Merge branch 'master' into blender2.8

Conflicts:
	source/blender/blenloader/intern/writefile.c

7 files changed, 14 insertions, 19 deletions

diff --git a/intern/cycles/device/device_cpu.cpp b/intern/cycles/device/device_cpu.cpp
index 273c3b48936..d1027f07510 100644
--- a/intern/cycles/device/device_cpu.cpp
+++ b/intern/cycles/device/device_cpu.cpp
@@ -855,9 +855,8 @@ int2 CPUSplitKernel::split_kernel_local_size()
 	return make_int2(1, 1);
 }
 
-int2 CPUSplitKernel::split_kernel_global_size(device_memory& /*kg*/, device_memory& /*data*/, DeviceTask *task) {
-	/* TODO(mai): this needs investigation but cpu gives incorrect render if global size doesnt match tile size */
-	return task->requested_tile_size;
+int2 CPUSplitKernel::split_kernel_global_size(device_memory& /*kg*/, device_memory& /*data*/, DeviceTask * /*task*/) {
+	return make_int2(64, 1);
 }
 
 uint64_t CPUSplitKernel::state_buffer_size(device_memory& kernel_globals, device_memory& /*data*/, size_t num_threads) {
diff --git a/intern/cycles/device/device_split_kernel.cpp b/intern/cycles/device/device_split_kernel.cpp
index 8925ef47b2e..7a4d8c419b5 100644
--- a/intern/cycles/device/device_split_kernel.cpp
+++ b/intern/cycles/device/device_split_kernel.cpp
@@ -142,6 +142,7 @@ bool DeviceSplitKernel::path_trace(DeviceTask *task,
 
 	/* Number of elements in the global state buffer */
 	int num_global_elements = global_size[0] * global_size[1];
+	assert(num_global_elements % WORK_POOL_SIZE == 0);
 
 	/* Allocate all required global memory once. */
 	if(first_tile) {
diff --git a/intern/cycles/kernel/kernel_random.h b/intern/cycles/kernel/kernel_random.h
index e773753396f..ad3fa32a809 100644
--- a/intern/cycles/kernel/kernel_random.h
+++ b/intern/cycles/kernel/kernel_random.h
@@ -191,14 +191,14 @@ ccl_device void path_rng_end(KernelGlobals *kg, ccl_global uint *rng_state, RNG
 
 /* Linear Congruential Generator */
 
-ccl_device_forceinline float path_rng_1D(KernelGlobals *kg, RNG& rng, int sample, int num_samples, int dimension)
+ccl_device_forceinline float path_rng_1D(KernelGlobals *kg, RNG *rng, int sample, int num_samples, int dimension)
 {
 	/* implicit mod 2^32 */
-	rng = (1103515245*(rng) + 12345);
-	return (float)rng * (1.0f/(float)0xFFFFFFFF);
+	*rng = (1103515245*(*rng) + 12345);
+	return (float)*rng * (1.0f/(float)0xFFFFFFFF);
 }
 
-ccl_device_inline void path_rng_2D(KernelGlobals *kg, RNG& rng, int sample, int num_samples, int dimension, float *fx, float *fy)
+ccl_device_inline void path_rng_2D(KernelGlobals *kg, RNG *rng, int sample, int num_samples, int dimension, float *fx, float *fy)
 {
 	*fx = path_rng_1D(kg, rng, sample, num_samples, dimension);
 	*fy = path_rng_1D(kg, rng, sample, num_samples, dimension + 1);
diff --git a/intern/cycles/kernel/kernel_types.h b/intern/cycles/kernel/kernel_types.h
index b6b891627bf..245832ae163 100644
--- a/intern/cycles/kernel/kernel_types.h
+++ b/intern/cycles/kernel/kernel_types.h
@@ -32,11 +32,6 @@
 #  define ccl_addr_space
 #endif
 
-#if defined(__SPLIT_KERNEL__) && !defined(__COMPUTE_DEVICE_GPU__)
-/* TODO(mai): need to investigate how this effects the kernel, as cpu kernel crashes without this right now */
-#define __COMPUTE_DEVICE_GPU__
-#endif
-
 CCL_NAMESPACE_BEGIN
 
 /* constants */
diff --git a/intern/cycles/kernel/kernel_volume.h b/intern/cycles/kernel/kernel_volume.h
index 608350a9038..dcab138dc84 100644
--- a/intern/cycles/kernel/kernel_volume.h
+++ b/intern/cycles/kernel/kernel_volume.h
@@ -360,7 +360,7 @@ ccl_device VolumeIntegrateResult kernel_volume_integrate_homogeneous(
     ShaderData *sd,
     PathRadiance *L,
     ccl_addr_space float3 *throughput,
-	ccl_addr_space RNG *rng,
+    ccl_addr_space RNG *rng,
     bool probalistic_scatter)
 {
 	VolumeShaderCoefficients coeff;
diff --git a/intern/cycles/kernel/split/kernel_shader_eval.h b/intern/cycles/kernel/split/kernel_shader_eval.h
index fc966b77b2c..b739f86338c 100644
--- a/intern/cycles/kernel/split/kernel_shader_eval.h
+++ b/intern/cycles/kernel/split/kernel_shader_eval.h
@@ -38,11 +38,11 @@ ccl_device void kernel_shader_eval(KernelGlobals *kg,
 	                          kernel_split_params.queue_size,
 	                          0);
 
-	if(ray_index == QUEUE_EMPTY_SLOT) {
-		return;
+	char enqueue_flag = 0;
+	if((ray_index != QUEUE_EMPTY_SLOT) && IS_STATE(kernel_split_state.ray_state, ray_index, RAY_TO_REGENERATE)) {
+		enqueue_flag = 1;
 	}
 
-	char enqueue_flag = (IS_STATE(kernel_split_state.ray_state, ray_index, RAY_TO_REGENERATE)) ? 1 : 0;
 	enqueue_ray_index_local(ray_index,
 	                        QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS,
 	                        enqueue_flag,
@@ -52,7 +52,7 @@ ccl_device void kernel_shader_eval(KernelGlobals *kg,
 	                        kernel_split_params.queue_index);
 
 	/* Continue on with shader evaluation. */
-	if(IS_STATE(kernel_split_state.ray_state, ray_index, RAY_ACTIVE)) {
+	if((ray_index != QUEUE_EMPTY_SLOT) && IS_STATE(kernel_split_state.ray_state, ray_index, RAY_ACTIVE)) {
 		Intersection isect = kernel_split_state.isect[ray_index];
 		ccl_global uint *rng = &kernel_split_state.rng[ray_index];
 		ccl_global PathState *state = &kernel_split_state.path_state[ray_index];
diff --git a/intern/cycles/kernel/split/kernel_subsurface_scatter.h b/intern/cycles/kernel/split/kernel_subsurface_scatter.h
index 709a296c9a0..4eaa7f56332 100644
--- a/intern/cycles/kernel/split/kernel_subsurface_scatter.h
+++ b/intern/cycles/kernel/split/kernel_subsurface_scatter.h
@@ -46,12 +46,12 @@ ccl_device void kernel_subsurface_scatter(KernelGlobals *kg,
 	}
 #endif
 
+	char enqueue_flag = 0;
+
 #ifndef __COMPUTE_DEVICE_GPU__
 	if(ray_index != QUEUE_EMPTY_SLOT) {
 #endif
 
-
-	char enqueue_flag = 0;
 	ccl_global char *ray_state = kernel_split_state.ray_state;
 	ccl_global PathState *state = &kernel_split_state.path_state[ray_index];
 	PathRadiance *L = &kernel_split_state.path_radiance[ray_index];