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: luisgpm at linux dot vnet dot ibm dot com (Luis Machado), gdb-patches at sourceware dot org
- Date: Thu, 12 Apr 2007 16:58:21 +0200 (CEST)
- Subject: Re: [patch] "single step" atomic instruction sequences as a whole.
Daniel Jacobowitz wrote:
> On Thu, Apr 12, 2007 at 10:25:26AM -0300, Luis Machado wrote:
> > The typos were corrected and the gdbarch.[c|h] files were updated with
> > gdbarch.sh, though the comment was just added to gdbarch.h, not
> > gdbarch.c, is this correct?
>
> Yes, it is. This version is fine.
I've checked this in now.
While we're on this topic, I'm wondering whether we could do some
further simplification on the single-step code. In particular,
now every single-step implementation makes use of the
insert_single_step_breakpoint helper to insert breakpoints,
and the !insert_breakpoint_p side of every single-step implementation
consist of a sole call to the remove_single_step_breakpoints helper.
It might be a good idea to actually enforce that behaviour. In fact,
we could just *remove* the insert_breakpoint_p flag of the gdbarch
callback, call that callback solely in insert the breakpoints (which
*must* use the insert_single_step_breakpoint helper), and just call
remove_single_step_breakpoints directly from common code instead of
the gdbarch callback to remove the breakpoints.
That would allow future changes to the details of single-step
breakpoint implementation in common code without having to go
through all implementations every time ...
What do you think?
Bye,
Ulrich
--
Dr. Ulrich Weigand
GNU Toolchain for Linux on System z and Cell BE
Ulrich.Weigand@de.ibm.com