Make Servers truly Thread Safe

-Rendering server now uses a split RID allocate/initialize internally, this allows generating RIDs immediately but initialization to happen later on the proper thread (as rendering APIs generally requiere to call on the right thread).
-RenderingServerWrapMT is no more, multithreading is done in RenderingServerDefault.
-Some functions like texture or mesh creation, when renderer supports it, can register and return immediately (so no waiting for server API to flush, and saving staging and command buffer memory).
-3D physics server changed to be made multithread friendly.
-Added PhysicsServer3DWrapMT to use 3D physics server from multiple threads.
-Disablet Bullet (too much effort to make multithread friendly, this needs to be fixed eventually).

1 files changed, 31 insertions, 18 deletions

diff --git a/servers/rendering/renderer_canvas_cull.cpp b/servers/rendering/renderer_canvas_cull.cpp
index e1ce52661c..e34bc9acb8 100644
--- a/servers/rendering/renderer_canvas_cull.cpp
+++ b/servers/rendering/renderer_canvas_cull.cpp
@@ -97,7 +97,7 @@ void _collect_ysort_children(RendererCanvasCull::Item *p_canvas_item, Transform2
 	}
 }
 
-void _mark_ysort_dirty(RendererCanvasCull::Item *ysort_owner, RID_PtrOwner<RendererCanvasCull::Item> &canvas_item_owner) {
+void _mark_ysort_dirty(RendererCanvasCull::Item *ysort_owner, RID_PtrOwner<RendererCanvasCull::Item, true> &canvas_item_owner) {
 	do {
 		ysort_owner->ysort_children_count = -1;
 		ysort_owner = canvas_item_owner.owns(ysort_owner->parent) ? canvas_item_owner.getornull(ysort_owner->parent) : nullptr;
@@ -356,12 +356,12 @@ bool RendererCanvasCull::was_sdf_used() {
 	return sdf_used;
 }
 
-RID RendererCanvasCull::canvas_create() {
+RID RendererCanvasCull::canvas_allocate() {
+	return canvas_owner.allocate_rid();
+}
+void RendererCanvasCull::canvas_initialize(RID p_rid) {
 	Canvas *canvas = memnew(Canvas);
-	ERR_FAIL_COND_V(!canvas, RID());
-	RID rid = canvas_owner.make_rid(canvas);
-
-	return rid;
+	canvas_owner.initialize_rid(p_rid, canvas);
 }
 
 void RendererCanvasCull::canvas_set_item_mirroring(RID p_canvas, RID p_item, const Point2 &p_mirroring) {
@@ -393,11 +393,12 @@ void RendererCanvasCull::canvas_set_parent(RID p_canvas, RID p_parent, float p_s
 	canvas->parent_scale = p_scale;
 }
 
-RID RendererCanvasCull::canvas_item_create() {
+RID RendererCanvasCull::canvas_item_allocate() {
+	return canvas_item_owner.allocate_rid();
+}
+void RendererCanvasCull::canvas_item_initialize(RID p_rid) {
 	Item *canvas_item = memnew(Item);
-	ERR_FAIL_COND_V(!canvas_item, RID());
-
-	return canvas_item_owner.make_rid(canvas_item);
+	canvas_item_owner.initialize_rid(p_rid, canvas_item);
 }
 
 void RendererCanvasCull::canvas_item_set_parent(RID p_item, RID p_parent) {
@@ -1075,10 +1076,13 @@ void RendererCanvasCull::canvas_item_set_canvas_group_mode(RID p_item, RS::Canva
 	}
 }
 
-RID RendererCanvasCull::canvas_light_create() {
+RID RendererCanvasCull::canvas_light_allocate() {
+	return canvas_light_owner.allocate_rid();
+}
+void RendererCanvasCull::canvas_light_initialize(RID p_rid) {
 	RendererCanvasRender::Light *clight = memnew(RendererCanvasRender::Light);
 	clight->light_internal = RSG::canvas_render->light_create();
-	return canvas_light_owner.make_rid(clight);
+	return canvas_light_owner.initialize_rid(p_rid, clight);
 }
 
 void RendererCanvasCull::canvas_light_set_mode(RID p_light, RS::CanvasLightMode p_mode) {
@@ -1268,10 +1272,13 @@ void RendererCanvasCull::canvas_light_set_shadow_smooth(RID p_light, float p_smo
 	clight->shadow_smooth = p_smooth;
 }
 
-RID RendererCanvasCull::canvas_light_occluder_create() {
+RID RendererCanvasCull::canvas_light_occluder_allocate() {
+	return canvas_light_occluder_owner.allocate_rid();
+}
+void RendererCanvasCull::canvas_light_occluder_initialize(RID p_rid) {
 	RendererCanvasRender::LightOccluderInstance *occluder = memnew(RendererCanvasRender::LightOccluderInstance);
 
-	return canvas_light_occluder_owner.make_rid(occluder);
+	return canvas_light_occluder_owner.initialize_rid(p_rid, occluder);
 }
 
 void RendererCanvasCull::canvas_light_occluder_attach_to_canvas(RID p_occluder, RID p_canvas) {
@@ -1349,10 +1356,13 @@ void RendererCanvasCull::canvas_light_occluder_set_light_mask(RID p_occluder, in
 	occluder->light_mask = p_mask;
 }
 
-RID RendererCanvasCull::canvas_occluder_polygon_create() {
+RID RendererCanvasCull::canvas_occluder_polygon_allocate() {
+	return canvas_light_occluder_polygon_owner.allocate_rid();
+}
+void RendererCanvasCull::canvas_occluder_polygon_initialize(RID p_rid) {
 	LightOccluderPolygon *occluder_poly = memnew(LightOccluderPolygon);
 	occluder_poly->occluder = RSG::canvas_render->occluder_polygon_create();
-	return canvas_light_occluder_polygon_owner.make_rid(occluder_poly);
+	return canvas_light_occluder_polygon_owner.initialize_rid(p_rid, occluder_poly);
 }
 
 void RendererCanvasCull::canvas_occluder_polygon_set_shape(RID p_occluder_polygon, const Vector<Vector2> &p_shape, bool p_closed) {
@@ -1393,8 +1403,11 @@ void RendererCanvasCull::canvas_set_shadow_texture_size(int p_size) {
 	RSG::canvas_render->set_shadow_texture_size(p_size);
 }
 
-RID RendererCanvasCull::canvas_texture_create() {
-	return RSG::storage->canvas_texture_create();
+RID RendererCanvasCull::canvas_texture_allocate() {
+	return RSG::storage->canvas_texture_allocate();
+}
+void RendererCanvasCull::canvas_texture_initialize(RID p_rid) {
+	RSG::storage->canvas_texture_initialize(p_rid);
 }
 
 void RendererCanvasCull::canvas_texture_set_channel(RID p_canvas_texture, RS::CanvasTextureChannel p_channel, RID p_texture) {