[Piglit] [PATCH 06/18] shader_runner: Refactor handling of fb commands.

[Marek Olšák]

On Wed, Nov 9, 2016 at 11:41 PM, Francisco Jerez <currojerez at riseup.net> wrote:
>>>>> @@ -3622,7 +3636,6 @@ piglit_init(int argc, char **argv)
>>>>>                         sso_in_use = false;
>>>>>                         prog_err_info = NULL;
>>>>>                         vao = 0;
>>>>> -                       fbo = 0;
>>>>
>>>> This breaks some piglit tests with --process-isolation 0. Adding
>>>> "draw_fbo = 0" here fixes that. Is that the right fix?
>>>>
>>>
>>> Interesting, I guess the test breaks while trying to deallocate the FBO
>>> From the previous test?  Or does it break in some other way?  In the
>>> former case I guess that setting 'draw_fbo = 0' would work around the
>>> issue, but it would probably also lead to FBO object leaks.
>>
>> No idea what's causing the failures, but releasing the FBO should be
>> really easy at the end of that block.
>>
>
> Does the attached patch fix the problem for you?  I didn't get a full
> piglit run to work with --process-isolation 0, so I couldn't confirm
> whether it fixes the issue.

Yes, it fixes the issue. Thanks.

Reviewed-by: Marek Olšák <marek.olsak at amd.com>

Marek