Kernel panic on PowerEdge 1950 under certain stress load

[Benjie Chen]

We got this bt after setting kernel dump stuff up. For some reason kgdb did
not find debugging symbols, even though the kernel was compiled with -g.
Hopefully this helps and I will go dig some more to find why we don't have a
debug symbol.

#0  0xc0674fae in doadump ()
#1  0xc067550a in boot ()
#2  0xc0675831 in panic ()
#3  0xc088e29c in trap_fatal ()
#4  0xc088da0e in trap ()
#5  0xc0879d4a in calltrap ()
#6  0xc066c731 in _mtx_lock_sleep ()
#7  0xc06bba96 in unp_gc ()
#8  0xc06961b3 in taskqueue_run ()
#9  0xc0696696 in taskqueue_thread_loop ()
#10 0xc065ec4d in fork_exit ()
#11 0xc0879dac in fork_trampoline ()




On 9/24/07, Ivan Voras <ivoras at freebsd.org> wrote:
>
> On 24/09/2007, Benjie Chen <benjie at addgene.org> wrote:
> > Ivan and Kris,
> >
> > I will try to get a kernel trace -- it may not happen for awhile since I
> am
> > not in the office and working remotely for awhile so it may not be easy
> to
> > get a trace... but I will check.
>
> It's fairly easy:
>
> 1) add lines like the following in rc.conf:
> dumpdev="/dev/amrd0s1b"
> dumpdir="/storage1/crashdumps"
>
> (dumpdev is your swap partition, which must be larger than your RAM,
> dumpdir is where the crash dumps will be saved, also needs to be
> larger than RAM)
>
> 2) add these lines to sysctl.conf:
>
> debug.debugger_on_panic=0
> debug.trace_on_panic=1
>
> (These will cause the panic message and backtrace to be automatically
> recorded in the message buffer saved to the kernel crash dump. If not
> told otherwise, the machine will then reboot and at the next boot
> generate a crash dump in your dumpdir. When you get the kernel crash
> dump, run "kgdb vmcore.0 /boot/kernel/kernel" as root and then you can
> inspect the trace, core dump, etc. You don't need any special
> knowledge of (k)gdb for this. The one useful command to you may be
> "bt" - generate a backtrace).
>
> > It looks like the problem reported by that link, and some of the links
> from
> > there though...
>
> I'm trying to solve that one with Craig Rodrigues. So far it's been
> without success but we know where the problem is.
>



-- 
Benjie Chen, Ph.D.
Addgene, a better way to share plasmids
www.addgene.org