On Tue, Dec 04, 2018 at 03:22:03PM +0100, Paul Kocialkowski wrote:
> In prevision of adding support for YUV formats, set the YUV to RGB
> colorspace conversion coefficients if required and don't bypass the
> CSC engine when converting.
>
> The BT601 coefficients from the A33 BSP are copied over from the backend
> code. Because of module inter-dependency, we can't have the frontend use
> these coefficients from the backend directly.
>
> Signed-off-by: Paul Kocialkowski <paul.kocialkowski@xxxxxxxxxxx>
> ---
> drivers/gpu/drm/sun4i/sun4i_frontend.c | 45 ++++++++++++++++++++++++--
> drivers/gpu/drm/sun4i/sun4i_frontend.h | 2 ++
> 2 files changed, 45 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/gpu/drm/sun4i/sun4i_frontend.c b/drivers/gpu/drm/sun4i/sun4i_frontend.c
> index d0eca4b8784d..1737a4394401 100644
> --- a/drivers/gpu/drm/sun4i/sun4i_frontend.c
> +++ b/drivers/gpu/drm/sun4i/sun4i_frontend.c
> @@ -48,6 +48,28 @@ static const u32 sun4i_frontend_horz_coef[64] = {
> 0x03ff0000, 0x0000fd41, 0x01ff0000, 0x0000fe42,
> };
>
> +/*
> + * These coefficients are taken from the A33 BSP from Allwinner.
> + *
> + * The first three values of each row are coded as 13-bit signed fixed-point
> + * numbers, with 10 bits for the fractional part. The fourth value is a
> + * constant coded as a 14-bit signed fixed-point number with 4 bits for the
> + * fractional part.
> + *
> + * The values in table order give the following colorspace translation:
> + * G = 1.164 * Y - 0.391 * U - 0.813 * V + 135
> + * R = 1.164 * Y + 1.596 * V - 222
> + * B = 1.164 * Y + 2.018 * U + 276
> + *
> + * This seems to be a conversion from Y[16:235] UV[16:240] to RGB[0:255],
> + * following the BT601 spec.
> + */
> +static const u32 sunxi_bt601_yuv2rgb_coef[12] = {
> + 0x000004a7, 0x00001e6f, 0x00001cbf, 0x00000877,
> + 0x000004a7, 0x00000000, 0x00000662, 0x00003211,
> + 0x000004a7, 0x00000812, 0x00000000, 0x00002eb1,
> +};
> +
I'd really prefer not to duplicate that structure. IIRC, we already
have a dependency between the frontend and the backend (or the other
way around), maybe we should just move that structure to either so
that we don't have a circular dependency?
Maxime
--
Maxime Ripard, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com
Attachment:
signature.asc
Description: PGP signature
_______________________________________________ dri-devel mailing list dri-devel@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/dri-devel
