panic: Fatal trap 12: page fault while in kernel mode

[Doug White]

On Thu, 17 Feb 2005, Rong-En Fan wrote:

> On Wed, 16 Feb 2005 15:36:25 -0800 (PST), Doug White
> <dwhite at gumbysoft.com> wrote:
> > On Wed, 16 Feb 2005, Rong-En Fan wrote:
> >
> > > Hello,
> > >
> > > This is a 5.3-RELEASE-p5/amd64 on IBM X236 (EM64T) with 2GB RAM
> > > and a LSI 21320 rmpt(4) running at 160MB/s with a hardware
> > > RAID (da0, da1). HTT is enabled. When I run benchmark/blogbench on
> > > /da0/ I can *reproduce* this panic again and again:
> > > (I'm getting a dump now, let me fsck first)
> > > kernel conf & dmesg (boot -v) are at
> > >  http://rafan.infor.org/tmp/236/
> >
> > I only have an 2x244 Opteron box so I'm not sure if this is a problem with
> > KSE or with hyperthreading.  I'll try the benchmark anyway and see if I
> > can reproduce.
> >
> > Looks like I'll need to rebuild first, I'm getting the "exiting from
> > __thread_start" error...

I got a good -CURRENT build and run this test. It appears to get stuck in
an endless loop at the end but no panics result.  I also ran it on a i386
-CURRENT machine for comparison and that completed, so this program
appears to have 64-bit cleanliness problems.

I'll see if I can build a RELENG_5 or 5.3 amd64 box and run the same
diagnostic. Its possible its a bug thats been fixed in CURRENT but not
backported yet.

> If I use machdep.hlt_logical_cpus=1, I got the same panic.
> And when I use kgdb to read the kernel dump, I see only
> #1 ?????? (??) in backtrace.
>
> I just reinstall the system to 5.3-p5, i386. It does not
> panic and finsih the test two times. I'll run more to see if is
> panics.
>
> Regards,
> Rong-En Fan
>

-- 
Doug White                    |  FreeBSD: The Power to Serve
dwhite at gumbysoft.com          |  www.FreeBSD.org