On Thu, May 01, 2014 at 05:21:26PM +0200, Otto Moerbeek wrote:
> On Thu, May 01, 2014 at 09:18:07AM -0600, Theo de Raadt wrote:
>
> > > On Thu, May 01, 2014 at 05:03:12PM +0200, Peter J. Philipp wrote:
> > >
> > > > Hi,
> > > >
> > > > I recently bought a new computer and it runs OpenBSD (latest snapshot,
> > > > -current) natively. Everything is fine except a program I develop on
> > > > and it crashes according to gdb with a SIGBUS.
> > > >
> > > > When I run this program on another amd64 computer (vmware fusion on mac,
> > > > OpenBSD 5.5-stable), I do not get the SIGBUS's and the program behaves
> > > > normally.
> > > >
> > > > So I'm wondering why no coredumps? SIGBUS is supposed to dump core.
> > > >
> > > > I have:
> > > >
> > > > # ls /var/crash
> > > > minfree
> > > > # sysctl -a|grep suid
> > > > kern.nosuidcoredump=2
> > > > # ls -ld /var/crash
> > > > drwxrwxrwt 2 root wheel 512 Apr 30 21:05 /var/crash
> > > >
> > > > is this not enough to make my program which setresuid()'s after fork,
> > > > core?
> > >
> > > your program also has to be running in a dir where it can write. It
> > > will not automatically write to /var/crash, that's for kernel dumps.
> >
> > Not true. He is using nosuidcoredump=2. And it appears it got broken
> > a while back.
> >
> > I am working on something even better, but not willing to share it yet :-)
>
> Oops, wasn't paying attention.
>
> -Otto
For solving one problem (debugging the program), you could run
directly from within gdb or attach gdb while it's running.
-Otto
