runner

Signed-off-by: Larry Gritz <lg@larrygritz.com>
Set up workflows to have subsequent PR pushes cancel previous in-progress jobs
2022-10-10 22:40:36 -07:00 · 2022-10-10 22:30:58 -07:00 · 2022-10-10 22:18:46 -07:00 · 2022-10-10 22:08:46 -07:00 · 2022-10-10 22:05:08 -07:00 · 2022-10-10 21:54:57 -07:00
277 changed files with 15833 additions and 13176 deletions
@@ -6,3 +6,40 @@ bdb473cbdb119f3d9060d48f75f965cbd45f6d7a

 # More clang-format: liboslquery, oslinfo, oslc, osl.imageio, osltoy (#1220)
 bf366d7a90c46599b82071dac6dbef7c5e3ca157
+
+# Turn on clang-format for cpp files in testsuite (#1269)
+00c895fc642eb89685797c292048bdff15d675e7
+5e848fb761261521a3ee7cd8a4f6814f32fe4a47
+
+# CI: Change our clang-format standard to clang11 (#1372)
+91d22ebe2f65d4ed0599e421172a15d99fe0e1b4
+
+# More clang-format (this time, testrender and liboslnoise) (#1511)
+08706b7cbac781b6a9389780428016be985d9d89
+
+# clang-format #1509
+b4026fd1695c08532557b3ba1c2142323087ccf9
+1f73b075c99861afa42fac44567b37f797bc72f2
+7f2e83794e6fa63d5180653cffb77be2cc38b32b
+28399fefc890442b5ac020e7a90cee0d5bf9f03c
+e56851056d33233b0bb8324faac251907099961a
+1c57fc61d8ab7c3f309be4335e4c71c26399edd4
+46d451ce74d00e904874df476ebeca19bb119ea9
+c1045f2c2845a98e661c7181932a031eb3c76063
+
+# More clang-format (#1512)
+8c6ffdd911b9eb7db4adf41ab88891360452d7c5
+
+# style: more clang-format (#1515)
+102430c83554fe7e328cd8c3317db00507bcfaba
+
+# Revert some formatting (#1518)
+21eac9d9215d071b0a7b9ce2c32f189ed0c62c4a
+
+# Another round of clang-format (#1540)
+38015b6be28643254aea530154b7860136cf5e15
+89d5db2b5c5fda79184d8245d903e580fc6a4e9c
+
+# clang-format: the very last piece of the puzzle (#1555)
+4ee62cd2ef05543f818f4d95a19efc1d684a7381
+5be7c6bd4e7bd2e560cb6be8c06f11985384a87a
@@ -0,0 +1,132 @@
+# Copyright 2008-present Contributors to the Open Shading Language project.
+# SPDX-License-Identifier: BSD-3-Clause
+# https://github.com/AcademySoftwareFoundation/OpenShadingLanguage
+
+name: Analysis
+
+on:
+  schedule:
+    # Run nightly while we're still working our way through the warnings
+    - cron: "0 8 * * *"
+    # Run unconditionally once weekly
+    # - cron: "0 0 * * 0"
+  push:
+    branches:
+      - main
+      - '*analysis*'
+  # Allow manual kicking off of the workflow from github.com
+  workflow_dispatch:
+
+permissions: read-all
+
+# Allow subsequent pushes to the same PR or REF to cancel any previous jobs.
+concurrency:
+  group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }}
+  cancel-in-progress: true
+
+
+jobs:
+
+  aswf:
+    name: "SonarCloud Analysis"
+    if: |
+      github.repository == 'AcademySoftwareFoundation/OpenShadingLanguage'
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - desc: sonar gcc9/C++14 llvm13 py3.9 boost1.76 exr3.1 oiio2.3 avx2
+            nametag: static-analysis-sonar
+            os: ubuntu-20.04-8c-32g-300h
+            # os: ubuntu-latest
+            vfxyear: 2022
+            vfxsuffix: -clang13
+            cxx_std: 14
+            openimageio_ver: master
+            python_ver: 3.9
+            pybind11_ver: v2.9.0
+            simd: avx2,f16c
+            batched: b8_AVX2,b16_AVX512
+            coverage: 1
+            # skip_tests: 1
+            sonar: 1
+            setenvs: export SONAR_SCANNER_VERSION=4.7.0.2747
+                            SONAR_SERVER_URL="https://sonarcloud.io"
+                            BUILD_WRAPPER_OUT_DIR=bw_output
+                            OSL_CMAKE_BUILD_WRAPPER="build-wrapper-linux-x86-64 --out-dir bw_output"
+                            CMAKE_BUILD_TYPE=Debug
+                            CMAKE_UNITY_BUILD=OFF
+                            CODECOV=1
+                            TESTRENDER_AA=1
+                            OSL_TESTSUITE_SKIP_DIFF=1
+                            CTEST_TEST_TIMEOUT=1200
+                            OSL_CMAKE_FLAGS="-DOSL_TEST_BIG_TIMEOUT=1200 -DUSE_QT=0"
+                            CTEST_EXCLUSIONS="broken|noise-reg.regress|noise-gabor-reg.regress"
+
+    runs-on: ${{ matrix.os }}
+    container:
+      image: aswf/ci-osl:${{matrix.vfxyear}}${{matrix.vfxsuffix}}
+    env:
+      CXX: ${{matrix.cxx_compiler}}
+      CC: ${{matrix.cc_compiler}}
+      CMAKE_CXX_STANDARD: ${{matrix.cxx_std}}
+      FMT_VERSION: ${{matrix.fmt_ver}}
+      OPENEXR_VERSION: ${{matrix.openexr_ver}}
+      OPENIMAGEIO_VERSION: ${{matrix.openimageio_ver}}
+      PYBIND11_VERSION: ${{matrix.pybind11_ver}}
+      PYTHON_VERSION: ${{matrix.python_ver}}
+      USE_BATCHED: ${{matrix.batched}}
+      USE_SIMD: ${{matrix.simd}}
+      PARALLEL: 24
+      # DEBUG_CI: 1
+    steps:
+      # We would like to use harden-runner, but it flags too many false
+      # positives, every time we download a dependency. We should use it only
+      # on CI runs where we are producing artifacts that users might rely on.
+      # - name: Harden Runner
+      #   uses: step-security/harden-runner@248ae51c2e8cc9622ecf50685c8bf7150c6e8813 # v1.4.3
+      #   with:
+      #     egress-policy: audit # TODO: change to 'egress-policy: block' after couple of runs
+      - uses: actions/checkout@2541b1294d2704b0964813337f33b291d3f8596b # v3.0.2
+        with:
+          fetch-depth: '0'
+      - name: Prepare ccache timestamp
+        id: ccache_cache_keys
+        run: echo "::set-output name=date::`date -u +'%Y-%m-%dT%H:%M:%SZ'`"
+      - name: ccache
+        id: ccache
+        uses: actions/cache@c3f1317a9e7b1ef106c153ac8c0f00fed3ddbc0d # v3.0.4
+        with:
+          path: /tmp/ccache
+          key: ${{github.job}}-${{matrix.nametag}}-${{steps.ccache_cache_keys.outputs.date}}
+          restore-keys: ${{github.job}}-
+      - name: Build setup
+        run: |
+            ${{matrix.setenvs}}
+            src/build-scripts/ci-startup.bash
+      - name: Remove existing OIIO
+        if: matrix.openimageio_ver != ''
+        run: |
+            sudo rm -rf /usr/local/include/OpenImageIO
+            sudo rm -rf /usr/local/lib*/cmake/OpenImageIO
+            sudo rm -rf /usr/local/lib*/libOpenImageIO*
+            sudo rm -rf /usr/local/lib*/python3.9/site-packages/OpenImageIO*
+      - name: Dependencies
+        run: |
+            ${{matrix.depcmds}}
+            src/build-scripts/gh-installdeps.bash
+      - name: Build
+        run: src/build-scripts/ci-build.bash
+      - name: Testsuite
+        if: matrix.skip_tests != '1'
+        run: src/build-scripts/ci-test.bash
+      - name: Code coverage
+        if: matrix.coverage == '1'
+        run: src/build-scripts/ci-coverage.bash
+      - name: Sonar-scanner
+        if: matrix.sonar == 1
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          SONAR_TOKEN: ${{ secrets.SONAR_TOKEN }}
+        run: |
+            sonar-scanner -X --define sonar.host.url="${{ env.SONAR_SERVER_URL }}" --define sonar.cfamily.build-wrapper-output="build/bw_output" --define sonar.cfamily.gcov.reportsPath="_coverage"
@@ -23,6 +23,13 @@ on:
    # Full nightly build
    - cron: "0 6 * * *"

+permissions: read-all
+
+# Allow subsequent pushes to the same PR or REF to cancel any previous jobs.
+concurrency:
+  group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }}
+  cancel-in-progress: true
+

 jobs:

@@ -38,7 +45,7 @@ jobs:
            vfxyear: 2019
            vfxsuffix: -clang9
            cxx_std: 14
-            openimageio_ver: v2.2.17.0
+            openimageio_ver: v2.3.17.0
            python_ver: 2.7
            pybind11_ver: v2.4.2
          - desc: clang9/C++14 llvm9 oiio-release boost1.66 avx2 exr2.3 py2.7
@@ -70,7 +77,7 @@ jobs:
            os: ubuntu-latest
            vfxyear: 2020
            cxx_std: 14
-            openimageio_ver: v2.2.17.0
+            openimageio_ver: v2.3.17.0
            python_ver: 3.7
            pybind11_ver: v2.5.0
            simd: sse4.2
@@ -81,7 +88,7 @@ jobs:
            vfxyear: 2021
            vfxsuffix: -clang11
            cxx_std: 17
-            openimageio_ver: dev-2.2
+            openimageio_ver: dev-2.3
            python_ver: 3.7
            pybind11_ver: v2.7.0
            simd: avx2,f16c
@@ -110,6 +117,7 @@ jobs:
            batched: b8_AVX2,b8_AVX512,b16_AVX512
            setenvs: USE_OPENVDB=0
          - desc: icc/C++17 llvm13 py3.9 boost1.76 exr3.1 oiio-master avx2
+            nametag: linux-icc
            os: ubuntu-latest
            vfxyear: 2022
            vfxsuffix: -clang13
@@ -125,7 +133,8 @@ jobs:
            setenvs: export OSL_CMAKE_FLAGS="-DSTOP_ON_WARNING=OFF -DEXTRA_CPP_ARGS=-fp-model=consistent"
                            OPENIMAGEIO_CMAKE_FLAGS=-DBUILD_FMT_VERSION=7.1.3
                            USE_OPENVDB=0
-          - desc:  icx/C++17 llvm13 py3.9 boost1.76 exr3.1 oiio2.3 avx2
+          - desc: icx/C++17 llvm13 py3.9 boost1.76 exr3.1 oiio2.3 avx2
+            nametag: linux-icx
            os: ubuntu-latest
            vfxyear: 2022
            vfxsuffix: -clang13
@@ -145,12 +154,72 @@ jobs:
            vfxyear: 2019
            vfxsuffix: -clang9
            cxx_std: 14
-            openimageio_ver: v2.2.17.0
+            openimageio_ver: v2.3.17.0
            python_ver: 2.7
            pybind11_ver: v2.6.2
            simd: 0
            setenvs: export PUGIXML_VERSION=v1.8

+          # Address and leak sanitizers
+          - desc: sanitizers
+            nametag: sanitizer
+            os: ubuntu-latest
+            vfxyear: 2022
+            vfxsuffix: -clang13
+            cc_compiler: clang
+            cxx_compiler: clang++
+            cxx_std: 17
+            openimageio_ver: master
+            python_ver: 3.9
+            pybind11_ver: v2.9.0
+            simd: avx2,f16c
+            batched: b8_AVX2
+            setenvs: export OSL_CMAKE_FLAGS="-DSANITIZE=address -DOSL_TEST_BIG_TIMEOUT=1200"
+                            CMAKE_BUILD_TYPE=Debug
+                            CTEST_TEST_TIMEOUT=1200
+                            CTEST_EXCLUSIONS="broken|noise-reg.regress|noise-gabor-reg.regress|example-.*deformer|spline-reg.regress.batched|osl-imageio"
+                            OSL_OPTIONS=llvm_debugging_symbols=1
+                            TESTRENDER_AA=1
+                            OSL_TESTSUITE_SKIP_DIFF=1
+            # Sanitizer notes:
+            # - There are a handful of tests we disable entirely with
+            #   CTEST_EXCLUSIONS because they are irrelevant or just cannot be
+            #   made to be clean (and that's ok).
+            # - Tests with sanitizers run slow, so we increase timeouts, and
+            #   also set AA to 1 for render tests to speed them up.
+            #   OSL_TESTSUITE_SKIP_DIFF prevents making the AA=1 image result
+            #   fail, but that's ok because every other CI test is checking
+            #   the results; here we have no reason to believe results will
+            #   differ from other tests, we just need to run with sanitizers
+            #   turned on.
+            # - In testing.cmake, we rig texture tests to disable the leak
+            #   sanitizer, because we can't eliminate some completely benign
+            #   leaks from the texture system, and because they are called
+            #   from JITed code, we don't get a clean stack trace of things
+            #   that happen in OIIO and therefore we just can't enumerate the
+            #   allowed exceptions. So we trust that OIIO itself tests the
+            #   texture system adequately and is able to ignore the benign
+            #   "leaks."
+
+            # Test formatting. This test entry doesn't do a full build, it
+            # just runs clang-format on everything, and passes if nothing is
+            # misformatted. Upon failure, the build artifact will be the full
+            # source code with the formatting fixed (diffs will also appear in
+            # the console output).
+          - desc: "clang-format"
+            nametag: clang-format
+            os: ubuntu-latest
+            vfxyear: 2022
+            vfxsuffix: -clang14
+            cxx_std: 17
+            openimageio_ver: release
+            python_ver: 3.9
+            simd: avx2,f16c
+            batched: b8_AVX2
+            skip_tests: 1
+            setenvs: export BUILDTARGET=clang-format
+                            OPENIMAGEIO_CMAKE_FLAGS=-DUSE_PYTHON=0
+
    runs-on: ${{matrix.os}}
    container:
      image: aswftesting/ci-osl:${{matrix.vfxyear}}${{matrix.vfxsuffix}}
@@ -166,13 +235,20 @@ jobs:
      USE_BATCHED: ${{matrix.batched}}
      USE_SIMD: ${{matrix.simd}}
    steps:
-      - uses: actions/checkout@v2
+      # We would like to use harden-runner, but it flags too many false
+      # positives, every time we download a dependency. We should use it only
+      # on CI runs where we are producing artifacts that users might rely on.
+      # - name: Harden Runner
+      #   uses: step-security/harden-runner@248ae51c2e8cc9622ecf50685c8bf7150c6e8813 # v1.4.3
+      #   with:
+      #     egress-policy: audit # TODO: change to 'egress-policy: block' after couple of runs
+      - uses: actions/checkout@2541b1294d2704b0964813337f33b291d3f8596b # v3.0.2
      - name: Prepare ccache timestamp
        id: ccache_cache_keys
        run: echo "::set-output name=date::`date -u +'%Y-%m-%dT%H:%M:%SZ'`"
      - name: ccache
        id: ccache
-        uses: actions/cache@v2
+        uses: actions/cache@c3f1317a9e7b1ef106c153ac8c0f00fed3ddbc0d # v3.0.4
        with:
          path: /tmp/ccache
          key: ${{github.job}}-${{matrix.nametag}}-${{steps.ccache_cache_keys.outputs.date}}
@@ -198,7 +274,7 @@ jobs:
      - name: Testsuite
        if: matrix.skip_tests != '1'
        run: src/build-scripts/ci-test.bash
-      - uses: actions/upload-artifact@v2
+      - uses: actions/upload-artifact@3cea5372237819ed00197afe530f5a7ea3e805c8 # v3.1.0
        if: failure()
        with:
          name: osl-${{github.job}}-${{matrix.nametag}}
@@ -242,32 +318,32 @@ jobs:
            setenvs: export LLVM_VERSION=10.0.0
                            OPENIMAGEIO_CMAKE_FLAGS="-DBUILD_FMT_VERSION=7.0.1"
                            PUGIXML_VERSION=v1.10
-          - desc: latest releases gcc11/C++17 llvm13 boost1.71 exr3.1 py3.8 avx2 batch-b16avx512
+          - desc: latest releases gcc11/C++17 llvm13 boost1.71 exr3.1 py3.9 avx2 batch-b16avx512
            nametag: linux-latest-releases
            os: ubuntu-20.04
            cxx_compiler: g++-11
            cxx_std: 17
-            fmt_ver: 8.1.1
+            fmt_ver: 9.1.0
            openexr_ver: v3.1.5
            openimageio_ver: master
-            pybind11_ver: v2.9.2
-            python_ver: 3.8
+            pybind11_ver: v2.10.0
+            python_ver: 3.9
            simd: avx2,f16c
            batched: b8_AVX2,b8_AVX512,b16_AVX512
            setenvs: export LLVM_VERSION=13.0.0
                            LLVM_DISTRO_NAME=ubuntu-20.04
                            OPENCOLORIO_VERSION=v2.1.2
                            PUGIXML_VERSION=v1.11.4
-          - desc: bleeding edge gcc11/C++17 llvm14 oiio/ocio/exr/pybind-master boost1.71 py3.9 avx2 batch-b16avx512
+          - desc: bleeding edge gcc12/C++17 llvm14 oiio/ocio/exr/pybind-master boost1.71 py3.10 avx2 batch-b16avx512
            nametag: linux-bleeding-edge
-            os: ubuntu-20.04
-            cxx_compiler: g++-11
+            os: ubuntu-22.04
+            cxx_compiler: g++-12
            cxx_std: 17
            fmt_ver: master
            openexr_ver: main
            openimageio_ver: master
            pybind11_ver: master
-            python_ver: 3.9
+            python_ver: "3.10"
            simd: avx2,f16c
            batched: b8_AVX2,b8_AVX512,b16_AVX512
            setenvs: export LLVM_VERSION=14.0.0
@@ -275,7 +351,7 @@ jobs:
                            OPENCOLORIO_VERSION=main
                            PUGIXML_VERSION=master
          - desc: clang14/C++17 llvm14 boost1.71 exr3.1 py3.8 avx2 batch-b16avx512
-            nametag: linux-latest-releases
+            nametag: linux-latest-releases-clang
            os: ubuntu-20.04
            cxx_compiler: clang++
            cc_compiler: clang
@@ -305,13 +381,20 @@ jobs:
      USE_BATCHED: ${{matrix.batched}}
      USE_SIMD: ${{matrix.simd}}
    steps:
-      - uses: actions/checkout@v2
+      # We would like to use harden-runner, but it flags too many false
+      # positives, every time we download a dependency. We should use it only
+      # on CI runs where we are producing artifacts that users might rely on.
+      # - name: Harden Runner
+      #   uses: step-security/harden-runner@248ae51c2e8cc9622ecf50685c8bf7150c6e8813 # v1.4.3
+      #   with:
+      #     egress-policy: audit # TODO: change to 'egress-policy: block' after couple of runs
+      - uses: actions/checkout@2541b1294d2704b0964813337f33b291d3f8596b # v3.0.2
      - name: Prepare ccache timestamp
        id: ccache_cache_keys
        run: echo "::set-output name=date::`date -u +'%Y-%m-%dT%H:%M:%SZ'`"
      - name: ccache
        id: ccache
-        uses: actions/cache@v2
+        uses: actions/cache@c3f1317a9e7b1ef106c153ac8c0f00fed3ddbc0d # v3.0.4
        with:
          path: /tmp/ccache
          key: ${{github.job}}-${{matrix.nametag}}-${{steps.ccache_cache_keys.outputs.date}}
@@ -329,7 +412,7 @@ jobs:
      - name: Testsuite
        if: matrix.skip_tests != '1'
        run: src/build-scripts/ci-test.bash
-      - uses: actions/upload-artifact@v2
+      - uses: actions/upload-artifact@3cea5372237819ed00197afe530f5a7ea3e805c8 # v3.1.0
        if: failure()
        with:
          name: osl-${{github.job}}-${{matrix.nametag}}
@@ -342,88 +425,71 @@ jobs:
  macos:
    name: "${{matrix.os}} appleclang${{matrix.aclang}}/C++${{matrix.cxx_std}} py${{matrix.python_ver}} ${{matrix.desc}}"
    strategy:
+      fail-fast: false
      matrix:
        include:
-          - desc: llvm11
-            os: macos-10.15
-            nametag: macos10-py39
-            cxx_std: 14
-            python_ver: 3.9
-            aclang: 12
-          - desc: llvm11
+          - desc: MacOS-11
            os: macos-11
            nametag: macos11-py39
+            cc_compiler: /usr/local/opt/llvm@14/bin/clang
+            cxx_compiler: /usr/local/opt/llvm@14/bin/clang++
            cxx_std: 17
-            python_ver: 3.9
+            python_ver: "3.10"
            aclang: 13
+            setenvs: export LLVMBREWVER="@14" DO_BREW_UPDATE=1
+          - desc: MacOS-12
+            os: macos-12
+            nametag: macos11-p310
+            cc_compiler: /usr/local/opt/llvm@15/bin/clang
+            cxx_compiler: /usr/local/opt/llvm@15/bin/clang++
+            cxx_std: 17
+            python_ver: "3.10"
+            aclang: 13
+            setenvs: export LLVMBREWVER="@15" DO_BREW_UPDATE=1
    runs-on: ${{matrix.os}}
    env:
-      CC: clang
-      CXX: clang++
+      CXX: ${{matrix.cxx_compiler}}
+      CC: ${{matrix.cc_compiler}}
      CMAKE_CXX_STANDARD: ${{matrix.cxx_std}}
      PYTHON_VERSION: ${{matrix.python_ver}}
+      OPENIMAGEIO_VERSION: ${{matrix.openimageio_ver}}
    steps:
-      - uses: actions/checkout@v2
+      - uses: actions/checkout@2541b1294d2704b0964813337f33b291d3f8596b # v3.0.2
      - name: Prepare ccache timestamp
        id: ccache_cache_keys
        run: echo "::set-output name=date::`date -u +'%Y-%m-%dT%H:%M:%SZ'`"
      - name: ccache
        id: ccache
-        uses: actions/cache@v2
+        uses: actions/cache@c3f1317a9e7b1ef106c153ac8c0f00fed3ddbc0d # v3.0.4
        with:
          path: /Users/runner/.ccache
-          key: ${{github.job}}-${{matrix.os}}-${{steps.ccache_cache_keys.outputs.date}}
-          restore-keys: ${{github.job}}-${{matrix.os}}-
+          key: ${{github.job}}-${{matrix.nametag}}-${{steps.ccache_cache_keys.outputs.date}}
+          restore-keys: ${{github.job}}-${{matrix.nametag}}-
      - name: Build setup
-        run: src/build-scripts/ci-startup.bash
+        run: |
+            ${{matrix.setenvs}}
+            src/build-scripts/ci-startup.bash
      - name: Dependencies
        run: |
            source src/build-scripts/install_homebrew_deps.bash
-            # OPENIMAGEIO_CMAKE_FLAGS="-DOIIO_BUILD_TESTS=0 -DUSE_OPENGL=0"
-            # source src/build-scripts/build_openimageio.bash
-            brew install --display-times -q openimageio
-            PYTHONPATH=$PYTHONPATH:/usr/local/python${PYTHON_VERSION}/site-packages
+            if [[ "$OPENIMAGEIO_VERSION" != "" ]] ; then \
+                OPENIMAGEIO_CMAKE_FLAGS="-DOIIO_BUILD_TESTS=0 -DUSE_OPENGL=0" ; \
+                source src/build-scripts/build_openimageio.bash ; \
+            else \
+                brew install --display-times -q openimageio ; \
+                PYTHONPATH=$PYTHONPATH:/usr/local/python${PYTHON_VERSION}/site-packages ; \
+            fi
+            src/build-scripts/save-env.bash
      - name: Build
        run: |
            src/build-scripts/ci-build.bash
      - name: Testsuite
        run: src/build-scripts/ci-test.bash
-      - uses: actions/upload-artifact@v2
+      - uses: actions/upload-artifact@3cea5372237819ed00197afe530f5a7ea3e805c8 # v3.1.0
        if: failure()
        with:
-          name: osl-${{github.job}}
+          name: osl-${{github.job}}-${{matrix.nametag}}
          path: |
            build/*.cmake
            build/CMake*
            build/testsuite/*/*.*
-
-
-  clang-format:
-    # Test formatting. This test entry doesn't do a full build, it just runs
-    # clang-format on everything, and passes if nothing is misformatted.
-    # Upon failure, the build artifact will be the full source code with the
-    # formatting fixed (diffs will also appear in the console output).
-    name: "clang-format verification"
-    runs-on: ubuntu-18.04
-    container:
-      image: aswf/ci-osl:2021-clang11
-    env:
-      BUILDTARGET: run-clang-format
-      CMAKE_CXX_STANDARD: 17
-      PYTHON_VERSION: 3.7
-    steps:
-      - uses: actions/checkout@v2
-      - name: Build setup
-        run: |
-            src/build-scripts/ci-startup.bash
-      - name: Dependencies
-        run: |
-            src/build-scripts/gh-installdeps.bash
-      - name: Build
-        run: |
-            src/build-scripts/ci-build.bash
-      - uses: actions/upload-artifact@v2
-        if: failure()
-        with:
-          name: osl-${{github.job}}
-          path: src/*/*.{cpp,h}
@@ -0,0 +1,62 @@
+name: Scorecards supply-chain security
+on:
+  # Only the default branch is supported.
+  branch_protection_rule:
+  schedule:
+    - cron: '44 20 * * 6'
+  push:
+    branches: [ "main", "scorecard" ]
+
+# Declare default permissions as read only.
+permissions: read-all
+
+jobs:
+  analysis:
+    name: Scorecards analysis
+    runs-on: ubuntu-latest
+    permissions:
+      # Needed to upload the results to code-scanning dashboard.
+      security-events: write
+      # Used to receive a badge. (Upcoming feature)
+      id-token: write
+      # Needs for private repositories.
+      contents: read
+      actions: read
+    
+    steps:
+      - name: "Checkout code"
+        uses: actions/checkout@a12a3943b4bdde767164f792f33f40b04645d846 # v3.0.0
+        with:
+          persist-credentials: false
+
+      - name: "Run analysis"
+        uses: ossf/scorecard-action@3e15ea8318eee9b333819ec77a36aca8d39df13e # v1.1.1
+        with:
+          results_file: results.sarif
+          results_format: sarif
+          # (Optional) Read-only PAT token. Uncomment the `repo_token` line below if:
+          # - you want to enable the Branch-Protection check on a *public* repository, or
+          # - you are installing Scorecards on a *private* repository
+          # To create the PAT, follow the steps in https://github.com/ossf/scorecard-action#authentication-with-pat.
+          # repo_token: ${{ secrets.SCORECARD_READ_TOKEN }}
+
+          # Publish the results for public repositories to enable scorecard badges. For more details, see
+          # https://github.com/ossf/scorecard-action#publishing-results. 
+          # For private repositories, `publish_results` will automatically be set to `false`, regardless 
+          # of the value entered here.
+          publish_results: true
+
+      # Upload the results as artifacts (optional). Commenting out will disable uploads of run results in SARIF
+      # format to the repository Actions tab.
+      - name: "Upload artifact"
+        uses: actions/upload-artifact@6673cd052c4cd6fcf4b4e6e60ea986c889389535 # v3.0.0
+        with:
+          name: SARIF file
+          path: results.sarif
+          retention-days: 5
+      
+      # Upload the results to GitHub's code scanning dashboard.
+      - name: "Upload to code-scanning"
+        uses: github/codeql-action/upload-sarif@5f532563584d71fdef14ee64d17bafb34f751ce5 # v1.0.26
+        with:
+          sarif_file: results.sarif
@@ -9,4 +9,5 @@ build
 *.pyc
 *.log
 .vscode
+.envrc

@@ -50,4 +50,4 @@ This is a checklist for TSC's to review as part of the intake process. The TSC s
  - [X]	License scan completed and no issues found
    - Minor issues were found but already fixed in PR https://github.com/AcademySoftwareFoundation/OpenShadingLanguage/pull/1145
  - [X] Code repository imported to ASWF GitHub organization
-  - [ ] Developer Certificate of Origin past commit signoff done and DCO Probot enabled.
+  - [X] Developer Certificate of Origin post commit signoff done and DCO Probot enabled.
@@ -1,8 +1,84 @@
 <!-- SPDX-License-Identifier: CC-BY-4.0 -->
 <!-- Copyright Contributors to the Open Shading Language Project. -->

-Release 1.12 -- ?? (compared to 1.11)
+Release 1.13 -- ?? 2023 ?? (compared to 1.12)
 --------------------------------------------------
+Dependency and standards requirements changes:
+* OptiX 6.0 support has been removed. For GPU rendering with OptiX, a minimum
+  of OptiX 7.0 is required.
+* The minimum version OpenImageIO has been raised from 2.2 to 2.3. #1591
+  (1.13.0.3)
+
+OSL Language and oslc compiler:
+* Bug fix: shader params whose default values involve "init ops" could be
+  incorrectly initialized if their values came from instance parameters set to
+  zero. #1578 (1.13.0.2)
+
+OSL Standard library:
+
+API changes, new options, new ShadingSystem features (for renderer writers):
+* A new `shadingStateUniform` pointer field was added to ShaderGlobals for
+  planned future expansion and improved GPU support. #1585 (1.13.1.0)
+
+SIMD batched shading mode:
+* Fix some array overruns (asymptomatic, but still potentially buggy) in
+  common_ancestor_between. #1577 (1.13.0.2)
+
+OptiX rendering:
+* OptiX 6.0 support has been removed. For GPU rendering with OptiX, a minimum
+  of OptiX 7.0 is required.
+* Big overhaul/simplification of how strings are represented on the GPU: the
+  extra level of indirection to global variables is removed, and strings are
+  simply represented in ordinary variables as the hash of their characters.
+  #1531 (1.13.0.0) #1553 (1.13.0.1)
+* Better PTX cache hit rate by eliminating several sources of not being
+  strictly deterministic in the PTX text we were generating. #1566 #1570
+  (1.13.0.1)
+
+Performance improvements:
+
+Bug fixes and other improvements (internals):
+* Fix memory leak in oslc ASTvariable_declaration. #1576 (1.13.0.2)
+* Fix memory leaks in LPE code in error conditions. #1593 #1594 (1.13.0.3)
+
+Internals/developer concerns:
+* platform.h:
+    - New `OSL_ALLOCA` macro to allocate on the stack. #1589 (1.13.0.3)
+
+Build & test system improvements:
+* CMake build system and scripts:
+    - The version number is now a cache variable, and so can be overridden at
+      build time using `-DOSL_VERSION=...`. Use with extreme caution! #1579
+      (1.13.0.2)
+* Dependency version support:
+    - Fix some issues that came up with the new icx 2022.0 compiler. #1601
+      (1.13.1.0)
+* Testing and Continuous integration (CI) systems:
+    - Dynamic analysis using address and leak sanitizers. #1581 (1.13.0.3)
+    - `testshade --help` now prints TextureSystem options and all hardware
+      information. #1584 (1.13.0.3)
+    - Improvements to SonarCloud and coverage analysis. #1596 (1.13.0.3)
+* Platform support:
+
+Documentation:
+* `doc/RELEASING.md` documents our release process and versioning policies.
+  #1572 (1.13.0.2)
+
+
+Release 1.12 -- 1 Oct 2022 (compared to 1.11)
+----------------------------------------------
+Big Deal Changes:
+* **Batch shading**: A fully operational "batch shading" support when using
+  CPUs supporting the Intel AVX2 or AVX512 instruction set architectures,
+  allows shading 8 or 16 points at a time, accelerated by using SIMD
+  instructions. (Changes contributed by Intel.)
+* **OptiX**: A reasonable subset of the OSL language can run on NVIDIA GPUs
+  for Cuda or OptiX-based renderers.
+* **Standardized material closures**: Synchronize with MaterialX on a minimum
+  standard set of material closures. This is reflected in both the OSL spec as
+  well as reference implementations of those closures in testrender. #1533
+  #1536 #1537 #1538 #1539 #1541 #1542 #1547 #1557
+
 Dependency and standards requirements changes:
 * The minimum (and default) compilation mode is now C++14. C++17 and 20 are
  also supported. #1362 #1369 (1.12.2)
@@ -33,6 +109,8 @@ OSL Standard library:
 * Fix previously-broken color conversions from RGB to xyY. #1410 (1.12.3)
 * Fully removed MaterialX shaders (get those from the MaterialX project).
  #1450 (1.12.3)
+* Fix incorrect implementation of `matrix4 * vector4` in vector4.h #1513
+  (1.12.5.0)

 API changes, new options, new ShadingSystem features (for renderer writers):
 * Custom experimental llvm optimization levels 10, 11, 12, and 13. These
@@ -58,7 +136,7 @@ API changes, new options, new ShadingSystem features (for renderer writers):
  triggered for those cases that cannot be resolved by the end of the "runtime
  optimization" phase. #1497 (1.12.4.6)

-Continued work on experimental SIMD batched shading mode:
+SIMD batched shading mode:
 * Added support for masked operations to LLVMUtil. #1248 #1250 (1.12.0.0)
 * Add interface to ShadingSystem for batched execution. #1272 (1.12.0.1)
 * Add interface to batched RendererServices. #1276 (1.12.0.1)
@@ -83,8 +161,9 @@ Continued work on experimental SIMD batched shading mode:
 * Batched isnan, isinf, isfinite. #1456 (1.12.4.2)
 * Batched userdata and output placement. #1455 (1.12.4.2)
 * Batched pointcloud functions. #1464 (1.12.4.4)
+* Batched closures (#1500)

-Continued work on experimental OptiX rendering:
+OptiX rendering:
 * Explicitly set the OptiX pipeline stack size. #1254 (1.12.0.0)
 * CI tests now at least compile and build with USE_OPTIX=1 (though not yet
  run the tests). #1281 (1.12.0.1)
@@ -101,10 +180,12 @@ Continued work on experimental OptiX rendering:
 * Fix race conditions when compiling for OptiX 7. #1411 (1.12.3)
 * osl_get_attribute() uses device strings when compiling for OptiX.
  #1435 (1.12.3)
+* Fix issue with OptiX + clang 14 with duplicate symbols. #1561 (1.12.6.0)

 Performance improvements:
 * Less mutex locking around use and retrieval of ColorSystem related to
  doing color transforms. #1405 (1.12.3)
+* Constant folding of `startswith()`. #1507 (1.12.5.0)

 Bug fixes and other improvements (internals):
 * Fix derivatives of texture calls when only derivatives of the "alpha"
@@ -142,8 +223,14 @@ Bug fixes and other improvements (internals):
 * Print all 4 levels of version numbers in help messages. #1477 (1.12.4.4)
 * Fix GPU code generation crash. #1479 (1.12.4.4)
 * Work to transition internals to use modern std::format notation for
-  formatted output. #1487 #1490 (1.12.4.6)
-* Fix testrender memory error on teardown. #1499 (1.12.4.6)
+  formatted output. #1487 #1490 (1.12.4.6) #1504 #1506 (1.12.5.0)
+* testrender improvements:
+    * Fix testrender memory error on teardown. #1499 (1.12.4.6)
+    * Modernize the sampler in testrender, improving quality and performance.
+      #1534 (1.12.6.0)
+    * Switch to cone tracing for derivatives. #1543 (1.12.6.0)
+* Fix error that prevented correct typecheck of ternary operator. #1552
+  (1.12.6.0)

 Internals/developer concerns:
 * Use the `final` keyword in certain internal classes where applicable.
@@ -167,6 +254,12 @@ Internals/developer concerns:
  to be used in OSL 1.12. #1494 (1.12.5.0)
 * Work on gradually converting printf-style formatted strings to fmt /
  std::format style. #1493 (1.12.5.0)
+* LLVM_Util: allow naming of llvm symbols to help with IR debugging. #1530
+  (1.12.6.0)
+* clang-format the code base. #1508 #1509 #1511 #1512 #1515 #1518 #1540
+  (1.12.5.0) #1555 #1562 #1568 (1.12.6.0)
+* A round of spell-checking all the comments in the code base. #1550
+  (1.12.5.0)

 Build & test system improvements:
 * CMake build system and scripts:
@@ -208,12 +301,14 @@ Build & test system improvements:
      (1.12.4.4)
 * Dependency version support:
    - Build properly against Cuda 11 and OptiX 7.1. #1232 (1.12.0.1)
+      and up to OptiX 7.4 #1559 (1.12.6.0)
    - PugiXML build fixes on some systems. #1262 (1.12.0.1/1.11.8)
    - Cuda/OptiX back end: Add `__CUDADEVRT_INTERNAL__` define to bitcode
      generation, needed to avoid duplicate cudaMalloc symbols with CUDA9+
      #1271 (1.12.0.1)
    - Build against LLVM 11 #1274 (1.12.0.1), LLVM 12 #1351 (1.12.1.0)
      #1412 (1.12.3), LLVM 13 #1420 (1.12.3), LLVM 14 #1492 (1.12.4.6)
+      LLVM 15 #1592 (1.12.6.2)
    - Fix build break against recent OIIO master change where m_mutex field
      was removed from ImageInput. #1281 (1.12.0.1/1.11.9)
    - Work to ensure that OIIO will build correctly against the upcoming
@@ -232,6 +327,8 @@ Build & test system improvements:
      compilers. #1459 #1473 #1475. (1.12.4.4)
    - Fixes for recent OIIO TextureSystem changes that hide Imath types.
      #1476 (1.12.4.4)
+    - Changes to work correctly with OpenImageIO 2.4 and its master being
+      the start of 2.5. #1569 (1.12.6.0)
 * Testing and Continuous integration (CI) systems:
    - Eliminate the old Travis CI and Appveyor. #1334 #1338 (1.12.1)
    - Use ccache + GitHub 'cache' action to greatly speed up CI runs. #1335
@@ -240,14 +337,15 @@ Build & test system improvements:
      scripts for CentOS and Ubuntu ito a single script. #1338 (1.12.1)
    - CI tests clang11, also batched shading with clang. #1379 (1.12.2)
    - CI speeds up OIIO builds by not building its tests. #1380 (1.12.2)
-    - CI tests gcc11. #1381 (1.12.2)
+    - CI tests gcc11 #1381 (1.12.2) and gcc12 #1565 (1.12.6)
    - Deal with OpenColorIO changing its master branch to "main". #1429
      (1.12.3)
    - Fix tests for texture3d now that OIIO has dropped support for Field3D.
      #1437 (1.12.3)
    - Test against LLVM13/clang13 for the VFX Platform 2022 tests. #1434
      (1.12.3)
-    - Test against MacOS-11. #1434 (1.12.3)
+    - CI test against MacOS-11 #1434 (1.12.3) and MacOS-12 #1554 (1.12.6)
+      #1522 (1.12.5)
    - Add testsuite coverage of environment(). #1438 (1.12.3)
    - Use a project-specific environment variable `OSL_CI` instead of the more
      generic `CI` to avoid clashes with other systems. #1446 (1.12.3)
@@ -262,6 +360,10 @@ Build & test system improvements:
      #1501 (1.12.5.0)
    - Overhaul of GHA ci.yml to use strategy matrix. #1474 (1.12.4.4)
    - Test against clang 14 and fix new warnings. #1498 (1.12.5.0)
+    - Test against clang 15 and fix new warnings. #1586 (1.12.6.2)
+    - A scorecards workflow guards against a variety of security issues of
+      the CI system itself.  #1529 (1.12.6.0)
+    - Enable SonarCloud static analysis. #1551 #1556 #1563 (1.12.6.0)
 * Platform support:
    - Various Windows compile fixes. #1263 #1285 (1.12.0.1)
    - Windows+Cuda build fixes. #1292 (1.12.0.1)
@@ -281,6 +383,9 @@ Documentation:
  #1326 (1.12.1)
 * Fix missing explanation of the optional "errormessage" parameter to
  environment() and texture3d(). #1442 (1.12.3/1.11.17)
+* A SECURITY.md file explains how to confidentially report vulnerabilities.
+  #1529 (1.12.6)
+* Document all the new MaterialX closures. #1564



@@ -748,7 +853,7 @@ Build & test system improvements:
  use clang library components for preprocessing. #1089 (1.11.3)
 * Add an optional build-time `EXTRA_WARNINGS` option (default: OFF) that
  enables even more picky than usual warnings. Fix many warnings that cropped
-  up during testing, mostly related to unused function paramters. #1106 (1.11.3)
+  up during testing, mostly related to unused function parameters. #1106 (1.11.3)
 * The testsuite can run in parallel, with a number of threads set by the
  environment variable `CTEST_PARALLEL_LEVEL`. #1119 (1.11.4)
 * oslconfig.h is now generated by a cmake configure_file step, from
@@ -3,7 +3,9 @@
 # https://github.com/AcademySoftwareFoundation/OpenShadingLanguage

 cmake_minimum_required (VERSION 3.12)
-project (OSL VERSION 1.12.5.0
+
+set (OSL_VERSION "1.13.1.0" CACHE STRING "Version")
+project (OSL VERSION ${OSL_VERSION}
         LANGUAGES CXX C
         HOMEPAGE_URL "https://github.com/AcademySoftwareFoundation/OpenShadingLanguage")
 set (PROJ_NAME ${PROJECT_NAME})    # short name
@@ -173,9 +173,16 @@ it really only applies to the particular file in which it appears.

 #### Formatting

-NEVER alter somebody else's code to reformat just because you found
-something that violates the rules. Let the group/author/leader know, and
-resist the temptation to change it yourself.
+We use clang-format to enforce formatting rules for the vast majority of the
+code base. It is automatically run during CI and is a CI failure if the code
+does not conform. There are, however, parts of the code base that are not
+clang-formatted, either because we have not yet put them under the
+clang-format regime, or because individual modules were deemed to be less
+clear with clang-format than with hand formatting.
+
+For non-clang-format regions of code, NEVER alter somebody else's code to
+reformat just because you found something that violates the rules. Let the
+group/author/leader know, and resist the temptation to change it yourself.

 Each line of text in your code should be at most 80 characters long.
 Exceptions are allowed for those rare cases where letting a line be longer
@@ -217,13 +224,10 @@ Function calls should have a space between the function name and the opening
 parenthesis, NO space inside the parenthesis, except for a single blank
 space between each argument. For example:

-    x = foo (a, b);     // Yes, this is always ok
+    x = foo(a, b);      // Yes
    
    x = foo ( a, b );   // No
    x = foo (a,b);      // No
-    x = foo(a, b);      // No
-    x = foo(a);         // Occasionally, this just looks better, when the function name is short,
-                        //    and there's just one very short argument.  What can I say, I do it too.

 Function declarations: function names should begin at column 0 for a full
 function definition. (It's ok to violate this rule for very short inline
@@ -233,7 +237,7 @@ functions within class definitions.)
 Here is a short code fragment that shows some of these rules in action:

    static int
-    function (int a, int b)
+    function(int a, int b)
    {
        int x = a + b;
        if (a == 0 || b == 0) {
@@ -244,7 +248,7 @@ Here is a short code fragment that shows some of these rules in action:
        }
        for (int i = 0;  i < 3;  ++i) {
            x += a * i;
-            x *= foo (i);  // function call
+            x *= foo(i);  // function call
        }
        return x;
    }
@@ -12,16 +12,16 @@ Dependencies
 OSL requires the following dependencies or tools.
 NEW or CHANGED dependencies since the last major release are **bold**.

-* Build system: [CMake](https://cmake.org/) 3.12 or newer (tested through 3.23)
+* Build system: [CMake](https://cmake.org/) 3.12 or newer (tested through 3.24)

-* A suitable **C++14 or C++17** compiler to build OSL itself, which may be any of:
-   - **GCC 6.1 or newer** (tested through gcc 11.2)
-   - Clang 3.4 or newer (tested through clang 14)
-   - Microsoft Visual Studio **2017 or newer**
-   - Intel C++ compiler icc **version 17 or newer** or LLVM-based icx compiler
-     **version 2022 or newer**.
+* A suitable C++14 or C++17 compiler to build OSL itself, which may be any of:
+   - GCC 6.1 or newer (tested through gcc 12.1)
+   - Clang 3.4 or newer (tested through clang 15)
+   - Microsoft Visual Studio 2017 or newer
+   - Intel C++ compiler icc version 17 or newer or LLVM-based icx compiler
+     version 2022 or newer.

-* **[OpenImageIO](http://openimageio.org) 2.2 or newer** (tested through 2.3)
+* **[OpenImageIO](http://openimageio.org) 2.3 or newer** (tested through 2.4)

    OSL uses OIIO both for its texture mapping functionality as well as
    numerous utility classes.  If you are integrating OSL into an existing
@@ -40,10 +40,14 @@ NEW or CHANGED dependencies since the last major release are **bold**.
    $OpenImageIO_ROOT/lib to be in your LD_LIBRARY_PATH (or
    DYLD_LIBRARY_PATH on OS X).

-* **[LLVM](http://www.llvm.org) 9, 10, 11, 12, 13, or 14**, including clang
-  libraries.
+* [LLVM](http://www.llvm.org) 9, 10, 11, 12, 13, 14, or 15, including
+  clang libraries.

-* [Boost](https://www.boost.org) 1.55 or newer (tested through boost 1.78)
+* (optional) For GPU rendering on NVIDIA GPUs:
+    * [OptiX](https://developer.nvidia.com/rtx/ray-tracing/optix) 7.0 or higher.
+    * [Cuda](https://developer.nvidia.com/cuda-downloads) 8.0 or higher.
+
+* [Boost](https://www.boost.org) 1.55 or newer (tested through boost 1.80)
 * [Ilmbase or Imath](https://github.com/AcademySoftwareFoundation/openexr) 2.3
   or newer (recommended: 2.4 or higher; tested through 3.1)
 * [Flex](https://github.com/westes/flex) 2.5.35 or newer and
@@ -57,8 +61,8 @@ NEW or CHANGED dependencies since the last major release are **bold**.
  be operative.
 * (optional) Python: If you are building the Python bindings or running the
  testsuite:
-    * Python >= 2.7 (tested against 2.7, 3.7, 3.8, 3.9)
-    * pybind11 >= 2.4.2 (Tested through 2.9)
+    * Python >= 2.7 (tested against 2.7, 3.7, 3.8, 3.9, 3.10)
+    * pybind11 >= 2.4.2 (Tested through 2.10)
    * NumPy
 * (optional) Qt >= 5.6 (tested through 5.15).  If not found at build time,
  the `osltoy` application will be disabled.
@@ -93,19 +97,14 @@ Here are the steps to check out, build, and test the OSL distribution:
   full errors, not warnings.

 4. After compilation, you'll end up with a full OSL distribution in
-   dist/ARCH, where ARCH is the architecture you are building on, one of
-   "linux", "linux64", "macosx", "windows", or "windows64".
+   dist/

-   Note: The default is to make an optimized "release" build.  If
-   instead type 'make debug' at the top level, you will end up with
-   a debug build (no optimization, full symbols) in "dist/ARCH.debug".
-
-5. Add the "dist/ARCH/bin" to your $PATH, and "dist/ARCH/lib" to your
-   $LD_LIBRAY_PATH (or $DYLD_LIBRARY_PATH on OS X), or copy the contents
+5. Add the "dist/bin" to your $PATH, and "dist/lib" to your
+   $LD_LIBRAY_PATH (or $DYLD_LIBRARY_PATH on MacOS), or copy the contents
   of those files to appropriate directories.  Public include files
   (those needed when building applications that incorporate OSL)
-   can be found in "dist/ARCH/include", and documentation can be found
-   in "dist/ARCH/share/doc".
+   can be found in "dist/include", and documentation can be found
+   in "dist/share/doc".

 6. After building (and setting your library path), you can run the
   test suite with:
@@ -294,7 +294,7 @@ package_source: install
 # 'make clang-format' runs clang-format on all source files (if it's installed)
 clang-format: config
 	@ ( cd ${build_dir} ; \
-	    ${CMAKE} --build . --target run-clang-format --config ${CMAKE_BUILD_TYPE} \
+	    ${CMAKE} --build . --target clang-format --config ${CMAKE_BUILD_TYPE} \
 	  )


@@ -393,6 +393,18 @@ Films using OSL (grouped by year of release date):
  Matrix Resurrections
 * **(2022 / upcoming)**
  Hotel Transylvania 4: Transformania,
+  Death on the Nile,
+  Uncharted,
+  Turning Red,
+  The Adam Project,
+  Doctor Strange in the Multiverse of Madness,
+  Love, Death, and Robots: In Vaulted Halls Entombed,
+  Jurassic World: Dominion,
+  Lightyear,
+  Thor: Love and Thunder,
+  The Sea Beast,
+  DC League of Super Pets,
+  Bullet Train,
  ...


@@ -427,9 +439,13 @@ certain is a legit problem in the code, and for which you can give clear
 instructions for how to reproduce, should be [reported as
 issues](https://github.com/AcademySoftwareFoundation/OpenShadingLanguage/issues).

-If confidentiality precludes a public question or issue, you may contact the
-project administrator privately at [lg@imageworks.com](lg@imageworks.com),
-including for security-related issues.
+If you think you've found a potential vulnerability in OSL, please
+confidentially report it by emailing the project administrators at
+[security@openshadinglanguage.org](security@openshadinglanguage.org).
+
+If any other issue requires confidentiality that precludes a public question
+or issue, you may contact the project administrator privately at
+[lg@larrygritz.com](lg@larrygritz.com).


 Contributing
@@ -453,7 +469,6 @@ Contacts, Links, and References

 [OSL home page](http://openshadinglanguage.org)

-[Sony Pictures Imageworks main open source page](http://opensource.imageworks.com)



@@ -466,19 +481,18 @@ The current project leadership is documented in the
 Many people have contributed features, bug fixes, and other changes to OSL
 over the years: Steve Agland, Shane Ambler, Martijn Berger, Farchad
 Bidgolirad, Nicholas Bishop, Solomon Boulos, Stefan Bruens, Stefan Büttner,
-Matthaus G. Chajdas, Clark Chen, Alejandro Conty,
-Dieter De Baets, Thomas Dinges, Mads Drøschler,
-Luke Emrose, Louis Feng, Mark Final, Henri Fousse, Syoyo Fujita,
-Tim Grant, Larry Gritz, Nicolas Guiard, Euan Haahr, Derek Haase, Sven-Hendrik
-Haase, John Haddon, Niklas Harrysson, Daniel Heckenberg, Thiago Ize, Matt
-Johnson, Ronan Keryell, Chris Kulla, Elvic Liang, Max Liani, Adam Martinez,
-John Mertic, Bastien Montagne, Steena Monteiro, Alexis Oblet, Erich Ocean,
-Mikko Ohtamaa, Jino Park, Alexei Pawlow, Jay Reynolds, Declan Russell, Benoit
-Ruiz, Patrick Scheibe, Alex Schworer, Jonathan Scruggs, Sergey Sharybin,
-Sandip Shukla, Cliff Stein, Stephan Steinbach, Luya Tshimbalanga, Esteban
-Tovagliari, Brecht Van Lommel, Alexander von Knorring, Alex Wells, Roman
-Zulak. (Listed alphabetically; if we've left anybody out, it is inadvertent,
-please let us know.)
+Matthaus G. Chajdas, Clark Chen, Alejandro Conty, Dieter De Baets, Thomas
+Dinges, Mads Drøschler, Luke Emrose, Louis Feng, Mark Final, Henri Fousse,
+Stephen Friedman, Syoyo Fujita, Tim Grant, Larry Gritz, Nicolas Guiard, Euan
+Haahr, Derek Haase, Sven-Hendrik Haase, John Haddon, Niklas Harrysson, Daniel
+Heckenberg, Adrien Herubel, Thiago Ize, Matt Johnson, Ronan Keryell, Chris
+Kulla, Elvic Liang, Max Liani, Adam Martinez, John Mertic, Bastien Montagne,
+Steena Monteiro, Alexis Oblet, Erich Ocean, Mikko Ohtamaa, Jino Park, Alexei
+Pawlow, Jay Reynolds, Declan Russell, Benoit Ruiz, Patrick Scheibe, Alex
+Schworer, Jonathan Scruggs, Sergey Sharybin, Sandip Shukla, Cliff Stein,
+Stephan Steinbach, Luya Tshimbalanga, Esteban Tovagliari, Brecht Van Lommel,
+Alexander von Knorring, Alex Wells, Roman Zulak. (Listed alphabetically; if
+we've left anybody out, it is inadvertent, please let us know.)

 We cannot possibly express sufficient gratitude to the managers at Sony
 Pictures Imageworks who allowed this project to proceed, supported it
@@ -0,0 +1,27 @@
+# Security Policy
+
+## Supported Versions
+
+This gives guidance about which branches are supported with patches to
+security vulnerabilities.
+
+| Version / branch  | Supported                                            |
+| --------- | ---------------------------------------------------- |
+| main      | :white_check_mark: :construction: ALL fixes immediately, but this is a branch under development with a frequently unstable ABI and occasionally unstable API. |
+| 1.12.x    | :white_check_mark: All fixes that can be backported without breaking ABI compatibility. New tagged releases monthly. |
+| 1.11.x    | :warning: Only the most critical fixes, only if they can be easily backported. |
+| <= 1.10.x | :x: No longer receiving patches of any kind. |
+
+
+## Reporting a Vulnerability
+
+If you think you've found a potential vulnerability in OSL, please report it
+by emailing the project administrators at
+[security@openshadinglanguage.org](security@openshadinglanguage.org). Only the
+project administrators have access to these messages. Include detailed steps to
+reproduce the issue, and any other information that could aid an
+investigation. Our policy is to respond to vulnerability reports within 14
+days.
+
+Our policy is to address critical security vulnerabilities rapidly and post
+patches as quickly as possible.
@@ -2,7 +2,7 @@

 The OSL Documentation Project is a work in progress.

-The plan is to adopt the OCIO Documentation Project's methodology for maintaining and building documention of various kinds as it becomes the blueprint for the ASWF.
+The plan is to adopt the OCIO Documentation Project's methodology for maintaining and building documentation of various kinds as it becomes the blueprint for the ASWF.

 There is much work to be done here. 

@@ -0,0 +1,341 @@
+<!-- SPDX-License-Identifier: CC-BY-4.0 -->
+<!-- Copyright Contributors to the Open Shading Language Project. -->
+
+# Release Procedures for Open Shading Language
+
+## Version nomenclature and release cadence
+
+### The meaning of the version parts
+
+Our release numbers are of the form `MAJOR.MINOR.PATCH.TWEAK`, and we follow
+an almost-semantic-versioning scheme.
+
+MAJOR changes are those that break backward-compatibility of the public APIs,
+requiring existing application code to change their source code where it calls
+OSL API functions in order to compile and work correctly.
+
+MINOR changes guarantee API backward-compatibility with prior releases that
+have the same major number, but can introduce ABI or link incompatibilities.
+Applications using OSL need to be recompiled to move to a new minor release,
+but should not need to change their source code. Minor releases are allowed
+to drop support for sufficiently old compilers and dependencies.
+
+PATCH changes guarantee both API and ABI/link backward-compatibility with
+prior releases that have the same minor number. Features and new public API
+calls may be added (thus deviating from strict semantic versioning), but only
+if they preserve ABI/link compatibility. This means that standalone functions
+or static methods may be added, but existing functions will not change their
+parameters, and no new virtual methods or struct data layouts change for patch
+releases (for classes in public header files). Application code can be
+re-linked to a new patch release without needing to be recompiled. We also
+guarantee that any compilers or dependencies supported for the first release
+of a minor series will continue to be supported for all patches to that minor
+series.
+
+TWEAK changes are restricted to critical bug fixes or build breaks that do
+not alter public APIs in any way.
+
+### Cadence of releases
+
+Major (arbitrarily incompatible) releases happen rarely, usually with multiple
+years between them.
+
+Minor (API backward-compatible) releases are scheduled annually, targeted at
+late summer (SIGGRAPH time) to be in beta, and early autumn for a full
+release.
+
+Patch (API and ABI backward-compatible) releases are scheduled monthly,
+usually aiming for the first day of each month.
+
+Tweak (fully forward and backward compatible) releases are not regularly
+scheduled, and only happen sporadically, when a critical fix is needed in
+between scheduled patch releases.
+
+### Things are different in "main"
+
+The "main" branch is where ongoing development occurs without any
+compatibility guarantees. We don't always know if the next year's release from
+main will turn out to be major (API breaking) or minor (only ABI breaking).
+
+Some studios or products need the most up-to-date enhancements, and thus tend
+to build OSL from the "main" branch rather than using tagged releases. For
+these users, we try to set version numbers and occasionally tag in main, to
+send a signal about which points in main seem safe to rely on various
+compatibility levels. However, the rules are slightly different than we use
+for releases.
+
+In the "main" branch:
+
+- TWEAK changes within "main" only guarantee ABI back-compatibility, but may
+  have additions or non-bug-fix behavior changes (akin to patch releases
+  within a release branch).
+- PATCH changes within "main" are allowed to break ABI or API (changes that
+  would require minor or major releases, if they were not in "main").
+- Beware of unmarked breaks in compatibility of commits in main that are
+  in between developer preview tags.
+
+
+---
+
+## Branch and tag names
+
+**Branches** are places where commits containing new development may be added
+(and thus the branch markers will move over time). Branch names obey the
+following conventions:
+
+- `main` is the area for arbitrary changes intended for the next year's major
+  or minor release.
+- `dev-1.2` are the areas for staging additions to the next month's patch
+  release for the designated minor version series.
+- `release` is a special branch marker that is always set to the latest tag of
+  the currently supported, stable release family. People can count on
+  "release" always being the latest stable release, whatever that may be at
+  the time.
+
+**Tags** indicate releases or developer previews, and once a commit is
+tagged, it is permanent and will never be moved. Tag names obey the following
+conventions:
+
+- `v1.2.3.4` is a specific public stable release, corresponding to official
+  major, minor, public, or tweak releases.
+- `v1.2.3.4-alpha`, `v1.2.3.4-beta`, or `v1.2.3.4-RC1` are pre-releases for an
+  upcoming public stable release, at the alpha, beta, or release candidate
+  stages, respectively.
+- `v1.2.3.4-dev` is a "developer preview" within "main." It is not an official
+  supported release, and while the numbering may indicate compatibility versus
+  other recent developer previews, they should not be assumed to be compatible
+  with any prior official stable releases.
+
+*Very* occasionally, other branch or tag names (with custom suffixes or
+different naming schemes) may be used for special purposes by particular
+developers or users, but should not be considered reliable or permanent, and
+in general should be ignored by anyone who didn't make the tag or branch.
+
+---
+
+## Release Procedures
+
+### Soft freeze for monthly tweak releases:
+
+We freely backport *safe*, non-ABI-breaking changes by cherry-picking them
+from main to the supported branch continually throughout most of the month.
+
+In the seven days before the new month, we do a "soft freeze" where only
+important bug fixes are backported. Hold on to any other patches you wish to
+backport until after the monthly patch release (i.e., save them for the next
+month's release so you never have a release that contains changes that are
+only a few days old, and therefore not as well tested). You may, of course,
+continue to backport fixes to documentation or other non-code parts of the
+project during this soft freeze.
+
+### Branching for annual major/minor releases
+
+By mid-June (about 1-2 months before the beta of a new annual big release),
+you should branch and mark it as an alpha.
+
+At the tip of "main," create a new branch named `dev-MAJOR.MINOR` that will be
+the staging area for the next release.
+
+Customarily, at that point we add a commit to main (but not to the new
+dev-MAJ.MIN branch) that bumps main's version to the next minor level, to
+avoid any confusion between builds from main versus what will be the next
+release branch. This starts the divergence between main and the release
+branch, and henceforth, any big or compatibility-breaking changes will only be
+committed to main and not backported to the release branch. (Though we are
+loose about this rule during the alpha period and allow continued breaking
+changes in the alpha, but by the time we call it beta, we allow no more
+breaking changes.)
+
+
+### Prior to a release
+
+1. Update CI dependencies: Ensure that the "latest" CI test specifies the
+   current releases of the dependencies and compilers (check their home
+   pages), and that the [INSTALL.md](../INSTALL.md) file correctly documents
+   the version ranges that we expect to work and that we actively test.
+
+2. Release notes: [CHANGES.md](../CHANGES.md) should be updated to reflect
+   all the new features and fixes. Looking at the notes from older releases
+   should provide a good example of what we're aiming for.
+
+   - Patch and tweak releases can be brief one-line descriptions (often the
+     first line of the commit messages involved), presented in a single
+     section in either chronological or any other order that seems right.
+   - Minor or major releases have a much more extensive, prose-based
+     description of all changes since the last minor release, broken into
+     sections (dependency changes, major new features, enhancements and fixes,
+     internals and developer goodies, testing/CI/ports, etc.) and ordered for
+     readability and relevance.
+   
+3. Ensure docs are up to date:
+
+   - Actually read the [README.md](../README.md), [INSTALL.md](../INSTALL.md),
+     and other repository docs and make sure all the information is up to
+     date. Check the "contributor credits" in the README against the recent
+     repo history to be sure new contributors are listed. Also check the
+     list of films and make sure it includes any recent productions that are
+     known to have used OSL.
+   - Skim the primary user documentation to look for any obvious errors,
+     especially the parts that describe new features. In the case of PDF docs
+     (such as osl-languagespec.pdf), make sure that a new PDF book is built if
+     the documentation source has been changed since the last release.
+   - If we are using any online docs (like readthedocs), make sure it is
+     building and looks correct on the web site.
+
+4. Make sure the the top-level CMakeLists.txt file is updated:
+
+   - The version on the `project()` call should be correct.
+   - The `PROJECT_VERSION_RELEASE_TYPE` variable should be set to "alpha" or
+     "beta" if appropriate, "RC" for release candidate, or empty string `""`
+     for most supported releases, and "dev" only for the main branch or
+     developer previews.
+   - The `${PROJECT_NAME}_SUPPORTED_RELEASE` variable should be `ON` for any
+     release branch, `OFF` for main.
+
+5. Make sure everything passes the usual CI workflow. Also check the daily or
+   weekly "analysis" workflows to make sure there aren't any important
+   warnings that should be fixed.
+
+
+### Staging alpha / beta / release candidate (annual minor/major releases only)
+
+For monthly patch releases, we don't have any formal alpha/beta/RC stages.
+
+For the annual minor or major releases we have a staged release:
+
+- `alpha` is everything after we have branched, in which we think all the big
+  new features have been added, and while we will make continued changes and
+  fixes, we are aiming for increased stability and to minimize additional
+  breaking changes prior to the release.
+- `beta` should be about one month prior to the final release date. We create
+  a tag `v1.2.3.4-beta` and announce the beta (see below "Making the
+  release"). Once we tag a beta, we try very hard not to allow any further
+  changes that change the API or are not ABI-compatible (i.e., safe fixes
+  only).
+- `RC` or release candidates around one week prior to the final release date.
+  This is a trial run, hopefully identical to the final release (except for
+  removing the RC label and making a new tag). The only changes allowed after
+  RC are fixes to the most critical bugs or build breaks that would be bad for
+  users to encounter. If critical problems are found with RC1, they should be
+  fixed and immediately re-tagged as RC2, etc. All other fixes should be
+  postponed until the next month's patch release.
+
+### Making the release
+
+A final release is made on the scheduled date, provided the latest RC has
+survived for at least a few days of testing without finding any critical
+problems. Don't make the release final until you are sure there are no
+truly critical bugs or build breaks that users will encounter.
+
+The following are the steps for making the release:
+
+1. Edit the top-level CMakeLists.txt to remove any RC designation
+   (i.e., `PROJECT_VERSION_RELEASE_TYPE` should be set to `""`).
+
+2. Edit CHANGES.md to reflect the correct date of the release and ensure it
+   includes any last-minute changes that were made during beta or release
+   candidate stages.
+
+3. Push it to **your** GitHub, make sure it passes CI.
+   
+4. Tag the release: `git tag v1.2.3.4` (no more beta, RC, or dev suffix).
+   
+5. If this will now be the recommended stable release, move the `release`
+   branch marker to the same position.
+
+6. Push it to GitHub: `git push aswf release --tags`
+
+   (This example assumes "aswf" is the name of the remote for the GitHub
+   `AcademySoftwareFoundation/OpenShadingLanguage` repo.)
+
+7. Draft a release on GitHub: On
+   https://github.com/AcademySoftwareFoundation/OpenShadingLanguage/releases
+   select "Draft a new release." Choose the new tag you just pushed. Make the
+   release title "Open Shading Language v1.2.3.4" (and beta, etc., designation
+   if applicable). In the description, paste the release notes for this
+   release (from Changes.md). If this is a beta or release candidate, check
+   "this is a pre-release" box at the bottom.
+
+8. Announce the release on the [osl-dev mail list](https://lists.aswf.io/g/osl-dev)
+   with the subject "Release: OSL v1.2.3.4" and using one of these templates
+   for the body of the email.
+
+   For a monthly patch release:
+
+    > We have tagged v1.2.3.4 as the latest production release and moved the
+    > "release" branch marker to that point.  This is guaranteed to be API,
+    > ABI, and link back-compatible with prior 1.2 releases. Release notes
+    > are below.
+    >
+    > (Paste the full set of 1.2 changes here, just copy the appropriate
+    > part of CHANGES.md)
+
+   For an annual major/minor release:
+
+    > OSL version 1.2 has been released! Officially tagged as "v1.2.3.4", we
+    > have also moved the "release" branch tag to this position. Henceforth,
+    > 1.2 is the supported production release family. The API is now frozen --
+    > we promise that subsequent 1.2.x releases (which should happen monthly)
+    > will not break back-compatibility of API, ABI, or linkage, compared to
+    > this release. Please note that this release is *not* ABI or link
+    > compatible with 1.1 or older releases.
+    > 
+    > Release notes for 1.2 outlining all the changes since last year's
+    > release are below.
+    > 
+    > Please note that a few of the build and runtime dependencies have
+    > changed their minimum supported versions. (List here any important
+    > changes to dependencies, compilers, or C++ standard that users should be
+    > aware of.)
+    > 
+    > (List here anything else that people should know about this release
+    > family that may be surprising if they haven't followed the last year of
+    > development closely, or that they must know even if they are too lazy to
+    > read the release notes. If this is a major release that is not
+    > backward-compatible with prior versions, warn about that here.)
+    > 
+    > Enjoy, and please report any problems. We will continue to make patch
+    > releases to the 1.2 family roughly monthly, which will contain bug fixes
+    > and non-breaking enhancements.
+    > 
+    > The older 1.1 series of releases is now considered obsolete. We will
+    > continue for now to make 1.1 patch releases, but over time, these will
+    > become less frequent and be reserved for only the most critical bug
+    > fixes.
+    > 
+    > The "main" branch is now progressing toward an eventual 1.3 release next
+    > summer. As usual, you are welcome to use main for real work, but we do
+    > not make any compatibility guarantees and don't guarantee continuing API
+    > compatibility in main.
+    >
+    > (Paste the full set of 1.2 changes here, just copy the appropriate
+    > part of CHANGES.md)
+
+   For a beta leading up to the annual major/minor release:
+
+    > OSL version 1.2 is now in beta, tagged as "v1.2.3.4-beta". We
+    > will try very hard not to make any further API or ABI changes between
+    > now and the final release (unless it is absolutely necessary to fix
+    > an important problem identified during beta testing). The final 1.2
+    > release is scheduled for [DATE GOES HERE], so please try building and
+    > testing the beta so we are sure to find any problems.
+    > 
+    > Release notes for 1.2 outlining all the changes since last year's
+    > release are below.
+    > 
+    > (Paste the full set of 1.2 changes here, just copy the appropriate
+    > part of CHANGES.md)
+
+
+### After the release
+
+Odds and ends to do after the tag is pushed and the announcements are sent:
+
+- Re-read RELEASING.md and ensure that the instructions match what you
+  have done. Update as necessary.
+
+- For annual beta releases, post a short message to the "Open Shading
+  Language" channel of the ASWF slack. It can be a one-line announcement and
+  then just a link to the GitHub release page for this version.
+
+
@@ -181,14 +181,14 @@ else ifeq (${platform}, macosx)
        MY_CMAKE_FLAGS += \
           -DCMAKE_C_COMPILER=gcc-9 -DCMAKE_CXX_COMPILER=g++-9
        USE_LIBCPLUSPLUS := 0
-    else ifeq (${COMPILER},clang11)
+    else ifeq (${COMPILER},clang14)
        MY_CMAKE_FLAGS += \
-            -DCMAKE_C_COMPILER=/usr/local/opt/llvm@11/bin/clang \
-            -DCMAKE_CXX_COMPILER=/usr/local/opt/llvm@11/bin/clang++
-    else ifeq (${COMPILER},clang12)
+            -DCMAKE_C_COMPILER=/usr/local/opt/llvm@14/bin/clang \
+            -DCMAKE_CXX_COMPILER=/usr/local/opt/llvm@14/bin/clang++
+    else ifeq (${COMPILER},clang15)
        MY_CMAKE_FLAGS += \
-            -DCMAKE_C_COMPILER=/usr/local/opt/llvm@12/bin/clang \
-            -DCMAKE_CXX_COMPILER=/usr/local/opt/llvm@12/bin/clang++
+            -DCMAKE_C_COMPILER=/usr/local/opt/llvm@15/bin/clang \
+            -DCMAKE_CXX_COMPILER=/usr/local/opt/llvm@15/bin/clang++
    else ifeq (${COMPILER},clang)
        MY_CMAKE_FLAGS += -DCMAKE_C_COMPILER=clang -DCMAKE_CXX_COMPILER=clang++
    else ifeq (${COMPILER},gcc)
@@ -0,0 +1,28 @@
+# SPDX-License-Identifier: BSD-3-Clause
+# Copyright Contributors to the Open Shading Language Project.
+
+sonar.host.url=https://sonarcloud.io
+
+# Required metadata
+sonar.organization=academysoftwarefoundation
+sonar.projectName=OpenShadingLanguage
+sonar.projectVersion=1.12
+sonar.projectKey=AcademySoftwareFoundation_OpenShadingLanguage
+
+# Project links
+sonar.links.homepage=https://github.com/AcademySoftwareFoundation/OpenShadingLanguage
+sonar.links.ci=https://github.com/AcademySoftwareFoundation/OpenShadingLanguage/actions
+sonar.links.scm=https://github.com/AcademySoftwareFoundation/OpenShadingLanguage
+sonar.links.issue=https://github.com/AcademySoftwareFoundation/OpenShadingLanguage/issues
+
+# Source properties
+sonar.sources=src
+sonar.sourceEncoding=UTF-8
+sonar.exclusions=src/include/OSL/Imath/*.h,src/build-scripts/*
+
+# C/C++ analyzer properties
+sonar.cfamily.build-wrapper-output=build/bw_output
+sonar.cfamily.gcov.reportsPath=_coverage
+
+sonar.cfamily.cache.enabled=true
+sonar.cfamily.cache.path=/tmp/sonarcache
@@ -12,6 +12,10 @@ if [[ "$USE_SIMD" != "" ]] ; then
    OSL_CMAKE_FLAGS="$OSL_CMAKE_FLAGS -DUSE_SIMD=$USE_SIMD"
 fi

+if [[ -n "$CODECOV" ]] ; then
+    OSL_CMAKE_FLAGS="$OSL_CMAKE_FLAGS -DCODECOV=${CODECOV}"
+fi
+
 pushd build
 cmake .. -G "$CMAKE_GENERATOR" \
        -DCMAKE_BUILD_TYPE="${CMAKE_BUILD_TYPE}" \
@@ -23,7 +27,16 @@ cmake .. -G "$CMAKE_GENERATOR" \
        -DCMAKE_INSTALL_LIBDIR="$OSL_ROOT/lib" \
        -DCMAKE_CXX_STANDARD="$CMAKE_CXX_STANDARD" \
        $OSL_CMAKE_FLAGS -DVERBOSE=1
-time cmake --build . --target ${BUILDTARGET:=install} --config ${CMAKE_BUILD_TYPE}
+
+# Save a copy of the generated files for debugging broken CI builds.
+mkdir cmake-save || /bin/true
+cp -r CMake* *.cmake cmake-save
+
+: ${BUILDTARGET:=install}
+if [[ "$BUILDTARGET" != "none" ]] ; then
+    echo "Parallel build ${CMAKE_BUILD_PARALLEL_LEVEL} of target ${BUILDTARGET}"
+    time ${OSL_CMAKE_BUILD_WRAPPER} cmake --build . --target ${BUILDTARGET} --config ${CMAKE_BUILD_TYPE}
+fi
 popd

 if [[ "${DEBUG_CI:=0}" != "0" ]] ; then
@@ -0,0 +1,34 @@
+#!/usr/bin/env bash
+
+# Run code coverage analysis
+# This assumes that the build occurred with CODECOV=1 andtests have already
+# fully run.
+
+set -ex
+
+echo "Performing code coverage analysis"
+mkdir _coverage
+pushd _coverage
+
+ls -R ../build/src/
+
+# Remove files or directories we want to exclude from code coverage analysis,
+# generally because we don't expect to execute any of their code during our
+# CI. (Because it's only used interactively, or is fallback code only used
+# when certain dependencies are not found.) Including these when we know we
+# won't be calling them during coverage gathering will just make our coverage
+# percentage look artifically low.
+rm -f ../build/src/osltoy/CMakeFiles/osltoy.dir/*.cpp.{gcno,gcda}
+
+
+# The sed command below converts from:
+#   ../build/src/liboslexec/CMakeFiles/oslexec.dir/foo.gcno
+# to:
+#   ../src/liboslexec/foo.cpp
+
+for g in $(find ../build -name "*.gcno" -type f); do
+    echo "Processing $g"
+    echo "dirname $g = $(dirname $g) to " `$(echo "$g" | sed -e 's/\/build\//\//' -e 's/\.gcno/\.cpp/' -e 's/\.cpp\.cpp/\.cpp/' -e 's/\/CMakeFiles.*\.dir\//\//')`
+    gcov -l -p -o $(dirname "$g") $(echo "$g" | sed -e 's/\/build\//\//' -e 's/\.gcno/\.cpp/' -e 's/\/CMakeFiles.*\.dir\//\//')
+done
+popd
@@ -23,7 +23,7 @@ export DYLD_LIBRARY_PATH=$OSL_ROOT/lib:$DYLD_LIBRARY_PATH
 export LD_LIBRARY_PATH=$OSL_ROOT/lib:$LD_LIBRARY_PATH
 export OIIO_LIBRARY_PATH=$OSL_ROOT/lib:${OIIO_LIBRARY_PATH}
 export LSAN_OPTIONS=suppressions=$PWD/src/build-scripts/nosanitize.txt
-export ASAN_OPTIONS=print_suppressions=0
+export ASAN_OPTIONS=print_suppressions=0:detect_odr_violation=1

 export USE_PYTHON=${USE_PYTHON:=1}
 export PYTHON_VERSION=${PYTHON_VERSION:="2.7"}
@@ -61,6 +61,7 @@ echo "uname -a: " `uname -a`
 echo "uname -m: " `uname -m`
 echo "uname -s: " `uname -s`
 echo "uname -n: " `uname -n`
+echo "nproc " `nproc`
 pwd
 ls
 env | sort
@@ -8,12 +8,26 @@
 # any command in it fails. This is crucial for CI tests.
 set -ex

+: ${CTEST_EXCLUSIONS:="broken"}
+: ${CTEST_TEST_TIMEOUT:=60}
+
 $OSL_ROOT/bin/testshade --help

-echo "Parallel test " ${CTEST_PARALLEL_LEVEL}
+echo "Parallel test ${CTEST_PARALLEL_LEVEL}"
+echo "Default timeout ${CTEST_TEST_TIMEOUT}"
+echo "Test exclusions '${CTEST_EXCLUSIONS}'"
+echo "CTEST_ARGS '${CTEST_ARGS}'"
+
 pushd build
-time ctest -C ${CMAKE_BUILD_TYPE} -E broken --force-new-ctest-process \
-    --output-on-failure --timeout ${CTEST_TEST_TIMEOUT:=60} ${CTEST_ARGS}
+
+ctest -C ${CMAKE_BUILD_TYPE} --force-new-ctest-process --output-on-failure \
+    --timeout ${CTEST_TEST_TIMEOUT} -E "${CTEST_EXCLUSIONS}" ${CTEST_ARGS} \
+  || \
+ctest -C ${CMAKE_BUILD_TYPE} --force-new-ctest-process \
+    --rerun-failed --repeat until-pass:5 -R render --output-on-failure \
+    --timeout ${CTEST_TEST_TIMEOUT} -E "${CTEST_EXCLUSIONS}" ${CTEST_ARGS}
+# The weird construct above allows the render-* tests to run multiple times
+
 popd


@@ -73,7 +73,7 @@ from __future__ import print_function, absolute_import
 ###
 ### Only one main doc region will be output by the `docdeep` program,
 ### specified with the `-d` command line argument. Any other doc-regions
-### will not be included in the documention output of that run.
+### will not be included in the documentation output of that run.
 ###
 ### However, one doc-region may *include* the text of another doc-region
 ### as follows:
@@ -114,7 +114,7 @@ from __future__ import print_function, absolute_import
 ### For pre-code comments, the contiguous comment region associated with
 ### a declaration immediately precedes the declaration. Such a comment
 ### set has its first line start with `///>`.  For post-code comments,
-### the documentaiton comments follows the code declaration, and this is
+### the documentation comments follows the code declaration, and this is
 ### designated by having its first line start with `///<`.  I like to
 ### just remember that the `<` or `>` points in the direction of the code
 ### declaration that the comment applies to.
@@ -42,8 +42,12 @@ if [[ "$ASWF_ORG" != ""  ]] ; then

    if [[ "$CXX" == "icpc" || "$CC" == "icc" || "$USE_ICC" != "" || "$CXX" == "icpx" || "$CC" == "icx" || "$USE_ICX" != "" ]] ; then
        sudo cp src/build-scripts/oneAPI.repo /etc/yum.repos.d
-        sudo yum install -y intel-oneapi-compiler-dpcpp-cpp-and-cpp-classic
-        set +e; source /opt/intel/oneapi/setvars.sh; set -e
+        sudo yum install -y intel-oneapi-compiler-dpcpp-cpp-and-cpp-classic-2022.1.0-3768
+        # Because multiple (possibly newer) versions of oneAPI may be installed,
+        # use a config file to specify compiler and tbb versions
+        # NOTE: oneAPI components have independent version numbering.
+        set +e; source /opt/intel/oneapi/setvars.sh --config oneapi_2022.1.0.cfg; set -e
+
        if [[ "$CXX" == "icpc" || "$CC" == "icc" || "$USE_ICC" != "" ]] ; then
            echo "Verifying installation of Intel(r) C++ Compiler:"
            icpc --version
@@ -64,15 +68,27 @@ else
        git cmake ninja-build ccache g++ \
        libboost-dev libboost-thread-dev libboost-filesystem-dev \
        libilmbase-dev libopenexr-dev \
-        python-dev python-numpy \
        libtiff-dev libgif-dev libpng-dev \
        flex bison libbison-dev \
        libpugixml-dev \
-        libopencolorio-dev \
-        qt5-default
+        libopencolorio-dev
+
+    time sudo apt-get -q install -y \
+        qt5-default || /bin/true

    export CMAKE_PREFIX_PATH=/usr/lib/x86_64-linux-gnu:$CMAKE_PREFIX_PATH

+    # Nonstandard python versions
+    if [[ "${PYTHON_VERSION}" == "3.9" ]] ; then
+        time sudo apt-get -q install -y python3.9-dev python3-numpy
+        pip3 --version
+        pip3 install numpy
+    elif [[ "$PYTHON_VERSION" == "2.7" ]] ; then
+        time sudo apt-get -q install -y python-dev python-numpy
+    else
+        pip3 install numpy
+    fi
+
    if [[ "$CXX" == "g++-6" ]] ; then
        time sudo apt-get install -y g++-6
    elif [[ "$CXX" == "g++-7" ]] ; then
@@ -85,6 +101,8 @@ else
        time sudo apt-get install -y g++-10
    elif [[ "$CXX" == "g++-11" ]] ; then
        time sudo apt-get install -y g++-11
+    elif [[ "$CXX" == "g++-12" ]] ; then
+        time sudo apt-get install -y g++-12
    fi

    if [[ "$CXX" == "icpc" || "$CC" == "icc" || "$USE_ICC" != "" || "$CXX" == "icpx" || "$CC" == "icx" || "$USE_ICX" != "" ]] ; then
@@ -92,22 +110,11 @@ else
        sudo apt-key add GPG-PUB-KEY-INTEL-SW-PRODUCTS-2023.PUB
        echo "deb https://apt.repos.intel.com/oneapi all main" | sudo tee /etc/apt/sources.list.d/oneAPI.list
        time sudo apt-get update
-        time sudo apt-get install -y intel-oneapi-compiler-dpcpp-cpp-and-cpp-classic
-        set +e; source /opt/intel/oneapi/setvars.sh; set -e
-        if [[ "$CXX" == "icpc" || "$CC" == "icc" || "$USE_ICC" != "" ]] ; then
-            icpc --version
-        fi
-        if [[ "$CXX" == "icpx" || "$CC" == "icx" || "$USE_ICX" != "" ]] ; then
-            icpx --version
-        fi
-    fi
-
-    # Nonstandard python versions
-    if [[ "${PYTHON_VERSION}" == "3.8" ]] ; then
-        #time sudo apt-get -q install -y python3.8-dev python3-numpy
-        echo "skip?"
-    elif [[ "${PYTHON_VERSION}" == "3.9" ]] ; then
-        time sudo apt-get -q install -y python3.9-dev python3-numpy
+        time sudo apt-get install -y intel-oneapi-compiler-dpcpp-cpp-and-cpp-classic=2022.1.0-3768
+        # Because multiple (possibly newer) versions of oneAPI may be installed,
+        # use a config file to specify compiler and tbb versions
+        # NOTE: oneAPI components have independent version numbering.
+        set +e; source /opt/intel/oneapi/setvars.sh --config oneapi_2022.1.0.cfg; set -e
    fi

    source src/build-scripts/build_llvm.bash
@@ -164,7 +171,7 @@ if [[ "$OPENIMAGEIO_VERSION" != "" ]] ; then
    # Don't let warnings in OIIO break OSL's CI run
    export OPENIMAGEIO_CMAKE_FLAGS+=" -DSTOP_ON_WARNING=OFF"
    export OPENIMAGEIO_CMAKE_FLAGS+=" -DUSE_OPENGL=0"
-    if [[ $OPENIMAGEIO_VERSION == master ]] ; then
+    if [[ $OPENIMAGEIO_VERSION == master && "${OPENIMAGEIO_UNITY:-1}" != "0" ]] ; then
        # Speed up the OIIO build by doing a "unity" build. (Note: this is
        # only a savings in CI where there are only 1-2 cores available, and
        # is only supported for OIIO 2.3.14 and later)
@@ -4,7 +4,7 @@
 # SPDX-License-Identifier: BSD-3-Clause
 # https://github.com/AcademySoftwareFoundation/OpenShadingLanguage

-# This script, which assumes it is runnign on a Mac OSX with Homebrew
+# This script, which assumes it is running on a Mac OSX with Homebrew
 # installed, does a "brew install" in all packages reasonably needed by
 # OSL and its dependencies.

@@ -20,16 +20,20 @@ if [[ `which brew` == "" ]] ; then
 fi


-#brew update >/dev/null
+if [[ "$DO_BREW_UPDATE" != "" ]] ; then
+    brew update >/dev/null
+fi
 echo ""
 echo "Before my brew installs:"
 brew list --versions

 brew install --display-times -q gcc ccache cmake ninja boost || true
 brew link --overwrite gcc
+brew install --display-times -q python@${PYTHON_VERSION} || true
 brew unlink python@2.7 || true
-brew unlink python@3.9 || true
 brew unlink python@3.8 || true
+brew unlink python@3.9 || true
+brew unlink python@3.10 || true
 brew link --overwrite --force python@${PYTHON_VERSION} || true
 brew upgrade --display-times -q cmake || true
 brew install --display-times -q ilmbase openexr
@@ -1,5 +0,0 @@
-leak:Imf_2_2
-leak:Imf_2_3
-leak:Imf_2_4
-leak:libIlmImf
-leak:oslgram.y
@@ -0,0 +1,3 @@
+default=exclude
+compiler=2022.1.0
+tbb=2021.6.0
@@ -1,3 +1,4 @@
+#!/usr/bin/env python
 # Copyright Contributors to the Open Shading Language project.
 # SPDX-License-Identifier: BSD-3-Clause
 # https://github.com/AcademySoftwareFoundation/OpenShadingLanguage
@@ -22,7 +22,7 @@ find_dependency(OpenImageIO @OpenImageIO_VERSION@ REQUIRED)
 # Compute the installation prefix relative to this file
 get_filename_component(_IMPORT_PREFIX "${CMAKE_CURRENT_LIST_DIR}/../../../" ABSOLUTE)

-# There's no guarantee that CMAKE_INSTALL_XXXDIR is a relative path so we extract the final component to handle asbolute paths
+# There's no guarantee that CMAKE_INSTALL_XXXDIR is a relative path so we extract the final component to handle absolute paths
 get_filename_component(_INCLUDE_DIR_NAME "@CMAKE_INSTALL_INCLUDEDIR@" NAME)
 get_filename_component(_LIB_DIR_NAME "@CMAKE_INSTALL_LIBDIR@" NAME)

@@ -226,7 +226,7 @@ endif ()



-# We wil use this for ccache and timing
+# We will use this for ccache and timing
 set (MY_RULE_LAUNCH "")


@@ -376,11 +376,9 @@ include (CheckCXXSourceRuns)
 option (CODECOV "Build code coverage tests" OFF)
 if (CODECOV AND (CMAKE_COMPILER_IS_GNUCC OR CMAKE_COMPILER_IS_CLANG))
    message (STATUS "Compiling for code coverage analysis")
-    add_compile_options ("-ftest-coverage -fprofile-arcs -O0")
+    add_compile_options (-ftest-coverage -fprofile-arcs)
    add_definitions ("-D${PROJ_NAME}_CODE_COVERAGE=1")
-    set (CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -ftest-coverage -fprofile-arcs")
-    set (CMAKE_MODULE_LINKER_FLAGS "${CMAKE_MODULE_LINKER_FLAGS} -ftest-coverage -fprofile-arcs")
-    set (CMAKE_SHARED_LINKER_FLAGS "${CMAKE_SHARED_LINKER_FLAGS} -ftest-coverage -fprofile-arcs")
+    link_libraries(gcov)
 endif ()


@@ -394,22 +392,9 @@ if (SANITIZE AND (CMAKE_COMPILER_IS_GNUCC OR CMAKE_COMPILER_IS_CLANG))
    foreach (feature ${SANITIZE_FEATURE_LIST})
        message (STATUS "  sanitize feature: ${feature}")
        add_compile_options (-fsanitize=${feature})
-        set (CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -fsanitize=${feature}")
-        set (CMAKE_MODULE_LINKER_FLAGS "${CMAKE_MODULE_LINKER_FLAGS} -fsanitize=${feature}")
-        set (CMAKE_SHARED_LINKER_FLAGS "${CMAKE_SHARED_LINKER_FLAGS} -fsanitize=${feature}")
+        add_link_options (-fsanitize=${feature})
    endforeach()
    add_compile_options (-g -fno-omit-frame-pointer)
-    if (${CMAKE_SYSTEM_NAME} STREQUAL "Linux")
-        set (SANITIZE_ON_LINUX 1)
-    endif ()
-    if (CMAKE_COMPILER_IS_GNUCC AND ${CMAKE_SYSTEM_NAME} STREQUAL "Linux")
-        add_compile_options ("-fuse-ld=gold")
-        set (CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -fuse-ld=gold")
-        set (CMAKE_MODULE_LINKER_FLAGS "${CMAKE_MODULE_LINKER_FLAGS} -fuse-ld=gold")
-        set (CMAKE_SHARED_LINKER_FLAGS "${CMAKE_SHARED_LINKER_FLAGS} -fuse-ld=gold")
-        set (SANITIZE_LIBRARIES "asan;pthread")
-        # set (SANITIZE_LIBRARIES "asan" "ubsan")
-    endif()
    if (CMAKE_COMPILER_IS_GNUCC)
        # turn on glibcxx extra annotations to find vector writes past end
        add_definitions ("-D_GLIBCXX_SANITIZE_VECTOR=1")
@@ -475,29 +460,7 @@ endif ()
 # is being built as a subproject.
 if (PROJECT_IS_TOP_LEVEL)
    set (CLANG_FORMAT_EXE_HINT "" CACHE PATH "clang-format executable's directory (will search if not specified")
-    set (CLANG_FORMAT_INCLUDES  # "src/*.cpp"
-                                "src/*.h"
-                                "src/*.cu"
-                                "src/include/*.h"
-                                "src/liboslcomp/*.cpp"
-                                "src/liboslcomp/*.h"
-                                "src/liboslexec/batch*.cpp"
-                                "src/liboslexec/wide/*.cpp"
-                                "src/liboslexec/wide/*.h"
-                                "src/liboslexec/rs_fallback.cpp"
-                                "src/liboslquery/*.cpp"
-                                "src/liboslnoise/*.cpp"
-                                "src/liboslquery/*.h"
-                                "src/oslinfo/*.cpp"
-                                "src/oslc/*.cpp"
-                                "src/osl.imageio/*.cpp"
-                                "src/osltoy/*.cpp"
-                                "src/osltoy/*.h"
-                                "src/testrender/*.h"
-                                "src/testrender/*.cpp"
-                                "src/testshade/*.h"
-                                "src/testshade/*.cpp"
-                                "testsuite/*.cpp"
+    set (CLANG_FORMAT_INCLUDES  "src/*.cpp" "src/*.h" "src/*.cu"
        CACHE STRING "Glob patterns to include for clang-format")
    set (CLANG_FORMAT_EXCLUDES
            # Files "imported and modified" that we don't want to reformat, so
@@ -508,7 +471,6 @@ if (PROJECT_IS_TOP_LEVEL)
            "testsuite/*.h"
            # Header files in shaders are OSL, not C++
            "src/shaders/*.h"
-            ""
         CACHE STRING "Glob patterns to exclude for clang-format")
    find_program (CLANG_FORMAT_EXE
                  NAMES clang-format bin/clang-format
@@ -529,7 +491,7 @@ if (PROJECT_IS_TOP_LEVEL)
        # message (STATUS "clang-format file list: ${FILES_TO_FORMAT}")
        file (COPY ${CMAKE_CURRENT_SOURCE_DIR}/.clang-format
              DESTINATION ${CMAKE_CURRENT_BINARY_DIR})
-        add_custom_target (run-clang-format
+        add_custom_target (clang-format
            COMMAND "${CLANG_FORMAT_EXE}" -i -style=file ${FILES_TO_FORMAT} )
    else ()
        message (STATUS "clang-format not found.")
@@ -88,7 +88,7 @@ checked_find_package (OpenEXR REQUIRED
                     )
 # Force Imath includes to be before everything else to ensure that we have
 # the right Imath/OpenEXR version, not some older version in the system
-# library. This shoudn't be necessary, except for the common case of people
+# library. This shouldn't be necessary, except for the common case of people
 # building against Imath/OpenEXR 3.x when there is still a system-level
 # install version of 2.x.
 include_directories(BEFORE ${IMATH_INCLUDES})
@@ -105,7 +105,7 @@ endif ()

 # OpenImageIO
 checked_find_package (OpenImageIO REQUIRED
-                      VERSION_MIN 2.2.6
+                      VERSION_MIN 2.3.17
                      DEFINITIONS -DOIIO_HIDE_FORMAT=1)

 checked_find_package (pugixml REQUIRED
@@ -136,6 +136,14 @@ if (APPLE AND LLVM_VERSION VERSION_EQUAL 10.0.1 AND EXISTS "/usr/local/Cellar/ll
             "    brew upgrade llvm \n"
             "${ColorReset}\n")
 endif ()
+if (LLVM_VERSION VERSION_GREATER_EQUAL 15.0
+    AND (   (CMAKE_COMPILER_IS_APPLECLANG AND APPLECLANG_VERSION_STRING VERSION_LESS 15.0)
+         OR (CMAKE_COMPILER_IS_CLANG AND CLANG_VERSION_STRING VERSION_LESS 15.0)))
+    message (WARNING
+         "${ColorYellow}"
+         "If you are using LLVM 15 or higher, you should also use clang version "
+         "15 or higher, or you may get build errors.${ColorReset}\n")
+endif ()

 checked_find_package (partio)

@@ -198,7 +206,7 @@ if (USE_CUDA OR USE_OPTIX)
    # OptiX setup
    if (USE_OPTIX)
        checked_find_package (OptiX REQUIRED
-                              VERSION_MIN 5.1)
+                              VERSION_MIN 7.0)
        include_directories (BEFORE "${OPTIX_INCLUDES}")
        if (NOT USE_LLVM_BITCODE OR NOT USE_FAST_MATH)
            message (FATAL_ERROR "Enabling OptiX requires USE_LLVM_BITCODE=1 and USE_FAST_MATH=1")
@@ -82,6 +82,11 @@ function ( EMBED_LLVM_BITCODE_IN_CPP src_list suffix output_name list_to_append_
        list (TRANSFORM OpenImageIO_INCLUDES PREPEND -I
            OUTPUT_VARIABLE ALL_OpenImageIO_INCLUDES)

+        if (${LLVM_VERSION} VERSION_GREATER_EQUAL 15.0)
+            # Until we fully support opaque pointers, we need to disable
+            # them when using LLVM 15.
+            list (APPEND LLVM_COMPILE_FLAGS -Xclang -no-opaque-pointers)
+        endif ()

        # Command to turn the .cpp file into LLVM assembly language .s, into
        # LLVM bitcode .bc, then back into a C++ file with the bc embedded!
@@ -103,8 +108,8 @@ function ( EMBED_LLVM_BITCODE_IN_CPP src_list suffix output_name list_to_append_
            -O3 -fno-math-errno -S -emit-llvm ${extra_clang_args}
            -o ${src_asm} ${src}
        COMMAND ${LLVM_AS_TOOL} -f -o ${src_bc} ${src_asm}
-        # Do NOT setup a MAIN_DEPENDENCY because only 1 may exist 
-        # and we may have the several outputs dependant on the same source 
+        # Do NOT setup a MAIN_DEPENDENCY because only 1 may exist
+        # and we may have the several outputs dependent on the same source
        DEPENDS ${src} ${exec_headers} ${PROJECT_PUBLIC_HEADERS}
        WORKING_DIRECTORY "${CMAKE_CURRENT_SOURCE_DIR}" )
    endforeach ()
@@ -132,7 +137,7 @@ function ( EMBED_LLVM_BITCODE_IN_CPP src_list suffix output_name list_to_append_
        ${exec_headers} ${PROJECT_PUBLIC_HEADERS}
        WORKING_DIRECTORY "${CMAKE_CURRENT_SOURCE_DIR}" )

-    # add generated .cpp with embedded bitcode to the list of soures
+    # add generated .cpp with embedded bitcode to the list of sources
    set ( ${list_to_append_cpp} ${${list_to_append_cpp}} ${src_bc_cpp} PARENT_SCOPE )

 endfunction ( )
@@ -10,7 +10,7 @@
 # PARTIO_LIBRARIES - library to link against
 # PARTIO_FOUND - true if Partio was found.

-# Hack! perfer a config if it can be found
+# Hack! prefer a config if it can be found
 find_package(partio CONFIG)
 if (PARTIO_FOUND OR partio_FOUND)
    if (VERBOSE)
@@ -1,4 +1,4 @@
-# Copyright Contributors to the Open Shading Languge project.
+# Copyright Contributors to the Open Shading Language project.
 # SPDX-License-Identifier: BSD-3-Clause
 # https://github.com/AcademySoftwareFoundation/OpenShadingLanguage/

@@ -77,7 +77,7 @@ macro (setup_python_module)
                           PRIVATE
                                pybind11::pybind11
                                Python::Python
-                                ${lib_LIBS} ${SANITIZE_LIBRARIES})
+                                ${lib_LIBS})
    if (APPLE)
        set_target_properties (${target_name} PROPERTIES LINK_FLAGS "-undefined dynamic_lookup")
    endif ()
@@ -1,4 +1,4 @@
-# Copyright Contributors to the Open Shading Languge project.
+# Copyright Contributors to the Open Shading Language project.
 # SPDX-License-Identifier: BSD-3-Clause
 # https://github.com/AcademySoftwareFoundation/OpenShadingLanguage/

@@ -16,6 +16,10 @@ add_custom_command (OUTPUT "${CMAKE_BINARY_DIR}/testsuite/runtest.py"
                    MAIN_DEPENDENCY "${CMAKE_SOURCE_DIR}/testsuite/runtest.py")
 add_custom_target ( CopyFiles ALL DEPENDS "${CMAKE_BINARY_DIR}/testsuite/runtest.py" )

+
+set (OSL_TEST_BIG_TIMEOUT 800 CACHE STRING "Timeout for tests that take a long time")
+
+
 # add_one_testsuite() - set up one testsuite entry
 #
 # Usage:
@@ -26,10 +30,12 @@ add_custom_target ( CopyFiles ALL DEPENDS "${CMAKE_BINARY_DIR}/testsuite/runtest
 #                  [COMMAND cmd...] - optional override of launch command
 #                 )
 #
+
 macro (add_one_testsuite testname testsrcdir)
    cmake_parse_arguments (_tst "" "" "ENV;COMMAND" ${ARGN})
    set (testsuite "${CMAKE_SOURCE_DIR}/testsuite")
    set (testdir "${CMAKE_BINARY_DIR}/testsuite/${testname}")
+    file (MAKE_DIRECTORY "${testdir}")
    if (NOT _tst_COMMAND)
        set (_tst_COMMAND ${Python_EXECUTABLE} "${testsuite}/runtest.py" ${testdir})
        if (MSVC_IDE)
@@ -45,7 +51,17 @@ macro (add_one_testsuite testname testsrcdir)
              OSL_TESTSUITE_SRC=${testsrcdir}
              OSL_TESTSUITE_CUR=${testdir}
         )
-    file (MAKE_DIRECTORY "${testdir}")
+
+    # We need to suppress the leak sanitizer for "texture" tests, because of a
+    # minor leak inside OIIO (140 bytes! -- inconsequential) but which we
+    # can't use the suppressions list becuase the leaks come from calls made
+    # inside JITed code for which there is not a good stack trace. Any real
+    # leaks from OIIO or its dependencies we assume will be caught by lsan on
+    # the OIIO side, where there is no JIT to complicate things.
+    if (${testname} MATCHES "^texture" OR ${testname} MATCHES "gettextureinfo")
+        list (APPEND _tst_ENV ASAN_OPTIONS=detect_leaks=0:$ENV{ASAN_OPTIONS})
+    endif ()
+
    add_test ( NAME ${testname} COMMAND ${_tst_COMMAND} )
    # message ("Test -- env ${_tst_ENV} cmd ${_tst_COMMAND}")
    set_tests_properties (${testname} PROPERTIES ENVIRONMENT "${_tst_ENV}" )
@@ -53,8 +69,7 @@ macro (add_one_testsuite testname testsrcdir)
    # from piling up together, allocate a few cores each.
    if (${testname} MATCHES "^render-")
        set_tests_properties (${testname} PROPERTIES LABELS render
-                              TIMEOUT 240
-                              PROCESSORS 4 COST 10 TIMEOUT 240)
+                              PROCESSORS 4 COST 10 TIMEOUT ${OSL_TEST_BIG_TIMEOUT})
    endif ()
    # Some labeling for fun
    if (${testname} MATCHES "^texture")
@@ -76,7 +91,7 @@ macro (add_one_testsuite testname testsrcdir)
        # These are batched shading regression tests. Some are quite
        # long, so give them a higher cost and timeout.
        set_tests_properties (${testname} PROPERTIES LABELS batchregression
-                              COST 15 TIMEOUT 300)
+                              COST 15 TIMEOUT ${OSL_TEST_BIG_TIMEOUT})
    endif ()
 endmacro ()

@@ -107,7 +122,7 @@ macro ( TESTSUITE )

        # Run the test unoptimized, unless it matches a few patterns that
        # we don't test unoptimized (or has an OPTIMIZEONLY marker file).
-        if (NOT _testname MATCHES "optix" 
+        if (NOT _testname MATCHES "optix"
            AND NOT EXISTS "${_testsrcdir}/NOSCALAR"
            AND NOT EXISTS "${_testsrcdir}/BATCHED_REGRESSION"
            AND NOT EXISTS "${_testsrcdir}/OPTIMIZEONLY")
@@ -124,6 +139,9 @@ macro ( TESTSUITE )
            add_one_testsuite ("${_testname}.opt" "${_testsrcdir}"
                               ENV TESTSHADE_OPT=2 )
        endif ()
+        if (_testname MATCHES "render-" AND ${CMAKE_BUILD_TYPE} STREQUAL Debug)
+            set_tests_properties (${_testname} PROPERTIES TIMEOUT ${OSL_TEST_BIG_TIMEOUT})
+        endif ()
        # When building for OptiX support, also run it in OptiX mode
        # if there is an OPTIX marker file in the directory.
        # If an environment variable $TESTSUITE_OPTIX is nonzero, then
@@ -142,7 +160,7 @@ macro ( TESTSUITE )
            add_one_testsuite ("${_testname}.optix.opt" "${_testsrcdir}"
                               ENV TESTSHADE_OPT=2 TESTSHADE_OPTIX=1 )
        endif ()
-        
+
        if (BUILD_BATCHED)
            # When building for Batched support, also run it in Batched mode
            # if there is an BATCHED marker file in the directory.
@@ -156,7 +174,7 @@ macro ( TESTSUITE )
                add_one_testsuite ("${_testname}.batched" "${_testsrcdir}"
                                   ENV TESTSHADE_OPT=0 TESTSHADE_BATCHED=1 )
            endif ()
-            
+
            if ((EXISTS "${_testsrcdir}/BATCHED" OR test_all_batched OR _testname MATCHES "batched")
                AND NOT EXISTS "${_testsrcdir}/BATCHED_REGRESSION"
                AND NOT EXISTS "${_testsrcdir}/NOBATCHED"
@@ -165,7 +183,7 @@ macro ( TESTSUITE )
                add_one_testsuite ("${_testname}.batched.opt" "${_testsrcdir}"
                                   ENV TESTSHADE_OPT=2 TESTSHADE_BATCHED=1 )
            endif ()
-            
+
            # When building for Batched support, optionally run a regression test
            # if there is an BATCHED_REGRESSION marker file in the directory.
            if ((EXISTS "${_testsrcdir}/BATCHED_REGRESSION")
@@ -211,7 +229,7 @@ macro (osl_add_all_tests)
    # special installed tests.
    TESTSUITE ( aastep allowconnect-err andor-reg and-or-not-synonyms
                arithmetic area-reg arithmetic-reg
-                array array-reg array-copy-reg array-derivs array-range 
+                array array-reg array-copy-reg array-derivs array-range
                array-aassign array-assign-reg array-length-reg
                bitwise-and-reg bitwise-or-reg bitwise-shl-reg  bitwise-shr-reg bitwise-xor-reg
                blackbody blackbody-reg blendmath breakcont breakcont-reg
@@ -220,7 +238,7 @@ macro (osl_add_all_tests)
                calculatenormal-reg
                cellnoise closure closure-array color color-reg colorspace comparison
                complement-reg compile-buffer compassign-reg
-                component-range 
+                component-range
                control-flow-reg connect-components
                const-array-params const-array-fill
                debugnan debug-uninit
@@ -239,7 +257,9 @@ macro (osl_add_all_tests)
                globals-needed
                group-outputs groupstring
                hash hashnoise hex hyperb
-                ieee_fp ieee_fp-reg if if-reg incdec initlist initops intbits isconnected
+                ieee_fp ieee_fp-reg if if-reg incdec initlist
+                initops initops-instance-clash
+                intbits isconnected
                isconstant
                layers layers-Ciassign layers-entry layers-lazy layers-lazyerror
                layers-nonlazycopy layers-repeatedoutputs
@@ -259,7 +279,7 @@ macro (osl_add_all_tests)
                noise-perlin noise-simplex
                noise-reg
                normalize-reg
-                pnoise pnoise-cell pnoise-gabor 
+                pnoise pnoise-cell pnoise-gabor
                pnoise-generic pnoise-perlin
                pnoise-reg
                operator-overloading
@@ -291,11 +311,20 @@ macro (osl_add_all_tests)
                raytype raytype-reg raytype-specialized regex-reg reparam
                render-background render-bumptest
                render-cornell render-furnace-diffuse
+                render-mx-furnace-burley-diffuse
+                render-mx-furnace-oren-nayar
+                render-mx-furnace-sheen
+                render-mx-burley-diffuse
+                render-mx-dielectric render-mx-dielectric-glass
+                render-mx-conductor
+                render-mx-generalized-schlick render-mx-generalized-schlick-glass
+                render-mx-layer
+                render-mx-sheen
                render-microfacet render-oren-nayar
                render-uv render-veachmis render-ward
                select select-reg shaderglobals shortcircuit
-                smoothstep-reg 
-                spline spline-reg splineinverse splineinverse-ident 
+                smoothstep-reg
+                spline spline-reg splineinverse splineinverse-ident
                splineinverse-knots-ascend-reg splineinverse-knots-descend-reg
                spline-boundarybug spline-derivbug
                split-reg
@@ -319,7 +348,7 @@ macro (osl_add_all_tests)
                trailing-commas
                transcendental-reg
                transitive-assign
-                transform transform-reg transformc transformc-reg trig trig-reg 
+                transform transform-reg transformc transformc-reg trig trig-reg
                typecast
                unknown-instruction
                userdata userdata-passthrough
@@ -339,10 +368,10 @@ macro (osl_add_all_tests)
    endif ()

    # Add tests that require the Python bindings if we built them.
-    # We also exclude these tests if this is a sanitizer build on Linux,
-    # because the Python interpreter itself won't be linked with the right asan
+    # We also exclude these tests if this is a sanitizer build, because the
+    # Python interpreter itself won't be linked with the right asan
    # libraries to run correctly.
-    if (USE_PYTHON AND NOT SANITIZE_ON_LINUX)
+    if (USE_PYTHON AND NOT SANITIZE)
        TESTSUITE ( python-oslquery )
    endif ()

@@ -363,11 +392,13 @@ macro (osl_add_all_tests)
        TESTSUITE ( testoptix testoptix-noise example-cuda)
    endif ()

-    # Some regression tests have alot of combinations and may need more time to finish 
+    # Some regression tests have a lot of combinations and may need more time to finish
    if (BUILD_BATCHED)
-        set_tests_properties (arithmetic-reg.regress.batched.opt PROPERTIES TIMEOUT 800)
-        set_tests_properties (transform-reg.regress.batched.opt PROPERTIES TIMEOUT 800)
+        set_tests_properties (arithmetic-reg.regress.batched.opt
+                              PROPERTIES TIMEOUT ${OSL_TEST_BIG_TIMEOUT})
+        set_tests_properties (transform-reg.regress.batched.opt
+                              PROPERTIES TIMEOUT ${OSL_TEST_BIG_TIMEOUT})
    endif ()
-    set_tests_properties (matrix-reg.regress.rsbitcode.opt PROPERTIES TIMEOUT 800)
-        
+    set_tests_properties (matrix-reg.regress.rsbitcode.opt
+                          PROPERTIES TIMEOUT ${OSL_TEST_BIG_TIMEOUT})
 endmacro()
@@ -66,7 +66,7 @@ All rights reserved.
 Editor: Larry Gritz \\
 \emph{lg@imageworks.com}
 }
-\date{{\large Date: 24 Nov 2021 \\
+\date{{\large Date: 27 Aug 2022 \\
 % (with corrections, 24 Nov 2021)
 }
 \bigskip
@@ -1431,7 +1431,7 @@ least 32 bits.

 Integer constants are constructed the same way as in C.  The following
 are examples of {\cf int} constants: {\cf 1}, {\cf -32}, etc. Integer
-constants may be specified as hexidecimal, for example: {\cf 0x01cf}.
+constants may be specified as hexadecimal, for example: {\cf 0x01cf}.

 Unlike C, no unsigned, bool, char, short, or long types are supplied.
 This is to simplify the process of writing shaders (as well as
@@ -3778,7 +3778,7 @@ Computes the \emph{inverse} of the {\cf spline()} function, i.e., returns
 the value $x$ for which \\
   \spc {\cf spline (basis, x, y...)}\\ 
 would return value
-$v$.  Results are undefined if the knots do not specifiy a monotonic
+$v$.  Results are undefined if the knots do not specify a monotonic
 (only increasing or only decreasing) set of values.

 Note that the combination of {\cf spline} and {\cf splineinverse} makes
@@ -3905,7 +3905,7 @@ multi-component types such as \color, the format will be applied to each
 of the components).  The {\cf \%s} expects a {\cf string} or 
 {\cf closure} argument.

-All of the substition commands follow the usual C/C++ formatting rules,
+All of the substitution commands follow the usual C/C++ formatting rules,
 so format commands such as {\cf "\%6.2f"}, etc., should work as
 expected.
 \apiend
@@ -3930,7 +3930,7 @@ and other diagnostic information.
 \indexapi{fprintf()}
 The {\cf fprintf()} function works similarly to {\cf printf}, but rather
 than printing to the default text output stream, the results will be
-concatentated onto the end of the text file named by {\cf filename}.
+concatenated onto the end of the text file named by {\cf filename}.
 \apiend

 \apiitem{string {\ce concat} (string s1, ..., string sN)}
@@ -4404,7 +4404,7 @@ will not be modified and {\cf gettextureinfo} will return {\cf 0}.
 The version of {\cf gettextureinfo} that takes {\cf s} and {\cf t} parameters
 retrieves information about the texture file that will be used for those
 texture coordinates. This can be useful for UDIM textures that may use
-different texture files for different regions, based on the coodinates. For
+different texture files for different regions, based on the corodinates. For
 regular, non-UDIM textures, the coordinates, if supplied, will be ignored.
 When UDIM textures are queried without coordinates supplied, it will succeed
 and return the texture info only if that parameter is found and has the same
@@ -4611,7 +4611,7 @@ Example:
 \index{material closure functions}
 \index{functions!closures}

-For \closurecolor functions, the return ``value'' is symbolic and is may
+For \closurecolor functions, the return ``value'' is symbolic and may
 be passed to an output variable or assigned to \Ci, to be evaluated at a
 later time convenient to the renderer in order to compute the exitant
 radiance in the direction {\cf -I}.  But the shader itself cannot
@@ -4620,15 +4620,695 @@ examine the numeric values of the \colorclosure.
 The intent of this specification is to give a minimal but useful set of
 material closures that you can expect any renderer implementation to
 provide.  Individual renderers may supply additional closures that are
-specific to the workings of that renderer.  Additionally, invidual
+specific to the workings of that renderer.  Additionally, individual
 renderers may allow additional parameters or controls on the standard
 closures, passed as token/value pairs following the required arguments
 (much like the optional arguments to the {\cf texture()} function).
 Consult the documentation for your specific renderer for details.

+OSL's standard material closures are by synchronized to match the names and
+properties of the physically-based shading nodes of MaterialX v1.38
+(\url{https://www.materialx.org/}).
+
+\subsection{Surface BSDF closures}
+
+\apiitem{\closurecolor\ {\ce oren_nayar_diffuse_bsdf}(normal N, color albedo, float roughness)}
+\indexapi{oren_nayar_diffuse_bsdf()}
+
+Constructs a diffuse reflection BSDF based on the Oren-Nayar reflectance
+model.
+
+\noindent Parameters include:
+
+  \apiitem{N}
+  \vspace{12pt}
+  Normal vector of the surface point being shaded.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{albedo}
+  \vspace{12pt}
+  Surface albedo.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{roughness}
+  \vspace{12pt}
+  Surface roughness [0,1]. A value of 0.0 gives Lambertian reflectance.
+  \apiend
+%   \vspace{-16pt}
+  
+  % \apiitem{label}
+  % \vspace{12pt}
+  % Optional string parameter to name this component. For use in AOVs / LPEs.
+  % \apiend
+
+The Oren-Nayar reflection model is described in  M.\ Oren and S.\ K.\
+Nayar, ``Generalization of Lambert's Reflectance Model,'' Proceedings of
+SIGGRAPH 1994, pp.239-246 (July, 1994).
+
+\apiend


-\subsection{Surface material closures}
+\apiitem{\colorclosure\ {\ce burley_diffuse_bsdf} (normal N, color albedo, float roughness)}
+\indexapi{burley_diffuse_bsdf()}
+
+Constructs a diffuse reflection BSDF based on the corresponding component of 
+the Disney Principled shading model.
+
+\noindent Parameters include:
+
+  \apiitem{N}
+  \vspace{12pt}
+  Normal vector of the surface point being shaded.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{albedo}
+  \vspace{12pt}
+  Surface albedo.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{roughness}
+  \vspace{12pt}
+  Surface roughness [0,1]. A value of 0.0 gives Lambertian reflectance.
+  \apiend
+%   \vspace{-16pt}
+  
+  % \apiitem{label}
+  % \vspace{12pt}
+  % Optional string parameter to name this component. For use in AOVs / LPEs.
+  % \apiend
+
+\apiend
+
+
+\apiitem{\colorclosure\ {\ce dielectric_bsdf} (normal N, vector U, color reflection_tint, \\
+\spc\spc color transmission_tint, float roughness_x, float roughness_y, \\
+\spc\spc float ior, string distribution)}
+\indexapi{dielectric_bsdf()}
+
+Constructs a reflection and/or transmission BSDF based on a microfacet
+reflectance model and a Fresnel curve for dielectrics. The two tint parameters
+control the contribution of each reflection/transmission lobe. The tints
+should remain 100\% white for a physically correct dielectric, but can be
+tweaked for artistic control or set to 0.0 for disabling a lobe.
+
+The closure may be vertically layered over a base BSDF for the surface beneath
+the dielectric layer. This is done using the {\cf layer()} closure. By
+chaining multiple {\cf dielectric_bsdf} closures you can describe a surface
+with multiple specular lobes. If transmission is enabled (transmission_tint
+$>$ 0.0) the closure may be layered over a VDF closure describing the surface
+interior to handle absorption and scattering inside the medium.
+
+\noindent Parameters include:
+
+  \apiitem{N}
+  \vspace{12pt}
+  Normal vector of the surface point being shaded.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{U}
+  \vspace{12pt}
+  Tangent vector of the surface point being shaded.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{reflection_tint}
+  \vspace{12pt}
+  Weight per color channel for the reflection lobe. Should be (1,1,1) for a physically-correct dielectric surface, but can be tweaked for artistic control. Set to (0,0,0) to disable reflection.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{transmission_tint}
+  \vspace{12pt}
+  Weight per color channel for the transmission lobe. Should be (1,1,1) for a physically-correct dielectric surface, but can be tweaked for artistic control. Set to (0,0,0) to disable transmission.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{roughness_x}
+  \vspace{12pt}
+  Surface roughness in the U direction with a perceptually linear response over its range.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{roughness_y}
+  \vspace{12pt}
+  Surface roughness in the V direction with a perceptually linear response
+  over its range.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{ior}
+  \vspace{12pt}
+  Refraction index.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{distribution}
+  \vspace{12pt}
+  Microfacet distribution. An implementation is expected to support the
+  following distributions: \qkw{ggx}
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{thinfilm_thickness}
+  \vspace{12pt}
+  Optional float parameter for thickness of an iridescent thin film layer on
+  top of this BSDF. Given in nanometers.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{thinfilm_ior}
+  \vspace{12pt}
+  Optional float parameter for refraction index of the thin film layer.
+  \apiend
+%   \vspace{-16pt}
+  
+  % \apiitem{label}
+  % \vspace{12pt}
+  % Optional string parameter to name this component. For use in AOVs / LPEs.
+  % \apiend
+
+\apiend
+
+
+\apiitem{\colorclosure\ {\ce conductor_bsdf} (normal N, vector U, float roughness_x, \\
+\spc\spc float roughness_y, color ior, color extinction, string distribution)}
+\indexapi{conductor_bsdf()}
+
+Constructs a reflection BSDF based on a microfacet reflectance model. Uses a
+Fresnel curve with complex refraction index for conductors/metals. If an
+artistic parametrization is preferred the {\cf artistic_ior()} utility
+function can be used to convert from artistic to physical parameters.
+
+\noindent Parameters include:
+
+  \apiitem{N}
+  \vspace{12pt}
+  Normal vector of the surface point being shaded.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{U}
+  \vspace{12pt}
+  Tangent vector of the surface point being shaded.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{roughness_x}
+  \vspace{12pt}
+  Surface roughness in the U direction with a perceptually linear response over its range.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{roughness_y}
+  \vspace{12pt}
+  Surface roughness in the V direction with a perceptually linear response over its range.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{ior}
+  \vspace{12pt}
+  Refraction index.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{extinction}
+  \vspace{12pt}
+  Extinction coefficient.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{distribution}
+  \vspace{12pt}
+  Microfacet distribution. An implementation is expected to support the
+  following distributions: \qkw{ggx}
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{thinfilm_thickness}
+  \vspace{12pt}
+  Optional float parameter for thickness of an iridescent thin film layer on
+  top of this BSDF. Given in nanometers.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{thinfilm_ior}
+  \vspace{12pt}
+  Optional float parameter for refraction index of the thin film layer.
+  \apiend
+%   \vspace{-16pt}
+  
+  % \apiitem{label}
+  % \vspace{12pt}
+  % Optional string parameter to name this component. For use in AOVs / LPEs.
+  % \apiend
+
+\apiend
+
+
+\apiitem{\colorclosure\ {\ce generalized_schlick_bsdf} (normal N, vector U, \\
+\spc\spc  color reflection_tint, color transmission_tint, \\
+\spc\spc float roughness_x, float roughness_y, color f0, color f90, \\
+\spc\spc float exponent, string distribution)}
+\indexapi{generalized_schlick_bsdf()}
+
+Constructs a reflection and/or transmission BSDF based on a microfacet
+reflectance model and a generalized Schlick Fresnel curve. The two tint
+parameters control the contribution of each reflection/transmission lobe.
+
+The closure may be vertically layered over a base BSDF for the surface beneath
+the dielectric layer. This is done using the layer() closure. By chaining
+multiple {\cf dielectric_bsdf} closures you can describe a surface with multiple
+specular lobes. If transmission is enabled (transmission_tint $>$ 0.0) the
+closure may be layered over a VDF closure describing the surface interior to
+handle absorption and scattering inside the medium.
+
+\noindent Parameters include:
+
+  \apiitem{N}
+  \vspace{12pt}
+  Normal vector of the surface point being shaded.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{U}
+  \vspace{12pt}
+  Tangent vector of the surface point being shaded.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{reflection_tint}
+  \vspace{12pt}
+  Weight per color channel for the reflection lobe. Set to (0,0,0) to disable
+  reflection.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{transmission_tint}
+  \vspace{12pt}
+  Weight per color channel for the transmission lobe. Set to (0,0,0) to
+  disable transmission.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{roughness_x}
+  \vspace{12pt}
+  Surface roughness in the U direction with a perceptually linear response
+  over its range.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{roughness_y}
+  \vspace{12pt}
+  Surface roughness in the V direction with a perceptually linear response
+  over its range.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{f0}
+  \vspace{12pt}
+  Reflectivity per color channel at facing angles.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{f90}
+  \vspace{12pt}
+  Reflectivity per color channel at grazing angles.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{exponent}
+  \vspace{12pt}
+  Variable exponent for the Schlick Fresnel curve, the default value should be
+  5.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{distribution}
+  \vspace{12pt}
+  Microfacet distribution. An implementation is expected to support the
+  following distributions: \qkw{ggx}
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{thinfilm_thickness}
+  \vspace{12pt}
+  Optional float parameter for thickness of an iridescent thin film layer on
+  top of this BSDF. Given in nanometers.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{thinfilm_ior}
+  \vspace{12pt}
+  Optional float parameter for refraction index of the thin film layer.
+  \apiend
+%   \vspace{-16pt}
+  
+  % \apiitem{label}
+  % \vspace{12pt}
+  % Optional string parameter to name this component. For use in AOVs / LPEs.
+  % \apiend
+
+\apiend
+
+
+
+\apiitem{\colorclosure\ {\ce translucent_bsdf} (normal N, color albedo)}
+\indexapi{translucent_bsdf()}
+
+Constructs a translucent (diffuse transmission) BSDF based on the Lambert
+reflectance model.
+
+\noindent Parameters include:
+
+  \apiitem{N}
+  \vspace{12pt}
+  Normal vector of the surface point being shaded.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{albedo}
+  \vspace{12pt}
+  Surface albedo.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{roughness}
+  \vspace{12pt}
+  Surface roughness [0,1]. A value of 0.0 gives Lambertian reflectance.
+  \apiend
+%   \vspace{-16pt}
+  
+  % \apiitem{label}
+  % \vspace{12pt}
+  % Optional string parameter to name this component. For use in AOVs / LPEs.
+  % \apiend
+
+\apiend
+
+
+\apiitem{\colorclosure\ {\ce transparent_bsdf} ( )}
+\indexapi{transparent_bsdf()}
+
+Constructs a closure that represents straight transmission through a surface.
+
+\apiend
+
+
+
+\apiitem{\colorclosure\ {\ce subsurface_bssrdf} ( )}
+\indexapi{subsurface_bssrdf}
+Constructs a BSSRDF for subsurface scattering within a homogeneous medium.
+
+\noindent Parameters include:
+
+  \apiitem{N}
+  \vspace{12pt}
+  Normal vector of the surface point being shaded.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{albedo}
+  \vspace{12pt}
+  Single-scattering albedo of the medium.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{transmission_depth}
+  \vspace{12pt}
+  Distance travelled inside the medium by white light before its color becomes
+  transmission_color by Beer's law. Given in scene length units, range
+  [0,infinity). Together with transmission_color this determines the
+  extinction coefficient of the medium.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{transmission_color}
+  \vspace{12pt}
+  Desired color resulting from white light transmitted a distance of
+  'transmission_depth' through the medium. Together with transmission_depth
+  this determines the extinction coefficient of the medium.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{anisotropy}
+  \vspace{12pt}
+  Scattering anisotropy [-1,1]. Negative values give backwards scattering,
+  positive values give forward scattering, and 0.0 gives uniform scattering.
+  \apiend
+  
+%   \vspace{-16pt}
+  
+  % \apiitem{label}
+  % \vspace{12pt}
+  % Optional string parameter to name this component. For use in AOVs / LPEs.
+  % \apiend
+\apiend
+
+
+\apiitem{\colorclosure\ {\ce sheen_bsdf} (normal N, color albedo, float roughness)}
+\indexapi{sheen_bsdf()}
+
+Constructs a microfacet BSDF for the back-scattering properties of cloth-like
+materials. This closure may be vertically layered over a base BSDF, where
+energy that is not reflected will be transmitted to the base closure.
+
+\noindent Parameters include:
+
+  \apiitem{N}
+  \vspace{12pt}
+  Normal vector of the surface point being shaded.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{albedo}
+  \vspace{12pt}
+  Surface albedo.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{roughness}
+  \vspace{12pt}
+  Surface roughness [0,1].
+  \apiend
+%   \vspace{-16pt}
+  
+  % \apiitem{label}
+  % \vspace{12pt}
+  % Optional string parameter to name this component. For use in AOVs / LPEs.
+  % \apiend
+
+\apiend
+
+
+
+\subsection{Volumetric material closures}
+
+\apiitem{\colorclosure\ {\ce anisotropic_vdf} (color albedo, color extinction, float anisotropy)}
+\indexapi{anisotropic_vdf}
+
+Constructs a VDF scattering light for a general participating medium, based on
+the Henyey-Greenstein phase function. Forward, backward and uniform scattering
+is supported and controlled by the anisotropy input.
+
+\noindent Parameters include:
+
+  \apiitem{albedo}
+  \vspace{12pt}
+  Single-scattering albedo of the medium.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{extinction}
+  \vspace{12pt}
+  Volume extinction coefficient.
+  \apiend
+  
+  \apiitem{anisotropy}
+  \vspace{12pt}
+  Scattering anisotropy [-1,1]. Negative values give backwards scattering,
+  positive values give forward scattering, and 0.0 gives uniform scattering.
+  \apiend
+  
+%   \vspace{-16pt}
+  
+  % \apiitem{label}
+  % \vspace{12pt}
+  % Optional string parameter to name this component. For use in AOVs / LPEs.
+  % \apiend
+\apiend
+
+
+\apiitem{\colorclosure\ {\ce medium_vdf} (color albedo, float transmission_depth, \\
+\spc\spc color transmission_color, float anisotropy, float ior, int priority)}
+\indexapi{medium_vdf}
+
+Constructs a VDF for light passing through a dielectric homogeneous medium,
+such as glass or liquids. The parameters {\cf transmission_depth} and
+{\cf transmission_color} control the extinction coefficient of the medium in an
+artist-friendly way. A priority can be set to determine the ordering of
+overlapping media.
+
+\noindent Parameters include:
+
+  \apiitem{albedo}
+  \vspace{12pt}
+  Single-scattering albedo of the medium.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{transmission_depth}
+  \vspace{12pt}
+  Distance travelled inside the medium by white light before its color becomes
+  transmission_color by Beer's law. Given in scene length units, range
+  [0,infinity). Together with transmission_color this determines the
+  extinction coefficient of the medium.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{transmission_color}
+  \vspace{12pt}
+  Desired color resulting from white light transmitted a distance of
+  'transmission_depth' through the medium. Together with transmission_depth
+  this determines the extinction coefficient of the medium.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{anisotropy}
+  \vspace{12pt}
+  Scattering anisotropy [-1,1]. Negative values give backwards scattering,
+  positive values give forward scattering, and 0.0 gives uniform scattering.
+  \apiend
+  
+  \apiitem{ior}
+  \vspace{12pt}
+  Refraction index of the medium.
+  \apiend
+  \vspace{-16pt}
+  
+  \apiitem{priority}
+  \vspace{12pt}
+  Priority of this medium (for nested dielectrics).
+  \apiend
+  \vspace{-16pt}
+  
+%   \vspace{-16pt}
+  
+  % \apiitem{label}
+  % \vspace{12pt}
+  % Optional string parameter to name this component. For use in AOVs / LPEs.
+  % \apiend
+\apiend
+
+
+\subsection{Light emission closures}
+
+\apiitem{\colorclosure\ {\ce uniform_edf} (color emittance)}
+\indexapi{uniform_edf}
+Constructs an EDF emitting light uniformly in all directions.
+This is used to represent a glowing/emissive material.
+When called in the context of a surface shader group, it implies that
+light is emitted in a full hemisphere centered around the surface
+normal. When called in the context of a volume shader group, it implies
+that light is emitted evenly in all directions around the point being
+shaded.
+
+The {\cf emittance} parameter is the amount of emission and
+has units of radiance (e.g.,
+$\mathrm{W}\cdot\mathrm{sr}^{-1}\cdot\mathrm{m}^{-2}$). This means that a surface
+directly seen by the camera will directly reproduce the closure weight in the
+final pixel, regardless of being a surface or a volume.
+
+For an emissive surface, if you divide the return value of {\cf uniform_edf()} by
+{\cf surfacearea() * M_PI}, then you can easily specify the total emissive
+power of the light (e.g., $\mathrm{W}$), regardless of its physical size.
+\apiend
+
+
+\subsection{Layering and Signaling closures}
+
+\apiitem{\colorclosure\ {\ce layer} (closure color top, closure color base)}
+\indexapi{layer()}
+Vertically layer a layerable BSDF such as {\cf dielectric_bsdf},
+{\cf generalized_schlick_bsdf} or {\cf sheen_bsdf} over a BSDF or VDF. The
+implementation is target specific, but a standard way of handling this is by
+albedo scaling, using {\cf base*(1-reflectance(top)) + top}, where
+reflectance() calculates the directional albedo of a given top BSDF.
+\apiend
+
+
+\apiitem{\colorclosure\ {\ce holdout} ( )}
+\indexapi{holdout()}
+Returns a \colorclosure that does not represent any additional light
+reflection from the surface, but does signal to the renderer that 
+the surface is a \emph{holdout object} (appears transparent in
+the final output yet hides objects behind it).  ``Partial holdouts''
+may be designated by weighting the {\cf holdout()} closure by a
+weight that is less than 1.0.
+\apiend
+
+
+\apiitem{\colorclosure\ {\ce debug} (string outputname)}
+\indexapi{debug()}
+Returns a \colorclosure that does not represent any additional light
+reflection from the surface, but does signal to the renderer to add
+the weight of the closure (which may be a {\cf float} or a \color)
+to the named output (i.e., AOV).
+\apiend
+
+
+\subsection{Material utility functions}
+
+\apiitem{void {\ce artistic_ior} (color reflectivity, color edge_tint, \\
+\spc\spc output color ior, output color extinction)}
+\indexapi{artistic_ior()}
+Converts the artistic parameterization reflectivity and edge_tint to complex IOR values.
+To be used with the conductor_bsdf() closure.
+
+\noindent Parameters include:
+
+\apiitem{reflectivity}
+\vspace{12pt}
+Reflectivity per color channel at facing angles (\emph{r} parameter in [OG14])
+\apiend
+\vspace{-16pt}
+
+\apiitem{edge_tint}
+\vspace{12pt}
+Color bias for grazing angles (\emph{g} parameter in [OG14]).
+NOTE: This is not equal to 'f90' in a Schlick Fresnel parameterization.
+\apiend
+\vspace{-16pt}
+
+\apiitem{ior}
+\vspace{12pt}
+Output refraction index.
+\apiend
+\vspace{-16pt}
+  
+\apiitem{extinction}
+\vspace{12pt}
+Output extinction coefficient.
+\apiend
+%   \vspace{-16pt}
+
+Reference: [OG14] Ole Gulbrandsen, "Artist Friendly Metallic Fresnel", Journal
+of Computer Graphics Tools 3(4), 2014.
+\url{http://jcgt.org/published/0003/04/03/paper.pdf}
+\apiend
+
+
+\subsection{Deprecated closures}
+
+These were described in the original OSL language specification, but
+beginning with OSL 1.12, these are considered deprecated. Support for
+them will be removed entirely in OSL 2.0.
+
+\subsubsection{Deprecated Surface closures}

 \apiitem{\colorclosure\ {\ce diffuse} (normal N)}
 \indexapi{diffuse()}
@@ -4664,7 +5344,7 @@ Nayar, ``Generalization of Lambert's Reflectance Model,'' Proceedings of
 SIGGRAPH 1994, pp.239-246 (July, 1994).

 \begin{comment}
-Like all \closurecolors, the return ``value'' is symbolic and is may be
+Like all \closurecolors, the return ``value'' is symbolic and may be
 evaluated at a later time convenient to the renderer in order to compute
 the exitant radiance in the direction {\cf -I}.  But aside from the
 fact that the shader cannot examine the numeric values of the
@@ -4745,33 +5425,6 @@ implemented as follows:
 \end{comment}
 \apiend

-
-\apiitem{\colorclosure\ {\ce microfacet} (string distribution, normal N, vector
-U, \\ \bigspc\bigspc float xalpha, float yalpha, float eta, int refract)}
-\indexapi{microfacet()}
-
-Returns a \colorclosure that represents scattering on the
-surface using some microfacet distribution such as \qkw{beckmann} or \qkw{ggx} (as
-described in Walter et al., ``Microfacet Models for Refraction through Rough
-Surfaces,'' Eurographics Symposium on Rendering, 2007).
-Both {\cf xalpha} and {\cf yalpha} control how rough the microstructure of the
-surface is, possibly in an anisotropic way if they are different. In the
-anisotropic case {\cf U} must be a vector orthogonal to {\cf N} to specify the
-direction.  These values are expected to be the alpha values of the distribution
-as it appears in the literature, not a universal roughness measure.
-The {\cf eta} parameter is the index of refraction of the material, and
-{\cf refract} is a enumeration flag that controls if the BSDF is reflective ({\cf 0}),
-refractive ({\cf 1}) or both at once ({\cf 2}). The implementation must not rely on
-{\cf U} when the roughness is isotropic or the given vector is zero.
-
-The list of supported distributions will change from renderer to renderer, and
-also the behavior when an unsupported distribution is requested. It is up to the
-implementation to issue an error or to default to a supported one. Also it is
-expected that the alias \qkw{default} must work as a distribution in any renderer.
-
-\apiend
-
-
 \apiitem{\colorclosure\ {\ce microfacet} (string distribution, normal N,
 \\ \bigspc\bigspc float alpha, float eta, int refract)}
 \indexapi{microfacet()}
@@ -4823,7 +5476,6 @@ implemented as follows:
 \end{code}
 \apiend

-
 \apiitem{\colorclosure\ {\ce transparent} ( )}
 \indexapi{transparent()}

@@ -4867,8 +5519,7 @@ scattering} exhibited by the surface.
 \apiend
 \end{comment}

-
-\subsection{Volumetric material closures}
+\subsubsection{Deprecated Volumetric closures}

 \apiitem{\colorclosure\ {\ce isotropic} ( )}
 \indexapi{isotropic}
@@ -4877,18 +5528,16 @@ volumetric material, scattering light evenly in all directions,
 regardless of its original direction.
 \apiend

-
 \apiitem{\colorclosure\ {\ce henyey_greenstein} (float g)}
 \indexapi{henyey_greenstein}
 Returns a \colorclosure that represents the directional volumetric
 scattering by small suspended particles.  The {\cf g} parameter is the
-anisotropy factor, ranging from $-1$ to $1$), with positive values
+anisotropy factor, in the range $(-1, 1)$, with positive values
 indicating predominantly forward-scattering, negative values indicating
 predominantly back-scattering, and value of $g=0$ resulting in isotropic
 scattering.
 \apiend

-
 \apiitem{\colorclosure\ {\ce absorption} ( )}
 \indexapi{absorption}
 Returns a \colorclosure that does not represent any additional
@@ -4898,8 +5547,7 @@ volumetric material, scattering light evenly in all directions,
 regardless of its original direction.
 \apiend

-
-\subsection{Light emission closures}
+\subsubsection{Deprecated Emission closures}

 \apiitem{\colorclosure\ {\ce emission} ( )}
 \indexapi{emission}
@@ -4915,7 +5563,7 @@ $\mathrm{W}\cdot\mathrm{sr}^{-1}\cdot\mathrm{m}^{-2}$). This means that a surfac
 directly seen by the camera will directly reproduce the closure weight in the
 final pixel, regardless of being a surface or a volume.

-For surface emission, If you divide {\cf emission()} by 
+For an emissive surface, if you divide the return value of {\cf emission()} by
 {\cf surfacearea() * M_PI}, then you can easily specify the total emissive
 power of the light (e.g., $\mathrm{W}$), regardless of its physical size.
 \apiend
@@ -4930,30 +5578,6 @@ up from an HDRI environment map.
 \apiend


-\subsection{Signaling closures}
-
-\apiitem{\colorclosure\ {\ce holdout} ( )}
-\indexapi{holdout()}
-Returns a \colorclosure that does not represent any additional light
-reflection from the surface, but does signal to the renderer that 
-the surface is a \emph{holdout object} (appears transparent in
-the final output yet hides objects behind it).  ``Partial holdouts''
-may be designated by weighting the {\cf holdout()} closure by a
-weight that is less than 1.0.
-\apiend
-
-
-\apiitem{\colorclosure\ {\ce debug} (string outputname)}
-\indexapi{debug()}
-Returns a \colorclosure that does not represent any additional light
-reflection from the surface, but does signal to the renderer to add
-the weight of the closure (which may be a {\cf float} or a \color)
-to the named output (i.e., AOV).
-\apiend
-
-
-
-
 \newpage
 \section{Renderer state and message passing}
 \label{sec:stdlib:state}
@@ -5750,7 +6374,7 @@ programming language design)?

 This spec is written assuming that \color should explicitly be
 3-component RGB.  Alternately, we could add global \inttype\ {\cf ncomps} that gives the
-numer of components in a \color, and eliminate any built-in operations
+number of components in a \color, and eliminate any built-in operations
 and library functions that seem to require or imply a 3-component color
 model (including type casting between \color and the 3-component
 \point-like types).
@@ -5862,7 +6486,7 @@ anybody want to argue for any of the following to be included?

 We have defined structures as being allowed to consist only of basic
 types and arrays thereof --- not structures of structures.  This
-restriction makes for a simpler grammar and implementation, simplier
+restriction makes for a simpler grammar and implementation, simpler
 renderer-side APIs for passing data to shaders, and simpler APIs for
 libraries that query parameters of compiled shaders.  But if people
 strongly feel like structures-of-structures is an important feature, we
@@ -5948,7 +6572,7 @@ about it.

 \subsection{Nomenclature}

-Does anybody have a perference for \dPdu and \dPdv (borrowing the
+Does anybody have a preference for \dPdu and \dPdv (borrowing the
 RenderMan names, but awkward for primitives without a consistent $u,v$
 parameterization?), versus something more generic such as {\cf T} and
 {\cf B} (tangent and bitangent, without explicitly saying how they are
@@ -6134,6 +6758,9 @@ full control in the shader.
  may be shared among multiple geometric primitives.  Also sometimes
  called \emph{graphics state}.

+\item[BSDF.] Bidirectional scattering distribution function, a
+  function that describes light scattering properties of a surface.
+
 \item[Built-in function.] A function callable from within a shader, where
  the implementation of the function is provided by the renderer (as
  opposed to a function that the shader author writes in \langname
@@ -6144,7 +6771,7 @@ full control in the shader.
  at a later time to yield a final numeric value.

 \item[Connection.] A routing of the value of an \emph{output parameter}
-  of one \emph{shader layer} to an \emph{input paramter} of another
+  of one \emph{shader layer} to an \emph{input parameter} of another
  shader layer within the same \emph{shader group}.

 \item[Default parameter value.] The initial value of a \emph{shader
@@ -6155,6 +6782,9 @@ full control in the shader.
  is explicitly given in the code for that shader, and may either be
  a constant or a computed expression.

+\item[EDF.] Emission distribution function, a function that describes
+    the distribution of light emitted by a light source.
+
 \item[Geometric primitive.] A single shape, such as a NURBS patch, a
  polygon or subdivision mesh, a hair primitive, etc.

@@ -6190,7 +6820,7 @@ full control in the shader.

 \item[Primitive variable.] A named variable, and values, attached to an
  individual geometric primitive.  Primitive variables may have one of
-  serveral \emph{interpolation methods} --- such as a single value for
+  several \emph{interpolation methods} --- such as a single value for
  the whole primitive, a value for each piece or face of the primitive,
  or per-vertex values that are smoothly interpolated across the
  surface.
@@ -6199,12 +6829,12 @@ full control in the shader.
  that is visible and directly callable by the renderer, as opposed to
  merely being called from other code within the shader.  Public methods
  must be \emph{top-level} (not defined within other functions) and must
-  be preceeded by the {\cf public} keyword.
+  be preceded by the {\cf public} keyword.

 \item[Shader.] A small self-contained program written in \langname, used
  to extend the functionality of a renderer with custom behavior of
  materials and lights.  A particular shader may have multiple
-  \emph{shader instances} within a scene, each of which has ts unique
+  \emph{shader instances} within a scene, each of which has its unique
  \emph{instance parameters}, transformation, etc.

 \item[Shader function.] A function written in \langname that may be
@@ -6239,6 +6869,9 @@ full control in the shader.
 \item[Shading.] The computations within a renderer that implement the
  behavior and visual appearance of materials and lights.

+\item[VDF.] Volumetric distribution function, a function that describes
+  the scattering properties of a volume.
+
 \end{description}

 \end{appendix}
@@ -11,7 +11,7 @@
 _testshade_ basics
 ==================================================

-The OSL software distribution inculdes `testshade`, a command-line utility
+The OSL software distribution includes `testshade`, a command-line utility
 that was written for OSL unit tests, and for a "test harness" in which we
 can execute shaders for debugging (the shaders or OSL itself) in isolation
 from any particular rendering system.
@@ -28,7 +28,7 @@ itself, it has a number of potential uses in a production studio, including:
 * Baking procedural patterns into textures.

 * Benchmarking shaders and evaluating their performance, including
-  comparitive tests such as "is it faster to code it this way or that way?"
+  comparative tests such as "is it faster to code it this way or that way?"

 * Exploring the inner workings of OSL's runtime optimizer, to answer
  questions like: "will this idiom optimize away at runtime when I use
@@ -176,7 +176,7 @@ Helpful command line arguments:

 For the sake of simplicity, this document presents almost all examples as if
 they consisted of just a single shader node. But `testshade` is able to
-specify and execut entire shader groups (networks of shader nodes). The
+specify and execute entire shader groups (networks of shader nodes). The
 remainder of this section will explain how this can be done.

 Let's make a simple shader network as an example.
@@ -582,7 +582,7 @@ single texture lookup.



-## Controling optimization
+## Controlling optimization

 You can control the optimization level overall:

@@ -638,7 +638,7 @@ that they are not optimized away so as to lead you to an inaccurate
 reflection of what optimizations would really happen at render time. At the
 same time, be careful not to inadvertently set `lockgeom=0` for parameters
 that will typically be set to default values, instance values, or
-connections from layers that are likely to be analized and found to emit
+connections from layers that are likely to be analyzed and found to emit
 constants for theor outputs -- you also don't want your benchmarks to
 incorrectly disable optimizations what would typically happen at render
 time.
@@ -72,7 +72,7 @@ public:
    AccumRule(int outidx, bool toalpha)
        : m_outidx(outidx), m_save_to_alpha(toalpha) {};

-    /// Called to accumulate from AccumAutomata. It will select the right ouput from
+    /// Called to accumulate from AccumAutomata. It will select the right output from
    /// the given vector based in the AOV index number (they are guaranteed to match)
    void accum(const Color3& color, std::vector<AovOutput>& outputs) const;

@@ -36,20 +36,20 @@ OSL_NAMESPACE_ENTER
 // stored in the table, in the 16 bytes preceding the characters.

 struct DeviceString {
-#if defined(__CUDA_ARCH__) && OPTIX_VERSION >= 70000
+#ifdef __CUDA_ARCH__
    size_t m_chars;
 #else
    const char* m_chars;
 #endif

-#if defined(__CUDA_ARCH__) && OPTIX_VERSION >= 70000
+#ifdef __CUDA_ARCH__
    OSL_HOSTDEVICE DeviceString() {}
    OSL_HOSTDEVICE DeviceString(uint64_t i) : m_chars(i) {}
 #endif

    OSL_HOSTDEVICE uint64_t hash() const
    {
-#if defined(__CUDA_ARCH__) && OPTIX_VERSION >= 70000
+#ifdef __CUDA_ARCH__
        return m_chars;
 #else
        return *(uint64_t*)(m_chars - sizeof(uint64_t) - sizeof(uint64_t));
@@ -58,7 +58,7 @@ struct DeviceString {

    // In OptiX 7 we don't store the string's length. Make this a compile
    // time error.
-#if !(defined(__CUDA_ARCH__) && OPTIX_VERSION >= 70000)
+#ifndef __CUDA_ARCH__
    OSL_HOSTDEVICE uint64_t length() const
    {
        return *(uint64_t*)(m_chars - sizeof(uint64_t));
@@ -67,8 +67,11 @@ struct DeviceString {

    // In OptiX 7 we can't return the string's contents. Make this a compile
    // time error.
-#if !(defined(__CUDA_ARCH__) && OPTIX_VERSION >= 70000)
-    OSL_HOSTDEVICE const char* c_str() const { return m_chars; }
+#ifndef __CUDA_ARCH__
+    OSL_HOSTDEVICE const char* c_str() const
+    {
+        return m_chars;
+    }
 #endif

    OSL_HOSTDEVICE bool operator==(const DeviceString& other) const
@@ -81,7 +84,7 @@ struct DeviceString {
        return m_chars != other.m_chars;
    }

-#if defined(__CUDA_ARCH__) && OPTIX_VERSION >= 70000
+#ifdef __CUDA_ARCH__

    OSL_HOSTDEVICE bool operator==(const size_t other) const
    {
@@ -93,7 +96,10 @@ struct DeviceString {
        return hash() != other;
    }

-    OSL_HOSTDEVICE operator size_t() const { return hash(); }
+    OSL_HOSTDEVICE operator size_t() const
+    {
+        return hash();
+    }

 #endif
 };
@@ -123,9 +129,6 @@ typedef DeviceString StringParam;
 namespace DeviceStrings {
 #    define STRDECL(str, var_name) \
        extern __device__ OSL_NAMESPACE::DeviceString var_name;
-#    if OPTIX_VERSION < 70000
-#        include <OSL/strdecls.h>
-#    endif
 #    undef STRDECL
 }  // namespace DeviceStrings
 #else
@@ -140,10 +143,6 @@ namespace DeviceStrings {
 OSL_NAMESPACE_EXIT


-#ifdef __CUDA_ARCH__
-#    if OPTIX_VERSION < 70000
-namespace StringParams = OSL_NAMESPACE::DeviceStrings;
-#    endif
-#else
+#ifndef __CUDA_ARCH__
 namespace StringParams = OSL_NAMESPACE::Strings;
 #endif
@@ -223,7 +223,7 @@ public:
 // delegating access to element storage to the DualStorage base class
 template<
         class T,           // Base data type
-         int PARTIALS=1     // Number of dimentions of partial derivs
+         int PARTIALS=1     // Number of dimensions of partial derivs
        >
 class Dual : public DualStorage<T, PARTIALS> {
    static const int elements = PARTIALS+1;   // main value + partials
@@ -108,6 +108,12 @@ public:
    void jit_aggressive(bool val) { m_jit_aggressive = val; }
    bool jit_aggressive() const { return m_jit_aggressive; }

+    // Select whether the representation of a ustring is going to be
+    // the character pointer, or the hash.
+    enum class UstringRep { charptr, hash };
+    UstringRep ustring_rep() const { return m_ustring_rep; }
+    void ustring_rep(UstringRep rep);
+
    /// Return a reference to the current context.
    llvm::LLVMContext& context() const { return *m_llvm_context; }

@@ -148,15 +154,6 @@ public:
    void debug_pop_inlined_function();
    void debug_set_location(OIIO::ustring sourcefile, int sourceline);

-    /// Create a new function (that will later be populated with
-    /// instructions) with up to 4 args.
-    OSL_DEPRECATED("Use make_function flavor that takes a cspan (1.12)")
-    llvm::Function* make_function(const std::string& name, bool fastcall,
-                                  llvm::Type* rettype, llvm::Type* arg1 = NULL,
-                                  llvm::Type* arg2 = NULL,
-                                  llvm::Type* arg3 = NULL,
-                                  llvm::Type* arg4 = NULL);
-
    /// Create a new function (that will later be populated with
    /// instructions) with a cspan of args.
    llvm::Function* make_function(const std::string& name, bool fastcall,
@@ -223,7 +220,7 @@ public:

    // Look up the TargetISA enum by name. Special names: "host" means
    // figure out what this host is, "none" or "" mean to use the baseline
-    // or no special ops not availabe on all flavors of this family of CPUs.
+    // or no special ops not available on all flavors of this family of CPUs.
    static TargetISA lookup_isa_by_name(string_view target_name = "");

    // Name for this TargetISA enum.
@@ -283,12 +280,7 @@ public:
        Linkage default_linkage = Linkage::Internal,
        std::string* out_err    = nullptr);

-    // OLD, might deprecate later
-    /// Change symbols in the module that are marked as having external
-    /// linkage to an alternate linkage that allows them to be discarded if
-    /// not used within the module. Only do this for functions that start
-    /// with prefix, and that DON'T match anything in the two exceptions
-    /// lists.
+    OSL_DEPRECATED("prune_and_internalize_module is better (1.13)")
    void internalize_module_functions(
        const std::string& prefix, const std::vector<std::string>& exceptions,
        const std::vector<std::string>& moreexceptions);
@@ -311,7 +303,7 @@ public:

    /// Create a new LLVM basic block (for the current function) and return
    /// its handle.
-    llvm::BasicBlock* new_basic_block(const std::string& name = std::string());
+    llvm::BasicBlock* new_basic_block(const std::string& name = {});

    /// Save the return block pointer when entering a function. If
    /// after==NULL, generate a new basic block for where to go after the
@@ -383,13 +375,13 @@ public:
    // vs. relying on a conditional result already in a register.
    //
    // To handle 'continue', we allocate a 'continue mask' that is
-    // initially populated with a 0 representing the deactived lanes by
+    // initially populated with a 0 representing the deactivated lanes by
    // 'continue' operations. We require a location on the stack for each
    // 'continue mask' in order for a 'continue' statement to modify it and
    // cause a subset of the lanes to be deactivated.  If an continue was
    // processed, the 'continue mask' is where it will be stored and pulled
    // from when applying to the unwinding conditional mask stack. When a
-    // loop's scope is exited, the continue mask is popped so it would't be
+    // loop's scope is exited, the continue mask is popped so it wouldn't be
    // applied to an outer loop.
    //
    // The logic of when and how to actually apply these masks is handled
@@ -543,8 +535,8 @@ public:
    llvm::Type* type_triple() const { return m_llvm_type_triple; }
    llvm::Type* type_matrix() const { return m_llvm_type_matrix; }
    llvm::Type* type_typedesc() const { return m_llvm_type_longlong; }
+    llvm::Type* type_ustring() { return m_llvm_type_ustring; }
    llvm::PointerType* type_void_ptr() const { return m_llvm_type_void_ptr; }
-    llvm::PointerType* type_string() { return m_llvm_type_char_ptr; }
    llvm::PointerType* type_ustring_ptr() const
    {
        return m_llvm_type_ustring_ptr;
@@ -579,7 +571,8 @@ public:
    llvm::Type* type_wide_triple() const { return m_llvm_type_wide_triple; }
    llvm::Type* type_wide_matrix() const { return m_llvm_type_wide_matrix; }
    llvm::Type* type_wide_void_ptr() const { return m_llvm_type_wide_void_ptr; }
-    llvm::Type* type_wide_string() const
+    llvm::Type* type_wide_ustring() const { return m_llvm_type_wide_ustring; }
+    llvm::PointerType* type_wide_ustring_ptr() const
    {
        return m_llvm_type_wide_ustring_ptr;
    }
@@ -629,7 +622,7 @@ public:
                            bool is_packed          = false);

    /// Return the llvm::Type that is a pointer to the given llvm type.
-    llvm::Type* type_ptr(llvm::Type* type);
+    llvm::PointerType* type_ptr(llvm::Type* type);

    /// Return the llvm::Type that is the wide version of the given llvm type
    llvm::Type* type_wide(llvm::Type* type);
@@ -660,26 +653,26 @@ public:
    /// Return the human-readable name of the type of the llvm value.
    std::string llvm_typenameof(llvm::Value* val) const;

-    /// Return an llvm::Value holding the given floating point constant.
+    /// Return an llvm::Constant holding the given floating point constant.
    llvm::Constant* constant(float f);

-    /// Return an llvm::Value holding the given integer constant.
+    /// Return an llvm::Constant holding the given integer constant.
    llvm::Constant* constant(int i);

-    /// Return an llvm::Value holding the given integer constant.
+    /// Return an llvm::Constant holding the given integer constant.
    llvm::Constant* constant8(int i);
    llvm::Constant* constant16(uint16_t i);
    llvm::Constant* constant64(uint64_t i);
    llvm::Constant* constant128(uint64_t i);
    llvm::Constant* constant128(uint64_t left, uint64_t right);

-    // Return an llvm::value holding an explicit signed int64_t constant.
+    // Return an llvm::Constant holding an explicit signed int64_t constant.
    llvm::Constant* constanti64(int64_t i);

-    /// Return an llvm::Value holding the given size_t constant.
+    /// Return an llvm::Constant holding the given size_t constant.
    llvm::Constant* constant(size_t i);

-    /// Return an llvm::Value holding the given bool constant.
+    /// Return an llvm::Constant holding the given bool constant.
    /// Change the name so it doesn't get mixed up with int.
    llvm::Constant* constant_bool(bool b);

@@ -687,7 +680,8 @@ public:
    /// If the type specified is NULL, it will make a 'void *'.
    llvm::Value* constant_ptr(void* p, llvm::PointerType* type = NULL);

-    /// Return an llvm::Value holding the given string constant.
+    /// Return an llvm::Value holding the given string constant (as
+    /// determined by the ustring_rep).
    llvm::Value* constant(ustring s);
    llvm::Value* constant(string_view s) { return constant(ustring(s)); }

@@ -735,19 +729,23 @@ public:

    /// Cast the pointer variable specified by val to the kind of pointer
    /// described by type (as an llvm pointer type).
-    llvm::Value* ptr_cast(llvm::Value* val, llvm::Type* type);
-    llvm::Value* ptr_cast(llvm::Value* val, llvm::PointerType* type)
+    llvm::Value* ptr_cast(llvm::Value* val, llvm::Type* type,
+                          const std::string& llname = {});
+    llvm::Value* ptr_cast(llvm::Value* val, llvm::PointerType* type,
+                          const std::string& llname = {})
    {
-        return ptr_cast(val, (llvm::Type*)type);
+        return ptr_cast(val, (llvm::Type*)type, llname);
    }

    /// Cast the pointer variable specified by val to a pointer to the type
    /// described by type (as an llvm data type).
-    llvm::Value* ptr_to_cast(llvm::Value* val, llvm::Type* type);
+    llvm::Value* ptr_to_cast(llvm::Value* val, llvm::Type* type,
+                             const std::string& llname = {});

    /// Cast the pointer variable specified by val to a pointer to the given
    /// data type, return the llvm::Value of the new pointer.
-    llvm::Value* ptr_cast(llvm::Value* val, const OIIO::TypeDesc& type);
+    llvm::Value* ptr_cast(llvm::Value* val, const OIIO::TypeDesc& type,
+                          const std::string& llname = {});

    llvm::Value* wide_ptr_cast(llvm::Value* val, const OIIO::TypeDesc& type);

@@ -757,7 +755,7 @@ public:

    /// Cast the pointer variable specified by val to a pointer of type
    /// void* return the llvm::Value of the new pointer.
-    llvm::Value* void_ptr(llvm::Value* val);
+    llvm::Value* void_ptr(llvm::Value* val, const std::string& llname = {});

    /// Generate a pointer that is (ptrtype)((char *)ptr + offset).
    /// If ptrtype is NULL, just return a void*.
@@ -865,9 +863,10 @@ public:

    /// Dereference a pointer:  return *ptr
    /// type is the type of the thing being pointed to.
-    llvm::Value* op_load(llvm::Type* type, llvm::Value* ptr);
+    llvm::Value* op_load(llvm::Type* type, llvm::Value* ptr,
+                         const std::string& llname = {});
    // Blind pointer version that's deprecated as of LLVM13:
-    llvm::Value* op_load(llvm::Value* ptr);
+    llvm::Value* op_load(llvm::Value* ptr, const std::string& llname = {});

    llvm::Value* op_gather(llvm::Value* ptr, llvm::Value* index);

@@ -898,24 +897,30 @@ public:
    /// llvm::Value, which can be generated from either a constant or a
    /// runtime-computed integer element index. `type` is the type of the data
    /// we're retrieving.
-    llvm::Value* GEP(llvm::Type* type, llvm::Value* ptr, llvm::Value* elem);
+    llvm::Value* GEP(llvm::Type* type, llvm::Value* ptr, llvm::Value* elem,
+                     const std::string& llname = {});
    // Blind pointer version that's deprecated as of LLVM13:
-    llvm::Value* GEP(llvm::Value* ptr, llvm::Value* elem);
+    llvm::Value* GEP(llvm::Value* ptr, llvm::Value* elem,
+                     const std::string& llname = {});

    /// Generate a GEP (get element pointer) with an integer element
    /// offset. `type` is the type of the data we're retrieving.
-    llvm::Value* GEP(llvm::Type* type, llvm::Value* ptr, int elem);
+    llvm::Value* GEP(llvm::Type* type, llvm::Value* ptr, int elem,
+                     const std::string& llname = {});
    // Blind pointer version that's deprecated as of LLVM13:
-    llvm::Value* GEP(llvm::Value* ptr, int elem);
+    llvm::Value* GEP(llvm::Value* ptr, int elem,
+                     const std::string& llname = {});

    /// Generate a GEP (get element pointer) with two integer element
    /// offsets.  This is just a special (and common) case of GEP where
    /// we have a 2-level hierarchy and we have fixed element indices
    /// that are known at compile time.  `type` is the type of the data we're
    /// retrieving.
-    llvm::Value* GEP(llvm::Type* type, llvm::Value* ptr, int elem1, int elem2);
+    llvm::Value* GEP(llvm::Type* type, llvm::Value* ptr, int elem1, int elem2,
+                     const std::string& llname = {});
    // Blind pointer version that's deprecated as of LLVM13:
-    llvm::Value* GEP(llvm::Value* ptr, int elem1, int elem2);
+    llvm::Value* GEP(llvm::Value* ptr, int elem1, int elem2,
+                     const std::string& llname = {});

    // Arithmetic ops.  It auto-detects the type (int vs float).
    // ...
@@ -1009,9 +1014,10 @@ private:
    IRBuilder& builder();

    int m_debug;
-    bool m_dumpasm        = false;
-    bool m_jit_fma        = false;
-    bool m_jit_aggressive = false;
+    bool m_dumpasm           = false;
+    bool m_jit_fma           = false;
+    bool m_jit_aggressive    = false;
+    UstringRep m_ustring_rep = UstringRep::charptr;
    PerThreadInfo::Impl* m_thread;
    llvm::LLVMContext* m_llvm_context;
    llvm::Module* m_llvm_module;
@@ -1027,6 +1033,9 @@ private:
    std::vector<llvm::BasicBlock*> m_loop_after_block;  // stack for break
    std::vector<llvm::BasicBlock*> m_loop_step_block;   // stack for continue

+    // Incrementing serial numbers
+    int m_next_serial_bb = 0;
+
    llvm::Type* m_llvm_type_float;
    llvm::Type* m_llvm_type_double;
    llvm::Type* m_llvm_type_int;
@@ -1039,8 +1048,8 @@ private:
    llvm::Type* m_llvm_type_void;
    llvm::Type* m_llvm_type_triple;
    llvm::Type* m_llvm_type_matrix;
+    llvm::Type* m_llvm_type_ustring;
    llvm::PointerType* m_llvm_type_void_ptr;
-    llvm::PointerType* m_llvm_type_ustring_ptr;
    llvm::PointerType* m_llvm_type_char_ptr;
    llvm::PointerType* m_llvm_type_bool_ptr;
    llvm::PointerType* m_llvm_type_int_ptr;
@@ -1049,6 +1058,7 @@ private:
    llvm::PointerType* m_llvm_type_triple_ptr;
    llvm::PointerType* m_llvm_type_matrix_ptr;
    llvm::PointerType* m_llvm_type_double_ptr;
+    llvm::PointerType* m_llvm_type_ustring_ptr;

    int m_vector_width;
    llvm::Type* m_llvm_type_wide_float;
@@ -1060,11 +1070,12 @@ private:
    llvm::Type* m_llvm_type_wide_triple;
    llvm::Type* m_llvm_type_wide_matrix;
    llvm::Type* m_llvm_type_wide_void_ptr;
-    llvm::Type* m_llvm_type_wide_ustring_ptr;
+    llvm::Type* m_llvm_type_wide_ustring;
    llvm::PointerType* m_llvm_type_wide_char_ptr;
    llvm::PointerType* m_llvm_type_wide_int_ptr;
    llvm::PointerType* m_llvm_type_wide_bool_ptr;
    llvm::PointerType* m_llvm_type_wide_float_ptr;
+    llvm::PointerType* m_llvm_type_wide_ustring_ptr;
    llvm::Type* m_llvm_type_native_mask;

    bool m_supports_masked_stores           = false;
@@ -4,7 +4,6 @@

 #pragma once

-#include <immintrin.h>
 #include <type_traits>

 #include <OSL/oslconfig.h>
@@ -23,6 +22,8 @@ using std::countr_zero;

 #elif OSL_INTEL_CLASSIC_COMPILER_VERSION

+#include <immintrin.h>
+
 OSL_FORCEINLINE int popcount(uint32_t x) noexcept { return _mm_popcnt_u32(x);}
 OSL_FORCEINLINE int popcount(uint64_t x) noexcept { return _mm_popcnt_u64(x); }
 OSL_FORCEINLINE int countr_zero(uint32_t x) noexcept { return _bit_scan_forward(x); }
@@ -170,13 +171,22 @@ public:
    }


-    OSL_FORCEINLINE ValueType value() const { return m_value; }
+    OSL_FORCEINLINE ValueType value() const
+    {
+        return m_value;
+    }

    // count number of active bits
-    OSL_FORCEINLINE int count() const { return OSL::popcount(m_value); }
+    OSL_FORCEINLINE int count() const
+    {
+        return OSL::popcount(m_value);
+    }

    // NOTE: undefined result if no bits are on
-    OSL_FORCEINLINE int first_on() const { return OSL::countr_zero(m_value); }
+    OSL_FORCEINLINE int first_on() const
+    {
+        return OSL::countr_zero(m_value);
+    }

    OSL_FORCEINLINE Mask invert() const
    {
@@ -229,7 +239,10 @@ public:
        return (m_value != static_cast<ValueType>(0));
    }

-    OSL_FORCEINLINE bool any_off() const { return (m_value < valid_bits); }
+    OSL_FORCEINLINE bool any_off() const
+    {
+        return (m_value < valid_bits);
+    }

    OSL_FORCEINLINE bool any_off(const Mask& mask) const
    {
@@ -244,27 +257,39 @@ public:
    // So really only set_on or set_off is required
    // Choose to not provide a generic set(int lane, bool flag)

-    OSL_FORCEINLINE void set_on(int lane) { m_value |= (1 << lane); }
+    OSL_FORCEINLINE void set_on(int lane)
+    {
+        m_value |= (1 << lane);
+    }

    OSL_FORCEINLINE void set_on_if(int lane, bool cond)
    {
        m_value |= (cond << lane);
    }

-    OSL_FORCEINLINE void set_all_on() { m_value = valid_bits; }
+    OSL_FORCEINLINE void set_all_on()
+    {
+        m_value = valid_bits;
+    }
    OSL_FORCEINLINE void set_count_on(int count)
    {
        m_value = valid_bits >> (width - count);
    }

-    OSL_FORCEINLINE void set_off(int lane) { m_value &= (~(1 << lane)); }
+    OSL_FORCEINLINE void set_off(int lane)
+    {
+        m_value &= (~(1 << lane));
+    }

    OSL_FORCEINLINE void set_off_if(int lane, bool cond)
    {
        m_value &= (~(cond << lane));
    }

-    OSL_FORCEINLINE void set_all_off() { m_value = static_cast<ValueType>(0); }
+    OSL_FORCEINLINE void set_all_off()
+    {
+        m_value = static_cast<ValueType>(0);
+    }

    OSL_FORCEINLINE bool operator==(const Mask& other) const
    {
@@ -298,7 +323,10 @@ public:
        return Mask(m_value | other.m_value);
    }

-    OSL_FORCEINLINE Mask operator~() const { return invert(); }
+    OSL_FORCEINLINE Mask operator~() const
+    {
+        return invert();
+    }


    template<int MinOccupancyT, int MaxOccupancyT = width, typename FunctorT>
@@ -109,8 +109,14 @@ OSL_ALIGNAS(16) ClosureComponent : public ClosureColor
    Vec3 w;  ///< Weight of this component

    /// Handy method for getting the parameter memory as a void*.
-    OSL_HOSTDEVICE void* data() { return (char*)(this + 1); }
-    OSL_HOSTDEVICE const void* data() const { return (const char*)(this + 1); }
+    OSL_HOSTDEVICE void* data()
+    {
+        return (char*)(this + 1);
+    }
+    OSL_HOSTDEVICE const void* data() const
+    {
+        return (const char*)(this + 1);
+    }

    /// Handy methods for extracting the underlying parameters as a struct
    template<typename T> OSL_HOSTDEVICE const T* as() const
@@ -102,7 +102,7 @@ using OIIO::TextureOpt;
 // And some other things we borrow from OIIO...
 using OIIO::ErrorHandler;
 using OIIO::ustring;
-using OIIO::ustringHash;
+using OIIO::ustringhash;
 using OIIO::string_view;
 using OIIO::span;
 using OIIO::cspan;
@@ -225,3 +225,30 @@ using ConstIndex = std::integral_constant<int, N>;


 OSL_NAMESPACE_EXIT
+
+
+namespace std {  // not necessary in C++17, then we can just say std::hash
+#ifndef OIIO_USTRING_HAS_STDHASH
+// Not a new enough OIIO to define std::hash<ustring>, so we'll do it
+template<> struct hash<OSL::ustring> {
+    std::size_t operator()(OSL::ustring u) const noexcept { return u.hash(); }
+};
+
+template<> struct hash<OSL::ustringhash> {
+    OSL_HOSTDEVICE constexpr std::size_t
+    operator()(OSL::ustringhash u) const noexcept
+    {
+        return u.hash();
+    }
+};
+#endif
+
+// Not necessary once minimum OIIO defines operator< for ustringhash
+template<> struct less<OSL::ustringhash> {
+    OSL_HOSTDEVICE constexpr bool operator()(OSL::ustringhash u,
+                                             OSL::ustringhash v) const noexcept
+    {
+        return u.hash() < v.hash();
+    }
+};
+}  // namespace std
@@ -50,7 +50,7 @@ namespace pvt {
 class ShadingSystemImpl;
 }

-#if defined(__CUDA_ARCH__) && OPTIX_VERSION >= 70000
+#ifdef __CUDA_ARCH__
 #    define STRING_PARAMS(x) UStringHash::Hash(__OSL_STRINGIFY(x))
 #else
 #    define STRING_PARAMS(x) StringParams::x
@@ -58,11 +58,7 @@ class ShadingSystemImpl;

 namespace Strings {
 #ifdef __CUDA_ARCH__
-#    if OPTIX_VERSION >= 70000
-#        define STRDECL(str, var_name)
-#    else
-#        define STRDECL(str, var_name) extern __device__ ustring var_name;
-#    endif
+#    define STRDECL(str, var_name)
 #else
 // Any strings referenced inside of a libsoslexec/wide/*.cpp
 // or liboslnoise/wide/*.cpp will need OSLEXECPUBLIC
@@ -235,8 +231,7 @@ public:
    ///                              for debugging. (0)
    ///    int llvm_dumpasm       Print the CPU assembly code from the JIT (0)
    ///    string llvm_prune_ir_strategy  Strategy for pruning unnecessary
-    ///                              IR (choices: "prune" [default],
-    ///                              "internalize", or "none").
+    ///                              IR (choices: "prune" [default], or "none").
    ///    int max_local_mem_KB   Error if shader group needs more than this
    ///                              much local storage to execute (1024K)
    ///    string debug_groupname Name of shader group -- debug only this one
@@ -677,6 +672,15 @@ public:
                 ShaderGlobals& globals, void* userdata_base_ptr,
                 void* output_base_ptr, bool run = true);

+    /// Future execute signature that will be range based. Shader globals will be
+    /// obtained from renderer services.
+#if 0  // TODO in future PR
+    bool execute(ShadingContext& ctx, ShaderGroup& group, OpaqueShadingStateUniformPtr ssu,
+                 int shadeindex_begin_at, int shadeindex_end_before
+                 void* userdata_base_ptr,
+                 void* output_base_ptr, bool run = true);
+#endif
+
    // DEPRECATED(2.0): no shadeindex or base pointers
    bool execute(ShadingContext& ctx, ShaderGroup& group,
                 ShaderGlobals& globals, bool run = true)
@@ -141,6 +141,19 @@ inline OSL_HOSTDEVICE float bits_to_01 (unsigned int bits) {
 }


+OSL_FORCEINLINE OSL_HOSTDEVICE int
+bitcast_to_uint (float x)
+{
+#if OPENIMAGEIO_VERSION >= 20500
+    return OIIO::bitcast<unsigned int, float>(x);
+#else
+    // obsolete call
+    return OIIO::bit_cast<float, unsigned int>(x);
+#endif
+}
+
+
+
 #ifndef __CUDA_ARCH__
 // Perform a bjmix (see OpenImageIO/hash.h) on 4 sets of values at once.
 OSL_FORCEINLINE void
@@ -231,7 +244,7 @@ reference_inthash (const unsigned int k[N]) {
 	OSL_FORCEINLINE OSL_HOSTDEVICE unsigned int
 	inthash (const unsigned int k0, const unsigned int k1) {
 		// now hash the data!
-		unsigned int start_val = 0xdeadbeef + (2 << 2) + 13;
+		unsigned int start_val = 0xdeadbeef + (2 << 2) + 13;  //NOSONAR

 		unsigned int a = start_val + k0;
 		unsigned int b = start_val + k1;
@@ -467,7 +480,7 @@ struct HashNoise: public IntHashNoiseBase<HashNoise>  {
    static OSL_FORCEINLINE OSL_HOSTDEVICE unsigned int
    transformToUint(float val)
    {
-        return OIIO::bit_cast<float,unsigned int>(val);
+        return bitcast_to_uint(val);
    }
 };

@@ -544,7 +557,6 @@ using PeriodicHashNoise = PeriodicAdaptionOf<HashNoise>;



-
 inline OSL_HOSTDEVICE int
 inthashi (int x)
 {
@@ -557,7 +569,7 @@ inline OSL_HOSTDEVICE int
 inthashf (float x)
 {
    return static_cast<int>(
-		inthash(OIIO::bit_cast<float,unsigned int>(x))
+		inthash(bitcast_to_uint(x))
 	);
 }

@@ -566,8 +578,8 @@ inthashf (float x, float y)
 {
    return static_cast<int>(
 		inthash(
-			OIIO::bit_cast<float,unsigned int>(x),
-			OIIO::bit_cast<float,unsigned int>(y)
+			bitcast_to_uint(x),
+			bitcast_to_uint(y)
 		)
 	);
 }
@@ -578,9 +590,9 @@ inthashf (const float *x)
 {
    return static_cast<int>(
 		inthash(
-			OIIO::bit_cast<float,unsigned int>(x[0]),
-			OIIO::bit_cast<float,unsigned int>(x[1]),
-			OIIO::bit_cast<float,unsigned int>(x[2])
+			bitcast_to_uint(x[0]),
+			bitcast_to_uint(x[1]),
+			bitcast_to_uint(x[2])
 		)
 	);
 }
@@ -591,10 +603,10 @@ inthashf (const float *x, float y)
 {
    return static_cast<int>(
 		inthash(
-			OIIO::bit_cast<float,unsigned int>(x[0]),
-			OIIO::bit_cast<float,unsigned int>(x[1]),
-			OIIO::bit_cast<float,unsigned int>(x[2]),
-			OIIO::bit_cast<float,unsigned int>(y)
+			bitcast_to_uint(x[0]),
+			bitcast_to_uint(x[1]),
+			bitcast_to_uint(x[2]),
+			bitcast_to_uint(y)
 		)
 	);
 }
@@ -1154,16 +1166,16 @@ private:
    OSL_FORCEINLINE OSL_HOSTDEVICE HashScalarPeriodic () {}
 public:
    OSL_FORCEINLINE OSL_HOSTDEVICE HashScalarPeriodic (float px) {
-        m_px = OIIO::ifloor(px); if (m_px < 1) m_px = 1;
+        m_px = OIIO::ifloor(px); if (m_px < 1) m_px = 1;  //NOSONAR
    }
    OSL_FORCEINLINE OSL_HOSTDEVICE HashScalarPeriodic (float px, float py) {
        m_px = OIIO::ifloor(px); if (m_px < 1) m_px = 1;
-        m_py = OIIO::ifloor(py); if (m_py < 1) m_py = 1;
+        m_py = OIIO::ifloor(py); if (m_py < 1) m_py = 1;  //NOSONAR
    }
    OSL_FORCEINLINE OSL_HOSTDEVICE HashScalarPeriodic (float px, float py, float pz) {
        m_px = OIIO::ifloor(px); if (m_px < 1) m_px = 1;
        m_py = OIIO::ifloor(py); if (m_py < 1) m_py = 1;
-        m_pz = OIIO::ifloor(pz); if (m_pz < 1) m_pz = 1;
+        m_pz = OIIO::ifloor(pz); if (m_pz < 1) m_pz = 1;  //NOSONAR
    }
    OSL_FORCEINLINE OSL_HOSTDEVICE HashScalarPeriodic (float px, float py, float pz, float pw) {
        m_px = OIIO::ifloor(px); if (m_px < 1) m_px = 1;
@@ -1225,16 +1237,16 @@ public:

 struct HashVectorPeriodic {
 	OSL_FORCEINLINE OSL_HOSTDEVICE HashVectorPeriodic (float px) {
-        m_px = OIIO::ifloor(px); if (m_px < 1) m_px = 1;
+        m_px = OIIO::ifloor(px); if (m_px < 1) m_px = 1;  //NOSONAR
    }
 	OSL_FORCEINLINE OSL_HOSTDEVICE HashVectorPeriodic (float px, float py) {
        m_px = OIIO::ifloor(px); if (m_px < 1) m_px = 1;
-        m_py = OIIO::ifloor(py); if (m_py < 1) m_py = 1;
+        m_py = OIIO::ifloor(py); if (m_py < 1) m_py = 1;  //NOSONAR
    }
 	OSL_FORCEINLINE OSL_HOSTDEVICE HashVectorPeriodic (float px, float py, float pz) {
        m_px = OIIO::ifloor(px); if (m_px < 1) m_px = 1;
        m_py = OIIO::ifloor(py); if (m_py < 1) m_py = 1;
-        m_pz = OIIO::ifloor(pz); if (m_pz < 1) m_pz = 1;
+        m_pz = OIIO::ifloor(pz); if (m_pz < 1) m_pz = 1;  //NOSONAR
    }
 	OSL_FORCEINLINE OSL_HOSTDEVICE HashVectorPeriodic (float px, float py, float pz, float pw) {
        m_px = OIIO::ifloor(px); if (m_px < 1) m_px = 1;
@@ -1881,7 +1893,7 @@ OSL_FORCEINLINE OSL_HOSTDEVICE void perlin (Vec3 &result, const H &hash,
                            u, v, w);
    result = scale3 (result);
 #else
-    static_assert(std::is_same<Vec3i, decltype(hash(X, Y, Z))>::value, "This re-implementation was developed for Hashs returning a vector type only");
+    static_assert(std::is_same<Vec3i, decltype(hash(X, Y, Z))>::value, "This re-implementation was developed for hashes returning a vector type only");
    // Want to avoid repeating the same hash work 3 times in a row
    // so rather than executing the vector version, we will use HashScalar
    // and directly mask its results on a per component basis
@@ -61,7 +61,7 @@ class OSOReaderQuery;  // Just so OSLQuery can friend OSLReaderQuery
 /// ~~~
 ///
 /// And the `Parameter` structure will hold all the information you need
-/// about that paramter. For example:
+/// about that parameter. For example:
 ///
 /// ~~~
 ///     std::cout << "Parameter " << p->name
@@ -123,7 +123,7 @@ public:
    }
    ///< Construct an OSLQuery and open a compiled shader from a disk file.
    /// The `shadername` may be either the name of the `.oso` file, or the
-    /// name of the shader. The optional `searchpath` paramter gives a
+    /// name of the shader. The optional `searchpath` parameter gives a
    /// colon-separated list of directories to search for compiled shaders.

    OSL_DEPRECATED("Use ShadingSystem::oslquery(group,layernum)")
@@ -482,11 +482,19 @@
 #    define OSL_DASSERT_MSG(x, ...) ((void)sizeof(x)) /*NOLINT*/
 #endif

-#if defined(_MSC_VER) && !defined(__INTEL_COMPILER)
-    #define OSL_STACK_ARRAY(TYPE, NAME, DYNAMIC_COUNT) TYPE * NAME = reinterpret_cast<TYPE *>(alloca(sizeof(TYPE)*(DYNAMIC_COUNT)))
+
+/// OSL_ALLOCA is used to allocate smallish amount of memory on the stack,
+/// equivalent of C99 type var_name[size].
+///
+/// NOTE: in a debug build, this will assert for allocations >= 1MB, which is
+/// much too big. Hopefully this will keep us from abusing alloca and having
+/// stack overflows. The rule of thumb is that it's ok to use alloca for small
+/// things of bounded size, but not for anything that could be arbitrarily
+/// big, which should instead use heap allocation.
+#if defined(__GNUC__)
+#    define OSL_ALLOCA(type, size) (assert(size < (1<<20)), (size) != 0 ? (reinterpret_cast<type*>(__builtin_alloca((size) * sizeof(type)))) : nullptr)
 #else
-    // Utilize variable length array compiler extension when available
-    #define OSL_STACK_ARRAY(TYPE, NAME, DYNAMIC_COUNT) TYPE NAME[DYNAMIC_COUNT]
+#    define OSL_ALLOCA(type, size) (assert(size < (1<<20)), (size) != 0 ? (reinterpret_cast<type*>(alloca((size) * sizeof(type)))) : nullptr)
 #endif

 OSL_NAMESPACE_ENTER
@@ -405,15 +405,6 @@ public:
    /// Return a pointer to the texture system (if available).
    virtual TextureSystem *texturesys () const;

-    /// Register a string with the renderer, so that the renderer can arrange
-    /// to make the string available in GPU memory as needed. Optionally specify
-    /// a variable name to associate with the string value.
-    virtual uint64_t register_string (const std::string& str,
-                                      const std::string& var_name)
-    {
-        return 0;
-    }
-
    virtual uint64_t register_global (const std::string& var_name,
                                      uint64_t           value)
    {
@@ -145,7 +145,7 @@ namespace sfm
        //return static_cast<int>(rintf(x));
        return rintf(x);
    #else
-        // emulate rounding by adding/substracting 0.5
+        // emulate rounding by adding/subtracting 0.5
        return static_cast<int>(x + copysignf(0.5f, x));

        // Other possible factorings
@@ -441,7 +441,7 @@ namespace sfm

    // Considering having functionally equivalent versions of Vec3, Color3, Matrix44
    // with slight modifications to inlining and implementation to avoid aliasing and
-    // improve likelyhood of proper privation of local variables within a SIMD loop
+    // improve likelihood of proper privation of local variables within a SIMD loop

 }  // namespace sfm

@@ -17,6 +17,9 @@ class RendererServices;
 /// coordinate transformation.
 typedef const void* TransformationPtr;

+/// Type for an opaque pointer to whatever the Shading System uses to represent
+/// Uniform State.
+typedef const void* OpaqueShadingStateUniformPtr;


 /// The ShaderGlobals structure represents the state describing a particular
@@ -83,6 +86,7 @@ struct ShaderGlobals {
    /// Back-pointer to the ShadingContext (set and used by OSL itself --
    /// renderers shouldn't mess with this at all).
    ShadingContext* context;
+    OpaqueShadingStateUniformPtr shadingStateUniform;

    /// Pointer to the RendererServices object. This is how OSL finds its
    /// way back to the renderer for callbacks.
@@ -194,17 +194,17 @@ template<int WidthT> struct alignas(WidthT * sizeof(float)) VecReg {
 };

 static_assert(std::alignment_of<VecReg<16>>::value == 64,
-              "Unexepected alignment");
+              "Unexpected alignment");
 static_assert(std::alignment_of<VecReg<8>>::value == 32,
-              "Unexepected alignment");
+              "Unexpected alignment");
 static_assert(std::alignment_of<VecReg<4>>::value == 16,
-              "Unexepected alignment");
+              "Unexpected alignment");
 static_assert(std::alignment_of<VecReg<16>>::value == VecReg<16>::alignment,
-              "Unexepected alignment");
+              "Unexpected alignment");
 static_assert(std::alignment_of<VecReg<8>>::value == VecReg<8>::alignment,
-              "Unexepected alignment");
+              "Unexpected alignment");
 static_assert(std::alignment_of<VecReg<4>>::value == VecReg<4>::alignment,
-              "Unexepected alignment");
+              "Unexpected alignment");


 template<typename BuiltinT, int WidthT>
@@ -1535,7 +1535,7 @@ assume_aligned(Block<DataT, WidthT>* block_ptr)
 {
    static_assert(std::alignment_of<Block<DataT, WidthT>>::value
                      == std::alignment_of<VecReg<WidthT>>::value,
-                  "Unexepected alignment");
+                  "Unexpected alignment");
    return assume_aligned<VecReg<WidthT>::alignment>(block_ptr);
 }

@@ -1545,7 +1545,7 @@ assume_aligned(const Block<DataT, WidthT>* block_ptr)
 {
    static_assert(std::alignment_of<Block<DataT, WidthT>>::value
                      == std::alignment_of<VecReg<WidthT>>::value,
-                  "Unexepected alignment");
+                  "Unexpected alignment");
    return assume_aligned<VecReg<WidthT>::alignment>(block_ptr);
 }

@@ -2089,7 +2089,7 @@ struct UniformAsWide : public pvt::UniformAsWideImpl<DataT, WidthT> {
 template<typename DataT>
 struct WideTraits;  // undefined, all types used should be specialized
 //{
-//static bool mathes(const TypeDesc &) { return false; }
+//static bool matches(const TypeDesc &) { return false; }
 //};

 template<> struct WideTraits<float> {
@@ -2128,7 +2128,7 @@ template<> struct WideTraits<ustring> {
    }
 };

-// We let Vec3 match any vector semantics as we don't have a seperate Point or Normal classes
+// We let Vec3 match any vector semantics as we don't have a separate Point or Normal classes
 template<> struct WideTraits<Vec3> {
    static bool matches(const TypeDesc& type_desc)
    {
@@ -2229,9 +2229,15 @@ template<typename DataT, int WidthT> struct MaskedLaneProxy {
    // visibility to end user whose IDE
    // might display these methods vs. free
    // functions
-    OSL_FORCEINLINE bool is_on() const { return m_mask.is_on(m_lane); }
+    OSL_FORCEINLINE bool is_on() const
+    {
+        return m_mask.is_on(m_lane);
+    }

-    OSL_FORCEINLINE bool is_off() const { return m_mask.is_off(m_lane); }
+    OSL_FORCEINLINE bool is_off() const
+    {
+        return m_mask.is_off(m_lane);
+    }

 private:
    Block<DataT, WidthT>& m_ref_wide_data;
@@ -2293,9 +2299,15 @@ struct MaskedArrayLaneProxy {
    // visibility to end user whose IDE
    // might display these methods vs. free
    // functions
-    OSL_FORCEINLINE bool is_on() const { return m_mask.is_on(m_lane); }
+    OSL_FORCEINLINE bool is_on() const
+    {
+        return m_mask.is_on(m_lane);
+    }

-    OSL_FORCEINLINE bool is_off() const { return m_mask.is_off(m_lane); }
+    OSL_FORCEINLINE bool is_off() const
+    {
+        return m_mask.is_off(m_lane);
+    }

    OSL_FORCEINLINE MaskedLaneProxy<DataT, WidthT>
    operator[](int array_index) const
@@ -2727,19 +2739,24 @@ foreach_unique(Wide<DataT, WidthT> wdata, Mask<WidthT> data_mask, FunctorT f)
    do {
        ActiveLane lead_lane(remaining_mask.first_on());
        DataT lead_data = wdata[lead_lane];
-        Mask<WidthT> matching_lanes(false);
+        // Use int instead of Mask<> to allow reduction clause in openmp simd declaration
+        int matching_lanes_bits { 0 };
        OSL_FORCEINLINE_BLOCK
        {
-            OSL_OMP_PRAGMA(omp simd simdlen(WidthT))
+            OSL_OMP_PRAGMA(omp simd simdlen(WidthT)
+                               reduction(|
+                                         : matching_lanes_bits))
            for (int lane = 0; lane < WidthT; ++lane) {
                // NOTE: the comparison ignores the remaining_mask
                bool lane_matches = (lead_data == wdata[lane]);
                // NOTE: using bitwise & to avoid branches
                if (lane_matches & remaining_mask[lane]) {
-                    matching_lanes.set_on(lane);
+                    // inline of Mask::set_on(lane)
+                    matching_lanes_bits |= 1 << lane;
                }
            }
        }
+        Mask<WidthT> matching_lanes { matching_lanes_bits };

        f(lead_data, matching_lanes);
        remaining_mask &= ~matching_lanes;
@@ -19,7 +19,7 @@ namespace __OSL_WIDE_PVT {

 OSL_USING_DATA_WIDTH(__OSL_WIDTH)

-// Foward declaration, implementation is in liboslnoise/wide_gabornoise.h
+// Forward declaration, implementation is in liboslnoise/wide_gabornoise.h
 struct DisabledFilterPolicy {
    static constexpr bool active = false;
 };
@@ -34,7 +34,7 @@ OSL_NOINLINE void
 wide_gabor(Masked<Dual2<float>> wResult, Wide<const Dual2<float>> wX,
           NoiseParams const* opt, Block<Vec3>* opt_varying_direction);

-// Foward declaration, implementation is in wide_gabor.h
+// Forward declaration, implementation is in wide_gabor.h
 template<int AnisotropicT, typename FilterPolicyT>
 OSL_NOINLINE void
 wide_gabor(Masked<Dual2<float>> wResult, Wide<const Dual2<float>> wX,
@@ -46,20 +46,20 @@ OSL_NOINLINE void
 wide_gabor(Masked<Dual2<float>> wResult, Wide<const Dual2<Vec3>> wP,
           NoiseParams const* opt, Block<Vec3>* opt_varying_direction);

-// Foward declaration, implementation is in wide_gabor.h
+// Forward declaration, implementation is in wide_gabor.h
 template<int AnisotropicT, typename FilterPolicyT>
 OSL_NOINLINE void
 wide_gabor3(Masked<Dual2<Vec3>> wResult, Wide<const Dual2<float>> wX,
            NoiseParams const* opt, Block<Vec3>* opt_varying_direction);

-// Foward declaration, implementation is in wide_gabor.h
+// Forward declaration, implementation is in wide_gabor.h
 template<int AnisotropicT, typename FilterPolicyT>
 OSL_NOINLINE void
 wide_gabor3(Masked<Dual2<Vec3>> wResult, Wide<const Dual2<float>> wX,
            Wide<const Dual2<float>> wY, NoiseParams const* opt,
            Block<Vec3>* opt_varying_direction);

-// Foward declaration, implementation is in wide_gabor.h
+// Forward declaration, implementation is in wide_gabor.h
 template<int AnisotropicT, typename FilterPolicyT>
 OSL_NOINLINE void
 wide_gabor3(Masked<Dual2<Vec3>> wResult, Wide<const Dual2<Vec3>> wP,
@@ -71,14 +71,14 @@ wide_gabor3(Masked<Dual2<Vec3>> wResult, Wide<const Dual2<Vec3>> wP,
 // Periodic gabor(s)
 //////////////////////////////////////////////////////////////

-// Foward declaration, implementation is in wide_gabor.h
+// Forward declaration, implementation is in wide_gabor.h
 template<int AnisotropicT, typename FilterPolicyT>
 OSL_NOINLINE void
 wide_pgabor(Masked<Dual2<float>> wResult, Wide<const Dual2<float>> wX,
            Wide<const float> wXp, NoiseParams const* opt,
            Block<Vec3>* opt_varying_direction);

-// Foward declaration, implementation is in wide_gabor.h
+// Forward declaration, implementation is in wide_gabor.h
 template<int AnisotropicT, typename FilterPolicyT>
 OSL_NOINLINE void
 wide_pgabor(Masked<Dual2<float>> wResult, Wide<const Dual2<float>> wX,
@@ -86,7 +86,7 @@ wide_pgabor(Masked<Dual2<float>> wResult, Wide<const Dual2<float>> wX,
            Wide<const float> wYp, NoiseParams const* opt,
            Block<Vec3>* opt_varying_direction);

-// Foward declaration, implementation is in wide_gabor.h
+// Forward declaration, implementation is in wide_gabor.h
 template<int AnisotropicT, typename FilterPolicyT>
 OSL_NOINLINE void
 wide_pgabor(Masked<Dual2<float>> wResult, Wide<const Dual2<Vec3>> wP,
@@ -95,14 +95,14 @@ wide_pgabor(Masked<Dual2<float>> wResult, Wide<const Dual2<Vec3>> wP,



-// Foward declaration, implementation is in wide_gabor.h
+// Forward declaration, implementation is in wide_gabor.h
 template<int AnisotropicT, typename FilterPolicyT>
 OSL_NOINLINE void
 wide_pgabor3(Masked<Dual2<Vec3>> wResult, Wide<const Dual2<float>> wX,
             Wide<const float> wXp, NoiseParams const* opt,
             Block<Vec3>* opt_varying_direction);

-// Foward declaration, implementation is in wide_gabor.h
+// Forward declaration, implementation is in wide_gabor.h
 template<int AnisotropicT, typename FilterPolicyT>
 OSL_NOINLINE void
 wide_pgabor3(Masked<Dual2<Vec3>> wResult, Wide<const Dual2<float>> wX,
@@ -110,7 +110,7 @@ wide_pgabor3(Masked<Dual2<Vec3>> wResult, Wide<const Dual2<float>> wX,
             Wide<const float> wYp, NoiseParams const* opt,
             Block<Vec3>* opt_varying_direction);

-// Foward declaration, implementation is in wide_gabor.h
+// Forward declaration, implementation is in wide_gabor.h
 template<int AnisotropicT, typename FilterPolicyT>
 OSL_NOINLINE void
 wide_pgabor3(Masked<Dual2<Vec3>> wResult, Wide<const Dual2<Vec3>> wP,
@@ -5,6 +5,7 @@

 #pragma once

+#include <OSL/oslconfig.h>
 #include <OSL/oslversion.h>

 #ifdef OSL_USE_OPTIX
@@ -13,14 +14,18 @@
 #    ifdef _WIN32
 #        define NOMINMAX
 #    endif
-#    if (OPTIX_VERSION < 70000)
-#        include <optix_world.h>
-#    endif
 #else
 #    include <stdlib.h>
 #endif


+#if !defined(OSL_USE_OPTIX) && !defined(__CUDA_ARCH__)
+using CUdeviceptr = void*;
+using float3      = OSL::Vec3;
+#endif
+
+
+
 OSL_NAMESPACE_ENTER

 #ifdef OSL_USE_OPTIX
@@ -31,14 +36,12 @@ OSL_NAMESPACE_ENTER


 // TODO clean this up once OptiX6 support is dropped
-#    if (OPTIX_VERSION < 70000)
-namespace optix = ::optix;
-#    else
 namespace optix {
 typedef OptixDeviceContext Context;
 typedef cudaTextureObject_t TextureSampler;
+using ::CUdeviceptr;
+using ::float3;
 }  // namespace optix
-#    endif

 using ::cudaFree;
 using ::cudaMalloc;
@@ -49,13 +52,19 @@ using ::cudaMalloc;
 ////////////////////////////////////////////////////////////////////////
 // If OptiX is not available, make reasonable proxies just so that we
 // don't have to litter the code with quite as many #ifdef's.
-// OSL::optix::
+// Mostly under OSL::optix::
+
+

 namespace optix {

 typedef void* Context;
 typedef void* Program;
 typedef void* TextureSampler;
+typedef void* CUdeviceptr;
+using float3 = OSL::Vec3;
+
+
 struct Exception {
    static const char* what() { return "OSL compiled without Optix."; }
 };
@@ -558,7 +558,7 @@ public:
    ///
    void scope(int s) { m_scope = s; }

-    /// Return teh AST node containing the declaration of this symbol.
+    /// Return the AST node containing the declaration of this symbol.
    /// Use with care!
    ASTNode* node() const { return m_node; }

@@ -633,55 +633,133 @@ public:
    }


-    void dataoffset(int d) { m_dataoffset = d; }
-    int dataoffset() const { return m_dataoffset; }
+    void dataoffset(int d)
+    {
+        m_dataoffset = d;
+    }
+    int dataoffset() const
+    {
+        return m_dataoffset;
+    }

-    void wide_dataoffset(int d) { m_wide_dataoffset = d; }
-    int wide_dataoffset() const { return m_wide_dataoffset; }
+    void wide_dataoffset(int d)
+    {
+        m_wide_dataoffset = d;
+    }
+    int wide_dataoffset() const
+    {
+        return m_wide_dataoffset;
+    }

-    SymArena arena() const { return static_cast<SymArena>(m_arena); }
+    SymArena arena() const
+    {
+        return static_cast<SymArena>(m_arena);
+    }

-    void initializers(int d) { m_initializers = d; }
-    int initializers() const { return m_initializers; }
+    void initializers(int d)
+    {
+        m_initializers = d;
+    }
+    int initializers() const
+    {
+        return m_initializers;
+    }

-    bool has_derivs() const { return m_has_derivs; }
-    void has_derivs(bool new_derivs) { m_has_derivs = new_derivs; }
-    int size() const { return m_size; }
-    void size(size_t newsize) { m_size = (int)newsize; }
+    bool has_derivs() const
+    {
+        return m_has_derivs;
+    }
+    void has_derivs(bool new_derivs)
+    {
+        m_has_derivs = new_derivs;
+    }
+    int size() const
+    {
+        return m_size;
+    }
+    void size(size_t newsize)
+    {
+        m_size = (int)newsize;
+    }

    /// Return the size for each point, including derivs.
    ///
-    int derivsize() const { return m_has_derivs ? 3 * m_size : m_size; }
+    int derivsize() const
+    {
+        return m_has_derivs ? 3 * m_size : m_size;
+    }

-    bool connected() const { return valuesource() == ConnectedVal; }
-    bool connected_down() const { return m_connected_down; }
-    void connected_down(bool c) { m_connected_down = c; }
+    bool connected() const
+    {
+        return valuesource() == ConnectedVal;
+    }
+    bool connected_down() const
+    {
+        return m_connected_down;
+    }
+    void connected_down(bool c)
+    {
+        m_connected_down = c;
+    }

    /// Where did the symbol's value come from?
    ///
    enum ValueSource { DefaultVal, InstanceVal, GeomVal, ConnectedVal };

-    ValueSource valuesource() const { return (ValueSource)m_valuesource; }
-    void valuesource(ValueSource v) { m_valuesource = v; }
+    ValueSource valuesource() const
+    {
+        return (ValueSource)m_valuesource;
+    }
+    void valuesource(ValueSource v)
+    {
+        m_valuesource = v;
+    }
    const char* valuesourcename() const;
    static const char* valuesourcename(ValueSource v);

-    int fieldid() const { return m_fieldid; }
-    void fieldid(int id) { m_fieldid = id; }
+    int fieldid() const
+    {
+        return m_fieldid;
+    }
+    void fieldid(int id)
+    {
+        m_fieldid = id;
+    }

-    int layer() const { return m_layer; }
-    void layer(int id) { m_layer = id; }
+    int layer() const
+    {
+        return m_layer;
+    }
+    void layer(int id)
+    {
+        m_layer = id;
+    }

-    int initbegin() const { return m_initbegin; }
-    void initbegin(int i) { m_initbegin = i; }
-    int initend() const { return m_initend; }
-    void initend(int i) { m_initend = i; }
+    int initbegin() const
+    {
+        return m_initbegin;
+    }
+    void initbegin(int i)
+    {
+        m_initbegin = i;
+    }
+    int initend() const
+    {
+        return m_initend;
+    }
+    void initend(int i)
+    {
+        m_initend = i;
+    }
    void set_initrange(int b = 0, int e = 0)
    {
        m_initbegin = b;
        m_initend   = e;
    }
-    bool has_init_ops() const { return m_initbegin != m_initend; }
+    bool has_init_ops() const
+    {
+        return m_initbegin != m_initend;
+    }

    /// Clear read/write usage info.
    ///
@@ -725,15 +803,42 @@ public:
        }
    }

-    int firstread() const { return m_firstread; }
-    int lastread() const { return m_lastread; }
-    int firstwrite() const { return m_firstwrite; }
-    int lastwrite() const { return m_lastwrite; }
-    int firstuse() const { return std::min(firstread(), firstwrite()); }
-    int lastuse() const { return std::max(lastread(), lastwrite()); }
-    bool everread() const { return lastread() >= 0; }
-    bool everwritten() const { return lastwrite() >= 0; }
-    bool everused() const { return everread() || everwritten(); }
+    int firstread() const
+    {
+        return m_firstread;
+    }
+    int lastread() const
+    {
+        return m_lastread;
+    }
+    int firstwrite() const
+    {
+        return m_firstwrite;
+    }
+    int lastwrite() const
+    {
+        return m_lastwrite;
+    }
+    int firstuse() const
+    {
+        return std::min(firstread(), firstwrite());
+    }
+    int lastuse() const
+    {
+        return std::max(lastread(), lastwrite());
+    }
+    bool everread() const
+    {
+        return lastread() >= 0;
+    }
+    bool everwritten() const
+    {
+        return lastwrite() >= 0;
+    }
+    bool everused() const
+    {
+        return everread() || everwritten();
+    }
    // everused_in_group is an even more stringent test -- not only must
    // the symbol not be used within the shader but it also must not be
    // used elsewhere in the group, by being connected to something downstream
@@ -754,31 +859,67 @@ public:
        m_lastwrite  = last;
    }

-    bool initialized() const { return m_initialized; }
-    void initialized(bool init) { m_initialized = init; }
+    bool initialized() const
+    {
+        return m_initialized;
+    }
+    void initialized(bool init)
+    {
+        m_initialized = init;
+    }

-    bool lockgeom() const { return m_lockgeom; }
-    void lockgeom(bool lock) { m_lockgeom = lock; }
+    bool lockgeom() const
+    {
+        return m_lockgeom;
+    }
+    void lockgeom(bool lock)
+    {
+        m_lockgeom = lock;
+    }

-    bool allowconnect() const { return m_allowconnect; }
-    void allowconnect(bool val) { m_allowconnect = val; }
+    bool allowconnect() const
+    {
+        return m_allowconnect;
+    }
+    void allowconnect(bool val)
+    {
+        m_allowconnect = val;
+    }

-    int arraylen() const { return m_typespec.arraylength(); }
+    int arraylen() const
+    {
+        return m_typespec.arraylength();
+    }
    void arraylen(int len)
    {
        m_typespec.make_array(len);
        m_size = m_typespec.simpletype().size();
    }

-    bool renderer_output() const { return m_renderer_output; }
-    void renderer_output(bool v) { m_renderer_output = v; }
+    bool renderer_output() const
+    {
+        return m_renderer_output;
+    }
+    void renderer_output(bool v)
+    {
+        m_renderer_output = v;
+    }

    // When not uniform a symbol will have a varying value under batched
    // execution and must use a Wide data type to hold different values
    // for each data lane executing
-    bool is_uniform() const { return m_is_uniform; }
-    bool is_varying() const { return (m_is_uniform == 0); }
-    void make_varying() { m_is_uniform = false; }
+    bool is_uniform() const
+    {
+        return m_is_uniform;
+    }
+    bool is_varying() const
+    {
+        return (m_is_uniform == 0);
+    }
+    void make_varying()
+    {
+        m_is_uniform = false;
+    }

    // Results of a compare_op and other ops with logically boolean
    // results under certain conditions could be forced to be represented
@@ -792,14 +933,32 @@ public:
    // The value of forced_llvm_bool() is currently only respected during
    // batched execution.  Forced bools should not be coalesced with regular
    // ints, only other forced bools.
-    bool forced_llvm_bool() const { return m_forced_llvm_bool; }
-    void forced_llvm_bool(bool v) { m_forced_llvm_bool = v; }
+    bool forced_llvm_bool() const
+    {
+        return m_forced_llvm_bool;
+    }
+    void forced_llvm_bool(bool v)
+    {
+        m_forced_llvm_bool = v;
+    }

-    bool readonly() const { return m_readonly; }
-    void readonly(bool v) { m_readonly = v; }
+    bool readonly() const
+    {
+        return m_readonly;
+    }
+    void readonly(bool v)
+    {
+        m_readonly = v;
+    }

-    bool is_constant() const { return symtype() == SymTypeConst; }
-    bool is_temp() const { return symtype() == SymTypeTemp; }
+    bool is_constant() const
+    {
+        return symtype() == SymTypeConst;
+    }
+    bool is_temp() const
+    {
+        return symtype() == SymTypeTemp;
+    }

    // Retrieve the const float value (must be a const float!)
    float get_float(int index = 0) const
@@ -855,7 +1014,7 @@ public:

    // Stream output. Note that print/print_vals assume that any string
    // values are "raw" and they will be converted to C source code "escaped
-    // string" notation for printing. For example, a newline characer will
+    // string" notation for printing. For example, a newline character will
    // be rendered into the stream as the two character sequence '\n'.
    std::ostream& print(std::ostream& out, int maxvals = 100000000) const;
    std::ostream& print_vals(std::ostream& out, int maxvals = 100000000) const;
@@ -35,7 +35,7 @@ std::atomic<int> node_counts_peak[ASTNode::_last_node];
 class ScopeExit {
 public:
    typedef std::function<void()> Task;
-    explicit ScopeExit(Task&& task) : m_task(std::forward<Task>(task)) {}
+    explicit ScopeExit(Task&& task) : m_task(std::move(task)) {}
    ~ScopeExit() { m_task(); }

 private:
@@ -578,8 +578,14 @@ ASTvariable_declaration::ASTvariable_declaration(OSLCompilerImpl* comp,
    if (symtype == SymTypeLocal && Strutil::starts_with(name, "__debug_tmp__"))
        symtype = SymTypeTemp;
    m_sym = new Symbol(name, type, symtype, this);
-    if (!m_ismetadata)
+    if (m_ismetadata) {
+        // Metadata doesn't go in the symbol table, so we need to retain
+        // an owning pointer so it doesn't leak.
+        m_sym_owned.reset(m_sym);
+    } else {
+        // Put in the symbol table, after which symtab is the owner
        m_compiler->symtab().insert(m_sym);
+    }

    // A struct really makes several subvariables
    if (type.is_structure() || type.is_structure_array()) {
@@ -375,7 +375,7 @@ protected:
    void codegen_initlist(ref init, TypeSpec type, Symbol* sym);

    // Special code generation for structure initializers.
-    // It's in the ASTNode base class because it's used from mutiple
+    // It's in the ASTNode base class because it's used from multiple
    // subclasses.
    Symbol* codegen_struct_initializers(ref init, Symbol* sym,
                                        bool is_constructor = false,
@@ -520,7 +520,7 @@ public:

 private:
    ustring m_name;     ///< Name of the symbol (unmangled)
-    Symbol* m_sym;      ///< Ptr to the symbol this declares
+    Symbol* m_sym;      ///< Ptr to the symbol this declares (non-owning)
    bool m_isparam;     ///< Is this a parameter?
    bool m_isoutput;    ///< Is this an output parameter?
    bool m_ismetadata;  ///< Is this declaration a piece of metadata?
@@ -529,6 +529,7 @@ private:
    // specific initializers.
    typedef std::pair<ustring, ASTNode*> NamedInit;
    std::vector<NamedInit> m_struct_field_inits;
+    std::unique_ptr<Symbol> m_sym_owned;  // if we own the symbol
 };


@@ -949,7 +950,7 @@ private:
    /// return false and call an appropriate error().
    bool typecheck_printf_args(const char* format, ASTNode* arg);

-    /// Handle "funcion calls" that are really struct ctrs.
+    /// Handle "function calls" that are really struct ctrs.
    TypeSpec typecheck_struct_constructor();

    /// Is the argument number 'arg' read by the op?
@@ -918,7 +918,7 @@ ASTNode::codegen_initlist(ref init, TypeSpec type, Symbol* sym)
    }

    if (paraminit) {
-        // Warn early about struct array paramters.
+        // Warn early about struct array parameters.
        // Handling this will likely need changes to oso format.
        if (type.is_structure_array()) {
            errorfmt("array of struct are not allowed as parameters");
@@ -1305,7 +1305,7 @@ OSLCompilerImpl::track_variable_dependencies()
    // This naivite will never miss a dependency, but it may
    // overestimate dependencies.  (Hence we call this "conservative"
    // rather than "wrong.")  We deem this acceptable for now, since
-    // it's so much easer to implement the conservative dependency
+    // it's so much easier to implement the conservative dependency
    // analysis, and it's not yet clear that getting it closer to
    // optimal will have any performance impact on final shaders. Also
    // because this is probably no worse than the "dependency slop" that
@@ -490,7 +490,7 @@ private:
    int m_main_method_start;     ///< Instruction where 'main' starts
    bool m_declaring_shader_formals;  ///< Are we declaring shader formals?
    std::set<std::pair<ustring, int>> m_nowarn_lines;  ///< Lines for 'nowarn'
-    typedef std::unordered_map<ustring, std::string, ustringHash> FCMap;
+    typedef std::unordered_map<ustring, std::string> FCMap;
    FCMap m_filecontents_map;   ///< Contents of source files we've seen
    ustring m_last_sourcefile;  ///< Last filename for retrieve_source
    std::string* m_last_filecontents = nullptr;  //< Last file contents
@@ -28,6 +28,10 @@
 #pragma clang diagnostic ignored "-Wparentheses-equality"
 #endif

+#if (OSL_CLANG_VERSION >= 150000) || (OSL_INTEL_CLANG_VERSION >= 140000)
+#pragma clang diagnostic ignored "-Wunused-but-set-variable"
+#endif
+
 #if defined(__GNUC__) || defined(__clang__)
 #pragma GCC visibility push(hidden)
 #endif
@@ -186,7 +186,7 @@ private:
 ///
 class SymbolTable {
 public:
-    typedef std::unordered_map<ustring, Symbol*, ustringHash> ScopeTable;
+    typedef std::unordered_map<ustring, Symbol*> ScopeTable;
    typedef std::vector<ScopeTable> ScopeTableStack;
    typedef SymbolPtrVec::iterator iterator;
    typedef SymbolPtrVec::const_iterator const_iterator;
@@ -478,7 +478,7 @@ ASTbinary_expression::typecheck(TypeSpec expected)
            TypeSpec returntype = m_compiler->type_from_code(code, &advance);
            code += advance;
            if (code[0] && check_simple_arg(l, code, true) && code[0]
-                && check_simple_arg(r, code, true) && !code[0]) {
+                && check_simple_arg(r, code, true) && !code[0]) {  // NOSONAR
                return m_typespec = returntype;
            }
        }
@@ -618,7 +618,7 @@ ASTternary_expression::typecheck(TypeSpec expected)
        errorfmt("Cannot use an array as a condition");

    // No arrays
-    if (t.is_array() || t.is_array()) {
+    if (t.is_array() || f.is_array()) {
        errorfmt("Not allowed: '{} ? {} : {}'", c, t, f);
        return TypeSpec();
    }
@@ -678,7 +678,7 @@ ASTtype_constructor::typecheck(TypeSpec expected, bool report, bool bind)
    if (expected.is_float()) {
        patterns    = float_patterns;
        argexpected = TypeDesc::FLOAT;
-        // ^^^ Since simetimes tht constructor `float(expr)` is used as a
+        // ^^^ Since sometimes the constructor `float(expr)` is used as a
        // synonym for a cast `(float)expr`, we know that ambiguously typed
        // expressions should favor disambiguating to a float.
    } else if (expected.is_triple()) {
@@ -762,7 +762,7 @@ public:

    ~TypeAdjuster()
    {
-        // Commit infered types of all ASTcompound_initializers scanned.
+        // Commit inferred types of all ASTcompound_initializers scanned.
        if (m_success) {
            for (auto&& initer : m_adjust) {
                ASTcompound_initializer* ciptr = std::get<0>(initer);
@@ -992,9 +992,9 @@ private:
                   const StructSpec::FieldSpec* field = nullptr)
    {
        if (node->typecheck(expected) != expected &&
-            // Alllow assignment with comparable type
+            // Allow assignment with comparable type
            !assignable(expected, node->typespec()) &&
-            // Alllow closure assignments to '0'
+            // Allow closure assignments to '0'
            !(expected.is_closure() && !node->typespec().is_closure()
              && node->typespec().is_int_or_float()
              && node->nodetype() == literal_node
@@ -1065,6 +1065,8 @@ bool
 ASTNode::check_arglist(const char* /*funcname*/, ASTNode::ref arg,
                       const char* formals, bool coerce, bool bind)
 {
+    if (!formals)
+        return false;
    // std::cerr << "ca " << funcname << " formals='" << formals << "\n";
    for (; arg; arg = arg->next()) {
        if (!*formals)  // More formal args, but no more actual args
@@ -1619,7 +1621,7 @@ class CandidateFunctions {

        default:
            // TODO: Scoring default function arguments would go here
-            // Curently an unused formal argument, so no match at all.
+            // Currently an unused formal argument, so no match at all.
            return kNoMatch;
        }
        OSL_DASSERT(*formals == 0);
@@ -1994,7 +1996,7 @@ ASTfunction_call::typecheck(TypeSpec expected)
            any_args_are_compound_initializers = true;
    }

-    // Save the currently choosen symbol for error reporting later
+    // Save the currently chosen symbol for error reporting later
    FunctionSymbol* poly = func();

    CandidateFunctions candidates(m_compiler, expected, args(), poly);
@@ -209,7 +209,7 @@ macro ( CUDA_SHADEOPS_COMPILE srclist )
        DEPENDS ${shadeops_bc_list} ${exec_headers} ${PROJECT_PUBLIC_HEADERS} ${shadeops_srcs}
        WORKING_DIRECTORY "${CMAKE_CURRENT_SOURCE_DIR}" )

-    # Serialize the linked bitcode into a CPP file and add it to the list of liboslexec soures
+    # Serialize the linked bitcode into a CPP file and add it to the list of liboslexec sources
    add_custom_command ( OUTPUT ${shadeops_bc_cuda_cpp}
        COMMAND ${Python_EXECUTABLE} "${CMAKE_SOURCE_DIR}/src/build-scripts/serialize-bc.py"
            ${linked_shadeops_bc} ${shadeops_bc_cuda_cpp} "osl_llvm_compiled_ops_cuda"
@@ -222,14 +222,14 @@ if (USE_LLVM_BITCODE)
    set (llvm_ops_srcs llvm_ops.cpp)
    EMBED_LLVM_BITCODE_IN_CPP ( "${llvm_ops_srcs}" "_host" "osl_llvm_compiled_ops" lib_src "")

-    set (rs_dependant_ops_srcs
+    set (rs_dependent_ops_srcs
         opmatrix.cpp
         )
    # Achieve the effect of absorbing osl_llvm_compiled_ops by adding its 
-    # sources to rs_dependant_ops_srcs which avoids having to do it at runtime.
-    list (APPEND rs_dependant_ops_srcs ${llvm_ops_srcs})
+    # sources to rs_dependent_ops_srcs which avoids having to do it at runtime.
+    list (APPEND rs_dependent_ops_srcs ${llvm_ops_srcs})

-    EMBED_LLVM_BITCODE_IN_CPP ( "${rs_dependant_ops_srcs}" "_host_rs" "osl_llvm_compiled_rs_dependant_ops" lib_src "")
+    EMBED_LLVM_BITCODE_IN_CPP ( "${rs_dependent_ops_srcs}" "_host_rs" "osl_llvm_compiled_rs_dependent_ops" lib_src "")

    # Optionally repeat the bitcode compilation with CUDA-specific options
    if (CUDA_FOUND)
@@ -352,7 +352,7 @@ foreach(batched_target ${BATCHED_TARGET_LIST})
    endif ()
    
    if (CMAKE_COMPILER_IS_CLANG OR CMAKE_COMPILER_IS_APPLECLANG)
-        # REQUIRES CLANG 7+ to use sucessfully vectorize with OpenMP simd
+        # REQUIRES CLANG 7+ to use successfully vectorize with OpenMP simd
        
        if (${TARGET_OPT_ISA} STREQUAL "AVX512")
            list (APPEND TARGET_CXX_OPTS "-march=skylake-avx512")
@@ -370,7 +370,13 @@ foreach(batched_target ${BATCHED_TARGET_LIST})
        endif ()
        
        if (${TARGET_OPT_FMA} STREQUAL "noFMA")
-            list (APPEND TARGET_CXX_OPTS "-mno-fma")
+            # Do NOT use "-mno-fma" as it may lower ISA to one without any FMA
+            # instructions (AVX512->AVX2).  Our intent is to not generate 
+            # FMA's but still use the available instructions and register 
+            # width. However math library calls may use still end up using 
+            # FMA's, but lets try turning of fused math ops because many OSL 
+            # math functions are from OIIO/IMath headers generate non FMA
+            list (APPEND TARGET_CXX_OPTS "-ffp-contract=off")
        endif ()
        
        # large SIMD function loops will exceed llvm's -inline-threshold default of 225.
@@ -388,7 +394,8 @@ foreach(batched_target ${BATCHED_TARGET_LIST})
        list (APPEND TARGET_CXX_OPTS "-fno-unroll-loops")

        if (CMAKE_COMPILER_IS_INTELCLANG)
-            list (APPEND TARGET_CXX_OPTS "-qopenmp-simd")
+            # utilize Intel compiler's optimized openmp simd generation vs. open source "-qopenmp-simd"
+            list (APPEND TARGET_CXX_OPTS "-fiopenmp-simd")
        else ()
            list (APPEND TARGET_CXX_OPTS "-fopenmp")
        endif ()
@@ -434,7 +441,13 @@ foreach(batched_target ${BATCHED_TARGET_LIST})
        endif ()
        
        if (${TARGET_OPT_FMA} STREQUAL "noFMA")
-            list (APPEND TARGET_CXX_OPTS "-mno-fma")
+            # Do NOT use "-mno-fma" as it may lower ISA to one without any FMA
+            # instructions (AVX512->AVX2).  Our intent is to not generate 
+            # FMA's but still use the available instructions and register 
+            # width. However math library calls may use still end up using 
+            # FMA's, but lets try turning of fused math ops because many OSL 
+            # math functions are from OIIO/IMath headers generate non FMA
+            list (APPEND TARGET_CXX_OPTS "-ffp-contract=off")
        endif ()
        
        list (APPEND TARGET_CXX_OPTS "-fopenmp-simd")
@@ -23,7 +23,7 @@ public:
    MyAov(const TestPath* test, int id)
    {
        // Init for the test case array. For each test case set a bool
-        // in m_expected marking wether this AOV should get some color
+        // in m_expected marking whether this AOV should get some color
        // from that test or not.
        for (int i = 0; test[i].path[0]; ++i) {  // iterate test cases
            m_expected.push_back(false);         // false by default
@@ -88,7 +88,7 @@ simulate(Accumulator& accum, const char** events, size_t testno)
 int
 main()
 {
-    // Some constants to avoid refering to AOV's by number
+    // Some constants to avoid referring to AOV's by number
    const int beauty       = 0;
    const int diffuse2_3   = 1;
    const int light3       = 2;
@@ -138,7 +138,7 @@ NdfAutomata::symbolsFrom(const IntSet& states, SymbolSet& out_symbols,
 {
    for (IntSet::const_iterator i = states.begin(); i != states.end(); ++i) {
        const State* state = m_states[*i];
-        // For every state we have to go thorugh all the symbols in the transition table
+        // For every state we have to go through all the symbols in the transition table
        // m_symbol_trans and add them to the output
        for (SymbolToIntList::const_iterator j = state->m_symbol_trans.begin();
             j != state->m_symbol_trans.end(); ++j)
@@ -4,12 +4,12 @@

 #pragma once

-#include <set>
-#include <map>
 #include <list>
-#include <vector>
+#include <map>
+#include <set>
 #include <unordered_map>
 #include <unordered_set>
+#include <vector>

 #include <OSL/oslconfig.h>

@@ -20,20 +20,21 @@ OSL_NAMESPACE_ENTER
 // General container for all symbol sets

 // General container for integer sets
-typedef std::set<int> IntSet; // probably faster to test for equality, unions and so
+typedef std::set<int>
+    IntSet;  // probably faster to test for equality, unions and so

-typedef std::unordered_set<ustring, ustringHash> SymbolSet;
+typedef std::unordered_set<ustring> SymbolSet;
 // This is for the transition table used in DfAutomata::State
-typedef std::unordered_map<ustring, int, ustringHash> SymbolToInt;
+typedef std::unordered_map<ustring, int> SymbolToInt;
 // And this is for the transition table in NdfAutomata which
 // has several movements for each symbol
-typedef std::unordered_map<ustring, IntSet, ustringHash> SymbolToIntList;
+typedef std::unordered_map<ustring, IntSet> SymbolToIntList;
 typedef std::unordered_map<int, int> HashIntInt;

 // For the rules in the deterministic states, we don't need a real set
 // cause when converting from the NDF automata we will never find the same
 // rule twice
-typedef std::vector<void *> RuleSet;
+typedef std::vector<void*> RuleSet;

 // The lambda symbol (empty word)
 extern ustring lambda;
@@ -45,9 +46,9 @@ extern ustring lambda;
 /// any symbol which is not listed in its black list (m_minus).
 struct Wildcard {
    Wildcard() {};
-    Wildcard(SymbolSet &minus):m_minus(minus) {};
+    Wildcard(SymbolSet& minus) : m_minus(minus) {};

-    bool matches(ustring symbol)const
+    bool matches(ustring symbol) const
    {
        // true if the given symbol is not in m_minus
        return (m_minus.find(symbol) == m_minus.end());
@@ -63,109 +64,110 @@ struct Wildcard {
 /// symbol and allowed lambda transitions too. Light path expressions
 /// build this automata easily. It is later converted to a DF automata.
 class NdfAutomata {
+public:
+    // Basic state for NDF automata
+    class State {
+        friend class NdfAutomata;
+
    public:
-
-        // Basic state for NDF automata
-        class State {
-            friend class NdfAutomata;
-            public:
-
-                State(int id)
-                {
-                    m_id = id;
-                    m_wildcard_trans = -1;
-                    m_wildcard = NULL;
-                    m_rule = NULL;
-                };
-                ~State() { delete m_wildcard; };
-
-                /// Get the set of state id's reachable by the given symbol
-                ///
-                /// It doesn't clean the given result set, so you can use this
-                /// function to accumulate states.
-                void getTransitions (ustring symbol, IntSet &out_states)const;
-
-                /// Get all the lambda transitions
-                ///
-                /// Returns the state collection as a pair of iterators [begin, end)
-                std::pair <IntSet::const_iterator, IntSet::const_iterator>
-                    getLambdaTransitions ()const;
-
-                /// Add a standar transition (also valid for lambda)
-                void addTransition (ustring symbol, State *state);
-                /// Add a wildcard transition
-                ///
-                /// Note that the state is going to take ownership of the wildcard
-                /// pointer. So it should be newly allocated on heap and you don't
-                /// have to delete ir
-                void addWildcardTransition(Wildcard *wildcard, State *state);
-
-                /// For final states, this sets the associated rule. Which
-                /// can be any kind of pointer
-                void setRule(void *rule) { m_rule = rule; };
-
-                void *getRule()const { return m_rule; };
-                int  getId()const { return m_id; };
-
-                /// For debuging purposes
-                std::string tostr()const;
-
-            protected:
-
-                // State id (i.e index in the states vector)
-                int m_id;
-                // Transitions by single symbols
-                SymbolToIntList m_symbol_trans;
-                // In case m_wildcard is not NULL this holds the destination
-                // of the wildcard transition
-                int m_wildcard_trans;
-                // Wildcard (NULL if no wildcard transition present)
-                Wildcard *m_wildcard;
-                // Associated rule for final states, NULL if not final
-                void *m_rule;
+        State(int id)
+        {
+            m_id             = id;
+            m_wildcard_trans = -1;
+            m_wildcard       = NULL;
+            m_rule           = NULL;
        };
+        ~State() { delete m_wildcard; };

-
-
-        NdfAutomata() { newState(); };
-        ~NdfAutomata();
-
-        const State *getInitial()const { return m_states[0]; };
-        State *getInitial() { return m_states[0]; };
-
-        /// Creates a new state qith a valid id
-        State *newState();
-
-        const State *getState(int i)const { return m_states[i]; };
-        size_t size()const { return m_states.size(); };
-
-        /// return all the symbols that lead to other states starting
-        /// in any of the states in the given set. Also if there are
-        /// wildcard transitions, give back a new wildcard that guarantees
-        /// that its matches match all the wildcards in the set.
-        /// So the symbols returned in out_symbols do NOT overlap those that
-        /// match the returned wildcard (if present). And the union of out_symbols
-        /// and those matched by the wildcard, are all the valid transitions from
-        /// the given state set
-        void symbolsFrom(const IntSet &states, SymbolSet &out_symbols, Wildcard *&wildcard)const;
-
-        /// Get the set of states that are reachable from the given state set using the given symbol
-        void transitionsFrom(const IntSet &states, ustring symbol, IntSet &out_states)const;
-        /// Get the set of states that are reachable from the given state set by wildcard
-        void wildcardTransitionsFrom(const IntSet &states, IntSet &out_states)const;
-
-        /// Perform a lambda closure of a state set
+        /// Get the set of state id's reachable by the given symbol
        ///
-        /// In other words, complete the given set so it includes all the additional
-        /// states that are reachable by the lambda symbol
-        void lambdaClosure(IntSet &states)const;
+        /// It doesn't clean the given result set, so you can use this
+        /// function to accumulate states.
+        void getTransitions(ustring symbol, IntSet& out_states) const;

-        /// for debuging purposes
-        std::string tostr()const;
+        /// Get all the lambda transitions
+        ///
+        /// Returns the state collection as a pair of iterators [begin, end)
+        std::pair<IntSet::const_iterator, IntSet::const_iterator>
+        getLambdaTransitions() const;
+
+        /// Add a standard transition (also valid for lambda)
+        void addTransition(ustring symbol, State* state);
+        /// Add a wildcard transition
+        ///
+        /// Note that the state is going to take ownership of the wildcard
+        /// pointer. So it should be newly allocated on heap and you don't
+        /// have to delete ir
+        void addWildcardTransition(Wildcard* wildcard, State* state);
+
+        /// For final states, this sets the associated rule. Which
+        /// can be any kind of pointer
+        void setRule(void* rule) { m_rule = rule; };
+
+        void* getRule() const { return m_rule; };
+        int getId() const { return m_id; };
+
+        /// For debugging purposes
+        std::string tostr() const;

    protected:
-        /// Vector of states
-        std::vector<State *> m_states;
+        // State id (i.e index in the states vector)
+        int m_id;
+        // Transitions by single symbols
+        SymbolToIntList m_symbol_trans;
+        // In case m_wildcard is not NULL this holds the destination
+        // of the wildcard transition
+        int m_wildcard_trans;
+        // Wildcard (NULL if no wildcard transition present)
+        Wildcard* m_wildcard;
+        // Associated rule for final states, NULL if not final
+        void* m_rule;
+    };
+
+
+
+    NdfAutomata() { newState(); };
+    ~NdfAutomata();
+
+    const State* getInitial() const { return m_states[0]; };
+    State* getInitial() { return m_states[0]; };
+
+    /// Creates a new state with a valid id
+    State* newState();
+
+    const State* getState(int i) const { return m_states[i]; };
+    size_t size() const { return m_states.size(); };
+
+    /// return all the symbols that lead to other states starting
+    /// in any of the states in the given set. Also if there are
+    /// wildcard transitions, give back a new wildcard that guarantees
+    /// that its matches match all the wildcards in the set.
+    /// So the symbols returned in out_symbols do NOT overlap those that
+    /// match the returned wildcard (if present). And the union of out_symbols
+    /// and those matched by the wildcard, are all the valid transitions from
+    /// the given state set
+    void symbolsFrom(const IntSet& states, SymbolSet& out_symbols,
+                     Wildcard*& wildcard) const;
+
+    /// Get the set of states that are reachable from the given state set using the given symbol
+    void transitionsFrom(const IntSet& states, ustring symbol,
+                         IntSet& out_states) const;
+    /// Get the set of states that are reachable from the given state set by wildcard
+    void wildcardTransitionsFrom(const IntSet& states,
+                                 IntSet& out_states) const;
+
+    /// Perform a lambda closure of a state set
+    ///
+    /// In other words, complete the given set so it includes all the additional
+    /// states that are reachable by the lambda symbol
+    void lambdaClosure(IntSet& states) const;
+
+    /// for debugging purposes
+    std::string tostr() const;
+
+protected:
+    /// Vector of states
+    std::vector<State*> m_states;
 };


@@ -177,7 +179,8 @@ class NdfAutomata {
 typedef std::vector<int> StateSetKey;

 // Compute the unique key for the given set of states
-void keyFromStateSet(const IntSet &states, StateSetKey &out_key);
+void
+keyFromStateSet(const IntSet& states, StateSetKey& out_key);



@@ -188,84 +191,83 @@ void keyFromStateSet(const IntSet &states, StateSetKey &out_key);
 /// time parsing of light path expressions
 class DfAutomata {
    friend class DfOptimizedAutomata;
+
+public:
+    /// Simple state for a deterministic automata
+    class State {
+        friend class DfAutomata;
+        friend class DfOptimizedAutomata;
+
    public:
-        /// Simple state for a deterministic automata
-        class State {
-            friend class DfAutomata;
-            friend class DfOptimizedAutomata;
-            public:
+        State(int id)
+        {
+            m_id             = id;
+            m_wildcard_trans = -1;
+        }

-                State(int id)
-                {
-                    m_id = id;
-                    m_wildcard_trans = -1;
-                }
+        /// Get the transition for a symbol
+        //
+        /// Returns -1 if no transitions with that symbol. That means the
+        /// symbol is not recognized at this point of the automata
+        int getTransition(ustring symbol) const;

-                /// Get the transition for a symbol
-                //
-                /// Returns -1 if no transitions with that symbol. That means the
-                /// symbol is not recognized at this point of the automata
-                int getTransition(ustring symbol)const;
+        // Same semantics as in NdfAutomata::State
+        void addTransition(ustring symbol, State* state);
+        // WARNING: this has an optimized representation and has to be called
+        // always AFTER all the normal transitions have been added
+        void addWildcardTransition(Wildcard* wildcard, State* state);

-                // Same semantics as in NdfAutomata::State
-                void addTransition(ustring symbol, State *state);
-                // WARNING: this has an optimized representation and has to be called
-                // always AFTER all the normal transitions have been added
-                void addWildcardTransition(Wildcard *wildcard, State *state);
-
-                /// Optimize transitions
-                ///
-                /// Sometimes the existing wildcard can be extended to wrap some of
-                /// the single symbol transitions. This function performs that optimization
-                void removeUselessTransitions();
-
-                void addRule(void *rule) { m_rules.push_back(rule); };
-                const RuleSet &getRules()const { return m_rules; };
-                int  getId()const { return m_id; };
-                std::string tostr()const;
-
-            protected:
-
-                int m_id;
-                // Only one transition per symbol here
-                SymbolToInt m_symbol_trans;
-                int m_wildcard_trans;
-                // A final state here might contain several final states
-                // from the original NDF automata. Therefore, we need a list
-                // of rules here
-                RuleSet m_rules;
-        };
-
-
-        DfAutomata() {};
-        ~DfAutomata();
-
-        State *newState();
-        const State *getState(int i)const { return m_states[i]; };
-        size_t size()const { return m_states.size(); };
-        // In case somebody wants to reuse the same object and recreate
-        // an automata, we provide this method
-        void clear();
-
-        /// Colapse all the equivalent states into single ones
-        void removeEquivalentStates();
-        /// Go through all the states and perform removeUselessTransitions
-        /// method call on them
+        /// Optimize transitions
+        ///
+        /// Sometimes the existing wildcard can be extended to wrap some of
+        /// the single symbol transitions. This function performs that optimization
        void removeUselessTransitions();

-        /// For debuging purposes
-        std::string tostr()const;
+        void addRule(void* rule) { m_rules.push_back(rule); };
+        const RuleSet& getRules() const { return m_rules; };
+        int getId() const { return m_id; };
+        std::string tostr() const;

    protected:
+        int m_id;
+        // Only one transition per symbol here
+        SymbolToInt m_symbol_trans;
+        int m_wildcard_trans;
+        // A final state here might contain several final states
+        // from the original NDF automata. Therefore, we need a list
+        // of rules here
+        RuleSet m_rules;
+    };

-        /// Return true if two states are equivalent
-        ///
-        /// Two states are equivalent if their transition tables lead
-        /// to the same states for the same symbols
-        bool equivalent(const State *dfstateA, const State *dfstateB);

-        // State vector with the automata
-        std::vector<State *> m_states;
+    DfAutomata() {};
+    ~DfAutomata();
+
+    State* newState();
+    const State* getState(int i) const { return m_states[i]; };
+    size_t size() const { return m_states.size(); };
+    // In case somebody wants to reuse the same object and recreate
+    // an automata, we provide this method
+    void clear();
+
+    /// Colapse all the equivalent states into single ones
+    void removeEquivalentStates();
+    /// Go through all the states and perform removeUselessTransitions
+    /// method call on them
+    void removeUselessTransitions();
+
+    /// For debugging purposes
+    std::string tostr() const;
+
+protected:
+    /// Return true if two states are equivalent
+    ///
+    /// Two states are equivalent if their transition tables lead
+    /// to the same states for the same symbols
+    bool equivalent(const State* dfstateA, const State* dfstateB);
+
+    // State vector with the automata
+    std::vector<State*> m_states;
 };


@@ -274,40 +276,42 @@ class DfAutomata {
 /// of state sets that have already been found to be new states of the
 /// deterministic automata
 class StateSetRecord {
-    public:
-        StateSetRecord(const NdfAutomata &ndfautomata, DfAutomata &dfautomata):
-            m_ndfautomata(ndfautomata), m_dfautomata(dfautomata)
-            {};
+public:
+    StateSetRecord(const NdfAutomata& ndfautomata, DfAutomata& dfautomata)
+        : m_ndfautomata(ndfautomata), m_dfautomata(dfautomata) {};

-        // A new found state is defined by the deterministic state created
-        // for it and the state(int) set in the original automata
-        typedef std::pair<DfAutomata::State *, IntSet> Discovery;
-        // The type that will index our new created states indexed by the set key
-        typedef std::map<StateSetKey, DfAutomata::State *> StateSetMap;
+    // A new found state is defined by the deterministic state created
+    // for it and the state(int) set in the original automata
+    typedef std::pair<DfAutomata::State*, IntSet> Discovery;
+    // The type that will index our new created states indexed by the set key
+    typedef std::map<StateSetKey, DfAutomata::State*> StateSetMap;

-        /// Take a state set and build a new df state (or return existing one)
-        /// Also, if it was newly created, append it to the discovered list so we
-        /// can iterate over it later
-        DfAutomata::State *ensureState(const IntSet &newstates, std::list<Discovery> &discovered);
+    /// Take a state set and build a new df state (or return existing one)
+    /// Also, if it was newly created, append it to the discovered list so we
+    /// can iterate over it later
+    DfAutomata::State* ensureState(const IntSet& newstates,
+                                   std::list<Discovery>& discovered);

-    private:
+private:
+    /// Gather all the rules from the original automata in the given sets (if any)
+    /// and put them in the dfstate rule set
+    void getRulesFromSet(DfAutomata::State* dfstate,
+                         const NdfAutomata& ndfautomata,
+                         const IntSet& ndfstates);

-        /// Gather all the rules from the original automata in the given sets (if any)
-        /// and put them in the dfstate rule set
-        void getRulesFromSet(DfAutomata::State *dfstate, const NdfAutomata &ndfautomata, const IntSet &ndfstates);
-
-        const NdfAutomata &m_ndfautomata;
-        DfAutomata &m_dfautomata;
-        StateSetMap m_key_to_dfstate;
+    const NdfAutomata& m_ndfautomata;
+    DfAutomata& m_dfautomata;
+    StateSetMap m_key_to_dfstate;
 };



-/// NDF to DF automata convertion
+/// NDF to DF automata conversion
 ///
 /// This function is the most important pice of the whole process. It takes
 /// a non-deterministic finite automata and computes an equivalent deterministic
 /// one. It is equivalente in the sense that they  both recognize the same language
-void ndfautoToDfauto(const NdfAutomata &ndfautomata, DfAutomata &dfautomata);
+void
+ndfautoToDfauto(const NdfAutomata& ndfautomata, DfAutomata& dfautomata);

 OSL_NAMESPACE_EXIT
@@ -87,6 +87,12 @@ BackendLLVM::BackendLLVM(ShadingSystemImpl& shadingsys, ShaderGroup& group,
 #endif
    m_use_optix      = shadingsys.renderer()->supports("OptiX");
    m_use_rs_bitcode = !shadingsys.m_rs_bitcode.empty();
+    m_name_llvm_syms = shadingsys.m_llvm_output_bitcode;
+
+    // Select the appropriate ustring representation
+    ll.ustring_rep(m_use_optix ? LLVM_Util::UstringRep::hash
+                               : LLVM_Util::UstringRep::charptr);
+
    ll.dumpasm(shadingsys.m_llvm_dumpasm);
    ll.jit_fma(shadingsys.m_llvm_jit_fma);
    ll.jit_aggressive(shadingsys.m_llvm_jit_aggressive);
@@ -135,7 +141,7 @@ BackendLLVM::llvm_pass_type(const TypeSpec& typespec)
    else if (t == TypeDesc::INT)
        lt = ll.type_int();
    else if (t == TypeDesc::STRING)
-        lt = (llvm::Type*)ll.type_string();
+        lt = (llvm::Type*)ll.type_ustring();
    else if (t.aggregate == TypeDesc::VEC3)
        lt = (llvm::Type*)ll.type_void_ptr();  //llvm_type_triple_ptr();
    else if (t.aggregate == TypeDesc::MATRIX44)
@@ -230,6 +236,7 @@ static ustring fields[] = { ustring("P"),
                            ustring("tracedata"),
                            ustring("objdata"),
                            ustring("shadingcontext"),
+                            ustring("shadingStateUniform"),
                            ustring("renderer"),
                            ustring("object2common"),
                            ustring("shader2common"),
@@ -259,7 +266,8 @@ BackendLLVM::llvm_global_symbol_ptr(ustring name)
    // the ShaderGlobals struct.
    int sg_index = ShaderGlobalNameToIndex(name);
    OSL_ASSERT(sg_index >= 0);
-    return ll.void_ptr(ll.GEP(sg_ptr(), 0, sg_index));
+    return ll.void_ptr(ll.GEP(sg_ptr(), 0, sg_index),
+                       llnamefmt("glob_{}_voidptr", name));
 }


@@ -322,7 +330,7 @@ BackendLLVM::getOrAllocateLLVMSymbol(const Symbol& sym)

    if (map_iter == named_values().end()) {
        llvm::Value* a = llvm_alloca(sym.typespec(), sym.has_derivs(),
-                                     mangled_name);
+                                     llnamefmt("{}_mem", mangled_name));
        named_values()[mangled_name] = a;
        return a;
    }
@@ -332,8 +340,9 @@ BackendLLVM::getOrAllocateLLVMSymbol(const Symbol& sym)


 llvm::Value*
-BackendLLVM::addCUDAVariable(const std::string& name, int size, int alignment,
-                             const void* data, TypeDesc type)
+BackendLLVM::addCUDAGlobalVariable(const std::string& name, int size,
+                                   int alignment, const void* init_data,
+                                   TypeDesc type)
 {
    OSL_ASSERT(use_optix()
               && "This function is only supposed to be used with OptiX!");
@@ -341,40 +350,16 @@ BackendLLVM::addCUDAVariable(const std::string& name, int size, int alignment,
    llvm::Constant* constant = nullptr;

    if (type == TypeDesc::TypeFloat) {
-        constant = llvm::ConstantFP::get(llvm::Type::getFloatTy(
-                                             ll.module()->getContext()),
-                                         *(float*)data);
+        constant = ll.constant(*(const float*)init_data);
    } else if (type == TypeDesc::TypeInt) {
-        constant = llvm::ConstantInt::get(llvm::Type::getInt32Ty(
-                                              ll.module()->getContext()),
-                                          *(int*)data);
+        constant = ll.constant(*(const int*)init_data);
    } else if (type == TypeDesc::TypeString) {
-        // Register the string with the OptiX renderer. The renderer will add
-        // the string to a global table and create an OptiX variable to hold the
-        // char*.
-        shadingsys().renderer()->register_string(((ustring*)data)->string(),
-                                                 name);
-
-#if OPTIX_VERSION < 70000
-        // Leave the variable uninitialized to prevent raw pointers from
-        // appearing in the generated code. The OptiX renderer will set the
-        // variable to the string address before the kernel is launched.
-        constant            = llvm::ConstantInt::get(llvm::Type::getInt64Ty(
-                                                         ll.module()->getContext()),
-                                                     0);
-        m_varname_map[name] = ((ustring*)data)->string();
-#else
-        // DeviceStrings use the ustring hash to identify themselves.  These hashes will
-        // match on both the device and host side.
-        constant            = llvm::ConstantInt::get(llvm::Type::getInt64Ty(
-                                                         ll.module()->getContext()),
-                                                     ((ustring*)data)->hash());
-        m_varname_map[name] = ((ustring*)data)->string();
-#endif
-
+        constant = ll.constant(((const ustring*)init_data)->hash());
+        // N.B. Since this is the OptiX side specifically, we will represent
+        // strings as the ustringhash, so we know it as a constant.
    } else {
        // Handle unspecified types as generic byte arrays
-        llvm::ArrayRef<uint8_t> arr_ref((uint8_t*)data, size);
+        llvm::ArrayRef<uint8_t> arr_ref((uint8_t*)init_data, size);
        constant = llvm::ConstantDataArray::get(ll.module()->getContext(),
                                                arr_ref);
    }
@@ -390,9 +375,7 @@ BackendLLVM::addCUDAVariable(const std::string& name, int size, int alignment,
    g_var->setAlignment(alignment);
 #endif

-#if (OPTIX_VERSION < 70000)
-    g_var->setLinkage(llvm::GlobalValue::ExternalLinkage);
-#endif
+    g_var->setLinkage(llvm::GlobalValue::PrivateLinkage);
    g_var->setVisibility(llvm::GlobalValue::DefaultVisibility);
    g_var->setInitializer(constant);
    m_const_map[name] = g_var;
@@ -402,155 +385,32 @@ BackendLLVM::addCUDAVariable(const std::string& name, int size, int alignment,



-void
-BackendLLVM::createOptixMetadata(const std::string& name, const Symbol& sym)
-{
-    // Create additional variables with the semantic information needed by OptiX
-    // to access the global variable created above.
-    //
-    // There is no need to retain pointers to these variables, since they are not
-    // accessed during execution. They are only used internally by OptiX.
-    //
-    // Refer to the OptiX API documentation and optix_defines.h in the OptiX SDK
-    // for more information.
-
-    OSL_ASSERT(use_optix()
-               && "This function is only supported when using OptiX!");
-
-    auto mangle_name = [](const std::string& name, const std::string& prefix) {
-        return fmtformat("_ZN{}rti_internal_{}{}{}E", prefix.size() + 13,
-                         prefix, name.size(), name);
-    };
-
-    std::string optix_type;
-    const TypeDesc type = sym.typespec().simpletype();
-    if (!sym.typespec().is_array()) {
-        optix_type =
-            // Documented built-in types
-            (type == TypeDesc::TypeInt)      ? "int"
-            : (type == TypeDesc::TypeFloat)  ? "float"
-            : (type == TypeDesc::TypePoint)  ? "float3"
-            : (type == TypeDesc::TypeVector) ? "float3"
-            : (type == TypeDesc::TypeNormal) ? "float3"
-            : (type == TypeDesc::TypeColor)  ? "float3"
-            : (type == TypeDesc::TypeMatrix) ? "matrix"
-            : (type == TypeDesc::TypeString)
-                ? "uint64_t"
-                :
-                // Catch-all for types that fall through, if there are any.
-                type.c_str();
-
-        // NB: TypeMatrix is assumed to be 4x4 and will be treated by OptiX as a
-        //     user datatype (i.e., a generic struct).
-    } else {
-        // OptiX should handle int and float vectors between 2 and 4 dimensions
-        // with no problem. Larger arrays, or arrays of other element types,
-        // will be treated as a user datatype and will not work natively with
-        // OptiX's variable mechanism.
-        optix_type = fmtformat("{}{}", sym.typespec().elementtype(),
-                               sym.typespec().arraylength());
-    }
-
-    struct rti_typeinfo {
-        unsigned int kind = 0x796152;  // _OPTIX_VARIABLE
-        unsigned int size;
-    } type_info;
-    type_info.size = sym.size();
-
-    int type_enum = 0x1337;  // _OPTIX_TYPE_ENUM_UNKNOWN
-    char zero     = 0;
-
-    addCUDAVariable(mangle_name(name, "typeinfo"), 8, 4, &type_info);
-    addCUDAVariable(mangle_name(name, "typename"), optix_type.size() + 1, 16,
-                    optix_type.data());
-    addCUDAVariable(mangle_name(name, "typeenum"), 4, 4, &type_enum,
-                    TypeDesc::TypeInt);
-    addCUDAVariable(mangle_name(name, "semantic"), 1, 1, &zero);
-    addCUDAVariable(mangle_name(name, "annotation"), 1, 1, &zero);
-}
-
-
-
 llvm::Value*
-BackendLLVM::getOrAllocateCUDAVariable(const Symbol& sym, bool addMetadata)
+BackendLLVM::getOrAllocateCUDAVariable(const Symbol& sym)
 {
    OSL_ASSERT(use_optix()
               && "This function is only supported when using OptiX!");

-    std::ostringstream ss;
-    ss.imbue(std::locale::classic());  // force C locale
-    if (sym.typespec().is_string()) {
-        // Use the ustring hash to create a name for the symbol that's based on
-        // the string contents
-        //
-        // TODO: Collisions between variable names are unlikely, but still
-        //       possible. Using something like a counter to handle collisions
-        //       is unattractive because it depends on the order in which
-        //       strings are encountered at run time.
-        //
-        //       For now I am simply appending the length to the hash, but a
-        //       more robust solution may be called for.
+    // We need to sanitize the symbol name for PTX compatibility. Also, if the
+    // sym name starts with a dollar sign, which are not allowed in PTX
+    // variable names, then prepend another underscore.
+    std::string sym_name = Strutil::replace(sym.name(), ".", "_", true);

-        ss << "ds_"
-#if OPTIX_VERSION >= 70000
-           << group().name() << "_" << group().id() << "_"
-#endif
-           << std::setbase(16) << std::setfill('0') << std::setw(16)
-           << sym.get_string().hash() << "_" << std::setbase(16)
-           << std::setfill('0') << std::setw(4) << sym.get_string().length();
-
-        auto it                   = m_varname_map.find(ss.str());
-        const std::string old_str = (it != m_varname_map.end()) ? it->second
-                                                                : "";
-
-        if (old_str != "" && old_str != sym.get_string().string()) {
-            std::cerr << "Warning: variable name collision between " << old_str
-                      << " and " << sym.get_string() << std::endl;
-        }
-    } else {
-        // We need to sanitize the symbol name for PTX compatibility
-        std::string sym_name = sym.name().c_str();
-        std::replace(sym_name.begin(), sym_name.end(), '.', '_');
-
-        std::string var_name = fmtformat("{}_{}_{}_{}_{}", sym_name,
-                                         group().name(), group().id(),
-                                         inst()->layername(), sym.layer());
-
-        // Leading dollar signs are not allowed in PTX variable names,
-        // so prepend an underscore.
-        if (var_name[0] == '$') {
-            ss << '_';
-        }
-
-        ss << var_name;
-    }
-
-    const std::string name = ss.str();
+    std::string name
+        = fmtformat("{}{}_{}_{}_{}", sym_name.front() == '$' ? "_" : "",
+                    sym_name, group().name(), inst()->layername(), sym.layer());

    // Return the Value if it has already been allocated
-    std::map<std::string, llvm::GlobalVariable*>::iterator it
-        = get_const_map().find(name);
-
-    if (it != get_const_map().end()) {
+    auto it = get_const_map().find(name);
+    if (it != get_const_map().end())
        return it->second;
-    }
-
-#if (OPTIX_VERSION < 70000)
-    // Add the extra metadata needed to make the variable visible to OptiX.
-    if (addMetadata || sym.typespec().is_string())
-        createOptixMetadata(name, sym);
-#endif

    // TODO: Figure out the actual CUDA alignment requirements for the various
    //       OSL types. For now, be somewhat conservative and assume 8 for
    //       non-scalar types.
    int alignment = (sym.typespec().is_scalarnum()) ? 4 : 8;
-
-    llvm::Value* cuda_var = addCUDAVariable(name, sym.size(), alignment,
-                                            sym.data(),
-                                            sym.typespec().simpletype());
-
-    return cuda_var;
+    return addCUDAGlobalVariable(name, sym.size(), alignment, sym.data(),
+                                 sym.typespec().simpletype());
 }


@@ -568,38 +428,37 @@ BackendLLVM::llvm_get_pointer(const Symbol& sym, int deriv,

    llvm::Value* result = NULL;
    if (sym.symtype() == SymTypeConst) {
+        TypeSpec elemtype = sym.typespec().elementtype();
        if (use_optix()) {
            // Check the constant map for the named Symbol; if it's found, then
            // a GlobalVariable has been created for it
            llvm::Value* ptr = getOrAllocateCUDAVariable(sym);
            if (ptr) {
-                llvm::Type* cast_type
-                    = (!sym.typespec().is_string())
-                          ? ll.type_ptr(llvm_type(sym.typespec().elementtype()))
-                          : ll.type_void_ptr();
-
-                result = ll.ptr_cast(ptr, cast_type);
+                result = llvm_ptr_cast(ptr, llvm_typedesc(elemtype),
+                                       llnamefmt("cast_to_{}_", sym.typespec()));
+                // N.B. llvm_typedesc() for a string will know that on OptiX,
+                // strings are actually represented as just the hash.
            }
        } else {
-            // For constants, start with *OUR* pointer to the constant values.
-            result = ll.ptr_cast(ll.constant_ptr(sym.data()),
-                                 ll.type_ptr(
-                                     llvm_type(sym.typespec().elementtype())));
+            // For constants on the CPU side, start with *OUR* pointer to the
+            // constant values.
+            result = ll.constant_ptr(sym.data(),
+                                     ll.type_ptr(llvm_type(elemtype)));
        }
-
    } else {
-        // Start with the initial pointer to the variable's memory location
+        // If the symbol is not a SymTypeConst, then start with the initial
+        // pointer to the variable's memory location.
        result = getLLVMSymbolBase(sym);
    }
    if (!result)
        return NULL;  // Error

-    // If it's an array or we're dealing with derivatives, step to the
-    // right element.
+    // If it's an array or we're dealing with derivatives, step to the right
+    // element.
    TypeDesc t = sym.typespec().simpletype();
    if (t.arraylen || has_derivs) {
        int d = deriv * std::max(1, t.arraylen);
-        if (arrayindex)
+        if (arrayindex && d)
            arrayindex = ll.op_add(arrayindex, ll.constant(d));
        else
            arrayindex = ll.constant(d);
@@ -645,13 +504,14 @@ BackendLLVM::llvm_load_value(const Symbol& sym, int deriv,
            return ll.constant(sym.get_float(component));
        }
        if (sym.typespec().is_string()) {
-            return ll.constant(sym.get_string());
+            return llvm_load_string(sym.get_string());
        }
        OSL_ASSERT(0 && "unhandled constant type");
    }

    return llvm_load_value(llvm_get_pointer(sym), sym.typespec(), deriv,
-                           arrayindex, component, cast);
+                           arrayindex, component, cast,
+                           llnamefmt("{}_", sym.name()));
 }


@@ -659,7 +519,7 @@ BackendLLVM::llvm_load_value(const Symbol& sym, int deriv,
 llvm::Value*
 BackendLLVM::llvm_load_value(llvm::Value* ptr, const TypeSpec& type, int deriv,
                             llvm::Value* arrayindex, int component,
-                             TypeDesc cast)
+                             TypeDesc cast, const std::string& llname)
 {
    if (!ptr)
        return NULL;  // Error
@@ -681,7 +541,7 @@ BackendLLVM::llvm_load_value(llvm::Value* ptr, const TypeSpec& type, int deriv,
        ptr = ll.GEP(ptr, 0, component);

    // Now grab the value
-    llvm::Value* result = ll.op_load(ptr);
+    llvm::Value* result = ll.op_load(ptr, llname);

    if (type.is_closure_based())
        return result;
@@ -691,46 +551,20 @@ BackendLLVM::llvm_load_value(llvm::Value* ptr, const TypeSpec& type, int deriv,
        result = ll.op_float_to_int(result);
    else if (type.is_int() && cast == TypeDesc::TypeFloat)
        result = ll.op_int_to_float(result);
-    else if (type.is_string() && cast == TypeDesc::LONGLONG)
+    else if (type.is_string() && cast == TypeDesc::LONGLONG) {
        result = ll.ptr_to_cast(result, ll.type_longlong());
-
-    return result;
-}
-
-
-
-llvm::Value*
-BackendLLVM::llvm_load_device_string(const Symbol& sym, bool follow)
-{
-    // TODO: need to make this work with arrays of strings
-    OSL_ASSERT(use_optix() && "This is only intended to be used with CUDA");
-
-    // Recover the userdata index for non-constant parameters
-    int userdata_index = find_userdata_index(sym);
-
-    llvm::Value* val = NULL;
-    if (sym.symtype() == SymTypeLocal || sym.symtype() == SymTypeTemp) {
-        // Handle temporary and local variables
-        val = getOrAllocateLLVMSymbol(sym);
-        val = ll.ptr_cast(val, ll.type_longlong_ptr());
-    } else if (userdata_index < 0) {
-        // Handle non-varying variables
-        OSL_DASSERT(sym.data() && "NULL data in non-varying string");
-        val = getOrAllocateCUDAVariable(sym);
-    } else {
-        // Handle potentially varying variables
-        val = ll.ptr_cast(groupdata_field_ptr(2 + userdata_index),
-                          ll.type_longlong_ptr());
+        // FIXME -- is this case ever used? Seems strange.
    }

-    // It's preferable to deal with device strings "symbolically" through the
-    // CUDA variable (essentially a char**), which helps keep the code
-    // portable. But sometimes it's necessary to handle the underlying char*
-    // directly, e.g. when printing or writing out a closure param.
-    if (follow)
-        val = ll.int_to_ptr_cast(ll.op_load(val));
+    // We still disguise hashed strings as char*. Ideally, we would only do
+    // that if the rep were a charptr, but we disguise the hashes as char*'s
+    // also to avoid ugliness with function signatures differing between CPU
+    // and GPU. Maybe some day we'll use the hash representation on both
+    // sides?
+    if (type.is_string() && ll.ustring_rep() != LLVM_Util::UstringRep::charptr)
+        result = ll.int_to_ptr_cast(result);

-    return val;
+    return result;
 }


@@ -764,13 +598,8 @@ BackendLLVM::llvm_load_constant_value(const Symbol& sym, int arrayindex,
        int ncomps = (int)sym.typespec().aggregate();
        return ll.constant(sym.get_float(ncomps * arrayindex + component));
    }
-    if (sym.typespec().is_string() && use_optix()) {
-        OSL_DASSERT((arrayindex == 0)
-                    && "String arrays are not currently supported in OptiX");
-        return llvm_load_device_string(sym, /*follow*/ false);
-    }
    if (sym.typespec().is_string()) {
-        return ll.constant(sym.get_string(arrayindex));
+        return llvm_load_string(sym.get_string(arrayindex));
    }

    OSL_ASSERT(0 && "unhandled constant type");
@@ -920,16 +749,20 @@ BackendLLVM::llvm_store_component_value(llvm::Value* new_val, const Symbol& sym,
 llvm::Value*
 BackendLLVM::groupdata_field_ref(int fieldnum)
 {
-    return ll.GEP(groupdata_ptr(), 0, fieldnum);
+    return ll.GEP(groupdata_ptr(), 0, fieldnum,
+                  llnamefmt("{}_ref", m_groupdata_field_names[fieldnum]));
 }


 llvm::Value*
 BackendLLVM::groupdata_field_ptr(int fieldnum, TypeDesc type)
 {
-    llvm::Value* result = ll.void_ptr(groupdata_field_ref(fieldnum));
+    llvm::Value* result = groupdata_field_ref(fieldnum);
+    std::string llname = llnamefmt("{}_ptr", m_groupdata_field_names[fieldnum]);
    if (type != TypeDesc::UNKNOWN)
-        result = ll.ptr_to_cast(result, llvm_type(type));
+        result = ll.ptr_to_cast(result, llvm_type(type), llname);
+    else
+        result = ll.void_ptr(result, llname);
    return result;
 }

@@ -972,8 +805,6 @@ BackendLLVM::llvm_call_function(const char* name, cspan<const Symbol*> args,
        const Symbol& s = *(args[i]);
        if (s.typespec().is_closure())
            valargs[i] = llvm_load_value(s);
-        else if (use_optix() && s.typespec().is_string())
-            valargs[i] = llvm_load_device_string(s, /*follow*/ true);
        else if (s.typespec().simpletype().aggregate > 1
                 || (deriv_ptrs && s.has_derivs()))
            valargs[i] = llvm_void_ptr(s);
@@ -983,6 +814,8 @@ BackendLLVM::llvm_call_function(const char* name, cspan<const Symbol*> args,
    return ll.call_function(name, cspan<llvm::Value*>(valargs, args.size()));
 }

+
+
 llvm::Value*
 BackendLLVM::llvm_test_nonzero(Symbol& val, bool test_derivs)
 {
@@ -1056,8 +889,8 @@ BackendLLVM::llvm_assign_impl(Symbol& Result, Symbol& Src, int arrayindex,
    // the size difference is intended (by the optimizer).
    if (result_t.is_array() && src_t.is_array() && arrayindex == -1) {
        OSL_DASSERT(assignable(result_t.elementtype(), src_t.elementtype()));
-        llvm::Value* resultptr = llvm_void_ptr(Result);
-        llvm::Value* srcptr    = llvm_void_ptr(Src);
+        llvm::Value* resultptr = llvm_get_pointer(Result);
+        llvm::Value* srcptr    = llvm_get_pointer(Src);
        int len                = std::min(Result.size(), Src.size());
        int align              = result_t.is_closure_based()
                                     ? (int)sizeof(void*)
@@ -1078,10 +911,7 @@ BackendLLVM::llvm_assign_impl(Symbol& Result, Symbol& Src, int arrayindex,
    TypeDesc basetype        = TypeDesc::BASETYPE(rt.basetype);
    const int num_components = rt.aggregate;
    const bool singlechan    = (srccomp != -1) || (dstcomp != -1);
-    if (use_optix() && Src.typespec().is_string()) {
-        llvm::Value* src = llvm_load_device_string(Src, /*follow*/ true);
-        llvm_store_value(ll.ptr_cast(src, ll.type_void_ptr()), Result);
-    } else if (!singlechan) {
+    if (!singlechan) {
        for (int i = 0; i < num_components; ++i) {
            llvm::Value* src_val
                = Src.is_constant()
@@ -4,15 +4,15 @@

 #pragma once

-#include <vector>
 #include <map>
+#include <vector>

 #include "oslexec_pvt.h"
 using namespace OSL;
 using namespace OSL::pvt;

-#include "runtimeoptimize.h"
 #include <OSL/llvm_util.h>
+#include "runtimeoptimize.h"

 // additional includes for creating global OptiX variables
 #include "llvm/IR/Constants.h"
@@ -22,7 +22,7 @@ using namespace OSL::pvt;

 OSL_NAMESPACE_ENTER

-namespace pvt {   // OSL::pvt
+namespace pvt {  // OSL::pvt



@@ -30,16 +30,16 @@ namespace pvt {   // OSL::pvt
 /// code to implement a shader group.
 class BackendLLVM final : public OSOProcessorBase {
 public:
-    BackendLLVM (ShadingSystemImpl &shadingsys, ShaderGroup &group,
-                ShadingContext *context);
+    BackendLLVM(ShadingSystemImpl& shadingsys, ShaderGroup& group,
+                ShadingContext* context);

-    virtual ~BackendLLVM ();
+    virtual ~BackendLLVM();

-    virtual void set_inst (int layer);
+    virtual void set_inst(int layer);

    /// Create an llvm function for the whole shader group, JIT it,
    /// and store the llvm::Function* handle to it with the ShaderGroup.
-    virtual void run ();
+    virtual void run();


    /// What LLVM debug level are we at?
@@ -47,33 +47,33 @@ public:

    /// Set up a bunch of static things we'll need for the whole group.
    ///
-    void initialize_llvm_group ();
+    void initialize_llvm_group();

-    int layer_remap (int origlayer) const { return m_layer_remap[origlayer]; }
+    int layer_remap(int origlayer) const { return m_layer_remap[origlayer]; }

    /// Create an llvm function for the current shader instance.
    /// This will end up being the group entry if 'groupentry' is true.
-    llvm::Function* build_llvm_instance (bool groupentry);
+    llvm::Function* build_llvm_instance(bool groupentry);

    /// Create an llvm function for group initialization code.
-    llvm::Function* build_llvm_init ();
+    llvm::Function* build_llvm_init();

    /// Build up LLVM IR code for the given range [begin,end) or
    /// opcodes, putting them (initially) into basic block bb (or the
    /// current basic block if bb==NULL).
-    bool build_llvm_code (int beginop, int endop, llvm::BasicBlock *bb=NULL);
+    bool build_llvm_code(int beginop, int endop, llvm::BasicBlock* bb = NULL);

    typedef std::map<std::string, llvm::Value*> AllocationMap;

-    void llvm_assign_initial_value (const Symbol& sym, bool force = false);
-    llvm::LLVMContext &llvm_context () const { return ll.context(); }
-    AllocationMap &named_values () { return m_named_values; }
+    void llvm_assign_initial_value(const Symbol& sym, bool force = false);
+    llvm::LLVMContext& llvm_context() const { return ll.context(); }
+    AllocationMap& named_values() { return m_named_values; }

    /// Return an llvm::Value* corresponding to the address of the given
    /// symbol element, with derivative (0=value, 1=dx, 2=dy) and array
    /// index (NULL if it's not an array).
-    llvm::Value *llvm_get_pointer (const Symbol& sym, int deriv=0,
-                                   llvm::Value *arrayindex=NULL);
+    llvm::Value* llvm_get_pointer(const Symbol& sym, int deriv = 0,
+                                  llvm::Value* arrayindex = NULL);

    /// Return the llvm::Value* corresponding to the given element
    /// value, with derivative (0=value, 1=dx, 2=dy), array index (NULL
@@ -83,9 +83,9 @@ public:
    /// return the scalar -- this allows automatic casting to triples.
    /// Finally, auto-cast int<->float if requested (no conversion is
    /// performed if cast is the default of UNKNOWN).
-    llvm::Value *llvm_load_value (const Symbol& sym, int deriv,
-                                  llvm::Value *arrayindex, int component,
-                                  TypeDesc cast=TypeDesc::UNKNOWN);
+    llvm::Value* llvm_load_value(const Symbol& sym, int deriv,
+                                 llvm::Value* arrayindex, int component,
+                                 TypeDesc cast = TypeDesc::UNKNOWN);


    /// Given an llvm::Value* of a pointer (and the type of the data
@@ -97,62 +97,71 @@ public:
    /// and it's a scalar, return the scalar -- this allows automatic
    /// casting to triples.  Finally, auto-cast int<->float if requested
    /// (no conversion is performed if cast is the default of UNKNOWN).
-    llvm::Value *llvm_load_value (llvm::Value *ptr, const TypeSpec &type,
-                              int deriv, llvm::Value *arrayindex,
-                              int component, TypeDesc cast=TypeDesc::UNKNOWN);
+    llvm::Value* llvm_load_value(llvm::Value* ptr, const TypeSpec& type,
+                                 int deriv, llvm::Value* arrayindex,
+                                 int component,
+                                 TypeDesc cast             = TypeDesc::UNKNOWN,
+                                 const std::string& llname = {});

    /// Just like llvm_load_value, but when both the symbol and the
    /// array index are known to be constants.  This can even handle
    /// pulling constant-indexed elements out of constant arrays.  Use
    /// arrayindex==-1 to indicate that it's not an array dereference.
-    llvm::Value *llvm_load_constant_value (const Symbol& sym,
-                                           int arrayindex, int component,
-                                           TypeDesc cast=TypeDesc::UNKNOWN);
+    llvm::Value* llvm_load_constant_value(const Symbol& sym, int arrayindex,
+                                          int component,
+                                          TypeDesc cast = TypeDesc::UNKNOWN);

    /// llvm_load_value with non-constant component designation.  Does
    /// not work with arrays or do type casts!
-    llvm::Value *llvm_load_component_value (const Symbol& sym, int deriv,
-                                            llvm::Value *component);
+    llvm::Value* llvm_load_component_value(const Symbol& sym, int deriv,
+                                           llvm::Value* component);

    /// Non-array version of llvm_load_value, with default deriv &
    /// component.
-    llvm::Value *llvm_load_value (const Symbol& sym, int deriv = 0,
-                                  int component = 0,
-                                  TypeDesc cast=TypeDesc::UNKNOWN) {
-        return (use_optix() && ! sym.typespec().is_closure() && sym.typespec().is_string())
-            ? llvm_load_device_string (sym, /*follow*/ true)
-            : llvm_load_value (sym, deriv, NULL, component, cast);
+    llvm::Value* llvm_load_value(const Symbol& sym, int deriv = 0,
+                                 int component = 0,
+                                 TypeDesc cast = TypeDesc::UNKNOWN)
+    {
+        return llvm_load_value(sym, deriv, NULL, component, cast);
    }

-    /// Load the address of a global device-side string pointer, which may
-    /// reside in a global variable, the groupdata struct, or a local value.
-    llvm::Value *llvm_load_device_string (const Symbol& sym, bool follow);
-
    /// Convenience function to load a string for CPU or GPU device
-    llvm::Value *llvm_load_string (const Symbol& sym) {
+    llvm::Value* llvm_load_string(const Symbol& sym)
+    {
        OSL_DASSERT(sym.typespec().is_string());
-        return use_optix()
-            ? llvm_load_device_string(sym, /*follow*/ true)
-            : llvm_load_value(sym);
+        return llvm_load_value(sym);
+    }
+
+    /// Convenience function to load a constant string for CPU or GPU device.
+    /// On the GPU, we use the ustring hash, not the character pointer.
+    llvm::Value* llvm_load_string(ustring str) { return ll.constant(str); }
+
+    llvm::Value* llvm_load_string(string_view str)
+    {
+        return llvm_load_string(ustring(str));
    }

    /// Legacy version
    ///
-    llvm::Value *loadLLVMValue (const Symbol& sym, int component=0,
-                                int deriv=0, TypeDesc cast=TypeDesc::UNKNOWN) {
-        return llvm_load_value (sym, deriv, NULL, component, cast);
+    llvm::Value* loadLLVMValue(const Symbol& sym, int component = 0,
+                               int deriv = 0, TypeDesc cast = TypeDesc::UNKNOWN)
+    {
+        return llvm_load_value(sym, deriv, NULL, component, cast);
    }

-    /// Return an llvm::Value* that is either a scalar and derivs is
-    /// false, or a pointer to sym's values (if sym is an aggregate or
-    /// derivs == true).  Furthermore, if deriv == true and sym doesn't
-    /// have derivs, coerce it into a variable with zero derivs.
-    llvm::Value *llvm_load_arg (const Symbol& sym, bool derivs);
+    /// Return an llvm::Value* in the form that we will pass a float-based
+    /// symbol as a function argument to any "built-in" OSL function -- as a
+    /// simple value if the symbol is a scalar and no derivs are needed, or as
+    /// a pointer to the data in all other cases (aggregates, arrays, or
+    /// derivs needed). If deriv == true and sym doesn't have derivs, coerce
+    /// it into a variable having derivatives set to 0.0.
+    llvm::Value* llvm_load_arg(const Symbol& sym, bool derivs);

    /// Just like llvm_load_arg(sym,deriv), except use use sym's derivs
    /// as-is, no coercion.
-    llvm::Value *llvm_load_arg (const Symbol& sym) {
-        return llvm_load_arg (sym, sym.has_derivs());
+    llvm::Value* llvm_load_arg(const Symbol& sym)
+    {
+        return llvm_load_arg(sym, sym.has_derivs());
    }

    /// Store new_val into the given symbol, given the derivative
@@ -161,8 +170,8 @@ public:
    /// symbol doesn't have a deriv, it's a nop.  If the component >0
    /// and it's a scalar, set the scalar.  Returns true if ok, false
    /// upon failure.
-    bool llvm_store_value (llvm::Value *new_val, const Symbol& sym, int deriv,
-                           llvm::Value *arrayindex, int component);
+    bool llvm_store_value(llvm::Value* new_val, const Symbol& sym, int deriv,
+                          llvm::Value* arrayindex, int component);

    /// Store new_val into the memory pointed to by dst_ptr, given the
    /// derivative (0=value, 1=dx, 2=dy), array index (NULL if it's not
@@ -170,150 +179,160 @@ public:
    /// and the symbol doesn't have a deriv, it's a nop.  If the
    /// component >0 and it's a scalar, set the scalar.  Returns true if
    /// ok, false upon failure.
-    bool llvm_store_value (llvm::Value* new_val, llvm::Value* dst_ptr,
-                           const TypeSpec &type, int deriv,
-                           llvm::Value* arrayindex, int component);
+    bool llvm_store_value(llvm::Value* new_val, llvm::Value* dst_ptr,
+                          const TypeSpec& type, int deriv,
+                          llvm::Value* arrayindex, int component);

    /// Non-array version of llvm_store_value, with default deriv &
    /// component.
-    bool llvm_store_value (llvm::Value *new_val, const Symbol& sym,
-                           int deriv=0, int component=0) {
-        return llvm_store_value (new_val, sym, deriv, NULL, component);
+    bool llvm_store_value(llvm::Value* new_val, const Symbol& sym,
+                          int deriv = 0, int component = 0)
+    {
+        return llvm_store_value(new_val, sym, deriv, NULL, component);
    }

    /// llvm_store_value with non-constant component designation.  Does
    /// not work with arrays or do type casts!
-    bool llvm_store_component_value (llvm::Value *new_val, const Symbol& sym,
-                                     int deriv, llvm::Value *component);
+    bool llvm_store_component_value(llvm::Value* new_val, const Symbol& sym,
+                                    int deriv, llvm::Value* component);

    /// Legacy version
    ///
-    bool storeLLVMValue (llvm::Value* new_val, const Symbol& sym,
-                         int component=0, int deriv=0) {
-        return llvm_store_value (new_val, sym, deriv, component);
+    bool storeLLVMValue(llvm::Value* new_val, const Symbol& sym,
+                        int component = 0, int deriv = 0)
+    {
+        return llvm_store_value(new_val, sym, deriv, component);
    }

    /// Generate an alloca instruction to allocate space for the given
    /// type, with derivs if derivs==true, and return the its pointer.
-    llvm::Value *llvm_alloca (const TypeSpec &type, bool derivs,
-                              const std::string &name="", int align=0);
+    llvm::Value* llvm_alloca(const TypeSpec& type, bool derivs,
+                             const std::string& name = "", int align = 0);

    /// Given the OSL symbol, return the llvm::Value* corresponding to the
    /// address of the start of that symbol (first element, first component,
    /// and just the plain value if it has derivatives).  This is retrieved
    /// from the allocation map if already there; and if not yet in the
    /// map, the symbol is alloca'd and placed in the map.
-    llvm::Value *getOrAllocateLLVMSymbol (const Symbol& sym);
+    llvm::Value* getOrAllocateLLVMSymbol(const Symbol& sym);

    /// Allocate a CUDA variable for the given OSL symbol and return a pointer
    /// to the corresponding LLVM GlobalVariable, or return the pointer if it
    /// has already been allocated.
-    llvm::Value *getOrAllocateCUDAVariable (const Symbol& sym, bool addMetadata=false);
+    llvm::Value* getOrAllocateCUDAVariable(const Symbol& sym);

-    /// Create a CUDA global variable and add it to the current Module
-    llvm::Value *addCUDAVariable (const std::string& name, int size, int alignment,
-                                  const void* data, TypeDesc type=TypeDesc::UNKNOWN);
-
-    /// Create the extra semantic information needed for OptiX variables
-    void createOptixMetadata (const std::string& name, const Symbol& sym );
+    /// Create a named CUDA global variable with the given type, size, and
+    /// alignment, and add it to the current Module. It will be initialized
+    /// with data pointed to by init_data. A record will be also added to
+    /// m_const_map, and will be retrieved by subsequent calls to
+    /// getOrAllocateCUDAVariable().
+    llvm::Value* addCUDAGlobalVariable(const std::string& name, int size,
+                                       int alignment, const void* init_data,
+                                       TypeDesc type = TypeDesc::UNKNOWN);

    /// Retrieve an llvm::Value that is a pointer holding the start address
    /// of the specified symbol. This always works for globals and params;
    /// for stack variables (locals/temps) is succeeds only if the symbol is
    /// already in the allocation table (will fail otherwise). This method
    /// is not designed to retrieve constants.
-    llvm::Value *getLLVMSymbolBase (const Symbol &sym);
+    llvm::Value* getLLVMSymbolBase(const Symbol& sym);

    /// Retrieve the named global ("P", "N", etc.).
-    llvm::Value *llvm_global_symbol_ptr (ustring name);
+    llvm::Value* llvm_global_symbol_ptr(ustring name);

    /// Test whether val is nonzero, return the llvm::Value* that's the
    /// result of a CreateICmpNE or CreateFCmpUNE (depending on the
    /// type).  If test_derivs is true, it it also tests whether the
    /// derivs are zero.
-    llvm::Value *llvm_test_nonzero (Symbol &val, bool test_derivs = false);
+    llvm::Value* llvm_test_nonzero(Symbol& val, bool test_derivs = false);

-    /// Implementaiton of Simple assignment.  If arrayindex >= 0, in
+    /// Implementation of Simple assignment.  If arrayindex >= 0, in
    /// designates a particular array index to assign.
-    bool llvm_assign_impl (Symbol &Result, Symbol &Src, int arrayindex = -1,
-                           int srccomp = -1, int dstcomp = -1);
+    bool llvm_assign_impl(Symbol& Result, Symbol& Src, int arrayindex = -1,
+                          int srccomp = -1, int dstcomp = -1);


    /// Convert the name of a global (and its derivative index) into the
    /// field number of the ShaderGlobals struct.
-    int ShaderGlobalNameToIndex (ustring name);
+    int ShaderGlobalNameToIndex(ustring name);

    /// Return the LLVM type handle for the ShaderGlobals struct.
    ///
-    llvm::Type *llvm_type_sg ();
+    llvm::Type* llvm_type_sg();

    /// Return the LLVM type handle for a pointer to a
    /// ShaderGlobals struct.
-    llvm::Type *llvm_type_sg_ptr ();
+    llvm::Type* llvm_type_sg_ptr();

    /// Return the ShaderGlobals pointer.
    ///
-    llvm::Value *sg_ptr () const { return m_llvm_shaderglobals_ptr; }
+    llvm::Value* sg_ptr() const { return m_llvm_shaderglobals_ptr; }

-    llvm::Type *llvm_type_closure_component ();
-    llvm::Type *llvm_type_closure_component_ptr ();
+    llvm::Type* llvm_type_closure_component();
+    llvm::Type* llvm_type_closure_component_ptr();

    /// Return the ShaderGlobals pointer cast as a void*.
    ///
-    llvm::Value *sg_void_ptr () {
-        return ll.void_ptr (m_llvm_shaderglobals_ptr);
+    llvm::Value* sg_void_ptr() { return ll.void_ptr(m_llvm_shaderglobals_ptr); }
+
+    /// Cast the pointer variable specified by val to a pointer to the
+    /// basic type comprising `type`.
+    llvm::Value* llvm_ptr_cast(llvm::Value* val, const TypeSpec& type,
+                               const std::string& llname = {})
+    {
+        return ll.ptr_cast(val, type.simpletype(), llname);
    }

-    llvm::Value *llvm_ptr_cast (llvm::Value* val, const TypeSpec &type) {
-        return ll.ptr_cast (val, type.simpletype());
-    }
-
-    llvm::Value *llvm_void_ptr (const Symbol &sym, int deriv=0) {
-        return ll.void_ptr (llvm_get_pointer(sym, deriv));
+    llvm::Value* llvm_void_ptr(const Symbol& sym, int deriv = 0)
+    {
+        return ll.void_ptr(llvm_get_pointer(sym, deriv),
+                           llnamefmt("{}_voidptr", sym.mangled()));
    }

    /// Return the LLVM type handle for a structure of the common group
    /// data that holds all the shader params.
-    llvm::Type *llvm_type_groupdata ();
+    llvm::Type* llvm_type_groupdata();

    /// Return the LLVM type handle for a pointer to the common group
    /// data that holds all the shader params.
-    llvm::Type *llvm_type_groupdata_ptr ();
+    llvm::Type* llvm_type_groupdata_ptr();

    /// Return the group data pointer.
    ///
-    llvm::Value *groupdata_ptr () const { return m_llvm_groupdata_ptr; }
+    llvm::Value* groupdata_ptr() const { return m_llvm_groupdata_ptr; }

    /// Return the group data pointer cast as a void*.
    ///
-    llvm::Value *groupdata_void_ptr () {
-        return ll.void_ptr (m_llvm_groupdata_ptr);
+    llvm::Value* groupdata_void_ptr()
+    {
+        return ll.void_ptr(m_llvm_groupdata_ptr);
    }

    /// Return a reference to the specified field within the group data.
-    llvm::Value *groupdata_field_ref (int fieldnum);
+    llvm::Value* groupdata_field_ref(int fieldnum);

    /// Return a pointer to the specified field within the group data,
    /// optionally cast to pointer to a particular data type.
-    llvm::Value *groupdata_field_ptr (int fieldnum,
-                                      TypeDesc type = TypeDesc::UNKNOWN);
+    llvm::Value* groupdata_field_ptr(int fieldnum,
+                                     TypeDesc type = TypeDesc::UNKNOWN);

    /// Return the userdata base pointer.
-    llvm::Value *userdata_base_ptr () const { return m_llvm_userdata_base_ptr; }
+    llvm::Value* userdata_base_ptr() const { return m_llvm_userdata_base_ptr; }

    /// Return the output base pointer.
-    llvm::Value *output_base_ptr () const { return m_llvm_output_base_ptr; }
+    llvm::Value* output_base_ptr() const { return m_llvm_output_base_ptr; }

    /// Return the shade index
-    llvm::Value *shadeindex () const { return m_llvm_shadeindex; }
+    llvm::Value* shadeindex() const { return m_llvm_shadeindex; }

    // For a symloc, compute the llvm::Value of the pointer to its true,
    // offset location from the base pointer for shade index `sindex`
    // (which should already be a i64, or if nullptr, then use
    // m_llvm_shadeindex and convert it to i64).
-    llvm::Value *symloc_ptr(const SymLocationDesc* symloc,
+    llvm::Value* symloc_ptr(const SymLocationDesc* symloc,
                            llvm::Value* base_ptr,
-                            llvm::Value* sindex = nullptr) {
+                            llvm::Value* sindex = nullptr)
+    {
        llvm::Value* offset = ll.constanti64(symloc->offset);
        llvm::Value* stride = ll.constanti64(symloc->stride);
        if (!sindex)
@@ -324,47 +343,47 @@ public:

    /// Return a ref to the bool where the "layer_run" flag is stored for
    /// the specified layer.
-    llvm::Value *layer_run_ref (int layer);
+    llvm::Value* layer_run_ref(int layer);

    /// Return a ref to the bool where the "userdata_initialized" flag is
    /// stored for the specified userdata index.
-    llvm::Value *userdata_initialized_ref (int userdata_index=0);
+    llvm::Value* userdata_initialized_ref(int userdata_index = 0);

    /// Generate LLVM code to zero out the variable (including derivs)
    ///
-    void llvm_assign_zero (const Symbol &sym);
+    void llvm_assign_zero(const Symbol& sym);

    /// Generate LLVM code to zero out the derivatives of sym.
    ///
-    void llvm_zero_derivs (const Symbol &sym);
+    void llvm_zero_derivs(const Symbol& sym);

    /// Generate LLVM code to zero out the derivatives of an array
    /// only for the first count elements of it.
    ///
-    void llvm_zero_derivs (const Symbol &sym, llvm::Value *count);
+    void llvm_zero_derivs(const Symbol& sym, llvm::Value* count);

    /// Generate a debugging printf at shader execution time.
-    void llvm_gen_debug_printf (string_view message);
+    void llvm_gen_debug_printf(string_view message);

    /// Generate a warning message at shader execution time.
-    void llvm_gen_warning (string_view message);
+    void llvm_gen_warning(string_view message);

    /// Generate an error message at shader execution time.
-    void llvm_gen_error (string_view message);
+    void llvm_gen_error(string_view message);

    /// Generate code to call the given layer.  If 'unconditional' is
    /// true, call it without even testing if the layer has already been
    /// called.
-    void llvm_call_layer (int layer, bool unconditional = false);
+    void llvm_call_layer(int layer, bool unconditional = false);

    /// Execute the upstream connection (if any, and if not yet run) that
    /// establishes the value of symbol sym, which has index 'symindex'
    /// within the current layer rop.inst().  If already_run is not NULL,
-    /// it points to a vector of layer indices that are known to have been 
+    /// it points to a vector of layer indices that are known to have been
    /// run -- those can be skipped without dynamically checking their
    /// execution status.
-    void llvm_run_connected_layers (Symbol &sym, int symindex, int opnum = -1,
-                                    std::set<int> *already_run = NULL);
+    void llvm_run_connected_layers(Symbol& sym, int symindex, int opnum = -1,
+                                   std::set<int>* already_run = NULL);

    /// Generate code for a call to the named function with the given
    /// arg list as symbols -- float & ints will be passed by value,
@@ -372,133 +391,161 @@ public:
    /// is true, pass pointers even for floats if they have derivs.
    /// Return an llvm::Value* corresponding to the return value of the
    /// function, if any.
-    llvm::Value *llvm_call_function (const char *name, cspan<const Symbol *> args,
-                                     bool deriv_ptrs=false);
-    llvm::Value *llvm_call_function (const char *name, const Symbol &A,
-                                     bool deriv_ptrs=false) {
-        return llvm_call_function (name, { &A }, deriv_ptrs);
+    llvm::Value* llvm_call_function(const char* name, cspan<const Symbol*> args,
+                                    bool deriv_ptrs = false);
+    llvm::Value* llvm_call_function(const char* name, const Symbol& A,
+                                    bool deriv_ptrs = false)
+    {
+        return llvm_call_function(name, { &A }, deriv_ptrs);
    }
-    llvm::Value *llvm_call_function (const char *name, const Symbol &A,
-                                     const Symbol &B, bool deriv_ptrs=false) {
-        return llvm_call_function (name, { &A, &B }, deriv_ptrs);
+    llvm::Value* llvm_call_function(const char* name, const Symbol& A,
+                                    const Symbol& B, bool deriv_ptrs = false)
+    {
+        return llvm_call_function(name, { &A, &B }, deriv_ptrs);
    }
-    llvm::Value *llvm_call_function (const char *name, const Symbol &A,
-                                     const Symbol &B, const Symbol &C,
-                                     bool deriv_ptrs=false) {
-        return llvm_call_function (name, { &A, &B, &C }, deriv_ptrs);
+    llvm::Value* llvm_call_function(const char* name, const Symbol& A,
+                                    const Symbol& B, const Symbol& C,
+                                    bool deriv_ptrs = false)
+    {
+        return llvm_call_function(name, { &A, &B, &C }, deriv_ptrs);
    }

-    TypeDesc llvm_typedesc (const TypeSpec &typespec) {
-        return typespec.is_closure_based()
-           ? TypeDesc(TypeDesc::PTR, typespec.arraylength())
-           : typespec.simpletype();
+    TypeDesc llvm_typedesc(const TypeSpec& typespec)
+    {
+        if (typespec.is_closure_based())
+            return TypeDesc(TypeDesc::PTR, typespec.arraylength());
+        else if (use_optix() && typespec.is_string_based()) {
+            // On the OptiX side, we use the uint64 hash to represent a string
+            return TypeDesc(TypeDesc::UINT64, typespec.arraylength());
+        } else
+            return typespec.simpletype();
    }

    /// Generate the appropriate llvm type definition for a TypeSpec
    /// (this is the actual type, for example when we allocate it).
    /// Allocates ptrs for closures.
-    llvm::Type *llvm_type (const TypeSpec &typespec) {
-        return ll.llvm_type (llvm_typedesc(typespec));
+    llvm::Type* llvm_type(const TypeSpec& typespec)
+    {
+        return ll.llvm_type(llvm_typedesc(typespec));
    }

    /// Generate the parameter-passing llvm type definition for an OSL
    /// TypeSpec.
-    llvm::Type *llvm_pass_type (const TypeSpec &typespec);
+    llvm::Type* llvm_pass_type(const TypeSpec& typespec);

-    llvm::PointerType *llvm_type_prepare_closure_func() { return m_llvm_type_prepare_closure_func; }
-    llvm::PointerType *llvm_type_setup_closure_func() { return m_llvm_type_setup_closure_func; }
-
-    /// Return the basic block of the exit for the whole instance.
-    ///
-    bool llvm_has_exit_instance_block () const {
-        return m_exit_instance_block;
+    llvm::PointerType* llvm_type_prepare_closure_func()
+    {
+        return m_llvm_type_prepare_closure_func;
+    }
+    llvm::PointerType* llvm_type_setup_closure_func()
+    {
+        return m_llvm_type_setup_closure_func;
    }

    /// Return the basic block of the exit for the whole instance.
    ///
-    llvm::BasicBlock *llvm_exit_instance_block () {
-        if (! m_exit_instance_block) {
-            std::string name = fmtformat("{}_{}_exit_", inst()->layername(),
+    bool llvm_has_exit_instance_block() const { return m_exit_instance_block; }
+
+    /// Return the basic block of the exit for the whole instance.
+    ///
+    llvm::BasicBlock* llvm_exit_instance_block()
+    {
+        if (!m_exit_instance_block) {
+            std::string name = llnamefmt("{}_{}_exit_", inst()->layername(),
                                         inst()->id());
-            m_exit_instance_block = ll.new_basic_block (name);
+            m_exit_instance_block = ll.new_basic_block(name);
        }
        return m_exit_instance_block;
    }

    /// Check for inf/nan in all written-to arguments of the op
-    void llvm_generate_debugnan (const Opcode &op);
+    void llvm_generate_debugnan(const Opcode& op);
    /// Check for uninitialized values in all read-from arguments to the op
-    void llvm_generate_debug_uninit (const Opcode &op);
+    void llvm_generate_debug_uninit(const Opcode& op);
    /// Print debugging line for the op
-    void llvm_generate_debug_op_printf (const Opcode &op);
+    void llvm_generate_debug_op_printf(const Opcode& op);

-    llvm::Function *layer_func () const { return ll.current_function(); }
+    llvm::Function* layer_func() const { return ll.current_function(); }

    /// Call this when JITing a texture-like call, to track how many.
-    void generated_texture_call (bool handle) {
+    void generated_texture_call(bool handle)
+    {
        shadingsys().m_stat_tex_calls_codegened += 1;
        if (handle)
            shadingsys().m_stat_tex_calls_as_handles += 1;
    }

    /// Return the mapping from symbol names to GlobalVariables.
-    std::map<std::string,llvm::GlobalVariable*>& get_const_map() { return m_const_map; }
+    std::map<std::string, llvm::GlobalVariable*>& get_const_map()
+    {
+        return m_const_map;
+    }

    /// Return whether or not we are compiling for an OptiX-based renderer.
    bool use_optix() { return m_use_optix; }

    /// Return if we should compile against free function versions of Renderer Service.
-    bool use_rs_bitcode() {return m_use_rs_bitcode; }
+    bool use_rs_bitcode() { return m_use_rs_bitcode; }

    /// Return the userdata index for the given Symbol.  Return -1 if the Symbol
    /// is not an input parameter or is constant and therefore doesn't have an
    /// entry in the groupdata struct.
-    int find_userdata_index (const Symbol& sym);
+    int find_userdata_index(const Symbol& sym);

    LLVM_Util ll;

-private:
-    std::vector<int> m_layer_remap;     ///< Remapping of layer ordering
-    std::set<int> m_layers_already_run; ///< List of layers run
-    int m_num_used_layers;              ///< Number of layers actually used
+    // Utility for constructing names for llvm symbols. It creates a formatted
+    // string if the shading system's "llvm_output_bitcode" option is set,
+    // otherwise it takes a shortcut and returns an empty string (since nobody
+    // is going to see the pretty bitcode anyway).
+    template<typename Str, typename... Args>
+    OSL_NODISCARD inline std::string llnamefmt(const Str& fmt,
+                                               Args&&... args) const
+    {
+        return m_name_llvm_syms ? fmtformat(fmt, std::forward<Args>(args)...)
+                                : std::string();
+    }

-    double m_stat_total_llvm_time;        ///<   total time spent on LLVM
-    double m_stat_llvm_setup_time;        ///<     llvm setup time
-    double m_stat_llvm_irgen_time;        ///<     llvm IR generation time
-    double m_stat_llvm_opt_time;          ///<     llvm IR optimization time
-    double m_stat_llvm_jit_time;          ///<     llvm JIT time
+private:
+    std::vector<int> m_layer_remap;      ///< Remapping of layer ordering
+    std::set<int> m_layers_already_run;  ///< List of layers run
+    int m_num_used_layers;               ///< Number of layers actually used
+
+    double m_stat_total_llvm_time;  ///<   total time spent on LLVM
+    double m_stat_llvm_setup_time;  ///<     llvm setup time
+    double m_stat_llvm_irgen_time;  ///<     llvm IR generation time
+    double m_stat_llvm_opt_time;    ///<     llvm IR optimization time
+    double m_stat_llvm_jit_time;    ///<     llvm JIT time

    // LLVM stuff
    AllocationMap m_named_values;
-    std::map<const Symbol*,int> m_param_order_map;
-    llvm::Value *m_llvm_shaderglobals_ptr;
-    llvm::Value *m_llvm_groupdata_ptr;
-    llvm::Value *m_llvm_userdata_base_ptr;
-    llvm::Value *m_llvm_output_base_ptr;
-    llvm::Value *m_llvm_shadeindex;
-    llvm::BasicBlock * m_exit_instance_block;  // exit point for the instance
-    llvm::Type *m_llvm_type_sg;  // LLVM type of ShaderGlobals struct
-    llvm::Type *m_llvm_type_groupdata;  // LLVM type of group data
-    llvm::Type *m_llvm_type_closure_component; // LLVM type for ClosureComponent
-    llvm::PointerType *m_llvm_type_prepare_closure_func;
-    llvm::PointerType *m_llvm_type_setup_closure_func;
-    int m_llvm_local_mem;             // Amount of memory we use for locals
+    std::map<const Symbol*, int> m_param_order_map;
+    llvm::Value* m_llvm_shaderglobals_ptr;
+    llvm::Value* m_llvm_groupdata_ptr;
+    llvm::Value* m_llvm_userdata_base_ptr;
+    llvm::Value* m_llvm_output_base_ptr;
+    llvm::Value* m_llvm_shadeindex;
+    llvm::BasicBlock* m_exit_instance_block;  // exit point for the instance
+    llvm::Type* m_llvm_type_sg;         // LLVM type of ShaderGlobals struct
+    llvm::Type* m_llvm_type_groupdata;  // LLVM type of group data
+    llvm::Type* m_llvm_type_closure_component;  // LLVM type for ClosureComponent
+    llvm::PointerType* m_llvm_type_prepare_closure_func;
+    llvm::PointerType* m_llvm_type_setup_closure_func;
+    int m_llvm_local_mem;   // Amount of memory we use for locals
+    bool m_name_llvm_syms;  // Whether to name LLVM symbols

    // A mapping from symbol names to llvm::GlobalVariables
-    std::map<std::string,llvm::GlobalVariable*> m_const_map;
+    std::map<std::string, llvm::GlobalVariable*> m_const_map;

-    // A mapping from canonical strings to string variable names, used to
-    // detect collisions that might occur due to using the string hash to
-    // create variable names.
-    std::map<std::string,std::string>           m_varname_map;
+    // Name of each indexed field in the groupdata, mostly for debugging.
+    std::vector<std::string> m_groupdata_field_names;

-    bool m_use_optix;                   ///< Compile for OptiX?
-
-    bool m_use_rs_bitcode;              /// To use free function versions of Renderer Service functions.
+    bool m_use_optix;  ///< Compile for OptiX?
+    bool m_use_rs_bitcode;  /// To use free function versions of Renderer Service functions.

    friend class ShadingSystemImpl;
 };


-}; // namespace pvt
+};  // namespace pvt
 OSL_NAMESPACE_EXIT
@@ -185,7 +185,7 @@ struct Symbols4ProtectedShaderGlobals {

    SymbolPtrVec varying_symbols;

-    void check_if_varing(Symbol& s)
+    void check_if_varying(Symbol& s)
    {
        if (!is_shader_global_uniform_by_name(s.name())) {
            s.make_varying();
@@ -202,13 +202,13 @@ struct Symbols4ProtectedShaderGlobals {
        , surfacearea(Strings::op_surfacearea, TypeSpec(), SymTypeGlobal)
        , time(Strings::time, TypeSpec(), SymTypeGlobal)
    {
-        check_if_varing(shader2common);
-        check_if_varing(object2common);
-        check_if_varing(flipHandedness);
-        check_if_varing(raytype);
-        check_if_varing(backfacing);
-        check_if_varing(surfacearea);
-        check_if_varing(time);
+        check_if_varying(shader2common);
+        check_if_varying(object2common);
+        check_if_varying(flipHandedness);
+        check_if_varying(raytype);
+        check_if_varying(backfacing);
+        check_if_varying(surfacearea);
+        check_if_varying(time);
    }
 };

@@ -240,7 +240,7 @@ private:
    static constexpr OpcodeTest no_test = nullptr;

    // Map opcode names to a Dependency info
-    std::unordered_map<ustring, Dependency, ustringHash> m_lookup;
+    std::unordered_map<ustring, Dependency> m_lookup;

    /// Return the ptr to the symbol that is the argnum-th argument to the
    /// given op in the instance.
@@ -620,6 +620,20 @@ public:
    }

    Position common_ancestor_between(Position read_pos, Position write_pos)
+    {
+        // Early out when read and write are at the same position
+        if (read_pos == write_pos)
+            return read_pos;
+        // Early out if either position is the root of the dependency tree
+        if (read_pos == end_pos() || write_pos == end_pos())
+            return end_pos();
+        return search_for_common_ancestor_between(read_pos, write_pos);
+    }
+
+    // Because this function makes use of stack arrays, ensure is NOT inlined
+    // inside a loop to avoid repeatedly reserve stack space each iteration
+    OSL_NOINLINE Position search_for_common_ancestor_between(Position read_pos,
+                                                             Position write_pos)
    {
        // To find common anscenstor
        // Walk up both paths building 2 stacks of positions
@@ -629,8 +643,8 @@ public:
        auto write_iter = begin_at(write_pos);

        const int read_count = read_iter.depth() + 1;
-        OSL_STACK_ARRAY(Position, read_path, read_count);
-        int read_depth = 0;
+        Position* read_path  = OSL_ALLOCA(Position, read_count);
+        int read_depth       = 0;
        {
            // Loop structure is to allow
            // writing the end position into the read path
@@ -644,8 +658,8 @@ public:
        }

        const int writeCount = write_iter.depth() + 1;
-        OSL_STACK_ARRAY(Position, write_path, writeCount);
-        int write_depth = 0;
+        Position* write_path = OSL_ALLOCA(Position, writeCount);
+        int write_depth      = 0;
        {
            // Loop structure is to allow
            // writing the end position into the write path
@@ -657,29 +671,53 @@ public:
                ++write_iter;
            }
        }
-        const int levelCount  = std::min(read_depth, write_depth);
-        const int read_level  = read_depth - 1;
-        const int write_level = write_depth - 1;
-        OSL_ASSERT(write_path[write_level] == end_pos());
-        OSL_ASSERT(read_path[read_level] == end_pos());

-        int level = 0;
-        OSL_ASSERT(read_path[read_level - level]
-                   == write_path[write_level - level]);
-        do {
-            ++level;
-            if (read_path[read_level - level]()
-                != write_path[write_level - level]())
+        const int walkCount = std::min(read_depth, write_depth);
+        const int read_top  = read_depth - 1;
+        const int write_top = write_depth - 1;
+        OSL_DASSERT(write_top < writeCount);
+        OSL_DASSERT(write_path[write_top] == end_pos());
+        OSL_DASSERT(read_top < read_count);
+        OSL_DASSERT(read_path[read_top] == end_pos());
+
+        int walkDepth = 0;
+        // Verify top of read and write stacks match positions
+        OSL_DASSERT((read_top - walkDepth) >= 0);
+        OSL_DASSERT((read_top - walkDepth) < read_count);
+        OSL_DASSERT((write_top - walkDepth) >= 0);
+        OSL_DASSERT((write_top - walkDepth) < writeCount);
+        OSL_DASSERT(read_path[read_top - walkDepth]
+                    == write_path[write_top - walkDepth]);
+        // That position should technically be the root (no conditions in the dependency tree)
+        OSL_DASSERT(read_path[read_top - walkDepth] == end_pos());
+        // Because we know walkDepth 0 never differs, we can start iterating
+        // at walkDepth 1
+
+        // Walk down the read and write paths until they differ
+        for (walkDepth = 1; walkDepth < walkCount; ++walkDepth) {
+            OSL_DASSERT((read_top - walkDepth) >= 0);
+            OSL_DASSERT((read_top - walkDepth) < read_count);
+            OSL_DASSERT((write_top - walkDepth) >= 0);
+            OSL_DASSERT((write_top - walkDepth) < writeCount);
+            if (read_path[read_top - walkDepth]()
+                != write_path[write_top - walkDepth]()) {
                break;
-        } while (level < levelCount);
-        --level;
-        OSL_ASSERT(read_path[read_level - level]
-                   == write_path[write_level - level]);
-        OSL_ASSERT(read_level - level >= 0);
-        return read_path[read_level - level];
+            }
+        }
+        // The previous walk depth's position was the common ancestor
+        --walkDepth;
+        OSL_DASSERT((read_top - walkDepth) >= 0);
+        OSL_DASSERT((read_top - walkDepth) < read_count);
+        OSL_DASSERT((write_top - walkDepth) >= 0);
+        OSL_DASSERT((write_top - walkDepth) < writeCount);
+        OSL_DASSERT(read_path[read_top - walkDepth]
+                    == write_path[write_top - walkDepth]);
+        return read_path[read_top - walkDepth];
    }
 };

+
+
 // Historically tracks stack of functions
 // The Position returned by top_pos changes and function scopes are pushed and popped.
 // However any given position is invariant as scopes change
@@ -1053,8 +1091,10 @@ typedef std::vector<WriteEvent> WriteChronology;

 class ReadEvent {
    DependencyTreeTracker::Position m_pos_in_tree;
+#ifdef OSL_DEV
    int m_op_num;
    int m_loop_op_index;
+#endif

 public:
    ReadEvent(DependencyTreeTracker::Position pos_in_tree_
@@ -1077,9 +1117,15 @@ public:
    }

 #ifdef OSL_DEV
-    OSL_FORCEINLINE int op_num() const { return m_op_num; }
+    OSL_FORCEINLINE int op_num() const
+    {
+        return m_op_num;
+    }

-    OSL_FORCEINLINE int loop_op_index() const { return m_loop_op_index; }
+    OSL_FORCEINLINE int loop_op_index() const
+    {
+        return m_loop_op_index;
+    }
 #endif
 };

@@ -1595,7 +1641,7 @@ struct Analyzer {
            // Process arguments written to to handle early outs, cyclical reads,
            // and record a WriteEvent to a potentially unmasked operations remembering
            // the exact stack of conditional dependencies at that point.
-            // When we processed reads (above), we ensured that any reads occuring at a higher  that on the stoCheck for
+            // When we processed reads (above), we ensured that any reads occurring at a higher  that on the stoCheck for
            for (int write_index = 0; write_index < write_count;
                 ++write_index) {
                const Symbol* symbolWrittenTo
@@ -1642,7 +1688,7 @@ struct Analyzer {
                }
            }

-            // Special case for op_getattribute which normaly would be
+            // Special case for op_getattribute which normally would be
            // implicitly varying unless BatchedRendererServices says it
            // is uniform.
            if (opcode.opname() == Strings::op_getattribute) {
@@ -188,7 +188,7 @@ BatchedBackendLLVM::llvm_pass_type(const TypeSpec& typespec)
    else if (t == TypeDesc::INT)
        lt = ll.type_int();
    else if (t == TypeDesc::STRING)
-        lt = (llvm::Type*)ll.type_string();
+        lt = (llvm::Type*)ll.type_ustring();
    else if (t.aggregate == TypeDesc::VEC3)
        lt = (llvm::Type*)ll.type_void_ptr();  //llvm_type_triple_ptr();
    else if (t.aggregate == TypeDesc::MATRIX44)
@@ -867,7 +867,7 @@ BatchedBackendLLVM::llvm_load_value(llvm::Value* ptr, const TypeSpec& type,
                || (ll.llvm_typeof(result) == ll.type_float())
                || (ll.llvm_typeof(result) == ll.type_triple())
                || (ll.llvm_typeof(result) == ll.type_int())
-                || (ll.llvm_typeof(result) == (llvm::Type*)ll.type_string())
+                || (ll.llvm_typeof(result) == ll.type_ustring())
                || (ll.llvm_typeof(result) == ll.type_matrix())
                || (ll.llvm_typeof(result) == ll.type_longlong())) {
                result = ll.widen_value(result);
@@ -888,7 +888,7 @@ BatchedBackendLLVM::llvm_load_value(llvm::Value* ptr, const TypeSpec& type,
                    (ll.llvm_typeof(result) == ll.type_wide_float())
                    || (ll.llvm_typeof(result) == ll.type_wide_int())
                    || (ll.llvm_typeof(result) == ll.type_wide_triple())
-                    || (ll.llvm_typeof(result) == ll.type_wide_string())
+                    || (ll.llvm_typeof(result) == ll.type_wide_ustring())
                    || (ll.llvm_typeof(result) == ll.type_wide_bool())
                    || (ll.llvm_typeof(result) == ll.type_wide_matrix())
                    || (ll.llvm_typeof(result) == ll.type_wide_longlong()));
@@ -1218,7 +1218,8 @@ BatchedBackendLLVM::llvm_store_value(llvm::Value* new_val, llvm::Value* dst_ptr,
        if (!type.is_closure_based() && t.aggregate > 1)
            dst_ptr = ll.GEP(dst_ptr, 0, component);

-        if (ll.type_ptr(ll.llvm_typeof(new_val)) != ll.llvm_typeof(dst_ptr)) {
+        if ((const llvm::Type*)ll.type_ptr(ll.llvm_typeof(new_val))
+            != ll.llvm_typeof(dst_ptr)) {
            std::cerr << " new_val type=";
            {
                llvm::raw_os_ostream os_cerr(std::cerr);
@@ -1231,7 +1232,7 @@ BatchedBackendLLVM::llvm_store_value(llvm::Value* new_val, llvm::Value* dst_ptr,
            }
            std::cerr << std::endl;
        }
-        OSL_ASSERT(ll.type_ptr(ll.llvm_typeof(new_val))
+        OSL_ASSERT((const llvm::Type*)ll.type_ptr(ll.llvm_typeof(new_val))
                   == ll.llvm_typeof(dst_ptr));

        // Finally, store the value.
@@ -1766,7 +1767,8 @@ BatchedBackendLLVM::llvm_test_nonzero(const Symbol& val, bool test_derivs)
        llvm::Value* llvmValue = llvm_get_pointer(val);
        //OSL_DEV_ONLY(std::cout << "llvmValue type=" << ll.llvm_typenameof(llvmValue) << std::endl);

-        if (ll.llvm_typeof(llvmValue) == ll.type_ptr(ll.type_bool())) {
+        if (ll.llvm_typeof(llvmValue)
+            == (const llvm::Type*)ll.type_ptr(ll.type_bool())) {
            return ll.op_ne(llvm_load_value(val), ll.constant_bool(0));
        } else {
            return ll.op_ne(llvm_load_value(val), ll.constant(0));
@@ -1846,7 +1848,7 @@ BatchedBackendLLVM::llvm_assign_impl(const Symbol& Result, const Symbol& Src,
        return true;
    }

-    // memcpy complicated by promotion of uniform to wide during assignment, dissallow
+    // memcpy complicated by promotion of uniform to wide during assignment, disallow

    // The following code handles f=f, f=i, v=v, v=f, v=i, m=m, s=s.
    // Remember that llvm_load_value will automatically convert scalar->triple.
@@ -1904,7 +1906,7 @@ BatchedBackendLLVM::llvm_assign_impl(const Symbol& Result, const Symbol& Src,
            // The llvm_load_value above should have handled bool to int conversions
            // when the basetype == Typedesc::INT

-            // write source float into all compnents when dstcomp == -1, otherwise
+            // write source float into all components when dstcomp == -1, otherwise
            // the single element requested.
            if (dstcomp == -1) {
                for (int i = 0; i < num_components; ++i)
@@ -164,7 +164,7 @@ public:
    llvm::Value* llvm_load_mask(const Symbol& cond);

    /// Return an llvm::Value* that is either a scalar and derivs is
-    /// false, or a pointer to sym's values (if sym is an aggreate or
+    /// false, or a pointer to sym's values (if sym is an aggregate or
    /// derivs == true).  Furthermore, if deriv == true and sym doesn't
    /// have derivs, coerce it into a variable with zero derivs.
    llvm::Value* llvm_load_arg(const Symbol& sym, bool derivs,
@@ -278,9 +278,9 @@ public:

    public:
        TempScope(BatchedBackendLLVM& backend);
-        TempScope(const TempScope& other)  = delete;
-        TempScope(const TempScope&& other) = delete;
-        TempScope& operator=(const TempScope& other) = delete;
+        TempScope(const TempScope& other)             = delete;
+        TempScope(const TempScope&& other)            = delete;
+        TempScope& operator=(const TempScope& other)  = delete;
        TempScope& operator=(const TempScope&& other) = delete;
        ~TempScope();
    };
@@ -331,7 +331,7 @@ public:
    /// derivs are zero.
    llvm::Value* llvm_test_nonzero(const Symbol& val, bool test_derivs = false);

-    /// Implementaiton of Simple assignment.  If arrayindex >= 0, in
+    /// Implementation of Simple assignment.  If arrayindex >= 0, in
    /// designates a particular array index to assign.
    bool llvm_assign_impl(const Symbol& Result, const Symbol& Src,
                          int arrayindex = -1, int srccomp = -1,
@@ -489,7 +489,7 @@ public:

    // Encapsulate creation of function names that encode parameter types,
    // including if each is varying or uniform, and if a mask is required.
-    // Utilize llvm::Twine to efficently combine multiple strings
+    // Utilize llvm::Twine to efficiently combine multiple strings
    // Usage is to start with the function name then append arguments,
    // and at any point masking can be turned on/off as well as batching.
    // The effect of disabling batching would be the function name is
@@ -533,7 +533,7 @@ public:
        ArgVector m_args;

    public:
-        FuncSpec(const FuncSpec&) = delete;
+        FuncSpec(const FuncSpec&)            = delete;
        FuncSpec& operator=(const FuncSpec&) = delete;

        FuncSpec(const llvm::Twine& name)
@@ -557,7 +557,7 @@ LLVMGEN(llvm_gen_arraylength)
    int len = A.typespec().is_unsized_array() ? A.initializers()
                                              : A.typespec().arraylength();

-    // Even thought the array's size should be uniform accross all lanes,
+    // Even thought the array's size should be uniform across all lanes,
    // the symbol its being stored in maybe varying
    llvm::Value* const_length = Result.is_uniform() ? rop.ll.constant(len)
                                                    : rop.ll.wide_constant(len);
@@ -605,7 +605,7 @@ LLVMGEN(llvm_gen_aref)
        } else {
            BatchedBackendLLVM::TempScope temp_scope(rop);

-            // We need a copy of the indices incase the range check clamps them
+            // We need a copy of the indices in case the range check clamps them
            llvm::Value* loc_clamped_wide_index = rop.getOrAllocateTemp(
                TypeSpec(TypeDesc::INT), false /*derivs*/, false /*is_uniform*/,
                false /*forceBool*/,
@@ -687,7 +687,7 @@ LLVMGEN(llvm_gen_aassign)
            }
        } else {
            BatchedBackendLLVM::TempScope temp_scope(rop);
-            // We need a copy of the indices incase the range check clamps them
+            // We need a copy of the indices in case the range check clamps them
            llvm::Value* loc_clamped_wide_index = rop.getOrAllocateTemp(
                TypeSpec(TypeDesc::INT), false /*derivs*/, false /*is_uniform*/,
                false /*forceBool*/,
@@ -1110,10 +1110,10 @@ LLVMGEN(llvm_gen_if)
        // contain a call to a lower level, those must not be executed
        // if the mask is all off
 //#define SKIP_TESTING_IF_ANY_LANES_ACTIVE_TO_AVOID_BRANCH 1
-// NOTE: if SKIP_TESTING_IF_ANY_LANES_ACTIVE_TO_AVOID_BRANCH is defineed then
+// NOTE: if SKIP_TESTING_IF_ANY_LANES_ACTIVE_TO_AVOID_BRANCH is defined then
 // library and layer functions may get called with an entirely empty mask
 #ifndef SKIP_TESTING_IF_ANY_LANES_ACTIVE_TO_AVOID_BRANCH
-        // We use the combined mask stack + the if condition's mask we aready pushed
+        // We use the combined mask stack + the if condition's mask we already pushed
        llvm::Value* anyThenLanesActive = rop.ll.test_if_mask_is_non_zero(
            rop.ll.current_mask());
 #endif
@@ -2302,7 +2302,7 @@ LLVMGEN(llvm_gen_compref)
        if (rop.inst()->master()->range_checking()) {
            BatchedBackendLLVM::TempScope temp_scope(rop);

-            // We need a copy of the indices incase the range check clamps them
+            // We need a copy of the indices in case the range check clamps them
            llvm::Value* loc_clamped_wide_index = rop.getOrAllocateTemp(
                TypeSpec(TypeDesc::INT), false /*derivs*/, false /*is_uniform*/,
                false /*forceBool*/,
@@ -2418,7 +2418,7 @@ LLVMGEN(llvm_gen_compassign)
        if (rop.inst()->master()->range_checking()) {
            BatchedBackendLLVM::TempScope temp_scope(rop);

-            // We need a copy of the indices incase the range check clamps them
+            // We need a copy of the indices in case the range check clamps them
            llvm::Value* loc_clamped_wide_index = rop.getOrAllocateTemp(
                TypeSpec(TypeDesc::INT), false /*derivs*/, false /*is_uniform*/,
                false /*forceBool*/,
@@ -2534,7 +2534,7 @@ LLVMGEN(llvm_gen_mxcompref)
            }
        } else {
            BatchedBackendLLVM::TempScope temp_scope(rop);
-            // We need a copy of the indices incase the range check clamps them
+            // We need a copy of the indices in case the range check clamps them
            llvm::Value* loc_clamped_wide_index = rop.getOrAllocateTemp(
                TypeSpec(TypeDesc::INT), false /*derivs*/, false /*is_uniform*/,
                false /*forceBool*/,
@@ -2636,7 +2636,7 @@ LLVMGEN(llvm_gen_mxcompassign)
            }
        } else {
            BatchedBackendLLVM::TempScope temp_scope(rop);
-            // We need a copy of the indices incase the range check clamps them
+            // We need a copy of the indices in case the range check clamps them
            llvm::Value* loc_clamped_wide_index = rop.getOrAllocateTemp(
                TypeSpec(TypeDesc::INT), false /*derivs*/, false /*is_uniform*/,
                false /*forceBool*/,
@@ -2955,7 +2955,7 @@ LLVMGEN(llvm_gen_compare_op)
            if ((typeOfA == rop.ll.type_bool() && typeOfB == rop.ll.type_int())
                || (typeOfA == rop.ll.type_wide_bool()
                    && typeOfB == rop.ll.type_wide_int())) {
-                // TODO: could optimize for contant 0 and 1 and skip the comparison
+                // TODO: could optimize for constant 0 and 1 and skip the comparison
                a = rop.ll.op_bool_to_int(a);
            }
            if ((typeOfB == rop.ll.type_bool() && typeOfA == rop.ll.type_int())
@@ -3547,7 +3547,7 @@ LLVMGEN(llvm_gen_transform)

            FuncSpec func_spec("build_transform_matrix");
            func_spec.arg_varying(TypeDesc::TypeMatrix44);
-            // Ignore derivatives if uneeded or unsupplied
+            // Ignore derivatives if unneeded or unsupplied
            func_spec.arg(*From, from_is_uniform);
            func_spec.arg(*To, to_is_uniform);
            func_spec.mask();
@@ -3570,7 +3570,7 @@ LLVMGEN(llvm_gen_transform)
            FuncSpec func_spec(op.opname().c_str());
            func_spec.unbatch();
            //std::string func_name = std::string("osl_") + op.opname().c_str() + "_";
-            // Ignore derivatives if uneeded or unsupplied
+            // Ignore derivatives if unneeded or unsupplied
            bool has_derivs = (Result->has_derivs() && P->has_derivs());
            func_spec.arg(*P, has_derivs, P_is_uniform);
            // The matrix is always varying if we looked it up,
@@ -5750,7 +5750,7 @@ llvm_batched_noise_options(BatchedBackendLLVM& rop, int opnum,
            rop.ll.call_function("osl_noiseparams_set_do_filter", opt,
                                 rop.llvm_load_value(Val));
        } else if (name == Strings::direction && Val.typespec().is_triple()) {
-            // We are not going to bin by varing direction
+            // We are not going to bin by varying direction
            // instead we will pass a pointer to its wide value
            // as an extra parameter.
            // If it is null, then the uniform value from noise options
@@ -7181,7 +7181,7 @@ LLVMGEN(llvm_gen_closure)
    llvm::Value* comp_wide_ptr = rop.ll.op_load(comp_wide_ptr_ptr);

    // This is an unrolled vector-width loop.  It was unrolled because it's
-    // was more expedient to write the for in C++ than IR, but if compiliation
+    // was more expedient to write the for in C++ than IR, but if compilation
    // of the extra ops becomes problematic, this choice can be revisited
    for (int lane_index = 0; lane_index < rop.vector_width(); ++lane_index) {
        llvm::Value* comp_ptr = rop.ll.op_extract(comp_wide_ptr, lane_index);
@@ -7706,7 +7706,7 @@ LLVMGEN(llvm_gen_pointcloud_write)
                                         TypeDesc::NOSEMANTICS, nattrs } };
    TypeSpec valuesArrayType { TypeDesc { TypeDesc::PTR, TypeDesc::SCALAR,
                                          TypeDesc::NOSEMANTICS, nattrs } };
-    llvm::Value* names  = rop.ll.op_alloca(rop.ll.type_string(), nattrs);
+    llvm::Value* names  = rop.ll.op_alloca(rop.ll.type_ustring(), nattrs);
    llvm::Value* types  = rop.ll.op_alloca(rop.ll.type_typedesc(), nattrs);
    llvm::Value* values = rop.ll.op_alloca(rop.ll.type_void_ptr(), nattrs);

@@ -8191,7 +8191,7 @@ LLVMGEN(llvm_gen_split)
            llvm::Value* wide_sep      = rop.ll.wide_constant("");
            llvm::Value* temp_wide_sep = rop.getOrAllocateTemp(
                TypeSpec(TypeDesc::STRING), false /*derivs*/,
-                false /*op_is_uniform*/, false /*forceBool*/, "wide seperator");
+                false /*op_is_uniform*/, false /*forceBool*/, "wide separator");
            rop.ll.op_unmasked_store(wide_sep, temp_wide_sep);
            args.push_back(rop.ll.void_ptr(temp_wide_sep));
        }
@@ -31,7 +31,7 @@
 #include "batched_backendllvm.h"
 #include "oslexec_pvt.h"

-// Create extrenal declarations for all built-in funcs we may call from LLVM
+// Create external declarations for all built-in funcs we may call from LLVM
 #define DECL(name, signature) extern "C" void name();
 #include "builtindecl.h"
 #undef DECL
@@ -760,11 +760,10 @@ BatchedBackendLLVM::llvm_type_batched_trace_options()
    std::vector<llvm::Type*> sg_types;

    // Uniform values of the batch
-    sg_types.push_back(ll.type_float());  // mindist
-    sg_types.push_back(ll.type_float());  // maxdist
-    sg_types.push_back(ll.type_int());    // shade
-    sg_types.push_back(
-        reinterpret_cast<llvm::Type*>(ll.type_string()));  // traceset
+    sg_types.push_back(ll.type_float());    // mindist
+    sg_types.push_back(ll.type_float());    // maxdist
+    sg_types.push_back(ll.type_int());      // shade
+    sg_types.push_back(ll.type_ustring());  // traceset

    m_llvm_type_batched_trace_options = ll.type_struct(sg_types, "TraceOptions",
                                                       false /*is_packed*/);
@@ -894,7 +893,7 @@ BatchedBackendLLVM::llvm_type_groupdata()
            fields.push_back(llvm_wide_type(ts));

            // Alignment
-            // TODO:  this isn't quite right, cant rely on batch size to == ISA SIMD requirements
+            // TODO:  this isn't quite right, can't rely on batch size to == ISA SIMD requirements
            size_t base_size = sym.typespec().is_closure_based()
                                   ? sizeof(void*)
                                   : sym.typespec().simpletype().basesize();
@@ -1126,7 +1125,7 @@ BatchedBackendLLVM::llvm_assign_initial_value(
                        // For ISA without a native mask (AVX & AVX2), this gather op will
                        // clamp the indices of masked off lanes to 0.
                        // This means the user data base pointer + sym_offset
-                        // must be dereferencable with a shadeindex of 0.
+                        // must be dereferenceable with a shadeindex of 0.
                        llvm::Value* wide_val
                            = ll.op_gather(userdata_sym_base_ptr, wide_index);

@@ -1252,8 +1251,7 @@ BatchedBackendLLVM::llvm_assign_initial_value(
                            data_type = ll.type_float();
                        } else if (elemtype.is_string()) {
                            ptr_type  = ll.type_ustring_ptr();
-                            data_type = static_cast<llvm::Type*>(
-                                ll.type_string());
+                            data_type = ll.type_ustring();
                        } else if (elemtype.is_int()) {
                            ptr_type  = ll.type_int_ptr();
                            data_type = ll.type_int();
@@ -2693,21 +2691,9 @@ BatchedBackendLLVM::run()
    // entry points, as well as for all the external functions that are
    // just declarations (not definitions) in the module (which we have
    // conveniently stashed in external_function_names).
-#if 0
-    std::vector<std::string> entry_function_names;
-    entry_function_names.push_back (ll.func_name(init_func));
-    for (int layer = 0; layer < nlayers; ++layer) {
-        // set_inst (layer);
-        llvm::Function* f = funcs[layer];
-        if (f && group().is_entry_layer(layer))
-            entry_function_names.push_back (ll.func_name(f));
-    }
-    ll.internalize_module_functions ("osl_", external_function_names, entry_function_names);
-#else
    std::unordered_set<llvm::Function*> external_functions;
    external_functions.insert(init_func);
    for (int layer = 0; layer < nlayers; ++layer) {
-        // set_inst (layer);
        llvm::Function* f = funcs[layer];
        // If we plan to call bitcode_string of a layer's function after optimization
        // it may not exist after optimization unless we treat it as external.
@@ -2716,7 +2702,6 @@ BatchedBackendLLVM::run()
        }
    }
    ll.prune_and_internalize_module(external_functions);
-#endif

    // Debug code to dump the pre-optimized bitcode to a file
    if (llvm_debug() >= 2 || shadingsys().llvm_output_bitcode()) {
@@ -117,7 +117,6 @@ BatchedRendererServices<WidthT>::resolve_udim_uniform(
    if (!texture_handle)
        texture_handle = texturesys()->get_texture_handle(filename,
                                                          texture_thread_info);
-#if OPENIMAGEIO_VERSION >= 20307
    if (texturesys()->is_udim(texture_handle)) {
        // Newer versions of the TextureSystem interface are able to determine the
        // specific UDIM tile we're using.
@@ -131,9 +130,9 @@ BatchedRendererServices<WidthT>::resolve_udim_uniform(
            udim_handle = texture_handle;
        }
        return udim_handle;
-    } else
-#endif
+    } else {
        return texture_handle;
+    }
 }


@@ -151,7 +150,6 @@ BatchedRendererServices<WidthT>::resolve_udim(
    if (!texture_handle)
        texture_handle = texturesys()->get_texture_handle(filename,
                                                          texture_thread_info);
-#if OPENIMAGEIO_VERSION >= 20307
    if (texturesys()->is_udim(texture_handle)) {
        // Newer versions of the TextureSystem interface are able to determine the
        // specific UDIM tile we're using.
@@ -167,9 +165,9 @@ BatchedRendererServices<WidthT>::resolve_udim(
            }
            wresult[l] = udim_handle;
        });
-    } else
-#endif
+    } else {
        assign_all(wresult, texture_handle);
+    }
 }


@@ -11,119 +11,118 @@


 #ifndef DECL
-#error Do not include this file unless DECL is defined
+#    error Do not include this file unless DECL is defined
 #endif



-#define NOISE_IMPL(name)                           \
-    DECL (osl_ ## name ## _ff,  "ff")              \
-    DECL (osl_ ## name ## _fff, "fff")             \
-    DECL (osl_ ## name ## _fv,  "fv")              \
-    DECL (osl_ ## name ## _fvf, "fvf")             \
-    DECL (osl_ ## name ## _vf,  "xvf")             \
-    DECL (osl_ ## name ## _vff, "xvff")            \
-    DECL (osl_ ## name ## _vv,  "xvv")             \
-    DECL (osl_ ## name ## _vvf, "xvvf")
+#define NOISE_IMPL(name)           \
+    DECL(osl_##name##_ff, "ff")    \
+    DECL(osl_##name##_fff, "fff")  \
+    DECL(osl_##name##_fv, "fv")    \
+    DECL(osl_##name##_fvf, "fvf")  \
+    DECL(osl_##name##_vf, "xvf")   \
+    DECL(osl_##name##_vff, "xvff") \
+    DECL(osl_##name##_vv, "xvv")   \
+    DECL(osl_##name##_vvf, "xvvf")

-#define NOISE_DERIV_IMPL(name)                     \
-    DECL (osl_ ## name ## _dfdf,   "xXX")          \
-    DECL (osl_ ## name ## _dfdff,  "xXXf")         \
-    DECL (osl_ ## name ## _dffdf,  "xXfX")         \
-    DECL (osl_ ## name ## _dfdfdf, "xXXX")         \
-    DECL (osl_ ## name ## _dfdv,   "xXv")          \
-    DECL (osl_ ## name ## _dfdvf,  "xXvf")         \
-    DECL (osl_ ## name ## _dfvdf,  "xXvX")         \
-    DECL (osl_ ## name ## _dfdvdf, "xXvX")         \
-    DECL (osl_ ## name ## _dvdf,   "xvX")          \
-    DECL (osl_ ## name ## _dvdff,  "xvXf")         \
-    DECL (osl_ ## name ## _dvfdf,  "xvfX")         \
-    DECL (osl_ ## name ## _dvdfdf, "xvXX")         \
-    DECL (osl_ ## name ## _dvdv,   "xvv")          \
-    DECL (osl_ ## name ## _dvdvf,  "xvvf")         \
-    DECL (osl_ ## name ## _dvvdf,  "xvvX")         \
-    DECL (osl_ ## name ## _dvdvdf, "xvvX")
+#define NOISE_DERIV_IMPL(name)        \
+    DECL(osl_##name##_dfdf, "xXX")    \
+    DECL(osl_##name##_dfdff, "xXXf")  \
+    DECL(osl_##name##_dffdf, "xXfX")  \
+    DECL(osl_##name##_dfdfdf, "xXXX") \
+    DECL(osl_##name##_dfdv, "xXv")    \
+    DECL(osl_##name##_dfdvf, "xXvf")  \
+    DECL(osl_##name##_dfvdf, "xXvX")  \
+    DECL(osl_##name##_dfdvdf, "xXvX") \
+    DECL(osl_##name##_dvdf, "xvX")    \
+    DECL(osl_##name##_dvdff, "xvXf")  \
+    DECL(osl_##name##_dvfdf, "xvfX")  \
+    DECL(osl_##name##_dvdfdf, "xvXX") \
+    DECL(osl_##name##_dvdv, "xvv")    \
+    DECL(osl_##name##_dvdvf, "xvvf")  \
+    DECL(osl_##name##_dvvdf, "xvvX")  \
+    DECL(osl_##name##_dvdvdf, "xvvX")

-#define GENERIC_NOISE_DERIV_IMPL(name)             \
-    DECL (osl_ ## name ## _dfdf,   "xsXXXX")       \
-    DECL (osl_ ## name ## _dfdfdf, "xsXXXXX")      \
-    DECL (osl_ ## name ## _dfdv,   "xsXXXX")       \
-    DECL (osl_ ## name ## _dfdvdf, "xsXXXXX")      \
-    DECL (osl_ ## name ## _dvdf,   "xsXXXX")       \
-    DECL (osl_ ## name ## _dvdfdf, "xsXXXXX")      \
-    DECL (osl_ ## name ## _dvdv,   "xsXXXX")       \
-    DECL (osl_ ## name ## _dvdvdf, "xsXXXXX")
+#define GENERIC_NOISE_DERIV_IMPL(name)   \
+    DECL(osl_##name##_dfdf, "xsXXXX")    \
+    DECL(osl_##name##_dfdfdf, "xsXXXXX") \
+    DECL(osl_##name##_dfdv, "xsXXXX")    \
+    DECL(osl_##name##_dfdvdf, "xsXXXXX") \
+    DECL(osl_##name##_dvdf, "xsXXXX")    \
+    DECL(osl_##name##_dvdfdf, "xsXXXXX") \
+    DECL(osl_##name##_dvdv, "xsXXXX")    \
+    DECL(osl_##name##_dvdvdf, "xsXXXXX")

-#define PNOISE_IMPL(name)                          \
-    DECL (osl_ ## name ## _fff,   "fff")           \
-    DECL (osl_ ## name ## _fffff, "fffff")         \
-    DECL (osl_ ## name ## _fvv,   "fvv")           \
-    DECL (osl_ ## name ## _fvfvf, "fvfvf")         \
-    DECL (osl_ ## name ## _vff,   "xvff")          \
-    DECL (osl_ ## name ## _vffff, "xvffff")        \
-    DECL (osl_ ## name ## _vvv,   "xvvv")          \
-    DECL (osl_ ## name ## _vvfvf, "xvvfvf")
+#define PNOISE_IMPL(name)              \
+    DECL(osl_##name##_fff, "fff")      \
+    DECL(osl_##name##_fffff, "fffff")  \
+    DECL(osl_##name##_fvv, "fvv")      \
+    DECL(osl_##name##_fvfvf, "fvfvf")  \
+    DECL(osl_##name##_vff, "xvff")     \
+    DECL(osl_##name##_vffff, "xvffff") \
+    DECL(osl_##name##_vvv, "xvvv")     \
+    DECL(osl_##name##_vvfvf, "xvvfvf")

-#define PNOISE_DERIV_IMPL(name)                    \
-    DECL (osl_ ## name ## _dfdff,    "xXXf")       \
-    DECL (osl_ ## name ## _dfdffff,  "xXXfff")     \
-    DECL (osl_ ## name ## _dffdfff,  "xXfXff")     \
-    DECL (osl_ ## name ## _dfdfdfff, "xXXXff")     \
-    DECL (osl_ ## name ## _dfdvv,    "xXXv")       \
-    DECL (osl_ ## name ## _dfdvfvf,  "xXvfvf")     \
-    DECL (osl_ ## name ## _dfvdfvf,  "xXvXvf")     \
-    DECL (osl_ ## name ## _dfdvdfvf, "xXvXvf")     \
-    DECL (osl_ ## name ## _dvdff,    "xvXf")       \
-    DECL (osl_ ## name ## _dvdffff,  "xvXfff")     \
-    DECL (osl_ ## name ## _dvfdfff,  "xvfXff")     \
-    DECL (osl_ ## name ## _dvdfdfff, "xvXXff")     \
-    DECL (osl_ ## name ## _dvdvv,    "xvvv")       \
-    DECL (osl_ ## name ## _dvdvfvf,  "xvvfvf")     \
-    DECL (osl_ ## name ## _dvvdfvf,  "xvvXvf")     \
-    DECL (osl_ ## name ## _dvdvdfvf, "xvvXvf")
+#define PNOISE_DERIV_IMPL(name)           \
+    DECL(osl_##name##_dfdff, "xXXf")      \
+    DECL(osl_##name##_dfdffff, "xXXfff")  \
+    DECL(osl_##name##_dffdfff, "xXfXff")  \
+    DECL(osl_##name##_dfdfdfff, "xXXXff") \
+    DECL(osl_##name##_dfdvv, "xXXv")      \
+    DECL(osl_##name##_dfdvfvf, "xXvfvf")  \
+    DECL(osl_##name##_dfvdfvf, "xXvXvf")  \
+    DECL(osl_##name##_dfdvdfvf, "xXvXvf") \
+    DECL(osl_##name##_dvdff, "xvXf")      \
+    DECL(osl_##name##_dvdffff, "xvXfff")  \
+    DECL(osl_##name##_dvfdfff, "xvfXff")  \
+    DECL(osl_##name##_dvdfdfff, "xvXXff") \
+    DECL(osl_##name##_dvdvv, "xvvv")      \
+    DECL(osl_##name##_dvdvfvf, "xvvfvf")  \
+    DECL(osl_##name##_dvvdfvf, "xvvXvf")  \
+    DECL(osl_##name##_dvdvdfvf, "xvvXvf")

-#define GENERIC_PNOISE_DERIV_IMPL(name)            \
-    DECL (osl_ ## name ## _dfdff,    "xsXXfXX")    \
-    DECL (osl_ ## name ## _dfdfdfff, "xsXXXffXX")  \
-    DECL (osl_ ## name ## _dfdvv,    "xsXXvXX")    \
-    DECL (osl_ ## name ## _dfdvdfvf, "xsXvXvfXX")  \
-    DECL (osl_ ## name ## _dvdff,    "xsvXfXX")    \
-    DECL (osl_ ## name ## _dvdfdfff, "xsvXXffXX")  \
-    DECL (osl_ ## name ## _dvdvv,    "xsvvvXX")    \
-    DECL (osl_ ## name ## _dvdvdfvf, "xsvvXvfXX")
+#define GENERIC_PNOISE_DERIV_IMPL(name)      \
+    DECL(osl_##name##_dfdff, "xsXXfXX")      \
+    DECL(osl_##name##_dfdfdfff, "xsXXXffXX") \
+    DECL(osl_##name##_dfdvv, "xsXXvXX")      \
+    DECL(osl_##name##_dfdvdfvf, "xsXvXvfXX") \
+    DECL(osl_##name##_dvdff, "xsvXfXX")      \
+    DECL(osl_##name##_dvdfdfff, "xsvXXffXX") \
+    DECL(osl_##name##_dvdvv, "xsvvvXX")      \
+    DECL(osl_##name##_dvdvdfvf, "xsvvXvfXX")

-#define UNARY_OP_IMPL(name)                        \
-    DECL (osl_ ## name ## _ff,  "ff")              \
-    DECL (osl_ ## name ## _dfdf, "xXX")            \
-    DECL (osl_ ## name ## _vv,  "xXX")             \
-    DECL (osl_ ## name ## _dvdv, "xXX")
+#define UNARY_OP_IMPL(name)        \
+    DECL(osl_##name##_ff, "ff")    \
+    DECL(osl_##name##_dfdf, "xXX") \
+    DECL(osl_##name##_vv, "xXX")   \
+    DECL(osl_##name##_dvdv, "xXX")

-#define BINARY_OP_IMPL(name)                       \
-    DECL (osl_ ## name ## _fff,    "fff")          \
-    DECL (osl_ ## name ## _dfdfdf, "xXXX")         \
-    DECL (osl_ ## name ## _dffdf,  "xXfX")         \
-    DECL (osl_ ## name ## _dfdff,  "xXXf")         \
-    DECL (osl_ ## name ## _vvv,    "xXXX")         \
-    DECL (osl_ ## name ## _dvdvdv, "xXXX")         \
-    DECL (osl_ ## name ## _dvvdv,  "xXXX")         \
-    DECL (osl_ ## name ## _dvdvv,  "xXXX")
+#define BINARY_OP_IMPL(name)          \
+    DECL(osl_##name##_fff, "fff")     \
+    DECL(osl_##name##_dfdfdf, "xXXX") \
+    DECL(osl_##name##_dffdf, "xXfX")  \
+    DECL(osl_##name##_dfdff, "xXXf")  \
+    DECL(osl_##name##_vvv, "xXXX")    \
+    DECL(osl_##name##_dvdvdv, "xXXX") \
+    DECL(osl_##name##_dvvdv, "xXXX")  \
+    DECL(osl_##name##_dvdvv, "xXXX")



-
-DECL (osl_add_closure_closure, "CXCC")
-DECL (osl_mul_closure_float, "CXCf")
-DECL (osl_mul_closure_color, "CXCc")
-DECL (osl_allocate_closure_component, "CXii")
-DECL (osl_allocate_weighted_closure_component, "CXiiX")
-DECL (osl_closure_to_string, "sXC")
-DECL (osl_format, "ss*")
-DECL (osl_printf, "xXs*")
-DECL (osl_fprintf, "xXss*")
-DECL (osl_error, "xXs*")
-DECL (osl_warning, "xXs*")
-DECL (osl_split, "isXsii")
-DECL (osl_incr_layers_executed, "xX")
+DECL(osl_add_closure_closure, "CXCC")
+DECL(osl_mul_closure_float, "CXCf")
+DECL(osl_mul_closure_color, "CXCc")
+DECL(osl_allocate_closure_component, "CXii")
+DECL(osl_allocate_weighted_closure_component, "CXiiX")
+DECL(osl_closure_to_string, "sXC")
+DECL(osl_format, "ss*")
+DECL(osl_printf, "xXs*")
+DECL(osl_fprintf, "xXss*")
+DECL(osl_error, "xXs*")
+DECL(osl_warning, "xXs*")
+DECL(osl_split, "isXsii")
+DECL(osl_incr_layers_executed, "xX")

 NOISE_IMPL(cellnoise)
 //NOISE_DERIV_IMPL(cellnoise)
@@ -151,70 +150,70 @@ PNOISE_IMPL(psnoise)
 PNOISE_DERIV_IMPL(psnoise)
 GENERIC_PNOISE_DERIV_IMPL(gaborpnoise)
 GENERIC_PNOISE_DERIV_IMPL(genericpnoise)
-DECL (osl_noiseparams_set_anisotropic, "xXi")
-DECL (osl_noiseparams_set_do_filter, "xXi")
-DECL (osl_noiseparams_set_direction, "xXv")
-DECL (osl_noiseparams_set_bandwidth, "xXf")
-DECL (osl_noiseparams_set_impulses, "xXf")
-DECL (osl_count_noise, "xX")
-DECL (osl_hash_ii,  "ii")
-DECL (osl_hash_if,  "if")
-DECL (osl_hash_iff, "iff")
-DECL (osl_hash_iv,  "iX")
-DECL (osl_hash_ivf, "iXf")
+DECL(osl_noiseparams_set_anisotropic, "xXi")
+DECL(osl_noiseparams_set_do_filter, "xXi")
+DECL(osl_noiseparams_set_direction, "xXv")
+DECL(osl_noiseparams_set_bandwidth, "xXf")
+DECL(osl_noiseparams_set_impulses, "xXf")
+DECL(osl_count_noise, "xX")
+DECL(osl_hash_ii, "ii")
+DECL(osl_hash_if, "if")
+DECL(osl_hash_iff, "iff")
+DECL(osl_hash_iv, "iX")
+DECL(osl_hash_ivf, "iXf")

-DECL (osl_spline_fff, "xXXXXii")
-DECL (osl_spline_dfdfdf, "xXXXXii")
-DECL (osl_spline_dfdff, "xXXXXii")
-DECL (osl_spline_dffdf, "xXXXXii")
-DECL (osl_spline_vfv, "xXXXXii")
-DECL (osl_spline_dvdfdv, "xXXXXii")
-DECL (osl_spline_dvdfv, "xXXXXii")
-DECL (osl_spline_dvfdv, "xXXXXii")
-DECL (osl_splineinverse_fff, "xXXXXii")
-DECL (osl_splineinverse_dfdfdf, "xXXXXii")
-DECL (osl_splineinverse_dfdff, "xXXXXii")
-DECL (osl_splineinverse_dffdf, "xXXXXii")
-DECL (osl_setmessage, "xXsLXisi")
-DECL (osl_getmessage, "iXssLXiisi")
-DECL (osl_pointcloud_search, "iXsXfiiXXii*")
-DECL (osl_pointcloud_get, "iXsXisLX")
-DECL (osl_pointcloud_write, "iXsXiXXX")
-DECL (osl_pointcloud_write_helper, "xXXXisLX")
-DECL (osl_blackbody_vf, "xXXf")
-DECL (osl_wavelength_color_vf, "xXXf")
-DECL (osl_luminance_fv, "xXXX")
-DECL (osl_luminance_dfdv, "xXXX")
-DECL (osl_prepend_color_from, "xXXs")
-DECL (osl_prepend_matrix_from, "iXXs")
-DECL (osl_get_matrix, "iXXs")
-DECL (osl_get_inverse_matrix, "iXXs")
-DECL (osl_transform_triple, "iXXiXiXXi")
-DECL (osl_transform_triple_nonlinear, "iXXiXiXXi")
-DECL (osl_transform_vmv, "xXXX")
-DECL (osl_transform_dvmdv, "xXXX")
-DECL (osl_transformv_vmv, "xXXX")
-DECL (osl_transformv_dvmdv, "xXXX")
-DECL (osl_transformn_vmv, "xXXX")
-DECL (osl_transformn_dvmdv, "xXXX")
-DECL (osl_transformc, "iXXiXiXX")
+DECL(osl_spline_fff, "xXXXXii")
+DECL(osl_spline_dfdfdf, "xXXXXii")
+DECL(osl_spline_dfdff, "xXXXXii")
+DECL(osl_spline_dffdf, "xXXXXii")
+DECL(osl_spline_vfv, "xXXXXii")
+DECL(osl_spline_dvdfdv, "xXXXXii")
+DECL(osl_spline_dvdfv, "xXXXXii")
+DECL(osl_spline_dvfdv, "xXXXXii")
+DECL(osl_splineinverse_fff, "xXXXXii")
+DECL(osl_splineinverse_dfdfdf, "xXXXXii")
+DECL(osl_splineinverse_dfdff, "xXXXXii")
+DECL(osl_splineinverse_dffdf, "xXXXXii")
+DECL(osl_setmessage, "xXsLXisi")
+DECL(osl_getmessage, "iXssLXiisi")
+DECL(osl_pointcloud_search, "iXsXfiiXXii*")
+DECL(osl_pointcloud_get, "iXsXisLX")
+DECL(osl_pointcloud_write, "iXsXiXXX")
+DECL(osl_pointcloud_write_helper, "xXXXisLX")
+DECL(osl_blackbody_vf, "xXXf")
+DECL(osl_wavelength_color_vf, "xXXf")
+DECL(osl_luminance_fv, "xXXX")
+DECL(osl_luminance_dfdv, "xXXX")
+DECL(osl_prepend_color_from, "xXXs")
+DECL(osl_prepend_matrix_from, "iXXs")
+DECL(osl_get_matrix, "iXXs")
+DECL(osl_get_inverse_matrix, "iXXs")
+DECL(osl_transform_triple, "iXXiXiXXi")
+DECL(osl_transform_triple_nonlinear, "iXXiXiXXi")
+DECL(osl_transform_vmv, "xXXX")
+DECL(osl_transform_dvmdv, "xXXX")
+DECL(osl_transformv_vmv, "xXXX")
+DECL(osl_transformv_dvmdv, "xXXX")
+DECL(osl_transformn_vmv, "xXXX")
+DECL(osl_transformn_dvmdv, "xXXX")
+DECL(osl_transformc, "iXXiXiXX")

-DECL (osl_dict_find_iis, "iXiX")
-DECL (osl_dict_find_iss, "iXXX")
-DECL (osl_dict_next, "iXi")
-DECL (osl_dict_value, "iXiXLX")
-DECL (osl_raytype_name, "iXs")
+DECL(osl_dict_find_iis, "iXiX")
+DECL(osl_dict_find_iss, "iXXX")
+DECL(osl_dict_next, "iXi")
+DECL(osl_dict_value, "iXiXLX")
+DECL(osl_raytype_name, "iXs")
 #ifdef OSL_LLVM_NO_BITCODE
-DECL (osl_range_check, "iiiXXXiXiXX")
+DECL(osl_range_check, "iiiXXXiXiXX")
 #endif
-DECL (osl_range_check_err, "iiiXXXiXiXX")
-DECL (osl_naninf_check, "xiXiXXiXiiX")
-DECL (osl_uninit_check, "xLXXXiXiXXiXiXii")
-DECL (osl_get_attribute, "iXiXXiiLX")
-DECL (osl_bind_interpolated_param, "iXXLiXiXiXi")
-DECL (osl_get_texture_options, "XX");
-DECL (osl_get_noise_options, "XX");
-DECL (osl_get_trace_options, "XX");
+DECL(osl_range_check_err, "iiiXXXiXiXX")
+DECL(osl_naninf_check, "xiXiXXiXiiX")
+DECL(osl_uninit_check, "xLXXXiXiXXiXiXii")
+DECL(osl_get_attribute, "iXiXXiiLX")
+DECL(osl_bind_interpolated_param, "iXXLiXiXiXi")
+DECL(osl_get_texture_options, "XX");
+DECL(osl_get_noise_options, "XX");
+DECL(osl_get_trace_options, "XX");


 // The following are defined inside llvm_ops.cpp. Only include these
@@ -231,17 +230,17 @@ UNARY_OP_IMPL(sinh)
 UNARY_OP_IMPL(cosh)
 UNARY_OP_IMPL(tanh)

-DECL (osl_safe_div_iii, "iii")
-DECL (osl_safe_div_fff, "fff")
-DECL (osl_safe_mod_iii, "iii")
-DECL (osl_sincos_fff, "xfXX")
-DECL (osl_sincos_dfdff, "xXXX")
-DECL (osl_sincos_dffdf, "xXXX")
-DECL (osl_sincos_dfdfdf, "xXXX")
-DECL (osl_sincos_vvv, "xXXX")
-DECL (osl_sincos_dvdvv, "xXXX")
-DECL (osl_sincos_dvvdv, "xXXX")
-DECL (osl_sincos_dvdvdv, "xXXX")
+DECL(osl_safe_div_iii, "iii")
+DECL(osl_safe_div_fff, "fff")
+DECL(osl_safe_mod_iii, "iii")
+DECL(osl_sincos_fff, "xfXX")
+DECL(osl_sincos_dfdff, "xXXX")
+DECL(osl_sincos_dffdf, "xXXX")
+DECL(osl_sincos_dfdfdf, "xXXX")
+DECL(osl_sincos_vvv, "xXXX")
+DECL(osl_sincos_dvdvv, "xXXX")
+DECL(osl_sincos_dvvdv, "xXXX")
+DECL(osl_sincos_dvdvdv, "xXXX")

 UNARY_OP_IMPL(log)
 UNARY_OP_IMPL(log2)
@@ -253,138 +252,138 @@ BINARY_OP_IMPL(pow)
 UNARY_OP_IMPL(erf)
 UNARY_OP_IMPL(erfc)

-DECL (osl_pow_vvf, "xXXf")
-DECL (osl_pow_dvdvdf, "xXXX")
-DECL (osl_pow_dvvdf, "xXXX")
-DECL (osl_pow_dvdvf, "xXXf")
+DECL(osl_pow_vvf, "xXXf")
+DECL(osl_pow_dvdvdf, "xXXX")
+DECL(osl_pow_dvvdf, "xXXX")
+DECL(osl_pow_dvdvf, "xXXf")

 UNARY_OP_IMPL(sqrt)
 UNARY_OP_IMPL(inversesqrt)
 UNARY_OP_IMPL(cbrt)

-DECL (osl_logb_ff, "ff")
-DECL (osl_logb_vv, "xXX")
+DECL(osl_logb_ff, "ff")
+DECL(osl_logb_vv, "xXX")

-DECL (osl_floor_ff, "ff")
-DECL (osl_floor_vv, "xXX")
-DECL (osl_ceil_ff, "ff")
-DECL (osl_ceil_vv, "xXX")
-DECL (osl_round_ff, "ff")
-DECL (osl_round_vv, "xXX")
-DECL (osl_trunc_ff, "ff")
-DECL (osl_trunc_vv, "xXX")
-DECL (osl_sign_ff, "ff")
-DECL (osl_sign_vv, "xXX")
-DECL (osl_step_fff, "fff")
-DECL (osl_step_vvv, "xXXX")
+DECL(osl_floor_ff, "ff")
+DECL(osl_floor_vv, "xXX")
+DECL(osl_ceil_ff, "ff")
+DECL(osl_ceil_vv, "xXX")
+DECL(osl_round_ff, "ff")
+DECL(osl_round_vv, "xXX")
+DECL(osl_trunc_ff, "ff")
+DECL(osl_trunc_vv, "xXX")
+DECL(osl_sign_ff, "ff")
+DECL(osl_sign_vv, "xXX")
+DECL(osl_step_fff, "fff")
+DECL(osl_step_vvv, "xXXX")

-DECL (osl_isnan_if, "if")
-DECL (osl_isinf_if, "if")
-DECL (osl_isfinite_if, "if")
-DECL (osl_abs_ii, "ii")
-DECL (osl_fabs_ii, "ii")
+DECL(osl_isnan_if, "if")
+DECL(osl_isinf_if, "if")
+DECL(osl_isfinite_if, "if")
+DECL(osl_abs_ii, "ii")
+DECL(osl_fabs_ii, "ii")

 UNARY_OP_IMPL(abs)
 UNARY_OP_IMPL(fabs)
 BINARY_OP_IMPL(fmod)

-DECL (osl_smoothstep_ffff, "ffff")
-DECL (osl_smoothstep_dfffdf, "xXffX")
-DECL (osl_smoothstep_dffdff, "xXfXf")
-DECL (osl_smoothstep_dffdfdf, "xXfXX")
-DECL (osl_smoothstep_dfdfff, "xXXff")
-DECL (osl_smoothstep_dfdffdf, "xXXfX")
-DECL (osl_smoothstep_dfdfdff, "xXXXf")
-DECL (osl_smoothstep_dfdfdfdf, "xXXXX")
+DECL(osl_smoothstep_ffff, "ffff")
+DECL(osl_smoothstep_dfffdf, "xXffX")
+DECL(osl_smoothstep_dffdff, "xXfXf")
+DECL(osl_smoothstep_dffdfdf, "xXfXX")
+DECL(osl_smoothstep_dfdfff, "xXXff")
+DECL(osl_smoothstep_dfdffdf, "xXXfX")
+DECL(osl_smoothstep_dfdfdff, "xXXXf")
+DECL(osl_smoothstep_dfdfdfdf, "xXXXX")

-DECL (osl_dot_fvv, "fXX")
-DECL (osl_dot_dfdvdv, "xXXX")
-DECL (osl_dot_dfdvv, "xXXX")
-DECL (osl_dot_dfvdv, "xXXX")
-DECL (osl_cross_vvv, "xXXX")
-DECL (osl_cross_dvdvdv, "xXXX")
-DECL (osl_cross_dvdvv, "xXXX")
-DECL (osl_cross_dvvdv, "xXXX")
-DECL (osl_length_fv, "fX")
-DECL (osl_length_dfdv, "xXX")
-DECL (osl_distance_fvv, "fXX")
-DECL (osl_distance_dfdvdv, "xXXX")
-DECL (osl_distance_dfdvv, "xXXX")
-DECL (osl_distance_dfvdv, "xXXX")
-DECL (osl_normalize_vv, "xXX")
-DECL (osl_normalize_dvdv, "xXX")
+DECL(osl_dot_fvv, "fXX")
+DECL(osl_dot_dfdvdv, "xXXX")
+DECL(osl_dot_dfdvv, "xXXX")
+DECL(osl_dot_dfvdv, "xXXX")
+DECL(osl_cross_vvv, "xXXX")
+DECL(osl_cross_dvdvdv, "xXXX")
+DECL(osl_cross_dvdvv, "xXXX")
+DECL(osl_cross_dvvdv, "xXXX")
+DECL(osl_length_fv, "fX")
+DECL(osl_length_dfdv, "xXX")
+DECL(osl_distance_fvv, "fXX")
+DECL(osl_distance_dfdvdv, "xXXX")
+DECL(osl_distance_dfdvv, "xXXX")
+DECL(osl_distance_dfvdv, "xXXX")
+DECL(osl_normalize_vv, "xXX")
+DECL(osl_normalize_dvdv, "xXX")
 #endif


-DECL (osl_mul_mmm, "xXXX")
-DECL (osl_mul_mmf, "xXXf")
+DECL(osl_mul_mmm, "xXXX")
+DECL(osl_mul_mmf, "xXXf")

-DECL (osl_div_mmm, "xXXX")
-DECL (osl_div_mmf, "xXXf")
-DECL (osl_div_mfm, "xXfX")
+DECL(osl_div_mmm, "xXXX")
+DECL(osl_div_mmf, "xXXf")
+DECL(osl_div_mfm, "xXfX")

-DECL (osl_get_from_to_matrix, "iXXss")
-DECL (osl_transpose_mm, "xXX")
-DECL (osl_determinant_fm, "fX")
+DECL(osl_get_from_to_matrix, "iXXss")
+DECL(osl_transpose_mm, "xXX")
+DECL(osl_determinant_fm, "fX")

-DECL (osl_concat_sss, "sss")
-DECL (osl_strlen_is, "is")
-DECL (osl_hash_is, "is")
-DECL (osl_getchar_isi, "isi");
-DECL (osl_startswith_iss, "iss")
-DECL (osl_endswith_iss, "iss")
-DECL (osl_stoi_is, "is")
-DECL (osl_stof_fs, "fs")
-DECL (osl_substr_ssii, "ssii")
-DECL (osl_regex_impl, "iXsXisi")
+DECL(osl_concat_sss, "sss")
+DECL(osl_strlen_is, "is")
+DECL(osl_hash_is, "is")
+DECL(osl_getchar_isi, "isi");
+DECL(osl_startswith_iss, "iss")
+DECL(osl_endswith_iss, "iss")
+DECL(osl_stoi_is, "is")
+DECL(osl_stof_fs, "fs")
+DECL(osl_substr_ssii, "ssii")
+DECL(osl_regex_impl, "iXsXisi")

 // Used by wide code generator, but are uniform calls
-DECL (osl_texture_decode_wrapmode, "iX");
-DECL (osl_texture_decode_interpmode, "iX");
+DECL(osl_texture_decode_wrapmode, "iX");
+DECL(osl_texture_decode_interpmode, "iX");

-DECL (osl_texture_set_firstchannel, "xXi")
-DECL (osl_texture_set_swrap, "xXs")
-DECL (osl_texture_set_twrap, "xXs")
-DECL (osl_texture_set_rwrap, "xXs")
-DECL (osl_texture_set_stwrap, "xXs")
-DECL (osl_texture_set_swrap_code, "xXi")
-DECL (osl_texture_set_twrap_code, "xXi")
-DECL (osl_texture_set_rwrap_code, "xXi")
-DECL (osl_texture_set_stwrap_code, "xXi")
-DECL (osl_texture_set_sblur, "xXf")
-DECL (osl_texture_set_tblur, "xXf")
-DECL (osl_texture_set_rblur, "xXf")
-DECL (osl_texture_set_stblur, "xXf")
-DECL (osl_texture_set_swidth, "xXf")
-DECL (osl_texture_set_twidth, "xXf")
-DECL (osl_texture_set_rwidth, "xXf")
-DECL (osl_texture_set_stwidth, "xXf")
-DECL (osl_texture_set_fill, "xXf")
-DECL (osl_texture_set_time, "xXf")
-DECL (osl_texture_set_interp, "xXs")
-DECL (osl_texture_set_interp_code, "xXi")
-DECL (osl_texture_set_subimage, "xXi")
-DECL (osl_texture_set_subimagename, "xXs")
-DECL (osl_texture_set_missingcolor_arena, "xXX")
-DECL (osl_texture_set_missingcolor_alpha, "xXif")
-DECL (osl_texture, "iXXXXffffffiXXXXXXX")
-DECL (osl_texture3d, "iXXXXXXXXiXXXXXXX")
-DECL (osl_environment, "iXXXXXXXiXXXXXXX")
-DECL (osl_get_textureinfo, "iXXXXiiiXX")
-DECL (osl_get_textureinfo_st, "iXXXffXiiiXX")
+DECL(osl_texture_set_firstchannel, "xXi")
+DECL(osl_texture_set_swrap, "xXs")
+DECL(osl_texture_set_twrap, "xXs")
+DECL(osl_texture_set_rwrap, "xXs")
+DECL(osl_texture_set_stwrap, "xXs")
+DECL(osl_texture_set_swrap_code, "xXi")
+DECL(osl_texture_set_twrap_code, "xXi")
+DECL(osl_texture_set_rwrap_code, "xXi")
+DECL(osl_texture_set_stwrap_code, "xXi")
+DECL(osl_texture_set_sblur, "xXf")
+DECL(osl_texture_set_tblur, "xXf")
+DECL(osl_texture_set_rblur, "xXf")
+DECL(osl_texture_set_stblur, "xXf")
+DECL(osl_texture_set_swidth, "xXf")
+DECL(osl_texture_set_twidth, "xXf")
+DECL(osl_texture_set_rwidth, "xXf")
+DECL(osl_texture_set_stwidth, "xXf")
+DECL(osl_texture_set_fill, "xXf")
+DECL(osl_texture_set_time, "xXf")
+DECL(osl_texture_set_interp, "xXs")
+DECL(osl_texture_set_interp_code, "xXi")
+DECL(osl_texture_set_subimage, "xXi")
+DECL(osl_texture_set_subimagename, "xXs")
+DECL(osl_texture_set_missingcolor_arena, "xXX")
+DECL(osl_texture_set_missingcolor_alpha, "xXif")
+DECL(osl_texture, "iXXXXffffffiXXXXXXX")
+DECL(osl_texture3d, "iXXXXXXXXiXXXXXXX")
+DECL(osl_environment, "iXXXXXXXiXXXXXXX")
+DECL(osl_get_textureinfo, "iXXXXiiiXX")
+DECL(osl_get_textureinfo_st, "iXXXffXiiiXX")

-DECL (osl_trace_set_mindist, "xXf")
-DECL (osl_trace_set_maxdist, "xXf")
-DECL (osl_trace_set_shade, "xXi")
-DECL (osl_trace_set_traceset, "xXs")
-DECL (osl_trace, "iXXXXXXXX")
+DECL(osl_trace_set_mindist, "xXf")
+DECL(osl_trace_set_maxdist, "xXf")
+DECL(osl_trace_set_shade, "xXi")
+DECL(osl_trace_set_traceset, "xXs")
+DECL(osl_trace, "iXXXXXXXX")

 #ifdef OSL_LLVM_NO_BITCODE
-DECL (osl_calculatenormal, "xXXX")
-DECL (osl_area, "fX")
-DECL (osl_filterwidth_fdf, "fX")
-DECL (osl_filterwidth_vdv, "xXX")
-DECL (osl_raytype_bit, "iXi")
+DECL(osl_calculatenormal, "xXXX")
+DECL(osl_area, "fX")
+DECL(osl_filterwidth_fdf, "fX")
+DECL(osl_filterwidth_vdv, "xXX")
+DECL(osl_raytype_bit, "iXi")
 #endif


@@ -253,7 +253,7 @@ DECL(__OSL_MASKED_OP2(hash, Wi, Wv), "xXXi")
 DECL(__OSL_MASKED_OP3(hash, Wi, Wv, Wf), "xXXXi")

 // Need wide for combinations of the 3 parameters allowed to be uniform
-// caveat, some combos are unreachable/uneeded
+// caveat, some combos are unreachable/unneeded
 // When result has a derivative, there is
 // no "easy" have a input parameter be non-derivative based on code
 // analysis promoting all inputs to be derivative base.
@@ -381,7 +381,7 @@ DECL(__OSL_OP(get_attribute_uniform), "iXiXXiiXX")
 // TODO:  shouldn't bind_interpolated_param be MASKED?  change name to reflect
 DECL(__OSL_OP(bind_interpolated_param), "iXXLiXiXiXii")

-//DECL (osl_get_texture_options, "XX") // uneeded
+//DECL (osl_get_texture_options, "XX") // unneeded
 DECL(__OSL_OP(get_noise_options), "XX")

 // The following are defined inside llvm_ops.cpp. Only include these
@@ -8,9 +8,10 @@

 #include <OpenImageIO/sysutil.h>

-#include <OSL/oslconfig.h>
-#include <OSL/oslclosure.h>
 #include <OSL/genclosure.h>
+#include <OSL/oslclosure.h>
+#include <OSL/oslconfig.h>
+
 #include "oslexec_pvt.h"


@@ -4,8 +4,8 @@

 #pragma once

-#include <vector>
 #include <list>
+#include <vector>

 #include <OpenImageIO/thread.h>

@@ -27,48 +27,48 @@ namespace pvt {
 /// a new allocation tries to fit it in one of the allocated blocks, but
 /// if it won't fit anywhere, it makes a new block and adds it to the
 /// head of the list.
-template<class T>
-class ConstantPool {
+template<class T> class ConstantPool {
 public:
    /// Allocate a new pool of T's.  The quanta, if supplied, is the
    /// number of T's to malloc at a time.
-    ConstantPool (size_t quanta = 1000000) : m_quanta(quanta), m_total(0) { }
+    ConstantPool(size_t quanta = 1000000) : m_quanta(quanta), m_total(0) {}

-    ~ConstantPool () { }
+    ~ConstantPool() {}

    /// Allocate space enough for n T's, and return a pointer to the
    /// start of that space.
-    T * alloc (size_t n) {
-        OIIO::lock_guard lock (m_mutex);
+    T* alloc(size_t n)
+    {
+        OIIO::lock_guard lock(m_mutex);
        // Check each block in the block list to see if it has enough space
        for (auto&& block : m_block_list) {
            size_t s = block.size();
-            if ((s+n) <= block.capacity()) {
+            if ((s + n) <= block.capacity()) {
                // Enough space in this block.  Use it.
-                block.resize (s+n);
+                block.resize(s + n);
                return &block[s];
            }
        }
        // If we got here, there were no mini-blocks in the list with enough
        // space.  Make a new one.
-        m_block_list.push_front (block_t());
-        block_t &block (m_block_list.front());
-        size_t s = std::max (m_quanta, n);
-        block.reserve (s);
+        m_block_list.push_front(block_t());
+        block_t& block(m_block_list.front());
+        size_t s = std::max(m_quanta, n);
+        block.reserve(s);
        m_total += s * sizeof(T);
-        block.resize (n);
+        block.resize(n);
        return &block[0];
    }

 private:
    typedef std::vector<T> block_t;   ///< Type of block
    std::list<block_t> m_block_list;  ///< List of memory blocks
-    size_t m_quanta;   ///< How big each memory block is (in T's, not bytes)
-    size_t m_total;    ///< Total memory allocated (bytes!)
+    size_t m_quanta;      ///< How big each memory block is (in T's, not bytes)
+    size_t m_total;       ///< Total memory allocated (bytes!)
    OIIO::mutex m_mutex;  ///< Thread-safe lock
 };



-}; // namespace OSL::pvt
+};  // namespace pvt
 OSL_NAMESPACE_EXIT
@@ -2410,7 +2410,7 @@ DECLFOLDER(constfold_gettextureinfo)
        ustring filename = Filename.get_string();
        ustring dataname = Dataname.get_string();
        TypeDesc t       = Data.typespec().simpletype();
-        void* mydata     = OIIO_ALLOCA(char, t.size());
+        void* mydata     = OSL_ALLOCA(char, t.size());
        // FIXME(ptex) -- exclude folding of ptex, since these things
        // can vary per face.
        ustring errormessage;
@@ -2424,7 +2424,7 @@ DECLFOLDER(constfold_gettextureinfo)
        //       assign data [retrieved values]
        //       assign result 1
        // into this for failure:
-        //       error "%s" errormesage
+        //       error "%s" errormessage
        //       assign result 0
        if (result) {
            int oldresultarg = rop.inst()->args()[op.firstarg() + 0];
@@ -2776,7 +2776,7 @@ DECLFOLDER(constfold_pointcloud_get)
        return 0;

    // Must transfer to size_t array
-    size_t* indices = OIIO_ALLOCA(size_t, count);
+    size_t* indices = OSL_ALLOCA(size_t, count);
    for (int i = 0; i < count; ++i)
        indices[i] = Indices.get_int(i);

@@ -128,9 +128,10 @@ ShadingContext::execute_init(ShaderGroup& sgroup, int shadeindex,
        RunLLVMGroupFunc run_func = sgroup.llvm_compiled_init();
        if (!run_func)
            return false;
-        ssg.context  = this;
-        ssg.renderer = renderer();
-        ssg.Ci       = NULL;
+        ssg.context             = this;
+        ssg.shadingStateUniform = &(shadingsys().m_shading_state_uniform);
+        ssg.renderer            = renderer();
+        ssg.Ci                  = NULL;
        run_func(&ssg, m_heap.get(), userdata_base_ptr, output_base_ptr,
                 shadeindex);
    }
@@ -115,7 +115,7 @@ private:
    };

    typedef std::unordered_map<Query, QueryResult, QueryHash> QueryMap;
-    typedef std::unordered_map<ustring, int, ustringHash> DocMap;
+    typedef std::unordered_map<ustring, int> DocMap;

    ShadingContext* m_context;  // back-pointer to shading context

@@ -301,7 +301,7 @@ ShaderInstance::parameters(const ParamValueList& params)
            if (paramtype.arraylen < 0) {
                // An array of definite size was supplied to a parameter
                // that was an array of indefinite size. Magic! The trick
-                // here is that we need to allocate paramter space at the
+                // here is that we need to allocate parameter space at the
                // END of the ordinary param storage, since when we assigned
                // data offsets to each parameter, we didn't know the length
                // needed to allocate this param in its proper spot.
@@ -330,12 +330,19 @@ ShaderInstance::parameters(const ParamValueList& params)
                // definite length in subsequent rerenders. Don't do that.
            } else {
                // If the instance value is the same as the master's default,
-                // just skip the parameter, let it "keep" the default.
-                // Note that this can't/shouldn't happen for the indefinite-
-                // sized array case, which is why we have it in the 'else'
-                // clause of that test.
+                // just skip the parameter, let it "keep" the default by
+                // marking the source as DefaultVal.
+                //
+                // N.B. Beware the situation where it has init ops, and so the
+                // "default value" slot only coincidentally has the same value
+                // as the instance value.  We can't mark it as DefaultVal in
+                // that case, because the init ops need to be run.
+                //
+                // Note that this case also can't/shouldn't happen for the
+                // indefinite- sized array case, which is why we have it in
+                // the 'else' clause of that test.
                void* defaultdata = m_master->param_default_storage(i);
-                if (lockgeom
+                if (lockgeom && !sm->has_init_ops()
                    && memcmp(defaultdata, data, valuetype.size()) == 0) {
                    // Must reset valuesource to default, in case the parameter
                    // was set already, and now is being changed back to default.
@@ -459,11 +466,9 @@ ShaderInstance::evaluate_writes_globals_and_userdata_params()
    // userdata_params as accurately as we can based on what we know from
    // the symbol overrides. This is very important to get instance merging
    // working correctly.
-    int p = 0;
    for (auto&& s : m_instoverrides) {
        if (!s.lockgeom())
            userdata_params(true);
-        ++p;
    }
 }

@@ -6,14 +6,14 @@

 #include <OSL/oslconfig.h>

-#include <llvm/Pass.h>
 #include <llvm/IR/Constants.h>
-#include <llvm/IR/IRBuilder.h>
-#include <llvm/IR/Function.h>
-#include <llvm/IR/Type.h>
 #include <llvm/IR/DerivedTypes.h>
+#include <llvm/IR/Function.h>
+#include <llvm/IR/IRBuilder.h>
 #include <llvm/IR/Instruction.h>
 #include <llvm/IR/Instructions.h>
+#include <llvm/IR/Type.h>
+#include <llvm/Pass.h>
 #include <llvm/Support/Casting.h>

 OSL_NAMESPACE_ENTER
@@ -37,16 +37,18 @@ namespace {
 // how instruction lowering happens then this pass may not be necessary.
 // Also if future LLVM version takes on the work of this optimization pass,
 // then it may be removed.
-template <int WidthT>
+template<int WidthT>
 class PreventBitMasksFromBeingLiveinsToBasicBlocks final
-    : public llvm::FunctionPass
-{
-    typedef llvm::IRBuilder<llvm::ConstantFolder, llvm::IRBuilderDefaultInserter> IRBuilder;
+    : public llvm::FunctionPass {
+    typedef llvm::IRBuilder<llvm::ConstantFolder, llvm::IRBuilderDefaultInserter>
+        IRBuilder;
    llvm::Type* m_llvm_mask_type;
    llvm::Type* m_native_mask_type;
    llvm::Constant* m_wide_zero_initializer;
-    std::unordered_map<llvm::Instruction*, llvm::Value*> m_native_mask_by_producing_inst;
-    std::unordered_map<llvm::Value*, llvm::Value*> m_llvm_mask_from_livein_by_native_mask;
+    std::unordered_map<llvm::Instruction*, llvm::Value*>
+        m_native_mask_by_producing_inst;
+    std::unordered_map<llvm::Value*, llvm::Value*>
+        m_llvm_mask_from_livein_by_native_mask;
    std::vector<llvm::Instruction*> m_phiNodesWithNativeMasks;

 public:
@@ -57,11 +59,13 @@ public:
        , m_llvm_mask_type(nullptr)
        , m_native_mask_type(nullptr)
        , m_wide_zero_initializer(nullptr)
-    {}
+    {
+    }

-    bool doInitialization (llvm::Module &M) override {
-        llvm::Type* llvm_type_bool = llvm::Type::getInt1Ty (M.getContext());
-        llvm::Type* llvm_type_int32 = llvm::Type::getInt32Ty (M.getContext());
+    bool doInitialization(llvm::Module& M) override
+    {
+        llvm::Type* llvm_type_bool  = llvm::Type::getInt1Ty(M.getContext());
+        llvm::Type* llvm_type_int32 = llvm::Type::getInt32Ty(M.getContext());

 #if OSL_LLVM_VERSION >= 110
        m_llvm_mask_type = llvm::FixedVectorType::get(llvm_type_bool, WidthT);
@@ -72,39 +76,45 @@ public:
        // what the native mask type should be.  And if necessary maintain a
        // 16 bit and 32 bit native mask representation to be passed as a
        // livein.
-        m_native_mask_type = llvm::FixedVectorType::get(llvm_type_int32, WidthT);
-#  if OSL_LLVM_VERSION >= 112
-        m_wide_zero_initializer = llvm::ConstantDataVector::getSplat(WidthT,
-                llvm::ConstantInt::get(M.getContext(), llvm::APInt(32,0)));
-#  else
+        m_native_mask_type = llvm::FixedVectorType::get(llvm_type_int32,
+                                                        WidthT);
+#    if OSL_LLVM_VERSION >= 112
+        m_wide_zero_initializer = llvm::ConstantDataVector::getSplat(
+            WidthT, llvm::ConstantInt::get(M.getContext(), llvm::APInt(32, 0)));
+#    else
        m_wide_zero_initializer = llvm::ConstantVector::getSplat(
-                                    llvm::ElementCount(WidthT, false),
-                                    llvm::ConstantInt::get (M.getContext(), llvm::APInt(32,0)));
-#  endif
+            llvm::ElementCount(WidthT, false),
+            llvm::ConstantInt::get(M.getContext(), llvm::APInt(32, 0)));
+#    endif
 #else
-        m_llvm_mask_type = llvm::VectorType::get(llvm_type_bool, WidthT);
+        m_llvm_mask_type   = llvm::VectorType::get(llvm_type_bool, WidthT);
        m_native_mask_type = llvm::VectorType::get(llvm_type_int32, WidthT);
-        m_wide_zero_initializer = llvm::ConstantVector::getSplat(WidthT, llvm::ConstantInt::get (M.getContext(), llvm::APInt(32,0)));
+        m_wide_zero_initializer = llvm::ConstantVector::getSplat(
+            WidthT, llvm::ConstantInt::get(M.getContext(), llvm::APInt(32, 0)));
 #endif
-        return false; // I don't think we modified the module
+        return false;  // I don't think we modified the module
    }

-    bool runOnFunction(llvm::Function &F) override {
-        OSL_DEV_ONLY(llvm::errs() << ">>>>>>>>>>>>>>>>>>PreventBitMasksFromBeingLiveinsToBasicBlocks<" << WidthT << ">:");
+    bool runOnFunction(llvm::Function& F) override
+    {
+        OSL_DEV_ONLY(
+            llvm::errs()
+            << ">>>>>>>>>>>>>>>>>>PreventBitMasksFromBeingLiveinsToBasicBlocks<"
+            << WidthT << ">:");
        OSL_DEV_ONLY(llvm::errs().write_escaped(F.getName()) << '\n');

        m_native_mask_by_producing_inst.clear();

        bool changed = false;

-        for (llvm::BasicBlock &bb : F) {
+        for (llvm::BasicBlock& bb : F) {
            OSL_DEV_ONLY(llvm::errs() << ">>>>>>>>>Basic Block: ");
            OSL_DEV_ONLY(llvm::errs().write_escaped(bb.getName()) << '\n');

            m_phiNodesWithNativeMasks.clear();
            m_llvm_mask_from_livein_by_native_mask.clear();

-            for (llvm::Instruction &inst : bb) {
+            for (llvm::Instruction& inst : bb) {
                // We could possibly identify all the instruction types that
                // we "think" could be using a mask, but feel just looking
                // at the operand types is cheaper.
@@ -120,21 +130,24 @@ public:
                    // back to llvm masks.
                    IRBuilder builder(&bb);
                    builder.SetInsertPoint(&inst);
-                    for (llvm::Instruction* phiNodeWithNativeMask : m_phiNodesWithNativeMasks) {
-                        llvm::Value* llvm_mask = builder.CreateICmpSLT (phiNodeWithNativeMask,m_wide_zero_initializer);
+                    for (llvm::Instruction* phiNodeWithNativeMask :
+                         m_phiNodesWithNativeMasks) {
+                        llvm::Value* llvm_mask
+                            = builder.CreateICmpSLT(phiNodeWithNativeMask,
+                                                    m_wide_zero_initializer);
                        // Now we need to replace all uses of phiNodeWithNativeMask with our converted llvm mask
                        auto use_iter = phiNodeWithNativeMask->use_begin(),
-                                           use_end = phiNodeWithNativeMask->use_end();
-                        for (; use_iter != use_end;)
-                        {
-                            llvm::Use &use = *use_iter;
+                             use_end  = phiNodeWithNativeMask->use_end();
+                        for (; use_iter != use_end;) {
+                            llvm::Use& use = *use_iter;
                            ++use_iter;
-                            auto *user = llvm::dyn_cast<llvm::Instruction>(use.getUser());
-                            // We need to skip the use wher the user is our conversion to a llvm_mask
+                            auto* user = llvm::dyn_cast<llvm::Instruction>(
+                                use.getUser());
+                            // We need to skip the use where the user is our conversion to a llvm_mask
                            if (user && user == llvm_mask)
                                continue;
                            use.set(llvm_mask);
-                       }
+                        }
                    }

                    m_phiNodesWithNativeMasks.clear();
@@ -151,7 +164,8 @@ public:
                bool phiNodeHadOperandReplacedInLastPassOverOperands;

                OSL_DEV_ONLY(llvm::errs() << ">>>>Op: ");
-                OSL_DEV_ONLY(llvm::errs().write_escaped(inst.getOpcodeName()) << '\n');
+                OSL_DEV_ONLY(llvm::errs().write_escaped(inst.getOpcodeName())
+                             << '\n');

                // As we may have multiple operands that need to be replaced
                // for a phi node, once we discover a qualifying replacement
@@ -167,25 +181,32 @@ public:

                    // As we could be replacing operands, but not the number of operands
                    // we use index vs. iterator based loop construct
-                    for (unsigned operand_i = 0, num_operands = inst.getNumOperands();
-                         operand_i != num_operands;
-                         ++operand_i) {
-
-                        llvm::Value *op_val = inst.getOperand (operand_i);
+                    for (unsigned operand_i    = 0,
+                                  num_operands = inst.getNumOperands();
+                         operand_i != num_operands; ++operand_i) {
+                        llvm::Value* op_val = inst.getOperand(operand_i);
                        if (op_val != nullptr) {
-                            llvm::Type* op_type =  op_val->getType();
+                            llvm::Type* op_type = op_val->getType();
                            if (op_type == m_llvm_mask_type) {
                                // See if the value came from an instruction
-                                auto *producing_instr = llvm::dyn_cast<llvm::Instruction>(op_val);
+                                auto* producing_instr
+                                    = llvm::dyn_cast<llvm::Instruction>(op_val);
                                if (producing_instr != nullptr) {
-                                    llvm::BasicBlock* producing_bb = producing_instr->getParent();
+                                    llvm::BasicBlock* producing_bb
+                                        = producing_instr->getParent();

                                    // If a phi node needs to have one operand promoted,
                                    // it will need all operands promoted even if they were
                                    // produced in the same basic block, so the types match
-                                    if (phiNodeHadOperandReplaced || producing_bb != &bb) {
-                                        OSL_DEV_ONLY(llvm::errs() << "Uses llvm mask <" <<WidthT << "xi1> from different Basic Block: ");
-                                        OSL_DEV_ONLY(llvm::errs().write_escaped(op_val->getName()) << '\n');
+                                    if (phiNodeHadOperandReplaced
+                                        || producing_bb != &bb) {
+                                        OSL_DEV_ONLY(
+                                            llvm::errs()
+                                            << "Uses llvm mask <" << WidthT
+                                            << "xi1> from different Basic Block: ");
+                                        OSL_DEV_ONLY(llvm::errs().write_escaped(
+                                                         op_val->getName())
+                                                     << '\n');
                                        changed = true;

                                        // We can't let an unrepresentable data type of <16xi1>, <8xi1> or <4xi1>
@@ -195,37 +216,68 @@ public:
                                        llvm::Value* native_mask = nullptr;
                                        {
                                            // We may have already created a native mask for this instruction
-                                            auto search_result = m_native_mask_by_producing_inst.find(producing_instr);
-                                            if (search_result == m_native_mask_by_producing_inst.end()) {
-
+                                            auto search_result
+                                                = m_native_mask_by_producing_inst
+                                                      .find(producing_instr);
+                                            if (search_result
+                                                == m_native_mask_by_producing_inst
+                                                       .end()) {
                                                // Scan producing basic block for an existing sign extend instruction
                                                // for the producing_instr.  As the existence of a basic block means
                                                // we most likely have a branch higher up whose test required a sign extend
-                                                for (llvm::Instruction &other_inst : *producing_bb) {
-                                                    auto *existing_sign_ext = llvm::dyn_cast<llvm::SExtInst>(&other_inst);
+                                                for (llvm::Instruction&
+                                                         other_inst :
+                                                     *producing_bb) {
+                                                    auto* existing_sign_ext
+                                                        = llvm::dyn_cast<
+                                                            llvm::SExtInst>(
+                                                            &other_inst);
                                                    if (existing_sign_ext) {
-                                                        OSL_ASSERT(existing_sign_ext->getNumOperands() == 1);
-                                                        llvm::Value *existing_operand = existing_sign_ext->getOperand (0);
-                                                        if (existing_operand == producing_instr) {
-                                                            OSL_ASSERT(existing_sign_ext->getType() == m_native_mask_type);
-                                                            OSL_DEV_ONLY(llvm::errs() << "Using existing sign ext in producing bb\n");
+                                                        OSL_ASSERT(
+                                                            existing_sign_ext
+                                                                ->getNumOperands()
+                                                            == 1);
+                                                        llvm::Value*
+                                                            existing_operand
+                                                            = existing_sign_ext
+                                                                  ->getOperand(
+                                                                      0);
+                                                        if (existing_operand
+                                                            == producing_instr) {
+                                                            OSL_ASSERT(
+                                                                existing_sign_ext
+                                                                    ->getType()
+                                                                == m_native_mask_type);
+                                                            OSL_DEV_ONLY(
+                                                                llvm::errs()
+                                                                << "Using existing sign ext in producing bb\n");

-                                                            native_mask = existing_sign_ext;
+                                                            native_mask
+                                                                = existing_sign_ext;
                                                            break;
                                                        }
                                                    }
                                                }
                                                if (nullptr == native_mask) {
-                                                    IRBuilder builder(producing_bb);
-                                                    builder.SetInsertPoint(&producing_bb->back());
+                                                    IRBuilder builder(
+                                                        producing_bb);
+                                                    builder.SetInsertPoint(
+                                                        &producing_bb->back());

-                                                    native_mask = builder.CreateSExt(producing_instr, m_native_mask_type);
+                                                    native_mask
+                                                        = builder.CreateSExt(
+                                                            producing_instr,
+                                                            m_native_mask_type);
                                                    // NOTE: we did add instructions, but not to the BB we are currently
                                                    // iterating over instructions
                                                }
-                                                m_native_mask_by_producing_inst.insert(std::make_pair(producing_instr,native_mask));
+                                                m_native_mask_by_producing_inst
+                                                    .insert(std::make_pair(
+                                                        producing_instr,
+                                                        native_mask));
                                            } else {
-                                                native_mask = search_result->second;
+                                                native_mask
+                                                    = search_result->second;
                                            }
                                        }
                                        OSL_ASSERT(native_mask);
@@ -239,16 +291,26 @@ public:
                                            llvm::Value* llvm_mask = nullptr;
                                            {
                                                // We may have already created a llvm mask in this basic block for the native mask
-                                                auto search_result = m_llvm_mask_from_livein_by_native_mask.find(native_mask);
-                                                if (search_result == m_llvm_mask_from_livein_by_native_mask.end()) {
+                                                auto search_result
+                                                    = m_llvm_mask_from_livein_by_native_mask
+                                                          .find(native_mask);
+                                                if (search_result
+                                                    == m_llvm_mask_from_livein_by_native_mask
+                                                           .end()) {
                                                    // Insert the conversion from native to llvm mask
                                                    // somewhere before the 1st instruction that needs to use it in this basic block
                                                    IRBuilder builder(&bb);
-                                                    builder.SetInsertPoint(&inst);
+                                                    builder.SetInsertPoint(
+                                                        &inst);

-                                                    llvm_mask = builder.CreateICmpSLT (native_mask,m_wide_zero_initializer);
+                                                    llvm_mask = builder.CreateICmpSLT(
+                                                        native_mask,
+                                                        m_wide_zero_initializer);

-                                                    m_llvm_mask_from_livein_by_native_mask.insert(std::make_pair(native_mask,llvm_mask));
+                                                    m_llvm_mask_from_livein_by_native_mask
+                                                        .insert(std::make_pair(
+                                                            native_mask,
+                                                            llvm_mask));

                                                    // NOTE: we did add instructions to this bb,
                                                    // but the should be inserted before the current inst.
@@ -257,12 +319,13 @@ public:
                                                    // iterator should not invalidate the iterator

                                                } else {
-                                                    llvm_mask = search_result->second;
+                                                    llvm_mask
+                                                        = search_result->second;
                                                }
                                            }
-                                            inst.setOperand (operand_i, llvm_mask);
+                                            inst.setOperand(operand_i,
+                                                            llvm_mask);
                                        } else {
-
                                            // Phi nodes have to exist at the top of a basic block which precludes
                                            // our basic algorithm from just converting the native mask back to a llvm mask
                                            // before the phi node.
@@ -270,24 +333,37 @@ public:
                                            // and insert an instruction to convert it after the block of phi nodes
                                            // However, this now means we will need to find all uses of the phi node's
                                            // result and replace them with our converted llvm_mask
-                                            inst.setOperand (operand_i, native_mask);
+                                            inst.setOperand(operand_i,
+                                                            native_mask);
                                            phiNodeHadOperandReplaced = true;
-                                            phiNodeHadOperandReplacedInLastPassOverOperands = true;
+                                            phiNodeHadOperandReplacedInLastPassOverOperands
+                                                = true;
                                        }
                                        // We should be able to continue iterating over the rest of the operands
                                    }
                                } else if (phiNodeHadOperandReplaced) {
-                                    auto *constant = llvm::dyn_cast<llvm::Constant>(op_val);
+                                    auto* constant
+                                        = llvm::dyn_cast<llvm::Constant>(
+                                            op_val);
                                    if (constant) {
-                                        OSL_DEV_ONLY(llvm::errs() << "Uses constant llvm mask <" <<WidthT << "xi1> \n");
+                                        OSL_DEV_ONLY(
+                                            llvm::errs()
+                                            << "Uses constant llvm mask <"
+                                            << WidthT << "xi1> \n");

                                        // Should handle promoting whatever the constant value is (most likely zeroinitializer)
                                        llvm::ConstantFolder Folder;
-                                        auto *signExtConstant = Folder.CreateCast(llvm::Instruction::SExt, constant, m_native_mask_type);
+                                        auto* signExtConstant
+                                            = Folder.CreateCast(
+                                                llvm::Instruction::SExt,
+                                                constant, m_native_mask_type);

-                                        inst.setOperand (operand_i, signExtConstant);
+                                        inst.setOperand(operand_i,
+                                                        signExtConstant);
                                    } else {
-                                        OSL_ASSERT(0 && "Unhandled/Unexpected llvm mask type");
+                                        OSL_ASSERT(
+                                            0
+                                            && "Unhandled/Unexpected llvm mask type");
                                    }
                                }
                            }
@@ -321,11 +397,9 @@ public:
 // including this file will need its own static members defined. LLVM will
 // assign IDs when they get registered, so this initialization value is not
 // important.
-template<>
-char PreventBitMasksFromBeingLiveinsToBasicBlocks<8>::ID = 0;
+template<> char PreventBitMasksFromBeingLiveinsToBasicBlocks<8>::ID = 0;

-template<>
-char PreventBitMasksFromBeingLiveinsToBasicBlocks<16>::ID = 0;
+template<> char PreventBitMasksFromBeingLiveinsToBasicBlocks<16>::ID = 0;

 }  // end of anonymous namespace

@@ -72,7 +72,7 @@ private:
    int m_oso_major, m_oso_minor;     ///< oso file format version
    int m_sym_default_index;          ///< Next sym default value to fill in
    bool m_errors;                    ///< Did we hit any errors?
-    typedef std::unordered_map<ustring, int, ustringHash> UstringIntMap;
+    typedef std::unordered_map<ustring, int> UstringIntMap;
    UstringIntMap m_symmap;  ///< map sym name to index
 };

@@ -12,23 +12,23 @@ OSL_NAMESPACE_ENTER

 static ustring udot(".");

-Parser::Parser(const std::vector<ustring> *user_events,
-               const std::vector<ustring> *user_scatterings)
+Parser::Parser(const std::vector<ustring>* user_events,
+               const std::vector<ustring>* user_scatterings)
 {
    m_ingroup = false;
-    m_error = "";
+    m_error   = "";

-    m_label_position[Labels::CAMERA]      = 0;
-    m_label_position[Labels::LIGHT]        = 0;
+    m_label_position[Labels::CAMERA]     = 0;
+    m_label_position[Labels::LIGHT]      = 0;
    m_label_position[Labels::BACKGROUND] = 0;
-    m_label_position[Labels::VOLUME]      = 0;
-    m_label_position[Labels::TRANSMIT]    = 0;
-    m_label_position[Labels::REFLECT]     = 0;
-    m_label_position[Labels::OBJECT]      = 0;
-    m_label_position[Labels::DIFFUSE]     = 1;
-    m_label_position[Labels::GLOSSY]      = 1;
-    m_label_position[Labels::SINGULAR]    = 1;
-    m_label_position[Labels::STRAIGHT]    = 1;
+    m_label_position[Labels::VOLUME]     = 0;
+    m_label_position[Labels::TRANSMIT]   = 0;
+    m_label_position[Labels::REFLECT]    = 0;
+    m_label_position[Labels::OBJECT]     = 0;
+    m_label_position[Labels::DIFFUSE]    = 1;
+    m_label_position[Labels::GLOSSY]     = 1;
+    m_label_position[Labels::SINGULAR]   = 1;
+    m_label_position[Labels::STRAIGHT]   = 1;

    m_basic_labels.insert(Labels::CAMERA);
    m_basic_labels.insert(Labels::LIGHT);
@@ -47,42 +47,40 @@ Parser::Parser(const std::vector<ustring> *user_events,
    m_minus_stop.insert(Labels::STOP);

    if (user_events)
-      for (size_t i = 0; i < user_events->size(); ++i)
-      {
-          m_label_position[(*user_events)[i]] = 0;
-          m_basic_labels.insert((*user_events)[i]);
-      }
-   if (user_scatterings)
-      for (size_t i = 0; i < user_scatterings->size(); ++i)
-      {
-          m_label_position[(*user_scatterings)[i]] = 1;
-          m_basic_labels.insert((*user_scatterings)[i]);
-      }
+        for (auto&& e : *user_events) {
+            m_label_position[e] = 0;
+            m_basic_labels.insert(e);
+        }
+    if (user_scatterings)
+        for (auto&& s : *user_scatterings) {
+            m_label_position[s] = 1;
+            m_basic_labels.insert(s);
+        }
 }


-LPexp *
-Parser::buildStop(LPexp *etype, LPexp *scatter, const std::list<LPexp*> &custom)
+LPexp*
+Parser::buildStop(LPexp* etype, LPexp* scatter, const std::list<LPexp*>& custom)
 {
-    lpexp::Cat *cat = new lpexp::Cat();
+    lpexp::Cat* cat = new lpexp::Cat();
    cat->append(etype);
    cat->append(scatter);
-    for (std::list<LPexp*>::const_iterator i = custom.begin(); i != custom.end(); ++i)
-        cat->append(*i);
+    for (const auto& p : custom)
+        cat->append(p);

    if (custom.size() < 5)
-        cat->append (new lpexp::Repeat (new lpexp::Wildexp (m_basic_labels)));
+        cat->append(new lpexp::Repeat(new lpexp::Wildexp(m_basic_labels)));
    cat->append(new lpexp::Symbol(Labels::STOP));
    return cat;
 }



-LPexp *
+LPexp*
 Parser::parseSymbol()
 {
    bool iscustom = false;
-    ustring sym = parseRawSymbol(iscustom);
+    ustring sym   = parseRawSymbol(iscustom);
    if (m_ingroup) {
        if (sym == udot)
            return new lpexp::Wildexp(m_minus_stop);
@@ -90,25 +88,26 @@ Parser::parseSymbol()
            return new lpexp::Symbol(sym);
    } else {
        if (iscustom) {
-            std::list<LPexp *> custom;
+            std::list<LPexp*> custom;
            custom.push_back(new lpexp::Symbol(sym));
-            return buildStop(new lpexp::Wildexp(m_minus_stop), new lpexp::Wildexp(m_minus_stop), custom);
+            return buildStop(new lpexp::Wildexp(m_minus_stop),
+                             new lpexp::Wildexp(m_minus_stop), custom);
        } else {
-            LPexp *basics[2] = {NULL, NULL};
+            LPexp* basics[2] = { NULL, NULL };
            if (sym != ".") {
                SymbolToInt::const_iterator i = m_label_position.find(sym);
                if (i == m_label_position.end()) {
-                    m_error = std::string("Unrecognized basic label: ") + sym.c_str();
-                    return NULL;
+                    m_error = fmtformat("Unrecognized basic label: {}", sym);
+                    return nullptr;
                }
-                int pos = i->second;
+                int pos     = i->second;
                basics[pos] = new lpexp::Symbol(sym);
            }
            for (int k = 0; k < 2; ++k)
                if (!basics[k])
                    basics[k] = new lpexp::Wildexp(m_minus_stop);
-            std::list<LPexp *> empty;
-            return buildStop (basics[0], basics[1], empty);
+            std::list<LPexp*> empty;
+            return buildStop(basics[0], basics[1], empty);
        }
    }
 }
@@ -116,7 +115,7 @@ Parser::parseSymbol()


 ustring
-Parser::parseRawSymbol(bool &iscustom)
+Parser::parseRawSymbol(bool& iscustom)
 {
    std::string sym;
    if (head() == '\'') {
@@ -144,47 +143,49 @@ Parser::parseRawSymbol(bool &iscustom)



-LPexp *
+LPexp*
 Parser::parseCat()
 {
    //lpexp::Cat *cat = new lpexp::Cat();
-    std::vector<LPexp *> explist;
+    std::vector<LPexp*> explist;
    char endchar;
    if (head() == '(') {
        next();
        endchar = ')';
-    }
-    else endchar = 0;
+    } else
+        endchar = 0;

-    while (hasInput() &&  head() != endchar) {
+    while (hasInput() && head() != endchar) {
        if (head() == '|') {
            if (!explist.size()) {
                m_error = "No left expression to or with |";
-                for (size_t i=0; i < explist.size(); ++i)
-                    delete explist[i];
-                return NULL;
+                for (auto& exp : explist)
+                    delete exp;
+                return nullptr;
            }
            next();
-            LPexp *e = _parse();
+            LPexp* e = _parse();
            if (error()) {
-                for (size_t i=0; i < explist.size(); ++i)
-                    delete explist[i];
-                return NULL;
+                delete e;
+                for (auto& exp : explist)
+                    delete exp;
+                return nullptr;
            }
            if (explist.back()->getType() == lpexp::OR)
                ((lpexp::Orlist*)explist.back())->append(e);
            else {
-                lpexp::Orlist *orexp = new lpexp::Orlist();
+                lpexp::Orlist* orexp = new lpexp::Orlist();
                orexp->append(explist.back());
                orexp->append(e);
                explist[explist.size() - 1] = orexp;
            }
        } else {
-            LPexp *e = _parse();
+            LPexp* e = _parse();
            if (error()) {
-                for (size_t i=0; i < explist.size(); ++i)
-                    delete explist[i];
-                return NULL;
+                delete e;
+                for (auto& exp : explist)
+                    delete exp;
+                return nullptr;
            }
            explist.push_back(e);
        }
@@ -193,41 +194,47 @@ Parser::parseCat()
        next();
    else if (endchar != 0) {
        m_error = "Reached end of line looking for )";
-        for (size_t i=0; i < explist.size(); ++i)
-            delete explist[i];
-        return NULL;
+        for (auto& exp : explist)
+            delete exp;
+        return nullptr;
    }
-    lpexp::Cat *cat = new lpexp::Cat();
-    for (size_t i=0; i < explist.size(); ++i)
-        cat->append(explist[i]);
+    lpexp::Cat* cat = new lpexp::Cat();
+    for (auto& exp : explist)
+        cat->append(exp);
    return cat;
 }



-LPexp *
+LPexp*
 Parser::parseGroup()
 {
    OSL_DASSERT(head() == '<');
    if (m_ingroup) {
        m_error = "No groups allowed inside of groups";
-        return NULL;
+        return nullptr;
    }
-    int basicpos = 0;
-    LPexp *basics[2] = {NULL, NULL};
-    std::list<LPexp *> custom;
-#define THROWAWAY() do{\
-    for (int i=0;i<2;++i) if(basics[i]) delete basics[i];\
-    for (std::list<LPexp *>::iterator i = custom.begin();i!=custom.end();++i) delete *i;\
-    m_ingroup = false;\
-    return NULL;\
-    }while(0)
+    int basicpos     = 0;
+    LPexp* basics[2] = { nullptr, nullptr };
+    std::list<LPexp*> custom;
+#define THROWAWAY()            \
+    do {                       \
+        delete basics[0];      \
+        delete basics[1];      \
+        for (auto& p : custom) \
+            delete p;          \
+        m_ingroup = false;     \
+        return nullptr;        \
+    } while (0)

    m_ingroup = true;
    next();
    while (hasInput() && head() != '>') {
-        LPexp *e = _parse();
-        if (error()) THROWAWAY();
+        LPexp* e = _parse();
+        if (error()) {
+            delete e;
+            THROWAWAY();
+        }
        if (basicpos < 2)
            basics[basicpos++] = e;
        else
@@ -247,65 +254,69 @@ Parser::parseGroup()



-LPexp *
+LPexp*
 Parser::parseNegor()
 {
-    OSL_DASSERT (head() == '^');
+    OSL_DASSERT(head() == '^');
    SymbolSet symlist;
-    symlist.insert(Labels::STOP); // never allowed
+    symlist.insert(Labels::STOP);  // never allowed
    int pos = -1;
    next();
    while (hasInput() && head() != ']') {
        bool iscustom = false;
-        ustring sym = parseRawSymbol(iscustom);
-        if (error()) return NULL;
+        ustring sym   = parseRawSymbol(iscustom);
+        if (error())
+            return nullptr;
        symlist.insert(sym);
        if (iscustom) {
-            if (symlist.size() > 2  && pos != -1)
-                std::cerr << "[pathexp] you are mixing labels of different type in [...]" << std::endl;
+            if (symlist.size() > 2 && pos != -1)
+                std::cerr
+                    << "[pathexp] you are mixing labels of different type in [...]"
+                    << std::endl;
            pos = -1;
        } else {
            SymbolToInt::const_iterator found = m_label_position.find(sym);
            if (found == m_label_position.end()) {
                m_error = "Unrecognized basic label";
-                return NULL;
+                return nullptr;
            }
-            if (symlist.size() > 2  && found->second != pos)
-                std::cerr << "[pathexp] you are mixing labels of different type in [...]" << std::endl;
+            if (symlist.size() > 2 && found->second != pos)
+                std::cerr
+                    << "[pathexp] you are mixing labels of different type in [...]"
+                    << std::endl;
            pos = found->second;
        }
    }
    if (!hasInput()) {
-        m_error = "Reached end of line looking for ] to end an negative or list'";
-        return NULL;
+        m_error
+            = "Reached end of line looking for ] to end an negative or list'";
+        return nullptr;
    }
    if (symlist.size() < 2) {
        m_error = "Empty or list [^] not allowed";
-        return NULL;
+        return nullptr;
    }
    next();
-    lpexp::Wildexp *wildcard = new lpexp::Wildexp(symlist);
+    lpexp::Wildexp* wildcard = new lpexp::Wildexp(symlist);
    if (m_ingroup)
        return wildcard;
    else {
-        std::list<LPexp *> custom;
-        if (pos < 0) { // is a custom label
+        std::list<LPexp*> custom;
+        if (pos < 0) {  // is a custom label
            custom.push_back(wildcard);
-            return buildStop(new lpexp::Wildexp(m_minus_stop), new lpexp::Wildexp(m_minus_stop), custom);
+            return buildStop(new lpexp::Wildexp(m_minus_stop),
+                             new lpexp::Wildexp(m_minus_stop), custom);
        } else {
-            LPexp *basics[2] = {NULL, NULL};
-            basics[pos] = wildcard;
-            for (int i = 0; i < 2; ++i)
-                if (!basics[i])
-                    basics[i] = new lpexp::Wildexp(m_minus_stop);
-            return buildStop(basics[0], basics[1], custom);
+            LPexp* w2 = new lpexp::Wildexp(m_minus_stop);
+            return buildStop(pos == 0 ? wildcard : w2, pos == 0 ? w2 : wildcard,
+                             custom);
        }
    }
 }



-LPexp *
+LPexp*
 Parser::parseOrlist()
 {
    OSL_DASSERT(head() == '[');
@@ -313,19 +324,20 @@ Parser::parseOrlist()
    if (hasInput() && head() == '^')
        return parseNegor();
    else {
-        lpexp::Orlist *orlist = new lpexp::Orlist();
+        lpexp::Orlist* orlist = new lpexp::Orlist();
        while (hasInput() && head() != ']') {
-            LPexp *e = _parse();
+            LPexp* e = _parse();
            if (error()) {
+                delete e;
                delete orlist;
-                return NULL;
+                return nullptr;
            }
            orlist->append(e);
        }
        if (!hasInput()) {
            m_error = "Reached end of line looking for ] to end an or list";
            delete orlist;
-            return NULL;
+            return nullptr;
        }
        next();
        return orlist;
@@ -360,29 +372,32 @@ Parser::parseRange()
    }
    next();
    if (secondnum.size())
-        return std::pair<int, int>(atoi(firstnum.c_str()), atoi(secondnum.c_str()));
+        return std::pair<int, int>(OIIO::Strutil::stoi(firstnum),
+                                   OIIO::Strutil::stoi(secondnum));
    else
-        return std::pair<int, int>(atoi(firstnum.c_str()), atoi(firstnum.c_str()));
+        return std::pair<int, int>(OIIO::Strutil::stoi(firstnum),
+                                   OIIO::Strutil::stoi(firstnum));
 }



-LPexp *
-Parser::parseModifier(LPexp *e)
+LPexp*
+Parser::parseModifier(LPexp* e)
 {
    if (hasInput()) {
        if (m_ingroup && (head() == '*' || head() == '{' || head() == '+')) {
            m_error = std::string("Repetitions not allowed inside '<...>'");
-            return NULL;
+            return nullptr;
        }
        if (head() == '*') {
            next();
            return new lpexp::Repeat(e);
        } else if (head() == '{') {
            std::pair<int, int> range = parseRange();
-            if (error()) return NULL;
+            if (error())
+                return nullptr;
            if (range.second < 0) {
-                lpexp::Cat *cat = new lpexp::Cat();
+                lpexp::Cat* cat = new lpexp::Cat();
                cat->append(new lpexp::NRepeat(e, range.first, range.first));
                cat->append(new lpexp::Repeat(e->clone()));
                return cat;
@@ -390,23 +405,22 @@ Parser::parseModifier(LPexp *e)
                return new lpexp::NRepeat(e, range.first, range.second);
        } else if (head() == '+') {
            next();
-            lpexp::Cat *cat = new lpexp::Cat();
+            lpexp::Cat* cat = new lpexp::Cat();
            cat->append(e);
            cat->append(new lpexp::Repeat(e->clone()));
            return cat;
        } else
            return e;
-    }
-    else
+    } else
        return e;
 }



-LPexp *
+LPexp*
 Parser::_parse()
 {
-    LPexp *e;
+    LPexp* e = nullptr;
    if (head() == '(')
        e = parseCat();
    else if (head() == '[')
@@ -415,24 +429,26 @@ Parser::_parse()
        e = parseGroup();
    else
        e = parseSymbol();
-    if (error())
-        return NULL;
+    if (error()) {
+        delete e;
+        return nullptr;
+    }
    return parseModifier(e);
 }



-LPexp *
-Parser::parse(const char *text)
+LPexp*
+Parser::parse(const char* text)
 {
-    m_error = "";
-    m_text = text;
-    m_pos = 0;
+    m_error   = "";
+    m_text    = text;
+    m_pos     = 0;
    m_ingroup = false;
    if (hasInput())
        return parseCat();
    else
-        return NULL;
+        return nullptr;
 }


@@ -33,92 +33,91 @@ using lpexp::LPexp;
 ///    exp        := wildexp | borexp | sorexp | catexp | group | repeat | repeat+ | mnrepeat
 ///    topexp    := exp | topexp exp
 ///
-class Parser
-{
-    public:
+class Parser {
+public:
+    Parser(const std::vector<ustring>* user_events      = NULL,
+           const std::vector<ustring>* user_scatterings = NULL);

-        Parser(const std::vector<ustring> *user_events = NULL,
-               const std::vector<ustring> *user_scatterings = NULL);
+    /// Parse a string and return the resulting light path expression tree or NULL if failed
+    LPexp* parse(const char* text);

-        /// Parse a string and return the resulting light path expression tree or NULL if failed
-        LPexp *parse(const char *text);
+    /// Check for error in the last parsed string
+    bool error() const { return m_error.size() > 0; };
+    /// Get the error string
+    const char* getErrorMsg() const { return m_error.c_str(); };
+    /// Get the position of the string where the error appeared
+    int getErrorPos() const { return m_pos; };

-        /// Check for error in the last parsed string
-        bool error()const { return m_error.size() > 0; };
-        /// Get the error string
-        const char *getErrorMsg()const { return m_error.c_str(); };
-        /// Get the position of the string where the error appeared
-        int getErrorPos()const { return m_pos; };
+private:
+    /// Current char being parsed
+    char head() const { return m_text[m_pos]; };
+    /// Any input (including head) left?
+    bool hasInput() const { return m_text.size() > m_pos; };
+    /// Go to the next char in the string
+    void next() { m_pos++; };

-    private:
+    /// build the complete pattern for a pathtracing stop
+    ///
+    /// That means the three basic label match expressions, plus those for the custom labels (if any),
+    /// plus an additional wildcard to eat extra custom labels at the end, and finally the stop mark.
+    /// So you provide the regexps for the basic three labels:
+    ///
+    ///      \param etype         Event type (CFVL)
+    ///      \param dir            Direction (RD)
+    ///      \param scatter      Scattering type (DGSs)
+    ///
+    /// That of course can be wildcards or any other expression willing to match them. And then you can
+    /// provide extra expressions to match the custom labels:
+    ///
+    ///      \param custom        Custom label regexps
+    ///
+    /// And the function will automatically add a "[^stop_mark]*stop_mark" at the end.
+    ///
+    LPexp* buildStop(LPexp* etype, LPexp* scatter,
+                     const std::list<LPexp*>& custom);
+    /// Given that a symbol is ready in head() to parse, parse it
+    LPexp* parseSymbol();
+    /// Given that a symbol is ready in head() to parse, parse it as a ustring
+    /// and report it was a custom symbol in the iscustom flag
+    ustring parseRawSymbol(bool& iscustom);
+    /// Given that the beginning of a concatenation of regexps is ready to
+    /// parse, parse it and it can be optionally be enclosed in parentheses ()
+    LPexp* parseCat();
+    /// Given that a fully qualified group like <.RD'custom'> is ready to parse, parse it
+    LPexp* parseGroup();
+    /// Given that a ^abcde] (note missing [) is ready to parse, parse it
+    LPexp* parseNegor();
+    /// Given that [abcde] is ready to parse, parse it, but if it finds that it was actually
+    /// [^abcd], fall back to parseNegor
+    LPexp* parseOrlist();
+    /// Given that a range like {5,7} or {2,} or {3} is ready to parse, parse it and return
+    /// the range. Second number being -1 if if was {2,} and equals the first if it was {3}
+    std::pair<int, int> parseRange();
+    /// Take an already parsed lpexp and parse its possible modifier (*={}) if present and
+    /// return the new lpexp
+    LPexp* parseModifier(LPexp* e);
+    /// Generic parse whatever comes next (just one item)
+    LPexp* _parse();

-        /// Current char being parsed
-        char head()const { return m_text[m_pos]; };
-        /// Any input (including head) left?
-        bool hasInput()const { return m_text.size() > m_pos; };
-        /// Go to the next char in the string
-        void next() { m_pos++; };
+    // error string
+    std::string m_error;
+    // True if we are actually parsing a group <>, since otherwise everything gets
+    // automatically converted to a group, this prevents that happening
+    bool m_ingroup;
+    // maps each basic label to its expected position in the appearance order,
+    // for instance the direction label can't be in the first pos of a group.
+    // This way we know where to put the expression when the user writes just
+    // S, which translates to <..S>
+    SymbolToInt m_label_position;
+    // The set of the basic labels
+    SymbolSet m_basic_labels;
+    // The black list for our wildcards, that have to exclude always the stop mark
+    SymbolSet m_minus_stop;

-        /// build the complete pattern for a pathtracing stop
-        ///
-        /// That means the three basic label match expressions, plus those for the custom labels (if any),
-        /// plus an additional wildcard to eat extra custom labels at the end, and finally the stop mark.
-        /// So you provide the regexps for the basic three labels:
-        ///
-        ///      \param etype         Event type (CFVL)
-        ///      \param dir            Direction (RD)
-        ///      \param scatter      Scattering type (DGSs)
-        ///
-        /// That of course can be wildcards or any other expression willing to match them. And then you can
-        /// provide extra expressions to match the custom labels:
-        ///
-        ///      \param custom        Custom label regexps
-        ///
-        /// And the function will automatically add a "[^stop_mark]*stop_mark" at the end.
-        ///
-        LPexp *buildStop(LPexp *etype, LPexp *scatter, const std::list<LPexp*> &custom);
-        /// Gicen that a symbol is ready in head() to parse, parse it
-        LPexp *parseSymbol();
-        /// Gicen that a symbol is ready in head() to parse, parse it as a ustring
-        /// and report it was a custom symbol in the iscustom flag
-        ustring parseRawSymbol(bool &iscustom);
-        /// Given that the begining of a concatenation of regexps is ready to parse, parse it
-        /// and it can be optionally be enclosed in parentheses ()
-        LPexp *parseCat();
-        /// Given that a fully qualified group like <.RD'custom'> is ready to parse, parse it
-        LPexp *parseGroup();
-        /// Given that a ^abcde] (note missing [) is ready to parse, parse it
-        LPexp *parseNegor();
-        /// Given that [abcde] is ready to parse, parse it, but if it finds that it was actually
-        /// [^abcd], fall back to parseNegor
-        LPexp *parseOrlist();
-        /// Given that a range like {5,7} or {2,} or {3} is ready to parse, parse it and return
-        /// the range. Second number being -1 if if was {2,} and equals the first if it was {3}
-        std::pair<int, int> parseRange();
-        /// Take an already parsed lpexp and parse its possible modifier (*={}) if present and
-        /// return the new lpexp
-        LPexp *parseModifier(LPexp *e);
-        /// Generic parse whatever comes next (just one item)
-        LPexp *_parse();
-
-        // error string
-        std::string    m_error;
-        // True if we are actually parsing a group <>, since otherwise everything gets
-        // automatically converted to a group, this prevents that happening
-        bool             m_ingroup;
-        // maps each basic label to its expected possition in the appearance order, for instance
-        // the direction label can't be in the first pos of a group. This way we know where to put the
-        // expression when the user writes just S, which translates to <..S>
-        SymbolToInt m_label_position;
-        // The set of the basic labels
-        SymbolSet    m_basic_labels;
-        // The black list for our wildcards, that have to exclude always the stop mark
-        SymbolSet    m_minus_stop;
-
-        // Current text being parsed
-        std::string    m_text;
-        // Current position in the text for head()
-        size_t          m_pos;
+    // Current text being parsed
+    std::string m_text;
+    // Current position in the text for head()
+    size_t m_pos;
 };


@@ -75,7 +75,7 @@ lpexp::Symbol::genAuto(NdfAutomata& automata) const
 lpexp::FirstLast
 lpexp::Wildexp::genAuto(NdfAutomata& automata) const
 {
-    // Same as the Symbol lpexp but with a wildcard insted of a symbol
+    // Same as the Symbol lpexp but with a wildcard instead of a symbol
    NdfAutomata::State* begin = automata.newState();
    NdfAutomata::State* end   = automata.newState();
    begin->addWildcardTransition(new Wildcard(m_wildcard), end);
@@ -13,17 +13,10 @@ namespace lpexp {

 // This is just a pair of states, see the use of the function genAuto in LPexp
 // for a justification of this type that we use throughout all the regexp code
-typedef std::pair<NdfAutomata::State *, NdfAutomata::State *> FirstLast;
+typedef std::pair<NdfAutomata::State*, NdfAutomata::State*> FirstLast;

 /// LPexp atom type for the getType method
-typedef enum {
-    CAT,
-    OR,
-    SYMBOL,
-    WILDCARD,
-    REPEAT,
-    NREPEAT
-}Regtype;
+typedef enum { CAT, OR, SYMBOL, WILDCARD, REPEAT, NREPEAT } Regtype;



@@ -42,56 +35,56 @@ typedef enum {
 ///     NREPEAT  Bounded repetition of the child expression like (exp){n,m}
 ///
 class LPexp {
-    public:
-        virtual ~LPexp() {};
+public:
+    virtual ~LPexp() {};

-        /// Generate automata states for this subtree
-        ///
-        /// This method recursively builds all the needed automata states of
-        /// the tree rooted by this node and returns the begin and end states
-        /// for it. That means that if it were the only thing in the automata,
-        /// making retvalue.first initial state and retvalue.second final state,
-        /// would be the right thing to do.
-        ///
-        virtual FirstLast genAuto(NdfAutomata &automata)const = 0;
-        /// Get the type for this node
-        virtual Regtype getType()const = 0;
-        /// For the parser's convenience. It is easy to implement things like a+
-        /// as aa*. So the amount of regexp classes gets reduced. For doing that
-        /// it needs an abstract clone function
-        virtual LPexp * clone()const = 0;
+    /// Generate automata states for this subtree
+    ///
+    /// This method recursively builds all the needed automata states of
+    /// the tree rooted by this node and returns the begin and end states
+    /// for it. That means that if it were the only thing in the automata,
+    /// making retvalue.first initial state and retvalue.second final state,
+    /// would be the right thing to do.
+    ///
+    virtual FirstLast genAuto(NdfAutomata& automata) const = 0;
+    /// Get the type for this node
+    virtual Regtype getType() const = 0;
+    /// For the parser's convenience. It is easy to implement things like a+
+    /// as aa*. So the amount of regexp classes gets reduced. For doing that
+    /// it needs an abstract clone function
+    virtual LPexp* clone() const = 0;
 };



 /// LPexp concatenation
 class Cat final : public LPexp {
-    public:
-        virtual ~Cat();
-        void append(LPexp *regexp);
-        virtual FirstLast genAuto(NdfAutomata &automata)const;
-        virtual Regtype getType()const { return CAT; };
-        virtual LPexp * clone()const;
+public:
+    virtual ~Cat();
+    void append(LPexp* regexp);
+    virtual FirstLast genAuto(NdfAutomata& automata) const;
+    virtual Regtype getType() const { return CAT; };
+    virtual LPexp* clone() const;

-    protected:
-        std::list<LPexp *> m_children;
+protected:
+    std::list<LPexp*> m_children;
 };



 /// Basic symbol like G or 'customlabel'
 class Symbol final : public LPexp {
-    public:
-        Symbol(ustring sym) { m_sym = sym; };
-        virtual ~Symbol() {};
+public:
+    Symbol(ustring sym) { m_sym = sym; };
+    virtual ~Symbol() {};

-        virtual FirstLast genAuto(NdfAutomata &automata)const;
-        virtual Regtype getType()const { return SYMBOL; };
-        virtual LPexp * clone()const { return new Symbol(*this); };
+    virtual FirstLast genAuto(NdfAutomata& automata) const;
+    virtual Regtype getType() const { return SYMBOL; };
+    virtual LPexp* clone() const { return new Symbol(*this); };

-    protected:
-        // All symbols are unique ustrings
-        ustring m_sym;
+protected:
+    // All symbols are unique ustrings
+    ustring m_sym;
 };


@@ -100,63 +93,67 @@ class Symbol final : public LPexp {
 ///
 /// Named like this to avoid confusion with the automata Wildcard class
 class Wildexp final : public LPexp {
-    public:
-        Wildexp(SymbolSet &minus):m_wildcard(minus) {};
-        virtual ~Wildexp() {};
+public:
+    Wildexp(SymbolSet& minus) : m_wildcard(minus) {};
+    virtual ~Wildexp() {};

-        virtual FirstLast genAuto(NdfAutomata &automata)const;
-        virtual Regtype getType()const { return WILDCARD; };
-        virtual LPexp * clone()const { return new Wildexp(*this); };
+    virtual FirstLast genAuto(NdfAutomata& automata) const;
+    virtual Regtype getType() const { return WILDCARD; };
+    virtual LPexp* clone() const { return new Wildexp(*this); };

-    protected:
-        // And internally we use the automata's Wildcard type
-        Wildcard m_wildcard;
+protected:
+    // And internally we use the automata's Wildcard type
+    Wildcard m_wildcard;
 };



 /// Ored list of expressions
 class Orlist final : public LPexp {
-    public:
-        virtual ~Orlist();
-        void append(LPexp *regexp);
-        virtual FirstLast genAuto(NdfAutomata &automata)const;
-        virtual Regtype getType()const { return OR; };
-        virtual LPexp * clone()const;
+public:
+    virtual ~Orlist();
+    void append(LPexp* regexp);
+    virtual FirstLast genAuto(NdfAutomata& automata) const;
+    virtual Regtype getType() const { return OR; };
+    virtual LPexp* clone() const;

-    protected:
-        std::list<LPexp *> m_children;
+protected:
+    std::list<LPexp*> m_children;
 };



 // Unlimited repeat: (exp)*
 class Repeat final : public LPexp {
-    public:
-        Repeat(LPexp *child):m_child(child) {};
-        virtual ~Repeat() { delete m_child; };
-        virtual FirstLast genAuto(NdfAutomata &automata)const;
-        virtual Regtype getType()const { return REPEAT; };
-        virtual LPexp * clone()const { return new Repeat(m_child->clone()); };
+public:
+    Repeat(LPexp* child) : m_child(child) {};
+    virtual ~Repeat() { delete m_child; };
+    virtual FirstLast genAuto(NdfAutomata& automata) const;
+    virtual Regtype getType() const { return REPEAT; };
+    virtual LPexp* clone() const { return new Repeat(m_child->clone()); };

-    protected:
-        LPexp *m_child;
+protected:
+    LPexp* m_child;
 };



 // Bounded repeat: (exp){m,n}
 class NRepeat final : public LPexp {
-    public:
-        NRepeat(LPexp *child, int min, int max):m_child(child),m_min(min),m_max(max) {};
-        virtual ~NRepeat() { delete m_child; };
-        virtual FirstLast genAuto(NdfAutomata &automata)const;
-        virtual Regtype getType()const { return NREPEAT; };
-        virtual LPexp * clone()const { return new NRepeat(m_child->clone(), m_min, m_max); };
+public:
+    NRepeat(LPexp* child, int min, int max)
+        : m_child(child), m_min(min), m_max(max) {};
+    virtual ~NRepeat() { delete m_child; };
+    virtual FirstLast genAuto(NdfAutomata& automata) const;
+    virtual Regtype getType() const { return NREPEAT; };
+    virtual LPexp* clone() const
+    {
+        return new NRepeat(m_child->clone(), m_min, m_max);
+    };

-    protected:
-        LPexp *m_child;
-        int m_min, m_max;
+protected:
+    LPexp* m_child;
+    int m_min, m_max;
 };


@@ -167,19 +164,18 @@ class NRepeat final : public LPexp {
 /// a LPexp. It actually binds a light path expression to a certain rule.
 /// Making the begin state initial and the end state final. It can't be
 /// nested in other light path expressions, it is the root of the tree.
-class Rule
-{
-    public:
-        Rule(LPexp *child, void *rule):m_child(child), m_rule(rule) {};
-        virtual ~Rule() { delete m_child; };
-        void genAuto(NdfAutomata &automata)const;
+class Rule {
+public:
+    Rule(LPexp* child, void* rule) : m_child(child), m_rule(rule) {};
+    virtual ~Rule() { delete m_child; };
+    void genAuto(NdfAutomata& automata) const;

-    protected:
-        LPexp *m_child;
-        // Anonymous pointer to the associated object for this rule
-        void *m_rule;
+protected:
+    LPexp* m_child;
+    // Anonymous pointer to the associated object for this rule
+    void* m_rule;
 };

-} // namespace regexp
+}  // namespace lpexp

 OSL_NAMESPACE_EXIT
@@ -38,6 +38,7 @@ namespace pvt {
 #define IlluminantE    0.33333333, 0.33333333  /* CIE equal-energy illuminant */
 #define IlluminantACES 0.32168, 0.33767        /* For ACES, approximate D60 */

+namespace {  // anon namespace to avoid duplicate OptiX symbols
 OSL_CONSTANT_DATA const static ColorSystem::Chroma k_color_systems[13] = {
   // Index, Name        xRed    yRed   xGreen  yGreen   xBlue   yBlue    White point
   /* 0  Rec709     */ { 0.64,   0.33,   0.30,   0.60,   0.15,   0.06,   IlluminantD65 },
@@ -54,6 +55,7 @@ OSL_CONSTANT_DATA const static ColorSystem::Chroma k_color_systems[13] = {
   /* 11 ACES2065-1 */ { 0.7347, 0.2653, 0.0,    1.0,    0.0001, -0.077, IlluminantACES },
   /* 12 ACEScg     */ { 0.713,  0.293,  0.165,  0.83,   0.128,  0.044,  IlluminantACES },
 };
+}  // namespace

 // clang-format on

@@ -231,17 +233,12 @@ rend_get_userdata(StringParam name, void* data, int data_size,
 OSL_HOSTDEVICE void
 ColorSystem::error(StringParam src, StringParam dst, Context sg) const
 {
-#    if !defined(__CUDA_ARCH__) || OPTIX_VERSION < 70000
+#    ifndef __CUDA_ARCH__
    const char* args[2] = { src.c_str(), dst.c_str() };
    osl_printf(sg,
               (char*)  // FIXME!
               "ERROR: Unknown color space transformation \"%s\" -> \"%s\"\n",
               args);
-#    else
-    uint64_t fmt_hash = UStringHash::Hash(
-        "ERROR: Unknown color space transformation \"%s\" -> \"%s\"\n");
-    uint64_t args[3] = { 2 * sizeof(uint64_t), dst, src };
-    osl_printf(sg, (char*)fmt_hash, args);
 #    endif
 }

@@ -268,9 +265,11 @@ ColorSystem::error(StringParam src, StringParam dst, Context context) const
 }

 static inline const ColorSystem&
-op_color_colorsystem(void* sg)
+op_color_colorsystem(void* sg_)
 {
-    return ((ShaderGlobals*)sg)->context->shadingsys().colorsystem();
+    ShaderGlobals* sg        = (ShaderGlobals*)sg_;
+    ShadingStateUniform* ssu = (ShadingStateUniform*)sg->shadingStateUniform;
+    return ssu->m_colorsystem;
 }

 static inline OSL::ShadingContext*
@@ -11,10 +11,10 @@
 ///
 /////////////////////////////////////////////////////////////////////////

-#include <OSL/dual.h>
-#include <OSL/dual_vec.h>
 #include <OSL/Imathx/Imathx.h>
 #include <OSL/device_string.h>
+#include <OSL/dual.h>
+#include <OSL/dual_vec.h>

 #include <OpenImageIO/color.h>

@@ -37,96 +37,107 @@ public:
    // A colour system is defined by the CIE x and y coordinates of its
    // three primary illuminants and its white point.
    struct Chroma {
-        float  xRed, yRed,
-               xGreen, yGreen,
-               xBlue, yBlue,
-               xWhite, yWhite;
+        float xRed, yRed, xGreen, yGreen, xBlue, yBlue, xWhite, yWhite;
    };

    OSL_HOSTDEVICE static const Chroma* fromString(StringParam colorspace);

    /// Convert an XYZ color to RGB in our preferred color space.
-    template <typename T> OSL_HOSTDEVICE T
-    XYZ_to_RGB (const T &XYZ) const { return XYZ * m_XYZ2RGB; }
+    template<typename T> OSL_HOSTDEVICE T XYZ_to_RGB(const T& XYZ) const
+    {
+        return XYZ * m_XYZ2RGB;
+    }

    /// Convert an RGB color in our preferred color space to XYZ.
-    template <typename T> OSL_HOSTDEVICE T
-    RGB_to_XYZ (const T &RGB) const { return RGB * m_RGB2XYZ; }
+    template<typename T> OSL_HOSTDEVICE T RGB_to_XYZ(const T& RGB) const
+    {
+        return RGB * m_RGB2XYZ;
+    }

    /// Return the luminance of an RGB color in the current color space.
-    OSL_HOSTDEVICE float
-    luminance (const Color3 &RGB) const { return RGB.dot(m_luminance_scale); }
+    OSL_HOSTDEVICE float luminance(const Color3& RGB) const
+    {
+        return RGB.dot(m_luminance_scale);
+    }

    /// Return the luminance scale  of the current color space.
-    const Color3 & luminance_scale() const { return m_luminance_scale; }
+    const Color3& luminance_scale() const
+    {
+        return m_luminance_scale;
+    }

    /// Return the RGB in the current color space for blackbody radiation
    /// at temperature T (in Kelvin).
-    OSL_HOSTDEVICE inline Color3
-    blackbody_rgb (float T /*Kelvin*/) const;
+    OSL_HOSTDEVICE inline Color3 blackbody_rgb(float T /*Kelvin*/) const;

    // Interface to access underlying optimized lookup table for blackbody
    // When can_lookup_blackbody() returns true,
    // then lookup_blackbody_rgb can be safely called,
    // otherwise the compute_blackbody_rgb is required.
-    OSL_HOSTDEVICE inline bool
-    can_lookup_blackbody(float T /*Kelvin*/) const;
+    OSL_HOSTDEVICE inline bool can_lookup_blackbody(float T /*Kelvin*/) const;

-    OSL_HOSTDEVICE inline Color3
-    lookup_blackbody_rgb (float T /*Kelvin*/) const;
+    OSL_HOSTDEVICE inline Color3 lookup_blackbody_rgb(float T /*Kelvin*/) const;

    // Expensive real computation without lookup table
-    OSL_HOSTDEVICE inline Color3
-    compute_blackbody_rgb (float T /*Kelvin*/) const;
+    OSL_HOSTDEVICE inline Color3 compute_blackbody_rgb(float T /*Kelvin*/) const;


    /// Set the current color space.
-    OSL_HOSTDEVICE bool
-    set_colorspace (StringParam colorspace);
+    OSL_HOSTDEVICE bool set_colorspace(StringParam colorspace);

-    OSL_HOSTDEVICE Color3
-    to_rgb (StringParam fromspace, const Color3& C, Context) const;
+    OSL_HOSTDEVICE Color3 to_rgb(StringParam fromspace, const Color3& C,
+                                 Context) const;

-    OSL_HOSTDEVICE Color3
-    from_rgb (StringParam fromspace, const Color3& C, Context) const;
+    OSL_HOSTDEVICE Color3 from_rgb(StringParam fromspace, const Color3& C,
+                                   Context) const;

-    OSL_HOSTDEVICE Dual2<Color3>
-    transformc (StringParam fromspace, StringParam tospace,
-                const Dual2<Color3>& color, Context ctx) const;
+    OSL_HOSTDEVICE Dual2<Color3> transformc(StringParam fromspace,
+                                            StringParam tospace,
+                                            const Dual2<Color3>& color,
+                                            Context ctx) const;

-    OSL_HOSTDEVICE Color3
-    transformc (StringParam fromspace, StringParam tospace,
-                const Color3& color, Context ctx) const;
+    OSL_HOSTDEVICE Color3 transformc(StringParam fromspace, StringParam tospace,
+                                     const Color3& color, Context ctx) const;

-    OSL_HOSTDEVICE Dual2<Color3>
-    ocio_transform (StringParam fromspace, StringParam tospace, const Dual2<Color3>& C, Context) const;
+    OSL_HOSTDEVICE Dual2<Color3> ocio_transform(StringParam fromspace,
+                                                StringParam tospace,
+                                                const Dual2<Color3>& C,
+                                                Context) const;

-    OSL_HOSTDEVICE Color3
-    ocio_transform (StringParam fromspace, StringParam tospace, const Color3& C, Context) const;
+    OSL_HOSTDEVICE Color3 ocio_transform(StringParam fromspace,
+                                         StringParam tospace, const Color3& C,
+                                         Context) const;


-    OSL_HOSTDEVICE const StringParam& colorspace() const { return m_colorspace; }
+    OSL_HOSTDEVICE const StringParam& colorspace() const
+    {
+        return m_colorspace;
+    }

    OSL_HOSTDEVICE void error(StringParam src, StringParam dst, Context) const;

 private:
-    template <typename Color> OSL_HOSTDEVICE inline Color
-    transformc (StringParam fromspace, StringParam tospace, const Color& C, Context) const;
+    template<typename Color>
+    OSL_HOSTDEVICE inline Color transformc(StringParam fromspace,
+                                           StringParam tospace, const Color& C,
+                                           Context) const;

-    template <typename Color> OSL_HOSTDEVICE inline Color
-    ocio_transform (StringParam fromspace, StringParam tospace, const Color& C, Context) const;
+    template<typename Color>
+    OSL_HOSTDEVICE inline Color ocio_transform(StringParam fromspace,
+                                               StringParam tospace,
+                                               const Color& C, Context) const;


    // Derived/cached calculations from options:
-    Color3 m_Red, m_Green, m_Blue;   ///< Color primaries (xyY)
-    Color3 m_White;                  ///< White point (xyY)
-    Matrix33 m_XYZ2RGB;              ///< XYZ to RGB conversion matrix
-    Matrix33 m_RGB2XYZ;              ///< RGB to XYZ conversion matrix
-    Color3 m_luminance_scale;        ///< Scaling for RGB->luma
-    Color3 m_blackbody_table[317];   ///< Precomputed blackbody table
+    Color3 m_Red, m_Green, m_Blue;  ///< Color primaries (xyY)
+    Color3 m_White;                 ///< White point (xyY)
+    Matrix33 m_XYZ2RGB;             ///< XYZ to RGB conversion matrix
+    Matrix33 m_RGB2XYZ;             ///< RGB to XYZ conversion matrix
+    Color3 m_luminance_scale;       ///< Scaling for RGB->luma
+    Color3 m_blackbody_table[317];  ///< Precomputed blackbody table

    // Keep this last so the CUDA device string can be easily set
-    StringParam m_colorspace;        ///< What RGB colors mean
+    StringParam m_colorspace;  ///< What RGB colors mean
 };


@@ -138,9 +149,10 @@ public:
                                              ShadingSystemImpl* shadingsys);

 private:
-    const OIIO::ColorConfig& colorconfig (ShadingSystemImpl* shadingsys);
+    const OIIO::ColorConfig& colorconfig(ShadingSystemImpl* shadingsys);

-    std::shared_ptr<OIIO::ColorConfig> m_colorconfig; ///< OIIO/OCIO color configuration
+    std::shared_ptr<OIIO::ColorConfig>
+        m_colorconfig;  ///< OIIO/OCIO color configuration

    // 1-item cache for the last requested custom color conversion processor
    OIIO::ColorProcessorHandle m_last_colorproc;
@@ -149,7 +161,7 @@ private:
 #endif
 };

-} // namespace pvt
+}  // namespace pvt


 OSL_NAMESPACE_EXIT
@@ -22,15 +22,16 @@
 OSL_NAMESPACE_ENTER

 #ifdef __OSL_WIDE_PVT
-    namespace __OSL_WIDE_PVT {
+namespace __OSL_WIDE_PVT {
 #else
-    namespace pvt {
+namespace pvt {
 #endif


 namespace {

-OSL_HOSTDEVICE inline void clamp_zero (Color3 &c)
+OSL_HOSTDEVICE inline void
+clamp_zero(Color3& c)
 {
    if (c.x < 0.0f)
        c.x = 0.0f;
@@ -48,8 +49,9 @@ OSL_HOSTDEVICE inline void clamp_zero (Color3 &c)
 //        cie_colour_match[(lambda - 380) / 5][2] = zBar
 //OSL_CONSTANT_DATA const float cie_colour_match[81][3] =
 // Choose to access 1d array vs 2d to allow better code generation of gathers
-OSL_CONSTANT_DATA const float cie_colour_match[81*3] OSL_ALIGNAS(64) =
-{
+namespace {  // anon namespace to avoid duplicate OptiX symbols
+static OSL_CONSTANT_DATA const float cie_colour_match[81 * 3] OSL_ALIGNAS(64) = {
+    // clang-format off
    0.0014,0.0000,0.0065, 0.0022,0.0001,0.0105, 0.0042,0.0001,0.0201,
    0.0076,0.0002,0.0362, 0.0143,0.0004,0.0679, 0.0232,0.0006,0.1102,
    0.0435,0.0012,0.2074, 0.0776,0.0022,0.3713, 0.1344,0.0040,0.6456,
@@ -77,22 +79,23 @@ OSL_CONSTANT_DATA const float cie_colour_match[81*3] OSL_ALIGNAS(64) =
    0.0007,0.0002,0.0000, 0.0005,0.0002,0.0000, 0.0003,0.0001,0.0000,
    0.0002,0.0001,0.0000, 0.0002,0.0001,0.0000, 0.0001,0.0000,0.0000,
    0.0001,0.0000,0.0000, 0.0001,0.0000,0.0000, 0.0000,0.0000,0.0000
+    // clang-format on
 };
-
+}  // namespace


 // For a given wavelength lambda (in nm), return the XYZ triple giving the
 // XYZ color corresponding to that single wavelength;
 OSL_HOSTDEVICE static Color3
-wavelength_color_XYZ (float lambda_nm)
+wavelength_color_XYZ(float lambda_nm)
 {
-    float ii = (lambda_nm-380.0f) / 5.0f;  // scaled 0..80
-    int i = (int) ii;
+    float ii = (lambda_nm - 380.0f) / 5.0f;  // scaled 0..80
+    int i    = (int)ii;
    // NOTE: bitwise OR to avoid branchiness logical OR introduces.
    // Also when left as logical OR, vectorizing with clang 11 did not
    // mask off gathering of out of range index values causing segfaults
    if ((i < 0) | (i >= 80))
-        return Color3(0.0f,0.0f,0.0f);
+        return Color3(0.0f, 0.0f, 0.0f);
    float remainder = ii - i;
    // Do not separate address calculation so that compiler can see that the
    // base pointer is uniform for all data lanes and 32bit indices can be
@@ -101,10 +104,15 @@ wavelength_color_XYZ (float lambda_nm)
    // Furthermore, have all gathers use the same indices by having different
    // base registers for each of the 6 components to avoid need to manipulate
    // indices between gathers.
-    constexpr int stride = sizeof(Color3)/sizeof(float);
-    const int si = stride*i;
-    Color3 XYZ = OIIO::lerp (Color3((cie_colour_match+0)[si], (cie_colour_match+1)[si], (cie_colour_match+2)[si]),
-                             Color3((cie_colour_match+3)[si], (cie_colour_match+4)[si], (cie_colour_match+5)[si]), remainder);
+    constexpr int stride = sizeof(Color3) / sizeof(float);
+    const int si         = stride * i;
+    Color3 XYZ           = OIIO::lerp(Color3((cie_colour_match + 0)[si],
+                                             (cie_colour_match + 1)[si],
+                                             (cie_colour_match + 2)[si]),
+                                      Color3((cie_colour_match + 3)[si],
+                                             (cie_colour_match + 4)[si],
+                                             (cie_colour_match + 5)[si]),
+                                      remainder);
 #if 0
    float n = (XYZ.x + XYZ.y + XYZ.z);
    float n_inv = (n >= 1.0e-6f ? 1.0f/n : 0.0f);
@@ -130,27 +138,32 @@ wavelength_color_XYZ (float lambda_nm)
 // I tuned this a bit, and with the current values we can have all
 // blackbody results accurate to within 0.1% with a table size of 317
 // (about 5 KB of data).
-#define BB_DRAPER 800.0f /* really 798K, below this visible BB is negligible */
+#define BB_DRAPER          800.0f /* really 798K, below this visible BB is negligible */
 #define BB_MAX_TABLE_RANGE 12000.0f /* max temp for which we use the table */
-#define BB_TABLE_XPOWER 1.5f       // NOTE: not used, hardcoded into expressions below
-#define BB_TABLE_YPOWER 5.0f       // NOTE: decode is hardcoded
+#define BB_TABLE_XPOWER \
+    1.5f  // NOTE: not used, hardcoded into expressions below
+#define BB_TABLE_YPOWER  5.0f  // NOTE: decode is hardcoded
 #define BB_TABLE_SPACING 2.0f

-OSL_HOSTDEVICE inline float BB_TABLE_MAP(float i) {
+OSL_HOSTDEVICE inline float
+BB_TABLE_MAP(float i)
+{
    // return powf (i, BB_TABLE_XPOWER) * BB_TABLE_SPACING + BB_DRAPER;
    float is = sqrtf(i);
-    float ip = is * is * is; // ^3/2
+    float ip = is * is * is;  // ^3/2
    return ip * BB_TABLE_SPACING + BB_DRAPER;
 }

-OSL_HOSTDEVICE inline float BB_TABLE_UNMAP(float T) {
+OSL_HOSTDEVICE inline float
+BB_TABLE_UNMAP(float T)
+{
    // return powf ((T - BB_DRAPER) / BB_TABLE_SPACING, 1.0f/BB_TABLE_XPOWER);
-    float t  = (T - BB_DRAPER) / BB_TABLE_SPACING;
+    float t = (T - BB_DRAPER) / BB_TABLE_SPACING;
    // Clang doesn't have vectorizing version of cbrtf,
    // so use OIIO's instead
    float ic = OIIO::fast_cbrt(t);

-    return ic * ic; // ^2/3
+    return ic * ic;  // ^2/3
 }


@@ -173,26 +186,31 @@ OSL_HOSTDEVICE inline float BB_TABLE_UNMAP(float T) {
 // W/m^2 in the range [wavelength,wavelength+dwavelength].
 class bb_spectrum {
 public:
-    OSL_HOSTDEVICE bb_spectrum (float temperature=5000) : m_temp(temperature) { }
+    OSL_HOSTDEVICE bb_spectrum(float temperature = 5000) : m_temp(temperature)
+    {
+    }

-    OSL_HOSTDEVICE float operator() (float wavelength_nm) const {
+    OSL_HOSTDEVICE float operator()(float wavelength_nm) const
+    {
        // Testing showed < .0000534058f differences between single and double
        // precision.  So choose to perform math using single precision.
-        float wlm = wavelength_nm * 1e-9f;   // Wavelength in meters
-        const float c1 = 3.74183e-16f; // 2*pi*h*c^2, W*m^2
-        const float c2 = 1.4388e-2f;   // h*c/k, m*K
-                                       // h is Planck's const, k is Boltzmann's
+        float wlm      = wavelength_nm * 1e-9f;  // Wavelength in meters
+        const float c1 = 3.74183e-16f;           // 2*pi*h*c^2, W*m^2
+        const float c2 = 1.4388e-2f;             // h*c/k, m*K
+            // h is Planck's const, k is Boltzmann's
        //const float inverse_of_wlm5 = std::pow(wlm,-5.0f);
        // Rather than calling powf, just implement exact power and inverse
        // directly here.  Confirmed with ICC that identical optimized assembly
        // is produced to std::powf and more SIMD friendly to other compilers.
-        const float wlm2 = wlm*wlm;
-        const float wlm4 = wlm2*wlm2;
-        const float wlm5 = wlm4*wlm;
-        const float inverse_of_wlm5 = 1.0f/wlm5;
+        const float wlm2            = wlm * wlm;
+        const float wlm4            = wlm2 * wlm2;
+        const float wlm5            = wlm4 * wlm;
+        const float inverse_of_wlm5 = 1.0f / wlm5;
        // Clang doesn't have vectorizing version of expm1, so use OIIO::fast_expm1
-        return float((c1 * inverse_of_wlm5) / OIIO::fast_expm1(c2 / (wlm * m_temp)));
+        return float((c1 * inverse_of_wlm5)
+                     / OIIO::fast_expm1(c2 / (wlm * m_temp)));
    }
+
 private:
    float m_temp;
 };
@@ -200,9 +218,9 @@ private:

 // Integrate the CIE color matching values, weighted by function
 // spec_intens(lambda_nm), returning the aggregate XYZ color.
-template<class SPECTRUM> OSL_HOSTDEVICE
-static Color3
-spectrum_to_XYZ (const SPECTRUM &spec_intens)
+template<class SPECTRUM>
+OSL_HOSTDEVICE static Color3
+spectrum_to_XYZ(const SPECTRUM& spec_intens)
 {
    float X = 0, Y = 0, Z = 0;
    const float dlambda = 5.0f * 1e-9;  // in meters
@@ -210,80 +228,81 @@ spectrum_to_XYZ (const SPECTRUM &spec_intens)
        float lambda = 380.0f + 5.0f * i;
        // N.B. spec_intens returns result in W/m^2 but it's a differential,
        // needs to be scaled by dlambda!
-        float Me = spec_intens(lambda) * dlambda;
-        constexpr int stride = sizeof(Color3)/sizeof(float);
-        const int si = stride*i;
+        float Me             = spec_intens(lambda) * dlambda;
+        constexpr int stride = sizeof(Color3) / sizeof(float);
+        const int si         = stride * i;
        X += Me * (cie_colour_match + 0)[si];
        Y += Me * (cie_colour_match + 1)[si];
        Z += Me * (cie_colour_match + 2)[si];
    }
-    return Color3 (X, Y, Z);
+    return Color3(X, Y, Z);
 }


-template <typename COLOR3> OSL_HOSTDEVICE
-static COLOR3
-hsv_to_rgb (const COLOR3& hsv)
+template<typename COLOR3>
+OSL_HOSTDEVICE static COLOR3
+hsv_to_rgb(const COLOR3& hsv)
 {
    // Reference for this technique: Foley & van Dam
    using FLOAT = typename ScalarFromVec<COLOR3>::type;
    FLOAT h = comp_x(hsv), s = comp_y(hsv), v = comp_z(hsv);
    if (s < 0.0001f) {
-      return make_Color3 (v, v, v);
+        return make_Color3(v, v, v);
    } else {
-        using std::floor;   // to pick up the float one
        using OIIO::ifloor;
-        h = 6.0f * (h - floor(h));  // expand to [0..6)
-        int hi = ifloor(h);
+        using std::floor;                 // to pick up the float one
+        h       = 6.0f * (h - floor(h));  // expand to [0..6)
+        int hi  = ifloor(h);
        FLOAT f = h - FLOAT(hi);
-        FLOAT p = v * (1.0f-s);
-        FLOAT q = v * (1.0f-s*f);
-        FLOAT t = v * (1.0f-s*(1.0f-f));
+        FLOAT p = v * (1.0f - s);
+        FLOAT q = v * (1.0f - s * f);
+        FLOAT t = v * (1.0f - s * (1.0f - f));
 #ifdef __OSL_WIDE_PVT
        // Avoid switch statement vectorizor doesn't like
        // Also avoid if/else nest which some optimizers might
        // convert back into a switch statement
-#   if OSL_ANY_CLANG && !OSL_INTEL_CLASSIC_COMPILER_VERSION && !OSL_INTEL_LLVM_COMPILER_VERSION
+#    if OSL_ANY_CLANG && !OSL_INTEL_CLASSIC_COMPILER_VERSION \
+        && !OSL_INTEL_LLVM_COMPILER_VERSION
        // Clang was still transforming series of if's back into a switch.
        // Alternate between == and <= comparisons to avoid
-#       define __OSL_ASC_EQ <=
-#   else
-#       define __OSL_ASC_EQ ==
-#   endif
+#        define __OSL_ASC_EQ <=
+#    else
+#        define __OSL_ASC_EQ ==
+#    endif
        if (hi == 0) {
-            return make_Color3 (v, t, p);
+            return make_Color3(v, t, p);
        }
        if (hi __OSL_ASC_EQ 1) {
-            return make_Color3 (q, v, p);
+            return make_Color3(q, v, p);
        }
        if (hi == 2) {
-            return make_Color3 (p, v, t);
+            return make_Color3(p, v, t);
        }
        if (hi __OSL_ASC_EQ 3) {
-            return make_Color3 (p, q, v);
+            return make_Color3(p, q, v);
        }
        if (hi == 4) {
-            return make_Color3 (t, p, v);
+            return make_Color3(t, p, v);
        }
-        return make_Color3 (v, p, q);
+        return make_Color3(v, p, q);
 #    undef __OSL_ASC_EQ
 #else
        // serial execution might be faster with switch
        switch (hi) {
-        case 0 : return make_Color3 (v, t, p);
-        case 1 : return make_Color3 (q, v, p);
-        case 2 : return make_Color3 (p, v, t);
-        case 3 : return make_Color3 (p, q, v);
-        case 4 : return make_Color3 (t, p, v);
-        default: return make_Color3 (v, p, q);
+        case 0: return make_Color3(v, t, p);
+        case 1: return make_Color3(q, v, p);
+        case 2: return make_Color3(p, v, t);
+        case 3: return make_Color3(p, q, v);
+        case 4: return make_Color3(t, p, v);
+        default: return make_Color3(v, p, q);
        }
 #endif
    }
 }

-template <typename COLOR3> OSL_HOSTDEVICE
-static inline COLOR3
-rgb_to_hsv (const COLOR3& rgb)
+template<typename COLOR3>
+OSL_HOSTDEVICE static inline COLOR3
+rgb_to_hsv(const COLOR3& rgb)
 {
    // See Foley & van Dam
    using FLOAT = typename ScalarFromVec<COLOR3>::type;
@@ -291,17 +310,16 @@ rgb_to_hsv (const COLOR3& rgb)
    // NOTE: use sfm::min_val and sfm::max_val instead of std::min and
    //       std::max which return references
    //       that can interfere with vectorization
-    FLOAT mincomp = sfm::min_val (r, sfm::min_val (g, b));
-    FLOAT maxcomp = sfm::max_val (r, sfm::max_val (g, b));
-    FLOAT delta = maxcomp - mincomp;  // chroma
-    FLOAT v = maxcomp;
-    FLOAT s = 0.0f;
+    FLOAT mincomp = sfm::min_val(r, sfm::min_val(g, b));
+    FLOAT maxcomp = sfm::max_val(r, sfm::max_val(g, b));
+    FLOAT delta   = maxcomp - mincomp;  // chroma
+    FLOAT v       = maxcomp;
+    FLOAT s       = 0.0f;
    if (maxcomp > 0.0f)
        s = delta / maxcomp;
    FLOAT h = 0.0f;
-    if (s > 0.0f)
-    {
-#if 0   // Reference version
+    if (s > 0.0f) {
+#if 0  // Reference version
        if      (r >= maxcomp) h = (g-b) / delta;
        else if (g >= maxcomp) h = 2.0f + (b-r) / delta;
        else                   h = 4.0f + (r-g) / delta;
@@ -314,57 +332,67 @@ rgb_to_hsv (const COLOR3& rgb)
        // which was already going to be a divisor.
        float k;
        FLOAT x, y;
-        if      (r >= maxcomp) { k = 0.0f/6.0f; x = g; y = b; }
-        else if (g >= maxcomp) { k = 2.0f/6.0f; x = b; y = r; }
-        else                   { k = 4.0f/6.0f; x = r; y = g; }
-        h = k + (x-y) / (6.0f*delta);
+        if (r >= maxcomp) {
+            k = 0.0f / 6.0f;
+            x = g;
+            y = b;
+        } else if (g >= maxcomp) {
+            k = 2.0f / 6.0f;
+            x = b;
+            y = r;
+        } else {
+            k = 4.0f / 6.0f;
+            x = r;
+            y = g;
+        }
+        h = k + (x - y) / (6.0f * delta);
 #endif
        if (h < 0.0f)
            h += 1.0f;
    }
-    return make_Color3 (h, s, v);
+    return make_Color3(h, s, v);
 }

-template <typename COLOR3> OSL_HOSTDEVICE
-static COLOR3
-hsl_to_rgb (const COLOR3& hsl)
+template<typename COLOR3>
+OSL_HOSTDEVICE static COLOR3
+hsl_to_rgb(const COLOR3& hsl)
 {
    using FLOAT = typename ScalarFromVec<COLOR3>::type;
    FLOAT h = comp_x(hsl), s = comp_y(hsl), l = comp_z(hsl);
    // Easiest to convert hsl -> hsv, then hsv -> RGB (per Foley & van Dam)
    FLOAT v = (l <= 0.5f) ? (l * (1.0f + s)) : (l * (1.0f - s) + s);
    if (v <= 0.0f) {
-        return make_Color3 (0.0f, 0.0f, 0.0f);
+        return make_Color3(0.0f, 0.0f, 0.0f);
    } else {
        FLOAT min = 2.0f * l - v;
-        s = (v - min) / v;
-        return hsv_to_rgb (make_Color3(h, s, v));
+        s         = (v - min) / v;
+        return hsv_to_rgb(make_Color3(h, s, v));
    }
 }

-template <typename COLOR3> OSL_HOSTDEVICE
-static COLOR3
-rgb_to_hsl (const COLOR3& rgb)
+template<typename COLOR3>
+OSL_HOSTDEVICE static COLOR3
+rgb_to_hsl(const COLOR3& rgb)
 {
    // See Foley & van Dam
    // First convert rgb to hsv, then to hsl
    using FLOAT = typename ScalarFromVec<COLOR3>::type;
    // NOTE: use sfm::min_val instead of std::min which returns references
    //       that can interfere with vectorization
-    FLOAT minval = sfm::min_val (comp_x(rgb), sfm::min_val (comp_y(rgb), comp_z(rgb)));
-    COLOR3 hsv = rgb_to_hsv (rgb);
-    FLOAT maxval = comp_z(hsv);   // v == maxval
-    FLOAT h = comp_x(hsv), s, l = (minval+maxval) / 2.0f;
-    if (equalVal (minval, maxval))
+    FLOAT minval = sfm::min_val(comp_x(rgb),
+                                sfm::min_val(comp_y(rgb), comp_z(rgb)));
+    COLOR3 hsv   = rgb_to_hsv(rgb);
+    FLOAT maxval = comp_z(hsv);  // v == maxval
+    FLOAT h = comp_x(hsv), s, l = (minval + maxval) / 2.0f;
+    if (equalVal(minval, maxval))
        s = 0.0f;  // special 'achromatic' case, hue is 0
-#if 0 // reference version
+#if 0              // reference version
    else if (l <= 0.5f)
        s = (maxval - minval) / (maxval + minval);
    else
        s = (maxval - minval) / (2.0f - maxval - minval)
 #else
-    else
-    {
+    else {
        // Share computation of min2max and divide vs. repeating in 2 branches
        // which both could execute masked
        FLOAT min2max = (maxval - minval);
@@ -376,90 +404,90 @@ rgb_to_hsl (const COLOR3& rgb)
        s = min2max / divisor;
    }
 #endif
-    return make_Color3 (h, s, l);
+    return make_Color3(h, s, l);
 }

-template <typename COLOR3> OSL_HOSTDEVICE
-static COLOR3
-YIQ_to_rgb (const COLOR3& YIQ)
+template<typename COLOR3>
+OSL_HOSTDEVICE static COLOR3
+YIQ_to_rgb(const COLOR3& YIQ)
 {
-    return YIQ * Matrix33(1.0000,  1.0000,  1.0000,
-                          0.9557, -0.2716, -1.1082,
-                          0.6199, -0.6469,  1.7051);
+    return YIQ
+           * Matrix33(1.0000, 1.0000, 1.0000, 0.9557, -0.2716, -1.1082, 0.6199,
+                      -0.6469, 1.7051);
 }

-template <typename COLOR3> OSL_HOSTDEVICE
-static COLOR3
-rgb_to_YIQ (const COLOR3& rgb)
+template<typename COLOR3>
+OSL_HOSTDEVICE static COLOR3
+rgb_to_YIQ(const COLOR3& rgb)
 {
-    return rgb * Matrix33(0.299,  0.596,  0.212,
-                          0.587, -0.275, -0.523,
-                          0.114, -0.321,  0.311);
+    return rgb
+           * Matrix33(0.299, 0.596, 0.212, 0.587, -0.275, -0.523, 0.114, -0.321,
+                      0.311);
 }

-template <typename COLOR3> OSL_HOSTDEVICE
-static COLOR3
-XYZ_to_xyY (const COLOR3 &XYZ)
+template<typename COLOR3>
+OSL_HOSTDEVICE static COLOR3
+XYZ_to_xyY(const COLOR3& XYZ)
 {
    using FLOAT = typename ScalarFromVec<COLOR3>::type;
-    FLOAT n = (comp_x(XYZ) + comp_y(XYZ) + comp_z(XYZ));
-    FLOAT n_inv = (n >= 1.0e-6f ? 1.0f/n : 0.0f);
-    return make_Color3 (comp_x(XYZ)*n_inv, comp_y(XYZ)*n_inv, comp_y(XYZ));
+    FLOAT n     = (comp_x(XYZ) + comp_y(XYZ) + comp_z(XYZ));
+    FLOAT n_inv = (n >= 1.0e-6f ? 1.0f / n : 0.0f);
+    return make_Color3(comp_x(XYZ) * n_inv, comp_y(XYZ) * n_inv, comp_y(XYZ));
    // N.B. http://brucelindbloom.com/ suggests returning xy of the
    // reference white in the X+Y+Z==0 case.
 }

-template <typename COLOR3> OSL_HOSTDEVICE
-static COLOR3
-xyY_to_XYZ (const COLOR3 &xyY)
+template<typename COLOR3>
+OSL_HOSTDEVICE static COLOR3
+xyY_to_XYZ(const COLOR3& xyY)
 {
    using FLOAT = typename ScalarFromVec<COLOR3>::type;
-    FLOAT Y = comp_z(xyY);
-    FLOAT Y_y = (comp_y(xyY) > 1.0e-6f ? Y/comp_y(xyY) : 0.0f);
-    FLOAT X = Y_y * comp_x(xyY);
-    FLOAT Z = Y_y * (1.0f - comp_x(xyY) - comp_y(xyY));
-    return make_Color3 (X, Y, Z);
+    FLOAT Y     = comp_z(xyY);
+    FLOAT Y_y   = (comp_y(xyY) > 1.0e-6f ? Y / comp_y(xyY) : 0.0f);
+    FLOAT X     = Y_y * comp_x(xyY);
+    FLOAT Z     = Y_y * (1.0f - comp_x(xyY) - comp_y(xyY));
+    return make_Color3(X, Y, Z);
 }

-template <typename COLOR3> OSL_HOSTDEVICE
-static COLOR3
-sRGB_to_linear (const COLOR3& srgb)
+template<typename COLOR3>
+OSL_HOSTDEVICE static COLOR3
+sRGB_to_linear(const COLOR3& srgb)
 {
    // See Foley & van Dam
    using FLOAT = typename ScalarFromVec<COLOR3>::type;
    using namespace OIIO;
    //using safe_pow = std::conditional<is_Dual<COLOR3>::value, OSL::safe_pow, OIIO::safe_pow>::type;
    FLOAT r = comp_x(srgb), g = comp_y(srgb), b = comp_z(srgb);
-    auto convert = [] (FLOAT x) -> FLOAT {
-        return (x <= 0.04045f) ?
-                     (x * (1.0f / 12.92f)) :
-                     safe_pow((x + 0.055f) * (1.0f / 1.055f), FLOAT(2.4f));
+    auto convert = [](FLOAT x) -> FLOAT {
+        return (x <= 0.04045f)
+                   ? (x * (1.0f / 12.92f))
+                   : safe_pow((x + 0.055f) * (1.0f / 1.055f), FLOAT(2.4f));
    };
-    return make_Color3 (convert(r), convert(g), convert(b));
+    return make_Color3(convert(r), convert(g), convert(b));
 }

-template <typename COLOR3> OSL_HOSTDEVICE
-static COLOR3
-linear_to_sRGB (const COLOR3& rgb)
+template<typename COLOR3>
+OSL_HOSTDEVICE static COLOR3
+linear_to_sRGB(const COLOR3& rgb)
 {
    // See Foley & van Dam
    using FLOAT = typename ScalarFromVec<COLOR3>::type;
    using namespace OIIO;
    //using safe_pow = std::conditional<is_Dual<COLOR3>::value, OSL::safe_pow, OIIO::safe_pow>::type;
    FLOAT r = comp_x(rgb), g = comp_y(rgb), b = comp_z(rgb);
-    auto convert = [] (FLOAT x) -> FLOAT {
-        return (x <= 0.0031308f) ?
-                      (12.92f * x)           :
-                      (1.055f * safe_pow(x, FLOAT(1.f / 2.4f)) - 0.055f);
+    auto convert = [](FLOAT x) -> FLOAT {
+        return (x <= 0.0031308f)
+                   ? (12.92f * x)
+                   : (1.055f * safe_pow(x, FLOAT(1.f / 2.4f)) - 0.055f);
    };

-    return make_Color3 (convert(r), convert(g), convert(b));
+    return make_Color3(convert(r), convert(g), convert(b));
 }

 OSL_HOSTDEVICE inline Color3
-colpow (const Color3 &c, float p)
+colpow(const Color3& c, float p)
 {
-    return Color3 (powf(c.x,p), powf(c.y,p), powf(c.z,p));
+    return Color3(powf(c.x, p), powf(c.y, p), powf(c.z, p));
 }

 }  // namespace __OSL_WIDE_PVT or pvt
@@ -468,11 +496,11 @@ namespace pvt {


 #ifdef __OSL_WIDE_PVT
-    using __OSL_WIDE_PVT::clamp_zero;
-    using __OSL_WIDE_PVT::bb_spectrum;
-    using __OSL_WIDE_PVT::spectrum_to_XYZ;
-    using __OSL_WIDE_PVT::BB_TABLE_MAP;
-    using __OSL_WIDE_PVT::BB_TABLE_UNMAP;
+using __OSL_WIDE_PVT::bb_spectrum;
+using __OSL_WIDE_PVT::BB_TABLE_MAP;
+using __OSL_WIDE_PVT::BB_TABLE_UNMAP;
+using __OSL_WIDE_PVT::clamp_zero;
+using __OSL_WIDE_PVT::spectrum_to_XYZ;
 #endif


@@ -483,15 +511,15 @@ ColorSystem::can_lookup_blackbody(float T /*Kelvin*/) const
 }

 OSL_HOSTDEVICE Color3
-ColorSystem::lookup_blackbody_rgb (float T /*Kelvin*/) const
+ColorSystem::lookup_blackbody_rgb(float T /*Kelvin*/) const
 {
    if (T < BB_DRAPER)
-        return Color3(1.0e-6f,0.0f,0.0f);  // very very dim red
+        return Color3(1.0e-6f, 0.0f, 0.0f);  // very very dim red

    // can_lookup_blackbody() should have checked before now
    OSL_DASSERT(T < BB_MAX_TABLE_RANGE);
-    float t = BB_TABLE_UNMAP(T);
-    int ti = static_cast<int>(t);
+    float t         = BB_TABLE_UNMAP(T);
+    int ti          = static_cast<int>(t);
    float remainder = t - ti;
 #ifdef __OSL_WIDE_PVT
    // Choose to access Color3 lookup table as 3 float components.
@@ -500,40 +528,40 @@ ColorSystem::lookup_blackbody_rgb (float T /*Kelvin*/) const
    // dereferencing Color3 which was using 64bit gathers which can take
    // multiple registers to hold the 64bit offsets and multiple gather
    // instructions who's results need to be merged into a single register.
-    const float *blackbody_components = reinterpret_cast<const float *>(m_blackbody_table);
-    constexpr int stride = sizeof(Color3)/sizeof(float);
-    int sti = stride*ti;
-    Color3 rgb = OIIO::lerp (
-        // Have all gathers use the same indices by having different base registers
+    const float* blackbody_components = reinterpret_cast<const float*>(
+        m_blackbody_table);
+    constexpr int stride = sizeof(Color3) / sizeof(float);
+    int sti              = stride * ti;
+    Color3 rgb           = OIIO::lerp(
+                  // Have all gathers use the same indices by having different base registers
        // for each of the 6 components
-        Color3((blackbody_components+0)[sti],
-               (blackbody_components+1)[sti],
-               (blackbody_components+2)[sti]),
-        Color3((blackbody_components+3)[sti],
-               (blackbody_components+4)[sti],
-               (blackbody_components+5)[sti]),
-       remainder);
+        Color3((blackbody_components + 0)[sti], (blackbody_components + 1)[sti],
+                         (blackbody_components + 2)[sti]),
+        Color3((blackbody_components + 3)[sti], (blackbody_components + 4)[sti],
+                         (blackbody_components + 5)[sti]),
+        remainder);
 #else
-    Color3 rgb = OIIO::lerp (m_blackbody_table[ti], m_blackbody_table[ti+1], remainder);
+    Color3 rgb = OIIO::lerp(m_blackbody_table[ti], m_blackbody_table[ti + 1],
+                            remainder);
 #endif
    //return colpow(rgb, BB_TABLE_YPOWER);
    Color3 rgb2 = rgb * rgb;
    Color3 rgb4 = rgb2 * rgb2;
-    return rgb4 * rgb; // ^5
+    return rgb4 * rgb;  // ^5
 }

 OSL_HOSTDEVICE Color3
-ColorSystem::compute_blackbody_rgb (float T /*Kelvin*/) const
+ColorSystem::compute_blackbody_rgb(float T /*Kelvin*/) const
 {
    // compute for real
-    bb_spectrum spec (T);
-    Color3 rgb = XYZ_to_RGB (spectrum_to_XYZ (spec));
-    clamp_zero (rgb);
+    bb_spectrum spec(T);
+    Color3 rgb = XYZ_to_RGB(spectrum_to_XYZ(spec));
+    clamp_zero(rgb);
    return rgb;
 }

 OSL_HOSTDEVICE Color3
-ColorSystem::blackbody_rgb (float T) const
+ColorSystem::blackbody_rgb(float T) const
 {
    if (can_lookup_blackbody(T)) {
        return lookup_blackbody_rgb(T);
@@ -544,4 +572,3 @@ ColorSystem::blackbody_rgb (float T) const

 }  // namespace pvt
 OSL_NAMESPACE_EXIT
-
@@ -138,10 +138,10 @@ osl_transformn_dvmdv(void* result, void* M_, void* v_)
 OSL_SHADEOP int
 osl_get_matrix(void* sg_, void* r, const char* from)
 {
-    ShaderGlobals* sg   = (ShaderGlobals*)sg_;
-    ShadingContext* ctx = (ShadingContext*)sg->context;
+    ShaderGlobals* sg        = (ShaderGlobals*)sg_;
+    ShadingStateUniform* ssu = (ShadingStateUniform*)sg->shadingStateUniform;
    if (HDSTR(from) == STRING_PARAMS(common)
-        || HDSTR(from) == ctx->shadingsys().commonspace_synonym()) {
+        || HDSTR(from) == ssu->m_commonspace_synonym) {
        MAT(r).makeIdentity();
        return true;
    }
@@ -157,7 +157,7 @@ osl_get_matrix(void* sg_, void* r, const char* from)
    if (!ok) {
        MAT(r).makeIdentity();
        ShadingContext* ctx = (ShadingContext*)((ShaderGlobals*)sg)->context;
-        if (ctx->shadingsys().unknown_coordsys_error())
+        if (ssu->m_unknown_coordsys_error)
            ctx->errorfmt("Unknown transformation \"{}\"", from);
    }
    return ok;
@@ -168,10 +168,10 @@ osl_get_matrix(void* sg_, void* r, const char* from)
 OSL_SHADEOP int
 osl_get_inverse_matrix(void* sg_, void* r, const char* to)
 {
-    ShaderGlobals* sg   = (ShaderGlobals*)sg_;
-    ShadingContext* ctx = (ShadingContext*)sg->context;
+    ShaderGlobals* sg        = (ShaderGlobals*)sg_;
+    ShadingStateUniform* ssu = (ShadingStateUniform*)sg->shadingStateUniform;
    if (HDSTR(to) == STRING_PARAMS(common)
-        || HDSTR(to) == ctx->shadingsys().commonspace_synonym()) {
+        || HDSTR(to) == ssu->m_commonspace_synonym) {
        MAT(r).makeIdentity();
        return true;
    }
@@ -189,7 +189,7 @@ osl_get_inverse_matrix(void* sg_, void* r, const char* to)
    if (!ok) {
        MAT(r).makeIdentity();
        ShadingContext* ctx = (ShadingContext*)((ShaderGlobals*)sg)->context;
-        if (ctx->shadingsys().unknown_coordsys_error())
+        if (ssu->m_unknown_coordsys_error)
            ctx->errorfmt("Unknown transformation \"{}\"", to);
    }
    return ok;
@@ -207,18 +207,21 @@ osl_get_inverse_matrix(void* sg_, void* r, const char* to);


 OSL_SHADEOP OSL_HOSTDEVICE int
-osl_prepend_matrix_from(void* sg, void* r, const char* from)
+osl_prepend_matrix_from(void* sg_, void* r, const char* from)
 {
+    ShaderGlobals* sg = (ShaderGlobals*)sg_;
    Matrix44 m;
-    bool ok = osl_get_matrix((ShaderGlobals*)sg, &m, from);
+    bool ok = osl_get_matrix(sg, &m, from);
    if (ok)
        MAT(r) = m * MAT(r);
 #ifndef __CUDACC__
    // TODO: How do we manage this in OptiX?
    else {
-        ShadingContext* ctx = (ShadingContext*)((ShaderGlobals*)sg)->context;
-        if (ctx->shadingsys().unknown_coordsys_error())
+        ShadingStateUniform* ssu = (ShadingStateUniform*)sg->shadingStateUniform;
+        if (ssu->m_unknown_coordsys_error) {
+            ShadingContext* ctx = (ShadingContext*)(sg)->context;
            ctx->errorfmt("Unknown transformation \"{}\"", from);
+        }
    }
 #endif
    return ok;
@@ -111,7 +111,7 @@ osl_getmessage(ShaderGlobals* sg, const char* source_, const char* name_,
                sg->context->errorfmt(
                    "message \"{}\" was set by layer #{} ({}:{})"
                    " but is being queried by layer #{} ({}:{})"
-                    " - messages may only be transfered from nodes "
+                    " - messages may only be transferred from nodes "
                    "that appear earlier in the shading network",
                    name, m->layeridx, m->sourcefile, m->sourceline, layeridx,
                    sourcefile, sourceline);
--- a/Show More
+++ b/Show More