1 files changed, 54 insertions, 5 deletions

diff --git a/lib/Driver/ToolChains/HIP.cpp b/lib/Driver/ToolChains/HIP.cpp
index 03acf45a9b313..868765cf88e5a 100644
--- a/lib/Driver/ToolChains/HIP.cpp
+++ b/lib/Driver/ToolChains/HIP.cpp
@@ -24,6 +24,12 @@ using namespace clang::driver::tools;
 using namespace clang;
 using namespace llvm::opt;
 
+#if _WIN32 || _WIN64
+#define NULL_FILE "nul"
+#else
+#define NULL_FILE "/dev/null"
+#endif
+
 namespace {
 
 static void addBCLib(Compilation &C, const ArgList &Args,
@@ -81,8 +87,8 @@ const char *AMDGCN::Linker::constructLLVMLinkCommand(
     else
       FlushDenormalControlBC = "oclc_daz_opt_off.amdgcn.bc";
 
-    BCLibs.append({"opencl.amdgcn.bc",
-                   "ocml.amdgcn.bc", "ockl.amdgcn.bc", "irif.amdgcn.bc",
+    BCLibs.append({"hip.amdgcn.bc", "opencl.amdgcn.bc",
+                   "ocml.amdgcn.bc", "ockl.amdgcn.bc",
                    "oclc_finite_only_off.amdgcn.bc",
                    FlushDenormalControlBC,
                    "oclc_correctly_rounded_sqrt_on.amdgcn.bc",
@@ -154,7 +160,7 @@ const char *AMDGCN::Linker::constructLlcCommand(
     llvm::StringRef OutputFilePrefix, const char *InputFileName) const {
   // Construct llc command.
   ArgStringList LlcArgs{InputFileName, "-mtriple=amdgcn-amd-amdhsa",
-                        "-filetype=obj",
+                        "-filetype=obj", "-mattr=-code-object-v3",
                         Args.MakeArgString("-mcpu=" + SubArchName), "-o"};
   std::string LlcOutputFileName =
       C.getDriver().GetTemporaryPath(OutputFilePrefix, "o");
@@ -184,6 +190,40 @@ void AMDGCN::Linker::constructLldCommand(Compilation &C, const JobAction &JA,
   C.addCommand(llvm::make_unique<Command>(JA, *this, Lld, LldArgs, Inputs));
 }
 
+// Construct a clang-offload-bundler command to bundle code objects for
+// different GPU's into a HIP fat binary.
+void AMDGCN::constructHIPFatbinCommand(Compilation &C, const JobAction &JA,
+                  StringRef OutputFileName, const InputInfoList &Inputs,
+                  const llvm::opt::ArgList &Args, const Tool& T) {
+  // Construct clang-offload-bundler command to bundle object files for
+  // for different GPU archs.
+  ArgStringList BundlerArgs;
+  BundlerArgs.push_back(Args.MakeArgString("-type=o"));
+
+  // ToDo: Remove the dummy host binary entry which is required by
+  // clang-offload-bundler.
+  std::string BundlerTargetArg = "-targets=host-x86_64-unknown-linux";
+  std::string BundlerInputArg = "-inputs=" NULL_FILE;
+
+  for (const auto &II : Inputs) {
+    const auto* A = II.getAction();
+    BundlerTargetArg = BundlerTargetArg + ",hip-amdgcn-amd-amdhsa-" +
+                       StringRef(A->getOffloadingArch()).str();
+    BundlerInputArg = BundlerInputArg + "," + II.getFilename();
+  }
+  BundlerArgs.push_back(Args.MakeArgString(BundlerTargetArg));
+  BundlerArgs.push_back(Args.MakeArgString(BundlerInputArg));
+
+  auto BundlerOutputArg =
+      Args.MakeArgString(std::string("-outputs=").append(OutputFileName));
+  BundlerArgs.push_back(BundlerOutputArg);
+
+  SmallString<128> BundlerPath(C.getDriver().Dir);
+  llvm::sys::path::append(BundlerPath, "clang-offload-bundler");
+  const char *Bundler = Args.MakeArgString(BundlerPath);
+  C.addCommand(llvm::make_unique<Command>(JA, T, Bundler, BundlerArgs, Inputs));
+}
+
 // For amdgcn the inputs of the linker job are device bitcode and output is
 // object file. It calls llvm-link, opt, llc, then lld steps.
 void AMDGCN::Linker::ConstructJob(Compilation &C, const JobAction &JA,
@@ -192,6 +232,9 @@ void AMDGCN::Linker::ConstructJob(Compilation &C, const JobAction &JA,
                                    const ArgList &Args,
                                    const char *LinkingOutput) const {
 
+  if (JA.getType() == types::TY_HIP_FATBIN)
+    return constructHIPFatbinCommand(C, JA, Output.getFilename(), Inputs, Args, *this);
+
   assert(getToolChain().getTriple().getArch() == llvm::Triple::amdgcn &&
          "Unsupported target");
 
@@ -244,9 +287,15 @@ void HIPToolChain::addClangTargetOptions(
                          options::OPT_fno_cuda_approx_transcendentals, false))
     CC1Args.push_back("-fcuda-approx-transcendentals");
 
-  if (DriverArgs.hasFlag(options::OPT_fcuda_rdc, options::OPT_fno_cuda_rdc,
+  if (DriverArgs.hasFlag(options::OPT_fgpu_rdc, options::OPT_fno_gpu_rdc,
                          false))
-    CC1Args.push_back("-fcuda-rdc");
+    CC1Args.push_back("-fgpu-rdc");
+
+  // Default to "hidden" visibility, as object level linking will not be
+  // supported for the foreseeable future.
+  if (!DriverArgs.hasArg(options::OPT_fvisibility_EQ,
+                         options::OPT_fvisibility_ms_compat))
+    CC1Args.append({"-fvisibility", "hidden"});
 }
 
 llvm::opt::DerivedArgList *