Bug#582088: linux-image-2.6.32-5-686: null pointer dereference in __mark_inode_dirty+0x15/0x10b

[Thiemo Nagel <thiemo.nagel@googlemail.com>]

Dear Jonathan,

On 10/21/2011 01:26 AM, Jonathan Nieder wrote:
> > I came across this BUG yesterday on my Asus Eee PC 1000HG.  I had probably
> > done 10-15 suspend/resume cycles that day.  I think I had mounted curlftpfs
> > over a somewhat flaky 3G connection earlier that day, but I didn't access it
> > at the time of the BUG (though I cannot speak for cron jobs, etc.).  Apart
> > from that, I was only ext3 and the usual system stuff (tmpfs, proc, sysfs,
> > devpts, binfmt_misc).
> [...]
> > BUG: unable to handle kernel NULL pointer dereference at 00000008
> > IP: [<c10c90f1>] __mark_inode_dirty+0x15/0x10b
>
> In other words, __mark_inode_dirty was called with (sb->s_op == NULL).
> Weird.
>
> Did this ever happen again?

I've found "BUG: unable to handle kernel NULL pointer" 7 times in the 2 
years of log files that I keep.  It was associated with 
__mark_inode_dirty only once, though.  Other occurrences have been:

(null) (Debian 2.6.32-27)
strcmp+0x6/0x19 (Debian 2.6.32-27)
__d_lookup+0xb5/0xd3 (Debian 2.6.32-15)
drm_mm_put_block+0x1e/0x123 (Debian 2.6.32-13)
rw_verify_area+0x43/0xac (2.6.31.5)

Cheers,
Thiemo