Today is not a good day for -current on alpha

[Bernd Walter]

On Wed, Aug 20, 2003 at 09:03:21PM +0200, Wilko Bulte wrote:
> On Wed, Aug 20, 2003 at 12:00:12PM -0700, Marcel Moolenaar wrote:
> > On Wed, Aug 20, 2003 at 08:07:36PM +0200, Wilko Bulte wrote:
> > > On Wed, Aug 20, 2003 at 07:10:24PM +0200, Bernd Walter wrote:
> > > > On Mon, Aug 18, 2003 at 10:15:45PM +0200, Wilko Bulte wrote:
> > > > > Second time today on my quad AS4100:
> > > > > 
> > > > > db> trace
> > > > > sysctl_vm_zone() at sysctl_vm_zone+0x370
> > > > > sysctl_root() at sysctl_root+0x16c
> > > > > userland_sysctl() at userland_sysctl+0x170
> > > > > __sysctl() at __sysctl+0xe8
> > > > > syscall() at syscall+0x3b8
> > > > > XentSys() at XentSys+0x64
> > > > > --- syscall (202, FreeBSD ELF64, __sysctl) ---
> > > > > --- user mode ---
> > > > > db> 
> > > > 
> > > > Was this a missalignment panic?
> > > > I get one reproduceable when going multi with a kernel from 17th.
> > > > Trace looks the same.
> > > 
> > > No, it was:
> > > 
> > >  > ^M
> > >  >     trap entry     = 0x2 (memory management fault)^M
> > >  >     cpuid          = 1^M
> > 
> > JFYI: both may be the result of the same bug. Uninitialized
> > variables (pointers) have a tendency to cause different kinds
> > of faults depending on the garbage.
> 
> Bosko provided me with a patch, which worked, and which he in the
> meantime committed. My alpha is now happily building world with its
> patched kernel.

Great - I'll resup.

-- 
B.Walter                   BWCT                http://www.bwct.de
ticso at bwct.de                                  info at bwct.de