This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [patch, rfc, rft] Multi-threaded single-step vs. breakpoint problems (prepare_to_proceed)
- From: Daniel Jacobowitz <drow at false dot org>
- To: Ulrich Weigand <uweigand at de dot ibm dot com>
- Cc: gdb-patches at sourceware dot org
- Date: Thu, 2 Aug 2007 15:42:20 -0400
- Subject: Re: [patch, rfc, rft] Multi-threaded single-step vs. breakpoint problems (prepare_to_proceed)
- References: <200708011952.l71JqO6e015264@d12av02.megacenter.de.ibm.com>
On Wed, Aug 01, 2007 at 09:52:21PM +0200, Ulrich Weigand wrote:
> The main changes to prepare_to_proceed are do only ever switch threads
> if we stopped due to hitting a breakpoint which is still set, and the
> user switched threads in the meantime. If we're also about to single-
> step, we remember which thread we were supposed to step, and switch
> back to it in handle_inferior_event after we're stepped past that
> breakpoint.
I've no objection and the patch seems fine to me. I wonder what
happens if the other thread hit a watchpoint, or some other
non-breakpoint event, though; on some platforms watchpoints must be
stepped past similarly.
--
Daniel Jacobowitz
CodeSourcery