reproducible panic - 6.0-STABLE-SNAP011 - vm_fault: fault on nofault entry

[V. T. Mueller]

Hello,


I was searching the list archives, but could not find anything that 
looked really related. So here is my problem:

We need to move several services from 5.4 i386 to amd64 and thought 
6.0 would work. All tests so far ended in panics or kernel traps, 
though.

Our test hardware is a Supermicro H8D8A 
http://www.supermicro.com/Aplus/motherboard/Opteron/8131/H8DA8.cfm 
with two single-core Opteron 246, eight GB of RAM and one ICP-Vortex 
GDT8500RZ zero-channel RAID controller (iir). Board and controller 
firmware are up-to-date; hardware was fully tested and proven 
error-free.

Booting from 6.0-RELEASE-amd64-bootonly.iso results in kernel trap 
12 depending on ACPI on/off. When not dying of sync with the 
mentioned kernel trap 12, it panics the same way as describe below.

Booting from 6.0-STABLE-SNAP011-bootonly.iso first works, but panics 
at "writing partition information to disk da0" with:

panic: vm_fault: fault on nofault entry, addr: ffffffffb19fb000

This is happening independantly from parameters like swap size 
and/or the patition scheme attempted. Disabling ACPI does not alter 
this behaviour, too. Besides, 6.1-BETA1 is the same (7.0-CURRENT 
dies early during bootup).

Has anyone seen s/th like this before? Any hints where the root of 
this evil might be located? Is there anything I can do to help 
tracking the problem down to its cause?

TIA,
vt