4 files changed, 32 insertions, 9 deletions

diff --git a/clang/lib/CodeGen/CodeGenModule.cpp b/clang/lib/CodeGen/CodeGenModule.cpp
index 87a1d45abf4..b0e3b0bb98d 100644
--- a/clang/lib/CodeGen/CodeGenModule.cpp
+++ b/clang/lib/CodeGen/CodeGenModule.cpp
@@ -2415,7 +2415,8 @@ void CodeGenModule::EmitGlobal(GlobalDecl GD) {
       if (!Global->hasAttr<CUDADeviceAttr>() &&
           !Global->hasAttr<CUDAGlobalAttr>() &&
           !Global->hasAttr<CUDAConstantAttr>() &&
-          !Global->hasAttr<CUDASharedAttr>())
+          !Global->hasAttr<CUDASharedAttr>() &&
+          !(LangOpts.HIP && Global->hasAttr<HIPPinnedShadowAttr>()))
         return;
     } else {
       // We need to emit host-side 'shadows' for all global
@@ -3781,7 +3782,12 @@ void CodeGenModule::EmitGlobalVarDefinition(const VarDecl *D,
       !getLangOpts().CUDAIsDevice &&
       (D->hasAttr<CUDAConstantAttr>() || D->hasAttr<CUDADeviceAttr>() ||
        D->hasAttr<CUDASharedAttr>());
-  if (getLangOpts().CUDA && (IsCUDASharedVar || IsCUDAShadowVar))
+  // HIP pinned shadow of initialized host-side global variables are also
+  // left undefined.
+  bool IsHIPPinnedShadowVar =
+      getLangOpts().CUDAIsDevice && D->hasAttr<HIPPinnedShadowAttr>();
+  if (getLangOpts().CUDA &&
+      (IsCUDASharedVar || IsCUDAShadowVar || IsHIPPinnedShadowVar))
     Init = llvm::UndefValue::get(getTypes().ConvertType(ASTTy));
   else if (!InitExpr) {
     // This is a tentative definition; tentative definitions are
@@ -3892,7 +3898,8 @@ void CodeGenModule::EmitGlobalVarDefinition(const VarDecl *D,
       // global variables become internal definitions. These have to
       // be internal in order to prevent name conflicts with global
       // host variables with the same name in a different TUs.
-      if (D->hasAttr<CUDADeviceAttr>() || D->hasAttr<CUDAConstantAttr>()) {
+      if (D->hasAttr<CUDADeviceAttr>() || D->hasAttr<CUDAConstantAttr>() ||
+          D->hasAttr<HIPPinnedShadowAttr>()) {
         Linkage = llvm::GlobalValue::InternalLinkage;
 
         // Shadow variables and their properties must be registered
@@ -3916,7 +3923,8 @@ void CodeGenModule::EmitGlobalVarDefinition(const VarDecl *D,
     }
   }
 
-  GV->setInitializer(Init);
+  if (!IsHIPPinnedShadowVar)
+    GV->setInitializer(Init);
   if (emitter) emitter->finalize(GV);
 
   // If it is safe to mark the global 'constant', do so now.
diff --git a/clang/lib/CodeGen/TargetInfo.cpp b/clang/lib/CodeGen/TargetInfo.cpp
index 8563a18b2ec..22f70a08b0f 100644
--- a/clang/lib/CodeGen/TargetInfo.cpp
+++ b/clang/lib/CodeGen/TargetInfo.cpp
@@ -7874,12 +7874,24 @@ static bool requiresAMDGPUProtectedVisibility(const Decl *D,
   return D->hasAttr<OpenCLKernelAttr>() ||
          (isa<FunctionDecl>(D) && D->hasAttr<CUDAGlobalAttr>()) ||
          (isa<VarDecl>(D) &&
-          (D->hasAttr<CUDADeviceAttr>() || D->hasAttr<CUDAConstantAttr>()));
+          (D->hasAttr<CUDADeviceAttr>() || D->hasAttr<CUDAConstantAttr>() ||
+           D->hasAttr<HIPPinnedShadowAttr>()));
+}
+
+static bool requiresAMDGPUDefaultVisibility(const Decl *D,
+                                            llvm::GlobalValue *GV) {
+  if (GV->getVisibility() != llvm::GlobalValue::HiddenVisibility)
+    return false;
+
+  return isa<VarDecl>(D) && D->hasAttr<HIPPinnedShadowAttr>();
 }
 
 void AMDGPUTargetCodeGenInfo::setTargetAttributes(
     const Decl *D, llvm::GlobalValue *GV, CodeGen::CodeGenModule &M) const {
-  if (requiresAMDGPUProtectedVisibility(D, GV)) {
+  if (requiresAMDGPUDefaultVisibility(D, GV)) {
+    GV->setVisibility(llvm::GlobalValue::DefaultVisibility);
+    GV->setDSOLocal(false);
+  } else if (requiresAMDGPUProtectedVisibility(D, GV)) {
     GV->setVisibility(llvm::GlobalValue::ProtectedVisibility);
     GV->setDSOLocal(true);
   }
diff --git a/clang/lib/Driver/ToolChains/HIP.cpp b/clang/lib/Driver/ToolChains/HIP.cpp
index a60485ab03b..2ec97e798fd 100644
--- a/clang/lib/Driver/ToolChains/HIP.cpp
+++ b/clang/lib/Driver/ToolChains/HIP.cpp
@@ -170,9 +170,8 @@ void AMDGCN::Linker::constructLldCommand(Compilation &C, const JobAction &JA,
                                           const char *InputFileName) const {
   // Construct lld command.
   // The output from ld.lld is an HSA code object file.
-  ArgStringList LldArgs{"-flavor",    "gnu", "--no-undefined",
-                        "-shared",    "-o",  Output.getFilename(),
-                        InputFileName};
+  ArgStringList LldArgs{
+      "-flavor", "gnu", "-shared", "-o", Output.getFilename(), InputFileName};
   SmallString<128> LldPath(C.getDriver().Dir);
   llvm::sys::path::append(LldPath, "lld");
   const char *Lld = Args.MakeArgString(LldPath);
diff --git a/clang/lib/Sema/SemaDeclAttr.cpp b/clang/lib/Sema/SemaDeclAttr.cpp
index 8670f74d956..5a1712ff13a 100644
--- a/clang/lib/Sema/SemaDeclAttr.cpp
+++ b/clang/lib/Sema/SemaDeclAttr.cpp
@@ -6786,6 +6786,10 @@ static void ProcessDeclAttribute(Sema &S, Scope *scope, Decl *D,
   case ParsedAttr::AT_CUDAHost:
     handleSimpleAttributeWithExclusions<CUDAHostAttr, CUDAGlobalAttr>(S, D, AL);
     break;
+  case ParsedAttr::AT_HIPPinnedShadow:
+    handleSimpleAttributeWithExclusions<HIPPinnedShadowAttr, CUDADeviceAttr,
+                                        CUDAConstantAttr>(S, D, AL);
+    break;
   case ParsedAttr::AT_GNUInline:
     handleGNUInlineAttr(S, D, AL);
     break;