Re: [PATCH v3 6/6] Fix reverse-step and reverse-next over undebuggable solib code

[Pedro Alves <palves at redhat dot com>]

(This patch doesn't look specific to ARM, and thus could/should
be split out from the rest of the series.)

On 08/13/2014 02:12 PM, Omair Javaid wrote:
> This patch fixes failures to reverse-step or reverse-next over solib functions in absence of line information.

Could you describe the failure a little more please?  What happens instead?

> The problem is fixed by making sure that in solib functions we keep doing reverse single stepping in absence of line information.
> 
> Tested with no regressions on arm, aarch64 and x86_64.

Is there a test in the testsuite that fails before this patch?
If not, could you add one?

There's another similarly looking piece of code above
with the same comment: "Reverse stepping through solib trampolines
that I'm wondering whether it needs treatment as well.

Thanks!

> 
> gdb:
> 
> 2014-08-13  Omair Javaid  <omair.javaid@linaro.org>
> 
> 	* infrun.c (process_event_stop_test): Updated.

Same remark as previous patches here.

> 
> ---
>  gdb/infrun.c | 5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
> 
> diff --git a/gdb/infrun.c b/gdb/infrun.c
> index c18267f..db8f15b 100644
> --- a/gdb/infrun.c
> +++ b/gdb/infrun.c
> @@ -4905,12 +4905,15 @@ process_event_stop_test (struct execution_control_state *ecs)
>        return;
>      }
>  
> +  stop_pc_sal = find_pc_line (stop_pc, 0);
> +
>    /* Reverse stepping through solib trampolines.  */
>  
>    if (execution_direction == EXEC_REVERSE
>        && ecs->event_thread->control.step_over_calls != STEP_OVER_NONE)
>      {
>        if (gdbarch_skip_trampoline_code (gdbarch, frame, stop_pc)
> +	  || stop_pc_sal.line == 0
>  	  || (ecs->stop_func_start == 0
>  	      && in_solib_dynsym_resolve_code (stop_pc)))
>  	{
> @@ -4939,8 +4942,6 @@ process_event_stop_test (struct execution_control_state *ecs)
>  	}
>      }
>  
> -  stop_pc_sal = find_pc_line (stop_pc, 0);
> -
>    /* NOTE: tausq/2004-05-24: This if block used to be done before all
>       the trampoline processing logic, however, there are some trampolines 
>       that have no names, so we should do trampoline handling first.  */
> 

Thanks,
Pedro Alves