diff options
Diffstat (limited to 'fs/btrfs/volumes.c')
-rw-r--r-- | fs/btrfs/volumes.c | 3746 |
1 files changed, 2278 insertions, 1468 deletions
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); +} |