xref: /llvm-project/llvm/lib/CodeGen/AsmPrinter/CodeViewDebug.cpp (revision 55baeefd54b1e89551545b01b82c5fbf041bb0ad)
1 //===- llvm/lib/CodeGen/AsmPrinter/CodeViewDebug.cpp ----------------------===//
2 //
3 //                     The LLVM Compiler Infrastructure
4 //
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
7 //
8 //===----------------------------------------------------------------------===//
9 //
10 // This file contains support for writing Microsoft CodeView debug info.
11 //
12 //===----------------------------------------------------------------------===//
13 
14 #include "CodeViewDebug.h"
15 #include "DwarfExpression.h"
16 #include "llvm/ADT/APSInt.h"
17 #include "llvm/ADT/ArrayRef.h"
18 #include "llvm/ADT/DenseMap.h"
19 #include "llvm/ADT/DenseSet.h"
20 #include "llvm/ADT/MapVector.h"
21 #include "llvm/ADT/None.h"
22 #include "llvm/ADT/Optional.h"
23 #include "llvm/ADT/STLExtras.h"
24 #include "llvm/ADT/SmallString.h"
25 #include "llvm/ADT/SmallVector.h"
26 #include "llvm/ADT/StringRef.h"
27 #include "llvm/ADT/TinyPtrVector.h"
28 #include "llvm/ADT/Triple.h"
29 #include "llvm/ADT/Twine.h"
30 #include "llvm/BinaryFormat/COFF.h"
31 #include "llvm/BinaryFormat/Dwarf.h"
32 #include "llvm/CodeGen/AsmPrinter.h"
33 #include "llvm/CodeGen/LexicalScopes.h"
34 #include "llvm/CodeGen/MachineFunction.h"
35 #include "llvm/CodeGen/MachineInstr.h"
36 #include "llvm/CodeGen/MachineModuleInfo.h"
37 #include "llvm/CodeGen/MachineOperand.h"
38 #include "llvm/CodeGen/TargetFrameLowering.h"
39 #include "llvm/CodeGen/TargetLoweringObjectFile.h"
40 #include "llvm/CodeGen/TargetRegisterInfo.h"
41 #include "llvm/CodeGen/TargetSubtargetInfo.h"
42 #include "llvm/Config/llvm-config.h"
43 #include "llvm/DebugInfo/CodeView/CVTypeVisitor.h"
44 #include "llvm/DebugInfo/CodeView/CodeView.h"
45 #include "llvm/DebugInfo/CodeView/ContinuationRecordBuilder.h"
46 #include "llvm/DebugInfo/CodeView/DebugInlineeLinesSubsection.h"
47 #include "llvm/DebugInfo/CodeView/Line.h"
48 #include "llvm/DebugInfo/CodeView/SymbolRecord.h"
49 #include "llvm/DebugInfo/CodeView/TypeDumpVisitor.h"
50 #include "llvm/DebugInfo/CodeView/TypeIndex.h"
51 #include "llvm/DebugInfo/CodeView/TypeRecord.h"
52 #include "llvm/DebugInfo/CodeView/TypeTableCollection.h"
53 #include "llvm/IR/Constants.h"
54 #include "llvm/IR/DataLayout.h"
55 #include "llvm/IR/DebugInfoMetadata.h"
56 #include "llvm/IR/DebugLoc.h"
57 #include "llvm/IR/Function.h"
58 #include "llvm/IR/GlobalValue.h"
59 #include "llvm/IR/GlobalVariable.h"
60 #include "llvm/IR/Metadata.h"
61 #include "llvm/IR/Module.h"
62 #include "llvm/MC/MCAsmInfo.h"
63 #include "llvm/MC/MCContext.h"
64 #include "llvm/MC/MCSectionCOFF.h"
65 #include "llvm/MC/MCStreamer.h"
66 #include "llvm/MC/MCSymbol.h"
67 #include "llvm/Support/BinaryByteStream.h"
68 #include "llvm/Support/BinaryStreamReader.h"
69 #include "llvm/Support/Casting.h"
70 #include "llvm/Support/CommandLine.h"
71 #include "llvm/Support/Compiler.h"
72 #include "llvm/Support/Endian.h"
73 #include "llvm/Support/Error.h"
74 #include "llvm/Support/ErrorHandling.h"
75 #include "llvm/Support/FormatVariadic.h"
76 #include "llvm/Support/SMLoc.h"
77 #include "llvm/Support/ScopedPrinter.h"
78 #include "llvm/Target/TargetMachine.h"
79 #include <algorithm>
80 #include <cassert>
81 #include <cctype>
82 #include <cstddef>
83 #include <cstdint>
84 #include <iterator>
85 #include <limits>
86 #include <string>
87 #include <utility>
88 #include <vector>
89 
90 using namespace llvm;
91 using namespace llvm::codeview;
92 
93 static cl::opt<bool> EmitDebugGlobalHashes("emit-codeview-ghash-section",
94                                            cl::ReallyHidden, cl::init(false));
95 
96 CodeViewDebug::CodeViewDebug(AsmPrinter *AP)
97     : DebugHandlerBase(AP), OS(*Asm->OutStreamer), TypeTable(Allocator) {
98   // If module doesn't have named metadata anchors or COFF debug section
99   // is not available, skip any debug info related stuff.
100   if (!MMI->getModule()->getNamedMetadata("llvm.dbg.cu") ||
101       !AP->getObjFileLowering().getCOFFDebugSymbolsSection()) {
102     Asm = nullptr;
103     return;
104   }
105 
106   // Tell MMI that we have debug info.
107   MMI->setDebugInfoAvailability(true);
108 }
109 
110 StringRef CodeViewDebug::getFullFilepath(const DIFile *File) {
111   std::string &Filepath = FileToFilepathMap[File];
112   if (!Filepath.empty())
113     return Filepath;
114 
115   StringRef Dir = File->getDirectory(), Filename = File->getFilename();
116 
117   // Clang emits directory and relative filename info into the IR, but CodeView
118   // operates on full paths.  We could change Clang to emit full paths too, but
119   // that would increase the IR size and probably not needed for other users.
120   // For now, just concatenate and canonicalize the path here.
121   if (Filename.find(':') == 1)
122     Filepath = Filename;
123   else
124     Filepath = (Dir + "\\" + Filename).str();
125 
126   // Canonicalize the path.  We have to do it textually because we may no longer
127   // have access the file in the filesystem.
128   // First, replace all slashes with backslashes.
129   std::replace(Filepath.begin(), Filepath.end(), '/', '\\');
130 
131   // Remove all "\.\" with "\".
132   size_t Cursor = 0;
133   while ((Cursor = Filepath.find("\\.\\", Cursor)) != std::string::npos)
134     Filepath.erase(Cursor, 2);
135 
136   // Replace all "\XXX\..\" with "\".  Don't try too hard though as the original
137   // path should be well-formatted, e.g. start with a drive letter, etc.
138   Cursor = 0;
139   while ((Cursor = Filepath.find("\\..\\", Cursor)) != std::string::npos) {
140     // Something's wrong if the path starts with "\..\", abort.
141     if (Cursor == 0)
142       break;
143 
144     size_t PrevSlash = Filepath.rfind('\\', Cursor - 1);
145     if (PrevSlash == std::string::npos)
146       // Something's wrong, abort.
147       break;
148 
149     Filepath.erase(PrevSlash, Cursor + 3 - PrevSlash);
150     // The next ".." might be following the one we've just erased.
151     Cursor = PrevSlash;
152   }
153 
154   // Remove all duplicate backslashes.
155   Cursor = 0;
156   while ((Cursor = Filepath.find("\\\\", Cursor)) != std::string::npos)
157     Filepath.erase(Cursor, 1);
158 
159   return Filepath;
160 }
161 
162 unsigned CodeViewDebug::maybeRecordFile(const DIFile *F) {
163   StringRef FullPath = getFullFilepath(F);
164   unsigned NextId = FileIdMap.size() + 1;
165   auto Insertion = FileIdMap.insert(std::make_pair(FullPath, NextId));
166   if (Insertion.second) {
167     // We have to compute the full filepath and emit a .cv_file directive.
168     ArrayRef<uint8_t> ChecksumAsBytes;
169     FileChecksumKind CSKind = FileChecksumKind::None;
170     if (F->getChecksum()) {
171       std::string Checksum = fromHex(F->getChecksum()->Value);
172       void *CKMem = OS.getContext().allocate(Checksum.size(), 1);
173       memcpy(CKMem, Checksum.data(), Checksum.size());
174       ChecksumAsBytes = ArrayRef<uint8_t>(
175           reinterpret_cast<const uint8_t *>(CKMem), Checksum.size());
176       switch (F->getChecksum()->Kind) {
177       case DIFile::CSK_MD5:  CSKind = FileChecksumKind::MD5; break;
178       case DIFile::CSK_SHA1: CSKind = FileChecksumKind::SHA1; break;
179       }
180     }
181     bool Success = OS.EmitCVFileDirective(NextId, FullPath, ChecksumAsBytes,
182                                           static_cast<unsigned>(CSKind));
183     (void)Success;
184     assert(Success && ".cv_file directive failed");
185   }
186   return Insertion.first->second;
187 }
188 
189 CodeViewDebug::InlineSite &
190 CodeViewDebug::getInlineSite(const DILocation *InlinedAt,
191                              const DISubprogram *Inlinee) {
192   auto SiteInsertion = CurFn->InlineSites.insert({InlinedAt, InlineSite()});
193   InlineSite *Site = &SiteInsertion.first->second;
194   if (SiteInsertion.second) {
195     unsigned ParentFuncId = CurFn->FuncId;
196     if (const DILocation *OuterIA = InlinedAt->getInlinedAt())
197       ParentFuncId =
198           getInlineSite(OuterIA, InlinedAt->getScope()->getSubprogram())
199               .SiteFuncId;
200 
201     Site->SiteFuncId = NextFuncId++;
202     OS.EmitCVInlineSiteIdDirective(
203         Site->SiteFuncId, ParentFuncId, maybeRecordFile(InlinedAt->getFile()),
204         InlinedAt->getLine(), InlinedAt->getColumn(), SMLoc());
205     Site->Inlinee = Inlinee;
206     InlinedSubprograms.insert(Inlinee);
207     getFuncIdForSubprogram(Inlinee);
208   }
209   return *Site;
210 }
211 
212 static StringRef getPrettyScopeName(const DIScope *Scope) {
213   StringRef ScopeName = Scope->getName();
214   if (!ScopeName.empty())
215     return ScopeName;
216 
217   switch (Scope->getTag()) {
218   case dwarf::DW_TAG_enumeration_type:
219   case dwarf::DW_TAG_class_type:
220   case dwarf::DW_TAG_structure_type:
221   case dwarf::DW_TAG_union_type:
222     return "<unnamed-tag>";
223   case dwarf::DW_TAG_namespace:
224     return "`anonymous namespace'";
225   }
226 
227   return StringRef();
228 }
229 
230 static const DISubprogram *getQualifiedNameComponents(
231     const DIScope *Scope, SmallVectorImpl<StringRef> &QualifiedNameComponents) {
232   const DISubprogram *ClosestSubprogram = nullptr;
233   while (Scope != nullptr) {
234     if (ClosestSubprogram == nullptr)
235       ClosestSubprogram = dyn_cast<DISubprogram>(Scope);
236     StringRef ScopeName = getPrettyScopeName(Scope);
237     if (!ScopeName.empty())
238       QualifiedNameComponents.push_back(ScopeName);
239     Scope = Scope->getScope().resolve();
240   }
241   return ClosestSubprogram;
242 }
243 
244 static std::string getQualifiedName(ArrayRef<StringRef> QualifiedNameComponents,
245                                     StringRef TypeName) {
246   std::string FullyQualifiedName;
247   for (StringRef QualifiedNameComponent :
248        llvm::reverse(QualifiedNameComponents)) {
249     FullyQualifiedName.append(QualifiedNameComponent);
250     FullyQualifiedName.append("::");
251   }
252   FullyQualifiedName.append(TypeName);
253   return FullyQualifiedName;
254 }
255 
256 static std::string getFullyQualifiedName(const DIScope *Scope, StringRef Name) {
257   SmallVector<StringRef, 5> QualifiedNameComponents;
258   getQualifiedNameComponents(Scope, QualifiedNameComponents);
259   return getQualifiedName(QualifiedNameComponents, Name);
260 }
261 
262 struct CodeViewDebug::TypeLoweringScope {
263   TypeLoweringScope(CodeViewDebug &CVD) : CVD(CVD) { ++CVD.TypeEmissionLevel; }
264   ~TypeLoweringScope() {
265     // Don't decrement TypeEmissionLevel until after emitting deferred types, so
266     // inner TypeLoweringScopes don't attempt to emit deferred types.
267     if (CVD.TypeEmissionLevel == 1)
268       CVD.emitDeferredCompleteTypes();
269     --CVD.TypeEmissionLevel;
270   }
271   CodeViewDebug &CVD;
272 };
273 
274 static std::string getFullyQualifiedName(const DIScope *Ty) {
275   const DIScope *Scope = Ty->getScope().resolve();
276   return getFullyQualifiedName(Scope, getPrettyScopeName(Ty));
277 }
278 
279 TypeIndex CodeViewDebug::getScopeIndex(const DIScope *Scope) {
280   // No scope means global scope and that uses the zero index.
281   if (!Scope || isa<DIFile>(Scope))
282     return TypeIndex();
283 
284   assert(!isa<DIType>(Scope) && "shouldn't make a namespace scope for a type");
285 
286   // Check if we've already translated this scope.
287   auto I = TypeIndices.find({Scope, nullptr});
288   if (I != TypeIndices.end())
289     return I->second;
290 
291   // Build the fully qualified name of the scope.
292   std::string ScopeName = getFullyQualifiedName(Scope);
293   StringIdRecord SID(TypeIndex(), ScopeName);
294   auto TI = TypeTable.writeLeafType(SID);
295   return recordTypeIndexForDINode(Scope, TI);
296 }
297 
298 TypeIndex CodeViewDebug::getFuncIdForSubprogram(const DISubprogram *SP) {
299   assert(SP);
300 
301   // Check if we've already translated this subprogram.
302   auto I = TypeIndices.find({SP, nullptr});
303   if (I != TypeIndices.end())
304     return I->second;
305 
306   // The display name includes function template arguments. Drop them to match
307   // MSVC.
308   StringRef DisplayName = SP->getName().split('<').first;
309 
310   const DIScope *Scope = SP->getScope().resolve();
311   TypeIndex TI;
312   if (const auto *Class = dyn_cast_or_null<DICompositeType>(Scope)) {
313     // If the scope is a DICompositeType, then this must be a method. Member
314     // function types take some special handling, and require access to the
315     // subprogram.
316     TypeIndex ClassType = getTypeIndex(Class);
317     MemberFuncIdRecord MFuncId(ClassType, getMemberFunctionType(SP, Class),
318                                DisplayName);
319     TI = TypeTable.writeLeafType(MFuncId);
320   } else {
321     // Otherwise, this must be a free function.
322     TypeIndex ParentScope = getScopeIndex(Scope);
323     FuncIdRecord FuncId(ParentScope, getTypeIndex(SP->getType()), DisplayName);
324     TI = TypeTable.writeLeafType(FuncId);
325   }
326 
327   return recordTypeIndexForDINode(SP, TI);
328 }
329 
330 TypeIndex CodeViewDebug::getMemberFunctionType(const DISubprogram *SP,
331                                                const DICompositeType *Class) {
332   // Always use the method declaration as the key for the function type. The
333   // method declaration contains the this adjustment.
334   if (SP->getDeclaration())
335     SP = SP->getDeclaration();
336   assert(!SP->getDeclaration() && "should use declaration as key");
337 
338   // Key the MemberFunctionRecord into the map as {SP, Class}. It won't collide
339   // with the MemberFuncIdRecord, which is keyed in as {SP, nullptr}.
340   auto I = TypeIndices.find({SP, Class});
341   if (I != TypeIndices.end())
342     return I->second;
343 
344   // Make sure complete type info for the class is emitted *after* the member
345   // function type, as the complete class type is likely to reference this
346   // member function type.
347   TypeLoweringScope S(*this);
348   const bool IsStaticMethod = (SP->getFlags() & DINode::FlagStaticMember) != 0;
349   TypeIndex TI = lowerTypeMemberFunction(
350       SP->getType(), Class, SP->getThisAdjustment(), IsStaticMethod);
351   return recordTypeIndexForDINode(SP, TI, Class);
352 }
353 
354 TypeIndex CodeViewDebug::recordTypeIndexForDINode(const DINode *Node,
355                                                   TypeIndex TI,
356                                                   const DIType *ClassTy) {
357   auto InsertResult = TypeIndices.insert({{Node, ClassTy}, TI});
358   (void)InsertResult;
359   assert(InsertResult.second && "DINode was already assigned a type index");
360   return TI;
361 }
362 
363 unsigned CodeViewDebug::getPointerSizeInBytes() {
364   return MMI->getModule()->getDataLayout().getPointerSizeInBits() / 8;
365 }
366 
367 void CodeViewDebug::recordLocalVariable(LocalVariable &&Var,
368                                         const DILocation *InlinedAt) {
369   if (InlinedAt) {
370     // This variable was inlined. Associate it with the InlineSite.
371     const DISubprogram *Inlinee = Var.DIVar->getScope()->getSubprogram();
372     InlineSite &Site = getInlineSite(InlinedAt, Inlinee);
373     Site.InlinedLocals.emplace_back(Var);
374   } else {
375     // This variable goes in the main ProcSym.
376     CurFn->Locals.emplace_back(Var);
377   }
378 }
379 
380 static void addLocIfNotPresent(SmallVectorImpl<const DILocation *> &Locs,
381                                const DILocation *Loc) {
382   auto B = Locs.begin(), E = Locs.end();
383   if (std::find(B, E, Loc) == E)
384     Locs.push_back(Loc);
385 }
386 
387 void CodeViewDebug::maybeRecordLocation(const DebugLoc &DL,
388                                         const MachineFunction *MF) {
389   // Skip this instruction if it has the same location as the previous one.
390   if (!DL || DL == PrevInstLoc)
391     return;
392 
393   const DIScope *Scope = DL.get()->getScope();
394   if (!Scope)
395     return;
396 
397   // Skip this line if it is longer than the maximum we can record.
398   LineInfo LI(DL.getLine(), DL.getLine(), /*IsStatement=*/true);
399   if (LI.getStartLine() != DL.getLine() || LI.isAlwaysStepInto() ||
400       LI.isNeverStepInto())
401     return;
402 
403   ColumnInfo CI(DL.getCol(), /*EndColumn=*/0);
404   if (CI.getStartColumn() != DL.getCol())
405     return;
406 
407   if (!CurFn->HaveLineInfo)
408     CurFn->HaveLineInfo = true;
409   unsigned FileId = 0;
410   if (PrevInstLoc.get() && PrevInstLoc->getFile() == DL->getFile())
411     FileId = CurFn->LastFileId;
412   else
413     FileId = CurFn->LastFileId = maybeRecordFile(DL->getFile());
414   PrevInstLoc = DL;
415 
416   unsigned FuncId = CurFn->FuncId;
417   if (const DILocation *SiteLoc = DL->getInlinedAt()) {
418     const DILocation *Loc = DL.get();
419 
420     // If this location was actually inlined from somewhere else, give it the ID
421     // of the inline call site.
422     FuncId =
423         getInlineSite(SiteLoc, Loc->getScope()->getSubprogram()).SiteFuncId;
424 
425     // Ensure we have links in the tree of inline call sites.
426     bool FirstLoc = true;
427     while ((SiteLoc = Loc->getInlinedAt())) {
428       InlineSite &Site =
429           getInlineSite(SiteLoc, Loc->getScope()->getSubprogram());
430       if (!FirstLoc)
431         addLocIfNotPresent(Site.ChildSites, Loc);
432       FirstLoc = false;
433       Loc = SiteLoc;
434     }
435     addLocIfNotPresent(CurFn->ChildSites, Loc);
436   }
437 
438   OS.EmitCVLocDirective(FuncId, FileId, DL.getLine(), DL.getCol(),
439                         /*PrologueEnd=*/false, /*IsStmt=*/false,
440                         DL->getFilename(), SMLoc());
441 }
442 
443 void CodeViewDebug::emitCodeViewMagicVersion() {
444   OS.EmitValueToAlignment(4);
445   OS.AddComment("Debug section magic");
446   OS.EmitIntValue(COFF::DEBUG_SECTION_MAGIC, 4);
447 }
448 
449 void CodeViewDebug::endModule() {
450   if (!Asm || !MMI->hasDebugInfo())
451     return;
452 
453   assert(Asm != nullptr);
454 
455   // The COFF .debug$S section consists of several subsections, each starting
456   // with a 4-byte control code (e.g. 0xF1, 0xF2, etc) and then a 4-byte length
457   // of the payload followed by the payload itself.  The subsections are 4-byte
458   // aligned.
459 
460   // Use the generic .debug$S section, and make a subsection for all the inlined
461   // subprograms.
462   switchToDebugSectionForSymbol(nullptr);
463 
464   MCSymbol *CompilerInfo = beginCVSubsection(DebugSubsectionKind::Symbols);
465   emitCompilerInformation();
466   endCVSubsection(CompilerInfo);
467 
468   emitInlineeLinesSubsection();
469 
470   // Emit per-function debug information.
471   for (auto &P : FnDebugInfo)
472     if (!P.first->isDeclarationForLinker())
473       emitDebugInfoForFunction(P.first, *P.second);
474 
475   // Emit global variable debug information.
476   setCurrentSubprogram(nullptr);
477   emitDebugInfoForGlobals();
478 
479   // Emit retained types.
480   emitDebugInfoForRetainedTypes();
481 
482   // Switch back to the generic .debug$S section after potentially processing
483   // comdat symbol sections.
484   switchToDebugSectionForSymbol(nullptr);
485 
486   // Emit UDT records for any types used by global variables.
487   if (!GlobalUDTs.empty()) {
488     MCSymbol *SymbolsEnd = beginCVSubsection(DebugSubsectionKind::Symbols);
489     emitDebugInfoForUDTs(GlobalUDTs);
490     endCVSubsection(SymbolsEnd);
491   }
492 
493   // This subsection holds a file index to offset in string table table.
494   OS.AddComment("File index to string table offset subsection");
495   OS.EmitCVFileChecksumsDirective();
496 
497   // This subsection holds the string table.
498   OS.AddComment("String table");
499   OS.EmitCVStringTableDirective();
500 
501   // Emit type information and hashes last, so that any types we translate while
502   // emitting function info are included.
503   emitTypeInformation();
504 
505   if (EmitDebugGlobalHashes)
506     emitTypeGlobalHashes();
507 
508   clear();
509 }
510 
511 static void emitNullTerminatedSymbolName(MCStreamer &OS, StringRef S,
512     unsigned MaxFixedRecordLength = 0xF00) {
513   // The maximum CV record length is 0xFF00. Most of the strings we emit appear
514   // after a fixed length portion of the record. The fixed length portion should
515   // always be less than 0xF00 (3840) bytes, so truncate the string so that the
516   // overall record size is less than the maximum allowed.
517   SmallString<32> NullTerminatedString(
518       S.take_front(MaxRecordLength - MaxFixedRecordLength - 1));
519   NullTerminatedString.push_back('\0');
520   OS.EmitBytes(NullTerminatedString);
521 }
522 
523 void CodeViewDebug::emitTypeInformation() {
524   if (TypeTable.empty())
525     return;
526 
527   // Start the .debug$T section with 0x4.
528   OS.SwitchSection(Asm->getObjFileLowering().getCOFFDebugTypesSection());
529   emitCodeViewMagicVersion();
530 
531   SmallString<8> CommentPrefix;
532   if (OS.isVerboseAsm()) {
533     CommentPrefix += '\t';
534     CommentPrefix += Asm->MAI->getCommentString();
535     CommentPrefix += ' ';
536   }
537 
538   TypeTableCollection Table(TypeTable.records());
539   Optional<TypeIndex> B = Table.getFirst();
540   while (B) {
541     // This will fail if the record data is invalid.
542     CVType Record = Table.getType(*B);
543 
544     if (OS.isVerboseAsm()) {
545       // Emit a block comment describing the type record for readability.
546       SmallString<512> CommentBlock;
547       raw_svector_ostream CommentOS(CommentBlock);
548       ScopedPrinter SP(CommentOS);
549       SP.setPrefix(CommentPrefix);
550       TypeDumpVisitor TDV(Table, &SP, false);
551 
552       Error E = codeview::visitTypeRecord(Record, *B, TDV);
553       if (E) {
554         logAllUnhandledErrors(std::move(E), errs(), "error: ");
555         llvm_unreachable("produced malformed type record");
556       }
557       // emitRawComment will insert its own tab and comment string before
558       // the first line, so strip off our first one. It also prints its own
559       // newline.
560       OS.emitRawComment(
561           CommentOS.str().drop_front(CommentPrefix.size() - 1).rtrim());
562     }
563     OS.EmitBinaryData(Record.str_data());
564     B = Table.getNext(*B);
565   }
566 }
567 
568 void CodeViewDebug::emitTypeGlobalHashes() {
569   if (TypeTable.empty())
570     return;
571 
572   // Start the .debug$H section with the version and hash algorithm, currently
573   // hardcoded to version 0, SHA1.
574   OS.SwitchSection(Asm->getObjFileLowering().getCOFFGlobalTypeHashesSection());
575 
576   OS.EmitValueToAlignment(4);
577   OS.AddComment("Magic");
578   OS.EmitIntValue(COFF::DEBUG_HASHES_SECTION_MAGIC, 4);
579   OS.AddComment("Section Version");
580   OS.EmitIntValue(0, 2);
581   OS.AddComment("Hash Algorithm");
582   OS.EmitIntValue(uint16_t(GlobalTypeHashAlg::SHA1), 2);
583 
584   TypeIndex TI(TypeIndex::FirstNonSimpleIndex);
585   for (const auto &GHR : TypeTable.hashes()) {
586     if (OS.isVerboseAsm()) {
587       // Emit an EOL-comment describing which TypeIndex this hash corresponds
588       // to, as well as the stringified SHA1 hash.
589       SmallString<32> Comment;
590       raw_svector_ostream CommentOS(Comment);
591       CommentOS << formatv("{0:X+} [{1}]", TI.getIndex(), GHR);
592       OS.AddComment(Comment);
593       ++TI;
594     }
595     assert(GHR.Hash.size() % 20 == 0);
596     StringRef S(reinterpret_cast<const char *>(GHR.Hash.data()),
597                 GHR.Hash.size());
598     OS.EmitBinaryData(S);
599   }
600 }
601 
602 static SourceLanguage MapDWLangToCVLang(unsigned DWLang) {
603   switch (DWLang) {
604   case dwarf::DW_LANG_C:
605   case dwarf::DW_LANG_C89:
606   case dwarf::DW_LANG_C99:
607   case dwarf::DW_LANG_C11:
608   case dwarf::DW_LANG_ObjC:
609     return SourceLanguage::C;
610   case dwarf::DW_LANG_C_plus_plus:
611   case dwarf::DW_LANG_C_plus_plus_03:
612   case dwarf::DW_LANG_C_plus_plus_11:
613   case dwarf::DW_LANG_C_plus_plus_14:
614     return SourceLanguage::Cpp;
615   case dwarf::DW_LANG_Fortran77:
616   case dwarf::DW_LANG_Fortran90:
617   case dwarf::DW_LANG_Fortran03:
618   case dwarf::DW_LANG_Fortran08:
619     return SourceLanguage::Fortran;
620   case dwarf::DW_LANG_Pascal83:
621     return SourceLanguage::Pascal;
622   case dwarf::DW_LANG_Cobol74:
623   case dwarf::DW_LANG_Cobol85:
624     return SourceLanguage::Cobol;
625   case dwarf::DW_LANG_Java:
626     return SourceLanguage::Java;
627   case dwarf::DW_LANG_D:
628     return SourceLanguage::D;
629   default:
630     // There's no CodeView representation for this language, and CV doesn't
631     // have an "unknown" option for the language field, so we'll use MASM,
632     // as it's very low level.
633     return SourceLanguage::Masm;
634   }
635 }
636 
637 namespace {
638 struct Version {
639   int Part[4];
640 };
641 } // end anonymous namespace
642 
643 // Takes a StringRef like "clang 4.0.0.0 (other nonsense 123)" and parses out
644 // the version number.
645 static Version parseVersion(StringRef Name) {
646   Version V = {{0}};
647   int N = 0;
648   for (const char C : Name) {
649     if (isdigit(C)) {
650       V.Part[N] *= 10;
651       V.Part[N] += C - '0';
652     } else if (C == '.') {
653       ++N;
654       if (N >= 4)
655         return V;
656     } else if (N > 0)
657       return V;
658   }
659   return V;
660 }
661 
662 static CPUType mapArchToCVCPUType(Triple::ArchType Type) {
663   switch (Type) {
664   case Triple::ArchType::x86:
665     return CPUType::Pentium3;
666   case Triple::ArchType::x86_64:
667     return CPUType::X64;
668   case Triple::ArchType::thumb:
669     return CPUType::Thumb;
670   case Triple::ArchType::aarch64:
671     return CPUType::ARM64;
672   default:
673     report_fatal_error("target architecture doesn't map to a CodeView CPUType");
674   }
675 }
676 
677 void CodeViewDebug::emitCompilerInformation() {
678   MCContext &Context = MMI->getContext();
679   MCSymbol *CompilerBegin = Context.createTempSymbol(),
680            *CompilerEnd = Context.createTempSymbol();
681   OS.AddComment("Record length");
682   OS.emitAbsoluteSymbolDiff(CompilerEnd, CompilerBegin, 2);
683   OS.EmitLabel(CompilerBegin);
684   OS.AddComment("Record kind: S_COMPILE3");
685   OS.EmitIntValue(SymbolKind::S_COMPILE3, 2);
686   uint32_t Flags = 0;
687 
688   NamedMDNode *CUs = MMI->getModule()->getNamedMetadata("llvm.dbg.cu");
689   const MDNode *Node = *CUs->operands().begin();
690   const auto *CU = cast<DICompileUnit>(Node);
691 
692   // The low byte of the flags indicates the source language.
693   Flags = MapDWLangToCVLang(CU->getSourceLanguage());
694   // TODO:  Figure out which other flags need to be set.
695 
696   OS.AddComment("Flags and language");
697   OS.EmitIntValue(Flags, 4);
698 
699   OS.AddComment("CPUType");
700   CPUType CPU =
701       mapArchToCVCPUType(Triple(MMI->getModule()->getTargetTriple()).getArch());
702   OS.EmitIntValue(static_cast<uint64_t>(CPU), 2);
703 
704   StringRef CompilerVersion = CU->getProducer();
705   Version FrontVer = parseVersion(CompilerVersion);
706   OS.AddComment("Frontend version");
707   for (int N = 0; N < 4; ++N)
708     OS.EmitIntValue(FrontVer.Part[N], 2);
709 
710   // Some Microsoft tools, like Binscope, expect a backend version number of at
711   // least 8.something, so we'll coerce the LLVM version into a form that
712   // guarantees it'll be big enough without really lying about the version.
713   int Major = 1000 * LLVM_VERSION_MAJOR +
714               10 * LLVM_VERSION_MINOR +
715               LLVM_VERSION_PATCH;
716   // Clamp it for builds that use unusually large version numbers.
717   Major = std::min<int>(Major, std::numeric_limits<uint16_t>::max());
718   Version BackVer = {{ Major, 0, 0, 0 }};
719   OS.AddComment("Backend version");
720   for (int N = 0; N < 4; ++N)
721     OS.EmitIntValue(BackVer.Part[N], 2);
722 
723   OS.AddComment("Null-terminated compiler version string");
724   emitNullTerminatedSymbolName(OS, CompilerVersion);
725 
726   OS.EmitLabel(CompilerEnd);
727 }
728 
729 void CodeViewDebug::emitInlineeLinesSubsection() {
730   if (InlinedSubprograms.empty())
731     return;
732 
733   OS.AddComment("Inlinee lines subsection");
734   MCSymbol *InlineEnd = beginCVSubsection(DebugSubsectionKind::InlineeLines);
735 
736   // We emit the checksum info for files.  This is used by debuggers to
737   // determine if a pdb matches the source before loading it.  Visual Studio,
738   // for instance, will display a warning that the breakpoints are not valid if
739   // the pdb does not match the source.
740   OS.AddComment("Inlinee lines signature");
741   OS.EmitIntValue(unsigned(InlineeLinesSignature::Normal), 4);
742 
743   for (const DISubprogram *SP : InlinedSubprograms) {
744     assert(TypeIndices.count({SP, nullptr}));
745     TypeIndex InlineeIdx = TypeIndices[{SP, nullptr}];
746 
747     OS.AddBlankLine();
748     unsigned FileId = maybeRecordFile(SP->getFile());
749     OS.AddComment("Inlined function " + SP->getName() + " starts at " +
750                   SP->getFilename() + Twine(':') + Twine(SP->getLine()));
751     OS.AddBlankLine();
752     OS.AddComment("Type index of inlined function");
753     OS.EmitIntValue(InlineeIdx.getIndex(), 4);
754     OS.AddComment("Offset into filechecksum table");
755     OS.EmitCVFileChecksumOffsetDirective(FileId);
756     OS.AddComment("Starting line number");
757     OS.EmitIntValue(SP->getLine(), 4);
758   }
759 
760   endCVSubsection(InlineEnd);
761 }
762 
763 void CodeViewDebug::emitInlinedCallSite(const FunctionInfo &FI,
764                                         const DILocation *InlinedAt,
765                                         const InlineSite &Site) {
766   MCSymbol *InlineBegin = MMI->getContext().createTempSymbol(),
767            *InlineEnd = MMI->getContext().createTempSymbol();
768 
769   assert(TypeIndices.count({Site.Inlinee, nullptr}));
770   TypeIndex InlineeIdx = TypeIndices[{Site.Inlinee, nullptr}];
771 
772   // SymbolRecord
773   OS.AddComment("Record length");
774   OS.emitAbsoluteSymbolDiff(InlineEnd, InlineBegin, 2);   // RecordLength
775   OS.EmitLabel(InlineBegin);
776   OS.AddComment("Record kind: S_INLINESITE");
777   OS.EmitIntValue(SymbolKind::S_INLINESITE, 2); // RecordKind
778 
779   OS.AddComment("PtrParent");
780   OS.EmitIntValue(0, 4);
781   OS.AddComment("PtrEnd");
782   OS.EmitIntValue(0, 4);
783   OS.AddComment("Inlinee type index");
784   OS.EmitIntValue(InlineeIdx.getIndex(), 4);
785 
786   unsigned FileId = maybeRecordFile(Site.Inlinee->getFile());
787   unsigned StartLineNum = Site.Inlinee->getLine();
788 
789   OS.EmitCVInlineLinetableDirective(Site.SiteFuncId, FileId, StartLineNum,
790                                     FI.Begin, FI.End);
791 
792   OS.EmitLabel(InlineEnd);
793 
794   emitLocalVariableList(Site.InlinedLocals);
795 
796   // Recurse on child inlined call sites before closing the scope.
797   for (const DILocation *ChildSite : Site.ChildSites) {
798     auto I = FI.InlineSites.find(ChildSite);
799     assert(I != FI.InlineSites.end() &&
800            "child site not in function inline site map");
801     emitInlinedCallSite(FI, ChildSite, I->second);
802   }
803 
804   // Close the scope.
805   OS.AddComment("Record length");
806   OS.EmitIntValue(2, 2);                                  // RecordLength
807   OS.AddComment("Record kind: S_INLINESITE_END");
808   OS.EmitIntValue(SymbolKind::S_INLINESITE_END, 2); // RecordKind
809 }
810 
811 void CodeViewDebug::switchToDebugSectionForSymbol(const MCSymbol *GVSym) {
812   // If we have a symbol, it may be in a section that is COMDAT. If so, find the
813   // comdat key. A section may be comdat because of -ffunction-sections or
814   // because it is comdat in the IR.
815   MCSectionCOFF *GVSec =
816       GVSym ? dyn_cast<MCSectionCOFF>(&GVSym->getSection()) : nullptr;
817   const MCSymbol *KeySym = GVSec ? GVSec->getCOMDATSymbol() : nullptr;
818 
819   MCSectionCOFF *DebugSec = cast<MCSectionCOFF>(
820       Asm->getObjFileLowering().getCOFFDebugSymbolsSection());
821   DebugSec = OS.getContext().getAssociativeCOFFSection(DebugSec, KeySym);
822 
823   OS.SwitchSection(DebugSec);
824 
825   // Emit the magic version number if this is the first time we've switched to
826   // this section.
827   if (ComdatDebugSections.insert(DebugSec).second)
828     emitCodeViewMagicVersion();
829 }
830 
831 void CodeViewDebug::emitDebugInfoForFunction(const Function *GV,
832                                              FunctionInfo &FI) {
833   // For each function there is a separate subsection which holds the PC to
834   // file:line table.
835   const MCSymbol *Fn = Asm->getSymbol(GV);
836   assert(Fn);
837 
838   // Switch to the to a comdat section, if appropriate.
839   switchToDebugSectionForSymbol(Fn);
840 
841   std::string FuncName;
842   auto *SP = GV->getSubprogram();
843   assert(SP);
844   setCurrentSubprogram(SP);
845 
846   // If we have a display name, build the fully qualified name by walking the
847   // chain of scopes.
848   if (!SP->getName().empty())
849     FuncName =
850         getFullyQualifiedName(SP->getScope().resolve(), SP->getName());
851 
852   // If our DISubprogram name is empty, use the mangled name.
853   if (FuncName.empty())
854     FuncName = GlobalValue::dropLLVMManglingEscape(GV->getName());
855 
856   // Emit FPO data, but only on 32-bit x86. No other platforms use it.
857   if (Triple(MMI->getModule()->getTargetTriple()).getArch() == Triple::x86)
858     OS.EmitCVFPOData(Fn);
859 
860   // Emit a symbol subsection, required by VS2012+ to find function boundaries.
861   OS.AddComment("Symbol subsection for " + Twine(FuncName));
862   MCSymbol *SymbolsEnd = beginCVSubsection(DebugSubsectionKind::Symbols);
863   {
864     MCSymbol *ProcRecordBegin = MMI->getContext().createTempSymbol(),
865              *ProcRecordEnd = MMI->getContext().createTempSymbol();
866     OS.AddComment("Record length");
867     OS.emitAbsoluteSymbolDiff(ProcRecordEnd, ProcRecordBegin, 2);
868     OS.EmitLabel(ProcRecordBegin);
869 
870     if (GV->hasLocalLinkage()) {
871       OS.AddComment("Record kind: S_LPROC32_ID");
872       OS.EmitIntValue(unsigned(SymbolKind::S_LPROC32_ID), 2);
873     } else {
874       OS.AddComment("Record kind: S_GPROC32_ID");
875       OS.EmitIntValue(unsigned(SymbolKind::S_GPROC32_ID), 2);
876     }
877 
878     // These fields are filled in by tools like CVPACK which run after the fact.
879     OS.AddComment("PtrParent");
880     OS.EmitIntValue(0, 4);
881     OS.AddComment("PtrEnd");
882     OS.EmitIntValue(0, 4);
883     OS.AddComment("PtrNext");
884     OS.EmitIntValue(0, 4);
885     // This is the important bit that tells the debugger where the function
886     // code is located and what's its size:
887     OS.AddComment("Code size");
888     OS.emitAbsoluteSymbolDiff(FI.End, Fn, 4);
889     OS.AddComment("Offset after prologue");
890     OS.EmitIntValue(0, 4);
891     OS.AddComment("Offset before epilogue");
892     OS.EmitIntValue(0, 4);
893     OS.AddComment("Function type index");
894     OS.EmitIntValue(getFuncIdForSubprogram(GV->getSubprogram()).getIndex(), 4);
895     OS.AddComment("Function section relative address");
896     OS.EmitCOFFSecRel32(Fn, /*Offset=*/0);
897     OS.AddComment("Function section index");
898     OS.EmitCOFFSectionIndex(Fn);
899     OS.AddComment("Flags");
900     OS.EmitIntValue(0, 1);
901     // Emit the function display name as a null-terminated string.
902     OS.AddComment("Function name");
903     // Truncate the name so we won't overflow the record length field.
904     emitNullTerminatedSymbolName(OS, FuncName);
905     OS.EmitLabel(ProcRecordEnd);
906 
907     emitLocalVariableList(FI.Locals);
908 
909     // Emit inlined call site information. Only emit functions inlined directly
910     // into the parent function. We'll emit the other sites recursively as part
911     // of their parent inline site.
912     for (const DILocation *InlinedAt : FI.ChildSites) {
913       auto I = FI.InlineSites.find(InlinedAt);
914       assert(I != FI.InlineSites.end() &&
915              "child site not in function inline site map");
916       emitInlinedCallSite(FI, InlinedAt, I->second);
917     }
918 
919     for (auto Annot : FI.Annotations) {
920       MCSymbol *Label = Annot.first;
921       MDTuple *Strs = cast<MDTuple>(Annot.second);
922       MCSymbol *AnnotBegin = MMI->getContext().createTempSymbol(),
923                *AnnotEnd = MMI->getContext().createTempSymbol();
924       OS.AddComment("Record length");
925       OS.emitAbsoluteSymbolDiff(AnnotEnd, AnnotBegin, 2);
926       OS.EmitLabel(AnnotBegin);
927       OS.AddComment("Record kind: S_ANNOTATION");
928       OS.EmitIntValue(SymbolKind::S_ANNOTATION, 2);
929       OS.EmitCOFFSecRel32(Label, /*Offset=*/0);
930       // FIXME: Make sure we don't overflow the max record size.
931       OS.EmitCOFFSectionIndex(Label);
932       OS.EmitIntValue(Strs->getNumOperands(), 2);
933       for (Metadata *MD : Strs->operands()) {
934         // MDStrings are null terminated, so we can do EmitBytes and get the
935         // nice .asciz directive.
936         StringRef Str = cast<MDString>(MD)->getString();
937         assert(Str.data()[Str.size()] == '\0' && "non-nullterminated MDString");
938         OS.EmitBytes(StringRef(Str.data(), Str.size() + 1));
939       }
940       OS.EmitLabel(AnnotEnd);
941     }
942 
943     if (SP != nullptr)
944       emitDebugInfoForUDTs(LocalUDTs);
945 
946     // We're done with this function.
947     OS.AddComment("Record length");
948     OS.EmitIntValue(0x0002, 2);
949     OS.AddComment("Record kind: S_PROC_ID_END");
950     OS.EmitIntValue(unsigned(SymbolKind::S_PROC_ID_END), 2);
951   }
952   endCVSubsection(SymbolsEnd);
953 
954   // We have an assembler directive that takes care of the whole line table.
955   OS.EmitCVLinetableDirective(FI.FuncId, Fn, FI.End);
956 }
957 
958 CodeViewDebug::LocalVarDefRange
959 CodeViewDebug::createDefRangeMem(uint16_t CVRegister, int Offset) {
960   LocalVarDefRange DR;
961   DR.InMemory = -1;
962   DR.DataOffset = Offset;
963   assert(DR.DataOffset == Offset && "truncation");
964   DR.IsSubfield = 0;
965   DR.StructOffset = 0;
966   DR.CVRegister = CVRegister;
967   return DR;
968 }
969 
970 CodeViewDebug::LocalVarDefRange
971 CodeViewDebug::createDefRangeGeneral(uint16_t CVRegister, bool InMemory,
972                                      int Offset, bool IsSubfield,
973                                      uint16_t StructOffset) {
974   LocalVarDefRange DR;
975   DR.InMemory = InMemory;
976   DR.DataOffset = Offset;
977   DR.IsSubfield = IsSubfield;
978   DR.StructOffset = StructOffset;
979   DR.CVRegister = CVRegister;
980   return DR;
981 }
982 
983 void CodeViewDebug::collectVariableInfoFromMFTable(
984     DenseSet<InlinedVariable> &Processed) {
985   const MachineFunction &MF = *Asm->MF;
986   const TargetSubtargetInfo &TSI = MF.getSubtarget();
987   const TargetFrameLowering *TFI = TSI.getFrameLowering();
988   const TargetRegisterInfo *TRI = TSI.getRegisterInfo();
989 
990   for (const MachineFunction::VariableDbgInfo &VI : MF.getVariableDbgInfo()) {
991     if (!VI.Var)
992       continue;
993     assert(VI.Var->isValidLocationForIntrinsic(VI.Loc) &&
994            "Expected inlined-at fields to agree");
995 
996     Processed.insert(InlinedVariable(VI.Var, VI.Loc->getInlinedAt()));
997     LexicalScope *Scope = LScopes.findLexicalScope(VI.Loc);
998 
999     // If variable scope is not found then skip this variable.
1000     if (!Scope)
1001       continue;
1002 
1003     // If the variable has an attached offset expression, extract it.
1004     // FIXME: Try to handle DW_OP_deref as well.
1005     int64_t ExprOffset = 0;
1006     if (VI.Expr)
1007       if (!VI.Expr->extractIfOffset(ExprOffset))
1008         continue;
1009 
1010     // Get the frame register used and the offset.
1011     unsigned FrameReg = 0;
1012     int FrameOffset = TFI->getFrameIndexReference(*Asm->MF, VI.Slot, FrameReg);
1013     uint16_t CVReg = TRI->getCodeViewRegNum(FrameReg);
1014 
1015     // Calculate the label ranges.
1016     LocalVarDefRange DefRange =
1017         createDefRangeMem(CVReg, FrameOffset + ExprOffset);
1018     for (const InsnRange &Range : Scope->getRanges()) {
1019       const MCSymbol *Begin = getLabelBeforeInsn(Range.first);
1020       const MCSymbol *End = getLabelAfterInsn(Range.second);
1021       End = End ? End : Asm->getFunctionEnd();
1022       DefRange.Ranges.emplace_back(Begin, End);
1023     }
1024 
1025     LocalVariable Var;
1026     Var.DIVar = VI.Var;
1027     Var.DefRanges.emplace_back(std::move(DefRange));
1028     recordLocalVariable(std::move(Var), VI.Loc->getInlinedAt());
1029   }
1030 }
1031 
1032 static bool canUseReferenceType(const DbgVariableLocation &Loc) {
1033   return !Loc.LoadChain.empty() && Loc.LoadChain.back() == 0;
1034 }
1035 
1036 static bool needsReferenceType(const DbgVariableLocation &Loc) {
1037   return Loc.LoadChain.size() == 2 && Loc.LoadChain.back() == 0;
1038 }
1039 
1040 void CodeViewDebug::calculateRanges(
1041     LocalVariable &Var, const DbgValueHistoryMap::InstrRanges &Ranges) {
1042   const TargetRegisterInfo *TRI = Asm->MF->getSubtarget().getRegisterInfo();
1043 
1044   // Calculate the definition ranges.
1045   for (auto I = Ranges.begin(), E = Ranges.end(); I != E; ++I) {
1046     const InsnRange &Range = *I;
1047     const MachineInstr *DVInst = Range.first;
1048     assert(DVInst->isDebugValue() && "Invalid History entry");
1049     // FIXME: Find a way to represent constant variables, since they are
1050     // relatively common.
1051     Optional<DbgVariableLocation> Location =
1052         DbgVariableLocation::extractFromMachineInstruction(*DVInst);
1053     if (!Location)
1054       continue;
1055 
1056     // CodeView can only express variables in register and variables in memory
1057     // at a constant offset from a register. However, for variables passed
1058     // indirectly by pointer, it is common for that pointer to be spilled to a
1059     // stack location. For the special case of one offseted load followed by a
1060     // zero offset load (a pointer spilled to the stack), we change the type of
1061     // the local variable from a value type to a reference type. This tricks the
1062     // debugger into doing the load for us.
1063     if (Var.UseReferenceType) {
1064       // We're using a reference type. Drop the last zero offset load.
1065       if (canUseReferenceType(*Location))
1066         Location->LoadChain.pop_back();
1067       else
1068         continue;
1069     } else if (needsReferenceType(*Location)) {
1070       // This location can't be expressed without switching to a reference type.
1071       // Start over using that.
1072       Var.UseReferenceType = true;
1073       Var.DefRanges.clear();
1074       calculateRanges(Var, Ranges);
1075       return;
1076     }
1077 
1078     // We can only handle a register or an offseted load of a register.
1079     if (Location->Register == 0 || Location->LoadChain.size() > 1)
1080       continue;
1081     {
1082       LocalVarDefRange DR;
1083       DR.CVRegister = TRI->getCodeViewRegNum(Location->Register);
1084       DR.InMemory = !Location->LoadChain.empty();
1085       DR.DataOffset =
1086           !Location->LoadChain.empty() ? Location->LoadChain.back() : 0;
1087       if (Location->FragmentInfo) {
1088         DR.IsSubfield = true;
1089         DR.StructOffset = Location->FragmentInfo->OffsetInBits / 8;
1090       } else {
1091         DR.IsSubfield = false;
1092         DR.StructOffset = 0;
1093       }
1094 
1095       if (Var.DefRanges.empty() ||
1096           Var.DefRanges.back().isDifferentLocation(DR)) {
1097         Var.DefRanges.emplace_back(std::move(DR));
1098       }
1099     }
1100 
1101     // Compute the label range.
1102     const MCSymbol *Begin = getLabelBeforeInsn(Range.first);
1103     const MCSymbol *End = getLabelAfterInsn(Range.second);
1104     if (!End) {
1105       // This range is valid until the next overlapping bitpiece. In the
1106       // common case, ranges will not be bitpieces, so they will overlap.
1107       auto J = std::next(I);
1108       const DIExpression *DIExpr = DVInst->getDebugExpression();
1109       while (J != E &&
1110              !DIExpr->fragmentsOverlap(J->first->getDebugExpression()))
1111         ++J;
1112       if (J != E)
1113         End = getLabelBeforeInsn(J->first);
1114       else
1115         End = Asm->getFunctionEnd();
1116     }
1117 
1118     // If the last range end is our begin, just extend the last range.
1119     // Otherwise make a new range.
1120     SmallVectorImpl<std::pair<const MCSymbol *, const MCSymbol *>> &R =
1121         Var.DefRanges.back().Ranges;
1122     if (!R.empty() && R.back().second == Begin)
1123       R.back().second = End;
1124     else
1125       R.emplace_back(Begin, End);
1126 
1127     // FIXME: Do more range combining.
1128   }
1129 }
1130 
1131 void CodeViewDebug::collectVariableInfo(const DISubprogram *SP) {
1132   DenseSet<InlinedVariable> Processed;
1133   // Grab the variable info that was squirreled away in the MMI side-table.
1134   collectVariableInfoFromMFTable(Processed);
1135 
1136   for (const auto &I : DbgValues) {
1137     InlinedVariable IV = I.first;
1138     if (Processed.count(IV))
1139       continue;
1140     const DILocalVariable *DIVar = IV.first;
1141     const DILocation *InlinedAt = IV.second;
1142 
1143     // Instruction ranges, specifying where IV is accessible.
1144     const auto &Ranges = I.second;
1145 
1146     LexicalScope *Scope = nullptr;
1147     if (InlinedAt)
1148       Scope = LScopes.findInlinedScope(DIVar->getScope(), InlinedAt);
1149     else
1150       Scope = LScopes.findLexicalScope(DIVar->getScope());
1151     // If variable scope is not found then skip this variable.
1152     if (!Scope)
1153       continue;
1154 
1155     LocalVariable Var;
1156     Var.DIVar = DIVar;
1157 
1158     calculateRanges(Var, Ranges);
1159     recordLocalVariable(std::move(Var), InlinedAt);
1160   }
1161 }
1162 
1163 void CodeViewDebug::beginFunctionImpl(const MachineFunction *MF) {
1164   const Function &GV = MF->getFunction();
1165   auto Insertion = FnDebugInfo.insert({&GV, llvm::make_unique<FunctionInfo>()});
1166   assert(!Insertion.second && "emitting function twice");
1167   CurFn = Insertion.first->second.get();
1168   CurFn->FuncId = NextFuncId++;
1169   CurFn->Begin = Asm->getFunctionBegin();
1170 
1171   OS.EmitCVFuncIdDirective(CurFn->FuncId);
1172 
1173   // Find the end of the function prolog.  First known non-DBG_VALUE and
1174   // non-frame setup location marks the beginning of the function body.
1175   // FIXME: is there a simpler a way to do this? Can we just search
1176   // for the first instruction of the function, not the last of the prolog?
1177   DebugLoc PrologEndLoc;
1178   bool EmptyPrologue = true;
1179   for (const auto &MBB : *MF) {
1180     for (const auto &MI : MBB) {
1181       if (!MI.isMetaInstruction() && !MI.getFlag(MachineInstr::FrameSetup) &&
1182           MI.getDebugLoc()) {
1183         PrologEndLoc = MI.getDebugLoc();
1184         break;
1185       } else if (!MI.isMetaInstruction()) {
1186         EmptyPrologue = false;
1187       }
1188     }
1189   }
1190 
1191   // Record beginning of function if we have a non-empty prologue.
1192   if (PrologEndLoc && !EmptyPrologue) {
1193     DebugLoc FnStartDL = PrologEndLoc.getFnDebugLoc();
1194     maybeRecordLocation(FnStartDL, MF);
1195   }
1196 }
1197 
1198 static bool shouldEmitUdt(const DIType *T) {
1199   if (!T)
1200     return false;
1201 
1202   // MSVC does not emit UDTs for typedefs that are scoped to classes.
1203   if (T->getTag() == dwarf::DW_TAG_typedef) {
1204     if (DIScope *Scope = T->getScope().resolve()) {
1205       switch (Scope->getTag()) {
1206       case dwarf::DW_TAG_structure_type:
1207       case dwarf::DW_TAG_class_type:
1208       case dwarf::DW_TAG_union_type:
1209         return false;
1210       }
1211     }
1212   }
1213 
1214   while (true) {
1215     if (!T || T->isForwardDecl())
1216       return false;
1217 
1218     const DIDerivedType *DT = dyn_cast<DIDerivedType>(T);
1219     if (!DT)
1220       return true;
1221     T = DT->getBaseType().resolve();
1222   }
1223   return true;
1224 }
1225 
1226 void CodeViewDebug::addToUDTs(const DIType *Ty) {
1227   // Don't record empty UDTs.
1228   if (Ty->getName().empty())
1229     return;
1230   if (!shouldEmitUdt(Ty))
1231     return;
1232 
1233   SmallVector<StringRef, 5> QualifiedNameComponents;
1234   const DISubprogram *ClosestSubprogram = getQualifiedNameComponents(
1235       Ty->getScope().resolve(), QualifiedNameComponents);
1236 
1237   std::string FullyQualifiedName =
1238       getQualifiedName(QualifiedNameComponents, getPrettyScopeName(Ty));
1239 
1240   if (ClosestSubprogram == nullptr) {
1241     GlobalUDTs.emplace_back(std::move(FullyQualifiedName), Ty);
1242   } else if (ClosestSubprogram == CurrentSubprogram) {
1243     LocalUDTs.emplace_back(std::move(FullyQualifiedName), Ty);
1244   }
1245 
1246   // TODO: What if the ClosestSubprogram is neither null or the current
1247   // subprogram?  Currently, the UDT just gets dropped on the floor.
1248   //
1249   // The current behavior is not desirable.  To get maximal fidelity, we would
1250   // need to perform all type translation before beginning emission of .debug$S
1251   // and then make LocalUDTs a member of FunctionInfo
1252 }
1253 
1254 TypeIndex CodeViewDebug::lowerType(const DIType *Ty, const DIType *ClassTy) {
1255   // Generic dispatch for lowering an unknown type.
1256   switch (Ty->getTag()) {
1257   case dwarf::DW_TAG_array_type:
1258     return lowerTypeArray(cast<DICompositeType>(Ty));
1259   case dwarf::DW_TAG_typedef:
1260     return lowerTypeAlias(cast<DIDerivedType>(Ty));
1261   case dwarf::DW_TAG_base_type:
1262     return lowerTypeBasic(cast<DIBasicType>(Ty));
1263   case dwarf::DW_TAG_pointer_type:
1264     if (cast<DIDerivedType>(Ty)->getName() == "__vtbl_ptr_type")
1265       return lowerTypeVFTableShape(cast<DIDerivedType>(Ty));
1266     LLVM_FALLTHROUGH;
1267   case dwarf::DW_TAG_reference_type:
1268   case dwarf::DW_TAG_rvalue_reference_type:
1269     return lowerTypePointer(cast<DIDerivedType>(Ty));
1270   case dwarf::DW_TAG_ptr_to_member_type:
1271     return lowerTypeMemberPointer(cast<DIDerivedType>(Ty));
1272   case dwarf::DW_TAG_restrict_type:
1273   case dwarf::DW_TAG_const_type:
1274   case dwarf::DW_TAG_volatile_type:
1275   // TODO: add support for DW_TAG_atomic_type here
1276     return lowerTypeModifier(cast<DIDerivedType>(Ty));
1277   case dwarf::DW_TAG_subroutine_type:
1278     if (ClassTy) {
1279       // The member function type of a member function pointer has no
1280       // ThisAdjustment.
1281       return lowerTypeMemberFunction(cast<DISubroutineType>(Ty), ClassTy,
1282                                      /*ThisAdjustment=*/0,
1283                                      /*IsStaticMethod=*/false);
1284     }
1285     return lowerTypeFunction(cast<DISubroutineType>(Ty));
1286   case dwarf::DW_TAG_enumeration_type:
1287     return lowerTypeEnum(cast<DICompositeType>(Ty));
1288   case dwarf::DW_TAG_class_type:
1289   case dwarf::DW_TAG_structure_type:
1290     return lowerTypeClass(cast<DICompositeType>(Ty));
1291   case dwarf::DW_TAG_union_type:
1292     return lowerTypeUnion(cast<DICompositeType>(Ty));
1293   case dwarf::DW_TAG_unspecified_type:
1294     return TypeIndex::None();
1295   default:
1296     // Use the null type index.
1297     return TypeIndex();
1298   }
1299 }
1300 
1301 TypeIndex CodeViewDebug::lowerTypeAlias(const DIDerivedType *Ty) {
1302   DITypeRef UnderlyingTypeRef = Ty->getBaseType();
1303   TypeIndex UnderlyingTypeIndex = getTypeIndex(UnderlyingTypeRef);
1304   StringRef TypeName = Ty->getName();
1305 
1306   addToUDTs(Ty);
1307 
1308   if (UnderlyingTypeIndex == TypeIndex(SimpleTypeKind::Int32Long) &&
1309       TypeName == "HRESULT")
1310     return TypeIndex(SimpleTypeKind::HResult);
1311   if (UnderlyingTypeIndex == TypeIndex(SimpleTypeKind::UInt16Short) &&
1312       TypeName == "wchar_t")
1313     return TypeIndex(SimpleTypeKind::WideCharacter);
1314 
1315   return UnderlyingTypeIndex;
1316 }
1317 
1318 TypeIndex CodeViewDebug::lowerTypeArray(const DICompositeType *Ty) {
1319   DITypeRef ElementTypeRef = Ty->getBaseType();
1320   TypeIndex ElementTypeIndex = getTypeIndex(ElementTypeRef);
1321   // IndexType is size_t, which depends on the bitness of the target.
1322   TypeIndex IndexType = getPointerSizeInBytes() == 8
1323                             ? TypeIndex(SimpleTypeKind::UInt64Quad)
1324                             : TypeIndex(SimpleTypeKind::UInt32Long);
1325 
1326   uint64_t ElementSize = getBaseTypeSize(ElementTypeRef) / 8;
1327 
1328   // Add subranges to array type.
1329   DINodeArray Elements = Ty->getElements();
1330   for (int i = Elements.size() - 1; i >= 0; --i) {
1331     const DINode *Element = Elements[i];
1332     assert(Element->getTag() == dwarf::DW_TAG_subrange_type);
1333 
1334     const DISubrange *Subrange = cast<DISubrange>(Element);
1335     assert(Subrange->getLowerBound() == 0 &&
1336            "codeview doesn't support subranges with lower bounds");
1337     int64_t Count = -1;
1338     if (auto *CI = Subrange->getCount().dyn_cast<ConstantInt*>())
1339       Count = CI->getSExtValue();
1340 
1341     // Forward declarations of arrays without a size and VLAs use a count of -1.
1342     // Emit a count of zero in these cases to match what MSVC does for arrays
1343     // without a size. MSVC doesn't support VLAs, so it's not clear what we
1344     // should do for them even if we could distinguish them.
1345     if (Count == -1)
1346       Count = 0;
1347 
1348     // Update the element size and element type index for subsequent subranges.
1349     ElementSize *= Count;
1350 
1351     // If this is the outermost array, use the size from the array. It will be
1352     // more accurate if we had a VLA or an incomplete element type size.
1353     uint64_t ArraySize =
1354         (i == 0 && ElementSize == 0) ? Ty->getSizeInBits() / 8 : ElementSize;
1355 
1356     StringRef Name = (i == 0) ? Ty->getName() : "";
1357     ArrayRecord AR(ElementTypeIndex, IndexType, ArraySize, Name);
1358     ElementTypeIndex = TypeTable.writeLeafType(AR);
1359   }
1360 
1361   return ElementTypeIndex;
1362 }
1363 
1364 TypeIndex CodeViewDebug::lowerTypeBasic(const DIBasicType *Ty) {
1365   TypeIndex Index;
1366   dwarf::TypeKind Kind;
1367   uint32_t ByteSize;
1368 
1369   Kind = static_cast<dwarf::TypeKind>(Ty->getEncoding());
1370   ByteSize = Ty->getSizeInBits() / 8;
1371 
1372   SimpleTypeKind STK = SimpleTypeKind::None;
1373   switch (Kind) {
1374   case dwarf::DW_ATE_address:
1375     // FIXME: Translate
1376     break;
1377   case dwarf::DW_ATE_boolean:
1378     switch (ByteSize) {
1379     case 1:  STK = SimpleTypeKind::Boolean8;   break;
1380     case 2:  STK = SimpleTypeKind::Boolean16;  break;
1381     case 4:  STK = SimpleTypeKind::Boolean32;  break;
1382     case 8:  STK = SimpleTypeKind::Boolean64;  break;
1383     case 16: STK = SimpleTypeKind::Boolean128; break;
1384     }
1385     break;
1386   case dwarf::DW_ATE_complex_float:
1387     switch (ByteSize) {
1388     case 2:  STK = SimpleTypeKind::Complex16;  break;
1389     case 4:  STK = SimpleTypeKind::Complex32;  break;
1390     case 8:  STK = SimpleTypeKind::Complex64;  break;
1391     case 10: STK = SimpleTypeKind::Complex80;  break;
1392     case 16: STK = SimpleTypeKind::Complex128; break;
1393     }
1394     break;
1395   case dwarf::DW_ATE_float:
1396     switch (ByteSize) {
1397     case 2:  STK = SimpleTypeKind::Float16;  break;
1398     case 4:  STK = SimpleTypeKind::Float32;  break;
1399     case 6:  STK = SimpleTypeKind::Float48;  break;
1400     case 8:  STK = SimpleTypeKind::Float64;  break;
1401     case 10: STK = SimpleTypeKind::Float80;  break;
1402     case 16: STK = SimpleTypeKind::Float128; break;
1403     }
1404     break;
1405   case dwarf::DW_ATE_signed:
1406     switch (ByteSize) {
1407     case 1:  STK = SimpleTypeKind::SignedCharacter; break;
1408     case 2:  STK = SimpleTypeKind::Int16Short;      break;
1409     case 4:  STK = SimpleTypeKind::Int32;           break;
1410     case 8:  STK = SimpleTypeKind::Int64Quad;       break;
1411     case 16: STK = SimpleTypeKind::Int128Oct;       break;
1412     }
1413     break;
1414   case dwarf::DW_ATE_unsigned:
1415     switch (ByteSize) {
1416     case 1:  STK = SimpleTypeKind::UnsignedCharacter; break;
1417     case 2:  STK = SimpleTypeKind::UInt16Short;       break;
1418     case 4:  STK = SimpleTypeKind::UInt32;            break;
1419     case 8:  STK = SimpleTypeKind::UInt64Quad;        break;
1420     case 16: STK = SimpleTypeKind::UInt128Oct;        break;
1421     }
1422     break;
1423   case dwarf::DW_ATE_UTF:
1424     switch (ByteSize) {
1425     case 2: STK = SimpleTypeKind::Character16; break;
1426     case 4: STK = SimpleTypeKind::Character32; break;
1427     }
1428     break;
1429   case dwarf::DW_ATE_signed_char:
1430     if (ByteSize == 1)
1431       STK = SimpleTypeKind::SignedCharacter;
1432     break;
1433   case dwarf::DW_ATE_unsigned_char:
1434     if (ByteSize == 1)
1435       STK = SimpleTypeKind::UnsignedCharacter;
1436     break;
1437   default:
1438     break;
1439   }
1440 
1441   // Apply some fixups based on the source-level type name.
1442   if (STK == SimpleTypeKind::Int32 && Ty->getName() == "long int")
1443     STK = SimpleTypeKind::Int32Long;
1444   if (STK == SimpleTypeKind::UInt32 && Ty->getName() == "long unsigned int")
1445     STK = SimpleTypeKind::UInt32Long;
1446   if (STK == SimpleTypeKind::UInt16Short &&
1447       (Ty->getName() == "wchar_t" || Ty->getName() == "__wchar_t"))
1448     STK = SimpleTypeKind::WideCharacter;
1449   if ((STK == SimpleTypeKind::SignedCharacter ||
1450        STK == SimpleTypeKind::UnsignedCharacter) &&
1451       Ty->getName() == "char")
1452     STK = SimpleTypeKind::NarrowCharacter;
1453 
1454   return TypeIndex(STK);
1455 }
1456 
1457 TypeIndex CodeViewDebug::lowerTypePointer(const DIDerivedType *Ty,
1458                                           PointerOptions PO) {
1459   TypeIndex PointeeTI = getTypeIndex(Ty->getBaseType());
1460 
1461   // Pointers to simple types without any options can use SimpleTypeMode, rather
1462   // than having a dedicated pointer type record.
1463   if (PointeeTI.isSimple() && PO == PointerOptions::None &&
1464       PointeeTI.getSimpleMode() == SimpleTypeMode::Direct &&
1465       Ty->getTag() == dwarf::DW_TAG_pointer_type) {
1466     SimpleTypeMode Mode = Ty->getSizeInBits() == 64
1467                               ? SimpleTypeMode::NearPointer64
1468                               : SimpleTypeMode::NearPointer32;
1469     return TypeIndex(PointeeTI.getSimpleKind(), Mode);
1470   }
1471 
1472   PointerKind PK =
1473       Ty->getSizeInBits() == 64 ? PointerKind::Near64 : PointerKind::Near32;
1474   PointerMode PM = PointerMode::Pointer;
1475   switch (Ty->getTag()) {
1476   default: llvm_unreachable("not a pointer tag type");
1477   case dwarf::DW_TAG_pointer_type:
1478     PM = PointerMode::Pointer;
1479     break;
1480   case dwarf::DW_TAG_reference_type:
1481     PM = PointerMode::LValueReference;
1482     break;
1483   case dwarf::DW_TAG_rvalue_reference_type:
1484     PM = PointerMode::RValueReference;
1485     break;
1486   }
1487 
1488   PointerRecord PR(PointeeTI, PK, PM, PO, Ty->getSizeInBits() / 8);
1489   return TypeTable.writeLeafType(PR);
1490 }
1491 
1492 static PointerToMemberRepresentation
1493 translatePtrToMemberRep(unsigned SizeInBytes, bool IsPMF, unsigned Flags) {
1494   // SizeInBytes being zero generally implies that the member pointer type was
1495   // incomplete, which can happen if it is part of a function prototype. In this
1496   // case, use the unknown model instead of the general model.
1497   if (IsPMF) {
1498     switch (Flags & DINode::FlagPtrToMemberRep) {
1499     case 0:
1500       return SizeInBytes == 0 ? PointerToMemberRepresentation::Unknown
1501                               : PointerToMemberRepresentation::GeneralFunction;
1502     case DINode::FlagSingleInheritance:
1503       return PointerToMemberRepresentation::SingleInheritanceFunction;
1504     case DINode::FlagMultipleInheritance:
1505       return PointerToMemberRepresentation::MultipleInheritanceFunction;
1506     case DINode::FlagVirtualInheritance:
1507       return PointerToMemberRepresentation::VirtualInheritanceFunction;
1508     }
1509   } else {
1510     switch (Flags & DINode::FlagPtrToMemberRep) {
1511     case 0:
1512       return SizeInBytes == 0 ? PointerToMemberRepresentation::Unknown
1513                               : PointerToMemberRepresentation::GeneralData;
1514     case DINode::FlagSingleInheritance:
1515       return PointerToMemberRepresentation::SingleInheritanceData;
1516     case DINode::FlagMultipleInheritance:
1517       return PointerToMemberRepresentation::MultipleInheritanceData;
1518     case DINode::FlagVirtualInheritance:
1519       return PointerToMemberRepresentation::VirtualInheritanceData;
1520     }
1521   }
1522   llvm_unreachable("invalid ptr to member representation");
1523 }
1524 
1525 TypeIndex CodeViewDebug::lowerTypeMemberPointer(const DIDerivedType *Ty,
1526                                                 PointerOptions PO) {
1527   assert(Ty->getTag() == dwarf::DW_TAG_ptr_to_member_type);
1528   TypeIndex ClassTI = getTypeIndex(Ty->getClassType());
1529   TypeIndex PointeeTI = getTypeIndex(Ty->getBaseType(), Ty->getClassType());
1530   PointerKind PK = getPointerSizeInBytes() == 8 ? PointerKind::Near64
1531                                                 : PointerKind::Near32;
1532   bool IsPMF = isa<DISubroutineType>(Ty->getBaseType());
1533   PointerMode PM = IsPMF ? PointerMode::PointerToMemberFunction
1534                          : PointerMode::PointerToDataMember;
1535 
1536   assert(Ty->getSizeInBits() / 8 <= 0xff && "pointer size too big");
1537   uint8_t SizeInBytes = Ty->getSizeInBits() / 8;
1538   MemberPointerInfo MPI(
1539       ClassTI, translatePtrToMemberRep(SizeInBytes, IsPMF, Ty->getFlags()));
1540   PointerRecord PR(PointeeTI, PK, PM, PO, SizeInBytes, MPI);
1541   return TypeTable.writeLeafType(PR);
1542 }
1543 
1544 /// Given a DWARF calling convention, get the CodeView equivalent. If we don't
1545 /// have a translation, use the NearC convention.
1546 static CallingConvention dwarfCCToCodeView(unsigned DwarfCC) {
1547   switch (DwarfCC) {
1548   case dwarf::DW_CC_normal:             return CallingConvention::NearC;
1549   case dwarf::DW_CC_BORLAND_msfastcall: return CallingConvention::NearFast;
1550   case dwarf::DW_CC_BORLAND_thiscall:   return CallingConvention::ThisCall;
1551   case dwarf::DW_CC_BORLAND_stdcall:    return CallingConvention::NearStdCall;
1552   case dwarf::DW_CC_BORLAND_pascal:     return CallingConvention::NearPascal;
1553   case dwarf::DW_CC_LLVM_vectorcall:    return CallingConvention::NearVector;
1554   }
1555   return CallingConvention::NearC;
1556 }
1557 
1558 TypeIndex CodeViewDebug::lowerTypeModifier(const DIDerivedType *Ty) {
1559   ModifierOptions Mods = ModifierOptions::None;
1560   PointerOptions PO = PointerOptions::None;
1561   bool IsModifier = true;
1562   const DIType *BaseTy = Ty;
1563   while (IsModifier && BaseTy) {
1564     // FIXME: Need to add DWARF tags for __unaligned and _Atomic
1565     switch (BaseTy->getTag()) {
1566     case dwarf::DW_TAG_const_type:
1567       Mods |= ModifierOptions::Const;
1568       PO |= PointerOptions::Const;
1569       break;
1570     case dwarf::DW_TAG_volatile_type:
1571       Mods |= ModifierOptions::Volatile;
1572       PO |= PointerOptions::Volatile;
1573       break;
1574     case dwarf::DW_TAG_restrict_type:
1575       // Only pointer types be marked with __restrict. There is no known flag
1576       // for __restrict in LF_MODIFIER records.
1577       PO |= PointerOptions::Restrict;
1578       break;
1579     default:
1580       IsModifier = false;
1581       break;
1582     }
1583     if (IsModifier)
1584       BaseTy = cast<DIDerivedType>(BaseTy)->getBaseType().resolve();
1585   }
1586 
1587   // Check if the inner type will use an LF_POINTER record. If so, the
1588   // qualifiers will go in the LF_POINTER record. This comes up for types like
1589   // 'int *const' and 'int *__restrict', not the more common cases like 'const
1590   // char *'.
1591   if (BaseTy) {
1592     switch (BaseTy->getTag()) {
1593     case dwarf::DW_TAG_pointer_type:
1594     case dwarf::DW_TAG_reference_type:
1595     case dwarf::DW_TAG_rvalue_reference_type:
1596       return lowerTypePointer(cast<DIDerivedType>(BaseTy), PO);
1597     case dwarf::DW_TAG_ptr_to_member_type:
1598       return lowerTypeMemberPointer(cast<DIDerivedType>(BaseTy), PO);
1599     default:
1600       break;
1601     }
1602   }
1603 
1604   TypeIndex ModifiedTI = getTypeIndex(BaseTy);
1605 
1606   // Return the base type index if there aren't any modifiers. For example, the
1607   // metadata could contain restrict wrappers around non-pointer types.
1608   if (Mods == ModifierOptions::None)
1609     return ModifiedTI;
1610 
1611   ModifierRecord MR(ModifiedTI, Mods);
1612   return TypeTable.writeLeafType(MR);
1613 }
1614 
1615 TypeIndex CodeViewDebug::lowerTypeFunction(const DISubroutineType *Ty) {
1616   SmallVector<TypeIndex, 8> ReturnAndArgTypeIndices;
1617   for (DITypeRef ArgTypeRef : Ty->getTypeArray())
1618     ReturnAndArgTypeIndices.push_back(getTypeIndex(ArgTypeRef));
1619 
1620   // MSVC uses type none for variadic argument.
1621   if (ReturnAndArgTypeIndices.size() > 1 &&
1622       ReturnAndArgTypeIndices.back() == TypeIndex::Void()) {
1623     ReturnAndArgTypeIndices.back() = TypeIndex::None();
1624   }
1625   TypeIndex ReturnTypeIndex = TypeIndex::Void();
1626   ArrayRef<TypeIndex> ArgTypeIndices = None;
1627   if (!ReturnAndArgTypeIndices.empty()) {
1628     auto ReturnAndArgTypesRef = makeArrayRef(ReturnAndArgTypeIndices);
1629     ReturnTypeIndex = ReturnAndArgTypesRef.front();
1630     ArgTypeIndices = ReturnAndArgTypesRef.drop_front();
1631   }
1632 
1633   ArgListRecord ArgListRec(TypeRecordKind::ArgList, ArgTypeIndices);
1634   TypeIndex ArgListIndex = TypeTable.writeLeafType(ArgListRec);
1635 
1636   CallingConvention CC = dwarfCCToCodeView(Ty->getCC());
1637 
1638   ProcedureRecord Procedure(ReturnTypeIndex, CC, FunctionOptions::None,
1639                             ArgTypeIndices.size(), ArgListIndex);
1640   return TypeTable.writeLeafType(Procedure);
1641 }
1642 
1643 TypeIndex CodeViewDebug::lowerTypeMemberFunction(const DISubroutineType *Ty,
1644                                                  const DIType *ClassTy,
1645                                                  int ThisAdjustment,
1646                                                  bool IsStaticMethod) {
1647   // Lower the containing class type.
1648   TypeIndex ClassType = getTypeIndex(ClassTy);
1649 
1650   SmallVector<TypeIndex, 8> ReturnAndArgTypeIndices;
1651   for (DITypeRef ArgTypeRef : Ty->getTypeArray())
1652     ReturnAndArgTypeIndices.push_back(getTypeIndex(ArgTypeRef));
1653 
1654   // MSVC uses type none for variadic argument.
1655   if (ReturnAndArgTypeIndices.size() > 1 &&
1656       ReturnAndArgTypeIndices.back() == TypeIndex::Void()) {
1657     ReturnAndArgTypeIndices.back() = TypeIndex::None();
1658   }
1659   TypeIndex ReturnTypeIndex = TypeIndex::Void();
1660   ArrayRef<TypeIndex> ArgTypeIndices = None;
1661   if (!ReturnAndArgTypeIndices.empty()) {
1662     auto ReturnAndArgTypesRef = makeArrayRef(ReturnAndArgTypeIndices);
1663     ReturnTypeIndex = ReturnAndArgTypesRef.front();
1664     ArgTypeIndices = ReturnAndArgTypesRef.drop_front();
1665   }
1666   TypeIndex ThisTypeIndex;
1667   if (!IsStaticMethod && !ArgTypeIndices.empty()) {
1668     ThisTypeIndex = ArgTypeIndices.front();
1669     ArgTypeIndices = ArgTypeIndices.drop_front();
1670   }
1671 
1672   ArgListRecord ArgListRec(TypeRecordKind::ArgList, ArgTypeIndices);
1673   TypeIndex ArgListIndex = TypeTable.writeLeafType(ArgListRec);
1674 
1675   CallingConvention CC = dwarfCCToCodeView(Ty->getCC());
1676 
1677   // TODO: Need to use the correct values for FunctionOptions.
1678   MemberFunctionRecord MFR(ReturnTypeIndex, ClassType, ThisTypeIndex, CC,
1679                            FunctionOptions::None, ArgTypeIndices.size(),
1680                            ArgListIndex, ThisAdjustment);
1681   return TypeTable.writeLeafType(MFR);
1682 }
1683 
1684 TypeIndex CodeViewDebug::lowerTypeVFTableShape(const DIDerivedType *Ty) {
1685   unsigned VSlotCount =
1686       Ty->getSizeInBits() / (8 * Asm->MAI->getCodePointerSize());
1687   SmallVector<VFTableSlotKind, 4> Slots(VSlotCount, VFTableSlotKind::Near);
1688 
1689   VFTableShapeRecord VFTSR(Slots);
1690   return TypeTable.writeLeafType(VFTSR);
1691 }
1692 
1693 static MemberAccess translateAccessFlags(unsigned RecordTag, unsigned Flags) {
1694   switch (Flags & DINode::FlagAccessibility) {
1695   case DINode::FlagPrivate:   return MemberAccess::Private;
1696   case DINode::FlagPublic:    return MemberAccess::Public;
1697   case DINode::FlagProtected: return MemberAccess::Protected;
1698   case 0:
1699     // If there was no explicit access control, provide the default for the tag.
1700     return RecordTag == dwarf::DW_TAG_class_type ? MemberAccess::Private
1701                                                  : MemberAccess::Public;
1702   }
1703   llvm_unreachable("access flags are exclusive");
1704 }
1705 
1706 static MethodOptions translateMethodOptionFlags(const DISubprogram *SP) {
1707   if (SP->isArtificial())
1708     return MethodOptions::CompilerGenerated;
1709 
1710   // FIXME: Handle other MethodOptions.
1711 
1712   return MethodOptions::None;
1713 }
1714 
1715 static MethodKind translateMethodKindFlags(const DISubprogram *SP,
1716                                            bool Introduced) {
1717   if (SP->getFlags() & DINode::FlagStaticMember)
1718     return MethodKind::Static;
1719 
1720   switch (SP->getVirtuality()) {
1721   case dwarf::DW_VIRTUALITY_none:
1722     break;
1723   case dwarf::DW_VIRTUALITY_virtual:
1724     return Introduced ? MethodKind::IntroducingVirtual : MethodKind::Virtual;
1725   case dwarf::DW_VIRTUALITY_pure_virtual:
1726     return Introduced ? MethodKind::PureIntroducingVirtual
1727                       : MethodKind::PureVirtual;
1728   default:
1729     llvm_unreachable("unhandled virtuality case");
1730   }
1731 
1732   return MethodKind::Vanilla;
1733 }
1734 
1735 static TypeRecordKind getRecordKind(const DICompositeType *Ty) {
1736   switch (Ty->getTag()) {
1737   case dwarf::DW_TAG_class_type:     return TypeRecordKind::Class;
1738   case dwarf::DW_TAG_structure_type: return TypeRecordKind::Struct;
1739   }
1740   llvm_unreachable("unexpected tag");
1741 }
1742 
1743 /// Return ClassOptions that should be present on both the forward declaration
1744 /// and the defintion of a tag type.
1745 static ClassOptions getCommonClassOptions(const DICompositeType *Ty) {
1746   ClassOptions CO = ClassOptions::None;
1747 
1748   // MSVC always sets this flag, even for local types. Clang doesn't always
1749   // appear to give every type a linkage name, which may be problematic for us.
1750   // FIXME: Investigate the consequences of not following them here.
1751   if (!Ty->getIdentifier().empty())
1752     CO |= ClassOptions::HasUniqueName;
1753 
1754   // Put the Nested flag on a type if it appears immediately inside a tag type.
1755   // Do not walk the scope chain. Do not attempt to compute ContainsNestedClass
1756   // here. That flag is only set on definitions, and not forward declarations.
1757   const DIScope *ImmediateScope = Ty->getScope().resolve();
1758   if (ImmediateScope && isa<DICompositeType>(ImmediateScope))
1759     CO |= ClassOptions::Nested;
1760 
1761   // Put the Scoped flag on function-local types.
1762   for (const DIScope *Scope = ImmediateScope; Scope != nullptr;
1763        Scope = Scope->getScope().resolve()) {
1764     if (isa<DISubprogram>(Scope)) {
1765       CO |= ClassOptions::Scoped;
1766       break;
1767     }
1768   }
1769 
1770   return CO;
1771 }
1772 
1773 void CodeViewDebug::addUDTSrcLine(const DIType *Ty, TypeIndex TI) {
1774   switch (Ty->getTag()) {
1775   case dwarf::DW_TAG_class_type:
1776   case dwarf::DW_TAG_structure_type:
1777   case dwarf::DW_TAG_union_type:
1778   case dwarf::DW_TAG_enumeration_type:
1779     break;
1780   default:
1781     return;
1782   }
1783 
1784   if (const auto *File = Ty->getFile()) {
1785     StringIdRecord SIDR(TypeIndex(0x0), getFullFilepath(File));
1786     TypeIndex SIDI = TypeTable.writeLeafType(SIDR);
1787 
1788     UdtSourceLineRecord USLR(TI, SIDI, Ty->getLine());
1789     TypeTable.writeLeafType(USLR);
1790   }
1791 }
1792 
1793 TypeIndex CodeViewDebug::lowerTypeEnum(const DICompositeType *Ty) {
1794   ClassOptions CO = getCommonClassOptions(Ty);
1795   TypeIndex FTI;
1796   unsigned EnumeratorCount = 0;
1797 
1798   if (Ty->isForwardDecl()) {
1799     CO |= ClassOptions::ForwardReference;
1800   } else {
1801     ContinuationRecordBuilder ContinuationBuilder;
1802     ContinuationBuilder.begin(ContinuationRecordKind::FieldList);
1803     for (const DINode *Element : Ty->getElements()) {
1804       // We assume that the frontend provides all members in source declaration
1805       // order, which is what MSVC does.
1806       if (auto *Enumerator = dyn_cast_or_null<DIEnumerator>(Element)) {
1807         EnumeratorRecord ER(MemberAccess::Public,
1808                             APSInt::getUnsigned(Enumerator->getValue()),
1809                             Enumerator->getName());
1810         ContinuationBuilder.writeMemberType(ER);
1811         EnumeratorCount++;
1812       }
1813     }
1814     FTI = TypeTable.insertRecord(ContinuationBuilder);
1815   }
1816 
1817   std::string FullName = getFullyQualifiedName(Ty);
1818 
1819   EnumRecord ER(EnumeratorCount, CO, FTI, FullName, Ty->getIdentifier(),
1820                 getTypeIndex(Ty->getBaseType()));
1821   TypeIndex EnumTI = TypeTable.writeLeafType(ER);
1822 
1823   addUDTSrcLine(Ty, EnumTI);
1824 
1825   return EnumTI;
1826 }
1827 
1828 //===----------------------------------------------------------------------===//
1829 // ClassInfo
1830 //===----------------------------------------------------------------------===//
1831 
1832 struct llvm::ClassInfo {
1833   struct MemberInfo {
1834     const DIDerivedType *MemberTypeNode;
1835     uint64_t BaseOffset;
1836   };
1837   // [MemberInfo]
1838   using MemberList = std::vector<MemberInfo>;
1839 
1840   using MethodsList = TinyPtrVector<const DISubprogram *>;
1841   // MethodName -> MethodsList
1842   using MethodsMap = MapVector<MDString *, MethodsList>;
1843 
1844   /// Base classes.
1845   std::vector<const DIDerivedType *> Inheritance;
1846 
1847   /// Direct members.
1848   MemberList Members;
1849   // Direct overloaded methods gathered by name.
1850   MethodsMap Methods;
1851 
1852   TypeIndex VShapeTI;
1853 
1854   std::vector<const DIType *> NestedTypes;
1855 };
1856 
1857 void CodeViewDebug::clear() {
1858   assert(CurFn == nullptr);
1859   FileIdMap.clear();
1860   FnDebugInfo.clear();
1861   FileToFilepathMap.clear();
1862   LocalUDTs.clear();
1863   GlobalUDTs.clear();
1864   TypeIndices.clear();
1865   CompleteTypeIndices.clear();
1866 }
1867 
1868 void CodeViewDebug::collectMemberInfo(ClassInfo &Info,
1869                                       const DIDerivedType *DDTy) {
1870   if (!DDTy->getName().empty()) {
1871     Info.Members.push_back({DDTy, 0});
1872     return;
1873   }
1874 
1875   // An unnamed member may represent a nested struct or union. Attempt to
1876   // interpret the unnamed member as a DICompositeType possibly wrapped in
1877   // qualifier types. Add all the indirect fields to the current record if that
1878   // succeeds, and drop the member if that fails.
1879   assert((DDTy->getOffsetInBits() % 8) == 0 && "Unnamed bitfield member!");
1880   uint64_t Offset = DDTy->getOffsetInBits();
1881   const DIType *Ty = DDTy->getBaseType().resolve();
1882   bool FullyResolved = false;
1883   while (!FullyResolved) {
1884     switch (Ty->getTag()) {
1885     case dwarf::DW_TAG_const_type:
1886     case dwarf::DW_TAG_volatile_type:
1887       // FIXME: we should apply the qualifier types to the indirect fields
1888       // rather than dropping them.
1889       Ty = cast<DIDerivedType>(Ty)->getBaseType().resolve();
1890       break;
1891     default:
1892       FullyResolved = true;
1893       break;
1894     }
1895   }
1896 
1897   const DICompositeType *DCTy = dyn_cast<DICompositeType>(Ty);
1898   if (!DCTy)
1899     return;
1900 
1901   ClassInfo NestedInfo = collectClassInfo(DCTy);
1902   for (const ClassInfo::MemberInfo &IndirectField : NestedInfo.Members)
1903     Info.Members.push_back(
1904         {IndirectField.MemberTypeNode, IndirectField.BaseOffset + Offset});
1905 }
1906 
1907 ClassInfo CodeViewDebug::collectClassInfo(const DICompositeType *Ty) {
1908   ClassInfo Info;
1909   // Add elements to structure type.
1910   DINodeArray Elements = Ty->getElements();
1911   for (auto *Element : Elements) {
1912     // We assume that the frontend provides all members in source declaration
1913     // order, which is what MSVC does.
1914     if (!Element)
1915       continue;
1916     if (auto *SP = dyn_cast<DISubprogram>(Element)) {
1917       Info.Methods[SP->getRawName()].push_back(SP);
1918     } else if (auto *DDTy = dyn_cast<DIDerivedType>(Element)) {
1919       if (DDTy->getTag() == dwarf::DW_TAG_member) {
1920         collectMemberInfo(Info, DDTy);
1921       } else if (DDTy->getTag() == dwarf::DW_TAG_inheritance) {
1922         Info.Inheritance.push_back(DDTy);
1923       } else if (DDTy->getTag() == dwarf::DW_TAG_pointer_type &&
1924                  DDTy->getName() == "__vtbl_ptr_type") {
1925         Info.VShapeTI = getTypeIndex(DDTy);
1926       } else if (DDTy->getTag() == dwarf::DW_TAG_typedef) {
1927         Info.NestedTypes.push_back(DDTy);
1928       } else if (DDTy->getTag() == dwarf::DW_TAG_friend) {
1929         // Ignore friend members. It appears that MSVC emitted info about
1930         // friends in the past, but modern versions do not.
1931       }
1932     } else if (auto *Composite = dyn_cast<DICompositeType>(Element)) {
1933       Info.NestedTypes.push_back(Composite);
1934     }
1935     // Skip other unrecognized kinds of elements.
1936   }
1937   return Info;
1938 }
1939 
1940 TypeIndex CodeViewDebug::lowerTypeClass(const DICompositeType *Ty) {
1941   // First, construct the forward decl.  Don't look into Ty to compute the
1942   // forward decl options, since it might not be available in all TUs.
1943   TypeRecordKind Kind = getRecordKind(Ty);
1944   ClassOptions CO =
1945       ClassOptions::ForwardReference | getCommonClassOptions(Ty);
1946   std::string FullName = getFullyQualifiedName(Ty);
1947   ClassRecord CR(Kind, 0, CO, TypeIndex(), TypeIndex(), TypeIndex(), 0,
1948                  FullName, Ty->getIdentifier());
1949   TypeIndex FwdDeclTI = TypeTable.writeLeafType(CR);
1950   if (!Ty->isForwardDecl())
1951     DeferredCompleteTypes.push_back(Ty);
1952   return FwdDeclTI;
1953 }
1954 
1955 TypeIndex CodeViewDebug::lowerCompleteTypeClass(const DICompositeType *Ty) {
1956   // Construct the field list and complete type record.
1957   TypeRecordKind Kind = getRecordKind(Ty);
1958   ClassOptions CO = getCommonClassOptions(Ty);
1959   TypeIndex FieldTI;
1960   TypeIndex VShapeTI;
1961   unsigned FieldCount;
1962   bool ContainsNestedClass;
1963   std::tie(FieldTI, VShapeTI, FieldCount, ContainsNestedClass) =
1964       lowerRecordFieldList(Ty);
1965 
1966   if (ContainsNestedClass)
1967     CO |= ClassOptions::ContainsNestedClass;
1968 
1969   std::string FullName = getFullyQualifiedName(Ty);
1970 
1971   uint64_t SizeInBytes = Ty->getSizeInBits() / 8;
1972 
1973   ClassRecord CR(Kind, FieldCount, CO, FieldTI, TypeIndex(), VShapeTI,
1974                  SizeInBytes, FullName, Ty->getIdentifier());
1975   TypeIndex ClassTI = TypeTable.writeLeafType(CR);
1976 
1977   addUDTSrcLine(Ty, ClassTI);
1978 
1979   addToUDTs(Ty);
1980 
1981   return ClassTI;
1982 }
1983 
1984 TypeIndex CodeViewDebug::lowerTypeUnion(const DICompositeType *Ty) {
1985   ClassOptions CO =
1986       ClassOptions::ForwardReference | getCommonClassOptions(Ty);
1987   std::string FullName = getFullyQualifiedName(Ty);
1988   UnionRecord UR(0, CO, TypeIndex(), 0, FullName, Ty->getIdentifier());
1989   TypeIndex FwdDeclTI = TypeTable.writeLeafType(UR);
1990   if (!Ty->isForwardDecl())
1991     DeferredCompleteTypes.push_back(Ty);
1992   return FwdDeclTI;
1993 }
1994 
1995 TypeIndex CodeViewDebug::lowerCompleteTypeUnion(const DICompositeType *Ty) {
1996   ClassOptions CO = ClassOptions::Sealed | getCommonClassOptions(Ty);
1997   TypeIndex FieldTI;
1998   unsigned FieldCount;
1999   bool ContainsNestedClass;
2000   std::tie(FieldTI, std::ignore, FieldCount, ContainsNestedClass) =
2001       lowerRecordFieldList(Ty);
2002 
2003   if (ContainsNestedClass)
2004     CO |= ClassOptions::ContainsNestedClass;
2005 
2006   uint64_t SizeInBytes = Ty->getSizeInBits() / 8;
2007   std::string FullName = getFullyQualifiedName(Ty);
2008 
2009   UnionRecord UR(FieldCount, CO, FieldTI, SizeInBytes, FullName,
2010                  Ty->getIdentifier());
2011   TypeIndex UnionTI = TypeTable.writeLeafType(UR);
2012 
2013   addUDTSrcLine(Ty, UnionTI);
2014 
2015   addToUDTs(Ty);
2016 
2017   return UnionTI;
2018 }
2019 
2020 std::tuple<TypeIndex, TypeIndex, unsigned, bool>
2021 CodeViewDebug::lowerRecordFieldList(const DICompositeType *Ty) {
2022   // Manually count members. MSVC appears to count everything that generates a
2023   // field list record. Each individual overload in a method overload group
2024   // contributes to this count, even though the overload group is a single field
2025   // list record.
2026   unsigned MemberCount = 0;
2027   ClassInfo Info = collectClassInfo(Ty);
2028   ContinuationRecordBuilder ContinuationBuilder;
2029   ContinuationBuilder.begin(ContinuationRecordKind::FieldList);
2030 
2031   // Create base classes.
2032   for (const DIDerivedType *I : Info.Inheritance) {
2033     if (I->getFlags() & DINode::FlagVirtual) {
2034       // Virtual base.
2035       // FIXME: Emit VBPtrOffset when the frontend provides it.
2036       unsigned VBPtrOffset = 0;
2037       // FIXME: Despite the accessor name, the offset is really in bytes.
2038       unsigned VBTableIndex = I->getOffsetInBits() / 4;
2039       auto RecordKind = (I->getFlags() & DINode::FlagIndirectVirtualBase) == DINode::FlagIndirectVirtualBase
2040                             ? TypeRecordKind::IndirectVirtualBaseClass
2041                             : TypeRecordKind::VirtualBaseClass;
2042       VirtualBaseClassRecord VBCR(
2043           RecordKind, translateAccessFlags(Ty->getTag(), I->getFlags()),
2044           getTypeIndex(I->getBaseType()), getVBPTypeIndex(), VBPtrOffset,
2045           VBTableIndex);
2046 
2047       ContinuationBuilder.writeMemberType(VBCR);
2048       MemberCount++;
2049     } else {
2050       assert(I->getOffsetInBits() % 8 == 0 &&
2051              "bases must be on byte boundaries");
2052       BaseClassRecord BCR(translateAccessFlags(Ty->getTag(), I->getFlags()),
2053                           getTypeIndex(I->getBaseType()),
2054                           I->getOffsetInBits() / 8);
2055       ContinuationBuilder.writeMemberType(BCR);
2056       MemberCount++;
2057     }
2058   }
2059 
2060   // Create members.
2061   for (ClassInfo::MemberInfo &MemberInfo : Info.Members) {
2062     const DIDerivedType *Member = MemberInfo.MemberTypeNode;
2063     TypeIndex MemberBaseType = getTypeIndex(Member->getBaseType());
2064     StringRef MemberName = Member->getName();
2065     MemberAccess Access =
2066         translateAccessFlags(Ty->getTag(), Member->getFlags());
2067 
2068     if (Member->isStaticMember()) {
2069       StaticDataMemberRecord SDMR(Access, MemberBaseType, MemberName);
2070       ContinuationBuilder.writeMemberType(SDMR);
2071       MemberCount++;
2072       continue;
2073     }
2074 
2075     // Virtual function pointer member.
2076     if ((Member->getFlags() & DINode::FlagArtificial) &&
2077         Member->getName().startswith("_vptr$")) {
2078       VFPtrRecord VFPR(getTypeIndex(Member->getBaseType()));
2079       ContinuationBuilder.writeMemberType(VFPR);
2080       MemberCount++;
2081       continue;
2082     }
2083 
2084     // Data member.
2085     uint64_t MemberOffsetInBits =
2086         Member->getOffsetInBits() + MemberInfo.BaseOffset;
2087     if (Member->isBitField()) {
2088       uint64_t StartBitOffset = MemberOffsetInBits;
2089       if (const auto *CI =
2090               dyn_cast_or_null<ConstantInt>(Member->getStorageOffsetInBits())) {
2091         MemberOffsetInBits = CI->getZExtValue() + MemberInfo.BaseOffset;
2092       }
2093       StartBitOffset -= MemberOffsetInBits;
2094       BitFieldRecord BFR(MemberBaseType, Member->getSizeInBits(),
2095                          StartBitOffset);
2096       MemberBaseType = TypeTable.writeLeafType(BFR);
2097     }
2098     uint64_t MemberOffsetInBytes = MemberOffsetInBits / 8;
2099     DataMemberRecord DMR(Access, MemberBaseType, MemberOffsetInBytes,
2100                          MemberName);
2101     ContinuationBuilder.writeMemberType(DMR);
2102     MemberCount++;
2103   }
2104 
2105   // Create methods
2106   for (auto &MethodItr : Info.Methods) {
2107     StringRef Name = MethodItr.first->getString();
2108 
2109     std::vector<OneMethodRecord> Methods;
2110     for (const DISubprogram *SP : MethodItr.second) {
2111       TypeIndex MethodType = getMemberFunctionType(SP, Ty);
2112       bool Introduced = SP->getFlags() & DINode::FlagIntroducedVirtual;
2113 
2114       unsigned VFTableOffset = -1;
2115       if (Introduced)
2116         VFTableOffset = SP->getVirtualIndex() * getPointerSizeInBytes();
2117 
2118       Methods.push_back(OneMethodRecord(
2119           MethodType, translateAccessFlags(Ty->getTag(), SP->getFlags()),
2120           translateMethodKindFlags(SP, Introduced),
2121           translateMethodOptionFlags(SP), VFTableOffset, Name));
2122       MemberCount++;
2123     }
2124     assert(!Methods.empty() && "Empty methods map entry");
2125     if (Methods.size() == 1)
2126       ContinuationBuilder.writeMemberType(Methods[0]);
2127     else {
2128       // FIXME: Make this use its own ContinuationBuilder so that
2129       // MethodOverloadList can be split correctly.
2130       MethodOverloadListRecord MOLR(Methods);
2131       TypeIndex MethodList = TypeTable.writeLeafType(MOLR);
2132 
2133       OverloadedMethodRecord OMR(Methods.size(), MethodList, Name);
2134       ContinuationBuilder.writeMemberType(OMR);
2135     }
2136   }
2137 
2138   // Create nested classes.
2139   for (const DIType *Nested : Info.NestedTypes) {
2140     NestedTypeRecord R(getTypeIndex(DITypeRef(Nested)), Nested->getName());
2141     ContinuationBuilder.writeMemberType(R);
2142     MemberCount++;
2143   }
2144 
2145   TypeIndex FieldTI = TypeTable.insertRecord(ContinuationBuilder);
2146   return std::make_tuple(FieldTI, Info.VShapeTI, MemberCount,
2147                          !Info.NestedTypes.empty());
2148 }
2149 
2150 TypeIndex CodeViewDebug::getVBPTypeIndex() {
2151   if (!VBPType.getIndex()) {
2152     // Make a 'const int *' type.
2153     ModifierRecord MR(TypeIndex::Int32(), ModifierOptions::Const);
2154     TypeIndex ModifiedTI = TypeTable.writeLeafType(MR);
2155 
2156     PointerKind PK = getPointerSizeInBytes() == 8 ? PointerKind::Near64
2157                                                   : PointerKind::Near32;
2158     PointerMode PM = PointerMode::Pointer;
2159     PointerOptions PO = PointerOptions::None;
2160     PointerRecord PR(ModifiedTI, PK, PM, PO, getPointerSizeInBytes());
2161     VBPType = TypeTable.writeLeafType(PR);
2162   }
2163 
2164   return VBPType;
2165 }
2166 
2167 TypeIndex CodeViewDebug::getTypeIndex(DITypeRef TypeRef, DITypeRef ClassTyRef) {
2168   const DIType *Ty = TypeRef.resolve();
2169   const DIType *ClassTy = ClassTyRef.resolve();
2170 
2171   // The null DIType is the void type. Don't try to hash it.
2172   if (!Ty)
2173     return TypeIndex::Void();
2174 
2175   // Check if we've already translated this type. Don't try to do a
2176   // get-or-create style insertion that caches the hash lookup across the
2177   // lowerType call. It will update the TypeIndices map.
2178   auto I = TypeIndices.find({Ty, ClassTy});
2179   if (I != TypeIndices.end())
2180     return I->second;
2181 
2182   TypeLoweringScope S(*this);
2183   TypeIndex TI = lowerType(Ty, ClassTy);
2184   return recordTypeIndexForDINode(Ty, TI, ClassTy);
2185 }
2186 
2187 TypeIndex CodeViewDebug::getTypeIndexForReferenceTo(DITypeRef TypeRef) {
2188   DIType *Ty = TypeRef.resolve();
2189   PointerRecord PR(getTypeIndex(Ty),
2190                    getPointerSizeInBytes() == 8 ? PointerKind::Near64
2191                                                 : PointerKind::Near32,
2192                    PointerMode::LValueReference, PointerOptions::None,
2193                    Ty->getSizeInBits() / 8);
2194   return TypeTable.writeLeafType(PR);
2195 }
2196 
2197 TypeIndex CodeViewDebug::getCompleteTypeIndex(DITypeRef TypeRef) {
2198   const DIType *Ty = TypeRef.resolve();
2199 
2200   // The null DIType is the void type. Don't try to hash it.
2201   if (!Ty)
2202     return TypeIndex::Void();
2203 
2204   // If this is a non-record type, the complete type index is the same as the
2205   // normal type index. Just call getTypeIndex.
2206   switch (Ty->getTag()) {
2207   case dwarf::DW_TAG_class_type:
2208   case dwarf::DW_TAG_structure_type:
2209   case dwarf::DW_TAG_union_type:
2210     break;
2211   default:
2212     return getTypeIndex(Ty);
2213   }
2214 
2215   // Check if we've already translated the complete record type.  Lowering a
2216   // complete type should never trigger lowering another complete type, so we
2217   // can reuse the hash table lookup result.
2218   const auto *CTy = cast<DICompositeType>(Ty);
2219   auto InsertResult = CompleteTypeIndices.insert({CTy, TypeIndex()});
2220   if (!InsertResult.second)
2221     return InsertResult.first->second;
2222 
2223   TypeLoweringScope S(*this);
2224 
2225   // Make sure the forward declaration is emitted first. It's unclear if this
2226   // is necessary, but MSVC does it, and we should follow suit until we can show
2227   // otherwise.
2228   TypeIndex FwdDeclTI = getTypeIndex(CTy);
2229 
2230   // Just use the forward decl if we don't have complete type info. This might
2231   // happen if the frontend is using modules and expects the complete definition
2232   // to be emitted elsewhere.
2233   if (CTy->isForwardDecl())
2234     return FwdDeclTI;
2235 
2236   TypeIndex TI;
2237   switch (CTy->getTag()) {
2238   case dwarf::DW_TAG_class_type:
2239   case dwarf::DW_TAG_structure_type:
2240     TI = lowerCompleteTypeClass(CTy);
2241     break;
2242   case dwarf::DW_TAG_union_type:
2243     TI = lowerCompleteTypeUnion(CTy);
2244     break;
2245   default:
2246     llvm_unreachable("not a record");
2247   }
2248 
2249   InsertResult.first->second = TI;
2250   return TI;
2251 }
2252 
2253 /// Emit all the deferred complete record types. Try to do this in FIFO order,
2254 /// and do this until fixpoint, as each complete record type typically
2255 /// references
2256 /// many other record types.
2257 void CodeViewDebug::emitDeferredCompleteTypes() {
2258   SmallVector<const DICompositeType *, 4> TypesToEmit;
2259   while (!DeferredCompleteTypes.empty()) {
2260     std::swap(DeferredCompleteTypes, TypesToEmit);
2261     for (const DICompositeType *RecordTy : TypesToEmit)
2262       getCompleteTypeIndex(RecordTy);
2263     TypesToEmit.clear();
2264   }
2265 }
2266 
2267 void CodeViewDebug::emitLocalVariableList(ArrayRef<LocalVariable> Locals) {
2268   // Get the sorted list of parameters and emit them first.
2269   SmallVector<const LocalVariable *, 6> Params;
2270   for (const LocalVariable &L : Locals)
2271     if (L.DIVar->isParameter())
2272       Params.push_back(&L);
2273   std::sort(Params.begin(), Params.end(),
2274             [](const LocalVariable *L, const LocalVariable *R) {
2275               return L->DIVar->getArg() < R->DIVar->getArg();
2276             });
2277   for (const LocalVariable *L : Params)
2278     emitLocalVariable(*L);
2279 
2280   // Next emit all non-parameters in the order that we found them.
2281   for (const LocalVariable &L : Locals)
2282     if (!L.DIVar->isParameter())
2283       emitLocalVariable(L);
2284 }
2285 
2286 void CodeViewDebug::emitLocalVariable(const LocalVariable &Var) {
2287   // LocalSym record, see SymbolRecord.h for more info.
2288   MCSymbol *LocalBegin = MMI->getContext().createTempSymbol(),
2289            *LocalEnd = MMI->getContext().createTempSymbol();
2290   OS.AddComment("Record length");
2291   OS.emitAbsoluteSymbolDiff(LocalEnd, LocalBegin, 2);
2292   OS.EmitLabel(LocalBegin);
2293 
2294   OS.AddComment("Record kind: S_LOCAL");
2295   OS.EmitIntValue(unsigned(SymbolKind::S_LOCAL), 2);
2296 
2297   LocalSymFlags Flags = LocalSymFlags::None;
2298   if (Var.DIVar->isParameter())
2299     Flags |= LocalSymFlags::IsParameter;
2300   if (Var.DefRanges.empty())
2301     Flags |= LocalSymFlags::IsOptimizedOut;
2302 
2303   OS.AddComment("TypeIndex");
2304   TypeIndex TI = Var.UseReferenceType
2305                      ? getTypeIndexForReferenceTo(Var.DIVar->getType())
2306                      : getCompleteTypeIndex(Var.DIVar->getType());
2307   OS.EmitIntValue(TI.getIndex(), 4);
2308   OS.AddComment("Flags");
2309   OS.EmitIntValue(static_cast<uint16_t>(Flags), 2);
2310   // Truncate the name so we won't overflow the record length field.
2311   emitNullTerminatedSymbolName(OS, Var.DIVar->getName());
2312   OS.EmitLabel(LocalEnd);
2313 
2314   // Calculate the on disk prefix of the appropriate def range record. The
2315   // records and on disk formats are described in SymbolRecords.h. BytePrefix
2316   // should be big enough to hold all forms without memory allocation.
2317   SmallString<20> BytePrefix;
2318   for (const LocalVarDefRange &DefRange : Var.DefRanges) {
2319     BytePrefix.clear();
2320     if (DefRange.InMemory) {
2321       uint16_t RegRelFlags = 0;
2322       if (DefRange.IsSubfield) {
2323         RegRelFlags = DefRangeRegisterRelSym::IsSubfieldFlag |
2324                       (DefRange.StructOffset
2325                        << DefRangeRegisterRelSym::OffsetInParentShift);
2326       }
2327       DefRangeRegisterRelSym Sym(S_DEFRANGE_REGISTER_REL);
2328       Sym.Hdr.Register = DefRange.CVRegister;
2329       Sym.Hdr.Flags = RegRelFlags;
2330       Sym.Hdr.BasePointerOffset = DefRange.DataOffset;
2331       ulittle16_t SymKind = ulittle16_t(S_DEFRANGE_REGISTER_REL);
2332       BytePrefix +=
2333           StringRef(reinterpret_cast<const char *>(&SymKind), sizeof(SymKind));
2334       BytePrefix +=
2335           StringRef(reinterpret_cast<const char *>(&Sym.Hdr), sizeof(Sym.Hdr));
2336     } else {
2337       assert(DefRange.DataOffset == 0 && "unexpected offset into register");
2338       if (DefRange.IsSubfield) {
2339         // Unclear what matters here.
2340         DefRangeSubfieldRegisterSym Sym(S_DEFRANGE_SUBFIELD_REGISTER);
2341         Sym.Hdr.Register = DefRange.CVRegister;
2342         Sym.Hdr.MayHaveNoName = 0;
2343         Sym.Hdr.OffsetInParent = DefRange.StructOffset;
2344 
2345         ulittle16_t SymKind = ulittle16_t(S_DEFRANGE_SUBFIELD_REGISTER);
2346         BytePrefix += StringRef(reinterpret_cast<const char *>(&SymKind),
2347                                 sizeof(SymKind));
2348         BytePrefix += StringRef(reinterpret_cast<const char *>(&Sym.Hdr),
2349                                 sizeof(Sym.Hdr));
2350       } else {
2351         // Unclear what matters here.
2352         DefRangeRegisterSym Sym(S_DEFRANGE_REGISTER);
2353         Sym.Hdr.Register = DefRange.CVRegister;
2354         Sym.Hdr.MayHaveNoName = 0;
2355         ulittle16_t SymKind = ulittle16_t(S_DEFRANGE_REGISTER);
2356         BytePrefix += StringRef(reinterpret_cast<const char *>(&SymKind),
2357                                 sizeof(SymKind));
2358         BytePrefix += StringRef(reinterpret_cast<const char *>(&Sym.Hdr),
2359                                 sizeof(Sym.Hdr));
2360       }
2361     }
2362     OS.EmitCVDefRangeDirective(DefRange.Ranges, BytePrefix);
2363   }
2364 }
2365 
2366 void CodeViewDebug::endFunctionImpl(const MachineFunction *MF) {
2367   const Function &GV = MF->getFunction();
2368   assert(FnDebugInfo.count(&GV));
2369   assert(CurFn == FnDebugInfo[&GV].get());
2370 
2371   collectVariableInfo(GV.getSubprogram());
2372 
2373   // Don't emit anything if we don't have any line tables.
2374   if (!CurFn->HaveLineInfo) {
2375     FnDebugInfo.erase(&GV);
2376     CurFn = nullptr;
2377     return;
2378   }
2379 
2380   CurFn->Annotations = MF->getCodeViewAnnotations();
2381 
2382   CurFn->End = Asm->getFunctionEnd();
2383 
2384   CurFn = nullptr;
2385 }
2386 
2387 void CodeViewDebug::beginInstruction(const MachineInstr *MI) {
2388   DebugHandlerBase::beginInstruction(MI);
2389 
2390   // Ignore DBG_VALUE locations and function prologue.
2391   if (!Asm || !CurFn || MI->isDebugValue() ||
2392       MI->getFlag(MachineInstr::FrameSetup))
2393     return;
2394 
2395   // If the first instruction of a new MBB has no location, find the first
2396   // instruction with a location and use that.
2397   DebugLoc DL = MI->getDebugLoc();
2398   if (!DL && MI->getParent() != PrevInstBB) {
2399     for (const auto &NextMI : *MI->getParent()) {
2400       if (NextMI.isDebugValue())
2401         continue;
2402       DL = NextMI.getDebugLoc();
2403       if (DL)
2404         break;
2405     }
2406   }
2407   PrevInstBB = MI->getParent();
2408 
2409   // If we still don't have a debug location, don't record a location.
2410   if (!DL)
2411     return;
2412 
2413   maybeRecordLocation(DL, Asm->MF);
2414 }
2415 
2416 MCSymbol *CodeViewDebug::beginCVSubsection(DebugSubsectionKind Kind) {
2417   MCSymbol *BeginLabel = MMI->getContext().createTempSymbol(),
2418            *EndLabel = MMI->getContext().createTempSymbol();
2419   OS.EmitIntValue(unsigned(Kind), 4);
2420   OS.AddComment("Subsection size");
2421   OS.emitAbsoluteSymbolDiff(EndLabel, BeginLabel, 4);
2422   OS.EmitLabel(BeginLabel);
2423   return EndLabel;
2424 }
2425 
2426 void CodeViewDebug::endCVSubsection(MCSymbol *EndLabel) {
2427   OS.EmitLabel(EndLabel);
2428   // Every subsection must be aligned to a 4-byte boundary.
2429   OS.EmitValueToAlignment(4);
2430 }
2431 
2432 void CodeViewDebug::emitDebugInfoForUDTs(
2433     ArrayRef<std::pair<std::string, const DIType *>> UDTs) {
2434   for (const auto &UDT : UDTs) {
2435     const DIType *T = UDT.second;
2436     assert(shouldEmitUdt(T));
2437 
2438     MCSymbol *UDTRecordBegin = MMI->getContext().createTempSymbol(),
2439              *UDTRecordEnd = MMI->getContext().createTempSymbol();
2440     OS.AddComment("Record length");
2441     OS.emitAbsoluteSymbolDiff(UDTRecordEnd, UDTRecordBegin, 2);
2442     OS.EmitLabel(UDTRecordBegin);
2443 
2444     OS.AddComment("Record kind: S_UDT");
2445     OS.EmitIntValue(unsigned(SymbolKind::S_UDT), 2);
2446 
2447     OS.AddComment("Type");
2448     OS.EmitIntValue(getCompleteTypeIndex(T).getIndex(), 4);
2449 
2450     emitNullTerminatedSymbolName(OS, UDT.first);
2451     OS.EmitLabel(UDTRecordEnd);
2452   }
2453 }
2454 
2455 void CodeViewDebug::emitDebugInfoForGlobals() {
2456   DenseMap<const DIGlobalVariableExpression *, const GlobalVariable *>
2457       GlobalMap;
2458   for (const GlobalVariable &GV : MMI->getModule()->globals()) {
2459     SmallVector<DIGlobalVariableExpression *, 1> GVEs;
2460     GV.getDebugInfo(GVEs);
2461     for (const auto *GVE : GVEs)
2462       GlobalMap[GVE] = &GV;
2463   }
2464 
2465   NamedMDNode *CUs = MMI->getModule()->getNamedMetadata("llvm.dbg.cu");
2466   for (const MDNode *Node : CUs->operands()) {
2467     const auto *CU = cast<DICompileUnit>(Node);
2468 
2469     // First, emit all globals that are not in a comdat in a single symbol
2470     // substream. MSVC doesn't like it if the substream is empty, so only open
2471     // it if we have at least one global to emit.
2472     switchToDebugSectionForSymbol(nullptr);
2473     MCSymbol *EndLabel = nullptr;
2474     for (const auto *GVE : CU->getGlobalVariables()) {
2475       if (const auto *GV = GlobalMap.lookup(GVE))
2476         if (!GV->hasComdat() && !GV->isDeclarationForLinker()) {
2477           if (!EndLabel) {
2478             OS.AddComment("Symbol subsection for globals");
2479             EndLabel = beginCVSubsection(DebugSubsectionKind::Symbols);
2480           }
2481           // FIXME: emitDebugInfoForGlobal() doesn't handle DIExpressions.
2482           emitDebugInfoForGlobal(GVE->getVariable(), GV, Asm->getSymbol(GV));
2483         }
2484     }
2485     if (EndLabel)
2486       endCVSubsection(EndLabel);
2487 
2488     // Second, emit each global that is in a comdat into its own .debug$S
2489     // section along with its own symbol substream.
2490     for (const auto *GVE : CU->getGlobalVariables()) {
2491       if (const auto *GV = GlobalMap.lookup(GVE)) {
2492         if (GV->hasComdat()) {
2493           MCSymbol *GVSym = Asm->getSymbol(GV);
2494           OS.AddComment("Symbol subsection for " +
2495                         Twine(GlobalValue::dropLLVMManglingEscape(GV->getName())));
2496           switchToDebugSectionForSymbol(GVSym);
2497           EndLabel = beginCVSubsection(DebugSubsectionKind::Symbols);
2498           // FIXME: emitDebugInfoForGlobal() doesn't handle DIExpressions.
2499           emitDebugInfoForGlobal(GVE->getVariable(), GV, GVSym);
2500           endCVSubsection(EndLabel);
2501         }
2502       }
2503     }
2504   }
2505 }
2506 
2507 void CodeViewDebug::emitDebugInfoForRetainedTypes() {
2508   NamedMDNode *CUs = MMI->getModule()->getNamedMetadata("llvm.dbg.cu");
2509   for (const MDNode *Node : CUs->operands()) {
2510     for (auto *Ty : cast<DICompileUnit>(Node)->getRetainedTypes()) {
2511       if (DIType *RT = dyn_cast<DIType>(Ty)) {
2512         getTypeIndex(RT);
2513         // FIXME: Add to global/local DTU list.
2514       }
2515     }
2516   }
2517 }
2518 
2519 void CodeViewDebug::emitDebugInfoForGlobal(const DIGlobalVariable *DIGV,
2520                                            const GlobalVariable *GV,
2521                                            MCSymbol *GVSym) {
2522   // DataSym record, see SymbolRecord.h for more info.
2523   // FIXME: Thread local data, etc
2524   MCSymbol *DataBegin = MMI->getContext().createTempSymbol(),
2525            *DataEnd = MMI->getContext().createTempSymbol();
2526   const unsigned FixedLengthOfThisRecord = 12;
2527   OS.AddComment("Record length");
2528   OS.emitAbsoluteSymbolDiff(DataEnd, DataBegin, 2);
2529   OS.EmitLabel(DataBegin);
2530   if (DIGV->isLocalToUnit()) {
2531     if (GV->isThreadLocal()) {
2532       OS.AddComment("Record kind: S_LTHREAD32");
2533       OS.EmitIntValue(unsigned(SymbolKind::S_LTHREAD32), 2);
2534     } else {
2535       OS.AddComment("Record kind: S_LDATA32");
2536       OS.EmitIntValue(unsigned(SymbolKind::S_LDATA32), 2);
2537     }
2538   } else {
2539     if (GV->isThreadLocal()) {
2540       OS.AddComment("Record kind: S_GTHREAD32");
2541       OS.EmitIntValue(unsigned(SymbolKind::S_GTHREAD32), 2);
2542     } else {
2543       OS.AddComment("Record kind: S_GDATA32");
2544       OS.EmitIntValue(unsigned(SymbolKind::S_GDATA32), 2);
2545     }
2546   }
2547   OS.AddComment("Type");
2548   OS.EmitIntValue(getCompleteTypeIndex(DIGV->getType()).getIndex(), 4);
2549   OS.AddComment("DataOffset");
2550   OS.EmitCOFFSecRel32(GVSym, /*Offset=*/0);
2551   OS.AddComment("Segment");
2552   OS.EmitCOFFSectionIndex(GVSym);
2553   OS.AddComment("Name");
2554   emitNullTerminatedSymbolName(OS, DIGV->getName(), FixedLengthOfThisRecord);
2555   OS.EmitLabel(DataEnd);
2556 }
2557