nir: Add pass to combine adjacent scoped memory barriers

SPIR-V generates very granular barriers, however HW and backends might
not necessarily take advantage of those.  This pass provides a general
mechanism to combine such barriers.

Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3224>

4 files changed, 95 insertions, 0 deletions

diff --git a/src/compiler/Makefile.sources b/src/compiler/Makefile.sources
index b44b2b5f6ff..28562a79701 100644
--- a/src/compiler/Makefile.sources
+++ b/src/compiler/Makefile.sources
@@ -296,6 +296,7 @@ NIR_FILES = \
 	nir/nir_move_vec_src_uses_to_dest.c \
 	nir/nir_normalize_cubemap_coords.c \
 	nir/nir_opt_access.c \
+	nir/nir_opt_barriers.c \
 	nir/nir_opt_combine_stores.c \
 	nir/nir_opt_comparison_pre.c \
 	nir/nir_opt_conditional_discard.c \
diff --git a/src/compiler/nir/meson.build b/src/compiler/nir/meson.build
index 80f2fc3d824..37c64cab9fd 100644
--- a/src/compiler/nir/meson.build
+++ b/src/compiler/nir/meson.build
@@ -177,6 +177,7 @@ files_libnir = files(
   'nir_move_vec_src_uses_to_dest.c',
   'nir_normalize_cubemap_coords.c',
   'nir_opt_access.c',
+  'nir_opt_barriers.c',
   'nir_opt_combine_stores.c',
   'nir_opt_comparison_pre.c',
   'nir_opt_conditional_discard.c',
diff --git a/src/compiler/nir/nir.h b/src/compiler/nir/nir.h
index ec281ad5317..1fe0264cbe5 100644
--- a/src/compiler/nir/nir.h
+++ b/src/compiler/nir/nir.h
@@ -4240,6 +4240,17 @@ bool nir_opt_algebraic_before_ffma(nir_shader *shader);
 bool nir_opt_algebraic_late(nir_shader *shader);
 bool nir_opt_constant_folding(nir_shader *shader);
 
+/* Try to combine a and b into a.  Return true if combination was possible,
+ * which will result in b being removed by the pass.  Return false if
+ * combination wasn't possible.
+ */
+typedef bool (*nir_combine_memory_barrier_cb)(
+   nir_intrinsic_instr *a, nir_intrinsic_instr *b, void *data);
+
+bool nir_opt_combine_memory_barriers(nir_shader *shader,
+                                     nir_combine_memory_barrier_cb combine_cb,
+                                     void *data);
+
 bool nir_opt_combine_stores(nir_shader *shader, nir_variable_mode modes);
 
 bool nir_copy_prop(nir_shader *shader);
diff --git a/src/compiler/nir/nir_opt_barriers.c b/src/compiler/nir/nir_opt_barriers.c
new file mode 100644
index 00000000000..c1de0ed9163
--- /dev/null
+++ b/src/compiler/nir/nir_opt_barriers.c
@@ -0,0 +1,82 @@
+/*
+ * Copyright © 2020 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include "nir.h"
+
+static bool
+nir_opt_combine_memory_barriers_impl(
+   nir_function_impl *impl, nir_combine_memory_barrier_cb combine_cb, void *data)
+{
+   bool progress = false;
+
+   nir_foreach_block(block, impl) {
+      nir_intrinsic_instr *prev = NULL;
+
+      nir_foreach_instr_safe(instr, block) {
+         if (instr->type != nir_instr_type_intrinsic) {
+            prev = NULL;
+            continue;
+         }
+
+         nir_intrinsic_instr *current = nir_instr_as_intrinsic(instr);
+         if (current->intrinsic != nir_intrinsic_scoped_memory_barrier) {
+            prev = NULL;
+            continue;
+         }
+
+         if (prev && combine_cb(prev, current, data)) {
+            nir_instr_remove(&current->instr);
+            progress = true;
+         } else {
+            prev = current;
+         }
+      }
+   }
+
+   if (progress) {
+      nir_metadata_preserve(impl, nir_metadata_block_index |
+                                  nir_metadata_dominance |
+                                  nir_metadata_live_ssa_defs);
+   }
+
+   return progress;
+}
+
+/* Combine adjacent scoped memory barriers. */
+bool
+nir_opt_combine_memory_barriers(
+   nir_shader *shader, nir_combine_memory_barrier_cb combine_cb, void *data)
+{
+   assert(combine_cb);
+
+   bool progress = false;
+
+   nir_foreach_function(function, shader) {
+      if (function->impl &&
+          nir_opt_combine_memory_barriers_impl(function->impl, combine_cb, data)) {
+         progress = true;
+      }
+   }
+
+   return progress;
+}