cp_payload is set differently [1] when extended node bitmap feature is
enabled. Commit b79c3ba4ea broke fsck on f2fs file systems created on
2+ TB device with extended node bitmap feature enabled.
As the sanity check is for checking overflows, fix this to assume the max
possible cp_payload size under the extended node bitmap.
Link: https://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs-tools.git/tree/mkfs/f2fs_format.c?h=v1.15.0#n372 [1]
Fixes: b79c3ba4ea ("fsck.f2fs: sanity check cp_payload before reading checkpoint")
Reported-by: Alexander Koskovich <akoskovich@pm.me>
Signed-off-by: Juhyung Park <qkrwngud825@gmail.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Keep sanity_check_raw_super flow in fsck.f2fs being consistent w/
sanity_check_raw_super() in f2fs kernel module, otherwise, it outputs
different error information when mounting/fscking the same image which
contains corrupted superblock.
Signed-off-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
There is no need to recalculate ADDRS_PER_INODE and ADDRS_PER_BLOCK,
especially in a for loop.
Signed-off-by: Sheng Yong <shengyong@oppo.com>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
If i_namelen is corrupted, there may be an overflow when doing memcpy.
Signed-off-by: Sheng Yong <shengyong@oppo.com>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Option "-g" for dump.f2fs is unavailable and not used in dump.f2fs.
Signed-off-by: Sheng Yong <shengyong@oppo.com>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
As f2fs becomes more resilient for GCs, let's give the marginal overprovision
space back to user.
Fix an issue where reserved_space > ovp_space, reported by Shinichiro.
Signed-off-by: Shinichiro Kawasaki <shinichiro.kawasaki@wdc.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
In auto/preen mode, let's trigger repairing if filesystem has
inconsistent errors.
Signed-off-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
In auto/preen mode, let's trigger repairing if filesystem was forced
to stop.
Signed-off-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Change to calculate and export max image size with valid blocks
used of image rather than last used block of image.
[FSCK] Max image size: xx MB, Free space: xx MB
Signed-off-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Update allocation policy for ro feature:
- hot_data: allocating blocks by LBA ascending order
- hot_node: allocating blocks by LBA descending order
Signed-off-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Make the argument list match the format string. Use PRIu64 for uint64_t
and %zu for size_t.
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
We now use walltime for monthly period check. However the walltime is:
* unstable(timestamp register reset) and settable(modified by user)
* unreasonable(e.g: device power-off for one month, no data changed)
When the walltime changes to the past before one month or the future
after one month, the period check in next fsck will fail to skip or
start a full scan. So, let's use the elapsed_time in checkpoint as
current time for period check.
Signed-off-by: Weichao Guo <guoweichao@oppo.com>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Since the size of struct f2fs_summary_block equals F2FS_BLKSIZE, use
F2FS_BLKSIZE instead of PAGE_CACHE_SIZE as the size of struct
f2fs_summary_block.
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Use F2FS_BLKSIZE instead of PAGE_SIZE for dev_read_block() buffers since
dev_read_block() reads F2FS_BLKSIZE bytes.
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Make.f2fs supports manually specifying overprovision, and we expect
resize.f2fs to support it as well.
This change add a new '-o' option to manually specify overprovision,
and fix to check free space before grow. Otherwise, after grow,
kernel may report below error message when we mount the image if -o
parameter is specified during resize:
F2FS-fs (loop0): invalid crc_offset: 0
F2FS-fs (loop0): Wrong valid_user_blocks: 16404, user_block_count: 13312
F2FS-fs (loop0): Failed to get valid F2FS checkpoint
mount(2) system call failed: Structure needs cleaning.
Signed-off-by: liuchao12 <liuchao12@xiaomi.com>
Signed-off-by: qixiaoyu1 <qixiaoyu1@xiaomi.com>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Change-Id: I2ad0ecc5546a1712b7aa36230e91c3e0b1beb7e0
The kernel enforced not to have compression and inline_data bits
together, but old kernel created files with them. Let's disable
compression in that case.
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Usage:
dump.f2fs -I [inode nid] /dev/sda
This feature can be useful for some bugs caused by system crash.
We not only need dump current valid node page, but alse the history
data in disk, which can give some clues for status change of the inode.
Signed-off-by: Yufen Yu <yuyufen@huawei.com>
Reviewed-by: Chao Yu <chao@kernel.org>
[Jaegeuk Kim: fix build error]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
commit 299c0b566ad71f0a9ca0cc359d1d174de9068294
category: bugfix
issue: #I6VAS0
CVE: NA
Signed-off-by: DongSenhao <dongsenhao2@huawei.com>
---------------------------------------
The new fsck progress should not be used for file_map.
Reported-by: Michael Benedict <michaelbenedicttjahya@gmail.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Signed-off-by: dongsenhao <dongsenhao2@huawei.com>
commit 0b9b89fda2e9fbaf17a11fa15332e7ae95cf5261
category: bugfix
issue: #I6VAS0
CVE: NA
Signed-off-by: DongSenhao <dongsenhao2@huawei.com>
---------------------------------------
Compressed cluster content includes compress cluster header, so
calculation of dstlen parameter in LZ4_compress_fast_extState()
should consider this design, otherwise compressed data may overflow
destination buffer during compression.
Signed-off-by: Chao Yu <chao.yu@oppo.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Signed-off-by: dongsenhao <dongsenhao2@huawei.com>
commit 7a1206adf33595a50b669d4027b72ad0786fb12f
category: bugfix
issue: #I6VAS0
CVE: NA
Signed-off-by: DongSenhao <dongsenhao2@huawei.com>
---------------------------------------
Just like in the Linux kernel source code, annotate switch/case
fallthrough.
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Signed-off-by: dongsenhao <dongsenhao2@huawei.com>
commit fdff1ab3157c2787d908e013b9d42db84184e196
category: bugfix
issue: #I6VAS0
CVE: NA
Signed-off-by: DongSenhao <dongsenhao2@huawei.com>
---------------------------------------
Since 'remained_blkentries' is unsigned, the assert statement that
verifies whether that variable is positive will never fail. Hence
remove that assert statement.
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Signed-off-by: dongsenhao <dongsenhao2@huawei.com>
