Re: [PATCH 7/8] Resume the inferior with signal rather than stepping over

[Yao Qi <qiyaoltc@gmail.com>]

Pedro Alves <palves@redhat.com> writes:

> Once the handler returns, it'll retrap the same breakpoint we already stopped
> for, and thus I think we'll count a spurious tracepoint/breakpoint hit.
>
> Sounds like we'll need to do like GDB does and remember that we are advancing
> past a signal handler, and need to get back to stepping over the breakpoint
> if/when the handler returns successfully?

I think it is no longer an issue as we live with the spurious double
trap, right?

>>   
>> +  /* On software single step target, resume the inferior with signal
>> +     rather than stepping over.  */
>> +  if (can_software_single_step ()
>> +      && lwp->pending_signals != NULL
>> +      && LWP_SIGNAL_CAN_BE_DELIVERED (lwp))
>> +    {
>> +      if (debug_threads)
>> +	debug_printf ("Need step over [LWP %ld]? Ignoring, has pending"
>> +		      " signals.\n",
>> +		      lwpid_of (thread));
>> +
>> +      return 0;
>
> I notice this missed restoring the current thread (below).
>

The code below is to switch current_thread to thread, not restoring.
Since we don't switch current_thread here, we don't have to do anything.

>> +    }
>> +
>>     saved_thread = current_thread;
>>     current_thread = thread;

-- 
Yao (齐尧)