On 06/24/2016 08:45 AM, Roland Scheidegger wrote:
For what formats is this really needed? I think that usually if you have
a rgb surface, the corresponding rgbx format should be used instead of
rgba (which implicitly has the alpha == 1 property for blending). But
maybe some formats are missing...

This all came about while finishing up support for GL_ARB_copy_image. Briefly, supporting RGBX caused quite a bit of grief that could be avoided by disabling the RGBX format and using RGBA instead. But this leads to the blending issue (which, again, is a corner case not likely to be hit in practice).

Some 'fallback' paths for GL_ARB_copy_image use pipe_context::blit() and involves creating RGBA8_UNORM sampler views and surface views of the underlying resource. Since our svga3d device doesn't have RGBX8_UINT, RGBX8_SNORM, etc. we use RGBA8_UINT, RGBA8_SNORM, etc. but we do support RGBX8_UNORM. So copying from GL_RGB8UI to GL_RGB8, for example, involves creating a RGBA8_UNORM view of a RGBX8_TYPLESS surface, which IIRC, is disallowed by our svga3d device.

Before I quit on Friday, however, I think I found another piece of this puzzle that might make RGBX work. I'll try to get back on that today.

-Brian


Am 24.06.2016 um 04:07 schrieb Brian Paul:
If the user requests an RGB drawing surface but we actually create an
RGBA surface, we need it to act as if A=1.  For blending, this means
adjusting the blending terms to use 1/0 instead of DST_ALPHA/INV_DST_ALPHA.
Drivers can use this flag to determine when that's needed.

A previous patch I posted last year did this entirely in the state tracker
but it involved making blend state dependent on the framebuffer state.
This approach avoids that dependency.
---
  src/gallium/include/pipe/p_state.h | 1 +
  1 file changed, 1 insertion(+)

diff --git a/src/gallium/include/pipe/p_state.h 
b/src/gallium/include/pipe/p_state.h
index 9c69355..8b0c3a2 100644
--- a/src/gallium/include/pipe/p_state.h
+++ b/src/gallium/include/pipe/p_state.h
@@ -387,6 +387,7 @@ struct pipe_surface
     unsigned height;              /**< logical height in pixels */

     unsigned writable:1;          /**< writable shader resource */
+   unsigned alpha_one:1;    /**< Should an RGBA surface should act like RGB? */

     union {
        struct {



_______________________________________________
mesa-dev mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Reply via email to