWriteback);
+IMPLEMENT_BTRFS_PAGE_OPS(ordered, SetPageOrdered, ClearPageOrdered,
+			 PageOrdered);
+IMPLEMENT_BTRFS_PAGE_OPS(checked, SetPageChecked, ClearPageChecked, PageChecked);
+
+/*
+ * Make sure not only the page dirty bit is cleared, but also subpage dirty bit
+ * is cleared.
+ */
+void btrfs_page_assert_not_dirty(const struct btrfs_fs_info *fs_info,
+				 struct page *page)
+{
+	struct btrfs_subpage *subpage = (struct btrfs_subpage *)page->private;
+
+	if (!IS_ENABLED(CONFIG_BTRFS_ASSERT))
+		return;
+
+	ASSERT(!PageDirty(page));
+	if (!btrfs_is_subpage(fs_info, page))
+		return;
+
+	ASSERT(PagePrivate(page) && page->private);
+	ASSERT(subpage_test_bitmap_all_zero(fs_info, subpage, dirty));
+}
+
+/*
+ * Handle different locked pages with different page sizes:
+ *
+ * - Page locked by plain lock_page()
+ *   It should not have any subpage::writers count.
+ *   Can be unlocked by unlock_page().
+ *   This is the most common locked page for __extent_writepage() called
+ *   inside extent_write_cache_pages().
+ *   Rarer cases include the @locked_page from extent_write_locked_range().
+ *
+ * - Page locked by lock_delalloc_pages()
+ *   There is only one caller, all pages except @locked_page for
+ *   extent_write_locked_range().
+ *   In this case, we have to call subpage helper to handle the case.
+ */
+void btrfs_page_unlock_writer(struct btrfs_fs_info *fs_info, struct page *page,
+			      u64 start, u32 len)
+{
+	struct btrfs_subpage *subpage;
+
+	ASSERT(PageLocked(page));
+	/* For non-subpage case, we just unlock the page */
+	if (!btrfs_is_subpage(fs_info, page))
+		return unlock_page(page);
+
+	ASSERT(PagePrivate(page) && page->private);
+	subpage = (struct btrfs_subpage *)page->private;
+
+	/*
+	 * For subpage case, there are two types of locked page.  With or
+	 * without writers number.
+	 *
+	 * Since we own the page lock, no one else could touch subpage::writers
+	 * and we are safe to do several atomic operations without spinlock.
+	 */
+	if (atomic_read(&subpage->writers) == 0)
+		/* No writers, locked by plain lock_page() */
+		return unlock_page(page);
+
+	/* Have writers, use proper subpage helper to end it */
+	btrfs_page_end_writer_lock(fs_info, page, start, len);
+}
diff --git a/fs/btrfs/subpage.h b/fs/btrfs/subpage.h
new file mode 100644
index 000000000000..0e80ad336904
--- /dev/null
+++ b/fs/btrfs/subpage.h
@@ -0,0 +1,158 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+
+#ifndef BTRFS_SUBPAGE_H
+#define BTRFS_SUBPAGE_H
+
+#include <linux/spinlock.h>
+
+/*
+ * Extra info for subpapge bitmap.
+ *
+ * For subpage we pack all uptodate/error/dirty/writeback/ordered bitmaps into
+ * one larger bitmap.
+ *
+ * This structure records how they are organized in the bitmap:
+ *
+ * /- uptodate_offset	/- error_offset	/- dirty_offset
+ * |			|		|
+ * v			v		v
+ * |u|u|u|u|........|u|u|e|e|.......|e|e| ...	|o|o|
+ * |<- bitmap_nr_bits ->|
+ * |<--------------- total_nr_bits ---------------->|
+ */
+struct btrfs_subpage_info {
+	/* Number of bits for each bitmap */
+	unsigned int bitmap_nr_bits;
+
+	/* Total number of bits for the whole bitmap */
+	unsigned int total_nr_bits;
+
+	/*
+	 * *_start indicates where the bitmap starts, the length is always
+	 * @bitmap_size, which is calculated from PAGE_SIZE / sectorsize.
+	 */
+	unsigned int uptodate_offset;
+	unsigned int error_offset;
+	unsigned int dirty_offset;
+	unsigned int writeback_offset;
+	unsigned int ordered_offset;
+	unsigned int checked_offset;
+};
+
+/*
+ * Structure to trace status of each sector inside a page, attached to
+ * page::private for both data and metadata inodes.
+ */
+struct btrfs_subpage {
+	/* Common members for both data and metadata pages */
+	spinlock_t lock;
+	/*
+	 * Both data and metadata needs to track how many readers are for the
+	 * page.
+	 * Data relies on @readers to unlock the page when last reader finished.
+	 * While metadata doesn't need page unlock, it needs to prevent
+	 * page::private get cleared before the last end_page_read().
+	 */
+	atomic_t readers;
+	union {
+		/*
+		 * Structures only used by metadata
+		 *
+		 * @eb_refs should only be operated under private_lock, as it
+		 * manages whether the subpage can be detached.
+		 */
+		atomic_t eb_refs;
+
+		/* Structures only used by data */
+		atomic_t writers;
+	};
+	unsigned long bitmaps[];
+};
+
+enum btrfs_subpage_type {
+	BTRFS_SUBPAGE_METADATA,
+	BTRFS_SUBPAGE_DATA,
+};
+
+bool btrfs_is_subpage(const struct btrfs_fs_info *fs_info, struct page *page);
+
+void btrfs_init_subpage_info(struct btrfs_subpage_info *subpage_info, u32 sectorsize);
+int btrfs_attach_subpage(const struct btrfs_fs_info *fs_info,
+			 struct page *page, enum btrfs_subpage_type type);
+void btrfs_detach_subpage(const struct btrfs_fs_info *fs_info,
+			  struct page *page);
+
+/* Allocate additional data where page represents more than one sector */
+struct btrfs_subpage *btrfs_alloc_subpage(const struct btrfs_fs_info *fs_info,
+					  enum btrfs_subpage_type type);
+void btrfs_free_subpage(struct btrfs_subpage *subpage);
+
+void btrfs_page_inc_eb_refs(const struct btrfs_fs_info *fs_info,
+			    struct page *page);
+void btrfs_page_dec_eb_refs(const struct btrfs_fs_info *fs_info,
+			    struct page *page);
+
+void btrfs_subpage_start_reader(const struct btrfs_fs_info *fs_info,
+		struct page *page, u64 start, u32 len);
+void btrfs_subpage_end_reader(const struct btrfs_fs_info *fs_info,
+		struct page *page, u64 start, u32 len);
+
+void btrfs_subpage_start_writer(const struct btrfs_fs_info *fs_info,
+		struct page *page, u64 start, u32 len);
+bool btrfs_subpage_end_and_test_writer(const struct btrfs_fs_info *fs_info,
+		struct page *page, u64 start, u32 len);
+int btrfs_page_start_writer_lock(const struct btrfs_fs_info *fs_info,
+		struct page *page, u64 start, u32 len);
+void btrfs_page_end_writer_lock(const struct btrfs_fs_info *fs_info,
+		struct page *page, u64 start, u32 len);
+
+/*
+ * Template for subpage related operations.
+ *
+ * btrfs_subpage_*() are for call sites where the page has subpage attached and
+ * the range is ensured to be inside the page.
+ *
+ * btrfs_page_*() are for call sites where the page can either be subpage
+ * specific or regular page. The function will handle both cases.
+ * But the range still needs to be inside the page.
+ *
+ * btrfs_page_clamp_*() are similar to btrfs_page_*(), except the range doesn't
+ * need to be inside the page. Those functions will truncate the range
+ * automatically.
+ */
+#define DECLARE_BTRFS_SUBPAGE_OPS(name)					\
+void btrfs_subpage_set_##name(const struct btrfs_fs_info *fs_info,	\
+		struct page *page, u64 start, u32 len);			\
+void btrfs_subpage_clear_##name(const struct btrfs_fs_info *fs_info,	\
+		struct page *page, u64 start, u32 len);			\
+bool btrfs_subpage_test_##name(const struct btrfs_fs_info *fs_info,	\
+		struct page *page, u64 start, u32 len);			\
+void btrfs_page_set_##name(const struct btrfs_fs_info *fs_info,		\
+		struct page *page, u64 start, u32 len);			\
+void btrfs_page_clear_##name(const struct btrfs_fs_info *fs_info,	\
+		struct page *page, u64 start, u32 len);			\
+bool btrfs_page_test_##name(const struct btrfs_fs_info *fs_info,	\
+		struct page *page, u64 start, u32 len);			\
+void btrfs_page_clamp_set_##name(const struct btrfs_fs_info *fs_info,	\
+		struct page *page, u64 start, u32 len);			\
+void btrfs_page_clamp_clear_##name(const struct btrfs_fs_info *fs_info,	\
+		struct page *page, u64 start, u32 len);			\
+bool btrfs_page_clamp_test_##name(const struct btrfs_fs_info *fs_info,	\
+		struct page *page, u64 start, u32 len);
+
+DECLARE_BTRFS_SUBPAGE_OPS(uptodate);
+DECLARE_BTRFS_SUBPAGE_OPS(error);
+DECLARE_BTRFS_SUBPAGE_OPS(dirty);
+DECLARE_BTRFS_SUBPAGE_OPS(writeback);
+DECLARE_BTRFS_SUBPAGE_OPS(ordered);
+DECLARE_BTRFS_SUBPAGE_OPS(checked);
+
+bool btrfs_subpage_clear_and_test_dirty(const struct btrfs_fs_info *fs_info,
+		struct page *page, u64 start, u32 len);
+
+void btrfs_page_assert_not_dirty(const struct btrfs_fs_info *fs_info,
+				 struct page *page);
+void btrfs_page_unlock_writer(struct btrfs_fs_info *fs_info, struct page *page,
+			      u64 start, u32 len);
+
+#endif
diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c
index 67c63858812a..5942b9384088 100644
--- a/fs/btrfs/super.c
+++ b/fs/btrfs/super.c
@@ -23,7 +23,6 @@
 #include <linux/miscdevice.h>
 #include <linux/magic.h>
 #include <linux/slab.h>
-#include <linux/cleancache.h>
 #include <linux/ratelimit.h>
 #include <linux/crc32c.h>
 #include <linux/btrfs.h>
@@ -44,11 +43,12 @@
 #include "backref.h"
 #include "space-info.h"
 #include "sysfs.h"
+#include "zoned.h"
 #include "tests/btrfs-tests.h"
 #include "block-group.h"
 #include "discard.h"
-
 #include "qgroup.h"
+#include "raid56.h"
 #define CREATE_TRACE_POINTS
 #include <trace/events/btrfs.h>
 
@@ -67,28 +67,98 @@ static struct file_system_type btrfs_root_fs_type;
 
 static int btrfs_remount(struct super_block *sb, int *flags, char *data);
 
+#ifdef CONFIG_PRINTK
+
+#define STATE_STRING_PREFACE	": state "
+#define STATE_STRING_BUF_LEN	(sizeof(STATE_STRING_PREFACE) + BTRFS_FS_STATE_COUNT)
+
+/*
+ * Characters to print to indicate error conditions or uncommon filesystem state.
+ * RO is not an error.
+ */
+static const char fs_state_chars[] = {
+	[BTRFS_FS_STATE_ERROR]			= 'E',
+	[BTRFS_FS_STATE_REMOUNTING]		= 'M',
+	[BTRFS_FS_STATE_RO]			= 0,
+	[BTRFS_FS_STATE_TRANS_ABORTED]		= 'A',
+	[BTRFS_FS_STATE_DEV_REPLACING]		= 'R',
+	[BTRFS_FS_STATE_DUMMY_FS_INFO]		= 0,
+	[BTRFS_FS_STATE_NO_CSUMS]		= 'C',
+	[BTRFS_FS_STATE_LOG_CLEANUP_ERROR]	= 'L',
+};
+
+static void btrfs_state_to_string(const struct btrfs_fs_info *info, char *buf)
+{
+	unsigned int bit;
+	bool states_printed = false;
+	unsigned long fs_state = READ_ONCE(info->fs_state);
+	char *curr = buf;
+
+	memcpy(curr, STATE_STRING_PREFACE, sizeof(STATE_STRING_PREFACE));
+	curr += sizeof(STATE_STRING_PREFACE) - 1;
+
+	for_each_set_bit(bit, &fs_state, sizeof(fs_state)) {
+		WARN_ON_ONCE(bit >= BTRFS_FS_STATE_COUNT);
+		if ((bit < BTRFS_FS_STATE_COUNT) && fs_state_chars[bit]) {
+			*curr++ = fs_state_chars[bit];
+			states_printed = true;
+		}
+	}
+
+	/* If no states were printed, reset the buffer */
+	if (!states_printed)
+		curr = buf;
+
+	*curr++ = 0;
+}
+#endif
+
+/*
+ * Generally the error codes correspond to their respective errors, but there
+ * are a few special cases.
+ *
+ * EUCLEAN: Any sort of corruption that we encounter.  The tree-checker for
+ *          instance will return EUCLEAN if any of the blocks are corrupted in
+ *          a way that is problematic.  We want to reserve EUCLEAN for these
+ *          sort of corruptions.
+ *
+ * EROFS: If we check BTRFS_FS_STATE_ERROR and fail out with a return error, we
+ *        need to use EROFS for this case.  We will have no idea of the
+ *        original failure, that will have been reported at the time we tripped
+ *        over the error.  Each subsequent error that doesn't have any context
+ *        of the original error should use EROFS when handling BTRFS_FS_STATE_ERROR.
+ */
 const char * __attribute_const__ btrfs_decode_error(int errno)
 {
 	char *errstr = "unknown";
 
 	switch (errno) {
-	case -EIO:
+	case -ENOENT:		/* -2 */
+		errstr = "No such entry";
+		break;
+	case -EIO:		/* -5 */
 		errstr = "IO failure";
 		break;
-	case -ENOMEM:
+	case -ENOMEM:		/* -12*/
 		errstr = "Out of memory";
 		break;
-	case -EROFS:
-		errstr = "Readonly filesystem";
-		break;
-	case -EEXIST:
+	case -EEXIST:		/* -17 */
 		errstr = "Object already exists";
 		break;
-	case -ENOSPC:
+	case -ENOSPC:		/* -28 */
 		errstr = "No space left";
 		break;
-	case -ENOENT:
-		errstr = "No such entry";
+	case -EROFS:		/* -30 */
+		errstr = "Readonly filesystem";
+		break;
+	case -EOPNOTSUPP:	/* -95 */
+		errstr = "Operation not supported";
+		break;
+	case -EUCLEAN:		/* -117 */
+		errstr = "Filesystem corrupted";
+		break;
+	case -EDQUOT:		/* -122 */
+		errstr = "Quota exceeded";
 		break;
 	}
 
@@ -105,6 +175,7 @@ void __btrfs_handle_fs_error(struct btrfs_fs_info *fs_info, const char *function
 {
 	struct super_block *sb = fs_info->sb;
 #ifdef CONFIG_PRINTK
+	char statestr[STATE_STRING_BUF_LEN];
 	const char *errstr;
 #endif
 
@@ -117,6 +188,7 @@ void __btrfs_handle_fs_error(struct btrfs_fs_info *fs_info, const char *function
 
 #ifdef CONFIG_PRINTK
 	errstr = btrfs_decode_error(errno);
+	btrfs_state_to_string(fs_info, statestr);
 	if (fmt) {
 		struct va_format vaf;
 		va_list args;
@@ -125,12 +197,12 @@ void __btrfs_handle_fs_error(struct btrfs_fs_info *fs_info, const char *function
 		vaf.fmt = fmt;
 		vaf.va = &args;
 
-		pr_crit("BTRFS: error (device %s) in %s:%d: errno=%d %s (%pV)\n",
-			sb->s_id, function, line, errno, errstr, &vaf);
+		pr_crit("BTRFS: error (device %s%s) in %s:%d: errno=%d %s (%pV)\n",
+			sb->s_id, statestr, function, line, errno, errstr, &vaf);
 		va_end(args);
 	} else {
-		pr_crit("BTRFS: error (device %s) in %s:%d: errno=%d %s\n",
-			sb->s_id, function, line, errno, errstr);
+		pr_crit("BTRFS: error (device %s%s) in %s:%d: errno=%d %s\n",
+			sb->s_id, statestr, function, line, errno, errstr);
 	}
 #endif
 
@@ -150,7 +222,7 @@ void __btrfs_handle_fs_error(struct btrfs_fs_info *fs_info, const char *function
 	btrfs_discard_stop(fs_info);
 
 	/* btrfs handle error by forcing the filesystem readonly */
-	sb->s_flags |= SB_RDONLY;
+	btrfs_set_sb_rdonly(sb);
 	btrfs_info(fs_info, "forced readonly");
 	/*
 	 * Note that a running device replace operation is not canceled here
@@ -190,7 +262,7 @@ static struct ratelimit_state printk_limits[] = {
 	RATELIMIT_STATE_INIT(printk_limits[7], DEFAULT_RATELIMIT_INTERVAL, 100),
 };
 
-void __cold btrfs_printk(const struct btrfs_fs_info *fs_info, const char *fmt, ...)
+void __cold _btrfs_printk(const struct btrfs_fs_info *fs_info, const char *fmt, ...)
 {
 	char lvl[PRINTK_MAX_SINGLE_HEADER_LEN + 1] = "\0";
 	struct va_format vaf;
@@ -216,14 +288,48 @@ void __cold btrfs_printk(const struct btrfs_fs_info *fs_info, const char *fmt, .
 	vaf.fmt = fmt;
 	vaf.va = &args;
 
-	if (__ratelimit(ratelimit))
-		printk("%sBTRFS %s (device %s): %pV\n", lvl, type,
-			fs_info ? fs_info->sb->s_id : "<unknown>", &vaf);
+	if (__ratelimit(ratelimit)) {
+		if (fs_info) {
+			char statestr[STATE_STRING_BUF_LEN];
+
+			btrfs_state_to_string(fs_info, statestr);
+			_printk("%sBTRFS %s (device %s%s): %pV\n", lvl, type,
+				fs_info->sb->s_id, statestr, &vaf);
+		} else {
+			_printk("%sBTRFS %s: %pV\n", lvl, type, &vaf);
+		}
+	}
 
 	va_end(args);
 }
 #endif
 
+#if BITS_PER_LONG == 32
+void __cold btrfs_warn_32bit_limit(struct btrfs_fs_info *fs_info)
+{
+	if (!test_and_set_bit(BTRFS_FS_32BIT_WARN, &fs_info->flags)) {
+		btrfs_warn(fs_info, "reaching 32bit limit for logical addresses");
+		btrfs_warn(fs_info,
+"due to page cache limit on 32bit systems, btrfs can't access metadata at or beyond %lluT",
+			   BTRFS_32BIT_MAX_FILE_SIZE >> 40);
+		btrfs_warn(fs_info,
+			   "please consider upgrading to 64bit kernel/hardware");
+	}
+}
+
+void __cold btrfs_err_32bit_limit(struct btrfs_fs_info *fs_info)
+{
+	if (!test_and_set_bit(BTRFS_FS_32BIT_ERROR, &fs_info->flags)) {
+		btrfs_err(fs_info, "reached 32bit limit for logical addresses");
+		btrfs_err(fs_info,
+"due to page cache limit on 32bit systems, metadata beyond %lluT can't be accessed",
+			  BTRFS_32BIT_MAX_FILE_SIZE >> 40);
+		btrfs_err(fs_info,
+			   "please consider upgrading to 64bit kernel/hardware");
+	}
+}
+#endif
+
 /*
  * We only mark the transaction aborted and then set the file system read-only.
  * This will prevent new transactions from starting or trying to join this
@@ -240,23 +346,14 @@ void __cold btrfs_printk(const struct btrfs_fs_info *fs_info, const char *fmt, .
 __cold
 void __btrfs_abort_transaction(struct btrfs_trans_handle *trans,
 			       const char *function,
-			       unsigned int line, int errno)
+			       unsigned int line, int errno, bool first_hit)
 {
 	struct btrfs_fs_info *fs_info = trans->fs_info;
 
-	trans->aborted = errno;
-	/* Nothing used. The other threads that have joined this
-	 * transaction may be able to continue. */
-	if (!trans->dirty && list_empty(&trans->new_bgs)) {
-		const char *errstr;
-
-		errstr = btrfs_decode_error(errno);
-		btrfs_warn(fs_info,
-		           "%s:%d: Aborting unused transaction(%s).",
-		           function, line, errstr);
-		return;
-	}
+	WRITE_ONCE(trans->aborted, errno);
 	WRITE_ONCE(trans->transaction->aborted, errno);
+	if (first_hit && errno == -ENOSPC)
+		btrfs_dump_space_info_for_trans_abort(fs_info);
 	/* Wake up anybody who may be waiting on this transaction */
 	wake_up(&fs_info->transaction_wait);
 	wake_up(&fs_info->transaction_blocked_wait);
@@ -309,7 +406,6 @@ enum {
 	Opt_device,
 	Opt_fatal_errors,
 	Opt_flushoncommit, Opt_noflushoncommit,
-	Opt_inode_cache, Opt_noinode_cache,
 	Opt_max_inline,
 	Opt_barrier, Opt_nobarrier,
 	Opt_datacow, Opt_nodatacow,
@@ -317,7 +413,6 @@ enum {
 	Opt_defrag, Opt_nodefrag,
 	Opt_discard, Opt_nodiscard,
 	Opt_discard_mode,
-	Opt_nologreplay,
 	Opt_norecovery,
 	Opt_ratio,
 	Opt_rescan_uuid_tree,
@@ -331,13 +426,19 @@ enum {
 	Opt_subvolid,
 	Opt_thread_pool,
 	Opt_treelog, Opt_notreelog,
-	Opt_usebackuproot,
 	Opt_user_subvol_rm_allowed,
 
+	/* Rescue options */
+	Opt_rescue,
+	Opt_usebackuproot,
+	Opt_nologreplay,
+	Opt_ignorebadroots,
+	Opt_ignoredatacsums,
+	Opt_rescue_all,
+
 	/* Deprecated options */
-	Opt_alloc_start,
 	Opt_recovery,
-	Opt_subvolrootid,
+	Opt_inode_cache, Opt_noinode_cache,
 
 	/* Debugging options */
 	Opt_check_integrity,
@@ -381,7 +482,6 @@ static const match_table_t tokens = {
 	{Opt_discard, "discard"},
 	{Opt_discard_mode, "discard=%s"},
 	{Opt_nodiscard, "nodiscard"},
-	{Opt_nologreplay, "nologreplay"},
 	{Opt_norecovery, "norecovery"},
 	{Opt_ratio, "metadata_ratio=%u"},
 	{Opt_rescan_uuid_tree, "rescan_uuid_tree"},
@@ -399,13 +499,17 @@ static const match_table_t tokens = {
 	{Opt_thread_pool, "thread_pool=%u"},
 	{Opt_treelog, "treelog"},
 	{Opt_notreelog, "notreelog"},
-	{Opt_usebackuproot, "usebackuproot"},
 	{Opt_user_subvol_rm_allowed, "user_subvol_rm_allowed"},
 
+	/* Rescue options */
+	{Opt_rescue, "rescue=%s"},
+	/* Deprecated, with alias rescue=nologreplay */
+	{Opt_nologreplay, "nologreplay"},
+	/* Deprecated, with alias rescue=usebackuproot */
+	{Opt_usebackuproot, "usebackuproot"},
+
 	/* Deprecated options */
-	{Opt_alloc_start, "alloc_start=%s"},
 	{Opt_recovery, "recovery"},
-	{Opt_subvolrootid, "subvolrootid=%d"},
 
 	/* Debugging options */
 	{Opt_check_integrity, "check_int"},
@@ -424,6 +528,88 @@ static const match_table_t tokens = {
 	{Opt_err, NULL},
 };
 
+static const match_table_t rescue_tokens = {
+	{Opt_usebackuproot, "usebackuproot"},
+	{Opt_nologreplay, "nologreplay"},
+	{Opt_ignorebadroots, "ignorebadroots"},
+	{Opt_ignorebadroots, "ibadroots"},
+	{Opt_ignoredatacsums, "ignoredatacsums"},
+	{Opt_ignoredatacsums, "idatacsums"},
+	{Opt_rescue_all, "all"},
+	{Opt_err, NULL},
+};
+
+static bool check_ro_option(struct btrfs_fs_info *fs_info, unsigned long opt,
+			    const char *opt_name)
+{
+	if (fs_info->mount_opt & opt) {
+		btrfs_err(fs_info, "%s must be used with ro mount option",
+			  opt_name);
+		return true;
+	}
+	return false;
+}
+
+static int parse_rescue_options(struct btrfs_fs_info *info, const char *options)
+{
+	char *opts;
+	char *orig;
+	char *p;
+	substring_t args[MAX_OPT_ARGS];
+	int ret = 0;
+
+	opts = kstrdup(options, GFP_KERNEL);
+	if (!opts)
+		return -ENOMEM;
+	orig = opts;
+
+	while ((p = strsep(&opts, ":")) != NULL) {
+		int token;
+
+		if (!*p)
+			continue;
+		token = match_token(p, rescue_tokens, args);
+		switch (token){
+		case Opt_usebackuproot:
+			btrfs_info(info,
+				   "trying to use backup root at mount time");
+			btrfs_set_opt(info->mount_opt, USEBACKUPROOT);
+			break;
+		case Opt_nologreplay:
+			btrfs_set_and_info(info, NOLOGREPLAY,
+					   "disabling log replay at mount time");
+			break;
+		case Opt_ignorebadroots:
+			btrfs_set_and_info(info, IGNOREBADROOTS,
+					   "ignoring bad roots");
+			break;
+		case Opt_ignoredatacsums:
+			btrfs_set_and_info(info, IGNOREDATACSUMS,
+					   "ignoring data csums");
+			break;
+		case Opt_rescue_all:
+			btrfs_info(info, "enabling all of the rescue options");
+			btrfs_set_and_info(info, IGNOREDATACSUMS,
+					   "ignoring data csums");
+			btrfs_set_and_info(info, IGNOREBADROOTS,
+					   "ignoring bad roots");
+			btrfs_set_and_info(info, NOLOGREPLAY,
+					   "disabling log replay at mount time");
+			break;
+		case Opt_err:
+			btrfs_info(info, "unrecognized rescue option '%s'", p);
+			ret = -EINVAL;
+			goto out;
+		default:
+			break;
+		}
+
+	}
+out:
+	kfree(orig);
+	return ret;
+}
+
 /*
  * Regular mount options parser.  Everything that is needed only when
  * reading in a new superblock is parsed here.
@@ -434,20 +620,27 @@ int btrfs_parse_options(struct btrfs_fs_info *info, char *options,
 {
 	substring_t args[MAX_OPT_ARGS];
 	char *p, *num;
-	u64 cache_gen;
 	int intarg;
 	int ret = 0;
 	char *compress_type;
 	bool compress_force = false;
 	enum btrfs_compression_type saved_compress_type;
+	int saved_compress_level;
 	bool saved_compress_force;
 	int no_compress = 0;
+	const bool remounting = test_bit(BTRFS_FS_STATE_REMOUNTING, &info->fs_state);
 
-	cache_gen = btrfs_super_cache_generation(info->super_copy);
 	if (btrfs_fs_compat_ro(info, FREE_SPACE_TREE))
 		btrfs_set_opt(info->mount_opt, FREE_SPACE_TREE);
-	else if (cache_gen)
-		btrfs_set_opt(info->mount_opt, SPACE_CACHE);
+	else if (btrfs_free_space_cache_v1_active(info)) {
+		if (btrfs_is_zoned(info)) {
+			btrfs_info(info,
+			"zoned: clearing existing space cache");
+			btrfs_set_super_cache_generation(info->super_copy, 0);
+		} else {
+			btrfs_set_opt(info->mount_opt, SPACE_CACHE);
+		}
+	}
 
 	/*
 	 * Even the options are empty, we still need to do extra check
@@ -470,7 +663,6 @@ int btrfs_parse_options(struct btrfs_fs_info *info, char *options,
 		case Opt_subvol:
 		case Opt_subvol_empty:
 		case Opt_subvolid:
-		case Opt_subvolrootid:
 		case Opt_device:
 			/*
 			 * These are parsed by btrfs_parse_subvol_options or
@@ -514,7 +706,7 @@ int btrfs_parse_options(struct btrfs_fs_info *info, char *options,
 		case Opt_compress_force:
 		case Opt_compress_force_type:
 			compress_force = true;
-			/* Fallthrough */
+			fallthrough;
 		case Opt_compress:
 		case Opt_compress_type:
 			saved_compress_type = btrfs_test_opt(info,
@@ -522,6 +714,7 @@ int btrfs_parse_options(struct btrfs_fs_info *info, char *options,
 				info->compress_type : BTRFS_COMPRESS_NONE;
 			saved_compress_force =
 				btrfs_test_opt(info, FORCE_COMPRESS);
+			saved_compress_level = info->compress_level;
 			if (token == Opt_compress ||
 			    token == Opt_compress_force ||
 			    strncmp(args[0].from, "zlib", 4) == 0) {
@@ -547,6 +740,7 @@ int btrfs_parse_options(struct btrfs_fs_info *info, char *options,
 			} else if (strncmp(args[0].from, "lzo", 3) == 0) {
 				compress_type = "lzo";
 				info->compress_type = BTRFS_COMPRESS_LZO;
+				info->compress_level = 0;
 				btrfs_set_opt(info->mount_opt, COMPRESS);
 				btrfs_clear_opt(info->mount_opt, NODATACOW);
 				btrfs_clear_opt(info->mount_opt, NODATASUM);
@@ -566,11 +760,15 @@ int btrfs_parse_options(struct btrfs_fs_info *info, char *options,
 				no_compress = 0;
 			} else if (strncmp(args[0].from, "no", 2) == 0) {
 				compress_type = "no";
+				info->compress_level = 0;
+				info->compress_type = 0;
 				btrfs_clear_opt(info->mount_opt, COMPRESS);
 				btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS);
 				compress_force = false;
 				no_compress++;
 			} else {
+				btrfs_err(info, "unrecognized compression value %s",
+					  args[0].from);
 				ret = -EINVAL;
 				goto out;
 			}
@@ -586,11 +784,11 @@ int btrfs_parse_options(struct btrfs_fs_info *info, char *options,
 				 */
 				btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS);
 			}
-			if ((btrfs_test_opt(info, COMPRESS) &&
-			     (info->compress_type != saved_compress_type ||
-			      compress_force != saved_compress_force)) ||
-			    (!btrfs_test_opt(info, COMPRESS) &&
-			     no_compress == 1)) {
+			if (no_compress == 1) {
+				btrfs_info(info, "use no compression");
+			} else if ((info->compress_type != saved_compress_type) ||
+				   (compress_force != saved_compress_force) ||
+				   (info->compress_level != saved_compress_level)) {
 				btrfs_info(info, "%s %s compression, level %d",
 					   (compress_force) ? "force" : "use",
 					   compress_type, info->compress_level);
@@ -613,7 +811,7 @@ int btrfs_parse_options(struct btrfs_fs_info *info, char *options,
 			btrfs_set_opt(info->mount_opt, NOSSD);
 			btrfs_clear_and_info(info, SSD,
 					     "not using ssd optimizations");
-			/* Fallthrough */
+			fallthrough;
 		case Opt_nossd_spread:
 			btrfs_clear_and_info(info, SSD_SPREAD,
 					     "not using spread ssd allocation scheme");
@@ -629,8 +827,11 @@ int btrfs_parse_options(struct btrfs_fs_info *info, char *options,
 		case Opt_thread_pool:
 			ret = match_int(&args[0], &intarg);
 			if (ret) {
+				btrfs_err(info, "unrecognized thread_pool value %s",
+					  args[0].from);
 				goto out;
 			} else if (intarg == 0) {
+				btrfs_err(info, "invalid value 0 for thread_pool");
 				ret = -EINVAL;
 				goto out;
 			}
@@ -654,10 +855,6 @@ int btrfs_parse_options(struct btrfs_fs_info *info, char *options,
 				goto out;
 			}
 			break;
-		case Opt_alloc_start:
-			btrfs_info(info,
-				"option alloc_start is obsolete, ignored");
-			break;
 		case Opt_acl:
 #ifdef CONFIG_BTRFS_FS_POSIX_ACL
 			info->sb->s_flags |= SB_POSIXACL;
@@ -680,6 +877,8 @@ int btrfs_parse_options(struct btrfs_fs_info *info, char *options,
 			break;
 		case Opt_norecovery:
 		case Opt_nologreplay:
+			btrfs_warn(info,
+		"'nologreplay' is deprecated, use 'rescue=nologreplay' instead");
 			btrfs_set_and_info(info, NOLOGREPLAY,
 					   "disabling log replay at mount time");
 			break;
@@ -693,8 +892,11 @@ int btrfs_parse_options(struct btrfs_fs_info *info, char *options,
 			break;
 		case Opt_ratio:
 			ret = match_int(&args[0], &intarg);
-			if (ret)
+			if (ret) {
+				btrfs_err(info, "unrecognized metadata_ratio value %s",
+					  args[0].from);
 				goto out;
+			}
 			info->metadata_ratio = intarg;
 			btrfs_info(info, "metadata ratio %u",
 				   info->metadata_ratio);
@@ -711,6 +913,8 @@ int btrfs_parse_options(struct btrfs_fs_info *info, char *options,
 				btrfs_set_and_info(info, DISCARD_ASYNC,
 						   "turning on async discard");
 			} else {
+				btrfs_err(info, "unrecognized discard mode value %s",
+					  args[0].from);
 				ret = -EINVAL;
 				goto out;
 			}
@@ -723,6 +927,14 @@ int btrfs_parse_options(struct btrfs_fs_info *info, char *options,
 			break;
 		case Opt_space_cache:
 		case Opt_space_cache_version:
+			/*
+			 * We already set FREE_SPACE_TREE above because we have
+			 * compat_ro(FREE_SPACE_TREE) set, and we aren't going
+			 * to allow v1 to be set for extent tree v2, simply
+			 * ignore this setting if we're extent tree v2.
+			 */
+			if (btrfs_fs_incompat(info, EXTENT_TREE_V2))
+				break;
 			if (token == Opt_space_cache ||
 			    strcmp(args[0].from, "v1") == 0) {
 				btrfs_clear_opt(info->mount_opt,
@@ -735,6 +947,8 @@ int btrfs_parse_options(struct btrfs_fs_info *info, char *options,
 				btrfs_set_and_info(info, FREE_SPACE_TREE,
 						   "enabling free space tree");
 			} else {
+				btrfs_err(info, "unrecognized space_cache value %s",
+					  args[0].from);
 				ret = -EINVAL;
 				goto out;
 			}
@@ -743,6 +957,12 @@ int btrfs_parse_options(struct btrfs_fs_info *info, char *options,
 			btrfs_set_opt(info->mount_opt, RESCAN_UUID_TREE);
 			break;
 		case Opt_no_space_cache:
+			/*
+			 * We cannot operate without the free space tree with
+			 * extent tree v2, ignore this option.
+			 */
+			if (btrfs_fs_incompat(info, EXTENT_TREE_V2))
+				break;
 			if (btrfs_test_opt(info, SPACE_CACHE)) {
 				btrfs_clear_and_info(info, SPACE_CACHE,
 					     "disabling disk space caching");
@@ -753,14 +973,17 @@ int btrfs_parse_options(struct btrfs_fs_info *info, char *options,
 			}
 			break;
 		case Opt_inode_cache:
-			btrfs_set_pending_and_info(info, INODE_MAP_CACHE,
-					   "enabling inode map caching");
-			break;
 		case Opt_noinode_cache:
-			btrfs_clear_pending_and_info(info, INODE_MAP_CACHE,
-					     "disabling inode map caching");
+			btrfs_warn(info,
+	"the 'inode_cache' option is deprecated and has no effect since 5.11");
 			break;
 		case Opt_clear_cache:
+			/*
+			 * We cannot clear the free space tree with extent tree
+			 * v2, ignore this option.
+			 */
+			if (btrfs_fs_incompat(info, EXTENT_TREE_V2))
+				break;
 			btrfs_set_and_info(info, CLEAR_CACHE,
 					   "force clearing of disk cache");
 			break;
@@ -782,10 +1005,11 @@ int btrfs_parse_options(struct btrfs_fs_info *info, char *options,
 					     "disabling auto defrag");
 			break;
 		case Opt_recovery:
-			btrfs_warn(info,
-				   "'recovery' is deprecated, use 'usebackuproot' instead");
-			/* fall through */
 		case Opt_usebackuproot:
+			btrfs_warn(info,
+			"'%s' is deprecated, use 'rescue=usebackuproot' instead",
+				   token == Opt_recovery ? "recovery" :
+				   "usebackuproot");
 			btrfs_info(info,
 				   "trying to use backup root at mount time");
 			btrfs_set_opt(info->mount_opt, USEBACKUPROOT);
@@ -797,8 +1021,7 @@ int btrfs_parse_options(struct btrfs_fs_info *info, char *options,
 		case Opt_check_integrity_including_extent_data:
 			btrfs_info(info,
 				   "enabling check integrity including extent data");
-			btrfs_set_opt(info->mount_opt,
-				      CHECK_INTEGRITY_INCLUDING_EXTENT_DATA);
+			btrfs_set_opt(info->mount_opt, CHECK_INTEGRITY_DATA);
 			btrfs_set_opt(info->mount_opt, CHECK_INTEGRITY);
 			break;
 		case Opt_check_integrity:
@@ -807,8 +1030,12 @@ int btrfs_parse_options(struct btrfs_fs_info *info, char *options,
 			break;
 		case Opt_check_integrity_print_mask:
 			ret = match_int(&args[0], &intarg);
-			if (ret)
+			if (ret) {
+				btrfs_err(info,
+				"unrecognized check_integrity_print_mask value %s",
+					args[0].from);
 				goto out;
+			}
 			info->check_integrity_print_mask = intarg;
 			btrfs_info(info, "check_integrity_print_mask 0x%x",
 				   info->check_integrity_print_mask);
@@ -823,13 +1050,15 @@ int btrfs_parse_options(struct btrfs_fs_info *info, char *options,
 			goto out;
 #endif
 		case Opt_fatal_errors:
-			if (strcmp(args[0].from, "panic") == 0)
+			if (strcmp(args[0].from, "panic") == 0) {
 				btrfs_set_opt(info->mount_opt,
 					      PANIC_ON_FATAL_ERROR);
-			else if (strcmp(args[0].from, "bug") == 0)
+			} else if (strcmp(args[0].from, "bug") == 0) {
 				btrfs_clear_opt(info->mount_opt,
 					      PANIC_ON_FATAL_ERROR);
-			else {
+			} else {
+				btrfs_err(info, "unrecognized fatal_errors value %s",
+					  args[0].from);
 				ret = -EINVAL;
 				goto out;
 			}
@@ -837,8 +1066,12 @@ int btrfs_parse_options(struct btrfs_fs_info *info, char *options,
 		case Opt_commit_interval:
 			intarg = 0;
 			ret = match_int(&args[0], &intarg);
-			if (ret)
+			if (ret) {
+				btrfs_err(info, "unrecognized commit_interval value %s",
+					  args[0].from);
+				ret = -EINVAL;
 				goto out;
+			}
 			if (intarg == 0) {
 				btrfs_info(info,
 					   "using default commit interval %us",
@@ -850,6 +1083,14 @@ int btrfs_parse_options(struct btrfs_fs_info *info, char *options,
 			}
 			info->commit_interval = intarg;
 			break;
+		case Opt_rescue:
+			ret = parse_rescue_options(info, args[0].from);
+			if (ret < 0) {
+				btrfs_err(info, "unrecognized rescue value %s",
+					  args[0].from);
+				goto out;
+			}
+			break;
 #ifdef CONFIG_BTRFS_DEBUG
 		case Opt_fragment_all:
 			btrfs_info(info, "fragmenting all space");
@@ -873,7 +1114,7 @@ int btrfs_parse_options(struct btrfs_fs_info *info, char *options,
 			break;
 #endif
 		case Opt_err:
-			btrfs_info(info, "unrecognized mount option '%s'", p);
+			btrfs_err(info, "unrecognized mount option '%s'", p);
 			ret = -EINVAL;
 			goto out;
 		default:
@@ -881,14 +1122,14 @@ int btrfs_parse_options(struct btrfs_fs_info *info, char *options,
 		}
 	}
 check:
-	/*
-	 * Extra check for current option against current flag
-	 */
-	if (btrfs_test_opt(info, NOLOGREPLAY) && !(new_flags & SB_RDONLY)) {
-		btrfs_err(info,
-			  "nologreplay must be used with ro mount option");
+	/* We're read-only, don't have to check. */
+	if (new_flags & SB_RDONLY)
+		goto out;
+
+	if (check_ro_option(info, BTRFS_MOUNT_NOLOGREPLAY, "nologreplay") ||
+	    check_ro_option(info, BTRFS_MOUNT_IGNOREBADROOTS, "ignorebadroots") ||
+	    check_ro_option(info, BTRFS_MOUNT_IGNOREDATACSUMS, "ignoredatacsums"))
 		ret = -EINVAL;
-	}
 out:
 	if (btrfs_fs_compat_ro(info, FREE_SPACE_TREE) &&
 	    !btrfs_test_opt(info, FREE_SPACE_TREE) &&
@@ -897,10 +1138,14 @@ out:
 		ret = -EINVAL;
 
 	}
-	if (!ret && btrfs_test_opt(info, SPACE_CACHE))
-		btrfs_info(info, "disk space caching is enabled");
-	if (!ret && btrfs_test_opt(info, FREE_SPACE_TREE))
-		btrfs_info(info, "using free space tree");
+	if (!ret)
+		ret = btrfs_check_mountopts_zoned(info);
+	if (!ret && !remounting) {
+		if (btrfs_test_opt(info, SPACE_CACHE))
+			btrfs_info(info, "disk space caching is enabled");
+		if (btrfs_test_opt(info, FREE_SPACE_TREE))
+			btrfs_info(info, "using free space tree");
+	}
 	return ret;
 }
 
@@ -1011,9 +1256,6 @@ static int btrfs_parse_subvol_options(const char *options, char **subvol_name,
 
 			*subvol_objectid = subvolid;
 			break;
-		case Opt_subvolrootid:
-			pr_warn("BTRFS: 'subvolrootid' mount option is deprecated and has no effect\n");
-			break;
 		default:
 			break;
 		}
@@ -1024,11 +1266,11 @@ out:
 	return error;
 }
 
-static char *get_subvol_name_from_objectid(struct btrfs_fs_info *fs_info,
-					   u64 subvol_objectid)
+char *btrfs_get_subvol_name_from_objectid(struct btrfs_fs_info *fs_info,
+					  u64 subvol_objectid)
 {
 	struct btrfs_root *root = fs_info->tree_root;
-	struct btrfs_root *fs_root;
+	struct btrfs_root *fs_root = NULL;
 	struct btrfs_root_ref *root_ref;
 	struct btrfs_inode_ref *inode_ref;
 	struct btrfs_key key;
@@ -1043,7 +1285,6 @@ static char *get_subvol_name_from_objectid(struct btrfs_fs_info *fs_info,
 		ret = -ENOMEM;
 		goto err;
 	}
-	path->leave_spinning = 1;
 
 	name = kmalloc(PATH_MAX, GFP_KERNEL);
 	if (!name) {
@@ -1062,21 +1303,14 @@ static char *get_subvol_name_from_objectid(struct btrfs_fs_info *fs_info,
 		key.type = BTRFS_ROOT_BACKREF_KEY;
 		key.offset = (u64)-1;
 
-		ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
+		ret = btrfs_search_backwards(root, &key, path);
 		if (ret < 0) {
 			goto err;
 		} else if (ret > 0) {
-			ret = btrfs_previous_item(root, path, subvol_objectid,
-						  BTRFS_ROOT_BACKREF_KEY);
-			if (ret < 0) {
-				goto err;
-			} else if (ret > 0) {
-				ret = -ENOENT;
-				goto err;
-			}
+			ret = -ENOENT;
+			goto err;
 		}
 
-		btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
 		subvol_objectid = key.offset;
 
 		root_ref = btrfs_item_ptr(path->nodes[0], path->slots[0],
@@ -1093,12 +1327,10 @@ static char *get_subvol_name_from_objectid(struct btrfs_fs_info *fs_info,
 		dirid = btrfs_root_ref_dirid(path->nodes[0], root_ref);
 		btrfs_release_path(path);
 
-		key.objectid = subvol_objectid;
-		key.type = BTRFS_ROOT_ITEM_KEY;
-		key.offset = (u64)-1;
-		fs_root = btrfs_read_fs_root_no_name(fs_info, &key);
+		fs_root = btrfs_get_fs_root(fs_info, subvol_objectid, true);
 		if (IS_ERR(fs_root)) {
 			ret = PTR_ERR(fs_root);
+			fs_root = NULL;
 			goto err;
 		}
 
@@ -1111,21 +1343,14 @@ static char *get_subvol_name_from_objectid(struct btrfs_fs_info *fs_info,
 			key.type = BTRFS_INODE_REF_KEY;
 			key.offset = (u64)-1;
 
-			ret = btrfs_search_slot(NULL, fs_root, &key, path, 0, 0);
+			ret = btrfs_search_backwards(fs_root, &key, path);
 			if (ret < 0) {
 				goto err;
 			} else if (ret > 0) {
-				ret = btrfs_previous_item(fs_root, path, dirid,
-							  BTRFS_INODE_REF_KEY);
-				if (ret < 0) {
-					goto err;
-				} else if (ret > 0) {
-					ret = -ENOENT;
-					goto err;
-				}
+				ret = -ENOENT;
+				goto err;
 			}
 
-			btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
 			dirid = key.offset;
 
 			inode_ref = btrfs_item_ptr(path->nodes[0],
@@ -1143,6 +1368,8 @@ static char *get_subvol_name_from_objectid(struct btrfs_fs_info *fs_info,
 			ptr[0] = '/';
 			btrfs_release_path(path);
 		}
+		btrfs_put_root(fs_root);
+		fs_root = NULL;
 	}
 
 	btrfs_free_path(path);
@@ -1155,6 +1382,7 @@ static char *get_subvol_name_from_objectid(struct btrfs_fs_info *fs_info,
 	return name;
 
 err:
+	btrfs_put_root(fs_root);
 	btrfs_free_path(path);
 	kfree(name);
 	return ERR_PTR(ret);
@@ -1171,7 +1399,6 @@ static int get_default_subvol_objectid(struct btrfs_fs_info *fs_info, u64 *objec
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
-	path->leave_spinning = 1;
 
 	/*
 	 * Find the "default" dir item which points to the root item that we
@@ -1207,7 +1434,6 @@ static int btrfs_fill_super(struct super_block *sb,
 {
 	struct inode *inode;
 	struct btrfs_fs_info *fs_info = btrfs_sb(sb);
-	struct btrfs_key key;
 	int err;
 
 	sb->s_maxbytes = MAX_LFS_FILESIZE;
@@ -1215,6 +1441,9 @@ static int btrfs_fill_super(struct super_block *sb,
 	sb->s_op = &btrfs_super_ops;
 	sb->s_d_op = &btrfs_dentry_operations;
 	sb->s_export_op = &btrfs_export_ops;
+#ifdef CONFIG_FS_VERITY
+	sb->s_vop = &btrfs_verityops;
+#endif
 	sb->s_xattr = btrfs_xattr_handlers;
 	sb->s_time_gran = 1;
 #ifdef CONFIG_BTRFS_FS_POSIX_ACL
@@ -1235,10 +1464,7 @@ static int btrfs_fill_super(struct super_block *sb,
 		return err;
 	}
 
-	key.objectid = BTRFS_FIRST_FREE_OBJECTID;
-	key.type = BTRFS_INODE_ITEM_KEY;
-	key.offset = 0;
-	inode = btrfs_iget(sb, &key, fs_info->fs_root);
+	inode = btrfs_iget(sb, BTRFS_FIRST_FREE_OBJECTID, fs_info->fs_root);
 	if (IS_ERR(inode)) {
 		err = PTR_ERR(inode);
 		goto fail_close;
@@ -1250,7 +1476,6 @@ static int btrfs_fill_super(struct super_block *sb,
 		goto fail_close;
 	}
 
-	cleancache_init_fs(sb);
 	sb->s_flags |= SB_ACTIVE;
 	return 0;
 
@@ -1302,10 +1527,18 @@ int btrfs_sync_fs(struct super_block *sb, int wait)
 	return btrfs_commit_transaction(trans);
 }
 
+static void print_rescue_option(struct seq_file *seq, const char *s, bool *printed)
+{
+	seq_printf(seq, "%s%s", (*printed) ? ":" : ",rescue=", s);
+	*printed = true;
+}
+
 static int btrfs_show_options(struct seq_file *seq, struct dentry *dentry)
 {
 	struct btrfs_fs_info *info = btrfs_sb(dentry->d_sb);
 	const char *compress_type;
+	const char *subvol_name;
+	bool printed = false;
 
 	if (btrfs_test_opt(info, DEGRADED))
 		seq_puts(seq, ",degraded");
@@ -1338,7 +1571,13 @@ static int btrfs_show_options(struct seq_file *seq, struct dentry *dentry)
 	if (btrfs_test_opt(info, NOTREELOG))
 		seq_puts(seq, ",notreelog");
 	if (btrfs_test_opt(info, NOLOGREPLAY))
-		seq_puts(seq, ",nologreplay");
+		print_rescue_option(seq, "nologreplay", &printed);
+	if (btrfs_test_opt(info, USEBACKUPROOT))
+		print_rescue_option(seq, "usebackuproot", &printed);
+	if (btrfs_test_opt(info, IGNOREBADROOTS))
+		print_rescue_option(seq, "ignorebadroots", &printed);
+	if (btrfs_test_opt(info, IGNOREDATACSUMS))
+		print_rescue_option(seq, "ignoredatacsums", &printed);
 	if (btrfs_test_opt(info, FLUSHONCOMMIT))
 		seq_puts(seq, ",flushoncommit");
 	if (btrfs_test_opt(info, DISCARD_SYNC))
@@ -1347,9 +1586,9 @@ static int btrfs_show_options(struct seq_file *seq, struct dentry *dentry)
 		seq_puts(seq, ",discard=async");
 	if (!(info->sb->s_flags & SB_POSIXACL))
 		seq_puts(seq, ",noacl");
-	if (btrfs_test_opt(info, SPACE_CACHE))
+	if (btrfs_free_space_cache_v1_active(info))
 		seq_puts(seq, ",space_cache");
-	else if (btrfs_test_opt(info, FREE_SPACE_TREE))
+	else if (btrfs_fs_compat_ro(info, FREE_SPACE_TREE))
 		seq_puts(seq, ",space_cache=v2");
 	else
 		seq_puts(seq, ",nospace_cache");
@@ -1363,12 +1602,10 @@ static int btrfs_show_options(struct seq_file *seq, struct dentry *dentry)
 		seq_puts(seq, ",enospc_debug");
 	if (btrfs_test_opt(info, AUTO_DEFRAG))
 		seq_puts(seq, ",autodefrag");
-	if (btrfs_test_opt(info, INODE_MAP_CACHE))
-		seq_puts(seq, ",inode_cache");
 	if (btrfs_test_opt(info, SKIP_BALANCE))
 		seq_puts(seq, ",skip_balance");
 #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
-	if (btrfs_test_opt(info, CHECK_INTEGRITY_INCLUDING_EXTENT_DATA))
+	if (btrfs_test_opt(info, CHECK_INTEGRITY_DATA))
 		seq_puts(seq, ",check_int_data");
 	else if (btrfs_test_opt(info, CHECK_INTEGRITY))
 		seq_puts(seq, ",check_int");
@@ -1392,8 +1629,13 @@ static int btrfs_show_options(struct seq_file *seq, struct dentry *dentry)
 		seq_puts(seq, ",ref_verify");
 	seq_printf(seq, ",subvolid=%llu",
 		  BTRFS_I(d_inode(dentry))->root->root_key.objectid);
-	seq_puts(seq, ",subvol=");
-	seq_dentry(seq, dentry, " \t\n\\");
+	subvol_name = btrfs_get_subvol_name_from_objectid(info,
+			BTRFS_I(d_inode(dentry))->root->root_key.objectid);
+	if (!IS_ERR(subvol_name)) {
+		seq_puts(seq, ",subvol=");
+		seq_escape(seq, subvol_name, " \t\n\\");
+		kfree(subvol_name);
+	}
 	return 0;
 }
 
@@ -1438,8 +1680,8 @@ static struct dentry *mount_subvol(const char *subvol_name, u64 subvol_objectid,
 				goto out;
 			}
 		}
-		subvol_name = get_subvol_name_from_objectid(btrfs_sb(mnt->mnt_sb),
-							    subvol_objectid);
+		subvol_name = btrfs_get_subvol_name_from_objectid(
+					btrfs_sb(mnt->mnt_sb), subvol_objectid);
 		if (IS_ERR(subvol_name)) {
 			root = ERR_CAST(subvol_name);
 			subvol_name = NULL;
@@ -1518,14 +1760,17 @@ static struct dentry *btrfs_mount_root(struct file_system_type *fs_type,
 	/*
 	 * Setup a dummy root and fs_info for test/set super.  This is because
 	 * we don't actually fill this stuff out until open_ctree, but we need
-	 * it for searching for existing supers, so this lets us do that and
-	 * then open_ctree will properly initialize everything later.
+	 * then open_ctree will properly initialize the file system specific
+	 * settings later.  btrfs_init_fs_info initializes the static elements
+	 * of the fs_info (locks and such) to make cleanup easier if we find a
+	 * superblock with our given fs_devices later on at sget() time.
 	 */
 	fs_info = kvzalloc(sizeof(struct btrfs_fs_info), GFP_KERNEL);
 	if (!fs_info) {
 		error = -ENOMEM;
 		goto error_sec_opts;
 	}
+	btrfs_init_fs_info(fs_info);
 
 	fs_info->super_copy = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_KERNEL);
 	fs_info->super_for_commit = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_KERNEL);
@@ -1561,7 +1806,7 @@ static struct dentry *btrfs_mount_root(struct file_system_type *fs_type,
 		goto error_close_devices;
 	}
 
-	bdev = fs_devices->latest_bdev;
+	bdev = fs_devices->latest_dev->bdev;
 	s = sget(fs_type, btrfs_test_super, btrfs_set_super, flags | SB_NOSEC,
 		 fs_info);
 	if (IS_ERR(s)) {
@@ -1571,11 +1816,13 @@ static struct dentry *btrfs_mount_root(struct file_system_type *fs_type,
 
 	if (s->s_root) {
 		btrfs_close_devices(fs_devices);
-		free_fs_info(fs_info);
+		btrfs_free_fs_info(fs_info);
 		if ((flags ^ s->s_flags) & SB_RDONLY)
 			error = -EBUSY;
 	} else {
 		snprintf(s->s_id, sizeof(s->s_id), "%pg", bdev);
+		shrinker_debugfs_rename(&s->s_shrink, "sb-%s:%s", fs_type->name,
+					s->s_id);
 		btrfs_sb(s)->bdev_holder = fs_type;
 		if (!strstr(crc32c_impl(), "generic"))
 			set_bit(BTRFS_FS_CSUM_IMPL_FAST, &fs_info->flags);
@@ -1594,7 +1841,7 @@ static struct dentry *btrfs_mount_root(struct file_system_type *fs_type,
 error_close_devices:
 	btrfs_close_devices(fs_devices);
 error_fs_info:
-	free_fs_info(fs_info);
+	btrfs_free_fs_info(fs_info);
 error_sec_opts:
 	security_free_mnt_opts(&new_sec_opts);
 	return ERR_PTR(error);
@@ -1689,23 +1936,12 @@ static void btrfs_resize_thread_pool(struct btrfs_fs_info *fs_info,
 	       old_pool_size, new_pool_size);
 
 	btrfs_workqueue_set_max(fs_info->workers, new_pool_size);
+	btrfs_workqueue_set_max(fs_info->hipri_workers, new_pool_size);
 	btrfs_workqueue_set_max(fs_info->delalloc_workers, new_pool_size);
 	btrfs_workqueue_set_max(fs_info->caching_workers, new_pool_size);
-	btrfs_workqueue_set_max(fs_info->endio_workers, new_pool_size);
-	btrfs_workqueue_set_max(fs_info->endio_meta_workers, new_pool_size);
-	btrfs_workqueue_set_max(fs_info->endio_meta_write_workers,
-				new_pool_size);
 	btrfs_workqueue_set_max(fs_info->endio_write_workers, new_pool_size);
 	btrfs_workqueue_set_max(fs_info->endio_freespace_worker, new_pool_size);
 	btrfs_workqueue_set_max(fs_info->delayed_workers, new_pool_size);
-	btrfs_workqueue_set_max(fs_info->readahead_workers, new_pool_size);
-	btrfs_workqueue_set_max(fs_info->scrub_wr_completion_workers,
-				new_pool_size);
-}
-
-static inline void btrfs_remount_prepare(struct btrfs_fs_info *fs_info)
-{
-	set_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state);
 }
 
 static inline void btrfs_remount_begin(struct btrfs_fs_info *fs_info,
@@ -1725,6 +1961,8 @@ static inline void btrfs_remount_begin(struct btrfs_fs_info *fs_info,
 static inline void btrfs_remount_cleanup(struct btrfs_fs_info *fs_info,
 					 unsigned long old_opts)
 {
+	const bool cache_opt = btrfs_test_opt(fs_info, SPACE_CACHE);
+
 	/*
 	 * We need to cleanup all defragable inodes if the autodefragment is
 	 * close or the filesystem is read only.
@@ -1742,13 +1980,14 @@ static inline void btrfs_remount_cleanup(struct btrfs_fs_info *fs_info,
 		 !btrfs_test_opt(fs_info, DISCARD_ASYNC))
 		btrfs_discard_cleanup(fs_info);
 
-	clear_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state);
+	/* If we toggled space cache */
+	if (cache_opt != btrfs_free_space_cache_v1_active(fs_info))
+		btrfs_set_free_space_cache_v1_active(fs_info, cache_opt);
 }
 
 static int btrfs_remount(struct super_block *sb, int *flags, char *data)
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(sb);
-	struct btrfs_root *root = fs_info->tree_root;
 	unsigned old_flags = sb->s_flags;
 	unsigned long old_opts = fs_info->mount_opt;
 	unsigned long old_compress_type = fs_info->compress_type;
@@ -1758,7 +1997,7 @@ static int btrfs_remount(struct super_block *sb, int *flags, char *data)
 	int ret;
 
 	sync_filesystem(sb);
-	btrfs_remount_prepare(fs_info);
+	set_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state);
 
 	if (data) {
 		void *new_sec_opts = NULL;
@@ -1775,10 +2014,30 @@ static int btrfs_remount(struct super_block *sb, int *flags, char *data)
 	if (ret)
 		goto restore;
 
+	ret = btrfs_check_features(fs_info, sb);
+	if (ret < 0)
+		goto restore;
+
 	btrfs_remount_begin(fs_info, old_opts, *flags);
 	btrfs_resize_thread_pool(fs_info,
 		fs_info->thread_pool_size, old_thread_pool_size);
 
+	if ((bool)btrfs_test_opt(fs_info, FREE_SPACE_TREE) !=
+	    (bool)btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE) &&
+	    (!sb_rdonly(sb) || (*flags & SB_RDONLY))) {
+		btrfs_warn(fs_info,
+		"remount supports changing free space tree only from ro to rw");
+		/* Make sure free space cache options match the state on disk */
+		if (btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE)) {
+			btrfs_set_opt(fs_info->mount_opt, FREE_SPACE_TREE);
+			btrfs_clear_opt(fs_info->mount_opt, SPACE_CACHE);
+		}
+		if (btrfs_free_space_cache_v1_active(fs_info)) {
+			btrfs_clear_opt(fs_info->mount_opt, FREE_SPACE_TREE);
+			btrfs_set_opt(fs_info->mount_opt, SPACE_CACHE);
+		}
+	}
+
 	if ((bool)(*flags & SB_RDONLY) == sb_rdonly(sb))
 		goto out;
 
@@ -1788,6 +2047,7 @@ static int btrfs_remount(struct super_block *sb, int *flags, char *data)
 		 * the filesystem is busy.
 		 */
 		cancel_work_sync(&fs_info->async_reclaim_work);
+		cancel_work_sync(&fs_info->async_data_reclaim_work);
 
 		btrfs_discard_cleanup(fs_info);
 
@@ -1796,7 +2056,7 @@ static int btrfs_remount(struct super_block *sb, int *flags, char *data)
 		/* avoid complains from lockdep et al. */
 		up(&fs_info->uuid_tree_rescan_sem);
 
-		sb->s_flags |= SB_RDONLY;
+		btrfs_set_sb_rdonly(sb);
 
 		/*
 		 * Setting SB_RDONLY will put the cleaner thread to
@@ -1807,15 +2067,47 @@ static int btrfs_remount(struct super_block *sb, int *flags, char *data)
 		 */
 		btrfs_delete_unused_bgs(fs_info);
 
+		/*
+		 * The cleaner task could be already running before we set the
+		 * flag BTRFS_FS_STATE_RO (and SB_RDONLY in the superblock).
+		 * We must make sure that after we finish the remount, i.e. after
+		 * we call btrfs_commit_super(), the cleaner can no longer start
+		 * a transaction - either because it was dropping a dead root,
+		 * running delayed iputs or deleting an unused block group (the
+		 * cleaner picked a block group from the list of unused block
+		 * groups before we were able to in the previous call to
+		 * btrfs_delete_unused_bgs()).
+		 */
+		wait_on_bit(&fs_info->flags, BTRFS_FS_CLEANER_RUNNING,
+			    TASK_UNINTERRUPTIBLE);
+
+		/*
+		 * We've set the superblock to RO mode, so we might have made
+		 * the cleaner task sleep without running all pending delayed
+		 * iputs. Go through all the delayed iputs here, so that if an
+		 * unmount happens without remounting RW we don't end up at
+		 * finishing close_ctree() with a non-empty list of delayed
+		 * iputs.
+		 */
+		btrfs_run_delayed_iputs(fs_info);
+
 		btrfs_dev_replace_suspend_for_unmount(fs_info);
 		btrfs_scrub_cancel(fs_info);
 		btrfs_pause_balance(fs_info);
 
+		/*
+		 * Pause the qgroup rescan worker if it is running. We don't want
+		 * it to be still running after we are in RO mode, as after that,
+		 * by the time we unmount, it might have left a transaction open,
+		 * so we would leak the transaction and/or crash.
+		 */
+		btrfs_qgroup_wait_for_completion(fs_info, false);
+
 		ret = btrfs_commit_super(fs_info);
 		if (ret)
 			goto restore;
 	} else {
-		if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
+		if (BTRFS_FS_ERROR(fs_info)) {
 			btrfs_err(fs_info,
 				"Remounting read-write after error is not allowed");
 			ret = -EINVAL;
@@ -1840,52 +2132,39 @@ static int btrfs_remount(struct super_block *sb, int *flags, char *data)
 			goto restore;
 		}
 
-		ret = btrfs_cleanup_fs_roots(fs_info);
-		if (ret)
-			goto restore;
-
-		/* recover relocation */
-		mutex_lock(&fs_info->cleaner_mutex);
-		ret = btrfs_recover_relocation(root);
-		mutex_unlock(&fs_info->cleaner_mutex);
-		if (ret)
-			goto restore;
-
-		ret = btrfs_resume_balance_async(fs_info);
+		/*
+		 * NOTE: when remounting with a change that does writes, don't
+		 * put it anywhere above this point, as we are not sure to be
+		 * safe to write until we pass the above checks.
+		 */
+		ret = btrfs_start_pre_rw_mount(fs_info);
 		if (ret)
 			goto restore;
 
-		ret = btrfs_resume_dev_replace_async(fs_info);
-		if (ret) {
-			btrfs_warn(fs_info, "failed to resume dev_replace");
-			goto restore;
-		}
-
-		btrfs_qgroup_rescan_resume(fs_info);
-
-		if (!fs_info->uuid_root) {
-			btrfs_info(fs_info, "creating UUID tree");
-			ret = btrfs_create_uuid_tree(fs_info);
-			if (ret) {
-				btrfs_warn(fs_info,
-					   "failed to create the UUID tree %d",
-					   ret);
-				goto restore;
-			}
-		}
-		sb->s_flags &= ~SB_RDONLY;
+		btrfs_clear_sb_rdonly(sb);
 
 		set_bit(BTRFS_FS_OPEN, &fs_info->flags);
 	}
 out:
+	/*
+	 * We need to set SB_I_VERSION here otherwise it'll get cleared by VFS,
+	 * since the absence of the flag means it can be toggled off by remount.
+	 */
+	*flags |= SB_I_VERSION;
+
 	wake_up_process(fs_info->transaction_kthread);
 	btrfs_remount_cleanup(fs_info, old_opts);
+	btrfs_clear_oneshot_options(fs_info);
+	clear_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state);
+
 	return 0;
 
 restore:
 	/* We've hit an error - don't reset SB_RDONLY */
 	if (sb_rdonly(sb))
 		old_flags |= SB_RDONLY;
+	if (!(old_flags & SB_RDONLY))
+		clear_bit(BTRFS_FS_STATE_RO, &fs_info->fs_state);
 	sb->s_flags = old_flags;
 	fs_info->mount_opt = old_opts;
 	fs_info->compress_type = old_compress_type;
@@ -1894,20 +2173,21 @@ restore:
 		old_thread_pool_size, fs_info->thread_pool_size);
 	fs_info->metadata_ratio = old_metadata_ratio;
 	btrfs_remount_cleanup(fs_info, old_opts);
+	clear_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state);
+
 	return ret;
 }
 
 /* Used to sort the devices by max_avail(descending sort) */
-static inline int btrfs_cmp_device_free_bytes(const void *dev_info1,
-				       const void *dev_info2)
+static int btrfs_cmp_device_free_bytes(const void *a, const void *b)
 {
-	if (((struct btrfs_device_info *)dev_info1)->max_avail >
-	    ((struct btrfs_device_info *)dev_info2)->max_avail)
+	const struct btrfs_device_info *dev_info1 = a;
+	const struct btrfs_device_info *dev_info2 = b;
+
+	if (dev_info1->max_avail > dev_info2->max_avail)
 		return -1;
-	else if (((struct btrfs_device_info *)dev_info1)->max_avail <
-		 ((struct btrfs_device_info *)dev_info2)->max_avail)
+	else if (dev_info1->max_avail < dev_info2->max_avail)
 		return 1;
-	else
 	return 0;
 }
 
@@ -1966,12 +2246,8 @@ static inline int btrfs_calc_avail_data_space(struct btrfs_fs_info *fs_info,
 
 	if (type & BTRFS_BLOCK_GROUP_RAID0)
 		num_stripes = nr_devices;
-	else if (type & BTRFS_BLOCK_GROUP_RAID1)
-		num_stripes = 2;
-	else if (type & BTRFS_BLOCK_GROUP_RAID1C3)
-		num_stripes = 3;
-	else if (type & BTRFS_BLOCK_GROUP_RAID1C4)
-		num_stripes = 4;
+	else if (type & BTRFS_BLOCK_GROUP_RAID1_MASK)
+		num_stripes = rattr->ncopies;
 	else if (type & BTRFS_BLOCK_GROUP_RAID10)
 		num_stripes = 4;
 
@@ -1995,17 +2271,13 @@ static inline int btrfs_calc_avail_data_space(struct btrfs_fs_info *fs_info,
 		avail_space = rounddown(avail_space, BTRFS_STRIPE_LEN);
 
 		/*
-		 * In order to avoid overwriting the superblock on the drive,
-		 * btrfs starts at an offset of at least 1MB when doing chunk
-		 * allocation.
-		 *
-		 * This ensures we have at least min_stripe_size free space
-		 * after excluding 1MB.
+		 * Ensure we have at least min_stripe_size on top of the
+		 * reserved space on the device.
 		 */
-		if (avail_space <= SZ_1M + min_stripe_size)
+		if (avail_space <= BTRFS_DEVICE_RANGE_RESERVED + min_stripe_size)
 			continue;
 
-		avail_space -= SZ_1M;
+		avail_space -= BTRFS_DEVICE_RANGE_RESERVED;
 
 		devices_info[i].dev = device;
 		devices_info[i].max_avail = avail_space;
@@ -2062,7 +2334,7 @@ static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
 	u64 total_used = 0;
 	u64 total_free_data = 0;
 	u64 total_free_meta = 0;
-	int bits = dentry->d_sb->s_blocksize_bits;
+	u32 bits = fs_info->sectorsize_bits;
 	__be32 *fsid = (__be32 *)fs_info->fs_devices->fsid;
 	unsigned factor = 1;
 	struct btrfs_block_rsv *block_rsv = &fs_info->global_block_rsv;
@@ -2070,8 +2342,7 @@ static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
 	u64 thresh = 0;
 	int mixed = 0;
 
-	rcu_read_lock();
-	list_for_each_entry_rcu(found, &fs_info->space_info, list) {
+	list_for_each_entry(found, &fs_info->space_info, list) {
 		if (found->flags & BTRFS_BLOCK_GROUP_DATA) {
 			int i;
 
@@ -2100,8 +2371,6 @@ static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
 		total_used += found->disk_used;
 	}
 
-	rcu_read_unlock();
-
 	buf->f_blocks = div_u64(btrfs_super_total_bytes(disk_super), factor);
 	buf->f_blocks >>= bits;
 	buf->f_bfree = buf->f_blocks - (div_u64(total_used, factor) >> bits);
@@ -2170,7 +2439,7 @@ static void btrfs_kill_super(struct super_block *sb)
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(sb);
 	kill_anon_super(sb);
-	free_fs_info(fs_info);
+	btrfs_free_fs_info(fs_info);
 }
 
 static struct file_system_type btrfs_fs_type = {
@@ -2186,7 +2455,7 @@ static struct file_system_type btrfs_root_fs_type = {
 	.name		= "btrfs",
 	.mount		= btrfs_mount_root,
 	.kill_sb	= btrfs_kill_super,
-	.fs_flags	= FS_REQUIRES_DEV | FS_BINARY_MOUNTDATA,
+	.fs_flags	= FS_REQUIRES_DEV | FS_BINARY_MOUNTDATA | FS_ALLOW_IDMAP,
 };
 
 MODULE_ALIAS_FS("btrfs");
@@ -2203,13 +2472,14 @@ static int btrfs_control_open(struct inode *inode, struct file *file)
 }
 
 /*
- * used by btrfsctl to scan devices when no FS is mounted
+ * Used by /dev/btrfs-control for devices ioctls.
  */
 static long btrfs_control_ioctl(struct file *file, unsigned int cmd,
 				unsigned long arg)
 {
 	struct btrfs_ioctl_vol_args *vol;
 	struct btrfs_device *device = NULL;
+	dev_t devt = 0;
 	int ret = -ENOTTY;
 
 	if (!capable(CAP_SYS_ADMIN))
@@ -2229,7 +2499,12 @@ static long btrfs_control_ioctl(struct file *file, unsigned int cmd,
 		mutex_unlock(&uuid_mutex);
 		break;
 	case BTRFS_IOC_FORGET_DEV:
-		ret = btrfs_forget_devices(vol->name);
+		if (vol->name[0] != 0) {
+			ret = lookup_bdev(vol->name, &devt);
+			if (ret)
+				break;
+		}
+		ret = btrfs_forget_devices(devt);
 		break;
 	case BTRFS_IOC_DEVICES_READY:
 		mutex_lock(&uuid_mutex);
@@ -2276,48 +2551,103 @@ static int btrfs_freeze(struct super_block *sb)
 	return btrfs_commit_transaction(trans);
 }
 
+static int check_dev_super(struct btrfs_device *dev)
+{
+	struct btrfs_fs_info *fs_info = dev->fs_info;
+	struct btrfs_super_block *sb;
+	u16 csum_type;
+	int ret = 0;
+
+	/* This should be called with fs still frozen. */
+	ASSERT(test_bit(BTRFS_FS_FROZEN, &fs_info->flags));
+
+	/* Missing dev, no need to check. */
+	if (!dev->bdev)
+		return 0;
+
+	/* Only need to check the primary super block. */
+	sb = btrfs_read_dev_one_super(dev->bdev, 0, true);
+	if (IS_ERR(sb))
+		return PTR_ERR(sb);
+
+	/* Verify the checksum. */
+	csum_type = btrfs_super_csum_type(sb);
+	if (csum_type != btrfs_super_csum_type(fs_info->super_copy)) {
+		btrfs_err(fs_info, "csum type changed, has %u expect %u",
+			  csum_type, btrfs_super_csum_type(fs_info->super_copy));
+		ret = -EUCLEAN;
+		goto out;
+	}
+
+	if (btrfs_check_super_csum(fs_info, sb)) {
+		btrfs_err(fs_info, "csum for on-disk super block no longer matches");
+		ret = -EUCLEAN;
+		goto out;
+	}
+
+	/* Btrfs_validate_super() includes fsid check against super->fsid. */
+	ret = btrfs_validate_super(fs_info, sb, 0);
+	if (ret < 0)
+		goto out;
+
+	if (btrfs_super_generation(sb) != fs_info->last_trans_committed) {
+		btrfs_err(fs_info, "transid mismatch, has %llu expect %llu",
+			btrfs_super_generation(sb),
+			fs_info->last_trans_committed);
+		ret = -EUCLEAN;
+		goto out;
+	}
+out:
+	btrfs_release_disk_super(sb);
+	return ret;
+}
+
 static int btrfs_unfreeze(struct super_block *sb)
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(sb);
+	struct btrfs_device *device;
+	int ret = 0;
 
+	/*
+	 * Make sure the fs is not changed by accident (like hibernation then
+	 * modified by other OS).
+	 * If we found anything wrong, we mark the fs error immediately.
+	 *
+	 * And since the fs is frozen, no one can modify the fs yet, thus
+	 * we don't need to hold device_list_mutex.
+	 */
+	list_for_each_entry(device, &fs_info->fs_devices->devices, dev_list) {
+		ret = check_dev_super(device);
+		if (ret < 0) {
+			btrfs_handle_fs_error(fs_info, ret,
+				"super block on devid %llu got modified unexpectedly",
+				device->devid);
+			break;
+		}
+	}
 	clear_bit(BTRFS_FS_FROZEN, &fs_info->flags);
+
+	/*
+	 * We still return 0, to allow VFS layer to unfreeze the fs even the
+	 * above checks failed. Since the fs is either fine or read-only, we're
+	 * safe to continue, without causing further damage.
+	 */
 	return 0;
 }
 
 static int btrfs_show_devname(struct seq_file *m, struct dentry *root)
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(root->d_sb);
-	struct btrfs_fs_devices *cur_devices;
-	struct btrfs_device *dev, *first_dev = NULL;
-	struct list_head *head;
 
 	/*
-	 * Lightweight locking of the devices. We should not need
-	 * device_list_mutex here as we only read the device data and the list
-	 * is protected by RCU.  Even if a device is deleted during the list
-	 * traversals, we'll get valid data, the freeing callback will wait at
-	 * least until the rcu_read_unlock.
+	 * There should be always a valid pointer in latest_dev, it may be stale
+	 * for a short moment in case it's being deleted but still valid until
+	 * the end of RCU grace period.
 	 */
 	rcu_read_lock();
-	cur_devices = fs_info->fs_devices;
-	while (cur_devices) {
-		head = &cur_devices->devices;
-		list_for_each_entry_rcu(dev, head, dev_list) {
-			if (test_bit(BTRFS_DEV_STATE_MISSING, &dev->dev_state))
-				continue;
-			if (!dev->name)
-				continue;
-			if (!first_dev || dev->devid < first_dev->devid)
-				first_dev = dev;
-		}
-		cur_devices = cur_devices->seed;
-	}
-
-	if (first_dev)
-		seq_escape(m, rcu_str_deref(first_dev->name), " \t\n\\");
-	else
-		WARN_ON(1);
+	seq_escape(m, rcu_str_deref(fs_info->fs_devices->latest_dev->name), " \t\n\\");
 	rcu_read_unlock();
+
 	return 0;
 }
 
@@ -2379,6 +2709,16 @@ static void __init btrfs_print_mod_info(void)
 #ifdef CONFIG_BTRFS_FS_REF_VERIFY
 			", ref-verify=on"
 #endif
+#ifdef CONFIG_BLK_DEV_ZONED
+			", zoned=yes"
+#else
+			", zoned=no"
+#endif
+#ifdef CONFIG_FS_VERITY
+			", fsverity=yes"
+#else
+			", fsverity=no"
+#endif
 			;
 	pr_info("Btrfs loaded, crc32c=%s%s\n", crc32c_impl(), options);
 }
@@ -2399,17 +2739,21 @@ static int __init init_btrfs_fs(void)
 	if (err)
 		goto free_compress;
 
-	err = extent_io_init();
+	err = extent_state_init_cachep();
 	if (err)
 		goto free_cachep;
 
-	err = extent_state_cache_init();
+	err = extent_buffer_init_cachep();
+	if (err)
+		goto free_extent_cachep;
+
+	err = btrfs_bioset_init();
 	if (err)
-		goto free_extent_io;
+		goto free_eb_cachep;
 
 	err = extent_map_init();
 	if (err)
-		goto free_extent_state_cache;
+		goto free_bioset;
 
 	err = ordered_data_init();
 	if (err)
@@ -2431,15 +2775,9 @@ static int __init init_btrfs_fs(void)
 	if (err)
 		goto free_delayed_ref;
 
-	err = btrfs_end_io_wq_init();
-	if (err)
-		goto free_prelim_ref;
-
 	err = btrfs_interface_init();
 	if (err)
-		goto free_end_io_wq;
-
-	btrfs_init_lockdep();
+		goto free_prelim_ref;
 
 	btrfs_print_mod_info();
 
@@ -2455,8 +2793,6 @@ static int __init init_btrfs_fs(void)
 
 unregister_ioctl:
 	btrfs_interface_exit();
-free_end_io_wq:
-	btrfs_end_io_wq_exit();
 free_prelim_ref:
 	btrfs_prelim_ref_exit();
 free_delayed_ref:
@@ -2469,10 +2805,12 @@ free_ordered_data:
 	ordered_data_exit();
 free_extent_map:
 	extent_map_exit();
-free_extent_state_cache:
-	extent_state_cache_exit();
-free_extent_io:
-	extent_io_exit();
+free_bioset:
+	btrfs_bioset_exit();
+free_eb_cachep:
+	extent_buffer_free_cachep();
+free_extent_cachep:
+	extent_state_free_cachep();
 free_cachep:
 	btrfs_destroy_cachep();
 free_compress:
@@ -2491,10 +2829,10 @@ static void __exit exit_btrfs_fs(void)
 	btrfs_prelim_ref_exit();
 	ordered_data_exit();
 	extent_map_exit();
-	extent_state_cache_exit();
-	extent_io_exit();
+	btrfs_bioset_exit();
+	extent_state_free_cachep();
+	extent_buffer_free_cachep();
 	btrfs_interface_exit();
-	btrfs_end_io_wq_exit();
 	unregister_filesystem(&btrfs_fs_type);
 	btrfs_exit_sysfs();
 	btrfs_cleanup_fs_uuids();
diff --git a/fs/btrfs/sysfs.c b/fs/btrfs/sysfs.c
index 3c10e78924d0..699b54b3acaa 100644
--- a/fs/btrfs/sysfs.c
+++ b/fs/btrfs/sysfs.c
@@ -14,11 +14,34 @@
 #include "ctree.h"
 #include "discard.h"
 #include "disk-io.h"
+#include "send.h"
 #include "transaction.h"
 #include "sysfs.h"
 #include "volumes.h"
 #include "space-info.h"
 #include "block-group.h"
+#include "qgroup.h"
+#include "misc.h"
+
+/*
+ * Structure name                       Path
+ * --------------------------------------------------------------------------
+ * btrfs_supported_static_feature_attrs /sys/fs/btrfs/features
+ * btrfs_supported_feature_attrs	/sys/fs/btrfs/features and
+ *					/sys/fs/btrfs/<uuid>/features
+ * btrfs_attrs				/sys/fs/btrfs/<uuid>
+ * devid_attrs				/sys/fs/btrfs/<uuid>/devinfo/<devid>
+ * allocation_attrs			/sys/fs/btrfs/<uuid>/allocation
+ * qgroup_attrs				/sys/fs/btrfs/<uuid>/qgroups/<level>_<qgroupid>
+ * space_info_attrs			/sys/fs/btrfs/<uuid>/allocation/<bg-type>
+ * raid_attrs				/sys/fs/btrfs/<uuid>/allocation/<bg-type>/<bg-profile>
+ * discard_attrs			/sys/fs/btrfs/<uuid>/discard
+ *
+ * When built with BTRFS_CONFIG_DEBUG:
+ *
+ * btrfs_debug_feature_attrs		/sys/fs/btrfs/debug
+ * btrfs_debug_mount_attrs		/sys/fs/btrfs/<uuid>/debug
+ */
 
 struct btrfs_feature_attr {
 	struct kobj_attribute kobj_attr;
@@ -39,6 +62,10 @@ struct raid_kobject {
 	.store	= _store,						\
 }
 
+#define BTRFS_ATTR_W(_prefix, _name, _store)			        \
+	static struct kobj_attribute btrfs_attr_##_prefix##_##_name =	\
+			__INIT_KOBJ_ATTR(_name, 0200, NULL, _store)
+
 #define BTRFS_ATTR_RW(_prefix, _name, _show, _store)			\
 	static struct kobj_attribute btrfs_attr_##_prefix##_##_name =	\
 			__INIT_KOBJ_ATTR(_name, 0644, _show, _store)
@@ -70,6 +97,7 @@ static struct btrfs_feature_attr btrfs_attr_features_##_name = {	     \
 
 static inline struct btrfs_fs_info *to_fs_info(struct kobject *kobj);
 static inline struct btrfs_fs_devices *to_fs_devs(struct kobject *kobj);
+static struct kobject *get_btrfs_kobj(struct kobject *kobj);
 
 static struct btrfs_feature_attr *to_btrfs_feature_attr(struct kobj_attribute *a)
 {
@@ -155,7 +183,7 @@ static ssize_t btrfs_feature_attr_show(struct kobject *kobj,
 	} else
 		val = can_modify_feature(fa);
 
-	return snprintf(buf, PAGE_SIZE, "%d\n", val);
+	return sysfs_emit(buf, "%d\n", val);
 }
 
 static ssize_t btrfs_feature_attr_store(struct kobject *kobj,
@@ -248,27 +276,41 @@ static umode_t btrfs_feature_visible(struct kobject *kobj,
 	return mode;
 }
 
-BTRFS_FEAT_ATTR_INCOMPAT(mixed_backref, MIXED_BACKREF);
 BTRFS_FEAT_ATTR_INCOMPAT(default_subvol, DEFAULT_SUBVOL);
 BTRFS_FEAT_ATTR_INCOMPAT(mixed_groups, MIXED_GROUPS);
 BTRFS_FEAT_ATTR_INCOMPAT(compress_lzo, COMPRESS_LZO);
 BTRFS_FEAT_ATTR_INCOMPAT(compress_zstd, COMPRESS_ZSTD);
-BTRFS_FEAT_ATTR_INCOMPAT(big_metadata, BIG_METADATA);
 BTRFS_FEAT_ATTR_INCOMPAT(extended_iref, EXTENDED_IREF);
 BTRFS_FEAT_ATTR_INCOMPAT(raid56, RAID56);
 BTRFS_FEAT_ATTR_INCOMPAT(skinny_metadata, SKINNY_METADATA);
 BTRFS_FEAT_ATTR_INCOMPAT(no_holes, NO_HOLES);
 BTRFS_FEAT_ATTR_INCOMPAT(metadata_uuid, METADATA_UUID);
 BTRFS_FEAT_ATTR_COMPAT_RO(free_space_tree, FREE_SPACE_TREE);
+BTRFS_FEAT_ATTR_COMPAT_RO(block_group_tree, BLOCK_GROUP_TREE);
 BTRFS_FEAT_ATTR_INCOMPAT(raid1c34, RAID1C34);
+#ifdef CONFIG_BLK_DEV_ZONED
+BTRFS_FEAT_ATTR_INCOMPAT(zoned, ZONED);
+#endif
+#ifdef CONFIG_BTRFS_DEBUG
+/* Remove once support for extent tree v2 is feature complete */
+BTRFS_FEAT_ATTR_INCOMPAT(extent_tree_v2, EXTENT_TREE_V2);
+#endif
+#ifdef CONFIG_FS_VERITY
+BTRFS_FEAT_ATTR_COMPAT_RO(verity, VERITY);
+#endif
 
+/*
+ * Features which depend on feature bits and may differ between each fs.
+ *
+ * /sys/fs/btrfs/features      - all available features implemented by this version
+ * /sys/fs/btrfs/UUID/features - features of the fs which are enabled or
+ *                               can be changed on a mounted filesystem.
+ */
 static struct attribute *btrfs_supported_feature_attrs[] = {
-	BTRFS_FEAT_ATTR_PTR(mixed_backref),
 	BTRFS_FEAT_ATTR_PTR(default_subvol),
 	BTRFS_FEAT_ATTR_PTR(mixed_groups),
 	BTRFS_FEAT_ATTR_PTR(compress_lzo),
 	BTRFS_FEAT_ATTR_PTR(compress_zstd),
-	BTRFS_FEAT_ATTR_PTR(big_metadata),
 	BTRFS_FEAT_ATTR_PTR(extended_iref),
 	BTRFS_FEAT_ATTR_PTR(raid56),
 	BTRFS_FEAT_ATTR_PTR(skinny_metadata),
@@ -276,16 +318,19 @@ static struct attribute *btrfs_supported_feature_attrs[] = {
 	BTRFS_FEAT_ATTR_PTR(metadata_uuid),
 	BTRFS_FEAT_ATTR_PTR(free_space_tree),
 	BTRFS_FEAT_ATTR_PTR(raid1c34),
+	BTRFS_FEAT_ATTR_PTR(block_group_tree),
+#ifdef CONFIG_BLK_DEV_ZONED
+	BTRFS_FEAT_ATTR_PTR(zoned),
+#endif
+#ifdef CONFIG_BTRFS_DEBUG
+	BTRFS_FEAT_ATTR_PTR(extent_tree_v2),
+#endif
+#ifdef CONFIG_FS_VERITY
+	BTRFS_FEAT_ATTR_PTR(verity),
+#endif
 	NULL
 };
 
-/*
- * Features which depend on feature bits and may differ between each fs.
- *
- * /sys/fs/btrfs/features lists all available features of this kernel while
- * /sys/fs/btrfs/UUID/features shows features of the fs which are enabled or
- * can be changed online.
- */
 static const struct attribute_group btrfs_feature_attr_group = {
 	.name = "features",
 	.is_visible = btrfs_feature_visible,
@@ -295,7 +340,7 @@ static const struct attribute_group btrfs_feature_attr_group = {
 static ssize_t rmdir_subvol_show(struct kobject *kobj,
 				 struct kobj_attribute *ka, char *buf)
 {
-	return snprintf(buf, PAGE_SIZE, "0\n");
+	return sysfs_emit(buf, "0\n");
 }
 BTRFS_ATTR(static_feature, rmdir_subvol, rmdir_subvol_show);
 
@@ -310,39 +355,86 @@ static ssize_t supported_checksums_show(struct kobject *kobj,
 		 * This "trick" only works as long as 'enum btrfs_csum_type' has
 		 * no holes in it
 		 */
-		ret += snprintf(buf + ret, PAGE_SIZE - ret, "%s%s",
-				(i == 0 ? "" : " "), btrfs_super_csum_name(i));
+		ret += sysfs_emit_at(buf, ret, "%s%s", (i == 0 ? "" : " "),
+				     btrfs_super_csum_name(i));
 
 	}
 
-	ret += snprintf(buf + ret, PAGE_SIZE - ret, "\n");
+	ret += sysfs_emit_at(buf, ret, "\n");
 	return ret;
 }
 BTRFS_ATTR(static_feature, supported_checksums, supported_checksums_show);
 
-static struct attribute *btrfs_supported_static_feature_attrs[] = {
-	BTRFS_ATTR_PTR(static_feature, rmdir_subvol),
-	BTRFS_ATTR_PTR(static_feature, supported_checksums),
-	NULL
+static ssize_t send_stream_version_show(struct kobject *kobj,
+					struct kobj_attribute *ka, char *buf)
+{
+	return sysfs_emit(buf, "%d\n", BTRFS_SEND_STREAM_VERSION);
+}
+BTRFS_ATTR(static_feature, send_stream_version, send_stream_version_show);
+
+static const char *rescue_opts[] = {
+	"usebackuproot",
+	"nologreplay",
+	"ignorebadroots",
+	"ignoredatacsums",
+	"all",
 };
 
+static ssize_t supported_rescue_options_show(struct kobject *kobj,
+					     struct kobj_attribute *a,
+					     char *buf)
+{
+	ssize_t ret = 0;
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(rescue_opts); i++)
+		ret += sysfs_emit_at(buf, ret, "%s%s", (i ? " " : ""), rescue_opts[i]);
+	ret += sysfs_emit_at(buf, ret, "\n");
+	return ret;
+}
+BTRFS_ATTR(static_feature, supported_rescue_options,
+	   supported_rescue_options_show);
+
+static ssize_t supported_sectorsizes_show(struct kobject *kobj,
+					  struct kobj_attribute *a,
+					  char *buf)
+{
+	ssize_t ret = 0;
+
+	/* An artificial limit to only support 4K and PAGE_SIZE */
+	if (PAGE_SIZE > SZ_4K)
+		ret += sysfs_emit_at(buf, ret, "%u ", SZ_4K);
+	ret += sysfs_emit_at(buf, ret, "%lu\n", PAGE_SIZE);
+
+	return ret;
+}
+BTRFS_ATTR(static_feature, supported_sectorsizes,
+	   supported_sectorsizes_show);
+
 /*
  * Features which only depend on kernel version.
  *
  * These are listed in /sys/fs/btrfs/features along with
- * btrfs_feature_attr_group
+ * btrfs_supported_feature_attrs.
  */
+static struct attribute *btrfs_supported_static_feature_attrs[] = {
+	BTRFS_ATTR_PTR(static_feature, rmdir_subvol),
+	BTRFS_ATTR_PTR(static_feature, supported_checksums),
+	BTRFS_ATTR_PTR(static_feature, send_stream_version),
+	BTRFS_ATTR_PTR(static_feature, supported_rescue_options),
+	BTRFS_ATTR_PTR(static_feature, supported_sectorsizes),
+	NULL
+};
+
 static const struct attribute_group btrfs_static_feature_attr_group = {
 	.name = "features",
 	.attrs = btrfs_supported_static_feature_attrs,
 };
 
-#ifdef CONFIG_BTRFS_DEBUG
-
 /*
  * Discard statistics and tunables
  */
-#define discard_to_fs_info(_kobj)	to_fs_info((_kobj)->parent->parent)
+#define discard_to_fs_info(_kobj)	to_fs_info(get_btrfs_kobj(_kobj))
 
 static ssize_t btrfs_discardable_bytes_show(struct kobject *kobj,
 					    struct kobj_attribute *a,
@@ -350,7 +442,7 @@ static ssize_t btrfs_discardable_bytes_show(struct kobject *kobj,
 {
 	struct btrfs_fs_info *fs_info = discard_to_fs_info(kobj);
 
-	return snprintf(buf, PAGE_SIZE, "%lld\n",
+	return sysfs_emit(buf, "%lld\n",
 			atomic64_read(&fs_info->discard_ctl.discardable_bytes));
 }
 BTRFS_ATTR(discard, discardable_bytes, btrfs_discardable_bytes_show);
@@ -361,7 +453,7 @@ static ssize_t btrfs_discardable_extents_show(struct kobject *kobj,
 {
 	struct btrfs_fs_info *fs_info = discard_to_fs_info(kobj);
 
-	return snprintf(buf, PAGE_SIZE, "%d\n",
+	return sysfs_emit(buf, "%d\n",
 			atomic_read(&fs_info->discard_ctl.discardable_extents));
 }
 BTRFS_ATTR(discard, discardable_extents, btrfs_discardable_extents_show);
@@ -372,8 +464,8 @@ static ssize_t btrfs_discard_bitmap_bytes_show(struct kobject *kobj,
 {
 	struct btrfs_fs_info *fs_info = discard_to_fs_info(kobj);
 
-	return snprintf(buf, PAGE_SIZE, "%lld\n",
-			fs_info->discard_ctl.discard_bitmap_bytes);
+	return sysfs_emit(buf, "%llu\n",
+			  fs_info->discard_ctl.discard_bitmap_bytes);
 }
 BTRFS_ATTR(discard, discard_bitmap_bytes, btrfs_discard_bitmap_bytes_show);
 
@@ -383,7 +475,7 @@ static ssize_t btrfs_discard_bytes_saved_show(struct kobject *kobj,
 {
 	struct btrfs_fs_info *fs_info = discard_to_fs_info(kobj);
 
-	return snprintf(buf, PAGE_SIZE, "%lld\n",
+	return sysfs_emit(buf, "%lld\n",
 		atomic64_read(&fs_info->discard_ctl.discard_bytes_saved));
 }
 BTRFS_ATTR(discard, discard_bytes_saved, btrfs_discard_bytes_saved_show);
@@ -394,8 +486,8 @@ static ssize_t btrfs_discard_extent_bytes_show(struct kobject *kobj,
 {
 	struct btrfs_fs_info *fs_info = discard_to_fs_info(kobj);
 
-	return snprintf(buf, PAGE_SIZE, "%lld\n",
-			fs_info->discard_ctl.discard_extent_bytes);
+	return sysfs_emit(buf, "%llu\n",
+			  fs_info->discard_ctl.discard_extent_bytes);
 }
 BTRFS_ATTR(discard, discard_extent_bytes, btrfs_discard_extent_bytes_show);
 
@@ -405,8 +497,8 @@ static ssize_t btrfs_discard_iops_limit_show(struct kobject *kobj,
 {
 	struct btrfs_fs_info *fs_info = discard_to_fs_info(kobj);
 
-	return snprintf(buf, PAGE_SIZE, "%u\n",
-			READ_ONCE(fs_info->discard_ctl.iops_limit));
+	return sysfs_emit(buf, "%u\n",
+			  READ_ONCE(fs_info->discard_ctl.iops_limit));
 }
 
 static ssize_t btrfs_discard_iops_limit_store(struct kobject *kobj,
@@ -423,7 +515,8 @@ static ssize_t btrfs_discard_iops_limit_store(struct kobject *kobj,
 		return -EINVAL;
 
 	WRITE_ONCE(discard_ctl->iops_limit, iops_limit);
-
+	btrfs_discard_calc_delay(discard_ctl);
+	btrfs_discard_schedule_work(discard_ctl, true);
 	return len;
 }
 BTRFS_ATTR_RW(discard, iops_limit, btrfs_discard_iops_limit_show,
@@ -435,8 +528,8 @@ static ssize_t btrfs_discard_kbps_limit_show(struct kobject *kobj,
 {
 	struct btrfs_fs_info *fs_info = discard_to_fs_info(kobj);
 
-	return snprintf(buf, PAGE_SIZE, "%u\n",
-			READ_ONCE(fs_info->discard_ctl.kbps_limit));
+	return sysfs_emit(buf, "%u\n",
+			  READ_ONCE(fs_info->discard_ctl.kbps_limit));
 }
 
 static ssize_t btrfs_discard_kbps_limit_store(struct kobject *kobj,
@@ -453,7 +546,7 @@ static ssize_t btrfs_discard_kbps_limit_store(struct kobject *kobj,
 		return -EINVAL;
 
 	WRITE_ONCE(discard_ctl->kbps_limit, kbps_limit);
-
+	btrfs_discard_schedule_work(discard_ctl, true);
 	return len;
 }
 BTRFS_ATTR_RW(discard, kbps_limit, btrfs_discard_kbps_limit_show,
@@ -465,8 +558,8 @@ static ssize_t btrfs_discard_max_discard_size_show(struct kobject *kobj,
 {
 	struct btrfs_fs_info *fs_info = discard_to_fs_info(kobj);
 
-	return snprintf(buf, PAGE_SIZE, "%llu\n",
-			READ_ONCE(fs_info->discard_ctl.max_discard_size));
+	return sysfs_emit(buf, "%llu\n",
+			  READ_ONCE(fs_info->discard_ctl.max_discard_size));
 }
 
 static ssize_t btrfs_discard_max_discard_size_store(struct kobject *kobj,
@@ -489,7 +582,12 @@ static ssize_t btrfs_discard_max_discard_size_store(struct kobject *kobj,
 BTRFS_ATTR_RW(discard, max_discard_size, btrfs_discard_max_discard_size_show,
 	      btrfs_discard_max_discard_size_store);
 
-static const struct attribute *discard_debug_attrs[] = {
+/*
+ * Per-filesystem stats for discard (when mounted with discard=async).
+ *
+ * Path: /sys/fs/btrfs/<uuid>/discard/
+ */
+static const struct attribute *discard_attrs[] = {
 	BTRFS_ATTR_PTR(discard, discardable_bytes),
 	BTRFS_ATTR_PTR(discard, discardable_extents),
 	BTRFS_ATTR_PTR(discard, discard_bitmap_bytes),
@@ -501,16 +599,22 @@ static const struct attribute *discard_debug_attrs[] = {
 	NULL,
 };
 
+#ifdef CONFIG_BTRFS_DEBUG
+
 /*
- * Runtime debugging exported via sysfs
+ * Per-filesystem runtime debugging exported via sysfs.
  *
- * /sys/fs/btrfs/debug - applies to module or all filesystems
- * /sys/fs/btrfs/UUID  - applies only to the given filesystem
+ * Path: /sys/fs/btrfs/UUID/debug/
  */
 static const struct attribute *btrfs_debug_mount_attrs[] = {
 	NULL,
 };
 
+/*
+ * Runtime debugging exported via sysfs, applies to all mounted filesystems.
+ *
+ * Path: /sys/fs/btrfs/debug
+ */
 static struct attribute *btrfs_debug_feature_attrs[] = {
 	NULL
 };
@@ -530,7 +634,7 @@ static ssize_t btrfs_show_u64(u64 *value_ptr, spinlock_t *lock, char *buf)
 	val = *value_ptr;
 	if (lock)
 		spin_unlock(lock);
-	return snprintf(buf, PAGE_SIZE, "%llu\n", val);
+	return sysfs_emit(buf, "%llu\n", val);
 }
 
 static ssize_t global_rsv_size_show(struct kobject *kobj,
@@ -576,9 +680,14 @@ static ssize_t raid_bytes_show(struct kobject *kobj,
 			val += block_group->used;
 	}
 	up_read(&sinfo->groups_sem);
-	return snprintf(buf, PAGE_SIZE, "%llu\n", val);
+	return sysfs_emit(buf, "%llu\n", val);
 }
 
+/*
+ * Allocation information about block group profiles.
+ *
+ * Path: /sys/fs/btrfs/<uuid>/allocation/<bg-type>/<bg-profile>/
+ */
 static struct attribute *raid_attrs[] = {
 	BTRFS_ATTR_PTR(raid, total_bytes),
 	BTRFS_ATTR_PTR(raid, used_bytes),
@@ -607,15 +716,112 @@ static ssize_t btrfs_space_info_show_##field(struct kobject *kobj,	\
 }									\
 BTRFS_ATTR(space_info, field, btrfs_space_info_show_##field)
 
-static ssize_t btrfs_space_info_show_total_bytes_pinned(struct kobject *kobj,
-						       struct kobj_attribute *a,
-						       char *buf)
+static ssize_t btrfs_chunk_size_show(struct kobject *kobj,
+				     struct kobj_attribute *a, char *buf)
 {
 	struct btrfs_space_info *sinfo = to_space_info(kobj);
-	s64 val = percpu_counter_sum(&sinfo->total_bytes_pinned);
-	return snprintf(buf, PAGE_SIZE, "%lld\n", val);
+
+	return sysfs_emit(buf, "%llu\n", READ_ONCE(sinfo->chunk_size));
 }
 
+/*
+ * Store new chunk size in space info. Can be called on a read-only filesystem.
+ *
+ * If the new chunk size value is larger than 10% of free space it is reduced
+ * to match that limit. Alignment must be to 256M and the system chunk size
+ * cannot be set.
+ */
+static ssize_t btrfs_chunk_size_store(struct kobject *kobj,
+				      struct kobj_attribute *a,
+				      const char *buf, size_t len)
+{
+	struct btrfs_space_info *space_info = to_space_info(kobj);
+	struct btrfs_fs_info *fs_info = to_fs_info(get_btrfs_kobj(kobj));
+	char *retptr;
+	u64 val;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -EPERM;
+
+	if (!fs_info->fs_devices)
+		return -EINVAL;
+
+	if (btrfs_is_zoned(fs_info))
+		return -EINVAL;
+
+	/* System block type must not be changed. */
+	if (space_info->flags & BTRFS_BLOCK_GROUP_SYSTEM)
+		return -EPERM;
+
+	val = memparse(buf, &retptr);
+	/* There could be trailing '\n', also catch any typos after the value */
+	retptr = skip_spaces(retptr);
+	if (*retptr != 0 || val == 0)
+		return -EINVAL;
+
+	val = min(val, BTRFS_MAX_DATA_CHUNK_SIZE);
+
+	/* Limit stripe size to 10% of available space. */
+	val = min(div_factor(fs_info->fs_devices->total_rw_bytes, 1), val);
+
+	/* Must be multiple of 256M. */
+	val &= ~((u64)SZ_256M - 1);
+
+	/* Must be at least 256M. */
+	if (val < SZ_256M)
+		return -EINVAL;
+
+	btrfs_update_space_info_chunk_size(space_info, val);
+
+	return len;
+}
+
+#ifdef CONFIG_BTRFS_DEBUG
+/*
+ * Request chunk allocation with current chunk size.
+ */
+static ssize_t btrfs_force_chunk_alloc_store(struct kobject *kobj,
+					     struct kobj_attribute *a,
+					     const char *buf, size_t len)
+{
+	struct btrfs_space_info *space_info = to_space_info(kobj);
+	struct btrfs_fs_info *fs_info = to_fs_info(get_btrfs_kobj(kobj));
+	struct btrfs_trans_handle *trans;
+	bool val;
+	int ret;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -EPERM;
+
+	if (sb_rdonly(fs_info->sb))
+		return -EROFS;
+
+	ret = kstrtobool(buf, &val);
+	if (ret)
+		return ret;
+
+	if (!val)
+		return -EINVAL;
+
+	/*
+	 * This is unsafe to be called from sysfs context and may cause
+	 * unexpected problems.
+	 */
+	trans = btrfs_start_transaction(fs_info->tree_root, 0);
+	if (IS_ERR(trans))
+		return PTR_ERR(trans);
+	ret = btrfs_force_chunk_alloc(trans, space_info->flags);
+	btrfs_end_transaction(trans);
+
+	if (ret == 1)
+		return len;
+
+	return -ENOSPC;
+}
+BTRFS_ATTR_W(space_info, force_chunk_alloc, btrfs_force_chunk_alloc_store);
+
+#endif
+
 SPACE_INFO_ATTR(flags);
 SPACE_INFO_ATTR(total_bytes);
 SPACE_INFO_ATTR(bytes_used);
@@ -623,11 +829,49 @@ SPACE_INFO_ATTR(bytes_pinned);
 SPACE_INFO_ATTR(bytes_reserved);
 SPACE_INFO_ATTR(bytes_may_use);
 SPACE_INFO_ATTR(bytes_readonly);
+SPACE_INFO_ATTR(bytes_zone_unusable);
 SPACE_INFO_ATTR(disk_used);
 SPACE_INFO_ATTR(disk_total);
-BTRFS_ATTR(space_info, total_bytes_pinned,
-	   btrfs_space_info_show_total_bytes_pinned);
+BTRFS_ATTR_RW(space_info, chunk_size, btrfs_chunk_size_show, btrfs_chunk_size_store);
+
+static ssize_t btrfs_sinfo_bg_reclaim_threshold_show(struct kobject *kobj,
+						     struct kobj_attribute *a,
+						     char *buf)
+{
+	struct btrfs_space_info *space_info = to_space_info(kobj);
+
+	return sysfs_emit(buf, "%d\n", READ_ONCE(space_info->bg_reclaim_threshold));
+}
+
+static ssize_t btrfs_sinfo_bg_reclaim_threshold_store(struct kobject *kobj,
+						      struct kobj_attribute *a,
+						      const char *buf, size_t len)
+{
+	struct btrfs_space_info *space_info = to_space_info(kobj);
+	int thresh;
+	int ret;
 
+	ret = kstrtoint(buf, 10, &thresh);
+	if (ret)
+		return ret;
+
+	if (thresh < 0 || thresh > 100)
+		return -EINVAL;
+
+	WRITE_ONCE(space_info->bg_reclaim_threshold, thresh);
+
+	return len;
+}
+
+BTRFS_ATTR_RW(space_info, bg_reclaim_threshold,
+	      btrfs_sinfo_bg_reclaim_threshold_show,
+	      btrfs_sinfo_bg_reclaim_threshold_store);
+
+/*
+ * Allocation information about block group types.
+ *
+ * Path: /sys/fs/btrfs/<uuid>/allocation/<bg-type>/
+ */
 static struct attribute *space_info_attrs[] = {
 	BTRFS_ATTR_PTR(space_info, flags),
 	BTRFS_ATTR_PTR(space_info, total_bytes),
@@ -636,9 +880,14 @@ static struct attribute *space_info_attrs[] = {
 	BTRFS_ATTR_PTR(space_info, bytes_reserved),
 	BTRFS_ATTR_PTR(space_info, bytes_may_use),
 	BTRFS_ATTR_PTR(space_info, bytes_readonly),
+	BTRFS_ATTR_PTR(space_info, bytes_zone_unusable),
 	BTRFS_ATTR_PTR(space_info, disk_used),
 	BTRFS_ATTR_PTR(space_info, disk_total),
-	BTRFS_ATTR_PTR(space_info, total_bytes_pinned),
+	BTRFS_ATTR_PTR(space_info, bg_reclaim_threshold),
+	BTRFS_ATTR_PTR(space_info, chunk_size),
+#ifdef CONFIG_BTRFS_DEBUG
+	BTRFS_ATTR_PTR(space_info, force_chunk_alloc),
+#endif
 	NULL,
 };
 ATTRIBUTE_GROUPS(space_info);
@@ -646,7 +895,6 @@ ATTRIBUTE_GROUPS(space_info);
 static void space_info_release(struct kobject *kobj)
 {
 	struct btrfs_space_info *sinfo = to_space_info(kobj);
-	percpu_counter_destroy(&sinfo->total_bytes_pinned);
 	kfree(sinfo);
 }
 
@@ -656,6 +904,11 @@ static struct kobj_type space_info_ktype = {
 	.default_groups = space_info_groups,
 };
 
+/*
+ * Allocation information about block groups.
+ *
+ * Path: /sys/fs/btrfs/<uuid>/allocation/
+ */
 static const struct attribute *allocation_attrs[] = {
 	BTRFS_ATTR_PTR(allocation, global_rsv_reserved),
 	BTRFS_ATTR_PTR(allocation, global_rsv_size),
@@ -670,7 +923,7 @@ static ssize_t btrfs_label_show(struct kobject *kobj,
 	ssize_t ret;
 
 	spin_lock(&fs_info->super_lock);
-	ret = snprintf(buf, PAGE_SIZE, label[0] ? "%s\n" : "%s", label);
+	ret = sysfs_emit(buf, label[0] ? "%s\n" : "%s", label);
 	spin_unlock(&fs_info->super_lock);
 
 	return ret;
@@ -718,7 +971,7 @@ static ssize_t btrfs_nodesize_show(struct kobject *kobj,
 {
 	struct btrfs_fs_info *fs_info = to_fs_info(kobj);
 
-	return snprintf(buf, PAGE_SIZE, "%u\n", fs_info->super_copy->nodesize);
+	return sysfs_emit(buf, "%u\n", fs_info->super_copy->nodesize);
 }
 
 BTRFS_ATTR(, nodesize, btrfs_nodesize_show);
@@ -728,19 +981,59 @@ static ssize_t btrfs_sectorsize_show(struct kobject *kobj,
 {
 	struct btrfs_fs_info *fs_info = to_fs_info(kobj);
 
-	return snprintf(buf, PAGE_SIZE, "%u\n",
-			fs_info->super_copy->sectorsize);
+	return sysfs_emit(buf, "%u\n", fs_info->super_copy->sectorsize);
 }
 
 BTRFS_ATTR(, sectorsize, btrfs_sectorsize_show);
 
+static ssize_t btrfs_commit_stats_show(struct kobject *kobj,
+				       struct kobj_attribute *a, char *buf)
+{
+	struct btrfs_fs_info *fs_info = to_fs_info(kobj);
+
+	return sysfs_emit(buf,
+		"commits %llu\n"
+		"last_commit_ms %llu\n"
+		"max_commit_ms %llu\n"
+		"total_commit_ms %llu\n",
+		fs_info->commit_stats.commit_count,
+		div_u64(fs_info->commit_stats.last_commit_dur, NSEC_PER_MSEC),
+		div_u64(fs_info->commit_stats.max_commit_dur, NSEC_PER_MSEC),
+		div_u64(fs_info->commit_stats.total_commit_dur, NSEC_PER_MSEC));
+}
+
+static ssize_t btrfs_commit_stats_store(struct kobject *kobj,
+					struct kobj_attribute *a,
+					const char *buf, size_t len)
+{
+	struct btrfs_fs_info *fs_info = to_fs_info(kobj);
+	unsigned long val;
+	int ret;
+
+	if (!fs_info)
+		return -EPERM;
+
+	if (!capable(CAP_SYS_RESOURCE))
+		return -EPERM;
+
+	ret = kstrtoul(buf, 10, &val);
+	if (ret)
+		return ret;
+	if (val)
+		return -EINVAL;
+
+	WRITE_ONCE(fs_info->commit_stats.max_commit_dur, 0);
+
+	return len;
+}
+BTRFS_ATTR_RW(, commit_stats, btrfs_commit_stats_show, btrfs_commit_stats_store);
+
 static ssize_t btrfs_clone_alignment_show(struct kobject *kobj,
 				struct kobj_attribute *a, char *buf)
 {
 	struct btrfs_fs_info *fs_info = to_fs_info(kobj);
 
-	return snprintf(buf, PAGE_SIZE, "%u\n",
-			fs_info->super_copy->sectorsize);
+	return sysfs_emit(buf, "%u\n", fs_info->super_copy->sectorsize);
 }
 
 BTRFS_ATTR(, clone_alignment, btrfs_clone_alignment_show);
@@ -752,7 +1045,7 @@ static ssize_t quota_override_show(struct kobject *kobj,
 	int quota_override;
 
 	quota_override = test_bit(BTRFS_FS_QUOTA_OVERRIDE, &fs_info->flags);
-	return snprintf(buf, PAGE_SIZE, "%d\n", quota_override);
+	return sysfs_emit(buf, "%d\n", quota_override);
 }
 
 static ssize_t quota_override_store(struct kobject *kobj,
@@ -790,8 +1083,7 @@ static ssize_t btrfs_metadata_uuid_show(struct kobject *kobj,
 {
 	struct btrfs_fs_info *fs_info = to_fs_info(kobj);
 
-	return snprintf(buf, PAGE_SIZE, "%pU\n",
-			fs_info->fs_devices->metadata_uuid);
+	return sysfs_emit(buf, "%pU\n", fs_info->fs_devices->metadata_uuid);
 }
 
 BTRFS_ATTR(, metadata_uuid, btrfs_metadata_uuid_show);
@@ -802,13 +1094,145 @@ static ssize_t btrfs_checksum_show(struct kobject *kobj,
 	struct btrfs_fs_info *fs_info = to_fs_info(kobj);
 	u16 csum_type = btrfs_super_csum_type(fs_info->super_copy);
 
-	return snprintf(buf, PAGE_SIZE, "%s (%s)\n",
-			btrfs_super_csum_name(csum_type),
-			crypto_shash_driver_name(fs_info->csum_shash));
+	return sysfs_emit(buf, "%s (%s)\n",
+			  btrfs_super_csum_name(csum_type),
+			  crypto_shash_driver_name(fs_info->csum_shash));
 }
 
 BTRFS_ATTR(, checksum, btrfs_checksum_show);
 
+static ssize_t btrfs_exclusive_operation_show(struct kobject *kobj,
+		struct kobj_attribute *a, char *buf)
+{
+	struct btrfs_fs_info *fs_info = to_fs_info(kobj);
+	const char *str;
+
+	switch (READ_ONCE(fs_info->exclusive_operation)) {
+		case  BTRFS_EXCLOP_NONE:
+			str = "none\n";
+			break;
+		case BTRFS_EXCLOP_BALANCE:
+			str = "balance\n";
+			break;
+		case BTRFS_EXCLOP_BALANCE_PAUSED:
+			str = "balance paused\n";
+			break;
+		case BTRFS_EXCLOP_DEV_ADD:
+			str = "device add\n";
+			break;
+		case BTRFS_EXCLOP_DEV_REMOVE:
+			str = "device remove\n";
+			break;
+		case BTRFS_EXCLOP_DEV_REPLACE:
+			str = "device replace\n";
+			break;
+		case BTRFS_EXCLOP_RESIZE:
+			str = "resize\n";
+			break;
+		case BTRFS_EXCLOP_SWAP_ACTIVATE:
+			str = "swap activate\n";
+			break;
+		default:
+			str = "UNKNOWN\n";
+			break;
+	}
+	return sysfs_emit(buf, "%s", str);
+}
+BTRFS_ATTR(, exclusive_operation, btrfs_exclusive_operation_show);
+
+static ssize_t btrfs_generation_show(struct kobject *kobj,
+				     struct kobj_attribute *a, char *buf)
+{
+	struct btrfs_fs_info *fs_info = to_fs_info(kobj);
+
+	return sysfs_emit(buf, "%llu\n", fs_info->generation);
+}
+BTRFS_ATTR(, generation, btrfs_generation_show);
+
+static const char * const btrfs_read_policy_name[] = { "pid" };
+
+static ssize_t btrfs_read_policy_show(struct kobject *kobj,
+				      struct kobj_attribute *a, char *buf)
+{
+	struct btrfs_fs_devices *fs_devices = to_fs_devs(kobj);
+	ssize_t ret = 0;
+	int i;
+
+	for (i = 0; i < BTRFS_NR_READ_POLICY; i++) {
+		if (fs_devices->read_policy == i)
+			ret += scnprintf(buf + ret, PAGE_SIZE - ret, "%s[%s]",
+					 (ret == 0 ? "" : " "),
+					 btrfs_read_policy_name[i]);
+		else
+			ret += scnprintf(buf + ret, PAGE_SIZE - ret, "%s%s",
+					 (ret == 0 ? "" : " "),
+					 btrfs_read_policy_name[i]);
+	}
+
+	ret += scnprintf(buf + ret, PAGE_SIZE - ret, "\n");
+
+	return ret;
+}
+
+static ssize_t btrfs_read_policy_store(struct kobject *kobj,
+				       struct kobj_attribute *a,
+				       const char *buf, size_t len)
+{
+	struct btrfs_fs_devices *fs_devices = to_fs_devs(kobj);
+	int i;
+
+	for (i = 0; i < BTRFS_NR_READ_POLICY; i++) {
+		if (sysfs_streq(buf, btrfs_read_policy_name[i])) {
+			if (i != fs_devices->read_policy) {
+				fs_devices->read_policy = i;
+				btrfs_info(fs_devices->fs_info,
+					   "read policy set to '%s'",
+					   btrfs_read_policy_name[i]);
+			}
+			return len;
+		}
+	}
+
+	return -EINVAL;
+}
+BTRFS_ATTR_RW(, read_policy, btrfs_read_policy_show, btrfs_read_policy_store);
+
+static ssize_t btrfs_bg_reclaim_threshold_show(struct kobject *kobj,
+					       struct kobj_attribute *a,
+					       char *buf)
+{
+	struct btrfs_fs_info *fs_info = to_fs_info(kobj);
+
+	return sysfs_emit(buf, "%d\n", READ_ONCE(fs_info->bg_reclaim_threshold));
+}
+
+static ssize_t btrfs_bg_reclaim_threshold_store(struct kobject *kobj,
+						struct kobj_attribute *a,
+						const char *buf, size_t len)
+{
+	struct btrfs_fs_info *fs_info = to_fs_info(kobj);
+	int thresh;
+	int ret;
+
+	ret = kstrtoint(buf, 10, &thresh);
+	if (ret)
+		return ret;
+
+	if (thresh != 0 && (thresh <= 50 || thresh > 100))
+		return -EINVAL;
+
+	WRITE_ONCE(fs_info->bg_reclaim_threshold, thresh);
+
+	return len;
+}
+BTRFS_ATTR_RW(, bg_reclaim_threshold, btrfs_bg_reclaim_threshold_show,
+	      btrfs_bg_reclaim_threshold_store);
+
+/*
+ * Per-filesystem information and stats.
+ *
+ * Path: /sys/fs/btrfs/<uuid>/
+ */
 static const struct attribute *btrfs_attrs[] = {
 	BTRFS_ATTR_PTR(, label),
 	BTRFS_ATTR_PTR(, nodesize),
@@ -817,6 +1241,11 @@ static const struct attribute *btrfs_attrs[] = {
 	BTRFS_ATTR_PTR(, quota_override),
 	BTRFS_ATTR_PTR(, metadata_uuid),
 	BTRFS_ATTR_PTR(, checksum),
+	BTRFS_ATTR_PTR(, exclusive_operation),
+	BTRFS_ATTR_PTR(, generation),
+	BTRFS_ATTR_PTR(, read_policy),
+	BTRFS_ATTR_PTR(, bg_reclaim_threshold),
+	BTRFS_ATTR_PTR(, commit_stats),
 	NULL,
 };
 
@@ -847,11 +1276,26 @@ static inline struct btrfs_fs_info *to_fs_info(struct kobject *kobj)
 	return to_fs_devs(kobj)->fs_info;
 }
 
+static struct kobject *get_btrfs_kobj(struct kobject *kobj)
+{
+	while (kobj) {
+		if (kobj->ktype == &btrfs_ktype)
+			return kobj;
+		kobj = kobj->parent;
+	}
+	return NULL;
+}
+
 #define NUM_FEATURE_BITS 64
 #define BTRFS_FEATURE_NAME_MAX 13
 static char btrfs_unknown_feature_names[FEAT_MAX][NUM_FEATURE_BITS][BTRFS_FEATURE_NAME_MAX];
 static struct btrfs_feature_attr btrfs_feature_attrs[FEAT_MAX][NUM_FEATURE_BITS];
 
+static_assert(ARRAY_SIZE(btrfs_unknown_feature_names) ==
+	      ARRAY_SIZE(btrfs_feature_attrs));
+static_assert(ARRAY_SIZE(btrfs_unknown_feature_names[0]) ==
+	      ARRAY_SIZE(btrfs_feature_attrs[0]));
+
 static const u64 supported_feature_masks[FEAT_MAX] = {
 	[FEAT_COMPAT]    = BTRFS_FEATURE_COMPAT_SUPP,
 	[FEAT_COMPAT_RO] = BTRFS_FEATURE_COMPAT_RO_SUPP,
@@ -935,22 +1379,37 @@ void btrfs_sysfs_remove_fsid(struct btrfs_fs_devices *fs_devs)
 	}
 }
 
+static void btrfs_sysfs_remove_fs_devices(struct btrfs_fs_devices *fs_devices)
+{
+	struct btrfs_device *device;
+	struct btrfs_fs_devices *seed;
+
+	list_for_each_entry(device, &fs_devices->devices, dev_list)
+		btrfs_sysfs_remove_device(device);
+
+	list_for_each_entry(seed, &fs_devices->seed_list, seed_list) {
+		list_for_each_entry(device, &seed->devices, dev_list)
+			btrfs_sysfs_remove_device(device);
+	}
+}
+
 void btrfs_sysfs_remove_mounted(struct btrfs_fs_info *fs_info)
 {
-	btrfs_reset_fs_info_ptr(fs_info);
+	struct kobject *fsid_kobj = &fs_info->fs_devices->fsid_kobj;
+
+	sysfs_remove_link(fsid_kobj, "bdi");
 
 	if (fs_info->space_info_kobj) {
 		sysfs_remove_files(fs_info->space_info_kobj, allocation_attrs);
 		kobject_del(fs_info->space_info_kobj);
 		kobject_put(fs_info->space_info_kobj);
 	}
-#ifdef CONFIG_BTRFS_DEBUG
-	if (fs_info->discard_debug_kobj) {
-		sysfs_remove_files(fs_info->discard_debug_kobj,
-				   discard_debug_attrs);
-		kobject_del(fs_info->discard_debug_kobj);
-		kobject_put(fs_info->discard_debug_kobj);
+	if (fs_info->discard_kobj) {
+		sysfs_remove_files(fs_info->discard_kobj, discard_attrs);
+		kobject_del(fs_info->discard_kobj);
+		kobject_put(fs_info->discard_kobj);
 	}
+#ifdef CONFIG_BTRFS_DEBUG
 	if (fs_info->debug_kobj) {
 		sysfs_remove_files(fs_info->debug_kobj, btrfs_debug_mount_attrs);
 		kobject_del(fs_info->debug_kobj);
@@ -958,9 +1417,9 @@ void btrfs_sysfs_remove_mounted(struct btrfs_fs_info *fs_info)
 	}
 #endif
 	addrm_unknown_feature_attrs(fs_info, false);
-	sysfs_remove_group(&fs_info->fs_devices->fsid_kobj, &btrfs_feature_attr_group);
-	sysfs_remove_files(&fs_info->fs_devices->fsid_kobj, btrfs_attrs);
-	btrfs_sysfs_rm_device_link(fs_info->fs_devices, NULL);
+	sysfs_remove_group(fsid_kobj, &btrfs_feature_attr_group);
+	sysfs_remove_files(fsid_kobj, btrfs_attrs);
+	btrfs_sysfs_remove_fs_devices(fs_info->fs_devices);
 }
 
 static const char * const btrfs_feature_set_names[FEAT_MAX] = {
@@ -969,7 +1428,7 @@ static const char * const btrfs_feature_set_names[FEAT_MAX] = {
 	[FEAT_INCOMPAT]	 = "incompat",
 };
 
-const char * const btrfs_feature_set_name(enum btrfs_feature_set set)
+const char *btrfs_feature_set_name(enum btrfs_feature_set set)
 {
 	return btrfs_feature_set_names[set];
 }
@@ -992,7 +1451,7 @@ char *btrfs_printable_features(enum btrfs_feature_set set, u64 flags)
 			continue;
 
 		name = btrfs_feature_attrs[set][i].kobj_attr.attr.name;
-		len += snprintf(str + len, bufsize - len, "%s%s",
+		len += scnprintf(str + len, bufsize - len, "%s%s",
 				len ? "," : "", name);
 	}
 
@@ -1004,11 +1463,6 @@ static void init_feature_attrs(void)
 	struct btrfs_feature_attr *fa;
 	int set, i;
 
-	BUILD_BUG_ON(ARRAY_SIZE(btrfs_unknown_feature_names) !=
-		     ARRAY_SIZE(btrfs_feature_attrs));
-	BUILD_BUG_ON(ARRAY_SIZE(btrfs_unknown_feature_names[0]) !=
-		     ARRAY_SIZE(btrfs_feature_attrs[0]));
-
 	memset(btrfs_feature_attrs, 0, sizeof(btrfs_feature_attrs));
 	memset(btrfs_unknown_feature_names, 0,
 	       sizeof(btrfs_unknown_feature_names));
@@ -1075,17 +1529,38 @@ void btrfs_sysfs_add_block_group_type(struct btrfs_block_group *cache)
 
 	rkobj->flags = cache->flags;
 	kobject_init(&rkobj->kobj, &btrfs_raid_ktype);
+
+	/*
+	 * We call this either on mount, or if we've created a block group for a
+	 * new index type while running (i.e. when restriping).  The running
+	 * case is tricky because we could race with other threads, so we need
+	 * to have this check to make sure we didn't already init the kobject.
+	 *
+	 * We don't have to protect on the free side because it only happens on
+	 * unmount.
+	 */
+	spin_lock(&space_info->lock);
+	if (space_info->block_group_kobjs[index]) {
+		spin_unlock(&space_info->lock);
+		kobject_put(&rkobj->kobj);
+		return;
+	} else {
+		space_info->block_group_kobjs[index] = &rkobj->kobj;
+	}
+	spin_unlock(&space_info->lock);
+
 	ret = kobject_add(&rkobj->kobj, &space_info->kobj, "%s",
 			  btrfs_bg_type_to_raid_name(rkobj->flags));
 	memalloc_nofs_restore(nofs_flag);
 	if (ret) {
+		spin_lock(&space_info->lock);
+		space_info->block_group_kobjs[index] = NULL;
+		spin_unlock(&space_info->lock);
 		kobject_put(&rkobj->kobj);
 		btrfs_warn(fs_info,
 			"failed to add kobject for block cache, ignoring");
 		return;
 	}
-
-	space_info->block_group_kobjs[index] = &rkobj->kobj;
 }
 
 /*
@@ -1124,7 +1599,7 @@ static const char *alloc_name(u64 flags)
 	default:
 		WARN_ON(1);
 		return "invalid-combination";
-	};
+	}
 }
 
 /*
@@ -1147,48 +1622,25 @@ int btrfs_sysfs_add_space_info_type(struct btrfs_fs_info *fs_info,
 	return 0;
 }
 
-/* when one_device is NULL, it removes all device links */
-
-int btrfs_sysfs_rm_device_link(struct btrfs_fs_devices *fs_devices,
-		struct btrfs_device *one_device)
+void btrfs_sysfs_remove_device(struct btrfs_device *device)
 {
-	struct hd_struct *disk;
-	struct kobject *disk_kobj;
+	struct kobject *devices_kobj;
 
-	if (!fs_devices->devices_kobj)
-		return -EINVAL;
-
-	if (one_device) {
-		if (one_device->bdev) {
-			disk = one_device->bdev->bd_part;
-			disk_kobj = &part_to_dev(disk)->kobj;
-			sysfs_remove_link(fs_devices->devices_kobj,
-					  disk_kobj->name);
-		}
-
-		kobject_del(&one_device->devid_kobj);
-		kobject_put(&one_device->devid_kobj);
-
-		wait_for_completion(&one_device->kobj_unregister);
-
-		return 0;
-	}
-
-	list_for_each_entry(one_device, &fs_devices->devices, dev_list) {
+	/*
+	 * Seed fs_devices devices_kobj aren't used, fetch kobject from the
+	 * fs_info::fs_devices.
+	 */
+	devices_kobj = device->fs_info->fs_devices->devices_kobj;
+	ASSERT(devices_kobj);
 
-		if (one_device->bdev) {
-			disk = one_device->bdev->bd_part;
-			disk_kobj = &part_to_dev(disk)->kobj;
-			sysfs_remove_link(fs_devices->devices_kobj,
-					  disk_kobj->name);
-		}
-		kobject_del(&one_device->devid_kobj);
-		kobject_put(&one_device->devid_kobj);
+	if (device->bdev)
+		sysfs_remove_link(devices_kobj, bdev_kobj(device->bdev)->name);
 
-		wait_for_completion(&one_device->kobj_unregister);
+	if (device->devid_kobj.state_initialized) {
+		kobject_del(&device->devid_kobj);
+		kobject_put(&device->devid_kobj);
+		wait_for_completion(&device->kobj_unregister);
 	}
-
-	return 0;
 }
 
 static ssize_t btrfs_devinfo_in_fs_metadata_show(struct kobject *kobj,
@@ -1201,11 +1653,11 @@ static ssize_t btrfs_devinfo_in_fs_metadata_show(struct kobject *kobj,
 
 	val = !!test_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &device->dev_state);
 
-	return snprintf(buf, PAGE_SIZE, "%d\n", val);
+	return sysfs_emit(buf, "%d\n", val);
 }
 BTRFS_ATTR(devid, in_fs_metadata, btrfs_devinfo_in_fs_metadata_show);
 
-static ssize_t btrfs_sysfs_missing_show(struct kobject *kobj,
+static ssize_t btrfs_devinfo_missing_show(struct kobject *kobj,
 					struct kobj_attribute *a, char *buf)
 {
 	int val;
@@ -1214,9 +1666,9 @@ static ssize_t btrfs_sysfs_missing_show(struct kobject *kobj,
 
 	val = !!test_bit(BTRFS_DEV_STATE_MISSING, &device->dev_state);
 
-	return snprintf(buf, PAGE_SIZE, "%d\n", val);
+	return sysfs_emit(buf, "%d\n", val);
 }
-BTRFS_ATTR(devid, missing, btrfs_sysfs_missing_show);
+BTRFS_ATTR(devid, missing, btrfs_devinfo_missing_show);
 
 static ssize_t btrfs_devinfo_replace_target_show(struct kobject *kobj,
 					         struct kobj_attribute *a,
@@ -1228,10 +1680,36 @@ static ssize_t btrfs_devinfo_replace_target_show(struct kobject *kobj,
 
 	val = !!test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state);
 
-	return snprintf(buf, PAGE_SIZE, "%d\n", val);
+	return sysfs_emit(buf, "%d\n", val);
 }
 BTRFS_ATTR(devid, replace_target, btrfs_devinfo_replace_target_show);
 
+static ssize_t btrfs_devinfo_scrub_speed_max_show(struct kobject *kobj,
+					     struct kobj_attribute *a,
+					     char *buf)
+{
+	struct btrfs_device *device = container_of(kobj, struct btrfs_device,
+						   devid_kobj);
+
+	return sysfs_emit(buf, "%llu\n", READ_ONCE(device->scrub_speed_max));
+}
+
+static ssize_t btrfs_devinfo_scrub_speed_max_store(struct kobject *kobj,
+					      struct kobj_attribute *a,
+					      const char *buf, size_t len)
+{
+	struct btrfs_device *device = container_of(kobj, struct btrfs_device,
+						   devid_kobj);
+	char *endptr;
+	unsigned long long limit;
+
+	limit = memparse(buf, &endptr);
+	WRITE_ONCE(device->scrub_speed_max, limit);
+	return len;
+}
+BTRFS_ATTR_RW(devid, scrub_speed_max, btrfs_devinfo_scrub_speed_max_show,
+	      btrfs_devinfo_scrub_speed_max_store);
+
 static ssize_t btrfs_devinfo_writeable_show(struct kobject *kobj,
 					    struct kobj_attribute *a, char *buf)
 {
@@ -1241,14 +1719,60 @@ static ssize_t btrfs_devinfo_writeable_show(struct kobject *kobj,
 
 	val = !!test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state);
 
-	return snprintf(buf, PAGE_SIZE, "%d\n", val);
+	return sysfs_emit(buf, "%d\n", val);
 }
 BTRFS_ATTR(devid, writeable, btrfs_devinfo_writeable_show);
 
+static ssize_t btrfs_devinfo_fsid_show(struct kobject *kobj,
+				       struct kobj_attribute *a, char *buf)
+{
+	struct btrfs_device *device = container_of(kobj, struct btrfs_device,
+						   devid_kobj);
+
+	return sysfs_emit(buf, "%pU\n", device->fs_devices->fsid);
+}
+BTRFS_ATTR(devid, fsid, btrfs_devinfo_fsid_show);
+
+static ssize_t btrfs_devinfo_error_stats_show(struct kobject *kobj,
+		struct kobj_attribute *a, char *buf)
+{
+	struct btrfs_device *device = container_of(kobj, struct btrfs_device,
+						   devid_kobj);
+
+	if (!device->dev_stats_valid)
+		return sysfs_emit(buf, "invalid\n");
+
+	/*
+	 * Print all at once so we get a snapshot of all values from the same
+	 * time. Keep them in sync and in order of definition of
+	 * btrfs_dev_stat_values.
+	 */
+	return sysfs_emit(buf,
+		"write_errs %d\n"
+		"read_errs %d\n"
+		"flush_errs %d\n"
+		"corruption_errs %d\n"
+		"generation_errs %d\n",
+		btrfs_dev_stat_read(device, BTRFS_DEV_STAT_WRITE_ERRS),
+		btrfs_dev_stat_read(device, BTRFS_DEV_STAT_READ_ERRS),
+		btrfs_dev_stat_read(device, BTRFS_DEV_STAT_FLUSH_ERRS),
+		btrfs_dev_stat_read(device, BTRFS_DEV_STAT_CORRUPTION_ERRS),
+		btrfs_dev_stat_read(device, BTRFS_DEV_STAT_GENERATION_ERRS));
+}
+BTRFS_ATTR(devid, error_stats, btrfs_devinfo_error_stats_show);
+
+/*
+ * Information about one device.
+ *
+ * Path: /sys/fs/btrfs/<uuid>/devinfo/<devid>/
+ */
 static struct attribute *devid_attrs[] = {
+	BTRFS_ATTR_PTR(devid, error_stats),
+	BTRFS_ATTR_PTR(devid, fsid),
 	BTRFS_ATTR_PTR(devid, in_fs_metadata),
 	BTRFS_ATTR_PTR(devid, missing),
 	BTRFS_ATTR_PTR(devid, replace_target),
+	BTRFS_ATTR_PTR(devid, scrub_speed_max),
 	BTRFS_ATTR_PTR(devid, writeable),
 	NULL
 };
@@ -1269,41 +1793,76 @@ static struct kobj_type devid_ktype = {
 	.release	= btrfs_release_devid_kobj,
 };
 
-int btrfs_sysfs_add_device_link(struct btrfs_fs_devices *fs_devices,
-				struct btrfs_device *one_device)
+int btrfs_sysfs_add_device(struct btrfs_device *device)
 {
-	int error = 0;
-	struct btrfs_device *dev;
-
-	list_for_each_entry(dev, &fs_devices->devices, dev_list) {
+	int ret;
+	unsigned int nofs_flag;
+	struct kobject *devices_kobj;
+	struct kobject *devinfo_kobj;
 
-		if (one_device && one_device != dev)
-			continue;
+	/*
+	 * Make sure we use the fs_info::fs_devices to fetch the kobjects even
+	 * for the seed fs_devices
+	 */
+	devices_kobj = device->fs_info->fs_devices->devices_kobj;
+	devinfo_kobj = device->fs_info->fs_devices->devinfo_kobj;
+	ASSERT(devices_kobj);
+	ASSERT(devinfo_kobj);
 
-		if (dev->bdev) {
-			struct hd_struct *disk;
-			struct kobject *disk_kobj;
+	nofs_flag = memalloc_nofs_save();
 
-			disk = dev->bdev->bd_part;
-			disk_kobj = &part_to_dev(disk)->kobj;
+	if (device->bdev) {
+		struct kobject *disk_kobj = bdev_kobj(device->bdev);
 
-			error = sysfs_create_link(fs_devices->devices_kobj,
-						  disk_kobj, disk_kobj->name);
-			if (error)
-				break;
+		ret = sysfs_create_link(devices_kobj, disk_kobj, disk_kobj->name);
+		if (ret) {
+			btrfs_warn(device->fs_info,
+				"creating sysfs device link for devid %llu failed: %d",
+				device->devid, ret);
+			goto out;
 		}
+	}
 
-		init_completion(&dev->kobj_unregister);
-		error = kobject_init_and_add(&dev->devid_kobj, &devid_ktype,
-					     fs_devices->devinfo_kobj, "%llu",
-					     dev->devid);
-		if (error) {
-			kobject_put(&dev->devid_kobj);
-			break;
+	init_completion(&device->kobj_unregister);
+	ret = kobject_init_and_add(&device->devid_kobj, &devid_ktype,
+				   devinfo_kobj, "%llu", device->devid);
+	if (ret) {
+		kobject_put(&device->devid_kobj);
+		btrfs_warn(device->fs_info,
+			   "devinfo init for devid %llu failed: %d",
+			   device->devid, ret);
+	}
+
+out:
+	memalloc_nofs_restore(nofs_flag);
+	return ret;
+}
+
+static int btrfs_sysfs_add_fs_devices(struct btrfs_fs_devices *fs_devices)
+{
+	int ret;
+	struct btrfs_device *device;
+	struct btrfs_fs_devices *seed;
+
+	list_for_each_entry(device, &fs_devices->devices, dev_list) {
+		ret = btrfs_sysfs_add_device(device);
+		if (ret)
+			goto fail;
+	}
+
+	list_for_each_entry(seed, &fs_devices->seed_list, seed_list) {
+		list_for_each_entry(device, &seed->devices, dev_list) {
+			ret = btrfs_sysfs_add_device(device);
+			if (ret)
+				goto fail;
 		}
 	}
 
-	return error;
+	return 0;
+
+fail:
+	btrfs_sysfs_remove_fs_devices(fs_devices);
+	return ret;
 }
 
 void btrfs_kobject_uevent(struct block_device *bdev, enum kobject_action action)
@@ -1317,8 +1876,8 @@ void btrfs_kobject_uevent(struct block_device *bdev, enum kobject_action action)
 			&disk_to_dev(bdev->bd_disk)->kobj);
 }
 
-void btrfs_sysfs_update_sprout_fsid(struct btrfs_fs_devices *fs_devices,
-				    const u8 *fsid)
+void btrfs_sysfs_update_sprout_fsid(struct btrfs_fs_devices *fs_devices)
+
 {
 	char fsid_buf[BTRFS_UUID_UNPARSED_SIZE];
 
@@ -1326,7 +1885,7 @@ void btrfs_sysfs_update_sprout_fsid(struct btrfs_fs_devices *fs_devices,
 	 * Sprouting changes fsid of the mounted filesystem, rename the fsid
 	 * directory
 	 */
-	snprintf(fsid_buf, BTRFS_UUID_UNPARSED_SIZE, "%pU", fsid);
+	snprintf(fsid_buf, BTRFS_UUID_UNPARSED_SIZE, "%pU", fs_devices->fsid);
 	if (kobject_rename(&fs_devices->fsid_kobj, fsid_buf))
 		btrfs_warn(fs_devices->fs_info,
 				"sysfs: failed to create fsid for sprout");
@@ -1371,7 +1930,7 @@ int btrfs_sysfs_add_fsid(struct btrfs_fs_devices *fs_devs)
 	if (!fs_devs->devices_kobj) {
 		btrfs_err(fs_devs->fs_info,
 			  "failed to init sysfs device interface");
-		kobject_put(&fs_devs->fsid_kobj);
+		btrfs_sysfs_remove_fsid(fs_devs);
 		return -ENOMEM;
 	}
 
@@ -1393,15 +1952,13 @@ int btrfs_sysfs_add_mounted(struct btrfs_fs_info *fs_info)
 	struct btrfs_fs_devices *fs_devs = fs_info->fs_devices;
 	struct kobject *fsid_kobj = &fs_devs->fsid_kobj;
 
-	btrfs_set_fs_info_ptr(fs_info);
-
-	error = btrfs_sysfs_add_device_link(fs_devs, NULL);
+	error = btrfs_sysfs_add_fs_devices(fs_devs);
 	if (error)
 		return error;
 
 	error = sysfs_create_files(fsid_kobj, btrfs_attrs);
 	if (error) {
-		btrfs_sysfs_rm_device_link(fs_devs, NULL);
+		btrfs_sysfs_remove_fs_devices(fs_devs);
 		return error;
 	}
 
@@ -1420,25 +1977,27 @@ int btrfs_sysfs_add_mounted(struct btrfs_fs_info *fs_info)
 	error = sysfs_create_files(fs_info->debug_kobj, btrfs_debug_mount_attrs);
 	if (error)
 		goto failure;
+#endif
 
 	/* Discard directory */
-	fs_info->discard_debug_kobj = kobject_create_and_add("discard",
-						     fs_info->debug_kobj);
-	if (!fs_info->discard_debug_kobj) {
+	fs_info->discard_kobj = kobject_create_and_add("discard", fsid_kobj);
+	if (!fs_info->discard_kobj) {
 		error = -ENOMEM;
 		goto failure;
 	}
 
-	error = sysfs_create_files(fs_info->discard_debug_kobj,
-				   discard_debug_attrs);
+	error = sysfs_create_files(fs_info->discard_kobj, discard_attrs);
 	if (error)
 		goto failure;
-#endif
 
 	error = addrm_unknown_feature_attrs(fs_info, true);
 	if (error)
 		goto failure;
 
+	error = sysfs_create_link(fsid_kobj, &fs_info->sb->s_bdi->dev->kobj, "bdi");
+	if (error)
+		goto failure;
+
 	fs_info->space_info_kobj = kobject_create_and_add("allocation",
 						  fsid_kobj);
 	if (!fs_info->space_info_kobj) {
@@ -1456,6 +2015,256 @@ failure:
 	return error;
 }
 
+static ssize_t qgroup_enabled_show(struct kobject *qgroups_kobj,
+				   struct kobj_attribute *a,
+				   char *buf)
+{
+	struct btrfs_fs_info *fs_info = to_fs_info(qgroups_kobj->parent);
+	bool enabled;
+
+	spin_lock(&fs_info->qgroup_lock);
+	enabled = fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_ON;
+	spin_unlock(&fs_info->qgroup_lock);
+
+	return sysfs_emit(buf, "%d\n", enabled);
+}
+BTRFS_ATTR(qgroups, enabled, qgroup_enabled_show);
+
+static ssize_t qgroup_inconsistent_show(struct kobject *qgroups_kobj,
+					struct kobj_attribute *a,
+					char *buf)
+{
+	struct btrfs_fs_info *fs_info = to_fs_info(qgroups_kobj->parent);
+	bool inconsistent;
+
+	spin_lock(&fs_info->qgroup_lock);
+	inconsistent = (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT);
+	spin_unlock(&fs_info->qgroup_lock);
+
+	return sysfs_emit(buf, "%d\n", inconsistent);
+}
+BTRFS_ATTR(qgroups, inconsistent, qgroup_inconsistent_show);
+
+static ssize_t qgroup_drop_subtree_thres_show(struct kobject *qgroups_kobj,
+					      struct kobj_attribute *a,
+					      char *buf)
+{
+	struct btrfs_fs_info *fs_info = to_fs_info(qgroups_kobj->parent);
+	u8 result;
+
+	spin_lock(&fs_info->qgroup_lock);
+	result = fs_info->qgroup_drop_subtree_thres;
+	spin_unlock(&fs_info->qgroup_lock);
+
+	return sysfs_emit(buf, "%d\n", result);
+}
+
+static ssize_t qgroup_drop_subtree_thres_store(struct kobject *qgroups_kobj,
+					       struct kobj_attribute *a,
+					       const char *buf, size_t len)
+{
+	struct btrfs_fs_info *fs_info = to_fs_info(qgroups_kobj->parent);
+	u8 new_thres;
+	int ret;
+
+	ret = kstrtou8(buf, 10, &new_thres);
+	if (ret)
+		return -EINVAL;
+
+	if (new_thres > BTRFS_MAX_LEVEL)
+		return -EINVAL;
+
+	spin_lock(&fs_info->qgroup_lock);
+	fs_info->qgroup_drop_subtree_thres = new_thres;
+	spin_unlock(&fs_info->qgroup_lock);
+
+	return len;
+}
+BTRFS_ATTR_RW(qgroups, drop_subtree_threshold, qgroup_drop_subtree_thres_show,
+	      qgroup_drop_subtree_thres_store);
+
+/*
+ * Qgroups global info
+ *
+ * Path: /sys/fs/btrfs/<uuid>/qgroups/
+ */
+static struct attribute *qgroups_attrs[] = {
+	BTRFS_ATTR_PTR(qgroups, enabled),
+	BTRFS_ATTR_PTR(qgroups, inconsistent),
+	BTRFS_ATTR_PTR(qgroups, drop_subtree_threshold),
+	NULL
+};
+ATTRIBUTE_GROUPS(qgroups);
+
+static void qgroups_release(struct kobject *kobj)
+{
+	kfree(kobj);
+}
+
+static struct kobj_type qgroups_ktype = {
+	.sysfs_ops = &kobj_sysfs_ops,
+	.default_groups = qgroups_groups,
+	.release = qgroups_release,
+};
+
+static inline struct btrfs_fs_info *qgroup_kobj_to_fs_info(struct kobject *kobj)
+{
+	return to_fs_info(kobj->parent->parent);
+}
+
+#define QGROUP_ATTR(_member, _show_name)					\
+static ssize_t btrfs_qgroup_show_##_member(struct kobject *qgroup_kobj,		\
+					   struct kobj_attribute *a,		\
+					   char *buf)				\
+{										\
+	struct btrfs_fs_info *fs_info = qgroup_kobj_to_fs_info(qgroup_kobj);	\
+	struct btrfs_qgroup *qgroup = container_of(qgroup_kobj,			\
+			struct btrfs_qgroup, kobj);				\
+	return btrfs_show_u64(&qgroup->_member, &fs_info->qgroup_lock, buf);	\
+}										\
+BTRFS_ATTR(qgroup, _show_name, btrfs_qgroup_show_##_member)
+
+#define QGROUP_RSV_ATTR(_name, _type)						\
+static ssize_t btrfs_qgroup_rsv_show_##_name(struct kobject *qgroup_kobj,	\
+					     struct kobj_attribute *a,		\
+					     char *buf)				\
+{										\
+	struct btrfs_fs_info *fs_info = qgroup_kobj_to_fs_info(qgroup_kobj);	\
+	struct btrfs_qgroup *qgroup = container_of(qgroup_kobj,			\
+			struct btrfs_qgroup, kobj);				\
+	return btrfs_show_u64(&qgroup->rsv.values[_type],			\
+			&fs_info->qgroup_lock, buf);				\
+}										\
+BTRFS_ATTR(qgroup, rsv_##_name, btrfs_qgroup_rsv_show_##_name)
+
+QGROUP_ATTR(rfer, referenced);
+QGROUP_ATTR(excl, exclusive);
+QGROUP_ATTR(max_rfer, max_referenced);
+QGROUP_ATTR(max_excl, max_exclusive);
+QGROUP_ATTR(lim_flags, limit_flags);
+QGROUP_RSV_ATTR(data, BTRFS_QGROUP_RSV_DATA);
+QGROUP_RSV_ATTR(meta_pertrans, BTRFS_QGROUP_RSV_META_PERTRANS);
+QGROUP_RSV_ATTR(meta_prealloc, BTRFS_QGROUP_RSV_META_PREALLOC);
+
+/*
+ * Qgroup information.
+ *
+ * Path: /sys/fs/btrfs/<uuid>/qgroups/<level>_<qgroupid>/
+ */
+static struct attribute *qgroup_attrs[] = {
+	BTRFS_ATTR_PTR(qgroup, referenced),
+	BTRFS_ATTR_PTR(qgroup, exclusive),
+	BTRFS_ATTR_PTR(qgroup, max_referenced),
+	BTRFS_ATTR_PTR(qgroup, max_exclusive),
+	BTRFS_ATTR_PTR(qgroup, limit_flags),
+	BTRFS_ATTR_PTR(qgroup, rsv_data),
+	BTRFS_ATTR_PTR(qgroup, rsv_meta_pertrans),
+	BTRFS_ATTR_PTR(qgroup, rsv_meta_prealloc),
+	NULL
+};
+ATTRIBUTE_GROUPS(qgroup);
+
+static void qgroup_release(struct kobject *kobj)
+{
+	struct btrfs_qgroup *qgroup = container_of(kobj, struct btrfs_qgroup, kobj);
+
+	memset(&qgroup->kobj, 0, sizeof(*kobj));
+}
+
+static struct kobj_type qgroup_ktype = {
+	.sysfs_ops = &kobj_sysfs_ops,
+	.release = qgroup_release,
+	.default_groups = qgroup_groups,
+};
+
+int btrfs_sysfs_add_one_qgroup(struct btrfs_fs_info *fs_info,
+				struct btrfs_qgroup *qgroup)
+{
+	struct kobject *qgroups_kobj = fs_info->qgroups_kobj;
+	int ret;
+
+	if (test_bit(BTRFS_FS_STATE_DUMMY_FS_INFO, &fs_info->fs_state))
+		return 0;
+	if (qgroup->kobj.state_initialized)
+		return 0;
+	if (!qgroups_kobj)
+		return -EINVAL;
+
+	ret = kobject_init_and_add(&qgroup->kobj, &qgroup_ktype, qgroups_kobj,
+			"%hu_%llu", btrfs_qgroup_level(qgroup->qgroupid),
+			btrfs_qgroup_subvolid(qgroup->qgroupid));
+	if (ret < 0)
+		kobject_put(&qgroup->kobj);
+
+	return ret;
+}
+
+void btrfs_sysfs_del_qgroups(struct btrfs_fs_info *fs_info)
+{
+	struct btrfs_qgroup *qgroup;
+	struct btrfs_qgroup *next;
+
+	if (test_bit(BTRFS_FS_STATE_DUMMY_FS_INFO, &fs_info->fs_state))
+		return;
+
+	rbtree_postorder_for_each_entry_safe(qgroup, next,
+					     &fs_info->qgroup_tree, node)
+		btrfs_sysfs_del_one_qgroup(fs_info, qgroup);
+	if (fs_info->qgroups_kobj) {
+		kobject_del(fs_info->qgroups_kobj);
+		kobject_put(fs_info->qgroups_kobj);
+		fs_info->qgroups_kobj = NULL;
+	}
+}
+
+/* Called when qgroups get initialized, thus there is no need for locking */
+int btrfs_sysfs_add_qgroups(struct btrfs_fs_info *fs_info)
+{
+	struct kobject *fsid_kobj = &fs_info->fs_devices->fsid_kobj;
+	struct btrfs_qgroup *qgroup;
+	struct btrfs_qgroup *next;
+	int ret = 0;
+
+	if (test_bit(BTRFS_FS_STATE_DUMMY_FS_INFO, &fs_info->fs_state))
+		return 0;
+
+	ASSERT(fsid_kobj);
+	if (fs_info->qgroups_kobj)
+		return 0;
+
+	fs_info->qgroups_kobj = kzalloc(sizeof(struct kobject), GFP_KERNEL);
+	if (!fs_info->qgroups_kobj)
+		return -ENOMEM;
+
+	ret = kobject_init_and_add(fs_info->qgroups_kobj, &qgroups_ktype,
+				   fsid_kobj, "qgroups");
+	if (ret < 0)
+		goto out;
+
+	rbtree_postorder_for_each_entry_safe(qgroup, next,
+					     &fs_info->qgroup_tree, node) {
+		ret = btrfs_sysfs_add_one_qgroup(fs_info, qgroup);
+		if (ret < 0)
+			goto out;
+	}
+
+out:
+	if (ret < 0)
+		btrfs_sysfs_del_qgroups(fs_info);
+	return ret;
+}
+
+void btrfs_sysfs_del_one_qgroup(struct btrfs_fs_info *fs_info,
+				struct btrfs_qgroup *qgroup)
+{
+	if (test_bit(BTRFS_FS_STATE_DUMMY_FS_INFO, &fs_info->fs_state))
+		return;
+
+	if (qgroup->kobj.state_initialized) {
+		kobject_del(&qgroup->kobj);
+		kobject_put(&qgroup->kobj);
+	}
+}
 
 /*
  * Change per-fs features in /sys/fs/btrfs/UUID/features to match current
@@ -1466,12 +2275,16 @@ void btrfs_sysfs_feature_update(struct btrfs_fs_info *fs_info,
 {
 	struct btrfs_fs_devices *fs_devs;
 	struct kobject *fsid_kobj;
-	u64 features;
-	int ret;
+	u64 __maybe_unused features;
+	int __maybe_unused ret;
 
 	if (!fs_info)
 		return;
 
+	/*
+	 * See 14e46e04958df74 and e410e34fad913dd, feature bit updates are not
+	 * safe when called from some contexts (eg. balance)
+	 */
 	features = get_features(fs_info, set);
 	ASSERT(bit & supported_feature_masks[set]);
 
@@ -1532,4 +2345,3 @@ void __cold btrfs_exit_sysfs(void)
 #endif
 	kset_unregister(btrfs_kset);
 }
-
diff --git a/fs/btrfs/sysfs.h b/fs/btrfs/sysfs.h
index c68582add92e..bacef43f7267 100644
--- a/fs/btrfs/sysfs.h
+++ b/fs/btrfs/sysfs.h
@@ -13,15 +13,12 @@ enum btrfs_feature_set {
 };
 
 char *btrfs_printable_features(enum btrfs_feature_set set, u64 flags);
-const char * const btrfs_feature_set_name(enum btrfs_feature_set set);
-int btrfs_sysfs_add_device_link(struct btrfs_fs_devices *fs_devices,
-		struct btrfs_device *one_device);
-int btrfs_sysfs_rm_device_link(struct btrfs_fs_devices *fs_devices,
-                struct btrfs_device *one_device);
+const char *btrfs_feature_set_name(enum btrfs_feature_set set);
+int btrfs_sysfs_add_device(struct btrfs_device *device);
+void btrfs_sysfs_remove_device(struct btrfs_device *device);
 int btrfs_sysfs_add_fsid(struct btrfs_fs_devices *fs_devs);
 void btrfs_sysfs_remove_fsid(struct btrfs_fs_devices *fs_devs);
-void btrfs_sysfs_update_sprout_fsid(struct btrfs_fs_devices *fs_devices,
-				    const u8 *fsid);
+void btrfs_sysfs_update_sprout_fsid(struct btrfs_fs_devices *fs_devices);
 void btrfs_sysfs_feature_update(struct btrfs_fs_info *fs_info,
 		u64 bit, enum btrfs_feature_set set);
 void btrfs_kobject_uevent(struct block_device *bdev, enum kobject_action action);
@@ -36,4 +33,11 @@ int btrfs_sysfs_add_space_info_type(struct btrfs_fs_info *fs_info,
 void btrfs_sysfs_remove_space_info(struct btrfs_space_info *space_info);
 void btrfs_sysfs_update_devid(struct btrfs_device *device);
 
+int btrfs_sysfs_add_one_qgroup(struct btrfs_fs_info *fs_info,
+				struct btrfs_qgroup *qgroup);
+void btrfs_sysfs_del_qgroups(struct btrfs_fs_info *fs_info);
+int btrfs_sysfs_add_qgroups(struct btrfs_fs_info *fs_info);
+void btrfs_sysfs_del_one_qgroup(struct btrfs_fs_info *fs_info,
+				struct btrfs_qgroup *qgroup);
+
 #endif
diff --git a/fs/btrfs/tests/btrfs-tests.c b/fs/btrfs/tests/btrfs-tests.c
index 84fb3fa940a6..d43cb5242fec 100644
--- a/fs/btrfs/tests/btrfs-tests.c
+++ b/fs/btrfs/tests/btrfs-tests.c
@@ -55,8 +55,15 @@ struct inode *btrfs_new_test_inode(void)
 	struct inode *inode;
 
 	inode = new_inode(test_mnt->mnt_sb);
-	if (inode)
-		inode_init_owner(inode, NULL, S_IFREG);
+	if (!inode)
+		return NULL;
+
+	inode->i_mode = S_IFREG;
+	inode->i_ino = BTRFS_FIRST_FREE_OBJECTID;
+	BTRFS_I(inode)->location.type = BTRFS_INODE_ITEM_KEY;
+	BTRFS_I(inode)->location.objectid = BTRFS_FIRST_FREE_OBJECTID;
+	BTRFS_I(inode)->location.offset = 0;
+	inode_init_owner(&init_user_ns, inode, NULL, S_IFREG);
 
 	return inode;
 }
@@ -120,6 +127,8 @@ struct btrfs_fs_info *btrfs_alloc_dummy_fs_info(u32 nodesize, u32 sectorsize)
 		kfree(fs_info);
 		return NULL;
 	}
+	INIT_LIST_HEAD(&fs_info->fs_devices->devices);
+
 	fs_info->super_copy = kzalloc(sizeof(struct btrfs_super_block),
 				      GFP_KERNEL);
 	if (!fs_info->super_copy) {
@@ -128,39 +137,11 @@ struct btrfs_fs_info *btrfs_alloc_dummy_fs_info(u32 nodesize, u32 sectorsize)
 		return NULL;
 	}
 
+	btrfs_init_fs_info(fs_info);
+
 	fs_info->nodesize = nodesize;
 	fs_info->sectorsize = sectorsize;
-
-	if (init_srcu_struct(&fs_info->subvol_srcu)) {
-		kfree(fs_info->fs_devices);
-		kfree(fs_info->super_copy);
-		kfree(fs_info);
-		return NULL;
-	}
-
-	spin_lock_init(&fs_info->buffer_lock);
-	spin_lock_init(&fs_info->qgroup_lock);
-	spin_lock_init(&fs_info->super_lock);
-	spin_lock_init(&fs_info->fs_roots_radix_lock);
-	mutex_init(&fs_info->qgroup_ioctl_lock);
-	mutex_init(&fs_info->qgroup_rescan_lock);
-	rwlock_init(&fs_info->tree_mod_log_lock);
-	fs_info->running_transaction = NULL;
-	fs_info->qgroup_tree = RB_ROOT;
-	fs_info->qgroup_ulist = NULL;
-	atomic64_set(&fs_info->tree_mod_seq, 0);
-	INIT_LIST_HEAD(&fs_info->dirty_qgroups);
-	INIT_LIST_HEAD(&fs_info->dead_roots);
-	INIT_LIST_HEAD(&fs_info->tree_mod_seq_list);
-	INIT_LIST_HEAD(&fs_info->fs_devices->devices);
-	INIT_RADIX_TREE(&fs_info->buffer_radix, GFP_ATOMIC);
-	INIT_RADIX_TREE(&fs_info->fs_roots_radix, GFP_ATOMIC);
-	extent_io_tree_init(fs_info, &fs_info->freed_extents[0],
-			    IO_TREE_FS_INFO_FREED_EXTENTS0, NULL);
-	extent_io_tree_init(fs_info, &fs_info->freed_extents[1],
-			    IO_TREE_FS_INFO_FREED_EXTENTS1, NULL);
-	extent_map_tree_init(&fs_info->mapping_tree);
-	fs_info->pinned_extents = &fs_info->freed_extents[0];
+	fs_info->sectorsize_bits = ilog2(sectorsize);
 	set_bit(BTRFS_FS_STATE_DUMMY_FS_INFO, &fs_info->fs_state);
 
 	test_mnt->mnt_sb->s_fs_info = fs_info;
@@ -210,24 +191,22 @@ void btrfs_free_dummy_fs_info(struct btrfs_fs_info *fs_info)
 	}
 	btrfs_free_qgroup_config(fs_info);
 	btrfs_free_fs_roots(fs_info);
-	cleanup_srcu_struct(&fs_info->subvol_srcu);
 	kfree(fs_info->super_copy);
+	btrfs_check_leaked_roots(fs_info);
+	btrfs_extent_buffer_leak_debug_check(fs_info);
 	kfree(fs_info->fs_devices);
 	kfree(fs_info);
 }
 
 void btrfs_free_dummy_root(struct btrfs_root *root)
 {
-	if (!root)
+	if (IS_ERR_OR_NULL(root))
 		return;
 	/* Will be freed by btrfs_free_fs_roots */
 	if (WARN_ON(test_bit(BTRFS_ROOT_IN_RADIX, &root->state)))
 		return;
-	if (root->node) {
-		/* One for allocate_extent_buffer */
-		free_extent_buffer(root->node);
-	}
-	kfree(root);
+	btrfs_global_root_delete(root);
+	btrfs_put_root(root);
 }
 
 struct btrfs_block_group *
@@ -254,7 +233,7 @@ btrfs_alloc_dummy_block_group(struct btrfs_fs_info *fs_info,
 	INIT_LIST_HEAD(&cache->list);
 	INIT_LIST_HEAD(&cache->cluster_list);
 	INIT_LIST_HEAD(&cache->bg_list);
-	btrfs_init_free_space_ctl(cache);
+	btrfs_init_free_space_ctl(cache, cache->free_space_ctl);
 	mutex_init(&cache->free_space_lock);
 
 	return cache;
@@ -264,7 +243,7 @@ void btrfs_free_dummy_block_group(struct btrfs_block_group *cache)
 {
 	if (!cache)
 		return;
-	__btrfs_remove_free_space_cache(cache->free_space_ctl);
+	btrfs_remove_free_space_cache(cache);
 	kfree(cache->free_space_ctl);
 	kfree(cache);
 }
diff --git a/fs/btrfs/tests/extent-buffer-tests.c b/fs/btrfs/tests/extent-buffer-tests.c
index a1b9f9b5978e..b7d181a08eab 100644
--- a/fs/btrfs/tests/extent-buffer-tests.c
+++ b/fs/btrfs/tests/extent-buffer-tests.c
@@ -15,7 +15,6 @@ static int test_btrfs_split_item(u32 sectorsize, u32 nodesize)
 	struct btrfs_path *path = NULL;
 	struct btrfs_root *root = NULL;
 	struct extent_buffer *eb;
-	struct btrfs_item *item;
 	char *value = "mary had a little lamb";
 	char *split1 = "mary had a little";
 	char *split2 = " lamb";
@@ -48,7 +47,8 @@ static int test_btrfs_split_item(u32 sectorsize, u32 nodesize)
 		goto out;
 	}
 
-	path->nodes[0] = eb = alloc_dummy_extent_buffer(fs_info, nodesize);
+	eb = alloc_dummy_extent_buffer(fs_info, nodesize);
+	path->nodes[0] = eb;
 	if (!eb) {
 		test_std_err(TEST_ALLOC_EXTENT_BUFFER);
 		ret = -ENOMEM;
@@ -60,9 +60,7 @@ static int test_btrfs_split_item(u32 sectorsize, u32 nodesize)
 	key.type = BTRFS_EXTENT_CSUM_KEY;
 	key.offset = 0;
 
-	setup_items_for_insert(root, path, &key, &value_len, value_len,
-			       value_len + sizeof(struct btrfs_item), 1);
-	item = btrfs_item_nr(0);
+	btrfs_setup_item_for_insert(root, path, &key, value_len);
 	write_extent_buffer(eb, value, btrfs_item_ptr_offset(eb, 0),
 			    value_len);
 
@@ -91,8 +89,7 @@ static int test_btrfs_split_item(u32 sectorsize, u32 nodesize)
 		goto out;
 	}
 
-	item = btrfs_item_nr(0);
-	if (btrfs_item_size(eb, item) != strlen(split1)) {
+	if (btrfs_item_size(eb, 0) != strlen(split1)) {
 		test_err("invalid len in the first split");
 		ret = -EINVAL;
 		goto out;
@@ -116,8 +113,7 @@ static int test_btrfs_split_item(u32 sectorsize, u32 nodesize)
 		goto out;
 	}
 
-	item = btrfs_item_nr(1);
-	if (btrfs_item_size(eb, item) != strlen(split2)) {
+	if (btrfs_item_size(eb, 1) != strlen(split2)) {
 		test_err("invalid len in the second split");
 		ret = -EINVAL;
 		goto out;
@@ -148,8 +144,7 @@ static int test_btrfs_split_item(u32 sectorsize, u32 nodesize)
 		goto out;
 	}
 
-	item = btrfs_item_nr(0);
-	if (btrfs_item_size(eb, item) != strlen(split3)) {
+	if (btrfs_item_size(eb, 0) != strlen(split3)) {
 		test_err("invalid len in the first split");
 		ret = -EINVAL;
 		goto out;
@@ -172,8 +167,7 @@ static int test_btrfs_split_item(u32 sectorsize, u32 nodesize)
 		goto out;
 	}
 
-	item = btrfs_item_nr(1);
-	if (btrfs_item_size(eb, item) != strlen(split4)) {
+	if (btrfs_item_size(eb, 1) != strlen(split4)) {
 		test_err("invalid len in the second split");
 		ret = -EINVAL;
 		goto out;
@@ -196,8 +190,7 @@ static int test_btrfs_split_item(u32 sectorsize, u32 nodesize)
 		goto out;
 	}
 
-	item = btrfs_item_nr(2);
-	if (btrfs_item_size(eb, item) != strlen(split2)) {
+	if (btrfs_item_size(eb, 2) != strlen(split2)) {
 		test_err("invalid len in the second split");
 		ret = -EINVAL;
 		goto out;
diff --git a/fs/btrfs/tests/extent-io-tests.c b/fs/btrfs/tests/extent-io-tests.c
index df7ce874a74b..350da449db08 100644
--- a/fs/btrfs/tests/extent-io-tests.c
+++ b/fs/btrfs/tests/extent-io-tests.c
@@ -4,6 +4,7 @@
  */
 
 #include <linux/pagemap.h>
+#include <linux/pagevec.h>
 #include <linux/sched.h>
 #include <linux/slab.h>
 #include <linux/sizes.h>
@@ -20,42 +21,90 @@ static noinline int process_page_range(struct inode *inode, u64 start, u64 end,
 				       unsigned long flags)
 {
 	int ret;
-	struct page *pages[16];
+	struct folio_batch fbatch;
 	unsigned long index = start >> PAGE_SHIFT;
 	unsigned long end_index = end >> PAGE_SHIFT;
-	unsigned long nr_pages = end_index - index + 1;
 	int i;
 	int count = 0;
 	int loops = 0;
 
-	while (nr_pages > 0) {
-		ret = find_get_pages_contig(inode->i_mapping, index,
-				     min_t(unsigned long, nr_pages,
-				     ARRAY_SIZE(pages)), pages);
+	folio_batch_init(&fbatch);
+
+	while (index <= end_index) {
+		ret = filemap_get_folios_contig(inode->i_mapping, &index,
+				end_index, &fbatch);
 		for (i = 0; i < ret; i++) {
+			struct folio *folio = fbatch.folios[i];
+
 			if (flags & PROCESS_TEST_LOCKED &&
-			    !PageLocked(pages[i]))
+			    !folio_test_locked(folio))
 				count++;
-			if (flags & PROCESS_UNLOCK && PageLocked(pages[i]))
-				unlock_page(pages[i]);
-			put_page(pages[i]);
+			if (flags & PROCESS_UNLOCK && folio_test_locked(folio))
+				folio_unlock(folio);
 			if (flags & PROCESS_RELEASE)
-				put_page(pages[i]);
+				folio_put(folio);
 		}
-		nr_pages -= ret;
-		index += ret;
+		folio_batch_release(&fbatch);
 		cond_resched();
 		loops++;
 		if (loops > 100000) {
 			printk(KERN_ERR
-		"stuck in a loop, start %llu, end %llu, nr_pages %lu, ret %d\n",
-				start, end, nr_pages, ret);
+		"stuck in a loop, start %llu, end %llu, ret %d\n",
+				start, end, ret);
 			break;
 		}
 	}
+
 	return count;
 }
 
+#define STATE_FLAG_STR_LEN			256
+
+#define PRINT_ONE_FLAG(state, dest, cur, name)				\
+({									\
+	if (state->state & EXTENT_##name)				\
+		cur += scnprintf(dest + cur, STATE_FLAG_STR_LEN - cur,	\
+				 "%s" #name, cur == 0 ? "" : "|");	\
+})
+
+static void extent_flag_to_str(const struct extent_state *state, char *dest)
+{
+	int cur = 0;
+
+	dest[0] = 0;
+	PRINT_ONE_FLAG(state, dest, cur, DIRTY);
+	PRINT_ONE_FLAG(state, dest, cur, UPTODATE);
+	PRINT_ONE_FLAG(state, dest, cur, LOCKED);
+	PRINT_ONE_FLAG(state, dest, cur, NEW);
+	PRINT_ONE_FLAG(state, dest, cur, DELALLOC);
+	PRINT_ONE_FLAG(state, dest, cur, DEFRAG);
+	PRINT_ONE_FLAG(state, dest, cur, BOUNDARY);
+	PRINT_ONE_FLAG(state, dest, cur, NODATASUM);
+	PRINT_ONE_FLAG(state, dest, cur, CLEAR_META_RESV);
+	PRINT_ONE_FLAG(state, dest, cur, NEED_WAIT);
+	PRINT_ONE_FLAG(state, dest, cur, NORESERVE);
+	PRINT_ONE_FLAG(state, dest, cur, QGROUP_RESERVED);
+	PRINT_ONE_FLAG(state, dest, cur, CLEAR_DATA_RESV);
+}
+
+static void dump_extent_io_tree(const struct extent_io_tree *tree)
+{
+	struct rb_node *node;
+	char flags_str[STATE_FLAG_STR_LEN];
+
+	node = rb_first(&tree->state);
+	test_msg("io tree content:");
+	while (node) {
+		struct extent_state *state;
+
+		state = rb_entry(node, struct extent_state, rb_node);
+		extent_flag_to_str(state, flags_str);
+		test_msg("  start=%llu len=%llu flags=%s", state->start,
+			 state->end + 1 - state->start, flags_str);
+		node = rb_next(node);
+	}
+}
+
 static int test_find_delalloc(u32 sectorsize)
 {
 	struct inode *inode;
@@ -112,7 +161,7 @@ static int test_find_delalloc(u32 sectorsize)
 	 */
 	set_extent_delalloc(tmp, 0, sectorsize - 1, 0, NULL);
 	start = 0;
-	end = 0;
+	end = start + PAGE_SIZE - 1;
 	found = find_lock_delalloc_range(inode, locked_page, &start,
 					 &end);
 	if (!found) {
@@ -124,7 +173,7 @@ static int test_find_delalloc(u32 sectorsize)
 			sectorsize - 1, start, end);
 		goto out_bits;
 	}
-	unlock_extent(tmp, start, end);
+	unlock_extent(tmp, start, end, NULL);
 	unlock_page(locked_page);
 	put_page(locked_page);
 
@@ -143,7 +192,7 @@ static int test_find_delalloc(u32 sectorsize)
 	}
 	set_extent_delalloc(tmp, sectorsize, max_bytes - 1, 0, NULL);
 	start = test_start;
-	end = 0;
+	end = start + PAGE_SIZE - 1;
 	found = find_lock_delalloc_range(inode, locked_page, &start,
 					 &end);
 	if (!found) {
@@ -160,7 +209,7 @@ static int test_find_delalloc(u32 sectorsize)
 		test_err("there were unlocked pages in the range");
 		goto out_bits;
 	}
-	unlock_extent(tmp, start, end);
+	unlock_extent(tmp, start, end, NULL);
 	/* locked_page was unlocked above */
 	put_page(locked_page);
 
@@ -177,14 +226,14 @@ static int test_find_delalloc(u32 sectorsize)
 		goto out_bits;
 	}
 	start = test_start;
-	end = 0;
+	end = start + PAGE_SIZE - 1;
 	found = find_lock_delalloc_range(inode, locked_page, &start,
 					 &end);
 	if (found) {
 		test_err("found range when we shouldn't have");
 		goto out_bits;
 	}
-	if (end != (u64)-1) {
+	if (end != test_start + PAGE_SIZE - 1) {
 		test_err("did not return the proper end offset");
 		goto out_bits;
 	}
@@ -198,7 +247,7 @@ static int test_find_delalloc(u32 sectorsize)
 	 */
 	set_extent_delalloc(tmp, max_bytes, total_dirty - 1, 0, NULL);
 	start = test_start;
-	end = 0;
+	end = start + PAGE_SIZE - 1;
 	found = find_lock_delalloc_range(inode, locked_page, &start,
 					 &end);
 	if (!found) {
@@ -215,7 +264,7 @@ static int test_find_delalloc(u32 sectorsize)
 		test_err("pages in range were not all locked");
 		goto out_bits;
 	}
-	unlock_extent(tmp, start, end);
+	unlock_extent(tmp, start, end, NULL);
 
 	/*
 	 * Now to test where we run into a page that is no longer dirty in the
@@ -233,7 +282,7 @@ static int test_find_delalloc(u32 sectorsize)
 	/* We unlocked it in the previous test */
 	lock_page(locked_page);
 	start = test_start;
-	end = 0;
+	end = start + PAGE_SIZE - 1;
 	/*
 	 * Currently if we fail to find dirty pages in the delalloc range we
 	 * will adjust max_bytes down to PAGE_SIZE and then re-search.  If
@@ -258,6 +307,8 @@ static int test_find_delalloc(u32 sectorsize)
 	}
 	ret = 0;
 out_bits:
+	if (ret)
+		dump_extent_io_tree(tmp);
 	clear_extent_bits(tmp, 0, total_dirty - 1, (unsigned)-1);
 out:
 	if (locked_page)
@@ -379,54 +430,50 @@ static int __test_eb_bitmaps(unsigned long *bitmap, struct extent_buffer *eb,
 static int test_eb_bitmaps(u32 sectorsize, u32 nodesize)
 {
 	struct btrfs_fs_info *fs_info;
-	unsigned long len;
 	unsigned long *bitmap = NULL;
 	struct extent_buffer *eb = NULL;
 	int ret;
 
 	test_msg("running extent buffer bitmap tests");
 
-	/*
-	 * In ppc64, sectorsize can be 64K, thus 4 * 64K will be larger than
-	 * BTRFS_MAX_METADATA_BLOCKSIZE.
-	 */
-	len = (sectorsize < BTRFS_MAX_METADATA_BLOCKSIZE)
-		? sectorsize * 4 : sectorsize;
-
-	fs_info = btrfs_alloc_dummy_fs_info(len, len);
+	fs_info = btrfs_alloc_dummy_fs_info(nodesize, sectorsize);
 	if (!fs_info) {
 		test_std_err(TEST_ALLOC_FS_INFO);
 		return -ENOMEM;
 	}
 
-	bitmap = kmalloc(len, GFP_KERNEL);
+	bitmap = kmalloc(nodesize, GFP_KERNEL);
 	if (!bitmap) {
 		test_err("couldn't allocate test bitmap");
 		ret = -ENOMEM;
 		goto out;
 	}
 
-	eb = __alloc_dummy_extent_buffer(fs_info, 0, len);
+	eb = __alloc_dummy_extent_buffer(fs_info, 0, nodesize);
 	if (!eb) {
 		test_std_err(TEST_ALLOC_ROOT);
 		ret = -ENOMEM;
 		goto out;
 	}
 
-	ret = __test_eb_bitmaps(bitmap, eb, len);
+	ret = __test_eb_bitmaps(bitmap, eb, nodesize);
 	if (ret)
 		goto out;
 
-	/* Do it over again with an extent buffer which isn't page-aligned. */
 	free_extent_buffer(eb);
-	eb = __alloc_dummy_extent_buffer(fs_info, nodesize / 2, len);
+
+	/*
+	 * Test again for case where the tree block is sectorsize aligned but
+	 * not nodesize aligned.
+	 */
+	eb = __alloc_dummy_extent_buffer(fs_info, sectorsize, nodesize);
 	if (!eb) {
 		test_std_err(TEST_ALLOC_ROOT);
 		ret = -ENOMEM;
 		goto out;
 	}
 
-	ret = __test_eb_bitmaps(bitmap, eb, len);
+	ret = __test_eb_bitmaps(bitmap, eb, nodesize);
 out:
 	free_extent_buffer(eb);
 	kfree(bitmap);
@@ -538,6 +585,8 @@ static int test_find_first_clear_extent_bit(void)
 
 	ret = 0;
 out:
+	if (ret)
+		dump_extent_io_tree(&tree);
 	clear_extent_bits(&tree, 0, (u64)-1, CHUNK_TRIMMED | CHUNK_ALLOCATED);
 
 	return ret;
diff --git a/fs/btrfs/tests/extent-map-tests.c b/fs/btrfs/tests/extent-map-tests.c
index 57379e96ccc9..c5b3a631bf4f 100644
--- a/fs/btrfs/tests/extent-map-tests.c
+++ b/fs/btrfs/tests/extent-map-tests.c
@@ -15,6 +15,7 @@ static void free_extent_map_tree(struct extent_map_tree *em_tree)
 	struct extent_map *em;
 	struct rb_node *node;
 
+	write_lock(&em_tree->lock);
 	while (!RB_EMPTY_ROOT(&em_tree->map.rb_root)) {
 		node = rb_first_cached(&em_tree->map);
 		em = rb_entry(node, struct extent_map, rb_node);
@@ -32,6 +33,7 @@ static void free_extent_map_tree(struct extent_map_tree *em_tree)
 #endif
 		free_extent_map(em);
 	}
+	write_unlock(&em_tree->lock);
 }
 
 /*
@@ -507,7 +509,7 @@ static int test_rmap_block(struct btrfs_fs_info *fs_info,
 		goto out_free;
 	}
 
-	ret = btrfs_rmap_block(fs_info, em->start, btrfs_sb_offset(1),
+	ret = btrfs_rmap_block(fs_info, em->start, NULL, btrfs_sb_offset(1),
 			       &logical, &out_ndaddrs, &out_stripe_len);
 	if (ret || (out_ndaddrs == 0 && test->expected_mapped_addr)) {
 		test_err("didn't rmap anything but expected %d",
@@ -557,7 +559,7 @@ int btrfs_test_extent_map(void)
 		{
 			/*
 			 * Test a chunk with 2 data stripes one of which
-			 * interesects the physical address of the super block
+			 * intersects the physical address of the super block
 			 * is correctly recognised.
 			 */
 			.raid_type = BTRFS_BLOCK_GROUP_RAID1,
diff --git a/fs/btrfs/tests/free-space-tests.c b/fs/btrfs/tests/free-space-tests.c
index aebdf23f0cdd..ebf68fcd2149 100644
--- a/fs/btrfs/tests/free-space-tests.c
+++ b/fs/btrfs/tests/free-space-tests.c
@@ -82,7 +82,7 @@ static int test_extents(struct btrfs_block_group *cache)
 	}
 
 	/* Cleanup */
-	__btrfs_remove_free_space_cache(cache->free_space_ctl);
+	btrfs_remove_free_space_cache(cache);
 
 	return 0;
 }
@@ -149,7 +149,7 @@ static int test_bitmaps(struct btrfs_block_group *cache, u32 sectorsize)
 		return -1;
 	}
 
-	__btrfs_remove_free_space_cache(cache->free_space_ctl);
+	btrfs_remove_free_space_cache(cache);
 
 	return 0;
 }
@@ -230,7 +230,7 @@ static int test_bitmaps_and_extents(struct btrfs_block_group *cache,
 		return -1;
 	}
 
-	__btrfs_remove_free_space_cache(cache->free_space_ctl);
+	btrfs_remove_free_space_cache(cache);
 
 	/* Now with the extent entry offset into the bitmap */
 	ret = test_add_free_space_entry(cache, SZ_4M, SZ_4M, 1);
@@ -266,7 +266,7 @@ static int test_bitmaps_and_extents(struct btrfs_block_group *cache,
 	 *      [ bitmap ]
 	 *        [ del ]
 	 */
-	__btrfs_remove_free_space_cache(cache->free_space_ctl);
+	btrfs_remove_free_space_cache(cache);
 	ret = test_add_free_space_entry(cache, bitmap_offset + SZ_4M, SZ_4M, 1);
 	if (ret) {
 		test_err("couldn't add bitmap %d", ret);
@@ -291,7 +291,7 @@ static int test_bitmaps_and_extents(struct btrfs_block_group *cache,
 		return -1;
 	}
 
-	__btrfs_remove_free_space_cache(cache->free_space_ctl);
+	btrfs_remove_free_space_cache(cache);
 
 	/*
 	 * This blew up before, we have part of the free space in a bitmap and
@@ -317,7 +317,7 @@ static int test_bitmaps_and_extents(struct btrfs_block_group *cache,
 		return ret;
 	}
 
-	__btrfs_remove_free_space_cache(cache->free_space_ctl);
+	btrfs_remove_free_space_cache(cache);
 	return 0;
 }
 
@@ -399,7 +399,6 @@ test_steal_space_from_bitmap_to_extent(struct btrfs_block_group *cache,
 	u64 offset;
 	u64 max_extent_size;
 	const struct btrfs_free_space_op test_free_space_ops = {
-		.recalc_thresholds = cache->free_space_ctl->op->recalc_thresholds,
 		.use_bitmap = test_use_bitmap,
 	};
 	const struct btrfs_free_space_op *orig_free_space_ops;
@@ -630,7 +629,7 @@ test_steal_space_from_bitmap_to_extent(struct btrfs_block_group *cache,
 	if (ret)
 		return ret;
 
-	__btrfs_remove_free_space_cache(cache->free_space_ctl);
+	btrfs_remove_free_space_cache(cache);
 
 	/*
 	 * Now test a similar scenario, but where our extent entry is located
@@ -820,11 +819,189 @@ test_steal_space_from_bitmap_to_extent(struct btrfs_block_group *cache,
 		return ret;
 
 	cache->free_space_ctl->op = orig_free_space_ops;
-	__btrfs_remove_free_space_cache(cache->free_space_ctl);
+	btrfs_remove_free_space_cache(cache);
 
 	return 0;
 }
 
+static bool bytes_index_use_bitmap(struct btrfs_free_space_ctl *ctl,
+				   struct btrfs_free_space *info)
+{
+	return true;
+}
+
+static int test_bytes_index(struct btrfs_block_group *cache, u32 sectorsize)
+{
+	const struct btrfs_free_space_op test_free_space_ops = {
+		.use_bitmap = bytes_index_use_bitmap,
+	};
+	const struct btrfs_free_space_op *orig_free_space_ops;
+	struct btrfs_free_space_ctl *ctl = cache->free_space_ctl;
+	struct btrfs_free_space *entry;
+	struct rb_node *node;
+	u64 offset, max_extent_size, bytes;
+	int ret, i;
+
+	test_msg("running bytes index tests");
+
+	/* First just validate that it does everything in order. */
+	offset = 0;
+	for (i = 0; i < 10; i++) {
+		bytes = (i + 1) * SZ_1M;
+		ret = test_add_free_space_entry(cache, offset, bytes, 0);
+		if (ret) {
+			test_err("couldn't add extent entry %d\n", ret);
+			return ret;
+		}
+		offset += bytes + sectorsize;
+	}
+
+	for (node = rb_first_cached(&ctl->free_space_bytes), i = 9; node;
+	     node = rb_next(node), i--) {
+		entry = rb_entry(node, struct btrfs_free_space, bytes_index);
+		bytes = (i + 1) * SZ_1M;
+		if (entry->bytes != bytes) {
+			test_err("invalid bytes index order, found %llu expected %llu",
+				 entry->bytes, bytes);
+			return -EINVAL;
+		}
+	}
+
+	/* Now validate bitmaps do the correct thing. */
+	btrfs_remove_free_space_cache(cache);
+	for (i = 0; i < 2; i++) {
+		offset = i * BITS_PER_BITMAP * sectorsize;
+		bytes = (i + 1) * SZ_1M;
+		ret = test_add_free_space_entry(cache, offset, bytes, 1);
+		if (ret) {
+			test_err("couldn't add bitmap entry");
+			return ret;
+		}
+	}
+
+	for (node = rb_first_cached(&ctl->free_space_bytes), i = 1; node;
+	     node = rb_next(node), i--) {
+		entry = rb_entry(node, struct btrfs_free_space, bytes_index);
+		bytes = (i + 1) * SZ_1M;
+		if (entry->bytes != bytes) {
+			test_err("invalid bytes index order, found %llu expected %llu",
+				 entry->bytes, bytes);
+			return -EINVAL;
+		}
+	}
+
+	/* Now validate bitmaps with different ->max_extent_size. */
+	btrfs_remove_free_space_cache(cache);
+	orig_free_space_ops = cache->free_space_ctl->op;
+	cache->free_space_ctl->op = &test_free_space_ops;
+
+	ret = test_add_free_space_entry(cache, 0, sectorsize, 1);
+	if (ret) {
+		test_err("couldn't add bitmap entry");
+		return ret;
+	}
+
+	offset = BITS_PER_BITMAP * sectorsize;
+	ret = test_add_free_space_entry(cache, offset, sectorsize, 1);
+	if (ret) {
+		test_err("couldn't add bitmap_entry");
+		return ret;
+	}
+
+	/*
+	 * Now set a bunch of sectorsize extents in the first entry so it's
+	 * ->bytes is large.
+	 */
+	for (i = 2; i < 20; i += 2) {
+		offset = sectorsize * i;
+		ret = btrfs_add_free_space(cache, offset, sectorsize);
+		if (ret) {
+			test_err("error populating sparse bitmap %d", ret);
+			return ret;
+		}
+	}
+
+	/*
+	 * Now set a contiguous extent in the second bitmap so its
+	 * ->max_extent_size is larger than the first bitmaps.
+	 */
+	offset = (BITS_PER_BITMAP * sectorsize) + sectorsize;
+	ret = btrfs_add_free_space(cache, offset, sectorsize);
+	if (ret) {
+		test_err("error adding contiguous extent %d", ret);
+		return ret;
+	}
+
+	/*
+	 * Since we don't set ->max_extent_size unless we search everything
+	 * should be indexed on bytes.
+	 */
+	entry = rb_entry(rb_first_cached(&ctl->free_space_bytes),
+			 struct btrfs_free_space, bytes_index);
+	if (entry->bytes != (10 * sectorsize)) {
+		test_err("error, wrong entry in the first slot in bytes_index");
+		return -EINVAL;
+	}
+
+	max_extent_size = 0;
+	offset = btrfs_find_space_for_alloc(cache, cache->start, sectorsize * 3,
+					    0, &max_extent_size);
+	if (offset != 0) {
+		test_err("found space to alloc even though we don't have enough space");
+		return -EINVAL;
+	}
+
+	if (max_extent_size != (2 * sectorsize)) {
+		test_err("got the wrong max_extent size %llu expected %llu",
+			 max_extent_size, (unsigned long long)(2 * sectorsize));
+		return -EINVAL;
+	}
+
+	/*
+	 * The search should have re-arranged the bytes index to use the
+	 * ->max_extent_size, validate it's now what we expect it to be.
+	 */
+	entry = rb_entry(rb_first_cached(&ctl->free_space_bytes),
+			 struct btrfs_free_space, bytes_index);
+	if (entry->bytes != (2 * sectorsize)) {
+		test_err("error, the bytes index wasn't recalculated properly");
+		return -EINVAL;
+	}
+
+	/* Add another sectorsize to re-arrange the tree back to ->bytes. */
+	offset = (BITS_PER_BITMAP * sectorsize) - sectorsize;
+	ret = btrfs_add_free_space(cache, offset, sectorsize);
+	if (ret) {
+		test_err("error adding extent to the sparse entry %d", ret);
+		return ret;
+	}
+
+	entry = rb_entry(rb_first_cached(&ctl->free_space_bytes),
+			 struct btrfs_free_space, bytes_index);
+	if (entry->bytes != (11 * sectorsize)) {
+		test_err("error, wrong entry in the first slot in bytes_index");
+		return -EINVAL;
+	}
+
+	/*
+	 * Now make sure we find our correct entry after searching that will
+	 * result in a re-arranging of the tree.
+	 */
+	max_extent_size = 0;
+	offset = btrfs_find_space_for_alloc(cache, cache->start, sectorsize * 2,
+					    0, &max_extent_size);
+	if (offset != (BITS_PER_BITMAP * sectorsize)) {
+		test_err("error, found %llu instead of %llu for our alloc",
+			 offset,
+			 (unsigned long long)(BITS_PER_BITMAP * sectorsize));
+		return -EINVAL;
+	}
+
+	cache->free_space_ctl->op = orig_free_space_ops;
+	btrfs_remove_free_space_cache(cache);
+	return 0;
+}
+
 int btrfs_test_free_space_cache(u32 sectorsize, u32 nodesize)
 {
 	struct btrfs_fs_info *fs_info;
@@ -859,7 +1036,10 @@ int btrfs_test_free_space_cache(u32 sectorsize, u32 nodesize)
 		goto out;
 	}
 
-	root->fs_info->extent_root = root;
+	root->root_key.objectid = BTRFS_EXTENT_TREE_OBJECTID;
+	root->root_key.type = BTRFS_ROOT_ITEM_KEY;
+	root->root_key.offset = 0;
+	btrfs_global_root_insert(root);
 
 	ret = test_extents(cache);
 	if (ret)
@@ -872,6 +1052,9 @@ int btrfs_test_free_space_cache(u32 sectorsize, u32 nodesize)
 		goto out;
 
 	ret = test_steal_space_from_bitmap_to_extent(cache, sectorsize);
+	if (ret)
+		goto out;
+	ret = test_bytes_index(cache, sectorsize);
 out:
 	btrfs_free_dummy_block_group(cache);
 	btrfs_free_dummy_root(root);
diff --git a/fs/btrfs/tests/free-space-tree-tests.c b/fs/btrfs/tests/free-space-tree-tests.c
index 914eea5ba6a7..13734ed43bfc 100644
--- a/fs/btrfs/tests/free-space-tree-tests.c
+++ b/fs/btrfs/tests/free-space-tree-tests.c
@@ -60,8 +60,6 @@ static int __check_free_space_extents(struct btrfs_trans_handle *trans,
 				if (prev_bit == 0 && bit == 1) {
 					extent_start = offset;
 				} else if (prev_bit == 1 && bit == 0) {
-					if (i >= num_extents)
-						goto invalid;
 					if (i >= num_extents ||
 					    extent_start != extents[i].start ||
 					    offset - extent_start != extents[i].length)
@@ -448,7 +446,10 @@ static int run_test(test_func_t test_func, int bitmaps, u32 sectorsize,
 
 	btrfs_set_super_compat_ro_flags(root->fs_info->super_copy,
 					BTRFS_FEATURE_COMPAT_RO_FREE_SPACE_TREE);
-	root->fs_info->free_space_root = root;
+	root->root_key.objectid = BTRFS_FREE_SPACE_TREE_OBJECTID;
+	root->root_key.type = BTRFS_ROOT_ITEM_KEY;
+	root->root_key.offset = 0;
+	btrfs_global_root_insert(root);
 	root->fs_info->tree_root = root;
 
 	root->node = alloc_test_extent_buffer(root->fs_info, nodesize);
diff --git a/fs/btrfs/tests/inode-tests.c b/fs/btrfs/tests/inode-tests.c
index 24a8c714f56c..625f7d398368 100644
--- a/fs/btrfs/tests/inode-tests.c
+++ b/fs/btrfs/tests/inode-tests.c
@@ -33,8 +33,7 @@ static void insert_extent(struct btrfs_root *root, u64 start, u64 len,
 	key.type = BTRFS_EXTENT_DATA_KEY;
 	key.offset = start;
 
-	setup_items_for_insert(root, &path, &key, &value_len, value_len,
-			       value_len + sizeof(struct btrfs_item), 1);
+	btrfs_setup_item_for_insert(root, &path, &key, value_len);
 	fi = btrfs_item_ptr(leaf, slot, struct btrfs_file_extent_item);
 	btrfs_set_file_extent_generation(leaf, fi, 1);
 	btrfs_set_file_extent_type(leaf, fi, type);
@@ -64,8 +63,7 @@ static void insert_inode_item_key(struct btrfs_root *root)
 	key.type = BTRFS_INODE_ITEM_KEY;
 	key.offset = 0;
 
-	setup_items_for_insert(root, &path, &key, &value_len, value_len,
-			       value_len + sizeof(struct btrfs_item), 1);
+	btrfs_setup_item_for_insert(root, &path, &key, value_len);
 }
 
 /*
@@ -234,11 +232,6 @@ static noinline int test_btrfs_get_extent(u32 sectorsize, u32 nodesize)
 		return ret;
 	}
 
-	inode->i_mode = S_IFREG;
-	BTRFS_I(inode)->location.type = BTRFS_INODE_ITEM_KEY;
-	BTRFS_I(inode)->location.objectid = BTRFS_FIRST_FREE_OBJECTID;
-	BTRFS_I(inode)->location.offset = 0;
-
 	fs_info = btrfs_alloc_dummy_fs_info(nodesize, sectorsize);
 	if (!fs_info) {
 		test_std_err(TEST_ALLOC_FS_INFO);
@@ -274,7 +267,7 @@ static noinline int test_btrfs_get_extent(u32 sectorsize, u32 nodesize)
 		goto out;
 	}
 	free_extent_map(em);
-	btrfs_drop_extent_cache(BTRFS_I(inode), 0, (u64)-1, 0);
+	btrfs_drop_extent_map_range(BTRFS_I(inode), 0, (u64)-1, false);
 
 	/*
 	 * All of the magic numbers are based on the mapping setup in
@@ -837,10 +830,6 @@ static int test_hole_first(u32 sectorsize, u32 nodesize)
 		return ret;
 	}
 
-	BTRFS_I(inode)->location.type = BTRFS_INODE_ITEM_KEY;
-	BTRFS_I(inode)->location.objectid = BTRFS_FIRST_FREE_OBJECTID;
-	BTRFS_I(inode)->location.offset = 0;
-
 	fs_info = btrfs_alloc_dummy_fs_info(nodesize, sectorsize);
 	if (!fs_info) {
 		test_std_err(TEST_ALLOC_FS_INFO);
@@ -951,11 +940,10 @@ static int test_extent_accounting(u32 sectorsize, u32 nodesize)
 	}
 
 	BTRFS_I(inode)->root = root;
-	btrfs_test_inode_set_ops(inode);
 
 	/* [BTRFS_MAX_EXTENT_SIZE] */
-	ret = btrfs_set_extent_delalloc(inode, 0, BTRFS_MAX_EXTENT_SIZE - 1, 0,
-					NULL);
+	ret = btrfs_set_extent_delalloc(BTRFS_I(inode), 0,
+					BTRFS_MAX_EXTENT_SIZE - 1, 0, NULL);
 	if (ret) {
 		test_err("btrfs_set_extent_delalloc returned %d", ret);
 		goto out;
@@ -968,7 +956,7 @@ static int test_extent_accounting(u32 sectorsize, u32 nodesize)
 	}
 
 	/* [BTRFS_MAX_EXTENT_SIZE][sectorsize] */
-	ret = btrfs_set_extent_delalloc(inode, BTRFS_MAX_EXTENT_SIZE,
+	ret = btrfs_set_extent_delalloc(BTRFS_I(inode), BTRFS_MAX_EXTENT_SIZE,
 					BTRFS_MAX_EXTENT_SIZE + sectorsize - 1,
 					0, NULL);
 	if (ret) {
@@ -986,7 +974,8 @@ static int test_extent_accounting(u32 sectorsize, u32 nodesize)
 	ret = clear_extent_bit(&BTRFS_I(inode)->io_tree,
 			       BTRFS_MAX_EXTENT_SIZE >> 1,
 			       (BTRFS_MAX_EXTENT_SIZE >> 1) + sectorsize - 1,
-			       EXTENT_DELALLOC | EXTENT_UPTODATE, 0, 0, NULL);
+			       EXTENT_DELALLOC | EXTENT_DELALLOC_NEW |
+			       EXTENT_UPTODATE, NULL);
 	if (ret) {
 		test_err("clear_extent_bit returned %d", ret);
 		goto out;
@@ -999,7 +988,7 @@ static int test_extent_accounting(u32 sectorsize, u32 nodesize)
 	}
 
 	/* [BTRFS_MAX_EXTENT_SIZE][sectorsize] */
-	ret = btrfs_set_extent_delalloc(inode, BTRFS_MAX_EXTENT_SIZE >> 1,
+	ret = btrfs_set_extent_delalloc(BTRFS_I(inode), BTRFS_MAX_EXTENT_SIZE >> 1,
 					(BTRFS_MAX_EXTENT_SIZE >> 1)
 					+ sectorsize - 1,
 					0, NULL);
@@ -1017,7 +1006,7 @@ static int test_extent_accounting(u32 sectorsize, u32 nodesize)
 	/*
 	 * [BTRFS_MAX_EXTENT_SIZE+sectorsize][sectorsize HOLE][BTRFS_MAX_EXTENT_SIZE+sectorsize]
 	 */
-	ret = btrfs_set_extent_delalloc(inode,
+	ret = btrfs_set_extent_delalloc(BTRFS_I(inode),
 			BTRFS_MAX_EXTENT_SIZE + 2 * sectorsize,
 			(BTRFS_MAX_EXTENT_SIZE << 1) + 3 * sectorsize - 1,
 			0, NULL);
@@ -1035,7 +1024,7 @@ static int test_extent_accounting(u32 sectorsize, u32 nodesize)
 	/*
 	* [BTRFS_MAX_EXTENT_SIZE+sectorsize][sectorsize][BTRFS_MAX_EXTENT_SIZE+sectorsize]
 	*/
-	ret = btrfs_set_extent_delalloc(inode,
+	ret = btrfs_set_extent_delalloc(BTRFS_I(inode),
 			BTRFS_MAX_EXTENT_SIZE + sectorsize,
 			BTRFS_MAX_EXTENT_SIZE + 2 * sectorsize - 1, 0, NULL);
 	if (ret) {
@@ -1053,7 +1042,8 @@ static int test_extent_accounting(u32 sectorsize, u32 nodesize)
 	ret = clear_extent_bit(&BTRFS_I(inode)->io_tree,
 			       BTRFS_MAX_EXTENT_SIZE + sectorsize,
 			       BTRFS_MAX_EXTENT_SIZE + 2 * sectorsize - 1,
-			       EXTENT_DELALLOC | EXTENT_UPTODATE, 0, 0, NULL);
+			       EXTENT_DELALLOC | EXTENT_DELALLOC_NEW |
+			       EXTENT_UPTODATE, NULL);
 	if (ret) {
 		test_err("clear_extent_bit returned %d", ret);
 		goto out;
@@ -1069,7 +1059,7 @@ static int test_extent_accounting(u32 sectorsize, u32 nodesize)
 	 * Refill the hole again just for good measure, because I thought it
 	 * might fail and I'd rather satisfy my paranoia at this point.
 	 */
-	ret = btrfs_set_extent_delalloc(inode,
+	ret = btrfs_set_extent_delalloc(BTRFS_I(inode),
 			BTRFS_MAX_EXTENT_SIZE + sectorsize,
 			BTRFS_MAX_EXTENT_SIZE + 2 * sectorsize - 1, 0, NULL);
 	if (ret) {
@@ -1085,7 +1075,8 @@ static int test_extent_accounting(u32 sectorsize, u32 nodesize)
 
 	/* Empty */
 	ret = clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, (u64)-1,
-			       EXTENT_DELALLOC | EXTENT_UPTODATE, 0, 0, NULL);
+			       EXTENT_DELALLOC | EXTENT_DELALLOC_NEW |
+			       EXTENT_UPTODATE, NULL);
 	if (ret) {
 		test_err("clear_extent_bit returned %d", ret);
 		goto out;
@@ -1100,7 +1091,8 @@ static int test_extent_accounting(u32 sectorsize, u32 nodesize)
 out:
 	if (ret)
 		clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, (u64)-1,
-				 EXTENT_DELALLOC | EXTENT_UPTODATE, 0, 0, NULL);
+				 EXTENT_DELALLOC | EXTENT_DELALLOC_NEW |
+				 EXTENT_UPTODATE, NULL);
 	iput(inode);
 	btrfs_free_dummy_root(root);
 	btrfs_free_dummy_fs_info(fs_info);
diff --git a/fs/btrfs/tests/qgroup-tests.c b/fs/btrfs/tests/qgroup-tests.c
index ac035a6fa003..63676ea19f29 100644
--- a/fs/btrfs/tests/qgroup-tests.c
+++ b/fs/btrfs/tests/qgroup-tests.c
@@ -36,7 +36,6 @@ static int insert_normal_tree_ref(struct btrfs_root *root, u64 bytenr,
 		return -ENOMEM;
 	}
 
-	path->leave_spinning = 1;
 	ret = btrfs_insert_empty_item(&trans, root, path, &ins, size);
 	if (ret) {
 		test_err("couldn't insert ref %d", ret);
@@ -86,7 +85,6 @@ static int add_tree_ref(struct btrfs_root *root, u64 bytenr, u64 num_bytes,
 		return -ENOMEM;
 	}
 
-	path->leave_spinning = 1;
 	ret = btrfs_search_slot(&trans, root, &key, path, 0, 1);
 	if (ret) {
 		test_err("couldn't find extent ref");
@@ -135,7 +133,6 @@ static int remove_extent_item(struct btrfs_root *root, u64 bytenr,
 		test_std_err(TEST_ALLOC_ROOT);
 		return -ENOMEM;
 	}
-	path->leave_spinning = 1;
 
 	ret = btrfs_search_slot(&trans, root, &key, path, -1, 1);
 	if (ret) {
@@ -170,7 +167,6 @@ static int remove_extent_ref(struct btrfs_root *root, u64 bytenr,
 		return -ENOMEM;
 	}
 
-	path->leave_spinning = 1;
 	ret = btrfs_search_slot(&trans, root, &key, path, 0, 1);
 	if (ret) {
 		test_err("couldn't find extent ref");
@@ -227,24 +223,22 @@ static int test_no_shared_qgroup(struct btrfs_root *root,
 	 * we can only call btrfs_qgroup_account_extent() directly to test
 	 * quota.
 	 */
-	ret = btrfs_find_all_roots(&trans, fs_info, nodesize, 0, &old_roots,
-			false);
+	ret = btrfs_find_all_roots(&trans, fs_info, nodesize, 0, &old_roots, false);
 	if (ret) {
-		ulist_free(old_roots);
 		test_err("couldn't find old roots: %d", ret);
 		return ret;
 	}
 
 	ret = insert_normal_tree_ref(root, nodesize, nodesize, 0,
 				BTRFS_FS_TREE_OBJECTID);
-	if (ret)
+	if (ret) {
+		ulist_free(old_roots);
 		return ret;
+	}
 
-	ret = btrfs_find_all_roots(&trans, fs_info, nodesize, 0, &new_roots,
-			false);
+	ret = btrfs_find_all_roots(&trans, fs_info, nodesize, 0, &new_roots, false);
 	if (ret) {
 		ulist_free(old_roots);
-		ulist_free(new_roots);
 		test_err("couldn't find old roots: %d", ret);
 		return ret;
 	}
@@ -256,31 +250,31 @@ static int test_no_shared_qgroup(struct btrfs_root *root,
 		return ret;
 	}
 
+	/* btrfs_qgroup_account_extent() always frees the ulists passed to it. */
+	old_roots = NULL;
+	new_roots = NULL;
+
 	if (btrfs_verify_qgroup_counts(fs_info, BTRFS_FS_TREE_OBJECTID,
 				nodesize, nodesize)) {
 		test_err("qgroup counts didn't match expected values");
 		return -EINVAL;
 	}
-	old_roots = NULL;
-	new_roots = NULL;
 
-	ret = btrfs_find_all_roots(&trans, fs_info, nodesize, 0, &old_roots,
-			false);
+	ret = btrfs_find_all_roots(&trans, fs_info, nodesize, 0, &old_roots, false);
 	if (ret) {
-		ulist_free(old_roots);
 		test_err("couldn't find old roots: %d", ret);
 		return ret;
 	}
 
 	ret = remove_extent_item(root, nodesize, nodesize);
-	if (ret)
+	if (ret) {
+		ulist_free(old_roots);
 		return -EINVAL;
+	}
 
-	ret = btrfs_find_all_roots(&trans, fs_info, nodesize, 0, &new_roots,
-			false);
+	ret = btrfs_find_all_roots(&trans, fs_info, nodesize, 0, &new_roots, false);
 	if (ret) {
 		ulist_free(old_roots);
-		ulist_free(new_roots);
 		test_err("couldn't find old roots: %d", ret);
 		return ret;
 	}
@@ -328,24 +322,22 @@ static int test_multiple_refs(struct btrfs_root *root,
 		return ret;
 	}
 
-	ret = btrfs_find_all_roots(&trans, fs_info, nodesize, 0, &old_roots,
-			false);
+	ret = btrfs_find_all_roots(&trans, fs_info, nodesize, 0, &old_roots, false);
 	if (ret) {
-		ulist_free(old_roots);
 		test_err("couldn't find old roots: %d", ret);
 		return ret;
 	}
 
 	ret = insert_normal_tree_ref(root, nodesize, nodesize, 0,
 				BTRFS_FS_TREE_OBJECTID);
-	if (ret)
+	if (ret) {
+		ulist_free(old_roots);
 		return ret;
+	}
 
-	ret = btrfs_find_all_roots(&trans, fs_info, nodesize, 0, &new_roots,
-			false);
+	ret = btrfs_find_all_roots(&trans, fs_info, nodesize, 0, &new_roots, false);
 	if (ret) {
 		ulist_free(old_roots);
-		ulist_free(new_roots);
 		test_err("couldn't find old roots: %d", ret);
 		return ret;
 	}
@@ -363,24 +355,22 @@ static int test_multiple_refs(struct btrfs_root *root,
 		return -EINVAL;
 	}
 
-	ret = btrfs_find_all_roots(&trans, fs_info, nodesize, 0, &old_roots,
-			false);
+	ret = btrfs_find_all_roots(&trans, fs_info, nodesize, 0, &old_roots, false);
 	if (ret) {
-		ulist_free(old_roots);
 		test_err("couldn't find old roots: %d", ret);
 		return ret;
 	}
 
 	ret = add_tree_ref(root, nodesize, nodesize, 0,
 			BTRFS_FIRST_FREE_OBJECTID);
-	if (ret)
+	if (ret) {
+		ulist_free(old_roots);
 		return ret;
+	}
 
-	ret = btrfs_find_all_roots(&trans, fs_info, nodesize, 0, &new_roots,
-			false);
+	ret = btrfs_find_all_roots(&trans, fs_info, nodesize, 0, &new_roots, false);
 	if (ret) {
 		ulist_free(old_roots);
-		ulist_free(new_roots);
 		test_err("couldn't find old roots: %d", ret);
 		return ret;
 	}
@@ -404,24 +394,22 @@ static int test_multiple_refs(struct btrfs_root *root,
 		return -EINVAL;
 	}
 
-	ret = btrfs_find_all_roots(&trans, fs_info, nodesize, 0, &old_roots,
-			false);
+	ret = btrfs_find_all_roots(&trans, fs_info, nodesize, 0, &old_roots, false);
 	if (ret) {
-		ulist_free(old_roots);
 		test_err("couldn't find old roots: %d", ret);
 		return ret;
 	}
 
 	ret = remove_extent_ref(root, nodesize, nodesize, 0,
 				BTRFS_FIRST_FREE_OBJECTID);
-	if (ret)
+	if (ret) {
+		ulist_free(old_roots);
 		return ret;
+	}
 
-	ret = btrfs_find_all_roots(&trans, fs_info, nodesize, 0, &new_roots,
-			false);
+	ret = btrfs_find_all_roots(&trans, fs_info, nodesize, 0, &new_roots, false);
 	if (ret) {
 		ulist_free(old_roots);
-		ulist_free(new_roots);
 		test_err("couldn't find old roots: %d", ret);
 		return ret;
 	}
@@ -469,7 +457,10 @@ int btrfs_test_qgroups(u32 sectorsize, u32 nodesize)
 	}
 
 	/* We are using this root as our extent root */
-	root->fs_info->extent_root = root;
+	root->root_key.objectid = BTRFS_EXTENT_TREE_OBJECTID;
+	root->root_key.type = BTRFS_ROOT_ITEM_KEY;
+	root->root_key.offset = 0;
+	btrfs_global_root_insert(root);
 
 	/*
 	 * Some of the paths we test assume we have a filled out fs_info, so we
@@ -507,6 +498,7 @@ int btrfs_test_qgroups(u32 sectorsize, u32 nodesize)
 		test_err("couldn't insert fs root %d", ret);
 		goto out;
 	}
+	btrfs_put_root(tmp_root);
 
 	tmp_root = btrfs_alloc_dummy_root(fs_info);
 	if (IS_ERR(tmp_root)) {
@@ -521,6 +513,7 @@ int btrfs_test_qgroups(u32 sectorsize, u32 nodesize)
 		test_err("couldn't insert fs root %d", ret);
 		goto out;
 	}
+	btrfs_put_root(tmp_root);
 
 	test_msg("running qgroup tests");
 	ret = test_no_shared_qgroup(root, sectorsize, nodesize);
diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index beb6c69cd1e5..d1f1da6820fb 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -10,17 +10,19 @@
 #include <linux/pagemap.h>
 #include <linux/blkdev.h>
 #include <linux/uuid.h>
+#include <linux/timekeeping.h>
 #include "misc.h"
 #include "ctree.h"
 #include "disk-io.h"
 #include "transaction.h"
 #include "locking.h"
 #include "tree-log.h"
-#include "inode-map.h"
 #include "volumes.h"
 #include "dev-replace.h"
 #include "qgroup.h"
 #include "block-group.h"
+#include "space-info.h"
+#include "zoned.h"
 
 #define BTRFS_ROOT_TRANS_TAG 0
 
@@ -107,6 +109,11 @@ static const unsigned int btrfs_blocked_trans_types[TRANS_STATE_MAX] = {
 					   __TRANS_JOIN |
 					   __TRANS_JOIN_NOLOCK |
 					   __TRANS_JOIN_NOSTART),
+	[TRANS_STATE_SUPER_COMMITTED]	= (__TRANS_START |
+					   __TRANS_ATTACH |
+					   __TRANS_JOIN |
+					   __TRANS_JOIN_NOLOCK |
+					   __TRANS_JOIN_NOSTART),
 	[TRANS_STATE_COMPLETED]		= (__TRANS_START |
 					   __TRANS_ATTACH |
 					   __TRANS_JOIN |
@@ -141,7 +148,7 @@ void btrfs_put_transaction(struct btrfs_transaction *transaction)
 						 struct btrfs_block_group,
 						 bg_list);
 			list_del_init(&cache->bg_list);
-			btrfs_put_block_group_trimming(cache);
+			btrfs_unfreeze_block_group(cache);
 			btrfs_put_block_group(cache);
 		}
 		WARN_ON(!list_empty(&transaction->dev_update_list));
@@ -155,14 +162,22 @@ static noinline void switch_commit_roots(struct btrfs_trans_handle *trans)
 	struct btrfs_fs_info *fs_info = trans->fs_info;
 	struct btrfs_root *root, *tmp;
 
+	/*
+	 * At this point no one can be using this transaction to modify any tree
+	 * and no one can start another transaction to modify any tree either.
+	 */
+	ASSERT(cur_trans->state == TRANS_STATE_COMMIT_DOING);
+
 	down_write(&fs_info->commit_root_sem);
+
+	if (test_bit(BTRFS_FS_RELOC_RUNNING, &fs_info->flags))
+		fs_info->last_reloc_trans = trans->transid;
+
 	list_for_each_entry_safe(root, tmp, &cur_trans->switch_commits,
 				 dirty_list) {
 		list_del_init(&root->dirty_list);
 		free_extent_buffer(root->commit_root);
 		root->commit_root = btrfs_root_node(root);
-		if (is_fstree(root->root_key.objectid))
-			btrfs_unpin_free_ino(root);
 		extent_io_tree_release(&root->dirty_log_pages);
 		btrfs_qgroup_clean_swapped_blocks(root);
 	}
@@ -179,6 +194,7 @@ static noinline void switch_commit_roots(struct btrfs_trans_handle *trans)
 		spin_lock(&cur_trans->dropped_roots_lock);
 	}
 	spin_unlock(&cur_trans->dropped_roots_lock);
+
 	up_write(&fs_info->commit_root_sem);
 }
 
@@ -208,8 +224,11 @@ static inline int extwriter_counter_read(struct btrfs_transaction *trans)
 }
 
 /*
- * To be called after all the new block groups attached to the transaction
- * handle have been created (btrfs_create_pending_block_groups()).
+ * To be called after doing the chunk btree updates right after allocating a new
+ * chunk (after btrfs_chunk_alloc_add_chunk_item() is called), when removing a
+ * chunk after all chunk btree updates and after finishing the second phase of
+ * chunk allocation (btrfs_create_pending_block_groups()) in case some block
+ * group had its chunk item insertion delayed to the second phase.
  */
 void btrfs_trans_release_chunk_metadata(struct btrfs_trans_handle *trans)
 {
@@ -218,10 +237,8 @@ void btrfs_trans_release_chunk_metadata(struct btrfs_trans_handle *trans)
 	if (!trans->chunk_bytes_reserved)
 		return;
 
-	WARN_ON_ONCE(!list_empty(&trans->new_bgs));
-
 	btrfs_block_rsv_release(fs_info, &fs_info->chunk_block_rsv,
-				trans->chunk_bytes_reserved);
+				trans->chunk_bytes_reserved, NULL);
 	trans->chunk_bytes_reserved = 0;
 }
 
@@ -236,14 +253,14 @@ static noinline int join_transaction(struct btrfs_fs_info *fs_info,
 	spin_lock(&fs_info->trans_lock);
 loop:
 	/* The file system has been taken offline. No new transactions. */
-	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
+	if (BTRFS_FS_ERROR(fs_info)) {
 		spin_unlock(&fs_info->trans_lock);
 		return -EROFS;
 	}
 
 	cur_trans = fs_info->running_transaction;
 	if (cur_trans) {
-		if (cur_trans->aborted) {
+		if (TRANS_ABORTED(cur_trans)) {
 			spin_unlock(&fs_info->trans_lock);
 			return cur_trans->aborted;
 		}
@@ -255,6 +272,8 @@ loop:
 		atomic_inc(&cur_trans->num_writers);
 		extwriter_counter_inc(cur_trans, type);
 		spin_unlock(&fs_info->trans_lock);
+		btrfs_lockdep_acquire(fs_info, btrfs_trans_num_writers);
+		btrfs_lockdep_acquire(fs_info, btrfs_trans_num_extwriters);
 		return 0;
 	}
 	spin_unlock(&fs_info->trans_lock);
@@ -276,21 +295,30 @@ loop:
 	if (!cur_trans)
 		return -ENOMEM;
 
+	btrfs_lockdep_acquire(fs_info, btrfs_trans_num_writers);
+	btrfs_lockdep_acquire(fs_info, btrfs_trans_num_extwriters);
+
 	spin_lock(&fs_info->trans_lock);
 	if (fs_info->running_transaction) {
 		/*
 		 * someone started a transaction after we unlocked.  Make sure
 		 * to redo the checks above
 		 */
+		btrfs_lockdep_release(fs_info, btrfs_trans_num_extwriters);
+		btrfs_lockdep_release(fs_info, btrfs_trans_num_writers);
 		kfree(cur_trans);
 		goto loop;
-	} else if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
+	} else if (BTRFS_FS_ERROR(fs_info)) {
 		spin_unlock(&fs_info->trans_lock);
+		btrfs_lockdep_release(fs_info, btrfs_trans_num_extwriters);
+		btrfs_lockdep_release(fs_info, btrfs_trans_num_writers);
 		kfree(cur_trans);
 		return -EROFS;
 	}
 
 	cur_trans->fs_info = fs_info;
+	atomic_set(&cur_trans->pending_ordered, 0);
+	init_waitqueue_head(&cur_trans->pending_wait);
 	atomic_set(&cur_trans->num_writers, 1);
 	extwriter_counter_init(cur_trans, type);
 	init_waitqueue_head(&cur_trans->writer_wait);
@@ -333,9 +361,13 @@ loop:
 	spin_lock_init(&cur_trans->dirty_bgs_lock);
 	INIT_LIST_HEAD(&cur_trans->deleted_bgs);
 	spin_lock_init(&cur_trans->dropped_roots_lock);
+	INIT_LIST_HEAD(&cur_trans->releasing_ebs);
+	spin_lock_init(&cur_trans->releasing_ebs_lock);
 	list_add_tail(&cur_trans->list, &fs_info->trans_list);
 	extent_io_tree_init(fs_info, &cur_trans->dirty_pages,
-			IO_TREE_TRANS_DIRTY_PAGES, fs_info->btree_inode);
+			IO_TREE_TRANS_DIRTY_PAGES, NULL);
+	extent_io_tree_init(fs_info, &cur_trans->pinned_extents,
+			IO_TREE_FS_PINNED_EXTENTS, NULL);
 	fs_info->generation++;
 	cur_trans->transid = fs_info->generation;
 	fs_info->running_transaction = cur_trans;
@@ -346,20 +378,20 @@ loop:
 }
 
 /*
- * this does all the record keeping required to make sure that a reference
- * counted root is properly recorded in a given transaction.  This is required
- * to make sure the old root from before we joined the transaction is deleted
- * when the transaction commits
+ * This does all the record keeping required to make sure that a shareable root
+ * is properly recorded in a given transaction.  This is required to make sure
+ * the old root from before we joined the transaction is deleted when the
+ * transaction commits.
  */
 static int record_root_in_trans(struct btrfs_trans_handle *trans,
 			       struct btrfs_root *root,
 			       int force)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
+	int ret = 0;
 
-	if ((test_bit(BTRFS_ROOT_REF_COWS, &root->state) &&
+	if ((test_bit(BTRFS_ROOT_SHAREABLE, &root->state) &&
 	    root->last_trans < trans->transid) || force) {
-		WARN_ON(root == fs_info->extent_root);
 		WARN_ON(!force && root->commit_root != root->node);
 
 		/*
@@ -404,11 +436,11 @@ static int record_root_in_trans(struct btrfs_trans_handle *trans,
 		 * lock.  smp_wmb() makes sure that all the writes above are
 		 * done before we pop in the zero below
 		 */
-		btrfs_init_reloc_root(trans, root);
+		ret = btrfs_init_reloc_root(trans, root);
 		smp_mb__before_atomic();
 		clear_bit(BTRFS_ROOT_IN_TRANS_SETUP, &root->state);
 	}
-	return 0;
+	return ret;
 }
 
 
@@ -435,8 +467,9 @@ int btrfs_record_root_in_trans(struct btrfs_trans_handle *trans,
 			       struct btrfs_root *root)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
+	int ret;
 
-	if (!test_bit(BTRFS_ROOT_REF_COWS, &root->state))
+	if (!test_bit(BTRFS_ROOT_SHAREABLE, &root->state))
 		return 0;
 
 	/*
@@ -449,17 +482,17 @@ int btrfs_record_root_in_trans(struct btrfs_trans_handle *trans,
 		return 0;
 
 	mutex_lock(&fs_info->reloc_mutex);
-	record_root_in_trans(trans, root, 0);
+	ret = record_root_in_trans(trans, root, 0);
 	mutex_unlock(&fs_info->reloc_mutex);
 
-	return 0;
+	return ret;
 }
 
 static inline int is_transaction_blocked(struct btrfs_transaction *trans)
 {
 	return (trans->state >= TRANS_STATE_COMMIT_START &&
 		trans->state < TRANS_STATE_UNBLOCKED &&
-		!trans->aborted);
+		!TRANS_ABORTED(trans));
 }
 
 /* wait for commit against the current transaction to become unblocked
@@ -476,9 +509,10 @@ static void wait_current_trans(struct btrfs_fs_info *fs_info)
 		refcount_inc(&cur_trans->use_count);
 		spin_unlock(&fs_info->trans_lock);
 
+		btrfs_might_wait_for_state(fs_info, BTRFS_LOCKDEP_TRANS_UNBLOCKED);
 		wait_event(fs_info->transaction_wait,
 			   cur_trans->state >= TRANS_STATE_UNBLOCKED ||
-			   cur_trans->aborted);
+			   TRANS_ABORTED(cur_trans));
 		btrfs_put_transaction(cur_trans);
 	} else {
 		spin_unlock(&fs_info->trans_lock);
@@ -501,7 +535,7 @@ static inline bool need_reserve_reloc_root(struct btrfs_root *root)
 	struct btrfs_fs_info *fs_info = root->fs_info;
 
 	if (!fs_info->reloc_ctl ||
-	    !test_bit(BTRFS_ROOT_REF_COWS, &root->state) ||
+	    !test_bit(BTRFS_ROOT_SHAREABLE, &root->state) ||
 	    root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID ||
 	    root->reloc_root)
 		return false;
@@ -521,12 +555,10 @@ start_transaction(struct btrfs_root *root, unsigned int num_items,
 	u64 num_bytes = 0;
 	u64 qgroup_reserved = 0;
 	bool reloc_reserved = false;
+	bool do_chunk_alloc = false;
 	int ret;
 
-	/* Send isn't supposed to start transactions. */
-	ASSERT(current->journal_info != BTRFS_SEND_TRANS_STUB);
-
-	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state))
+	if (BTRFS_FS_ERROR(fs_info))
 		return ERR_PTR(-EROFS);
 
 	if (current->journal_info) {
@@ -561,7 +593,8 @@ start_transaction(struct btrfs_root *root, unsigned int num_items,
 		 * refill that amount for whatever is missing in the reserve.
 		 */
 		num_bytes = btrfs_calc_insert_metadata_size(fs_info, num_items);
-		if (delayed_refs_rsv->full == 0) {
+		if (flush == BTRFS_RESERVE_FLUSH_ALL &&
+		    btrfs_block_rsv_full(delayed_refs_rsv) == 0) {
 			delayed_refs_bytes = num_bytes;
 			num_bytes <<= 1;
 		}
@@ -574,7 +607,7 @@ start_transaction(struct btrfs_root *root, unsigned int num_items,
 			reloc_reserved = true;
 		}
 
-		ret = btrfs_block_rsv_add(root, rsv, num_bytes, flush);
+		ret = btrfs_block_rsv_add(fs_info, rsv, num_bytes, flush);
 		if (ret)
 			goto reserve_fail;
 		if (delayed_refs_bytes) {
@@ -582,8 +615,11 @@ start_transaction(struct btrfs_root *root, unsigned int num_items,
 							  delayed_refs_bytes);
 			num_bytes -= delayed_refs_bytes;
 		}
+
+		if (rsv->space_info->force_alloc)
+			do_chunk_alloc = true;
 	} else if (num_items == 0 && flush == BTRFS_RESERVE_FLUSH_ALL &&
-		   !delayed_refs_rsv->full) {
+		   !btrfs_block_rsv_full(delayed_refs_rsv)) {
 		/*
 		 * Some people call with btrfs_start_transaction(root, 0)
 		 * because they can be throttled, but have some other mechanism
@@ -635,12 +671,10 @@ again:
 
 	h->transid = cur_trans->transid;
 	h->transaction = cur_trans;
-	h->root = root;
 	refcount_set(&h->use_count, 1);
 	h->fs_info = root->fs_info;
 
 	h->type = type;
-	h->can_flush_pending_bgs = true;
 	INIT_LIST_HEAD(&h->new_bgs);
 
 	smp_mb();
@@ -660,10 +694,41 @@ again:
 	}
 
 got_it:
-	btrfs_record_root_in_trans(h, root);
-
 	if (!current->journal_info)
 		current->journal_info = h;
+
+	/*
+	 * If the space_info is marked ALLOC_FORCE then we'll get upgraded to
+	 * ALLOC_FORCE the first run through, and then we won't allocate for
+	 * anybody else who races in later.  We don't care about the return
+	 * value here.
+	 */
+	if (do_chunk_alloc && num_bytes) {
+		u64 flags = h->block_rsv->space_info->flags;
+
+		btrfs_chunk_alloc(h, btrfs_get_alloc_profile(fs_info, flags),
+				  CHUNK_ALLOC_NO_FORCE);
+	}
+
+	/*
+	 * btrfs_record_root_in_trans() needs to alloc new extents, and may
+	 * call btrfs_join_transaction() while we're also starting a
+	 * transaction.
+	 *
+	 * Thus it need to be called after current->journal_info initialized,
+	 * or we can deadlock.
+	 */
+	ret = btrfs_record_root_in_trans(h, root);
+	if (ret) {
+		/*
+		 * The transaction handle is fully initialized and linked with
+		 * other structures so it needs to be ended in case of errors,
+		 * not just freed.
+		 */
+		btrfs_end_transaction(h);
+		return ERR_PTR(ret);
+	}
+
 	return h;
 
 join_fail:
@@ -673,7 +738,7 @@ join_fail:
 alloc_fail:
 	if (num_bytes)
 		btrfs_block_rsv_release(fs_info, &fs_info->trans_block_rsv,
-					num_bytes);
+					num_bytes, NULL);
 reserve_fail:
 	btrfs_qgroup_free_meta_pertrans(root, qgroup_reserved);
 	return ERR_PTR(ret);
@@ -688,43 +753,10 @@ struct btrfs_trans_handle *btrfs_start_transaction(struct btrfs_root *root,
 
 struct btrfs_trans_handle *btrfs_start_transaction_fallback_global_rsv(
 					struct btrfs_root *root,
-					unsigned int num_items,
-					int min_factor)
+					unsigned int num_items)
 {
-	struct btrfs_fs_info *fs_info = root->fs_info;
-	struct btrfs_trans_handle *trans;
-	u64 num_bytes;
-	int ret;
-
-	/*
-	 * We have two callers: unlink and block group removal.  The
-	 * former should succeed even if we will temporarily exceed
-	 * quota and the latter operates on the extent root so
-	 * qgroup enforcement is ignored anyway.
-	 */
-	trans = start_transaction(root, num_items, TRANS_START,
-				  BTRFS_RESERVE_FLUSH_ALL, false);
-	if (!IS_ERR(trans) || PTR_ERR(trans) != -ENOSPC)
-		return trans;
-
-	trans = btrfs_start_transaction(root, 0);
-	if (IS_ERR(trans))
-		return trans;
-
-	num_bytes = btrfs_calc_insert_metadata_size(fs_info, num_items);
-	ret = btrfs_cond_migrate_bytes(fs_info, &fs_info->trans_block_rsv,
-				       num_bytes, min_factor);
-	if (ret) {
-		btrfs_end_transaction(trans);
-		return ERR_PTR(ret);
-	}
-
-	trans->block_rsv = &fs_info->trans_block_rsv;
-	trans->bytes_reserved = num_bytes;
-	trace_btrfs_space_reservation(fs_info, "transaction",
-				      trans->transid, num_bytes, 1);
-
-	return trans;
+	return start_transaction(root, num_items, TRANS_START,
+				 BTRFS_RESERVE_FLUSH_ALL_STEAL, false);
 }
 
 struct btrfs_trans_handle *btrfs_join_transaction(struct btrfs_root *root)
@@ -788,10 +820,50 @@ btrfs_attach_transaction_barrier(struct btrfs_root *root)
 	return trans;
 }
 
-/* wait for a transaction commit to be fully complete */
-static noinline void wait_for_commit(struct btrfs_transaction *commit)
+/* Wait for a transaction commit to reach at least the given state. */
+static noinline void wait_for_commit(struct btrfs_transaction *commit,
+				     const enum btrfs_trans_state min_state)
 {
-	wait_event(commit->commit_wait, commit->state == TRANS_STATE_COMPLETED);
+	struct btrfs_fs_info *fs_info = commit->fs_info;
+	u64 transid = commit->transid;
+	bool put = false;
+
+	/*
+	 * At the moment this function is called with min_state either being
+	 * TRANS_STATE_COMPLETED or TRANS_STATE_SUPER_COMMITTED.
+	 */
+	if (min_state == TRANS_STATE_COMPLETED)
+		btrfs_might_wait_for_state(fs_info, BTRFS_LOCKDEP_TRANS_COMPLETED);
+	else
+		btrfs_might_wait_for_state(fs_info, BTRFS_LOCKDEP_TRANS_SUPER_COMMITTED);
+
+	while (1) {
+		wait_event(commit->commit_wait, commit->state >= min_state);
+		if (put)
+			btrfs_put_transaction(commit);
+
+		if (min_state < TRANS_STATE_COMPLETED)
+			break;
+
+		/*
+		 * A transaction isn't really completed until all of the
+		 * previous transactions are completed, but with fsync we can
+		 * end up with SUPER_COMMITTED transactions before a COMPLETED
+		 * transaction. Wait for those.
+		 */
+
+		spin_lock(&fs_info->trans_lock);
+		commit = list_first_entry_or_null(&fs_info->trans_list,
+						  struct btrfs_transaction,
+						  list);
+		if (!commit || commit->transid > transid) {
+			spin_unlock(&fs_info->trans_lock);
+			break;
+		}
+		refcount_inc(&commit->use_count);
+		put = true;
+		spin_unlock(&fs_info->trans_lock);
+	}
 }
 
 int btrfs_wait_for_commit(struct btrfs_fs_info *fs_info, u64 transid)
@@ -846,7 +918,7 @@ int btrfs_wait_for_commit(struct btrfs_fs_info *fs_info, u64 transid)
 			goto out;  /* nothing committing|committed */
 	}
 
-	wait_for_commit(cur_trans);
+	wait_for_commit(cur_trans, TRANS_STATE_COMPLETED);
 	btrfs_put_transaction(cur_trans);
 out:
 	return ret;
@@ -857,24 +929,23 @@ void btrfs_throttle(struct btrfs_fs_info *fs_info)
 	wait_current_trans(fs_info);
 }
 
-static int should_end_transaction(struct btrfs_trans_handle *trans)
+static bool should_end_transaction(struct btrfs_trans_handle *trans)
 {
 	struct btrfs_fs_info *fs_info = trans->fs_info;
 
 	if (btrfs_check_space_for_delayed_refs(fs_info))
-		return 1;
+		return true;
 
 	return !!btrfs_block_rsv_check(&fs_info->global_block_rsv, 5);
 }
 
-int btrfs_should_end_transaction(struct btrfs_trans_handle *trans)
+bool btrfs_should_end_transaction(struct btrfs_trans_handle *trans)
 {
 	struct btrfs_transaction *cur_trans = trans->transaction;
 
-	smp_mb();
 	if (cur_trans->state >= TRANS_STATE_COMMIT_START ||
-	    cur_trans->delayed_refs.flushing)
-		return 1;
+	    test_bit(BTRFS_DELAYED_REFS_FLUSHING, &cur_trans->delayed_refs.flags))
+		return true;
 
 	return should_end_transaction(trans);
 }
@@ -896,7 +967,7 @@ static void btrfs_trans_release_metadata(struct btrfs_trans_handle *trans)
 	trace_btrfs_space_reservation(fs_info, "transaction",
 				      trans->transid, trans->bytes_reserved, 0);
 	btrfs_block_rsv_release(fs_info, trans->block_rsv,
-				trans->bytes_reserved);
+				trans->bytes_reserved, NULL);
 	trans->bytes_reserved = 0;
 }
 
@@ -929,6 +1000,10 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans,
 	extwriter_counter_dec(cur_trans, trans->type);
 
 	cond_wake_up(&cur_trans->writer_wait);
+
+	btrfs_lockdep_release(info, btrfs_trans_num_extwriters);
+	btrfs_lockdep_release(info, btrfs_trans_num_writers);
+
 	btrfs_put_transaction(cur_trans);
 
 	if (current->journal_info == trans)
@@ -937,10 +1012,12 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans,
 	if (throttle)
 		btrfs_run_delayed_iputs(info);
 
-	if (trans->aborted ||
-	    test_bit(BTRFS_FS_STATE_ERROR, &info->fs_state)) {
+	if (TRANS_ABORTED(trans) || BTRFS_FS_ERROR(info)) {
 		wake_up_process(info->transaction_kthread);
-		err = -EIO;
+		if (TRANS_ABORTED(trans))
+			err = trans->aborted;
+		else
+			err = -EROFS;
 	}
 
 	kmem_cache_free(btrfs_trans_handle_cachep, trans);
@@ -1039,7 +1116,7 @@ static int __btrfs_wait_marked_extents(struct btrfs_fs_info *fs_info,
 		 * it's safe to do it (through extent_io_tree_release()).
 		 */
 		err = clear_extent_bit(dirty_pages, start, end,
-				       EXTENT_NEED_WAIT, 0, 0, &cached_state);
+				       EXTENT_NEED_WAIT, &cached_state);
 		if (err == -ENOMEM)
 			err = 0;
 		if (!err)
@@ -1180,19 +1257,21 @@ static noinline int commit_cowonly_roots(struct btrfs_trans_handle *trans)
 	struct extent_buffer *eb;
 	int ret;
 
+	/*
+	 * At this point no one can be using this transaction to modify any tree
+	 * and no one can start another transaction to modify any tree either.
+	 */
+	ASSERT(trans->transaction->state == TRANS_STATE_COMMIT_DOING);
+
 	eb = btrfs_lock_root_node(fs_info->tree_root);
 	ret = btrfs_cow_block(trans, fs_info->tree_root, eb, NULL,
-			      0, &eb);
+			      0, &eb, BTRFS_NESTING_COW);
 	btrfs_tree_unlock(eb);
 	free_extent_buffer(eb);
 
 	if (ret)
 		return ret;
 
-	ret = btrfs_run_delayed_refs(trans, (unsigned long)-1);
-	if (ret)
-		return ret;
-
 	ret = btrfs_run_dev_stats(trans);
 	if (ret)
 		return ret;
@@ -1207,10 +1286,6 @@ static noinline int commit_cowonly_roots(struct btrfs_trans_handle *trans)
 	if (ret)
 		return ret;
 
-	/* run_qgroups might have added some more refs */
-	ret = btrfs_run_delayed_refs(trans, (unsigned long)-1);
-	if (ret)
-		return ret;
 again:
 	while (!list_empty(&fs_info->dirty_cowonly_roots)) {
 		struct btrfs_root *root;
@@ -1219,21 +1294,29 @@ again:
 		root = list_entry(next, struct btrfs_root, dirty_list);
 		clear_bit(BTRFS_ROOT_DIRTY, &root->state);
 
-		if (root != fs_info->extent_root)
-			list_add_tail(&root->dirty_list,
-				      &trans->transaction->switch_commits);
+		list_add_tail(&root->dirty_list,
+			      &trans->transaction->switch_commits);
 		ret = update_cowonly_root(trans, root);
 		if (ret)
 			return ret;
-		ret = btrfs_run_delayed_refs(trans, (unsigned long)-1);
-		if (ret)
-			return ret;
 	}
 
+	/* Now flush any delayed refs generated by updating all of the roots */
+	ret = btrfs_run_delayed_refs(trans, (unsigned long)-1);
+	if (ret)
+		return ret;
+
 	while (!list_empty(dirty_bgs) || !list_empty(io_bgs)) {
 		ret = btrfs_write_dirty_block_groups(trans);
 		if (ret)
 			return ret;
+
+		/*
+		 * We're writing the dirty block groups, which could generate
+		 * delayed refs, which could generate more dirty block groups,
+		 * so we want to keep this flushing in this loop to make sure
+		 * everything gets run.
+		 */
 		ret = btrfs_run_delayed_refs(trans, (unsigned long)-1);
 		if (ret)
 			return ret;
@@ -1242,9 +1325,6 @@ again:
 	if (!list_empty(&fs_info->dirty_cowonly_roots))
 		goto again;
 
-	list_add_tail(&fs_info->extent_root->dirty_list,
-		      &trans->transaction->switch_commits);
-
 	/* Update dev-replace pointer once everything is committed */
 	fs_info->dev_replace.committed_cursor_left =
 		fs_info->dev_replace.cursor_left_last_write_of_item;
@@ -1253,6 +1333,32 @@ again:
 }
 
 /*
+ * If we had a pending drop we need to see if there are any others left in our
+ * dead roots list, and if not clear our bit and wake any waiters.
+ */
+void btrfs_maybe_wake_unfinished_drop(struct btrfs_fs_info *fs_info)
+{
+	/*
+	 * We put the drop in progress roots at the front of the list, so if the
+	 * first entry doesn't have UNFINISHED_DROP set we can wake everybody
+	 * up.
+	 */
+	spin_lock(&fs_info->trans_lock);
+	if (!list_empty(&fs_info->dead_roots)) {
+		struct btrfs_root *root = list_first_entry(&fs_info->dead_roots,
+							   struct btrfs_root,
+							   root_list);
+		if (test_bit(BTRFS_ROOT_UNFINISHED_DROP, &root->state)) {
+			spin_unlock(&fs_info->trans_lock);
+			return;
+		}
+	}
+	spin_unlock(&fs_info->trans_lock);
+
+	btrfs_wake_unfinished_drop(fs_info);
+}
+
+/*
  * dead roots are old snapshots that need to be deleted.  This allocates
  * a dirty root struct and adds it into the list of dead roots that need to
  * be deleted
@@ -1262,13 +1368,21 @@ void btrfs_add_dead_root(struct btrfs_root *root)
 	struct btrfs_fs_info *fs_info = root->fs_info;
 
 	spin_lock(&fs_info->trans_lock);
-	if (list_empty(&root->root_list))
-		list_add_tail(&root->root_list, &fs_info->dead_roots);
+	if (list_empty(&root->root_list)) {
+		btrfs_grab_root(root);
+
+		/* We want to process the partially complete drops first. */
+		if (test_bit(BTRFS_ROOT_UNFINISHED_DROP, &root->state))
+			list_add(&root->root_list, &fs_info->dead_roots);
+		else
+			list_add_tail(&root->root_list, &fs_info->dead_roots);
+	}
 	spin_unlock(&fs_info->trans_lock);
 }
 
 /*
- * update all the cowonly tree roots on disk
+ * Update each subvolume root and its relocation root, if it exists, in the tree
+ * of tree roots. Also free log roots if they exist.
  */
 static noinline int commit_fs_roots(struct btrfs_trans_handle *trans)
 {
@@ -1276,7 +1390,12 @@ static noinline int commit_fs_roots(struct btrfs_trans_handle *trans)
 	struct btrfs_root *gang[8];
 	int i;
 	int ret;
-	int err = 0;
+
+	/*
+	 * At this point no one can be using this transaction to modify any tree
+	 * and no one can start another transaction to modify any tree either.
+	 */
+	ASSERT(trans->transaction->state == TRANS_STATE_COMMIT_DOING);
 
 	spin_lock(&fs_info->fs_roots_radix_lock);
 	while (1) {
@@ -1288,15 +1407,25 @@ static noinline int commit_fs_roots(struct btrfs_trans_handle *trans)
 			break;
 		for (i = 0; i < ret; i++) {
 			struct btrfs_root *root = gang[i];
+			int ret2;
+
+			/*
+			 * At this point we can neither have tasks logging inodes
+			 * from a root nor trying to commit a log tree.
+			 */
+			ASSERT(atomic_read(&root->log_writers) == 0);
+			ASSERT(atomic_read(&root->log_commit[0]) == 0);
+			ASSERT(atomic_read(&root->log_commit[1]) == 0);
+
 			radix_tree_tag_clear(&fs_info->fs_roots_radix,
 					(unsigned long)root->root_key.objectid,
 					BTRFS_ROOT_TRANS_TAG);
 			spin_unlock(&fs_info->fs_roots_radix_lock);
 
 			btrfs_free_log(trans, root);
-			btrfs_update_reloc_root(trans, root);
-
-			btrfs_save_ino_cache(root, trans);
+			ret2 = btrfs_update_reloc_root(trans, root);
+			if (ret2)
+				return ret2;
 
 			/* see comments in should_cow_block() */
 			clear_bit(BTRFS_ROOT_FORCE_COW, &root->state);
@@ -1309,17 +1438,17 @@ static noinline int commit_fs_roots(struct btrfs_trans_handle *trans)
 						    root->node);
 			}
 
-			err = btrfs_update_root(trans, fs_info->tree_root,
+			ret2 = btrfs_update_root(trans, fs_info->tree_root,
 						&root->root_key,
 						&root->root_item);
+			if (ret2)
+				return ret2;
 			spin_lock(&fs_info->fs_roots_radix_lock);
-			if (err)
-				break;
 			btrfs_qgroup_free_meta_all_pertrans(root);
 		}
 	}
 	spin_unlock(&fs_info->fs_roots_radix_lock);
-	return err;
+	return 0;
 }
 
 /*
@@ -1337,8 +1466,10 @@ int btrfs_defrag_root(struct btrfs_root *root)
 
 	while (1) {
 		trans = btrfs_start_transaction(root, 0);
-		if (IS_ERR(trans))
-			return PTR_ERR(trans);
+		if (IS_ERR(trans)) {
+			ret = PTR_ERR(trans);
+			break;
+		}
 
 		ret = btrfs_defrag_leaves(trans, root);
 
@@ -1389,13 +1520,26 @@ static int qgroup_account_snapshot(struct btrfs_trans_handle *trans,
 	 * recorded root will never be updated again, causing an outdated root
 	 * item.
 	 */
-	record_root_in_trans(trans, src, 1);
+	ret = record_root_in_trans(trans, src, 1);
+	if (ret)
+		return ret;
 
 	/*
-	 * We are going to commit transaction, see btrfs_commit_transaction()
-	 * comment for reason locking tree_log_mutex
+	 * btrfs_qgroup_inherit relies on a consistent view of the usage for the
+	 * src root, so we must run the delayed refs here.
+	 *
+	 * However this isn't particularly fool proof, because there's no
+	 * synchronization keeping us from changing the tree after this point
+	 * before we do the qgroup_inherit, or even from making changes while
+	 * we're doing the qgroup_inherit.  But that's a problem for the future,
+	 * for now flush the delayed refs to narrow the race window where the
+	 * qgroup counters could end up wrong.
 	 */
-	mutex_lock(&fs_info->tree_log_mutex);
+	ret = btrfs_run_delayed_refs(trans, (unsigned long)-1);
+	if (ret) {
+		btrfs_abort_transaction(trans, ret);
+		return ret;
+	}
 
 	ret = commit_fs_roots(trans);
 	if (ret)
@@ -1432,8 +1576,6 @@ static int qgroup_account_snapshot(struct btrfs_trans_handle *trans,
 			"Error while writing out transaction for qgroup");
 
 out:
-	mutex_unlock(&fs_info->tree_log_mutex);
-
 	/*
 	 * Force parent root to be updated, as we recorded it before so its
 	 * last_trans == cur_transid.
@@ -1441,7 +1583,7 @@ out:
 	 * insert_dir_item()
 	 */
 	if (!ret)
-		record_root_in_trans(trans, parent, 1);
+		ret = record_root_in_trans(trans, parent, 1);
 	return ret;
 }
 
@@ -1477,7 +1619,6 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	u64 index = 0;
 	u64 objectid;
 	u64 root_flags;
-	uuid_le new_uuid;
 
 	ASSERT(pending->path);
 	path = pending->path;
@@ -1485,7 +1626,7 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	ASSERT(pending->root_item);
 	new_root_item = pending->root_item;
 
-	pending->error = btrfs_find_free_objectid(tree_root, &objectid);
+	pending->error = btrfs_get_free_objectid(tree_root, &objectid);
 	if (pending->error)
 		goto no_free_objectid;
 
@@ -1498,7 +1639,7 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	btrfs_reloc_pre_snapshot(pending, &to_reserve);
 
 	if (to_reserve > 0) {
-		pending->error = btrfs_block_rsv_add(root,
+		pending->error = btrfs_block_rsv_add(fs_info,
 						     &pending->block_rsv,
 						     to_reserve,
 						     BTRFS_RESERVE_NO_FLUSH);
@@ -1519,8 +1660,9 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	dentry = pending->dentry;
 	parent_inode = pending->dir;
 	parent_root = BTRFS_I(parent_inode)->root;
-	record_root_in_trans(trans, parent_root, 0);
-
+	ret = record_root_in_trans(trans, parent_root, 0);
+	if (ret)
+		goto fail;
 	cur_time = current_time(parent_inode);
 
 	/*
@@ -1556,7 +1698,11 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 		goto fail;
 	}
 
-	record_root_in_trans(trans, root, 0);
+	ret = record_root_in_trans(trans, root, 0);
+	if (ret) {
+		btrfs_abort_transaction(trans, ret);
+		goto fail;
+	}
 	btrfs_set_root_last_snapshot(&root->root_item, trans->transid);
 	memcpy(new_root_item, &root->root_item, sizeof(*new_root_item));
 	btrfs_check_and_init_root_item(new_root_item);
@@ -1570,8 +1716,7 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 
 	btrfs_set_root_generation_v2(new_root_item,
 			trans->transid);
-	uuid_le_gen(&new_uuid);
-	memcpy(new_root_item->uuid, new_uuid.b, BTRFS_UUID_SIZE);
+	generate_random_guid(new_root_item->uuid);
 	memcpy(new_root_item->parent_uuid, root->root_item.uuid,
 			BTRFS_UUID_SIZE);
 	if (!(root_flags & BTRFS_ROOT_SUBVOL_RDONLY)) {
@@ -1587,7 +1732,8 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	btrfs_set_root_otransid(new_root_item, trans->transid);
 
 	old = btrfs_lock_root_node(root);
-	ret = btrfs_cow_block(trans, root, old, NULL, 0, &old);
+	ret = btrfs_cow_block(trans, root, old, NULL, 0, &old,
+			      BTRFS_NESTING_COW);
 	if (ret) {
 		btrfs_tree_unlock(old);
 		free_extent_buffer(old);
@@ -1595,8 +1741,6 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 		goto fail;
 	}
 
-	btrfs_set_lock_blocking_write(old);
-
 	ret = btrfs_copy_root(trans, root, old, &tmp, objectid);
 	/* clean up in any case */
 	btrfs_tree_unlock(old);
@@ -1633,9 +1777,10 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	}
 
 	key.offset = (u64)-1;
-	pending->snap = btrfs_read_fs_root_no_name(fs_info, &key);
+	pending->snap = btrfs_get_new_fs_root(fs_info, objectid, pending->anon_dev);
 	if (IS_ERR(pending->snap)) {
 		ret = PTR_ERR(pending->snap);
+		pending->snap = NULL;
 		btrfs_abort_transaction(trans, ret);
 		goto fail;
 	}
@@ -1646,12 +1791,6 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 		goto fail;
 	}
 
-	ret = btrfs_run_delayed_refs(trans, (unsigned long)-1);
-	if (ret) {
-		btrfs_abort_transaction(trans, ret);
-		goto fail;
-	}
-
 	/*
 	 * Do special qgroup accounting for snapshot, as we do some qgroup
 	 * snapshot hack to do fast snapshot.
@@ -1675,14 +1814,15 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 
 	btrfs_i_size_write(BTRFS_I(parent_inode), parent_inode->i_size +
 					 dentry->d_name.len * 2);
-	parent_inode->i_mtime = parent_inode->i_ctime =
-		current_time(parent_inode);
-	ret = btrfs_update_inode_fallback(trans, parent_root, parent_inode);
+	parent_inode->i_mtime = current_time(parent_inode);
+	parent_inode->i_ctime = parent_inode->i_mtime;
+	ret = btrfs_update_inode_fallback(trans, parent_root, BTRFS_I(parent_inode));
 	if (ret) {
 		btrfs_abort_transaction(trans, ret);
 		goto fail;
 	}
-	ret = btrfs_uuid_tree_add(trans, new_uuid.b, BTRFS_UUID_KEY_SUBVOL,
+	ret = btrfs_uuid_tree_add(trans, new_root_item->uuid,
+				  BTRFS_UUID_KEY_SUBVOL,
 				  objectid);
 	if (ret) {
 		btrfs_abort_transaction(trans, ret);
@@ -1698,12 +1838,6 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 		}
 	}
 
-	ret = btrfs_run_delayed_refs(trans, (unsigned long)-1);
-	if (ret) {
-		btrfs_abort_transaction(trans, ret);
-		goto fail;
-	}
-
 fail:
 	pending->error = ret;
 dir_item_existed:
@@ -1756,6 +1890,8 @@ static void update_super_roots(struct btrfs_fs_info *fs_info)
 	super->root_level = root_item->level;
 	if (btrfs_test_opt(fs_info, SPACE_CACHE))
 		super->cache_generation = root_item->generation;
+	else if (test_bit(BTRFS_FS_CLEANUP_SPACE_CACHE_V1, &fs_info->flags))
+		super->cache_generation = 0;
 	if (test_bit(BTRFS_FS_UPDATE_UUID_TREE_GEN, &fs_info->flags))
 		super->uuid_tree_generation = root_item->generation;
 }
@@ -1786,74 +1922,14 @@ int btrfs_transaction_blocked(struct btrfs_fs_info *info)
 	return ret;
 }
 
-/*
- * wait for the current transaction commit to start and block subsequent
- * transaction joins
- */
-static void wait_current_trans_commit_start(struct btrfs_fs_info *fs_info,
-					    struct btrfs_transaction *trans)
-{
-	wait_event(fs_info->transaction_blocked_wait,
-		   trans->state >= TRANS_STATE_COMMIT_START || trans->aborted);
-}
-
-/*
- * wait for the current transaction to start and then become unblocked.
- * caller holds ref.
- */
-static void wait_current_trans_commit_start_and_unblock(
-					struct btrfs_fs_info *fs_info,
-					struct btrfs_transaction *trans)
-{
-	wait_event(fs_info->transaction_wait,
-		   trans->state >= TRANS_STATE_UNBLOCKED || trans->aborted);
-}
-
-/*
- * commit transactions asynchronously. once btrfs_commit_transaction_async
- * returns, any subsequent transaction will not be allowed to join.
- */
-struct btrfs_async_commit {
-	struct btrfs_trans_handle *newtrans;
-	struct work_struct work;
-};
-
-static void do_async_commit(struct work_struct *work)
-{
-	struct btrfs_async_commit *ac =
-		container_of(work, struct btrfs_async_commit, work);
-
-	/*
-	 * We've got freeze protection passed with the transaction.
-	 * Tell lockdep about it.
-	 */
-	if (ac->newtrans->type & __TRANS_FREEZABLE)
-		__sb_writers_acquired(ac->newtrans->fs_info->sb, SB_FREEZE_FS);
-
-	current->journal_info = ac->newtrans;
-
-	btrfs_commit_transaction(ac->newtrans);
-	kfree(ac);
-}
-
-int btrfs_commit_transaction_async(struct btrfs_trans_handle *trans,
-				   int wait_for_unblock)
+void btrfs_commit_transaction_async(struct btrfs_trans_handle *trans)
 {
 	struct btrfs_fs_info *fs_info = trans->fs_info;
-	struct btrfs_async_commit *ac;
 	struct btrfs_transaction *cur_trans;
 
-	ac = kmalloc(sizeof(*ac), GFP_NOFS);
-	if (!ac)
-		return -ENOMEM;
-
-	INIT_WORK(&ac->work, do_async_commit);
-	ac->newtrans = btrfs_join_transaction(trans->root);
-	if (IS_ERR(ac->newtrans)) {
-		int err = PTR_ERR(ac->newtrans);
-		kfree(ac);
-		return err;
-	}
+	/* Kick the transaction kthread. */
+	set_bit(BTRFS_FS_COMMIT_TRANS, &fs_info->flags);
+	wake_up_process(fs_info->transaction_kthread);
 
 	/* take transaction reference */
 	cur_trans = trans->transaction;
@@ -1862,28 +1938,16 @@ int btrfs_commit_transaction_async(struct btrfs_trans_handle *trans,
 	btrfs_end_transaction(trans);
 
 	/*
-	 * Tell lockdep we've released the freeze rwsem, since the
-	 * async commit thread will be the one to unlock it.
+	 * Wait for the current transaction commit to start and block
+	 * subsequent transaction joins
 	 */
-	if (ac->newtrans->type & __TRANS_FREEZABLE)
-		__sb_writers_release(fs_info->sb, SB_FREEZE_FS);
-
-	schedule_work(&ac->work);
-
-	/* wait for transaction to start and unblock */
-	if (wait_for_unblock)
-		wait_current_trans_commit_start_and_unblock(fs_info, cur_trans);
-	else
-		wait_current_trans_commit_start(fs_info, cur_trans);
-
-	if (current->journal_info == trans)
-		current->journal_info = NULL;
-
+	btrfs_might_wait_for_state(fs_info, BTRFS_LOCKDEP_TRANS_COMMIT_START);
+	wait_event(fs_info->transaction_blocked_wait,
+		   cur_trans->state >= TRANS_STATE_COMMIT_START ||
+		   TRANS_ABORTED(cur_trans));
 	btrfs_put_transaction(cur_trans);
-	return 0;
 }
 
-
 static void cleanup_transaction(struct btrfs_trans_handle *trans, int err)
 {
 	struct btrfs_fs_info *fs_info = trans->fs_info;
@@ -1902,15 +1966,31 @@ static void cleanup_transaction(struct btrfs_trans_handle *trans, int err)
 	 */
 	BUG_ON(list_empty(&cur_trans->list));
 
-	list_del_init(&cur_trans->list);
 	if (cur_trans == fs_info->running_transaction) {
 		cur_trans->state = TRANS_STATE_COMMIT_DOING;
 		spin_unlock(&fs_info->trans_lock);
+
+		/*
+		 * The thread has already released the lockdep map as reader
+		 * already in btrfs_commit_transaction().
+		 */
+		btrfs_might_wait_for_event(fs_info, btrfs_trans_num_writers);
 		wait_event(cur_trans->writer_wait,
 			   atomic_read(&cur_trans->num_writers) == 1);
 
 		spin_lock(&fs_info->trans_lock);
 	}
+
+	/*
+	 * Now that we know no one else is still using the transaction we can
+	 * remove the transaction from the list of transactions. This avoids
+	 * the transaction kthread from cleaning up the transaction while some
+	 * other task is still using it, which could result in a use-after-free
+	 * on things like log trees, as it forces the transaction kthread to
+	 * wait for this transaction to be cleaned up by us.
+	 */
+	list_del_init(&cur_trans->list);
+
 	spin_unlock(&fs_info->trans_lock);
 
 	btrfs_cleanup_one_transaction(trans->transaction, fs_info);
@@ -1925,7 +2005,7 @@ static void cleanup_transaction(struct btrfs_trans_handle *trans, int err)
 	btrfs_put_transaction(cur_trans);
 	btrfs_put_transaction(cur_trans);
 
-	trace_btrfs_transaction_commit(trans->root);
+	trace_btrfs_transaction_commit(fs_info);
 
 	if (current->journal_info == trans)
 		current->journal_info = NULL;
@@ -1949,63 +2029,64 @@ static void btrfs_cleanup_pending_block_groups(struct btrfs_trans_handle *trans)
        }
 }
 
-static inline int btrfs_start_delalloc_flush(struct btrfs_trans_handle *trans)
+static inline int btrfs_start_delalloc_flush(struct btrfs_fs_info *fs_info)
 {
-	struct btrfs_fs_info *fs_info = trans->fs_info;
-
 	/*
-	 * We use writeback_inodes_sb here because if we used
+	 * We use try_to_writeback_inodes_sb() here because if we used
 	 * btrfs_start_delalloc_roots we would deadlock with fs freeze.
 	 * Currently are holding the fs freeze lock, if we do an async flush
 	 * we'll do btrfs_join_transaction() and deadlock because we need to
 	 * wait for the fs freeze lock.  Using the direct flushing we benefit
 	 * from already being in a transaction and our join_transaction doesn't
 	 * have to re-take the fs freeze lock.
+	 *
+	 * Note that try_to_writeback_inodes_sb() will only trigger writeback
+	 * if it can read lock sb->s_umount. It will always be able to lock it,
+	 * except when the filesystem is being unmounted or being frozen, but in
+	 * those cases sync_filesystem() is called, which results in calling
+	 * writeback_inodes_sb() while holding a write lock on sb->s_umount.
+	 * Note that we don't call writeback_inodes_sb() directly, because it
+	 * will emit a warning if sb->s_umount is not locked.
 	 */
-	if (btrfs_test_opt(fs_info, FLUSHONCOMMIT)) {
-		writeback_inodes_sb(fs_info->sb, WB_REASON_SYNC);
-	} else {
-		struct btrfs_pending_snapshot *pending;
-		struct list_head *head = &trans->transaction->pending_snapshots;
-
-		/*
-		 * Flush dellaloc for any root that is going to be snapshotted.
-		 * This is done to avoid a corrupted version of files, in the
-		 * snapshots, that had both buffered and direct IO writes (even
-		 * if they were done sequentially) due to an unordered update of
-		 * the inode's size on disk.
-		 */
-		list_for_each_entry(pending, head, list) {
-			int ret;
-
-			ret = btrfs_start_delalloc_snapshot(pending->root);
-			if (ret)
-				return ret;
-		}
-	}
+	if (btrfs_test_opt(fs_info, FLUSHONCOMMIT))
+		try_to_writeback_inodes_sb(fs_info->sb, WB_REASON_SYNC);
 	return 0;
 }
 
-static inline void btrfs_wait_delalloc_flush(struct btrfs_trans_handle *trans)
+static inline void btrfs_wait_delalloc_flush(struct btrfs_fs_info *fs_info)
 {
-	struct btrfs_fs_info *fs_info = trans->fs_info;
-
-	if (btrfs_test_opt(fs_info, FLUSHONCOMMIT)) {
+	if (btrfs_test_opt(fs_info, FLUSHONCOMMIT))
 		btrfs_wait_ordered_roots(fs_info, U64_MAX, 0, (u64)-1);
-	} else {
-		struct btrfs_pending_snapshot *pending;
-		struct list_head *head = &trans->transaction->pending_snapshots;
+}
 
-		/*
-		 * Wait for any dellaloc that we started previously for the roots
-		 * that are going to be snapshotted. This is to avoid a corrupted
-		 * version of files in the snapshots that had both buffered and
-		 * direct IO writes (even if they were done sequentially).
-		 */
-		list_for_each_entry(pending, head, list)
-			btrfs_wait_ordered_extents(pending->root,
-						   U64_MAX, 0, U64_MAX);
-	}
+/*
+ * Add a pending snapshot associated with the given transaction handle to the
+ * respective handle. This must be called after the transaction commit started
+ * and while holding fs_info->trans_lock.
+ * This serves to guarantee a caller of btrfs_commit_transaction() that it can
+ * safely free the pending snapshot pointer in case btrfs_commit_transaction()
+ * returns an error.
+ */
+static void add_pending_snapshot(struct btrfs_trans_handle *trans)
+{
+	struct btrfs_transaction *cur_trans = trans->transaction;
+
+	if (!trans->pending_snapshot)
+		return;
+
+	lockdep_assert_held(&trans->fs_info->trans_lock);
+	ASSERT(cur_trans->state >= TRANS_STATE_COMMIT_START);
+
+	list_add(&trans->pending_snapshot->list, &cur_trans->pending_snapshots);
+}
+
+static void update_commit_stats(struct btrfs_fs_info *fs_info, ktime_t interval)
+{
+	fs_info->commit_stats.commit_count++;
+	fs_info->commit_stats.last_commit_dur = interval;
+	fs_info->commit_stats.max_commit_dur =
+			max_t(u64, fs_info->commit_stats.max_commit_dur, interval);
+	fs_info->commit_stats.total_commit_dur += interval;
 }
 
 int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
@@ -2014,53 +2095,38 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 	struct btrfs_transaction *cur_trans = trans->transaction;
 	struct btrfs_transaction *prev_trans = NULL;
 	int ret;
+	ktime_t start_time;
+	ktime_t interval;
 
 	ASSERT(refcount_read(&trans->use_count) == 1);
-
-	/*
-	 * Some places just start a transaction to commit it.  We need to make
-	 * sure that if this commit fails that the abort code actually marks the
-	 * transaction as failed, so set trans->dirty to make the abort code do
-	 * the right thing.
-	 */
-	trans->dirty = true;
+	btrfs_trans_state_lockdep_acquire(fs_info, BTRFS_LOCKDEP_TRANS_COMMIT_START);
 
 	/* Stop the commit early if ->aborted is set */
-	if (unlikely(READ_ONCE(cur_trans->aborted))) {
+	if (TRANS_ABORTED(cur_trans)) {
 		ret = cur_trans->aborted;
-		btrfs_end_transaction(trans);
-		return ret;
+		goto lockdep_trans_commit_start_release;
 	}
 
 	btrfs_trans_release_metadata(trans);
 	trans->block_rsv = NULL;
 
-	/* make a pass through all the delayed refs we have so far
-	 * any runnings procs may add more while we are here
-	 */
-	ret = btrfs_run_delayed_refs(trans, 0);
-	if (ret) {
-		btrfs_end_transaction(trans);
-		return ret;
-	}
-
-	cur_trans = trans->transaction;
-
 	/*
-	 * set the flushing flag so procs in this transaction have to
-	 * start sending their work down.
+	 * We only want one transaction commit doing the flushing so we do not
+	 * waste a bunch of time on lock contention on the extent root node.
 	 */
-	cur_trans->delayed_refs.flushing = 1;
-	smp_wmb();
+	if (!test_and_set_bit(BTRFS_DELAYED_REFS_FLUSHING,
+			      &cur_trans->delayed_refs.flags)) {
+		/*
+		 * Make a pass through all the delayed refs we have so far.
+		 * Any running threads may add more while we are here.
+		 */
+		ret = btrfs_run_delayed_refs(trans, 0);
+		if (ret)
+			goto lockdep_trans_commit_start_release;
+	}
 
 	btrfs_create_pending_block_groups(trans);
 
-	ret = btrfs_run_delayed_refs(trans, 0);
-	if (ret) {
-		btrfs_end_transaction(trans);
-		return ret;
-	}
-
 	if (!test_bit(BTRFS_TRANS_DIRTY_BG_RUN, &cur_trans->flags)) {
 		int run_it = 0;
 
@@ -2085,22 +2151,29 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 
 		if (run_it) {
 			ret = btrfs_start_dirty_block_groups(trans);
-			if (ret) {
-				btrfs_end_transaction(trans);
-				return ret;
-			}
+			if (ret)
+				goto lockdep_trans_commit_start_release;
 		}
 	}
 
 	spin_lock(&fs_info->trans_lock);
 	if (cur_trans->state >= TRANS_STATE_COMMIT_START) {
+		enum btrfs_trans_state want_state = TRANS_STATE_COMPLETED;
+
+		add_pending_snapshot(trans);
+
 		spin_unlock(&fs_info->trans_lock);
 		refcount_inc(&cur_trans->use_count);
-		ret = btrfs_end_transaction(trans);
 
-		wait_for_commit(cur_trans);
+		if (trans->in_fsync)
+			want_state = TRANS_STATE_SUPER_COMMITTED;
+
+		btrfs_trans_state_lockdep_release(fs_info,
+						  BTRFS_LOCKDEP_TRANS_COMMIT_START);
+		ret = btrfs_end_transaction(trans);
+		wait_for_commit(cur_trans, want_state);
 
-		if (unlikely(cur_trans->aborted))
+		if (TRANS_ABORTED(cur_trans))
 			ret = cur_trans->aborted;
 
 		btrfs_put_transaction(cur_trans);
@@ -2110,20 +2183,27 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 
 	cur_trans->state = TRANS_STATE_COMMIT_START;
 	wake_up(&fs_info->transaction_blocked_wait);
+	btrfs_trans_state_lockdep_release(fs_info, BTRFS_LOCKDEP_TRANS_COMMIT_START);
 
 	if (cur_trans->list.prev != &fs_info->trans_list) {
+		enum btrfs_trans_state want_state = TRANS_STATE_COMPLETED;
+
+		if (trans->in_fsync)
+			want_state = TRANS_STATE_SUPER_COMMITTED;
+
 		prev_trans = list_entry(cur_trans->list.prev,
 					struct btrfs_transaction, list);
-		if (prev_trans->state != TRANS_STATE_COMPLETED) {
+		if (prev_trans->state < want_state) {
 			refcount_inc(&prev_trans->use_count);
 			spin_unlock(&fs_info->trans_lock);
 
-			wait_for_commit(prev_trans);
-			ret = prev_trans->aborted;
+			wait_for_commit(prev_trans, want_state);
+
+			ret = READ_ONCE(prev_trans->aborted);
 
 			btrfs_put_transaction(prev_trans);
 			if (ret)
-				goto cleanup_transaction;
+				goto lockdep_release;
 		} else {
 			spin_unlock(&fs_info->trans_lock);
 		}
@@ -2135,31 +2215,55 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 		 * abort to prevent writing a new superblock that reflects a
 		 * corrupt state (pointing to trees with unwritten nodes/leafs).
 		 */
-		if (test_bit(BTRFS_FS_STATE_TRANS_ABORTED, &fs_info->fs_state)) {
+		if (BTRFS_FS_ERROR(fs_info)) {
 			ret = -EROFS;
-			goto cleanup_transaction;
+			goto lockdep_release;
 		}
 	}
 
+	/*
+	 * Get the time spent on the work done by the commit thread and not
+	 * the time spent waiting on a previous commit
+	 */
+	start_time = ktime_get_ns();
+
 	extwriter_counter_dec(cur_trans, trans->type);
 
-	ret = btrfs_start_delalloc_flush(trans);
+	ret = btrfs_start_delalloc_flush(fs_info);
 	if (ret)
-		goto cleanup_transaction;
+		goto lockdep_release;
 
 	ret = btrfs_run_delayed_items(trans);
 	if (ret)
-		goto cleanup_transaction;
+		goto lockdep_release;
 
+	/*
+	 * The thread has started/joined the transaction thus it holds the
+	 * lockdep map as a reader. It has to release it before acquiring the
+	 * lockdep map as a writer.
+	 */
+	btrfs_lockdep_release(fs_info, btrfs_trans_num_extwriters);
+	btrfs_might_wait_for_event(fs_info, btrfs_trans_num_extwriters);
 	wait_event(cur_trans->writer_wait,
 		   extwriter_counter_read(cur_trans) == 0);
 
 	/* some pending stuffs might be added after the previous flush. */
 	ret = btrfs_run_delayed_items(trans);
-	if (ret)
+	if (ret) {
+		btrfs_lockdep_release(fs_info, btrfs_trans_num_writers);
 		goto cleanup_transaction;
+	}
 
-	btrfs_wait_delalloc_flush(trans);
+	btrfs_wait_delalloc_flush(fs_info);
+
+	/*
+	 * Wait for all ordered extents started by a fast fsync that joined this
+	 * transaction. Otherwise if this transaction commits before the ordered
+	 * extents complete we lose logged data after a power failure.
+	 */
+	btrfs_might_wait_for_event(fs_info, btrfs_trans_pending_ordered);
+	wait_event(cur_trans->pending_wait,
+		   atomic_read(&cur_trans->pending_ordered) == 0);
 
 	btrfs_scrub_pause(fs_info);
 	/*
@@ -2168,14 +2272,40 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 	 * COMMIT_DOING so make sure to wait for num_writers to == 1 again.
 	 */
 	spin_lock(&fs_info->trans_lock);
+	add_pending_snapshot(trans);
 	cur_trans->state = TRANS_STATE_COMMIT_DOING;
 	spin_unlock(&fs_info->trans_lock);
+
+	/*
+	 * The thread has started/joined the transaction thus it holds the
+	 * lockdep map as a reader. It has to release it before acquiring the
+	 * lockdep map as a writer.
+	 */
+	btrfs_lockdep_release(fs_info, btrfs_trans_num_writers);
+	btrfs_might_wait_for_event(fs_info, btrfs_trans_num_writers);
 	wait_event(cur_trans->writer_wait,
 		   atomic_read(&cur_trans->num_writers) == 1);
 
-	/* ->aborted might be set after the previous check, so check it */
-	if (unlikely(READ_ONCE(cur_trans->aborted))) {
+	/*
+	 * Make lockdep happy by acquiring the state locks after
+	 * btrfs_trans_num_writers is released. If we acquired the state locks
+	 * before releasing the btrfs_trans_num_writers lock then lockdep would
+	 * complain because we did not follow the reverse order unlocking rule.
+	 */
+	btrfs_trans_state_lockdep_acquire(fs_info, BTRFS_LOCKDEP_TRANS_COMPLETED);
+	btrfs_trans_state_lockdep_acquire(fs_info, BTRFS_LOCKDEP_TRANS_SUPER_COMMITTED);
+	btrfs_trans_state_lockdep_acquire(fs_info, BTRFS_LOCKDEP_TRANS_UNBLOCKED);
+
+	/*
+	 * We've started the commit, clear the flag in case we were triggered to
+	 * do an async commit but somebody else started before the transaction
+	 * kthread could do the work.
+	 */
+	clear_bit(BTRFS_FS_COMMIT_TRANS, &fs_info->flags);
+
+	if (TRANS_ABORTED(cur_trans)) {
 		ret = cur_trans->aborted;
+		btrfs_trans_state_lockdep_release(fs_info, BTRFS_LOCKDEP_TRANS_UNBLOCKED);
 		goto scrub_continue;
 	}
 	/*
@@ -2191,10 +2321,8 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 	 * core function of the snapshot creation.
 	 */
 	ret = create_pending_snapshots(trans);
-	if (ret) {
-		mutex_unlock(&fs_info->reloc_mutex);
-		goto scrub_continue;
-	}
+	if (ret)
+		goto unlock_reloc;
 
 	/*
 	 * We insert the dir indexes of the snapshots and update the inode
@@ -2207,16 +2335,12 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 	 * the nodes and leaves.
 	 */
 	ret = btrfs_run_delayed_items(trans);
-	if (ret) {
-		mutex_unlock(&fs_info->reloc_mutex);
-		goto scrub_continue;
-	}
+	if (ret)
+		goto unlock_reloc;
 
 	ret = btrfs_run_delayed_refs(trans, (unsigned long)-1);
-	if (ret) {
-		mutex_unlock(&fs_info->reloc_mutex);
-		goto scrub_continue;
-	}
+	if (ret)
+		goto unlock_reloc;
 
 	/*
 	 * make sure none of the code above managed to slip in a
@@ -2226,27 +2350,9 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 
 	WARN_ON(cur_trans != trans->transaction);
 
-	/* btrfs_commit_tree_roots is responsible for getting the
-	 * various roots consistent with each other.  Every pointer
-	 * in the tree of tree roots has to point to the most up to date
-	 * root for every subvolume and other tree.  So, we have to keep
-	 * the tree logging code from jumping in and changing any
-	 * of the trees.
-	 *
-	 * At this point in the commit, there can't be any tree-log
-	 * writers, but a little lower down we drop the trans mutex
-	 * and let new people in.  By holding the tree_log_mutex
-	 * from now until after the super is written, we avoid races
-	 * with the tree-log code.
-	 */
-	mutex_lock(&fs_info->tree_log_mutex);
-
 	ret = commit_fs_roots(trans);
-	if (ret) {
-		mutex_unlock(&fs_info->tree_log_mutex);
-		mutex_unlock(&fs_info->reloc_mutex);
-		goto scrub_continue;
-	}
+	if (ret)
+		goto unlock_reloc;
 
 	/*
 	 * Since the transaction is done, we can apply the pending changes
@@ -2260,47 +2366,26 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 	btrfs_free_log_root_tree(trans, fs_info);
 
 	/*
-	 * commit_fs_roots() can call btrfs_save_ino_cache(), which generates
-	 * new delayed refs. Must handle them or qgroup can be wrong.
-	 */
-	ret = btrfs_run_delayed_refs(trans, (unsigned long)-1);
-	if (ret) {
-		mutex_unlock(&fs_info->tree_log_mutex);
-		mutex_unlock(&fs_info->reloc_mutex);
-		goto scrub_continue;
-	}
-
-	/*
 	 * Since fs roots are all committed, we can get a quite accurate
 	 * new_roots. So let's do quota accounting.
 	 */
 	ret = btrfs_qgroup_account_extents(trans);
-	if (ret < 0) {
-		mutex_unlock(&fs_info->tree_log_mutex);
-		mutex_unlock(&fs_info->reloc_mutex);
-		goto scrub_continue;
-	}
+	if (ret < 0)
+		goto unlock_reloc;
 
 	ret = commit_cowonly_roots(trans);
-	if (ret) {
-		mutex_unlock(&fs_info->tree_log_mutex);
-		mutex_unlock(&fs_info->reloc_mutex);
-		goto scrub_continue;
-	}
+	if (ret)
+		goto unlock_reloc;
 
 	/*
 	 * The tasks which save the space cache and inode cache may also
 	 * update ->aborted, check it.
 	 */
-	if (unlikely(READ_ONCE(cur_trans->aborted))) {
+	if (TRANS_ABORTED(cur_trans)) {
 		ret = cur_trans->aborted;
-		mutex_unlock(&fs_info->tree_log_mutex);
-		mutex_unlock(&fs_info->reloc_mutex);
-		goto scrub_continue;
+		goto unlock_reloc;
 	}
 
-	btrfs_prepare_extent_commit(fs_info);
-
 	cur_trans = fs_info->running_transaction;
 
 	btrfs_set_root_node(&fs_info->tree_root->root_item,
@@ -2313,6 +2398,13 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 	list_add_tail(&fs_info->chunk_root->dirty_list,
 		      &cur_trans->switch_commits);
 
+	if (btrfs_fs_incompat(fs_info, EXTENT_TREE_V2)) {
+		btrfs_set_root_node(&fs_info->block_group_root->root_item,
+				    fs_info->block_group_root->node);
+		list_add_tail(&fs_info->block_group_root->dirty_list,
+			      &cur_trans->switch_commits);
+	}
+
 	switch_commit_roots(trans);
 
 	ASSERT(list_empty(&cur_trans->dirty_bgs));
@@ -2331,6 +2423,16 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 
 	btrfs_trans_release_chunk_metadata(trans);
 
+	/*
+	 * Before changing the transaction state to TRANS_STATE_UNBLOCKED and
+	 * setting fs_info->running_transaction to NULL, lock tree_log_mutex to
+	 * make sure that before we commit our superblock, no other task can
+	 * start a new transaction and commit a log tree before we commit our
+	 * superblock. Anyone trying to commit a log tree locks this mutex before
+	 * writing its superblock.
+	 */
+	mutex_lock(&fs_info->tree_log_mutex);
+
 	spin_lock(&fs_info->trans_lock);
 	cur_trans->state = TRANS_STATE_UNBLOCKED;
 	fs_info->running_transaction = NULL;
@@ -2338,6 +2440,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 	mutex_unlock(&fs_info->reloc_mutex);
 
 	wake_up(&fs_info->transaction_wait);
+	btrfs_trans_state_lockdep_release(fs_info, BTRFS_LOCKDEP_TRANS_UNBLOCKED);
 
 	ret = btrfs_write_and_wait_transaction(trans);
 	if (ret) {
@@ -2347,6 +2450,13 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 		goto scrub_continue;
 	}
 
+	/*
+	 * At this point, we should have written all the tree blocks allocated
+	 * in this transaction. So it's now safe to free the redirtyied extent
+	 * buffers.
+	 */
+	btrfs_free_redirty_list(cur_trans);
+
 	ret = write_all_supers(fs_info, 0);
 	/*
 	 * the super is written, we can safely allow the tree-loggers
@@ -2356,6 +2466,14 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 	if (ret)
 		goto scrub_continue;
 
+	/*
+	 * We needn't acquire the lock here because there is no other task
+	 * which can change it.
+	 */
+	cur_trans->state = TRANS_STATE_SUPER_COMMITTED;
+	wake_up(&cur_trans->commit_wait);
+	btrfs_trans_state_lockdep_release(fs_info, BTRFS_LOCKDEP_TRANS_SUPER_COMMITTED);
+
 	btrfs_finish_extent_commit(trans);
 
 	if (test_bit(BTRFS_TRANS_HAVE_FREE_BGS, &cur_trans->flags))
@@ -2368,7 +2486,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 	 */
 	cur_trans->state = TRANS_STATE_COMPLETED;
 	wake_up(&cur_trans->commit_wait);
-	clear_bit(BTRFS_FS_NEED_ASYNC_COMMIT, &fs_info->flags);
+	btrfs_trans_state_lockdep_release(fs_info, BTRFS_LOCKDEP_TRANS_COMPLETED);
 
 	spin_lock(&fs_info->trans_lock);
 	list_del_init(&cur_trans->list);
@@ -2380,7 +2498,9 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 	if (trans->type & __TRANS_FREEZABLE)
 		sb_end_intwrite(fs_info->sb);
 
-	trace_btrfs_transaction_commit(trans->root);
+	trace_btrfs_transaction_commit(fs_info);
+
+	interval = ktime_get_ns() - start_time;
 
 	btrfs_scrub_continue(fs_info);
 
@@ -2389,9 +2509,16 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 
 	kmem_cache_free(btrfs_trans_handle_cachep, trans);
 
+	update_commit_stats(fs_info, interval);
+
 	return ret;
 
+unlock_reloc:
+	mutex_unlock(&fs_info->reloc_mutex);
+	btrfs_trans_state_lockdep_release(fs_info, BTRFS_LOCKDEP_TRANS_UNBLOCKED);
 scrub_continue:
+	btrfs_trans_state_lockdep_release(fs_info, BTRFS_LOCKDEP_TRANS_SUPER_COMMITTED);
+	btrfs_trans_state_lockdep_release(fs_info, BTRFS_LOCKDEP_TRANS_COMPLETED);
 	btrfs_scrub_continue(fs_info);
 cleanup_transaction:
 	btrfs_trans_release_metadata(trans);
@@ -2404,6 +2531,16 @@ cleanup_transaction:
 	cleanup_transaction(trans, ret);
 
 	return ret;
+
+lockdep_release:
+	btrfs_lockdep_release(fs_info, btrfs_trans_num_extwriters);
+	btrfs_lockdep_release(fs_info, btrfs_trans_num_writers);
+	goto cleanup_transaction;
+
+lockdep_trans_commit_start_release:
+	btrfs_trans_state_lockdep_release(fs_info, BTRFS_LOCKDEP_TRANS_COMMIT_START);
+	btrfs_end_transaction(trans);
+	return ret;
 }
 
 /*
@@ -2416,10 +2553,10 @@ cleanup_transaction:
  * because btrfs_commit_super will poke cleaner thread and it will process it a
  * few seconds later.
  */
-int btrfs_clean_one_deleted_snapshot(struct btrfs_root *root)
+int btrfs_clean_one_deleted_snapshot(struct btrfs_fs_info *fs_info)
 {
+	struct btrfs_root *root;
 	int ret;
-	struct btrfs_fs_info *fs_info = root->fs_info;
 
 	spin_lock(&fs_info->trans_lock);
 	if (list_empty(&fs_info->dead_roots)) {
@@ -2437,10 +2574,11 @@ int btrfs_clean_one_deleted_snapshot(struct btrfs_root *root)
 
 	if (btrfs_header_backref_rev(root->node) <
 			BTRFS_MIXED_BACKREF_REV)
-		ret = btrfs_drop_snapshot(root, NULL, 0, 0);
+		ret = btrfs_drop_snapshot(root, 0, 0);
 	else
-		ret = btrfs_drop_snapshot(root, NULL, 1, 0);
+		ret = btrfs_drop_snapshot(root, 1, 0);
 
+	btrfs_put_root(root);
 	return (ret < 0) ? 0 : 1;
 }
 
@@ -2453,16 +2591,6 @@ void btrfs_apply_pending_changes(struct btrfs_fs_info *fs_info)
 	if (!prev)
 		return;
 
-	bit = 1 << BTRFS_PENDING_SET_INODE_MAP_CACHE;
-	if (prev & bit)
-		btrfs_set_opt(fs_info->mount_opt, INODE_MAP_CACHE);
-	prev &= ~bit;
-
-	bit = 1 << BTRFS_PENDING_CLEAR_INODE_MAP_CACHE;
-	if (prev & bit)
-		btrfs_clear_opt(fs_info->mount_opt, INODE_MAP_CACHE);
-	prev &= ~bit;
-
 	bit = 1 << BTRFS_PENDING_COMMIT;
 	if (prev & bit)
 		btrfs_debug(fs_info, "pending commit done");
diff --git a/fs/btrfs/transaction.h b/fs/btrfs/transaction.h
index 49f7196368f5..970ff316069d 100644
--- a/fs/btrfs/transaction.h
+++ b/fs/btrfs/transaction.h
@@ -16,6 +16,7 @@ enum btrfs_trans_state {
 	TRANS_STATE_COMMIT_START,
 	TRANS_STATE_COMMIT_DOING,
 	TRANS_STATE_UNBLOCKED,
+	TRANS_STATE_SUPER_COMMITTED,
 	TRANS_STATE_COMPLETED,
 	TRANS_STATE_MAX,
 };
@@ -71,6 +72,7 @@ struct btrfs_transaction {
 	 */
 	struct list_head io_bgs;
 	struct list_head dropped_roots;
+	struct extent_io_tree pinned_extents;
 
 	/*
 	 * we need to make sure block group deletion doesn't race with
@@ -84,6 +86,16 @@ struct btrfs_transaction {
 	spinlock_t dropped_roots_lock;
 	struct btrfs_delayed_ref_root delayed_refs;
 	struct btrfs_fs_info *fs_info;
+
+	/*
+	 * Number of ordered extents the transaction must wait for before
+	 * committing. These are ordered extents started by a fast fsync.
+	 */
+	atomic_t pending_ordered;
+	wait_queue_head_t pending_wait;
+
+	spinlock_t releasing_ebs_lock;
+	struct list_head releasing_ebs;
 };
 
 #define __TRANS_FREEZABLE	(1U << 0)
@@ -103,8 +115,6 @@ struct btrfs_transaction {
 
 #define TRANS_EXTWRITERS	(__TRANS_START | __TRANS_ATTACH)
 
-#define BTRFS_SEND_TRANS_STUB	((void *)1)
-
 struct btrfs_trans_handle {
 	u64 transid;
 	u64 bytes_reserved;
@@ -113,19 +123,32 @@ struct btrfs_trans_handle {
 	struct btrfs_transaction *transaction;
 	struct btrfs_block_rsv *block_rsv;
 	struct btrfs_block_rsv *orig_rsv;
+	/* Set by a task that wants to create a snapshot. */
+	struct btrfs_pending_snapshot *pending_snapshot;
 	refcount_t use_count;
 	unsigned int type;
+	/*
+	 * Error code of transaction abort, set outside of locks and must use
+	 * the READ_ONCE/WRITE_ONCE access
+	 */
 	short aborted;
 	bool adding_csums;
 	bool allocating_chunk;
-	bool can_flush_pending_bgs;
+	bool removing_chunk;
 	bool reloc_reserved;
-	bool dirty;
-	struct btrfs_root *root;
+	bool in_fsync;
 	struct btrfs_fs_info *fs_info;
 	struct list_head new_bgs;
 };
 
+/*
+ * The abort status can be changed between calls and is not protected by locks.
+ * This accepts btrfs_transaction and btrfs_trans_handle as types. Once it's
+ * set to a non-zero value it does not change, so the macro should be in checks
+ * but is not necessary for further reads of the value.
+ */
+#define TRANS_ABORTED(trans)		(unlikely(READ_ONCE((trans)->aborted)))
+
 struct btrfs_pending_snapshot {
 	struct dentry *dentry;
 	struct inode *dir;
@@ -138,18 +161,20 @@ struct btrfs_pending_snapshot {
 	struct btrfs_block_rsv block_rsv;
 	/* extra metadata reservation for relocation */
 	int error;
+	/* Preallocated anonymous block device number */
+	dev_t anon_dev;
 	bool readonly;
 	struct list_head list;
 };
 
 static inline void btrfs_set_inode_last_trans(struct btrfs_trans_handle *trans,
-					      struct inode *inode)
+					      struct btrfs_inode *inode)
 {
-	spin_lock(&BTRFS_I(inode)->lock);
-	BTRFS_I(inode)->last_trans = trans->transaction->transid;
-	BTRFS_I(inode)->last_sub_trans = BTRFS_I(inode)->root->log_transid;
-	BTRFS_I(inode)->last_log_commit = BTRFS_I(inode)->root->last_log_commit;
-	spin_unlock(&BTRFS_I(inode)->lock);
+	spin_lock(&inode->lock);
+	inode->last_trans = trans->transaction->transid;
+	inode->last_sub_trans = inode->root->log_transid;
+	inode->last_log_commit = inode->last_sub_trans - 1;
+	spin_unlock(&inode->lock);
 }
 
 /*
@@ -180,8 +205,7 @@ struct btrfs_trans_handle *btrfs_start_transaction(struct btrfs_root *root,
 						   unsigned int num_items);
 struct btrfs_trans_handle *btrfs_start_transaction_fallback_global_rsv(
 					struct btrfs_root *root,
-					unsigned int num_items,
-					int min_factor);
+					unsigned int num_items);
 struct btrfs_trans_handle *btrfs_join_transaction(struct btrfs_root *root);
 struct btrfs_trans_handle *btrfs_join_transaction_spacecache(struct btrfs_root *root);
 struct btrfs_trans_handle *btrfs_join_transaction_nostart(struct btrfs_root *root);
@@ -192,26 +216,12 @@ int btrfs_wait_for_commit(struct btrfs_fs_info *fs_info, u64 transid);
 
 void btrfs_add_dead_root(struct btrfs_root *root);
 int btrfs_defrag_root(struct btrfs_root *root);
-int btrfs_clean_one_deleted_snapshot(struct btrfs_root *root);
+void btrfs_maybe_wake_unfinished_drop(struct btrfs_fs_info *fs_info);
+int btrfs_clean_one_deleted_snapshot(struct btrfs_fs_info *fs_info);
 int btrfs_commit_transaction(struct btrfs_trans_handle *trans);
-int btrfs_commit_transaction_async(struct btrfs_trans_handle *trans,
-				   int wait_for_unblock);
-
-/*
- * Try to commit transaction asynchronously, so this is safe to call
- * even holding a spinlock.
- *
- * It's done by informing transaction_kthread to commit transaction without
- * waiting for commit interval.
- */
-static inline void btrfs_commit_transaction_locksafe(
-		struct btrfs_fs_info *fs_info)
-{
-	set_bit(BTRFS_FS_NEED_ASYNC_COMMIT, &fs_info->flags);
-	wake_up_process(fs_info->transaction_kthread);
-}
+void btrfs_commit_transaction_async(struct btrfs_trans_handle *trans);
 int btrfs_end_transaction_throttle(struct btrfs_trans_handle *trans);
-int btrfs_should_end_transaction(struct btrfs_trans_handle *trans);
+bool btrfs_should_end_transaction(struct btrfs_trans_handle *trans);
 void btrfs_throttle(struct btrfs_fs_info *fs_info);
 int btrfs_record_root_in_trans(struct btrfs_trans_handle *trans,
 				struct btrfs_root *root);
diff --git a/fs/btrfs/tree-checker.c b/fs/btrfs/tree-checker.c
index a92f8a6dd192..43f905ab0a18 100644
--- a/fs/btrfs/tree-checker.c
+++ b/fs/btrfs/tree-checker.c
@@ -24,6 +24,7 @@
 #include "compression.h"
 #include "volumes.h"
 #include "misc.h"
+#include "btrfs_inode.h"
 
 /*
  * Error message should follow the following format:
@@ -100,7 +101,8 @@ static void file_extent_err(const struct extent_buffer *eb, int slot,
  */
 #define CHECK_FE_ALIGNED(leaf, slot, fi, name, alignment)		      \
 ({									      \
-	if (!IS_ALIGNED(btrfs_file_extent_##name((leaf), (fi)), (alignment))) \
+	if (unlikely(!IS_ALIGNED(btrfs_file_extent_##name((leaf), (fi)),      \
+				 (alignment))))				      \
 		file_extent_err((leaf), (slot),				      \
 	"invalid %s for file extent, have %llu, should be aligned to %u",     \
 			(#name), btrfs_file_extent_##name((leaf), (fi)),      \
@@ -200,10 +202,10 @@ static int check_extent_data_item(struct extent_buffer *leaf,
 	struct btrfs_fs_info *fs_info = leaf->fs_info;
 	struct btrfs_file_extent_item *fi;
 	u32 sectorsize = fs_info->sectorsize;
-	u32 item_size = btrfs_item_size_nr(leaf, slot);
+	u32 item_size = btrfs_item_size(leaf, slot);
 	u64 extent_end;
 
-	if (!IS_ALIGNED(key->offset, sectorsize)) {
+	if (unlikely(!IS_ALIGNED(key->offset, sectorsize))) {
 		file_extent_err(leaf, slot,
 "unaligned file_offset for file extent, have %llu should be aligned to %u",
 			key->offset, sectorsize);
@@ -216,7 +218,7 @@ static int check_extent_data_item(struct extent_buffer *leaf,
 	 * But if objectids mismatch, it means we have a missing
 	 * INODE_ITEM.
 	 */
-	if (!check_prev_ino(leaf, key, slot, prev_key))
+	if (unlikely(!check_prev_ino(leaf, key, slot, prev_key)))
 		return -EUCLEAN;
 
 	fi = btrfs_item_ptr(leaf, slot, struct btrfs_file_extent_item);
@@ -225,14 +227,15 @@ static int check_extent_data_item(struct extent_buffer *leaf,
 	 * Make sure the item contains at least inline header, so the file
 	 * extent type is not some garbage.
 	 */
-	if (item_size < BTRFS_FILE_EXTENT_INLINE_DATA_START) {
+	if (unlikely(item_size < BTRFS_FILE_EXTENT_INLINE_DATA_START)) {
 		file_extent_err(leaf, slot,
 				"invalid item size, have %u expect [%zu, %u)",
 				item_size, BTRFS_FILE_EXTENT_INLINE_DATA_START,
 				SZ_4K);
 		return -EUCLEAN;
 	}
-	if (btrfs_file_extent_type(leaf, fi) >= BTRFS_NR_FILE_EXTENT_TYPES) {
+	if (unlikely(btrfs_file_extent_type(leaf, fi) >=
+		     BTRFS_NR_FILE_EXTENT_TYPES)) {
 		file_extent_err(leaf, slot,
 		"invalid type for file extent, have %u expect range [0, %u]",
 			btrfs_file_extent_type(leaf, fi),
@@ -244,14 +247,15 @@ static int check_extent_data_item(struct extent_buffer *leaf,
 	 * Support for new compression/encryption must introduce incompat flag,
 	 * and must be caught in open_ctree().
 	 */
-	if (btrfs_file_extent_compression(leaf, fi) >= BTRFS_NR_COMPRESS_TYPES) {
+	if (unlikely(btrfs_file_extent_compression(leaf, fi) >=
+		     BTRFS_NR_COMPRESS_TYPES)) {
 		file_extent_err(leaf, slot,
 	"invalid compression for file extent, have %u expect range [0, %u]",
 			btrfs_file_extent_compression(leaf, fi),
 			BTRFS_NR_COMPRESS_TYPES - 1);
 		return -EUCLEAN;
 	}
-	if (btrfs_file_extent_encryption(leaf, fi)) {
+	if (unlikely(btrfs_file_extent_encryption(leaf, fi))) {
 		file_extent_err(leaf, slot,
 			"invalid encryption for file extent, have %u expect 0",
 			btrfs_file_extent_encryption(leaf, fi));
@@ -259,7 +263,7 @@ static int check_extent_data_item(struct extent_buffer *leaf,
 	}
 	if (btrfs_file_extent_type(leaf, fi) == BTRFS_FILE_EXTENT_INLINE) {
 		/* Inline extent must have 0 as key offset */
-		if (key->offset) {
+		if (unlikely(key->offset)) {
 			file_extent_err(leaf, slot,
 		"invalid file_offset for inline file extent, have %llu expect 0",
 				key->offset);
@@ -272,8 +276,8 @@ static int check_extent_data_item(struct extent_buffer *leaf,
 			return 0;
 
 		/* Uncompressed inline extent size must match item size */
-		if (item_size != BTRFS_FILE_EXTENT_INLINE_DATA_START +
-		    btrfs_file_extent_ram_bytes(leaf, fi)) {
+		if (unlikely(item_size != BTRFS_FILE_EXTENT_INLINE_DATA_START +
+					  btrfs_file_extent_ram_bytes(leaf, fi))) {
 			file_extent_err(leaf, slot,
 	"invalid ram_bytes for uncompressed inline extent, have %u expect %llu",
 				item_size, BTRFS_FILE_EXTENT_INLINE_DATA_START +
@@ -284,22 +288,22 @@ static int check_extent_data_item(struct extent_buffer *leaf,
 	}
 
 	/* Regular or preallocated extent has fixed item size */
-	if (item_size != sizeof(*fi)) {
+	if (unlikely(item_size != sizeof(*fi))) {
 		file_extent_err(leaf, slot,
 	"invalid item size for reg/prealloc file extent, have %u expect %zu",
 			item_size, sizeof(*fi));
 		return -EUCLEAN;
 	}
-	if (CHECK_FE_ALIGNED(leaf, slot, fi, ram_bytes, sectorsize) ||
-	    CHECK_FE_ALIGNED(leaf, slot, fi, disk_bytenr, sectorsize) ||
-	    CHECK_FE_ALIGNED(leaf, slot, fi, disk_num_bytes, sectorsize) ||
-	    CHECK_FE_ALIGNED(leaf, slot, fi, offset, sectorsize) ||
-	    CHECK_FE_ALIGNED(leaf, slot, fi, num_bytes, sectorsize))
+	if (unlikely(CHECK_FE_ALIGNED(leaf, slot, fi, ram_bytes, sectorsize) ||
+		     CHECK_FE_ALIGNED(leaf, slot, fi, disk_bytenr, sectorsize) ||
+		     CHECK_FE_ALIGNED(leaf, slot, fi, disk_num_bytes, sectorsize) ||
+		     CHECK_FE_ALIGNED(leaf, slot, fi, offset, sectorsize) ||
+		     CHECK_FE_ALIGNED(leaf, slot, fi, num_bytes, sectorsize)))
 		return -EUCLEAN;
 
 	/* Catch extent end overflow */
-	if (check_add_overflow(btrfs_file_extent_num_bytes(leaf, fi),
-			       key->offset, &extent_end)) {
+	if (unlikely(check_add_overflow(btrfs_file_extent_num_bytes(leaf, fi),
+					key->offset, &extent_end))) {
 		file_extent_err(leaf, slot,
 	"extent end overflow, have file offset %llu extent num bytes %llu",
 				key->offset,
@@ -320,7 +324,7 @@ static int check_extent_data_item(struct extent_buffer *leaf,
 		prev_fi = btrfs_item_ptr(leaf, slot - 1,
 					 struct btrfs_file_extent_item);
 		prev_end = file_extent_end(leaf, prev_key, prev_fi);
-		if (prev_end > key->offset) {
+		if (unlikely(prev_end > key->offset)) {
 			file_extent_err(leaf, slot - 1,
 "file extent end range (%llu) goes beyond start offset (%llu) of the next file extent",
 					prev_end, key->offset);
@@ -336,34 +340,34 @@ static int check_csum_item(struct extent_buffer *leaf, struct btrfs_key *key,
 {
 	struct btrfs_fs_info *fs_info = leaf->fs_info;
 	u32 sectorsize = fs_info->sectorsize;
-	u32 csumsize = btrfs_super_csum_size(fs_info->super_copy);
+	const u32 csumsize = fs_info->csum_size;
 
-	if (key->objectid != BTRFS_EXTENT_CSUM_OBJECTID) {
+	if (unlikely(key->objectid != BTRFS_EXTENT_CSUM_OBJECTID)) {
 		generic_err(leaf, slot,
 		"invalid key objectid for csum item, have %llu expect %llu",
 			key->objectid, BTRFS_EXTENT_CSUM_OBJECTID);
 		return -EUCLEAN;
 	}
-	if (!IS_ALIGNED(key->offset, sectorsize)) {
+	if (unlikely(!IS_ALIGNED(key->offset, sectorsize))) {
 		generic_err(leaf, slot,
 	"unaligned key offset for csum item, have %llu should be aligned to %u",
 			key->offset, sectorsize);
 		return -EUCLEAN;
 	}
-	if (!IS_ALIGNED(btrfs_item_size_nr(leaf, slot), csumsize)) {
+	if (unlikely(!IS_ALIGNED(btrfs_item_size(leaf, slot), csumsize))) {
 		generic_err(leaf, slot,
 	"unaligned item size for csum item, have %u should be aligned to %u",
-			btrfs_item_size_nr(leaf, slot), csumsize);
+			btrfs_item_size(leaf, slot), csumsize);
 		return -EUCLEAN;
 	}
 	if (slot > 0 && prev_key->type == BTRFS_EXTENT_CSUM_KEY) {
 		u64 prev_csum_end;
 		u32 prev_item_size;
 
-		prev_item_size = btrfs_item_size_nr(leaf, slot - 1);
+		prev_item_size = btrfs_item_size(leaf, slot - 1);
 		prev_csum_end = (prev_item_size / csumsize) * sectorsize;
 		prev_csum_end += prev_key->offset;
-		if (prev_csum_end > key->offset) {
+		if (unlikely(prev_csum_end > key->offset)) {
 			generic_err(leaf, slot - 1,
 "csum end range (%llu) goes beyond the start range (%llu) of the next csum item",
 				    prev_csum_end, key->offset);
@@ -388,15 +392,16 @@ static int check_inode_key(struct extent_buffer *leaf, struct btrfs_key *key,
 
 	/* For XATTR_ITEM, location key should be all 0 */
 	if (item_key.type == BTRFS_XATTR_ITEM_KEY) {
-		if (key->type != 0 || key->objectid != 0 || key->offset != 0)
+		if (unlikely(key->objectid != 0 || key->type != 0 ||
+			     key->offset != 0))
 			return -EUCLEAN;
 		return 0;
 	}
 
-	if ((key->objectid < BTRFS_FIRST_FREE_OBJECTID ||
-	     key->objectid > BTRFS_LAST_FREE_OBJECTID) &&
-	    key->objectid != BTRFS_ROOT_TREE_DIR_OBJECTID &&
-	    key->objectid != BTRFS_FREE_INO_OBJECTID) {
+	if (unlikely((key->objectid < BTRFS_FIRST_FREE_OBJECTID ||
+		      key->objectid > BTRFS_LAST_FREE_OBJECTID) &&
+		     key->objectid != BTRFS_ROOT_TREE_DIR_OBJECTID &&
+		     key->objectid != BTRFS_FREE_INO_OBJECTID)) {
 		if (is_inode_item) {
 			generic_err(leaf, slot,
 	"invalid key objectid: has %llu expect %llu or [%llu, %llu] or %llu",
@@ -414,7 +419,7 @@ static int check_inode_key(struct extent_buffer *leaf, struct btrfs_key *key,
 		}
 		return -EUCLEAN;
 	}
-	if (key->offset != 0) {
+	if (unlikely(key->offset != 0)) {
 		if (is_inode_item)
 			inode_item_err(leaf, slot,
 				       "invalid key offset: has %llu expect 0",
@@ -438,7 +443,7 @@ static int check_root_key(struct extent_buffer *leaf, struct btrfs_key *key,
 	is_root_item = (item_key.type == BTRFS_ROOT_ITEM_KEY);
 
 	/* No such tree id */
-	if (key->objectid == 0) {
+	if (unlikely(key->objectid == 0)) {
 		if (is_root_item)
 			generic_err(leaf, slot, "invalid root id 0");
 		else
@@ -448,7 +453,7 @@ static int check_root_key(struct extent_buffer *leaf, struct btrfs_key *key,
 	}
 
 	/* DIR_ITEM/INDEX/INODE_REF is not allowed to point to non-fs trees */
-	if (!is_fstree(key->objectid) && !is_root_item) {
+	if (unlikely(!is_fstree(key->objectid) && !is_root_item)) {
 		dir_item_err(leaf, slot,
 		"invalid location key objectid, have %llu expect [%llu, %llu]",
 				key->objectid, BTRFS_FIRST_FREE_OBJECTID,
@@ -464,7 +469,8 @@ static int check_root_key(struct extent_buffer *leaf, struct btrfs_key *key,
 	 * So here we only check offset for reloc tree whose key->offset must
 	 * be a valid tree.
 	 */
-	if (key->objectid == BTRFS_TREE_RELOC_OBJECTID && key->offset == 0) {
+	if (unlikely(key->objectid == BTRFS_TREE_RELOC_OBJECTID &&
+		     key->offset == 0)) {
 		generic_err(leaf, slot, "invalid root id 0 for reloc tree");
 		return -EUCLEAN;
 	}
@@ -477,11 +483,12 @@ static int check_dir_item(struct extent_buffer *leaf,
 {
 	struct btrfs_fs_info *fs_info = leaf->fs_info;
 	struct btrfs_dir_item *di;
-	u32 item_size = btrfs_item_size_nr(leaf, slot);
+	u32 item_size = btrfs_item_size(leaf, slot);
 	u32 cur = 0;
 
-	if (!check_prev_ino(leaf, key, slot, prev_key))
+	if (unlikely(!check_prev_ino(leaf, key, slot, prev_key)))
 		return -EUCLEAN;
+
 	di = btrfs_item_ptr(leaf, slot, struct btrfs_dir_item);
 	while (cur < item_size) {
 		struct btrfs_key location_key;
@@ -494,7 +501,7 @@ static int check_dir_item(struct extent_buffer *leaf,
 		int ret;
 
 		/* header itself should not cross item boundary */
-		if (cur + sizeof(*di) > item_size) {
+		if (unlikely(cur + sizeof(*di) > item_size)) {
 			dir_item_err(leaf, slot,
 		"dir item header crosses item boundary, have %zu boundary %u",
 				cur + sizeof(*di), item_size);
@@ -505,12 +512,12 @@ static int check_dir_item(struct extent_buffer *leaf,
 		btrfs_dir_item_key_to_cpu(leaf, di, &location_key);
 		if (location_key.type == BTRFS_ROOT_ITEM_KEY) {
 			ret = check_root_key(leaf, &location_key, slot);
-			if (ret < 0)
+			if (unlikely(ret < 0))
 				return ret;
 		} else if (location_key.type == BTRFS_INODE_ITEM_KEY ||
 			   location_key.type == 0) {
 			ret = check_inode_key(leaf, &location_key, slot);
-			if (ret < 0)
+			if (unlikely(ret < 0))
 				return ret;
 		} else {
 			dir_item_err(leaf, slot,
@@ -522,22 +529,22 @@ static int check_dir_item(struct extent_buffer *leaf,
 
 		/* dir type check */
 		dir_type = btrfs_dir_type(leaf, di);
-		if (dir_type >= BTRFS_FT_MAX) {
+		if (unlikely(dir_type >= BTRFS_FT_MAX)) {
 			dir_item_err(leaf, slot,
 			"invalid dir item type, have %u expect [0, %u)",
 				dir_type, BTRFS_FT_MAX);
 			return -EUCLEAN;
 		}
 
-		if (key->type == BTRFS_XATTR_ITEM_KEY &&
-		    dir_type != BTRFS_FT_XATTR) {
+		if (unlikely(key->type == BTRFS_XATTR_ITEM_KEY &&
+			     dir_type != BTRFS_FT_XATTR)) {
 			dir_item_err(leaf, slot,
 		"invalid dir item type for XATTR key, have %u expect %u",
 				dir_type, BTRFS_FT_XATTR);
 			return -EUCLEAN;
 		}
-		if (dir_type == BTRFS_FT_XATTR &&
-		    key->type != BTRFS_XATTR_ITEM_KEY) {
+		if (unlikely(dir_type == BTRFS_FT_XATTR &&
+			     key->type != BTRFS_XATTR_ITEM_KEY)) {
 			dir_item_err(leaf, slot,
 			"xattr dir type found for non-XATTR key");
 			return -EUCLEAN;
@@ -550,13 +557,13 @@ static int check_dir_item(struct extent_buffer *leaf,
 		/* Name/data length check */
 		name_len = btrfs_dir_name_len(leaf, di);
 		data_len = btrfs_dir_data_len(leaf, di);
-		if (name_len > max_name_len) {
+		if (unlikely(name_len > max_name_len)) {
 			dir_item_err(leaf, slot,
 			"dir item name len too long, have %u max %u",
 				name_len, max_name_len);
 			return -EUCLEAN;
 		}
-		if (name_len + data_len > BTRFS_MAX_XATTR_SIZE(fs_info)) {
+		if (unlikely(name_len + data_len > BTRFS_MAX_XATTR_SIZE(fs_info))) {
 			dir_item_err(leaf, slot,
 			"dir item name and data len too long, have %u max %u",
 				name_len + data_len,
@@ -564,7 +571,7 @@ static int check_dir_item(struct extent_buffer *leaf,
 			return -EUCLEAN;
 		}
 
-		if (data_len && dir_type != BTRFS_FT_XATTR) {
+		if (unlikely(data_len && dir_type != BTRFS_FT_XATTR)) {
 			dir_item_err(leaf, slot,
 			"dir item with invalid data len, have %u expect 0",
 				data_len);
@@ -574,7 +581,7 @@ static int check_dir_item(struct extent_buffer *leaf,
 		total_size = sizeof(*di) + name_len + data_len;
 
 		/* header and name/data should not cross item boundary */
-		if (cur + total_size > item_size) {
+		if (unlikely(cur + total_size > item_size)) {
 			dir_item_err(leaf, slot,
 		"dir item data crosses item boundary, have %u boundary %u",
 				cur + total_size, item_size);
@@ -592,7 +599,7 @@ static int check_dir_item(struct extent_buffer *leaf,
 			read_extent_buffer(leaf, namebuf,
 					(unsigned long)(di + 1), name_len);
 			name_hash = btrfs_name_hash(namebuf, name_len);
-			if (key->offset != name_hash) {
+			if (unlikely(key->offset != name_hash)) {
 				dir_item_err(leaf, slot,
 		"name hash mismatch with key, have 0x%016x expect 0x%016llx",
 					name_hash, key->offset);
@@ -632,8 +639,10 @@ static void block_group_err(const struct extent_buffer *eb, int slot,
 static int check_block_group_item(struct extent_buffer *leaf,
 				  struct btrfs_key *key, int slot)
 {
+	struct btrfs_fs_info *fs_info = leaf->fs_info;
 	struct btrfs_block_group_item bgi;
-	u32 item_size = btrfs_item_size_nr(leaf, slot);
+	u32 item_size = btrfs_item_size(leaf, slot);
+	u64 chunk_objectid;
 	u64 flags;
 	u64 type;
 
@@ -641,13 +650,13 @@ static int check_block_group_item(struct extent_buffer *leaf,
 	 * Here we don't really care about alignment since extent allocator can
 	 * handle it.  We care more about the size.
 	 */
-	if (key->offset == 0) {
+	if (unlikely(key->offset == 0)) {
 		block_group_err(leaf, slot,
 				"invalid block group size 0");
 		return -EUCLEAN;
 	}
 
-	if (item_size != sizeof(bgi)) {
+	if (unlikely(item_size != sizeof(bgi))) {
 		block_group_err(leaf, slot,
 			"invalid item size, have %u expect %zu",
 				item_size, sizeof(bgi));
@@ -656,8 +665,23 @@ static int check_block_group_item(struct extent_buffer *leaf,
 
 	read_extent_buffer(leaf, &bgi, btrfs_item_ptr_offset(leaf, slot),
 			   sizeof(bgi));
-	if (btrfs_stack_block_group_chunk_objectid(&bgi) !=
-	    BTRFS_FIRST_CHUNK_TREE_OBJECTID) {
+	chunk_objectid = btrfs_stack_block_group_chunk_objectid(&bgi);
+	if (btrfs_fs_incompat(fs_info, EXTENT_TREE_V2)) {
+		/*
+		 * We don't init the nr_global_roots until we load the global
+		 * roots, so this could be 0 at mount time.  If it's 0 we'll
+		 * just assume we're fine, and later we'll check against our
+		 * actual value.
+		 */
+		if (unlikely(fs_info->nr_global_roots &&
+			     chunk_objectid >= fs_info->nr_global_roots)) {
+			block_group_err(leaf, slot,
+	"invalid block group global root id, have %llu, needs to be <= %llu",
+					chunk_objectid,
+					fs_info->nr_global_roots);
+			return -EUCLEAN;
+		}
+	} else if (unlikely(chunk_objectid != BTRFS_FIRST_CHUNK_TREE_OBJECTID)) {
 		block_group_err(leaf, slot,
 		"invalid block group chunk objectid, have %llu expect %llu",
 				btrfs_stack_block_group_chunk_objectid(&bgi),
@@ -665,7 +689,7 @@ static int check_block_group_item(struct extent_buffer *leaf,
 		return -EUCLEAN;
 	}
 
-	if (btrfs_stack_block_group_used(&bgi) > key->offset) {
+	if (unlikely(btrfs_stack_block_group_used(&bgi) > key->offset)) {
 		block_group_err(leaf, slot,
 			"invalid block group used, have %llu expect [0, %llu)",
 				btrfs_stack_block_group_used(&bgi), key->offset);
@@ -673,7 +697,7 @@ static int check_block_group_item(struct extent_buffer *leaf,
 	}
 
 	flags = btrfs_stack_block_group_flags(&bgi);
-	if (hweight64(flags & BTRFS_BLOCK_GROUP_PROFILE_MASK) > 1) {
+	if (unlikely(hweight64(flags & BTRFS_BLOCK_GROUP_PROFILE_MASK) > 1)) {
 		block_group_err(leaf, slot,
 "invalid profile flags, have 0x%llx (%lu bits set) expect no more than 1 bit set",
 			flags & BTRFS_BLOCK_GROUP_PROFILE_MASK,
@@ -682,11 +706,11 @@ static int check_block_group_item(struct extent_buffer *leaf,
 	}
 
 	type = flags & BTRFS_BLOCK_GROUP_TYPE_MASK;
-	if (type != BTRFS_BLOCK_GROUP_DATA &&
-	    type != BTRFS_BLOCK_GROUP_METADATA &&
-	    type != BTRFS_BLOCK_GROUP_SYSTEM &&
-	    type != (BTRFS_BLOCK_GROUP_METADATA |
-			   BTRFS_BLOCK_GROUP_DATA)) {
+	if (unlikely(type != BTRFS_BLOCK_GROUP_DATA &&
+		     type != BTRFS_BLOCK_GROUP_METADATA &&
+		     type != BTRFS_BLOCK_GROUP_SYSTEM &&
+		     type != (BTRFS_BLOCK_GROUP_METADATA |
+			      BTRFS_BLOCK_GROUP_DATA))) {
 		block_group_err(leaf, slot,
 "invalid type, have 0x%llx (%lu bits set) expect either 0x%llx, 0x%llx, 0x%llx or 0x%llx",
 			type, hweight64(type),
@@ -754,50 +778,75 @@ int btrfs_check_chunk_valid(struct extent_buffer *leaf,
 {
 	struct btrfs_fs_info *fs_info = leaf->fs_info;
 	u64 length;
+	u64 chunk_end;
 	u64 stripe_len;
 	u16 num_stripes;
 	u16 sub_stripes;
 	u64 type;
 	u64 features;
 	bool mixed = false;
+	int raid_index;
+	int nparity;
+	int ncopies;
 
 	length = btrfs_chunk_length(leaf, chunk);
 	stripe_len = btrfs_chunk_stripe_len(leaf, chunk);
 	num_stripes = btrfs_chunk_num_stripes(leaf, chunk);
 	sub_stripes = btrfs_chunk_sub_stripes(leaf, chunk);
 	type = btrfs_chunk_type(leaf, chunk);
+	raid_index = btrfs_bg_flags_to_raid_index(type);
+	ncopies = btrfs_raid_array[raid_index].ncopies;
+	nparity = btrfs_raid_array[raid_index].nparity;
 
-	if (!num_stripes) {
+	if (unlikely(!num_stripes)) {
 		chunk_err(leaf, chunk, logical,
 			  "invalid chunk num_stripes, have %u", num_stripes);
 		return -EUCLEAN;
 	}
-	if (!IS_ALIGNED(logical, fs_info->sectorsize)) {
+	if (unlikely(num_stripes < ncopies)) {
+		chunk_err(leaf, chunk, logical,
+			  "invalid chunk num_stripes < ncopies, have %u < %d",
+			  num_stripes, ncopies);
+		return -EUCLEAN;
+	}
+	if (unlikely(nparity && num_stripes == nparity)) {
+		chunk_err(leaf, chunk, logical,
+			  "invalid chunk num_stripes == nparity, have %u == %d",
+			  num_stripes, nparity);
+		return -EUCLEAN;
+	}
+	if (unlikely(!IS_ALIGNED(logical, fs_info->sectorsize))) {
 		chunk_err(leaf, chunk, logical,
 		"invalid chunk logical, have %llu should aligned to %u",
 			  logical, fs_info->sectorsize);
 		return -EUCLEAN;
 	}
-	if (btrfs_chunk_sector_size(leaf, chunk) != fs_info->sectorsize) {
+	if (unlikely(btrfs_chunk_sector_size(leaf, chunk) != fs_info->sectorsize)) {
 		chunk_err(leaf, chunk, logical,
 			  "invalid chunk sectorsize, have %u expect %u",
 			  btrfs_chunk_sector_size(leaf, chunk),
 			  fs_info->sectorsize);
 		return -EUCLEAN;
 	}
-	if (!length || !IS_ALIGNED(length, fs_info->sectorsize)) {
+	if (unlikely(!length || !IS_ALIGNED(length, fs_info->sectorsize))) {
 		chunk_err(leaf, chunk, logical,
 			  "invalid chunk length, have %llu", length);
 		return -EUCLEAN;
 	}
-	if (!is_power_of_2(stripe_len) || stripe_len != BTRFS_STRIPE_LEN) {
+	if (unlikely(check_add_overflow(logical, length, &chunk_end))) {
+		chunk_err(leaf, chunk, logical,
+"invalid chunk logical start and length, have logical start %llu length %llu",
+			  logical, length);
+		return -EUCLEAN;
+	}
+	if (unlikely(!is_power_of_2(stripe_len) || stripe_len != BTRFS_STRIPE_LEN)) {
 		chunk_err(leaf, chunk, logical,
 			  "invalid chunk stripe length: %llu",
 			  stripe_len);
 		return -EUCLEAN;
 	}
-	if (~(BTRFS_BLOCK_GROUP_TYPE_MASK | BTRFS_BLOCK_GROUP_PROFILE_MASK) &
-	    type) {
+	if (unlikely(type & ~(BTRFS_BLOCK_GROUP_TYPE_MASK |
+			      BTRFS_BLOCK_GROUP_PROFILE_MASK))) {
 		chunk_err(leaf, chunk, logical,
 			  "unrecognized chunk type: 0x%llx",
 			  ~(BTRFS_BLOCK_GROUP_TYPE_MASK |
@@ -806,22 +855,23 @@ int btrfs_check_chunk_valid(struct extent_buffer *leaf,
 		return -EUCLEAN;
 	}
 
-	if (!has_single_bit_set(type & BTRFS_BLOCK_GROUP_PROFILE_MASK) &&
-	    (type & BTRFS_BLOCK_GROUP_PROFILE_MASK) != 0) {
+	if (unlikely(!has_single_bit_set(type & BTRFS_BLOCK_GROUP_PROFILE_MASK) &&
+		     (type & BTRFS_BLOCK_GROUP_PROFILE_MASK) != 0)) {
 		chunk_err(leaf, chunk, logical,
 		"invalid chunk profile flag: 0x%llx, expect 0 or 1 bit set",
 			  type & BTRFS_BLOCK_GROUP_PROFILE_MASK);
 		return -EUCLEAN;
 	}
-	if ((type & BTRFS_BLOCK_GROUP_TYPE_MASK) == 0) {
+	if (unlikely((type & BTRFS_BLOCK_GROUP_TYPE_MASK) == 0)) {
 		chunk_err(leaf, chunk, logical,
 	"missing chunk type flag, have 0x%llx one bit must be set in 0x%llx",
 			  type, BTRFS_BLOCK_GROUP_TYPE_MASK);
 		return -EUCLEAN;
 	}
 
-	if ((type & BTRFS_BLOCK_GROUP_SYSTEM) &&
-	    (type & (BTRFS_BLOCK_GROUP_METADATA | BTRFS_BLOCK_GROUP_DATA))) {
+	if (unlikely((type & BTRFS_BLOCK_GROUP_SYSTEM) &&
+		     (type & (BTRFS_BLOCK_GROUP_METADATA |
+			      BTRFS_BLOCK_GROUP_DATA)))) {
 		chunk_err(leaf, chunk, logical,
 			  "system chunk with data or metadata type: 0x%llx",
 			  type);
@@ -833,20 +883,30 @@ int btrfs_check_chunk_valid(struct extent_buffer *leaf,
 		mixed = true;
 
 	if (!mixed) {
-		if ((type & BTRFS_BLOCK_GROUP_METADATA) &&
-		    (type & BTRFS_BLOCK_GROUP_DATA)) {
+		if (unlikely((type & BTRFS_BLOCK_GROUP_METADATA) &&
+			     (type & BTRFS_BLOCK_GROUP_DATA))) {
 			chunk_err(leaf, chunk, logical,
 			"mixed chunk type in non-mixed mode: 0x%llx", type);
 			return -EUCLEAN;
 		}
 	}
 
-	if ((type & BTRFS_BLOCK_GROUP_RAID10 && sub_stripes != 2) ||
-	    (type & BTRFS_BLOCK_GROUP_RAID1 && num_stripes != 2) ||
-	    (type & BTRFS_BLOCK_GROUP_RAID5 && num_stripes < 2) ||
-	    (type & BTRFS_BLOCK_GROUP_RAID6 && num_stripes < 3) ||
-	    (type & BTRFS_BLOCK_GROUP_DUP && num_stripes != 2) ||
-	    ((type & BTRFS_BLOCK_GROUP_PROFILE_MASK) == 0 && num_stripes != 1)) {
+	if (unlikely((type & BTRFS_BLOCK_GROUP_RAID10 &&
+		      sub_stripes != btrfs_raid_array[BTRFS_RAID_RAID10].sub_stripes) ||
+		     (type & BTRFS_BLOCK_GROUP_RAID1 &&
+		      num_stripes != btrfs_raid_array[BTRFS_RAID_RAID1].devs_min) ||
+		     (type & BTRFS_BLOCK_GROUP_RAID1C3 &&
+		      num_stripes != btrfs_raid_array[BTRFS_RAID_RAID1C3].devs_min) ||
+		     (type & BTRFS_BLOCK_GROUP_RAID1C4 &&
+		      num_stripes != btrfs_raid_array[BTRFS_RAID_RAID1C4].devs_min) ||
+		     (type & BTRFS_BLOCK_GROUP_RAID5 &&
+		      num_stripes < btrfs_raid_array[BTRFS_RAID_RAID5].devs_min) ||
+		     (type & BTRFS_BLOCK_GROUP_RAID6 &&
+		      num_stripes < btrfs_raid_array[BTRFS_RAID_RAID6].devs_min) ||
+		     (type & BTRFS_BLOCK_GROUP_DUP &&
+		      num_stripes != btrfs_raid_array[BTRFS_RAID_DUP].dev_stripes) ||
+		     ((type & BTRFS_BLOCK_GROUP_PROFILE_MASK) == 0 &&
+		      num_stripes != btrfs_raid_array[BTRFS_RAID_SINGLE].dev_stripes))) {
 		chunk_err(leaf, chunk, logical,
 			"invalid num_stripes:sub_stripes %u:%u for profile %llu",
 			num_stripes, sub_stripes,
@@ -869,10 +929,10 @@ static int check_leaf_chunk_item(struct extent_buffer *leaf,
 {
 	int num_stripes;
 
-	if (btrfs_item_size_nr(leaf, slot) < sizeof(struct btrfs_chunk)) {
+	if (unlikely(btrfs_item_size(leaf, slot) < sizeof(struct btrfs_chunk))) {
 		chunk_err(leaf, chunk, key->offset,
 			"invalid chunk item size: have %u expect [%zu, %u)",
-			btrfs_item_size_nr(leaf, slot),
+			btrfs_item_size(leaf, slot),
 			sizeof(struct btrfs_chunk),
 			BTRFS_LEAF_DATA_SIZE(leaf->fs_info));
 		return -EUCLEAN;
@@ -883,11 +943,11 @@ static int check_leaf_chunk_item(struct extent_buffer *leaf,
 	if (num_stripes == 0)
 		goto out;
 
-	if (btrfs_chunk_item_size(num_stripes) !=
-	    btrfs_item_size_nr(leaf, slot)) {
+	if (unlikely(btrfs_chunk_item_size(num_stripes) !=
+		     btrfs_item_size(leaf, slot))) {
 		chunk_err(leaf, chunk, key->offset,
 			"invalid chunk item size: have %u expect %lu",
-			btrfs_item_size_nr(leaf, slot),
+			btrfs_item_size(leaf, slot),
 			btrfs_chunk_item_size(num_stripes));
 		return -EUCLEAN;
 	}
@@ -922,15 +982,23 @@ static int check_dev_item(struct extent_buffer *leaf,
 			  struct btrfs_key *key, int slot)
 {
 	struct btrfs_dev_item *ditem;
+	const u32 item_size = btrfs_item_size(leaf, slot);
 
-	if (key->objectid != BTRFS_DEV_ITEMS_OBJECTID) {
+	if (unlikely(key->objectid != BTRFS_DEV_ITEMS_OBJECTID)) {
 		dev_item_err(leaf, slot,
 			     "invalid objectid: has=%llu expect=%llu",
 			     key->objectid, BTRFS_DEV_ITEMS_OBJECTID);
 		return -EUCLEAN;
 	}
+
+	if (unlikely(item_size != sizeof(*ditem))) {
+		dev_item_err(leaf, slot, "invalid item size: has %u expect %zu",
+			     item_size, sizeof(*ditem));
+		return -EUCLEAN;
+	}
+
 	ditem = btrfs_item_ptr(leaf, slot, struct btrfs_dev_item);
-	if (btrfs_device_id(leaf, ditem) != key->offset) {
+	if (unlikely(btrfs_device_id(leaf, ditem) != key->offset)) {
 		dev_item_err(leaf, slot,
 			     "devid mismatch: key has=%llu item has=%llu",
 			     key->offset, btrfs_device_id(leaf, ditem));
@@ -942,8 +1010,8 @@ static int check_dev_item(struct extent_buffer *leaf,
 	 * it can be 0 for device removal. Device size check can only be done
 	 * by dev extents check.
 	 */
-	if (btrfs_device_bytes_used(leaf, ditem) >
-	    btrfs_device_total_bytes(leaf, ditem)) {
+	if (unlikely(btrfs_device_bytes_used(leaf, ditem) >
+		     btrfs_device_total_bytes(leaf, ditem))) {
 		dev_item_err(leaf, slot,
 			     "invalid bytes used: have %llu expect [0, %llu]",
 			     btrfs_device_bytes_used(leaf, ditem),
@@ -957,10 +1025,6 @@ static int check_dev_item(struct extent_buffer *leaf,
 	return 0;
 }
 
-/* Inode item error output has the same format as dir_item_err() */
-#define inode_item_err(eb, slot, fmt, ...)			\
-	dir_item_err(eb, slot, fmt, __VA_ARGS__)
-
 static int check_inode_item(struct extent_buffer *leaf,
 			    struct btrfs_key *key, int slot)
 {
@@ -968,17 +1032,26 @@ static int check_inode_item(struct extent_buffer *leaf,
 	struct btrfs_inode_item *iitem;
 	u64 super_gen = btrfs_super_generation(fs_info->super_copy);
 	u32 valid_mask = (S_IFMT | S_ISUID | S_ISGID | S_ISVTX | 0777);
+	const u32 item_size = btrfs_item_size(leaf, slot);
 	u32 mode;
 	int ret;
+	u32 flags;
+	u32 ro_flags;
 
 	ret = check_inode_key(leaf, key, slot);
-	if (ret < 0)
+	if (unlikely(ret < 0))
 		return ret;
 
+	if (unlikely(item_size != sizeof(*iitem))) {
+		generic_err(leaf, slot, "invalid item size: has %u expect %zu",
+			    item_size, sizeof(*iitem));
+		return -EUCLEAN;
+	}
+
 	iitem = btrfs_item_ptr(leaf, slot, struct btrfs_inode_item);
 
 	/* Here we use super block generation + 1 to handle log tree */
-	if (btrfs_inode_generation(leaf, iitem) > super_gen + 1) {
+	if (unlikely(btrfs_inode_generation(leaf, iitem) > super_gen + 1)) {
 		inode_item_err(leaf, slot,
 			"invalid inode generation: has %llu expect (0, %llu]",
 			       btrfs_inode_generation(leaf, iitem),
@@ -986,9 +1059,9 @@ static int check_inode_item(struct extent_buffer *leaf,
 		return -EUCLEAN;
 	}
 	/* Note for ROOT_TREE_DIR_ITEM, mkfs could set its transid 0 */
-	if (btrfs_inode_transid(leaf, iitem) > super_gen + 1) {
+	if (unlikely(btrfs_inode_transid(leaf, iitem) > super_gen + 1)) {
 		inode_item_err(leaf, slot,
-			"invalid inode generation: has %llu expect [0, %llu]",
+			"invalid inode transid: has %llu expect [0, %llu]",
 			       btrfs_inode_transid(leaf, iitem), super_gen + 1);
 		return -EUCLEAN;
 	}
@@ -999,7 +1072,7 @@ static int check_inode_item(struct extent_buffer *leaf,
 	 * anything in the fs. So here we skip the check.
 	 */
 	mode = btrfs_inode_mode(leaf, iitem);
-	if (mode & ~valid_mask) {
+	if (unlikely(mode & ~valid_mask)) {
 		inode_item_err(leaf, slot,
 			       "unknown mode bit detected: 0x%x",
 			       mode & ~valid_mask);
@@ -1012,24 +1085,30 @@ static int check_inode_item(struct extent_buffer *leaf,
 	 * FIFO/CHR/DIR/REG.  Only needs to check BLK, LNK and SOCKS
 	 */
 	if (!has_single_bit_set(mode & S_IFMT)) {
-		if (!S_ISLNK(mode) && !S_ISBLK(mode) && !S_ISSOCK(mode)) {
+		if (unlikely(!S_ISLNK(mode) && !S_ISBLK(mode) && !S_ISSOCK(mode))) {
 			inode_item_err(leaf, slot,
 			"invalid mode: has 0%o expect valid S_IF* bit(s)",
 				       mode & S_IFMT);
 			return -EUCLEAN;
 		}
 	}
-	if (S_ISDIR(mode) && btrfs_inode_nlink(leaf, iitem) > 1) {
+	if (unlikely(S_ISDIR(mode) && btrfs_inode_nlink(leaf, iitem) > 1)) {
 		inode_item_err(leaf, slot,
 		       "invalid nlink: has %u expect no more than 1 for dir",
 			btrfs_inode_nlink(leaf, iitem));
 		return -EUCLEAN;
 	}
-	if (btrfs_inode_flags(leaf, iitem) & ~BTRFS_INODE_FLAG_MASK) {
+	btrfs_inode_split_flags(btrfs_inode_flags(leaf, iitem), &flags, &ro_flags);
+	if (unlikely(flags & ~BTRFS_INODE_FLAG_MASK)) {
 		inode_item_err(leaf, slot,
-			       "unknown flags detected: 0x%llx",
-			       btrfs_inode_flags(leaf, iitem) &
-			       ~BTRFS_INODE_FLAG_MASK);
+			       "unknown incompat flags detected: 0x%x", flags);
+		return -EUCLEAN;
+	}
+	if (unlikely(!sb_rdonly(fs_info->sb) &&
+		     (ro_flags & ~BTRFS_INODE_RO_FLAG_MASK))) {
+		inode_item_err(leaf, slot,
+			"unknown ro-compat flags detected on writeable mount: 0x%x",
+			ro_flags);
 		return -EUCLEAN;
 	}
 	return 0;
@@ -1039,43 +1118,52 @@ static int check_root_item(struct extent_buffer *leaf, struct btrfs_key *key,
 			   int slot)
 {
 	struct btrfs_fs_info *fs_info = leaf->fs_info;
-	struct btrfs_root_item ri;
+	struct btrfs_root_item ri = { 0 };
 	const u64 valid_root_flags = BTRFS_ROOT_SUBVOL_RDONLY |
 				     BTRFS_ROOT_SUBVOL_DEAD;
 	int ret;
 
 	ret = check_root_key(leaf, key, slot);
-	if (ret < 0)
+	if (unlikely(ret < 0))
 		return ret;
 
-	if (btrfs_item_size_nr(leaf, slot) != sizeof(ri)) {
+	if (unlikely(btrfs_item_size(leaf, slot) != sizeof(ri) &&
+		     btrfs_item_size(leaf, slot) !=
+		     btrfs_legacy_root_item_size())) {
 		generic_err(leaf, slot,
-			    "invalid root item size, have %u expect %zu",
-			    btrfs_item_size_nr(leaf, slot), sizeof(ri));
+			    "invalid root item size, have %u expect %zu or %u",
+			    btrfs_item_size(leaf, slot), sizeof(ri),
+			    btrfs_legacy_root_item_size());
+		return -EUCLEAN;
 	}
 
+	/*
+	 * For legacy root item, the members starting at generation_v2 will be
+	 * all filled with 0.
+	 * And since we allow geneartion_v2 as 0, it will still pass the check.
+	 */
 	read_extent_buffer(leaf, &ri, btrfs_item_ptr_offset(leaf, slot),
-			   sizeof(ri));
+			   btrfs_item_size(leaf, slot));
 
 	/* Generation related */
-	if (btrfs_root_generation(&ri) >
-	    btrfs_super_generation(fs_info->super_copy) + 1) {
+	if (unlikely(btrfs_root_generation(&ri) >
+		     btrfs_super_generation(fs_info->super_copy) + 1)) {
 		generic_err(leaf, slot,
 			"invalid root generation, have %llu expect (0, %llu]",
 			    btrfs_root_generation(&ri),
 			    btrfs_super_generation(fs_info->super_copy) + 1);
 		return -EUCLEAN;
 	}
-	if (btrfs_root_generation_v2(&ri) >
-	    btrfs_super_generation(fs_info->super_copy) + 1) {
+	if (unlikely(btrfs_root_generation_v2(&ri) >
+		     btrfs_super_generation(fs_info->super_copy) + 1)) {
 		generic_err(leaf, slot,
 		"invalid root v2 generation, have %llu expect (0, %llu]",
 			    btrfs_root_generation_v2(&ri),
 			    btrfs_super_generation(fs_info->super_copy) + 1);
 		return -EUCLEAN;
 	}
-	if (btrfs_root_last_snapshot(&ri) >
-	    btrfs_super_generation(fs_info->super_copy) + 1) {
+	if (unlikely(btrfs_root_last_snapshot(&ri) >
+		     btrfs_super_generation(fs_info->super_copy) + 1)) {
 		generic_err(leaf, slot,
 		"invalid root last_snapshot, have %llu expect (0, %llu]",
 			    btrfs_root_last_snapshot(&ri),
@@ -1084,27 +1172,27 @@ static int check_root_item(struct extent_buffer *leaf, struct btrfs_key *key,
 	}
 
 	/* Alignment and level check */
-	if (!IS_ALIGNED(btrfs_root_bytenr(&ri), fs_info->sectorsize)) {
+	if (unlikely(!IS_ALIGNED(btrfs_root_bytenr(&ri), fs_info->sectorsize))) {
 		generic_err(leaf, slot,
 		"invalid root bytenr, have %llu expect to be aligned to %u",
 			    btrfs_root_bytenr(&ri), fs_info->sectorsize);
 		return -EUCLEAN;
 	}
-	if (btrfs_root_level(&ri) >= BTRFS_MAX_LEVEL) {
+	if (unlikely(btrfs_root_level(&ri) >= BTRFS_MAX_LEVEL)) {
 		generic_err(leaf, slot,
 			    "invalid root level, have %u expect [0, %u]",
 			    btrfs_root_level(&ri), BTRFS_MAX_LEVEL - 1);
 		return -EUCLEAN;
 	}
-	if (ri.drop_level >= BTRFS_MAX_LEVEL) {
+	if (unlikely(btrfs_root_drop_level(&ri) >= BTRFS_MAX_LEVEL)) {
 		generic_err(leaf, slot,
 			    "invalid root level, have %u expect [0, %u]",
-			    ri.drop_level, BTRFS_MAX_LEVEL - 1);
+			    btrfs_root_drop_level(&ri), BTRFS_MAX_LEVEL - 1);
 		return -EUCLEAN;
 	}
 
 	/* Flags check */
-	if (btrfs_root_flags(&ri) & ~valid_root_flags) {
+	if (unlikely(btrfs_root_flags(&ri) & ~valid_root_flags)) {
 		generic_err(leaf, slot,
 			    "invalid root flags, have 0x%llx expect mask 0x%llx",
 			    btrfs_root_flags(&ri), valid_root_flags);
@@ -1145,27 +1233,28 @@ static void extent_err(const struct extent_buffer *eb, int slot,
 }
 
 static int check_extent_item(struct extent_buffer *leaf,
-			     struct btrfs_key *key, int slot)
+			     struct btrfs_key *key, int slot,
+			     struct btrfs_key *prev_key)
 {
 	struct btrfs_fs_info *fs_info = leaf->fs_info;
 	struct btrfs_extent_item *ei;
 	bool is_tree_block = false;
 	unsigned long ptr;	/* Current pointer inside inline refs */
 	unsigned long end;	/* Extent item end */
-	const u32 item_size = btrfs_item_size_nr(leaf, slot);
+	const u32 item_size = btrfs_item_size(leaf, slot);
 	u64 flags;
 	u64 generation;
 	u64 total_refs;		/* Total refs in btrfs_extent_item */
 	u64 inline_refs = 0;	/* found total inline refs */
 
-	if (key->type == BTRFS_METADATA_ITEM_KEY &&
-	    !btrfs_fs_incompat(fs_info, SKINNY_METADATA)) {
+	if (unlikely(key->type == BTRFS_METADATA_ITEM_KEY &&
+		     !btrfs_fs_incompat(fs_info, SKINNY_METADATA))) {
 		generic_err(leaf, slot,
 "invalid key type, METADATA_ITEM type invalid when SKINNY_METADATA feature disabled");
 		return -EUCLEAN;
 	}
 	/* key->objectid is the bytenr for both key types */
-	if (!IS_ALIGNED(key->objectid, fs_info->sectorsize)) {
+	if (unlikely(!IS_ALIGNED(key->objectid, fs_info->sectorsize))) {
 		generic_err(leaf, slot,
 		"invalid key objectid, have %llu expect to be aligned to %u",
 			   key->objectid, fs_info->sectorsize);
@@ -1173,8 +1262,8 @@ static int check_extent_item(struct extent_buffer *leaf,
 	}
 
 	/* key->offset is tree level for METADATA_ITEM_KEY */
-	if (key->type == BTRFS_METADATA_ITEM_KEY &&
-	    key->offset >= BTRFS_MAX_LEVEL) {
+	if (unlikely(key->type == BTRFS_METADATA_ITEM_KEY &&
+		     key->offset >= BTRFS_MAX_LEVEL)) {
 		extent_err(leaf, slot,
 			   "invalid tree level, have %llu expect [0, %u]",
 			   key->offset, BTRFS_MAX_LEVEL - 1);
@@ -1200,7 +1289,7 @@ static int check_extent_item(struct extent_buffer *leaf,
 	 *         Either using btrfs_extent_inline_ref::offset, or specific
 	 *         data structure.
 	 */
-	if (item_size < sizeof(*ei)) {
+	if (unlikely(item_size < sizeof(*ei))) {
 		extent_err(leaf, slot,
 			   "invalid item size, have %u expect [%zu, %u)",
 			   item_size, sizeof(*ei),
@@ -1214,15 +1303,16 @@ static int check_extent_item(struct extent_buffer *leaf,
 	flags = btrfs_extent_flags(leaf, ei);
 	total_refs = btrfs_extent_refs(leaf, ei);
 	generation = btrfs_extent_generation(leaf, ei);
-	if (generation > btrfs_super_generation(fs_info->super_copy) + 1) {
+	if (unlikely(generation >
+		     btrfs_super_generation(fs_info->super_copy) + 1)) {
 		extent_err(leaf, slot,
 			   "invalid generation, have %llu expect (0, %llu]",
 			   generation,
 			   btrfs_super_generation(fs_info->super_copy) + 1);
 		return -EUCLEAN;
 	}
-	if (!has_single_bit_set(flags & (BTRFS_EXTENT_FLAG_DATA |
-					 BTRFS_EXTENT_FLAG_TREE_BLOCK))) {
+	if (unlikely(!has_single_bit_set(flags & (BTRFS_EXTENT_FLAG_DATA |
+						  BTRFS_EXTENT_FLAG_TREE_BLOCK)))) {
 		extent_err(leaf, slot,
 		"invalid extent flag, have 0x%llx expect 1 bit set in 0x%llx",
 			flags, BTRFS_EXTENT_FLAG_DATA |
@@ -1231,26 +1321,31 @@ static int check_extent_item(struct extent_buffer *leaf,
 	}
 	is_tree_block = !!(flags & BTRFS_EXTENT_FLAG_TREE_BLOCK);
 	if (is_tree_block) {
-		if (key->type == BTRFS_EXTENT_ITEM_KEY &&
-		    key->offset != fs_info->nodesize) {
+		if (unlikely(key->type == BTRFS_EXTENT_ITEM_KEY &&
+			     key->offset != fs_info->nodesize)) {
 			extent_err(leaf, slot,
 				   "invalid extent length, have %llu expect %u",
 				   key->offset, fs_info->nodesize);
 			return -EUCLEAN;
 		}
 	} else {
-		if (key->type != BTRFS_EXTENT_ITEM_KEY) {
+		if (unlikely(key->type != BTRFS_EXTENT_ITEM_KEY)) {
 			extent_err(leaf, slot,
 			"invalid key type, have %u expect %u for data backref",
 				   key->type, BTRFS_EXTENT_ITEM_KEY);
 			return -EUCLEAN;
 		}
-		if (!IS_ALIGNED(key->offset, fs_info->sectorsize)) {
+		if (unlikely(!IS_ALIGNED(key->offset, fs_info->sectorsize))) {
 			extent_err(leaf, slot,
 			"invalid extent length, have %llu expect aligned to %u",
 				   key->offset, fs_info->sectorsize);
 			return -EUCLEAN;
 		}
+		if (unlikely(flags & BTRFS_BLOCK_FLAG_FULL_BACKREF)) {
+			extent_err(leaf, slot,
+			"invalid extent flag, data has full backref set");
+			return -EUCLEAN;
+		}
 	}
 	ptr = (unsigned long)(struct btrfs_extent_item *)(ei + 1);
 
@@ -1259,7 +1354,7 @@ static int check_extent_item(struct extent_buffer *leaf,
 		struct btrfs_tree_block_info *info;
 
 		info = (struct btrfs_tree_block_info *)ptr;
-		if (btrfs_tree_block_level(leaf, info) >= BTRFS_MAX_LEVEL) {
+		if (unlikely(btrfs_tree_block_level(leaf, info) >= BTRFS_MAX_LEVEL)) {
 			extent_err(leaf, slot,
 			"invalid tree block info level, have %u expect [0, %u]",
 				   btrfs_tree_block_level(leaf, info),
@@ -1278,7 +1373,7 @@ static int check_extent_item(struct extent_buffer *leaf,
 		u64 inline_offset;
 		u8 inline_type;
 
-		if (ptr + sizeof(*iref) > end) {
+		if (unlikely(ptr + sizeof(*iref) > end)) {
 			extent_err(leaf, slot,
 "inline ref item overflows extent item, ptr %lu iref size %zu end %lu",
 				   ptr, sizeof(*iref), end);
@@ -1287,7 +1382,7 @@ static int check_extent_item(struct extent_buffer *leaf,
 		iref = (struct btrfs_extent_inline_ref *)ptr;
 		inline_type = btrfs_extent_inline_ref_type(leaf, iref);
 		inline_offset = btrfs_extent_inline_ref_offset(leaf, iref);
-		if (ptr + btrfs_extent_inline_ref_size(inline_type) > end) {
+		if (unlikely(ptr + btrfs_extent_inline_ref_size(inline_type) > end)) {
 			extent_err(leaf, slot,
 "inline ref item overflows extent item, ptr %lu iref size %u end %lu",
 				   ptr, inline_type, end);
@@ -1301,7 +1396,8 @@ static int check_extent_item(struct extent_buffer *leaf,
 			break;
 		/* Contains parent bytenr */
 		case BTRFS_SHARED_BLOCK_REF_KEY:
-			if (!IS_ALIGNED(inline_offset, fs_info->sectorsize)) {
+			if (unlikely(!IS_ALIGNED(inline_offset,
+						 fs_info->sectorsize))) {
 				extent_err(leaf, slot,
 		"invalid tree parent bytenr, have %llu expect aligned to %u",
 					   inline_offset, fs_info->sectorsize);
@@ -1316,7 +1412,8 @@ static int check_extent_item(struct extent_buffer *leaf,
 		case BTRFS_EXTENT_DATA_REF_KEY:
 			dref = (struct btrfs_extent_data_ref *)(&iref->offset);
 			dref_offset = btrfs_extent_data_ref_offset(leaf, dref);
-			if (!IS_ALIGNED(dref_offset, fs_info->sectorsize)) {
+			if (unlikely(!IS_ALIGNED(dref_offset,
+						 fs_info->sectorsize))) {
 				extent_err(leaf, slot,
 		"invalid data ref offset, have %llu expect aligned to %u",
 					   dref_offset, fs_info->sectorsize);
@@ -1327,7 +1424,8 @@ static int check_extent_item(struct extent_buffer *leaf,
 		/* Contains parent bytenr and ref count */
 		case BTRFS_SHARED_DATA_REF_KEY:
 			sref = (struct btrfs_shared_data_ref *)(iref + 1);
-			if (!IS_ALIGNED(inline_offset, fs_info->sectorsize)) {
+			if (unlikely(!IS_ALIGNED(inline_offset,
+						 fs_info->sectorsize))) {
 				extent_err(leaf, slot,
 		"invalid data parent bytenr, have %llu expect aligned to %u",
 					   inline_offset, fs_info->sectorsize);
@@ -1343,19 +1441,39 @@ static int check_extent_item(struct extent_buffer *leaf,
 		ptr += btrfs_extent_inline_ref_size(inline_type);
 	}
 	/* No padding is allowed */
-	if (ptr != end) {
+	if (unlikely(ptr != end)) {
 		extent_err(leaf, slot,
 			   "invalid extent item size, padding bytes found");
 		return -EUCLEAN;
 	}
 
 	/* Finally, check the inline refs against total refs */
-	if (inline_refs > total_refs) {
+	if (unlikely(inline_refs > total_refs)) {
 		extent_err(leaf, slot,
 			"invalid extent refs, have %llu expect >= inline %llu",
 			   total_refs, inline_refs);
 		return -EUCLEAN;
 	}
+
+	if ((prev_key->type == BTRFS_EXTENT_ITEM_KEY) ||
+	    (prev_key->type == BTRFS_METADATA_ITEM_KEY)) {
+		u64 prev_end = prev_key->objectid;
+
+		if (prev_key->type == BTRFS_METADATA_ITEM_KEY)
+			prev_end += fs_info->nodesize;
+		else
+			prev_end += prev_key->offset;
+
+		if (unlikely(prev_end > key->objectid)) {
+			extent_err(leaf, slot,
+	"previous extent [%llu %u %llu] overlaps current extent [%llu %u %llu]",
+				   prev_key->objectid, prev_key->type,
+				   prev_key->offset, key->objectid, key->type,
+				   key->offset);
+			return -EUCLEAN;
+		}
+	}
+
 	return 0;
 }
 
@@ -1367,21 +1485,21 @@ static int check_simple_keyed_refs(struct extent_buffer *leaf,
 	if (key->type == BTRFS_SHARED_DATA_REF_KEY)
 		expect_item_size = sizeof(struct btrfs_shared_data_ref);
 
-	if (btrfs_item_size_nr(leaf, slot) != expect_item_size) {
+	if (unlikely(btrfs_item_size(leaf, slot) != expect_item_size)) {
 		generic_err(leaf, slot,
 		"invalid item size, have %u expect %u for key type %u",
-			    btrfs_item_size_nr(leaf, slot),
+			    btrfs_item_size(leaf, slot),
 			    expect_item_size, key->type);
 		return -EUCLEAN;
 	}
-	if (!IS_ALIGNED(key->objectid, leaf->fs_info->sectorsize)) {
+	if (unlikely(!IS_ALIGNED(key->objectid, leaf->fs_info->sectorsize))) {
 		generic_err(leaf, slot,
 "invalid key objectid for shared block ref, have %llu expect aligned to %u",
 			    key->objectid, leaf->fs_info->sectorsize);
 		return -EUCLEAN;
 	}
-	if (key->type != BTRFS_TREE_BLOCK_REF_KEY &&
-	    !IS_ALIGNED(key->offset, leaf->fs_info->sectorsize)) {
+	if (unlikely(key->type != BTRFS_TREE_BLOCK_REF_KEY &&
+		     !IS_ALIGNED(key->offset, leaf->fs_info->sectorsize))) {
 		extent_err(leaf, slot,
 		"invalid tree parent bytenr, have %llu expect aligned to %u",
 			   key->offset, leaf->fs_info->sectorsize);
@@ -1395,41 +1513,35 @@ static int check_extent_data_ref(struct extent_buffer *leaf,
 {
 	struct btrfs_extent_data_ref *dref;
 	unsigned long ptr = btrfs_item_ptr_offset(leaf, slot);
-	const unsigned long end = ptr + btrfs_item_size_nr(leaf, slot);
+	const unsigned long end = ptr + btrfs_item_size(leaf, slot);
 
-	if (btrfs_item_size_nr(leaf, slot) % sizeof(*dref) != 0) {
+	if (unlikely(btrfs_item_size(leaf, slot) % sizeof(*dref) != 0)) {
 		generic_err(leaf, slot,
 	"invalid item size, have %u expect aligned to %zu for key type %u",
-			    btrfs_item_size_nr(leaf, slot),
+			    btrfs_item_size(leaf, slot),
 			    sizeof(*dref), key->type);
+		return -EUCLEAN;
 	}
-	if (!IS_ALIGNED(key->objectid, leaf->fs_info->sectorsize)) {
+	if (unlikely(!IS_ALIGNED(key->objectid, leaf->fs_info->sectorsize))) {
 		generic_err(leaf, slot,
 "invalid key objectid for shared block ref, have %llu expect aligned to %u",
 			    key->objectid, leaf->fs_info->sectorsize);
 		return -EUCLEAN;
 	}
 	for (; ptr < end; ptr += sizeof(*dref)) {
-		u64 root_objectid;
-		u64 owner;
 		u64 offset;
-		u64 hash;
 
+		/*
+		 * We cannot check the extent_data_ref hash due to possible
+		 * overflow from the leaf due to hash collisions.
+		 */
 		dref = (struct btrfs_extent_data_ref *)ptr;
-		root_objectid = btrfs_extent_data_ref_root(leaf, dref);
-		owner = btrfs_extent_data_ref_objectid(leaf, dref);
 		offset = btrfs_extent_data_ref_offset(leaf, dref);
-		hash = hash_extent_data_ref(root_objectid, owner, offset);
-		if (hash != key->offset) {
-			extent_err(leaf, slot,
-	"invalid extent data ref hash, item has 0x%016llx key has 0x%016llx",
-				   hash, key->offset);
-			return -EUCLEAN;
-		}
-		if (!IS_ALIGNED(offset, leaf->fs_info->sectorsize)) {
+		if (unlikely(!IS_ALIGNED(offset, leaf->fs_info->sectorsize))) {
 			extent_err(leaf, slot,
 	"invalid extent data backref offset, have %llu expect aligned to %u",
 				   offset, leaf->fs_info->sectorsize);
+			return -EUCLEAN;
 		}
 	}
 	return 0;
@@ -1445,23 +1557,23 @@ static int check_inode_ref(struct extent_buffer *leaf,
 	unsigned long ptr;
 	unsigned long end;
 
-	if (!check_prev_ino(leaf, key, slot, prev_key))
+	if (unlikely(!check_prev_ino(leaf, key, slot, prev_key)))
 		return -EUCLEAN;
 	/* namelen can't be 0, so item_size == sizeof() is also invalid */
-	if (btrfs_item_size_nr(leaf, slot) <= sizeof(*iref)) {
+	if (unlikely(btrfs_item_size(leaf, slot) <= sizeof(*iref))) {
 		inode_ref_err(leaf, slot,
 			"invalid item size, have %u expect (%zu, %u)",
-			btrfs_item_size_nr(leaf, slot),
+			btrfs_item_size(leaf, slot),
 			sizeof(*iref), BTRFS_LEAF_DATA_SIZE(leaf->fs_info));
 		return -EUCLEAN;
 	}
 
 	ptr = btrfs_item_ptr_offset(leaf, slot);
-	end = ptr + btrfs_item_size_nr(leaf, slot);
+	end = ptr + btrfs_item_size(leaf, slot);
 	while (ptr < end) {
 		u16 namelen;
 
-		if (ptr + sizeof(iref) > end) {
+		if (unlikely(ptr + sizeof(iref) > end)) {
 			inode_ref_err(leaf, slot,
 			"inode ref overflow, ptr %lu end %lu inode_ref_size %zu",
 				ptr, end, sizeof(iref));
@@ -1470,7 +1582,7 @@ static int check_inode_ref(struct extent_buffer *leaf,
 
 		iref = (struct btrfs_inode_ref *)ptr;
 		namelen = btrfs_inode_ref_name_len(leaf, iref);
-		if (ptr + sizeof(*iref) + namelen > end) {
+		if (unlikely(ptr + sizeof(*iref) + namelen > end)) {
 			inode_ref_err(leaf, slot,
 				"inode ref overflow, ptr %lu end %lu namelen %u",
 				ptr, end, namelen);
@@ -1530,7 +1642,7 @@ static int check_leaf_item(struct extent_buffer *leaf,
 		break;
 	case BTRFS_EXTENT_ITEM_KEY:
 	case BTRFS_METADATA_ITEM_KEY:
-		ret = check_extent_item(leaf, key, slot);
+		ret = check_extent_item(leaf, key, slot, prev_key);
 		break;
 	case BTRFS_TREE_BLOCK_REF_KEY:
 	case BTRFS_SHARED_DATA_REF_KEY:
@@ -1553,7 +1665,7 @@ static int check_leaf(struct extent_buffer *leaf, bool check_item_data)
 	u32 nritems = btrfs_header_nritems(leaf);
 	int slot;
 
-	if (btrfs_header_level(leaf) != 0) {
+	if (unlikely(btrfs_header_level(leaf) != 0)) {
 		generic_err(leaf, 0,
 			"invalid level for leaf, have %d expect 0",
 			btrfs_header_level(leaf));
@@ -1572,27 +1684,39 @@ static int check_leaf(struct extent_buffer *leaf, bool check_item_data)
 		u64 owner = btrfs_header_owner(leaf);
 
 		/* These trees must never be empty */
-		if (owner == BTRFS_ROOT_TREE_OBJECTID ||
-		    owner == BTRFS_CHUNK_TREE_OBJECTID ||
-		    owner == BTRFS_EXTENT_TREE_OBJECTID ||
-		    owner == BTRFS_DEV_TREE_OBJECTID ||
-		    owner == BTRFS_FS_TREE_OBJECTID ||
-		    owner == BTRFS_DATA_RELOC_TREE_OBJECTID) {
+		if (unlikely(owner == BTRFS_ROOT_TREE_OBJECTID ||
+			     owner == BTRFS_CHUNK_TREE_OBJECTID ||
+			     owner == BTRFS_DEV_TREE_OBJECTID ||
+			     owner == BTRFS_FS_TREE_OBJECTID ||
+			     owner == BTRFS_DATA_RELOC_TREE_OBJECTID)) {
 			generic_err(leaf, 0,
 			"invalid root, root %llu must never be empty",
 				    owner);
 			return -EUCLEAN;
 		}
+
 		/* Unknown tree */
-		if (owner == 0) {
+		if (unlikely(owner == 0)) {
 			generic_err(leaf, 0,
 				"invalid owner, root 0 is not defined");
 			return -EUCLEAN;
 		}
+
+		/* EXTENT_TREE_V2 can have empty extent trees. */
+		if (btrfs_fs_incompat(fs_info, EXTENT_TREE_V2))
+			return 0;
+
+		if (unlikely(owner == BTRFS_EXTENT_TREE_OBJECTID)) {
+			generic_err(leaf, 0,
+			"invalid root, root %llu must never be empty",
+				    owner);
+			return -EUCLEAN;
+		}
+
 		return 0;
 	}
 
-	if (nritems == 0)
+	if (unlikely(nritems == 0))
 		return 0;
 
 	/*
@@ -1608,12 +1732,13 @@ static int check_leaf(struct extent_buffer *leaf, bool check_item_data)
 	 */
 	for (slot = 0; slot < nritems; slot++) {
 		u32 item_end_expected;
+		u64 item_data_end;
 		int ret;
 
 		btrfs_item_key_to_cpu(leaf, &key, slot);
 
 		/* Make sure the keys are in the right order */
-		if (btrfs_comp_cpu_keys(&prev_key, &key) >= 0) {
+		if (unlikely(btrfs_comp_cpu_keys(&prev_key, &key) >= 0)) {
 			generic_err(leaf, slot,
 	"bad key order, prev (%llu %u %llu) current (%llu %u %llu)",
 				prev_key.objectid, prev_key.type,
@@ -1622,6 +1747,8 @@ static int check_leaf(struct extent_buffer *leaf, bool check_item_data)
 			return -EUCLEAN;
 		}
 
+		item_data_end = (u64)btrfs_item_offset(leaf, slot) +
+				btrfs_item_size(leaf, slot);
 		/*
 		 * Make sure the offset and ends are right, remember that the
 		 * item data starts at the end of the leaf and grows towards the
@@ -1630,13 +1757,12 @@ static int check_leaf(struct extent_buffer *leaf, bool check_item_data)
 		if (slot == 0)
 			item_end_expected = BTRFS_LEAF_DATA_SIZE(fs_info);
 		else
-			item_end_expected = btrfs_item_offset_nr(leaf,
+			item_end_expected = btrfs_item_offset(leaf,
 								 slot - 1);
-		if (btrfs_item_end_nr(leaf, slot) != item_end_expected) {
+		if (unlikely(item_data_end != item_end_expected)) {
 			generic_err(leaf, slot,
-				"unexpected item end, have %u expect %u",
-				btrfs_item_end_nr(leaf, slot),
-				item_end_expected);
+				"unexpected item end, have %llu expect %u",
+				item_data_end, item_end_expected);
 			return -EUCLEAN;
 		}
 
@@ -1645,18 +1771,16 @@ static int check_leaf(struct extent_buffer *leaf, bool check_item_data)
 		 * just in case all the items are consistent to each other, but
 		 * all point outside of the leaf.
 		 */
-		if (btrfs_item_end_nr(leaf, slot) >
-		    BTRFS_LEAF_DATA_SIZE(fs_info)) {
+		if (unlikely(item_data_end > BTRFS_LEAF_DATA_SIZE(fs_info))) {
 			generic_err(leaf, slot,
-			"slot end outside of leaf, have %u expect range [0, %u]",
-				btrfs_item_end_nr(leaf, slot),
-				BTRFS_LEAF_DATA_SIZE(fs_info));
+			"slot end outside of leaf, have %llu expect range [0, %u]",
+				item_data_end, BTRFS_LEAF_DATA_SIZE(fs_info));
 			return -EUCLEAN;
 		}
 
 		/* Also check if the item pointer overlaps with btrfs item. */
-		if (btrfs_item_nr_offset(slot) + sizeof(struct btrfs_item) >
-		    btrfs_item_ptr_offset(leaf, slot)) {
+		if (unlikely(btrfs_item_ptr_offset(leaf, slot) <
+			     btrfs_item_nr_offset(slot) + sizeof(struct btrfs_item))) {
 			generic_err(leaf, slot,
 		"slot overlaps with its data, item end %lu data start %lu",
 				btrfs_item_nr_offset(slot) +
@@ -1671,7 +1795,7 @@ static int check_leaf(struct extent_buffer *leaf, bool check_item_data)
 			 * criteria
 			 */
 			ret = check_leaf_item(leaf, &key, slot, &prev_key);
-			if (ret < 0)
+			if (unlikely(ret < 0))
 				return ret;
 		}
 
@@ -1704,13 +1828,13 @@ int btrfs_check_node(struct extent_buffer *node)
 	u64 bytenr;
 	int ret = 0;
 
-	if (level <= 0 || level >= BTRFS_MAX_LEVEL) {
+	if (unlikely(level <= 0 || level >= BTRFS_MAX_LEVEL)) {
 		generic_err(node, 0,
 			"invalid level for node, have %d expect [1, %d]",
 			level, BTRFS_MAX_LEVEL - 1);
 		return -EUCLEAN;
 	}
-	if (nr == 0 || nr > BTRFS_NODEPTRS_PER_BLOCK(fs_info)) {
+	if (unlikely(nr == 0 || nr > BTRFS_NODEPTRS_PER_BLOCK(fs_info))) {
 		btrfs_crit(fs_info,
 "corrupt node: root=%llu block=%llu, nritems too %s, have %lu expect range [1,%u]",
 			   btrfs_header_owner(node), node->start,
@@ -1724,13 +1848,13 @@ int btrfs_check_node(struct extent_buffer *node)
 		btrfs_node_key_to_cpu(node, &key, slot);
 		btrfs_node_key_to_cpu(node, &next_key, slot + 1);
 
-		if (!bytenr) {
+		if (unlikely(!bytenr)) {
 			generic_err(node, slot,
 				"invalid NULL node pointer");
 			ret = -EUCLEAN;
 			goto out;
 		}
-		if (!IS_ALIGNED(bytenr, fs_info->sectorsize)) {
+		if (unlikely(!IS_ALIGNED(bytenr, fs_info->sectorsize))) {
 			generic_err(node, slot,
 			"unaligned pointer, have %llu should be aligned to %u",
 				bytenr, fs_info->sectorsize);
@@ -1738,7 +1862,7 @@ int btrfs_check_node(struct extent_buffer *node)
 			goto out;
 		}
 
-		if (btrfs_comp_cpu_keys(&key, &next_key) >= 0) {
+		if (unlikely(btrfs_comp_cpu_keys(&key, &next_key) >= 0)) {
 			generic_err(node, slot,
 	"bad key order, current (%llu %u %llu) next (%llu %u %llu)",
 				key.objectid, key.type, key.offset,
@@ -1752,3 +1876,58 @@ out:
 	return ret;
 }
 ALLOW_ERROR_INJECTION(btrfs_check_node, ERRNO);
+
+int btrfs_check_eb_owner(const struct extent_buffer *eb, u64 root_owner)
+{
+	const bool is_subvol = is_fstree(root_owner);
+	const u64 eb_owner = btrfs_header_owner(eb);
+
+	/*
+	 * Skip dummy fs, as selftests don't create unique ebs for each dummy
+	 * root.
+	 */
+	if (test_bit(BTRFS_FS_STATE_DUMMY_FS_INFO, &eb->fs_info->fs_state))
+		return 0;
+	/*
+	 * There are several call sites (backref walking, qgroup, and data
+	 * reloc) passing 0 as @root_owner, as they are not holding the
+	 * tree root.  In that case, we can not do a reliable ownership check,
+	 * so just exit.
+	 */
+	if (root_owner == 0)
+		return 0;
+	/*
+	 * These trees use key.offset as their owner, our callers don't have
+	 * the extra capacity to pass key.offset here.  So we just skip them.
+	 */
+	if (root_owner == BTRFS_TREE_LOG_OBJECTID ||
+	    root_owner == BTRFS_TREE_RELOC_OBJECTID)
+		return 0;
+
+	if (!is_subvol) {
+		/* For non-subvolume trees, the eb owner should match root owner */
+		if (unlikely(root_owner != eb_owner)) {
+			btrfs_crit(eb->fs_info,
+"corrupted %s, root=%llu block=%llu owner mismatch, have %llu expect %llu",
+				btrfs_header_level(eb) == 0 ? "leaf" : "node",
+				root_owner, btrfs_header_bytenr(eb), eb_owner,
+				root_owner);
+			return -EUCLEAN;
+		}
+		return 0;
+	}
+
+	/*
+	 * For subvolume trees, owners can mismatch, but they should all belong
+	 * to subvolume trees.
+	 */
+	if (unlikely(is_subvol != is_fstree(eb_owner))) {
+		btrfs_crit(eb->fs_info,
+"corrupted %s, root=%llu block=%llu owner mismatch, have %llu expect [%llu, %llu]",
+			btrfs_header_level(eb) == 0 ? "leaf" : "node",
+			root_owner, btrfs_header_bytenr(eb), eb_owner,
+			BTRFS_FIRST_FREE_OBJECTID, BTRFS_LAST_FREE_OBJECTID);
+		return -EUCLEAN;
+	}
+	return 0;
+}
diff --git a/fs/btrfs/tree-checker.h b/fs/btrfs/tree-checker.h
index 32fecc9dc1dd..ece497e26558 100644
--- a/fs/btrfs/tree-checker.h
+++ b/fs/btrfs/tree-checker.h
@@ -25,5 +25,6 @@ int btrfs_check_node(struct extent_buffer *node);
 
 int btrfs_check_chunk_valid(struct extent_buffer *leaf,
 			    struct btrfs_chunk *chunk, u64 logical);
+int btrfs_check_eb_owner(const struct extent_buffer *eb, u64 root_owner);
 
 #endif
diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index 5f9e2dd413af..b6cf39f4e7e4 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -27,15 +27,7 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 	int next_key_ret = 0;
 	u64 last_ret = 0;
 
-	if (root->fs_info->extent_root == root) {
-		/*
-		 * there's recursion here right now in the tree locking,
-		 * we can't defrag the extent root without deadlock
-		 */
-		goto out;
-	}
-
-	if (!test_bit(BTRFS_ROOT_REF_COWS, &root->state))
+	if (!test_bit(BTRFS_ROOT_SHAREABLE, &root->state))
 		goto out;
 
 	path = btrfs_alloc_path();
@@ -52,7 +44,6 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 		u32 nritems;
 
 		root_node = btrfs_lock_root_node(root);
-		btrfs_set_lock_blocking_write(root_node);
 		nritems = btrfs_header_nritems(root_node);
 		root->defrag_max.objectid = 0;
 		/* from above we know this is not a leaf */
@@ -133,10 +124,9 @@ out:
 		ret = 0;
 	}
 done:
-	if (ret != -EAGAIN) {
+	if (ret != -EAGAIN)
 		memset(&root->defrag_progress, 0,
 		       sizeof(root->defrag_progress));
-		root->defrag_trans_start = trans->transid;
-	}
+
 	return ret;
 }
diff --git a/fs/btrfs/tree-log.c b/fs/btrfs/tree-log.c
index 7dd7552f53a4..813986e38258 100644
--- a/fs/btrfs/tree-log.c
+++ b/fs/btrfs/tree-log.c
@@ -17,7 +17,12 @@
 #include "backref.h"
 #include "compression.h"
 #include "qgroup.h"
-#include "inode-map.h"
+#include "block-group.h"
+#include "space-info.h"
+#include "zoned.h"
+#include "inode-item.h"
+
+#define MAX_CONFLICT_INODES 10
 
 /* magic values for the inode_only field in btrfs_log_inode:
  *
@@ -28,8 +33,6 @@
 enum {
 	LOG_INODE_ALL,
 	LOG_INODE_EXISTS,
-	LOG_OTHER_INODE,
-	LOG_OTHER_INODE_ALL,
 };
 
 /*
@@ -92,10 +95,8 @@ enum {
 };
 
 static int btrfs_log_inode(struct btrfs_trans_handle *trans,
-			   struct btrfs_root *root, struct btrfs_inode *inode,
+			   struct btrfs_inode *inode,
 			   int inode_only,
-			   const loff_t start,
-			   const loff_t end,
 			   struct btrfs_log_ctx *ctx);
 static int link_to_fixup_dir(struct btrfs_trans_handle *trans,
 			     struct btrfs_root *root,
@@ -105,6 +106,7 @@ static noinline int replay_dir_deletes(struct btrfs_trans_handle *trans,
 				       struct btrfs_root *log,
 				       struct btrfs_path *path,
 				       u64 dirid, int del_all);
+static void wait_log_commit(struct btrfs_root *root, int transid);
 
 /*
  * tree logging is a special write ahead log used to make sure that
@@ -139,16 +141,45 @@ static int start_log_trans(struct btrfs_trans_handle *trans,
 			   struct btrfs_log_ctx *ctx)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
+	struct btrfs_root *tree_root = fs_info->tree_root;
+	const bool zoned = btrfs_is_zoned(fs_info);
 	int ret = 0;
+	bool created = false;
+
+	/*
+	 * First check if the log root tree was already created. If not, create
+	 * it before locking the root's log_mutex, just to keep lockdep happy.
+	 */
+	if (!test_bit(BTRFS_ROOT_HAS_LOG_TREE, &tree_root->state)) {
+		mutex_lock(&tree_root->log_mutex);
+		if (!fs_info->log_root_tree) {
+			ret = btrfs_init_log_root_tree(trans, fs_info);
+			if (!ret) {
+				set_bit(BTRFS_ROOT_HAS_LOG_TREE, &tree_root->state);
+				created = true;
+			}
+		}
+		mutex_unlock(&tree_root->log_mutex);
+		if (ret)
+			return ret;
+	}
 
 	mutex_lock(&root->log_mutex);
 
+again:
 	if (root->log_root) {
+		int index = (root->log_transid + 1) % 2;
+
 		if (btrfs_need_log_full_commit(trans)) {
-			ret = -EAGAIN;
+			ret = BTRFS_LOG_FORCE_COMMIT;
 			goto out;
 		}
 
+		if (zoned && atomic_read(&root->log_commit[index])) {
+			wait_log_commit(root, root->log_transid - 1);
+			goto again;
+		}
+
 		if (!root->log_start_pid) {
 			clear_bit(BTRFS_ROOT_MULTI_LOG_TASKS, &root->state);
 			root->log_start_pid = current->pid;
@@ -156,24 +187,28 @@ static int start_log_trans(struct btrfs_trans_handle *trans,
 			set_bit(BTRFS_ROOT_MULTI_LOG_TASKS, &root->state);
 		}
 	} else {
-		mutex_lock(&fs_info->tree_log_mutex);
-		if (!fs_info->log_root_tree)
-			ret = btrfs_init_log_root_tree(trans, fs_info);
-		mutex_unlock(&fs_info->tree_log_mutex);
-		if (ret)
+		/*
+		 * This means fs_info->log_root_tree was already created
+		 * for some other FS trees. Do the full commit not to mix
+		 * nodes from multiple log transactions to do sequential
+		 * writing.
+		 */
+		if (zoned && !created) {
+			ret = BTRFS_LOG_FORCE_COMMIT;
 			goto out;
+		}
 
 		ret = btrfs_add_log_tree(trans, root);
 		if (ret)
 			goto out;
 
+		set_bit(BTRFS_ROOT_HAS_LOG_TREE, &root->state);
 		clear_bit(BTRFS_ROOT_MULTI_LOG_TASKS, &root->state);
 		root->log_start_pid = current->pid;
 	}
 
-	atomic_inc(&root->log_batch);
 	atomic_inc(&root->log_writers);
-	if (ctx) {
+	if (!ctx->logging_new_name) {
 		int index = root->log_transid % 2;
 		list_add_tail(&ctx->list, &root->log_ctxs[index]);
 		ctx->log_transid = root->log_transid;
@@ -191,11 +226,22 @@ out:
  */
 static int join_running_log_trans(struct btrfs_root *root)
 {
+	const bool zoned = btrfs_is_zoned(root->fs_info);
 	int ret = -ENOENT;
 
+	if (!test_bit(BTRFS_ROOT_HAS_LOG_TREE, &root->state))
+		return ret;
+
 	mutex_lock(&root->log_mutex);
+again:
 	if (root->log_root) {
+		int index = (root->log_transid + 1) % 2;
+
 		ret = 0;
+		if (zoned && atomic_read(&root->log_commit[index])) {
+			wait_log_commit(root, root->log_transid - 1);
+			goto again;
+		}
 		atomic_inc(&root->log_writers);
 	}
 	mutex_unlock(&root->log_mutex);
@@ -209,9 +255,7 @@ static int join_running_log_trans(struct btrfs_root *root)
  */
 void btrfs_pin_log_trans(struct btrfs_root *root)
 {
-	mutex_lock(&root->log_mutex);
 	atomic_inc(&root->log_writers);
-	mutex_unlock(&root->log_mutex);
 }
 
 /*
@@ -226,12 +270,6 @@ void btrfs_end_log_trans(struct btrfs_root *root)
 	}
 }
 
-static int btrfs_write_tree_block(struct extent_buffer *buf)
-{
-	return filemap_fdatawrite_range(buf->pages[0]->mapping, buf->start,
-					buf->start + buf->len - 1);
-}
-
 static void btrfs_wait_tree_block_writeback(struct extent_buffer *buf)
 {
 	filemap_fdatawait_range(buf->pages[0]->mapping,
@@ -250,16 +288,6 @@ struct walk_control {
 	 */
 	int free;
 
-	/* should we write out the extent buffer?  This is used
-	 * while flushing the log tree to disk during a sync
-	 */
-	int write;
-
-	/* should we wait for the extent buffer io to finish?  Also used
-	 * while flushing the log tree to disk for a sync
-	 */
-	int wait;
-
 	/* pin only walk, we record which extents on disk belong to the
 	 * log trees
 	 */
@@ -305,45 +333,29 @@ static int process_one_buffer(struct btrfs_root *log,
 	 * pin down any logged extents, so we have to read the block.
 	 */
 	if (btrfs_fs_incompat(fs_info, MIXED_GROUPS)) {
-		ret = btrfs_read_buffer(eb, gen, level, NULL);
+		ret = btrfs_read_extent_buffer(eb, gen, level, NULL);
 		if (ret)
 			return ret;
 	}
 
-	if (wc->pin)
-		ret = btrfs_pin_extent_for_log_replay(fs_info, eb->start,
+	if (wc->pin) {
+		ret = btrfs_pin_extent_for_log_replay(wc->trans, eb->start,
 						      eb->len);
+		if (ret)
+			return ret;
 
-	if (!ret && btrfs_buffer_uptodate(eb, gen, 0)) {
-		if (wc->pin && btrfs_header_level(eb) == 0)
+		if (btrfs_buffer_uptodate(eb, gen, 0) &&
+		    btrfs_header_level(eb) == 0)
 			ret = btrfs_exclude_logged_extents(eb);
-		if (wc->write)
-			btrfs_write_tree_block(eb);
-		if (wc->wait)
-			btrfs_wait_tree_block_writeback(eb);
 	}
 	return ret;
 }
 
-/*
- * Item overwrite used by replay and tree logging.  eb, slot and key all refer
- * to the src data we are copying out.
- *
- * root is the tree we are copying into, and path is a scratch
- * path for use in this function (it should be released on entry and
- * will be released on exit).
- *
- * If the key is already in the destination tree the existing item is
- * overwritten.  If the existing item isn't big enough, it is extended.
- * If it is too large, it is truncated.
- *
- * If the key isn't in the destination yet, a new item is inserted.
- */
-static noinline int overwrite_item(struct btrfs_trans_handle *trans,
-				   struct btrfs_root *root,
-				   struct btrfs_path *path,
-				   struct extent_buffer *eb, int slot,
-				   struct btrfs_key *key)
+static int do_overwrite_item(struct btrfs_trans_handle *trans,
+			     struct btrfs_root *root,
+			     struct btrfs_path *path,
+			     struct extent_buffer *eb, int slot,
+			     struct btrfs_key *key)
 {
 	int ret;
 	u32 item_size;
@@ -357,18 +369,30 @@ static noinline int overwrite_item(struct btrfs_trans_handle *trans,
 	if (root->root_key.objectid != BTRFS_TREE_LOG_OBJECTID)
 		overwrite_root = 1;
 
-	item_size = btrfs_item_size_nr(eb, slot);
+	item_size = btrfs_item_size(eb, slot);
 	src_ptr = btrfs_item_ptr_offset(eb, slot);
 
-	/* look for the key in the destination tree */
-	ret = btrfs_search_slot(NULL, root, key, path, 0, 0);
-	if (ret < 0)
-		return ret;
+	/* Our caller must have done a search for the key for us. */
+	ASSERT(path->nodes[0] != NULL);
+
+	/*
+	 * And the slot must point to the exact key or the slot where the key
+	 * should be at (the first item with a key greater than 'key')
+	 */
+	if (path->slots[0] < btrfs_header_nritems(path->nodes[0])) {
+		struct btrfs_key found_key;
+
+		btrfs_item_key_to_cpu(path->nodes[0], &found_key, path->slots[0]);
+		ret = btrfs_comp_cpu_keys(&found_key, key);
+		ASSERT(ret >= 0);
+	} else {
+		ret = 1;
+	}
 
 	if (ret == 0) {
 		char *src_copy;
 		char *dst_copy;
-		u32 dst_size = btrfs_item_size_nr(path->nodes[0],
+		u32 dst_size = btrfs_item_size(path->nodes[0],
 						  path->slots[0]);
 		if (dst_size != item_size)
 			goto insert;
@@ -462,7 +486,7 @@ insert:
 	/* make sure any existing item is the correct size */
 	if (ret == -EEXIST || ret == -EOVERFLOW) {
 		u32 found_size;
-		found_size = btrfs_item_size_nr(path->nodes[0],
+		found_size = btrfs_item_size(path->nodes[0],
 						path->slots[0]);
 		if (found_size > item_size)
 			btrfs_truncate_item(path, item_size, 1);
@@ -503,13 +527,8 @@ insert:
 			 */
 			if (S_ISREG(btrfs_inode_mode(eb, src_item)) &&
 			    S_ISREG(btrfs_inode_mode(dst_eb, dst_item)) &&
-			    ino_size != 0) {
-				struct btrfs_map_token token;
-
-				btrfs_init_map_token(&token, dst_eb);
-				btrfs_set_token_inode_size(dst_eb, dst_item,
-							   ino_size, &token);
-			}
+			    ino_size != 0)
+				btrfs_set_inode_size(dst_eb, dst_item, ino_size);
 			goto no_copy;
 		}
 
@@ -547,19 +566,45 @@ no_copy:
 }
 
 /*
+ * Item overwrite used by replay and tree logging.  eb, slot and key all refer
+ * to the src data we are copying out.
+ *
+ * root is the tree we are copying into, and path is a scratch
+ * path for use in this function (it should be released on entry and
+ * will be released on exit).
+ *
+ * If the key is already in the destination tree the existing item is
+ * overwritten.  If the existing item isn't big enough, it is extended.
+ * If it is too large, it is truncated.
+ *
+ * If the key isn't in the destination yet, a new item is inserted.
+ */
+static int overwrite_item(struct btrfs_trans_handle *trans,
+			  struct btrfs_root *root,
+			  struct btrfs_path *path,
+			  struct extent_buffer *eb, int slot,
+			  struct btrfs_key *key)
+{
+	int ret;
+
+	/* Look for the key in the destination tree. */
+	ret = btrfs_search_slot(NULL, root, key, path, 0, 0);
+	if (ret < 0)
+		return ret;
+
+	return do_overwrite_item(trans, root, path, eb, slot, key);
+}
+
+/*
  * simple helper to read an inode off the disk from a given root
  * This can only be called for subvolume roots and not for the log
  */
 static noinline struct inode *read_one_inode(struct btrfs_root *root,
 					     u64 objectid)
 {
-	struct btrfs_key key;
 	struct inode *inode;
 
-	key.objectid = objectid;
-	key.type = BTRFS_INODE_ITEM_KEY;
-	key.offset = 0;
-	inode = btrfs_iget(root->fs_info->sb, &key, root);
+	inode = btrfs_iget(root->fs_info->sb, objectid, root);
 	if (IS_ERR(inode))
 		inode = NULL;
 	return inode;
@@ -583,6 +628,7 @@ static noinline int replay_one_extent(struct btrfs_trans_handle *trans,
 				      struct extent_buffer *eb, int slot,
 				      struct btrfs_key *key)
 {
+	struct btrfs_drop_extents_args drop_args = { 0 };
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	int found_type;
 	u64 extent_end;
@@ -660,7 +706,10 @@ static noinline int replay_one_extent(struct btrfs_trans_handle *trans,
 	btrfs_release_path(path);
 
 	/* drop any overlapping extents */
-	ret = btrfs_drop_extents(trans, root, inode, start, extent_end, 1);
+	drop_args.start = start;
+	drop_args.end = extent_end;
+	drop_args.drop_cache = true;
+	ret = btrfs_drop_extents(trans, root, BTRFS_I(inode), &drop_args);
 	if (ret)
 		goto out;
 
@@ -715,13 +764,15 @@ static noinline int replay_one_extent(struct btrfs_trans_handle *trans,
 			 */
 			ret = btrfs_lookup_data_extent(fs_info, ins.objectid,
 						ins.offset);
-			if (ret == 0) {
+			if (ret < 0) {
+				goto out;
+			} else if (ret == 0) {
 				btrfs_init_generic_ref(&ref,
 						BTRFS_ADD_DELAYED_REF,
 						ins.objectid, ins.offset, 0);
 				btrfs_init_data_ref(&ref,
 						root->root_key.objectid,
-						key->objectid, offset);
+						key->objectid, offset, 0, false);
 				ret = btrfs_inc_extent_ref(trans, &ref);
 				if (ret)
 					goto out;
@@ -750,7 +801,7 @@ static noinline int replay_one_extent(struct btrfs_trans_handle *trans,
 
 			ret = btrfs_lookup_csums_range(root->log_root,
 						csum_start, csum_end - 1,
-						&ordered_sums, 0);
+						&ordered_sums, 0, false);
 			if (ret)
 				goto out;
 			/*
@@ -804,17 +855,21 @@ static noinline int replay_one_extent(struct btrfs_trans_handle *trans,
 			 */
 			while (!list_empty(&ordered_sums)) {
 				struct btrfs_ordered_sum *sums;
+				struct btrfs_root *csum_root;
+
 				sums = list_entry(ordered_sums.next,
 						struct btrfs_ordered_sum,
 						list);
+				csum_root = btrfs_csum_root(fs_info,
+							    sums->bytenr);
 				if (!ret)
-					ret = btrfs_del_csums(trans,
-							      fs_info->csum_root,
+					ret = btrfs_del_csums(trans, csum_root,
 							      sums->bytenr,
 							      sums->len);
 				if (!ret)
 					ret = btrfs_csum_file_blocks(trans,
-						fs_info->csum_root, sums);
+								     csum_root,
+								     sums);
 				list_del(&sums->list);
 				kfree(sums);
 			}
@@ -830,15 +885,39 @@ static noinline int replay_one_extent(struct btrfs_trans_handle *trans,
 			goto out;
 	}
 
-	inode_add_bytes(inode, nbytes);
+	ret = btrfs_inode_set_file_extent_range(BTRFS_I(inode), start,
+						extent_end - start);
+	if (ret)
+		goto out;
+
 update_inode:
-	ret = btrfs_update_inode(trans, root, inode);
+	btrfs_update_inode_bytes(BTRFS_I(inode), nbytes, drop_args.bytes_found);
+	ret = btrfs_update_inode(trans, root, BTRFS_I(inode));
 out:
-	if (inode)
-		iput(inode);
+	iput(inode);
 	return ret;
 }
 
+static int unlink_inode_for_log_replay(struct btrfs_trans_handle *trans,
+				       struct btrfs_inode *dir,
+				       struct btrfs_inode *inode,
+				       const char *name,
+				       int name_len)
+{
+	int ret;
+
+	ret = btrfs_unlink_inode(trans, dir, inode, name, name_len);
+	if (ret)
+		return ret;
+	/*
+	 * Whenever we need to check if a name exists or not, we check the
+	 * fs/subvolume tree. So after an unlink we must run delayed items, so
+	 * that future checks for a name during log replay see that the name
+	 * does not exists anymore.
+	 */
+	return btrfs_run_delayed_items(trans);
+}
+
 /*
  * when cleaning up conflicts between the directory names in the
  * subvolume, directory names in the log and directory names in the
@@ -848,11 +927,11 @@ out:
  * item
  */
 static noinline int drop_one_dir_item(struct btrfs_trans_handle *trans,
-				      struct btrfs_root *root,
 				      struct btrfs_path *path,
 				      struct btrfs_inode *dir,
 				      struct btrfs_dir_item *di)
 {
+	struct btrfs_root *root = dir->root;
 	struct inode *inode;
 	char *name;
 	int name_len;
@@ -881,12 +960,8 @@ static noinline int drop_one_dir_item(struct btrfs_trans_handle *trans,
 	if (ret)
 		goto out;
 
-	ret = btrfs_unlink_inode(trans, root, dir, BTRFS_I(inode), name,
+	ret = unlink_inode_for_log_replay(trans, dir, BTRFS_I(inode), name,
 			name_len);
-	if (ret)
-		goto out;
-	else
-		ret = btrfs_run_delayed_items(trans);
 out:
 	kfree(name);
 	iput(inode);
@@ -894,9 +969,11 @@ out:
 }
 
 /*
- * helper function to see if a given name and sequence number found
- * in an inode back reference are already in a directory and correctly
- * point to this inode
+ * See if a given name and sequence number found in an inode back reference are
+ * already in a directory and correctly point to this inode.
+ *
+ * Returns: < 0 on error, 0 if the directory entry does not exists and 1 if it
+ * exists.
  */
 static noinline int inode_in_dir(struct btrfs_root *root,
 				 struct btrfs_path *path,
@@ -905,29 +982,34 @@ static noinline int inode_in_dir(struct btrfs_root *root,
 {
 	struct btrfs_dir_item *di;
 	struct btrfs_key location;
-	int match = 0;
+	int ret = 0;
 
 	di = btrfs_lookup_dir_index_item(NULL, root, path, dirid,
 					 index, name, name_len, 0);
-	if (di && !IS_ERR(di)) {
+	if (IS_ERR(di)) {
+		ret = PTR_ERR(di);
+		goto out;
+	} else if (di) {
 		btrfs_dir_item_key_to_cpu(path->nodes[0], di, &location);
 		if (location.objectid != objectid)
 			goto out;
-	} else
+	} else {
 		goto out;
-	btrfs_release_path(path);
+	}
 
+	btrfs_release_path(path);
 	di = btrfs_lookup_dir_item(NULL, root, path, dirid, name, name_len, 0);
-	if (di && !IS_ERR(di)) {
-		btrfs_dir_item_key_to_cpu(path->nodes[0], di, &location);
-		if (location.objectid != objectid)
-			goto out;
-	} else
+	if (IS_ERR(di)) {
+		ret = PTR_ERR(di);
 		goto out;
-	match = 1;
+	} else if (di) {
+		btrfs_dir_item_key_to_cpu(path->nodes[0], di, &location);
+		if (location.objectid == objectid)
+			ret = 1;
+	}
 out:
 	btrfs_release_path(path);
-	return match;
+	return ret;
 }
 
 /*
@@ -981,8 +1063,7 @@ static inline int __add_inode_ref(struct btrfs_trans_handle *trans,
 				  struct btrfs_inode *dir,
 				  struct btrfs_inode *inode,
 				  u64 inode_objectid, u64 parent_objectid,
-				  u64 ref_index, char *name, int namelen,
-				  int *search_done)
+				  u64 ref_index, char *name, int namelen)
 {
 	int ret;
 	char *victim_name;
@@ -1016,7 +1097,7 @@ again:
 		 * otherwise they must be unlinked as a conflict
 		 */
 		ptr = btrfs_item_ptr_offset(leaf, path->slots[0]);
-		ptr_end = ptr + btrfs_item_size_nr(leaf, path->slots[0]);
+		ptr_end = ptr + btrfs_item_size(leaf, path->slots[0]);
 		while (ptr < ptr_end) {
 			victim_ref = (struct btrfs_inode_ref *)ptr;
 			victim_name_len = btrfs_inode_ref_name_len(leaf,
@@ -1039,27 +1120,17 @@ again:
 				inc_nlink(&inode->vfs_inode);
 				btrfs_release_path(path);
 
-				ret = btrfs_unlink_inode(trans, root, dir, inode,
+				ret = unlink_inode_for_log_replay(trans, dir, inode,
 						victim_name, victim_name_len);
 				kfree(victim_name);
 				if (ret)
 					return ret;
-				ret = btrfs_run_delayed_items(trans);
-				if (ret)
-					return ret;
-				*search_done = 1;
 				goto again;
 			}
 			kfree(victim_name);
 
 			ptr = (unsigned long)(victim_ref + 1) + victim_name_len;
 		}
-
-		/*
-		 * NOTE: we have searched root tree and checked the
-		 * corresponding ref, it does not need to check again.
-		 */
-		*search_done = 1;
 	}
 	btrfs_release_path(path);
 
@@ -1067,7 +1138,9 @@ again:
 	extref = btrfs_lookup_inode_extref(NULL, root, path, name, namelen,
 					   inode_objectid, parent_objectid, 0,
 					   0);
-	if (!IS_ERR_OR_NULL(extref)) {
+	if (IS_ERR(extref)) {
+		return PTR_ERR(extref);
+	} else if (extref) {
 		u32 item_size;
 		u32 cur_offset = 0;
 		unsigned long base;
@@ -1075,7 +1148,7 @@ again:
 
 		leaf = path->nodes[0];
 
-		item_size = btrfs_item_size_nr(leaf, path->slots[0]);
+		item_size = btrfs_item_size(leaf, path->slots[0]);
 		base = btrfs_item_ptr_offset(leaf, path->slots[0]);
 
 		while (cur_offset < item_size) {
@@ -1101,6 +1174,7 @@ again:
 					     parent_objectid, victim_name,
 					     victim_name_len);
 			if (ret < 0) {
+				kfree(victim_name);
 				return ret;
 			} else if (!ret) {
 				ret = -ENOENT;
@@ -1110,35 +1184,32 @@ again:
 					inc_nlink(&inode->vfs_inode);
 					btrfs_release_path(path);
 
-					ret = btrfs_unlink_inode(trans, root,
+					ret = unlink_inode_for_log_replay(trans,
 							BTRFS_I(victim_parent),
 							inode,
 							victim_name,
 							victim_name_len);
-					if (!ret)
-						ret = btrfs_run_delayed_items(
-								  trans);
 				}
 				iput(victim_parent);
 				kfree(victim_name);
 				if (ret)
 					return ret;
-				*search_done = 1;
 				goto again;
 			}
 			kfree(victim_name);
 next:
 			cur_offset += victim_name_len + sizeof(*extref);
 		}
-		*search_done = 1;
 	}
 	btrfs_release_path(path);
 
 	/* look for a conflicting sequence number */
 	di = btrfs_lookup_dir_index_item(trans, root, path, btrfs_ino(dir),
 					 ref_index, name, namelen, 0);
-	if (di && !IS_ERR(di)) {
-		ret = drop_one_dir_item(trans, root, path, dir, di);
+	if (IS_ERR(di)) {
+		return PTR_ERR(di);
+	} else if (di) {
+		ret = drop_one_dir_item(trans, path, dir, di);
 		if (ret)
 			return ret;
 	}
@@ -1147,8 +1218,10 @@ next:
 	/* look for a conflicting name */
 	di = btrfs_lookup_dir_item(trans, root, path, btrfs_ino(dir),
 				   name, namelen, 0);
-	if (di && !IS_ERR(di)) {
-		ret = drop_one_dir_item(trans, root, path, dir, di);
+	if (IS_ERR(di)) {
+		return PTR_ERR(di);
+	} else if (di) {
+		ret = drop_one_dir_item(trans, path, dir, di);
 		if (ret)
 			return ret;
 	}
@@ -1233,7 +1306,7 @@ again:
 
 	eb = path->nodes[0];
 	ref_ptr = btrfs_item_ptr_offset(eb, path->slots[0]);
-	ref_end = ref_ptr + btrfs_item_size_nr(eb, path->slots[0]);
+	ref_end = ref_ptr + btrfs_item_size(eb, path->slots[0]);
 	while (ref_ptr < ref_end) {
 		char *name = NULL;
 		int namelen;
@@ -1268,7 +1341,7 @@ again:
 				kfree(name);
 				goto out;
 			}
-			ret = btrfs_unlink_inode(trans, root, BTRFS_I(dir),
+			ret = unlink_inode_for_log_replay(trans, BTRFS_I(dir),
 						 inode, name, namelen);
 			kfree(name);
 			iput(dir);
@@ -1290,106 +1363,6 @@ again:
 	return ret;
 }
 
-static int btrfs_inode_ref_exists(struct inode *inode, struct inode *dir,
-				  const u8 ref_type, const char *name,
-				  const int namelen)
-{
-	struct btrfs_key key;
-	struct btrfs_path *path;
-	const u64 parent_id = btrfs_ino(BTRFS_I(dir));
-	int ret;
-
-	path = btrfs_alloc_path();
-	if (!path)
-		return -ENOMEM;
-
-	key.objectid = btrfs_ino(BTRFS_I(inode));
-	key.type = ref_type;
-	if (key.type == BTRFS_INODE_REF_KEY)
-		key.offset = parent_id;
-	else
-		key.offset = btrfs_extref_hash(parent_id, name, namelen);
-
-	ret = btrfs_search_slot(NULL, BTRFS_I(inode)->root, &key, path, 0, 0);
-	if (ret < 0)
-		goto out;
-	if (ret > 0) {
-		ret = 0;
-		goto out;
-	}
-	if (key.type == BTRFS_INODE_EXTREF_KEY)
-		ret = !!btrfs_find_name_in_ext_backref(path->nodes[0],
-				path->slots[0], parent_id, name, namelen);
-	else
-		ret = !!btrfs_find_name_in_backref(path->nodes[0], path->slots[0],
-						   name, namelen);
-
-out:
-	btrfs_free_path(path);
-	return ret;
-}
-
-static int add_link(struct btrfs_trans_handle *trans, struct btrfs_root *root,
-		    struct inode *dir, struct inode *inode, const char *name,
-		    int namelen, u64 ref_index)
-{
-	struct btrfs_dir_item *dir_item;
-	struct btrfs_key key;
-	struct btrfs_path *path;
-	struct inode *other_inode = NULL;
-	int ret;
-
-	path = btrfs_alloc_path();
-	if (!path)
-		return -ENOMEM;
-
-	dir_item = btrfs_lookup_dir_item(NULL, root, path,
-					 btrfs_ino(BTRFS_I(dir)),
-					 name, namelen, 0);
-	if (!dir_item) {
-		btrfs_release_path(path);
-		goto add_link;
-	} else if (IS_ERR(dir_item)) {
-		ret = PTR_ERR(dir_item);
-		goto out;
-	}
-
-	/*
-	 * Our inode's dentry collides with the dentry of another inode which is
-	 * in the log but not yet processed since it has a higher inode number.
-	 * So delete that other dentry.
-	 */
-	btrfs_dir_item_key_to_cpu(path->nodes[0], dir_item, &key);
-	btrfs_release_path(path);
-	other_inode = read_one_inode(root, key.objectid);
-	if (!other_inode) {
-		ret = -ENOENT;
-		goto out;
-	}
-	ret = btrfs_unlink_inode(trans, root, BTRFS_I(dir), BTRFS_I(other_inode),
-				 name, namelen);
-	if (ret)
-		goto out;
-	/*
-	 * If we dropped the link count to 0, bump it so that later the iput()
-	 * on the inode will not free it. We will fixup the link count later.
-	 */
-	if (other_inode->i_nlink == 0)
-		inc_nlink(other_inode);
-
-	ret = btrfs_run_delayed_items(trans);
-	if (ret)
-		goto out;
-add_link:
-	ret = btrfs_add_link(trans, BTRFS_I(dir), BTRFS_I(inode),
-			     name, namelen, 0, ref_index);
-out:
-	iput(other_inode);
-	btrfs_free_path(path);
-
-	return ret;
-}
-
 /*
  * replay one inode back reference item found in the log tree.
  * eb, slot and key refer to the buffer and key found in the log tree.
@@ -1410,7 +1383,6 @@ static noinline int add_inode_ref(struct btrfs_trans_handle *trans,
 	char *name = NULL;
 	int namelen;
 	int ret;
-	int search_done = 0;
 	int log_ref_ver = 0;
 	u64 parent_objectid;
 	u64 inode_objectid;
@@ -1418,7 +1390,7 @@ static noinline int add_inode_ref(struct btrfs_trans_handle *trans,
 	int ref_struct_size;
 
 	ref_ptr = btrfs_item_ptr_offset(eb, slot);
-	ref_end = ref_ptr + btrfs_item_size_nr(eb, slot);
+	ref_end = ref_ptr + btrfs_item_size(eb, slot);
 
 	if (key->type == BTRFS_INODE_EXTREF_KEY) {
 		struct btrfs_inode_extref *r;
@@ -1472,10 +1444,12 @@ static noinline int add_inode_ref(struct btrfs_trans_handle *trans,
 		if (ret)
 			goto out;
 
-		/* if we already have a perfect match, we're done */
-		if (!inode_in_dir(root, path, btrfs_ino(BTRFS_I(dir)),
-					btrfs_ino(BTRFS_I(inode)), ref_index,
-					name, namelen)) {
+		ret = inode_in_dir(root, path, btrfs_ino(BTRFS_I(dir)),
+				   btrfs_ino(BTRFS_I(inode)), ref_index,
+				   name, namelen);
+		if (ret < 0) {
+			goto out;
+		} else if (ret == 0) {
 			/*
 			 * look for a conflicting back reference in the
 			 * metadata. if we find one we have to unlink that name
@@ -1483,56 +1457,27 @@ static noinline int add_inode_ref(struct btrfs_trans_handle *trans,
 			 * overwrite any existing back reference, and we don't
 			 * want to create dangling pointers in the directory.
 			 */
-
-			if (!search_done) {
-				ret = __add_inode_ref(trans, root, path, log,
-						      BTRFS_I(dir),
-						      BTRFS_I(inode),
-						      inode_objectid,
-						      parent_objectid,
-						      ref_index, name, namelen,
-						      &search_done);
-				if (ret) {
-					if (ret == 1)
-						ret = 0;
-					goto out;
-				}
-			}
-
-			/*
-			 * If a reference item already exists for this inode
-			 * with the same parent and name, but different index,
-			 * drop it and the corresponding directory index entries
-			 * from the parent before adding the new reference item
-			 * and dir index entries, otherwise we would fail with
-			 * -EEXIST returned from btrfs_add_link() below.
-			 */
-			ret = btrfs_inode_ref_exists(inode, dir, key->type,
-						     name, namelen);
-			if (ret > 0) {
-				ret = btrfs_unlink_inode(trans, root,
-							 BTRFS_I(dir),
-							 BTRFS_I(inode),
-							 name, namelen);
-				/*
-				 * If we dropped the link count to 0, bump it so
-				 * that later the iput() on the inode will not
-				 * free it. We will fixup the link count later.
-				 */
-				if (!ret && inode->i_nlink == 0)
-					inc_nlink(inode);
-			}
-			if (ret < 0)
+			ret = __add_inode_ref(trans, root, path, log,
+					      BTRFS_I(dir), BTRFS_I(inode),
+					      inode_objectid, parent_objectid,
+					      ref_index, name, namelen);
+			if (ret) {
+				if (ret == 1)
+					ret = 0;
 				goto out;
+			}
 
 			/* insert our name */
-			ret = add_link(trans, root, dir, inode, name, namelen,
-				       ref_index);
+			ret = btrfs_add_link(trans, BTRFS_I(dir), BTRFS_I(inode),
+					     name, namelen, 0, ref_index);
 			if (ret)
 				goto out;
 
-			btrfs_update_inode(trans, root, inode);
+			ret = btrfs_update_inode(trans, root, BTRFS_I(inode));
+			if (ret)
+				goto out;
 		}
+		/* Else, ret == 1, we already have a perfect match, we're done. */
 
 		ref_ptr = (unsigned long)(ref_ptr + ref_struct_size) + namelen;
 		kfree(name);
@@ -1566,18 +1511,6 @@ out:
 	return ret;
 }
 
-static int insert_orphan_item(struct btrfs_trans_handle *trans,
-			      struct btrfs_root *root, u64 ino)
-{
-	int ret;
-
-	ret = btrfs_insert_orphan_item(trans, root, ino);
-	if (ret == -EEXIST)
-		ret = 0;
-
-	return ret;
-}
-
 static int count_inode_extrefs(struct btrfs_root *root,
 		struct btrfs_inode *inode, struct btrfs_path *path)
 {
@@ -1599,7 +1532,7 @@ static int count_inode_extrefs(struct btrfs_root *root,
 			break;
 
 		leaf = path->nodes[0];
-		item_size = btrfs_item_size_nr(leaf, path->slots[0]);
+		item_size = btrfs_item_size(leaf, path->slots[0]);
 		ptr = btrfs_item_ptr_offset(leaf, path->slots[0]);
 		cur_offset = 0;
 
@@ -1653,7 +1586,7 @@ process_slot:
 		    key.type != BTRFS_INODE_REF_KEY)
 			break;
 		ptr = btrfs_item_ptr_offset(path->nodes[0], path->slots[0]);
-		ptr_end = ptr + btrfs_item_size_nr(path->nodes[0],
+		ptr_end = ptr + btrfs_item_size(path->nodes[0],
 						   path->slots[0]);
 		while (ptr < ptr_end) {
 			struct btrfs_inode_ref *ref;
@@ -1718,7 +1651,9 @@ static noinline int fixup_inode_link_count(struct btrfs_trans_handle *trans,
 
 	if (nlink != inode->i_nlink) {
 		set_nlink(inode, nlink);
-		btrfs_update_inode(trans, root, inode);
+		ret = btrfs_update_inode(trans, root, BTRFS_I(inode));
+		if (ret)
+			goto out;
 	}
 	BTRFS_I(inode)->index_cnt = (u64)-1;
 
@@ -1729,7 +1664,9 @@ static noinline int fixup_inode_link_count(struct btrfs_trans_handle *trans,
 			if (ret)
 				goto out;
 		}
-		ret = insert_orphan_item(trans, root, ino);
+		ret = btrfs_insert_orphan_item(trans, root, ino);
+		if (ret == -EEXIST)
+			ret = 0;
 	}
 
 out:
@@ -1754,6 +1691,7 @@ static noinline int fixup_inode_link_counts(struct btrfs_trans_handle *trans,
 			break;
 
 		if (ret == 1) {
+			ret = 0;
 			if (path->slots[0] == 0)
 				break;
 			path->slots[0]--;
@@ -1766,17 +1704,19 @@ static noinline int fixup_inode_link_counts(struct btrfs_trans_handle *trans,
 
 		ret = btrfs_del_item(trans, root, path);
 		if (ret)
-			goto out;
+			break;
 
 		btrfs_release_path(path);
 		inode = read_one_inode(root, key.offset);
-		if (!inode)
-			return -EIO;
+		if (!inode) {
+			ret = -EIO;
+			break;
+		}
 
 		ret = fixup_inode_link_count(trans, root, inode);
 		iput(inode);
 		if (ret)
-			goto out;
+			break;
 
 		/*
 		 * fixup on a directory may create new entries,
@@ -1785,8 +1725,6 @@ static noinline int fixup_inode_link_counts(struct btrfs_trans_handle *trans,
 		 */
 		key.offset = (u64)-1;
 	}
-	ret = 0;
-out:
 	btrfs_release_path(path);
 	return ret;
 }
@@ -1822,11 +1760,9 @@ static noinline int link_to_fixup_dir(struct btrfs_trans_handle *trans,
 			set_nlink(inode, 1);
 		else
 			inc_nlink(inode);
-		ret = btrfs_update_inode(trans, root, inode);
+		ret = btrfs_update_inode(trans, root, BTRFS_I(inode));
 	} else if (ret == -EEXIST) {
 		ret = 0;
-	} else {
-		BUG(); /* Logic Error */
 	}
 	iput(inode);
 
@@ -1868,6 +1804,34 @@ static noinline int insert_one_name(struct btrfs_trans_handle *trans,
 	return ret;
 }
 
+static int delete_conflicting_dir_entry(struct btrfs_trans_handle *trans,
+					struct btrfs_inode *dir,
+					struct btrfs_path *path,
+					struct btrfs_dir_item *dst_di,
+					const struct btrfs_key *log_key,
+					u8 log_type,
+					bool exists)
+{
+	struct btrfs_key found_key;
+
+	btrfs_dir_item_key_to_cpu(path->nodes[0], dst_di, &found_key);
+	/* The existing dentry points to the same inode, don't delete it. */
+	if (found_key.objectid == log_key->objectid &&
+	    found_key.type == log_key->type &&
+	    found_key.offset == log_key->offset &&
+	    btrfs_dir_type(path->nodes[0], dst_di) == log_type)
+		return 1;
+
+	/*
+	 * Don't drop the conflicting directory entry if the inode for the new
+	 * entry doesn't exist.
+	 */
+	if (!exists)
+		return 0;
+
+	return drop_one_dir_item(trans, path, dir, dst_di);
+}
+
 /*
  * take a single entry in a log directory item and replay it into
  * the subvolume.
@@ -1893,14 +1857,17 @@ static noinline int replay_one_name(struct btrfs_trans_handle *trans,
 {
 	char *name;
 	int name_len;
-	struct btrfs_dir_item *dst_di;
-	struct btrfs_key found_key;
+	struct btrfs_dir_item *dir_dst_di;
+	struct btrfs_dir_item *index_dst_di;
+	bool dir_dst_matches = false;
+	bool index_dst_matches = false;
 	struct btrfs_key log_key;
+	struct btrfs_key search_key;
 	struct inode *dir;
 	u8 log_type;
-	int exists;
-	int ret = 0;
-	bool update_size = (key->type == BTRFS_DIR_INDEX_KEY);
+	bool exists;
+	int ret;
+	bool update_size = true;
 	bool name_added = false;
 
 	dir = read_one_inode(root, key->objectid);
@@ -1919,79 +1886,60 @@ static noinline int replay_one_name(struct btrfs_trans_handle *trans,
 		   name_len);
 
 	btrfs_dir_item_key_to_cpu(eb, di, &log_key);
-	exists = btrfs_lookup_inode(trans, root, path, &log_key, 0);
-	if (exists == 0)
-		exists = 1;
-	else
-		exists = 0;
+	ret = btrfs_lookup_inode(trans, root, path, &log_key, 0);
 	btrfs_release_path(path);
+	if (ret < 0)
+		goto out;
+	exists = (ret == 0);
+	ret = 0;
 
-	if (key->type == BTRFS_DIR_ITEM_KEY) {
-		dst_di = btrfs_lookup_dir_item(trans, root, path, key->objectid,
-				       name, name_len, 1);
-	} else if (key->type == BTRFS_DIR_INDEX_KEY) {
-		dst_di = btrfs_lookup_dir_index_item(trans, root, path,
-						     key->objectid,
-						     key->offset, name,
-						     name_len, 1);
-	} else {
-		/* Corruption */
-		ret = -EINVAL;
+	dir_dst_di = btrfs_lookup_dir_item(trans, root, path, key->objectid,
+					   name, name_len, 1);
+	if (IS_ERR(dir_dst_di)) {
+		ret = PTR_ERR(dir_dst_di);
 		goto out;
-	}
-	if (IS_ERR_OR_NULL(dst_di)) {
-		/* we need a sequence number to insert, so we only
-		 * do inserts for the BTRFS_DIR_INDEX_KEY types
-		 */
-		if (key->type != BTRFS_DIR_INDEX_KEY)
+	} else if (dir_dst_di) {
+		ret = delete_conflicting_dir_entry(trans, BTRFS_I(dir), path,
+						   dir_dst_di, &log_key, log_type,
+						   exists);
+		if (ret < 0)
 			goto out;
-		goto insert;
+		dir_dst_matches = (ret == 1);
 	}
 
-	btrfs_dir_item_key_to_cpu(path->nodes[0], dst_di, &found_key);
-	/* the existing item matches the logged item */
-	if (found_key.objectid == log_key.objectid &&
-	    found_key.type == log_key.type &&
-	    found_key.offset == log_key.offset &&
-	    btrfs_dir_type(path->nodes[0], dst_di) == log_type) {
-		update_size = false;
+	btrfs_release_path(path);
+
+	index_dst_di = btrfs_lookup_dir_index_item(trans, root, path,
+						   key->objectid, key->offset,
+						   name, name_len, 1);
+	if (IS_ERR(index_dst_di)) {
+		ret = PTR_ERR(index_dst_di);
 		goto out;
+	} else if (index_dst_di) {
+		ret = delete_conflicting_dir_entry(trans, BTRFS_I(dir), path,
+						   index_dst_di, &log_key,
+						   log_type, exists);
+		if (ret < 0)
+			goto out;
+		index_dst_matches = (ret == 1);
 	}
 
-	/*
-	 * don't drop the conflicting directory entry if the inode
-	 * for the new entry doesn't exist
-	 */
-	if (!exists)
-		goto out;
+	btrfs_release_path(path);
 
-	ret = drop_one_dir_item(trans, root, path, BTRFS_I(dir), dst_di);
-	if (ret)
+	if (dir_dst_matches && index_dst_matches) {
+		ret = 0;
+		update_size = false;
 		goto out;
-
-	if (key->type == BTRFS_DIR_INDEX_KEY)
-		goto insert;
-out:
-	btrfs_release_path(path);
-	if (!ret && update_size) {
-		btrfs_i_size_write(BTRFS_I(dir), dir->i_size + name_len * 2);
-		ret = btrfs_update_inode(trans, root, dir);
 	}
-	kfree(name);
-	iput(dir);
-	if (!ret && name_added)
-		ret = 1;
-	return ret;
 
-insert:
 	/*
 	 * Check if the inode reference exists in the log for the given name,
 	 * inode and parent inode
 	 */
-	found_key.objectid = log_key.objectid;
-	found_key.type = BTRFS_INODE_REF_KEY;
-	found_key.offset = key->objectid;
-	ret = backref_in_log(root->log_root, &found_key, 0, name, name_len);
+	search_key.objectid = log_key.objectid;
+	search_key.type = BTRFS_INODE_REF_KEY;
+	search_key.offset = key->objectid;
+	ret = backref_in_log(root->log_root, &search_key, 0, name, name_len);
 	if (ret < 0) {
 	        goto out;
 	} else if (ret) {
@@ -2001,10 +1949,10 @@ insert:
 	        goto out;
 	}
 
-	found_key.objectid = log_key.objectid;
-	found_key.type = BTRFS_INODE_EXTREF_KEY;
-	found_key.offset = key->objectid;
-	ret = backref_in_log(root->log_root, &found_key, key->objectid, name,
+	search_key.objectid = log_key.objectid;
+	search_key.type = BTRFS_INODE_EXTREF_KEY;
+	search_key.offset = key->objectid;
+	ret = backref_in_log(root->log_root, &search_key, key->objectid, name,
 			     name_len);
 	if (ret < 0) {
 		goto out;
@@ -2023,87 +1971,76 @@ insert:
 		name_added = true;
 	update_size = false;
 	ret = 0;
-	goto out;
+
+out:
+	if (!ret && update_size) {
+		btrfs_i_size_write(BTRFS_I(dir), dir->i_size + name_len * 2);
+		ret = btrfs_update_inode(trans, root, BTRFS_I(dir));
+	}
+	kfree(name);
+	iput(dir);
+	if (!ret && name_added)
+		ret = 1;
+	return ret;
 }
 
-/*
- * find all the names in a directory item and reconcile them into
- * the subvolume.  Only BTRFS_DIR_ITEM_KEY types will have more than
- * one name in a directory item, but the same code gets used for
- * both directory index types
- */
+/* Replay one dir item from a BTRFS_DIR_INDEX_KEY key. */
 static noinline int replay_one_dir_item(struct btrfs_trans_handle *trans,
 					struct btrfs_root *root,
 					struct btrfs_path *path,
 					struct extent_buffer *eb, int slot,
 					struct btrfs_key *key)
 {
-	int ret = 0;
-	u32 item_size = btrfs_item_size_nr(eb, slot);
+	int ret;
 	struct btrfs_dir_item *di;
-	int name_len;
-	unsigned long ptr;
-	unsigned long ptr_end;
-	struct btrfs_path *fixup_path = NULL;
-
-	ptr = btrfs_item_ptr_offset(eb, slot);
-	ptr_end = ptr + item_size;
-	while (ptr < ptr_end) {
-		di = (struct btrfs_dir_item *)ptr;
-		name_len = btrfs_dir_name_len(eb, di);
-		ret = replay_one_name(trans, root, path, eb, di, key);
-		if (ret < 0)
-			break;
-		ptr = (unsigned long)(di + 1);
-		ptr += name_len;
 
-		/*
-		 * If this entry refers to a non-directory (directories can not
-		 * have a link count > 1) and it was added in the transaction
-		 * that was not committed, make sure we fixup the link count of
-		 * the inode it the entry points to. Otherwise something like
-		 * the following would result in a directory pointing to an
-		 * inode with a wrong link that does not account for this dir
-		 * entry:
-		 *
-		 * mkdir testdir
-		 * touch testdir/foo
-		 * touch testdir/bar
-		 * sync
-		 *
-		 * ln testdir/bar testdir/bar_link
-		 * ln testdir/foo testdir/foo_link
-		 * xfs_io -c "fsync" testdir/bar
-		 *
-		 * <power failure>
-		 *
-		 * mount fs, log replay happens
-		 *
-		 * File foo would remain with a link count of 1 when it has two
-		 * entries pointing to it in the directory testdir. This would
-		 * make it impossible to ever delete the parent directory has
-		 * it would result in stale dentries that can never be deleted.
-		 */
-		if (ret == 1 && btrfs_dir_type(eb, di) != BTRFS_FT_DIR) {
-			struct btrfs_key di_key;
+	/* We only log dir index keys, which only contain a single dir item. */
+	ASSERT(key->type == BTRFS_DIR_INDEX_KEY);
 
-			if (!fixup_path) {
-				fixup_path = btrfs_alloc_path();
-				if (!fixup_path) {
-					ret = -ENOMEM;
-					break;
-				}
-			}
+	di = btrfs_item_ptr(eb, slot, struct btrfs_dir_item);
+	ret = replay_one_name(trans, root, path, eb, di, key);
+	if (ret < 0)
+		return ret;
 
-			btrfs_dir_item_key_to_cpu(eb, di, &di_key);
-			ret = link_to_fixup_dir(trans, root, fixup_path,
-						di_key.objectid);
-			if (ret)
-				break;
-		}
-		ret = 0;
+	/*
+	 * If this entry refers to a non-directory (directories can not have a
+	 * link count > 1) and it was added in the transaction that was not
+	 * committed, make sure we fixup the link count of the inode the entry
+	 * points to. Otherwise something like the following would result in a
+	 * directory pointing to an inode with a wrong link that does not account
+	 * for this dir entry:
+	 *
+	 * mkdir testdir
+	 * touch testdir/foo
+	 * touch testdir/bar
+	 * sync
+	 *
+	 * ln testdir/bar testdir/bar_link
+	 * ln testdir/foo testdir/foo_link
+	 * xfs_io -c "fsync" testdir/bar
+	 *
+	 * <power failure>
+	 *
+	 * mount fs, log replay happens
+	 *
+	 * File foo would remain with a link count of 1 when it has two entries
+	 * pointing to it in the directory testdir. This would make it impossible
+	 * to ever delete the parent directory has it would result in stale
+	 * dentries that can never be deleted.
+	 */
+	if (ret == 1 && btrfs_dir_type(eb, di) != BTRFS_FT_DIR) {
+		struct btrfs_path *fixup_path;
+		struct btrfs_key di_key;
+
+		fixup_path = btrfs_alloc_path();
+		if (!fixup_path)
+			return -ENOMEM;
+
+		btrfs_dir_item_key_to_cpu(eb, di, &di_key);
+		ret = link_to_fixup_dir(trans, root, fixup_path, di_key.objectid);
+		btrfs_free_path(fixup_path);
 	}
-	btrfs_free_path(fixup_path);
+
 	return ret;
 }
 
@@ -2120,7 +2057,7 @@ static noinline int replay_one_dir_item(struct btrfs_trans_handle *trans,
  */
 static noinline int find_dir_range(struct btrfs_root *root,
 				   struct btrfs_path *path,
-				   u64 dirid, int key_type,
+				   u64 dirid,
 				   u64 *start_ret, u64 *end_ret)
 {
 	struct btrfs_key key;
@@ -2133,7 +2070,7 @@ static noinline int find_dir_range(struct btrfs_root *root,
 		return 1;
 
 	key.objectid = dirid;
-	key.type = key_type;
+	key.type = BTRFS_DIR_LOG_INDEX_KEY;
 	key.offset = *start_ret;
 
 	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
@@ -2147,7 +2084,7 @@ static noinline int find_dir_range(struct btrfs_root *root,
 	if (ret != 0)
 		btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
 
-	if (key.type != key_type || key.objectid != dirid) {
+	if (key.type != BTRFS_DIR_LOG_INDEX_KEY || key.objectid != dirid) {
 		ret = 1;
 		goto next;
 	}
@@ -2174,7 +2111,7 @@ next:
 
 	btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
 
-	if (key.type != key_type || key.objectid != dirid) {
+	if (key.type != BTRFS_DIR_LOG_INDEX_KEY || key.objectid != dirid) {
 		ret = 1;
 		goto out;
 	}
@@ -2195,105 +2132,85 @@ out:
  * to is unlinked
  */
 static noinline int check_item_in_log(struct btrfs_trans_handle *trans,
-				      struct btrfs_root *root,
 				      struct btrfs_root *log,
 				      struct btrfs_path *path,
 				      struct btrfs_path *log_path,
 				      struct inode *dir,
 				      struct btrfs_key *dir_key)
 {
+	struct btrfs_root *root = BTRFS_I(dir)->root;
 	int ret;
 	struct extent_buffer *eb;
 	int slot;
-	u32 item_size;
 	struct btrfs_dir_item *di;
-	struct btrfs_dir_item *log_di;
 	int name_len;
-	unsigned long ptr;
-	unsigned long ptr_end;
 	char *name;
-	struct inode *inode;
+	struct inode *inode = NULL;
 	struct btrfs_key location;
 
-again:
+	/*
+	 * Currently we only log dir index keys. Even if we replay a log created
+	 * by an older kernel that logged both dir index and dir item keys, all
+	 * we need to do is process the dir index keys, we (and our caller) can
+	 * safely ignore dir item keys (key type BTRFS_DIR_ITEM_KEY).
+	 */
+	ASSERT(dir_key->type == BTRFS_DIR_INDEX_KEY);
+
 	eb = path->nodes[0];
 	slot = path->slots[0];
-	item_size = btrfs_item_size_nr(eb, slot);
-	ptr = btrfs_item_ptr_offset(eb, slot);
-	ptr_end = ptr + item_size;
-	while (ptr < ptr_end) {
-		di = (struct btrfs_dir_item *)ptr;
-		name_len = btrfs_dir_name_len(eb, di);
-		name = kmalloc(name_len, GFP_NOFS);
-		if (!name) {
-			ret = -ENOMEM;
-			goto out;
-		}
-		read_extent_buffer(eb, name, (unsigned long)(di + 1),
-				  name_len);
-		log_di = NULL;
-		if (log && dir_key->type == BTRFS_DIR_ITEM_KEY) {
-			log_di = btrfs_lookup_dir_item(trans, log, log_path,
-						       dir_key->objectid,
-						       name, name_len, 0);
-		} else if (log && dir_key->type == BTRFS_DIR_INDEX_KEY) {
-			log_di = btrfs_lookup_dir_index_item(trans, log,
-						     log_path,
-						     dir_key->objectid,
-						     dir_key->offset,
-						     name, name_len, 0);
-		}
-		if (!log_di || log_di == ERR_PTR(-ENOENT)) {
-			btrfs_dir_item_key_to_cpu(eb, di, &location);
-			btrfs_release_path(path);
-			btrfs_release_path(log_path);
-			inode = read_one_inode(root, location.objectid);
-			if (!inode) {
-				kfree(name);
-				return -EIO;
-			}
+	di = btrfs_item_ptr(eb, slot, struct btrfs_dir_item);
+	name_len = btrfs_dir_name_len(eb, di);
+	name = kmalloc(name_len, GFP_NOFS);
+	if (!name) {
+		ret = -ENOMEM;
+		goto out;
+	}
 
-			ret = link_to_fixup_dir(trans, root,
-						path, location.objectid);
-			if (ret) {
-				kfree(name);
-				iput(inode);
-				goto out;
-			}
+	read_extent_buffer(eb, name, (unsigned long)(di + 1), name_len);
 
-			inc_nlink(inode);
-			ret = btrfs_unlink_inode(trans, root, BTRFS_I(dir),
-					BTRFS_I(inode), name, name_len);
-			if (!ret)
-				ret = btrfs_run_delayed_items(trans);
-			kfree(name);
-			iput(inode);
-			if (ret)
-				goto out;
+	if (log) {
+		struct btrfs_dir_item *log_di;
 
-			/* there might still be more names under this key
-			 * check and repeat if required
-			 */
-			ret = btrfs_search_slot(NULL, root, dir_key, path,
-						0, 0);
-			if (ret == 0)
-				goto again;
+		log_di = btrfs_lookup_dir_index_item(trans, log, log_path,
+						     dir_key->objectid,
+						     dir_key->offset,
+						     name, name_len, 0);
+		if (IS_ERR(log_di)) {
+			ret = PTR_ERR(log_di);
+			goto out;
+		} else if (log_di) {
+			/* The dentry exists in the log, we have nothing to do. */
 			ret = 0;
 			goto out;
-		} else if (IS_ERR(log_di)) {
-			kfree(name);
-			return PTR_ERR(log_di);
 		}
-		btrfs_release_path(log_path);
-		kfree(name);
+	}
 
-		ptr = (unsigned long)(di + 1);
-		ptr += name_len;
+	btrfs_dir_item_key_to_cpu(eb, di, &location);
+	btrfs_release_path(path);
+	btrfs_release_path(log_path);
+	inode = read_one_inode(root, location.objectid);
+	if (!inode) {
+		ret = -EIO;
+		goto out;
 	}
-	ret = 0;
+
+	ret = link_to_fixup_dir(trans, root, path, location.objectid);
+	if (ret)
+		goto out;
+
+	inc_nlink(inode);
+	ret = unlink_inode_for_log_replay(trans, BTRFS_I(dir), BTRFS_I(inode),
+					  name, name_len);
+	/*
+	 * Unlike dir item keys, dir index keys can only have one name (entry) in
+	 * them, as there are no key collisions since each key has a unique offset
+	 * (an index number), so we're done.
+	 */
 out:
 	btrfs_release_path(path);
 	btrfs_release_path(log_path);
+	kfree(name);
+	iput(inode);
 	return ret;
 }
 
@@ -2336,7 +2253,7 @@ process_leaf:
 		}
 
 		di = btrfs_item_ptr(path->nodes[0], i, struct btrfs_dir_item);
-		total_size = btrfs_item_size_nr(path->nodes[0], i);
+		total_size = btrfs_item_size(path->nodes[0], i);
 		cur = 0;
 		while (cur < total_size) {
 			u16 name_len = btrfs_dir_name_len(path->nodes[0], di);
@@ -2413,7 +2330,6 @@ static noinline int replay_dir_deletes(struct btrfs_trans_handle *trans,
 {
 	u64 range_start;
 	u64 range_end;
-	int key_type = BTRFS_DIR_LOG_ITEM_KEY;
 	int ret = 0;
 	struct btrfs_key dir_key;
 	struct btrfs_key found_key;
@@ -2421,7 +2337,7 @@ static noinline int replay_dir_deletes(struct btrfs_trans_handle *trans,
 	struct inode *dir;
 
 	dir_key.objectid = dirid;
-	dir_key.type = BTRFS_DIR_ITEM_KEY;
+	dir_key.type = BTRFS_DIR_INDEX_KEY;
 	log_path = btrfs_alloc_path();
 	if (!log_path)
 		return -ENOMEM;
@@ -2435,16 +2351,18 @@ static noinline int replay_dir_deletes(struct btrfs_trans_handle *trans,
 		btrfs_free_path(log_path);
 		return 0;
 	}
-again:
+
 	range_start = 0;
 	range_end = 0;
 	while (1) {
 		if (del_all)
 			range_end = (u64)-1;
 		else {
-			ret = find_dir_range(log, path, dirid, key_type,
+			ret = find_dir_range(log, path, dirid,
 					     &range_start, &range_end);
-			if (ret != 0)
+			if (ret < 0)
+				goto out;
+			else if (ret > 0)
 				break;
 		}
 
@@ -2467,13 +2385,15 @@ again:
 			btrfs_item_key_to_cpu(path->nodes[0], &found_key,
 					      path->slots[0]);
 			if (found_key.objectid != dirid ||
-			    found_key.type != dir_key.type)
-				goto next_type;
+			    found_key.type != dir_key.type) {
+				ret = 0;
+				goto out;
+			}
 
 			if (found_key.offset > range_end)
 				break;
 
-			ret = check_item_in_log(trans, root, log, path,
+			ret = check_item_in_log(trans, log, path,
 						log_path, dir,
 						&found_key);
 			if (ret)
@@ -2487,15 +2407,7 @@ again:
 			break;
 		range_start = range_end + 1;
 	}
-
-next_type:
 	ret = 0;
-	if (key_type == BTRFS_DIR_LOG_ITEM_KEY) {
-		key_type = BTRFS_DIR_LOG_INDEX_KEY;
-		dir_key.type = BTRFS_DIR_INDEX_KEY;
-		btrfs_release_path(path);
-		goto again;
-	}
 out:
 	btrfs_release_path(path);
 	btrfs_free_path(log_path);
@@ -2524,7 +2436,7 @@ static int replay_one_buffer(struct btrfs_root *log, struct extent_buffer *eb,
 	int i;
 	int ret;
 
-	ret = btrfs_read_buffer(eb, gen, level, NULL);
+	ret = btrfs_read_extent_buffer(eb, gen, level, NULL);
 	if (ret)
 		return ret;
 
@@ -2588,6 +2500,7 @@ static int replay_one_buffer(struct btrfs_root *log, struct extent_buffer *eb,
 			 * those prealloc extents just after replaying them.
 			 */
 			if (S_ISREG(mode)) {
+				struct btrfs_drop_extents_args drop_args = { 0 };
 				struct inode *inode;
 				u64 from;
 
@@ -2598,12 +2511,18 @@ static int replay_one_buffer(struct btrfs_root *log, struct extent_buffer *eb,
 				}
 				from = ALIGN(i_size_read(inode),
 					     root->fs_info->sectorsize);
-				ret = btrfs_drop_extents(wc->trans, root, inode,
-							 from, (u64)-1, 1);
+				drop_args.start = from;
+				drop_args.end = (u64)-1;
+				drop_args.drop_cache = true;
+				ret = btrfs_drop_extents(wc->trans, root,
+							 BTRFS_I(inode),
+							 &drop_args);
 				if (!ret) {
+					inode_sub_bytes(inode,
+							drop_args.bytes_found);
 					/* Update the inode's nbytes. */
 					ret = btrfs_update_inode(wc->trans,
-								 root, inode);
+							root, BTRFS_I(inode));
 				}
 				iput(inode);
 				if (ret)
@@ -2648,29 +2567,51 @@ static int replay_one_buffer(struct btrfs_root *log, struct extent_buffer *eb,
 						eb, i, &key);
 			if (ret)
 				break;
-		} else if (key.type == BTRFS_DIR_ITEM_KEY) {
-			ret = replay_one_dir_item(wc->trans, root, path,
-						  eb, i, &key);
-			if (ret)
-				break;
 		}
+		/*
+		 * We don't log BTRFS_DIR_ITEM_KEY keys anymore, only the
+		 * BTRFS_DIR_INDEX_KEY items which we use to derive the
+		 * BTRFS_DIR_ITEM_KEY items. If we are replaying a log from an
+		 * older kernel with such keys, ignore them.
+		 */
 	}
 	btrfs_free_path(path);
 	return ret;
 }
 
+/*
+ * Correctly adjust the reserved bytes occupied by a log tree extent buffer
+ */
+static void unaccount_log_buffer(struct btrfs_fs_info *fs_info, u64 start)
+{
+	struct btrfs_block_group *cache;
+
+	cache = btrfs_lookup_block_group(fs_info, start);
+	if (!cache) {
+		btrfs_err(fs_info, "unable to find block group for %llu", start);
+		return;
+	}
+
+	spin_lock(&cache->space_info->lock);
+	spin_lock(&cache->lock);
+	cache->reserved -= fs_info->nodesize;
+	cache->space_info->bytes_reserved -= fs_info->nodesize;
+	spin_unlock(&cache->lock);
+	spin_unlock(&cache->space_info->lock);
+
+	btrfs_put_block_group(cache);
+}
+
 static noinline int walk_down_log_tree(struct btrfs_trans_handle *trans,
 				   struct btrfs_root *root,
 				   struct btrfs_path *path, int *level,
 				   struct walk_control *wc)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
-	u64 root_owner;
 	u64 bytenr;
 	u64 ptr_gen;
 	struct extent_buffer *next;
 	struct extent_buffer *cur;
-	struct extent_buffer *parent;
 	u32 blocksize;
 	int ret = 0;
 
@@ -2690,10 +2631,9 @@ static noinline int walk_down_log_tree(struct btrfs_trans_handle *trans,
 		btrfs_node_key_to_cpu(cur, &first_key, path->slots[*level]);
 		blocksize = fs_info->nodesize;
 
-		parent = path->nodes[*level];
-		root_owner = btrfs_header_owner(parent);
-
-		next = btrfs_find_create_tree_block(fs_info, bytenr);
+		next = btrfs_find_create_tree_block(fs_info, bytenr,
+						    btrfs_header_owner(cur),
+						    *level - 1);
 		if (IS_ERR(next))
 			return PTR_ERR(next);
 
@@ -2707,7 +2647,7 @@ static noinline int walk_down_log_tree(struct btrfs_trans_handle *trans,
 
 			path->slots[*level]++;
 			if (wc->free) {
-				ret = btrfs_read_buffer(next, ptr_gen,
+				ret = btrfs_read_extent_buffer(next, ptr_gen,
 							*level - 1, &first_key);
 				if (ret) {
 					free_extent_buffer(next);
@@ -2716,28 +2656,27 @@ static noinline int walk_down_log_tree(struct btrfs_trans_handle *trans,
 
 				if (trans) {
 					btrfs_tree_lock(next);
-					btrfs_set_lock_blocking_write(next);
 					btrfs_clean_tree_block(next);
 					btrfs_wait_tree_block_writeback(next);
 					btrfs_tree_unlock(next);
+					ret = btrfs_pin_reserved_extent(trans,
+							bytenr, blocksize);
+					if (ret) {
+						free_extent_buffer(next);
+						return ret;
+					}
+					btrfs_redirty_list_add(
+						trans->transaction, next);
 				} else {
 					if (test_and_clear_bit(EXTENT_BUFFER_DIRTY, &next->bflags))
 						clear_extent_buffer_dirty(next);
-				}
-
-				WARN_ON(root_owner !=
-					BTRFS_TREE_LOG_OBJECTID);
-				ret = btrfs_pin_reserved_extent(fs_info,
-							bytenr, blocksize);
-				if (ret) {
-					free_extent_buffer(next);
-					return ret;
+					unaccount_log_buffer(fs_info, bytenr);
 				}
 			}
 			free_extent_buffer(next);
 			continue;
 		}
-		ret = btrfs_read_buffer(next, ptr_gen, *level - 1, &first_key);
+		ret = btrfs_read_extent_buffer(next, ptr_gen, *level - 1, &first_key);
 		if (ret) {
 			free_extent_buffer(next);
 			return ret;
@@ -2762,7 +2701,6 @@ static noinline int walk_up_log_tree(struct btrfs_trans_handle *trans,
 				 struct walk_control *wc)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
-	u64 root_owner;
 	int i;
 	int slot;
 	int ret;
@@ -2775,13 +2713,6 @@ static noinline int walk_up_log_tree(struct btrfs_trans_handle *trans,
 			WARN_ON(*level == 0);
 			return 0;
 		} else {
-			struct extent_buffer *parent;
-			if (path->nodes[*level] == root->node)
-				parent = path->nodes[*level];
-			else
-				parent = path->nodes[*level + 1];
-
-			root_owner = btrfs_header_owner(parent);
 			ret = wc->process_func(root, path->nodes[*level], wc,
 				 btrfs_header_generation(path->nodes[*level]),
 				 *level);
@@ -2795,21 +2726,23 @@ static noinline int walk_up_log_tree(struct btrfs_trans_handle *trans,
 
 				if (trans) {
 					btrfs_tree_lock(next);
-					btrfs_set_lock_blocking_write(next);
 					btrfs_clean_tree_block(next);
 					btrfs_wait_tree_block_writeback(next);
 					btrfs_tree_unlock(next);
+					ret = btrfs_pin_reserved_extent(trans,
+						     path->nodes[*level]->start,
+						     path->nodes[*level]->len);
+					if (ret)
+						return ret;
+					btrfs_redirty_list_add(trans->transaction,
+							       next);
 				} else {
 					if (test_and_clear_bit(EXTENT_BUFFER_DIRTY, &next->bflags))
 						clear_extent_buffer_dirty(next);
-				}
 
-				WARN_ON(root_owner != BTRFS_TREE_LOG_OBJECTID);
-				ret = btrfs_pin_reserved_extent(fs_info,
-						path->nodes[*level]->start,
-						path->nodes[*level]->len);
-				if (ret)
-					return ret;
+					unaccount_log_buffer(fs_info,
+						path->nodes[*level]->start);
+				}
 			}
 			free_extent_buffer(path->nodes[*level]);
 			path->nodes[*level] = NULL;
@@ -2876,19 +2809,19 @@ static int walk_log_tree(struct btrfs_trans_handle *trans,
 
 			if (trans) {
 				btrfs_tree_lock(next);
-				btrfs_set_lock_blocking_write(next);
 				btrfs_clean_tree_block(next);
 				btrfs_wait_tree_block_writeback(next);
 				btrfs_tree_unlock(next);
+				ret = btrfs_pin_reserved_extent(trans,
+						next->start, next->len);
+				if (ret)
+					goto out;
+				btrfs_redirty_list_add(trans->transaction, next);
 			} else {
 				if (test_and_clear_bit(EXTENT_BUFFER_DIRTY, &next->bflags))
 					clear_extent_buffer_dirty(next);
+				unaccount_log_buffer(fs_info, next->start);
 			}
-
-			ret = btrfs_pin_reserved_extent(fs_info, next->start,
-							next->len);
-			if (ret)
-				goto out;
 		}
 	}
 
@@ -2964,9 +2897,6 @@ static void wait_for_writer(struct btrfs_root *root)
 static inline void btrfs_remove_log_ctx(struct btrfs_root *root,
 					struct btrfs_log_ctx *ctx)
 {
-	if (!ctx)
-		return;
-
 	mutex_lock(&root->log_mutex);
 	list_del_init(&ctx->list);
 	mutex_unlock(&root->log_mutex);
@@ -2986,8 +2916,6 @@ static inline void btrfs_remove_all_log_ctxs(struct btrfs_root *root,
 		list_del_init(&ctx->list);
 		ctx->log_ret = error;
 	}
-
-	INIT_LIST_HEAD(&root->log_ctxs[index]);
 }
 
 /*
@@ -3016,6 +2944,8 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 	int log_transid = 0;
 	struct btrfs_log_ctx root_log_ctx;
 	struct blk_plug plug;
+	u64 log_root_start;
+	u64 log_root_level;
 
 	mutex_lock(&root->log_mutex);
 	log_transid = ctx->log_transid;
@@ -3053,7 +2983,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 
 	/* bail out if we need to do a full commit */
 	if (btrfs_need_log_full_commit(trans)) {
-		ret = -EAGAIN;
+		ret = BTRFS_LOG_FORCE_COMMIT;
 		mutex_unlock(&root->log_mutex);
 		goto out;
 	}
@@ -3068,6 +2998,17 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 	 */
 	blk_start_plug(&plug);
 	ret = btrfs_write_marked_extents(fs_info, &log->dirty_log_pages, mark);
+	/*
+	 * -EAGAIN happens when someone, e.g., a concurrent transaction
+	 *  commit, writes a dirty extent in this tree-log commit. This
+	 *  concurrent write will create a hole writing out the extents,
+	 *  and we cannot proceed on a zoned filesystem, requiring
+	 *  sequential writing. While we can bail out to a full commit
+	 *  here, but we can continue hoping the concurrent writing fills
+	 *  the hole.
+	 */
+	if (ret == -EAGAIN && btrfs_is_zoned(fs_info))
+		ret = 0;
 	if (ret) {
 		blk_finish_plug(&plug);
 		btrfs_abort_transaction(trans, ret);
@@ -3102,32 +3043,33 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 	 */
 	mutex_unlock(&root->log_mutex);
 
+	if (btrfs_is_zoned(fs_info)) {
+		mutex_lock(&fs_info->tree_root->log_mutex);
+		if (!log_root_tree->node) {
+			ret = btrfs_alloc_log_tree_node(trans, log_root_tree);
+			if (ret) {
+				mutex_unlock(&fs_info->tree_root->log_mutex);
+				blk_finish_plug(&plug);
+				goto out;
+			}
+		}
+		mutex_unlock(&fs_info->tree_root->log_mutex);
+	}
+
 	btrfs_init_log_ctx(&root_log_ctx, NULL);
 
 	mutex_lock(&log_root_tree->log_mutex);
-	atomic_inc(&log_root_tree->log_batch);
-	atomic_inc(&log_root_tree->log_writers);
 
 	index2 = log_root_tree->log_transid % 2;
 	list_add_tail(&root_log_ctx.list, &log_root_tree->log_ctxs[index2]);
 	root_log_ctx.log_transid = log_root_tree->log_transid;
 
-	mutex_unlock(&log_root_tree->log_mutex);
-
-	mutex_lock(&log_root_tree->log_mutex);
-
 	/*
 	 * Now we are safe to update the log_root_tree because we're under the
 	 * log_mutex, and we're a current writer so we're holding the commit
 	 * open until we drop the log_mutex.
 	 */
 	ret = update_log_root(trans, log, &new_root_item);
-
-	if (atomic_dec_and_test(&log_root_tree->log_writers)) {
-		/* atomic_dec_and_test implies a barrier */
-		cond_wake_up_nomb(&log_root_tree->log_writer_wait);
-	}
-
 	if (ret) {
 		if (!list_empty(&root_log_ctx.list))
 			list_del_init(&root_log_ctx.list);
@@ -3142,7 +3084,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 		}
 		btrfs_wait_tree_log_extents(log, mark);
 		mutex_unlock(&log_root_tree->log_mutex);
-		ret = -EAGAIN;
+		ret = BTRFS_LOG_FORCE_COMMIT;
 		goto out;
 	}
 
@@ -3173,8 +3115,6 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 				root_log_ctx.log_transid - 1);
 	}
 
-	wait_for_writer(log_root_tree);
-
 	/*
 	 * now that we've moved on to the tree of log tree roots,
 	 * check the full commit flag again
@@ -3183,7 +3123,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 		blk_finish_plug(&plug);
 		btrfs_wait_tree_log_extents(log, mark);
 		mutex_unlock(&log_root_tree->log_mutex);
-		ret = -EAGAIN;
+		ret = BTRFS_LOG_FORCE_COMMIT;
 		goto out_wake_log_root;
 	}
 
@@ -3191,7 +3131,17 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 					 &log_root_tree->dirty_log_pages,
 					 EXTENT_DIRTY | EXTENT_NEW);
 	blk_finish_plug(&plug);
-	if (ret) {
+	/*
+	 * As described above, -EAGAIN indicates a hole in the extents. We
+	 * cannot wait for these write outs since the waiting cause a
+	 * deadlock. Bail out to the full commit instead.
+	 */
+	if (ret == -EAGAIN && btrfs_is_zoned(fs_info)) {
+		btrfs_set_log_full_commit(trans);
+		btrfs_wait_tree_log_extents(log, mark);
+		mutex_unlock(&log_root_tree->log_mutex);
+		goto out_wake_log_root;
+	} else if (ret) {
 		btrfs_set_log_full_commit(trans);
 		btrfs_abort_transaction(trans, ret);
 		mutex_unlock(&log_root_tree->log_mutex);
@@ -3207,32 +3157,63 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 		goto out_wake_log_root;
 	}
 
-	btrfs_set_super_log_root(fs_info->super_for_commit,
-				 log_root_tree->node->start);
-	btrfs_set_super_log_root_level(fs_info->super_for_commit,
-				       btrfs_header_level(log_root_tree->node));
-
+	log_root_start = log_root_tree->node->start;
+	log_root_level = btrfs_header_level(log_root_tree->node);
 	log_root_tree->log_transid++;
 	mutex_unlock(&log_root_tree->log_mutex);
 
 	/*
-	 * Nobody else is going to jump in and write the ctree
-	 * super here because the log_commit atomic below is protecting
-	 * us.  We must be called with a transaction handle pinning
-	 * the running transaction open, so a full commit can't hop
-	 * in and cause problems either.
+	 * Here we are guaranteed that nobody is going to write the superblock
+	 * for the current transaction before us and that neither we do write
+	 * our superblock before the previous transaction finishes its commit
+	 * and writes its superblock, because:
+	 *
+	 * 1) We are holding a handle on the current transaction, so no body
+	 *    can commit it until we release the handle;
+	 *
+	 * 2) Before writing our superblock we acquire the tree_log_mutex, so
+	 *    if the previous transaction is still committing, and hasn't yet
+	 *    written its superblock, we wait for it to do it, because a
+	 *    transaction commit acquires the tree_log_mutex when the commit
+	 *    begins and releases it only after writing its superblock.
+	 */
+	mutex_lock(&fs_info->tree_log_mutex);
+
+	/*
+	 * The previous transaction writeout phase could have failed, and thus
+	 * marked the fs in an error state.  We must not commit here, as we
+	 * could have updated our generation in the super_for_commit and
+	 * writing the super here would result in transid mismatches.  If there
+	 * is an error here just bail.
 	 */
+	if (BTRFS_FS_ERROR(fs_info)) {
+		ret = -EIO;
+		btrfs_set_log_full_commit(trans);
+		btrfs_abort_transaction(trans, ret);
+		mutex_unlock(&fs_info->tree_log_mutex);
+		goto out_wake_log_root;
+	}
+
+	btrfs_set_super_log_root(fs_info->super_for_commit, log_root_start);
+	btrfs_set_super_log_root_level(fs_info->super_for_commit, log_root_level);
 	ret = write_all_supers(fs_info, 1);
+	mutex_unlock(&fs_info->tree_log_mutex);
 	if (ret) {
 		btrfs_set_log_full_commit(trans);
 		btrfs_abort_transaction(trans, ret);
 		goto out_wake_log_root;
 	}
 
-	mutex_lock(&root->log_mutex);
-	if (root->last_log_commit < log_transid)
-		root->last_log_commit = log_transid;
-	mutex_unlock(&root->log_mutex);
+	/*
+	 * We know there can only be one task here, since we have not yet set
+	 * root->log_commit[index1] to 0 and any task attempting to sync the
+	 * log must wait for the previous log transaction to commit if it's
+	 * still in progress or wait for the current log transaction commit if
+	 * someone else already started it. We use <= and not < because the
+	 * first log transaction has an ID of 0.
+	 */
+	ASSERT(root->last_log_commit <= log_transid);
+	root->last_log_commit = log_transid;
 
 out_wake_log_root:
 	mutex_lock(&log_root_tree->log_mutex);
@@ -3273,18 +3254,44 @@ static void free_log_tree(struct btrfs_trans_handle *trans,
 		.process_func = process_one_buffer
 	};
 
-	ret = walk_log_tree(trans, log, &wc);
-	if (ret) {
-		if (trans)
-			btrfs_abort_transaction(trans, ret);
-		else
-			btrfs_handle_fs_error(log->fs_info, ret, NULL);
+	if (log->node) {
+		ret = walk_log_tree(trans, log, &wc);
+		if (ret) {
+			/*
+			 * We weren't able to traverse the entire log tree, the
+			 * typical scenario is getting an -EIO when reading an
+			 * extent buffer of the tree, due to a previous writeback
+			 * failure of it.
+			 */
+			set_bit(BTRFS_FS_STATE_LOG_CLEANUP_ERROR,
+				&log->fs_info->fs_state);
+
+			/*
+			 * Some extent buffers of the log tree may still be dirty
+			 * and not yet written back to storage, because we may
+			 * have updates to a log tree without syncing a log tree,
+			 * such as during rename and link operations. So flush
+			 * them out and wait for their writeback to complete, so
+			 * that we properly cleanup their state and pages.
+			 */
+			btrfs_write_marked_extents(log->fs_info,
+						   &log->dirty_log_pages,
+						   EXTENT_DIRTY | EXTENT_NEW);
+			btrfs_wait_tree_log_extents(log,
+						    EXTENT_DIRTY | EXTENT_NEW);
+
+			if (trans)
+				btrfs_abort_transaction(trans, ret);
+			else
+				btrfs_handle_fs_error(log->fs_info, ret, NULL);
+		}
 	}
 
 	clear_extent_bits(&log->dirty_log_pages, 0, (u64)-1,
 			  EXTENT_DIRTY | EXTENT_NEW | EXTENT_NEED_WAIT);
-	free_extent_buffer(log->node);
-	kfree(log);
+	extent_io_tree_release(&log->log_csum_range);
+
+	btrfs_put_root(log);
 }
 
 /*
@@ -3296,6 +3303,7 @@ int btrfs_free_log(struct btrfs_trans_handle *trans, struct btrfs_root *root)
 	if (root->log_root) {
 		free_log_tree(trans, root->log_root);
 		root->log_root = NULL;
+		clear_bit(BTRFS_ROOT_HAS_LOG_TREE, &root->state);
 	}
 	return 0;
 }
@@ -3306,32 +3314,162 @@ int btrfs_free_log_root_tree(struct btrfs_trans_handle *trans,
 	if (fs_info->log_root_tree) {
 		free_log_tree(trans, fs_info->log_root_tree);
 		fs_info->log_root_tree = NULL;
+		clear_bit(BTRFS_ROOT_HAS_LOG_TREE, &fs_info->tree_root->state);
 	}
 	return 0;
 }
 
 /*
- * Check if an inode was logged in the current transaction. We can't always rely
- * on an inode's logged_trans value, because it's an in-memory only field and
- * therefore not persisted. This means that its value is lost if the inode gets
- * evicted and loaded again from disk (in which case it has a value of 0, and
- * certainly it is smaller then any possible transaction ID), when that happens
- * the full_sync flag is set in the inode's runtime flags, so on that case we
- * assume eviction happened and ignore the logged_trans value, assuming the
- * worst case, that the inode was logged before in the current transaction.
+ * Check if an inode was logged in the current transaction. This correctly deals
+ * with the case where the inode was logged but has a logged_trans of 0, which
+ * happens if the inode is evicted and loaded again, as logged_trans is an in
+ * memory only field (not persisted).
+ *
+ * Returns 1 if the inode was logged before in the transaction, 0 if it was not,
+ * and < 0 on error.
  */
-static bool inode_logged(struct btrfs_trans_handle *trans,
-			 struct btrfs_inode *inode)
+static int inode_logged(struct btrfs_trans_handle *trans,
+			struct btrfs_inode *inode,
+			struct btrfs_path *path_in)
 {
+	struct btrfs_path *path = path_in;
+	struct btrfs_key key;
+	int ret;
+
 	if (inode->logged_trans == trans->transid)
-		return true;
+		return 1;
+
+	/*
+	 * If logged_trans is not 0, then we know the inode logged was not logged
+	 * in this transaction, so we can return false right away.
+	 */
+	if (inode->logged_trans > 0)
+		return 0;
+
+	/*
+	 * If no log tree was created for this root in this transaction, then
+	 * the inode can not have been logged in this transaction. In that case
+	 * set logged_trans to anything greater than 0 and less than the current
+	 * transaction's ID, to avoid the search below in a future call in case
+	 * a log tree gets created after this.
+	 */
+	if (!test_bit(BTRFS_ROOT_HAS_LOG_TREE, &inode->root->state)) {
+		inode->logged_trans = trans->transid - 1;
+		return 0;
+	}
+
+	/*
+	 * We have a log tree and the inode's logged_trans is 0. We can't tell
+	 * for sure if the inode was logged before in this transaction by looking
+	 * only at logged_trans. We could be pessimistic and assume it was, but
+	 * that can lead to unnecessarily logging an inode during rename and link
+	 * operations, and then further updating the log in followup rename and
+	 * link operations, specially if it's a directory, which adds latency
+	 * visible to applications doing a series of rename or link operations.
+	 *
+	 * A logged_trans of 0 here can mean several things:
+	 *
+	 * 1) The inode was never logged since the filesystem was mounted, and may
+	 *    or may have not been evicted and loaded again;
+	 *
+	 * 2) The inode was logged in a previous transaction, then evicted and
+	 *    then loaded again;
+	 *
+	 * 3) The inode was logged in the current transaction, then evicted and
+	 *    then loaded again.
+	 *
+	 * For cases 1) and 2) we don't want to return true, but we need to detect
+	 * case 3) and return true. So we do a search in the log root for the inode
+	 * item.
+	 */
+	key.objectid = btrfs_ino(inode);
+	key.type = BTRFS_INODE_ITEM_KEY;
+	key.offset = 0;
+
+	if (!path) {
+		path = btrfs_alloc_path();
+		if (!path)
+			return -ENOMEM;
+	}
+
+	ret = btrfs_search_slot(NULL, inode->root->log_root, &key, path, 0, 0);
+
+	if (path_in)
+		btrfs_release_path(path);
+	else
+		btrfs_free_path(path);
+
+	/*
+	 * Logging an inode always results in logging its inode item. So if we
+	 * did not find the item we know the inode was not logged for sure.
+	 */
+	if (ret < 0) {
+		return ret;
+	} else if (ret > 0) {
+		/*
+		 * Set logged_trans to a value greater than 0 and less then the
+		 * current transaction to avoid doing the search in future calls.
+		 */
+		inode->logged_trans = trans->transid - 1;
+		return 0;
+	}
+
+	/*
+	 * The inode was previously logged and then evicted, set logged_trans to
+	 * the current transacion's ID, to avoid future tree searches as long as
+	 * the inode is not evicted again.
+	 */
+	inode->logged_trans = trans->transid;
 
-	if (inode->last_trans == trans->transid &&
-	    test_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &inode->runtime_flags) &&
-	    !test_bit(BTRFS_FS_LOG_RECOVERING, &trans->fs_info->flags))
-		return true;
+	/*
+	 * If it's a directory, then we must set last_dir_index_offset to the
+	 * maximum possible value, so that the next attempt to log the inode does
+	 * not skip checking if dir index keys found in modified subvolume tree
+	 * leaves have been logged before, otherwise it would result in attempts
+	 * to insert duplicate dir index keys in the log tree. This must be done
+	 * because last_dir_index_offset is an in-memory only field, not persisted
+	 * in the inode item or any other on-disk structure, so its value is lost
+	 * once the inode is evicted.
+	 */
+	if (S_ISDIR(inode->vfs_inode.i_mode))
+		inode->last_dir_index_offset = (u64)-1;
+
+	return 1;
+}
+
+/*
+ * Delete a directory entry from the log if it exists.
+ *
+ * Returns < 0 on error
+ *           1 if the entry does not exists
+ *           0 if the entry existed and was successfully deleted
+ */
+static int del_logged_dentry(struct btrfs_trans_handle *trans,
+			     struct btrfs_root *log,
+			     struct btrfs_path *path,
+			     u64 dir_ino,
+			     const char *name, int name_len,
+			     u64 index)
+{
+	struct btrfs_dir_item *di;
+
+	/*
+	 * We only log dir index items of a directory, so we don't need to look
+	 * for dir item keys.
+	 */
+	di = btrfs_lookup_dir_index_item(trans, log, path, dir_ino,
+					 index, name, name_len, -1);
+	if (IS_ERR(di))
+		return PTR_ERR(di);
+	else if (!di)
+		return 1;
 
-	return false;
+	/*
+	 * We do not need to update the size field of the directory's
+	 * inode item because on log replay we update the field to reflect
+	 * all existing entries in the directory (see overwrite_item()).
+	 */
+	return btrfs_delete_one_dir_name(trans, log, path, di);
 }
 
 /*
@@ -3355,143 +3493,74 @@ static bool inode_logged(struct btrfs_trans_handle *trans,
  * This optimizations allows us to avoid relogging the entire inode
  * or the entire directory.
  */
-int btrfs_del_dir_entries_in_log(struct btrfs_trans_handle *trans,
-				 struct btrfs_root *root,
-				 const char *name, int name_len,
-				 struct btrfs_inode *dir, u64 index)
+void btrfs_del_dir_entries_in_log(struct btrfs_trans_handle *trans,
+				  struct btrfs_root *root,
+				  const char *name, int name_len,
+				  struct btrfs_inode *dir, u64 index)
 {
-	struct btrfs_root *log;
-	struct btrfs_dir_item *di;
 	struct btrfs_path *path;
 	int ret;
-	int err = 0;
-	int bytes_del = 0;
-	u64 dir_ino = btrfs_ino(dir);
 
-	if (!inode_logged(trans, dir))
-		return 0;
+	ret = inode_logged(trans, dir, NULL);
+	if (ret == 0)
+		return;
+	else if (ret < 0) {
+		btrfs_set_log_full_commit(trans);
+		return;
+	}
 
 	ret = join_running_log_trans(root);
 	if (ret)
-		return 0;
+		return;
 
 	mutex_lock(&dir->log_mutex);
 
-	log = root->log_root;
 	path = btrfs_alloc_path();
 	if (!path) {
-		err = -ENOMEM;
+		ret = -ENOMEM;
 		goto out_unlock;
 	}
 
-	di = btrfs_lookup_dir_item(trans, log, path, dir_ino,
-				   name, name_len, -1);
-	if (IS_ERR(di)) {
-		err = PTR_ERR(di);
-		goto fail;
-	}
-	if (di) {
-		ret = btrfs_delete_one_dir_name(trans, log, path, di);
-		bytes_del += name_len;
-		if (ret) {
-			err = ret;
-			goto fail;
-		}
-	}
-	btrfs_release_path(path);
-	di = btrfs_lookup_dir_index_item(trans, log, path, dir_ino,
-					 index, name, name_len, -1);
-	if (IS_ERR(di)) {
-		err = PTR_ERR(di);
-		goto fail;
-	}
-	if (di) {
-		ret = btrfs_delete_one_dir_name(trans, log, path, di);
-		bytes_del += name_len;
-		if (ret) {
-			err = ret;
-			goto fail;
-		}
-	}
-
-	/* update the directory size in the log to reflect the names
-	 * we have removed
-	 */
-	if (bytes_del) {
-		struct btrfs_key key;
-
-		key.objectid = dir_ino;
-		key.offset = 0;
-		key.type = BTRFS_INODE_ITEM_KEY;
-		btrfs_release_path(path);
-
-		ret = btrfs_search_slot(trans, log, &key, path, 0, 1);
-		if (ret < 0) {
-			err = ret;
-			goto fail;
-		}
-		if (ret == 0) {
-			struct btrfs_inode_item *item;
-			u64 i_size;
-
-			item = btrfs_item_ptr(path->nodes[0], path->slots[0],
-					      struct btrfs_inode_item);
-			i_size = btrfs_inode_size(path->nodes[0], item);
-			if (i_size > bytes_del)
-				i_size -= bytes_del;
-			else
-				i_size = 0;
-			btrfs_set_inode_size(path->nodes[0], item, i_size);
-			btrfs_mark_buffer_dirty(path->nodes[0]);
-		} else
-			ret = 0;
-		btrfs_release_path(path);
-	}
-fail:
+	ret = del_logged_dentry(trans, root->log_root, path, btrfs_ino(dir),
+				name, name_len, index);
 	btrfs_free_path(path);
 out_unlock:
 	mutex_unlock(&dir->log_mutex);
-	if (ret == -ENOSPC) {
+	if (ret < 0)
 		btrfs_set_log_full_commit(trans);
-		ret = 0;
-	} else if (ret < 0)
-		btrfs_abort_transaction(trans, ret);
-
 	btrfs_end_log_trans(root);
-
-	return err;
 }
 
 /* see comments for btrfs_del_dir_entries_in_log */
-int btrfs_del_inode_ref_in_log(struct btrfs_trans_handle *trans,
-			       struct btrfs_root *root,
-			       const char *name, int name_len,
-			       struct btrfs_inode *inode, u64 dirid)
+void btrfs_del_inode_ref_in_log(struct btrfs_trans_handle *trans,
+				struct btrfs_root *root,
+				const char *name, int name_len,
+				struct btrfs_inode *inode, u64 dirid)
 {
 	struct btrfs_root *log;
 	u64 index;
 	int ret;
 
-	if (!inode_logged(trans, inode))
-		return 0;
+	ret = inode_logged(trans, inode, NULL);
+	if (ret == 0)
+		return;
+	else if (ret < 0) {
+		btrfs_set_log_full_commit(trans);
+		return;
+	}
 
 	ret = join_running_log_trans(root);
 	if (ret)
-		return 0;
+		return;
 	log = root->log_root;
 	mutex_lock(&inode->log_mutex);
 
 	ret = btrfs_del_inode_ref(trans, log, name, name_len, btrfs_ino(inode),
 				  dirid, &index);
 	mutex_unlock(&inode->log_mutex);
-	if (ret == -ENOSPC) {
+	if (ret < 0 && ret != -ENOENT)
 		btrfs_set_log_full_commit(trans);
-		ret = 0;
-	} else if (ret < 0 && ret != -ENOENT)
-		btrfs_abort_transaction(trans, ret);
 	btrfs_end_log_trans(root);
-
-	return ret;
 }
 
 /*
@@ -3502,7 +3571,7 @@ int btrfs_del_inode_ref_in_log(struct btrfs_trans_handle *trans,
 static noinline int insert_dir_log_key(struct btrfs_trans_handle *trans,
 				       struct btrfs_root *log,
 				       struct btrfs_path *path,
-				       int key_type, u64 dirid,
+				       u64 dirid,
 				       u64 first_offset, u64 last_offset)
 {
 	int ret;
@@ -3511,49 +3580,236 @@ static noinline int insert_dir_log_key(struct btrfs_trans_handle *trans,
 
 	key.objectid = dirid;
 	key.offset = first_offset;
-	if (key_type == BTRFS_DIR_ITEM_KEY)
-		key.type = BTRFS_DIR_LOG_ITEM_KEY;
-	else
-		key.type = BTRFS_DIR_LOG_INDEX_KEY;
+	key.type = BTRFS_DIR_LOG_INDEX_KEY;
 	ret = btrfs_insert_empty_item(trans, log, path, &key, sizeof(*item));
-	if (ret)
+	/*
+	 * -EEXIST is fine and can happen sporadically when we are logging a
+	 * directory and have concurrent insertions in the subvolume's tree for
+	 * items from other inodes and that result in pushing off some dir items
+	 * from one leaf to another in order to accommodate for the new items.
+	 * This results in logging the same dir index range key.
+	 */
+	if (ret && ret != -EEXIST)
 		return ret;
 
 	item = btrfs_item_ptr(path->nodes[0], path->slots[0],
 			      struct btrfs_dir_log_item);
+	if (ret == -EEXIST) {
+		const u64 curr_end = btrfs_dir_log_end(path->nodes[0], item);
+
+		/*
+		 * btrfs_del_dir_entries_in_log() might have been called during
+		 * an unlink between the initial insertion of this key and the
+		 * current update, or we might be logging a single entry deletion
+		 * during a rename, so set the new last_offset to the max value.
+		 */
+		last_offset = max(last_offset, curr_end);
+	}
 	btrfs_set_dir_log_end(path->nodes[0], item, last_offset);
 	btrfs_mark_buffer_dirty(path->nodes[0]);
 	btrfs_release_path(path);
 	return 0;
 }
 
+static int flush_dir_items_batch(struct btrfs_trans_handle *trans,
+				 struct btrfs_root *log,
+				 struct extent_buffer *src,
+				 struct btrfs_path *dst_path,
+				 int start_slot,
+				 int count)
+{
+	char *ins_data = NULL;
+	struct btrfs_item_batch batch;
+	struct extent_buffer *dst;
+	unsigned long src_offset;
+	unsigned long dst_offset;
+	struct btrfs_key key;
+	u32 item_size;
+	int ret;
+	int i;
+
+	ASSERT(count > 0);
+	batch.nr = count;
+
+	if (count == 1) {
+		btrfs_item_key_to_cpu(src, &key, start_slot);
+		item_size = btrfs_item_size(src, start_slot);
+		batch.keys = &key;
+		batch.data_sizes = &item_size;
+		batch.total_data_size = item_size;
+	} else {
+		struct btrfs_key *ins_keys;
+		u32 *ins_sizes;
+
+		ins_data = kmalloc(count * sizeof(u32) +
+				   count * sizeof(struct btrfs_key), GFP_NOFS);
+		if (!ins_data)
+			return -ENOMEM;
+
+		ins_sizes = (u32 *)ins_data;
+		ins_keys = (struct btrfs_key *)(ins_data + count * sizeof(u32));
+		batch.keys = ins_keys;
+		batch.data_sizes = ins_sizes;
+		batch.total_data_size = 0;
+
+		for (i = 0; i < count; i++) {
+			const int slot = start_slot + i;
+
+			btrfs_item_key_to_cpu(src, &ins_keys[i], slot);
+			ins_sizes[i] = btrfs_item_size(src, slot);
+			batch.total_data_size += ins_sizes[i];
+		}
+	}
+
+	ret = btrfs_insert_empty_items(trans, log, dst_path, &batch);
+	if (ret)
+		goto out;
+
+	dst = dst_path->nodes[0];
+	/*
+	 * Copy all the items in bulk, in a single copy operation. Item data is
+	 * organized such that it's placed at the end of a leaf and from right
+	 * to left. For example, the data for the second item ends at an offset
+	 * that matches the offset where the data for the first item starts, the
+	 * data for the third item ends at an offset that matches the offset
+	 * where the data of the second items starts, and so on.
+	 * Therefore our source and destination start offsets for copy match the
+	 * offsets of the last items (highest slots).
+	 */
+	dst_offset = btrfs_item_ptr_offset(dst, dst_path->slots[0] + count - 1);
+	src_offset = btrfs_item_ptr_offset(src, start_slot + count - 1);
+	copy_extent_buffer(dst, src, dst_offset, src_offset, batch.total_data_size);
+	btrfs_release_path(dst_path);
+out:
+	kfree(ins_data);
+
+	return ret;
+}
+
+static int process_dir_items_leaf(struct btrfs_trans_handle *trans,
+				  struct btrfs_inode *inode,
+				  struct btrfs_path *path,
+				  struct btrfs_path *dst_path,
+				  struct btrfs_log_ctx *ctx,
+				  u64 *last_old_dentry_offset)
+{
+	struct btrfs_root *log = inode->root->log_root;
+	struct extent_buffer *src = path->nodes[0];
+	const int nritems = btrfs_header_nritems(src);
+	const u64 ino = btrfs_ino(inode);
+	bool last_found = false;
+	int batch_start = 0;
+	int batch_size = 0;
+	int i;
+
+	for (i = path->slots[0]; i < nritems; i++) {
+		struct btrfs_dir_item *di;
+		struct btrfs_key key;
+		int ret;
+
+		btrfs_item_key_to_cpu(src, &key, i);
+
+		if (key.objectid != ino || key.type != BTRFS_DIR_INDEX_KEY) {
+			last_found = true;
+			break;
+		}
+
+		di = btrfs_item_ptr(src, i, struct btrfs_dir_item);
+		ctx->last_dir_item_offset = key.offset;
+
+		/*
+		 * Skip ranges of items that consist only of dir item keys created
+		 * in past transactions. However if we find a gap, we must log a
+		 * dir index range item for that gap, so that index keys in that
+		 * gap are deleted during log replay.
+		 */
+		if (btrfs_dir_transid(src, di) < trans->transid) {
+			if (key.offset > *last_old_dentry_offset + 1) {
+				ret = insert_dir_log_key(trans, log, dst_path,
+						 ino, *last_old_dentry_offset + 1,
+						 key.offset - 1);
+				if (ret < 0)
+					return ret;
+			}
+
+			*last_old_dentry_offset = key.offset;
+			continue;
+		}
+
+		/* If we logged this dir index item before, we can skip it. */
+		if (key.offset <= inode->last_dir_index_offset)
+			continue;
+
+		/*
+		 * We must make sure that when we log a directory entry, the
+		 * corresponding inode, after log replay, has a matching link
+		 * count. For example:
+		 *
+		 * touch foo
+		 * mkdir mydir
+		 * sync
+		 * ln foo mydir/bar
+		 * xfs_io -c "fsync" mydir
+		 * <crash>
+		 * <mount fs and log replay>
+		 *
+		 * Would result in a fsync log that when replayed, our file inode
+		 * would have a link count of 1, but we get two directory entries
+		 * pointing to the same inode. After removing one of the names,
+		 * it would not be possible to remove the other name, which
+		 * resulted always in stale file handle errors, and would not be
+		 * possible to rmdir the parent directory, since its i_size could
+		 * never be decremented to the value BTRFS_EMPTY_DIR_SIZE,
+		 * resulting in -ENOTEMPTY errors.
+		 */
+		if (!ctx->log_new_dentries) {
+			struct btrfs_key di_key;
+
+			btrfs_dir_item_key_to_cpu(src, di, &di_key);
+			if (di_key.type != BTRFS_ROOT_ITEM_KEY)
+				ctx->log_new_dentries = true;
+		}
+
+		if (batch_size == 0)
+			batch_start = i;
+		batch_size++;
+	}
+
+	if (batch_size > 0) {
+		int ret;
+
+		ret = flush_dir_items_batch(trans, log, src, dst_path,
+					    batch_start, batch_size);
+		if (ret < 0)
+			return ret;
+	}
+
+	return last_found ? 1 : 0;
+}
+
 /*
  * log all the items included in the current transaction for a given
  * directory.  This also creates the range items in the log tree required
  * to replay anything deleted before the fsync
  */
 static noinline int log_dir_items(struct btrfs_trans_handle *trans,
-			  struct btrfs_root *root, struct btrfs_inode *inode,
+			  struct btrfs_inode *inode,
 			  struct btrfs_path *path,
-			  struct btrfs_path *dst_path, int key_type,
+			  struct btrfs_path *dst_path,
 			  struct btrfs_log_ctx *ctx,
 			  u64 min_offset, u64 *last_offset_ret)
 {
 	struct btrfs_key min_key;
+	struct btrfs_root *root = inode->root;
 	struct btrfs_root *log = root->log_root;
-	struct extent_buffer *src;
 	int err = 0;
 	int ret;
-	int i;
-	int nritems;
-	u64 first_offset = min_offset;
+	u64 last_old_dentry_offset = min_offset - 1;
 	u64 last_offset = (u64)-1;
 	u64 ino = btrfs_ino(inode);
 
-	log = root->log_root;
-
 	min_key.objectid = ino;
-	min_key.type = key_type;
+	min_key.type = BTRFS_DIR_INDEX_KEY;
 	min_key.offset = min_offset;
 
 	ret = btrfs_search_forward(root, &min_key, path, trans->transid);
@@ -3562,9 +3818,10 @@ static noinline int log_dir_items(struct btrfs_trans_handle *trans,
 	 * we didn't find anything from this transaction, see if there
 	 * is anything at all
 	 */
-	if (ret != 0 || min_key.objectid != ino || min_key.type != key_type) {
+	if (ret != 0 || min_key.objectid != ino ||
+	    min_key.type != BTRFS_DIR_INDEX_KEY) {
 		min_key.objectid = ino;
-		min_key.type = key_type;
+		min_key.type = BTRFS_DIR_INDEX_KEY;
 		min_key.offset = (u64)-1;
 		btrfs_release_path(path);
 		ret = btrfs_search_slot(NULL, root, &min_key, path, 0, 0);
@@ -3572,7 +3829,7 @@ static noinline int log_dir_items(struct btrfs_trans_handle *trans,
 			btrfs_release_path(path);
 			return ret;
 		}
-		ret = btrfs_previous_item(root, path, ino, key_type);
+		ret = btrfs_previous_item(root, path, ino, BTRFS_DIR_INDEX_KEY);
 
 		/* if ret == 0 there are items for this type,
 		 * create a range to tell us the last key of this type.
@@ -3581,29 +3838,31 @@ static noinline int log_dir_items(struct btrfs_trans_handle *trans,
 		 */
 		if (ret == 0) {
 			struct btrfs_key tmp;
+
 			btrfs_item_key_to_cpu(path->nodes[0], &tmp,
 					      path->slots[0]);
-			if (key_type == tmp.type)
-				first_offset = max(min_offset, tmp.offset) + 1;
+			if (tmp.type == BTRFS_DIR_INDEX_KEY)
+				last_old_dentry_offset = tmp.offset;
 		}
 		goto done;
 	}
 
 	/* go backward to find any previous key */
-	ret = btrfs_previous_item(root, path, ino, key_type);
+	ret = btrfs_previous_item(root, path, ino, BTRFS_DIR_INDEX_KEY);
 	if (ret == 0) {
 		struct btrfs_key tmp;
+
 		btrfs_item_key_to_cpu(path->nodes[0], &tmp, path->slots[0]);
-		if (key_type == tmp.type) {
-			first_offset = tmp.offset;
-			ret = overwrite_item(trans, log, dst_path,
-					     path->nodes[0], path->slots[0],
-					     &tmp);
-			if (ret) {
-				err = ret;
-				goto done;
-			}
-		}
+		/*
+		 * The dir index key before the first one we found that needs to
+		 * be logged might be in a previous leaf, and there might be a
+		 * gap between these keys, meaning that we had deletions that
+		 * happened. So the key range item we log (key type
+		 * BTRFS_DIR_LOG_INDEX_KEY) must cover a range that starts at the
+		 * previous key's offset plus 1, so that those deletes are replayed.
+		 */
+		if (tmp.type == BTRFS_DIR_INDEX_KEY)
+			last_old_dentry_offset = tmp.offset;
 	}
 	btrfs_release_path(path);
 
@@ -3615,6 +3874,7 @@ static noinline int log_dir_items(struct btrfs_trans_handle *trans,
 	 * search and this search we'll not find the key again and can just
 	 * bail.
 	 */
+search:
 	ret = btrfs_search_slot(NULL, root, &min_key, path, 0, 0);
 	if (ret != 0)
 		goto done;
@@ -3624,55 +3884,14 @@ static noinline int log_dir_items(struct btrfs_trans_handle *trans,
 	 * from our directory
 	 */
 	while (1) {
-		struct btrfs_key tmp;
-		src = path->nodes[0];
-		nritems = btrfs_header_nritems(src);
-		for (i = path->slots[0]; i < nritems; i++) {
-			struct btrfs_dir_item *di;
-
-			btrfs_item_key_to_cpu(src, &min_key, i);
-
-			if (min_key.objectid != ino || min_key.type != key_type)
-				goto done;
-			ret = overwrite_item(trans, log, dst_path, src, i,
-					     &min_key);
-			if (ret) {
+		ret = process_dir_items_leaf(trans, inode, path, dst_path, ctx,
+					     &last_old_dentry_offset);
+		if (ret != 0) {
+			if (ret < 0)
 				err = ret;
-				goto done;
-			}
-
-			/*
-			 * We must make sure that when we log a directory entry,
-			 * the corresponding inode, after log replay, has a
-			 * matching link count. For example:
-			 *
-			 * touch foo
-			 * mkdir mydir
-			 * sync
-			 * ln foo mydir/bar
-			 * xfs_io -c "fsync" mydir
-			 * <crash>
-			 * <mount fs and log replay>
-			 *
-			 * Would result in a fsync log that when replayed, our
-			 * file inode would have a link count of 1, but we get
-			 * two directory entries pointing to the same inode.
-			 * After removing one of the names, it would not be
-			 * possible to remove the other name, which resulted
-			 * always in stale file handle errors, and would not
-			 * be possible to rmdir the parent directory, since
-			 * its i_size could never decrement to the value
-			 * BTRFS_EMPTY_DIR_SIZE, resulting in -ENOTEMPTY errors.
-			 */
-			di = btrfs_item_ptr(src, i, struct btrfs_dir_item);
-			btrfs_dir_item_key_to_cpu(src, di, &tmp);
-			if (ctx &&
-			    (btrfs_dir_transid(src, di) == trans->transid ||
-			     btrfs_dir_type(src, di) == BTRFS_FT_DIR) &&
-			    tmp.type != BTRFS_ROOT_ITEM_KEY)
-				ctx->log_new_dentries = true;
+			goto done;
 		}
-		path->slots[0] = nritems;
+		path->slots[0] = btrfs_header_nritems(path->nodes[0]);
 
 		/*
 		 * look ahead to the next item and see if it is also
@@ -3686,21 +3905,29 @@ static noinline int log_dir_items(struct btrfs_trans_handle *trans,
 				err = ret;
 			goto done;
 		}
-		btrfs_item_key_to_cpu(path->nodes[0], &tmp, path->slots[0]);
-		if (tmp.objectid != ino || tmp.type != key_type) {
+		btrfs_item_key_to_cpu(path->nodes[0], &min_key, path->slots[0]);
+		if (min_key.objectid != ino || min_key.type != BTRFS_DIR_INDEX_KEY) {
 			last_offset = (u64)-1;
 			goto done;
 		}
 		if (btrfs_header_generation(path->nodes[0]) != trans->transid) {
-			ret = overwrite_item(trans, log, dst_path,
-					     path->nodes[0], path->slots[0],
-					     &tmp);
-			if (ret)
-				err = ret;
-			else
-				last_offset = tmp.offset;
+			/*
+			 * The next leaf was not changed in the current transaction
+			 * and has at least one dir index key.
+			 * We check for the next key because there might have been
+			 * one or more deletions between the last key we logged and
+			 * that next key. So the key range item we log (key type
+			 * BTRFS_DIR_LOG_INDEX_KEY) must end at the next key's
+			 * offset minus 1, so that those deletes are replayed.
+			 */
+			last_offset = min_key.offset - 1;
 			goto done;
 		}
+		if (need_resched()) {
+			btrfs_release_path(path);
+			cond_resched();
+			goto search;
+		}
 	}
 done:
 	btrfs_release_path(path);
@@ -3709,18 +3936,91 @@ done:
 	if (err == 0) {
 		*last_offset_ret = last_offset;
 		/*
-		 * insert the log range keys to indicate where the log
-		 * is valid
+		 * In case the leaf was changed in the current transaction but
+		 * all its dir items are from a past transaction, the last item
+		 * in the leaf is a dir item and there's no gap between that last
+		 * dir item and the first one on the next leaf (which did not
+		 * change in the current transaction), then we don't need to log
+		 * a range, last_old_dentry_offset is == to last_offset.
 		 */
-		ret = insert_dir_log_key(trans, log, path, key_type,
-					 ino, first_offset, last_offset);
-		if (ret)
-			err = ret;
+		ASSERT(last_old_dentry_offset <= last_offset);
+		if (last_old_dentry_offset < last_offset) {
+			ret = insert_dir_log_key(trans, log, path, ino,
+						 last_old_dentry_offset + 1,
+						 last_offset);
+			if (ret)
+				err = ret;
+		}
 	}
 	return err;
 }
 
 /*
+ * If the inode was logged before and it was evicted, then its
+ * last_dir_index_offset is (u64)-1, so we don't the value of the last index
+ * key offset. If that's the case, search for it and update the inode. This
+ * is to avoid lookups in the log tree every time we try to insert a dir index
+ * key from a leaf changed in the current transaction, and to allow us to always
+ * do batch insertions of dir index keys.
+ */
+static int update_last_dir_index_offset(struct btrfs_inode *inode,
+					struct btrfs_path *path,
+					const struct btrfs_log_ctx *ctx)
+{
+	const u64 ino = btrfs_ino(inode);
+	struct btrfs_key key;
+	int ret;
+
+	lockdep_assert_held(&inode->log_mutex);
+
+	if (inode->last_dir_index_offset != (u64)-1)
+		return 0;
+
+	if (!ctx->logged_before) {
+		inode->last_dir_index_offset = BTRFS_DIR_START_INDEX - 1;
+		return 0;
+	}
+
+	key.objectid = ino;
+	key.type = BTRFS_DIR_INDEX_KEY;
+	key.offset = (u64)-1;
+
+	ret = btrfs_search_slot(NULL, inode->root->log_root, &key, path, 0, 0);
+	/*
+	 * An error happened or we actually have an index key with an offset
+	 * value of (u64)-1. Bail out, we're done.
+	 */
+	if (ret <= 0)
+		goto out;
+
+	ret = 0;
+	inode->last_dir_index_offset = BTRFS_DIR_START_INDEX - 1;
+
+	/*
+	 * No dir index items, bail out and leave last_dir_index_offset with
+	 * the value right before the first valid index value.
+	 */
+	if (path->slots[0] == 0)
+		goto out;
+
+	/*
+	 * btrfs_search_slot() left us at one slot beyond the slot with the last
+	 * index key, or beyond the last key of the directory that is not an
+	 * index key. If we have an index key before, set last_dir_index_offset
+	 * to its offset value, otherwise leave it with a value right before the
+	 * first valid index value, as it means we have an empty directory.
+	 */
+	btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0] - 1);
+	if (key.objectid == ino && key.type == BTRFS_DIR_INDEX_KEY)
+		inode->last_dir_index_offset = key.offset;
+
+out:
+	btrfs_release_path(path);
+
+	return ret;
+}
+
+/*
  * logging directories is very similar to logging inodes, We find all the items
  * from the current transaction and write them to the log.
  *
@@ -3733,7 +4033,7 @@ done:
  * key logged by this transaction.
  */
 static noinline int log_directory_changes(struct btrfs_trans_handle *trans,
-			  struct btrfs_root *root, struct btrfs_inode *inode,
+			  struct btrfs_inode *inode,
 			  struct btrfs_path *path,
 			  struct btrfs_path *dst_path,
 			  struct btrfs_log_ctx *ctx)
@@ -3741,13 +4041,17 @@ static noinline int log_directory_changes(struct btrfs_trans_handle *trans,
 	u64 min_key;
 	u64 max_key;
 	int ret;
-	int key_type = BTRFS_DIR_ITEM_KEY;
 
-again:
-	min_key = 0;
+	ret = update_last_dir_index_offset(inode, path, ctx);
+	if (ret)
+		return ret;
+
+	min_key = BTRFS_DIR_START_INDEX;
 	max_key = 0;
+	ctx->last_dir_item_offset = inode->last_dir_index_offset;
+
 	while (1) {
-		ret = log_dir_items(trans, root, inode, path, dst_path, key_type,
+		ret = log_dir_items(trans, inode, path, dst_path,
 				ctx, min_key, &max_key);
 		if (ret)
 			return ret;
@@ -3756,10 +4060,8 @@ again:
 		min_key = max_key + 1;
 	}
 
-	if (key_type == BTRFS_DIR_ITEM_KEY) {
-		key_type = BTRFS_DIR_INDEX_KEY;
-		goto again;
-	}
+	inode->last_dir_index_offset = ctx->last_dir_item_offset;
+
 	return 0;
 }
 
@@ -3769,17 +4071,18 @@ again:
  * This cannot be run for file data extents because it does not
  * free the extents they point to.
  */
-static int drop_objectid_items(struct btrfs_trans_handle *trans,
+static int drop_inode_items(struct btrfs_trans_handle *trans,
 				  struct btrfs_root *log,
 				  struct btrfs_path *path,
-				  u64 objectid, int max_key_type)
+				  struct btrfs_inode *inode,
+				  int max_key_type)
 {
 	int ret;
 	struct btrfs_key key;
 	struct btrfs_key found_key;
 	int start_slot;
 
-	key.objectid = objectid;
+	key.objectid = btrfs_ino(inode);
 	key.type = max_key_type;
 	key.offset = (u64)-1;
 
@@ -3796,13 +4099,12 @@ static int drop_objectid_items(struct btrfs_trans_handle *trans,
 		btrfs_item_key_to_cpu(path->nodes[0], &found_key,
 				      path->slots[0]);
 
-		if (found_key.objectid != objectid)
+		if (found_key.objectid != key.objectid)
 			break;
 
 		found_key.offset = 0;
 		found_key.type = 0;
-		ret = btrfs_bin_search(path->nodes[0], &found_key, 0,
-				       &start_slot);
+		ret = btrfs_bin_search(path->nodes[0], &found_key, &start_slot);
 		if (ret < 0)
 			break;
 
@@ -3822,6 +4124,21 @@ static int drop_objectid_items(struct btrfs_trans_handle *trans,
 	return ret;
 }
 
+static int truncate_inode_items(struct btrfs_trans_handle *trans,
+				struct btrfs_root *log_root,
+				struct btrfs_inode *inode,
+				u64 new_size, u32 min_type)
+{
+	struct btrfs_truncate_control control = {
+		.new_size = new_size,
+		.ino = btrfs_ino(inode),
+		.min_type = min_type,
+		.skip_ref_updates = true,
+	};
+
+	return btrfs_truncate_inode_items(trans, log_root, &control);
+}
+
 static void fill_inode_item(struct btrfs_trans_handle *trans,
 			    struct extent_buffer *leaf,
 			    struct btrfs_inode_item *item,
@@ -3829,6 +4146,7 @@ static void fill_inode_item(struct btrfs_trans_handle *trans,
 			    u64 logged_isize)
 {
 	struct btrfs_map_token token;
+	u64 flags;
 
 	btrfs_init_map_token(&token, leaf);
 
@@ -3838,56 +4156,89 @@ static void fill_inode_item(struct btrfs_trans_handle *trans,
 		 * just to say 'this inode exists' and a logging
 		 * to say 'update this inode with these values'
 		 */
-		btrfs_set_token_inode_generation(leaf, item, 0, &token);
-		btrfs_set_token_inode_size(leaf, item, logged_isize, &token);
+		btrfs_set_token_inode_generation(&token, item, 0);
+		btrfs_set_token_inode_size(&token, item, logged_isize);
 	} else {
-		btrfs_set_token_inode_generation(leaf, item,
-						 BTRFS_I(inode)->generation,
-						 &token);
-		btrfs_set_token_inode_size(leaf, item, inode->i_size, &token);
-	}
-
-	btrfs_set_token_inode_uid(leaf, item, i_uid_read(inode), &token);
-	btrfs_set_token_inode_gid(leaf, item, i_gid_read(inode), &token);
-	btrfs_set_token_inode_mode(leaf, item, inode->i_mode, &token);
-	btrfs_set_token_inode_nlink(leaf, item, inode->i_nlink, &token);
-
-	btrfs_set_token_timespec_sec(leaf, &item->atime,
-				     inode->i_atime.tv_sec, &token);
-	btrfs_set_token_timespec_nsec(leaf, &item->atime,
-				      inode->i_atime.tv_nsec, &token);
-
-	btrfs_set_token_timespec_sec(leaf, &item->mtime,
-				     inode->i_mtime.tv_sec, &token);
-	btrfs_set_token_timespec_nsec(leaf, &item->mtime,
-				      inode->i_mtime.tv_nsec, &token);
-
-	btrfs_set_token_timespec_sec(leaf, &item->ctime,
-				     inode->i_ctime.tv_sec, &token);
-	btrfs_set_token_timespec_nsec(leaf, &item->ctime,
-				      inode->i_ctime.tv_nsec, &token);
-
-	btrfs_set_token_inode_nbytes(leaf, item, inode_get_bytes(inode),
-				     &token);
-
-	btrfs_set_token_inode_sequence(leaf, item,
-				       inode_peek_iversion(inode), &token);
-	btrfs_set_token_inode_transid(leaf, item, trans->transid, &token);
-	btrfs_set_token_inode_rdev(leaf, item, inode->i_rdev, &token);
-	btrfs_set_token_inode_flags(leaf, item, BTRFS_I(inode)->flags, &token);
-	btrfs_set_token_inode_block_group(leaf, item, 0, &token);
+		btrfs_set_token_inode_generation(&token, item,
+						 BTRFS_I(inode)->generation);
+		btrfs_set_token_inode_size(&token, item, inode->i_size);
+	}
+
+	btrfs_set_token_inode_uid(&token, item, i_uid_read(inode));
+	btrfs_set_token_inode_gid(&token, item, i_gid_read(inode));
+	btrfs_set_token_inode_mode(&token, item, inode->i_mode);
+	btrfs_set_token_inode_nlink(&token, item, inode->i_nlink);
+
+	btrfs_set_token_timespec_sec(&token, &item->atime,
+				     inode->i_atime.tv_sec);
+	btrfs_set_token_timespec_nsec(&token, &item->atime,
+				      inode->i_atime.tv_nsec);
+
+	btrfs_set_token_timespec_sec(&token, &item->mtime,
+				     inode->i_mtime.tv_sec);
+	btrfs_set_token_timespec_nsec(&token, &item->mtime,
+				      inode->i_mtime.tv_nsec);
+
+	btrfs_set_token_timespec_sec(&token, &item->ctime,
+				     inode->i_ctime.tv_sec);
+	btrfs_set_token_timespec_nsec(&token, &item->ctime,
+				      inode->i_ctime.tv_nsec);
+
+	/*
+	 * We do not need to set the nbytes field, in fact during a fast fsync
+	 * its value may not even be correct, since a fast fsync does not wait
+	 * for ordered extent completion, which is where we update nbytes, it
+	 * only waits for writeback to complete. During log replay as we find
+	 * file extent items and replay them, we adjust the nbytes field of the
+	 * inode item in subvolume tree as needed (see overwrite_item()).
+	 */
+
+	btrfs_set_token_inode_sequence(&token, item, inode_peek_iversion(inode));
+	btrfs_set_token_inode_transid(&token, item, trans->transid);
+	btrfs_set_token_inode_rdev(&token, item, inode->i_rdev);
+	flags = btrfs_inode_combine_flags(BTRFS_I(inode)->flags,
+					  BTRFS_I(inode)->ro_flags);
+	btrfs_set_token_inode_flags(&token, item, flags);
+	btrfs_set_token_inode_block_group(&token, item, 0);
 }
 
 static int log_inode_item(struct btrfs_trans_handle *trans,
 			  struct btrfs_root *log, struct btrfs_path *path,
-			  struct btrfs_inode *inode)
+			  struct btrfs_inode *inode, bool inode_item_dropped)
 {
 	struct btrfs_inode_item *inode_item;
 	int ret;
 
-	ret = btrfs_insert_empty_item(trans, log, path,
-				      &inode->location, sizeof(*inode_item));
-	if (ret && ret != -EEXIST)
+	/*
+	 * If we are doing a fast fsync and the inode was logged before in the
+	 * current transaction, then we know the inode was previously logged and
+	 * it exists in the log tree. For performance reasons, in this case use
+	 * btrfs_search_slot() directly with ins_len set to 0 so that we never
+	 * attempt a write lock on the leaf's parent, which adds unnecessary lock
+	 * contention in case there are concurrent fsyncs for other inodes of the
+	 * same subvolume. Using btrfs_insert_empty_item() when the inode item
+	 * already exists can also result in unnecessarily splitting a leaf.
+	 */
+	if (!inode_item_dropped && inode->logged_trans == trans->transid) {
+		ret = btrfs_search_slot(trans, log, &inode->location, path, 0, 1);
+		ASSERT(ret <= 0);
+		if (ret > 0)
+			ret = -ENOENT;
+	} else {
+		/*
+		 * This means it is the first fsync in the current transaction,
+		 * so the inode item is not in the log and we need to insert it.
+		 * We can never get -EEXIST because we are only called for a fast
+		 * fsync and in case an inode eviction happens after the inode was
+		 * logged before in the current transaction, when we load again
+		 * the inode, we set BTRFS_INODE_NEEDS_FULL_SYNC on its runtime
+		 * flags and set ->logged_trans to 0.
+		 */
+		ret = btrfs_insert_empty_item(trans, log, path, &inode->location,
+					      sizeof(*inode_item));
+		ASSERT(ret != -EEXIST);
+	}
+	if (ret)
 		return ret;
 	inode_item = btrfs_item_ptr(path->nodes[0], path->slots[0],
 				    struct btrfs_inode_item);
@@ -3898,12 +4249,33 @@ static int log_inode_item(struct btrfs_trans_handle *trans,
 }
 
 static int log_csums(struct btrfs_trans_handle *trans,
+		     struct btrfs_inode *inode,
 		     struct btrfs_root *log_root,
 		     struct btrfs_ordered_sum *sums)
 {
+	const u64 lock_end = sums->bytenr + sums->len - 1;
+	struct extent_state *cached_state = NULL;
 	int ret;
 
 	/*
+	 * If this inode was not used for reflink operations in the current
+	 * transaction with new extents, then do the fast path, no need to
+	 * worry about logging checksum items with overlapping ranges.
+	 */
+	if (inode->last_reflink_trans < trans->transid)
+		return btrfs_csum_file_blocks(trans, log_root, sums);
+
+	/*
+	 * Serialize logging for checksums. This is to avoid racing with the
+	 * same checksum being logged by another task that is logging another
+	 * file which happens to refer to the same extent as well. Such races
+	 * can leave checksum items in the log with overlapping ranges.
+	 */
+	ret = lock_extent(&log_root->log_csum_range, sums->bytenr, lock_end,
+			  &cached_state);
+	if (ret)
+		return ret;
+	/*
 	 * Due to extent cloning, we might have logged a csum item that covers a
 	 * subrange of a cloned extent, and later we can end up logging a csum
 	 * item for a larger subrange of the same extent or the entire range.
@@ -3913,10 +4285,13 @@ static int log_csums(struct btrfs_trans_handle *trans,
 	 * trim and adjust) any existing csum items in the log for this range.
 	 */
 	ret = btrfs_del_csums(trans, log_root, sums->bytenr, sums->len);
-	if (ret)
-		return ret;
+	if (!ret)
+		ret = btrfs_csum_file_blocks(trans, log_root, sums);
+
+	unlock_extent(&log_root->log_csum_range, sums->bytenr, lock_end,
+		      &cached_state);
 
-	return btrfs_csum_file_blocks(trans, log_root, sums);
+	return ret;
 }
 
 static noinline int copy_items(struct btrfs_trans_handle *trans,
@@ -3926,22 +4301,18 @@ static noinline int copy_items(struct btrfs_trans_handle *trans,
 			       int start_slot, int nr, int inode_only,
 			       u64 logged_isize)
 {
-	struct btrfs_fs_info *fs_info = trans->fs_info;
-	unsigned long src_offset;
-	unsigned long dst_offset;
 	struct btrfs_root *log = inode->root->log_root;
 	struct btrfs_file_extent_item *extent;
-	struct btrfs_inode_item *inode_item;
 	struct extent_buffer *src = src_path->nodes[0];
-	int ret;
+	int ret = 0;
 	struct btrfs_key *ins_keys;
 	u32 *ins_sizes;
+	struct btrfs_item_batch batch;
 	char *ins_data;
 	int i;
-	struct list_head ordered_sums;
-	int skip_csum = inode->flags & BTRFS_INODE_NODATASUM;
-
-	INIT_LIST_HEAD(&ordered_sums);
+	int dst_index;
+	const bool skip_csum = (inode->flags & BTRFS_INODE_NODATASUM);
+	const u64 i_size = i_size_read(&inode->vfs_inode);
 
 	ins_data = kmalloc(nr * sizeof(struct btrfs_key) +
 			   nr * sizeof(u32), GFP_NOFS);
@@ -3950,27 +4321,155 @@ static noinline int copy_items(struct btrfs_trans_handle *trans,
 
 	ins_sizes = (u32 *)ins_data;
 	ins_keys = (struct btrfs_key *)(ins_data + nr * sizeof(u32));
+	batch.keys = ins_keys;
+	batch.data_sizes = ins_sizes;
+	batch.total_data_size = 0;
+	batch.nr = 0;
 
+	dst_index = 0;
 	for (i = 0; i < nr; i++) {
-		ins_sizes[i] = btrfs_item_size_nr(src, i + start_slot);
-		btrfs_item_key_to_cpu(src, ins_keys + i, i + start_slot);
-	}
-	ret = btrfs_insert_empty_items(trans, log, dst_path,
-				       ins_keys, ins_sizes, nr);
-	if (ret) {
-		kfree(ins_data);
-		return ret;
+		const int src_slot = start_slot + i;
+		struct btrfs_root *csum_root;
+		struct btrfs_ordered_sum *sums;
+		struct btrfs_ordered_sum *sums_next;
+		LIST_HEAD(ordered_sums);
+		u64 disk_bytenr;
+		u64 disk_num_bytes;
+		u64 extent_offset;
+		u64 extent_num_bytes;
+		bool is_old_extent;
+
+		btrfs_item_key_to_cpu(src, &ins_keys[dst_index], src_slot);
+
+		if (ins_keys[dst_index].type != BTRFS_EXTENT_DATA_KEY)
+			goto add_to_batch;
+
+		extent = btrfs_item_ptr(src, src_slot,
+					struct btrfs_file_extent_item);
+
+		is_old_extent = (btrfs_file_extent_generation(src, extent) <
+				 trans->transid);
+
+		/*
+		 * Don't copy extents from past generations. That would make us
+		 * log a lot more metadata for common cases like doing only a
+		 * few random writes into a file and then fsync it for the first
+		 * time or after the full sync flag is set on the inode. We can
+		 * get leaves full of extent items, most of which are from past
+		 * generations, so we can skip them - as long as the inode has
+		 * not been the target of a reflink operation in this transaction,
+		 * as in that case it might have had file extent items with old
+		 * generations copied into it. We also must always log prealloc
+		 * extents that start at or beyond eof, otherwise we would lose
+		 * them on log replay.
+		 */
+		if (is_old_extent &&
+		    ins_keys[dst_index].offset < i_size &&
+		    inode->last_reflink_trans < trans->transid)
+			continue;
+
+		if (skip_csum)
+			goto add_to_batch;
+
+		/* Only regular extents have checksums. */
+		if (btrfs_file_extent_type(src, extent) != BTRFS_FILE_EXTENT_REG)
+			goto add_to_batch;
+
+		/*
+		 * If it's an extent created in a past transaction, then its
+		 * checksums are already accessible from the committed csum tree,
+		 * no need to log them.
+		 */
+		if (is_old_extent)
+			goto add_to_batch;
+
+		disk_bytenr = btrfs_file_extent_disk_bytenr(src, extent);
+		/* If it's an explicit hole, there are no checksums. */
+		if (disk_bytenr == 0)
+			goto add_to_batch;
+
+		disk_num_bytes = btrfs_file_extent_disk_num_bytes(src, extent);
+
+		if (btrfs_file_extent_compression(src, extent)) {
+			extent_offset = 0;
+			extent_num_bytes = disk_num_bytes;
+		} else {
+			extent_offset = btrfs_file_extent_offset(src, extent);
+			extent_num_bytes = btrfs_file_extent_num_bytes(src, extent);
+		}
+
+		csum_root = btrfs_csum_root(trans->fs_info, disk_bytenr);
+		disk_bytenr += extent_offset;
+		ret = btrfs_lookup_csums_range(csum_root, disk_bytenr,
+					       disk_bytenr + extent_num_bytes - 1,
+					       &ordered_sums, 0, false);
+		if (ret)
+			goto out;
+
+		list_for_each_entry_safe(sums, sums_next, &ordered_sums, list) {
+			if (!ret)
+				ret = log_csums(trans, inode, log, sums);
+			list_del(&sums->list);
+			kfree(sums);
+		}
+		if (ret)
+			goto out;
+
+add_to_batch:
+		ins_sizes[dst_index] = btrfs_item_size(src, src_slot);
+		batch.total_data_size += ins_sizes[dst_index];
+		batch.nr++;
+		dst_index++;
 	}
 
-	for (i = 0; i < nr; i++, dst_path->slots[0]++) {
-		dst_offset = btrfs_item_ptr_offset(dst_path->nodes[0],
-						   dst_path->slots[0]);
+	/*
+	 * We have a leaf full of old extent items that don't need to be logged,
+	 * so we don't need to do anything.
+	 */
+	if (batch.nr == 0)
+		goto out;
+
+	ret = btrfs_insert_empty_items(trans, log, dst_path, &batch);
+	if (ret)
+		goto out;
+
+	dst_index = 0;
+	for (i = 0; i < nr; i++) {
+		const int src_slot = start_slot + i;
+		const int dst_slot = dst_path->slots[0] + dst_index;
+		struct btrfs_key key;
+		unsigned long src_offset;
+		unsigned long dst_offset;
+
+		/*
+		 * We're done, all the remaining items in the source leaf
+		 * correspond to old file extent items.
+		 */
+		if (dst_index >= batch.nr)
+			break;
+
+		btrfs_item_key_to_cpu(src, &key, src_slot);
 
-		src_offset = btrfs_item_ptr_offset(src, start_slot + i);
+		if (key.type != BTRFS_EXTENT_DATA_KEY)
+			goto copy_item;
+
+		extent = btrfs_item_ptr(src, src_slot,
+					struct btrfs_file_extent_item);
 
-		if (ins_keys[i].type == BTRFS_INODE_ITEM_KEY) {
-			inode_item = btrfs_item_ptr(dst_path->nodes[0],
-						    dst_path->slots[0],
+		/* See the comment in the previous loop, same logic. */
+		if (btrfs_file_extent_generation(src, extent) < trans->transid &&
+		    key.offset < i_size &&
+		    inode->last_reflink_trans < trans->transid)
+			continue;
+
+copy_item:
+		dst_offset = btrfs_item_ptr_offset(dst_path->nodes[0], dst_slot);
+		src_offset = btrfs_item_ptr_offset(src, src_slot);
+
+		if (key.type == BTRFS_INODE_ITEM_KEY) {
+			struct btrfs_inode_item *inode_item;
+
+			inode_item = btrfs_item_ptr(dst_path->nodes[0], dst_slot,
 						    struct btrfs_inode_item);
 			fill_inode_item(trans, dst_path->nodes[0], inode_item,
 					&inode->vfs_inode,
@@ -3978,80 +4477,24 @@ static noinline int copy_items(struct btrfs_trans_handle *trans,
 					logged_isize);
 		} else {
 			copy_extent_buffer(dst_path->nodes[0], src, dst_offset,
-					   src_offset, ins_sizes[i]);
+					   src_offset, ins_sizes[dst_index]);
 		}
 
-		/* take a reference on file data extents so that truncates
-		 * or deletes of this inode don't have to relog the inode
-		 * again
-		 */
-		if (ins_keys[i].type == BTRFS_EXTENT_DATA_KEY &&
-		    !skip_csum) {
-			int found_type;
-			extent = btrfs_item_ptr(src, start_slot + i,
-						struct btrfs_file_extent_item);
-
-			if (btrfs_file_extent_generation(src, extent) < trans->transid)
-				continue;
-
-			found_type = btrfs_file_extent_type(src, extent);
-			if (found_type == BTRFS_FILE_EXTENT_REG) {
-				u64 ds, dl, cs, cl;
-				ds = btrfs_file_extent_disk_bytenr(src,
-								extent);
-				/* ds == 0 is a hole */
-				if (ds == 0)
-					continue;
-
-				dl = btrfs_file_extent_disk_num_bytes(src,
-								extent);
-				cs = btrfs_file_extent_offset(src, extent);
-				cl = btrfs_file_extent_num_bytes(src,
-								extent);
-				if (btrfs_file_extent_compression(src,
-								  extent)) {
-					cs = 0;
-					cl = dl;
-				}
-
-				ret = btrfs_lookup_csums_range(
-						fs_info->csum_root,
-						ds + cs, ds + cs + cl - 1,
-						&ordered_sums, 0);
-				if (ret) {
-					btrfs_release_path(dst_path);
-					kfree(ins_data);
-					return ret;
-				}
-			}
-		}
+		dst_index++;
 	}
 
 	btrfs_mark_buffer_dirty(dst_path->nodes[0]);
 	btrfs_release_path(dst_path);
+out:
 	kfree(ins_data);
 
-	/*
-	 * we have to do this after the loop above to avoid changing the
-	 * log tree while trying to change the log tree.
-	 */
-	ret = 0;
-	while (!list_empty(&ordered_sums)) {
-		struct btrfs_ordered_sum *sums = list_entry(ordered_sums.next,
-						   struct btrfs_ordered_sum,
-						   list);
-		if (!ret)
-			ret = log_csums(trans, log, sums);
-		list_del(&sums->list);
-		kfree(sums);
-	}
-
 	return ret;
 }
 
-static int extent_cmp(void *priv, struct list_head *a, struct list_head *b)
+static int extent_cmp(void *priv, const struct list_head *a,
+		      const struct list_head *b)
 {
-	struct extent_map *em1, *em2;
+	const struct extent_map *em1, *em2;
 
 	em1 = list_entry(a, struct extent_map, list);
 	em2 = list_entry(b, struct extent_map, list);
@@ -4066,10 +4509,15 @@ static int extent_cmp(void *priv, struct list_head *a, struct list_head *b)
 static int log_extent_csums(struct btrfs_trans_handle *trans,
 			    struct btrfs_inode *inode,
 			    struct btrfs_root *log_root,
-			    const struct extent_map *em)
+			    const struct extent_map *em,
+			    struct btrfs_log_ctx *ctx)
 {
+	struct btrfs_ordered_extent *ordered;
+	struct btrfs_root *csum_root;
 	u64 csum_offset;
 	u64 csum_len;
+	u64 mod_start = em->mod_start;
+	u64 mod_len = em->mod_len;
 	LIST_HEAD(ordered_sums);
 	int ret = 0;
 
@@ -4078,20 +4526,79 @@ static int log_extent_csums(struct btrfs_trans_handle *trans,
 	    em->block_start == EXTENT_MAP_HOLE)
 		return 0;
 
+	list_for_each_entry(ordered, &ctx->ordered_extents, log_list) {
+		const u64 ordered_end = ordered->file_offset + ordered->num_bytes;
+		const u64 mod_end = mod_start + mod_len;
+		struct btrfs_ordered_sum *sums;
+
+		if (mod_len == 0)
+			break;
+
+		if (ordered_end <= mod_start)
+			continue;
+		if (mod_end <= ordered->file_offset)
+			break;
+
+		/*
+		 * We are going to copy all the csums on this ordered extent, so
+		 * go ahead and adjust mod_start and mod_len in case this ordered
+		 * extent has already been logged.
+		 */
+		if (ordered->file_offset > mod_start) {
+			if (ordered_end >= mod_end)
+				mod_len = ordered->file_offset - mod_start;
+			/*
+			 * If we have this case
+			 *
+			 * |--------- logged extent ---------|
+			 *       |----- ordered extent ----|
+			 *
+			 * Just don't mess with mod_start and mod_len, we'll
+			 * just end up logging more csums than we need and it
+			 * will be ok.
+			 */
+		} else {
+			if (ordered_end < mod_end) {
+				mod_len = mod_end - ordered_end;
+				mod_start = ordered_end;
+			} else {
+				mod_len = 0;
+			}
+		}
+
+		/*
+		 * To keep us from looping for the above case of an ordered
+		 * extent that falls inside of the logged extent.
+		 */
+		if (test_and_set_bit(BTRFS_ORDERED_LOGGED_CSUM, &ordered->flags))
+			continue;
+
+		list_for_each_entry(sums, &ordered->list, list) {
+			ret = log_csums(trans, inode, log_root, sums);
+			if (ret)
+				return ret;
+		}
+	}
+
+	/* We're done, found all csums in the ordered extents. */
+	if (mod_len == 0)
+		return 0;
+
 	/* If we're compressed we have to save the entire range of csums. */
 	if (em->compress_type) {
 		csum_offset = 0;
 		csum_len = max(em->block_len, em->orig_block_len);
 	} else {
-		csum_offset = em->mod_start - em->start;
-		csum_len = em->mod_len;
+		csum_offset = mod_start - em->start;
+		csum_len = mod_len;
 	}
 
 	/* block start is already adjusted for the file extent offset. */
-	ret = btrfs_lookup_csums_range(trans->fs_info->csum_root,
+	csum_root = btrfs_csum_root(trans->fs_info, em->block_start);
+	ret = btrfs_lookup_csums_range(csum_root,
 				       em->block_start + csum_offset,
 				       em->block_start + csum_offset +
-				       csum_len - 1, &ordered_sums, 0);
+				       csum_len - 1, &ordered_sums, 0, false);
 	if (ret)
 		return ret;
 
@@ -4100,7 +4607,7 @@ static int log_extent_csums(struct btrfs_trans_handle *trans,
 						   struct btrfs_ordered_sum,
 						   list);
 		if (!ret)
-			ret = log_csums(trans, log_root, sums);
+			ret = log_csums(trans, inode, log_root, sums);
 		list_del(&sums->list);
 		kfree(sums);
 	}
@@ -4109,83 +4616,79 @@ static int log_extent_csums(struct btrfs_trans_handle *trans,
 }
 
 static int log_one_extent(struct btrfs_trans_handle *trans,
-			  struct btrfs_inode *inode, struct btrfs_root *root,
+			  struct btrfs_inode *inode,
 			  const struct extent_map *em,
 			  struct btrfs_path *path,
 			  struct btrfs_log_ctx *ctx)
 {
-	struct btrfs_root *log = root->log_root;
-	struct btrfs_file_extent_item *fi;
+	struct btrfs_drop_extents_args drop_args = { 0 };
+	struct btrfs_root *log = inode->root->log_root;
+	struct btrfs_file_extent_item fi = { 0 };
 	struct extent_buffer *leaf;
-	struct btrfs_map_token token;
 	struct btrfs_key key;
 	u64 extent_offset = em->start - em->orig_start;
 	u64 block_len;
 	int ret;
-	int extent_inserted = 0;
 
-	ret = log_extent_csums(trans, inode, log, em);
-	if (ret)
-		return ret;
+	btrfs_set_stack_file_extent_generation(&fi, trans->transid);
+	if (test_bit(EXTENT_FLAG_PREALLOC, &em->flags))
+		btrfs_set_stack_file_extent_type(&fi, BTRFS_FILE_EXTENT_PREALLOC);
+	else
+		btrfs_set_stack_file_extent_type(&fi, BTRFS_FILE_EXTENT_REG);
+
+	block_len = max(em->block_len, em->orig_block_len);
+	if (em->compress_type != BTRFS_COMPRESS_NONE) {
+		btrfs_set_stack_file_extent_disk_bytenr(&fi, em->block_start);
+		btrfs_set_stack_file_extent_disk_num_bytes(&fi, block_len);
+	} else if (em->block_start < EXTENT_MAP_LAST_BYTE) {
+		btrfs_set_stack_file_extent_disk_bytenr(&fi, em->block_start -
+							extent_offset);
+		btrfs_set_stack_file_extent_disk_num_bytes(&fi, block_len);
+	}
+
+	btrfs_set_stack_file_extent_offset(&fi, extent_offset);
+	btrfs_set_stack_file_extent_num_bytes(&fi, em->len);
+	btrfs_set_stack_file_extent_ram_bytes(&fi, em->ram_bytes);
+	btrfs_set_stack_file_extent_compression(&fi, em->compress_type);
 
-	ret = __btrfs_drop_extents(trans, log, &inode->vfs_inode, path, em->start,
-				   em->start + em->len, NULL, 0, 1,
-				   sizeof(*fi), &extent_inserted);
+	ret = log_extent_csums(trans, inode, log, em, ctx);
 	if (ret)
 		return ret;
 
-	if (!extent_inserted) {
+	/*
+	 * If this is the first time we are logging the inode in the current
+	 * transaction, we can avoid btrfs_drop_extents(), which is expensive
+	 * because it does a deletion search, which always acquires write locks
+	 * for extent buffers at levels 2, 1 and 0. This not only wastes time
+	 * but also adds significant contention in a log tree, since log trees
+	 * are small, with a root at level 2 or 3 at most, due to their short
+	 * life span.
+	 */
+	if (ctx->logged_before) {
+		drop_args.path = path;
+		drop_args.start = em->start;
+		drop_args.end = em->start + em->len;
+		drop_args.replace_extent = true;
+		drop_args.extent_item_size = sizeof(fi);
+		ret = btrfs_drop_extents(trans, log, inode, &drop_args);
+		if (ret)
+			return ret;
+	}
+
+	if (!drop_args.extent_inserted) {
 		key.objectid = btrfs_ino(inode);
 		key.type = BTRFS_EXTENT_DATA_KEY;
 		key.offset = em->start;
 
 		ret = btrfs_insert_empty_item(trans, log, path, &key,
-					      sizeof(*fi));
+					      sizeof(fi));
 		if (ret)
 			return ret;
 	}
 	leaf = path->nodes[0];
-	btrfs_init_map_token(&token, leaf);
-	fi = btrfs_item_ptr(leaf, path->slots[0],
-			    struct btrfs_file_extent_item);
-
-	btrfs_set_token_file_extent_generation(leaf, fi, trans->transid,
-					       &token);
-	if (test_bit(EXTENT_FLAG_PREALLOC, &em->flags))
-		btrfs_set_token_file_extent_type(leaf, fi,
-						 BTRFS_FILE_EXTENT_PREALLOC,
-						 &token);
-	else
-		btrfs_set_token_file_extent_type(leaf, fi,
-						 BTRFS_FILE_EXTENT_REG,
-						 &token);
-
-	block_len = max(em->block_len, em->orig_block_len);
-	if (em->compress_type != BTRFS_COMPRESS_NONE) {
-		btrfs_set_token_file_extent_disk_bytenr(leaf, fi,
-							em->block_start,
-							&token);
-		btrfs_set_token_file_extent_disk_num_bytes(leaf, fi, block_len,
-							   &token);
-	} else if (em->block_start < EXTENT_MAP_LAST_BYTE) {
-		btrfs_set_token_file_extent_disk_bytenr(leaf, fi,
-							em->block_start -
-							extent_offset, &token);
-		btrfs_set_token_file_extent_disk_num_bytes(leaf, fi, block_len,
-							   &token);
-	} else {
-		btrfs_set_token_file_extent_disk_bytenr(leaf, fi, 0, &token);
-		btrfs_set_token_file_extent_disk_num_bytes(leaf, fi, 0,
-							   &token);
-	}
-
-	btrfs_set_token_file_extent_offset(leaf, fi, extent_offset, &token);
-	btrfs_set_token_file_extent_num_bytes(leaf, fi, em->len, &token);
-	btrfs_set_token_file_extent_ram_bytes(leaf, fi, em->ram_bytes, &token);
-	btrfs_set_token_file_extent_compression(leaf, fi, em->compress_type,
-						&token);
-	btrfs_set_token_file_extent_encryption(leaf, fi, 0, &token);
-	btrfs_set_token_file_extent_other_encoding(leaf, fi, 0, &token);
+	write_extent_buffer(leaf, &fi,
+			    btrfs_item_ptr_offset(leaf, path->slots[0]),
+			    sizeof(fi));
 	btrfs_mark_buffer_dirty(leaf);
 
 	btrfs_release_path(path);
@@ -4195,7 +4698,7 @@ static int log_one_extent(struct btrfs_trans_handle *trans,
 
 /*
  * Log all prealloc extents beyond the inode's i_size to make sure we do not
- * lose them after doing a fast fsync and replaying the log. We scan the
+ * lose them after doing a full/fast fsync and replaying the log. We scan the
  * subvolume's root instead of iterating the inode's extent map tree because
  * otherwise we can log incorrect extent items based on extent map conversion.
  * That can happen due to the fact that extent maps are merged when they
@@ -4211,6 +4714,9 @@ static int btrfs_log_prealloc_extents(struct btrfs_trans_handle *trans,
 	const u64 ino = btrfs_ino(inode);
 	struct btrfs_path *dst_path = NULL;
 	bool dropped_extents = false;
+	u64 truncate_offset = i_size;
+	struct extent_buffer *leaf;
+	int slot;
 	int ins_nr = 0;
 	int start_slot;
 	int ret;
@@ -4225,9 +4731,43 @@ static int btrfs_log_prealloc_extents(struct btrfs_trans_handle *trans,
 	if (ret < 0)
 		goto out;
 
+	/*
+	 * We must check if there is a prealloc extent that starts before the
+	 * i_size and crosses the i_size boundary. This is to ensure later we
+	 * truncate down to the end of that extent and not to the i_size, as
+	 * otherwise we end up losing part of the prealloc extent after a log
+	 * replay and with an implicit hole if there is another prealloc extent
+	 * that starts at an offset beyond i_size.
+	 */
+	ret = btrfs_previous_item(root, path, ino, BTRFS_EXTENT_DATA_KEY);
+	if (ret < 0)
+		goto out;
+
+	if (ret == 0) {
+		struct btrfs_file_extent_item *ei;
+
+		leaf = path->nodes[0];
+		slot = path->slots[0];
+		ei = btrfs_item_ptr(leaf, slot, struct btrfs_file_extent_item);
+
+		if (btrfs_file_extent_type(leaf, ei) ==
+		    BTRFS_FILE_EXTENT_PREALLOC) {
+			u64 extent_end;
+
+			btrfs_item_key_to_cpu(leaf, &key, slot);
+			extent_end = key.offset +
+				btrfs_file_extent_num_bytes(leaf, ei);
+
+			if (extent_end > i_size)
+				truncate_offset = extent_end;
+		}
+	} else {
+		ret = 0;
+	}
+
 	while (true) {
-		struct extent_buffer *leaf = path->nodes[0];
-		int slot = path->slots[0];
+		leaf = path->nodes[0];
+		slot = path->slots[0];
 
 		if (slot >= btrfs_header_nritems(leaf)) {
 			if (ins_nr > 0) {
@@ -4261,13 +4801,9 @@ static int btrfs_log_prealloc_extents(struct btrfs_trans_handle *trans,
 			 * Avoid logging extent items logged in past fsync calls
 			 * and leading to duplicate keys in the log tree.
 			 */
-			do {
-				ret = btrfs_truncate_inode_items(trans,
-							 root->log_root,
-							 &inode->vfs_inode,
-							 i_size,
-							 BTRFS_EXTENT_DATA_KEY);
-			} while (ret == -EAGAIN);
+			ret = truncate_inode_items(trans, root->log_root, inode,
+						   truncate_offset,
+						   BTRFS_EXTENT_DATA_KEY);
 			if (ret)
 				goto out;
 			dropped_extents = true;
@@ -4284,12 +4820,9 @@ static int btrfs_log_prealloc_extents(struct btrfs_trans_handle *trans,
 			}
 		}
 	}
-	if (ins_nr > 0) {
+	if (ins_nr > 0)
 		ret = copy_items(trans, inode, dst_path, path,
 				 start_slot, ins_nr, 1, 0);
-		if (ret > 0)
-			ret = 0;
-	}
 out:
 	btrfs_release_path(path);
 	btrfs_free_path(dst_path);
@@ -4297,43 +4830,23 @@ out:
 }
 
 static int btrfs_log_changed_extents(struct btrfs_trans_handle *trans,
-				     struct btrfs_root *root,
 				     struct btrfs_inode *inode,
 				     struct btrfs_path *path,
-				     struct btrfs_log_ctx *ctx,
-				     const u64 start,
-				     const u64 end)
+				     struct btrfs_log_ctx *ctx)
 {
+	struct btrfs_ordered_extent *ordered;
+	struct btrfs_ordered_extent *tmp;
 	struct extent_map *em, *n;
 	struct list_head extents;
 	struct extent_map_tree *tree = &inode->extent_tree;
-	u64 test_gen;
 	int ret = 0;
 	int num = 0;
 
 	INIT_LIST_HEAD(&extents);
 
 	write_lock(&tree->lock);
-	test_gen = root->fs_info->last_trans_committed;
 
 	list_for_each_entry_safe(em, n, &tree->modified_extents, list) {
-		/*
-		 * Skip extents outside our logging range. It's important to do
-		 * it for correctness because if we don't ignore them, we may
-		 * log them before their ordered extent completes, and therefore
-		 * we could log them without logging their respective checksums
-		 * (the checksum items are added to the csum tree at the very
-		 * end of btrfs_finish_ordered_io()). Also leave such extents
-		 * outside of our range in the list, since we may have another
-		 * ranged fsync in the near future that needs them. If an extent
-		 * outside our range corresponds to a hole, log it to avoid
-		 * leaving gaps between extents (fsck will complain when we are
-		 * not using the NO_HOLES feature).
-		 */
-		if ((em->start > end || em->start + em->len <= start) &&
-		    em->block_start != EXTENT_MAP_HOLE)
-			continue;
-
 		list_del_init(&em->list);
 		/*
 		 * Just an arbitrary number, this can be really CPU intensive
@@ -4347,7 +4860,7 @@ static int btrfs_log_changed_extents(struct btrfs_trans_handle *trans,
 			goto process;
 		}
 
-		if (em->generation <= test_gen)
+		if (em->generation < trans->transid)
 			continue;
 
 		/* We log prealloc extents beyond eof later. */
@@ -4381,7 +4894,7 @@ process:
 
 		write_unlock(&tree->lock);
 
-		ret = log_one_extent(trans, inode, root, em, path, ctx);
+		ret = log_one_extent(trans, inode, em, path, ctx);
 		write_lock(&tree->lock);
 		clear_em_logging(tree, em);
 		free_extent_map(em);
@@ -4389,11 +4902,34 @@ process:
 	WARN_ON(!list_empty(&extents));
 	write_unlock(&tree->lock);
 
-	btrfs_release_path(path);
 	if (!ret)
 		ret = btrfs_log_prealloc_extents(trans, inode, path);
+	if (ret)
+		return ret;
 
-	return ret;
+	/*
+	 * We have logged all extents successfully, now make sure the commit of
+	 * the current transaction waits for the ordered extents to complete
+	 * before it commits and wipes out the log trees, otherwise we would
+	 * lose data if an ordered extents completes after the transaction
+	 * commits and a power failure happens after the transaction commit.
+	 */
+	list_for_each_entry_safe(ordered, tmp, &ctx->ordered_extents, log_list) {
+		list_del_init(&ordered->log_list);
+		set_bit(BTRFS_ORDERED_LOGGED, &ordered->flags);
+
+		if (!test_bit(BTRFS_ORDERED_COMPLETE, &ordered->flags)) {
+			spin_lock_irq(&inode->ordered_tree.lock);
+			if (!test_bit(BTRFS_ORDERED_COMPLETE, &ordered->flags)) {
+				set_bit(BTRFS_ORDERED_PENDING, &ordered->flags);
+				atomic_inc(&trans->transaction->pending_ordered);
+			}
+			spin_unlock_irq(&inode->ordered_tree.lock);
+		}
+		btrfs_put_ordered_extent(ordered);
+	}
+
+	return 0;
 }
 
 static int logged_inode_size(struct btrfs_root *log, struct btrfs_inode *inode,
@@ -4446,16 +4982,20 @@ static int logged_inode_size(struct btrfs_root *log, struct btrfs_inode *inode,
  * with a journal, ext3/4, xfs, f2fs, etc).
  */
 static int btrfs_log_all_xattrs(struct btrfs_trans_handle *trans,
-				struct btrfs_root *root,
 				struct btrfs_inode *inode,
 				struct btrfs_path *path,
 				struct btrfs_path *dst_path)
 {
+	struct btrfs_root *root = inode->root;
 	int ret;
 	struct btrfs_key key;
 	const u64 ino = btrfs_ino(inode);
 	int ins_nr = 0;
 	int start_slot = 0;
+	bool found_xattrs = false;
+
+	if (test_bit(BTRFS_INODE_NO_XATTRS, &inode->runtime_flags))
+		return 0;
 
 	key.objectid = ino;
 	key.type = BTRFS_XATTR_ITEM_KEY;
@@ -4494,6 +5034,7 @@ static int btrfs_log_all_xattrs(struct btrfs_trans_handle *trans,
 			start_slot = slot;
 		ins_nr++;
 		path->slots[0]++;
+		found_xattrs = true;
 		cond_resched();
 	}
 	if (ins_nr > 0) {
@@ -4503,6 +5044,9 @@ static int btrfs_log_all_xattrs(struct btrfs_trans_handle *trans,
 			return ret;
 	}
 
+	if (!found_xattrs)
+		set_bit(BTRFS_INODE_NO_XATTRS, &inode->runtime_flags);
+
 	return 0;
 }
 
@@ -4516,10 +5060,10 @@ static int btrfs_log_all_xattrs(struct btrfs_trans_handle *trans,
  * truncate operation that changes the inode's size.
  */
 static int btrfs_log_holes(struct btrfs_trans_handle *trans,
-			   struct btrfs_root *root,
 			   struct btrfs_inode *inode,
 			   struct btrfs_path *path)
 {
+	struct btrfs_root *root = inode->root;
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct btrfs_key key;
 	const u64 ino = btrfs_ino(inode);
@@ -4539,9 +5083,7 @@ static int btrfs_log_holes(struct btrfs_trans_handle *trans,
 		return ret;
 
 	while (true) {
-		struct btrfs_file_extent_item *extent;
 		struct extent_buffer *leaf = path->nodes[0];
-		u64 len;
 
 		if (path->slots[0] >= btrfs_header_nritems(path->nodes[0])) {
 			ret = btrfs_next_leaf(root, path);
@@ -4568,10 +5110,9 @@ static int btrfs_log_holes(struct btrfs_trans_handle *trans,
 			 * leafs from the log root.
 			 */
 			btrfs_release_path(path);
-			ret = btrfs_insert_file_extent(trans, root->log_root,
-						       ino, prev_extent_end, 0,
-						       0, hole_len, 0, hole_len,
-						       0, 0, 0);
+			ret = btrfs_insert_hole_extent(trans, root->log_root,
+						       ino, prev_extent_end,
+						       hole_len);
 			if (ret < 0)
 				return ret;
 
@@ -4590,18 +5131,7 @@ static int btrfs_log_holes(struct btrfs_trans_handle *trans,
 			leaf = path->nodes[0];
 		}
 
-		extent = btrfs_item_ptr(leaf, path->slots[0],
-					struct btrfs_file_extent_item);
-		if (btrfs_file_extent_type(leaf, extent) ==
-		    BTRFS_FILE_EXTENT_INLINE) {
-			len = btrfs_file_extent_ram_bytes(leaf, extent);
-			prev_extent_end = ALIGN(key.offset + len,
-						fs_info->sectorsize);
-		} else {
-			len = btrfs_file_extent_num_bytes(leaf, extent);
-			prev_extent_end = key.offset + len;
-		}
-
+		prev_extent_end = btrfs_file_extent_end(path);
 		path->slots[0]++;
 		cond_resched();
 	}
@@ -4611,10 +5141,8 @@ static int btrfs_log_holes(struct btrfs_trans_handle *trans,
 
 		btrfs_release_path(path);
 		hole_len = ALIGN(i_size - prev_extent_end, fs_info->sectorsize);
-		ret = btrfs_insert_file_extent(trans, root->log_root,
-					       ino, prev_extent_end, 0, 0,
-					       hole_len, 0, hole_len,
-					       0, 0, 0);
+		ret = btrfs_insert_hole_extent(trans, root->log_root, ino,
+					       prev_extent_end, hole_len);
 		if (ret < 0)
 			return ret;
 	}
@@ -4674,7 +5202,7 @@ static int btrfs_check_ref_name_override(struct extent_buffer *eb,
 	struct btrfs_path *search_path;
 	char *name = NULL;
 	u32 name_len = 0;
-	u32 item_size = btrfs_item_size_nr(eb, slot);
+	u32 item_size = btrfs_item_size(eb, slot);
 	u32 cur_offset = 0;
 	unsigned long ptr = btrfs_item_ptr_offset(eb, slot);
 
@@ -4757,115 +5285,461 @@ out:
 	return ret;
 }
 
+/*
+ * Check if we need to log an inode. This is used in contexts where while
+ * logging an inode we need to log another inode (either that it exists or in
+ * full mode). This is used instead of btrfs_inode_in_log() because the later
+ * requires the inode to be in the log and have the log transaction committed,
+ * while here we do not care if the log transaction was already committed - our
+ * caller will commit the log later - and we want to avoid logging an inode
+ * multiple times when multiple tasks have joined the same log transaction.
+ */
+static bool need_log_inode(const struct btrfs_trans_handle *trans,
+			   const struct btrfs_inode *inode)
+{
+	/*
+	 * If a directory was not modified, no dentries added or removed, we can
+	 * and should avoid logging it.
+	 */
+	if (S_ISDIR(inode->vfs_inode.i_mode) && inode->last_trans < trans->transid)
+		return false;
+
+	/*
+	 * If this inode does not have new/updated/deleted xattrs since the last
+	 * time it was logged and is flagged as logged in the current transaction,
+	 * we can skip logging it. As for new/deleted names, those are updated in
+	 * the log by link/unlink/rename operations.
+	 * In case the inode was logged and then evicted and reloaded, its
+	 * logged_trans will be 0, in which case we have to fully log it since
+	 * logged_trans is a transient field, not persisted.
+	 */
+	if (inode->logged_trans == trans->transid &&
+	    !test_bit(BTRFS_INODE_COPY_EVERYTHING, &inode->runtime_flags))
+		return false;
+
+	return true;
+}
+
+struct btrfs_dir_list {
+	u64 ino;
+	struct list_head list;
+};
+
+/*
+ * Log the inodes of the new dentries of a directory.
+ * See process_dir_items_leaf() for details about why it is needed.
+ * This is a recursive operation - if an existing dentry corresponds to a
+ * directory, that directory's new entries are logged too (same behaviour as
+ * ext3/4, xfs, f2fs, reiserfs, nilfs2). Note that when logging the inodes
+ * the dentries point to we do not acquire their VFS lock, otherwise lockdep
+ * complains about the following circular lock dependency / possible deadlock:
+ *
+ *        CPU0                                        CPU1
+ *        ----                                        ----
+ * lock(&type->i_mutex_dir_key#3/2);
+ *                                            lock(sb_internal#2);
+ *                                            lock(&type->i_mutex_dir_key#3/2);
+ * lock(&sb->s_type->i_mutex_key#14);
+ *
+ * Where sb_internal is the lock (a counter that works as a lock) acquired by
+ * sb_start_intwrite() in btrfs_start_transaction().
+ * Not acquiring the VFS lock of the inodes is still safe because:
+ *
+ * 1) For regular files we log with a mode of LOG_INODE_EXISTS. It's possible
+ *    that while logging the inode new references (names) are added or removed
+ *    from the inode, leaving the logged inode item with a link count that does
+ *    not match the number of logged inode reference items. This is fine because
+ *    at log replay time we compute the real number of links and correct the
+ *    link count in the inode item (see replay_one_buffer() and
+ *    link_to_fixup_dir());
+ *
+ * 2) For directories we log with a mode of LOG_INODE_ALL. It's possible that
+ *    while logging the inode's items new index items (key type
+ *    BTRFS_DIR_INDEX_KEY) are added to fs/subvol tree and the logged inode item
+ *    has a size that doesn't match the sum of the lengths of all the logged
+ *    names - this is ok, not a problem, because at log replay time we set the
+ *    directory's i_size to the correct value (see replay_one_name() and
+ *    do_overwrite_item()).
+ */
+static int log_new_dir_dentries(struct btrfs_trans_handle *trans,
+				struct btrfs_inode *start_inode,
+				struct btrfs_log_ctx *ctx)
+{
+	struct btrfs_root *root = start_inode->root;
+	struct btrfs_fs_info *fs_info = root->fs_info;
+	struct btrfs_path *path;
+	LIST_HEAD(dir_list);
+	struct btrfs_dir_list *dir_elem;
+	u64 ino = btrfs_ino(start_inode);
+	int ret = 0;
+
+	/*
+	 * If we are logging a new name, as part of a link or rename operation,
+	 * don't bother logging new dentries, as we just want to log the names
+	 * of an inode and that any new parents exist.
+	 */
+	if (ctx->logging_new_name)
+		return 0;
+
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+
+	while (true) {
+		struct extent_buffer *leaf;
+		struct btrfs_key min_key;
+		bool continue_curr_inode = true;
+		int nritems;
+		int i;
+
+		min_key.objectid = ino;
+		min_key.type = BTRFS_DIR_INDEX_KEY;
+		min_key.offset = 0;
+again:
+		btrfs_release_path(path);
+		ret = btrfs_search_forward(root, &min_key, path, trans->transid);
+		if (ret < 0) {
+			break;
+		} else if (ret > 0) {
+			ret = 0;
+			goto next;
+		}
+
+		leaf = path->nodes[0];
+		nritems = btrfs_header_nritems(leaf);
+		for (i = path->slots[0]; i < nritems; i++) {
+			struct btrfs_dir_item *di;
+			struct btrfs_key di_key;
+			struct inode *di_inode;
+			int log_mode = LOG_INODE_EXISTS;
+			int type;
+
+			btrfs_item_key_to_cpu(leaf, &min_key, i);
+			if (min_key.objectid != ino ||
+			    min_key.type != BTRFS_DIR_INDEX_KEY) {
+				continue_curr_inode = false;
+				break;
+			}
+
+			di = btrfs_item_ptr(leaf, i, struct btrfs_dir_item);
+			type = btrfs_dir_type(leaf, di);
+			if (btrfs_dir_transid(leaf, di) < trans->transid)
+				continue;
+			btrfs_dir_item_key_to_cpu(leaf, di, &di_key);
+			if (di_key.type == BTRFS_ROOT_ITEM_KEY)
+				continue;
+
+			btrfs_release_path(path);
+			di_inode = btrfs_iget(fs_info->sb, di_key.objectid, root);
+			if (IS_ERR(di_inode)) {
+				ret = PTR_ERR(di_inode);
+				goto out;
+			}
+
+			if (!need_log_inode(trans, BTRFS_I(di_inode))) {
+				btrfs_add_delayed_iput(di_inode);
+				break;
+			}
+
+			ctx->log_new_dentries = false;
+			if (type == BTRFS_FT_DIR)
+				log_mode = LOG_INODE_ALL;
+			ret = btrfs_log_inode(trans, BTRFS_I(di_inode),
+					      log_mode, ctx);
+			btrfs_add_delayed_iput(di_inode);
+			if (ret)
+				goto out;
+			if (ctx->log_new_dentries) {
+				dir_elem = kmalloc(sizeof(*dir_elem), GFP_NOFS);
+				if (!dir_elem) {
+					ret = -ENOMEM;
+					goto out;
+				}
+				dir_elem->ino = di_key.objectid;
+				list_add_tail(&dir_elem->list, &dir_list);
+			}
+			break;
+		}
+
+		if (continue_curr_inode && min_key.offset < (u64)-1) {
+			min_key.offset++;
+			goto again;
+		}
+
+next:
+		if (list_empty(&dir_list))
+			break;
+
+		dir_elem = list_first_entry(&dir_list, struct btrfs_dir_list, list);
+		ino = dir_elem->ino;
+		list_del(&dir_elem->list);
+		kfree(dir_elem);
+	}
+out:
+	btrfs_free_path(path);
+	if (ret) {
+		struct btrfs_dir_list *next;
+
+		list_for_each_entry_safe(dir_elem, next, &dir_list, list)
+			kfree(dir_elem);
+	}
+
+	return ret;
+}
+
 struct btrfs_ino_list {
 	u64 ino;
 	u64 parent;
 	struct list_head list;
 };
 
-static int log_conflicting_inodes(struct btrfs_trans_handle *trans,
-				  struct btrfs_root *root,
-				  struct btrfs_path *path,
-				  struct btrfs_log_ctx *ctx,
-				  u64 ino, u64 parent)
+static void free_conflicting_inodes(struct btrfs_log_ctx *ctx)
+{
+	struct btrfs_ino_list *curr;
+	struct btrfs_ino_list *next;
+
+	list_for_each_entry_safe(curr, next, &ctx->conflict_inodes, list) {
+		list_del(&curr->list);
+		kfree(curr);
+	}
+}
+
+static int conflicting_inode_is_dir(struct btrfs_root *root, u64 ino,
+				    struct btrfs_path *path)
+{
+	struct btrfs_key key;
+	int ret;
+
+	key.objectid = ino;
+	key.type = BTRFS_INODE_ITEM_KEY;
+	key.offset = 0;
+
+	path->search_commit_root = 1;
+	path->skip_locking = 1;
+
+	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
+	if (WARN_ON_ONCE(ret > 0)) {
+		/*
+		 * We have previously found the inode through the commit root
+		 * so this should not happen. If it does, just error out and
+		 * fallback to a transaction commit.
+		 */
+		ret = -ENOENT;
+	} else if (ret == 0) {
+		struct btrfs_inode_item *item;
+
+		item = btrfs_item_ptr(path->nodes[0], path->slots[0],
+				      struct btrfs_inode_item);
+		if (S_ISDIR(btrfs_inode_mode(path->nodes[0], item)))
+			ret = 1;
+	}
+
+	btrfs_release_path(path);
+	path->search_commit_root = 0;
+	path->skip_locking = 0;
+
+	return ret;
+}
+
+static int add_conflicting_inode(struct btrfs_trans_handle *trans,
+				 struct btrfs_root *root,
+				 struct btrfs_path *path,
+				 u64 ino, u64 parent,
+				 struct btrfs_log_ctx *ctx)
 {
 	struct btrfs_ino_list *ino_elem;
-	LIST_HEAD(inode_list);
-	int ret = 0;
+	struct inode *inode;
+
+	/*
+	 * It's rare to have a lot of conflicting inodes, in practice it is not
+	 * common to have more than 1 or 2. We don't want to collect too many,
+	 * as we could end up logging too many inodes (even if only in
+	 * LOG_INODE_EXISTS mode) and slow down other fsyncs or transaction
+	 * commits.
+	 */
+	if (ctx->num_conflict_inodes >= MAX_CONFLICT_INODES)
+		return BTRFS_LOG_FORCE_COMMIT;
+
+	inode = btrfs_iget(root->fs_info->sb, ino, root);
+	/*
+	 * If the other inode that had a conflicting dir entry was deleted in
+	 * the current transaction then we either:
+	 *
+	 * 1) Log the parent directory (later after adding it to the list) if
+	 *    the inode is a directory. This is because it may be a deleted
+	 *    subvolume/snapshot or it may be a regular directory that had
+	 *    deleted subvolumes/snapshots (or subdirectories that had them),
+	 *    and at the moment we can't deal with dropping subvolumes/snapshots
+	 *    during log replay. So we just log the parent, which will result in
+	 *    a fallback to a transaction commit if we are dealing with those
+	 *    cases (last_unlink_trans will match the current transaction);
+	 *
+	 * 2) Do nothing if it's not a directory. During log replay we simply
+	 *    unlink the conflicting dentry from the parent directory and then
+	 *    add the dentry for our inode. Like this we can avoid logging the
+	 *    parent directory (and maybe fallback to a transaction commit in
+	 *    case it has a last_unlink_trans == trans->transid, due to moving
+	 *    some inode from it to some other directory).
+	 */
+	if (IS_ERR(inode)) {
+		int ret = PTR_ERR(inode);
+
+		if (ret != -ENOENT)
+			return ret;
+
+		ret = conflicting_inode_is_dir(root, ino, path);
+		/* Not a directory or we got an error. */
+		if (ret <= 0)
+			return ret;
+
+		/* Conflicting inode is a directory, so we'll log its parent. */
+		ino_elem = kmalloc(sizeof(*ino_elem), GFP_NOFS);
+		if (!ino_elem)
+			return -ENOMEM;
+		ino_elem->ino = ino;
+		ino_elem->parent = parent;
+		list_add_tail(&ino_elem->list, &ctx->conflict_inodes);
+		ctx->num_conflict_inodes++;
+
+		return 0;
+	}
+
+	/*
+	 * If the inode was already logged skip it - otherwise we can hit an
+	 * infinite loop. Example:
+	 *
+	 * From the commit root (previous transaction) we have the following
+	 * inodes:
+	 *
+	 * inode 257 a directory
+	 * inode 258 with references "zz" and "zz_link" on inode 257
+	 * inode 259 with reference "a" on inode 257
+	 *
+	 * And in the current (uncommitted) transaction we have:
+	 *
+	 * inode 257 a directory, unchanged
+	 * inode 258 with references "a" and "a2" on inode 257
+	 * inode 259 with reference "zz_link" on inode 257
+	 * inode 261 with reference "zz" on inode 257
+	 *
+	 * When logging inode 261 the following infinite loop could
+	 * happen if we don't skip already logged inodes:
+	 *
+	 * - we detect inode 258 as a conflicting inode, with inode 261
+	 *   on reference "zz", and log it;
+	 *
+	 * - we detect inode 259 as a conflicting inode, with inode 258
+	 *   on reference "a", and log it;
+	 *
+	 * - we detect inode 258 as a conflicting inode, with inode 259
+	 *   on reference "zz_link", and log it - again! After this we
+	 *   repeat the above steps forever.
+	 *
+	 * Here we can use need_log_inode() because we only need to log the
+	 * inode in LOG_INODE_EXISTS mode and rename operations update the log,
+	 * so that the log ends up with the new name and without the old name.
+	 */
+	if (!need_log_inode(trans, BTRFS_I(inode))) {
+		btrfs_add_delayed_iput(inode);
+		return 0;
+	}
+
+	btrfs_add_delayed_iput(inode);
 
 	ino_elem = kmalloc(sizeof(*ino_elem), GFP_NOFS);
 	if (!ino_elem)
 		return -ENOMEM;
 	ino_elem->ino = ino;
 	ino_elem->parent = parent;
-	list_add_tail(&ino_elem->list, &inode_list);
+	list_add_tail(&ino_elem->list, &ctx->conflict_inodes);
+	ctx->num_conflict_inodes++;
 
-	while (!list_empty(&inode_list)) {
-		struct btrfs_fs_info *fs_info = root->fs_info;
-		struct btrfs_key key;
-		struct inode *inode;
+	return 0;
+}
 
-		ino_elem = list_first_entry(&inode_list, struct btrfs_ino_list,
-					    list);
-		ino = ino_elem->ino;
-		parent = ino_elem->parent;
-		list_del(&ino_elem->list);
-		kfree(ino_elem);
-		if (ret)
-			continue;
+static int log_conflicting_inodes(struct btrfs_trans_handle *trans,
+				  struct btrfs_root *root,
+				  struct btrfs_log_ctx *ctx)
+{
+	struct btrfs_fs_info *fs_info = root->fs_info;
+	int ret = 0;
 
-		btrfs_release_path(path);
+	/*
+	 * Conflicting inodes are logged by the first call to btrfs_log_inode(),
+	 * otherwise we could have unbounded recursion of btrfs_log_inode()
+	 * calls. This check guarantees we can have only 1 level of recursion.
+	 */
+	if (ctx->logging_conflict_inodes)
+		return 0;
 
-		key.objectid = ino;
-		key.type = BTRFS_INODE_ITEM_KEY;
-		key.offset = 0;
-		inode = btrfs_iget(fs_info->sb, &key, root);
+	ctx->logging_conflict_inodes = true;
+
+	/*
+	 * New conflicting inodes may be found and added to the list while we
+	 * are logging a conflicting inode, so keep iterating while the list is
+	 * not empty.
+	 */
+	while (!list_empty(&ctx->conflict_inodes)) {
+		struct btrfs_ino_list *curr;
+		struct inode *inode;
+		u64 ino;
+		u64 parent;
+
+		curr = list_first_entry(&ctx->conflict_inodes,
+					struct btrfs_ino_list, list);
+		ino = curr->ino;
+		parent = curr->parent;
+		list_del(&curr->list);
+		kfree(curr);
+
+		inode = btrfs_iget(fs_info->sb, ino, root);
 		/*
 		 * If the other inode that had a conflicting dir entry was
 		 * deleted in the current transaction, we need to log its parent
-		 * directory.
+		 * directory. See the comment at add_conflicting_inode().
 		 */
 		if (IS_ERR(inode)) {
 			ret = PTR_ERR(inode);
-			if (ret == -ENOENT) {
-				key.objectid = parent;
-				inode = btrfs_iget(fs_info->sb, &key, root);
-				if (IS_ERR(inode)) {
-					ret = PTR_ERR(inode);
-				} else {
-					ret = btrfs_log_inode(trans, root,
-						      BTRFS_I(inode),
-						      LOG_OTHER_INODE_ALL,
-						      0, LLONG_MAX, ctx);
-					btrfs_add_delayed_iput(inode);
-				}
+			if (ret != -ENOENT)
+				break;
+
+			inode = btrfs_iget(fs_info->sb, parent, root);
+			if (IS_ERR(inode)) {
+				ret = PTR_ERR(inode);
+				break;
 			}
+
+			/*
+			 * Always log the directory, we cannot make this
+			 * conditional on need_log_inode() because the directory
+			 * might have been logged in LOG_INODE_EXISTS mode or
+			 * the dir index of the conflicting inode is not in a
+			 * dir index key range logged for the directory. So we
+			 * must make sure the deletion is recorded.
+			 */
+			ret = btrfs_log_inode(trans, BTRFS_I(inode),
+					      LOG_INODE_ALL, ctx);
+			btrfs_add_delayed_iput(inode);
+			if (ret)
+				break;
 			continue;
 		}
+
 		/*
-		 * If the inode was already logged skip it - otherwise we can
-		 * hit an infinite loop. Example:
-		 *
-		 * From the commit root (previous transaction) we have the
-		 * following inodes:
-		 *
-		 * inode 257 a directory
-		 * inode 258 with references "zz" and "zz_link" on inode 257
-		 * inode 259 with reference "a" on inode 257
-		 *
-		 * And in the current (uncommitted) transaction we have:
-		 *
-		 * inode 257 a directory, unchanged
-		 * inode 258 with references "a" and "a2" on inode 257
-		 * inode 259 with reference "zz_link" on inode 257
-		 * inode 261 with reference "zz" on inode 257
-		 *
-		 * When logging inode 261 the following infinite loop could
-		 * happen if we don't skip already logged inodes:
+		 * Here we can use need_log_inode() because we only need to log
+		 * the inode in LOG_INODE_EXISTS mode and rename operations
+		 * update the log, so that the log ends up with the new name and
+		 * without the old name.
 		 *
-		 * - we detect inode 258 as a conflicting inode, with inode 261
-		 *   on reference "zz", and log it;
-		 *
-		 * - we detect inode 259 as a conflicting inode, with inode 258
-		 *   on reference "a", and log it;
-		 *
-		 * - we detect inode 258 as a conflicting inode, with inode 259
-		 *   on reference "zz_link", and log it - again! After this we
-		 *   repeat the above steps forever.
-		 */
-		spin_lock(&BTRFS_I(inode)->lock);
-		/*
-		 * Check the inode's logged_trans only instead of
-		 * btrfs_inode_in_log(). This is because the last_log_commit of
-		 * the inode is not updated when we only log that it exists and
-		 * and it has the full sync bit set (see btrfs_log_inode()).
+		 * We did this check at add_conflicting_inode(), but here we do
+		 * it again because if some other task logged the inode after
+		 * that, we can avoid doing it again.
 		 */
-		if (BTRFS_I(inode)->logged_trans == trans->transid) {
-			spin_unlock(&BTRFS_I(inode)->lock);
+		if (!need_log_inode(trans, BTRFS_I(inode))) {
 			btrfs_add_delayed_iput(inode);
 			continue;
 		}
-		spin_unlock(&BTRFS_I(inode)->lock);
+
 		/*
 		 * We are safe logging the other inode without acquiring its
 		 * lock as long as we log with the LOG_INODE_EXISTS mode. We
@@ -4873,68 +5747,539 @@ static int log_conflicting_inodes(struct btrfs_trans_handle *trans,
 		 * well because during a rename we pin the log and update the
 		 * log with the new name before we unpin it.
 		 */
-		ret = btrfs_log_inode(trans, root, BTRFS_I(inode),
-				      LOG_OTHER_INODE, 0, LLONG_MAX, ctx);
-		if (ret) {
-			btrfs_add_delayed_iput(inode);
-			continue;
-		}
+		ret = btrfs_log_inode(trans, BTRFS_I(inode), LOG_INODE_EXISTS, ctx);
+		btrfs_add_delayed_iput(inode);
+		if (ret)
+			break;
+	}
 
-		key.objectid = ino;
-		key.type = BTRFS_INODE_REF_KEY;
-		key.offset = 0;
-		ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
-		if (ret < 0) {
-			btrfs_add_delayed_iput(inode);
-			continue;
+	ctx->logging_conflict_inodes = false;
+	if (ret)
+		free_conflicting_inodes(ctx);
+
+	return ret;
+}
+
+static int copy_inode_items_to_log(struct btrfs_trans_handle *trans,
+				   struct btrfs_inode *inode,
+				   struct btrfs_key *min_key,
+				   const struct btrfs_key *max_key,
+				   struct btrfs_path *path,
+				   struct btrfs_path *dst_path,
+				   const u64 logged_isize,
+				   const int inode_only,
+				   struct btrfs_log_ctx *ctx,
+				   bool *need_log_inode_item)
+{
+	const u64 i_size = i_size_read(&inode->vfs_inode);
+	struct btrfs_root *root = inode->root;
+	int ins_start_slot = 0;
+	int ins_nr = 0;
+	int ret;
+
+	while (1) {
+		ret = btrfs_search_forward(root, min_key, path, trans->transid);
+		if (ret < 0)
+			return ret;
+		if (ret > 0) {
+			ret = 0;
+			break;
 		}
+again:
+		/* Note, ins_nr might be > 0 here, cleanup outside the loop */
+		if (min_key->objectid != max_key->objectid)
+			break;
+		if (min_key->type > max_key->type)
+			break;
 
-		while (true) {
-			struct extent_buffer *leaf = path->nodes[0];
-			int slot = path->slots[0];
+		if (min_key->type == BTRFS_INODE_ITEM_KEY) {
+			*need_log_inode_item = false;
+		} else if (min_key->type == BTRFS_EXTENT_DATA_KEY &&
+			   min_key->offset >= i_size) {
+			/*
+			 * Extents at and beyond eof are logged with
+			 * btrfs_log_prealloc_extents().
+			 * Only regular files have BTRFS_EXTENT_DATA_KEY keys,
+			 * and no keys greater than that, so bail out.
+			 */
+			break;
+		} else if ((min_key->type == BTRFS_INODE_REF_KEY ||
+			    min_key->type == BTRFS_INODE_EXTREF_KEY) &&
+			   (inode->generation == trans->transid ||
+			    ctx->logging_conflict_inodes)) {
 			u64 other_ino = 0;
 			u64 other_parent = 0;
 
-			if (slot >= btrfs_header_nritems(leaf)) {
-				ret = btrfs_next_leaf(root, path);
-				if (ret < 0) {
-					break;
-				} else if (ret > 0) {
-					ret = 0;
-					break;
+			ret = btrfs_check_ref_name_override(path->nodes[0],
+					path->slots[0], min_key, inode,
+					&other_ino, &other_parent);
+			if (ret < 0) {
+				return ret;
+			} else if (ret > 0 &&
+				   other_ino != btrfs_ino(BTRFS_I(ctx->inode))) {
+				if (ins_nr > 0) {
+					ins_nr++;
+				} else {
+					ins_nr = 1;
+					ins_start_slot = path->slots[0];
 				}
-				continue;
-			}
+				ret = copy_items(trans, inode, dst_path, path,
+						 ins_start_slot, ins_nr,
+						 inode_only, logged_isize);
+				if (ret < 0)
+					return ret;
+				ins_nr = 0;
 
-			btrfs_item_key_to_cpu(leaf, &key, slot);
-			if (key.objectid != ino ||
-			    (key.type != BTRFS_INODE_REF_KEY &&
-			     key.type != BTRFS_INODE_EXTREF_KEY)) {
-				ret = 0;
-				break;
+				btrfs_release_path(path);
+				ret = add_conflicting_inode(trans, root, path,
+							    other_ino,
+							    other_parent, ctx);
+				if (ret)
+					return ret;
+				goto next_key;
 			}
+		} else if (min_key->type == BTRFS_XATTR_ITEM_KEY) {
+			/* Skip xattrs, logged later with btrfs_log_all_xattrs() */
+			if (ins_nr == 0)
+				goto next_slot;
+			ret = copy_items(trans, inode, dst_path, path,
+					 ins_start_slot,
+					 ins_nr, inode_only, logged_isize);
+			if (ret < 0)
+				return ret;
+			ins_nr = 0;
+			goto next_slot;
+		}
+
+		if (ins_nr && ins_start_slot + ins_nr == path->slots[0]) {
+			ins_nr++;
+			goto next_slot;
+		} else if (!ins_nr) {
+			ins_start_slot = path->slots[0];
+			ins_nr = 1;
+			goto next_slot;
+		}
 
-			ret = btrfs_check_ref_name_override(leaf, slot, &key,
-					BTRFS_I(inode), &other_ino,
-					&other_parent);
+		ret = copy_items(trans, inode, dst_path, path, ins_start_slot,
+				 ins_nr, inode_only, logged_isize);
+		if (ret < 0)
+			return ret;
+		ins_nr = 1;
+		ins_start_slot = path->slots[0];
+next_slot:
+		path->slots[0]++;
+		if (path->slots[0] < btrfs_header_nritems(path->nodes[0])) {
+			btrfs_item_key_to_cpu(path->nodes[0], min_key,
+					      path->slots[0]);
+			goto again;
+		}
+		if (ins_nr) {
+			ret = copy_items(trans, inode, dst_path, path,
+					 ins_start_slot, ins_nr, inode_only,
+					 logged_isize);
 			if (ret < 0)
+				return ret;
+			ins_nr = 0;
+		}
+		btrfs_release_path(path);
+next_key:
+		if (min_key->offset < (u64)-1) {
+			min_key->offset++;
+		} else if (min_key->type < max_key->type) {
+			min_key->type++;
+			min_key->offset = 0;
+		} else {
+			break;
+		}
+
+		/*
+		 * We may process many leaves full of items for our inode, so
+		 * avoid monopolizing a cpu for too long by rescheduling while
+		 * not holding locks on any tree.
+		 */
+		cond_resched();
+	}
+	if (ins_nr) {
+		ret = copy_items(trans, inode, dst_path, path, ins_start_slot,
+				 ins_nr, inode_only, logged_isize);
+		if (ret)
+			return ret;
+	}
+
+	if (inode_only == LOG_INODE_ALL && S_ISREG(inode->vfs_inode.i_mode)) {
+		/*
+		 * Release the path because otherwise we might attempt to double
+		 * lock the same leaf with btrfs_log_prealloc_extents() below.
+		 */
+		btrfs_release_path(path);
+		ret = btrfs_log_prealloc_extents(trans, inode, dst_path);
+	}
+
+	return ret;
+}
+
+static int insert_delayed_items_batch(struct btrfs_trans_handle *trans,
+				      struct btrfs_root *log,
+				      struct btrfs_path *path,
+				      const struct btrfs_item_batch *batch,
+				      const struct btrfs_delayed_item *first_item)
+{
+	const struct btrfs_delayed_item *curr = first_item;
+	int ret;
+
+	ret = btrfs_insert_empty_items(trans, log, path, batch);
+	if (ret)
+		return ret;
+
+	for (int i = 0; i < batch->nr; i++) {
+		char *data_ptr;
+
+		data_ptr = btrfs_item_ptr(path->nodes[0], path->slots[0], char);
+		write_extent_buffer(path->nodes[0], &curr->data,
+				    (unsigned long)data_ptr, curr->data_len);
+		curr = list_next_entry(curr, log_list);
+		path->slots[0]++;
+	}
+
+	btrfs_release_path(path);
+
+	return 0;
+}
+
+static int log_delayed_insertion_items(struct btrfs_trans_handle *trans,
+				       struct btrfs_inode *inode,
+				       struct btrfs_path *path,
+				       const struct list_head *delayed_ins_list,
+				       struct btrfs_log_ctx *ctx)
+{
+	/* 195 (4095 bytes of keys and sizes) fits in a single 4K page. */
+	const int max_batch_size = 195;
+	const int leaf_data_size = BTRFS_LEAF_DATA_SIZE(trans->fs_info);
+	const u64 ino = btrfs_ino(inode);
+	struct btrfs_root *log = inode->root->log_root;
+	struct btrfs_item_batch batch = {
+		.nr = 0,
+		.total_data_size = 0,
+	};
+	const struct btrfs_delayed_item *first = NULL;
+	const struct btrfs_delayed_item *curr;
+	char *ins_data;
+	struct btrfs_key *ins_keys;
+	u32 *ins_sizes;
+	u64 curr_batch_size = 0;
+	int batch_idx = 0;
+	int ret;
+
+	/* We are adding dir index items to the log tree. */
+	lockdep_assert_held(&inode->log_mutex);
+
+	/*
+	 * We collect delayed items before copying index keys from the subvolume
+	 * to the log tree. However just after we collected them, they may have
+	 * been flushed (all of them or just some of them), and therefore we
+	 * could have copied them from the subvolume tree to the log tree.
+	 * So find the first delayed item that was not yet logged (they are
+	 * sorted by index number).
+	 */
+	list_for_each_entry(curr, delayed_ins_list, log_list) {
+		if (curr->index > inode->last_dir_index_offset) {
+			first = curr;
+			break;
+		}
+	}
+
+	/* Empty list or all delayed items were already logged. */
+	if (!first)
+		return 0;
+
+	ins_data = kmalloc(max_batch_size * sizeof(u32) +
+			   max_batch_size * sizeof(struct btrfs_key), GFP_NOFS);
+	if (!ins_data)
+		return -ENOMEM;
+	ins_sizes = (u32 *)ins_data;
+	batch.data_sizes = ins_sizes;
+	ins_keys = (struct btrfs_key *)(ins_data + max_batch_size * sizeof(u32));
+	batch.keys = ins_keys;
+
+	curr = first;
+	while (!list_entry_is_head(curr, delayed_ins_list, log_list)) {
+		const u32 curr_size = curr->data_len + sizeof(struct btrfs_item);
+
+		if (curr_batch_size + curr_size > leaf_data_size ||
+		    batch.nr == max_batch_size) {
+			ret = insert_delayed_items_batch(trans, log, path,
+							 &batch, first);
+			if (ret)
+				goto out;
+			batch_idx = 0;
+			batch.nr = 0;
+			batch.total_data_size = 0;
+			curr_batch_size = 0;
+			first = curr;
+		}
+
+		ins_sizes[batch_idx] = curr->data_len;
+		ins_keys[batch_idx].objectid = ino;
+		ins_keys[batch_idx].type = BTRFS_DIR_INDEX_KEY;
+		ins_keys[batch_idx].offset = curr->index;
+		curr_batch_size += curr_size;
+		batch.total_data_size += curr->data_len;
+		batch.nr++;
+		batch_idx++;
+		curr = list_next_entry(curr, log_list);
+	}
+
+	ASSERT(batch.nr >= 1);
+	ret = insert_delayed_items_batch(trans, log, path, &batch, first);
+
+	curr = list_last_entry(delayed_ins_list, struct btrfs_delayed_item,
+			       log_list);
+	inode->last_dir_index_offset = curr->index;
+out:
+	kfree(ins_data);
+
+	return ret;
+}
+
+static int log_delayed_deletions_full(struct btrfs_trans_handle *trans,
+				      struct btrfs_inode *inode,
+				      struct btrfs_path *path,
+				      const struct list_head *delayed_del_list,
+				      struct btrfs_log_ctx *ctx)
+{
+	const u64 ino = btrfs_ino(inode);
+	const struct btrfs_delayed_item *curr;
+
+	curr = list_first_entry(delayed_del_list, struct btrfs_delayed_item,
+				log_list);
+
+	while (!list_entry_is_head(curr, delayed_del_list, log_list)) {
+		u64 first_dir_index = curr->index;
+		u64 last_dir_index;
+		const struct btrfs_delayed_item *next;
+		int ret;
+
+		/*
+		 * Find a range of consecutive dir index items to delete. Like
+		 * this we log a single dir range item spanning several contiguous
+		 * dir items instead of logging one range item per dir index item.
+		 */
+		next = list_next_entry(curr, log_list);
+		while (!list_entry_is_head(next, delayed_del_list, log_list)) {
+			if (next->index != curr->index + 1)
 				break;
-			if (ret > 0) {
-				ino_elem = kmalloc(sizeof(*ino_elem), GFP_NOFS);
-				if (!ino_elem) {
-					ret = -ENOMEM;
-					break;
-				}
-				ino_elem->ino = other_ino;
-				ino_elem->parent = other_parent;
-				list_add_tail(&ino_elem->list, &inode_list);
-				ret = 0;
-			}
-			path->slots[0]++;
+			curr = next;
+			next = list_next_entry(next, log_list);
 		}
-		btrfs_add_delayed_iput(inode);
+
+		last_dir_index = curr->index;
+		ASSERT(last_dir_index >= first_dir_index);
+
+		ret = insert_dir_log_key(trans, inode->root->log_root, path,
+					 ino, first_dir_index, last_dir_index);
+		if (ret)
+			return ret;
+		curr = list_next_entry(curr, log_list);
+	}
+
+	return 0;
+}
+
+static int batch_delete_dir_index_items(struct btrfs_trans_handle *trans,
+					struct btrfs_inode *inode,
+					struct btrfs_path *path,
+					struct btrfs_log_ctx *ctx,
+					const struct list_head *delayed_del_list,
+					const struct btrfs_delayed_item *first,
+					const struct btrfs_delayed_item **last_ret)
+{
+	const struct btrfs_delayed_item *next;
+	struct extent_buffer *leaf = path->nodes[0];
+	const int last_slot = btrfs_header_nritems(leaf) - 1;
+	int slot = path->slots[0] + 1;
+	const u64 ino = btrfs_ino(inode);
+
+	next = list_next_entry(first, log_list);
+
+	while (slot < last_slot &&
+	       !list_entry_is_head(next, delayed_del_list, log_list)) {
+		struct btrfs_key key;
+
+		btrfs_item_key_to_cpu(leaf, &key, slot);
+		if (key.objectid != ino ||
+		    key.type != BTRFS_DIR_INDEX_KEY ||
+		    key.offset != next->index)
+			break;
+
+		slot++;
+		*last_ret = next;
+		next = list_next_entry(next, log_list);
+	}
+
+	return btrfs_del_items(trans, inode->root->log_root, path,
+			       path->slots[0], slot - path->slots[0]);
+}
+
+static int log_delayed_deletions_incremental(struct btrfs_trans_handle *trans,
+					     struct btrfs_inode *inode,
+					     struct btrfs_path *path,
+					     const struct list_head *delayed_del_list,
+					     struct btrfs_log_ctx *ctx)
+{
+	struct btrfs_root *log = inode->root->log_root;
+	const struct btrfs_delayed_item *curr;
+	u64 last_range_start;
+	u64 last_range_end = 0;
+	struct btrfs_key key;
+
+	key.objectid = btrfs_ino(inode);
+	key.type = BTRFS_DIR_INDEX_KEY;
+	curr = list_first_entry(delayed_del_list, struct btrfs_delayed_item,
+				log_list);
+
+	while (!list_entry_is_head(curr, delayed_del_list, log_list)) {
+		const struct btrfs_delayed_item *last = curr;
+		u64 first_dir_index = curr->index;
+		u64 last_dir_index;
+		bool deleted_items = false;
+		int ret;
+
+		key.offset = curr->index;
+		ret = btrfs_search_slot(trans, log, &key, path, -1, 1);
+		if (ret < 0) {
+			return ret;
+		} else if (ret == 0) {
+			ret = batch_delete_dir_index_items(trans, inode, path, ctx,
+							   delayed_del_list, curr,
+							   &last);
+			if (ret)
+				return ret;
+			deleted_items = true;
+		}
+
+		btrfs_release_path(path);
+
+		/*
+		 * If we deleted items from the leaf, it means we have a range
+		 * item logging their range, so no need to add one or update an
+		 * existing one. Otherwise we have to log a dir range item.
+		 */
+		if (deleted_items)
+			goto next_batch;
+
+		last_dir_index = last->index;
+		ASSERT(last_dir_index >= first_dir_index);
+		/*
+		 * If this range starts right after where the previous one ends,
+		 * then we want to reuse the previous range item and change its
+		 * end offset to the end of this range. This is just to minimize
+		 * leaf space usage, by avoiding adding a new range item.
+		 */
+		if (last_range_end != 0 && first_dir_index == last_range_end + 1)
+			first_dir_index = last_range_start;
+
+		ret = insert_dir_log_key(trans, log, path, key.objectid,
+					 first_dir_index, last_dir_index);
+		if (ret)
+			return ret;
+
+		last_range_start = first_dir_index;
+		last_range_end = last_dir_index;
+next_batch:
+		curr = list_next_entry(last, log_list);
+	}
+
+	return 0;
+}
+
+static int log_delayed_deletion_items(struct btrfs_trans_handle *trans,
+				      struct btrfs_inode *inode,
+				      struct btrfs_path *path,
+				      const struct list_head *delayed_del_list,
+				      struct btrfs_log_ctx *ctx)
+{
+	/*
+	 * We are deleting dir index items from the log tree or adding range
+	 * items to it.
+	 */
+	lockdep_assert_held(&inode->log_mutex);
+
+	if (list_empty(delayed_del_list))
+		return 0;
+
+	if (ctx->logged_before)
+		return log_delayed_deletions_incremental(trans, inode, path,
+							 delayed_del_list, ctx);
+
+	return log_delayed_deletions_full(trans, inode, path, delayed_del_list,
+					  ctx);
+}
+
+/*
+ * Similar logic as for log_new_dir_dentries(), but it iterates over the delayed
+ * items instead of the subvolume tree.
+ */
+static int log_new_delayed_dentries(struct btrfs_trans_handle *trans,
+				    struct btrfs_inode *inode,
+				    const struct list_head *delayed_ins_list,
+				    struct btrfs_log_ctx *ctx)
+{
+	const bool orig_log_new_dentries = ctx->log_new_dentries;
+	struct btrfs_fs_info *fs_info = trans->fs_info;
+	struct btrfs_delayed_item *item;
+	int ret = 0;
+
+	/*
+	 * No need for the log mutex, plus to avoid potential deadlocks or
+	 * lockdep annotations due to nesting of delayed inode mutexes and log
+	 * mutexes.
+	 */
+	lockdep_assert_not_held(&inode->log_mutex);
+
+	ASSERT(!ctx->logging_new_delayed_dentries);
+	ctx->logging_new_delayed_dentries = true;
+
+	list_for_each_entry(item, delayed_ins_list, log_list) {
+		struct btrfs_dir_item *dir_item;
+		struct inode *di_inode;
+		struct btrfs_key key;
+		int log_mode = LOG_INODE_EXISTS;
+
+		dir_item = (struct btrfs_dir_item *)item->data;
+		btrfs_disk_key_to_cpu(&key, &dir_item->location);
+
+		if (key.type == BTRFS_ROOT_ITEM_KEY)
+			continue;
+
+		di_inode = btrfs_iget(fs_info->sb, key.objectid, inode->root);
+		if (IS_ERR(di_inode)) {
+			ret = PTR_ERR(di_inode);
+			break;
+		}
+
+		if (!need_log_inode(trans, BTRFS_I(di_inode))) {
+			btrfs_add_delayed_iput(di_inode);
+			continue;
+		}
+
+		if (btrfs_stack_dir_type(dir_item) == BTRFS_FT_DIR)
+			log_mode = LOG_INODE_ALL;
+
+		ctx->log_new_dentries = false;
+		ret = btrfs_log_inode(trans, BTRFS_I(di_inode), log_mode, ctx);
+
+		if (!ret && ctx->log_new_dentries)
+			ret = log_new_dir_dentries(trans, BTRFS_I(di_inode), ctx);
+
+		btrfs_add_delayed_iput(di_inode);
+
+		if (ret)
+			break;
 	}
 
+	ctx->log_new_dentries = orig_log_new_dentries;
+	ctx->logging_new_delayed_dentries = false;
+
 	return ret;
 }
 
@@ -4953,30 +6298,26 @@ static int log_conflicting_inodes(struct btrfs_trans_handle *trans,
  * This handles both files and directories.
  */
 static int btrfs_log_inode(struct btrfs_trans_handle *trans,
-			   struct btrfs_root *root, struct btrfs_inode *inode,
+			   struct btrfs_inode *inode,
 			   int inode_only,
-			   const loff_t start,
-			   const loff_t end,
 			   struct btrfs_log_ctx *ctx)
 {
-	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct btrfs_path *path;
 	struct btrfs_path *dst_path;
 	struct btrfs_key min_key;
 	struct btrfs_key max_key;
-	struct btrfs_root *log = root->log_root;
-	int err = 0;
+	struct btrfs_root *log = inode->root->log_root;
 	int ret;
-	int nritems;
-	int ins_start_slot = 0;
-	int ins_nr;
 	bool fast_search = false;
 	u64 ino = btrfs_ino(inode);
 	struct extent_map_tree *em_tree = &inode->extent_tree;
 	u64 logged_isize = 0;
 	bool need_log_inode_item = true;
 	bool xattrs_logged = false;
-	bool recursive_logging = false;
+	bool inode_item_dropped = true;
+	bool full_dir_logging = false;
+	LIST_HEAD(delayed_ins_list);
+	LIST_HEAD(delayed_del_list);
 
 	path = btrfs_alloc_path();
 	if (!path)
@@ -5004,33 +6345,81 @@ static int btrfs_log_inode(struct btrfs_trans_handle *trans,
 		max_key.type = (u8)-1;
 	max_key.offset = (u64)-1;
 
+	if (S_ISDIR(inode->vfs_inode.i_mode) && inode_only == LOG_INODE_ALL)
+		full_dir_logging = true;
+
 	/*
-	 * Only run delayed items if we are a dir or a new file.
-	 * Otherwise commit the delayed inode only, which is needed in
-	 * order for the log replay code to mark inodes for link count
-	 * fixup (create temporary BTRFS_TREE_LOG_FIXUP_OBJECTID items).
+	 * If we are logging a directory while we are logging dentries of the
+	 * delayed items of some other inode, then we need to flush the delayed
+	 * items of this directory and not log the delayed items directly. This
+	 * is to prevent more than one level of recursion into btrfs_log_inode()
+	 * by having something like this:
+	 *
+	 *     $ mkdir -p a/b/c/d/e/f/g/h/...
+	 *     $ xfs_io -c "fsync" a
+	 *
+	 * Where all directories in the path did not exist before and are
+	 * created in the current transaction.
+	 * So in such a case we directly log the delayed items of the main
+	 * directory ("a") without flushing them first, while for each of its
+	 * subdirectories we flush their delayed items before logging them.
+	 * This prevents a potential unbounded recursion like this:
+	 *
+	 * btrfs_log_inode()
+	 *   log_new_delayed_dentries()
+	 *      btrfs_log_inode()
+	 *        log_new_delayed_dentries()
+	 *          btrfs_log_inode()
+	 *            log_new_delayed_dentries()
+	 *              (...)
+	 *
+	 * We have thresholds for the maximum number of delayed items to have in
+	 * memory, and once they are hit, the items are flushed asynchronously.
+	 * However the limit is quite high, so lets prevent deep levels of
+	 * recursion to happen by limiting the maximum depth to be 1.
 	 */
-	if (S_ISDIR(inode->vfs_inode.i_mode) ||
-	    inode->generation > fs_info->last_trans_committed)
+	if (full_dir_logging && ctx->logging_new_delayed_dentries) {
 		ret = btrfs_commit_inode_delayed_items(trans, inode);
-	else
-		ret = btrfs_commit_inode_delayed_inode(inode);
-
-	if (ret) {
-		btrfs_free_path(path);
-		btrfs_free_path(dst_path);
-		return ret;
+		if (ret)
+			goto out;
 	}
 
-	if (inode_only == LOG_OTHER_INODE || inode_only == LOG_OTHER_INODE_ALL) {
-		recursive_logging = true;
-		if (inode_only == LOG_OTHER_INODE)
-			inode_only = LOG_INODE_EXISTS;
-		else
-			inode_only = LOG_INODE_ALL;
-		mutex_lock_nested(&inode->log_mutex, SINGLE_DEPTH_NESTING);
-	} else {
-		mutex_lock(&inode->log_mutex);
+	mutex_lock(&inode->log_mutex);
+
+	/*
+	 * For symlinks, we must always log their content, which is stored in an
+	 * inline extent, otherwise we could end up with an empty symlink after
+	 * log replay, which is invalid on linux (symlink(2) returns -ENOENT if
+	 * one attempts to create an empty symlink).
+	 * We don't need to worry about flushing delalloc, because when we create
+	 * the inline extent when the symlink is created (we never have delalloc
+	 * for symlinks).
+	 */
+	if (S_ISLNK(inode->vfs_inode.i_mode))
+		inode_only = LOG_INODE_ALL;
+
+	/*
+	 * Before logging the inode item, cache the value returned by
+	 * inode_logged(), because after that we have the need to figure out if
+	 * the inode was previously logged in this transaction.
+	 */
+	ret = inode_logged(trans, inode, path);
+	if (ret < 0)
+		goto out_unlock;
+	ctx->logged_before = (ret == 1);
+	ret = 0;
+
+	/*
+	 * This is for cases where logging a directory could result in losing a
+	 * a file after replaying the log. For example, if we move a file from a
+	 * directory A to a directory B, then fsync directory A, we have no way
+	 * to known the file was moved from A to B, so logging just A would
+	 * result in losing the file after a log replay.
+	 */
+	if (full_dir_logging && inode->last_unlink_trans >= trans->transid) {
+		btrfs_set_log_full_commit(trans);
+		ret = BTRFS_LOG_FORCE_COMMIT;
+		goto out_unlock;
 	}
 
 	/*
@@ -5038,13 +6427,12 @@ static int btrfs_log_inode(struct btrfs_trans_handle *trans,
 	 * copies of everything.
 	 */
 	if (S_ISDIR(inode->vfs_inode.i_mode)) {
-		int max_key_type = BTRFS_DIR_LOG_INDEX_KEY;
-
-		if (inode_only == LOG_INODE_EXISTS)
-			max_key_type = BTRFS_XATTR_ITEM_KEY;
-		ret = drop_objectid_items(trans, log, path, ino, max_key_type);
+		clear_bit(BTRFS_INODE_COPY_EVERYTHING, &inode->runtime_flags);
+		if (ctx->logged_before)
+			ret = drop_inode_items(trans, log, path, inode,
+					       BTRFS_XATTR_ITEM_KEY);
 	} else {
-		if (inode_only == LOG_INODE_EXISTS) {
+		if (inode_only == LOG_INODE_EXISTS && ctx->logged_before) {
 			/*
 			 * Make sure the new inode item we write to the log has
 			 * the same isize as the current one (if it exists).
@@ -5058,27 +6446,25 @@ static int btrfs_log_inode(struct btrfs_trans_handle *trans,
 			 * (zeroes), as if an expanding truncate happened,
 			 * instead of getting a file of 4Kb only.
 			 */
-			err = logged_inode_size(log, inode, path, &logged_isize);
-			if (err)
+			ret = logged_inode_size(log, inode, path, &logged_isize);
+			if (ret)
 				goto out_unlock;
 		}
 		if (test_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
 			     &inode->runtime_flags)) {
 			if (inode_only == LOG_INODE_EXISTS) {
 				max_key.type = BTRFS_XATTR_ITEM_KEY;
-				ret = drop_objectid_items(trans, log, path, ino,
-							  max_key.type);
+				if (ctx->logged_before)
+					ret = drop_inode_items(trans, log, path,
+							       inode, max_key.type);
 			} else {
 				clear_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
 					  &inode->runtime_flags);
 				clear_bit(BTRFS_INODE_COPY_EVERYTHING,
 					  &inode->runtime_flags);
-				while(1) {
-					ret = btrfs_truncate_inode_items(trans,
-						log, &inode->vfs_inode, 0, 0);
-					if (ret != -EAGAIN)
-						break;
-				}
+				if (ctx->logged_before)
+					ret = truncate_inode_items(trans, log,
+								   inode, 0, 0);
 			}
 		} else if (test_and_clear_bit(BTRFS_INODE_COPY_EVERYTHING,
 					      &inode->runtime_flags) ||
@@ -5086,520 +6472,163 @@ static int btrfs_log_inode(struct btrfs_trans_handle *trans,
 			if (inode_only == LOG_INODE_ALL)
 				fast_search = true;
 			max_key.type = BTRFS_XATTR_ITEM_KEY;
-			ret = drop_objectid_items(trans, log, path, ino,
-						  max_key.type);
+			if (ctx->logged_before)
+				ret = drop_inode_items(trans, log, path, inode,
+						       max_key.type);
 		} else {
 			if (inode_only == LOG_INODE_ALL)
 				fast_search = true;
+			inode_item_dropped = false;
 			goto log_extents;
 		}
 
 	}
-	if (ret) {
-		err = ret;
+	if (ret)
 		goto out_unlock;
-	}
 
-	while (1) {
-		ins_nr = 0;
-		ret = btrfs_search_forward(root, &min_key,
-					   path, trans->transid);
-		if (ret < 0) {
-			err = ret;
-			goto out_unlock;
-		}
-		if (ret != 0)
-			break;
-again:
-		/* note, ins_nr might be > 0 here, cleanup outside the loop */
-		if (min_key.objectid != ino)
-			break;
-		if (min_key.type > max_key.type)
-			break;
-
-		if (min_key.type == BTRFS_INODE_ITEM_KEY)
-			need_log_inode_item = false;
-
-		if ((min_key.type == BTRFS_INODE_REF_KEY ||
-		     min_key.type == BTRFS_INODE_EXTREF_KEY) &&
-		    inode->generation == trans->transid &&
-		    !recursive_logging) {
-			u64 other_ino = 0;
-			u64 other_parent = 0;
-
-			ret = btrfs_check_ref_name_override(path->nodes[0],
-					path->slots[0], &min_key, inode,
-					&other_ino, &other_parent);
-			if (ret < 0) {
-				err = ret;
-				goto out_unlock;
-			} else if (ret > 0 && ctx &&
-				   other_ino != btrfs_ino(BTRFS_I(ctx->inode))) {
-				if (ins_nr > 0) {
-					ins_nr++;
-				} else {
-					ins_nr = 1;
-					ins_start_slot = path->slots[0];
-				}
-				ret = copy_items(trans, inode, dst_path, path,
-						 ins_start_slot,
-						 ins_nr, inode_only,
-						 logged_isize);
-				if (ret < 0) {
-					err = ret;
-					goto out_unlock;
-				}
-				ins_nr = 0;
-
-				err = log_conflicting_inodes(trans, root, path,
-						ctx, other_ino, other_parent);
-				if (err)
-					goto out_unlock;
-				btrfs_release_path(path);
-				goto next_key;
-			}
-		}
-
-		/* Skip xattrs, we log them later with btrfs_log_all_xattrs() */
-		if (min_key.type == BTRFS_XATTR_ITEM_KEY) {
-			if (ins_nr == 0)
-				goto next_slot;
-			ret = copy_items(trans, inode, dst_path, path,
-					 ins_start_slot,
-					 ins_nr, inode_only, logged_isize);
-			if (ret < 0) {
-				err = ret;
-				goto out_unlock;
-			}
-			ins_nr = 0;
-			goto next_slot;
-		}
-
-		if (ins_nr && ins_start_slot + ins_nr == path->slots[0]) {
-			ins_nr++;
-			goto next_slot;
-		} else if (!ins_nr) {
-			ins_start_slot = path->slots[0];
-			ins_nr = 1;
-			goto next_slot;
-		}
-
-		ret = copy_items(trans, inode, dst_path, path,
-				 ins_start_slot, ins_nr, inode_only,
-				 logged_isize);
-		if (ret < 0) {
-			err = ret;
-			goto out_unlock;
-		}
-		ins_nr = 1;
-		ins_start_slot = path->slots[0];
-next_slot:
-
-		nritems = btrfs_header_nritems(path->nodes[0]);
-		path->slots[0]++;
-		if (path->slots[0] < nritems) {
-			btrfs_item_key_to_cpu(path->nodes[0], &min_key,
-					      path->slots[0]);
-			goto again;
-		}
-		if (ins_nr) {
-			ret = copy_items(trans, inode, dst_path, path,
-					 ins_start_slot,
-					 ins_nr, inode_only, logged_isize);
-			if (ret < 0) {
-				err = ret;
-				goto out_unlock;
-			}
-			ins_nr = 0;
-		}
-		btrfs_release_path(path);
-next_key:
-		if (min_key.offset < (u64)-1) {
-			min_key.offset++;
-		} else if (min_key.type < max_key.type) {
-			min_key.type++;
-			min_key.offset = 0;
-		} else {
-			break;
-		}
-	}
-	if (ins_nr) {
-		ret = copy_items(trans, inode, dst_path, path,
-				 ins_start_slot, ins_nr, inode_only,
-				 logged_isize);
-		if (ret < 0) {
-			err = ret;
-			goto out_unlock;
-		}
-		ins_nr = 0;
-	}
+	/*
+	 * If we are logging a directory in full mode, collect the delayed items
+	 * before iterating the subvolume tree, so that we don't miss any new
+	 * dir index items in case they get flushed while or right after we are
+	 * iterating the subvolume tree.
+	 */
+	if (full_dir_logging && !ctx->logging_new_delayed_dentries)
+		btrfs_log_get_delayed_items(inode, &delayed_ins_list,
+					    &delayed_del_list);
+
+	ret = copy_inode_items_to_log(trans, inode, &min_key, &max_key,
+				      path, dst_path, logged_isize,
+				      inode_only, ctx,
+				      &need_log_inode_item);
+	if (ret)
+		goto out_unlock;
 
 	btrfs_release_path(path);
 	btrfs_release_path(dst_path);
-	err = btrfs_log_all_xattrs(trans, root, inode, path, dst_path);
-	if (err)
+	ret = btrfs_log_all_xattrs(trans, inode, path, dst_path);
+	if (ret)
 		goto out_unlock;
 	xattrs_logged = true;
 	if (max_key.type >= BTRFS_EXTENT_DATA_KEY && !fast_search) {
 		btrfs_release_path(path);
 		btrfs_release_path(dst_path);
-		err = btrfs_log_holes(trans, root, inode, path);
-		if (err)
+		ret = btrfs_log_holes(trans, inode, path);
+		if (ret)
 			goto out_unlock;
 	}
 log_extents:
 	btrfs_release_path(path);
 	btrfs_release_path(dst_path);
 	if (need_log_inode_item) {
-		err = log_inode_item(trans, log, dst_path, inode);
-		if (!err && !xattrs_logged) {
-			err = btrfs_log_all_xattrs(trans, root, inode, path,
-						   dst_path);
+		ret = log_inode_item(trans, log, dst_path, inode, inode_item_dropped);
+		if (ret)
+			goto out_unlock;
+		/*
+		 * If we are doing a fast fsync and the inode was logged before
+		 * in this transaction, we don't need to log the xattrs because
+		 * they were logged before. If xattrs were added, changed or
+		 * deleted since the last time we logged the inode, then we have
+		 * already logged them because the inode had the runtime flag
+		 * BTRFS_INODE_COPY_EVERYTHING set.
+		 */
+		if (!xattrs_logged && inode->logged_trans < trans->transid) {
+			ret = btrfs_log_all_xattrs(trans, inode, path, dst_path);
+			if (ret)
+				goto out_unlock;
 			btrfs_release_path(path);
 		}
-		if (err)
-			goto out_unlock;
 	}
 	if (fast_search) {
-		ret = btrfs_log_changed_extents(trans, root, inode, dst_path,
-						ctx, start, end);
-		if (ret) {
-			err = ret;
+		ret = btrfs_log_changed_extents(trans, inode, dst_path, ctx);
+		if (ret)
 			goto out_unlock;
-		}
 	} else if (inode_only == LOG_INODE_ALL) {
 		struct extent_map *em, *n;
 
 		write_lock(&em_tree->lock);
-		/*
-		 * We can't just remove every em if we're called for a ranged
-		 * fsync - that is, one that doesn't cover the whole possible
-		 * file range (0 to LLONG_MAX). This is because we can have
-		 * em's that fall outside the range we're logging and therefore
-		 * their ordered operations haven't completed yet
-		 * (btrfs_finish_ordered_io() not invoked yet). This means we
-		 * didn't get their respective file extent item in the fs/subvol
-		 * tree yet, and need to let the next fast fsync (one which
-		 * consults the list of modified extent maps) find the em so
-		 * that it logs a matching file extent item and waits for the
-		 * respective ordered operation to complete (if it's still
-		 * running).
-		 *
-		 * Removing every em outside the range we're logging would make
-		 * the next fast fsync not log their matching file extent items,
-		 * therefore making us lose data after a log replay.
-		 */
-		list_for_each_entry_safe(em, n, &em_tree->modified_extents,
-					 list) {
-			const u64 mod_end = em->mod_start + em->mod_len - 1;
-
-			if (em->mod_start >= start && mod_end <= end)
-				list_del_init(&em->list);
-		}
+		list_for_each_entry_safe(em, n, &em_tree->modified_extents, list)
+			list_del_init(&em->list);
 		write_unlock(&em_tree->lock);
 	}
 
-	if (inode_only == LOG_INODE_ALL && S_ISDIR(inode->vfs_inode.i_mode)) {
-		ret = log_directory_changes(trans, root, inode, path, dst_path,
-					ctx);
-		if (ret) {
-			err = ret;
+	if (full_dir_logging) {
+		ret = log_directory_changes(trans, inode, path, dst_path, ctx);
+		if (ret)
+			goto out_unlock;
+		ret = log_delayed_insertion_items(trans, inode, path,
+						  &delayed_ins_list, ctx);
+		if (ret)
+			goto out_unlock;
+		ret = log_delayed_deletion_items(trans, inode, path,
+						 &delayed_del_list, ctx);
+		if (ret)
 			goto out_unlock;
-		}
 	}
 
-	/*
-	 * Don't update last_log_commit if we logged that an inode exists after
-	 * it was loaded to memory (full_sync bit set).
-	 * This is to prevent data loss when we do a write to the inode, then
-	 * the inode gets evicted after all delalloc was flushed, then we log
-	 * it exists (due to a rename for example) and then fsync it. This last
-	 * fsync would do nothing (not logging the extents previously written).
-	 */
 	spin_lock(&inode->lock);
 	inode->logged_trans = trans->transid;
-	if (inode_only != LOG_INODE_EXISTS ||
-	    !test_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &inode->runtime_flags))
+	/*
+	 * Don't update last_log_commit if we logged that an inode exists.
+	 * We do this for three reasons:
+	 *
+	 * 1) We might have had buffered writes to this inode that were
+	 *    flushed and had their ordered extents completed in this
+	 *    transaction, but we did not previously log the inode with
+	 *    LOG_INODE_ALL. Later the inode was evicted and after that
+	 *    it was loaded again and this LOG_INODE_EXISTS log operation
+	 *    happened. We must make sure that if an explicit fsync against
+	 *    the inode is performed later, it logs the new extents, an
+	 *    updated inode item, etc, and syncs the log. The same logic
+	 *    applies to direct IO writes instead of buffered writes.
+	 *
+	 * 2) When we log the inode with LOG_INODE_EXISTS, its inode item
+	 *    is logged with an i_size of 0 or whatever value was logged
+	 *    before. If later the i_size of the inode is increased by a
+	 *    truncate operation, the log is synced through an fsync of
+	 *    some other inode and then finally an explicit fsync against
+	 *    this inode is made, we must make sure this fsync logs the
+	 *    inode with the new i_size, the hole between old i_size and
+	 *    the new i_size, and syncs the log.
+	 *
+	 * 3) If we are logging that an ancestor inode exists as part of
+	 *    logging a new name from a link or rename operation, don't update
+	 *    its last_log_commit - otherwise if an explicit fsync is made
+	 *    against an ancestor, the fsync considers the inode in the log
+	 *    and doesn't sync the log, resulting in the ancestor missing after
+	 *    a power failure unless the log was synced as part of an fsync
+	 *    against any other unrelated inode.
+	 */
+	if (inode_only != LOG_INODE_EXISTS)
 		inode->last_log_commit = inode->last_sub_trans;
 	spin_unlock(&inode->lock);
-out_unlock:
-	mutex_unlock(&inode->log_mutex);
-
-	btrfs_free_path(path);
-	btrfs_free_path(dst_path);
-	return err;
-}
-
-/*
- * Check if we must fallback to a transaction commit when logging an inode.
- * This must be called after logging the inode and is used only in the context
- * when fsyncing an inode requires the need to log some other inode - in which
- * case we can't lock the i_mutex of each other inode we need to log as that
- * can lead to deadlocks with concurrent fsync against other inodes (as we can
- * log inodes up or down in the hierarchy) or rename operations for example. So
- * we take the log_mutex of the inode after we have logged it and then check for
- * its last_unlink_trans value - this is safe because any task setting
- * last_unlink_trans must take the log_mutex and it must do this before it does
- * the actual unlink operation, so if we do this check before a concurrent task
- * sets last_unlink_trans it means we've logged a consistent version/state of
- * all the inode items, otherwise we are not sure and must do a transaction
- * commit (the concurrent task might have only updated last_unlink_trans before
- * we logged the inode or it might have also done the unlink).
- */
-static bool btrfs_must_commit_transaction(struct btrfs_trans_handle *trans,
-					  struct btrfs_inode *inode)
-{
-	struct btrfs_fs_info *fs_info = inode->root->fs_info;
-	bool ret = false;
-
-	mutex_lock(&inode->log_mutex);
-	if (inode->last_unlink_trans > fs_info->last_trans_committed) {
-		/*
-		 * Make sure any commits to the log are forced to be full
-		 * commits.
-		 */
-		btrfs_set_log_full_commit(trans);
-		ret = true;
-	}
-	mutex_unlock(&inode->log_mutex);
-
-	return ret;
-}
-
-/*
- * follow the dentry parent pointers up the chain and see if any
- * of the directories in it require a full commit before they can
- * be logged.  Returns zero if nothing special needs to be done or 1 if
- * a full commit is required.
- */
-static noinline int check_parent_dirs_for_sync(struct btrfs_trans_handle *trans,
-					       struct btrfs_inode *inode,
-					       struct dentry *parent,
-					       struct super_block *sb,
-					       u64 last_committed)
-{
-	int ret = 0;
-	struct dentry *old_parent = NULL;
 
 	/*
-	 * for regular files, if its inode is already on disk, we don't
-	 * have to worry about the parents at all.  This is because
-	 * we can use the last_unlink_trans field to record renames
-	 * and other fun in this file.
+	 * Reset the last_reflink_trans so that the next fsync does not need to
+	 * go through the slower path when logging extents and their checksums.
 	 */
-	if (S_ISREG(inode->vfs_inode.i_mode) &&
-	    inode->generation <= last_committed &&
-	    inode->last_unlink_trans <= last_committed)
-		goto out;
-
-	if (!S_ISDIR(inode->vfs_inode.i_mode)) {
-		if (!parent || d_really_is_negative(parent) || sb != parent->d_sb)
-			goto out;
-		inode = BTRFS_I(d_inode(parent));
-	}
-
-	while (1) {
-		if (btrfs_must_commit_transaction(trans, inode)) {
-			ret = 1;
-			break;
-		}
-
-		if (!parent || d_really_is_negative(parent) || sb != parent->d_sb)
-			break;
+	if (inode_only == LOG_INODE_ALL)
+		inode->last_reflink_trans = 0;
 
-		if (IS_ROOT(parent)) {
-			inode = BTRFS_I(d_inode(parent));
-			if (btrfs_must_commit_transaction(trans, inode))
-				ret = 1;
-			break;
-		}
-
-		parent = dget_parent(parent);
-		dput(old_parent);
-		old_parent = parent;
-		inode = BTRFS_I(d_inode(parent));
-
-	}
-	dput(old_parent);
+out_unlock:
+	mutex_unlock(&inode->log_mutex);
 out:
-	return ret;
-}
-
-struct btrfs_dir_list {
-	u64 ino;
-	struct list_head list;
-};
-
-/*
- * Log the inodes of the new dentries of a directory. See log_dir_items() for
- * details about the why it is needed.
- * This is a recursive operation - if an existing dentry corresponds to a
- * directory, that directory's new entries are logged too (same behaviour as
- * ext3/4, xfs, f2fs, reiserfs, nilfs2). Note that when logging the inodes
- * the dentries point to we do not lock their i_mutex, otherwise lockdep
- * complains about the following circular lock dependency / possible deadlock:
- *
- *        CPU0                                        CPU1
- *        ----                                        ----
- * lock(&type->i_mutex_dir_key#3/2);
- *                                            lock(sb_internal#2);
- *                                            lock(&type->i_mutex_dir_key#3/2);
- * lock(&sb->s_type->i_mutex_key#14);
- *
- * Where sb_internal is the lock (a counter that works as a lock) acquired by
- * sb_start_intwrite() in btrfs_start_transaction().
- * Not locking i_mutex of the inodes is still safe because:
- *
- * 1) For regular files we log with a mode of LOG_INODE_EXISTS. It's possible
- *    that while logging the inode new references (names) are added or removed
- *    from the inode, leaving the logged inode item with a link count that does
- *    not match the number of logged inode reference items. This is fine because
- *    at log replay time we compute the real number of links and correct the
- *    link count in the inode item (see replay_one_buffer() and
- *    link_to_fixup_dir());
- *
- * 2) For directories we log with a mode of LOG_INODE_ALL. It's possible that
- *    while logging the inode's items new items with keys BTRFS_DIR_ITEM_KEY and
- *    BTRFS_DIR_INDEX_KEY are added to fs/subvol tree and the logged inode item
- *    has a size that doesn't match the sum of the lengths of all the logged
- *    names. This does not result in a problem because if a dir_item key is
- *    logged but its matching dir_index key is not logged, at log replay time we
- *    don't use it to replay the respective name (see replay_one_name()). On the
- *    other hand if only the dir_index key ends up being logged, the respective
- *    name is added to the fs/subvol tree with both the dir_item and dir_index
- *    keys created (see replay_one_name()).
- *    The directory's inode item with a wrong i_size is not a problem as well,
- *    since we don't use it at log replay time to set the i_size in the inode
- *    item of the fs/subvol tree (see overwrite_item()).
- */
-static int log_new_dir_dentries(struct btrfs_trans_handle *trans,
-				struct btrfs_root *root,
-				struct btrfs_inode *start_inode,
-				struct btrfs_log_ctx *ctx)
-{
-	struct btrfs_fs_info *fs_info = root->fs_info;
-	struct btrfs_root *log = root->log_root;
-	struct btrfs_path *path;
-	LIST_HEAD(dir_list);
-	struct btrfs_dir_list *dir_elem;
-	int ret = 0;
-
-	path = btrfs_alloc_path();
-	if (!path)
-		return -ENOMEM;
-
-	dir_elem = kmalloc(sizeof(*dir_elem), GFP_NOFS);
-	if (!dir_elem) {
-		btrfs_free_path(path);
-		return -ENOMEM;
-	}
-	dir_elem->ino = btrfs_ino(start_inode);
-	list_add_tail(&dir_elem->list, &dir_list);
-
-	while (!list_empty(&dir_list)) {
-		struct extent_buffer *leaf;
-		struct btrfs_key min_key;
-		int nritems;
-		int i;
-
-		dir_elem = list_first_entry(&dir_list, struct btrfs_dir_list,
-					    list);
-		if (ret)
-			goto next_dir_inode;
-
-		min_key.objectid = dir_elem->ino;
-		min_key.type = BTRFS_DIR_ITEM_KEY;
-		min_key.offset = 0;
-again:
-		btrfs_release_path(path);
-		ret = btrfs_search_forward(log, &min_key, path, trans->transid);
-		if (ret < 0) {
-			goto next_dir_inode;
-		} else if (ret > 0) {
-			ret = 0;
-			goto next_dir_inode;
-		}
-
-process_leaf:
-		leaf = path->nodes[0];
-		nritems = btrfs_header_nritems(leaf);
-		for (i = path->slots[0]; i < nritems; i++) {
-			struct btrfs_dir_item *di;
-			struct btrfs_key di_key;
-			struct inode *di_inode;
-			struct btrfs_dir_list *new_dir_elem;
-			int log_mode = LOG_INODE_EXISTS;
-			int type;
-
-			btrfs_item_key_to_cpu(leaf, &min_key, i);
-			if (min_key.objectid != dir_elem->ino ||
-			    min_key.type != BTRFS_DIR_ITEM_KEY)
-				goto next_dir_inode;
-
-			di = btrfs_item_ptr(leaf, i, struct btrfs_dir_item);
-			type = btrfs_dir_type(leaf, di);
-			if (btrfs_dir_transid(leaf, di) < trans->transid &&
-			    type != BTRFS_FT_DIR)
-				continue;
-			btrfs_dir_item_key_to_cpu(leaf, di, &di_key);
-			if (di_key.type == BTRFS_ROOT_ITEM_KEY)
-				continue;
+	btrfs_free_path(path);
+	btrfs_free_path(dst_path);
 
-			btrfs_release_path(path);
-			di_inode = btrfs_iget(fs_info->sb, &di_key, root);
-			if (IS_ERR(di_inode)) {
-				ret = PTR_ERR(di_inode);
-				goto next_dir_inode;
-			}
+	if (ret)
+		free_conflicting_inodes(ctx);
+	else
+		ret = log_conflicting_inodes(trans, inode->root, ctx);
 
-			if (btrfs_inode_in_log(BTRFS_I(di_inode), trans->transid)) {
-				btrfs_add_delayed_iput(di_inode);
-				break;
-			}
+	if (full_dir_logging && !ctx->logging_new_delayed_dentries) {
+		if (!ret)
+			ret = log_new_delayed_dentries(trans, inode,
+						       &delayed_ins_list, ctx);
 
-			ctx->log_new_dentries = false;
-			if (type == BTRFS_FT_DIR || type == BTRFS_FT_SYMLINK)
-				log_mode = LOG_INODE_ALL;
-			ret = btrfs_log_inode(trans, root, BTRFS_I(di_inode),
-					      log_mode, 0, LLONG_MAX, ctx);
-			if (!ret &&
-			    btrfs_must_commit_transaction(trans, BTRFS_I(di_inode)))
-				ret = 1;
-			btrfs_add_delayed_iput(di_inode);
-			if (ret)
-				goto next_dir_inode;
-			if (ctx->log_new_dentries) {
-				new_dir_elem = kmalloc(sizeof(*new_dir_elem),
-						       GFP_NOFS);
-				if (!new_dir_elem) {
-					ret = -ENOMEM;
-					goto next_dir_inode;
-				}
-				new_dir_elem->ino = di_key.objectid;
-				list_add_tail(&new_dir_elem->list, &dir_list);
-			}
-			break;
-		}
-		if (i == nritems) {
-			ret = btrfs_next_leaf(log, path);
-			if (ret < 0) {
-				goto next_dir_inode;
-			} else if (ret > 0) {
-				ret = 0;
-				goto next_dir_inode;
-			}
-			goto process_leaf;
-		}
-		if (min_key.offset < (u64)-1) {
-			min_key.offset++;
-			goto again;
-		}
-next_dir_inode:
-		list_del(&dir_elem->list);
-		kfree(dir_elem);
+		btrfs_log_put_delayed_items(inode, &delayed_ins_list,
+					    &delayed_del_list);
 	}
 
-	btrfs_free_path(path);
 	return ret;
 }
 
@@ -5648,7 +6677,7 @@ static int btrfs_log_all_parents(struct btrfs_trans_handle *trans,
 		if (key.objectid != ino || key.type > BTRFS_INODE_EXTREF_KEY)
 			break;
 
-		item_size = btrfs_item_size_nr(leaf, slot);
+		item_size = btrfs_item_size(leaf, slot);
 		ptr = btrfs_item_ptr_offset(leaf, slot);
 		while (cur_offset < item_size) {
 			struct btrfs_key inode_key;
@@ -5672,7 +6701,8 @@ static int btrfs_log_all_parents(struct btrfs_trans_handle *trans,
 				cur_offset = item_size;
 			}
 
-			dir_inode = btrfs_iget(fs_info->sb, &inode_key, root);
+			dir_inode = btrfs_iget(fs_info->sb, inode_key.objectid,
+					       root);
 			/*
 			 * If the parent inode was deleted, return an error to
 			 * fallback to a transaction commit. This is to prevent
@@ -5701,15 +6731,16 @@ static int btrfs_log_all_parents(struct btrfs_trans_handle *trans,
 				goto out;
 			}
 
-			if (ctx)
-				ctx->log_new_dentries = false;
-			ret = btrfs_log_inode(trans, root, BTRFS_I(dir_inode),
-					      LOG_INODE_ALL, 0, LLONG_MAX, ctx);
-			if (!ret &&
-			    btrfs_must_commit_transaction(trans, BTRFS_I(dir_inode)))
-				ret = 1;
-			if (!ret && ctx && ctx->log_new_dentries)
-				ret = log_new_dir_dentries(trans, root,
+			if (!need_log_inode(trans, BTRFS_I(dir_inode))) {
+				btrfs_add_delayed_iput(dir_inode);
+				continue;
+			}
+
+			ctx->log_new_dentries = false;
+			ret = btrfs_log_inode(trans, BTRFS_I(dir_inode),
+					      LOG_INODE_ALL, ctx);
+			if (!ret && ctx->log_new_dentries)
+				ret = log_new_dir_dentries(trans,
 						   BTRFS_I(dir_inode), ctx);
 			btrfs_add_delayed_iput(dir_inode);
 			if (ret)
@@ -5734,26 +6765,28 @@ static int log_new_ancestors(struct btrfs_trans_handle *trans,
 
 	while (true) {
 		struct btrfs_fs_info *fs_info = root->fs_info;
-		const u64 last_committed = fs_info->last_trans_committed;
 		struct extent_buffer *leaf = path->nodes[0];
 		int slot = path->slots[0];
 		struct btrfs_key search_key;
 		struct inode *inode;
+		u64 ino;
 		int ret = 0;
 
 		btrfs_release_path(path);
 
+		ino = found_key.offset;
+
 		search_key.objectid = found_key.offset;
 		search_key.type = BTRFS_INODE_ITEM_KEY;
 		search_key.offset = 0;
-		inode = btrfs_iget(fs_info->sb, &search_key, root);
+		inode = btrfs_iget(fs_info->sb, ino, root);
 		if (IS_ERR(inode))
 			return PTR_ERR(inode);
 
-		if (BTRFS_I(inode)->generation > last_committed)
-			ret = btrfs_log_inode(trans, root, BTRFS_I(inode),
-					      LOG_INODE_EXISTS,
-					      0, LLONG_MAX, ctx);
+		if (BTRFS_I(inode)->generation >= trans->transid &&
+		    need_log_inode(trans, BTRFS_I(inode)))
+			ret = btrfs_log_inode(trans, BTRFS_I(inode),
+					      LOG_INODE_EXISTS, ctx);
 		btrfs_add_delayed_iput(inode);
 		if (ret)
 			return ret;
@@ -5792,7 +6825,6 @@ static int log_new_ancestors_fast(struct btrfs_trans_handle *trans,
 				  struct btrfs_log_ctx *ctx)
 {
 	struct btrfs_root *root = inode->root;
-	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct dentry *old_parent = NULL;
 	struct super_block *sb = inode->vfs_inode.i_sb;
 	int ret = 0;
@@ -5806,9 +6838,10 @@ static int log_new_ancestors_fast(struct btrfs_trans_handle *trans,
 		if (root != inode->root)
 			break;
 
-		if (inode->generation > fs_info->last_trans_committed) {
-			ret = btrfs_log_inode(trans, root, inode,
-					LOG_INODE_EXISTS, 0, LLONG_MAX, ctx);
+		if (inode->generation >= trans->transid &&
+		    need_log_inode(trans, inode)) {
+			ret = btrfs_log_inode(trans, inode,
+					      LOG_INODE_EXISTS, ctx);
 			if (ret)
 				break;
 		}
@@ -5916,51 +6949,31 @@ out:
 static int btrfs_log_inode_parent(struct btrfs_trans_handle *trans,
 				  struct btrfs_inode *inode,
 				  struct dentry *parent,
-				  const loff_t start,
-				  const loff_t end,
 				  int inode_only,
 				  struct btrfs_log_ctx *ctx)
 {
 	struct btrfs_root *root = inode->root;
 	struct btrfs_fs_info *fs_info = root->fs_info;
-	struct super_block *sb;
 	int ret = 0;
-	u64 last_committed = fs_info->last_trans_committed;
 	bool log_dentries = false;
 
-	sb = inode->vfs_inode.i_sb;
-
 	if (btrfs_test_opt(fs_info, NOTREELOG)) {
-		ret = 1;
-		goto end_no_trans;
-	}
-
-	/*
-	 * The prev transaction commit doesn't complete, we need do
-	 * full commit by ourselves.
-	 */
-	if (fs_info->last_trans_log_full_commit >
-	    fs_info->last_trans_committed) {
-		ret = 1;
+		ret = BTRFS_LOG_FORCE_COMMIT;
 		goto end_no_trans;
 	}
 
 	if (btrfs_root_refs(&root->root_item) == 0) {
-		ret = 1;
+		ret = BTRFS_LOG_FORCE_COMMIT;
 		goto end_no_trans;
 	}
 
-	ret = check_parent_dirs_for_sync(trans, inode, parent, sb,
-			last_committed);
-	if (ret)
-		goto end_no_trans;
-
 	/*
 	 * Skip already logged inodes or inodes corresponding to tmpfiles
 	 * (since logging them is pointless, a link count of 0 means they
 	 * will never be accessible).
 	 */
-	if (btrfs_inode_in_log(inode, trans->transid) ||
+	if ((btrfs_inode_in_log(inode, trans->transid) &&
+	     list_empty(&ctx->ordered_extents)) ||
 	    inode->vfs_inode.i_nlink == 0) {
 		ret = BTRFS_NO_LOG_SYNC;
 		goto end_no_trans;
@@ -5970,7 +6983,7 @@ static int btrfs_log_inode_parent(struct btrfs_trans_handle *trans,
 	if (ret)
 		goto end_no_trans;
 
-	ret = btrfs_log_inode(trans, root, inode, inode_only, start, end, ctx);
+	ret = btrfs_log_inode(trans, inode, inode_only, ctx);
 	if (ret)
 		goto end_trans;
 
@@ -5981,13 +6994,13 @@ static int btrfs_log_inode_parent(struct btrfs_trans_handle *trans,
 	 * and other fun in this file.
 	 */
 	if (S_ISREG(inode->vfs_inode.i_mode) &&
-	    inode->generation <= last_committed &&
-	    inode->last_unlink_trans <= last_committed) {
+	    inode->generation < trans->transid &&
+	    inode->last_unlink_trans < trans->transid) {
 		ret = 0;
 		goto end_trans;
 	}
 
-	if (S_ISDIR(inode->vfs_inode.i_mode) && ctx && ctx->log_new_dentries)
+	if (S_ISDIR(inode->vfs_inode.i_mode) && ctx->log_new_dentries)
 		log_dentries = true;
 
 	/*
@@ -6031,7 +7044,7 @@ static int btrfs_log_inode_parent(struct btrfs_trans_handle *trans,
 	 * but the file inode does not have a matching BTRFS_INODE_REF_KEY item
 	 * and has a link count of 2.
 	 */
-	if (inode->last_unlink_trans > last_committed) {
+	if (inode->last_unlink_trans >= trans->transid) {
 		ret = btrfs_log_all_parents(trans, inode, ctx);
 		if (ret)
 			goto end_trans;
@@ -6042,13 +7055,13 @@ static int btrfs_log_inode_parent(struct btrfs_trans_handle *trans,
 		goto end_trans;
 
 	if (log_dentries)
-		ret = log_new_dir_dentries(trans, root, inode, ctx);
+		ret = log_new_dir_dentries(trans, inode, ctx);
 	else
 		ret = 0;
 end_trans:
 	if (ret < 0) {
 		btrfs_set_log_full_commit(trans);
-		ret = 1;
+		ret = BTRFS_LOG_FORCE_COMMIT;
 	}
 
 	if (ret)
@@ -6066,15 +7079,13 @@ end_no_trans:
  */
 int btrfs_log_dentry_safe(struct btrfs_trans_handle *trans,
 			  struct dentry *dentry,
-			  const loff_t start,
-			  const loff_t end,
 			  struct btrfs_log_ctx *ctx)
 {
 	struct dentry *parent = dget_parent(dentry);
 	int ret;
 
 	ret = btrfs_log_inode_parent(trans, BTRFS_I(d_inode(dentry)), parent,
-				     start, end, LOG_INODE_ALL, ctx);
+				     LOG_INODE_ALL, ctx);
 	dput(parent);
 
 	return ret;
@@ -6091,7 +7102,6 @@ int btrfs_recover_log_trees(struct btrfs_root *log_root_tree)
 	struct btrfs_trans_handle *trans;
 	struct btrfs_key key;
 	struct btrfs_key found_key;
-	struct btrfs_key tmp_key;
 	struct btrfs_root *log;
 	struct btrfs_fs_info *fs_info = log_root_tree->fs_info;
 	struct walk_control wc = {
@@ -6116,8 +7126,7 @@ int btrfs_recover_log_trees(struct btrfs_root *log_root_tree)
 
 	ret = walk_log_tree(trans, log_root_tree, &wc);
 	if (ret) {
-		btrfs_handle_fs_error(fs_info, ret,
-			"Failed to pin buffers while recovering log root tree.");
+		btrfs_abort_transaction(trans, ret);
 		goto error;
 	}
 
@@ -6130,8 +7139,7 @@ again:
 		ret = btrfs_search_slot(NULL, log_root_tree, &key, path, 0, 0);
 
 		if (ret < 0) {
-			btrfs_handle_fs_error(fs_info, ret,
-				    "Couldn't find tree log root.");
+			btrfs_abort_transaction(trans, ret);
 			goto error;
 		}
 		if (ret > 0) {
@@ -6145,19 +7153,15 @@ again:
 		if (found_key.objectid != BTRFS_TREE_LOG_OBJECTID)
 			break;
 
-		log = btrfs_read_fs_root(log_root_tree, &found_key);
+		log = btrfs_read_tree_root(log_root_tree, &found_key);
 		if (IS_ERR(log)) {
 			ret = PTR_ERR(log);
-			btrfs_handle_fs_error(fs_info, ret,
-				    "Couldn't read tree log root.");
+			btrfs_abort_transaction(trans, ret);
 			goto error;
 		}
 
-		tmp_key.objectid = found_key.offset;
-		tmp_key.type = BTRFS_ROOT_ITEM_KEY;
-		tmp_key.offset = (u64)-1;
-
-		wc.replay_dest = btrfs_read_fs_root_no_name(fs_info, &tmp_key);
+		wc.replay_dest = btrfs_get_fs_root(fs_info, found_key.offset,
+						   true);
 		if (IS_ERR(wc.replay_dest)) {
 			ret = PTR_ERR(wc.replay_dest);
 
@@ -6173,27 +7177,30 @@ again:
 			 * each subsequent pass.
 			 */
 			if (ret == -ENOENT)
-				ret = btrfs_pin_extent_for_log_replay(fs_info,
+				ret = btrfs_pin_extent_for_log_replay(trans,
 							log->node->start,
 							log->node->len);
-			free_extent_buffer(log->node);
-			free_extent_buffer(log->commit_root);
-			kfree(log);
+			btrfs_put_root(log);
 
 			if (!ret)
 				goto next;
-			btrfs_handle_fs_error(fs_info, ret,
-				"Couldn't read target root for tree log recovery.");
+			btrfs_abort_transaction(trans, ret);
 			goto error;
 		}
 
 		wc.replay_dest->log_root = log;
-		btrfs_record_root_in_trans(trans, wc.replay_dest);
-		ret = walk_log_tree(trans, log, &wc);
+		ret = btrfs_record_root_in_trans(trans, wc.replay_dest);
+		if (ret)
+			/* The loop needs to continue due to the root refs */
+			btrfs_abort_transaction(trans, ret);
+		else
+			ret = walk_log_tree(trans, log, &wc);
 
 		if (!ret && wc.stage == LOG_WALK_REPLAY_ALL) {
 			ret = fixup_inode_link_counts(trans, wc.replay_dest,
 						      path);
+			if (ret)
+				btrfs_abort_transaction(trans, ret);
 		}
 
 		if (!ret && wc.stage == LOG_WALK_REPLAY_ALL) {
@@ -6209,14 +7216,14 @@ again:
 			 * root->objectid_mutex is not acquired as log replay
 			 * could only happen during mount.
 			 */
-			ret = btrfs_find_highest_objectid(root,
-						  &root->highest_objectid);
+			ret = btrfs_init_root_free_objectid(root);
+			if (ret)
+				btrfs_abort_transaction(trans, ret);
 		}
 
 		wc.replay_dest->log_root = NULL;
-		free_extent_buffer(log->node);
-		free_extent_buffer(log->commit_root);
-		kfree(log);
+		btrfs_put_root(wc.replay_dest);
+		btrfs_put_root(log);
 
 		if (ret)
 			goto error;
@@ -6247,15 +7254,15 @@ next:
 	if (ret)
 		return ret;
 
-	free_extent_buffer(log_root_tree->node);
 	log_root_tree->log_root = NULL;
 	clear_bit(BTRFS_FS_LOG_RECOVERING, &fs_info->flags);
-	kfree(log_root_tree);
+	btrfs_put_root(log_root_tree);
 
 	return 0;
 error:
 	if (wc.trans)
 		btrfs_end_transaction(wc.trans);
+	clear_bit(BTRFS_FS_LOG_RECOVERING, &fs_info->flags);
 	btrfs_free_path(path);
 	return ret;
 }
@@ -6342,28 +7349,31 @@ void btrfs_record_snapshot_destroy(struct btrfs_trans_handle *trans,
 	mutex_unlock(&dir->log_mutex);
 }
 
-/*
- * Call this after adding a new name for a file and it will properly
- * update the log to reflect the new name.
+/**
+ * Update the log after adding a new name for an inode.
  *
- * @ctx can not be NULL when @sync_log is false, and should be NULL when it's
- * true (because it's not used).
+ * @trans:              Transaction handle.
+ * @old_dentry:         The dentry associated with the old name and the old
+ *                      parent directory.
+ * @old_dir:            The inode of the previous parent directory for the case
+ *                      of a rename. For a link operation, it must be NULL.
+ * @old_dir_index:      The index number associated with the old name, meaningful
+ *                      only for rename operations (when @old_dir is not NULL).
+ *                      Ignored for link operations.
+ * @parent:             The dentry associated with the directory under which the
+ *                      new name is located.
  *
- * Return value depends on whether @sync_log is true or false.
- * When true: returns BTRFS_NEED_TRANS_COMMIT if the transaction needs to be
- *            committed by the caller, and BTRFS_DONT_NEED_TRANS_COMMIT
- *            otherwise.
- * When false: returns BTRFS_DONT_NEED_LOG_SYNC if the caller does not need to
- *             to sync the log, BTRFS_NEED_LOG_SYNC if it needs to sync the log,
- *             or BTRFS_NEED_TRANS_COMMIT if the transaction needs to be
- *             committed (without attempting to sync the log).
+ * Call this after adding a new name for an inode, as a result of a link or
+ * rename operation, and it will properly update the log to reflect the new name.
  */
-int btrfs_log_new_name(struct btrfs_trans_handle *trans,
-			struct btrfs_inode *inode, struct btrfs_inode *old_dir,
-			struct dentry *parent,
-			bool sync_log, struct btrfs_log_ctx *ctx)
+void btrfs_log_new_name(struct btrfs_trans_handle *trans,
+			struct dentry *old_dentry, struct btrfs_inode *old_dir,
+			u64 old_dir_index, struct dentry *parent)
 {
-	struct btrfs_fs_info *fs_info = trans->fs_info;
+	struct btrfs_inode *inode = BTRFS_I(d_inode(old_dentry));
+	struct btrfs_root *root = inode->root;
+	struct btrfs_log_ctx ctx;
+	bool log_pinned = false;
 	int ret;
 
 	/*
@@ -6377,36 +7387,112 @@ int btrfs_log_new_name(struct btrfs_trans_handle *trans,
 	 * if this inode hasn't been logged and directory we're renaming it
 	 * from hasn't been logged, we don't need to log it
 	 */
-	if (inode->logged_trans <= fs_info->last_trans_committed &&
-	    (!old_dir || old_dir->logged_trans <= fs_info->last_trans_committed))
-		return sync_log ? BTRFS_DONT_NEED_TRANS_COMMIT :
-			BTRFS_DONT_NEED_LOG_SYNC;
-
-	if (sync_log) {
-		struct btrfs_log_ctx ctx2;
-
-		btrfs_init_log_ctx(&ctx2, &inode->vfs_inode);
-		ret = btrfs_log_inode_parent(trans, inode, parent, 0, LLONG_MAX,
-					     LOG_INODE_EXISTS, &ctx2);
-		if (ret == BTRFS_NO_LOG_SYNC)
-			return BTRFS_DONT_NEED_TRANS_COMMIT;
-		else if (ret)
-			return BTRFS_NEED_TRANS_COMMIT;
-
-		ret = btrfs_sync_log(trans, inode->root, &ctx2);
-		if (ret)
-			return BTRFS_NEED_TRANS_COMMIT;
-		return BTRFS_DONT_NEED_TRANS_COMMIT;
+	ret = inode_logged(trans, inode, NULL);
+	if (ret < 0) {
+		goto out;
+	} else if (ret == 0) {
+		if (!old_dir)
+			return;
+		/*
+		 * If the inode was not logged and we are doing a rename (old_dir is not
+		 * NULL), check if old_dir was logged - if it was not we can return and
+		 * do nothing.
+		 */
+		ret = inode_logged(trans, old_dir, NULL);
+		if (ret < 0)
+			goto out;
+		else if (ret == 0)
+			return;
 	}
+	ret = 0;
+
+	/*
+	 * If we are doing a rename (old_dir is not NULL) from a directory that
+	 * was previously logged, make sure that on log replay we get the old
+	 * dir entry deleted. This is needed because we will also log the new
+	 * name of the renamed inode, so we need to make sure that after log
+	 * replay we don't end up with both the new and old dir entries existing.
+	 */
+	if (old_dir && old_dir->logged_trans == trans->transid) {
+		struct btrfs_root *log = old_dir->root->log_root;
+		struct btrfs_path *path;
+
+		ASSERT(old_dir_index >= BTRFS_DIR_START_INDEX);
+
+		/*
+		 * We have two inodes to update in the log, the old directory and
+		 * the inode that got renamed, so we must pin the log to prevent
+		 * anyone from syncing the log until we have updated both inodes
+		 * in the log.
+		 */
+		ret = join_running_log_trans(root);
+		/*
+		 * At least one of the inodes was logged before, so this should
+		 * not fail, but if it does, it's not serious, just bail out and
+		 * mark the log for a full commit.
+		 */
+		if (WARN_ON_ONCE(ret < 0))
+			goto out;
+		log_pinned = true;
 
-	ASSERT(ctx);
-	ret = btrfs_log_inode_parent(trans, inode, parent, 0, LLONG_MAX,
-				     LOG_INODE_EXISTS, ctx);
-	if (ret == BTRFS_NO_LOG_SYNC)
-		return BTRFS_DONT_NEED_LOG_SYNC;
-	else if (ret)
-		return BTRFS_NEED_TRANS_COMMIT;
+		path = btrfs_alloc_path();
+		if (!path) {
+			ret = -ENOMEM;
+			goto out;
+		}
 
-	return BTRFS_NEED_LOG_SYNC;
+		/*
+		 * Other concurrent task might be logging the old directory,
+		 * as it can be triggered when logging other inode that had or
+		 * still has a dentry in the old directory. We lock the old
+		 * directory's log_mutex to ensure the deletion of the old
+		 * name is persisted, because during directory logging we
+		 * delete all BTRFS_DIR_LOG_INDEX_KEY keys and the deletion of
+		 * the old name's dir index item is in the delayed items, so
+		 * it could be missed by an in progress directory logging.
+		 */
+		mutex_lock(&old_dir->log_mutex);
+		ret = del_logged_dentry(trans, log, path, btrfs_ino(old_dir),
+					old_dentry->d_name.name,
+					old_dentry->d_name.len, old_dir_index);
+		if (ret > 0) {
+			/*
+			 * The dentry does not exist in the log, so record its
+			 * deletion.
+			 */
+			btrfs_release_path(path);
+			ret = insert_dir_log_key(trans, log, path,
+						 btrfs_ino(old_dir),
+						 old_dir_index, old_dir_index);
+		}
+		mutex_unlock(&old_dir->log_mutex);
+
+		btrfs_free_path(path);
+		if (ret < 0)
+			goto out;
+	}
+
+	btrfs_init_log_ctx(&ctx, &inode->vfs_inode);
+	ctx.logging_new_name = true;
+	/*
+	 * We don't care about the return value. If we fail to log the new name
+	 * then we know the next attempt to sync the log will fallback to a full
+	 * transaction commit (due to a call to btrfs_set_log_full_commit()), so
+	 * we don't need to worry about getting a log committed that has an
+	 * inconsistent state after a rename operation.
+	 */
+	btrfs_log_inode_parent(trans, inode, parent, LOG_INODE_EXISTS, &ctx);
+	ASSERT(list_empty(&ctx.conflict_inodes));
+out:
+	/*
+	 * If an error happened mark the log for a full commit because it's not
+	 * consistent and up to date or we couldn't find out if one of the
+	 * inodes was logged before in this transaction. Do it before unpinning
+	 * the log, to avoid any races with someone else trying to commit it.
+	 */
+	if (ret < 0)
+		btrfs_set_log_full_commit(trans);
+	if (log_pinned)
+		btrfs_end_log_trans(root);
 }
 
diff --git a/fs/btrfs/tree-log.h b/fs/btrfs/tree-log.h
index 132e43d29034..aed1e05e9879 100644
--- a/fs/btrfs/tree-log.h
+++ b/fs/btrfs/tree-log.h
@@ -12,12 +12,26 @@
 /* return value for btrfs_log_dentry_safe that means we don't need to log it at all */
 #define BTRFS_NO_LOG_SYNC 256
 
+/* We can't use the tree log for whatever reason, force a transaction commit */
+#define BTRFS_LOG_FORCE_COMMIT				(1)
+
 struct btrfs_log_ctx {
 	int log_ret;
 	int log_transid;
 	bool log_new_dentries;
+	bool logging_new_name;
+	bool logging_new_delayed_dentries;
+	/* Indicate if the inode being logged was logged before. */
+	bool logged_before;
+	/* Tracks the last logged dir item/index key offset. */
+	u64 last_dir_item_offset;
 	struct inode *inode;
 	struct list_head list;
+	/* Only used for fast fsyncs. */
+	struct list_head ordered_extents;
+	struct list_head conflict_inodes;
+	int num_conflict_inodes;
+	bool logging_conflict_inodes;
 };
 
 static inline void btrfs_init_log_ctx(struct btrfs_log_ctx *ctx,
@@ -26,8 +40,28 @@ static inline void btrfs_init_log_ctx(struct btrfs_log_ctx *ctx,
 	ctx->log_ret = 0;
 	ctx->log_transid = 0;
 	ctx->log_new_dentries = false;
+	ctx->logging_new_name = false;
+	ctx->logging_new_delayed_dentries = false;
+	ctx->logged_before = false;
 	ctx->inode = inode;
 	INIT_LIST_HEAD(&ctx->list);
+	INIT_LIST_HEAD(&ctx->ordered_extents);
+	INIT_LIST_HEAD(&ctx->conflict_inodes);
+	ctx->num_conflict_inodes = 0;
+	ctx->logging_conflict_inodes = false;
+}
+
+static inline void btrfs_release_log_ctx_extents(struct btrfs_log_ctx *ctx)
+{
+	struct btrfs_ordered_extent *ordered;
+	struct btrfs_ordered_extent *tmp;
+
+	ASSERT(inode_is_locked(ctx->inode));
+
+	list_for_each_entry_safe(ordered, tmp, &ctx->ordered_extents, log_list) {
+		list_del_init(&ordered->log_list);
+		btrfs_put_ordered_extent(ordered);
+	}
 }
 
 static inline void btrfs_set_log_full_commit(struct btrfs_trans_handle *trans)
@@ -49,17 +83,15 @@ int btrfs_free_log_root_tree(struct btrfs_trans_handle *trans,
 int btrfs_recover_log_trees(struct btrfs_root *tree_root);
 int btrfs_log_dentry_safe(struct btrfs_trans_handle *trans,
 			  struct dentry *dentry,
-			  const loff_t start,
-			  const loff_t end,
 			  struct btrfs_log_ctx *ctx);
-int btrfs_del_dir_entries_in_log(struct btrfs_trans_handle *trans,
-				 struct btrfs_root *root,
-				 const char *name, int name_len,
-				 struct btrfs_inode *dir, u64 index);
-int btrfs_del_inode_ref_in_log(struct btrfs_trans_handle *trans,
-			       struct btrfs_root *root,
-			       const char *name, int name_len,
-			       struct btrfs_inode *inode, u64 dirid);
+void btrfs_del_dir_entries_in_log(struct btrfs_trans_handle *trans,
+				  struct btrfs_root *root,
+				  const char *name, int name_len,
+				  struct btrfs_inode *dir, u64 index);
+void btrfs_del_inode_ref_in_log(struct btrfs_trans_handle *trans,
+				struct btrfs_root *root,
+				const char *name, int name_len,
+				struct btrfs_inode *inode, u64 dirid);
 void btrfs_end_log_trans(struct btrfs_root *root);
 void btrfs_pin_log_trans(struct btrfs_root *root);
 void btrfs_record_unlink_dir(struct btrfs_trans_handle *trans,
@@ -67,16 +99,8 @@ void btrfs_record_unlink_dir(struct btrfs_trans_handle *trans,
 			     int for_rename);
 void btrfs_record_snapshot_destroy(struct btrfs_trans_handle *trans,
 				   struct btrfs_inode *dir);
-/* Return values for btrfs_log_new_name() */
-enum {
-	BTRFS_DONT_NEED_TRANS_COMMIT,
-	BTRFS_NEED_TRANS_COMMIT,
-	BTRFS_DONT_NEED_LOG_SYNC,
-	BTRFS_NEED_LOG_SYNC,
-};
-int btrfs_log_new_name(struct btrfs_trans_handle *trans,
-			struct btrfs_inode *inode, struct btrfs_inode *old_dir,
-			struct dentry *parent,
-			bool sync_log, struct btrfs_log_ctx *ctx);
+void btrfs_log_new_name(struct btrfs_trans_handle *trans,
+			struct dentry *old_dentry, struct btrfs_inode *old_dir,
+			u64 old_dir_index, struct dentry *parent);
 
 #endif
diff --git a/fs/btrfs/tree-mod-log.c b/fs/btrfs/tree-mod-log.c
new file mode 100644
index 000000000000..8a3a14686d3e
--- /dev/null
+++ b/fs/btrfs/tree-mod-log.c
@@ -0,0 +1,929 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include "tree-mod-log.h"
+#include "disk-io.h"
+
+struct tree_mod_root {
+	u64 logical;
+	u8 level;
+};
+
+struct tree_mod_elem {
+	struct rb_node node;
+	u64 logical;
+	u64 seq;
+	enum btrfs_mod_log_op op;
+
+	/*
+	 * This is used for BTRFS_MOD_LOG_KEY_* and BTRFS_MOD_LOG_MOVE_KEYS
+	 * operations.
+	 */
+	int slot;
+
+	/* This is used for BTRFS_MOD_LOG_KEY* and BTRFS_MOD_LOG_ROOT_REPLACE. */
+	u64 generation;
+
+	/* Those are used for op == BTRFS_MOD_LOG_KEY_{REPLACE,REMOVE}. */
+	struct btrfs_disk_key key;
+	u64 blockptr;
+
+	/* This is used for op == BTRFS_MOD_LOG_MOVE_KEYS. */
+	struct {
+		int dst_slot;
+		int nr_items;
+	} move;
+
+	/* This is used for op == BTRFS_MOD_LOG_ROOT_REPLACE. */
+	struct tree_mod_root old_root;
+};
+
+/*
+ * Pull a new tree mod seq number for our operation.
+ */
+static inline u64 btrfs_inc_tree_mod_seq(struct btrfs_fs_info *fs_info)
+{
+	return atomic64_inc_return(&fs_info->tree_mod_seq);
+}
+
+/*
+ * This adds a new blocker to the tree mod log's blocker list if the @elem
+ * passed does not already have a sequence number set. So when a caller expects
+ * to record tree modifications, it should ensure to set elem->seq to zero
+ * before calling btrfs_get_tree_mod_seq.
+ * Returns a fresh, unused tree log modification sequence number, even if no new
+ * blocker was added.
+ */
+u64 btrfs_get_tree_mod_seq(struct btrfs_fs_info *fs_info,
+			   struct btrfs_seq_list *elem)
+{
+	write_lock(&fs_info->tree_mod_log_lock);
+	if (!elem->seq) {
+		elem->seq = btrfs_inc_tree_mod_seq(fs_info);
+		list_add_tail(&elem->list, &fs_info->tree_mod_seq_list);
+		set_bit(BTRFS_FS_TREE_MOD_LOG_USERS, &fs_info->flags);
+	}
+	write_unlock(&fs_info->tree_mod_log_lock);
+
+	return elem->seq;
+}
+
+void btrfs_put_tree_mod_seq(struct btrfs_fs_info *fs_info,
+			    struct btrfs_seq_list *elem)
+{
+	struct rb_root *tm_root;
+	struct rb_node *node;
+	struct rb_node *next;
+	struct tree_mod_elem *tm;
+	u64 min_seq = BTRFS_SEQ_LAST;
+	u64 seq_putting = elem->seq;
+
+	if (!seq_putting)
+		return;
+
+	write_lock(&fs_info->tree_mod_log_lock);
+	list_del(&elem->list);
+	elem->seq = 0;
+
+	if (list_empty(&fs_info->tree_mod_seq_list)) {
+		clear_bit(BTRFS_FS_TREE_MOD_LOG_USERS, &fs_info->flags);
+	} else {
+		struct btrfs_seq_list *first;
+
+		first = list_first_entry(&fs_info->tree_mod_seq_list,
+					 struct btrfs_seq_list, list);
+		if (seq_putting > first->seq) {
+			/*
+			 * Blocker with lower sequence number exists, we cannot
+			 * remove anything from the log.
+			 */
+			write_unlock(&fs_info->tree_mod_log_lock);
+			return;
+		}
+		min_seq = first->seq;
+	}
+
+	/*
+	 * Anything that's lower than the lowest existing (read: blocked)
+	 * sequence number can be removed from the tree.
+	 */
+	tm_root = &fs_info->tree_mod_log;
+	for (node = rb_first(tm_root); node; node = next) {
+		next = rb_next(node);
+		tm = rb_entry(node, struct tree_mod_elem, node);
+		if (tm->seq >= min_seq)
+			continue;
+		rb_erase(node, tm_root);
+		kfree(tm);
+	}
+	write_unlock(&fs_info->tree_mod_log_lock);
+}
+
+/*
+ * Key order of the log:
+ *       node/leaf start address -> sequence
+ *
+ * The 'start address' is the logical address of the *new* root node for root
+ * replace operations, or the logical address of the affected block for all
+ * other operations.
+ */
+static noinline int tree_mod_log_insert(struct btrfs_fs_info *fs_info,
+					struct tree_mod_elem *tm)
+{
+	struct rb_root *tm_root;
+	struct rb_node **new;
+	struct rb_node *parent = NULL;
+	struct tree_mod_elem *cur;
+
+	lockdep_assert_held_write(&fs_info->tree_mod_log_lock);
+
+	tm->seq = btrfs_inc_tree_mod_seq(fs_info);
+
+	tm_root = &fs_info->tree_mod_log;
+	new = &tm_root->rb_node;
+	while (*new) {
+		cur = rb_entry(*new, struct tree_mod_elem, node);
+		parent = *new;
+		if (cur->logical < tm->logical)
+			new = &((*new)->rb_left);
+		else if (cur->logical > tm->logical)
+			new = &((*new)->rb_right);
+		else if (cur->seq < tm->seq)
+			new = &((*new)->rb_left);
+		else if (cur->seq > tm->seq)
+			new = &((*new)->rb_right);
+		else
+			return -EEXIST;
+	}
+
+	rb_link_node(&tm->node, parent, new);
+	rb_insert_color(&tm->node, tm_root);
+	return 0;
+}
+
+/*
+ * Determines if logging can be omitted. Returns true if it can. Otherwise, it
+ * returns false with the tree_mod_log_lock acquired. The caller must hold
+ * this until all tree mod log insertions are recorded in the rb tree and then
+ * write unlock fs_info::tree_mod_log_lock.
+ */
+static inline bool tree_mod_dont_log(struct btrfs_fs_info *fs_info,
+				    struct extent_buffer *eb)
+{
+	if (!test_bit(BTRFS_FS_TREE_MOD_LOG_USERS, &fs_info->flags))
+		return true;
+	if (eb && btrfs_header_level(eb) == 0)
+		return true;
+
+	write_lock(&fs_info->tree_mod_log_lock);
+	if (list_empty(&(fs_info)->tree_mod_seq_list)) {
+		write_unlock(&fs_info->tree_mod_log_lock);
+		return true;
+	}
+
+	return false;
+}
+
+/* Similar to tree_mod_dont_log, but doesn't acquire any locks. */
+static inline bool tree_mod_need_log(const struct btrfs_fs_info *fs_info,
+				    struct extent_buffer *eb)
+{
+	if (!test_bit(BTRFS_FS_TREE_MOD_LOG_USERS, &fs_info->flags))
+		return false;
+	if (eb && btrfs_header_level(eb) == 0)
+		return false;
+
+	return true;
+}
+
+static struct tree_mod_elem *alloc_tree_mod_elem(struct extent_buffer *eb,
+						 int slot,
+						 enum btrfs_mod_log_op op,
+						 gfp_t flags)
+{
+	struct tree_mod_elem *tm;
+
+	tm = kzalloc(sizeof(*tm), flags);
+	if (!tm)
+		return NULL;
+
+	tm->logical = eb->start;
+	if (op != BTRFS_MOD_LOG_KEY_ADD) {
+		btrfs_node_key(eb, &tm->key, slot);
+		tm->blockptr = btrfs_node_blockptr(eb, slot);
+	}
+	tm->op = op;
+	tm->slot = slot;
+	tm->generation = btrfs_node_ptr_generation(eb, slot);
+	RB_CLEAR_NODE(&tm->node);
+
+	return tm;
+}
+
+int btrfs_tree_mod_log_insert_key(struct extent_buffer *eb, int slot,
+				  enum btrfs_mod_log_op op, gfp_t flags)
+{
+	struct tree_mod_elem *tm;
+	int ret;
+
+	if (!tree_mod_need_log(eb->fs_info, eb))
+		return 0;
+
+	tm = alloc_tree_mod_elem(eb, slot, op, flags);
+	if (!tm)
+		return -ENOMEM;
+
+	if (tree_mod_dont_log(eb->fs_info, eb)) {
+		kfree(tm);
+		return 0;
+	}
+
+	ret = tree_mod_log_insert(eb->fs_info, tm);
+	write_unlock(&eb->fs_info->tree_mod_log_lock);
+	if (ret)
+		kfree(tm);
+
+	return ret;
+}
+
+int btrfs_tree_mod_log_insert_move(struct extent_buffer *eb,
+				   int dst_slot, int src_slot,
+				   int nr_items)
+{
+	struct tree_mod_elem *tm = NULL;
+	struct tree_mod_elem **tm_list = NULL;
+	int ret = 0;
+	int i;
+	bool locked = false;
+
+	if (!tree_mod_need_log(eb->fs_info, eb))
+		return 0;
+
+	tm_list = kcalloc(nr_items, sizeof(struct tree_mod_elem *), GFP_NOFS);
+	if (!tm_list)
+		return -ENOMEM;
+
+	tm = kzalloc(sizeof(*tm), GFP_NOFS);
+	if (!tm) {
+		ret = -ENOMEM;
+		goto free_tms;
+	}
+
+	tm->logical = eb->start;
+	tm->slot = src_slot;
+	tm->move.dst_slot = dst_slot;
+	tm->move.nr_items = nr_items;
+	tm->op = BTRFS_MOD_LOG_MOVE_KEYS;
+
+	for (i = 0; i + dst_slot < src_slot && i < nr_items; i++) {
+		tm_list[i] = alloc_tree_mod_elem(eb, i + dst_slot,
+				BTRFS_MOD_LOG_KEY_REMOVE_WHILE_MOVING, GFP_NOFS);
+		if (!tm_list[i]) {
+			ret = -ENOMEM;
+			goto free_tms;
+		}
+	}
+
+	if (tree_mod_dont_log(eb->fs_info, eb))
+		goto free_tms;
+	locked = true;
+
+	/*
+	 * When we override something during the move, we log these removals.
+	 * This can only happen when we move towards the beginning of the
+	 * buffer, i.e. dst_slot < src_slot.
+	 */
+	for (i = 0; i + dst_slot < src_slot && i < nr_items; i++) {
+		ret = tree_mod_log_insert(eb->fs_info, tm_list[i]);
+		if (ret)
+			goto free_tms;
+	}
+
+	ret = tree_mod_log_insert(eb->fs_info, tm);
+	if (ret)
+		goto free_tms;
+	write_unlock(&eb->fs_info->tree_mod_log_lock);
+	kfree(tm_list);
+
+	return 0;
+
+free_tms:
+	for (i = 0; i < nr_items; i++) {
+		if (tm_list[i] && !RB_EMPTY_NODE(&tm_list[i]->node))
+			rb_erase(&tm_list[i]->node, &eb->fs_info->tree_mod_log);
+		kfree(tm_list[i]);
+	}
+	if (locked)
+		write_unlock(&eb->fs_info->tree_mod_log_lock);
+	kfree(tm_list);
+	kfree(tm);
+
+	return ret;
+}
+
+static inline int tree_mod_log_free_eb(struct btrfs_fs_info *fs_info,
+				       struct tree_mod_elem **tm_list,
+				       int nritems)
+{
+	int i, j;
+	int ret;
+
+	for (i = nritems - 1; i >= 0; i--) {
+		ret = tree_mod_log_insert(fs_info, tm_list[i]);
+		if (ret) {
+			for (j = nritems - 1; j > i; j--)
+				rb_erase(&tm_list[j]->node,
+					 &fs_info->tree_mod_log);
+			return ret;
+		}
+	}
+
+	return 0;
+}
+
+int btrfs_tree_mod_log_insert_root(struct extent_buffer *old_root,
+				   struct extent_buffer *new_root,
+				   bool log_removal)
+{
+	struct btrfs_fs_info *fs_info = old_root->fs_info;
+	struct tree_mod_elem *tm = NULL;
+	struct tree_mod_elem **tm_list = NULL;
+	int nritems = 0;
+	int ret = 0;
+	int i;
+
+	if (!tree_mod_need_log(fs_info, NULL))
+		return 0;
+
+	if (log_removal && btrfs_header_level(old_root) > 0) {
+		nritems = btrfs_header_nritems(old_root);
+		tm_list = kcalloc(nritems, sizeof(struct tree_mod_elem *),
+				  GFP_NOFS);
+		if (!tm_list) {
+			ret = -ENOMEM;
+			goto free_tms;
+		}
+		for (i = 0; i < nritems; i++) {
+			tm_list[i] = alloc_tree_mod_elem(old_root, i,
+			    BTRFS_MOD_LOG_KEY_REMOVE_WHILE_FREEING, GFP_NOFS);
+			if (!tm_list[i]) {
+				ret = -ENOMEM;
+				goto free_tms;
+			}
+		}
+	}
+
+	tm = kzalloc(sizeof(*tm), GFP_NOFS);
+	if (!tm) {
+		ret = -ENOMEM;
+		goto free_tms;
+	}
+
+	tm->logical = new_root->start;
+	tm->old_root.logical = old_root->start;
+	tm->old_root.level = btrfs_header_level(old_root);
+	tm->generation = btrfs_header_generation(old_root);
+	tm->op = BTRFS_MOD_LOG_ROOT_REPLACE;
+
+	if (tree_mod_dont_log(fs_info, NULL))
+		goto free_tms;
+
+	if (tm_list)
+		ret = tree_mod_log_free_eb(fs_info, tm_list, nritems);
+	if (!ret)
+		ret = tree_mod_log_insert(fs_info, tm);
+
+	write_unlock(&fs_info->tree_mod_log_lock);
+	if (ret)
+		goto free_tms;
+	kfree(tm_list);
+
+	return ret;
+
+free_tms:
+	if (tm_list) {
+		for (i = 0; i < nritems; i++)
+			kfree(tm_list[i]);
+		kfree(tm_list);
+	}
+	kfree(tm);
+
+	return ret;
+}
+
+static struct tree_mod_elem *__tree_mod_log_search(struct btrfs_fs_info *fs_info,
+						   u64 start, u64 min_seq,
+						   bool smallest)
+{
+	struct rb_root *tm_root;
+	struct rb_node *node;
+	struct tree_mod_elem *cur = NULL;
+	struct tree_mod_elem *found = NULL;
+
+	read_lock(&fs_info->tree_mod_log_lock);
+	tm_root = &fs_info->tree_mod_log;
+	node = tm_root->rb_node;
+	while (node) {
+		cur = rb_entry(node, struct tree_mod_elem, node);
+		if (cur->logical < start) {
+			node = node->rb_left;
+		} else if (cur->logical > start) {
+			node = node->rb_right;
+		} else if (cur->seq < min_seq) {
+			node = node->rb_left;
+		} else if (!smallest) {
+			/* We want the node with the highest seq */
+			if (found)
+				BUG_ON(found->seq > cur->seq);
+			found = cur;
+			node = node->rb_left;
+		} else if (cur->seq > min_seq) {
+			/* We want the node with the smallest seq */
+			if (found)
+				BUG_ON(found->seq < cur->seq);
+			found = cur;
+			node = node->rb_right;
+		} else {
+			found = cur;
+			break;
+		}
+	}
+	read_unlock(&fs_info->tree_mod_log_lock);
+
+	return found;
+}
+
+/*
+ * This returns the element from the log with the smallest time sequence
+ * value that's in the log (the oldest log item). Any element with a time
+ * sequence lower than min_seq will be ignored.
+ */
+static struct tree_mod_elem *tree_mod_log_search_oldest(struct btrfs_fs_info *fs_info,
+							u64 start, u64 min_seq)
+{
+	return __tree_mod_log_search(fs_info, start, min_seq, true);
+}
+
+/*
+ * This returns the element from the log with the largest time sequence
+ * value that's in the log (the most recent log item). Any element with
+ * a time sequence lower than min_seq will be ignored.
+ */
+static struct tree_mod_elem *tree_mod_log_search(struct btrfs_fs_info *fs_info,
+						 u64 start, u64 min_seq)
+{
+	return __tree_mod_log_search(fs_info, start, min_seq, false);
+}
+
+int btrfs_tree_mod_log_eb_copy(struct extent_buffer *dst,
+			       struct extent_buffer *src,
+			       unsigned long dst_offset,
+			       unsigned long src_offset,
+			       int nr_items)
+{
+	struct btrfs_fs_info *fs_info = dst->fs_info;
+	int ret = 0;
+	struct tree_mod_elem **tm_list = NULL;
+	struct tree_mod_elem **tm_list_add, **tm_list_rem;
+	int i;
+	bool locked = false;
+
+	if (!tree_mod_need_log(fs_info, NULL))
+		return 0;
+
+	if (btrfs_header_level(dst) == 0 && btrfs_header_level(src) == 0)
+		return 0;
+
+	tm_list = kcalloc(nr_items * 2, sizeof(struct tree_mod_elem *),
+			  GFP_NOFS);
+	if (!tm_list)
+		return -ENOMEM;
+
+	tm_list_add = tm_list;
+	tm_list_rem = tm_list + nr_items;
+	for (i = 0; i < nr_items; i++) {
+		tm_list_rem[i] = alloc_tree_mod_elem(src, i + src_offset,
+		    BTRFS_MOD_LOG_KEY_REMOVE, GFP_NOFS);
+		if (!tm_list_rem[i]) {
+			ret = -ENOMEM;
+			goto free_tms;
+		}
+
+		tm_list_add[i] = alloc_tree_mod_elem(dst, i + dst_offset,
+						BTRFS_MOD_LOG_KEY_ADD, GFP_NOFS);
+		if (!tm_list_add[i]) {
+			ret = -ENOMEM;
+			goto free_tms;
+		}
+	}
+
+	if (tree_mod_dont_log(fs_info, NULL))
+		goto free_tms;
+	locked = true;
+
+	for (i = 0; i < nr_items; i++) {
+		ret = tree_mod_log_insert(fs_info, tm_list_rem[i]);
+		if (ret)
+			goto free_tms;
+		ret = tree_mod_log_insert(fs_info, tm_list_add[i]);
+		if (ret)
+			goto free_tms;
+	}
+
+	write_unlock(&fs_info->tree_mod_log_lock);
+	kfree(tm_list);
+
+	return 0;
+
+free_tms:
+	for (i = 0; i < nr_items * 2; i++) {
+		if (tm_list[i] && !RB_EMPTY_NODE(&tm_list[i]->node))
+			rb_erase(&tm_list[i]->node, &fs_info->tree_mod_log);
+		kfree(tm_list[i]);
+	}
+	if (locked)
+		write_unlock(&fs_info->tree_mod_log_lock);
+	kfree(tm_list);
+
+	return ret;
+}
+
+int btrfs_tree_mod_log_free_eb(struct extent_buffer *eb)
+{
+	struct tree_mod_elem **tm_list = NULL;
+	int nritems = 0;
+	int i;
+	int ret = 0;
+
+	if (!tree_mod_need_log(eb->fs_info, eb))
+		return 0;
+
+	nritems = btrfs_header_nritems(eb);
+	tm_list = kcalloc(nritems, sizeof(struct tree_mod_elem *), GFP_NOFS);
+	if (!tm_list)
+		return -ENOMEM;
+
+	for (i = 0; i < nritems; i++) {
+		tm_list[i] = alloc_tree_mod_elem(eb, i,
+		    BTRFS_MOD_LOG_KEY_REMOVE_WHILE_FREEING, GFP_NOFS);
+		if (!tm_list[i]) {
+			ret = -ENOMEM;
+			goto free_tms;
+		}
+	}
+
+	if (tree_mod_dont_log(eb->fs_info, eb))
+		goto free_tms;
+
+	ret = tree_mod_log_free_eb(eb->fs_info, tm_list, nritems);
+	write_unlock(&eb->fs_info->tree_mod_log_lock);
+	if (ret)
+		goto free_tms;
+	kfree(tm_list);
+
+	return 0;
+
+free_tms:
+	for (i = 0; i < nritems; i++)
+		kfree(tm_list[i]);
+	kfree(tm_list);
+
+	return ret;
+}
+
+/*
+ * Returns the logical address of the oldest predecessor of the given root.
+ * Entries older than time_seq are ignored.
+ */
+static struct tree_mod_elem *tree_mod_log_oldest_root(struct extent_buffer *eb_root,
+						      u64 time_seq)
+{
+	struct tree_mod_elem *tm;
+	struct tree_mod_elem *found = NULL;
+	u64 root_logical = eb_root->start;
+	bool looped = false;
+
+	if (!time_seq)
+		return NULL;
+
+	/*
+	 * The very last operation that's logged for a root is the replacement
+	 * operation (if it is replaced at all). This has the logical address
+	 * of the *new* root, making it the very first operation that's logged
+	 * for this root.
+	 */
+	while (1) {
+		tm = tree_mod_log_search_oldest(eb_root->fs_info, root_logical,
+						time_seq);
+		if (!looped && !tm)
+			return NULL;
+		/*
+		 * If there are no tree operation for the oldest root, we simply
+		 * return it. This should only happen if that (old) root is at
+		 * level 0.
+		 */
+		if (!tm)
+			break;
+
+		/*
+		 * If there's an operation that's not a root replacement, we
+		 * found the oldest version of our root. Normally, we'll find a
+		 * BTRFS_MOD_LOG_KEY_REMOVE_WHILE_FREEING operation here.
+		 */
+		if (tm->op != BTRFS_MOD_LOG_ROOT_REPLACE)
+			break;
+
+		found = tm;
+		root_logical = tm->old_root.logical;
+		looped = true;
+	}
+
+	/* If there's no old root to return, return what we found instead */
+	if (!found)
+		found = tm;
+
+	return found;
+}
+
+
+/*
+ * tm is a pointer to the first operation to rewind within eb. Then, all
+ * previous operations will be rewound (until we reach something older than
+ * time_seq).
+ */
+static void tree_mod_log_rewind(struct btrfs_fs_info *fs_info,
+				struct extent_buffer *eb,
+				u64 time_seq,
+				struct tree_mod_elem *first_tm)
+{
+	u32 n;
+	struct rb_node *next;
+	struct tree_mod_elem *tm = first_tm;
+	unsigned long o_dst;
+	unsigned long o_src;
+	unsigned long p_size = sizeof(struct btrfs_key_ptr);
+
+	n = btrfs_header_nritems(eb);
+	read_lock(&fs_info->tree_mod_log_lock);
+	while (tm && tm->seq >= time_seq) {
+		/*
+		 * All the operations are recorded with the operator used for
+		 * the modification. As we're going backwards, we do the
+		 * opposite of each operation here.
+		 */
+		switch (tm->op) {
+		case BTRFS_MOD_LOG_KEY_REMOVE_WHILE_FREEING:
+			BUG_ON(tm->slot < n);
+			fallthrough;
+		case BTRFS_MOD_LOG_KEY_REMOVE_WHILE_MOVING:
+		case BTRFS_MOD_LOG_KEY_REMOVE:
+			btrfs_set_node_key(eb, &tm->key, tm->slot);
+			btrfs_set_node_blockptr(eb, tm->slot, tm->blockptr);
+			btrfs_set_node_ptr_generation(eb, tm->slot,
+						      tm->generation);
+			n++;
+			break;
+		case BTRFS_MOD_LOG_KEY_REPLACE:
+			BUG_ON(tm->slot >= n);
+			btrfs_set_node_key(eb, &tm->key, tm->slot);
+			btrfs_set_node_blockptr(eb, tm->slot, tm->blockptr);
+			btrfs_set_node_ptr_generation(eb, tm->slot,
+						      tm->generation);
+			break;
+		case BTRFS_MOD_LOG_KEY_ADD:
+			/* if a move operation is needed it's in the log */
+			n--;
+			break;
+		case BTRFS_MOD_LOG_MOVE_KEYS:
+			o_dst = btrfs_node_key_ptr_offset(tm->slot);
+			o_src = btrfs_node_key_ptr_offset(tm->move.dst_slot);
+			memmove_extent_buffer(eb, o_dst, o_src,
+					      tm->move.nr_items * p_size);
+			break;
+		case BTRFS_MOD_LOG_ROOT_REPLACE:
+			/*
+			 * This operation is special. For roots, this must be
+			 * handled explicitly before rewinding.
+			 * For non-roots, this operation may exist if the node
+			 * was a root: root A -> child B; then A gets empty and
+			 * B is promoted to the new root. In the mod log, we'll
+			 * have a root-replace operation for B, a tree block
+			 * that is no root. We simply ignore that operation.
+			 */
+			break;
+		}
+		next = rb_next(&tm->node);
+		if (!next)
+			break;
+		tm = rb_entry(next, struct tree_mod_elem, node);
+		if (tm->logical != first_tm->logical)
+			break;
+	}
+	read_unlock(&fs_info->tree_mod_log_lock);
+	btrfs_set_header_nritems(eb, n);
+}
+
+/*
+ * Called with eb read locked. If the buffer cannot be rewound, the same buffer
+ * is returned. If rewind operations happen, a fresh buffer is returned. The
+ * returned buffer is always read-locked. If the returned buffer is not the
+ * input buffer, the lock on the input buffer is released and the input buffer
+ * is freed (its refcount is decremented).
+ */
+struct extent_buffer *btrfs_tree_mod_log_rewind(struct btrfs_fs_info *fs_info,
+						struct btrfs_path *path,
+						struct extent_buffer *eb,
+						u64 time_seq)
+{
+	struct extent_buffer *eb_rewin;
+	struct tree_mod_elem *tm;
+
+	if (!time_seq)
+		return eb;
+
+	if (btrfs_header_level(eb) == 0)
+		return eb;
+
+	tm = tree_mod_log_search(fs_info, eb->start, time_seq);
+	if (!tm)
+		return eb;
+
+	if (tm->op == BTRFS_MOD_LOG_KEY_REMOVE_WHILE_FREEING) {
+		BUG_ON(tm->slot != 0);
+		eb_rewin = alloc_dummy_extent_buffer(fs_info, eb->start);
+		if (!eb_rewin) {
+			btrfs_tree_read_unlock(eb);
+			free_extent_buffer(eb);
+			return NULL;
+		}
+		btrfs_set_header_bytenr(eb_rewin, eb->start);
+		btrfs_set_header_backref_rev(eb_rewin,
+					     btrfs_header_backref_rev(eb));
+		btrfs_set_header_owner(eb_rewin, btrfs_header_owner(eb));
+		btrfs_set_header_level(eb_rewin, btrfs_header_level(eb));
+	} else {
+		eb_rewin = btrfs_clone_extent_buffer(eb);
+		if (!eb_rewin) {
+			btrfs_tree_read_unlock(eb);
+			free_extent_buffer(eb);
+			return NULL;
+		}
+	}
+
+	btrfs_tree_read_unlock(eb);
+	free_extent_buffer(eb);
+
+	btrfs_set_buffer_lockdep_class(btrfs_header_owner(eb_rewin),
+				       eb_rewin, btrfs_header_level(eb_rewin));
+	btrfs_tree_read_lock(eb_rewin);
+	tree_mod_log_rewind(fs_info, eb_rewin, time_seq, tm);
+	WARN_ON(btrfs_header_nritems(eb_rewin) >
+		BTRFS_NODEPTRS_PER_BLOCK(fs_info));
+
+	return eb_rewin;
+}
+
+/*
+ * Rewind the state of @root's root node to the given @time_seq value.
+ * If there are no changes, the current root->root_node is returned. If anything
+ * changed in between, there's a fresh buffer allocated on which the rewind
+ * operations are done. In any case, the returned buffer is read locked.
+ * Returns NULL on error (with no locks held).
+ */
+struct extent_buffer *btrfs_get_old_root(struct btrfs_root *root, u64 time_seq)
+{
+	struct btrfs_fs_info *fs_info = root->fs_info;
+	struct tree_mod_elem *tm;
+	struct extent_buffer *eb = NULL;
+	struct extent_buffer *eb_root;
+	u64 eb_root_owner = 0;
+	struct extent_buffer *old;
+	struct tree_mod_root *old_root = NULL;
+	u64 old_generation = 0;
+	u64 logical;
+	int level;
+
+	eb_root = btrfs_read_lock_root_node(root);
+	tm = tree_mod_log_oldest_root(eb_root, time_seq);
+	if (!tm)
+		return eb_root;
+
+	if (tm->op == BTRFS_MOD_LOG_ROOT_REPLACE) {
+		old_root = &tm->old_root;
+		old_generation = tm->generation;
+		logical = old_root->logical;
+		level = old_root->level;
+	} else {
+		logical = eb_root->start;
+		level = btrfs_header_level(eb_root);
+	}
+
+	tm = tree_mod_log_search(fs_info, logical, time_seq);
+	if (old_root && tm && tm->op != BTRFS_MOD_LOG_KEY_REMOVE_WHILE_FREEING) {
+		btrfs_tree_read_unlock(eb_root);
+		free_extent_buffer(eb_root);
+		old = read_tree_block(fs_info, logical, root->root_key.objectid,
+				      0, level, NULL);
+		if (WARN_ON(IS_ERR(old) || !extent_buffer_uptodate(old))) {
+			if (!IS_ERR(old))
+				free_extent_buffer(old);
+			btrfs_warn(fs_info,
+				   "failed to read tree block %llu from get_old_root",
+				   logical);
+		} else {
+			struct tree_mod_elem *tm2;
+
+			btrfs_tree_read_lock(old);
+			eb = btrfs_clone_extent_buffer(old);
+			/*
+			 * After the lookup for the most recent tree mod operation
+			 * above and before we locked and cloned the extent buffer
+			 * 'old', a new tree mod log operation may have been added.
+			 * So lookup for a more recent one to make sure the number
+			 * of mod log operations we replay is consistent with the
+			 * number of items we have in the cloned extent buffer,
+			 * otherwise we can hit a BUG_ON when rewinding the extent
+			 * buffer.
+			 */
+			tm2 = tree_mod_log_search(fs_info, logical, time_seq);
+			btrfs_tree_read_unlock(old);
+			free_extent_buffer(old);
+			ASSERT(tm2);
+			ASSERT(tm2 == tm || tm2->seq > tm->seq);
+			if (!tm2 || tm2->seq < tm->seq) {
+				free_extent_buffer(eb);
+				return NULL;
+			}
+			tm = tm2;
+		}
+	} else if (old_root) {
+		eb_root_owner = btrfs_header_owner(eb_root);
+		btrfs_tree_read_unlock(eb_root);
+		free_extent_buffer(eb_root);
+		eb = alloc_dummy_extent_buffer(fs_info, logical);
+	} else {
+		eb = btrfs_clone_extent_buffer(eb_root);
+		btrfs_tree_read_unlock(eb_root);
+		free_extent_buffer(eb_root);
+	}
+
+	if (!eb)
+		return NULL;
+	if (old_root) {
+		btrfs_set_header_bytenr(eb, eb->start);
+		btrfs_set_header_backref_rev(eb, BTRFS_MIXED_BACKREF_REV);
+		btrfs_set_header_owner(eb, eb_root_owner);
+		btrfs_set_header_level(eb, old_root->level);
+		btrfs_set_header_generation(eb, old_generation);
+	}
+	btrfs_set_buffer_lockdep_class(btrfs_header_owner(eb), eb,
+				       btrfs_header_level(eb));
+	btrfs_tree_read_lock(eb);
+	if (tm)
+		tree_mod_log_rewind(fs_info, eb, time_seq, tm);
+	else
+		WARN_ON(btrfs_header_level(eb) != 0);
+	WARN_ON(btrfs_header_nritems(eb) > BTRFS_NODEPTRS_PER_BLOCK(fs_info));
+
+	return eb;
+}
+
+int btrfs_old_root_level(struct btrfs_root *root, u64 time_seq)
+{
+	struct tree_mod_elem *tm;
+	int level;
+	struct extent_buffer *eb_root = btrfs_root_node(root);
+
+	tm = tree_mod_log_oldest_root(eb_root, time_seq);
+	if (tm && tm->op == BTRFS_MOD_LOG_ROOT_REPLACE)
+		level = tm->old_root.level;
+	else
+		level = btrfs_header_level(eb_root);
+
+	free_extent_buffer(eb_root);
+
+	return level;
+}
+
+/*
+ * Return the lowest sequence number in the tree modification log.
+ *
+ * Return the sequence number of the oldest tree modification log user, which
+ * corresponds to the lowest sequence number of all existing users. If there are
+ * no users it returns 0.
+ */
+u64 btrfs_tree_mod_log_lowest_seq(struct btrfs_fs_info *fs_info)
+{
+	u64 ret = 0;
+
+	read_lock(&fs_info->tree_mod_log_lock);
+	if (!list_empty(&fs_info->tree_mod_seq_list)) {
+		struct btrfs_seq_list *elem;
+
+		elem = list_first_entry(&fs_info->tree_mod_seq_list,
+					struct btrfs_seq_list, list);
+		ret = elem->seq;
+	}
+	read_unlock(&fs_info->tree_mod_log_lock);
+
+	return ret;
+}
diff --git a/fs/btrfs/tree-mod-log.h b/fs/btrfs/tree-mod-log.h
new file mode 100644
index 000000000000..12605d19621b
--- /dev/null
+++ b/fs/btrfs/tree-mod-log.h
@@ -0,0 +1,53 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#ifndef BTRFS_TREE_MOD_LOG_H
+#define BTRFS_TREE_MOD_LOG_H
+
+#include "ctree.h"
+
+/* Represents a tree mod log user. */
+struct btrfs_seq_list {
+	struct list_head list;
+	u64 seq;
+};
+
+#define BTRFS_SEQ_LIST_INIT(name) { .list = LIST_HEAD_INIT((name).list), .seq = 0 }
+#define BTRFS_SEQ_LAST            ((u64)-1)
+
+enum btrfs_mod_log_op {
+	BTRFS_MOD_LOG_KEY_REPLACE,
+	BTRFS_MOD_LOG_KEY_ADD,
+	BTRFS_MOD_LOG_KEY_REMOVE,
+	BTRFS_MOD_LOG_KEY_REMOVE_WHILE_FREEING,
+	BTRFS_MOD_LOG_KEY_REMOVE_WHILE_MOVING,
+	BTRFS_MOD_LOG_MOVE_KEYS,
+	BTRFS_MOD_LOG_ROOT_REPLACE,
+};
+
+u64 btrfs_get_tree_mod_seq(struct btrfs_fs_info *fs_info,
+			   struct btrfs_seq_list *elem);
+void btrfs_put_tree_mod_seq(struct btrfs_fs_info *fs_info,
+			    struct btrfs_seq_list *elem);
+int btrfs_tree_mod_log_insert_root(struct extent_buffer *old_root,
+				   struct extent_buffer *new_root,
+				   bool log_removal);
+int btrfs_tree_mod_log_insert_key(struct extent_buffer *eb, int slot,
+				  enum btrfs_mod_log_op op, gfp_t flags);
+int btrfs_tree_mod_log_free_eb(struct extent_buffer *eb);
+struct extent_buffer *btrfs_tree_mod_log_rewind(struct btrfs_fs_info *fs_info,
+						struct btrfs_path *path,
+						struct extent_buffer *eb,
+						u64 time_seq);
+struct extent_buffer *btrfs_get_old_root(struct btrfs_root *root, u64 time_seq);
+int btrfs_old_root_level(struct btrfs_root *root, u64 time_seq);
+int btrfs_tree_mod_log_eb_copy(struct extent_buffer *dst,
+			       struct extent_buffer *src,
+			       unsigned long dst_offset,
+			       unsigned long src_offset,
+			       int nr_items);
+int btrfs_tree_mod_log_insert_move(struct extent_buffer *eb,
+				   int dst_slot, int src_slot,
+				   int nr_items);
+u64 btrfs_tree_mod_log_lowest_seq(struct btrfs_fs_info *fs_info);
+
+#endif
diff --git a/fs/btrfs/uuid-tree.c b/fs/btrfs/uuid-tree.c
index 76b84f2397b1..b458452a1aaf 100644
--- a/fs/btrfs/uuid-tree.c
+++ b/fs/btrfs/uuid-tree.c
@@ -52,7 +52,7 @@ static int btrfs_uuid_tree_lookup(struct btrfs_root *uuid_root, u8 *uuid,
 
 	eb = path->nodes[0];
 	slot = path->slots[0];
-	item_size = btrfs_item_size_nr(eb, slot);
+	item_size = btrfs_item_size(eb, slot);
 	offset = btrfs_item_ptr_offset(eb, slot);
 	ret = -ENOENT;
 
@@ -125,12 +125,11 @@ int btrfs_uuid_tree_add(struct btrfs_trans_handle *trans, u8 *uuid, u8 type,
 		eb = path->nodes[0];
 		slot = path->slots[0];
 		offset = btrfs_item_ptr_offset(eb, slot);
-		offset += btrfs_item_size_nr(eb, slot) - sizeof(subid_le);
+		offset += btrfs_item_size(eb, slot) - sizeof(subid_le);
 	} else {
 		btrfs_warn(fs_info,
 			   "insert uuid item failed %d (0x%016llx, 0x%016llx) type %u!",
-			   ret, (unsigned long long)key.objectid,
-			   (unsigned long long)key.offset, type);
+			   ret, key.objectid, key.offset, type);
 		goto out;
 	}
 
@@ -187,7 +186,7 @@ int btrfs_uuid_tree_remove(struct btrfs_trans_handle *trans, u8 *uuid, u8 type,
 	eb = path->nodes[0];
 	slot = path->slots[0];
 	offset = btrfs_item_ptr_offset(eb, slot);
-	item_size = btrfs_item_size_nr(eb, slot);
+	item_size = btrfs_item_size(eb, slot);
 	if (!IS_ALIGNED(item_size, sizeof(u64))) {
 		btrfs_warn(fs_info, "uuid item with illegal size %lu!",
 			   (unsigned long)item_size);
@@ -209,7 +208,7 @@ int btrfs_uuid_tree_remove(struct btrfs_trans_handle *trans, u8 *uuid, u8 type,
 		goto out;
 	}
 
-	item_size = btrfs_item_size_nr(eb, slot);
+	item_size = btrfs_item_size(eb, slot);
 	if (item_size == sizeof(subid)) {
 		ret = btrfs_del_item(trans, uuid_root, path);
 		goto out;
@@ -246,9 +245,49 @@ out:
 	return ret;
 }
 
-int btrfs_uuid_tree_iterate(struct btrfs_fs_info *fs_info,
-			    int (*check_func)(struct btrfs_fs_info *, u8 *, u8,
-					      u64))
+/*
+ * Check if there's an matching subvolume for given UUID
+ *
+ * Return:
+ * 0	check succeeded, the entry is not outdated
+ * > 0	if the check failed, the caller should remove the entry
+ * < 0	if an error occurred
+ */
+static int btrfs_check_uuid_tree_entry(struct btrfs_fs_info *fs_info,
+				       u8 *uuid, u8 type, u64 subvolid)
+{
+	int ret = 0;
+	struct btrfs_root *subvol_root;
+
+	if (type != BTRFS_UUID_KEY_SUBVOL &&
+	    type != BTRFS_UUID_KEY_RECEIVED_SUBVOL)
+		goto out;
+
+	subvol_root = btrfs_get_fs_root(fs_info, subvolid, true);
+	if (IS_ERR(subvol_root)) {
+		ret = PTR_ERR(subvol_root);
+		if (ret == -ENOENT)
+			ret = 1;
+		goto out;
+	}
+
+	switch (type) {
+	case BTRFS_UUID_KEY_SUBVOL:
+		if (memcmp(uuid, subvol_root->root_item.uuid, BTRFS_UUID_SIZE))
+			ret = 1;
+		break;
+	case BTRFS_UUID_KEY_RECEIVED_SUBVOL:
+		if (memcmp(uuid, subvol_root->root_item.received_uuid,
+			   BTRFS_UUID_SIZE))
+			ret = 1;
+		break;
+	}
+	btrfs_put_root(subvol_root);
+out:
+	return ret;
+}
+
+int btrfs_uuid_tree_iterate(struct btrfs_fs_info *fs_info)
 {
 	struct btrfs_root *root = fs_info->uuid_root;
 	struct btrfs_key key;
@@ -278,6 +317,10 @@ again_search_slot:
 	}
 
 	while (1) {
+		if (btrfs_fs_closing(fs_info)) {
+			ret = -EINTR;
+			goto out;
+		}
 		cond_resched();
 		leaf = path->nodes[0];
 		slot = path->slots[0];
@@ -288,7 +331,7 @@ again_search_slot:
 			goto skip;
 
 		offset = btrfs_item_ptr_offset(leaf, slot);
-		item_size = btrfs_item_size_nr(leaf, slot);
+		item_size = btrfs_item_size(leaf, slot);
 		if (!IS_ALIGNED(item_size, sizeof(u64))) {
 			btrfs_warn(fs_info,
 				   "uuid item with illegal size %lu!",
@@ -305,7 +348,8 @@ again_search_slot:
 			read_extent_buffer(leaf, &subid_le, offset,
 					   sizeof(subid_le));
 			subid_cpu = le64_to_cpu(subid_le);
-			ret = check_func(fs_info, uuid, key.type, subid_cpu);
+			ret = btrfs_check_uuid_tree_entry(fs_info, uuid,
+							  key.type, subid_cpu);
 			if (ret < 0)
 				goto out;
 			if (ret > 0) {
diff --git a/fs/btrfs/verity.c b/fs/btrfs/verity.c
new file mode 100644
index 000000000000..ee00e33c309e
--- /dev/null
+++ b/fs/btrfs/verity.c
@@ -0,0 +1,812 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <linux/init.h>
+#include <linux/fs.h>
+#include <linux/slab.h>
+#include <linux/rwsem.h>
+#include <linux/xattr.h>
+#include <linux/security.h>
+#include <linux/posix_acl_xattr.h>
+#include <linux/iversion.h>
+#include <linux/fsverity.h>
+#include <linux/sched/mm.h>
+#include "ctree.h"
+#include "btrfs_inode.h"
+#include "transaction.h"
+#include "disk-io.h"
+#include "locking.h"
+
+/*
+ * Implementation of the interface defined in struct fsverity_operations.
+ *
+ * The main question is how and where to store the verity descriptor and the
+ * Merkle tree. We store both in dedicated btree items in the filesystem tree,
+ * together with the rest of the inode metadata. This means we'll need to do
+ * extra work to encrypt them once encryption is supported in btrfs, but btrfs
+ * has a lot of careful code around i_size and it seems better to make a new key
+ * type than try and adjust all of our expectations for i_size.
+ *
+ * Note that this differs from the implementation in ext4 and f2fs, where
+ * this data is stored as if it were in the file, but past EOF. However, btrfs
+ * does not have a widespread mechanism for caching opaque metadata pages, so we
+ * do pretend that the Merkle tree pages themselves are past EOF for the
+ * purposes of caching them (as opposed to creating a virtual inode).
+ *
+ * fs verity items are stored under two different key types on disk.
+ * The descriptor items:
+ * [ inode objectid, BTRFS_VERITY_DESC_ITEM_KEY, offset ]
+ *
+ * At offset 0, we store a btrfs_verity_descriptor_item which tracks the
+ * size of the descriptor item and some extra data for encryption.
+ * Starting at offset 1, these hold the generic fs verity descriptor.
+ * The latter are opaque to btrfs, we just read and write them as a blob for
+ * the higher level verity code.  The most common descriptor size is 256 bytes.
+ *
+ * The merkle tree items:
+ * [ inode objectid, BTRFS_VERITY_MERKLE_ITEM_KEY, offset ]
+ *
+ * These also start at offset 0, and correspond to the merkle tree bytes.
+ * So when fsverity asks for page 0 of the merkle tree, we pull up one page
+ * starting at offset 0 for this key type.  These are also opaque to btrfs,
+ * we're blindly storing whatever fsverity sends down.
+ *
+ * Another important consideration is the fact that the Merkle tree data scales
+ * linearly with the size of the file (with 4K pages/blocks and SHA-256, it's
+ * ~1/127th the size) so for large files, writing the tree can be a lengthy
+ * operation. For that reason, we guard the whole enable verity operation
+ * (between begin_enable_verity and end_enable_verity) with an orphan item.
+ * Again, because the data can be pretty large, it's quite possible that we
+ * could run out of space writing it, so we try our best to handle errors by
+ * stopping and rolling back rather than aborting the victim transaction.
+ */
+
+#define MERKLE_START_ALIGN			65536
+
+/*
+ * Compute the logical file offset where we cache the Merkle tree.
+ *
+ * @inode:  inode of the verity file
+ *
+ * For the purposes of caching the Merkle tree pages, as required by
+ * fs-verity, it is convenient to do size computations in terms of a file
+ * offset, rather than in terms of page indices.
+ *
+ * Use 64K to be sure it's past the last page in the file, even with 64K pages.
+ * That rounding operation itself can overflow loff_t, so we do it in u64 and
+ * check.
+ *
+ * Returns the file offset on success, negative error code on failure.
+ */
+static loff_t merkle_file_pos(const struct inode *inode)
+{
+	u64 sz = inode->i_size;
+	u64 rounded = round_up(sz, MERKLE_START_ALIGN);
+
+	if (rounded > inode->i_sb->s_maxbytes)
+		return -EFBIG;
+
+	return rounded;
+}
+
+/*
+ * Drop all the items for this inode with this key_type.
+ *
+ * @inode:     inode to drop items for
+ * @key_type:  type of items to drop (BTRFS_VERITY_DESC_ITEM or
+ *             BTRFS_VERITY_MERKLE_ITEM)
+ *
+ * Before doing a verity enable we cleanup any existing verity items.
+ * This is also used to clean up if a verity enable failed half way through.
+ *
+ * Returns number of dropped items on success, negative error code on failure.
+ */
+static int drop_verity_items(struct btrfs_inode *inode, u8 key_type)
+{
+	struct btrfs_trans_handle *trans;
+	struct btrfs_root *root = inode->root;
+	struct btrfs_path *path;
+	struct btrfs_key key;
+	int count = 0;
+	int ret;
+
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+
+	while (1) {
+		/* 1 for the item being dropped */
+		trans = btrfs_start_transaction(root, 1);
+		if (IS_ERR(trans)) {
+			ret = PTR_ERR(trans);
+			goto out;
+		}
+
+		/*
+		 * Walk backwards through all the items until we find one that
+		 * isn't from our key type or objectid
+		 */
+		key.objectid = btrfs_ino(inode);
+		key.type = key_type;
+		key.offset = (u64)-1;
+
+		ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
+		if (ret > 0) {
+			ret = 0;
+			/* No more keys of this type, we're done */
+			if (path->slots[0] == 0)
+				break;
+			path->slots[0]--;
+		} else if (ret < 0) {
+			btrfs_end_transaction(trans);
+			goto out;
+		}
+
+		btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
+
+		/* No more keys of this type, we're done */
+		if (key.objectid != btrfs_ino(inode) || key.type != key_type)
+			break;
+
+		/*
+		 * This shouldn't be a performance sensitive function because
+		 * it's not used as part of truncate.  If it ever becomes
+		 * perf sensitive, change this to walk forward and bulk delete
+		 * items
+		 */
+		ret = btrfs_del_items(trans, root, path, path->slots[0], 1);
+		if (ret) {
+			btrfs_end_transaction(trans);
+			goto out;
+		}
+		count++;
+		btrfs_release_path(path);
+		btrfs_end_transaction(trans);
+	}
+	ret = count;
+	btrfs_end_transaction(trans);
+out:
+	btrfs_free_path(path);
+	return ret;
+}
+
+/*
+ * Drop all verity items
+ *
+ * @inode:  inode to drop verity items for
+ *
+ * In most contexts where we are dropping verity items, we want to do it for all
+ * the types of verity items, not a particular one.
+ *
+ * Returns: 0 on success, negative error code on failure.
+ */
+int btrfs_drop_verity_items(struct btrfs_inode *inode)
+{
+	int ret;
+
+	ret = drop_verity_items(inode, BTRFS_VERITY_DESC_ITEM_KEY);
+	if (ret < 0)
+		return ret;
+	ret = drop_verity_items(inode, BTRFS_VERITY_MERKLE_ITEM_KEY);
+	if (ret < 0)
+		return ret;
+
+	return 0;
+}
+
+/*
+ * Insert and write inode items with a given key type and offset.
+ *
+ * @inode:     inode to insert for
+ * @key_type:  key type to insert
+ * @offset:    item offset to insert at
+ * @src:       source data to write
+ * @len:       length of source data to write
+ *
+ * Write len bytes from src into items of up to 2K length.
+ * The inserted items will have key (ino, key_type, offset + off) where off is
+ * consecutively increasing from 0 up to the last item ending at offset + len.
+ *
+ * Returns 0 on success and a negative error code on failure.
+ */
+static int write_key_bytes(struct btrfs_inode *inode, u8 key_type, u64 offset,
+			   const char *src, u64 len)
+{
+	struct btrfs_trans_handle *trans;
+	struct btrfs_path *path;
+	struct btrfs_root *root = inode->root;
+	struct extent_buffer *leaf;
+	struct btrfs_key key;
+	unsigned long copy_bytes;
+	unsigned long src_offset = 0;
+	void *data;
+	int ret = 0;
+
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+
+	while (len > 0) {
+		/* 1 for the new item being inserted */
+		trans = btrfs_start_transaction(root, 1);
+		if (IS_ERR(trans)) {
+			ret = PTR_ERR(trans);
+			break;
+		}
+
+		key.objectid = btrfs_ino(inode);
+		key.type = key_type;
+		key.offset = offset;
+
+		/*
+		 * Insert 2K at a time mostly to be friendly for smaller leaf
+		 * size filesystems
+		 */
+		copy_bytes = min_t(u64, len, 2048);
+
+		ret = btrfs_insert_empty_item(trans, root, path, &key, copy_bytes);
+		if (ret) {
+			btrfs_end_transaction(trans);
+			break;
+		}
+
+		leaf = path->nodes[0];
+
+		data = btrfs_item_ptr(leaf, path->slots[0], void);
+		write_extent_buffer(leaf, src + src_offset,
+				    (unsigned long)data, copy_bytes);
+		offset += copy_bytes;
+		src_offset += copy_bytes;
+		len -= copy_bytes;
+
+		btrfs_release_path(path);
+		btrfs_end_transaction(trans);
+	}
+
+	btrfs_free_path(path);
+	return ret;
+}
+
+/*
+ * Read inode items of the given key type and offset from the btree.
+ *
+ * @inode:      inode to read items of
+ * @key_type:   key type to read
+ * @offset:     item offset to read from
+ * @dest:       Buffer to read into. This parameter has slightly tricky
+ *              semantics.  If it is NULL, the function will not do any copying
+ *              and will just return the size of all the items up to len bytes.
+ *              If dest_page is passed, then the function will kmap_local the
+ *              page and ignore dest, but it must still be non-NULL to avoid the
+ *              counting-only behavior.
+ * @len:        length in bytes to read
+ * @dest_page:  copy into this page instead of the dest buffer
+ *
+ * Helper function to read items from the btree.  This returns the number of
+ * bytes read or < 0 for errors.  We can return short reads if the items don't
+ * exist on disk or aren't big enough to fill the desired length.  Supports
+ * reading into a provided buffer (dest) or into the page cache
+ *
+ * Returns number of bytes read or a negative error code on failure.
+ */
+static int read_key_bytes(struct btrfs_inode *inode, u8 key_type, u64 offset,
+			  char *dest, u64 len, struct page *dest_page)
+{
+	struct btrfs_path *path;
+	struct btrfs_root *root = inode->root;
+	struct extent_buffer *leaf;
+	struct btrfs_key key;
+	u64 item_end;
+	u64 copy_end;
+	int copied = 0;
+	u32 copy_offset;
+	unsigned long copy_bytes;
+	unsigned long dest_offset = 0;
+	void *data;
+	char *kaddr = dest;
+	int ret;
+
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+
+	if (dest_page)
+		path->reada = READA_FORWARD;
+
+	key.objectid = btrfs_ino(inode);
+	key.type = key_type;
+	key.offset = offset;
+
+	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
+	if (ret < 0) {
+		goto out;
+	} else if (ret > 0) {
+		ret = 0;
+		if (path->slots[0] == 0)
+			goto out;
+		path->slots[0]--;
+	}
+
+	while (len > 0) {
+		leaf = path->nodes[0];
+		btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
+
+		if (key.objectid != btrfs_ino(inode) || key.type != key_type)
+			break;
+
+		item_end = btrfs_item_size(leaf, path->slots[0]) + key.offset;
+
+		if (copied > 0) {
+			/*
+			 * Once we've copied something, we want all of the items
+			 * to be sequential
+			 */
+			if (key.offset != offset)
+				break;
+		} else {
+			/*
+			 * Our initial offset might be in the middle of an
+			 * item.  Make sure it all makes sense.
+			 */
+			if (key.offset > offset)
+				break;
+			if (item_end <= offset)
+				break;
+		}
+
+		/* desc = NULL to just sum all the item lengths */
+		if (!dest)
+			copy_end = item_end;
+		else
+			copy_end = min(offset + len, item_end);
+
+		/* Number of bytes in this item we want to copy */
+		copy_bytes = copy_end - offset;
+
+		/* Offset from the start of item for copying */
+		copy_offset = offset - key.offset;
+
+		if (dest) {
+			if (dest_page)
+				kaddr = kmap_local_page(dest_page);
+
+			data = btrfs_item_ptr(leaf, path->slots[0], void);
+			read_extent_buffer(leaf, kaddr + dest_offset,
+					   (unsigned long)data + copy_offset,
+					   copy_bytes);
+
+			if (dest_page)
+				kunmap_local(kaddr);
+		}
+
+		offset += copy_bytes;
+		dest_offset += copy_bytes;
+		len -= copy_bytes;
+		copied += copy_bytes;
+
+		path->slots[0]++;
+		if (path->slots[0] >= btrfs_header_nritems(path->nodes[0])) {
+			/*
+			 * We've reached the last slot in this leaf and we need
+			 * to go to the next leaf.
+			 */
+			ret = btrfs_next_leaf(root, path);
+			if (ret < 0) {
+				break;
+			} else if (ret > 0) {
+				ret = 0;
+				break;
+			}
+		}
+	}
+out:
+	btrfs_free_path(path);
+	if (!ret)
+		ret = copied;
+	return ret;
+}
+
+/*
+ * Delete an fsverity orphan
+ *
+ * @trans:  transaction to do the delete in
+ * @inode:  inode to orphan
+ *
+ * Capture verity orphan specific logic that is repeated in the couple places
+ * we delete verity orphans. Specifically, handling ENOENT and ignoring inodes
+ * with 0 links.
+ *
+ * Returns zero on success or a negative error code on failure.
+ */
+static int del_orphan(struct btrfs_trans_handle *trans, struct btrfs_inode *inode)
+{
+	struct btrfs_root *root = inode->root;
+	int ret;
+
+	/*
+	 * If the inode has no links, it is either already unlinked, or was
+	 * created with O_TMPFILE. In either case, it should have an orphan from
+	 * that other operation. Rather than reference count the orphans, we
+	 * simply ignore them here, because we only invoke the verity path in
+	 * the orphan logic when i_nlink is 1.
+	 */
+	if (!inode->vfs_inode.i_nlink)
+		return 0;
+
+	ret = btrfs_del_orphan_item(trans, root, btrfs_ino(inode));
+	if (ret == -ENOENT)
+		ret = 0;
+	return ret;
+}
+
+/*
+ * Rollback in-progress verity if we encounter an error.
+ *
+ * @inode:  inode verity had an error for
+ *
+ * We try to handle recoverable errors while enabling verity by rolling it back
+ * and just failing the operation, rather than having an fs level error no
+ * matter what. However, any error in rollback is unrecoverable.
+ *
+ * Returns 0 on success, negative error code on failure.
+ */
+static int rollback_verity(struct btrfs_inode *inode)
+{
+	struct btrfs_trans_handle *trans = NULL;
+	struct btrfs_root *root = inode->root;
+	int ret;
+
+	ASSERT(inode_is_locked(&inode->vfs_inode));
+	truncate_inode_pages(inode->vfs_inode.i_mapping, inode->vfs_inode.i_size);
+	clear_bit(BTRFS_INODE_VERITY_IN_PROGRESS, &inode->runtime_flags);
+	ret = btrfs_drop_verity_items(inode);
+	if (ret) {
+		btrfs_handle_fs_error(root->fs_info, ret,
+				"failed to drop verity items in rollback %llu",
+				(u64)inode->vfs_inode.i_ino);
+		goto out;
+	}
+
+	/*
+	 * 1 for updating the inode flag
+	 * 1 for deleting the orphan
+	 */
+	trans = btrfs_start_transaction(root, 2);
+	if (IS_ERR(trans)) {
+		ret = PTR_ERR(trans);
+		trans = NULL;
+		btrfs_handle_fs_error(root->fs_info, ret,
+			"failed to start transaction in verity rollback %llu",
+			(u64)inode->vfs_inode.i_ino);
+		goto out;
+	}
+	inode->ro_flags &= ~BTRFS_INODE_RO_VERITY;
+	btrfs_sync_inode_flags_to_i_flags(&inode->vfs_inode);
+	ret = btrfs_update_inode(trans, root, inode);
+	if (ret) {
+		btrfs_abort_transaction(trans, ret);
+		goto out;
+	}
+	ret = del_orphan(trans, inode);
+	if (ret) {
+		btrfs_abort_transaction(trans, ret);
+		goto out;
+	}
+out:
+	if (trans)
+		btrfs_end_transaction(trans);
+	return ret;
+}
+
+/*
+ * Finalize making the file a valid verity file
+ *
+ * @inode:      inode to be marked as verity
+ * @desc:       contents of the verity descriptor to write (not NULL)
+ * @desc_size:  size of the verity descriptor
+ *
+ * Do the actual work of finalizing verity after successfully writing the Merkle
+ * tree:
+ *
+ * - write out the descriptor items
+ * - mark the inode with the verity flag
+ * - delete the orphan item
+ * - mark the ro compat bit
+ * - clear the in progress bit
+ *
+ * Returns 0 on success, negative error code on failure.
+ */
+static int finish_verity(struct btrfs_inode *inode, const void *desc,
+			 size_t desc_size)
+{
+	struct btrfs_trans_handle *trans = NULL;
+	struct btrfs_root *root = inode->root;
+	struct btrfs_verity_descriptor_item item;
+	int ret;
+
+	/* Write out the descriptor item */
+	memset(&item, 0, sizeof(item));
+	btrfs_set_stack_verity_descriptor_size(&item, desc_size);
+	ret = write_key_bytes(inode, BTRFS_VERITY_DESC_ITEM_KEY, 0,
+			      (const char *)&item, sizeof(item));
+	if (ret)
+		goto out;
+
+	/* Write out the descriptor itself */
+	ret = write_key_bytes(inode, BTRFS_VERITY_DESC_ITEM_KEY, 1,
+			      desc, desc_size);
+	if (ret)
+		goto out;
+
+	/*
+	 * 1 for updating the inode flag
+	 * 1 for deleting the orphan
+	 */
+	trans = btrfs_start_transaction(root, 2);
+	if (IS_ERR(trans)) {
+		ret = PTR_ERR(trans);
+		goto out;
+	}
+	inode->ro_flags |= BTRFS_INODE_RO_VERITY;
+	btrfs_sync_inode_flags_to_i_flags(&inode->vfs_inode);
+	ret = btrfs_update_inode(trans, root, inode);
+	if (ret)
+		goto end_trans;
+	ret = del_orphan(trans, inode);
+	if (ret)
+		goto end_trans;
+	clear_bit(BTRFS_INODE_VERITY_IN_PROGRESS, &inode->runtime_flags);
+	btrfs_set_fs_compat_ro(root->fs_info, VERITY);
+end_trans:
+	btrfs_end_transaction(trans);
+out:
+	return ret;
+
+}
+
+/*
+ * fsverity op that begins enabling verity.
+ *
+ * @filp:  file to enable verity on
+ *
+ * Begin enabling fsverity for the file. We drop any existing verity items, add
+ * an orphan and set the in progress bit.
+ *
+ * Returns 0 on success, negative error code on failure.
+ */
+static int btrfs_begin_enable_verity(struct file *filp)
+{
+	struct btrfs_inode *inode = BTRFS_I(file_inode(filp));
+	struct btrfs_root *root = inode->root;
+	struct btrfs_trans_handle *trans;
+	int ret;
+
+	ASSERT(inode_is_locked(file_inode(filp)));
+
+	if (test_bit(BTRFS_INODE_VERITY_IN_PROGRESS, &inode->runtime_flags))
+		return -EBUSY;
+
+	/*
+	 * This should almost never do anything, but theoretically, it's
+	 * possible that we failed to enable verity on a file, then were
+	 * interrupted or failed while rolling back, failed to cleanup the
+	 * orphan, and finally attempt to enable verity again.
+	 */
+	ret = btrfs_drop_verity_items(inode);
+	if (ret)
+		return ret;
+
+	/* 1 for the orphan item */
+	trans = btrfs_start_transaction(root, 1);
+	if (IS_ERR(trans))
+		return PTR_ERR(trans);
+
+	ret = btrfs_orphan_add(trans, inode);
+	if (!ret)
+		set_bit(BTRFS_INODE_VERITY_IN_PROGRESS, &inode->runtime_flags);
+	btrfs_end_transaction(trans);
+
+	return 0;
+}
+
+/*
+ * fsverity op that ends enabling verity.
+ *
+ * @filp:              file we are finishing enabling verity on
+ * @desc:              verity descriptor to write out (NULL in error conditions)
+ * @desc_size:         size of the verity descriptor (variable with signatures)
+ * @merkle_tree_size:  size of the merkle tree in bytes
+ *
+ * If desc is null, then VFS is signaling an error occurred during verity
+ * enable, and we should try to rollback. Otherwise, attempt to finish verity.
+ *
+ * Returns 0 on success, negative error code on error.
+ */
+static int btrfs_end_enable_verity(struct file *filp, const void *desc,
+				   size_t desc_size, u64 merkle_tree_size)
+{
+	struct btrfs_inode *inode = BTRFS_I(file_inode(filp));
+	int ret = 0;
+	int rollback_ret;
+
+	ASSERT(inode_is_locked(file_inode(filp)));
+
+	if (desc == NULL)
+		goto rollback;
+
+	ret = finish_verity(inode, desc, desc_size);
+	if (ret)
+		goto rollback;
+	return ret;
+
+rollback:
+	rollback_ret = rollback_verity(inode);
+	if (rollback_ret)
+		btrfs_err(inode->root->fs_info,
+			  "failed to rollback verity items: %d", rollback_ret);
+	return ret;
+}
+
+/*
+ * fsverity op that gets the struct fsverity_descriptor.
+ *
+ * @inode:     inode to get the descriptor of
+ * @buf:       output buffer for the descriptor contents
+ * @buf_size:  size of the output buffer. 0 to query the size
+ *
+ * fsverity does a two pass setup for reading the descriptor, in the first pass
+ * it calls with buf_size = 0 to query the size of the descriptor, and then in
+ * the second pass it actually reads the descriptor off disk.
+ *
+ * Returns the size on success or a negative error code on failure.
+ */
+int btrfs_get_verity_descriptor(struct inode *inode, void *buf, size_t buf_size)
+{
+	u64 true_size;
+	int ret = 0;
+	struct btrfs_verity_descriptor_item item;
+
+	memset(&item, 0, sizeof(item));
+	ret = read_key_bytes(BTRFS_I(inode), BTRFS_VERITY_DESC_ITEM_KEY, 0,
+			     (char *)&item, sizeof(item), NULL);
+	if (ret < 0)
+		return ret;
+
+	if (item.reserved[0] != 0 || item.reserved[1] != 0)
+		return -EUCLEAN;
+
+	true_size = btrfs_stack_verity_descriptor_size(&item);
+	if (true_size > INT_MAX)
+		return -EUCLEAN;
+
+	if (buf_size == 0)
+		return true_size;
+	if (buf_size < true_size)
+		return -ERANGE;
+
+	ret = read_key_bytes(BTRFS_I(inode), BTRFS_VERITY_DESC_ITEM_KEY, 1,
+			     buf, buf_size, NULL);
+	if (ret < 0)
+		return ret;
+	if (ret != true_size)
+		return -EIO;
+
+	return true_size;
+}
+
+/*
+ * fsverity op that reads and caches a merkle tree page.
+ *
+ * @inode:         inode to read a merkle tree page for
+ * @index:         page index relative to the start of the merkle tree
+ * @num_ra_pages:  number of pages to readahead. Optional, we ignore it
+ *
+ * The Merkle tree is stored in the filesystem btree, but its pages are cached
+ * with a logical position past EOF in the inode's mapping.
+ *
+ * Returns the page we read, or an ERR_PTR on error.
+ */
+static struct page *btrfs_read_merkle_tree_page(struct inode *inode,
+						pgoff_t index,
+						unsigned long num_ra_pages)
+{
+	struct page *page;
+	u64 off = (u64)index << PAGE_SHIFT;
+	loff_t merkle_pos = merkle_file_pos(inode);
+	int ret;
+
+	if (merkle_pos < 0)
+		return ERR_PTR(merkle_pos);
+	if (merkle_pos > inode->i_sb->s_maxbytes - off - PAGE_SIZE)
+		return ERR_PTR(-EFBIG);
+	index += merkle_pos >> PAGE_SHIFT;
+again:
+	page = find_get_page_flags(inode->i_mapping, index, FGP_ACCESSED);
+	if (page) {
+		if (PageUptodate(page))
+			return page;
+
+		lock_page(page);
+		/*
+		 * We only insert uptodate pages, so !Uptodate has to be
+		 * an error
+		 */
+		if (!PageUptodate(page)) {
+			unlock_page(page);
+			put_page(page);
+			return ERR_PTR(-EIO);
+		}
+		unlock_page(page);
+		return page;
+	}
+
+	page = __page_cache_alloc(mapping_gfp_constraint(inode->i_mapping, ~__GFP_FS));
+	if (!page)
+		return ERR_PTR(-ENOMEM);
+
+	/*
+	 * Merkle item keys are indexed from byte 0 in the merkle tree.
+	 * They have the form:
+	 *
+	 * [ inode objectid, BTRFS_MERKLE_ITEM_KEY, offset in bytes ]
+	 */
+	ret = read_key_bytes(BTRFS_I(inode), BTRFS_VERITY_MERKLE_ITEM_KEY, off,
+			     page_address(page), PAGE_SIZE, page);
+	if (ret < 0) {
+		put_page(page);
+		return ERR_PTR(ret);
+	}
+	if (ret < PAGE_SIZE)
+		memzero_page(page, ret, PAGE_SIZE - ret);
+
+	SetPageUptodate(page);
+	ret = add_to_page_cache_lru(page, inode->i_mapping, index, GFP_NOFS);
+
+	if (!ret) {
+		/* Inserted and ready for fsverity */
+		unlock_page(page);
+	} else {
+		put_page(page);
+		/* Did someone race us into inserting this page? */
+		if (ret == -EEXIST)
+			goto again;
+		page = ERR_PTR(ret);
+	}
+	return page;
+}
+
+/*
+ * fsverity op that writes a Merkle tree block into the btree.
+ *
+ * @inode:          inode to write a Merkle tree block for
+ * @buf:            Merkle tree data block to write
+ * @index:          index of the block in the Merkle tree
+ * @log_blocksize:  log base 2 of the Merkle tree block size
+ *
+ * Note that the block size could be different from the page size, so it is not
+ * safe to assume that index is a page index.
+ *
+ * Returns 0 on success or negative error code on failure
+ */
+static int btrfs_write_merkle_tree_block(struct inode *inode, const void *buf,
+					u64 index, int log_blocksize)
+{
+	u64 off = index << log_blocksize;
+	u64 len = 1ULL << log_blocksize;
+	loff_t merkle_pos = merkle_file_pos(inode);
+
+	if (merkle_pos < 0)
+		return merkle_pos;
+	if (merkle_pos > inode->i_sb->s_maxbytes - off - len)
+		return -EFBIG;
+
+	return write_key_bytes(BTRFS_I(inode), BTRFS_VERITY_MERKLE_ITEM_KEY,
+			       off, buf, len);
+}
+
+const struct fsverity_operations btrfs_verityops = {
+	.begin_enable_verity     = btrfs_begin_enable_verity,
+	.end_enable_verity       = btrfs_end_enable_verity,
+	.get_verity_descriptor   = btrfs_get_verity_descriptor,
+	.read_merkle_tree_page   = btrfs_read_merkle_tree_page,
+	.write_merkle_tree_block = btrfs_write_merkle_tree_block,
+};
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 9cfc668f91f4..635f45f1a2ef 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -4,9 +4,9 @@
  */
 
 #include <linux/sched.h>
+#include <linux/sched/mm.h>
 #include <linux/bio.h>
 #include <linux/slab.h>
-#include <linux/buffer_head.h>
 #include <linux/blkdev.h>
 #include <linux/ratelimit.h>
 #include <linux/kthread.h>
@@ -14,6 +14,7 @@
 #include <linux/semaphore.h>
 #include <linux/uuid.h>
 #include <linux/list_sort.h>
+#include <linux/namei.h>
 #include "misc.h"
 #include "ctree.h"
 #include "extent_map.h"
@@ -31,13 +32,20 @@
 #include "space-info.h"
 #include "block-group.h"
 #include "discard.h"
+#include "zoned.h"
+
+static struct bio_set btrfs_bioset;
+
+#define BTRFS_BLOCK_GROUP_STRIPE_MASK	(BTRFS_BLOCK_GROUP_RAID0 | \
+					 BTRFS_BLOCK_GROUP_RAID10 | \
+					 BTRFS_BLOCK_GROUP_RAID56_MASK)
 
 const struct btrfs_raid_attr btrfs_raid_array[BTRFS_NR_RAID_TYPES] = {
 	[BTRFS_RAID_RAID10] = {
 		.sub_stripes	= 2,
 		.dev_stripes	= 1,
 		.devs_max	= 0,	/* 0 == as many as possible */
-		.devs_min	= 4,
+		.devs_min	= 2,
 		.tolerated_failures = 1,
 		.devs_increment	= 2,
 		.ncopies	= 2,
@@ -102,7 +110,7 @@ const struct btrfs_raid_attr btrfs_raid_array[BTRFS_NR_RAID_TYPES] = {
 		.sub_stripes	= 1,
 		.dev_stripes	= 1,
 		.devs_max	= 0,
-		.devs_min	= 2,
+		.devs_min	= 1,
 		.tolerated_failures = 0,
 		.devs_increment	= 1,
 		.ncopies	= 1,
@@ -152,6 +160,20 @@ const struct btrfs_raid_attr btrfs_raid_array[BTRFS_NR_RAID_TYPES] = {
 	},
 };
 
+/*
+ * Convert block group flags (BTRFS_BLOCK_GROUP_*) to btrfs_raid_types, which
+ * can be used as index to access btrfs_raid_array[].
+ */
+enum btrfs_raid_types __attribute_const__ btrfs_bg_flags_to_raid_index(u64 flags)
+{
+	const u64 profile = (flags & BTRFS_BLOCK_GROUP_PROFILE_MASK);
+
+	if (!profile)
+		return BTRFS_RAID_SINGLE;
+
+	return BTRFS_BG_FLAG_TO_INDEX(profile);
+}
+
 const char *btrfs_bg_type_to_raid_name(u64 flags)
 {
 	const int index = btrfs_bg_flags_to_raid_index(flags);
@@ -162,6 +184,13 @@ const char *btrfs_bg_type_to_raid_name(u64 flags)
 	return btrfs_raid_array[index].raid_name;
 }
 
+int btrfs_nr_parity_stripes(u64 type)
+{
+	enum btrfs_raid_types index = btrfs_bg_flags_to_raid_index(type);
+
+	return btrfs_raid_array[index].nparity;
+}
+
 /*
  * Fill @buf with textual description of @bg_flags, no more than @size_buf
  * bytes including terminating null byte.
@@ -218,13 +247,12 @@ out_overflow:;
 
 static int init_first_rw_device(struct btrfs_trans_handle *trans);
 static int btrfs_relocate_sys_chunks(struct btrfs_fs_info *fs_info);
-static void btrfs_dev_stat_print_on_error(struct btrfs_device *dev);
 static void btrfs_dev_stat_print_on_load(struct btrfs_device *device);
 static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
-			     enum btrfs_map_op op,
-			     u64 logical, u64 *length,
-			     struct btrfs_bio **bbio_ret,
-			     int mirror_num, int need_raid_map);
+			     enum btrfs_map_op op, u64 logical, u64 *length,
+			     struct btrfs_io_context **bioc_ret,
+			     struct btrfs_io_stripe *smap,
+			     int *mirror_num_ret, int need_raid_map);
 
 /*
  * Device locking
@@ -246,7 +274,9 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
  *
  * global::fs_devs - add, remove, updates to the global list
  *
- * does not protect: manipulation of the fs_devices::devices list!
+ * does not protect: manipulation of the fs_devices::devices list in general
+ * but in mount context it could be used to exclude list modifications by eg.
+ * scan ioctl
  *
  * btrfs_device::name - renames (write side), read is RCU
  *
@@ -259,6 +289,9 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
  * may be used to exclude some operations from running concurrently without any
  * modifications to the list (see write_all_supers)
  *
+ * Is not required at mount and close times, because our device list is
+ * protected by the uuid_mutex at that point.
+ *
  * balance_mutex
  * -------------
  * protects balance structures (status, state) and context accessed from
@@ -281,14 +314,13 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
  * ============
  *
  * uuid_mutex
- *   volume_mutex
- *     device_list_mutex
- *       chunk_mutex
- *     balance_mutex
+ *   device_list_mutex
+ *     chunk_mutex
+ *   balance_mutex
  *
  *
- * Exclusive operations, BTRFS_FS_EXCL_OP
- * ======================================
+ * Exclusive operations
+ * ====================
  *
  * Maintains the exclusivity of the following operations that apply to the
  * whole filesystem and cannot run in parallel.
@@ -314,11 +346,11 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
  * - system power-cycle and filesystem mounted as read-only
  * - filesystem or device errors leading to forced read-only
  *
- * BTRFS_FS_EXCL_OP flag is set and cleared using atomic operations.
- * During the course of Paused state, the BTRFS_FS_EXCL_OP remains set.
+ * The status of exclusive operation is set and cleared atomically.
+ * During the course of Paused state, fs_info::exclusive_operation remains set.
  * A device operation in Paused or Running state can be canceled or resumed
  * either by ioctl (Balance only) or when remounted as read-write.
- * BTRFS_FS_EXCL_OP flag is cleared when the device operation is canceled or
+ * The exclusive status is cleared when the device operation is canceled or
  * completed.
  */
 
@@ -352,6 +384,7 @@ static struct btrfs_fs_devices *alloc_fs_devices(const u8 *fsid,
 	INIT_LIST_HEAD(&fs_devs->devices);
 	INIT_LIST_HEAD(&fs_devs->alloc_list);
 	INIT_LIST_HEAD(&fs_devs->fs_list);
+	INIT_LIST_HEAD(&fs_devs->seed_list);
 	if (fsid)
 		memcpy(fs_devs->fsid, fsid, BTRFS_FSID_SIZE);
 
@@ -368,7 +401,7 @@ void btrfs_free_device(struct btrfs_device *device)
 	WARN_ON(!list_empty(&device->post_commit_list));
 	rcu_string_free(device->name);
 	extent_io_tree_release(&device->alloc_state);
-	bio_put(device->flush_bio);
+	btrfs_destroy_dev_zone_info(device);
 	kfree(device);
 }
 
@@ -397,43 +430,6 @@ void __exit btrfs_cleanup_fs_uuids(void)
 	}
 }
 
-/*
- * Returns a pointer to a new btrfs_device on success; ERR_PTR() on error.
- * Returned struct is not linked onto any lists and must be destroyed using
- * btrfs_free_device.
- */
-static struct btrfs_device *__alloc_device(void)
-{
-	struct btrfs_device *dev;
-
-	dev = kzalloc(sizeof(*dev), GFP_KERNEL);
-	if (!dev)
-		return ERR_PTR(-ENOMEM);
-
-	/*
-	 * Preallocate a bio that's always going to be used for flushing device
-	 * barriers and matches the device lifespan
-	 */
-	dev->flush_bio = bio_alloc_bioset(GFP_KERNEL, 0, NULL);
-	if (!dev->flush_bio) {
-		kfree(dev);
-		return ERR_PTR(-ENOMEM);
-	}
-
-	INIT_LIST_HEAD(&dev->dev_list);
-	INIT_LIST_HEAD(&dev->dev_alloc_list);
-	INIT_LIST_HEAD(&dev->post_commit_list);
-
-	atomic_set(&dev->reada_in_flight, 0);
-	atomic_set(&dev->dev_stats_ccnt, 0);
-	btrfs_device_data_ordered_init(dev);
-	INIT_RADIX_TREE(&dev->reada_zones, GFP_NOFS & ~__GFP_DIRECT_RECLAIM);
-	INIT_RADIX_TREE(&dev->reada_extents, GFP_NOFS & ~__GFP_DIRECT_RECLAIM);
-	extent_io_tree_init(NULL, &dev->alloc_state, 0, NULL);
-
-	return dev;
-}
-
 static noinline struct btrfs_fs_devices *find_fsid(
 		const u8 *fsid, const u8 *metadata_fsid)
 {
@@ -500,7 +496,7 @@ static struct btrfs_fs_devices *find_fsid_with_metadata_uuid(
 static int
 btrfs_get_bdev_and_sb(const char *device_path, fmode_t flags, void *holder,
 		      int flush, struct block_device **bdev,
-		      struct buffer_head **bh)
+		      struct btrfs_super_block **disk_super)
 {
 	int ret;
 
@@ -512,16 +508,16 @@ btrfs_get_bdev_and_sb(const char *device_path, fmode_t flags, void *holder,
 	}
 
 	if (flush)
-		filemap_write_and_wait((*bdev)->bd_inode->i_mapping);
+		sync_blockdev(*bdev);
 	ret = set_blocksize(*bdev, BTRFS_BDEV_BLOCKSIZE);
 	if (ret) {
 		blkdev_put(*bdev, flags);
 		goto error;
 	}
 	invalidate_bdev(*bdev);
-	*bh = btrfs_read_dev_super(*bdev);
-	if (IS_ERR(*bh)) {
-		ret = PTR_ERR(*bh);
+	*disk_super = btrfs_read_dev_super(*bdev);
+	if (IS_ERR(*disk_super)) {
+		ret = PTR_ERR(*disk_super);
 		blkdev_put(*bdev, flags);
 		goto error;
 	}
@@ -530,40 +526,31 @@ btrfs_get_bdev_and_sb(const char *device_path, fmode_t flags, void *holder,
 
 error:
 	*bdev = NULL;
-	*bh = NULL;
 	return ret;
 }
 
-static bool device_path_matched(const char *path, struct btrfs_device *device)
-{
-	int found;
-
-	rcu_read_lock();
-	found = strcmp(rcu_str_deref(device->name), path);
-	rcu_read_unlock();
-
-	return found == 0;
-}
-
-/*
- *  Search and remove all stale (devices which are not mounted) devices.
+/**
+ *  Search and remove all stale devices (which are not mounted).
  *  When both inputs are NULL, it will search and release all stale devices.
- *  path:	Optional. When provided will it release all unmounted devices
- *		matching this path only.
- *  skip_dev:	Optional. Will skip this device when searching for the stale
+ *
+ *  @devt:	Optional. When provided will it release all unmounted devices
+ *		matching this devt only.
+ *  @skip_device:  Optional. Will skip this device when searching for the stale
  *		devices.
- *  Return:	0 for success or if @path is NULL.
- * 		-EBUSY if @path is a mounted device.
- * 		-ENOENT if @path does not match any device in the list.
+ *
+ *  Return:	0 for success or if @devt is 0.
+ *		-EBUSY if @devt is a mounted device.
+ *		-ENOENT if @devt does not match any device in the list.
  */
-static int btrfs_free_stale_devices(const char *path,
-				     struct btrfs_device *skip_device)
+static int btrfs_free_stale_devices(dev_t devt, struct btrfs_device *skip_device)
 {
 	struct btrfs_fs_devices *fs_devices, *tmp_fs_devices;
 	struct btrfs_device *device, *tmp_device;
 	int ret = 0;
 
-	if (path)
+	lockdep_assert_held(&uuid_mutex);
+
+	if (devt)
 		ret = -ENOENT;
 
 	list_for_each_entry_safe(fs_devices, tmp_fs_devices, &fs_uuids, fs_list) {
@@ -573,13 +560,11 @@ static int btrfs_free_stale_devices(const char *path,
 					 &fs_devices->devices, dev_list) {
 			if (skip_device && skip_device == device)
 				continue;
-			if (path && !device->name)
-				continue;
-			if (path && !device_path_matched(path, device))
+			if (devt && devt != device->devt)
 				continue;
 			if (fs_devices->opened) {
 				/* for an already deleted device return 0 */
-				if (path && ret != 0)
+				if (devt && ret != 0)
 					ret = -EBUSY;
 				break;
 			}
@@ -590,8 +575,6 @@ static int btrfs_free_stale_devices(const char *path,
 			btrfs_free_device(device);
 
 			ret = 0;
-			if (fs_devices->num_devices == 0)
-				break;
 		}
 		mutex_unlock(&fs_devices->device_list_mutex);
 
@@ -605,13 +588,16 @@ static int btrfs_free_stale_devices(const char *path,
 	return ret;
 }
 
+/*
+ * This is only used on mount, and we are protected from competing things
+ * messing with our fs_devices by the uuid_mutex, thus we do not need the
+ * fs_devices->device_list_mutex here.
+ */
 static int btrfs_open_one_device(struct btrfs_fs_devices *fs_devices,
 			struct btrfs_device *device, fmode_t flags,
 			void *holder)
 {
-	struct request_queue *q;
 	struct block_device *bdev;
-	struct buffer_head *bh;
 	struct btrfs_super_block *disk_super;
 	u64 devid;
 	int ret;
@@ -622,17 +608,16 @@ static int btrfs_open_one_device(struct btrfs_fs_devices *fs_devices,
 		return -EINVAL;
 
 	ret = btrfs_get_bdev_and_sb(device->name->str, flags, holder, 1,
-				    &bdev, &bh);
+				    &bdev, &disk_super);
 	if (ret)
 		return ret;
 
-	disk_super = (struct btrfs_super_block *)bh->b_data;
 	devid = btrfs_stack_device_id(&disk_super->dev_item);
 	if (devid != device->devid)
-		goto error_brelse;
+		goto error_free_page;
 
 	if (memcmp(device->uuid, disk_super->dev_item.uuid, BTRFS_UUID_SIZE))
-		goto error_brelse;
+		goto error_free_page;
 
 	device->generation = btrfs_super_generation(disk_super);
 
@@ -641,7 +626,7 @@ static int btrfs_open_one_device(struct btrfs_fs_devices *fs_devices,
 		    BTRFS_FEATURE_INCOMPAT_METADATA_UUID) {
 			pr_err(
 		"BTRFS: Invalid seeding and uuid-changed device detected\n");
-			goto error_brelse;
+			goto error_free_page;
 		}
 
 		clear_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state);
@@ -653,8 +638,7 @@ static int btrfs_open_one_device(struct btrfs_fs_devices *fs_devices,
 			set_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state);
 	}
 
-	q = bdev_get_queue(bdev);
-	if (!blk_queue_nonrot(q))
+	if (!bdev_nonrot(bdev))
 		fs_devices->rotating = true;
 
 	device->bdev = bdev;
@@ -667,12 +651,12 @@ static int btrfs_open_one_device(struct btrfs_fs_devices *fs_devices,
 		fs_devices->rw_devices++;
 		list_add_tail(&device->dev_alloc_list, &fs_devices->alloc_list);
 	}
-	brelse(bh);
+	btrfs_release_disk_super(disk_super);
 
 	return 0;
 
-error_brelse:
-	brelse(bh);
+error_free_page:
+	btrfs_release_disk_super(disk_super);
 	blkdev_put(bdev, flags);
 
 	return -EINVAL;
@@ -709,7 +693,7 @@ static struct btrfs_fs_devices *find_fsid_changed(
 
 	/*
 	 * Handles the case where scanned device is part of an fs that had
-	 * multiple successful changes of FSID but curently device didn't
+	 * multiple successful changes of FSID but currently device didn't
 	 * observe it. Meaning our fsid will be different than theirs. We need
 	 * to handle two subcases :
 	 *  1 - The fs still continues to have different METADATA/FSID uuids.
@@ -778,11 +762,17 @@ static noinline struct btrfs_device *device_list_add(const char *path,
 	struct rcu_string *name;
 	u64 found_transid = btrfs_super_generation(disk_super);
 	u64 devid = btrfs_stack_device_id(&disk_super->dev_item);
+	dev_t path_devt;
+	int error;
 	bool has_metadata_uuid = (btrfs_super_incompat_flags(disk_super) &
 		BTRFS_FEATURE_INCOMPAT_METADATA_UUID);
 	bool fsid_change_in_progress = (btrfs_super_flags(disk_super) &
 					BTRFS_SUPER_FLAG_CHANGING_FSID_V2);
 
+	error = lookup_bdev(path, &path_devt);
+	if (error)
+		return ERR_PTR(error);
+
 	if (fsid_change_in_progress) {
 		if (!has_metadata_uuid)
 			fs_devices = find_fsid_inprogress(disk_super);
@@ -814,9 +804,13 @@ static noinline struct btrfs_device *device_list_add(const char *path,
 
 		device = NULL;
 	} else {
+		struct btrfs_dev_lookup_args args = {
+			.devid = devid,
+			.uuid = disk_super->dev_item.uuid,
+		};
+
 		mutex_lock(&fs_devices->device_list_mutex);
-		device = btrfs_find_device(fs_devices, devid,
-				disk_super->dev_item.uuid, NULL, false);
+		device = btrfs_find_device(fs_devices, &args);
 
 		/*
 		 * If this disk has been pulled into an fs devices created by
@@ -861,6 +855,7 @@ static noinline struct btrfs_device *device_list_add(const char *path,
 			return ERR_PTR(-ENOMEM);
 		}
 		rcu_assign_pointer(device->name, name);
+		device->devt = path_devt;
 
 		list_add_rcu(&device->dev_list, &fs_devices->devices);
 		fs_devices->num_devices++;
@@ -921,30 +916,27 @@ static noinline struct btrfs_device *device_list_add(const char *path,
 		/*
 		 * We are going to replace the device path for a given devid,
 		 * make sure it's the same device if the device is mounted
+		 *
+		 * NOTE: the device->fs_info may not be reliable here so pass
+		 * in a NULL to message helpers instead. This avoids a possible
+		 * use-after-free when the fs_info and fs_info->sb are already
+		 * torn down.
 		 */
 		if (device->bdev) {
-			struct block_device *path_bdev;
-
-			path_bdev = lookup_bdev(path);
-			if (IS_ERR(path_bdev)) {
+			if (device->devt != path_devt) {
 				mutex_unlock(&fs_devices->device_list_mutex);
-				return ERR_CAST(path_bdev);
-			}
-
-			if (device->bdev != path_bdev) {
-				bdput(path_bdev);
-				mutex_unlock(&fs_devices->device_list_mutex);
-				btrfs_warn_in_rcu(device->fs_info,
-			"duplicate device fsid:devid for %pU:%llu old:%s new:%s",
-					disk_super->fsid, devid,
-					rcu_str_deref(device->name), path);
+				btrfs_warn_in_rcu(NULL,
+	"duplicate device %s devid %llu generation %llu scanned by %s (%d)",
+						  path, devid, found_transid,
+						  current->comm,
+						  task_pid_nr(current));
 				return ERR_PTR(-EEXIST);
 			}
-			bdput(path_bdev);
-			btrfs_info_in_rcu(device->fs_info,
-				"device fsid %pU devid %llu moved old:%s new:%s",
-				disk_super->fsid, devid,
-				rcu_str_deref(device->name), path);
+			btrfs_info_in_rcu(NULL,
+	"devid %llu device path %s changed to %s scanned by %s (%d)",
+					  devid, rcu_str_deref(device->name),
+					  path, current->comm,
+					  task_pid_nr(current));
 		}
 
 		name = rcu_string_strdup(path, GFP_NOFS);
@@ -958,6 +950,7 @@ static noinline struct btrfs_device *device_list_add(const char *path,
 			fs_devices->missing_devices--;
 			clear_bit(BTRFS_DEV_STATE_MISSING, &device->dev_state);
 		}
+		device->devt = path_devt;
 	}
 
 	/*
@@ -985,11 +978,12 @@ static struct btrfs_fs_devices *clone_fs_devices(struct btrfs_fs_devices *orig)
 	struct btrfs_device *orig_dev;
 	int ret = 0;
 
+	lockdep_assert_held(&uuid_mutex);
+
 	fs_devices = alloc_fs_devices(orig->fsid, NULL);
 	if (IS_ERR(fs_devices))
 		return fs_devices;
 
-	mutex_lock(&orig->device_list_mutex);
 	fs_devices->total_devices = orig->total_devices;
 
 	list_for_each_entry(orig_dev, &orig->devices, dev_list) {
@@ -1017,58 +1011,54 @@ static struct btrfs_fs_devices *clone_fs_devices(struct btrfs_fs_devices *orig)
 			rcu_assign_pointer(device->name, name);
 		}
 
+		if (orig_dev->zone_info) {
+			struct btrfs_zoned_device_info *zone_info;
+
+			zone_info = btrfs_clone_dev_zone_info(orig_dev);
+			if (!zone_info) {
+				btrfs_free_device(device);
+				ret = -ENOMEM;
+				goto error;
+			}
+			device->zone_info = zone_info;
+		}
+
 		list_add(&device->dev_list, &fs_devices->devices);
 		device->fs_devices = fs_devices;
 		fs_devices->num_devices++;
 	}
-	mutex_unlock(&orig->device_list_mutex);
 	return fs_devices;
 error:
-	mutex_unlock(&orig->device_list_mutex);
 	free_fs_devices(fs_devices);
 	return ERR_PTR(ret);
 }
 
-/*
- * After we have read the system tree and know devids belonging to
- * this filesystem, remove the device which does not belong there.
- */
-void btrfs_free_extra_devids(struct btrfs_fs_devices *fs_devices, int step)
+static void __btrfs_free_extra_devids(struct btrfs_fs_devices *fs_devices,
+				      struct btrfs_device **latest_dev)
 {
 	struct btrfs_device *device, *next;
-	struct btrfs_device *latest_dev = NULL;
 
-	mutex_lock(&uuid_mutex);
-again:
 	/* This is the initialized path, it is safe to release the devices. */
 	list_for_each_entry_safe(device, next, &fs_devices->devices, dev_list) {
-		if (test_bit(BTRFS_DEV_STATE_IN_FS_METADATA,
-							&device->dev_state)) {
+		if (test_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &device->dev_state)) {
 			if (!test_bit(BTRFS_DEV_STATE_REPLACE_TGT,
-			     &device->dev_state) &&
-			     (!latest_dev ||
-			      device->generation > latest_dev->generation)) {
-				latest_dev = device;
+				      &device->dev_state) &&
+			    !test_bit(BTRFS_DEV_STATE_MISSING,
+				      &device->dev_state) &&
+			    (!*latest_dev ||
+			     device->generation > (*latest_dev)->generation)) {
+				*latest_dev = device;
 			}
 			continue;
 		}
 
-		if (device->devid == BTRFS_DEV_REPLACE_DEVID) {
-			/*
-			 * In the first step, keep the device which has
-			 * the correct fsid and the devid that is used
-			 * for the dev_replace procedure.
-			 * In the second step, the dev_replace state is
-			 * read from the device tree and it is known
-			 * whether the procedure is really active or
-			 * not, which means whether this device is
-			 * used or whether it should be removed.
-			 */
-			if (step == 0 || test_bit(BTRFS_DEV_STATE_REPLACE_TGT,
-						  &device->dev_state)) {
-				continue;
-			}
-		}
+		/*
+		 * We have already validated the presence of BTRFS_DEV_REPLACE_DEVID,
+		 * in btrfs_init_dev_replace() so just continue.
+		 */
+		if (device->devid == BTRFS_DEV_REPLACE_DEVID)
+			continue;
+
 		if (device->bdev) {
 			blkdev_put(device->bdev, device->mode);
 			device->bdev = NULL;
@@ -1077,21 +1067,31 @@ again:
 		if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state)) {
 			list_del_init(&device->dev_alloc_list);
 			clear_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state);
-			if (!test_bit(BTRFS_DEV_STATE_REPLACE_TGT,
-				      &device->dev_state))
-				fs_devices->rw_devices--;
+			fs_devices->rw_devices--;
 		}
 		list_del_init(&device->dev_list);
 		fs_devices->num_devices--;
 		btrfs_free_device(device);
 	}
 
-	if (fs_devices->seed) {
-		fs_devices = fs_devices->seed;
-		goto again;
-	}
+}
+
+/*
+ * After we have read the system tree and know devids belonging to this
+ * filesystem, remove the device which does not belong there.
+ */
+void btrfs_free_extra_devids(struct btrfs_fs_devices *fs_devices)
+{
+	struct btrfs_device *latest_dev = NULL;
+	struct btrfs_fs_devices *seed_dev;
+
+	mutex_lock(&uuid_mutex);
+	__btrfs_free_extra_devids(fs_devices, &latest_dev);
+
+	list_for_each_entry(seed_dev, &fs_devices->seed_list, seed_list)
+		__btrfs_free_extra_devids(seed_dev, &latest_dev);
 
-	fs_devices->latest_bdev = latest_dev->bdev;
+	fs_devices->latest_dev = latest_dev;
 
 	mutex_unlock(&uuid_mutex);
 }
@@ -1119,8 +1119,13 @@ static void btrfs_close_one_device(struct btrfs_device *device)
 		fs_devices->rw_devices--;
 	}
 
-	if (test_bit(BTRFS_DEV_STATE_MISSING, &device->dev_state))
+	if (device->devid == BTRFS_DEV_REPLACE_DEVID)
+		clear_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state);
+
+	if (test_bit(BTRFS_DEV_STATE_MISSING, &device->dev_state)) {
+		clear_bit(BTRFS_DEV_STATE_MISSING, &device->dev_state);
 		fs_devices->missing_devices--;
+	}
 
 	btrfs_close_bdev(device);
 	if (device->bdev) {
@@ -1128,60 +1133,67 @@ static void btrfs_close_one_device(struct btrfs_device *device)
 		device->bdev = NULL;
 	}
 	clear_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state);
+	btrfs_destroy_dev_zone_info(device);
 
 	device->fs_info = NULL;
 	atomic_set(&device->dev_stats_ccnt, 0);
 	extent_io_tree_release(&device->alloc_state);
 
+	/*
+	 * Reset the flush error record. We might have a transient flush error
+	 * in this mount, and if so we aborted the current transaction and set
+	 * the fs to an error state, guaranteeing no super blocks can be further
+	 * committed. However that error might be transient and if we unmount the
+	 * filesystem and mount it again, we should allow the mount to succeed
+	 * (btrfs_check_rw_degradable() should not fail) - if after mounting the
+	 * filesystem again we still get flush errors, then we will again abort
+	 * any transaction and set the error state, guaranteeing no commits of
+	 * unsafe super blocks.
+	 */
+	device->last_flush_error = 0;
+
 	/* Verify the device is back in a pristine state  */
 	ASSERT(!test_bit(BTRFS_DEV_STATE_FLUSH_SENT, &device->dev_state));
 	ASSERT(!test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state));
 	ASSERT(list_empty(&device->dev_alloc_list));
 	ASSERT(list_empty(&device->post_commit_list));
-	ASSERT(atomic_read(&device->reada_in_flight) == 0);
 }
 
-static int close_fs_devices(struct btrfs_fs_devices *fs_devices)
+static void close_fs_devices(struct btrfs_fs_devices *fs_devices)
 {
 	struct btrfs_device *device, *tmp;
 
+	lockdep_assert_held(&uuid_mutex);
+
 	if (--fs_devices->opened > 0)
-		return 0;
+		return;
 
-	mutex_lock(&fs_devices->device_list_mutex);
-	list_for_each_entry_safe(device, tmp, &fs_devices->devices, dev_list) {
+	list_for_each_entry_safe(device, tmp, &fs_devices->devices, dev_list)
 		btrfs_close_one_device(device);
-	}
-	mutex_unlock(&fs_devices->device_list_mutex);
 
 	WARN_ON(fs_devices->open_devices);
 	WARN_ON(fs_devices->rw_devices);
 	fs_devices->opened = 0;
 	fs_devices->seeding = false;
-
-	return 0;
+	fs_devices->fs_info = NULL;
 }
 
-int btrfs_close_devices(struct btrfs_fs_devices *fs_devices)
+void btrfs_close_devices(struct btrfs_fs_devices *fs_devices)
 {
-	struct btrfs_fs_devices *seed_devices = NULL;
-	int ret;
+	LIST_HEAD(list);
+	struct btrfs_fs_devices *tmp;
 
 	mutex_lock(&uuid_mutex);
-	ret = close_fs_devices(fs_devices);
-	if (!fs_devices->opened) {
-		seed_devices = fs_devices->seed;
-		fs_devices->seed = NULL;
-	}
-	mutex_unlock(&uuid_mutex);
+	close_fs_devices(fs_devices);
+	if (!fs_devices->opened)
+		list_splice_init(&fs_devices->seed_list, &list);
 
-	while (seed_devices) {
-		fs_devices = seed_devices;
-		seed_devices = fs_devices->seed;
+	list_for_each_entry_safe(fs_devices, tmp, &list, seed_list) {
 		close_fs_devices(fs_devices);
+		list_del(&fs_devices->seed_list);
 		free_fs_devices(fs_devices);
 	}
-	return ret;
+	mutex_unlock(&uuid_mutex);
 }
 
 static int open_fs_devices(struct btrfs_fs_devices *fs_devices,
@@ -1189,33 +1201,40 @@ static int open_fs_devices(struct btrfs_fs_devices *fs_devices,
 {
 	struct btrfs_device *device;
 	struct btrfs_device *latest_dev = NULL;
-	int ret = 0;
+	struct btrfs_device *tmp_device;
 
 	flags |= FMODE_EXCL;
 
-	list_for_each_entry(device, &fs_devices->devices, dev_list) {
-		/* Just open everything we can; ignore failures here */
-		if (btrfs_open_one_device(fs_devices, device, flags, holder))
-			continue;
+	list_for_each_entry_safe(device, tmp_device, &fs_devices->devices,
+				 dev_list) {
+		int ret;
 
-		if (!latest_dev ||
-		    device->generation > latest_dev->generation)
+		ret = btrfs_open_one_device(fs_devices, device, flags, holder);
+		if (ret == 0 &&
+		    (!latest_dev || device->generation > latest_dev->generation)) {
 			latest_dev = device;
+		} else if (ret == -ENODATA) {
+			fs_devices->num_devices--;
+			list_del(&device->dev_list);
+			btrfs_free_device(device);
+		}
 	}
-	if (fs_devices->open_devices == 0) {
-		ret = -EINVAL;
-		goto out;
-	}
+	if (fs_devices->open_devices == 0)
+		return -EINVAL;
+
 	fs_devices->opened = 1;
-	fs_devices->latest_bdev = latest_dev->bdev;
+	fs_devices->latest_dev = latest_dev;
 	fs_devices->total_rw_bytes = 0;
-out:
-	return ret;
+	fs_devices->chunk_alloc_policy = BTRFS_CHUNK_ALLOC_REGULAR;
+	fs_devices->read_policy = BTRFS_READ_POLICY_PID;
+
+	return 0;
 }
 
-static int devid_cmp(void *priv, struct list_head *a, struct list_head *b)
+static int devid_cmp(void *priv, const struct list_head *a,
+		     const struct list_head *b)
 {
-	struct btrfs_device *dev1, *dev2;
+	const struct btrfs_device *dev1, *dev2;
 
 	dev1 = list_entry(a, struct btrfs_device, dev_list);
 	dev2 = list_entry(b, struct btrfs_device, dev_list);
@@ -1233,8 +1252,14 @@ int btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
 	int ret;
 
 	lockdep_assert_held(&uuid_mutex);
+	/*
+	 * The device_list_mutex cannot be taken here in case opening the
+	 * underlying device takes further locks like open_mutex.
+	 *
+	 * We also don't need the lock here as this is called during mount and
+	 * exclusion is provided by uuid_mutex
+	 */
 
-	mutex_lock(&fs_devices->device_list_mutex);
 	if (fs_devices->opened) {
 		fs_devices->opened++;
 		ret = 0;
@@ -1242,68 +1267,67 @@ int btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
 		list_sort(NULL, &fs_devices->devices, devid_cmp);
 		ret = open_fs_devices(fs_devices, flags, holder);
 	}
-	mutex_unlock(&fs_devices->device_list_mutex);
 
 	return ret;
 }
 
-static void btrfs_release_disk_super(struct page *page)
+void btrfs_release_disk_super(struct btrfs_super_block *super)
 {
-	kunmap(page);
+	struct page *page = virt_to_page(super);
+
 	put_page(page);
 }
 
-static int btrfs_read_disk_super(struct block_device *bdev, u64 bytenr,
-				 struct page **page,
-				 struct btrfs_super_block **disk_super)
+static struct btrfs_super_block *btrfs_read_disk_super(struct block_device *bdev,
+						       u64 bytenr, u64 bytenr_orig)
 {
+	struct btrfs_super_block *disk_super;
+	struct page *page;
 	void *p;
 	pgoff_t index;
 
 	/* make sure our super fits in the device */
-	if (bytenr + PAGE_SIZE >= i_size_read(bdev->bd_inode))
-		return 1;
+	if (bytenr + PAGE_SIZE >= bdev_nr_bytes(bdev))
+		return ERR_PTR(-EINVAL);
 
 	/* make sure our super fits in the page */
-	if (sizeof(**disk_super) > PAGE_SIZE)
-		return 1;
+	if (sizeof(*disk_super) > PAGE_SIZE)
+		return ERR_PTR(-EINVAL);
 
 	/* make sure our super doesn't straddle pages on disk */
 	index = bytenr >> PAGE_SHIFT;
-	if ((bytenr + sizeof(**disk_super) - 1) >> PAGE_SHIFT != index)
-		return 1;
+	if ((bytenr + sizeof(*disk_super) - 1) >> PAGE_SHIFT != index)
+		return ERR_PTR(-EINVAL);
 
 	/* pull in the page with our super */
-	*page = read_cache_page_gfp(bdev->bd_inode->i_mapping,
-				   index, GFP_KERNEL);
+	page = read_cache_page_gfp(bdev->bd_inode->i_mapping, index, GFP_KERNEL);
 
-	if (IS_ERR_OR_NULL(*page))
-		return 1;
+	if (IS_ERR(page))
+		return ERR_CAST(page);
 
-	p = kmap(*page);
+	p = page_address(page);
 
 	/* align our pointer to the offset of the super block */
-	*disk_super = p + offset_in_page(bytenr);
+	disk_super = p + offset_in_page(bytenr);
 
-	if (btrfs_super_bytenr(*disk_super) != bytenr ||
-	    btrfs_super_magic(*disk_super) != BTRFS_MAGIC) {
-		btrfs_release_disk_super(*page);
-		return 1;
+	if (btrfs_super_bytenr(disk_super) != bytenr_orig ||
+	    btrfs_super_magic(disk_super) != BTRFS_MAGIC) {
+		btrfs_release_disk_super(p);
+		return ERR_PTR(-EINVAL);
 	}
 
-	if ((*disk_super)->label[0] &&
-		(*disk_super)->label[BTRFS_LABEL_SIZE - 1])
-		(*disk_super)->label[BTRFS_LABEL_SIZE - 1] = '\0';
+	if (disk_super->label[0] && disk_super->label[BTRFS_LABEL_SIZE - 1])
+		disk_super->label[BTRFS_LABEL_SIZE - 1] = 0;
 
-	return 0;
+	return disk_super;
 }
 
-int btrfs_forget_devices(const char *path)
+int btrfs_forget_devices(dev_t devt)
 {
 	int ret;
 
 	mutex_lock(&uuid_mutex);
-	ret = btrfs_free_stale_devices(strlen(path) ? path : NULL, NULL);
+	ret = btrfs_free_stale_devices(devt, NULL);
 	mutex_unlock(&uuid_mutex);
 
 	return ret;
@@ -1321,8 +1345,8 @@ struct btrfs_device *btrfs_scan_one_device(const char *path, fmode_t flags,
 	bool new_device_added = false;
 	struct btrfs_device *device = NULL;
 	struct block_device *bdev;
-	struct page *page;
-	u64 bytenr;
+	u64 bytenr, bytenr_orig;
+	int ret;
 
 	lockdep_assert_held(&uuid_mutex);
 
@@ -1332,25 +1356,30 @@ struct btrfs_device *btrfs_scan_one_device(const char *path, fmode_t flags,
 	 * So, we need to add a special mount option to scan for
 	 * later supers, using BTRFS_SUPER_MIRROR_MAX instead
 	 */
-	bytenr = btrfs_sb_offset(0);
 	flags |= FMODE_EXCL;
 
 	bdev = blkdev_get_by_path(path, flags, holder);
 	if (IS_ERR(bdev))
 		return ERR_CAST(bdev);
 
-	if (btrfs_read_disk_super(bdev, bytenr, &page, &disk_super)) {
-		device = ERR_PTR(-EINVAL);
+	bytenr_orig = btrfs_sb_offset(0);
+	ret = btrfs_sb_log_location_bdev(bdev, 0, READ, &bytenr);
+	if (ret) {
+		device = ERR_PTR(ret);
 		goto error_bdev_put;
 	}
 
-	device = device_list_add(path, disk_super, &new_device_added);
-	if (!IS_ERR(device)) {
-		if (new_device_added)
-			btrfs_free_stale_devices(path, device);
+	disk_super = btrfs_read_disk_super(bdev, bytenr, bytenr_orig);
+	if (IS_ERR(disk_super)) {
+		device = ERR_CAST(disk_super);
+		goto error_bdev_put;
 	}
 
-	btrfs_release_disk_super(page);
+	device = device_list_add(path, disk_super, &new_device_added);
+	if (!IS_ERR(device) && new_device_added)
+		btrfs_free_stale_devices(device->devt, device);
+
+	btrfs_release_disk_super(disk_super);
 
 error_bdev_put:
 	blkdev_put(bdev, flags);
@@ -1383,6 +1412,120 @@ static bool contains_pending_extent(struct btrfs_device *device, u64 *start,
 	return false;
 }
 
+static u64 dev_extent_search_start(struct btrfs_device *device, u64 start)
+{
+	switch (device->fs_devices->chunk_alloc_policy) {
+	case BTRFS_CHUNK_ALLOC_REGULAR:
+		return max_t(u64, start, BTRFS_DEVICE_RANGE_RESERVED);
+	case BTRFS_CHUNK_ALLOC_ZONED:
+		/*
+		 * We don't care about the starting region like regular
+		 * allocator, because we anyway use/reserve the first two zones
+		 * for superblock logging.
+		 */
+		return ALIGN(start, device->zone_info->zone_size);
+	default:
+		BUG();
+	}
+}
+
+static bool dev_extent_hole_check_zoned(struct btrfs_device *device,
+					u64 *hole_start, u64 *hole_size,
+					u64 num_bytes)
+{
+	u64 zone_size = device->zone_info->zone_size;
+	u64 pos;
+	int ret;
+	bool changed = false;
+
+	ASSERT(IS_ALIGNED(*hole_start, zone_size));
+
+	while (*hole_size > 0) {
+		pos = btrfs_find_allocatable_zones(device, *hole_start,
+						   *hole_start + *hole_size,
+						   num_bytes);
+		if (pos != *hole_start) {
+			*hole_size = *hole_start + *hole_size - pos;
+			*hole_start = pos;
+			changed = true;
+			if (*hole_size < num_bytes)
+				break;
+		}
+
+		ret = btrfs_ensure_empty_zones(device, pos, num_bytes);
+
+		/* Range is ensured to be empty */
+		if (!ret)
+			return changed;
+
+		/* Given hole range was invalid (outside of device) */
+		if (ret == -ERANGE) {
+			*hole_start += *hole_size;
+			*hole_size = 0;
+			return true;
+		}
+
+		*hole_start += zone_size;
+		*hole_size -= zone_size;
+		changed = true;
+	}
+
+	return changed;
+}
+
+/**
+ * dev_extent_hole_check - check if specified hole is suitable for allocation
+ * @device:	the device which we have the hole
+ * @hole_start: starting position of the hole
+ * @hole_size:	the size of the hole
+ * @num_bytes:	the size of the free space that we need
+ *
+ * This function may modify @hole_start and @hole_size to reflect the suitable
+ * position for allocation. Returns 1 if hole position is updated, 0 otherwise.
+ */
+static bool dev_extent_hole_check(struct btrfs_device *device, u64 *hole_start,
+				  u64 *hole_size, u64 num_bytes)
+{
+	bool changed = false;
+	u64 hole_end = *hole_start + *hole_size;
+
+	for (;;) {
+		/*
+		 * Check before we set max_hole_start, otherwise we could end up
+		 * sending back this offset anyway.
+		 */
+		if (contains_pending_extent(device, hole_start, *hole_size)) {
+			if (hole_end >= *hole_start)
+				*hole_size = hole_end - *hole_start;
+			else
+				*hole_size = 0;
+			changed = true;
+		}
+
+		switch (device->fs_devices->chunk_alloc_policy) {
+		case BTRFS_CHUNK_ALLOC_REGULAR:
+			/* No extra check */
+			break;
+		case BTRFS_CHUNK_ALLOC_ZONED:
+			if (dev_extent_hole_check_zoned(device, hole_start,
+							hole_size, num_bytes)) {
+				changed = true;
+				/*
+				 * The changed hole can contain pending extent.
+				 * Loop again to check that.
+				 */
+				continue;
+			}
+			break;
+		default:
+			BUG();
+		}
+
+		break;
+	}
+
+	return changed;
+}
 
 /*
  * find_free_dev_extent_start - find free space in the specified device
@@ -1409,7 +1552,7 @@ static bool contains_pending_extent(struct btrfs_device *device, u64 *start,
  * check to ensure dev extents are not double allocated.
  * This makes the function safe to allocate dev extents but may not report
  * correct usable device space, as device extent freed in current transaction
- * is not reported as avaiable.
+ * is not reported as available.
  */
 static int find_free_dev_extent_start(struct btrfs_device *device,
 				u64 num_bytes, u64 search_start, u64 *start,
@@ -1429,12 +1572,10 @@ static int find_free_dev_extent_start(struct btrfs_device *device,
 	int slot;
 	struct extent_buffer *l;
 
-	/*
-	 * We don't want to overwrite the superblock on the drive nor any area
-	 * used by the boot loader (grub for example), so we make sure to start
-	 * at an offset of at least 1MB.
-	 */
-	search_start = max_t(u64, search_start, SZ_1M);
+	search_start = dev_extent_search_start(device, search_start);
+
+	WARN_ON(device->zone_info &&
+		!IS_ALIGNED(num_bytes, device->zone_info->zone_size));
 
 	path = btrfs_alloc_path();
 	if (!path)
@@ -1458,14 +1599,9 @@ again:
 	key.offset = search_start;
 	key.type = BTRFS_DEV_EXTENT_KEY;
 
-	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
+	ret = btrfs_search_backwards(root, &key, path);
 	if (ret < 0)
 		goto out;
-	if (ret > 0) {
-		ret = btrfs_previous_item(root, path, key.objectid, key.type);
-		if (ret < 0)
-			goto out;
-	}
 
 	while (1) {
 		l = path->nodes[0];
@@ -1492,18 +1628,8 @@ again:
 
 		if (key.offset > search_start) {
 			hole_size = key.offset - search_start;
-
-			/*
-			 * Have to check before we set max_hole_start, otherwise
-			 * we could end up sending back this offset anyway.
-			 */
-			if (contains_pending_extent(device, &search_start,
-						    hole_size)) {
-				if (key.offset >= search_start)
-					hole_size = key.offset - search_start;
-				else
-					hole_size = 0;
-			}
+			dev_extent_hole_check(device, &search_start, &hole_size,
+					      num_bytes);
 
 			if (hole_size > max_hole_size) {
 				max_hole_start = search_start;
@@ -1542,8 +1668,8 @@ next:
 	 */
 	if (search_end > search_start) {
 		hole_size = search_end - search_start;
-
-		if (contains_pending_extent(device, &search_start, hole_size)) {
+		if (dev_extent_hole_check(device, &search_start, &hole_size,
+					  num_bytes)) {
 			btrfs_release_path(path);
 			goto again;
 		}
@@ -1616,61 +1742,14 @@ again:
 		extent = btrfs_item_ptr(leaf, path->slots[0],
 					struct btrfs_dev_extent);
 	} else {
-		btrfs_handle_fs_error(fs_info, ret, "Slot search failed");
 		goto out;
 	}
 
 	*dev_extent_len = btrfs_dev_extent_length(leaf, extent);
 
 	ret = btrfs_del_item(trans, root, path);
-	if (ret) {
-		btrfs_handle_fs_error(fs_info, ret,
-				      "Failed to remove dev extent item");
-	} else {
+	if (ret == 0)
 		set_bit(BTRFS_TRANS_HAVE_FREE_BGS, &trans->transaction->flags);
-	}
-out:
-	btrfs_free_path(path);
-	return ret;
-}
-
-static int btrfs_alloc_dev_extent(struct btrfs_trans_handle *trans,
-				  struct btrfs_device *device,
-				  u64 chunk_offset, u64 start, u64 num_bytes)
-{
-	int ret;
-	struct btrfs_path *path;
-	struct btrfs_fs_info *fs_info = device->fs_info;
-	struct btrfs_root *root = fs_info->dev_root;
-	struct btrfs_dev_extent *extent;
-	struct extent_buffer *leaf;
-	struct btrfs_key key;
-
-	WARN_ON(!test_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &device->dev_state));
-	WARN_ON(test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state));
-	path = btrfs_alloc_path();
-	if (!path)
-		return -ENOMEM;
-
-	key.objectid = device->devid;
-	key.offset = start;
-	key.type = BTRFS_DEV_EXTENT_KEY;
-	ret = btrfs_insert_empty_item(trans, root, path, &key,
-				      sizeof(*extent));
-	if (ret)
-		goto out;
-
-	leaf = path->nodes[0];
-	extent = btrfs_item_ptr(leaf, path->slots[0],
-				struct btrfs_dev_extent);
-	btrfs_set_dev_extent_chunk_tree(leaf, extent,
-					BTRFS_CHUNK_TREE_OBJECTID);
-	btrfs_set_dev_extent_chunk_objectid(leaf, extent,
-					    BTRFS_FIRST_CHUNK_TREE_OBJECTID);
-	btrfs_set_dev_extent_chunk_offset(leaf, extent, chunk_offset);
-
-	btrfs_set_dev_extent_length(leaf, extent, num_bytes);
-	btrfs_mark_buffer_dirty(leaf);
 out:
 	btrfs_free_path(path);
 	return ret;
@@ -1760,8 +1839,10 @@ static int btrfs_add_dev_item(struct btrfs_trans_handle *trans,
 	key.type = BTRFS_DEV_ITEM_KEY;
 	key.offset = device->devid;
 
+	btrfs_reserve_chunk_metadata(trans, true);
 	ret = btrfs_insert_empty_item(trans, trans->fs_info->chunk_root, path,
 				      &key, sizeof(*dev_item));
+	btrfs_trans_release_chunk_metadata(trans);
 	if (ret)
 		goto out;
 
@@ -1799,58 +1880,52 @@ out:
 /*
  * Function to update ctime/mtime for a given device path.
  * Mainly used for ctime/mtime based probe like libblkid.
+ *
+ * We don't care about errors here, this is just to be kind to userspace.
  */
-static void update_dev_time(const char *path_name)
+static void update_dev_time(const char *device_path)
 {
-	struct file *filp;
+	struct path path;
+	struct timespec64 now;
+	int ret;
 
-	filp = filp_open(path_name, O_RDWR, 0);
-	if (IS_ERR(filp))
+	ret = kern_path(device_path, LOOKUP_FOLLOW, &path);
+	if (ret)
 		return;
-	file_update_time(filp);
-	filp_close(filp, NULL);
+
+	now = current_time(d_inode(path.dentry));
+	inode_update_time(d_inode(path.dentry), &now, S_MTIME | S_CTIME);
+	path_put(&path);
 }
 
-static int btrfs_rm_dev_item(struct btrfs_device *device)
+static int btrfs_rm_dev_item(struct btrfs_trans_handle *trans,
+			     struct btrfs_device *device)
 {
 	struct btrfs_root *root = device->fs_info->chunk_root;
 	int ret;
 	struct btrfs_path *path;
 	struct btrfs_key key;
-	struct btrfs_trans_handle *trans;
 
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
 
-	trans = btrfs_start_transaction(root, 0);
-	if (IS_ERR(trans)) {
-		btrfs_free_path(path);
-		return PTR_ERR(trans);
-	}
 	key.objectid = BTRFS_DEV_ITEMS_OBJECTID;
 	key.type = BTRFS_DEV_ITEM_KEY;
 	key.offset = device->devid;
 
+	btrfs_reserve_chunk_metadata(trans, false);
 	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
+	btrfs_trans_release_chunk_metadata(trans);
 	if (ret) {
 		if (ret > 0)
 			ret = -ENOENT;
-		btrfs_abort_transaction(trans, ret);
-		btrfs_end_transaction(trans);
 		goto out;
 	}
 
 	ret = btrfs_del_item(trans, root, path);
-	if (ret) {
-		btrfs_abort_transaction(trans, ret);
-		btrfs_end_transaction(trans);
-	}
-
 out:
 	btrfs_free_path(path);
-	if (!ret)
-		ret = btrfs_commit_transaction(trans);
 	return ret;
 }
 
@@ -1878,12 +1953,8 @@ static int btrfs_check_raid_min_devices(struct btrfs_fs_info *fs_info,
 		if (!(all_avail & btrfs_raid_array[i].bg_flag))
 			continue;
 
-		if (num_devices < btrfs_raid_array[i].devs_min) {
-			int ret = btrfs_raid_array[i].mindev_error;
-
-			if (ret)
-				return ret;
-		}
+		if (num_devices < btrfs_raid_array[i].devs_min)
+			return btrfs_raid_array[i].mindev_error;
 	}
 
 	return 0;
@@ -1905,30 +1976,27 @@ static struct btrfs_device * btrfs_find_next_active_device(
 }
 
 /*
- * Helper function to check if the given device is part of s_bdev / latest_bdev
+ * Helper function to check if the given device is part of s_bdev / latest_dev
  * and replace it with the provided or the next active device, in the context
  * where this function called, there should be always be another device (or
  * this_dev) which is active.
  */
 void __cold btrfs_assign_next_active_device(struct btrfs_device *device,
-				     struct btrfs_device *this_dev)
+					    struct btrfs_device *next_device)
 {
 	struct btrfs_fs_info *fs_info = device->fs_info;
-	struct btrfs_device *next_device;
 
-	if (this_dev)
-		next_device = this_dev;
-	else
+	if (!next_device)
 		next_device = btrfs_find_next_active_device(fs_info->fs_devices,
-								device);
+							    device);
 	ASSERT(next_device);
 
 	if (fs_info->sb->s_bdev &&
 			(fs_info->sb->s_bdev == device->bdev))
 		fs_info->sb->s_bdev = next_device->bdev;
 
-	if (fs_info->fs_devices->latest_bdev == device->bdev)
-		fs_info->fs_devices->latest_bdev = next_device->bdev;
+	if (fs_info->fs_devices->latest_dev->bdev == device->bdev)
+		fs_info->fs_devices->latest_dev = next_device;
 }
 
 /*
@@ -1949,52 +2017,100 @@ static u64 btrfs_num_devices(struct btrfs_fs_info *fs_info)
 	return num_devices;
 }
 
-int btrfs_rm_device(struct btrfs_fs_info *fs_info, const char *device_path,
-		u64 devid)
+void btrfs_scratch_superblocks(struct btrfs_fs_info *fs_info,
+			       struct block_device *bdev,
+			       const char *device_path)
 {
+	struct btrfs_super_block *disk_super;
+	int copy_num;
+
+	if (!bdev)
+		return;
+
+	for (copy_num = 0; copy_num < BTRFS_SUPER_MIRROR_MAX; copy_num++) {
+		struct page *page;
+		int ret;
+
+		disk_super = btrfs_read_dev_one_super(bdev, copy_num, false);
+		if (IS_ERR(disk_super))
+			continue;
+
+		if (bdev_is_zoned(bdev)) {
+			btrfs_reset_sb_log_zones(bdev, copy_num);
+			continue;
+		}
+
+		memset(&disk_super->magic, 0, sizeof(disk_super->magic));
+
+		page = virt_to_page(disk_super);
+		set_page_dirty(page);
+		lock_page(page);
+		/* write_on_page() unlocks the page */
+		ret = write_one_page(page);
+		if (ret)
+			btrfs_warn(fs_info,
+				"error clearing superblock number %d (%d)",
+				copy_num, ret);
+		btrfs_release_disk_super(disk_super);
+
+	}
+
+	/* Notify udev that device has changed */
+	btrfs_kobject_uevent(bdev, KOBJ_CHANGE);
+
+	/* Update ctime/mtime for device path for libblkid */
+	update_dev_time(device_path);
+}
+
+int btrfs_rm_device(struct btrfs_fs_info *fs_info,
+		    struct btrfs_dev_lookup_args *args,
+		    struct block_device **bdev, fmode_t *mode)
+{
+	struct btrfs_trans_handle *trans;
 	struct btrfs_device *device;
 	struct btrfs_fs_devices *cur_devices;
 	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
 	u64 num_devices;
 	int ret = 0;
 
-	mutex_lock(&uuid_mutex);
+	if (btrfs_fs_incompat(fs_info, EXTENT_TREE_V2)) {
+		btrfs_err(fs_info, "device remove not supported on extent tree v2 yet");
+		return -EINVAL;
+	}
 
+	/*
+	 * The device list in fs_devices is accessed without locks (neither
+	 * uuid_mutex nor device_list_mutex) as it won't change on a mounted
+	 * filesystem and another device rm cannot run.
+	 */
 	num_devices = btrfs_num_devices(fs_info);
 
 	ret = btrfs_check_raid_min_devices(fs_info, num_devices - 1);
 	if (ret)
-		goto out;
-
-	device = btrfs_find_device_by_devspec(fs_info, devid, device_path);
+		return ret;
 
-	if (IS_ERR(device)) {
-		if (PTR_ERR(device) == -ENOENT &&
-		    strcmp(device_path, "missing") == 0)
+	device = btrfs_find_device(fs_info->fs_devices, args);
+	if (!device) {
+		if (args->missing)
 			ret = BTRFS_ERROR_DEV_MISSING_NOT_FOUND;
 		else
-			ret = PTR_ERR(device);
-		goto out;
+			ret = -ENOENT;
+		return ret;
 	}
 
 	if (btrfs_pinned_by_swapfile(fs_info, device)) {
 		btrfs_warn_in_rcu(fs_info,
 		  "cannot remove device %s (devid %llu) due to active swapfile",
 				  rcu_str_deref(device->name), device->devid);
-		ret = -ETXTBSY;
-		goto out;
+		return -ETXTBSY;
 	}
 
-	if (test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state)) {
-		ret = BTRFS_ERROR_DEV_TGT_REPLACE;
-		goto out;
-	}
+	if (test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state))
+		return BTRFS_ERROR_DEV_TGT_REPLACE;
 
 	if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state) &&
-	    fs_info->fs_devices->rw_devices == 1) {
-		ret = BTRFS_ERROR_DEV_ONLY_WRITABLE;
-		goto out;
-	}
+	    fs_info->fs_devices->rw_devices == 1)
+		return BTRFS_ERROR_DEV_ONLY_WRITABLE;
 
 	if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state)) {
 		mutex_lock(&fs_info->chunk_mutex);
@@ -2003,20 +2119,26 @@ int btrfs_rm_device(struct btrfs_fs_info *fs_info, const char *device_path,
 		mutex_unlock(&fs_info->chunk_mutex);
 	}
 
-	mutex_unlock(&uuid_mutex);
 	ret = btrfs_shrink_device(device, 0);
-	mutex_lock(&uuid_mutex);
 	if (ret)
 		goto error_undo;
 
-	/*
-	 * TODO: the superblock still includes this device in its num_devices
-	 * counter although write_all_supers() is not locked out. This
-	 * could give a filesystem state which requires a degraded mount.
-	 */
-	ret = btrfs_rm_dev_item(device);
-	if (ret)
+	trans = btrfs_start_transaction(fs_info->chunk_root, 0);
+	if (IS_ERR(trans)) {
+		ret = PTR_ERR(trans);
 		goto error_undo;
+	}
+
+	ret = btrfs_rm_dev_item(trans, device);
+	if (ret) {
+		/* Any error in dev item removal is critical */
+		btrfs_crit(fs_info,
+			   "failed to remove device item for devid %llu: %d",
+			   device->devid, ret);
+		btrfs_abort_transaction(trans, ret);
+		btrfs_end_transaction(trans);
+		return ret;
+	}
 
 	clear_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &device->dev_state);
 	btrfs_scrub_cancel_dev(device);
@@ -2034,7 +2156,7 @@ int btrfs_rm_device(struct btrfs_fs_info *fs_info, const char *device_path,
 	/*
 	 * In normal cases the cur_devices == fs_devices. But in case
 	 * of deleting a seed device, the cur_devices should point to
-	 * its own fs_devices listed under the fs_devices->seed.
+	 * its own fs_devices listed under the fs_devices->seed_list.
 	 */
 	cur_devices = device->fs_devices;
 	mutex_lock(&fs_devices->device_list_mutex);
@@ -2054,7 +2176,7 @@ int btrfs_rm_device(struct btrfs_fs_info *fs_info, const char *device_path,
 	if (device->bdev) {
 		cur_devices->open_devices--;
 		/* remove sysfs entry */
-		btrfs_sysfs_rm_device_link(fs_devices, device);
+		btrfs_sysfs_remove_device(device);
 	}
 
 	num_devices = btrfs_super_num_devices(fs_info->super_copy) - 1;
@@ -2062,32 +2184,45 @@ int btrfs_rm_device(struct btrfs_fs_info *fs_info, const char *device_path,
 	mutex_unlock(&fs_devices->device_list_mutex);
 
 	/*
-	 * at this point, the device is zero sized and detached from
-	 * the devices list.  All that's left is to zero out the old
-	 * supers and free the device.
+	 * At this point, the device is zero sized and detached from the
+	 * devices list.  All that's left is to zero out the old supers and
+	 * free the device.
+	 *
+	 * We cannot call btrfs_close_bdev() here because we're holding the sb
+	 * write lock, and blkdev_put() will pull in the ->open_mutex on the
+	 * block device and it's dependencies.  Instead just flush the device
+	 * and let the caller do the final blkdev_put.
 	 */
-	if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state))
-		btrfs_scratch_superblocks(device->bdev, device->name->str);
+	if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state)) {
+		btrfs_scratch_superblocks(fs_info, device->bdev,
+					  device->name->str);
+		if (device->bdev) {
+			sync_blockdev(device->bdev);
+			invalidate_bdev(device->bdev);
+		}
+	}
 
-	btrfs_close_bdev(device);
+	*bdev = device->bdev;
+	*mode = device->mode;
 	synchronize_rcu();
 	btrfs_free_device(device);
 
-	if (cur_devices->open_devices == 0) {
-		while (fs_devices) {
-			if (fs_devices->seed == cur_devices) {
-				fs_devices->seed = cur_devices->seed;
-				break;
-			}
-			fs_devices = fs_devices->seed;
-		}
-		cur_devices->seed = NULL;
-		close_fs_devices(cur_devices);
+	/*
+	 * This can happen if cur_devices is the private seed devices list.  We
+	 * cannot call close_fs_devices() here because it expects the uuid_mutex
+	 * to be held, but in fact we don't need that for the private
+	 * seed_devices, we can simply decrement cur_devices->opened and then
+	 * remove it from our list and free the fs_devices.
+	 */
+	if (cur_devices->num_devices == 0) {
+		list_del_init(&cur_devices->seed_list);
+		ASSERT(cur_devices->opened == 1);
+		cur_devices->opened--;
 		free_fs_devices(cur_devices);
 	}
 
-out:
-	mutex_unlock(&uuid_mutex);
+	ret = btrfs_commit_transaction(trans);
+
 	return ret;
 
 error_undo:
@@ -2098,7 +2233,7 @@ error_undo:
 		device->fs_devices->rw_devices++;
 		mutex_unlock(&fs_info->chunk_mutex);
 	}
-	goto out;
+	return ret;
 }
 
 void btrfs_rm_dev_replace_remove_srcdev(struct btrfs_device *srcdev)
@@ -2130,13 +2265,9 @@ void btrfs_rm_dev_replace_remove_srcdev(struct btrfs_device *srcdev)
 
 void btrfs_rm_dev_replace_free_srcdev(struct btrfs_device *srcdev)
 {
-	struct btrfs_fs_info *fs_info = srcdev->fs_info;
 	struct btrfs_fs_devices *fs_devices = srcdev->fs_devices;
 
-	if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &srcdev->dev_state)) {
-		/* zero out the old super if it is writable */
-		btrfs_scratch_superblocks(srcdev->bdev, srcdev->name->str);
-	}
+	mutex_lock(&uuid_mutex);
 
 	btrfs_close_bdev(srcdev);
 	synchronize_rcu();
@@ -2144,8 +2275,6 @@ void btrfs_rm_dev_replace_free_srcdev(struct btrfs_device *srcdev)
 
 	/* if this is no devs we rather delete the fs_devices */
 	if (!fs_devices->num_devices) {
-		struct btrfs_fs_devices *tmp_fs_devices;
-
 		/*
 		 * On a mounted FS, num_devices can't be zero unless it's a
 		 * seed. In case of a seed device being replaced, the replace
@@ -2154,18 +2283,11 @@ void btrfs_rm_dev_replace_free_srcdev(struct btrfs_device *srcdev)
 		 */
 		ASSERT(fs_devices->seeding);
 
-		tmp_fs_devices = fs_info->fs_devices;
-		while (tmp_fs_devices) {
-			if (tmp_fs_devices->seed == fs_devices) {
-				tmp_fs_devices->seed = fs_devices->seed;
-				break;
-			}
-			tmp_fs_devices = tmp_fs_devices->seed;
-		}
-		fs_devices->seed = NULL;
+		list_del_init(&fs_devices->seed_list);
 		close_fs_devices(fs_devices);
 		free_fs_devices(fs_devices);
 	}
+	mutex_unlock(&uuid_mutex);
 }
 
 void btrfs_destroy_dev_replace_tgtdev(struct btrfs_device *tgtdev)
@@ -2174,7 +2296,7 @@ void btrfs_destroy_dev_replace_tgtdev(struct btrfs_device *tgtdev)
 
 	mutex_lock(&fs_devices->device_list_mutex);
 
-	btrfs_sysfs_rm_device_link(fs_devices, tgtdev);
+	btrfs_sysfs_remove_device(tgtdev);
 
 	if (tgtdev->bdev)
 		fs_devices->open_devices--;
@@ -2187,110 +2309,139 @@ void btrfs_destroy_dev_replace_tgtdev(struct btrfs_device *tgtdev)
 
 	mutex_unlock(&fs_devices->device_list_mutex);
 
-	/*
-	 * The update_dev_time() with in btrfs_scratch_superblocks()
-	 * may lead to a call to btrfs_show_devname() which will try
-	 * to hold device_list_mutex. And here this device
-	 * is already out of device list, so we don't have to hold
-	 * the device_list_mutex lock.
-	 */
-	btrfs_scratch_superblocks(tgtdev->bdev, tgtdev->name->str);
+	btrfs_scratch_superblocks(tgtdev->fs_info, tgtdev->bdev,
+				  tgtdev->name->str);
 
 	btrfs_close_bdev(tgtdev);
 	synchronize_rcu();
 	btrfs_free_device(tgtdev);
 }
 
-static struct btrfs_device *btrfs_find_device_by_path(
-		struct btrfs_fs_info *fs_info, const char *device_path)
+/**
+ * Populate args from device at path
+ *
+ * @fs_info:	the filesystem
+ * @args:	the args to populate
+ * @path:	the path to the device
+ *
+ * This will read the super block of the device at @path and populate @args with
+ * the devid, fsid, and uuid.  This is meant to be used for ioctls that need to
+ * lookup a device to operate on, but need to do it before we take any locks.
+ * This properly handles the special case of "missing" that a user may pass in,
+ * and does some basic sanity checks.  The caller must make sure that @path is
+ * properly NUL terminated before calling in, and must call
+ * btrfs_put_dev_args_from_path() in order to free up the temporary fsid and
+ * uuid buffers.
+ *
+ * Return: 0 for success, -errno for failure
+ */
+int btrfs_get_dev_args_from_path(struct btrfs_fs_info *fs_info,
+				 struct btrfs_dev_lookup_args *args,
+				 const char *path)
 {
-	int ret = 0;
 	struct btrfs_super_block *disk_super;
-	u64 devid;
-	u8 *dev_uuid;
 	struct block_device *bdev;
-	struct buffer_head *bh;
-	struct btrfs_device *device;
+	int ret;
 
-	ret = btrfs_get_bdev_and_sb(device_path, FMODE_READ,
-				    fs_info->bdev_holder, 0, &bdev, &bh);
-	if (ret)
-		return ERR_PTR(ret);
-	disk_super = (struct btrfs_super_block *)bh->b_data;
-	devid = btrfs_stack_device_id(&disk_super->dev_item);
-	dev_uuid = disk_super->dev_item.uuid;
+	if (!path || !path[0])
+		return -EINVAL;
+	if (!strcmp(path, "missing")) {
+		args->missing = true;
+		return 0;
+	}
+
+	args->uuid = kzalloc(BTRFS_UUID_SIZE, GFP_KERNEL);
+	args->fsid = kzalloc(BTRFS_FSID_SIZE, GFP_KERNEL);
+	if (!args->uuid || !args->fsid) {
+		btrfs_put_dev_args_from_path(args);
+		return -ENOMEM;
+	}
+
+	ret = btrfs_get_bdev_and_sb(path, FMODE_READ, fs_info->bdev_holder, 0,
+				    &bdev, &disk_super);
+	if (ret) {
+		btrfs_put_dev_args_from_path(args);
+		return ret;
+	}
+
+	args->devid = btrfs_stack_device_id(&disk_super->dev_item);
+	memcpy(args->uuid, disk_super->dev_item.uuid, BTRFS_UUID_SIZE);
 	if (btrfs_fs_incompat(fs_info, METADATA_UUID))
-		device = btrfs_find_device(fs_info->fs_devices, devid, dev_uuid,
-					   disk_super->metadata_uuid, true);
+		memcpy(args->fsid, disk_super->metadata_uuid, BTRFS_FSID_SIZE);
 	else
-		device = btrfs_find_device(fs_info->fs_devices, devid, dev_uuid,
-					   disk_super->fsid, true);
-
-	brelse(bh);
-	if (!device)
-		device = ERR_PTR(-ENOENT);
+		memcpy(args->fsid, disk_super->fsid, BTRFS_FSID_SIZE);
+	btrfs_release_disk_super(disk_super);
 	blkdev_put(bdev, FMODE_READ);
-	return device;
+	return 0;
 }
 
 /*
- * Lookup a device given by device id, or the path if the id is 0.
+ * Only use this jointly with btrfs_get_dev_args_from_path() because we will
+ * allocate our ->uuid and ->fsid pointers, everybody else uses local variables
+ * that don't need to be freed.
  */
+void btrfs_put_dev_args_from_path(struct btrfs_dev_lookup_args *args)
+{
+	kfree(args->uuid);
+	kfree(args->fsid);
+	args->uuid = NULL;
+	args->fsid = NULL;
+}
+
 struct btrfs_device *btrfs_find_device_by_devspec(
 		struct btrfs_fs_info *fs_info, u64 devid,
 		const char *device_path)
 {
+	BTRFS_DEV_LOOKUP_ARGS(args);
 	struct btrfs_device *device;
+	int ret;
 
 	if (devid) {
-		device = btrfs_find_device(fs_info->fs_devices, devid, NULL,
-					   NULL, true);
+		args.devid = devid;
+		device = btrfs_find_device(fs_info->fs_devices, &args);
 		if (!device)
 			return ERR_PTR(-ENOENT);
 		return device;
 	}
 
-	if (!device_path || !device_path[0])
-		return ERR_PTR(-EINVAL);
-
-	if (strcmp(device_path, "missing") == 0) {
-		/* Find first missing device */
-		list_for_each_entry(device, &fs_info->fs_devices->devices,
-				    dev_list) {
-			if (test_bit(BTRFS_DEV_STATE_IN_FS_METADATA,
-				     &device->dev_state) && !device->bdev)
-				return device;
-		}
+	ret = btrfs_get_dev_args_from_path(fs_info, &args, device_path);
+	if (ret)
+		return ERR_PTR(ret);
+	device = btrfs_find_device(fs_info->fs_devices, &args);
+	btrfs_put_dev_args_from_path(&args);
+	if (!device)
 		return ERR_PTR(-ENOENT);
-	}
-
-	return btrfs_find_device_by_path(fs_info, device_path);
+	return device;
 }
 
-/*
- * does all the dirty work required for changing file system's UUID.
- */
-static int btrfs_prepare_sprout(struct btrfs_fs_info *fs_info)
+static struct btrfs_fs_devices *btrfs_init_sprout(struct btrfs_fs_info *fs_info)
 {
 	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
 	struct btrfs_fs_devices *old_devices;
 	struct btrfs_fs_devices *seed_devices;
-	struct btrfs_super_block *disk_super = fs_info->super_copy;
-	struct btrfs_device *device;
-	u64 super_flags;
 
 	lockdep_assert_held(&uuid_mutex);
 	if (!fs_devices->seeding)
-		return -EINVAL;
+		return ERR_PTR(-EINVAL);
 
+	/*
+	 * Private copy of the seed devices, anchored at
+	 * fs_info->fs_devices->seed_list
+	 */
 	seed_devices = alloc_fs_devices(NULL, NULL);
 	if (IS_ERR(seed_devices))
-		return PTR_ERR(seed_devices);
+		return seed_devices;
 
+	/*
+	 * It's necessary to retain a copy of the original seed fs_devices in
+	 * fs_uuids so that filesystems which have been seeded can successfully
+	 * reference the seed device from open_seed_devices. This also supports
+	 * multiple fs seed.
+	 */
 	old_devices = clone_fs_devices(fs_devices);
 	if (IS_ERR(old_devices)) {
 		kfree(seed_devices);
-		return PTR_ERR(old_devices);
+		return old_devices;
 	}
 
 	list_add(&old_devices->fs_list, &fs_uuids);
@@ -2301,33 +2452,60 @@ static int btrfs_prepare_sprout(struct btrfs_fs_info *fs_info)
 	INIT_LIST_HEAD(&seed_devices->alloc_list);
 	mutex_init(&seed_devices->device_list_mutex);
 
-	mutex_lock(&fs_devices->device_list_mutex);
+	return seed_devices;
+}
+
+/*
+ * Splice seed devices into the sprout fs_devices.
+ * Generate a new fsid for the sprouted read-write filesystem.
+ */
+static void btrfs_setup_sprout(struct btrfs_fs_info *fs_info,
+			       struct btrfs_fs_devices *seed_devices)
+{
+	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
+	struct btrfs_super_block *disk_super = fs_info->super_copy;
+	struct btrfs_device *device;
+	u64 super_flags;
+
+	/*
+	 * We are updating the fsid, the thread leading to device_list_add()
+	 * could race, so uuid_mutex is needed.
+	 */
+	lockdep_assert_held(&uuid_mutex);
+
+	/*
+	 * The threads listed below may traverse dev_list but can do that without
+	 * device_list_mutex:
+	 * - All device ops and balance - as we are in btrfs_exclop_start.
+	 * - Various dev_list readers - are using RCU.
+	 * - btrfs_ioctl_fitrim() - is using RCU.
+	 *
+	 * For-read threads as below are using device_list_mutex:
+	 * - Readonly scrub btrfs_scrub_dev()
+	 * - Readonly scrub btrfs_scrub_progress()
+	 * - btrfs_get_dev_stats()
+	 */
+	lockdep_assert_held(&fs_devices->device_list_mutex);
+
 	list_splice_init_rcu(&fs_devices->devices, &seed_devices->devices,
 			      synchronize_rcu);
 	list_for_each_entry(device, &seed_devices->devices, dev_list)
 		device->fs_devices = seed_devices;
 
-	mutex_lock(&fs_info->chunk_mutex);
-	list_splice_init(&fs_devices->alloc_list, &seed_devices->alloc_list);
-	mutex_unlock(&fs_info->chunk_mutex);
-
 	fs_devices->seeding = false;
 	fs_devices->num_devices = 0;
 	fs_devices->open_devices = 0;
 	fs_devices->missing_devices = 0;
 	fs_devices->rotating = false;
-	fs_devices->seed = seed_devices;
+	list_add(&seed_devices->seed_list, &fs_devices->seed_list);
 
 	generate_random_uuid(fs_devices->fsid);
 	memcpy(fs_devices->metadata_uuid, fs_devices->fsid, BTRFS_FSID_SIZE);
 	memcpy(disk_super->fsid, fs_devices->fsid, BTRFS_FSID_SIZE);
-	mutex_unlock(&fs_devices->device_list_mutex);
 
 	super_flags = btrfs_super_flags(disk_super) &
 		      ~BTRFS_SUPER_FLAG_SEEDING;
 	btrfs_set_super_flags(disk_super, super_flags);
-
-	return 0;
 }
 
 /*
@@ -2335,6 +2513,7 @@ static int btrfs_prepare_sprout(struct btrfs_fs_info *fs_info)
  */
 static int btrfs_finish_sprout(struct btrfs_trans_handle *trans)
 {
+	BTRFS_DEV_LOOKUP_ARGS(args);
 	struct btrfs_fs_info *fs_info = trans->fs_info;
 	struct btrfs_root *root = fs_info->chunk_root;
 	struct btrfs_path *path;
@@ -2344,7 +2523,6 @@ static int btrfs_finish_sprout(struct btrfs_trans_handle *trans)
 	struct btrfs_key key;
 	u8 fs_uuid[BTRFS_FSID_SIZE];
 	u8 dev_uuid[BTRFS_UUID_SIZE];
-	u64 devid;
 	int ret;
 
 	path = btrfs_alloc_path();
@@ -2356,7 +2534,9 @@ static int btrfs_finish_sprout(struct btrfs_trans_handle *trans)
 	key.type = BTRFS_DEV_ITEM_KEY;
 
 	while (1) {
+		btrfs_reserve_chunk_metadata(trans, false);
 		ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
+		btrfs_trans_release_chunk_metadata(trans);
 		if (ret < 0)
 			goto error;
 
@@ -2381,13 +2561,14 @@ next_slot:
 
 		dev_item = btrfs_item_ptr(leaf, path->slots[0],
 					  struct btrfs_dev_item);
-		devid = btrfs_device_id(leaf, dev_item);
+		args.devid = btrfs_device_id(leaf, dev_item);
 		read_extent_buffer(leaf, dev_uuid, btrfs_device_uuid(dev_item),
 				   BTRFS_UUID_SIZE);
 		read_extent_buffer(leaf, fs_uuid, btrfs_device_fsid(dev_item),
 				   BTRFS_FSID_SIZE);
-		device = btrfs_find_device(fs_info->fs_devices, devid, dev_uuid,
-					   fs_uuid, true);
+		args.uuid = dev_uuid;
+		args.fsid = fs_uuid;
+		device = btrfs_find_device(fs_info->fs_devices, &args);
 		BUG_ON(!device); /* Logic error */
 
 		if (device->fs_devices->seeding) {
@@ -2408,18 +2589,18 @@ error:
 int btrfs_init_new_device(struct btrfs_fs_info *fs_info, const char *device_path)
 {
 	struct btrfs_root *root = fs_info->dev_root;
-	struct request_queue *q;
 	struct btrfs_trans_handle *trans;
 	struct btrfs_device *device;
 	struct block_device *bdev;
 	struct super_block *sb = fs_info->sb;
 	struct rcu_string *name;
 	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
+	struct btrfs_fs_devices *seed_devices;
 	u64 orig_super_total_bytes;
 	u64 orig_super_num_devices;
-	int seeding_dev = 0;
 	int ret = 0;
-	bool unlocked = false;
+	bool seeding_dev = false;
+	bool locked = false;
 
 	if (sb_rdonly(sb) && !fs_devices->seeding)
 		return -EROFS;
@@ -2429,24 +2610,29 @@ int btrfs_init_new_device(struct btrfs_fs_info *fs_info, const char *device_path
 	if (IS_ERR(bdev))
 		return PTR_ERR(bdev);
 
+	if (!btrfs_check_device_zone_type(fs_info, bdev)) {
+		ret = -EINVAL;
+		goto error;
+	}
+
 	if (fs_devices->seeding) {
-		seeding_dev = 1;
+		seeding_dev = true;
 		down_write(&sb->s_umount);
 		mutex_lock(&uuid_mutex);
+		locked = true;
 	}
 
-	filemap_write_and_wait(bdev->bd_inode->i_mapping);
+	sync_blockdev(bdev);
 
-	mutex_lock(&fs_devices->device_list_mutex);
-	list_for_each_entry(device, &fs_devices->devices, dev_list) {
+	rcu_read_lock();
+	list_for_each_entry_rcu(device, &fs_devices->devices, dev_list) {
 		if (device->bdev == bdev) {
 			ret = -EEXIST;
-			mutex_unlock(
-				&fs_devices->device_list_mutex);
+			rcu_read_unlock();
 			goto error;
 		}
 	}
-	mutex_unlock(&fs_devices->device_list_mutex);
+	rcu_read_unlock();
 
 	device = btrfs_alloc_device(fs_info, NULL, NULL);
 	if (IS_ERR(device)) {
@@ -2462,24 +2648,31 @@ int btrfs_init_new_device(struct btrfs_fs_info *fs_info, const char *device_path
 	}
 	rcu_assign_pointer(device->name, name);
 
+	device->fs_info = fs_info;
+	device->bdev = bdev;
+	ret = lookup_bdev(device_path, &device->devt);
+	if (ret)
+		goto error_free_device;
+
+	ret = btrfs_get_dev_zone_info(device, false);
+	if (ret)
+		goto error_free_device;
+
 	trans = btrfs_start_transaction(root, 0);
 	if (IS_ERR(trans)) {
 		ret = PTR_ERR(trans);
-		goto error_free_device;
+		goto error_free_zone;
 	}
 
-	q = bdev_get_queue(bdev);
 	set_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state);
 	device->generation = trans->transid;
 	device->io_width = fs_info->sectorsize;
 	device->io_align = fs_info->sectorsize;
 	device->sector_size = fs_info->sectorsize;
-	device->total_bytes = round_down(i_size_read(bdev->bd_inode),
-					 fs_info->sectorsize);
+	device->total_bytes =
+		round_down(bdev_nr_bytes(bdev), fs_info->sectorsize);
 	device->disk_total_bytes = device->total_bytes;
 	device->commit_total_bytes = device->total_bytes;
-	device->fs_info = fs_info;
-	device->bdev = bdev;
 	set_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &device->dev_state);
 	clear_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state);
 	device->mode = FMODE_EXCL;
@@ -2487,17 +2680,26 @@ int btrfs_init_new_device(struct btrfs_fs_info *fs_info, const char *device_path
 	set_blocksize(device->bdev, BTRFS_BDEV_BLOCKSIZE);
 
 	if (seeding_dev) {
-		sb->s_flags &= ~SB_RDONLY;
-		ret = btrfs_prepare_sprout(fs_info);
-		if (ret) {
+		btrfs_clear_sb_rdonly(sb);
+
+		/* GFP_KERNEL allocation must not be under device_list_mutex */
+		seed_devices = btrfs_init_sprout(fs_info);
+		if (IS_ERR(seed_devices)) {
+			ret = PTR_ERR(seed_devices);
 			btrfs_abort_transaction(trans, ret);
 			goto error_trans;
 		}
 	}
 
+	mutex_lock(&fs_devices->device_list_mutex);
+	if (seeding_dev) {
+		btrfs_setup_sprout(fs_info, seed_devices);
+		btrfs_assign_next_active_device(fs_info->fs_devices->latest_dev,
+						device);
+	}
+
 	device->fs_devices = fs_devices;
 
-	mutex_lock(&fs_devices->device_list_mutex);
 	mutex_lock(&fs_info->chunk_mutex);
 	list_add_rcu(&device->dev_list, &fs_devices->devices);
 	list_add(&device->dev_alloc_list, &fs_devices->alloc_list);
@@ -2509,7 +2711,7 @@ int btrfs_init_new_device(struct btrfs_fs_info *fs_info, const char *device_path
 
 	atomic64_add(device->total_bytes, &fs_info->free_chunk_space);
 
-	if (!blk_queue_nonrot(q))
+	if (!bdev_nonrot(bdev))
 		fs_devices->rotating = true;
 
 	orig_super_total_bytes = btrfs_super_total_bytes(fs_info->super_copy);
@@ -2521,9 +2723,6 @@ int btrfs_init_new_device(struct btrfs_fs_info *fs_info, const char *device_path
 	btrfs_set_super_num_devices(fs_info->super_copy,
 				    orig_super_num_devices + 1);
 
-	/* add sysfs device entry */
-	btrfs_sysfs_add_device_link(fs_devices, device);
-
 	/*
 	 * we've got more storage, clear any full flags on the space
 	 * infos
@@ -2531,6 +2730,10 @@ int btrfs_init_new_device(struct btrfs_fs_info *fs_info, const char *device_path
 	btrfs_clear_space_info_full(fs_info);
 
 	mutex_unlock(&fs_info->chunk_mutex);
+
+	/* Add sysfs device entry */
+	btrfs_sysfs_add_device(device);
+
 	mutex_unlock(&fs_devices->device_list_mutex);
 
 	if (seeding_dev) {
@@ -2556,8 +2759,11 @@ int btrfs_init_new_device(struct btrfs_fs_info *fs_info, const char *device_path
 			goto error_sysfs;
 		}
 
-		btrfs_sysfs_update_sprout_fsid(fs_devices,
-				fs_info->fs_devices->fsid);
+		/*
+		 * fs_devices now represents the newly sprouted filesystem and
+		 * its fsid has been changed by btrfs_sprout_splice().
+		 */
+		btrfs_sysfs_update_sprout_fsid(fs_devices);
 	}
 
 	ret = btrfs_commit_transaction(trans);
@@ -2565,7 +2771,7 @@ int btrfs_init_new_device(struct btrfs_fs_info *fs_info, const char *device_path
 	if (seeding_dev) {
 		mutex_unlock(&uuid_mutex);
 		up_write(&sb->s_umount);
-		unlocked = true;
+		locked = false;
 
 		if (ret) /* transaction commit */
 			return ret;
@@ -2585,12 +2791,22 @@ int btrfs_init_new_device(struct btrfs_fs_info *fs_info, const char *device_path
 		ret = btrfs_commit_transaction(trans);
 	}
 
-	/* Update ctime/mtime for libblkid */
+	/*
+	 * Now that we have written a new super block to this device, check all
+	 * other fs_devices list if device_path alienates any other scanned
+	 * device.
+	 * We can ignore the return value as it typically returns -EINVAL and
+	 * only succeeds if the device was an alien.
+	 */
+	btrfs_forget_devices(device->devt);
+
+	/* Update ctime/mtime for blkid or udev */
 	update_dev_time(device_path);
+
 	return ret;
 
 error_sysfs:
-	btrfs_sysfs_rm_device_link(fs_devices, device);
+	btrfs_sysfs_remove_device(device);
 	mutex_lock(&fs_info->fs_devices->device_list_mutex);
 	mutex_lock(&fs_info->chunk_mutex);
 	list_del_rcu(&device->dev_list);
@@ -2609,14 +2825,16 @@ error_sysfs:
 	mutex_unlock(&fs_info->fs_devices->device_list_mutex);
 error_trans:
 	if (seeding_dev)
-		sb->s_flags |= SB_RDONLY;
+		btrfs_set_sb_rdonly(sb);
 	if (trans)
 		btrfs_end_transaction(trans);
+error_free_zone:
+	btrfs_destroy_dev_zone_info(device);
 error_free_device:
 	btrfs_free_device(device);
 error:
 	blkdev_put(bdev, FMODE_EXCL);
-	if (seeding_dev && !unlocked) {
+	if (locked) {
 		mutex_unlock(&uuid_mutex);
 		up_write(&sb->s_umount);
 	}
@@ -2676,6 +2894,7 @@ int btrfs_grow_device(struct btrfs_trans_handle *trans,
 	struct btrfs_super_block *super_copy = fs_info->super_copy;
 	u64 old_total;
 	u64 diff;
+	int ret;
 
 	if (!test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state))
 		return -EACCES;
@@ -2704,7 +2923,11 @@ int btrfs_grow_device(struct btrfs_trans_handle *trans,
 			      &trans->transaction->dev_update_list);
 	mutex_unlock(&fs_info->chunk_mutex);
 
-	return btrfs_update_device(trans, device);
+	btrfs_reserve_chunk_metadata(trans, false);
+	ret = btrfs_update_device(trans, device);
+	btrfs_trans_release_chunk_metadata(trans);
+
+	return ret;
 }
 
 static int btrfs_free_chunk(struct btrfs_trans_handle *trans, u64 chunk_offset)
@@ -2755,7 +2978,7 @@ static int btrfs_del_sys_chunk(struct btrfs_fs_info *fs_info, u64 chunk_offset)
 	u32 cur;
 	struct btrfs_key key;
 
-	mutex_lock(&fs_info->chunk_mutex);
+	lockdep_assert_held(&fs_info->chunk_mutex);
 	array_size = btrfs_super_sys_array_size(super_copy);
 
 	ptr = super_copy->sys_chunk_array;
@@ -2785,7 +3008,6 @@ static int btrfs_del_sys_chunk(struct btrfs_fs_info *fs_info, u64 chunk_offset)
 			cur += len;
 		}
 	}
-	mutex_unlock(&fs_info->chunk_mutex);
 	return ret;
 }
 
@@ -2825,6 +3047,29 @@ struct extent_map *btrfs_get_chunk_map(struct btrfs_fs_info *fs_info,
 	return em;
 }
 
+static int remove_chunk_item(struct btrfs_trans_handle *trans,
+			     struct map_lookup *map, u64 chunk_offset)
+{
+	int i;
+
+	/*
+	 * Removing chunk items and updating the device items in the chunks btree
+	 * requires holding the chunk_mutex.
+	 * See the comment at btrfs_chunk_alloc() for the details.
+	 */
+	lockdep_assert_held(&trans->fs_info->chunk_mutex);
+
+	for (i = 0; i < map->num_stripes; i++) {
+		int ret;
+
+		ret = btrfs_update_device(trans, map->stripes[i].dev);
+		if (ret)
+			return ret;
+	}
+
+	return btrfs_free_chunk(trans, chunk_offset);
+}
+
 int btrfs_remove_chunk(struct btrfs_trans_handle *trans, u64 chunk_offset)
 {
 	struct btrfs_fs_info *fs_info = trans->fs_info;
@@ -2845,14 +3090,16 @@ int btrfs_remove_chunk(struct btrfs_trans_handle *trans, u64 chunk_offset)
 		return PTR_ERR(em);
 	}
 	map = em->map_lookup;
-	mutex_lock(&fs_info->chunk_mutex);
-	check_system_chunk(trans, map->type);
-	mutex_unlock(&fs_info->chunk_mutex);
 
 	/*
-	 * Take the device list mutex to prevent races with the final phase of
-	 * a device replace operation that replaces the device object associated
-	 * with map stripes (dev-replace.c:btrfs_dev_replace_finishing()).
+	 * First delete the device extent items from the devices btree.
+	 * We take the device_list_mutex to avoid racing with the finishing phase
+	 * of a device replace operation. See the comment below before acquiring
+	 * fs_info->chunk_mutex. Note that here we do not acquire the chunk_mutex
+	 * because that can result in a deadlock when deleting the device extent
+	 * items from the devices btree - COWing an extent buffer from the btree
+	 * may result in allocating a new metadata chunk, which would attempt to
+	 * lock again fs_info->chunk_mutex.
 	 */
 	mutex_lock(&fs_devices->device_list_mutex);
 	for (i = 0; i < map->num_stripes; i++) {
@@ -2874,18 +3121,73 @@ int btrfs_remove_chunk(struct btrfs_trans_handle *trans, u64 chunk_offset)
 			btrfs_clear_space_info_full(fs_info);
 			mutex_unlock(&fs_info->chunk_mutex);
 		}
+	}
+	mutex_unlock(&fs_devices->device_list_mutex);
 
-		ret = btrfs_update_device(trans, device);
+	/*
+	 * We acquire fs_info->chunk_mutex for 2 reasons:
+	 *
+	 * 1) Just like with the first phase of the chunk allocation, we must
+	 *    reserve system space, do all chunk btree updates and deletions, and
+	 *    update the system chunk array in the superblock while holding this
+	 *    mutex. This is for similar reasons as explained on the comment at
+	 *    the top of btrfs_chunk_alloc();
+	 *
+	 * 2) Prevent races with the final phase of a device replace operation
+	 *    that replaces the device object associated with the map's stripes,
+	 *    because the device object's id can change at any time during that
+	 *    final phase of the device replace operation
+	 *    (dev-replace.c:btrfs_dev_replace_finishing()), so we could grab the
+	 *    replaced device and then see it with an ID of
+	 *    BTRFS_DEV_REPLACE_DEVID, which would cause a failure when updating
+	 *    the device item, which does not exists on the chunk btree.
+	 *    The finishing phase of device replace acquires both the
+	 *    device_list_mutex and the chunk_mutex, in that order, so we are
+	 *    safe by just acquiring the chunk_mutex.
+	 */
+	trans->removing_chunk = true;
+	mutex_lock(&fs_info->chunk_mutex);
+
+	check_system_chunk(trans, map->type);
+
+	ret = remove_chunk_item(trans, map, chunk_offset);
+	/*
+	 * Normally we should not get -ENOSPC since we reserved space before
+	 * through the call to check_system_chunk().
+	 *
+	 * Despite our system space_info having enough free space, we may not
+	 * be able to allocate extents from its block groups, because all have
+	 * an incompatible profile, which will force us to allocate a new system
+	 * block group with the right profile, or right after we called
+	 * check_system_space() above, a scrub turned the only system block group
+	 * with enough free space into RO mode.
+	 * This is explained with more detail at do_chunk_alloc().
+	 *
+	 * So if we get -ENOSPC, allocate a new system chunk and retry once.
+	 */
+	if (ret == -ENOSPC) {
+		const u64 sys_flags = btrfs_system_alloc_profile(fs_info);
+		struct btrfs_block_group *sys_bg;
+
+		sys_bg = btrfs_create_chunk(trans, sys_flags);
+		if (IS_ERR(sys_bg)) {
+			ret = PTR_ERR(sys_bg);
+			btrfs_abort_transaction(trans, ret);
+			goto out;
+		}
+
+		ret = btrfs_chunk_alloc_add_chunk_item(trans, sys_bg);
 		if (ret) {
-			mutex_unlock(&fs_devices->device_list_mutex);
 			btrfs_abort_transaction(trans, ret);
 			goto out;
 		}
-	}
-	mutex_unlock(&fs_devices->device_list_mutex);
 
-	ret = btrfs_free_chunk(trans, chunk_offset);
-	if (ret) {
+		ret = remove_chunk_item(trans, map, chunk_offset);
+		if (ret) {
+			btrfs_abort_transaction(trans, ret);
+			goto out;
+		}
+	} else if (ret) {
 		btrfs_abort_transaction(trans, ret);
 		goto out;
 	}
@@ -2900,6 +3202,15 @@ int btrfs_remove_chunk(struct btrfs_trans_handle *trans, u64 chunk_offset)
 		}
 	}
 
+	mutex_unlock(&fs_info->chunk_mutex);
+	trans->removing_chunk = false;
+
+	/*
+	 * We are done with chunk btree updates and deletions, so release the
+	 * system space we previously reserved (with check_system_chunk()).
+	 */
+	btrfs_trans_release_chunk_metadata(trans);
+
 	ret = btrfs_remove_block_group(trans, chunk_offset, em);
 	if (ret) {
 		btrfs_abort_transaction(trans, ret);
@@ -2907,18 +3218,29 @@ int btrfs_remove_chunk(struct btrfs_trans_handle *trans, u64 chunk_offset)
 	}
 
 out:
+	if (trans->removing_chunk) {
+		mutex_unlock(&fs_info->chunk_mutex);
+		trans->removing_chunk = false;
+	}
 	/* once for us */
 	free_extent_map(em);
 	return ret;
 }
 
-static int btrfs_relocate_chunk(struct btrfs_fs_info *fs_info, u64 chunk_offset)
+int btrfs_relocate_chunk(struct btrfs_fs_info *fs_info, u64 chunk_offset)
 {
 	struct btrfs_root *root = fs_info->chunk_root;
 	struct btrfs_trans_handle *trans;
 	struct btrfs_block_group *block_group;
+	u64 length;
 	int ret;
 
+	if (btrfs_fs_incompat(fs_info, EXTENT_TREE_V2)) {
+		btrfs_err(fs_info,
+			  "relocate: not supported on extent tree v2 yet");
+		return -EINVAL;
+	}
+
 	/*
 	 * Prevent races with automatic removal of unused block groups.
 	 * After we relocate and before we remove the chunk with offset
@@ -2931,7 +3253,7 @@ static int btrfs_relocate_chunk(struct btrfs_fs_info *fs_info, u64 chunk_offset)
 	 * we release the path used to search the chunk/dev tree and before
 	 * the current task acquires this mutex and calls us.
 	 */
-	lockdep_assert_held(&fs_info->delete_unused_bgs_mutex);
+	lockdep_assert_held(&fs_info->reclaim_bgs_lock);
 
 	/* step one, relocate all the extents inside this chunk */
 	btrfs_scrub_pause(fs_info);
@@ -2944,8 +3266,23 @@ static int btrfs_relocate_chunk(struct btrfs_fs_info *fs_info, u64 chunk_offset)
 	if (!block_group)
 		return -ENOENT;
 	btrfs_discard_cancel_work(&fs_info->discard_ctl, block_group);
+	length = block_group->length;
 	btrfs_put_block_group(block_group);
 
+	/*
+	 * On a zoned file system, discard the whole block group, this will
+	 * trigger a REQ_OP_ZONE_RESET operation on the device zone. If
+	 * resetting the zone fails, don't treat it as a fatal problem from the
+	 * filesystem's point of view.
+	 */
+	if (btrfs_is_zoned(fs_info)) {
+		ret = btrfs_discard_extent(fs_info, chunk_offset, length, NULL);
+		if (ret)
+			btrfs_info(fs_info,
+				"failed to reset zone %llu after relocation",
+				chunk_offset);
+	}
+
 	trans = btrfs_start_trans_remove_block_group(root->fs_info,
 						     chunk_offset);
 	if (IS_ERR(trans)) {
@@ -2986,10 +3323,10 @@ again:
 	key.type = BTRFS_CHUNK_ITEM_KEY;
 
 	while (1) {
-		mutex_lock(&fs_info->delete_unused_bgs_mutex);
+		mutex_lock(&fs_info->reclaim_bgs_lock);
 		ret = btrfs_search_slot(NULL, chunk_root, &key, path, 0, 0);
 		if (ret < 0) {
-			mutex_unlock(&fs_info->delete_unused_bgs_mutex);
+			mutex_unlock(&fs_info->reclaim_bgs_lock);
 			goto error;
 		}
 		BUG_ON(ret == 0); /* Corruption */
@@ -2997,7 +3334,7 @@ again:
 		ret = btrfs_previous_item(chunk_root, path, key.objectid,
 					  key.type);
 		if (ret)
-			mutex_unlock(&fs_info->delete_unused_bgs_mutex);
+			mutex_unlock(&fs_info->reclaim_bgs_lock);
 		if (ret < 0)
 			goto error;
 		if (ret > 0)
@@ -3018,7 +3355,7 @@ again:
 			else
 				BUG_ON(ret);
 		}
-		mutex_unlock(&fs_info->delete_unused_bgs_mutex);
+		mutex_unlock(&fs_info->reclaim_bgs_lock);
 
 		if (found_key.offset == 0)
 			break;
@@ -3145,7 +3482,7 @@ static int del_balance_item(struct btrfs_fs_info *fs_info)
 	if (!path)
 		return -ENOMEM;
 
-	trans = btrfs_start_transaction(root, 0);
+	trans = btrfs_start_transaction_fallback_global_rsv(root, 0);
 	if (IS_ERR(trans)) {
 		btrfs_free_path(path);
 		return PTR_ERR(trans);
@@ -3331,10 +3668,7 @@ static u64 calc_data_stripes(u64 type, int num_stripes)
 	const int ncopies = btrfs_raid_array[index].ncopies;
 	const int nparity = btrfs_raid_array[index].nparity;
 
-	if (nparity)
-		return num_stripes - nparity;
-	else
-		return num_stripes / ncopies;
+	return (num_stripes - nparity) / ncopies;
 }
 
 /* [pstart, pend) */
@@ -3558,10 +3892,10 @@ again:
 			goto error;
 		}
 
-		mutex_lock(&fs_info->delete_unused_bgs_mutex);
+		mutex_lock(&fs_info->reclaim_bgs_lock);
 		ret = btrfs_search_slot(NULL, chunk_root, &key, path, 0, 0);
 		if (ret < 0) {
-			mutex_unlock(&fs_info->delete_unused_bgs_mutex);
+			mutex_unlock(&fs_info->reclaim_bgs_lock);
 			goto error;
 		}
 
@@ -3575,7 +3909,7 @@ again:
 		ret = btrfs_previous_item(chunk_root, path, 0,
 					  BTRFS_CHUNK_ITEM_KEY);
 		if (ret) {
-			mutex_unlock(&fs_info->delete_unused_bgs_mutex);
+			mutex_unlock(&fs_info->reclaim_bgs_lock);
 			ret = 0;
 			break;
 		}
@@ -3585,7 +3919,7 @@ again:
 		btrfs_item_key_to_cpu(leaf, &found_key, slot);
 
 		if (found_key.objectid != key.objectid) {
-			mutex_unlock(&fs_info->delete_unused_bgs_mutex);
+			mutex_unlock(&fs_info->reclaim_bgs_lock);
 			break;
 		}
 
@@ -3602,12 +3936,12 @@ again:
 
 		btrfs_release_path(path);
 		if (!ret) {
-			mutex_unlock(&fs_info->delete_unused_bgs_mutex);
+			mutex_unlock(&fs_info->reclaim_bgs_lock);
 			goto loop;
 		}
 
 		if (counting) {
-			mutex_unlock(&fs_info->delete_unused_bgs_mutex);
+			mutex_unlock(&fs_info->reclaim_bgs_lock);
 			spin_lock(&fs_info->balance_lock);
 			bctl->stat.expected++;
 			spin_unlock(&fs_info->balance_lock);
@@ -3632,7 +3966,7 @@ again:
 					count_meta < bctl->meta.limit_min)
 				|| ((chunk_type & BTRFS_BLOCK_GROUP_SYSTEM) &&
 					count_sys < bctl->sys.limit_min)) {
-			mutex_unlock(&fs_info->delete_unused_bgs_mutex);
+			mutex_unlock(&fs_info->reclaim_bgs_lock);
 			goto loop;
 		}
 
@@ -3646,7 +3980,7 @@ again:
 			ret = btrfs_may_alloc_data_chunk(fs_info,
 							 found_key.offset);
 			if (ret < 0) {
-				mutex_unlock(&fs_info->delete_unused_bgs_mutex);
+				mutex_unlock(&fs_info->reclaim_bgs_lock);
 				goto error;
 			} else if (ret == 1) {
 				chunk_reserved = 1;
@@ -3654,7 +3988,7 @@ again:
 		}
 
 		ret = btrfs_relocate_chunk(fs_info, found_key.offset);
-		mutex_unlock(&fs_info->delete_unused_bgs_mutex);
+		mutex_unlock(&fs_info->reclaim_bgs_lock);
 		if (ret == -ENOSPC) {
 			enospc_errors++;
 		} else if (ret == -ETXTBSY) {
@@ -3723,13 +4057,25 @@ static inline int balance_need_close(struct btrfs_fs_info *fs_info)
 		 atomic_read(&fs_info->balance_cancel_req) == 0);
 }
 
-/* Non-zero return value signifies invalidity */
-static inline int validate_convert_profile(struct btrfs_balance_args *bctl_arg,
-		u64 allowed)
+/*
+ * Validate target profile against allowed profiles and return true if it's OK.
+ * Otherwise print the error message and return false.
+ */
+static inline int validate_convert_profile(struct btrfs_fs_info *fs_info,
+		const struct btrfs_balance_args *bargs,
+		u64 allowed, const char *type)
 {
-	return ((bctl_arg->flags & BTRFS_BALANCE_ARGS_CONVERT) &&
-		(!alloc_profile_is_valid(bctl_arg->target, 1) ||
-		 (bctl_arg->target & ~allowed)));
+	if (!(bargs->flags & BTRFS_BALANCE_ARGS_CONVERT))
+		return true;
+
+	/* Profile is valid and does not have bits outside of the allowed set */
+	if (alloc_profile_is_valid(bargs->target, 1) &&
+	    (bargs->target & ~allowed) == 0)
+		return true;
+
+	btrfs_err(fs_info, "balance: invalid convert %s profile %s",
+			type, btrfs_bg_type_to_raid_name(bargs->target));
+	return false;
 }
 
 /*
@@ -3904,7 +4250,7 @@ int btrfs_balance(struct btrfs_fs_info *fs_info,
 
 	if (btrfs_fs_closing(fs_info) ||
 	    atomic_read(&fs_info->balance_pause_req) ||
-	    atomic_read(&fs_info->balance_cancel_req)) {
+	    btrfs_should_cancel_balance(fs_info)) {
 		ret = -EINVAL;
 		goto out;
 	}
@@ -3931,7 +4277,7 @@ int btrfs_balance(struct btrfs_fs_info *fs_info,
 
 	/*
 	 * rw_devices will not change at the moment, device add/delete/replace
-	 * are excluded by EXCL_OP
+	 * are exclusive
 	 */
 	num_devices = fs_info->fs_devices->rw_devices;
 
@@ -3945,24 +4291,9 @@ int btrfs_balance(struct btrfs_fs_info *fs_info,
 		if (num_devices >= btrfs_raid_array[i].devs_min)
 			allowed |= btrfs_raid_array[i].bg_flag;
 
-	if (validate_convert_profile(&bctl->data, allowed)) {
-		btrfs_err(fs_info,
-			  "balance: invalid convert data profile %s",
-			  btrfs_bg_type_to_raid_name(bctl->data.target));
-		ret = -EINVAL;
-		goto out;
-	}
-	if (validate_convert_profile(&bctl->meta, allowed)) {
-		btrfs_err(fs_info,
-			  "balance: invalid convert metadata profile %s",
-			  btrfs_bg_type_to_raid_name(bctl->meta.target));
-		ret = -EINVAL;
-		goto out;
-	}
-	if (validate_convert_profile(&bctl->sys, allowed)) {
-		btrfs_err(fs_info,
-			  "balance: invalid convert system profile %s",
-			  btrfs_bg_type_to_raid_name(bctl->sys.target));
+	if (!validate_convert_profile(fs_info, &bctl->data, allowed, "data") ||
+	    !validate_convert_profile(fs_info, &bctl->meta, allowed, "metadata") ||
+	    !validate_convert_profile(fs_info, &bctl->sys,  allowed, "system")) {
 		ret = -EINVAL;
 		goto out;
 	}
@@ -4017,14 +4348,6 @@ int btrfs_balance(struct btrfs_fs_info *fs_info,
 				btrfs_bg_type_to_raid_name(data_target));
 	}
 
-	if (fs_info->send_in_progress) {
-		btrfs_warn_rl(fs_info,
-"cannot run balance while send operations are in progress (%d in progress)",
-			      fs_info->send_in_progress);
-		ret = -EAGAIN;
-		goto out;
-	}
-
 	ret = insert_balance_item(fs_info, bctl);
 	if (ret && ret != -EEXIST)
 		goto out;
@@ -4050,9 +4373,26 @@ int btrfs_balance(struct btrfs_fs_info *fs_info,
 	ret = __btrfs_balance(fs_info);
 
 	mutex_lock(&fs_info->balance_mutex);
-	if (ret == -ECANCELED && atomic_read(&fs_info->balance_pause_req))
+	if (ret == -ECANCELED && atomic_read(&fs_info->balance_pause_req)) {
 		btrfs_info(fs_info, "balance: paused");
-	else if (ret == -ECANCELED && atomic_read(&fs_info->balance_cancel_req))
+		btrfs_exclop_balance(fs_info, BTRFS_EXCLOP_BALANCE_PAUSED);
+	}
+	/*
+	 * Balance can be canceled by:
+	 *
+	 * - Regular cancel request
+	 *   Then ret == -ECANCELED and balance_cancel_req > 0
+	 *
+	 * - Fatal signal to "btrfs" process
+	 *   Either the signal caught by wait_reserve_ticket() and callers
+	 *   got -EINTR, or caught by btrfs_should_cancel_balance() and
+	 *   got -ECANCELED.
+	 *   Either way, in this case balance_cancel_req = 0, and
+	 *   ret == -EINTR or ret == -ECANCELED.
+	 *
+	 * So here we only check the return value to catch canceled balance.
+	 */
+	else if (ret == -ECANCELED || ret == -EINTR)
 		btrfs_info(fs_info, "balance: canceled");
 	else
 		btrfs_info(fs_info, "balance: ended with status: %d", ret);
@@ -4067,7 +4407,7 @@ int btrfs_balance(struct btrfs_fs_info *fs_info,
 	if ((ret && ret != -ECANCELED && ret != -ENOSPC) ||
 	    balance_need_close(fs_info)) {
 		reset_balance_state(fs_info);
-		clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
+		btrfs_exclop_finish(fs_info);
 	}
 
 	wake_up(&fs_info->balance_wait_q);
@@ -4078,7 +4418,7 @@ out:
 		reset_balance_state(fs_info);
 	else
 		kfree(bctl);
-	clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
+	btrfs_exclop_finish(fs_info);
 
 	return ret;
 }
@@ -4088,10 +4428,12 @@ static int balance_kthread(void *data)
 	struct btrfs_fs_info *fs_info = data;
 	int ret = 0;
 
+	sb_start_write(fs_info->sb);
 	mutex_lock(&fs_info->balance_mutex);
 	if (fs_info->balance_ctl)
 		ret = btrfs_balance(fs_info, fs_info->balance_ctl, NULL);
 	mutex_unlock(&fs_info->balance_mutex);
+	sb_end_write(fs_info->sb);
 
 	return ret;
 }
@@ -4112,6 +4454,10 @@ int btrfs_resume_balance_async(struct btrfs_fs_info *fs_info)
 		return 0;
 	}
 
+	spin_lock(&fs_info->super_lock);
+	ASSERT(fs_info->exclusive_operation == BTRFS_EXCLOP_BALANCE_PAUSED);
+	fs_info->exclusive_operation = BTRFS_EXCLOP_BALANCE;
+	spin_unlock(&fs_info->super_lock);
 	/*
 	 * A ro->rw remount sequence should continue with the paused balance
 	 * regardless of who pauses it, system or the user as of now, so set
@@ -4180,10 +4526,12 @@ int btrfs_recover_balance(struct btrfs_fs_info *fs_info)
 	 * is in a paused state and must have fs_info::balance_ctl properly
 	 * set up.
 	 */
-	if (test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags))
+	if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_BALANCE_PAUSED))
 		btrfs_warn(fs_info,
 	"balance: cannot set exclusive op status, resume manually");
 
+	btrfs_release_path(path);
+
 	mutex_lock(&fs_info->balance_mutex);
 	BUG_ON(fs_info->balance_ctl);
 	spin_lock(&fs_info->balance_lock);
@@ -4262,7 +4610,7 @@ int btrfs_cancel_balance(struct btrfs_fs_info *fs_info)
 
 		if (fs_info->balance_ctl) {
 			reset_balance_state(fs_info);
-			clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
+			btrfs_exclop_finish(fs_info);
 			btrfs_info(fs_info, "balance: canceled");
 		}
 	}
@@ -4274,7 +4622,7 @@ int btrfs_cancel_balance(struct btrfs_fs_info *fs_info)
 	return 0;
 }
 
-static int btrfs_uuid_scan_kthread(void *data)
+int btrfs_uuid_scan_kthread(void *data)
 {
 	struct btrfs_fs_info *fs_info = data;
 	struct btrfs_root *root = fs_info->tree_root;
@@ -4286,6 +4634,7 @@ static int btrfs_uuid_scan_kthread(void *data)
 	struct btrfs_root_item root_item;
 	u32 item_size;
 	struct btrfs_trans_handle *trans = NULL;
+	bool closing = false;
 
 	path = btrfs_alloc_path();
 	if (!path) {
@@ -4298,6 +4647,10 @@ static int btrfs_uuid_scan_kthread(void *data)
 	key.offset = 0;
 
 	while (1) {
+		if (btrfs_fs_closing(fs_info)) {
+			closing = true;
+			break;
+		}
 		ret = btrfs_search_forward(root, &key, path,
 				BTRFS_OLDEST_GENERATION);
 		if (ret) {
@@ -4314,7 +4667,7 @@ static int btrfs_uuid_scan_kthread(void *data)
 
 		eb = path->nodes[0];
 		slot = path->slots[0];
-		item_size = btrfs_item_size_nr(eb, slot);
+		item_size = btrfs_item_size(eb, slot);
 		if (item_size < sizeof(root_item))
 			goto skip;
 
@@ -4344,6 +4697,7 @@ static int btrfs_uuid_scan_kthread(void *data)
 			goto skip;
 		}
 update_tree:
+		btrfs_release_path(path);
 		if (!btrfs_is_empty_uuid(root_item.uuid)) {
 			ret = btrfs_uuid_tree_add(trans, root_item.uuid,
 						  BTRFS_UUID_KEY_SUBVOL,
@@ -4368,6 +4722,7 @@ update_tree:
 		}
 
 skip:
+		btrfs_release_path(path);
 		if (trans) {
 			ret = btrfs_end_transaction(trans);
 			trans = NULL;
@@ -4375,7 +4730,6 @@ skip:
 				break;
 		}
 
-		btrfs_release_path(path);
 		if (key.offset < (u64)-1) {
 			key.offset++;
 		} else if (key.type < BTRFS_ROOT_ITEM_KEY) {
@@ -4397,76 +4751,12 @@ out:
 		btrfs_end_transaction(trans);
 	if (ret)
 		btrfs_warn(fs_info, "btrfs_uuid_scan_kthread failed %d", ret);
-	else
+	else if (!closing)
 		set_bit(BTRFS_FS_UPDATE_UUID_TREE_GEN, &fs_info->flags);
 	up(&fs_info->uuid_tree_rescan_sem);
 	return 0;
 }
 
-/*
- * Callback for btrfs_uuid_tree_iterate().
- * returns:
- * 0	check succeeded, the entry is not outdated.
- * < 0	if an error occurred.
- * > 0	if the check failed, which means the caller shall remove the entry.
- */
-static int btrfs_check_uuid_tree_entry(struct btrfs_fs_info *fs_info,
-				       u8 *uuid, u8 type, u64 subid)
-{
-	struct btrfs_key key;
-	int ret = 0;
-	struct btrfs_root *subvol_root;
-
-	if (type != BTRFS_UUID_KEY_SUBVOL &&
-	    type != BTRFS_UUID_KEY_RECEIVED_SUBVOL)
-		goto out;
-
-	key.objectid = subid;
-	key.type = BTRFS_ROOT_ITEM_KEY;
-	key.offset = (u64)-1;
-	subvol_root = btrfs_read_fs_root_no_name(fs_info, &key);
-	if (IS_ERR(subvol_root)) {
-		ret = PTR_ERR(subvol_root);
-		if (ret == -ENOENT)
-			ret = 1;
-		goto out;
-	}
-
-	switch (type) {
-	case BTRFS_UUID_KEY_SUBVOL:
-		if (memcmp(uuid, subvol_root->root_item.uuid, BTRFS_UUID_SIZE))
-			ret = 1;
-		break;
-	case BTRFS_UUID_KEY_RECEIVED_SUBVOL:
-		if (memcmp(uuid, subvol_root->root_item.received_uuid,
-			   BTRFS_UUID_SIZE))
-			ret = 1;
-		break;
-	}
-
-out:
-	return ret;
-}
-
-static int btrfs_uuid_rescan_kthread(void *data)
-{
-	struct btrfs_fs_info *fs_info = (struct btrfs_fs_info *)data;
-	int ret;
-
-	/*
-	 * 1st step is to iterate through the existing UUID tree and
-	 * to delete all entries that contain outdated data.
-	 * 2nd step is to add all missing entries to the UUID tree.
-	 */
-	ret = btrfs_uuid_tree_iterate(fs_info, btrfs_check_uuid_tree_entry);
-	if (ret < 0) {
-		btrfs_warn(fs_info, "iterating uuid_tree failed %d", ret);
-		up(&fs_info->uuid_tree_rescan_sem);
-		return ret;
-	}
-	return btrfs_uuid_scan_kthread(data);
-}
-
 int btrfs_create_uuid_tree(struct btrfs_fs_info *fs_info)
 {
 	struct btrfs_trans_handle *trans;
@@ -4509,22 +4799,6 @@ int btrfs_create_uuid_tree(struct btrfs_fs_info *fs_info)
 	return 0;
 }
 
-int btrfs_check_uuid_tree(struct btrfs_fs_info *fs_info)
-{
-	struct task_struct *task;
-
-	down(&fs_info->uuid_tree_rescan_sem);
-	task = kthread_run(btrfs_uuid_rescan_kthread, fs_info, "btrfs-uuid");
-	if (IS_ERR(task)) {
-		/* fs_info->update_uuid_tree_gen remains 0 in all error case */
-		btrfs_warn(fs_info, "failed to start uuid_rescan task");
-		up(&fs_info->uuid_tree_rescan_sem);
-		return PTR_ERR(task);
-	}
-
-	return 0;
-}
-
 /*
  * shrinking a device means finding all of the device extents past
  * the new size, and then following the back refs to the chunks.
@@ -4599,19 +4873,18 @@ again:
 	key.type = BTRFS_DEV_EXTENT_KEY;
 
 	do {
-		mutex_lock(&fs_info->delete_unused_bgs_mutex);
+		mutex_lock(&fs_info->reclaim_bgs_lock);
 		ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
 		if (ret < 0) {
-			mutex_unlock(&fs_info->delete_unused_bgs_mutex);
+			mutex_unlock(&fs_info->reclaim_bgs_lock);
 			goto done;
 		}
 
 		ret = btrfs_previous_item(root, path, 0, key.type);
-		if (ret)
-			mutex_unlock(&fs_info->delete_unused_bgs_mutex);
-		if (ret < 0)
-			goto done;
 		if (ret) {
+			mutex_unlock(&fs_info->reclaim_bgs_lock);
+			if (ret < 0)
+				goto done;
 			ret = 0;
 			btrfs_release_path(path);
 			break;
@@ -4622,7 +4895,7 @@ again:
 		btrfs_item_key_to_cpu(l, &key, path->slots[0]);
 
 		if (key.objectid != device->devid) {
-			mutex_unlock(&fs_info->delete_unused_bgs_mutex);
+			mutex_unlock(&fs_info->reclaim_bgs_lock);
 			btrfs_release_path(path);
 			break;
 		}
@@ -4631,7 +4904,7 @@ again:
 		length = btrfs_dev_extent_length(l, dev_extent);
 
 		if (key.offset + length <= new_size) {
-			mutex_unlock(&fs_info->delete_unused_bgs_mutex);
+			mutex_unlock(&fs_info->reclaim_bgs_lock);
 			btrfs_release_path(path);
 			break;
 		}
@@ -4647,12 +4920,12 @@ again:
 		 */
 		ret = btrfs_may_alloc_data_chunk(fs_info, chunk_offset);
 		if (ret < 0) {
-			mutex_unlock(&fs_info->delete_unused_bgs_mutex);
+			mutex_unlock(&fs_info->reclaim_bgs_lock);
 			goto done;
 		}
 
 		ret = btrfs_relocate_chunk(fs_info, chunk_offset);
-		mutex_unlock(&fs_info->delete_unused_bgs_mutex);
+		mutex_unlock(&fs_info->reclaim_bgs_lock);
 		if (ret == -ENOSPC) {
 			failed++;
 		} else if (ret) {
@@ -4682,6 +4955,10 @@ again:
 	}
 
 	mutex_lock(&fs_info->chunk_mutex);
+	/* Clear all state bits beyond the shrunk device size */
+	clear_extent_bits(&device->alloc_state, new_size, (u64)-1,
+			  CHUNK_STATE_MASK);
+
 	btrfs_device_set_disk_total_bytes(device, new_size);
 	if (list_empty(&device->post_commit_list))
 		list_add_tail(&device->post_commit_list,
@@ -4692,8 +4969,10 @@ again:
 			round_down(old_total - diff, fs_info->sectorsize));
 	mutex_unlock(&fs_info->chunk_mutex);
 
+	btrfs_reserve_chunk_metadata(trans, false);
 	/* Now btrfs_update_device() will change the on-disk size. */
 	ret = btrfs_update_device(trans, device);
+	btrfs_trans_release_chunk_metadata(trans);
 	if (ret < 0) {
 		btrfs_abort_transaction(trans, ret);
 		btrfs_end_transaction(trans);
@@ -4722,13 +5001,12 @@ static int btrfs_add_system_chunk(struct btrfs_fs_info *fs_info,
 	u32 array_size;
 	u8 *ptr;
 
-	mutex_lock(&fs_info->chunk_mutex);
+	lockdep_assert_held(&fs_info->chunk_mutex);
+
 	array_size = btrfs_super_sys_array_size(super_copy);
 	if (array_size + item_size + sizeof(disk_key)
-			> BTRFS_SYSTEM_CHUNK_ARRAY_SIZE) {
-		mutex_unlock(&fs_info->chunk_mutex);
+			> BTRFS_SYSTEM_CHUNK_ARRAY_SIZE)
 		return -EFBIG;
-	}
 
 	ptr = super_copy->sys_chunk_array + array_size;
 	btrfs_cpu_key_to_disk(&disk_key, key);
@@ -4737,7 +5015,6 @@ static int btrfs_add_system_chunk(struct btrfs_fs_info *fs_info,
 	memcpy(ptr, chunk, item_size);
 	item_size += sizeof(disk_key);
 	btrfs_set_super_sys_array_size(super_copy, array_size + item_size);
-	mutex_unlock(&fs_info->chunk_mutex);
 
 	return 0;
 }
@@ -4777,96 +5054,137 @@ static void check_raid1c34_incompat_flag(struct btrfs_fs_info *info, u64 type)
 	btrfs_set_fs_incompat(info, RAID1C34);
 }
 
-static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
-			       u64 start, u64 type)
-{
-	struct btrfs_fs_info *info = trans->fs_info;
-	struct btrfs_fs_devices *fs_devices = info->fs_devices;
-	struct btrfs_device *device;
-	struct map_lookup *map = NULL;
-	struct extent_map_tree *em_tree;
-	struct extent_map *em;
-	struct btrfs_device_info *devices_info = NULL;
-	u64 total_avail;
-	int num_stripes;	/* total number of stripes to allocate */
-	int data_stripes;	/* number of stripes that count for
-				   block group size */
-	int sub_stripes;	/* sub_stripes info for map */
-	int dev_stripes;	/* stripes per dev */
-	int devs_max;		/* max devs to use */
-	int devs_min;		/* min devs needed */
-	int devs_increment;	/* ndevs has to be a multiple of this */
-	int ncopies;		/* how many copies to data has */
-	int nparity;		/* number of stripes worth of bytes to
-				   store parity information */
-	int ret;
+/*
+ * Structure used internally for btrfs_create_chunk() function.
+ * Wraps needed parameters.
+ */
+struct alloc_chunk_ctl {
+	u64 start;
+	u64 type;
+	/* Total number of stripes to allocate */
+	int num_stripes;
+	/* sub_stripes info for map */
+	int sub_stripes;
+	/* Stripes per device */
+	int dev_stripes;
+	/* Maximum number of devices to use */
+	int devs_max;
+	/* Minimum number of devices to use */
+	int devs_min;
+	/* ndevs has to be a multiple of this */
+	int devs_increment;
+	/* Number of copies */
+	int ncopies;
+	/* Number of stripes worth of bytes to store parity information */
+	int nparity;
 	u64 max_stripe_size;
 	u64 max_chunk_size;
+	u64 dev_extent_min;
 	u64 stripe_size;
 	u64 chunk_size;
 	int ndevs;
-	int i;
-	int j;
-	int index;
+};
 
-	BUG_ON(!alloc_profile_is_valid(type, 0));
+static void init_alloc_chunk_ctl_policy_regular(
+				struct btrfs_fs_devices *fs_devices,
+				struct alloc_chunk_ctl *ctl)
+{
+	struct btrfs_space_info *space_info;
 
-	if (list_empty(&fs_devices->alloc_list)) {
-		if (btrfs_test_opt(info, ENOSPC_DEBUG))
-			btrfs_debug(info, "%s: no writable device", __func__);
-		return -ENOSPC;
-	}
+	space_info = btrfs_find_space_info(fs_devices->fs_info, ctl->type);
+	ASSERT(space_info);
 
-	index = btrfs_bg_flags_to_raid_index(type);
+	ctl->max_chunk_size = READ_ONCE(space_info->chunk_size);
+	ctl->max_stripe_size = ctl->max_chunk_size;
 
-	sub_stripes = btrfs_raid_array[index].sub_stripes;
-	dev_stripes = btrfs_raid_array[index].dev_stripes;
-	devs_max = btrfs_raid_array[index].devs_max;
-	if (!devs_max)
-		devs_max = BTRFS_MAX_DEVS(info);
-	devs_min = btrfs_raid_array[index].devs_min;
-	devs_increment = btrfs_raid_array[index].devs_increment;
-	ncopies = btrfs_raid_array[index].ncopies;
-	nparity = btrfs_raid_array[index].nparity;
+	if (ctl->type & BTRFS_BLOCK_GROUP_SYSTEM)
+		ctl->devs_max = min_t(int, ctl->devs_max, BTRFS_MAX_DEVS_SYS_CHUNK);
 
+	/* We don't want a chunk larger than 10% of writable space */
+	ctl->max_chunk_size = min(div_factor(fs_devices->total_rw_bytes, 1),
+				  ctl->max_chunk_size);
+	ctl->dev_extent_min = BTRFS_STRIPE_LEN * ctl->dev_stripes;
+}
+
+static void init_alloc_chunk_ctl_policy_zoned(
+				      struct btrfs_fs_devices *fs_devices,
+				      struct alloc_chunk_ctl *ctl)
+{
+	u64 zone_size = fs_devices->fs_info->zone_size;
+	u64 limit;
+	int min_num_stripes = ctl->devs_min * ctl->dev_stripes;
+	int min_data_stripes = (min_num_stripes - ctl->nparity) / ctl->ncopies;
+	u64 min_chunk_size = min_data_stripes * zone_size;
+	u64 type = ctl->type;
+
+	ctl->max_stripe_size = zone_size;
 	if (type & BTRFS_BLOCK_GROUP_DATA) {
-		max_stripe_size = SZ_1G;
-		max_chunk_size = BTRFS_MAX_DATA_CHUNK_SIZE;
+		ctl->max_chunk_size = round_down(BTRFS_MAX_DATA_CHUNK_SIZE,
+						 zone_size);
 	} else if (type & BTRFS_BLOCK_GROUP_METADATA) {
-		/* for larger filesystems, use larger metadata chunks */
-		if (fs_devices->total_rw_bytes > 50ULL * SZ_1G)
-			max_stripe_size = SZ_1G;
-		else
-			max_stripe_size = SZ_256M;
-		max_chunk_size = max_stripe_size;
+		ctl->max_chunk_size = ctl->max_stripe_size;
 	} else if (type & BTRFS_BLOCK_GROUP_SYSTEM) {
-		max_stripe_size = SZ_32M;
-		max_chunk_size = 2 * max_stripe_size;
-		devs_max = min_t(int, devs_max, BTRFS_MAX_DEVS_SYS_CHUNK);
+		ctl->max_chunk_size = 2 * ctl->max_stripe_size;
+		ctl->devs_max = min_t(int, ctl->devs_max,
+				      BTRFS_MAX_DEVS_SYS_CHUNK);
 	} else {
-		btrfs_err(info, "invalid chunk type 0x%llx requested",
-		       type);
 		BUG();
 	}
 
 	/* We don't want a chunk larger than 10% of writable space */
-	max_chunk_size = min(div_factor(fs_devices->total_rw_bytes, 1),
-			     max_chunk_size);
+	limit = max(round_down(div_factor(fs_devices->total_rw_bytes, 1),
+			       zone_size),
+		    min_chunk_size);
+	ctl->max_chunk_size = min(limit, ctl->max_chunk_size);
+	ctl->dev_extent_min = zone_size * ctl->dev_stripes;
+}
+
+static void init_alloc_chunk_ctl(struct btrfs_fs_devices *fs_devices,
+				 struct alloc_chunk_ctl *ctl)
+{
+	int index = btrfs_bg_flags_to_raid_index(ctl->type);
+
+	ctl->sub_stripes = btrfs_raid_array[index].sub_stripes;
+	ctl->dev_stripes = btrfs_raid_array[index].dev_stripes;
+	ctl->devs_max = btrfs_raid_array[index].devs_max;
+	if (!ctl->devs_max)
+		ctl->devs_max = BTRFS_MAX_DEVS(fs_devices->fs_info);
+	ctl->devs_min = btrfs_raid_array[index].devs_min;
+	ctl->devs_increment = btrfs_raid_array[index].devs_increment;
+	ctl->ncopies = btrfs_raid_array[index].ncopies;
+	ctl->nparity = btrfs_raid_array[index].nparity;
+	ctl->ndevs = 0;
+
+	switch (fs_devices->chunk_alloc_policy) {
+	case BTRFS_CHUNK_ALLOC_REGULAR:
+		init_alloc_chunk_ctl_policy_regular(fs_devices, ctl);
+		break;
+	case BTRFS_CHUNK_ALLOC_ZONED:
+		init_alloc_chunk_ctl_policy_zoned(fs_devices, ctl);
+		break;
+	default:
+		BUG();
+	}
+}
 
-	devices_info = kcalloc(fs_devices->rw_devices, sizeof(*devices_info),
-			       GFP_NOFS);
-	if (!devices_info)
-		return -ENOMEM;
+static int gather_device_info(struct btrfs_fs_devices *fs_devices,
+			      struct alloc_chunk_ctl *ctl,
+			      struct btrfs_device_info *devices_info)
+{
+	struct btrfs_fs_info *info = fs_devices->fs_info;
+	struct btrfs_device *device;
+	u64 total_avail;
+	u64 dev_extent_want = ctl->max_stripe_size * ctl->dev_stripes;
+	int ret;
+	int ndevs = 0;
+	u64 max_avail;
+	u64 dev_offset;
 
 	/*
 	 * in the first pass through the devices list, we gather information
 	 * about the available holes on each device.
 	 */
-	ndevs = 0;
 	list_for_each_entry(device, &fs_devices->alloc_list, dev_alloc_list) {
-		u64 max_avail;
-		u64 dev_offset;
-
 		if (!test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state)) {
 			WARN(1, KERN_ERR
 			       "BTRFS: read-only device in alloc_list\n");
@@ -4884,24 +5202,23 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
 			total_avail = 0;
 
 		/* If there is no space on this device, skip it. */
-		if (total_avail == 0)
+		if (total_avail < ctl->dev_extent_min)
 			continue;
 
-		ret = find_free_dev_extent(device,
-					   max_stripe_size * dev_stripes,
-					   &dev_offset, &max_avail);
+		ret = find_free_dev_extent(device, dev_extent_want, &dev_offset,
+					   &max_avail);
 		if (ret && ret != -ENOSPC)
-			goto error;
+			return ret;
 
 		if (ret == 0)
-			max_avail = max_stripe_size * dev_stripes;
+			max_avail = dev_extent_want;
 
-		if (max_avail < BTRFS_STRIPE_LEN * dev_stripes) {
+		if (max_avail < ctl->dev_extent_min) {
 			if (btrfs_test_opt(info, ENOSPC_DEBUG))
 				btrfs_debug(info,
-			"%s: devid %llu has no free space, have=%llu want=%u",
+			"%s: devid %llu has no free space, have=%llu want=%llu",
 					    __func__, device->devid, max_avail,
-					    BTRFS_STRIPE_LEN * dev_stripes);
+					    ctl->dev_extent_min);
 			continue;
 		}
 
@@ -4916,6 +5233,7 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
 		devices_info[ndevs].dev = device;
 		++ndevs;
 	}
+	ctl->ndevs = ndevs;
 
 	/*
 	 * now sort the devices by hole size / available space
@@ -4923,23 +5241,14 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
 	sort(devices_info, ndevs, sizeof(struct btrfs_device_info),
 	     btrfs_cmp_device_info, NULL);
 
-	/*
-	 * Round down to number of usable stripes, devs_increment can be any
-	 * number so we can't use round_down()
-	 */
-	ndevs -= ndevs % devs_increment;
-
-	if (ndevs < devs_min) {
-		ret = -ENOSPC;
-		if (btrfs_test_opt(info, ENOSPC_DEBUG)) {
-			btrfs_debug(info,
-	"%s: not enough devices with free space: have=%d minimum required=%d",
-				    __func__, ndevs, devs_min);
-		}
-		goto error;
-	}
+	return 0;
+}
 
-	ndevs = min(ndevs, devs_max);
+static int decide_stripe_size_regular(struct alloc_chunk_ctl *ctl,
+				      struct btrfs_device_info *devices_info)
+{
+	/* Number of stripes that count for block group size */
+	int data_stripes;
 
 	/*
 	 * The primary goal is to maximize the number of stripes, so use as
@@ -4948,73 +5257,154 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
 	 * The DUP profile stores more than one stripe per device, the
 	 * max_avail is the total size so we have to adjust.
 	 */
-	stripe_size = div_u64(devices_info[ndevs - 1].max_avail, dev_stripes);
-	num_stripes = ndevs * dev_stripes;
+	ctl->stripe_size = div_u64(devices_info[ctl->ndevs - 1].max_avail,
+				   ctl->dev_stripes);
+	ctl->num_stripes = ctl->ndevs * ctl->dev_stripes;
 
-	/*
-	 * this will have to be fixed for RAID1 and RAID10 over
-	 * more drives
-	 */
-	data_stripes = (num_stripes - nparity) / ncopies;
+	/* This will have to be fixed for RAID1 and RAID10 over more drives */
+	data_stripes = (ctl->num_stripes - ctl->nparity) / ctl->ncopies;
 
 	/*
-	 * Use the number of data stripes to figure out how big this chunk
-	 * is really going to be in terms of logical address space,
-	 * and compare that answer with the max chunk size. If it's higher,
-	 * we try to reduce stripe_size.
+	 * Use the number of data stripes to figure out how big this chunk is
+	 * really going to be in terms of logical address space, and compare
+	 * that answer with the max chunk size. If it's higher, we try to
+	 * reduce stripe_size.
 	 */
-	if (stripe_size * data_stripes > max_chunk_size) {
+	if (ctl->stripe_size * data_stripes > ctl->max_chunk_size) {
 		/*
 		 * Reduce stripe_size, round it up to a 16MB boundary again and
 		 * then use it, unless it ends up being even bigger than the
 		 * previous value we had already.
 		 */
-		stripe_size = min(round_up(div_u64(max_chunk_size,
-						   data_stripes), SZ_16M),
-				  stripe_size);
+		ctl->stripe_size = min(round_up(div_u64(ctl->max_chunk_size,
+							data_stripes), SZ_16M),
+				       ctl->stripe_size);
 	}
 
-	/* align to BTRFS_STRIPE_LEN */
-	stripe_size = round_down(stripe_size, BTRFS_STRIPE_LEN);
+	/* Stripe size should not go beyond 1G. */
+	ctl->stripe_size = min_t(u64, ctl->stripe_size, SZ_1G);
 
-	map = kmalloc(map_lookup_size(num_stripes), GFP_NOFS);
-	if (!map) {
-		ret = -ENOMEM;
-		goto error;
+	/* Align to BTRFS_STRIPE_LEN */
+	ctl->stripe_size = round_down(ctl->stripe_size, BTRFS_STRIPE_LEN);
+	ctl->chunk_size = ctl->stripe_size * data_stripes;
+
+	return 0;
+}
+
+static int decide_stripe_size_zoned(struct alloc_chunk_ctl *ctl,
+				    struct btrfs_device_info *devices_info)
+{
+	u64 zone_size = devices_info[0].dev->zone_info->zone_size;
+	/* Number of stripes that count for block group size */
+	int data_stripes;
+
+	/*
+	 * It should hold because:
+	 *    dev_extent_min == dev_extent_want == zone_size * dev_stripes
+	 */
+	ASSERT(devices_info[ctl->ndevs - 1].max_avail == ctl->dev_extent_min);
+
+	ctl->stripe_size = zone_size;
+	ctl->num_stripes = ctl->ndevs * ctl->dev_stripes;
+	data_stripes = (ctl->num_stripes - ctl->nparity) / ctl->ncopies;
+
+	/* stripe_size is fixed in zoned filesysmte. Reduce ndevs instead. */
+	if (ctl->stripe_size * data_stripes > ctl->max_chunk_size) {
+		ctl->ndevs = div_u64(div_u64(ctl->max_chunk_size * ctl->ncopies,
+					     ctl->stripe_size) + ctl->nparity,
+				     ctl->dev_stripes);
+		ctl->num_stripes = ctl->ndevs * ctl->dev_stripes;
+		data_stripes = (ctl->num_stripes - ctl->nparity) / ctl->ncopies;
+		ASSERT(ctl->stripe_size * data_stripes <= ctl->max_chunk_size);
 	}
-	map->num_stripes = num_stripes;
 
-	for (i = 0; i < ndevs; ++i) {
-		for (j = 0; j < dev_stripes; ++j) {
-			int s = i * dev_stripes + j;
+	ctl->chunk_size = ctl->stripe_size * data_stripes;
+
+	return 0;
+}
+
+static int decide_stripe_size(struct btrfs_fs_devices *fs_devices,
+			      struct alloc_chunk_ctl *ctl,
+			      struct btrfs_device_info *devices_info)
+{
+	struct btrfs_fs_info *info = fs_devices->fs_info;
+
+	/*
+	 * Round down to number of usable stripes, devs_increment can be any
+	 * number so we can't use round_down() that requires power of 2, while
+	 * rounddown is safe.
+	 */
+	ctl->ndevs = rounddown(ctl->ndevs, ctl->devs_increment);
+
+	if (ctl->ndevs < ctl->devs_min) {
+		if (btrfs_test_opt(info, ENOSPC_DEBUG)) {
+			btrfs_debug(info,
+	"%s: not enough devices with free space: have=%d minimum required=%d",
+				    __func__, ctl->ndevs, ctl->devs_min);
+		}
+		return -ENOSPC;
+	}
+
+	ctl->ndevs = min(ctl->ndevs, ctl->devs_max);
+
+	switch (fs_devices->chunk_alloc_policy) {
+	case BTRFS_CHUNK_ALLOC_REGULAR:
+		return decide_stripe_size_regular(ctl, devices_info);
+	case BTRFS_CHUNK_ALLOC_ZONED:
+		return decide_stripe_size_zoned(ctl, devices_info);
+	default:
+		BUG();
+	}
+}
+
+static struct btrfs_block_group *create_chunk(struct btrfs_trans_handle *trans,
+			struct alloc_chunk_ctl *ctl,
+			struct btrfs_device_info *devices_info)
+{
+	struct btrfs_fs_info *info = trans->fs_info;
+	struct map_lookup *map = NULL;
+	struct extent_map_tree *em_tree;
+	struct btrfs_block_group *block_group;
+	struct extent_map *em;
+	u64 start = ctl->start;
+	u64 type = ctl->type;
+	int ret;
+	int i;
+	int j;
+
+	map = kmalloc(map_lookup_size(ctl->num_stripes), GFP_NOFS);
+	if (!map)
+		return ERR_PTR(-ENOMEM);
+	map->num_stripes = ctl->num_stripes;
+
+	for (i = 0; i < ctl->ndevs; ++i) {
+		for (j = 0; j < ctl->dev_stripes; ++j) {
+			int s = i * ctl->dev_stripes + j;
 			map->stripes[s].dev = devices_info[i].dev;
 			map->stripes[s].physical = devices_info[i].dev_offset +
-						   j * stripe_size;
+						   j * ctl->stripe_size;
 		}
 	}
 	map->stripe_len = BTRFS_STRIPE_LEN;
 	map->io_align = BTRFS_STRIPE_LEN;
 	map->io_width = BTRFS_STRIPE_LEN;
 	map->type = type;
-	map->sub_stripes = sub_stripes;
+	map->sub_stripes = ctl->sub_stripes;
 
-	chunk_size = stripe_size * data_stripes;
-
-	trace_btrfs_chunk_alloc(info, map, start, chunk_size);
+	trace_btrfs_chunk_alloc(info, map, start, ctl->chunk_size);
 
 	em = alloc_extent_map();
 	if (!em) {
 		kfree(map);
-		ret = -ENOMEM;
-		goto error;
+		return ERR_PTR(-ENOMEM);
 	}
 	set_bit(EXTENT_FLAG_FS_MAPPING, &em->flags);
 	em->map_lookup = map;
 	em->start = start;
-	em->len = chunk_size;
+	em->len = ctl->chunk_size;
 	em->block_start = 0;
 	em->block_len = em->len;
-	em->orig_block_len = stripe_size;
+	em->orig_block_len = ctl->stripe_size;
 
 	em_tree = &info->mapping_tree;
 	write_lock(&em_tree->lock);
@@ -5022,31 +5412,32 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
 	if (ret) {
 		write_unlock(&em_tree->lock);
 		free_extent_map(em);
-		goto error;
+		return ERR_PTR(ret);
 	}
 	write_unlock(&em_tree->lock);
 
-	ret = btrfs_make_block_group(trans, 0, type, start, chunk_size);
-	if (ret)
+	block_group = btrfs_make_block_group(trans, 0, type, start, ctl->chunk_size);
+	if (IS_ERR(block_group))
 		goto error_del_extent;
 
 	for (i = 0; i < map->num_stripes; i++) {
 		struct btrfs_device *dev = map->stripes[i].dev;
 
-		btrfs_device_set_bytes_used(dev, dev->bytes_used + stripe_size);
+		btrfs_device_set_bytes_used(dev,
+					    dev->bytes_used + ctl->stripe_size);
 		if (list_empty(&dev->post_commit_list))
 			list_add_tail(&dev->post_commit_list,
 				      &trans->transaction->dev_update_list);
 	}
 
-	atomic64_sub(stripe_size * map->num_stripes, &info->free_chunk_space);
+	atomic64_sub(ctl->stripe_size * map->num_stripes,
+		     &info->free_chunk_space);
 
 	free_extent_map(em);
 	check_raid56_incompat_flag(info, type);
 	check_raid1c34_incompat_flag(info, type);
 
-	kfree(devices_info);
-	return 0;
+	return block_group;
 
 error_del_extent:
 	write_lock(&em_tree->lock);
@@ -5057,82 +5448,151 @@ error_del_extent:
 	free_extent_map(em);
 	/* One for the tree reference */
 	free_extent_map(em);
-error:
+
+	return block_group;
+}
+
+struct btrfs_block_group *btrfs_create_chunk(struct btrfs_trans_handle *trans,
+					    u64 type)
+{
+	struct btrfs_fs_info *info = trans->fs_info;
+	struct btrfs_fs_devices *fs_devices = info->fs_devices;
+	struct btrfs_device_info *devices_info = NULL;
+	struct alloc_chunk_ctl ctl;
+	struct btrfs_block_group *block_group;
+	int ret;
+
+	lockdep_assert_held(&info->chunk_mutex);
+
+	if (!alloc_profile_is_valid(type, 0)) {
+		ASSERT(0);
+		return ERR_PTR(-EINVAL);
+	}
+
+	if (list_empty(&fs_devices->alloc_list)) {
+		if (btrfs_test_opt(info, ENOSPC_DEBUG))
+			btrfs_debug(info, "%s: no writable device", __func__);
+		return ERR_PTR(-ENOSPC);
+	}
+
+	if (!(type & BTRFS_BLOCK_GROUP_TYPE_MASK)) {
+		btrfs_err(info, "invalid chunk type 0x%llx requested", type);
+		ASSERT(0);
+		return ERR_PTR(-EINVAL);
+	}
+
+	ctl.start = find_next_chunk(info);
+	ctl.type = type;
+	init_alloc_chunk_ctl(fs_devices, &ctl);
+
+	devices_info = kcalloc(fs_devices->rw_devices, sizeof(*devices_info),
+			       GFP_NOFS);
+	if (!devices_info)
+		return ERR_PTR(-ENOMEM);
+
+	ret = gather_device_info(fs_devices, &ctl, devices_info);
+	if (ret < 0) {
+		block_group = ERR_PTR(ret);
+		goto out;
+	}
+
+	ret = decide_stripe_size(fs_devices, &ctl, devices_info);
+	if (ret < 0) {
+		block_group = ERR_PTR(ret);
+		goto out;
+	}
+
+	block_group = create_chunk(trans, &ctl, devices_info);
+
+out:
 	kfree(devices_info);
-	return ret;
+	return block_group;
 }
 
-int btrfs_finish_chunk_alloc(struct btrfs_trans_handle *trans,
-			     u64 chunk_offset, u64 chunk_size)
+/*
+ * This function, btrfs_chunk_alloc_add_chunk_item(), typically belongs to the
+ * phase 1 of chunk allocation. It belongs to phase 2 only when allocating system
+ * chunks.
+ *
+ * See the comment at btrfs_chunk_alloc() for details about the chunk allocation
+ * phases.
+ */
+int btrfs_chunk_alloc_add_chunk_item(struct btrfs_trans_handle *trans,
+				     struct btrfs_block_group *bg)
 {
 	struct btrfs_fs_info *fs_info = trans->fs_info;
-	struct btrfs_root *extent_root = fs_info->extent_root;
 	struct btrfs_root *chunk_root = fs_info->chunk_root;
 	struct btrfs_key key;
-	struct btrfs_device *device;
 	struct btrfs_chunk *chunk;
 	struct btrfs_stripe *stripe;
 	struct extent_map *em;
 	struct map_lookup *map;
 	size_t item_size;
-	u64 dev_offset;
-	u64 stripe_size;
-	int i = 0;
-	int ret = 0;
+	int i;
+	int ret;
 
-	em = btrfs_get_chunk_map(fs_info, chunk_offset, chunk_size);
-	if (IS_ERR(em))
-		return PTR_ERR(em);
+	/*
+	 * We take the chunk_mutex for 2 reasons:
+	 *
+	 * 1) Updates and insertions in the chunk btree must be done while holding
+	 *    the chunk_mutex, as well as updating the system chunk array in the
+	 *    superblock. See the comment on top of btrfs_chunk_alloc() for the
+	 *    details;
+	 *
+	 * 2) To prevent races with the final phase of a device replace operation
+	 *    that replaces the device object associated with the map's stripes,
+	 *    because the device object's id can change at any time during that
+	 *    final phase of the device replace operation
+	 *    (dev-replace.c:btrfs_dev_replace_finishing()), so we could grab the
+	 *    replaced device and then see it with an ID of BTRFS_DEV_REPLACE_DEVID,
+	 *    which would cause a failure when updating the device item, which does
+	 *    not exists, or persisting a stripe of the chunk item with such ID.
+	 *    Here we can't use the device_list_mutex because our caller already
+	 *    has locked the chunk_mutex, and the final phase of device replace
+	 *    acquires both mutexes - first the device_list_mutex and then the
+	 *    chunk_mutex. Using any of those two mutexes protects us from a
+	 *    concurrent device replace.
+	 */
+	lockdep_assert_held(&fs_info->chunk_mutex);
+
+	em = btrfs_get_chunk_map(fs_info, bg->start, bg->length);
+	if (IS_ERR(em)) {
+		ret = PTR_ERR(em);
+		btrfs_abort_transaction(trans, ret);
+		return ret;
+	}
 
 	map = em->map_lookup;
 	item_size = btrfs_chunk_item_size(map->num_stripes);
-	stripe_size = em->orig_block_len;
 
 	chunk = kzalloc(item_size, GFP_NOFS);
 	if (!chunk) {
 		ret = -ENOMEM;
+		btrfs_abort_transaction(trans, ret);
 		goto out;
 	}
 
-	/*
-	 * Take the device list mutex to prevent races with the final phase of
-	 * a device replace operation that replaces the device object associated
-	 * with the map's stripes, because the device object's id can change
-	 * at any time during that final phase of the device replace operation
-	 * (dev-replace.c:btrfs_dev_replace_finishing()).
-	 */
-	mutex_lock(&fs_info->fs_devices->device_list_mutex);
 	for (i = 0; i < map->num_stripes; i++) {
-		device = map->stripes[i].dev;
-		dev_offset = map->stripes[i].physical;
+		struct btrfs_device *device = map->stripes[i].dev;
 
 		ret = btrfs_update_device(trans, device);
 		if (ret)
-			break;
-		ret = btrfs_alloc_dev_extent(trans, device, chunk_offset,
-					     dev_offset, stripe_size);
-		if (ret)
-			break;
-	}
-	if (ret) {
-		mutex_unlock(&fs_info->fs_devices->device_list_mutex);
-		goto out;
+			goto out;
 	}
 
 	stripe = &chunk->stripe;
 	for (i = 0; i < map->num_stripes; i++) {
-		device = map->stripes[i].dev;
-		dev_offset = map->stripes[i].physical;
+		struct btrfs_device *device = map->stripes[i].dev;
+		const u64 dev_offset = map->stripes[i].physical;
 
 		btrfs_set_stack_stripe_devid(stripe, device->devid);
 		btrfs_set_stack_stripe_offset(stripe, dev_offset);
 		memcpy(stripe->dev_uuid, device->uuid, BTRFS_UUID_SIZE);
 		stripe++;
 	}
-	mutex_unlock(&fs_info->fs_devices->device_list_mutex);
 
-	btrfs_set_stack_chunk_length(chunk, chunk_size);
-	btrfs_set_stack_chunk_owner(chunk, extent_root->root_key.objectid);
+	btrfs_set_stack_chunk_length(chunk, bg->length);
+	btrfs_set_stack_chunk_owner(chunk, BTRFS_EXTENT_TREE_OBJECTID);
 	btrfs_set_stack_chunk_stripe_len(chunk, map->stripe_len);
 	btrfs_set_stack_chunk_type(chunk, map->type);
 	btrfs_set_stack_chunk_num_stripes(chunk, map->num_stripes);
@@ -5143,15 +5603,18 @@ int btrfs_finish_chunk_alloc(struct btrfs_trans_handle *trans,
 
 	key.objectid = BTRFS_FIRST_CHUNK_TREE_OBJECTID;
 	key.type = BTRFS_CHUNK_ITEM_KEY;
-	key.offset = chunk_offset;
+	key.offset = bg->start;
 
 	ret = btrfs_insert_item(trans, chunk_root, &key, chunk, item_size);
-	if (ret == 0 && map->type & BTRFS_BLOCK_GROUP_SYSTEM) {
-		/*
-		 * TODO: Cleanup of inserted chunk root in case of
-		 * failure.
-		 */
+	if (ret)
+		goto out;
+
+	set_bit(BLOCK_GROUP_FLAG_CHUNK_ITEM_INSERTED, &bg->runtime_flags);
+
+	if (map->type & BTRFS_BLOCK_GROUP_SYSTEM) {
 		ret = btrfs_add_system_chunk(fs_info, &key, chunk, item_size);
+		if (ret)
+			goto out;
 	}
 
 out:
@@ -5160,40 +5623,45 @@ out:
 	return ret;
 }
 
-/*
- * Chunk allocation falls into two parts. The first part does work
- * that makes the new allocated chunk usable, but does not do any operation
- * that modifies the chunk tree. The second part does the work that
- * requires modifying the chunk tree. This division is important for the
- * bootstrap process of adding storage to a seed btrfs.
- */
-int btrfs_alloc_chunk(struct btrfs_trans_handle *trans, u64 type)
-{
-	u64 chunk_offset;
-
-	lockdep_assert_held(&trans->fs_info->chunk_mutex);
-	chunk_offset = find_next_chunk(trans->fs_info);
-	return __btrfs_alloc_chunk(trans, chunk_offset, type);
-}
-
 static noinline int init_first_rw_device(struct btrfs_trans_handle *trans)
 {
 	struct btrfs_fs_info *fs_info = trans->fs_info;
-	u64 chunk_offset;
-	u64 sys_chunk_offset;
 	u64 alloc_profile;
-	int ret;
+	struct btrfs_block_group *meta_bg;
+	struct btrfs_block_group *sys_bg;
+
+	/*
+	 * When adding a new device for sprouting, the seed device is read-only
+	 * so we must first allocate a metadata and a system chunk. But before
+	 * adding the block group items to the extent, device and chunk btrees,
+	 * we must first:
+	 *
+	 * 1) Create both chunks without doing any changes to the btrees, as
+	 *    otherwise we would get -ENOSPC since the block groups from the
+	 *    seed device are read-only;
+	 *
+	 * 2) Add the device item for the new sprout device - finishing the setup
+	 *    of a new block group requires updating the device item in the chunk
+	 *    btree, so it must exist when we attempt to do it. The previous step
+	 *    ensures this does not fail with -ENOSPC.
+	 *
+	 * After that we can add the block group items to their btrees:
+	 * update existing device item in the chunk btree, add a new block group
+	 * item to the extent btree, add a new chunk item to the chunk btree and
+	 * finally add the new device extent items to the devices btree.
+	 */
 
-	chunk_offset = find_next_chunk(fs_info);
 	alloc_profile = btrfs_metadata_alloc_profile(fs_info);
-	ret = __btrfs_alloc_chunk(trans, chunk_offset, alloc_profile);
-	if (ret)
-		return ret;
+	meta_bg = btrfs_create_chunk(trans, alloc_profile);
+	if (IS_ERR(meta_bg))
+		return PTR_ERR(meta_bg);
 
-	sys_chunk_offset = find_next_chunk(fs_info);
 	alloc_profile = btrfs_system_alloc_profile(fs_info);
-	ret = __btrfs_alloc_chunk(trans, sys_chunk_offset, alloc_profile);
-	return ret;
+	sys_bg = btrfs_create_chunk(trans, alloc_profile);
+	if (IS_ERR(sys_bg))
+		return PTR_ERR(sys_bg);
+
+	return 0;
 }
 
 static inline int btrfs_chunk_max_errors(struct map_lookup *map)
@@ -5203,17 +5671,17 @@ static inline int btrfs_chunk_max_errors(struct map_lookup *map)
 	return btrfs_raid_array[index].tolerated_failures;
 }
 
-int btrfs_chunk_readonly(struct btrfs_fs_info *fs_info, u64 chunk_offset)
+bool btrfs_chunk_writeable(struct btrfs_fs_info *fs_info, u64 chunk_offset)
 {
 	struct extent_map *em;
 	struct map_lookup *map;
-	int readonly = 0;
 	int miss_ndevs = 0;
 	int i;
+	bool ret = true;
 
 	em = btrfs_get_chunk_map(fs_info, chunk_offset, 1);
 	if (IS_ERR(em))
-		return 1;
+		return false;
 
 	map = em->map_lookup;
 	for (i = 0; i < map->num_stripes; i++) {
@@ -5224,21 +5692,20 @@ int btrfs_chunk_readonly(struct btrfs_fs_info *fs_info, u64 chunk_offset)
 		}
 		if (!test_bit(BTRFS_DEV_STATE_WRITEABLE,
 					&map->stripes[i].dev->dev_state)) {
-			readonly = 1;
+			ret = false;
 			goto end;
 		}
 	}
 
 	/*
-	 * If the number of missing devices is larger than max errors,
-	 * we can not write the data into that chunk successfully, so
-	 * set it readonly.
+	 * If the number of missing devices is larger than max errors, we can
+	 * not write the data into that chunk successfully.
 	 */
 	if (miss_ndevs > btrfs_chunk_max_errors(map))
-		readonly = 1;
+		ret = false;
 end:
 	free_extent_map(em);
-	return readonly;
+	return ret;
 }
 
 void btrfs_mapping_tree_free(struct extent_map_tree *tree)
@@ -5264,7 +5731,8 @@ int btrfs_num_copies(struct btrfs_fs_info *fs_info, u64 logical, u64 len)
 {
 	struct extent_map *em;
 	struct map_lookup *map;
-	int ret;
+	enum btrfs_raid_types index;
+	int ret = 1;
 
 	em = btrfs_get_chunk_map(fs_info, logical, len);
 	if (IS_ERR(em))
@@ -5277,10 +5745,11 @@ int btrfs_num_copies(struct btrfs_fs_info *fs_info, u64 logical, u64 len)
 		return 1;
 
 	map = em->map_lookup;
-	if (map->type & (BTRFS_BLOCK_GROUP_DUP | BTRFS_BLOCK_GROUP_RAID1_MASK))
-		ret = map->num_stripes;
-	else if (map->type & BTRFS_BLOCK_GROUP_RAID10)
-		ret = map->sub_stripes;
+	index = btrfs_bg_flags_to_raid_index(map->type);
+
+	/* Non-RAID56, use their ncopies from btrfs_raid_array. */
+	if (!(map->type & BTRFS_BLOCK_GROUP_RAID56_MASK))
+		ret = btrfs_raid_array[index].ncopies;
 	else if (map->type & BTRFS_BLOCK_GROUP_RAID5)
 		ret = 2;
 	else if (map->type & BTRFS_BLOCK_GROUP_RAID6)
@@ -5292,8 +5761,6 @@ int btrfs_num_copies(struct btrfs_fs_info *fs_info, u64 logical, u64 len)
 		 * stripe under reconstruction.
 		 */
 		ret = map->num_stripes;
-	else
-		ret = 1;
 	free_extent_map(em);
 
 	down_read(&fs_info->dev_replace.rwsem);
@@ -5312,6 +5779,9 @@ unsigned long btrfs_full_stripe_len(struct btrfs_fs_info *fs_info,
 	struct map_lookup *map;
 	unsigned long len = fs_info->sectorsize;
 
+	if (!btrfs_fs_incompat(fs_info, RAID56))
+		return len;
+
 	em = btrfs_get_chunk_map(fs_info, logical, len);
 
 	if (!WARN_ON(IS_ERR(em))) {
@@ -5329,6 +5799,9 @@ int btrfs_is_parity_mirror(struct btrfs_fs_info *fs_info, u64 logical, u64 len)
 	struct map_lookup *map;
 	int ret = 0;
 
+	if (!btrfs_fs_incompat(fs_info, RAID56))
+		return 0;
+
 	em = btrfs_get_chunk_map(fs_info, logical, len);
 
 	if(!WARN_ON(IS_ERR(em))) {
@@ -5358,7 +5831,18 @@ static int find_live_mirror(struct btrfs_fs_info *fs_info,
 	else
 		num_stripes = map->num_stripes;
 
-	preferred_mirror = first + current->pid % num_stripes;
+	switch (fs_info->fs_devices->read_policy) {
+	default:
+		/* Shouldn't happen, just warn and use pid instead of failing */
+		btrfs_warn_rl(fs_info,
+			      "unknown read_policy type %u, reset to pid",
+			      fs_info->fs_devices->read_policy);
+		fs_info->fs_devices->read_policy = BTRFS_READ_POLICY_PID;
+		fallthrough;
+	case BTRFS_READ_POLICY_PID:
+		preferred_mirror = first + (current->pid % num_stripes);
+		break;
+	}
 
 	if (dev_replace_is_ongoing &&
 	    fs_info->dev_replace.cont_reading_from_srcdev_mode ==
@@ -5389,85 +5873,77 @@ static int find_live_mirror(struct btrfs_fs_info *fs_info,
 	return preferred_mirror;
 }
 
-static inline int parity_smaller(u64 a, u64 b)
-{
-	return a > b;
-}
-
 /* Bubble-sort the stripe set to put the parity/syndrome stripes last */
-static void sort_parity_stripes(struct btrfs_bio *bbio, int num_stripes)
+static void sort_parity_stripes(struct btrfs_io_context *bioc, int num_stripes)
 {
-	struct btrfs_bio_stripe s;
 	int i;
-	u64 l;
 	int again = 1;
 
 	while (again) {
 		again = 0;
 		for (i = 0; i < num_stripes - 1; i++) {
-			if (parity_smaller(bbio->raid_map[i],
-					   bbio->raid_map[i+1])) {
-				s = bbio->stripes[i];
-				l = bbio->raid_map[i];
-				bbio->stripes[i] = bbio->stripes[i+1];
-				bbio->raid_map[i] = bbio->raid_map[i+1];
-				bbio->stripes[i+1] = s;
-				bbio->raid_map[i+1] = l;
-
+			/* Swap if parity is on a smaller index */
+			if (bioc->raid_map[i] > bioc->raid_map[i + 1]) {
+				swap(bioc->stripes[i], bioc->stripes[i + 1]);
+				swap(bioc->raid_map[i], bioc->raid_map[i + 1]);
 				again = 1;
 			}
 		}
 	}
 }
 
-static struct btrfs_bio *alloc_btrfs_bio(int total_stripes, int real_stripes)
+static struct btrfs_io_context *alloc_btrfs_io_context(struct btrfs_fs_info *fs_info,
+						       int total_stripes,
+						       int real_stripes)
 {
-	struct btrfs_bio *bbio = kzalloc(
-		 /* the size of the btrfs_bio */
-		sizeof(struct btrfs_bio) +
-		/* plus the variable array for the stripes */
-		sizeof(struct btrfs_bio_stripe) * (total_stripes) +
-		/* plus the variable array for the tgt dev */
+	struct btrfs_io_context *bioc = kzalloc(
+		 /* The size of btrfs_io_context */
+		sizeof(struct btrfs_io_context) +
+		/* Plus the variable array for the stripes */
+		sizeof(struct btrfs_io_stripe) * (total_stripes) +
+		/* Plus the variable array for the tgt dev */
 		sizeof(int) * (real_stripes) +
 		/*
-		 * plus the raid_map, which includes both the tgt dev
-		 * and the stripes
+		 * Plus the raid_map, which includes both the tgt dev
+		 * and the stripes.
 		 */
 		sizeof(u64) * (total_stripes),
 		GFP_NOFS|__GFP_NOFAIL);
 
-	atomic_set(&bbio->error, 0);
-	refcount_set(&bbio->refs, 1);
+	refcount_set(&bioc->refs, 1);
+
+	bioc->fs_info = fs_info;
+	bioc->tgtdev_map = (int *)(bioc->stripes + total_stripes);
+	bioc->raid_map = (u64 *)(bioc->tgtdev_map + real_stripes);
 
-	return bbio;
+	return bioc;
 }
 
-void btrfs_get_bbio(struct btrfs_bio *bbio)
+void btrfs_get_bioc(struct btrfs_io_context *bioc)
 {
-	WARN_ON(!refcount_read(&bbio->refs));
-	refcount_inc(&bbio->refs);
+	WARN_ON(!refcount_read(&bioc->refs));
+	refcount_inc(&bioc->refs);
 }
 
-void btrfs_put_bbio(struct btrfs_bio *bbio)
+void btrfs_put_bioc(struct btrfs_io_context *bioc)
 {
-	if (!bbio)
+	if (!bioc)
 		return;
-	if (refcount_dec_and_test(&bbio->refs))
-		kfree(bbio);
+	if (refcount_dec_and_test(&bioc->refs))
+		kfree(bioc);
 }
 
-/* can REQ_OP_DISCARD be sent with other REQ like REQ_OP_WRITE? */
 /*
  * Please note that, discard won't be sent to target device of device
  * replace.
  */
-static int __btrfs_map_block_for_discard(struct btrfs_fs_info *fs_info,
-					 u64 logical, u64 *length_ret,
-					 struct btrfs_bio **bbio_ret)
+struct btrfs_discard_stripe *btrfs_map_discard(struct btrfs_fs_info *fs_info,
+					       u64 logical, u64 *length_ret,
+					       u32 *num_stripes)
 {
 	struct extent_map *em;
 	struct map_lookup *map;
-	struct btrfs_bio *bbio;
+	struct btrfs_discard_stripe *stripes;
 	u64 length = *length_ret;
 	u64 offset;
 	u64 stripe_nr;
@@ -5476,29 +5952,26 @@ static int __btrfs_map_block_for_discard(struct btrfs_fs_info *fs_info,
 	u64 stripe_cnt;
 	u64 stripe_len;
 	u64 stripe_offset;
-	u64 num_stripes;
 	u32 stripe_index;
 	u32 factor = 0;
 	u32 sub_stripes = 0;
 	u64 stripes_per_dev = 0;
 	u32 remaining_stripes = 0;
 	u32 last_stripe = 0;
-	int ret = 0;
+	int ret;
 	int i;
 
-	/* discard always return a bbio */
-	ASSERT(bbio_ret);
-
 	em = btrfs_get_chunk_map(fs_info, logical, length);
 	if (IS_ERR(em))
-		return PTR_ERR(em);
+		return ERR_CAST(em);
 
 	map = em->map_lookup;
+
 	/* we don't discard raid56 yet */
 	if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) {
 		ret = -EOPNOTSUPP;
-		goto out;
-	}
+		goto out_free_map;
+}
 
 	offset = logical - em->start;
 	length = min_t(u64, em->start + em->len - logical, length);
@@ -5524,7 +5997,7 @@ static int __btrfs_map_block_for_discard(struct btrfs_fs_info *fs_info,
 	 * device we have to walk to find the data, and stripe_index is
 	 * the number of our device in the stripe array
 	 */
-	num_stripes = 1;
+	*num_stripes = 1;
 	stripe_index = 0;
 	if (map->type & (BTRFS_BLOCK_GROUP_RAID0 |
 			 BTRFS_BLOCK_GROUP_RAID10)) {
@@ -5534,7 +6007,7 @@ static int __btrfs_map_block_for_discard(struct btrfs_fs_info *fs_info,
 			sub_stripes = map->sub_stripes;
 
 		factor = map->num_stripes / sub_stripes;
-		num_stripes = min_t(u64, map->num_stripes,
+		*num_stripes = min_t(u64, map->num_stripes,
 				    sub_stripes * stripe_cnt);
 		stripe_nr = div_u64_rem(stripe_nr, factor, &stripe_index);
 		stripe_index *= sub_stripes;
@@ -5544,32 +6017,30 @@ static int __btrfs_map_block_for_discard(struct btrfs_fs_info *fs_info,
 		last_stripe *= sub_stripes;
 	} else if (map->type & (BTRFS_BLOCK_GROUP_RAID1_MASK |
 				BTRFS_BLOCK_GROUP_DUP)) {
-		num_stripes = map->num_stripes;
+		*num_stripes = map->num_stripes;
 	} else {
 		stripe_nr = div_u64_rem(stripe_nr, map->num_stripes,
 					&stripe_index);
 	}
 
-	bbio = alloc_btrfs_bio(num_stripes, 0);
-	if (!bbio) {
+	stripes = kcalloc(*num_stripes, sizeof(*stripes), GFP_NOFS);
+	if (!stripes) {
 		ret = -ENOMEM;
-		goto out;
+		goto out_free_map;
 	}
 
-	for (i = 0; i < num_stripes; i++) {
-		bbio->stripes[i].physical =
+	for (i = 0; i < *num_stripes; i++) {
+		stripes[i].physical =
 			map->stripes[stripe_index].physical +
 			stripe_offset + stripe_nr * map->stripe_len;
-		bbio->stripes[i].dev = map->stripes[stripe_index].dev;
+		stripes[i].dev = map->stripes[stripe_index].dev;
 
 		if (map->type & (BTRFS_BLOCK_GROUP_RAID0 |
 				 BTRFS_BLOCK_GROUP_RAID10)) {
-			bbio->stripes[i].length = stripes_per_dev *
-				map->stripe_len;
+			stripes[i].length = stripes_per_dev * map->stripe_len;
 
 			if (i / sub_stripes < remaining_stripes)
-				bbio->stripes[i].length +=
-					map->stripe_len;
+				stripes[i].length += map->stripe_len;
 
 			/*
 			 * Special for the first stripe and
@@ -5580,19 +6051,17 @@ static int __btrfs_map_block_for_discard(struct btrfs_fs_info *fs_info,
 			 *    off     end_off
 			 */
 			if (i < sub_stripes)
-				bbio->stripes[i].length -=
-					stripe_offset;
+				stripes[i].length -= stripe_offset;
 
 			if (stripe_index >= last_stripe &&
 			    stripe_index <= (last_stripe +
 					     sub_stripes - 1))
-				bbio->stripes[i].length -=
-					stripe_end_offset;
+				stripes[i].length -= stripe_end_offset;
 
 			if (i == sub_stripes - 1)
 				stripe_offset = 0;
 		} else {
-			bbio->stripes[i].length = length;
+			stripes[i].length = length;
 		}
 
 		stripe_index++;
@@ -5602,12 +6071,11 @@ static int __btrfs_map_block_for_discard(struct btrfs_fs_info *fs_info,
 		}
 	}
 
-	*bbio_ret = bbio;
-	bbio->map_type = map->type;
-	bbio->num_stripes = num_stripes;
-out:
 	free_extent_map(em);
-	return ret;
+	return stripes;
+out_free_map:
+	free_extent_map(em);
+	return ERR_PTR(ret);
 }
 
 /*
@@ -5628,7 +6096,7 @@ static int get_extra_mirror_from_replace(struct btrfs_fs_info *fs_info,
 					 u64 srcdev_devid, int *mirror_num,
 					 u64 *physical)
 {
-	struct btrfs_bio *bbio = NULL;
+	struct btrfs_io_context *bioc = NULL;
 	int num_stripes;
 	int index_srcdev = 0;
 	int found = 0;
@@ -5637,20 +6105,20 @@ static int get_extra_mirror_from_replace(struct btrfs_fs_info *fs_info,
 	int ret = 0;
 
 	ret = __btrfs_map_block(fs_info, BTRFS_MAP_GET_READ_MIRRORS,
-				logical, &length, &bbio, 0, 0);
+				logical, &length, &bioc, NULL, NULL, 0);
 	if (ret) {
-		ASSERT(bbio == NULL);
+		ASSERT(bioc == NULL);
 		return ret;
 	}
 
-	num_stripes = bbio->num_stripes;
+	num_stripes = bioc->num_stripes;
 	if (*mirror_num > num_stripes) {
 		/*
 		 * BTRFS_MAP_GET_READ_MIRRORS does not contain this mirror,
 		 * that means that the requested area is not left of the left
 		 * cursor
 		 */
-		btrfs_put_bbio(bbio);
+		btrfs_put_bioc(bioc);
 		return -EIO;
 	}
 
@@ -5660,7 +6128,7 @@ static int get_extra_mirror_from_replace(struct btrfs_fs_info *fs_info,
 	 * pointer to the one of the target drive.
 	 */
 	for (i = 0; i < num_stripes; i++) {
-		if (bbio->stripes[i].dev->devid != srcdev_devid)
+		if (bioc->stripes[i].dev->devid != srcdev_devid)
 			continue;
 
 		/*
@@ -5668,15 +6136,15 @@ static int get_extra_mirror_from_replace(struct btrfs_fs_info *fs_info,
 		 * mirror with the lowest physical address
 		 */
 		if (found &&
-		    physical_of_found <= bbio->stripes[i].physical)
+		    physical_of_found <= bioc->stripes[i].physical)
 			continue;
 
 		index_srcdev = i;
 		found = 1;
-		physical_of_found = bbio->stripes[i].physical;
+		physical_of_found = bioc->stripes[i].physical;
 	}
 
-	btrfs_put_bbio(bbio);
+	btrfs_put_bioc(bioc);
 
 	ASSERT(found);
 	if (!found)
@@ -5687,12 +6155,30 @@ static int get_extra_mirror_from_replace(struct btrfs_fs_info *fs_info,
 	return ret;
 }
 
+static bool is_block_group_to_copy(struct btrfs_fs_info *fs_info, u64 logical)
+{
+	struct btrfs_block_group *cache;
+	bool ret;
+
+	/* Non zoned filesystem does not use "to_copy" flag */
+	if (!btrfs_is_zoned(fs_info))
+		return false;
+
+	cache = btrfs_lookup_block_group(fs_info, logical);
+
+	ret = test_bit(BLOCK_GROUP_FLAG_TO_COPY, &cache->runtime_flags);
+
+	btrfs_put_block_group(cache);
+	return ret;
+}
+
 static void handle_ops_on_dev_replace(enum btrfs_map_op op,
-				      struct btrfs_bio **bbio_ret,
+				      struct btrfs_io_context **bioc_ret,
 				      struct btrfs_dev_replace *dev_replace,
+				      u64 logical,
 				      int *num_stripes_ret, int *max_errors_ret)
 {
-	struct btrfs_bio *bbio = *bbio_ret;
+	struct btrfs_io_context *bioc = *bioc_ret;
 	u64 srcdev_devid = dev_replace->srcdev->devid;
 	int tgtdev_indexes = 0;
 	int num_stripes = *num_stripes_ret;
@@ -5703,6 +6189,13 @@ static void handle_ops_on_dev_replace(enum btrfs_map_op op,
 		int index_where_to_add;
 
 		/*
+		 * A block group which have "to_copy" set will eventually
+		 * copied by dev-replace process. We can avoid cloning IO here.
+		 */
+		if (is_block_group_to_copy(dev_replace->srcdev->fs_info, logical))
+			return;
+
+		/*
 		 * duplicate the write operations while the dev replace
 		 * procedure is running. Since the copying of the old disk to
 		 * the new disk takes place at run time while the filesystem is
@@ -5715,17 +6208,16 @@ static void handle_ops_on_dev_replace(enum btrfs_map_op op,
 		 */
 		index_where_to_add = num_stripes;
 		for (i = 0; i < num_stripes; i++) {
-			if (bbio->stripes[i].dev->devid == srcdev_devid) {
+			if (bioc->stripes[i].dev->devid == srcdev_devid) {
 				/* write to new disk, too */
-				struct btrfs_bio_stripe *new =
-					bbio->stripes + index_where_to_add;
-				struct btrfs_bio_stripe *old =
-					bbio->stripes + i;
+				struct btrfs_io_stripe *new =
+					bioc->stripes + index_where_to_add;
+				struct btrfs_io_stripe *old =
+					bioc->stripes + i;
 
 				new->physical = old->physical;
-				new->length = old->length;
 				new->dev = dev_replace->tgtdev;
-				bbio->tgtdev_map[i] = index_where_to_add;
+				bioc->tgtdev_map[i] = index_where_to_add;
 				index_where_to_add++;
 				max_errors++;
 				tgtdev_indexes++;
@@ -5745,30 +6237,27 @@ static void handle_ops_on_dev_replace(enum btrfs_map_op op,
 		 * full copy of the source drive.
 		 */
 		for (i = 0; i < num_stripes; i++) {
-			if (bbio->stripes[i].dev->devid == srcdev_devid) {
+			if (bioc->stripes[i].dev->devid == srcdev_devid) {
 				/*
 				 * In case of DUP, in order to keep it simple,
 				 * only add the mirror with the lowest physical
 				 * address
 				 */
 				if (found &&
-				    physical_of_found <=
-				     bbio->stripes[i].physical)
+				    physical_of_found <= bioc->stripes[i].physical)
 					continue;
 				index_srcdev = i;
 				found = 1;
-				physical_of_found = bbio->stripes[i].physical;
+				physical_of_found = bioc->stripes[i].physical;
 			}
 		}
 		if (found) {
-			struct btrfs_bio_stripe *tgtdev_stripe =
-				bbio->stripes + num_stripes;
+			struct btrfs_io_stripe *tgtdev_stripe =
+				bioc->stripes + num_stripes;
 
 			tgtdev_stripe->physical = physical_of_found;
-			tgtdev_stripe->length =
-				bbio->stripes[index_srcdev].length;
 			tgtdev_stripe->dev = dev_replace->tgtdev;
-			bbio->tgtdev_map[index_srcdev] = num_stripes;
+			bioc->tgtdev_map[index_srcdev] = num_stripes;
 
 			tgtdev_indexes++;
 			num_stripes++;
@@ -5777,8 +6266,8 @@ static void handle_ops_on_dev_replace(enum btrfs_map_op op,
 
 	*num_stripes_ret = num_stripes;
 	*max_errors_ret = max_errors;
-	bbio->num_tgtdevs = tgtdev_indexes;
-	*bbio_ret = bbio;
+	bioc->num_tgtdevs = tgtdev_indexes;
+	*bioc_ret = bioc;
 }
 
 static bool need_full_stripe(enum btrfs_map_op op)
@@ -5787,60 +6276,50 @@ static bool need_full_stripe(enum btrfs_map_op op)
 }
 
 /*
- * btrfs_get_io_geometry - calculates the geomery of a particular (address, len)
- *		       tuple. This information is used to calculate how big a
- *		       particular bio can get before it straddles a stripe.
+ * Calculate the geometry of a particular (address, len) tuple. This
+ * information is used to calculate how big a particular bio can get before it
+ * straddles a stripe.
  *
- * @fs_info - the filesystem
- * @logical - address that we want to figure out the geometry of
- * @len	    - the length of IO we are going to perform, starting at @logical
- * @op      - type of operation - write or read
- * @io_geom - pointer used to return values
+ * @fs_info: the filesystem
+ * @em:      mapping containing the logical extent
+ * @op:      type of operation - write or read
+ * @logical: address that we want to figure out the geometry of
+ * @io_geom: pointer used to return values
  *
  * Returns < 0 in case a chunk for the given logical address cannot be found,
  * usually shouldn't happen unless @logical is corrupted, 0 otherwise.
  */
-int btrfs_get_io_geometry(struct btrfs_fs_info *fs_info, enum btrfs_map_op op,
-			u64 logical, u64 len, struct btrfs_io_geometry *io_geom)
+int btrfs_get_io_geometry(struct btrfs_fs_info *fs_info, struct extent_map *em,
+			  enum btrfs_map_op op, u64 logical,
+			  struct btrfs_io_geometry *io_geom)
 {
-	struct extent_map *em;
 	struct map_lookup *map;
+	u64 len;
 	u64 offset;
 	u64 stripe_offset;
 	u64 stripe_nr;
-	u64 stripe_len;
+	u32 stripe_len;
 	u64 raid56_full_stripe_start = (u64)-1;
 	int data_stripes;
-	int ret = 0;
 
 	ASSERT(op != BTRFS_MAP_DISCARD);
 
-	em = btrfs_get_chunk_map(fs_info, logical, len);
-	if (IS_ERR(em))
-		return PTR_ERR(em);
-
 	map = em->map_lookup;
 	/* Offset of this logical address in the chunk */
 	offset = logical - em->start;
 	/* Len of a stripe in a chunk */
 	stripe_len = map->stripe_len;
-	/* Stripe wher this block falls in */
-	stripe_nr = div64_u64(offset, stripe_len);
-	/* Offset of stripe in the chunk */
-	stripe_offset = stripe_nr * stripe_len;
-	if (offset < stripe_offset) {
-		btrfs_crit(fs_info,
-"stripe math has gone wrong, stripe_offset=%llu offset=%llu start=%llu logical=%llu stripe_len=%llu",
-			stripe_offset, offset, em->start, logical, stripe_len);
-		ret = -EINVAL;
-		goto out;
-	}
+	/*
+	 * Stripe_nr is where this block falls in
+	 * stripe_offset is the offset of this block in its stripe.
+	 */
+	stripe_nr = div64_u64_rem(offset, stripe_len, &stripe_offset);
+	ASSERT(stripe_offset < U32_MAX);
 
-	/* stripe_offset is the offset of this block in its stripe */
-	stripe_offset = offset - stripe_offset;
 	data_stripes = nr_data_stripes(map);
 
-	if (map->type & BTRFS_BLOCK_GROUP_PROFILE_MASK) {
+	/* Only stripe based profiles needs to check against stripe length. */
+	if (map->type & BTRFS_BLOCK_GROUP_STRIPE_MASK) {
 		u64 max_len = stripe_len - stripe_offset;
 
 		/*
@@ -5880,17 +6359,22 @@ int btrfs_get_io_geometry(struct btrfs_fs_info *fs_info, enum btrfs_map_op op,
 	io_geom->stripe_offset = stripe_offset;
 	io_geom->raid56_stripe_offset = raid56_full_stripe_start;
 
-out:
-	/* once for us */
-	free_extent_map(em);
-	return ret;
+	return 0;
+}
+
+static void set_io_stripe(struct btrfs_io_stripe *dst, const struct map_lookup *map,
+		          u32 stripe_index, u64 stripe_offset, u64 stripe_nr)
+{
+	dst->dev = map->stripes[stripe_index].dev;
+	dst->physical = map->stripes[stripe_index].physical +
+			stripe_offset + stripe_nr * map->stripe_len;
 }
 
 static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
-			     enum btrfs_map_op op,
-			     u64 logical, u64 *length,
-			     struct btrfs_bio **bbio_ret,
-			     int mirror_num, int need_raid_map)
+			     enum btrfs_map_op op, u64 logical, u64 *length,
+			     struct btrfs_io_context **bioc_ret,
+			     struct btrfs_io_stripe *smap,
+			     int *mirror_num_ret, int need_raid_map)
 {
 	struct extent_map *em;
 	struct map_lookup *map;
@@ -5901,10 +6385,11 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
 	int data_stripes;
 	int i;
 	int ret = 0;
+	int mirror_num = (mirror_num_ret ? *mirror_num_ret : 0);
 	int num_stripes;
 	int max_errors = 0;
 	int tgtdev_indexes = 0;
-	struct btrfs_bio *bbio = NULL;
+	struct btrfs_io_context *bioc = NULL;
 	struct btrfs_dev_replace *dev_replace = &fs_info->dev_replace;
 	int dev_replace_is_ongoing = 0;
 	int num_alloc_stripes;
@@ -5913,18 +6398,16 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
 	u64 raid56_full_stripe_start = (u64)-1;
 	struct btrfs_io_geometry geom;
 
-	ASSERT(bbio_ret);
+	ASSERT(bioc_ret);
+	ASSERT(op != BTRFS_MAP_DISCARD);
 
-	if (op == BTRFS_MAP_DISCARD)
-		return __btrfs_map_block_for_discard(fs_info, logical,
-						     length, bbio_ret);
+	em = btrfs_get_chunk_map(fs_info, logical, *length);
+	ASSERT(!IS_ERR(em));
 
-	ret = btrfs_get_io_geometry(fs_info, op, logical, *length, &geom);
+	ret = btrfs_get_io_geometry(fs_info, em, op, logical, &geom);
 	if (ret < 0)
 		return ret;
 
-	em = btrfs_get_chunk_map(fs_info, logical, *length);
-	ASSERT(!IS_ERR(em));
 	map = em->map_lookup;
 
 	*length = geom.len;
@@ -6003,6 +6486,7 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
 		}
 
 	} else if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) {
+		ASSERT(map->stripe_len == BTRFS_STRIPE_LEN);
 		if (need_raid_map && (need_full_stripe(op) || mirror_num > 1)) {
 			/* push stripe_nr back to the start of the full stripe */
 			stripe_nr = div64_u64(raid56_full_stripe_start,
@@ -6010,9 +6494,12 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
 
 			/* RAID[56] write or recovery. Return all stripes */
 			num_stripes = map->num_stripes;
-			max_errors = nr_parity_stripes(map);
+			max_errors = btrfs_chunk_max_errors(map);
 
-			*length = map->stripe_len;
+			/* Return the length to the full stripe end */
+			*length = min(logical + *length,
+				      raid56_full_stripe_start + em->start +
+				      data_stripes * stripe_len) - logical;
 			stripe_index = 0;
 			stripe_offset = 0;
 		} else {
@@ -6059,68 +6546,78 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
 		tgtdev_indexes = num_stripes;
 	}
 
-	bbio = alloc_btrfs_bio(num_alloc_stripes, tgtdev_indexes);
-	if (!bbio) {
+	/*
+	 * If this I/O maps to a single device, try to return the device and
+	 * physical block information on the stack instead of allocating an
+	 * I/O context structure.
+	 */
+	if (smap && num_alloc_stripes == 1 &&
+	    !((map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) && mirror_num > 1) &&
+	    (!need_full_stripe(op) || !dev_replace_is_ongoing ||
+	     !dev_replace->tgtdev)) {
+		if (patch_the_first_stripe_for_dev_replace) {
+			smap->dev = dev_replace->tgtdev;
+			smap->physical = physical_to_patch_in_first_stripe;
+			*mirror_num_ret = map->num_stripes + 1;
+		} else {
+			set_io_stripe(smap, map, stripe_index, stripe_offset,
+				      stripe_nr);
+			*mirror_num_ret = mirror_num;
+		}
+		*bioc_ret = NULL;
+		ret = 0;
+		goto out;
+	}
+
+	bioc = alloc_btrfs_io_context(fs_info, num_alloc_stripes, tgtdev_indexes);
+	if (!bioc) {
 		ret = -ENOMEM;
 		goto out;
 	}
-	if (dev_replace_is_ongoing && dev_replace->tgtdev != NULL)
-		bbio->tgtdev_map = (int *)(bbio->stripes + num_alloc_stripes);
 
-	/* build raid_map */
+	for (i = 0; i < num_stripes; i++) {
+		set_io_stripe(&bioc->stripes[i], map, stripe_index, stripe_offset,
+			      stripe_nr);
+		stripe_index++;
+	}
+
+	/* Build raid_map */
 	if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK && need_raid_map &&
 	    (need_full_stripe(op) || mirror_num > 1)) {
 		u64 tmp;
 		unsigned rot;
 
-		bbio->raid_map = (u64 *)((void *)bbio->stripes +
-				 sizeof(struct btrfs_bio_stripe) *
-				 num_alloc_stripes +
-				 sizeof(int) * tgtdev_indexes);
-
 		/* Work out the disk rotation on this stripe-set */
 		div_u64_rem(stripe_nr, num_stripes, &rot);
 
 		/* Fill in the logical address of each stripe */
 		tmp = stripe_nr * data_stripes;
 		for (i = 0; i < data_stripes; i++)
-			bbio->raid_map[(i+rot) % num_stripes] =
+			bioc->raid_map[(i + rot) % num_stripes] =
 				em->start + (tmp + i) * map->stripe_len;
 
-		bbio->raid_map[(i+rot) % map->num_stripes] = RAID5_P_STRIPE;
+		bioc->raid_map[(i + rot) % map->num_stripes] = RAID5_P_STRIPE;
 		if (map->type & BTRFS_BLOCK_GROUP_RAID6)
-			bbio->raid_map[(i+rot+1) % num_stripes] =
+			bioc->raid_map[(i + rot + 1) % num_stripes] =
 				RAID6_Q_STRIPE;
-	}
 
-
-	for (i = 0; i < num_stripes; i++) {
-		bbio->stripes[i].physical =
-			map->stripes[stripe_index].physical +
-			stripe_offset +
-			stripe_nr * map->stripe_len;
-		bbio->stripes[i].dev =
-			map->stripes[stripe_index].dev;
-		stripe_index++;
+		sort_parity_stripes(bioc, num_stripes);
 	}
 
 	if (need_full_stripe(op))
 		max_errors = btrfs_chunk_max_errors(map);
 
-	if (bbio->raid_map)
-		sort_parity_stripes(bbio, num_stripes);
-
 	if (dev_replace_is_ongoing && dev_replace->tgtdev != NULL &&
 	    need_full_stripe(op)) {
-		handle_ops_on_dev_replace(op, &bbio, dev_replace, &num_stripes,
-					  &max_errors);
+		handle_ops_on_dev_replace(op, &bioc, dev_replace, logical,
+					  &num_stripes, &max_errors);
 	}
 
-	*bbio_ret = bbio;
-	bbio->map_type = map->type;
-	bbio->num_stripes = num_stripes;
-	bbio->max_errors = max_errors;
-	bbio->mirror_num = mirror_num;
+	*bioc_ret = bioc;
+	bioc->map_type = map->type;
+	bioc->num_stripes = num_stripes;
+	bioc->max_errors = max_errors;
+	bioc->mirror_num = mirror_num;
 
 	/*
 	 * this is the case that REQ_READ && dev_replace_is_ongoing &&
@@ -6129,9 +6626,9 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
 	 */
 	if (patch_the_first_stripe_for_dev_replace && num_stripes > 0) {
 		WARN_ON(num_stripes > 1);
-		bbio->stripes[0].dev = dev_replace->tgtdev;
-		bbio->stripes[0].physical = physical_to_patch_in_first_stripe;
-		bbio->mirror_num = map->num_stripes + 1;
+		bioc->stripes[0].dev = dev_replace->tgtdev;
+		bioc->stripes[0].physical = physical_to_patch_in_first_stripe;
+		bioc->mirror_num = map->num_stripes + 1;
 	}
 out:
 	if (dev_replace_is_ongoing) {
@@ -6145,173 +6642,245 @@ out:
 
 int btrfs_map_block(struct btrfs_fs_info *fs_info, enum btrfs_map_op op,
 		      u64 logical, u64 *length,
-		      struct btrfs_bio **bbio_ret, int mirror_num)
+		      struct btrfs_io_context **bioc_ret, int mirror_num)
 {
-	return __btrfs_map_block(fs_info, op, logical, length, bbio_ret,
-				 mirror_num, 0);
+	return __btrfs_map_block(fs_info, op, logical, length, bioc_ret,
+				 NULL, &mirror_num, 0);
 }
 
 /* For Scrub/replace */
 int btrfs_map_sblock(struct btrfs_fs_info *fs_info, enum btrfs_map_op op,
 		     u64 logical, u64 *length,
-		     struct btrfs_bio **bbio_ret)
+		     struct btrfs_io_context **bioc_ret)
 {
-	return __btrfs_map_block(fs_info, op, logical, length, bbio_ret, 0, 1);
+	return __btrfs_map_block(fs_info, op, logical, length, bioc_ret,
+				 NULL, NULL, 1);
 }
 
-static inline void btrfs_end_bbio(struct btrfs_bio *bbio, struct bio *bio)
+/*
+ * Initialize a btrfs_bio structure.  This skips the embedded bio itself as it
+ * is already initialized by the block layer.
+ */
+static inline void btrfs_bio_init(struct btrfs_bio *bbio,
+				  btrfs_bio_end_io_t end_io, void *private)
 {
-	bio->bi_private = bbio->private;
-	bio->bi_end_io = bbio->end_io;
-	bio_endio(bio);
+	memset(bbio, 0, offsetof(struct btrfs_bio, bio));
+	bbio->end_io = end_io;
+	bbio->private = private;
+}
 
-	btrfs_put_bbio(bbio);
+/*
+ * Allocate a btrfs_bio structure.  The btrfs_bio is the main I/O container for
+ * btrfs, and is used for all I/O submitted through btrfs_submit_bio.
+ *
+ * Just like the underlying bio_alloc_bioset it will not fail as it is backed by
+ * a mempool.
+ */
+struct bio *btrfs_bio_alloc(unsigned int nr_vecs, blk_opf_t opf,
+			    btrfs_bio_end_io_t end_io, void *private)
+{
+	struct bio *bio;
+
+	bio = bio_alloc_bioset(NULL, nr_vecs, opf, GFP_NOFS, &btrfs_bioset);
+	btrfs_bio_init(btrfs_bio(bio), end_io, private);
+	return bio;
 }
 
-static void btrfs_end_bio(struct bio *bio)
+struct bio *btrfs_bio_clone_partial(struct bio *orig, u64 offset, u64 size,
+				    btrfs_bio_end_io_t end_io, void *private)
 {
-	struct btrfs_bio *bbio = bio->bi_private;
-	int is_orig_bio = 0;
+	struct bio *bio;
+	struct btrfs_bio *bbio;
 
-	if (bio->bi_status) {
-		atomic_inc(&bbio->error);
-		if (bio->bi_status == BLK_STS_IOERR ||
-		    bio->bi_status == BLK_STS_TARGET) {
-			unsigned int stripe_index =
-				btrfs_io_bio(bio)->stripe_index;
-			struct btrfs_device *dev;
-
-			BUG_ON(stripe_index >= bbio->num_stripes);
-			dev = bbio->stripes[stripe_index].dev;
-			if (dev->bdev) {
-				if (bio_op(bio) == REQ_OP_WRITE)
-					btrfs_dev_stat_inc_and_print(dev,
-						BTRFS_DEV_STAT_WRITE_ERRS);
-				else if (!(bio->bi_opf & REQ_RAHEAD))
-					btrfs_dev_stat_inc_and_print(dev,
-						BTRFS_DEV_STAT_READ_ERRS);
-				if (bio->bi_opf & REQ_PREFLUSH)
-					btrfs_dev_stat_inc_and_print(dev,
-						BTRFS_DEV_STAT_FLUSH_ERRS);
-			}
-		}
+	ASSERT(offset <= UINT_MAX && size <= UINT_MAX);
+
+	bio = bio_alloc_clone(orig->bi_bdev, orig, GFP_NOFS, &btrfs_bioset);
+	bbio = btrfs_bio(bio);
+	btrfs_bio_init(bbio, end_io, private);
+
+	bio_trim(bio, offset >> 9, size >> 9);
+	bbio->iter = bio->bi_iter;
+	return bio;
+}
+
+static void btrfs_log_dev_io_error(struct bio *bio, struct btrfs_device *dev)
+{
+	if (!dev || !dev->bdev)
+		return;
+	if (bio->bi_status != BLK_STS_IOERR && bio->bi_status != BLK_STS_TARGET)
+		return;
+
+	if (btrfs_op(bio) == BTRFS_MAP_WRITE)
+		btrfs_dev_stat_inc_and_print(dev, BTRFS_DEV_STAT_WRITE_ERRS);
+	if (!(bio->bi_opf & REQ_RAHEAD))
+		btrfs_dev_stat_inc_and_print(dev, BTRFS_DEV_STAT_READ_ERRS);
+	if (bio->bi_opf & REQ_PREFLUSH)
+		btrfs_dev_stat_inc_and_print(dev, BTRFS_DEV_STAT_FLUSH_ERRS);
+}
+
+static struct workqueue_struct *btrfs_end_io_wq(struct btrfs_fs_info *fs_info,
+						struct bio *bio)
+{
+	if (bio->bi_opf & REQ_META)
+		return fs_info->endio_meta_workers;
+	return fs_info->endio_workers;
+}
+
+static void btrfs_end_bio_work(struct work_struct *work)
+{
+	struct btrfs_bio *bbio =
+		container_of(work, struct btrfs_bio, end_io_work);
+
+	bbio->end_io(bbio);
+}
+
+static void btrfs_simple_end_io(struct bio *bio)
+{
+	struct btrfs_fs_info *fs_info = bio->bi_private;
+	struct btrfs_bio *bbio = btrfs_bio(bio);
+
+	btrfs_bio_counter_dec(fs_info);
+
+	if (bio->bi_status)
+		btrfs_log_dev_io_error(bio, bbio->device);
+
+	if (bio_op(bio) == REQ_OP_READ) {
+		INIT_WORK(&bbio->end_io_work, btrfs_end_bio_work);
+		queue_work(btrfs_end_io_wq(fs_info, bio), &bbio->end_io_work);
+	} else {
+		bbio->end_io(bbio);
 	}
+}
 
-	if (bio == bbio->orig_bio)
-		is_orig_bio = 1;
+static void btrfs_raid56_end_io(struct bio *bio)
+{
+	struct btrfs_io_context *bioc = bio->bi_private;
+	struct btrfs_bio *bbio = btrfs_bio(bio);
 
-	btrfs_bio_counter_dec(bbio->fs_info);
+	btrfs_bio_counter_dec(bioc->fs_info);
+	bbio->mirror_num = bioc->mirror_num;
+	bbio->end_io(bbio);
 
-	if (atomic_dec_and_test(&bbio->stripes_pending)) {
-		if (!is_orig_bio) {
-			bio_put(bio);
-			bio = bbio->orig_bio;
-		}
+	btrfs_put_bioc(bioc);
+}
 
-		btrfs_io_bio(bio)->mirror_num = bbio->mirror_num;
-		/* only send an error to the higher layers if it is
-		 * beyond the tolerance of the btrfs bio
-		 */
-		if (atomic_read(&bbio->error) > bbio->max_errors) {
-			bio->bi_status = BLK_STS_IOERR;
-		} else {
-			/*
-			 * this bio is actually up to date, we didn't
-			 * go over the max number of errors
-			 */
-			bio->bi_status = BLK_STS_OK;
-		}
+static void btrfs_orig_write_end_io(struct bio *bio)
+{
+	struct btrfs_io_stripe *stripe = bio->bi_private;
+	struct btrfs_io_context *bioc = stripe->bioc;
+	struct btrfs_bio *bbio = btrfs_bio(bio);
+
+	btrfs_bio_counter_dec(bioc->fs_info);
 
-		btrfs_end_bbio(bbio, bio);
-	} else if (!is_orig_bio) {
-		bio_put(bio);
+	if (bio->bi_status) {
+		atomic_inc(&bioc->error);
+		btrfs_log_dev_io_error(bio, stripe->dev);
 	}
+
+	/*
+	 * Only send an error to the higher layers if it is beyond the tolerance
+	 * threshold.
+	 */
+	if (atomic_read(&bioc->error) > bioc->max_errors)
+		bio->bi_status = BLK_STS_IOERR;
+	else
+		bio->bi_status = BLK_STS_OK;
+
+	bbio->end_io(bbio);
+	btrfs_put_bioc(bioc);
 }
 
-static void submit_stripe_bio(struct btrfs_bio *bbio, struct bio *bio,
-			      u64 physical, int dev_nr)
+static void btrfs_clone_write_end_io(struct bio *bio)
 {
-	struct btrfs_device *dev = bbio->stripes[dev_nr].dev;
-	struct btrfs_fs_info *fs_info = bbio->fs_info;
+	struct btrfs_io_stripe *stripe = bio->bi_private;
+
+	if (bio->bi_status) {
+		atomic_inc(&stripe->bioc->error);
+		btrfs_log_dev_io_error(bio, stripe->dev);
+	}
+
+	/* Pass on control to the original bio this one was cloned from */
+	bio_endio(stripe->bioc->orig_bio);
+	bio_put(bio);
+}
+
+static void btrfs_submit_dev_bio(struct btrfs_device *dev, struct bio *bio)
+{
+	if (!dev || !dev->bdev ||
+	    test_bit(BTRFS_DEV_STATE_MISSING, &dev->dev_state) ||
+	    (btrfs_op(bio) == BTRFS_MAP_WRITE &&
+	     !test_bit(BTRFS_DEV_STATE_WRITEABLE, &dev->dev_state))) {
+		bio_io_error(bio);
+		return;
+	}
 
-	bio->bi_private = bbio;
-	btrfs_io_bio(bio)->stripe_index = dev_nr;
-	bio->bi_end_io = btrfs_end_bio;
-	bio->bi_iter.bi_sector = physical >> 9;
-	btrfs_debug_in_rcu(fs_info,
-	"btrfs_map_bio: rw %d 0x%x, sector=%llu, dev=%lu (%s id %llu), size=%u",
-		bio_op(bio), bio->bi_opf, (u64)bio->bi_iter.bi_sector,
-		(u_long)dev->bdev->bd_dev, rcu_str_deref(dev->name), dev->devid,
-		bio->bi_iter.bi_size);
 	bio_set_dev(bio, dev->bdev);
 
-	btrfs_bio_counter_inc_noblocked(fs_info);
+	/*
+	 * For zone append writing, bi_sector must point the beginning of the
+	 * zone
+	 */
+	if (bio_op(bio) == REQ_OP_ZONE_APPEND) {
+		u64 physical = bio->bi_iter.bi_sector << SECTOR_SHIFT;
+
+		if (btrfs_dev_is_sequential(dev, physical)) {
+			u64 zone_start = round_down(physical,
+						    dev->fs_info->zone_size);
 
-	btrfsic_submit_bio(bio);
+			bio->bi_iter.bi_sector = zone_start >> SECTOR_SHIFT;
+		} else {
+			bio->bi_opf &= ~REQ_OP_ZONE_APPEND;
+			bio->bi_opf |= REQ_OP_WRITE;
+		}
+	}
+	btrfs_debug_in_rcu(dev->fs_info,
+	"%s: rw %d 0x%x, sector=%llu, dev=%lu (%s id %llu), size=%u",
+		__func__, bio_op(bio), bio->bi_opf, bio->bi_iter.bi_sector,
+		(unsigned long)dev->bdev->bd_dev, rcu_str_deref(dev->name),
+		dev->devid, bio->bi_iter.bi_size);
+
+	btrfsic_check_bio(bio);
+	submit_bio(bio);
 }
 
-static void bbio_error(struct btrfs_bio *bbio, struct bio *bio, u64 logical)
+static void btrfs_submit_mirrored_bio(struct btrfs_io_context *bioc, int dev_nr)
 {
-	atomic_inc(&bbio->error);
-	if (atomic_dec_and_test(&bbio->stripes_pending)) {
-		/* Should be the original bio. */
-		WARN_ON(bio != bbio->orig_bio);
+	struct bio *orig_bio = bioc->orig_bio, *bio;
 
-		btrfs_io_bio(bio)->mirror_num = bbio->mirror_num;
-		bio->bi_iter.bi_sector = logical >> 9;
-		if (atomic_read(&bbio->error) > bbio->max_errors)
-			bio->bi_status = BLK_STS_IOERR;
-		else
-			bio->bi_status = BLK_STS_OK;
-		btrfs_end_bbio(bbio, bio);
+	ASSERT(bio_op(orig_bio) != REQ_OP_READ);
+
+	/* Reuse the bio embedded into the btrfs_bio for the last mirror */
+	if (dev_nr == bioc->num_stripes - 1) {
+		bio = orig_bio;
+		bio->bi_end_io = btrfs_orig_write_end_io;
+	} else {
+		bio = bio_alloc_clone(NULL, orig_bio, GFP_NOFS, &fs_bio_set);
+		bio_inc_remaining(orig_bio);
+		bio->bi_end_io = btrfs_clone_write_end_io;
 	}
+
+	bio->bi_private = &bioc->stripes[dev_nr];
+	bio->bi_iter.bi_sector = bioc->stripes[dev_nr].physical >> SECTOR_SHIFT;
+	bioc->stripes[dev_nr].bioc = bioc;
+	btrfs_submit_dev_bio(bioc->stripes[dev_nr].dev, bio);
 }
 
-blk_status_t btrfs_map_bio(struct btrfs_fs_info *fs_info, struct bio *bio,
-			   int mirror_num)
+void btrfs_submit_bio(struct btrfs_fs_info *fs_info, struct bio *bio, int mirror_num)
 {
-	struct btrfs_device *dev;
-	struct bio *first_bio = bio;
-	u64 logical = (u64)bio->bi_iter.bi_sector << 9;
-	u64 length = 0;
-	u64 map_length;
+	u64 logical = bio->bi_iter.bi_sector << 9;
+	u64 length = bio->bi_iter.bi_size;
+	u64 map_length = length;
+	struct btrfs_io_context *bioc = NULL;
+	struct btrfs_io_stripe smap;
 	int ret;
-	int dev_nr;
-	int total_devs;
-	struct btrfs_bio *bbio = NULL;
-
-	length = bio->bi_iter.bi_size;
-	map_length = length;
 
 	btrfs_bio_counter_inc_blocked(fs_info);
-	ret = __btrfs_map_block(fs_info, btrfs_op(bio), logical,
-				&map_length, &bbio, mirror_num, 1);
+	ret = __btrfs_map_block(fs_info, btrfs_op(bio), logical, &map_length,
+				&bioc, &smap, &mirror_num, 1);
 	if (ret) {
 		btrfs_bio_counter_dec(fs_info);
-		return errno_to_blk_status(ret);
-	}
-
-	total_devs = bbio->num_stripes;
-	bbio->orig_bio = first_bio;
-	bbio->private = first_bio->bi_private;
-	bbio->end_io = first_bio->bi_end_io;
-	bbio->fs_info = fs_info;
-	atomic_set(&bbio->stripes_pending, bbio->num_stripes);
-
-	if ((bbio->map_type & BTRFS_BLOCK_GROUP_RAID56_MASK) &&
-	    ((bio_op(bio) == REQ_OP_WRITE) || (mirror_num > 1))) {
-		/* In this case, map_length has been set to the length of
-		   a single stripe; not the whole write */
-		if (bio_op(bio) == REQ_OP_WRITE) {
-			ret = raid56_parity_write(fs_info, bio, bbio,
-						  map_length);
-		} else {
-			ret = raid56_parity_recover(fs_info, bio, bbio,
-						    map_length, mirror_num, 1);
-		}
-
-		btrfs_bio_counter_dec(fs_info);
-		return errno_to_blk_status(ret);
+		btrfs_bio_end_io(btrfs_bio(bio), errno_to_blk_status(ret));
+		return;
 	}
 
 	if (map_length < length) {
@@ -6321,26 +6890,58 @@ blk_status_t btrfs_map_bio(struct btrfs_fs_info *fs_info, struct bio *bio,
 		BUG();
 	}
 
-	for (dev_nr = 0; dev_nr < total_devs; dev_nr++) {
-		dev = bbio->stripes[dev_nr].dev;
-		if (!dev || !dev->bdev || test_bit(BTRFS_DEV_STATE_MISSING,
-						   &dev->dev_state) ||
-		    (bio_op(first_bio) == REQ_OP_WRITE &&
-		    !test_bit(BTRFS_DEV_STATE_WRITEABLE, &dev->dev_state))) {
-			bbio_error(bbio, first_bio, logical);
-			continue;
-		}
-
-		if (dev_nr < total_devs - 1)
-			bio = btrfs_bio_clone(first_bio);
+	if (!bioc) {
+		/* Single mirror read/write fast path */
+		btrfs_bio(bio)->mirror_num = mirror_num;
+		btrfs_bio(bio)->device = smap.dev;
+		bio->bi_iter.bi_sector = smap.physical >> SECTOR_SHIFT;
+		bio->bi_private = fs_info;
+		bio->bi_end_io = btrfs_simple_end_io;
+		btrfs_submit_dev_bio(smap.dev, bio);
+	} else if (bioc->map_type & BTRFS_BLOCK_GROUP_RAID56_MASK) {
+		/* Parity RAID write or read recovery */
+		bio->bi_private = bioc;
+		bio->bi_end_io = btrfs_raid56_end_io;
+		if (bio_op(bio) == REQ_OP_READ)
+			raid56_parity_recover(bio, bioc, mirror_num);
 		else
-			bio = first_bio;
+			raid56_parity_write(bio, bioc);
+	} else {
+		/* Write to multiple mirrors */
+		int total_devs = bioc->num_stripes;
+		int dev_nr;
 
-		submit_stripe_bio(bbio, bio, bbio->stripes[dev_nr].physical,
-				  dev_nr);
+		bioc->orig_bio = bio;
+		for (dev_nr = 0; dev_nr < total_devs; dev_nr++)
+			btrfs_submit_mirrored_bio(bioc, dev_nr);
 	}
-	btrfs_bio_counter_dec(fs_info);
-	return BLK_STS_OK;
+}
+
+static bool dev_args_match_fs_devices(const struct btrfs_dev_lookup_args *args,
+				      const struct btrfs_fs_devices *fs_devices)
+{
+	if (args->fsid == NULL)
+		return true;
+	if (memcmp(fs_devices->metadata_uuid, args->fsid, BTRFS_FSID_SIZE) == 0)
+		return true;
+	return false;
+}
+
+static bool dev_args_match_device(const struct btrfs_dev_lookup_args *args,
+				  const struct btrfs_device *device)
+{
+	if (args->missing) {
+		if (test_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &device->dev_state) &&
+		    !device->bdev)
+			return true;
+		return false;
+	}
+
+	if (device->devid != args->devid)
+		return false;
+	if (args->uuid && memcmp(device->uuid, args->uuid, BTRFS_UUID_SIZE) != 0)
+		return false;
+	return true;
 }
 
 /*
@@ -6349,31 +6950,29 @@ blk_status_t btrfs_map_bio(struct btrfs_fs_info *fs_info, struct bio *bio,
  *
  * If devid and uuid are both specified, the match must be exact, otherwise
  * only devid is used.
- *
- * If @seed is true, traverse through the seed devices.
  */
-struct btrfs_device *btrfs_find_device(struct btrfs_fs_devices *fs_devices,
-				       u64 devid, u8 *uuid, u8 *fsid,
-				       bool seed)
+struct btrfs_device *btrfs_find_device(const struct btrfs_fs_devices *fs_devices,
+				       const struct btrfs_dev_lookup_args *args)
 {
 	struct btrfs_device *device;
+	struct btrfs_fs_devices *seed_devs;
 
-	while (fs_devices) {
-		if (!fsid ||
-		    !memcmp(fs_devices->metadata_uuid, fsid, BTRFS_FSID_SIZE)) {
-			list_for_each_entry(device, &fs_devices->devices,
-					    dev_list) {
-				if (device->devid == devid &&
-				    (!uuid || memcmp(device->uuid, uuid,
-						     BTRFS_UUID_SIZE) == 0))
-					return device;
-			}
+	if (dev_args_match_fs_devices(args, fs_devices)) {
+		list_for_each_entry(device, &fs_devices->devices, dev_list) {
+			if (dev_args_match_device(args, device))
+				return device;
+		}
+	}
+
+	list_for_each_entry(seed_devs, &fs_devices->seed_list, seed_list) {
+		if (!dev_args_match_fs_devices(args, seed_devs))
+			continue;
+		list_for_each_entry(device, &seed_devs->devices, dev_list) {
+			if (dev_args_match_device(args, device))
+				return device;
 		}
-		if (seed)
-			fs_devices = fs_devices->seed;
-		else
-			return NULL;
 	}
+
 	return NULL;
 }
 
@@ -6381,8 +6980,17 @@ static struct btrfs_device *add_missing_dev(struct btrfs_fs_devices *fs_devices,
 					    u64 devid, u8 *dev_uuid)
 {
 	struct btrfs_device *device;
+	unsigned int nofs_flag;
 
+	/*
+	 * We call this under the chunk_mutex, so we want to use NOFS for this
+	 * allocation, however we don't want to change btrfs_alloc_device() to
+	 * always do NOFS because we use it in a lot of other GFP_KERNEL safe
+	 * places.
+	 */
+	nofs_flag = memalloc_nofs_save();
 	device = btrfs_alloc_device(NULL, &devid, dev_uuid);
+	memalloc_nofs_restore(nofs_flag);
 	if (IS_ERR(device))
 		return device;
 
@@ -6419,9 +7027,18 @@ struct btrfs_device *btrfs_alloc_device(struct btrfs_fs_info *fs_info,
 	if (WARN_ON(!devid && !fs_info))
 		return ERR_PTR(-EINVAL);
 
-	dev = __alloc_device();
-	if (IS_ERR(dev))
-		return dev;
+	dev = kzalloc(sizeof(*dev), GFP_KERNEL);
+	if (!dev)
+		return ERR_PTR(-ENOMEM);
+
+	INIT_LIST_HEAD(&dev->dev_list);
+	INIT_LIST_HEAD(&dev->dev_alloc_list);
+	INIT_LIST_HEAD(&dev->post_commit_list);
+
+	atomic_set(&dev->dev_stats_ccnt, 0);
+	btrfs_device_data_ordered_init(dev);
+	extent_io_tree_init(fs_info, &dev->alloc_state,
+			    IO_TREE_DEVICE_ALLOC_STATE, NULL);
 
 	if (devid)
 		tmp = *devid;
@@ -6455,24 +7072,79 @@ static void btrfs_report_missing_device(struct btrfs_fs_info *fs_info,
 			      devid, uuid);
 }
 
-static u64 calc_stripe_length(u64 type, u64 chunk_len, int num_stripes)
+u64 btrfs_calc_stripe_length(const struct extent_map *em)
 {
-	int index = btrfs_bg_flags_to_raid_index(type);
-	int ncopies = btrfs_raid_array[index].ncopies;
-	const int nparity = btrfs_raid_array[index].nparity;
-	int data_stripes;
+	const struct map_lookup *map = em->map_lookup;
+	const int data_stripes = calc_data_stripes(map->type, map->num_stripes);
 
-	if (nparity)
-		data_stripes = num_stripes - nparity;
-	else
-		data_stripes = num_stripes / ncopies;
+	return div_u64(em->len, data_stripes);
+}
+
+#if BITS_PER_LONG == 32
+/*
+ * Due to page cache limit, metadata beyond BTRFS_32BIT_MAX_FILE_SIZE
+ * can't be accessed on 32bit systems.
+ *
+ * This function do mount time check to reject the fs if it already has
+ * metadata chunk beyond that limit.
+ */
+static int check_32bit_meta_chunk(struct btrfs_fs_info *fs_info,
+				  u64 logical, u64 length, u64 type)
+{
+	if (!(type & BTRFS_BLOCK_GROUP_METADATA))
+		return 0;
+
+	if (logical + length < MAX_LFS_FILESIZE)
+		return 0;
+
+	btrfs_err_32bit_limit(fs_info);
+	return -EOVERFLOW;
+}
+
+/*
+ * This is to give early warning for any metadata chunk reaching
+ * BTRFS_32BIT_EARLY_WARN_THRESHOLD.
+ * Although we can still access the metadata, it's not going to be possible
+ * once the limit is reached.
+ */
+static void warn_32bit_meta_chunk(struct btrfs_fs_info *fs_info,
+				  u64 logical, u64 length, u64 type)
+{
+	if (!(type & BTRFS_BLOCK_GROUP_METADATA))
+		return;
+
+	if (logical + length < BTRFS_32BIT_EARLY_WARN_THRESHOLD)
+		return;
+
+	btrfs_warn_32bit_limit(fs_info);
+}
+#endif
+
+static struct btrfs_device *handle_missing_device(struct btrfs_fs_info *fs_info,
+						  u64 devid, u8 *uuid)
+{
+	struct btrfs_device *dev;
+
+	if (!btrfs_test_opt(fs_info, DEGRADED)) {
+		btrfs_report_missing_device(fs_info, devid, uuid, true);
+		return ERR_PTR(-ENOENT);
+	}
+
+	dev = add_missing_dev(fs_info->fs_devices, devid, uuid);
+	if (IS_ERR(dev)) {
+		btrfs_err(fs_info, "failed to init missing device %llu: %ld",
+			  devid, PTR_ERR(dev));
+		return dev;
+	}
+	btrfs_report_missing_device(fs_info, devid, uuid, false);
 
-	return div_u64(chunk_len, data_stripes);
+	return dev;
 }
 
 static int read_one_chunk(struct btrfs_key *key, struct extent_buffer *leaf,
 			  struct btrfs_chunk *chunk)
 {
+	BTRFS_DEV_LOOKUP_ARGS(args);
 	struct btrfs_fs_info *fs_info = leaf->fs_info;
 	struct extent_map_tree *map_tree = &fs_info->mapping_tree;
 	struct map_lookup *map;
@@ -6480,15 +7152,26 @@ static int read_one_chunk(struct btrfs_key *key, struct extent_buffer *leaf,
 	u64 logical;
 	u64 length;
 	u64 devid;
+	u64 type;
 	u8 uuid[BTRFS_UUID_SIZE];
+	int index;
 	int num_stripes;
 	int ret;
 	int i;
 
 	logical = key->offset;
 	length = btrfs_chunk_length(leaf, chunk);
+	type = btrfs_chunk_type(leaf, chunk);
+	index = btrfs_bg_flags_to_raid_index(type);
 	num_stripes = btrfs_chunk_num_stripes(leaf, chunk);
 
+#if BITS_PER_LONG == 32
+	ret = check_32bit_meta_chunk(fs_info, logical, length, type);
+	if (ret < 0)
+		return ret;
+	warn_32bit_meta_chunk(fs_info, logical, length, type);
+#endif
+
 	/*
 	 * Only need to verify chunk item if we're reading from sys chunk array,
 	 * as chunk item in tree block is already verified by tree-checker.
@@ -6532,42 +7215,39 @@ static int read_one_chunk(struct btrfs_key *key, struct extent_buffer *leaf,
 	map->io_width = btrfs_chunk_io_width(leaf, chunk);
 	map->io_align = btrfs_chunk_io_align(leaf, chunk);
 	map->stripe_len = btrfs_chunk_stripe_len(leaf, chunk);
-	map->type = btrfs_chunk_type(leaf, chunk);
-	map->sub_stripes = btrfs_chunk_sub_stripes(leaf, chunk);
+	map->type = type;
+	/*
+	 * We can't use the sub_stripes value, as for profiles other than
+	 * RAID10, they may have 0 as sub_stripes for filesystems created by
+	 * older mkfs (<v5.4).
+	 * In that case, it can cause divide-by-zero errors later.
+	 * Since currently sub_stripes is fixed for each profile, let's
+	 * use the trusted value instead.
+	 */
+	map->sub_stripes = btrfs_raid_array[index].sub_stripes;
 	map->verified_stripes = 0;
-	em->orig_block_len = calc_stripe_length(map->type, em->len,
-						map->num_stripes);
+	em->orig_block_len = btrfs_calc_stripe_length(em);
 	for (i = 0; i < num_stripes; i++) {
 		map->stripes[i].physical =
 			btrfs_stripe_offset_nr(leaf, chunk, i);
 		devid = btrfs_stripe_devid_nr(leaf, chunk, i);
+		args.devid = devid;
 		read_extent_buffer(leaf, uuid, (unsigned long)
 				   btrfs_stripe_dev_uuid_nr(chunk, i),
 				   BTRFS_UUID_SIZE);
-		map->stripes[i].dev = btrfs_find_device(fs_info->fs_devices,
-							devid, uuid, NULL, true);
-		if (!map->stripes[i].dev &&
-		    !btrfs_test_opt(fs_info, DEGRADED)) {
-			free_extent_map(em);
-			btrfs_report_missing_device(fs_info, devid, uuid, true);
-			return -ENOENT;
-		}
+		args.uuid = uuid;
+		map->stripes[i].dev = btrfs_find_device(fs_info->fs_devices, &args);
 		if (!map->stripes[i].dev) {
-			map->stripes[i].dev =
-				add_missing_dev(fs_info->fs_devices, devid,
-						uuid);
+			map->stripes[i].dev = handle_missing_device(fs_info,
+								    devid, uuid);
 			if (IS_ERR(map->stripes[i].dev)) {
 				free_extent_map(em);
-				btrfs_err(fs_info,
-					"failed to init missing dev %llu: %ld",
-					devid, PTR_ERR(map->stripes[i].dev));
 				return PTR_ERR(map->stripes[i].dev);
 			}
-			btrfs_report_missing_device(fs_info, devid, uuid, false);
 		}
+
 		set_bit(BTRFS_DEV_STATE_IN_FS_METADATA,
 				&(map->stripes[i].dev->dev_state));
-
 	}
 
 	write_lock(&map_tree->lock);
@@ -6615,13 +7295,11 @@ static struct btrfs_fs_devices *open_seed_devices(struct btrfs_fs_info *fs_info,
 	lockdep_assert_held(&uuid_mutex);
 	ASSERT(fsid);
 
-	fs_devices = fs_info->fs_devices->seed;
-	while (fs_devices) {
+	/* This will match only for multi-device seed fs */
+	list_for_each_entry(fs_devices, &fs_info->fs_devices->seed_list, seed_list)
 		if (!memcmp(fs_devices->fsid, fsid, BTRFS_FSID_SIZE))
 			return fs_devices;
 
-		fs_devices = fs_devices->seed;
-	}
 
 	fs_devices = find_fsid(fsid, NULL);
 	if (!fs_devices) {
@@ -6637,6 +7315,10 @@ static struct btrfs_fs_devices *open_seed_devices(struct btrfs_fs_info *fs_info,
 		return fs_devices;
 	}
 
+	/*
+	 * Upon first call for a seed fs fsid, just create a private copy of the
+	 * respective fs_devices and anchor it at fs_info->fs_devices->seed_list
+	 */
 	fs_devices = clone_fs_devices(fs_devices);
 	if (IS_ERR(fs_devices))
 		return fs_devices;
@@ -6644,26 +7326,24 @@ static struct btrfs_fs_devices *open_seed_devices(struct btrfs_fs_info *fs_info,
 	ret = open_fs_devices(fs_devices, FMODE_READ, fs_info->bdev_holder);
 	if (ret) {
 		free_fs_devices(fs_devices);
-		fs_devices = ERR_PTR(ret);
-		goto out;
+		return ERR_PTR(ret);
 	}
 
 	if (!fs_devices->seeding) {
 		close_fs_devices(fs_devices);
 		free_fs_devices(fs_devices);
-		fs_devices = ERR_PTR(-EINVAL);
-		goto out;
+		return ERR_PTR(-EINVAL);
 	}
 
-	fs_devices->seed = fs_info->fs_devices->seed;
-	fs_info->fs_devices->seed = fs_devices;
-out:
+	list_add(&fs_devices->seed_list, &fs_info->fs_devices->seed_list);
+
 	return fs_devices;
 }
 
 static int read_one_dev(struct extent_buffer *leaf,
 			struct btrfs_dev_item *dev_item)
 {
+	BTRFS_DEV_LOOKUP_ARGS(args);
 	struct btrfs_fs_info *fs_info = leaf->fs_info;
 	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
 	struct btrfs_device *device;
@@ -6673,10 +7353,13 @@ static int read_one_dev(struct extent_buffer *leaf,
 	u8 dev_uuid[BTRFS_UUID_SIZE];
 
 	devid = btrfs_device_id(leaf, dev_item);
+	args.devid = devid;
 	read_extent_buffer(leaf, dev_uuid, btrfs_device_uuid(dev_item),
 			   BTRFS_UUID_SIZE);
 	read_extent_buffer(leaf, fs_uuid, btrfs_device_fsid(dev_item),
 			   BTRFS_FSID_SIZE);
+	args.uuid = dev_uuid;
+	args.fsid = fs_uuid;
 
 	if (memcmp(fs_uuid, fs_devices->metadata_uuid, BTRFS_FSID_SIZE)) {
 		fs_devices = open_seed_devices(fs_info, fs_uuid);
@@ -6684,8 +7367,7 @@ static int read_one_dev(struct extent_buffer *leaf,
 			return PTR_ERR(fs_devices);
 	}
 
-	device = btrfs_find_device(fs_info->fs_devices, devid, dev_uuid,
-				   fs_uuid, true);
+	device = btrfs_find_device(fs_info->fs_devices, &args);
 	if (!device) {
 		if (!btrfs_test_opt(fs_info, DEGRADED)) {
 			btrfs_report_missing_device(fs_info, devid,
@@ -6748,6 +7430,16 @@ static int read_one_dev(struct extent_buffer *leaf,
 	}
 
 	fill_device_from_item(leaf, dev_item, device);
+	if (device->bdev) {
+		u64 max_total_bytes = bdev_nr_bytes(device->bdev);
+
+		if (device->total_bytes > max_total_bytes) {
+			btrfs_err(fs_info,
+			"device total_bytes should be at most %llu but found %llu",
+				  max_total_bytes, device->total_bytes);
+			return -EINVAL;
+		}
+	}
 	set_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &device->dev_state);
 	if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state) &&
 	   !test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state)) {
@@ -6761,7 +7453,6 @@ static int read_one_dev(struct extent_buffer *leaf,
 
 int btrfs_read_sys_array(struct btrfs_fs_info *fs_info)
 {
-	struct btrfs_root *root = fs_info->tree_root;
 	struct btrfs_super_block *super_copy = fs_info->super_copy;
 	struct extent_buffer *sb;
 	struct btrfs_disk_key *disk_key;
@@ -6777,30 +7468,16 @@ int btrfs_read_sys_array(struct btrfs_fs_info *fs_info)
 	struct btrfs_key key;
 
 	ASSERT(BTRFS_SUPER_INFO_SIZE <= fs_info->nodesize);
+
 	/*
-	 * This will create extent buffer of nodesize, superblock size is
-	 * fixed to BTRFS_SUPER_INFO_SIZE. If nodesize > sb size, this will
-	 * overallocate but we can keep it as-is, only the first page is used.
+	 * We allocated a dummy extent, just to use extent buffer accessors.
+	 * There will be unused space after BTRFS_SUPER_INFO_SIZE, but
+	 * that's fine, we will not go beyond system chunk array anyway.
 	 */
-	sb = btrfs_find_create_tree_block(fs_info, BTRFS_SUPER_INFO_OFFSET);
-	if (IS_ERR(sb))
-		return PTR_ERR(sb);
+	sb = alloc_dummy_extent_buffer(fs_info, BTRFS_SUPER_INFO_OFFSET);
+	if (!sb)
+		return -ENOMEM;
 	set_extent_buffer_uptodate(sb);
-	btrfs_set_buffer_lockdep_class(root->root_key.objectid, sb, 0);
-	/*
-	 * The sb extent buffer is artificial and just used to read the system array.
-	 * set_extent_buffer_uptodate() call does not properly mark all it's
-	 * pages up-to-date when the page is larger: extent does not cover the
-	 * whole page and consequently check_page_uptodate does not find all
-	 * the page's extents up-to-date (the hole beyond sb),
-	 * write_extent_buffer then triggers a WARN_ON.
-	 *
-	 * Regular short extents go through mark_extent_buffer_dirty/writeback cycle,
-	 * but sb spans only this function. Add an explicit SetPageUptodate call
-	 * to silence the warning eg. on PowerPC 64.
-	 */
-	if (PAGE_SIZE > BTRFS_SUPER_INFO_SIZE)
-		SetPageUptodate(sb->pages[0]);
 
 	write_extent_buffer(sb, super_copy, 0, BTRFS_SUPER_INFO_SIZE);
 	array_size = btrfs_super_sys_array_size(super_copy);
@@ -6945,6 +7622,15 @@ out:
 	return ret;
 }
 
+static void readahead_tree_node_children(struct extent_buffer *node)
+{
+	int i;
+	const int nr_items = btrfs_header_nritems(node);
+
+	for (i = 0; i < nr_items; i++)
+		btrfs_readahead_node_child(node, i);
+}
+
 int btrfs_read_chunk_tree(struct btrfs_fs_info *fs_info)
 {
 	struct btrfs_root *root = fs_info->chunk_root;
@@ -6954,7 +7640,9 @@ int btrfs_read_chunk_tree(struct btrfs_fs_info *fs_info)
 	struct btrfs_key found_key;
 	int ret;
 	int slot;
+	int iter_ret = 0;
 	u64 total_dev = 0;
+	u64 last_ra_node = 0;
 
 	path = btrfs_alloc_path();
 	if (!path)
@@ -6965,7 +7653,27 @@ int btrfs_read_chunk_tree(struct btrfs_fs_info *fs_info)
 	 * otherwise we don't need it.
 	 */
 	mutex_lock(&uuid_mutex);
-	mutex_lock(&fs_info->chunk_mutex);
+
+	/*
+	 * It is possible for mount and umount to race in such a way that
+	 * we execute this code path, but open_fs_devices failed to clear
+	 * total_rw_bytes. We certainly want it cleared before reading the
+	 * device items, so clear it here.
+	 */
+	fs_info->fs_devices->total_rw_bytes = 0;
+
+	/*
+	 * Lockdep complains about possible circular locking dependency between
+	 * a disk's open_mutex (struct gendisk.open_mutex), the rw semaphores
+	 * used for freeze procection of a fs (struct super_block.s_writers),
+	 * which we take when starting a transaction, and extent buffers of the
+	 * chunk tree if we call read_one_dev() while holding a lock on an
+	 * extent buffer of the chunk tree. Since we are mounting the filesystem
+	 * and at this point there can't be any concurrent task modifying the
+	 * chunk tree, to keep it simple, just skip locking on the chunk tree.
+	 */
+	ASSERT(!test_bit(BTRFS_FS_OPEN, &fs_info->flags));
+	path->skip_locking = 1;
 
 	/*
 	 * Read all device items, and then all the chunk items. All
@@ -6976,21 +7684,18 @@ int btrfs_read_chunk_tree(struct btrfs_fs_info *fs_info)
 	key.objectid = BTRFS_DEV_ITEMS_OBJECTID;
 	key.offset = 0;
 	key.type = 0;
-	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
-	if (ret < 0)
-		goto error;
-	while (1) {
+	btrfs_for_each_slot(root, &key, &found_key, path, iter_ret) {
+		struct extent_buffer *node = path->nodes[1];
+
 		leaf = path->nodes[0];
 		slot = path->slots[0];
-		if (slot >= btrfs_header_nritems(leaf)) {
-			ret = btrfs_next_leaf(root, path);
-			if (ret == 0)
-				continue;
-			if (ret < 0)
-				goto error;
-			break;
+
+		if (node) {
+			if (last_ra_node != node->start) {
+				readahead_tree_node_children(node);
+				last_ra_node = node->start;
+			}
 		}
-		btrfs_item_key_to_cpu(leaf, &found_key, slot);
 		if (found_key.type == BTRFS_DEV_ITEM_KEY) {
 			struct btrfs_dev_item *dev_item;
 			dev_item = btrfs_item_ptr(leaf, slot,
@@ -7001,12 +7706,25 @@ int btrfs_read_chunk_tree(struct btrfs_fs_info *fs_info)
 			total_dev++;
 		} else if (found_key.type == BTRFS_CHUNK_ITEM_KEY) {
 			struct btrfs_chunk *chunk;
+
+			/*
+			 * We are only called at mount time, so no need to take
+			 * fs_info->chunk_mutex. Plus, to avoid lockdep warnings,
+			 * we always lock first fs_info->chunk_mutex before
+			 * acquiring any locks on the chunk tree. This is a
+			 * requirement for chunk allocation, see the comment on
+			 * top of btrfs_chunk_alloc() for details.
+			 */
 			chunk = btrfs_item_ptr(leaf, slot, struct btrfs_chunk);
 			ret = read_one_chunk(&found_key, leaf, chunk);
 			if (ret)
 				goto error;
 		}
-		path->slots[0]++;
+	}
+	/* Catch error found during iteration */
+	if (iter_ret < 0) {
+		ret = iter_ret;
+		goto error;
 	}
 
 	/*
@@ -7014,12 +7732,12 @@ int btrfs_read_chunk_tree(struct btrfs_fs_info *fs_info)
 	 * do another round of validation checks.
 	 */
 	if (total_dev != fs_info->fs_devices->total_devices) {
-		btrfs_err(fs_info,
-	   "super_num_devices %llu mismatch with num_devices %llu found here",
+		btrfs_warn(fs_info,
+"super block num_devices %llu mismatch with DEV_ITEM count %llu, will be repaired on next transaction commit",
 			  btrfs_super_num_devices(fs_info->super_copy),
 			  total_dev);
-		ret = -EINVAL;
-		goto error;
+		fs_info->fs_devices->total_devices = total_dev;
+		btrfs_set_super_num_devices(fs_info->super_copy, total_dev);
 	}
 	if (btrfs_super_total_bytes(fs_info->super_copy) <
 	    fs_info->fs_devices->total_rw_bytes) {
@@ -7032,26 +7750,37 @@ int btrfs_read_chunk_tree(struct btrfs_fs_info *fs_info)
 	}
 	ret = 0;
 error:
-	mutex_unlock(&fs_info->chunk_mutex);
 	mutex_unlock(&uuid_mutex);
 
 	btrfs_free_path(path);
 	return ret;
 }
 
-void btrfs_init_devices_late(struct btrfs_fs_info *fs_info)
+int btrfs_init_devices_late(struct btrfs_fs_info *fs_info)
 {
-	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
+	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices, *seed_devs;
 	struct btrfs_device *device;
+	int ret = 0;
 
-	while (fs_devices) {
-		mutex_lock(&fs_devices->device_list_mutex);
-		list_for_each_entry(device, &fs_devices->devices, dev_list)
+	fs_devices->fs_info = fs_info;
+
+	mutex_lock(&fs_devices->device_list_mutex);
+	list_for_each_entry(device, &fs_devices->devices, dev_list)
+		device->fs_info = fs_info;
+
+	list_for_each_entry(seed_devs, &fs_devices->seed_list, seed_list) {
+		list_for_each_entry(device, &seed_devs->devices, dev_list) {
 			device->fs_info = fs_info;
-		mutex_unlock(&fs_devices->device_list_mutex);
+			ret = btrfs_get_dev_zone_info(device, false);
+			if (ret)
+				break;
+		}
 
-		fs_devices = fs_devices->seed;
+		seed_devs->fs_info = fs_info;
 	}
+	mutex_unlock(&fs_devices->device_list_mutex);
+
+	return ret;
 }
 
 static u64 btrfs_dev_stats_value(const struct extent_buffer *eb,
@@ -7077,17 +7806,56 @@ static void btrfs_set_dev_stats_value(struct extent_buffer *eb,
 			    sizeof(val));
 }
 
-int btrfs_init_dev_stats(struct btrfs_fs_info *fs_info)
+static int btrfs_device_init_dev_stats(struct btrfs_device *device,
+				       struct btrfs_path *path)
 {
-	struct btrfs_key key;
-	struct btrfs_root *dev_root = fs_info->dev_root;
-	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
+	struct btrfs_dev_stats_item *ptr;
 	struct extent_buffer *eb;
-	int slot;
-	int ret = 0;
+	struct btrfs_key key;
+	int item_size;
+	int i, ret, slot;
+
+	if (!device->fs_info->dev_root)
+		return 0;
+
+	key.objectid = BTRFS_DEV_STATS_OBJECTID;
+	key.type = BTRFS_PERSISTENT_ITEM_KEY;
+	key.offset = device->devid;
+	ret = btrfs_search_slot(NULL, device->fs_info->dev_root, &key, path, 0, 0);
+	if (ret) {
+		for (i = 0; i < BTRFS_DEV_STAT_VALUES_MAX; i++)
+			btrfs_dev_stat_set(device, i, 0);
+		device->dev_stats_valid = 1;
+		btrfs_release_path(path);
+		return ret < 0 ? ret : 0;
+	}
+	slot = path->slots[0];
+	eb = path->nodes[0];
+	item_size = btrfs_item_size(eb, slot);
+
+	ptr = btrfs_item_ptr(eb, slot, struct btrfs_dev_stats_item);
+
+	for (i = 0; i < BTRFS_DEV_STAT_VALUES_MAX; i++) {
+		if (item_size >= (1 + i) * sizeof(__le64))
+			btrfs_dev_stat_set(device, i,
+					   btrfs_dev_stats_value(eb, ptr, i));
+		else
+			btrfs_dev_stat_set(device, i, 0);
+	}
+
+	device->dev_stats_valid = 1;
+	btrfs_dev_stat_print_on_load(device);
+	btrfs_release_path(path);
+
+	return 0;
+}
+
+int btrfs_init_dev_stats(struct btrfs_fs_info *fs_info)
+{
+	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices, *seed_devs;
 	struct btrfs_device *device;
 	struct btrfs_path *path = NULL;
-	int i;
+	int ret = 0;
 
 	path = btrfs_alloc_path();
 	if (!path)
@@ -7095,43 +7863,22 @@ int btrfs_init_dev_stats(struct btrfs_fs_info *fs_info)
 
 	mutex_lock(&fs_devices->device_list_mutex);
 	list_for_each_entry(device, &fs_devices->devices, dev_list) {
-		int item_size;
-		struct btrfs_dev_stats_item *ptr;
-
-		key.objectid = BTRFS_DEV_STATS_OBJECTID;
-		key.type = BTRFS_PERSISTENT_ITEM_KEY;
-		key.offset = device->devid;
-		ret = btrfs_search_slot(NULL, dev_root, &key, path, 0, 0);
-		if (ret) {
-			for (i = 0; i < BTRFS_DEV_STAT_VALUES_MAX; i++)
-				btrfs_dev_stat_set(device, i, 0);
-			device->dev_stats_valid = 1;
-			btrfs_release_path(path);
-			continue;
-		}
-		slot = path->slots[0];
-		eb = path->nodes[0];
-		item_size = btrfs_item_size_nr(eb, slot);
-
-		ptr = btrfs_item_ptr(eb, slot,
-				     struct btrfs_dev_stats_item);
-
-		for (i = 0; i < BTRFS_DEV_STAT_VALUES_MAX; i++) {
-			if (item_size >= (1 + i) * sizeof(__le64))
-				btrfs_dev_stat_set(device, i,
-					btrfs_dev_stats_value(eb, ptr, i));
-			else
-				btrfs_dev_stat_set(device, i, 0);
+		ret = btrfs_device_init_dev_stats(device, path);
+		if (ret)
+			goto out;
+	}
+	list_for_each_entry(seed_devs, &fs_devices->seed_list, seed_list) {
+		list_for_each_entry(device, &seed_devs->devices, dev_list) {
+			ret = btrfs_device_init_dev_stats(device, path);
+			if (ret)
+				goto out;
 		}
-
-		device->dev_stats_valid = 1;
-		btrfs_dev_stat_print_on_load(device);
-		btrfs_release_path(path);
 	}
+out:
 	mutex_unlock(&fs_devices->device_list_mutex);
 
 	btrfs_free_path(path);
-	return ret < 0 ? ret : 0;
+	return ret;
 }
 
 static int update_dev_stat_item(struct btrfs_trans_handle *trans,
@@ -7162,7 +7909,7 @@ static int update_dev_stat_item(struct btrfs_trans_handle *trans,
 	}
 
 	if (ret == 0 &&
-	    btrfs_item_size_nr(path->nodes[0], path->slots[0]) < sizeof(*ptr)) {
+	    btrfs_item_size(path->nodes[0], path->slots[0]) < sizeof(*ptr)) {
 		/* need to delete old one and insert a new one */
 		ret = btrfs_del_item(trans, dev_root, path);
 		if (ret != 0) {
@@ -7242,11 +7989,7 @@ int btrfs_run_dev_stats(struct btrfs_trans_handle *trans)
 void btrfs_dev_stat_inc_and_print(struct btrfs_device *dev, int index)
 {
 	btrfs_dev_stat_inc(dev, index);
-	btrfs_dev_stat_print_on_error(dev);
-}
 
-static void btrfs_dev_stat_print_on_error(struct btrfs_device *dev)
-{
 	if (!dev->dev_stats_valid)
 		return;
 	btrfs_err_rl_in_rcu(dev->fs_info,
@@ -7282,13 +8025,14 @@ static void btrfs_dev_stat_print_on_load(struct btrfs_device *dev)
 int btrfs_get_dev_stats(struct btrfs_fs_info *fs_info,
 			struct btrfs_ioctl_get_dev_stats *stats)
 {
+	BTRFS_DEV_LOOKUP_ARGS(args);
 	struct btrfs_device *dev;
 	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
 	int i;
 
 	mutex_lock(&fs_devices->device_list_mutex);
-	dev = btrfs_find_device(fs_info->fs_devices, stats->devid, NULL, NULL,
-				true);
+	args.devid = stats->devid;
+	dev = btrfs_find_device(fs_info->fs_devices, &args);
 	mutex_unlock(&fs_devices->device_list_mutex);
 
 	if (!dev) {
@@ -7317,36 +8061,6 @@ int btrfs_get_dev_stats(struct btrfs_fs_info *fs_info,
 	return 0;
 }
 
-void btrfs_scratch_superblocks(struct block_device *bdev, const char *device_path)
-{
-	struct buffer_head *bh;
-	struct btrfs_super_block *disk_super;
-	int copy_num;
-
-	if (!bdev)
-		return;
-
-	for (copy_num = 0; copy_num < BTRFS_SUPER_MIRROR_MAX;
-		copy_num++) {
-
-		if (btrfs_read_dev_one_super(bdev, copy_num, &bh))
-			continue;
-
-		disk_super = (struct btrfs_super_block *)bh->b_data;
-
-		memset(&disk_super->magic, 0, sizeof(disk_super->magic));
-		set_buffer_dirty(bh);
-		sync_dirty_buffer(bh);
-		brelse(bh);
-	}
-
-	/* Notify udev that device has changed */
-	btrfs_kobject_uevent(bdev, KOBJ_CHANGE);
-
-	/* Update ctime/mtime for device path for libblkid */
-	update_dev_time(device_path);
-}
-
 /*
  * Update the size and bytes used for each device where it changed.  This is
  * delayed since we would otherwise get errors while writing out the
@@ -7378,24 +8092,6 @@ void btrfs_commit_device_sizes(struct btrfs_transaction *trans)
 	mutex_unlock(&trans->fs_info->chunk_mutex);
 }
 
-void btrfs_set_fs_info_ptr(struct btrfs_fs_info *fs_info)
-{
-	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
-	while (fs_devices) {
-		fs_devices->fs_info = fs_info;
-		fs_devices = fs_devices->seed;
-	}
-}
-
-void btrfs_reset_fs_info_ptr(struct btrfs_fs_info *fs_info)
-{
-	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
-	while (fs_devices) {
-		fs_devices->fs_info = NULL;
-		fs_devices = fs_devices->seed;
-	}
-}
-
 /*
  * Multiplicity factor for simple profiles: DUP, RAID1-like and RAID10.
  */
@@ -7412,6 +8108,7 @@ static int verify_one_dev_extent(struct btrfs_fs_info *fs_info,
 				 u64 chunk_offset, u64 devid,
 				 u64 physical_offset, u64 physical_len)
 {
+	struct btrfs_dev_lookup_args args = { .devid = devid };
 	struct extent_map_tree *em_tree = &fs_info->mapping_tree;
 	struct extent_map *em;
 	struct map_lookup *map;
@@ -7434,7 +8131,7 @@ static int verify_one_dev_extent(struct btrfs_fs_info *fs_info,
 	}
 
 	map = em->map_lookup;
-	stripe_len = calc_stripe_length(map->type, em->len, map->num_stripes);
+	stripe_len = btrfs_calc_stripe_length(em);
 	if (physical_len != stripe_len) {
 		btrfs_err(fs_info,
 "dev extent physical offset %llu on devid %llu length doesn't match chunk %llu, have %llu expect %llu",
@@ -7444,6 +8141,16 @@ static int verify_one_dev_extent(struct btrfs_fs_info *fs_info,
 		goto out;
 	}
 
+	/*
+	 * Very old mkfs.btrfs (before v4.1) will not respect the reserved
+	 * space. Although kernel can handle it without problem, better to warn
+	 * the users.
+	 */
+	if (physical_offset < BTRFS_DEVICE_RANGE_RESERVED)
+		btrfs_warn(fs_info,
+		"devid %llu physical %llu len %llu inside the reserved space",
+			   devid, physical_offset, physical_len);
+
 	for (i = 0; i < map->num_stripes; i++) {
 		if (map->stripes[i].dev->devid == devid &&
 		    map->stripes[i].physical == physical_offset) {
@@ -7466,26 +8173,14 @@ static int verify_one_dev_extent(struct btrfs_fs_info *fs_info,
 		ret = -EUCLEAN;
 	}
 
-	/* Make sure no dev extent is beyond device bondary */
-	dev = btrfs_find_device(fs_info->fs_devices, devid, NULL, NULL, true);
+	/* Make sure no dev extent is beyond device boundary */
+	dev = btrfs_find_device(fs_info->fs_devices, &args);
 	if (!dev) {
 		btrfs_err(fs_info, "failed to find devid %llu", devid);
 		ret = -EUCLEAN;
 		goto out;
 	}
 
-	/* It's possible this device is a dummy for seed device */
-	if (dev->disk_total_bytes == 0) {
-		dev = btrfs_find_device(fs_info->fs_devices->seed, devid, NULL,
-					NULL, false);
-		if (!dev) {
-			btrfs_err(fs_info, "failed to find seed devid %llu",
-				  devid);
-			ret = -EUCLEAN;
-			goto out;
-		}
-	}
-
 	if (physical_offset + physical_len > dev->disk_total_bytes) {
 		btrfs_err(fs_info,
 "dev extent devid %llu physical offset %llu len %llu is beyond device boundary %llu",
@@ -7494,6 +8189,20 @@ static int verify_one_dev_extent(struct btrfs_fs_info *fs_info,
 		ret = -EUCLEAN;
 		goto out;
 	}
+
+	if (dev->zone_info) {
+		u64 zone_size = dev->zone_info->zone_size;
+
+		if (!IS_ALIGNED(physical_offset, zone_size) ||
+		    !IS_ALIGNED(physical_len, zone_size)) {
+			btrfs_err(fs_info,
+"zoned: dev extent devid %llu physical offset %llu len %llu is not aligned to device zone",
+				  devid, physical_offset, physical_len);
+			ret = -EUCLEAN;
+			goto out;
+		}
+	}
+
 out:
 	free_extent_map(em);
 	return ret;
@@ -7540,6 +8249,19 @@ int btrfs_verify_dev_extents(struct btrfs_fs_info *fs_info)
 	u64 prev_dev_ext_end = 0;
 	int ret = 0;
 
+	/*
+	 * We don't have a dev_root because we mounted with ignorebadroots and
+	 * failed to load the root, so we want to skip the verification in this
+	 * case for sure.
+	 *
+	 * However if the dev root is fine, but the tree itself is corrupted
+	 * we'd still fail to mount.  This verification is only to make sure
+	 * writes can happen safely, so instead just bypass this check
+	 * completely in the case of IGNOREBADROOTS.
+	 */
+	if (btrfs_test_opt(fs_info, IGNOREBADROOTS))
+		return 0;
+
 	key.objectid = 1;
 	key.type = BTRFS_DEV_EXTENT_KEY;
 	key.offset = 0;
@@ -7554,7 +8276,7 @@ int btrfs_verify_dev_extents(struct btrfs_fs_info *fs_info)
 		goto out;
 
 	if (path->slots[0] >= btrfs_header_nritems(path->nodes[0])) {
-		ret = btrfs_next_item(root, path);
+		ret = btrfs_next_leaf(root, path);
 		if (ret < 0)
 			goto out;
 		/* No dev extents at all? Not good */
@@ -7637,3 +8359,91 @@ bool btrfs_pinned_by_swapfile(struct btrfs_fs_info *fs_info, void *ptr)
 	spin_unlock(&fs_info->swapfile_pins_lock);
 	return node != NULL;
 }
+
+static int relocating_repair_kthread(void *data)
+{
+	struct btrfs_block_group *cache = data;
+	struct btrfs_fs_info *fs_info = cache->fs_info;
+	u64 target;
+	int ret = 0;
+
+	target = cache->start;
+	btrfs_put_block_group(cache);
+
+	sb_start_write(fs_info->sb);
+	if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_BALANCE)) {
+		btrfs_info(fs_info,
+			   "zoned: skip relocating block group %llu to repair: EBUSY",
+			   target);
+		sb_end_write(fs_info->sb);
+		return -EBUSY;
+	}
+
+	mutex_lock(&fs_info->reclaim_bgs_lock);
+
+	/* Ensure block group still exists */
+	cache = btrfs_lookup_block_group(fs_info, target);
+	if (!cache)
+		goto out;
+
+	if (!test_bit(BLOCK_GROUP_FLAG_RELOCATING_REPAIR, &cache->runtime_flags))
+		goto out;
+
+	ret = btrfs_may_alloc_data_chunk(fs_info, target);
+	if (ret < 0)
+		goto out;
+
+	btrfs_info(fs_info,
+		   "zoned: relocating block group %llu to repair IO failure",
+		   target);
+	ret = btrfs_relocate_chunk(fs_info, target);
+
+out:
+	if (cache)
+		btrfs_put_block_group(cache);
+	mutex_unlock(&fs_info->reclaim_bgs_lock);
+	btrfs_exclop_finish(fs_info);
+	sb_end_write(fs_info->sb);
+
+	return ret;
+}
+
+bool btrfs_repair_one_zone(struct btrfs_fs_info *fs_info, u64 logical)
+{
+	struct btrfs_block_group *cache;
+
+	if (!btrfs_is_zoned(fs_info))
+		return false;
+
+	/* Do not attempt to repair in degraded state */
+	if (btrfs_test_opt(fs_info, DEGRADED))
+		return true;
+
+	cache = btrfs_lookup_block_group(fs_info, logical);
+	if (!cache)
+		return true;
+
+	if (test_and_set_bit(BLOCK_GROUP_FLAG_RELOCATING_REPAIR, &cache->runtime_flags)) {
+		btrfs_put_block_group(cache);
+		return true;
+	}
+
+	kthread_run(relocating_repair_kthread, cache,
+		    "btrfs-relocating-repair");
+
+	return true;
+}
+
+int __init btrfs_bioset_init(void)
+{
+	if (bioset_init(&btrfs_bioset, BIO_POOL_SIZE,
+			offsetof(struct btrfs_bio, bio),
+			BIOSET_NEED_BVECS))
+		return -ENOMEM;
+	return 0;
+}
+
+void __cold btrfs_bioset_exit(void)
+{
+	bioset_exit(&btrfs_bioset);
+}
diff --git a/fs/btrfs/volumes.h b/fs/btrfs/volumes.h
index f01552a0785e..099def5613b8 100644
--- a/fs/btrfs/volumes.h
+++ b/fs/btrfs/volumes.h
@@ -17,7 +17,39 @@ extern struct mutex uuid_mutex;
 
 #define BTRFS_STRIPE_LEN	SZ_64K
 
-struct buffer_head;
+/* Used by sanity check for btrfs_raid_types. */
+#define const_ffs(n) (__builtin_ctzll(n) + 1)
+
+/*
+ * The conversion from BTRFS_BLOCK_GROUP_* bits to btrfs_raid_type requires
+ * RAID0 always to be the lowest profile bit.
+ * Although it's part of on-disk format and should never change, do extra
+ * compile-time sanity checks.
+ */
+static_assert(const_ffs(BTRFS_BLOCK_GROUP_RAID0) <
+	      const_ffs(BTRFS_BLOCK_GROUP_PROFILE_MASK & ~BTRFS_BLOCK_GROUP_RAID0));
+static_assert(const_ilog2(BTRFS_BLOCK_GROUP_RAID0) >
+	      ilog2(BTRFS_BLOCK_GROUP_TYPE_MASK));
+
+/* ilog2() can handle both constants and variables */
+#define BTRFS_BG_FLAG_TO_INDEX(profile)					\
+	ilog2((profile) >> (ilog2(BTRFS_BLOCK_GROUP_RAID0) - 1))
+
+enum btrfs_raid_types {
+	/* SINGLE is the special one as it doesn't have on-disk bit. */
+	BTRFS_RAID_SINGLE  = 0,
+
+	BTRFS_RAID_RAID0   = BTRFS_BG_FLAG_TO_INDEX(BTRFS_BLOCK_GROUP_RAID0),
+	BTRFS_RAID_RAID1   = BTRFS_BG_FLAG_TO_INDEX(BTRFS_BLOCK_GROUP_RAID1),
+	BTRFS_RAID_DUP	   = BTRFS_BG_FLAG_TO_INDEX(BTRFS_BLOCK_GROUP_DUP),
+	BTRFS_RAID_RAID10  = BTRFS_BG_FLAG_TO_INDEX(BTRFS_BLOCK_GROUP_RAID10),
+	BTRFS_RAID_RAID5   = BTRFS_BG_FLAG_TO_INDEX(BTRFS_BLOCK_GROUP_RAID5),
+	BTRFS_RAID_RAID6   = BTRFS_BG_FLAG_TO_INDEX(BTRFS_BLOCK_GROUP_RAID6),
+	BTRFS_RAID_RAID1C3 = BTRFS_BG_FLAG_TO_INDEX(BTRFS_BLOCK_GROUP_RAID1C3),
+	BTRFS_RAID_RAID1C4 = BTRFS_BG_FLAG_TO_INDEX(BTRFS_BLOCK_GROUP_RAID1C4),
+
+	BTRFS_NR_RAID_TYPES
+};
 
 struct btrfs_io_geometry {
 	/* remaining bytes before crossing a stripe */
@@ -25,11 +57,11 @@ struct btrfs_io_geometry {
 	/* offset of logical address in chunk */
 	u64 offset;
 	/* length of single IO stripe */
-	u64 stripe_len;
+	u32 stripe_len;
+	/* offset of address in stripe */
+	u32 stripe_offset;
 	/* number of stripe where address falls */
 	u64 stripe_nr;
-	/* offset of address in stripe */
-	u64 stripe_offset;
 	/* offset of raid56 stripe into the chunk */
 	u64 raid56_stripe_offset;
 };
@@ -52,6 +84,9 @@ struct btrfs_io_geometry {
 #define BTRFS_DEV_STATE_MISSING		(2)
 #define BTRFS_DEV_STATE_REPLACE_TGT	(3)
 #define BTRFS_DEV_STATE_FLUSH_SENT	(4)
+#define BTRFS_DEV_STATE_NO_READA	(5)
+
+struct btrfs_zoned_device_info;
 
 struct btrfs_device {
 	struct list_head dev_list; /* device_list_mutex */
@@ -60,15 +95,22 @@ struct btrfs_device {
 	struct btrfs_fs_devices *fs_devices;
 	struct btrfs_fs_info *fs_info;
 
-	struct rcu_string *name;
+	struct rcu_string __rcu *name;
 
 	u64 generation;
 
 	struct block_device *bdev;
 
+	struct btrfs_zoned_device_info *zone_info;
+
 	/* the mode sent to blkdev_get */
 	fmode_t mode;
 
+	/*
+	 * Device's major-minor number. Must be set even if the device is not
+	 * opened (bdev == NULL), unless the device is missing.
+	 */
+	dev_t devt;
 	unsigned long dev_state;
 	blk_status_t last_flush_error;
 
@@ -113,20 +155,13 @@ struct btrfs_device {
 	/* bytes used on the current transaction */
 	u64 commit_bytes_used;
 
-	/* for sending down flush barriers */
-	struct bio *flush_bio;
+	/* Bio used for flushing device barriers */
+	struct bio flush_bio;
 	struct completion flush_wait;
 
 	/* per-device scrub information */
 	struct scrub_ctx *scrub_ctx;
 
-	/* readahead state */
-	atomic_t reada_in_flight;
-	u64 reada_next;
-	struct reada_zone *reada_curr_zone;
-	struct radix_tree_root reada_zones;
-	struct radix_tree_root reada_extents;
-
 	/* disk I/O failure stats. For detailed description refer to
 	 * enum btrfs_dev_stat_values in ioctl.h */
 	int dev_stats_valid;
@@ -140,6 +175,34 @@ struct btrfs_device {
 	struct completion kobj_unregister;
 	/* For sysfs/FSID/devinfo/devid/ */
 	struct kobject devid_kobj;
+
+	/* Bandwidth limit for scrub, in bytes */
+	u64 scrub_speed_max;
+};
+
+/*
+ * Block group or device which contains an active swapfile. Used for preventing
+ * unsafe operations while a swapfile is active.
+ *
+ * These are sorted on (ptr, inode) (note that a block group or device can
+ * contain more than one swapfile). We compare the pointer values because we
+ * don't actually care what the object is, we just need a quick check whether
+ * the object exists in the rbtree.
+ */
+struct btrfs_swapfile_pin {
+	struct rb_node node;
+	void *ptr;
+	struct inode *inode;
+	/*
+	 * If true, ptr points to a struct btrfs_block_group. Otherwise, ptr
+	 * points to a struct btrfs_device.
+	 */
+	bool is_block_group;
+	/*
+	 * Only used when 'is_block_group' is true and it is the number of
+	 * extents used by a swapfile for this block group ('ptr' field).
+	 */
+	int bg_extent_count;
 };
 
 /*
@@ -209,23 +272,61 @@ BTRFS_DEVICE_GETSET_FUNCS(total_bytes);
 BTRFS_DEVICE_GETSET_FUNCS(disk_total_bytes);
 BTRFS_DEVICE_GETSET_FUNCS(bytes_used);
 
+enum btrfs_chunk_allocation_policy {
+	BTRFS_CHUNK_ALLOC_REGULAR,
+	BTRFS_CHUNK_ALLOC_ZONED,
+};
+
+/*
+ * Read policies for mirrored block group profiles, read picks the stripe based
+ * on these policies.
+ */
+enum btrfs_read_policy {
+	/* Use process PID to choose the stripe */
+	BTRFS_READ_POLICY_PID,
+	BTRFS_NR_READ_POLICY,
+};
+
 struct btrfs_fs_devices {
 	u8 fsid[BTRFS_FSID_SIZE]; /* FS specific uuid */
 	u8 metadata_uuid[BTRFS_FSID_SIZE];
 	bool fsid_change;
 	struct list_head fs_list;
 
+	/*
+	 * Number of devices under this fsid including missing and
+	 * replace-target device and excludes seed devices.
+	 */
 	u64 num_devices;
+
+	/*
+	 * The number of devices that successfully opened, including
+	 * replace-target, excludes seed devices.
+	 */
 	u64 open_devices;
+
+	/* The number of devices that are under the chunk allocation list. */
 	u64 rw_devices;
+
+	/* Count of missing devices under this fsid excluding seed device. */
 	u64 missing_devices;
 	u64 total_rw_bytes;
+
+	/*
+	 * Count of devices from btrfs_super_block::num_devices for this fsid,
+	 * which includes the seed device, excludes the transient replace-target
+	 * device.
+	 */
 	u64 total_devices;
 
 	/* Highest generation number of seen devices */
 	u64 latest_generation;
 
-	struct block_device *latest_bdev;
+	/*
+	 * The mount device or a device with highest generation after removal
+	 * or replace.
+	 */
+	struct btrfs_device *latest_dev;
 
 	/* all of the devices in the FS, protected by a mutex
 	 * so we can safely walk it to write out the supers without
@@ -244,7 +345,7 @@ struct btrfs_fs_devices {
 	 */
 	struct list_head alloc_list;
 
-	struct btrfs_fs_devices *seed;
+	struct list_head seed_list;
 	bool seeding;
 
 	int opened;
@@ -260,6 +361,11 @@ struct btrfs_fs_devices {
 	struct kobject *devices_kobj;
 	struct kobject *devinfo_kobj;
 	struct completion kobj_unregister;
+
+	enum btrfs_chunk_allocation_policy chunk_alloc_policy;
+
+	/* Policy used to read the mirrored stripes */
+	enum btrfs_read_policy read_policy;
 };
 
 #define BTRFS_BIO_INLINE_CSUM_SIZE	64
@@ -274,55 +380,125 @@ struct btrfs_fs_devices {
 				/ sizeof(struct btrfs_stripe) + 1)
 
 /*
- * we need the mirror number and stripe index to be passed around
- * the call chain while we are processing end_io (especially errors).
- * Really, what we need is a btrfs_bio structure that has this info
- * and is properly sized with its stripe array, but we're not there
- * quite yet.  We have our own btrfs bioset, and all of the bios
- * we allocate are actually btrfs_io_bios.  We'll cram as much of
- * struct btrfs_bio as we can into this over time.
+ * Maximum number of sectors for a single bio to limit the size of the
+ * checksum array.  This matches the number of bio_vecs per bio and thus the
+ * I/O size for buffered I/O.
  */
-struct btrfs_io_bio {
+#define BTRFS_MAX_BIO_SECTORS				(256)
+
+typedef void (*btrfs_bio_end_io_t)(struct btrfs_bio *bbio);
+
+/*
+ * Additional info to pass along bio.
+ *
+ * Mostly for btrfs specific features like csum and mirror_num.
+ */
+struct btrfs_bio {
 	unsigned int mirror_num;
-	unsigned int stripe_index;
-	u64 logical;
+	struct bvec_iter iter;
+
+	/* for direct I/O */
+	u64 file_offset;
+
+	/* @device is for stripe IO submission. */
+	struct btrfs_device *device;
 	u8 *csum;
 	u8 csum_inline[BTRFS_BIO_INLINE_CSUM_SIZE];
-	struct bvec_iter iter;
+
+	/* End I/O information supplied to btrfs_bio_alloc */
+	btrfs_bio_end_io_t end_io;
+	void *private;
+
+	/* For read end I/O handling */
+	struct work_struct end_io_work;
+
 	/*
 	 * This member must come last, bio_alloc_bioset will allocate enough
-	 * bytes for entire btrfs_io_bio but relies on bio being last.
+	 * bytes for entire btrfs_bio but relies on bio being last.
 	 */
 	struct bio bio;
 };
 
-static inline struct btrfs_io_bio *btrfs_io_bio(struct bio *bio)
+static inline struct btrfs_bio *btrfs_bio(struct bio *bio)
+{
+	return container_of(bio, struct btrfs_bio, bio);
+}
+
+int __init btrfs_bioset_init(void);
+void __cold btrfs_bioset_exit(void);
+
+struct bio *btrfs_bio_alloc(unsigned int nr_vecs, blk_opf_t opf,
+			    btrfs_bio_end_io_t end_io, void *private);
+struct bio *btrfs_bio_clone_partial(struct bio *orig, u64 offset, u64 size,
+				    btrfs_bio_end_io_t end_io, void *private);
+
+static inline void btrfs_bio_end_io(struct btrfs_bio *bbio, blk_status_t status)
 {
-	return container_of(bio, struct btrfs_io_bio, bio);
+	bbio->bio.bi_status = status;
+	bbio->end_io(bbio);
 }
 
-static inline void btrfs_io_bio_free_csum(struct btrfs_io_bio *io_bio)
+static inline void btrfs_bio_free_csum(struct btrfs_bio *bbio)
 {
-	if (io_bio->csum != io_bio->csum_inline) {
-		kfree(io_bio->csum);
-		io_bio->csum = NULL;
+	if (bbio->csum != bbio->csum_inline) {
+		kfree(bbio->csum);
+		bbio->csum = NULL;
 	}
 }
 
-struct btrfs_bio_stripe {
+/*
+ * Iterate through a btrfs_bio (@bbio) on a per-sector basis.
+ *
+ * bvl        - struct bio_vec
+ * bbio       - struct btrfs_bio
+ * iters      - struct bvec_iter
+ * bio_offset - unsigned int
+ */
+#define btrfs_bio_for_each_sector(fs_info, bvl, bbio, iter, bio_offset)	\
+	for ((iter) = (bbio)->iter, (bio_offset) = 0;			\
+	     (iter).bi_size &&					\
+	     (((bvl) = bio_iter_iovec((&(bbio)->bio), (iter))), 1);	\
+	     (bio_offset) += fs_info->sectorsize,			\
+	     bio_advance_iter_single(&(bbio)->bio, &(iter),		\
+	     (fs_info)->sectorsize))
+
+struct btrfs_io_stripe {
+	struct btrfs_device *dev;
+	union {
+		/* Block mapping */
+		u64 physical;
+		/* For the endio handler */
+		struct btrfs_io_context *bioc;
+	};
+};
+
+struct btrfs_discard_stripe {
 	struct btrfs_device *dev;
 	u64 physical;
-	u64 length; /* only used for discard mappings */
+	u64 length;
 };
 
-struct btrfs_bio {
+/*
+ * Context for IO subsmission for device stripe.
+ *
+ * - Track the unfinished mirrors for mirror based profiles
+ *   Mirror based profiles are SINGLE/DUP/RAID1/RAID10.
+ *
+ * - Contain the logical -> physical mapping info
+ *   Used by submit_stripe_bio() for mapping logical bio
+ *   into physical device address.
+ *
+ * - Contain device replace info
+ *   Used by handle_ops_on_dev_replace() to copy logical bios
+ *   into the new device.
+ *
+ * - Contain RAID56 full stripe logical bytenrs
+ */
+struct btrfs_io_context {
 	refcount_t refs;
-	atomic_t stripes_pending;
 	struct btrfs_fs_info *fs_info;
 	u64 map_type; /* get from map_lookup->type */
-	bio_end_io_t *end_io;
 	struct bio *orig_bio;
-	void *private;
 	atomic_t error;
 	int max_errors;
 	int num_stripes;
@@ -335,7 +511,7 @@ struct btrfs_bio {
 	 * so raid_map[0] is the start of our full stripe
 	 */
 	u64 *raid_map;
-	struct btrfs_bio_stripe stripes[];
+	struct btrfs_io_stripe stripes[];
 };
 
 struct btrfs_device_info {
@@ -366,15 +542,15 @@ struct map_lookup {
 	u64 type;
 	int io_align;
 	int io_width;
-	u64 stripe_len;
+	u32 stripe_len;
 	int num_stripes;
 	int sub_stripes;
 	int verified_stripes; /* For mount time dev extent verification */
-	struct btrfs_bio_stripe stripes[];
+	struct btrfs_io_stripe stripes[];
 };
 
 #define map_lookup_size(n) (sizeof(struct map_lookup) + \
-			    (sizeof(struct btrfs_bio_stripe) * (n)))
+			    (sizeof(struct btrfs_io_stripe) * (n)))
 
 struct btrfs_balance_args;
 struct btrfs_balance_progress;
@@ -388,6 +564,22 @@ struct btrfs_balance_control {
 	struct btrfs_balance_progress stat;
 };
 
+/*
+ * Search for a given device by the set parameters
+ */
+struct btrfs_dev_lookup_args {
+	u64 devid;
+	u8 *uuid;
+	u8 *fsid;
+	bool missing;
+};
+
+/* We have to initialize to -1 because BTRFS_DEV_REPLACE_DEVID is 0 */
+#define BTRFS_DEV_LOOKUP_ARGS_INIT { .devid = (u64)-1 }
+
+#define BTRFS_DEV_LOOKUP_ARGS(name) \
+	struct btrfs_dev_lookup_args name = BTRFS_DEV_LOOKUP_ARGS_INIT
+
 enum btrfs_map_op {
 	BTRFS_MAP_READ,
 	BTRFS_MAP_WRITE,
@@ -401,55 +593,65 @@ static inline enum btrfs_map_op btrfs_op(struct bio *bio)
 	case REQ_OP_DISCARD:
 		return BTRFS_MAP_DISCARD;
 	case REQ_OP_WRITE:
+	case REQ_OP_ZONE_APPEND:
 		return BTRFS_MAP_WRITE;
 	default:
 		WARN_ON_ONCE(1);
-		/* fall through */
+		fallthrough;
 	case REQ_OP_READ:
 		return BTRFS_MAP_READ;
 	}
 }
 
-void btrfs_get_bbio(struct btrfs_bio *bbio);
-void btrfs_put_bbio(struct btrfs_bio *bbio);
+void btrfs_get_bioc(struct btrfs_io_context *bioc);
+void btrfs_put_bioc(struct btrfs_io_context *bioc);
 int btrfs_map_block(struct btrfs_fs_info *fs_info, enum btrfs_map_op op,
 		    u64 logical, u64 *length,
-		    struct btrfs_bio **bbio_ret, int mirror_num);
+		    struct btrfs_io_context **bioc_ret, int mirror_num);
 int btrfs_map_sblock(struct btrfs_fs_info *fs_info, enum btrfs_map_op op,
 		     u64 logical, u64 *length,
-		     struct btrfs_bio **bbio_ret);
-int btrfs_get_io_geometry(struct btrfs_fs_info *fs_info, enum btrfs_map_op op,
-		u64 logical, u64 len, struct btrfs_io_geometry *io_geom);
+		     struct btrfs_io_context **bioc_ret);
+struct btrfs_discard_stripe *btrfs_map_discard(struct btrfs_fs_info *fs_info,
+					       u64 logical, u64 *length_ret,
+					       u32 *num_stripes);
+int btrfs_get_io_geometry(struct btrfs_fs_info *fs_info, struct extent_map *map,
+			  enum btrfs_map_op op, u64 logical,
+			  struct btrfs_io_geometry *io_geom);
 int btrfs_read_sys_array(struct btrfs_fs_info *fs_info);
 int btrfs_read_chunk_tree(struct btrfs_fs_info *fs_info);
-int btrfs_alloc_chunk(struct btrfs_trans_handle *trans, u64 type);
+struct btrfs_block_group *btrfs_create_chunk(struct btrfs_trans_handle *trans,
+					    u64 type);
 void btrfs_mapping_tree_free(struct extent_map_tree *tree);
-blk_status_t btrfs_map_bio(struct btrfs_fs_info *fs_info, struct bio *bio,
-			   int mirror_num);
+void btrfs_submit_bio(struct btrfs_fs_info *fs_info, struct bio *bio, int mirror_num);
 int btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
 		       fmode_t flags, void *holder);
 struct btrfs_device *btrfs_scan_one_device(const char *path,
 					   fmode_t flags, void *holder);
-int btrfs_forget_devices(const char *path);
-int btrfs_close_devices(struct btrfs_fs_devices *fs_devices);
-void btrfs_free_extra_devids(struct btrfs_fs_devices *fs_devices, int step);
+int btrfs_forget_devices(dev_t devt);
+void btrfs_close_devices(struct btrfs_fs_devices *fs_devices);
+void btrfs_free_extra_devids(struct btrfs_fs_devices *fs_devices);
 void btrfs_assign_next_active_device(struct btrfs_device *device,
 				     struct btrfs_device *this_dev);
 struct btrfs_device *btrfs_find_device_by_devspec(struct btrfs_fs_info *fs_info,
 						  u64 devid,
 						  const char *devpath);
+int btrfs_get_dev_args_from_path(struct btrfs_fs_info *fs_info,
+				 struct btrfs_dev_lookup_args *args,
+				 const char *path);
 struct btrfs_device *btrfs_alloc_device(struct btrfs_fs_info *fs_info,
 					const u64 *devid,
 					const u8 *uuid);
+void btrfs_put_dev_args_from_path(struct btrfs_dev_lookup_args *args);
 void btrfs_free_device(struct btrfs_device *device);
 int btrfs_rm_device(struct btrfs_fs_info *fs_info,
-		    const char *device_path, u64 devid);
+		    struct btrfs_dev_lookup_args *args,
+		    struct block_device **bdev, fmode_t *mode);
 void __exit btrfs_cleanup_fs_uuids(void);
 int btrfs_num_copies(struct btrfs_fs_info *fs_info, u64 logical, u64 len);
 int btrfs_grow_device(struct btrfs_trans_handle *trans,
 		      struct btrfs_device *device, u64 new_size);
-struct btrfs_device *btrfs_find_device(struct btrfs_fs_devices *fs_devices,
-				       u64 devid, u8 *uuid, u8 *fsid, bool seed);
+struct btrfs_device *btrfs_find_device(const struct btrfs_fs_devices *fs_devices,
+				       const struct btrfs_dev_lookup_args *args);
 int btrfs_shrink_device(struct btrfs_device *device, u64 new_size);
 int btrfs_init_new_device(struct btrfs_fs_info *fs_info, const char *path);
 int btrfs_balance(struct btrfs_fs_info *fs_info,
@@ -459,31 +661,34 @@ void btrfs_describe_block_groups(u64 flags, char *buf, u32 size_buf);
 int btrfs_resume_balance_async(struct btrfs_fs_info *fs_info);
 int btrfs_recover_balance(struct btrfs_fs_info *fs_info);
 int btrfs_pause_balance(struct btrfs_fs_info *fs_info);
+int btrfs_relocate_chunk(struct btrfs_fs_info *fs_info, u64 chunk_offset);
 int btrfs_cancel_balance(struct btrfs_fs_info *fs_info);
 int btrfs_create_uuid_tree(struct btrfs_fs_info *fs_info);
-int btrfs_check_uuid_tree(struct btrfs_fs_info *fs_info);
-int btrfs_chunk_readonly(struct btrfs_fs_info *fs_info, u64 chunk_offset);
+int btrfs_uuid_scan_kthread(void *data);
+bool btrfs_chunk_writeable(struct btrfs_fs_info *fs_info, u64 chunk_offset);
 int find_free_dev_extent(struct btrfs_device *device, u64 num_bytes,
 			 u64 *start, u64 *max_avail);
 void btrfs_dev_stat_inc_and_print(struct btrfs_device *dev, int index);
 int btrfs_get_dev_stats(struct btrfs_fs_info *fs_info,
 			struct btrfs_ioctl_get_dev_stats *stats);
-void btrfs_init_devices_late(struct btrfs_fs_info *fs_info);
+int btrfs_init_devices_late(struct btrfs_fs_info *fs_info);
 int btrfs_init_dev_stats(struct btrfs_fs_info *fs_info);
 int btrfs_run_dev_stats(struct btrfs_trans_handle *trans);
 void btrfs_rm_dev_replace_remove_srcdev(struct btrfs_device *srcdev);
 void btrfs_rm_dev_replace_free_srcdev(struct btrfs_device *srcdev);
 void btrfs_destroy_dev_replace_tgtdev(struct btrfs_device *tgtdev);
-void btrfs_scratch_superblocks(struct block_device *bdev, const char *device_path);
 int btrfs_is_parity_mirror(struct btrfs_fs_info *fs_info,
 			   u64 logical, u64 len);
 unsigned long btrfs_full_stripe_len(struct btrfs_fs_info *fs_info,
 				    u64 logical);
-int btrfs_finish_chunk_alloc(struct btrfs_trans_handle *trans,
-			     u64 chunk_offset, u64 chunk_size);
+u64 btrfs_calc_stripe_length(const struct extent_map *em);
+int btrfs_nr_parity_stripes(u64 type);
+int btrfs_chunk_alloc_add_chunk_item(struct btrfs_trans_handle *trans,
+				     struct btrfs_block_group *bg);
 int btrfs_remove_chunk(struct btrfs_trans_handle *trans, u64 chunk_offset);
 struct extent_map *btrfs_get_chunk_map(struct btrfs_fs_info *fs_info,
 				       u64 logical, u64 length);
+void btrfs_release_disk_super(struct btrfs_super_block *super);
 
 static inline void btrfs_dev_stat_inc(struct btrfs_device *dev,
 				      int index)
@@ -536,42 +741,21 @@ static inline void btrfs_dev_stat_set(struct btrfs_device *dev,
 	atomic_inc(&dev->dev_stats_ccnt);
 }
 
-/*
- * Convert block group flags (BTRFS_BLOCK_GROUP_*) to btrfs_raid_types, which
- * can be used as index to access btrfs_raid_array[].
- */
-static inline enum btrfs_raid_types btrfs_bg_flags_to_raid_index(u64 flags)
-{
-	if (flags & BTRFS_BLOCK_GROUP_RAID10)
-		return BTRFS_RAID_RAID10;
-	else if (flags & BTRFS_BLOCK_GROUP_RAID1)
-		return BTRFS_RAID_RAID1;
-	else if (flags & BTRFS_BLOCK_GROUP_RAID1C3)
-		return BTRFS_RAID_RAID1C3;
-	else if (flags & BTRFS_BLOCK_GROUP_RAID1C4)
-		return BTRFS_RAID_RAID1C4;
-	else if (flags & BTRFS_BLOCK_GROUP_DUP)
-		return BTRFS_RAID_DUP;
-	else if (flags & BTRFS_BLOCK_GROUP_RAID0)
-		return BTRFS_RAID_RAID0;
-	else if (flags & BTRFS_BLOCK_GROUP_RAID5)
-		return BTRFS_RAID_RAID5;
-	else if (flags & BTRFS_BLOCK_GROUP_RAID6)
-		return BTRFS_RAID_RAID6;
-
-	return BTRFS_RAID_SINGLE; /* BTRFS_BLOCK_GROUP_SINGLE */
-}
-
 void btrfs_commit_device_sizes(struct btrfs_transaction *trans);
 
 struct list_head * __attribute_const__ btrfs_get_fs_uuids(void);
-void btrfs_set_fs_info_ptr(struct btrfs_fs_info *fs_info);
-void btrfs_reset_fs_info_ptr(struct btrfs_fs_info *fs_info);
 bool btrfs_check_rw_degradable(struct btrfs_fs_info *fs_info,
 					struct btrfs_device *failing_dev);
+void btrfs_scratch_superblocks(struct btrfs_fs_info *fs_info,
+			       struct block_device *bdev,
+			       const char *device_path);
 
+enum btrfs_raid_types __attribute_const__ btrfs_bg_flags_to_raid_index(u64 flags);
 int btrfs_bg_type_to_factor(u64 flags);
 const char *btrfs_bg_type_to_raid_name(u64 flags);
 int btrfs_verify_dev_extents(struct btrfs_fs_info *fs_info);
+bool btrfs_repair_one_zone(struct btrfs_fs_info *fs_info, u64 logical);
+
+bool btrfs_pinned_by_swapfile(struct btrfs_fs_info *fs_info, void *ptr);
 
 #endif
diff --git a/fs/btrfs/xattr.c b/fs/btrfs/xattr.c
index 95d9aebff2c4..5bb8d8c86311 100644
--- a/fs/btrfs/xattr.c
+++ b/fs/btrfs/xattr.c
@@ -138,7 +138,7 @@ int btrfs_setxattr(struct btrfs_trans_handle *trans, struct inode *inode,
 		 * matches our target xattr, so lets check.
 		 */
 		ret = 0;
-		btrfs_assert_tree_locked(path->nodes[0]);
+		btrfs_assert_tree_write_locked(path->nodes[0]);
 		di = btrfs_match_dir_item_name(fs_info, path, name, name_len);
 		if (!di && !(flags & XATTR_REPLACE)) {
 			ret = -ENOSPC;
@@ -168,9 +168,8 @@ int btrfs_setxattr(struct btrfs_trans_handle *trans, struct inode *inode,
 		const int slot = path->slots[0];
 		struct extent_buffer *leaf = path->nodes[0];
 		const u16 old_data_len = btrfs_dir_data_len(leaf, di);
-		const u32 item_size = btrfs_item_size_nr(leaf, slot);
+		const u32 item_size = btrfs_item_size(leaf, slot);
 		const u32 data_size = sizeof(*di) + name_len + size;
-		struct btrfs_item *item;
 		unsigned long data_ptr;
 		char *ptr;
 
@@ -196,9 +195,8 @@ int btrfs_setxattr(struct btrfs_trans_handle *trans, struct inode *inode,
 			btrfs_extend_item(path, data_size);
 		}
 
-		item = btrfs_item_nr(slot);
 		ptr = btrfs_item_ptr(leaf, slot, char);
-		ptr += btrfs_item_size(leaf, item) - data_size;
+		ptr += btrfs_item_size(leaf, slot) - data_size;
 		di = (struct btrfs_dir_item *)ptr;
 		btrfs_set_dir_data_len(leaf, di, size);
 		data_ptr = ((unsigned long)(di + 1)) + name_len;
@@ -213,9 +211,11 @@ int btrfs_setxattr(struct btrfs_trans_handle *trans, struct inode *inode,
 	}
 out:
 	btrfs_free_path(path);
-	if (!ret)
+	if (!ret) {
 		set_bit(BTRFS_INODE_COPY_EVERYTHING,
 			&BTRFS_I(inode)->runtime_flags);
+		clear_bit(BTRFS_INODE_NO_XATTRS, &BTRFS_I(inode)->runtime_flags);
+	}
 	return ret;
 }
 
@@ -227,11 +227,33 @@ int btrfs_setxattr_trans(struct inode *inode, const char *name,
 {
 	struct btrfs_root *root = BTRFS_I(inode)->root;
 	struct btrfs_trans_handle *trans;
+	const bool start_trans = (current->journal_info == NULL);
 	int ret;
 
-	trans = btrfs_start_transaction(root, 2);
-	if (IS_ERR(trans))
-		return PTR_ERR(trans);
+	if (start_trans) {
+		/*
+		 * 1 unit for inserting/updating/deleting the xattr
+		 * 1 unit for the inode item update
+		 */
+		trans = btrfs_start_transaction(root, 2);
+		if (IS_ERR(trans))
+			return PTR_ERR(trans);
+	} else {
+		/*
+		 * This can happen when smack is enabled and a directory is being
+		 * created. It happens through d_instantiate_new(), which calls
+		 * smack_d_instantiate(), which in turn calls __vfs_setxattr() to
+		 * set the transmute xattr (XATTR_NAME_SMACKTRANSMUTE) on the
+		 * inode. We have already reserved space for the xattr and inode
+		 * update at btrfs_mkdir(), so just use the transaction handle.
+		 * We don't join or start a transaction, as that will reset the
+		 * block_rsv of the handle and trigger a warning for the start
+		 * case.
+		 */
+		ASSERT(strncmp(name, XATTR_SECURITY_PREFIX,
+			       XATTR_SECURITY_PREFIX_LEN) == 0);
+		trans = current->journal_info;
+	}
 
 	ret = btrfs_setxattr(trans, inode, name, value, size, flags);
 	if (ret)
@@ -239,19 +261,23 @@ int btrfs_setxattr_trans(struct inode *inode, const char *name,
 
 	inode_inc_iversion(inode);
 	inode->i_ctime = current_time(inode);
-	ret = btrfs_update_inode(trans, root, inode);
-	BUG_ON(ret);
+	ret = btrfs_update_inode(trans, root, BTRFS_I(inode));
+	if (ret)
+		btrfs_abort_transaction(trans, ret);
 out:
-	btrfs_end_transaction(trans);
+	if (start_trans)
+		btrfs_end_transaction(trans);
 	return ret;
 }
 
 ssize_t btrfs_listxattr(struct dentry *dentry, char *buffer, size_t size)
 {
+	struct btrfs_key found_key;
 	struct btrfs_key key;
 	struct inode *inode = d_inode(dentry);
 	struct btrfs_root *root = BTRFS_I(inode)->root;
 	struct btrfs_path *path;
+	int iter_ret = 0;
 	int ret = 0;
 	size_t total_size = 0, size_left = size;
 
@@ -270,47 +296,26 @@ ssize_t btrfs_listxattr(struct dentry *dentry, char *buffer, size_t size)
 	path->reada = READA_FORWARD;
 
 	/* search for our xattrs */
-	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
-	if (ret < 0)
-		goto err;
-
-	while (1) {
+	btrfs_for_each_slot(root, &key, &found_key, path, iter_ret) {
 		struct extent_buffer *leaf;
 		int slot;
 		struct btrfs_dir_item *di;
-		struct btrfs_key found_key;
 		u32 item_size;
 		u32 cur;
 
 		leaf = path->nodes[0];
 		slot = path->slots[0];
 
-		/* this is where we start walking through the path */
-		if (slot >= btrfs_header_nritems(leaf)) {
-			/*
-			 * if we've reached the last slot in this leaf we need
-			 * to go to the next leaf and reset everything
-			 */
-			ret = btrfs_next_leaf(root, path);
-			if (ret < 0)
-				goto err;
-			else if (ret > 0)
-				break;
-			continue;
-		}
-
-		btrfs_item_key_to_cpu(leaf, &found_key, slot);
-
 		/* check to make sure this item is what we want */
 		if (found_key.objectid != key.objectid)
 			break;
 		if (found_key.type > BTRFS_XATTR_ITEM_KEY)
 			break;
 		if (found_key.type < BTRFS_XATTR_ITEM_KEY)
-			goto next_item;
+			continue;
 
 		di = btrfs_item_ptr(leaf, slot, struct btrfs_dir_item);
-		item_size = btrfs_item_size_nr(leaf, slot);
+		item_size = btrfs_item_size(leaf, slot);
 		cur = 0;
 		while (cur < item_size) {
 			u16 name_len = btrfs_dir_name_len(leaf, di);
@@ -327,8 +332,8 @@ ssize_t btrfs_listxattr(struct dentry *dentry, char *buffer, size_t size)
 				goto next;
 
 			if (!buffer || (name_len + 1) > size_left) {
-				ret = -ERANGE;
-				goto err;
+			        iter_ret = -ERANGE;
+				break;
 			}
 
 			read_extent_buffer(leaf, buffer, name_ptr, name_len);
@@ -340,12 +345,13 @@ next:
 			cur += this_len;
 			di = (struct btrfs_dir_item *)((char *)di + this_len);
 		}
-next_item:
-		path->slots[0]++;
 	}
-	ret = total_size;
 
-err:
+	if (iter_ret < 0)
+		ret = iter_ret;
+	else
+		ret = total_size;
+
 	btrfs_free_path(path);
 
 	return ret;
@@ -360,15 +366,20 @@ static int btrfs_xattr_handler_get(const struct xattr_handler *handler,
 }
 
 static int btrfs_xattr_handler_set(const struct xattr_handler *handler,
+				   struct user_namespace *mnt_userns,
 				   struct dentry *unused, struct inode *inode,
 				   const char *name, const void *buffer,
 				   size_t size, int flags)
 {
+	if (btrfs_root_readonly(BTRFS_I(inode)->root))
+		return -EROFS;
+
 	name = xattr_full_name(handler, name);
 	return btrfs_setxattr_trans(inode, name, buffer, size, flags);
 }
 
 static int btrfs_xattr_handler_set_prop(const struct xattr_handler *handler,
+					struct user_namespace *mnt_userns,
 					struct dentry *unused, struct inode *inode,
 					const char *name, const void *value,
 					size_t size, int flags)
@@ -378,10 +389,13 @@ static int btrfs_xattr_handler_set_prop(const struct xattr_handler *handler,
 	struct btrfs_root *root = BTRFS_I(inode)->root;
 
 	name = xattr_full_name(handler, name);
-	ret = btrfs_validate_prop(name, value, size);
+	ret = btrfs_validate_prop(BTRFS_I(inode), name, value, size);
 	if (ret)
 		return ret;
 
+	if (btrfs_ignore_prop(BTRFS_I(inode), name))
+		return 0;
+
 	trans = btrfs_start_transaction(root, 2);
 	if (IS_ERR(trans))
 		return PTR_ERR(trans);
@@ -390,8 +404,9 @@ static int btrfs_xattr_handler_set_prop(const struct xattr_handler *handler,
 	if (!ret) {
 		inode_inc_iversion(inode);
 		inode->i_ctime = current_time(inode);
-		ret = btrfs_update_inode(trans, root, inode);
-		BUG_ON(ret);
+		ret = btrfs_update_inode(trans, root, BTRFS_I(inode));
+		if (ret)
+			btrfs_abort_transaction(trans, ret);
 	}
 
 	btrfs_end_transaction(trans);
diff --git a/fs/btrfs/zlib.c b/fs/btrfs/zlib.c
index 05615a1099db..b4f44662cda7 100644
--- a/fs/btrfs/zlib.c
+++ b/fs/btrfs/zlib.c
@@ -97,7 +97,7 @@ int zlib_compress_pages(struct list_head *ws, struct address_space *mapping,
 {
 	struct workspace *workspace = list_entry(ws, struct workspace, list);
 	int ret;
-	char *data_in;
+	char *data_in = NULL;
 	char *cpage_out;
 	int nr_pages = 0;
 	struct page *in_page = NULL;
@@ -121,12 +121,12 @@ int zlib_compress_pages(struct list_head *ws, struct address_space *mapping,
 	workspace->strm.total_in = 0;
 	workspace->strm.total_out = 0;
 
-	out_page = alloc_page(GFP_NOFS | __GFP_HIGHMEM);
+	out_page = alloc_page(GFP_NOFS);
 	if (out_page == NULL) {
 		ret = -ENOMEM;
 		goto out;
 	}
-	cpage_out = kmap(out_page);
+	cpage_out = page_address(out_page);
 	pages[0] = out_page;
 	nr_pages = 1;
 
@@ -148,26 +148,26 @@ int zlib_compress_pages(struct list_head *ws, struct address_space *mapping,
 				int i;
 
 				for (i = 0; i < in_buf_pages; i++) {
-					if (in_page) {
-						kunmap(in_page);
+					if (data_in) {
+						kunmap_local(data_in);
 						put_page(in_page);
 					}
 					in_page = find_get_page(mapping,
 								start >> PAGE_SHIFT);
-					data_in = kmap(in_page);
+					data_in = kmap_local_page(in_page);
 					memcpy(workspace->buf + i * PAGE_SIZE,
 					       data_in, PAGE_SIZE);
 					start += PAGE_SIZE;
 				}
 				workspace->strm.next_in = workspace->buf;
 			} else {
-				if (in_page) {
-					kunmap(in_page);
+				if (data_in) {
+					kunmap_local(data_in);
 					put_page(in_page);
 				}
 				in_page = find_get_page(mapping,
 							start >> PAGE_SHIFT);
-				data_in = kmap(in_page);
+				data_in = kmap_local_page(in_page);
 				start += PAGE_SIZE;
 				workspace->strm.next_in = data_in;
 			}
@@ -196,18 +196,16 @@ int zlib_compress_pages(struct list_head *ws, struct address_space *mapping,
 		 * the stream end if required
 		 */
 		if (workspace->strm.avail_out == 0) {
-			kunmap(out_page);
 			if (nr_pages == nr_dest_pages) {
-				out_page = NULL;
 				ret = -E2BIG;
 				goto out;
 			}
-			out_page = alloc_page(GFP_NOFS | __GFP_HIGHMEM);
+			out_page = alloc_page(GFP_NOFS);
 			if (out_page == NULL) {
 				ret = -ENOMEM;
 				goto out;
 			}
-			cpage_out = kmap(out_page);
+			cpage_out = page_address(out_page);
 			pages[nr_pages] = out_page;
 			nr_pages++;
 			workspace->strm.avail_out = PAGE_SIZE;
@@ -234,18 +232,16 @@ int zlib_compress_pages(struct list_head *ws, struct address_space *mapping,
 			goto out;
 		} else if (workspace->strm.avail_out == 0) {
 			/* get another page for the stream end */
-			kunmap(out_page);
 			if (nr_pages == nr_dest_pages) {
-				out_page = NULL;
 				ret = -E2BIG;
 				goto out;
 			}
-			out_page = alloc_page(GFP_NOFS | __GFP_HIGHMEM);
+			out_page = alloc_page(GFP_NOFS);
 			if (out_page == NULL) {
 				ret = -ENOMEM;
 				goto out;
 			}
-			cpage_out = kmap(out_page);
+			cpage_out = page_address(out_page);
 			pages[nr_pages] = out_page;
 			nr_pages++;
 			workspace->strm.avail_out = PAGE_SIZE;
@@ -264,13 +260,11 @@ int zlib_compress_pages(struct list_head *ws, struct address_space *mapping,
 	*total_in = workspace->strm.total_in;
 out:
 	*out_pages = nr_pages;
-	if (out_page)
-		kunmap(out_page);
-
-	if (in_page) {
-		kunmap(in_page);
+	if (data_in) {
+		kunmap_local(data_in);
 		put_page(in_page);
 	}
+
 	return ret;
 }
 
@@ -286,10 +280,8 @@ int zlib_decompress_bio(struct list_head *ws, struct compressed_bio *cb)
 	unsigned long total_pages_in = DIV_ROUND_UP(srclen, PAGE_SIZE);
 	unsigned long buf_start;
 	struct page **pages_in = cb->compressed_pages;
-	u64 disk_start = cb->start;
-	struct bio *orig_bio = cb->orig_bio;
 
-	data_in = kmap(pages_in[page_in_index]);
+	data_in = kmap_local_page(pages_in[page_in_index]);
 	workspace->strm.next_in = data_in;
 	workspace->strm.avail_in = min_t(size_t, srclen, PAGE_SIZE);
 	workspace->strm.total_in = 0;
@@ -311,7 +303,7 @@ int zlib_decompress_bio(struct list_head *ws, struct compressed_bio *cb)
 
 	if (Z_OK != zlib_inflateInit2(&workspace->strm, wbits)) {
 		pr_warn("BTRFS: inflateInit failed\n");
-		kunmap(pages_in[page_in_index]);
+		kunmap_local(data_in);
 		return -EIO;
 	}
 	while (workspace->strm.total_in < srclen) {
@@ -326,9 +318,8 @@ int zlib_decompress_bio(struct list_head *ws, struct compressed_bio *cb)
 		if (buf_start == total_out)
 			break;
 
-		ret2 = btrfs_decompress_buf2page(workspace->buf, buf_start,
-						 total_out, disk_start,
-						 orig_bio);
+		ret2 = btrfs_decompress_buf2page(workspace->buf,
+				total_out - buf_start, cb, buf_start);
 		if (ret2 == 0) {
 			ret = 0;
 			goto done;
@@ -339,17 +330,16 @@ int zlib_decompress_bio(struct list_head *ws, struct compressed_bio *cb)
 
 		if (workspace->strm.avail_in == 0) {
 			unsigned long tmp;
-			kunmap(pages_in[page_in_index]);
+			kunmap_local(data_in);
 			page_in_index++;
 			if (page_in_index >= total_pages_in) {
 				data_in = NULL;
 				break;
 			}
-			data_in = kmap(pages_in[page_in_index]);
+			data_in = kmap_local_page(pages_in[page_in_index]);
 			workspace->strm.next_in = data_in;
 			tmp = srclen - workspace->strm.total_in;
-			workspace->strm.avail_in = min(tmp,
-							   PAGE_SIZE);
+			workspace->strm.avail_in = min(tmp, PAGE_SIZE);
 		}
 	}
 	if (ret != Z_STREAM_END)
@@ -359,9 +349,9 @@ int zlib_decompress_bio(struct list_head *ws, struct compressed_bio *cb)
 done:
 	zlib_inflateEnd(&workspace->strm);
 	if (data_in)
-		kunmap(pages_in[page_in_index]);
+		kunmap_local(data_in);
 	if (!ret)
-		zero_fill_bio(orig_bio);
+		zero_fill_bio(cb->orig_bio);
 	return ret;
 }
 
@@ -375,7 +365,6 @@ int zlib_decompress(struct list_head *ws, unsigned char *data_in,
 	unsigned long bytes_left;
 	unsigned long total_out = 0;
 	unsigned long pg_offset = 0;
-	char *kaddr;
 
 	destlen = min_t(unsigned long, destlen, PAGE_SIZE);
 	bytes_left = destlen;
@@ -432,9 +421,8 @@ int zlib_decompress(struct list_head *ws, unsigned char *data_in,
 			    PAGE_SIZE - (buf_offset % PAGE_SIZE));
 		bytes = min(bytes, bytes_left);
 
-		kaddr = kmap_atomic(dest_page);
-		memcpy(kaddr + pg_offset, workspace->buf + buf_offset, bytes);
-		kunmap_atomic(kaddr);
+		memcpy_to_page(dest_page, pg_offset,
+			       workspace->buf + buf_offset, bytes);
 
 		pg_offset += bytes;
 		bytes_left -= bytes;
@@ -456,9 +444,7 @@ next:
 	 * end of the inline extent (destlen) to the end of the page
 	 */
 	if (pg_offset < destlen) {
-		kaddr = kmap_atomic(dest_page);
-		memset(kaddr + pg_offset, 0, destlen - pg_offset);
-		kunmap_atomic(kaddr);
+		memzero_page(dest_page, pg_offset, destlen - pg_offset);
 	}
 	return ret;
 }
diff --git a/fs/btrfs/zoned.c b/fs/btrfs/zoned.c
new file mode 100644
index 000000000000..1912abf6d020
--- /dev/null
+++ b/fs/btrfs/zoned.c
@@ -0,0 +1,2348 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <linux/bitops.h>
+#include <linux/slab.h>
+#include <linux/blkdev.h>
+#include <linux/sched/mm.h>
+#include <linux/atomic.h>
+#include <linux/vmalloc.h>
+#include "ctree.h"
+#include "volumes.h"
+#include "zoned.h"
+#include "rcu-string.h"
+#include "disk-io.h"
+#include "block-group.h"
+#include "transaction.h"
+#include "dev-replace.h"
+#include "space-info.h"
+
+/* Maximum number of zones to report per blkdev_report_zones() call */
+#define BTRFS_REPORT_NR_ZONES   4096
+/* Invalid allocation pointer value for missing devices */
+#define WP_MISSING_DEV ((u64)-1)
+/* Pseudo write pointer value for conventional zone */
+#define WP_CONVENTIONAL ((u64)-2)
+
+/*
+ * Location of the first zone of superblock logging zone pairs.
+ *
+ * - primary superblock:    0B (zone 0)
+ * - first copy:          512G (zone starting at that offset)
+ * - second copy:           4T (zone starting at that offset)
+ */
+#define BTRFS_SB_LOG_PRIMARY_OFFSET	(0ULL)
+#define BTRFS_SB_LOG_FIRST_OFFSET	(512ULL * SZ_1G)
+#define BTRFS_SB_LOG_SECOND_OFFSET	(4096ULL * SZ_1G)
+
+#define BTRFS_SB_LOG_FIRST_SHIFT	const_ilog2(BTRFS_SB_LOG_FIRST_OFFSET)
+#define BTRFS_SB_LOG_SECOND_SHIFT	const_ilog2(BTRFS_SB_LOG_SECOND_OFFSET)
+
+/* Number of superblock log zones */
+#define BTRFS_NR_SB_LOG_ZONES 2
+
+/*
+ * Minimum of active zones we need:
+ *
+ * - BTRFS_SUPER_MIRROR_MAX zones for superblock mirrors
+ * - 3 zones to ensure at least one zone per SYSTEM, META and DATA block group
+ * - 1 zone for tree-log dedicated block group
+ * - 1 zone for relocation
+ */
+#define BTRFS_MIN_ACTIVE_ZONES		(BTRFS_SUPER_MIRROR_MAX + 5)
+
+/*
+ * Minimum / maximum supported zone size. Currently, SMR disks have a zone
+ * size of 256MiB, and we are expecting ZNS drives to be in the 1-4GiB range.
+ * We do not expect the zone size to become larger than 8GiB or smaller than
+ * 4MiB in the near future.
+ */
+#define BTRFS_MAX_ZONE_SIZE		SZ_8G
+#define BTRFS_MIN_ZONE_SIZE		SZ_4M
+
+#define SUPER_INFO_SECTORS	((u64)BTRFS_SUPER_INFO_SIZE >> SECTOR_SHIFT)
+
+static inline bool sb_zone_is_full(const struct blk_zone *zone)
+{
+	return (zone->cond == BLK_ZONE_COND_FULL) ||
+		(zone->wp + SUPER_INFO_SECTORS > zone->start + zone->capacity);
+}
+
+static int copy_zone_info_cb(struct blk_zone *zone, unsigned int idx, void *data)
+{
+	struct blk_zone *zones = data;
+
+	memcpy(&zones[idx], zone, sizeof(*zone));
+
+	return 0;
+}
+
+static int sb_write_pointer(struct block_device *bdev, struct blk_zone *zones,
+			    u64 *wp_ret)
+{
+	bool empty[BTRFS_NR_SB_LOG_ZONES];
+	bool full[BTRFS_NR_SB_LOG_ZONES];
+	sector_t sector;
+	int i;
+
+	for (i = 0; i < BTRFS_NR_SB_LOG_ZONES; i++) {
+		ASSERT(zones[i].type != BLK_ZONE_TYPE_CONVENTIONAL);
+		empty[i] = (zones[i].cond == BLK_ZONE_COND_EMPTY);
+		full[i] = sb_zone_is_full(&zones[i]);
+	}
+
+	/*
+	 * Possible states of log buffer zones
+	 *
+	 *           Empty[0]  In use[0]  Full[0]
+	 * Empty[1]         *          0        1
+	 * In use[1]        x          x        1
+	 * Full[1]          0          0        C
+	 *
+	 * Log position:
+	 *   *: Special case, no superblock is written
+	 *   0: Use write pointer of zones[0]
+	 *   1: Use write pointer of zones[1]
+	 *   C: Compare super blocks from zones[0] and zones[1], use the latest
+	 *      one determined by generation
+	 *   x: Invalid state
+	 */
+
+	if (empty[0] && empty[1]) {
+		/* Special case to distinguish no superblock to read */
+		*wp_ret = zones[0].start << SECTOR_SHIFT;
+		return -ENOENT;
+	} else if (full[0] && full[1]) {
+		/* Compare two super blocks */
+		struct address_space *mapping = bdev->bd_inode->i_mapping;
+		struct page *page[BTRFS_NR_SB_LOG_ZONES];
+		struct btrfs_super_block *super[BTRFS_NR_SB_LOG_ZONES];
+		int i;
+
+		for (i = 0; i < BTRFS_NR_SB_LOG_ZONES; i++) {
+			u64 bytenr;
+
+			bytenr = ((zones[i].start + zones[i].len)
+				   << SECTOR_SHIFT) - BTRFS_SUPER_INFO_SIZE;
+
+			page[i] = read_cache_page_gfp(mapping,
+					bytenr >> PAGE_SHIFT, GFP_NOFS);
+			if (IS_ERR(page[i])) {
+				if (i == 1)
+					btrfs_release_disk_super(super[0]);
+				return PTR_ERR(page[i]);
+			}
+			super[i] = page_address(page[i]);
+		}
+
+		if (super[0]->generation > super[1]->generation)
+			sector = zones[1].start;
+		else
+			sector = zones[0].start;
+
+		for (i = 0; i < BTRFS_NR_SB_LOG_ZONES; i++)
+			btrfs_release_disk_super(super[i]);
+	} else if (!full[0] && (empty[1] || full[1])) {
+		sector = zones[0].wp;
+	} else if (full[0]) {
+		sector = zones[1].wp;
+	} else {
+		return -EUCLEAN;
+	}
+	*wp_ret = sector << SECTOR_SHIFT;
+	return 0;
+}
+
+/*
+ * Get the first zone number of the superblock mirror
+ */
+static inline u32 sb_zone_number(int shift, int mirror)
+{
+	u64 zone;
+
+	ASSERT(mirror < BTRFS_SUPER_MIRROR_MAX);
+	switch (mirror) {
+	case 0: zone = 0; break;
+	case 1: zone = 1ULL << (BTRFS_SB_LOG_FIRST_SHIFT - shift); break;
+	case 2: zone = 1ULL << (BTRFS_SB_LOG_SECOND_SHIFT - shift); break;
+	}
+
+	ASSERT(zone <= U32_MAX);
+
+	return (u32)zone;
+}
+
+static inline sector_t zone_start_sector(u32 zone_number,
+					 struct block_device *bdev)
+{
+	return (sector_t)zone_number << ilog2(bdev_zone_sectors(bdev));
+}
+
+static inline u64 zone_start_physical(u32 zone_number,
+				      struct btrfs_zoned_device_info *zone_info)
+{
+	return (u64)zone_number << zone_info->zone_size_shift;
+}
+
+/*
+ * Emulate blkdev_report_zones() for a non-zoned device. It slices up the block
+ * device into static sized chunks and fake a conventional zone on each of
+ * them.
+ */
+static int emulate_report_zones(struct btrfs_device *device, u64 pos,
+				struct blk_zone *zones, unsigned int nr_zones)
+{
+	const sector_t zone_sectors = device->fs_info->zone_size >> SECTOR_SHIFT;
+	sector_t bdev_size = bdev_nr_sectors(device->bdev);
+	unsigned int i;
+
+	pos >>= SECTOR_SHIFT;
+	for (i = 0; i < nr_zones; i++) {
+		zones[i].start = i * zone_sectors + pos;
+		zones[i].len = zone_sectors;
+		zones[i].capacity = zone_sectors;
+		zones[i].wp = zones[i].start + zone_sectors;
+		zones[i].type = BLK_ZONE_TYPE_CONVENTIONAL;
+		zones[i].cond = BLK_ZONE_COND_NOT_WP;
+
+		if (zones[i].wp >= bdev_size) {
+			i++;
+			break;
+		}
+	}
+
+	return i;
+}
+
+static int btrfs_get_dev_zones(struct btrfs_device *device, u64 pos,
+			       struct blk_zone *zones, unsigned int *nr_zones)
+{
+	struct btrfs_zoned_device_info *zinfo = device->zone_info;
+	u32 zno;
+	int ret;
+
+	if (!*nr_zones)
+		return 0;
+
+	if (!bdev_is_zoned(device->bdev)) {
+		ret = emulate_report_zones(device, pos, zones, *nr_zones);
+		*nr_zones = ret;
+		return 0;
+	}
+
+	/* Check cache */
+	if (zinfo->zone_cache) {
+		unsigned int i;
+
+		ASSERT(IS_ALIGNED(pos, zinfo->zone_size));
+		zno = pos >> zinfo->zone_size_shift;
+		/*
+		 * We cannot report zones beyond the zone end. So, it is OK to
+		 * cap *nr_zones to at the end.
+		 */
+		*nr_zones = min_t(u32, *nr_zones, zinfo->nr_zones - zno);
+
+		for (i = 0; i < *nr_zones; i++) {
+			struct blk_zone *zone_info;
+
+			zone_info = &zinfo->zone_cache[zno + i];
+			if (!zone_info->len)
+				break;
+		}
+
+		if (i == *nr_zones) {
+			/* Cache hit on all the zones */
+			memcpy(zones, zinfo->zone_cache + zno,
+			       sizeof(*zinfo->zone_cache) * *nr_zones);
+			return 0;
+		}
+	}
+
+	ret = blkdev_report_zones(device->bdev, pos >> SECTOR_SHIFT, *nr_zones,
+				  copy_zone_info_cb, zones);
+	if (ret < 0) {
+		btrfs_err_in_rcu(device->fs_info,
+				 "zoned: failed to read zone %llu on %s (devid %llu)",
+				 pos, rcu_str_deref(device->name),
+				 device->devid);
+		return ret;
+	}
+	*nr_zones = ret;
+	if (!ret)
+		return -EIO;
+
+	/* Populate cache */
+	if (zinfo->zone_cache)
+		memcpy(zinfo->zone_cache + zno, zones,
+		       sizeof(*zinfo->zone_cache) * *nr_zones);
+
+	return 0;
+}
+
+/* The emulated zone size is determined from the size of device extent */
+static int calculate_emulated_zone_size(struct btrfs_fs_info *fs_info)
+{
+	struct btrfs_path *path;
+	struct btrfs_root *root = fs_info->dev_root;
+	struct btrfs_key key;
+	struct extent_buffer *leaf;
+	struct btrfs_dev_extent *dext;
+	int ret = 0;
+
+	key.objectid = 1;
+	key.type = BTRFS_DEV_EXTENT_KEY;
+	key.offset = 0;
+
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+
+	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
+	if (ret < 0)
+		goto out;
+
+	if (path->slots[0] >= btrfs_header_nritems(path->nodes[0])) {
+		ret = btrfs_next_leaf(root, path);
+		if (ret < 0)
+			goto out;
+		/* No dev extents at all? Not good */
+		if (ret > 0) {
+			ret = -EUCLEAN;
+			goto out;
+		}
+	}
+
+	leaf = path->nodes[0];
+	dext = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_dev_extent);
+	fs_info->zone_size = btrfs_dev_extent_length(leaf, dext);
+	ret = 0;
+
+out:
+	btrfs_free_path(path);
+
+	return ret;
+}
+
+int btrfs_get_dev_zone_info_all_devices(struct btrfs_fs_info *fs_info)
+{
+	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
+	struct btrfs_device *device;
+	int ret = 0;
+
+	/* fs_info->zone_size might not set yet. Use the incomapt flag here. */
+	if (!btrfs_fs_incompat(fs_info, ZONED))
+		return 0;
+
+	mutex_lock(&fs_devices->device_list_mutex);
+	list_for_each_entry(device, &fs_devices->devices, dev_list) {
+		/* We can skip reading of zone info for missing devices */
+		if (!device->bdev)
+			continue;
+
+		ret = btrfs_get_dev_zone_info(device, true);
+		if (ret)
+			break;
+	}
+	mutex_unlock(&fs_devices->device_list_mutex);
+
+	return ret;
+}
+
+int btrfs_get_dev_zone_info(struct btrfs_device *device, bool populate_cache)
+{
+	struct btrfs_fs_info *fs_info = device->fs_info;
+	struct btrfs_zoned_device_info *zone_info = NULL;
+	struct block_device *bdev = device->bdev;
+	unsigned int max_active_zones;
+	unsigned int nactive;
+	sector_t nr_sectors;
+	sector_t sector = 0;
+	struct blk_zone *zones = NULL;
+	unsigned int i, nreported = 0, nr_zones;
+	sector_t zone_sectors;
+	char *model, *emulated;
+	int ret;
+
+	/*
+	 * Cannot use btrfs_is_zoned here, since fs_info::zone_size might not
+	 * yet be set.
+	 */
+	if (!btrfs_fs_incompat(fs_info, ZONED))
+		return 0;
+
+	if (device->zone_info)
+		return 0;
+
+	zone_info = kzalloc(sizeof(*zone_info), GFP_KERNEL);
+	if (!zone_info)
+		return -ENOMEM;
+
+	device->zone_info = zone_info;
+
+	if (!bdev_is_zoned(bdev)) {
+		if (!fs_info->zone_size) {
+			ret = calculate_emulated_zone_size(fs_info);
+			if (ret)
+				goto out;
+		}
+
+		ASSERT(fs_info->zone_size);
+		zone_sectors = fs_info->zone_size >> SECTOR_SHIFT;
+	} else {
+		zone_sectors = bdev_zone_sectors(bdev);
+	}
+
+	/* Check if it's power of 2 (see is_power_of_2) */
+	ASSERT(zone_sectors != 0 && (zone_sectors & (zone_sectors - 1)) == 0);
+	zone_info->zone_size = zone_sectors << SECTOR_SHIFT;
+
+	/* We reject devices with a zone size larger than 8GB */
+	if (zone_info->zone_size > BTRFS_MAX_ZONE_SIZE) {
+		btrfs_err_in_rcu(fs_info,
+		"zoned: %s: zone size %llu larger than supported maximum %llu",
+				 rcu_str_deref(device->name),
+				 zone_info->zone_size, BTRFS_MAX_ZONE_SIZE);
+		ret = -EINVAL;
+		goto out;
+	} else if (zone_info->zone_size < BTRFS_MIN_ZONE_SIZE) {
+		btrfs_err_in_rcu(fs_info,
+		"zoned: %s: zone size %llu smaller than supported minimum %u",
+				 rcu_str_deref(device->name),
+				 zone_info->zone_size, BTRFS_MIN_ZONE_SIZE);
+		ret = -EINVAL;
+		goto out;
+	}
+
+	nr_sectors = bdev_nr_sectors(bdev);
+	zone_info->zone_size_shift = ilog2(zone_info->zone_size);
+	zone_info->nr_zones = nr_sectors >> ilog2(zone_sectors);
+	/*
+	 * We limit max_zone_append_size also by max_segments *
+	 * PAGE_SIZE. Technically, we can have multiple pages per segment. But,
+	 * since btrfs adds the pages one by one to a bio, and btrfs cannot
+	 * increase the metadata reservation even if it increases the number of
+	 * extents, it is safe to stick with the limit.
+	 *
+	 * With the zoned emulation, we can have non-zoned device on the zoned
+	 * mode. In this case, we don't have a valid max zone append size. So,
+	 * use max_segments * PAGE_SIZE as the pseudo max_zone_append_size.
+	 */
+	if (bdev_is_zoned(bdev)) {
+		zone_info->max_zone_append_size = min_t(u64,
+			(u64)bdev_max_zone_append_sectors(bdev) << SECTOR_SHIFT,
+			(u64)bdev_max_segments(bdev) << PAGE_SHIFT);
+	} else {
+		zone_info->max_zone_append_size =
+			(u64)bdev_max_segments(bdev) << PAGE_SHIFT;
+	}
+	if (!IS_ALIGNED(nr_sectors, zone_sectors))
+		zone_info->nr_zones++;
+
+	max_active_zones = bdev_max_active_zones(bdev);
+	if (max_active_zones && max_active_zones < BTRFS_MIN_ACTIVE_ZONES) {
+		btrfs_err_in_rcu(fs_info,
+"zoned: %s: max active zones %u is too small, need at least %u active zones",
+				 rcu_str_deref(device->name), max_active_zones,
+				 BTRFS_MIN_ACTIVE_ZONES);
+		ret = -EINVAL;
+		goto out;
+	}
+	zone_info->max_active_zones = max_active_zones;
+
+	zone_info->seq_zones = bitmap_zalloc(zone_info->nr_zones, GFP_KERNEL);
+	if (!zone_info->seq_zones) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	zone_info->empty_zones = bitmap_zalloc(zone_info->nr_zones, GFP_KERNEL);
+	if (!zone_info->empty_zones) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	zone_info->active_zones = bitmap_zalloc(zone_info->nr_zones, GFP_KERNEL);
+	if (!zone_info->active_zones) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	zones = kcalloc(BTRFS_REPORT_NR_ZONES, sizeof(struct blk_zone), GFP_KERNEL);
+	if (!zones) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	/*
+	 * Enable zone cache only for a zoned device. On a non-zoned device, we
+	 * fill the zone info with emulated CONVENTIONAL zones, so no need to
+	 * use the cache.
+	 */
+	if (populate_cache && bdev_is_zoned(device->bdev)) {
+		zone_info->zone_cache = vzalloc(sizeof(struct blk_zone) *
+						zone_info->nr_zones);
+		if (!zone_info->zone_cache) {
+			btrfs_err_in_rcu(device->fs_info,
+				"zoned: failed to allocate zone cache for %s",
+				rcu_str_deref(device->name));
+			ret = -ENOMEM;
+			goto out;
+		}
+	}
+
+	/* Get zones type */
+	nactive = 0;
+	while (sector < nr_sectors) {
+		nr_zones = BTRFS_REPORT_NR_ZONES;
+		ret = btrfs_get_dev_zones(device, sector << SECTOR_SHIFT, zones,
+					  &nr_zones);
+		if (ret)
+			goto out;
+
+		for (i = 0; i < nr_zones; i++) {
+			if (zones[i].type == BLK_ZONE_TYPE_SEQWRITE_REQ)
+				__set_bit(nreported, zone_info->seq_zones);
+			switch (zones[i].cond) {
+			case BLK_ZONE_COND_EMPTY:
+				__set_bit(nreported, zone_info->empty_zones);
+				break;
+			case BLK_ZONE_COND_IMP_OPEN:
+			case BLK_ZONE_COND_EXP_OPEN:
+			case BLK_ZONE_COND_CLOSED:
+				__set_bit(nreported, zone_info->active_zones);
+				nactive++;
+				break;
+			}
+			nreported++;
+		}
+		sector = zones[nr_zones - 1].start + zones[nr_zones - 1].len;
+	}
+
+	if (nreported != zone_info->nr_zones) {
+		btrfs_err_in_rcu(device->fs_info,
+				 "inconsistent number of zones on %s (%u/%u)",
+				 rcu_str_deref(device->name), nreported,
+				 zone_info->nr_zones);
+		ret = -EIO;
+		goto out;
+	}
+
+	if (max_active_zones) {
+		if (nactive > max_active_zones) {
+			btrfs_err_in_rcu(device->fs_info,
+			"zoned: %u active zones on %s exceeds max_active_zones %u",
+					 nactive, rcu_str_deref(device->name),
+					 max_active_zones);
+			ret = -EIO;
+			goto out;
+		}
+		atomic_set(&zone_info->active_zones_left,
+			   max_active_zones - nactive);
+	}
+
+	/* Validate superblock log */
+	nr_zones = BTRFS_NR_SB_LOG_ZONES;
+	for (i = 0; i < BTRFS_SUPER_MIRROR_MAX; i++) {
+		u32 sb_zone;
+		u64 sb_wp;
+		int sb_pos = BTRFS_NR_SB_LOG_ZONES * i;
+
+		sb_zone = sb_zone_number(zone_info->zone_size_shift, i);
+		if (sb_zone + 1 >= zone_info->nr_zones)
+			continue;
+
+		ret = btrfs_get_dev_zones(device,
+					  zone_start_physical(sb_zone, zone_info),
+					  &zone_info->sb_zones[sb_pos],
+					  &nr_zones);
+		if (ret)
+			goto out;
+
+		if (nr_zones != BTRFS_NR_SB_LOG_ZONES) {
+			btrfs_err_in_rcu(device->fs_info,
+	"zoned: failed to read super block log zone info at devid %llu zone %u",
+					 device->devid, sb_zone);
+			ret = -EUCLEAN;
+			goto out;
+		}
+
+		/*
+		 * If zones[0] is conventional, always use the beginning of the
+		 * zone to record superblock. No need to validate in that case.
+		 */
+		if (zone_info->sb_zones[BTRFS_NR_SB_LOG_ZONES * i].type ==
+		    BLK_ZONE_TYPE_CONVENTIONAL)
+			continue;
+
+		ret = sb_write_pointer(device->bdev,
+				       &zone_info->sb_zones[sb_pos], &sb_wp);
+		if (ret != -ENOENT && ret) {
+			btrfs_err_in_rcu(device->fs_info,
+			"zoned: super block log zone corrupted devid %llu zone %u",
+					 device->devid, sb_zone);
+			ret = -EUCLEAN;
+			goto out;
+		}
+	}
+
+
+	kfree(zones);
+
+	switch (bdev_zoned_model(bdev)) {
+	case BLK_ZONED_HM:
+		model = "host-managed zoned";
+		emulated = "";
+		break;
+	case BLK_ZONED_HA:
+		model = "host-aware zoned";
+		emulated = "";
+		break;
+	case BLK_ZONED_NONE:
+		model = "regular";
+		emulated = "emulated ";
+		break;
+	default:
+		/* Just in case */
+		btrfs_err_in_rcu(fs_info, "zoned: unsupported model %d on %s",
+				 bdev_zoned_model(bdev),
+				 rcu_str_deref(device->name));
+		ret = -EOPNOTSUPP;
+		goto out_free_zone_info;
+	}
+
+	btrfs_info_in_rcu(fs_info,
+		"%s block device %s, %u %szones of %llu bytes",
+		model, rcu_str_deref(device->name), zone_info->nr_zones,
+		emulated, zone_info->zone_size);
+
+	return 0;
+
+out:
+	kfree(zones);
+out_free_zone_info:
+	btrfs_destroy_dev_zone_info(device);
+
+	return ret;
+}
+
+void btrfs_destroy_dev_zone_info(struct btrfs_device *device)
+{
+	struct btrfs_zoned_device_info *zone_info = device->zone_info;
+
+	if (!zone_info)
+		return;
+
+	bitmap_free(zone_info->active_zones);
+	bitmap_free(zone_info->seq_zones);
+	bitmap_free(zone_info->empty_zones);
+	vfree(zone_info->zone_cache);
+	kfree(zone_info);
+	device->zone_info = NULL;
+}
+
+struct btrfs_zoned_device_info *btrfs_clone_dev_zone_info(struct btrfs_device *orig_dev)
+{
+	struct btrfs_zoned_device_info *zone_info;
+
+	zone_info = kmemdup(orig_dev->zone_info, sizeof(*zone_info), GFP_KERNEL);
+	if (!zone_info)
+		return NULL;
+
+	zone_info->seq_zones = bitmap_zalloc(zone_info->nr_zones, GFP_KERNEL);
+	if (!zone_info->seq_zones)
+		goto out;
+
+	bitmap_copy(zone_info->seq_zones, orig_dev->zone_info->seq_zones,
+		    zone_info->nr_zones);
+
+	zone_info->empty_zones = bitmap_zalloc(zone_info->nr_zones, GFP_KERNEL);
+	if (!zone_info->empty_zones)
+		goto out;
+
+	bitmap_copy(zone_info->empty_zones, orig_dev->zone_info->empty_zones,
+		    zone_info->nr_zones);
+
+	zone_info->active_zones = bitmap_zalloc(zone_info->nr_zones, GFP_KERNEL);
+	if (!zone_info->active_zones)
+		goto out;
+
+	bitmap_copy(zone_info->active_zones, orig_dev->zone_info->active_zones,
+		    zone_info->nr_zones);
+	zone_info->zone_cache = NULL;
+
+	return zone_info;
+
+out:
+	bitmap_free(zone_info->seq_zones);
+	bitmap_free(zone_info->empty_zones);
+	bitmap_free(zone_info->active_zones);
+	kfree(zone_info);
+	return NULL;
+}
+
+int btrfs_get_dev_zone(struct btrfs_device *device, u64 pos,
+		       struct blk_zone *zone)
+{
+	unsigned int nr_zones = 1;
+	int ret;
+
+	ret = btrfs_get_dev_zones(device, pos, zone, &nr_zones);
+	if (ret != 0 || !nr_zones)
+		return ret ? ret : -EIO;
+
+	return 0;
+}
+
+static int btrfs_check_for_zoned_device(struct btrfs_fs_info *fs_info)
+{
+	struct btrfs_device *device;
+
+	list_for_each_entry(device, &fs_info->fs_devices->devices, dev_list) {
+		if (device->bdev &&
+		    bdev_zoned_model(device->bdev) == BLK_ZONED_HM) {
+			btrfs_err(fs_info,
+				"zoned: mode not enabled but zoned device found: %pg",
+				device->bdev);
+			return -EINVAL;
+		}
+	}
+
+	return 0;
+}
+
+int btrfs_check_zoned_mode(struct btrfs_fs_info *fs_info)
+{
+	struct btrfs_device *device;
+	u64 zone_size = 0;
+	u64 max_zone_append_size = 0;
+	int ret;
+
+	/*
+	 * Host-Managed devices can't be used without the ZONED flag.  With the
+	 * ZONED all devices can be used, using zone emulation if required.
+	 */
+	if (!btrfs_fs_incompat(fs_info, ZONED))
+		return btrfs_check_for_zoned_device(fs_info);
+
+	list_for_each_entry(device, &fs_info->fs_devices->devices, dev_list) {
+		struct btrfs_zoned_device_info *zone_info = device->zone_info;
+
+		if (!device->bdev)
+			continue;
+
+		if (!zone_size) {
+			zone_size = zone_info->zone_size;
+		} else if (zone_info->zone_size != zone_size) {
+			btrfs_err(fs_info,
+		"zoned: unequal block device zone sizes: have %llu found %llu",
+				  zone_info->zone_size, zone_size);
+			return -EINVAL;
+		}
+		if (!max_zone_append_size ||
+		    (zone_info->max_zone_append_size &&
+		     zone_info->max_zone_append_size < max_zone_append_size))
+			max_zone_append_size = zone_info->max_zone_append_size;
+	}
+
+	/*
+	 * stripe_size is always aligned to BTRFS_STRIPE_LEN in
+	 * btrfs_create_chunk(). Since we want stripe_len == zone_size,
+	 * check the alignment here.
+	 */
+	if (!IS_ALIGNED(zone_size, BTRFS_STRIPE_LEN)) {
+		btrfs_err(fs_info,
+			  "zoned: zone size %llu not aligned to stripe %u",
+			  zone_size, BTRFS_STRIPE_LEN);
+		return -EINVAL;
+	}
+
+	if (btrfs_fs_incompat(fs_info, MIXED_GROUPS)) {
+		btrfs_err(fs_info, "zoned: mixed block groups not supported");
+		return -EINVAL;
+	}
+
+	fs_info->zone_size = zone_size;
+	fs_info->max_zone_append_size = ALIGN_DOWN(max_zone_append_size,
+						   fs_info->sectorsize);
+	fs_info->fs_devices->chunk_alloc_policy = BTRFS_CHUNK_ALLOC_ZONED;
+	if (fs_info->max_zone_append_size < fs_info->max_extent_size)
+		fs_info->max_extent_size = fs_info->max_zone_append_size;
+
+	/*
+	 * Check mount options here, because we might change fs_info->zoned
+	 * from fs_info->zone_size.
+	 */
+	ret = btrfs_check_mountopts_zoned(fs_info);
+	if (ret)
+		return ret;
+
+	btrfs_info(fs_info, "zoned mode enabled with zone size %llu", zone_size);
+	return 0;
+}
+
+int btrfs_check_mountopts_zoned(struct btrfs_fs_info *info)
+{
+	if (!btrfs_is_zoned(info))
+		return 0;
+
+	/*
+	 * Space cache writing is not COWed. Disable that to avoid write errors
+	 * in sequential zones.
+	 */
+	if (btrfs_test_opt(info, SPACE_CACHE)) {
+		btrfs_err(info, "zoned: space cache v1 is not supported");
+		return -EINVAL;
+	}
+
+	if (btrfs_test_opt(info, NODATACOW)) {
+		btrfs_err(info, "zoned: NODATACOW not supported");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int sb_log_location(struct block_device *bdev, struct blk_zone *zones,
+			   int rw, u64 *bytenr_ret)
+{
+	u64 wp;
+	int ret;
+
+	if (zones[0].type == BLK_ZONE_TYPE_CONVENTIONAL) {
+		*bytenr_ret = zones[0].start << SECTOR_SHIFT;
+		return 0;
+	}
+
+	ret = sb_write_pointer(bdev, zones, &wp);
+	if (ret != -ENOENT && ret < 0)
+		return ret;
+
+	if (rw == WRITE) {
+		struct blk_zone *reset = NULL;
+
+		if (wp == zones[0].start << SECTOR_SHIFT)
+			reset = &zones[0];
+		else if (wp == zones[1].start << SECTOR_SHIFT)
+			reset = &zones[1];
+
+		if (reset && reset->cond != BLK_ZONE_COND_EMPTY) {
+			ASSERT(sb_zone_is_full(reset));
+
+			ret = blkdev_zone_mgmt(bdev, REQ_OP_ZONE_RESET,
+					       reset->start, reset->len,
+					       GFP_NOFS);
+			if (ret)
+				return ret;
+
+			reset->cond = BLK_ZONE_COND_EMPTY;
+			reset->wp = reset->start;
+		}
+	} else if (ret != -ENOENT) {
+		/*
+		 * For READ, we want the previous one. Move write pointer to
+		 * the end of a zone, if it is at the head of a zone.
+		 */
+		u64 zone_end = 0;
+
+		if (wp == zones[0].start << SECTOR_SHIFT)
+			zone_end = zones[1].start + zones[1].capacity;
+		else if (wp == zones[1].start << SECTOR_SHIFT)
+			zone_end = zones[0].start + zones[0].capacity;
+		if (zone_end)
+			wp = ALIGN_DOWN(zone_end << SECTOR_SHIFT,
+					BTRFS_SUPER_INFO_SIZE);
+
+		wp -= BTRFS_SUPER_INFO_SIZE;
+	}
+
+	*bytenr_ret = wp;
+	return 0;
+
+}
+
+int btrfs_sb_log_location_bdev(struct block_device *bdev, int mirror, int rw,
+			       u64 *bytenr_ret)
+{
+	struct blk_zone zones[BTRFS_NR_SB_LOG_ZONES];
+	sector_t zone_sectors;
+	u32 sb_zone;
+	int ret;
+	u8 zone_sectors_shift;
+	sector_t nr_sectors;
+	u32 nr_zones;
+
+	if (!bdev_is_zoned(bdev)) {
+		*bytenr_ret = btrfs_sb_offset(mirror);
+		return 0;
+	}
+
+	ASSERT(rw == READ || rw == WRITE);
+
+	zone_sectors = bdev_zone_sectors(bdev);
+	if (!is_power_of_2(zone_sectors))
+		return -EINVAL;
+	zone_sectors_shift = ilog2(zone_sectors);
+	nr_sectors = bdev_nr_sectors(bdev);
+	nr_zones = nr_sectors >> zone_sectors_shift;
+
+	sb_zone = sb_zone_number(zone_sectors_shift + SECTOR_SHIFT, mirror);
+	if (sb_zone + 1 >= nr_zones)
+		return -ENOENT;
+
+	ret = blkdev_report_zones(bdev, zone_start_sector(sb_zone, bdev),
+				  BTRFS_NR_SB_LOG_ZONES, copy_zone_info_cb,
+				  zones);
+	if (ret < 0)
+		return ret;
+	if (ret != BTRFS_NR_SB_LOG_ZONES)
+		return -EIO;
+
+	return sb_log_location(bdev, zones, rw, bytenr_ret);
+}
+
+int btrfs_sb_log_location(struct btrfs_device *device, int mirror, int rw,
+			  u64 *bytenr_ret)
+{
+	struct btrfs_zoned_device_info *zinfo = device->zone_info;
+	u32 zone_num;
+
+	/*
+	 * For a zoned filesystem on a non-zoned block device, use the same
+	 * super block locations as regular filesystem. Doing so, the super
+	 * block can always be retrieved and the zoned flag of the volume
+	 * detected from the super block information.
+	 */
+	if (!bdev_is_zoned(device->bdev)) {
+		*bytenr_ret = btrfs_sb_offset(mirror);
+		return 0;
+	}
+
+	zone_num = sb_zone_number(zinfo->zone_size_shift, mirror);
+	if (zone_num + 1 >= zinfo->nr_zones)
+		return -ENOENT;
+
+	return sb_log_location(device->bdev,
+			       &zinfo->sb_zones[BTRFS_NR_SB_LOG_ZONES * mirror],
+			       rw, bytenr_ret);
+}
+
+static inline bool is_sb_log_zone(struct btrfs_zoned_device_info *zinfo,
+				  int mirror)
+{
+	u32 zone_num;
+
+	if (!zinfo)
+		return false;
+
+	zone_num = sb_zone_number(zinfo->zone_size_shift, mirror);
+	if (zone_num + 1 >= zinfo->nr_zones)
+		return false;
+
+	if (!test_bit(zone_num, zinfo->seq_zones))
+		return false;
+
+	return true;
+}
+
+int btrfs_advance_sb_log(struct btrfs_device *device, int mirror)
+{
+	struct btrfs_zoned_device_info *zinfo = device->zone_info;
+	struct blk_zone *zone;
+	int i;
+
+	if (!is_sb_log_zone(zinfo, mirror))
+		return 0;
+
+	zone = &zinfo->sb_zones[BTRFS_NR_SB_LOG_ZONES * mirror];
+	for (i = 0; i < BTRFS_NR_SB_LOG_ZONES; i++) {
+		/* Advance the next zone */
+		if (zone->cond == BLK_ZONE_COND_FULL) {
+			zone++;
+			continue;
+		}
+
+		if (zone->cond == BLK_ZONE_COND_EMPTY)
+			zone->cond = BLK_ZONE_COND_IMP_OPEN;
+
+		zone->wp += SUPER_INFO_SECTORS;
+
+		if (sb_zone_is_full(zone)) {
+			/*
+			 * No room left to write new superblock. Since
+			 * superblock is written with REQ_SYNC, it is safe to
+			 * finish the zone now.
+			 *
+			 * If the write pointer is exactly at the capacity,
+			 * explicit ZONE_FINISH is not necessary.
+			 */
+			if (zone->wp != zone->start + zone->capacity) {
+				int ret;
+
+				ret = blkdev_zone_mgmt(device->bdev,
+						REQ_OP_ZONE_FINISH, zone->start,
+						zone->len, GFP_NOFS);
+				if (ret)
+					return ret;
+			}
+
+			zone->wp = zone->start + zone->len;
+			zone->cond = BLK_ZONE_COND_FULL;
+		}
+		return 0;
+	}
+
+	/* All the zones are FULL. Should not reach here. */
+	ASSERT(0);
+	return -EIO;
+}
+
+int btrfs_reset_sb_log_zones(struct block_device *bdev, int mirror)
+{
+	sector_t zone_sectors;
+	sector_t nr_sectors;
+	u8 zone_sectors_shift;
+	u32 sb_zone;
+	u32 nr_zones;
+
+	zone_sectors = bdev_zone_sectors(bdev);
+	zone_sectors_shift = ilog2(zone_sectors);
+	nr_sectors = bdev_nr_sectors(bdev);
+	nr_zones = nr_sectors >> zone_sectors_shift;
+
+	sb_zone = sb_zone_number(zone_sectors_shift + SECTOR_SHIFT, mirror);
+	if (sb_zone + 1 >= nr_zones)
+		return -ENOENT;
+
+	return blkdev_zone_mgmt(bdev, REQ_OP_ZONE_RESET,
+				zone_start_sector(sb_zone, bdev),
+				zone_sectors * BTRFS_NR_SB_LOG_ZONES, GFP_NOFS);
+}
+
+/**
+ * btrfs_find_allocatable_zones - find allocatable zones within a given region
+ *
+ * @device:	the device to allocate a region on
+ * @hole_start: the position of the hole to allocate the region
+ * @num_bytes:	size of wanted region
+ * @hole_end:	the end of the hole
+ * @return:	position of allocatable zones
+ *
+ * Allocatable region should not contain any superblock locations.
+ */
+u64 btrfs_find_allocatable_zones(struct btrfs_device *device, u64 hole_start,
+				 u64 hole_end, u64 num_bytes)
+{
+	struct btrfs_zoned_device_info *zinfo = device->zone_info;
+	const u8 shift = zinfo->zone_size_shift;
+	u64 nzones = num_bytes >> shift;
+	u64 pos = hole_start;
+	u64 begin, end;
+	bool have_sb;
+	int i;
+
+	ASSERT(IS_ALIGNED(hole_start, zinfo->zone_size));
+	ASSERT(IS_ALIGNED(num_bytes, zinfo->zone_size));
+
+	while (pos < hole_end) {
+		begin = pos >> shift;
+		end = begin + nzones;
+
+		if (end > zinfo->nr_zones)
+			return hole_end;
+
+		/* Check if zones in the region are all empty */
+		if (btrfs_dev_is_sequential(device, pos) &&
+		    find_next_zero_bit(zinfo->empty_zones, end, begin) != end) {
+			pos += zinfo->zone_size;
+			continue;
+		}
+
+		have_sb = false;
+		for (i = 0; i < BTRFS_SUPER_MIRROR_MAX; i++) {
+			u32 sb_zone;
+			u64 sb_pos;
+
+			sb_zone = sb_zone_number(shift, i);
+			if (!(end <= sb_zone ||
+			      sb_zone + BTRFS_NR_SB_LOG_ZONES <= begin)) {
+				have_sb = true;
+				pos = zone_start_physical(
+					sb_zone + BTRFS_NR_SB_LOG_ZONES, zinfo);
+				break;
+			}
+
+			/* We also need to exclude regular superblock positions */
+			sb_pos = btrfs_sb_offset(i);
+			if (!(pos + num_bytes <= sb_pos ||
+			      sb_pos + BTRFS_SUPER_INFO_SIZE <= pos)) {
+				have_sb = true;
+				pos = ALIGN(sb_pos + BTRFS_SUPER_INFO_SIZE,
+					    zinfo->zone_size);
+				break;
+			}
+		}
+		if (!have_sb)
+			break;
+	}
+
+	return pos;
+}
+
+static bool btrfs_dev_set_active_zone(struct btrfs_device *device, u64 pos)
+{
+	struct btrfs_zoned_device_info *zone_info = device->zone_info;
+	unsigned int zno = (pos >> zone_info->zone_size_shift);
+
+	/* We can use any number of zones */
+	if (zone_info->max_active_zones == 0)
+		return true;
+
+	if (!test_bit(zno, zone_info->active_zones)) {
+		/* Active zone left? */
+		if (atomic_dec_if_positive(&zone_info->active_zones_left) < 0)
+			return false;
+		if (test_and_set_bit(zno, zone_info->active_zones)) {
+			/* Someone already set the bit */
+			atomic_inc(&zone_info->active_zones_left);
+		}
+	}
+
+	return true;
+}
+
+static void btrfs_dev_clear_active_zone(struct btrfs_device *device, u64 pos)
+{
+	struct btrfs_zoned_device_info *zone_info = device->zone_info;
+	unsigned int zno = (pos >> zone_info->zone_size_shift);
+
+	/* We can use any number of zones */
+	if (zone_info->max_active_zones == 0)
+		return;
+
+	if (test_and_clear_bit(zno, zone_info->active_zones))
+		atomic_inc(&zone_info->active_zones_left);
+}
+
+int btrfs_reset_device_zone(struct btrfs_device *device, u64 physical,
+			    u64 length, u64 *bytes)
+{
+	int ret;
+
+	*bytes = 0;
+	ret = blkdev_zone_mgmt(device->bdev, REQ_OP_ZONE_RESET,
+			       physical >> SECTOR_SHIFT, length >> SECTOR_SHIFT,
+			       GFP_NOFS);
+	if (ret)
+		return ret;
+
+	*bytes = length;
+	while (length) {
+		btrfs_dev_set_zone_empty(device, physical);
+		btrfs_dev_clear_active_zone(device, physical);
+		physical += device->zone_info->zone_size;
+		length -= device->zone_info->zone_size;
+	}
+
+	return 0;
+}
+
+int btrfs_ensure_empty_zones(struct btrfs_device *device, u64 start, u64 size)
+{
+	struct btrfs_zoned_device_info *zinfo = device->zone_info;
+	const u8 shift = zinfo->zone_size_shift;
+	unsigned long begin = start >> shift;
+	unsigned long end = (start + size) >> shift;
+	u64 pos;
+	int ret;
+
+	ASSERT(IS_ALIGNED(start, zinfo->zone_size));
+	ASSERT(IS_ALIGNED(size, zinfo->zone_size));
+
+	if (end > zinfo->nr_zones)
+		return -ERANGE;
+
+	/* All the zones are conventional */
+	if (find_next_bit(zinfo->seq_zones, begin, end) == end)
+		return 0;
+
+	/* All the zones are sequential and empty */
+	if (find_next_zero_bit(zinfo->seq_zones, begin, end) == end &&
+	    find_next_zero_bit(zinfo->empty_zones, begin, end) == end)
+		return 0;
+
+	for (pos = start; pos < start + size; pos += zinfo->zone_size) {
+		u64 reset_bytes;
+
+		if (!btrfs_dev_is_sequential(device, pos) ||
+		    btrfs_dev_is_empty_zone(device, pos))
+			continue;
+
+		/* Free regions should be empty */
+		btrfs_warn_in_rcu(
+			device->fs_info,
+		"zoned: resetting device %s (devid %llu) zone %llu for allocation",
+			rcu_str_deref(device->name), device->devid, pos >> shift);
+		WARN_ON_ONCE(1);
+
+		ret = btrfs_reset_device_zone(device, pos, zinfo->zone_size,
+					      &reset_bytes);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+/*
+ * Calculate an allocation pointer from the extent allocation information
+ * for a block group consist of conventional zones. It is pointed to the
+ * end of the highest addressed extent in the block group as an allocation
+ * offset.
+ */
+static int calculate_alloc_pointer(struct btrfs_block_group *cache,
+				   u64 *offset_ret, bool new)
+{
+	struct btrfs_fs_info *fs_info = cache->fs_info;
+	struct btrfs_root *root;
+	struct btrfs_path *path;
+	struct btrfs_key key;
+	struct btrfs_key found_key;
+	int ret;
+	u64 length;
+
+	/*
+	 * Avoid  tree lookups for a new block group, there's no use for it.
+	 * It must always be 0.
+	 *
+	 * Also, we have a lock chain of extent buffer lock -> chunk mutex.
+	 * For new a block group, this function is called from
+	 * btrfs_make_block_group() which is already taking the chunk mutex.
+	 * Thus, we cannot call calculate_alloc_pointer() which takes extent
+	 * buffer locks to avoid deadlock.
+	 */
+	if (new) {
+		*offset_ret = 0;
+		return 0;
+	}
+
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+
+	key.objectid = cache->start + cache->length;
+	key.type = 0;
+	key.offset = 0;
+
+	root = btrfs_extent_root(fs_info, key.objectid);
+	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
+	/* We should not find the exact match */
+	if (!ret)
+		ret = -EUCLEAN;
+	if (ret < 0)
+		goto out;
+
+	ret = btrfs_previous_extent_item(root, path, cache->start);
+	if (ret) {
+		if (ret == 1) {
+			ret = 0;
+			*offset_ret = 0;
+		}
+		goto out;
+	}
+
+	btrfs_item_key_to_cpu(path->nodes[0], &found_key, path->slots[0]);
+
+	if (found_key.type == BTRFS_EXTENT_ITEM_KEY)
+		length = found_key.offset;
+	else
+		length = fs_info->nodesize;
+
+	if (!(found_key.objectid >= cache->start &&
+	       found_key.objectid + length <= cache->start + cache->length)) {
+		ret = -EUCLEAN;
+		goto out;
+	}
+	*offset_ret = found_key.objectid + length - cache->start;
+	ret = 0;
+
+out:
+	btrfs_free_path(path);
+	return ret;
+}
+
+int btrfs_load_block_group_zone_info(struct btrfs_block_group *cache, bool new)
+{
+	struct btrfs_fs_info *fs_info = cache->fs_info;
+	struct extent_map_tree *em_tree = &fs_info->mapping_tree;
+	struct extent_map *em;
+	struct map_lookup *map;
+	struct btrfs_device *device;
+	u64 logical = cache->start;
+	u64 length = cache->length;
+	int ret;
+	int i;
+	unsigned int nofs_flag;
+	u64 *alloc_offsets = NULL;
+	u64 *caps = NULL;
+	u64 *physical = NULL;
+	unsigned long *active = NULL;
+	u64 last_alloc = 0;
+	u32 num_sequential = 0, num_conventional = 0;
+
+	if (!btrfs_is_zoned(fs_info))
+		return 0;
+
+	/* Sanity check */
+	if (!IS_ALIGNED(length, fs_info->zone_size)) {
+		btrfs_err(fs_info,
+		"zoned: block group %llu len %llu unaligned to zone size %llu",
+			  logical, length, fs_info->zone_size);
+		return -EIO;
+	}
+
+	/* Get the chunk mapping */
+	read_lock(&em_tree->lock);
+	em = lookup_extent_mapping(em_tree, logical, length);
+	read_unlock(&em_tree->lock);
+
+	if (!em)
+		return -EINVAL;
+
+	map = em->map_lookup;
+
+	cache->physical_map = kmemdup(map, map_lookup_size(map->num_stripes), GFP_NOFS);
+	if (!cache->physical_map) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	alloc_offsets = kcalloc(map->num_stripes, sizeof(*alloc_offsets), GFP_NOFS);
+	if (!alloc_offsets) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	caps = kcalloc(map->num_stripes, sizeof(*caps), GFP_NOFS);
+	if (!caps) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	physical = kcalloc(map->num_stripes, sizeof(*physical), GFP_NOFS);
+	if (!physical) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	active = bitmap_zalloc(map->num_stripes, GFP_NOFS);
+	if (!active) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	for (i = 0; i < map->num_stripes; i++) {
+		bool is_sequential;
+		struct blk_zone zone;
+		struct btrfs_dev_replace *dev_replace = &fs_info->dev_replace;
+		int dev_replace_is_ongoing = 0;
+
+		device = map->stripes[i].dev;
+		physical[i] = map->stripes[i].physical;
+
+		if (device->bdev == NULL) {
+			alloc_offsets[i] = WP_MISSING_DEV;
+			continue;
+		}
+
+		is_sequential = btrfs_dev_is_sequential(device, physical[i]);
+		if (is_sequential)
+			num_sequential++;
+		else
+			num_conventional++;
+
+		/*
+		 * Consider a zone as active if we can allow any number of
+		 * active zones.
+		 */
+		if (!device->zone_info->max_active_zones)
+			__set_bit(i, active);
+
+		if (!is_sequential) {
+			alloc_offsets[i] = WP_CONVENTIONAL;
+			continue;
+		}
+
+		/*
+		 * This zone will be used for allocation, so mark this zone
+		 * non-empty.
+		 */
+		btrfs_dev_clear_zone_empty(device, physical[i]);
+
+		down_read(&dev_replace->rwsem);
+		dev_replace_is_ongoing = btrfs_dev_replace_is_ongoing(dev_replace);
+		if (dev_replace_is_ongoing && dev_replace->tgtdev != NULL)
+			btrfs_dev_clear_zone_empty(dev_replace->tgtdev, physical[i]);
+		up_read(&dev_replace->rwsem);
+
+		/*
+		 * The group is mapped to a sequential zone. Get the zone write
+		 * pointer to determine the allocation offset within the zone.
+		 */
+		WARN_ON(!IS_ALIGNED(physical[i], fs_info->zone_size));
+		nofs_flag = memalloc_nofs_save();
+		ret = btrfs_get_dev_zone(device, physical[i], &zone);
+		memalloc_nofs_restore(nofs_flag);
+		if (ret == -EIO || ret == -EOPNOTSUPP) {
+			ret = 0;
+			alloc_offsets[i] = WP_MISSING_DEV;
+			continue;
+		} else if (ret) {
+			goto out;
+		}
+
+		if (zone.type == BLK_ZONE_TYPE_CONVENTIONAL) {
+			btrfs_err_in_rcu(fs_info,
+	"zoned: unexpected conventional zone %llu on device %s (devid %llu)",
+				zone.start << SECTOR_SHIFT,
+				rcu_str_deref(device->name), device->devid);
+			ret = -EIO;
+			goto out;
+		}
+
+		caps[i] = (zone.capacity << SECTOR_SHIFT);
+
+		switch (zone.cond) {
+		case BLK_ZONE_COND_OFFLINE:
+		case BLK_ZONE_COND_READONLY:
+			btrfs_err(fs_info,
+		"zoned: offline/readonly zone %llu on device %s (devid %llu)",
+				  physical[i] >> device->zone_info->zone_size_shift,
+				  rcu_str_deref(device->name), device->devid);
+			alloc_offsets[i] = WP_MISSING_DEV;
+			break;
+		case BLK_ZONE_COND_EMPTY:
+			alloc_offsets[i] = 0;
+			break;
+		case BLK_ZONE_COND_FULL:
+			alloc_offsets[i] = caps[i];
+			break;
+		default:
+			/* Partially used zone */
+			alloc_offsets[i] =
+					((zone.wp - zone.start) << SECTOR_SHIFT);
+			__set_bit(i, active);
+			break;
+		}
+	}
+
+	if (num_sequential > 0)
+		cache->seq_zone = true;
+
+	if (num_conventional > 0) {
+		/* Zone capacity is always zone size in emulation */
+		cache->zone_capacity = cache->length;
+		ret = calculate_alloc_pointer(cache, &last_alloc, new);
+		if (ret) {
+			btrfs_err(fs_info,
+			"zoned: failed to determine allocation offset of bg %llu",
+				  cache->start);
+			goto out;
+		} else if (map->num_stripes == num_conventional) {
+			cache->alloc_offset = last_alloc;
+			set_bit(BLOCK_GROUP_FLAG_ZONE_IS_ACTIVE, &cache->runtime_flags);
+			goto out;
+		}
+	}
+
+	switch (map->type & BTRFS_BLOCK_GROUP_PROFILE_MASK) {
+	case 0: /* single */
+		if (alloc_offsets[0] == WP_MISSING_DEV) {
+			btrfs_err(fs_info,
+			"zoned: cannot recover write pointer for zone %llu",
+				physical[0]);
+			ret = -EIO;
+			goto out;
+		}
+		cache->alloc_offset = alloc_offsets[0];
+		cache->zone_capacity = caps[0];
+		if (test_bit(0, active))
+			set_bit(BLOCK_GROUP_FLAG_ZONE_IS_ACTIVE, &cache->runtime_flags);
+		break;
+	case BTRFS_BLOCK_GROUP_DUP:
+		if (map->type & BTRFS_BLOCK_GROUP_DATA) {
+			btrfs_err(fs_info, "zoned: profile DUP not yet supported on data bg");
+			ret = -EINVAL;
+			goto out;
+		}
+		if (alloc_offsets[0] == WP_MISSING_DEV) {
+			btrfs_err(fs_info,
+			"zoned: cannot recover write pointer for zone %llu",
+				physical[0]);
+			ret = -EIO;
+			goto out;
+		}
+		if (alloc_offsets[1] == WP_MISSING_DEV) {
+			btrfs_err(fs_info,
+			"zoned: cannot recover write pointer for zone %llu",
+				physical[1]);
+			ret = -EIO;
+			goto out;
+		}
+		if (alloc_offsets[0] != alloc_offsets[1]) {
+			btrfs_err(fs_info,
+			"zoned: write pointer offset mismatch of zones in DUP profile");
+			ret = -EIO;
+			goto out;
+		}
+		if (test_bit(0, active) != test_bit(1, active)) {
+			if (!btrfs_zone_activate(cache)) {
+				ret = -EIO;
+				goto out;
+			}
+		} else {
+			if (test_bit(0, active))
+				set_bit(BLOCK_GROUP_FLAG_ZONE_IS_ACTIVE,
+					&cache->runtime_flags);
+		}
+		cache->alloc_offset = alloc_offsets[0];
+		cache->zone_capacity = min(caps[0], caps[1]);
+		break;
+	case BTRFS_BLOCK_GROUP_RAID1:
+	case BTRFS_BLOCK_GROUP_RAID0:
+	case BTRFS_BLOCK_GROUP_RAID10:
+	case BTRFS_BLOCK_GROUP_RAID5:
+	case BTRFS_BLOCK_GROUP_RAID6:
+		/* non-single profiles are not supported yet */
+	default:
+		btrfs_err(fs_info, "zoned: profile %s not yet supported",
+			  btrfs_bg_type_to_raid_name(map->type));
+		ret = -EINVAL;
+		goto out;
+	}
+
+out:
+	if (cache->alloc_offset > fs_info->zone_size) {
+		btrfs_err(fs_info,
+			"zoned: invalid write pointer %llu in block group %llu",
+			cache->alloc_offset, cache->start);
+		ret = -EIO;
+	}
+
+	if (cache->alloc_offset > cache->zone_capacity) {
+		btrfs_err(fs_info,
+"zoned: invalid write pointer %llu (larger than zone capacity %llu) in block group %llu",
+			  cache->alloc_offset, cache->zone_capacity,
+			  cache->start);
+		ret = -EIO;
+	}
+
+	/* An extent is allocated after the write pointer */
+	if (!ret && num_conventional && last_alloc > cache->alloc_offset) {
+		btrfs_err(fs_info,
+			  "zoned: got wrong write pointer in BG %llu: %llu > %llu",
+			  logical, last_alloc, cache->alloc_offset);
+		ret = -EIO;
+	}
+
+	if (!ret) {
+		cache->meta_write_pointer = cache->alloc_offset + cache->start;
+		if (test_bit(BLOCK_GROUP_FLAG_ZONE_IS_ACTIVE, &cache->runtime_flags)) {
+			btrfs_get_block_group(cache);
+			spin_lock(&fs_info->zone_active_bgs_lock);
+			list_add_tail(&cache->active_bg_list,
+				      &fs_info->zone_active_bgs);
+			spin_unlock(&fs_info->zone_active_bgs_lock);
+		}
+	} else {
+		kfree(cache->physical_map);
+		cache->physical_map = NULL;
+	}
+	bitmap_free(active);
+	kfree(physical);
+	kfree(caps);
+	kfree(alloc_offsets);
+	free_extent_map(em);
+
+	return ret;
+}
+
+void btrfs_calc_zone_unusable(struct btrfs_block_group *cache)
+{
+	u64 unusable, free;
+
+	if (!btrfs_is_zoned(cache->fs_info))
+		return;
+
+	WARN_ON(cache->bytes_super != 0);
+	unusable = (cache->alloc_offset - cache->used) +
+		   (cache->length - cache->zone_capacity);
+	free = cache->zone_capacity - cache->alloc_offset;
+
+	/* We only need ->free_space in ALLOC_SEQ block groups */
+	cache->cached = BTRFS_CACHE_FINISHED;
+	cache->free_space_ctl->free_space = free;
+	cache->zone_unusable = unusable;
+}
+
+void btrfs_redirty_list_add(struct btrfs_transaction *trans,
+			    struct extent_buffer *eb)
+{
+	struct btrfs_fs_info *fs_info = eb->fs_info;
+
+	if (!btrfs_is_zoned(fs_info) ||
+	    btrfs_header_flag(eb, BTRFS_HEADER_FLAG_WRITTEN) ||
+	    !list_empty(&eb->release_list))
+		return;
+
+	set_extent_buffer_dirty(eb);
+	set_extent_bits_nowait(&trans->dirty_pages, eb->start,
+			       eb->start + eb->len - 1, EXTENT_DIRTY);
+	memzero_extent_buffer(eb, 0, eb->len);
+	set_bit(EXTENT_BUFFER_NO_CHECK, &eb->bflags);
+
+	spin_lock(&trans->releasing_ebs_lock);
+	list_add_tail(&eb->release_list, &trans->releasing_ebs);
+	spin_unlock(&trans->releasing_ebs_lock);
+	atomic_inc(&eb->refs);
+}
+
+void btrfs_free_redirty_list(struct btrfs_transaction *trans)
+{
+	spin_lock(&trans->releasing_ebs_lock);
+	while (!list_empty(&trans->releasing_ebs)) {
+		struct extent_buffer *eb;
+
+		eb = list_first_entry(&trans->releasing_ebs,
+				      struct extent_buffer, release_list);
+		list_del_init(&eb->release_list);
+		free_extent_buffer(eb);
+	}
+	spin_unlock(&trans->releasing_ebs_lock);
+}
+
+bool btrfs_use_zone_append(struct btrfs_inode *inode, u64 start)
+{
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
+	struct btrfs_block_group *cache;
+	bool ret = false;
+
+	if (!btrfs_is_zoned(fs_info))
+		return false;
+
+	if (!is_data_inode(&inode->vfs_inode))
+		return false;
+
+	/*
+	 * Using REQ_OP_ZONE_APPNED for relocation can break assumptions on the
+	 * extent layout the relocation code has.
+	 * Furthermore we have set aside own block-group from which only the
+	 * relocation "process" can allocate and make sure only one process at a
+	 * time can add pages to an extent that gets relocated, so it's safe to
+	 * use regular REQ_OP_WRITE for this special case.
+	 */
+	if (btrfs_is_data_reloc_root(inode->root))
+		return false;
+
+	cache = btrfs_lookup_block_group(fs_info, start);
+	ASSERT(cache);
+	if (!cache)
+		return false;
+
+	ret = cache->seq_zone;
+	btrfs_put_block_group(cache);
+
+	return ret;
+}
+
+void btrfs_record_physical_zoned(struct inode *inode, u64 file_offset,
+				 struct bio *bio)
+{
+	struct btrfs_ordered_extent *ordered;
+	const u64 physical = bio->bi_iter.bi_sector << SECTOR_SHIFT;
+
+	if (bio_op(bio) != REQ_OP_ZONE_APPEND)
+		return;
+
+	ordered = btrfs_lookup_ordered_extent(BTRFS_I(inode), file_offset);
+	if (WARN_ON(!ordered))
+		return;
+
+	ordered->physical = physical;
+	ordered->bdev = bio->bi_bdev;
+
+	btrfs_put_ordered_extent(ordered);
+}
+
+void btrfs_rewrite_logical_zoned(struct btrfs_ordered_extent *ordered)
+{
+	struct btrfs_inode *inode = BTRFS_I(ordered->inode);
+	struct btrfs_fs_info *fs_info = inode->root->fs_info;
+	struct extent_map_tree *em_tree;
+	struct extent_map *em;
+	struct btrfs_ordered_sum *sum;
+	u64 orig_logical = ordered->disk_bytenr;
+	u64 *logical = NULL;
+	int nr, stripe_len;
+
+	/* Zoned devices should not have partitions. So, we can assume it is 0 */
+	ASSERT(!bdev_is_partition(ordered->bdev));
+	if (WARN_ON(!ordered->bdev))
+		return;
+
+	if (WARN_ON(btrfs_rmap_block(fs_info, orig_logical, ordered->bdev,
+				     ordered->physical, &logical, &nr,
+				     &stripe_len)))
+		goto out;
+
+	WARN_ON(nr != 1);
+
+	if (orig_logical == *logical)
+		goto out;
+
+	ordered->disk_bytenr = *logical;
+
+	em_tree = &inode->extent_tree;
+	write_lock(&em_tree->lock);
+	em = search_extent_mapping(em_tree, ordered->file_offset,
+				   ordered->num_bytes);
+	em->block_start = *logical;
+	free_extent_map(em);
+	write_unlock(&em_tree->lock);
+
+	list_for_each_entry(sum, &ordered->list, list) {
+		if (*logical < orig_logical)
+			sum->bytenr -= orig_logical - *logical;
+		else
+			sum->bytenr += *logical - orig_logical;
+	}
+
+out:
+	kfree(logical);
+}
+
+bool btrfs_check_meta_write_pointer(struct btrfs_fs_info *fs_info,
+				    struct extent_buffer *eb,
+				    struct btrfs_block_group **cache_ret)
+{
+	struct btrfs_block_group *cache;
+	bool ret = true;
+
+	if (!btrfs_is_zoned(fs_info))
+		return true;
+
+	cache = btrfs_lookup_block_group(fs_info, eb->start);
+	if (!cache)
+		return true;
+
+	if (cache->meta_write_pointer != eb->start) {
+		btrfs_put_block_group(cache);
+		cache = NULL;
+		ret = false;
+	} else {
+		cache->meta_write_pointer = eb->start + eb->len;
+	}
+
+	*cache_ret = cache;
+
+	return ret;
+}
+
+void btrfs_revert_meta_write_pointer(struct btrfs_block_group *cache,
+				     struct extent_buffer *eb)
+{
+	if (!btrfs_is_zoned(eb->fs_info) || !cache)
+		return;
+
+	ASSERT(cache->meta_write_pointer == eb->start + eb->len);
+	cache->meta_write_pointer = eb->start;
+}
+
+int btrfs_zoned_issue_zeroout(struct btrfs_device *device, u64 physical, u64 length)
+{
+	if (!btrfs_dev_is_sequential(device, physical))
+		return -EOPNOTSUPP;
+
+	return blkdev_issue_zeroout(device->bdev, physical >> SECTOR_SHIFT,
+				    length >> SECTOR_SHIFT, GFP_NOFS, 0);
+}
+
+static int read_zone_info(struct btrfs_fs_info *fs_info, u64 logical,
+			  struct blk_zone *zone)
+{
+	struct btrfs_io_context *bioc = NULL;
+	u64 mapped_length = PAGE_SIZE;
+	unsigned int nofs_flag;
+	int nmirrors;
+	int i, ret;
+
+	ret = btrfs_map_sblock(fs_info, BTRFS_MAP_GET_READ_MIRRORS, logical,
+			       &mapped_length, &bioc);
+	if (ret || !bioc || mapped_length < PAGE_SIZE) {
+		ret = -EIO;
+		goto out_put_bioc;
+	}
+
+	if (bioc->map_type & BTRFS_BLOCK_GROUP_RAID56_MASK) {
+		ret = -EINVAL;
+		goto out_put_bioc;
+	}
+
+	nofs_flag = memalloc_nofs_save();
+	nmirrors = (int)bioc->num_stripes;
+	for (i = 0; i < nmirrors; i++) {
+		u64 physical = bioc->stripes[i].physical;
+		struct btrfs_device *dev = bioc->stripes[i].dev;
+
+		/* Missing device */
+		if (!dev->bdev)
+			continue;
+
+		ret = btrfs_get_dev_zone(dev, physical, zone);
+		/* Failing device */
+		if (ret == -EIO || ret == -EOPNOTSUPP)
+			continue;
+		break;
+	}
+	memalloc_nofs_restore(nofs_flag);
+out_put_bioc:
+	btrfs_put_bioc(bioc);
+	return ret;
+}
+
+/*
+ * Synchronize write pointer in a zone at @physical_start on @tgt_dev, by
+ * filling zeros between @physical_pos to a write pointer of dev-replace
+ * source device.
+ */
+int btrfs_sync_zone_write_pointer(struct btrfs_device *tgt_dev, u64 logical,
+				    u64 physical_start, u64 physical_pos)
+{
+	struct btrfs_fs_info *fs_info = tgt_dev->fs_info;
+	struct blk_zone zone;
+	u64 length;
+	u64 wp;
+	int ret;
+
+	if (!btrfs_dev_is_sequential(tgt_dev, physical_pos))
+		return 0;
+
+	ret = read_zone_info(fs_info, logical, &zone);
+	if (ret)
+		return ret;
+
+	wp = physical_start + ((zone.wp - zone.start) << SECTOR_SHIFT);
+
+	if (physical_pos == wp)
+		return 0;
+
+	if (physical_pos > wp)
+		return -EUCLEAN;
+
+	length = wp - physical_pos;
+	return btrfs_zoned_issue_zeroout(tgt_dev, physical_pos, length);
+}
+
+struct btrfs_device *btrfs_zoned_get_device(struct btrfs_fs_info *fs_info,
+					    u64 logical, u64 length)
+{
+	struct btrfs_device *device;
+	struct extent_map *em;
+	struct map_lookup *map;
+
+	em = btrfs_get_chunk_map(fs_info, logical, length);
+	if (IS_ERR(em))
+		return ERR_CAST(em);
+
+	map = em->map_lookup;
+	/* We only support single profile for now */
+	device = map->stripes[0].dev;
+
+	free_extent_map(em);
+
+	return device;
+}
+
+/**
+ * Activate block group and underlying device zones
+ *
+ * @block_group: the block group to activate
+ *
+ * Return: true on success, false otherwise
+ */
+bool btrfs_zone_activate(struct btrfs_block_group *block_group)
+{
+	struct btrfs_fs_info *fs_info = block_group->fs_info;
+	struct btrfs_space_info *space_info = block_group->space_info;
+	struct map_lookup *map;
+	struct btrfs_device *device;
+	u64 physical;
+	bool ret;
+	int i;
+
+	if (!btrfs_is_zoned(block_group->fs_info))
+		return true;
+
+	map = block_group->physical_map;
+
+	spin_lock(&space_info->lock);
+	spin_lock(&block_group->lock);
+	if (test_bit(BLOCK_GROUP_FLAG_ZONE_IS_ACTIVE, &block_group->runtime_flags)) {
+		ret = true;
+		goto out_unlock;
+	}
+
+	/* No space left */
+	if (btrfs_zoned_bg_is_full(block_group)) {
+		ret = false;
+		goto out_unlock;
+	}
+
+	for (i = 0; i < map->num_stripes; i++) {
+		device = map->stripes[i].dev;
+		physical = map->stripes[i].physical;
+
+		if (device->zone_info->max_active_zones == 0)
+			continue;
+
+		if (!btrfs_dev_set_active_zone(device, physical)) {
+			/* Cannot activate the zone */
+			ret = false;
+			goto out_unlock;
+		}
+	}
+
+	/* Successfully activated all the zones */
+	set_bit(BLOCK_GROUP_FLAG_ZONE_IS_ACTIVE, &block_group->runtime_flags);
+	space_info->active_total_bytes += block_group->length;
+	spin_unlock(&block_group->lock);
+	btrfs_try_granting_tickets(fs_info, space_info);
+	spin_unlock(&space_info->lock);
+
+	/* For the active block group list */
+	btrfs_get_block_group(block_group);
+
+	spin_lock(&fs_info->zone_active_bgs_lock);
+	list_add_tail(&block_group->active_bg_list, &fs_info->zone_active_bgs);
+	spin_unlock(&fs_info->zone_active_bgs_lock);
+
+	return true;
+
+out_unlock:
+	spin_unlock(&block_group->lock);
+	spin_unlock(&space_info->lock);
+	return ret;
+}
+
+static void wait_eb_writebacks(struct btrfs_block_group *block_group)
+{
+	struct btrfs_fs_info *fs_info = block_group->fs_info;
+	const u64 end = block_group->start + block_group->length;
+	struct radix_tree_iter iter;
+	struct extent_buffer *eb;
+	void __rcu **slot;
+
+	rcu_read_lock();
+	radix_tree_for_each_slot(slot, &fs_info->buffer_radix, &iter,
+				 block_group->start >> fs_info->sectorsize_bits) {
+		eb = radix_tree_deref_slot(slot);
+		if (!eb)
+			continue;
+		if (radix_tree_deref_retry(eb)) {
+			slot = radix_tree_iter_retry(&iter);
+			continue;
+		}
+
+		if (eb->start < block_group->start)
+			continue;
+		if (eb->start >= end)
+			break;
+
+		slot = radix_tree_iter_resume(slot, &iter);
+		rcu_read_unlock();
+		wait_on_extent_buffer_writeback(eb);
+		rcu_read_lock();
+	}
+	rcu_read_unlock();
+}
+
+static int do_zone_finish(struct btrfs_block_group *block_group, bool fully_written)
+{
+	struct btrfs_fs_info *fs_info = block_group->fs_info;
+	struct map_lookup *map;
+	const bool is_metadata = (block_group->flags &
+			(BTRFS_BLOCK_GROUP_METADATA | BTRFS_BLOCK_GROUP_SYSTEM));
+	int ret = 0;
+	int i;
+
+	spin_lock(&block_group->lock);
+	if (!test_bit(BLOCK_GROUP_FLAG_ZONE_IS_ACTIVE, &block_group->runtime_flags)) {
+		spin_unlock(&block_group->lock);
+		return 0;
+	}
+
+	/* Check if we have unwritten allocated space */
+	if (is_metadata &&
+	    block_group->start + block_group->alloc_offset > block_group->meta_write_pointer) {
+		spin_unlock(&block_group->lock);
+		return -EAGAIN;
+	}
+
+	/*
+	 * If we are sure that the block group is full (= no more room left for
+	 * new allocation) and the IO for the last usable block is completed, we
+	 * don't need to wait for the other IOs. This holds because we ensure
+	 * the sequential IO submissions using the ZONE_APPEND command for data
+	 * and block_group->meta_write_pointer for metadata.
+	 */
+	if (!fully_written) {
+		spin_unlock(&block_group->lock);
+
+		ret = btrfs_inc_block_group_ro(block_group, false);
+		if (ret)
+			return ret;
+
+		/* Ensure all writes in this block group finish */
+		btrfs_wait_block_group_reservations(block_group);
+		/* No need to wait for NOCOW writers. Zoned mode does not allow that */
+		btrfs_wait_ordered_roots(fs_info, U64_MAX, block_group->start,
+					 block_group->length);
+		/* Wait for extent buffers to be written. */
+		if (is_metadata)
+			wait_eb_writebacks(block_group);
+
+		spin_lock(&block_group->lock);
+
+		/*
+		 * Bail out if someone already deactivated the block group, or
+		 * allocated space is left in the block group.
+		 */
+		if (!test_bit(BLOCK_GROUP_FLAG_ZONE_IS_ACTIVE,
+			      &block_group->runtime_flags)) {
+			spin_unlock(&block_group->lock);
+			btrfs_dec_block_group_ro(block_group);
+			return 0;
+		}
+
+		if (block_group->reserved) {
+			spin_unlock(&block_group->lock);
+			btrfs_dec_block_group_ro(block_group);
+			return -EAGAIN;
+		}
+	}
+
+	clear_bit(BLOCK_GROUP_FLAG_ZONE_IS_ACTIVE, &block_group->runtime_flags);
+	block_group->alloc_offset = block_group->zone_capacity;
+	block_group->free_space_ctl->free_space = 0;
+	btrfs_clear_treelog_bg(block_group);
+	btrfs_clear_data_reloc_bg(block_group);
+	spin_unlock(&block_group->lock);
+
+	map = block_group->physical_map;
+	for (i = 0; i < map->num_stripes; i++) {
+		struct btrfs_device *device = map->stripes[i].dev;
+		const u64 physical = map->stripes[i].physical;
+
+		if (device->zone_info->max_active_zones == 0)
+			continue;
+
+		ret = blkdev_zone_mgmt(device->bdev, REQ_OP_ZONE_FINISH,
+				       physical >> SECTOR_SHIFT,
+				       device->zone_info->zone_size >> SECTOR_SHIFT,
+				       GFP_NOFS);
+
+		if (ret)
+			return ret;
+
+		btrfs_dev_clear_active_zone(device, physical);
+	}
+
+	if (!fully_written)
+		btrfs_dec_block_group_ro(block_group);
+
+	spin_lock(&fs_info->zone_active_bgs_lock);
+	ASSERT(!list_empty(&block_group->active_bg_list));
+	list_del_init(&block_group->active_bg_list);
+	spin_unlock(&fs_info->zone_active_bgs_lock);
+
+	/* For active_bg_list */
+	btrfs_put_block_group(block_group);
+
+	clear_and_wake_up_bit(BTRFS_FS_NEED_ZONE_FINISH, &fs_info->flags);
+
+	return 0;
+}
+
+int btrfs_zone_finish(struct btrfs_block_group *block_group)
+{
+	if (!btrfs_is_zoned(block_group->fs_info))
+		return 0;
+
+	return do_zone_finish(block_group, false);
+}
+
+bool btrfs_can_activate_zone(struct btrfs_fs_devices *fs_devices, u64 flags)
+{
+	struct btrfs_fs_info *fs_info = fs_devices->fs_info;
+	struct btrfs_device *device;
+	bool ret = false;
+
+	if (!btrfs_is_zoned(fs_info))
+		return true;
+
+	/* Check if there is a device with active zones left */
+	mutex_lock(&fs_info->chunk_mutex);
+	list_for_each_entry(device, &fs_devices->alloc_list, dev_alloc_list) {
+		struct btrfs_zoned_device_info *zinfo = device->zone_info;
+
+		if (!device->bdev)
+			continue;
+
+		if (!zinfo->max_active_zones ||
+		    atomic_read(&zinfo->active_zones_left)) {
+			ret = true;
+			break;
+		}
+	}
+	mutex_unlock(&fs_info->chunk_mutex);
+
+	if (!ret)
+		set_bit(BTRFS_FS_NEED_ZONE_FINISH, &fs_info->flags);
+
+	return ret;
+}
+
+void btrfs_zone_finish_endio(struct btrfs_fs_info *fs_info, u64 logical, u64 length)
+{
+	struct btrfs_block_group *block_group;
+	u64 min_alloc_bytes;
+
+	if (!btrfs_is_zoned(fs_info))
+		return;
+
+	block_group = btrfs_lookup_block_group(fs_info, logical);
+	ASSERT(block_group);
+
+	/* No MIXED_BG on zoned btrfs. */
+	if (block_group->flags & BTRFS_BLOCK_GROUP_DATA)
+		min_alloc_bytes = fs_info->sectorsize;
+	else
+		min_alloc_bytes = fs_info->nodesize;
+
+	/* Bail out if we can allocate more data from this block group. */
+	if (logical + length + min_alloc_bytes <=
+	    block_group->start + block_group->zone_capacity)
+		goto out;
+
+	do_zone_finish(block_group, true);
+
+out:
+	btrfs_put_block_group(block_group);
+}
+
+static void btrfs_zone_finish_endio_workfn(struct work_struct *work)
+{
+	struct btrfs_block_group *bg =
+		container_of(work, struct btrfs_block_group, zone_finish_work);
+
+	wait_on_extent_buffer_writeback(bg->last_eb);
+	free_extent_buffer(bg->last_eb);
+	btrfs_zone_finish_endio(bg->fs_info, bg->start, bg->length);
+	btrfs_put_block_group(bg);
+}
+
+void btrfs_schedule_zone_finish_bg(struct btrfs_block_group *bg,
+				   struct extent_buffer *eb)
+{
+	if (!bg->seq_zone || eb->start + eb->len * 2 <= bg->start + bg->zone_capacity)
+		return;
+
+	if (WARN_ON(bg->zone_finish_work.func == btrfs_zone_finish_endio_workfn)) {
+		btrfs_err(bg->fs_info, "double scheduling of bg %llu zone finishing",
+			  bg->start);
+		return;
+	}
+
+	/* For the work */
+	btrfs_get_block_group(bg);
+	atomic_inc(&eb->refs);
+	bg->last_eb = eb;
+	INIT_WORK(&bg->zone_finish_work, btrfs_zone_finish_endio_workfn);
+	queue_work(system_unbound_wq, &bg->zone_finish_work);
+}
+
+void btrfs_clear_data_reloc_bg(struct btrfs_block_group *bg)
+{
+	struct btrfs_fs_info *fs_info = bg->fs_info;
+
+	spin_lock(&fs_info->relocation_bg_lock);
+	if (fs_info->data_reloc_bg == bg->start)
+		fs_info->data_reloc_bg = 0;
+	spin_unlock(&fs_info->relocation_bg_lock);
+}
+
+void btrfs_free_zone_cache(struct btrfs_fs_info *fs_info)
+{
+	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
+	struct btrfs_device *device;
+
+	if (!btrfs_is_zoned(fs_info))
+		return;
+
+	mutex_lock(&fs_devices->device_list_mutex);
+	list_for_each_entry(device, &fs_devices->devices, dev_list) {
+		if (device->zone_info) {
+			vfree(device->zone_info->zone_cache);
+			device->zone_info->zone_cache = NULL;
+		}
+	}
+	mutex_unlock(&fs_devices->device_list_mutex);
+}
+
+bool btrfs_zoned_should_reclaim(struct btrfs_fs_info *fs_info)
+{
+	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
+	struct btrfs_device *device;
+	u64 used = 0;
+	u64 total = 0;
+	u64 factor;
+
+	ASSERT(btrfs_is_zoned(fs_info));
+
+	if (fs_info->bg_reclaim_threshold == 0)
+		return false;
+
+	mutex_lock(&fs_devices->device_list_mutex);
+	list_for_each_entry(device, &fs_devices->devices, dev_list) {
+		if (!device->bdev)
+			continue;
+
+		total += device->disk_total_bytes;
+		used += device->bytes_used;
+	}
+	mutex_unlock(&fs_devices->device_list_mutex);
+
+	factor = div64_u64(used * 100, total);
+	return factor >= fs_info->bg_reclaim_threshold;
+}
+
+void btrfs_zoned_release_data_reloc_bg(struct btrfs_fs_info *fs_info, u64 logical,
+				       u64 length)
+{
+	struct btrfs_block_group *block_group;
+
+	if (!btrfs_is_zoned(fs_info))
+		return;
+
+	block_group = btrfs_lookup_block_group(fs_info, logical);
+	/* It should be called on a previous data relocation block group. */
+	ASSERT(block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA));
+
+	spin_lock(&block_group->lock);
+	if (!test_bit(BLOCK_GROUP_FLAG_ZONED_DATA_RELOC, &block_group->runtime_flags))
+		goto out;
+
+	/* All relocation extents are written. */
+	if (block_group->start + block_group->alloc_offset == logical + length) {
+		/* Now, release this block group for further allocations. */
+		clear_bit(BLOCK_GROUP_FLAG_ZONED_DATA_RELOC,
+			  &block_group->runtime_flags);
+	}
+
+out:
+	spin_unlock(&block_group->lock);
+	btrfs_put_block_group(block_group);
+}
+
+int btrfs_zone_finish_one_bg(struct btrfs_fs_info *fs_info)
+{
+	struct btrfs_block_group *block_group;
+	struct btrfs_block_group *min_bg = NULL;
+	u64 min_avail = U64_MAX;
+	int ret;
+
+	spin_lock(&fs_info->zone_active_bgs_lock);
+	list_for_each_entry(block_group, &fs_info->zone_active_bgs,
+			    active_bg_list) {
+		u64 avail;
+
+		spin_lock(&block_group->lock);
+		if (block_group->reserved ||
+		    (block_group->flags & BTRFS_BLOCK_GROUP_SYSTEM)) {
+			spin_unlock(&block_group->lock);
+			continue;
+		}
+
+		avail = block_group->zone_capacity - block_group->alloc_offset;
+		if (min_avail > avail) {
+			if (min_bg)
+				btrfs_put_block_group(min_bg);
+			min_bg = block_group;
+			min_avail = avail;
+			btrfs_get_block_group(min_bg);
+		}
+		spin_unlock(&block_group->lock);
+	}
+	spin_unlock(&fs_info->zone_active_bgs_lock);
+
+	if (!min_bg)
+		return 0;
+
+	ret = btrfs_zone_finish(min_bg);
+	btrfs_put_block_group(min_bg);
+
+	return ret < 0 ? ret : 1;
+}
+
+int btrfs_zoned_activate_one_bg(struct btrfs_fs_info *fs_info,
+				struct btrfs_space_info *space_info,
+				bool do_finish)
+{
+	struct btrfs_block_group *bg;
+	int index;
+
+	if (!btrfs_is_zoned(fs_info) || (space_info->flags & BTRFS_BLOCK_GROUP_DATA))
+		return 0;
+
+	/* No more block groups to activate */
+	if (space_info->active_total_bytes == space_info->total_bytes)
+		return 0;
+
+	for (;;) {
+		int ret;
+		bool need_finish = false;
+
+		down_read(&space_info->groups_sem);
+		for (index = 0; index < BTRFS_NR_RAID_TYPES; index++) {
+			list_for_each_entry(bg, &space_info->block_groups[index],
+					    list) {
+				if (!spin_trylock(&bg->lock))
+					continue;
+				if (btrfs_zoned_bg_is_full(bg) ||
+				    test_bit(BLOCK_GROUP_FLAG_ZONE_IS_ACTIVE,
+					     &bg->runtime_flags)) {
+					spin_unlock(&bg->lock);
+					continue;
+				}
+				spin_unlock(&bg->lock);
+
+				if (btrfs_zone_activate(bg)) {
+					up_read(&space_info->groups_sem);
+					return 1;
+				}
+
+				need_finish = true;
+			}
+		}
+		up_read(&space_info->groups_sem);
+
+		if (!do_finish || !need_finish)
+			break;
+
+		ret = btrfs_zone_finish_one_bg(fs_info);
+		if (ret == 0)
+			break;
+		if (ret < 0)
+			return ret;
+	}
+
+	return 0;
+}
diff --git a/fs/btrfs/zoned.h b/fs/btrfs/zoned.h
new file mode 100644
index 000000000000..8bd16d40b7c6
--- /dev/null
+++ b/fs/btrfs/zoned.h
@@ -0,0 +1,420 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+
+#ifndef BTRFS_ZONED_H
+#define BTRFS_ZONED_H
+
+#include <linux/types.h>
+#include <linux/blkdev.h>
+#include "volumes.h"
+#include "disk-io.h"
+#include "block-group.h"
+#include "btrfs_inode.h"
+
+#define BTRFS_DEFAULT_RECLAIM_THRESH           			(75)
+
+struct btrfs_zoned_device_info {
+	/*
+	 * Number of zones, zone size and types of zones if bdev is a
+	 * zoned block device.
+	 */
+	u64 zone_size;
+	u8  zone_size_shift;
+	u64 max_zone_append_size;
+	u32 nr_zones;
+	unsigned int max_active_zones;
+	atomic_t active_zones_left;
+	unsigned long *seq_zones;
+	unsigned long *empty_zones;
+	unsigned long *active_zones;
+	struct blk_zone *zone_cache;
+	struct blk_zone sb_zones[2 * BTRFS_SUPER_MIRROR_MAX];
+};
+
+#ifdef CONFIG_BLK_DEV_ZONED
+int btrfs_get_dev_zone(struct btrfs_device *device, u64 pos,
+		       struct blk_zone *zone);
+int btrfs_get_dev_zone_info_all_devices(struct btrfs_fs_info *fs_info);
+int btrfs_get_dev_zone_info(struct btrfs_device *device, bool populate_cache);
+void btrfs_destroy_dev_zone_info(struct btrfs_device *device);
+struct btrfs_zoned_device_info *btrfs_clone_dev_zone_info(struct btrfs_device *orig_dev);
+int btrfs_check_zoned_mode(struct btrfs_fs_info *fs_info);
+int btrfs_check_mountopts_zoned(struct btrfs_fs_info *info);
+int btrfs_sb_log_location_bdev(struct block_device *bdev, int mirror, int rw,
+			       u64 *bytenr_ret);
+int btrfs_sb_log_location(struct btrfs_device *device, int mirror, int rw,
+			  u64 *bytenr_ret);
+int btrfs_advance_sb_log(struct btrfs_device *device, int mirror);
+int btrfs_reset_sb_log_zones(struct block_device *bdev, int mirror);
+u64 btrfs_find_allocatable_zones(struct btrfs_device *device, u64 hole_start,
+				 u64 hole_end, u64 num_bytes);
+int btrfs_reset_device_zone(struct btrfs_device *device, u64 physical,
+			    u64 length, u64 *bytes);
+int btrfs_ensure_empty_zones(struct btrfs_device *device, u64 start, u64 size);
+int btrfs_load_block_group_zone_info(struct btrfs_block_group *cache, bool new);
+void btrfs_calc_zone_unusable(struct btrfs_block_group *cache);
+void btrfs_redirty_list_add(struct btrfs_transaction *trans,
+			    struct extent_buffer *eb);
+void btrfs_free_redirty_list(struct btrfs_transaction *trans);
+bool btrfs_use_zone_append(struct btrfs_inode *inode, u64 start);
+void btrfs_record_physical_zoned(struct inode *inode, u64 file_offset,
+				 struct bio *bio);
+void btrfs_rewrite_logical_zoned(struct btrfs_ordered_extent *ordered);
+bool btrfs_check_meta_write_pointer(struct btrfs_fs_info *fs_info,
+				    struct extent_buffer *eb,
+				    struct btrfs_block_group **cache_ret);
+void btrfs_revert_meta_write_pointer(struct btrfs_block_group *cache,
+				     struct extent_buffer *eb);
+int btrfs_zoned_issue_zeroout(struct btrfs_device *device, u64 physical, u64 length);
+int btrfs_sync_zone_write_pointer(struct btrfs_device *tgt_dev, u64 logical,
+				  u64 physical_start, u64 physical_pos);
+struct btrfs_device *btrfs_zoned_get_device(struct btrfs_fs_info *fs_info,
+					    u64 logical, u64 length);
+bool btrfs_zone_activate(struct btrfs_block_group *block_group);
+int btrfs_zone_finish(struct btrfs_block_group *block_group);
+bool btrfs_can_activate_zone(struct btrfs_fs_devices *fs_devices, u64 flags);
+void btrfs_zone_finish_endio(struct btrfs_fs_info *fs_info, u64 logical,
+			     u64 length);
+void btrfs_schedule_zone_finish_bg(struct btrfs_block_group *bg,
+				   struct extent_buffer *eb);
+void btrfs_clear_data_reloc_bg(struct btrfs_block_group *bg);
+void btrfs_free_zone_cache(struct btrfs_fs_info *fs_info);
+bool btrfs_zoned_should_reclaim(struct btrfs_fs_info *fs_info);
+void btrfs_zoned_release_data_reloc_bg(struct btrfs_fs_info *fs_info, u64 logical,
+				       u64 length);
+int btrfs_zone_finish_one_bg(struct btrfs_fs_info *fs_info);
+int btrfs_zoned_activate_one_bg(struct btrfs_fs_info *fs_info,
+				struct btrfs_space_info *space_info, bool do_finish);
+#else /* CONFIG_BLK_DEV_ZONED */
+static inline int btrfs_get_dev_zone(struct btrfs_device *device, u64 pos,
+				     struct blk_zone *zone)
+{
+	return 0;
+}
+
+static inline int btrfs_get_dev_zone_info_all_devices(struct btrfs_fs_info *fs_info)
+{
+	return 0;
+}
+
+static inline int btrfs_get_dev_zone_info(struct btrfs_device *device,
+					  bool populate_cache)
+{
+	return 0;
+}
+
+static inline void btrfs_destroy_dev_zone_info(struct btrfs_device *device) { }
+
+/*
+ * In case the kernel is compiled without CONFIG_BLK_DEV_ZONED we'll never call
+ * into btrfs_clone_dev_zone_info() so it's safe to return NULL here.
+ */
+static inline struct btrfs_zoned_device_info *btrfs_clone_dev_zone_info(
+						 struct btrfs_device *orig_dev)
+{
+	return NULL;
+}
+
+static inline int btrfs_check_zoned_mode(const struct btrfs_fs_info *fs_info)
+{
+	if (!btrfs_is_zoned(fs_info))
+		return 0;
+
+	btrfs_err(fs_info, "zoned block devices support is not enabled");
+	return -EOPNOTSUPP;
+}
+
+static inline int btrfs_check_mountopts_zoned(struct btrfs_fs_info *info)
+{
+	return 0;
+}
+
+static inline int btrfs_sb_log_location_bdev(struct block_device *bdev,
+					     int mirror, int rw, u64 *bytenr_ret)
+{
+	*bytenr_ret = btrfs_sb_offset(mirror);
+	return 0;
+}
+
+static inline int btrfs_sb_log_location(struct btrfs_device *device, int mirror,
+					int rw, u64 *bytenr_ret)
+{
+	*bytenr_ret = btrfs_sb_offset(mirror);
+	return 0;
+}
+
+static inline int btrfs_advance_sb_log(struct btrfs_device *device, int mirror)
+{
+	return 0;
+}
+
+static inline int btrfs_reset_sb_log_zones(struct block_device *bdev, int mirror)
+{
+	return 0;
+}
+
+static inline u64 btrfs_find_allocatable_zones(struct btrfs_device *device,
+					       u64 hole_start, u64 hole_end,
+					       u64 num_bytes)
+{
+	return hole_start;
+}
+
+static inline int btrfs_reset_device_zone(struct btrfs_device *device,
+					  u64 physical, u64 length, u64 *bytes)
+{
+	*bytes = 0;
+	return 0;
+}
+
+static inline int btrfs_ensure_empty_zones(struct btrfs_device *device,
+					   u64 start, u64 size)
+{
+	return 0;
+}
+
+static inline int btrfs_load_block_group_zone_info(
+		struct btrfs_block_group *cache, bool new)
+{
+	return 0;
+}
+
+static inline void btrfs_calc_zone_unusable(struct btrfs_block_group *cache) { }
+
+static inline void btrfs_redirty_list_add(struct btrfs_transaction *trans,
+					  struct extent_buffer *eb) { }
+static inline void btrfs_free_redirty_list(struct btrfs_transaction *trans) { }
+
+static inline bool btrfs_use_zone_append(struct btrfs_inode *inode, u64 start)
+{
+	return false;
+}
+
+static inline void btrfs_record_physical_zoned(struct inode *inode,
+					       u64 file_offset, struct bio *bio)
+{
+}
+
+static inline void btrfs_rewrite_logical_zoned(
+				struct btrfs_ordered_extent *ordered) { }
+
+static inline bool btrfs_check_meta_write_pointer(struct btrfs_fs_info *fs_info,
+			       struct extent_buffer *eb,
+			       struct btrfs_block_group **cache_ret)
+{
+	return true;
+}
+
+static inline void btrfs_revert_meta_write_pointer(
+						struct btrfs_block_group *cache,
+						struct extent_buffer *eb)
+{
+}
+
+static inline int btrfs_zoned_issue_zeroout(struct btrfs_device *device,
+					    u64 physical, u64 length)
+{
+	return -EOPNOTSUPP;
+}
+
+static inline int btrfs_sync_zone_write_pointer(struct btrfs_device *tgt_dev,
+						u64 logical, u64 physical_start,
+						u64 physical_pos)
+{
+	return -EOPNOTSUPP;
+}
+
+static inline struct btrfs_device *btrfs_zoned_get_device(
+						  struct btrfs_fs_info *fs_info,
+						  u64 logical, u64 length)
+{
+	return ERR_PTR(-EOPNOTSUPP);
+}
+
+static inline bool btrfs_zone_activate(struct btrfs_block_group *block_group)
+{
+	return true;
+}
+
+static inline int btrfs_zone_finish(struct btrfs_block_group *block_group)
+{
+	return 0;
+}
+
+static inline bool btrfs_can_activate_zone(struct btrfs_fs_devices *fs_devices,
+					   u64 flags)
+{
+	return true;
+}
+
+static inline void btrfs_zone_finish_endio(struct btrfs_fs_info *fs_info,
+					   u64 logical, u64 length) { }
+
+static inline void btrfs_schedule_zone_finish_bg(struct btrfs_block_group *bg,
+						 struct extent_buffer *eb) { }
+
+static inline void btrfs_clear_data_reloc_bg(struct btrfs_block_group *bg) { }
+
+static inline void btrfs_free_zone_cache(struct btrfs_fs_info *fs_info) { }
+
+static inline bool btrfs_zoned_should_reclaim(struct btrfs_fs_info *fs_info)
+{
+	return false;
+}
+
+static inline void btrfs_zoned_release_data_reloc_bg(struct btrfs_fs_info *fs_info,
+						     u64 logical, u64 length) { }
+
+static inline int btrfs_zone_finish_one_bg(struct btrfs_fs_info *fs_info)
+{
+	return 1;
+}
+
+static inline int btrfs_zoned_activate_one_bg(struct btrfs_fs_info *fs_info,
+					      struct btrfs_space_info *space_info,
+					      bool do_finish)
+{
+	/* Consider all the block groups are active */
+	return 0;
+}
+
+#endif
+
+static inline bool btrfs_dev_is_sequential(struct btrfs_device *device, u64 pos)
+{
+	struct btrfs_zoned_device_info *zone_info = device->zone_info;
+
+	if (!zone_info)
+		return false;
+
+	return test_bit(pos >> zone_info->zone_size_shift, zone_info->seq_zones);
+}
+
+static inline bool btrfs_dev_is_empty_zone(struct btrfs_device *device, u64 pos)
+{
+	struct btrfs_zoned_device_info *zone_info = device->zone_info;
+
+	if (!zone_info)
+		return true;
+
+	return test_bit(pos >> zone_info->zone_size_shift, zone_info->empty_zones);
+}
+
+static inline void btrfs_dev_set_empty_zone_bit(struct btrfs_device *device,
+						u64 pos, bool set)
+{
+	struct btrfs_zoned_device_info *zone_info = device->zone_info;
+	unsigned int zno;
+
+	if (!zone_info)
+		return;
+
+	zno = pos >> zone_info->zone_size_shift;
+	if (set)
+		set_bit(zno, zone_info->empty_zones);
+	else
+		clear_bit(zno, zone_info->empty_zones);
+}
+
+static inline void btrfs_dev_set_zone_empty(struct btrfs_device *device, u64 pos)
+{
+	btrfs_dev_set_empty_zone_bit(device, pos, true);
+}
+
+static inline void btrfs_dev_clear_zone_empty(struct btrfs_device *device, u64 pos)
+{
+	btrfs_dev_set_empty_zone_bit(device, pos, false);
+}
+
+static inline bool btrfs_check_device_zone_type(const struct btrfs_fs_info *fs_info,
+						struct block_device *bdev)
+{
+	if (btrfs_is_zoned(fs_info)) {
+		/*
+		 * We can allow a regular device on a zoned filesystem, because
+		 * we will emulate the zoned capabilities.
+		 */
+		if (!bdev_is_zoned(bdev))
+			return true;
+
+		return fs_info->zone_size ==
+			(bdev_zone_sectors(bdev) << SECTOR_SHIFT);
+	}
+
+	/* Do not allow Host Manged zoned device */
+	return bdev_zoned_model(bdev) != BLK_ZONED_HM;
+}
+
+static inline bool btrfs_check_super_location(struct btrfs_device *device, u64 pos)
+{
+	/*
+	 * On a non-zoned device, any address is OK. On a zoned device,
+	 * non-SEQUENTIAL WRITE REQUIRED zones are capable.
+	 */
+	return device->zone_info == NULL || !btrfs_dev_is_sequential(device, pos);
+}
+
+static inline bool btrfs_can_zone_reset(struct btrfs_device *device,
+					u64 physical, u64 length)
+{
+	u64 zone_size;
+
+	if (!btrfs_dev_is_sequential(device, physical))
+		return false;
+
+	zone_size = device->zone_info->zone_size;
+	if (!IS_ALIGNED(physical, zone_size) || !IS_ALIGNED(length, zone_size))
+		return false;
+
+	return true;
+}
+
+static inline void btrfs_zoned_meta_io_lock(struct btrfs_fs_info *fs_info)
+{
+	if (!btrfs_is_zoned(fs_info))
+		return;
+	mutex_lock(&fs_info->zoned_meta_io_lock);
+}
+
+static inline void btrfs_zoned_meta_io_unlock(struct btrfs_fs_info *fs_info)
+{
+	if (!btrfs_is_zoned(fs_info))
+		return;
+	mutex_unlock(&fs_info->zoned_meta_io_lock);
+}
+
+static inline void btrfs_clear_treelog_bg(struct btrfs_block_group *bg)
+{
+	struct btrfs_fs_info *fs_info = bg->fs_info;
+
+	if (!btrfs_is_zoned(fs_info))
+		return;
+
+	spin_lock(&fs_info->treelog_bg_lock);
+	if (fs_info->treelog_bg == bg->start)
+		fs_info->treelog_bg = 0;
+	spin_unlock(&fs_info->treelog_bg_lock);
+}
+
+static inline void btrfs_zoned_data_reloc_lock(struct btrfs_inode *inode)
+{
+	struct btrfs_root *root = inode->root;
+
+	if (btrfs_is_data_reloc_root(root) && btrfs_is_zoned(root->fs_info))
+		mutex_lock(&root->fs_info->zoned_data_reloc_io_lock);
+}
+
+static inline void btrfs_zoned_data_reloc_unlock(struct btrfs_inode *inode)
+{
+	struct btrfs_root *root = inode->root;
+
+	if (btrfs_is_data_reloc_root(root) && btrfs_is_zoned(root->fs_info))
+		mutex_unlock(&root->fs_info->zoned_data_reloc_io_lock);
+}
+
+static inline bool btrfs_zoned_bg_is_full(const struct btrfs_block_group *bg)
+{
+	ASSERT(btrfs_is_zoned(bg->fs_info));
+	return (bg->alloc_offset == bg->zone_capacity);
+}
+
+#endif
diff --git a/fs/btrfs/zstd.c b/fs/btrfs/zstd.c
index 9a4871636c6c..35a0224d4eb7 100644
--- a/fs/btrfs/zstd.c
+++ b/fs/btrfs/zstd.c
@@ -28,10 +28,10 @@
 /* 307s to avoid pathologically clashing with transaction commit */
 #define ZSTD_BTRFS_RECLAIM_JIFFIES (307 * HZ)
 
-static ZSTD_parameters zstd_get_btrfs_parameters(unsigned int level,
+static zstd_parameters zstd_get_btrfs_parameters(unsigned int level,
 						 size_t src_len)
 {
-	ZSTD_parameters params = ZSTD_getParams(level, src_len, 0);
+	zstd_parameters params = zstd_get_params(level, src_len);
 
 	if (params.cParams.windowLog > ZSTD_BTRFS_MAX_WINDOWLOG)
 		params.cParams.windowLog = ZSTD_BTRFS_MAX_WINDOWLOG;
@@ -48,8 +48,8 @@ struct workspace {
 	unsigned long last_used; /* jiffies */
 	struct list_head list;
 	struct list_head lru_list;
-	ZSTD_inBuffer in_buf;
-	ZSTD_outBuffer out_buf;
+	zstd_in_buffer in_buf;
+	zstd_out_buffer out_buf;
 };
 
 /*
@@ -93,22 +93,26 @@ static inline struct workspace *list_to_workspace(struct list_head *list)
 
 void zstd_free_workspace(struct list_head *ws);
 struct list_head *zstd_alloc_workspace(unsigned int level);
-/*
- * zstd_reclaim_timer_fn - reclaim timer
+
+/**
+ * Timer callback to free unused workspaces.
+ *
  * @t: timer
  *
  * This scans the lru_list and attempts to reclaim any workspace that hasn't
  * been used for ZSTD_BTRFS_RECLAIM_JIFFIES.
+ *
+ * The context is softirq and does not need the _bh locking primitives.
  */
 static void zstd_reclaim_timer_fn(struct timer_list *timer)
 {
 	unsigned long reclaim_threshold = jiffies - ZSTD_BTRFS_RECLAIM_JIFFIES;
 	struct list_head *pos, *next;
 
-	spin_lock_bh(&wsm.lock);
+	spin_lock(&wsm.lock);
 
 	if (list_empty(&wsm.lru_list)) {
-		spin_unlock_bh(&wsm.lock);
+		spin_unlock(&wsm.lock);
 		return;
 	}
 
@@ -137,7 +141,7 @@ static void zstd_reclaim_timer_fn(struct timer_list *timer)
 	if (!list_empty(&wsm.lru_list))
 		mod_timer(&wsm.timer, jiffies + ZSTD_BTRFS_RECLAIM_JIFFIES);
 
-	spin_unlock_bh(&wsm.lock);
+	spin_unlock(&wsm.lock);
 }
 
 /*
@@ -155,12 +159,12 @@ static void zstd_calc_ws_mem_sizes(void)
 	unsigned int level;
 
 	for (level = 1; level <= ZSTD_BTRFS_MAX_LEVEL; level++) {
-		ZSTD_parameters params =
+		zstd_parameters params =
 			zstd_get_btrfs_parameters(level, ZSTD_BTRFS_MAX_INPUT);
 		size_t level_size =
 			max_t(size_t,
-			      ZSTD_CStreamWorkspaceBound(params.cParams),
-			      ZSTD_DStreamWorkspaceBound(ZSTD_BTRFS_MAX_INPUT));
+			      zstd_cstream_workspace_bound(&params.cParams),
+			      zstd_dstream_workspace_bound(ZSTD_BTRFS_MAX_INPUT));
 
 		max_size = max_t(size_t, max_size, level_size);
 		zstd_ws_mem_sizes[level - 1] = max_size;
@@ -371,7 +375,7 @@ int zstd_compress_pages(struct list_head *ws, struct address_space *mapping,
 		unsigned long *total_in, unsigned long *total_out)
 {
 	struct workspace *workspace = list_entry(ws, struct workspace, list);
-	ZSTD_CStream *stream;
+	zstd_cstream *stream;
 	int ret = 0;
 	int nr_pages = 0;
 	struct page *in_page = NULL;  /* The current page to read */
@@ -381,7 +385,7 @@ int zstd_compress_pages(struct list_head *ws, struct address_space *mapping,
 	unsigned long len = *total_out;
 	const unsigned long nr_dest_pages = *out_pages;
 	unsigned long max_out = nr_dest_pages * PAGE_SIZE;
-	ZSTD_parameters params = zstd_get_btrfs_parameters(workspace->req_level,
+	zstd_parameters params = zstd_get_btrfs_parameters(workspace->req_level,
 							   len);
 
 	*out_pages = 0;
@@ -389,40 +393,40 @@ int zstd_compress_pages(struct list_head *ws, struct address_space *mapping,
 	*total_in = 0;
 
 	/* Initialize the stream */
-	stream = ZSTD_initCStream(params, len, workspace->mem,
+	stream = zstd_init_cstream(&params, len, workspace->mem,
 			workspace->size);
 	if (!stream) {
-		pr_warn("BTRFS: ZSTD_initCStream failed\n");
+		pr_warn("BTRFS: zstd_init_cstream failed\n");
 		ret = -EIO;
 		goto out;
 	}
 
 	/* map in the first page of input data */
 	in_page = find_get_page(mapping, start >> PAGE_SHIFT);
-	workspace->in_buf.src = kmap(in_page);
+	workspace->in_buf.src = kmap_local_page(in_page);
 	workspace->in_buf.pos = 0;
 	workspace->in_buf.size = min_t(size_t, len, PAGE_SIZE);
 
 
 	/* Allocate and map in the output buffer */
-	out_page = alloc_page(GFP_NOFS | __GFP_HIGHMEM);
+	out_page = alloc_page(GFP_NOFS);
 	if (out_page == NULL) {
 		ret = -ENOMEM;
 		goto out;
 	}
 	pages[nr_pages++] = out_page;
-	workspace->out_buf.dst = kmap(out_page);
+	workspace->out_buf.dst = page_address(out_page);
 	workspace->out_buf.pos = 0;
 	workspace->out_buf.size = min_t(size_t, max_out, PAGE_SIZE);
 
 	while (1) {
 		size_t ret2;
 
-		ret2 = ZSTD_compressStream(stream, &workspace->out_buf,
+		ret2 = zstd_compress_stream(stream, &workspace->out_buf,
 				&workspace->in_buf);
-		if (ZSTD_isError(ret2)) {
-			pr_debug("BTRFS: ZSTD_compressStream returned %d\n",
-					ZSTD_getErrorCode(ret2));
+		if (zstd_is_error(ret2)) {
+			pr_debug("BTRFS: zstd_compress_stream returned %d\n",
+					zstd_get_error_code(ret2));
 			ret = -EIO;
 			goto out;
 		}
@@ -446,19 +450,17 @@ int zstd_compress_pages(struct list_head *ws, struct address_space *mapping,
 		if (workspace->out_buf.pos == workspace->out_buf.size) {
 			tot_out += PAGE_SIZE;
 			max_out -= PAGE_SIZE;
-			kunmap(out_page);
 			if (nr_pages == nr_dest_pages) {
-				out_page = NULL;
 				ret = -E2BIG;
 				goto out;
 			}
-			out_page = alloc_page(GFP_NOFS | __GFP_HIGHMEM);
+			out_page = alloc_page(GFP_NOFS);
 			if (out_page == NULL) {
 				ret = -ENOMEM;
 				goto out;
 			}
 			pages[nr_pages++] = out_page;
-			workspace->out_buf.dst = kmap(out_page);
+			workspace->out_buf.dst = page_address(out_page);
 			workspace->out_buf.pos = 0;
 			workspace->out_buf.size = min_t(size_t, max_out,
 							PAGE_SIZE);
@@ -473,13 +475,12 @@ int zstd_compress_pages(struct list_head *ws, struct address_space *mapping,
 		/* Check if we need more input */
 		if (workspace->in_buf.pos == workspace->in_buf.size) {
 			tot_in += PAGE_SIZE;
-			kunmap(in_page);
+			kunmap_local(workspace->in_buf.src);
 			put_page(in_page);
-
 			start += PAGE_SIZE;
 			len -= PAGE_SIZE;
 			in_page = find_get_page(mapping, start >> PAGE_SHIFT);
-			workspace->in_buf.src = kmap(in_page);
+			workspace->in_buf.src = kmap_local_page(in_page);
 			workspace->in_buf.pos = 0;
 			workspace->in_buf.size = min_t(size_t, len, PAGE_SIZE);
 		}
@@ -487,10 +488,10 @@ int zstd_compress_pages(struct list_head *ws, struct address_space *mapping,
 	while (1) {
 		size_t ret2;
 
-		ret2 = ZSTD_endStream(stream, &workspace->out_buf);
-		if (ZSTD_isError(ret2)) {
-			pr_debug("BTRFS: ZSTD_endStream returned %d\n",
-					ZSTD_getErrorCode(ret2));
+		ret2 = zstd_end_stream(stream, &workspace->out_buf);
+		if (zstd_is_error(ret2)) {
+			pr_debug("BTRFS: zstd_end_stream returned %d\n",
+					zstd_get_error_code(ret2));
 			ret = -EIO;
 			goto out;
 		}
@@ -506,19 +507,17 @@ int zstd_compress_pages(struct list_head *ws, struct address_space *mapping,
 
 		tot_out += PAGE_SIZE;
 		max_out -= PAGE_SIZE;
-		kunmap(out_page);
 		if (nr_pages == nr_dest_pages) {
-			out_page = NULL;
 			ret = -E2BIG;
 			goto out;
 		}
-		out_page = alloc_page(GFP_NOFS | __GFP_HIGHMEM);
+		out_page = alloc_page(GFP_NOFS);
 		if (out_page == NULL) {
 			ret = -ENOMEM;
 			goto out;
 		}
 		pages[nr_pages++] = out_page;
-		workspace->out_buf.dst = kmap(out_page);
+		workspace->out_buf.dst = page_address(out_page);
 		workspace->out_buf.pos = 0;
 		workspace->out_buf.size = min_t(size_t, max_out, PAGE_SIZE);
 	}
@@ -533,13 +532,10 @@ int zstd_compress_pages(struct list_head *ws, struct address_space *mapping,
 	*total_out = tot_out;
 out:
 	*out_pages = nr_pages;
-	/* Cleanup */
-	if (in_page) {
-		kunmap(in_page);
+	if (workspace->in_buf.src) {
+		kunmap_local(workspace->in_buf.src);
 		put_page(in_page);
 	}
-	if (out_page)
-		kunmap(out_page);
 	return ret;
 }
 
@@ -547,25 +543,23 @@ int zstd_decompress_bio(struct list_head *ws, struct compressed_bio *cb)
 {
 	struct workspace *workspace = list_entry(ws, struct workspace, list);
 	struct page **pages_in = cb->compressed_pages;
-	u64 disk_start = cb->start;
-	struct bio *orig_bio = cb->orig_bio;
 	size_t srclen = cb->compressed_len;
-	ZSTD_DStream *stream;
+	zstd_dstream *stream;
 	int ret = 0;
 	unsigned long page_in_index = 0;
 	unsigned long total_pages_in = DIV_ROUND_UP(srclen, PAGE_SIZE);
 	unsigned long buf_start;
 	unsigned long total_out = 0;
 
-	stream = ZSTD_initDStream(
+	stream = zstd_init_dstream(
 			ZSTD_BTRFS_MAX_INPUT, workspace->mem, workspace->size);
 	if (!stream) {
-		pr_debug("BTRFS: ZSTD_initDStream failed\n");
+		pr_debug("BTRFS: zstd_init_dstream failed\n");
 		ret = -EIO;
 		goto done;
 	}
 
-	workspace->in_buf.src = kmap(pages_in[page_in_index]);
+	workspace->in_buf.src = kmap_local_page(pages_in[page_in_index]);
 	workspace->in_buf.pos = 0;
 	workspace->in_buf.size = min_t(size_t, srclen, PAGE_SIZE);
 
@@ -576,11 +570,11 @@ int zstd_decompress_bio(struct list_head *ws, struct compressed_bio *cb)
 	while (1) {
 		size_t ret2;
 
-		ret2 = ZSTD_decompressStream(stream, &workspace->out_buf,
+		ret2 = zstd_decompress_stream(stream, &workspace->out_buf,
 				&workspace->in_buf);
-		if (ZSTD_isError(ret2)) {
-			pr_debug("BTRFS: ZSTD_decompressStream returned %d\n",
-					ZSTD_getErrorCode(ret2));
+		if (zstd_is_error(ret2)) {
+			pr_debug("BTRFS: zstd_decompress_stream returned %d\n",
+					zstd_get_error_code(ret2));
 			ret = -EIO;
 			goto done;
 		}
@@ -589,7 +583,7 @@ int zstd_decompress_bio(struct list_head *ws, struct compressed_bio *cb)
 		workspace->out_buf.pos = 0;
 
 		ret = btrfs_decompress_buf2page(workspace->out_buf.dst,
-				buf_start, total_out, disk_start, orig_bio);
+				total_out - buf_start, cb, buf_start);
 		if (ret == 0)
 			break;
 
@@ -601,23 +595,24 @@ int zstd_decompress_bio(struct list_head *ws, struct compressed_bio *cb)
 			break;
 
 		if (workspace->in_buf.pos == workspace->in_buf.size) {
-			kunmap(pages_in[page_in_index++]);
+			kunmap_local(workspace->in_buf.src);
+			page_in_index++;
 			if (page_in_index >= total_pages_in) {
 				workspace->in_buf.src = NULL;
 				ret = -EIO;
 				goto done;
 			}
 			srclen -= PAGE_SIZE;
-			workspace->in_buf.src = kmap(pages_in[page_in_index]);
+			workspace->in_buf.src = kmap_local_page(pages_in[page_in_index]);
 			workspace->in_buf.pos = 0;
 			workspace->in_buf.size = min_t(size_t, srclen, PAGE_SIZE);
 		}
 	}
 	ret = 0;
-	zero_fill_bio(orig_bio);
+	zero_fill_bio(cb->orig_bio);
 done:
 	if (workspace->in_buf.src)
-		kunmap(pages_in[page_in_index]);
+		kunmap_local(workspace->in_buf.src);
 	return ret;
 }
 
@@ -626,17 +621,16 @@ int zstd_decompress(struct list_head *ws, unsigned char *data_in,
 		size_t destlen)
 {
 	struct workspace *workspace = list_entry(ws, struct workspace, list);
-	ZSTD_DStream *stream;
+	zstd_dstream *stream;
 	int ret = 0;
 	size_t ret2;
 	unsigned long total_out = 0;
 	unsigned long pg_offset = 0;
-	char *kaddr;
 
-	stream = ZSTD_initDStream(
+	stream = zstd_init_dstream(
 			ZSTD_BTRFS_MAX_INPUT, workspace->mem, workspace->size);
 	if (!stream) {
-		pr_warn("BTRFS: ZSTD_initDStream failed\n");
+		pr_warn("BTRFS: zstd_init_dstream failed\n");
 		ret = -EIO;
 		goto finish;
 	}
@@ -660,15 +654,15 @@ int zstd_decompress(struct list_head *ws, unsigned char *data_in,
 
 		/* Check if the frame is over and we still need more input */
 		if (ret2 == 0) {
-			pr_debug("BTRFS: ZSTD_decompressStream ended early\n");
+			pr_debug("BTRFS: zstd_decompress_stream ended early\n");
 			ret = -EIO;
 			goto finish;
 		}
-		ret2 = ZSTD_decompressStream(stream, &workspace->out_buf,
+		ret2 = zstd_decompress_stream(stream, &workspace->out_buf,
 				&workspace->in_buf);
-		if (ZSTD_isError(ret2)) {
-			pr_debug("BTRFS: ZSTD_decompressStream returned %d\n",
-					ZSTD_getErrorCode(ret2));
+		if (zstd_is_error(ret2)) {
+			pr_debug("BTRFS: zstd_decompress_stream returned %d\n",
+					zstd_get_error_code(ret2));
 			ret = -EIO;
 			goto finish;
 		}
@@ -688,19 +682,15 @@ int zstd_decompress(struct list_head *ws, unsigned char *data_in,
 		bytes = min_t(unsigned long, destlen - pg_offset,
 				workspace->out_buf.size - buf_offset);
 
-		kaddr = kmap_atomic(dest_page);
-		memcpy(kaddr + pg_offset, workspace->out_buf.dst + buf_offset,
-				bytes);
-		kunmap_atomic(kaddr);
+		memcpy_to_page(dest_page, pg_offset,
+			       workspace->out_buf.dst + buf_offset, bytes);
 
 		pg_offset += bytes;
 	}
 	ret = 0;
 finish:
 	if (pg_offset < destlen) {
-		kaddr = kmap_atomic(dest_page);
-		memset(kaddr + pg_offset, 0, destlen - pg_offset);
-		kunmap_atomic(kaddr);
+		memzero_page(dest_page, pg_offset, destlen - pg_offset);
 	}
 	return ret;
 }