resize: check s_log_groups_per_flex before accessing flex groups

PATCH In review Eric Sandeen 3 participants 5 messages Mar 27, 2010 – Apr 4, 2010 ext4 file system

lore

Share & tools

lore

Eric Sandeen

author

2010-03-27 19:32 UTC #0 lore raw

PATCH 1/1 resize: check s_log_groups_per_flex before accessing flex groups

This is for kernel.org bug:

#13549, Kernel oops while online resizing of an ext4 filesystem

if groups_per_flex < 2, sbi->s_flex_groups[] doesn't get filled out, and every other access to this first tests s_log_groups_per_flex; same thing needs to happen in resize or we'll wander off into a null pointer.

Thanks to Christoph Biedl, who came up with the trivial testcase:

# truncate --size 128M fsfile
# mkfs.ext3 -F fsfile
# tune2fs -O extents,uninit_bg,dir_index,flex_bg,huge_file,dir_nlink,extra_isize fsfile
# e2fsck -yDf -C0 fsfile
# truncate --size 132M fsfile
# losetup /dev/loop0 fsfile
# mount /dev/loop0 mnt
# resize2fs -p /dev/loop0

Reported-by Alessandro Polverini <[email protected]>

Test-case-by Christoph Biedl <[email protected]>

Signed-off-by Eric Sandeen <[email protected]>

---

fs/ext4/resize.c +2 -1

Index: linux-2.6/fs/ext4/resize.c
===================================================================
@@ -930,7 +930,8 @@ int ext4_group_add(struct super_block *s
 	percpu_counter_add(&sbi->s_freeinodes_counter,
 			   EXT4_INODES_PER_GROUP(sb));
 
-	if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FLEX_BG)) {
+	if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FLEX_BG) &&
+	    sbi->s_log_groups_per_flex) {
 		ext4_group_t flex_group;
 		flex_group = ext4_flex_group(sbi, input->group);
 		atomic_add(input->free_blocks_count,

Andreas Dilger

2010-03-28 15:14 UTC #1 lore raw

Quoted 6 lines · Eric Sandeen

#13549, Kernel oops while online resizing of an ext4 filesystem
if groups_per_flex < 2, sbi->s_flex_groups[] doesn't get filled out, and every other access to this first tests s_log_groups_per_flex; same thing needs to happen in resize or we'll wander off into a null pointer.

Does it even make sense to set INCOMPAT_FLEX_BG if we only have a single group per flexbg? That is just a normal filesystem then. That would be a separate bug in mke2fs.

Quoted 22 lines

Reported-by Alessandro Polverini <[email protected]>
Test-case-by Christoph Biedl <[email protected]
Quoted 1 line
Signed-off-by Eric Sandeen <[email protected]>
---
Index: linux-2.6/fs/ext4/resize.c
===================================================================
fs/ext4/resize.c +2 -2
@@ -930,7 +930,8 @@ int ext4_group_add(struct super_block *s
 	percpu_counter_add(&sbi->s_freeinodes_counter,
 			   EXT4_INODES_PER_GROUP(sb));
 
-	if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FLEX_BG)) {
+	if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FLEX_BG) &&
+	    sbi->s_log_groups_per_flex) {
 		ext4_group_t flex_group;
 		flex_group = ext4_flex_group(sbi, input->group);
 		atomic_add(input->free_blocks_count,
 
--

Quoted 3 lines

To unsubscribe from this list: send the line "unsubscribe linux-ext4" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

Cheers, Andreas
--
Andreas Dilger
Principal Engineer, Lustre Group
Oracle Corporation Canada Inc.

Eric Sandeen

author

2010-03-28 15:27 UTC #2 lore raw

Quoted 11 lines · Andreas Dilger

Quoted 6 lines · Eric Sandeen
#13549, Kernel oops while online resizing of an ext4 filesystem
if groups_per_flex < 2, sbi->s_flex_groups[] doesn't get filled out, and every other access to this first tests s_log_groups_per_flex; same thing needs to happen in resize or we'll wander off into a null pointer.
Does it even make sense to set INCOMPAT_FLEX_BG if we only have a single group per flexbg? That is just a normal filesystem then. That would be a separate bug in mke2fs.

yes, I really wondered about that, but we have this check throughout the ext4 kernel code right now, so as a quick fix ...

(note in this case it was an ext3 fs converted to ext4, with tune2fs:)

# tune2fs -O extents,uninit_bg,dir_index,flex_bg,huge_file,dir_nlink,extra_isize fsfile

I haven't honestly looked at what it means to "turn on" flex_bg for a filesystem not originally mkfs'd with it. I'm not sure it does anything other than setting the flag, leaving flex group size == group size.

Thanks,
-Eric

[email protected]

maintainer

2010-04-04 02:11 UTC #3 lore raw

In reply to Andreas Dilger

Quoted 11 lines · Andreas Dilger

Quoted 6 lines · Eric Sandeen
#13549, Kernel oops while online resizing of an ext4 filesystem
if groups_per_flex < 2, sbi->s_flex_groups[] doesn't get filled out, and every other access to this first tests s_log_groups_per_flex; same thing needs to happen in resize or we'll wander off into a null pointer.
Does it even make sense to set INCOMPAT_FLEX_BG if we only have a single group per flexbg? That is just a normal filesystem then. That would be a separate bug in mke2fs.

Yes, it does make sense to set flex_bg in this case; it allows the group metadata to be stored outside of a blockgroup, which is helpful to e2fsck in some cases when it needs to relocate an inode table and there's no contiguous free space available in the block group.

- Ted

[email protected]

maintainer

2010-04-04 02:17 UTC #4 lore raw

In reply to Eric Sandeen

Quoted 8 lines · Eric Sandeen

This is for kernel.org bug:
#13549, Kernel oops while online resizing of an ext4 filesystem
if groups_per_flex < 2, sbi->s_flex_groups[] doesn't get filled out, and every other access to this first tests s_log_groups_per_flex; same thing needs to happen in resize or we'll wander off into a null pointer.

Added to the ext4 patch queue, thanks.

- Ted