Re: [PATCH 11/12] Use reinsert_breakpoint for vCont;s

[Pedro Alves <palves at redhat dot com>]

On 06/14/2016 02:14 PM, Yao Qi wrote:
> Pedro Alves <palves@redhat.com> writes:
> 
>>> @@ -4293,7 +4313,7 @@ linux_resume_one_lwp_throw (struct lwp_info *lwp,
>>>  
>>>        step = maybe_hw_step (thread);
>>>      }
>>> -  else
>>> +  else if (lwp->resume != NULL && lwp->resume->kind != resume_step)
>>>      {
>>>        /* If the thread isn't doing step-over, there shouldn't be any
>>>  	 reinsert breakpoints.  */
>>
>> Consider (non-stop RSP):
>>
>>  -> vCont;s:1
>>  <- OK
>>  -> vCont;s:2
>>  <- OK
>>
>> The handling of the second vCont sets thread 1's lwp->resume to NULL.
> 
> If so, the assert won't be called for thread 1.
> 
>> The lwp->resume pointer is only meaningful within linux_resume
>> and its callees.  (But this function is called in other contexts.)
>>
> 
> When I wrote the patch, it took me a while to think about this condition
> check.  I wanted to remove this condition and assert, but finally
> decided to leave it there, as it is not harmful.  If lwp->resume is only
> meaningful within linux_resume and its callees, how about remove the
> condition check and assert here?

Yes, if it's only for the assert, then let's remove it.

> 
>>> @@ -5009,12 +5033,52 @@ linux_resume (struct thread_resume *resume_info, size_t n)
>>>  	debug_printf ("Resuming, no pending status or step over needed\n");
>>>      }
>>>  
>>> +  /* Before we resume the threads, if resume_step is requested by GDB,
>>> +     stop all threads and install reinsert breakpoints.  */
>>
>> Looking again, I think the rationale for stopping threads should
>> be mentioned here, as it's not obvious.
>>
> 
> How about this,
> 
>   /* Before we resume the threads, if resume_step is requested by GDB,
>      we need to access the inferior memory to install reinsert
>      breakpoints, so stop all threads.  */

That doesn't tell the reader why we need to stop _all_ threads.  The
threads that are about to be resumed are obviously stopped, and
thus we could already _access_ inferior memory through them.

I guess this is about flushing instruction caches?

>>> @@ -5176,6 +5241,30 @@ proceed_all_lwps (void)
>>>    if (debug_threads)
>>>      debug_printf ("Proceeding, no step-over needed\n");
>>>  
>>> +  /* Re-install the reinsert breakpoints on software single step target
>>> +     if the client wants it step.  */
>>> +  if (can_software_single_step ())
>>
>> Not immediately obvious to why is this necessary.  Where were they
>> removed in the first place?  I'm it must be necessary, but maybe
>> extending the comment helps.
> 
> How about this
> 
>   /* On software single step target, we removed reinsert breakpoints
>      after we get any events from the inferior.  

Is that all events, even internal events?  From the patch, it seemed
like it was only before reporting an event to gdb.

> If the client wants
>      thread step, re-install these reinsert breakpoints.  */
> 

If we only remove before reporting an event to gdb, then I don't
understand this.  We already insert single-step breakpoints when
we process the resume request from gdb, no?

Thanks,
Pedro Alves