[Mesa-dev] [PATCH v2 6/9] radeonsi: Enable ARB_framebuffer_no_attachments

[Marek Olšák]

On Fri, Feb 5, 2016 at 5:46 PM, Ilia Mirkin <imirkin at alum.mit.edu> wrote:
> On Fri, Feb 5, 2016 at 11:43 AM, Marek Olšák <maraeo at gmail.com> wrote:
>> On Fri, Feb 5, 2016 at 12:44 PM, Edward O'Callaghan
>> <eocallaghan at alterapraxis.com> wrote:
>>> Signed-off-by: Edward O'Callaghan <eocallaghan at alterapraxis.com>
>>> ---
>>>  src/gallium/drivers/radeonsi/si_pipe.c | 4 +++-
>>>  1 file changed, 3 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/src/gallium/drivers/radeonsi/si_pipe.c b/src/gallium/drivers/radeonsi/si_pipe.c
>>> index d3f3677..a58a9d5 100644
>>> --- a/src/gallium/drivers/radeonsi/si_pipe.c
>>> +++ b/src/gallium/drivers/radeonsi/si_pipe.c
>>> @@ -361,7 +361,6 @@ static int si_get_param(struct pipe_screen* pscreen, enum pipe_cap param)
>>>         case PIPE_CAP_GENERATE_MIPMAP:
>>>         case PIPE_CAP_STRING_MARKER:
>>>         case PIPE_CAP_QUERY_BUFFER_OBJECT:
>>> -       case PIPE_CAP_MAX_FRAMEBUFFER_LAYERS:
>>>                 return 0;
>>>
>>>         case PIPE_CAP_MAX_SHADER_PATCH_VARYINGS:
>>> @@ -401,6 +400,9 @@ static int si_get_param(struct pipe_screen* pscreen, enum pipe_cap param)
>>>                 /* textures support 8192, but layered rendering supports 2048 */
>>>                 return 2048;
>>>
>>> +       case PIPE_CAP_MAX_FRAMEBUFFER_LAYERS:
>>> +               return 2048;
>>> +
>>
>> Others have probably said this already, but I think this new CAP is
>> unnecessary and PIPE_CAP_MAX_TEXTURE_ARRAY_LAYERS can be used instead.
>
> Not sure that we want to flip this on for all drivers that have array
> textures... I think a separate enable is needed, but of the boolean
> variety.

Oh yes, that would be better.

Marek