From b0eee29c7415dfa1599e4a02018a7e99465d4253 Mon Sep 17 00:00:00 2001
From: Yaxun Liu <Yaxun.Liu@amd.com>
Date: Thu, 29 Mar 2018 14:50:00 +0000
Subject: [PATCH] Disable emitting static extern C aliases for amdgcn target
 for CUDA

Patch by Greg Rodgers.
Revised and lit test added by Yaxun Liu.

Differential Revision: https://reviews.llvm.org/D44987

llvm-svn: 328793
---
 clang/lib/CodeGen/CodeGenModule.cpp |  4 +---
 clang/lib/CodeGen/TargetInfo.cpp    | 10 ++++++++++
 clang/lib/CodeGen/TargetInfo.h      |  5 +++++
 clang/test/CodeGenCUDA/alias.cu     |  3 +++
 4 files changed, 19 insertions(+), 3 deletions(-)

diff --git a/clang/lib/CodeGen/CodeGenModule.cpp b/clang/lib/CodeGen/CodeGenModule.cpp
index e1cb324..bc7ba90 100644
--- a/clang/lib/CodeGen/CodeGenModule.cpp
+++ b/clang/lib/CodeGen/CodeGenModule.cpp
@@ -4677,9 +4677,7 @@ static void EmitGlobalDeclMetadata(CodeGenModule &CGM,
 /// to such functions with an unmangled name from inline assembly within the
 /// same translation unit.
 void CodeGenModule::EmitStaticExternCAliases() {
-  // Don't do anything if we're generating CUDA device code -- the NVPTX
-  // assembly target doesn't support aliases.
-  if (Context.getTargetInfo().getTriple().isNVPTX())
+  if (!getTargetCodeGenInfo().shouldEmitStaticExternCAliases())
     return;
   for (auto &I : StaticExternCValues) {
     IdentifierInfo *Name = I.first;
diff --git a/clang/lib/CodeGen/TargetInfo.cpp b/clang/lib/CodeGen/TargetInfo.cpp
index 8f3676e..b6f265a 100644
--- a/clang/lib/CodeGen/TargetInfo.cpp
+++ b/clang/lib/CodeGen/TargetInfo.cpp
@@ -6154,6 +6154,7 @@ public:
 
   void setTargetAttributes(const Decl *D, llvm::GlobalValue *GV,
                            CodeGen::CodeGenModule &M) const override;
+  bool shouldEmitStaticExternCAliases() const override;
 
 private:
   // Adds a NamedMDNode with F, Name, and Operand as operands, and adds the
@@ -6275,6 +6276,10 @@ void NVPTXTargetCodeGenInfo::addNVVMMetadata(llvm::Function *F, StringRef Name,
   // Append metadata to nvvm.annotations
   MD->addOperand(llvm::MDNode::get(Ctx, MDVals));
 }
+
+bool NVPTXTargetCodeGenInfo::shouldEmitStaticExternCAliases() const {
+  return false;
+}
 }
 
 //===----------------------------------------------------------------------===//
@@ -7646,6 +7651,7 @@ public:
   createEnqueuedBlockKernel(CodeGenFunction &CGF,
                             llvm::Function *BlockInvokeFunc,
                             llvm::Value *BlockLiteral) const override;
+  bool shouldEmitStaticExternCAliases() const override;
 };
 }
 
@@ -7777,6 +7783,10 @@ AMDGPUTargetCodeGenInfo::getLLVMSyncScopeID(SyncScope S,
   return C.getOrInsertSyncScopeID(Name);
 }
 
+bool AMDGPUTargetCodeGenInfo::shouldEmitStaticExternCAliases() const {
+  return false;
+}
+
 //===----------------------------------------------------------------------===//
 // SPARC v8 ABI Implementation.
 // Based on the SPARC Compliance Definition version 2.4.1.
diff --git a/clang/lib/CodeGen/TargetInfo.h b/clang/lib/CodeGen/TargetInfo.h
index f7a813e..533e145 100644
--- a/clang/lib/CodeGen/TargetInfo.h
+++ b/clang/lib/CodeGen/TargetInfo.h
@@ -296,6 +296,11 @@ public:
   createEnqueuedBlockKernel(CodeGenFunction &CGF,
                             llvm::Function *BlockInvokeFunc,
                             llvm::Value *BlockLiteral) const;
+
+  /// \return true if the target supports alias from the unmangled name to the
+  /// mangled name of functions declared within an extern "C" region and marked
+  /// as 'used', and having internal linkage.
+  virtual bool shouldEmitStaticExternCAliases() const { return true; }
 };
 
 } // namespace CodeGen
diff --git a/clang/test/CodeGenCUDA/alias.cu b/clang/test/CodeGenCUDA/alias.cu
index 6efff6b..56e595e 100644
--- a/clang/test/CodeGenCUDA/alias.cu
+++ b/clang/test/CodeGenCUDA/alias.cu
@@ -1,8 +1,11 @@
 // REQUIRES: x86-registered-target
 // REQUIRES: nvptx-registered-target
+// REQUIRES: amdgpu-registered-target
 
 // RUN: %clang_cc1 -fcuda-is-device -triple nvptx-nvidia-cuda -emit-llvm \
 // RUN:   -o - %s | FileCheck %s
+// RUN: %clang_cc1 -fcuda-is-device -triple amdgcn -emit-llvm \
+// RUN:   -o - %s | FileCheck %s
 
 #include "Inputs/cuda.h"
 
-- 
2.7.4