dom0 crash on csched_acct

["guilherme m. schroeder" <guialemas@gmail.com>]

Hi,

I'm having some random crashes with XenServer 5.6 SP2 (Xen 3.4.2).
I thought that it was a problem with Xen 3.4 and tried XenServer 6.0
(Xen 4.1.1), but no luck.
I know that this list is about XenSource and (maybe) doesn't cover it,
but i'm trying to understand
what is happening.

All i have is this crash.log: http://cpro3845.publiccloud.com.br/crash.log

Kdump segfaults on XenServer, so no core to help.

I took a look on sched_credit.c +978 and this is what triggering the crash:

        BUG_ON( (sdom->weight * sdom->active_vcpu_count) > weight_left );

We have some VM's that have different vCPU priority values. Could be
this that is triggering the problem?
Also, there is a limit for vcpu_count on a host?

I'm still trying to reproduce the crash, but it's totally random.
Does anyone have an idea or where to start?

If more details are needed, let me know.

Thanks.