Random panics with 5.3-REL, SMP

[Doug White]

On Wed, 24 Nov 2004, Randy Bush wrote:

> > - Using a serial console to the box, you can reliably gather information
> >   without the core dump mechanism working.
>
> i caught one, but the rest happened and reboot's beastie wiped the console
>
>     Fatal trap 12: page fault while in kernel mode
>     cpuid = 0; apic id = 00
>     fault virtual address   = 0x41959c25

Thats a wierd address.  If I didn't know better, I'd say that there's a
MSB flip in there (4+8=c in hex).

A backtrace would be interesting to look at.

>     fault code              = supervisor write, page not present
>     instruction pointer     = 0x8:0xc04e813a
>     stack pointer           = 0x10:0xe784d978
>     frame pointer           = 0x10:0xe784d994
>     code segment            = base 0x0, limit 0xfffff, type 0x1b
> 			    = DPL 0, pres 1, def32 1, gran 1
>     processor eflags        = interrupt enabled, resume, IOPL = 0
>     current process         = 68301 (sshd)
>     trap number             = 12
>     panic: page fault
>     cpuid = 0
>     boot() called on cpu#0
>     Uptime: 9h29m10s
>     Cannot dump. No dump device defined.

Looks like you just need to set 'dumpdev' in rc.conf.  What is your boot
device and swap partition(s)?

>     Automatic reboot in 15 seconds - press a key on the console to abort
>     Rebooting...
>     cpu_reset called on cpu#0
>     cpu_reset: Stopping other CPUs
>
> since then, i installed a kernel with debugging symbols (blush), but
> have yet to get a usable savecore and the beastie got me too.

Set 'beastie_disable="YES"' in /boot/loader.conf to stop that from
happening. :)  if you turn on session logging in your comm program it
should capture the output before it scrolls off.

-- 
Doug White                    |  FreeBSD: The Power to Serve
dwhite at gumbysoft.com          |  www.FreeBSD.org