Re: [PATCH v3 5/7] fix issue: gdb hangs in the command following a commad returning with TARGET_WAITKIND_NO_HISTORY

[Luis Machado <luis.machado@linaro.org>]

On 3/30/21 11:52 PM, Zied Guermazi wrote:
> 
> This patch fixes an issue observed with btrace when replaying the execution.
> The issue was observed on ARMv7 processors.
> To reproduce the issue, the user needs to replay in forward direction
> until he reaches the end of history, then replay backwards
> (e.g a reverse-next) and then replay forwards. gdb hangs and the user
> can not issue new commands.
> This fix keeps the same behaviour of gdb as when TARGET_WAITKIND_NO_HISTORY
> is hit on other architectures.
> 
> gdb/ChangeLog
> 
> 	* infrun.c (set_step_over_info): add debug print.
> 	(handle_inferior_event): clear step over info
> 	in case TARGET_WAITKIND_NO_HISTORY
> 
> ---
>   gdb/ChangeLog | 6 ++++++
>   gdb/infrun.c  | 3 ++-
>   2 files changed, 8 insertions(+), 1 deletion(-)
> 
> diff --git a/gdb/ChangeLog b/gdb/ChangeLog
> index d22ec8b5976..d4ed6fe0255 100644
> --- a/gdb/ChangeLog
> +++ b/gdb/ChangeLog
> @@ -1,3 +1,9 @@
> +2021-02-25  Zied Guermazi  <zied.guermazi@trande.de>
> +
> +	* infrun.c (set_step_over_info): add debug print.
> +	(handle_inferior_event): clear step over info
> +	in case TARGET_WAITKIND_NO_HISTORY
> +
>   2021-02-25  Zied Guermazi  <zied.guermazi@trande.de>
>   
>   	* nat/linux-btrace.h (btrace_tinfo_etm): New.
> diff --git a/gdb/infrun.c b/gdb/infrun.c
> index a271220b261..9c2522376f2 100644
> --- a/gdb/infrun.c
> +++ b/gdb/infrun.c
> @@ -1308,6 +1308,7 @@ set_step_over_info (const address_space *aspace, CORE_ADDR address,
>   		    int nonsteppable_watchpoint_p,
>   		    int thread)
>   {
> +  infrun_debug_printf ("setting step over info");
>     step_over_info.aspace = aspace;
>     step_over_info.address = address;
>     step_over_info.nonsteppable_watchpoint_p = nonsteppable_watchpoint_p;
> @@ -5536,7 +5537,7 @@ handle_inferior_event (struct execution_control_state *ecs)
>         delete_just_stopped_threads_single_step_breakpoints ();
>         ecs->event_thread->suspend.stop_pc
>   	= regcache_read_pc (get_thread_regcache (inferior_thread ()));
> -
> +      clear_step_over_info ();
>         if (handle_stop_requested (ecs))
>   	return;
>   
> 

I'll defer this to folks more familiar with the infrun code, but this is 
one area that needs to be exercised carefully, as it is really complex 
with many execution paths.