vc4: Handle partial loads/stores of tiled textures.

Previously, we would load out the tile-aligned area, update the raster
copy, and store it back.  This was a huge cost for XPutImage calls to the
screen under glamor.

Instead, implement a general load/store path that walks over the source
x/y writing into the corresponding pixel of the destination (using clever
math from
https://fgiesen.wordpress.com/2011/01/17/texture-tiling-and-swizzling/).
If things are aligned, we go through the previous utile-at-a-time loop.

Improves x11perf -putimage10 performance by 139.777% +/- 2.83464% (n=5)
Improves x11perf -putimage100 performance by 383.908% +/- 22.6297% (n=11)
Improves x11perf -getimage10 performance by 2.75731% +/- 0.585054% (n=145)

1 files changed, 0 insertions, 13 deletions

diff --git a/src/gallium/drivers/vc4/vc4_tiling.c b/src/gallium/drivers/vc4/vc4_tiling.c
index 07e1c9c5f67..2da520eb4db 100644
--- a/src/gallium/drivers/vc4/vc4_tiling.c
+++ b/src/gallium/drivers/vc4/vc4_tiling.c
@@ -63,15 +63,6 @@ vc4_size_is_lt(uint32_t width, uint32_t height, int cpp)
                 height <= 4 * vc4_utile_height(cpp));
 }
 
-static void
-check_box_utile_alignment(const struct pipe_box *box, int cpp)
-{
-        assert(!(box->x & (vc4_utile_width(cpp) - 1)));
-        assert(!(box->y & (vc4_utile_height(cpp) - 1)));
-        assert(!(box->width & (vc4_utile_width(cpp) - 1)));
-        assert(!(box->height & (vc4_utile_height(cpp) - 1)));
-}
-
 /**
  * Takes a utile x and y (and the number of utiles of width of the image) and
  * returns the offset to the utile within a VC4_TILING_FORMAT_TF image.
@@ -216,8 +207,6 @@ vc4_load_tiled_image(void *dst, uint32_t dst_stride,
                      uint8_t tiling_format, int cpp,
                      const struct pipe_box *box)
 {
-        check_box_utile_alignment(box, cpp);
-
         if (tiling_format == VC4_TILING_FORMAT_LT) {
                 vc4_load_lt_image(dst, dst_stride,
                                   src, src_stride,
@@ -240,8 +229,6 @@ vc4_store_tiled_image(void *dst, uint32_t dst_stride,
                       uint8_t tiling_format, int cpp,
                       const struct pipe_box *box)
 {
-        check_box_utile_alignment(box, cpp);
-
         if (tiling_format == VC4_TILING_FORMAT_LT) {
                 vc4_store_lt_image(dst, dst_stride,
                                    src, src_stride,