Subject: Re: Page fault
To: Perry E. Metzger <perry@piermont.com>
From: Timothy A. Musson <timothy.musson@zin-tech.com>
List: netbsd-help
Date: 10/21/2003 13:41:39
At 12:18 PM 10/21/03 , Perry E. Metzger wrote:
>
>"Timothy A. Musson" <timothy.musson@zin-tech.com> writes:
>> We've had a page fault error on a production machine that caused a kernel
>> panic.
>[...]
>> Oct 17 22:41:50 <host> /netbsd: NetBSD 1.2.1 (<tag>) #6: Thu Feb 17
14:41:16
>> EST 2000
>
>1.2.1 is very, very, very old. I might suggest that the bug might be
>fixed in a newer NetBSD. FYI, 1.5 is the oldest branch that is still
>supported -- the more current release is the 1.6 branch.
>
>Perry
> 

Yes. Yes, it is :)
The catch, though, is that it's a "production machine". Even under the
circumstance that we were able to get funding to put the man hours required
for the upgrade (including porting drivers, etc), with the process we have
to go through it would literally be about a year before we could put the
code update into the field. And, with our new units coming out, that
funding isn't going to appear.

In any case, I don't expect (or even want) anyone digging through source. I
was just hoping that some guru might recognize the pattern of the error and
be able to say "ahh, that looks like it's probably a [corrupted pointer,
page miss, <whatever>]". We're hoping to say in a report that a radiation
hit caused "X" which caused a vm_fault which caused a kernel panic
(reboot). If we can't find any words for "X", that's acceptable, just not
as desirable.

-Tim