[Mesa-dev] [PATCH 3/5] i965/fs: Allow saturate propagation to propagate negations into ADDs.

[Matt Turner]

Allows us to transform

   add      res  src0   src1
   mov.sat  dst  -res

into

   add.sat  dst  -src0 -src1

No shader-db changes.
---
 .../dri/i965/brw_fs_saturate_propagation.cpp       | 11 ++++++
 .../dri/i965/test_fs_saturate_propagation.cpp      | 45 ++++++++++++++++++++--
 2 files changed, 52 insertions(+), 4 deletions(-)

diff --git a/src/mesa/drivers/dri/i965/brw_fs_saturate_propagation.cpp b/src/mesa/drivers/dri/i965/brw_fs_saturate_propagation.cpp
index b9ad6fb..7c82509 100644
--- a/src/mesa/drivers/dri/i965/brw_fs_saturate_propagation.cpp
+++ b/src/mesa/drivers/dri/i965/brw_fs_saturate_propagation.cpp
@@ -86,6 +86,17 @@ opt_saturate_propagation_local(fs_visitor *v, bblock_t *block)
                      if (scan_inst->opcode == BRW_OPCODE_MUL) {
                         scan_inst->src[0].negate = !scan_inst->src[0].negate;
                         inst->src[0].negate = false;
+                     } else if (scan_inst->opcode == BRW_OPCODE_ADD) {
+                        if (scan_inst->src[1].file == IMM) {
+                           if (!brw_negate_immediate(scan_inst->src[1].type,
+                                                     &scan_inst->src[1].as_brw_reg())) {
+                              break;
+                           }
+                        } else {
+                           scan_inst->src[1].negate = !scan_inst->src[1].negate;
+                        }
+                        scan_inst->src[0].negate = !scan_inst->src[0].negate;
+                        inst->src[0].negate = false;
                      } else {
                         break;
                      }
diff --git a/src/mesa/drivers/dri/i965/test_fs_saturate_propagation.cpp b/src/mesa/drivers/dri/i965/test_fs_saturate_propagation.cpp
index 30c1472..7b3b9e5 100644
--- a/src/mesa/drivers/dri/i965/test_fs_saturate_propagation.cpp
+++ b/src/mesa/drivers/dri/i965/test_fs_saturate_propagation.cpp
@@ -215,14 +215,13 @@ TEST_F(saturate_propagation_test, neg_mov_sat)
    fs_reg dst0 = v->vgrf(glsl_type::float_type);
    fs_reg dst1 = v->vgrf(glsl_type::float_type);
    fs_reg src0 = v->vgrf(glsl_type::float_type);
-   fs_reg src1 = v->vgrf(glsl_type::float_type);
-   bld.ADD(dst0, src0, src1);
+   bld.RNDU(dst0, src0);
    dst0.negate = true;
    set_saturate(true, bld.MOV(dst1, dst0));
 
    /* = Before =
     *
-    * 0: add(8)        dst0  src0  src1
+    * 0: rndu(8)       dst0  src0
     * 1: mov.sat(8)    dst1  -dst0
     *
     * = After =
@@ -238,12 +237,50 @@ TEST_F(saturate_propagation_test, neg_mov_sat)
    EXPECT_FALSE(saturate_propagation(v));
    EXPECT_EQ(0, block0->start_ip);
    EXPECT_EQ(1, block0->end_ip);
-   EXPECT_EQ(BRW_OPCODE_ADD, instruction(block0, 0)->opcode);
+   EXPECT_EQ(BRW_OPCODE_RNDU, instruction(block0, 0)->opcode);
    EXPECT_FALSE(instruction(block0, 0)->saturate);
    EXPECT_EQ(BRW_OPCODE_MOV, instruction(block0, 1)->opcode);
    EXPECT_TRUE(instruction(block0, 1)->saturate);
 }
 
+TEST_F(saturate_propagation_test, add_neg_mov_sat)
+{
+   const fs_builder &bld = v->bld;
+   fs_reg dst0 = v->vgrf(glsl_type::float_type);
+   fs_reg dst1 = v->vgrf(glsl_type::float_type);
+   fs_reg src0 = v->vgrf(glsl_type::float_type);
+   fs_reg src1 = v->vgrf(glsl_type::float_type);
+   bld.ADD(dst0, src0, src1);
+   dst0.negate = true;
+   set_saturate(true, bld.MOV(dst1, dst0));
+
+   /* = Before =
+    *
+    * 0: add(8)        dst0  src0  src1
+    * 1: mov.sat(8)    dst1  -dst0
+    *
+    * = After =
+    * 0: add.sat(8)    dst0  -src0 -src1
+    * 1: mov(8)        dst1  dst0
+    */
+
+   v->calculate_cfg();
+   bblock_t *block0 = v->cfg->blocks[0];
+
+   EXPECT_EQ(0, block0->start_ip);
+   EXPECT_EQ(1, block0->end_ip);
+
+   EXPECT_TRUE(saturate_propagation(v));
+   EXPECT_EQ(0, block0->start_ip);
+   EXPECT_EQ(1, block0->end_ip);
+   EXPECT_EQ(BRW_OPCODE_ADD, instruction(block0, 0)->opcode);
+   EXPECT_TRUE(instruction(block0, 0)->saturate);
+   EXPECT_TRUE(instruction(block0, 0)->src[0].negate);
+   EXPECT_TRUE(instruction(block0, 0)->src[1].negate);
+   EXPECT_EQ(BRW_OPCODE_MOV, instruction(block0, 1)->opcode);
+   EXPECT_FALSE(instruction(block0, 1)->saturate);
+}
+
 TEST_F(saturate_propagation_test, mul_neg_mov_sat)
 {
    const fs_builder &bld = v->bld;
-- 
2.4.10