>> Fixes: 8a3f5711ad74 ("exfat: reduce FAT chain traversal")
> This issue caused by this patch ? If yes, Could you elaborate how this
patch make this issue ?
Yes.
This issue caused by the change in this patch.
- return dentry;
+ p_dir->dir = exfat_sector_to_cluster(sbi, es->bh[0]->b_blocknr);
+ p_dir->size -= dentry / sbi->dentries_per_clu;
+
+ return dentry & (sbi->dentries_per_clu - 1);
}
'dentry' is -EIO or -ENOMEM when reading directory entries fails.
"dentry & (sbi->dentries_per_clu - 1)" makes the return value a positive value,
so that exfat_add_entry() always thinks that the directory entry is read successfully.
> I can not reproduce it using C-reproducer. Have you reproduced it ?
This issue occurs when reading directory entries fails(this can be confirmed by
https://syzkaller.appspot.com/text?tag=Patch&x=1068bd30580000).
Reproducing it requires a disk with bad blocks, I can not reproduce it too.
