Re: [RFA] dummy frame handling cleanup, plus inferior fun call signal handling improvement

[Doug Evans <dje at google dot com>]

On Tue, Nov 18, 2008 at 4:58 AM, Doug Evans <dje@google.com> wrote:
> Hi.  This patch was in progress when I wrote
> http://sourceware.org/ml/gdb-patches/2008-11/msg00391.html
> This patch subsumes that one.
>
> There are two things this patch does:
> 1) properly pop dummy frames more often
> 2) make the inferior resumable after an inferior function call gets a
>   signal without having to remember to do "signal 0"
>
> 1) properly pop dummy frames more often
>
> Ulrich asked:
>> I agree that it would be better if call_function_by_hand were to call
>> dummy_frame_pop in this case.  However, why exactly is this a bug?
>
> It's a bug because it's confusing to not pop the frame in the places
> where one is able to.  Plus it's, well, unclean.
> I recognize that the dummy frame stack can't be precisely managed because
> the user can do things like:
>
> set $orig_pc = $pc
> set $orig_sp = $sp
> break my_fun
> call my_fun()
> set $pc = $orig_pc
> set $sp = $orig_sp
> continue
>
> [This doesn't always work, but you get the idea.]
> At the "continue", the inferior function call no longer exists and
> gdb's mechanisms for removing the dummy frame from dummy_frame_stack
> will never trigger (until the program is resumed and cleanup_dummy_frames
> is called).  But we can still keep things clean
> and unconfusing for the common case.

For completeness' sake,
To catch cases like these one could do things like compare $sp with
$dummy_frame_stack.*.regcache.sp every time execution stops.