Merge remote-tracking branch 'intel_llvm/sycl' into matrixuse

Author: yubingex007-a11y

.github/workflows/sycl_resolve_test_matrix.yml renamed to .github/workflows/sycl_gen_test_matrix.yml

@@ -24,16 +24,21 @@ on:
         type: string
         required: false
         default: ${{ github.run_id }}-${{ github.run_attempt }}
+      ref:
+        description: "Commit to use for checkout or github.sha by default"
+        type: string
+        required: false
+        default: ${{ github.sha }}
     outputs:
       lts_matrix:
         description: "Generated Matrix"
-        value: ${{ jobs.resolve_matrix.outputs.lts_matrix }}
+        value: ${{ jobs.test_matrix.outputs.lts_matrix }}
       lts_aws_matrix:
         description: "Generated Matrix AWS subset"
-        value: ${{ jobs.resolve_matrix.outputs.lts_aws_matrix }}
+        value: ${{ jobs.test_matrix.outputs.lts_aws_matrix }}
 jobs:
-  resolve_matrix:
-    name: Resolve Test Matrix
+  test_matrix:
+    name: Generate Test Matrix
     runs-on: ubuntu-latest
     outputs:
       lts_matrix: ${{ steps.work.outputs.lts_matrix }}
@@ -42,11 +47,11 @@ jobs:
     - name: Download scripts and configs
       shell: bash
       run: |
-        wget raw.githubusercontent.com/intel/llvm/${{ github.sha }}/devops/scripts/generate_test_matrix.js
-        wget raw.githubusercontent.com/intel/llvm/${{ github.sha }}/devops/test_configs.json
+        wget raw.githubusercontent.com/intel/llvm/${{ inputs.ref }}/devops/scripts/generate_test_matrix.js
+        wget raw.githubusercontent.com/intel/llvm/${{ inputs.ref }}/devops/test_configs.json
         wget raw.githubusercontent.com/intel/llvm/sycl/devops/dependencies.json
         mv dependencies.json dependencies.sycl.json
-        wget raw.githubusercontent.com/intel/llvm/${{ github.sha }}/devops/dependencies.json
+        wget raw.githubusercontent.com/intel/llvm/${{ inputs.ref }}/devops/dependencies.json
     - id: work
       uses: actions/github-script@v6
       name: Generate matrix

.github/workflows/sycl_nightly.yml

@@ -10,16 +10,16 @@ on:
       - '.github/workflows/sycl_nightly.yml'
 
 jobs:
-  resolve_matrix:
-    name: Resolve Test Matrix
-    uses: ./.github/workflows/sycl_resolve_test_matrix.yml
+  test_matrix:
+    name: Generate Test Matrix
+    uses: ./.github/workflows/sycl_gen_test_matrix.yml
     with:
       lts_config: "ocl_gen9;ocl_x64"
 
   ubuntu2004_build_test:
     if: github.repository == 'intel/llvm'
     uses: ./.github/workflows/sycl_linux_build_and_test.yml
-    needs: resolve_matrix
+    needs: test_matrix
     secrets: inherit
     with:
       build_cache_root: "/__w/"
@@ -29,7 +29,7 @@ jobs:
   ubuntu2004_opaque_pointers_build_test:
     if: github.repository == 'intel/llvm'
     uses: ./.github/workflows/sycl_linux_build_and_test.yml
-    needs: resolve_matrix
+    needs: test_matrix
     secrets: inherit
     with:
       build_cache_root: "/__w/"

.github/workflows/sycl_post_commit.yml

@@ -12,21 +12,21 @@ on:
 
 jobs:
   # This job generates matrix of tests for LLVM Test Suite
-  resolve_matrix:
-    name: Resolve Test Matrix
-    uses: ./.github/workflows/sycl_resolve_test_matrix.yml
+  test_matrix:
+    name: Generate Test Matrix
+    uses: ./.github/workflows/sycl_gen_test_matrix.yml
     with:
       lts_config: "l0_gen9"
   linux_default:
     name: Linux Default
-    needs: resolve_matrix
+    needs: test_matrix
     uses: ./.github/workflows/sycl_linux_build_and_test.yml
     secrets: inherit
     with:
       build_cache_root: "/__w/llvm"
       build_artifact_suffix: "post_commit"
-      lts_matrix: ${{ needs.resolve_matrix.outputs.lts_matrix }}
-      lts_aws_matrix: ${{ needs.resolve_matrix.outputs.lts_aws_matrix }}
+      lts_matrix: ${{ needs.test_matrix.outputs.lts_matrix }}
+      lts_aws_matrix: ${{ needs.test_matrix.outputs.lts_aws_matrix }}
   linux_no_assert:
     name: Linux (no assert)
     uses: ./.github/workflows/sycl_linux_build_and_test.yml

.github/workflows/sycl_precommit.yml

@@ -38,17 +38,18 @@ jobs:
       uses: ./devops/actions/clang-format
 
   # This job generates matrix of tests for LLVM Test Suite
-  resolve_matrix:
-    name: Resolve Test Matrix
-    uses: ./.github/workflows/sycl_resolve_test_matrix.yml
+  test_matrix:
+    name: Generate Test Matrix
+    uses: ./.github/workflows/sycl_gen_test_matrix.yml
     with:
+      ref: ${{ github.event.pull_request.head.sha }}
       lts_config: "hip_amdgpu;ocl_x64;ocl_gen9;l0_gen9;esimd_emu;cuda"
 
   linux_default:
     name: Linux
     # Only build and test patches, that have passed all linter checks, because
     # the next commit is likely to be a follow-up on that job.
-    needs: [lint, resolve_matrix]
+    needs: [lint, test_matrix]
     if: always() && (success() || contains(github.event.pull_request.labels.*.name, 'ignore-lint'))
     uses: ./.github/workflows/sycl_linux_build_and_test.yml
     secrets: inherit
@@ -58,5 +59,5 @@ jobs:
       build_cache_size: "8G"
       build_artifact_suffix: "default"
       build_cache_suffix: "default"
-      lts_matrix: ${{ needs.resolve_matrix.outputs.lts_matrix }}
-      lts_aws_matrix: ${{ needs.resolve_matrix.outputs.lts_aws_matrix }}
+      lts_matrix: ${{ needs.test_matrix.outputs.lts_matrix }}
+      lts_aws_matrix: ${{ needs.test_matrix.outputs.lts_aws_matrix }}

buildbot/configure.py

@@ -44,6 +44,10 @@ def do_configure(args):
     sycl_enable_xpti_tracing = 'ON'
     xpti_enable_werror = 'OFF'
 
+    # lld is needed on Windows or for the HIP plugin on AMD
+    if platform.system() == 'Windows' or (args.hip and args.hip_platform == 'AMD'):
+        llvm_enable_projects += ';lld'
+
     # replace not append, so ARM ^ X86
     if args.arm:
         llvm_targets_to_build = 'ARM;AArch64'
@@ -65,8 +69,6 @@ def do_configure(args):
             llvm_targets_to_build += ';AMDGPU'
             libclc_targets_to_build += libclc_amd_target_names
 
-            # The HIP plugin for AMD uses lld for linking
-            llvm_enable_projects += ';lld'
         elif args.hip_platform == 'NVIDIA' and not args.cuda:
             llvm_targets_to_build += ';NVPTX'
             libclc_targets_to_build += libclc_nvidia_target_names

buildbot/dependency.conf

@@ -4,8 +4,8 @@ ocl_cpu_rt_ver=2022.14.8.0.04
 # https://github.com/intel/llvm/releases/download/2022-WW33/win-oclcpuexp-2022.14.8.0.04_rel.zip
 ocl_cpu_rt_ver_win=2022.14.8.0.04
 # Same GPU driver supports Level Zero and OpenCL
-# https://github.com/intel/compute-runtime/releases/tag/22.31.23852
-ocl_gpu_rt_ver=22.31.23852
+# https://github.com/intel/compute-runtime/releases/tag/22.35.24055
+ocl_gpu_rt_ver=22.35.24055
 # Same GPU driver supports Level Zero and OpenCL
 # https://downloadmirror.intel.com/723911/igfx_win_101.1404.zip
 ocl_gpu_rt_ver_win=101.1404
@@ -31,7 +31,7 @@ ocloc_ver_win=101.1404
 [DRIVER VERSIONS]
 cpu_driver_lin=2022.14.8.0.04
 cpu_driver_win=2022.14.8.0.04
-gpu_driver_lin=22.31.23852
+gpu_driver_lin=22.35.24055
 gpu_driver_win=101.1404
 fpga_driver_lin=2022.14.8.0.04
 fpga_driver_win=2022.14.8.0.04

clang/include/clang/Basic/Attr.td

@@ -1281,6 +1281,23 @@ def SYCLSpecialClass: InheritableAttr {
   let Documentation = [SYCLSpecialClassDocs];
 }
 
+def SYCLType: InheritableAttr {
+  let Spellings = [CXX11<"__sycl_detail__", "sycl_type">];
+  let Subjects = SubjectList<[CXXRecord, Enum], ErrorDiag>;
+  let LangOpts = [SYCLIsDevice, SYCLIsHost];
+  let Args = [EnumArgument<"Type", "SYCLType",
+                           ["accessor", "local_accessor", "spec_constant",
+                            "specialization_id", "kernel_handler", "buffer_location",
+                            "no_alias", "accessor_property_list", "group",
+                            "private_memory", "aspect"],
+                           ["accessor", "local_accessor", "spec_constant",
+                            "specialization_id", "kernel_handler", "buffer_location",
+                            "no_alias", "accessor_property_list", "group",
+                            "private_memory", "aspect"]>];
+  // Only used internally by SYCL implementation
+  let Documentation = [InternalOnly];
+}
+
 def SYCLDeviceHas : InheritableAttr {
   let Spellings = [CXX11<"sycl", "device_has">];
   let Subjects = SubjectList<[Function], ErrorDiag>;

clang/include/clang/Basic/DiagnosticSemaKinds.td

@@ -4098,6 +4098,8 @@ def warn_transparent_union_attribute_zero_fields : Warning<
 def warn_attribute_type_not_supported : Warning<
   "%0 attribute argument not supported: %1">,
   InGroup<IgnoredAttributes>;
+def err_attribute_argument_not_supported : Error<
+  "%0 attribute argument %1 is not supported">;
 def warn_attribute_unknown_visibility : Warning<"unknown visibility %0">,
   InGroup<IgnoredAttributes>;
 def warn_attribute_protected_visibility :

clang/include/clang/Driver/Options.td

@@ -956,6 +956,12 @@ def cuda_include_ptx_EQ : Joined<["--"], "cuda-include-ptx=">, Flags<[NoXarchOpt
   HelpText<"Include PTX for the following GPU architecture (e.g. sm_35) or 'all'. May be specified more than once.">;
 def no_cuda_include_ptx_EQ : Joined<["--"], "no-cuda-include-ptx=">, Flags<[NoXarchOption]>,
   HelpText<"Do not include PTX for the following GPU architecture (e.g. sm_35) or 'all'. May be specified more than once.">;
+def fno_bundle_offload_arch : Flag<["-"], "fno-bundle-offload-arch">, 
+  HelpText<"Specify that the offload bundler should not identify a bundle with "
+            "specific arch. For example, the bundle for `nvptx64-nvidia-cuda-sm_80` "
+            "uses the bundle tag `nvptx64-nvidia-cuda` when used. "
+            "This allows .o files to contain .bc bundles that are unspecific "
+            "to a particular arch version.">;
 def offload_arch_EQ : Joined<["--"], "offload-arch=">, Flags<[NoXarchOption]>,
   HelpText<"CUDA offloading device architecture (e.g. sm_35), or HIP offloading target ID in the form of a "
            "device architecture followed by target ID features delimited by a colon. Each target ID feature "

clang/include/clang/Sema/Sema.h

@@ -10858,6 +10858,9 @@ class Sema final {
   ReqdWorkGroupSizeAttr *
   MergeReqdWorkGroupSizeAttr(Decl *D, const ReqdWorkGroupSizeAttr &A);
 
+  SYCLTypeAttr *MergeSYCLTypeAttr(Decl *D, const AttributeCommonInfo &CI,
+                                  SYCLTypeAttr::SYCLType TypeName);
+
   /// Only called on function definitions; if there is a MSVC #pragma optimize
   /// in scope, consider changing the function's attributes based on the
   /// optimization list passed to the pragma.
@@ -13546,12 +13549,16 @@ class Sema final {
     const CXXRecordDecl *RecTy = Ty->getAsCXXRecordDecl();
     if (!RecTy)
       return false;
+
+    if (RecTy->hasAttr<AttrTy>())
+      return true;
+
     if (auto *CTSD = dyn_cast<ClassTemplateSpecializationDecl>(RecTy)) {
       ClassTemplateDecl *Template = CTSD->getSpecializedTemplate();
       if (CXXRecordDecl *RD = Template->getTemplatedDecl())
         return RD->hasAttr<AttrTy>();
     }
-    return RecTy->hasAttr<AttrTy>();
+    return false;
   }
 
 private:
@@ -13975,7 +13982,7 @@ class Sema final {
     KernelConstStaticVariable
   };
 
-  bool isKnownGoodSYCLDecl(const Decl *D);
+  bool isDeclAllowedInSYCLDeviceCode(const Decl *D);
   void checkSYCLDeviceVarDecl(VarDecl *Var);
   void copySYCLKernelAttrs(const CXXRecordDecl *KernelObj);
   void ConstructOpenCLKernel(FunctionDecl *KernelCallerFunc, MangleContext &MC);

clang/lib/CodeGen/CodeGenFunction.cpp

@@ -1600,7 +1600,7 @@ void CodeGenFunction::GenerateCode(GlobalDecl GD, llvm::Function *Fn,
   StartFunction(GD, ResTy, Fn, FnInfo, Args, Loc, BodyRange.getBegin());
   if (!getLangOpts().OptRecordFile.empty()) {
     SyclOptReportHandler &SyclOptReport = CGM.getDiags().getSYCLOptReport();
-    if (Fn && SyclOptReport.HasOptReportInfo(FD)) {
+    if (SyclOptReport.HasOptReportInfo(FD)) {
       llvm::OptimizationRemarkEmitter ORE(Fn);
       for (auto ORI : llvm::enumerate(SyclOptReport.GetInfo(FD))) {
         llvm::DiagnosticLocation DL =