Merge remote-tracking branch 'origin/sycl' into fix-buffer-shadow

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

@@ -59,8 +59,9 @@ jobs:
 
       - name: Test Container
         run: |
-          for image in ${{ steps.vars.outputs.container-name-tag }} ${{  steps.vars.outputs.container-name }}; do
-            podman run --rm -it $image /usr/bin/bash -x -c 'cd $HOME && printf '\''#include <iostream>\nint main(int argc, char **argv) { std::cout << "Hello\\n"; }'\'' | clang++ -x c++ - && ./a.out | grep Hello'
+          for image in ${{ steps.vars.outputs.container-name-tag }}; do
+            # Use --pull=never to ensure we are testing the just built image.
+            podman run --pull=never --rm -it $image /usr/bin/bash -x -c 'cd $HOME && printf '\''#include <iostream>\nint main(int argc, char **argv) { std::cout << "Hello\\n"; }'\'' | clang++ -x c++ - && ./a.out | grep Hello'
           done
 
   push-ci-container:

.github/workflows/new-issues.yml

@@ -15,7 +15,7 @@ jobs:
     steps:
       - uses: llvm/actions/issue-labeler@main
         with:
-          repo-token: ${{ secrets.GITHUB_TOKEN }}
+          repo-token: ${{ secrets.ISSUE_SUBSCRIBER_TOKEN }}
           configuration-path: .github/new-issues-labeler.yml
           include-title: 1
           include-body: 0

.github/workflows/release-binaries.yml

@@ -83,7 +83,7 @@ jobs:
         USER_TOKEN: ${{ secrets.RELEASE_TASKS_USER_TOKEN }}
       shell: bash
       run: |
-        ./llvm/utils/release/./github-upload-release.py --token "$GITHUB_TOKEN" --user ${{ github.actor }} --user-token "$USER_TOKEN" check-permissions
+        ./llvm/utils/release/./github-upload-release.py --token "$GITHUB_TOKEN" --user "$GITHUB_ACTOR" --user-token "$USER_TOKEN" check-permissions
 
     - name: Collect Variables
       id: vars
@@ -102,8 +102,8 @@ jobs:
           release_version="$trimmed"
           ref="llvmorg-$release_version"
         else
-          release_version="${{ (github.event_name == 'pull_request' && format('PR{0}', github.event.pull_request.number)) || 'CI'}}-${{ github.sha }}"
-          ref=${{ github.sha }}
+          release_version="${{ (github.event_name == 'pull_request' && format('PR{0}', github.event.pull_request.number)) || 'CI'}}-$GITHUB_SHA"
+          ref="$GITHUB_SHA"
         fi
         if [ -n "${{ inputs.upload }}" ]; then
           upload="${{ inputs.upload }}"
@@ -114,20 +114,20 @@ jobs:
         echo "ref=$ref" >> $GITHUB_OUTPUT
         echo "upload=$upload" >> $GITHUB_OUTPUT
 
-        release_binary_basename="LLVM-$release_version-${{ runner.os }}-${{ runner.arch }}"
+        release_binary_basename="LLVM-$release_version-$RUNNER_OS-$RUNNER_ARCH"
         echo "release-binary-basename=$release_binary_basename" >> $GITHUB_OUTPUT
         echo "release-binary-filename=$release_binary_basename.tar.xz" >> $GITHUB_OUTPUT
 
         # Detect necessary CMake flags
-        target="${{ runner.os }}-${{ runner.arch }}"
+        target="$RUNNER_OS-$RUNNER_ARCH"
         echo "enable-pgo=false" >> $GITHUB_OUTPUT
         target_cmake_flags="-DLLVM_RELEASE_ENABLE_PGO=OFF"
         # The macOS builds try to cross compile some libraries so we need to
         # add extra CMake args to disable them.
         # See https://github.com/llvm/llvm-project/issues/99767
-        if [ "${{ runner.os }}" = "macOS" ]; then
+        if [ "$RUNNER_OS" = "macOS" ]; then
           target_cmake_flags="$target_cmake_flags -DBOOTSTRAP_COMPILER_RT_ENABLE_IOS=OFF"
-          if [ "${{ runner.arch }}" = "ARM64" ]; then
+          if [ "$RUNNER_ARCH" = "ARM64" ]; then
             arches=arm64
           else
             arches=x86_64
@@ -137,7 +137,7 @@ jobs:
 
         build_flang="true"
 
-        if [ "${{ runner.os }}" = "Windows" ]; then
+        if [ "$RUNNER_OS" = "Windows" ]; then
           # The build times out on Windows, so we need to disable LTO.
           target_cmake_flags="$target_cmake_flags -DLLVM_RELEASE_ENABLE_LTO=OFF"
         fi

.github/workflows/sycl-linux-build.yml

@@ -50,9 +50,6 @@ on:
         description: 'Artifacts retention period'
         type: string
         default: 3
-      ref:
-        type: string
-        required: false
 
     outputs:
       build_conclusion:
@@ -146,7 +143,6 @@ jobs:
       with:
         sparse-checkout: |
           devops/actions
-        ref: ${{ inputs.ref || github.sha }}
     # Cleanup will be run after all actions are completed.
     - name: Register cleanup after job is finished
       uses: ./devops/actions/cleanup

.github/workflows/sycl-linux-precommit.yml

@@ -21,6 +21,7 @@ on:
     - '.github/workflows/sycl-windows-*.yml'
     - '.github/workflows/sycl-macos-*.yml'
     - '.github/workflows/sycl-nightly.yml'
+    - '.github/workflows/sycl-rel-nightly.yml'
     - 'devops/containers/**'
     - 'devops/actions/build_container/**'
 
@@ -157,13 +158,13 @@ jobs:
           - name: E2E tests on Intel Ponte Vecchio GPU
             runner: '["Linux", "pvc"]'
             image: ghcr.io/intel/llvm/ubuntu2404_intel_drivers:latest
-            image_options: -u 1001 --device=/dev/dri --privileged --cap-add SYS_ADMIN
+            image_options: -u 1001 --device=/dev/dri -v /dev/dri/by-path:/dev/dri/by-path --privileged --cap-add SYS_ADMIN
             target_devices: level_zero:gpu;opencl:gpu
             extra_lit_opts: -j 50
           - name: E2E tests with dev igc on Intel Ponte Vecchio GPU
             runner: '["Linux", "pvc"]'
             image: ghcr.io/intel/llvm/ubuntu2404_intel_drivers:devigc
-            image_options: -u 1001 --device=/dev/dri --privileged --cap-add SYS_ADMIN
+            image_options: -u 1001 --device=/dev/dri -v /dev/dri/by-path:/dev/dri/by-path --privileged --cap-add SYS_ADMIN
             target_devices: level_zero:gpu;opencl:gpu
             use_igc_dev: true
             extra_lit_opts: -j 50

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

@@ -222,7 +222,7 @@ jobs:
         merge_ref: ${{ inputs.merge_ref }}
         cache_path: "/__w/repo_cache/"
     - name: Checkout SYCL CTS tests
-      if: inputs.tests_selector == 'cts'
+      if: inputs.tests_selector == 'cts' && inputs.cts_testing_mode != 'run-only'
       uses: ./devops/actions/cached_checkout
       with:
         path: khronos_sycl_cts
@@ -231,7 +231,7 @@ jobs:
         default_branch: 'main'
         cache_path: "/__w/repo_cache/"
     - name: SYCL CTS GIT submodules init
-      if: inputs.tests_selector == 'cts'
+      if: inputs.tests_selector == 'cts' && inputs.cts_testing_mode != 'run-only'
       run: |
         git -C khronos_sycl_cts submodule update --init
     - name: Install drivers

.github/workflows/sycl-nightly.yml

@@ -186,14 +186,12 @@ jobs:
             image: ghcr.io/intel/llvm/ubuntu2404_intel_drivers:latest
             image_options: -u 1001 --device=/dev/dri --privileged --cap-add SYS_ADMIN
             target_devices: opencl:cpu
-            tests_selector: cts
 
           - name: SYCL-CTS on L0 gen12
             runner: '["Linux", "gen12"]'
             image: ghcr.io/intel/llvm/ubuntu2404_intel_drivers:latest
             image_options: -u 1001 --device=/dev/dri --privileged --cap-add SYS_ADMIN
             target_devices: level_zero:gpu
-            tests_selector: cts
     uses: ./.github/workflows/sycl-linux-run-tests.yml
     with:
       name: ${{ matrix.name }}
@@ -202,9 +200,8 @@ jobs:
       image: ${{ matrix.image }}
       image_options: ${{ matrix.image_options }}
       target_devices: ${{ matrix.target_devices }}
-      tests_selector: ${{ matrix.tests_selector }}
+      tests_selector: cts
       ref: ${{ github.sha }}
-      merge_ref: ''
       sycl_toolchain_artifact: sycl_linux_default
       sycl_toolchain_archive: ${{ needs.ubuntu2204_build.outputs.artifact_archive_name }}
       sycl_toolchain_decompress_command: ${{ needs.ubuntu2204_build.outputs.artifact_decompress_command }}

.github/workflows/sycl-rel-nightly.yml

@@ -36,10 +36,11 @@ jobs:
     secrets: inherit
     with:
       build_cache_root: "/__w/"
-      build_artifact_suffix: v6
+      build_artifact_suffix: default
       build_configure_extra_args: '--hip --cuda'
+      build_image: ghcr.io/intel/llvm/ubuntu2204_build:latest
       merge_ref: ''
-      ref: sycl-rel-6_0_0
+      build_ref: sycl-rel-6_0_0
 
       # We upload the build for people to download/use, override its name and
       # prefer widespread gzip compression.
@@ -118,10 +119,12 @@ jobs:
     if: ${{ github.repository == 'intel/llvm' && needs.check_for_new_commits.outputs.is_new_commit != 'false' }}
     uses: ./.github/workflows/sycl-windows-build.yml
     with:
+      merge_ref: ''
+      build_ref: sycl-rel-6_0_0
+
       # We upload both Linux/Windows build via Github's "Releases"
       # functionality, make sure Linux/Windows names follow the same pattern.
       artifact_archive_name: sycl_windows.tar.gz
-      build_ref: sycl-rel-6_0_0
 
   e2e-win:
     needs: build-win
@@ -154,7 +157,7 @@ jobs:
     with:
       name: CUDA E2E
       runner: '["aws_cuda-${{ github.run_id }}-${{ github.run_attempt }}"]'
-      image: ghcr.io/intel/llvm/ubuntu2204_build:latest-0300ac924620a51f76c4929794637b82790f12ab
+      image: ghcr.io/intel/llvm/ubuntu2204_build:latest
       image_options: -u 1001 --gpus all --cap-add SYS_ADMIN --env NVIDIA_DISABLE_REQUIRE=1
       target_devices: cuda:gpu
       ref: sycl-rel-6_0_0

.github/workflows/sycl-windows-build.yml

@@ -18,6 +18,12 @@ on:
         description: 'Filter matches for the changed files in the PR'
         default: '[llvm, clang, sycl, llvm_spirv, xptifw, libclc, libdevice]'
         required: false
+      merge_ref:
+        description: |
+          Commit-ish to merge post-checkout if non-empty. Must be reachable from
+          the default_branch input paramter.
+        type: string
+        default: 'FETCH_HEAD'
       artifact_archive_name:
         type: string
         default: llvm_sycl.tar.gz
@@ -100,6 +106,7 @@ jobs:
       with:
         path: src
         ref: ${{ inputs.build_ref || github.sha }}
+        merge_ref: ${{ inputs.merge_ref }}
         cache_path: "D:\\\\github\\\\_work\\\\repo_cache\\\\"
     - name: Configure
       shell: cmd

.github/workflows/sycl-windows-precommit.yml

@@ -20,6 +20,7 @@ on:
     - '.github/workflows/sycl-precommit-aws.yml'
     - '.github/workflows/sycl-macos-*.yml'
     - '.github/workflows/sycl-nightly.yml'
+    - '.github/workflows/sycl-rel-nightly.yml'
     - 'devops/containers/**'
     - 'devops/actions/build_container/**'
 

SECURITY.md

@@ -1,5 +1,17 @@
-# Reporting LLVM Security Issues
+# Security Policy
 
-To report security issues in LLVM, please follow the steps outlined on the
-[LLVM Security Group](https://llvm.org/docs/Security.html#how-to-report-a-security-issue)
-page.
+Intel is committed to rapidly addressing security vulnerabilities affecting our
+customers and providing clear guidance on the solution, impact, severity and
+mitigation.
+
+## Reporting a Vulnerability
+
+`intel/llvm` project is based upon the upstream
+[`llvm/llvm-project`](https://github.com/llvm/llvm-project) and leverages lots
+of functionality provided by it. If an issue you found concerns functionality
+which is available in upstream, then it is more efficient to directly report it
+there, see corresponding `SECURITY.md` for their policy.
+
+If you are unsure, or if you know that the issue is specific to additions
+developed in `intel/llvm`, then please report any security vulnerabilities
+[utilizing the guidelines here](https://www.intel.com/content/www/us/en/security-center/vulnerability-handling-guidelines.html).

bolt/docs/CommandLineArgumentReference.md

@@ -931,15 +931,6 @@
 
   Remove redundant Address-Size override prefix
 
-### BOLT options in relocation mode:
-
-- `--align-macro-fusion=<value>`
-
-  Fix instruction alignment for macro-fusion (x86 relocation mode)
-  - `none`: do not insert alignment no-ops for macro-fusion
-  - `hot`: only insert alignment no-ops on hot execution paths (default)
-  - `all`: always align instructions to allow macro-fusion
-
 ### BOLT instrumentation options:
 
 `llvm-bolt <executable> -instrument [-o outputfile] <instrumented-executable>`

bolt/include/bolt/Core/BinaryData.h

@@ -169,6 +169,11 @@ class BinaryData {
     return Parent && (Parent == BD || Parent->isAncestorOf(BD));
   }
 
+  void updateSize(uint64_t N) {
+    if (N > Size)
+      Size = N;
+  }
+
   void setIsMoveable(bool Flag) { IsMoveable = Flag; }
   void setSection(BinarySection &NewSection);
   void setOutputSection(BinarySection &NewSection) {

bolt/lib/Core/BinaryContext.cpp

@@ -1076,6 +1076,7 @@ MCSymbol *BinaryContext::registerNameAtAddress(StringRef Name, uint64_t Address,
     BD = GAI->second;
     if (!BD->hasName(Name)) {
       GlobalSymbols[Name] = BD;
+      BD->updateSize(Size);
       BD->Symbols.push_back(Symbol);
     }
   }

bolt/lib/Rewrite/LinuxKernelRewriter.cpp

@@ -21,6 +21,8 @@
 #include "llvm/Support/CommandLine.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/Errc.h"
+#include "llvm/Support/ErrorOr.h"
+#include <regex>
 
 #define DEBUG_TYPE "bolt-linux"
 
@@ -89,6 +91,34 @@ static cl::opt<bool>
 
 } // namespace opts
 
+/// Linux kernel version
+struct LKVersion {
+  LKVersion() {}
+  LKVersion(unsigned Major, unsigned Minor, unsigned Rev)
+      : Major(Major), Minor(Minor), Rev(Rev) {}
+
+  bool operator<(const LKVersion &Other) const {
+    return std::make_tuple(Major, Minor, Rev) <
+           std::make_tuple(Other.Major, Other.Minor, Other.Rev);
+  }
+
+  bool operator>(const LKVersion &Other) const { return Other < *this; }
+
+  bool operator<=(const LKVersion &Other) const { return !(*this > Other); }
+
+  bool operator>=(const LKVersion &Other) const { return !(*this < Other); }
+
+  bool operator==(const LKVersion &Other) const {
+    return Major == Other.Major && Minor == Other.Minor && Rev == Other.Rev;
+  }
+
+  bool operator!=(const LKVersion &Other) const { return !(*this == Other); }
+
+  unsigned Major{0};
+  unsigned Minor{0};
+  unsigned Rev{0};
+};
+
 /// Linux Kernel supports stack unwinding using ORC (oops rewind capability).
 /// ORC state at every IP can be described by the following data structure.
 struct ORCState {
@@ -148,6 +178,8 @@ class AddressExtractor : public DataExtractor {
 };
 
 class LinuxKernelRewriter final : public MetadataRewriter {
+  LKVersion LinuxKernelVersion;
+
   /// Information required for updating metadata referencing an instruction.
   struct InstructionFixup {
     BinarySection &Section; // Section referencing the instruction.
@@ -249,6 +281,8 @@ class LinuxKernelRewriter final : public MetadataRewriter {
   ErrorOr<BinarySection &> PCIFixupSection = std::errc::bad_address;
   static constexpr size_t PCI_FIXUP_ENTRY_SIZE = 16;
 
+  Error detectLinuxKernelVersion();
+
   /// Process linux kernel special sections and their relocations.
   void processLKSections();
 
@@ -314,6 +348,9 @@ class LinuxKernelRewriter final : public MetadataRewriter {
       : MetadataRewriter("linux-kernel-rewriter", BC) {}
 
   Error preCFGInitializer() override {
+    if (Error E = detectLinuxKernelVersion())
+      return E;
+
     processLKSections();
 
     if (Error E = processSMPLocks())
@@ -394,6 +431,28 @@ class LinuxKernelRewriter final : public MetadataRewriter {
   }
 };
 
+Error LinuxKernelRewriter::detectLinuxKernelVersion() {
+  if (BinaryData *BD = BC.getBinaryDataByName("linux_banner")) {
+    const BinarySection &Section = BD->getSection();
+    const std::string S =
+        Section.getContents().substr(BD->getOffset(), BD->getSize()).str();
+
+    const std::regex Re(R"---(Linux version ((\d+)\.(\d+)(\.(\d+))?))---");
+    std::smatch Match;
+    if (std::regex_search(S, Match, Re)) {
+      const unsigned Major = std::stoi(Match[2].str());
+      const unsigned Minor = std::stoi(Match[3].str());
+      const unsigned Rev = Match[5].matched ? std::stoi(Match[5].str()) : 0;
+      LinuxKernelVersion = LKVersion(Major, Minor, Rev);
+      BC.outs() << "BOLT-INFO: Linux kernel version is " << Match[1].str()
+                << "\n";
+      return Error::success();
+    }
+  }
+  return createStringError(errc::executable_format_error,
+                           "Linux kernel version is unknown");
+}
+
 void LinuxKernelRewriter::processLKSections() {
   processLKKSymtab();
   processLKKSymtab(true);