Re: [PATCH i-g-t v2 13/18] lib/vebox: Document that 2:10:10:10 is unusable with VEBOX

[Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>]

Reviewed-by: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>

On 18.9.2024 15.05, Ville Syrjala wrote:
> From: Ville Syrjälä <ville.syrjala@linux.intel.com>
> 
> VEBOX has a specific format for R10G10B10A2_UNORM, but it
> just hangs the engine if you try to use it.
> 
> For most platforms bsepc says: "Valid format only for when
> IECP is enabled and in output state only. Not supported in
> HSB mode."
> 
> Document this fact so that the next fella doesn't try to
> implement it like I did :/
> 
> Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
> ---
>   lib/veboxcopy_gen12.c | 12 ++++++++++++
>   1 file changed, 12 insertions(+)
> 
> diff --git a/lib/veboxcopy_gen12.c b/lib/veboxcopy_gen12.c
> index af52a96578d9..4e3aefaf1f44 100644
> --- a/lib/veboxcopy_gen12.c
> +++ b/lib/veboxcopy_gen12.c
> @@ -29,6 +29,7 @@
>   #define YCRCB_NORMAL	0
>   #define PLANAR_420_8	4
>   #define PACKED_444A_8	5
> +#define R10G10B10A2_UNORM	7
>   #define R8G8B8A8_UNORM	8
>   #define PACKED_444_16	9
>   #define PLANAR_420_16	12
> @@ -168,6 +169,8 @@ static uint32_t compression_format(int format, struct intel_buf *buf)
>   	switch (format) {
>   	case R16G16B16A16:
>   		return 0x1;
> +	case R10G10B10A2_UNORM:
> +		return 0xe;
>   	case R8G8B8A8_UNORM:
>   		return 0xa;
>   	case PLANAR_420_8:
> @@ -343,6 +346,15 @@ void gen12_vebox_copyfunc(struct intel_bb *ibb,
>   		break;
>   	case 32:
>   		igt_assert(!src->format_is_yuv_semiplanar);
> +		/*
> +		 * R10G10B10A2_UNORM is not usable.
> +		 *
> +		 * Bspec says: "Valid format only for when IECP is enabled
> +		 * and in output state only. Not supported in  HSB mode."
> +		 *
> +		 * VEBOX hangs if you try use it.
> +		 */
> +		igt_assert(src->depth != 30);
>   		format = src->format_is_yuv ? PACKED_444A_8 :
>   					      R8G8B8A8_UNORM;
>   		break;