intel/fs: Make alpha test work with MRT and sample mask

Fix the order of src0_alpha and sample mask in fb payload.
From SKL PRM Volume 7, "Data Payload Register Order
for Render Target Write Messages":
 Type   S0A  oM  sZ  oS  M2     M3       M4
 SIMD8   1   1   0   0   s0A    oM       R
 SIMD16  1   1   0   0   1/0s0A 3/2s0A   oM

It also fixes working of alpha to coverage with sample mask
on GEN6 since now they are in correct order.

Signed-off-by: Danylo Piliaiev <danylo.piliaiev@globallogic.com>
Signed-off-by: Francisco Jerez <currojerez@riseup.net>
Reviewed-by:  Francisco Jerez <currojerez@riseup.net>

diff --git a/src/intel/compiler/brw_fs.cpp b/src/intel/compiler/brw_fs.cpp
index 635f72721d9af558336e158f74c10b370dcb5bb7..0c2439d9daf528e9c8cc3aa76c345c0be77549e8 100644 (file)
--- a/src/intel/compiler/brw_fs.cpp
+++ b/src/intel/compiler/brw_fs.cpp
@@ -4281,6 +4281,23 @@ lower_fb_write_logical_send(const fs_builder &bld, fs_inst *inst,
       length++;
    }
 
+   if (src0_alpha.file != BAD_FILE) {
+      for (unsigned i = 0; i < bld.dispatch_width() / 8; i++) {
+         const fs_builder &ubld = bld.exec_all().group(8, i)
+                                    .annotate("FB write src0 alpha");
+         const fs_reg tmp = ubld.vgrf(BRW_REGISTER_TYPE_F);
+         ubld.MOV(tmp, horiz_offset(src0_alpha, i * 8));
+         setup_color_payload(ubld, key, &sources[length], tmp, 1);
+         length++;
+      }
+   } else if (prog_data->replicate_alpha && inst->target != 0) {
+      /* Handle the case when fragment shader doesn't write to draw buffer
+       * zero. No need to call setup_color_payload() for src0_alpha because
+       * alpha value will be undefined.
+       */
+      length += bld.dispatch_width() / 8;
+   }
+
    if (sample_mask.file != BAD_FILE) {
       sources[length] = fs_reg(VGRF, bld.shader->alloc.allocate(1),
                                BRW_REGISTER_TYPE_UD);
@@ -4304,24 +4321,6 @@ lower_fb_write_logical_send(const fs_builder &bld, fs_inst *inst,
 
    payload_header_size = length;
 
-   if (src0_alpha.file != BAD_FILE) {
-      /* FIXME: This is being passed at the wrong location in the payload and
-       * doesn't work when gl_SampleMask and MRTs are used simultaneously.
-       * It's supposed to be immediately before oMask but there seems to be no
-       * reasonable way to pass them in the correct order because LOAD_PAYLOAD
-       * requires header sources to form a contiguous segment at the beginning
-       * of the message and src0_alpha has per-channel semantics.
-       */
-      setup_color_payload(bld, key, &sources[length], src0_alpha, 1);
-      length++;
-   } else if (prog_data->replicate_alpha && inst->target != 0) {
-      /* Handle the case when fragment shader doesn't write to draw buffer
-       * zero. No need to call setup_color_payload() for src0_alpha because
-       * alpha value will be undefined.
-       */
-      length++;
-   }
-
    setup_color_payload(bld, key, &sources[length], color0, components);
    length += 4;