[Mesa-dev] [PATCH 07/23] mesa: glGetProgramResourceLocationIndex

[Tapani Pälli]

On 03/17/2015 04:20 PM, Ilia Mirkin wrote:
> On Tue, Mar 17, 2015 at 5:13 AM, Tapani Pälli <tapani.palli at intel.com> wrote:
>>
>>
>> On 03/16/2015 08:08 PM, Ilia Mirkin wrote:
>>>
>>> On Fri, Mar 13, 2015 at 4:37 AM, Tapani Pälli <tapani.palli at intel.com>
>>> wrote:
>>>> +/**
>>>> + * Returns output index for dual source blending.
>>>> + */
>>>>    GLint GLAPIENTRY
>>>>    _mesa_GetProgramResourceLocationIndex(GLuint program, GLenum
>>>> programInterface,
>>>>                                          const GLchar *name)
>>>>    {
>>>> -   return -1;
>>>> +   GET_CURRENT_CONTEXT(ctx);
>>>> +   struct gl_shader_program *shProg =
>>>> +      lookup_linked_program(program,
>>>> "glGetProgramResourceLocationIndex");
>>>> +
>>>> +   if (!shProg || invalid_array_element_syntax(name))
>>>> +      return -1;
>>>> +
>>>> +   /* From the GL_ARB_program_interface_query spec:
>>>> +    *
>>>> +    * "For GetProgramResourceLocationIndex, <programInterface> must be
>>>> +    * PROGRAM_OUTPUT."
>>>> +    */
>>>
>>>
>>> And presumably it must be a program with a fragment shader (which
>>> might not be there for a no-rast or compute pipeline).
>>
>>
>> spec says that -1 is returned:
>>
>> "If <program> has been successfully linked but contains no fragment shader,
>> no error will be generated but -1 will be returned."
>>
>> this is what happens with the implementation.
>
> Can you explain how the current implementation takes care of that?
>

My earlier suggestion for this was that 
_mesa_program_resource_location_index returns -1 because resource cannot 
be found. But now I see that one could call this for any output of any 
stage and then it would likely return 0 or whatever is the uninitialized 
state of ir_variable::data.index. Will have to fix this, thanks for 
spotting this!

// Tapani