Help? 6.1-S: Fatal trap 12: page fault while in kernel mode

[Maxim Konovalov]

On Fri, 16 Jun 2006, 08:45-0700, David Wolfskill wrote:

> On Thu, Jun 15, 2006 at 04:22:40PM -0700, David Wolfskill wrote:
> > I had one of these [kernel panics] a couple of weeks ago or so...
> > ...[upgrade to -STABLE as of 15 June; repeat panic]...
>
> The message to which I'm replying (posted to -stable) has the
> particulars about the panic in question, and the machine in question is
> still sitting at the DDB prompt, if anyone wishes to work with me on
> that.
>
> But the reason for this message is to report that I upgraded the other
> test machines -- identical confguration: 2x3 GHz Xeons w/ 4 GB RAM;
> kernel config is called "SMP_PAE_DDB" for a fairly good reason -- to
> today's -CURRENT, then started the same test that cause -STABLE to crash
> & burn within a couple of minutes.
>
> That was 30 minutes ago; the test is still running on
>
> FreeBSD localhost 7.0-CURRENT FreeBSD 7.0-CURRENT #1: Fri Jun 16 07:28:18 PDT 2006     dhw at localhost:/usr/obj/usr/src/sys/SMP_PAE_DDB  i386
>
> As I commented in email to some colleagues, "color me surprised."
>
> I've suggested to the vendor (the program under test on the box is
> from a vendor, built under & for FreeBSD 5.x; I'm using the
> misc/compat5x port) that they consider trying this themselves, and
> perhaps also take advantage of John Birrell's work to date on the
> FreeBSD port of DTrace.
>
> I'm still not too keen to run a production workload on a -CURRENT
> platform.  I don't know if whatever is causing -CURRENT to keep running
> while -STABLE dies is an MFC candidate, but it seems to me that
> identifying the salient change(s) would be helpful in figuring that out.
>
> Any suggestions for how to go about doing that?

"trace" in ddb would be good start.  Do you really need PAE?

-- 
Maxim Konovalov