fs: sysv: check sbi->s_firstdatazone in complete_read_super
authorLiu Shixin <liushixin2@huawei.com>
Fri, 29 Apr 2022 21:38:04 +0000 (14:38 -0700)
committerakpm <akpm@linux-foundation.org>
Fri, 29 Apr 2022 21:38:04 +0000 (14:38 -0700)
sbi->s_firstinodezone is initialized to 2 and sbi->s_firstdatazone is read
from sbd.  There's no guarantee that sbi->s_firstdatazone must bigger than
sbi->s_firstinodezone.  If sbi->s_firstdatazone less than 2, the
filesystem can still be mounted unexpetly.  At this point, sbi->s_ninodes
flip to very large value and this filesystem is broken.  We can observe
this by executing 'df' command.  When we execute, we will get an error
message:

"sysv_count_free_inodes: unable to read inode table"

Link: https://lkml.kernel.org/r/20220330104215.530223-1-liushixin2@huawei.com
Signed-off-by: Liu Shixin <liushixin2@huawei.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
fs/sysv/super.c

index d1def07..3365a30 100644 (file)
@@ -312,7 +312,9 @@ static int complete_read_super(struct super_block *sb, int silent, int size)
        sbi->s_firstinodezone = 2;
 
        flavour_setup[sbi->s_type](sbi, &sb->s_max_links);
-       
+       if (sbi->s_firstdatazone < sbi->s_firstinodezone)
+               return 0;
+
        sbi->s_ndatazones = sbi->s_nzones - sbi->s_firstdatazone;
        sbi->s_inodes_per_block = bsize >> 6;
        sbi->s_inodes_per_block_1 = (bsize >> 6)-1;