This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [RFC/RFA/i386] pb reading insns if breakpoints still inserted
- From: Daniel Jacobowitz <drow at false dot org>
- To: Mark Kettenis <mark dot kettenis at xs4all dot nl>
- Cc: brobecker at adacore dot com, gdb-patches at sources dot redhat dot com
- Date: Fri, 5 May 2006 14:16:13 -0400
- Subject: Re: [RFC/RFA/i386] pb reading insns if breakpoints still inserted
- References: <20060428171154.GP17613@adacore.com> <200605011636.k41GaO8I027381@elgar.sibelius.xs4all.nl>
On Mon, May 01, 2006 at 06:36:24PM +0200, Mark Kettenis wrote:
> Hmm, I'm not entirely happy with the name, and I agree with jimb that
> using current_frame here doesn't make sense. I was planning to look
> into getting rid of the prologue skipper, but that isn't really going
> to solve the problems.
>
> After another day of thinking I came to the conclusion that jimb is
> probably right about the premature deprecation of memory_read_nobpt.
> If we undeprecate it, you wouldn't really need your new read_insn
> function, since it really would be the same as memory_read_nobpt.
>
> What do the other (global) maintainers think?
I see two choices.
1. Undeprecate deprecated_read_memory_nobpt. Use it.
2. Make all memory reads skip breakpoints by default, and provide a
different function which doesn't.
I think (1) is the best choice for now. (2) is going to require a lot
of poking around dark corners and testing on strange targets. I'd
recommend anyone interested start by cutting down on the number of
memory reading functions - there's quite a lot of them!
So, I'm in favor.
--
Daniel Jacobowitz
CodeSourcery