Revert "[Serialization] Remove delta encoding optimization (llvm#145670)" (swiftlang#11622)

This reverts commit e90ab0e. Removing the delta encoding optimization led to a 5% in PCH sizes and builders ran out of disk space. 

 Conflicts:
	clang/lib/Serialization/ASTWriter.cpp

Temporary workaround for rdar://162016610.

Author: qiongsiwu

clang/include/clang/Serialization/ASTReader.h

@@ -488,6 +488,8 @@ class ASTReader
   using ModuleReverseIterator = ModuleManager::ModuleReverseIterator;
 
 private:
+  using LocSeq = SourceLocationSequence;
+
   /// The receiver of some callbacks invoked by ASTReader.
   std::unique_ptr<ASTReaderListener> Listener;
 
@@ -2450,16 +2452,18 @@ class ASTReader
   /// Read a source location from raw form and return it in its
   /// originating module file's source location space.
   std::pair<SourceLocation, unsigned>
-  ReadUntranslatedSourceLocation(RawLocEncoding Raw) const {
-    return SourceLocationEncoding::decode(Raw);
+  ReadUntranslatedSourceLocation(RawLocEncoding Raw,
+                                 LocSeq *Seq = nullptr) const {
+    return SourceLocationEncoding::decode(Raw, Seq);
   }
 
   /// Read a source location from raw form.
-  SourceLocation ReadSourceLocation(ModuleFile &MF, RawLocEncoding Raw) const {
+  SourceLocation ReadSourceLocation(ModuleFile &MF, RawLocEncoding Raw,
+                                    LocSeq *Seq = nullptr) const {
     if (!MF.ModuleOffsetMap.empty())
       ReadModuleOffsetMap(MF);
 
-    auto [Loc, ModuleFileIndex] = ReadUntranslatedSourceLocation(Raw);
+    auto [Loc, ModuleFileIndex] = ReadUntranslatedSourceLocation(Raw, Seq);
     ModuleFile *OwningModuleFile =
         ModuleFileIndex == 0 ? &MF : MF.TransitiveImports[ModuleFileIndex - 1];
 
@@ -2487,9 +2491,9 @@ class ASTReader
 
   /// Read a source location.
   SourceLocation ReadSourceLocation(ModuleFile &ModuleFile,
-                                    const RecordDataImpl &Record,
-                                    unsigned &Idx) {
-    return ReadSourceLocation(ModuleFile, Record[Idx++]);
+                                    const RecordDataImpl &Record, unsigned &Idx,
+                                    LocSeq *Seq = nullptr) {
+    return ReadSourceLocation(ModuleFile, Record[Idx++], Seq);
   }
 
   /// Read a FileID.
@@ -2508,7 +2512,7 @@ class ASTReader
 
   /// Read a source range.
   SourceRange ReadSourceRange(ModuleFile &F, const RecordData &Record,
-                              unsigned &Idx);
+                              unsigned &Idx, LocSeq *Seq = nullptr);
 
   static llvm::BitVector ReadBitVector(const RecordData &Record,
                                        const StringRef Blob);

clang/include/clang/Serialization/ASTRecordReader.h

@@ -32,6 +32,7 @@ class OMPChildren;
 class ASTRecordReader
     : public serialization::DataStreamBasicReader<ASTRecordReader> {
   using ModuleFile = serialization::ModuleFile;
+  using LocSeq = SourceLocationSequence;
 
   ASTReader *Reader;
   ModuleFile *F;
@@ -159,7 +160,7 @@ class ASTRecordReader
   TypeSourceInfo *readTypeSourceInfo();
 
   /// Reads the location information for a type.
-  void readTypeLoc(TypeLoc TL);
+  void readTypeLoc(TypeLoc TL, LocSeq *Seq = nullptr);
 
   /// Map a local type ID within a given AST file to a global type ID.
   serialization::TypeID getGlobalTypeID(serialization::TypeID LocalID) const {
@@ -292,13 +293,13 @@ class ASTRecordReader
   void readOpenACCRoutineDeclAttr(OpenACCRoutineDeclAttr *A);
 
   /// Read a source location, advancing Idx.
-  SourceLocation readSourceLocation() {
-    return Reader->ReadSourceLocation(*F, Record, Idx);
+  SourceLocation readSourceLocation(LocSeq *Seq = nullptr) {
+    return Reader->ReadSourceLocation(*F, Record, Idx, Seq);
   }
 
   /// Read a source range, advancing Idx.
-  SourceRange readSourceRange() {
-    return Reader->ReadSourceRange(*F, Record, Idx);
+  SourceRange readSourceRange(LocSeq *Seq = nullptr) {
+    return Reader->ReadSourceRange(*F, Record, Idx, Seq);
   }
 
   /// Read an arbitrary constant value, advancing Idx.

clang/include/clang/Serialization/ASTRecordWriter.h

@@ -29,6 +29,7 @@ class TypeLoc;
 /// An object for streaming information to a record.
 class ASTRecordWriter
     : public serialization::DataStreamBasicWriter<ASTRecordWriter> {
+  using LocSeq = SourceLocationSequence;
 
   ASTWriter *Writer;
   ASTWriter::RecordDataImpl *Record;
@@ -146,8 +147,8 @@ class ASTRecordWriter
   void AddFunctionDefinition(const FunctionDecl *FD);
 
   /// Emit a source location.
-  void AddSourceLocation(SourceLocation Loc) {
-    return Writer->AddSourceLocation(Loc, *Record);
+  void AddSourceLocation(SourceLocation Loc, LocSeq *Seq = nullptr) {
+    return Writer->AddSourceLocation(Loc, *Record, Seq);
   }
   void writeSourceLocation(SourceLocation Loc) {
     AddSourceLocation(Loc);
@@ -174,8 +175,8 @@ class ASTRecordWriter
   }
 
   /// Emit a source range.
-  void AddSourceRange(SourceRange Range) {
-    return Writer->AddSourceRange(Range, *Record);
+  void AddSourceRange(SourceRange Range, LocSeq *Seq = nullptr) {
+    return Writer->AddSourceRange(Range, *Record, Seq);
   }
 
   void writeBool(bool Value) {
@@ -245,7 +246,7 @@ class ASTRecordWriter
   void AddTypeSourceInfo(TypeSourceInfo *TInfo);
 
   /// Emits source location information for a type. Does not emit the type.
-  void AddTypeLoc(TypeLoc TL);
+  void AddTypeLoc(TypeLoc TL, LocSeq *Seq = nullptr);
 
   /// Emits a template argument location info.
   void AddTemplateArgumentLocInfo(TemplateArgument::ArgKind Kind,

clang/include/clang/Serialization/ASTWriter.h

@@ -115,6 +115,8 @@ class ASTWriter : public ASTDeserializationListener,
   using TypeIdxMap = llvm::DenseMap<QualType, serialization::TypeIdx,
                                     serialization::UnsafeQualTypeDenseMapInfo>;
 
+  using LocSeq = SourceLocationSequence;
+
   /// The bitstream writer used to emit this precompiled header.
   llvm::BitstreamWriter &Stream;
 
@@ -733,14 +735,16 @@ class ASTWriter : public ASTDeserializationListener,
   void AddFileID(FileID FID, RecordDataImpl &Record);
 
   /// Emit a source location.
-  void AddSourceLocation(SourceLocation Loc, RecordDataImpl &Record);
+  void AddSourceLocation(SourceLocation Loc, RecordDataImpl &Record,
+                         LocSeq *Seq = nullptr);
 
   /// Return the raw encodings for source locations.
   SourceLocationEncoding::RawLocEncoding
-  getRawSourceLocationEncoding(SourceLocation Loc);
+  getRawSourceLocationEncoding(SourceLocation Loc, LocSeq *Seq = nullptr);
 
   /// Emit a source range.
-  void AddSourceRange(SourceRange Range, RecordDataImpl &Record);
+  void AddSourceRange(SourceRange Range, RecordDataImpl &Record,
+                      LocSeq *Seq = nullptr);
 
   /// Emit a reference to an identifier.
   void AddIdentifierRef(const IdentifierInfo *II, RecordDataImpl &Record);

clang/include/clang/Serialization/SourceLocationEncoding.h

@@ -25,6 +25,8 @@
 //  * C: The macro bit. We rotate it to the lowest bit so that we can save some
 //  space in case the index of the module file is 0.
 //
+// Specially, if the index of the module file is 0, we allow to encode a
+// sequence of locations we store only differences between successive elements.
 //
 //===----------------------------------------------------------------------===//
 
@@ -36,6 +38,7 @@
 #include <climits>
 
 namespace clang {
+class SourceLocationSequence;
 
 /// Serialized encoding of SourceLocations without context.
 /// Optimized to have small unsigned values (=> small after VBR encoding).
@@ -51,22 +54,119 @@ class SourceLocationEncoding {
   static UIntTy decodeRaw(UIntTy Raw) {
     return (Raw >> 1) | (Raw << (UIntBits - 1));
   }
+  friend SourceLocationSequence;
 
 public:
   using RawLocEncoding = uint64_t;
 
   static RawLocEncoding encode(SourceLocation Loc, UIntTy BaseOffset,
-                               unsigned BaseModuleFileIndex);
-  static std::pair<SourceLocation, unsigned> decode(RawLocEncoding);
+                               unsigned BaseModuleFileIndex,
+                               SourceLocationSequence * = nullptr);
+  static std::pair<SourceLocation, unsigned>
+  decode(RawLocEncoding, SourceLocationSequence * = nullptr);
+};
+
+/// Serialized encoding of a sequence of SourceLocations.
+///
+/// Optimized to produce small values when locations with the sequence are
+/// similar. Each element can be delta-encoded against the last nonzero element.
+///
+/// Sequences should be started by creating a SourceLocationSequence::State,
+/// and then passed around as SourceLocationSequence*. Example:
+///
+///   // establishes a sequence
+///   void EmitTopLevelThing() {
+///     SourceLocationSequence::State Seq;
+///     EmitContainedThing(Seq);
+///     EmitRecursiveThing(Seq);
+///   }
+///
+///   // optionally part of a sequence
+///   void EmitContainedThing(SourceLocationSequence *Seq = nullptr) {
+///     Record.push_back(SourceLocationEncoding::encode(SomeLoc, Seq));
+///   }
+///
+///   // establishes a sequence if there isn't one already
+///   void EmitRecursiveThing(SourceLocationSequence *ParentSeq = nullptr) {
+///     SourceLocationSequence::State Seq(ParentSeq);
+///     Record.push_back(SourceLocationEncoding::encode(SomeLoc, Seq));
+///     EmitRecursiveThing(Seq);
+///   }
+///
+class SourceLocationSequence {
+  using UIntTy = SourceLocation::UIntTy;
+  using EncodedTy = uint64_t;
+  constexpr static auto UIntBits = SourceLocationEncoding::UIntBits;
+  static_assert(sizeof(EncodedTy) > sizeof(UIntTy), "Need one extra bit!");
+
+  // Prev stores the rotated last nonzero location.
+  UIntTy &Prev;
+
+  // Zig-zag encoding turns small signed integers into small unsigned integers.
+  // 0 => 0, -1 => 1, 1 => 2, -2 => 3, ...
+  static UIntTy zigZag(UIntTy V) {
+    UIntTy Sign = (V & (1 << (UIntBits - 1))) ? UIntTy(-1) : UIntTy(0);
+    return Sign ^ (V << 1);
+  }
+  static UIntTy zagZig(UIntTy V) { return (V >> 1) ^ -(V & 1); }
+
+  SourceLocationSequence(UIntTy &Prev) : Prev(Prev) {}
+
+  EncodedTy encodeRaw(UIntTy Raw) {
+    if (Raw == 0)
+      return 0;
+    UIntTy Rotated = SourceLocationEncoding::encodeRaw(Raw);
+    if (Prev == 0)
+      return Prev = Rotated;
+    UIntTy Delta = Rotated - Prev;
+    Prev = Rotated;
+    // Exactly one 33 bit value is possible! (1 << 32).
+    // This is because we have two representations of zero: trivial & relative.
+    return 1 + EncodedTy{zigZag(Delta)};
+  }
+  UIntTy decodeRaw(EncodedTy Encoded) {
+    if (Encoded == 0)
+      return 0;
+    if (Prev == 0)
+      return SourceLocationEncoding::decodeRaw(Prev = Encoded);
+    return SourceLocationEncoding::decodeRaw(Prev += zagZig(Encoded - 1));
+  }
+
+public:
+  SourceLocation decode(EncodedTy Encoded) {
+    return SourceLocation::getFromRawEncoding(decodeRaw(Encoded));
+  }
+  EncodedTy encode(SourceLocation Loc) {
+    return encodeRaw(Loc.getRawEncoding());
+  }
+
+  class State;
+};
+
+/// This object establishes a SourceLocationSequence.
+class SourceLocationSequence::State {
+  UIntTy Prev = 0;
+  SourceLocationSequence Seq;
+
+public:
+  // If Parent is provided and non-null, then this root becomes part of that
+  // enclosing sequence instead of establishing a new one.
+  State(SourceLocationSequence *Parent = nullptr)
+      : Seq(Parent ? Parent->Prev : Prev) {}
+
+  // Implicit conversion for uniform use of roots vs propagated sequences.
+  operator SourceLocationSequence *() { return &Seq; }
 };
 
 inline SourceLocationEncoding::RawLocEncoding
 SourceLocationEncoding::encode(SourceLocation Loc, UIntTy BaseOffset,
-                               unsigned BaseModuleFileIndex) {
+                               unsigned BaseModuleFileIndex,
+                               SourceLocationSequence *Seq) {
   // If the source location is a local source location, we can try to optimize
   // the similar sequences to only record the differences.
   if (!BaseOffset)
-    return encodeRaw(Loc.getRawEncoding());
+    return Seq ? Seq->encode(Loc) : encodeRaw(Loc.getRawEncoding());
+
   if (Loc.isInvalid())
     return 0;
 
@@ -83,11 +183,13 @@ SourceLocationEncoding::encode(SourceLocation Loc, UIntTy BaseOffset,
   return Encoded;
 }
 inline std::pair<SourceLocation, unsigned>
-SourceLocationEncoding::decode(RawLocEncoding Encoded) {
+SourceLocationEncoding::decode(RawLocEncoding Encoded,
+                               SourceLocationSequence *Seq) {
   unsigned ModuleFileIndex = Encoded >> 32;
 
   if (!ModuleFileIndex)
-    return {SourceLocation::getFromRawEncoding(decodeRaw(Encoded)),
+    return {Seq ? Seq->decode(Encoded)
+                : SourceLocation::getFromRawEncoding(decodeRaw(Encoded)),
             ModuleFileIndex};
 
   Encoded &= llvm::maskTrailingOnes<RawLocEncoding>(32);

clang/lib/Serialization/ASTReader.cpp

@@ -1935,9 +1935,10 @@ bool ASTReader::ReadSLocEntry(int ID) {
   }
 
   case SM_SLOC_EXPANSION_ENTRY: {
-    SourceLocation SpellingLoc = ReadSourceLocation(*F, Record[1]);
-    SourceLocation ExpansionBegin = ReadSourceLocation(*F, Record[2]);
-    SourceLocation ExpansionEnd = ReadSourceLocation(*F, Record[3]);
+    LocSeq::State Seq;
+    SourceLocation SpellingLoc = ReadSourceLocation(*F, Record[1], Seq);
+    SourceLocation ExpansionBegin = ReadSourceLocation(*F, Record[2], Seq);
+    SourceLocation ExpansionEnd = ReadSourceLocation(*F, Record[3], Seq);
     SourceMgr.createExpansionLoc(SpellingLoc, ExpansionBegin, ExpansionEnd,
                                  Record[5], Record[4], ID,
                                  BaseOffset + Record[0]);
@@ -7192,10 +7193,13 @@ QualType ASTReader::readTypeRecord(TypeID ID) {
 namespace clang {
 
 class TypeLocReader : public TypeLocVisitor<TypeLocReader> {
+  using LocSeq = SourceLocationSequence;
+
   ASTRecordReader &Reader;
+  LocSeq *Seq;
 
-  SourceLocation readSourceLocation() { return Reader.readSourceLocation(); }
-  SourceRange readSourceRange() { return Reader.readSourceRange(); }
+  SourceLocation readSourceLocation() { return Reader.readSourceLocation(Seq); }
+  SourceRange readSourceRange() { return Reader.readSourceRange(Seq); }
 
   TypeSourceInfo *GetTypeSourceInfo() {
     return Reader.readTypeSourceInfo();
@@ -7210,7 +7214,8 @@ class TypeLocReader : public TypeLocVisitor<TypeLocReader> {
   }
 
 public:
-  TypeLocReader(ASTRecordReader &Reader) : Reader(Reader) {}
+  TypeLocReader(ASTRecordReader &Reader, LocSeq *Seq)
+      : Reader(Reader), Seq(Seq) {}
 
   // We want compile-time assurance that we've enumerated all of
   // these, so unfortunately we have to declare them first, then
@@ -7585,8 +7590,9 @@ void TypeLocReader::VisitDependentBitIntTypeLoc(
   TL.setNameLoc(readSourceLocation());
 }
 
-void ASTRecordReader::readTypeLoc(TypeLoc TL) {
-  TypeLocReader TLR(*this);
+void ASTRecordReader::readTypeLoc(TypeLoc TL, LocSeq *ParentSeq) {
+  LocSeq::State Seq(ParentSeq);
+  TypeLocReader TLR(*this, Seq);
   for (; !TL.isNull(); TL = TL.getNextTypeLoc())
     TLR.Visit(TL);
 }
@@ -10157,9 +10163,9 @@ ASTRecordReader::readNestedNameSpecifierLoc() {
 }
 
 SourceRange ASTReader::ReadSourceRange(ModuleFile &F, const RecordData &Record,
-                                       unsigned &Idx) {
-  SourceLocation beg = ReadSourceLocation(F, Record, Idx);
-  SourceLocation end = ReadSourceLocation(F, Record, Idx);
+                                       unsigned &Idx, LocSeq *Seq) {
+  SourceLocation beg = ReadSourceLocation(F, Record, Idx, Seq);
+  SourceLocation end = ReadSourceLocation(F, Record, Idx, Seq);
   return SourceRange(beg, end);
 }