Decoding stack traces

jason andrade jason at rtfmconsult.com
Fri Oct 24 04:02:01 CDT 2003


On Fri, 24 Oct 2003, John Reuning wrote:

> I'm having issues with a 2550, but this isn't really a hardware
> question.  Does anyone have pointers on how to decode a linux kernel
> stack trace?  I've found a few things on the web, but specifically, I'm
> trying to figure out which module or kernel component is triggering a
> crash.
>
> These aren't oopses or panics, they get dumped to console and look like:
>
> do_IRQ: stack overflow: 1928
> c02638e0 00000788 00000000 f7ffc5e0 00000000 00000001 00000004 c0245884
>        f7ffc5e0 001f85a0 00000004 00000000 00000001 00000004 04000000
> f5510018
>        00000018 ffffff10 c0108c7d 00000010 00000246 f551ceac c034a880
> 00000004
> Call Trace: [<c02638e0>] .rodata.str1.1 [kernel] 0x77b
>
>
> It just scrolls endlessly on the console until I power cycle the system.

hi john,

welcome to what i think is a bug i identified sometime in march and was confirmed
(i think) by howard owen at cisco. i'm taking a guess though.

if you search for 'stack trace' in bugzilla you should pick this up or
alternately for 'broken compiler' or something like that.

if i am guessing correctly you have a RH 7.X OS (or derived from) on the box
and you are using gcc-2.96-113.  the problem seems to be that -113 produces
broken kernels which have the above error.  kernels built with -112 don't
exhibit these symptoms.  bugzilla has a more indepth explanation of how
this is broken (by howard).

the redhat errata kernels used to be ok until they started being produced
on a system that had the compiler updated.  this happened after 2.4.18-27.7.x
all kernels after that have not been stable under conditions such as high
network load for us and most of our stuff is back to 2.4.18-27.7.x.

i don't know whether there is or will be a fix for this.

regards,

-jason




More information about the Linux-PowerEdge mailing list