• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
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/TargetRegisterInfo.h"
40 #include "llvm/CodeGen/TargetSubtargetInfo.h"
41 #include "llvm/Config/llvm-config.h"
42 #include "llvm/DebugInfo/CodeView/CVTypeVisitor.h"
43 #include "llvm/DebugInfo/CodeView/CodeView.h"
44 #include "llvm/DebugInfo/CodeView/ContinuationRecordBuilder.h"
45 #include "llvm/DebugInfo/CodeView/DebugInlineeLinesSubsection.h"
46 #include "llvm/DebugInfo/CodeView/Line.h"
47 #include "llvm/DebugInfo/CodeView/SymbolRecord.h"
48 #include "llvm/DebugInfo/CodeView/TypeDumpVisitor.h"
49 #include "llvm/DebugInfo/CodeView/TypeIndex.h"
50 #include "llvm/DebugInfo/CodeView/TypeRecord.h"
51 #include "llvm/DebugInfo/CodeView/TypeTableCollection.h"
52 #include "llvm/IR/Constants.h"
53 #include "llvm/IR/DataLayout.h"
54 #include "llvm/IR/DebugInfoMetadata.h"
55 #include "llvm/IR/DebugLoc.h"
56 #include "llvm/IR/Function.h"
57 #include "llvm/IR/GlobalValue.h"
58 #include "llvm/IR/GlobalVariable.h"
59 #include "llvm/IR/Metadata.h"
60 #include "llvm/IR/Module.h"
61 #include "llvm/MC/MCAsmInfo.h"
62 #include "llvm/MC/MCContext.h"
63 #include "llvm/MC/MCSectionCOFF.h"
64 #include "llvm/MC/MCStreamer.h"
65 #include "llvm/MC/MCSymbol.h"
66 #include "llvm/Support/BinaryByteStream.h"
67 #include "llvm/Support/BinaryStreamReader.h"
68 #include "llvm/Support/Casting.h"
69 #include "llvm/Support/CommandLine.h"
70 #include "llvm/Support/Compiler.h"
71 #include "llvm/Support/Endian.h"
72 #include "llvm/Support/Error.h"
73 #include "llvm/Support/ErrorHandling.h"
74 #include "llvm/Support/FormatVariadic.h"
75 #include "llvm/Support/SMLoc.h"
76 #include "llvm/Support/ScopedPrinter.h"
77 #include "llvm/Target/TargetLoweringObjectFile.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 
CodeViewDebug(AsmPrinter * AP)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 
getFullFilepath(const DIFile * File)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   // If this is a Unix-style path, just use it as is. Don't try to canonicalize
118   // it textually because one of the path components could be a symlink.
119   if (!Dir.empty() && Dir[0] == '/') {
120     Filepath = Dir;
121     if (Dir.back() != '/')
122       Filepath += '/';
123     Filepath += Filename;
124     return Filepath;
125   }
126 
127   // Clang emits directory and relative filename info into the IR, but CodeView
128   // operates on full paths.  We could change Clang to emit full paths too, but
129   // that would increase the IR size and probably not needed for other users.
130   // For now, just concatenate and canonicalize the path here.
131   if (Filename.find(':') == 1)
132     Filepath = Filename;
133   else
134     Filepath = (Dir + "\\" + Filename).str();
135 
136   // Canonicalize the path.  We have to do it textually because we may no longer
137   // have access the file in the filesystem.
138   // First, replace all slashes with backslashes.
139   std::replace(Filepath.begin(), Filepath.end(), '/', '\\');
140 
141   // Remove all "\.\" with "\".
142   size_t Cursor = 0;
143   while ((Cursor = Filepath.find("\\.\\", Cursor)) != std::string::npos)
144     Filepath.erase(Cursor, 2);
145 
146   // Replace all "\XXX\..\" with "\".  Don't try too hard though as the original
147   // path should be well-formatted, e.g. start with a drive letter, etc.
148   Cursor = 0;
149   while ((Cursor = Filepath.find("\\..\\", Cursor)) != std::string::npos) {
150     // Something's wrong if the path starts with "\..\", abort.
151     if (Cursor == 0)
152       break;
153 
154     size_t PrevSlash = Filepath.rfind('\\', Cursor - 1);
155     if (PrevSlash == std::string::npos)
156       // Something's wrong, abort.
157       break;
158 
159     Filepath.erase(PrevSlash, Cursor + 3 - PrevSlash);
160     // The next ".." might be following the one we've just erased.
161     Cursor = PrevSlash;
162   }
163 
164   // Remove all duplicate backslashes.
165   Cursor = 0;
166   while ((Cursor = Filepath.find("\\\\", Cursor)) != std::string::npos)
167     Filepath.erase(Cursor, 1);
168 
169   return Filepath;
170 }
171 
maybeRecordFile(const DIFile * F)172 unsigned CodeViewDebug::maybeRecordFile(const DIFile *F) {
173   StringRef FullPath = getFullFilepath(F);
174   unsigned NextId = FileIdMap.size() + 1;
175   auto Insertion = FileIdMap.insert(std::make_pair(FullPath, NextId));
176   if (Insertion.second) {
177     // We have to compute the full filepath and emit a .cv_file directive.
178     ArrayRef<uint8_t> ChecksumAsBytes;
179     FileChecksumKind CSKind = FileChecksumKind::None;
180     if (F->getChecksum()) {
181       std::string Checksum = fromHex(F->getChecksum()->Value);
182       void *CKMem = OS.getContext().allocate(Checksum.size(), 1);
183       memcpy(CKMem, Checksum.data(), Checksum.size());
184       ChecksumAsBytes = ArrayRef<uint8_t>(
185           reinterpret_cast<const uint8_t *>(CKMem), Checksum.size());
186       switch (F->getChecksum()->Kind) {
187       case DIFile::CSK_MD5:  CSKind = FileChecksumKind::MD5; break;
188       case DIFile::CSK_SHA1: CSKind = FileChecksumKind::SHA1; break;
189       }
190     }
191     bool Success = OS.EmitCVFileDirective(NextId, FullPath, ChecksumAsBytes,
192                                           static_cast<unsigned>(CSKind));
193     (void)Success;
194     assert(Success && ".cv_file directive failed");
195   }
196   return Insertion.first->second;
197 }
198 
199 CodeViewDebug::InlineSite &
getInlineSite(const DILocation * InlinedAt,const DISubprogram * Inlinee)200 CodeViewDebug::getInlineSite(const DILocation *InlinedAt,
201                              const DISubprogram *Inlinee) {
202   auto SiteInsertion = CurFn->InlineSites.insert({InlinedAt, InlineSite()});
203   InlineSite *Site = &SiteInsertion.first->second;
204   if (SiteInsertion.second) {
205     unsigned ParentFuncId = CurFn->FuncId;
206     if (const DILocation *OuterIA = InlinedAt->getInlinedAt())
207       ParentFuncId =
208           getInlineSite(OuterIA, InlinedAt->getScope()->getSubprogram())
209               .SiteFuncId;
210 
211     Site->SiteFuncId = NextFuncId++;
212     OS.EmitCVInlineSiteIdDirective(
213         Site->SiteFuncId, ParentFuncId, maybeRecordFile(InlinedAt->getFile()),
214         InlinedAt->getLine(), InlinedAt->getColumn(), SMLoc());
215     Site->Inlinee = Inlinee;
216     InlinedSubprograms.insert(Inlinee);
217     getFuncIdForSubprogram(Inlinee);
218   }
219   return *Site;
220 }
221 
getPrettyScopeName(const DIScope * Scope)222 static StringRef getPrettyScopeName(const DIScope *Scope) {
223   StringRef ScopeName = Scope->getName();
224   if (!ScopeName.empty())
225     return ScopeName;
226 
227   switch (Scope->getTag()) {
228   case dwarf::DW_TAG_enumeration_type:
229   case dwarf::DW_TAG_class_type:
230   case dwarf::DW_TAG_structure_type:
231   case dwarf::DW_TAG_union_type:
232     return "<unnamed-tag>";
233   case dwarf::DW_TAG_namespace:
234     return "`anonymous namespace'";
235   }
236 
237   return StringRef();
238 }
239 
getQualifiedNameComponents(const DIScope * Scope,SmallVectorImpl<StringRef> & QualifiedNameComponents)240 static const DISubprogram *getQualifiedNameComponents(
241     const DIScope *Scope, SmallVectorImpl<StringRef> &QualifiedNameComponents) {
242   const DISubprogram *ClosestSubprogram = nullptr;
243   while (Scope != nullptr) {
244     if (ClosestSubprogram == nullptr)
245       ClosestSubprogram = dyn_cast<DISubprogram>(Scope);
246     StringRef ScopeName = getPrettyScopeName(Scope);
247     if (!ScopeName.empty())
248       QualifiedNameComponents.push_back(ScopeName);
249     Scope = Scope->getScope().resolve();
250   }
251   return ClosestSubprogram;
252 }
253 
getQualifiedName(ArrayRef<StringRef> QualifiedNameComponents,StringRef TypeName)254 static std::string getQualifiedName(ArrayRef<StringRef> QualifiedNameComponents,
255                                     StringRef TypeName) {
256   std::string FullyQualifiedName;
257   for (StringRef QualifiedNameComponent :
258        llvm::reverse(QualifiedNameComponents)) {
259     FullyQualifiedName.append(QualifiedNameComponent);
260     FullyQualifiedName.append("::");
261   }
262   FullyQualifiedName.append(TypeName);
263   return FullyQualifiedName;
264 }
265 
getFullyQualifiedName(const DIScope * Scope,StringRef Name)266 static std::string getFullyQualifiedName(const DIScope *Scope, StringRef Name) {
267   SmallVector<StringRef, 5> QualifiedNameComponents;
268   getQualifiedNameComponents(Scope, QualifiedNameComponents);
269   return getQualifiedName(QualifiedNameComponents, Name);
270 }
271 
272 struct CodeViewDebug::TypeLoweringScope {
TypeLoweringScopeCodeViewDebug::TypeLoweringScope273   TypeLoweringScope(CodeViewDebug &CVD) : CVD(CVD) { ++CVD.TypeEmissionLevel; }
~TypeLoweringScopeCodeViewDebug::TypeLoweringScope274   ~TypeLoweringScope() {
275     // Don't decrement TypeEmissionLevel until after emitting deferred types, so
276     // inner TypeLoweringScopes don't attempt to emit deferred types.
277     if (CVD.TypeEmissionLevel == 1)
278       CVD.emitDeferredCompleteTypes();
279     --CVD.TypeEmissionLevel;
280   }
281   CodeViewDebug &CVD;
282 };
283 
getFullyQualifiedName(const DIScope * Ty)284 static std::string getFullyQualifiedName(const DIScope *Ty) {
285   const DIScope *Scope = Ty->getScope().resolve();
286   return getFullyQualifiedName(Scope, getPrettyScopeName(Ty));
287 }
288 
getScopeIndex(const DIScope * Scope)289 TypeIndex CodeViewDebug::getScopeIndex(const DIScope *Scope) {
290   // No scope means global scope and that uses the zero index.
291   if (!Scope || isa<DIFile>(Scope))
292     return TypeIndex();
293 
294   assert(!isa<DIType>(Scope) && "shouldn't make a namespace scope for a type");
295 
296   // Check if we've already translated this scope.
297   auto I = TypeIndices.find({Scope, nullptr});
298   if (I != TypeIndices.end())
299     return I->second;
300 
301   // Build the fully qualified name of the scope.
302   std::string ScopeName = getFullyQualifiedName(Scope);
303   StringIdRecord SID(TypeIndex(), ScopeName);
304   auto TI = TypeTable.writeLeafType(SID);
305   return recordTypeIndexForDINode(Scope, TI);
306 }
307 
getFuncIdForSubprogram(const DISubprogram * SP)308 TypeIndex CodeViewDebug::getFuncIdForSubprogram(const DISubprogram *SP) {
309   assert(SP);
310 
311   // Check if we've already translated this subprogram.
312   auto I = TypeIndices.find({SP, nullptr});
313   if (I != TypeIndices.end())
314     return I->second;
315 
316   // The display name includes function template arguments. Drop them to match
317   // MSVC.
318   StringRef DisplayName = SP->getName().split('<').first;
319 
320   const DIScope *Scope = SP->getScope().resolve();
321   TypeIndex TI;
322   if (const auto *Class = dyn_cast_or_null<DICompositeType>(Scope)) {
323     // If the scope is a DICompositeType, then this must be a method. Member
324     // function types take some special handling, and require access to the
325     // subprogram.
326     TypeIndex ClassType = getTypeIndex(Class);
327     MemberFuncIdRecord MFuncId(ClassType, getMemberFunctionType(SP, Class),
328                                DisplayName);
329     TI = TypeTable.writeLeafType(MFuncId);
330   } else {
331     // Otherwise, this must be a free function.
332     TypeIndex ParentScope = getScopeIndex(Scope);
333     FuncIdRecord FuncId(ParentScope, getTypeIndex(SP->getType()), DisplayName);
334     TI = TypeTable.writeLeafType(FuncId);
335   }
336 
337   return recordTypeIndexForDINode(SP, TI);
338 }
339 
getMemberFunctionType(const DISubprogram * SP,const DICompositeType * Class)340 TypeIndex CodeViewDebug::getMemberFunctionType(const DISubprogram *SP,
341                                                const DICompositeType *Class) {
342   // Always use the method declaration as the key for the function type. The
343   // method declaration contains the this adjustment.
344   if (SP->getDeclaration())
345     SP = SP->getDeclaration();
346   assert(!SP->getDeclaration() && "should use declaration as key");
347 
348   // Key the MemberFunctionRecord into the map as {SP, Class}. It won't collide
349   // with the MemberFuncIdRecord, which is keyed in as {SP, nullptr}.
350   auto I = TypeIndices.find({SP, Class});
351   if (I != TypeIndices.end())
352     return I->second;
353 
354   // Make sure complete type info for the class is emitted *after* the member
355   // function type, as the complete class type is likely to reference this
356   // member function type.
357   TypeLoweringScope S(*this);
358   const bool IsStaticMethod = (SP->getFlags() & DINode::FlagStaticMember) != 0;
359   TypeIndex TI = lowerTypeMemberFunction(
360       SP->getType(), Class, SP->getThisAdjustment(), IsStaticMethod);
361   return recordTypeIndexForDINode(SP, TI, Class);
362 }
363 
recordTypeIndexForDINode(const DINode * Node,TypeIndex TI,const DIType * ClassTy)364 TypeIndex CodeViewDebug::recordTypeIndexForDINode(const DINode *Node,
365                                                   TypeIndex TI,
366                                                   const DIType *ClassTy) {
367   auto InsertResult = TypeIndices.insert({{Node, ClassTy}, TI});
368   (void)InsertResult;
369   assert(InsertResult.second && "DINode was already assigned a type index");
370   return TI;
371 }
372 
getPointerSizeInBytes()373 unsigned CodeViewDebug::getPointerSizeInBytes() {
374   return MMI->getModule()->getDataLayout().getPointerSizeInBits() / 8;
375 }
376 
recordLocalVariable(LocalVariable && Var,const LexicalScope * LS)377 void CodeViewDebug::recordLocalVariable(LocalVariable &&Var,
378                                         const LexicalScope *LS) {
379   if (const DILocation *InlinedAt = LS->getInlinedAt()) {
380     // This variable was inlined. Associate it with the InlineSite.
381     const DISubprogram *Inlinee = Var.DIVar->getScope()->getSubprogram();
382     InlineSite &Site = getInlineSite(InlinedAt, Inlinee);
383     Site.InlinedLocals.emplace_back(Var);
384   } else {
385     // This variable goes into the corresponding lexical scope.
386     ScopeVariables[LS].emplace_back(Var);
387   }
388 }
389 
addLocIfNotPresent(SmallVectorImpl<const DILocation * > & Locs,const DILocation * Loc)390 static void addLocIfNotPresent(SmallVectorImpl<const DILocation *> &Locs,
391                                const DILocation *Loc) {
392   auto B = Locs.begin(), E = Locs.end();
393   if (std::find(B, E, Loc) == E)
394     Locs.push_back(Loc);
395 }
396 
maybeRecordLocation(const DebugLoc & DL,const MachineFunction * MF)397 void CodeViewDebug::maybeRecordLocation(const DebugLoc &DL,
398                                         const MachineFunction *MF) {
399   // Skip this instruction if it has the same location as the previous one.
400   if (!DL || DL == PrevInstLoc)
401     return;
402 
403   const DIScope *Scope = DL.get()->getScope();
404   if (!Scope)
405     return;
406 
407   // Skip this line if it is longer than the maximum we can record.
408   LineInfo LI(DL.getLine(), DL.getLine(), /*IsStatement=*/true);
409   if (LI.getStartLine() != DL.getLine() || LI.isAlwaysStepInto() ||
410       LI.isNeverStepInto())
411     return;
412 
413   ColumnInfo CI(DL.getCol(), /*EndColumn=*/0);
414   if (CI.getStartColumn() != DL.getCol())
415     return;
416 
417   if (!CurFn->HaveLineInfo)
418     CurFn->HaveLineInfo = true;
419   unsigned FileId = 0;
420   if (PrevInstLoc.get() && PrevInstLoc->getFile() == DL->getFile())
421     FileId = CurFn->LastFileId;
422   else
423     FileId = CurFn->LastFileId = maybeRecordFile(DL->getFile());
424   PrevInstLoc = DL;
425 
426   unsigned FuncId = CurFn->FuncId;
427   if (const DILocation *SiteLoc = DL->getInlinedAt()) {
428     const DILocation *Loc = DL.get();
429 
430     // If this location was actually inlined from somewhere else, give it the ID
431     // of the inline call site.
432     FuncId =
433         getInlineSite(SiteLoc, Loc->getScope()->getSubprogram()).SiteFuncId;
434 
435     // Ensure we have links in the tree of inline call sites.
436     bool FirstLoc = true;
437     while ((SiteLoc = Loc->getInlinedAt())) {
438       InlineSite &Site =
439           getInlineSite(SiteLoc, Loc->getScope()->getSubprogram());
440       if (!FirstLoc)
441         addLocIfNotPresent(Site.ChildSites, Loc);
442       FirstLoc = false;
443       Loc = SiteLoc;
444     }
445     addLocIfNotPresent(CurFn->ChildSites, Loc);
446   }
447 
448   OS.EmitCVLocDirective(FuncId, FileId, DL.getLine(), DL.getCol(),
449                         /*PrologueEnd=*/false, /*IsStmt=*/false,
450                         DL->getFilename(), SMLoc());
451 }
452 
emitCodeViewMagicVersion()453 void CodeViewDebug::emitCodeViewMagicVersion() {
454   OS.EmitValueToAlignment(4);
455   OS.AddComment("Debug section magic");
456   OS.EmitIntValue(COFF::DEBUG_SECTION_MAGIC, 4);
457 }
458 
endModule()459 void CodeViewDebug::endModule() {
460   if (!Asm || !MMI->hasDebugInfo())
461     return;
462 
463   assert(Asm != nullptr);
464 
465   // The COFF .debug$S section consists of several subsections, each starting
466   // with a 4-byte control code (e.g. 0xF1, 0xF2, etc) and then a 4-byte length
467   // of the payload followed by the payload itself.  The subsections are 4-byte
468   // aligned.
469 
470   // Use the generic .debug$S section, and make a subsection for all the inlined
471   // subprograms.
472   switchToDebugSectionForSymbol(nullptr);
473 
474   MCSymbol *CompilerInfo = beginCVSubsection(DebugSubsectionKind::Symbols);
475   emitCompilerInformation();
476   endCVSubsection(CompilerInfo);
477 
478   emitInlineeLinesSubsection();
479 
480   // Emit per-function debug information.
481   for (auto &P : FnDebugInfo)
482     if (!P.first->isDeclarationForLinker())
483       emitDebugInfoForFunction(P.first, *P.second);
484 
485   // Emit global variable debug information.
486   setCurrentSubprogram(nullptr);
487   emitDebugInfoForGlobals();
488 
489   // Emit retained types.
490   emitDebugInfoForRetainedTypes();
491 
492   // Switch back to the generic .debug$S section after potentially processing
493   // comdat symbol sections.
494   switchToDebugSectionForSymbol(nullptr);
495 
496   // Emit UDT records for any types used by global variables.
497   if (!GlobalUDTs.empty()) {
498     MCSymbol *SymbolsEnd = beginCVSubsection(DebugSubsectionKind::Symbols);
499     emitDebugInfoForUDTs(GlobalUDTs);
500     endCVSubsection(SymbolsEnd);
501   }
502 
503   // This subsection holds a file index to offset in string table table.
504   OS.AddComment("File index to string table offset subsection");
505   OS.EmitCVFileChecksumsDirective();
506 
507   // This subsection holds the string table.
508   OS.AddComment("String table");
509   OS.EmitCVStringTableDirective();
510 
511   // Emit type information and hashes last, so that any types we translate while
512   // emitting function info are included.
513   emitTypeInformation();
514 
515   if (EmitDebugGlobalHashes)
516     emitTypeGlobalHashes();
517 
518   clear();
519 }
520 
emitNullTerminatedSymbolName(MCStreamer & OS,StringRef S,unsigned MaxFixedRecordLength=0xF00)521 static void emitNullTerminatedSymbolName(MCStreamer &OS, StringRef S,
522     unsigned MaxFixedRecordLength = 0xF00) {
523   // The maximum CV record length is 0xFF00. Most of the strings we emit appear
524   // after a fixed length portion of the record. The fixed length portion should
525   // always be less than 0xF00 (3840) bytes, so truncate the string so that the
526   // overall record size is less than the maximum allowed.
527   SmallString<32> NullTerminatedString(
528       S.take_front(MaxRecordLength - MaxFixedRecordLength - 1));
529   NullTerminatedString.push_back('\0');
530   OS.EmitBytes(NullTerminatedString);
531 }
532 
emitTypeInformation()533 void CodeViewDebug::emitTypeInformation() {
534   if (TypeTable.empty())
535     return;
536 
537   // Start the .debug$T or .debug$P section with 0x4.
538   OS.SwitchSection(Asm->getObjFileLowering().getCOFFDebugTypesSection());
539   emitCodeViewMagicVersion();
540 
541   SmallString<8> CommentPrefix;
542   if (OS.isVerboseAsm()) {
543     CommentPrefix += '\t';
544     CommentPrefix += Asm->MAI->getCommentString();
545     CommentPrefix += ' ';
546   }
547 
548   TypeTableCollection Table(TypeTable.records());
549   Optional<TypeIndex> B = Table.getFirst();
550   while (B) {
551     // This will fail if the record data is invalid.
552     CVType Record = Table.getType(*B);
553 
554     if (OS.isVerboseAsm()) {
555       // Emit a block comment describing the type record for readability.
556       SmallString<512> CommentBlock;
557       raw_svector_ostream CommentOS(CommentBlock);
558       ScopedPrinter SP(CommentOS);
559       SP.setPrefix(CommentPrefix);
560       TypeDumpVisitor TDV(Table, &SP, false);
561 
562       Error E = codeview::visitTypeRecord(Record, *B, TDV);
563       if (E) {
564         logAllUnhandledErrors(std::move(E), errs(), "error: ");
565         llvm_unreachable("produced malformed type record");
566       }
567       // emitRawComment will insert its own tab and comment string before
568       // the first line, so strip off our first one. It also prints its own
569       // newline.
570       OS.emitRawComment(
571           CommentOS.str().drop_front(CommentPrefix.size() - 1).rtrim());
572     }
573     OS.EmitBinaryData(Record.str_data());
574     B = Table.getNext(*B);
575   }
576 }
577 
emitTypeGlobalHashes()578 void CodeViewDebug::emitTypeGlobalHashes() {
579   if (TypeTable.empty())
580     return;
581 
582   // Start the .debug$H section with the version and hash algorithm, currently
583   // hardcoded to version 0, SHA1.
584   OS.SwitchSection(Asm->getObjFileLowering().getCOFFGlobalTypeHashesSection());
585 
586   OS.EmitValueToAlignment(4);
587   OS.AddComment("Magic");
588   OS.EmitIntValue(COFF::DEBUG_HASHES_SECTION_MAGIC, 4);
589   OS.AddComment("Section Version");
590   OS.EmitIntValue(0, 2);
591   OS.AddComment("Hash Algorithm");
592   OS.EmitIntValue(uint16_t(GlobalTypeHashAlg::SHA1_8), 2);
593 
594   TypeIndex TI(TypeIndex::FirstNonSimpleIndex);
595   for (const auto &GHR : TypeTable.hashes()) {
596     if (OS.isVerboseAsm()) {
597       // Emit an EOL-comment describing which TypeIndex this hash corresponds
598       // to, as well as the stringified SHA1 hash.
599       SmallString<32> Comment;
600       raw_svector_ostream CommentOS(Comment);
601       CommentOS << formatv("{0:X+} [{1}]", TI.getIndex(), GHR);
602       OS.AddComment(Comment);
603       ++TI;
604     }
605     assert(GHR.Hash.size() == 8);
606     StringRef S(reinterpret_cast<const char *>(GHR.Hash.data()),
607                 GHR.Hash.size());
608     OS.EmitBinaryData(S);
609   }
610 }
611 
MapDWLangToCVLang(unsigned DWLang)612 static SourceLanguage MapDWLangToCVLang(unsigned DWLang) {
613   switch (DWLang) {
614   case dwarf::DW_LANG_C:
615   case dwarf::DW_LANG_C89:
616   case dwarf::DW_LANG_C99:
617   case dwarf::DW_LANG_C11:
618   case dwarf::DW_LANG_ObjC:
619     return SourceLanguage::C;
620   case dwarf::DW_LANG_C_plus_plus:
621   case dwarf::DW_LANG_C_plus_plus_03:
622   case dwarf::DW_LANG_C_plus_plus_11:
623   case dwarf::DW_LANG_C_plus_plus_14:
624     return SourceLanguage::Cpp;
625   case dwarf::DW_LANG_Fortran77:
626   case dwarf::DW_LANG_Fortran90:
627   case dwarf::DW_LANG_Fortran03:
628   case dwarf::DW_LANG_Fortran08:
629     return SourceLanguage::Fortran;
630   case dwarf::DW_LANG_Pascal83:
631     return SourceLanguage::Pascal;
632   case dwarf::DW_LANG_Cobol74:
633   case dwarf::DW_LANG_Cobol85:
634     return SourceLanguage::Cobol;
635   case dwarf::DW_LANG_Java:
636     return SourceLanguage::Java;
637   case dwarf::DW_LANG_D:
638     return SourceLanguage::D;
639   default:
640     // There's no CodeView representation for this language, and CV doesn't
641     // have an "unknown" option for the language field, so we'll use MASM,
642     // as it's very low level.
643     return SourceLanguage::Masm;
644   }
645 }
646 
647 namespace {
648 struct Version {
649   int Part[4];
650 };
651 } // end anonymous namespace
652 
653 // Takes a StringRef like "clang 4.0.0.0 (other nonsense 123)" and parses out
654 // the version number.
parseVersion(StringRef Name)655 static Version parseVersion(StringRef Name) {
656   Version V = {{0}};
657   int N = 0;
658   for (const char C : Name) {
659     if (isdigit(C)) {
660       V.Part[N] *= 10;
661       V.Part[N] += C - '0';
662     } else if (C == '.') {
663       ++N;
664       if (N >= 4)
665         return V;
666     } else if (N > 0)
667       return V;
668   }
669   return V;
670 }
671 
mapArchToCVCPUType(Triple::ArchType Type)672 static CPUType mapArchToCVCPUType(Triple::ArchType Type) {
673   switch (Type) {
674   case Triple::ArchType::x86:
675     return CPUType::Pentium3;
676   case Triple::ArchType::x86_64:
677     return CPUType::X64;
678   case Triple::ArchType::thumb:
679     return CPUType::Thumb;
680   case Triple::ArchType::aarch64:
681     return CPUType::ARM64;
682   default:
683     report_fatal_error("target architecture doesn't map to a CodeView CPUType");
684   }
685 }
686 
emitCompilerInformation()687 void CodeViewDebug::emitCompilerInformation() {
688   MCContext &Context = MMI->getContext();
689   MCSymbol *CompilerBegin = Context.createTempSymbol(),
690            *CompilerEnd = Context.createTempSymbol();
691   OS.AddComment("Record length");
692   OS.emitAbsoluteSymbolDiff(CompilerEnd, CompilerBegin, 2);
693   OS.EmitLabel(CompilerBegin);
694   OS.AddComment("Record kind: S_COMPILE3");
695   OS.EmitIntValue(SymbolKind::S_COMPILE3, 2);
696   uint32_t Flags = 0;
697 
698   NamedMDNode *CUs = MMI->getModule()->getNamedMetadata("llvm.dbg.cu");
699   const MDNode *Node = *CUs->operands().begin();
700   const auto *CU = cast<DICompileUnit>(Node);
701 
702   // The low byte of the flags indicates the source language.
703   Flags = MapDWLangToCVLang(CU->getSourceLanguage());
704   // TODO:  Figure out which other flags need to be set.
705 
706   OS.AddComment("Flags and language");
707   OS.EmitIntValue(Flags, 4);
708 
709   OS.AddComment("CPUType");
710   CPUType CPU =
711       mapArchToCVCPUType(Triple(MMI->getModule()->getTargetTriple()).getArch());
712   OS.EmitIntValue(static_cast<uint64_t>(CPU), 2);
713 
714   StringRef CompilerVersion = CU->getProducer();
715   Version FrontVer = parseVersion(CompilerVersion);
716   OS.AddComment("Frontend version");
717   for (int N = 0; N < 4; ++N)
718     OS.EmitIntValue(FrontVer.Part[N], 2);
719 
720   // Some Microsoft tools, like Binscope, expect a backend version number of at
721   // least 8.something, so we'll coerce the LLVM version into a form that
722   // guarantees it'll be big enough without really lying about the version.
723   int Major = 1000 * LLVM_VERSION_MAJOR +
724               10 * LLVM_VERSION_MINOR +
725               LLVM_VERSION_PATCH;
726   // Clamp it for builds that use unusually large version numbers.
727   Major = std::min<int>(Major, std::numeric_limits<uint16_t>::max());
728   Version BackVer = {{ Major, 0, 0, 0 }};
729   OS.AddComment("Backend version");
730   for (int N = 0; N < 4; ++N)
731     OS.EmitIntValue(BackVer.Part[N], 2);
732 
733   OS.AddComment("Null-terminated compiler version string");
734   emitNullTerminatedSymbolName(OS, CompilerVersion);
735 
736   OS.EmitLabel(CompilerEnd);
737 }
738 
emitInlineeLinesSubsection()739 void CodeViewDebug::emitInlineeLinesSubsection() {
740   if (InlinedSubprograms.empty())
741     return;
742 
743   OS.AddComment("Inlinee lines subsection");
744   MCSymbol *InlineEnd = beginCVSubsection(DebugSubsectionKind::InlineeLines);
745 
746   // We emit the checksum info for files.  This is used by debuggers to
747   // determine if a pdb matches the source before loading it.  Visual Studio,
748   // for instance, will display a warning that the breakpoints are not valid if
749   // the pdb does not match the source.
750   OS.AddComment("Inlinee lines signature");
751   OS.EmitIntValue(unsigned(InlineeLinesSignature::Normal), 4);
752 
753   for (const DISubprogram *SP : InlinedSubprograms) {
754     assert(TypeIndices.count({SP, nullptr}));
755     TypeIndex InlineeIdx = TypeIndices[{SP, nullptr}];
756 
757     OS.AddBlankLine();
758     unsigned FileId = maybeRecordFile(SP->getFile());
759     OS.AddComment("Inlined function " + SP->getName() + " starts at " +
760                   SP->getFilename() + Twine(':') + Twine(SP->getLine()));
761     OS.AddBlankLine();
762     OS.AddComment("Type index of inlined function");
763     OS.EmitIntValue(InlineeIdx.getIndex(), 4);
764     OS.AddComment("Offset into filechecksum table");
765     OS.EmitCVFileChecksumOffsetDirective(FileId);
766     OS.AddComment("Starting line number");
767     OS.EmitIntValue(SP->getLine(), 4);
768   }
769 
770   endCVSubsection(InlineEnd);
771 }
772 
emitInlinedCallSite(const FunctionInfo & FI,const DILocation * InlinedAt,const InlineSite & Site)773 void CodeViewDebug::emitInlinedCallSite(const FunctionInfo &FI,
774                                         const DILocation *InlinedAt,
775                                         const InlineSite &Site) {
776   MCSymbol *InlineBegin = MMI->getContext().createTempSymbol(),
777            *InlineEnd = MMI->getContext().createTempSymbol();
778 
779   assert(TypeIndices.count({Site.Inlinee, nullptr}));
780   TypeIndex InlineeIdx = TypeIndices[{Site.Inlinee, nullptr}];
781 
782   // SymbolRecord
783   OS.AddComment("Record length");
784   OS.emitAbsoluteSymbolDiff(InlineEnd, InlineBegin, 2);   // RecordLength
785   OS.EmitLabel(InlineBegin);
786   OS.AddComment("Record kind: S_INLINESITE");
787   OS.EmitIntValue(SymbolKind::S_INLINESITE, 2); // RecordKind
788 
789   OS.AddComment("PtrParent");
790   OS.EmitIntValue(0, 4);
791   OS.AddComment("PtrEnd");
792   OS.EmitIntValue(0, 4);
793   OS.AddComment("Inlinee type index");
794   OS.EmitIntValue(InlineeIdx.getIndex(), 4);
795 
796   unsigned FileId = maybeRecordFile(Site.Inlinee->getFile());
797   unsigned StartLineNum = Site.Inlinee->getLine();
798 
799   OS.EmitCVInlineLinetableDirective(Site.SiteFuncId, FileId, StartLineNum,
800                                     FI.Begin, FI.End);
801 
802   OS.EmitLabel(InlineEnd);
803 
804   emitLocalVariableList(Site.InlinedLocals);
805 
806   // Recurse on child inlined call sites before closing the scope.
807   for (const DILocation *ChildSite : Site.ChildSites) {
808     auto I = FI.InlineSites.find(ChildSite);
809     assert(I != FI.InlineSites.end() &&
810            "child site not in function inline site map");
811     emitInlinedCallSite(FI, ChildSite, I->second);
812   }
813 
814   // Close the scope.
815   OS.AddComment("Record length");
816   OS.EmitIntValue(2, 2);                                  // RecordLength
817   OS.AddComment("Record kind: S_INLINESITE_END");
818   OS.EmitIntValue(SymbolKind::S_INLINESITE_END, 2); // RecordKind
819 }
820 
switchToDebugSectionForSymbol(const MCSymbol * GVSym)821 void CodeViewDebug::switchToDebugSectionForSymbol(const MCSymbol *GVSym) {
822   // If we have a symbol, it may be in a section that is COMDAT. If so, find the
823   // comdat key. A section may be comdat because of -ffunction-sections or
824   // because it is comdat in the IR.
825   MCSectionCOFF *GVSec =
826       GVSym ? dyn_cast<MCSectionCOFF>(&GVSym->getSection()) : nullptr;
827   const MCSymbol *KeySym = GVSec ? GVSec->getCOMDATSymbol() : nullptr;
828 
829   MCSectionCOFF *DebugSec = cast<MCSectionCOFF>(
830       Asm->getObjFileLowering().getCOFFDebugSymbolsSection());
831   DebugSec = OS.getContext().getAssociativeCOFFSection(DebugSec, KeySym);
832 
833   OS.SwitchSection(DebugSec);
834 
835   // Emit the magic version number if this is the first time we've switched to
836   // this section.
837   if (ComdatDebugSections.insert(DebugSec).second)
838     emitCodeViewMagicVersion();
839 }
840 
841 // Emit an S_THUNK32/S_END symbol pair for a thunk routine.
842 // The only supported thunk ordinal is currently the standard type.
emitDebugInfoForThunk(const Function * GV,FunctionInfo & FI,const MCSymbol * Fn)843 void CodeViewDebug::emitDebugInfoForThunk(const Function *GV,
844                                           FunctionInfo &FI,
845                                           const MCSymbol *Fn) {
846   std::string FuncName = GlobalValue::dropLLVMManglingEscape(GV->getName());
847   const ThunkOrdinal ordinal = ThunkOrdinal::Standard; // Only supported kind.
848 
849   OS.AddComment("Symbol subsection for " + Twine(FuncName));
850   MCSymbol *SymbolsEnd = beginCVSubsection(DebugSubsectionKind::Symbols);
851 
852   // Emit S_THUNK32
853   MCSymbol *ThunkRecordBegin = MMI->getContext().createTempSymbol(),
854            *ThunkRecordEnd   = MMI->getContext().createTempSymbol();
855   OS.AddComment("Record length");
856   OS.emitAbsoluteSymbolDiff(ThunkRecordEnd, ThunkRecordBegin, 2);
857   OS.EmitLabel(ThunkRecordBegin);
858   OS.AddComment("Record kind: S_THUNK32");
859   OS.EmitIntValue(unsigned(SymbolKind::S_THUNK32), 2);
860   OS.AddComment("PtrParent");
861   OS.EmitIntValue(0, 4);
862   OS.AddComment("PtrEnd");
863   OS.EmitIntValue(0, 4);
864   OS.AddComment("PtrNext");
865   OS.EmitIntValue(0, 4);
866   OS.AddComment("Thunk section relative address");
867   OS.EmitCOFFSecRel32(Fn, /*Offset=*/0);
868   OS.AddComment("Thunk section index");
869   OS.EmitCOFFSectionIndex(Fn);
870   OS.AddComment("Code size");
871   OS.emitAbsoluteSymbolDiff(FI.End, Fn, 2);
872   OS.AddComment("Ordinal");
873   OS.EmitIntValue(unsigned(ordinal), 1);
874   OS.AddComment("Function name");
875   emitNullTerminatedSymbolName(OS, FuncName);
876   // Additional fields specific to the thunk ordinal would go here.
877   OS.EmitLabel(ThunkRecordEnd);
878 
879   // Local variables/inlined routines are purposely omitted here.  The point of
880   // marking this as a thunk is so Visual Studio will NOT stop in this routine.
881 
882   // Emit S_PROC_ID_END
883   const unsigned RecordLengthForSymbolEnd = 2;
884   OS.AddComment("Record length");
885   OS.EmitIntValue(RecordLengthForSymbolEnd, 2);
886   OS.AddComment("Record kind: S_PROC_ID_END");
887   OS.EmitIntValue(unsigned(SymbolKind::S_PROC_ID_END), 2);
888 
889   endCVSubsection(SymbolsEnd);
890 }
891 
emitDebugInfoForFunction(const Function * GV,FunctionInfo & FI)892 void CodeViewDebug::emitDebugInfoForFunction(const Function *GV,
893                                              FunctionInfo &FI) {
894   // For each function there is a separate subsection which holds the PC to
895   // file:line table.
896   const MCSymbol *Fn = Asm->getSymbol(GV);
897   assert(Fn);
898 
899   // Switch to the to a comdat section, if appropriate.
900   switchToDebugSectionForSymbol(Fn);
901 
902   std::string FuncName;
903   auto *SP = GV->getSubprogram();
904   assert(SP);
905   setCurrentSubprogram(SP);
906 
907   if (SP->isThunk()) {
908     emitDebugInfoForThunk(GV, FI, Fn);
909     return;
910   }
911 
912   // If we have a display name, build the fully qualified name by walking the
913   // chain of scopes.
914   if (!SP->getName().empty())
915     FuncName =
916         getFullyQualifiedName(SP->getScope().resolve(), SP->getName());
917 
918   // If our DISubprogram name is empty, use the mangled name.
919   if (FuncName.empty())
920     FuncName = GlobalValue::dropLLVMManglingEscape(GV->getName());
921 
922   // Emit FPO data, but only on 32-bit x86. No other platforms use it.
923   if (Triple(MMI->getModule()->getTargetTriple()).getArch() == Triple::x86)
924     OS.EmitCVFPOData(Fn);
925 
926   // Emit a symbol subsection, required by VS2012+ to find function boundaries.
927   OS.AddComment("Symbol subsection for " + Twine(FuncName));
928   MCSymbol *SymbolsEnd = beginCVSubsection(DebugSubsectionKind::Symbols);
929   {
930     MCSymbol *ProcRecordBegin = MMI->getContext().createTempSymbol(),
931              *ProcRecordEnd = MMI->getContext().createTempSymbol();
932     OS.AddComment("Record length");
933     OS.emitAbsoluteSymbolDiff(ProcRecordEnd, ProcRecordBegin, 2);
934     OS.EmitLabel(ProcRecordBegin);
935 
936     if (GV->hasLocalLinkage()) {
937       OS.AddComment("Record kind: S_LPROC32_ID");
938       OS.EmitIntValue(unsigned(SymbolKind::S_LPROC32_ID), 2);
939     } else {
940       OS.AddComment("Record kind: S_GPROC32_ID");
941       OS.EmitIntValue(unsigned(SymbolKind::S_GPROC32_ID), 2);
942     }
943 
944     // These fields are filled in by tools like CVPACK which run after the fact.
945     OS.AddComment("PtrParent");
946     OS.EmitIntValue(0, 4);
947     OS.AddComment("PtrEnd");
948     OS.EmitIntValue(0, 4);
949     OS.AddComment("PtrNext");
950     OS.EmitIntValue(0, 4);
951     // This is the important bit that tells the debugger where the function
952     // code is located and what's its size:
953     OS.AddComment("Code size");
954     OS.emitAbsoluteSymbolDiff(FI.End, Fn, 4);
955     OS.AddComment("Offset after prologue");
956     OS.EmitIntValue(0, 4);
957     OS.AddComment("Offset before epilogue");
958     OS.EmitIntValue(0, 4);
959     OS.AddComment("Function type index");
960     OS.EmitIntValue(getFuncIdForSubprogram(GV->getSubprogram()).getIndex(), 4);
961     OS.AddComment("Function section relative address");
962     OS.EmitCOFFSecRel32(Fn, /*Offset=*/0);
963     OS.AddComment("Function section index");
964     OS.EmitCOFFSectionIndex(Fn);
965     OS.AddComment("Flags");
966     OS.EmitIntValue(0, 1);
967     // Emit the function display name as a null-terminated string.
968     OS.AddComment("Function name");
969     // Truncate the name so we won't overflow the record length field.
970     emitNullTerminatedSymbolName(OS, FuncName);
971     OS.EmitLabel(ProcRecordEnd);
972 
973     emitLocalVariableList(FI.Locals);
974     emitLexicalBlockList(FI.ChildBlocks, FI);
975 
976     // Emit inlined call site information. Only emit functions inlined directly
977     // into the parent function. We'll emit the other sites recursively as part
978     // of their parent inline site.
979     for (const DILocation *InlinedAt : FI.ChildSites) {
980       auto I = FI.InlineSites.find(InlinedAt);
981       assert(I != FI.InlineSites.end() &&
982              "child site not in function inline site map");
983       emitInlinedCallSite(FI, InlinedAt, I->second);
984     }
985 
986     for (auto Annot : FI.Annotations) {
987       MCSymbol *Label = Annot.first;
988       MDTuple *Strs = cast<MDTuple>(Annot.second);
989       MCSymbol *AnnotBegin = MMI->getContext().createTempSymbol(),
990                *AnnotEnd = MMI->getContext().createTempSymbol();
991       OS.AddComment("Record length");
992       OS.emitAbsoluteSymbolDiff(AnnotEnd, AnnotBegin, 2);
993       OS.EmitLabel(AnnotBegin);
994       OS.AddComment("Record kind: S_ANNOTATION");
995       OS.EmitIntValue(SymbolKind::S_ANNOTATION, 2);
996       OS.EmitCOFFSecRel32(Label, /*Offset=*/0);
997       // FIXME: Make sure we don't overflow the max record size.
998       OS.EmitCOFFSectionIndex(Label);
999       OS.EmitIntValue(Strs->getNumOperands(), 2);
1000       for (Metadata *MD : Strs->operands()) {
1001         // MDStrings are null terminated, so we can do EmitBytes and get the
1002         // nice .asciz directive.
1003         StringRef Str = cast<MDString>(MD)->getString();
1004         assert(Str.data()[Str.size()] == '\0' && "non-nullterminated MDString");
1005         OS.EmitBytes(StringRef(Str.data(), Str.size() + 1));
1006       }
1007       OS.EmitLabel(AnnotEnd);
1008     }
1009 
1010     if (SP != nullptr)
1011       emitDebugInfoForUDTs(LocalUDTs);
1012 
1013     // We're done with this function.
1014     OS.AddComment("Record length");
1015     OS.EmitIntValue(0x0002, 2);
1016     OS.AddComment("Record kind: S_PROC_ID_END");
1017     OS.EmitIntValue(unsigned(SymbolKind::S_PROC_ID_END), 2);
1018   }
1019   endCVSubsection(SymbolsEnd);
1020 
1021   // We have an assembler directive that takes care of the whole line table.
1022   OS.EmitCVLinetableDirective(FI.FuncId, Fn, FI.End);
1023 }
1024 
1025 CodeViewDebug::LocalVarDefRange
createDefRangeMem(uint16_t CVRegister,int Offset)1026 CodeViewDebug::createDefRangeMem(uint16_t CVRegister, int Offset) {
1027   LocalVarDefRange DR;
1028   DR.InMemory = -1;
1029   DR.DataOffset = Offset;
1030   assert(DR.DataOffset == Offset && "truncation");
1031   DR.IsSubfield = 0;
1032   DR.StructOffset = 0;
1033   DR.CVRegister = CVRegister;
1034   return DR;
1035 }
1036 
1037 CodeViewDebug::LocalVarDefRange
createDefRangeGeneral(uint16_t CVRegister,bool InMemory,int Offset,bool IsSubfield,uint16_t StructOffset)1038 CodeViewDebug::createDefRangeGeneral(uint16_t CVRegister, bool InMemory,
1039                                      int Offset, bool IsSubfield,
1040                                      uint16_t StructOffset) {
1041   LocalVarDefRange DR;
1042   DR.InMemory = InMemory;
1043   DR.DataOffset = Offset;
1044   DR.IsSubfield = IsSubfield;
1045   DR.StructOffset = StructOffset;
1046   DR.CVRegister = CVRegister;
1047   return DR;
1048 }
1049 
collectVariableInfoFromMFTable(DenseSet<InlinedVariable> & Processed)1050 void CodeViewDebug::collectVariableInfoFromMFTable(
1051     DenseSet<InlinedVariable> &Processed) {
1052   const MachineFunction &MF = *Asm->MF;
1053   const TargetSubtargetInfo &TSI = MF.getSubtarget();
1054   const TargetFrameLowering *TFI = TSI.getFrameLowering();
1055   const TargetRegisterInfo *TRI = TSI.getRegisterInfo();
1056 
1057   for (const MachineFunction::VariableDbgInfo &VI : MF.getVariableDbgInfo()) {
1058     if (!VI.Var)
1059       continue;
1060     assert(VI.Var->isValidLocationForIntrinsic(VI.Loc) &&
1061            "Expected inlined-at fields to agree");
1062 
1063     Processed.insert(InlinedVariable(VI.Var, VI.Loc->getInlinedAt()));
1064     LexicalScope *Scope = LScopes.findLexicalScope(VI.Loc);
1065 
1066     // If variable scope is not found then skip this variable.
1067     if (!Scope)
1068       continue;
1069 
1070     // If the variable has an attached offset expression, extract it.
1071     // FIXME: Try to handle DW_OP_deref as well.
1072     int64_t ExprOffset = 0;
1073     if (VI.Expr)
1074       if (!VI.Expr->extractIfOffset(ExprOffset))
1075         continue;
1076 
1077     // Get the frame register used and the offset.
1078     unsigned FrameReg = 0;
1079     int FrameOffset = TFI->getFrameIndexReference(*Asm->MF, VI.Slot, FrameReg);
1080     uint16_t CVReg = TRI->getCodeViewRegNum(FrameReg);
1081 
1082     // Calculate the label ranges.
1083     LocalVarDefRange DefRange =
1084         createDefRangeMem(CVReg, FrameOffset + ExprOffset);
1085     for (const InsnRange &Range : Scope->getRanges()) {
1086       const MCSymbol *Begin = getLabelBeforeInsn(Range.first);
1087       const MCSymbol *End = getLabelAfterInsn(Range.second);
1088       End = End ? End : Asm->getFunctionEnd();
1089       DefRange.Ranges.emplace_back(Begin, End);
1090     }
1091 
1092     LocalVariable Var;
1093     Var.DIVar = VI.Var;
1094     Var.DefRanges.emplace_back(std::move(DefRange));
1095     recordLocalVariable(std::move(Var), Scope);
1096   }
1097 }
1098 
canUseReferenceType(const DbgVariableLocation & Loc)1099 static bool canUseReferenceType(const DbgVariableLocation &Loc) {
1100   return !Loc.LoadChain.empty() && Loc.LoadChain.back() == 0;
1101 }
1102 
needsReferenceType(const DbgVariableLocation & Loc)1103 static bool needsReferenceType(const DbgVariableLocation &Loc) {
1104   return Loc.LoadChain.size() == 2 && Loc.LoadChain.back() == 0;
1105 }
1106 
calculateRanges(LocalVariable & Var,const DbgValueHistoryMap::InstrRanges & Ranges)1107 void CodeViewDebug::calculateRanges(
1108     LocalVariable &Var, const DbgValueHistoryMap::InstrRanges &Ranges) {
1109   const TargetRegisterInfo *TRI = Asm->MF->getSubtarget().getRegisterInfo();
1110 
1111   // Calculate the definition ranges.
1112   for (auto I = Ranges.begin(), E = Ranges.end(); I != E; ++I) {
1113     const InsnRange &Range = *I;
1114     const MachineInstr *DVInst = Range.first;
1115     assert(DVInst->isDebugValue() && "Invalid History entry");
1116     // FIXME: Find a way to represent constant variables, since they are
1117     // relatively common.
1118     Optional<DbgVariableLocation> Location =
1119         DbgVariableLocation::extractFromMachineInstruction(*DVInst);
1120     if (!Location)
1121       continue;
1122 
1123     // CodeView can only express variables in register and variables in memory
1124     // at a constant offset from a register. However, for variables passed
1125     // indirectly by pointer, it is common for that pointer to be spilled to a
1126     // stack location. For the special case of one offseted load followed by a
1127     // zero offset load (a pointer spilled to the stack), we change the type of
1128     // the local variable from a value type to a reference type. This tricks the
1129     // debugger into doing the load for us.
1130     if (Var.UseReferenceType) {
1131       // We're using a reference type. Drop the last zero offset load.
1132       if (canUseReferenceType(*Location))
1133         Location->LoadChain.pop_back();
1134       else
1135         continue;
1136     } else if (needsReferenceType(*Location)) {
1137       // This location can't be expressed without switching to a reference type.
1138       // Start over using that.
1139       Var.UseReferenceType = true;
1140       Var.DefRanges.clear();
1141       calculateRanges(Var, Ranges);
1142       return;
1143     }
1144 
1145     // We can only handle a register or an offseted load of a register.
1146     if (Location->Register == 0 || Location->LoadChain.size() > 1)
1147       continue;
1148     {
1149       LocalVarDefRange DR;
1150       DR.CVRegister = TRI->getCodeViewRegNum(Location->Register);
1151       DR.InMemory = !Location->LoadChain.empty();
1152       DR.DataOffset =
1153           !Location->LoadChain.empty() ? Location->LoadChain.back() : 0;
1154       if (Location->FragmentInfo) {
1155         DR.IsSubfield = true;
1156         DR.StructOffset = Location->FragmentInfo->OffsetInBits / 8;
1157       } else {
1158         DR.IsSubfield = false;
1159         DR.StructOffset = 0;
1160       }
1161 
1162       if (Var.DefRanges.empty() ||
1163           Var.DefRanges.back().isDifferentLocation(DR)) {
1164         Var.DefRanges.emplace_back(std::move(DR));
1165       }
1166     }
1167 
1168     // Compute the label range.
1169     const MCSymbol *Begin = getLabelBeforeInsn(Range.first);
1170     const MCSymbol *End = getLabelAfterInsn(Range.second);
1171     if (!End) {
1172       // This range is valid until the next overlapping bitpiece. In the
1173       // common case, ranges will not be bitpieces, so they will overlap.
1174       auto J = std::next(I);
1175       const DIExpression *DIExpr = DVInst->getDebugExpression();
1176       while (J != E &&
1177              !DIExpr->fragmentsOverlap(J->first->getDebugExpression()))
1178         ++J;
1179       if (J != E)
1180         End = getLabelBeforeInsn(J->first);
1181       else
1182         End = Asm->getFunctionEnd();
1183     }
1184 
1185     // If the last range end is our begin, just extend the last range.
1186     // Otherwise make a new range.
1187     SmallVectorImpl<std::pair<const MCSymbol *, const MCSymbol *>> &R =
1188         Var.DefRanges.back().Ranges;
1189     if (!R.empty() && R.back().second == Begin)
1190       R.back().second = End;
1191     else
1192       R.emplace_back(Begin, End);
1193 
1194     // FIXME: Do more range combining.
1195   }
1196 }
1197 
collectVariableInfo(const DISubprogram * SP)1198 void CodeViewDebug::collectVariableInfo(const DISubprogram *SP) {
1199   DenseSet<InlinedVariable> Processed;
1200   // Grab the variable info that was squirreled away in the MMI side-table.
1201   collectVariableInfoFromMFTable(Processed);
1202 
1203   for (const auto &I : DbgValues) {
1204     InlinedVariable IV = I.first;
1205     if (Processed.count(IV))
1206       continue;
1207     const DILocalVariable *DIVar = IV.first;
1208     const DILocation *InlinedAt = IV.second;
1209 
1210     // Instruction ranges, specifying where IV is accessible.
1211     const auto &Ranges = I.second;
1212 
1213     LexicalScope *Scope = nullptr;
1214     if (InlinedAt)
1215       Scope = LScopes.findInlinedScope(DIVar->getScope(), InlinedAt);
1216     else
1217       Scope = LScopes.findLexicalScope(DIVar->getScope());
1218     // If variable scope is not found then skip this variable.
1219     if (!Scope)
1220       continue;
1221 
1222     LocalVariable Var;
1223     Var.DIVar = DIVar;
1224 
1225     calculateRanges(Var, Ranges);
1226     recordLocalVariable(std::move(Var), Scope);
1227   }
1228 }
1229 
beginFunctionImpl(const MachineFunction * MF)1230 void CodeViewDebug::beginFunctionImpl(const MachineFunction *MF) {
1231   const Function &GV = MF->getFunction();
1232   auto Insertion = FnDebugInfo.insert({&GV, llvm::make_unique<FunctionInfo>()});
1233   assert(Insertion.second && "function already has info");
1234   CurFn = Insertion.first->second.get();
1235   CurFn->FuncId = NextFuncId++;
1236   CurFn->Begin = Asm->getFunctionBegin();
1237 
1238   OS.EmitCVFuncIdDirective(CurFn->FuncId);
1239 
1240   // Find the end of the function prolog.  First known non-DBG_VALUE and
1241   // non-frame setup location marks the beginning of the function body.
1242   // FIXME: is there a simpler a way to do this? Can we just search
1243   // for the first instruction of the function, not the last of the prolog?
1244   DebugLoc PrologEndLoc;
1245   bool EmptyPrologue = true;
1246   for (const auto &MBB : *MF) {
1247     for (const auto &MI : MBB) {
1248       if (!MI.isMetaInstruction() && !MI.getFlag(MachineInstr::FrameSetup) &&
1249           MI.getDebugLoc()) {
1250         PrologEndLoc = MI.getDebugLoc();
1251         break;
1252       } else if (!MI.isMetaInstruction()) {
1253         EmptyPrologue = false;
1254       }
1255     }
1256   }
1257 
1258   // Record beginning of function if we have a non-empty prologue.
1259   if (PrologEndLoc && !EmptyPrologue) {
1260     DebugLoc FnStartDL = PrologEndLoc.getFnDebugLoc();
1261     maybeRecordLocation(FnStartDL, MF);
1262   }
1263 }
1264 
shouldEmitUdt(const DIType * T)1265 static bool shouldEmitUdt(const DIType *T) {
1266   if (!T)
1267     return false;
1268 
1269   // MSVC does not emit UDTs for typedefs that are scoped to classes.
1270   if (T->getTag() == dwarf::DW_TAG_typedef) {
1271     if (DIScope *Scope = T->getScope().resolve()) {
1272       switch (Scope->getTag()) {
1273       case dwarf::DW_TAG_structure_type:
1274       case dwarf::DW_TAG_class_type:
1275       case dwarf::DW_TAG_union_type:
1276         return false;
1277       }
1278     }
1279   }
1280 
1281   while (true) {
1282     if (!T || T->isForwardDecl())
1283       return false;
1284 
1285     const DIDerivedType *DT = dyn_cast<DIDerivedType>(T);
1286     if (!DT)
1287       return true;
1288     T = DT->getBaseType().resolve();
1289   }
1290   return true;
1291 }
1292 
addToUDTs(const DIType * Ty)1293 void CodeViewDebug::addToUDTs(const DIType *Ty) {
1294   // Don't record empty UDTs.
1295   if (Ty->getName().empty())
1296     return;
1297   if (!shouldEmitUdt(Ty))
1298     return;
1299 
1300   SmallVector<StringRef, 5> QualifiedNameComponents;
1301   const DISubprogram *ClosestSubprogram = getQualifiedNameComponents(
1302       Ty->getScope().resolve(), QualifiedNameComponents);
1303 
1304   std::string FullyQualifiedName =
1305       getQualifiedName(QualifiedNameComponents, getPrettyScopeName(Ty));
1306 
1307   if (ClosestSubprogram == nullptr) {
1308     GlobalUDTs.emplace_back(std::move(FullyQualifiedName), Ty);
1309   } else if (ClosestSubprogram == CurrentSubprogram) {
1310     LocalUDTs.emplace_back(std::move(FullyQualifiedName), Ty);
1311   }
1312 
1313   // TODO: What if the ClosestSubprogram is neither null or the current
1314   // subprogram?  Currently, the UDT just gets dropped on the floor.
1315   //
1316   // The current behavior is not desirable.  To get maximal fidelity, we would
1317   // need to perform all type translation before beginning emission of .debug$S
1318   // and then make LocalUDTs a member of FunctionInfo
1319 }
1320 
lowerType(const DIType * Ty,const DIType * ClassTy)1321 TypeIndex CodeViewDebug::lowerType(const DIType *Ty, const DIType *ClassTy) {
1322   // Generic dispatch for lowering an unknown type.
1323   switch (Ty->getTag()) {
1324   case dwarf::DW_TAG_array_type:
1325     return lowerTypeArray(cast<DICompositeType>(Ty));
1326   case dwarf::DW_TAG_typedef:
1327     return lowerTypeAlias(cast<DIDerivedType>(Ty));
1328   case dwarf::DW_TAG_base_type:
1329     return lowerTypeBasic(cast<DIBasicType>(Ty));
1330   case dwarf::DW_TAG_pointer_type:
1331     if (cast<DIDerivedType>(Ty)->getName() == "__vtbl_ptr_type")
1332       return lowerTypeVFTableShape(cast<DIDerivedType>(Ty));
1333     LLVM_FALLTHROUGH;
1334   case dwarf::DW_TAG_reference_type:
1335   case dwarf::DW_TAG_rvalue_reference_type:
1336     return lowerTypePointer(cast<DIDerivedType>(Ty));
1337   case dwarf::DW_TAG_ptr_to_member_type:
1338     return lowerTypeMemberPointer(cast<DIDerivedType>(Ty));
1339   case dwarf::DW_TAG_restrict_type:
1340   case dwarf::DW_TAG_const_type:
1341   case dwarf::DW_TAG_volatile_type:
1342   // TODO: add support for DW_TAG_atomic_type here
1343     return lowerTypeModifier(cast<DIDerivedType>(Ty));
1344   case dwarf::DW_TAG_subroutine_type:
1345     if (ClassTy) {
1346       // The member function type of a member function pointer has no
1347       // ThisAdjustment.
1348       return lowerTypeMemberFunction(cast<DISubroutineType>(Ty), ClassTy,
1349                                      /*ThisAdjustment=*/0,
1350                                      /*IsStaticMethod=*/false);
1351     }
1352     return lowerTypeFunction(cast<DISubroutineType>(Ty));
1353   case dwarf::DW_TAG_enumeration_type:
1354     return lowerTypeEnum(cast<DICompositeType>(Ty));
1355   case dwarf::DW_TAG_class_type:
1356   case dwarf::DW_TAG_structure_type:
1357     return lowerTypeClass(cast<DICompositeType>(Ty));
1358   case dwarf::DW_TAG_union_type:
1359     return lowerTypeUnion(cast<DICompositeType>(Ty));
1360   case dwarf::DW_TAG_unspecified_type:
1361     return TypeIndex::None();
1362   default:
1363     // Use the null type index.
1364     return TypeIndex();
1365   }
1366 }
1367 
lowerTypeAlias(const DIDerivedType * Ty)1368 TypeIndex CodeViewDebug::lowerTypeAlias(const DIDerivedType *Ty) {
1369   DITypeRef UnderlyingTypeRef = Ty->getBaseType();
1370   TypeIndex UnderlyingTypeIndex = getTypeIndex(UnderlyingTypeRef);
1371   StringRef TypeName = Ty->getName();
1372 
1373   addToUDTs(Ty);
1374 
1375   if (UnderlyingTypeIndex == TypeIndex(SimpleTypeKind::Int32Long) &&
1376       TypeName == "HRESULT")
1377     return TypeIndex(SimpleTypeKind::HResult);
1378   if (UnderlyingTypeIndex == TypeIndex(SimpleTypeKind::UInt16Short) &&
1379       TypeName == "wchar_t")
1380     return TypeIndex(SimpleTypeKind::WideCharacter);
1381 
1382   return UnderlyingTypeIndex;
1383 }
1384 
lowerTypeArray(const DICompositeType * Ty)1385 TypeIndex CodeViewDebug::lowerTypeArray(const DICompositeType *Ty) {
1386   DITypeRef ElementTypeRef = Ty->getBaseType();
1387   TypeIndex ElementTypeIndex = getTypeIndex(ElementTypeRef);
1388   // IndexType is size_t, which depends on the bitness of the target.
1389   TypeIndex IndexType = getPointerSizeInBytes() == 8
1390                             ? TypeIndex(SimpleTypeKind::UInt64Quad)
1391                             : TypeIndex(SimpleTypeKind::UInt32Long);
1392 
1393   uint64_t ElementSize = getBaseTypeSize(ElementTypeRef) / 8;
1394 
1395   // Add subranges to array type.
1396   DINodeArray Elements = Ty->getElements();
1397   for (int i = Elements.size() - 1; i >= 0; --i) {
1398     const DINode *Element = Elements[i];
1399     assert(Element->getTag() == dwarf::DW_TAG_subrange_type);
1400 
1401     const DISubrange *Subrange = cast<DISubrange>(Element);
1402     assert(Subrange->getLowerBound() == 0 &&
1403            "codeview doesn't support subranges with lower bounds");
1404     int64_t Count = -1;
1405     if (auto *CI = Subrange->getCount().dyn_cast<ConstantInt*>())
1406       Count = CI->getSExtValue();
1407 
1408     // Forward declarations of arrays without a size and VLAs use a count of -1.
1409     // Emit a count of zero in these cases to match what MSVC does for arrays
1410     // without a size. MSVC doesn't support VLAs, so it's not clear what we
1411     // should do for them even if we could distinguish them.
1412     if (Count == -1)
1413       Count = 0;
1414 
1415     // Update the element size and element type index for subsequent subranges.
1416     ElementSize *= Count;
1417 
1418     // If this is the outermost array, use the size from the array. It will be
1419     // more accurate if we had a VLA or an incomplete element type size.
1420     uint64_t ArraySize =
1421         (i == 0 && ElementSize == 0) ? Ty->getSizeInBits() / 8 : ElementSize;
1422 
1423     StringRef Name = (i == 0) ? Ty->getName() : "";
1424     ArrayRecord AR(ElementTypeIndex, IndexType, ArraySize, Name);
1425     ElementTypeIndex = TypeTable.writeLeafType(AR);
1426   }
1427 
1428   return ElementTypeIndex;
1429 }
1430 
lowerTypeBasic(const DIBasicType * Ty)1431 TypeIndex CodeViewDebug::lowerTypeBasic(const DIBasicType *Ty) {
1432   TypeIndex Index;
1433   dwarf::TypeKind Kind;
1434   uint32_t ByteSize;
1435 
1436   Kind = static_cast<dwarf::TypeKind>(Ty->getEncoding());
1437   ByteSize = Ty->getSizeInBits() / 8;
1438 
1439   SimpleTypeKind STK = SimpleTypeKind::None;
1440   switch (Kind) {
1441   case dwarf::DW_ATE_address:
1442     // FIXME: Translate
1443     break;
1444   case dwarf::DW_ATE_boolean:
1445     switch (ByteSize) {
1446     case 1:  STK = SimpleTypeKind::Boolean8;   break;
1447     case 2:  STK = SimpleTypeKind::Boolean16;  break;
1448     case 4:  STK = SimpleTypeKind::Boolean32;  break;
1449     case 8:  STK = SimpleTypeKind::Boolean64;  break;
1450     case 16: STK = SimpleTypeKind::Boolean128; break;
1451     }
1452     break;
1453   case dwarf::DW_ATE_complex_float:
1454     switch (ByteSize) {
1455     case 2:  STK = SimpleTypeKind::Complex16;  break;
1456     case 4:  STK = SimpleTypeKind::Complex32;  break;
1457     case 8:  STK = SimpleTypeKind::Complex64;  break;
1458     case 10: STK = SimpleTypeKind::Complex80;  break;
1459     case 16: STK = SimpleTypeKind::Complex128; break;
1460     }
1461     break;
1462   case dwarf::DW_ATE_float:
1463     switch (ByteSize) {
1464     case 2:  STK = SimpleTypeKind::Float16;  break;
1465     case 4:  STK = SimpleTypeKind::Float32;  break;
1466     case 6:  STK = SimpleTypeKind::Float48;  break;
1467     case 8:  STK = SimpleTypeKind::Float64;  break;
1468     case 10: STK = SimpleTypeKind::Float80;  break;
1469     case 16: STK = SimpleTypeKind::Float128; break;
1470     }
1471     break;
1472   case dwarf::DW_ATE_signed:
1473     switch (ByteSize) {
1474     case 1:  STK = SimpleTypeKind::SignedCharacter; break;
1475     case 2:  STK = SimpleTypeKind::Int16Short;      break;
1476     case 4:  STK = SimpleTypeKind::Int32;           break;
1477     case 8:  STK = SimpleTypeKind::Int64Quad;       break;
1478     case 16: STK = SimpleTypeKind::Int128Oct;       break;
1479     }
1480     break;
1481   case dwarf::DW_ATE_unsigned:
1482     switch (ByteSize) {
1483     case 1:  STK = SimpleTypeKind::UnsignedCharacter; break;
1484     case 2:  STK = SimpleTypeKind::UInt16Short;       break;
1485     case 4:  STK = SimpleTypeKind::UInt32;            break;
1486     case 8:  STK = SimpleTypeKind::UInt64Quad;        break;
1487     case 16: STK = SimpleTypeKind::UInt128Oct;        break;
1488     }
1489     break;
1490   case dwarf::DW_ATE_UTF:
1491     switch (ByteSize) {
1492     case 2: STK = SimpleTypeKind::Character16; break;
1493     case 4: STK = SimpleTypeKind::Character32; break;
1494     }
1495     break;
1496   case dwarf::DW_ATE_signed_char:
1497     if (ByteSize == 1)
1498       STK = SimpleTypeKind::SignedCharacter;
1499     break;
1500   case dwarf::DW_ATE_unsigned_char:
1501     if (ByteSize == 1)
1502       STK = SimpleTypeKind::UnsignedCharacter;
1503     break;
1504   default:
1505     break;
1506   }
1507 
1508   // Apply some fixups based on the source-level type name.
1509   if (STK == SimpleTypeKind::Int32 && Ty->getName() == "long int")
1510     STK = SimpleTypeKind::Int32Long;
1511   if (STK == SimpleTypeKind::UInt32 && Ty->getName() == "long unsigned int")
1512     STK = SimpleTypeKind::UInt32Long;
1513   if (STK == SimpleTypeKind::UInt16Short &&
1514       (Ty->getName() == "wchar_t" || Ty->getName() == "__wchar_t"))
1515     STK = SimpleTypeKind::WideCharacter;
1516   if ((STK == SimpleTypeKind::SignedCharacter ||
1517        STK == SimpleTypeKind::UnsignedCharacter) &&
1518       Ty->getName() == "char")
1519     STK = SimpleTypeKind::NarrowCharacter;
1520 
1521   return TypeIndex(STK);
1522 }
1523 
lowerTypePointer(const DIDerivedType * Ty,PointerOptions PO)1524 TypeIndex CodeViewDebug::lowerTypePointer(const DIDerivedType *Ty,
1525                                           PointerOptions PO) {
1526   TypeIndex PointeeTI = getTypeIndex(Ty->getBaseType());
1527 
1528   // Pointers to simple types without any options can use SimpleTypeMode, rather
1529   // than having a dedicated pointer type record.
1530   if (PointeeTI.isSimple() && PO == PointerOptions::None &&
1531       PointeeTI.getSimpleMode() == SimpleTypeMode::Direct &&
1532       Ty->getTag() == dwarf::DW_TAG_pointer_type) {
1533     SimpleTypeMode Mode = Ty->getSizeInBits() == 64
1534                               ? SimpleTypeMode::NearPointer64
1535                               : SimpleTypeMode::NearPointer32;
1536     return TypeIndex(PointeeTI.getSimpleKind(), Mode);
1537   }
1538 
1539   PointerKind PK =
1540       Ty->getSizeInBits() == 64 ? PointerKind::Near64 : PointerKind::Near32;
1541   PointerMode PM = PointerMode::Pointer;
1542   switch (Ty->getTag()) {
1543   default: llvm_unreachable("not a pointer tag type");
1544   case dwarf::DW_TAG_pointer_type:
1545     PM = PointerMode::Pointer;
1546     break;
1547   case dwarf::DW_TAG_reference_type:
1548     PM = PointerMode::LValueReference;
1549     break;
1550   case dwarf::DW_TAG_rvalue_reference_type:
1551     PM = PointerMode::RValueReference;
1552     break;
1553   }
1554 
1555   PointerRecord PR(PointeeTI, PK, PM, PO, Ty->getSizeInBits() / 8);
1556   return TypeTable.writeLeafType(PR);
1557 }
1558 
1559 static PointerToMemberRepresentation
translatePtrToMemberRep(unsigned SizeInBytes,bool IsPMF,unsigned Flags)1560 translatePtrToMemberRep(unsigned SizeInBytes, bool IsPMF, unsigned Flags) {
1561   // SizeInBytes being zero generally implies that the member pointer type was
1562   // incomplete, which can happen if it is part of a function prototype. In this
1563   // case, use the unknown model instead of the general model.
1564   if (IsPMF) {
1565     switch (Flags & DINode::FlagPtrToMemberRep) {
1566     case 0:
1567       return SizeInBytes == 0 ? PointerToMemberRepresentation::Unknown
1568                               : PointerToMemberRepresentation::GeneralFunction;
1569     case DINode::FlagSingleInheritance:
1570       return PointerToMemberRepresentation::SingleInheritanceFunction;
1571     case DINode::FlagMultipleInheritance:
1572       return PointerToMemberRepresentation::MultipleInheritanceFunction;
1573     case DINode::FlagVirtualInheritance:
1574       return PointerToMemberRepresentation::VirtualInheritanceFunction;
1575     }
1576   } else {
1577     switch (Flags & DINode::FlagPtrToMemberRep) {
1578     case 0:
1579       return SizeInBytes == 0 ? PointerToMemberRepresentation::Unknown
1580                               : PointerToMemberRepresentation::GeneralData;
1581     case DINode::FlagSingleInheritance:
1582       return PointerToMemberRepresentation::SingleInheritanceData;
1583     case DINode::FlagMultipleInheritance:
1584       return PointerToMemberRepresentation::MultipleInheritanceData;
1585     case DINode::FlagVirtualInheritance:
1586       return PointerToMemberRepresentation::VirtualInheritanceData;
1587     }
1588   }
1589   llvm_unreachable("invalid ptr to member representation");
1590 }
1591 
lowerTypeMemberPointer(const DIDerivedType * Ty,PointerOptions PO)1592 TypeIndex CodeViewDebug::lowerTypeMemberPointer(const DIDerivedType *Ty,
1593                                                 PointerOptions PO) {
1594   assert(Ty->getTag() == dwarf::DW_TAG_ptr_to_member_type);
1595   TypeIndex ClassTI = getTypeIndex(Ty->getClassType());
1596   TypeIndex PointeeTI = getTypeIndex(Ty->getBaseType(), Ty->getClassType());
1597   PointerKind PK = getPointerSizeInBytes() == 8 ? PointerKind::Near64
1598                                                 : PointerKind::Near32;
1599   bool IsPMF = isa<DISubroutineType>(Ty->getBaseType());
1600   PointerMode PM = IsPMF ? PointerMode::PointerToMemberFunction
1601                          : PointerMode::PointerToDataMember;
1602 
1603   assert(Ty->getSizeInBits() / 8 <= 0xff && "pointer size too big");
1604   uint8_t SizeInBytes = Ty->getSizeInBits() / 8;
1605   MemberPointerInfo MPI(
1606       ClassTI, translatePtrToMemberRep(SizeInBytes, IsPMF, Ty->getFlags()));
1607   PointerRecord PR(PointeeTI, PK, PM, PO, SizeInBytes, MPI);
1608   return TypeTable.writeLeafType(PR);
1609 }
1610 
1611 /// Given a DWARF calling convention, get the CodeView equivalent. If we don't
1612 /// have a translation, use the NearC convention.
dwarfCCToCodeView(unsigned DwarfCC)1613 static CallingConvention dwarfCCToCodeView(unsigned DwarfCC) {
1614   switch (DwarfCC) {
1615   case dwarf::DW_CC_normal:             return CallingConvention::NearC;
1616   case dwarf::DW_CC_BORLAND_msfastcall: return CallingConvention::NearFast;
1617   case dwarf::DW_CC_BORLAND_thiscall:   return CallingConvention::ThisCall;
1618   case dwarf::DW_CC_BORLAND_stdcall:    return CallingConvention::NearStdCall;
1619   case dwarf::DW_CC_BORLAND_pascal:     return CallingConvention::NearPascal;
1620   case dwarf::DW_CC_LLVM_vectorcall:    return CallingConvention::NearVector;
1621   }
1622   return CallingConvention::NearC;
1623 }
1624 
lowerTypeModifier(const DIDerivedType * Ty)1625 TypeIndex CodeViewDebug::lowerTypeModifier(const DIDerivedType *Ty) {
1626   ModifierOptions Mods = ModifierOptions::None;
1627   PointerOptions PO = PointerOptions::None;
1628   bool IsModifier = true;
1629   const DIType *BaseTy = Ty;
1630   while (IsModifier && BaseTy) {
1631     // FIXME: Need to add DWARF tags for __unaligned and _Atomic
1632     switch (BaseTy->getTag()) {
1633     case dwarf::DW_TAG_const_type:
1634       Mods |= ModifierOptions::Const;
1635       PO |= PointerOptions::Const;
1636       break;
1637     case dwarf::DW_TAG_volatile_type:
1638       Mods |= ModifierOptions::Volatile;
1639       PO |= PointerOptions::Volatile;
1640       break;
1641     case dwarf::DW_TAG_restrict_type:
1642       // Only pointer types be marked with __restrict. There is no known flag
1643       // for __restrict in LF_MODIFIER records.
1644       PO |= PointerOptions::Restrict;
1645       break;
1646     default:
1647       IsModifier = false;
1648       break;
1649     }
1650     if (IsModifier)
1651       BaseTy = cast<DIDerivedType>(BaseTy)->getBaseType().resolve();
1652   }
1653 
1654   // Check if the inner type will use an LF_POINTER record. If so, the
1655   // qualifiers will go in the LF_POINTER record. This comes up for types like
1656   // 'int *const' and 'int *__restrict', not the more common cases like 'const
1657   // char *'.
1658   if (BaseTy) {
1659     switch (BaseTy->getTag()) {
1660     case dwarf::DW_TAG_pointer_type:
1661     case dwarf::DW_TAG_reference_type:
1662     case dwarf::DW_TAG_rvalue_reference_type:
1663       return lowerTypePointer(cast<DIDerivedType>(BaseTy), PO);
1664     case dwarf::DW_TAG_ptr_to_member_type:
1665       return lowerTypeMemberPointer(cast<DIDerivedType>(BaseTy), PO);
1666     default:
1667       break;
1668     }
1669   }
1670 
1671   TypeIndex ModifiedTI = getTypeIndex(BaseTy);
1672 
1673   // Return the base type index if there aren't any modifiers. For example, the
1674   // metadata could contain restrict wrappers around non-pointer types.
1675   if (Mods == ModifierOptions::None)
1676     return ModifiedTI;
1677 
1678   ModifierRecord MR(ModifiedTI, Mods);
1679   return TypeTable.writeLeafType(MR);
1680 }
1681 
lowerTypeFunction(const DISubroutineType * Ty)1682 TypeIndex CodeViewDebug::lowerTypeFunction(const DISubroutineType *Ty) {
1683   SmallVector<TypeIndex, 8> ReturnAndArgTypeIndices;
1684   for (DITypeRef ArgTypeRef : Ty->getTypeArray())
1685     ReturnAndArgTypeIndices.push_back(getTypeIndex(ArgTypeRef));
1686 
1687   // MSVC uses type none for variadic argument.
1688   if (ReturnAndArgTypeIndices.size() > 1 &&
1689       ReturnAndArgTypeIndices.back() == TypeIndex::Void()) {
1690     ReturnAndArgTypeIndices.back() = TypeIndex::None();
1691   }
1692   TypeIndex ReturnTypeIndex = TypeIndex::Void();
1693   ArrayRef<TypeIndex> ArgTypeIndices = None;
1694   if (!ReturnAndArgTypeIndices.empty()) {
1695     auto ReturnAndArgTypesRef = makeArrayRef(ReturnAndArgTypeIndices);
1696     ReturnTypeIndex = ReturnAndArgTypesRef.front();
1697     ArgTypeIndices = ReturnAndArgTypesRef.drop_front();
1698   }
1699 
1700   ArgListRecord ArgListRec(TypeRecordKind::ArgList, ArgTypeIndices);
1701   TypeIndex ArgListIndex = TypeTable.writeLeafType(ArgListRec);
1702 
1703   CallingConvention CC = dwarfCCToCodeView(Ty->getCC());
1704 
1705   ProcedureRecord Procedure(ReturnTypeIndex, CC, FunctionOptions::None,
1706                             ArgTypeIndices.size(), ArgListIndex);
1707   return TypeTable.writeLeafType(Procedure);
1708 }
1709 
lowerTypeMemberFunction(const DISubroutineType * Ty,const DIType * ClassTy,int ThisAdjustment,bool IsStaticMethod)1710 TypeIndex CodeViewDebug::lowerTypeMemberFunction(const DISubroutineType *Ty,
1711                                                  const DIType *ClassTy,
1712                                                  int ThisAdjustment,
1713                                                  bool IsStaticMethod) {
1714   // Lower the containing class type.
1715   TypeIndex ClassType = getTypeIndex(ClassTy);
1716 
1717   SmallVector<TypeIndex, 8> ReturnAndArgTypeIndices;
1718   for (DITypeRef ArgTypeRef : Ty->getTypeArray())
1719     ReturnAndArgTypeIndices.push_back(getTypeIndex(ArgTypeRef));
1720 
1721   // MSVC uses type none for variadic argument.
1722   if (ReturnAndArgTypeIndices.size() > 1 &&
1723       ReturnAndArgTypeIndices.back() == TypeIndex::Void()) {
1724     ReturnAndArgTypeIndices.back() = TypeIndex::None();
1725   }
1726   TypeIndex ReturnTypeIndex = TypeIndex::Void();
1727   ArrayRef<TypeIndex> ArgTypeIndices = None;
1728   if (!ReturnAndArgTypeIndices.empty()) {
1729     auto ReturnAndArgTypesRef = makeArrayRef(ReturnAndArgTypeIndices);
1730     ReturnTypeIndex = ReturnAndArgTypesRef.front();
1731     ArgTypeIndices = ReturnAndArgTypesRef.drop_front();
1732   }
1733   TypeIndex ThisTypeIndex;
1734   if (!IsStaticMethod && !ArgTypeIndices.empty()) {
1735     ThisTypeIndex = ArgTypeIndices.front();
1736     ArgTypeIndices = ArgTypeIndices.drop_front();
1737   }
1738 
1739   ArgListRecord ArgListRec(TypeRecordKind::ArgList, ArgTypeIndices);
1740   TypeIndex ArgListIndex = TypeTable.writeLeafType(ArgListRec);
1741 
1742   CallingConvention CC = dwarfCCToCodeView(Ty->getCC());
1743 
1744   // TODO: Need to use the correct values for FunctionOptions.
1745   MemberFunctionRecord MFR(ReturnTypeIndex, ClassType, ThisTypeIndex, CC,
1746                            FunctionOptions::None, ArgTypeIndices.size(),
1747                            ArgListIndex, ThisAdjustment);
1748   return TypeTable.writeLeafType(MFR);
1749 }
1750 
lowerTypeVFTableShape(const DIDerivedType * Ty)1751 TypeIndex CodeViewDebug::lowerTypeVFTableShape(const DIDerivedType *Ty) {
1752   unsigned VSlotCount =
1753       Ty->getSizeInBits() / (8 * Asm->MAI->getCodePointerSize());
1754   SmallVector<VFTableSlotKind, 4> Slots(VSlotCount, VFTableSlotKind::Near);
1755 
1756   VFTableShapeRecord VFTSR(Slots);
1757   return TypeTable.writeLeafType(VFTSR);
1758 }
1759 
translateAccessFlags(unsigned RecordTag,unsigned Flags)1760 static MemberAccess translateAccessFlags(unsigned RecordTag, unsigned Flags) {
1761   switch (Flags & DINode::FlagAccessibility) {
1762   case DINode::FlagPrivate:   return MemberAccess::Private;
1763   case DINode::FlagPublic:    return MemberAccess::Public;
1764   case DINode::FlagProtected: return MemberAccess::Protected;
1765   case 0:
1766     // If there was no explicit access control, provide the default for the tag.
1767     return RecordTag == dwarf::DW_TAG_class_type ? MemberAccess::Private
1768                                                  : MemberAccess::Public;
1769   }
1770   llvm_unreachable("access flags are exclusive");
1771 }
1772 
translateMethodOptionFlags(const DISubprogram * SP)1773 static MethodOptions translateMethodOptionFlags(const DISubprogram *SP) {
1774   if (SP->isArtificial())
1775     return MethodOptions::CompilerGenerated;
1776 
1777   // FIXME: Handle other MethodOptions.
1778 
1779   return MethodOptions::None;
1780 }
1781 
translateMethodKindFlags(const DISubprogram * SP,bool Introduced)1782 static MethodKind translateMethodKindFlags(const DISubprogram *SP,
1783                                            bool Introduced) {
1784   if (SP->getFlags() & DINode::FlagStaticMember)
1785     return MethodKind::Static;
1786 
1787   switch (SP->getVirtuality()) {
1788   case dwarf::DW_VIRTUALITY_none:
1789     break;
1790   case dwarf::DW_VIRTUALITY_virtual:
1791     return Introduced ? MethodKind::IntroducingVirtual : MethodKind::Virtual;
1792   case dwarf::DW_VIRTUALITY_pure_virtual:
1793     return Introduced ? MethodKind::PureIntroducingVirtual
1794                       : MethodKind::PureVirtual;
1795   default:
1796     llvm_unreachable("unhandled virtuality case");
1797   }
1798 
1799   return MethodKind::Vanilla;
1800 }
1801 
getRecordKind(const DICompositeType * Ty)1802 static TypeRecordKind getRecordKind(const DICompositeType *Ty) {
1803   switch (Ty->getTag()) {
1804   case dwarf::DW_TAG_class_type:     return TypeRecordKind::Class;
1805   case dwarf::DW_TAG_structure_type: return TypeRecordKind::Struct;
1806   }
1807   llvm_unreachable("unexpected tag");
1808 }
1809 
1810 /// Return ClassOptions that should be present on both the forward declaration
1811 /// and the defintion of a tag type.
getCommonClassOptions(const DICompositeType * Ty)1812 static ClassOptions getCommonClassOptions(const DICompositeType *Ty) {
1813   ClassOptions CO = ClassOptions::None;
1814 
1815   // MSVC always sets this flag, even for local types. Clang doesn't always
1816   // appear to give every type a linkage name, which may be problematic for us.
1817   // FIXME: Investigate the consequences of not following them here.
1818   if (!Ty->getIdentifier().empty())
1819     CO |= ClassOptions::HasUniqueName;
1820 
1821   // Put the Nested flag on a type if it appears immediately inside a tag type.
1822   // Do not walk the scope chain. Do not attempt to compute ContainsNestedClass
1823   // here. That flag is only set on definitions, and not forward declarations.
1824   const DIScope *ImmediateScope = Ty->getScope().resolve();
1825   if (ImmediateScope && isa<DICompositeType>(ImmediateScope))
1826     CO |= ClassOptions::Nested;
1827 
1828   // Put the Scoped flag on function-local types.
1829   for (const DIScope *Scope = ImmediateScope; Scope != nullptr;
1830        Scope = Scope->getScope().resolve()) {
1831     if (isa<DISubprogram>(Scope)) {
1832       CO |= ClassOptions::Scoped;
1833       break;
1834     }
1835   }
1836 
1837   return CO;
1838 }
1839 
addUDTSrcLine(const DIType * Ty,TypeIndex TI)1840 void CodeViewDebug::addUDTSrcLine(const DIType *Ty, TypeIndex TI) {
1841   switch (Ty->getTag()) {
1842   case dwarf::DW_TAG_class_type:
1843   case dwarf::DW_TAG_structure_type:
1844   case dwarf::DW_TAG_union_type:
1845   case dwarf::DW_TAG_enumeration_type:
1846     break;
1847   default:
1848     return;
1849   }
1850 
1851   if (const auto *File = Ty->getFile()) {
1852     StringIdRecord SIDR(TypeIndex(0x0), getFullFilepath(File));
1853     TypeIndex SIDI = TypeTable.writeLeafType(SIDR);
1854 
1855     UdtSourceLineRecord USLR(TI, SIDI, Ty->getLine());
1856     TypeTable.writeLeafType(USLR);
1857   }
1858 }
1859 
lowerTypeEnum(const DICompositeType * Ty)1860 TypeIndex CodeViewDebug::lowerTypeEnum(const DICompositeType *Ty) {
1861   ClassOptions CO = getCommonClassOptions(Ty);
1862   TypeIndex FTI;
1863   unsigned EnumeratorCount = 0;
1864 
1865   if (Ty->isForwardDecl()) {
1866     CO |= ClassOptions::ForwardReference;
1867   } else {
1868     ContinuationRecordBuilder ContinuationBuilder;
1869     ContinuationBuilder.begin(ContinuationRecordKind::FieldList);
1870     for (const DINode *Element : Ty->getElements()) {
1871       // We assume that the frontend provides all members in source declaration
1872       // order, which is what MSVC does.
1873       if (auto *Enumerator = dyn_cast_or_null<DIEnumerator>(Element)) {
1874         EnumeratorRecord ER(MemberAccess::Public,
1875                             APSInt::getUnsigned(Enumerator->getValue()),
1876                             Enumerator->getName());
1877         ContinuationBuilder.writeMemberType(ER);
1878         EnumeratorCount++;
1879       }
1880     }
1881     FTI = TypeTable.insertRecord(ContinuationBuilder);
1882   }
1883 
1884   std::string FullName = getFullyQualifiedName(Ty);
1885 
1886   EnumRecord ER(EnumeratorCount, CO, FTI, FullName, Ty->getIdentifier(),
1887                 getTypeIndex(Ty->getBaseType()));
1888   TypeIndex EnumTI = TypeTable.writeLeafType(ER);
1889 
1890   addUDTSrcLine(Ty, EnumTI);
1891 
1892   return EnumTI;
1893 }
1894 
1895 //===----------------------------------------------------------------------===//
1896 // ClassInfo
1897 //===----------------------------------------------------------------------===//
1898 
1899 struct llvm::ClassInfo {
1900   struct MemberInfo {
1901     const DIDerivedType *MemberTypeNode;
1902     uint64_t BaseOffset;
1903   };
1904   // [MemberInfo]
1905   using MemberList = std::vector<MemberInfo>;
1906 
1907   using MethodsList = TinyPtrVector<const DISubprogram *>;
1908   // MethodName -> MethodsList
1909   using MethodsMap = MapVector<MDString *, MethodsList>;
1910 
1911   /// Base classes.
1912   std::vector<const DIDerivedType *> Inheritance;
1913 
1914   /// Direct members.
1915   MemberList Members;
1916   // Direct overloaded methods gathered by name.
1917   MethodsMap Methods;
1918 
1919   TypeIndex VShapeTI;
1920 
1921   std::vector<const DIType *> NestedTypes;
1922 };
1923 
clear()1924 void CodeViewDebug::clear() {
1925   assert(CurFn == nullptr);
1926   FileIdMap.clear();
1927   FnDebugInfo.clear();
1928   FileToFilepathMap.clear();
1929   LocalUDTs.clear();
1930   GlobalUDTs.clear();
1931   TypeIndices.clear();
1932   CompleteTypeIndices.clear();
1933 }
1934 
collectMemberInfo(ClassInfo & Info,const DIDerivedType * DDTy)1935 void CodeViewDebug::collectMemberInfo(ClassInfo &Info,
1936                                       const DIDerivedType *DDTy) {
1937   if (!DDTy->getName().empty()) {
1938     Info.Members.push_back({DDTy, 0});
1939     return;
1940   }
1941 
1942   // An unnamed member may represent a nested struct or union. Attempt to
1943   // interpret the unnamed member as a DICompositeType possibly wrapped in
1944   // qualifier types. Add all the indirect fields to the current record if that
1945   // succeeds, and drop the member if that fails.
1946   assert((DDTy->getOffsetInBits() % 8) == 0 && "Unnamed bitfield member!");
1947   uint64_t Offset = DDTy->getOffsetInBits();
1948   const DIType *Ty = DDTy->getBaseType().resolve();
1949   bool FullyResolved = false;
1950   while (!FullyResolved) {
1951     switch (Ty->getTag()) {
1952     case dwarf::DW_TAG_const_type:
1953     case dwarf::DW_TAG_volatile_type:
1954       // FIXME: we should apply the qualifier types to the indirect fields
1955       // rather than dropping them.
1956       Ty = cast<DIDerivedType>(Ty)->getBaseType().resolve();
1957       break;
1958     default:
1959       FullyResolved = true;
1960       break;
1961     }
1962   }
1963 
1964   const DICompositeType *DCTy = dyn_cast<DICompositeType>(Ty);
1965   if (!DCTy)
1966     return;
1967 
1968   ClassInfo NestedInfo = collectClassInfo(DCTy);
1969   for (const ClassInfo::MemberInfo &IndirectField : NestedInfo.Members)
1970     Info.Members.push_back(
1971         {IndirectField.MemberTypeNode, IndirectField.BaseOffset + Offset});
1972 }
1973 
collectClassInfo(const DICompositeType * Ty)1974 ClassInfo CodeViewDebug::collectClassInfo(const DICompositeType *Ty) {
1975   ClassInfo Info;
1976   // Add elements to structure type.
1977   DINodeArray Elements = Ty->getElements();
1978   for (auto *Element : Elements) {
1979     // We assume that the frontend provides all members in source declaration
1980     // order, which is what MSVC does.
1981     if (!Element)
1982       continue;
1983     if (auto *SP = dyn_cast<DISubprogram>(Element)) {
1984       Info.Methods[SP->getRawName()].push_back(SP);
1985     } else if (auto *DDTy = dyn_cast<DIDerivedType>(Element)) {
1986       if (DDTy->getTag() == dwarf::DW_TAG_member) {
1987         collectMemberInfo(Info, DDTy);
1988       } else if (DDTy->getTag() == dwarf::DW_TAG_inheritance) {
1989         Info.Inheritance.push_back(DDTy);
1990       } else if (DDTy->getTag() == dwarf::DW_TAG_pointer_type &&
1991                  DDTy->getName() == "__vtbl_ptr_type") {
1992         Info.VShapeTI = getTypeIndex(DDTy);
1993       } else if (DDTy->getTag() == dwarf::DW_TAG_typedef) {
1994         Info.NestedTypes.push_back(DDTy);
1995       } else if (DDTy->getTag() == dwarf::DW_TAG_friend) {
1996         // Ignore friend members. It appears that MSVC emitted info about
1997         // friends in the past, but modern versions do not.
1998       }
1999     } else if (auto *Composite = dyn_cast<DICompositeType>(Element)) {
2000       Info.NestedTypes.push_back(Composite);
2001     }
2002     // Skip other unrecognized kinds of elements.
2003   }
2004   return Info;
2005 }
2006 
shouldAlwaysEmitCompleteClassType(const DICompositeType * Ty)2007 static bool shouldAlwaysEmitCompleteClassType(const DICompositeType *Ty) {
2008   // This routine is used by lowerTypeClass and lowerTypeUnion to determine
2009   // if a complete type should be emitted instead of a forward reference.
2010   return Ty->getName().empty() && Ty->getIdentifier().empty() &&
2011       !Ty->isForwardDecl();
2012 }
2013 
lowerTypeClass(const DICompositeType * Ty)2014 TypeIndex CodeViewDebug::lowerTypeClass(const DICompositeType *Ty) {
2015   // Emit the complete type for unnamed structs.  C++ classes with methods
2016   // which have a circular reference back to the class type are expected to
2017   // be named by the front-end and should not be "unnamed".  C unnamed
2018   // structs should not have circular references.
2019   if (shouldAlwaysEmitCompleteClassType(Ty)) {
2020     // If this unnamed complete type is already in the process of being defined
2021     // then the description of the type is malformed and cannot be emitted
2022     // into CodeView correctly so report a fatal error.
2023     auto I = CompleteTypeIndices.find(Ty);
2024     if (I != CompleteTypeIndices.end() && I->second == TypeIndex())
2025       report_fatal_error("cannot debug circular reference to unnamed type");
2026     return getCompleteTypeIndex(Ty);
2027   }
2028 
2029   // First, construct the forward decl.  Don't look into Ty to compute the
2030   // forward decl options, since it might not be available in all TUs.
2031   TypeRecordKind Kind = getRecordKind(Ty);
2032   ClassOptions CO =
2033       ClassOptions::ForwardReference | getCommonClassOptions(Ty);
2034   std::string FullName = getFullyQualifiedName(Ty);
2035   ClassRecord CR(Kind, 0, CO, TypeIndex(), TypeIndex(), TypeIndex(), 0,
2036                  FullName, Ty->getIdentifier());
2037   TypeIndex FwdDeclTI = TypeTable.writeLeafType(CR);
2038   if (!Ty->isForwardDecl())
2039     DeferredCompleteTypes.push_back(Ty);
2040   return FwdDeclTI;
2041 }
2042 
lowerCompleteTypeClass(const DICompositeType * Ty)2043 TypeIndex CodeViewDebug::lowerCompleteTypeClass(const DICompositeType *Ty) {
2044   // Construct the field list and complete type record.
2045   TypeRecordKind Kind = getRecordKind(Ty);
2046   ClassOptions CO = getCommonClassOptions(Ty);
2047   TypeIndex FieldTI;
2048   TypeIndex VShapeTI;
2049   unsigned FieldCount;
2050   bool ContainsNestedClass;
2051   std::tie(FieldTI, VShapeTI, FieldCount, ContainsNestedClass) =
2052       lowerRecordFieldList(Ty);
2053 
2054   if (ContainsNestedClass)
2055     CO |= ClassOptions::ContainsNestedClass;
2056 
2057   std::string FullName = getFullyQualifiedName(Ty);
2058 
2059   uint64_t SizeInBytes = Ty->getSizeInBits() / 8;
2060 
2061   ClassRecord CR(Kind, FieldCount, CO, FieldTI, TypeIndex(), VShapeTI,
2062                  SizeInBytes, FullName, Ty->getIdentifier());
2063   TypeIndex ClassTI = TypeTable.writeLeafType(CR);
2064 
2065   addUDTSrcLine(Ty, ClassTI);
2066 
2067   addToUDTs(Ty);
2068 
2069   return ClassTI;
2070 }
2071 
lowerTypeUnion(const DICompositeType * Ty)2072 TypeIndex CodeViewDebug::lowerTypeUnion(const DICompositeType *Ty) {
2073   // Emit the complete type for unnamed unions.
2074   if (shouldAlwaysEmitCompleteClassType(Ty))
2075     return getCompleteTypeIndex(Ty);
2076 
2077   ClassOptions CO =
2078       ClassOptions::ForwardReference | getCommonClassOptions(Ty);
2079   std::string FullName = getFullyQualifiedName(Ty);
2080   UnionRecord UR(0, CO, TypeIndex(), 0, FullName, Ty->getIdentifier());
2081   TypeIndex FwdDeclTI = TypeTable.writeLeafType(UR);
2082   if (!Ty->isForwardDecl())
2083     DeferredCompleteTypes.push_back(Ty);
2084   return FwdDeclTI;
2085 }
2086 
lowerCompleteTypeUnion(const DICompositeType * Ty)2087 TypeIndex CodeViewDebug::lowerCompleteTypeUnion(const DICompositeType *Ty) {
2088   ClassOptions CO = ClassOptions::Sealed | getCommonClassOptions(Ty);
2089   TypeIndex FieldTI;
2090   unsigned FieldCount;
2091   bool ContainsNestedClass;
2092   std::tie(FieldTI, std::ignore, FieldCount, ContainsNestedClass) =
2093       lowerRecordFieldList(Ty);
2094 
2095   if (ContainsNestedClass)
2096     CO |= ClassOptions::ContainsNestedClass;
2097 
2098   uint64_t SizeInBytes = Ty->getSizeInBits() / 8;
2099   std::string FullName = getFullyQualifiedName(Ty);
2100 
2101   UnionRecord UR(FieldCount, CO, FieldTI, SizeInBytes, FullName,
2102                  Ty->getIdentifier());
2103   TypeIndex UnionTI = TypeTable.writeLeafType(UR);
2104 
2105   addUDTSrcLine(Ty, UnionTI);
2106 
2107   addToUDTs(Ty);
2108 
2109   return UnionTI;
2110 }
2111 
2112 std::tuple<TypeIndex, TypeIndex, unsigned, bool>
lowerRecordFieldList(const DICompositeType * Ty)2113 CodeViewDebug::lowerRecordFieldList(const DICompositeType *Ty) {
2114   // Manually count members. MSVC appears to count everything that generates a
2115   // field list record. Each individual overload in a method overload group
2116   // contributes to this count, even though the overload group is a single field
2117   // list record.
2118   unsigned MemberCount = 0;
2119   ClassInfo Info = collectClassInfo(Ty);
2120   ContinuationRecordBuilder ContinuationBuilder;
2121   ContinuationBuilder.begin(ContinuationRecordKind::FieldList);
2122 
2123   // Create base classes.
2124   for (const DIDerivedType *I : Info.Inheritance) {
2125     if (I->getFlags() & DINode::FlagVirtual) {
2126       // Virtual base.
2127       unsigned VBPtrOffset = I->getVBPtrOffset();
2128       // FIXME: Despite the accessor name, the offset is really in bytes.
2129       unsigned VBTableIndex = I->getOffsetInBits() / 4;
2130       auto RecordKind = (I->getFlags() & DINode::FlagIndirectVirtualBase) == DINode::FlagIndirectVirtualBase
2131                             ? TypeRecordKind::IndirectVirtualBaseClass
2132                             : TypeRecordKind::VirtualBaseClass;
2133       VirtualBaseClassRecord VBCR(
2134           RecordKind, translateAccessFlags(Ty->getTag(), I->getFlags()),
2135           getTypeIndex(I->getBaseType()), getVBPTypeIndex(), VBPtrOffset,
2136           VBTableIndex);
2137 
2138       ContinuationBuilder.writeMemberType(VBCR);
2139       MemberCount++;
2140     } else {
2141       assert(I->getOffsetInBits() % 8 == 0 &&
2142              "bases must be on byte boundaries");
2143       BaseClassRecord BCR(translateAccessFlags(Ty->getTag(), I->getFlags()),
2144                           getTypeIndex(I->getBaseType()),
2145                           I->getOffsetInBits() / 8);
2146       ContinuationBuilder.writeMemberType(BCR);
2147       MemberCount++;
2148     }
2149   }
2150 
2151   // Create members.
2152   for (ClassInfo::MemberInfo &MemberInfo : Info.Members) {
2153     const DIDerivedType *Member = MemberInfo.MemberTypeNode;
2154     TypeIndex MemberBaseType = getTypeIndex(Member->getBaseType());
2155     StringRef MemberName = Member->getName();
2156     MemberAccess Access =
2157         translateAccessFlags(Ty->getTag(), Member->getFlags());
2158 
2159     if (Member->isStaticMember()) {
2160       StaticDataMemberRecord SDMR(Access, MemberBaseType, MemberName);
2161       ContinuationBuilder.writeMemberType(SDMR);
2162       MemberCount++;
2163       continue;
2164     }
2165 
2166     // Virtual function pointer member.
2167     if ((Member->getFlags() & DINode::FlagArtificial) &&
2168         Member->getName().startswith("_vptr$")) {
2169       VFPtrRecord VFPR(getTypeIndex(Member->getBaseType()));
2170       ContinuationBuilder.writeMemberType(VFPR);
2171       MemberCount++;
2172       continue;
2173     }
2174 
2175     // Data member.
2176     uint64_t MemberOffsetInBits =
2177         Member->getOffsetInBits() + MemberInfo.BaseOffset;
2178     if (Member->isBitField()) {
2179       uint64_t StartBitOffset = MemberOffsetInBits;
2180       if (const auto *CI =
2181               dyn_cast_or_null<ConstantInt>(Member->getStorageOffsetInBits())) {
2182         MemberOffsetInBits = CI->getZExtValue() + MemberInfo.BaseOffset;
2183       }
2184       StartBitOffset -= MemberOffsetInBits;
2185       BitFieldRecord BFR(MemberBaseType, Member->getSizeInBits(),
2186                          StartBitOffset);
2187       MemberBaseType = TypeTable.writeLeafType(BFR);
2188     }
2189     uint64_t MemberOffsetInBytes = MemberOffsetInBits / 8;
2190     DataMemberRecord DMR(Access, MemberBaseType, MemberOffsetInBytes,
2191                          MemberName);
2192     ContinuationBuilder.writeMemberType(DMR);
2193     MemberCount++;
2194   }
2195 
2196   // Create methods
2197   for (auto &MethodItr : Info.Methods) {
2198     StringRef Name = MethodItr.first->getString();
2199 
2200     std::vector<OneMethodRecord> Methods;
2201     for (const DISubprogram *SP : MethodItr.second) {
2202       TypeIndex MethodType = getMemberFunctionType(SP, Ty);
2203       bool Introduced = SP->getFlags() & DINode::FlagIntroducedVirtual;
2204 
2205       unsigned VFTableOffset = -1;
2206       if (Introduced)
2207         VFTableOffset = SP->getVirtualIndex() * getPointerSizeInBytes();
2208 
2209       Methods.push_back(OneMethodRecord(
2210           MethodType, translateAccessFlags(Ty->getTag(), SP->getFlags()),
2211           translateMethodKindFlags(SP, Introduced),
2212           translateMethodOptionFlags(SP), VFTableOffset, Name));
2213       MemberCount++;
2214     }
2215     assert(!Methods.empty() && "Empty methods map entry");
2216     if (Methods.size() == 1)
2217       ContinuationBuilder.writeMemberType(Methods[0]);
2218     else {
2219       // FIXME: Make this use its own ContinuationBuilder so that
2220       // MethodOverloadList can be split correctly.
2221       MethodOverloadListRecord MOLR(Methods);
2222       TypeIndex MethodList = TypeTable.writeLeafType(MOLR);
2223 
2224       OverloadedMethodRecord OMR(Methods.size(), MethodList, Name);
2225       ContinuationBuilder.writeMemberType(OMR);
2226     }
2227   }
2228 
2229   // Create nested classes.
2230   for (const DIType *Nested : Info.NestedTypes) {
2231     NestedTypeRecord R(getTypeIndex(DITypeRef(Nested)), Nested->getName());
2232     ContinuationBuilder.writeMemberType(R);
2233     MemberCount++;
2234   }
2235 
2236   TypeIndex FieldTI = TypeTable.insertRecord(ContinuationBuilder);
2237   return std::make_tuple(FieldTI, Info.VShapeTI, MemberCount,
2238                          !Info.NestedTypes.empty());
2239 }
2240 
getVBPTypeIndex()2241 TypeIndex CodeViewDebug::getVBPTypeIndex() {
2242   if (!VBPType.getIndex()) {
2243     // Make a 'const int *' type.
2244     ModifierRecord MR(TypeIndex::Int32(), ModifierOptions::Const);
2245     TypeIndex ModifiedTI = TypeTable.writeLeafType(MR);
2246 
2247     PointerKind PK = getPointerSizeInBytes() == 8 ? PointerKind::Near64
2248                                                   : PointerKind::Near32;
2249     PointerMode PM = PointerMode::Pointer;
2250     PointerOptions PO = PointerOptions::None;
2251     PointerRecord PR(ModifiedTI, PK, PM, PO, getPointerSizeInBytes());
2252     VBPType = TypeTable.writeLeafType(PR);
2253   }
2254 
2255   return VBPType;
2256 }
2257 
getTypeIndex(DITypeRef TypeRef,DITypeRef ClassTyRef)2258 TypeIndex CodeViewDebug::getTypeIndex(DITypeRef TypeRef, DITypeRef ClassTyRef) {
2259   const DIType *Ty = TypeRef.resolve();
2260   const DIType *ClassTy = ClassTyRef.resolve();
2261 
2262   // The null DIType is the void type. Don't try to hash it.
2263   if (!Ty)
2264     return TypeIndex::Void();
2265 
2266   // Check if we've already translated this type. Don't try to do a
2267   // get-or-create style insertion that caches the hash lookup across the
2268   // lowerType call. It will update the TypeIndices map.
2269   auto I = TypeIndices.find({Ty, ClassTy});
2270   if (I != TypeIndices.end())
2271     return I->second;
2272 
2273   TypeLoweringScope S(*this);
2274   TypeIndex TI = lowerType(Ty, ClassTy);
2275   return recordTypeIndexForDINode(Ty, TI, ClassTy);
2276 }
2277 
getTypeIndexForReferenceTo(DITypeRef TypeRef)2278 TypeIndex CodeViewDebug::getTypeIndexForReferenceTo(DITypeRef TypeRef) {
2279   DIType *Ty = TypeRef.resolve();
2280   PointerRecord PR(getTypeIndex(Ty),
2281                    getPointerSizeInBytes() == 8 ? PointerKind::Near64
2282                                                 : PointerKind::Near32,
2283                    PointerMode::LValueReference, PointerOptions::None,
2284                    Ty->getSizeInBits() / 8);
2285   return TypeTable.writeLeafType(PR);
2286 }
2287 
getCompleteTypeIndex(DITypeRef TypeRef)2288 TypeIndex CodeViewDebug::getCompleteTypeIndex(DITypeRef TypeRef) {
2289   const DIType *Ty = TypeRef.resolve();
2290 
2291   // The null DIType is the void type. Don't try to hash it.
2292   if (!Ty)
2293     return TypeIndex::Void();
2294 
2295   // If this is a non-record type, the complete type index is the same as the
2296   // normal type index. Just call getTypeIndex.
2297   switch (Ty->getTag()) {
2298   case dwarf::DW_TAG_class_type:
2299   case dwarf::DW_TAG_structure_type:
2300   case dwarf::DW_TAG_union_type:
2301     break;
2302   default:
2303     return getTypeIndex(Ty);
2304   }
2305 
2306   // Check if we've already translated the complete record type.
2307   const auto *CTy = cast<DICompositeType>(Ty);
2308   auto InsertResult = CompleteTypeIndices.insert({CTy, TypeIndex()});
2309   if (!InsertResult.second)
2310     return InsertResult.first->second;
2311 
2312   TypeLoweringScope S(*this);
2313 
2314   // Make sure the forward declaration is emitted first. It's unclear if this
2315   // is necessary, but MSVC does it, and we should follow suit until we can show
2316   // otherwise.
2317   // We only emit a forward declaration for named types.
2318   if (!CTy->getName().empty() || !CTy->getIdentifier().empty()) {
2319     TypeIndex FwdDeclTI = getTypeIndex(CTy);
2320 
2321     // Just use the forward decl if we don't have complete type info. This
2322     // might happen if the frontend is using modules and expects the complete
2323     // definition to be emitted elsewhere.
2324     if (CTy->isForwardDecl())
2325       return FwdDeclTI;
2326   }
2327 
2328   TypeIndex TI;
2329   switch (CTy->getTag()) {
2330   case dwarf::DW_TAG_class_type:
2331   case dwarf::DW_TAG_structure_type:
2332     TI = lowerCompleteTypeClass(CTy);
2333     break;
2334   case dwarf::DW_TAG_union_type:
2335     TI = lowerCompleteTypeUnion(CTy);
2336     break;
2337   default:
2338     llvm_unreachable("not a record");
2339   }
2340 
2341   // Update the type index associated with this CompositeType.  This cannot
2342   // use the 'InsertResult' iterator above because it is potentially
2343   // invalidated by map insertions which can occur while lowering the class
2344   // type above.
2345   CompleteTypeIndices[CTy] = TI;
2346   return TI;
2347 }
2348 
2349 /// Emit all the deferred complete record types. Try to do this in FIFO order,
2350 /// and do this until fixpoint, as each complete record type typically
2351 /// references
2352 /// many other record types.
emitDeferredCompleteTypes()2353 void CodeViewDebug::emitDeferredCompleteTypes() {
2354   SmallVector<const DICompositeType *, 4> TypesToEmit;
2355   while (!DeferredCompleteTypes.empty()) {
2356     std::swap(DeferredCompleteTypes, TypesToEmit);
2357     for (const DICompositeType *RecordTy : TypesToEmit)
2358       getCompleteTypeIndex(RecordTy);
2359     TypesToEmit.clear();
2360   }
2361 }
2362 
emitLocalVariableList(ArrayRef<LocalVariable> Locals)2363 void CodeViewDebug::emitLocalVariableList(ArrayRef<LocalVariable> Locals) {
2364   // Get the sorted list of parameters and emit them first.
2365   SmallVector<const LocalVariable *, 6> Params;
2366   for (const LocalVariable &L : Locals)
2367     if (L.DIVar->isParameter())
2368       Params.push_back(&L);
2369   llvm::sort(Params.begin(), Params.end(),
2370              [](const LocalVariable *L, const LocalVariable *R) {
2371                return L->DIVar->getArg() < R->DIVar->getArg();
2372              });
2373   for (const LocalVariable *L : Params)
2374     emitLocalVariable(*L);
2375 
2376   // Next emit all non-parameters in the order that we found them.
2377   for (const LocalVariable &L : Locals)
2378     if (!L.DIVar->isParameter())
2379       emitLocalVariable(L);
2380 }
2381 
emitLocalVariable(const LocalVariable & Var)2382 void CodeViewDebug::emitLocalVariable(const LocalVariable &Var) {
2383   // LocalSym record, see SymbolRecord.h for more info.
2384   MCSymbol *LocalBegin = MMI->getContext().createTempSymbol(),
2385            *LocalEnd = MMI->getContext().createTempSymbol();
2386   OS.AddComment("Record length");
2387   OS.emitAbsoluteSymbolDiff(LocalEnd, LocalBegin, 2);
2388   OS.EmitLabel(LocalBegin);
2389 
2390   OS.AddComment("Record kind: S_LOCAL");
2391   OS.EmitIntValue(unsigned(SymbolKind::S_LOCAL), 2);
2392 
2393   LocalSymFlags Flags = LocalSymFlags::None;
2394   if (Var.DIVar->isParameter())
2395     Flags |= LocalSymFlags::IsParameter;
2396   if (Var.DefRanges.empty())
2397     Flags |= LocalSymFlags::IsOptimizedOut;
2398 
2399   OS.AddComment("TypeIndex");
2400   TypeIndex TI = Var.UseReferenceType
2401                      ? getTypeIndexForReferenceTo(Var.DIVar->getType())
2402                      : getCompleteTypeIndex(Var.DIVar->getType());
2403   OS.EmitIntValue(TI.getIndex(), 4);
2404   OS.AddComment("Flags");
2405   OS.EmitIntValue(static_cast<uint16_t>(Flags), 2);
2406   // Truncate the name so we won't overflow the record length field.
2407   emitNullTerminatedSymbolName(OS, Var.DIVar->getName());
2408   OS.EmitLabel(LocalEnd);
2409 
2410   // Calculate the on disk prefix of the appropriate def range record. The
2411   // records and on disk formats are described in SymbolRecords.h. BytePrefix
2412   // should be big enough to hold all forms without memory allocation.
2413   SmallString<20> BytePrefix;
2414   for (const LocalVarDefRange &DefRange : Var.DefRanges) {
2415     BytePrefix.clear();
2416     if (DefRange.InMemory) {
2417       uint16_t RegRelFlags = 0;
2418       if (DefRange.IsSubfield) {
2419         RegRelFlags = DefRangeRegisterRelSym::IsSubfieldFlag |
2420                       (DefRange.StructOffset
2421                        << DefRangeRegisterRelSym::OffsetInParentShift);
2422       }
2423       DefRangeRegisterRelSym Sym(S_DEFRANGE_REGISTER_REL);
2424       Sym.Hdr.Register = DefRange.CVRegister;
2425       Sym.Hdr.Flags = RegRelFlags;
2426       Sym.Hdr.BasePointerOffset = DefRange.DataOffset;
2427       ulittle16_t SymKind = ulittle16_t(S_DEFRANGE_REGISTER_REL);
2428       BytePrefix +=
2429           StringRef(reinterpret_cast<const char *>(&SymKind), sizeof(SymKind));
2430       BytePrefix +=
2431           StringRef(reinterpret_cast<const char *>(&Sym.Hdr), sizeof(Sym.Hdr));
2432     } else {
2433       assert(DefRange.DataOffset == 0 && "unexpected offset into register");
2434       if (DefRange.IsSubfield) {
2435         // Unclear what matters here.
2436         DefRangeSubfieldRegisterSym Sym(S_DEFRANGE_SUBFIELD_REGISTER);
2437         Sym.Hdr.Register = DefRange.CVRegister;
2438         Sym.Hdr.MayHaveNoName = 0;
2439         Sym.Hdr.OffsetInParent = DefRange.StructOffset;
2440 
2441         ulittle16_t SymKind = ulittle16_t(S_DEFRANGE_SUBFIELD_REGISTER);
2442         BytePrefix += StringRef(reinterpret_cast<const char *>(&SymKind),
2443                                 sizeof(SymKind));
2444         BytePrefix += StringRef(reinterpret_cast<const char *>(&Sym.Hdr),
2445                                 sizeof(Sym.Hdr));
2446       } else {
2447         // Unclear what matters here.
2448         DefRangeRegisterSym Sym(S_DEFRANGE_REGISTER);
2449         Sym.Hdr.Register = DefRange.CVRegister;
2450         Sym.Hdr.MayHaveNoName = 0;
2451         ulittle16_t SymKind = ulittle16_t(S_DEFRANGE_REGISTER);
2452         BytePrefix += StringRef(reinterpret_cast<const char *>(&SymKind),
2453                                 sizeof(SymKind));
2454         BytePrefix += StringRef(reinterpret_cast<const char *>(&Sym.Hdr),
2455                                 sizeof(Sym.Hdr));
2456       }
2457     }
2458     OS.EmitCVDefRangeDirective(DefRange.Ranges, BytePrefix);
2459   }
2460 }
2461 
emitLexicalBlockList(ArrayRef<LexicalBlock * > Blocks,const FunctionInfo & FI)2462 void CodeViewDebug::emitLexicalBlockList(ArrayRef<LexicalBlock *> Blocks,
2463                                          const FunctionInfo& FI) {
2464   for (LexicalBlock *Block : Blocks)
2465     emitLexicalBlock(*Block, FI);
2466 }
2467 
2468 /// Emit an S_BLOCK32 and S_END record pair delimiting the contents of a
2469 /// lexical block scope.
emitLexicalBlock(const LexicalBlock & Block,const FunctionInfo & FI)2470 void CodeViewDebug::emitLexicalBlock(const LexicalBlock &Block,
2471                                      const FunctionInfo& FI) {
2472   MCSymbol *RecordBegin = MMI->getContext().createTempSymbol(),
2473            *RecordEnd   = MMI->getContext().createTempSymbol();
2474 
2475   // Lexical block symbol record.
2476   OS.AddComment("Record length");
2477   OS.emitAbsoluteSymbolDiff(RecordEnd, RecordBegin, 2);   // Record Length
2478   OS.EmitLabel(RecordBegin);
2479   OS.AddComment("Record kind: S_BLOCK32");
2480   OS.EmitIntValue(SymbolKind::S_BLOCK32, 2);              // Record Kind
2481   OS.AddComment("PtrParent");
2482   OS.EmitIntValue(0, 4);                                  // PtrParent
2483   OS.AddComment("PtrEnd");
2484   OS.EmitIntValue(0, 4);                                  // PtrEnd
2485   OS.AddComment("Code size");
2486   OS.emitAbsoluteSymbolDiff(Block.End, Block.Begin, 4);   // Code Size
2487   OS.AddComment("Function section relative address");
2488   OS.EmitCOFFSecRel32(Block.Begin, /*Offset=*/0);         // Func Offset
2489   OS.AddComment("Function section index");
2490   OS.EmitCOFFSectionIndex(FI.Begin);                      // Func Symbol
2491   OS.AddComment("Lexical block name");
2492   emitNullTerminatedSymbolName(OS, Block.Name);           // Name
2493   OS.EmitLabel(RecordEnd);
2494 
2495   // Emit variables local to this lexical block.
2496   emitLocalVariableList(Block.Locals);
2497 
2498   // Emit lexical blocks contained within this block.
2499   emitLexicalBlockList(Block.Children, FI);
2500 
2501   // Close the lexical block scope.
2502   OS.AddComment("Record length");
2503   OS.EmitIntValue(2, 2);                                  // Record Length
2504   OS.AddComment("Record kind: S_END");
2505   OS.EmitIntValue(SymbolKind::S_END, 2);                  // Record Kind
2506 }
2507 
2508 /// Convenience routine for collecting lexical block information for a list
2509 /// of lexical scopes.
collectLexicalBlockInfo(SmallVectorImpl<LexicalScope * > & Scopes,SmallVectorImpl<LexicalBlock * > & Blocks,SmallVectorImpl<LocalVariable> & Locals)2510 void CodeViewDebug::collectLexicalBlockInfo(
2511         SmallVectorImpl<LexicalScope *> &Scopes,
2512         SmallVectorImpl<LexicalBlock *> &Blocks,
2513         SmallVectorImpl<LocalVariable> &Locals) {
2514   for (LexicalScope *Scope : Scopes)
2515     collectLexicalBlockInfo(*Scope, Blocks, Locals);
2516 }
2517 
2518 /// Populate the lexical blocks and local variable lists of the parent with
2519 /// information about the specified lexical scope.
collectLexicalBlockInfo(LexicalScope & Scope,SmallVectorImpl<LexicalBlock * > & ParentBlocks,SmallVectorImpl<LocalVariable> & ParentLocals)2520 void CodeViewDebug::collectLexicalBlockInfo(
2521     LexicalScope &Scope,
2522     SmallVectorImpl<LexicalBlock *> &ParentBlocks,
2523     SmallVectorImpl<LocalVariable> &ParentLocals) {
2524   if (Scope.isAbstractScope())
2525     return;
2526 
2527   auto LocalsIter = ScopeVariables.find(&Scope);
2528   if (LocalsIter == ScopeVariables.end()) {
2529     // This scope does not contain variables and can be eliminated.
2530     collectLexicalBlockInfo(Scope.getChildren(), ParentBlocks, ParentLocals);
2531     return;
2532   }
2533   SmallVectorImpl<LocalVariable> &Locals = LocalsIter->second;
2534 
2535   const DILexicalBlock *DILB = dyn_cast<DILexicalBlock>(Scope.getScopeNode());
2536   if (!DILB) {
2537     // This scope is not a lexical block and can be eliminated, but keep any
2538     // local variables it contains.
2539     ParentLocals.append(Locals.begin(), Locals.end());
2540     collectLexicalBlockInfo(Scope.getChildren(), ParentBlocks, ParentLocals);
2541     return;
2542   }
2543 
2544   const SmallVectorImpl<InsnRange> &Ranges = Scope.getRanges();
2545   if (Ranges.size() != 1 || !getLabelAfterInsn(Ranges.front().second)) {
2546     // This lexical block scope has too many address ranges to represent in the
2547     // current CodeView format or does not have a valid address range.
2548     // Eliminate this lexical scope and promote any locals it contains to the
2549     // parent scope.
2550     //
2551     // For lexical scopes with multiple address ranges you may be tempted to
2552     // construct a single range covering every instruction where the block is
2553     // live and everything in between.  Unfortunately, Visual Studio only
2554     // displays variables from the first matching lexical block scope.  If the
2555     // first lexical block contains exception handling code or cold code which
2556     // is moved to the bottom of the routine creating a single range covering
2557     // nearly the entire routine, then it will hide all other lexical blocks
2558     // and the variables they contain.
2559     //
2560     ParentLocals.append(Locals.begin(), Locals.end());
2561     collectLexicalBlockInfo(Scope.getChildren(), ParentBlocks, ParentLocals);
2562     return;
2563   }
2564 
2565   // Create a new CodeView lexical block for this lexical scope.  If we've
2566   // seen this DILexicalBlock before then the scope tree is malformed and
2567   // we can handle this gracefully by not processing it a second time.
2568   auto BlockInsertion = CurFn->LexicalBlocks.insert({DILB, LexicalBlock()});
2569   if (!BlockInsertion.second)
2570     return;
2571 
2572   // Create a lexical block containing the local variables and collect the
2573   // the lexical block information for the children.
2574   const InsnRange &Range = Ranges.front();
2575   assert(Range.first && Range.second);
2576   LexicalBlock &Block = BlockInsertion.first->second;
2577   Block.Begin = getLabelBeforeInsn(Range.first);
2578   Block.End = getLabelAfterInsn(Range.second);
2579   assert(Block.Begin && "missing label for scope begin");
2580   assert(Block.End && "missing label for scope end");
2581   Block.Name = DILB->getName();
2582   Block.Locals = std::move(Locals);
2583   ParentBlocks.push_back(&Block);
2584   collectLexicalBlockInfo(Scope.getChildren(), Block.Children, Block.Locals);
2585 }
2586 
endFunctionImpl(const MachineFunction * MF)2587 void CodeViewDebug::endFunctionImpl(const MachineFunction *MF) {
2588   const Function &GV = MF->getFunction();
2589   assert(FnDebugInfo.count(&GV));
2590   assert(CurFn == FnDebugInfo[&GV].get());
2591 
2592   collectVariableInfo(GV.getSubprogram());
2593 
2594   // Build the lexical block structure to emit for this routine.
2595   if (LexicalScope *CFS = LScopes.getCurrentFunctionScope())
2596     collectLexicalBlockInfo(*CFS, CurFn->ChildBlocks, CurFn->Locals);
2597 
2598   // Clear the scope and variable information from the map which will not be
2599   // valid after we have finished processing this routine.  This also prepares
2600   // the map for the subsequent routine.
2601   ScopeVariables.clear();
2602 
2603   // Don't emit anything if we don't have any line tables.
2604   // Thunks are compiler-generated and probably won't have source correlation.
2605   if (!CurFn->HaveLineInfo && !GV.getSubprogram()->isThunk()) {
2606     FnDebugInfo.erase(&GV);
2607     CurFn = nullptr;
2608     return;
2609   }
2610 
2611   CurFn->Annotations = MF->getCodeViewAnnotations();
2612 
2613   CurFn->End = Asm->getFunctionEnd();
2614 
2615   CurFn = nullptr;
2616 }
2617 
beginInstruction(const MachineInstr * MI)2618 void CodeViewDebug::beginInstruction(const MachineInstr *MI) {
2619   DebugHandlerBase::beginInstruction(MI);
2620 
2621   // Ignore DBG_VALUE and DBG_LABEL locations and function prologue.
2622   if (!Asm || !CurFn || MI->isDebugInstr() ||
2623       MI->getFlag(MachineInstr::FrameSetup))
2624     return;
2625 
2626   // If the first instruction of a new MBB has no location, find the first
2627   // instruction with a location and use that.
2628   DebugLoc DL = MI->getDebugLoc();
2629   if (!DL && MI->getParent() != PrevInstBB) {
2630     for (const auto &NextMI : *MI->getParent()) {
2631       if (NextMI.isDebugInstr())
2632         continue;
2633       DL = NextMI.getDebugLoc();
2634       if (DL)
2635         break;
2636     }
2637   }
2638   PrevInstBB = MI->getParent();
2639 
2640   // If we still don't have a debug location, don't record a location.
2641   if (!DL)
2642     return;
2643 
2644   maybeRecordLocation(DL, Asm->MF);
2645 }
2646 
beginCVSubsection(DebugSubsectionKind Kind)2647 MCSymbol *CodeViewDebug::beginCVSubsection(DebugSubsectionKind Kind) {
2648   MCSymbol *BeginLabel = MMI->getContext().createTempSymbol(),
2649            *EndLabel = MMI->getContext().createTempSymbol();
2650   OS.EmitIntValue(unsigned(Kind), 4);
2651   OS.AddComment("Subsection size");
2652   OS.emitAbsoluteSymbolDiff(EndLabel, BeginLabel, 4);
2653   OS.EmitLabel(BeginLabel);
2654   return EndLabel;
2655 }
2656 
endCVSubsection(MCSymbol * EndLabel)2657 void CodeViewDebug::endCVSubsection(MCSymbol *EndLabel) {
2658   OS.EmitLabel(EndLabel);
2659   // Every subsection must be aligned to a 4-byte boundary.
2660   OS.EmitValueToAlignment(4);
2661 }
2662 
emitDebugInfoForUDTs(ArrayRef<std::pair<std::string,const DIType * >> UDTs)2663 void CodeViewDebug::emitDebugInfoForUDTs(
2664     ArrayRef<std::pair<std::string, const DIType *>> UDTs) {
2665   for (const auto &UDT : UDTs) {
2666     const DIType *T = UDT.second;
2667     assert(shouldEmitUdt(T));
2668 
2669     MCSymbol *UDTRecordBegin = MMI->getContext().createTempSymbol(),
2670              *UDTRecordEnd = MMI->getContext().createTempSymbol();
2671     OS.AddComment("Record length");
2672     OS.emitAbsoluteSymbolDiff(UDTRecordEnd, UDTRecordBegin, 2);
2673     OS.EmitLabel(UDTRecordBegin);
2674 
2675     OS.AddComment("Record kind: S_UDT");
2676     OS.EmitIntValue(unsigned(SymbolKind::S_UDT), 2);
2677 
2678     OS.AddComment("Type");
2679     OS.EmitIntValue(getCompleteTypeIndex(T).getIndex(), 4);
2680 
2681     emitNullTerminatedSymbolName(OS, UDT.first);
2682     OS.EmitLabel(UDTRecordEnd);
2683   }
2684 }
2685 
emitDebugInfoForGlobals()2686 void CodeViewDebug::emitDebugInfoForGlobals() {
2687   DenseMap<const DIGlobalVariableExpression *, const GlobalVariable *>
2688       GlobalMap;
2689   for (const GlobalVariable &GV : MMI->getModule()->globals()) {
2690     SmallVector<DIGlobalVariableExpression *, 1> GVEs;
2691     GV.getDebugInfo(GVEs);
2692     for (const auto *GVE : GVEs)
2693       GlobalMap[GVE] = &GV;
2694   }
2695 
2696   NamedMDNode *CUs = MMI->getModule()->getNamedMetadata("llvm.dbg.cu");
2697   for (const MDNode *Node : CUs->operands()) {
2698     const auto *CU = cast<DICompileUnit>(Node);
2699 
2700     // First, emit all globals that are not in a comdat in a single symbol
2701     // substream. MSVC doesn't like it if the substream is empty, so only open
2702     // it if we have at least one global to emit.
2703     switchToDebugSectionForSymbol(nullptr);
2704     MCSymbol *EndLabel = nullptr;
2705     for (const auto *GVE : CU->getGlobalVariables()) {
2706       if (const auto *GV = GlobalMap.lookup(GVE))
2707         if (!GV->hasComdat() && !GV->isDeclarationForLinker()) {
2708           if (!EndLabel) {
2709             OS.AddComment("Symbol subsection for globals");
2710             EndLabel = beginCVSubsection(DebugSubsectionKind::Symbols);
2711           }
2712           // FIXME: emitDebugInfoForGlobal() doesn't handle DIExpressions.
2713           emitDebugInfoForGlobal(GVE->getVariable(), GV, Asm->getSymbol(GV));
2714         }
2715     }
2716     if (EndLabel)
2717       endCVSubsection(EndLabel);
2718 
2719     // Second, emit each global that is in a comdat into its own .debug$S
2720     // section along with its own symbol substream.
2721     for (const auto *GVE : CU->getGlobalVariables()) {
2722       if (const auto *GV = GlobalMap.lookup(GVE)) {
2723         if (GV->hasComdat()) {
2724           MCSymbol *GVSym = Asm->getSymbol(GV);
2725           OS.AddComment("Symbol subsection for " +
2726                         Twine(GlobalValue::dropLLVMManglingEscape(GV->getName())));
2727           switchToDebugSectionForSymbol(GVSym);
2728           EndLabel = beginCVSubsection(DebugSubsectionKind::Symbols);
2729           // FIXME: emitDebugInfoForGlobal() doesn't handle DIExpressions.
2730           emitDebugInfoForGlobal(GVE->getVariable(), GV, GVSym);
2731           endCVSubsection(EndLabel);
2732         }
2733       }
2734     }
2735   }
2736 }
2737 
emitDebugInfoForRetainedTypes()2738 void CodeViewDebug::emitDebugInfoForRetainedTypes() {
2739   NamedMDNode *CUs = MMI->getModule()->getNamedMetadata("llvm.dbg.cu");
2740   for (const MDNode *Node : CUs->operands()) {
2741     for (auto *Ty : cast<DICompileUnit>(Node)->getRetainedTypes()) {
2742       if (DIType *RT = dyn_cast<DIType>(Ty)) {
2743         getTypeIndex(RT);
2744         // FIXME: Add to global/local DTU list.
2745       }
2746     }
2747   }
2748 }
2749 
emitDebugInfoForGlobal(const DIGlobalVariable * DIGV,const GlobalVariable * GV,MCSymbol * GVSym)2750 void CodeViewDebug::emitDebugInfoForGlobal(const DIGlobalVariable *DIGV,
2751                                            const GlobalVariable *GV,
2752                                            MCSymbol *GVSym) {
2753   // DataSym record, see SymbolRecord.h for more info.
2754   // FIXME: Thread local data, etc
2755   MCSymbol *DataBegin = MMI->getContext().createTempSymbol(),
2756            *DataEnd = MMI->getContext().createTempSymbol();
2757   const unsigned FixedLengthOfThisRecord = 12;
2758   OS.AddComment("Record length");
2759   OS.emitAbsoluteSymbolDiff(DataEnd, DataBegin, 2);
2760   OS.EmitLabel(DataBegin);
2761   if (DIGV->isLocalToUnit()) {
2762     if (GV->isThreadLocal()) {
2763       OS.AddComment("Record kind: S_LTHREAD32");
2764       OS.EmitIntValue(unsigned(SymbolKind::S_LTHREAD32), 2);
2765     } else {
2766       OS.AddComment("Record kind: S_LDATA32");
2767       OS.EmitIntValue(unsigned(SymbolKind::S_LDATA32), 2);
2768     }
2769   } else {
2770     if (GV->isThreadLocal()) {
2771       OS.AddComment("Record kind: S_GTHREAD32");
2772       OS.EmitIntValue(unsigned(SymbolKind::S_GTHREAD32), 2);
2773     } else {
2774       OS.AddComment("Record kind: S_GDATA32");
2775       OS.EmitIntValue(unsigned(SymbolKind::S_GDATA32), 2);
2776     }
2777   }
2778   OS.AddComment("Type");
2779   OS.EmitIntValue(getCompleteTypeIndex(DIGV->getType()).getIndex(), 4);
2780   OS.AddComment("DataOffset");
2781   OS.EmitCOFFSecRel32(GVSym, /*Offset=*/0);
2782   OS.AddComment("Segment");
2783   OS.EmitCOFFSectionIndex(GVSym);
2784   OS.AddComment("Name");
2785   emitNullTerminatedSymbolName(OS, DIGV->getName(), FixedLengthOfThisRecord);
2786   OS.EmitLabel(DataEnd);
2787 }
2788