[Mesa-dev] [RFC] mesa/st: better colormask check for clear fallback

[Rob Clark]

For RGB surfaces (for example) we don't really care that the colormask
is 0x7 instead of 0xf.  This should not trigger clear_with_quad()
slowpath.

Signed-off-by: Rob Clark <robdclark at gmail.com>
---
This avoids some mem->gmem for webgl and probably other things using
a RGB surface.  This should be ok for 24b packed RGB, but not sure
about 32b RGBx.  In my case, I have a WRITEMASK for the thing I use
for clears, which I think should make this work properly for RGBx
as well.  (And, well, I probably don't even need the whole
clear_with_quad() path for colormask in the first place.)

 src/mesa/state_tracker/st_cb_clear.c | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

diff --git a/src/mesa/state_tracker/st_cb_clear.c b/src/mesa/state_tracker/st_cb_clear.c
index 3b51bd2c8a7..88fc12789e3 100644
--- a/src/mesa/state_tracker/st_cb_clear.c
+++ b/src/mesa/state_tracker/st_cb_clear.c
@@ -392,12 +392,18 @@ st_Clear(struct gl_context *ctx, GLbitfield mask)
             if (!strb || !strb->surface)
                continue;
 
-            if (!GET_COLORMASK(ctx->Color.ColorMask, colormask_index))
+            unsigned colormask =
+               GET_COLORMASK(ctx->Color.ColorMask, colormask_index);
+
+            if (!colormask)
                continue;
 
+            unsigned surf_colormask =
+               util_format_colormask(util_format_description(strb->surface->format));
+
             if (is_scissor_enabled(ctx, rb) ||
                 is_window_rectangle_enabled(ctx) ||
-                GET_COLORMASK(ctx->Color.ColorMask, colormask_index) != 0xf)
+                ((colormask & surf_colormask) != surf_colormask))
                quad_buffers |= PIPE_CLEAR_COLOR0 << i;
             else
                clear_buffers |= PIPE_CLEAR_COLOR0 << i;
-- 
2.19.1