Re: [PATCH] Fix dprintf work not right if it is pending

[Hui Zhu <teawater at gmail dot com>]

On Wed, Apr 3, 2013 at 2:05 AM, Keith Seitz <keiths@redhat.com> wrote:
> On 03/29/2013 12:42 AM, Hui Zhu wrote:
>
>>>> +  breakpoint_re_set_default (b);
>>>> +
>>>> +  if (b->extra_string != NULL)
>>>> +    update_dprintf_command_list (b);
>>>> +}
>>>> +
>>>
>>>
>>>
>>> This will update the command list every time breakpoints are reset and
>>> could
>>> be limited to only those needing updating. Is there perhaps a reason to
>>> always do this?
>>
>>
>> I think it need, because it need to generate different commands with
>> different status for example:
>>        if (target_can_run_breakpoint_commands ())
>>         printf_line = xstrprintf ("agent-printf %s", dprintf_args);
>>
>
> I'm not understanding this example. How is this likely to change whenever
> breakpoints are reset? Is there perhaps a way to add a test to demonstrate
> this requirement?

If the target that GDB connect to is changed  (for example, we set
dprintf and used it with local host target.  Then we changed to use a
remote target), GDB need set different commands of dprintfs.
The code that I post is from function update_dprintf_command_list.

I updated dprintf-pending.exp for this test.

>
> Otherwise, this patch looks good to me, and recommend that a maintainer give
> this a final review.
>

Thanks for your help.

> Keith

Best,
Hui

Attachment: dprintf-pending.txt
Description: Text document

Attachment: dprintf-pending-test.txt
Description: Text document