Re: Oops after 30 days of uptime

[Willy Tarreau <w@1wt.eu>]

Hi !

On Sun, Sep 03, 2006 at 10:03:22PM +0200, Ondrej Zary wrote:
> > On Friday 01 September 2006 19:00, Patrick McHardy wrote:
> > > Ondrej Zary wrote:
> > > > Hello,
> > > > my home router crashed after about a month. It does this sometimes but
> > > > this time I was able to capture the oops. Here is the result of running
> > > > ksymoops on it (took a photo of the screen and then manually converted
> > > > to plain-text). Does it look like a bug or something other?
> > > >
> > > >
> > > > Code;  c01eeb9e <init_or_cleanup+15e/160>
> > > > 00000000 <_EIP>:
> > > > Code;  c01eeb9e <init_or_cleanup+15e/160>   <=====
> > > >    0:   8b 5e 18                  mov    0x18(%esi),%ebx   <=====
> > > > Code;  c01eeba1 <ip_conntrack_protocol_register+1/70>
> > > >    3:   11 d8                     adc    %ebx,%eax
> > >
> > > This looks like a bug in some out of tree protocol module (2.4 only
> > > contains the built-in protocols). Did you apply any netfilter patches?
> >
> > No patches, it's clean 2.4.31.
> > Hopefully I typed all the numbers correctly...
> 
> Checked all numbers and it's correct. Can this be a hardware problem?

Given that esi was 0xc1ffffe8, the oops was triggered by a crossed page
boundary (0xc2000000). This does not look like a hardware problem, but
rather a bug somewhere with too small a malloc for some structure. It
is really hard to tell where. I would suspect some classical bugs such
as kmalloc(sizeof(p)) instead of kmalloc(sizeof(*p)), but it's just
pure guess. I'll try to figure out what the code section was to find
the structure name and check its allocation path.

> -- 
> Ondrej Zary

Regards,
Willy

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/