source/blender/blenkernel/BKE_pbvh_pixels.hh

@@ -4,8 +4,6 @@
 
 #pragma once
 
-#include <functional>
-
 #include "BLI_math_vector.hh"
 #include "BLI_rect.h"
 #include "BLI_vector.hh"

source/blender/blenlib/BLI_mesh_boolean.hh

@@ -11,8 +11,8 @@
 /* The boolean functions in Blenlib use exact arithmetic, so require GMP. */
 #ifdef WITH_GMP
 
+#  include "BLI_function_ref.hh"
 #  include "BLI_mesh_intersect.hh"
-#  include <functional>
 
 namespace blender::meshintersect {
 
@@ -46,7 +46,7 @@ enum class BoolOpType {
 IMesh boolean_mesh(IMesh &imesh,
                    BoolOpType op,
                    int nshapes,
-                   std::function<int(int)> shape_fn,
+                   FunctionRef<int(int)> shape_fn,
                    bool use_self,
                    bool hole_tolerant,
                    IMesh *imesh_triangulated,
@@ -60,7 +60,7 @@ IMesh boolean_mesh(IMesh &imesh,
 IMesh boolean_trimesh(IMesh &tm_in,
                       BoolOpType op,
                       int nshapes,
-                      std::function<int(int)> shape_fn,
+                      FunctionRef<int(int)> shape_fn,
                       bool use_self,
                       bool hole_tolerant,
                       IMeshArena *arena);

source/blender/blenlib/BLI_mesh_intersect.hh

@@ -16,6 +16,7 @@
 #  include <iosfwd>
 
 #  include "BLI_array.hh"
+#  include "BLI_function_ref.hh"
 #  include "BLI_index_range.hh"
 #  include "BLI_map.hh"
 #  include "BLI_math_mpq.hh"
@@ -403,7 +404,7 @@ IMesh trimesh_self_intersect(const IMesh &tm_in, IMeshArena *arena);
 
 IMesh trimesh_nary_intersect(const IMesh &tm_in,
                              int nshapes,
-                             std::function<int(int)> shape_fn,
+                             FunctionRef<int(int)> shape_fn,
                              bool use_self,
                              IMeshArena *arena);
 

source/blender/blenlib/intern/mesh_boolean.cc

@@ -2198,7 +2198,7 @@ static void propagate_windings_and_in_output_volume(PatchesInfo &pinfo,
                                                     int c_ambient,
                                                     BoolOpType op,
                                                     int nshapes,
-                                                    std::function<int(int)> shape_fn)
+                                                    FunctionRef<int(int)> shape_fn)
 {
   int dbg_level = 0;
   if (dbg_level > 0) {
@@ -2514,12 +2514,12 @@ static double3 calc_point_inside_tri_db(const Face &tri)
 class InsideShapeTestData {
  public:
   const IMesh &tm;
-  std::function<int(int)> shape_fn;
+  FunctionRef<int(int)> shape_fn;
   int nshapes;
   /* A per-shape vector of parity of hits of that shape. */
   Array<int> hit_parity;
 
-  InsideShapeTestData(const IMesh &tm, std::function<int(int)> shape_fn, int nshapes)
+  InsideShapeTestData(const IMesh &tm, FunctionRef<int(int)> shape_fn, int nshapes)
       : tm(tm), shape_fn(shape_fn), nshapes(nshapes)
   {
   }
@@ -2578,7 +2578,7 @@ static void inside_shape_callback(void *userdata,
  * \param tree: Contains all the triangles of \a tm and can be used for fast ray-casting.
  */
 static void test_tri_inside_shapes(const IMesh &tm,
-                                   std::function<int(int)> shape_fn,
+                                   FunctionRef<int(int)> shape_fn,
                                    int nshapes,
                                    int test_t_index,
                                    BVHTree *tree,
@@ -2726,11 +2726,8 @@ static void raycast_add_flipped(Vector<Face *> &out_faces, Face &tri, IMeshArena
  * when the input is not PWN, some patches can be both inside and outside
  * some shapes (e.g., a plane cutting through Suzanne's open eyes).
  */
-static IMesh raycast_tris_boolean(const IMesh &tm,
-                                  BoolOpType op,
-                                  int nshapes,
-                                  std::function<int(int)> shape_fn,
-                                  IMeshArena *arena)
+static IMesh raycast_tris_boolean(
+    const IMesh &tm, BoolOpType op, int nshapes, FunctionRef<int(int)> shape_fn, IMeshArena *arena)
 {
   constexpr int dbg_level = 0;
   if (dbg_level > 0) {
@@ -2804,7 +2801,7 @@ static IMesh raycast_tris_boolean(const IMesh &tm,
 static IMesh raycast_patches_boolean(const IMesh &tm,
                                      BoolOpType op,
                                      int nshapes,
-                                     std::function<int(int)> shape_fn,
+                                     FunctionRef<int(int)> shape_fn,
                                      const PatchesInfo &pinfo,
                                      IMeshArena *arena)
 {
@@ -3543,7 +3540,7 @@ static IMesh polymesh_from_trimesh_with_dissolve(const IMesh &tm_out,
 IMesh boolean_trimesh(IMesh &tm_in,
                       BoolOpType op,
                       int nshapes,
-                      std::function<int(int)> shape_fn,
+                      FunctionRef<int(int)> shape_fn,
                       bool use_self,
                       bool hole_tolerant,
                       IMeshArena *arena)
@@ -3691,7 +3688,7 @@ static void dump_test_spec(IMesh &imesh)
 IMesh boolean_mesh(IMesh &imesh,
                    BoolOpType op,
                    int nshapes,
-                   std::function<int(int)> shape_fn,
+                   FunctionRef<int(int)> shape_fn,
                    bool use_self,
                    bool hole_tolerant,
                    IMesh *imesh_triangulated,

source/blender/blenlib/intern/mesh_intersect.cc

@@ -2935,7 +2935,7 @@ IMesh trimesh_self_intersect(const IMesh &tm_in, IMeshArena *arena)
 
 IMesh trimesh_nary_intersect(const IMesh &tm_in,
                              int nshapes,
-                             std::function<int(int)> shape_fn,
+                             const FunctionRef<int(int)> shape_fn,
                              bool use_self,
                              IMeshArena *arena)
 {

source/blender/bmesh/tools/bmesh_boolean.cc

@@ -8,6 +8,8 @@
  * Main functions for boolean on a #BMesh (used by the tool and modifier)
  */
 
+#include <functional>
+
 #include "BLI_array.hh"
 #include "BLI_math_mpq.hh"
 #include "BLI_mesh_boolean.hh"

source/blender/draw/engines/eevee_next/shaders/eevee_depth_of_field_resolve_comp.glsl

@@ -20,7 +20,7 @@ shared uint shared_max_slight_focus_abs_coc;
  */
 float dof_slight_focus_coc_tile_get(vec2 frag_coord)
 {
-  if (all(equal(gl_LocalInvocationID, uvec3(0)))) {
+  if (gl_LocalInvocationIndex == 0u) {
     shared_max_slight_focus_abs_coc = floatBitsToUint(0.0);
   }
   barrier();

source/blender/draw/engines/eevee_next/shaders/eevee_depth_of_field_tiles_flatten_comp.glsl

@@ -33,7 +33,7 @@ uint dof_tile_large_coc_uint = floatBitsToUint(dof_tile_large_coc);
 
 void main()
 {
-  if (all(equal(gl_LocalInvocationID.xy, uvec2(0)))) {
+  if (gl_LocalInvocationIndex == 0u) {
     /* NOTE: Min/Max flipped because of inverted fg_coc sign. */
     fg_min_coc = floatBitsToUint(0.0);
     fg_max_coc = dof_tile_large_coc_uint;
@@ -61,7 +61,7 @@ void main()
 
   barrier();
 
-  if (all(equal(gl_LocalInvocationID.xy, uvec2(0)))) {
+  if (gl_LocalInvocationIndex == 0u) {
     if (fg_max_intersectable_coc == dof_tile_large_coc_uint) {
       fg_max_intersectable_coc = floatBitsToUint(0.0);
     }

source/blender/draw/engines/eevee_next/shaders/eevee_motion_blur_flatten_comp.glsl

@@ -37,7 +37,7 @@ uvec2 unpack_payload(uint payload)
 
 void main()
 {
-  if (all(equal(gl_LocalInvocationID.xy, uvec2(0)))) {
+  if (gl_LocalInvocationIndex == 0u) {
     payload_prev = 0u;
     payload_next = 0u;
   }
@@ -99,7 +99,7 @@ void main()
   }
   barrier();
 
-  if (all(equal(gl_LocalInvocationID.xy, uvec2(0)))) {
+  if (gl_LocalInvocationIndex == 0u) {
     ivec2 tile_co = ivec2(gl_WorkGroupID.xy);
     imageStore(out_tiles_img, tile_co, vec4(max_motion_prev, max_motion_next));
   }

source/blender/draw/engines/eevee_next/shaders/eevee_ray_tile_classify_comp.glsl

@@ -23,7 +23,7 @@ float ray_roughness_factor(RayTraceData raytrace, float roughness)
 
 void main()
 {
-  if (all(equal(gl_LocalInvocationID, uvec3(0)))) {
+  if (gl_LocalInvocationIndex == 0u) {
     /* Init shared variables. */
     for (int i = 0; i < 3; i++) {
       tile_contains_ray_tracing[i] = 0;
@@ -73,7 +73,7 @@ void main()
 
   barrier();
 
-  if (all(equal(gl_LocalInvocationID, uvec3(0)))) {
+  if (gl_LocalInvocationIndex == 0u) {
     ivec2 denoise_tile_co = ivec2(gl_WorkGroupID.xy);
     ivec2 tracing_tile_co = denoise_tile_co / uniform_buf.raytrace.resolution_scale;
 

source/blender/draw/engines/eevee_next/shaders/eevee_reflection_probe_update_irradiance_comp.glsl

@@ -56,7 +56,7 @@ void main()
   cooefs[store_index][3] = cooef.L1.Mp1;
 
   barrier();
-  if (gl_LocalInvocationID.x == 0) {
+  if (gl_LocalInvocationIndex == 0u) {
     /* Join results */
     vec4 result[4];
     result[0] = vec4(0.0);

source/blender/draw/engines/eevee_next/shaders/eevee_shadow_tilemap_bounds_comp.glsl

@@ -52,7 +52,7 @@ void main()
       local_max = max(local_max, z);
     }
 
-    if (gl_LocalInvocationID.x == 0) {
+    if (gl_LocalInvocationIndex == 0u) {
       global_min = floatBitsToOrderedInt(FLT_MAX);
       global_max = floatBitsToOrderedInt(-FLT_MAX);
     }
@@ -69,7 +69,7 @@ void main()
 
     barrier();
 
-    if (gl_LocalInvocationID.x == 0) {
+    if (gl_LocalInvocationIndex == 0u) {
       /* Final result. Min/Max of the whole dispatch. */
       atomicMin(light_buf[l_idx].clip_near, global_min);
       atomicMax(light_buf[l_idx].clip_far, global_max);

source/blender/draw/engines/eevee_next/shaders/eevee_shadow_tilemap_finalize_comp.glsl

@@ -68,7 +68,7 @@ void main()
     ShadowTileData tile = shadow_tile_unpack(tiles_buf[tile_index]);
 
     /* Compute update area. */
-    if (all(equal(gl_LocalInvocationID, uvec3(0)))) {
+    if (gl_LocalInvocationIndex == 0u) {
       rect_min_x = SHADOW_TILEMAP_RES;
       rect_min_y = SHADOW_TILEMAP_RES;
       rect_max_x = 0;
@@ -96,7 +96,7 @@ void main()
     ivec2 viewport_size = viewport_size_get(viewport_index);
 
     /* Issue one view if there is an update in the LOD. */
-    if (all(equal(gl_LocalInvocationID, uvec3(0)))) {
+    if (gl_LocalInvocationIndex == 0u) {
       bool lod_has_update = rect_min.x < rect_max.x;
       if (lod_has_update) {
         view_index = atomicAdd(statistics_buf.view_needed_count, 1);

source/blender/draw/engines/eevee_next/shaders/eevee_shadow_tilemap_init_comp.glsl

@@ -34,7 +34,7 @@ void main()
 
   barrier();
 
-  if (all(equal(gl_LocalInvocationID, uvec3(0)))) {
+  if (gl_LocalInvocationIndex == 0u) {
     /* Reset shift to not tag for update more than once per sync cycle. */
     tilemaps_buf[tilemap_index].grid_shift = ivec2(0);
 

source/blender/draw/engines/eevee_next/shaders/eevee_subsurface_setup_comp.glsl

@@ -17,7 +17,7 @@ void main(void)
 {
   ivec2 texel = ivec2(gl_GlobalInvocationID.xy);
 
-  if (all(equal(gl_LocalInvocationID.xy, uvec2(0)))) {
+  if (gl_LocalInvocationIndex == 0u) {
     has_visible_sss = 0u;
   }
 
@@ -53,7 +53,7 @@ void main(void)
 
   barrier();
 
-  if (all(equal(gl_LocalInvocationID.xy, uvec2(0)))) {
+  if (gl_LocalInvocationIndex == 0u) {
     if (has_visible_sss > 0u) {
       uint tile_id = atomicAdd(convolve_dispatch_buf.num_groups_x, 1u);
       convolve_tile_buf[tile_id] = packUvec2x16(gl_WorkGroupID.xy);

source/blender/draw/engines/eevee_next/shaders/infos/eevee_irradiance_cache_info.hh

@@ -87,6 +87,7 @@ GPU_SHADER_CREATE_INFO(eevee_surfel_light)
 
 GPU_SHADER_CREATE_INFO(eevee_surfel_cluster_build)
     .local_group_size(SURFEL_GROUP_SIZE)
+    .builtins(BuiltinBits::TEXTURE_ATOMIC)
     .additional_info("eevee_shared", "eevee_surfel_common", "draw_view")
     .image(0, GPU_R32I, Qualifier::READ_WRITE, ImageType::INT_3D_ATOMIC, "cluster_list_img")
     .compute_source("eevee_surfel_cluster_build_comp.glsl")

source/blender/draw/engines/overlay/overlay_next_private.hh

@@ -8,6 +8,8 @@
 
 #pragma once
 
+#include "BLI_function_ref.hh"
+
 #include "DRW_gpu_wrapper.hh"
 #include "DRW_render.h"
 #include "UI_resources.hh"
@@ -129,7 +131,7 @@ class ShaderModule {
   }
   ShaderPtr selectable_shader(const char *create_info_name);
   ShaderPtr selectable_shader(const char *create_info_name,
-                              std::function<void(gpu::shader::ShaderCreateInfo &info)> patch);
+                              FunctionRef<void(gpu::shader::ShaderCreateInfo &info)> patch);
 };
 
 struct Resources : public select::SelectMap {

source/blender/draw/engines/overlay/overlay_next_shader.cc

@@ -34,7 +34,8 @@ ShaderModule::ShaderPtr ShaderModule::selectable_shader(const char *create_info_
 }
 
 ShaderModule::ShaderPtr ShaderModule::selectable_shader(
-    const char *create_info_name, std::function<void(gpu::shader::ShaderCreateInfo &info)> patch)
+    const char *create_info_name,
+    const FunctionRef<void(gpu::shader::ShaderCreateInfo &info)> patch)
 {
   gpu::shader::ShaderCreateInfo info = *reinterpret_cast<const gpu::shader::ShaderCreateInfo *>(
       GPU_shader_create_info_get(create_info_name));

source/blender/gpu/metal/mtl_batch.mm

@@ -157,11 +157,14 @@ int MTLBatch::prepare_vertex_binding(MTLVertBuf *verts,
         /* Vertex/instance buffers provided have attribute data for attributes which are not needed
          * by this particular shader. This shader only needs binding information for the attributes
          * has in the shader interface. */
-        MTL_LOG_WARNING(
-            "MTLBatch: Could not find attribute with name '%s' (defined in active vertex format) "
-            "in the shader interface for shader '%s'",
-            name,
-            interface->get_name());
+        if (StringRefNull(name) != "dummy") {
+          MTL_LOG_WARNING(
+              "MTLBatch: Could not find attribute with name '%s' (defined in active vertex "
+              "format) "
+              "in the shader interface for shader '%s'",
+              name,
+              interface->get_name());
+        }
         continue;
       }
 

source/blender/gpu/metal/mtl_shader_generator.mm

@@ -50,6 +50,8 @@ char *MSLGeneratorInterface::msl_patch_default = nullptr;
 #define FRAGMENT_OUT_STRUCT_NAME "FragmentOut"
 #define FRAGMENT_TILE_IN_STRUCT_NAME "FragmentTileIn"
 
+#define ATOMIC_DEFINE_STR "#define MTL_SUPPORTS_TEXTURE_ATOMICS 1\n"
+
 /* -------------------------------------------------------------------- */
 /** \name Shader Translation utility functions.
  * \{ */
@@ -1060,6 +1062,13 @@ bool MTLShader::generate_msl_from_glsl(const shader::ShaderCreateInfo *info)
   std::stringstream ss_vertex;
   std::stringstream ss_fragment;
 
+  if (bool(info->builtins_ & BuiltinBits::TEXTURE_ATOMIC) &&
+      MTLBackend::get_capabilities().supports_texture_atomics)
+  {
+    ss_vertex << ATOMIC_DEFINE_STR;
+    ss_fragment << ATOMIC_DEFINE_STR;
+  }
+
   /* Generate specialization constants. */
   generate_specialization_constant_declarations(info, ss_vertex);
   generate_specialization_constant_declarations(info, ss_fragment);
@@ -1521,6 +1530,11 @@ bool MTLShader::generate_msl_from_glsl_compute(const shader::ShaderCreateInfo *i
 
   ss_compute << "#define GPU_ARB_texture_cube_map_array 1\n"
                 "#define GPU_ARB_shader_draw_parameters 1\n";
+  if (bool(info->builtins_ & BuiltinBits::TEXTURE_ATOMIC) &&
+      MTLBackend::get_capabilities().supports_texture_atomics)
+  {
+    ss_compute << ATOMIC_DEFINE_STR;
+  }
 
   generate_specialization_constant_declarations(info, ss_compute);
 
@@ -4027,7 +4041,7 @@ std::string MSLTextureResource::get_msl_wrapper_type_str() const
     case ImageType::INT_2D_ARRAY_ATOMIC:
     case ImageType::UINT_2D_ARRAY_ATOMIC: {
       if (supports_native_atomics) {
-        return "_mtl_combined_image_sampler_2d";
+        return "_mtl_combined_image_sampler_2d_array";
       }
       else {
         return "_mtl_combined_image_sampler_2d_array_atomic_fallback";

source/blender/gpu/shaders/metal/mtl_shader_defines.msl

@@ -13,13 +13,6 @@
 #pragma clang diagnostic ignored "-Wunused-variable"
 #pragma clang diagnostic ignored "-Wcomment"
 
-/** Feature support. */
-/* Native texture atomics supported in Metal 3.1+. Without these, we will use a fallback
- * implementation based on SSBOS.*/
-#if __METAL_VERSION__ >= 310
-#  define MTL_SUPPORTS_TEXTURE_ATOMICS 1
-#endif
-
 /* Base instance with offsets. */
 #define gpu_BaseInstance gl_BaseInstanceARB
 #define gpu_InstanceIndex (gl_InstanceID + gpu_BaseInstance)

source/blender/io/wavefront_obj/exporter/obj_export_file_writer.cc

@@ -336,7 +336,7 @@ static int get_smooth_group(const OBJMesh &mesh, const OBJExportParams &params,
 void OBJWriter::write_poly_elements(FormatHandler &fh,
                                     const IndexOffsets &offsets,
                                     const OBJMesh &obj_mesh_data,
-                                    std::function<const char *(int)> matname_fn)
+                                    FunctionRef<const char *(int)> matname_fn)
 {
   const func_vert_uv_normal_indices poly_element_writer = get_poly_element_writer(
       obj_mesh_data.tot_uv_vertices());

source/blender/io/wavefront_obj/exporter/obj_export_file_writer.hh

@@ -99,7 +99,7 @@ class OBJWriter : NonMovable, NonCopyable {
   void write_poly_elements(FormatHandler &fh,
                            const IndexOffsets &offsets,
                            const OBJMesh &obj_mesh_data,
-                           std::function<const char *(int)> matname_fn);
+                           FunctionRef<const char *(int)> matname_fn);
   /**
    * Write loose edges of a mesh as "l v1 v2".
    */

source/blender/io/wavefront_obj/exporter/obj_exporter.cc

@@ -213,7 +213,7 @@ static void write_mesh_objects(Vector<std::unique_ptr<OBJMesh>> exportable_as_me
         /* This function takes a 0-indexed slot index for the obj_mesh object and
          * returns the material name that we are using in the .obj file for it. */
         const auto *obj_mtlindices = mtlindices.is_empty() ? nullptr : &mtlindices[i];
-        std::function<const char *(int)> matname_fn = [&](int s) -> const char * {
+        auto matname_fn = [&](int s) -> const char * {
           if (!obj_mtlindices || s < 0 || s >= obj_mtlindices->size()) {
             return nullptr;
           }