Merge remote-tracking branch 'intel/sycl' into steffen/group_algs_sup…

…port_shuffles

.clang-tidy

@@ -1,4 +1,4 @@
-Checks: '-*,clang-diagnostic-*,llvm-*,misc-*,-misc-const-correctness,-misc-unused-parameters,-misc-non-private-member-variables-in-classes,-misc-no-recursion,-misc-use-anonymous-namespace,readability-identifier-naming'
+Checks: '-*,clang-diagnostic-*,llvm-*,misc-*,-misc-const-correctness,-misc-unused-parameters,-misc-non-private-member-variables-in-classes,-misc-no-recursion,-misc-use-anonymous-namespace,readability-identifier-naming,-misc-include-cleaner'
 CheckOptions:
   - key:             readability-identifier-naming.ClassCase
     value:           CamelCase

.github/ISSUE_TEMPLATE/2-bug-report-cuda.yml

@@ -1,5 +1,8 @@
 name: CUDA-specific bug report
-description: Please use this template in case of CUDA-specific issue
+description: >
+  Please use this template in case of CUDA-specific issue.
+  Check https://developer.codeplay.com/products/oneapi/nvidia/latest/guides/troubleshooting
+  before submitting a bug report.
 labels: ["bug", "cuda"]
 body:
   - type: textarea

.github/ISSUE_TEMPLATE/3-bug-report-hip.yml

@@ -1,5 +1,8 @@
 name: HIP-specific bug report
-description: Please use this template in case of HIP-specific issue
+description: >
+  Please use this template in case of HIP-specific issue.
+  Check https://developer.codeplay.com/products/oneapi/amd/latest/guides/troubleshooting
+  before submitting a bug report.
 labels: ["bug", "hip"]
 body:
   - type: textarea

.github/new-prs-labeler.yml

@@ -846,9 +846,32 @@ backend:PowerPC:
   - clang/lib/Driver/ToolChains/Arch/PPC.*
   - clang/test/CodeGen/PowerPC/**
 
+backend:SystemZ:
+  - llvm/include/llvm/BinaryFormat/ELFRelocs/SystemZ*
+  - llvm/include/llvm/BinaryFormat/GOFF.h
+  - llvm/include/llvm/IR/IntrinsicsSystemZ.td
+  - llvm/lib/Target/SystemZ/**
+  - llvm/test/Analysis/**/SystemZ/**
+  - llvm/test/CodeGen/SystemZ/**
+  - llvm/test/DebugInfo/SystemZ/**
+  - llvm/test/ExecutionEngine/**/SystemZ/**
+  - llvm/test/MC/Disassembler/SystemZ/**
+  - llvm/test/MC/GOFF/**
+  - llvm/test/MC/SystemZ/**
+  - llvm/test/Transforms/**/SystemZ/**
+  - clang/include/clang/Basic/BuiltinsSystemZ.*
+  - clang/lib/Basic/Targets/SystemZ.*
+  - clang/lib/CodeGen/Targets/SystemZ.cpp
+  - clang/lib/Driver/ToolChains/ZOS*
+  - clang/lib/Driver/ToolChains/Arch/SystemZ.*
+  - clang/test/CodeGen/SystemZ/**
+
 third-party:unittests:
   - third-party/unittests/**
 
+third-party:benchmark:
+  - third-party/benchmark/**
+
 llvm:binary-utilities:
   - llvm/docs/CommandGuide/llvm-*
   - llvm/include/llvm/BinaryFormat/**

.github/workflows/build-ci-container.yml

@@ -77,14 +77,18 @@ jobs:
           cp ./.github/workflows/containers/github-action-ci/storage.conf ~/.config/containers/storage.conf
           podman info
 
+        # Download the container image into /mnt/podman rather than
+        # $GITHUB_WORKSPACE to avoid space limitations on the default drive
+        # and use the permissions setup for /mnt/podman.
       - name: Download stage1-toolchain
         uses: actions/download-artifact@v4
         with:
           name: stage1-toolchain
+          path: /mnt/podman
 
       - name: Load stage1-toolchain
         run: |
-          podman load -i stage1-toolchain.tar
+          podman load -i /mnt/podman/stage1-toolchain.tar
 
       - name: Build Container
         working-directory: ./.github/workflows/containers/github-action-ci/

.github/workflows/containers/github-action-ci/stage2.Dockerfile

@@ -12,11 +12,13 @@ COPY --from=stage2-toolchain $LLVM_SYSROOT $LLVM_SYSROOT
 # Need to install curl for hendrikmuhs/ccache-action
 # Need nodejs for some of the GitHub actions.
 # Need perl-modules for clang analyzer tests.
+# Need git for SPIRV-Tools tests.
 RUN apt-get update && \
     apt-get install -y \
     binutils \
     cmake \
     curl \
+    git \
     libstdc++-11-dev \
     ninja-build \
     nodejs \

.github/workflows/sycl-linux-run-tests.yml

@@ -297,17 +297,39 @@ jobs:
       if: inputs.tests_selector == 'cts'
       env:
         ONEAPI_DEVICE_SELECTOR: ${{ inputs.target_devices }}
+      # By-default GitHub actions execute the "run" shell script with -e option,
+      # so the execution terminates if any command returns a non-zero status.
+      # Since we're using a loop to run all test-binaries separately, some test
+      # may fail and terminate the execution. Setting "shell" value to override
+      # the default behavior.
+      # https://docs.github.com/en/actions/using-workflows/workflow-syntax-for-github-actions#custom-shell
+      shell: bash {0}
       # FIXME: For some reason the "sub_group api" test-case is failing with
       # SIGSEGV while running test_all, so running each binary separately.
       # BTW test_all requires a lot of resources to build it, so probably it'll
       # be better to build each binary also separately.
       # run: |
       #   ./build-cts/bin/test_all $CTS_TEST_ARGS
       run: |
+        status=""
+        failed_suites=""
         for i in `ls -1 ./build-cts/bin`; do
           if [ "$i" != "test_all" ]; then
             echo "::group::Running $i"
             build-cts/bin/$i
+            if [ $? -ne 0 ]; then
+              status=1
+              if [$failed_suites == ""]; then
+                failed_suites=$i
+              else
+                failed_suites="$failed_suites, $i"
+              fi
+            fi
             echo "::endgroup::"
           fi
         done
+        if [ -n "$status" ]; then
+          echo "Failed suite(s): $failed_suites"
+          exit 1
+        fi
+        exit 0

.github/workflows/sycl-nightly.yml

@@ -33,47 +33,54 @@ jobs:
             image: ghcr.io/intel/llvm/ubuntu2204_build:latest
             image_options: -u 1001 --device=/dev/dri --device=/dev/kfd
             target_devices: ext_oneapi_hip:gpu
+            tests_selector: e2e
 
           - name: Intel L0 GPU
             runner: '["Linux", "gen12"]'
             image: ghcr.io/intel/llvm/ubuntu2204_intel_drivers:latest
             image_options: -u 1001 --device=/dev/dri --privileged --cap-add SYS_ADMIN
             target_devices: ext_oneapi_level_zero:gpu
             reset_gpu: true
+            tests_selector: e2e
 
           - name: Intel OCL GPU
             runner: '["Linux", "gen12"]'
             image: ghcr.io/intel/llvm/ubuntu2204_intel_drivers:latest
             image_options: -u 1001 --device=/dev/dri --privileged --cap-add SYS_ADMIN
             target_devices: opencl:gpu
             reset_gpu: true
+            tests_selector: e2e
 
           - name: OCL CPU (AMD)
             runner: '["Linux", "amdgpu"]'
             image: ghcr.io/intel/llvm/ubuntu2204_intel_drivers:latest
             image_options: -u 1001
             target_devices: opencl:cpu
+            tests_selector: e2e
 
           - name: OCL CPU (Intel/GEN12)
             runner: '["Linux", "gen12"]'
             image: ghcr.io/intel/llvm/ubuntu2204_intel_drivers:latest
             image_options: -u 1001 --privileged --cap-add SYS_ADMIN
             target_devices: opencl:cpu
+            tests_selector: e2e
 
           - name: OCL CPU (Intel/Arc)
             runner: '["Linux", "arc"]'
             image: ghcr.io/intel/llvm/ubuntu2204_intel_drivers:latest
             image_options: -u 1001
             target_devices: opencl:cpu
+            tests_selector: e2e
 
           - name: Self-hosted CUDA
             runner: '["Linux", "cuda"]'
             image: ghcr.io/intel/llvm/ubuntu2204_build:latest
             image_options: -u 1001 --gpus all --cap-add SYS_ADMIN
             target_devices: ext_oneapi_cuda:gpu
+            tests_selector: e2e
 
           - name: SYCL-CTS
-            runner: '["cts-cpu"]'
+            runner: '["Linux", "gen12"]'
             image: ghcr.io/intel/llvm/ubuntu2204_intel_drivers:latest
             image_options: -u 1001 --device=/dev/dri --privileged --cap-add SYS_ADMIN
             target_devices: opencl:cpu

bolt/include/bolt/Core/BinaryContext.h

@@ -997,6 +997,10 @@ class BinaryContext {
     return getUniqueSectionByName(".gdb_index");
   }
 
+  ErrorOr<BinarySection &> getDebugNamesSection() const {
+    return getUniqueSectionByName(".debug_names");
+  }
+
   /// @}
 
   /// Register \p TargetFunction as a fragment of \p Function if checks pass:

bolt/include/bolt/Core/BinaryFunction.h

@@ -2056,6 +2056,14 @@ class BinaryFunction {
   /// Returns false if disassembly failed.
   Error disassemble();
 
+  /// An external interface to register a branch while the function is in
+  /// disassembled state. Allows to make custom modifications to the
+  /// disassembler. E.g., a pre-CFG pass can add an instruction and register
+  /// a branch that will later be used during the CFG construction.
+  ///
+  /// Return a label at the branch destination.
+  MCSymbol *registerBranch(uint64_t Src, uint64_t Dst);
+
   Error handlePCRelOperand(MCInst &Instruction, uint64_t Address,
                            uint64_t Size);
 

bolt/include/bolt/Core/DIEBuilder.h

@@ -16,6 +16,7 @@
 #define BOLT_CORE_DIE_BUILDER_H
 
 #include "bolt/Core/BinaryContext.h"
+#include "bolt/Core/DebugNames.h"
 #include "llvm/CodeGen/DIE.h"
 #include "llvm/DebugInfo/DWARF/DWARFAbbreviationDeclaration.h"
 #include "llvm/DebugInfo/DWARF/DWARFDie.h"
@@ -127,6 +128,7 @@ class DIEBuilder {
   DWARFUnit *SkeletonCU{nullptr};
   uint64_t UnitSize{0};
   llvm::DenseSet<uint64_t> AllProcessed;
+  DWARF5AcceleratorTable &DebugNamesTable;
 
   /// Returns current state of the DIEBuilder
   State &getState() { return *BuilderState.get(); }
@@ -206,8 +208,8 @@ class DIEBuilder {
   /// Update references once the layout is finalized.
   void updateReferences();
 
-  /// Update the Offset and Size of DIE.
-  uint32_t computeDIEOffset(const DWARFUnit &CU, DIE &Die, uint32_t &CurOffset);
+  /// Update the Offset and Size of DIE, populate DebugNames table.
+  uint32_t finalizeDIEs(DWARFUnit &CU, DIE &Die, uint32_t &CurOffset);
 
   void registerUnit(DWARFUnit &DU, bool NeedSort);
 
@@ -269,6 +271,7 @@ class DIEBuilder {
 
 public:
   DIEBuilder(BinaryContext &BC, DWARFContext *DwarfContext,
+             DWARF5AcceleratorTable &DebugNamesTable,
              DWARFUnit *SkeletonCU = nullptr);
 
   /// Returns enum to what we are currently processing.

bolt/include/bolt/Core/DebugData.h

@@ -439,6 +439,8 @@ class DebugStrOffsetsWriter {
   /// Update Str offset in .debug_str in .debug_str_offsets.
   void updateAddressMap(uint32_t Index, uint32_t Address);
 
+  /// Get offset for given index in original .debug_str_offsets section.
+  uint64_t getOffset(uint32_t Index) const { return StrOffsets[Index]; }
   /// Writes out current sections entry into .debug_str_offsets.
   void finalizeSection(DWARFUnit &Unit, DIEBuilder &DIEBldr);
 
@@ -463,7 +465,7 @@ class DebugStrOffsetsWriter {
   std::unique_ptr<DebugStrOffsetsBufferVector> StrOffsetsBuffer;
   std::unique_ptr<raw_svector_ostream> StrOffsetsStream;
   std::map<uint32_t, uint32_t> IndexToAddressMap;
-  std::vector<uint32_t> StrOffsets;
+  SmallVector<uint32_t, 5> StrOffsets;
   std::unordered_map<uint64_t, uint64_t> ProcessedBaseOffsets;
   bool StrOffsetSectionWasModified = false;
 };
@@ -484,11 +486,12 @@ class DebugStrWriter {
   /// Returns False if no strings were added to .debug_str.
   bool isInitialized() const { return !StrBuffer->empty(); }
 
+  /// Initializes Buffer and Stream.
+  void initialize();
+
 private:
   /// Mutex used for parallel processing of debug info.
   std::mutex WriterMutex;
-  /// Initializes Buffer and Stream.
-  void initialize();
   /// Creates internal data structures.
   void create();
   std::unique_ptr<DebugStrBufferVector> StrBuffer;