3 files changed, 42 insertions, 12 deletions

diff --git a/clang/lib/CodeGen/Targets/AMDGPU.cpp b/clang/lib/CodeGen/Targets/AMDGPU.cpp
index 16d5919..e4ad078 100644
--- a/clang/lib/CodeGen/Targets/AMDGPU.cpp
+++ b/clang/lib/CodeGen/Targets/AMDGPU.cpp
@@ -439,9 +439,11 @@ void AMDGPUTargetCodeGenInfo::setTargetAttributes(
     return;
 
   const FunctionDecl *FD = dyn_cast_or_null<FunctionDecl>(D);
-  if (FD)
+  if (FD) {
     setFunctionDeclAttributes(FD, F, M);
-
+    if (FD->hasAttr<DeviceKernelAttr>() && !M.getLangOpts().OpenCL)
+      F->setCallingConv(getDeviceKernelCallingConv());
+  }
   if (!getABIInfo().getCodeGenOpts().EmitIEEENaNCompliantInsts)
     F->addFnAttr("amdgpu-ieee", "false");
 }
@@ -508,6 +510,10 @@ AMDGPUTargetCodeGenInfo::getLLVMSyncScopeID(const LangOptions &LangOpts,
   case SyncScope::WavefrontScope:
     Name = "wavefront";
     break;
+  case SyncScope::HIPCluster:
+  case SyncScope::ClusterScope:
+    Name = "cluster";
+    break;
   case SyncScope::HIPWorkgroup:
   case SyncScope::OpenCLWorkGroup:
   case SyncScope::WorkgroupScope:
@@ -654,7 +660,7 @@ llvm::Value *AMDGPUTargetCodeGenInfo::createEnqueuedBlockKernel(
   // kernel address (only the kernel descriptor).
   auto *F = llvm::Function::Create(FT, llvm::GlobalValue::InternalLinkage, Name,
                                    &Mod);
-  F->setCallingConv(llvm::CallingConv::AMDGPU_KERNEL);
+  F->setCallingConv(getDeviceKernelCallingConv());
 
   llvm::AttrBuilder KernelAttrs(C);
   // FIXME: The invoke isn't applying the right attributes either
diff --git a/clang/lib/CodeGen/Targets/NVPTX.cpp b/clang/lib/CodeGen/Targets/NVPTX.cpp
index 53f2fc4..f6715861 100644
--- a/clang/lib/CodeGen/Targets/NVPTX.cpp
+++ b/clang/lib/CodeGen/Targets/NVPTX.cpp
@@ -264,7 +264,7 @@ void NVPTXTargetCodeGenInfo::setTargetAttributes(
       // And kernel functions are not subject to inlining
       F->addFnAttr(llvm::Attribute::NoInline);
       if (FD->hasAttr<CUDAGlobalAttr>()) {
-        F->setCallingConv(llvm::CallingConv::PTX_Kernel);
+        F->setCallingConv(getDeviceKernelCallingConv());
 
         for (auto IV : llvm::enumerate(FD->parameters()))
           if (IV.value()->hasAttr<CUDAGridConstantAttr>())
@@ -278,7 +278,7 @@ void NVPTXTargetCodeGenInfo::setTargetAttributes(
   }
   // Attach kernel metadata directly if compiling for NVPTX.
   if (FD->hasAttr<DeviceKernelAttr>())
-    F->setCallingConv(llvm::CallingConv::PTX_Kernel);
+    F->setCallingConv(getDeviceKernelCallingConv());
 }
 
 void NVPTXTargetCodeGenInfo::addNVVMMetadata(llvm::GlobalValue *GV,
diff --git a/clang/lib/CodeGen/Targets/SPIR.cpp b/clang/lib/CodeGen/Targets/SPIR.cpp
index 3f6d4e0..15d0b35 100644
--- a/clang/lib/CodeGen/Targets/SPIR.cpp
+++ b/clang/lib/CodeGen/Targets/SPIR.cpp
@@ -64,6 +64,8 @@ public:
   llvm::Constant *getNullPointer(const CodeGen::CodeGenModule &CGM,
                                  llvm::PointerType *T,
                                  QualType QT) const override;
+  void setTargetAttributes(const Decl *D, llvm::GlobalValue *GV,
+                           CodeGen::CodeGenModule &M) const override;
 };
 class SPIRVTargetCodeGenInfo : public CommonSPIRTargetCodeGenInfo {
 public:
@@ -93,6 +95,8 @@ inline StringRef mapClangSyncScopeToLLVM(SyncScope Scope) {
   case SyncScope::OpenCLSubGroup:
   case SyncScope::WavefrontScope:
     return "subgroup";
+  case SyncScope::HIPCluster:
+  case SyncScope::ClusterScope:
   case SyncScope::HIPWorkgroup:
   case SyncScope::OpenCLWorkGroup:
   case SyncScope::WorkgroupScope:
@@ -266,6 +270,22 @@ CommonSPIRTargetCodeGenInfo::getNullPointer(const CodeGen::CodeGenModule &CGM,
       llvm::ConstantPointerNull::get(NPT), PT);
 }
 
+void CommonSPIRTargetCodeGenInfo::setTargetAttributes(
+    const Decl *D, llvm::GlobalValue *GV, CodeGen::CodeGenModule &M) const {
+  if (M.getLangOpts().OpenCL || GV->isDeclaration())
+    return;
+
+  const FunctionDecl *FD = dyn_cast<FunctionDecl>(D);
+  if (!FD)
+    return;
+
+  llvm::Function *F = dyn_cast<llvm::Function>(GV);
+  assert(F && "Expected GlobalValue to be a Function");
+
+  if (FD->hasAttr<DeviceKernelAttr>())
+    F->setCallingConv(getDeviceKernelCallingConv());
+}
+
 LangAS
 SPIRVTargetCodeGenInfo::getGlobalVarAddressSpace(CodeGenModule &CGM,
                                                  const VarDecl *D) const {
@@ -290,19 +310,23 @@ SPIRVTargetCodeGenInfo::getGlobalVarAddressSpace(CodeGenModule &CGM,
 
 void SPIRVTargetCodeGenInfo::setTargetAttributes(
     const Decl *D, llvm::GlobalValue *GV, CodeGen::CodeGenModule &M) const {
-  if (!M.getLangOpts().HIP ||
-      M.getTarget().getTriple().getVendor() != llvm::Triple::AMD)
-    return;
   if (GV->isDeclaration())
     return;
 
-  auto F = dyn_cast<llvm::Function>(GV);
-  if (!F)
+  const FunctionDecl *FD = dyn_cast_or_null<FunctionDecl>(D);
+  if (!FD)
     return;
 
-  auto FD = dyn_cast_or_null<FunctionDecl>(D);
-  if (!FD)
+  llvm::Function *F = dyn_cast<llvm::Function>(GV);
+  assert(F && "Expected GlobalValue to be a Function");
+
+  if (FD->hasAttr<DeviceKernelAttr>())
+    F->setCallingConv(getDeviceKernelCallingConv());
+
+  if (!M.getLangOpts().HIP ||
+      M.getTarget().getTriple().getVendor() != llvm::Triple::AMD)
     return;
+
   if (!FD->hasAttr<CUDAGlobalAttr>())
     return;