[clang][modules] NFC: Only sort interesting identifiers

[jansvoboda11]

Summary

In 9c254184 ASTWriter stopped writing identifiers that are not interesting. Taking it a bit further, we don't need to sort the whole identifier table, just the interesting identifiers. This reduces the size of sorted vector from ~10k (including lots of builtins) to 2 (__VA_ARGS__ and __VA_OPT__) in a typical Xcode project, improving clang-scan-deps performance.

[ributzka]

clang/lib/Serialization/ASTWriter.cpp

3646

Would it make sense to reduce the size of the SmallVector too?

[jansvoboda11]

clang/lib/Serialization/ASTWriter.cpp

3646

That would most likely regress serialization of real (non-scanning) PCMs, so I'd prefer to leave this be.

clang/lib/Serialization/ASTWriter.cpp

ASTIdentifierTableTrait Trait(
         *this, PP, IdResolver, IsModule,
         (getLangOpts().CPlusPlus && IsModule) ? &InterestingIdents : nullptr);
 
     // Look for any identifiers that were named while processing the
     // headers, but are otherwise not needed. We add these to the hash
     // table to enable checking of the predefines buffer in the case
     // where the user adds new macro definitions when building the AST
     // file.
     SmallVector<const IdentifierInfo *, 128> IIs;
     for (const auto &ID : PP.getIdentifierTable())
+      if (Trait.isInterestingNonMacroIdentifier(ID.second))
         IIs.push_back(ID.second);
-    // Sort the identifiers lexicographically before getting them references so
+    // Sort the identifiers lexicographically before getting the references so
     // that their order is stable.
     llvm::sort(IIs, llvm::deref<std::less<>>());
     for (const IdentifierInfo *II : IIs)
-      if (Trait.isInterestingNonMacroIdentifier(II))
-        getIdentifierRef(II);
 
     // Create the on-disk hash table representation. We only store offsets
     // for identifiers that appear here for the first time.
     IdentifierOffsets.resize(NextIdentID - FirstIdentID);
     for (auto IdentIDPair : IdentifierIDs) {
       auto *II = const_cast<IdentifierInfo *>(IdentIDPair.first);
       IdentID ID = IdentIDPair.second;
       assert(II && "NULL identifier in identifier table");