Re: PROBLEM: OOPSes in PREEMPT SMP for AMD Opteron Dual-Core with Memhole Mapping (non tainted kernel)

[Jacob Martin <martin@cs.uga.edu>]

On Wednesday 15 June 2005 03:51 pm, you wrote:
> On Mon, Jun 13, 2005 at 07:53:16PM +0000, Jacob Martin wrote:
> > On Monday 13 June 2005 10:06 am, Andi Kleen wrote:
> > > On Sun, Jun 12, 2005 at 03:29:50PM -0400, Jacob Martin wrote:
> > > > Hardware memhole mapping never seems to work, or causes lockups right
> > > > away.  I need to test it further though.
> > > >
> > > > I have discovered that with the following features enabled:
> > > >
> > > > 1.  Software memhole mapping
> > > > 2.  Continuous,
> > > >
> > > > linux sees the entire 4GB of memory.  However, when things start
> > > > getting requested from the upper half, there are Oopses generated. 
> > > > Attached are two Oopses that occurred under the test scenario
> > > > described.
> > >
> > > What happens when you boot with numa=off or with numa=noacpi ?
> >
> > You got it!  It seems to be working just fine without it compiled into
> > the kernel.
>
> Not compiled what in the kernel? I just wanted you to boot
> the kernel with these options.

I took NUMA support out of the kernel.  I will recompile NUMA in and reboot 
with the options.  I'm very stable now without NUMA in the kernel.

> > > The system seems to believe it has memory in an area not covered
> > > by mem_map.
> >
> > I think you hit it right on the head.
> >
> > I enabled NUMA because I had anticipated upgrading later.  So I guess if
> > you don't actually have NUMA set up hardware-wise, and enable this
> > module, then you will have problems.
>
> No, it should work fine in theory.

Hmm, ok I'll work with you to try and verify that what happened really 
happened because of NUMA.

> > Maybe a simple update to the kernel "K8 NUMA support" Processor feature's
> > help section should be made to note this?  Or, is there something that
> > could be fixed somewhere.  I wouldn't mind helping, it was baffling me
> > for two weeks.
>
> It is something that must be either fixed or workarounded (if it's a bug
> in your BIOS, which is quite possible)
>
> Can you send me the full dmesg from a numa boot again?

Yes, I'll send it to you as soon as this experiment I am running finishes 
(unless you know how to sleep a process to disk for reenactment after boot?).
> -Andi
> -
> 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/

-- 
Jacob Martin