Hi Julian,
I'm sorry, I already recompiled the kernel with a lower
value for IP_VS_SVC_TAB_BITS. And it seems to work. But
if you would help me debugging the oops, I'd gladly
reconstruct the former kernel state and try to reproduce
the problem. Then we could start debugging with the new
stack addresses and the correct symbol table.
Let me know, aehm, actually I'll do it anyway!
Julian Anastasov wrote:
>
> Hello,
>
> On Tue, 25 Apr 2000, Ratz wrote:
>
> > Hi Wensong,
> >
> > Since several days I get those kernel messages running ipvs-0.9.9 patch
> > on kernel 2.2.14.
> > It seems to have no impact on the functionality of the service itself,
> > so I think my test
> > machine is somewhat f*ed up. Here's the message:
> >
> > IP_VS: WRR scheduling module loaded.
> > .
> > .
> > -> After hours
> > .
> > .
> > Unable to handle kernel NULL pointer dereference at virtual address
> > 0000002b
> > current->tss.cr3 = 025e7000, %cr3 = 025e7000
> > *pde = 00000000
> > Oops: 0000
> > CPU: 0
> > EIP: 0010:[<0000002b>]
>
> Can you try gdb /usr/src/linux/vmlinux ?
>
> disassemble 0x80109391
> For 85000000, 801093f4, 801a2458, 801a3b2e too.
>
> Or just try with ksymoops:
> /usr/src/linux/Documentation/oops-tracing.txt
or Linux Device Drivers by Alessandro Rubini!
Thanks alot for your help,
Roberto Nibali, ratz
|