linux/fs/f2fs/f2fs.h

4726 lines
153 KiB
C
Raw Normal View History

/* SPDX-License-Identifier: GPL-2.0 */
/*
* fs/f2fs/f2fs.h
*
* Copyright (c) 2012 Samsung Electronics Co., Ltd.
* http://www.samsung.com/
*/
#ifndef _LINUX_F2FS_H
#define _LINUX_F2FS_H
#include <linux/uio.h>
#include <linux/types.h>
#include <linux/page-flags.h>
#include <linux/slab.h>
#include <linux/crc32.h>
#include <linux/magic.h>
#include <linux/kobject.h>
#include <linux/sched.h>
#include <linux/cred.h>
#include <linux/sched/mm.h>
#include <linux/vmalloc.h>
#include <linux/bio.h>
#include <linux/blkdev.h>
#include <linux/quotaops.h>
#include <linux/part_stat.h>
#include <linux/rw_hint.h>
#include <crypto/hash.h>
#include <linux/fscrypt.h>
f2fs: add fs-verity support Add fs-verity support to f2fs. fs-verity is a filesystem feature that enables transparent integrity protection and authentication of read-only files. It uses a dm-verity like mechanism at the file level: a Merkle tree is used to verify any block in the file in log(filesize) time. It is implemented mainly by helper functions in fs/verity/. See Documentation/filesystems/fsverity.rst for the full documentation. The f2fs support for fs-verity consists of: - Adding a filesystem feature flag and an inode flag for fs-verity. - Implementing the fsverity_operations to support enabling verity on an inode and reading/writing the verity metadata. - Updating ->readpages() to verify data as it's read from verity files and to support reading verity metadata pages. - Updating ->write_begin(), ->write_end(), and ->writepages() to support writing verity metadata pages. - Calling the fs-verity hooks for ->open(), ->setattr(), and ->ioctl(). Like ext4, f2fs stores the verity metadata (Merkle tree and fsverity_descriptor) past the end of the file, starting at the first 64K boundary beyond i_size. This approach works because (a) verity files are readonly, and (b) pages fully beyond i_size aren't visible to userspace but can be read/written internally by f2fs with only some relatively small changes to f2fs. Extended attributes cannot be used because (a) f2fs limits the total size of an inode's xattr entries to 4096 bytes, which wouldn't be enough for even a single Merkle tree block, and (b) f2fs encryption doesn't encrypt xattrs, yet the verity metadata *must* be encrypted when the file is because it contains hashes of the plaintext data. Acked-by: Jaegeuk Kim <jaegeuk@kernel.org> Acked-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Eric Biggers <ebiggers@google.com>
2019-07-22 16:26:24 +00:00
#include <linux/fsverity.h>
struct pagevec;
#ifdef CONFIG_F2FS_CHECK_FS
#define f2fs_bug_on(sbi, condition) BUG_ON(condition)
#else
#define f2fs_bug_on(sbi, condition) \
do { \
if (WARN_ON(condition)) \
set_sbi_flag(sbi, SBI_NEED_FSCK); \
} while (0)
#endif
enum {
FAULT_KMALLOC,
FAULT_KVMALLOC,
FAULT_PAGE_ALLOC,
FAULT_PAGE_GET,
FAULT_ALLOC_BIO, /* it's obsolete due to bio_alloc() will never fail */
FAULT_ALLOC_NID,
FAULT_ORPHAN,
FAULT_BLOCK,
FAULT_DIR_DEPTH,
FAULT_EVICT_INODE,
FAULT_TRUNCATE,
FAULT_READ_IO,
FAULT_CHECKPOINT,
FAULT_DISCARD,
FAULT_WRITE_IO,
FAULT_SLAB_ALLOC,
FAULT_DQUOT_INIT,
FAULT_LOCK_OP,
f2fs: introduce FAULT_BLKADDR_CONSISTENCE We will encounter below inconsistent status when FAULT_BLKADDR type fault injection is on. Info: checkpoint state = d6 : nat_bits crc fsck compacted_summary orphan_inodes sudden-power-off [ASSERT] (fsck_chk_inode_blk:1254) --> ino: 0x1c100 has i_blocks: 000000c0, but has 191 blocks [FIX] (fsck_chk_inode_blk:1260) --> [0x1c100] i_blocks=0x000000c0 -> 0xbf [FIX] (fsck_chk_inode_blk:1269) --> [0x1c100] i_compr_blocks=0x00000026 -> 0x27 [ASSERT] (fsck_chk_inode_blk:1254) --> ino: 0x1cadb has i_blocks: 0000002f, but has 46 blocks [FIX] (fsck_chk_inode_blk:1260) --> [0x1cadb] i_blocks=0x0000002f -> 0x2e [FIX] (fsck_chk_inode_blk:1269) --> [0x1cadb] i_compr_blocks=0x00000011 -> 0x12 [ASSERT] (fsck_chk_inode_blk:1254) --> ino: 0x1c62c has i_blocks: 00000002, but has 1 blocks [FIX] (fsck_chk_inode_blk:1260) --> [0x1c62c] i_blocks=0x00000002 -> 0x1 After we inject fault into f2fs_is_valid_blkaddr() during truncation, a) it missed to increase @nr_free or @valid_blocks b) it can cause in blkaddr leak in truncated dnode Which may cause inconsistent status. This patch separates FAULT_BLKADDR_CONSISTENCE from FAULT_BLKADDR, and rename FAULT_BLKADDR to FAULT_BLKADDR_VALIDITY so that we can: a) use FAULT_BLKADDR_CONSISTENCE in f2fs_truncate_data_blocks_range() to simulate inconsistent issue independently, then it can verify fsck repair flow. b) FAULT_BLKADDR_VALIDITY fault will not cause any inconsistent status, we can just use it to check error path handling in kernel side. Reviewed-by: Daeho Jeong <daehojeong@google.com> Signed-off-by: Chao Yu <chao@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2024-01-12 19:41:32 +00:00
FAULT_BLKADDR_VALIDITY,
FAULT_BLKADDR_CONSISTENCE,
FAULT_NO_SEGMENT,
FAULT_MAX,
};
#ifdef CONFIG_F2FS_FAULT_INJECTION
#define F2FS_ALL_FAULT_TYPE (GENMASK(FAULT_MAX - 1, 0))
struct f2fs_fault_info {
atomic_t inject_ops;
int inject_rate;
unsigned int inject_type;
};
extern const char *f2fs_fault_name[FAULT_MAX];
#define IS_FAULT_SET(fi, type) ((fi)->inject_type & BIT(type))
/* maximum retry count for injected failure */
#define DEFAULT_FAILURE_RETRY_COUNT 8
#else
#define DEFAULT_FAILURE_RETRY_COUNT 1
#endif
/*
* For mount options
*/
#define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000001
#define F2FS_MOUNT_DISCARD 0x00000002
#define F2FS_MOUNT_NOHEAP 0x00000004
#define F2FS_MOUNT_XATTR_USER 0x00000008
#define F2FS_MOUNT_POSIX_ACL 0x00000010
#define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000020
#define F2FS_MOUNT_INLINE_XATTR 0x00000040
#define F2FS_MOUNT_INLINE_DATA 0x00000080
#define F2FS_MOUNT_INLINE_DENTRY 0x00000100
#define F2FS_MOUNT_FLUSH_MERGE 0x00000200
#define F2FS_MOUNT_NOBARRIER 0x00000400
#define F2FS_MOUNT_FASTBOOT 0x00000800
#define F2FS_MOUNT_READ_EXTENT_CACHE 0x00001000
#define F2FS_MOUNT_DATA_FLUSH 0x00002000
#define F2FS_MOUNT_FAULT_INJECTION 0x00004000
#define F2FS_MOUNT_USRQUOTA 0x00008000
#define F2FS_MOUNT_GRPQUOTA 0x00010000
#define F2FS_MOUNT_PRJQUOTA 0x00020000
#define F2FS_MOUNT_QUOTA 0x00040000
#define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00080000
#define F2FS_MOUNT_RESERVE_ROOT 0x00100000
#define F2FS_MOUNT_DISABLE_CHECKPOINT 0x00200000
#define F2FS_MOUNT_NORECOVERY 0x00400000
#define F2FS_MOUNT_ATGC 0x00800000
#define F2FS_MOUNT_MERGE_CHECKPOINT 0x01000000
#define F2FS_MOUNT_GC_MERGE 0x02000000
#define F2FS_MOUNT_COMPRESS_CACHE 0x04000000
#define F2FS_MOUNT_AGE_EXTENT_CACHE 0x08000000
#define F2FS_OPTION(sbi) ((sbi)->mount_opt)
#define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option)
#define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option)
#define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option)
#define ver_after(a, b) (typecheck(unsigned long long, a) && \
typecheck(unsigned long long, b) && \
((long long)((a) - (b)) > 0))
typedef u32 block_t; /*
* should not change u32, since it is the on-disk block
* address format, __le32.
*/
typedef u32 nid_t;
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>
2019-11-01 10:07:14 +00:00
#define COMPRESS_EXT_NUM 16
enum blkzone_allocation_policy {
BLKZONE_ALLOC_PRIOR_SEQ, /* Prioritize writing to sequential zones */
BLKZONE_ALLOC_ONLY_SEQ, /* Only allow writing to sequential zones */
BLKZONE_ALLOC_PRIOR_CONV, /* Prioritize writing to conventional zones */
};
/*
* An implementation of an rwsem that is explicitly unfair to readers. This
* prevents priority inversion when a low-priority reader acquires the read lock
* while sleeping on the write lock but the write lock is needed by
* higher-priority clients.
*/
struct f2fs_rwsem {
struct rw_semaphore internal_rwsem;
#ifdef CONFIG_F2FS_UNFAIR_RWSEM
wait_queue_head_t read_waiters;
#endif
};
struct f2fs_mount_info {
unsigned int opt;
block_t root_reserved_blocks; /* root reserved blocks */
kuid_t s_resuid; /* reserved blocks for uid */
kgid_t s_resgid; /* reserved blocks for gid */
int active_logs; /* # of active logs */
int inline_xattr_size; /* inline xattr size */
#ifdef CONFIG_F2FS_FAULT_INJECTION
struct f2fs_fault_info fault_info; /* For fault injection */
#endif
#ifdef CONFIG_QUOTA
/* Names of quota files with journalled quota */
char *s_qf_names[MAXQUOTAS];
int s_jquota_fmt; /* Format of quota to use */
#endif
/* For which write hints are passed down to block layer */
int alloc_mode; /* segment allocation policy */
int fsync_mode; /* fsync policy */
int fs_mode; /* fs mode: LFS or ADAPTIVE */
int bggc_mode; /* bggc mode: off, on or sync */
int memory_mode; /* memory mode */
int errors; /* errors parameter */
f2fs: introduce discard_unit mount option As James Z reported in bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=213877 [1.] One-line summary of the problem: Mount multiple SMR block devices exceed certain number cause system non-response [2.] Full description of the problem/report: Created some F2FS on SMR devices (mkfs.f2fs -m), then mounted in sequence. Each device is the same Model: HGST HSH721414AL (Size 14TB). Empirically, found that when the amount of SMR device * 1.5Gb > System RAM, the system ran out of memory and hung. No dmesg output. For example, 24 SMR Disk need 24*1.5GB = 36GB. A system with 32G RAM can only mount 21 devices, the 22nd device will be a reproducible cause of system hang. The number of SMR devices with other FS mounted on this system does not interfere with the result above. [3.] Keywords (i.e., modules, networking, kernel): F2FS, SMR, Memory [4.] Kernel information [4.1.] Kernel version (uname -a): Linux 5.13.4-200.fc34.x86_64 #1 SMP Tue Jul 20 20:27:29 UTC 2021 x86_64 x86_64 x86_64 GNU/Linux [4.2.] Kernel .config file: Default Fedora 34 with f2fs-tools-1.14.0-2.fc34.x86_64 [5.] Most recent kernel version which did not have the bug: None [6.] Output of Oops.. message (if applicable) with symbolic information resolved (see Documentation/admin-guide/oops-tracing.rst) None [7.] A small shell script or example program which triggers the problem (if possible) mount /dev/sdX /mnt/0X [8.] Memory consumption With 24 * 14T SMR Block device with F2FS free -g total used free shared buff/cache available Mem: 46 36 0 0 10 10 Swap: 0 0 0 With 3 * 14T SMR Block device with F2FS free -g total used free shared buff/cache available Mem: 7 5 0 0 1 1 Swap: 7 0 7 The root cause is, there are three bitmaps: - cur_valid_map - ckpt_valid_map - discard_map and each of them will cost ~500MB memory, {cur, ckpt}_valid_map are necessary, but discard_map is optional, since this bitmap will only be useful in mountpoint that small discard is enabled. For a blkzoned device such as SMR or ZNS devices, f2fs will only issue discard for a section(zone) when all blocks of that section are invalid, so, for such device, we don't need small discard functionality at all. This patch introduces a new mountoption "discard_unit=block|segment| section" to support issuing discard with different basic unit which is aligned to block, segment or section, so that user can specify "discard_unit=segment" or "discard_unit=section" to disable small discard functionality. Note that this mount option can not be changed by remount() due to related metadata need to be initialized during mount(). In order to save memory, let's use "discard_unit=section" for blkzoned device by default. Signed-off-by: Chao Yu <chao@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2021-08-03 00:15:43 +00:00
int discard_unit; /*
* discard command's offset/size should
* be aligned to this unit: block,
* segment or section
*/
fscrypt: handle test_dummy_encryption in more logical way The behavior of the test_dummy_encryption mount option is that when a new file (or directory or symlink) is created in an unencrypted directory, it's automatically encrypted using a dummy encryption policy. That's it; in particular, the encryption (or lack thereof) of existing files (or directories or symlinks) doesn't change. Unfortunately the implementation of test_dummy_encryption is a bit weird and confusing. When test_dummy_encryption is enabled and a file is being created in an unencrypted directory, we set up an encryption key (->i_crypt_info) for the directory. This isn't actually used to do any encryption, however, since the directory is still unencrypted! Instead, ->i_crypt_info is only used for inheriting the encryption policy. One consequence of this is that the filesystem ends up providing a "dummy context" (policy + nonce) instead of a "dummy policy". In commit ed318a6cc0b6 ("fscrypt: support test_dummy_encryption=v2"), I mistakenly thought this was required. However, actually the nonce only ends up being used to derive a key that is never used. Another consequence of this implementation is that it allows for 'inode->i_crypt_info != NULL && !IS_ENCRYPTED(inode)', which is an edge case that can be forgotten about. For example, currently FS_IOC_GET_ENCRYPTION_POLICY on an unencrypted directory may return the dummy encryption policy when the filesystem is mounted with test_dummy_encryption. That seems like the wrong thing to do, since again, the directory itself is not actually encrypted. Therefore, switch to a more logical and maintainable implementation where the dummy encryption policy inheritance is done without setting up keys for unencrypted directories. This involves: - Adding a function fscrypt_policy_to_inherit() which returns the encryption policy to inherit from a directory. This can be a real policy, a dummy policy, or no policy. - Replacing struct fscrypt_dummy_context, ->get_dummy_context(), etc. with struct fscrypt_dummy_policy, ->get_dummy_policy(), etc. - Making fscrypt_fname_encrypted_size() take an fscrypt_policy instead of an inode. Acked-by: Jaegeuk Kim <jaegeuk@kernel.org> Acked-by: Jeff Layton <jlayton@kernel.org> Link: https://lore.kernel.org/r/20200917041136.178600-13-ebiggers@kernel.org Signed-off-by: Eric Biggers <ebiggers@google.com>
2020-09-17 04:11:35 +00:00
struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */
block_t unusable_cap_perc; /* percentage for cap */
block_t unusable_cap; /* Amount of space allowed to be
* unusable when disabling checkpoint
*/
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>
2019-11-01 10:07:14 +00:00
/* For compression */
unsigned char compress_algorithm; /* algorithm type */
unsigned char compress_log_size; /* cluster log size */
unsigned char compress_level; /* compress level */
bool compress_chksum; /* compressed data chksum */
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>
2019-11-01 10:07:14 +00:00
unsigned char compress_ext_cnt; /* extension count */
unsigned char nocompress_ext_cnt; /* nocompress extension count */
int compress_mode; /* compression mode */
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>
2019-11-01 10:07:14 +00:00
unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */
unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */
};
#define F2FS_FEATURE_ENCRYPT 0x00000001
#define F2FS_FEATURE_BLKZONED 0x00000002
#define F2FS_FEATURE_ATOMIC_WRITE 0x00000004
#define F2FS_FEATURE_EXTRA_ATTR 0x00000008
#define F2FS_FEATURE_PRJQUOTA 0x00000010
#define F2FS_FEATURE_INODE_CHKSUM 0x00000020
#define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x00000040
#define F2FS_FEATURE_QUOTA_INO 0x00000080
#define F2FS_FEATURE_INODE_CRTIME 0x00000100
#define F2FS_FEATURE_LOST_FOUND 0x00000200
#define F2FS_FEATURE_VERITY 0x00000400
#define F2FS_FEATURE_SB_CHKSUM 0x00000800
#define F2FS_FEATURE_CASEFOLD 0x00001000
#define F2FS_FEATURE_COMPRESSION 0x00002000
#define F2FS_FEATURE_RO 0x00004000
#define __F2FS_HAS_FEATURE(raw_super, mask) \
((raw_super->feature & cpu_to_le32(mask)) != 0)
#define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask)
/*
* Default values for user and/or group using reserved blocks
*/
#define F2FS_DEF_RESUID 0
#define F2FS_DEF_RESGID 0
/*
* For checkpoint manager
*/
enum {
NAT_BITMAP,
SIT_BITMAP
};
#define CP_UMOUNT 0x00000001
#define CP_FASTBOOT 0x00000002
#define CP_SYNC 0x00000004
#define CP_RECOVERY 0x00000008
#define CP_DISCARD 0x00000010
#define CP_TRIMMED 0x00000020
#define CP_PAUSE 0x00000040
#define CP_RESIZE 0x00000080
#define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */
#define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */
#define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */
#define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */
#define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */
#define DEF_CP_INTERVAL 60 /* 60 secs */
#define DEF_IDLE_INTERVAL 5 /* 5 secs */
#define DEF_DISABLE_INTERVAL 5 /* 5 secs */
#define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */
#define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */
struct cp_control {
int reason;
__u64 trim_start;
__u64 trim_end;
__u64 trim_minlen;
};
/*
* indicate meta/data type
*/
enum {
META_CP,
META_NAT,
META_SIT,
META_SSA,
META_MAX,
META_POR,
f2fs: introduce DATA_GENERIC_ENHANCE Previously, f2fs_is_valid_blkaddr(, blkaddr, DATA_GENERIC) will check whether @blkaddr locates in main area or not. That check is weak, since the block address in range of main area can point to the address which is not valid in segment info table, and we can not detect such condition, we may suffer worse corruption as system continues running. So this patch introduce DATA_GENERIC_ENHANCE to enhance the sanity check which trigger SIT bitmap check rather than only range check. This patch did below changes as wel: - set SBI_NEED_FSCK in f2fs_is_valid_blkaddr(). - get rid of is_valid_data_blkaddr() to avoid panic if blkaddr is invalid. - introduce verify_fio_blkaddr() to wrap fio {new,old}_blkaddr validation check. - spread blkaddr check in: * f2fs_get_node_info() * __read_out_blkaddrs() * f2fs_submit_page_read() * ra_data_block() * do_recover_data() This patch can fix bug reported from bugzilla below: https://bugzilla.kernel.org/show_bug.cgi?id=203215 https://bugzilla.kernel.org/show_bug.cgi?id=203223 https://bugzilla.kernel.org/show_bug.cgi?id=203231 https://bugzilla.kernel.org/show_bug.cgi?id=203235 https://bugzilla.kernel.org/show_bug.cgi?id=203241 = Update by Jaegeuk Kim = DATA_GENERIC_ENHANCE enhanced to validate block addresses on read/write paths. But, xfstest/generic/446 compalins some generated kernel messages saying invalid bitmap was detected when reading a block. The reaons is, when we get the block addresses from extent_cache, there is no lock to synchronize it from truncating the blocks in parallel. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-04-15 07:26:32 +00:00
DATA_GENERIC, /* check range only */
DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */
DATA_GENERIC_ENHANCE_READ, /*
* strong check on range and segment
* bitmap but no warning due to race
* condition of read on truncated area
* by extent_cache
*/
f2fs: fix to do sanity check on destination blkaddr during recovery As Wenqing Liu reported in bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=216456 loop5: detected capacity change from 0 to 131072 F2FS-fs (loop5): recover_inode: ino = 6, name = hln, inline = 1 F2FS-fs (loop5): recover_data: ino = 6 (i_size: recover) err = 0 F2FS-fs (loop5): recover_inode: ino = 6, name = hln, inline = 1 F2FS-fs (loop5): recover_data: ino = 6 (i_size: recover) err = 0 F2FS-fs (loop5): recover_inode: ino = 6, name = hln, inline = 1 F2FS-fs (loop5): recover_data: ino = 6 (i_size: recover) err = 0 F2FS-fs (loop5): Bitmap was wrongly set, blk:5634 ------------[ cut here ]------------ WARNING: CPU: 3 PID: 1013 at fs/f2fs/segment.c:2198 RIP: 0010:update_sit_entry+0xa55/0x10b0 [f2fs] Call Trace: <TASK> f2fs_do_replace_block+0xa98/0x1890 [f2fs] f2fs_replace_block+0xeb/0x180 [f2fs] recover_data+0x1a69/0x6ae0 [f2fs] f2fs_recover_fsync_data+0x120d/0x1fc0 [f2fs] f2fs_fill_super+0x4665/0x61e0 [f2fs] mount_bdev+0x2cf/0x3b0 legacy_get_tree+0xed/0x1d0 vfs_get_tree+0x81/0x2b0 path_mount+0x47e/0x19d0 do_mount+0xce/0xf0 __x64_sys_mount+0x12c/0x1a0 do_syscall_64+0x38/0x90 entry_SYSCALL_64_after_hwframe+0x63/0xcd If we enable CONFIG_F2FS_CHECK_FS config, it will trigger a kernel panic instead of warning. The root cause is: in fuzzed image, SIT table is inconsistent with inode mapping table, result in triggering such warning during SIT table update. This patch introduces a new flag DATA_GENERIC_ENHANCE_UPDATE, w/ this flag, data block recovery flow can check destination blkaddr's validation in SIT table, and skip f2fs_replace_block() to avoid inconsistent status. Cc: stable@vger.kernel.org Reported-by: Wenqing Liu <wenqingliu0120@gmail.com> Signed-off-by: Chao Yu <chao@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2022-09-13 02:08:41 +00:00
DATA_GENERIC_ENHANCE_UPDATE, /*
* strong check on range and segment
* bitmap for update case
*/
META_GENERIC,
};
/* for the list of ino */
enum {
ORPHAN_INO, /* for orphan ino list */
APPEND_INO, /* for append ino list */
UPDATE_INO, /* for update ino list */
TRANS_DIR_INO, /* for transactions dir ino list */
XATTR_DIR_INO, /* for xattr updated dir ino list */
FLUSH_INO, /* for multiple device flushing */
MAX_INO_ENTRY, /* max. list */
};
struct ino_entry {
struct list_head list; /* list head */
nid_t ino; /* inode number */
unsigned int dirty_device; /* dirty device bitmap */
};
/* for the list of inodes to be GCed */
struct inode_entry {
struct list_head list; /* list head */
struct inode *inode; /* vfs inode pointer */
};
f2fs: fix to avoid broken of dnode block list f2fs recovery flow is relying on dnode block link list, it means fsynced file recovery depends on previous dnode's persistence in the list, so during fsync() we should wait on all regular inode's dnode writebacked before issuing flush. By this way, we can avoid dnode block list being broken by out-of-order IO submission due to IO scheduler or driver. Sheng Yong helps to do the test with this patch: Target:/data (f2fs, -) 64MB / 32768KB / 4KB / 8 1 / PERSIST / Index Base: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 867.82 204.15 41440.03 41370.54 680.8 1025.94 1031.08 2 871.87 205.87 41370.3 40275.2 791.14 1065.84 1101.7 3 866.52 205.69 41795.67 40596.16 694.69 1037.16 1031.48 Avg 868.7366667 205.2366667 41535.33333 40747.3 722.21 1042.98 1054.753333 After: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 798.81 202.5 41143 40613.87 602.71 838.08 913.83 2 805.79 206.47 40297.2 41291.46 604.44 840.75 924.27 3 814.83 206.17 41209.57 40453.62 602.85 834.66 927.91 Avg 806.4766667 205.0466667 40883.25667 40786.31667 603.3333333 837.83 922.0033333 Patched/Original: 0.928332713 0.999074239 0.984300676 1.000957528 0.835398753 0.803303994 0.874141189 It looks like atomic write will suffer performance regression. I suspect that the criminal is that we forcing to wait all dnode being in storage cache before we issue PREFLUSH+FUA. BTW, will commit ("f2fs: don't need to wait for node writes for atomic write") cause the problem: we will lose data of last transaction after SPO, even if atomic write return no error: - atomic_open(); - write() P1, P2, P3; - atomic_commit(); - writeback data: P1, P2, P3; - writeback node: N1, N2, N3; <--- If N1, N2 is not writebacked, N3 with fsync_mark is writebacked, In SPOR, we won't find N3 since node chain is broken, turns out that losing last transaction. - preflush + fua; - power-cut If we don't wait dnode writeback for atomic_write: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 779.91 206.03 41621.5 40333.16 716.9 1038.21 1034.85 2 848.51 204.35 40082.44 39486.17 791.83 1119.96 1083.77 3 772.12 206.27 41335.25 41599.65 723.29 1055.07 971.92 Avg 800.18 205.55 41013.06333 40472.99333 744.0066667 1071.08 1030.18 Patched/Original: 0.92108464 1.001526693 0.987425886 0.993268102 1.030180511 1.026942031 0.976702294 SQLite's performance recovers. Jaegeuk: "Practically, I don't see db corruption becase of this. We can excuse to lose the last transaction." Finally, we decide to keep original implementation of atomic write interface sematics that we don't wait all dnode writeback before preflush+fua submission. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-02 15:03:19 +00:00
struct fsync_node_entry {
struct list_head list; /* list head */
struct page *page; /* warm node page pointer */
unsigned int seq_id; /* sequence id */
};
f2fs: introduce checkpoint_merge mount option We've added a new mount options, "checkpoint_merge" and "nocheckpoint_merge", which creates a kernel daemon and makes it to merge concurrent checkpoint requests as much as possible to eliminate redundant checkpoint issues. Plus, we can eliminate the sluggish issue caused by slow checkpoint operation when the checkpoint is done in a process context in a cgroup having low i/o budget and cpu shares. To make this do better, we set the default i/o priority of the kernel daemon to "3", to give one higher priority than other kernel threads. The below verification result explains this. The basic idea has come from https://opensource.samsung.com. [Verification] Android Pixel Device(ARM64, 7GB RAM, 256GB UFS) Create two I/O cgroups (fg w/ weight 100, bg w/ wight 20) Set "strict_guarantees" to "1" in BFQ tunables In "fg" cgroup, - thread A => trigger 1000 checkpoint operations "for i in `seq 1 1000`; do touch test_dir1/file; fsync test_dir1; done" - thread B => gererating async. I/O "fio --rw=write --numjobs=1 --bs=128k --runtime=3600 --time_based=1 --filename=test_img --name=test" In "bg" cgroup, - thread C => trigger repeated checkpoint operations "echo $$ > /dev/blkio/bg/tasks; while true; do touch test_dir2/file; fsync test_dir2; done" We've measured thread A's execution time. [ w/o patch ] Elapsed Time: Avg. 68 seconds [ w/ patch ] Elapsed Time: Avg. 48 seconds Reported-by: kernel test robot <lkp@intel.com> Reported-by: Dan Carpenter <dan.carpenter@oracle.com> [Jaegeuk Kim: fix the return value in f2fs_start_ckpt_thread, reported by Dan] Signed-off-by: Daeho Jeong <daehojeong@google.com> Signed-off-by: Sungjong Seo <sj1557.seo@samsung.com> Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2021-01-19 00:00:42 +00:00
struct ckpt_req {
struct completion wait; /* completion for checkpoint done */
struct llist_node llnode; /* llist_node to be linked in wait queue */
int ret; /* return code of checkpoint */
ktime_t queue_time; /* request queued time */
};
struct ckpt_req_control {
struct task_struct *f2fs_issue_ckpt; /* checkpoint task */
int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */
f2fs: introduce checkpoint_merge mount option We've added a new mount options, "checkpoint_merge" and "nocheckpoint_merge", which creates a kernel daemon and makes it to merge concurrent checkpoint requests as much as possible to eliminate redundant checkpoint issues. Plus, we can eliminate the sluggish issue caused by slow checkpoint operation when the checkpoint is done in a process context in a cgroup having low i/o budget and cpu shares. To make this do better, we set the default i/o priority of the kernel daemon to "3", to give one higher priority than other kernel threads. The below verification result explains this. The basic idea has come from https://opensource.samsung.com. [Verification] Android Pixel Device(ARM64, 7GB RAM, 256GB UFS) Create two I/O cgroups (fg w/ weight 100, bg w/ wight 20) Set "strict_guarantees" to "1" in BFQ tunables In "fg" cgroup, - thread A => trigger 1000 checkpoint operations "for i in `seq 1 1000`; do touch test_dir1/file; fsync test_dir1; done" - thread B => gererating async. I/O "fio --rw=write --numjobs=1 --bs=128k --runtime=3600 --time_based=1 --filename=test_img --name=test" In "bg" cgroup, - thread C => trigger repeated checkpoint operations "echo $$ > /dev/blkio/bg/tasks; while true; do touch test_dir2/file; fsync test_dir2; done" We've measured thread A's execution time. [ w/o patch ] Elapsed Time: Avg. 68 seconds [ w/ patch ] Elapsed Time: Avg. 48 seconds Reported-by: kernel test robot <lkp@intel.com> Reported-by: Dan Carpenter <dan.carpenter@oracle.com> [Jaegeuk Kim: fix the return value in f2fs_start_ckpt_thread, reported by Dan] Signed-off-by: Daeho Jeong <daehojeong@google.com> Signed-off-by: Sungjong Seo <sj1557.seo@samsung.com> Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2021-01-19 00:00:42 +00:00
wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */
atomic_t issued_ckpt; /* # of actually issued ckpts */
atomic_t total_ckpt; /* # of total ckpts */
atomic_t queued_ckpt; /* # of queued ckpts */
struct llist_head issue_list; /* list for command issue */
spinlock_t stat_lock; /* lock for below checkpoint time stats */
unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */
unsigned int peak_time; /* peak wait time in msec until now */
};
/* for the bitmap indicate blocks to be discarded */
struct discard_entry {
struct list_head list; /* list head */
block_t start_blkaddr; /* start blockaddr of current segment */
unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */
};
/* minimum discard granularity, unit: block count */
#define MIN_DISCARD_GRANULARITY 1
/* default discard granularity of inner discard thread, unit: block count */
#define DEFAULT_DISCARD_GRANULARITY 16
/* default maximum discard granularity of ordered discard, unit: block count */
#define DEFAULT_MAX_ORDERED_DISCARD_GRANULARITY 16
/* max discard pend list number */
#define MAX_PLIST_NUM 512
#define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \
(MAX_PLIST_NUM - 1) : ((blk_num) - 1))
enum {
D_PREP, /* initial */
D_PARTIAL, /* partially submitted */
D_SUBMIT, /* all submitted */
D_DONE, /* finished */
};
struct discard_info {
block_t lstart; /* logical start address */
block_t len; /* length */
block_t start; /* actual start address in dev */
};
struct discard_cmd {
struct rb_node rb_node; /* rb node located in rb-tree */
struct discard_info di; /* discard info */
struct list_head list; /* command list */
struct completion wait; /* compleation */
struct block_device *bdev; /* bdev */
unsigned short ref; /* reference count */
unsigned char state; /* state */
unsigned char queued; /* queued discard */
int error; /* bio error */
spinlock_t lock; /* for state/bio_ref updating */
unsigned short bio_ref; /* bio reference count */
};
enum {
DPOLICY_BG,
DPOLICY_FORCE,
DPOLICY_FSTRIM,
DPOLICY_UMOUNT,
MAX_DPOLICY,
};
enum {
DPOLICY_IO_AWARE_DISABLE, /* force to not be aware of IO */
DPOLICY_IO_AWARE_ENABLE, /* force to be aware of IO */
DPOLICY_IO_AWARE_MAX,
};
struct discard_policy {
int type; /* type of discard */
unsigned int min_interval; /* used for candidates exist */
unsigned int mid_interval; /* used for device busy */
unsigned int max_interval; /* used for candidates not exist */
unsigned int max_requests; /* # of discards issued per round */
unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */
bool io_aware; /* issue discard in idle time */
bool sync; /* submit discard with REQ_SYNC flag */
bool ordered; /* issue discard by lba order */
bool timeout; /* discard timeout for put_super */
unsigned int granularity; /* discard granularity */
};
struct discard_cmd_control {
struct task_struct *f2fs_issue_discard; /* discard thread */
struct list_head entry_list; /* 4KB discard entry list */
struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */
struct list_head wait_list; /* store on-flushing entries */
struct list_head fstrim_list; /* in-flight discard from fstrim */
wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */
struct mutex cmd_lock;
unsigned int nr_discards; /* # of discards in the list */
unsigned int max_discards; /* max. discards to be issued */
unsigned int max_discard_request; /* max. discard request per round */
unsigned int min_discard_issue_time; /* min. interval between discard issue */
unsigned int mid_discard_issue_time; /* mid. interval between discard issue */
unsigned int max_discard_issue_time; /* max. interval between discard issue */
unsigned int discard_io_aware_gran; /* minimum discard granularity not be aware of I/O */
unsigned int discard_urgent_util; /* utilization which issue discard proactively */
unsigned int discard_granularity; /* discard granularity */
unsigned int max_ordered_discard; /* maximum discard granularity issued by lba order */
unsigned int discard_io_aware; /* io_aware policy */
unsigned int undiscard_blks; /* # of undiscard blocks */
unsigned int next_pos; /* next discard position */
atomic_t issued_discard; /* # of issued discard */
atomic_t queued_discard; /* # of queued discard */
atomic_t discard_cmd_cnt; /* # of cached cmd count */
struct rb_root_cached root; /* root of discard rb-tree */
bool rbtree_check; /* config for consistence check */
bool discard_wake; /* to wake up discard thread */
};
/* for the list of fsync inodes, used only during recovery */
struct fsync_inode_entry {
struct list_head list; /* list head */
struct inode *inode; /* vfs inode pointer */
block_t blkaddr; /* block address locating the last fsync */
block_t last_dentry; /* block address locating the last dentry */
};
#define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats))
#define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits))
#define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne)
#define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid)
#define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se)
#define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno)
#define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl))
#define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl))
static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i)
{
int before = nats_in_cursum(journal);
journal->n_nats = cpu_to_le16(before + i);
return before;
}
static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i)
{
int before = sits_in_cursum(journal);
journal->n_sits = cpu_to_le16(before + i);
return before;
}
static inline bool __has_cursum_space(struct f2fs_journal *journal,
int size, int type)
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
{
if (type == NAT_JOURNAL)
return size <= MAX_NAT_JENTRIES(journal);
return size <= MAX_SIT_JENTRIES(journal);
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
}
/* for inline stuff */
#define DEF_INLINE_RESERVED_SIZE 1
f2fs: enhance on-disk inode structure scalability This patch add new flag F2FS_EXTRA_ATTR storing in inode.i_inline to indicate that on-disk structure of current inode is extended. In order to extend, we changed the inode structure a bit: Original one: struct f2fs_inode { ... struct f2fs_extent i_ext; __le32 i_addr[DEF_ADDRS_PER_INODE]; __le32 i_nid[DEF_NIDS_PER_INODE]; } Extended one: struct f2fs_inode { ... struct f2fs_extent i_ext; union { struct { __le16 i_extra_isize; __le16 i_padding; __le32 i_extra_end[0]; }; __le32 i_addr[DEF_ADDRS_PER_INODE]; }; __le32 i_nid[DEF_NIDS_PER_INODE]; } Once F2FS_EXTRA_ATTR is set, we will steal four bytes in the head of i_addr field for storing i_extra_isize and i_padding. with i_extra_isize, we can calculate actual size of reserved space in i_addr, available attribute fields included in total extra attribute fields for current inode can be described as below: +--------------------+ | .i_mode | | ... | | .i_ext | +--------------------+ | .i_extra_isize |-----+ | .i_padding | | | .i_prjid | | | .i_atime_extra | | | .i_ctime_extra | | | .i_mtime_extra |<----+ | .i_inode_cs |<----- store blkaddr/inline from here | .i_xattr_cs | | ... | +--------------------+ | | | block address | | | +--------------------+ | .i_nid | +--------------------+ | node_footer | | (nid, ino, offset) | +--------------------+ Hence, with this patch, we would enhance scalability of f2fs inode for storing more newly added attribute. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-07-18 16:19:06 +00:00
static inline int get_extra_isize(struct inode *inode);
f2fs: support flexible inline xattr size Now, in product, more and more features based on file encryption were introduced, their demand of xattr space is increasing, however, inline xattr has fixed-size of 200 bytes, once inline xattr space is full, new increased xattr data would occupy additional xattr block which may bring us more space usage and performance regression during persisting. In order to resolve above issue, it's better to expand inline xattr size flexibly according to user's requirement. So this patch introduces new filesystem feature 'flexible inline xattr', and new mount option 'inline_xattr_size=%u', once mkfs enables the feature, we can use the option to make f2fs supporting flexible inline xattr size. To support this feature, we add extra attribute i_inline_xattr_size in inode layout, indicating that how many space inline xattr borrows from block address mapping space in inode layout, by this, we can easily locate and store flexible-sized inline xattr data in inode. Inode disk layout: +----------------------+ | .i_mode | | ... | | .i_ext | +----------------------+ | .i_extra_isize | | .i_inline_xattr_size |-----------+ | ... | | +----------------------+ | | .i_addr | | | - block address or | | | - inline data | | +----------------------+<---+ v | inline xattr | +---inline xattr range +----------------------+<---+ | .i_nid | +----------------------+ | node_footer | | (nid, ino, offset) | +----------------------+ Note that, we have to cnosider backward compatibility which reserved inline_data space, 200 bytes, all the time, reported by Sheng Yong. Previous inline data or directory always reserved 200 bytes in inode layout, even if inline_xattr is disabled. In order to keep inline_dentry's structure for backward compatibility, we get the space back only from inline_data. Signed-off-by: Chao Yu <yuchao0@huawei.com> Reported-by: Sheng Yong <shengyong1@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-09-06 13:59:50 +00:00
static inline int get_inline_xattr_addrs(struct inode *inode);
#define MAX_INLINE_DATA(inode) (sizeof(__le32) * \
(CUR_ADDRS_PER_INODE(inode) - \
get_inline_xattr_addrs(inode) - \
f2fs: support flexible inline xattr size Now, in product, more and more features based on file encryption were introduced, their demand of xattr space is increasing, however, inline xattr has fixed-size of 200 bytes, once inline xattr space is full, new increased xattr data would occupy additional xattr block which may bring us more space usage and performance regression during persisting. In order to resolve above issue, it's better to expand inline xattr size flexibly according to user's requirement. So this patch introduces new filesystem feature 'flexible inline xattr', and new mount option 'inline_xattr_size=%u', once mkfs enables the feature, we can use the option to make f2fs supporting flexible inline xattr size. To support this feature, we add extra attribute i_inline_xattr_size in inode layout, indicating that how many space inline xattr borrows from block address mapping space in inode layout, by this, we can easily locate and store flexible-sized inline xattr data in inode. Inode disk layout: +----------------------+ | .i_mode | | ... | | .i_ext | +----------------------+ | .i_extra_isize | | .i_inline_xattr_size |-----------+ | ... | | +----------------------+ | | .i_addr | | | - block address or | | | - inline data | | +----------------------+<---+ v | inline xattr | +---inline xattr range +----------------------+<---+ | .i_nid | +----------------------+ | node_footer | | (nid, ino, offset) | +----------------------+ Note that, we have to cnosider backward compatibility which reserved inline_data space, 200 bytes, all the time, reported by Sheng Yong. Previous inline data or directory always reserved 200 bytes in inode layout, even if inline_xattr is disabled. In order to keep inline_dentry's structure for backward compatibility, we get the space back only from inline_data. Signed-off-by: Chao Yu <yuchao0@huawei.com> Reported-by: Sheng Yong <shengyong1@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-09-06 13:59:50 +00:00
DEF_INLINE_RESERVED_SIZE))
/* for inline dir */
#define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \
((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
BITS_PER_BYTE + 1))
#define INLINE_DENTRY_BITMAP_SIZE(inode) \
DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE)
#define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \
((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
NR_INLINE_DENTRY(inode) + \
INLINE_DENTRY_BITMAP_SIZE(inode)))
/*
* For INODE and NODE manager
*/
/* for directory operations */
2020-05-07 07:59:04 +00:00
struct f2fs_filename {
/*
* The filename the user specified. This is NULL for some
* filesystem-internal operations, e.g. converting an inline directory
* to a non-inline one, or roll-forward recovering an encrypted dentry.
*/
const struct qstr *usr_fname;
/*
* The on-disk filename. For encrypted directories, this is encrypted.
* This may be NULL for lookups in an encrypted dir without the key.
*/
struct fscrypt_str disk_name;
/* The dirhash of this filename */
f2fs_hash_t hash;
#ifdef CONFIG_FS_ENCRYPTION
/*
* For lookups in encrypted directories: either the buffer backing
* disk_name, or a buffer that holds the decoded no-key name.
*/
struct fscrypt_str crypto_buf;
#endif
#if IS_ENABLED(CONFIG_UNICODE)
2020-05-07 07:59:04 +00:00
/*
* For casefolded directories: the casefolded name, but it's left NULL
* if the original name is not valid Unicode, if the original name is
* "." or "..", if the directory is both casefolded and encrypted and
* its encryption key is unavailable, or if the filesystem is doing an
* internal operation where usr_fname is also NULL. In all these cases
* we fall back to treating the name as an opaque byte sequence.
2020-05-07 07:59:04 +00:00
*/
struct qstr cf_name;
2020-05-07 07:59:04 +00:00
#endif
};
struct f2fs_dentry_ptr {
struct inode *inode;
void *bitmap;
struct f2fs_dir_entry *dentry;
__u8 (*filename)[F2FS_SLOT_LEN];
int max;
int nr_bitmap;
};
static inline void make_dentry_ptr_block(struct inode *inode,
struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t)
{
d->inode = inode;
d->max = NR_DENTRY_IN_BLOCK;
d->nr_bitmap = SIZE_OF_DENTRY_BITMAP;
d->bitmap = t->dentry_bitmap;
d->dentry = t->dentry;
d->filename = t->filename;
}
static inline void make_dentry_ptr_inline(struct inode *inode,
struct f2fs_dentry_ptr *d, void *t)
{
int entry_cnt = NR_INLINE_DENTRY(inode);
int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode);
int reserved_size = INLINE_RESERVED_SIZE(inode);
d->inode = inode;
d->max = entry_cnt;
d->nr_bitmap = bitmap_size;
d->bitmap = t;
d->dentry = t + bitmap_size + reserved_size;
d->filename = t + bitmap_size + reserved_size +
SIZE_OF_DIR_ENTRY * entry_cnt;
}
/*
* XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1
* as its node offset to distinguish from index node blocks.
* But some bits are used to mark the node block.
*/
#define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \
>> OFFSET_BIT_SHIFT)
enum {
ALLOC_NODE, /* allocate a new node page if needed */
LOOKUP_NODE, /* look up a node without readahead */
LOOKUP_NODE_RA, /*
* look up a node with readahead called
* by get_data_block.
*/
};
#define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */
/* congestion wait timeout value, default: 20ms */
#define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20))
/* maximum retry quota flush count */
#define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8
f2fs: avoid infinite loop to flush node pages xfstests/generic/475 can give EIO all the time which give an infinite loop to flush node page like below. Let's avoid it. [16418.518551] Call Trace: [16418.518553] ? dm_submit_bio+0x48/0x400 [16418.518574] ? submit_bio_checks+0x1ac/0x5a0 [16418.525207] __submit_bio+0x1a9/0x230 [16418.525210] ? kmem_cache_alloc+0x29e/0x3c0 [16418.525223] submit_bio_noacct+0xa8/0x2b0 [16418.525226] submit_bio+0x4d/0x130 [16418.525238] __submit_bio+0x49/0x310 [f2fs] [16418.525339] ? bio_add_page+0x6a/0x90 [16418.525344] f2fs_submit_page_bio+0x134/0x1f0 [f2fs] [16418.525365] read_node_page+0x125/0x1b0 [f2fs] [16418.525388] __get_node_page.part.0+0x58/0x3f0 [f2fs] [16418.525409] __get_node_page+0x2f/0x60 [f2fs] [16418.525431] f2fs_get_dnode_of_data+0x423/0x860 [f2fs] [16418.525452] ? asm_sysvec_apic_timer_interrupt+0x12/0x20 [16418.525458] ? __mod_memcg_state.part.0+0x2a/0x30 [16418.525465] ? __mod_memcg_lruvec_state+0x27/0x40 [16418.525467] ? __xa_set_mark+0x57/0x70 [16418.525472] f2fs_do_write_data_page+0x10e/0x7b0 [f2fs] [16418.525493] f2fs_write_single_data_page+0x555/0x830 [f2fs] [16418.525514] ? sysvec_apic_timer_interrupt+0x4e/0x90 [16418.525518] ? asm_sysvec_apic_timer_interrupt+0x12/0x20 [16418.525523] f2fs_write_cache_pages+0x303/0x880 [f2fs] [16418.525545] ? blk_flush_plug_list+0x47/0x100 [16418.525548] f2fs_write_data_pages+0xfd/0x320 [f2fs] [16418.525569] do_writepages+0xd5/0x210 [16418.525648] filemap_fdatawrite_wbc+0x7d/0xc0 [16418.525655] filemap_fdatawrite+0x50/0x70 [16418.525658] f2fs_sync_dirty_inodes+0xa4/0x230 [f2fs] [16418.525679] f2fs_write_checkpoint+0x16d/0x1720 [f2fs] [16418.525699] ? ttwu_do_wakeup+0x1c/0x160 [16418.525709] ? ttwu_do_activate+0x6d/0xd0 [16418.525711] ? __wait_for_common+0x11d/0x150 [16418.525715] kill_f2fs_super+0xca/0x100 [f2fs] [16418.525733] deactivate_locked_super+0x3b/0xb0 [16418.525739] deactivate_super+0x40/0x50 [16418.525741] cleanup_mnt+0x139/0x190 [16418.525747] __cleanup_mnt+0x12/0x20 [16418.525749] task_work_run+0x6d/0xa0 [16418.525765] exit_to_user_mode_prepare+0x1ad/0x1b0 [16418.525771] syscall_exit_to_user_mode+0x27/0x50 [16418.525774] do_syscall_64+0x48/0xc0 [16418.525776] entry_SYSCALL_64_after_hwframe+0x44/0xae Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2022-03-29 23:25:54 +00:00
/* maximum retry of EIO'ed page */
#define MAX_RETRY_PAGE_EIO 100
f2fs: avoid an infinite loop in f2fs_sync_dirty_inodes If one read IO is always failing, we can fall into an infinite loop in f2fs_sync_dirty_inodes. This happens during xfstests/generic/475. [ 142.803335] Buffer I/O error on dev dm-1, logical block 8388592, async page read ... [ 382.887210] submit_bio_noacct+0xdd/0x2a0 [ 382.887213] submit_bio+0x80/0x110 [ 382.887223] __submit_bio+0x4d/0x300 [f2fs] [ 382.887282] f2fs_submit_page_bio+0x125/0x200 [f2fs] [ 382.887299] __get_meta_page+0xc9/0x280 [f2fs] [ 382.887315] f2fs_get_meta_page+0x13/0x20 [f2fs] [ 382.887331] f2fs_get_node_info+0x317/0x3c0 [f2fs] [ 382.887350] f2fs_do_write_data_page+0x327/0x6f0 [f2fs] [ 382.887367] f2fs_write_single_data_page+0x5b7/0x960 [f2fs] [ 382.887386] f2fs_write_cache_pages+0x302/0x890 [f2fs] [ 382.887405] ? preempt_count_add+0x7a/0xc0 [ 382.887408] f2fs_write_data_pages+0xfd/0x320 [f2fs] [ 382.887425] ? _raw_spin_unlock+0x1a/0x30 [ 382.887428] do_writepages+0xd3/0x1d0 [ 382.887432] filemap_fdatawrite_wbc+0x69/0x90 [ 382.887434] filemap_fdatawrite+0x50/0x70 [ 382.887437] f2fs_sync_dirty_inodes+0xa4/0x270 [f2fs] [ 382.887453] f2fs_write_checkpoint+0x189/0x1640 [f2fs] [ 382.887469] ? schedule_timeout+0x114/0x150 [ 382.887471] ? ttwu_do_activate+0x6d/0xb0 [ 382.887473] ? preempt_count_add+0x7a/0xc0 [ 382.887476] kill_f2fs_super+0xca/0x100 [f2fs] [ 382.887491] deactivate_locked_super+0x35/0xa0 [ 382.887494] deactivate_super+0x40/0x50 [ 382.887497] cleanup_mnt+0x139/0x190 [ 382.887499] __cleanup_mnt+0x12/0x20 [ 382.887501] task_work_run+0x64/0xa0 [ 382.887505] exit_to_user_mode_prepare+0x1b7/0x1c0 [ 382.887508] syscall_exit_to_user_mode+0x27/0x50 [ 382.887510] do_syscall_64+0x48/0xc0 [ 382.887513] entry_SYSCALL_64_after_hwframe+0x44/0xae Reviewed-by: Chao Yu <chao@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2022-03-02 19:39:08 +00:00
#define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */
#define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */
/* dirty segments threshold for triggering CP */
#define DEFAULT_DIRTY_THRESHOLD 4
#define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS
#define RECOVERY_MIN_RA_BLOCKS 1
#define F2FS_ONSTACK_PAGES 16 /* nr of onstack pages */
/* for in-memory extent cache entry */
#define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */
/* number of extent info in extent cache we try to shrink */
#define READ_EXTENT_CACHE_SHRINK_NUMBER 128
f2fs: add block_age-based extent cache This patch introduces a runtime hot/cold data separation method for f2fs, in order to improve the accuracy for data temperature classification, reduce the garbage collection overhead after long-term data updates. Enhanced hot/cold data separation can record data block update frequency as "age" of the extent per inode, and take use of the age info to indicate better temperature type for data block allocation: - It records total data blocks allocated since mount; - When file extent has been updated, it calculate the count of data blocks allocated since last update as the age of the extent; - Before the data block allocated, it searches for the age info and chooses the suitable segment for allocation. Test and result: - Prepare: create about 30000 files * 3% for cold files (with cold file extension like .apk, from 3M to 10M) * 50% for warm files (with random file extension like .FcDxq, from 1K to 4M) * 47% for hot files (with hot file extension like .db, from 1K to 256K) - create(5%)/random update(90%)/delete(5%) the files * total write amount is about 70G * fsync will be called for .db files, and buffered write will be used for other files The storage of test device is large enough(128G) so that it will not switch to SSR mode during the test. Benefit: dirty segment count increment reduce about 14% - before: Dirty +21110 - after: Dirty +18286 Signed-off-by: qixiaoyu1 <qixiaoyu1@xiaomi.com> Signed-off-by: xiongping1 <xiongping1@xiaomi.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2022-12-02 01:37:15 +00:00
/* number of age extent info in extent cache we try to shrink */
#define AGE_EXTENT_CACHE_SHRINK_NUMBER 128
#define LAST_AGE_WEIGHT 30
#define SAME_AGE_REGION 1024
/*
* Define data block with age less than 1GB as hot data
* define data block with age less than 10GB but more than 1GB as warm data
*/
#define DEF_HOT_DATA_AGE_THRESHOLD 262144
#define DEF_WARM_DATA_AGE_THRESHOLD 2621440
/* extent cache type */
enum extent_type {
EX_READ,
f2fs: add block_age-based extent cache This patch introduces a runtime hot/cold data separation method for f2fs, in order to improve the accuracy for data temperature classification, reduce the garbage collection overhead after long-term data updates. Enhanced hot/cold data separation can record data block update frequency as "age" of the extent per inode, and take use of the age info to indicate better temperature type for data block allocation: - It records total data blocks allocated since mount; - When file extent has been updated, it calculate the count of data blocks allocated since last update as the age of the extent; - Before the data block allocated, it searches for the age info and chooses the suitable segment for allocation. Test and result: - Prepare: create about 30000 files * 3% for cold files (with cold file extension like .apk, from 3M to 10M) * 50% for warm files (with random file extension like .FcDxq, from 1K to 4M) * 47% for hot files (with hot file extension like .db, from 1K to 256K) - create(5%)/random update(90%)/delete(5%) the files * total write amount is about 70G * fsync will be called for .db files, and buffered write will be used for other files The storage of test device is large enough(128G) so that it will not switch to SSR mode during the test. Benefit: dirty segment count increment reduce about 14% - before: Dirty +21110 - after: Dirty +18286 Signed-off-by: qixiaoyu1 <qixiaoyu1@xiaomi.com> Signed-off-by: xiongping1 <xiongping1@xiaomi.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2022-12-02 01:37:15 +00:00
EX_BLOCK_AGE,
NR_EXTENT_CACHES,
};
struct extent_info {
unsigned int fofs; /* start offset in a file */
unsigned int len; /* length of the extent */
union {
/* read extent_cache */
struct {
/* start block address of the extent */
block_t blk;
#ifdef CONFIG_F2FS_FS_COMPRESSION
/* physical extent length of compressed blocks */
unsigned int c_len;
#endif
};
f2fs: add block_age-based extent cache This patch introduces a runtime hot/cold data separation method for f2fs, in order to improve the accuracy for data temperature classification, reduce the garbage collection overhead after long-term data updates. Enhanced hot/cold data separation can record data block update frequency as "age" of the extent per inode, and take use of the age info to indicate better temperature type for data block allocation: - It records total data blocks allocated since mount; - When file extent has been updated, it calculate the count of data blocks allocated since last update as the age of the extent; - Before the data block allocated, it searches for the age info and chooses the suitable segment for allocation. Test and result: - Prepare: create about 30000 files * 3% for cold files (with cold file extension like .apk, from 3M to 10M) * 50% for warm files (with random file extension like .FcDxq, from 1K to 4M) * 47% for hot files (with hot file extension like .db, from 1K to 256K) - create(5%)/random update(90%)/delete(5%) the files * total write amount is about 70G * fsync will be called for .db files, and buffered write will be used for other files The storage of test device is large enough(128G) so that it will not switch to SSR mode during the test. Benefit: dirty segment count increment reduce about 14% - before: Dirty +21110 - after: Dirty +18286 Signed-off-by: qixiaoyu1 <qixiaoyu1@xiaomi.com> Signed-off-by: xiongping1 <xiongping1@xiaomi.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2022-12-02 01:37:15 +00:00
/* block age extent_cache */
struct {
/* block age of the extent */
unsigned long long age;
/* last total blocks allocated */
unsigned long long last_blocks;
};
};
};
struct extent_node {
struct rb_node rb_node; /* rb node located in rb-tree */
struct extent_info ei; /* extent info */
struct list_head list; /* node in global extent list of sbi */
struct extent_tree *et; /* extent tree pointer */
};
struct extent_tree {
nid_t ino; /* inode number */
enum extent_type type; /* keep the extent tree type */
struct rb_root_cached root; /* root of extent info rb-tree */
struct extent_node *cached_en; /* recently accessed extent node */
struct list_head list; /* to be used by sbi->zombie_list */
rwlock_t lock; /* protect extent info rb-tree */
atomic_t node_cnt; /* # of extent node in rb-tree*/
bool largest_updated; /* largest extent updated */
struct extent_info largest; /* largest cached extent for EX_READ */
};
struct extent_tree_info {
struct radix_tree_root extent_tree_root;/* cache extent cache entries */
struct mutex extent_tree_lock; /* locking extent radix tree */
struct list_head extent_list; /* lru list for shrinker */
spinlock_t extent_lock; /* locking extent lru list */
atomic_t total_ext_tree; /* extent tree count */
struct list_head zombie_list; /* extent zombie tree list */
atomic_t total_zombie_tree; /* extent zombie tree count */
atomic_t total_ext_node; /* extent info count */
};
/*
* State of block returned by f2fs_map_blocks.
*/
#define F2FS_MAP_NEW (1U << 0)
#define F2FS_MAP_MAPPED (1U << 1)
#define F2FS_MAP_DELALLOC (1U << 2)
#define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\
F2FS_MAP_DELALLOC)
struct f2fs_map_blocks {
struct block_device *m_bdev; /* for multi-device dio */
block_t m_pblk;
block_t m_lblk;
unsigned int m_len;
unsigned int m_flags;
pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */
pgoff_t *m_next_extent; /* point to next possible extent */
int m_seg_type;
f2fs: fix out-place-update DIO write In get_more_blocks(), we may override @create as below code: create = dio->op == REQ_OP_WRITE; if (dio->flags & DIO_SKIP_HOLES) { if (fs_startblk <= ((i_size_read(dio->inode) - 1) >> i_blkbits)) create = 0; } But in f2fs_map_blocks(), we only trigger f2fs_balance_fs() if @create is 1, so in LFS mode, dio overwrite under LFS mode can easily run out of free segments, result in below panic. Call Trace: allocate_segment_by_default+0xa8/0x270 [f2fs] f2fs_allocate_data_block+0x1ea/0x5c0 [f2fs] __allocate_data_block+0x306/0x480 [f2fs] f2fs_map_blocks+0x6f6/0x920 [f2fs] __get_data_block+0x4f/0xb0 [f2fs] get_data_block_dio_write+0x50/0x60 [f2fs] do_blockdev_direct_IO+0xcd5/0x21e0 __blockdev_direct_IO+0x3a/0x3c f2fs_direct_IO+0x1ff/0x4a0 [f2fs] generic_file_direct_write+0xd9/0x160 __generic_file_write_iter+0xbb/0x1e0 f2fs_file_write_iter+0xaf/0x220 [f2fs] __vfs_write+0xd0/0x130 vfs_write+0xb2/0x1b0 SyS_pwrite64+0x69/0xa0 ? vtime_user_exit+0x29/0x70 do_syscall_64+0x6e/0x160 entry_SYSCALL64_slow_path+0x25/0x25 RIP: new_curseg+0x36f/0x380 [f2fs] RSP: ffffac570393f7a8 So this patch introduces a parameter map.m_may_create to indicate that f2fs_map_blocks() is called from write or read path, which can give the right hint to let f2fs_map_blocks() trigger OPU allocation and call f2fs_balanc_fs() correctly. BTW, it disables physical address preallocation for direct IO in f2fs_preallocate_blocks, which is redundant to OPU allocation of f2fs_map_blocks. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-11-13 06:33:45 +00:00
bool m_may_create; /* indicate it is from write path */
bool m_multidev_dio; /* indicate it allows multi-device dio */
};
/* for flag in get_data_block */
enum {
F2FS_GET_BLOCK_DEFAULT,
F2FS_GET_BLOCK_FIEMAP,
F2FS_GET_BLOCK_BMAP,
F2FS_GET_BLOCK_DIO,
F2FS_GET_BLOCK_PRE_DIO,
F2FS_GET_BLOCK_PRE_AIO,
F2FS_GET_BLOCK_PRECACHE,
};
/*
* i_advise uses FADVISE_XXX_BIT. We can add additional hints later.
*/
#define FADVISE_COLD_BIT 0x01
#define FADVISE_LOST_PINO_BIT 0x02
#define FADVISE_ENCRYPT_BIT 0x04
#define FADVISE_ENC_NAME_BIT 0x08
#define FADVISE_KEEP_SIZE_BIT 0x10
#define FADVISE_HOT_BIT 0x20
f2fs: add fs-verity support Add fs-verity support to f2fs. fs-verity is a filesystem feature that enables transparent integrity protection and authentication of read-only files. It uses a dm-verity like mechanism at the file level: a Merkle tree is used to verify any block in the file in log(filesize) time. It is implemented mainly by helper functions in fs/verity/. See Documentation/filesystems/fsverity.rst for the full documentation. The f2fs support for fs-verity consists of: - Adding a filesystem feature flag and an inode flag for fs-verity. - Implementing the fsverity_operations to support enabling verity on an inode and reading/writing the verity metadata. - Updating ->readpages() to verify data as it's read from verity files and to support reading verity metadata pages. - Updating ->write_begin(), ->write_end(), and ->writepages() to support writing verity metadata pages. - Calling the fs-verity hooks for ->open(), ->setattr(), and ->ioctl(). Like ext4, f2fs stores the verity metadata (Merkle tree and fsverity_descriptor) past the end of the file, starting at the first 64K boundary beyond i_size. This approach works because (a) verity files are readonly, and (b) pages fully beyond i_size aren't visible to userspace but can be read/written internally by f2fs with only some relatively small changes to f2fs. Extended attributes cannot be used because (a) f2fs limits the total size of an inode's xattr entries to 4096 bytes, which wouldn't be enough for even a single Merkle tree block, and (b) f2fs encryption doesn't encrypt xattrs, yet the verity metadata *must* be encrypted when the file is because it contains hashes of the plaintext data. Acked-by: Jaegeuk Kim <jaegeuk@kernel.org> Acked-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Eric Biggers <ebiggers@google.com>
2019-07-22 16:26:24 +00:00
#define FADVISE_VERITY_BIT 0x40
#define FADVISE_TRUNC_BIT 0x80
#define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT)
#define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT)
#define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT)
#define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT)
#define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT)
#define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT)
#define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT)
#define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT)
#define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT)
#define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT)
#define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT)
#define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT)
#define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT)
#define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT)
#define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT)
#define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT)
f2fs: add fs-verity support Add fs-verity support to f2fs. fs-verity is a filesystem feature that enables transparent integrity protection and authentication of read-only files. It uses a dm-verity like mechanism at the file level: a Merkle tree is used to verify any block in the file in log(filesize) time. It is implemented mainly by helper functions in fs/verity/. See Documentation/filesystems/fsverity.rst for the full documentation. The f2fs support for fs-verity consists of: - Adding a filesystem feature flag and an inode flag for fs-verity. - Implementing the fsverity_operations to support enabling verity on an inode and reading/writing the verity metadata. - Updating ->readpages() to verify data as it's read from verity files and to support reading verity metadata pages. - Updating ->write_begin(), ->write_end(), and ->writepages() to support writing verity metadata pages. - Calling the fs-verity hooks for ->open(), ->setattr(), and ->ioctl(). Like ext4, f2fs stores the verity metadata (Merkle tree and fsverity_descriptor) past the end of the file, starting at the first 64K boundary beyond i_size. This approach works because (a) verity files are readonly, and (b) pages fully beyond i_size aren't visible to userspace but can be read/written internally by f2fs with only some relatively small changes to f2fs. Extended attributes cannot be used because (a) f2fs limits the total size of an inode's xattr entries to 4096 bytes, which wouldn't be enough for even a single Merkle tree block, and (b) f2fs encryption doesn't encrypt xattrs, yet the verity metadata *must* be encrypted when the file is because it contains hashes of the plaintext data. Acked-by: Jaegeuk Kim <jaegeuk@kernel.org> Acked-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Eric Biggers <ebiggers@google.com>
2019-07-22 16:26:24 +00:00
#define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT)
#define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT)
#define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT)
#define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT)
#define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT)
#define DEF_DIR_LEVEL 0
/* used for f2fs_inode_info->flags */
enum {
FI_NEW_INODE, /* indicate newly allocated inode */
FI_DIRTY_INODE, /* indicate inode is dirty or not */
FI_AUTO_RECOVER, /* indicate inode is recoverable */
FI_DIRTY_DIR, /* indicate directory has dirty pages */
FI_INC_LINK, /* need to increment i_nlink */
FI_ACL_MODE, /* indicate acl mode */
FI_NO_ALLOC, /* should not allocate any blocks */
FI_FREE_NID, /* free allocated nide */
FI_NO_EXTENT, /* not to use the extent cache */
FI_INLINE_XATTR, /* used for inline xattr */
FI_INLINE_DATA, /* used for inline data*/
FI_INLINE_DENTRY, /* used for inline dentry */
FI_APPEND_WRITE, /* inode has appended data */
FI_UPDATE_WRITE, /* inode has in-place-update data */
FI_NEED_IPU, /* used for ipu per file */
FI_ATOMIC_FILE, /* indicate atomic file */
FI_DATA_EXIST, /* indicate data exists */
FI_INLINE_DOTS, /* indicate inline dot dentries */
FI_SKIP_WRITES, /* should skip data page writeback */
FI_OPU_WRITE, /* used for opu per file */
FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */
FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */
FI_HOT_DATA, /* indicate file is hot */
FI_EXTRA_ATTR, /* indicate file has extra attribute */
FI_PROJ_INHERIT, /* indicate file inherits projectid */
FI_PIN_FILE, /* indicate file should not be gced */
FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
FI_COMPRESSED_FILE, /* indicate file's data can be compressed */
FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */
FI_MMAP_FILE, /* indicate file was mmapped */
FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */
FI_COMPRESS_RELEASED, /* compressed blocks were released */
f2fs: swap: support migrating swapfile in aligned write mode This patch supports to migrate swapfile in aligned write mode during swapon in order to keep swapfile being aligned to section as much as possible, then pinned swapfile will locates fully filled section which may not affected by GC. However, for the case that swapfile's size is not aligned to section size, it will still leave last extent in file's tail as unaligned due to its size is smaller than section size, like case #2. case #1 xfs_io -f /mnt/f2fs/file -c "pwrite 0 4M" -c "fsync" Before swapon: EXT: FILE-OFFSET BLOCK-RANGE TOTAL FLAGS 0: [0..3047]: 1123352..1126399 3048 0x1000 1: [3048..7143]: 237568..241663 4096 0x1000 2: [7144..8191]: 245760..246807 1048 0x1001 After swapon: EXT: FILE-OFFSET BLOCK-RANGE TOTAL FLAGS 0: [0..8191]: 249856..258047 8192 0x1001 Kmsg: F2FS-fs (zram0): Swapfile (2) is not align to section: 1) creat(), 2) ioctl(F2FS_IOC_SET_PIN_FILE), 3) fallocate(2097152 * n) case #2 xfs_io -f /mnt/f2fs/file -c "pwrite 0 3M" -c "fsync" Before swapon: EXT: FILE-OFFSET BLOCK-RANGE TOTAL FLAGS 0: [0..3047]: 246808..249855 3048 0x1000 1: [3048..6143]: 237568..240663 3096 0x1001 After swapon: EXT: FILE-OFFSET BLOCK-RANGE TOTAL FLAGS 0: [0..4095]: 258048..262143 4096 0x1000 1: [4096..6143]: 238616..240663 2048 0x1001 Kmsg: F2FS-fs (zram0): Swapfile: last extent is not aligned to section F2FS-fs (zram0): Swapfile (2) is not align to section: 1) creat(), 2) ioctl(F2FS_IOC_SET_PIN_FILE), 3) fallocate(2097152 * n) Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2021-05-26 06:29:27 +00:00
FI_ALIGNED_WRITE, /* enable aligned write */
f2fs: fix null-ptr-deref in f2fs_get_dnode_of_data There is issue as follows when test f2fs atomic write: F2FS-fs (loop0): Can't find valid F2FS filesystem in 2th superblock F2FS-fs (loop0): invalid crc_offset: 0 F2FS-fs (loop0): f2fs_check_nid_range: out-of-range nid=1, run fsck to fix. F2FS-fs (loop0): f2fs_check_nid_range: out-of-range nid=2, run fsck to fix. ================================================================== BUG: KASAN: null-ptr-deref in f2fs_get_dnode_of_data+0xac/0x16d0 Read of size 8 at addr 0000000000000028 by task rep/1990 CPU: 4 PID: 1990 Comm: rep Not tainted 5.19.0-rc6-next-20220715 #266 Call Trace: <TASK> dump_stack_lvl+0x6e/0x91 print_report.cold+0x49a/0x6bb kasan_report+0xa8/0x130 f2fs_get_dnode_of_data+0xac/0x16d0 f2fs_do_write_data_page+0x2a5/0x1030 move_data_page+0x3c5/0xdf0 do_garbage_collect+0x2015/0x36c0 f2fs_gc+0x554/0x1d30 f2fs_balance_fs+0x7f5/0xda0 f2fs_write_single_data_page+0xb66/0xdc0 f2fs_write_cache_pages+0x716/0x1420 f2fs_write_data_pages+0x84f/0x9a0 do_writepages+0x130/0x3a0 filemap_fdatawrite_wbc+0x87/0xa0 file_write_and_wait_range+0x157/0x1c0 f2fs_do_sync_file+0x206/0x12d0 f2fs_sync_file+0x99/0xc0 vfs_fsync_range+0x75/0x140 f2fs_file_write_iter+0xd7b/0x1850 vfs_write+0x645/0x780 ksys_write+0xf1/0x1e0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x63/0xcd As 3db1de0e582c commit changed atomic write way which new a cow_inode for atomic write file, and also mark cow_inode as FI_ATOMIC_FILE. When f2fs_do_write_data_page write cow_inode will use cow_inode's cow_inode which is NULL. Then will trigger null-ptr-deref. To solve above issue, introduce FI_COW_FILE flag for COW inode. Fiexes: 3db1de0e582c("f2fs: change the current atomic write way") Signed-off-by: Ye Bin <yebin10@huawei.com> Reviewed-by: Chao Yu <chao@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2022-08-01 11:26:04 +00:00
FI_COW_FILE, /* indicate COW file */
FI_ATOMIC_COMMITTED, /* indicate atomic commit completed except disk sync */
FI_ATOMIC_REPLACE, /* indicate atomic replace */
2024-05-29 10:01:03 +00:00
FI_OPENED_FILE, /* indicate file has been opened */
FI_MAX, /* max flag, never be used */
};
struct f2fs_inode_info {
struct inode vfs_inode; /* serve a vfs inode */
unsigned long i_flags; /* keep an inode flags for ioctl */
unsigned char i_advise; /* use to give file attribute hints */
f2fs: introduce large directory support This patch introduces an i_dir_level field to support large directory. Previously, f2fs maintains multi-level hash tables to find a dentry quickly from a bunch of chiild dentries in a directory, and the hash tables consist of the following tree structure as below. In Documentation/filesystems/f2fs.txt, ---------------------- A : bucket B : block N : MAX_DIR_HASH_DEPTH ---------------------- level #0 | A(2B) | level #1 | A(2B) - A(2B) | level #2 | A(2B) - A(2B) - A(2B) - A(2B) . | . . . . level #N/2 | A(2B) - A(2B) - A(2B) - A(2B) - A(2B) - ... - A(2B) . | . . . . level #N | A(4B) - A(4B) - A(4B) - A(4B) - A(4B) - ... - A(4B) But, if we can guess that a directory will handle a number of child files, we don't need to traverse the tree from level #0 to #N all the time. Since the lower level tables contain relatively small number of dentries, the miss ratio of the target dentry is likely to be high. In order to avoid that, we can configure the hash tables sparsely from level #0 like this. level #0 | A(2B) - A(2B) - A(2B) - A(2B) level #1 | A(2B) - A(2B) - A(2B) - A(2B) - A(2B) - ... - A(2B) . | . . . . level #N/2 | A(2B) - A(2B) - A(2B) - A(2B) - A(2B) - ... - A(2B) . | . . . . level #N | A(4B) - A(4B) - A(4B) - A(4B) - A(4B) - ... - A(4B) With this structure, we can skip the ineffective tree searches in lower level hash tables. This patch adds just a facility for this by introducing i_dir_level in f2fs_inode. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2014-02-27 09:20:00 +00:00
unsigned char i_dir_level; /* use for dentry level for large dir */
union {
unsigned int i_current_depth; /* only for directory depth */
unsigned short i_gc_failures; /* for gc failure statistic */
};
unsigned int i_pino; /* parent inode number */
umode_t i_acl_mode; /* keep file acl mode temporarily */
/* Use below internally in f2fs*/
unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */
struct f2fs_rwsem i_sem; /* protect fi info */
atomic_t dirty_pages; /* # of dirty pages */
f2fs_hash_t chash; /* hash value of given file name */
unsigned int clevel; /* maximum level of given file name */
struct task_struct *task; /* lookup and create consistency */
struct task_struct *cp_task; /* separate cp/wb IO stats*/
struct task_struct *wb_task; /* indicate inode is in context of writeback */
nid_t i_xattr_nid; /* node id that contains xattrs */
loff_t last_disk_size; /* lastly written file size */
f2fs: cover last_disk_size update with spinlock This change solves below hangtask issue: INFO: task kworker/u16:1:58 blocked for more than 122 seconds. Not tainted 5.6.0-rc2-00590-g9983bdae4974e #11 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. kworker/u16:1 D 0 58 2 0x00000000 Workqueue: writeback wb_workfn (flush-179:0) Backtrace: (__schedule) from [<c0913234>] (schedule+0x78/0xf4) (schedule) from [<c017ec74>] (rwsem_down_write_slowpath+0x24c/0x4c0) (rwsem_down_write_slowpath) from [<c0915f2c>] (down_write+0x6c/0x70) (down_write) from [<c0435b80>] (f2fs_write_single_data_page+0x608/0x7ac) (f2fs_write_single_data_page) from [<c0435fd8>] (f2fs_write_cache_pages+0x2b4/0x7c4) (f2fs_write_cache_pages) from [<c043682c>] (f2fs_write_data_pages+0x344/0x35c) (f2fs_write_data_pages) from [<c0267ee8>] (do_writepages+0x3c/0xd4) (do_writepages) from [<c0310cbc>] (__writeback_single_inode+0x44/0x454) (__writeback_single_inode) from [<c03112d0>] (writeback_sb_inodes+0x204/0x4b0) (writeback_sb_inodes) from [<c03115cc>] (__writeback_inodes_wb+0x50/0xe4) (__writeback_inodes_wb) from [<c03118f4>] (wb_writeback+0x294/0x338) (wb_writeback) from [<c0312dac>] (wb_workfn+0x35c/0x54c) (wb_workfn) from [<c014f2b8>] (process_one_work+0x214/0x544) (process_one_work) from [<c014f634>] (worker_thread+0x4c/0x574) (worker_thread) from [<c01564fc>] (kthread+0x144/0x170) (kthread) from [<c01010e8>] (ret_from_fork+0x14/0x2c) Reported-and-tested-by: Ondřej Jirman <megi@xff.cz> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2020-02-27 11:30:03 +00:00
spinlock_t i_size_lock; /* protect last_disk_size */
#ifdef CONFIG_QUOTA
struct dquot __rcu *i_dquot[MAXQUOTAS];
/* quota space reservation, managed internally by quota code */
qsize_t i_reserved_quota;
#endif
struct list_head dirty_list; /* dirty list for dirs and files */
struct list_head gdirty_list; /* linked in global dirty list */
struct task_struct *atomic_write_task; /* store atomic write task */
struct extent_tree *extent_tree[NR_EXTENT_CACHES];
/* cached extent_tree entry */
union {
struct inode *cow_inode; /* copy-on-write inode for atomic write */
struct inode *atomic_inode;
/* point to atomic_inode, available only for cow_inode */
};
/* avoid racing between foreground op and gc */
struct f2fs_rwsem i_gc_rwsem[2];
struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */
f2fs: enhance on-disk inode structure scalability This patch add new flag F2FS_EXTRA_ATTR storing in inode.i_inline to indicate that on-disk structure of current inode is extended. In order to extend, we changed the inode structure a bit: Original one: struct f2fs_inode { ... struct f2fs_extent i_ext; __le32 i_addr[DEF_ADDRS_PER_INODE]; __le32 i_nid[DEF_NIDS_PER_INODE]; } Extended one: struct f2fs_inode { ... struct f2fs_extent i_ext; union { struct { __le16 i_extra_isize; __le16 i_padding; __le32 i_extra_end[0]; }; __le32 i_addr[DEF_ADDRS_PER_INODE]; }; __le32 i_nid[DEF_NIDS_PER_INODE]; } Once F2FS_EXTRA_ATTR is set, we will steal four bytes in the head of i_addr field for storing i_extra_isize and i_padding. with i_extra_isize, we can calculate actual size of reserved space in i_addr, available attribute fields included in total extra attribute fields for current inode can be described as below: +--------------------+ | .i_mode | | ... | | .i_ext | +--------------------+ | .i_extra_isize |-----+ | .i_padding | | | .i_prjid | | | .i_atime_extra | | | .i_ctime_extra | | | .i_mtime_extra |<----+ | .i_inode_cs |<----- store blkaddr/inline from here | .i_xattr_cs | | ... | +--------------------+ | | | block address | | | +--------------------+ | .i_nid | +--------------------+ | node_footer | | (nid, ino, offset) | +--------------------+ Hence, with this patch, we would enhance scalability of f2fs inode for storing more newly added attribute. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-07-18 16:19:06 +00:00
int i_extra_isize; /* size of extra space located in i_addr */
kprojid_t i_projid; /* id for project quota */
f2fs: support flexible inline xattr size Now, in product, more and more features based on file encryption were introduced, their demand of xattr space is increasing, however, inline xattr has fixed-size of 200 bytes, once inline xattr space is full, new increased xattr data would occupy additional xattr block which may bring us more space usage and performance regression during persisting. In order to resolve above issue, it's better to expand inline xattr size flexibly according to user's requirement. So this patch introduces new filesystem feature 'flexible inline xattr', and new mount option 'inline_xattr_size=%u', once mkfs enables the feature, we can use the option to make f2fs supporting flexible inline xattr size. To support this feature, we add extra attribute i_inline_xattr_size in inode layout, indicating that how many space inline xattr borrows from block address mapping space in inode layout, by this, we can easily locate and store flexible-sized inline xattr data in inode. Inode disk layout: +----------------------+ | .i_mode | | ... | | .i_ext | +----------------------+ | .i_extra_isize | | .i_inline_xattr_size |-----------+ | ... | | +----------------------+ | | .i_addr | | | - block address or | | | - inline data | | +----------------------+<---+ v | inline xattr | +---inline xattr range +----------------------+<---+ | .i_nid | +----------------------+ | node_footer | | (nid, ino, offset) | +----------------------+ Note that, we have to cnosider backward compatibility which reserved inline_data space, 200 bytes, all the time, reported by Sheng Yong. Previous inline data or directory always reserved 200 bytes in inode layout, even if inline_xattr is disabled. In order to keep inline_dentry's structure for backward compatibility, we get the space back only from inline_data. Signed-off-by: Chao Yu <yuchao0@huawei.com> Reported-by: Sheng Yong <shengyong1@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-09-06 13:59:50 +00:00
int i_inline_xattr_size; /* inline xattr size */
struct timespec64 i_crtime; /* inode creation time */
struct timespec64 i_disk_time[3];/* inode disk times */
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>
2019-11-01 10:07:14 +00:00
/* for file compress */
atomic_t i_compr_blocks; /* # of compressed blocks */
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>
2019-11-01 10:07:14 +00:00
unsigned char i_compress_algorithm; /* algorithm type */
unsigned char i_log_cluster_size; /* log of cluster size */
unsigned char i_compress_level; /* compress level (lz4hc,zstd) */
unsigned char i_compress_flag; /* compress flag */
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>
2019-11-01 10:07:14 +00:00
unsigned int i_cluster_size; /* cluster size */
unsigned int atomic_write_cnt;
loff_t original_i_size; /* original i_size before atomic write */
};
static inline void get_read_extent_info(struct extent_info *ext,
struct f2fs_extent *i_ext)
{
ext->fofs = le32_to_cpu(i_ext->fofs);
ext->blk = le32_to_cpu(i_ext->blk);
ext->len = le32_to_cpu(i_ext->len);
}
static inline void set_raw_read_extent(struct extent_info *ext,
struct f2fs_extent *i_ext)
{
i_ext->fofs = cpu_to_le32(ext->fofs);
i_ext->blk = cpu_to_le32(ext->blk);
i_ext->len = cpu_to_le32(ext->len);
}
static inline bool __is_discard_mergeable(struct discard_info *back,
struct discard_info *front, unsigned int max_len)
{
return (back->lstart + back->len == front->lstart) &&
(back->len + front->len <= max_len);
}
static inline bool __is_discard_back_mergeable(struct discard_info *cur,
struct discard_info *back, unsigned int max_len)
{
return __is_discard_mergeable(back, cur, max_len);
}
static inline bool __is_discard_front_mergeable(struct discard_info *cur,
struct discard_info *front, unsigned int max_len)
{
return __is_discard_mergeable(cur, front, max_len);
}
/*
* For free nid management
*/
enum nid_state {
FREE_NID, /* newly added to free nid list */
PREALLOC_NID, /* it is preallocated */
MAX_NID_STATE,
};
enum nat_state {
TOTAL_NAT,
DIRTY_NAT,
RECLAIMABLE_NAT,
MAX_NAT_STATE,
};
struct f2fs_nm_info {
block_t nat_blkaddr; /* base disk address of NAT */
nid_t max_nid; /* maximum possible node ids */
nid_t available_nids; /* # of available node ids */
nid_t next_scan_nid; /* the next nid to be scanned */
nid_t max_rf_node_blocks; /* max # of nodes for recovery */
unsigned int ram_thresh; /* control the memory footprint */
unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */
unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */
/* NAT cache management */
struct radix_tree_root nat_root;/* root of the nat entry cache */
struct radix_tree_root nat_set_root;/* root of the nat set cache */
struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */
struct list_head nat_entries; /* cached nat entry list (clean) */
spinlock_t nat_list_lock; /* protect clean nat entry list */
unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */
unsigned int nat_blocks; /* # of nat blocks */
/* free node ids management */
struct radix_tree_root free_nid_root;/* root of the free_nid cache */
struct list_head free_nid_list; /* list for free nids excluding preallocated nids */
unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */
spinlock_t nid_list_lock; /* protect nid lists ops */
struct mutex build_lock; /* lock for build free nids */
unsigned char **free_nid_bitmap;
unsigned char *nat_block_bitmap;
unsigned short *free_nid_count; /* free nid count of NAT block */
/* for checkpoint */
char *nat_bitmap; /* NAT bitmap pointer */
unsigned int nat_bits_blocks; /* # of nat bits blocks */
unsigned char *nat_bits; /* NAT bits blocks */
unsigned char *full_nat_bits; /* full NAT pages */
unsigned char *empty_nat_bits; /* empty NAT pages */
#ifdef CONFIG_F2FS_CHECK_FS
char *nat_bitmap_mir; /* NAT bitmap mirror */
#endif
int bitmap_size; /* bitmap size */
};
/*
* this structure is used as one of function parameters.
* all the information are dedicated to a given direct node block determined
* by the data offset in a file.
*/
struct dnode_of_data {
struct inode *inode; /* vfs inode pointer */
struct page *inode_page; /* its inode page, NULL is possible */
struct page *node_page; /* cached direct node page */
nid_t nid; /* node id of the direct node block */
unsigned int ofs_in_node; /* data offset in the node page */
bool inode_page_locked; /* inode page is locked or not */
bool node_changed; /* is node block changed */
char cur_level; /* level of hole node page */
char max_level; /* level of current page located */
block_t data_blkaddr; /* block address of the node block */
};
static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode,
struct page *ipage, struct page *npage, nid_t nid)
{
memset(dn, 0, sizeof(*dn));
dn->inode = inode;
dn->inode_page = ipage;
dn->node_page = npage;
dn->nid = nid;
}
/*
* For SIT manager
*
* By default, there are 6 active log areas across the whole main area.
* When considering hot and cold data separation to reduce cleaning overhead,
* we split 3 for data logs and 3 for node logs as hot, warm, and cold types,
* respectively.
* In the current design, you should not change the numbers intentionally.
* Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6
* logs individually according to the underlying devices. (default: 6)
* Just in case, on-disk layout covers maximum 16 logs that consist of 8 for
* data and 8 for node logs.
*/
#define NR_CURSEG_DATA_TYPE (3)
#define NR_CURSEG_NODE_TYPE (3)
f2fs: support age threshold based garbage collection There are several issues in current background GC algorithm: - valid blocks is one of key factors during cost overhead calculation, so if segment has less valid block, however even its age is young or it locates hot segment, CB algorithm will still choose the segment as victim, it's not appropriate. - GCed data/node will go to existing logs, no matter in-there datas' update frequency is the same or not, it may mix hot and cold data again. - GC alloctor mainly use LFS type segment, it will cost free segment more quickly. This patch introduces a new algorithm named age threshold based garbage collection to solve above issues, there are three steps mainly: 1. select a source victim: - set an age threshold, and select candidates beased threshold: e.g. 0 means youngest, 100 means oldest, if we set age threshold to 80 then select dirty segments which has age in range of [80, 100] as candiddates; - set candidate_ratio threshold, and select candidates based the ratio, so that we can shrink candidates to those oldest segments; - select target segment with fewest valid blocks in order to migrate blocks with minimum cost; 2. select a target victim: - select candidates beased age threshold; - set candidate_radius threshold, search candidates whose age is around source victims, searching radius should less than the radius threshold. - select target segment with most valid blocks in order to avoid migrating current target segment. 3. merge valid blocks from source victim into target victim with SSR alloctor. Test steps: - create 160 dirty segments: * half of them have 128 valid blocks per segment * left of them have 384 valid blocks per segment - run background GC Benefit: GC count and block movement count both decrease obviously: - Before: - Valid: 86 - Dirty: 1 - Prefree: 11 - Free: 6001 (6001) GC calls: 162 (BG: 220) - data segments : 160 (160) - node segments : 2 (2) Try to move 41454 blocks (BG: 41454) - data blocks : 40960 (40960) - node blocks : 494 (494) IPU: 0 blocks SSR: 0 blocks in 0 segments LFS: 41364 blocks in 81 segments - After: - Valid: 87 - Dirty: 0 - Prefree: 4 - Free: 6008 (6008) GC calls: 75 (BG: 76) - data segments : 74 (74) - node segments : 1 (1) Try to move 12813 blocks (BG: 12813) - data blocks : 12544 (12544) - node blocks : 269 (269) IPU: 0 blocks SSR: 12032 blocks in 77 segments LFS: 855 blocks in 2 segments Signed-off-by: Chao Yu <yuchao0@huawei.com> [Jaegeuk Kim: fix a bug along with pinfile in-mem segment & clean up] Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2020-08-04 13:14:49 +00:00
#define NR_CURSEG_INMEM_TYPE (2)
#define NR_CURSEG_RO_TYPE (2)
#define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE)
#define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE)
enum {
CURSEG_HOT_DATA = 0, /* directory entry blocks */
CURSEG_WARM_DATA, /* data blocks */
CURSEG_COLD_DATA, /* multimedia or GCed data blocks */
CURSEG_HOT_NODE, /* direct node blocks of directory files */
CURSEG_WARM_NODE, /* direct node blocks of normal files */
CURSEG_COLD_NODE, /* indirect node blocks */
NR_PERSISTENT_LOG, /* number of persistent log */
CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG,
/* pinned file that needs consecutive block address */
f2fs: support age threshold based garbage collection There are several issues in current background GC algorithm: - valid blocks is one of key factors during cost overhead calculation, so if segment has less valid block, however even its age is young or it locates hot segment, CB algorithm will still choose the segment as victim, it's not appropriate. - GCed data/node will go to existing logs, no matter in-there datas' update frequency is the same or not, it may mix hot and cold data again. - GC alloctor mainly use LFS type segment, it will cost free segment more quickly. This patch introduces a new algorithm named age threshold based garbage collection to solve above issues, there are three steps mainly: 1. select a source victim: - set an age threshold, and select candidates beased threshold: e.g. 0 means youngest, 100 means oldest, if we set age threshold to 80 then select dirty segments which has age in range of [80, 100] as candiddates; - set candidate_ratio threshold, and select candidates based the ratio, so that we can shrink candidates to those oldest segments; - select target segment with fewest valid blocks in order to migrate blocks with minimum cost; 2. select a target victim: - select candidates beased age threshold; - set candidate_radius threshold, search candidates whose age is around source victims, searching radius should less than the radius threshold. - select target segment with most valid blocks in order to avoid migrating current target segment. 3. merge valid blocks from source victim into target victim with SSR alloctor. Test steps: - create 160 dirty segments: * half of them have 128 valid blocks per segment * left of them have 384 valid blocks per segment - run background GC Benefit: GC count and block movement count both decrease obviously: - Before: - Valid: 86 - Dirty: 1 - Prefree: 11 - Free: 6001 (6001) GC calls: 162 (BG: 220) - data segments : 160 (160) - node segments : 2 (2) Try to move 41454 blocks (BG: 41454) - data blocks : 40960 (40960) - node blocks : 494 (494) IPU: 0 blocks SSR: 0 blocks in 0 segments LFS: 41364 blocks in 81 segments - After: - Valid: 87 - Dirty: 0 - Prefree: 4 - Free: 6008 (6008) GC calls: 75 (BG: 76) - data segments : 74 (74) - node segments : 1 (1) Try to move 12813 blocks (BG: 12813) - data blocks : 12544 (12544) - node blocks : 269 (269) IPU: 0 blocks SSR: 12032 blocks in 77 segments LFS: 855 blocks in 2 segments Signed-off-by: Chao Yu <yuchao0@huawei.com> [Jaegeuk Kim: fix a bug along with pinfile in-mem segment & clean up] Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2020-08-04 13:14:49 +00:00
CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */
NO_CHECK_TYPE, /* number of persistent & inmem log */
};
struct flush_cmd {
struct completion wait;
struct llist_node llnode;
nid_t ino;
int ret;
};
struct flush_cmd_control {
struct task_struct *f2fs_issue_flush; /* flush thread */
wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */
atomic_t issued_flush; /* # of issued flushes */
atomic_t queued_flush; /* # of queued flushes */
struct llist_head issue_list; /* list for command issue */
struct llist_node *dispatch_list; /* list for command dispatch */
};
struct f2fs_sm_info {
struct sit_info *sit_info; /* whole segment information */
struct free_segmap_info *free_info; /* free segment information */
struct dirty_seglist_info *dirty_info; /* dirty segment information */
struct curseg_info *curseg_array; /* active segment information */
struct f2fs_rwsem curseg_lock; /* for preventing curseg change */
block_t seg0_blkaddr; /* block address of 0'th segment */
block_t main_blkaddr; /* start block address of main area */
block_t ssa_blkaddr; /* start block address of SSA area */
unsigned int segment_count; /* total # of segments */
unsigned int main_segments; /* # of segments in main area */
unsigned int reserved_segments; /* # of reserved segments */
f2fs: fix to reserve space for IO align feature https://bugzilla.kernel.org/show_bug.cgi?id=204137 With below script, we will hit panic during new segment allocation: DISK=bingo.img MOUNT_DIR=/mnt/f2fs dd if=/dev/zero of=$DISK bs=1M count=105 mkfs.f2fe -a 1 -o 19 -t 1 -z 1 -f -q $DISK mount -t f2fs $DISK $MOUNT_DIR -o "noinline_dentry,flush_merge,noextent_cache,mode=lfs,io_bits=7,fsync_mode=strict" for (( i = 0; i < 4096; i++ )); do name=`head /dev/urandom | tr -dc A-Za-z0-9 | head -c 10` mkdir $MOUNT_DIR/$name done umount $MOUNT_DIR rm $DISK --- Core dump --- Call Trace: allocate_segment_by_default+0x9d/0x100 [f2fs] f2fs_allocate_data_block+0x3c0/0x5c0 [f2fs] do_write_page+0x62/0x110 [f2fs] f2fs_outplace_write_data+0x43/0xc0 [f2fs] f2fs_do_write_data_page+0x386/0x560 [f2fs] __write_data_page+0x706/0x850 [f2fs] f2fs_write_cache_pages+0x267/0x6a0 [f2fs] f2fs_write_data_pages+0x19c/0x2e0 [f2fs] do_writepages+0x1c/0x70 __filemap_fdatawrite_range+0xaa/0xe0 filemap_fdatawrite+0x1f/0x30 f2fs_sync_dirty_inodes+0x74/0x1f0 [f2fs] block_operations+0xdc/0x350 [f2fs] f2fs_write_checkpoint+0x104/0x1150 [f2fs] f2fs_sync_fs+0xa2/0x120 [f2fs] f2fs_balance_fs_bg+0x33c/0x390 [f2fs] f2fs_write_node_pages+0x4c/0x1f0 [f2fs] do_writepages+0x1c/0x70 __writeback_single_inode+0x45/0x320 writeback_sb_inodes+0x273/0x5c0 wb_writeback+0xff/0x2e0 wb_workfn+0xa1/0x370 process_one_work+0x138/0x350 worker_thread+0x4d/0x3d0 kthread+0x109/0x140 ret_from_fork+0x25/0x30 The root cause here is, with IO alignment feature enables, in worst case, we need F2FS_IO_SIZE() free blocks space for single one 4k write due to IO alignment feature will fill dummy pages to make IO being aligned. So we will easily run out of free segments during non-inline directory's data writeback, even in process of foreground GC. In order to fix this issue, I just propose to reserve additional free space for IO alignment feature to handle worst case of free space usage ratio during FGGC. Fixes: 0a595ebaaa6b ("f2fs: support IO alignment for DATA and NODE writes") Signed-off-by: Chao Yu <chao@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2021-12-11 13:27:36 +00:00
unsigned int additional_reserved_segments;/* reserved segs for IO align feature */
unsigned int ovp_segments; /* # of overprovision segments */
/* a threshold to reclaim prefree segments */
unsigned int rec_prefree_segments;
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
struct list_head sit_entry_set; /* sit entry set list */
unsigned int ipu_policy; /* in-place-update policy */
unsigned int min_ipu_util; /* in-place-update threshold */
unsigned int min_fsync_blocks; /* threshold for fsync */
unsigned int min_seq_blocks; /* threshold for sequential blocks */
unsigned int min_hot_blocks; /* threshold for hot block allocation */
unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */
/* for flush command control */
struct flush_cmd_control *fcc_info;
/* for discard command control */
struct discard_cmd_control *dcc_info;
};
/*
* For superblock
*/
/*
* COUNT_TYPE for monitoring
*
* f2fs monitors the number of several block types such as on-writeback,
* dirty dentry blocks, dirty node blocks, and dirty meta blocks.
*/
#define WB_DATA_TYPE(p, f) \
(f || f2fs_is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA)
enum count_type {
F2FS_DIRTY_DENTS,
F2FS_DIRTY_DATA,
F2FS_DIRTY_QDATA,
F2FS_DIRTY_NODES,
F2FS_DIRTY_META,
F2FS_DIRTY_IMETA,
F2FS_WB_CP_DATA,
F2FS_WB_DATA,
F2FS_RD_DATA,
F2FS_RD_NODE,
F2FS_RD_META,
F2FS_DIO_WRITE,
F2FS_DIO_READ,
NR_COUNT_TYPE,
};
/*
* The below are the page types of bios used in submit_bio().
* The available types are:
* DATA User data pages. It operates as async mode.
* NODE Node pages. It operates as async mode.
* META FS metadata pages such as SIT, NAT, CP.
* NR_PAGE_TYPE The number of page types.
* META_FLUSH Make sure the previous pages are written
* with waiting the bio's completion
* ... Only can be used with META.
*/
#define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type))
#define PAGE_TYPE_ON_MAIN(type) ((type) == DATA || (type) == NODE)
enum page_type {
2022-05-06 01:33:06 +00:00
DATA = 0,
NODE = 1, /* should not change this */
META,
NR_PAGE_TYPE,
META_FLUSH,
IPU, /* the below types are used by tracepoints only. */
OPU,
};
enum temp_type {
HOT = 0, /* must be zero for meta bio */
WARM,
COLD,
NR_TEMP_TYPE,
};
enum need_lock_type {
LOCK_REQ = 0,
LOCK_DONE,
LOCK_RETRY,
};
enum cp_reason_type {
CP_NO_NEEDED,
CP_NON_REGULAR,
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>
2019-11-01 10:07:14 +00:00
CP_COMPRESSED,
CP_HARDLINK,
CP_SB_NEED_CP,
CP_WRONG_PINO,
CP_NO_SPC_ROLL,
CP_NODE_NEED_CP,
CP_FASTBOOT_MODE,
CP_SPEC_LOG_NUM,
CP_RECOVER_DIR,
CP_XATTR_DIR,
};
enum iostat_type {
/* WRITE IO */
APP_DIRECT_IO, /* app direct write IOs */
APP_BUFFERED_IO, /* app buffered write IOs */
APP_WRITE_IO, /* app write IOs */
APP_MAPPED_IO, /* app mapped IOs */
APP_BUFFERED_CDATA_IO, /* app buffered write IOs on compressed file */
APP_MAPPED_CDATA_IO, /* app mapped write IOs on compressed file */
FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */
FS_CDATA_IO, /* data IOs from kworker/fsync/reclaimer on compressed file */
FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */
FS_META_IO, /* meta IOs from kworker/reclaimer */
FS_GC_DATA_IO, /* data IOs from forground gc */
FS_GC_NODE_IO, /* node IOs from forground gc */
FS_CP_DATA_IO, /* data IOs from checkpoint */
FS_CP_NODE_IO, /* node IOs from checkpoint */
FS_CP_META_IO, /* meta IOs from checkpoint */
/* READ IO */
APP_DIRECT_READ_IO, /* app direct read IOs */
APP_BUFFERED_READ_IO, /* app buffered read IOs */
APP_READ_IO, /* app read IOs */
APP_MAPPED_READ_IO, /* app mapped read IOs */
APP_BUFFERED_CDATA_READ_IO, /* app buffered read IOs on compressed file */
APP_MAPPED_CDATA_READ_IO, /* app mapped read IOs on compressed file */
FS_DATA_READ_IO, /* data read IOs */
FS_GDATA_READ_IO, /* data read IOs from background gc */
FS_CDATA_READ_IO, /* compressed data read IOs */
FS_NODE_READ_IO, /* node read IOs */
FS_META_READ_IO, /* meta read IOs */
/* other */
FS_DISCARD_IO, /* discard */
FS_FLUSH_IO, /* flush */
FS_ZONE_RESET_IO, /* zone reset */
NR_IO_TYPE,
};
struct f2fs_io_info {
struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */
nid_t ino; /* inode number */
enum page_type type; /* contains DATA/NODE/META/META_FLUSH */
enum temp_type temp; /* contains HOT/WARM/COLD */
enum req_op op; /* contains REQ_OP_ */
blk_opf_t op_flags; /* req_flag_bits */
block_t new_blkaddr; /* new block address to be written */
block_t old_blkaddr; /* old block address before Cow */
struct page *page; /* page to be written */
struct page *encrypted_page; /* encrypted page */
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>
2019-11-01 10:07:14 +00:00
struct page *compressed_page; /* compressed page */
struct list_head list; /* serialize IOs */
unsigned int compr_blocks; /* # of compressed block addresses */
unsigned int need_lock:8; /* indicate we need to lock cp_rwsem */
unsigned int version:8; /* version of the node */
unsigned int submitted:1; /* indicate IO submission */
unsigned int in_list:1; /* indicate fio is in io_list */
unsigned int is_por:1; /* indicate IO is from recovery or not */
unsigned int encrypted:1; /* indicate file is encrypted */
unsigned int meta_gc:1; /* require meta inode GC */
enum iostat_type io_type; /* io type */
struct writeback_control *io_wbc; /* writeback control */
f2fs: add bio cache for IPU SQLite in Wal mode may trigger sequential IPU write in db-wal file, after commit d1b3e72d5490 ("f2fs: submit bio of in-place-update pages"), we lost the chance of merging page in inner managed bio cache, result in submitting more small-sized IO. So let's add temporary bio in writepages() to cache mergeable write IO as much as possible. Test case: 1. xfs_io -f /mnt/f2fs/file -c "pwrite 0 65536" -c "fsync" 2. xfs_io -f /mnt/f2fs/file -c "pwrite 0 65536" -c "fsync" Before: f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65544, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65552, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65560, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65568, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65576, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65584, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65592, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65600, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65608, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65616, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65624, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65632, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65640, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65648, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65656, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65664, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), NODE, sector = 57352, size = 4096 After: f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65544, size = 65536 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), NODE, sector = 57368, size = 4096 Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-02-19 08:15:29 +00:00
struct bio **bio; /* bio for ipu */
sector_t *last_block; /* last block number in bio */
};
struct bio_entry {
struct bio *bio;
struct list_head list;
};
#define is_read_io(rw) ((rw) == READ)
struct f2fs_bio_info {
struct f2fs_sb_info *sbi; /* f2fs superblock */
struct bio *bio; /* bios to merge */
sector_t last_block_in_bio; /* last block number */
struct f2fs_io_info fio; /* store buffered io info. */
#ifdef CONFIG_BLK_DEV_ZONED
struct completion zone_wait; /* condition value for the previous open zone to close */
struct bio *zone_pending_bio; /* pending bio for the previous zone */
void *bi_private; /* previous bi_private for pending bio */
#endif
struct f2fs_rwsem io_rwsem; /* blocking op for bio */
spinlock_t io_lock; /* serialize DATA/NODE IOs */
struct list_head io_list; /* track fios */
struct list_head bio_list; /* bio entry list head */
struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */
};
#define FDEV(i) (sbi->devs[i])
#define RDEV(i) (raw_super->devs[i])
struct f2fs_dev_info {
struct file *bdev_file;
struct block_device *bdev;
char path[MAX_PATH_LEN];
unsigned int total_segments;
block_t start_blk;
block_t end_blk;
#ifdef CONFIG_BLK_DEV_ZONED
unsigned int nr_blkz; /* Total number of zones */
unsigned long *blkz_seq; /* Bitmap indicating sequential zones */
#endif
};
enum inode_type {
DIR_INODE, /* for dirty dir inode */
FILE_INODE, /* for dirty regular/symlink inode */
DIRTY_META, /* for all dirtied inode metadata */
NR_INODE_TYPE,
};
/* for inner inode cache management */
struct inode_management {
struct radix_tree_root ino_root; /* ino entry array */
spinlock_t ino_lock; /* for ino entry lock */
struct list_head ino_list; /* inode list head */
unsigned long ino_num; /* number of entries */
};
f2fs: support age threshold based garbage collection There are several issues in current background GC algorithm: - valid blocks is one of key factors during cost overhead calculation, so if segment has less valid block, however even its age is young or it locates hot segment, CB algorithm will still choose the segment as victim, it's not appropriate. - GCed data/node will go to existing logs, no matter in-there datas' update frequency is the same or not, it may mix hot and cold data again. - GC alloctor mainly use LFS type segment, it will cost free segment more quickly. This patch introduces a new algorithm named age threshold based garbage collection to solve above issues, there are three steps mainly: 1. select a source victim: - set an age threshold, and select candidates beased threshold: e.g. 0 means youngest, 100 means oldest, if we set age threshold to 80 then select dirty segments which has age in range of [80, 100] as candiddates; - set candidate_ratio threshold, and select candidates based the ratio, so that we can shrink candidates to those oldest segments; - select target segment with fewest valid blocks in order to migrate blocks with minimum cost; 2. select a target victim: - select candidates beased age threshold; - set candidate_radius threshold, search candidates whose age is around source victims, searching radius should less than the radius threshold. - select target segment with most valid blocks in order to avoid migrating current target segment. 3. merge valid blocks from source victim into target victim with SSR alloctor. Test steps: - create 160 dirty segments: * half of them have 128 valid blocks per segment * left of them have 384 valid blocks per segment - run background GC Benefit: GC count and block movement count both decrease obviously: - Before: - Valid: 86 - Dirty: 1 - Prefree: 11 - Free: 6001 (6001) GC calls: 162 (BG: 220) - data segments : 160 (160) - node segments : 2 (2) Try to move 41454 blocks (BG: 41454) - data blocks : 40960 (40960) - node blocks : 494 (494) IPU: 0 blocks SSR: 0 blocks in 0 segments LFS: 41364 blocks in 81 segments - After: - Valid: 87 - Dirty: 0 - Prefree: 4 - Free: 6008 (6008) GC calls: 75 (BG: 76) - data segments : 74 (74) - node segments : 1 (1) Try to move 12813 blocks (BG: 12813) - data blocks : 12544 (12544) - node blocks : 269 (269) IPU: 0 blocks SSR: 12032 blocks in 77 segments LFS: 855 blocks in 2 segments Signed-off-by: Chao Yu <yuchao0@huawei.com> [Jaegeuk Kim: fix a bug along with pinfile in-mem segment & clean up] Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2020-08-04 13:14:49 +00:00
/* for GC_AT */
struct atgc_management {
bool atgc_enabled; /* ATGC is enabled or not */
struct rb_root_cached root; /* root of victim rb-tree */
struct list_head victim_list; /* linked with all victim entries */
unsigned int victim_count; /* victim count in rb-tree */
unsigned int candidate_ratio; /* candidate ratio */
unsigned int max_candidate_count; /* max candidate count */
unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */
unsigned long long age_threshold; /* age threshold */
};
struct f2fs_gc_control {
unsigned int victim_segno; /* target victim segment number */
int init_gc_type; /* FG_GC or BG_GC */
bool no_bg_gc; /* check the space and stop bg_gc */
bool should_migrate_blocks; /* should migrate blocks */
bool err_gc_skipped; /* return EAGAIN if GC skipped */
unsigned int nr_free_secs; /* # of free sections to do GC */
};
/*
* For s_flag in struct f2fs_sb_info
* Modification on enum should be synchronized with s_flag array
*/
enum {
SBI_IS_DIRTY, /* dirty flag for checkpoint */
SBI_IS_CLOSE, /* specify unmounting */
SBI_NEED_FSCK, /* need fsck.f2fs to fix */
SBI_POR_DOING, /* recovery is doing or not */
SBI_NEED_SB_WRITE, /* need to recover superblock */
SBI_NEED_CP, /* need to checkpoint */
SBI_IS_SHUTDOWN, /* shutdown by ioctl */
f2fs: fix to flush all dirty inodes recovered in readonly fs generic/417 reported as blow: ------------[ cut here ]------------ kernel BUG at /home/yuchao/git/devf2fs/inode.c:695! invalid opcode: 0000 [#1] PREEMPT SMP CPU: 1 PID: 21697 Comm: umount Tainted: G W O 4.18.0-rc2+ #39 Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006 EIP: f2fs_evict_inode+0x556/0x580 [f2fs] Call Trace: ? _raw_spin_unlock+0x2c/0x50 evict+0xa8/0x170 dispose_list+0x34/0x40 evict_inodes+0x118/0x120 generic_shutdown_super+0x41/0x100 ? rcu_read_lock_sched_held+0x97/0xa0 kill_block_super+0x22/0x50 kill_f2fs_super+0x6f/0x80 [f2fs] deactivate_locked_super+0x3d/0x70 deactivate_super+0x40/0x60 cleanup_mnt+0x39/0x70 __cleanup_mnt+0x10/0x20 task_work_run+0x81/0xa0 exit_to_usermode_loop+0x59/0xa7 do_fast_syscall_32+0x1f5/0x22c entry_SYSENTER_32+0x53/0x86 EIP: f2fs_evict_inode+0x556/0x580 [f2fs] It can simply reproduced with scripts: Enable quota feature during mkfs. Testcase1: 1. mkfs.f2fs /dev/zram0 2. mount -t f2fs /dev/zram0 /mnt/f2fs 3. xfs_io -f /mnt/f2fs/file -c "pwrite 0 4k" -c "fsync" 4. godown /mnt/f2fs 5. umount /mnt/f2fs 6. mount -t f2fs -o ro /dev/zram0 /mnt/f2fs 7. umount /mnt/f2fs Testcase2: 1. mkfs.f2fs /dev/zram0 2. mount -t f2fs /dev/zram0 /mnt/f2fs 3. touch /mnt/f2fs/file 4. create process[pid = x] do: a) open /mnt/f2fs/file; b) unlink /mnt/f2fs/file 5. godown -f /mnt/f2fs 6. kill process[pid = x] 7. umount /mnt/f2fs 8. mount -t f2fs -o ro /dev/zram0 /mnt/f2fs 9. umount /mnt/f2fs The reason is: during recovery, i_{c,m}time of inode will be updated, then the inode can be set dirty w/o being tracked in sbi->inode_list[DIRTY_META] global list, so later write_checkpoint will not flush such dirty inode into node page. Once umount is called, sync_filesystem() in generic_shutdown_super() will skip syncng dirty inodes due to sb_rdonly check, leaving dirty inodes there. To solve this issue, during umount, add remove SB_RDONLY flag in sb->s_flags, to make sure sync_filesystem() will not be skipped. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-22 09:11:05 +00:00
SBI_IS_RECOVERED, /* recovered orphan/data */
SBI_CP_DISABLED, /* CP was disabled last mount */
SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */
SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */
SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */
SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */
f2fs: ioctl for removing a range from F2FS This ioctl shrinks a given length (aligned to sections) from end of the main area. Any cursegs and valid blocks will be moved out before invalidating the range. This feature can be used for adjusting partition sizes online. History of the patch: Sahitya Tummala: - Add this ioctl for f2fs_compat_ioctl() as well. - Fix debugfs status to reflect the online resize changes. - Fix potential race between online resize path and allocate new data block path or gc path. Others: - Rename some identifiers. - Add some error handling branches. - Clear sbi->next_victim_seg[BG_GC/FG_GC] in shrinking range. - Implement this interface as ext4's, and change the parameter from shrunk bytes to new block count of F2FS. - During resizing, force to empty sit_journal and forbid adding new entries to it, in order to avoid invalid segno in journal after resize. - Reduce sbi->user_block_count before resize starts. - Commit the updated superblock first, and then update in-memory metadata only when the former succeeds. - Target block count must align to sections. - Write checkpoint before and after committing the new superblock, w/o CP_FSCK_FLAG respectively, so that the FS can be fixed by fsck even if resize fails after the new superblock is committed. - In free_segment_range(), reduce granularity of gc_mutex. - Add protection on curseg migration. - Add freeze_bdev() and thaw_bdev() for resize fs. - Remove CUR_MAIN_SECS and use MAIN_SECS directly for allocation. - Recover super_block and FS metadata when resize fails. - No need to clear CP_FSCK_FLAG in update_ckpt_flags(). - Clean up the sb and fs metadata update functions for resize_fs. Geert Uytterhoeven: - Use div_u64*() for 64-bit divisions Arnd Bergmann: - Not all architectures support get_user() with a 64-bit argument: ERROR: "__get_user_bad" [fs/f2fs/f2fs.ko] undefined! Use copy_from_user() here, this will always work. Signed-off-by: Qiuyang Sun <sunqiuyang@huawei.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Sahitya Tummala <stummala@codeaurora.org> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-06-05 03:33:25 +00:00
SBI_IS_RESIZEFS, /* resizefs is in process */
SBI_IS_FREEZING, /* freezefs is in process */
SBI_IS_WRITABLE, /* remove ro mountoption transiently */
MAX_SBI_FLAG,
};
enum {
CP_TIME,
REQ_TIME,
DISCARD_TIME,
GC_TIME,
DISABLE_TIME,
UMOUNT_DISCARD_TIMEOUT,
MAX_TIME,
};
/* Note that you need to keep synchronization with this gc_mode_names array */
enum {
GC_NORMAL,
GC_IDLE_CB,
GC_IDLE_GREEDY,
f2fs: support age threshold based garbage collection There are several issues in current background GC algorithm: - valid blocks is one of key factors during cost overhead calculation, so if segment has less valid block, however even its age is young or it locates hot segment, CB algorithm will still choose the segment as victim, it's not appropriate. - GCed data/node will go to existing logs, no matter in-there datas' update frequency is the same or not, it may mix hot and cold data again. - GC alloctor mainly use LFS type segment, it will cost free segment more quickly. This patch introduces a new algorithm named age threshold based garbage collection to solve above issues, there are three steps mainly: 1. select a source victim: - set an age threshold, and select candidates beased threshold: e.g. 0 means youngest, 100 means oldest, if we set age threshold to 80 then select dirty segments which has age in range of [80, 100] as candiddates; - set candidate_ratio threshold, and select candidates based the ratio, so that we can shrink candidates to those oldest segments; - select target segment with fewest valid blocks in order to migrate blocks with minimum cost; 2. select a target victim: - select candidates beased age threshold; - set candidate_radius threshold, search candidates whose age is around source victims, searching radius should less than the radius threshold. - select target segment with most valid blocks in order to avoid migrating current target segment. 3. merge valid blocks from source victim into target victim with SSR alloctor. Test steps: - create 160 dirty segments: * half of them have 128 valid blocks per segment * left of them have 384 valid blocks per segment - run background GC Benefit: GC count and block movement count both decrease obviously: - Before: - Valid: 86 - Dirty: 1 - Prefree: 11 - Free: 6001 (6001) GC calls: 162 (BG: 220) - data segments : 160 (160) - node segments : 2 (2) Try to move 41454 blocks (BG: 41454) - data blocks : 40960 (40960) - node blocks : 494 (494) IPU: 0 blocks SSR: 0 blocks in 0 segments LFS: 41364 blocks in 81 segments - After: - Valid: 87 - Dirty: 0 - Prefree: 4 - Free: 6008 (6008) GC calls: 75 (BG: 76) - data segments : 74 (74) - node segments : 1 (1) Try to move 12813 blocks (BG: 12813) - data blocks : 12544 (12544) - node blocks : 269 (269) IPU: 0 blocks SSR: 12032 blocks in 77 segments LFS: 855 blocks in 2 segments Signed-off-by: Chao Yu <yuchao0@huawei.com> [Jaegeuk Kim: fix a bug along with pinfile in-mem segment & clean up] Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2020-08-04 13:14:49 +00:00
GC_IDLE_AT,
GC_URGENT_HIGH,
GC_URGENT_LOW,
GC_URGENT_MID,
MAX_GC_MODE,
};
enum {
BGGC_MODE_ON, /* background gc is on */
BGGC_MODE_OFF, /* background gc is off */
BGGC_MODE_SYNC, /*
* background gc is on, migrating blocks
* like foreground gc
*/
};
enum {
FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */
FS_MODE_LFS, /* use lfs allocation only */
FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */
FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */
};
enum {
ALLOC_MODE_DEFAULT, /* stay default */
ALLOC_MODE_REUSE, /* reuse segments as much as possible */
};
enum fsync_mode {
FSYNC_MODE_POSIX, /* fsync follows posix semantics */
FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */
FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */
};
enum {
COMPR_MODE_FS, /*
* automatically compress compression
* enabled files
*/
COMPR_MODE_USER, /*
* automatical compression is disabled.
* user can control the file compression
* using ioctls
*/
};
f2fs: introduce discard_unit mount option As James Z reported in bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=213877 [1.] One-line summary of the problem: Mount multiple SMR block devices exceed certain number cause system non-response [2.] Full description of the problem/report: Created some F2FS on SMR devices (mkfs.f2fs -m), then mounted in sequence. Each device is the same Model: HGST HSH721414AL (Size 14TB). Empirically, found that when the amount of SMR device * 1.5Gb > System RAM, the system ran out of memory and hung. No dmesg output. For example, 24 SMR Disk need 24*1.5GB = 36GB. A system with 32G RAM can only mount 21 devices, the 22nd device will be a reproducible cause of system hang. The number of SMR devices with other FS mounted on this system does not interfere with the result above. [3.] Keywords (i.e., modules, networking, kernel): F2FS, SMR, Memory [4.] Kernel information [4.1.] Kernel version (uname -a): Linux 5.13.4-200.fc34.x86_64 #1 SMP Tue Jul 20 20:27:29 UTC 2021 x86_64 x86_64 x86_64 GNU/Linux [4.2.] Kernel .config file: Default Fedora 34 with f2fs-tools-1.14.0-2.fc34.x86_64 [5.] Most recent kernel version which did not have the bug: None [6.] Output of Oops.. message (if applicable) with symbolic information resolved (see Documentation/admin-guide/oops-tracing.rst) None [7.] A small shell script or example program which triggers the problem (if possible) mount /dev/sdX /mnt/0X [8.] Memory consumption With 24 * 14T SMR Block device with F2FS free -g total used free shared buff/cache available Mem: 46 36 0 0 10 10 Swap: 0 0 0 With 3 * 14T SMR Block device with F2FS free -g total used free shared buff/cache available Mem: 7 5 0 0 1 1 Swap: 7 0 7 The root cause is, there are three bitmaps: - cur_valid_map - ckpt_valid_map - discard_map and each of them will cost ~500MB memory, {cur, ckpt}_valid_map are necessary, but discard_map is optional, since this bitmap will only be useful in mountpoint that small discard is enabled. For a blkzoned device such as SMR or ZNS devices, f2fs will only issue discard for a section(zone) when all blocks of that section are invalid, so, for such device, we don't need small discard functionality at all. This patch introduces a new mountoption "discard_unit=block|segment| section" to support issuing discard with different basic unit which is aligned to block, segment or section, so that user can specify "discard_unit=segment" or "discard_unit=section" to disable small discard functionality. Note that this mount option can not be changed by remount() due to related metadata need to be initialized during mount(). In order to save memory, let's use "discard_unit=section" for blkzoned device by default. Signed-off-by: Chao Yu <chao@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2021-08-03 00:15:43 +00:00
enum {
DISCARD_UNIT_BLOCK, /* basic discard unit is block */
DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */
DISCARD_UNIT_SECTION, /* basic discard unit is section */
};
enum {
MEMORY_MODE_NORMAL, /* memory mode for normal devices */
MEMORY_MODE_LOW, /* memory mode for low memry devices */
};
enum errors_option {
MOUNT_ERRORS_READONLY, /* remount fs ro on errors */
MOUNT_ERRORS_CONTINUE, /* continue on errors */
MOUNT_ERRORS_PANIC, /* panic on errors */
};
enum {
BACKGROUND,
FOREGROUND,
MAX_CALL_TYPE,
TOTAL_CALL = FOREGROUND,
};
f2fs: restructure f2fs page.private layout Restruct f2fs page private layout for below reasons: There are some cases that f2fs wants to set a flag in a page to indicate a specified status of page: a) page is in transaction list for atomic write b) page contains dummy data for aligned write c) page is migrating for GC d) page contains inline data for inline inode flush e) page belongs to merkle tree, and is verified for fsverity f) page is dirty and has filesystem/inode reference count for writeback g) page is temporary and has decompress io context reference for compression There are existed places in page structure we can use to store f2fs private status/data: - page.flags: PG_checked, PG_private - page.private However it was a mess when we using them, which may cause potential confliction: page.private PG_private PG_checked page._refcount (+1 at most) a) -1 set +1 b) -2 set c), d), e) set f) 0 set +1 g) pointer set The other problem is page.flags has no free slot, if we can avoid set zero to page.private and set PG_private flag, then we use non-zero value to indicate PG_private status, so that we may have chance to reclaim PG_private slot for other usage. [1] The other concern is f2fs has bad scalability in aspect of indicating more page status. So in this patch, let's restructure f2fs' page.private as below to solve above issues: Layout A: lowest bit should be 1 | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | bit 0 PAGE_PRIVATE_NOT_POINTER bit 1 PAGE_PRIVATE_ATOMIC_WRITE bit 2 PAGE_PRIVATE_DUMMY_WRITE bit 3 PAGE_PRIVATE_ONGOING_MIGRATION bit 4 PAGE_PRIVATE_INLINE_INODE bit 5 PAGE_PRIVATE_REF_RESOURCE bit 6- f2fs private data Layout B: lowest bit should be 0 page.private is a wrapped pointer. After the change: page.private PG_private PG_checked page._refcount (+1 at most) a) 11 set +1 b) 101 set +1 c) 1001 set +1 d) 10001 set +1 e) set f) 100001 set +1 g) pointer set +1 [1] https://lore.kernel.org/linux-f2fs-devel/20210422154705.GO3596236@casper.infradead.org/T/#u Cc: Matthew Wilcox <willy@infradead.org> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2021-04-28 09:20:31 +00:00
static inline int f2fs_test_bit(unsigned int nr, char *addr);
static inline void f2fs_set_bit(unsigned int nr, char *addr);
static inline void f2fs_clear_bit(unsigned int nr, char *addr);
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>
2019-11-01 10:07:14 +00:00
/*
f2fs: restructure f2fs page.private layout Restruct f2fs page private layout for below reasons: There are some cases that f2fs wants to set a flag in a page to indicate a specified status of page: a) page is in transaction list for atomic write b) page contains dummy data for aligned write c) page is migrating for GC d) page contains inline data for inline inode flush e) page belongs to merkle tree, and is verified for fsverity f) page is dirty and has filesystem/inode reference count for writeback g) page is temporary and has decompress io context reference for compression There are existed places in page structure we can use to store f2fs private status/data: - page.flags: PG_checked, PG_private - page.private However it was a mess when we using them, which may cause potential confliction: page.private PG_private PG_checked page._refcount (+1 at most) a) -1 set +1 b) -2 set c), d), e) set f) 0 set +1 g) pointer set The other problem is page.flags has no free slot, if we can avoid set zero to page.private and set PG_private flag, then we use non-zero value to indicate PG_private status, so that we may have chance to reclaim PG_private slot for other usage. [1] The other concern is f2fs has bad scalability in aspect of indicating more page status. So in this patch, let's restructure f2fs' page.private as below to solve above issues: Layout A: lowest bit should be 1 | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | bit 0 PAGE_PRIVATE_NOT_POINTER bit 1 PAGE_PRIVATE_ATOMIC_WRITE bit 2 PAGE_PRIVATE_DUMMY_WRITE bit 3 PAGE_PRIVATE_ONGOING_MIGRATION bit 4 PAGE_PRIVATE_INLINE_INODE bit 5 PAGE_PRIVATE_REF_RESOURCE bit 6- f2fs private data Layout B: lowest bit should be 0 page.private is a wrapped pointer. After the change: page.private PG_private PG_checked page._refcount (+1 at most) a) 11 set +1 b) 101 set +1 c) 1001 set +1 d) 10001 set +1 e) set f) 100001 set +1 g) pointer set +1 [1] https://lore.kernel.org/linux-f2fs-devel/20210422154705.GO3596236@casper.infradead.org/T/#u Cc: Matthew Wilcox <willy@infradead.org> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2021-04-28 09:20:31 +00:00
* Layout of f2fs page.private:
*
* Layout A: lowest bit should be 1
* | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... |
* bit 0 PAGE_PRIVATE_NOT_POINTER
* bit 1 PAGE_PRIVATE_ONGOING_MIGRATION
* bit 2 PAGE_PRIVATE_INLINE_INODE
* bit 3 PAGE_PRIVATE_REF_RESOURCE
f2fs: atomic: fix to avoid racing w/ GC Case #1: SQLite App GC Thread Kworker Shrinker - f2fs_ioc_start_atomic_write - f2fs_ioc_commit_atomic_write - f2fs_commit_atomic_write - filemap_write_and_wait_range : write atomic_file's data to cow_inode echo 3 > drop_caches to drop atomic_file's cache. - f2fs_gc - gc_data_segment - move_data_page - set_page_dirty - writepages - f2fs_do_write_data_page : overwrite atomic_file's data to cow_inode - f2fs_down_write(&fi->i_gc_rwsem[WRITE]) - __f2fs_commit_atomic_write - f2fs_up_write(&fi->i_gc_rwsem[WRITE]) Case #2: SQLite App GC Thread Kworker - f2fs_ioc_start_atomic_write - __writeback_single_inode - do_writepages - f2fs_write_cache_pages - f2fs_write_single_data_page - f2fs_do_write_data_page : write atomic_file's data to cow_inode - f2fs_gc - gc_data_segment - move_data_page - set_page_dirty - writepages - f2fs_do_write_data_page : overwrite atomic_file's data to cow_inode - f2fs_ioc_commit_atomic_write In above cases racing in between atomic_write and GC, previous data in atomic_file may be overwrited to cow_file, result in data corruption. This patch introduces PAGE_PRIVATE_ATOMIC_WRITE bit flag in page.private, and use it to indicate that there is last dirty data in atomic file, and the data should be writebacked into cow_file, if the flag is not tagged in page, we should never write data across files. Fixes: 3db1de0e582c ("f2fs: change the current atomic write way") Cc: Daeho Jeong <daehojeong@google.com> Signed-off-by: Chao Yu <chao@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2024-06-25 03:13:48 +00:00
* bit 4 PAGE_PRIVATE_ATOMIC_WRITE
* bit 5- f2fs private data
f2fs: restructure f2fs page.private layout Restruct f2fs page private layout for below reasons: There are some cases that f2fs wants to set a flag in a page to indicate a specified status of page: a) page is in transaction list for atomic write b) page contains dummy data for aligned write c) page is migrating for GC d) page contains inline data for inline inode flush e) page belongs to merkle tree, and is verified for fsverity f) page is dirty and has filesystem/inode reference count for writeback g) page is temporary and has decompress io context reference for compression There are existed places in page structure we can use to store f2fs private status/data: - page.flags: PG_checked, PG_private - page.private However it was a mess when we using them, which may cause potential confliction: page.private PG_private PG_checked page._refcount (+1 at most) a) -1 set +1 b) -2 set c), d), e) set f) 0 set +1 g) pointer set The other problem is page.flags has no free slot, if we can avoid set zero to page.private and set PG_private flag, then we use non-zero value to indicate PG_private status, so that we may have chance to reclaim PG_private slot for other usage. [1] The other concern is f2fs has bad scalability in aspect of indicating more page status. So in this patch, let's restructure f2fs' page.private as below to solve above issues: Layout A: lowest bit should be 1 | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | bit 0 PAGE_PRIVATE_NOT_POINTER bit 1 PAGE_PRIVATE_ATOMIC_WRITE bit 2 PAGE_PRIVATE_DUMMY_WRITE bit 3 PAGE_PRIVATE_ONGOING_MIGRATION bit 4 PAGE_PRIVATE_INLINE_INODE bit 5 PAGE_PRIVATE_REF_RESOURCE bit 6- f2fs private data Layout B: lowest bit should be 0 page.private is a wrapped pointer. After the change: page.private PG_private PG_checked page._refcount (+1 at most) a) 11 set +1 b) 101 set +1 c) 1001 set +1 d) 10001 set +1 e) set f) 100001 set +1 g) pointer set +1 [1] https://lore.kernel.org/linux-f2fs-devel/20210422154705.GO3596236@casper.infradead.org/T/#u Cc: Matthew Wilcox <willy@infradead.org> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2021-04-28 09:20:31 +00:00
*
* Layout B: lowest bit should be 0
* page.private is a wrapped pointer.
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>
2019-11-01 10:07:14 +00:00
*/
f2fs: restructure f2fs page.private layout Restruct f2fs page private layout for below reasons: There are some cases that f2fs wants to set a flag in a page to indicate a specified status of page: a) page is in transaction list for atomic write b) page contains dummy data for aligned write c) page is migrating for GC d) page contains inline data for inline inode flush e) page belongs to merkle tree, and is verified for fsverity f) page is dirty and has filesystem/inode reference count for writeback g) page is temporary and has decompress io context reference for compression There are existed places in page structure we can use to store f2fs private status/data: - page.flags: PG_checked, PG_private - page.private However it was a mess when we using them, which may cause potential confliction: page.private PG_private PG_checked page._refcount (+1 at most) a) -1 set +1 b) -2 set c), d), e) set f) 0 set +1 g) pointer set The other problem is page.flags has no free slot, if we can avoid set zero to page.private and set PG_private flag, then we use non-zero value to indicate PG_private status, so that we may have chance to reclaim PG_private slot for other usage. [1] The other concern is f2fs has bad scalability in aspect of indicating more page status. So in this patch, let's restructure f2fs' page.private as below to solve above issues: Layout A: lowest bit should be 1 | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | bit 0 PAGE_PRIVATE_NOT_POINTER bit 1 PAGE_PRIVATE_ATOMIC_WRITE bit 2 PAGE_PRIVATE_DUMMY_WRITE bit 3 PAGE_PRIVATE_ONGOING_MIGRATION bit 4 PAGE_PRIVATE_INLINE_INODE bit 5 PAGE_PRIVATE_REF_RESOURCE bit 6- f2fs private data Layout B: lowest bit should be 0 page.private is a wrapped pointer. After the change: page.private PG_private PG_checked page._refcount (+1 at most) a) 11 set +1 b) 101 set +1 c) 1001 set +1 d) 10001 set +1 e) set f) 100001 set +1 g) pointer set +1 [1] https://lore.kernel.org/linux-f2fs-devel/20210422154705.GO3596236@casper.infradead.org/T/#u Cc: Matthew Wilcox <willy@infradead.org> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2021-04-28 09:20:31 +00:00
enum {
PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */
PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */
PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */
PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */
f2fs: atomic: fix to avoid racing w/ GC Case #1: SQLite App GC Thread Kworker Shrinker - f2fs_ioc_start_atomic_write - f2fs_ioc_commit_atomic_write - f2fs_commit_atomic_write - filemap_write_and_wait_range : write atomic_file's data to cow_inode echo 3 > drop_caches to drop atomic_file's cache. - f2fs_gc - gc_data_segment - move_data_page - set_page_dirty - writepages - f2fs_do_write_data_page : overwrite atomic_file's data to cow_inode - f2fs_down_write(&fi->i_gc_rwsem[WRITE]) - __f2fs_commit_atomic_write - f2fs_up_write(&fi->i_gc_rwsem[WRITE]) Case #2: SQLite App GC Thread Kworker - f2fs_ioc_start_atomic_write - __writeback_single_inode - do_writepages - f2fs_write_cache_pages - f2fs_write_single_data_page - f2fs_do_write_data_page : write atomic_file's data to cow_inode - f2fs_gc - gc_data_segment - move_data_page - set_page_dirty - writepages - f2fs_do_write_data_page : overwrite atomic_file's data to cow_inode - f2fs_ioc_commit_atomic_write In above cases racing in between atomic_write and GC, previous data in atomic_file may be overwrited to cow_file, result in data corruption. This patch introduces PAGE_PRIVATE_ATOMIC_WRITE bit flag in page.private, and use it to indicate that there is last dirty data in atomic file, and the data should be writebacked into cow_file, if the flag is not tagged in page, we should never write data across files. Fixes: 3db1de0e582c ("f2fs: change the current atomic write way") Cc: Daeho Jeong <daehojeong@google.com> Signed-off-by: Chao Yu <chao@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2024-06-25 03:13:48 +00:00
PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */
f2fs: restructure f2fs page.private layout Restruct f2fs page private layout for below reasons: There are some cases that f2fs wants to set a flag in a page to indicate a specified status of page: a) page is in transaction list for atomic write b) page contains dummy data for aligned write c) page is migrating for GC d) page contains inline data for inline inode flush e) page belongs to merkle tree, and is verified for fsverity f) page is dirty and has filesystem/inode reference count for writeback g) page is temporary and has decompress io context reference for compression There are existed places in page structure we can use to store f2fs private status/data: - page.flags: PG_checked, PG_private - page.private However it was a mess when we using them, which may cause potential confliction: page.private PG_private PG_checked page._refcount (+1 at most) a) -1 set +1 b) -2 set c), d), e) set f) 0 set +1 g) pointer set The other problem is page.flags has no free slot, if we can avoid set zero to page.private and set PG_private flag, then we use non-zero value to indicate PG_private status, so that we may have chance to reclaim PG_private slot for other usage. [1] The other concern is f2fs has bad scalability in aspect of indicating more page status. So in this patch, let's restructure f2fs' page.private as below to solve above issues: Layout A: lowest bit should be 1 | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | bit 0 PAGE_PRIVATE_NOT_POINTER bit 1 PAGE_PRIVATE_ATOMIC_WRITE bit 2 PAGE_PRIVATE_DUMMY_WRITE bit 3 PAGE_PRIVATE_ONGOING_MIGRATION bit 4 PAGE_PRIVATE_INLINE_INODE bit 5 PAGE_PRIVATE_REF_RESOURCE bit 6- f2fs private data Layout B: lowest bit should be 0 page.private is a wrapped pointer. After the change: page.private PG_private PG_checked page._refcount (+1 at most) a) 11 set +1 b) 101 set +1 c) 1001 set +1 d) 10001 set +1 e) set f) 100001 set +1 g) pointer set +1 [1] https://lore.kernel.org/linux-f2fs-devel/20210422154705.GO3596236@casper.infradead.org/T/#u Cc: Matthew Wilcox <willy@infradead.org> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2021-04-28 09:20:31 +00:00
PAGE_PRIVATE_MAX
};
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>
2019-11-01 10:07:14 +00:00
/* For compression */
enum compress_algorithm_type {
COMPRESS_LZO,
COMPRESS_LZ4,
COMPRESS_ZSTD,
COMPRESS_LZORLE,
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>
2019-11-01 10:07:14 +00:00
COMPRESS_MAX,
};
enum compress_flag {
COMPRESS_CHKSUM,
COMPRESS_MAX_FLAG,
};
#define COMPRESS_WATERMARK 20
#define COMPRESS_PERCENT 20
#define COMPRESS_DATA_RESERVED_SIZE 4
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>
2019-11-01 10:07:14 +00:00
struct compress_data {
__le32 clen; /* compressed data size */
__le32 chksum; /* compressed data chksum */
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>
2019-11-01 10:07:14 +00:00
__le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */
u8 cdata[]; /* compressed data */
};
#define COMPRESS_HEADER_SIZE (sizeof(struct compress_data))
#define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000
#define F2FS_ZSTD_DEFAULT_CLEVEL 1
#define COMPRESS_LEVEL_OFFSET 8
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>
2019-11-01 10:07:14 +00:00
/* compress context */
struct compress_ctx {
struct inode *inode; /* inode the context belong to */
pgoff_t cluster_idx; /* cluster index number */
unsigned int cluster_size; /* page count in cluster */
unsigned int log_cluster_size; /* log of cluster size */
struct page **rpages; /* pages store raw data in cluster */
unsigned int nr_rpages; /* total page number in rpages */
struct page **cpages; /* pages store compressed data in cluster */
unsigned int nr_cpages; /* total page number in cpages */
unsigned int valid_nr_cpages; /* valid page number in cpages */
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>
2019-11-01 10:07:14 +00:00
void *rbuf; /* virtual mapped address on rpages */
struct compress_data *cbuf; /* virtual mapped address on cpages */
size_t rlen; /* valid data length in rbuf */
size_t clen; /* valid data length in cbuf */
void *private; /* payload buffer for specified compression algorithm */
void *private2; /* extra payload buffer */
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>
2019-11-01 10:07:14 +00:00
};
/* compress context for write IO path */
struct compress_io_ctx {
u32 magic; /* magic number to indicate page is compressed */
struct inode *inode; /* inode the context belong to */
struct page **rpages; /* pages store raw data in cluster */
unsigned int nr_rpages; /* total page number in rpages */
atomic_t pending_pages; /* in-flight compressed page count */
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>
2019-11-01 10:07:14 +00:00
};
/* Context for decompressing one cluster on the read IO path */
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>
2019-11-01 10:07:14 +00:00
struct decompress_io_ctx {
u32 magic; /* magic number to indicate page is compressed */
struct inode *inode; /* inode the context belong to */
pgoff_t cluster_idx; /* cluster index number */
unsigned int cluster_size; /* page count in cluster */
unsigned int log_cluster_size; /* log of cluster size */
struct page **rpages; /* pages store raw data in cluster */
unsigned int nr_rpages; /* total page number in rpages */
struct page **cpages; /* pages store compressed data in cluster */
unsigned int nr_cpages; /* total page number in cpages */
struct page **tpages; /* temp pages to pad holes in cluster */
void *rbuf; /* virtual mapped address on rpages */
struct compress_data *cbuf; /* virtual mapped address on cpages */
size_t rlen; /* valid data length in rbuf */
size_t clen; /* valid data length in cbuf */
/*
* The number of compressed pages remaining to be read in this cluster.
* This is initially nr_cpages. It is decremented by 1 each time a page
* has been read (or failed to be read). When it reaches 0, the cluster
* is decompressed (or an error is reported).
*
* If an error occurs before all the pages have been submitted for I/O,
* then this will never reach 0. In this case the I/O submitter is
* responsible for calling f2fs_decompress_end_io() instead.
*/
atomic_t remaining_pages;
/*
* Number of references to this decompress_io_ctx.
*
* One reference is held for I/O completion. This reference is dropped
* after the pagecache pages are updated and unlocked -- either after
* decompression (and verity if enabled), or after an error.
*
* In addition, each compressed page holds a reference while it is in a
* bio. These references are necessary prevent compressed pages from
* being freed while they are still in a bio.
*/
refcount_t refcnt;
bool failed; /* IO error occurred before decompression? */
bool need_verity; /* need fs-verity verification after decompression? */
void *private; /* payload buffer for specified decompression algorithm */
void *private2; /* extra payload buffer */
struct work_struct verity_work; /* work to verify the decompressed pages */
struct work_struct free_work; /* work for late free this structure itself */
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>
2019-11-01 10:07:14 +00:00
};
#define NULL_CLUSTER ((unsigned int)(~0))
#define MIN_COMPRESS_LOG_SIZE 2
#define MAX_COMPRESS_LOG_SIZE 8
f2fs: allocate proper size memory for zstd decompress As 5kft <5kft@5kft.org> reported: kworker/u9:3: page allocation failure: order:9, mode:0x40c40(GFP_NOFS|__GFP_COMP), nodemask=(null),cpuset=/,mems_allowed=0 CPU: 3 PID: 8168 Comm: kworker/u9:3 Tainted: G C 5.8.3-sunxi #trunk Hardware name: Allwinner sun8i Family Workqueue: f2fs_post_read_wq f2fs_post_read_work [<c010d6d5>] (unwind_backtrace) from [<c0109a55>] (show_stack+0x11/0x14) [<c0109a55>] (show_stack) from [<c056d489>] (dump_stack+0x75/0x84) [<c056d489>] (dump_stack) from [<c0243b53>] (warn_alloc+0xa3/0x104) [<c0243b53>] (warn_alloc) from [<c024473b>] (__alloc_pages_nodemask+0xb87/0xc40) [<c024473b>] (__alloc_pages_nodemask) from [<c02267c5>] (kmalloc_order+0x19/0x38) [<c02267c5>] (kmalloc_order) from [<c02267fd>] (kmalloc_order_trace+0x19/0x90) [<c02267fd>] (kmalloc_order_trace) from [<c047c665>] (zstd_init_decompress_ctx+0x21/0x88) [<c047c665>] (zstd_init_decompress_ctx) from [<c047e9cf>] (f2fs_decompress_pages+0x97/0x228) [<c047e9cf>] (f2fs_decompress_pages) from [<c045d0ab>] (__read_end_io+0xfb/0x130) [<c045d0ab>] (__read_end_io) from [<c045d141>] (f2fs_post_read_work+0x61/0x84) [<c045d141>] (f2fs_post_read_work) from [<c0130b2f>] (process_one_work+0x15f/0x3b0) [<c0130b2f>] (process_one_work) from [<c0130e7b>] (worker_thread+0xfb/0x3e0) [<c0130e7b>] (worker_thread) from [<c0135c3b>] (kthread+0xeb/0x10c) [<c0135c3b>] (kthread) from [<c0100159>] zstd may allocate large size memory for {,de}compression, it may cause file copy failure on low-end device which has very few memory. For decompression, let's just allocate proper size memory based on current file's cluster size instead of max cluster size. Reported-by: 5kft <5kft@5kft.org> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2020-09-02 07:01:52 +00:00
#define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size))
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>
2019-11-01 10:07:14 +00:00
struct f2fs_sb_info {
struct super_block *sb; /* pointer to VFS super block */
struct proc_dir_entry *s_proc; /* proc entry */
struct f2fs_super_block *raw_super; /* raw super block pointer */
struct f2fs_rwsem sb_lock; /* lock for raw super block */
int valid_super_block; /* valid super block no */
unsigned long s_flag; /* flags for sbi */
struct mutex writepages; /* mutex for writepages() */
#ifdef CONFIG_BLK_DEV_ZONED
unsigned int blocks_per_blkz; /* F2FS blocks per zone */
unsigned int max_open_zones; /* max open zone resources of the zoned device */
/* For adjust the priority writing position of data in zone UFS */
unsigned int blkzone_alloc_policy;
#endif
/* for node-related operations */
struct f2fs_nm_info *nm_info; /* node manager */
struct inode *node_inode; /* cache node blocks */
/* for segment-related operations */
struct f2fs_sm_info *sm_info; /* segment manager */
/* for bio operations */
struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */
/* keep migration IO order for LFS mode */
struct f2fs_rwsem io_order_lock;
f2fs: avoid infinite loop to flush node pages xfstests/generic/475 can give EIO all the time which give an infinite loop to flush node page like below. Let's avoid it. [16418.518551] Call Trace: [16418.518553] ? dm_submit_bio+0x48/0x400 [16418.518574] ? submit_bio_checks+0x1ac/0x5a0 [16418.525207] __submit_bio+0x1a9/0x230 [16418.525210] ? kmem_cache_alloc+0x29e/0x3c0 [16418.525223] submit_bio_noacct+0xa8/0x2b0 [16418.525226] submit_bio+0x4d/0x130 [16418.525238] __submit_bio+0x49/0x310 [f2fs] [16418.525339] ? bio_add_page+0x6a/0x90 [16418.525344] f2fs_submit_page_bio+0x134/0x1f0 [f2fs] [16418.525365] read_node_page+0x125/0x1b0 [f2fs] [16418.525388] __get_node_page.part.0+0x58/0x3f0 [f2fs] [16418.525409] __get_node_page+0x2f/0x60 [f2fs] [16418.525431] f2fs_get_dnode_of_data+0x423/0x860 [f2fs] [16418.525452] ? asm_sysvec_apic_timer_interrupt+0x12/0x20 [16418.525458] ? __mod_memcg_state.part.0+0x2a/0x30 [16418.525465] ? __mod_memcg_lruvec_state+0x27/0x40 [16418.525467] ? __xa_set_mark+0x57/0x70 [16418.525472] f2fs_do_write_data_page+0x10e/0x7b0 [f2fs] [16418.525493] f2fs_write_single_data_page+0x555/0x830 [f2fs] [16418.525514] ? sysvec_apic_timer_interrupt+0x4e/0x90 [16418.525518] ? asm_sysvec_apic_timer_interrupt+0x12/0x20 [16418.525523] f2fs_write_cache_pages+0x303/0x880 [f2fs] [16418.525545] ? blk_flush_plug_list+0x47/0x100 [16418.525548] f2fs_write_data_pages+0xfd/0x320 [f2fs] [16418.525569] do_writepages+0xd5/0x210 [16418.525648] filemap_fdatawrite_wbc+0x7d/0xc0 [16418.525655] filemap_fdatawrite+0x50/0x70 [16418.525658] f2fs_sync_dirty_inodes+0xa4/0x230 [f2fs] [16418.525679] f2fs_write_checkpoint+0x16d/0x1720 [f2fs] [16418.525699] ? ttwu_do_wakeup+0x1c/0x160 [16418.525709] ? ttwu_do_activate+0x6d/0xd0 [16418.525711] ? __wait_for_common+0x11d/0x150 [16418.525715] kill_f2fs_super+0xca/0x100 [f2fs] [16418.525733] deactivate_locked_super+0x3b/0xb0 [16418.525739] deactivate_super+0x40/0x50 [16418.525741] cleanup_mnt+0x139/0x190 [16418.525747] __cleanup_mnt+0x12/0x20 [16418.525749] task_work_run+0x6d/0xa0 [16418.525765] exit_to_user_mode_prepare+0x1ad/0x1b0 [16418.525771] syscall_exit_to_user_mode+0x27/0x50 [16418.525774] do_syscall_64+0x48/0xc0 [16418.525776] entry_SYSCALL_64_after_hwframe+0x44/0xae Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2022-03-29 23:25:54 +00:00
pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */
int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */
/* for checkpoint */
struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */
int cur_cp_pack; /* remain current cp pack */
spinlock_t cp_lock; /* for flag in ckpt */
struct inode *meta_inode; /* cache meta blocks */
struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */
struct f2fs_rwsem cp_rwsem; /* blocking FS operations */
struct f2fs_rwsem node_write; /* locking node writes */
struct f2fs_rwsem node_change; /* locking node change */
wait_queue_head_t cp_wait;
unsigned long last_time[MAX_TIME]; /* to store time in jiffies */
long interval_time[MAX_TIME]; /* to store thresholds */
f2fs: introduce checkpoint_merge mount option We've added a new mount options, "checkpoint_merge" and "nocheckpoint_merge", which creates a kernel daemon and makes it to merge concurrent checkpoint requests as much as possible to eliminate redundant checkpoint issues. Plus, we can eliminate the sluggish issue caused by slow checkpoint operation when the checkpoint is done in a process context in a cgroup having low i/o budget and cpu shares. To make this do better, we set the default i/o priority of the kernel daemon to "3", to give one higher priority than other kernel threads. The below verification result explains this. The basic idea has come from https://opensource.samsung.com. [Verification] Android Pixel Device(ARM64, 7GB RAM, 256GB UFS) Create two I/O cgroups (fg w/ weight 100, bg w/ wight 20) Set "strict_guarantees" to "1" in BFQ tunables In "fg" cgroup, - thread A => trigger 1000 checkpoint operations "for i in `seq 1 1000`; do touch test_dir1/file; fsync test_dir1; done" - thread B => gererating async. I/O "fio --rw=write --numjobs=1 --bs=128k --runtime=3600 --time_based=1 --filename=test_img --name=test" In "bg" cgroup, - thread C => trigger repeated checkpoint operations "echo $$ > /dev/blkio/bg/tasks; while true; do touch test_dir2/file; fsync test_dir2; done" We've measured thread A's execution time. [ w/o patch ] Elapsed Time: Avg. 68 seconds [ w/ patch ] Elapsed Time: Avg. 48 seconds Reported-by: kernel test robot <lkp@intel.com> Reported-by: Dan Carpenter <dan.carpenter@oracle.com> [Jaegeuk Kim: fix the return value in f2fs_start_ckpt_thread, reported by Dan] Signed-off-by: Daeho Jeong <daehojeong@google.com> Signed-off-by: Sungjong Seo <sj1557.seo@samsung.com> Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2021-01-19 00:00:42 +00:00
struct ckpt_req_control cprc_info; /* for checkpoint request control */
struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */
f2fs: fix to avoid broken of dnode block list f2fs recovery flow is relying on dnode block link list, it means fsynced file recovery depends on previous dnode's persistence in the list, so during fsync() we should wait on all regular inode's dnode writebacked before issuing flush. By this way, we can avoid dnode block list being broken by out-of-order IO submission due to IO scheduler or driver. Sheng Yong helps to do the test with this patch: Target:/data (f2fs, -) 64MB / 32768KB / 4KB / 8 1 / PERSIST / Index Base: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 867.82 204.15 41440.03 41370.54 680.8 1025.94 1031.08 2 871.87 205.87 41370.3 40275.2 791.14 1065.84 1101.7 3 866.52 205.69 41795.67 40596.16 694.69 1037.16 1031.48 Avg 868.7366667 205.2366667 41535.33333 40747.3 722.21 1042.98 1054.753333 After: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 798.81 202.5 41143 40613.87 602.71 838.08 913.83 2 805.79 206.47 40297.2 41291.46 604.44 840.75 924.27 3 814.83 206.17 41209.57 40453.62 602.85 834.66 927.91 Avg 806.4766667 205.0466667 40883.25667 40786.31667 603.3333333 837.83 922.0033333 Patched/Original: 0.928332713 0.999074239 0.984300676 1.000957528 0.835398753 0.803303994 0.874141189 It looks like atomic write will suffer performance regression. I suspect that the criminal is that we forcing to wait all dnode being in storage cache before we issue PREFLUSH+FUA. BTW, will commit ("f2fs: don't need to wait for node writes for atomic write") cause the problem: we will lose data of last transaction after SPO, even if atomic write return no error: - atomic_open(); - write() P1, P2, P3; - atomic_commit(); - writeback data: P1, P2, P3; - writeback node: N1, N2, N3; <--- If N1, N2 is not writebacked, N3 with fsync_mark is writebacked, In SPOR, we won't find N3 since node chain is broken, turns out that losing last transaction. - preflush + fua; - power-cut If we don't wait dnode writeback for atomic_write: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 779.91 206.03 41621.5 40333.16 716.9 1038.21 1034.85 2 848.51 204.35 40082.44 39486.17 791.83 1119.96 1083.77 3 772.12 206.27 41335.25 41599.65 723.29 1055.07 971.92 Avg 800.18 205.55 41013.06333 40472.99333 744.0066667 1071.08 1030.18 Patched/Original: 0.92108464 1.001526693 0.987425886 0.993268102 1.030180511 1.026942031 0.976702294 SQLite's performance recovers. Jaegeuk: "Practically, I don't see db corruption becase of this. We can excuse to lose the last transaction." Finally, we decide to keep original implementation of atomic write interface sematics that we don't wait all dnode writeback before preflush+fua submission. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-02 15:03:19 +00:00
spinlock_t fsync_node_lock; /* for node entry lock */
struct list_head fsync_node_list; /* node list head */
unsigned int fsync_seg_id; /* sequence id */
unsigned int fsync_node_num; /* number of node entries */
/* for orphan inode, use 0'th array */
unsigned int max_orphans; /* max orphan inodes */
/* for inode management */
struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */
spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */
f2fs: fix to avoid deadloop if data_flush is on As Hagbard Celine reported: [ 615.697824] INFO: task kworker/u16:5:344 blocked for more than 120 seconds. [ 615.697825] Not tainted 5.0.15-gentoo-f2fslog #4 [ 615.697826] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. [ 615.697827] kworker/u16:5 D 0 344 2 0x80000000 [ 615.697831] Workqueue: writeback wb_workfn (flush-259:0) [ 615.697832] Call Trace: [ 615.697836] ? __schedule+0x2c5/0x8b0 [ 615.697839] schedule+0x32/0x80 [ 615.697841] schedule_preempt_disabled+0x14/0x20 [ 615.697842] __mutex_lock.isra.8+0x2ba/0x4d0 [ 615.697845] ? log_store+0xf5/0x260 [ 615.697848] f2fs_write_data_pages+0x133/0x320 [ 615.697851] ? trace_hardirqs_on+0x2c/0xe0 [ 615.697854] do_writepages+0x41/0xd0 [ 615.697857] __filemap_fdatawrite_range+0x81/0xb0 [ 615.697859] f2fs_sync_dirty_inodes+0x1dd/0x200 [ 615.697861] f2fs_balance_fs_bg+0x2a7/0x2c0 [ 615.697863] ? up_read+0x5/0x20 [ 615.697865] ? f2fs_do_write_data_page+0x2cb/0x940 [ 615.697867] f2fs_balance_fs+0xe5/0x2c0 [ 615.697869] __write_data_page+0x1c8/0x6e0 [ 615.697873] f2fs_write_cache_pages+0x1e0/0x450 [ 615.697878] f2fs_write_data_pages+0x14b/0x320 [ 615.697880] ? trace_hardirqs_on+0x2c/0xe0 [ 615.697883] do_writepages+0x41/0xd0 [ 615.697885] __filemap_fdatawrite_range+0x81/0xb0 [ 615.697887] f2fs_sync_dirty_inodes+0x1dd/0x200 [ 615.697889] f2fs_balance_fs_bg+0x2a7/0x2c0 [ 615.697891] f2fs_write_node_pages+0x51/0x220 [ 615.697894] do_writepages+0x41/0xd0 [ 615.697897] __writeback_single_inode+0x3d/0x3d0 [ 615.697899] writeback_sb_inodes+0x1e8/0x410 [ 615.697902] __writeback_inodes_wb+0x5d/0xb0 [ 615.697904] wb_writeback+0x28f/0x340 [ 615.697906] ? cpumask_next+0x16/0x20 [ 615.697908] wb_workfn+0x33e/0x420 [ 615.697911] process_one_work+0x1a1/0x3d0 [ 615.697913] worker_thread+0x30/0x380 [ 615.697915] ? process_one_work+0x3d0/0x3d0 [ 615.697916] kthread+0x116/0x130 [ 615.697918] ? kthread_create_worker_on_cpu+0x70/0x70 [ 615.697921] ret_from_fork+0x3a/0x50 There is still deadloop in below condition: d A - do_writepages - f2fs_write_node_pages - f2fs_balance_fs_bg - f2fs_sync_dirty_inodes - f2fs_write_cache_pages - mutex_lock(&sbi->writepages) -- lock once - __write_data_page - f2fs_balance_fs_bg - f2fs_sync_dirty_inodes - f2fs_write_data_pages - mutex_lock(&sbi->writepages) -- lock again Thread A Thread B - do_writepages - f2fs_write_node_pages - f2fs_balance_fs_bg - f2fs_sync_dirty_inodes - .cp_task = current - f2fs_sync_dirty_inodes - .cp_task = current - filemap_fdatawrite - .cp_task = NULL - filemap_fdatawrite - f2fs_write_cache_pages - enter f2fs_balance_fs_bg since .cp_task is NULL - .cp_task = NULL Change as below to avoid this: - add condition to avoid holding .writepages mutex lock in path of data flush - introduce mutex lock sbi.flush_lock to exclude concurrent data flush in background. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-05-20 09:36:59 +00:00
struct mutex flush_lock; /* for flush exclusion */
/* for extent tree cache */
struct extent_tree_info extent_tree[NR_EXTENT_CACHES];
f2fs: add block_age-based extent cache This patch introduces a runtime hot/cold data separation method for f2fs, in order to improve the accuracy for data temperature classification, reduce the garbage collection overhead after long-term data updates. Enhanced hot/cold data separation can record data block update frequency as "age" of the extent per inode, and take use of the age info to indicate better temperature type for data block allocation: - It records total data blocks allocated since mount; - When file extent has been updated, it calculate the count of data blocks allocated since last update as the age of the extent; - Before the data block allocated, it searches for the age info and chooses the suitable segment for allocation. Test and result: - Prepare: create about 30000 files * 3% for cold files (with cold file extension like .apk, from 3M to 10M) * 50% for warm files (with random file extension like .FcDxq, from 1K to 4M) * 47% for hot files (with hot file extension like .db, from 1K to 256K) - create(5%)/random update(90%)/delete(5%) the files * total write amount is about 70G * fsync will be called for .db files, and buffered write will be used for other files The storage of test device is large enough(128G) so that it will not switch to SSR mode during the test. Benefit: dirty segment count increment reduce about 14% - before: Dirty +21110 - after: Dirty +18286 Signed-off-by: qixiaoyu1 <qixiaoyu1@xiaomi.com> Signed-off-by: xiongping1 <xiongping1@xiaomi.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2022-12-02 01:37:15 +00:00
atomic64_t allocated_data_blocks; /* for block age extent_cache */
/* The threshold used for hot and warm data seperation*/
unsigned int hot_data_age_threshold;
unsigned int warm_data_age_threshold;
unsigned int last_age_weight;
/* basic filesystem units */
unsigned int log_sectors_per_block; /* log2 sectors per block */
unsigned int log_blocksize; /* log2 block size */
unsigned int blocksize; /* block size */
unsigned int root_ino_num; /* root inode number*/
unsigned int node_ino_num; /* node inode number*/
unsigned int meta_ino_num; /* meta inode number*/
unsigned int log_blocks_per_seg; /* log2 blocks per segment */
unsigned int blocks_per_seg; /* blocks per segment */
unsigned int unusable_blocks_per_sec; /* unusable blocks per section */
unsigned int segs_per_sec; /* segments per section */
unsigned int secs_per_zone; /* sections per zone */
unsigned int total_sections; /* total section count */
unsigned int total_node_count; /* total node block count */
unsigned int total_valid_node_count; /* valid node block count */
int dir_level; /* directory level */
bool readdir_ra; /* readahead inode in readdir */
u64 max_io_bytes; /* max io bytes to merge IOs */
block_t user_block_count; /* # of user blocks */
block_t total_valid_block_count; /* # of valid blocks */
block_t discard_blks; /* discard command candidats */
block_t last_valid_block_count; /* for recovery */
block_t reserved_blocks; /* configurable reserved blocks */
block_t current_reserved_blocks; /* current reserved blocks */
/* Additional tracking for no checkpoint mode */
block_t unusable_block_count; /* # of blocks saved by last cp */
unsigned int nquota_files; /* # of quota sysfile */
struct f2fs_rwsem quota_sem; /* blocking cp for flags */
/* # of pages, see count_type */
atomic_t nr_pages[NR_COUNT_TYPE];
/* # of allocated blocks */
struct percpu_counter alloc_valid_block_count;
/* # of node block writes as roll forward recovery */
struct percpu_counter rf_node_block_count;
/* writeback control */
atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */
/* valid inode count */
struct percpu_counter total_valid_inode_count;
struct f2fs_mount_info mount_opt; /* mount options */
/* for cleaning operations */
struct f2fs_rwsem gc_lock; /*
* semaphore for GC, avoid
* race between GC and GC or CP
*/
struct f2fs_gc_kthread *gc_thread; /* GC thread */
f2fs: support age threshold based garbage collection There are several issues in current background GC algorithm: - valid blocks is one of key factors during cost overhead calculation, so if segment has less valid block, however even its age is young or it locates hot segment, CB algorithm will still choose the segment as victim, it's not appropriate. - GCed data/node will go to existing logs, no matter in-there datas' update frequency is the same or not, it may mix hot and cold data again. - GC alloctor mainly use LFS type segment, it will cost free segment more quickly. This patch introduces a new algorithm named age threshold based garbage collection to solve above issues, there are three steps mainly: 1. select a source victim: - set an age threshold, and select candidates beased threshold: e.g. 0 means youngest, 100 means oldest, if we set age threshold to 80 then select dirty segments which has age in range of [80, 100] as candiddates; - set candidate_ratio threshold, and select candidates based the ratio, so that we can shrink candidates to those oldest segments; - select target segment with fewest valid blocks in order to migrate blocks with minimum cost; 2. select a target victim: - select candidates beased age threshold; - set candidate_radius threshold, search candidates whose age is around source victims, searching radius should less than the radius threshold. - select target segment with most valid blocks in order to avoid migrating current target segment. 3. merge valid blocks from source victim into target victim with SSR alloctor. Test steps: - create 160 dirty segments: * half of them have 128 valid blocks per segment * left of them have 384 valid blocks per segment - run background GC Benefit: GC count and block movement count both decrease obviously: - Before: - Valid: 86 - Dirty: 1 - Prefree: 11 - Free: 6001 (6001) GC calls: 162 (BG: 220) - data segments : 160 (160) - node segments : 2 (2) Try to move 41454 blocks (BG: 41454) - data blocks : 40960 (40960) - node blocks : 494 (494) IPU: 0 blocks SSR: 0 blocks in 0 segments LFS: 41364 blocks in 81 segments - After: - Valid: 87 - Dirty: 0 - Prefree: 4 - Free: 6008 (6008) GC calls: 75 (BG: 76) - data segments : 74 (74) - node segments : 1 (1) Try to move 12813 blocks (BG: 12813) - data blocks : 12544 (12544) - node blocks : 269 (269) IPU: 0 blocks SSR: 12032 blocks in 77 segments LFS: 855 blocks in 2 segments Signed-off-by: Chao Yu <yuchao0@huawei.com> [Jaegeuk Kim: fix a bug along with pinfile in-mem segment & clean up] Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2020-08-04 13:14:49 +00:00
struct atgc_management am; /* atgc management */
unsigned int cur_victim_sec; /* current victim section num */
unsigned int gc_mode; /* current GC state */
unsigned int next_victim_seg[2]; /* next segment in victim section */
spinlock_t gc_remaining_trials_lock;
/* remaining trial count for GC_URGENT_* and GC_IDLE_* */
unsigned int gc_remaining_trials;
/* for skip statistic */
unsigned long long skipped_gc_rwsem; /* FG_GC only */
/* threshold for gc trials on pinned files */
unsigned short gc_pin_file_threshold;
struct f2fs_rwsem pin_sem;
/* maximum # of trials to find a victim segment for SSR and GC */
unsigned int max_victim_search;
/* migration granularity of garbage collection, unit: segment */
unsigned int migration_granularity;
/*
* for stat information.
* one is for the LFS mode, and the other is for the SSR mode.
*/
#ifdef CONFIG_F2FS_STAT_FS
struct f2fs_stat_info *stat_info; /* FS status information */
atomic_t meta_count[META_MAX]; /* # of meta blocks */
unsigned int segment_count[2]; /* # of allocated segments */
unsigned int block_count[2]; /* # of allocated blocks */
atomic_t inplace_count; /* # of inplace update */
/* # of lookup extent cache */
atomic64_t total_hit_ext[NR_EXTENT_CACHES];
/* # of hit rbtree extent node */
atomic64_t read_hit_rbtree[NR_EXTENT_CACHES];
/* # of hit cached extent node */
atomic64_t read_hit_cached[NR_EXTENT_CACHES];
/* # of hit largest extent node in read extent cache */
atomic64_t read_hit_largest;
atomic_t inline_xattr; /* # of inline_xattr inodes */
atomic_t inline_inode; /* # of inline_data inodes */
atomic_t inline_dir; /* # of inline_dentry inodes */
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>
2019-11-01 10:07:14 +00:00
atomic_t compr_inode; /* # of compressed inodes */
atomic64_t compr_blocks; /* # of compressed blocks */
atomic_t swapfile_inode; /* # of swapfile inodes */
atomic_t atomic_files; /* # of opened atomic file */
atomic_t max_aw_cnt; /* max # of atomic writes */
unsigned int io_skip_bggc; /* skip background gc for in-flight IO */
unsigned int other_skip_bggc; /* skip background gc for other reasons */
unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */
atomic_t cp_call_count[MAX_CALL_TYPE]; /* # of cp call */
#endif
spinlock_t stat_lock; /* lock for stat operations */
/* to attach REQ_META|REQ_FUA flags */
unsigned int data_io_flag;
unsigned int node_io_flag;
/* For sysfs support */
struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */
struct completion s_kobj_unregister;
struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */
struct completion s_stat_kobj_unregister;
struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */
struct completion s_feature_list_kobj_unregister;
/* For shrinker support */
struct list_head s_list;
struct mutex umount_mutex;
unsigned int shrinker_run_no;
/* For multi devices */
int s_ndevs; /* number of devices */
struct f2fs_dev_info *devs; /* for device list */
unsigned int dirty_device; /* for checkpoint data flush */
spinlock_t dev_lock; /* protect dirty_device */
bool aligned_blksize; /* all devices has the same logical blksize */
/* For write statistics */
u64 sectors_written_start;
u64 kbytes_written;
/* Reference to checksum algorithm driver via cryptoapi */
struct crypto_shash *s_chksum_driver;
/* Precomputed FS UUID checksum for seeding other checksums */
__u32 s_chksum_seed;
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>
2019-11-01 10:07:14 +00:00
struct workqueue_struct *post_read_wq; /* post read workqueue */
/*
* If we are in irq context, let's update error information into
* on-disk superblock in the work.
*/
struct work_struct s_error_work;
unsigned char errors[MAX_F2FS_ERRORS]; /* error flags */
unsigned char stop_reason[MAX_STOP_REASON]; /* stop reason */
spinlock_t error_lock; /* protect errors/stop_reason array */
bool error_dirty; /* errors of sb is dirty */
struct kmem_cache *inline_xattr_slab; /* inline xattr entry */
unsigned int inline_xattr_slab_size; /* default inline xattr slab size */
/* For reclaimed segs statistics per each GC mode */
unsigned int gc_segment_mode; /* GC state for reclaimed segments */
unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */
unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */
int max_fragment_chunk; /* max chunk size for block fragmentation mode */
int max_fragment_hole; /* max hole size for block fragmentation mode */
/* For atomic write statistics */
atomic64_t current_atomic_write;
s64 peak_atomic_write;
u64 committed_atomic_block;
u64 revoked_atomic_block;
#ifdef CONFIG_F2FS_FS_COMPRESSION
struct kmem_cache *page_array_slab; /* page array entry */
unsigned int page_array_slab_size; /* default page array slab size */
/* For runtime compression statistics */
u64 compr_written_block;
u64 compr_saved_block;
u32 compr_new_inode;
/* For compressed block cache */
struct inode *compress_inode; /* cache compressed blocks */
unsigned int compress_percent; /* cache page percentage */
unsigned int compress_watermark; /* cache page watermark */
atomic_t compress_page_hit; /* cache hit count */
#endif
#ifdef CONFIG_F2FS_IOSTAT
/* For app/fs IO statistics */
spinlock_t iostat_lock;
unsigned long long iostat_count[NR_IO_TYPE];
unsigned long long iostat_bytes[NR_IO_TYPE];
unsigned long long prev_iostat_bytes[NR_IO_TYPE];
bool iostat_enable;
unsigned long iostat_next_period;
unsigned int iostat_period_ms;
f2fs: introduce periodic iostat io latency traces Whenever we notice some sluggish issues on our machines, we are always curious about how well all types of I/O in the f2fs filesystem are handled. But, it's hard to get this kind of real data. First of all, we need to reproduce the issue while turning on the profiling tool like blktrace, but the issue doesn't happen again easily. Second, with the intervention of any tools, the overall timing of the issue will be slightly changed and it sometimes makes us hard to figure it out. So, I added the feature printing out IO latency statistics tracepoint events, which are minimal things to understand filesystem's I/O related behaviors, into F2FS_IOSTAT kernel config. With "iostat_enable" sysfs node on, we can get this statistics info in a periodic way and it would cause the least overhead. [samples] f2fs_ckpt-254:1-507 [003] .... 2842.439683: f2fs_iostat_latency: dev = (254,11), iotype [peak lat.(ms)/avg lat.(ms)/count], rd_data [136/1/801], rd_node [136/1/1704], rd_meta [4/2/4], wr_sync_data [164/16/3331], wr_sync_node [152/3/648], wr_sync_meta [160/2/4243], wr_async_data [24/13/15], wr_async_node [0/0/0], wr_async_meta [0/0/0] f2fs_ckpt-254:1-507 [002] .... 2845.450514: f2fs_iostat_latency: dev = (254,11), iotype [peak lat.(ms)/avg lat.(ms)/count], rd_data [60/3/456], rd_node [60/3/1258], rd_meta [0/0/1], wr_sync_data [120/12/2285], wr_sync_node [88/5/428], wr_sync_meta [52/6/2990], wr_async_data [4/1/3], wr_async_node [0/0/0], wr_async_meta [0/0/0] Signed-off-by: Daeho Jeong <daehojeong@google.com> Reviewed-by: Chao Yu <chao@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2021-08-20 22:29:09 +00:00
/* For io latency related statistics info in one iostat period */
spinlock_t iostat_lat_lock;
struct iostat_lat_info *iostat_io_lat;
#endif
};
/* Definitions to access f2fs_sb_info */
#define SEGS_TO_BLKS(sbi, segs) \
((segs) << (sbi)->log_blocks_per_seg)
#define BLKS_TO_SEGS(sbi, blks) \
((blks) >> (sbi)->log_blocks_per_seg)
#define BLKS_PER_SEG(sbi) ((sbi)->blocks_per_seg)
#define BLKS_PER_SEC(sbi) (SEGS_TO_BLKS(sbi, (sbi)->segs_per_sec))
#define SEGS_PER_SEC(sbi) ((sbi)->segs_per_sec)
__printf(3, 4)
void f2fs_printk(struct f2fs_sb_info *sbi, bool limit_rate, const char *fmt, ...);
#define f2fs_err(sbi, fmt, ...) \
f2fs_printk(sbi, false, KERN_ERR fmt, ##__VA_ARGS__)
#define f2fs_warn(sbi, fmt, ...) \
f2fs_printk(sbi, false, KERN_WARNING fmt, ##__VA_ARGS__)
#define f2fs_notice(sbi, fmt, ...) \
f2fs_printk(sbi, false, KERN_NOTICE fmt, ##__VA_ARGS__)
#define f2fs_info(sbi, fmt, ...) \
f2fs_printk(sbi, false, KERN_INFO fmt, ##__VA_ARGS__)
#define f2fs_debug(sbi, fmt, ...) \
f2fs_printk(sbi, false, KERN_DEBUG fmt, ##__VA_ARGS__)
#define f2fs_err_ratelimited(sbi, fmt, ...) \
f2fs_printk(sbi, true, KERN_ERR fmt, ##__VA_ARGS__)
#define f2fs_warn_ratelimited(sbi, fmt, ...) \
f2fs_printk(sbi, true, KERN_WARNING fmt, ##__VA_ARGS__)
#define f2fs_info_ratelimited(sbi, fmt, ...) \
f2fs_printk(sbi, true, KERN_INFO fmt, ##__VA_ARGS__)
#ifdef CONFIG_F2FS_FAULT_INJECTION
#define time_to_inject(sbi, type) __time_to_inject(sbi, type, __func__, \
__builtin_return_address(0))
static inline bool __time_to_inject(struct f2fs_sb_info *sbi, int type,
const char *func, const char *parent_func)
{
struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
if (!ffi->inject_rate)
return false;
if (!IS_FAULT_SET(ffi, type))
return false;
atomic_inc(&ffi->inject_ops);
if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) {
atomic_set(&ffi->inject_ops, 0);
f2fs_info_ratelimited(sbi, "inject %s in %s of %pS",
f2fs_fault_name[type], func, parent_func);
return true;
}
return false;
}
#else
static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
{
return false;
}
#endif
/*
* Test if the mounted volume is a multi-device volume.
* - For a single regular disk volume, sbi->s_ndevs is 0.
* - For a single zoned disk volume, sbi->s_ndevs is 1.
* - For a multi-device volume, sbi->s_ndevs is always 2 or more.
*/
static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi)
{
return sbi->s_ndevs > 1;
}
static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type)
{
unsigned long now = jiffies;
sbi->last_time[type] = now;
/* DISCARD_TIME and GC_TIME are based on REQ_TIME */
if (type == REQ_TIME) {
sbi->last_time[DISCARD_TIME] = now;
sbi->last_time[GC_TIME] = now;
}
}
static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type)
{
unsigned long interval = sbi->interval_time[type] * HZ;
return time_after(jiffies, sbi->last_time[type] + interval);
}
static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi,
int type)
{
unsigned long interval = sbi->interval_time[type] * HZ;
unsigned int wait_ms = 0;
long delta;
delta = (sbi->last_time[type] + interval) - jiffies;
if (delta > 0)
wait_ms = jiffies_to_msecs(delta);
return wait_ms;
}
/*
* Inline functions
*/
static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc,
const void *address, unsigned int length)
{
struct {
struct shash_desc shash;
char ctx[4];
} desc;
int err;
BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx));
desc.shash.tfm = sbi->s_chksum_driver;
*(u32 *)desc.ctx = crc;
err = crypto_shash_update(&desc.shash, address, length);
BUG_ON(err);
return *(u32 *)desc.ctx;
}
static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address,
unsigned int length)
{
return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length);
}
static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc,
void *buf, size_t buf_size)
{
return f2fs_crc32(sbi, buf, buf_size) == blk_crc;
}
static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc,
const void *address, unsigned int length)
{
return __f2fs_crc32(sbi, crc, address, length);
}
static inline struct f2fs_inode_info *F2FS_I(struct inode *inode)
{
return container_of(inode, struct f2fs_inode_info, vfs_inode);
}
static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb)
{
return sb->s_fs_info;
}
static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode)
{
return F2FS_SB(inode->i_sb);
}
static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping)
{
return F2FS_I_SB(mapping->host);
}
static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page)
{
return F2FS_M_SB(page_file_mapping(page));
}
static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi)
{
return (struct f2fs_super_block *)(sbi->raw_super);
}
static inline struct f2fs_super_block *F2FS_SUPER_BLOCK(struct folio *folio,
pgoff_t index)
{
pgoff_t idx_in_folio = index % (1 << folio_order(folio));
return (struct f2fs_super_block *)
(page_address(folio_page(folio, idx_in_folio)) +
F2FS_SUPER_OFFSET);
}
static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi)
{
return (struct f2fs_checkpoint *)(sbi->ckpt);
}
static inline struct f2fs_node *F2FS_NODE(struct page *page)
{
return (struct f2fs_node *)page_address(page);
}
static inline struct f2fs_inode *F2FS_INODE(struct page *page)
{
return &((struct f2fs_node *)page_address(page))->i;
}
static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi)
{
return (struct f2fs_nm_info *)(sbi->nm_info);
}
static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi)
{
return (struct f2fs_sm_info *)(sbi->sm_info);
}
static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi)
{
return (struct sit_info *)(SM_I(sbi)->sit_info);
}
static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi)
{
return (struct free_segmap_info *)(SM_I(sbi)->free_info);
}
static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi)
{
return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info);
}
static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi)
{
return sbi->meta_inode->i_mapping;
}
static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi)
{
return sbi->node_inode->i_mapping;
}
static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type)
{
return test_bit(type, &sbi->s_flag);
}
static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
{
set_bit(type, &sbi->s_flag);
}
static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
{
clear_bit(type, &sbi->s_flag);
}
static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp)
{
return le64_to_cpu(cp->checkpoint_ver);
}
static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type)
{
if (type < F2FS_MAX_QUOTAS)
return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]);
return 0;
}
static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp)
{
size_t crc_offset = le32_to_cpu(cp->checksum_offset);
return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset)));
}
static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
{
unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
return ckpt_flags & f;
}
static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
{
return __is_set_ckpt_flags(F2FS_CKPT(sbi), f);
}
static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
{
unsigned int ckpt_flags;
ckpt_flags = le32_to_cpu(cp->ckpt_flags);
ckpt_flags |= f;
cp->ckpt_flags = cpu_to_le32(ckpt_flags);
}
static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
{
f2fs: use spin_{,un}lock_irq{save,restore} generic/361 reports below warning, this is because: once, there is someone entering into critical region of sbi.cp_lock, if write_end_io. f2fs_stop_checkpoint is invoked from an triggered IRQ, we will encounter deadlock. So this patch changes to use spin_{,un}lock_irq{save,restore} to create critical region without IRQ enabled to avoid potential deadlock. irq event stamp: 83391573 loop: Write error at byte offset 438729728, length 1024. hardirqs last enabled at (83391573): [<c1809752>] restore_all+0xf/0x65 hardirqs last disabled at (83391572): [<c1809eac>] reschedule_interrupt+0x30/0x3c loop: Write error at byte offset 438860288, length 1536. softirqs last enabled at (83389244): [<c180cc4e>] __do_softirq+0x1ae/0x476 softirqs last disabled at (83389237): [<c101ca7c>] do_softirq_own_stack+0x2c/0x40 loop: Write error at byte offset 438990848, length 2048. ================================ WARNING: inconsistent lock state 4.12.0-rc2+ #30 Tainted: G O -------------------------------- inconsistent {HARDIRQ-ON-W} -> {IN-HARDIRQ-W} usage. xfs_io/7959 [HC1[1]:SC0[0]:HE0:SE1] takes: (&(&sbi->cp_lock)->rlock){?.+...}, at: [<f96f96cc>] f2fs_stop_checkpoint+0x1c/0x50 [f2fs] {HARDIRQ-ON-W} state was registered at: __lock_acquire+0x527/0x7b0 lock_acquire+0xae/0x220 _raw_spin_lock+0x42/0x50 do_checkpoint+0x165/0x9e0 [f2fs] write_checkpoint+0x33f/0x740 [f2fs] __f2fs_sync_fs+0x92/0x1f0 [f2fs] f2fs_sync_fs+0x12/0x20 [f2fs] sync_filesystem+0x67/0x80 generic_shutdown_super+0x27/0x100 kill_block_super+0x22/0x50 kill_f2fs_super+0x3a/0x40 [f2fs] deactivate_locked_super+0x3d/0x70 deactivate_super+0x40/0x60 cleanup_mnt+0x39/0x70 __cleanup_mnt+0x10/0x20 task_work_run+0x69/0x80 exit_to_usermode_loop+0x57/0x85 do_fast_syscall_32+0x18c/0x1b0 entry_SYSENTER_32+0x4c/0x7b irq event stamp: 1957420 hardirqs last enabled at (1957419): [<c1808f37>] _raw_spin_unlock_irq+0x27/0x50 hardirqs last disabled at (1957420): [<c1809f9c>] call_function_single_interrupt+0x30/0x3c softirqs last enabled at (1953784): [<c180cc4e>] __do_softirq+0x1ae/0x476 softirqs last disabled at (1953773): [<c101ca7c>] do_softirq_own_stack+0x2c/0x40 other info that might help us debug this: Possible unsafe locking scenario: CPU0 ---- lock(&(&sbi->cp_lock)->rlock); <Interrupt> lock(&(&sbi->cp_lock)->rlock); *** DEADLOCK *** 2 locks held by xfs_io/7959: #0: (sb_writers#13){.+.+.+}, at: [<c11fd7ca>] vfs_write+0x16a/0x190 #1: (&sb->s_type->i_mutex_key#16){+.+.+.}, at: [<f96e33f5>] f2fs_file_write_iter+0x25/0x140 [f2fs] stack backtrace: CPU: 2 PID: 7959 Comm: xfs_io Tainted: G O 4.12.0-rc2+ #30 Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006 Call Trace: dump_stack+0x5f/0x92 print_usage_bug+0x1d3/0x1dd ? check_usage_backwards+0xe0/0xe0 mark_lock+0x23d/0x280 __lock_acquire+0x699/0x7b0 ? __this_cpu_preempt_check+0xf/0x20 ? trace_hardirqs_off_caller+0x91/0xe0 lock_acquire+0xae/0x220 ? f2fs_stop_checkpoint+0x1c/0x50 [f2fs] _raw_spin_lock+0x42/0x50 ? f2fs_stop_checkpoint+0x1c/0x50 [f2fs] f2fs_stop_checkpoint+0x1c/0x50 [f2fs] f2fs_write_end_io+0x147/0x150 [f2fs] bio_endio+0x7a/0x1e0 blk_update_request+0xad/0x410 blk_mq_end_request+0x16/0x60 lo_complete_rq+0x3c/0x70 __blk_mq_complete_request_remote+0x11/0x20 flush_smp_call_function_queue+0x6d/0x120 ? debug_smp_processor_id+0x12/0x20 generic_smp_call_function_single_interrupt+0x12/0x30 smp_call_function_single_interrupt+0x25/0x40 call_function_single_interrupt+0x37/0x3c EIP: _raw_spin_unlock_irq+0x2d/0x50 EFLAGS: 00000296 CPU: 2 EAX: 00000001 EBX: d2ccc51c ECX: 00000001 EDX: c1aacebd ESI: 00000000 EDI: 00000000 EBP: c96c9d1c ESP: c96c9d18 DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068 ? inherit_task_group.isra.98.part.99+0x6b/0xb0 __add_to_page_cache_locked+0x1d4/0x290 add_to_page_cache_lru+0x38/0xb0 pagecache_get_page+0x8e/0x200 f2fs_write_begin+0x96/0xf00 [f2fs] ? trace_hardirqs_on_caller+0xdd/0x1c0 ? current_time+0x17/0x50 ? trace_hardirqs_on+0xb/0x10 generic_perform_write+0xa9/0x170 __generic_file_write_iter+0x1a2/0x1f0 ? f2fs_preallocate_blocks+0x137/0x160 [f2fs] f2fs_file_write_iter+0x6e/0x140 [f2fs] ? __lock_acquire+0x429/0x7b0 __vfs_write+0xc1/0x140 vfs_write+0x9b/0x190 SyS_pwrite64+0x63/0xa0 do_fast_syscall_32+0xa1/0x1b0 entry_SYSENTER_32+0x4c/0x7b EIP: 0xb7786c61 EFLAGS: 00000293 CPU: 2 EAX: ffffffda EBX: 00000003 ECX: 08416000 EDX: 00001000 ESI: 18b24000 EDI: 00000000 EBP: 00000003 ESP: bf9b36b0 DS: 007b ES: 007b FS: 0000 GS: 0033 SS: 007b Fixes: aaec2b1d1879 ("f2fs: introduce cp_lock to protect updating of ckpt_flags") Cc: stable@vger.kernel.org Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-07-07 06:10:15 +00:00
unsigned long flags;
spin_lock_irqsave(&sbi->cp_lock, flags);
__set_ckpt_flags(F2FS_CKPT(sbi), f);
f2fs: use spin_{,un}lock_irq{save,restore} generic/361 reports below warning, this is because: once, there is someone entering into critical region of sbi.cp_lock, if write_end_io. f2fs_stop_checkpoint is invoked from an triggered IRQ, we will encounter deadlock. So this patch changes to use spin_{,un}lock_irq{save,restore} to create critical region without IRQ enabled to avoid potential deadlock. irq event stamp: 83391573 loop: Write error at byte offset 438729728, length 1024. hardirqs last enabled at (83391573): [<c1809752>] restore_all+0xf/0x65 hardirqs last disabled at (83391572): [<c1809eac>] reschedule_interrupt+0x30/0x3c loop: Write error at byte offset 438860288, length 1536. softirqs last enabled at (83389244): [<c180cc4e>] __do_softirq+0x1ae/0x476 softirqs last disabled at (83389237): [<c101ca7c>] do_softirq_own_stack+0x2c/0x40 loop: Write error at byte offset 438990848, length 2048. ================================ WARNING: inconsistent lock state 4.12.0-rc2+ #30 Tainted: G O -------------------------------- inconsistent {HARDIRQ-ON-W} -> {IN-HARDIRQ-W} usage. xfs_io/7959 [HC1[1]:SC0[0]:HE0:SE1] takes: (&(&sbi->cp_lock)->rlock){?.+...}, at: [<f96f96cc>] f2fs_stop_checkpoint+0x1c/0x50 [f2fs] {HARDIRQ-ON-W} state was registered at: __lock_acquire+0x527/0x7b0 lock_acquire+0xae/0x220 _raw_spin_lock+0x42/0x50 do_checkpoint+0x165/0x9e0 [f2fs] write_checkpoint+0x33f/0x740 [f2fs] __f2fs_sync_fs+0x92/0x1f0 [f2fs] f2fs_sync_fs+0x12/0x20 [f2fs] sync_filesystem+0x67/0x80 generic_shutdown_super+0x27/0x100 kill_block_super+0x22/0x50 kill_f2fs_super+0x3a/0x40 [f2fs] deactivate_locked_super+0x3d/0x70 deactivate_super+0x40/0x60 cleanup_mnt+0x39/0x70 __cleanup_mnt+0x10/0x20 task_work_run+0x69/0x80 exit_to_usermode_loop+0x57/0x85 do_fast_syscall_32+0x18c/0x1b0 entry_SYSENTER_32+0x4c/0x7b irq event stamp: 1957420 hardirqs last enabled at (1957419): [<c1808f37>] _raw_spin_unlock_irq+0x27/0x50 hardirqs last disabled at (1957420): [<c1809f9c>] call_function_single_interrupt+0x30/0x3c softirqs last enabled at (1953784): [<c180cc4e>] __do_softirq+0x1ae/0x476 softirqs last disabled at (1953773): [<c101ca7c>] do_softirq_own_stack+0x2c/0x40 other info that might help us debug this: Possible unsafe locking scenario: CPU0 ---- lock(&(&sbi->cp_lock)->rlock); <Interrupt> lock(&(&sbi->cp_lock)->rlock); *** DEADLOCK *** 2 locks held by xfs_io/7959: #0: (sb_writers#13){.+.+.+}, at: [<c11fd7ca>] vfs_write+0x16a/0x190 #1: (&sb->s_type->i_mutex_key#16){+.+.+.}, at: [<f96e33f5>] f2fs_file_write_iter+0x25/0x140 [f2fs] stack backtrace: CPU: 2 PID: 7959 Comm: xfs_io Tainted: G O 4.12.0-rc2+ #30 Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006 Call Trace: dump_stack+0x5f/0x92 print_usage_bug+0x1d3/0x1dd ? check_usage_backwards+0xe0/0xe0 mark_lock+0x23d/0x280 __lock_acquire+0x699/0x7b0 ? __this_cpu_preempt_check+0xf/0x20 ? trace_hardirqs_off_caller+0x91/0xe0 lock_acquire+0xae/0x220 ? f2fs_stop_checkpoint+0x1c/0x50 [f2fs] _raw_spin_lock+0x42/0x50 ? f2fs_stop_checkpoint+0x1c/0x50 [f2fs] f2fs_stop_checkpoint+0x1c/0x50 [f2fs] f2fs_write_end_io+0x147/0x150 [f2fs] bio_endio+0x7a/0x1e0 blk_update_request+0xad/0x410 blk_mq_end_request+0x16/0x60 lo_complete_rq+0x3c/0x70 __blk_mq_complete_request_remote+0x11/0x20 flush_smp_call_function_queue+0x6d/0x120 ? debug_smp_processor_id+0x12/0x20 generic_smp_call_function_single_interrupt+0x12/0x30 smp_call_function_single_interrupt+0x25/0x40 call_function_single_interrupt+0x37/0x3c EIP: _raw_spin_unlock_irq+0x2d/0x50 EFLAGS: 00000296 CPU: 2 EAX: 00000001 EBX: d2ccc51c ECX: 00000001 EDX: c1aacebd ESI: 00000000 EDI: 00000000 EBP: c96c9d1c ESP: c96c9d18 DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068 ? inherit_task_group.isra.98.part.99+0x6b/0xb0 __add_to_page_cache_locked+0x1d4/0x290 add_to_page_cache_lru+0x38/0xb0 pagecache_get_page+0x8e/0x200 f2fs_write_begin+0x96/0xf00 [f2fs] ? trace_hardirqs_on_caller+0xdd/0x1c0 ? current_time+0x17/0x50 ? trace_hardirqs_on+0xb/0x10 generic_perform_write+0xa9/0x170 __generic_file_write_iter+0x1a2/0x1f0 ? f2fs_preallocate_blocks+0x137/0x160 [f2fs] f2fs_file_write_iter+0x6e/0x140 [f2fs] ? __lock_acquire+0x429/0x7b0 __vfs_write+0xc1/0x140 vfs_write+0x9b/0x190 SyS_pwrite64+0x63/0xa0 do_fast_syscall_32+0xa1/0x1b0 entry_SYSENTER_32+0x4c/0x7b EIP: 0xb7786c61 EFLAGS: 00000293 CPU: 2 EAX: ffffffda EBX: 00000003 ECX: 08416000 EDX: 00001000 ESI: 18b24000 EDI: 00000000 EBP: 00000003 ESP: bf9b36b0 DS: 007b ES: 007b FS: 0000 GS: 0033 SS: 007b Fixes: aaec2b1d1879 ("f2fs: introduce cp_lock to protect updating of ckpt_flags") Cc: stable@vger.kernel.org Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-07-07 06:10:15 +00:00
spin_unlock_irqrestore(&sbi->cp_lock, flags);
}
static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
{
unsigned int ckpt_flags;
ckpt_flags = le32_to_cpu(cp->ckpt_flags);
ckpt_flags &= (~f);
cp->ckpt_flags = cpu_to_le32(ckpt_flags);
}
static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
{
f2fs: use spin_{,un}lock_irq{save,restore} generic/361 reports below warning, this is because: once, there is someone entering into critical region of sbi.cp_lock, if write_end_io. f2fs_stop_checkpoint is invoked from an triggered IRQ, we will encounter deadlock. So this patch changes to use spin_{,un}lock_irq{save,restore} to create critical region without IRQ enabled to avoid potential deadlock. irq event stamp: 83391573 loop: Write error at byte offset 438729728, length 1024. hardirqs last enabled at (83391573): [<c1809752>] restore_all+0xf/0x65 hardirqs last disabled at (83391572): [<c1809eac>] reschedule_interrupt+0x30/0x3c loop: Write error at byte offset 438860288, length 1536. softirqs last enabled at (83389244): [<c180cc4e>] __do_softirq+0x1ae/0x476 softirqs last disabled at (83389237): [<c101ca7c>] do_softirq_own_stack+0x2c/0x40 loop: Write error at byte offset 438990848, length 2048. ================================ WARNING: inconsistent lock state 4.12.0-rc2+ #30 Tainted: G O -------------------------------- inconsistent {HARDIRQ-ON-W} -> {IN-HARDIRQ-W} usage. xfs_io/7959 [HC1[1]:SC0[0]:HE0:SE1] takes: (&(&sbi->cp_lock)->rlock){?.+...}, at: [<f96f96cc>] f2fs_stop_checkpoint+0x1c/0x50 [f2fs] {HARDIRQ-ON-W} state was registered at: __lock_acquire+0x527/0x7b0 lock_acquire+0xae/0x220 _raw_spin_lock+0x42/0x50 do_checkpoint+0x165/0x9e0 [f2fs] write_checkpoint+0x33f/0x740 [f2fs] __f2fs_sync_fs+0x92/0x1f0 [f2fs] f2fs_sync_fs+0x12/0x20 [f2fs] sync_filesystem+0x67/0x80 generic_shutdown_super+0x27/0x100 kill_block_super+0x22/0x50 kill_f2fs_super+0x3a/0x40 [f2fs] deactivate_locked_super+0x3d/0x70 deactivate_super+0x40/0x60 cleanup_mnt+0x39/0x70 __cleanup_mnt+0x10/0x20 task_work_run+0x69/0x80 exit_to_usermode_loop+0x57/0x85 do_fast_syscall_32+0x18c/0x1b0 entry_SYSENTER_32+0x4c/0x7b irq event stamp: 1957420 hardirqs last enabled at (1957419): [<c1808f37>] _raw_spin_unlock_irq+0x27/0x50 hardirqs last disabled at (1957420): [<c1809f9c>] call_function_single_interrupt+0x30/0x3c softirqs last enabled at (1953784): [<c180cc4e>] __do_softirq+0x1ae/0x476 softirqs last disabled at (1953773): [<c101ca7c>] do_softirq_own_stack+0x2c/0x40 other info that might help us debug this: Possible unsafe locking scenario: CPU0 ---- lock(&(&sbi->cp_lock)->rlock); <Interrupt> lock(&(&sbi->cp_lock)->rlock); *** DEADLOCK *** 2 locks held by xfs_io/7959: #0: (sb_writers#13){.+.+.+}, at: [<c11fd7ca>] vfs_write+0x16a/0x190 #1: (&sb->s_type->i_mutex_key#16){+.+.+.}, at: [<f96e33f5>] f2fs_file_write_iter+0x25/0x140 [f2fs] stack backtrace: CPU: 2 PID: 7959 Comm: xfs_io Tainted: G O 4.12.0-rc2+ #30 Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006 Call Trace: dump_stack+0x5f/0x92 print_usage_bug+0x1d3/0x1dd ? check_usage_backwards+0xe0/0xe0 mark_lock+0x23d/0x280 __lock_acquire+0x699/0x7b0 ? __this_cpu_preempt_check+0xf/0x20 ? trace_hardirqs_off_caller+0x91/0xe0 lock_acquire+0xae/0x220 ? f2fs_stop_checkpoint+0x1c/0x50 [f2fs] _raw_spin_lock+0x42/0x50 ? f2fs_stop_checkpoint+0x1c/0x50 [f2fs] f2fs_stop_checkpoint+0x1c/0x50 [f2fs] f2fs_write_end_io+0x147/0x150 [f2fs] bio_endio+0x7a/0x1e0 blk_update_request+0xad/0x410 blk_mq_end_request+0x16/0x60 lo_complete_rq+0x3c/0x70 __blk_mq_complete_request_remote+0x11/0x20 flush_smp_call_function_queue+0x6d/0x120 ? debug_smp_processor_id+0x12/0x20 generic_smp_call_function_single_interrupt+0x12/0x30 smp_call_function_single_interrupt+0x25/0x40 call_function_single_interrupt+0x37/0x3c EIP: _raw_spin_unlock_irq+0x2d/0x50 EFLAGS: 00000296 CPU: 2 EAX: 00000001 EBX: d2ccc51c ECX: 00000001 EDX: c1aacebd ESI: 00000000 EDI: 00000000 EBP: c96c9d1c ESP: c96c9d18 DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068 ? inherit_task_group.isra.98.part.99+0x6b/0xb0 __add_to_page_cache_locked+0x1d4/0x290 add_to_page_cache_lru+0x38/0xb0 pagecache_get_page+0x8e/0x200 f2fs_write_begin+0x96/0xf00 [f2fs] ? trace_hardirqs_on_caller+0xdd/0x1c0 ? current_time+0x17/0x50 ? trace_hardirqs_on+0xb/0x10 generic_perform_write+0xa9/0x170 __generic_file_write_iter+0x1a2/0x1f0 ? f2fs_preallocate_blocks+0x137/0x160 [f2fs] f2fs_file_write_iter+0x6e/0x140 [f2fs] ? __lock_acquire+0x429/0x7b0 __vfs_write+0xc1/0x140 vfs_write+0x9b/0x190 SyS_pwrite64+0x63/0xa0 do_fast_syscall_32+0xa1/0x1b0 entry_SYSENTER_32+0x4c/0x7b EIP: 0xb7786c61 EFLAGS: 00000293 CPU: 2 EAX: ffffffda EBX: 00000003 ECX: 08416000 EDX: 00001000 ESI: 18b24000 EDI: 00000000 EBP: 00000003 ESP: bf9b36b0 DS: 007b ES: 007b FS: 0000 GS: 0033 SS: 007b Fixes: aaec2b1d1879 ("f2fs: introduce cp_lock to protect updating of ckpt_flags") Cc: stable@vger.kernel.org Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-07-07 06:10:15 +00:00
unsigned long flags;
spin_lock_irqsave(&sbi->cp_lock, flags);
__clear_ckpt_flags(F2FS_CKPT(sbi), f);
f2fs: use spin_{,un}lock_irq{save,restore} generic/361 reports below warning, this is because: once, there is someone entering into critical region of sbi.cp_lock, if write_end_io. f2fs_stop_checkpoint is invoked from an triggered IRQ, we will encounter deadlock. So this patch changes to use spin_{,un}lock_irq{save,restore} to create critical region without IRQ enabled to avoid potential deadlock. irq event stamp: 83391573 loop: Write error at byte offset 438729728, length 1024. hardirqs last enabled at (83391573): [<c1809752>] restore_all+0xf/0x65 hardirqs last disabled at (83391572): [<c1809eac>] reschedule_interrupt+0x30/0x3c loop: Write error at byte offset 438860288, length 1536. softirqs last enabled at (83389244): [<c180cc4e>] __do_softirq+0x1ae/0x476 softirqs last disabled at (83389237): [<c101ca7c>] do_softirq_own_stack+0x2c/0x40 loop: Write error at byte offset 438990848, length 2048. ================================ WARNING: inconsistent lock state 4.12.0-rc2+ #30 Tainted: G O -------------------------------- inconsistent {HARDIRQ-ON-W} -> {IN-HARDIRQ-W} usage. xfs_io/7959 [HC1[1]:SC0[0]:HE0:SE1] takes: (&(&sbi->cp_lock)->rlock){?.+...}, at: [<f96f96cc>] f2fs_stop_checkpoint+0x1c/0x50 [f2fs] {HARDIRQ-ON-W} state was registered at: __lock_acquire+0x527/0x7b0 lock_acquire+0xae/0x220 _raw_spin_lock+0x42/0x50 do_checkpoint+0x165/0x9e0 [f2fs] write_checkpoint+0x33f/0x740 [f2fs] __f2fs_sync_fs+0x92/0x1f0 [f2fs] f2fs_sync_fs+0x12/0x20 [f2fs] sync_filesystem+0x67/0x80 generic_shutdown_super+0x27/0x100 kill_block_super+0x22/0x50 kill_f2fs_super+0x3a/0x40 [f2fs] deactivate_locked_super+0x3d/0x70 deactivate_super+0x40/0x60 cleanup_mnt+0x39/0x70 __cleanup_mnt+0x10/0x20 task_work_run+0x69/0x80 exit_to_usermode_loop+0x57/0x85 do_fast_syscall_32+0x18c/0x1b0 entry_SYSENTER_32+0x4c/0x7b irq event stamp: 1957420 hardirqs last enabled at (1957419): [<c1808f37>] _raw_spin_unlock_irq+0x27/0x50 hardirqs last disabled at (1957420): [<c1809f9c>] call_function_single_interrupt+0x30/0x3c softirqs last enabled at (1953784): [<c180cc4e>] __do_softirq+0x1ae/0x476 softirqs last disabled at (1953773): [<c101ca7c>] do_softirq_own_stack+0x2c/0x40 other info that might help us debug this: Possible unsafe locking scenario: CPU0 ---- lock(&(&sbi->cp_lock)->rlock); <Interrupt> lock(&(&sbi->cp_lock)->rlock); *** DEADLOCK *** 2 locks held by xfs_io/7959: #0: (sb_writers#13){.+.+.+}, at: [<c11fd7ca>] vfs_write+0x16a/0x190 #1: (&sb->s_type->i_mutex_key#16){+.+.+.}, at: [<f96e33f5>] f2fs_file_write_iter+0x25/0x140 [f2fs] stack backtrace: CPU: 2 PID: 7959 Comm: xfs_io Tainted: G O 4.12.0-rc2+ #30 Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006 Call Trace: dump_stack+0x5f/0x92 print_usage_bug+0x1d3/0x1dd ? check_usage_backwards+0xe0/0xe0 mark_lock+0x23d/0x280 __lock_acquire+0x699/0x7b0 ? __this_cpu_preempt_check+0xf/0x20 ? trace_hardirqs_off_caller+0x91/0xe0 lock_acquire+0xae/0x220 ? f2fs_stop_checkpoint+0x1c/0x50 [f2fs] _raw_spin_lock+0x42/0x50 ? f2fs_stop_checkpoint+0x1c/0x50 [f2fs] f2fs_stop_checkpoint+0x1c/0x50 [f2fs] f2fs_write_end_io+0x147/0x150 [f2fs] bio_endio+0x7a/0x1e0 blk_update_request+0xad/0x410 blk_mq_end_request+0x16/0x60 lo_complete_rq+0x3c/0x70 __blk_mq_complete_request_remote+0x11/0x20 flush_smp_call_function_queue+0x6d/0x120 ? debug_smp_processor_id+0x12/0x20 generic_smp_call_function_single_interrupt+0x12/0x30 smp_call_function_single_interrupt+0x25/0x40 call_function_single_interrupt+0x37/0x3c EIP: _raw_spin_unlock_irq+0x2d/0x50 EFLAGS: 00000296 CPU: 2 EAX: 00000001 EBX: d2ccc51c ECX: 00000001 EDX: c1aacebd ESI: 00000000 EDI: 00000000 EBP: c96c9d1c ESP: c96c9d18 DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068 ? inherit_task_group.isra.98.part.99+0x6b/0xb0 __add_to_page_cache_locked+0x1d4/0x290 add_to_page_cache_lru+0x38/0xb0 pagecache_get_page+0x8e/0x200 f2fs_write_begin+0x96/0xf00 [f2fs] ? trace_hardirqs_on_caller+0xdd/0x1c0 ? current_time+0x17/0x50 ? trace_hardirqs_on+0xb/0x10 generic_perform_write+0xa9/0x170 __generic_file_write_iter+0x1a2/0x1f0 ? f2fs_preallocate_blocks+0x137/0x160 [f2fs] f2fs_file_write_iter+0x6e/0x140 [f2fs] ? __lock_acquire+0x429/0x7b0 __vfs_write+0xc1/0x140 vfs_write+0x9b/0x190 SyS_pwrite64+0x63/0xa0 do_fast_syscall_32+0xa1/0x1b0 entry_SYSENTER_32+0x4c/0x7b EIP: 0xb7786c61 EFLAGS: 00000293 CPU: 2 EAX: ffffffda EBX: 00000003 ECX: 08416000 EDX: 00001000 ESI: 18b24000 EDI: 00000000 EBP: 00000003 ESP: bf9b36b0 DS: 007b ES: 007b FS: 0000 GS: 0033 SS: 007b Fixes: aaec2b1d1879 ("f2fs: introduce cp_lock to protect updating of ckpt_flags") Cc: stable@vger.kernel.org Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-07-07 06:10:15 +00:00
spin_unlock_irqrestore(&sbi->cp_lock, flags);
}
#define init_f2fs_rwsem(sem) \
do { \
static struct lock_class_key __key; \
\
__init_f2fs_rwsem((sem), #sem, &__key); \
} while (0)
static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem,
const char *sem_name, struct lock_class_key *key)
{
__init_rwsem(&sem->internal_rwsem, sem_name, key);
#ifdef CONFIG_F2FS_UNFAIR_RWSEM
init_waitqueue_head(&sem->read_waiters);
#endif
}
static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem)
{
return rwsem_is_locked(&sem->internal_rwsem);
}
static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem)
{
return rwsem_is_contended(&sem->internal_rwsem);
}
static inline void f2fs_down_read(struct f2fs_rwsem *sem)
{
#ifdef CONFIG_F2FS_UNFAIR_RWSEM
wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem));
#else
down_read(&sem->internal_rwsem);
#endif
}
static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem)
{
return down_read_trylock(&sem->internal_rwsem);
}
static inline void f2fs_up_read(struct f2fs_rwsem *sem)
{
up_read(&sem->internal_rwsem);
}
static inline void f2fs_down_write(struct f2fs_rwsem *sem)
{
down_write(&sem->internal_rwsem);
}
f2fs: avoid false alarm of circular locking ====================================================== WARNING: possible circular locking dependency detected 6.5.0-rc5-syzkaller-00353-gae545c3283dc #0 Not tainted ------------------------------------------------------ syz-executor273/5027 is trying to acquire lock: ffff888077fe1fb0 (&fi->i_sem){+.+.}-{3:3}, at: f2fs_down_write fs/f2fs/f2fs.h:2133 [inline] ffff888077fe1fb0 (&fi->i_sem){+.+.}-{3:3}, at: f2fs_add_inline_entry+0x300/0x6f0 fs/f2fs/inline.c:644 but task is already holding lock: ffff888077fe07c8 (&fi->i_xattr_sem){.+.+}-{3:3}, at: f2fs_down_read fs/f2fs/f2fs.h:2108 [inline] ffff888077fe07c8 (&fi->i_xattr_sem){.+.+}-{3:3}, at: f2fs_add_dentry+0x92/0x230 fs/f2fs/dir.c:783 which lock already depends on the new lock. the existing dependency chain (in reverse order) is: -> #1 (&fi->i_xattr_sem){.+.+}-{3:3}: down_read+0x9c/0x470 kernel/locking/rwsem.c:1520 f2fs_down_read fs/f2fs/f2fs.h:2108 [inline] f2fs_getxattr+0xb1e/0x12c0 fs/f2fs/xattr.c:532 __f2fs_get_acl+0x5a/0x900 fs/f2fs/acl.c:179 f2fs_acl_create fs/f2fs/acl.c:377 [inline] f2fs_init_acl+0x15c/0xb30 fs/f2fs/acl.c:420 f2fs_init_inode_metadata+0x159/0x1290 fs/f2fs/dir.c:558 f2fs_add_regular_entry+0x79e/0xb90 fs/f2fs/dir.c:740 f2fs_add_dentry+0x1de/0x230 fs/f2fs/dir.c:788 f2fs_do_add_link+0x190/0x280 fs/f2fs/dir.c:827 f2fs_add_link fs/f2fs/f2fs.h:3554 [inline] f2fs_mkdir+0x377/0x620 fs/f2fs/namei.c:781 vfs_mkdir+0x532/0x7e0 fs/namei.c:4117 do_mkdirat+0x2a9/0x330 fs/namei.c:4140 __do_sys_mkdir fs/namei.c:4160 [inline] __se_sys_mkdir fs/namei.c:4158 [inline] __x64_sys_mkdir+0xf2/0x140 fs/namei.c:4158 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x38/0xb0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd -> #0 (&fi->i_sem){+.+.}-{3:3}: check_prev_add kernel/locking/lockdep.c:3142 [inline] check_prevs_add kernel/locking/lockdep.c:3261 [inline] validate_chain kernel/locking/lockdep.c:3876 [inline] __lock_acquire+0x2e3d/0x5de0 kernel/locking/lockdep.c:5144 lock_acquire kernel/locking/lockdep.c:5761 [inline] lock_acquire+0x1ae/0x510 kernel/locking/lockdep.c:5726 down_write+0x93/0x200 kernel/locking/rwsem.c:1573 f2fs_down_write fs/f2fs/f2fs.h:2133 [inline] f2fs_add_inline_entry+0x300/0x6f0 fs/f2fs/inline.c:644 f2fs_add_dentry+0xa6/0x230 fs/f2fs/dir.c:784 f2fs_do_add_link+0x190/0x280 fs/f2fs/dir.c:827 f2fs_add_link fs/f2fs/f2fs.h:3554 [inline] f2fs_mkdir+0x377/0x620 fs/f2fs/namei.c:781 vfs_mkdir+0x532/0x7e0 fs/namei.c:4117 ovl_do_mkdir fs/overlayfs/overlayfs.h:196 [inline] ovl_mkdir_real+0xb5/0x370 fs/overlayfs/dir.c:146 ovl_workdir_create+0x3de/0x820 fs/overlayfs/super.c:309 ovl_make_workdir fs/overlayfs/super.c:711 [inline] ovl_get_workdir fs/overlayfs/super.c:864 [inline] ovl_fill_super+0xdab/0x6180 fs/overlayfs/super.c:1400 vfs_get_super+0xf9/0x290 fs/super.c:1152 vfs_get_tree+0x88/0x350 fs/super.c:1519 do_new_mount fs/namespace.c:3335 [inline] path_mount+0x1492/0x1ed0 fs/namespace.c:3662 do_mount fs/namespace.c:3675 [inline] __do_sys_mount fs/namespace.c:3884 [inline] __se_sys_mount fs/namespace.c:3861 [inline] __x64_sys_mount+0x293/0x310 fs/namespace.c:3861 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x38/0xb0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd other info that might help us debug this: Possible unsafe locking scenario: CPU0 CPU1 ---- ---- rlock(&fi->i_xattr_sem); lock(&fi->i_sem); lock(&fi->i_xattr_sem); lock(&fi->i_sem); Cc: <stable@vger.kernel.org> Reported-and-tested-by: syzbot+e5600587fa9cbf8e3826@syzkaller.appspotmail.com Fixes: 5eda1ad1aaff "f2fs: fix deadlock in i_xattr_sem and inode page lock" Tested-by: Guenter Roeck <linux@roeck-us.net> Reviewed-by: Chao Yu <chao@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2023-08-18 18:34:32 +00:00
#ifdef CONFIG_DEBUG_LOCK_ALLOC
static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass)
{
down_read_nested(&sem->internal_rwsem, subclass);
}
static inline void f2fs_down_write_nested(struct f2fs_rwsem *sem, int subclass)
{
down_write_nested(&sem->internal_rwsem, subclass);
}
#else
#define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem)
#define f2fs_down_write_nested(sem, subclass) f2fs_down_write(sem)
#endif
static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem)
{
return down_write_trylock(&sem->internal_rwsem);
}
static inline void f2fs_up_write(struct f2fs_rwsem *sem)
{
up_write(&sem->internal_rwsem);
#ifdef CONFIG_F2FS_UNFAIR_RWSEM
wake_up_all(&sem->read_waiters);
#endif
}
f2fs: use rw_sem instead of fs_lock(locks mutex) The fs_locks is used to block other ops(ex, recovery) when doing checkpoint. And each other operate routine(besides checkpoint) needs to acquire a fs_lock, there is a terrible problem here, if these are too many concurrency threads acquiring fs_lock, so that they will block each other and may lead to some performance problem, but this is not the phenomenon we want to see. Though there are some optimization patches introduced to enhance the usage of fs_lock, but the thorough solution is using a *rw_sem* to replace the fs_lock. Checkpoint routine takes write_sem, and other ops take read_sem, so that we can block other ops(ex, recovery) when doing checkpoint, and other ops will not disturb each other, this can avoid the problem described above completely. Because of the weakness of rw_sem, the above change may introduce a potential problem that the checkpoint thread might get starved if other threads are intensively locking the read semaphore for I/O.(Pointed out by Xu Jin) In order to avoid this, a wait_list is introduced, the appending read semaphore ops will be dropped into the wait_list if checkpoint thread is waiting for write semaphore, and will be waked up when checkpoint thread gives up write semaphore. Thanks to Kim's previous review and test, and will be very glad to see other guys' performance tests about this patch. V2: -fix the potential starvation problem. -use more suitable func name suggested by Xu Jin. Signed-off-by: Gu Zheng <guz.fnst@cn.fujitsu.com> [Jaegeuk Kim: adjust minor coding standard] Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-09-27 10:08:30 +00:00
static inline void f2fs_lock_op(struct f2fs_sb_info *sbi)
f2fs: introduce a new global lock scheme In the previous version, f2fs uses global locks according to the usage types, such as directory operations, block allocation, block write, and so on. Reference the following lock types in f2fs.h. enum lock_type { RENAME, /* for renaming operations */ DENTRY_OPS, /* for directory operations */ DATA_WRITE, /* for data write */ DATA_NEW, /* for data allocation */ DATA_TRUNC, /* for data truncate */ NODE_NEW, /* for node allocation */ NODE_TRUNC, /* for node truncate */ NODE_WRITE, /* for node write */ NR_LOCK_TYPE, }; In that case, we lose the performance under the multi-threading environment, since every types of operations must be conducted one at a time. In order to address the problem, let's share the locks globally with a mutex array regardless of any types. So, let users grab a mutex and perform their jobs in parallel as much as possbile. For this, I propose a new global lock scheme as follows. 0. Data structure - f2fs_sb_info -> mutex_lock[NR_GLOBAL_LOCKS] - f2fs_sb_info -> node_write 1. mutex_lock_op(sbi) - try to get an avaiable lock from the array. - returns the index of the gottern lock variable. 2. mutex_unlock_op(sbi, index of the lock) - unlock the given index of the lock. 3. mutex_lock_all(sbi) - grab all the locks in the array before the checkpoint. 4. mutex_unlock_all(sbi) - release all the locks in the array after checkpoint. 5. block_operations() - call mutex_lock_all() - sync_dirty_dir_inodes() - grab node_write - sync_node_pages() Note that, the pairs of mutex_lock_op()/mutex_unlock_op() and mutex_lock_all()/mutex_unlock_all() should be used together. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-22 07:21:29 +00:00
{
f2fs_down_read(&sbi->cp_rwsem);
f2fs: introduce a new global lock scheme In the previous version, f2fs uses global locks according to the usage types, such as directory operations, block allocation, block write, and so on. Reference the following lock types in f2fs.h. enum lock_type { RENAME, /* for renaming operations */ DENTRY_OPS, /* for directory operations */ DATA_WRITE, /* for data write */ DATA_NEW, /* for data allocation */ DATA_TRUNC, /* for data truncate */ NODE_NEW, /* for node allocation */ NODE_TRUNC, /* for node truncate */ NODE_WRITE, /* for node write */ NR_LOCK_TYPE, }; In that case, we lose the performance under the multi-threading environment, since every types of operations must be conducted one at a time. In order to address the problem, let's share the locks globally with a mutex array regardless of any types. So, let users grab a mutex and perform their jobs in parallel as much as possbile. For this, I propose a new global lock scheme as follows. 0. Data structure - f2fs_sb_info -> mutex_lock[NR_GLOBAL_LOCKS] - f2fs_sb_info -> node_write 1. mutex_lock_op(sbi) - try to get an avaiable lock from the array. - returns the index of the gottern lock variable. 2. mutex_unlock_op(sbi, index of the lock) - unlock the given index of the lock. 3. mutex_lock_all(sbi) - grab all the locks in the array before the checkpoint. 4. mutex_unlock_all(sbi) - release all the locks in the array after checkpoint. 5. block_operations() - call mutex_lock_all() - sync_dirty_dir_inodes() - grab node_write - sync_node_pages() Note that, the pairs of mutex_lock_op()/mutex_unlock_op() and mutex_lock_all()/mutex_unlock_all() should be used together. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-22 07:21:29 +00:00
}
static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi)
{
if (time_to_inject(sbi, FAULT_LOCK_OP))
return 0;
return f2fs_down_read_trylock(&sbi->cp_rwsem);
}
f2fs: use rw_sem instead of fs_lock(locks mutex) The fs_locks is used to block other ops(ex, recovery) when doing checkpoint. And each other operate routine(besides checkpoint) needs to acquire a fs_lock, there is a terrible problem here, if these are too many concurrency threads acquiring fs_lock, so that they will block each other and may lead to some performance problem, but this is not the phenomenon we want to see. Though there are some optimization patches introduced to enhance the usage of fs_lock, but the thorough solution is using a *rw_sem* to replace the fs_lock. Checkpoint routine takes write_sem, and other ops take read_sem, so that we can block other ops(ex, recovery) when doing checkpoint, and other ops will not disturb each other, this can avoid the problem described above completely. Because of the weakness of rw_sem, the above change may introduce a potential problem that the checkpoint thread might get starved if other threads are intensively locking the read semaphore for I/O.(Pointed out by Xu Jin) In order to avoid this, a wait_list is introduced, the appending read semaphore ops will be dropped into the wait_list if checkpoint thread is waiting for write semaphore, and will be waked up when checkpoint thread gives up write semaphore. Thanks to Kim's previous review and test, and will be very glad to see other guys' performance tests about this patch. V2: -fix the potential starvation problem. -use more suitable func name suggested by Xu Jin. Signed-off-by: Gu Zheng <guz.fnst@cn.fujitsu.com> [Jaegeuk Kim: adjust minor coding standard] Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-09-27 10:08:30 +00:00
static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi)
{
f2fs_up_read(&sbi->cp_rwsem);
}
f2fs: use rw_sem instead of fs_lock(locks mutex) The fs_locks is used to block other ops(ex, recovery) when doing checkpoint. And each other operate routine(besides checkpoint) needs to acquire a fs_lock, there is a terrible problem here, if these are too many concurrency threads acquiring fs_lock, so that they will block each other and may lead to some performance problem, but this is not the phenomenon we want to see. Though there are some optimization patches introduced to enhance the usage of fs_lock, but the thorough solution is using a *rw_sem* to replace the fs_lock. Checkpoint routine takes write_sem, and other ops take read_sem, so that we can block other ops(ex, recovery) when doing checkpoint, and other ops will not disturb each other, this can avoid the problem described above completely. Because of the weakness of rw_sem, the above change may introduce a potential problem that the checkpoint thread might get starved if other threads are intensively locking the read semaphore for I/O.(Pointed out by Xu Jin) In order to avoid this, a wait_list is introduced, the appending read semaphore ops will be dropped into the wait_list if checkpoint thread is waiting for write semaphore, and will be waked up when checkpoint thread gives up write semaphore. Thanks to Kim's previous review and test, and will be very glad to see other guys' performance tests about this patch. V2: -fix the potential starvation problem. -use more suitable func name suggested by Xu Jin. Signed-off-by: Gu Zheng <guz.fnst@cn.fujitsu.com> [Jaegeuk Kim: adjust minor coding standard] Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-09-27 10:08:30 +00:00
static inline void f2fs_lock_all(struct f2fs_sb_info *sbi)
{
f2fs_down_write(&sbi->cp_rwsem);
f2fs: introduce a new global lock scheme In the previous version, f2fs uses global locks according to the usage types, such as directory operations, block allocation, block write, and so on. Reference the following lock types in f2fs.h. enum lock_type { RENAME, /* for renaming operations */ DENTRY_OPS, /* for directory operations */ DATA_WRITE, /* for data write */ DATA_NEW, /* for data allocation */ DATA_TRUNC, /* for data truncate */ NODE_NEW, /* for node allocation */ NODE_TRUNC, /* for node truncate */ NODE_WRITE, /* for node write */ NR_LOCK_TYPE, }; In that case, we lose the performance under the multi-threading environment, since every types of operations must be conducted one at a time. In order to address the problem, let's share the locks globally with a mutex array regardless of any types. So, let users grab a mutex and perform their jobs in parallel as much as possbile. For this, I propose a new global lock scheme as follows. 0. Data structure - f2fs_sb_info -> mutex_lock[NR_GLOBAL_LOCKS] - f2fs_sb_info -> node_write 1. mutex_lock_op(sbi) - try to get an avaiable lock from the array. - returns the index of the gottern lock variable. 2. mutex_unlock_op(sbi, index of the lock) - unlock the given index of the lock. 3. mutex_lock_all(sbi) - grab all the locks in the array before the checkpoint. 4. mutex_unlock_all(sbi) - release all the locks in the array after checkpoint. 5. block_operations() - call mutex_lock_all() - sync_dirty_dir_inodes() - grab node_write - sync_node_pages() Note that, the pairs of mutex_lock_op()/mutex_unlock_op() and mutex_lock_all()/mutex_unlock_all() should be used together. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-22 07:21:29 +00:00
}
f2fs: use rw_sem instead of fs_lock(locks mutex) The fs_locks is used to block other ops(ex, recovery) when doing checkpoint. And each other operate routine(besides checkpoint) needs to acquire a fs_lock, there is a terrible problem here, if these are too many concurrency threads acquiring fs_lock, so that they will block each other and may lead to some performance problem, but this is not the phenomenon we want to see. Though there are some optimization patches introduced to enhance the usage of fs_lock, but the thorough solution is using a *rw_sem* to replace the fs_lock. Checkpoint routine takes write_sem, and other ops take read_sem, so that we can block other ops(ex, recovery) when doing checkpoint, and other ops will not disturb each other, this can avoid the problem described above completely. Because of the weakness of rw_sem, the above change may introduce a potential problem that the checkpoint thread might get starved if other threads are intensively locking the read semaphore for I/O.(Pointed out by Xu Jin) In order to avoid this, a wait_list is introduced, the appending read semaphore ops will be dropped into the wait_list if checkpoint thread is waiting for write semaphore, and will be waked up when checkpoint thread gives up write semaphore. Thanks to Kim's previous review and test, and will be very glad to see other guys' performance tests about this patch. V2: -fix the potential starvation problem. -use more suitable func name suggested by Xu Jin. Signed-off-by: Gu Zheng <guz.fnst@cn.fujitsu.com> [Jaegeuk Kim: adjust minor coding standard] Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2013-09-27 10:08:30 +00:00
static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi)
f2fs: introduce a new global lock scheme In the previous version, f2fs uses global locks according to the usage types, such as directory operations, block allocation, block write, and so on. Reference the following lock types in f2fs.h. enum lock_type { RENAME, /* for renaming operations */ DENTRY_OPS, /* for directory operations */ DATA_WRITE, /* for data write */ DATA_NEW, /* for data allocation */ DATA_TRUNC, /* for data truncate */ NODE_NEW, /* for node allocation */ NODE_TRUNC, /* for node truncate */ NODE_WRITE, /* for node write */ NR_LOCK_TYPE, }; In that case, we lose the performance under the multi-threading environment, since every types of operations must be conducted one at a time. In order to address the problem, let's share the locks globally with a mutex array regardless of any types. So, let users grab a mutex and perform their jobs in parallel as much as possbile. For this, I propose a new global lock scheme as follows. 0. Data structure - f2fs_sb_info -> mutex_lock[NR_GLOBAL_LOCKS] - f2fs_sb_info -> node_write 1. mutex_lock_op(sbi) - try to get an avaiable lock from the array. - returns the index of the gottern lock variable. 2. mutex_unlock_op(sbi, index of the lock) - unlock the given index of the lock. 3. mutex_lock_all(sbi) - grab all the locks in the array before the checkpoint. 4. mutex_unlock_all(sbi) - release all the locks in the array after checkpoint. 5. block_operations() - call mutex_lock_all() - sync_dirty_dir_inodes() - grab node_write - sync_node_pages() Note that, the pairs of mutex_lock_op()/mutex_unlock_op() and mutex_lock_all()/mutex_unlock_all() should be used together. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-22 07:21:29 +00:00
{
f2fs_up_write(&sbi->cp_rwsem);
}
static inline int __get_cp_reason(struct f2fs_sb_info *sbi)
{
int reason = CP_SYNC;
if (test_opt(sbi, FASTBOOT))
reason = CP_FASTBOOT;
if (is_sbi_flag_set(sbi, SBI_IS_CLOSE))
reason = CP_UMOUNT;
return reason;
}
static inline bool __remain_node_summaries(int reason)
{
return (reason & (CP_UMOUNT | CP_FASTBOOT));
}
static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi)
{
return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) ||
is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG));
}
/*
* Check whether the inode has blocks or not
*/
static inline int F2FS_HAS_BLOCKS(struct inode *inode)
{
block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0;
return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block;
}
static inline bool f2fs_has_xattr_block(unsigned int ofs)
{
return ofs == XATTR_NODE_OFFSET;
}
static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi,
struct inode *inode, bool cap)
{
if (!inode)
return true;
if (!test_opt(sbi, RESERVE_ROOT))
return false;
if (IS_NOQUOTA(inode))
return true;
if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid()))
return true;
if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) &&
in_group_p(F2FS_OPTION(sbi).s_resgid))
return true;
if (cap && capable(CAP_SYS_RESOURCE))
return true;
return false;
}
static inline unsigned int get_available_block_count(struct f2fs_sb_info *sbi,
struct inode *inode, bool cap)
{
block_t avail_user_block_count;
avail_user_block_count = sbi->user_block_count -
sbi->current_reserved_blocks;
if (!__allow_reserved_blocks(sbi, inode, cap))
avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks;
if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
if (avail_user_block_count > sbi->unusable_block_count)
avail_user_block_count -= sbi->unusable_block_count;
else
avail_user_block_count = 0;
}
return avail_user_block_count;
}
static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool);
static inline int inc_valid_block_count(struct f2fs_sb_info *sbi,
struct inode *inode, blkcnt_t *count, bool partial)
{
long long diff = 0, release = 0;
block_t avail_user_block_count;
int ret;
ret = dquot_reserve_block(inode, *count);
if (ret)
return ret;
if (time_to_inject(sbi, FAULT_BLOCK)) {
release = *count;
goto release_quota;
}
/*
* let's increase this in prior to actual block count change in order
* for f2fs_sync_file to avoid data races when deciding checkpoint.
*/
percpu_counter_add(&sbi->alloc_valid_block_count, (*count));
spin_lock(&sbi->stat_lock);
f2fs: fix to reserve space for IO align feature https://bugzilla.kernel.org/show_bug.cgi?id=204137 With below script, we will hit panic during new segment allocation: DISK=bingo.img MOUNT_DIR=/mnt/f2fs dd if=/dev/zero of=$DISK bs=1M count=105 mkfs.f2fe -a 1 -o 19 -t 1 -z 1 -f -q $DISK mount -t f2fs $DISK $MOUNT_DIR -o "noinline_dentry,flush_merge,noextent_cache,mode=lfs,io_bits=7,fsync_mode=strict" for (( i = 0; i < 4096; i++ )); do name=`head /dev/urandom | tr -dc A-Za-z0-9 | head -c 10` mkdir $MOUNT_DIR/$name done umount $MOUNT_DIR rm $DISK --- Core dump --- Call Trace: allocate_segment_by_default+0x9d/0x100 [f2fs] f2fs_allocate_data_block+0x3c0/0x5c0 [f2fs] do_write_page+0x62/0x110 [f2fs] f2fs_outplace_write_data+0x43/0xc0 [f2fs] f2fs_do_write_data_page+0x386/0x560 [f2fs] __write_data_page+0x706/0x850 [f2fs] f2fs_write_cache_pages+0x267/0x6a0 [f2fs] f2fs_write_data_pages+0x19c/0x2e0 [f2fs] do_writepages+0x1c/0x70 __filemap_fdatawrite_range+0xaa/0xe0 filemap_fdatawrite+0x1f/0x30 f2fs_sync_dirty_inodes+0x74/0x1f0 [f2fs] block_operations+0xdc/0x350 [f2fs] f2fs_write_checkpoint+0x104/0x1150 [f2fs] f2fs_sync_fs+0xa2/0x120 [f2fs] f2fs_balance_fs_bg+0x33c/0x390 [f2fs] f2fs_write_node_pages+0x4c/0x1f0 [f2fs] do_writepages+0x1c/0x70 __writeback_single_inode+0x45/0x320 writeback_sb_inodes+0x273/0x5c0 wb_writeback+0xff/0x2e0 wb_workfn+0xa1/0x370 process_one_work+0x138/0x350 worker_thread+0x4d/0x3d0 kthread+0x109/0x140 ret_from_fork+0x25/0x30 The root cause here is, with IO alignment feature enables, in worst case, we need F2FS_IO_SIZE() free blocks space for single one 4k write due to IO alignment feature will fill dummy pages to make IO being aligned. So we will easily run out of free segments during non-inline directory's data writeback, even in process of foreground GC. In order to fix this issue, I just propose to reserve additional free space for IO alignment feature to handle worst case of free space usage ratio during FGGC. Fixes: 0a595ebaaa6b ("f2fs: support IO alignment for DATA and NODE writes") Signed-off-by: Chao Yu <chao@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2021-12-11 13:27:36 +00:00
avail_user_block_count = get_available_block_count(sbi, inode, true);
diff = (long long)sbi->total_valid_block_count + *count -
avail_user_block_count;
if (unlikely(diff > 0)) {
if (!partial) {
spin_unlock(&sbi->stat_lock);
release = *count;
goto enospc;
}
if (diff > *count)
diff = *count;
*count -= diff;
release = diff;
if (!*count) {
spin_unlock(&sbi->stat_lock);
goto enospc;
}
}
sbi->total_valid_block_count += (block_t)(*count);
spin_unlock(&sbi->stat_lock);
if (unlikely(release)) {
percpu_counter_sub(&sbi->alloc_valid_block_count, release);
dquot_release_reservation_block(inode, release);
}
f2fs_i_blocks_write(inode, *count, true, true);
return 0;
enospc:
percpu_counter_sub(&sbi->alloc_valid_block_count, release);
release_quota:
dquot_release_reservation_block(inode, release);
return -ENOSPC;
}
#define PAGE_PRIVATE_GET_FUNC(name, flagname) \
static inline bool page_private_##name(struct page *page) \
{ \
return PagePrivate(page) && \
test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \
test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \
}
#define PAGE_PRIVATE_SET_FUNC(name, flagname) \
static inline void set_page_private_##name(struct page *page) \
{ \
if (!PagePrivate(page)) \
attach_page_private(page, (void *)0); \
set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \
set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \
}
#define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \
static inline void clear_page_private_##name(struct page *page) \
{ \
clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \
if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER)) \
detach_page_private(page); \
}
PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER);
PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE);
PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION);
f2fs: atomic: fix to avoid racing w/ GC Case #1: SQLite App GC Thread Kworker Shrinker - f2fs_ioc_start_atomic_write - f2fs_ioc_commit_atomic_write - f2fs_commit_atomic_write - filemap_write_and_wait_range : write atomic_file's data to cow_inode echo 3 > drop_caches to drop atomic_file's cache. - f2fs_gc - gc_data_segment - move_data_page - set_page_dirty - writepages - f2fs_do_write_data_page : overwrite atomic_file's data to cow_inode - f2fs_down_write(&fi->i_gc_rwsem[WRITE]) - __f2fs_commit_atomic_write - f2fs_up_write(&fi->i_gc_rwsem[WRITE]) Case #2: SQLite App GC Thread Kworker - f2fs_ioc_start_atomic_write - __writeback_single_inode - do_writepages - f2fs_write_cache_pages - f2fs_write_single_data_page - f2fs_do_write_data_page : write atomic_file's data to cow_inode - f2fs_gc - gc_data_segment - move_data_page - set_page_dirty - writepages - f2fs_do_write_data_page : overwrite atomic_file's data to cow_inode - f2fs_ioc_commit_atomic_write In above cases racing in between atomic_write and GC, previous data in atomic_file may be overwrited to cow_file, result in data corruption. This patch introduces PAGE_PRIVATE_ATOMIC_WRITE bit flag in page.private, and use it to indicate that there is last dirty data in atomic file, and the data should be writebacked into cow_file, if the flag is not tagged in page, we should never write data across files. Fixes: 3db1de0e582c ("f2fs: change the current atomic write way") Cc: Daeho Jeong <daehojeong@google.com> Signed-off-by: Chao Yu <chao@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2024-06-25 03:13:48 +00:00
PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE);
PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE);
PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE);
PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION);
f2fs: atomic: fix to avoid racing w/ GC Case #1: SQLite App GC Thread Kworker Shrinker - f2fs_ioc_start_atomic_write - f2fs_ioc_commit_atomic_write - f2fs_commit_atomic_write - filemap_write_and_wait_range : write atomic_file's data to cow_inode echo 3 > drop_caches to drop atomic_file's cache. - f2fs_gc - gc_data_segment - move_data_page - set_page_dirty - writepages - f2fs_do_write_data_page : overwrite atomic_file's data to cow_inode - f2fs_down_write(&fi->i_gc_rwsem[WRITE]) - __f2fs_commit_atomic_write - f2fs_up_write(&fi->i_gc_rwsem[WRITE]) Case #2: SQLite App GC Thread Kworker - f2fs_ioc_start_atomic_write - __writeback_single_inode - do_writepages - f2fs_write_cache_pages - f2fs_write_single_data_page - f2fs_do_write_data_page : write atomic_file's data to cow_inode - f2fs_gc - gc_data_segment - move_data_page - set_page_dirty - writepages - f2fs_do_write_data_page : overwrite atomic_file's data to cow_inode - f2fs_ioc_commit_atomic_write In above cases racing in between atomic_write and GC, previous data in atomic_file may be overwrited to cow_file, result in data corruption. This patch introduces PAGE_PRIVATE_ATOMIC_WRITE bit flag in page.private, and use it to indicate that there is last dirty data in atomic file, and the data should be writebacked into cow_file, if the flag is not tagged in page, we should never write data across files. Fixes: 3db1de0e582c ("f2fs: change the current atomic write way") Cc: Daeho Jeong <daehojeong@google.com> Signed-off-by: Chao Yu <chao@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2024-06-25 03:13:48 +00:00
PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE);
PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE);
PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE);
PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION);
f2fs: atomic: fix to avoid racing w/ GC Case #1: SQLite App GC Thread Kworker Shrinker - f2fs_ioc_start_atomic_write - f2fs_ioc_commit_atomic_write - f2fs_commit_atomic_write - filemap_write_and_wait_range : write atomic_file's data to cow_inode echo 3 > drop_caches to drop atomic_file's cache. - f2fs_gc - gc_data_segment - move_data_page - set_page_dirty - writepages - f2fs_do_write_data_page : overwrite atomic_file's data to cow_inode - f2fs_down_write(&fi->i_gc_rwsem[WRITE]) - __f2fs_commit_atomic_write - f2fs_up_write(&fi->i_gc_rwsem[WRITE]) Case #2: SQLite App GC Thread Kworker - f2fs_ioc_start_atomic_write - __writeback_single_inode - do_writepages - f2fs_write_cache_pages - f2fs_write_single_data_page - f2fs_do_write_data_page : write atomic_file's data to cow_inode - f2fs_gc - gc_data_segment - move_data_page - set_page_dirty - writepages - f2fs_do_write_data_page : overwrite atomic_file's data to cow_inode - f2fs_ioc_commit_atomic_write In above cases racing in between atomic_write and GC, previous data in atomic_file may be overwrited to cow_file, result in data corruption. This patch introduces PAGE_PRIVATE_ATOMIC_WRITE bit flag in page.private, and use it to indicate that there is last dirty data in atomic file, and the data should be writebacked into cow_file, if the flag is not tagged in page, we should never write data across files. Fixes: 3db1de0e582c ("f2fs: change the current atomic write way") Cc: Daeho Jeong <daehojeong@google.com> Signed-off-by: Chao Yu <chao@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2024-06-25 03:13:48 +00:00
PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE);
static inline unsigned long get_page_private_data(struct page *page)
{
unsigned long data = page_private(page);
if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data))
return 0;
return data >> PAGE_PRIVATE_MAX;
}
static inline void set_page_private_data(struct page *page, unsigned long data)
{
if (!PagePrivate(page))
attach_page_private(page, (void *)0);
set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page));
page_private(page) |= data << PAGE_PRIVATE_MAX;
}
static inline void clear_page_private_data(struct page *page)
{
page_private(page) &= GENMASK(PAGE_PRIVATE_MAX - 1, 0);
if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER))
detach_page_private(page);
}
static inline void clear_page_private_all(struct page *page)
{
clear_page_private_data(page);
clear_page_private_reference(page);
clear_page_private_gcing(page);
clear_page_private_inline(page);
f2fs: atomic: fix to avoid racing w/ GC Case #1: SQLite App GC Thread Kworker Shrinker - f2fs_ioc_start_atomic_write - f2fs_ioc_commit_atomic_write - f2fs_commit_atomic_write - filemap_write_and_wait_range : write atomic_file's data to cow_inode echo 3 > drop_caches to drop atomic_file's cache. - f2fs_gc - gc_data_segment - move_data_page - set_page_dirty - writepages - f2fs_do_write_data_page : overwrite atomic_file's data to cow_inode - f2fs_down_write(&fi->i_gc_rwsem[WRITE]) - __f2fs_commit_atomic_write - f2fs_up_write(&fi->i_gc_rwsem[WRITE]) Case #2: SQLite App GC Thread Kworker - f2fs_ioc_start_atomic_write - __writeback_single_inode - do_writepages - f2fs_write_cache_pages - f2fs_write_single_data_page - f2fs_do_write_data_page : write atomic_file's data to cow_inode - f2fs_gc - gc_data_segment - move_data_page - set_page_dirty - writepages - f2fs_do_write_data_page : overwrite atomic_file's data to cow_inode - f2fs_ioc_commit_atomic_write In above cases racing in between atomic_write and GC, previous data in atomic_file may be overwrited to cow_file, result in data corruption. This patch introduces PAGE_PRIVATE_ATOMIC_WRITE bit flag in page.private, and use it to indicate that there is last dirty data in atomic file, and the data should be writebacked into cow_file, if the flag is not tagged in page, we should never write data across files. Fixes: 3db1de0e582c ("f2fs: change the current atomic write way") Cc: Daeho Jeong <daehojeong@google.com> Signed-off-by: Chao Yu <chao@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2024-06-25 03:13:48 +00:00
clear_page_private_atomic(page);
f2fs_bug_on(F2FS_P_SB(page), page_private(page));
}
static inline void dec_valid_block_count(struct f2fs_sb_info *sbi,
struct inode *inode,
block_t count)
{
blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK;
spin_lock(&sbi->stat_lock);
f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count);
sbi->total_valid_block_count -= (block_t)count;
if (sbi->reserved_blocks &&
sbi->current_reserved_blocks < sbi->reserved_blocks)
sbi->current_reserved_blocks = min(sbi->reserved_blocks,
sbi->current_reserved_blocks + count);
spin_unlock(&sbi->stat_lock);
if (unlikely(inode->i_blocks < sectors)) {
f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu",
inode->i_ino,
(unsigned long long)inode->i_blocks,
(unsigned long long)sectors);
set_sbi_flag(sbi, SBI_NEED_FSCK);
return;
}
f2fs_i_blocks_write(inode, count, false, true);
}
static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type)
{
atomic_inc(&sbi->nr_pages[count_type]);
if (count_type == F2FS_DIRTY_DENTS ||
count_type == F2FS_DIRTY_NODES ||
count_type == F2FS_DIRTY_META ||
count_type == F2FS_DIRTY_QDATA ||
count_type == F2FS_DIRTY_IMETA)
set_sbi_flag(sbi, SBI_IS_DIRTY);
}
static inline void inode_inc_dirty_pages(struct inode *inode)
{
atomic_inc(&F2FS_I(inode)->dirty_pages);
inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
if (IS_NOQUOTA(inode))
inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
}
static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type)
{
atomic_dec(&sbi->nr_pages[count_type]);
}
static inline void inode_dec_dirty_pages(struct inode *inode)
{
if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) &&
!S_ISLNK(inode->i_mode))
return;
atomic_dec(&F2FS_I(inode)->dirty_pages);
dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
if (IS_NOQUOTA(inode))
dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
}
static inline void inc_atomic_write_cnt(struct inode *inode)
{
struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
struct f2fs_inode_info *fi = F2FS_I(inode);
u64 current_write;
fi->atomic_write_cnt++;
atomic64_inc(&sbi->current_atomic_write);
current_write = atomic64_read(&sbi->current_atomic_write);
if (current_write > sbi->peak_atomic_write)
sbi->peak_atomic_write = current_write;
}
static inline void release_atomic_write_cnt(struct inode *inode)
{
struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
struct f2fs_inode_info *fi = F2FS_I(inode);
atomic64_sub(fi->atomic_write_cnt, &sbi->current_atomic_write);
fi->atomic_write_cnt = 0;
}
static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type)
{
return atomic_read(&sbi->nr_pages[count_type]);
}
static inline int get_dirty_pages(struct inode *inode)
{
return atomic_read(&F2FS_I(inode)->dirty_pages);
}
static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type)
{
return div_u64(get_pages(sbi, block_type) + BLKS_PER_SEC(sbi) - 1,
BLKS_PER_SEC(sbi));
}
static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi)
{
return sbi->total_valid_block_count;
}
static inline block_t discard_blocks(struct f2fs_sb_info *sbi)
{
return sbi->discard_blks;
}
static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag)
{
struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
/* return NAT or SIT bitmap */
if (flag == NAT_BITMAP)
return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
else if (flag == SIT_BITMAP)
return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
return 0;
}
static inline block_t __cp_payload(struct f2fs_sb_info *sbi)
{
return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload);
}
static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
{
struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
f2fs: Replace one-element array with flexible-array member There is a regular need in the kernel to provide a way to declare having a dynamically sized set of trailing elements in a structure. Kernel code should always use “flexible array members”[1] for these cases. The older style of one-element or zero-length arrays should no longer be used[2]. Refactor the code according to the use of a flexible-array member in struct f2fs_checkpoint, instead of a one-element arrays. Notice that a temporary pointer to void '*tmp_ptr' was used in order to fix the following errors when using a flexible array instead of a one element array in struct f2fs_checkpoint: CC [M] fs/f2fs/dir.o In file included from fs/f2fs/dir.c:13: fs/f2fs/f2fs.h: In function ‘__bitmap_ptr’: fs/f2fs/f2fs.h:2227:40: error: invalid use of flexible array member 2227 | return &ckpt->sit_nat_version_bitmap + offset + sizeof(__le32); | ^ fs/f2fs/f2fs.h:2227:49: error: invalid use of flexible array member 2227 | return &ckpt->sit_nat_version_bitmap + offset + sizeof(__le32); | ^ fs/f2fs/f2fs.h:2238:40: error: invalid use of flexible array member 2238 | return &ckpt->sit_nat_version_bitmap + offset; | ^ make[2]: *** [scripts/Makefile.build:287: fs/f2fs/dir.o] Error 1 make[1]: *** [scripts/Makefile.build:530: fs/f2fs] Error 2 make: *** [Makefile:1819: fs] Error 2 [1] https://en.wikipedia.org/wiki/Flexible_array_member [2] https://www.kernel.org/doc/html/v5.9/process/deprecated.html#zero-length-and-one-element-arrays Link: https://github.com/KSPP/linux/issues/79 Build-tested-by: kernel test robot <lkp@intel.com> Link: https://lore.kernel.org/lkml/603647e4.DeEFbl4eqljuwAUe%25lkp@intel.com/ Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org> Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2021-02-24 19:03:13 +00:00
void *tmp_ptr = &ckpt->sit_nat_version_bitmap;
int offset;
if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) {
offset = (flag == SIT_BITMAP) ?
le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0;
2019-04-22 09:33:53 +00:00
/*
* if large_nat_bitmap feature is enabled, leave checksum
* protection for all nat/sit bitmaps.
*/
f2fs: Replace one-element array with flexible-array member There is a regular need in the kernel to provide a way to declare having a dynamically sized set of trailing elements in a structure. Kernel code should always use “flexible array members”[1] for these cases. The older style of one-element or zero-length arrays should no longer be used[2]. Refactor the code according to the use of a flexible-array member in struct f2fs_checkpoint, instead of a one-element arrays. Notice that a temporary pointer to void '*tmp_ptr' was used in order to fix the following errors when using a flexible array instead of a one element array in struct f2fs_checkpoint: CC [M] fs/f2fs/dir.o In file included from fs/f2fs/dir.c:13: fs/f2fs/f2fs.h: In function ‘__bitmap_ptr’: fs/f2fs/f2fs.h:2227:40: error: invalid use of flexible array member 2227 | return &ckpt->sit_nat_version_bitmap + offset + sizeof(__le32); | ^ fs/f2fs/f2fs.h:2227:49: error: invalid use of flexible array member 2227 | return &ckpt->sit_nat_version_bitmap + offset + sizeof(__le32); | ^ fs/f2fs/f2fs.h:2238:40: error: invalid use of flexible array member 2238 | return &ckpt->sit_nat_version_bitmap + offset; | ^ make[2]: *** [scripts/Makefile.build:287: fs/f2fs/dir.o] Error 1 make[1]: *** [scripts/Makefile.build:530: fs/f2fs] Error 2 make: *** [Makefile:1819: fs] Error 2 [1] https://en.wikipedia.org/wiki/Flexible_array_member [2] https://www.kernel.org/doc/html/v5.9/process/deprecated.html#zero-length-and-one-element-arrays Link: https://github.com/KSPP/linux/issues/79 Build-tested-by: kernel test robot <lkp@intel.com> Link: https://lore.kernel.org/lkml/603647e4.DeEFbl4eqljuwAUe%25lkp@intel.com/ Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org> Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2021-02-24 19:03:13 +00:00
return tmp_ptr + offset + sizeof(__le32);
}
if (__cp_payload(sbi) > 0) {
if (flag == NAT_BITMAP)
return tmp_ptr;
else
return (unsigned char *)ckpt + F2FS_BLKSIZE;
} else {
offset = (flag == NAT_BITMAP) ?
le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0;
f2fs: Replace one-element array with flexible-array member There is a regular need in the kernel to provide a way to declare having a dynamically sized set of trailing elements in a structure. Kernel code should always use “flexible array members”[1] for these cases. The older style of one-element or zero-length arrays should no longer be used[2]. Refactor the code according to the use of a flexible-array member in struct f2fs_checkpoint, instead of a one-element arrays. Notice that a temporary pointer to void '*tmp_ptr' was used in order to fix the following errors when using a flexible array instead of a one element array in struct f2fs_checkpoint: CC [M] fs/f2fs/dir.o In file included from fs/f2fs/dir.c:13: fs/f2fs/f2fs.h: In function ‘__bitmap_ptr’: fs/f2fs/f2fs.h:2227:40: error: invalid use of flexible array member 2227 | return &ckpt->sit_nat_version_bitmap + offset + sizeof(__le32); | ^ fs/f2fs/f2fs.h:2227:49: error: invalid use of flexible array member 2227 | return &ckpt->sit_nat_version_bitmap + offset + sizeof(__le32); | ^ fs/f2fs/f2fs.h:2238:40: error: invalid use of flexible array member 2238 | return &ckpt->sit_nat_version_bitmap + offset; | ^ make[2]: *** [scripts/Makefile.build:287: fs/f2fs/dir.o] Error 1 make[1]: *** [scripts/Makefile.build:530: fs/f2fs] Error 2 make: *** [Makefile:1819: fs] Error 2 [1] https://en.wikipedia.org/wiki/Flexible_array_member [2] https://www.kernel.org/doc/html/v5.9/process/deprecated.html#zero-length-and-one-element-arrays Link: https://github.com/KSPP/linux/issues/79 Build-tested-by: kernel test robot <lkp@intel.com> Link: https://lore.kernel.org/lkml/603647e4.DeEFbl4eqljuwAUe%25lkp@intel.com/ Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org> Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2021-02-24 19:03:13 +00:00
return tmp_ptr + offset;
}
}
static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)
{
block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
if (sbi->cur_cp_pack == 2)
start_addr += BLKS_PER_SEG(sbi);
return start_addr;
}
static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi)
{
block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
if (sbi->cur_cp_pack == 1)
start_addr += BLKS_PER_SEG(sbi);
return start_addr;
}
static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi)
{
sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1;
}
static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi)
{
return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum);
}
extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync);
static inline int inc_valid_node_count(struct f2fs_sb_info *sbi,
struct inode *inode, bool is_inode)
{
block_t valid_block_count;
unsigned int valid_node_count;
unsigned int avail_user_block_count;
int err;
if (is_inode) {
if (inode) {
err = dquot_alloc_inode(inode);
if (err)
return err;
}
} else {
err = dquot_reserve_block(inode, 1);
if (err)
return err;
}
if (time_to_inject(sbi, FAULT_BLOCK))
goto enospc;
spin_lock(&sbi->stat_lock);
valid_block_count = sbi->total_valid_block_count + 1;
avail_user_block_count = get_available_block_count(sbi, inode, false);
f2fs: fix to reserve space for IO align feature https://bugzilla.kernel.org/show_bug.cgi?id=204137 With below script, we will hit panic during new segment allocation: DISK=bingo.img MOUNT_DIR=/mnt/f2fs dd if=/dev/zero of=$DISK bs=1M count=105 mkfs.f2fe -a 1 -o 19 -t 1 -z 1 -f -q $DISK mount -t f2fs $DISK $MOUNT_DIR -o "noinline_dentry,flush_merge,noextent_cache,mode=lfs,io_bits=7,fsync_mode=strict" for (( i = 0; i < 4096; i++ )); do name=`head /dev/urandom | tr -dc A-Za-z0-9 | head -c 10` mkdir $MOUNT_DIR/$name done umount $MOUNT_DIR rm $DISK --- Core dump --- Call Trace: allocate_segment_by_default+0x9d/0x100 [f2fs] f2fs_allocate_data_block+0x3c0/0x5c0 [f2fs] do_write_page+0x62/0x110 [f2fs] f2fs_outplace_write_data+0x43/0xc0 [f2fs] f2fs_do_write_data_page+0x386/0x560 [f2fs] __write_data_page+0x706/0x850 [f2fs] f2fs_write_cache_pages+0x267/0x6a0 [f2fs] f2fs_write_data_pages+0x19c/0x2e0 [f2fs] do_writepages+0x1c/0x70 __filemap_fdatawrite_range+0xaa/0xe0 filemap_fdatawrite+0x1f/0x30 f2fs_sync_dirty_inodes+0x74/0x1f0 [f2fs] block_operations+0xdc/0x350 [f2fs] f2fs_write_checkpoint+0x104/0x1150 [f2fs] f2fs_sync_fs+0xa2/0x120 [f2fs] f2fs_balance_fs_bg+0x33c/0x390 [f2fs] f2fs_write_node_pages+0x4c/0x1f0 [f2fs] do_writepages+0x1c/0x70 __writeback_single_inode+0x45/0x320 writeback_sb_inodes+0x273/0x5c0 wb_writeback+0xff/0x2e0 wb_workfn+0xa1/0x370 process_one_work+0x138/0x350 worker_thread+0x4d/0x3d0 kthread+0x109/0x140 ret_from_fork+0x25/0x30 The root cause here is, with IO alignment feature enables, in worst case, we need F2FS_IO_SIZE() free blocks space for single one 4k write due to IO alignment feature will fill dummy pages to make IO being aligned. So we will easily run out of free segments during non-inline directory's data writeback, even in process of foreground GC. In order to fix this issue, I just propose to reserve additional free space for IO alignment feature to handle worst case of free space usage ratio during FGGC. Fixes: 0a595ebaaa6b ("f2fs: support IO alignment for DATA and NODE writes") Signed-off-by: Chao Yu <chao@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2021-12-11 13:27:36 +00:00
if (unlikely(valid_block_count > avail_user_block_count)) {
spin_unlock(&sbi->stat_lock);
goto enospc;
}
valid_node_count = sbi->total_valid_node_count + 1;
if (unlikely(valid_node_count > sbi->total_node_count)) {
spin_unlock(&sbi->stat_lock);
goto enospc;
}
sbi->total_valid_node_count++;
sbi->total_valid_block_count++;
spin_unlock(&sbi->stat_lock);
if (inode) {
if (is_inode)
f2fs_mark_inode_dirty_sync(inode, true);
else
f2fs_i_blocks_write(inode, 1, true, true);
}
percpu_counter_inc(&sbi->alloc_valid_block_count);
return 0;
enospc:
if (is_inode) {
if (inode)
dquot_free_inode(inode);
} else {
dquot_release_reservation_block(inode, 1);
}
return -ENOSPC;
}
static inline void dec_valid_node_count(struct f2fs_sb_info *sbi,
struct inode *inode, bool is_inode)
{
spin_lock(&sbi->stat_lock);
if (unlikely(!sbi->total_valid_block_count ||
!sbi->total_valid_node_count)) {
f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u",
sbi->total_valid_block_count,
sbi->total_valid_node_count);
set_sbi_flag(sbi, SBI_NEED_FSCK);
} else {
sbi->total_valid_block_count--;
sbi->total_valid_node_count--;
}
if (sbi->reserved_blocks &&
sbi->current_reserved_blocks < sbi->reserved_blocks)
sbi->current_reserved_blocks++;
spin_unlock(&sbi->stat_lock);
if (is_inode) {
dquot_free_inode(inode);
} else {
if (unlikely(inode->i_blocks == 0)) {
f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu",
inode->i_ino,
(unsigned long long)inode->i_blocks);
set_sbi_flag(sbi, SBI_NEED_FSCK);
return;
}
f2fs_i_blocks_write(inode, 1, false, true);
}
}
static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi)
{
return sbi->total_valid_node_count;
}
static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi)
{
percpu_counter_inc(&sbi->total_valid_inode_count);
}
static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi)
{
percpu_counter_dec(&sbi->total_valid_inode_count);
}
static inline s64 valid_inode_count(struct f2fs_sb_info *sbi)
{
return percpu_counter_sum_positive(&sbi->total_valid_inode_count);
}
static inline struct page *f2fs_grab_cache_page(struct address_space *mapping,
pgoff_t index, bool for_write)
{
struct page *page;
unsigned int flags;
if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) {
if (!for_write)
page = find_get_page_flags(mapping, index,
FGP_LOCK | FGP_ACCESSED);
else
page = find_lock_page(mapping, index);
if (page)
return page;
if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC))
return NULL;
}
if (!for_write)
return grab_cache_page(mapping, index);
flags = memalloc_nofs_save();
page = grab_cache_page_write_begin(mapping, index);
memalloc_nofs_restore(flags);
return page;
}
static inline struct page *f2fs_pagecache_get_page(
struct address_space *mapping, pgoff_t index,
fgf_t fgp_flags, gfp_t gfp_mask)
{
if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET))
return NULL;
return pagecache_get_page(mapping, index, fgp_flags, gfp_mask);
}
static inline void f2fs_put_page(struct page *page, int unlock)
{
if (!page)
return;
if (unlock) {
f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page));
unlock_page(page);
}
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
put_page(page);
}
static inline void f2fs_put_dnode(struct dnode_of_data *dn)
{
if (dn->node_page)
f2fs_put_page(dn->node_page, 1);
if (dn->inode_page && dn->node_page != dn->inode_page)
f2fs_put_page(dn->inode_page, 0);
dn->node_page = NULL;
dn->inode_page = NULL;
}
static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name,
size_t size)
{
return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL);
}
static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep,
gfp_t flags)
{
void *entry;
entry = kmem_cache_alloc(cachep, flags);
if (!entry)
entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL);
return entry;
}
static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep,
gfp_t flags, bool nofail, struct f2fs_sb_info *sbi)
{
if (nofail)
return f2fs_kmem_cache_alloc_nofail(cachep, flags);
if (time_to_inject(sbi, FAULT_SLAB_ALLOC))
return NULL;
return kmem_cache_alloc(cachep, flags);
}
static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type)
{
if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) ||
get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) ||
get_pages(sbi, F2FS_WB_CP_DATA) ||
get_pages(sbi, F2FS_DIO_READ) ||
get_pages(sbi, F2FS_DIO_WRITE))
return true;
if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info &&
atomic_read(&SM_I(sbi)->dcc_info->queued_discard))
return true;
if (SM_I(sbi) && SM_I(sbi)->fcc_info &&
atomic_read(&SM_I(sbi)->fcc_info->queued_flush))
return true;
return false;
}
static inline bool is_idle(struct f2fs_sb_info *sbi, int type)
{
if (sbi->gc_mode == GC_URGENT_HIGH)
return true;
if (is_inflight_io(sbi, type))
return false;
if (sbi->gc_mode == GC_URGENT_MID)
return true;
if (sbi->gc_mode == GC_URGENT_LOW &&
(type == DISCARD_TIME || type == GC_TIME))
return true;
return f2fs_time_over(sbi, type);
}
static inline void f2fs_radix_tree_insert(struct radix_tree_root *root,
unsigned long index, void *item)
{
while (radix_tree_insert(root, index, item))
cond_resched();
}
#define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
static inline bool IS_INODE(struct page *page)
{
struct f2fs_node *p = F2FS_NODE(page);
return RAW_IS_INODE(p);
}
f2fs: enhance on-disk inode structure scalability This patch add new flag F2FS_EXTRA_ATTR storing in inode.i_inline to indicate that on-disk structure of current inode is extended. In order to extend, we changed the inode structure a bit: Original one: struct f2fs_inode { ... struct f2fs_extent i_ext; __le32 i_addr[DEF_ADDRS_PER_INODE]; __le32 i_nid[DEF_NIDS_PER_INODE]; } Extended one: struct f2fs_inode { ... struct f2fs_extent i_ext; union { struct { __le16 i_extra_isize; __le16 i_padding; __le32 i_extra_end[0]; }; __le32 i_addr[DEF_ADDRS_PER_INODE]; }; __le32 i_nid[DEF_NIDS_PER_INODE]; } Once F2FS_EXTRA_ATTR is set, we will steal four bytes in the head of i_addr field for storing i_extra_isize and i_padding. with i_extra_isize, we can calculate actual size of reserved space in i_addr, available attribute fields included in total extra attribute fields for current inode can be described as below: +--------------------+ | .i_mode | | ... | | .i_ext | +--------------------+ | .i_extra_isize |-----+ | .i_padding | | | .i_prjid | | | .i_atime_extra | | | .i_ctime_extra | | | .i_mtime_extra |<----+ | .i_inode_cs |<----- store blkaddr/inline from here | .i_xattr_cs | | ... | +--------------------+ | | | block address | | | +--------------------+ | .i_nid | +--------------------+ | node_footer | | (nid, ino, offset) | +--------------------+ Hence, with this patch, we would enhance scalability of f2fs inode for storing more newly added attribute. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-07-18 16:19:06 +00:00
static inline int offset_in_addr(struct f2fs_inode *i)
{
return (i->i_inline & F2FS_EXTRA_ATTR) ?
(le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0;
}
static inline __le32 *blkaddr_in_node(struct f2fs_node *node)
{
return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr;
}
f2fs: enhance on-disk inode structure scalability This patch add new flag F2FS_EXTRA_ATTR storing in inode.i_inline to indicate that on-disk structure of current inode is extended. In order to extend, we changed the inode structure a bit: Original one: struct f2fs_inode { ... struct f2fs_extent i_ext; __le32 i_addr[DEF_ADDRS_PER_INODE]; __le32 i_nid[DEF_NIDS_PER_INODE]; } Extended one: struct f2fs_inode { ... struct f2fs_extent i_ext; union { struct { __le16 i_extra_isize; __le16 i_padding; __le32 i_extra_end[0]; }; __le32 i_addr[DEF_ADDRS_PER_INODE]; }; __le32 i_nid[DEF_NIDS_PER_INODE]; } Once F2FS_EXTRA_ATTR is set, we will steal four bytes in the head of i_addr field for storing i_extra_isize and i_padding. with i_extra_isize, we can calculate actual size of reserved space in i_addr, available attribute fields included in total extra attribute fields for current inode can be described as below: +--------------------+ | .i_mode | | ... | | .i_ext | +--------------------+ | .i_extra_isize |-----+ | .i_padding | | | .i_prjid | | | .i_atime_extra | | | .i_ctime_extra | | | .i_mtime_extra |<----+ | .i_inode_cs |<----- store blkaddr/inline from here | .i_xattr_cs | | ... | +--------------------+ | | | block address | | | +--------------------+ | .i_nid | +--------------------+ | node_footer | | (nid, ino, offset) | +--------------------+ Hence, with this patch, we would enhance scalability of f2fs inode for storing more newly added attribute. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-07-18 16:19:06 +00:00
static inline int f2fs_has_extra_attr(struct inode *inode);
static inline unsigned int get_dnode_base(struct inode *inode,
struct page *node_page)
{
if (!IS_INODE(node_page))
return 0;
return inode ? get_extra_isize(inode) :
offset_in_addr(&F2FS_NODE(node_page)->i);
}
f2fs: enhance on-disk inode structure scalability This patch add new flag F2FS_EXTRA_ATTR storing in inode.i_inline to indicate that on-disk structure of current inode is extended. In order to extend, we changed the inode structure a bit: Original one: struct f2fs_inode { ... struct f2fs_extent i_ext; __le32 i_addr[DEF_ADDRS_PER_INODE]; __le32 i_nid[DEF_NIDS_PER_INODE]; } Extended one: struct f2fs_inode { ... struct f2fs_extent i_ext; union { struct { __le16 i_extra_isize; __le16 i_padding; __le32 i_extra_end[0]; }; __le32 i_addr[DEF_ADDRS_PER_INODE]; }; __le32 i_nid[DEF_NIDS_PER_INODE]; } Once F2FS_EXTRA_ATTR is set, we will steal four bytes in the head of i_addr field for storing i_extra_isize and i_padding. with i_extra_isize, we can calculate actual size of reserved space in i_addr, available attribute fields included in total extra attribute fields for current inode can be described as below: +--------------------+ | .i_mode | | ... | | .i_ext | +--------------------+ | .i_extra_isize |-----+ | .i_padding | | | .i_prjid | | | .i_atime_extra | | | .i_ctime_extra | | | .i_mtime_extra |<----+ | .i_inode_cs |<----- store blkaddr/inline from here | .i_xattr_cs | | ... | +--------------------+ | | | block address | | | +--------------------+ | .i_nid | +--------------------+ | node_footer | | (nid, ino, offset) | +--------------------+ Hence, with this patch, we would enhance scalability of f2fs inode for storing more newly added attribute. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-07-18 16:19:06 +00:00
static inline __le32 *get_dnode_addr(struct inode *inode,
struct page *node_page)
{
return blkaddr_in_node(F2FS_NODE(node_page)) +
get_dnode_base(inode, node_page);
}
f2fs: enhance on-disk inode structure scalability This patch add new flag F2FS_EXTRA_ATTR storing in inode.i_inline to indicate that on-disk structure of current inode is extended. In order to extend, we changed the inode structure a bit: Original one: struct f2fs_inode { ... struct f2fs_extent i_ext; __le32 i_addr[DEF_ADDRS_PER_INODE]; __le32 i_nid[DEF_NIDS_PER_INODE]; } Extended one: struct f2fs_inode { ... struct f2fs_extent i_ext; union { struct { __le16 i_extra_isize; __le16 i_padding; __le32 i_extra_end[0]; }; __le32 i_addr[DEF_ADDRS_PER_INODE]; }; __le32 i_nid[DEF_NIDS_PER_INODE]; } Once F2FS_EXTRA_ATTR is set, we will steal four bytes in the head of i_addr field for storing i_extra_isize and i_padding. with i_extra_isize, we can calculate actual size of reserved space in i_addr, available attribute fields included in total extra attribute fields for current inode can be described as below: +--------------------+ | .i_mode | | ... | | .i_ext | +--------------------+ | .i_extra_isize |-----+ | .i_padding | | | .i_prjid | | | .i_atime_extra | | | .i_ctime_extra | | | .i_mtime_extra |<----+ | .i_inode_cs |<----- store blkaddr/inline from here | .i_xattr_cs | | ... | +--------------------+ | | | block address | | | +--------------------+ | .i_nid | +--------------------+ | node_footer | | (nid, ino, offset) | +--------------------+ Hence, with this patch, we would enhance scalability of f2fs inode for storing more newly added attribute. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-07-18 16:19:06 +00:00
static inline block_t data_blkaddr(struct inode *inode,
struct page *node_page, unsigned int offset)
{
return le32_to_cpu(*(get_dnode_addr(inode, node_page) + offset));
}
static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn)
{
return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node);
}
static inline int f2fs_test_bit(unsigned int nr, char *addr)
{
int mask;
addr += (nr >> 3);
mask = BIT(7 - (nr & 0x07));
return mask & *addr;
}
static inline void f2fs_set_bit(unsigned int nr, char *addr)
{
int mask;
addr += (nr >> 3);
mask = BIT(7 - (nr & 0x07));
*addr |= mask;
}
static inline void f2fs_clear_bit(unsigned int nr, char *addr)
{
int mask;
addr += (nr >> 3);
mask = BIT(7 - (nr & 0x07));
*addr &= ~mask;
}
static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr)
{
int mask;
int ret;
addr += (nr >> 3);
mask = BIT(7 - (nr & 0x07));
ret = mask & *addr;
*addr |= mask;
return ret;
}
static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr)
{
int mask;
int ret;
addr += (nr >> 3);
mask = BIT(7 - (nr & 0x07));
ret = mask & *addr;
*addr &= ~mask;
return ret;
}
static inline void f2fs_change_bit(unsigned int nr, char *addr)
{
int mask;
addr += (nr >> 3);
mask = BIT(7 - (nr & 0x07));
*addr ^= mask;
}
/*
* On-disk inode flags (f2fs_inode::i_flags)
*/
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>
2019-11-01 10:07:14 +00:00
#define F2FS_COMPR_FL 0x00000004 /* Compress file */
#define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */
#define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
#define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */
#define F2FS_NODUMP_FL 0x00000040 /* do not dump file */
#define F2FS_NOATIME_FL 0x00000080 /* do not update atime */
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>
2019-11-01 10:07:14 +00:00
#define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */
#define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */
#define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
#define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
f2fs: Support case-insensitive file name lookups Modeled after commit b886ee3e778e ("ext4: Support case-insensitive file name lookups") """ This patch implements the actual support for case-insensitive file name lookups in f2fs, based on the feature bit and the encoding stored in the superblock. A filesystem that has the casefold feature set is able to configure directories with the +F (F2FS_CASEFOLD_FL) attribute, enabling lookups to succeed in that directory in a case-insensitive fashion, i.e: match a directory entry even if the name used by userspace is not a byte per byte match with the disk name, but is an equivalent case-insensitive version of the Unicode string. This operation is called a case-insensitive file name lookup. The feature is configured as an inode attribute applied to directories and inherited by its children. This attribute can only be enabled on empty directories for filesystems that support the encoding feature, thus preventing collision of file names that only differ by case. * dcache handling: For a +F directory, F2Fs only stores the first equivalent name dentry used in the dcache. This is done to prevent unintentional duplication of dentries in the dcache, while also allowing the VFS code to quickly find the right entry in the cache despite which equivalent string was used in a previous lookup, without having to resort to ->lookup(). d_hash() of casefolded directories is implemented as the hash of the casefolded string, such that we always have a well-known bucket for all the equivalencies of the same string. d_compare() uses the utf8_strncasecmp() infrastructure, which handles the comparison of equivalent, same case, names as well. For now, negative lookups are not inserted in the dcache, since they would need to be invalidated anyway, because we can't trust missing file dentries. This is bad for performance but requires some leveraging of the vfs layer to fix. We can live without that for now, and so does everyone else. * on-disk data: Despite using a specific version of the name as the internal representation within the dcache, the name stored and fetched from the disk is a byte-per-byte match with what the user requested, making this implementation 'name-preserving'. i.e. no actual information is lost when writing to storage. DX is supported by modifying the hashes used in +F directories to make them case/encoding-aware. The new disk hashes are calculated as the hash of the full casefolded string, instead of the string directly. This allows us to efficiently search for file names in the htree without requiring the user to provide an exact name. * Dealing with invalid sequences: By default, when a invalid UTF-8 sequence is identified, ext4 will treat it as an opaque byte sequence, ignoring the encoding and reverting to the old behavior for that unique file. This means that case-insensitive file name lookup will not work only for that file. An optional bit can be set in the superblock telling the filesystem code and userspace tools to enforce the encoding. When that optional bit is set, any attempt to create a file name using an invalid UTF-8 sequence will fail and return an error to userspace. * Normalization algorithm: The UTF-8 algorithms used to compare strings in f2fs is implemented in fs/unicode, and is based on a previous version developed by SGI. It implements the Canonical decomposition (NFD) algorithm described by the Unicode specification 12.1, or higher, combined with the elimination of ignorable code points (NFDi) and full case-folding (CF) as documented in fs/unicode/utf8_norm.c. NFD seems to be the best normalization method for F2FS because: - It has a lower cost than NFC/NFKC (which requires decomposing to NFD as an intermediary step) - It doesn't eliminate important semantic meaning like compatibility decompositions. Although: - This implementation is not completely linguistic accurate, because different languages have conflicting rules, which would require the specialization of the filesystem to a given locale, which brings all sorts of problems for removable media and for users who use more than one language. """ Signed-off-by: Daniel Rosenberg <drosen@google.com> Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-07-23 23:05:29 +00:00
#define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */
#define F2FS_QUOTA_DEFAULT_FL (F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL)
/* Flags that should be inherited by new inodes from their parent. */
#define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \
f2fs: Support case-insensitive file name lookups Modeled after commit b886ee3e778e ("ext4: Support case-insensitive file name lookups") """ This patch implements the actual support for case-insensitive file name lookups in f2fs, based on the feature bit and the encoding stored in the superblock. A filesystem that has the casefold feature set is able to configure directories with the +F (F2FS_CASEFOLD_FL) attribute, enabling lookups to succeed in that directory in a case-insensitive fashion, i.e: match a directory entry even if the name used by userspace is not a byte per byte match with the disk name, but is an equivalent case-insensitive version of the Unicode string. This operation is called a case-insensitive file name lookup. The feature is configured as an inode attribute applied to directories and inherited by its children. This attribute can only be enabled on empty directories for filesystems that support the encoding feature, thus preventing collision of file names that only differ by case. * dcache handling: For a +F directory, F2Fs only stores the first equivalent name dentry used in the dcache. This is done to prevent unintentional duplication of dentries in the dcache, while also allowing the VFS code to quickly find the right entry in the cache despite which equivalent string was used in a previous lookup, without having to resort to ->lookup(). d_hash() of casefolded directories is implemented as the hash of the casefolded string, such that we always have a well-known bucket for all the equivalencies of the same string. d_compare() uses the utf8_strncasecmp() infrastructure, which handles the comparison of equivalent, same case, names as well. For now, negative lookups are not inserted in the dcache, since they would need to be invalidated anyway, because we can't trust missing file dentries. This is bad for performance but requires some leveraging of the vfs layer to fix. We can live without that for now, and so does everyone else. * on-disk data: Despite using a specific version of the name as the internal representation within the dcache, the name stored and fetched from the disk is a byte-per-byte match with what the user requested, making this implementation 'name-preserving'. i.e. no actual information is lost when writing to storage. DX is supported by modifying the hashes used in +F directories to make them case/encoding-aware. The new disk hashes are calculated as the hash of the full casefolded string, instead of the string directly. This allows us to efficiently search for file names in the htree without requiring the user to provide an exact name. * Dealing with invalid sequences: By default, when a invalid UTF-8 sequence is identified, ext4 will treat it as an opaque byte sequence, ignoring the encoding and reverting to the old behavior for that unique file. This means that case-insensitive file name lookup will not work only for that file. An optional bit can be set in the superblock telling the filesystem code and userspace tools to enforce the encoding. When that optional bit is set, any attempt to create a file name using an invalid UTF-8 sequence will fail and return an error to userspace. * Normalization algorithm: The UTF-8 algorithms used to compare strings in f2fs is implemented in fs/unicode, and is based on a previous version developed by SGI. It implements the Canonical decomposition (NFD) algorithm described by the Unicode specification 12.1, or higher, combined with the elimination of ignorable code points (NFDi) and full case-folding (CF) as documented in fs/unicode/utf8_norm.c. NFD seems to be the best normalization method for F2FS because: - It has a lower cost than NFC/NFKC (which requires decomposing to NFD as an intermediary step) - It doesn't eliminate important semantic meaning like compatibility decompositions. Although: - This implementation is not completely linguistic accurate, because different languages have conflicting rules, which would require the specialization of the filesystem to a given locale, which brings all sorts of problems for removable media and for users who use more than one language. """ Signed-off-by: Daniel Rosenberg <drosen@google.com> Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-07-23 23:05:29 +00:00
F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
F2FS_CASEFOLD_FL)
/* Flags that are appropriate for regular files (all but dir-specific ones). */
f2fs: Support case-insensitive file name lookups Modeled after commit b886ee3e778e ("ext4: Support case-insensitive file name lookups") """ This patch implements the actual support for case-insensitive file name lookups in f2fs, based on the feature bit and the encoding stored in the superblock. A filesystem that has the casefold feature set is able to configure directories with the +F (F2FS_CASEFOLD_FL) attribute, enabling lookups to succeed in that directory in a case-insensitive fashion, i.e: match a directory entry even if the name used by userspace is not a byte per byte match with the disk name, but is an equivalent case-insensitive version of the Unicode string. This operation is called a case-insensitive file name lookup. The feature is configured as an inode attribute applied to directories and inherited by its children. This attribute can only be enabled on empty directories for filesystems that support the encoding feature, thus preventing collision of file names that only differ by case. * dcache handling: For a +F directory, F2Fs only stores the first equivalent name dentry used in the dcache. This is done to prevent unintentional duplication of dentries in the dcache, while also allowing the VFS code to quickly find the right entry in the cache despite which equivalent string was used in a previous lookup, without having to resort to ->lookup(). d_hash() of casefolded directories is implemented as the hash of the casefolded string, such that we always have a well-known bucket for all the equivalencies of the same string. d_compare() uses the utf8_strncasecmp() infrastructure, which handles the comparison of equivalent, same case, names as well. For now, negative lookups are not inserted in the dcache, since they would need to be invalidated anyway, because we can't trust missing file dentries. This is bad for performance but requires some leveraging of the vfs layer to fix. We can live without that for now, and so does everyone else. * on-disk data: Despite using a specific version of the name as the internal representation within the dcache, the name stored and fetched from the disk is a byte-per-byte match with what the user requested, making this implementation 'name-preserving'. i.e. no actual information is lost when writing to storage. DX is supported by modifying the hashes used in +F directories to make them case/encoding-aware. The new disk hashes are calculated as the hash of the full casefolded string, instead of the string directly. This allows us to efficiently search for file names in the htree without requiring the user to provide an exact name. * Dealing with invalid sequences: By default, when a invalid UTF-8 sequence is identified, ext4 will treat it as an opaque byte sequence, ignoring the encoding and reverting to the old behavior for that unique file. This means that case-insensitive file name lookup will not work only for that file. An optional bit can be set in the superblock telling the filesystem code and userspace tools to enforce the encoding. When that optional bit is set, any attempt to create a file name using an invalid UTF-8 sequence will fail and return an error to userspace. * Normalization algorithm: The UTF-8 algorithms used to compare strings in f2fs is implemented in fs/unicode, and is based on a previous version developed by SGI. It implements the Canonical decomposition (NFD) algorithm described by the Unicode specification 12.1, or higher, combined with the elimination of ignorable code points (NFDi) and full case-folding (CF) as documented in fs/unicode/utf8_norm.c. NFD seems to be the best normalization method for F2FS because: - It has a lower cost than NFC/NFKC (which requires decomposing to NFD as an intermediary step) - It doesn't eliminate important semantic meaning like compatibility decompositions. Although: - This implementation is not completely linguistic accurate, because different languages have conflicting rules, which would require the specialization of the filesystem to a given locale, which brings all sorts of problems for removable media and for users who use more than one language. """ Signed-off-by: Daniel Rosenberg <drosen@google.com> Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-07-23 23:05:29 +00:00
#define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
F2FS_CASEFOLD_FL))
/* Flags that are appropriate for non-directories/regular files. */
#define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL)
static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags)
{
if (S_ISDIR(mode))
return flags;
else if (S_ISREG(mode))
return flags & F2FS_REG_FLMASK;
else
return flags & F2FS_OTHER_FLMASK;
}
static inline void __mark_inode_dirty_flag(struct inode *inode,
int flag, bool set)
{
switch (flag) {
case FI_INLINE_XATTR:
case FI_INLINE_DATA:
case FI_INLINE_DENTRY:
case FI_NEW_INODE:
if (set)
return;
fallthrough;
case FI_DATA_EXIST:
case FI_INLINE_DOTS:
case FI_PIN_FILE:
case FI_COMPRESS_RELEASED:
case FI_ATOMIC_COMMITTED:
f2fs_mark_inode_dirty_sync(inode, true);
}
}
static inline void set_inode_flag(struct inode *inode, int flag)
{
set_bit(flag, F2FS_I(inode)->flags);
__mark_inode_dirty_flag(inode, flag, true);
}
static inline int is_inode_flag_set(struct inode *inode, int flag)
{
return test_bit(flag, F2FS_I(inode)->flags);
}
static inline void clear_inode_flag(struct inode *inode, int flag)
{
clear_bit(flag, F2FS_I(inode)->flags);
__mark_inode_dirty_flag(inode, flag, false);
}
f2fs: add fs-verity support Add fs-verity support to f2fs. fs-verity is a filesystem feature that enables transparent integrity protection and authentication of read-only files. It uses a dm-verity like mechanism at the file level: a Merkle tree is used to verify any block in the file in log(filesize) time. It is implemented mainly by helper functions in fs/verity/. See Documentation/filesystems/fsverity.rst for the full documentation. The f2fs support for fs-verity consists of: - Adding a filesystem feature flag and an inode flag for fs-verity. - Implementing the fsverity_operations to support enabling verity on an inode and reading/writing the verity metadata. - Updating ->readpages() to verify data as it's read from verity files and to support reading verity metadata pages. - Updating ->write_begin(), ->write_end(), and ->writepages() to support writing verity metadata pages. - Calling the fs-verity hooks for ->open(), ->setattr(), and ->ioctl(). Like ext4, f2fs stores the verity metadata (Merkle tree and fsverity_descriptor) past the end of the file, starting at the first 64K boundary beyond i_size. This approach works because (a) verity files are readonly, and (b) pages fully beyond i_size aren't visible to userspace but can be read/written internally by f2fs with only some relatively small changes to f2fs. Extended attributes cannot be used because (a) f2fs limits the total size of an inode's xattr entries to 4096 bytes, which wouldn't be enough for even a single Merkle tree block, and (b) f2fs encryption doesn't encrypt xattrs, yet the verity metadata *must* be encrypted when the file is because it contains hashes of the plaintext data. Acked-by: Jaegeuk Kim <jaegeuk@kernel.org> Acked-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Eric Biggers <ebiggers@google.com>
2019-07-22 16:26:24 +00:00
static inline bool f2fs_verity_in_progress(struct inode *inode)
{
return IS_ENABLED(CONFIG_FS_VERITY) &&
is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS);
}
static inline void set_acl_inode(struct inode *inode, umode_t mode)
{
F2FS_I(inode)->i_acl_mode = mode;
set_inode_flag(inode, FI_ACL_MODE);
f2fs_mark_inode_dirty_sync(inode, false);
}
static inline void f2fs_i_links_write(struct inode *inode, bool inc)
{
if (inc)
inc_nlink(inode);
else
drop_nlink(inode);
f2fs_mark_inode_dirty_sync(inode, true);
}
static inline void f2fs_i_blocks_write(struct inode *inode,
block_t diff, bool add, bool claim)
{
bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
/* add = 1, claim = 1 should be dquot_reserve_block in pair */
if (add) {
if (claim)
dquot_claim_block(inode, diff);
else
dquot_alloc_block_nofail(inode, diff);
} else {
dquot_free_block(inode, diff);
}
f2fs_mark_inode_dirty_sync(inode, true);
if (clean || recover)
set_inode_flag(inode, FI_AUTO_RECOVER);
}
static inline bool f2fs_is_atomic_file(struct inode *inode);
static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size)
{
bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
if (i_size_read(inode) == i_size)
return;
i_size_write(inode, i_size);
if (f2fs_is_atomic_file(inode))
return;
f2fs_mark_inode_dirty_sync(inode, true);
if (clean || recover)
set_inode_flag(inode, FI_AUTO_RECOVER);
}
static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth)
{
F2FS_I(inode)->i_current_depth = depth;
f2fs_mark_inode_dirty_sync(inode, true);
}
static inline void f2fs_i_gc_failures_write(struct inode *inode,
unsigned int count)
{
F2FS_I(inode)->i_gc_failures = count;
f2fs_mark_inode_dirty_sync(inode, true);
}
static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid)
{
F2FS_I(inode)->i_xattr_nid = xnid;
f2fs_mark_inode_dirty_sync(inode, true);
}
static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino)
{
F2FS_I(inode)->i_pino = pino;
f2fs_mark_inode_dirty_sync(inode, true);
}
static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri)
{
struct f2fs_inode_info *fi = F2FS_I(inode);
if (ri->i_inline & F2FS_INLINE_XATTR)
set_bit(FI_INLINE_XATTR, fi->flags);
if (ri->i_inline & F2FS_INLINE_DATA)
set_bit(FI_INLINE_DATA, fi->flags);
if (ri->i_inline & F2FS_INLINE_DENTRY)
set_bit(FI_INLINE_DENTRY, fi->flags);
if (ri->i_inline & F2FS_DATA_EXIST)
set_bit(FI_DATA_EXIST, fi->flags);
if (ri->i_inline & F2FS_INLINE_DOTS)
set_bit(FI_INLINE_DOTS, fi->flags);
f2fs: enhance on-disk inode structure scalability This patch add new flag F2FS_EXTRA_ATTR storing in inode.i_inline to indicate that on-disk structure of current inode is extended. In order to extend, we changed the inode structure a bit: Original one: struct f2fs_inode { ... struct f2fs_extent i_ext; __le32 i_addr[DEF_ADDRS_PER_INODE]; __le32 i_nid[DEF_NIDS_PER_INODE]; } Extended one: struct f2fs_inode { ... struct f2fs_extent i_ext; union { struct { __le16 i_extra_isize; __le16 i_padding; __le32 i_extra_end[0]; }; __le32 i_addr[DEF_ADDRS_PER_INODE]; }; __le32 i_nid[DEF_NIDS_PER_INODE]; } Once F2FS_EXTRA_ATTR is set, we will steal four bytes in the head of i_addr field for storing i_extra_isize and i_padding. with i_extra_isize, we can calculate actual size of reserved space in i_addr, available attribute fields included in total extra attribute fields for current inode can be described as below: +--------------------+ | .i_mode | | ... | | .i_ext | +--------------------+ | .i_extra_isize |-----+ | .i_padding | | | .i_prjid | | | .i_atime_extra | | | .i_ctime_extra | | | .i_mtime_extra |<----+ | .i_inode_cs |<----- store blkaddr/inline from here | .i_xattr_cs | | ... | +--------------------+ | | | block address | | | +--------------------+ | .i_nid | +--------------------+ | node_footer | | (nid, ino, offset) | +--------------------+ Hence, with this patch, we would enhance scalability of f2fs inode for storing more newly added attribute. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-07-18 16:19:06 +00:00
if (ri->i_inline & F2FS_EXTRA_ATTR)
set_bit(FI_EXTRA_ATTR, fi->flags);
if (ri->i_inline & F2FS_PIN_FILE)
set_bit(FI_PIN_FILE, fi->flags);
if (ri->i_inline & F2FS_COMPRESS_RELEASED)
set_bit(FI_COMPRESS_RELEASED, fi->flags);
}
static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri)
{
ri->i_inline = 0;
if (is_inode_flag_set(inode, FI_INLINE_XATTR))
ri->i_inline |= F2FS_INLINE_XATTR;
if (is_inode_flag_set(inode, FI_INLINE_DATA))
ri->i_inline |= F2FS_INLINE_DATA;
if (is_inode_flag_set(inode, FI_INLINE_DENTRY))
ri->i_inline |= F2FS_INLINE_DENTRY;
if (is_inode_flag_set(inode, FI_DATA_EXIST))
ri->i_inline |= F2FS_DATA_EXIST;
if (is_inode_flag_set(inode, FI_INLINE_DOTS))
ri->i_inline |= F2FS_INLINE_DOTS;
f2fs: enhance on-disk inode structure scalability This patch add new flag F2FS_EXTRA_ATTR storing in inode.i_inline to indicate that on-disk structure of current inode is extended. In order to extend, we changed the inode structure a bit: Original one: struct f2fs_inode { ... struct f2fs_extent i_ext; __le32 i_addr[DEF_ADDRS_PER_INODE]; __le32 i_nid[DEF_NIDS_PER_INODE]; } Extended one: struct f2fs_inode { ... struct f2fs_extent i_ext; union { struct { __le16 i_extra_isize; __le16 i_padding; __le32 i_extra_end[0]; }; __le32 i_addr[DEF_ADDRS_PER_INODE]; }; __le32 i_nid[DEF_NIDS_PER_INODE]; } Once F2FS_EXTRA_ATTR is set, we will steal four bytes in the head of i_addr field for storing i_extra_isize and i_padding. with i_extra_isize, we can calculate actual size of reserved space in i_addr, available attribute fields included in total extra attribute fields for current inode can be described as below: +--------------------+ | .i_mode | | ... | | .i_ext | +--------------------+ | .i_extra_isize |-----+ | .i_padding | | | .i_prjid | | | .i_atime_extra | | | .i_ctime_extra | | | .i_mtime_extra |<----+ | .i_inode_cs |<----- store blkaddr/inline from here | .i_xattr_cs | | ... | +--------------------+ | | | block address | | | +--------------------+ | .i_nid | +--------------------+ | node_footer | | (nid, ino, offset) | +--------------------+ Hence, with this patch, we would enhance scalability of f2fs inode for storing more newly added attribute. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-07-18 16:19:06 +00:00
if (is_inode_flag_set(inode, FI_EXTRA_ATTR))
ri->i_inline |= F2FS_EXTRA_ATTR;
if (is_inode_flag_set(inode, FI_PIN_FILE))
ri->i_inline |= F2FS_PIN_FILE;
if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED))
ri->i_inline |= F2FS_COMPRESS_RELEASED;
f2fs: enhance on-disk inode structure scalability This patch add new flag F2FS_EXTRA_ATTR storing in inode.i_inline to indicate that on-disk structure of current inode is extended. In order to extend, we changed the inode structure a bit: Original one: struct f2fs_inode { ... struct f2fs_extent i_ext; __le32 i_addr[DEF_ADDRS_PER_INODE]; __le32 i_nid[DEF_NIDS_PER_INODE]; } Extended one: struct f2fs_inode { ... struct f2fs_extent i_ext; union { struct { __le16 i_extra_isize; __le16 i_padding; __le32 i_extra_end[0]; }; __le32 i_addr[DEF_ADDRS_PER_INODE]; }; __le32 i_nid[DEF_NIDS_PER_INODE]; } Once F2FS_EXTRA_ATTR is set, we will steal four bytes in the head of i_addr field for storing i_extra_isize and i_padding. with i_extra_isize, we can calculate actual size of reserved space in i_addr, available attribute fields included in total extra attribute fields for current inode can be described as below: +--------------------+ | .i_mode | | ... | | .i_ext | +--------------------+ | .i_extra_isize |-----+ | .i_padding | | | .i_prjid | | | .i_atime_extra | | | .i_ctime_extra | | | .i_mtime_extra |<----+ | .i_inode_cs |<----- store blkaddr/inline from here | .i_xattr_cs | | ... | +--------------------+ | | | block address | | | +--------------------+ | .i_nid | +--------------------+ | node_footer | | (nid, ino, offset) | +--------------------+ Hence, with this patch, we would enhance scalability of f2fs inode for storing more newly added attribute. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-07-18 16:19:06 +00:00
}
static inline int f2fs_has_extra_attr(struct inode *inode)
{
return is_inode_flag_set(inode, FI_EXTRA_ATTR);
}
static inline int f2fs_has_inline_xattr(struct inode *inode)
{
return is_inode_flag_set(inode, FI_INLINE_XATTR);
}
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>
2019-11-01 10:07:14 +00:00
static inline int f2fs_compressed_file(struct inode *inode)
{
return S_ISREG(inode->i_mode) &&
is_inode_flag_set(inode, FI_COMPRESSED_FILE);
}
static inline bool f2fs_need_compress_data(struct inode *inode)
{
int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode;
if (!f2fs_compressed_file(inode))
return false;
if (compress_mode == COMPR_MODE_FS)
return true;
else if (compress_mode == COMPR_MODE_USER &&
is_inode_flag_set(inode, FI_ENABLE_COMPRESS))
return true;
return false;
}
static inline unsigned int addrs_per_page(struct inode *inode,
bool is_inode)
{
unsigned int addrs = is_inode ? (CUR_ADDRS_PER_INODE(inode) -
get_inline_xattr_addrs(inode)) : DEF_ADDRS_PER_BLOCK;
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>
2019-11-01 10:07:14 +00:00
if (f2fs_compressed_file(inode))
return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size);
return addrs;
}
f2fs: support flexible inline xattr size Now, in product, more and more features based on file encryption were introduced, their demand of xattr space is increasing, however, inline xattr has fixed-size of 200 bytes, once inline xattr space is full, new increased xattr data would occupy additional xattr block which may bring us more space usage and performance regression during persisting. In order to resolve above issue, it's better to expand inline xattr size flexibly according to user's requirement. So this patch introduces new filesystem feature 'flexible inline xattr', and new mount option 'inline_xattr_size=%u', once mkfs enables the feature, we can use the option to make f2fs supporting flexible inline xattr size. To support this feature, we add extra attribute i_inline_xattr_size in inode layout, indicating that how many space inline xattr borrows from block address mapping space in inode layout, by this, we can easily locate and store flexible-sized inline xattr data in inode. Inode disk layout: +----------------------+ | .i_mode | | ... | | .i_ext | +----------------------+ | .i_extra_isize | | .i_inline_xattr_size |-----------+ | ... | | +----------------------+ | | .i_addr | | | - block address or | | | - inline data | | +----------------------+<---+ v | inline xattr | +---inline xattr range +----------------------+<---+ | .i_nid | +----------------------+ | node_footer | | (nid, ino, offset) | +----------------------+ Note that, we have to cnosider backward compatibility which reserved inline_data space, 200 bytes, all the time, reported by Sheng Yong. Previous inline data or directory always reserved 200 bytes in inode layout, even if inline_xattr is disabled. In order to keep inline_dentry's structure for backward compatibility, we get the space back only from inline_data. Signed-off-by: Chao Yu <yuchao0@huawei.com> Reported-by: Sheng Yong <shengyong1@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-09-06 13:59:50 +00:00
static inline void *inline_xattr_addr(struct inode *inode, struct page *page)
{
struct f2fs_inode *ri = F2FS_INODE(page);
return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE -
get_inline_xattr_addrs(inode)]);
}
static inline int inline_xattr_size(struct inode *inode)
{
if (f2fs_has_inline_xattr(inode))
return get_inline_xattr_addrs(inode) * sizeof(__le32);
return 0;
}
/*
* Notice: check inline_data flag without inode page lock is unsafe.
* It could change at any time by f2fs_convert_inline_page().
*/
static inline int f2fs_has_inline_data(struct inode *inode)
{
return is_inode_flag_set(inode, FI_INLINE_DATA);
}
static inline int f2fs_exist_data(struct inode *inode)
{
return is_inode_flag_set(inode, FI_DATA_EXIST);
}
static inline int f2fs_has_inline_dots(struct inode *inode)
{
return is_inode_flag_set(inode, FI_INLINE_DOTS);
}
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>
2019-11-01 10:07:14 +00:00
static inline int f2fs_is_mmap_file(struct inode *inode)
{
return is_inode_flag_set(inode, FI_MMAP_FILE);
}
static inline bool f2fs_is_pinned_file(struct inode *inode)
{
return is_inode_flag_set(inode, FI_PIN_FILE);
}
static inline bool f2fs_is_atomic_file(struct inode *inode)
{
return is_inode_flag_set(inode, FI_ATOMIC_FILE);
}
f2fs: fix null-ptr-deref in f2fs_get_dnode_of_data There is issue as follows when test f2fs atomic write: F2FS-fs (loop0): Can't find valid F2FS filesystem in 2th superblock F2FS-fs (loop0): invalid crc_offset: 0 F2FS-fs (loop0): f2fs_check_nid_range: out-of-range nid=1, run fsck to fix. F2FS-fs (loop0): f2fs_check_nid_range: out-of-range nid=2, run fsck to fix. ================================================================== BUG: KASAN: null-ptr-deref in f2fs_get_dnode_of_data+0xac/0x16d0 Read of size 8 at addr 0000000000000028 by task rep/1990 CPU: 4 PID: 1990 Comm: rep Not tainted 5.19.0-rc6-next-20220715 #266 Call Trace: <TASK> dump_stack_lvl+0x6e/0x91 print_report.cold+0x49a/0x6bb kasan_report+0xa8/0x130 f2fs_get_dnode_of_data+0xac/0x16d0 f2fs_do_write_data_page+0x2a5/0x1030 move_data_page+0x3c5/0xdf0 do_garbage_collect+0x2015/0x36c0 f2fs_gc+0x554/0x1d30 f2fs_balance_fs+0x7f5/0xda0 f2fs_write_single_data_page+0xb66/0xdc0 f2fs_write_cache_pages+0x716/0x1420 f2fs_write_data_pages+0x84f/0x9a0 do_writepages+0x130/0x3a0 filemap_fdatawrite_wbc+0x87/0xa0 file_write_and_wait_range+0x157/0x1c0 f2fs_do_sync_file+0x206/0x12d0 f2fs_sync_file+0x99/0xc0 vfs_fsync_range+0x75/0x140 f2fs_file_write_iter+0xd7b/0x1850 vfs_write+0x645/0x780 ksys_write+0xf1/0x1e0 do_syscall_64+0x3b/0x90 entry_SYSCALL_64_after_hwframe+0x63/0xcd As 3db1de0e582c commit changed atomic write way which new a cow_inode for atomic write file, and also mark cow_inode as FI_ATOMIC_FILE. When f2fs_do_write_data_page write cow_inode will use cow_inode's cow_inode which is NULL. Then will trigger null-ptr-deref. To solve above issue, introduce FI_COW_FILE flag for COW inode. Fiexes: 3db1de0e582c("f2fs: change the current atomic write way") Signed-off-by: Ye Bin <yebin10@huawei.com> Reviewed-by: Chao Yu <chao@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2022-08-01 11:26:04 +00:00
static inline bool f2fs_is_cow_file(struct inode *inode)
{
return is_inode_flag_set(inode, FI_COW_FILE);
}
static inline void *inline_data_addr(struct inode *inode, struct page *page)
{
__le32 *addr = get_dnode_addr(inode, page);
return (void *)(addr + DEF_INLINE_RESERVED_SIZE);
}
static inline int f2fs_has_inline_dentry(struct inode *inode)
{
return is_inode_flag_set(inode, FI_INLINE_DENTRY);
}
static inline int is_file(struct inode *inode, int type)
{
return F2FS_I(inode)->i_advise & type;
}
static inline void set_file(struct inode *inode, int type)
{
if (is_file(inode, type))
return;
F2FS_I(inode)->i_advise |= type;
f2fs_mark_inode_dirty_sync(inode, true);
}
static inline void clear_file(struct inode *inode, int type)
{
if (!is_file(inode, type))
return;
F2FS_I(inode)->i_advise &= ~type;
f2fs_mark_inode_dirty_sync(inode, true);
}
static inline bool f2fs_is_time_consistent(struct inode *inode)
{
struct timespec64 ts = inode_get_atime(inode);
if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &ts))
return false;
ts = inode_get_ctime(inode);
if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &ts))
return false;
ts = inode_get_mtime(inode);
if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &ts))
return false;
return true;
}
static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync)
{
bool ret;
if (dsync) {
struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
spin_lock(&sbi->inode_lock[DIRTY_META]);
ret = list_empty(&F2FS_I(inode)->gdirty_list);
spin_unlock(&sbi->inode_lock[DIRTY_META]);
return ret;
}
if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) ||
file_keep_isize(inode) ||
i_size_read(inode) & ~PAGE_MASK)
return false;
if (!f2fs_is_time_consistent(inode))
return false;
f2fs: cover last_disk_size update with spinlock This change solves below hangtask issue: INFO: task kworker/u16:1:58 blocked for more than 122 seconds. Not tainted 5.6.0-rc2-00590-g9983bdae4974e #11 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. kworker/u16:1 D 0 58 2 0x00000000 Workqueue: writeback wb_workfn (flush-179:0) Backtrace: (__schedule) from [<c0913234>] (schedule+0x78/0xf4) (schedule) from [<c017ec74>] (rwsem_down_write_slowpath+0x24c/0x4c0) (rwsem_down_write_slowpath) from [<c0915f2c>] (down_write+0x6c/0x70) (down_write) from [<c0435b80>] (f2fs_write_single_data_page+0x608/0x7ac) (f2fs_write_single_data_page) from [<c0435fd8>] (f2fs_write_cache_pages+0x2b4/0x7c4) (f2fs_write_cache_pages) from [<c043682c>] (f2fs_write_data_pages+0x344/0x35c) (f2fs_write_data_pages) from [<c0267ee8>] (do_writepages+0x3c/0xd4) (do_writepages) from [<c0310cbc>] (__writeback_single_inode+0x44/0x454) (__writeback_single_inode) from [<c03112d0>] (writeback_sb_inodes+0x204/0x4b0) (writeback_sb_inodes) from [<c03115cc>] (__writeback_inodes_wb+0x50/0xe4) (__writeback_inodes_wb) from [<c03118f4>] (wb_writeback+0x294/0x338) (wb_writeback) from [<c0312dac>] (wb_workfn+0x35c/0x54c) (wb_workfn) from [<c014f2b8>] (process_one_work+0x214/0x544) (process_one_work) from [<c014f634>] (worker_thread+0x4c/0x574) (worker_thread) from [<c01564fc>] (kthread+0x144/0x170) (kthread) from [<c01010e8>] (ret_from_fork+0x14/0x2c) Reported-and-tested-by: Ondřej Jirman <megi@xff.cz> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2020-02-27 11:30:03 +00:00
spin_lock(&F2FS_I(inode)->i_size_lock);
ret = F2FS_I(inode)->last_disk_size == i_size_read(inode);
f2fs: cover last_disk_size update with spinlock This change solves below hangtask issue: INFO: task kworker/u16:1:58 blocked for more than 122 seconds. Not tainted 5.6.0-rc2-00590-g9983bdae4974e #11 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. kworker/u16:1 D 0 58 2 0x00000000 Workqueue: writeback wb_workfn (flush-179:0) Backtrace: (__schedule) from [<c0913234>] (schedule+0x78/0xf4) (schedule) from [<c017ec74>] (rwsem_down_write_slowpath+0x24c/0x4c0) (rwsem_down_write_slowpath) from [<c0915f2c>] (down_write+0x6c/0x70) (down_write) from [<c0435b80>] (f2fs_write_single_data_page+0x608/0x7ac) (f2fs_write_single_data_page) from [<c0435fd8>] (f2fs_write_cache_pages+0x2b4/0x7c4) (f2fs_write_cache_pages) from [<c043682c>] (f2fs_write_data_pages+0x344/0x35c) (f2fs_write_data_pages) from [<c0267ee8>] (do_writepages+0x3c/0xd4) (do_writepages) from [<c0310cbc>] (__writeback_single_inode+0x44/0x454) (__writeback_single_inode) from [<c03112d0>] (writeback_sb_inodes+0x204/0x4b0) (writeback_sb_inodes) from [<c03115cc>] (__writeback_inodes_wb+0x50/0xe4) (__writeback_inodes_wb) from [<c03118f4>] (wb_writeback+0x294/0x338) (wb_writeback) from [<c0312dac>] (wb_workfn+0x35c/0x54c) (wb_workfn) from [<c014f2b8>] (process_one_work+0x214/0x544) (process_one_work) from [<c014f634>] (worker_thread+0x4c/0x574) (worker_thread) from [<c01564fc>] (kthread+0x144/0x170) (kthread) from [<c01010e8>] (ret_from_fork+0x14/0x2c) Reported-and-tested-by: Ondřej Jirman <megi@xff.cz> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2020-02-27 11:30:03 +00:00
spin_unlock(&F2FS_I(inode)->i_size_lock);
return ret;
}
static inline bool f2fs_readonly(struct super_block *sb)
{
return sb_rdonly(sb);
}
static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi)
{
return is_set_ckpt_flags(sbi, CP_ERROR_FLAG);
}
static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi,
size_t size, gfp_t flags)
{
if (time_to_inject(sbi, FAULT_KMALLOC))
return NULL;
return kmalloc(size, flags);
}
static inline void *f2fs_getname(struct f2fs_sb_info *sbi)
{
if (time_to_inject(sbi, FAULT_KMALLOC))
return NULL;
return __getname();
}
static inline void f2fs_putname(char *buf)
{
__putname(buf);
}
static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi,
size_t size, gfp_t flags)
{
return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO);
}
static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi,
size_t size, gfp_t flags)
{
if (time_to_inject(sbi, FAULT_KVMALLOC))
return NULL;
return kvmalloc(size, flags);
}
static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi,
size_t size, gfp_t flags)
{
return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO);
}
f2fs: enhance on-disk inode structure scalability This patch add new flag F2FS_EXTRA_ATTR storing in inode.i_inline to indicate that on-disk structure of current inode is extended. In order to extend, we changed the inode structure a bit: Original one: struct f2fs_inode { ... struct f2fs_extent i_ext; __le32 i_addr[DEF_ADDRS_PER_INODE]; __le32 i_nid[DEF_NIDS_PER_INODE]; } Extended one: struct f2fs_inode { ... struct f2fs_extent i_ext; union { struct { __le16 i_extra_isize; __le16 i_padding; __le32 i_extra_end[0]; }; __le32 i_addr[DEF_ADDRS_PER_INODE]; }; __le32 i_nid[DEF_NIDS_PER_INODE]; } Once F2FS_EXTRA_ATTR is set, we will steal four bytes in the head of i_addr field for storing i_extra_isize and i_padding. with i_extra_isize, we can calculate actual size of reserved space in i_addr, available attribute fields included in total extra attribute fields for current inode can be described as below: +--------------------+ | .i_mode | | ... | | .i_ext | +--------------------+ | .i_extra_isize |-----+ | .i_padding | | | .i_prjid | | | .i_atime_extra | | | .i_ctime_extra | | | .i_mtime_extra |<----+ | .i_inode_cs |<----- store blkaddr/inline from here | .i_xattr_cs | | ... | +--------------------+ | | | block address | | | +--------------------+ | .i_nid | +--------------------+ | node_footer | | (nid, ino, offset) | +--------------------+ Hence, with this patch, we would enhance scalability of f2fs inode for storing more newly added attribute. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-07-18 16:19:06 +00:00
static inline int get_extra_isize(struct inode *inode)
{
f2fs: enhance on-disk inode structure scalability This patch add new flag F2FS_EXTRA_ATTR storing in inode.i_inline to indicate that on-disk structure of current inode is extended. In order to extend, we changed the inode structure a bit: Original one: struct f2fs_inode { ... struct f2fs_extent i_ext; __le32 i_addr[DEF_ADDRS_PER_INODE]; __le32 i_nid[DEF_NIDS_PER_INODE]; } Extended one: struct f2fs_inode { ... struct f2fs_extent i_ext; union { struct { __le16 i_extra_isize; __le16 i_padding; __le32 i_extra_end[0]; }; __le32 i_addr[DEF_ADDRS_PER_INODE]; }; __le32 i_nid[DEF_NIDS_PER_INODE]; } Once F2FS_EXTRA_ATTR is set, we will steal four bytes in the head of i_addr field for storing i_extra_isize and i_padding. with i_extra_isize, we can calculate actual size of reserved space in i_addr, available attribute fields included in total extra attribute fields for current inode can be described as below: +--------------------+ | .i_mode | | ... | | .i_ext | +--------------------+ | .i_extra_isize |-----+ | .i_padding | | | .i_prjid | | | .i_atime_extra | | | .i_ctime_extra | | | .i_mtime_extra |<----+ | .i_inode_cs |<----- store blkaddr/inline from here | .i_xattr_cs | | ... | +--------------------+ | | | block address | | | +--------------------+ | .i_nid | +--------------------+ | node_footer | | (nid, ino, offset) | +--------------------+ Hence, with this patch, we would enhance scalability of f2fs inode for storing more newly added attribute. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-07-18 16:19:06 +00:00
return F2FS_I(inode)->i_extra_isize / sizeof(__le32);
}
f2fs: support flexible inline xattr size Now, in product, more and more features based on file encryption were introduced, their demand of xattr space is increasing, however, inline xattr has fixed-size of 200 bytes, once inline xattr space is full, new increased xattr data would occupy additional xattr block which may bring us more space usage and performance regression during persisting. In order to resolve above issue, it's better to expand inline xattr size flexibly according to user's requirement. So this patch introduces new filesystem feature 'flexible inline xattr', and new mount option 'inline_xattr_size=%u', once mkfs enables the feature, we can use the option to make f2fs supporting flexible inline xattr size. To support this feature, we add extra attribute i_inline_xattr_size in inode layout, indicating that how many space inline xattr borrows from block address mapping space in inode layout, by this, we can easily locate and store flexible-sized inline xattr data in inode. Inode disk layout: +----------------------+ | .i_mode | | ... | | .i_ext | +----------------------+ | .i_extra_isize | | .i_inline_xattr_size |-----------+ | ... | | +----------------------+ | | .i_addr | | | - block address or | | | - inline data | | +----------------------+<---+ v | inline xattr | +---inline xattr range +----------------------+<---+ | .i_nid | +----------------------+ | node_footer | | (nid, ino, offset) | +----------------------+ Note that, we have to cnosider backward compatibility which reserved inline_data space, 200 bytes, all the time, reported by Sheng Yong. Previous inline data or directory always reserved 200 bytes in inode layout, even if inline_xattr is disabled. In order to keep inline_dentry's structure for backward compatibility, we get the space back only from inline_data. Signed-off-by: Chao Yu <yuchao0@huawei.com> Reported-by: Sheng Yong <shengyong1@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-09-06 13:59:50 +00:00
static inline int get_inline_xattr_addrs(struct inode *inode)
{
return F2FS_I(inode)->i_inline_xattr_size;
}
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
#define f2fs_get_inode_mode(i) \
((is_inode_flag_set(i, FI_ACL_MODE)) ? \
(F2FS_I(i)->i_acl_mode) : ((i)->i_mode))
#define F2FS_MIN_EXTRA_ATTR_SIZE (sizeof(__le32))
f2fs: enhance on-disk inode structure scalability This patch add new flag F2FS_EXTRA_ATTR storing in inode.i_inline to indicate that on-disk structure of current inode is extended. In order to extend, we changed the inode structure a bit: Original one: struct f2fs_inode { ... struct f2fs_extent i_ext; __le32 i_addr[DEF_ADDRS_PER_INODE]; __le32 i_nid[DEF_NIDS_PER_INODE]; } Extended one: struct f2fs_inode { ... struct f2fs_extent i_ext; union { struct { __le16 i_extra_isize; __le16 i_padding; __le32 i_extra_end[0]; }; __le32 i_addr[DEF_ADDRS_PER_INODE]; }; __le32 i_nid[DEF_NIDS_PER_INODE]; } Once F2FS_EXTRA_ATTR is set, we will steal four bytes in the head of i_addr field for storing i_extra_isize and i_padding. with i_extra_isize, we can calculate actual size of reserved space in i_addr, available attribute fields included in total extra attribute fields for current inode can be described as below: +--------------------+ | .i_mode | | ... | | .i_ext | +--------------------+ | .i_extra_isize |-----+ | .i_padding | | | .i_prjid | | | .i_atime_extra | | | .i_ctime_extra | | | .i_mtime_extra |<----+ | .i_inode_cs |<----- store blkaddr/inline from here | .i_xattr_cs | | ... | +--------------------+ | | | block address | | | +--------------------+ | .i_nid | +--------------------+ | node_footer | | (nid, ino, offset) | +--------------------+ Hence, with this patch, we would enhance scalability of f2fs inode for storing more newly added attribute. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-07-18 16:19:06 +00:00
#define F2FS_TOTAL_EXTRA_ATTR_SIZE \
(offsetof(struct f2fs_inode, i_extra_end) - \
offsetof(struct f2fs_inode, i_extra_isize)) \
#define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr))
#define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \
((offsetof(typeof(*(f2fs_inode)), field) + \
sizeof((f2fs_inode)->field)) \
<= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \
#define __is_large_section(sbi) (SEGS_PER_SEC(sbi) > 1)
#define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META)
f2fs: fix to do sanity check with block address in main area This patch add to do sanity check with below field: - cp_pack_total_block_count - blkaddr of data/node - extent info - Overview BUG() in verify_block_addr() when writing to a corrupted f2fs image - Reproduce (4.18 upstream kernel) - POC (poc.c) static void activity(char *mpoint) { char *foo_bar_baz; int err; static int buf[8192]; memset(buf, 0, sizeof(buf)); err = asprintf(&foo_bar_baz, "%s/foo/bar/baz", mpoint); int fd = open(foo_bar_baz, O_RDWR | O_TRUNC, 0777); if (fd >= 0) { write(fd, (char *)buf, sizeof(buf)); fdatasync(fd); close(fd); } } int main(int argc, char *argv[]) { activity(argv[1]); return 0; } - Kernel message [ 689.349473] F2FS-fs (loop0): Mounted with checkpoint version = 3 [ 699.728662] WARNING: CPU: 0 PID: 1309 at fs/f2fs/segment.c:2860 f2fs_inplace_write_data+0x232/0x240 [ 699.728670] Modules linked in: snd_hda_codec_generic snd_hda_intel snd_hda_codec snd_hwdep snd_hda_core snd_pcm snd_timer snd mac_hid i2c_piix4 soundcore ib_iser rdma_cm iw_cm ib_cm ib_core iscsi_tcp libiscsi_tcp libiscsi scsi_transport_iscsi raid10 raid456 async_raid6_recov async_memcpy async_pq async_xor async_tx raid1 raid0 multipath linear 8139too crct10dif_pclmul crc32_pclmul qxl drm_kms_helper syscopyarea aesni_intel sysfillrect sysimgblt fb_sys_fops ttm drm aes_x86_64 crypto_simd cryptd 8139cp glue_helper mii pata_acpi floppy [ 699.729056] CPU: 0 PID: 1309 Comm: a.out Not tainted 4.18.0-rc1+ #4 [ 699.729064] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014 [ 699.729074] RIP: 0010:f2fs_inplace_write_data+0x232/0x240 [ 699.729076] Code: ff e9 cf fe ff ff 49 8d 7d 10 e8 39 45 ad ff 4d 8b 7d 10 be 04 00 00 00 49 8d 7f 48 e8 07 49 ad ff 45 8b 7f 48 e9 fb fe ff ff <0f> 0b f0 41 80 4d 48 04 e9 65 fe ff ff 90 66 66 66 66 90 55 48 8d [ 699.729130] RSP: 0018:ffff8801f43af568 EFLAGS: 00010202 [ 699.729139] RAX: 000000000000003f RBX: ffff8801f43af7b8 RCX: ffffffffb88c9113 [ 699.729142] RDX: 0000000000000003 RSI: dffffc0000000000 RDI: ffff8802024e5540 [ 699.729144] RBP: ffff8801f43af590 R08: 0000000000000009 R09: ffffffffffffffe8 [ 699.729147] R10: 0000000000000001 R11: ffffed0039b0596a R12: ffff8802024e5540 [ 699.729149] R13: ffff8801f0335500 R14: ffff8801e3e7a700 R15: ffff8801e1ee4450 [ 699.729154] FS: 00007f9bf97f5700(0000) GS:ffff8801f6e00000(0000) knlGS:0000000000000000 [ 699.729156] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 699.729159] CR2: 00007f9bf925d170 CR3: 00000001f0c34000 CR4: 00000000000006f0 [ 699.729171] Call Trace: [ 699.729192] f2fs_do_write_data_page+0x2e2/0xe00 [ 699.729203] ? f2fs_should_update_outplace+0xd0/0xd0 [ 699.729238] ? memcg_drain_all_list_lrus+0x280/0x280 [ 699.729269] ? __radix_tree_replace+0xa3/0x120 [ 699.729276] __write_data_page+0x5c7/0xe30 [ 699.729291] ? kasan_check_read+0x11/0x20 [ 699.729310] ? page_mapped+0x8a/0x110 [ 699.729321] ? page_mkclean+0xe9/0x160 [ 699.729327] ? f2fs_do_write_data_page+0xe00/0xe00 [ 699.729331] ? invalid_page_referenced_vma+0x130/0x130 [ 699.729345] ? clear_page_dirty_for_io+0x332/0x450 [ 699.729351] f2fs_write_cache_pages+0x4ca/0x860 [ 699.729358] ? __write_data_page+0xe30/0xe30 [ 699.729374] ? percpu_counter_add_batch+0x22/0xa0 [ 699.729380] ? kasan_check_write+0x14/0x20 [ 699.729391] ? _raw_spin_lock+0x17/0x40 [ 699.729403] ? f2fs_mark_inode_dirty_sync.part.18+0x16/0x30 [ 699.729413] ? iov_iter_advance+0x113/0x640 [ 699.729418] ? f2fs_write_end+0x133/0x2e0 [ 699.729423] ? balance_dirty_pages_ratelimited+0x239/0x640 [ 699.729428] f2fs_write_data_pages+0x329/0x520 [ 699.729433] ? generic_perform_write+0x250/0x320 [ 699.729438] ? f2fs_write_cache_pages+0x860/0x860 [ 699.729454] ? current_time+0x110/0x110 [ 699.729459] ? f2fs_preallocate_blocks+0x1ef/0x370 [ 699.729464] do_writepages+0x37/0xb0 [ 699.729468] ? f2fs_write_cache_pages+0x860/0x860 [ 699.729472] ? do_writepages+0x37/0xb0 [ 699.729478] __filemap_fdatawrite_range+0x19a/0x1f0 [ 699.729483] ? delete_from_page_cache_batch+0x4e0/0x4e0 [ 699.729496] ? __vfs_write+0x2b2/0x410 [ 699.729501] file_write_and_wait_range+0x66/0xb0 [ 699.729506] f2fs_do_sync_file+0x1f9/0xd90 [ 699.729511] ? truncate_partial_data_page+0x290/0x290 [ 699.729521] ? __sb_end_write+0x30/0x50 [ 699.729526] ? vfs_write+0x20f/0x260 [ 699.729530] f2fs_sync_file+0x9a/0xb0 [ 699.729534] ? f2fs_do_sync_file+0xd90/0xd90 [ 699.729548] vfs_fsync_range+0x68/0x100 [ 699.729554] ? __fget_light+0xc9/0xe0 [ 699.729558] do_fsync+0x3d/0x70 [ 699.729562] __x64_sys_fdatasync+0x24/0x30 [ 699.729585] do_syscall_64+0x78/0x170 [ 699.729595] entry_SYSCALL_64_after_hwframe+0x44/0xa9 [ 699.729613] RIP: 0033:0x7f9bf930d800 [ 699.729615] Code: 00 f7 d8 64 89 01 48 83 c8 ff c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 83 3d 49 bf 2c 00 00 75 10 b8 4b 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 31 c3 48 83 ec 08 e8 be 78 01 00 48 89 04 24 [ 699.729668] RSP: 002b:00007ffee3606c68 EFLAGS: 00000246 ORIG_RAX: 000000000000004b [ 699.729673] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f9bf930d800 [ 699.729675] RDX: 0000000000008000 RSI: 00000000006010a0 RDI: 0000000000000003 [ 699.729678] RBP: 00007ffee3606ca0 R08: 0000000001503010 R09: 0000000000000000 [ 699.729680] R10: 00000000000002e8 R11: 0000000000000246 R12: 0000000000400610 [ 699.729683] R13: 00007ffee3606da0 R14: 0000000000000000 R15: 0000000000000000 [ 699.729687] ---[ end trace 4ce02f25ff7d3df5 ]--- [ 699.729782] ------------[ cut here ]------------ [ 699.729785] kernel BUG at fs/f2fs/segment.h:654! [ 699.731055] invalid opcode: 0000 [#1] SMP KASAN PTI [ 699.732104] CPU: 0 PID: 1309 Comm: a.out Tainted: G W 4.18.0-rc1+ #4 [ 699.733684] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014 [ 699.735611] RIP: 0010:f2fs_submit_page_bio+0x29b/0x730 [ 699.736649] Code: 54 49 8d bd 18 04 00 00 e8 b2 59 af ff 41 8b 8d 18 04 00 00 8b 45 b8 41 d3 e6 44 01 f0 4c 8d 73 14 41 39 c7 0f 82 37 fe ff ff <0f> 0b 65 8b 05 2c 04 77 47 89 c0 48 0f a3 05 52 c1 d5 01 0f 92 c0 [ 699.740524] RSP: 0018:ffff8801f43af508 EFLAGS: 00010283 [ 699.741573] RAX: 0000000000000000 RBX: ffff8801f43af7b8 RCX: ffffffffb88a7cef [ 699.743006] RDX: 0000000000000007 RSI: dffffc0000000000 RDI: ffff8801e3e7a64c [ 699.744426] RBP: ffff8801f43af558 R08: ffffed003e066b55 R09: ffffed003e066b55 [ 699.745833] R10: 0000000000000001 R11: ffffed003e066b54 R12: ffffea0007876940 [ 699.747256] R13: ffff8801f0335500 R14: ffff8801e3e7a600 R15: 0000000000000001 [ 699.748683] FS: 00007f9bf97f5700(0000) GS:ffff8801f6e00000(0000) knlGS:0000000000000000 [ 699.750293] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 699.751462] CR2: 00007f9bf925d170 CR3: 00000001f0c34000 CR4: 00000000000006f0 [ 699.752874] Call Trace: [ 699.753386] ? f2fs_inplace_write_data+0x93/0x240 [ 699.754341] f2fs_inplace_write_data+0xd2/0x240 [ 699.755271] f2fs_do_write_data_page+0x2e2/0xe00 [ 699.756214] ? f2fs_should_update_outplace+0xd0/0xd0 [ 699.757215] ? memcg_drain_all_list_lrus+0x280/0x280 [ 699.758209] ? __radix_tree_replace+0xa3/0x120 [ 699.759164] __write_data_page+0x5c7/0xe30 [ 699.760002] ? kasan_check_read+0x11/0x20 [ 699.760823] ? page_mapped+0x8a/0x110 [ 699.761573] ? page_mkclean+0xe9/0x160 [ 699.762345] ? f2fs_do_write_data_page+0xe00/0xe00 [ 699.763332] ? invalid_page_referenced_vma+0x130/0x130 [ 699.764374] ? clear_page_dirty_for_io+0x332/0x450 [ 699.765347] f2fs_write_cache_pages+0x4ca/0x860 [ 699.766276] ? __write_data_page+0xe30/0xe30 [ 699.767161] ? percpu_counter_add_batch+0x22/0xa0 [ 699.768112] ? kasan_check_write+0x14/0x20 [ 699.768951] ? _raw_spin_lock+0x17/0x40 [ 699.769739] ? f2fs_mark_inode_dirty_sync.part.18+0x16/0x30 [ 699.770885] ? iov_iter_advance+0x113/0x640 [ 699.771743] ? f2fs_write_end+0x133/0x2e0 [ 699.772569] ? balance_dirty_pages_ratelimited+0x239/0x640 [ 699.773680] f2fs_write_data_pages+0x329/0x520 [ 699.774603] ? generic_perform_write+0x250/0x320 [ 699.775544] ? f2fs_write_cache_pages+0x860/0x860 [ 699.776510] ? current_time+0x110/0x110 [ 699.777299] ? f2fs_preallocate_blocks+0x1ef/0x370 [ 699.778279] do_writepages+0x37/0xb0 [ 699.779026] ? f2fs_write_cache_pages+0x860/0x860 [ 699.779978] ? do_writepages+0x37/0xb0 [ 699.780755] __filemap_fdatawrite_range+0x19a/0x1f0 [ 699.781746] ? delete_from_page_cache_batch+0x4e0/0x4e0 [ 699.782820] ? __vfs_write+0x2b2/0x410 [ 699.783597] file_write_and_wait_range+0x66/0xb0 [ 699.784540] f2fs_do_sync_file+0x1f9/0xd90 [ 699.785381] ? truncate_partial_data_page+0x290/0x290 [ 699.786415] ? __sb_end_write+0x30/0x50 [ 699.787204] ? vfs_write+0x20f/0x260 [ 699.787941] f2fs_sync_file+0x9a/0xb0 [ 699.788694] ? f2fs_do_sync_file+0xd90/0xd90 [ 699.789572] vfs_fsync_range+0x68/0x100 [ 699.790360] ? __fget_light+0xc9/0xe0 [ 699.791128] do_fsync+0x3d/0x70 [ 699.791779] __x64_sys_fdatasync+0x24/0x30 [ 699.792614] do_syscall_64+0x78/0x170 [ 699.793371] entry_SYSCALL_64_after_hwframe+0x44/0xa9 [ 699.794406] RIP: 0033:0x7f9bf930d800 [ 699.795134] Code: 00 f7 d8 64 89 01 48 83 c8 ff c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 83 3d 49 bf 2c 00 00 75 10 b8 4b 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 31 c3 48 83 ec 08 e8 be 78 01 00 48 89 04 24 [ 699.798960] RSP: 002b:00007ffee3606c68 EFLAGS: 00000246 ORIG_RAX: 000000000000004b [ 699.800483] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f9bf930d800 [ 699.801923] RDX: 0000000000008000 RSI: 00000000006010a0 RDI: 0000000000000003 [ 699.803373] RBP: 00007ffee3606ca0 R08: 0000000001503010 R09: 0000000000000000 [ 699.804798] R10: 00000000000002e8 R11: 0000000000000246 R12: 0000000000400610 [ 699.806233] R13: 00007ffee3606da0 R14: 0000000000000000 R15: 0000000000000000 [ 699.807667] Modules linked in: snd_hda_codec_generic snd_hda_intel snd_hda_codec snd_hwdep snd_hda_core snd_pcm snd_timer snd mac_hid i2c_piix4 soundcore ib_iser rdma_cm iw_cm ib_cm ib_core iscsi_tcp libiscsi_tcp libiscsi scsi_transport_iscsi raid10 raid456 async_raid6_recov async_memcpy async_pq async_xor async_tx raid1 raid0 multipath linear 8139too crct10dif_pclmul crc32_pclmul qxl drm_kms_helper syscopyarea aesni_intel sysfillrect sysimgblt fb_sys_fops ttm drm aes_x86_64 crypto_simd cryptd 8139cp glue_helper mii pata_acpi floppy [ 699.817079] ---[ end trace 4ce02f25ff7d3df6 ]--- [ 699.818068] RIP: 0010:f2fs_submit_page_bio+0x29b/0x730 [ 699.819114] Code: 54 49 8d bd 18 04 00 00 e8 b2 59 af ff 41 8b 8d 18 04 00 00 8b 45 b8 41 d3 e6 44 01 f0 4c 8d 73 14 41 39 c7 0f 82 37 fe ff ff <0f> 0b 65 8b 05 2c 04 77 47 89 c0 48 0f a3 05 52 c1 d5 01 0f 92 c0 [ 699.822919] RSP: 0018:ffff8801f43af508 EFLAGS: 00010283 [ 699.823977] RAX: 0000000000000000 RBX: ffff8801f43af7b8 RCX: ffffffffb88a7cef [ 699.825436] RDX: 0000000000000007 RSI: dffffc0000000000 RDI: ffff8801e3e7a64c [ 699.826881] RBP: ffff8801f43af558 R08: ffffed003e066b55 R09: ffffed003e066b55 [ 699.828292] R10: 0000000000000001 R11: ffffed003e066b54 R12: ffffea0007876940 [ 699.829750] R13: ffff8801f0335500 R14: ffff8801e3e7a600 R15: 0000000000000001 [ 699.831192] FS: 00007f9bf97f5700(0000) GS:ffff8801f6e00000(0000) knlGS:0000000000000000 [ 699.832793] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 699.833981] CR2: 00007f9bf925d170 CR3: 00000001f0c34000 CR4: 00000000000006f0 [ 699.835556] ================================================================== [ 699.837029] BUG: KASAN: stack-out-of-bounds in update_stack_state+0x38c/0x3e0 [ 699.838462] Read of size 8 at addr ffff8801f43af970 by task a.out/1309 [ 699.840086] CPU: 0 PID: 1309 Comm: a.out Tainted: G D W 4.18.0-rc1+ #4 [ 699.841603] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014 [ 699.843475] Call Trace: [ 699.843982] dump_stack+0x7b/0xb5 [ 699.844661] print_address_description+0x70/0x290 [ 699.845607] kasan_report+0x291/0x390 [ 699.846351] ? update_stack_state+0x38c/0x3e0 [ 699.853831] __asan_load8+0x54/0x90 [ 699.854569] update_stack_state+0x38c/0x3e0 [ 699.855428] ? __read_once_size_nocheck.constprop.7+0x20/0x20 [ 699.856601] ? __save_stack_trace+0x5e/0x100 [ 699.857476] unwind_next_frame.part.5+0x18e/0x490 [ 699.858448] ? unwind_dump+0x290/0x290 [ 699.859217] ? clear_page_dirty_for_io+0x332/0x450 [ 699.860185] __unwind_start+0x106/0x190 [ 699.860974] __save_stack_trace+0x5e/0x100 [ 699.861808] ? __save_stack_trace+0x5e/0x100 [ 699.862691] ? unlink_anon_vmas+0xba/0x2c0 [ 699.863525] save_stack_trace+0x1f/0x30 [ 699.864312] save_stack+0x46/0xd0 [ 699.864993] ? __alloc_pages_slowpath+0x1420/0x1420 [ 699.865990] ? flush_tlb_mm_range+0x15e/0x220 [ 699.866889] ? kasan_check_write+0x14/0x20 [ 699.867724] ? __dec_node_state+0x92/0xb0 [ 699.868543] ? lock_page_memcg+0x85/0xf0 [ 699.869350] ? unlock_page_memcg+0x16/0x80 [ 699.870185] ? page_remove_rmap+0x198/0x520 [ 699.871048] ? mark_page_accessed+0x133/0x200 [ 699.871930] ? _cond_resched+0x1a/0x50 [ 699.872700] ? unmap_page_range+0xcd4/0xe50 [ 699.873551] ? rb_next+0x58/0x80 [ 699.874217] ? rb_next+0x58/0x80 [ 699.874895] __kasan_slab_free+0x13c/0x1a0 [ 699.875734] ? unlink_anon_vmas+0xba/0x2c0 [ 699.876563] kasan_slab_free+0xe/0x10 [ 699.877315] kmem_cache_free+0x89/0x1e0 [ 699.878095] unlink_anon_vmas+0xba/0x2c0 [ 699.878913] free_pgtables+0x101/0x1b0 [ 699.879677] exit_mmap+0x146/0x2a0 [ 699.880378] ? __ia32_sys_munmap+0x50/0x50 [ 699.881214] ? kasan_check_read+0x11/0x20 [ 699.882052] ? mm_update_next_owner+0x322/0x380 [ 699.882985] mmput+0x8b/0x1d0 [ 699.883602] do_exit+0x43a/0x1390 [ 699.884288] ? mm_update_next_owner+0x380/0x380 [ 699.885212] ? f2fs_sync_file+0x9a/0xb0 [ 699.885995] ? f2fs_do_sync_file+0xd90/0xd90 [ 699.886877] ? vfs_fsync_range+0x68/0x100 [ 699.887694] ? __fget_light+0xc9/0xe0 [ 699.888442] ? do_fsync+0x3d/0x70 [ 699.889118] ? __x64_sys_fdatasync+0x24/0x30 [ 699.889996] rewind_stack_do_exit+0x17/0x20 [ 699.890860] RIP: 0033:0x7f9bf930d800 [ 699.891585] Code: Bad RIP value. [ 699.892268] RSP: 002b:00007ffee3606c68 EFLAGS: 00000246 ORIG_RAX: 000000000000004b [ 699.893781] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f9bf930d800 [ 699.895220] RDX: 0000000000008000 RSI: 00000000006010a0 RDI: 0000000000000003 [ 699.896643] RBP: 00007ffee3606ca0 R08: 0000000001503010 R09: 0000000000000000 [ 699.898069] R10: 00000000000002e8 R11: 0000000000000246 R12: 0000000000400610 [ 699.899505] R13: 00007ffee3606da0 R14: 0000000000000000 R15: 0000000000000000 [ 699.901241] The buggy address belongs to the page: [ 699.902215] page:ffffea0007d0ebc0 count:0 mapcount:0 mapping:0000000000000000 index:0x0 [ 699.903811] flags: 0x2ffff0000000000() [ 699.904585] raw: 02ffff0000000000 0000000000000000 ffffffff07d00101 0000000000000000 [ 699.906125] raw: 0000000000000000 0000000000240000 00000000ffffffff 0000000000000000 [ 699.907673] page dumped because: kasan: bad access detected [ 699.909108] Memory state around the buggy address: [ 699.910077] ffff8801f43af800: 00 f1 f1 f1 f1 00 f4 f4 f4 f3 f3 f3 f3 00 00 00 [ 699.911528] ffff8801f43af880: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 [ 699.912953] >ffff8801f43af900: 00 00 00 00 00 00 00 00 f1 01 f4 f4 f4 f2 f2 f2 [ 699.914392] ^ [ 699.915758] ffff8801f43af980: f2 00 f4 f4 00 00 00 00 f2 00 00 00 00 00 00 00 [ 699.917193] ffff8801f43afa00: 00 00 00 00 00 00 00 00 00 f3 f3 f3 00 00 00 00 [ 699.918634] ================================================================== - Location https://elixir.bootlin.com/linux/v4.18-rc1/source/fs/f2fs/segment.h#L644 Reported-by Wen Xu <wen.xu@gatech.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-01 11:13:44 +00:00
bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
block_t blkaddr, int type);
static inline void verify_blkaddr(struct f2fs_sb_info *sbi,
block_t blkaddr, int type)
{
if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type))
f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.",
blkaddr, type);
}
static inline bool __is_valid_data_blkaddr(block_t blkaddr)
{
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>
2019-11-01 10:07:14 +00:00
if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR ||
blkaddr == COMPRESS_ADDR)
return false;
return true;
}
/*
* file.c
*/
int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync);
int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock);
int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock);
int f2fs_truncate(struct inode *inode);
int f2fs_getattr(struct mnt_idmap *idmap, const struct path *path,
struct kstat *stat, u32 request_mask, unsigned int flags);
int f2fs_setattr(struct mnt_idmap *idmap, struct dentry *dentry,
struct iattr *attr);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end);
void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count);
int f2fs_do_shutdown(struct f2fs_sb_info *sbi, unsigned int flag,
bool readonly);
int f2fs_precache_extents(struct inode *inode);
int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa);
int f2fs_fileattr_set(struct mnt_idmap *idmap,
struct dentry *dentry, struct fileattr *fa);
long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg);
long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid);
int f2fs_pin_file_control(struct inode *inode, bool inc);
/*
* inode.c
*/
void f2fs_set_inode_flags(struct inode *inode);
bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page);
void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page);
struct inode *f2fs_iget(struct super_block *sb, unsigned long ino);
struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink);
void f2fs_update_inode(struct inode *inode, struct page *node_page);
void f2fs_update_inode_page(struct inode *inode);
int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc);
void f2fs_evict_inode(struct inode *inode);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
void f2fs_handle_failed_inode(struct inode *inode);
/*
* namei.c
*/
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name,
bool hot, bool set);
struct dentry *f2fs_get_parent(struct dentry *child);
int f2fs_get_tmpfile(struct mnt_idmap *idmap, struct inode *dir,
struct inode **new_inode);
/*
* dir.c
*/
#if IS_ENABLED(CONFIG_UNICODE)
2020-05-07 07:59:04 +00:00
int f2fs_init_casefolded_name(const struct inode *dir,
struct f2fs_filename *fname);
void f2fs_free_casefolded_name(struct f2fs_filename *fname);
#else
static inline int f2fs_init_casefolded_name(const struct inode *dir,
struct f2fs_filename *fname)
{
return 0;
}
static inline void f2fs_free_casefolded_name(struct f2fs_filename *fname)
{
}
#endif /* CONFIG_UNICODE */
2020-05-07 07:59:04 +00:00
int f2fs_setup_filename(struct inode *dir, const struct qstr *iname,
int lookup, struct f2fs_filename *fname);
int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry,
struct f2fs_filename *fname);
void f2fs_free_filename(struct f2fs_filename *fname);
struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d,
const struct f2fs_filename *fname, int *max_slots);
int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d,
unsigned int start_pos, struct fscrypt_str *fstr);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent,
struct f2fs_dentry_ptr *d);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir,
2020-05-07 07:59:04 +00:00
const struct f2fs_filename *fname, struct page *dpage);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode,
unsigned int current_depth);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots);
void f2fs_drop_nlink(struct inode *dir, struct inode *inode);
struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
2020-05-07 07:59:04 +00:00
const struct f2fs_filename *fname,
struct page **res_page);
struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir,
const struct qstr *child, struct page **res_page);
struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p);
ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr,
struct page **page);
void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de,
struct page *page, struct inode *inode);
bool f2fs_has_enough_room(struct inode *dir, struct page *ipage,
2020-05-07 07:59:04 +00:00
const struct f2fs_filename *fname);
void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d,
2020-05-07 07:59:04 +00:00
const struct fscrypt_str *name, f2fs_hash_t name_hash,
unsigned int bit_pos);
2020-05-07 07:59:04 +00:00
int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname,
struct inode *inode, nid_t ino, umode_t mode);
2020-05-07 07:59:04 +00:00
int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname,
struct inode *inode, nid_t ino, umode_t mode);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
int f2fs_do_add_link(struct inode *dir, const struct qstr *name,
struct inode *inode, nid_t ino, umode_t mode);
void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page,
struct inode *dir, struct inode *inode);
int f2fs_do_tmpfile(struct inode *inode, struct inode *dir,
struct f2fs_filename *fname);
bool f2fs_empty_dir(struct inode *dir);
static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode)
{
if (fscrypt_is_nokey_name(dentry))
return -ENOKEY;
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name,
inode, inode->i_ino, inode->i_mode);
}
/*
* super.c
*/
int f2fs_inode_dirtied(struct inode *inode, bool sync);
void f2fs_inode_synced(struct inode *inode);
int f2fs_dquot_initialize(struct inode *inode);
int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly);
int f2fs_quota_sync(struct super_block *sb, int type);
loff_t max_file_blocks(struct inode *inode);
void f2fs_quota_off_umount(struct super_block *sb);
f2fs: fix scheduling while atomic in decompression path [ 16.945668][ C0] Call trace: [ 16.945678][ C0] dump_backtrace+0x110/0x204 [ 16.945706][ C0] dump_stack_lvl+0x84/0xbc [ 16.945735][ C0] __schedule_bug+0xb8/0x1ac [ 16.945756][ C0] __schedule+0x724/0xbdc [ 16.945778][ C0] schedule+0x154/0x258 [ 16.945793][ C0] bit_wait_io+0x48/0xa4 [ 16.945808][ C0] out_of_line_wait_on_bit+0x114/0x198 [ 16.945824][ C0] __sync_dirty_buffer+0x1f8/0x2e8 [ 16.945853][ C0] __f2fs_commit_super+0x140/0x1f4 [ 16.945881][ C0] f2fs_commit_super+0x110/0x28c [ 16.945898][ C0] f2fs_handle_error+0x1f4/0x2f4 [ 16.945917][ C0] f2fs_decompress_cluster+0xc4/0x450 [ 16.945942][ C0] f2fs_end_read_compressed_page+0xc0/0xfc [ 16.945959][ C0] f2fs_handle_step_decompress+0x118/0x1cc [ 16.945978][ C0] f2fs_read_end_io+0x168/0x2b0 [ 16.945993][ C0] bio_endio+0x25c/0x2c8 [ 16.946015][ C0] dm_io_dec_pending+0x3e8/0x57c [ 16.946052][ C0] clone_endio+0x134/0x254 [ 16.946069][ C0] bio_endio+0x25c/0x2c8 [ 16.946084][ C0] blk_update_request+0x1d4/0x478 [ 16.946103][ C0] scsi_end_request+0x38/0x4cc [ 16.946129][ C0] scsi_io_completion+0x94/0x184 [ 16.946147][ C0] scsi_finish_command+0xe8/0x154 [ 16.946164][ C0] scsi_complete+0x90/0x1d8 [ 16.946181][ C0] blk_done_softirq+0xa4/0x11c [ 16.946198][ C0] _stext+0x184/0x614 [ 16.946214][ C0] __irq_exit_rcu+0x78/0x144 [ 16.946234][ C0] handle_domain_irq+0xd4/0x154 [ 16.946260][ C0] gic_handle_irq.33881+0x5c/0x27c [ 16.946281][ C0] call_on_irq_stack+0x40/0x70 [ 16.946298][ C0] do_interrupt_handler+0x48/0xa4 [ 16.946313][ C0] el1_interrupt+0x38/0x68 [ 16.946346][ C0] el1h_64_irq_handler+0x20/0x30 [ 16.946362][ C0] el1h_64_irq+0x78/0x7c [ 16.946377][ C0] finish_task_switch+0xc8/0x3d8 [ 16.946394][ C0] __schedule+0x600/0xbdc [ 16.946408][ C0] preempt_schedule_common+0x34/0x5c [ 16.946423][ C0] preempt_schedule+0x44/0x48 [ 16.946438][ C0] process_one_work+0x30c/0x550 [ 16.946456][ C0] worker_thread+0x414/0x8bc [ 16.946472][ C0] kthread+0x16c/0x1e0 [ 16.946486][ C0] ret_from_fork+0x10/0x20 Fixes: bff139b49d9f ("f2fs: handle decompress only post processing in softirq") Fixes: 95fa90c9e5a7 ("f2fs: support recording errors into superblock") Reviewed-by: Chao Yu <chao@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2023-03-23 22:37:54 +00:00
void f2fs_save_errors(struct f2fs_sb_info *sbi, unsigned char flag);
void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason,
bool irq_context);
void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error);
void f2fs_handle_error_async(struct f2fs_sb_info *sbi, unsigned char error);
int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover);
int f2fs_sync_fs(struct super_block *sb, int sync);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi);
/*
* hash.c
*/
2020-05-07 07:59:04 +00:00
void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname);
/*
* node.c
*/
struct node_info;
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid);
bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type);
f2fs: fix to avoid broken of dnode block list f2fs recovery flow is relying on dnode block link list, it means fsynced file recovery depends on previous dnode's persistence in the list, so during fsync() we should wait on all regular inode's dnode writebacked before issuing flush. By this way, we can avoid dnode block list being broken by out-of-order IO submission due to IO scheduler or driver. Sheng Yong helps to do the test with this patch: Target:/data (f2fs, -) 64MB / 32768KB / 4KB / 8 1 / PERSIST / Index Base: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 867.82 204.15 41440.03 41370.54 680.8 1025.94 1031.08 2 871.87 205.87 41370.3 40275.2 791.14 1065.84 1101.7 3 866.52 205.69 41795.67 40596.16 694.69 1037.16 1031.48 Avg 868.7366667 205.2366667 41535.33333 40747.3 722.21 1042.98 1054.753333 After: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 798.81 202.5 41143 40613.87 602.71 838.08 913.83 2 805.79 206.47 40297.2 41291.46 604.44 840.75 924.27 3 814.83 206.17 41209.57 40453.62 602.85 834.66 927.91 Avg 806.4766667 205.0466667 40883.25667 40786.31667 603.3333333 837.83 922.0033333 Patched/Original: 0.928332713 0.999074239 0.984300676 1.000957528 0.835398753 0.803303994 0.874141189 It looks like atomic write will suffer performance regression. I suspect that the criminal is that we forcing to wait all dnode being in storage cache before we issue PREFLUSH+FUA. BTW, will commit ("f2fs: don't need to wait for node writes for atomic write") cause the problem: we will lose data of last transaction after SPO, even if atomic write return no error: - atomic_open(); - write() P1, P2, P3; - atomic_commit(); - writeback data: P1, P2, P3; - writeback node: N1, N2, N3; <--- If N1, N2 is not writebacked, N3 with fsync_mark is writebacked, In SPOR, we won't find N3 since node chain is broken, turns out that losing last transaction. - preflush + fua; - power-cut If we don't wait dnode writeback for atomic_write: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 779.91 206.03 41621.5 40333.16 716.9 1038.21 1034.85 2 848.51 204.35 40082.44 39486.17 791.83 1119.96 1083.77 3 772.12 206.27 41335.25 41599.65 723.29 1055.07 971.92 Avg 800.18 205.55 41013.06333 40472.99333 744.0066667 1071.08 1030.18 Patched/Original: 0.92108464 1.001526693 0.987425886 0.993268102 1.030180511 1.026942031 0.976702294 SQLite's performance recovers. Jaegeuk: "Practically, I don't see db corruption becase of this. We can excuse to lose the last transaction." Finally, we decide to keep original implementation of atomic write interface sematics that we don't wait all dnode writeback before preflush+fua submission. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-02 15:03:19 +00:00
bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page);
void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi);
void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page);
void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid);
bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid);
bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino);
int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid,
struct node_info *ni, bool checkpoint_context);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs);
int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode);
int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from);
int f2fs_truncate_xattr_node(struct inode *inode);
f2fs: fix to avoid broken of dnode block list f2fs recovery flow is relying on dnode block link list, it means fsynced file recovery depends on previous dnode's persistence in the list, so during fsync() we should wait on all regular inode's dnode writebacked before issuing flush. By this way, we can avoid dnode block list being broken by out-of-order IO submission due to IO scheduler or driver. Sheng Yong helps to do the test with this patch: Target:/data (f2fs, -) 64MB / 32768KB / 4KB / 8 1 / PERSIST / Index Base: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 867.82 204.15 41440.03 41370.54 680.8 1025.94 1031.08 2 871.87 205.87 41370.3 40275.2 791.14 1065.84 1101.7 3 866.52 205.69 41795.67 40596.16 694.69 1037.16 1031.48 Avg 868.7366667 205.2366667 41535.33333 40747.3 722.21 1042.98 1054.753333 After: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 798.81 202.5 41143 40613.87 602.71 838.08 913.83 2 805.79 206.47 40297.2 41291.46 604.44 840.75 924.27 3 814.83 206.17 41209.57 40453.62 602.85 834.66 927.91 Avg 806.4766667 205.0466667 40883.25667 40786.31667 603.3333333 837.83 922.0033333 Patched/Original: 0.928332713 0.999074239 0.984300676 1.000957528 0.835398753 0.803303994 0.874141189 It looks like atomic write will suffer performance regression. I suspect that the criminal is that we forcing to wait all dnode being in storage cache before we issue PREFLUSH+FUA. BTW, will commit ("f2fs: don't need to wait for node writes for atomic write") cause the problem: we will lose data of last transaction after SPO, even if atomic write return no error: - atomic_open(); - write() P1, P2, P3; - atomic_commit(); - writeback data: P1, P2, P3; - writeback node: N1, N2, N3; <--- If N1, N2 is not writebacked, N3 with fsync_mark is writebacked, In SPOR, we won't find N3 since node chain is broken, turns out that losing last transaction. - preflush + fua; - power-cut If we don't wait dnode writeback for atomic_write: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 779.91 206.03 41621.5 40333.16 716.9 1038.21 1034.85 2 848.51 204.35 40082.44 39486.17 791.83 1119.96 1083.77 3 772.12 206.27 41335.25 41599.65 723.29 1055.07 971.92 Avg 800.18 205.55 41013.06333 40472.99333 744.0066667 1071.08 1030.18 Patched/Original: 0.92108464 1.001526693 0.987425886 0.993268102 1.030180511 1.026942031 0.976702294 SQLite's performance recovers. Jaegeuk: "Practically, I don't see db corruption becase of this. We can excuse to lose the last transaction." Finally, we decide to keep original implementation of atomic write interface sematics that we don't wait all dnode writeback before preflush+fua submission. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-02 15:03:19 +00:00
int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi,
unsigned int seq_id);
bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
int f2fs_remove_inode_page(struct inode *inode);
struct page *f2fs_new_inode_page(struct inode *inode);
struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs);
void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid);
struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid);
struct page *f2fs_get_node_page_ra(struct page *parent, int start);
int f2fs_move_node_page(struct page *node_page, int gc_type);
void f2fs_flush_inline_data(struct f2fs_sb_info *sbi);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode,
f2fs: fix to avoid broken of dnode block list f2fs recovery flow is relying on dnode block link list, it means fsynced file recovery depends on previous dnode's persistence in the list, so during fsync() we should wait on all regular inode's dnode writebacked before issuing flush. By this way, we can avoid dnode block list being broken by out-of-order IO submission due to IO scheduler or driver. Sheng Yong helps to do the test with this patch: Target:/data (f2fs, -) 64MB / 32768KB / 4KB / 8 1 / PERSIST / Index Base: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 867.82 204.15 41440.03 41370.54 680.8 1025.94 1031.08 2 871.87 205.87 41370.3 40275.2 791.14 1065.84 1101.7 3 866.52 205.69 41795.67 40596.16 694.69 1037.16 1031.48 Avg 868.7366667 205.2366667 41535.33333 40747.3 722.21 1042.98 1054.753333 After: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 798.81 202.5 41143 40613.87 602.71 838.08 913.83 2 805.79 206.47 40297.2 41291.46 604.44 840.75 924.27 3 814.83 206.17 41209.57 40453.62 602.85 834.66 927.91 Avg 806.4766667 205.0466667 40883.25667 40786.31667 603.3333333 837.83 922.0033333 Patched/Original: 0.928332713 0.999074239 0.984300676 1.000957528 0.835398753 0.803303994 0.874141189 It looks like atomic write will suffer performance regression. I suspect that the criminal is that we forcing to wait all dnode being in storage cache before we issue PREFLUSH+FUA. BTW, will commit ("f2fs: don't need to wait for node writes for atomic write") cause the problem: we will lose data of last transaction after SPO, even if atomic write return no error: - atomic_open(); - write() P1, P2, P3; - atomic_commit(); - writeback data: P1, P2, P3; - writeback node: N1, N2, N3; <--- If N1, N2 is not writebacked, N3 with fsync_mark is writebacked, In SPOR, we won't find N3 since node chain is broken, turns out that losing last transaction. - preflush + fua; - power-cut If we don't wait dnode writeback for atomic_write: SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS) 1 779.91 206.03 41621.5 40333.16 716.9 1038.21 1034.85 2 848.51 204.35 40082.44 39486.17 791.83 1119.96 1083.77 3 772.12 206.27 41335.25 41599.65 723.29 1055.07 971.92 Avg 800.18 205.55 41013.06333 40472.99333 744.0066667 1071.08 1030.18 Patched/Original: 0.92108464 1.001526693 0.987425886 0.993268102 1.030180511 1.026942031 0.976702294 SQLite's performance recovers. Jaegeuk: "Practically, I don't see db corruption becase of this. We can excuse to lose the last transaction." Finally, we decide to keep original implementation of atomic write interface sematics that we don't wait all dnode writeback before preflush+fua submission. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-02 15:03:19 +00:00
struct writeback_control *wbc, bool atomic,
unsigned int *seq_id);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
int f2fs_sync_node_pages(struct f2fs_sb_info *sbi,
struct writeback_control *wbc,
bool do_balance, enum iostat_type io_type);
f2fs: fix to propagate return value of scan_nat_page() As Anatoly Trosinenko reported in bugzilla: How to reproduce: 1. Compile the 73fcb1a370c76 version of the kernel using the config attached 2. Unpack and mount the attached filesystem image as F2FS 3. The kernel will BUG() on mount (BUGs are explicitly enabled in config) [ 2.233612] F2FS-fs (sda): Found nat_bits in checkpoint [ 2.248422] ------------[ cut here ]------------ [ 2.248857] kernel BUG at fs/f2fs/node.c:1967! [ 2.249760] invalid opcode: 0000 [#1] SMP NOPTI [ 2.250219] Modules linked in: [ 2.251848] CPU: 0 PID: 944 Comm: mount Not tainted 4.17.0-rc5+ #1 [ 2.252331] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014 [ 2.253305] RIP: 0010:build_free_nids+0x337/0x3f0 [ 2.253672] RSP: 0018:ffffae7fc0857c50 EFLAGS: 00000246 [ 2.254080] RAX: 00000000ffffffff RBX: 0000000000000123 RCX: 0000000000000001 [ 2.254638] RDX: ffff9aa7063d5c00 RSI: 0000000000000122 RDI: ffff9aa705852e00 [ 2.255190] RBP: ffff9aa705852e00 R08: 0000000000000001 R09: ffff9aa7059090c0 [ 2.255719] R10: 0000000000000000 R11: 0000000000000000 R12: ffff9aa705852e00 [ 2.256242] R13: ffff9aa7063ad000 R14: ffff9aa705919000 R15: 0000000000000123 [ 2.256809] FS: 00000000023078c0(0000) GS:ffff9aa707800000(0000) knlGS:0000000000000000 [ 2.258654] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 2.259153] CR2: 00000000005511ae CR3: 0000000005872000 CR4: 00000000000006f0 [ 2.259801] Call Trace: [ 2.260583] build_node_manager+0x5cd/0x600 [ 2.260963] f2fs_fill_super+0x66a/0x17c0 [ 2.261300] ? f2fs_commit_super+0xe0/0xe0 [ 2.261622] mount_bdev+0x16e/0x1a0 [ 2.261899] mount_fs+0x30/0x150 [ 2.262398] vfs_kern_mount.part.28+0x4f/0xf0 [ 2.262743] do_mount+0x5d0/0xc60 [ 2.263010] ? _copy_from_user+0x37/0x60 [ 2.263313] ? memdup_user+0x39/0x60 [ 2.263692] ksys_mount+0x7b/0xd0 [ 2.263960] __x64_sys_mount+0x1c/0x20 [ 2.264268] do_syscall_64+0x43/0xf0 [ 2.264560] entry_SYSCALL_64_after_hwframe+0x44/0xa9 [ 2.265095] RIP: 0033:0x48d31a [ 2.265502] RSP: 002b:00007ffc6fe60a08 EFLAGS: 00000246 ORIG_RAX: 00000000000000a5 [ 2.266089] RAX: ffffffffffffffda RBX: 0000000000008000 RCX: 000000000048d31a [ 2.266607] RDX: 00007ffc6fe62fa5 RSI: 00007ffc6fe62f9d RDI: 00007ffc6fe62f94 [ 2.267130] RBP: 00000000023078a0 R08: 0000000000000000 R09: 0000000000000000 [ 2.267670] R10: 0000000000008000 R11: 0000000000000246 R12: 0000000000000000 [ 2.268192] R13: 0000000000000000 R14: 00007ffc6fe60c78 R15: 0000000000000000 [ 2.268767] Code: e8 5f c3 ff ff 83 c3 01 41 83 c7 01 81 fb c7 01 00 00 74 48 44 39 7d 04 76 42 48 63 c3 48 8d 04 c0 41 8b 44 06 05 83 f8 ff 75 c1 <0f> 0b 49 8b 45 50 48 8d b8 b0 00 00 00 e8 37 59 69 00 b9 01 00 [ 2.270434] RIP: build_free_nids+0x337/0x3f0 RSP: ffffae7fc0857c50 [ 2.271426] ---[ end trace ab20c06cd3c8fde4 ]--- During loading NAT entries, we will do sanity check, once the entry info is corrupted, it will cause BUG_ON directly to protect user data from being overwrited. In this case, it will be better to just return failure on mount() instead of panic, so that user can get hint from kmsg and try fsck for recovery immediately rather than after an abnormal reboot. https://bugzilla.kernel.org/show_bug.cgi?id=199769 Reported-by: Anatoly Trosinenko <anatoly.trosinenko@gmail.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-06-15 06:45:57 +00:00
int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid);
void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid);
void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid);
int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink);
int f2fs_recover_inline_xattr(struct inode *inode, struct page *page);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
int f2fs_recover_xattr_data(struct inode *inode, struct page *page);
int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page);
int f2fs_restore_node_summary(struct f2fs_sb_info *sbi,
unsigned int segno, struct f2fs_summary_block *sum);
void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi);
int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
int f2fs_build_node_manager(struct f2fs_sb_info *sbi);
void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi);
int __init f2fs_create_node_manager_caches(void);
void f2fs_destroy_node_manager_caches(void);
/*
* segment.c
*/
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
bool f2fs_need_SSR(struct f2fs_sb_info *sbi);
int f2fs_commit_atomic_write(struct inode *inode);
void f2fs_abort_atomic_write(struct inode *inode, bool clean);
void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need);
void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg);
int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi);
int f2fs_flush_device_cache(struct f2fs_sb_info *sbi);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free);
void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr);
bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr);
int f2fs_start_discard_thread(struct f2fs_sb_info *sbi);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi);
void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi);
bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi,
struct cp_control *cpc);
void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi);
block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi);
int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi);
int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra);
bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno);
int f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi);
int f2fs_reinit_atgc_curseg(struct f2fs_sb_info *sbi);
f2fs: support age threshold based garbage collection There are several issues in current background GC algorithm: - valid blocks is one of key factors during cost overhead calculation, so if segment has less valid block, however even its age is young or it locates hot segment, CB algorithm will still choose the segment as victim, it's not appropriate. - GCed data/node will go to existing logs, no matter in-there datas' update frequency is the same or not, it may mix hot and cold data again. - GC alloctor mainly use LFS type segment, it will cost free segment more quickly. This patch introduces a new algorithm named age threshold based garbage collection to solve above issues, there are three steps mainly: 1. select a source victim: - set an age threshold, and select candidates beased threshold: e.g. 0 means youngest, 100 means oldest, if we set age threshold to 80 then select dirty segments which has age in range of [80, 100] as candiddates; - set candidate_ratio threshold, and select candidates based the ratio, so that we can shrink candidates to those oldest segments; - select target segment with fewest valid blocks in order to migrate blocks with minimum cost; 2. select a target victim: - select candidates beased age threshold; - set candidate_radius threshold, search candidates whose age is around source victims, searching radius should less than the radius threshold. - select target segment with most valid blocks in order to avoid migrating current target segment. 3. merge valid blocks from source victim into target victim with SSR alloctor. Test steps: - create 160 dirty segments: * half of them have 128 valid blocks per segment * left of them have 384 valid blocks per segment - run background GC Benefit: GC count and block movement count both decrease obviously: - Before: - Valid: 86 - Dirty: 1 - Prefree: 11 - Free: 6001 (6001) GC calls: 162 (BG: 220) - data segments : 160 (160) - node segments : 2 (2) Try to move 41454 blocks (BG: 41454) - data blocks : 40960 (40960) - node blocks : 494 (494) IPU: 0 blocks SSR: 0 blocks in 0 segments LFS: 41364 blocks in 81 segments - After: - Valid: 87 - Dirty: 0 - Prefree: 4 - Free: 6008 (6008) GC calls: 75 (BG: 76) - data segments : 74 (74) - node segments : 1 (1) Try to move 12813 blocks (BG: 12813) - data blocks : 12544 (12544) - node blocks : 269 (269) IPU: 0 blocks SSR: 12032 blocks in 77 segments LFS: 855 blocks in 2 segments Signed-off-by: Chao Yu <yuchao0@huawei.com> [Jaegeuk Kim: fix a bug along with pinfile in-mem segment & clean up] Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2020-08-04 13:14:49 +00:00
void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi);
void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi);
int f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type,
f2fs: ioctl for removing a range from F2FS This ioctl shrinks a given length (aligned to sections) from end of the main area. Any cursegs and valid blocks will be moved out before invalidating the range. This feature can be used for adjusting partition sizes online. History of the patch: Sahitya Tummala: - Add this ioctl for f2fs_compat_ioctl() as well. - Fix debugfs status to reflect the online resize changes. - Fix potential race between online resize path and allocate new data block path or gc path. Others: - Rename some identifiers. - Add some error handling branches. - Clear sbi->next_victim_seg[BG_GC/FG_GC] in shrinking range. - Implement this interface as ext4's, and change the parameter from shrunk bytes to new block count of F2FS. - During resizing, force to empty sit_journal and forbid adding new entries to it, in order to avoid invalid segno in journal after resize. - Reduce sbi->user_block_count before resize starts. - Commit the updated superblock first, and then update in-memory metadata only when the former succeeds. - Target block count must align to sections. - Write checkpoint before and after committing the new superblock, w/o CP_FSCK_FLAG respectively, so that the FS can be fixed by fsck even if resize fails after the new superblock is committed. - In free_segment_range(), reduce granularity of gc_mutex. - Add protection on curseg migration. - Add freeze_bdev() and thaw_bdev() for resize fs. - Remove CUR_MAIN_SECS and use MAIN_SECS directly for allocation. - Recover super_block and FS metadata when resize fails. - No need to clear CP_FSCK_FLAG in update_ckpt_flags(). - Clean up the sb and fs metadata update functions for resize_fs. Geert Uytterhoeven: - Use div_u64*() for 64-bit divisions Arnd Bergmann: - Not all architectures support get_user() with a 64-bit argument: ERROR: "__get_user_bad" [fs/f2fs/f2fs.ko] undefined! Use copy_from_user() here, this will always work. Signed-off-by: Qiuyang Sun <sunqiuyang@huawei.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Sahitya Tummala <stummala@codeaurora.org> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-06-05 03:33:25 +00:00
unsigned int start, unsigned int end);
int f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force);
int f2fs_allocate_pinning_section(struct f2fs_sb_info *sbi);
int f2fs_allocate_new_segments(struct f2fs_sb_info *sbi);
int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi,
struct cp_control *cpc);
struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno);
void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src,
block_t blk_addr);
void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page,
enum iostat_type io_type);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio);
void f2fs_outplace_write_data(struct dnode_of_data *dn,
struct f2fs_io_info *fio);
int f2fs_inplace_write_data(struct f2fs_io_info *fio);
void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
block_t old_blkaddr, block_t new_blkaddr,
bool recover_curseg, bool recover_newaddr,
bool from_gc);
void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn,
block_t old_addr, block_t new_addr,
unsigned char version, bool recover_curseg,
bool recover_newaddr);
int f2fs_get_segment_temp(int seg_type);
int f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page,
block_t old_blkaddr, block_t *new_blkaddr,
struct f2fs_summary *sum, int type,
struct f2fs_io_info *fio);
void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino,
block_t blkaddr, unsigned int blkcnt);
void f2fs_wait_on_page_writeback(struct page *page,
enum page_type type, bool ordered, bool locked);
void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr);
void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr,
block_t len);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type,
unsigned int val, int alloc);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi);
int f2fs_check_write_pointer(struct f2fs_sb_info *sbi);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
int f2fs_build_segment_manager(struct f2fs_sb_info *sbi);
void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi);
int __init f2fs_create_segment_manager_caches(void);
void f2fs_destroy_segment_manager_caches(void);
int f2fs_rw_hint_to_seg_type(struct f2fs_sb_info *sbi, enum rw_hint hint);
enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi,
enum page_type type, enum temp_type temp);
f2fs: support zone capacity less than zone size NVMe Zoned Namespace devices can have zone-capacity less than zone-size. Zone-capacity indicates the maximum number of sectors that are usable in a zone beginning from the first sector of the zone. This makes the sectors sectors after the zone-capacity till zone-size to be unusable. This patch set tracks zone-size and zone-capacity in zoned devices and calculate the usable blocks per segment and usable segments per section. If zone-capacity is less than zone-size mark only those segments which start before zone-capacity as free segments. All segments at and beyond zone-capacity are treated as permanently used segments. In cases where zone-capacity does not align with segment size the last segment will start before zone-capacity and end beyond the zone-capacity of the zone. For such spanning segments only sectors within the zone-capacity are used. During writes and GC manage the usable segments in a section and usable blocks per segment. Segments which are beyond zone-capacity are never allocated, and do not need to be garbage collected, only the segments which are before zone-capacity needs to garbage collected. For spanning segments based on the number of usable blocks in that segment, write to blocks only up to zone-capacity. Zone-capacity is device specific and cannot be configured by the user. Since NVMe ZNS device zones are sequentially write only, a block device with conventional zones or any normal block device is needed along with the ZNS device for the metadata operations of F2fs. A typical nvme-cli output of a zoned device shows zone start and capacity and write pointer as below: SLBA: 0x0 WP: 0x0 Cap: 0x18800 State: EMPTY Type: SEQWRITE_REQ SLBA: 0x20000 WP: 0x20000 Cap: 0x18800 State: EMPTY Type: SEQWRITE_REQ SLBA: 0x40000 WP: 0x40000 Cap: 0x18800 State: EMPTY Type: SEQWRITE_REQ Here zone size is 64MB, capacity is 49MB, WP is at zone start as the zones are in EMPTY state. For each zone, only zone start + 49MB is usable area, any lba/sector after 49MB cannot be read or written to, the drive will fail any attempts to read/write. So, the second zone starts at 64MB and is usable till 113MB (64 + 49) and the range between 113 and 128MB is again unusable. The next zone starts at 128MB, and so on. Signed-off-by: Aravind Ramesh <aravind.ramesh@wdc.com> Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com> Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2020-07-16 12:56:56 +00:00
unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi,
unsigned int segno);
unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi,
unsigned int segno);
#define DEF_FRAGMENT_SIZE 4
#define MIN_FRAGMENT_SIZE 1
#define MAX_FRAGMENT_SIZE 512
static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi)
{
return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG ||
F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK;
}
/*
* checkpoint.c
*/
void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io,
unsigned char reason);
void f2fs_flush_ckpt_thread(struct f2fs_sb_info *sbi);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
f2fs: handle errors of f2fs_get_meta_page_nofail First problem is we hit BUG_ON() in f2fs_get_sum_page given EIO on f2fs_get_meta_page_nofail(). Quick fix was not to give any error with infinite loop, but syzbot caught a case where it goes to that loop from fuzzed image. In turned out we abused f2fs_get_meta_page_nofail() like in the below call stack. - f2fs_fill_super - f2fs_build_segment_manager - build_sit_entries - get_current_sit_page INFO: task syz-executor178:6870 can't die for more than 143 seconds. task:syz-executor178 state:R stack:26960 pid: 6870 ppid: 6869 flags:0x00004006 Call Trace: Showing all locks held in the system: 1 lock held by khungtaskd/1179: #0: ffffffff8a554da0 (rcu_read_lock){....}-{1:2}, at: debug_show_all_locks+0x53/0x260 kernel/locking/lockdep.c:6242 1 lock held by systemd-journal/3920: 1 lock held by in:imklog/6769: #0: ffff88809eebc130 (&f->f_pos_lock){+.+.}-{3:3}, at: __fdget_pos+0xe9/0x100 fs/file.c:930 1 lock held by syz-executor178/6870: #0: ffff8880925120e0 (&type->s_umount_key#47/1){+.+.}-{3:3}, at: alloc_super+0x201/0xaf0 fs/super.c:229 Actually, we didn't have to use _nofail in this case, since we could return error to mount(2) already with the error handler. As a result, this patch tries to 1) remove _nofail callers as much as possible, 2) deal with error case in last remaining caller, f2fs_get_sum_page(). Reported-by: syzbot+ee250ac8137be41d7b13@syzkaller.appspotmail.com Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2020-10-02 21:17:35 +00:00
struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index);
bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
block_t blkaddr, int type);
f2fs: introduce FAULT_BLKADDR_CONSISTENCE We will encounter below inconsistent status when FAULT_BLKADDR type fault injection is on. Info: checkpoint state = d6 : nat_bits crc fsck compacted_summary orphan_inodes sudden-power-off [ASSERT] (fsck_chk_inode_blk:1254) --> ino: 0x1c100 has i_blocks: 000000c0, but has 191 blocks [FIX] (fsck_chk_inode_blk:1260) --> [0x1c100] i_blocks=0x000000c0 -> 0xbf [FIX] (fsck_chk_inode_blk:1269) --> [0x1c100] i_compr_blocks=0x00000026 -> 0x27 [ASSERT] (fsck_chk_inode_blk:1254) --> ino: 0x1cadb has i_blocks: 0000002f, but has 46 blocks [FIX] (fsck_chk_inode_blk:1260) --> [0x1cadb] i_blocks=0x0000002f -> 0x2e [FIX] (fsck_chk_inode_blk:1269) --> [0x1cadb] i_compr_blocks=0x00000011 -> 0x12 [ASSERT] (fsck_chk_inode_blk:1254) --> ino: 0x1c62c has i_blocks: 00000002, but has 1 blocks [FIX] (fsck_chk_inode_blk:1260) --> [0x1c62c] i_blocks=0x00000002 -> 0x1 After we inject fault into f2fs_is_valid_blkaddr() during truncation, a) it missed to increase @nr_free or @valid_blocks b) it can cause in blkaddr leak in truncated dnode Which may cause inconsistent status. This patch separates FAULT_BLKADDR_CONSISTENCE from FAULT_BLKADDR, and rename FAULT_BLKADDR to FAULT_BLKADDR_VALIDITY so that we can: a) use FAULT_BLKADDR_CONSISTENCE in f2fs_truncate_data_blocks_range() to simulate inconsistent issue independently, then it can verify fsck repair flow. b) FAULT_BLKADDR_VALIDITY fault will not cause any inconsistent status, we can just use it to check error path handling in kernel side. Reviewed-by: Daeho Jeong <daehojeong@google.com> Signed-off-by: Chao Yu <chao@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2024-01-12 19:41:32 +00:00
bool f2fs_is_valid_blkaddr_raw(struct f2fs_sb_info *sbi,
block_t blkaddr, int type);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages,
int type, bool sync);
void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index,
unsigned int ra_blocks);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type,
long nr_to_write, enum iostat_type io_type);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all);
bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode);
void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
unsigned int devidx, int type);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
unsigned int devidx, int type);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi);
void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi);
void f2fs_add_orphan_inode(struct inode *inode);
void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino);
int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi);
int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi);
void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
void f2fs_remove_dirty_inode(struct inode *inode);
int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type,
bool from_cp);
void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type);
u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc);
void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi);
int __init f2fs_create_checkpoint_caches(void);
void f2fs_destroy_checkpoint_caches(void);
f2fs: introduce checkpoint_merge mount option We've added a new mount options, "checkpoint_merge" and "nocheckpoint_merge", which creates a kernel daemon and makes it to merge concurrent checkpoint requests as much as possible to eliminate redundant checkpoint issues. Plus, we can eliminate the sluggish issue caused by slow checkpoint operation when the checkpoint is done in a process context in a cgroup having low i/o budget and cpu shares. To make this do better, we set the default i/o priority of the kernel daemon to "3", to give one higher priority than other kernel threads. The below verification result explains this. The basic idea has come from https://opensource.samsung.com. [Verification] Android Pixel Device(ARM64, 7GB RAM, 256GB UFS) Create two I/O cgroups (fg w/ weight 100, bg w/ wight 20) Set "strict_guarantees" to "1" in BFQ tunables In "fg" cgroup, - thread A => trigger 1000 checkpoint operations "for i in `seq 1 1000`; do touch test_dir1/file; fsync test_dir1; done" - thread B => gererating async. I/O "fio --rw=write --numjobs=1 --bs=128k --runtime=3600 --time_based=1 --filename=test_img --name=test" In "bg" cgroup, - thread C => trigger repeated checkpoint operations "echo $$ > /dev/blkio/bg/tasks; while true; do touch test_dir2/file; fsync test_dir2; done" We've measured thread A's execution time. [ w/o patch ] Elapsed Time: Avg. 68 seconds [ w/ patch ] Elapsed Time: Avg. 48 seconds Reported-by: kernel test robot <lkp@intel.com> Reported-by: Dan Carpenter <dan.carpenter@oracle.com> [Jaegeuk Kim: fix the return value in f2fs_start_ckpt_thread, reported by Dan] Signed-off-by: Daeho Jeong <daehojeong@google.com> Signed-off-by: Sungjong Seo <sj1557.seo@samsung.com> Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2021-01-19 00:00:42 +00:00
int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi);
int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi);
void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi);
void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi);
/*
* data.c
*/
int __init f2fs_init_bioset(void);
void f2fs_destroy_bioset(void);
bool f2fs_is_cp_guaranteed(struct page *page);
int f2fs_init_bio_entry_cache(void);
void f2fs_destroy_bio_entry_cache(void);
void f2fs_submit_read_bio(struct f2fs_sb_info *sbi, struct bio *bio,
enum page_type type);
int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi);
void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type);
void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi,
struct inode *inode, struct page *page,
nid_t ino, enum page_type type);
void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi,
struct bio **bio, struct page *page);
void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi);
int f2fs_submit_page_bio(struct f2fs_io_info *fio);
f2fs: add bio cache for IPU SQLite in Wal mode may trigger sequential IPU write in db-wal file, after commit d1b3e72d5490 ("f2fs: submit bio of in-place-update pages"), we lost the chance of merging page in inner managed bio cache, result in submitting more small-sized IO. So let's add temporary bio in writepages() to cache mergeable write IO as much as possible. Test case: 1. xfs_io -f /mnt/f2fs/file -c "pwrite 0 65536" -c "fsync" 2. xfs_io -f /mnt/f2fs/file -c "pwrite 0 65536" -c "fsync" Before: f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65544, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65552, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65560, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65568, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65576, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65584, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65592, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65600, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65608, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65616, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65624, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65632, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65640, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65648, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65656, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65664, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), NODE, sector = 57352, size = 4096 After: f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65544, size = 65536 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), NODE, sector = 57368, size = 4096 Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-02-19 08:15:29 +00:00
int f2fs_merge_page_bio(struct f2fs_io_info *fio);
f2fs: fix to let caller retry allocating block address Configure io_bits with 2 and enable LFS mode, generic/013 reports below dmesg: BUG: unable to handle kernel NULL pointer dereference at 00000104 *pdpt = 0000000029b7b001 *pde = 0000000000000000 Oops: 0002 [#1] PREEMPT SMP Modules linked in: crc32_generic zram f2fs(O) rfcomm bnep bluetooth ecdh_generic snd_intel8x0 snd_ac97_codec ac97_bus snd_pcm snd_seq_midi snd_seq_midi_event snd_rawmidi snd_seq pcbc joydev snd_seq_device aesni_intel snd_timer aes_i586 snd crypto_simd cryptd soundcore i2c_piix4 serio_raw mac_hid video parport_pc ppdev lp parport hid_generic psmouse usbhid hid e1000 CPU: 0 PID: 11161 Comm: fsstress Tainted: G O 4.17.0-rc2 #38 Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006 EIP: f2fs_submit_page_write+0x28d/0x550 [f2fs] EFLAGS: 00010206 CPU: 0 EAX: e863dcd8 EBX: 00000000 ECX: 00000100 EDX: 00000200 ESI: e863dcf4 EDI: f6f82768 EBP: e863dbb0 ESP: e863db74 DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068 CR0: 80050033 CR2: 00000104 CR3: 29a62020 CR4: 000406f0 Call Trace: do_write_page+0x6f/0xc0 [f2fs] write_data_page+0x4a/0xd0 [f2fs] do_write_data_page+0x327/0x630 [f2fs] __write_data_page+0x34b/0x820 [f2fs] __f2fs_write_data_pages+0x42d/0x8c0 [f2fs] f2fs_write_data_pages+0x27/0x30 [f2fs] do_writepages+0x1a/0x70 __filemap_fdatawrite_range+0x94/0xd0 filemap_write_and_wait_range+0x3d/0xa0 __generic_file_write_iter+0x11a/0x1f0 f2fs_file_write_iter+0xdd/0x3b0 [f2fs] __vfs_write+0xd2/0x150 vfs_write+0x9b/0x190 ksys_write+0x45/0x90 sys_write+0x16/0x20 do_fast_syscall_32+0xaa/0x22c entry_SYSENTER_32+0x4c/0x7b EIP: 0xb7fc8c51 EFLAGS: 00000246 CPU: 0 EAX: ffffffda EBX: 00000003 ECX: 09cde000 EDX: 00001000 ESI: 00000003 EDI: 00001000 EBP: 00000000 ESP: bfbded38 DS: 007b ES: 007b FS: 0000 GS: 0033 SS: 007b Code: e8 f9 77 34 c9 8b 45 e0 8b 80 b8 00 00 00 39 45 d8 0f 84 bb 02 00 00 8b 45 e0 8b 80 b8 00 00 00 8d 50 d8 8b 08 89 55 f0 8b 50 04 <89> 51 04 89 0a c7 00 00 01 00 00 c7 40 04 00 02 00 00 8b 45 dc EIP: f2fs_submit_page_write+0x28d/0x550 [f2fs] SS:ESP: 0068:e863db74 CR2: 0000000000000104 ---[ end trace 4cac79c0d1305ee6 ]--- allocate_data_block will submit all sequential pending IOs sorted by a FIFO list, If we failed to submit other user's IO due to unaligned write, we will retry to allocate new block address for current IO, then it will initialize fio.list again, if fio was in the list before, it can break FIFO list, result in above panic. Thread A Thread B - do_write_page - allocate_data_block - list_add_tail : fioA cached in FIFO list. - do_write_page - allocate_data_block - list_add_tail : fioB cached in FIFO list. - f2fs_submit_page_write : fail to submit IO - allocate_data_block - INIT_LIST_HEAD - f2fs_submit_page_write - list_del <-- NULL pointer dereference This patch adds fio.retry parameter to indicate failure status for each IO, and avoid bailing out if there is still pending IO in FIFO list for fixing. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-28 15:47:18 +00:00
void f2fs_submit_page_write(struct f2fs_io_info *fio);
struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi,
block_t blk_addr, sector_t *sector);
int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr);
void f2fs_set_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr);
void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count);
int f2fs_reserve_new_block(struct dnode_of_data *dn);
int f2fs_get_block_locked(struct dnode_of_data *dn, pgoff_t index);
int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index,
f2fs: optimize iteration over sparse directories Wei Chen reports a kernel bug as blew: INFO: task syz-executor.0:29056 blocked for more than 143 seconds. Not tainted 5.15.0-rc5 #1 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. task:syz-executor.0 state:D stack:14632 pid:29056 ppid: 6574 flags:0x00000004 Call Trace: __schedule+0x4a1/0x1720 schedule+0x36/0xe0 rwsem_down_write_slowpath+0x322/0x7a0 fscrypt_ioctl_set_policy+0x11f/0x2a0 __f2fs_ioctl+0x1a9f/0x5780 f2fs_ioctl+0x89/0x3a0 __x64_sys_ioctl+0xe8/0x140 do_syscall_64+0x34/0xb0 entry_SYSCALL_64_after_hwframe+0x44/0xae Eric did some investigation on this issue, quoted from reply of Eric: "Well, the quality of this bug report has a lot to be desired (not on upstream kernel, reproducer is full of totally irrelevant stuff, not sent to the mailing list of the filesystem whose disk image is being fuzzed, etc.). But what is going on is that f2fs_empty_dir() doesn't consider the case of a directory with an extremely large i_size on a malicious disk image. Specifically, the reproducer mounts an f2fs image with a directory that has an i_size of 14814520042850357248, then calls FS_IOC_SET_ENCRYPTION_POLICY on it. That results in a call to f2fs_empty_dir() to check whether the directory is empty. f2fs_empty_dir() then iterates through all 3616826182336513 blocks the directory allegedly contains to check whether any contain anything. i_rwsem is held during this, so anything else that tries to take it will hang." In order to solve this issue, let's use f2fs_get_next_page_offset() to speed up iteration by skipping holes for all below functions: - f2fs_empty_dir - f2fs_readdir - find_in_level The way why we can speed up iteration was described in 'commit 3cf4574705b4 ("f2fs: introduce get_next_page_offset to speed up SEEK_DATA")'. Meanwhile, in f2fs_empty_dir(), let's use f2fs_find_data_page() instead f2fs_get_lock_data_page(), due to i_rwsem was held in caller of f2fs_empty_dir(), there shouldn't be any races, so it's fine to not lock dentry page during lookuping dirents in the page. Link: https://lore.kernel.org/lkml/536944df-a0ae-1dd8-148f-510b476e1347@kernel.org/T/ Reported-by: Wei Chen <harperchen1110@gmail.com> Cc: Eric Biggers <ebiggers@google.com> Signed-off-by: Chao Yu <chao@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2022-11-08 14:33:21 +00:00
blk_opf_t op_flags, bool for_write, pgoff_t *next_pgofs);
struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index,
pgoff_t *next_pgofs);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index,
bool for_write);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
struct page *f2fs_get_new_data_page(struct inode *inode,
struct page *ipage, pgoff_t index, bool new_i_size);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
int f2fs_do_write_data_page(struct f2fs_io_info *fio);
int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, int flag);
int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
u64 start, u64 len);
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>
2019-11-01 10:07:14 +00:00
int f2fs_encrypt_one_page(struct f2fs_io_info *fio);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio);
bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio);
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>
2019-11-01 10:07:14 +00:00
int f2fs_write_single_data_page(struct page *page, int *submitted,
struct bio **bio, sector_t *last_block,
struct writeback_control *wbc,
enum iostat_type io_type,
int compr_blocks, bool allow_balance);
void f2fs_write_failed(struct inode *inode, loff_t to);
void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length);
bool f2fs_release_folio(struct folio *folio, gfp_t wait);
bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len);
void f2fs_clear_page_cache_dirty_tag(struct page *page);
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>
2019-11-01 10:07:14 +00:00
int f2fs_init_post_read_processing(void);
void f2fs_destroy_post_read_processing(void);
int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi);
void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi);
extern const struct iomap_ops f2fs_iomap_ops;
/*
* gc.c
*/
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
int f2fs_start_gc_thread(struct f2fs_sb_info *sbi);
void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi);
block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode);
int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
void f2fs_build_gc_manager(struct f2fs_sb_info *sbi);
int f2fs_gc_range(struct f2fs_sb_info *sbi,
unsigned int start_seg, unsigned int end_seg,
bool dry_run, unsigned int dry_run_sections);
f2fs: fix to avoid NULL pointer dereference f2fs_write_end_io() butt3rflyh4ck reports a bug as below: When a thread always calls F2FS_IOC_RESIZE_FS to resize fs, if resize fs is failed, f2fs kernel thread would invoke callback function to update f2fs io info, it would call f2fs_write_end_io and may trigger null-ptr-deref in NODE_MAPPING. general protection fault, probably for non-canonical address KASAN: null-ptr-deref in range [0x0000000000000030-0x0000000000000037] RIP: 0010:NODE_MAPPING fs/f2fs/f2fs.h:1972 [inline] RIP: 0010:f2fs_write_end_io+0x727/0x1050 fs/f2fs/data.c:370 <TASK> bio_endio+0x5af/0x6c0 block/bio.c:1608 req_bio_endio block/blk-mq.c:761 [inline] blk_update_request+0x5cc/0x1690 block/blk-mq.c:906 blk_mq_end_request+0x59/0x4c0 block/blk-mq.c:1023 lo_complete_rq+0x1c6/0x280 drivers/block/loop.c:370 blk_complete_reqs+0xad/0xe0 block/blk-mq.c:1101 __do_softirq+0x1d4/0x8ef kernel/softirq.c:571 run_ksoftirqd kernel/softirq.c:939 [inline] run_ksoftirqd+0x31/0x60 kernel/softirq.c:931 smpboot_thread_fn+0x659/0x9e0 kernel/smpboot.c:164 kthread+0x33e/0x440 kernel/kthread.c:379 ret_from_fork+0x1f/0x30 arch/x86/entry/entry_64.S:308 The root cause is below race case can cause leaving dirty metadata in f2fs after filesystem is remount as ro: Thread A Thread B - f2fs_ioc_resize_fs - f2fs_readonly --- return false - f2fs_resize_fs - f2fs_remount - write_checkpoint - set f2fs as ro - free_segment_range - update meta_inode's data Then, if f2fs_put_super() fails to write_checkpoint due to readonly status, and meta_inode's dirty data will be writebacked after node_inode is put, finally, f2fs_write_end_io will access NULL pointer on sbi->node_inode. Thread A IRQ context - f2fs_put_super - write_checkpoint fails - iput(node_inode) - node_inode = NULL - iput(meta_inode) - write_inode_now - f2fs_write_meta_page - f2fs_write_end_io - NODE_MAPPING(sbi) : access NULL pointer on node_inode Fixes: b4b10061ef98 ("f2fs: refactor resize_fs to avoid meta updates in progress") Reported-by: butt3rflyh4ck <butterflyhuangxx@gmail.com> Closes: https://lore.kernel.org/r/1684480657-2375-1-git-send-email-yangtiezhu@loongson.cn Tested-by: butt3rflyh4ck <butterflyhuangxx@gmail.com> Signed-off-by: Chao Yu <chao@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2023-05-23 06:17:25 +00:00
int f2fs_resize_fs(struct file *filp, __u64 block_count);
f2fs: support age threshold based garbage collection There are several issues in current background GC algorithm: - valid blocks is one of key factors during cost overhead calculation, so if segment has less valid block, however even its age is young or it locates hot segment, CB algorithm will still choose the segment as victim, it's not appropriate. - GCed data/node will go to existing logs, no matter in-there datas' update frequency is the same or not, it may mix hot and cold data again. - GC alloctor mainly use LFS type segment, it will cost free segment more quickly. This patch introduces a new algorithm named age threshold based garbage collection to solve above issues, there are three steps mainly: 1. select a source victim: - set an age threshold, and select candidates beased threshold: e.g. 0 means youngest, 100 means oldest, if we set age threshold to 80 then select dirty segments which has age in range of [80, 100] as candiddates; - set candidate_ratio threshold, and select candidates based the ratio, so that we can shrink candidates to those oldest segments; - select target segment with fewest valid blocks in order to migrate blocks with minimum cost; 2. select a target victim: - select candidates beased age threshold; - set candidate_radius threshold, search candidates whose age is around source victims, searching radius should less than the radius threshold. - select target segment with most valid blocks in order to avoid migrating current target segment. 3. merge valid blocks from source victim into target victim with SSR alloctor. Test steps: - create 160 dirty segments: * half of them have 128 valid blocks per segment * left of them have 384 valid blocks per segment - run background GC Benefit: GC count and block movement count both decrease obviously: - Before: - Valid: 86 - Dirty: 1 - Prefree: 11 - Free: 6001 (6001) GC calls: 162 (BG: 220) - data segments : 160 (160) - node segments : 2 (2) Try to move 41454 blocks (BG: 41454) - data blocks : 40960 (40960) - node blocks : 494 (494) IPU: 0 blocks SSR: 0 blocks in 0 segments LFS: 41364 blocks in 81 segments - After: - Valid: 87 - Dirty: 0 - Prefree: 4 - Free: 6008 (6008) GC calls: 75 (BG: 76) - data segments : 74 (74) - node segments : 1 (1) Try to move 12813 blocks (BG: 12813) - data blocks : 12544 (12544) - node blocks : 269 (269) IPU: 0 blocks SSR: 12032 blocks in 77 segments LFS: 855 blocks in 2 segments Signed-off-by: Chao Yu <yuchao0@huawei.com> [Jaegeuk Kim: fix a bug along with pinfile in-mem segment & clean up] Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2020-08-04 13:14:49 +00:00
int __init f2fs_create_garbage_collection_cache(void);
void f2fs_destroy_garbage_collection_cache(void);
/* victim selection function for cleaning and SSR */
int f2fs_get_victim(struct f2fs_sb_info *sbi, unsigned int *result,
int gc_type, int type, char alloc_mode,
unsigned long long age);
/*
* recovery.c
*/
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only);
bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi);
int __init f2fs_create_recovery_cache(void);
void f2fs_destroy_recovery_cache(void);
/*
* debug.c
*/
#ifdef CONFIG_F2FS_STAT_FS
struct f2fs_stat_info {
struct list_head stat_list;
struct f2fs_sb_info *sbi;
int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs;
int main_area_segs, main_area_sections, main_area_zones;
unsigned long long hit_cached[NR_EXTENT_CACHES];
unsigned long long hit_rbtree[NR_EXTENT_CACHES];
unsigned long long total_ext[NR_EXTENT_CACHES];
unsigned long long hit_total[NR_EXTENT_CACHES];
int ext_tree[NR_EXTENT_CACHES];
int zombie_tree[NR_EXTENT_CACHES];
int ext_node[NR_EXTENT_CACHES];
/* to count memory footprint */
unsigned long long ext_mem[NR_EXTENT_CACHES];
/* for read extent cache */
unsigned long long hit_largest;
f2fs: add block_age-based extent cache This patch introduces a runtime hot/cold data separation method for f2fs, in order to improve the accuracy for data temperature classification, reduce the garbage collection overhead after long-term data updates. Enhanced hot/cold data separation can record data block update frequency as "age" of the extent per inode, and take use of the age info to indicate better temperature type for data block allocation: - It records total data blocks allocated since mount; - When file extent has been updated, it calculate the count of data blocks allocated since last update as the age of the extent; - Before the data block allocated, it searches for the age info and chooses the suitable segment for allocation. Test and result: - Prepare: create about 30000 files * 3% for cold files (with cold file extension like .apk, from 3M to 10M) * 50% for warm files (with random file extension like .FcDxq, from 1K to 4M) * 47% for hot files (with hot file extension like .db, from 1K to 256K) - create(5%)/random update(90%)/delete(5%) the files * total write amount is about 70G * fsync will be called for .db files, and buffered write will be used for other files The storage of test device is large enough(128G) so that it will not switch to SSR mode during the test. Benefit: dirty segment count increment reduce about 14% - before: Dirty +21110 - after: Dirty +18286 Signed-off-by: qixiaoyu1 <qixiaoyu1@xiaomi.com> Signed-off-by: xiongping1 <xiongping1@xiaomi.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2022-12-02 01:37:15 +00:00
/* for block age extent cache */
unsigned long long allocated_data_blocks;
int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta;
int ndirty_data, ndirty_qdata;
unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all;
int nats, dirty_nats, sits, dirty_sits;
int free_nids, avail_nids, alloc_nids;
int total_count, utilization;
int nr_wb_cp_data, nr_wb_data;
int nr_rd_data, nr_rd_node, nr_rd_meta;
int nr_dio_read, nr_dio_write;
unsigned int io_skip_bggc, other_skip_bggc;
int nr_flushing, nr_flushed, flush_list_empty;
int nr_discarding, nr_discarded;
int nr_discard_cmd;
unsigned int undiscard_blks;
f2fs: introduce checkpoint_merge mount option We've added a new mount options, "checkpoint_merge" and "nocheckpoint_merge", which creates a kernel daemon and makes it to merge concurrent checkpoint requests as much as possible to eliminate redundant checkpoint issues. Plus, we can eliminate the sluggish issue caused by slow checkpoint operation when the checkpoint is done in a process context in a cgroup having low i/o budget and cpu shares. To make this do better, we set the default i/o priority of the kernel daemon to "3", to give one higher priority than other kernel threads. The below verification result explains this. The basic idea has come from https://opensource.samsung.com. [Verification] Android Pixel Device(ARM64, 7GB RAM, 256GB UFS) Create two I/O cgroups (fg w/ weight 100, bg w/ wight 20) Set "strict_guarantees" to "1" in BFQ tunables In "fg" cgroup, - thread A => trigger 1000 checkpoint operations "for i in `seq 1 1000`; do touch test_dir1/file; fsync test_dir1; done" - thread B => gererating async. I/O "fio --rw=write --numjobs=1 --bs=128k --runtime=3600 --time_based=1 --filename=test_img --name=test" In "bg" cgroup, - thread C => trigger repeated checkpoint operations "echo $$ > /dev/blkio/bg/tasks; while true; do touch test_dir2/file; fsync test_dir2; done" We've measured thread A's execution time. [ w/o patch ] Elapsed Time: Avg. 68 seconds [ w/ patch ] Elapsed Time: Avg. 48 seconds Reported-by: kernel test robot <lkp@intel.com> Reported-by: Dan Carpenter <dan.carpenter@oracle.com> [Jaegeuk Kim: fix the return value in f2fs_start_ckpt_thread, reported by Dan] Signed-off-by: Daeho Jeong <daehojeong@google.com> Signed-off-by: Sungjong Seo <sj1557.seo@samsung.com> Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2021-01-19 00:00:42 +00:00
int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt;
unsigned int cur_ckpt_time, peak_ckpt_time;
int inline_xattr, inline_inode, inline_dir, append, update, orphans;
int compr_inode, swapfile_inode;
unsigned long long compr_blocks;
int aw_cnt, max_aw_cnt;
unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks;
unsigned int bimodal, avg_vblocks;
int util_free, util_valid, util_invalid;
int rsvd_segs, overp_segs;
int dirty_count, node_pages, meta_pages, compress_pages;
int compress_page_hit;
int prefree_count, free_segs, free_secs;
int cp_call_count[MAX_CALL_TYPE], cp_count;
int gc_call_count[MAX_CALL_TYPE];
int gc_segs[2][2];
int gc_secs[2][2];
int tot_blks, data_blks, node_blks;
int bg_data_blks, bg_node_blks;
int curseg[NR_CURSEG_TYPE];
int cursec[NR_CURSEG_TYPE];
int curzone[NR_CURSEG_TYPE];
unsigned int dirty_seg[NR_CURSEG_TYPE];
unsigned int full_seg[NR_CURSEG_TYPE];
unsigned int valid_blks[NR_CURSEG_TYPE];
unsigned int meta_count[META_MAX];
unsigned int segment_count[2];
unsigned int block_count[2];
unsigned int inplace_count;
unsigned long long base_mem, cache_mem, page_mem;
};
static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)
{
return (struct f2fs_stat_info *)sbi->stat_info;
}
#define stat_inc_cp_call_count(sbi, foreground) \
atomic_inc(&sbi->cp_call_count[(foreground)])
#define stat_inc_cp_count(sbi) (F2FS_STAT(sbi)->cp_count++)
#define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++)
#define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++)
#define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++)
#define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--)
#define stat_inc_total_hit(sbi, type) (atomic64_inc(&(sbi)->total_hit_ext[type]))
#define stat_inc_rbtree_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_rbtree[type]))
#define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest))
#define stat_inc_cached_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_cached[type]))
#define stat_inc_inline_xattr(inode) \
do { \
if (f2fs_has_inline_xattr(inode)) \
(atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \
} while (0)
#define stat_dec_inline_xattr(inode) \
do { \
if (f2fs_has_inline_xattr(inode)) \
(atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \
} while (0)
#define stat_inc_inline_inode(inode) \
do { \
if (f2fs_has_inline_data(inode)) \
(atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \
} while (0)
#define stat_dec_inline_inode(inode) \
do { \
if (f2fs_has_inline_data(inode)) \
(atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \
} while (0)
#define stat_inc_inline_dir(inode) \
do { \
if (f2fs_has_inline_dentry(inode)) \
(atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \
} while (0)
#define stat_dec_inline_dir(inode) \
do { \
if (f2fs_has_inline_dentry(inode)) \
(atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \
} while (0)
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>
2019-11-01 10:07:14 +00:00
#define stat_inc_compr_inode(inode) \
do { \
if (f2fs_compressed_file(inode)) \
(atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \
} while (0)
#define stat_dec_compr_inode(inode) \
do { \
if (f2fs_compressed_file(inode)) \
(atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \
} while (0)
#define stat_add_compr_blocks(inode, blocks) \
(atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks))
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>
2019-11-01 10:07:14 +00:00
#define stat_sub_compr_blocks(inode, blocks) \
(atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks))
#define stat_inc_swapfile_inode(inode) \
(atomic_inc(&F2FS_I_SB(inode)->swapfile_inode))
#define stat_dec_swapfile_inode(inode) \
(atomic_dec(&F2FS_I_SB(inode)->swapfile_inode))
#define stat_inc_atomic_inode(inode) \
(atomic_inc(&F2FS_I_SB(inode)->atomic_files))
#define stat_dec_atomic_inode(inode) \
(atomic_dec(&F2FS_I_SB(inode)->atomic_files))
#define stat_inc_meta_count(sbi, blkaddr) \
do { \
if (blkaddr < SIT_I(sbi)->sit_base_addr) \
atomic_inc(&(sbi)->meta_count[META_CP]); \
else if (blkaddr < NM_I(sbi)->nat_blkaddr) \
atomic_inc(&(sbi)->meta_count[META_SIT]); \
else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \
atomic_inc(&(sbi)->meta_count[META_NAT]); \
else if (blkaddr < SM_I(sbi)->main_blkaddr) \
atomic_inc(&(sbi)->meta_count[META_SSA]); \
} while (0)
#define stat_inc_seg_type(sbi, curseg) \
((sbi)->segment_count[(curseg)->alloc_type]++)
#define stat_inc_block_count(sbi, curseg) \
((sbi)->block_count[(curseg)->alloc_type]++)
#define stat_inc_inplace_blocks(sbi) \
(atomic_inc(&(sbi)->inplace_count))
#define stat_update_max_atomic_write(inode) \
do { \
int cur = atomic_read(&F2FS_I_SB(inode)->atomic_files); \
int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \
if (cur > max) \
atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \
} while (0)
#define stat_inc_gc_call_count(sbi, foreground) \
(F2FS_STAT(sbi)->gc_call_count[(foreground)]++)
#define stat_inc_gc_sec_count(sbi, type, gc_type) \
(F2FS_STAT(sbi)->gc_secs[(type)][(gc_type)]++)
#define stat_inc_gc_seg_count(sbi, type, gc_type) \
(F2FS_STAT(sbi)->gc_segs[(type)][(gc_type)]++)
#define stat_inc_tot_blk_count(si, blks) \
((si)->tot_blks += (blks))
#define stat_inc_data_blk_count(sbi, blks, gc_type) \
do { \
struct f2fs_stat_info *si = F2FS_STAT(sbi); \
stat_inc_tot_blk_count(si, blks); \
si->data_blks += (blks); \
si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
} while (0)
#define stat_inc_node_blk_count(sbi, blks, gc_type) \
do { \
struct f2fs_stat_info *si = F2FS_STAT(sbi); \
stat_inc_tot_blk_count(si, blks); \
si->node_blks += (blks); \
si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
} while (0)
int f2fs_build_stats(struct f2fs_sb_info *sbi);
void f2fs_destroy_stats(struct f2fs_sb_info *sbi);
void __init f2fs_create_root_stats(void);
void f2fs_destroy_root_stats(void);
void f2fs_update_sit_info(struct f2fs_sb_info *sbi);
#else
#define stat_inc_cp_call_count(sbi, foreground) do { } while (0)
#define stat_inc_cp_count(sbi) do { } while (0)
#define stat_io_skip_bggc_count(sbi) do { } while (0)
#define stat_other_skip_bggc_count(sbi) do { } while (0)
#define stat_inc_dirty_inode(sbi, type) do { } while (0)
#define stat_dec_dirty_inode(sbi, type) do { } while (0)
#define stat_inc_total_hit(sbi, type) do { } while (0)
#define stat_inc_rbtree_node_hit(sbi, type) do { } while (0)
#define stat_inc_largest_node_hit(sbi) do { } while (0)
#define stat_inc_cached_node_hit(sbi, type) do { } while (0)
#define stat_inc_inline_xattr(inode) do { } while (0)
#define stat_dec_inline_xattr(inode) do { } while (0)
#define stat_inc_inline_inode(inode) do { } while (0)
#define stat_dec_inline_inode(inode) do { } while (0)
#define stat_inc_inline_dir(inode) do { } while (0)
#define stat_dec_inline_dir(inode) do { } while (0)
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>
2019-11-01 10:07:14 +00:00
#define stat_inc_compr_inode(inode) do { } while (0)
#define stat_dec_compr_inode(inode) do { } while (0)
#define stat_add_compr_blocks(inode, blocks) do { } while (0)
#define stat_sub_compr_blocks(inode, blocks) do { } while (0)
#define stat_inc_swapfile_inode(inode) do { } while (0)
#define stat_dec_swapfile_inode(inode) do { } while (0)
#define stat_inc_atomic_inode(inode) do { } while (0)
#define stat_dec_atomic_inode(inode) do { } while (0)
#define stat_update_max_atomic_write(inode) do { } while (0)
#define stat_inc_meta_count(sbi, blkaddr) do { } while (0)
#define stat_inc_seg_type(sbi, curseg) do { } while (0)
#define stat_inc_block_count(sbi, curseg) do { } while (0)
#define stat_inc_inplace_blocks(sbi) do { } while (0)
#define stat_inc_gc_call_count(sbi, foreground) do { } while (0)
#define stat_inc_gc_sec_count(sbi, type, gc_type) do { } while (0)
#define stat_inc_gc_seg_count(sbi, type, gc_type) do { } while (0)
#define stat_inc_tot_blk_count(si, blks) do { } while (0)
#define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0)
#define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0)
static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; }
static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { }
static inline void __init f2fs_create_root_stats(void) { }
static inline void f2fs_destroy_root_stats(void) { }
static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {}
#endif
extern const struct file_operations f2fs_dir_operations;
extern const struct file_operations f2fs_file_operations;
extern const struct inode_operations f2fs_file_inode_operations;
extern const struct address_space_operations f2fs_dblock_aops;
extern const struct address_space_operations f2fs_node_aops;
extern const struct address_space_operations f2fs_meta_aops;
extern const struct inode_operations f2fs_dir_inode_operations;
extern const struct inode_operations f2fs_symlink_inode_operations;
extern const struct inode_operations f2fs_encrypted_symlink_inode_operations;
extern const struct inode_operations f2fs_special_inode_operations;
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
extern struct kmem_cache *f2fs_inode_entry_slab;
/*
* inline.c
*/
bool f2fs_may_inline_data(struct inode *inode);
bool f2fs_sanity_check_inline_data(struct inode *inode, struct page *ipage);
bool f2fs_may_inline_dentry(struct inode *inode);
void f2fs_do_read_inline_data(struct folio *folio, struct page *ipage);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
void f2fs_truncate_inline_inode(struct inode *inode,
struct page *ipage, u64 from);
int f2fs_read_inline_data(struct inode *inode, struct folio *folio);
int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page);
int f2fs_convert_inline_inode(struct inode *inode);
int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry);
int f2fs_write_inline_data(struct inode *inode, struct page *page);
int f2fs_recover_inline_data(struct inode *inode, struct page *npage);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir,
2020-05-07 07:59:04 +00:00
const struct f2fs_filename *fname,
struct page **res_page);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent,
struct page *ipage);
2020-05-07 07:59:04 +00:00
int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname,
struct inode *inode, nid_t ino, umode_t mode);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry,
struct page *page, struct inode *dir,
struct inode *inode);
bool f2fs_empty_inline_dir(struct inode *dir);
int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx,
struct fscrypt_str *fstr);
int f2fs_inline_data_fiemap(struct inode *inode,
struct fiemap_extent_info *fieinfo,
__u64 start, __u64 len);
/*
* shrinker.c
*/
unsigned long f2fs_shrink_count(struct shrinker *shrink,
struct shrink_control *sc);
unsigned long f2fs_shrink_scan(struct shrinker *shrink,
struct shrink_control *sc);
void f2fs_join_shrinker(struct f2fs_sb_info *sbi);
void f2fs_leave_shrinker(struct f2fs_sb_info *sbi);
/*
* extent_cache.c
*/
f2fs: fix to cover read extent cache access with lock syzbot reports a f2fs bug as below: BUG: KASAN: slab-use-after-free in sanity_check_extent_cache+0x370/0x410 fs/f2fs/extent_cache.c:46 Read of size 4 at addr ffff8880739ab220 by task syz-executor200/5097 CPU: 0 PID: 5097 Comm: syz-executor200 Not tainted 6.9.0-rc6-syzkaller #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 03/27/2024 Call Trace: <TASK> __dump_stack lib/dump_stack.c:88 [inline] dump_stack_lvl+0x241/0x360 lib/dump_stack.c:114 print_address_description mm/kasan/report.c:377 [inline] print_report+0x169/0x550 mm/kasan/report.c:488 kasan_report+0x143/0x180 mm/kasan/report.c:601 sanity_check_extent_cache+0x370/0x410 fs/f2fs/extent_cache.c:46 do_read_inode fs/f2fs/inode.c:509 [inline] f2fs_iget+0x33e1/0x46e0 fs/f2fs/inode.c:560 f2fs_nfs_get_inode+0x74/0x100 fs/f2fs/super.c:3237 generic_fh_to_dentry+0x9f/0xf0 fs/libfs.c:1413 exportfs_decode_fh_raw+0x152/0x5f0 fs/exportfs/expfs.c:444 exportfs_decode_fh+0x3c/0x80 fs/exportfs/expfs.c:584 do_handle_to_path fs/fhandle.c:155 [inline] handle_to_path fs/fhandle.c:210 [inline] do_handle_open+0x495/0x650 fs/fhandle.c:226 do_syscall_x64 arch/x86/entry/common.c:52 [inline] do_syscall_64+0xf5/0x240 arch/x86/entry/common.c:83 entry_SYSCALL_64_after_hwframe+0x77/0x7f We missed to cover sanity_check_extent_cache() w/ extent cache lock, so, below race case may happen, result in use after free issue. - f2fs_iget - do_read_inode - f2fs_init_read_extent_tree : add largest extent entry in to cache - shrink - f2fs_shrink_read_extent_tree - __shrink_extent_tree - __detach_extent_node : drop largest extent entry - sanity_check_extent_cache : access et->largest w/o lock let's refactor sanity_check_extent_cache() to avoid extent cache access and call it before f2fs_init_read_extent_tree() to fix this issue. Reported-by: syzbot+74ebe2104433e9dc610d@syzkaller.appspotmail.com Closes: https://lore.kernel.org/linux-f2fs-devel/00000000000009beea061740a531@google.com Signed-off-by: Chao Yu <chao@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2024-05-31 02:00:32 +00:00
bool sanity_check_extent_cache(struct inode *inode, struct page *ipage);
void f2fs_init_extent_tree(struct inode *inode);
void f2fs_drop_extent_tree(struct inode *inode);
void f2fs_destroy_extent_node(struct inode *inode);
void f2fs_destroy_extent_tree(struct inode *inode);
f2fs: clean up symbol namespace As Ted reported: "Hi, I was looking at f2fs's sources recently, and I noticed that there is a very large number of non-static symbols which don't have a f2fs prefix. There's well over a hundred (see attached below). As one example, in fs/f2fs/dir.c there is: unsigned char get_de_type(struct f2fs_dir_entry *de) This function is clearly only useful for f2fs, but it has a generic name. This means that if any other file system tries to have the same symbol name, there will be a symbol conflict and the kernel would not successfully build. It also means that when someone is looking f2fs sources, it's not at all obvious whether a function such as read_data_page(), invalidate_blocks(), is a generic kernel function found in the fs, mm, or block layers, or a f2fs specific function. You might want to fix this at some point. Hopefully Kent's bcachefs isn't similarly using genericly named functions, since that might cause conflicts with f2fs's functions --- but just as this would be a problem that we would rightly insist that Kent fix, this is something that we should have rightly insisted that f2fs should have fixed before it was integrated into the mainline kernel. acquire_orphan_inode add_ino_entry add_orphan_inode allocate_data_block allocate_new_segments alloc_nid alloc_nid_done alloc_nid_failed available_free_memory ...." This patch adds "f2fs_" prefix for all non-static symbols in order to: a) avoid conflict with other kernel generic symbols; b) to indicate the function is f2fs specific one instead of generic one; Reported-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-29 16:20:41 +00:00
void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi);
int __init f2fs_create_extent_cache(void);
void f2fs_destroy_extent_cache(void);
/* read extent cache ops */
void f2fs_init_read_extent_tree(struct inode *inode, struct page *ipage);
bool f2fs_lookup_read_extent_cache(struct inode *inode, pgoff_t pgofs,
struct extent_info *ei);
bool f2fs_lookup_read_extent_cache_block(struct inode *inode, pgoff_t index,
block_t *blkaddr);
void f2fs_update_read_extent_cache(struct dnode_of_data *dn);
void f2fs_update_read_extent_cache_range(struct dnode_of_data *dn,
pgoff_t fofs, block_t blkaddr, unsigned int len);
unsigned int f2fs_shrink_read_extent_tree(struct f2fs_sb_info *sbi,
int nr_shrink);
f2fs: add block_age-based extent cache This patch introduces a runtime hot/cold data separation method for f2fs, in order to improve the accuracy for data temperature classification, reduce the garbage collection overhead after long-term data updates. Enhanced hot/cold data separation can record data block update frequency as "age" of the extent per inode, and take use of the age info to indicate better temperature type for data block allocation: - It records total data blocks allocated since mount; - When file extent has been updated, it calculate the count of data blocks allocated since last update as the age of the extent; - Before the data block allocated, it searches for the age info and chooses the suitable segment for allocation. Test and result: - Prepare: create about 30000 files * 3% for cold files (with cold file extension like .apk, from 3M to 10M) * 50% for warm files (with random file extension like .FcDxq, from 1K to 4M) * 47% for hot files (with hot file extension like .db, from 1K to 256K) - create(5%)/random update(90%)/delete(5%) the files * total write amount is about 70G * fsync will be called for .db files, and buffered write will be used for other files The storage of test device is large enough(128G) so that it will not switch to SSR mode during the test. Benefit: dirty segment count increment reduce about 14% - before: Dirty +21110 - after: Dirty +18286 Signed-off-by: qixiaoyu1 <qixiaoyu1@xiaomi.com> Signed-off-by: xiongping1 <xiongping1@xiaomi.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2022-12-02 01:37:15 +00:00
/* block age extent cache ops */
void f2fs_init_age_extent_tree(struct inode *inode);
bool f2fs_lookup_age_extent_cache(struct inode *inode, pgoff_t pgofs,
struct extent_info *ei);
void f2fs_update_age_extent_cache(struct dnode_of_data *dn);
void f2fs_update_age_extent_cache_range(struct dnode_of_data *dn,
pgoff_t fofs, unsigned int len);
unsigned int f2fs_shrink_age_extent_tree(struct f2fs_sb_info *sbi,
int nr_shrink);
/*
* sysfs.c
*/
#define MIN_RA_MUL 2
#define MAX_RA_MUL 256
int __init f2fs_init_sysfs(void);
void f2fs_exit_sysfs(void);
int f2fs_register_sysfs(struct f2fs_sb_info *sbi);
void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi);
f2fs: add fs-verity support Add fs-verity support to f2fs. fs-verity is a filesystem feature that enables transparent integrity protection and authentication of read-only files. It uses a dm-verity like mechanism at the file level: a Merkle tree is used to verify any block in the file in log(filesize) time. It is implemented mainly by helper functions in fs/verity/. See Documentation/filesystems/fsverity.rst for the full documentation. The f2fs support for fs-verity consists of: - Adding a filesystem feature flag and an inode flag for fs-verity. - Implementing the fsverity_operations to support enabling verity on an inode and reading/writing the verity metadata. - Updating ->readpages() to verify data as it's read from verity files and to support reading verity metadata pages. - Updating ->write_begin(), ->write_end(), and ->writepages() to support writing verity metadata pages. - Calling the fs-verity hooks for ->open(), ->setattr(), and ->ioctl(). Like ext4, f2fs stores the verity metadata (Merkle tree and fsverity_descriptor) past the end of the file, starting at the first 64K boundary beyond i_size. This approach works because (a) verity files are readonly, and (b) pages fully beyond i_size aren't visible to userspace but can be read/written internally by f2fs with only some relatively small changes to f2fs. Extended attributes cannot be used because (a) f2fs limits the total size of an inode's xattr entries to 4096 bytes, which wouldn't be enough for even a single Merkle tree block, and (b) f2fs encryption doesn't encrypt xattrs, yet the verity metadata *must* be encrypted when the file is because it contains hashes of the plaintext data. Acked-by: Jaegeuk Kim <jaegeuk@kernel.org> Acked-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Eric Biggers <ebiggers@google.com>
2019-07-22 16:26:24 +00:00
/* verity.c */
extern const struct fsverity_operations f2fs_verityops;
/*
* crypto support
*/
static inline bool f2fs_encrypted_file(struct inode *inode)
{
return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode);
}
static inline void f2fs_set_encrypted_inode(struct inode *inode)
{
#ifdef CONFIG_FS_ENCRYPTION
file_set_encrypt(inode);
f2fs_set_inode_flags(inode);
#endif
}
/*
* Returns true if the reads of the inode's data need to undergo some
* postprocessing step, like decryption or authenticity verification.
*/
static inline bool f2fs_post_read_required(struct inode *inode)
{
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>
2019-11-01 10:07:14 +00:00
return f2fs_encrypted_file(inode) || fsverity_active(inode) ||
f2fs_compressed_file(inode);
}
static inline bool f2fs_used_in_atomic_write(struct inode *inode)
{
return f2fs_is_atomic_file(inode) || f2fs_is_cow_file(inode);
}
static inline bool f2fs_meta_inode_gc_required(struct inode *inode)
{
return f2fs_post_read_required(inode) || f2fs_used_in_atomic_write(inode);
}
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>
2019-11-01 10:07:14 +00:00
/*
* compress.c
*/
#ifdef CONFIG_F2FS_FS_COMPRESSION
bool f2fs_is_compressed_page(struct page *page);
struct page *f2fs_compress_control_page(struct page *page);
int f2fs_prepare_compress_overwrite(struct inode *inode,
struct page **pagep, pgoff_t index, void **fsdata);
bool f2fs_compress_write_end(struct inode *inode, void *fsdata,
pgoff_t index, unsigned copied);
int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock);
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>
2019-11-01 10:07:14 +00:00
void f2fs_compress_write_end_io(struct bio *bio, struct page *page);
bool f2fs_is_compress_backend_ready(struct inode *inode);
bool f2fs_is_compress_level_valid(int alg, int lvl);
int __init f2fs_init_compress_mempool(void);
void f2fs_destroy_compress_mempool(void);
void f2fs_decompress_cluster(struct decompress_io_ctx *dic, bool in_task);
void f2fs_end_read_compressed_page(struct page *page, bool failed,
block_t blkaddr, bool in_task);
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>
2019-11-01 10:07:14 +00:00
bool f2fs_cluster_is_empty(struct compress_ctx *cc);
bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index);
bool f2fs_all_cluster_page_ready(struct compress_ctx *cc, struct page **pages,
int index, int nr_pages, bool uptodate);
bool f2fs_sanity_check_cluster(struct dnode_of_data *dn);
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>
2019-11-01 10:07:14 +00:00
void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page);
int f2fs_write_multi_pages(struct compress_ctx *cc,
int *submitted,
struct writeback_control *wbc,
enum iostat_type io_type);
int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index);
void f2fs_update_read_extent_tree_range_compressed(struct inode *inode,
pgoff_t fofs, block_t blkaddr,
unsigned int llen, unsigned int c_len);
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>
2019-11-01 10:07:14 +00:00
int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret,
unsigned nr_pages, sector_t *last_block_in_bio,
struct readahead_control *rac, bool for_write);
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>
2019-11-01 10:07:14 +00:00
struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc);
void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed,
bool in_task);
void f2fs_put_page_dic(struct page *page, bool in_task);
unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn,
unsigned int ofs_in_node);
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>
2019-11-01 10:07:14 +00:00
int f2fs_init_compress_ctx(struct compress_ctx *cc);
void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse);
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>
2019-11-01 10:07:14 +00:00
void f2fs_init_compress_info(struct f2fs_sb_info *sbi);
int f2fs_init_compress_inode(struct f2fs_sb_info *sbi);
void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi);
int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi);
void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi);
int __init f2fs_init_compress_cache(void);
void f2fs_destroy_compress_cache(void);
struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi);
void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr);
void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page,
nid_t ino, block_t blkaddr);
bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page,
block_t blkaddr);
void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino);
#define inc_compr_inode_stat(inode) \
do { \
struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \
sbi->compr_new_inode++; \
} while (0)
#define add_compr_block_stat(inode, blocks) \
do { \
struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \
int diff = F2FS_I(inode)->i_cluster_size - blocks; \
sbi->compr_written_block += blocks; \
sbi->compr_saved_block += diff; \
} while (0)
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>
2019-11-01 10:07:14 +00:00
#else
static inline bool f2fs_is_compressed_page(struct page *page) { return false; }
static inline bool f2fs_is_compress_backend_ready(struct inode *inode)
{
if (!f2fs_compressed_file(inode))
return true;
/* not support compression */
return false;
}
static inline bool f2fs_is_compress_level_valid(int alg, int lvl) { return false; }
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>
2019-11-01 10:07:14 +00:00
static inline struct page *f2fs_compress_control_page(struct page *page)
{
WARN_ON_ONCE(1);
return ERR_PTR(-EINVAL);
}
static inline int __init f2fs_init_compress_mempool(void) { return 0; }
static inline void f2fs_destroy_compress_mempool(void) { }
static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic,
bool in_task) { }
static inline void f2fs_end_read_compressed_page(struct page *page,
bool failed, block_t blkaddr, bool in_task)
{
WARN_ON_ONCE(1);
}
static inline void f2fs_put_page_dic(struct page *page, bool in_task)
{
WARN_ON_ONCE(1);
}
static inline unsigned int f2fs_cluster_blocks_are_contiguous(
struct dnode_of_data *dn, unsigned int ofs_in_node) { return 0; }
static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; }
static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; }
static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { }
static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; }
static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { }
static inline int __init f2fs_init_compress_cache(void) { return 0; }
static inline void f2fs_destroy_compress_cache(void) { }
static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi,
block_t blkaddr) { }
static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi,
struct page *page, nid_t ino, block_t blkaddr) { }
static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi,
struct page *page, block_t blkaddr) { return false; }
static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi,
nid_t ino) { }
#define inc_compr_inode_stat(inode) do { } while (0)
static inline void f2fs_update_read_extent_tree_range_compressed(
struct inode *inode,
pgoff_t fofs, block_t blkaddr,
unsigned int llen, unsigned int c_len) { }
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>
2019-11-01 10:07:14 +00:00
#endif
static inline int set_compress_context(struct inode *inode)
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>
2019-11-01 10:07:14 +00:00
{
#ifdef CONFIG_F2FS_FS_COMPRESSION
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>
2019-11-01 10:07:14 +00:00
struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
struct f2fs_inode_info *fi = F2FS_I(inode);
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>
2019-11-01 10:07:14 +00:00
fi->i_compress_algorithm = F2FS_OPTION(sbi).compress_algorithm;
fi->i_log_cluster_size = F2FS_OPTION(sbi).compress_log_size;
fi->i_compress_flag = F2FS_OPTION(sbi).compress_chksum ?
BIT(COMPRESS_CHKSUM) : 0;
fi->i_cluster_size = BIT(fi->i_log_cluster_size);
if ((fi->i_compress_algorithm == COMPRESS_LZ4 ||
fi->i_compress_algorithm == COMPRESS_ZSTD) &&
F2FS_OPTION(sbi).compress_level)
fi->i_compress_level = F2FS_OPTION(sbi).compress_level;
fi->i_flags |= F2FS_COMPR_FL;
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>
2019-11-01 10:07:14 +00:00
set_inode_flag(inode, FI_COMPRESSED_FILE);
stat_inc_compr_inode(inode);
inc_compr_inode_stat(inode);
f2fs_mark_inode_dirty_sync(inode, true);
return 0;
#else
return -EOPNOTSUPP;
#endif
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>
2019-11-01 10:07:14 +00:00
}
static inline bool f2fs_disable_compressed_file(struct inode *inode)
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>
2019-11-01 10:07:14 +00:00
{
struct f2fs_inode_info *fi = F2FS_I(inode);
f2fs_down_write(&fi->i_sem);
if (!f2fs_compressed_file(inode)) {
f2fs_up_write(&fi->i_sem);
return true;
}
if (f2fs_is_mmap_file(inode) ||
(S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode))) {
f2fs_up_write(&fi->i_sem);
return false;
}
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>
2019-11-01 10:07:14 +00:00
fi->i_flags &= ~F2FS_COMPR_FL;
stat_dec_compr_inode(inode);
clear_inode_flag(inode, FI_COMPRESSED_FILE);
f2fs_mark_inode_dirty_sync(inode, true);
f2fs_up_write(&fi->i_sem);
return true;
}
#define F2FS_FEATURE_FUNCS(name, flagname) \
static inline bool f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \
{ \
return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \
}
F2FS_FEATURE_FUNCS(encrypt, ENCRYPT);
F2FS_FEATURE_FUNCS(blkzoned, BLKZONED);
F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR);
F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA);
F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM);
F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR);
F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO);
F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME);
F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND);
f2fs: add fs-verity support Add fs-verity support to f2fs. fs-verity is a filesystem feature that enables transparent integrity protection and authentication of read-only files. It uses a dm-verity like mechanism at the file level: a Merkle tree is used to verify any block in the file in log(filesize) time. It is implemented mainly by helper functions in fs/verity/. See Documentation/filesystems/fsverity.rst for the full documentation. The f2fs support for fs-verity consists of: - Adding a filesystem feature flag and an inode flag for fs-verity. - Implementing the fsverity_operations to support enabling verity on an inode and reading/writing the verity metadata. - Updating ->readpages() to verify data as it's read from verity files and to support reading verity metadata pages. - Updating ->write_begin(), ->write_end(), and ->writepages() to support writing verity metadata pages. - Calling the fs-verity hooks for ->open(), ->setattr(), and ->ioctl(). Like ext4, f2fs stores the verity metadata (Merkle tree and fsverity_descriptor) past the end of the file, starting at the first 64K boundary beyond i_size. This approach works because (a) verity files are readonly, and (b) pages fully beyond i_size aren't visible to userspace but can be read/written internally by f2fs with only some relatively small changes to f2fs. Extended attributes cannot be used because (a) f2fs limits the total size of an inode's xattr entries to 4096 bytes, which wouldn't be enough for even a single Merkle tree block, and (b) f2fs encryption doesn't encrypt xattrs, yet the verity metadata *must* be encrypted when the file is because it contains hashes of the plaintext data. Acked-by: Jaegeuk Kim <jaegeuk@kernel.org> Acked-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Eric Biggers <ebiggers@google.com>
2019-07-22 16:26:24 +00:00
F2FS_FEATURE_FUNCS(verity, VERITY);
F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM);
f2fs: include charset encoding information in the superblock Add charset encoding to f2fs to support casefolding. It is modeled after the same feature introduced in commit c83ad55eaa91 ("ext4: include charset encoding information in the superblock") Currently this is not compatible with encryption, similar to the current ext4 imlpementation. This will change in the future. >From the ext4 patch: """ The s_encoding field stores a magic number indicating the encoding format and version used globally by file and directory names in the filesystem. The s_encoding_flags defines policies for using the charset encoding, like how to handle invalid sequences. The magic number is mapped to the exact charset table, but the mapping is specific to ext4. Since we don't have any commitment to support old encodings, the only encoding I am supporting right now is utf8-12.1.0. The current implementation prevents the user from enabling encoding and per-directory encryption on the same filesystem at the same time. The incompatibility between these features lies in how we do efficient directory searches when we cannot be sure the encryption of the user provided fname will match the actual hash stored in the disk without decrypting every directory entry, because of normalization cases. My quickest solution is to simply block the concurrent use of these features for now, and enable it later, once we have a better solution. """ Signed-off-by: Daniel Rosenberg <drosen@google.com> Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-07-23 23:05:28 +00:00
F2FS_FEATURE_FUNCS(casefold, CASEFOLD);
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>
2019-11-01 10:07:14 +00:00
F2FS_FEATURE_FUNCS(compression, COMPRESSION);
F2FS_FEATURE_FUNCS(readonly, RO);
#ifdef CONFIG_BLK_DEV_ZONED
static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi,
block_t blkaddr)
{
unsigned int zno = blkaddr / sbi->blocks_per_blkz;
return test_bit(zno, FDEV(devi).blkz_seq);
}
#endif
static inline int f2fs_bdev_index(struct f2fs_sb_info *sbi,
struct block_device *bdev)
{
int i;
if (!f2fs_is_multi_device(sbi))
return 0;
for (i = 0; i < sbi->s_ndevs; i++)
if (FDEV(i).bdev == bdev)
return i;
WARN_ON(1);
return -1;
}
f2fs: fix to avoid NULL pointer dereference on se->discard_map https://bugzilla.kernel.org/show_bug.cgi?id=200951 These is a NULL pointer dereference issue reported in bugzilla: Hi, in the setup there is a SATA SSD connected to a SATA-to-USB bridge. The disc is "Samsung SSD 850 PRO 256G" which supports TRIM. There are four partitions: sda1: FAT /boot sda2: F2FS / sda3: F2FS /home sda4: F2FS The bridge is ASMT1153e which uses the "uas" driver. There is no TRIM pass-through, so, when mounting it reports: mounting with "discard" option, but the device does not support discard The USB host is USB3.0 and UASP capable. It is the one on RK3399. Given this everything works fine, except there is no TRIM support. In order to enable TRIM a new UDEV rule is added [1]: /etc/udev/rules.d/10-sata-bridge-trim.rules: ACTION=="add|change", ATTRS{idVendor}=="174c", ATTRS{idProduct}=="55aa", SUBSYSTEM=="scsi_disk", ATTR{provisioning_mode}="unmap" After reboot any F2FS write hangs forever and dmesg reports: Unable to handle kernel NULL pointer dereference Also tested on a x86_64 system: works fine even with TRIM enabled. same disc same bridge different usb host controller different cpu architecture not root filesystem Regards, Vicenç. [1] Post #5 in https://bbs.archlinux.org/viewtopic.php?id=236280 Unable to handle kernel NULL pointer dereference at virtual address 000000000000003e Mem abort info: ESR = 0x96000004 Exception class = DABT (current EL), IL = 32 bits SET = 0, FnV = 0 EA = 0, S1PTW = 0 Data abort info: ISV = 0, ISS = 0x00000004 CM = 0, WnR = 0 user pgtable: 4k pages, 48-bit VAs, pgdp = 00000000626e3122 [000000000000003e] pgd=0000000000000000 Internal error: Oops: 96000004 [#1] SMP Modules linked in: overlay snd_soc_hdmi_codec rc_cec dw_hdmi_i2s_audio dw_hdmi_cec snd_soc_simple_card snd_soc_simple_card_utils snd_soc_rockchip_i2s rockchip_rga snd_soc_rockchip_pcm rockchipdrm videobuf2_dma_sg v4l2_mem2mem rtc_rk808 videobuf2_memops analogix_dp videobuf2_v4l2 videobuf2_common dw_hdmi dw_wdt cec rc_core videodev drm_kms_helper media drm rockchip_thermal rockchip_saradc realtek drm_panel_orientation_quirks syscopyarea sysfillrect sysimgblt fb_sys_fops dwmac_rk stmmac_platform stmmac pwm_bl squashfs loop crypto_user gpio_keys hid_kensington CPU: 5 PID: 957 Comm: nvim Not tainted 4.19.0-rc1-1-ARCH #1 Hardware name: Sapphire-RK3399 Board (DT) pstate: 00000005 (nzcv daif -PAN -UAO) pc : update_sit_entry+0x304/0x4b0 lr : update_sit_entry+0x108/0x4b0 sp : ffff00000ca13bd0 x29: ffff00000ca13bd0 x28: 000000000000003e x27: 0000000000000020 x26: 0000000000080000 x25: 0000000000000048 x24: ffff8000ebb85cf8 x23: 0000000000000253 x22: 00000000ffffffff x21: 00000000000535f2 x20: 00000000ffffffdf x19: ffff8000eb9e6800 x18: ffff8000eb9e6be8 x17: 0000000007ce6926 x16: 000000001c83ffa8 x15: 0000000000000000 x14: ffff8000f602df90 x13: 0000000000000006 x12: 0000000000000040 x11: 0000000000000228 x10: 0000000000000000 x9 : 0000000000000000 x8 : 0000000000000000 x7 : 00000000000535f2 x6 : ffff8000ebff3440 x5 : ffff8000ebff3440 x4 : ffff8000ebe3a6c8 x3 : 00000000ffffffff x2 : 0000000000000020 x1 : 0000000000000000 x0 : ffff8000eb9e5800 Process nvim (pid: 957, stack limit = 0x0000000063a78320) Call trace: update_sit_entry+0x304/0x4b0 f2fs_invalidate_blocks+0x98/0x140 truncate_node+0x90/0x400 f2fs_remove_inode_page+0xe8/0x340 f2fs_evict_inode+0x2b0/0x408 evict+0xe0/0x1e0 iput+0x160/0x260 do_unlinkat+0x214/0x298 __arm64_sys_unlinkat+0x3c/0x68 el0_svc_handler+0x94/0x118 el0_svc+0x8/0xc Code: f9400800 b9488400 36080140 f9400f01 (387c4820) ---[ end trace a0f21a307118c477 ]--- The reason is it is possible to enable discard flag on block queue via UDEV, but during mount, f2fs will initialize se->discard_map only if this flag is set, once the flag is set after mount, f2fs may dereference NULL pointer on se->discard_map. So this patch does below changes to fix this issue: - initialize and update se->discard_map all the time. - don't clear DISCARD option if device has no QUEUE_FLAG_DISCARD flag during mount. - don't issue small discard on zoned block device. - introduce some functions to enhance the readability. Signed-off-by: Chao Yu <yuchao0@huawei.com> Tested-by: Vicente Bergas <vicencb@gmail.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-09-03 19:52:17 +00:00
static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi)
{
return f2fs_sb_has_blkzoned(sbi);
f2fs: fix to avoid NULL pointer dereference on se->discard_map https://bugzilla.kernel.org/show_bug.cgi?id=200951 These is a NULL pointer dereference issue reported in bugzilla: Hi, in the setup there is a SATA SSD connected to a SATA-to-USB bridge. The disc is "Samsung SSD 850 PRO 256G" which supports TRIM. There are four partitions: sda1: FAT /boot sda2: F2FS / sda3: F2FS /home sda4: F2FS The bridge is ASMT1153e which uses the "uas" driver. There is no TRIM pass-through, so, when mounting it reports: mounting with "discard" option, but the device does not support discard The USB host is USB3.0 and UASP capable. It is the one on RK3399. Given this everything works fine, except there is no TRIM support. In order to enable TRIM a new UDEV rule is added [1]: /etc/udev/rules.d/10-sata-bridge-trim.rules: ACTION=="add|change", ATTRS{idVendor}=="174c", ATTRS{idProduct}=="55aa", SUBSYSTEM=="scsi_disk", ATTR{provisioning_mode}="unmap" After reboot any F2FS write hangs forever and dmesg reports: Unable to handle kernel NULL pointer dereference Also tested on a x86_64 system: works fine even with TRIM enabled. same disc same bridge different usb host controller different cpu architecture not root filesystem Regards, Vicenç. [1] Post #5 in https://bbs.archlinux.org/viewtopic.php?id=236280 Unable to handle kernel NULL pointer dereference at virtual address 000000000000003e Mem abort info: ESR = 0x96000004 Exception class = DABT (current EL), IL = 32 bits SET = 0, FnV = 0 EA = 0, S1PTW = 0 Data abort info: ISV = 0, ISS = 0x00000004 CM = 0, WnR = 0 user pgtable: 4k pages, 48-bit VAs, pgdp = 00000000626e3122 [000000000000003e] pgd=0000000000000000 Internal error: Oops: 96000004 [#1] SMP Modules linked in: overlay snd_soc_hdmi_codec rc_cec dw_hdmi_i2s_audio dw_hdmi_cec snd_soc_simple_card snd_soc_simple_card_utils snd_soc_rockchip_i2s rockchip_rga snd_soc_rockchip_pcm rockchipdrm videobuf2_dma_sg v4l2_mem2mem rtc_rk808 videobuf2_memops analogix_dp videobuf2_v4l2 videobuf2_common dw_hdmi dw_wdt cec rc_core videodev drm_kms_helper media drm rockchip_thermal rockchip_saradc realtek drm_panel_orientation_quirks syscopyarea sysfillrect sysimgblt fb_sys_fops dwmac_rk stmmac_platform stmmac pwm_bl squashfs loop crypto_user gpio_keys hid_kensington CPU: 5 PID: 957 Comm: nvim Not tainted 4.19.0-rc1-1-ARCH #1 Hardware name: Sapphire-RK3399 Board (DT) pstate: 00000005 (nzcv daif -PAN -UAO) pc : update_sit_entry+0x304/0x4b0 lr : update_sit_entry+0x108/0x4b0 sp : ffff00000ca13bd0 x29: ffff00000ca13bd0 x28: 000000000000003e x27: 0000000000000020 x26: 0000000000080000 x25: 0000000000000048 x24: ffff8000ebb85cf8 x23: 0000000000000253 x22: 00000000ffffffff x21: 00000000000535f2 x20: 00000000ffffffdf x19: ffff8000eb9e6800 x18: ffff8000eb9e6be8 x17: 0000000007ce6926 x16: 000000001c83ffa8 x15: 0000000000000000 x14: ffff8000f602df90 x13: 0000000000000006 x12: 0000000000000040 x11: 0000000000000228 x10: 0000000000000000 x9 : 0000000000000000 x8 : 0000000000000000 x7 : 00000000000535f2 x6 : ffff8000ebff3440 x5 : ffff8000ebff3440 x4 : ffff8000ebe3a6c8 x3 : 00000000ffffffff x2 : 0000000000000020 x1 : 0000000000000000 x0 : ffff8000eb9e5800 Process nvim (pid: 957, stack limit = 0x0000000063a78320) Call trace: update_sit_entry+0x304/0x4b0 f2fs_invalidate_blocks+0x98/0x140 truncate_node+0x90/0x400 f2fs_remove_inode_page+0xe8/0x340 f2fs_evict_inode+0x2b0/0x408 evict+0xe0/0x1e0 iput+0x160/0x260 do_unlinkat+0x214/0x298 __arm64_sys_unlinkat+0x3c/0x68 el0_svc_handler+0x94/0x118 el0_svc+0x8/0xc Code: f9400800 b9488400 36080140 f9400f01 (387c4820) ---[ end trace a0f21a307118c477 ]--- The reason is it is possible to enable discard flag on block queue via UDEV, but during mount, f2fs will initialize se->discard_map only if this flag is set, once the flag is set after mount, f2fs may dereference NULL pointer on se->discard_map. So this patch does below changes to fix this issue: - initialize and update se->discard_map all the time. - don't clear DISCARD option if device has no QUEUE_FLAG_DISCARD flag during mount. - don't issue small discard on zoned block device. - introduce some functions to enhance the readability. Signed-off-by: Chao Yu <yuchao0@huawei.com> Tested-by: Vicente Bergas <vicencb@gmail.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-09-03 19:52:17 +00:00
}
static inline bool f2fs_bdev_support_discard(struct block_device *bdev)
{
return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev);
}
f2fs: fix to avoid NULL pointer dereference on se->discard_map https://bugzilla.kernel.org/show_bug.cgi?id=200951 These is a NULL pointer dereference issue reported in bugzilla: Hi, in the setup there is a SATA SSD connected to a SATA-to-USB bridge. The disc is "Samsung SSD 850 PRO 256G" which supports TRIM. There are four partitions: sda1: FAT /boot sda2: F2FS / sda3: F2FS /home sda4: F2FS The bridge is ASMT1153e which uses the "uas" driver. There is no TRIM pass-through, so, when mounting it reports: mounting with "discard" option, but the device does not support discard The USB host is USB3.0 and UASP capable. It is the one on RK3399. Given this everything works fine, except there is no TRIM support. In order to enable TRIM a new UDEV rule is added [1]: /etc/udev/rules.d/10-sata-bridge-trim.rules: ACTION=="add|change", ATTRS{idVendor}=="174c", ATTRS{idProduct}=="55aa", SUBSYSTEM=="scsi_disk", ATTR{provisioning_mode}="unmap" After reboot any F2FS write hangs forever and dmesg reports: Unable to handle kernel NULL pointer dereference Also tested on a x86_64 system: works fine even with TRIM enabled. same disc same bridge different usb host controller different cpu architecture not root filesystem Regards, Vicenç. [1] Post #5 in https://bbs.archlinux.org/viewtopic.php?id=236280 Unable to handle kernel NULL pointer dereference at virtual address 000000000000003e Mem abort info: ESR = 0x96000004 Exception class = DABT (current EL), IL = 32 bits SET = 0, FnV = 0 EA = 0, S1PTW = 0 Data abort info: ISV = 0, ISS = 0x00000004 CM = 0, WnR = 0 user pgtable: 4k pages, 48-bit VAs, pgdp = 00000000626e3122 [000000000000003e] pgd=0000000000000000 Internal error: Oops: 96000004 [#1] SMP Modules linked in: overlay snd_soc_hdmi_codec rc_cec dw_hdmi_i2s_audio dw_hdmi_cec snd_soc_simple_card snd_soc_simple_card_utils snd_soc_rockchip_i2s rockchip_rga snd_soc_rockchip_pcm rockchipdrm videobuf2_dma_sg v4l2_mem2mem rtc_rk808 videobuf2_memops analogix_dp videobuf2_v4l2 videobuf2_common dw_hdmi dw_wdt cec rc_core videodev drm_kms_helper media drm rockchip_thermal rockchip_saradc realtek drm_panel_orientation_quirks syscopyarea sysfillrect sysimgblt fb_sys_fops dwmac_rk stmmac_platform stmmac pwm_bl squashfs loop crypto_user gpio_keys hid_kensington CPU: 5 PID: 957 Comm: nvim Not tainted 4.19.0-rc1-1-ARCH #1 Hardware name: Sapphire-RK3399 Board (DT) pstate: 00000005 (nzcv daif -PAN -UAO) pc : update_sit_entry+0x304/0x4b0 lr : update_sit_entry+0x108/0x4b0 sp : ffff00000ca13bd0 x29: ffff00000ca13bd0 x28: 000000000000003e x27: 0000000000000020 x26: 0000000000080000 x25: 0000000000000048 x24: ffff8000ebb85cf8 x23: 0000000000000253 x22: 00000000ffffffff x21: 00000000000535f2 x20: 00000000ffffffdf x19: ffff8000eb9e6800 x18: ffff8000eb9e6be8 x17: 0000000007ce6926 x16: 000000001c83ffa8 x15: 0000000000000000 x14: ffff8000f602df90 x13: 0000000000000006 x12: 0000000000000040 x11: 0000000000000228 x10: 0000000000000000 x9 : 0000000000000000 x8 : 0000000000000000 x7 : 00000000000535f2 x6 : ffff8000ebff3440 x5 : ffff8000ebff3440 x4 : ffff8000ebe3a6c8 x3 : 00000000ffffffff x2 : 0000000000000020 x1 : 0000000000000000 x0 : ffff8000eb9e5800 Process nvim (pid: 957, stack limit = 0x0000000063a78320) Call trace: update_sit_entry+0x304/0x4b0 f2fs_invalidate_blocks+0x98/0x140 truncate_node+0x90/0x400 f2fs_remove_inode_page+0xe8/0x340 f2fs_evict_inode+0x2b0/0x408 evict+0xe0/0x1e0 iput+0x160/0x260 do_unlinkat+0x214/0x298 __arm64_sys_unlinkat+0x3c/0x68 el0_svc_handler+0x94/0x118 el0_svc+0x8/0xc Code: f9400800 b9488400 36080140 f9400f01 (387c4820) ---[ end trace a0f21a307118c477 ]--- The reason is it is possible to enable discard flag on block queue via UDEV, but during mount, f2fs will initialize se->discard_map only if this flag is set, once the flag is set after mount, f2fs may dereference NULL pointer on se->discard_map. So this patch does below changes to fix this issue: - initialize and update se->discard_map all the time. - don't clear DISCARD option if device has no QUEUE_FLAG_DISCARD flag during mount. - don't issue small discard on zoned block device. - introduce some functions to enhance the readability. Signed-off-by: Chao Yu <yuchao0@huawei.com> Tested-by: Vicente Bergas <vicencb@gmail.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-09-03 19:52:17 +00:00
static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi)
{
int i;
if (!f2fs_is_multi_device(sbi))
return f2fs_bdev_support_discard(sbi->sb->s_bdev);
for (i = 0; i < sbi->s_ndevs; i++)
if (f2fs_bdev_support_discard(FDEV(i).bdev))
return true;
return false;
f2fs: fix to avoid NULL pointer dereference on se->discard_map https://bugzilla.kernel.org/show_bug.cgi?id=200951 These is a NULL pointer dereference issue reported in bugzilla: Hi, in the setup there is a SATA SSD connected to a SATA-to-USB bridge. The disc is "Samsung SSD 850 PRO 256G" which supports TRIM. There are four partitions: sda1: FAT /boot sda2: F2FS / sda3: F2FS /home sda4: F2FS The bridge is ASMT1153e which uses the "uas" driver. There is no TRIM pass-through, so, when mounting it reports: mounting with "discard" option, but the device does not support discard The USB host is USB3.0 and UASP capable. It is the one on RK3399. Given this everything works fine, except there is no TRIM support. In order to enable TRIM a new UDEV rule is added [1]: /etc/udev/rules.d/10-sata-bridge-trim.rules: ACTION=="add|change", ATTRS{idVendor}=="174c", ATTRS{idProduct}=="55aa", SUBSYSTEM=="scsi_disk", ATTR{provisioning_mode}="unmap" After reboot any F2FS write hangs forever and dmesg reports: Unable to handle kernel NULL pointer dereference Also tested on a x86_64 system: works fine even with TRIM enabled. same disc same bridge different usb host controller different cpu architecture not root filesystem Regards, Vicenç. [1] Post #5 in https://bbs.archlinux.org/viewtopic.php?id=236280 Unable to handle kernel NULL pointer dereference at virtual address 000000000000003e Mem abort info: ESR = 0x96000004 Exception class = DABT (current EL), IL = 32 bits SET = 0, FnV = 0 EA = 0, S1PTW = 0 Data abort info: ISV = 0, ISS = 0x00000004 CM = 0, WnR = 0 user pgtable: 4k pages, 48-bit VAs, pgdp = 00000000626e3122 [000000000000003e] pgd=0000000000000000 Internal error: Oops: 96000004 [#1] SMP Modules linked in: overlay snd_soc_hdmi_codec rc_cec dw_hdmi_i2s_audio dw_hdmi_cec snd_soc_simple_card snd_soc_simple_card_utils snd_soc_rockchip_i2s rockchip_rga snd_soc_rockchip_pcm rockchipdrm videobuf2_dma_sg v4l2_mem2mem rtc_rk808 videobuf2_memops analogix_dp videobuf2_v4l2 videobuf2_common dw_hdmi dw_wdt cec rc_core videodev drm_kms_helper media drm rockchip_thermal rockchip_saradc realtek drm_panel_orientation_quirks syscopyarea sysfillrect sysimgblt fb_sys_fops dwmac_rk stmmac_platform stmmac pwm_bl squashfs loop crypto_user gpio_keys hid_kensington CPU: 5 PID: 957 Comm: nvim Not tainted 4.19.0-rc1-1-ARCH #1 Hardware name: Sapphire-RK3399 Board (DT) pstate: 00000005 (nzcv daif -PAN -UAO) pc : update_sit_entry+0x304/0x4b0 lr : update_sit_entry+0x108/0x4b0 sp : ffff00000ca13bd0 x29: ffff00000ca13bd0 x28: 000000000000003e x27: 0000000000000020 x26: 0000000000080000 x25: 0000000000000048 x24: ffff8000ebb85cf8 x23: 0000000000000253 x22: 00000000ffffffff x21: 00000000000535f2 x20: 00000000ffffffdf x19: ffff8000eb9e6800 x18: ffff8000eb9e6be8 x17: 0000000007ce6926 x16: 000000001c83ffa8 x15: 0000000000000000 x14: ffff8000f602df90 x13: 0000000000000006 x12: 0000000000000040 x11: 0000000000000228 x10: 0000000000000000 x9 : 0000000000000000 x8 : 0000000000000000 x7 : 00000000000535f2 x6 : ffff8000ebff3440 x5 : ffff8000ebff3440 x4 : ffff8000ebe3a6c8 x3 : 00000000ffffffff x2 : 0000000000000020 x1 : 0000000000000000 x0 : ffff8000eb9e5800 Process nvim (pid: 957, stack limit = 0x0000000063a78320) Call trace: update_sit_entry+0x304/0x4b0 f2fs_invalidate_blocks+0x98/0x140 truncate_node+0x90/0x400 f2fs_remove_inode_page+0xe8/0x340 f2fs_evict_inode+0x2b0/0x408 evict+0xe0/0x1e0 iput+0x160/0x260 do_unlinkat+0x214/0x298 __arm64_sys_unlinkat+0x3c/0x68 el0_svc_handler+0x94/0x118 el0_svc+0x8/0xc Code: f9400800 b9488400 36080140 f9400f01 (387c4820) ---[ end trace a0f21a307118c477 ]--- The reason is it is possible to enable discard flag on block queue via UDEV, but during mount, f2fs will initialize se->discard_map only if this flag is set, once the flag is set after mount, f2fs may dereference NULL pointer on se->discard_map. So this patch does below changes to fix this issue: - initialize and update se->discard_map all the time. - don't clear DISCARD option if device has no QUEUE_FLAG_DISCARD flag during mount. - don't issue small discard on zoned block device. - introduce some functions to enhance the readability. Signed-off-by: Chao Yu <yuchao0@huawei.com> Tested-by: Vicente Bergas <vicencb@gmail.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-09-03 19:52:17 +00:00
}
static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi)
{
return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) ||
f2fs_hw_should_discard(sbi);
}
static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi)
{
int i;
if (!f2fs_is_multi_device(sbi))
return bdev_read_only(sbi->sb->s_bdev);
for (i = 0; i < sbi->s_ndevs; i++)
if (bdev_read_only(FDEV(i).bdev))
return true;
return false;
}
static inline bool f2fs_dev_is_readonly(struct f2fs_sb_info *sbi)
{
return f2fs_sb_has_readonly(sbi) || f2fs_hw_is_readonly(sbi);
}
static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi)
{
return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS;
}
static inline bool f2fs_valid_pinned_area(struct f2fs_sb_info *sbi,
block_t blkaddr)
{
if (f2fs_sb_has_blkzoned(sbi)) {
int devi = f2fs_target_device_index(sbi, blkaddr);
return !bdev_is_zoned(FDEV(devi).bdev);
}
return true;
}
static inline bool f2fs_low_mem_mode(struct f2fs_sb_info *sbi)
{
return F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW;
}
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>
2019-11-01 10:07:14 +00:00
static inline bool f2fs_may_compress(struct inode *inode)
{
if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) ||
f2fs_is_atomic_file(inode) || f2fs_has_inline_data(inode) ||
f2fs_is_mmap_file(inode))
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>
2019-11-01 10:07:14 +00:00
return false;
return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode);
}
static inline void f2fs_i_compr_blocks_update(struct inode *inode,
u64 blocks, bool add)
{
struct f2fs_inode_info *fi = F2FS_I(inode);
int diff = fi->i_cluster_size - blocks;
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>
2019-11-01 10:07:14 +00:00
/* don't update i_compr_blocks if saved blocks were released */
if (!add && !atomic_read(&fi->i_compr_blocks))
return;
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>
2019-11-01 10:07:14 +00:00
if (add) {
atomic_add(diff, &fi->i_compr_blocks);
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>
2019-11-01 10:07:14 +00:00
stat_add_compr_blocks(inode, diff);
} else {
atomic_sub(diff, &fi->i_compr_blocks);
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>
2019-11-01 10:07:14 +00:00
stat_sub_compr_blocks(inode, diff);
}
f2fs_mark_inode_dirty_sync(inode, true);
}
static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi,
int flag)
{
if (!f2fs_is_multi_device(sbi))
return false;
if (flag != F2FS_GET_BLOCK_DIO)
return false;
return sbi->aligned_blksize;
}
static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx)
{
return fsverity_active(inode) &&
idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE);
}
#ifdef CONFIG_F2FS_FAULT_INJECTION
extern int f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned long rate,
unsigned long type);
#else
static inline int f2fs_build_fault_attr(struct f2fs_sb_info *sbi,
unsigned long rate, unsigned long type)
{
return 0;
}
#endif
static inline bool is_journalled_quota(struct f2fs_sb_info *sbi)
{
#ifdef CONFIG_QUOTA
if (f2fs_sb_has_quota_ino(sbi))
return true;
if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
return true;
#endif
return false;
}
f2fs: introduce discard_unit mount option As James Z reported in bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=213877 [1.] One-line summary of the problem: Mount multiple SMR block devices exceed certain number cause system non-response [2.] Full description of the problem/report: Created some F2FS on SMR devices (mkfs.f2fs -m), then mounted in sequence. Each device is the same Model: HGST HSH721414AL (Size 14TB). Empirically, found that when the amount of SMR device * 1.5Gb > System RAM, the system ran out of memory and hung. No dmesg output. For example, 24 SMR Disk need 24*1.5GB = 36GB. A system with 32G RAM can only mount 21 devices, the 22nd device will be a reproducible cause of system hang. The number of SMR devices with other FS mounted on this system does not interfere with the result above. [3.] Keywords (i.e., modules, networking, kernel): F2FS, SMR, Memory [4.] Kernel information [4.1.] Kernel version (uname -a): Linux 5.13.4-200.fc34.x86_64 #1 SMP Tue Jul 20 20:27:29 UTC 2021 x86_64 x86_64 x86_64 GNU/Linux [4.2.] Kernel .config file: Default Fedora 34 with f2fs-tools-1.14.0-2.fc34.x86_64 [5.] Most recent kernel version which did not have the bug: None [6.] Output of Oops.. message (if applicable) with symbolic information resolved (see Documentation/admin-guide/oops-tracing.rst) None [7.] A small shell script or example program which triggers the problem (if possible) mount /dev/sdX /mnt/0X [8.] Memory consumption With 24 * 14T SMR Block device with F2FS free -g total used free shared buff/cache available Mem: 46 36 0 0 10 10 Swap: 0 0 0 With 3 * 14T SMR Block device with F2FS free -g total used free shared buff/cache available Mem: 7 5 0 0 1 1 Swap: 7 0 7 The root cause is, there are three bitmaps: - cur_valid_map - ckpt_valid_map - discard_map and each of them will cost ~500MB memory, {cur, ckpt}_valid_map are necessary, but discard_map is optional, since this bitmap will only be useful in mountpoint that small discard is enabled. For a blkzoned device such as SMR or ZNS devices, f2fs will only issue discard for a section(zone) when all blocks of that section are invalid, so, for such device, we don't need small discard functionality at all. This patch introduces a new mountoption "discard_unit=block|segment| section" to support issuing discard with different basic unit which is aligned to block, segment or section, so that user can specify "discard_unit=segment" or "discard_unit=section" to disable small discard functionality. Note that this mount option can not be changed by remount() due to related metadata need to be initialized during mount(). In order to save memory, let's use "discard_unit=section" for blkzoned device by default. Signed-off-by: Chao Yu <chao@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2021-08-03 00:15:43 +00:00
static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi)
{
return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK;
}
static inline void f2fs_io_schedule_timeout(long timeout)
{
set_current_state(TASK_UNINTERRUPTIBLE);
io_schedule_timeout(timeout);
}
f2fs: avoid infinite loop to flush node pages xfstests/generic/475 can give EIO all the time which give an infinite loop to flush node page like below. Let's avoid it. [16418.518551] Call Trace: [16418.518553] ? dm_submit_bio+0x48/0x400 [16418.518574] ? submit_bio_checks+0x1ac/0x5a0 [16418.525207] __submit_bio+0x1a9/0x230 [16418.525210] ? kmem_cache_alloc+0x29e/0x3c0 [16418.525223] submit_bio_noacct+0xa8/0x2b0 [16418.525226] submit_bio+0x4d/0x130 [16418.525238] __submit_bio+0x49/0x310 [f2fs] [16418.525339] ? bio_add_page+0x6a/0x90 [16418.525344] f2fs_submit_page_bio+0x134/0x1f0 [f2fs] [16418.525365] read_node_page+0x125/0x1b0 [f2fs] [16418.525388] __get_node_page.part.0+0x58/0x3f0 [f2fs] [16418.525409] __get_node_page+0x2f/0x60 [f2fs] [16418.525431] f2fs_get_dnode_of_data+0x423/0x860 [f2fs] [16418.525452] ? asm_sysvec_apic_timer_interrupt+0x12/0x20 [16418.525458] ? __mod_memcg_state.part.0+0x2a/0x30 [16418.525465] ? __mod_memcg_lruvec_state+0x27/0x40 [16418.525467] ? __xa_set_mark+0x57/0x70 [16418.525472] f2fs_do_write_data_page+0x10e/0x7b0 [f2fs] [16418.525493] f2fs_write_single_data_page+0x555/0x830 [f2fs] [16418.525514] ? sysvec_apic_timer_interrupt+0x4e/0x90 [16418.525518] ? asm_sysvec_apic_timer_interrupt+0x12/0x20 [16418.525523] f2fs_write_cache_pages+0x303/0x880 [f2fs] [16418.525545] ? blk_flush_plug_list+0x47/0x100 [16418.525548] f2fs_write_data_pages+0xfd/0x320 [f2fs] [16418.525569] do_writepages+0xd5/0x210 [16418.525648] filemap_fdatawrite_wbc+0x7d/0xc0 [16418.525655] filemap_fdatawrite+0x50/0x70 [16418.525658] f2fs_sync_dirty_inodes+0xa4/0x230 [f2fs] [16418.525679] f2fs_write_checkpoint+0x16d/0x1720 [f2fs] [16418.525699] ? ttwu_do_wakeup+0x1c/0x160 [16418.525709] ? ttwu_do_activate+0x6d/0xd0 [16418.525711] ? __wait_for_common+0x11d/0x150 [16418.525715] kill_f2fs_super+0xca/0x100 [f2fs] [16418.525733] deactivate_locked_super+0x3b/0xb0 [16418.525739] deactivate_super+0x40/0x50 [16418.525741] cleanup_mnt+0x139/0x190 [16418.525747] __cleanup_mnt+0x12/0x20 [16418.525749] task_work_run+0x6d/0xa0 [16418.525765] exit_to_user_mode_prepare+0x1ad/0x1b0 [16418.525771] syscall_exit_to_user_mode+0x27/0x50 [16418.525774] do_syscall_64+0x48/0xc0 [16418.525776] entry_SYSCALL_64_after_hwframe+0x44/0xae Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2022-03-29 23:25:54 +00:00
static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, pgoff_t ofs,
enum page_type type)
{
if (unlikely(f2fs_cp_error(sbi)))
return;
if (ofs == sbi->page_eio_ofs[type]) {
if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO)
set_ckpt_flags(sbi, CP_ERROR_FLAG);
} else {
sbi->page_eio_ofs[type] = ofs;
sbi->page_eio_cnt[type] = 0;
}
}
static inline bool f2fs_is_readonly(struct f2fs_sb_info *sbi)
{
return f2fs_sb_has_readonly(sbi) || f2fs_readonly(sbi->sb);
}
static inline void f2fs_truncate_meta_inode_pages(struct f2fs_sb_info *sbi,
block_t blkaddr, unsigned int cnt)
{
bool need_submit = false;
int i = 0;
do {
struct page *page;
page = find_get_page(META_MAPPING(sbi), blkaddr + i);
if (page) {
if (folio_test_writeback(page_folio(page)))
need_submit = true;
f2fs_put_page(page, 0);
}
} while (++i < cnt && !need_submit);
if (need_submit)
f2fs_submit_merged_write_cond(sbi, sbi->meta_inode,
NULL, 0, DATA);
truncate_inode_pages_range(META_MAPPING(sbi),
F2FS_BLK_TO_BYTES((loff_t)blkaddr),
F2FS_BLK_END_BYTES((loff_t)(blkaddr + cnt - 1)));
}
static inline void f2fs_invalidate_internal_cache(struct f2fs_sb_info *sbi,
block_t blkaddr)
{
f2fs_truncate_meta_inode_pages(sbi, blkaddr, 1);
f2fs_invalidate_compress_page(sbi, blkaddr);
}
#define EFSBADCRC EBADMSG /* Bad CRC detected */
#define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
#endif /* _LINUX_F2FS_H */