[llvm] A Unified LTO Bitcode Frontend

Here's a high level summary of the changes in this patch. For more
information on rational, see the RFC.
(https://discourse.llvm.org/t/rfc-a-unified-lto-bitcode-frontend/61774).

  - Add config parameter to LTO backend, specifying which LTO mode is
    desired when using unified LTO.
  - Add unified LTO flag to the summary index for efficiency. Unified
    LTO modules can be detected without parsing the module.
  - Make sure that the ModuleID is generated by incorporating more types
    of symbols.

Differential Revision: https://reviews.llvm.org/D123803

9 files changed, 118 insertions, 33 deletions

diff --git a/llvm/lib/Analysis/ModuleSummaryAnalysis.cpp b/llvm/lib/Analysis/ModuleSummaryAnalysis.cpp
index fd125ac..165b8f1 100644
--- a/llvm/lib/Analysis/ModuleSummaryAnalysis.cpp
+++ b/llvm/lib/Analysis/ModuleSummaryAnalysis.cpp
@@ -786,10 +786,14 @@ ModuleSummaryIndex llvm::buildModuleSummaryIndex(
     std::function<const StackSafetyInfo *(const Function &F)> GetSSICallback) {
   assert(PSI);
   bool EnableSplitLTOUnit = false;
+  bool UnifiedLTO = false;
   if (auto *MD = mdconst::extract_or_null<ConstantInt>(
           M.getModuleFlag("EnableSplitLTOUnit")))
     EnableSplitLTOUnit = MD->getZExtValue();
-  ModuleSummaryIndex Index(/*HaveGVs=*/true, EnableSplitLTOUnit);
+  if (auto *MD =
+          mdconst::extract_or_null<ConstantInt>(M.getModuleFlag("UnifiedLTO")))
+    UnifiedLTO = MD->getZExtValue();
+  ModuleSummaryIndex Index(/*HaveGVs=*/true, EnableSplitLTOUnit, UnifiedLTO);
 
   // Identify the local values in the llvm.used and llvm.compiler.used sets,
   // which should not be exported as they would then require renaming and
diff --git a/llvm/lib/Bitcode/Reader/BitcodeReader.cpp b/llvm/lib/Bitcode/Reader/BitcodeReader.cpp
index ebc88c8..4f6c5f0 100644
--- a/llvm/lib/Bitcode/Reader/BitcodeReader.cpp
+++ b/llvm/lib/Bitcode/Reader/BitcodeReader.cpp
@@ -8033,14 +8033,17 @@ Expected<std::unique_ptr<ModuleSummaryIndex>> BitcodeModule::getSummary() {
   return std::move(Index);
 }
 
-static Expected<bool> getEnableSplitLTOUnitFlag(BitstreamCursor &Stream,
-                                                unsigned ID) {
+static Expected<std::pair<bool, bool>>
+getEnableSplitLTOUnitAndUnifiedFlag(BitstreamCursor &Stream,
+                                                 unsigned ID,
+                                                 BitcodeLTOInfo &LTOInfo) {
   if (Error Err = Stream.EnterSubBlock(ID))
     return std::move(Err);
   SmallVector<uint64_t, 64> Record;
 
   while (true) {
     BitstreamEntry Entry;
+    std::pair<bool, bool> Result = {false,false};
     if (Error E = Stream.advanceSkippingSubblocks().moveInto(Entry))
       return std::move(E);
 
@@ -8048,10 +8051,10 @@ static Expected<bool> getEnableSplitLTOUnitFlag(BitstreamCursor &Stream,
     case BitstreamEntry::SubBlock: // Handled for us already.
     case BitstreamEntry::Error:
       return error("Malformed block");
-    case BitstreamEntry::EndBlock:
-      // If no flags record found, conservatively return true to mimic
-      // behavior before this flag was added.
-      return true;
+    case BitstreamEntry::EndBlock: {
+      // If no flags record found, set both flags to false.
+      return Result;
+    }
     case BitstreamEntry::Record:
       // The interesting case.
       break;
@@ -8068,9 +8071,13 @@ static Expected<bool> getEnableSplitLTOUnitFlag(BitstreamCursor &Stream,
     case bitc::FS_FLAGS: { // [flags]
       uint64_t Flags = Record[0];
       // Scan flags.
-      assert(Flags <= 0x1ff && "Unexpected bits in flag");
+      assert(Flags <= 0x2ff && "Unexpected bits in flag");
+
+      bool EnableSplitLTOUnit = Flags & 0x8;
+      bool UnifiedLTO = Flags & 0x200;
+      Result = {EnableSplitLTOUnit, UnifiedLTO};
 
-      return Flags & 0x8;
+      return Result;
     }
     }
   }
@@ -8096,25 +8103,31 @@ Expected<BitcodeLTOInfo> BitcodeModule::getLTOInfo() {
       return error("Malformed block");
     case BitstreamEntry::EndBlock:
       return BitcodeLTOInfo{/*IsThinLTO=*/false, /*HasSummary=*/false,
-                            /*EnableSplitLTOUnit=*/false};
+                            /*EnableSplitLTOUnit=*/false, /*UnifiedLTO=*/false};
 
     case BitstreamEntry::SubBlock:
       if (Entry.ID == bitc::GLOBALVAL_SUMMARY_BLOCK_ID) {
-        Expected<bool> EnableSplitLTOUnit =
-            getEnableSplitLTOUnitFlag(Stream, Entry.ID);
-        if (!EnableSplitLTOUnit)
-          return EnableSplitLTOUnit.takeError();
-        return BitcodeLTOInfo{/*IsThinLTO=*/true, /*HasSummary=*/true,
-                              *EnableSplitLTOUnit};
+        BitcodeLTOInfo LTOInfo;
+        Expected<std::pair<bool, bool>> Flags =
+            getEnableSplitLTOUnitAndUnifiedFlag(Stream, Entry.ID, LTOInfo);
+        if (!Flags)
+          return Flags.takeError();
+        std::tie(LTOInfo.EnableSplitLTOUnit, LTOInfo.UnifiedLTO) = Flags.get();
+        LTOInfo.IsThinLTO = true;
+        LTOInfo.HasSummary = true;
+        return LTOInfo;
       }
 
       if (Entry.ID == bitc::FULL_LTO_GLOBALVAL_SUMMARY_BLOCK_ID) {
-        Expected<bool> EnableSplitLTOUnit =
-            getEnableSplitLTOUnitFlag(Stream, Entry.ID);
-        if (!EnableSplitLTOUnit)
-          return EnableSplitLTOUnit.takeError();
-        return BitcodeLTOInfo{/*IsThinLTO=*/false, /*HasSummary=*/true,
-                              *EnableSplitLTOUnit};
+        BitcodeLTOInfo LTOInfo;
+        Expected<std::pair<bool, bool>> Flags =
+            getEnableSplitLTOUnitAndUnifiedFlag(Stream, Entry.ID, LTOInfo);
+        if (!Flags)
+          return Flags.takeError();
+        std::tie(LTOInfo.EnableSplitLTOUnit, LTOInfo.UnifiedLTO) = Flags.get();
+        LTOInfo.IsThinLTO = false;
+        LTOInfo.HasSummary = true;
+        return LTOInfo;
       }
 
       // Ignore other sub-blocks.
diff --git a/llvm/lib/Bitcode/Writer/BitcodeWriter.cpp b/llvm/lib/Bitcode/Writer/BitcodeWriter.cpp
index bedae66..f2bfb4e 100644
--- a/llvm/lib/Bitcode/Writer/BitcodeWriter.cpp
+++ b/llvm/lib/Bitcode/Writer/BitcodeWriter.cpp
@@ -4086,6 +4086,9 @@ void ModuleBitcodeWriterBase::writePerModuleGlobalValueSummary() {
   // Bits 1-3 are set only in the combined index, skip them.
   if (Index->enableSplitLTOUnit())
     Flags |= 0x8;
+  if (Index->hasUnifiedLTO())
+    Flags |= 0x200;
+
   Stream.EmitRecord(bitc::FS_FLAGS, ArrayRef<uint64_t>{Flags});
 
   if (Index->begin() == Index->end()) {
@@ -4112,7 +4115,7 @@ void ModuleBitcodeWriterBase::writePerModuleGlobalValueSummary() {
   auto Abbv = std::make_shared<BitCodeAbbrev>();
   Abbv->Add(BitCodeAbbrevOp(bitc::FS_PERMODULE_PROFILE));
   Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::VBR, 8));   // valueid
-  Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::VBR, 6));   // flags
+  Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::VBR, 8));   // flags
   Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::VBR, 8));   // instcount
   Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::VBR, 4));   // fflags
   Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::VBR, 4));   // numrefs
diff --git a/llvm/lib/IR/ModuleSummaryIndex.cpp b/llvm/lib/IR/ModuleSummaryIndex.cpp
index 7cba03e..15fe342 100644
--- a/llvm/lib/IR/ModuleSummaryIndex.cpp
+++ b/llvm/lib/IR/ModuleSummaryIndex.cpp
@@ -109,11 +109,13 @@ uint64_t ModuleSummaryIndex::getFlags() const {
     Flags |= 0x80;
   if (withSupportsHotColdNew())
     Flags |= 0x100;
+  if (hasUnifiedLTO())
+    Flags |= 0x200;
   return Flags;
 }
 
 void ModuleSummaryIndex::setFlags(uint64_t Flags) {
-  assert(Flags <= 0x1ff && "Unexpected bits in flag");
+  assert(Flags <= 0x2ff && "Unexpected bits in flag");
   // 1 bit: WithGlobalValueDeadStripping flag.
   // Set on combined index only.
   if (Flags & 0x1)
@@ -151,6 +153,10 @@ void ModuleSummaryIndex::setFlags(uint64_t Flags) {
   // Set on combined index only.
   if (Flags & 0x100)
     setWithSupportsHotColdNew();
+  // 1 bit: WithUnifiedLTO flag.
+  // Set on combined index only.
+  if (Flags & 0x200)
+    setUnifiedLTO();
 }
 
 // Collect for the given module the list of function it defines
diff --git a/llvm/lib/LTO/LTO.cpp b/llvm/lib/LTO/LTO.cpp
index 65ba3cb..6803d6a 100644
--- a/llvm/lib/LTO/LTO.cpp
+++ b/llvm/lib/LTO/LTO.cpp
@@ -603,10 +603,10 @@ LTO::ThinLTOState::ThinLTOState(ThinBackend Backend)
 }
 
 LTO::LTO(Config Conf, ThinBackend Backend,
-         unsigned ParallelCodeGenParallelismLevel)
+         unsigned ParallelCodeGenParallelismLevel, LTOKind LTOMode)
     : Conf(std::move(Conf)),
       RegularLTO(ParallelCodeGenParallelismLevel, this->Conf),
-      ThinLTO(std::move(Backend)) {}
+      ThinLTO(std::move(Backend)), LTOMode(LTOMode) {}
 
 // Requires a destructor for MapVector<BitcodeModule>.
 LTO::~LTO() = default;
@@ -747,12 +747,25 @@ Error LTO::addModule(InputFile &Input, unsigned ModI,
     EnableSplitLTOUnit = LTOInfo->EnableSplitLTOUnit;
 
   BitcodeModule BM = Input.Mods[ModI];
+
+  if ((LTOMode == LTOK_UnifiedRegular || LTOMode == LTOK_UnifiedThin) &&
+      !LTOInfo->UnifiedLTO)
+    return make_error<StringError>(
+        "unified LTO compilation must use "
+        "compatible bitcode modules (use -funified-lto)",
+        inconvertibleErrorCode());
+
+  if (LTOInfo->UnifiedLTO && LTOMode == LTOK_Default)
+    LTOMode = LTOK_UnifiedThin;
+
+  bool IsThinLTO = LTOInfo->IsThinLTO && (LTOMode != LTOK_UnifiedRegular);
+
   auto ModSyms = Input.module_symbols(ModI);
   addModuleToGlobalRes(ModSyms, {ResI, ResE},
-                       LTOInfo->IsThinLTO ? ThinLTO.ModuleMap.size() + 1 : 0,
+                       IsThinLTO ? ThinLTO.ModuleMap.size() + 1 : 0,
                        LTOInfo->HasSummary);
 
-  if (LTOInfo->IsThinLTO)
+  if (IsThinLTO)
     return addThinLTO(BM, ModSyms, ResI, ResE);
 
   RegularLTO.EmptyCombinedModule = false;
@@ -820,6 +833,15 @@ LTO::addRegularLTO(BitcodeModule BM, ArrayRef<InputFile::Symbol> Syms,
 
   if (Error Err = M.materializeMetadata())
     return std::move(Err);
+
+  // If cfi.functions is present and we are in regular LTO mode, LowerTypeTests
+  // will rename local functions in the merged module as "<function name>.1".
+  // This causes linking errors, since other parts of the module expect the
+  // original function name.
+  if (LTOMode == LTOK_UnifiedRegular)
+    if (NamedMDNode *CfiFunctionsMD = M.getNamedMetadata("cfi.functions"))
+      M.eraseNamedMetadata(CfiFunctionsMD);
+
   UpgradeDebugInfo(M);
 
   ModuleSymbolTable SymTab;
@@ -1214,6 +1236,7 @@ Error LTO::runRegularLTO(AddStreamFn AddStream) {
       RegularLTO.CombinedModule->getContext(), Conf.RemarksFilename,
       Conf.RemarksPasses, Conf.RemarksFormat, Conf.RemarksWithHotness,
       Conf.RemarksHotnessThreshold);
+  LLVM_DEBUG(dbgs() << "Running regular LTO\n");
   if (!DiagFileOrErr)
     return DiagFileOrErr.takeError();
   DiagnosticOutputFile = std::move(*DiagFileOrErr);
@@ -1277,18 +1300,33 @@ Error LTO::runRegularLTO(AddStreamFn AddStream) {
 
   if (!Conf.CodeGenOnly) {
     for (const auto &R : GlobalResolutions) {
+      GlobalValue *GV =
+          RegularLTO.CombinedModule->getNamedValue(R.second.IRName);
       if (!R.second.isPrevailingIRSymbol())
         continue;
       if (R.second.Partition != 0 &&
           R.second.Partition != GlobalResolution::External)
         continue;
 
-      GlobalValue *GV =
-          RegularLTO.CombinedModule->getNamedValue(R.second.IRName);
       // Ignore symbols defined in other partitions.
       // Also skip declarations, which are not allowed to have internal linkage.
       if (!GV || GV->hasLocalLinkage() || GV->isDeclaration())
         continue;
+
+      // Symbols that are marked DLLImport or DLLExport should not be
+      // internalized, as they are either externally visible or referencing
+      // external symbols. Symbols that have AvailableExternally or Appending
+      // linkage might be used by future passes and should be kept as is.
+      // These linkages are seen in Unified regular LTO, because the process
+      // of creating split LTO units introduces symbols with that linkage into
+      // one of the created modules. Normally, only the ThinLTO backend would
+      // compile this module, but Unified Regular LTO processes both
+      // modules created by the splitting process as regular LTO modules.
+      if ((LTOMode == LTOKind::LTOK_UnifiedRegular) &&
+          ((GV->getDLLStorageClass() != GlobalValue::DefaultStorageClass) ||
+           GV->hasAvailableExternallyLinkage() || GV->hasAppendingLinkage()))
+        continue;
+
       GV->setUnnamedAddr(R.second.UnnamedAddr ? GlobalValue::UnnamedAddr::Global
                                               : GlobalValue::UnnamedAddr::None);
       if (EnableLTOInternalization && R.second.Partition == 0)
@@ -1606,6 +1644,7 @@ ThinBackend lto::createWriteIndexesThinBackend(
 
 Error LTO::runThinLTO(AddStreamFn AddStream, FileCache Cache,
                       const DenseSet<GlobalValue::GUID> &GUIDPreservedSymbols) {
+  LLVM_DEBUG(dbgs() << "Running ThinLTO\n");
   ThinLTO.CombinedIndex.releaseTemporaryMemory();
   timeTraceProfilerBegin("ThinLink", StringRef(""));
   auto TimeTraceScopeExit = llvm::make_scope_exit([]() {
diff --git a/llvm/lib/LTO/LTOBackend.cpp b/llvm/lib/LTO/LTOBackend.cpp
index 6eee789..897380c 100644
--- a/llvm/lib/LTO/LTOBackend.cpp
+++ b/llvm/lib/LTO/LTOBackend.cpp
@@ -504,6 +504,7 @@ Error lto::backend(const Config &C, AddStreamFn AddStream,
 
   std::unique_ptr<TargetMachine> TM = createTargetMachine(C, *TOrErr, Mod);
 
+  LLVM_DEBUG(dbgs() << "Running regular LTO\n");
   if (!C.CodeGenOnly) {
     if (!opt(C, TM.get(), 0, Mod, /*IsThinLTO=*/false,
              /*ExportSummary=*/&CombinedIndex, /*ImportSummary=*/nullptr,
@@ -566,6 +567,7 @@ Error lto::thinBackend(const Config &Conf, unsigned Task, AddStreamFn AddStream,
   // the module, if applicable.
   Mod.setPartialSampleProfileRatio(CombinedIndex);
 
+  LLVM_DEBUG(dbgs() << "Running ThinLTO\n");
   if (Conf.CodeGenOnly) {
     codegen(Conf, TM.get(), AddStream, Task, Mod, CombinedIndex);
     return finalizeOptimizationRemarks(std::move(DiagnosticOutputFile));
diff --git a/llvm/lib/Passes/PassBuilder.cpp b/llvm/lib/Passes/PassBuilder.cpp
index bbab55c..d86ec21 100644
--- a/llvm/lib/Passes/PassBuilder.cpp
+++ b/llvm/lib/Passes/PassBuilder.cpp
@@ -1379,7 +1379,13 @@ Error PassBuilder::parseModulePass(ModulePassManager &MPM,
     } else if (Matches[1] == "thinlto") {
       MPM.addPass(buildThinLTODefaultPipeline(L, nullptr));
     } else if (Matches[1] == "lto-pre-link") {
-      MPM.addPass(buildLTOPreLinkDefaultPipeline(L));
+      if (PTO.UnifiedLTO)
+        // When UnifiedLTO is enabled, use the ThinLTO pre-link pipeline. This
+        // avoids compile-time performance regressions and keeps the pre-link
+        // LTO pipeline "unified" for both LTO modes.
+        MPM.addPass(buildThinLTOPreLinkDefaultPipeline(L));
+      else
+        MPM.addPass(buildLTOPreLinkDefaultPipeline(L));
     } else {
       assert(Matches[1] == "lto" && "Not one of the matched options!");
       MPM.addPass(buildLTODefaultPipeline(L, nullptr));
diff --git a/llvm/lib/Passes/PassBuilderPipelines.cpp b/llvm/lib/Passes/PassBuilderPipelines.cpp
index 58a68c0..d3e625e 100644
--- a/llvm/lib/Passes/PassBuilderPipelines.cpp
+++ b/llvm/lib/Passes/PassBuilderPipelines.cpp
@@ -283,6 +283,7 @@ PipelineTuningOptions::PipelineTuningOptions() {
   LicmMssaOptCap = SetLicmMssaOptCap;
   LicmMssaNoAccForPromotionCap = SetLicmMssaNoAccForPromotionCap;
   CallGraphProfile = true;
+  UnifiedLTO = false;
   MergeFunctions = EnableMergeFunctions;
   InlinerThreshold = -1;
   EagerlyInvalidateAnalyses = EnableEagerlyInvalidateAnalyses;
diff --git a/llvm/lib/Transforms/IPO/ThinLTOBitcodeWriter.cpp b/llvm/lib/Transforms/IPO/ThinLTOBitcodeWriter.cpp
index a7adb74..574874b 100644
--- a/llvm/lib/Transforms/IPO/ThinLTOBitcodeWriter.cpp
+++ b/llvm/lib/Transforms/IPO/ThinLTOBitcodeWriter.cpp
@@ -259,26 +259,37 @@ static void cloneUsedGlobalVariables(const Module &SrcM, Module &DestM,
     appendToUsed(DestM, NewUsed);
 }
 
+bool enableUnifiedLTO(Module &M) {
+  bool UnifiedLTO = false;
+  if (auto *MD =
+          mdconst::extract_or_null<ConstantInt>(M.getModuleFlag("UnifiedLTO")))
+    UnifiedLTO = MD->getZExtValue();
+  return UnifiedLTO;
+}
+
 // If it's possible to split M into regular and thin LTO parts, do so and write
 // a multi-module bitcode file with the two parts to OS. Otherwise, write only a
 // regular LTO bitcode file to OS.
 void splitAndWriteThinLTOBitcode(
     raw_ostream &OS, raw_ostream *ThinLinkOS,
     function_ref<AAResults &(Function &)> AARGetter, Module &M) {
+  bool UnifiedLTO = enableUnifiedLTO(M);
   std::string ModuleId = getUniqueModuleId(&M);
   if (ModuleId.empty()) {
+    assert(!UnifiedLTO);
     // We couldn't generate a module ID for this module, write it out as a
     // regular LTO module with an index for summary-based dead stripping.
     ProfileSummaryInfo PSI(M);
     M.addModuleFlag(Module::Error, "ThinLTO", uint32_t(0));
     ModuleSummaryIndex Index = buildModuleSummaryIndex(M, nullptr, &PSI);
-    WriteBitcodeToFile(M, OS, /*ShouldPreserveUseListOrder=*/false, &Index);
+    WriteBitcodeToFile(M, OS, /*ShouldPreserveUseListOrder=*/false, &Index,
+                       /*UnifiedLTO=*/false);
 
     if (ThinLinkOS)
       // We don't have a ThinLTO part, but still write the module to the
       // ThinLinkOS if requested so that the expected output file is produced.
       WriteBitcodeToFile(M, *ThinLinkOS, /*ShouldPreserveUseListOrder=*/false,
-                         &Index);
+                         &Index, /*UnifiedLTO=*/false);
 
     return;
   }