fix missing type conversion

Only 3D in F1 mode with euclidean distance is support right now.

CMakeLists.txt

@@ -251,6 +251,16 @@ if(WITH_GHOST_X11)
   option(WITH_X11_ALPHA     "Enable X11 transparent background"                     ON)
 endif()
 
+if(UNIX AND NOT APPLE)
+  option(WITH_SYSTEM_GLEW "Use GLEW OpenGL wrapper library provided by the operating system" OFF)
+  option(WITH_SYSTEM_GLES "Use OpenGL ES library provided by the operating system"           ON)
+else()
+  # not an option for other OS's
+  set(WITH_SYSTEM_GLEW OFF)
+  set(WITH_SYSTEM_GLES OFF)
+endif()
+
+
 if(UNIX AND NOT APPLE)
   option(WITH_SYSTEM_EIGEN3 "Use the systems Eigen3 library" OFF)
 endif()
@@ -339,7 +349,7 @@ mark_as_advanced(WITH_SYSTEM_GLOG)
 option(WITH_FREESTYLE     "Enable Freestyle (advanced edges rendering)" ON)
 
 # Misc
-if(WIN32 OR APPLE)
+if(WIN32)
   option(WITH_INPUT_IME "Enable Input Method Editor (IME) for complex Asian character input" ON)
 endif()
 option(WITH_INPUT_NDOF "Enable NDOF input devices (SpaceNavigator and friends)" ON)
@@ -393,14 +403,14 @@ set(CYCLES_TEST_DEVICES CPU CACHE STRING "Run regression tests on the specified
 set(CYCLES_CUDA_BINARIES_ARCH sm_30 sm_35 sm_37 sm_50 sm_52 sm_60 sm_61 sm_70 sm_75 sm_86 compute_75 CACHE STRING "CUDA architectures to build binaries for")
 mark_as_advanced(CYCLES_CUDA_BINARIES_ARCH)
 unset(PLATFORM_DEFAULT)
-option(WITH_CYCLES_LOGGING      "Build Cycles with logging support" ON)
-option(WITH_CYCLES_DEBUG_NAN    "Build Cycles with additional asserts for detecting NaNs and invalid values" OFF)
+option(WITH_CYCLES_LOGGING  "Build Cycles with logging support" ON)
+option(WITH_CYCLES_DEBUG    "Build Cycles with extra debug capabilities" OFF)
 option(WITH_CYCLES_NATIVE_ONLY  "Build Cycles with native kernel only (which fits current CPU, use for development only)" OFF)
 option(WITH_CYCLES_KERNEL_ASAN  "Build Cycles kernels with address sanitizer when WITH_COMPILER_ASAN is on, even if it's very slow" OFF)
 mark_as_advanced(WITH_CYCLES_KERNEL_ASAN)
 mark_as_advanced(WITH_CYCLES_CUBIN_COMPILER)
 mark_as_advanced(WITH_CYCLES_LOGGING)
-mark_as_advanced(WITH_CYCLES_DEBUG_NAN)
+mark_as_advanced(WITH_CYCLES_DEBUG)
 mark_as_advanced(WITH_CYCLES_NATIVE_ONLY)
 
 option(WITH_CYCLES_DEVICE_CUDA              "Enable Cycles CUDA compute support" ON)
@@ -475,32 +485,15 @@ endif()
 
 # OpenGL
 
-if(UNIX AND NOT APPLE)
-  # GLEW can only built with either GLX or EGL support and most binary
-  # distributions are built with GLX support. So we always compile GLEW
-  # with EGL support manually, and the options are no longer available.
-  set(WITH_SYSTEM_GLEW OFF)
-  set(WITH_SYSTEM_GLES ON)
-
-  # Always use EGL instead of GLX, for X11, Wayland and headless.
-  set(WITH_GL_EGL ON)
-else()
-  # System GLEW and GLES were never an option on other platforms.
-  set(WITH_SYSTEM_GLEW OFF)
-  set(WITH_SYSTEM_GLES OFF)
-
-  # Experimental EGL option.
-  option(WITH_GL_EGL "Use the EGL OpenGL system library instead of the platform specific OpenGL system library (CGL or WGL)" OFF)
-  mark_as_advanced(WITH_GL_EGL)
-endif()
-
 option(WITH_OPENGL              "When off limits visibility of the opengl headers to just bf_gpu and gawain (temporary option for development purposes)" ON)
 option(WITH_GLEW_ES             "Switches to experimental copy of GLEW that has support for OpenGL ES. (temporary option for development purposes)" OFF)
+option(WITH_GL_EGL              "Use the EGL OpenGL system library instead of the platform specific OpenGL system library (CGL, glX, or WGL)"       OFF)
 option(WITH_GL_PROFILE_ES20     "Support using OpenGL ES 2.0. (through either EGL or the AGL/WGL/XGL 'es20' profile)"                               OFF)
 
 mark_as_advanced(
   WITH_OPENGL
   WITH_GLEW_ES
+  WITH_GL_EGL
   WITH_GL_PROFILE_ES20
 )
 
@@ -1712,18 +1705,22 @@ if(WITH_PYTHON)
   endif()
 endif()
 
-# Select C++17 as the standard for C++ projects.
-set(CMAKE_CXX_STANDARD 17)
-# If C++17 is not available, downgrading to an earlier standard is NOT OK.
-set(CMAKE_CXX_STANDARD_REQUIRED ON)
-# Do not enable compiler specific language extentions.
-set(CMAKE_CXX_EXTENSIONS OFF)
-
-# Make MSVC properly report the value of the __cplusplus preprocessor macro
-# Available MSVC 15.7 (1914) and up, without this it reports 199711L regardless
-# of the C++ standard chosen above.
-if(MSVC AND MSVC_VERSION GREATER 1913)
-  string(APPEND CMAKE_CXX_FLAGS " /Zc:__cplusplus")
+if(MSVC)
+  string(APPEND CMAKE_CXX_FLAGS " /std:c++17")
+  # Make MSVC properly report the value of the __cplusplus preprocessor macro
+  # Available MSVC 15.7 (1914) and up, without this it reports 199711L regardless
+  # of the C++ standard chosen above
+  if(MSVC_VERSION GREATER 1913)
+    string(APPEND CMAKE_CXX_FLAGS " /Zc:__cplusplus")
+  endif()
+elseif(
+  CMAKE_COMPILER_IS_GNUCC OR
+  CMAKE_C_COMPILER_ID MATCHES "Clang" OR
+  CMAKE_C_COMPILER_ID MATCHES "Intel"
+)
+  string(APPEND CMAKE_CXX_FLAGS " -std=c++17")
+else()
+  message(FATAL_ERROR "Unknown compiler ${CMAKE_C_COMPILER_ID}, can't enable C++17 build")
 endif()
 
 # Visual Studio has all standards it supports available by default
@@ -1918,7 +1915,6 @@ if(FIRST_RUN)
   info_cfg_option(WITH_IK_ITASC)
   info_cfg_option(WITH_IK_SOLVER)
   info_cfg_option(WITH_INPUT_NDOF)
-  info_cfg_option(WITH_INPUT_IME)
   info_cfg_option(WITH_INTERNATIONAL)
   info_cfg_option(WITH_OPENCOLLADA)
   info_cfg_option(WITH_OPENCOLORIO)

build_files/build_environment/CMakeLists.txt

@@ -56,7 +56,6 @@ else()
 endif()
 
 include(cmake/zlib.cmake)
-include(cmake/zstd.cmake)
 include(cmake/openal.cmake)
 include(cmake/png.cmake)
 include(cmake/jpeg.cmake)
@@ -82,11 +81,7 @@ if(UNIX)
 endif()
 include(cmake/openimageio.cmake)
 include(cmake/tiff.cmake)
-if(WIN32)
-  include(cmake/flexbison.cmake)
-else()
-  include(cmake/flex.cmake)
-endif()
+include(cmake/flexbison.cmake)
 include(cmake/osl.cmake)
 include(cmake/tbb.cmake)
 include(cmake/openvdb.cmake)
@@ -169,7 +164,6 @@ endif()
 if(UNIX AND NOT APPLE)
   include(cmake/libglu.cmake)
   include(cmake/mesa.cmake)
-  include(cmake/wayland_protocols.cmake)
 endif()
 
 include(cmake/harvest.cmake)

build_files/build_environment/cmake/download.cmake

@@ -87,10 +87,7 @@ download_source(LIBGLU)
 download_source(MESA)
 download_source(NASM)
 download_source(XR_OPENXR_SDK)
-download_source(WL_PROTOCOLS)
 download_source(ISPC)
 download_source(GMP)
 download_source(POTRACE)
 download_source(HARU)
-download_source(ZSTD)
-download_source(FLEX)

build_files/build_environment/cmake/ffmpeg.cmake

@@ -30,7 +30,6 @@ if(WIN32)
     --enable-w32threads
     --disable-pthreads
     --enable-libopenjpeg
-    --disable-mediafoundation
   )
   if("${CMAKE_SIZEOF_VOID_P}" EQUAL "4")
     set(FFMPEG_EXTRA_FLAGS

build_files/build_environment/cmake/flex.cmake

@@ -1,28 +0,0 @@
-# ***** BEGIN GPL LICENSE BLOCK *****
-#
-# This program is free software; you can redistribute it and/or
-# modify it under the terms of the GNU General Public License
-# as published by the Free Software Foundation; either version 2
-# of the License, or (at your option) any later version.
-#
-# This program is distributed in the hope that it will be useful,
-# but WITHOUT ANY WARRANTY; without even the implied warranty of
-# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-# GNU General Public License for more details.
-#
-# You should have received a copy of the GNU General Public License
-# along with this program; if not, write to the Free Software Foundation,
-# Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
-#
-# ***** END GPL LICENSE BLOCK *****
-
-ExternalProject_Add(external_flex
-  URL file://${PACKAGE_DIR}/${FLEX_FILE}
-  URL_HASH ${FLEX_HASH_TYPE}=${FLEX_HASH}
-  DOWNLOAD_DIR ${DOWNLOAD_DIR}
-  PREFIX ${BUILD_DIR}/flex
-  CONFIGURE_COMMAND ${CONFIGURE_ENV} && cd ${BUILD_DIR}/flex/src/external_flex/ && ${CONFIGURE_COMMAND} --prefix=${LIBDIR}/flex
-  BUILD_COMMAND ${CONFIGURE_ENV} && cd ${BUILD_DIR}/flex/src/external_flex/ && make -j${MAKE_THREADS}
-  INSTALL_COMMAND ${CONFIGURE_ENV} && cd ${BUILD_DIR}/flex/src/external_flex/ && make install
-  INSTALL_DIR ${LIBDIR}/flex
-)

build_files/build_environment/cmake/harvest.cmake

@@ -106,7 +106,6 @@ harvest(llvm/include llvm/include "*")
 harvest(llvm/bin llvm/bin "llvm-config")
 harvest(llvm/lib llvm/lib "libLLVM*.a")
 harvest(llvm/lib llvm/lib "libclang*.a")
-harvest(llvm/lib/clang llvm/lib/clang "*.h")
 if(APPLE)
   harvest(openmp/lib openmp/lib "*")
   harvest(openmp/include openmp/include "*.h")
@@ -127,8 +126,6 @@ if(UNIX AND NOT APPLE)
 
   harvest(xml2/include xml2/include "*.h")
   harvest(xml2/lib xml2/lib "*.a")
-
-  harvest(wayland-protocols/share/wayland-protocols wayland-protocols/share/wayland-protocols/ "*.xml")
 else()
   harvest(blosc/lib openvdb/lib "*.a")
   harvest(xml2/lib opencollada/lib "*.a")
@@ -193,8 +190,6 @@ harvest(potrace/include potrace/include "*.h")
 harvest(potrace/lib potrace/lib "*.a")
 harvest(haru/include haru/include "*.h")
 harvest(haru/lib haru/lib "*.a")
-harvest(zstd/include zstd/include "*.h")
-harvest(zstd/lib zstd/lib "*.a")
 
 if(UNIX AND NOT APPLE)
   harvest(libglu/lib mesa/lib "*.so*")

build_files/build_environment/cmake/ispc.cmake

@@ -29,13 +29,12 @@ elseif(APPLE)
   if("${CMAKE_HOST_SYSTEM_PROCESSOR}" STREQUAL "arm64")
     set(ISPC_EXTRA_ARGS_APPLE
       -DBISON_EXECUTABLE=/opt/homebrew/opt/bison/bin/bison
-      -DFLEX_EXECUTABLE=${LIBDIR}/flex/bin/flex
+      -DFLEX_EXECUTABLE=/opt/homebrew/opt/flex/bin/flex
       -DARM_ENABLED=On
     )
   else()
     set(ISPC_EXTRA_ARGS_APPLE
       -DBISON_EXECUTABLE=/usr/local/opt/bison/bin/bison
-      -DFLEX_EXECUTABLE=${LIBDIR}/flex/bin/flex
       -DARM_ENABLED=Off
     )
   endif()
@@ -44,7 +43,6 @@ elseif(UNIX)
     -DCMAKE_C_COMPILER=${LIBDIR}/llvm/bin/clang
     -DCMAKE_CXX_COMPILER=${LIBDIR}/llvm/bin/clang++
     -DARM_ENABLED=Off
-    -DFLEX_EXECUTABLE=${LIBDIR}/flex/bin/flex
   )
 endif()
 
@@ -84,9 +82,4 @@ if(WIN32)
     external_ispc
     external_flexbison
   )
-else()
-  add_dependencies(
-    external_ispc
-    external_flex
-  )
 endif()

build_files/build_environment/cmake/llvm.cmake

@@ -66,11 +66,7 @@ ExternalProject_Add(ll
 
 if(MSVC)
   if(BUILD_MODE STREQUAL Release)
-    set(LLVM_HARVEST_COMMAND
-      ${CMAKE_COMMAND} -E copy_directory ${LIBDIR}/llvm/lib ${HARVEST_TARGET}/llvm/lib &&
-      ${CMAKE_COMMAND} -E copy_directory ${LIBDIR}/llvm/include ${HARVEST_TARGET}/llvm/include &&
-      ${CMAKE_COMMAND} -E copy ${LIBDIR}/llvm/bin/clang-format.exe ${HARVEST_TARGET}/llvm/bin/clang-format.exe
-    )
+    set(LLVM_HARVEST_COMMAND ${CMAKE_COMMAND} -E copy_directory ${LIBDIR}/llvm/ ${HARVEST_TARGET}/llvm/ )
   else()
     set(LLVM_HARVEST_COMMAND
       ${CMAKE_COMMAND} -E copy_directory ${LIBDIR}/llvm/lib/ ${HARVEST_TARGET}/llvm/debug/lib/ &&

build_files/build_environment/cmake/openimagedenoise.cmake

@@ -45,6 +45,7 @@ ExternalProject_Add(external_openimagedenoise
   DOWNLOAD_DIR ${DOWNLOAD_DIR}
   URL_HASH ${OIDN_HASH_TYPE}=${OIDN_HASH}
   PREFIX ${BUILD_DIR}/openimagedenoise
+  PATCH_COMMAND ${PATCH_CMD} -p 1 -N -d ${BUILD_DIR}/openimagedenoise/src/external_openimagedenoise < ${PATCH_DIR}/oidn.diff
   CMAKE_ARGS -DCMAKE_INSTALL_PREFIX=${LIBDIR}/openimagedenoise ${DEFAULT_CMAKE_FLAGS} ${OIDN_EXTRA_ARGS}
   INSTALL_DIR ${LIBDIR}/openimagedenoise
 )

build_files/build_environment/cmake/openmp.cmake

@@ -22,8 +22,9 @@ ExternalProject_Add(external_openmp
   DOWNLOAD_DIR ${DOWNLOAD_DIR}
   URL_HASH ${OPENMP_HASH_TYPE}=${OPENMP_HASH}
   PREFIX ${BUILD_DIR}/openmp
+  PATCH_COMMAND ${PATCH_CMD} -p 1 -d ${BUILD_DIR}/openmp/src/external_openmp < ${PATCH_DIR}/openmp.diff
   CMAKE_ARGS -DCMAKE_INSTALL_PREFIX=${LIBDIR}/openmp ${DEFAULT_CMAKE_FLAGS}
-  INSTALL_COMMAND cd ${BUILD_DIR}/openmp/src/external_openmp-build && install_name_tool -id @rpath/libomp.dylib runtime/src/libomp.dylib && make install
+  INSTALL_COMMAND cd ${BUILD_DIR}/openmp/src/external_openmp-build && install_name_tool -id @executable_path/../Resources/lib/libomp.dylib runtime/src/libomp.dylib && make install
   INSTALL_DIR ${LIBDIR}/openmp
 )
 

build_files/build_environment/cmake/osl.cmake

@@ -20,10 +20,12 @@ if(WIN32)
   set(OSL_CMAKE_CXX_STANDARD_LIBRARIES "kernel32${LIBEXT} user32${LIBEXT} gdi32${LIBEXT} winspool${LIBEXT} shell32${LIBEXT} ole32${LIBEXT} oleaut32${LIBEXT} uuid${LIBEXT} comdlg32${LIBEXT} advapi32${LIBEXT} psapi${LIBEXT}")
   set(OSL_FLEX_BISON -DFLEX_EXECUTABLE=${LIBDIR}/flexbison/win_flex.exe -DBISON_EXECUTABLE=${LIBDIR}/flexbison/win_bison.exe)
   set(OSL_SIMD_FLAGS -DOIIO_NOSIMD=1 -DOIIO_SIMD=sse2)
+  SET(OSL_PLATFORM_FLAGS -DLINKSTATIC=ON)
 else()
   set(OSL_CMAKE_CXX_STANDARD_LIBRARIES)
   set(OSL_FLEX_BISON)
   set(OSL_OPENIMAGEIO_LIBRARY "${LIBDIR}/openimageio/lib/${LIBPREFIX}OpenImageIO${LIBEXT};${LIBDIR}/openimageio/lib/${LIBPREFIX}OpenImageIO_Util${LIBEXT};${LIBDIR}/png/lib/${LIBPREFIX}png16${LIBEXT};${LIBDIR}/jpg/lib/${LIBPREFIX}jpeg${LIBEXT};${LIBDIR}/tiff/lib/${LIBPREFIX}tiff${LIBEXT};${LIBDIR}/openexr/lib/${LIBPREFIX}IlmImf${OPENEXR_VERSION_POSTFIX}${LIBEXT}")
+  SET(OSL_PLATFORM_FLAGS)
 endif()
 
 set(OSL_ILMBASE_CUSTOM_LIBRARIES "${LIBDIR}/openexr/lib/Imath${OPENEXR_VERSION_POSTFIX}.lib^^${LIBDIR}/openexr/lib/Half{OPENEXR_VERSION_POSTFIX}.lib^^${LIBDIR}/openexr/lib/IlmThread${OPENEXR_VERSION_POSTFIX}.lib^^${LIBDIR}/openexr/lib/Iex${OPENEXR_VERSION_POSTFIX}.lib")
@@ -49,13 +51,12 @@ set(OSL_EXTRA_ARGS
   -DOpenImageIO_ROOT=${LIBDIR}/openimageio/
   -DOSL_BUILD_TESTS=OFF
   -DOSL_BUILD_MATERIALX=OFF
-  -DPNG_ROOT=${LIBDIR}/png
   -DZLIB_LIBRARY=${LIBDIR}/zlib/lib/${ZLIB_LIBRARY}
   -DZLIB_INCLUDE_DIR=${LIBDIR}/zlib/include/
   ${OSL_FLEX_BISON}
   -DCMAKE_CXX_STANDARD_LIBRARIES=${OSL_CMAKE_CXX_STANDARD_LIBRARIES}
   -DBUILD_SHARED_LIBS=OFF
-  -DLINKSTATIC=ON
+  ${OSL_PLATFORM_FLAGS}
   -DOSL_BUILD_PLUGINS=OFF
   -DSTOP_ON_WARNING=OFF
   -DUSE_LLVM_BITCODE=OFF
@@ -68,9 +69,13 @@ set(OSL_EXTRA_ARGS
   ${OSL_SIMD_FLAGS}
   -Dpugixml_ROOT=${LIBDIR}/pugixml
   -DUSE_PYTHON=OFF
-  -DCMAKE_CXX_STANDARD=14
 )
 
+# Apple arm64 uses LLVM 11, LLVM 10+ requires C++14
+if (APPLE AND "${CMAKE_OSX_ARCHITECTURES}" STREQUAL "arm64")
+  list(APPEND OSL_EXTRA_ARGS -DCMAKE_CXX_STANDARD=14)
+endif()
+
 ExternalProject_Add(external_osl
   URL file://${PACKAGE_DIR}/${OSL_FILE}
   DOWNLOAD_DIR ${DOWNLOAD_DIR}
@@ -88,20 +93,10 @@ add_dependencies(
   ll
   external_openexr
   external_zlib
+  external_flexbison
   external_openimageio
   external_pugixml
 )
-if(WIN32)
-  add_dependencies(
-    external_osl
-    external_flexbison
-  )
-else()
-  add_dependencies(
-    external_osl
-    external_flex
-  )
-endif()
 
 if(WIN32)
   if(BUILD_MODE STREQUAL Release)

build_files/build_environment/cmake/versions.cmake

@@ -152,20 +152,35 @@ set(OPENCOLORIO_HASH 1a2e3478b6cd9a1549f24e1b2205e3f0)
 set(OPENCOLORIO_HASH_TYPE MD5)
 set(OPENCOLORIO_FILE OpenColorIO-${OPENCOLORIO_VERSION}.tar.gz)
 
-set(LLVM_VERSION 12.0.0)
-set(LLVM_URI https://github.com/llvm/llvm-project/releases/download/llvmorg-${LLVM_VERSION}/llvm-project-${LLVM_VERSION}.src.tar.xz)
-set(LLVM_HASH 5a4fab4d7fc84aefffb118ac2c8a4fc0)
-set(LLVM_HASH_TYPE MD5)
-set(LLVM_FILE llvm-project-${LLVM_VERSION}.src.tar.xz)
+if(BLENDER_PLATFORM_ARM)
+  # Newer version required by ISPC with arm support.
+  set(LLVM_VERSION 11.0.1)
+  set(LLVM_URI https://github.com/llvm/llvm-project/releases/download/llvmorg-${LLVM_VERSION}/llvm-project-${LLVM_VERSION}.src.tar.xz)
+  set(LLVM_HASH e700af40ab83463e4e9ab0ba3708312e)
+  set(LLVM_HASH_TYPE MD5)
+  set(LLVM_FILE llvm-project-${LLVM_VERSION}.src.tar.xz)
 
-set(OPENMP_URI https://github.com/llvm/llvm-project/releases/download/llvmorg-${LLVM_VERSION}/openmp-${LLVM_VERSION}.src.tar.xz)
-set(OPENMP_HASH ac48ce3e4582ccb82f81ab59eb3fc9dc)
-set(OPENMP_HASH_TYPE MD5)
-set(OPENMP_FILE openmp-${LLVM_VERSION}.src.tar.xz)
+  set(OPENMP_VERSION 9.0.1)
+  set(OPENMP_URI https://github.com/llvm/llvm-project/releases/download/llvmorg-${OPENMP_VERSION}/openmp-${OPENMP_VERSION}.src.tar.xz)
+  set(OPENMP_HASH 6eade16057edbdecb3c4eef9daa2bfcf)
+  set(OPENMP_HASH_TYPE MD5)
+  set(OPENMP_FILE openmp-${OPENMP_VERSION}.src.tar.xz)
+else()
+  set(LLVM_VERSION 9.0.1)
+  set(LLVM_URI https://github.com/llvm/llvm-project/releases/download/llvmorg-${LLVM_VERSION}/llvm-project-${LLVM_VERSION}.tar.xz)
+  set(LLVM_HASH b4268e733dfe352960140dc07ef2efcb)
+  set(LLVM_HASH_TYPE MD5)
+  set(LLVM_FILE llvm-project-${LLVM_VERSION}.tar.xz)
 
-set(OPENIMAGEIO_VERSION 2.2.15.1)
+  set(OPENMP_URI https://github.com/llvm/llvm-project/releases/download/llvmorg-${LLVM_VERSION}/openmp-${LLVM_VERSION}.src.tar.xz)
+  set(OPENMP_HASH 6eade16057edbdecb3c4eef9daa2bfcf)
+  set(OPENMP_HASH_TYPE MD5)
+  set(OPENMP_FILE openmp-${LLVM_VERSION}.src.tar.xz)
+endif()
+
+set(OPENIMAGEIO_VERSION 2.1.15.0)
 set(OPENIMAGEIO_URI https://github.com/OpenImageIO/oiio/archive/Release-${OPENIMAGEIO_VERSION}.tar.gz)
-set(OPENIMAGEIO_HASH 3db5c5f0b3dc91597c75e5df09eb9072)
+set(OPENIMAGEIO_HASH f03aa5e3ac4795af04771ee4146e9832)
 set(OPENIMAGEIO_HASH_TYPE MD5)
 set(OPENIMAGEIO_FILE OpenImageIO-${OPENIMAGEIO_VERSION}.tar.gz)
 
@@ -175,9 +190,9 @@ set(TIFF_HASH 2165e7aba557463acc0664e71a3ed424)
 set(TIFF_HASH_TYPE MD5)
 set(TIFF_FILE tiff-${TIFF_VERSION}.tar.gz)
 
-set(OSL_VERSION 1.11.14.1)
+set(OSL_VERSION 1.11.10.0)
 set(OSL_URI https://github.com/imageworks/OpenShadingLanguage/archive/Release-${OSL_VERSION}.tar.gz)
-set(OSL_HASH 1abd7ce40481771a9fa937f19595d2f2)
+set(OSL_HASH dfdc23597aeef083832cbada62211756)
 set(OSL_HASH_TYPE MD5)
 set(OSL_FILE OpenShadingLanguage-${OSL_VERSION}.tar.gz)
 
@@ -201,9 +216,9 @@ set(OPENVDB_HASH 01b490be16cc0e15c690f9a153c21461)
 set(OPENVDB_HASH_TYPE MD5)
 set(OPENVDB_FILE openvdb-${OPENVDB_VERSION}.tar.gz)
 
-set(NANOVDB_GIT_UID dc37d8a631922e7bef46712947dc19b755f3e841)
+set(NANOVDB_GIT_UID e62f7a0bf1e27397223c61ddeaaf57edf111b77f)
 set(NANOVDB_URI https://github.com/AcademySoftwareFoundation/openvdb/archive/${NANOVDB_GIT_UID}.tar.gz)
-set(NANOVDB_HASH e7b9e863ec2f3b04ead171dec2322807)
+set(NANOVDB_HASH 90919510bc6ccd630fedc56f748cb199)
 set(NANOVDB_HASH_TYPE MD5)
 set(NANOVDB_FILE nano-vdb-${NANOVDB_GIT_UID}.tar.gz)
 
@@ -355,18 +370,12 @@ set(PUGIXML_HASH 0c208b0664c7fb822bf1b49ad035e8fd)
 set(PUGIXML_HASH_TYPE MD5)
 set(PUGIXML_FILE pugixml-${PUGIXML_VERSION}.tar.gz)
 
-set(FLEXBISON_VERSION 2.5.24)
+set(FLEXBISON_VERSION 2.5.5)
 set(FLEXBISON_URI http://prdownloads.sourceforge.net/winflexbison/win_flex_bison-${FLEXBISON_VERSION}.zip)
-set(FLEXBISON_HASH 6b549d43e34ece0e8ed05af92daa31c4)
+set(FLEXBISON_HASH d87a3938194520d904013abef3df10ce)
 set(FLEXBISON_HASH_TYPE MD5)
 set(FLEXBISON_FILE win_flex_bison-${FLEXBISON_VERSION}.zip)
 
-set(FLEX_VERSION 2.6.4)
-set(FLEX_URI https://github.com/westes/flex/releases/download/v${FLEX_VERSION}/flex-${FLEX_VERSION}.tar.gz)
-set(FLEX_HASH 2882e3179748cc9f9c23ec593d6adc8d)
-set(FLEX_HASH_TYPE MD5)
-set(FLEX_FILE flex-${FLEX_VERSION}.tar.gz)
-
 # Libraries to keep Python modules static on Linux.
 
 # NOTE: bzip.org domain does no longer belong to BZip 2 project, so we download
@@ -423,9 +432,9 @@ set(USD_HASH 1dd1e2092d085ed393c1f7c450a4155a)
 set(USD_HASH_TYPE MD5)
 set(USD_FILE usd-v${USD_VERSION}.tar.gz)
 
-set(OIDN_VERSION 1.4.1)
+set(OIDN_VERSION 1.4.0)
 set(OIDN_URI https://github.com/OpenImageDenoise/oidn/releases/download/v${OIDN_VERSION}/oidn-${OIDN_VERSION}.src.tar.gz)
-set(OIDN_HASH df4007b0ab93b1c41cdf223b075d01c0)
+set(OIDN_HASH 421824019becc5b664a22a2b98332bc5)
 set(OIDN_HASH_TYPE MD5)
 set(OIDN_FILE oidn-${OIDN_VERSION}.src.tar.gz)
 
@@ -435,10 +444,10 @@ set(LIBGLU_HASH 151aef599b8259efe9acd599c96ea2a3)
 set(LIBGLU_HASH_TYPE MD5)
 set(LIBGLU_FILE glu-${LIBGLU_VERSION}.tar.xz)
 
-set(MESA_VERSION 21.1.5)
+set(MESA_VERSION 20.3.4)
 set(MESA_URI ftp://ftp.freedesktop.org/pub/mesa/mesa-${MESA_VERSION}.tar.xz)
-set(MESA_HASH 022c7293074aeeced2278c872db4fa693147c70f8595b076cf3f1ef81520766d)
-set(MESA_HASH_TYPE SHA256)
+set(MESA_HASH 556338446aef8ae947a789b3e0b5e056)
+set(MESA_HASH_TYPE MD5)
 set(MESA_FILE mesa-${MESA_VERSION}.tar.xz)
 
 set(NASM_VERSION 2.15.02)
@@ -447,23 +456,25 @@ set(NASM_HASH aded8b796c996a486a56e0515c83e414116decc3b184d88043480b32eb0a8589)
 set(NASM_HASH_TYPE SHA256)
 set(NASM_FILE nasm-${NASM_VERSION}.tar.gz)
 
-set(XR_OPENXR_SDK_VERSION 1.0.17)
+set(XR_OPENXR_SDK_VERSION 1.0.14)
 set(XR_OPENXR_SDK_URI https://github.com/KhronosGroup/OpenXR-SDK/archive/release-${XR_OPENXR_SDK_VERSION}.tar.gz)
-set(XR_OPENXR_SDK_HASH bf0fd8828837edff01047474e90013e1)
+set(XR_OPENXR_SDK_HASH 0df6b2fd6045423451a77ff6bc3e1a75)
 set(XR_OPENXR_SDK_HASH_TYPE MD5)
 set(XR_OPENXR_SDK_FILE OpenXR-SDK-${XR_OPENXR_SDK_VERSION}.tar.gz)
 
-set(WL_PROTOCOLS_VERSION 1.21)
-set(WL_PROTOCOLS_FILE wayland-protocols-${WL_PROTOCOLS_VERSION}.tar.gz)
-set(WL_PROTOCOLS_URI https://gitlab.freedesktop.org/wayland/wayland-protocols/-/archive/${WL_PROTOCOLS_VERSION}/${WL_PROTOCOLS_FILE})
-set(WL_PROTOCOLS_HASH af5ca07e13517cdbab33504492cef54a)
-set(WL_PROTOCOLS_HASH_TYPE MD5)
-
-set(ISPC_VERSION v1.16.0)
-set(ISPC_URI https://github.com/ispc/ispc/archive/${ISPC_VERSION}.tar.gz)
-set(ISPC_HASH 2e3abedbc0ea9aaec17d6562c632454d)
-set(ISPC_HASH_TYPE MD5)
-set(ISPC_FILE ispc-${ISPC_VERSION}.tar.gz)
+if(BLENDER_PLATFORM_ARM)
+  # Unreleased version with macOS arm support.
+  set(ISPC_URI https://github.com/ispc/ispc/archive/f5949c055eb9eeb93696978a3da4bfb3a6a30b35.zip)
+  set(ISPC_HASH d382fea18d01dbd0cd05d9e1ede36d7d)
+  set(ISPC_HASH_TYPE MD5)
+  set(ISPC_FILE f5949c055eb9eeb93696978a3da4bfb3a6a30b35.zip)
+else()
+  set(ISPC_VERSION v1.14.1)
+  set(ISPC_URI https://github.com/ispc/ispc/archive/${ISPC_VERSION}.tar.gz)
+  set(ISPC_HASH 968fbc8dfd16a60ba4e32d2e0e03ea7a)
+  set(ISPC_HASH_TYPE MD5)
+  set(ISPC_FILE ispc-${ISPC_VERSION}.tar.gz)
+endif()
 
 set(GMP_VERSION 6.2.0)
 set(GMP_URI https://gmplib.org/download/gmp/gmp-${GMP_VERSION}.tar.xz)
@@ -483,11 +494,5 @@ set(HARU_HASH 4f916aa49c3069b3a10850013c507460)
 set(HARU_HASH_TYPE MD5)
 set(HARU_FILE libharu-${HARU_VERSION}.tar.gz)
 
-set(ZSTD_VERSION 1.5.0)
-set(ZSTD_URI https://github.com/facebook/zstd/releases/download/v${ZSTD_VERSION}/zstd-${ZSTD_VERSION}.tar.gz)
-set(ZSTD_HASH 5194fbfa781fcf45b98c5e849651aa7b3b0a008c6b72d4a0db760f3002291e94)
-set(ZSTD_HASH_TYPE SHA256)
-set(ZSTD_FILE zstd-${ZSTD_VERSION}.tar.gz)
-
 set(SSE2NEON_GIT https://github.com/DLTcollab/sse2neon.git)
 set(SSE2NEON_GIT_HASH fe5ff00bb8d19b327714a3c290f3e2ce81ba3525)

build_files/build_environment/cmake/wayland_protocols.cmake

@@ -1,27 +0,0 @@
-# ***** BEGIN GPL LICENSE BLOCK *****
-#
-# This program is free software; you can redistribute it and/or
-# modify it under the terms of the GNU General Public License
-# as published by the Free Software Foundation; either version 2
-# of the License, or (at your option) any later version.
-#
-# This program is distributed in the hope that it will be useful,
-# but WITHOUT ANY WARRANTY; without even the implied warranty of
-# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-# GNU General Public License for more details.
-#
-# You should have received a copy of the GNU General Public License
-# along with this program; if not, write to the Free Software Foundation,
-# Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
-#
-# ***** END GPL LICENSE BLOCK *****
-
-ExternalProject_Add(external_wayland_protocols
-  URL file://${PACKAGE_DIR}/${WL_PROTOCOLS_FILE}
-  DOWNLOAD_DIR ${DOWNLOAD_DIR}
-  URL_HASH ${WL_PROTOCOLS_HASH_TYPE}=${WL_PROTOCOLS_HASH}
-  PREFIX ${BUILD_DIR}/wayland-protocols
-  CONFIGURE_COMMAND meson --prefix ${LIBDIR}/wayland-protocols . ../external_wayland_protocols -Dtests=false
-  BUILD_COMMAND ninja
-  INSTALL_COMMAND ninja install
-)

build_files/build_environment/cmake/zstd.cmake

@@ -1,51 +0,0 @@
-# ***** BEGIN GPL LICENSE BLOCK *****
-#
-# This program is free software; you can redistribute it and/or
-# modify it under the terms of the GNU General Public License
-# as published by the Free Software Foundation; either version 2
-# of the License, or (at your option) any later version.
-#
-# This program is distributed in the hope that it will be useful,
-# but WITHOUT ANY WARRANTY; without even the implied warranty of
-# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-# GNU General Public License for more details.
-#
-# You should have received a copy of the GNU General Public License
-# along with this program; if not, write to the Free Software Foundation,
-# Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
-#
-# ***** END GPL LICENSE BLOCK *****
-
-set(ZSTD_EXTRA_ARGS
-  -DZSTD_BUILD_PROGRAMS=OFF
-  -DZSTD_BUILD_SHARED=OFF
-  -DZSTD_BUILD_STATIC=ON
-  -DZSTD_BUILD_TESTS=OFF
-  -DZSTD_LEGACY_SUPPORT=OFF
-  -DZSTD_LZ4_SUPPORT=OFF
-  -DZSTD_LZMA_SUPPORT=OFF
-  -DZSTD_MULTITHREAD_SUPPORT=ON
-  -DZSTD_PROGRAMS_LINK_SHARED=OFF
-  -DZSTD_USE_STATIC_RUNTIME=OFF
-  -DZSTD_ZLIB_SUPPORT=OFF
-)
-
-ExternalProject_Add(external_zstd
-  URL file://${PACKAGE_DIR}/${ZSTD_FILE}
-  DOWNLOAD_DIR ${DOWNLOAD_DIR}
-  URL_HASH ${ZSTD_HASH_TYPE}=${ZSTD_HASH}
-  PREFIX ${BUILD_DIR}/zstd
-  SOURCE_SUBDIR build/cmake
-  CMAKE_ARGS -DCMAKE_INSTALL_PREFIX=${LIBDIR}/zstd ${DEFAULT_CMAKE_FLAGS} ${ZSTD_EXTRA_ARGS}
-  INSTALL_DIR ${LIBDIR}/zstd
-)
-
-if(WIN32)
-  if(BUILD_MODE STREQUAL Release)
-    ExternalProject_Add_Step(external_zstd after_install
-      COMMAND ${CMAKE_COMMAND} -E copy ${LIBDIR}/zstd/lib/zstd_static${LIBEXT} ${HARVEST_TARGET}/zstd/lib/zstd_static${LIBEXT}
-      COMMAND ${CMAKE_COMMAND} -E copy_directory ${LIBDIR}/zstd/include/ ${HARVEST_TARGET}/zstd/include/
-      DEPENDEES install
-    )
-  endif()
-endif()

build_files/build_environment/install_deps.sh

@@ -474,25 +474,25 @@ OPENEXR_FORCE_REBUILD=false
 OPENEXR_SKIP=false
 _with_built_openexr=false
 
-OIIO_VERSION="2.2.15.1"
-OIIO_VERSION_SHORT="2.2"
+OIIO_VERSION="2.1.15.0"
+OIIO_VERSION_SHORT="2.1"
 OIIO_VERSION_MIN="2.1.12"
-OIIO_VERSION_MAX="2.3.0"
+OIIO_VERSION_MAX="2.2.10"
 OIIO_FORCE_BUILD=false
 OIIO_FORCE_REBUILD=false
 OIIO_SKIP=false
 
-LLVM_VERSION="12.0.0"
-LLVM_VERSION_SHORT="12.0"
-LLVM_VERSION_MIN="11.0"
-LLVM_VERSION_MAX="13.0"
+LLVM_VERSION="9.0.1"
+LLVM_VERSION_SHORT="9.0"
+LLVM_VERSION_MIN="6.0"
+LLVM_VERSION_MAX="12.0"
 LLVM_VERSION_FOUND=""
 LLVM_FORCE_BUILD=false
 LLVM_FORCE_REBUILD=false
 LLVM_SKIP=false
 
 # OSL needs to be compiled for now!
-OSL_VERSION="1.11.14.1"
+OSL_VERSION="1.11.10.0"
 OSL_VERSION_SHORT="1.11"
 OSL_VERSION_MIN="1.11"
 OSL_VERSION_MAX="2.0"
@@ -553,7 +553,7 @@ EMBREE_FORCE_BUILD=false
 EMBREE_FORCE_REBUILD=false
 EMBREE_SKIP=false
 
-OIDN_VERSION="1.4.1"
+OIDN_VERSION="1.4.0"
 OIDN_VERSION_SHORT="1.4"
 OIDN_VERSION_MIN="1.4.0"
 OIDN_VERSION_MAX="1.5"
@@ -561,7 +561,7 @@ OIDN_FORCE_BUILD=false
 OIDN_FORCE_REBUILD=false
 OIDN_SKIP=false
 
-ISPC_VERSION="1.16.0"
+ISPC_VERSION="1.14.1"
 
 FFMPEG_VERSION="4.4"
 FFMPEG_VERSION_SHORT="4.4"
@@ -572,7 +572,7 @@ FFMPEG_FORCE_REBUILD=false
 FFMPEG_SKIP=false
 _ffmpeg_list_sep=";"
 
-XR_OPENXR_VERSION="1.0.17"
+XR_OPENXR_VERSION="1.0.14"
 XR_OPENXR_VERSION_SHORT="1.0"
 XR_OPENXR_VERSION_MIN="1.0.8"
 XR_OPENXR_VERSION_MAX="2.0"
@@ -603,6 +603,9 @@ MP3LAME_DEV=""
 OPENJPEG_USE=false
 OPENJPEG_DEV=""
 
+# Whether to use system GLEW or not (OpenSubDiv needs recent glew to work).
+NO_SYSTEM_GLEW=false
+
 # Switch to english language, else some things (like check_package_DEB()) won't work!
 LANG_BACK=$LANG
 LANG=""
@@ -1070,7 +1073,7 @@ OPENVDB_SOURCE=( "https://github.com/AcademySoftwareFoundation/openvdb/archive/v
 #~ OPENVDB_SOURCE_REPO_BRANCH="dev"
 
 NANOVDB_USE_REPO=false
-NANOVDB_SOURCE_REPO_UID="dc37d8a631922e7bef46712947dc19b755f3e841"
+NANOVDB_SOURCE_REPO_UID="e62f7a0bf1e27397223c61ddeaaf57edf111b77f"
 NANOVDB_SOURCE=( "https://github.com/AcademySoftwareFoundation/openvdb/archive/${NANOVDB_SOURCE_REPO_UID}.tar.gz" )
 
 ALEMBIC_USE_REPO=false
@@ -1105,9 +1108,9 @@ FFMPEG_SOURCE=( "http://ffmpeg.org/releases/ffmpeg-$FFMPEG_VERSION.tar.bz2" )
 
 XR_OPENXR_USE_REPO=false
 XR_OPENXR_SOURCE=("https://github.com/KhronosGroup/OpenXR-SDK/archive/release-${XR_OPENXR_VERSION}.tar.gz")
-XR_OPENXR_SOURCE_REPO=("https://github.com/KhronosGroup/OpenXR-SDK.git")
-XR_OPENXR_REPO_UID="bf21ccb1007bb531b45d9978919a56ea5059c245"
-XR_OPENXR_REPO_BRANCH="master"
+#~ XR_OPENXR_SOURCE_REPO=("https://github.com/KhronosGroup/OpenXR-SDK.git")
+#~ XR_OPENXR_REPO_UID="5900c51562769b03bea699dc0352cae56acb6419d"
+#~ XR_OPENXR_REPO_BRANCH="master"
 
 # C++11 is required now
 CXXFLAGS_BACK=$CXXFLAGS
@@ -1125,8 +1128,7 @@ Those libraries should be available as packages in all recent distributions (opt
     * Basics of dev environment (cmake, gcc, svn , git, ...).
     * libjpeg, libpng, libtiff, [openjpeg2], [libopenal].
     * libx11, libxcursor, libxi, libxrandr, libxinerama (and other libx... as needed).
-    * libwayland-client0, libwayland-cursor0, libwayland-egl1, libxkbcommon0, libdbus-1-3, libegl1 (Wayland)
-    * libsqlite3, libzstd, libbz2, libssl, libfftw3, libxml2, libtinyxml, yasm, libyaml-cpp, flex.
+    * libsqlite3, libbz2, libssl, libfftw3, libxml2, libtinyxml, yasm, libyaml-cpp.
     * libsdl2, libglew, libpugixml, libpotrace, [libgmp], [libglewmx], fontconfig, [libharu/libhpdf].\""
 
 DEPS_SPECIFIC_INFO="\"BUILDABLE DEPENDENCIES:
@@ -2313,7 +2315,6 @@ compile_OSL() {
         tar -C $SRC --transform "s,(.*/?)OpenShadingLanguage-[^/]*(.*),\1OpenShadingLanguage-$OSL_VERSION\2,x" \
             -xf $_src.tar.gz
       fi
-      patch -d $_src -p1 < $SCRIPT_DIR/patches/osl.diff
     fi
 
     cd $_src
@@ -2336,6 +2337,7 @@ compile_OSL() {
 
     cmake_d="-D CMAKE_BUILD_TYPE=Release"
     cmake_d="$cmake_d -D CMAKE_INSTALL_PREFIX=$_inst"
+    cmake_d="$cmake_d -D BUILD_TESTING=OFF"
     cmake_d="$cmake_d -D STOP_ON_WARNING=OFF"
     cmake_d="$cmake_d -D OSL_BUILD_PLUGINS=OFF"
     cmake_d="$cmake_d -D OSL_BUILD_TESTS=OFF"
@@ -2343,10 +2345,7 @@ compile_OSL() {
     cmake_d="$cmake_d -D USE_PARTIO=OFF"
     cmake_d="$cmake_d -D OSL_BUILD_MATERIALX=OFF"
     cmake_d="$cmake_d -D USE_QT=OFF"
-    cmake_d="$cmake_d -D USE_Qt5=OFF"
     cmake_d="$cmake_d -D USE_PYTHON=OFF"
-    cmake_d="$cmake_d -D USE_PARTIO=OFF"
-    cmake_d="$cmake_d -D INSTALL_DOCS=OFF"
 
     if [ $(uname -m) != "aarch64" ]; then
       cmake_d="$cmake_d -D USE_SIMD=sse2"
@@ -2368,7 +2367,10 @@ compile_OSL() {
     fi
 
     if [ -d $INST/oiio ]; then
-      cmake_d="$cmake_d -D OpenImageIO_ROOT=$INST/oiio"
+      cmake_d="$cmake_d -D OPENIMAGEIO_ROOT_DIR=$INST/oiio"
+      # HACK! SIC!!!!
+      # Quiet incredible, but if root dir is given, path to lib is found, but not path to include...
+      cmake_d="$cmake_d -D OPENIMAGEIO_INCLUDE_DIR=$INST/oiio/include"
     fi
 
     if [ ! -z $LLVM_VERSION_FOUND ]; then
@@ -2735,7 +2737,7 @@ _init_openvdb() {
   _git=false
   _inst=$INST/openvdb-$OPENVDB_VERSION_SHORT
   _inst_shortcut=$INST/openvdb
-
+  
   _openvdb_source=$OPENVDB_SOURCE
   if [ "$WITH_NANOVDB" = true ]; then
     _openvdb_source=$NANOVDB_SOURCE
@@ -2840,7 +2842,7 @@ compile_OPENVDB() {
     if [ -d $INST/blosc ]; then
       cmake_d="$cmake_d -D Blosc_ROOT=$INST/blosc"
     fi
-
+  
     cmake $cmake_d ..
 
     make -j$THREADS install
@@ -3837,11 +3839,10 @@ install_DEB() {
   _packages="gawk cmake cmake-curses-gui build-essential libjpeg-dev libpng-dev libtiff-dev \
              git libfreetype6-dev libfontconfig-dev libx11-dev flex bison libxxf86vm-dev \
              libxcursor-dev libxi-dev wget libsqlite3-dev libxrandr-dev libxinerama-dev \
-             libwayland-dev wayland-protocols libegl-dev libxkbcommon-dev libdbus-1-dev linux-libc-dev \
              libbz2-dev libncurses5-dev libssl-dev liblzma-dev libreadline-dev \
              libopenal-dev libglew-dev yasm $THEORA_DEV $VORBIS_DEV $OGG_DEV \
              libsdl2-dev libfftw3-dev patch bzip2 libxml2-dev libtinyxml-dev libjemalloc-dev \
-             libgmp-dev libpugixml-dev libpotrace-dev libhpdf-dev libzstd-dev"
+             libgmp-dev libpugixml-dev libpotrace-dev libhpdf-dev"
              # libglewmx-dev  (broken in deb testing currently...)
 
   VORBIS_USE=true
@@ -3982,9 +3983,13 @@ install_DEB() {
     version_ge $_glew "1.7.0"
     if [ $? -eq 1 ]; then
       WARNING "OpenSubdiv disabled because GLEW-$_glew is not enough"
+      WARNING "Blender will not use system GLEW library"
       OSD_SKIP=true
+      NO_SYSTEM_GLEW=true
     else
       WARNING "OpenSubdiv will compile with GLEW-$_glew but with limited capability"
+      WARNING "Blender will not use system GLEW library"
+      NO_SYSTEM_GLEW=true
     fi
   fi
 
@@ -4105,8 +4110,6 @@ install_DEB() {
 
 
   PRINT ""
-  # Debian OIIO includes again libopencv, without even properly dealing with this dependency...
-  OIIO_FORCE_BUILD=true
   if [ "$OIIO_SKIP" = true ]; then
     WARNING "Skipping OpenImageIO installation, as requested..."
   elif [ "$OIIO_FORCE_BUILD" = true ]; then
@@ -4505,11 +4508,10 @@ install_RPM() {
   _packages="gcc gcc-c++ git make cmake tar bzip2 xz findutils flex bison fontconfig-devel \
              libtiff-devel libjpeg-devel libpng-devel sqlite-devel fftw-devel SDL2-devel \
              libX11-devel libXi-devel libXcursor-devel libXrandr-devel libXinerama-devel \
-             wayland-devel wayland-protocols-devel mesa-libEGL-devel libxkbcommon-devel dbus-devel kernel-headers \
              wget ncurses-devel readline-devel $OPENJPEG_DEV openal-soft-devel \
              glew-devel yasm $THEORA_DEV $VORBIS_DEV $OGG_DEV patch \
              libxml2-devel yaml-cpp-devel tinyxml-devel jemalloc-devel \
-             gmp-devel pugixml-devel potrace-devel libharu-devel libzstd-devel"
+             gmp-devel pugixml-devel potrace-devel libharu-devel"
 
   OPENJPEG_USE=true
   VORBIS_USE=true
@@ -5087,11 +5089,10 @@ install_ARCH() {
     BASE_DEVEL=`pacman -Sgq base-devel | sed -e 's/^gcc$/gcc-multilib/g' | paste -s -d' '`
   fi
 
-  _packages="$BASE_DEVEL git cmake fontconfig flex \
+  _packages="$BASE_DEVEL git cmake fontconfig \
              libxi libxcursor libxrandr libxinerama glew libpng libtiff wget openal \
              $OPENJPEG_DEV $VORBIS_DEV $OGG_DEV $THEORA_DEV yasm sdl2 fftw \
-             libxml2 yaml-cpp tinyxml python-requests jemalloc gmp potrace pugixml libharu \
-             zstd"
+             libxml2 yaml-cpp tinyxml python-requests jemalloc gmp potrace pugixml libharu"
 
   OPENJPEG_USE=true
   VORBIS_USE=true
@@ -5955,6 +5956,12 @@ print_info() {
     fi
   fi
 
+  if [ "$NO_SYSTEM_GLEW" = true ]; then
+    _1="-D WITH_SYSTEM_GLEW=OFF"
+    PRINT "  $_1"
+    _buildargs="$_buildargs $_1"
+  fi
+
   if [ "$FFMPEG_SKIP" = false ]; then
     _1="-D WITH_CODEC_FFMPEG=ON"
     _2="-D FFMPEG_LIBRARIES='avformat;avcodec;avutil;avdevice;swscale;swresample;lzma;rt;`print_info_ffmpeglink`'"

build_files/build_environment/patches/ffmpeg.diff

@@ -68,32 +68,3 @@
 +
      return ret;
  }
---- a/libavcodec/rl.c
-+++ b/libavcodec/rl.c
-@@ -71,7 +71,7 @@ av_cold void ff_rl_init(RLTable *rl,
- av_cold void ff_rl_init_vlc(RLTable *rl, unsigned static_size)
- {
-     int i, q;
--    VLC_TYPE table[1500][2] = {{0}};
-+    VLC_TYPE (*table)[2] = av_calloc(sizeof(VLC_TYPE), 1500 * 2);
-     VLC vlc = { .table = table, .table_allocated = static_size };
-     av_assert0(static_size <= FF_ARRAY_ELEMS(table));
-     init_vlc(&vlc, 9, rl->n + 1, &rl->table_vlc[0][1], 4, 2, &rl->table_vlc[0][0], 4, 2, INIT_VLC_USE_NEW_STATIC);
-@@ -80,8 +80,10 @@ av_cold void ff_rl_init_vlc(RLTable *rl, unsigned static_size)
-         int qmul = q * 2;
-         int qadd = (q - 1) | 1;
- 
--        if (!rl->rl_vlc[q])
-+        if (!rl->rl_vlc[q]){
-+            av_free(table);
-             return;
-+        }
- 
-         if (q == 0) {
-             qmul = 1;
-@@ -113,4 +115,5 @@ av_cold void ff_rl_init_vlc(RLTable *rl, unsigned static_size)
-             rl->rl_vlc[q][i].run   = run;
-         }
-     }
-+    av_free(table);
- }

build_files/build_environment/patches/oidn.diff

@@ -0,0 +1,10 @@
+--- external_openimagedenoise/cmake/oidn_ispc.cmake	2021-02-15 17:29:34.000000000 +0100
++++ external_openimagedenoise/cmake/oidn_ispc.cmake2	2021-02-15 17:29:28.000000000 +0100
+@@ -98,7 +98,7 @@
+   elseif(OIDN_ARCH STREQUAL "ARM64")
+     set(ISPC_ARCHITECTURE "aarch64")
+     if(APPLE)
+-      set(ISPC_TARGET_OS "--target-os=ios")
++      set(ISPC_TARGET_OS "--target-os=macos")
+     endif()
+   endif()

build_files/build_environment/patches/openimageio.diff

@@ -34,3 +34,24 @@ diff -Naur orig/src/include/OpenImageIO/platform.h external_openimageio/src/incl
  #    include <windows.h>
  #endif
  
+diff -Naur orig/src/libutil/ustring.cpp external_openimageio/src/libutil/ustring.cpp
+--- orig/src/libutil/ustring.cpp	2020-05-11 05:43:52.000000000 +0200
++++ external_openimageio/src/libutil/ustring.cpp	2020-11-26 12:06:08.000000000 +0100
+@@ -337,6 +337,8 @@
+     // the std::string to make it point to our chars!  In such a case, the
+     // destructor will be careful not to allow a deallocation.
+ 
++    // Disable internal std::string for Apple silicon based Macs
++#if !(defined(__APPLE__) && defined(__arm64__))
+ #if defined(__GNUC__) && !defined(_LIBCPP_VERSION)                             \
+     && defined(_GLIBCXX_USE_CXX11_ABI) && _GLIBCXX_USE_CXX11_ABI
+     // NEW gcc ABI
+@@ -382,7 +384,7 @@
+         return;
+     }
+ #endif
+-
++#endif
+     // Remaining cases - just assign the internal string.  This may result
+     // in double allocation for the chars.  If you care about that, do
+     // something special for your platform, much like we did for gcc and

build_files/build_environment/patches/openmp.diff

@@ -0,0 +1,23 @@
+diff --git a/runtime/src/z_Linux_asm.S b/runtime/src/z_Linux_asm.S
+index 0d8885e..42aa5ad 100644
+--- a/runtime/src/z_Linux_asm.S
++++ b/runtime/src/z_Linux_asm.S
+@@ -1540,10 +1540,12 @@ __kmp_unnamed_critical_addr:
+     .comm .gomp_critical_user_,32,8
+     .data
+     .align 8
+-    .global __kmp_unnamed_critical_addr
+-__kmp_unnamed_critical_addr:
++    .global ___kmp_unnamed_critical_addr
++___kmp_unnamed_critical_addr:
+     .8byte .gomp_critical_user_
+-    .size __kmp_unnamed_critical_addr,8
++#  if !(KMP_OS_DARWIN)
++    .size ___kmp_unnamed_critical_addr,8
++#  endif
+ #endif /* KMP_ARCH_PPC64 || KMP_ARCH_AARCH64 */
+ 
+ #if KMP_OS_LINUX
+
+
+

build_files/build_environment/patches/osl.diff

@@ -1,3 +1,18 @@
+diff -Naur OpenShadingLanguage-Release-1.9.9/src/cmake/flexbison.cmake.rej external_osl/src/cmake/flexbison.cmake.rej
+--- OpenShadingLanguage-Release-1.9.9/src/cmake/flexbison.cmake.rej	1969-12-31 17:00:00 -0700
++++ external_osl/src/cmake/flexbison.cmake.rej	2018-08-24 17:42:11 -0600
+@@ -0,0 +1,11 @@
++--- src/cmake/flexbison.cmake	2018-05-01 16:39:02 -0600
+++++ src/cmake/flexbison.cmake	2018-08-24 10:24:03 -0600
++@@ -77,7 +77,7 @@
++           DEPENDS ${${compiler_headers}}
++           WORKING_DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR} )
++         ADD_CUSTOM_COMMAND ( OUTPUT ${flexoutputcxx} 
++-          COMMAND ${FLEX_EXECUTABLE} -o ${flexoutputcxx} "${CMAKE_CURRENT_SOURCE_DIR}/${flexsrc}"
+++          COMMAND ${FLEX_EXECUTABLE} ${FLEX_EXTRA_OPTIONS} -o ${flexoutputcxx} "${CMAKE_CURRENT_SOURCE_DIR}/${flexsrc}"
++           MAIN_DEPENDENCY ${flexsrc}
++           DEPENDS ${${compiler_headers}}
++           WORKING_DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR} )
 diff -Naur OpenShadingLanguage-Release-1.9.9/src/include/OSL/llvm_util.h external_osl/src/include/OSL/llvm_util.h
 --- OpenShadingLanguage-Release-1.9.9/src/include/OSL/llvm_util.h	2018-05-01 16:39:02 -0600
 +++ external_osl/src/include/OSL/llvm_util.h	2018-08-25 14:05:00 -0600
@@ -48,50 +63,19 @@ diff -Naur org/CMakeLists.txt external_osl/CMakeLists.txt
 
  set (OSL_NO_DEFAULT_TEXTURESYSTEM OFF CACHE BOOL "Do not use create a raw OIIO::TextureSystem")
  if (OSL_NO_DEFAULT_TEXTURESYSTEM)
-diff --git a/CMakeLists.txt b/CMakeLists.txt
-index 990f50d69..46ef7351d 100644
---- a/CMakeLists.txt
-+++ b/CMakeLists.txt
-@@ -252,11 +252,9 @@ install (EXPORT OSL_EXPORTED_TARGETS
-         FILE ${OSL_TARGETS_EXPORT_NAME}
-         NAMESPACE ${PROJECT_NAME}::)
- 
--
--
--
--osl_add_all_tests()
--
-+if (${PROJECT_NAME}_BUILD_TESTS AND NOT ${PROJECT_NAME}_IS_SUBPROJECT)
-+	osl_add_all_tests()
-+endif ()
- 
- if (NOT ${PROJECT_NAME}_IS_SUBPROJECT)
-     include (packaging)
-diff -Naur external_osl_orig/src/cmake/externalpackages.cmake external_osl/src/cmake/externalpackages.cmake
---- external_osl_orig/src/cmake/externalpackages.cmake  2021-06-01 13:44:18 -0600
-+++ external_osl/src/cmake/externalpackages.cmake       2021-06-28 07:44:32 -0600
-@@ -80,6 +80,7 @@
-
-
- checked_find_package (ZLIB REQUIRED)  # Needed by several packages
-+checked_find_package (PNG REQUIRED)  # Needed since OIIO needs it
-
- # IlmBase & OpenEXR
- checked_find_package (OpenEXR REQUIRED
-diff -Naur external_osl_orig/src/liboslcomp/oslcomp.cpp external_osl/src/liboslcomp/oslcomp.cpp
---- external_osl_orig/src/liboslcomp/oslcomp.cpp        2021-06-01 13:44:18 -0600
-+++ external_osl/src/liboslcomp/oslcomp.cpp     2021-06-28 09:11:06 -0600
-@@ -21,6 +21,13 @@
- #if !defined(__STDC_CONSTANT_MACROS)
- #    define __STDC_CONSTANT_MACROS 1
+diff --git a/src/liboslexec/llvm_util.cpp b/src/liboslexec/llvm_util.cpp
+index 445f6400..3d468de2 100644
+--- a/src/liboslexec/llvm_util.cpp
++++ b/src/liboslexec/llvm_util.cpp
+@@ -3430,8 +3430,9 @@ LLVM_Util::call_function (llvm::Value *func, cspan<llvm::Value *> args)
+ #endif
+     //llvm_gen_debug_printf (std::string("start ") + std::string(name));
+ #if OSL_LLVM_VERSION >= 110
+-    OSL_DASSERT(llvm::isa<llvm::Function>(func));
+-    llvm::Value *r = builder().CreateCall(llvm::cast<llvm::Function>(func), llvm::ArrayRef<llvm::Value *>(args.data(), args.size()));
++    llvm::Value* r = builder().CreateCall(
++        llvm::cast<llvm::FunctionType>(func->getType()->getPointerElementType()), func,
++        llvm::ArrayRef<llvm::Value*>(args.data(), args.size()));
+ #else
+     llvm::Value *r = builder().CreateCall (func, llvm::ArrayRef<llvm::Value *>(args.data(), args.size()));
  #endif
-+
-+// clang uses CALLBACK in its templates which causes issues if it is already defined
-+#ifdef _WIN32 && defined(CALLBACK)
-+#    undef CALLBACK
-+#endif
-+
-+//
- #include <clang/Basic/TargetInfo.h>
- #include <clang/Frontend/CompilerInstance.h>
- #include <clang/Frontend/TextDiagnosticPrinter.h>

build_files/cmake/cmake_consistency_check_config.py

@@ -8,9 +8,6 @@ IGNORE_SOURCE = (
     # specific source files
     "extern/audaspace/",
 
-    # Use for `WIN32` only.
-    "source/creator/blender_launcher_win32.c",
-
     # specific source files
     "extern/bullet2/src/BulletCollision/CollisionDispatch/btBox2dBox2dCollisionAlgorithm.cpp",
     "extern/bullet2/src/BulletCollision/CollisionDispatch/btConvex2dConvex2dAlgorithm.cpp",

build_files/cmake/cmake_netbeans_project.py

@@ -82,7 +82,7 @@ def create_nb_project_main():
         make_exe = cmake_cache_var("CMAKE_MAKE_PROGRAM")
         make_exe_basename = os.path.basename(make_exe)
 
-        # --------------- NetBeans specific.
+        # --------------- NB specific
         defines = [("%s=%s" % cdef) if cdef[1] else cdef[0] for cdef in defines]
         defines += [cdef.replace("#define", "").strip() for cdef in cmake_compiler_defines()]
 

build_files/cmake/config/blender_developer.cmake

@@ -7,6 +7,7 @@
 set(WITH_ASSERT_ABORT                 ON  CACHE BOOL "" FORCE)
 set(WITH_BUILDINFO                    OFF CACHE BOOL "" FORCE)
 set(WITH_COMPILER_ASAN                ON  CACHE BOOL "" FORCE)
+set(WITH_CYCLES_DEBUG                 ON  CACHE BOOL "" FORCE)
 set(WITH_CYCLES_NATIVE_ONLY           ON  CACHE BOOL "" FORCE)
 set(WITH_DOC_MANPAGE                  OFF CACHE BOOL "" FORCE)
 set(WITH_GTESTS                       ON  CACHE BOOL "" FORCE)

build_files/cmake/config/blender_full.cmake

@@ -29,7 +29,6 @@ set(WITH_IMAGE_OPENEXR       ON  CACHE BOOL "" FORCE)
 set(WITH_IMAGE_OPENJPEG      ON  CACHE BOOL "" FORCE)
 set(WITH_IMAGE_TIFF          ON  CACHE BOOL "" FORCE)
 set(WITH_INPUT_NDOF          ON  CACHE BOOL "" FORCE)
-set(WITH_INPUT_IME           ON  CACHE BOOL "" FORCE)
 set(WITH_INTERNATIONAL       ON  CACHE BOOL "" FORCE)
 set(WITH_LIBMV               ON  CACHE BOOL "" FORCE)
 set(WITH_LIBMV_SCHUR_SPECIALIZATIONS ON CACHE BOOL "" FORCE)

build_files/cmake/config/blender_release.cmake

@@ -30,7 +30,6 @@ set(WITH_IMAGE_OPENEXR       ON  CACHE BOOL "" FORCE)
 set(WITH_IMAGE_OPENJPEG      ON  CACHE BOOL "" FORCE)
 set(WITH_IMAGE_TIFF          ON  CACHE BOOL "" FORCE)
 set(WITH_INPUT_NDOF          ON  CACHE BOOL "" FORCE)
-set(WITH_INPUT_IME           ON  CACHE BOOL "" FORCE)
 set(WITH_INTERNATIONAL       ON  CACHE BOOL "" FORCE)
 set(WITH_LIBMV               ON  CACHE BOOL "" FORCE)
 set(WITH_LIBMV_SCHUR_SPECIALIZATIONS ON CACHE BOOL "" FORCE)

build_files/cmake/platform/platform_apple.cmake

@@ -404,7 +404,7 @@ endif()
 
 # CMake FindOpenMP doesn't know about AppleClang before 3.12, so provide custom flags.
 if(WITH_OPENMP)
-  if(CMAKE_C_COMPILER_ID MATCHES "Clang")
+  if(CMAKE_C_COMPILER_ID MATCHES "Clang" AND CMAKE_C_COMPILER_VERSION VERSION_GREATER_EQUAL "7.0")
     # Use OpenMP from our precompiled libraries.
     message(STATUS "Using ${LIBDIR}/openmp for OpenMP")
     set(OPENMP_CUSTOM ON)
@@ -480,8 +480,10 @@ else()
   set(CMAKE_CXX_FLAGS_RELEASE "-O2 -mdynamic-no-pic")
 endif()
 
-# Clang has too low template depth of 128 for libmv.
-string(APPEND CMAKE_CXX_FLAGS " -ftemplate-depth=1024")
+if(${XCODE_VERSION} VERSION_EQUAL 5 OR ${XCODE_VERSION} VERSION_GREATER 5)
+  # Xcode 5 is always using CLANG, which has too low template depth of 128 for libmv
+  string(APPEND CMAKE_CXX_FLAGS " -ftemplate-depth=1024")
+endif()
 
 # Avoid conflicts with Luxrender, and other plug-ins that may use the same
 # libraries as Blender with a different version or build options.

build_files/cmake/platform/platform_unix.cmake

@@ -581,6 +581,8 @@ if(WITH_GHOST_WAYLAND)
   pkg_check_modules(wayland-cursor REQUIRED wayland-cursor)
   pkg_check_modules(dbus REQUIRED dbus-1)
 
+  set(WITH_GL_EGL ON)
+
   list(APPEND PLATFORM_LINKLIBS
     ${wayland-client_LINK_LIBRARIES}
     ${wayland-egl_LINK_LIBRARIES}

build_files/config/README.md

@@ -1,11 +1,8 @@
 Pipeline Config
 ===============
 
-This configuration file is used by buildbot new build pipeline for the `update-code` step.
+This configuration file is used by buildbot new pipeline for the `update-code` step.
 
-It will also be used by the `../utils/make_update.py` script in the near future.
+It will soon be used by the ../utils/make_update.py script.
 
-NOTES:
-* Keep both `yaml` and `json` files in sync until deployment of build pipeline updates.
-* The `json` file  be removed once all branches are running with the `yaml` file.
-* Expected buildbot pipeline update is *Friday, July 30th* or *Monday August, 2nd*.
+Both buildbot and developers will eventually use the same configuration file.

build_files/config/pipeline_config.json

@@ -41,7 +41,7 @@
             },
             "cuda11":
             {
-                "version": "11.4"
+                "version": "11.3"
             }
         },
         "cmake":

build_files/config/pipeline_config.yaml

@@ -1,70 +0,0 @@
-#
-# Used by Buildbot build pipeline make_update.py script only for now
-# We intended to update the make_update.py in the branches to use this file eventually
-#
-update-code:
-    git:
-        submodules:
-        -   branch: master
-            commit_id: HEAD
-            path: release/scripts/addons
-        -   branch: master
-            commit_id: HEAD
-            path: release/scripts/addons_contrib
-        -   branch: master
-            commit_id: HEAD
-            path: release/datafiles/locale
-        -   branch: master
-            commit_id: HEAD
-            path: source/tools
-    svn:
-        libraries:
-            darwin-arm64:
-                branch: trunk
-                commit_id: HEAD
-                path: lib/darwin_arm64
-            darwin-x86_64:
-                branch: trunk
-                commit_id: HEAD
-                path: lib/darwin
-            linux-x86_64:
-                branch: trunk
-                commit_id: HEAD
-                path: lib/linux_centos7_x86_64
-            windows-amd64:
-                branch: trunk
-                commit_id: HEAD
-                path: lib/win64_vc15
-        tests:
-            branch: trunk
-            commit_id: HEAD
-            path: lib/tests
-        benchmarks:
-            branch: trunk
-            commit_id: HEAD
-            path: lib/benchmarks
-
-#
-# Buildbot only configs
-#
-buildbot:
-    gcc:
-        version: '9.0.0'
-    cuda10:
-        version: '10.1.0'
-    cuda11:
-        version: '11.4.0'
-    optix:
-        version: '7.1.0'
-    cmake:
-        default:
-            version: any
-            overrides: {}
-        darwin-arm64:
-            overrides: {}
-        darwin-x86_64:
-            overrides: {}
-        linux-x86_64:
-            overrides: {}
-        windows-amd64:
-            overrides: {}

build_files/windows/autodetect_msvc.cmd

@@ -1,9 +1,9 @@
 echo No explicit msvc version requested, autodetecting version.
 
-call "%~dp0\detect_msvc2019.cmd"
+call "%~dp0\detect_msvc2017.cmd"
 if %ERRORLEVEL% EQU 0 goto DetectionComplete
 
-call "%~dp0\detect_msvc2017.cmd"
+call "%~dp0\detect_msvc2019.cmd"
 if %ERRORLEVEL% EQU 0 goto DetectionComplete
 
 call "%~dp0\detect_msvc2022.cmd"

doc/manpage/blender.1.py

@@ -122,7 +122,7 @@ is a full-featured 3D application. It supports the entirety of the 3D pipeline -
 '''modeling, rigging, animation, simulation, rendering, compositing, motion tracking, and video editing.
 
 Use Blender to create 3D images and animations, films and commercials, content for games, '''
-r'''architectural and industrial visualizations, and scientific visualizations.
+r'''architectural and industrial visualizatons, and scientific visualizations.
 
 https://www.blender.org''')
 

doc/python_api/examples/bpy.app.handlers.2.py

@@ -12,7 +12,6 @@ such cases, lock the interface (Render → Lock Interface or
 Below is an example of a mesh that is altered from a handler:
 """
 
-
 def frame_change_pre(scene):
     # A triangle that shifts in the z direction
     zshift = scene.frame_current * 0.1

doc/python_api/examples/bpy.app.timers.5.py

@@ -16,12 +16,10 @@ execution_queue = queue.Queue()
 def run_in_main_thread(function):
     execution_queue.put(function)
 
-
 def execute_queued_functions():
     while not execution_queue.empty():
         function = execution_queue.get()
         function()
     return 1.0
 
-
 bpy.app.timers.register(execute_queued_functions)

doc/python_api/examples/bpy.msgbus.1.py

@@ -31,13 +31,11 @@ owner = object()
 
 subscribe_to = bpy.context.object.location
 
-
 def msgbus_callback(*args):
     # This will print:
     # Something changed! (1, 2, 3)
     print("Something changed!", args)
 
-
 bpy.msgbus.subscribe_rna(
     key=subscribe_to,
     owner=owner,

doc/python_api/examples/bpy.types.Depsgraph.7.py

@@ -44,7 +44,7 @@ class OBJECT_OT_object_to_curve(bpy.types.Operator):
         # Remove temporary curve.
         obj.to_curve_clear()
         # Invoke to_curve() with applying modifiers.
-        curve_with_modifiers = obj.to_curve(depsgraph, apply_modifiers=True)
+        curve_with_modifiers = obj.to_curve(depsgraph, apply_modifiers = True)
         self.report({'INFO'}, f"{len(curve_with_modifiers.splines)} splines in new curve with modifiers.")
         # Remove temporary curve.
         obj.to_curve_clear()

doc/python_api/examples/gpu.6.py

@@ -21,7 +21,6 @@ batch = batch_for_shader(
     },
 )
 
-
 def draw():
     shader.bind()
     shader.uniform_sampler("image", texture)

doc/python_api/sphinx_doc_gen.py

@@ -1047,7 +1047,6 @@ context_type_map = {
     "annotation_data": ("GreasePencil", False),
     "annotation_data_owner": ("ID", False),
     "armature": ("Armature", False),
-    "asset_library": ("AssetLibraryReference", False),
     "bone": ("Bone", False),
     "brush": ("Brush", False),
     "camera": ("Camera", False),
@@ -1114,7 +1113,6 @@ context_type_map = {
     "texture_slot": ("MaterialTextureSlot", False),
     "texture_user": ("ID", False),
     "texture_user_property": ("Property", False),
-    "ui_list": ("UIList", False),
     "vertex_paint_object": ("Object", False),
     "view_layer": ("ViewLayer", False),
     "visible_bones": ("EditBone", True),

extern/audaspace/include/devices/SoftwareDevice.h

@@ -265,12 +265,6 @@ protected:
 	 */
 	void setSpecs(Specs specs);
 
-	/**
-	 * Sets the audio output specification of the device.
-	 * \param specs The output specification.
-	 */
-	void setSpecs(DeviceSpecs specs);
-
 	/**
 	 * Empty default constructor. To setup the device call the function create()
 	 * and to uninitialize call destroy().

extern/audaspace/include/respec/Mixer.h

@@ -87,12 +87,6 @@ public:
 	 */
 	void setSpecs(Specs specs);
 
-	/**
-	 * Sets the target specification for superposing.
-	 * \param specs The target specification.
-	 */
-	void setSpecs(DeviceSpecs specs);
-
 	/**
 	 * Mixes a buffer.
 	 * \param buffer The buffer to superpose.

extern/audaspace/plugins/pulseaudio/PulseAudioDevice.cpp

@@ -78,7 +78,6 @@ void PulseAudioDevice::runMixingThread()
 			if(shouldStop())
 			{
 				AUD_pa_stream_cork(m_stream, 1, nullptr, nullptr);
-				AUD_pa_stream_flush(m_stream, nullptr, nullptr);
 				doStop();
 				return;
 			}
@@ -87,10 +86,7 @@ void PulseAudioDevice::runMixingThread()
 		if(AUD_pa_stream_is_corked(m_stream))
 			AUD_pa_stream_cork(m_stream, 0, nullptr, nullptr);
 
-		// similar to AUD_pa_mainloop_iterate(m_mainloop, false, nullptr); except with a longer timeout
-		AUD_pa_mainloop_prepare(m_mainloop, 1 << 14);
-		AUD_pa_mainloop_poll(m_mainloop);
-		AUD_pa_mainloop_dispatch(m_mainloop);
+		AUD_pa_mainloop_iterate(m_mainloop, true, nullptr);
 	}
 }
 

extern/audaspace/plugins/pulseaudio/PulseAudioSymbols.h

@@ -24,7 +24,6 @@ PULSEAUDIO_SYMBOL(pa_context_unref);
 PULSEAUDIO_SYMBOL(pa_stream_begin_write);
 PULSEAUDIO_SYMBOL(pa_stream_connect_playback);
 PULSEAUDIO_SYMBOL(pa_stream_cork);
-PULSEAUDIO_SYMBOL(pa_stream_flush);
 PULSEAUDIO_SYMBOL(pa_stream_is_corked);
 PULSEAUDIO_SYMBOL(pa_stream_new);
 PULSEAUDIO_SYMBOL(pa_stream_set_buffer_attr);
@@ -36,6 +35,3 @@ PULSEAUDIO_SYMBOL(pa_mainloop_free);
 PULSEAUDIO_SYMBOL(pa_mainloop_get_api);
 PULSEAUDIO_SYMBOL(pa_mainloop_new);
 PULSEAUDIO_SYMBOL(pa_mainloop_iterate);
-PULSEAUDIO_SYMBOL(pa_mainloop_prepare);
-PULSEAUDIO_SYMBOL(pa_mainloop_poll);
-PULSEAUDIO_SYMBOL(pa_mainloop_dispatch);

extern/audaspace/plugins/wasapi/WASAPIDevice.cpp

@@ -31,81 +31,65 @@ template <class T> void SafeRelease(T **ppT)
 	}
 }
 
-HRESULT WASAPIDevice::setupRenderClient(IAudioRenderClient*& render_client, UINT32& buffer_size)
+void WASAPIDevice::runMixingThread()
 {
-	const IID IID_IAudioRenderClient = __uuidof(IAudioRenderClient);
-
+	UINT32 buffer_size;
 	UINT32 padding;
 	UINT32 length;
 	data_t* buffer;
 
-	HRESULT result;
+	IAudioRenderClient* render_client = nullptr;
 
-	if(FAILED(result = m_audio_client->GetBufferSize(&buffer_size)))
-		return result;
+	{
+		std::lock_guard<ILockable> lock(*this);
 
-	if(FAILED(result = m_audio_client->GetService(IID_IAudioRenderClient, reinterpret_cast<void**>(&render_client))))
-		return result;
+		const IID IID_IAudioRenderClient = __uuidof(IAudioRenderClient);
 
-	if(FAILED(result = m_audio_client->GetCurrentPadding(&padding)))
-		return result;
+		if(FAILED(m_audio_client->GetBufferSize(&buffer_size)))
+			goto init_error;
 
-	length = buffer_size - padding;
+		if(FAILED(m_audio_client->GetService(IID_IAudioRenderClient, reinterpret_cast<void**>(&render_client))))
+			goto init_error;
 
-	if(FAILED(result = render_client->GetBuffer(length, &buffer)))
-		return result;
+		if(FAILED(m_audio_client->GetCurrentPadding(&padding)))
+			goto init_error;
 
-	mix((data_t*)buffer, length);
+		length = buffer_size - padding;
 
-	if(FAILED(result = render_client->ReleaseBuffer(length, 0)))
-		return result;
+		if(FAILED(render_client->GetBuffer(length, &buffer)))
+			goto init_error;
+
+		mix((data_t*)buffer, length);
+
+		if(FAILED(render_client->ReleaseBuffer(length, 0)))
+		{
+			init_error:
+				SafeRelease(&render_client);
+				doStop();
+				return;
+		}
+	}
 
 	m_audio_client->Start();
 
-	return result;
-}
-
-void WASAPIDevice::runMixingThread()
-{
-	UINT32 buffer_size;
-
-	IAudioRenderClient* render_client = nullptr;
-
-	std::chrono::milliseconds sleep_duration;
-
-	bool run_init = true;
+	auto sleepDuration = std::chrono::milliseconds(buffer_size * 1000 / int(m_specs.rate) / 2);
 
 	for(;;)
 	{
-		HRESULT result = S_OK;
-
 		{
-			UINT32 padding;
-			UINT32 length;
-			data_t* buffer;
 			std::lock_guard<ILockable> lock(*this);
 
-			if(run_init)
-			{
-				result = setupRenderClient(render_client, buffer_size);
-
-				if(FAILED(result))
-					goto stop_thread;
-
-				sleep_duration = std::chrono::milliseconds(buffer_size * 1000 / int(m_specs.rate) / 2);
-			}
-
-			if(FAILED(result = m_audio_client->GetCurrentPadding(&padding)))
+			if(FAILED(m_audio_client->GetCurrentPadding(&padding)))
 				goto stop_thread;
 
 			length = buffer_size - padding;
 
-			if(FAILED(result = render_client->GetBuffer(length, &buffer)))
+			if(FAILED(render_client->GetBuffer(length, &buffer)))
 				goto stop_thread;
 
 			mix((data_t*)buffer, length);
 
-			if(FAILED(result = render_client->ReleaseBuffer(length, 0)))
+			if(FAILED(render_client->ReleaseBuffer(length, 0)))
 				goto stop_thread;
 
 			// stop thread
@@ -114,51 +98,53 @@ void WASAPIDevice::runMixingThread()
 				stop_thread:
 					m_audio_client->Stop();
 					SafeRelease(&render_client);
-
-					if(result == AUDCLNT_E_DEVICE_INVALIDATED)
-					{
-						DeviceSpecs specs = m_specs;
-						if(!setupDevice(specs))
-							result = S_FALSE;
-						else
-						{
-							setSpecs(specs);
-
-							run_init = true;
-						}
-					}
-
-					if(result != AUDCLNT_E_DEVICE_INVALIDATED)
-					{
-						doStop();
-						return;
-					}
+					doStop();
+					return;
 			}
 		}
 
-		std::this_thread::sleep_for(sleep_duration);
+		std::this_thread::sleep_for(sleepDuration);
 	}
 }
 
-bool WASAPIDevice::setupDevice(DeviceSpecs &specs)
+WASAPIDevice::WASAPIDevice(DeviceSpecs specs, int buffersize) :
+	m_imm_device_enumerator(nullptr),
+	m_imm_device(nullptr),
+	m_audio_client(nullptr),
+
+	m_wave_format_extensible({})
 {
-	SafeRelease(&m_audio_client);
-	SafeRelease(&m_imm_device);
+	// initialize COM if it hasn't happened yet
+	CoInitializeEx(nullptr, COINIT_MULTITHREADED);
 
+	const CLSID CLSID_MMDeviceEnumerator = __uuidof(MMDeviceEnumerator);
+	const IID IID_IMMDeviceEnumerator = __uuidof(IMMDeviceEnumerator);
 	const IID IID_IAudioClient = __uuidof(IAudioClient);
 
-	if(FAILED(m_imm_device_enumerator->GetDefaultAudioEndpoint(eRender, eMultimedia, &m_imm_device)))
-		return false;
-
-	if(FAILED(m_imm_device->Activate(IID_IAudioClient, CLSCTX_ALL, nullptr, reinterpret_cast<void**>(&m_audio_client))))
-		return false;
-
 	WAVEFORMATEXTENSIBLE wave_format_extensible_closest_match;
 	WAVEFORMATEXTENSIBLE* closest_match_pointer = &wave_format_extensible_closest_match;
 
+	HRESULT result;
+
 	REFERENCE_TIME minimum_time = 0;
 	REFERENCE_TIME buffer_duration;
 
+	if(FAILED(CoCreateInstance(CLSID_MMDeviceEnumerator, nullptr, CLSCTX_ALL, IID_IMMDeviceEnumerator, reinterpret_cast<void**>(&m_imm_device_enumerator))))
+		goto error;
+
+	if(FAILED(m_imm_device_enumerator->GetDefaultAudioEndpoint(eRender, eMultimedia, &m_imm_device)))
+		goto error;
+
+	if(FAILED(m_imm_device->Activate(IID_IAudioClient, CLSCTX_ALL, nullptr, reinterpret_cast<void**>(&m_audio_client))))
+		goto error;
+
+	if(specs.channels == CHANNELS_INVALID)
+		specs.channels = CHANNELS_STEREO;
+	if(specs.format == FORMAT_INVALID)
+		specs.format = FORMAT_FLOAT32;
+	if(specs.rate == RATE_INVALID)
+		specs.rate = RATE_48000;
+
 	switch(specs.format)
 	{
 	case FORMAT_U8:
@@ -217,14 +203,12 @@ bool WASAPIDevice::setupDevice(DeviceSpecs &specs)
 	m_wave_format_extensible.Format.cbSize = 22;
 	m_wave_format_extensible.Samples.wValidBitsPerSample = m_wave_format_extensible.Format.wBitsPerSample;
 
-	HRESULT result = m_audio_client->IsFormatSupported(AUDCLNT_SHAREMODE_SHARED, reinterpret_cast<const WAVEFORMATEX*>(&m_wave_format_extensible), reinterpret_cast<WAVEFORMATEX**>(&closest_match_pointer));
+	result = m_audio_client->IsFormatSupported(AUDCLNT_SHAREMODE_SHARED, reinterpret_cast<const WAVEFORMATEX*>(&m_wave_format_extensible), reinterpret_cast<WAVEFORMATEX**>(&closest_match_pointer));
 
 	if(result == S_FALSE)
 	{
-		bool errored = false;
-
 		if(closest_match_pointer->Format.wFormatTag != WAVE_FORMAT_EXTENSIBLE)
-			goto closest_match_error;
+			goto error;
 
 		specs.channels = Channels(closest_match_pointer->Format.nChannels);
 		specs.rate = closest_match_pointer->Format.nSamplesPerSec;
@@ -236,7 +220,7 @@ bool WASAPIDevice::setupDevice(DeviceSpecs &specs)
 			else if(closest_match_pointer->Format.wBitsPerSample == 64)
 				specs.format = FORMAT_FLOAT64;
 			else
-				goto closest_match_error;
+				goto error;
 		}
 		else if(closest_match_pointer->SubFormat == KSDATAFORMAT_SUBTYPE_PCM)
 		{
@@ -255,81 +239,44 @@ bool WASAPIDevice::setupDevice(DeviceSpecs &specs)
 				specs.format = FORMAT_S32;
 				break;
 			default:
-				goto closest_match_error;
+				goto error;
 				break;
 			}
 		}
 		else
-			goto closest_match_error;
+			goto error;
 
 		m_wave_format_extensible = *closest_match_pointer;
 
-		if(false)
-		{
-			closest_match_error:
-			errored = true;
-		}
-
 		if(closest_match_pointer != &wave_format_extensible_closest_match)
 		{
 			CoTaskMemFree(closest_match_pointer);
 			closest_match_pointer = &wave_format_extensible_closest_match;
 		}
-
-		if(errored)
-			return false;
 	}
 	else if(FAILED(result))
-		return false;
+		goto error;
 
 	if(FAILED(m_audio_client->GetDevicePeriod(nullptr, &minimum_time)))
-		return false;
+		goto error;
 
-	buffer_duration = REFERENCE_TIME(m_buffersize) * REFERENCE_TIME(10000000) / REFERENCE_TIME(specs.rate);
+	buffer_duration = REFERENCE_TIME(buffersize) * REFERENCE_TIME(10000000) / REFERENCE_TIME(specs.rate);
 
 	if(minimum_time > buffer_duration)
 		buffer_duration = minimum_time;
 
-	if(FAILED(m_audio_client->Initialize(AUDCLNT_SHAREMODE_SHARED, 0, buffer_duration, 0, reinterpret_cast<WAVEFORMATEX*>(&m_wave_format_extensible), nullptr)))
-		return false;
-
-	return true;
-}
-
-WASAPIDevice::WASAPIDevice(DeviceSpecs specs, int buffersize) :
-	m_buffersize(buffersize),
-	m_imm_device_enumerator(nullptr),
-	m_imm_device(nullptr),
-	m_audio_client(nullptr),
-
-	m_wave_format_extensible({})
-{
-	// initialize COM if it hasn't happened yet
-	CoInitializeEx(nullptr, COINIT_MULTITHREADED);
-
-	const CLSID CLSID_MMDeviceEnumerator = __uuidof(MMDeviceEnumerator);
-	const IID IID_IMMDeviceEnumerator = __uuidof(IMMDeviceEnumerator);
-
-	if(specs.channels == CHANNELS_INVALID)
-		specs.channels = CHANNELS_STEREO;
-	if(specs.format == FORMAT_INVALID)
-		specs.format = FORMAT_FLOAT32;
-	if(specs.rate == RATE_INVALID)
-		specs.rate = RATE_48000;
-
-	if(FAILED(CoCreateInstance(CLSID_MMDeviceEnumerator, nullptr, CLSCTX_ALL, IID_IMMDeviceEnumerator, reinterpret_cast<void**>(&m_imm_device_enumerator))))
-		goto error;
-
-	if(!setupDevice(specs))
-		goto error;
-
 	m_specs = specs;
 
+	if(FAILED(m_audio_client->Initialize(AUDCLNT_SHAREMODE_SHARED, 0, buffer_duration, 0, reinterpret_cast<WAVEFORMATEX*>(&m_wave_format_extensible), nullptr)))
+		goto error;
+
 	create();
 
 	return;
 
 	error:
+	if(closest_match_pointer != &wave_format_extensible_closest_match)
+		CoTaskMemFree(closest_match_pointer);
 	SafeRelease(&m_imm_device);
 	SafeRelease(&m_imm_device_enumerator);
 	SafeRelease(&m_audio_client);

extern/audaspace/plugins/wasapi/WASAPIDevice.h

@@ -43,21 +43,16 @@ AUD_NAMESPACE_BEGIN
 class AUD_PLUGIN_API WASAPIDevice : public ThreadedDevice
 {
 private:
-	int m_buffersize;
 	IMMDeviceEnumerator* m_imm_device_enumerator;
 	IMMDevice* m_imm_device;
 	IAudioClient* m_audio_client;
 	WAVEFORMATEXTENSIBLE m_wave_format_extensible;
 
-	AUD_LOCAL HRESULT setupRenderClient(IAudioRenderClient*& render_client, UINT32& buffer_size);
-
 	/**
 	 * Streaming thread main function.
 	 */
 	AUD_LOCAL void runMixingThread();
 
-	AUD_LOCAL bool setupDevice(DeviceSpecs& specs);
-
 	// delete copy constructor and operator=
 	WASAPIDevice(const WASAPIDevice&) = delete;
 	WASAPIDevice& operator=(const WASAPIDevice&) = delete;

extern/audaspace/src/devices/SoftwareDevice.cpp

@@ -756,7 +756,6 @@ void SoftwareDevice::mix(data_t* buffer, int length)
 			// get the buffer from the source
 			pos = 0;
 			len = length;
-			eos = false;
 
 			// update 3D Info
 			sound->update();
@@ -843,27 +842,6 @@ void SoftwareDevice::setSpecs(Specs specs)
 	{
 		sound->setSpecs(specs);
 	}
-
-	for(auto& sound : m_pausedSounds)
-	{
-		sound->setSpecs(specs);
-	}
-}
-
-void SoftwareDevice::setSpecs(DeviceSpecs specs)
-{
-	m_specs = specs;
-	m_mixer->setSpecs(specs);
-
-	for(auto& sound : m_playingSounds)
-	{
-		sound->setSpecs(specs.specs);
-	}
-
-	for(auto& sound : m_pausedSounds)
-	{
-		sound->setSpecs(specs.specs);
-	}
 }
 
 SoftwareDevice::SoftwareDevice()

extern/audaspace/src/respec/Mixer.cpp

@@ -21,25 +21,9 @@
 
 AUD_NAMESPACE_BEGIN
 
-Mixer::Mixer(DeviceSpecs specs)
+Mixer::Mixer(DeviceSpecs specs) :
+	m_specs(specs)
 {
-	setSpecs(specs);
-}
-
-DeviceSpecs Mixer::getSpecs() const
-{
-	return m_specs;
-}
-
-void Mixer::setSpecs(Specs specs)
-{
-	m_specs.specs = specs;
-}
-
-void Mixer::setSpecs(DeviceSpecs specs)
-{
-	m_specs = specs;
-
 	switch(m_specs.format)
 	{
 	case FORMAT_U8:
@@ -70,6 +54,16 @@ void Mixer::setSpecs(DeviceSpecs specs)
 	}
 }
 
+DeviceSpecs Mixer::getSpecs() const
+{
+	return m_specs;
+}
+
+void Mixer::setSpecs(Specs specs)
+{
+	m_specs.specs = specs;
+}
+
 void Mixer::clear(int length)
 {
 	m_buffer.assureSize(length * AUD_SAMPLE_SIZE(m_specs));

extern/glog/README.blender

@@ -7,4 +7,3 @@ Local modifications:
   checks for functions and so are needed.
 * Added special definitions of HAVE_SNPRINTF and HAVE_LIB_GFLAGS
   in Windows' specific config.h.
-* Silenced syscall deprecation warnings on macOS >= 10.12.

extern/glog/src/raw_logging.cc

@@ -59,7 +59,7 @@
 # include <unistd.h>
 #endif
 
-#if (defined(HAVE_SYSCALL_H) || defined(HAVE_SYS_SYSCALL_H)) && (!(defined OS_MACOSX))
+#if defined(HAVE_SYSCALL_H) || defined(HAVE_SYS_SYSCALL_H)
 # define safe_write(fd, s, len)  syscall(SYS_write, fd, s, len)
 #else
   // Not so safe, but what can you do?

extern/glog/src/utilities.cc

@@ -259,13 +259,7 @@ pid_t GetTID() {
 #endif
   static bool lacks_gettid = false;
   if (!lacks_gettid) {
-#ifdef OS_MACOSX
-    uint64_t tid64;
-    const int error = pthread_threadid_np(NULL, &tid64);
-    pid_t tid = error ? -1 : (pid_t)tid64;
-#else
     pid_t tid = syscall(__NR_gettid);
-#endif
     if (tid != -1) {
       return tid;
     }

extern/rangetree/intern/generic_alloc_impl.h

@@ -32,7 +32,7 @@
  * - #TPOOL_STRUCT: Name for pool struct name.
  * - #TPOOL_CHUNK_SIZE: Chunk size (optional), use 64kb when not defined.
  *
- * \note #TPOOL_ALLOC_TYPE must be at least `sizeof(void *)`.
+ * \note #TPOOL_ALLOC_TYPE must be at least ``sizeof(void *)``.
  *
  * Defines the API, uses #TPOOL_IMPL_PREFIX to prefix each function.
  *

intern/atomic/intern/atomic_ops_unix.h

@@ -49,9 +49,9 @@
 
 #include "atomic_ops_utils.h"
 
-#if defined(__arm__) || defined(__riscv)
-/* Attempt to fix compilation error on Debian armel and RISC-V kernels.
- * Both architectures do have both 32 and 64bit atomics, however
+#if defined(__arm__)
+/* Attempt to fix compilation error on Debian armel kernel.
+ * arm7 architecture does have both 32 and 64bit atomics, however
  * its gcc doesn't have __GCC_HAVE_SYNC_COMPARE_AND_SWAP_n defined.
  */
 #  define JE_FORCE_SYNC_COMPARE_AND_SWAP_1

intern/cycles/CMakeLists.txt

@@ -307,9 +307,9 @@ if(WITH_CYCLES_LOGGING)
   )
 endif()
 
-# NaN debugging
-if(WITH_CYCLES_DEBUG_NAN)
-  add_definitions(-DWITH_CYCLES_DEBUG_NAN)
+# Debugging capabilities (debug passes etc).
+if(WITH_CYCLES_DEBUG)
+  add_definitions(-DWITH_CYCLES_DEBUG)
 endif()
 
 if(NOT OPENIMAGEIO_PUGIXML_FOUND)

intern/cycles/blender/addon/engine.py

@@ -235,12 +235,9 @@ def system_info():
 
 
 def list_render_passes(scene, srl):
-    crl = srl.cycles
-
-    # Combined pass.
+    # Builtin Blender passes.
     yield ("Combined", "RGBA", 'COLOR')
 
-    # Data passes.
     if srl.use_pass_z:                     yield ("Depth",         "Z",    'VALUE')
     if srl.use_pass_mist:                  yield ("Mist",          "Z",    'VALUE')
     if srl.use_pass_normal:                yield ("Normal",        "XYZ",  'VECTOR')
@@ -248,8 +245,8 @@ def list_render_passes(scene, srl):
     if srl.use_pass_uv:                    yield ("UV",            "UVA",  'VECTOR')
     if srl.use_pass_object_index:          yield ("IndexOB",       "X",    'VALUE')
     if srl.use_pass_material_index:        yield ("IndexMA",       "X",    'VALUE')
-
-    # Light passes.
+    if srl.use_pass_shadow:                yield ("Shadow",        "RGB",  'COLOR')
+    if srl.use_pass_ambient_occlusion:     yield ("AO",            "RGB",  'COLOR')
     if srl.use_pass_diffuse_direct:        yield ("DiffDir",       "RGB",  'COLOR')
     if srl.use_pass_diffuse_indirect:      yield ("DiffInd",       "RGB",  'COLOR')
     if srl.use_pass_diffuse_color:         yield ("DiffCol",       "RGB",  'COLOR')
@@ -259,16 +256,19 @@ def list_render_passes(scene, srl):
     if srl.use_pass_transmission_direct:   yield ("TransDir",      "RGB",  'COLOR')
     if srl.use_pass_transmission_indirect: yield ("TransInd",      "RGB",  'COLOR')
     if srl.use_pass_transmission_color:    yield ("TransCol",      "RGB",  'COLOR')
-    if crl.use_pass_volume_direct:         yield ("VolumeDir",     "RGB",  'COLOR')
-    if crl.use_pass_volume_indirect:       yield ("VolumeInd",     "RGB",  'COLOR')
     if srl.use_pass_emit:                  yield ("Emit",          "RGB",  'COLOR')
     if srl.use_pass_environment:           yield ("Env",           "RGB",  'COLOR')
-    if srl.use_pass_shadow:                yield ("Shadow",        "RGB",  'COLOR')
-    if srl.use_pass_ambient_occlusion:     yield ("AO",            "RGB",  'COLOR')
 
-    # Debug passes.
+    # Cycles specific passes.
+    crl = srl.cycles
     if crl.pass_debug_render_time:             yield ("Debug Render Time",             "X",   'VALUE')
+    if crl.pass_debug_bvh_traversed_nodes:     yield ("Debug BVH Traversed Nodes",     "X",   'VALUE')
+    if crl.pass_debug_bvh_traversed_instances: yield ("Debug BVH Traversed Instances", "X",   'VALUE')
+    if crl.pass_debug_bvh_intersections:       yield ("Debug BVH Intersections",       "X",   'VALUE')
+    if crl.pass_debug_ray_bounces:             yield ("Debug Ray Bounces",             "X",   'VALUE')
     if crl.pass_debug_sample_count:            yield ("Debug Sample Count",            "X",   'VALUE')
+    if crl.use_pass_volume_direct:             yield ("VolumeDir",                     "RGB", 'COLOR')
+    if crl.use_pass_volume_indirect:           yield ("VolumeInd",                     "RGB", 'COLOR')
 
     # Cryptomatte passes.
     crypto_depth = (srl.pass_cryptomatte_depth + 1) // 2

intern/cycles/blender/addon/properties.py

@@ -263,7 +263,6 @@ class CyclesRenderSettings(bpy.types.PropertyGroup):
         name="Use Denoising",
         description="Denoise the rendered image",
         default=False,
-        update=update_render_passes,
     )
     use_preview_denoising: BoolProperty(
         name="Use Viewport Denoising",
@@ -483,7 +482,7 @@ class CyclesRenderSettings(bpy.types.PropertyGroup):
 
     transparent_max_bounces: IntProperty(
         name="Transparent Max Bounces",
-        description="Maximum number of transparent bounces. This is independent of maximum number of other bounces ",
+        description="Maximum number of transparent bounces",
         min=0, max=1024,
         default=8,
     )
@@ -1255,19 +1254,12 @@ class CyclesObjectSettings(bpy.types.PropertyGroup):
     )
 
     shadow_terminator_offset: FloatProperty(
-        name="Shadow Terminator Shading Offset",
+        name="Shadow Terminator Offset",
         description="Push the shadow terminator towards the light to hide artifacts on low poly geometry",
         min=0.0, max=1.0,
         default=0.0,
     )
 
-    shadow_terminator_geometry_offset: FloatProperty(
-        name="Shadow Terminator Geometry Offset",
-        description="Offset rays from the surface to reduce shadow terminator artifact on low poly geometry. Only affects triangles at grazing angles to light",
-        min=0.0, max=1.0,
-        default=0.1,
-    )
-
     is_shadow_catcher: BoolProperty(
         name="Shadow Catcher",
         description="Only render shadows on this object, for compositing renders into real footage",
@@ -1325,6 +1317,30 @@ class CyclesCurveRenderSettings(bpy.types.PropertyGroup):
 
 class CyclesRenderLayerSettings(bpy.types.PropertyGroup):
 
+    pass_debug_bvh_traversed_nodes: BoolProperty(
+        name="Debug BVH Traversed Nodes",
+        description="Store Debug BVH Traversed Nodes pass",
+        default=False,
+        update=update_render_passes,
+    )
+    pass_debug_bvh_traversed_instances: BoolProperty(
+        name="Debug BVH Traversed Instances",
+        description="Store Debug BVH Traversed Instances pass",
+        default=False,
+        update=update_render_passes,
+    )
+    pass_debug_bvh_intersections: BoolProperty(
+        name="Debug BVH Intersections",
+        description="Store Debug BVH Intersections",
+        default=False,
+        update=update_render_passes,
+    )
+    pass_debug_ray_bounces: BoolProperty(
+        name="Debug Ray Bounces",
+        description="Store Debug Ray Bounces pass",
+        default=False,
+        update=update_render_passes,
+    )
     pass_debug_render_time: BoolProperty(
         name="Debug Render Time",
         description="Render time in milliseconds per sample and pixel",

intern/cycles/blender/addon/ui.py

@@ -485,12 +485,10 @@ class CYCLES_RENDER_PT_light_paths_max_bounces(CyclesButtonsPanel, Panel):
         col = layout.column(align=True)
         col.prop(cscene, "diffuse_bounces", text="Diffuse")
         col.prop(cscene, "glossy_bounces", text="Glossy")
+        col.prop(cscene, "transparent_max_bounces", text="Transparency")
         col.prop(cscene, "transmission_bounces", text="Transmission")
         col.prop(cscene, "volume_bounces", text="Volume")
 
-        col = layout.column(align=True)
-        col.prop(cscene, "transparent_max_bounces", text="Transparent")
-
 
 class CYCLES_RENDER_PT_light_paths_clamping(CyclesButtonsPanel, Panel):
     bl_label = "Clamping"
@@ -936,6 +934,29 @@ class CYCLES_RENDER_PT_passes_crypto(CyclesButtonsPanel, ViewLayerCryptomattePan
     bl_parent_id = "CYCLES_RENDER_PT_passes"
 
 
+class CYCLES_RENDER_PT_passes_debug(CyclesButtonsPanel, Panel):
+    bl_label = "Debug"
+    bl_context = "view_layer"
+    bl_parent_id = "CYCLES_RENDER_PT_passes"
+
+    @classmethod
+    def poll(cls, context):
+        import _cycles
+        return _cycles.with_cycles_debug
+
+    def draw(self, context):
+        layout = self.layout
+        layout.use_property_split = True
+        layout.use_property_decorate = False
+
+        cycles_view_layer = context.view_layer.cycles
+
+        layout.prop(cycles_view_layer, "pass_debug_bvh_traversed_nodes")
+        layout.prop(cycles_view_layer, "pass_debug_bvh_traversed_instances")
+        layout.prop(cycles_view_layer, "pass_debug_bvh_intersections")
+        layout.prop(cycles_view_layer, "pass_debug_ray_bounces")
+
+
 class CYCLES_RENDER_PT_passes_aov(CyclesButtonsPanel, ViewLayerAOVPanel):
     bl_label = "Shader AOV"
     bl_context = "view_layer"
@@ -1113,7 +1134,7 @@ class CYCLES_PT_context_material(CyclesButtonsPanel, Panel):
             col = row.column(align=True)
             col.operator("object.material_slot_add", icon='ADD', text="")
             col.operator("object.material_slot_remove", icon='REMOVE', text="")
-            col.separator()
+
             col.menu("MATERIAL_MT_context_menu", icon='DOWNARROW_HLT', text="")
 
             if is_sortable:
@@ -1128,15 +1149,16 @@ class CYCLES_PT_context_material(CyclesButtonsPanel, Panel):
                 row.operator("object.material_slot_select", text="Select")
                 row.operator("object.material_slot_deselect", text="Deselect")
 
-        row = layout.row()
+        split = layout.split(factor=0.65)
 
         if ob:
-            row.template_ID(ob, "active_material", new="material.new")
+            split.template_ID(ob, "active_material", new="material.new")
+            row = split.row()
 
             if slot:
-                icon_link = 'MESH_DATA' if slot.link == 'DATA' else 'OBJECT_DATA'
-                row.prop(slot, "link", text="", icon=icon_link, icon_only=True)
-
+                row.prop(slot, "link", text="")
+            else:
+                row.label()
         elif mat:
             split.template_ID(space, "pin_id")
             split.separator()
@@ -1201,31 +1223,20 @@ class CYCLES_OBJECT_PT_shading(CyclesButtonsPanel, Panel):
 
     @classmethod
     def poll(cls, context):
-        if not CyclesButtonsPanel.poll(context):
-            return False
-
-        ob = context.object
-        return ob and has_geometry_visibility(ob)
-
-    def draw(self, context):
-        pass
-
-
-class CYCLES_OBJECT_PT_shading_shadow_terminator(CyclesButtonsPanel, Panel):
-    bl_label = "Shadow Terminator"
-    bl_parent_id = "CYCLES_OBJECT_PT_shading"
-    bl_context = "object"
+        return CyclesButtonsPanel.poll(context) and (context.object)
 
     def draw(self, context):
         layout = self.layout
         layout.use_property_split = True
 
-        flow = layout.grid_flow(row_major=False, columns=0, even_columns=True, even_rows=False, align=True)
-
+        flow = layout.grid_flow(row_major=False, columns=0, even_columns=True, even_rows=False, align=False)
+        layout = self.layout
         ob = context.object
         cob = ob.cycles
-        flow.prop(cob, "shadow_terminator_geometry_offset", text="Geometry Offset")
-        flow.prop(cob, "shadow_terminator_offset", text="Shading Offset")
+
+        if has_geometry_visibility(ob):
+            col = flow.column()
+            col.prop(cob, "shadow_terminator_offset")
 
 
 class CYCLES_OBJECT_PT_visibility(CyclesButtonsPanel, Panel):
@@ -2294,6 +2305,7 @@ classes = (
     CYCLES_RENDER_PT_passes_data,
     CYCLES_RENDER_PT_passes_light,
     CYCLES_RENDER_PT_passes_crypto,
+    CYCLES_RENDER_PT_passes_debug,
     CYCLES_RENDER_PT_passes_aov,
     CYCLES_RENDER_PT_filter,
     CYCLES_RENDER_PT_override,
@@ -2304,7 +2316,6 @@ classes = (
     CYCLES_PT_context_material,
     CYCLES_OBJECT_PT_motion_blur,
     CYCLES_OBJECT_PT_shading,
-    CYCLES_OBJECT_PT_shading_shadow_terminator,
     CYCLES_OBJECT_PT_visibility,
     CYCLES_OBJECT_PT_visibility_ray_visibility,
     CYCLES_OBJECT_PT_visibility_culling,

intern/cycles/blender/blender_image.cpp

@@ -137,9 +137,9 @@ bool BlenderImageLoader::load_pixels(const ImageMetaData &metadata,
       /* Premultiply, byte images are always straight for Blender. */
       unsigned char *cp = (unsigned char *)pixels;
       for (size_t i = 0; i < num_pixels; i++, cp += channels) {
-        cp[0] = (cp[0] * cp[3]) / 255;
-        cp[1] = (cp[1] * cp[3]) / 255;
-        cp[2] = (cp[2] * cp[3]) / 255;
+        cp[0] = (cp[0] * cp[3]) >> 8;
+        cp[1] = (cp[1] * cp[3]) >> 8;
+        cp[2] = (cp[2] * cp[3]) >> 8;
       }
     }
   }

intern/cycles/blender/blender_mesh.cpp

@@ -1055,45 +1055,10 @@ static BL::MeshSequenceCacheModifier object_mesh_cache_find(BL::Object &b_ob)
   return BL::MeshSequenceCacheModifier(PointerRNA_NULL);
 }
 
-/* Check whether some of "built-in" motion-related attributes are needed to be exported (includes
- * things like velocity from cache modifier, fluid simulation).
- *
- * NOTE: This code is run prior to object motion blur initialization. so can not access properties
- * set by `sync_object_motion_init()`. */
-static bool mesh_need_motion_attribute(BL::Object &b_ob, Scene *scene)
-{
-  const Scene::MotionType need_motion = scene->need_motion();
-  if (need_motion == Scene::MOTION_NONE) {
-    /* Simple case: neither motion pass nor motion blur is needed, no need in the motion related
-     * attributes. */
-    return false;
-  }
-
-  if (need_motion == Scene::MOTION_BLUR) {
-    /* A bit tricky and implicit case:
-     * - Motion blur is enabled in the scene, which implies specific number of time steps for
-     *   objects.
-     * - If the object has motion blur disabled on it, it will have 0 time steps.
-     * - Motion attribute expects non-zero time steps.
-     *
-     * Avoid adding motion attributes if the motion blur will enforce 0 motion steps. */
-    PointerRNA cobject = RNA_pointer_get(&b_ob.ptr, "cycles");
-    const bool use_motion = get_boolean(cobject, "use_motion_blur");
-    if (!use_motion) {
-      return false;
-    }
-  }
-
-  /* Motion pass which implies 3 motion steps, or motion blur which is not disabled on object
-   * level. */
-  return true;
-}
-
 static void sync_mesh_cached_velocities(BL::Object &b_ob, Scene *scene, Mesh *mesh)
 {
-  if (!mesh_need_motion_attribute(b_ob, scene)) {
+  if (scene->need_motion() == Scene::MOTION_NONE)
     return;
-  }
 
   BL::MeshSequenceCacheModifier b_mesh_cache = object_mesh_cache_find(b_ob);
 
@@ -1137,9 +1102,8 @@ static void sync_mesh_cached_velocities(BL::Object &b_ob, Scene *scene, Mesh *me
 
 static void sync_mesh_fluid_motion(BL::Object &b_ob, Scene *scene, Mesh *mesh)
 {
-  if (!mesh_need_motion_attribute(b_ob, scene)) {
+  if (scene->need_motion() == Scene::MOTION_NONE)
     return;
-  }
 
   BL::FluidDomainSettings b_fluid_domain = object_fluid_liquid_domain_find(b_ob);
 

intern/cycles/blender/blender_object.cpp

@@ -109,23 +109,23 @@ void BlenderSync::sync_object_motion_init(BL::Object &b_parent, BL::Object &b_ob
   }
 
   Geometry *geom = object->get_geometry();
+  geom->set_use_motion_blur(false);
+  geom->set_motion_steps(0);
 
-  int motion_steps = 0;
-  bool use_motion_blur = false;
+  uint motion_steps;
 
   if (need_motion == Scene::MOTION_BLUR) {
     motion_steps = object_motion_steps(b_parent, b_ob, Object::MAX_MOTION_STEPS);
+    geom->set_motion_steps(motion_steps);
     if (motion_steps && object_use_deform_motion(b_parent, b_ob)) {
-      use_motion_blur = true;
+      geom->set_use_motion_blur(true);
     }
   }
   else {
     motion_steps = 3;
+    geom->set_motion_steps(motion_steps);
   }
 
-  geom->set_use_motion_blur(use_motion_blur);
-  geom->set_motion_steps(motion_steps);
-
   motion.resize(motion_steps, transform_empty());
 
   if (motion_steps) {
@@ -290,12 +290,8 @@ Object *BlenderSync::sync_object(BL::Depsgraph &b_depsgraph,
   bool is_shadow_catcher = get_boolean(cobject, "is_shadow_catcher");
   object->set_is_shadow_catcher(is_shadow_catcher);
 
-  float shadow_terminator_shading_offset = get_float(cobject, "shadow_terminator_offset");
-  object->set_shadow_terminator_shading_offset(shadow_terminator_shading_offset);
-
-  float shadow_terminator_geometry_offset = get_float(cobject,
-                                                      "shadow_terminator_geometry_offset");
-  object->set_shadow_terminator_geometry_offset(shadow_terminator_geometry_offset);
+  float shadow_terminator_offset = get_float(cobject, "shadow_terminator_offset");
+  object->set_shadow_terminator_offset(shadow_terminator_offset);
 
   /* sync the asset name for Cryptomatte */
   BL::Object parent = b_ob.parent();

intern/cycles/blender/blender_python.cpp

@@ -1098,6 +1098,14 @@ void *CCL_python_module_init()
   PyModule_AddStringConstant(mod, "osl_version_string", "unknown");
 #endif
 
+#ifdef WITH_CYCLES_DEBUG
+  PyModule_AddObject(mod, "with_cycles_debug", Py_True);
+  Py_INCREF(Py_True);
+#else
+  PyModule_AddObject(mod, "with_cycles_debug", Py_False);
+  Py_INCREF(Py_False);
+#endif
+
 #ifdef WITH_NETWORK
   PyModule_AddObject(mod, "with_network", Py_True);
   Py_INCREF(Py_True);

intern/cycles/blender/blender_sync.cpp

@@ -404,6 +404,8 @@ void BlenderSync::sync_film(BL::SpaceView3D &b_v3d)
 
   Film *film = scene->film;
 
+  vector<Pass> prevpasses = scene->passes;
+
   if (b_v3d) {
     film->set_display_pass(update_viewport_display_passes(b_v3d, scene->passes));
   }
@@ -433,6 +435,11 @@ void BlenderSync::sync_film(BL::SpaceView3D &b_v3d)
         break;
     }
   }
+
+  if (!Pass::equals(prevpasses, scene->passes)) {
+    film->tag_passes_update(scene, prevpasses, false);
+    film->tag_modified();
+  }
 }
 
 /* Render Layer */
@@ -538,6 +545,12 @@ PassType BlenderSync::get_pass_type(BL::RenderPass &b_pass)
   MAP_PASS("BakePrimitive", PASS_BAKE_PRIMITIVE);
   MAP_PASS("BakeDifferential", PASS_BAKE_DIFFERENTIAL);
 
+#ifdef __KERNEL_DEBUG__
+  MAP_PASS("Debug BVH Traversed Nodes", PASS_BVH_TRAVERSED_NODES);
+  MAP_PASS("Debug BVH Traversed Instances", PASS_BVH_TRAVERSED_INSTANCES);
+  MAP_PASS("Debug BVH Intersections", PASS_BVH_INTERSECTIONS);
+  MAP_PASS("Debug Ray Bounces", PASS_RAY_BOUNCES);
+#endif
   MAP_PASS("Debug Render Time", PASS_RENDER_TIME);
   MAP_PASS("AdaptiveAuxBuffer", PASS_ADAPTIVE_AUX_BUFFER);
   MAP_PASS("Debug Sample Count", PASS_SAMPLE_COUNT);
@@ -590,10 +603,8 @@ vector<Pass> BlenderSync::sync_render_passes(BL::Scene &b_scene,
   for (BL::RenderPass &b_pass : b_rlay.passes) {
     PassType pass_type = get_pass_type(b_pass);
 
-    if (pass_type == PASS_MOTION &&
-        (b_view_layer.use_motion_blur() && b_scene.render().use_motion_blur())) {
+    if (pass_type == PASS_MOTION && view_layer.use_motion_blur)
       continue;
-    }
     if (pass_type != PASS_NONE)
       Pass::add(pass_type, passes, b_pass.name().c_str());
   }
@@ -635,6 +646,24 @@ vector<Pass> BlenderSync::sync_render_passes(BL::Scene &b_scene,
     }
   }
 
+#ifdef __KERNEL_DEBUG__
+  if (get_boolean(crl, "pass_debug_bvh_traversed_nodes")) {
+    b_engine.add_pass("Debug BVH Traversed Nodes", 1, "X", b_view_layer.name().c_str());
+    Pass::add(PASS_BVH_TRAVERSED_NODES, passes, "Debug BVH Traversed Nodes");
+  }
+  if (get_boolean(crl, "pass_debug_bvh_traversed_instances")) {
+    b_engine.add_pass("Debug BVH Traversed Instances", 1, "X", b_view_layer.name().c_str());
+    Pass::add(PASS_BVH_TRAVERSED_INSTANCES, passes, "Debug BVH Traversed Instances");
+  }
+  if (get_boolean(crl, "pass_debug_bvh_intersections")) {
+    b_engine.add_pass("Debug BVH Intersections", 1, "X", b_view_layer.name().c_str());
+    Pass::add(PASS_BVH_INTERSECTIONS, passes, "Debug BVH Intersections");
+  }
+  if (get_boolean(crl, "pass_debug_ray_bounces")) {
+    b_engine.add_pass("Debug Ray Bounces", 1, "X", b_view_layer.name().c_str());
+    Pass::add(PASS_RAY_BOUNCES, passes, "Debug Ray Bounces");
+  }
+#endif
   if (get_boolean(crl, "pass_debug_render_time")) {
     b_engine.add_pass("Debug Render Time", 1, "X", b_view_layer.name().c_str());
     Pass::add(PASS_RENDER_TIME, passes, "Debug Render Time");
@@ -718,13 +747,10 @@ vector<Pass> BlenderSync::sync_render_passes(BL::Scene &b_scene,
                                         DENOISING_CLEAN_ALL_PASSES);
   scene->film->set_denoising_prefiltered_pass(denoising.store_passes &&
                                               denoising.type == DENOISER_NLM);
-  scene->film->set_pass_alpha_threshold(b_view_layer.pass_alpha_threshold());
 
-  if (!Pass::equals(passes, scene->passes)) {
-    scene->film->tag_passes_update(scene, passes);
-    scene->film->tag_modified();
-    scene->integrator->tag_update(scene, Integrator::UPDATE_ALL);
-  }
+  scene->film->set_pass_alpha_threshold(b_view_layer.pass_alpha_threshold());
+  scene->film->tag_passes_update(scene, passes);
+  scene->integrator->tag_update(scene, Integrator::UPDATE_ALL);
 
   return passes;
 }

intern/cycles/blender/blender_viewport.cpp

@@ -40,8 +40,8 @@ BlenderViewportParameters::BlenderViewportParameters(BL::SpaceView3D &b_v3d)
   BL::View3DShading shading = b_v3d.shading();
   PointerRNA cshading = RNA_pointer_get(&shading.ptr, "cycles");
 
-  /* We only copy the shading parameters if we are in look-dev mode.
-   * Otherwise defaults are being used. These defaults mimic normal render settings. */
+  /* We only copy the shading parameters if we are in look dev mode. otherwise
+   * defaults are being used. These defaults mimic normal render settings */
   if (shading.type() == BL::View3DShading::type_RENDERED) {
     use_scene_world = shading.use_scene_world_render();
     use_scene_lights = shading.use_scene_lights_render();

intern/cycles/device/cuda/device_cuda_impl.cpp

@@ -351,6 +351,9 @@ string CUDADevice::compile_kernel_get_common_cflags(
   if (extra_cflags) {
     cflags += string(" ") + string(extra_cflags);
   }
+#  ifdef WITH_CYCLES_DEBUG
+  cflags += " -D__KERNEL_DEBUG__";
+#  endif
 
   if (split) {
     cflags += " -D__SPLIT__";
@@ -458,19 +461,18 @@ string CUDADevice::compile_kernel(const DeviceRequestedFeatures &requested_featu
 
   const int nvcc_cuda_version = cuewCompilerVersion();
   VLOG(1) << "Found nvcc " << nvcc << ", CUDA version " << nvcc_cuda_version << ".";
-  if (nvcc_cuda_version < 101) {
+  if (nvcc_cuda_version < 80) {
     printf(
         "Unsupported CUDA version %d.%d detected, "
-        "you need CUDA 10.1 or newer.\n",
+        "you need CUDA 8.0 or newer.\n",
         nvcc_cuda_version / 10,
         nvcc_cuda_version % 10);
     return string();
   }
-  else if (!(nvcc_cuda_version == 101 || nvcc_cuda_version == 102 || nvcc_cuda_version == 111 ||
-             nvcc_cuda_version == 112 || nvcc_cuda_version == 113 || nvcc_cuda_version == 114)) {
+  else if (!(nvcc_cuda_version == 101 || nvcc_cuda_version == 102)) {
     printf(
         "CUDA version %d.%d detected, build may succeed but only "
-        "CUDA 10.1 to 11.4 are officially supported.\n",
+        "CUDA 10.1 and 10.2 are officially supported.\n",
         nvcc_cuda_version / 10,
         nvcc_cuda_version % 10);
   }

intern/cycles/device/device_cpu.cpp

@@ -57,8 +57,8 @@
 #include "util/util_function.h"
 #include "util/util_logging.h"
 #include "util/util_map.h"
-#include "util/util_opengl.h"
 #include "util/util_openimagedenoise.h"
+#include "util/util_opengl.h"
 #include "util/util_optimization.h"
 #include "util/util_progress.h"
 #include "util/util_system.h"

intern/cycles/device/device_optix.cpp

@@ -233,6 +233,9 @@ class OptiXDevice : public CUDADevice {
               break;
           }
         };
+#  endif
+#  if OPTIX_ABI_VERSION >= 41 && defined(WITH_CYCLES_DEBUG)
+    options.validationMode = OPTIX_DEVICE_CONTEXT_VALIDATION_MODE_ALL;
 #  endif
     check_result_optix(optixDeviceContextCreate(cuContext, &options, &context));
 #  ifdef WITH_CYCLES_LOGGING
@@ -366,8 +369,13 @@ class OptiXDevice : public CUDADevice {
 
     OptixModuleCompileOptions module_options = {};
     module_options.maxRegisterCount = 0;  // Do not set an explicit register limit
+#  ifdef WITH_CYCLES_DEBUG
+    module_options.optLevel = OPTIX_COMPILE_OPTIMIZATION_LEVEL_0;
+    module_options.debugLevel = OPTIX_COMPILE_DEBUG_LEVEL_FULL;
+#  else
     module_options.optLevel = OPTIX_COMPILE_OPTIMIZATION_LEVEL_3;
     module_options.debugLevel = OPTIX_COMPILE_DEBUG_LEVEL_LINEINFO;
+#  endif
 
 #  if OPTIX_ABI_VERSION >= 41
     module_options.boundValues = nullptr;
@@ -570,7 +578,11 @@ class OptiXDevice : public CUDADevice {
 
     OptixPipelineLinkOptions link_options = {};
     link_options.maxTraceDepth = 1;
+#  ifdef WITH_CYCLES_DEBUG
+    link_options.debugLevel = OPTIX_COMPILE_DEBUG_LEVEL_FULL;
+#  else
     link_options.debugLevel = OPTIX_COMPILE_DEBUG_LEVEL_LINEINFO;
+#  endif
 #  if OPTIX_ABI_VERSION < 24
     link_options.overrideUsesMotionBlur = motion_blur;
 #  endif

intern/cycles/device/opencl/device_opencl_impl.cpp

@@ -1968,6 +1968,10 @@ string OpenCLDevice::kernel_build_options(const string *debug_src)
     build_options += "-D__KERNEL_OPENCL_DEBUG__ ";
   }
 
+#  ifdef WITH_CYCLES_DEBUG
+  build_options += "-D__KERNEL_DEBUG__ ";
+#  endif
+
 #  ifdef WITH_NANOVDB
   if (info.has_nanovdb) {
     build_options += "-DWITH_NANOVDB ";

intern/cycles/kernel/CMakeLists.txt

@@ -380,16 +380,11 @@ if(WITH_CYCLES_CUDA_BINARIES)
   set(CUDA_VERSION "${CUDA_VERSION_MAJOR}${CUDA_VERSION_MINOR}")
 
   # warn for other versions
-  if((CUDA_VERSION MATCHES "101") OR
-     (CUDA_VERSION MATCHES "102") OR
-     (CUDA_VERSION MATCHES "111") OR
-     (CUDA_VERSION MATCHES "112") OR
-     (CUDA_VERSION MATCHES "113") OR
-     (CUDA_VERSION MATCHES "114"))
+  if((CUDA_VERSION MATCHES "101") OR (CUDA_VERSION MATCHES "102") OR (CUDA_VERSION MATCHES "111"))
   else()
     message(WARNING
       "CUDA version ${CUDA_VERSION_MAJOR}.${CUDA_VERSION_MINOR} detected, "
-      "build may succeed but only CUDA 10.1 to 11.4 are officially supported")
+      "build may succeed but only CUDA 10.1, 10.2 and 11.1 are officially supported")
   endif()
 
   # build for each arch
@@ -444,6 +439,10 @@ if(WITH_CYCLES_CUDA_BINARIES)
       set(name ${name}_experimental)
     endif()
 
+    if(WITH_CYCLES_DEBUG)
+      set(cuda_flags ${cuda_flags} -D __KERNEL_DEBUG__)
+    endif()
+
     if(WITH_NANOVDB)
       set(cuda_flags ${cuda_flags}
         -D WITH_NANOVDB
@@ -558,6 +557,11 @@ if(WITH_CYCLES_DEVICE_OPTIX AND WITH_CYCLES_CUDA_BINARIES)
       --use_fast_math
       -o ${output})
 
+    if(WITH_CYCLES_DEBUG)
+      set(cuda_flags ${cuda_flags}
+        -D __KERNEL_DEBUG__)
+    endif()
+
     if(WITH_NANOVDB)
       set(cuda_flags ${cuda_flags}
         -D WITH_NANOVDB

intern/cycles/kernel/bvh/bvh_traversal.h

@@ -67,6 +67,8 @@ ccl_device_noinline bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg,
   isect->prim = PRIM_NONE;
   isect->object = OBJECT_NONE;
 
+  BVH_DEBUG_INIT();
+
   /* traversal loop */
   do {
     do {
@@ -116,6 +118,7 @@ ccl_device_noinline bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg,
             --stack_ptr;
           }
         }
+        BVH_DEBUG_NEXT_NODE();
       }
 
       /* if node is leaf, fetch triangle list */
@@ -135,6 +138,7 @@ ccl_device_noinline bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg,
           switch (type & PRIMITIVE_ALL) {
             case PRIMITIVE_TRIANGLE: {
               for (; prim_addr < prim_addr2; prim_addr++) {
+                BVH_DEBUG_NEXT_INTERSECTION();
                 kernel_assert(kernel_tex_fetch(__prim_type, prim_addr) == type);
                 if (triangle_intersect(kg, isect, P, dir, visibility, object, prim_addr)) {
                   /* shadow ray early termination */
@@ -147,6 +151,7 @@ ccl_device_noinline bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg,
 #if BVH_FEATURE(BVH_MOTION)
             case PRIMITIVE_MOTION_TRIANGLE: {
               for (; prim_addr < prim_addr2; prim_addr++) {
+                BVH_DEBUG_NEXT_INTERSECTION();
                 kernel_assert(kernel_tex_fetch(__prim_type, prim_addr) == type);
                 if (motion_triangle_intersect(
                         kg, isect, P, dir, ray->time, visibility, object, prim_addr)) {
@@ -164,6 +169,7 @@ ccl_device_noinline bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg,
             case PRIMITIVE_CURVE_RIBBON:
             case PRIMITIVE_MOTION_CURVE_RIBBON: {
               for (; prim_addr < prim_addr2; prim_addr++) {
+                BVH_DEBUG_NEXT_INTERSECTION();
                 const uint curve_type = kernel_tex_fetch(__prim_type, prim_addr);
                 kernel_assert((curve_type & PRIMITIVE_ALL) == (type & PRIMITIVE_ALL));
                 const bool hit = curve_intersect(
@@ -195,6 +201,8 @@ ccl_device_noinline bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg,
           traversal_stack[stack_ptr] = ENTRYPOINT_SENTINEL;
 
           node_addr = kernel_tex_fetch(__object_node, object);
+
+          BVH_DEBUG_NEXT_INSTANCE();
         }
       }
     } while (node_addr != ENTRYPOINT_SENTINEL);

intern/cycles/kernel/bvh/bvh_types.h

@@ -42,6 +42,33 @@ CCL_NAMESPACE_BEGIN
 
 #define BVH_FEATURE(f) (((BVH_FUNCTION_FEATURES) & (f)) != 0)
 
+/* Debugging helpers. */
+#ifdef __KERNEL_DEBUG__
+#  define BVH_DEBUG_INIT() \
+    do { \
+      isect->num_traversed_nodes = 0; \
+      isect->num_traversed_instances = 0; \
+      isect->num_intersections = 0; \
+    } while (0)
+#  define BVH_DEBUG_NEXT_NODE() \
+    do { \
+      ++isect->num_traversed_nodes; \
+    } while (0)
+#  define BVH_DEBUG_NEXT_INTERSECTION() \
+    do { \
+      ++isect->num_intersections; \
+    } while (0)
+#  define BVH_DEBUG_NEXT_INSTANCE() \
+    do { \
+      ++isect->num_traversed_instances; \
+    } while (0)
+#else /* __KERNEL_DEBUG__ */
+#  define BVH_DEBUG_INIT()
+#  define BVH_DEBUG_NEXT_NODE()
+#  define BVH_DEBUG_NEXT_INTERSECTION()
+#  define BVH_DEBUG_NEXT_INSTANCE()
+#endif /* __KERNEL_DEBUG__ */
+
 CCL_NAMESPACE_END
 
 #endif /* __BVH_TYPES__ */

intern/cycles/kernel/bvh/bvh_util.h

@@ -71,86 +71,6 @@ ccl_device_inline float3 ray_offset(float3 P, float3 Ng)
 #endif
 }
 
-/* This function should be used to compute a modified ray start position for
- * rays leaving from a surface. The algorithm slightly distorts flat surface
- * of a triangle. Surface is lifted by amount h along normal n in the incident
- * point. */
-
-ccl_device_inline float3 smooth_surface_offset(KernelGlobals *kg, ShaderData *sd, float3 Ng)
-{
-  float3 V[3], N[3];
-  triangle_vertices_and_normals(kg, sd->prim, V, N);
-
-  const float u = sd->u, v = sd->v;
-  const float w = 1 - u - v;
-  float3 P = V[0] * u + V[1] * v + V[2] * w; /* Local space */
-  float3 n = N[0] * u + N[1] * v + N[2] * w; /* We get away without normalization */
-
-  object_normal_transform(kg, sd, &n); /* Normal x scale, world space */
-
-  /* Parabolic approximation */
-  float a = dot(N[2] - N[0], V[0] - V[2]);
-  float b = dot(N[2] - N[1], V[1] - V[2]);
-  float c = dot(N[1] - N[0], V[1] - V[0]);
-  float h = a * u * (u - 1) + (a + b + c) * u * v + b * v * (v - 1);
-
-  /* Check flipped normals */
-  if (dot(n, Ng) > 0) {
-    /* Local linear envelope */
-    float h0 = max(max(dot(V[1] - V[0], N[0]), dot(V[2] - V[0], N[0])), 0.0f);
-    float h1 = max(max(dot(V[0] - V[1], N[1]), dot(V[2] - V[1], N[1])), 0.0f);
-    float h2 = max(max(dot(V[0] - V[2], N[2]), dot(V[1] - V[2], N[2])), 0.0f);
-    h0 = max(dot(V[0] - P, N[0]) + h0, 0.0f);
-    h1 = max(dot(V[1] - P, N[1]) + h1, 0.0f);
-    h2 = max(dot(V[2] - P, N[2]) + h2, 0.0f);
-    h = max(min(min(h0, h1), h2), h * 0.5f);
-  }
-  else {
-    float h0 = max(max(dot(V[0] - V[1], N[0]), dot(V[0] - V[2], N[0])), 0.0f);
-    float h1 = max(max(dot(V[1] - V[0], N[1]), dot(V[1] - V[2], N[1])), 0.0f);
-    float h2 = max(max(dot(V[2] - V[0], N[2]), dot(V[2] - V[1], N[2])), 0.0f);
-    h0 = max(dot(P - V[0], N[0]) + h0, 0.0f);
-    h1 = max(dot(P - V[1], N[1]) + h1, 0.0f);
-    h2 = max(dot(P - V[2], N[2]) + h2, 0.0f);
-    h = min(-min(min(h0, h1), h2), h * 0.5f);
-  }
-
-  return n * h;
-}
-
-/* Ray offset to avoid shadow terminator artifact. */
-
-ccl_device_inline float3 ray_offset_shadow(KernelGlobals *kg, ShaderData *sd, float3 L)
-{
-  float NL = dot(sd->N, L);
-  bool transmit = (NL < 0.0f);
-  float3 Ng = (transmit ? -sd->Ng : sd->Ng);
-  float3 P = ray_offset(sd->P, Ng);
-
-  if ((sd->type & PRIMITIVE_ALL_TRIANGLE) && (sd->shader & SHADER_SMOOTH_NORMAL)) {
-    const float offset_cutoff =
-        kernel_tex_fetch(__objects, sd->object).shadow_terminator_geometry_offset;
-    /* Do ray offset (heavy stuff) only for close to be terminated triangles:
-     * offset_cutoff = 0.1f means that 10-20% of rays will be affected. Also
-     * make a smooth transition near the threshold. */
-    if (offset_cutoff > 0.0f) {
-      float NgL = dot(Ng, L);
-      float offset_amount = 0.0f;
-      if (NL < offset_cutoff) {
-        offset_amount = clamp(2.0f - (NgL + NL) / offset_cutoff, 0.0f, 1.0f);
-      }
-      else {
-        offset_amount = clamp(1.0f - NgL / offset_cutoff, 0.0f, 1.0f);
-      }
-      if (offset_amount > 0.0f) {
-        P += smooth_surface_offset(kg, sd, Ng) * offset_amount;
-      }
-    }
-  }
-
-  return P;
-}
-
 #if defined(__VOLUME_RECORD_ALL__) || (defined(__SHADOW_RECORD_ALL__) && defined(__KERNEL_CPU__))
 /* ToDo: Move to another file? */
 ccl_device int intersections_compare(const void *a, const void *b)

intern/cycles/kernel/closure/bsdf.h

@@ -462,7 +462,7 @@ ccl_device_inline int bsdf_sample(KernelGlobals *kg,
   else {
     /* Shadow terminator offset. */
     const float frequency_multiplier =
-        kernel_tex_fetch(__objects, sd->object).shadow_terminator_shading_offset;
+        kernel_tex_fetch(__objects, sd->object).shadow_terminator_offset;
     if (frequency_multiplier > 1.0f) {
       *eval *= shift_cos_in(dot(*omega_in, sc->N), frequency_multiplier);
     }
@@ -488,9 +488,12 @@ ccl_device_inline
               const float3 omega_in,
               float *pdf)
 {
+  /* For curves use the smooth normal, particularly for ribbons the geometric
+   * normal gives too much darkening otherwise. */
+  const float3 Ng = (sd->type & PRIMITIVE_ALL_CURVE) ? sd->N : sd->Ng;
   float3 eval;
 
-  if (dot(sd->N, omega_in) >= 0.0f) {
+  if (dot(Ng, omega_in) >= 0.0f) {
     switch (sc->type) {
       case CLOSURE_BSDF_DIFFUSE_ID:
       case CLOSURE_BSDF_BSSRDF_ID:
@@ -586,7 +589,7 @@ ccl_device_inline
     }
     /* Shadow terminator offset. */
     const float frequency_multiplier =
-        kernel_tex_fetch(__objects, sd->object).shadow_terminator_shading_offset;
+        kernel_tex_fetch(__objects, sd->object).shadow_terminator_offset;
     if (frequency_multiplier > 1.0f) {
       eval *= shift_cos_in(dot(omega_in, sc->N), frequency_multiplier);
     }

intern/cycles/kernel/closure/bssrdf.h

@@ -432,9 +432,9 @@ ccl_device void bssrdf_sample(const ShaderClosure *sc, float xi, float *r, float
   xi *= bssrdf->channels;
 
   if (xi < 1.0f) {
-    radius = (bssrdf->radius.x > 0.0f) ? bssrdf->radius.x :
-             (bssrdf->radius.y > 0.0f) ? bssrdf->radius.y :
-                                         bssrdf->radius.z;
+    radius = (bssrdf->radius.x > 0.0f) ?
+                 bssrdf->radius.x :
+                 (bssrdf->radius.y > 0.0f) ? bssrdf->radius.y : bssrdf->radius.z;
   }
   else if (xi < 2.0f) {
     xi -= 1.0f;

intern/cycles/kernel/filter/filter_defines.h

@@ -52,14 +52,15 @@ typedef struct TileInfo {
         tile_buffer_6, tile_buffer_7, tile_buffer_8, tile_buffer_9
 #  define ccl_get_tile_buffer(id) \
     (id == 0 ? tile_buffer_1 : \
-     id == 1 ? tile_buffer_2 : \
-     id == 2 ? tile_buffer_3 : \
-     id == 3 ? tile_buffer_4 : \
-     id == 4 ? tile_buffer_5 : \
-     id == 5 ? tile_buffer_6 : \
-     id == 6 ? tile_buffer_7 : \
-     id == 7 ? tile_buffer_8 : \
-               tile_buffer_9)
+               id == 1 ? \
+               tile_buffer_2 : \
+               id == 2 ? \
+               tile_buffer_3 : \
+               id == 3 ? tile_buffer_4 : \
+                         id == 4 ? tile_buffer_5 : \
+                                   id == 5 ? tile_buffer_6 : \
+                                             id == 6 ? tile_buffer_7 : \
+                                                       id == 7 ? tile_buffer_8 : tile_buffer_9)
 #else
 #  ifdef __KERNEL_CUDA__
 #    define CCL_FILTER_TILE_INFO ccl_global TileInfo *tile_info

intern/cycles/kernel/geom/geom_curve_intersect.h

@@ -237,7 +237,7 @@ ccl_device bool curve_intersect_iterative(const float3 ray_dir,
         return false; /* Rejects NaNs */
       }
 
-      /* Back-face culling. */
+      /* Backface culling. */
       const float3 R = normalize(Q - P);
       const float3 U = dradiusdu * R + dPdu;
       const float3 V = cross(dPdu, R);
@@ -458,12 +458,10 @@ ccl_device_inline bool cylinder_culling_test(const float2 p1, const float2 p2, c
   return num * num <= r * r * den2;
 }
 
-/**
- * Intersects a ray with a quad with back-face culling
- * enabled. The quad v0,v1,v2,v3 is split into two triangles
- * v0,v1,v3 and v2,v3,v1. The edge v1,v2 decides which of the two
- * triangles gets intersected.
- */
+/*! Intersects a ray with a quad with backface culling
+ *  enabled. The quad v0,v1,v2,v3 is split into two triangles
+ *  v0,v1,v3 and v2,v3,v1. The edge v1,v2 decides which of the two
+ *  triangles gets intersected. */
 ccl_device_inline bool ribbon_intersect_quad(const float ray_tfar,
                                              const float3 quad_v0,
                                              const float3 quad_v1,

intern/cycles/kernel/geom/geom_primitive.h

@@ -138,7 +138,7 @@ ccl_device_inline float4 primitive_surface_attribute_float4(KernelGlobals *kg,
 ccl_device_inline bool primitive_is_volume_attribute(const ShaderData *sd,
                                                      const AttributeDescriptor desc)
 {
-  return sd->type == PRIMITIVE_VOLUME;
+  return (sd->object != OBJECT_NONE && desc.element == ATTR_ELEMENT_VOXEL);
 }
 
 ccl_device_inline float primitive_volume_attribute_float(KernelGlobals *kg,

intern/cycles/kernel/geom/geom_triangle.h

@@ -75,22 +75,6 @@ ccl_device_inline void triangle_vertices(KernelGlobals *kg, int prim, float3 P[3
   P[2] = float4_to_float3(kernel_tex_fetch(__prim_tri_verts, tri_vindex.w + 2));
 }
 
-/* Triangle vertex locations and vertex normals */
-
-ccl_device_inline void triangle_vertices_and_normals(KernelGlobals *kg,
-                                                     int prim,
-                                                     float3 P[3],
-                                                     float3 N[3])
-{
-  const uint4 tri_vindex = kernel_tex_fetch(__tri_vindex, prim);
-  P[0] = float4_to_float3(kernel_tex_fetch(__prim_tri_verts, tri_vindex.w + 0));
-  P[1] = float4_to_float3(kernel_tex_fetch(__prim_tri_verts, tri_vindex.w + 1));
-  P[2] = float4_to_float3(kernel_tex_fetch(__prim_tri_verts, tri_vindex.w + 2));
-  N[0] = float4_to_float3(kernel_tex_fetch(__tri_vnormal, tri_vindex.x));
-  N[1] = float4_to_float3(kernel_tex_fetch(__tri_vnormal, tri_vindex.y));
-  N[2] = float4_to_float3(kernel_tex_fetch(__tri_vnormal, tri_vindex.z));
-}
-
 /* Interpolate smooth vertex normal from vertices */
 
 ccl_device_inline float3

intern/cycles/kernel/geom/geom_volume.h

@@ -72,22 +72,14 @@ ccl_device float4 volume_attribute_float4(KernelGlobals *kg,
                                           const ShaderData *sd,
                                           const AttributeDescriptor desc)
 {
-  if (desc.element & (ATTR_ELEMENT_OBJECT | ATTR_ELEMENT_MESH)) {
-    return kernel_tex_fetch(__attributes_float3, desc.offset);
-  }
-  else if (desc.element == ATTR_ELEMENT_VOXEL) {
-    /* todo: optimize this so we don't have to transform both here and in
-     * kernel_tex_image_interp_3d when possible. Also could optimize for the
-     * common case where transform is translation/scale only. */
-    float3 P = sd->P;
-    object_inverse_position_transform(kg, sd, &P);
-    InterpolationType interp = (sd->flag & SD_VOLUME_CUBIC) ? INTERPOLATION_CUBIC :
-                                                              INTERPOLATION_NONE;
-    return kernel_tex_image_interp_3d(kg, desc.offset, P, interp);
-  }
-  else {
-    return make_float4(0.0f, 0.0f, 0.0f, 0.0f);
-  }
+  /* todo: optimize this so we don't have to transform both here and in
+   * kernel_tex_image_interp_3d when possible. Also could optimize for the
+   * common case where transform is translation/scale only. */
+  float3 P = sd->P;
+  object_inverse_position_transform(kg, sd, &P);
+  InterpolationType interp = (sd->flag & SD_VOLUME_CUBIC) ? INTERPOLATION_CUBIC :
+                                                            INTERPOLATION_NONE;
+  return kernel_tex_image_interp_3d(kg, desc.offset, P, interp);
 }
 
 #endif

intern/cycles/kernel/kernel_accumulate.h

@@ -225,6 +225,13 @@ ccl_device_inline void path_radiance_init(KernelGlobals *kg, PathRadiance *L)
   L->denoising_albedo = zero_float3();
   L->denoising_depth = 0.0f;
 #endif
+
+#ifdef __KERNEL_DEBUG__
+  L->debug_data.num_bvh_traversed_nodes = 0;
+  L->debug_data.num_bvh_traversed_instances = 0;
+  L->debug_data.num_bvh_intersections = 0;
+  L->debug_data.num_ray_bounces = 0;
+#endif
 }
 
 ccl_device_inline void path_radiance_bsdf_bounce(KernelGlobals *kg,
@@ -588,9 +595,7 @@ ccl_device_inline void path_radiance_sum_shadowcatcher(KernelGlobals *kg,
   float shadow;
 
   if (UNLIKELY(!isfinite_safe(path_total))) {
-#  ifdef __KERNEL_DEBUG_NAN__
     kernel_assert(!"Non-finite total radiance along the path");
-#  endif
     shadow = 0.0f;
   }
   else if (path_total == 0.0f) {
@@ -636,9 +641,7 @@ ccl_device_inline float3 path_radiance_clamp_and_sum(KernelGlobals *kg,
 
     /* Reject invalid value */
     if (!isfinite_safe(sum)) {
-#  ifdef __KERNEL_DEBUG_NAN__
       kernel_assert(!"Non-finite sum in path_radiance_clamp_and_sum!");
-#  endif
       L_sum = zero_float3();
 
       L->direct_diffuse = zero_float3();
@@ -664,9 +667,7 @@ ccl_device_inline float3 path_radiance_clamp_and_sum(KernelGlobals *kg,
     /* Reject invalid value */
     float sum = fabsf((L_sum).x) + fabsf((L_sum).y) + fabsf((L_sum).z);
     if (!isfinite_safe(sum)) {
-#ifdef __KERNEL_DEBUG_NAN__
       kernel_assert(!"Non-finite final sum in path_radiance_clamp_and_sum!");
-#endif
       L_sum = zero_float3();
     }
   }

intern/cycles/kernel/kernel_bake.h

@@ -40,7 +40,7 @@ ccl_device_noinline void compute_light_pass(
   /* Evaluate surface shader. */
   shader_eval_surface(kg, sd, &state, NULL, state.flag);
 
-  /* TODO: disable more closures we don't need besides transparent. */
+  /* TODO, disable more closures we don't need besides transparent */
   shader_bsdf_disable_transparency(kg, sd);
 
   /* Init ray. */

intern/cycles/kernel/kernel_emission.h

@@ -176,7 +176,8 @@ ccl_device_noinline_cpu bool direct_emission(KernelGlobals *kg,
 
   if (ls->shader & SHADER_CAST_SHADOW) {
     /* setup ray */
-    ray->P = ray_offset_shadow(kg, sd, ls->D);
+    bool transmit = (dot(sd->Ng, ls->D) < 0.0f);
+    ray->P = ray_offset(sd->P, (transmit) ? -sd->Ng : sd->Ng);
 
     if (ls->t == FLT_MAX) {
       /* distant light */

intern/cycles/kernel/kernel_passes.h

@@ -122,6 +122,31 @@ ccl_device_inline void kernel_update_denoising_features(KernelGlobals *kg,
 }
 #endif /* __DENOISING_FEATURES__ */
 
+#ifdef __KERNEL_DEBUG__
+ccl_device_inline void kernel_write_debug_passes(KernelGlobals *kg,
+                                                 ccl_global float *buffer,
+                                                 PathRadiance *L)
+{
+  int flag = kernel_data.film.pass_flag;
+  if (flag & PASSMASK(BVH_TRAVERSED_NODES)) {
+    kernel_write_pass_float(buffer + kernel_data.film.pass_bvh_traversed_nodes,
+                            L->debug_data.num_bvh_traversed_nodes);
+  }
+  if (flag & PASSMASK(BVH_TRAVERSED_INSTANCES)) {
+    kernel_write_pass_float(buffer + kernel_data.film.pass_bvh_traversed_instances,
+                            L->debug_data.num_bvh_traversed_instances);
+  }
+  if (flag & PASSMASK(BVH_INTERSECTIONS)) {
+    kernel_write_pass_float(buffer + kernel_data.film.pass_bvh_intersections,
+                            L->debug_data.num_bvh_intersections);
+  }
+  if (flag & PASSMASK(RAY_BOUNCES)) {
+    kernel_write_pass_float(buffer + kernel_data.film.pass_ray_bounces,
+                            L->debug_data.num_ray_bounces);
+  }
+}
+#endif /* __KERNEL_DEBUG__ */
+
 #ifdef __KERNEL_CPU__
 #  define WRITE_ID_SLOT(buffer, depth, id, matte_weight, name) \
     kernel_write_id_pass_cpu(buffer, depth * 2, id, matte_weight, kg->coverage_##name)
@@ -364,6 +389,10 @@ ccl_device_inline void kernel_write_result(KernelGlobals *kg,
   }
 #endif /* __DENOISING_FEATURES__ */
 
+#ifdef __KERNEL_DEBUG__
+  kernel_write_debug_passes(kg, buffer, L);
+#endif
+
   /* Adaptive Sampling. Fill the additional buffer with the odd samples and calculate our stopping
      criteria. This is the heuristic from "A hierarchical automatic stopping condition for Monte
      Carlo global illumination" except that here it is applied per pixel and not in hierarchical

intern/cycles/kernel/kernel_path.h

@@ -70,6 +70,15 @@ ccl_device_forceinline bool kernel_path_scene_intersect(KernelGlobals *kg,
 
   bool hit = scene_intersect(kg, ray, visibility, isect);
 
+#ifdef __KERNEL_DEBUG__
+  if (state->flag & PATH_RAY_CAMERA) {
+    L->debug_data.num_bvh_traversed_nodes += isect->num_traversed_nodes;
+    L->debug_data.num_bvh_traversed_instances += isect->num_traversed_instances;
+    L->debug_data.num_bvh_intersections += isect->num_intersections;
+  }
+  L->debug_data.num_ray_bounces++;
+#endif /* __KERNEL_DEBUG__ */
+
   return hit;
 }
 

intern/cycles/kernel/kernel_shader.h

@@ -1268,7 +1268,6 @@ ccl_device_inline void shader_eval_volume(KernelGlobals *kg,
   sd->num_closure_left = max_closures;
   sd->flag = 0;
   sd->object_flag = 0;
-  sd->type = PRIMITIVE_VOLUME;
 
   for (int i = 0; stack[i].shader != SHADER_NONE; i++) {
     /* setup shaderdata from stack. it's mostly setup already in

intern/cycles/kernel/kernel_types.h

@@ -182,8 +182,9 @@ CCL_NAMESPACE_BEGIN
 #  undef __SHADER_RAYTRACE__
 #endif
 
-#ifdef WITH_CYCLES_DEBUG_NAN
-#  define __KERNEL_DEBUG_NAN__
+/* Features that enable others */
+#ifdef WITH_CYCLES_DEBUG
+#  define __KERNEL_DEBUG__
 #endif
 
 #if defined(__SUBSURFACE__) || defined(__SHADER_RAYTRACE__)
@@ -355,6 +356,12 @@ typedef enum PassType {
   PASS_MATERIAL_ID,
   PASS_MOTION,
   PASS_MOTION_WEIGHT,
+#ifdef __KERNEL_DEBUG__
+  PASS_BVH_TRAVERSED_NODES,
+  PASS_BVH_TRAVERSED_INSTANCES,
+  PASS_BVH_INTERSECTIONS,
+  PASS_RAY_BOUNCES,
+#endif
   PASS_RENDER_TIME,
   PASS_CRYPTOMATTE,
   PASS_AOV_COLOR,
@@ -458,6 +465,18 @@ typedef enum DenoiseFlag {
   DENOISING_CLEAN_ALL_PASSES = (1 << 6) - 1,
 } DenoiseFlag;
 
+#ifdef __KERNEL_DEBUG__
+/* NOTE: This is a runtime-only struct, alignment is not
+ * really important here.
+ */
+typedef struct DebugData {
+  int num_bvh_traversed_nodes;
+  int num_bvh_traversed_instances;
+  int num_bvh_intersections;
+  int num_ray_bounces;
+} DebugData;
+#endif
+
 typedef ccl_addr_space struct PathRadianceState {
 #ifdef __PASSES__
   float3 diffuse;
@@ -533,6 +552,10 @@ typedef ccl_addr_space struct PathRadiance {
   float3 denoising_albedo;
   float denoising_depth;
 #endif /* __DENOISING_FEATURES__ */
+
+#ifdef __KERNEL_DEBUG__
+  DebugData debug_data;
+#endif /* __KERNEL_DEBUG__ */
 } PathRadiance;
 
 typedef struct BsdfEval {
@@ -648,6 +671,12 @@ typedef struct Intersection {
   int prim;
   int object;
   int type;
+
+#ifdef __KERNEL_DEBUG__
+  int num_traversed_nodes;
+  int num_traversed_instances;
+  int num_intersections;
+#endif
 } Intersection;
 
 /* Primitives */
@@ -660,24 +689,22 @@ typedef enum PrimitiveType {
   PRIMITIVE_MOTION_CURVE_THICK = (1 << 3),
   PRIMITIVE_CURVE_RIBBON = (1 << 4),
   PRIMITIVE_MOTION_CURVE_RIBBON = (1 << 5),
-  PRIMITIVE_VOLUME = (1 << 6),
   /* Lamp primitive is not included below on purpose,
    * since it is no real traceable primitive.
    */
-  PRIMITIVE_LAMP = (1 << 7),
+  PRIMITIVE_LAMP = (1 << 6),
 
   PRIMITIVE_ALL_TRIANGLE = (PRIMITIVE_TRIANGLE | PRIMITIVE_MOTION_TRIANGLE),
   PRIMITIVE_ALL_CURVE = (PRIMITIVE_CURVE_THICK | PRIMITIVE_MOTION_CURVE_THICK |
                          PRIMITIVE_CURVE_RIBBON | PRIMITIVE_MOTION_CURVE_RIBBON),
-  PRIMITIVE_ALL_VOLUME = (PRIMITIVE_VOLUME),
   PRIMITIVE_ALL_MOTION = (PRIMITIVE_MOTION_TRIANGLE | PRIMITIVE_MOTION_CURVE_THICK |
                           PRIMITIVE_MOTION_CURVE_RIBBON),
-  PRIMITIVE_ALL = (PRIMITIVE_ALL_TRIANGLE | PRIMITIVE_ALL_CURVE | PRIMITIVE_ALL_VOLUME),
+  PRIMITIVE_ALL = (PRIMITIVE_ALL_TRIANGLE | PRIMITIVE_ALL_CURVE),
 
   /* Total number of different traceable primitives.
    * NOTE: This is an actual value, not a bitflag.
    */
-  PRIMITIVE_NUM_TOTAL = 7,
+  PRIMITIVE_NUM_TOTAL = 6,
 } PrimitiveType;
 
 #define PRIMITIVE_PACK_SEGMENT(type, segment) ((segment << PRIMITIVE_NUM_TOTAL) | (type))
@@ -1236,6 +1263,13 @@ typedef struct KernelFilm {
   int pass_bake_differential;
   int pad;
 
+#ifdef __KERNEL_DEBUG__
+  int pass_bvh_traversed_nodes;
+  int pass_bvh_traversed_instances;
+  int pass_bvh_intersections;
+  int pass_ray_bounces;
+#endif
+
   /* viewport rendering options */
   int display_pass_stride;
   int display_pass_components;
@@ -1445,8 +1479,7 @@ typedef struct KernelObject {
   float cryptomatte_object;
   float cryptomatte_asset;
 
-  float shadow_terminator_shading_offset;
-  float shadow_terminator_geometry_offset;
+  float shadow_terminator_offset;
   float pad1, pad2, pad3;
 } KernelObject;
 static_assert_align(KernelObject, 16);

intern/cycles/kernel/shaders/node_voronoi_texture.osl

@@ -54,18 +54,14 @@ vector4 safe_divide(vector4 a, float b)
 }
 
 /*
- * Original code is under the MIT License, Copyright (c) 2013 Inigo Quilez.
- *
  * Smooth Voronoi:
+ *
  * - https://wiki.blender.org/wiki/User:OmarSquircleArt/GSoC2019/Documentation/Smooth_Voronoi
  *
- * Distance To Edge based on:
+ * Distance To Edge:
  *
- * - https://www.iquilezles.org/www/articles/voronoilines/voronoilines.htm
- * - https://www.shadertoy.com/view/ldl3W8
+ * - https://www.shadertoy.com/view/llG3zy
  *
- * With optimization to change -2..2 scan window to -1..1 for better performance,
- * as explained in https://www.shadertoy.com/view/llG3zy.
  */
 
 /* **** 1D Voronoi **** */

intern/cycles/kernel/svm/svm_vector_rotate.h

@@ -50,29 +50,24 @@ ccl_device void svm_node_vector_rotate(ShaderData *sd,
     }
     else {
       float3 axis;
-      float axis_length;
       switch (type) {
         case NODE_VECTOR_ROTATE_TYPE_AXIS_X:
           axis = make_float3(1.0f, 0.0f, 0.0f);
-          axis_length = 1.0f;
           break;
         case NODE_VECTOR_ROTATE_TYPE_AXIS_Y:
           axis = make_float3(0.0f, 1.0f, 0.0f);
-          axis_length = 1.0f;
           break;
         case NODE_VECTOR_ROTATE_TYPE_AXIS_Z:
           axis = make_float3(0.0f, 0.0f, 1.0f);
-          axis_length = 1.0f;
           break;
         default:
-          axis = stack_load_float3(stack, axis_stack_offset);
-          axis_length = len(axis);
+          axis = normalize(stack_load_float3(stack, axis_stack_offset));
           break;
       }
       float angle = stack_load_float(stack, angle_stack_offset);
       angle = invert ? -angle : angle;
-      result = (axis_length != 0.0f) ?
-                   rotate_around_axis(vector - center, axis / axis_length, angle) + center :
+      result = (len_squared(axis) != 0.0f) ?
+                   rotate_around_axis(vector - center, axis, angle) + center :
                    vector;
     }
 

intern/cycles/kernel/svm/svm_voronoi.h

@@ -17,19 +17,14 @@
 CCL_NAMESPACE_BEGIN
 
 /*
- * Original code is under the MIT License, Copyright (c) 2013 Inigo Quilez.
- *
  * Smooth Voronoi:
  *
  * - https://wiki.blender.org/wiki/User:OmarSquircleArt/GSoC2019/Documentation/Smooth_Voronoi
  *
- * Distance To Edge based on:
+ * Distance To Edge:
  *
- * - https://www.iquilezles.org/www/articles/voronoilines/voronoilines.htm
- * - https://www.shadertoy.com/view/ldl3W8
+ * - https://www.shadertoy.com/view/llG3zy
  *
- * With optimization to change -2..2 scan window to -1..1 for better performance,
- * as explained in https://www.shadertoy.com/view/llG3zy.
  */
 
 /* **** 1D Voronoi **** */

intern/cycles/render/attribute.cpp

@@ -20,7 +20,6 @@
 #include "render/mesh.h"
 
 #include "util/util_foreach.h"
-#include "util/util_logging.h"
 #include "util/util_transform.h"
 
 CCL_NAMESPACE_BEGIN
@@ -209,7 +208,6 @@ size_t Attribute::element_size(Geometry *geom, AttributePrimitive prim) const
     case ATTR_ELEMENT_VERTEX_MOTION:
       if (geom->geometry_type == Geometry::MESH) {
         Mesh *mesh = static_cast<Mesh *>(geom);
-        DCHECK_GT(mesh->get_motion_steps(), 0);
         size = (mesh->get_verts().size() + mesh->get_num_ngons()) * (mesh->get_motion_steps() - 1);
         if (prim == ATTR_PRIM_SUBD) {
           size -= mesh->get_num_subd_verts() * (mesh->get_motion_steps() - 1);
@@ -254,7 +252,6 @@ size_t Attribute::element_size(Geometry *geom, AttributePrimitive prim) const
     case ATTR_ELEMENT_CURVE_KEY_MOTION:
       if (geom->geometry_type == Geometry::HAIR) {
         Hair *hair = static_cast<Hair *>(geom);
-        DCHECK_GT(hair->get_motion_steps(), 0);
         size = hair->get_curve_keys().size() * (hair->get_motion_steps() - 1);
       }
       break;

intern/cycles/render/buffers.cpp

@@ -322,6 +322,15 @@ bool RenderBuffers::get_pass_rect(
           pixels[0] = saturate(f * scale_exposure);
         }
       }
+#ifdef WITH_CYCLES_DEBUG
+      else if (type == PASS_BVH_TRAVERSED_NODES || type == PASS_BVH_TRAVERSED_INSTANCES ||
+               type == PASS_BVH_INTERSECTIONS || type == PASS_RAY_BOUNCES) {
+        for (int i = 0; i < size; i++, in += pass_stride, pixels++) {
+          float f = *in;
+          pixels[0] = f * scale;
+        }
+      }
+#endif
       else {
         for (int i = 0; i < size; i++, in += pass_stride, pixels++) {
           float f = *in;

intern/cycles/render/camera.h

@@ -81,7 +81,7 @@ class Camera : public Node {
   /* ** Rolling shutter effect. ** */
   /* Defines rolling shutter effect type. */
   NODE_SOCKET_API(RollingShutterType, rolling_shutter_type)
-  /* Specifies exposure time of scan-lines when using
+  /* Specifies exposure time of scanlines when using
    * rolling shutter effect.
    */
   NODE_SOCKET_API(float, rolling_shutter_duration)

intern/cycles/render/colorspace.cpp

@@ -385,7 +385,7 @@ void ColorSpaceManager::free_memory()
 {
 #ifdef WITH_OCIO
   map_free_memory(cached_colorspaces);
-  map_free_memory(cached_processors);
+  map_free_memory(cached_colorspaces);
 #endif
 }
 

intern/cycles/render/constant_fold.cpp

@@ -48,7 +48,6 @@ void ConstantFolder::make_constant(float value) const
 
   foreach (ShaderInput *sock, output->links) {
     sock->set(value);
-    sock->constant_folded_in = true;
   }
 
   graph->disconnect(output);
@@ -60,7 +59,6 @@ void ConstantFolder::make_constant(float3 value) const
 
   foreach (ShaderInput *sock, output->links) {
     sock->set(value);
-    sock->constant_folded_in = true;
   }
 
   graph->disconnect(output);

intern/cycles/render/film.cpp

@@ -51,6 +51,12 @@ static NodeEnum *get_pass_type_enum()
   pass_type_enum.insert("material_id", PASS_MATERIAL_ID);
   pass_type_enum.insert("motion", PASS_MOTION);
   pass_type_enum.insert("motion_weight", PASS_MOTION_WEIGHT);
+#ifdef __KERNEL_DEBUG__
+  pass_type_enum.insert("traversed_nodes", PASS_BVH_TRAVERSED_NODES);
+  pass_type_enum.insert("traverse_instances", PASS_BVH_TRAVERSED_INSTANCES);
+  pass_type_enum.insert("bvh_intersections", PASS_BVH_INTERSECTIONS);
+  pass_type_enum.insert("ray_bounces", PASS_RAY_BOUNCES);
+#endif
   pass_type_enum.insert("render_time", PASS_RENDER_TIME);
   pass_type_enum.insert("cryptomatte", PASS_CRYPTOMATTE);
   pass_type_enum.insert("aov_color", PASS_AOV_COLOR);
@@ -194,6 +200,15 @@ void Pass::add(PassType type, vector<Pass> &passes, const char *name)
        */
       pass.components = 0;
       break;
+#ifdef WITH_CYCLES_DEBUG
+    case PASS_BVH_TRAVERSED_NODES:
+    case PASS_BVH_TRAVERSED_INSTANCES:
+    case PASS_BVH_INTERSECTIONS:
+    case PASS_RAY_BOUNCES:
+      pass.components = 1;
+      pass.exposure = false;
+      break;
+#endif
     case PASS_RENDER_TIME:
       /* This pass is handled entirely on the host side. */
       pass.components = 0;
@@ -555,6 +570,20 @@ void Film::device_update(Device *device, DeviceScene *dscene, Scene *scene)
         kfilm->pass_bake_differential = kfilm->pass_stride;
         break;
 
+#ifdef WITH_CYCLES_DEBUG
+      case PASS_BVH_TRAVERSED_NODES:
+        kfilm->pass_bvh_traversed_nodes = kfilm->pass_stride;
+        break;
+      case PASS_BVH_TRAVERSED_INSTANCES:
+        kfilm->pass_bvh_traversed_instances = kfilm->pass_stride;
+        break;
+      case PASS_BVH_INTERSECTIONS:
+        kfilm->pass_bvh_intersections = kfilm->pass_stride;
+        break;
+      case PASS_RAY_BOUNCES:
+        kfilm->pass_ray_bounces = kfilm->pass_stride;
+        break;
+#endif
       case PASS_RENDER_TIME:
         break;
       case PASS_CRYPTOMATTE:

intern/cycles/render/graph.h

@@ -79,11 +79,7 @@ enum ShaderNodeSpecialType {
 class ShaderInput {
  public:
   ShaderInput(const SocketType &socket_type_, ShaderNode *parent_)
-      : socket_type(socket_type_),
-        parent(parent_),
-        link(NULL),
-        stack_offset(SVM_STACK_INVALID),
-        constant_folded_in(false)
+      : socket_type(socket_type_), parent(parent_), link(NULL), stack_offset(SVM_STACK_INVALID)
   {
   }
 
@@ -115,10 +111,6 @@ class ShaderInput {
   ShaderNode *parent;
   ShaderOutput *link;
   int stack_offset; /* for SVM compiler */
-
-  /* Keeps track of whether a constant was folded in this socket, to avoid over-optimizing when the
-   * link is null. */
-  bool constant_folded_in;
 };
 
 /* Output
@@ -356,7 +348,7 @@ class ShaderGraph : public NodeOwner {
   /* This function is used to create a node of a specified type instead of
    * calling 'new', and sets the graph as the owner of the node.
    */
-  template<typename T, typename... Args> T *create_node(Args &&...args)
+  template<typename T, typename... Args> T *create_node(Args &&... args)
   {
     T *node = new T(args...);
     node->set_owner(this);

intern/cycles/render/image_vdb.cpp

@@ -16,10 +16,8 @@
 
 #include "render/image_vdb.h"
 
-#include "util/util_logging.h"
-#include "util/util_openvdb.h"
-
 #ifdef WITH_OPENVDB
+#  include <openvdb/openvdb.h>
 #  include <openvdb/tools/Dense.h>
 #endif
 #ifdef WITH_NANOVDB
@@ -28,57 +26,6 @@
 
 CCL_NAMESPACE_BEGIN
 
-#ifdef WITH_OPENVDB
-struct NumChannelsOp {
-  int num_channels = 0;
-
-  template<typename GridType, typename FloatGridType, typename FloatDataType, int channels>
-  bool operator()(const openvdb::GridBase::ConstPtr &)
-  {
-    num_channels = channels;
-    return true;
-  }
-};
-
-struct ToDenseOp {
-  openvdb::CoordBBox bbox;
-  void *pixels;
-
-  template<typename GridType, typename FloatGridType, typename FloatDataType, int channels>
-  bool operator()(const openvdb::GridBase::ConstPtr &grid)
-  {
-    openvdb::tools::Dense<FloatDataType, openvdb::tools::LayoutXYZ> dense(bbox,
-                                                                          (FloatDataType *)pixels);
-    openvdb::tools::copyToDense(*openvdb::gridConstPtrCast<GridType>(grid), dense);
-    return true;
-  }
-};
-
-#  ifdef WITH_NANOVDB
-struct ToNanoOp {
-  nanovdb::GridHandle<> nanogrid;
-
-  template<typename GridType, typename FloatGridType, typename FloatDataType, int channels>
-  bool operator()(const openvdb::GridBase::ConstPtr &grid)
-  {
-    if constexpr (!std::is_same_v<GridType, openvdb::MaskGrid>) {
-      try {
-        nanogrid = nanovdb::openToNanoVDB(
-            FloatGridType(*openvdb::gridConstPtrCast<GridType>(grid)));
-      }
-      catch (const std::exception &e) {
-        VLOG(1) << "Error converting OpenVDB to NanoVDB grid: " << e.what();
-      }
-      return true;
-    }
-    else {
-      return false;
-    }
-  }
-};
-#  endif
-#endif
-
 VDBImageLoader::VDBImageLoader(const string &grid_name) : grid_name(grid_name)
 {
 }
@@ -94,40 +41,98 @@ bool VDBImageLoader::load_metadata(const ImageDeviceFeatures &features, ImageMet
     return false;
   }
 
-  /* Get number of channels from type. */
-  NumChannelsOp op;
-  if (!openvdb::grid_type_operation(grid, op)) {
-    return false;
-  }
-
-  metadata.channels = op.num_channels;
-
-  /* Set data type. */
-#  ifdef WITH_NANOVDB
-  if (features.has_nanovdb) {
-    /* NanoVDB expects no inactive leaf nodes. */
-    /*openvdb::FloatGrid &pruned_grid = *openvdb::gridPtrCast<openvdb::FloatGrid>(grid);
-    openvdb::tools::pruneInactive(pruned_grid.tree());
-    nanogrid = nanovdb::openToNanoVDB(pruned_grid);*/
-    ToNanoOp op;
-    if (!openvdb::grid_type_operation(grid, op)) {
-      return false;
-    }
-    nanogrid = std::move(op.nanogrid);
-  }
-#  endif
-
-  /* Set dimensions. */
   bbox = grid->evalActiveVoxelBoundingBox();
   if (bbox.empty()) {
     return false;
   }
 
+  /* Set dimensions. */
   openvdb::Coord dim = bbox.dim();
   metadata.width = dim.x();
   metadata.height = dim.y();
   metadata.depth = dim.z();
 
+  /* Set data type. */
+  if (grid->isType<openvdb::FloatGrid>()) {
+    metadata.channels = 1;
+#  ifdef WITH_NANOVDB
+    if (features.has_nanovdb) {
+      nanogrid = nanovdb::openToNanoVDB(*openvdb::gridConstPtrCast<openvdb::FloatGrid>(grid));
+    }
+#  endif
+  }
+  else if (grid->isType<openvdb::Vec3fGrid>()) {
+    metadata.channels = 3;
+#  ifdef WITH_NANOVDB
+    if (features.has_nanovdb) {
+      nanogrid = nanovdb::openToNanoVDB(*openvdb::gridConstPtrCast<openvdb::Vec3fGrid>(grid));
+    }
+#  endif
+  }
+  else if (grid->isType<openvdb::BoolGrid>()) {
+    metadata.channels = 1;
+#  ifdef WITH_NANOVDB
+    if (features.has_nanovdb) {
+      nanogrid = nanovdb::openToNanoVDB(
+          openvdb::FloatGrid(*openvdb::gridConstPtrCast<openvdb::BoolGrid>(grid)));
+    }
+#  endif
+  }
+  else if (grid->isType<openvdb::DoubleGrid>()) {
+    metadata.channels = 1;
+#  ifdef WITH_NANOVDB
+    if (features.has_nanovdb) {
+      nanogrid = nanovdb::openToNanoVDB(
+          openvdb::FloatGrid(*openvdb::gridConstPtrCast<openvdb::DoubleGrid>(grid)));
+    }
+#  endif
+  }
+  else if (grid->isType<openvdb::Int32Grid>()) {
+    metadata.channels = 1;
+#  ifdef WITH_NANOVDB
+    if (features.has_nanovdb) {
+      nanogrid = nanovdb::openToNanoVDB(
+          openvdb::FloatGrid(*openvdb::gridConstPtrCast<openvdb::Int32Grid>(grid)));
+    }
+#  endif
+  }
+  else if (grid->isType<openvdb::Int64Grid>()) {
+    metadata.channels = 1;
+#  ifdef WITH_NANOVDB
+    if (features.has_nanovdb) {
+      nanogrid = nanovdb::openToNanoVDB(
+          openvdb::FloatGrid(*openvdb::gridConstPtrCast<openvdb::Int64Grid>(grid)));
+    }
+#  endif
+  }
+  else if (grid->isType<openvdb::Vec3IGrid>()) {
+    metadata.channels = 3;
+#  ifdef WITH_NANOVDB
+    if (features.has_nanovdb) {
+      nanogrid = nanovdb::openToNanoVDB(
+          openvdb::Vec3fGrid(*openvdb::gridConstPtrCast<openvdb::Vec3IGrid>(grid)));
+    }
+#  endif
+  }
+  else if (grid->isType<openvdb::Vec3dGrid>()) {
+    metadata.channels = 3;
+#  ifdef WITH_NANOVDB
+    if (features.has_nanovdb) {
+      nanogrid = nanovdb::openToNanoVDB(
+          openvdb::Vec3fGrid(*openvdb::gridConstPtrCast<openvdb::Vec3dGrid>(grid)));
+    }
+#  endif
+  }
+  else if (grid->isType<openvdb::MaskGrid>()) {
+    metadata.channels = 1;
+#  ifdef WITH_NANOVDB
+    return false;  // Unsupported
+#  endif
+  }
+  else {
+    return false;
+  }
+
 #  ifdef WITH_NANOVDB
   if (nanogrid) {
     metadata.byte_size = nanogrid.size();
@@ -195,10 +200,45 @@ bool VDBImageLoader::load_pixels(const ImageMetaData &, void *pixels, const size
   else
 #  endif
   {
-    ToDenseOp op;
-    op.pixels = pixels;
-    op.bbox = bbox;
-    openvdb::grid_type_operation(grid, op);
+    if (grid->isType<openvdb::FloatGrid>()) {
+      openvdb::tools::Dense<float, openvdb::tools::LayoutXYZ> dense(bbox, (float *)pixels);
+      openvdb::tools::copyToDense(*openvdb::gridConstPtrCast<openvdb::FloatGrid>(grid), dense);
+    }
+    else if (grid->isType<openvdb::Vec3fGrid>()) {
+      openvdb::tools::Dense<openvdb::Vec3f, openvdb::tools::LayoutXYZ> dense(
+          bbox, (openvdb::Vec3f *)pixels);
+      openvdb::tools::copyToDense(*openvdb::gridConstPtrCast<openvdb::Vec3fGrid>(grid), dense);
+    }
+    else if (grid->isType<openvdb::BoolGrid>()) {
+      openvdb::tools::Dense<float, openvdb::tools::LayoutXYZ> dense(bbox, (float *)pixels);
+      openvdb::tools::copyToDense(*openvdb::gridConstPtrCast<openvdb::BoolGrid>(grid), dense);
+    }
+    else if (grid->isType<openvdb::DoubleGrid>()) {
+      openvdb::tools::Dense<float, openvdb::tools::LayoutXYZ> dense(bbox, (float *)pixels);
+      openvdb::tools::copyToDense(*openvdb::gridConstPtrCast<openvdb::DoubleGrid>(grid), dense);
+    }
+    else if (grid->isType<openvdb::Int32Grid>()) {
+      openvdb::tools::Dense<float, openvdb::tools::LayoutXYZ> dense(bbox, (float *)pixels);
+      openvdb::tools::copyToDense(*openvdb::gridConstPtrCast<openvdb::Int32Grid>(grid), dense);
+    }
+    else if (grid->isType<openvdb::Int64Grid>()) {
+      openvdb::tools::Dense<float, openvdb::tools::LayoutXYZ> dense(bbox, (float *)pixels);
+      openvdb::tools::copyToDense(*openvdb::gridConstPtrCast<openvdb::Int64Grid>(grid), dense);
+    }
+    else if (grid->isType<openvdb::Vec3IGrid>()) {
+      openvdb::tools::Dense<openvdb::Vec3f, openvdb::tools::LayoutXYZ> dense(
+          bbox, (openvdb::Vec3f *)pixels);
+      openvdb::tools::copyToDense(*openvdb::gridConstPtrCast<openvdb::Vec3IGrid>(grid), dense);
+    }
+    else if (grid->isType<openvdb::Vec3dGrid>()) {
+      openvdb::tools::Dense<openvdb::Vec3f, openvdb::tools::LayoutXYZ> dense(
+          bbox, (openvdb::Vec3f *)pixels);
+      openvdb::tools::copyToDense(*openvdb::gridConstPtrCast<openvdb::Vec3dGrid>(grid), dense);
+    }
+    else if (grid->isType<openvdb::MaskGrid>()) {
+      openvdb::tools::Dense<float, openvdb::tools::LayoutXYZ> dense(bbox, (float *)pixels);
+      openvdb::tools::copyToDense(*openvdb::gridConstPtrCast<openvdb::MaskGrid>(grid), dense);
+    }
   }
   return true;
 #else

intern/cycles/render/object.cpp

@@ -93,8 +93,7 @@ NODE_DEFINE(Object)
   SOCKET_POINT(dupli_generated, "Dupli Generated", zero_float3());
   SOCKET_POINT2(dupli_uv, "Dupli UV", zero_float2());
   SOCKET_TRANSFORM_ARRAY(motion, "Motion", array<Transform>());
-  SOCKET_FLOAT(shadow_terminator_shading_offset, "Shadow Terminator Shading Offset", 0.0f);
-  SOCKET_FLOAT(shadow_terminator_geometry_offset, "Shadow Terminator Geometry Offset", 0.1f);
+  SOCKET_FLOAT(shadow_terminator_offset, "Terminator Offset", 0.0f);
   SOCKET_STRING(asset_name, "Asset Name", ustring());
 
   SOCKET_BOOLEAN(is_shadow_catcher, "Shadow Catcher", false);
@@ -508,9 +507,7 @@ void ObjectManager::device_update_object_transform(UpdateObjectTransformState *s
     kobject.cryptomatte_asset = util_hash_to_float(hash_asset);
   }
 
-  kobject.shadow_terminator_shading_offset = 1.0f /
-                                             (1.0f - 0.5f * ob->shadow_terminator_shading_offset);
-  kobject.shadow_terminator_geometry_offset = ob->shadow_terminator_geometry_offset;
+  kobject.shadow_terminator_offset = 1.0f / (1.0f - 0.5f * ob->shadow_terminator_offset);
 
   /* Object flag. */
   if (ob->use_holdout) {

intern/cycles/render/object.h

@@ -64,8 +64,7 @@ class Object : public Node {
   NODE_SOCKET_API(bool, hide_on_missing_motion)
   NODE_SOCKET_API(bool, use_holdout)
   NODE_SOCKET_API(bool, is_shadow_catcher)
-  NODE_SOCKET_API(float, shadow_terminator_shading_offset)
-  NODE_SOCKET_API(float, shadow_terminator_geometry_offset)
+  NODE_SOCKET_API(float, shadow_terminator_offset)
 
   NODE_SOCKET_API(float3, dupli_generated)
   NODE_SOCKET_API(float2, dupli_uv)

intern/cycles/render/osl.cpp

@@ -993,7 +993,7 @@ void OSLCompiler::parameter_array(const char *name, const float f[], int arrayle
 
 void OSLCompiler::parameter_color_array(const char *name, const array<float3> &f)
 {
-  /* NOTE: cycles float3 type is actually 4 floats! need to use an explicit array. */
+  /* NB: cycles float3 type is actually 4 floats! need to use an explicit array */
   array<float[3]> table(f.size());
 
   for (int i = 0; i < f.size(); ++i) {

intern/cycles/render/scene.h

@@ -209,7 +209,7 @@ class SceneParams {
 
   int curve_subdivisions()
   {
-    /* Matching the tessellation rate limit in Embree. */
+    /* Matching the tesselation rate limit in Embree. */
     return clamp(1 << hair_subdivisions, 1, 16);
   }
 };
@@ -299,7 +299,7 @@ class Scene : public NodeOwner {
    * node array (e.g. Scene::geometry for Geometry nodes) and tag the appropriate
    * manager for an update.
    */
-  template<typename T, typename... Args> T *create_node(Args &&...args)
+  template<typename T, typename... Args> T *create_node(Args &&... args)
   {
     T *node = new T(args...);
     node->set_owner(this);

intern/cycles/render/session.cpp

@@ -227,25 +227,66 @@ void Session::run_gpu()
   progress.set_render_start_time();
 
   while (!progress.get_cancel()) {
-    const bool no_tiles = !run_update_for_next_iteration();
+    /* advance to next tile */
+    bool no_tiles = !tile_manager.next();
 
+    DeviceKernelStatus kernel_state = DEVICE_KERNEL_UNKNOWN;
     if (no_tiles) {
-      if (params.background) {
-        /* if no work left and in background mode, we can stop immediately */
+      kernel_state = device->get_active_kernel_switch_state();
+    }
+
+    if (params.background) {
+      /* if no work left and in background mode, we can stop immediately */
+      if (no_tiles) {
         progress.set_status("Finished");
         break;
       }
     }
 
-    if (run_wait_for_work(no_tiles)) {
-      continue;
+    else if (no_tiles && kernel_state == DEVICE_KERNEL_FEATURE_KERNEL_AVAILABLE) {
+      reset_gpu(tile_manager.params, params.samples);
     }
 
-    if (progress.get_cancel()) {
-      break;
+    else {
+      /* if in interactive mode, and we are either paused or done for now,
+       * wait for pause condition notify to wake up again */
+      thread_scoped_lock pause_lock(pause_mutex);
+
+      if (!pause && !tile_manager.done()) {
+        /* reset could have happened after no_tiles was set, before this lock.
+         * in this case we shall not wait for pause condition
+         */
+      }
+      else if (pause || no_tiles) {
+        update_status_time(pause, no_tiles);
+
+        while (1) {
+          scoped_timer pause_timer;
+          pause_cond.wait(pause_lock);
+          if (pause) {
+            progress.add_skip_time(pause_timer, params.background);
+          }
+
+          update_status_time(pause, no_tiles);
+          progress.set_update();
+
+          if (!pause)
+            break;
+        }
+      }
+
+      if (progress.get_cancel())
+        break;
     }
 
     if (!no_tiles) {
+      /* update scene */
+      scoped_timer update_timer;
+      if (update_scene()) {
+        profiler.reset(scene->shaders.size(), scene->objects.size());
+      }
+      progress.add_skip_time(update_timer, params.background);
+
       if (!device->error_message().empty())
         progress.set_error(device->error_message());
 
@@ -688,27 +729,82 @@ void Session::run_cpu()
   last_update_time = time_dt();
   last_display_time = last_update_time;
 
+  {
+    /* reset once to start */
+    thread_scoped_lock reset_lock(delayed_reset.mutex);
+    thread_scoped_lock buffers_lock(buffers_mutex);
+    thread_scoped_lock display_lock(display_mutex);
+
+    reset_(delayed_reset.params, delayed_reset.samples);
+    delayed_reset.do_reset = false;
+  }
+
   while (!progress.get_cancel()) {
-    const bool no_tiles = !run_update_for_next_iteration();
+    /* advance to next tile */
+    bool no_tiles = !tile_manager.next();
     bool need_copy_to_display_buffer = false;
 
+    DeviceKernelStatus kernel_state = DEVICE_KERNEL_UNKNOWN;
     if (no_tiles) {
-      if (params.background) {
-        /* if no work left and in background mode, we can stop immediately */
+      kernel_state = device->get_active_kernel_switch_state();
+    }
+
+    if (params.background) {
+      /* if no work left and in background mode, we can stop immediately */
+      if (no_tiles) {
         progress.set_status("Finished");
         break;
       }
     }
 
-    if (run_wait_for_work(no_tiles)) {
-      continue;
+    else if (no_tiles && kernel_state == DEVICE_KERNEL_FEATURE_KERNEL_AVAILABLE) {
+      reset_cpu(tile_manager.params, params.samples);
     }
 
-    if (progress.get_cancel()) {
-      break;
+    else {
+      /* if in interactive mode, and we are either paused or done for now,
+       * wait for pause condition notify to wake up again */
+      thread_scoped_lock pause_lock(pause_mutex);
+
+      if (!pause && delayed_reset.do_reset) {
+        /* reset once to start */
+        thread_scoped_lock reset_lock(delayed_reset.mutex);
+        thread_scoped_lock buffers_lock(buffers_mutex);
+        thread_scoped_lock display_lock(display_mutex);
+
+        reset_(delayed_reset.params, delayed_reset.samples);
+        delayed_reset.do_reset = false;
+      }
+      else if (pause || no_tiles) {
+        update_status_time(pause, no_tiles);
+
+        while (1) {
+          scoped_timer pause_timer;
+          pause_cond.wait(pause_lock);
+          if (pause) {
+            progress.add_skip_time(pause_timer, params.background);
+          }
+
+          update_status_time(pause, no_tiles);
+          progress.set_update();
+
+          if (!pause)
+            break;
+        }
+      }
+
+      if (progress.get_cancel())
+        break;
     }
 
     if (!no_tiles) {
+      /* update scene */
+      scoped_timer update_timer;
+      if (update_scene()) {
+        profiler.reset(scene->shaders.size(), scene->objects.size());
+      }
+      progress.add_skip_time(update_timer, params.background);
+
       if (!device->error_message().empty())
         progress.set_error(device->error_message());
 
@@ -798,63 +894,6 @@ void Session::run()
     progress.set_update();
 }
 
-bool Session::run_update_for_next_iteration()
-{
-  thread_scoped_lock scene_lock(scene->mutex);
-  thread_scoped_lock reset_lock(delayed_reset.mutex);
-
-  if (delayed_reset.do_reset) {
-    thread_scoped_lock buffers_lock(buffers_mutex);
-    reset_(delayed_reset.params, delayed_reset.samples);
-    delayed_reset.do_reset = false;
-  }
-
-  const bool have_tiles = tile_manager.next();
-
-  if (have_tiles) {
-    scoped_timer update_timer;
-    if (update_scene()) {
-      profiler.reset(scene->shaders.size(), scene->objects.size());
-    }
-    progress.add_skip_time(update_timer, params.background);
-  }
-
-  return have_tiles;
-}
-
-bool Session::run_wait_for_work(bool no_tiles)
-{
-  /* In an offline rendering there is no pause, and no tiles will mean the job is fully done. */
-  if (params.background) {
-    return false;
-  }
-
-  thread_scoped_lock pause_lock(pause_mutex);
-
-  if (!pause && !no_tiles) {
-    return false;
-  }
-
-  update_status_time(pause, no_tiles);
-
-  while (true) {
-    scoped_timer pause_timer;
-    pause_cond.wait(pause_lock);
-    if (pause) {
-      progress.add_skip_time(pause_timer, params.background);
-    }
-
-    update_status_time(pause, no_tiles);
-    progress.set_update();
-
-    if (!pause) {
-      break;
-    }
-  }
-
-  return no_tiles;
-}
-
 bool Session::draw(BufferParams &buffer_params, DeviceDrawParams &draw_params)
 {
   if (device_use_gl)
@@ -973,6 +1012,8 @@ void Session::wait()
 
 bool Session::update_scene()
 {
+  thread_scoped_lock scene_lock(scene->mutex);
+
   /* update camera if dimensions changed for progressive render. the camera
    * knows nothing about progressive or cropped rendering, it just gets the
    * image dimensions passed in */

intern/cycles/render/session.h

@@ -178,9 +178,6 @@ class Session {
 
   void run();
 
-  bool run_update_for_next_iteration();
-  bool run_wait_for_work(bool no_tiles);
-
   void update_status_time(bool show_pause = false, bool show_done = false);
 
   void render(bool use_denoise);