Updates since review comments.

Author: Colin Davidson

scripts/PR_text.txt

@@ -0,0 +1,48 @@
+Bring Native CPU pipeline files into dpc++ directly.
+
+This brings the Native CPU compiler pipeline files directly under the ownership
+of intel/llvm. This removes the direct dependence on the oneAPI Construction
+Kit, although the history of those files still exists under intel/llvm and the
+originals still exist at
+https://github.com/uxlfoundation/oneapi-construction-kit.
+
+This was done as follows:
+
+In the oneAPI Construction Kit repo:
+
+```
+    git-filter-repo --paths-from-file $script_dir/native_cpu_paths.txt \
+    --path-rename modules/compiler/compiler_pipeline:llvm/lib/SYCLNativeCPUUtils/compiler_passes/compiler_pipeline \
+    --path-rename modules/compiler/vecz:llvm/lib/SYCLNativeCPUUtils/compiler_passes/vecz \
+    --path-rename modules/compiler/multi_llvm/include/multi_llvm:llvm/lib/SYCLNativeCPUUtils/compiler_passes/compiler_pipeline/include/multi_llvm \
+    --path-rename doc/modules/vecz/vecz.md:llvm/lib/SYCLNativeCPUUtils/compiler_passes/vecz/vecz.md \
+    --path-rename doc/modules/compiler/overview.rst:llvm/lib/SYCLNativeCPUUtils/compiler_passes/compiler_pipeline/docs/overview.rst \
+    --path-rename doc/modules/compiler/utils.rst:llvm/lib/SYCLNativeCPUUtils/compiler_passes/compiler_pipeline/docs/utils.rst
+```
+See https://github.com/uxlfoundation/oneapi-construction-kit/pull/969 for more info.
+
+Under intel/llvm:
+
+```
+    git checkout sycl
+    git checkout -b $llvm_branch
+    git fetch $ock_fork $ock_branch
+    # Update CMake to be more llvm like
+    # clang-format-20 on added files
+    # Add some new documentation to reflect the state of play.
+    git merge -m "[NATIVE_CPU][SYCL] Merge from oneAPI Construction Kit into native_cpu" --allow-unrelated-histories ock_fork/$ock_branch
+```
+
+
+`llvm/lib/SYCLNativeCPUUtils/CMakeLists.txt` was updated to remove the fetch
+content and the ugly install workaround. Missing is the lit tests for the
+pipeline tests as these refer to the use of a non-ported tool, `muxc`.
+CMakeLists.txt across was updated to use LLVM macros across `compiler_pipeline`
+and `vecz` ported code, as well as changing `NATIVECPU_USE_OCK` to be the
+default. It also applies clang-format-20 to all the .h and .cpp files brought
+in. Other than that the .cpp and .h files are unchanged. The lit tests cfg files
+were updated to fit with LLVM, and a small change was made to avoid use of
+`undef`.
+
+A brief explanation and limitations of the importing of the files are under
+`llvm/lib/SYCLNativeCPUUtils/compiler_passes/compiler_passes.rst`.

scripts/commit.txt

@@ -41,7 +41,7 @@ if [ "$ock_branch" != "" ]; then
         git fetch ock_fork $ock_branch
         git merge -m "[NATIVE_CPU][SYCL] Merge from oneAPI Construction Kit into native_cpu" --allow-unrelated-histories ock_fork/$ock_branch
         cp $script_dir/native_cpu_CMakeLists.txt $llvm_repo/llvm/lib/SYCLNativeCPUUtils/CMakeLists.txt
-        sed -i "s/DNATIVECPU_USE_OCK=Off/DNATIVECPU_USE_OCK=ON/" $llvm_repo/.github/workflows/sycl-linux-build.yml
+        sed -i -e 's/DLLVM_INSTALL_UTILS=ON \\/DLLVM_INSTALL_UTILS=ON/' -e "/DNATIVECPU_USE_OCK=Off/d" $llvm_repo/.github/workflows/sycl-linux-build.yml
         sed -i -e "s/config.name = 'Vecz'/config.name = \"Vecz\"/" -e "s/config.suffixes = \['.hlsl', '.ll'\]/config.suffixes = \[\".hlsl\", \".ll\"\]/" $llvm_repo/llvm/lib/SYCLNativeCPUUtils/compiler_passes/vecz/test/lit/lit.cfg.py
         for f in $( find $llvm_repo/llvm/lib/SYCLNativeCPUUtils/compiler_passes -name '*.h' -o -name '*.cpp' -o -name '*.inc'); do
            clang-format $f > /tmp/format.txt; cp /tmp/format.txt $f
@@ -57,8 +57,6 @@ if [ "$ock_branch" != "" ]; then
         cp $script_dir/native_cpu_vecz_test_lit_lit.cfg.py llvm/lib/SYCLNativeCPUUtils/compiler_passes/vecz/test/lit/lit.cfg.py
         cp $script_dir/native_cpu_vecz_test_lit_lit.site.cfg.py.in llvm/lib/SYCLNativeCPUUtils/compiler_passes/vecz/test/lit/lit.site.cfg.py.in
         cp $script_dir/native_cpu_vecz_test_lit_llvm_RISCV_lit.local.cfg llvm/lib/SYCLNativeCPUUtils/compiler_passes/vecz/test/lit/llvm/RISCV/lit.local.cfg
-        git rm $llvm_repo/llvm/lib/SYCLNativeCPUUtils/compiler_passes/vecz/test/lit/llvm/partial_linearization22-llvm18.ll
-        git rm $llvm_repo/llvm/lib/SYCLNativeCPUUtils/compiler_passes/vecz/test/lit/llvm/ScalableVectors/lit.local.cfg
         sed -i 's/{{undef|poison}}/poison/g' llvm/lib/SYCLNativeCPUUtils/compiler_passes/vecz/test/lit/llvm/packetization_debug_info.ll
         git add llvm/lib/SYCLNativeCPUUtils .github/workflows/sycl-linux-build.yml
         git commit -F $script_dir/native_cpu_commit_message.txt

scripts/create_native_cpu.sh

@@ -1,6 +1,6 @@
 [NATIVE_CPU][SYCL] Switch to using native_cpu compiler pipeline inline
 
-This brings the native cpu compiler pipeline files directly under the ownership
+This brings the Native CPU compiler pipeline files directly under the ownership
 of intel/llvm. This removes the direct dependence on the oneAPI Construction
 Kit, although the history of those files still exists under intel/llvm and the
 originals still exist at
@@ -9,15 +9,15 @@ https://github.com/uxlfoundation/oneapi-construction-kit.
 This is the post merge update to the the oneAPI Construction Kit move of the compiler
 pipeline files for Native CPU and replaces the original FetchContent method.
 
-This updates the CMakeLists.txt to be appropriate for llvm components,
-rather in the style of oneAPI Construction Kit. It also applies clang-format-20 to all the
-.h and .cpp files brought in. Other than that the .cpp and .h files are unchanged.
-
-A brief explanation and limitations of the importing of the files are under l
-lvm/lib/SYCLNativeCPUUtils/compiler_passes/compiler_passes.rst.
-
-It also updates the NATIVECPU_USE_OCK to be on in sycl-linux-build.yml. There is also a
-check-sycl-vecz target added for the vecz lit tests.
-
-
-
+`llvm/lib/SYCLNativeCPUUtils/CMakeLists.txt` was updated to remove the fetch
+content and the ugly install workaround. Missing is the lit tests for the
+pipeline tests as these refer to the use of a non-ported tool, `muxc`.
+CMakeLists.txt across was updated to use LLVM macros across `compiler_pipeline`
+and `vecz` ported code, as well as changing `NATIVECPU_USE_OCK` to be the
+default. It also applies clang-format-20 to all the .h and .cpp files brought
+in. Other than that the .cpp and .h files are unchanged. The lit tests cfg files
+were updated to fit with LLVM, and a small change was made to avoid use of
+`undef`.
+
+A brief explanation and limitations of the importing of the files are under
+`llvm/lib/SYCLNativeCPUUtils/compiler_passes/compiler_passes.rst`.

scripts/native_cpu_commit_message.txt

@@ -4,15 +4,15 @@ Compiler passes
 Introduction
 ------------
 
-Files under this directory are ported from the [oneAPI Construction
-Kit](https://github.com/uxlfoundation/oneapi-construction-kit). They are used by
-NativeCPU to help create a pipeline for turning a base kernel into something
-which can be executed across multiple work items, including auto-vectorization.
+Files under this directory are ported from the `oneAPI Construction Kit`_. They
+are used by Native CPU to help create a pipeline for turning a base kernel into
+something which can be executed across multiple work items, including
+auto-vectorization.
 
 These files are largely from the sub-directories
 **modules/compiler/compiler_pipeline**, **modules/compiler/vecz** and
 **modules/compiler/multi_llvm**. Only files that are used have been copied and
-the **Cmake** files have been updated to fit in with LLVM components.
+the **CMake** files have been updated to fit in with LLVM components.
 
 These sub-directories are used as follows:
 
@@ -50,15 +50,16 @@ copied across.
   these directories after the port.
 * **vecz** has a test tool **veczc** and associated **lit** tests. This tool if
   required should be moved under **llvm/tools** or **llvm/test**. This is also
-  requires `NATIVE_CPU_BUILD_VECZ_TEST_TOOLS` cmake option to build. This can be
+  requires `NATIVE_CPU_BUILD_VECZ_TEST_TOOLS` **CMake** option to build. This can be
   run using the target `check-sycl-vecz`.
 * **compiler_pipeline** has lit tests for the passes which have not been ported
   across. This is because they use a tool **muxc**, but these passes should be
-  able to be tested using opt. These lit tests can be found
-  [here](https://github.com/uxlfoundation/oneapi-construction-kit/tree/main/modules/compiler/test/lit/passes).
+  able to be tested using **opt**. These lit tests can be found in the
+  `pipeline pass tests`_.
 * There are many files that are unlikely to have any code coverage but because
   there are referred to in other files which we do need, they exist here. These
   should be pruned over time as a better understanding is made of what is
   required.
 
-.. _oneAPI Construction Kit: https://github.com/uxlfoundation/oneapi-construction-kit
+.. _oneAPI Construction Kit: https://github.com/uxlfoundation/oneapi-construction-kit
+.. _pipeline pass tests: https://github.com/uxlfoundation/oneapi-construction-kit/tree/main/modules/compiler/test/lit/passes

scripts/native_cpu_compiler_passes.rst

@@ -11,7 +11,7 @@ config.test_exec_root = r"@CURRENT_BINARY_DIR@"
 # Paths to helper utilities
 config.tools = [ ToolSubst('veczc') ]
 
-config.root.targets = frozenset('@LLVM_TARGETS_TO_BUILD@'.split(';'))
+config.targets = frozenset('@LLVM_TARGETS_TO_BUILD@'.split(';'))
 
 config.llvm_tools_dir = lit_config.substitute(path(r"@LLVM_TOOLS_DIR@"))