Kernel 2.6.21 BUG?

Tijnema ! tijnema at gmail.com
Sun Apr 29 15:25:16 PDT 2007


On 4/29/07, alupu at verizon.net <alupu at verizon.net> wrote:
> - Alex:
>  <snip>
>  BUG: at arch/i386/kernel/smp.c:546 smp_call_function()
>  [<c010d965>] smp_call_function+0x125/0x130
>  [<........>] ... (Twenty lines of this nature)
>  [<c0341570>] unknown_bootoption+0x0/0x240
>  <snip>
>
> - Alexander E. Patrakov:
> 1) Tell us the kernel line in your /boot/grub/menu.lst file
> 2) Complain to LKML <linux-kernel at vger.kernel.org>,
>   including 20 lines that you snipped
>
> - Tijnema:
> 3) Just go back to 2.6.20.x, unless you really need it.
>   SMP seems very buggy in 2.6.21.x.
>
> - Dave (a.k.a., linux23dragon):
>  Update to iptables-1.3.7 and iproute-2.6.20-070313
>  Just make sure that you compile against the raw kernel headers
>  when you compile iptables.  Works good
> ------------------------------------------
> MANY THANKS to one and all!
>
> Individual replies:
> - ALEXANDER: 1. I haven't touched that line since day one.
>  I don't see its relevance, anyway.
>  Even the 2.6.21 kernel works fine until you try to network.
>  2.  I'm a little shell-shocked since I last got entangled into kernel mailing lists.
>  If I manage to gather enough courage and energy I might submit a bug report.
> - TIJNEMA:  You might have a point there.  My machine has been into SMP since it was built.
>  I'm impressed with the speed you gather statistics;
>  the 2.6.21 had been out for only a couple of days when you posted.
> - DAVE: I don't use "iptables" (no firewall);
>  Cyberpeople are free to come and go as they please :-)
>  Strangely, nothing untoward in all these years
>  (true, the box is on-line only when I need to be on line).
>  My system is as straightforward as they come.
>  This is why this BUG is so annoying at this day and age
>  (it was not present all these years until 2.6.21).
>  However, I did install iproute-2.6.20-070313
>  (BTW, it won't compile if you don't use "the raw kernel headers",
>  '/linux-2.6.21/include', anyway).
>  Unfortunately, when trying to network the kernel didn't work good.
>  It still crashes on the slightest attempt at networking
>  but at least with a different screen - to keep it interesting, I suppose.
>  FWIW (maybe an astute and generous soul might detect a clue):
>  1. Process swapper ...
>  2. Stack ...
>    ...
>  5. Call Trace:
>    [<f88d2635>] sis900_interrupt+0x625/0x6a0 [sis900]
>    ...
> 19. =======================
> 20. Code: 00 ...
>     24 ...
>     c0 ...
> 23. EIP [c024fde4] skb_over_panic+0x54/0x60 SS:ESP 0068:c033debc
> 24. Kernel panic - not syncing: Fatal exception in interrupt
> (Note:  SiS900 is my board's Ethernet chip.
>  It's the one connecting to the ADSL router)
>
> I think my next move will be to stay put until some relief comes along (time heals all wounds):
> I just noticed 2.6.21.1 is out!  One never knows do one?
>
> Thanks,
> -- Alex


First it might be useful to post the whole log, and not only a few lines.

Second, there's a SiS900 patch applied to the kernel in 2.6.21.1, it
might fix your problem. As 2.6.21.x versions are bugfixing only.

Third, if it still doesn't, then you really need to go to the LKML as
Alexander suggested.

Tijnema



More information about the blfs-support mailing list