- 02 Feb, 2023 12 commits
-
-
Yangtao Li authored
Add iotype sanity check to avoid potential memory corruption. This is to fix the compile error below: fs/f2fs/iostat.c:231 __update_iostat_latency() error: buffer overflow 'io_lat->peak_lat[type]' 3 <= 3 vim +228 fs/f2fs/iostat.c 211 static inline void __update_iostat_latency(struct bio_iostat_ctx *iostat_ctx, 212 enum iostat_lat_type type) 213 { 214 unsigned long ts_diff; 215 unsigned int page_type = iostat_ctx->type; 216 struct f2fs_sb_info *sbi = iostat_ctx->sbi; 217 struct iostat_lat_info *io_lat = sbi->iostat_io_lat; 218 unsigned long flags; 219 220 if (!sbi->iostat_enable) 221 return; 222 223 ts_diff = jiffies - iostat_ctx->submit_ts; 224 if (page_type >= META_FLUSH) ^^^^^^^^^^ 225 page_type = META; 226 227 spin_lock_irqsave(&sbi->iostat_lat_lock, flags); @228 io_lat->sum_lat[type][page_type] += ts_diff; ^^^^^^^^^ Mixup between META_FLUSH and NR_PAGE_TYPE leads to memory corruption. Fixes: a4b68176 ("f2fs: introduce periodic iostat io latency traces") Reported-by:kernel test robot <lkp@intel.com> Reported-by:
Dan Carpenter <error27@gmail.com> Suggested-by:
Chao Yu <chao@kernel.org> Suggested-by:
Jaegeuk Kim <jaegeuk@kernel.org> Signed-off-by:
Yangtao Li <frank.li@vivo.com> Reviewed-by:
-
Jaegeuk Kim authored
If the storage gives a corrupted node block due to short power failure and reset, f2fs stops the entire operations by setting the checkpoint failure flag. Let's give more chances to live by re-issuing IOs for a while in such critical path. Cc: stable@vger.kernel.org Suggested-by:
Randall Huang <huangrandall@google.com> Suggested-by:
-
Chao Yu authored
Otherwise, 32-bits binary call ioctl(F2FS_IOC_START_ATOMIC_REPLACE) will fail in 64-bits kernel. Fixes: 41e8f85a ("f2fs: introduce F2FS_IOC_START_ATOMIC_REPLACE") Signed-off-by:
-
Chao Yu authored
.i_compress_level was introduced by commit 3fde13f8 ("f2fs: compress: support compress level"), but never be used. This patch updates as below: - load high 8-bits of on-disk .i_compress_flag to in-memory .i_compress_level - load low 8-bits of on-disk .i_compress_flag to in-memory .i_compress_flag - change type of in-memory .i_compress_flag from unsigned short to unsigned char. w/ above changes, we can avoid unneeded bit shift whenever during .init_compress_ctx(), and shrink size of struct f2fs_inode_info. Signed-off-by:
-
Chao Yu authored
This patch tries to use bitfield in struct f2fs_io_info to improve memory usage. struct f2fs_io_info { ... 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 retry:1; /* need to reallocate block address */ unsigned int encrypted:1; /* indicate file is encrypted */ unsigned int post_read:1; /* require post read */ ... }; After this patch, size of struct f2fs_io_info reduces from 136 to 120. [Nathan: fix a compile warning (single-bit-bitfield-constant-conversion)] Signed-off-by:Nathan Chancellor <nathan@kernel.org> Signed-off-by:
-
Christoph Hellwig authored
Factor the logic to log a path for reads and writs into a helper shared between the read_iter and write_iter methods. Signed-off-by:
Christoph Hellwig <hch@lst.de> Signed-off-by:
-
Christoph Hellwig authored
Just open code the logic in the only caller, where it is more obvious. Signed-off-by:
-
Christoph Hellwig authored
Remove __refresh_next_blkoff by opencoding the SSR vs LFS segment check in the only caller, and then add helpers for SSR block selection and blkoff randomization instead. Signed-off-by:
-
Christoph Hellwig authored
Just fold this trivial wrapper into the only caller. Signed-off-by:
-
Christoph Hellwig authored
Simplify the check whether to allocate a new segment or reuse an open one. Signed-off-by:
-
Christoph Hellwig authored
Add a helper to return the valid blocks on log and SSR segments, and replace the last two uses of curseg_blkoff with it. Signed-off-by:
-
Christoph Hellwig authored
For each loop add a local curseg_info pointer insted of looking it up for each of the three fields. Signed-off-by:
-
- 31 Jan, 2023 11 commits
-
-
Christoph Hellwig authored
This function just assigns a summary entry. This can be done entirely typesafe with an open code struct assignment that relies on array indexing. Signed-off-by:
-
Chao Yu authored
Commit 7a10f017 ("f2fs: don't give partially written atomic data from process crash") attempted to drop atomic write data after process crash, however, f2fs_abort_atomic_write() may be called from noncrash case, fix it by adding missed PF_EXITING check condition f2fs_file_flush(). - application crashs - do_exit - exit_signals -- sets PF_EXITING - exit_files - put_files_struct - close_files - filp_close - flush (f2fs_file_flush) - check atomic_write_task && PF_EXITING - f2fs_abort_atomic_write Fixes: 7a10f017 ("f2fs: don't give partially written atomic data from process crash") Signed-off-by:
-
Yangtao Li authored
Files created by truncate have a size but no blocks, so they can be allowed to set compression option. Fixes: e1e8debe ("f2fs: add F2FS_IOC_SET_COMPRESS_OPTION ioctl") Signed-off-by:
-
Eric Biggers authored
When converting an inline directory to a regular one, f2fs is leaking uninitialized memory to disk because it doesn't initialize the entire directory block. Fix this by zero-initializing the block. This bug was introduced by commit 4ec17d68 ("f2fs: avoid unneeded initializing when converting inline dentry"), which didn't consider the security implications of leaking uninitialized memory to disk. This was found by running xfstest generic/435 on a KMSAN-enabled kernel. Fixes: 4ec17d68 ("f2fs: avoid unneeded initializing when converting inline dentry") Cc: <stable@vger.kernel.org> # v4.3+ Signed-off-by:
Eric Biggers <ebiggers@google.com> Reviewed-by:
-
Alexander Potapenko authored
When aops->write_begin() does not initialize fsdata, KMSAN may report an error passing the latter to aops->write_end(). Fix this by unconditionally initializing fsdata. Suggested-by:
Eric Biggers <ebiggers@kernel.org> Fixes: 95ae251f ("f2fs: add fs-verity support") Signed-off-by:
Alexander Potapenko <glider@google.com> Reviewed-by:
-
Yangtao Li authored
hot_data_age_threshold is a non-zero positive number, and condition 2 includes condition 1, so there is no need to additionally judge whether t is 0. And let's remove it. Signed-off-by:
-
Yangtao Li authored
f2fs: return true if all cmd were issued or no cmd need to be issued for f2fs_issue_discard_timeout() f2fs_issue_discard_timeout() returns whether discard cmds are dropped, which does not match the meaning of the function. Let's change it to return whether all discard cmd are issued. After commit 4d674904 ("f2fs: Don't create discard thread when device doesn't support realtime discard"), f2fs_issue_discard_timeout() is alse called by f2fs_remount(). Since the comments of f2fs_issue_discard_timeout() doesn't make much sense, let's update it. Signed-off-by:
-
Yangtao Li authored
commit 3fde13f8 ("f2fs: compress: support compress level") introduce compress level, which macro(COMPRESS_LEVEL_OFFSET) is 8, But use wrong comment about compress level. Let's fix it. Signed-off-by:
-
Yangtao Li authored
Convert to show discard_unit only when has DISCARD opt. Signed-off-by:
-
Chao Yu authored
In do_read_inode(), sanity_check_inode() should be called after f2fs_init_read_extent_tree(), fix it. Fixes: 72840ccc ("f2fs: allocate the extent_cache by default") Signed-off-by:
-
Chao Yu authored
f2fs_rename() has checked CP_ERROR_FLAG, so remove redundant check in f2fs_create_whiteout(). Signed-off-by:
-
- 30 Jan, 2023 5 commits
-
-
Chao Yu authored
Otherwise, last .atomic_write_task will be remained in structure f2fs_inode_info, resulting in aborting atomic_write accidentally in race case. Meanwhile, clear original_i_size as well. Fixes: 7a10f017 ("f2fs: don't give partially written atomic data from process crash") Signed-off-by:
-
Chao Yu authored
Commit 3db1de0e ("f2fs: change the current atomic write way") removed old tracepoints, but it missed to add new one, this patch fixes to introduce trace_f2fs_replace_atomic_write_block to trace atomic_write commit flow. Fixes: 3db1de0e ("f2fs: change the current atomic write way") Signed-off-by:
-
Yangtao Li authored
The current discard_io_aware_gran is a fixed value, change it to be configurable through the sys node. Signed-off-by:
-
Yangtao Li authored
No need to initialize idx twice. BTW, remove the unnecessary cnt variable. Signed-off-by:
-
Yangtao Li authored
In this patch, it adds to account flush count. Signed-off-by:
-
- 11 Jan, 2023 10 commits
-
-
Yangtao Li authored
Previously, we supported to account iostat io_bytes, in this patch, it adds to account iostat count and avg_bytes: time: 1671648667 io_bytes count avg_bytes [WRITE] app buffered data: 31 2 15 Signed-off-by: -
Yangtao Li authored
discard_wake and gc_wake have only two values, 0 or 1. So there is no need to use int type to store them. BTW, move discard_wake to the end of the discard_cmd_control structure. Before: - sizeof(struct discard_cmd_control): 8392 After move: - sizeof(struct discard_cmd_control): 8384 Signed-off-by:
-
Yangtao Li authored
Commit 1cd2e6d5 ("f2fs: define MIN_DISCARD_GRANULARITY macro") introduce it, let's convert to use MIN_DISCARD_GRANULARITY macro. Signed-off-by:
-
Yangtao Li authored
There is no need to additionally use f2fs_show_injection_info() to output information. Concatenate time_to_inject() and __time_to_inject() via a macro. In the new __time_to_inject() function, pass in the caller function name and parent function. In this way, we no longer need the f2fs_show_injection_info() function, and let's remove it. Suggested-by:
-
Yangtao Li authored
For example, f2fs_collapse_range(), f2fs_collapse_range(), f2fs_insert_range(), the functions used in f2fs_fallocate() are all prefixed with f2fs_, so let's keep the name consistent. Signed-off-by:
-
Yangtao Li authored
Just cleanup. Signed-off-by:
-
Yangtao Li authored
f2fs_init_compress_mempool() only initializes the memory pool during the f2fs module init phase. Let's mark it as __init like any other function. Signed-off-by:
-
Yangtao Li authored
The current logic, regardless of whether CONFIG_BLK_DEV_ZONED is enabled or not, will judge whether discard_unit is SECTION, when f2fs_sb_has_blkzoned. In fact, when CONFIG_BLK_DEV_ZONED is not enabled, this judgment is a path that will never be accessed. At this time, -EINVAL will be returned in the parse_options function, accompanied by the message "Zoned block device support is not enabled". Let's wrap this discard_unit judgment with CONFIG_BLK_DEV_ZONED. Signed-off-by:
-
Zhang Qilong authored
We can start freeing cluster page(s) from which compression is not used. It will get better performance. Signed-off-by:
Zhang Qilong <zhangqilong3@huawei.com> Signed-off-by:
-
Yangtao Li authored
>From now on, f2fs also has its own patchwork link, thanks to Jaegeuk for starting this tool! Let's update it to f2fs entry. Signed-off-by:
-
- 06 Jan, 2023 2 commits
-
-
Yuwei Guan authored
Any of the following scenarios will send more than the number of max_requests at a time, which will not meet the design of the max_requests limit. - Set max_ordered_discard larger than discard_granularity from userspace. - It is a small size device, discard_granularity can be tuned to 1 in f2fs_tuning_parameters(). We need to deliver the accumulated @issued to __issue_discard_cmd_orderly() to meet the max_requests limit. BTW, convert the parameter type of @issued in __submit_discard_cmd(). Signed-off-by:
Yuwei Guan <Yuwei.Guan@zeekrlife.com> Cc: Bagas Sanjaya <bagasdotme@gmail.com> Reviewed-by:
-
Chao Yu authored
After below changes: commit 14db0b3c ("fscrypt: stop using PG_error to track error status") commit 98dc08ba ("fsverity: stop using PG_error to track error status") There is no place in f2fs we will set PG_error flag in page, let's remove other PG_error usage in f2fs, as a step towards freeing the PG_error flag for other uses. Cc: Eric Biggers <ebiggers@kernel.org> Signed-off-by:
-
