Buildsys (i386) oddity + segmentation fault

[Dan Williams]

On Fri, 2006-05-26 at 15:00 +0100, Jose Pedro Oliveira wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Ville Skyttä wrote:
> > There seems to be something strange with the buildsys (i386/devel): I
> > made some innocent changes to xmms and requested a build, but the i386
> > build seems to get repeatedly stuck in running/building state.
> > 
> > It builds fine in the x86_64 and ppc builders in the buildsys and I
> > can't reproduce any weirdness locally on i386.
> > 
> > I've killed/requeued the job already twice, but that hasn't helped:
> > http://buildsys.fedoraproject.org/build-status/job.psp?uid=9945
> > 
> > Could someone take a look?  TIA!
> > 
> 
> 
> And also at this segmentation fault?
> 
> Job failed on arch i386
> http://buildsys.fedoraproject.org/logs/fedora-development-extras/9911-multitail-4.0.4-1.fc6/
> - -------------------------------------------------
> ...
> cc -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions
> - -fstack-protector --param=ssp-buffer-size=4 -m32 -march=i386
> - -mtune=generic -fasynchronous-unwind-tables -D`uname` -O2 -Wall
> - -DVERSION=\"4.0.4\" -g  -DCONFIG_FILE=\"//etc/multitail.conf\"   -c -o
> scrollback.o scrollback.c
> make: *** [scrollback.o] Segmentation fault
> make: *** Waiting for unfinished jobs....
> error: Bad exit status from /var/tmp/rpm-tmp.26412 (%build)

I suppose it would be useful if we set coredump limits higher in the
chroot, and if the buildsys detected the message "segmentation fault"
and if the job failed, to keep the chroot around for further analysis.
Right? :)

Dan