nir: Add a pass for lowering CL-style image ops to texture ops

In CL 1.2, images are required to be either read-only or write-only.  We
can always translate the read-only image ops to texture ops.  In CL 2.0
(and an extension), the ability is added to have read-write images but
sampling (with a sampler) is only allowed on read-only images.  As long
as we only lower read-only images to texture ops, everything should stay
consistent.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6578>

diff --git a/src/compiler/Makefile.sources b/src/compiler/Makefile.sources
index d2a2a16..9b9d8af 100644 (file)
--- a/src/compiler/Makefile.sources
+++ b/src/compiler/Makefile.sources
@@ -243,6 +243,7 @@ NIR_FILES = \
        nir/nir_lower_bool_to_bitsize.c \
        nir/nir_lower_bool_to_float.c \
        nir/nir_lower_bool_to_int32.c \
+       nir/nir_lower_cl_images_to_tex.c \
        nir/nir_lower_clamp_color_outputs.c \
        nir/nir_lower_clip.c \
        nir/nir_lower_clip_cull_distance_arrays.c \

diff --git a/src/compiler/nir/meson.build b/src/compiler/nir/meson.build
index 71fed70..e563bd2 100644 (file)
--- a/src/compiler/nir/meson.build
+++ b/src/compiler/nir/meson.build
@@ -122,6 +122,7 @@ files_libnir = files(
   'nir_lower_bool_to_bitsize.c',
   'nir_lower_bool_to_float.c',
   'nir_lower_bool_to_int32.c',
+  'nir_lower_cl_images_to_tex.c',
   'nir_lower_clamp_color_outputs.c',
   'nir_lower_clip.c',
   'nir_lower_clip_cull_distance_arrays.c',

diff --git a/src/compiler/nir/nir.h b/src/compiler/nir/nir.h
index 7bc6f2a..d3ba1bf 100644 (file)
--- a/src/compiler/nir/nir.h
+++ b/src/compiler/nir/nir.h
@@ -4551,6 +4551,8 @@ typedef struct nir_lower_tex_options {
 bool nir_lower_tex(nir_shader *shader,
                    const nir_lower_tex_options *options);
 
+bool nir_lower_cl_images_to_tex(nir_shader *shader);
+
 enum nir_lower_non_uniform_access_type {
    nir_lower_non_uniform_ubo_access     = (1 << 0),
    nir_lower_non_uniform_ssbo_access    = (1 << 1),

diff --git a/src/compiler/nir/nir_lower_cl_images_to_tex.c b/src/compiler/nir/nir_lower_cl_images_to_tex.c
new file mode 100644 (file)
index 0000000..07064e5
--- /dev/null
+++ b/src/compiler/nir/nir_lower_cl_images_to_tex.c
@@ -0,0 +1,162 @@
+/*
+ * 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"
+#include "nir_builder.h"
+
+static bool
+lower_cl_images_to_tex_impl(nir_function_impl *impl)
+{
+   bool progress = false;
+
+   nir_builder b;
+   nir_builder_init(&b, impl);
+
+   nir_foreach_block(block, impl) {
+      nir_foreach_instr_safe(instr, block) {
+         if (instr->type != nir_instr_type_intrinsic)
+            continue;
+
+         nir_intrinsic_instr *intrin = nir_instr_as_intrinsic(instr);
+
+         unsigned num_srcs;
+         nir_texop texop;
+         switch (intrin->intrinsic) {
+         case nir_intrinsic_image_deref_load:
+            texop = nir_texop_txf;
+            num_srcs = 3;
+            break;
+         case nir_intrinsic_image_deref_size:
+            texop = nir_texop_txs;
+            num_srcs = 2;
+            break;
+         default:
+            /* Not an op we can lower */
+            continue;
+         }
+
+         /* In CL 1.2, images are required to be either read-only or
+          * write-only.  We can always translate the read-only image ops to
+          * texture ops.  In CL 2.0 (and an extension), the ability is added
+          * to have read-write images but sampling (with a sampler) is only
+          * allowed on read-only images.  As long as we only lower read-only
+          * images to texture ops, everything should stay consistent.
+          */
+         if (!(nir_intrinsic_access(intrin) & ACCESS_NON_WRITEABLE))
+            continue;
+
+         nir_deref_instr *deref = nir_src_as_deref(intrin->src[0]);
+
+         b.cursor = nir_instr_remove(&intrin->instr);
+
+         nir_tex_instr *tex = nir_tex_instr_create(b.shader, num_srcs);
+         tex->op = texop;
+
+         tex->sampler_dim = glsl_get_sampler_dim(deref->type);
+         tex->is_array = glsl_sampler_type_is_array(deref->type);
+         tex->is_shadow = false;
+
+         unsigned coord_components =
+            glsl_get_sampler_dim_coordinate_components(tex->sampler_dim);
+         if (glsl_sampler_type_is_array(deref->type))
+            tex->coord_components++;
+
+         tex->src[0].src_type = nir_tex_src_texture_deref;
+         tex->src[0].src = nir_src_for_ssa(&deref->dest.ssa);
+
+         switch (intrin->intrinsic) {
+         case nir_intrinsic_image_deref_load: {
+            assert(intrin->src[1].is_ssa);
+            tex->coord_components = coord_components;
+            nir_ssa_def *coord =
+               nir_channels(&b, intrin->src[1].ssa,
+                            (1 << tex->coord_components) - 1);
+            tex->src[1].src_type = nir_tex_src_coord;
+            tex->src[1].src = nir_src_for_ssa(coord);
+
+            assert(intrin->src[3].is_ssa);
+            nir_ssa_def *lod = intrin->src[3].ssa;
+            tex->src[2].src_type = nir_tex_src_lod;
+            tex->src[2].src = nir_src_for_ssa(lod);
+
+            assert(num_srcs == 3);
+
+            tex->dest_type = nir_intrinsic_type(intrin);
+            nir_ssa_dest_init(&tex->instr, &tex->dest, 4, 32, NULL);
+            break;
+         }
+
+         case nir_intrinsic_image_deref_size: {
+            assert(intrin->src[1].is_ssa);
+            nir_ssa_def *lod = intrin->src[1].ssa;
+            tex->src[1].src_type = nir_tex_src_lod;
+            tex->src[1].src = nir_src_for_ssa(lod);
+
+            assert(num_srcs == 2);
+
+            tex->dest_type = nir_type_uint32;
+            nir_ssa_dest_init(&tex->instr, &tex->dest,
+                              coord_components, 32, NULL);
+            break;
+         }
+
+         default:
+            unreachable("Unsupported intrinsic");
+         }
+
+         nir_builder_instr_insert(&b, &tex->instr);
+
+         nir_ssa_def *res = &tex->dest.ssa;
+         if (res->num_components != intrin->dest.ssa.num_components) {
+            unsigned num_components = intrin->dest.ssa.num_components;
+            res = nir_channels(&b, res, (1 << num_components) - 1);
+         }
+
+         nir_ssa_def_rewrite_uses(&intrin->dest.ssa, nir_src_for_ssa(res));
+         progress = true;
+      }
+   }
+
+   if (progress) {
+      nir_metadata_preserve(impl, nir_metadata_block_index |
+                                  nir_metadata_dominance);
+   } else {
+      nir_metadata_preserve(impl, nir_metadata_all);
+   }
+
+   return progress;
+}
+
+/** Lowers OpenCL image ops to texture ops for read-only images */
+bool
+nir_lower_cl_images_to_tex(nir_shader *shader)
+{
+   bool progress = false;
+
+   nir_foreach_function(function, shader) {
+      if (function->impl && lower_cl_images_to_tex_impl(function->impl))
+         progress = true;
+   }
+
+   return progress;
+}