svn merge ^/trunk/blender -r43733:43751

19 files changed, 90 insertions, 129 deletions

diff --git a/intern/boolop/intern/BOP_CarveInterface.cpp b/intern/boolop/intern/BOP_CarveInterface.cpp
index 274d9cac0cf..5a847ff26d4 100644
--- a/intern/boolop/intern/BOP_CarveInterface.cpp
+++ b/intern/boolop/intern/BOP_CarveInterface.cpp
@@ -150,12 +150,19 @@ static MeshSet<3> *Carve_unionIntersectingMeshes(MeshSet<3> *poly,
 		MeshSet<3> *right = Carve_getIntersectedOperand(orig_meshes, precomputedAABB, otherAABB);
 
 		try {
-			MeshSet<3> *result = csg.compute(left, right, carve::csg::CSG::UNION, NULL, carve::csg::CSG::CLASSIFY_EDGE);
+			if(left->meshes.size()==0) {
+				delete left;
 
-			delete left;
-			delete right;
+				left = right;
+			}
+			else {
+				MeshSet<3> *result = csg.compute(left, right, carve::csg::CSG::UNION, NULL, carve::csg::CSG::CLASSIFY_EDGE);
 
-			left = result;
+				delete left;
+				delete right;
+
+				left = result;
+			}
 		}
 		catch(carve::exception e) {
 			std::cerr << "CSG failed, exception " << e.str() << std::endl;
@@ -492,7 +499,7 @@ BoolOpState BOP_performBooleanOperation(BoolOpType                    opType,
                                         CSG_VertexIteratorDescriptor  obBVertices)
 {
 	carve::csg::CSG::OP op;
-	MeshSet<3> *left, *right, *output;
+	MeshSet<3> *left, *right, *output = NULL;
 	carve::csg::CSG csg;
 	carve::geom3d::Vector min, max;
 	carve::interpolate::FaceAttr<uint> oface_num;
@@ -517,6 +524,17 @@ BoolOpState BOP_performBooleanOperation(BoolOpType                    opType,
 
 	Carve_prepareOperands(&left, &right, oface_num);
 
+	if(left->meshes.size() == 0 || right->meshes.size()==0) {
+		// normally sohuldn't happen (zero-faces objects are handled by modifier itself), but
+		// unioning intersecting meshes which doesn't have consistent normals might lead to
+		// empty result which wouldn't work here
+
+		delete left;
+		delete right;
+
+		return BOP_ERROR;
+	}
+
 	min.x = max.x = left->vertex_storage[0].v.x;
 	min.y = max.y = left->vertex_storage[0].v.y;
 	min.z = max.z = left->vertex_storage[0].v.z;
diff --git a/intern/cycles/blender/blender_session.cpp b/intern/cycles/blender/blender_session.cpp
index 5e3102fd7c7..d9adc5480dc 100644
--- a/intern/cycles/blender/blender_session.cpp
+++ b/intern/cycles/blender/blender_session.cpp
@@ -206,7 +206,7 @@ void BlenderSession::render()
 		vector<Pass> passes;
 		Pass::add(PASS_COMBINED, passes);
 
-		if(session_params.device.type == DEVICE_CPU) { /* todo */
+		if(session_params.device.advanced_shading) {
 			BL::RenderLayer::passes_iterator b_pass_iter;
 			
 			for(b_rlay.passes.begin(b_pass_iter); b_pass_iter != b_rlay.passes.end(); ++b_pass_iter) {
diff --git a/intern/cycles/device/device.cpp b/intern/cycles/device/device.cpp
index e4beb4d7d8c..cceec8b8e5c 100644
--- a/intern/cycles/device/device.cpp
+++ b/intern/cycles/device/device.cpp
@@ -183,6 +183,9 @@ Device *Device::create(DeviceInfo& info, bool background, int threads)
 			return NULL;
 	}
 
+	if(device)
+		device->info = info;
+
 	return device;
 }
 
diff --git a/intern/cycles/device/device.h b/intern/cycles/device/device.h
index b8fea4c4c69..af2567498d9 100644
--- a/intern/cycles/device/device.h
+++ b/intern/cycles/device/device.h
@@ -51,6 +51,7 @@ public:
 	string id;
 	int num;
 	bool display_device;
+	bool advanced_shading;
 	vector<DeviceInfo> multi_devices;
 
 	DeviceInfo()
@@ -59,6 +60,7 @@ public:
 		id = "CPU";
 		num = 0;
 		display_device = false;
+		advanced_shading = true;
 	}
 };
 
@@ -101,10 +103,8 @@ protected:
 public:
 	virtual ~Device() {}
 
-	virtual bool support_full_kernel() = 0;
-
 	/* info */
-	virtual string description() = 0;
+	DeviceInfo info;
 	virtual const string& error_message() { return error_msg; }
 
 	/* regular memory */
diff --git a/intern/cycles/device/device_cpu.cpp b/intern/cycles/device/device_cpu.cpp
index 2ca599f6c67..da977ed8472 100644
--- a/intern/cycles/device/device_cpu.cpp
+++ b/intern/cycles/device/device_cpu.cpp
@@ -72,16 +72,11 @@ public:
 		kernel_globals_free(kg);
 	}
 
-	bool support_full_kernel()
+	bool support_advanced_shading()
 	{
 		return true;
 	}
 
-	string description()
-	{
-		return system_cpu_brand_string();
-	}
-
 	void mem_alloc(device_memory& mem, MemoryType type)
 	{
 		mem.device_pointer = mem.data_pointer;
@@ -271,6 +266,7 @@ void device_cpu_info(vector<DeviceInfo>& devices)
 	info.description = system_cpu_brand_string();
 	info.id = "CPU";
 	info.num = 0;
+	info.advanced_shading = true;
 
 	devices.insert(devices.begin(), info);
 }
diff --git a/intern/cycles/device/device_cuda.cpp b/intern/cycles/device/device_cuda.cpp
index 55b467fc856..0c08baae3ff 100644
--- a/intern/cycles/device/device_cuda.cpp
+++ b/intern/cycles/device/device_cuda.cpp
@@ -106,11 +106,6 @@ public:
 		}
 	}
 
-	static int cuda_align_up(int& offset, int alignment)
-	{
-		return (offset + alignment - 1) & ~(alignment - 1);
-	}
-
 #ifdef NDEBUG
 #define cuda_abort()
 #else
@@ -194,26 +189,6 @@ public:
 		cuda_assert(cuCtxDetach(cuContext))
 	}
 
-	bool support_full_kernel()
-	{
-		int major, minor;
-		cuDeviceComputeCapability(&major, &minor, cuDevId);
-
-		return (major >= 2);
-	}
-
-	string description()
-	{
-		/* print device information */
-		char deviceName[256];
-
-		cuda_push_context();
-		cuDeviceGetName(deviceName, 256, cuDevId);
-		cuda_pop_context();
-
-		return string("CUDA ") + deviceName;
-	}
-
 	bool support_device(bool experimental)
 	{
 		if(!experimental) {
@@ -505,7 +480,7 @@ public:
 		offset += sizeof(d_rng_state);
 
 		int sample = task.sample;
-		offset = cuda_align_up(offset, __alignof(sample));
+		offset = align_up(offset, __alignof(sample));
 
 		cuda_assert(cuParamSeti(cuPathTrace, offset, task.sample))
 		offset += sizeof(task.sample);
@@ -569,7 +544,7 @@ public:
 		offset += sizeof(d_buffer);
 
 		int sample = task.sample;
-		offset = cuda_align_up(offset, __alignof(sample));
+		offset = align_up(offset, __alignof(sample));
 
 		cuda_assert(cuParamSeti(cuFilmConvert, offset, task.sample))
 		offset += sizeof(task.sample);
@@ -638,7 +613,7 @@ public:
 		offset += sizeof(d_offset);
 
 		int shader_eval_type = task.shader_eval_type;
-		offset = cuda_align_up(offset, __alignof(shader_eval_type));
+		offset = align_up(offset, __alignof(shader_eval_type));
 
 		cuda_assert(cuParamSeti(cuDisplace, offset, task.shader_eval_type))
 		offset += sizeof(task.shader_eval_type);
@@ -881,6 +856,10 @@ void device_cuda_info(vector<DeviceInfo>& devices)
 		info.id = string_printf("CUDA_%d", num);
 		info.num = num;
 
+		int major, minor;
+		cuDeviceComputeCapability(&major, &minor, num);
+		info.advanced_shading = (major >= 2);
+
 		/* if device has a kernel timeout, assume it is used for display */
 		if(cuDeviceGetAttribute(&attr, CU_DEVICE_ATTRIBUTE_KERNEL_EXEC_TIMEOUT, num) == CUDA_SUCCESS && attr == 1) {
 			info.display_device = true;
diff --git a/intern/cycles/device/device_multi.cpp b/intern/cycles/device/device_multi.cpp
index 41d0e268526..375719133b8 100644
--- a/intern/cycles/device/device_multi.cpp
+++ b/intern/cycles/device/device_multi.cpp
@@ -76,16 +76,6 @@ public:
 			delete sub.device;
 	}
 
-	bool support_full_kernel()
-	{
-		foreach(SubDevice& sub, devices) {
-			if(!sub.device->support_full_kernel())
-				return false;
-		}
-
-		return true;
-	}
-
 	const string& error_message()
 	{
 		foreach(SubDevice& sub, devices) {
@@ -99,38 +89,6 @@ public:
 		return error_msg;
 	}
 
-	string description()
-	{
-		/* create map to find duplicate descriptions */
-		map<string, int> dupli_map;
-		map<string, int>::iterator dt;
-
-		foreach(SubDevice& sub, devices) {
-			string key = sub.device->description();
-
-			if(dupli_map.find(key) == dupli_map.end())
-				dupli_map[key] = 1;
-			else
-				dupli_map[key]++;
-		}
-
-		/* generate string */
-		stringstream desc;
-		bool first = true;
-
-		for(dt = dupli_map.begin(); dt != dupli_map.end(); dt++) {
-			if(!first) desc << ", ";
-			first = false;
-
-			if(dt->second > 1)
-				desc << dt->second << "x " << dt->first;
-			else
-				desc << dt->first;
-		}
-
-		return desc.str();
-	}
-
 	bool load_kernels(bool experimental)
 	{
 		foreach(SubDevice& sub, devices)
@@ -344,6 +302,8 @@ static void device_multi_add(vector<DeviceInfo>& devices, DeviceType type, bool
 	map<string, int>::iterator dt;
 	int num_added = 0, num_display = 0;
 
+	info.advanced_shading = true;
+
 	foreach(DeviceInfo& subinfo, devices) {
 		if(subinfo.type == type) {
 			if(subinfo.display_device) {
@@ -363,6 +323,8 @@ static void device_multi_add(vector<DeviceInfo>& devices, DeviceType type, bool
 			info.multi_devices.push_back(subinfo);
 			if(subinfo.display_device)
 				info.display_device = true;
+			if(!subinfo.advanced_shading)
+				info.advanced_shading = false;
 			num_added++;
 		}
 	}
diff --git a/intern/cycles/device/device_network.cpp b/intern/cycles/device/device_network.cpp
index 14518b1507e..931890b5859 100644
--- a/intern/cycles/device/device_network.cpp
+++ b/intern/cycles/device/device_network.cpp
@@ -57,24 +57,6 @@ public:
 	{
 	}
 
-	bool support_full_kernel()
-	{
-		return false;
-	}
-
-	string description()
-	{
-		RPCSend snd(socket, "description");
-		snd.write();
-
-		RPCReceive rcv(socket);
-		string desc_string;
-
-		*rcv.archive & desc_string;
-
-		return desc_string + " (remote)";
-	}
-
 	void mem_alloc(device_memory& mem, MemoryType type)
 	{
 #if 0
diff --git a/intern/cycles/device/device_opencl.cpp b/intern/cycles/device/device_opencl.cpp
index ccfd8544362..9a55f957895 100644
--- a/intern/cycles/device/device_opencl.cpp
+++ b/intern/cycles/device/device_opencl.cpp
@@ -453,20 +453,6 @@ public:
 			clReleaseContext(cxContext);
 	}
 
-	bool support_full_kernel()
-	{
-		return false;
-	}
-
-	string description()
-	{
-		char name[1024];
-
-		clGetDeviceInfo(cdDevice, CL_DEVICE_NAME, sizeof(name), &name, NULL);
-
-		return string("OpenCL ") + name;
-	}
-
 	void mem_alloc(device_memory& mem, MemoryType type)
 	{
 		size_t size = mem.memory_size();
@@ -750,6 +736,7 @@ void device_opencl_info(vector<DeviceInfo>& devices)
 		info.num = num;
 		/* we don't know if it's used for display, but assume it is */
 		info.display_device = true;
+		info.advanced_shading = false;
 
 		devices.push_back(info);
 	}
diff --git a/intern/cycles/kernel/kernel_emission.h b/intern/cycles/kernel/kernel_emission.h
index b3a5b2bfcb4..513a453b585 100644
--- a/intern/cycles/kernel/kernel_emission.h
+++ b/intern/cycles/kernel/kernel_emission.h
@@ -27,6 +27,7 @@ __device float3 direct_emissive_eval(KernelGlobals *kg, float rando,
 	ShaderData sd;
 	float3 eval;
 
+#ifdef __BACKGROUND_MIS__
 	if(ls->type == LIGHT_BACKGROUND) {
 		Ray ray;
 		ray.D = ls->D;
@@ -36,7 +37,9 @@ __device float3 direct_emissive_eval(KernelGlobals *kg, float rando,
 		shader_setup_from_background(kg, &sd, &ray);
 		eval = shader_eval_background(kg, &sd, 0);
 	}
-	else {
+	else
+#endif
+	{
 		shader_setup_from_sample(kg, &sd, ls->P, ls->Ng, I, ls->shader, ls->object, ls->prim, u, v);
 		ls->Ng = sd.Ng;
 
@@ -164,6 +167,7 @@ __device float3 indirect_background(KernelGlobals *kg, Ray *ray, int path_flag,
 	float3 L = shader_eval_background(kg, &sd, path_flag);
 	shader_release(kg, &sd);
 
+#ifdef __BACKGROUND_MIS__
 	/* check if background light exists or if we should skip pdf */
 	int res = kernel_data.integrator.pdf_background_res;
 
@@ -175,6 +179,7 @@ __device float3 indirect_background(KernelGlobals *kg, Ray *ray, int path_flag,
 
 		return L*mis_weight;
 	}
+#endif
 
 	return L;
 #else
diff --git a/intern/cycles/kernel/kernel_light.h b/intern/cycles/kernel/kernel_light.h
index 4c2b69c2716..aa125180a94 100644
--- a/intern/cycles/kernel/kernel_light.h
+++ b/intern/cycles/kernel/kernel_light.h
@@ -192,6 +192,7 @@ __device void regular_light_sample(KernelGlobals *kg, int point,
 		ls->D = -D;
 		ls->t = FLT_MAX;
 	}
+#ifdef __BACKGROUND_MIS__
 	else if(type == LIGHT_BACKGROUND) {
 		/* infinite area light (e.g. light dome or env light) */
 		float3 D = background_light_sample(kg, randu, randv, pdf);
@@ -201,6 +202,7 @@ __device void regular_light_sample(KernelGlobals *kg, int point,
 		ls->D = -D;
 		ls->t = FLT_MAX;
 	}
+#endif
 	else {
 		ls->P = make_float3(data0.y, data0.z, data0.w);
 
diff --git a/intern/cycles/kernel/kernel_passes.h b/intern/cycles/kernel/kernel_passes.h
index cfd73c98bad..a1b3b0e9038 100644
--- a/intern/cycles/kernel/kernel_passes.h
+++ b/intern/cycles/kernel/kernel_passes.h
@@ -36,7 +36,7 @@ __device_inline void kernel_write_pass_float4(__global float *buffer, int sample
 	*buf = (sample == 0)? value: *buf + value;
 }
 
-__device void kernel_write_data_passes(KernelGlobals *kg, __global float *buffer, PathRadiance *L,
+__device_inline void kernel_write_data_passes(KernelGlobals *kg, __global float *buffer, PathRadiance *L,
 	ShaderData *sd, int sample, int path_flag, float3 throughput)
 {
 #ifdef __PASSES__
@@ -86,7 +86,7 @@ __device void kernel_write_data_passes(KernelGlobals *kg, __global float *buffer
 #endif
 }
 
-__device void kernel_write_light_passes(KernelGlobals *kg, __global float *buffer, PathRadiance *L, int sample)
+__device_inline void kernel_write_light_passes(KernelGlobals *kg, __global float *buffer, PathRadiance *L, int sample)
 {
 #ifdef __PASSES__
 	int flag = kernel_data.film.pass_flag;
diff --git a/intern/cycles/kernel/kernel_shader.h b/intern/cycles/kernel/kernel_shader.h
index 1d2cf46aa56..0f04af1275d 100644
--- a/intern/cycles/kernel/kernel_shader.h
+++ b/intern/cycles/kernel/kernel_shader.h
@@ -26,7 +26,6 @@
  *
  */
 
-
 #ifdef __OSL__
 
 #include "osl_shader.h"
diff --git a/intern/cycles/kernel/kernel_types.h b/intern/cycles/kernel/kernel_types.h
index b4b1da83162..477b08f2f87 100644
--- a/intern/cycles/kernel/kernel_types.h
+++ b/intern/cycles/kernel/kernel_types.h
@@ -70,9 +70,8 @@ CCL_NAMESPACE_BEGIN
 #ifdef __KERNEL_ADV_SHADING__
 #define __MULTI_CLOSURE__
 #define __TRANSPARENT_SHADOWS__
-#ifdef __KERNEL_CPU__
 #define __PASSES__
-#endif
+#define __BACKGROUND_MIS__
 #endif
 
 //#define __MULTI_LIGHT__
@@ -185,6 +184,9 @@ typedef float3 PathThroughput;
 struct PathRadiance {
 	int use_light_pass;
 
+	float3 emission;
+	float3 background;
+
 	float3 indirect;
 	float3 direct_throughput;
 	float3 direct_emission;
@@ -200,9 +202,6 @@ struct PathRadiance {
 	float3 indirect_diffuse;
 	float3 indirect_glossy;
 	float3 indirect_transmission;
-
-	float3 emission;
-	float3 background;
 };
 
 struct BsdfEval {
diff --git a/intern/cycles/render/buffers.cpp b/intern/cycles/render/buffers.cpp
index 56219482ef0..dd0ebf7195c 100644
--- a/intern/cycles/render/buffers.cpp
+++ b/intern/cycles/render/buffers.cpp
@@ -71,7 +71,7 @@ int BufferParams::get_passes_size()
 	foreach(Pass& pass, passes)
 		size += pass.components;
 	
-	return size;
+	return align_up(size, 4);
 }
 
 /* Render Buffers */
@@ -130,7 +130,7 @@ bool RenderBuffers::copy_from_device()
 	if(!buffer.device_pointer)
 		return false;
 
-	device->mem_copy_from(buffer, 0, params.width, params.height, sizeof(float4));
+	device->mem_copy_from(buffer, 0, params.width, params.height, params.get_passes_size()*sizeof(float));
 
 	return true;
 }
diff --git a/intern/cycles/render/film.cpp b/intern/cycles/render/film.cpp
index bc51384b873..376e9d6d0ca 100644
--- a/intern/cycles/render/film.cpp
+++ b/intern/cycles/render/film.cpp
@@ -21,12 +21,20 @@
 #include "film.h"
 #include "scene.h"
 
+#include "util_algorithm.h"
 #include "util_foreach.h"
 
 CCL_NAMESPACE_BEGIN
 
 /* Pass */
 
+static bool compare_pass_order(const Pass& a, const Pass& b)
+{
+	if(a.components == b.components)
+		return (a.type < b.type);
+	return (a.components > b.components);
+}
+
 void Pass::add(PassType type, vector<Pass>& passes)
 {
 	Pass pass;
@@ -106,6 +114,10 @@ void Pass::add(PassType type, vector<Pass>& passes)
 	}
 
 	passes.push_back(pass);
+
+	/* order from by components, to ensure alignment so passes with size 4
+	   come first and then passes with size 1 */
+	sort(passes.begin(), passes.end(), compare_pass_order);
 }
 
 bool Pass::equals(const vector<Pass>& A, const vector<Pass>& B)
@@ -219,6 +231,8 @@ void Film::device_update(Device *device, DeviceScene *dscene)
 		kfilm->pass_stride += pass.components;
 	}
 
+	kfilm->pass_stride = align_up(kfilm->pass_stride, 4);
+
 	need_update = false;
 }
 
diff --git a/intern/cycles/render/light.cpp b/intern/cycles/render/light.cpp
index 777e764558f..405aa800457 100644
--- a/intern/cycles/render/light.cpp
+++ b/intern/cycles/render/light.cpp
@@ -402,6 +402,16 @@ void LightManager::device_update_points(Device *device, DeviceScene *dscene, Sce
 
 	float4 *light_data = dscene->light_data.resize(scene->lights.size()*LIGHT_SIZE);
 
+	if(!device->info.advanced_shading) {
+		/* remove unsupported light */
+		foreach(Light *light, scene->lights) {
+			if(light->type == LIGHT_BACKGROUND) {
+				scene->lights.erase(std::remove(scene->lights.begin(), scene->lights.end(), light), scene->lights.end());
+				break;
+			}
+		}
+	}
+
 	for(size_t i = 0; i < scene->lights.size(); i++) {
 		Light *light = scene->lights[i];
 		float3 co = light->co;
diff --git a/intern/cycles/render/svm.cpp b/intern/cycles/render/svm.cpp
index f088a8143cc..ae666ddfe68 100644
--- a/intern/cycles/render/svm.cpp
+++ b/intern/cycles/render/svm.cpp
@@ -58,7 +58,7 @@ void SVMShaderManager::device_update(Device *device, DeviceScene *dscene, Scene
 	}
 	
 	bool sunsky_done = false;
-	bool use_multi_closure = device->support_full_kernel();
+	bool use_multi_closure = device->info.advanced_shading;
 
 	for(i = 0; i < scene->shaders.size(); i++) {
 		Shader *shader = scene->shaders[i];
diff --git a/intern/cycles/util/util_types.h b/intern/cycles/util/util_types.h
index 2c0ae13ad2a..efdda98571a 100644
--- a/intern/cycles/util/util_types.h
+++ b/intern/cycles/util/util_types.h
@@ -277,6 +277,11 @@ __device float4 make_float4(float x, float y, float z, float w)
 	return a;
 }
 
+__device int align_up(int offset, int alignment)
+{
+	return (offset + alignment - 1) & ~(alignment - 1);
+}
+
 #endif
 
 CCL_NAMESPACE_END