i965: Add support for the DP2 opcode, which we use for dot(vec2, vec2).

The original glsl compiler would generate a.x * b.x + a.y * b.y, which
we would do mul+mul+add for instead of this mul+mac.

Fixes glsl-fs-dot-vec2.

4 files changed, 38 insertions, 0 deletions

diff --git a/src/mesa/drivers/dri/i965/brw_wm.h b/src/mesa/drivers/dri/i965/brw_wm.h
index 938557ff369..197b8754345 100644
--- a/src/mesa/drivers/dri/i965/brw_wm.h
+++ b/src/mesa/drivers/dri/i965/brw_wm.h
@@ -343,6 +343,11 @@ void emit_delta_xy(struct brw_compile *p,
 		   const struct brw_reg *dst,
 		   GLuint mask,
 		   const struct brw_reg *arg0);
+void emit_dp2(struct brw_compile *p,
+	      const struct brw_reg *dst,
+	      GLuint mask,
+	      const struct brw_reg *arg0,
+	      const struct brw_reg *arg1);
 void emit_dp3(struct brw_compile *p,
 	      const struct brw_reg *dst,
 	      GLuint mask,
diff --git a/src/mesa/drivers/dri/i965/brw_wm_emit.c b/src/mesa/drivers/dri/i965/brw_wm_emit.c
index 11f482bddf2..a90a2d3cf25 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_emit.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_emit.c
@@ -731,6 +731,27 @@ void emit_min(struct brw_compile *p,
 }
 
 
+void emit_dp2(struct brw_compile *p,
+	      const struct brw_reg *dst,
+	      GLuint mask,
+	      const struct brw_reg *arg0,
+	      const struct brw_reg *arg1)
+{
+   int dst_chan = _mesa_ffs(mask & WRITEMASK_XYZW) - 1;
+
+   if (!(mask & WRITEMASK_XYZW))
+      return; /* Do not emit dead code */
+
+   assert(is_power_of_two(mask & WRITEMASK_XYZW));
+
+   brw_MUL(p, brw_null_reg(), arg0[0], arg1[0]);
+
+   brw_set_saturate(p, (mask & SATURATE) ? 1 : 0);
+   brw_MAC(p, dst[dst_chan], arg0[1], arg1[1]);
+   brw_set_saturate(p, 0);
+}
+
+
 void emit_dp3(struct brw_compile *p,
 	      const struct brw_reg *dst,
 	      GLuint mask,
@@ -1584,6 +1605,10 @@ void brw_wm_emit( struct brw_wm_compile *c )
 	 emit_ddxy(p, dst, dst_flags, GL_FALSE, args[0]);
 	 break;
 
+      case OPCODE_DP2:
+	 emit_dp2(p, dst, dst_flags, args[0], args[1]);
+	 break;
+
       case OPCODE_DP3:
 	 emit_dp3(p, dst, dst_flags, args[0], args[1]);
 	 break;
diff --git a/src/mesa/drivers/dri/i965/brw_wm_glsl.c b/src/mesa/drivers/dri/i965/brw_wm_glsl.c
index e23ce7ad7a5..57be08a8d1d 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_glsl.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_glsl.c
@@ -1903,6 +1903,9 @@ static void brw_wm_emit_glsl(struct brw_context *brw, struct brw_wm_compile *c)
 	    case OPCODE_SWZ:
 		emit_alu1(p, brw_MOV, dst, dst_flags, args[0]);
 		break;
+	    case OPCODE_DP2:
+		emit_dp2(p, dst, dst_flags, args[0], args[1]);
+		break;
 	    case OPCODE_DP3:
 		emit_dp3(p, dst, dst_flags, args[0], args[1]);
 		break;
diff --git a/src/mesa/drivers/dri/i965/brw_wm_pass1.c b/src/mesa/drivers/dri/i965/brw_wm_pass1.c
index 8ee1f153b1f..962515a99e9 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_pass1.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_pass1.c
@@ -255,6 +255,11 @@ void brw_wm_pass1( struct brw_wm_compile *c )
 	 read2 = WRITEMASK_W; /* pixel w */
 	 break;
 
+      case OPCODE_DP2:
+	 read0 = WRITEMASK_XY;
+	 read1 = WRITEMASK_XY;
+	 break;
+
       case OPCODE_DP3:	
 	 read0 = WRITEMASK_XYZ;
 	 read1 = WRITEMASK_XYZ;