Merge branch 'sycl' into sycl-devops-pr/udit/pvc_e2e

Author: uditagarwal97

.github/workflows/sycl-containers-igc-dev.yaml

@@ -33,7 +33,7 @@ jobs:
             imagefile: ubuntu2404_intel_drivers
             tag: devigc
             build_args: |
-                  "use_latest=false"
+                  "use_unstable_driver=false"
                   "use_igc_dev=true"
     steps:
       - name: Checkout

.github/workflows/sycl-containers.yaml

@@ -54,22 +54,22 @@ jobs:
           - name: Intel Drivers Ubuntu 22.04 Docker image
             file: ubuntu2204_intel_drivers
             tag: latest
-            build_args: "use_latest=false"
+            build_args: "use_unstable_driver=false"
           - name: Intel Drivers Ubuntu 24.04 Docker image
             file: ubuntu2404_intel_drivers
             tag: latest
-            build_args: "use_latest=false"
+            build_args: "use_unstable_driver=false"
           - name: Intel Drivers (unstable) Ubuntu 24.04 Docker image
             file: ubuntu2404_intel_drivers
             tag: unstable
-            build_args: "use_latest=true"
+            build_args: "use_unstable_driver=true"
           - name: Build + Intel Drivers Ubuntu 22.04 Docker image
             file: ubuntu2204_intel_drivers
             tag: alldeps
             build_args: |
               base_image=ghcr.io/intel/llvm/ubuntu2204_build
               base_tag=latest
-              use_latest=false
+              use_unstable_driver=false
     steps:
       - name: Checkout
         uses: actions/checkout@v4

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

@@ -72,6 +72,53 @@ jobs:
           else
             echo 'arc_tests="Matrix/"' >> "$GITHUB_OUTPUT"
           fi
+
+  build_e2e_tests:
+    needs: [build]
+    if: ${{ always() && !cancelled() && needs.build.outputs.build_conclusion == 'success' }}
+    uses: ./.github/workflows/sycl-linux-run-tests.yml
+    with:
+      name: Build e2e tests
+      runner: '["Linux", "build"]'
+      image: ghcr.io/intel/llvm/ubuntu2404_intel_drivers:latest
+      image_options: -u 1001
+      ref: ${{ github.sha }}
+      merge_ref: ''
+      sycl_toolchain_artifact: sycl_linux_default
+      sycl_toolchain_archive: ${{ needs.build.outputs.artifact_archive_name }}
+      sycl_toolchain_decompress_command: ${{ needs.build.outputs.artifact_decompress_command }}
+      e2e_testing_mode: 'build-only'
+  run_prebuilt_e2e_tests:
+    needs: [build, build_e2e_tests]
+    if: ${{ always() && !cancelled() && needs.build.outputs.build_conclusion == 'success' }}
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - name: Intel
+            runner: '["Linux", "gen12"]'
+            image: ghcr.io/intel/llvm/ubuntu2404_intel_drivers:latest
+            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;opencl:cpu
+            reset_intel_gpu: true
+            extra_lit_opts: --param gpu-intel-gen12=True
+    uses: ./.github/workflows/sycl-linux-run-tests.yml
+    with:
+      name: ${{ matrix.name }}
+      runner: ${{ matrix.runner }}
+      image: ${{ matrix.image }}
+      image_options: ${{ matrix.image_options }}
+      target_devices: ${{ matrix.target_devices }}
+      extra_lit_opts: --param fallback-to-build-if-requires-build-and-run=True ${{ matrix.extra_lit_opts }}
+      reset_intel_gpu: ${{ matrix.reset_intel_gpu }}
+      ref: ${{ github.sha }}
+      merge_ref: ''
+      sycl_toolchain_artifact: sycl_linux_default
+      sycl_toolchain_archive: ${{ needs.build.outputs.artifact_archive_name }}
+      sycl_toolchain_decompress_command: ${{ needs.build.outputs.artifact_decompress_command }}
+      e2e_binaries_artifact: sycl_e2e_bin_default
+      e2e_testing_mode: 'run-only'
+
   test:
     needs: [build, detect_changes, determine_arc_tests]
     if: ${{ always() && !cancelled() && needs.build.outputs.build_conclusion == 'success' }}
@@ -90,13 +137,6 @@ jobs:
             image_options: -u 1001 --device=/dev/dri --device=/dev/kfd
             target_devices: ext_oneapi_hip:gpu
             reset_intel_gpu: false
-          - name: Intel
-            runner: '["Linux", "gen12"]'
-            image: ghcr.io/intel/llvm/ubuntu2404_intel_drivers:latest
-            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;opencl:cpu
-            reset_intel_gpu: true
-            extra_lit_opts: --param gpu-intel-gen12=True
           - name: E2E tests on Intel Arc A-Series Graphics
             runner: '["Linux", "arc"]'
             image: ghcr.io/intel/llvm/ubuntu2404_intel_drivers:latest

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

@@ -19,7 +19,7 @@ on:
 
       target_devices:
         type: string
-        required: True
+        required: False
       extra_cmake_args:
         type: string
         required: False
@@ -59,6 +59,30 @@ on:
         default: ''
         required: False
 
+      e2e_binaries_artifact:
+        description: |
+          By setting this the E2E binaries folder will not be created, rather it
+          will be downloaded and extracted from the specified artifact. When
+          running tests in `run-only` mode this must be provided.
+        type: string
+        default: ''
+        required: False
+      e2e_testing_mode:
+        description: |
+          Testing mode to run E2E tests in, can be either `full`, `build-only`
+          or `run-only`. In `build-only` mode an artifact of the E2E binaries
+          will be uploaded.
+        type: string
+        default: 'full'
+      artifact_suffix:
+        description: 'Suffix for E2E binaries artifact that is output when in `build-only`.'
+        type: string
+        default: 'default'
+      retention-days:
+        description: 'E2E binaries artifact retention period.'
+        type: string
+        default: 1
+
       reset_intel_gpu:
         type: string
         required: False
@@ -88,6 +112,7 @@ on:
           - '["Linux", "arc"]'
           - '["Linux", "pvc"]'
           - '["cts-cpu"]'
+          - '["Linux", "build"]'
       image:
         description: |
           Use option ending with ":build" for AMDGPU, ":latest" for the rest.
@@ -143,6 +168,11 @@ on:
         options:
           - false
           - true
+      e2e_testing_mode:
+        type: choice
+        options:
+          - "full"
+          - "build-only"
 
 permissions:
   contents: read
@@ -271,8 +301,19 @@ jobs:
             cat /usr/local/lib/igc/IGCTAG.txt
           fi
 
+    - name: Download E2E Binaries
+      if: inputs.e2e_binaries_artifact != ''
+      uses: actions/download-artifact@v4
+      with:
+        name: ${{ inputs.e2e_binaries_artifact }}
+    - name: Extract E2E Binaries
+      if: inputs.e2e_binaries_artifact != ''
+      run: |
+        mkdir build-e2e
+        tar -I 'zstd' -xf e2e_binaries.tar.zst -C build-e2e
+
     - name: Deduce E2E CMake options
-      if: inputs.tests_selector == 'e2e'
+      if: inputs.tests_selector == 'e2e' && inputs.e2e_binaries_artifact == ''
       id: cmake_opts
       shell: bash
       env:
@@ -282,14 +323,14 @@ jobs:
           echo "opts=$CMAKE_EXTRA_ARGS" >> $GITHUB_OUTPUT
         fi
     - name: Configure E2E tests
-      if: inputs.tests_selector == 'e2e'
+      if: inputs.tests_selector == 'e2e' && inputs.e2e_binaries_artifact == ''
       run: |
-        cmake -GNinja -B./build-e2e -S./llvm/sycl/test-e2e -DSYCL_TEST_E2E_TARGETS="${{ inputs.target_devices }}" -DCMAKE_CXX_COMPILER="$(which clang++)" -DLLVM_LIT="$PWD/llvm/llvm/utils/lit/lit.py" ${{ steps.cmake_opts.outputs.opts }}
+        cmake -GNinja -B./build-e2e -S./llvm/sycl/test-e2e -DCMAKE_CXX_COMPILER="$(which clang++)" -DLLVM_LIT="$PWD/llvm/llvm/utils/lit/lit.py" ${{ steps.cmake_opts.outputs.opts }}
     - name: SYCL End-to-end tests
       shell: bash {0}
       if: inputs.tests_selector == 'e2e'
       env:
-        LIT_OPTS: -v --no-progress-bar --show-unsupported --show-pass --show-xfail --max-time 3600 --time-tests ${{ inputs.extra_lit_opts }}
+        LIT_OPTS: -v --no-progress-bar --show-unsupported --show-pass --show-xfail --max-time 3600 --time-tests --param test-mode=${{ inputs.e2e_testing_mode }} --param sycl_devices=${{ inputs.target_devices }} ${{ inputs.extra_lit_opts }}
       run: |
         ninja -C build-e2e check-sycl-e2e > e2e.log 2>&1
         exit_code=$?
@@ -376,3 +417,13 @@ jobs:
         grep 'exit code: [^0]' -r logs >> $GITHUB_STEP_SUMMARY
 
         exit $ret
+    - name: Pack E2E binaries
+      if: ${{ always() && !cancelled() && inputs.e2e_testing_mode == 'build-only'}}
+      run: tar -I 'zstd -9' -cf e2e_binaries.tar.zst -C ./build-e2e .
+    - name: Upload E2E binaries
+      if: ${{ always() && !cancelled() && inputs.e2e_testing_mode == 'build-only'}}
+      uses: actions/upload-artifact@v4
+      with:
+        name: sycl_e2e_bin_${{ inputs.artifact_suffix }}
+        path: e2e_binaries.tar.zst
+        retention-days: ${{ inputs.retention-days }}

devops/containers/ubuntu2204_intel_drivers.Dockerfile

@@ -5,7 +5,7 @@ FROM $base_image:$base_tag
 
 ENV DEBIAN_FRONTEND=noninteractive
 
-ARG use_latest=true
+ARG use_unstable_driver=true
 
 USER root
 
@@ -18,7 +18,7 @@ COPY dependencies.json /
 RUN mkdir /runtimes
 ENV INSTALL_LOCATION=/runtimes
 RUN --mount=type=secret,id=github_token \
-    if [ "$use_latest" = "true" ]; then \
+    if [ "$use_unstable_driver" = "true" ]; then \
       install_driver_opt=" --use-latest"; \
     else \
       install_driver_opt=" dependencies.json"; \

devops/containers/ubuntu2404_intel_drivers.Dockerfile

@@ -5,7 +5,7 @@ FROM $base_image:$base_tag
 
 ENV DEBIAN_FRONTEND=noninteractive
 
-ARG use_latest=true
+ARG use_unstable_driver=true
 
 USER root
 
@@ -18,7 +18,7 @@ COPY dependencies.json /
 RUN mkdir /runtimes
 ENV INSTALL_LOCATION=/runtimes
 RUN --mount=type=secret,id=github_token \
-    if [ "$use_latest" = "true" ]; then \
+    if [ "$use_unstable_driver" = "true" ]; then \
       install_driver_opt=" --use-latest"; \
     else \
       install_driver_opt=" dependencies.json"; \

llvm/lib/SYCLLowerIR/SanitizerKernelMetadata.cpp

@@ -38,6 +38,29 @@ PreservedAnalyses SanitizerKernelMetadataPass::run(Module &M,
   auto &DL = M.getDataLayout();
   auto &Ctx = M.getContext();
 
+  // Fix device global type, by wrapping a structure type
+  {
+    assert(KernelMetadata->getValueType()->isArrayTy());
+
+    auto *KernelMetadataOld = KernelMetadata;
+
+    StructType *StructTypeWithArray = StructType::create(Ctx);
+    StructTypeWithArray->setBody(KernelMetadataOld->getValueType());
+
+    KernelMetadata = new GlobalVariable(
+        M, StructTypeWithArray, false, GlobalValue::ExternalLinkage,
+        ConstantStruct::get(StructTypeWithArray,
+                            KernelMetadataOld->getInitializer()),
+        "", nullptr, GlobalValue::NotThreadLocal, 1); // Global AddressSpace
+    KernelMetadata->takeName(KernelMetadataOld);
+    KernelMetadata->setUnnamedAddr(GlobalValue::UnnamedAddr::Local);
+    KernelMetadata->setDSOLocal(true);
+    KernelMetadata->copyAttributesFrom(KernelMetadataOld);
+    KernelMetadata->copyMetadata(KernelMetadataOld, 0);
+
+    KernelMetadataOld->eraseFromParent();
+  }
+
   // Fix attributes
   KernelMetadata->addAttribute(
       "sycl-device-global-size",

llvm/test/tools/sycl-post-link/device-sanitizer/asan.ll

@@ -18,7 +18,7 @@ $_ZTSZZ4mainENKUlRN4sycl3_V17handlerEE_clES2_E8MyKernel = comdat any
 
 @__asan_kernel = internal addrspace(1) constant [55 x i8] c"_ZTSZZ4mainENKUlRN4sycl3_V17handlerEE_clES2_E8MyKernel\00"
 @__AsanKernelMetadata = appending dso_local local_unnamed_addr addrspace(1) global [1 x { i64, i64 }] [{ i64, i64 } { i64 ptrtoint (ptr addrspace(1) @__asan_kernel to i64), i64 54 }] #2
-; CHECK-IR: @__AsanKernelMetadata {{.*}} !spirv.Decorations
+; CHECK-IR: @__AsanKernelMetadata = dso_local local_unnamed_addr addrspace(1) global %0 { {{.*}} }, !spirv.Decorations
 @__spirv_BuiltInGlobalInvocationId = external dso_local local_unnamed_addr addrspace(1) constant <3 x i64>, align 32
 @__asan_func = internal addrspace(2) constant [106 x i8] c"typeinfo name for main::'lambda'(sycl::_V1::handler&)::operator()(sycl::_V1::handler&) const::MyKernelR_4\00"
 

llvm/test/tools/sycl-post-link/device-sanitizer/msan.ll

@@ -18,7 +18,7 @@ $_ZTSZZ4mainENKUlRN4sycl3_V17handlerEE_clES2_E8MyKernel = comdat any
 
 @__msan_kernel = internal addrspace(1) constant [55 x i8] c"_ZTSZZ4mainENKUlRN4sycl3_V17handlerEE_clES2_E8MyKernel\00"
 @__MsanKernelMetadata = appending dso_local local_unnamed_addr addrspace(1) global [1 x { i64, i64 }] [{ i64, i64 } { i64 ptrtoint (ptr addrspace(1) @__msan_kernel to i64), i64 54 }] #0
-; CHECK-IR: @__MsanKernelMetadata {{.*}} !spirv.Decorations
+; CHECK-IR: @__MsanKernelMetadata = dso_local local_unnamed_addr addrspace(1) global %0 { {{.*}} }, !spirv.Decorations
 @__spirv_BuiltInGlobalInvocationId = external dso_local local_unnamed_addr addrspace(1) constant <3 x i64>, align 32
 @__asan_func = internal addrspace(2) constant [106 x i8] c"typeinfo name for main::'lambda'(sycl::_V1::handler&)::operator()(sycl::_V1::handler&) const::MyKernelR_4\00"
 

sycl/doc/design/KernelProgramCache.md

@@ -415,15 +415,16 @@ When adding a new program to cache, we check if the size of the program cache ex
 
 #### Persistent cache eviction
 
-Persistent cache eviction is going to be applied based on file last access
-(read/write) date (access time). On SYCL application shutdown phase cache
-eviction process is initiated which walks through cache directories as follows:
-
-- if the file is locked, go to the next file;
-- otherwise check file access time:
-  - if file access time is above threshold, delete the file and remove parent
-    directory while they are unlocked and empty;
-  - otherwise do nothing.
+Persistent cache eviction can be enabled using the SYCL_CACHE_MAX_SIZE environment variable and is based on the LRU strategy.
+
+- A new file, called `cache_size.txt`, is created at the root of the persistent cache directory. This file contains the total size of the cache in bytes. When a new item is added to the cache, the size of the item is added to the total size in the `cache_size.txt` file. When the total size exceeds the threshold, the eviction process is initiated.
+
+- Whenever a cache entry is added or accessed, the corresponding cache item directory is updated with the current time. This is done by creating a new file, called `<entry name>_access_time.txt`, in the cache item directory. This file contains the current time in nanoseconds since the epoch. When the eviction process is initiated, we use this file to determine the last access time of the cache item.
+
+- When a new item is added to the cache, we check if the total size exceeds the threshold. If so, we iterate through the cache item directories and delete the least recently accessed items until the total size is below half the cache size.
+
+Note that once the eviction is triggered, the cache size is reduced to half the cache size to avoid frequent eviction.
+
 
 ## Cache limitations