[7/8] Use reinsert_breakpoint for vCont;s

  This patch is to teach GDBserver using software single step to handle
vCont;s.  Simply speaking, if the thread's resume request is resume_step,
install reinsert breakpoint at the next pcs when GDBserver is about to
resume threads.  These reinsert breakpoints of a thread are removed,
when GDBserver gets an event from that thread.  Note that GDBserver may
or may not report this event back to GDB.

gdb/gdbserver:

2016-05-20  Yao Qi  <yao.qi@linaro.org>

	* linux-low.c (resume_stopped_resumed_lwps): If resume request
	is resume_step, either set step to 1 or assert that reinsert
	breakpoints are installed.
	(linux_wait_1): Stop all threads, remove reinsert breakpoints,
	and unstop them.
	(linux_resume_one_thread): If resume request is resume_step,
	either set step to 1 or assert that reinsert breakpoints are
	installed.
	(linux_resume): Install reinsert breakpoints if the thread is
	requested to resume_step.
	(proceed_one_lwp): If resume request is resume_step,
        either set step to 1 or assert that reinsert breakpoints are
        installed.
	(proceed_all_lwps): Install reinsert breakpoints if the thread is
        requested to resume_step.
---
 gdb/gdbserver/linux-low.c | 125 ++++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 121 insertions(+), 4 deletions(-)
  

On 05/20/2016 04:12 PM, Yao Qi wrote:
> This patch is to teach GDBserver using software single step to handle
> vCont;s.  Simply speaking, if the thread's resume request is resume_step,
> install reinsert breakpoint at the next pcs when GDBserver is about to
> resume threads.  These reinsert breakpoints of a thread are removed,
> when GDBserver gets an event from that thread.  Note that GDBserver may
> or may not report this event back to GDB.
> 
> gdb/gdbserver:
> 
> 2016-05-20  Yao Qi  <yao.qi@linaro.org>
> 
> 	* linux-low.c (resume_stopped_resumed_lwps): If resume request
> 	is resume_step, either set step to 1 or assert that reinsert
> 	breakpoints are installed.
> 	(linux_wait_1): Stop all threads, remove reinsert breakpoints,
> 	and unstop them.
> 	(linux_resume_one_thread): If resume request is resume_step,
> 	either set step to 1 or assert that reinsert breakpoints are
> 	installed.
> 	(linux_resume): Install reinsert breakpoints if the thread is
> 	requested to resume_step.
> 	(proceed_one_lwp): If resume request is resume_step,
>         either set step to 1 or assert that reinsert breakpoints are
>         installed.
> 	(proceed_all_lwps): Install reinsert breakpoints if the thread is
>         requested to resume_step.

Spaces vs tabs, I think.

> ---
>  gdb/gdbserver/linux-low.c | 125 ++++++++++++++++++++++++++++++++++++++++++++--
>  1 file changed, 121 insertions(+), 4 deletions(-)
> 
> diff --git a/gdb/gdbserver/linux-low.c b/gdb/gdbserver/linux-low.c
> index d5f7097..22cf51a 100644
> --- a/gdb/gdbserver/linux-low.c
> +++ b/gdb/gdbserver/linux-low.c
> @@ -2526,7 +2526,18 @@ resume_stopped_resumed_lwps (struct inferior_list_entry *entry)
>        && !lp->status_pending_p
>        && thread->last_status.kind == TARGET_WAITKIND_IGNORE)
>      {
> -      int step = thread->last_resume_kind == resume_step;
> +      int step = 0;
> +
> +      if (thread->last_resume_kind == resume_step)
> +	{

...

> +	  if (can_hardware_single_step ())
> +	    step = 1;
> +	  else
> +	    {
> +	      /* GDBserver must insert reinsert breakpoint for vCont;s.  */
> +	      gdb_assert (has_reinsert_breakpoints (thread));
> +	    }

This pattern seems to appear in many places in the series.
Maybe we could add a function to wrap it?  Something like:

static int
maybe_hw_step (struct thread_info *thread)
{
  if (can_hardware_single_step ())
    return 1;
  else
    {
      gdb_assert (has_reinsert_breakpoints (thread));
      return 0;
    }
}

?

> +	}
>  
>        if (debug_threads)
>  	debug_printf ("RSRL: resuming stopped-resumed LWP %s at %s: step=%d\n",
> @@ -3477,6 +3488,18 @@ linux_wait_1 (ptid_t ptid,
>        return ignore_event (ourstatus);
>      }
>  
> +  /* Remove reinsert breakpoints for resume_step.  */
> +  if (can_software_single_step ()
> +      && current_thread->last_resume_kind == resume_step

What if a vCont;t came in after the step had started and
before it finished?  Won't we have lost the resume_step 
linux_set_resume_request then?


> +      && has_reinsert_breakpoints (current_thread))
> +    {
> +      stop_all_lwps (1, event_child);
> +
> +      delete_reinsert_breakpoints (current_ptid);
> +
> +      unstop_all_lwps (1, event_child);
> +    }
> +
>    /* Note that all addresses are always "out of the step range" when
>       there's no range to begin with.  */
>    in_step_range = lwp_in_step_range (event_child);
> @@ -4796,7 +4819,6 @@ linux_resume_one_thread (struct inferior_list_entry *entry, void *arg)
>  {
>    struct thread_info *thread = (struct thread_info *) entry;
>    struct lwp_info *lwp = get_thread_lwp (thread);
> -  int step;
>    int leave_all_stopped = * (int *) arg;
>    int leave_pending;
>  
> @@ -4865,10 +4887,22 @@ linux_resume_one_thread (struct inferior_list_entry *entry, void *arg)
>  
>    if (!leave_pending)
>      {
> +      int step = 0;
> +
>        if (debug_threads)
>  	debug_printf ("resuming LWP %ld\n", lwpid_of (thread));
>  
> -      step = (lwp->resume->kind == resume_step);
> +      if (lwp->resume->kind == resume_step)
> +	{
> +	  if (can_hardware_single_step ())
> +	    step = 1;
> +	  else
> +	    {
> +	      /* GDBserver must insert reinsert breakpoint for vCont;s.  */
> +	      gdb_assert (has_reinsert_breakpoints (thread));
> +	    }
> +	}
> +
>        linux_resume_one_lwp (lwp, step, lwp->resume->sig, NULL);
>      }
>    else
> @@ -4909,6 +4943,7 @@ linux_resume (struct thread_resume *resume_info, size_t n)
>    struct thread_info *need_step_over = NULL;
>    int any_pending;
>    int leave_all_stopped;
> +  int resume_step_is_handled = 0;
>  
>    if (debug_threads)
>      {
> @@ -4952,12 +4987,57 @@ 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.  */
> +  if (!leave_all_stopped && can_software_single_step ())
> +    {
> +      struct inferior_list_entry *inf, *tmp;
> +
> +      if (debug_threads)
> +	debug_printf ("Handle resume_step.\n");
> +
> +      ALL_INFERIORS (&all_threads, inf, tmp)
> +	{
> +	  struct thread_info *thread = (struct thread_info *) inf;
> +	  struct lwp_info *lwp = get_thread_lwp (thread);
> +
> +	  if (lwp->resume != NULL && lwp->resume->kind == resume_step)
> +	    {
> +	      struct thread_info *saved_thread;
> +
> +	      if (!resume_step_is_handled)
> +		{
> +		  stop_all_lwps (0, NULL);
> +
> +		  if (debug_threads)
> +		    debug_printf ("Done stopping all threads.\n");
> +
> +		  resume_step_is_handled = 1;
> +		}
> +
> +	      saved_thread = current_thread;
> +	      current_thread = thread;
> +	      install_software_single_step_breakpoints (lwp);
> +	      current_thread = saved_thread;
> +

This current thread save/restore should move to
install_software_single_step_breakpoints, as per the other patch.


> +	      if (debug_threads)
> +		debug_printf ("Insert breakpoint for resume_step LWP %ld\n",
> +			      lwpid_of (thread));
> +	    }
> +	}
> +
> +      if (debug_threads)
> +	debug_printf ("Handle resume_step.  Done\n");
> +    }
> +
>    /* Even if we're leaving threads stopped, queue all signals we'd
>       otherwise deliver.  */
>    find_inferior (&all_threads, linux_resume_one_thread, &leave_all_stopped);
>  
>    if (need_step_over)
>      start_step_over (get_thread_lwp (need_step_over));
> +  else if (resume_step_is_handled)
> +    unstop_all_lwps (0, NULL);
>  
>    if (debug_threads)
>      {
> @@ -5054,7 +5134,14 @@ proceed_one_lwp (struct inferior_list_entry *entry, void *except)
>        if (debug_threads)
>  	debug_printf ("   stepping LWP %ld, client wants it stepping\n",
>  		      lwpid_of (thread));
> -      step = 1;
> +
> +      if (can_hardware_single_step ())
> +	step = 1;
> +      else
> +	{
> +	  /* GDBserver must insert reinsert breakpoint for vCont;s.  */
> +	  gdb_assert (has_reinsert_breakpoints (thread));
> +	}
>      }
>    else if (lwp->bp_reinsert != 0)
>      {
> @@ -5124,6 +5211,36 @@ 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 ())
> +    {
> +      struct inferior_list_entry *inf, *tmp;
> +
> +      ALL_INFERIORS (&all_threads, inf, tmp)
> +	{
> +	  struct thread_info *thread = (struct thread_info *) inf;
> +
> +	  if (thread->last_resume_kind == resume_step)
> +	    {
> +	      struct lwp_info *lwp = get_thread_lwp (thread);
> +	      struct thread_info *saved_thread;
> +
> +	      saved_thread = current_thread;
> +	      current_thread = thread;
> +
> +	      if (!has_reinsert_breakpoints (thread))
> +		install_software_single_step_breakpoints (lwp);
> +
> +	      current_thread = saved_thread;

Ditto.

> +
> +	      if (debug_threads)
> +		debug_printf ("Insert breakpoint for resume_step LWP %ld\n",
> +			      lwpid_of (thread));
> +	    }
> +	}
> +    }
> +
>    find_inferior (&all_threads, proceed_one_lwp, NULL);
>  }
>  
> 

Thanks,
Pedro Alves
  

Pedro Alves <palves@redhat.com> writes:

>> +  /* Remove reinsert breakpoints for resume_step.  */
>> +  if (can_software_single_step ()
>> +      && current_thread->last_resume_kind == resume_step
>
> What if a vCont;t came in after the step had started and
> before it finished?  Won't we have lost the resume_step 
> linux_set_resume_request then?

Do you mean a case like this?  GDB sends vCont;s, and then GDBserver
started step-over, in the middle of step-over, GDB sends vCont;t, so
thread->last_resume_kind is set from resume_step to resume_stop, and the
check here fails, so reinsert breakpoints are not deleted.
  

On 05/26/2016 05:52 PM, Yao Qi wrote:
> Pedro Alves <palves@redhat.com> writes:
> 
>>> +  /* Remove reinsert breakpoints for resume_step.  */
>>> +  if (can_software_single_step ()
>>> +      && current_thread->last_resume_kind == resume_step
>>
>> What if a vCont;t came in after the step had started and
>> before it finished?  Won't we have lost the resume_step 
>> linux_set_resume_request then?
> 
> Do you mean a case like this?  GDB sends vCont;s, and then GDBserver
> started step-over, in the middle of step-over, GDB sends vCont;t, so
> thread->last_resume_kind is set from resume_step to resume_stop, and the
> check here fails, so reinsert breakpoints are not deleted.

Exactly.

Thanks,
Pedro Alves