intel/fs: Add virtual instruction to load mask of live channels into flag register.

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Cc: 20.0 <mesa-stable@lists.freedesktop.org>

diff --git a/src/intel/compiler/brw_eu_defines.h b/src/intel/compiler/brw_eu_defines.h
index c83a478b9089e256ee279bbea768294482ef942e..779bab04235c0df66aa924bede36c9ed26d3d6d3 100644 (file)
--- a/src/intel/compiler/brw_eu_defines.h
+++ b/src/intel/compiler/brw_eu_defines.h
@@ -490,6 +490,12 @@ enum opcode {
     */
    SHADER_OPCODE_FIND_LIVE_CHANNEL,
 
+   /**
+    * Return the current execution mask in the specified flag subregister.
+    * Can be CSE'ed more easily than a plain MOV from the ce0 ARF register.
+    */
+   FS_OPCODE_LOAD_LIVE_CHANNELS,
+
    /**
     * Pick the channel from its first source register given by the index
     * specified as second source.  Useful for variable indexing of surfaces.

diff --git a/src/intel/compiler/brw_fs.cpp b/src/intel/compiler/brw_fs.cpp
index d54fa7a640b3cf1d622070acba62e688026a8ceb..1b582cda0cd6f7e9342b924540ae556acf6d05f3 100644 (file)
--- a/src/intel/compiler/brw_fs.cpp
+++ b/src/intel/compiler/brw_fs.cpp
@@ -1088,7 +1088,8 @@ fs_inst::flags_written() const
                             opcode != BRW_OPCODE_WHILE)) ||
        opcode == FS_OPCODE_FB_WRITE) {
       return flag_mask(this, 1);
-   } else if (opcode == SHADER_OPCODE_FIND_LIVE_CHANNEL) {
+   } else if (opcode == SHADER_OPCODE_FIND_LIVE_CHANNEL ||
+              opcode == FS_OPCODE_LOAD_LIVE_CHANNELS) {
       return flag_mask(this, 32);
    } else {
       return flag_mask(dst, size_written);

diff --git a/src/intel/compiler/brw_fs_cse.cpp b/src/intel/compiler/brw_fs_cse.cpp
index 4751e11bcfea8a2edf05020d06416ca5e5538f88..f02914e04f352f0d81f4d75581ec7d34d62c3c40 100644 (file)
--- a/src/intel/compiler/brw_fs_cse.cpp
+++ b/src/intel/compiler/brw_fs_cse.cpp
@@ -76,6 +76,7 @@ is_expression(const fs_visitor *v, const fs_inst *const inst)
    case FS_OPCODE_VARYING_PULL_CONSTANT_LOAD_LOGICAL:
    case FS_OPCODE_LINTERP:
    case SHADER_OPCODE_FIND_LIVE_CHANNEL:
+   case FS_OPCODE_LOAD_LIVE_CHANNELS:
    case SHADER_OPCODE_BROADCAST:
    case SHADER_OPCODE_MOV_INDIRECT:
    case SHADER_OPCODE_TEX_LOGICAL:

diff --git a/src/intel/compiler/brw_fs_generator.cpp b/src/intel/compiler/brw_fs_generator.cpp
index 4f7bc159f073dc68a62d3c27946caf0130b758f0..00b0f29ef7b9193162be6a4250ed9e66ca50a7c5 100644 (file)
--- a/src/intel/compiler/brw_fs_generator.cpp
+++ b/src/intel/compiler/brw_fs_generator.cpp
@@ -2224,7 +2224,16 @@ fs_generator::generate_code(const cfg_t *cfg, int dispatch_width,
          brw_find_live_channel(p, dst, mask);
          break;
       }
-
+      case FS_OPCODE_LOAD_LIVE_CHANNELS: {
+         assert(devinfo->gen >= 8);
+         assert(inst->force_writemask_all && inst->group == 0);
+         assert(inst->dst.file == BAD_FILE);
+         brw_set_default_exec_size(p, BRW_EXECUTE_1);
+         brw_MOV(p, retype(brw_flag_subreg(inst->flag_subreg),
+                           BRW_REGISTER_TYPE_UD),
+                 retype(brw_mask_reg(0), BRW_REGISTER_TYPE_UD));
+         break;
+      }
       case SHADER_OPCODE_BROADCAST:
          assert(inst->force_writemask_all);
          brw_broadcast(p, dst, src[0], src[1]);

diff --git a/src/intel/compiler/brw_shader.cpp b/src/intel/compiler/brw_shader.cpp
index 1906c1fff1b85734afdbcee6b2396615bcafdbac..0a3c3739aa2b7ae1d12c6626087810a5742e550b 100644 (file)
--- a/src/intel/compiler/brw_shader.cpp
+++ b/src/intel/compiler/brw_shader.cpp
@@ -364,6 +364,9 @@ brw_instruction_name(const struct gen_device_info *devinfo, enum opcode op)
 
    case SHADER_OPCODE_FIND_LIVE_CHANNEL:
       return "find_live_channel";
+   case FS_OPCODE_LOAD_LIVE_CHANNELS:
+      return "load_live_channels";
+
    case SHADER_OPCODE_BROADCAST:
       return "broadcast";
    case SHADER_OPCODE_SHUFFLE: