Re: How to produce effective backtraces from crashdump?

From: Dan Nelson <dnelson_at_allantgroup.com>
Date: Fri, 9 Apr 2004 09:27:25 -0500
In the last episode (Apr 09), Xin LI said:
> I'm encounting a very strange problem when trying to obtain
> backtraces from a recently compiled kernel (see attachments for my
> dmesg.boot, kernel configuration, and /etc/make.conf).
> 
> To make the long story short, here's my rough situation: I have built
> a customized kernel with makeoptions DEBUG=-g, and no special options
> in /etc/make.conf. When system is restarted after a panic, and when I
> open it with:
> 
> gdb -k /usr/obj/usr/src/sys/BEASTIE/kernel.debug /var/crash/vmcore.0
> 
> then have a "bt full" I got something like this:
> 
> (kgdb) bt full
> #0  doadump () at /usr/src/sys/kern/kern_shutdown.c:236
> No locals.
> #1  0xc04f63ae in ?? ()
> No symbol table info available.

I can't help you with your symbol problem, but if you add the DDB_TRACE
option to your kernel config, the kernel will print a simple backtrace
(function names and arguments) on panic, which is sometimes enough info
to track down the problem.

-- 
	Dan Nelson
	dnelson_at_allantgroup.com
Received on Fri Apr 09 2004 - 05:27:26 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:37:50 UTC