The ext4_fill_super may return with value 0 in case of an error. The
reason is that the ret variable is used to store the return value of a
called function and is not set back to an error value later on.
This issue caused a NULL pointer de-ref in vfs_kern_mount on one of my
machines when it tried to mount a partition:
BUG: unable to handle kernel NULL pointer dereference at 0000000000000090
IP: [<ffffffff8112e8dc>] vfs_kern_mount+0x7c/0x180
PGD 218480067 PUD 225323067 PMD 0
Oops: 0000 [#1] SMP
last sysfs file: /sys/devices/system/cpu/cpu5/cache/index2/shared_cpu_map
CPU 1
...
Process mount (pid: 3591, threadinfo ffff8802258fc000, task ffff88022586dac0)
Stack:
ffff8802258fde38 0000000000000000 ffff880225e299b0 ffff880200000001
ffffffff81a230e0 ffff880225e2c818 ffff8802258fde88 ffffffff8112ea52
ffff8802258fde98 0000000000000246 ffff88022586dac0 0000000000000060
Call Trace:
[<ffffffff8112ea52>] do_kern_mount+0x52/0x130
[<ffffffff811497ff>] do_mount+0x2bf/0x810
[<ffffffff810face3>] ? strndup_user+0x53/0x70
[<ffffffff81149de0>] sys_mount+0x90/0xe0
[<ffffffff81002fdb>] system_call_fastpath+0x16/0x1b
This patch fixes the issue by using the 'err' variable instead of 'ret'
for that function-call like it is done in for all other function-calls
in ext4_fill_super too.
If have seen this issue on 2.6.36 too, so this patch may be -stable
material as well.
Signed-off-by: Joerg Roedel <[email protected]>
---
fs/ext4/super.c | 4 ++--
1 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/fs/ext4/super.c b/fs/ext4/super.c
index 61182fe..499cf5d 100644
--- a/fs/ext4/super.c
+++ b/fs/ext4/super.c
@@ -3268,9 +3268,9 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent)
* Test whether we have more sectors than will fit in sector_t,
* and whether the max offset is addressable by the page cache.
*/
- ret = generic_check_addressable(sb->s_blocksize_bits,
+ err = generic_check_addressable(sb->s_blocksize_bits,
ext4_blocks_count(es));
- if (ret) {
+ if (err) {
ext4_msg(sb, KERN_ERR, "filesystem"
" too large to mount safely on this system");
if (sizeof(sector_t) < 8)
--
1.7.0.4
On Thu, Nov 18, 2010 at 10:47:12PM +0100, Joerg Roedel wrote:
> The ext4_fill_super may return with value 0 in case of an error. The
> reason is that the ret variable is used to store the return value of a
> called function and is not set back to an error value later on.
> This issue caused a NULL pointer de-ref in vfs_kern_mount on one of my
> machines when it tried to mount a partition:
>
> BUG: unable to handle kernel NULL pointer dereference at 0000000000000090
> IP: [<ffffffff8112e8dc>] vfs_kern_mount+0x7c/0x180
> PGD 218480067 PUD 225323067 PMD 0
> Oops: 0000 [#1] SMP
> last sysfs file: /sys/devices/system/cpu/cpu5/cache/index2/shared_cpu_map
> CPU 1
>
> ...
>
> Process mount (pid: 3591, threadinfo ffff8802258fc000, task ffff88022586dac0)
> Stack:
> ffff8802258fde38 0000000000000000 ffff880225e299b0 ffff880200000001
> ffffffff81a230e0 ffff880225e2c818 ffff8802258fde88 ffffffff8112ea52
> ffff8802258fde98 0000000000000246 ffff88022586dac0 0000000000000060
> Call Trace:
> [<ffffffff8112ea52>] do_kern_mount+0x52/0x130
> [<ffffffff811497ff>] do_mount+0x2bf/0x810
> [<ffffffff810face3>] ? strndup_user+0x53/0x70
> [<ffffffff81149de0>] sys_mount+0x90/0xe0
> [<ffffffff81002fdb>] system_call_fastpath+0x16/0x1b
>
> This patch fixes the issue by using the 'err' variable instead of 'ret'
> for that function-call like it is done in for all other function-calls
> in ext4_fill_super too.
> If have seen this issue on 2.6.36 too, so this patch may be -stable
> material as well.
Lukas and I have been trying to fix this bug for a couple of days now, and he
actually sent off an identical patch this morning:
http://www.spinics.net/lists/linux-ext4/msg21743.html
(I also sent off a similar cleanup of the ext3 counterpart.)
But, thanks for pointing this out! :)
--D
On Thu, Nov 18, 2010 at 02:21:57PM -0800, Darrick J. Wong wrote:
> On Thu, Nov 18, 2010 at 10:47:12PM +0100, Joerg Roedel wrote:
> > This patch fixes the issue by using the 'err' variable instead of 'ret'
> > for that function-call like it is done in for all other function-calls
> > in ext4_fill_super too.
> > If have seen this issue on 2.6.36 too, so this patch may be -stable
> > material as well.
>
> Lukas and I have been trying to fix this bug for a couple of days now, and he
> actually sent off an identical patch this morning:
> http://www.spinics.net/lists/linux-ext4/msg21743.html
>
> (I also sent off a similar cleanup of the ext3 counterpart.)
>
> But, thanks for pointing this out! :)
Oh ok, good to know. I ran into this issue today when I installed a new
kernel on that machine and the 'update-grub' command gave me an ooops.
Good to know it will be fixed soon :)
Joerg