Re: [f2fs-dev] [PATCH v2] f2fs: initialize ino_entry_info before checkpoint load

On 6/21/26 16:18, Deepanshu Kartikey wrote:
> On Sun, May 10, 2026 at 9:53 AM Deepanshu Kartikey
> <[email protected]> wrote:
>>
>> When f2fs_get_valid_checkpoint() fails during mount (e.g. due to an
>> invalid checkpoint CRC on a malformed image), f2fs_fill_super() takes
>> an error path that eventually calls iput() on the root inode. This
>> invokes f2fs_drop_inode() -> f2fs_exist_written_data(), which acquires
>> sbi->im[]->ino_lock. However, f2fs_init_ino_entry_info() has not run
>> yet at this point, so the spinlock is uninitialized and lockdep
>> complains:
>>
>>    F2FS-fs (loop0): invalid crc value
>>    F2FS-fs (loop0): Failed to get valid F2FS checkpoint
>>    INFO: trying to register non-static key.
>>    The code is fine but needs lockdep annotation, or maybe
>>    you didn't initialize this object before use?
>>    ...
>>     f2fs_exist_written_data+0x53/0x90 fs/f2fs/checkpoint.c:787
>>     f2fs_drop_inode+0xda/0xbf0 fs/f2fs/super.c:1852
>>     iput+0x651/0xe80 fs/inode.c:2009
>>     f2fs_fill_super+0x6047/0x7850 fs/f2fs/super.c:5461
>>
>> Move f2fs_init_ino_entry_info() to before f2fs_get_valid_checkpoint()
>> so that sbi->im[] is always fully initialized before any error path
>> can trigger iput() -> f2fs_drop_inode(). The init function only
>> depends on raw superblock fields (BLKS_PER_SEG, F2FS_CP_PACKS,
>> NR_CURSEG_PERSIST_TYPE, __cp_payload), which are populated well
>> before checkpoint load, so the move is safe.
>>
>> Fixes: 3063c80776e3 ("f2fs: another way to set large folio by remembering inode number")
>> Cc: [email protected]
>> Reported-by: [email protected]
>> Closes: https://syzkaller.appspot.com/bug?extid=eec8f2693d71386bd600
>> Tested-by: [email protected]
>> Signed-off-by: Deepanshu Kartikey <[email protected]>
>> ---
>> Changes in v2:
>>   - Add Fixes: tag (suggested by Chao Yu)
>>   - Add Cc: [email protected]
>> ---
>>   fs/f2fs/super.c | 9 +++++++--
>>   1 file changed, 7 insertions(+), 2 deletions(-)
>>
>> diff --git a/fs/f2fs/super.c b/fs/f2fs/super.c
>> index c6afdbd6e1cd..6a231a5b0d62 100644
>> --- a/fs/f2fs/super.c
>> +++ b/fs/f2fs/super.c
>> @@ -5140,6 +5140,13 @@ static int f2fs_fill_super(struct super_block *sb, struct fs_context *fc)
>>                  goto free_page_array_cache;
>>          }
>>
>> +       /*
>> +        * Initialize ino entry info early so f2fs_drop_inode ->
>> +        * f2fs_exist_written_data can safely take im->ino_lock if mount
>> +        * fails after this point and triggers iput on cleanup.
>> +        */
>> +       f2fs_init_ino_entry_info(sbi);
>> +
>>          err = f2fs_get_valid_checkpoint(sbi);
>>          if (err) {
>>                  f2fs_err(sbi, "Failed to get valid F2FS checkpoint");
>> @@ -5184,8 +5191,6 @@ static int f2fs_fill_super(struct super_block *sb, struct fs_context *fc)
>>
>>          f2fs_init_extent_cache_info(sbi);
>>
>> -       f2fs_init_ino_entry_info(sbi);
>> -
>>          f2fs_init_fsync_node_info(sbi);
>>
>>          /* setup checkpoint request control and start checkpoint issue thread */
>> --
>> 2.43.0
>>
> 
> Hi Chao,
> 
> Please let me know the status of this patch. This is already "Reviewed-by:" you.

https://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs.git/commit/?h=dev-test&id=065a6f8cd23a9297b543dcec913feb3cb787a25e

It was merged, however, I think it should be reverted because commit 3063c80776e3
("f2fs: another way to set large folio by remembering inode number") was removed
from dev branch, we won't call f2fs_exist_written_data() from f2fs_drop_inode().

Thanks,

> 
> Please let me know if anything needed from my side.
> 
> Thanks
> 
> Deepanshu Kartikey



_______________________________________________
Linux-f2fs-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel