Merge branch 'sycl' into counter-based-2

.github/workflows/sycl-linux-matrix-e2e-on-nightly.yml

@@ -29,7 +29,7 @@ jobs:
             runner: '["Linux", "gen12"]'
             image: ghcr.io/intel/llvm/sycl_ubuntu2204_nightly:latest
             image_options: -u 1001 --device=/dev/dri --privileged --cap-add SYS_ADMIN
-            target_devices: ext_oneapi_level_zero:gpu
+            target_devices: level_zero:gpu
             reset_gpu: true
 
           - name: Intel OCL GPU

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

@@ -90,15 +90,15 @@ jobs:
             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;opencl:gpu;opencl:cpu
+            target_devices: level_zero:gpu;opencl:gpu;opencl:cpu
             reset_gpu: true
             install_drivers: ${{ contains(needs.detect_changes.outputs.filters, 'drivers') }}
             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/ubuntu2204_intel_drivers:latest
             image_options: -u 1001 --device=/dev/dri --privileged --cap-add SYS_ADMIN
-            target_devices: ext_oneapi_level_zero:gpu;opencl:gpu
+            target_devices: level_zero:gpu;opencl:gpu
             reset_gpu: true
             install_drivers: ${{ contains(needs.detect_changes.outputs.filters, 'drivers') }}
             extra_lit_opts: --param matrix-xmx8=True --param gpu-intel-dg2=True
@@ -107,7 +107,7 @@ jobs:
             runner: '["Linux", "arc"]'
             image: ghcr.io/intel/llvm/ubuntu2204_intel_drivers:${{ contains(github.event.pull_request.labels.*.name, 'ci-no-devigc') && 'latest' || 'devigc' }}
             image_options: -u 1001 --device=/dev/dri --privileged --cap-add SYS_ADMIN
-            target_devices: ext_oneapi_level_zero:gpu;opencl:gpu
+            target_devices: level_zero:gpu;opencl:gpu
             reset_gpu: true
             install_drivers: >-
               ${{ contains(needs.detect_changes.outputs.filters, 'drivers') ||

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

@@ -105,7 +105,7 @@ on:
           - 'opencl:cpu'
           - 'opencl:gpu'
           - 'opencl:fpga'
-          - 'ext_oneapi_level_zero:gpu'
+          - 'level_zero:gpu'
           - 'ext_oneapi_hip:gpu'
       tests_selector:
         type: choice
@@ -318,7 +318,6 @@ jobs:
         -DSYCL_CTS_ENABLE_OPENCL_INTEROP_TESTS=OFF \
         -DSYCL_CTS_MEASURE_BUILD_TIMES=ON \
         -DDPCPP_INSTALL_DIR="$(dirname $(which clang++))/.." \
-        -DCMAKE_CXX_FLAGS="-Wno-missing-template-arg-list-after-template-kw" \
         $CMAKE_EXTRA_ARGS
         # Ignore errors so that if one category build fails others still have a
         # chance to finish and be executed at the run stage. Note that

.github/workflows/sycl-nightly.yml

@@ -48,7 +48,7 @@ jobs:
             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
+            target_devices: level_zero:gpu
             reset_gpu: true
             tests_selector: e2e
             extra_lit_opts: --param gpu-intel-gen12=True

.github/workflows/sycl-post-commit.yml

@@ -52,7 +52,7 @@ jobs:
           - name: Intel GEN12 Graphics with Level Zero
             runner: '["Linux", "gen12"]'
             extra_lit_opts: --param gpu-intel-gen12=True
-            target_devices: ext_oneapi_level_zero:gpu;opencl:fpga
+            target_devices: level_zero:gpu;opencl:fpga
           - name: Intel Arc A-Series Graphics with Level Zero
             runner: '["Linux", "arc"]'
             extra_lit_opts: --param matrix-xmx8=True --param gpu-intel-dg2=True
@@ -78,7 +78,7 @@ jobs:
       runner: ${{ matrix. runner }}
       image: ghcr.io/intel/llvm/ubuntu2204_intel_drivers:latest
       image_options: -u 1001 --device=/dev/dri --privileged --cap-add SYS_ADMIN
-      target_devices: ${{ matrix.target_devices || 'ext_oneapi_level_zero:gpu' }}
+      target_devices: ${{ matrix.target_devices || 'level_zero:gpu' }}
       reset_gpu: true
 
       extra_lit_opts: ${{ matrix.extra_lit_opts }}

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

@@ -79,7 +79,7 @@ jobs:
       shell: cmd
       run: |
         mkdir build-e2e
-        cmake -GNinja -B build-e2e -S.\llvm\sycl\test-e2e -DSYCL_TEST_E2E_TARGETS="ext_oneapi_level_zero:gpu" -DCMAKE_CXX_COMPILER="clang++" -DLEVEL_ZERO_LIBS_DIR="D:\github\level-zero_win-sdk\lib" -DLEVEL_ZERO_INCLUDE="D:\github\level-zero_win-sdk\include" -DLLVM_LIT="..\llvm\llvm\utils\lit\lit.py"
+        cmake -GNinja -B build-e2e -S.\llvm\sycl\test-e2e -DSYCL_TEST_E2E_TARGETS="level_zero:gpu" -DCMAKE_CXX_COMPILER="clang++" -DLEVEL_ZERO_LIBS_DIR="D:\github\level-zero_win-sdk\lib" -DLEVEL_ZERO_INCLUDE="D:\github\level-zero_win-sdk\include" -DLLVM_LIT="..\llvm\llvm\utils\lit\lit.py"
     - name: Run End-to-End tests
       shell: bash
       run: |

clang/include/clang/Driver/Options.td

@@ -4041,12 +4041,6 @@ def flink_huge_device_code : Flag<["-"], "flink-huge-device-code">,
 def fno_link_huge_device_code : Flag<["-"], "fno-link-huge-device-code">,
   Group<Link_Group>, HelpText<"Do not generate or use a custom linker script"
   " for huge device code sections (default)">;
-def fsycl_link_huge_device_code : Flag<["-"], "fsycl-link-huge-device-code">,
-  Group<sycl_Group>, Alias<flink_huge_device_code>, HelpText<"Generate and use a custom linker script for huge"
-  " device code sections (deprecated)">, Flags<[Deprecated]>;
-def fno_sycl_link_huge_device_code : Flag<["-"], "fno-sycl-link-huge-device-code">,
-  Group<sycl_Group>, Alias<fno_link_huge_device_code>, HelpText<"Do not generate or use a custom linker script"
-  " for huge device code sections (default) (deprecated)">, Flags<[Deprecated]>;
 defm sycl_id_queries_fit_in_int: BoolFOption<"sycl-id-queries-fit-in-int",
   LangOpts<"SYCLValueFitInMaxInt">, DefaultTrue,
   PosFlag<SetTrue, [], [ClangOption], "Assume">, NegFlag<SetFalse, [], [ClangOption], "Do not assume">,

clang/lib/Basic/Targets/NVPTX.cpp

@@ -175,6 +175,9 @@ void NVPTXTargetInfo::getTargetDefines(const LangOptions &Opts,
                                        MacroBuilder &Builder) const {
   Builder.defineMacro("__PTX__");
   Builder.defineMacro("__NVPTX__");
+  // PTXVersion is the value as parsed so 70 if +ptx70 is given.
+  // For symmetry with __CUDA_ARCH__, make it in the Major * 100 + Minor * 10.
+  Builder.defineMacro("__PTX_VERSION__", Twine(PTXVersion*10));
 
   // Skip setting architecture dependent macros if undefined.
   if (GPU == CudaArch::UNUSED && !HostTarget)

clang/lib/Driver/Driver.cpp

@@ -806,8 +806,10 @@ Driver::OpenMPRuntimeKind Driver::getOpenMPRuntime(const ArgList &Args) const {
 }
 
 static bool isValidSYCLTriple(llvm::Triple T) {
-  // NVPTX is valid for SYCL.
-  if (T.isNVPTX())
+  // 'nvptx64-nvidia-cuda' is the valid SYCL triple for NVidia GPUs.
+  if (T.getArch() == llvm::Triple::nvptx64 &&
+      T.getVendor() == llvm::Triple::NVIDIA &&
+      T.getOS() == llvm::Triple::CUDA && !T.hasEnvironment())
     return true;
 
   // AMDGCN is valid for SYCL
@@ -2407,11 +2409,14 @@ void Driver::PrintHelp(bool ShowHidden) const {
 llvm::Triple Driver::MakeSYCLDeviceTriple(StringRef TargetArch) const {
   SmallVector<StringRef, 5> SYCLAlias = {
       "spir",       "spir64",  "spir64_fpga", "spir64_x86_64",
-      "spir64_gen", "spirv32", "spirv64"};
+      "spir64_gen", "spirv32", "spirv64",     "nvptx64"};
   if (std::find(SYCLAlias.begin(), SYCLAlias.end(), TargetArch) !=
       SYCLAlias.end()) {
     llvm::Triple TT;
     TT.setArchName(TargetArch);
+    // Return the full SYCL target triple string for NVidia GPU targets.
+    if (TT.getArch() == llvm::Triple::nvptx64)
+      return llvm::Triple("nvptx64-nvidia-cuda");
     TT.setVendor(llvm::Triple::UnknownVendor);
     TT.setOS(llvm::Triple::UnknownOS);
     return TT;

clang/lib/Driver/ToolChains/Clang.cpp

@@ -10842,6 +10842,8 @@ void SYCLPostLink::ConstructJob(Compilation &C, const JobAction &JA,
   std::string OutputArg = Output.getFilename();
   if (T.getSubArch() == llvm::Triple::SPIRSubArch_gen && Device.data())
     OutputArg = ("intel_gpu_" + Device + "," + OutputArg).str();
+  else if (T.getSubArch() == llvm::Triple::SPIRSubArch_x86_64)
+    OutputArg = "spir64_x86_64," + OutputArg;
 
   const toolchains::SYCLToolChain &TC =
       static_cast<const toolchains::SYCLToolChain &>(getToolChain());

clang/lib/Sema/SemaDeclAttr.cpp

@@ -8322,90 +8322,6 @@ static void handleSYCLAddIRAnnotationsMemberAttr(Sema &S, Decl *D,
   S.AddSYCLAddIRAnnotationsMemberAttr(D, A, Args);
 }
 
-namespace {
-struct IntrinToName {
-  uint32_t Id;
-  int32_t FullName;
-  int32_t ShortName;
-};
-} // unnamed namespace
-
-static bool ArmBuiltinAliasValid(unsigned BuiltinID, StringRef AliasName,
-                                 ArrayRef<IntrinToName> Map,
-                                 const char *IntrinNames) {
-  AliasName.consume_front("__arm_");
-  const IntrinToName *It =
-      llvm::lower_bound(Map, BuiltinID, [](const IntrinToName &L, unsigned Id) {
-        return L.Id < Id;
-      });
-  if (It == Map.end() || It->Id != BuiltinID)
-    return false;
-  StringRef FullName(&IntrinNames[It->FullName]);
-  if (AliasName == FullName)
-    return true;
-  if (It->ShortName == -1)
-    return false;
-  StringRef ShortName(&IntrinNames[It->ShortName]);
-  return AliasName == ShortName;
-}
-
-static bool ArmMveAliasValid(unsigned BuiltinID, StringRef AliasName) {
-#include "clang/Basic/arm_mve_builtin_aliases.inc"
-  // The included file defines:
-  // - ArrayRef<IntrinToName> Map
-  // - const char IntrinNames[]
-  return ArmBuiltinAliasValid(BuiltinID, AliasName, Map, IntrinNames);
-}
-
-static bool ArmCdeAliasValid(unsigned BuiltinID, StringRef AliasName) {
-#include "clang/Basic/arm_cde_builtin_aliases.inc"
-  return ArmBuiltinAliasValid(BuiltinID, AliasName, Map, IntrinNames);
-}
-
-static bool ArmSveAliasValid(ASTContext &Context, unsigned BuiltinID,
-                             StringRef AliasName) {
-  if (Context.BuiltinInfo.isAuxBuiltinID(BuiltinID))
-    BuiltinID = Context.BuiltinInfo.getAuxBuiltinID(BuiltinID);
-  return BuiltinID >= AArch64::FirstSVEBuiltin &&
-         BuiltinID <= AArch64::LastSVEBuiltin;
-}
-
-static bool ArmSmeAliasValid(ASTContext &Context, unsigned BuiltinID,
-                             StringRef AliasName) {
-  if (Context.BuiltinInfo.isAuxBuiltinID(BuiltinID))
-    BuiltinID = Context.BuiltinInfo.getAuxBuiltinID(BuiltinID);
-  return BuiltinID >= AArch64::FirstSMEBuiltin &&
-         BuiltinID <= AArch64::LastSMEBuiltin;
-}
-
-static void handleArmBuiltinAliasAttr(Sema &S, Decl *D, const ParsedAttr &AL) {
-  if (!AL.isArgIdent(0)) {
-    S.Diag(AL.getLoc(), diag::err_attribute_argument_n_type)
-        << AL << 1 << AANT_ArgumentIdentifier;
-    return;
-  }
-
-  IdentifierInfo *Ident = AL.getArgAsIdent(0)->Ident;
-  unsigned BuiltinID = Ident->getBuiltinID();
-  StringRef AliasName = cast<FunctionDecl>(D)->getIdentifier()->getName();
-
-  bool IsAArch64 = S.Context.getTargetInfo().getTriple().isAArch64();
-  if ((IsAArch64 && !ArmSveAliasValid(S.Context, BuiltinID, AliasName) &&
-       !ArmSmeAliasValid(S.Context, BuiltinID, AliasName)) ||
-      (!IsAArch64 && !ArmMveAliasValid(BuiltinID, AliasName) &&
-       !ArmCdeAliasValid(BuiltinID, AliasName))) {
-    S.Diag(AL.getLoc(), diag::err_attribute_arm_builtin_alias);
-    return;
-  }
-
-  D->addAttr(::new (S.Context) ArmBuiltinAliasAttr(S.Context, AL, Ident));
-}
-
-static bool RISCVAliasValid(unsigned BuiltinID, StringRef AliasName) {
-  return BuiltinID >= RISCV::FirstRVVBuiltin &&
-         BuiltinID <= RISCV::LastRVVBuiltin;
-}
-
 static bool SYCLAliasValid(ASTContext &Context, unsigned BuiltinID) {
   constexpr llvm::StringLiteral Prefix = "__builtin_intel_sycl";
   return Context.BuiltinInfo.getName(BuiltinID).starts_with(Prefix);