5 files changed, 22 insertions, 27 deletions
diff --git a/llvm/tools/llvm-gpu-loader/amdhsa.cpp b/llvm/tools/llvm-gpu-loader/amdhsa.cpp
index be1b6b7..5715058 100644
--- a/llvm/tools/llvm-gpu-loader/amdhsa.cpp
+++ b/llvm/tools/llvm-gpu-loader/amdhsa.cpp
@@ -192,7 +192,7 @@ hsa_status_t launch_kernel(hsa_agent_t dev_agent, hsa_executable_t executable,
   // Initialize all the arguments (explicit and implicit) to zero, then set the
   // explicit arguments to the values created above.
   std::memset(args, 0, args_size);
-  std::memcpy(args, &kernel_args, sizeof(args_t));
+  std::memcpy(args, &kernel_args, std::is_empty_v<args_t> ? 0 : sizeof(args_t));
 
   // Initialize the necessary implicit arguments to the proper values.
   int dims = 1 + (params.num_blocks_y * params.num_threads_y != 1) +
@@ -563,7 +563,7 @@ int load_amdhsa(int argc, const char **argv, const char **envp, void *image,
   // Save the return value and perform basic clean-up.
   int ret = *static_cast<int *>(host_ret);
 
-  end_args_t fini_args = {ret};
+  end_args_t fini_args = {};
   if (hsa_status_t err = launch_kernel(
           dev_agent, executable, kernargs_pool, coarsegrained_pool, queue,
           server, single_threaded_params, "_end.kd", fini_args,
diff --git a/llvm/tools/llvm-gpu-loader/llvm-gpu-loader.h b/llvm/tools/llvm-gpu-loader/llvm-gpu-loader.h
index ed34d0b..08861c2 100644
--- a/llvm/tools/llvm-gpu-loader/llvm-gpu-loader.h
+++ b/llvm/tools/llvm-gpu-loader/llvm-gpu-loader.h
@@ -41,9 +41,7 @@ struct start_args_t {
 };
 
 /// The arguments to the '_end' kernel.
-struct end_args_t {
-  int argc;
-};
+struct end_args_t {};
 
 /// Generic interface to load the \p image and launch execution of the _start
 /// kernel on the target device. Copies \p argc and \p argv to the device.
diff --git a/llvm/tools/llvm-gpu-loader/nvptx.cpp b/llvm/tools/llvm-gpu-loader/nvptx.cpp
index 781a045..82b4552 100644
--- a/llvm/tools/llvm-gpu-loader/nvptx.cpp
+++ b/llvm/tools/llvm-gpu-loader/nvptx.cpp
@@ -177,7 +177,7 @@ CUresult launch_kernel(CUmodule binary, CUstream stream, rpc::Server &server,
     handle_error(err);
 
   // Set up the arguments to the '_start' kernel on the GPU.
-  uint64_t args_size = sizeof(args_t);
+  uint64_t args_size = std::is_empty_v<args_t> ? 0 : sizeof(args_t);
   void *args_config[] = {CU_LAUNCH_PARAM_BUFFER_POINTER, &kernel_args,
                          CU_LAUNCH_PARAM_BUFFER_SIZE, &args_size,
                          CU_LAUNCH_PARAM_END};
@@ -342,7 +342,7 @@ int load_nvptx(int argc, const char **argv, const char **envp, void *image,
   if (CUresult err = cuStreamSynchronize(stream))
     handle_error(err);
 
-  end_args_t fini_args = {host_ret};
+  end_args_t fini_args = {};
   if (CUresult err =
           launch_kernel(binary, stream, server, single_threaded_params, "_end",
                         fini_args, print_resource_usage))
diff --git a/llvm/tools/llvm-profgen/ProfiledBinary.cpp b/llvm/tools/llvm-profgen/ProfiledBinary.cpp
index 6865e36..94728ce 100644
--- a/llvm/tools/llvm-profgen/ProfiledBinary.cpp
+++ b/llvm/tools/llvm-profgen/ProfiledBinary.cpp
@@ -250,14 +250,12 @@ void ProfiledBinary::load() {
 
   DisassembleFunctionSet.insert_range(DisassembleFunctions);
 
-  if (auto *ELFObj = dyn_cast<ELFObjectFileBase>(Obj)) {
-    checkPseudoProbe(ELFObj);
-    if (UsePseudoProbes)
-      populateElfSymbolAddressList(ELFObj);
+  checkPseudoProbe(Obj);
+  if (UsePseudoProbes)
+    populateSymbolAddressList(Obj);
 
-    if (ShowDisassemblyOnly)
-      decodePseudoProbe(ELFObj);
-  }
+  if (ShowDisassemblyOnly)
+    decodePseudoProbe(Obj);
 
   // Disassemble the text sections.
   disassemble(Obj);
@@ -417,7 +415,7 @@ void ProfiledBinary::setPreferredTextSegmentAddresses(const ObjectFile *Obj) {
     llvm_unreachable("invalid object format");
 }
 
-void ProfiledBinary::checkPseudoProbe(const ELFObjectFileBase *Obj) {
+void ProfiledBinary::checkPseudoProbe(const ObjectFile *Obj) {
   if (UseDwarfCorrelation)
     return;
 
@@ -440,7 +438,7 @@ void ProfiledBinary::checkPseudoProbe(const ELFObjectFileBase *Obj) {
   UsePseudoProbes = HasProbeDescSection && HasPseudoProbeSection;
 }
 
-void ProfiledBinary::decodePseudoProbe(const ELFObjectFileBase *Obj) {
+void ProfiledBinary::decodePseudoProbe(const ObjectFile *Obj) {
   if (!UsePseudoProbes)
     return;
 
@@ -511,7 +509,7 @@ void ProfiledBinary::decodePseudoProbe(const ELFObjectFileBase *Obj) {
 void ProfiledBinary::decodePseudoProbe() {
   OwningBinary<Binary> OBinary = unwrapOrError(createBinary(Path), Path);
   Binary &ExeBinary = *OBinary.getBinary();
-  auto *Obj = cast<ELFObjectFileBase>(&ExeBinary);
+  auto *Obj = cast<ObjectFile>(&ExeBinary);
   decodePseudoProbe(Obj);
 }
 
@@ -809,8 +807,7 @@ void ProfiledBinary::checkUseFSDiscriminator(
   }
 }
 
-void ProfiledBinary::populateElfSymbolAddressList(
-    const ELFObjectFileBase *Obj) {
+void ProfiledBinary::populateSymbolAddressList(const ObjectFile *Obj) {
   // Create a mapping from virtual address to symbol GUID and the other way
   // around.
   StringRef FileName = Obj->getFileName();
diff --git a/llvm/tools/llvm-profgen/ProfiledBinary.h b/llvm/tools/llvm-profgen/ProfiledBinary.h
index e82fbab..5a814b7 100644
--- a/llvm/tools/llvm-profgen/ProfiledBinary.h
+++ b/llvm/tools/llvm-profgen/ProfiledBinary.h
@@ -228,19 +228,19 @@ class ProfiledBinary {
   // A list of binary functions that have samples.
   std::unordered_set<const BinaryFunction *> ProfiledFunctions;
 
-  // GUID to Elf symbol start address map
+  // GUID to symbol start address map
   DenseMap<uint64_t, uint64_t> SymbolStartAddrs;
 
   // These maps are for temporary use of warning diagnosis.
   DenseSet<int64_t> AddrsWithMultipleSymbols;
   DenseSet<std::pair<uint64_t, uint64_t>> AddrsWithInvalidInstruction;
 
-  // Start address to Elf symbol GUID map
+  // Start address to symbol GUID map
   std::unordered_multimap<uint64_t, uint64_t> StartAddrToSymMap;
 
   // An ordered map of mapping function's start address to function range
-  // relevant info. Currently to determine if the offset of ELF is the start of
-  // a real function, we leverage the function range info from DWARF.
+  // relevant info. Currently to determine if the offset of ELF/COFF is the
+  // start of a real function, we leverage the function range info from DWARF.
   std::map<uint64_t, FuncRange> StartAddrToFuncRangeMap;
 
   // Address to context location map. Used to expand the context.
@@ -335,9 +335,9 @@ class ProfiledBinary {
   void setPreferredTextSegmentAddresses(const object::COFFObjectFile *Obj,
                                         StringRef FileName);
 
-  void checkPseudoProbe(const object::ELFObjectFileBase *Obj);
+  void checkPseudoProbe(const object::ObjectFile *Obj);
 
-  void decodePseudoProbe(const object::ELFObjectFileBase *Obj);
+  void decodePseudoProbe(const object::ObjectFile *Obj);
 
   void checkUseFSDiscriminator(
       const object::ObjectFile *Obj,
@@ -353,8 +353,8 @@ class ProfiledBinary {
   // Load debug info from DWARF unit.
   void loadSymbolsFromDWARFUnit(DWARFUnit &CompilationUnit);
 
-  // Create elf symbol to its start address mapping.
-  void populateElfSymbolAddressList(const object::ELFObjectFileBase *O);
+  // Create symbol to its start address mapping.
+  void populateSymbolAddressList(const object::ObjectFile *O);
 
   // A function may be spilt into multiple non-continuous address ranges. We use
   // this to set whether start a function range is the real entry of the