[Mesa-dev] [PATCH] r600g: Fix ARB_texture_rgb10_a2ui support in big-endian

[Oded Gabbay]

On Mon, Feb 29, 2016 at 3:51 AM, Michel Dänzer <michel at daenzer.net> wrote:
> On 28.02.2016 05:48, Oded Gabbay wrote:
>> This patch enables the correct detection of PIPE_FORMAT_R10G10B10A2_UINT
>> and PIPE_FORMAT_B10G10R10A2_UINT formats in r600g in big-endian mode, by
>> adjusting the order of channels in various functions.
>>
>> This enables support for ARB_texture_rgb10_a2ui, which otherwise is not
>> detected as supported.
>>
>> Signed-off-by: Oded Gabbay <oded.gabbay at gmail.com>
>> ---
>>  src/gallium/drivers/r600/r600_state_common.c | 15 ++++++++++-----
>>  1 file changed, 10 insertions(+), 5 deletions(-)
>>
>> diff --git a/src/gallium/drivers/r600/r600_state_common.c b/src/gallium/drivers/r600/r600_state_common.c
>> index b231d1e..b02b6a9 100644
>> --- a/src/gallium/drivers/r600/r600_state_common.c
>> +++ b/src/gallium/drivers/r600/r600_state_common.c
>> @@ -2468,10 +2468,13 @@ uint32_t r600_translate_texformat(struct pipe_screen *screen,
>>                               result = FMT_1_5_5_5;
>>                               goto out_word4;
>>                       }
>> -                     if (desc->channel[0].size == 10 &&
>> -                         desc->channel[1].size == 10 &&
>> -                         desc->channel[2].size == 10 &&
>> -                         desc->channel[3].size == 2) {
>> +                     if ((desc->channel[1].size == 10 &&
>> +                               desc->channel[2].size == 10) &&
>> +                                     ((desc->channel[0].size == 10 &&
>> +                                       desc->channel[3].size == 2) ||
>> +                                      (R600_BIG_ENDIAN &&
>> +                                       desc->channel[0].size == 2 &&
>> +                                       desc->channel[3].size == 10))) {
>>                               result = FMT_2_10_10_10;
>>                               goto out_word4;
>>                       }
>> @@ -2694,9 +2697,11 @@ uint32_t r600_translate_colorformat(enum chip_class chip, enum pipe_format forma
>>                       }
>>               } else if (HAS_SIZE(5,5,5,1)) {
>>                       return V_0280A0_COLOR_1_5_5_5;
>> -             } else if (HAS_SIZE(10,10,10,2)) {
>> +             } else if (HAS_SIZE(10,10,10,2) ||
>> +                             (R600_BIG_ENDIAN && HAS_SIZE(2,10,10,10))) {
>
> Since the components of these formats cross byte boundaries, these
> formats should really be treated as packed by the core Gallium code, in
> which case this change probably wouldn't be necessary. In other words,
> this is probably working around a core Gallium bug.
>

so ?

>
>>               }
>> +
>>               break;
>>       }
>
> Don't add this blank line.
>
>
> --
> Earthling Michel Dänzer               |               http://www.amd.com
> Libre software enthusiast             |             Mesa and X developer