f2fs: support data compression

This patch tries to support compression in f2fs.

- New term named cluster is defined as basic unit of compression, file can
be divided into multiple clusters logically. One cluster includes 4 << n
(n >= 0) logical pages, compression size is also cluster size, each of
cluster can be compressed or not.

- In cluster metadata layout, one special flag is used to indicate cluster
is compressed one or normal one, for compressed cluster, following metadata
maps cluster to [1, 4 << n - 1] physical blocks, in where f2fs stores
data including compress header and compressed data.

- In order to eliminate write amplification during overwrite, F2FS only
support compression on write-once file, data can be compressed only when
all logical blocks in file are valid and cluster compress ratio is lower
than specified threshold.

- To enable compression on regular inode, there are three ways:
* chattr +c file
* chattr +c dir; touch dir/file
* mount w/ -o compress_extension=ext; touch file.ext

Compress metadata layout:
                             [Dnode Structure]
             +-----------------------------------------------+
             | cluster 1 | cluster 2 | ......... | cluster N |
             +-----------------------------------------------+
             .           .                       .           .
       .                       .                .                      .
  .         Compressed Cluster       .        .        Normal Cluster            .
+----------+---------+---------+---------+  +---------+---------+---------+---------+
|compr flag| block 1 | block 2 | block 3 |  | block 1 | block 2 | block 3 | block 4 |
+----------+---------+---------+---------+  +---------+---------+---------+---------+
           .                             .
         .                                           .
       .                                                           .
      +-------------+-------------+----------+----------------------------+
      | data length | data chksum | reserved |      compressed data       |
      +-------------+-------------+----------+----------------------------+

Changelog:

20190326:
- fix error handling of read_end_io().
- remove unneeded comments in f2fs_encrypt_one_page().

20190327:
- fix wrong use of f2fs_cluster_is_full() in f2fs_mpage_readpages().
- don't jump into loop directly to avoid uninitialized variables.
- add TODO tag in error path of f2fs_write_cache_pages().

20190328:
- fix wrong merge condition in f2fs_read_multi_pages().
- check compressed file in f2fs_post_read_required().

20190401
- allow overwrite on non-compressed cluster.
- check cluster meta before writing compressed data.

20190402
- don't preallocate blocks for compressed file.

- add lz4 compress algorithm
- process multiple post read works in one workqueue
  Now f2fs supports processing post read work in multiple workqueue,
  it shows low performance due to schedule overhead of multiple
  workqueue executing orderly.

20190921
- compress: support buffered overwrite
C: compress cluster flag
V: valid block address
N: NEW_ADDR

One cluster contain 4 blocks

 before overwrite   after overwrite

- VVVV		->	CVNN
- CVNN		->	VVVV

- CVNN		->	CVNN
- CVNN		->	CVVV

- CVVV		->	CVNN
- CVVV		->	CVVV

20191029
- add kconfig F2FS_FS_COMPRESSION to isolate compression related
codes, add kconfig F2FS_FS_{LZO,LZ4} to cover backend algorithm.
note that: will remove lzo backend if Jaegeuk agreed that too.
- update codes according to Eric's comments.

20191101
- apply fixes from Jaegeuk

20191113
- apply fixes from Jaegeuk
- split workqueue for fsverity

20191216
- apply fixes from Jaegeuk

20200117
- fix to avoid NULL pointer dereference

[Jaegeuk Kim]
- add tracepoint for f2fs_{,de}compress_pages()
- fix many bugs and add some compression stats
- fix overwrite/mmap bugs
- address 32bit build error, reported by Geert.
- bug fixes when handling errors and i_compressed_blocks

Reported-by: <noreply@ellerman.id.au>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>

1 files changed, 1176 insertions, 0 deletions

diff --git a/fs/f2fs/compress.c b/fs/f2fs/compress.c
new file mode 100644
index 000000000000..d8a64be90a50
--- /dev/null
+++ b/fs/f2fs/compress.c
@@ -0,0 +1,1176 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * f2fs compress support
+ *
+ * Copyright (c) 2019 Chao Yu <chao@kernel.org>
+ */
+
+#include <linux/fs.h>
+#include <linux/f2fs_fs.h>
+#include <linux/writeback.h>
+#include <linux/backing-dev.h>
+#include <linux/lzo.h>
+#include <linux/lz4.h>
+
+#include "f2fs.h"
+#include "node.h"
+#include <trace/events/f2fs.h>
+
+struct f2fs_compress_ops {
+	int (*init_compress_ctx)(struct compress_ctx *cc);
+	void (*destroy_compress_ctx)(struct compress_ctx *cc);
+	int (*compress_pages)(struct compress_ctx *cc);
+	int (*decompress_pages)(struct decompress_io_ctx *dic);
+};
+
+static unsigned int offset_in_cluster(struct compress_ctx *cc, pgoff_t index)
+{
+	return index & (cc->cluster_size - 1);
+}
+
+static pgoff_t cluster_idx(struct compress_ctx *cc, pgoff_t index)
+{
+	return index >> cc->log_cluster_size;
+}
+
+static pgoff_t start_idx_of_cluster(struct compress_ctx *cc)
+{
+	return cc->cluster_idx << cc->log_cluster_size;
+}
+
+bool f2fs_is_compressed_page(struct page *page)
+{
+	if (!PagePrivate(page))
+		return false;
+	if (!page_private(page))
+		return false;
+	if (IS_ATOMIC_WRITTEN_PAGE(page) || IS_DUMMY_WRITTEN_PAGE(page))
+		return false;
+	f2fs_bug_on(F2FS_M_SB(page->mapping),
+		*((u32 *)page_private(page)) != F2FS_COMPRESSED_PAGE_MAGIC);
+	return true;
+}
+
+static void f2fs_set_compressed_page(struct page *page,
+		struct inode *inode, pgoff_t index, void *data, refcount_t *r)
+{
+	SetPagePrivate(page);
+	set_page_private(page, (unsigned long)data);
+
+	/* i_crypto_info and iv index */
+	page->index = index;
+	page->mapping = inode->i_mapping;
+	if (r)
+		refcount_inc(r);
+}
+
+static void f2fs_put_compressed_page(struct page *page)
+{
+	set_page_private(page, (unsigned long)NULL);
+	ClearPagePrivate(page);
+	page->mapping = NULL;
+	unlock_page(page);
+	put_page(page);
+}
+
+static void f2fs_drop_rpages(struct compress_ctx *cc, int len, bool unlock)
+{
+	int i;
+
+	for (i = 0; i < len; i++) {
+		if (!cc->rpages[i])
+			continue;
+		if (unlock)
+			unlock_page(cc->rpages[i]);
+		else
+			put_page(cc->rpages[i]);
+	}
+}
+
+static void f2fs_put_rpages(struct compress_ctx *cc)
+{
+	f2fs_drop_rpages(cc, cc->cluster_size, false);
+}
+
+static void f2fs_unlock_rpages(struct compress_ctx *cc, int len)
+{
+	f2fs_drop_rpages(cc, len, true);
+}
+
+static void f2fs_put_rpages_mapping(struct compress_ctx *cc,
+				struct address_space *mapping,
+				pgoff_t start, int len)
+{
+	int i;
+
+	for (i = 0; i < len; i++) {
+		struct page *page = find_get_page(mapping, start + i);
+
+		put_page(page);
+		put_page(page);
+	}
+}
+
+static void f2fs_put_rpages_wbc(struct compress_ctx *cc,
+		struct writeback_control *wbc, bool redirty, int unlock)
+{
+	unsigned int i;
+
+	for (i = 0; i < cc->cluster_size; i++) {
+		if (!cc->rpages[i])
+			continue;
+		if (redirty)
+			redirty_page_for_writepage(wbc, cc->rpages[i]);
+		f2fs_put_page(cc->rpages[i], unlock);
+	}
+}
+
+struct page *f2fs_compress_control_page(struct page *page)
+{
+	return ((struct compress_io_ctx *)page_private(page))->rpages[0];
+}
+
+int f2fs_init_compress_ctx(struct compress_ctx *cc)
+{
+	struct f2fs_sb_info *sbi = F2FS_I_SB(cc->inode);
+
+	if (cc->nr_rpages)
+		return 0;
+
+	cc->rpages = f2fs_kzalloc(sbi, sizeof(struct page *) <<
+					cc->log_cluster_size, GFP_NOFS);
+	return cc->rpages ? 0 : -ENOMEM;
+}
+
+void f2fs_destroy_compress_ctx(struct compress_ctx *cc)
+{
+	kfree(cc->rpages);
+	cc->rpages = NULL;
+	cc->nr_rpages = 0;
+	cc->nr_cpages = 0;
+	cc->cluster_idx = NULL_CLUSTER;
+}
+
+void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page)
+{
+	unsigned int cluster_ofs;
+
+	if (!f2fs_cluster_can_merge_page(cc, page->index))
+		f2fs_bug_on(F2FS_I_SB(cc->inode), 1);
+
+	cluster_ofs = offset_in_cluster(cc, page->index);
+	cc->rpages[cluster_ofs] = page;
+	cc->nr_rpages++;
+	cc->cluster_idx = cluster_idx(cc, page->index);
+}
+
+#ifdef CONFIG_F2FS_FS_LZO
+static int lzo_init_compress_ctx(struct compress_ctx *cc)
+{
+	cc->private = f2fs_kvmalloc(F2FS_I_SB(cc->inode),
+				LZO1X_MEM_COMPRESS, GFP_NOFS);
+	if (!cc->private)
+		return -ENOMEM;
+
+	cc->clen = lzo1x_worst_compress(PAGE_SIZE << cc->log_cluster_size);
+	return 0;
+}
+
+static void lzo_destroy_compress_ctx(struct compress_ctx *cc)
+{
+	kvfree(cc->private);
+	cc->private = NULL;
+}
+
+static int lzo_compress_pages(struct compress_ctx *cc)
+{
+	int ret;
+
+	ret = lzo1x_1_compress(cc->rbuf, cc->rlen, cc->cbuf->cdata,
+					&cc->clen, cc->private);
+	if (ret != LZO_E_OK) {
+		printk_ratelimited("%sF2FS-fs (%s): lzo compress failed, ret:%d\n",
+				KERN_ERR, F2FS_I_SB(cc->inode)->sb->s_id, ret);
+		return -EIO;
+	}
+	return 0;
+}
+
+static int lzo_decompress_pages(struct decompress_io_ctx *dic)
+{
+	int ret;
+
+	ret = lzo1x_decompress_safe(dic->cbuf->cdata, dic->clen,
+						dic->rbuf, &dic->rlen);
+	if (ret != LZO_E_OK) {
+		printk_ratelimited("%sF2FS-fs (%s): lzo decompress failed, ret:%d\n",
+				KERN_ERR, F2FS_I_SB(dic->inode)->sb->s_id, ret);
+		return -EIO;
+	}
+
+	if (dic->rlen != PAGE_SIZE << dic->log_cluster_size) {
+		printk_ratelimited("%sF2FS-fs (%s): lzo invalid rlen:%zu, "
+					"expected:%lu\n", KERN_ERR,
+					F2FS_I_SB(dic->inode)->sb->s_id,
+					dic->rlen,
+					PAGE_SIZE << dic->log_cluster_size);
+		return -EIO;
+	}
+	return 0;
+}
+
+static const struct f2fs_compress_ops f2fs_lzo_ops = {
+	.init_compress_ctx	= lzo_init_compress_ctx,
+	.destroy_compress_ctx	= lzo_destroy_compress_ctx,
+	.compress_pages		= lzo_compress_pages,
+	.decompress_pages	= lzo_decompress_pages,
+};
+#endif
+
+#ifdef CONFIG_F2FS_FS_LZ4
+static int lz4_init_compress_ctx(struct compress_ctx *cc)
+{
+	cc->private = f2fs_kvmalloc(F2FS_I_SB(cc->inode),
+				LZ4_MEM_COMPRESS, GFP_NOFS);
+	if (!cc->private)
+		return -ENOMEM;
+
+	cc->clen = LZ4_compressBound(PAGE_SIZE << cc->log_cluster_size);
+	return 0;
+}
+
+static void lz4_destroy_compress_ctx(struct compress_ctx *cc)
+{
+	kvfree(cc->private);
+	cc->private = NULL;
+}
+
+static int lz4_compress_pages(struct compress_ctx *cc)
+{
+	int len;
+
+	len = LZ4_compress_default(cc->rbuf, cc->cbuf->cdata, cc->rlen,
+						cc->clen, cc->private);
+	if (!len) {
+		printk_ratelimited("%sF2FS-fs (%s): lz4 compress failed\n",
+				KERN_ERR, F2FS_I_SB(cc->inode)->sb->s_id);
+		return -EIO;
+	}
+	cc->clen = len;
+	return 0;
+}
+
+static int lz4_decompress_pages(struct decompress_io_ctx *dic)
+{
+	int ret;
+
+	ret = LZ4_decompress_safe(dic->cbuf->cdata, dic->rbuf,
+						dic->clen, dic->rlen);
+	if (ret < 0) {
+		printk_ratelimited("%sF2FS-fs (%s): lz4 decompress failed, ret:%d\n",
+				KERN_ERR, F2FS_I_SB(dic->inode)->sb->s_id, ret);
+		return -EIO;
+	}
+
+	if (ret != PAGE_SIZE << dic->log_cluster_size) {
+		printk_ratelimited("%sF2FS-fs (%s): lz4 invalid rlen:%zu, "
+					"expected:%lu\n", KERN_ERR,
+					F2FS_I_SB(dic->inode)->sb->s_id,
+					dic->rlen,
+					PAGE_SIZE << dic->log_cluster_size);
+		return -EIO;
+	}
+	return 0;
+}
+
+static const struct f2fs_compress_ops f2fs_lz4_ops = {
+	.init_compress_ctx	= lz4_init_compress_ctx,
+	.destroy_compress_ctx	= lz4_destroy_compress_ctx,
+	.compress_pages		= lz4_compress_pages,
+	.decompress_pages	= lz4_decompress_pages,
+};
+#endif
+
+static const struct f2fs_compress_ops *f2fs_cops[COMPRESS_MAX] = {
+#ifdef CONFIG_F2FS_FS_LZO
+	&f2fs_lzo_ops,
+#else
+	NULL,
+#endif
+#ifdef CONFIG_F2FS_FS_LZ4
+	&f2fs_lz4_ops,
+#else
+	NULL,
+#endif
+};
+
+bool f2fs_is_compress_backend_ready(struct inode *inode)
+{
+	if (!f2fs_compressed_file(inode))
+		return true;
+	return f2fs_cops[F2FS_I(inode)->i_compress_algorithm];
+}
+
+static struct page *f2fs_grab_page(void)
+{
+	struct page *page;
+
+	page = alloc_page(GFP_NOFS);
+	if (!page)
+		return NULL;
+	lock_page(page);
+	return page;
+}
+
+static int f2fs_compress_pages(struct compress_ctx *cc)
+{
+	struct f2fs_sb_info *sbi = F2FS_I_SB(cc->inode);
+	struct f2fs_inode_info *fi = F2FS_I(cc->inode);
+	const struct f2fs_compress_ops *cops =
+				f2fs_cops[fi->i_compress_algorithm];
+	unsigned int max_len, nr_cpages;
+	int i, ret;
+
+	trace_f2fs_compress_pages_start(cc->inode, cc->cluster_idx,
+				cc->cluster_size, fi->i_compress_algorithm);
+
+	ret = cops->init_compress_ctx(cc);
+	if (ret)
+		goto out;
+
+	max_len = COMPRESS_HEADER_SIZE + cc->clen;
+	cc->nr_cpages = DIV_ROUND_UP(max_len, PAGE_SIZE);
+
+	cc->cpages = f2fs_kzalloc(sbi, sizeof(struct page *) *
+					cc->nr_cpages, GFP_NOFS);
+	if (!cc->cpages) {
+		ret = -ENOMEM;
+		goto destroy_compress_ctx;
+	}
+
+	for (i = 0; i < cc->nr_cpages; i++) {
+		cc->cpages[i] = f2fs_grab_page();
+		if (!cc->cpages[i]) {
+			ret = -ENOMEM;
+			goto out_free_cpages;
+		}
+	}
+
+	cc->rbuf = vmap(cc->rpages, cc->cluster_size, VM_MAP, PAGE_KERNEL_RO);
+	if (!cc->rbuf) {
+		ret = -ENOMEM;
+		goto out_free_cpages;
+	}
+
+	cc->cbuf = vmap(cc->cpages, cc->nr_cpages, VM_MAP, PAGE_KERNEL);
+	if (!cc->cbuf) {
+		ret = -ENOMEM;
+		goto out_vunmap_rbuf;
+	}
+
+	ret = cops->compress_pages(cc);
+	if (ret)
+		goto out_vunmap_cbuf;
+
+	max_len = PAGE_SIZE * (cc->cluster_size - 1) - COMPRESS_HEADER_SIZE;
+
+	if (cc->clen > max_len) {
+		ret = -EAGAIN;
+		goto out_vunmap_cbuf;
+	}
+
+	cc->cbuf->clen = cpu_to_le32(cc->clen);
+	cc->cbuf->chksum = cpu_to_le32(0);
+
+	for (i = 0; i < COMPRESS_DATA_RESERVED_SIZE; i++)
+		cc->cbuf->reserved[i] = cpu_to_le32(0);
+
+	vunmap(cc->cbuf);
+	vunmap(cc->rbuf);
+
+	nr_cpages = DIV_ROUND_UP(cc->clen + COMPRESS_HEADER_SIZE, PAGE_SIZE);
+
+	for (i = nr_cpages; i < cc->nr_cpages; i++) {
+		f2fs_put_compressed_page(cc->cpages[i]);
+		cc->cpages[i] = NULL;
+	}
+
+	cc->nr_cpages = nr_cpages;
+
+	trace_f2fs_compress_pages_end(cc->inode, cc->cluster_idx,
+							cc->clen, ret);
+	return 0;
+
+out_vunmap_cbuf:
+	vunmap(cc->cbuf);
+out_vunmap_rbuf:
+	vunmap(cc->rbuf);
+out_free_cpages:
+	for (i = 0; i < cc->nr_cpages; i++) {
+		if (cc->cpages[i])
+			f2fs_put_compressed_page(cc->cpages[i]);
+	}
+	kfree(cc->cpages);
+	cc->cpages = NULL;
+destroy_compress_ctx:
+	cops->destroy_compress_ctx(cc);
+out:
+	trace_f2fs_compress_pages_end(cc->inode, cc->cluster_idx,
+							cc->clen, ret);
+	return ret;
+}
+
+void f2fs_decompress_pages(struct bio *bio, struct page *page, bool verity)
+{
+	struct decompress_io_ctx *dic =
+			(struct decompress_io_ctx *)page_private(page);
+	struct f2fs_sb_info *sbi = F2FS_I_SB(dic->inode);
+	struct f2fs_inode_info *fi= F2FS_I(dic->inode);
+	const struct f2fs_compress_ops *cops =
+			f2fs_cops[fi->i_compress_algorithm];
+	int ret;
+
+	dec_page_count(sbi, F2FS_RD_DATA);
+
+	if (bio->bi_status || PageError(page))
+		dic->failed = true;
+
+	if (refcount_dec_not_one(&dic->ref))
+		return;
+
+	trace_f2fs_decompress_pages_start(dic->inode, dic->cluster_idx,
+				dic->cluster_size, fi->i_compress_algorithm);
+
+	/* submit partial compressed pages */
+	if (dic->failed) {
+		ret = -EIO;
+		goto out_free_dic;
+	}
+
+	dic->rbuf = vmap(dic->tpages, dic->cluster_size, VM_MAP, PAGE_KERNEL);
+	if (!dic->rbuf) {
+		ret = -ENOMEM;
+		goto out_free_dic;
+	}
+
+	dic->cbuf = vmap(dic->cpages, dic->nr_cpages, VM_MAP, PAGE_KERNEL_RO);
+	if (!dic->cbuf) {
+		ret = -ENOMEM;
+		goto out_vunmap_rbuf;
+	}
+
+	dic->clen = le32_to_cpu(dic->cbuf->clen);
+	dic->rlen = PAGE_SIZE << dic->log_cluster_size;
+
+	if (dic->clen > PAGE_SIZE * dic->nr_cpages - COMPRESS_HEADER_SIZE) {
+		ret = -EFSCORRUPTED;
+		goto out_vunmap_cbuf;
+	}
+
+	ret = cops->decompress_pages(dic);
+
+out_vunmap_cbuf:
+	vunmap(dic->cbuf);
+out_vunmap_rbuf:
+	vunmap(dic->rbuf);
+out_free_dic:
+	if (!verity)
+		f2fs_decompress_end_io(dic->rpages, dic->cluster_size,
+								ret, false);
+
+	trace_f2fs_decompress_pages_end(dic->inode, dic->cluster_idx,
+							dic->clen, ret);
+	if (!verity)
+		f2fs_free_dic(dic);
+}
+
+static bool is_page_in_cluster(struct compress_ctx *cc, pgoff_t index)
+{
+	if (cc->cluster_idx == NULL_CLUSTER)
+		return true;
+	return cc->cluster_idx == cluster_idx(cc, index);
+}
+
+bool f2fs_cluster_is_empty(struct compress_ctx *cc)
+{
+	return cc->nr_rpages == 0;
+}
+
+static bool f2fs_cluster_is_full(struct compress_ctx *cc)
+{
+	return cc->cluster_size == cc->nr_rpages;
+}
+
+bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index)
+{
+	if (f2fs_cluster_is_empty(cc))
+		return true;
+	return is_page_in_cluster(cc, index);
+}
+
+static bool __cluster_may_compress(struct compress_ctx *cc)
+{
+	struct f2fs_sb_info *sbi = F2FS_I_SB(cc->inode);
+	loff_t i_size = i_size_read(cc->inode);
+	unsigned nr_pages = DIV_ROUND_UP(i_size, PAGE_SIZE);
+	int i;
+
+	for (i = 0; i < cc->cluster_size; i++) {
+		struct page *page = cc->rpages[i];
+
+		f2fs_bug_on(sbi, !page);
+
+		if (unlikely(f2fs_cp_error(sbi)))
+			return false;
+		if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING)))
+			return false;
+
+		/* beyond EOF */
+		if (page->index >= nr_pages)
+			return false;
+	}
+	return true;
+}
+
+/* return # of compressed block addresses */
+static int f2fs_compressed_blocks(struct compress_ctx *cc)
+{
+	struct dnode_of_data dn;
+	int ret;
+
+	set_new_dnode(&dn, cc->inode, NULL, NULL, 0);
+	ret = f2fs_get_dnode_of_data(&dn, start_idx_of_cluster(cc),
+							LOOKUP_NODE);
+	if (ret) {
+		if (ret == -ENOENT)
+			ret = 0;
+		goto fail;
+	}
+
+	if (dn.data_blkaddr == COMPRESS_ADDR) {
+		int i;
+
+		ret = 1;
+		for (i = 1; i < cc->cluster_size; i++) {
+			block_t blkaddr;
+
+			blkaddr = datablock_addr(dn.inode,
+					dn.node_page, dn.ofs_in_node + i);
+			if (blkaddr != NULL_ADDR)
+				ret++;
+		}
+	}
+fail:
+	f2fs_put_dnode(&dn);
+	return ret;
+}
+
+int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index)
+{
+	struct compress_ctx cc = {
+		.inode = inode,
+		.log_cluster_size = F2FS_I(inode)->i_log_cluster_size,
+		.cluster_size = F2FS_I(inode)->i_cluster_size,
+		.cluster_idx = index >> F2FS_I(inode)->i_log_cluster_size,
+	};
+
+	return f2fs_compressed_blocks(&cc);
+}
+
+static bool cluster_may_compress(struct compress_ctx *cc)
+{
+	if (!f2fs_compressed_file(cc->inode))
+		return false;
+	if (f2fs_is_atomic_file(cc->inode))
+		return false;
+	if (f2fs_is_mmap_file(cc->inode))
+		return false;
+	if (!f2fs_cluster_is_full(cc))
+		return false;
+	return __cluster_may_compress(cc);
+}
+
+static void set_cluster_writeback(struct compress_ctx *cc)
+{
+	int i;
+
+	for (i = 0; i < cc->cluster_size; i++) {
+		if (cc->rpages[i])
+			set_page_writeback(cc->rpages[i]);
+	}
+}
+
+static void set_cluster_dirty(struct compress_ctx *cc)
+{
+	int i;
+
+	for (i = 0; i < cc->cluster_size; i++)
+		if (cc->rpages[i])
+			set_page_dirty(cc->rpages[i]);
+}
+
+static int prepare_compress_overwrite(struct compress_ctx *cc,
+		struct page **pagep, pgoff_t index, void **fsdata)
+{
+	struct f2fs_sb_info *sbi = F2FS_I_SB(cc->inode);
+	struct address_space *mapping = cc->inode->i_mapping;
+	struct page *page;
+	struct dnode_of_data dn;
+	sector_t last_block_in_bio;
+	unsigned fgp_flag = FGP_LOCK | FGP_WRITE | FGP_CREAT;
+	pgoff_t start_idx = start_idx_of_cluster(cc);
+	int i, ret;
+	bool prealloc;
+
+retry:
+	ret = f2fs_compressed_blocks(cc);
+	if (ret <= 0)
+		return ret;
+
+	/* compressed case */
+	prealloc = (ret < cc->cluster_size);
+
+	ret = f2fs_init_compress_ctx(cc);
+	if (ret)
+		return ret;
+
+	/* keep page reference to avoid page reclaim */
+	for (i = 0; i < cc->cluster_size; i++) {
+		page = f2fs_pagecache_get_page(mapping, start_idx + i,
+							fgp_flag, GFP_NOFS);
+		if (!page) {
+			ret = -ENOMEM;
+			goto unlock_pages;
+		}
+
+		if (PageUptodate(page))
+			unlock_page(page);
+		else
+			f2fs_compress_ctx_add_page(cc, page);
+	}
+
+	if (!f2fs_cluster_is_empty(cc)) {
+		struct bio *bio = NULL;
+
+		ret = f2fs_read_multi_pages(cc, &bio, cc->cluster_size,
+						&last_block_in_bio, false);
+		f2fs_destroy_compress_ctx(cc);
+		if (ret)
+			goto release_pages;
+		if (bio)
+			f2fs_submit_bio(sbi, bio, DATA);
+
+		ret = f2fs_init_compress_ctx(cc);
+		if (ret)
+			goto release_pages;
+	}
+
+	for (i = 0; i < cc->cluster_size; i++) {
+		f2fs_bug_on(sbi, cc->rpages[i]);
+
+		page = find_lock_page(mapping, start_idx + i);
+		f2fs_bug_on(sbi, !page);
+
+		f2fs_wait_on_page_writeback(page, DATA, true, true);
+
+		f2fs_compress_ctx_add_page(cc, page);
+		f2fs_put_page(page, 0);
+
+		if (!PageUptodate(page)) {
+			f2fs_unlock_rpages(cc, i + 1);
+			f2fs_put_rpages_mapping(cc, mapping, start_idx,
+					cc->cluster_size);
+			f2fs_destroy_compress_ctx(cc);
+			goto retry;
+		}
+	}
+
+	if (prealloc) {
+		__do_map_lock(sbi, F2FS_GET_BLOCK_PRE_AIO, true);
+
+		set_new_dnode(&dn, cc->inode, NULL, NULL, 0);
+
+		for (i = cc->cluster_size - 1; i > 0; i--) {
+			ret = f2fs_get_block(&dn, start_idx + i);
+			if (ret) {
+				i = cc->cluster_size;
+				break;
+			}
+
+			if (dn.data_blkaddr != NEW_ADDR)
+				break;
+		}
+
+		__do_map_lock(sbi, F2FS_GET_BLOCK_PRE_AIO, false);
+	}
+
+	if (likely(!ret)) {
+		*fsdata = cc->rpages;
+		*pagep = cc->rpages[offset_in_cluster(cc, index)];
+		return cc->cluster_size;
+	}
+
+unlock_pages:
+	f2fs_unlock_rpages(cc, i);
+release_pages:
+	f2fs_put_rpages_mapping(cc, mapping, start_idx, i);
+	f2fs_destroy_compress_ctx(cc);
+	return ret;
+}
+
+int f2fs_prepare_compress_overwrite(struct inode *inode,
+		struct page **pagep, pgoff_t index, void **fsdata)
+{
+	struct compress_ctx cc = {
+		.inode = inode,
+		.log_cluster_size = F2FS_I(inode)->i_log_cluster_size,
+		.cluster_size = F2FS_I(inode)->i_cluster_size,
+		.cluster_idx = index >> F2FS_I(inode)->i_log_cluster_size,
+		.rpages = NULL,
+		.nr_rpages = 0,
+	};
+
+	return prepare_compress_overwrite(&cc, pagep, index, fsdata);
+}
+
+bool f2fs_compress_write_end(struct inode *inode, void *fsdata,
+					pgoff_t index, unsigned copied)
+
+{
+	struct compress_ctx cc = {
+		.log_cluster_size = F2FS_I(inode)->i_log_cluster_size,
+		.cluster_size = F2FS_I(inode)->i_cluster_size,
+		.rpages = fsdata,
+	};
+	bool first_index = (index == cc.rpages[0]->index);
+
+	if (copied)
+		set_cluster_dirty(&cc);
+
+	f2fs_put_rpages_wbc(&cc, NULL, false, 1);
+	f2fs_destroy_compress_ctx(&cc);
+
+	return first_index;
+}
+
+static int f2fs_write_compressed_pages(struct compress_ctx *cc,
+					int *submitted,
+					struct writeback_control *wbc,
+					enum iostat_type io_type)
+{
+	struct inode *inode = cc->inode;
+	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
+	struct f2fs_inode_info *fi = F2FS_I(inode);
+	struct f2fs_io_info fio = {
+		.sbi = sbi,
+		.ino = cc->inode->i_ino,
+		.type = DATA,
+		.op = REQ_OP_WRITE,
+		.op_flags = wbc_to_write_flags(wbc),
+		.old_blkaddr = NEW_ADDR,
+		.page = NULL,
+		.encrypted_page = NULL,
+		.compressed_page = NULL,
+		.submitted = false,
+		.need_lock = LOCK_RETRY,
+		.io_type = io_type,
+		.io_wbc = wbc,
+		.encrypted = f2fs_encrypted_file(cc->inode),
+	};
+	struct dnode_of_data dn;
+	struct node_info ni;
+	struct compress_io_ctx *cic;
+	pgoff_t start_idx = start_idx_of_cluster(cc);
+	unsigned int last_index = cc->cluster_size - 1;
+	loff_t psize;
+	int i, err;
+
+	set_new_dnode(&dn, cc->inode, NULL, NULL, 0);
+
+	f2fs_lock_op(sbi);
+
+	err = f2fs_get_dnode_of_data(&dn, start_idx, LOOKUP_NODE);
+	if (err)
+		goto out_unlock_op;
+
+	for (i = 0; i < cc->cluster_size; i++) {
+		if (datablock_addr(dn.inode, dn.node_page,
+					dn.ofs_in_node + i) == NULL_ADDR)
+			goto out_put_dnode;
+	}
+
+	psize = (loff_t)(cc->rpages[last_index]->index + 1) << PAGE_SHIFT;
+
+	err = f2fs_get_node_info(fio.sbi, dn.nid, &ni);
+	if (err)
+		goto out_put_dnode;
+
+	fio.version = ni.version;
+
+	cic = f2fs_kzalloc(sbi, sizeof(struct compress_io_ctx), GFP_NOFS);
+	if (!cic)
+		goto out_put_dnode;
+
+	cic->magic = F2FS_COMPRESSED_PAGE_MAGIC;
+	cic->inode = inode;
+	refcount_set(&cic->ref, 1);
+	cic->rpages = f2fs_kzalloc(sbi, sizeof(struct page *) <<
+			cc->log_cluster_size, GFP_NOFS);
+	if (!cic->rpages)
+		goto out_put_cic;
+
+	cic->nr_rpages = cc->cluster_size;
+
+	for (i = 0; i < cc->nr_cpages; i++) {
+		f2fs_set_compressed_page(cc->cpages[i], inode,
+					cc->rpages[i + 1]->index,
+					cic, i ? &cic->ref : NULL);
+		fio.compressed_page = cc->cpages[i];
+		if (fio.encrypted) {
+			fio.page = cc->rpages[i + 1];
+			err = f2fs_encrypt_one_page(&fio);
+			if (err)
+				goto out_destroy_crypt;
+			cc->cpages[i] = fio.encrypted_page;
+		}
+	}
+
+	set_cluster_writeback(cc);
+
+	for (i = 0; i < cc->cluster_size; i++)
+		cic->rpages[i] = cc->rpages[i];
+
+	for (i = 0; i < cc->cluster_size; i++, dn.ofs_in_node++) {
+		block_t blkaddr;
+
+		blkaddr = datablock_addr(dn.inode, dn.node_page,
+							dn.ofs_in_node);
+		fio.page = cic->rpages[i];
+		fio.old_blkaddr = blkaddr;
+
+		/* cluster header */
+		if (i == 0) {
+			if (blkaddr == COMPRESS_ADDR)
+				fio.compr_blocks++;
+			if (__is_valid_data_blkaddr(blkaddr))
+				f2fs_invalidate_blocks(sbi, blkaddr);
+			f2fs_update_data_blkaddr(&dn, COMPRESS_ADDR);
+			goto unlock_continue;
+		}
+
+		if (fio.compr_blocks && __is_valid_data_blkaddr(blkaddr))
+			fio.compr_blocks++;
+
+		if (i > cc->nr_cpages) {
+			if (__is_valid_data_blkaddr(blkaddr)) {
+				f2fs_invalidate_blocks(sbi, blkaddr);
+				f2fs_update_data_blkaddr(&dn, NEW_ADDR);
+			}
+			goto unlock_continue;
+		}
+
+		f2fs_bug_on(fio.sbi, blkaddr == NULL_ADDR);
+
+		if (fio.encrypted)
+			fio.encrypted_page = cc->cpages[i - 1];
+		else
+			fio.compressed_page = cc->cpages[i - 1];
+
+		cc->cpages[i - 1] = NULL;
+		f2fs_outplace_write_data(&dn, &fio);
+		(*submitted)++;
+unlock_continue:
+		inode_dec_dirty_pages(cc->inode);
+		unlock_page(fio.page);
+	}
+
+	if (fio.compr_blocks)
+		f2fs_i_compr_blocks_update(inode, fio.compr_blocks - 1, false);
+	f2fs_i_compr_blocks_update(inode, cc->nr_cpages, true);
+
+	set_inode_flag(cc->inode, FI_APPEND_WRITE);
+	if (cc->cluster_idx == 0)
+		set_inode_flag(inode, FI_FIRST_BLOCK_WRITTEN);
+
+	f2fs_put_dnode(&dn);
+	f2fs_unlock_op(sbi);
+
+	down_write(&fi->i_sem);
+	if (fi->last_disk_size < psize)
+		fi->last_disk_size = psize;
+	up_write(&fi->i_sem);
+
+	f2fs_put_rpages(cc);
+	f2fs_destroy_compress_ctx(cc);
+	return 0;
+
+out_destroy_crypt:
+	kfree(cic->rpages);
+
+	for (--i; i >= 0; i--)
+		fscrypt_finalize_bounce_page(&cc->cpages[i]);
+	for (i = 0; i < cc->nr_cpages; i++) {
+		if (!cc->cpages[i])
+			continue;
+		f2fs_put_page(cc->cpages[i], 1);
+	}
+out_put_cic:
+	kfree(cic);
+out_put_dnode:
+	f2fs_put_dnode(&dn);
+out_unlock_op:
+	f2fs_unlock_op(sbi);
+	return -EAGAIN;
+}
+
+void f2fs_compress_write_end_io(struct bio *bio, struct page *page)
+{
+	struct f2fs_sb_info *sbi = bio->bi_private;
+	struct compress_io_ctx *cic =
+			(struct compress_io_ctx *)page_private(page);
+	int i;
+
+	if (unlikely(bio->bi_status))
+		mapping_set_error(cic->inode->i_mapping, -EIO);
+
+	f2fs_put_compressed_page(page);
+
+	dec_page_count(sbi, F2FS_WB_DATA);
+
+	if (refcount_dec_not_one(&cic->ref))
+		return;
+
+	for (i = 0; i < cic->nr_rpages; i++) {
+		WARN_ON(!cic->rpages[i]);
+		clear_cold_data(cic->rpages[i]);
+		end_page_writeback(cic->rpages[i]);
+	}
+
+	kfree(cic->rpages);
+	kfree(cic);
+}
+
+static int f2fs_write_raw_pages(struct compress_ctx *cc,
+					int *submitted,
+					struct writeback_control *wbc,
+					enum iostat_type io_type)
+{
+	struct address_space *mapping = cc->inode->i_mapping;
+	int _submitted, compr_blocks, ret;
+	int i = -1, err = 0;
+
+	compr_blocks = f2fs_compressed_blocks(cc);
+	if (compr_blocks < 0) {
+		err = compr_blocks;
+		goto out_err;
+	}
+
+	for (i = 0; i < cc->cluster_size; i++) {
+		if (!cc->rpages[i])
+			continue;
+retry_write:
+		if (cc->rpages[i]->mapping != mapping) {
+			unlock_page(cc->rpages[i]);
+			continue;
+		}
+
+		BUG_ON(!PageLocked(cc->rpages[i]));
+
+		ret = f2fs_write_single_data_page(cc->rpages[i], &_submitted,
+						NULL, NULL, wbc, io_type,
+						compr_blocks);
+		if (ret) {
+			if (ret == AOP_WRITEPAGE_ACTIVATE) {
+				unlock_page(cc->rpages[i]);
+				ret = 0;
+			} else if (ret == -EAGAIN) {
+				ret = 0;
+				cond_resched();
+				congestion_wait(BLK_RW_ASYNC, HZ/50);
+				lock_page(cc->rpages[i]);
+				clear_page_dirty_for_io(cc->rpages[i]);
+				goto retry_write;
+			}
+			err = ret;
+			goto out_fail;
+		}
+
+		*submitted += _submitted;
+	}
+	return 0;
+
+out_fail:
+	/* TODO: revoke partially updated block addresses */
+	BUG_ON(compr_blocks);
+out_err:
+	for (++i; i < cc->cluster_size; i++) {
+		if (!cc->rpages[i])
+			continue;
+		redirty_page_for_writepage(wbc, cc->rpages[i]);
+		unlock_page(cc->rpages[i]);
+	}
+	return err;
+}
+
+int f2fs_write_multi_pages(struct compress_ctx *cc,
+					int *submitted,
+					struct writeback_control *wbc,
+					enum iostat_type io_type)
+{
+	struct f2fs_inode_info *fi = F2FS_I(cc->inode);
+	const struct f2fs_compress_ops *cops =
+			f2fs_cops[fi->i_compress_algorithm];
+	int err;
+
+	*submitted = 0;
+	if (cluster_may_compress(cc)) {
+		err = f2fs_compress_pages(cc);
+		if (err == -EAGAIN) {
+			goto write;
+		} else if (err) {
+			f2fs_put_rpages_wbc(cc, wbc, true, 1);
+			goto destroy_out;
+		}
+
+		err = f2fs_write_compressed_pages(cc, submitted,
+							wbc, io_type);
+		cops->destroy_compress_ctx(cc);
+		if (!err)
+			return 0;
+		f2fs_bug_on(F2FS_I_SB(cc->inode), err != -EAGAIN);
+	}
+write:
+	f2fs_bug_on(F2FS_I_SB(cc->inode), *submitted);
+
+	err = f2fs_write_raw_pages(cc, submitted, wbc, io_type);
+	f2fs_put_rpages_wbc(cc, wbc, false, 0);
+destroy_out:
+	f2fs_destroy_compress_ctx(cc);
+	return err;
+}
+
+struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc)
+{
+	struct f2fs_sb_info *sbi = F2FS_I_SB(cc->inode);
+	struct decompress_io_ctx *dic;
+	pgoff_t start_idx = start_idx_of_cluster(cc);
+	int i;
+
+	dic = f2fs_kzalloc(sbi, sizeof(struct decompress_io_ctx), GFP_NOFS);
+	if (!dic)
+		return ERR_PTR(-ENOMEM);
+
+	dic->rpages = f2fs_kzalloc(sbi, sizeof(struct page *) <<
+			cc->log_cluster_size, GFP_NOFS);
+	if (!dic->rpages) {
+		kfree(dic);
+		return ERR_PTR(-ENOMEM);
+	}
+
+	dic->magic = F2FS_COMPRESSED_PAGE_MAGIC;
+	dic->inode = cc->inode;
+	refcount_set(&dic->ref, 1);
+	dic->cluster_idx = cc->cluster_idx;
+	dic->cluster_size = cc->cluster_size;
+	dic->log_cluster_size = cc->log_cluster_size;
+	dic->nr_cpages = cc->nr_cpages;
+	dic->failed = false;
+
+	for (i = 0; i < dic->cluster_size; i++)
+		dic->rpages[i] = cc->rpages[i];
+	dic->nr_rpages = cc->cluster_size;
+
+	dic->cpages = f2fs_kzalloc(sbi, sizeof(struct page *) *
+					dic->nr_cpages, GFP_NOFS);
+	if (!dic->cpages)
+		goto out_free;
+
+	for (i = 0; i < dic->nr_cpages; i++) {
+		struct page *page;
+
+		page = f2fs_grab_page();
+		if (!page)
+			goto out_free;
+
+		f2fs_set_compressed_page(page, cc->inode,
+					start_idx + i + 1,
+					dic, i ? &dic->ref : NULL);
+		dic->cpages[i] = page;
+	}
+
+	dic->tpages = f2fs_kzalloc(sbi, sizeof(struct page *) *
+					dic->cluster_size, GFP_NOFS);
+	if (!dic->tpages)
+		goto out_free;
+
+	for (i = 0; i < dic->cluster_size; i++) {
+		if (cc->rpages[i])
+			continue;
+
+		dic->tpages[i] = f2fs_grab_page();
+		if (!dic->tpages[i])
+			goto out_free;
+	}
+
+	for (i = 0; i < dic->cluster_size; i++) {
+		if (dic->tpages[i])
+			continue;
+		dic->tpages[i] = cc->rpages[i];
+	}
+
+	return dic;
+
+out_free:
+	f2fs_free_dic(dic);
+	return ERR_PTR(-ENOMEM);
+}
+
+void f2fs_free_dic(struct decompress_io_ctx *dic)
+{
+	int i;
+
+	if (dic->tpages) {
+		for (i = 0; i < dic->cluster_size; i++) {
+			if (dic->rpages[i])
+				continue;
+			f2fs_put_page(dic->tpages[i], 1);
+		}
+		kfree(dic->tpages);
+	}
+
+	if (dic->cpages) {
+		for (i = 0; i < dic->nr_cpages; i++) {
+			if (!dic->cpages[i])
+				continue;
+			f2fs_put_compressed_page(dic->cpages[i]);
+		}
+		kfree(dic->cpages);
+	}
+
+	kfree(dic->rpages);
+	kfree(dic);
+}
+
+void f2fs_decompress_end_io(struct page **rpages,
+			unsigned int cluster_size, bool err, bool verity)
+{
+	int i;
+
+	for (i = 0; i < cluster_size; i++) {
+		struct page *rpage = rpages[i];
+
+		if (!rpage)
+			continue;
+
+		if (err || PageError(rpage)) {
+			ClearPageUptodate(rpage);
+			ClearPageError(rpage);
+		} else {
+			if (!verity || fsverity_verify_page(rpage))
+				SetPageUptodate(rpage);
+			else
+				SetPageError(rpage);
+		}
+		unlock_page(rpage);
+	}
+}