This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [RFA] Don't reset watchpoint block on solib load.
- From: Daniel Jacobowitz <dan at codesourcery dot com>
- To: Vladimir Prus <vladimir at codesourcery dot com>
- Cc: Jim Blandy <jimb at codesourcery dot com>, gdb-patches at sources dot redhat dot com
- Date: Tue, 29 Jan 2008 10:17:15 -0500
- Subject: Re: [RFA] Don't reset watchpoint block on solib load.
- References: <200711202013.47537.vladimir@codesourcery.com> <200711281858.51145.vladimir@codesourcery.com> <m38x2qh83j.fsf@codesourcery.com> <200801231257.27977.vladimir@codesourcery.com>
On Wed, Jan 23, 2008 at 12:57:27PM +0300, Vladimir Prus wrote:
> So, I guess the solutions are either:
>
> 1. Omit the move of condition parsing from this patch, or
> 2. Keep condition parsing in this patch, and then commit
> the use-multiple-locations-for-watchpoint patch which will
> immediately make condition only reparsed when breakpoint_re_set_one
> is called.
>
> Given that use-multiple-locations-for-watchpoint patch is
> approved except for the bit that makes reparsing of condition
> optional, it seems that (2) is less work.
All this sounds fine to me. I'm OK with this patch, which I believe
is the one here:
http://sourceware.org/ml/gdb-patches/2007-11/msg00522.html
> For convenience, I attach the current version of
> use-multiple-locations-for-watchpoint patch. The difference
> to the previous revision is as follows:
>
>
> @@ -966,7 +966,7 @@ update_watchpoint (struct breakpoint *b, int reparse)
> value_free (v);
> }
>
> - if (b->cond_string != NULL)
> + if (reparse && b->cond_string != NULL)
> {
> char *s = b->cond_string;
> if (b->loc->cond)
>
>
> Dan, is the patch still OK to commit?
Yes, it is.
--
Daniel Jacobowitz
CodeSourcery