Clean up external users of GlobalValue::getGUID(StringRef)

See https://discourse.llvm.org/t/rfc-keep-globalvalue-guids-stable/84801
for context.

This is a non-functional change which just changes the interface of
GlobalValue, in preparation for future functional changes. This part
touches a fair few users, so is split out for ease of review. Future
changes to the GlobalValue implementation can then be focused purely on
that class.

This does the following:

* Make global identifier computation nominally private. There are a few
  users we can't remove yet, but we rename the method to make it clear
  that new users shouldn't be added.
* Rename GlobalValue::getGUID(StringRef) to
  getGUIDAssumingExternalLinkage. This is simply making explicit at the
  callsite what is currently implicit.
* Where possible, migrate users to directly calling getGUID on a
  GlobalValue instance.
* Otherwise, where possible, have them call the newly renamed
  getGUIDAssumingExternalLinkage, to make the assumption explicit.
* There are a few cases where neither of the above are possible, as the
  caller saves and reconstructs the necessary information to compute the
  GUID themselves. We want to migrate these callers eventually, but for
  this first step we leave them be and allow them to call our nominally
  private global identifier computation method.

Author: orodley

llvm/include/llvm/IR/GlobalValue.h

@@ -570,6 +570,12 @@ class GlobalValue : public Constant {
     return Name;
   }
 
+  /// Declare a type to represent a global unique identifier for a global value.
+  /// This is a 64 bits hash that is used by PGO and ThinLTO to have a compact
+  /// unique way to identify a symbol.
+  using GUID = uint64_t;
+
+private:
   /// Return the modified name for a global value suitable to be
   /// used as the key for a global lookup (e.g. profile or ThinLTO).
   /// The value's original name is \c Name and has linkage of type
@@ -582,14 +588,18 @@ class GlobalValue : public Constant {
   /// used as the key for a global lookup (e.g. profile or ThinLTO).
   std::string getGlobalIdentifier() const;
 
-  /// Declare a type to represent a global unique identifier for a global value.
-  /// This is a 64 bits hash that is used by PGO and ThinLTO to have a compact
-  /// unique way to identify a symbol.
-  using GUID = uint64_t;
-
   /// Return a 64-bit global unique ID constructed from global value name
   /// (i.e. returned by getGlobalIdentifier()).
-  static GUID getGUID(StringRef GlobalName);
+  static GUID getGUID(StringRef GlobalIdentifier);
+
+public:
+  /// Return a 64-bit global unique ID constructed from the name of a global
+  /// symbol. Since this call doesn't supply the linkage or defining filename,
+  /// the GUID computation will assume that the global has external linkage.
+  static GUID getGUIDAssumingExternalLinkage(StringRef GlobalName) {
+    return getGUID(
+        getGlobalIdentifier(GlobalName, GlobalValue::ExternalLinkage, ""));
+  }
 
   /// Return a 64-bit global unique ID constructed from global value name
   /// (i.e. returned by getGlobalIdentifier()).

llvm/include/llvm/IR/ModuleSummaryIndex.h

@@ -1828,19 +1828,22 @@ class ModuleSummaryIndex {
   /// This accessor can mutate the map and therefore should not be used in
   /// the ThinLTO backends.
   TypeIdSummary &getOrInsertTypeIdSummary(StringRef TypeId) {
-    auto TidIter = TypeIdMap.equal_range(GlobalValue::getGUID(TypeId));
+    auto TidIter = TypeIdMap.equal_range(
+        GlobalValue::getGUIDAssumingExternalLinkage(TypeId));
     for (auto &[GUID, TypeIdPair] : make_range(TidIter))
       if (TypeIdPair.first == TypeId)
         return TypeIdPair.second;
-    auto It = TypeIdMap.insert({GlobalValue::getGUID(TypeId),
-                                {TypeIdSaver.save(TypeId), TypeIdSummary()}});
+    auto It =
+        TypeIdMap.insert({GlobalValue::getGUIDAssumingExternalLinkage(TypeId),
+                          {TypeIdSaver.save(TypeId), TypeIdSummary()}});
     return It->second.second;
   }
 
   /// This returns either a pointer to the type id summary (if present in the
   /// summary map) or null (if not present). This may be used when importing.
   const TypeIdSummary *getTypeIdSummary(StringRef TypeId) const {
-    auto TidIter = TypeIdMap.equal_range(GlobalValue::getGUID(TypeId));
+    auto TidIter = TypeIdMap.equal_range(
+        GlobalValue::getGUIDAssumingExternalLinkage(TypeId));
     for (const auto &[GUID, TypeIdPair] : make_range(TidIter))
       if (TypeIdPair.first == TypeId)
         return &TypeIdPair.second;

llvm/include/llvm/ProfileData/SampleProf.h

@@ -1299,7 +1299,7 @@ class FunctionSamples {
 static inline FunctionId getRepInFormat(StringRef Name) {
   if (Name.empty() || !FunctionSamples::UseMD5)
     return FunctionId(Name);
-  return FunctionId(Function::getGUID(Name));
+  return FunctionId(Function::getGUIDAssumingExternalLinkage(Name));
 }
 
 raw_ostream &operator<<(raw_ostream &OS, const FunctionSamples &FS);

llvm/lib/Analysis/CtxProfAnalysis.cpp

@@ -57,7 +57,7 @@ PreservedAnalyses AssignGUIDPass::run(Module &M, ModuleAnalysisManager &MAM) {
 GlobalValue::GUID AssignGUIDPass::getGUID(const Function &F) {
   if (F.isDeclaration()) {
     assert(GlobalValue::isExternalLinkage(F.getLinkage()));
-    return GlobalValue::getGUID(F.getGlobalIdentifier());
+    return F.getGUID();
   }
   auto *MD = F.getMetadata(GUIDMetadataName);
   assert(MD && "guid not found for defined function");

llvm/lib/Analysis/ModuleSummaryAnalysis.cpp

@@ -222,7 +222,8 @@ static void addIntrinsicToSummary(
     auto *TypeId = dyn_cast<MDString>(TypeMDVal->getMetadata());
     if (!TypeId)
       break;
-    GlobalValue::GUID Guid = GlobalValue::getGUID(TypeId->getString());
+    GlobalValue::GUID Guid =
+        GlobalValue::getGUIDAssumingExternalLinkage(TypeId->getString());
 
     // Produce a summary from type.test intrinsics. We only summarize type.test
     // intrinsics that are used other than by an llvm.assume intrinsic.
@@ -250,7 +251,8 @@ static void addIntrinsicToSummary(
     auto *TypeId = dyn_cast<MDString>(TypeMDVal->getMetadata());
     if (!TypeId)
       break;
-    GlobalValue::GUID Guid = GlobalValue::getGUID(TypeId->getString());
+    GlobalValue::GUID Guid =
+        GlobalValue::getGUIDAssumingExternalLinkage(TypeId->getString());
 
     SmallVector<DevirtCallSite, 4> DevirtCalls;
     SmallVector<Instruction *, 4> LoadedPtrs;
@@ -906,7 +908,8 @@ static void computeAliasSummary(ModuleSummaryIndex &Index, const GlobalAlias &A,
 
 // Set LiveRoot flag on entries matching the given value name.
 static void setLiveRoot(ModuleSummaryIndex &Index, StringRef Name) {
-  if (ValueInfo VI = Index.getValueInfo(GlobalValue::getGUID(Name)))
+  if (ValueInfo VI =
+          Index.getValueInfo(GlobalValue::getGUIDAssumingExternalLinkage(Name)))
     for (const auto &Summary : VI.getSummaryList())
       Summary->setLive(true);
 }

llvm/lib/AsmParser/LLParser.cpp

@@ -9391,11 +9391,9 @@ bool LLParser::addGlobalValueToIndex(
 
       VI = Index->getOrInsertValueInfo(GV);
     } else {
-      assert(
-          (!GlobalValue::isLocalLinkage(Linkage) || !SourceFileName.empty()) &&
-          "Need a source_filename to compute GUID for local");
-      GUID = GlobalValue::getGUID(
-          GlobalValue::getGlobalIdentifier(Name, Linkage, SourceFileName));
+      assert(GlobalValue::isExternalLinkage(Linkage) &&
+             "Cannot compute GUID for local");
+      GUID = GlobalValue::getGUIDAssumingExternalLinkage(Name);
       VI = Index->getOrInsertValueInfo(GUID, Index->saveString(Name));
     }
   }

llvm/lib/Bitcode/Reader/BitcodeReader.cpp

@@ -7221,7 +7221,7 @@ void ModuleSummaryIndexBitcodeReader::setValueGUID(
   auto ValueGUID = GlobalValue::getGUID(GlobalId);
   auto OriginalNameID = ValueGUID;
   if (GlobalValue::isLocalLinkage(Linkage))
-    OriginalNameID = GlobalValue::getGUID(ValueName);
+    OriginalNameID = GlobalValue::getGUIDAssumingExternalLinkage(ValueName);
   if (PrintSummaryGUIDs)
     dbgs() << "GUID " << ValueGUID << "(" << OriginalNameID << ") is "
            << ValueName << "\n";

llvm/lib/Bitcode/Writer/BitcodeWriter.cpp

@@ -5066,8 +5066,8 @@ void IndexBitcodeWriter::writeCombinedGlobalValueSummary() {
 
   if (!Index.cfiFunctionDefs().empty()) {
     for (auto &S : Index.cfiFunctionDefs()) {
-      if (DefOrUseGUIDs.contains(
-              GlobalValue::getGUID(GlobalValue::dropLLVMManglingEscape(S)))) {
+      if (DefOrUseGUIDs.contains(GlobalValue::getGUIDAssumingExternalLinkage(
+              GlobalValue::dropLLVMManglingEscape(S)))) {
         NameVals.push_back(StrtabBuilder.add(S));
         NameVals.push_back(S.size());
       }
@@ -5080,8 +5080,8 @@ void IndexBitcodeWriter::writeCombinedGlobalValueSummary() {
 
   if (!Index.cfiFunctionDecls().empty()) {
     for (auto &S : Index.cfiFunctionDecls()) {
-      if (DefOrUseGUIDs.contains(
-              GlobalValue::getGUID(GlobalValue::dropLLVMManglingEscape(S)))) {
+      if (DefOrUseGUIDs.contains(GlobalValue::getGUIDAssumingExternalLinkage(
+              GlobalValue::dropLLVMManglingEscape(S)))) {
         NameVals.push_back(StrtabBuilder.add(S));
         NameVals.push_back(S.size());
       }

llvm/lib/CodeGen/PseudoProbeInserter.cpp

@@ -129,7 +129,7 @@ class PseudoProbeInserter : public MachineFunctionPass {
 private:
   uint64_t getFuncGUID(Module *M, DILocation *DL) {
     auto Name = DL->getSubprogramLinkageName();
-    return Function::getGUID(Name);
+    return Function::getGUIDAssumingExternalLinkage(Name);
   }
 
   bool ShouldRun = false;

llvm/lib/IR/AsmWriter.cpp

@@ -3154,7 +3154,7 @@ void AssemblyWriter::printModuleSummaryIndex() {
 
   // Print the TypeIdCompatibleVtableMap entries.
   for (auto &TId : TheIndex->typeIdCompatibleVtableMap()) {
-    auto GUID = GlobalValue::getGUID(TId.first);
+    auto GUID = GlobalValue::getGUIDAssumingExternalLinkage(TId.first);
     Out << "^" << Machine.getTypeIdCompatibleVtableSlot(TId.first)
         << " = typeidCompatibleVTable: (name: \"" << TId.first << "\"";
     printTypeIdCompatibleVtableSummary(TId.second);