[OpenMP][Fix] Make the arch selector for x86_64 work

[jdoerfert]

Summary

The triple uses a bar "x86-64" instead of an underscore. Since we
have troubles accepting x86-64 as an identifier, we stick with
x86_64 in the frontend and translate it explicitly.

[tianshilei1992]

Is it because the triple is separated by -?

[jdoerfert]

In D93786#2470796, @tianshilei1992 wrote:

Is it because the triple is separated by -?

x86 and 64 is separated by - in the architecture list so getArchTypeForLLVMName will not recognize "x86_64".

clang/test/OpenMP/declare_variant_ast_x86_64.c

+// RUN: %clang_cc1 -verify -fopenmp -triple x86_64-unknown-unknown %s -ast-dump | FileCheck %s
+// expected-no-diagnostics
+
+#pragma omp begin declare variant match(device={arch(x86_64)})
+
+void bar() {}
+
+// CHECK: FunctionDecl {{.*}} bar[device={arch(x86_64)}] 'void ()'
+
+#pragma omp end declare variant

llvm/lib/Frontend/OpenMP/OMPContext.cpp

 //===- OMPContext.cpp ------ Collection of helpers for OpenMP contexts ----===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 /// \file
 ///
 /// This file implements helper functions and classes to deal with OpenMP
 /// contexts as used by `[begin/end] declare variant` and `metadirective`.
 ///
 //===----------------------------------------------------------------------===//
 
 #include "llvm/Frontend/OpenMP/OMPContext.h"
 #include "llvm/ADT/SetOperations.h"
+#include "llvm/ADT/StringRef.h"
 #include "llvm/ADT/StringSwitch.h"
+#include "llvm/ADT/Triple.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/raw_ostream.h"
 
 #define DEBUG_TYPE "openmp-ir-builder"
 
 using namespace llvm;
 using namespace omp;
 
case Triple::nvptx64:
     ActiveTraits.set(unsigned(TraitProperty::device_kind_gpu));
     break;
   default:
     break;
   }
 
   // Add the appropriate device architecture trait based on the triple.
 #define OMP_TRAIT_PROPERTY(Enum, TraitSetEnum, TraitSelectorEnum, Str)         \
-  if (TraitSelector::TraitSelectorEnum == TraitSelector::device_arch)          \
+  if (TraitSelector::TraitSelectorEnum == TraitSelector::device_arch) {        \
     if (TargetTriple.getArch() == TargetTriple.getArchTypeForLLVMName(Str))    \
-      ActiveTraits.set(unsigned(TraitProperty::Enum));
+      ActiveTraits.set(unsigned(TraitProperty::Enum));                         \
+    if (Str == "x86_64" && TargetTriple.getArch() == Triple::x86_64)           \
+      ActiveTraits.set(unsigned(TraitProperty::Enum));                         \
+  }
 #include "llvm/Frontend/OpenMP/OMPKinds.def"
 
   // TODO: What exactly do we want to see as device ISA trait?
   //       The discussion on the list did not seem to have come to an agreed
   //       upon solution.
 
   // LLVM is the "OpenMP vendor" but we could also interpret vendor as the
   // target vendor.