Re: Dumping core on SIGSEGV

[Owen Taylor <otaylor redhat com>]

beach@ataman.com writes:

> On Fri, Oct 16, 1998 at 01:52:22PM -0400, Jason A. Pfeil wrote:
> > Ummm...if gtk DOESN'T catch SIGSEGV, then the core IS dumped.  What seems
> > to be the problem here?  Signals are up to the application to catch
> > anyway.
> > 
> > --Jason
> > 
> > On Fri, 16 Oct 1998 beach@ataman.com wrote:
> > 
> > > 
> > > Why doesn't Gtk catch SIGSEGV instead of allowing a core dump?  This
> > > is making it difficult to use in development at times (e.g. no
> > > coredump files to work with means nothing to give to the debugger.)
> 
> Please excuse my horribly phrased question.  What I meant to say was:
> 
> Why *does* Gtk catch SIGSEGV.  I'd like it to coredump so that I can
> use the debugger.  What should I do?

Compile GTK+ with --enable-debug - that causes the SIGSEGV
handler to abort() instead of exit()'ing, which should
give you a core file.

[ And if you are developing with GTK+, you really, really,
  really, should be compiling with --enable-debug. ]

It catches SIGSEGV, mostly, because if it doesn't, on some
platforms it will leak shared memory segments. 

Regards,
                                        Owen