[Mesa-dev] [PATCH] nv50,nvc0: enable using LOAD from constbuf

[Ilia Mirkin]

This enables std430-style packing for UBOs which aren't otherwise marked
as std140.

There might be small register lifetime changes as a result of removed
duplicate loads in some cases, but this seems worth it overall.

Signed-off-by: Ilia Mirkin <imirkin at alum.mit.edu>
---
 src/gallium/drivers/nouveau/codegen/nv50_ir_from_tgsi.cpp | 2 ++
 src/gallium/drivers/nouveau/nv50/nv50_screen.c            | 2 +-
 src/gallium/drivers/nouveau/nvc0/nvc0_screen.c            | 2 +-
 3 files changed, 4 insertions(+), 2 deletions(-)

diff --git a/src/gallium/drivers/nouveau/codegen/nv50_ir_from_tgsi.cpp b/src/gallium/drivers/nouveau/codegen/nv50_ir_from_tgsi.cpp
index 34351dab518..dfddffc89da 100644
--- a/src/gallium/drivers/nouveau/codegen/nv50_ir_from_tgsi.cpp
+++ b/src/gallium/drivers/nouveau/codegen/nv50_ir_from_tgsi.cpp
@@ -403,6 +403,7 @@ static nv50_ir::DataFile translateFile(uint file)
 {
    switch (file) {
    case TGSI_FILE_CONSTANT:        return nv50_ir::FILE_MEMORY_CONST;
+   case TGSI_FILE_CONSTBUF:        return nv50_ir::FILE_MEMORY_CONST;
    case TGSI_FILE_INPUT:           return nv50_ir::FILE_SHADER_INPUT;
    case TGSI_FILE_OUTPUT:          return nv50_ir::FILE_SHADER_OUTPUT;
    case TGSI_FILE_TEMPORARY:       return nv50_ir::FILE_GPR;
@@ -2628,6 +2629,7 @@ Converter::handleLOAD(Value *dst0[4])
 
    switch (tgsi.getSrc(0).getFile()) {
    case TGSI_FILE_BUFFER:
+   case TGSI_FILE_CONSTBUF:
    case TGSI_FILE_MEMORY:
       for (c = 0; c < 4; ++c) {
          if (!dst0[c])
diff --git a/src/gallium/drivers/nouveau/nv50/nv50_screen.c b/src/gallium/drivers/nouveau/nv50/nv50_screen.c
index 2066cf3f6e1..5ee5a26b652 100644
--- a/src/gallium/drivers/nouveau/nv50/nv50_screen.c
+++ b/src/gallium/drivers/nouveau/nv50/nv50_screen.c
@@ -202,6 +202,7 @@ nv50_screen_get_param(struct pipe_screen *pscreen, enum pipe_cap param)
    case PIPE_CAP_TGSI_CLOCK:
    case PIPE_CAP_CAN_BIND_CONST_BUFFER_AS_VERTEX:
    case PIPE_CAP_ALLOW_MAPPED_BUFFERS_DURING_EXECUTION:
+   case PIPE_CAP_LOAD_CONSTBUF:
       return 1;
    case PIPE_CAP_SEAMLESS_CUBE_MAP:
       return 1; /* class_3d >= NVA0_3D_CLASS; */
@@ -275,7 +276,6 @@ nv50_screen_get_param(struct pipe_screen *pscreen, enum pipe_cap param)
    case PIPE_CAP_NIR_SAMPLERS_AS_DEREF:
    case PIPE_CAP_QUERY_SO_OVERFLOW:
    case PIPE_CAP_MEMOBJ:
-   case PIPE_CAP_LOAD_CONSTBUF:
    case PIPE_CAP_TGSI_ANY_REG_AS_ADDRESS:
    case PIPE_CAP_TILE_RASTER_ORDER:
    case PIPE_CAP_MAX_COMBINED_SHADER_OUTPUT_RESOURCES:
diff --git a/src/gallium/drivers/nouveau/nvc0/nvc0_screen.c b/src/gallium/drivers/nouveau/nvc0/nvc0_screen.c
index d62a5552536..3544afe044b 100644
--- a/src/gallium/drivers/nouveau/nvc0/nvc0_screen.c
+++ b/src/gallium/drivers/nouveau/nvc0/nvc0_screen.c
@@ -257,6 +257,7 @@ nvc0_screen_get_param(struct pipe_screen *pscreen, enum pipe_cap param)
    case PIPE_CAP_COMPUTE:
    case PIPE_CAP_CAN_BIND_CONST_BUFFER_AS_VERTEX:
    case PIPE_CAP_ALLOW_MAPPED_BUFFERS_DURING_EXECUTION:
+   case PIPE_CAP_LOAD_CONSTBUF:
       return 1;
    case PIPE_CAP_SEAMLESS_CUBE_MAP_PER_TEXTURE:
       return (class_3d >= NVE4_3D_CLASS) ? 1 : 0;
@@ -304,7 +305,6 @@ nvc0_screen_get_param(struct pipe_screen *pscreen, enum pipe_cap param)
    case PIPE_CAP_NIR_SAMPLERS_AS_DEREF:
    case PIPE_CAP_QUERY_SO_OVERFLOW:
    case PIPE_CAP_MEMOBJ:
-   case PIPE_CAP_LOAD_CONSTBUF:
    case PIPE_CAP_TGSI_ANY_REG_AS_ADDRESS:
    case PIPE_CAP_TILE_RASTER_ORDER:
    case PIPE_CAP_MAX_COMBINED_SHADER_OUTPUT_RESOURCES:
-- 
2.13.6