vmunix: trap: invalid memory write access from kernel mode

[Azim_3]

Hi All,

Today my server Alpha ES40(o/s 5.1B,firmware 6.3-2)got rebooted giving below mentioned error.

Oct 15 13:00:58 atd1 vmunix: trap: invalid memory write access from kernel mode
Oct 15 13:00:58 atd1 vmunix:
Oct 15 13:00:58 atd1 vmunix: faulting virtual address: 0x0000000000000010
Oct 15 13:00:58 atd1 vmunix: pc of faulting instruction: 0xffffffff000bb82c
Oct 15 13:00:58 atd1 vmunix: ra contents at time of fault: 0xffffffff000bb82c
Oct 15 13:00:58 atd1 vmunix: sp contents at time of fault: 0xfffffe05e5e4f620
Oct 15 13:00:58 atd1 vmunix:
Oct 15 13:00:58 atd1 vmunix: panic (cpu 2): kernel memory fault

Can any one pin point what is the actual cause of failure?
Is it related to h/w?
Attached herewith binary.errlog & messages.

[Azim_3]

Also attaching the crash-data file.

Thanks
Azim

[Michael Schulte zur Sur]

Hi,

what patch kit are you running?
It sounds like a software matter to me.

greetings,

Michael

[Azim_3]

Hello Michael,

The current installed patchkits are

KITNAME> OSF540, TCR540 on July 25th 2004

KITNAME> OSF540 on June 11th 2005

KITNAME> OSF540 on June 26th 2005

regds
azim

[David_854]

Azim,
Looking at the attached info, its better you call your software support.
It looks like a hardware issue.

David

[Johan Brusche]

There is nothing in the binary.errlog that points to a HW problem. The file is 1 year old, so you should recycle it via the command:
kill -USR1 `cat /var/run/binlogd.pid`

The messages file shows that the ES40 was booted on Oct 15 at 8:49 and 4 hrs later it paniced and rebooted again.

Was anything changed before the reboot at 8:49 ?

Please make sure that new_wire_method=1 in /etc/sysconfigtab . If not, reboot with this vm-parameter set to 1, I believe it will help to avoid this kind of panic.

Rgds,
__ Johan /.