[Pixman] [PATCH] sse2: Implement simple bilinear scaling for x8r8g8b8 to a8r8g8b8

[Søren Sandmann]

Chris Wilson <chris at chris-wilson.co.uk> writes:

>  static force_inline void
> +scaled_bilinear_scanline_sse2_0888_8888_SRC (uint32_t *       dst,
> +					     const uint32_t * mask,
> +					     const uint32_t * src_top,
> +					     const uint32_t * src_bottom,
> +					     int32_t          w,
> +					     int              wt,
> +					     int              wb,
> +					     pixman_fixed_t   vx,
> +					     pixman_fixed_t   unit_x,
> +					     pixman_fixed_t   max_vx,
> +					     pixman_bool_t    zero_src)

We usually use the name x888 for x8r8g8b8. 0888 is 24bpp RGB.

Other than that, looks good to me, although I'll say once again that to
really make bilinear scaling fast, we need to take advantage of
separability, as described here:

   http://lists.freedesktop.org/archives/pixman/2012-June/002140.html


Søren