Fatal trap 12: page fault while in kernel mode

NAKAJI Hiroyuki nakaji at jp.freebsd.org
Fri Dec 5 23:53:29 UTC 2008


Thanks for advice.

I noticed I had just upgraded the kernel only and the userland remains
old, that is, kernel is on Dec 3 while userland is on Oct 20. I'll do
installworld too, after next buildworld, buildkernel, installkernel and
reboot.

And, I found same panic this morning (JST), but before I read
Koshy's instruction, I typed 'reset' without doadump and I cannot show
you a full bt. I'll do it when next panic.

So, what I have to do are three:

1. Full upgrade to the latest kernel and userland (world)
2. Observe whether a panic occurs, and
3. When panic, save the crash dump and get full bt with kgdb

I hope it will not reach to the step three. Thanks.

>>>>> In <3a142e750812050613w4e9155bat950f03716aa58beb at mail.gmail.com> 
>>>>>	"Paul B. Mahol" <onemda at gmail.com> wrote:

Paul> Full bt would be better, also make sure that world and kernel are in
Paul> sync (because of recent libc changes).

>>>>> In <84dead720812050840j28c95014wf76095c3d6def985 at mail.gmail.com> 
>>>>>	"Joseph Koshy" <joseph.koshy at gmail.com> wrote:
> > Fatal trap 12: page fault while in kernel mode
> > cpuid = 0; apic id = 00
> > fault virtual address   = 0x0
> > fault code              = supervisor read, page not present
> > instruction pointer     = 0x20:0xc0a57285

> > Why this occurs and what do I have to check?

Koshy> The trap is being caused by a null pointer access.

Koshy> You may want to take a full dump and use kgdb(1) to investigate the matter:

Koshy>   ddb> call doadump
Koshy>   ddb> reset

Koshy> and after the reboot finishes,

Koshy>   # kgdb /boot/kernel/kernel /var/crash/vmcore.N

Koshy> where "vmcore.N" is the name of the dump file saved by savecore(8).


-- 
NAKAJI Hiroyuki


More information about the freebsd-current mailing list