This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [commit] build failure in inf-ptrace.c (pa-hpux)
On Wednesday 30 December 2009 20:07:20, Mark Kettenis wrote:
> > Date: Wed, 30 Dec 2009 23:46:28 +0400
> > From: Joel Brobecker <brobecker@adacore.com>
> >
> > Hello,
> >
> > inf-ptrace no longer builds on pa-hpux. Looks like a small thinko
> > while updating gazillions of files the same way. I hope I fixed it
> > right (notice the [...]->pspace = [...]->aspace on the second line
> > I changed ;-). I didn't see I first fixed the name of the variable...
> >
> > 2009-12-30 Joel Brobecker <brobecker@adacore.com>
> >
> > Fix build failure in inf-ptrace.c.
> > * inf-ptrace.c (inf_ptrace_follow_fork): Fix typo.
> >
> > Checked in.
Thanks.
On Wednesday 30 December 2009 20:07:20, Mark Kettenis wrote:
> > Date: Wed, 30 Dec 2009 23:46:28 +0400
> > From: Joel Brobecker <brobecker@adacore.com>
> For what it is worth, it also broke OpenBSD and I came up with the
> same fix.
You could have said something sooner. ;-)
> I'm not sure the fix is right though.
>
> > + child_inf->pspace = parent_inf->pspace;
> > + child_inf->aspace = parent_inf->aspace;
>
> Looking at the Linux code, this seesm to be right for vfork(), but the
> code in inf-ptrace.c handles fork().
> Pedro, Stan, what would be the right thing to do here?
It's also what linux does in the "set detach-on-fork on" (default) case
when following the child, notice the `|| detach_fork'
condition there. Neither OpenBSD nor pa-hpux implement
"set detach-on-fork off", that is, don't handle debugging more
than inferior simultaneouly, so this is fine. I assume one of
you confirmed following a fork works as good as it use to.
--
Pedro Alves