[SYCL][NativeCPU] Copy over more host/aux target data.

clang/lib/Basic/Targets/NativeCPU.cpp

@@ -60,50 +60,12 @@ NativeCPUTargetInfo::NativeCPUTargetInfo(const llvm::Triple &,
   }());
   if (HostTriple.getArch() != llvm::Triple::UnknownArch) {
     HostTarget = AllocateTarget(HostTriple, Opts);
-
-    // Copy properties from host target.
-    BoolWidth = HostTarget->getBoolWidth();
-    BoolAlign = HostTarget->getBoolAlign();
-    IntWidth = HostTarget->getIntWidth();
-    IntAlign = HostTarget->getIntAlign();
-    HalfWidth = HostTarget->getHalfWidth();
-    HalfAlign = HostTarget->getHalfAlign();
-    FloatWidth = HostTarget->getFloatWidth();
-    FloatAlign = HostTarget->getFloatAlign();
-    DoubleWidth = HostTarget->getDoubleWidth();
-    DoubleAlign = HostTarget->getDoubleAlign();
-    LongWidth = HostTarget->getLongWidth();
-    LongAlign = HostTarget->getLongAlign();
-    LongLongWidth = HostTarget->getLongLongWidth();
-    LongLongAlign = HostTarget->getLongLongAlign();
-    PointerWidth = HostTarget->getPointerWidth(LangAS::Default);
-    PointerAlign = HostTarget->getPointerAlign(LangAS::Default);
-    MinGlobalAlign = HostTarget->getMinGlobalAlign(/*TypeSize=*/0,
-                                                   /*HasNonWeakDef=*/true);
-    NewAlign = HostTarget->getNewAlign();
-    DefaultAlignForAttributeAligned =
-        HostTarget->getDefaultAlignForAttributeAligned();
-    SizeType = HostTarget->getSizeType();
-    PtrDiffType = HostTarget->getPtrDiffType(LangAS::Default);
-    IntMaxType = HostTarget->getIntMaxType();
-    WCharType = HostTarget->getWCharType();
-    WIntType = HostTarget->getWIntType();
-    Char16Type = HostTarget->getChar16Type();
-    Char32Type = HostTarget->getChar32Type();
-    Int64Type = HostTarget->getInt64Type();
-    SigAtomicType = HostTarget->getSigAtomicType();
-    ProcessIDType = HostTarget->getProcessIDType();
-
-    UseBitFieldTypeAlignment = HostTarget->useBitFieldTypeAlignment();
-    UseZeroLengthBitfieldAlignment =
-        HostTarget->useZeroLengthBitfieldAlignment();
-    UseExplicitBitFieldAlignment = HostTarget->useExplicitBitFieldAlignment();
-    ZeroLengthBitfieldBoundary = HostTarget->getZeroLengthBitfieldBoundary();
-
-    // This is a bit of a lie, but it controls __GCC_ATOMIC_XXX_LOCK_FREE, and
-    // we need those macros to be identical on host and device, because (among
-    // other things) they affect which standard library classes are defined,
-    // and we need all classes to be defined on both the host and device.
-    MaxAtomicInlineWidth = HostTarget->getMaxAtomicInlineWidth();
+    copyAuxTarget(&*HostTarget);
   }
 }
+
+void NativeCPUTargetInfo::setAuxTarget(const TargetInfo *Aux) {
+  assert(Aux && "Cannot invoke setAuxTarget without a valid auxiliary target!");
+  copyAuxTarget(Aux);
+  getTargetOpts() = Aux->getTargetOpts();
+}

clang/lib/Basic/Targets/NativeCPU.h

@@ -57,6 +57,8 @@ class LLVM_LIBRARY_VISIBILITY NativeCPUTargetInfo final : public TargetInfo {
   }
 
 protected:
+  void setAuxTarget(const TargetInfo *Aux) override;
+
   ArrayRef<const char *> getGCCRegNames() const override { return {}; }
 
   ArrayRef<TargetInfo::GCCRegAlias> getGCCRegAliases() const override {

clang/lib/Driver/Compilation.cpp

@@ -7,7 +7,6 @@
 //===----------------------------------------------------------------------===//
 
 #include "clang/Driver/Compilation.h"
-#include "ToolChains/SYCL.h"
 #include "clang/Basic/LLVM.h"
 #include "clang/Driver/Action.h"
 #include "clang/Driver/Driver.h"
@@ -128,8 +127,7 @@ Compilation::getArgsForToolChain(const ToolChain *TC, StringRef BoundArch,
     if (DeviceOffloadKind == Action::OFK_OpenMP ||
         DeviceOffloadKind == Action::OFK_SYCL) {
       const ToolChain *HostTC = getSingleOffloadToolChain<Action::OFK_Host>();
-      bool SameTripleAsHost = (TC->getTriple() == HostTC->getTriple()) ||
-                              isSYCLNativeCPU(TC->getTriple());
+      bool SameTripleAsHost = (TC->getTriple() == HostTC->getTriple());
       OffloadArgs = TC->TranslateOffloadTargetArgs(
           *TranslatedArgs, SameTripleAsHost, AllocatedArgs, DeviceOffloadKind);
     }

clang/test/CodeGenSYCL/native_cpu_target_features.cpp

@@ -0,0 +1,17 @@
+// RUN: %clang_cc1 -triple native_cpu -aux-triple x86_64-unknown-linux-gnu -fsycl-is-device -emit-llvm -fsycl-is-native-cpu -o - %s | FileCheck %s --check-prefixes=CHECK,NOAVX
+// RUN: %clang_cc1 -triple native_cpu -aux-triple x86_64-unknown-linux-gnu -aux-target-cpu skylake -fsycl-is-device -emit-llvm -fsycl-is-native-cpu -o - %s | FileCheck %s --check-prefixes=CHECK,AVX
+// RUN: %clang_cc1 -triple native_cpu -aux-triple x86_64-unknown-linux-gnu -aux-target-feature +avx -fsycl-is-device -emit-llvm -fsycl-is-native-cpu -o - %s | FileCheck %s --check-prefixes=CHECK,AVX
+
+#include "Inputs/sycl.hpp"
+using namespace sycl;
+
+class Test;
+int main() {
+  sycl::queue deviceQueue;
+  deviceQueue.submit([&](handler &h) { h.single_task<Test>([=] {}); });
+}
+
+// CHECK: void @_ZTS4Test() [[ATTRS:#[0-9]+]]
+// CHECK: [[ATTRS]] = {
+// NOAVX-NOT: "target-features"="{{[^"]*}}+avx{{[^"]*}}"
+// AVX-SAME: "target-features"="{{[^"]*}}+avx{{[^"]*}}"

clang/test/Driver/sycl-native-cpu.cpp

@@ -25,7 +25,7 @@
 // CHECK-OPTS: clang{{.*}}"-triple" "[[TRIPLE]]"{{.*}}"-fsycl-is-device"
 // CHECK-OPTS-NOT: -sycl-opt
 // CHECK-OPTS-SAME: "-Wno-override-module" "-mllvm" "-sycl-native-cpu-backend"
-// CHECK-OPTS-SAME: "-target-feature" "+v9.4a"
+// CHECK-OPTS-SAME: "-aux-target-feature" "+v9.4a"
 
 // RUN: %clangxx -fsycl -fsycl-targets=spir64 %s -### 2>&1 | FileCheck -check-prefix=CHECK-NONATIVECPU %s
 // CHECK-NONATIVECPU-NOT: "-D" "__SYCL_NATIVE_CPU__"

libclc/utils/prepare-builtins.cpp

@@ -116,7 +116,12 @@ int main(int argc, char **argv) {
   // functions were inlined prior to incompatible functions pass. Now that the
   // inliner runs later in the pipeline we have to remove all of the target
   // features, so libclc functions will not be earmarked for deletion.
-  if (M->getTargetTriple().str().find("amdgcn") != std::string::npos) {
+  //
+  // NativeCPU uses the same builtins for multiple host targets and should
+  // likewise not have features that limit the builtins to any particular
+  // target.
+  if (M->getTargetTriple().str().find("amdgcn") != std::string::npos ||
+      M->getTargetTriple().str() != "native_cpu") {
     AttributeMask AM;
     AM.addAttribute("target-features");
     AM.addAttribute("target-cpu");