Merge branch 'sycl' into amd-markup-change

Author: ayylol

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

@@ -51,30 +51,8 @@ jobs:
       changes: ${{ needs.detect_changes.outputs.filters }}
       e2e_binaries_artifact: sycl_e2e_bin_default
 
-  determine_arc_tests:
-    name: Decide which Arc tests to run
-    needs: [build, detect_changes]
-    if: ${{ always() && !cancelled() && needs.build.outputs.build_conclusion == 'success' }}
-    runs-on: [Linux, aux-tasks]
-    timeout-minutes: 3
-    outputs:
-      arc_tests: ${{ steps.arc_tests.outputs.arc_tests }}
-    steps:
-      - name: Determine Arc tests
-        id: arc_tests
-        run: |
-          if [ "${{ contains(needs.detect_changes.outputs.filters, 'devigccfg') }}" == "true" ]; then
-            echo 'arc_tests="(ESIMD|InvokeSimd|Matrix)/"' >> "$GITHUB_OUTPUT"
-          elif [ "${{ contains(needs.detect_changes.outputs.filters, 'drivers') }}" == "true" ]; then
-             echo 'arc_tests=""' >> "$GITHUB_OUTPUT"
-          elif [ "${{ contains(needs.detect_changes.outputs.filters, 'esimd') }}" == "true" ]; then
-            echo 'arc_tests="(ESIMD|InvokeSimd|Matrix)/"' >> "$GITHUB_OUTPUT"
-          else
-            echo 'arc_tests="Matrix/"' >> "$GITHUB_OUTPUT"
-          fi
-
   run_prebuilt_e2e_tests:
-    needs: [build, detect_changes, determine_arc_tests]
+    needs: [build, detect_changes]
     if: ${{ always() && !cancelled() && needs.build.outputs.build_conclusion == 'success' }}
     strategy:
       fail-fast: false
@@ -96,7 +74,19 @@ jobs:
             target_devices: level_zero:gpu;opencl:gpu
             reset_intel_gpu: true
             extra_lit_opts: --param matrix-xmx8=True
-            env: '{"LIT_FILTER":${{ needs.determine_arc_tests.outputs.arc_tests }} }'
+          - name: E2E tests with dev igc on Intel Arc A-Series Graphics
+            runner: '["Linux", "arc"]'
+            image: ghcr.io/intel/llvm/ubuntu2404_intel_drivers:devigc
+            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
+            reset_intel_gpu: true
+            extra_lit_opts: --param matrix-xmx8=True
+            use_igc_dev: true
+          - name: E2E tests on Intel Ponte Vecchio GPU
+            runner: '["Linux", "pvc"]'
+            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: Dev IGC on Intel Ponte Vecchio GPU
             runner: '["Linux", "pvc"]'
             image: ghcr.io/intel/llvm/ubuntu2404_intel_drivers:devigc
@@ -141,7 +131,7 @@ jobs:
       skip_run: ${{matrix.use_igc_dev && contains(github.event.pull_request.labels.*.name, 'ci-no-devigc') || 'false'}}
 
   test:
-    needs: [build, detect_changes, determine_arc_tests]
+    needs: [build, detect_changes]
     if: ${{ always() && !cancelled() && needs.build.outputs.build_conclusion == 'success' }}
     strategy:
       fail-fast: false
@@ -151,46 +141,13 @@ jobs:
             runner: '["Linux", "amdgpu"]'
             image_options: -u 1001 --device=/dev/dri --device=/dev/kfd
             target_devices: hip:gpu
-            reset_intel_gpu: false
-          - name: E2E tests with dev igc on Intel Arc A-Series Graphics
-            runner: '["Linux", "arc"]'
-            image: ghcr.io/intel/llvm/ubuntu2404_intel_drivers:devigc
-            image_options: -u 1001 --device=/dev/dri -v /dev/dri/by-path:/dev/dri/by-path --privileged --cap-add SYS_ADMIN
-            target_devices: level_zero:gpu;opencl:gpu
-            reset_intel_gpu: true
-            extra_lit_opts: --param matrix-xmx8=True
-            env: '{"LIT_FILTER":${{ needs.determine_arc_tests.outputs.arc_tests }} }'
-            use_igc_dev: true
-          - name: E2E tests on Intel Ponte Vecchio GPU
-            runner: '["Linux", "pvc"]'
-            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
 
     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 }}
-      reset_intel_gpu: ${{ matrix.reset_intel_gpu }}
-      extra_lit_opts: ${{ matrix.extra_lit_opts }}
-      env: ${{ matrix.env || '{}' }}
-
-      # Do not install drivers on AMD and CUDA runners.
-      install_igc_driver: >-
-        ${{ !contains(matrix.target_devices, 'cuda') &&
-        !contains(matrix.target_devices, 'hip') &&
-        contains(needs.detect_changes.outputs.filters, 'drivers') }}
-      install_dev_igc_driver: >-
-        ${{ !contains(matrix.target_devices, 'cuda') &&
-        !contains(matrix.target_devices, 'hip') &&
-        matrix.use_igc_dev &&
-        (contains(needs.detect_changes.outputs.filters, 'devigccfg') || contains(needs.detect_changes.outputs.filters, 'drivers')) ||
-        'false' }}
-      # Run only if the PR does not have the 'ci-no-devigc' label.
-      skip_run: ${{matrix.use_igc_dev && contains(github.event.pull_request.labels.*.name, 'ci-no-devigc') || 'false'}}
 
       ref: ${{ github.sha }}
 

.github/workflows/sycl-nightly.yml

@@ -60,46 +60,61 @@ jobs:
             runner: '["Linux", "amdgpu"]'
             image_options: -u 1001 --device=/dev/dri --device=/dev/kfd
             target_devices: hip:gpu
-            tests_selector: e2e
 
-          - name: Intel L0 GPU
+          - name: NVIDIA/CUDA
+            runner: '["Linux", "cuda"]'
+            image_options: -u 1001 --gpus all --cap-add SYS_ADMIN
+            target_devices: cuda:gpu
+
+          - name: Intel L0 Gen12 GPU
             runner: '["Linux", "gen12"]'
             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
             reset_intel_gpu: true
-            tests_selector: e2e
 
-          - name: Intel OCL GPU
+          - name: Intel L0 Ponte Vecchio GPU
+            runner: '["Linux", "pvc"]'
+            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
+
+          - name: Intel L0 Battlemage GPU
+            runner: '["Linux", "bmg"]'
+            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
+
+          - name: Intel L0 Arc A-Series GPU
+            runner: '["Linux", "arc"]'
+            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
+            reset_intel_gpu: true
+
+          - name: Intel OCL Gen12 GPU
             runner: '["Linux", "gen12"]'
             image_options: -u 1001 --device=/dev/dri -v /dev/dri/by-path:/dev/dri/by-path --privileged --cap-add SYS_ADMIN
             target_devices: opencl:gpu
             reset_intel_gpu: true
-            tests_selector: e2e
 
           - name: OCL CPU (AMD)
             runner: '["Linux", "amdcpu"]'
             image_options: -u 1001
             target_devices: opencl:cpu
-            tests_selector: e2e
 
           - name: OCL CPU (Intel/GEN12)
             runner: '["Linux", "gen12"]'
             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_options: -u 1001
             target_devices: opencl:cpu
-            tests_selector: e2e
     uses: ./.github/workflows/sycl-linux-run-tests.yml
     with:
       name: ${{ matrix.name }}
       runner: ${{ matrix.runner }}
       image_options: ${{ matrix.image_options }}
       target_devices: ${{ matrix.target_devices }}
-      tests_selector: ${{ matrix.tests_selector }}
+      tests_selector: e2e
       extra_lit_opts: "--param 'cxx_flags=-D_GLIBCXX_USE_CXX11_ABI=0' ${{ matrix.extra_lit_opts }}"
       reset_intel_gpu: ${{ matrix.reset_intel_gpu }}
       ref: ${{ github.sha }}

clang/include/clang/Basic/DiagnosticGroups.td

@@ -115,11 +115,10 @@ def ImplicitIntFloatConversion : DiagGroup<"implicit-int-float-conversion",
  [ImplicitConstIntFloatConversion]>;
 def ObjCSignedCharBoolImplicitFloatConversion :
   DiagGroup<"objc-signed-char-bool-implicit-float-conversion">;
-def ImplicitFloatSizeConversion :
-  DiagGroup<"implicit-float-size-conversion">;
+def SyclImplicitFloatSizeConversion : DiagGroup<"sycl-implicit-float-size-conversion">;
 def ImplicitFloatConversion : DiagGroup<"implicit-float-conversion",
   [ImplicitIntFloatConversion,
-   ObjCSignedCharBoolImplicitFloatConversion, ImplicitFloatSizeConversion]>;
+   ObjCSignedCharBoolImplicitFloatConversion, SyclImplicitFloatSizeConversion]>;
 def ImplicitFixedPointConversion : DiagGroup<"implicit-fixed-point-conversion">;
 
 def FloatOverflowConversion : DiagGroup<"float-overflow-conversion">;

clang/include/clang/Basic/DiagnosticSemaKinds.td

@@ -131,9 +131,9 @@ def warn_float_compare_literal : Warning<
 def warn_double_const_requires_fp64 : Warning<
   "double precision constant requires %select{cl_khr_fp64|cl_khr_fp64 and __opencl_c_fp64}0, "
   "casting to single precision">;
-def warn_imp_float_size_conversion : Warning<
+def warn_sycl_imp_float_size_conversion : Warning<
   "implicit conversion between floating point types of different sizes">,
-  InGroup<ImplicitFloatSizeConversion>, DefaultIgnore;
+  InGroup<SyclImplicitFloatSizeConversion>, DefaultIgnore;
 def err_half_const_requires_fp16 : Error<
   "half precision constant requires cl_khr_fp16">;
 

clang/include/clang/Driver/Options.td

@@ -1169,17 +1169,17 @@ defm cx_fortran_rules: BoolOptionWithoutMarshalling<"f", "cx-fortran-rules",
 
  defm offload_fp32_prec_div: BoolOption<"f", "offload-fp32-prec-div",
    LangOpts<"OffloadFP32PrecDiv">, DefaultTrue,
-   PosFlag<SetTrue, [], [ClangOption, CC1Option], "fdiv operations in offload device "
+   PosFlag<SetTrue, [], [ClangOption, CLOption, CC1Option], "fdiv operations in offload device "
    "code are required to return correctly rounded results.">,
-   NegFlag<SetFalse, [], [ClangOption, CC1Option], "fdiv operations in offload device "
+   NegFlag<SetFalse, [], [ClangOption, CLOption, CC1Option], "fdiv operations in offload device "
    "code are not required to return correctly rounded results.">>,
    Group<f_Group>;
 
  defm offload_fp32_prec_sqrt: BoolOption<"f", "offload-fp32-prec-sqrt",
    LangOpts<"OffloadFP32PrecSqrt">, DefaultTrue,
-   PosFlag<SetTrue, [], [ClangOption, CC1Option], "sqrt operations in offload device "
+   PosFlag<SetTrue, [], [ClangOption, CLOption, CC1Option], "sqrt operations in offload device "
    "code are required to return correctly rounded results.">,
-   NegFlag<SetFalse, [], [ClangOption, CC1Option], "sqrt operations in offload device "
+   NegFlag<SetFalse, [], [ClangOption, CLOption, CC1Option], "sqrt operations in offload device "
    "code are not required to return correctly rounded results.">>,
    Group<f_Group>;
 

clang/lib/CodeGen/BackendUtil.cpp

@@ -1044,6 +1044,8 @@ void EmitAssemblyHelper::RunOptimizationPipeline(
             /*FP64ConvEmu=*/CodeGenOpts.FP64ConvEmu,
             /*ExcludeAspects=*/{"fp64"}));
         MPM.addPass(SYCLPropagateJointMatrixUsagePass());
+        // Lowers static/dynamic local memory builtin calls.
+        MPM.addPass(SYCLLowerWGLocalMemoryPass());
       });
     else if (LangOpts.SYCLIsHost && !LangOpts.SYCLESIMDBuildHostCode)
       PB.registerPipelineStartEPCallback(
@@ -1206,10 +1208,6 @@ void EmitAssemblyHelper::RunOptimizationPipeline(
         MPM.addPass(SPIRITTAnnotationsPass());
       }
 
-      // Allocate static local memory in SYCL kernel scope for each allocation
-      // call.
-      MPM.addPass(SYCLLowerWGLocalMemoryPass());
-
       // Process properties and annotations
       MPM.addPass(CompileTimePropertiesPass());
 

clang/lib/Driver/ToolChains/SYCL.cpp

@@ -896,6 +896,8 @@ static llvm::SmallVector<StringRef, 16> SYCLDeviceLibList{
     "asan-cpu",
     "asan-dg2",
     "msan",
+    "msan-pvc",
+    "msan-cpu",
 #endif
     "imf",
     "imf-fp64",

clang/lib/Sema/SemaChecking.cpp

@@ -11680,27 +11680,33 @@ void Sema::CheckImplicitConversion(Expr *E, QualType T, SourceLocation CC,
                   result.Val,
                   Context.getFloatTypeSemantics(QualType(TargetBT, 0)),
                   Context.getFloatTypeSemantics(QualType(SourceBT, 0)))) {
-            if (getLangOpts().SYCLIsDevice)
-              SYCL().DiagIfDeviceCode(CC, diag::warn_imp_float_size_conversion);
-            else
-              DiagnoseImpCast(*this, E, T, CC,
-                              diag::warn_imp_float_size_conversion);
+            if (getLangOpts().isSYCL()) {
+              if (getLangOpts().SYCLIsDevice)
+                SYCL().DiagIfDeviceCode(
+                    CC, diag::warn_sycl_imp_float_size_conversion);
+              else
+                DiagnoseImpCast(*this, E, T, CC,
+                                diag::warn_sycl_imp_float_size_conversion);
+            }
             return;
           }
         }
 
         if (SourceMgr.isInSystemMacro(CC))
           return;
         // If there is a precision conversion between floating point types when
-        // -Wimplicit-float-size-conversion is passed but
+        // -Wsycl-implicit-float-size-conversion is passed but
         // -Wimplicit-float-conversion is not, make sure we emit at least a size
         // warning.
         if (Diags.isIgnored(diag::warn_impcast_float_precision, CC)) {
-          if (getLangOpts().SYCLIsDevice)
-            SYCL().DiagIfDeviceCode(CC, diag::warn_imp_float_size_conversion);
-          else
-            DiagnoseImpCast(*this, E, T, CC,
-                            diag::warn_imp_float_size_conversion);
+          if (getLangOpts().isSYCL()) {
+            if (getLangOpts().SYCLIsDevice)
+              SYCL().DiagIfDeviceCode(
+                  CC, diag::warn_sycl_imp_float_size_conversion);
+            else
+              DiagnoseImpCast(*this, E, T, CC,
+                              diag::warn_sycl_imp_float_size_conversion);
+          }
         }
         DiagnoseImpCast(*this, E, T, CC, diag::warn_impcast_float_precision);
       }

clang/test/CodeGenSYCL/group-local-memory.cpp

@@ -2,15 +2,24 @@
 // SYCL device target, and can be disabled with -fno-sycl-early-optimizations.
 // New pass manager doesn't print all passes tree, only module level.
 //
-// RUN: %clang_cc1 -O2 -fsycl-is-device -triple spir64-unknown-unknown %s -mdebug-pass Structure -emit-llvm -o /dev/null 2>&1 | FileCheck %s --check-prefix=CHECK-NEWPM-EARLYOPT
-// CHECK-NEWPM-EARLYOPT: ConstantMergePass
-// CHECK-NEWPM-EARLYOPT: SYCLMutatePrintfAddrspacePass
+// RUN: %clang_cc1 -O2 -fsycl-is-device -triple spir64-unknown-unknown %s -mdebug-pass Structure -emit-llvm -o /dev/null 2>&1 | FileCheck %s
+// CHECK: SYCLVirtualFunctionsAnalysisPass
+// CHECK: ESIMDVerifierPass
+// CHECK: SYCLConditionalCallOnDevicePass
+// CHECK: SYCLPropagateAspectsUsagePass
+// CHECK: SYCLPropagateJointMatrixUsagePass
+// CHECK: SYCLLowerWGLocalMemoryPass
+// CHECK: InferFunctionAttrsPass
+// CHECK: AlwaysInlinerPass
+// CHECK: ModuleInlinerWrapperPass
+// CHECK: ConstantMergePass
+// CHECK: SYCLMutatePrintfAddrspacePass
+// CHECK: SYCLPropagateAspectsUsagePass
+// CHECK: SYCLAddOptLevelAttributePass
+// CHECK: CompileTimePropertiesPass
+// CHECK: RecordSYCLAspectNamesPass
+// CHECK: CleanupSYCLMetadataPass
 //
-// RUN: %clang_cc1 -O2 -fsycl-is-device -triple spir64-unknown-unknown %s -mdebug-pass Structure -emit-llvm -fno-sycl-early-optimizations -o /dev/null 2>&1 | FileCheck %s --check-prefix=CHECK-NEWPM-NOEARLYOPT
-// CHECK-NEWPM-NOEARLYOPT-NOT: ConstantMergePass
-// CHECK-NEWPM-NOEARLYOPT: SYCLMutatePrintfAddrspacePass
-
-// Checks that the compile time properties pass is added into the compilation pipeline
-//
-// RUN: %clang_cc1 -O2 -fsycl-is-device -triple spir64-unknown-unknown %s -mdebug-pass Structure -emit-llvm -o /dev/null 2>&1 | FileCheck %s --check-prefix=CHECK-COMPTIMEPROPS
-// CHECK-COMPTIMEPROPS: Running pass: CompileTimePropertiesPass on [module]
+// RUN: %clang_cc1 -O2 -fsycl-is-device -triple spir64-unknown-unknown %s -mdebug-pass Structure -emit-llvm -fno-sycl-early-optimizations -o /dev/null 2>&1 | FileCheck %s --check-prefix=CHECK-NOEARLYOPT
+// CHECK-NOEARLYOPT-NOT: ConstantMergePass1
+// CHECK-NOEARLYOPT: SYCLMutatePrintfAddrspacePass