Page 1 of 1

2.6.23.9 hangs/oops/kernel bug

PostPosted: Wed Jan 23, 2008 2:37 am
by glaeken
Arch x86
2.6.23.9+grsec

After few hours (ok. 4 to 30)

I had sth like this:
dsl.one.pl/kernelbug.txt
dsl.one.pl/kernelbug_ksym.txt

I understand it's a developement version, BUT this kernel (2.6.23.9+grsec) works fine for me on another 2 boxes, or at least oops'es are very very rare there (althogh there have been some mysterious reboots of this machines, but not every 12 hours - say once for 2-3 months)

Re: 2.6.23.9 hangs/oops/kernel bug

PostPosted: Wed Jan 23, 2008 7:32 am
by PaX Team
glaeken wrote:I understand it's a developement version, BUT this kernel (2.6.23.9+grsec) works fine for me on another 2 boxes, or at least oops'es are very very rare there (althogh there have been some mysterious reboots of this machines, but not every 12 hours - say once for 2-3 months)
can you post your .config as well please? are you using SLAB (vs. SLUB/SLOB) on the other boxes as well? in any case, this looks like some internal SLAB accounting bug, i don't know yet how we could cause this. you could maybe enable the SLAB debugging options but this bug doesn't look like a corruption per se, 'just' some reference count mismatch.

Re: 2.6.23.9 hangs/oops/kernel bug

PostPosted: Fri Jan 25, 2008 11:11 am
by glaeken
This is config:
dsl.one.pl/config_kbug.gz

But don't get so engaged.. I changed to 2.6.19.2 AND changed almost whole hardware except two NIC's and hardrive. After about 27 hours of uptime I got kernel bug (I didn't have time to write it down because of automatic restart) but it wasn't in SLAB at all. I'm getting nuts here, just WTF!?
I used 2.6.19.2 for a long time, and 2.6.23.9 (on other machines) without a problem.
Now I changed the two NIC's and switched to 2.6.21.5 which was stable for about a year of uptime.

Thx for help anyway :)