Re: [PATCH/RFC] Signals & single-stepping

[Daniel Jacobowitz <drow@false.org>]

On Wed, Sep 30, 2009 at 04:40:57PM +0200, Mark Kettenis wrote:
> A while ago Daniel made a change to infcmd.c which broke hitting
> breakpoints in signal handlers while single-stepping on OpenBSD (and
> probably other ptrace-based native targets):
> 
> 2009-01-20  Daniel Jacobowitz  <dan@codesourcery.com>
> 
>         PR gdb/9346
>         * infcmd.c (signal_command): Do not specify a resume PC.
> 
> The problem is that breakpoints are removed for single-stepping.  So
> the diff below fixes things.  Now it isn't clear to me if this
> approach doesn't reintroduce the problem from the PR that Daniel tried
> to fix.
> 
> Comments?

Your patch doesn't reintroduce the problem from the PR, and the new
tests in interrupt.exp pass on x86_64-linux.  I would really love
someone else to volunteer to review it though - trap_expected confuses
me horribly.  I'd guess this change could lead to hitting (and
displaying) the breakpoint at the current PC a second time, which is
undesirable.  On the other hand it might fix some of the signal tests
on software single-step targets...

> 
> 
> Index: infrun.c
> ===================================================================
> RCS file: /cvs/src/src/gdb/infrun.c,v
> retrieving revision 1.410
> diff -u -p -r1.410 infrun.c
> --- infrun.c	29 Sep 2009 00:53:04 -0000	1.410
> +++ infrun.c	30 Sep 2009 12:55:46 -0000
> @@ -1604,6 +1604,11 @@ proceed (CORE_ADDR addr, enum target_sig
>    else if (!signal_program[tp->stop_signal])
>      tp->stop_signal = TARGET_SIGNAL_0;
>  
> +  if (tp->stop_signal != TARGET_SIGNAL_0) {
> +	  tp->trap_expected = 0;
> +	  insert_breakpoints ();
> +  }
> +
>    annotate_starting ();
>  
>    /* Make sure that output from GDB appears before output from the
> 

-- 
Daniel Jacobowitz
CodeSourcery