[gold] emit assembly listing from gold plugin on LTO stage

[dendibakh]

Summary

Sometimes it's useful to emit assembly after LTO stage to modify it manually. Emitting precodegen bitcode file (via save-temps plugin option) and then feeding it to llc doesn't always give the same binary as original.
This patch is simpler alternative to https://reviews.llvm.org/D24020.

[tejohnson]

Nit on patch description: use [gold] not [LTO] since this is about the gold plugin specifically, and not a change to the LTO API. Also, please add a gold test.

tools/gold/gold-plugin.cpp

1081

Rather than doing this copying, I think you could modify the invocation of getOutputFileName so that it sets Filename to output_name (or maybe output_name + ".s"?) in this case. I.e. see what it does when the output type is OT_SAVE_TEMPS. Then it should emit the assemble to the desired file directly.

[dendibakh]

Updated title, added test and removed copying files.
I decided not to add '.s' to the filename, because then user can also name it as "out.s" and it will be "out.s.s" which seems unnatural.

[tejohnson]

LGTM, with a change to the gold invocations to make the test more portable.

test/tools/gold/X86/emit-asm.ll

3

Please add "-m elf_x86_64" to the gold invocations.

[dendibakh]

Added "-m elf_x86_64" to the gold invocation.

@tejohnson, can you please commit this patch on my behalf?

[tejohnson]

In D56114#1344247, @dendibakh wrote:

Added "-m elf_x86_64" to the gold invocation.

@tejohnson, can you please commit this patch on my behalf?

Sure, will do so shortly.

test/tools/gold/X86/emit-asm.ll

+; RUN: llvm-as %s -o %t.o
+
+; RUN: %gold -plugin %llvmshlibdir/LLVMgold%shlibext \
+; RUN:    --plugin-opt=emit-asm \
+; RUN:    -shared %t.o -o %t2.s
+; RUN: FileCheck --input-file %t2.s %s
+
+; RUN: %gold -plugin %llvmshlibdir/LLVMgold%shlibext \
+; RUN:    --plugin-opt=emit-asm --plugin-opt=lto-partitions=2\
+; RUN:    -shared %t.o -o %t2.s
+; RUN: cat %t2.s %t2.s1 > %t3.s
+; RUN: FileCheck --input-file %t3.s %s
+
+target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
+target triple = "x86_64-unknown-linux-gnu"
+
+; CHECK-DAG: f1:
+define void @f1() {
+  ret void
+}
+
+; CHECK-DAG: f2:
+define void @f2() {
+  ret void
+}

tools/gold/gold-plugin.cpp

 static StringMap<ResolutionInfo> ResInfo;
 static std::vector<std::string> Cleanup;
 
 namespace options {
   enum OutputType {
     OT_NORMAL,
     OT_DISABLE,
     OT_BC_ONLY,
+    OT_ASM_ONLY,
     OT_SAVE_TEMPS
   };
   static OutputType TheOutputType = OT_NORMAL;
   static unsigned OptLevel = 2;
   // Default parallelism of 0 used to indicate that user did not specify.
   // Actual parallelism default value depends on implementation.
   // Currently only affects ThinLTO, where the default is
   // llvm::heavyweight_hardware_concurrency.
static void process_plugin_option(const char *opt_)
     } else if (opt.startswith("obj-path=")) {
       obj_path = opt.substr(strlen("obj-path="));
     } else if (opt == "emit-llvm") {
       TheOutputType = OT_BC_ONLY;
     } else if (opt == "save-temps") {
       TheOutputType = OT_SAVE_TEMPS;
     } else if (opt == "disable-output") {
       TheOutputType = OT_DISABLE;
+    } else if (opt == "emit-asm") {
+      TheOutputType = OT_ASM_ONLY;
     } else if (opt == "thinlto") {
       thinlto = true;
     } else if (opt == "thinlto-index-only") {
       thinlto_index_only = true;
     } else if (opt.startswith("thinlto-index-only=")) {
       thinlto_index_only = true;
       thinlto_linked_objects_file = opt.substr(strlen("thinlto-index-only="));
     } else if (opt == "thinlto-emit-imports-files") {
Conf.PostInternalizeModuleHook = [](size_t Task, const Module &M) {
       return false;
     };
     break;
 
   case options::OT_SAVE_TEMPS:
     check(Conf.addSaveTemps(output_name + ".",
                             /* UseInputModulePath */ true));
     break;
+  case options::OT_ASM_ONLY:
+    Conf.CGFileType = TargetMachine::CGFT_AssemblyFile;
+    break;
   }
 
   if (!options::sample_profile.empty())
     Conf.SampleProfile = options::sample_profile;
 
   Conf.DwoDir = options::dwo_dir;
 
   // Set up optimization remarks handling.
static std::vector<std::pair<SmallString<128>, bool>> runLTO() {
   }
 
   SmallString<128> Filename;
   // Note that getOutputFileName will append a unique ID for each task
   if (!options::obj_path.empty())
     Filename = options::obj_path;
   else if (options::TheOutputType == options::OT_SAVE_TEMPS)
     Filename = output_name + ".o";
+  else if (options::TheOutputType == options::OT_ASM_ONLY)
+    Filename = output_name;
   bool SaveTemps = !Filename.empty();
 
   size_t MaxTasks = Lto->getMaxTasks();
   std::vector<std::pair<SmallString<128>, bool>> Files(MaxTasks);
 
   auto AddStream =
       [&](size_t Task) -> std::unique_ptr<lto::NativeObjectStream> {
     Files[Task].second = !SaveTemps;
if (Modules.empty())
     return LDPS_OK;
 
   if (unsigned NumOpts = options::extra.size())
     cl::ParseCommandLineOptions(NumOpts, &options::extra[0]);
 
   std::vector<std::pair<SmallString<128>, bool>> Files = runLTO();
 
   if (options::TheOutputType == options::OT_DISABLE ||
-      options::TheOutputType == options::OT_BC_ONLY)
+      options::TheOutputType == options::OT_BC_ONLY ||
+      options::TheOutputType == options::OT_ASM_ONLY)
     return LDPS_OK;
 
   if (options::thinlto_index_only) {
     llvm_shutdown();
     cleanup_hook();
     exit(0);
   }
 
   for (const auto &F : Files)
     if (!F.first.empty())
       recordFile(F.first.str(), F.second);
 
   if (!options::extra_library_path.empty() &&
       set_extra_library_path(options::extra_library_path.c_str()) != LDPS_OK)
     message(LDPL_FATAL, "Unable to set the extra library path.");
 
   return LDPS_OK;
 }
 
 static ld_plugin_status all_symbols_read_hook(void) {
   ld_plugin_status Ret = allSymbolsReadHook();
   llvm_shutdown();
 
   if (options::TheOutputType == options::OT_BC_ONLY ||
+      options::TheOutputType == options::OT_ASM_ONLY ||
       options::TheOutputType == options::OT_DISABLE) {
     if (options::TheOutputType == options::OT_DISABLE) {
       // Remove the output file here since ld.bfd creates the output file
       // early.
       std::error_code EC = sys::fs::remove(output_name);
       if (EC)
         message(LDPL_ERROR, "Failed to delete '%s': %s", output_name.c_str(),
                 EC.message().c_str());