Cycles: Kernel address space changes for MSL

This is the first of a sequence of changes to support compiling Cycles kernels as MSL (Metal Shading Language) in preparation for a Metal GPU device implementation.

MSL requires that all pointer types be declared with explicit address space attributes (device, thread, etc...). There is already precedent for this with Cycles' address space macros (ccl_global, ccl_private, etc...), therefore the first step of MSL-enablement is to apply these consistently. Line-for-line this represents the largest change required to enable MSL. Applying this change first will simplify future patches as well as offering the emergent benefit of enhanced descriptiveness.

The vast majority of deltas in this patch fall into one of two cases:

- Ensuring ccl_private is specified for thread-local pointer types
- Ensuring ccl_global is specified for device-wide pointer types

Additionally, the ccl_addr_space qualifier can be removed. Prior to Cycles X, ccl_addr_space was used as a context-dependent address space qualifier, but now it is either redundant (e.g. in struct typedefs), or can be replaced by ccl_global in the case of pointer types. Associated function variants (e.g. lcg_step_float_addrspace) are also redundant.

In cases where address space qualifiers are chained with "const", this patch places the address space qualifier first. The rationale for this is that the choice of address space is likely to have the greater impact on runtime performance and overall architecture.

The final part of this patch is the addition of a metal/compat.h header. This is partially complete and will be extended in future patches, paving the way for the full Metal implementation.

Ref T92212

Reviewed By: brecht

Maniphest Tasks: T92212

Differential Revision: https://developer.blender.org/D12864

1 files changed, 25 insertions, 16 deletions

diff --git a/intern/cycles/kernel/kernel_light_background.h b/intern/cycles/kernel/kernel_light_background.h
index 493ed560bc6..3669ff50455 100644
--- a/intern/cycles/kernel/kernel_light_background.h
+++ b/intern/cycles/kernel/kernel_light_background.h
@@ -24,10 +24,10 @@ CCL_NAMESPACE_BEGIN
 
 #ifdef __BACKGROUND_MIS__
 
-ccl_device float3 background_map_sample(const KernelGlobals *kg,
+ccl_device float3 background_map_sample(ccl_global const KernelGlobals *kg,
                                         float randu,
                                         float randv,
-                                        float *pdf)
+                                        ccl_private float *pdf)
 {
   /* for the following, the CDF values are actually a pair of floats, with the
    * function value as X and the actual CDF as Y.  The last entry's function
@@ -109,7 +109,7 @@ ccl_device float3 background_map_sample(const KernelGlobals *kg,
 /* TODO(sergey): Same as above, after the release we should consider using
  * 'noinline' for all devices.
  */
-ccl_device float background_map_pdf(const KernelGlobals *kg, float3 direction)
+ccl_device float background_map_pdf(ccl_global const KernelGlobals *kg, float3 direction)
 {
   float2 uv = direction_to_equirectangular(direction);
   int res_x = kernel_data.background.map_res_x;
@@ -143,7 +143,11 @@ ccl_device float background_map_pdf(const KernelGlobals *kg, float3 direction)
 }
 
 ccl_device_inline bool background_portal_data_fetch_and_check_side(
-    const KernelGlobals *kg, float3 P, int index, float3 *lightpos, float3 *dir)
+    ccl_global const KernelGlobals *kg,
+    float3 P,
+    int index,
+    ccl_private float3 *lightpos,
+    ccl_private float3 *dir)
 {
   int portal = kernel_data.background.portal_offset + index;
   const ccl_global KernelLight *klight = &kernel_tex_fetch(__lights, portal);
@@ -158,8 +162,11 @@ ccl_device_inline bool background_portal_data_fetch_and_check_side(
   return false;
 }
 
-ccl_device_inline float background_portal_pdf(
-    const KernelGlobals *kg, float3 P, float3 direction, int ignore_portal, bool *is_possible)
+ccl_device_inline float background_portal_pdf(ccl_global const KernelGlobals *kg,
+                                              float3 P,
+                                              float3 direction,
+                                              int ignore_portal,
+                                              ccl_private bool *is_possible)
 {
   float portal_pdf = 0.0f;
 
@@ -219,7 +226,7 @@ ccl_device_inline float background_portal_pdf(
   return (num_possible > 0) ? portal_pdf / num_possible : 0.0f;
 }
 
-ccl_device int background_num_possible_portals(const KernelGlobals *kg, float3 P)
+ccl_device int background_num_possible_portals(ccl_global const KernelGlobals *kg, float3 P)
 {
   int num_possible_portals = 0;
   for (int p = 0; p < kernel_data.background.num_portals; p++) {
@@ -230,13 +237,13 @@ ccl_device int background_num_possible_portals(const KernelGlobals *kg, float3 P
   return num_possible_portals;
 }
 
-ccl_device float3 background_portal_sample(const KernelGlobals *kg,
+ccl_device float3 background_portal_sample(ccl_global const KernelGlobals *kg,
                                            float3 P,
                                            float randu,
                                            float randv,
                                            int num_possible,
-                                           int *sampled_portal,
-                                           float *pdf)
+                                           ccl_private int *sampled_portal,
+                                           ccl_private float *pdf)
 {
   /* Pick a portal, then re-normalize randv. */
   randv *= num_possible;
@@ -285,10 +292,10 @@ ccl_device float3 background_portal_sample(const KernelGlobals *kg,
   return zero_float3();
 }
 
-ccl_device_inline float3 background_sun_sample(const KernelGlobals *kg,
+ccl_device_inline float3 background_sun_sample(ccl_global const KernelGlobals *kg,
                                                float randu,
                                                float randv,
-                                               float *pdf)
+                                               ccl_private float *pdf)
 {
   float3 D;
   const float3 N = float4_to_float3(kernel_data.background.sun);
@@ -297,15 +304,15 @@ ccl_device_inline float3 background_sun_sample(const KernelGlobals *kg,
   return D;
 }
 
-ccl_device_inline float background_sun_pdf(const KernelGlobals *kg, float3 D)
+ccl_device_inline float background_sun_pdf(ccl_global const KernelGlobals *kg, float3 D)
 {
   const float3 N = float4_to_float3(kernel_data.background.sun);
   const float angle = kernel_data.background.sun.w;
   return pdf_uniform_cone(N, D, angle);
 }
 
-ccl_device_inline float3
-background_light_sample(const KernelGlobals *kg, float3 P, float randu, float randv, float *pdf)
+ccl_device_inline float3 background_light_sample(
+    ccl_global const KernelGlobals *kg, float3 P, float randu, float randv, ccl_private float *pdf)
 {
   float portal_method_pdf = kernel_data.background.portal_weight;
   float sun_method_pdf = kernel_data.background.sun_weight;
@@ -405,7 +412,9 @@ background_light_sample(const KernelGlobals *kg, float3 P, float randu, float ra
   return D;
 }
 
-ccl_device float background_light_pdf(const KernelGlobals *kg, float3 P, float3 direction)
+ccl_device float background_light_pdf(ccl_global const KernelGlobals *kg,
+                                      float3 P,
+                                      float3 direction)
 {
   float portal_method_pdf = kernel_data.background.portal_weight;
   float sun_method_pdf = kernel_data.background.sun_weight;