This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: thanks for the memory work
> X-Original-To: geoffk@foam.wonderslug.com
> From: Steven Bosscher <s.bosscher@student.tudelft.nl>
> Date: Thu, 27 Nov 2003 00:57:12 +0100
> On Wednesday 26 November 2003 20:00, Geoff Keating wrote:
> > Steven Bosscher <s.bosscher@student.tudelft.nl> writes:
> > > On Monday 24 November 2003 17:05, Jeff Law wrote:
> > > > For example, I'd like to ask the GC system "do you have any PHI nodes
> > > > which are marked? If so, what was the path by which they were
> > > > marked?".
> > >
> > > It's doable to tell _if_ any PHI nodes are marked, but it's very hard if
> > > at all possible to tell _when_ and along which path.
> >
> > Have you tried setting a breakpoint in the appropriate GC-marking
> > routine, and asking GDB what the backtrace was?
>
> Yes. gdb won't let me put breakpoints in the marking routines, claiming that
> the gt-* header files are not source files :-/
My usual response to that kind of GDB behaviour is to disassemble the
routine (using 'x/i gt_mx_foo') and place a breakpoint at the
instruction I want (using 'break *0x1234...). It'd probably be better
to work out why GDB is doing this and fix it, though. I'm pretty sure
GDB doesn't do it to me, but I use the Darwin GDB, which uses stabs
and which probably isn't very similar to the GDB you're using.
--
- Geoffrey Keating <geoffk@geoffk.org>