Skip to contentSkip to conversation

Fix potential Divide by Zero error in ext2_get_inode()

PATCH In review B. N. Poornima 2 participants 3 messages May 13, 2009 – May 14, 2009 ext2 file system

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 Fix potential Divide by Zero error in ext2_get_inode()

Found a line of code in ext2_get_inode function of inode.c in the ext2 filesystem that has the potential of hitting the divide by 0 error.
*************************************************
block_group = (ino - 1) / EXT2_INODES_PER_GROUP(sb);
*************************************************
There is no checking done here to verify if EXT2_INODES_PER_GROUP() returns 0. This could result in divide by zero error and panic the system. Below is the patch, built against 2.6.30-rc5, to correct the same:

fs/ext2/inode.c 2009-05-12 19:14:42.873991280 +0530 renamed from fs/ext2/inode.c 2009-05-09 05:44:14.000000000 +0530 +10 -1
Index: linux-2.6.30-rc5/fs/ext2/inode.c
===================================================================
@@ -1138,6 +1138,7 @@
       unsigned long block_group;
       unsigned long block;
       unsigned long offset;
+       unsigned long inodes_per_group;
       struct ext2_group_desc * gdp;
 
       *p = NULL;
@@ -1145,7 +1146,10 @@
           ino > le32_to_cpu(EXT2_SB(sb)->s_es->s_inodes_count))
               goto Einval;
 
-       block_group = (ino - 1) / EXT2_INODES_PER_GROUP(sb);
+       inodes_per_group = EXT2_INODES_PER_GROUP(sb);
+       if (!inodes_per_group)
+               goto Einval1;
+       block_group = (ino - 1) / inodes_per_group;
       gdp = ext2_get_group_desc(sb, block_group, NULL);
       if (!gdp)
               goto Egdp;
@@ -1166,6 +1170,11 @@
       ext2_error(sb, "ext2_get_inode", "bad inode number: %lu",
                  (unsigned long) ino);
       return ERR_PTR(-EINVAL);
+
+Einval1:
+       printk(KERN_ERR "Ext2-fs: Filesystem corrupted. Wrong inodes per group: %lu, run e2fsck\n", (unsigned long)inodes_per_group);
+       return ERR_PTR(-EINVAL);
+

Eio:
ext2_error(sb, "ext2_get_inode",
"unable to read inode block - inode=%lu, block=%lu",

#1 lore raw
Quoted 8 lines · On Wed, 13 May 2009 13:51:01 +0530 "B. N. Poornima" <[email protected]> wrote:

Found a line of code in ext2_get_inode function of inode.c in the ext2 filesystem that has the potential of hitting the divide by 0 error.
*************************************************
block_group = (ino - 1) / EXT2_INODES_PER_GROUP(sb);
*************************************************
There is no checking done here to verify if EXT2_INODES_PER_GROUP() returns 0. This could result in divide by zero error and panic the system. Below is the patch, built against 2.6.30-rc5, to correct the same:

EXT2_INODES_PER_GROUP() cannot return zero. See ext2_fill_super():

	if (sbi->s_inodes_per_block == 0 || sbi->s_inodes_per_group == 0)
		goto cantfind_ext2;
#2 lore raw
Quoted 12 lines · Andrew Morton
Quoted 7 lines · On Wed, 13 May 2009 13:51:01 +0530 "B. N. Poornima" <[email protected]> wrote:

Found a line of code in ext2_get_inode function of inode.c in the ext2 filesystem that has the potential of hitting the divide by 0 error.
*************************************************
block_group = (ino - 1) / EXT2_INODES_PER_GROUP(sb);
*************************************************
There is no checking done here to verify if EXT2_INODES_PER_GROUP() returns 0. This could result in divide by zero error and panic the system. Below is the patch, built against 2.6.30-rc5, to correct the same:

Quoted 6 lines

EXT2_INODES_PER_GROUP() cannot return zero. See ext2_fill_super():

	if (sbi->s_inodes_per_block == 0 || sbi->s_inodes_per_group == 0)
		goto cantfind_ext2;

Andrew,
Thanks for pointing that out!

Regards,
Poornima.