Merge branch 'master' into cycles_path_guiding

intern/cycles/device/optix/device_impl.cpp

@@ -854,12 +854,14 @@ bool OptiXDevice::load_osl_kernels()
         context, group_descs, 2, &group_options, nullptr, 0, &osl_groups[i * 2]));
   }
 
+  OptixStackSizes stack_size[NUM_PROGRAM_GROUPS] = {};
   vector<OptixStackSizes> osl_stack_size(osl_groups.size());
 
   /* Update SBT with new entries. */
   sbt_data.alloc(NUM_PROGRAM_GROUPS + osl_groups.size());
   for (int i = 0; i < NUM_PROGRAM_GROUPS; ++i) {
     optix_assert(optixSbtRecordPackHeader(groups[i], &sbt_data[i]));
+    optix_assert(optixProgramGroupGetStackSize(groups[i], &stack_size[i]));
   }
   for (size_t i = 0; i < osl_groups.size(); ++i) {
     if (osl_groups[i] != NULL) {
@@ -907,13 +909,15 @@ bool OptiXDevice::load_osl_kernels()
                                      0,
                                      &pipelines[PIP_SHADE]));
 
+    const unsigned int css = std::max(stack_size[PG_RGEN_SHADE_SURFACE_RAYTRACE].cssRG,
+                                      stack_size[PG_RGEN_SHADE_SURFACE_MNEE].cssRG);
     unsigned int dss = 0;
     for (unsigned int i = 0; i < osl_stack_size.size(); ++i) {
       dss = std::max(dss, osl_stack_size[i].dssDC);
     }
 
     optix_assert(optixPipelineSetStackSize(
-        pipelines[PIP_SHADE], 0, dss, 0, pipeline_options.usesMotionBlur ? 3 : 2));
+        pipelines[PIP_SHADE], 0, dss, css, pipeline_options.usesMotionBlur ? 3 : 2));
   }
 
   return !have_error();

intern/cycles/kernel/device/gpu/kernel.h

@@ -431,6 +431,7 @@ ccl_gpu_kernel_threads(GPU_PARALLEL_SORT_BLOCK_SIZE)
                                 metal_grid_id);
 #endif
 }
+ccl_gpu_kernel_postfix
 
 ccl_gpu_kernel_threads(GPU_PARALLEL_SORT_BLOCK_SIZE)
     ccl_gpu_kernel_signature(integrator_sort_write_pass,

intern/cycles/kernel/device/gpu/parallel_sorted_index.h

@@ -19,11 +19,10 @@ CCL_NAMESPACE_BEGIN
 #  define GPU_PARALLEL_SORTED_INDEX_DEFAULT_BLOCK_SIZE 512
 #endif
 #define GPU_PARALLEL_SORTED_INDEX_INACTIVE_KEY (~0)
+#define GPU_PARALLEL_SORT_BLOCK_SIZE 1024
 
 #if defined(__KERNEL_LOCAL_ATOMIC_SORT__)
 
-#  define GPU_PARALLEL_SORT_BLOCK_SIZE 1024
-
 #  define atomic_store_local(p, x) \
     atomic_store_explicit((threadgroup atomic_int *)p, x, memory_order_relaxed)
 #  define atomic_load_local(p) \

intern/cycles/kernel/device/oneapi/kernel.cpp

@@ -372,6 +372,16 @@ bool oneapi_enqueue_kernel(KernelContext *kernel_context,
               kg, cgh, global_size, local_size, args, oneapi_kernel_integrator_sorted_paths_array);
           break;
         }
+        case DEVICE_KERNEL_INTEGRATOR_SORT_BUCKET_PASS: {
+          oneapi_call(
+              kg, cgh, global_size, local_size, args, oneapi_kernel_integrator_sort_bucket_pass);
+          break;
+        }
+        case DEVICE_KERNEL_INTEGRATOR_SORT_WRITE_PASS: {
+          oneapi_call(
+              kg, cgh, global_size, local_size, args, oneapi_kernel_integrator_sort_write_pass);
+          break;
+        }
         case DEVICE_KERNEL_INTEGRATOR_COMPACT_PATHS_ARRAY: {
           oneapi_call(kg,
                       cgh,

source/blender/blenkernel/BKE_modifier.h

@@ -612,3 +612,25 @@ void BKE_modifier_blend_read_lib(struct BlendLibReader *reader, struct Object *o
 #ifdef __cplusplus
 }
 #endif
+
+#ifdef __cplusplus
+
+namespace blender::bke {
+
+/**
+ * A convenience class that can be used to set `ModifierData::execution_time` based on the lifetime
+ * of this class.
+ */
+class ScopedModifierTimer {
+ private:
+  ModifierData &md_;
+  double start_time_;
+
+ public:
+  ScopedModifierTimer(ModifierData &md);
+  ~ScopedModifierTimer();
+};
+
+}  // namespace blender::bke
+
+#endif

source/blender/blenkernel/intern/DerivedMesh.cc

@@ -760,6 +760,7 @@ static void mesh_calc_modifiers(struct Depsgraph *depsgraph,
       }
 
       if (mti->type == eModifierTypeType_OnlyDeform && !sculpt_dyntopo) {
+        blender::bke::ScopedModifierTimer modifier_timer{*md};
         if (!deformed_verts) {
           deformed_verts = BKE_mesh_vert_coords_alloc(mesh_input, &num_deformed_verts);
         }
@@ -846,6 +847,8 @@ static void mesh_calc_modifiers(struct Depsgraph *depsgraph,
       continue;
     }
 
+    blender::bke::ScopedModifierTimer modifier_timer{*md};
+
     /* Add orco mesh as layer if needed by this modifier. */
     if (mesh_final && mesh_orco && mti->requiredDataMask) {
       CustomData_MeshMasks mask = {0};
@@ -1323,6 +1326,8 @@ static void editbmesh_calc_modifiers(struct Depsgraph *depsgraph,
       continue;
     }
 
+    blender::bke::ScopedModifierTimer modifier_timer{*md};
+
     /* Add an orco mesh as layer if needed by this modifier. */
     if (mesh_final && mesh_orco && mti->requiredDataMask) {
       CustomData_MeshMasks mask = {0};

source/blender/blenkernel/intern/curves.cc

@@ -314,6 +314,8 @@ static void curves_evaluate_modifiers(struct Depsgraph *depsgraph,
       continue;
     }
 
+    blender::bke::ScopedModifierTimer modifier_timer{*md};
+
     if (mti->modifyGeometrySet != nullptr) {
       mti->modifyGeometrySet(md, &mectx, &geometry_set);
     }

source/blender/blenkernel/intern/displist.cc

@@ -613,6 +613,8 @@ void BKE_curve_calc_modifiers_pre(Depsgraph *depsgraph,
         continue;
       }
 
+      blender::bke::ScopedModifierTimer modifier_timer{*md};
+
       if (!deformedVerts) {
         deformedVerts = BKE_curve_nurbs_vert_coords_alloc(source_nurb, &numVerts);
       }
@@ -733,6 +735,8 @@ static GeometrySet curve_calc_modifiers_post(Depsgraph *depsgraph,
       continue;
     }
 
+    blender::bke::ScopedModifierTimer modifier_timer{*md};
+
     if (!geometry_set.has_mesh()) {
       geometry_set.replace_mesh(BKE_mesh_new_nomain(0, 0, 0, 0, 0));
     }

source/blender/blenkernel/intern/modifier.cc

@@ -11,6 +11,7 @@
 #define DNA_DEPRECATED_ALLOW
 
 #include <cfloat>
+#include <chrono>
 #include <cmath>
 #include <cstdarg>
 #include <cstddef>
@@ -1514,3 +1515,28 @@ void BKE_modifier_blend_read_lib(BlendLibReader *reader, Object *ob)
     }
   }
 }
+
+namespace blender::bke {
+
+using Clock = std::chrono::high_resolution_clock;
+
+static double get_current_time_in_seconds()
+{
+  return std::chrono::duration<double, std::chrono::seconds::period>(
+             Clock::now().time_since_epoch())
+      .count();
+}
+
+ScopedModifierTimer::ScopedModifierTimer(ModifierData &md) : md_(md)
+{
+  start_time_ = get_current_time_in_seconds();
+}
+
+ScopedModifierTimer::~ScopedModifierTimer()
+{
+  const double end_time = get_current_time_in_seconds();
+  const double duration = end_time - start_time_;
+  md_.execution_time = duration;
+}
+
+}  // namespace blender::bke

source/blender/blenkernel/intern/pointcloud.cc

@@ -380,6 +380,8 @@ static void pointcloud_evaluate_modifiers(struct Depsgraph *depsgraph,
       continue;
     }
 
+    blender::bke::ScopedModifierTimer modifier_timer{*md};
+
     if (mti->modifyGeometrySet) {
       mti->modifyGeometrySet(md, &mectx, &geometry_set);
     }

source/blender/blenkernel/intern/volume.cc

@@ -1108,6 +1108,8 @@ static void volume_evaluate_modifiers(struct Depsgraph *depsgraph,
       continue;
     }
 
+    blender::bke::ScopedModifierTimer modifier_timer{*md};
+
     if (mti->modifyGeometrySet) {
       mti->modifyGeometrySet(md, &mectx, &geometry_set);
     }

source/blender/editors/transform/transform.c

@@ -1484,6 +1484,12 @@ static void drawTransformPixel(const struct bContext *C, ARegion *region, void *
 
 void saveTransform(bContext *C, TransInfo *t, wmOperator *op)
 {
+  if (t->state == TRANS_CANCEL) {
+    /* No need to edit operator properties or tool settings if we are canceling the operation.
+     * These properties must match the original ones. */
+    return;
+  }
+
   ToolSettings *ts = CTX_data_tool_settings(C);
   PropertyRNA *prop;
 
@@ -1589,7 +1595,7 @@ void saveTransform(bContext *C, TransInfo *t, wmOperator *op)
     /* Update `ToolSettings` for properties that change during modal. */
     if (t->flag & T_MODAL) {
       /* Do we check for parameter? */
-      if (transformModeUseSnap(t) && !(t->tsnap.status & SNAP_FORCED)) {
+      if (transformModeUseSnap(t) && !(t->modifiers & MOD_SNAP_FORCED)) {
         if (!(t->modifiers & MOD_SNAP) != !(t->tsnap.flag & SCE_SNAP)) {
           /* Type is #eSnapFlag, but type must match various snap attributes in #ToolSettings. */
           short *snap_flag_ptr;

source/blender/editors/transform/transform.h

@@ -155,18 +155,18 @@ typedef enum {
   MOD_SNAP_INVERT = 1 << 3,
   MOD_CONSTRAINT_SELECT_PLANE = 1 << 4,
   MOD_NODE_ATTACH = 1 << 5,
+  MOD_SNAP_FORCED = 1 << 6,
 } eTModifier;
 ENUM_OPERATORS(eTModifier, MOD_NODE_ATTACH)
 
 /** #TransSnap.status */
 typedef enum eTSnap {
   SNAP_RESETTED = 0,
-  SNAP_FORCED = 1 << 0,
-  SNAP_SOURCE_FOUND = 1 << 1,
+  SNAP_SOURCE_FOUND = 1 << 0,
   /* Special flag for snap to grid. */
-  SNAP_TARGET_GRID_FOUND = 1 << 2,
-  SNAP_TARGET_FOUND = 1 << 3,
-  SNAP_MULTI_POINTS = 1 << 4,
+  SNAP_TARGET_GRID_FOUND = 1 << 1,
+  SNAP_TARGET_FOUND = 1 << 2,
+  SNAP_MULTI_POINTS = 1 << 3,
 } eTSnap;
 ENUM_OPERATORS(eTSnap, SNAP_MULTI_POINTS)
 

source/blender/editors/transform/transform_gizmo_3d.c

@@ -2688,7 +2688,7 @@ static wmGizmoGroup *gizmogroup_xform_find(TransInfo *t)
 
 void transform_gizmo_3d_model_from_constraint_and_mode_init(TransInfo *t)
 {
-  wmGizmo *gizmo_modal_current = WM_gizmomap_get_modal(t->region->gizmo_map);
+  wmGizmo *gizmo_modal_current = t->region ? WM_gizmomap_get_modal(t->region->gizmo_map) : NULL;
   if (!gizmo_modal_current || !ELEM(gizmo_modal_current->parent_gzgroup->type,
                                     g_GGT_xform_gizmo,
                                     g_GGT_xform_gizmo_context)) {

source/blender/editors/transform/transform_snap.cc

@@ -147,7 +147,7 @@ bool transformModeUseSnap(const TransInfo *t)
 
 static bool doForceIncrementSnap(const TransInfo *t)
 {
-  if (t->tsnap.status & SNAP_FORCED) {
+  if (t->modifiers & MOD_SNAP_FORCED) {
     return false;
   }
 
@@ -808,7 +808,8 @@ void initSnapping(TransInfo *t, wmOperator *op)
       if ((prop = RNA_struct_find_property(op->ptr, "snap_point")) &&
           RNA_property_is_set(op->ptr, prop)) {
         RNA_property_float_get_array(op->ptr, prop, t->tsnap.snap_target);
-        t->tsnap.status |= SNAP_FORCED | SNAP_TARGET_FOUND;
+        t->modifiers |= MOD_SNAP_FORCED;
+        t->tsnap.status |= SNAP_TARGET_FOUND;
       }
 
       /* snap align only defined in specific cases */

source/blender/makesdna/DNA_modifier_types.h

@@ -105,7 +105,8 @@ typedef struct ModifierData {
   struct ModifierData *next, *prev;
 
   int type, mode;
-  char _pad0[4];
+  /** Time in seconds that the modifier took to evaluate. This is only set on evaluated objects. */
+  float execution_time;
   short flag;
   /** An "expand" bit for each of the modifier's (sub)panels (#uiPanelDataExpansion). */
   short ui_expand_flag;

source/blender/makesrna/intern/rna_modifier.c

@@ -7355,6 +7355,14 @@ void RNA_def_modifier(BlenderRNA *brna)
   RNA_def_property_ui_icon(prop, ICON_SURFACE_DATA, 0);
   RNA_def_property_update(prop, 0, "rna_Modifier_update");
 
+  prop = RNA_def_property(srna, "execution_time", PROP_FLOAT, PROP_TIME_ABSOLUTE);
+  RNA_def_property_clear_flag(prop, PROP_EDITABLE);
+  RNA_def_property_ui_text(
+      prop,
+      "Execution Time",
+      "Time in seconds that the modifier took to evaluate. This is only set on evaluated objects. "
+      "If multiple modifiers run in parallel, execution time is not a reliable metric");
+
   /* types */
   rna_def_modifier_subsurf(brna);
   rna_def_modifier_lattice(brna);

source/blender/python/generic/bgl.c

@@ -40,6 +40,12 @@ static CLG_LogRef LOG = {"bgl"};
 
 static void report_deprecated_call(const char *function_name)
 {
+  /* Only report first 100 deprecated calls. BGL is typically used inside an handler that is
+   * triggered at refresh. */
+  static int times = 0;
+  while (times >= 100) {
+    return;
+  }
   char message[256];
   SNPRINTF(message,
            "'bgl.gl%s' is deprecated and will be removed in Blender 3.7. Report or update your "
@@ -47,6 +53,7 @@ static void report_deprecated_call(const char *function_name)
            function_name);
   CLOG_WARN(&LOG, "%s", message);
   PyErr_WarnEx(PyExc_DeprecationWarning, message, 1);
+  times++;
 }
 
 static void report_deprecated_call_to_user(void)