Re: [PATCH] Let gdbserver doesn't tell GDB it support target-side breakpoint conditions and commands if it doesn't support 'Z' packet

[Hui Zhu <hui_zhu@mentor.com>]

On 12/02/13 22:37, Pedro Alves wrote:
> On 12/02/2013 12:45 PM, Hui Zhu wrote:
>> On 11/29/13 23:10, Pedro Alves wrote:
>>>> On 11/28/2013 09:07 AM, Hui Zhu wrote:
>>>>
>>>>>> +  if (have_target_target_side_commands)
>>>>
>>>> This can now just be:
>>>>
>>>>     if (!VEC_empty (agent_expr_p, bp_tgt->tcommands))
>>>>
>>>> OK with that change.
>>>>
>>>>>> +    {
>>>>>> +      warning (_("\
>>>>>> +Target doesn't support breakpoints that have target side commands."));
>>>>
>>>> I was doing to suggest making this an error instead, that
>>>> insert_bp_location would print the error string, but that's
>>>> only true for hw breakpoints... insert_bp_location's error
>>>> handling is quite messy.  For instance, if this breakpoint
>>>> is in a a shared library, this will disable the breakpoint,
>>>> even though the cause of the error is clearly not that the
>>>> shared library disappeared (i.e., not a memory error).
>>>>
>>>>>> +      return -1;
>>>>>> +    }
>> Updated the patch according to your comments.
>
> But you switched to "error" anyway?  Above I was saying that I
> was going that suggest it, but then explained why I didn't think
> it would work.  Was I wrong?

The reason is insert_bp_location doesn't have code to handle this error.  I make a new patch that include this code.

Please help me review it.

>
>>
>> And I make a patch for dprintf.exp and mi-dprintf.exp to make test OK on the target that doesn't support "Zx" packets.
>>
>> The patches were tested and pass regression test on X86_64 and PPC.
>
> (It's best to stick to one patch per email, otherwise things
> end up confusing.  I suggest looking into git send-email.)
>

OK.  I will post new version of this patch in another mail.

Thanks,
Hui

2013-12-03  Hui Zhu  <hui@codesourcery.com>

	PR gdb/16101
	* breakpoint.c (insert_bp_location): output error message of
	software breakpoints.
	* remote.c (remote_insert_breakpoint): If this breakpoint has
	target-side commands but this stub doesn't support Z0 packets,
	throw error.

--- a/gdb/breakpoint.c
+++ b/gdb/breakpoint.c
@@ -2588,6 +2588,12 @@ insert_bp_location (struct bp_location *
                    if (hw_bp_err_string)
                      fprintf_unfiltered (tmp_error_stream, "%s.\n", hw_bp_err_string);
  		}
+	      else if (hw_bp_err_string != NULL)
+	        {
+		  fprintf_unfiltered (tmp_error_stream,
+				      "Cannot insert breakpoint %d:%s\n",
+				      bl->owner->number, hw_bp_err_string);
+		}
  	      else
  		{
  		  char *message = memory_error_message (TARGET_XFER_E_IO,
--- a/gdb/remote.c
+++ b/gdb/remote.c
@@ -8244,6 +8244,12 @@ remote_insert_breakpoint (struct gdbarch
  	}
      }
  
+  /* If this breakpoint has target-side commands but this stub doesn't
+     support Z0 packets, throw error.  */
+  if (!VEC_empty (agent_expr_p, bp_tgt->tcommands))
+    error (_("\
+Target doesn't support breakpoints that have target side commands."));
+
    return memory_insert_breakpoint (gdbarch, bp_tgt);
  }