This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [patch] "single step" atomic instruction sequences as a whole.
- From: "Ulrich Weigand" <uweigand at de dot ibm dot com>
- To: drow at false dot org (Daniel Jacobowitz)
- Cc: gdb-patches at sourceware dot org
- Date: Thu, 12 Apr 2007 22:09:10 +0200 (CEST)
- Subject: Re: [patch] "single step" atomic instruction sequences as a whole.
Daniel Jacobowitz wrote:
> On Thu, Apr 12, 2007 at 07:16:14PM +0200, Ulrich Weigand wrote:
> > I would think the "write_pc (next_pc)" statement in the
> > !insert_breakpoints_p case should be unnecessary. This should
> > already have been taken care of by infrun, shouldn't it?
>
> I'm afraid I don't know what this is for. You're probably correct,
> though.
I'd say we should remove it. The use of the contents of the
static variable next_pc from a previous invocation strikes me
as suspect anyway -- what if we're in another thread now?
I don't have a way to test on alpha-linux unfortunately. Do you?
> > Finally, all single-step implementations currently have to
> > fall back to global functions like read_pc (or current_regcache)
> > to find the target registers. I understand this is something
> > we should be moving away from, so if we're already changing
> > the signature, maybe we should pass in a regcache argument?
>
> Hmm, or a frame?
I thought "resume" (where this is called) too low-level for
a frame to make sense. You cannot single-step anywhere but
in the innermost frame, so a regcache seemed more appropriate.
(How would I actually get hold of a frame in resume?)
Bye,
Ulrich
--
Dr. Ulrich Weigand
GNU Toolchain for Linux on System z and Cell BE
Ulrich.Weigand@de.ibm.com