[PCH] Add an option to not write comments into PCH

[ilya-biryukov]

Summary

Will be used in clangd, see the follow-up change.
Clangd does not use comments read from PCH to avoid crashes due to
changed contents of the file. However, reading them considerably slows
down code completion on files with large preambles.

[ilya-biryukov]

Not sure about two things:

• Using PreprocessorOptions for plumbing this setting. I'd rather put it into FrontendOptions, but there seems to be no way to get those in the ASTWriter... Happy to search for the alternatives
• Lack of tests. I'm not sure how to properly test it in clang. I've added asserts in clangd, see the follow-up D48943. But happy to look into writing actual tests if that looks important.

Also not sure if someone else should look at this change, suggestions for reviewers are very welcome!

include/clang/Lex/PreprocessorOptions.h

public:
 
   /// True indicates that a preamble is being generated.
   ///
   /// When the lexer is done, one of the things that need to be preserved is the
   /// conditional #if stack, so the ASTWriter/ASTReader can save/restore it when
   /// processing the rest of the file.
   bool GeneratePreamble = false;
 
+  /// Whether to write comment locations into the PCH when building it.
+  /// Reading the comments from the PCH can be a performance hit even if the
+  /// clients don't use them.
+  bool WriteCommentListToPCH = true;
+
   /// The implicit PTH input included at the start of the translation unit, or
   /// empty.
   std::string ImplicitPTHInclude;
 
   /// If given, a PTH cache file to use for speeding up header parsing.
   std::string TokenCache;
 
   /// When enabled, preprocessor is in a mode for parsing a single file only.

lib/Serialization/ASTWriter.cpp

 #include "llvm/ADT/APInt.h"
 #include "llvm/ADT/APSInt.h"
 #include "llvm/ADT/ArrayRef.h"
 #include "llvm/ADT/DenseMap.h"
 #include "llvm/ADT/Hashing.h"
 #include "llvm/ADT/Optional.h"
 #include "llvm/ADT/PointerIntPair.h"
 #include "llvm/ADT/STLExtras.h"
+#include "llvm/ADT/ScopeExit.h"
 #include "llvm/ADT/SmallSet.h"
 #include "llvm/ADT/SmallString.h"
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/StringMap.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Bitcode/BitCodes.h"
 #include "llvm/Bitcode/BitstreamWriter.h"
 #include "llvm/Support/Casting.h"
void ASTWriter::WriteFileDeclIDsMap() {
   unsigned AbbrevCode = Stream.EmitAbbrev(std::move(Abbrev));
   RecordData::value_type Record[] = {FILE_SORTED_DECLS,
                                      FileGroupedDeclIDs.size()};
   Stream.EmitRecordWithBlob(AbbrevCode, Record, bytes(FileGroupedDeclIDs));
 }
 
 void ASTWriter::WriteComments() {
   Stream.EnterSubblock(COMMENTS_BLOCK_ID, 3);
+  auto _ = llvm::make_scope_exit([this] { Stream.ExitBlock(); });
+  if (!PP->getPreprocessorOpts().WriteCommentListToPCH)
+    return;
   ArrayRef<RawComment *> RawComments = Context->Comments.getComments();
   RecordData Record;
   for (const auto *I : RawComments) {
     Record.clear();
     AddSourceRange(I->getSourceRange(), Record);
     Record.push_back(I->getKind());
     Record.push_back(I->isTrailingComment());
     Record.push_back(I->isAlmostTrailingComment());
     Stream.EmitRecord(COMMENTS_RAW_COMMENT, Record);
   }
-  Stream.ExitBlock();
 }
 
 //===----------------------------------------------------------------------===//
 // Global Method Pool and Selector Serialization
 //===----------------------------------------------------------------------===//
 
 namespace {