Segfaults on powerpc platform

Panu Matilainen pmatilai at laiskiainen.org
Tue Apr 17 07:15:53 PDT 2007


On Tue, 17 Apr 2007, Tim Verhoeven wrote:

> Hi,
>
> I'm trying to run apt-rpm on a powerpc server but I'm having some
> issues. The OS on the server is RHEL4 and I've recompiled Dag's apt
> package on that system. That works fine.
>
> Now most apt command segfault, but some don't. For example doing a
> apt-get update works fine. It looks like doing any kind of transaction
> with the RPM databases it self will cause a segfault (installing
> packages, updating, removing, ...).
>
> Has anyone have a idea where to look or the best way to debug this ?
> I've tried both the latest stable (org 3.2) as the latest test release
> (org 3.90). I don't think it is a 64 bit issue since we have apt-rpm
> working fine on 32 and 64 Intel systems.

I'm not exactly surprised to hear about problems with PPC64:
1) it's totally untested as I don't have access to ppc hardware
2) it's kinda reverse from all other multilib systems (most userland is 
32bit although it's seen as 64bit platform)

Because of 2) almost everything on ppc64 ends up being foo.32bit virtual 
packages kludgery which don't behave quite the same as real ones. So 
things just might, well, blow up. However I would've expected grand royale 
confusion with dependencies, not segfaults.

The first step is to get backtraces from the crash(es), preferrably with 
apt-debuginfo package installed:
# gdb apt-get
(gdb) run <apt command here, eg "install pkg">
..and when it crashes, do
(gdb) bt

and send the output of the entire gdb session here.

 	- Panu -



More information about the Apt-Rpm mailing list