Home | History | Annotate | Download | only in llvm-dwp 
      1 #include "llvm/ADT/STLExtras.h"
      2 #include "llvm/ADT/StringSet.h"
      3 #include "llvm/CodeGen/AsmPrinter.h"
      4 #include "llvm/DebugInfo/DWARF/DWARFFormValue.h"
      5 #include "llvm/DebugInfo/DWARF/DWARFUnitIndex.h"
      6 #include "llvm/MC/MCAsmInfo.h"
      7 #include "llvm/MC/MCContext.h"
      8 #include "llvm/MC/MCInstrInfo.h"
      9 #include "llvm/MC/MCObjectFileInfo.h"
     10 #include "llvm/MC/MCRegisterInfo.h"
     11 #include "llvm/MC/MCSectionELF.h"
     12 #include "llvm/MC/MCStreamer.h"
     13 #include "llvm/MC/MCTargetOptionsCommandFlags.h"
     14 #include "llvm/Object/ObjectFile.h"
     15 #include "llvm/Support/DataExtractor.h"
     16 #include "llvm/Support/FileSystem.h"
     17 #include "llvm/Support/MathExtras.h"
     18 #include "llvm/Support/MemoryBuffer.h"
     19 #include "llvm/Support/Options.h"
     20 #include "llvm/Support/TargetRegistry.h"
     21 #include "llvm/Support/TargetSelect.h"
     22 #include "llvm/Support/raw_ostream.h"
     23 #include "llvm/Target/TargetMachine.h"
     24 #include <list>
     25 #include <memory>
     26 #include <unordered_set>
     27 
     28 using namespace llvm;
     29 using namespace llvm::object;
     30 using namespace cl;
     31 
     32 OptionCategory DwpCategory("Specific Options");
     33 static list<std::string> InputFiles(Positional, OneOrMore,
     34                                     desc("<input files>"), cat(DwpCategory));
     35 
     36 static opt<std::string> OutputFilename(Required, "o",
     37                                        desc("Specify the output file."),
     38                                        value_desc("filename"),
     39                                        cat(DwpCategory));
     40 
     41 static int error(const Twine &Error, const Twine &Context) {
     42   errs() << Twine("while processing ") + Context + ":\n";
     43   errs() << Twine("error: ") + Error + "\n";
     44   return 1;
     45 }
     46 
     47 static std::error_code
     48 writeStringsAndOffsets(MCStreamer &Out, StringMap<uint32_t> &Strings,
     49                        uint32_t &StringOffset, MCSection *StrSection,
     50                        MCSection *StrOffsetSection, StringRef CurStrSection,
     51                        StringRef CurStrOffsetSection) {
     52   // Could possibly produce an error or warning if one of these was non-null but
     53   // the other was null.
     54   if (CurStrSection.empty() || CurStrOffsetSection.empty())
     55     return std::error_code();
     56 
     57   DenseMap<uint32_t, uint32_t> OffsetRemapping;
     58 
     59   DataExtractor Data(CurStrSection, true, 0);
     60   uint32_t LocalOffset = 0;
     61   uint32_t PrevOffset = 0;
     62   while (const char *s = Data.getCStr(&LocalOffset)) {
     63     StringRef Str(s, LocalOffset - PrevOffset - 1);
     64     auto Pair = Strings.insert(std::make_pair(Str, StringOffset));
     65     if (Pair.second) {
     66       Out.SwitchSection(StrSection);
     67       Out.EmitBytes(
     68           StringRef(Pair.first->getKeyData(), Pair.first->getKeyLength() + 1));
     69       StringOffset += Str.size() + 1;
     70     }
     71     OffsetRemapping[PrevOffset] = Pair.first->second;
     72     PrevOffset = LocalOffset;
     73   }
     74 
     75   Data = DataExtractor(CurStrOffsetSection, true, 0);
     76 
     77   Out.SwitchSection(StrOffsetSection);
     78 
     79   uint32_t Offset = 0;
     80   uint64_t Size = CurStrOffsetSection.size();
     81   while (Offset < Size) {
     82     auto OldOffset = Data.getU32(&Offset);
     83     auto NewOffset = OffsetRemapping[OldOffset];
     84     Out.EmitIntValue(NewOffset, 4);
     85   }
     86 
     87   return std::error_code();
     88 }
     89 
     90 static uint32_t getCUAbbrev(StringRef Abbrev, uint64_t AbbrCode) {
     91   uint64_t CurCode;
     92   uint32_t Offset = 0;
     93   DataExtractor AbbrevData(Abbrev, true, 0);
     94   while ((CurCode = AbbrevData.getULEB128(&Offset)) != AbbrCode) {
     95     // Tag
     96     AbbrevData.getULEB128(&Offset);
     97     // DW_CHILDREN
     98     AbbrevData.getU8(&Offset);
     99     // Attributes
    100     while (AbbrevData.getULEB128(&Offset) | AbbrevData.getULEB128(&Offset))
    101       ;
    102   }
    103   return Offset;
    104 }
    105 
    106 static uint64_t getCUSignature(StringRef Abbrev, StringRef Info) {
    107   uint32_t Offset = 0;
    108   DataExtractor InfoData(Info, true, 0);
    109   InfoData.getU32(&Offset); // Length
    110   uint16_t Version = InfoData.getU16(&Offset);
    111   InfoData.getU32(&Offset); // Abbrev offset (should be zero)
    112   uint8_t AddrSize = InfoData.getU8(&Offset);
    113 
    114   uint32_t AbbrCode = InfoData.getULEB128(&Offset);
    115 
    116   DataExtractor AbbrevData(Abbrev, true, 0);
    117   uint32_t AbbrevOffset = getCUAbbrev(Abbrev, AbbrCode);
    118   uint64_t Tag = AbbrevData.getULEB128(&AbbrevOffset);
    119   (void)Tag;
    120   // FIXME: Real error handling
    121   assert(Tag == dwarf::DW_TAG_compile_unit);
    122   // DW_CHILDREN
    123   AbbrevData.getU8(&AbbrevOffset);
    124   uint32_t Name;
    125   uint32_t Form;
    126   while ((Name = AbbrevData.getULEB128(&AbbrevOffset)) |
    127              (Form = AbbrevData.getULEB128(&AbbrevOffset)) &&
    128          Name != dwarf::DW_AT_GNU_dwo_id) {
    129     DWARFFormValue::skipValue(Form, InfoData, &Offset, Version, AddrSize);
    130   }
    131   // FIXME: Real error handling
    132   assert(Name == dwarf::DW_AT_GNU_dwo_id);
    133   return InfoData.getU64(&Offset);
    134 }
    135 
    136 struct UnitIndexEntry {
    137   uint64_t Signature;
    138   DWARFUnitIndex::Entry::SectionContribution Contributions[8];
    139 };
    140 
    141 static void addAllTypes(MCStreamer &Out,
    142                         std::vector<UnitIndexEntry> &TypeIndexEntries,
    143                         MCSection *OutputTypes, StringRef Types,
    144                         const UnitIndexEntry &CUEntry, uint32_t &TypesOffset) {
    145   if (Types.empty())
    146     return;
    147 
    148   Out.SwitchSection(OutputTypes);
    149   uint32_t Offset = 0;
    150   DataExtractor Data(Types, true, 0);
    151   while (Data.isValidOffset(Offset)) {
    152     UnitIndexEntry Entry = CUEntry;
    153     // Zero out the debug_info contribution
    154     Entry.Contributions[0] = {};
    155     auto &C = Entry.Contributions[DW_SECT_TYPES - DW_SECT_INFO];
    156     C.Offset = TypesOffset;
    157     auto PrevOffset = Offset;
    158     // Length of the unit, including the 4 byte length field.
    159     C.Length = Data.getU32(&Offset) + 4;
    160 
    161     Data.getU16(&Offset); // Version
    162     Data.getU32(&Offset); // Abbrev offset
    163     Data.getU8(&Offset);  // Address size
    164     Entry.Signature = Data.getU64(&Offset);
    165     Offset = PrevOffset + C.Length;
    166 
    167     if (any_of(TypeIndexEntries, [&](const UnitIndexEntry &E) {
    168           return E.Signature == Entry.Signature;
    169         }))
    170       continue;
    171 
    172     Out.EmitBytes(Types.substr(PrevOffset, C.Length));
    173     TypesOffset += C.Length;
    174 
    175     TypeIndexEntries.push_back(Entry);
    176   }
    177 }
    178 
    179 static void
    180 writeIndexTable(MCStreamer &Out, ArrayRef<unsigned> ContributionOffsets,
    181                 ArrayRef<UnitIndexEntry> IndexEntries,
    182                 uint32_t DWARFUnitIndex::Entry::SectionContribution::*Field) {
    183   for (const auto &E : IndexEntries)
    184     for (size_t i = 0; i != array_lengthof(E.Contributions); ++i)
    185       if (ContributionOffsets[i])
    186         Out.EmitIntValue(E.Contributions[i].*Field, 4);
    187 }
    188 
    189 static void writeIndex(MCStreamer &Out, MCSection *Section,
    190                        ArrayRef<unsigned> ContributionOffsets,
    191                        ArrayRef<UnitIndexEntry> IndexEntries) {
    192   unsigned Columns = 0;
    193   for (auto &C : ContributionOffsets)
    194     if (C)
    195       ++Columns;
    196 
    197   std::vector<unsigned> Buckets(NextPowerOf2(3 * IndexEntries.size() / 2));
    198   uint64_t Mask = Buckets.size() - 1;
    199   for (size_t i = 0; i != IndexEntries.size(); ++i) {
    200     auto S = IndexEntries[i].Signature;
    201     auto H = S & Mask;
    202     while (Buckets[H]) {
    203       assert(S != IndexEntries[Buckets[H] - 1].Signature &&
    204              "Duplicate type unit");
    205       H += ((S >> 32) & Mask) | 1;
    206     }
    207     Buckets[H] = i + 1;
    208   }
    209 
    210   Out.SwitchSection(Section);
    211   Out.EmitIntValue(2, 4);                   // Version
    212   Out.EmitIntValue(Columns, 4);             // Columns
    213   Out.EmitIntValue(IndexEntries.size(), 4); // Num Units
    214   Out.EmitIntValue(Buckets.size(), 4);      // Num Buckets
    215 
    216   // Write the signatures.
    217   for (const auto &I : Buckets)
    218     Out.EmitIntValue(I ? IndexEntries[I - 1].Signature : 0, 8);
    219 
    220   // Write the indexes.
    221   for (const auto &I : Buckets)
    222     Out.EmitIntValue(I, 4);
    223 
    224   // Write the column headers (which sections will appear in the table)
    225   for (size_t i = 0; i != ContributionOffsets.size(); ++i)
    226     if (ContributionOffsets[i])
    227       Out.EmitIntValue(i + DW_SECT_INFO, 4);
    228 
    229   // Write the offsets.
    230   writeIndexTable(Out, ContributionOffsets, IndexEntries,
    231                   &DWARFUnitIndex::Entry::SectionContribution::Offset);
    232 
    233   // Write the lengths.
    234   writeIndexTable(Out, ContributionOffsets, IndexEntries,
    235                   &DWARFUnitIndex::Entry::SectionContribution::Length);
    236 }
    237 static std::error_code write(MCStreamer &Out, ArrayRef<std::string> Inputs) {
    238   const auto &MCOFI = *Out.getContext().getObjectFileInfo();
    239   MCSection *const StrSection = MCOFI.getDwarfStrDWOSection();
    240   MCSection *const StrOffsetSection = MCOFI.getDwarfStrOffDWOSection();
    241   MCSection *const TypesSection = MCOFI.getDwarfTypesDWOSection();
    242   const StringMap<std::pair<MCSection *, DWARFSectionKind>> KnownSections = {
    243       {"debug_info.dwo", {MCOFI.getDwarfInfoDWOSection(), DW_SECT_INFO}},
    244       {"debug_types.dwo", {MCOFI.getDwarfTypesDWOSection(), DW_SECT_TYPES}},
    245       {"debug_str_offsets.dwo", {StrOffsetSection, DW_SECT_STR_OFFSETS}},
    246       {"debug_str.dwo", {StrSection, static_cast<DWARFSectionKind>(0)}},
    247       {"debug_loc.dwo", {MCOFI.getDwarfLocDWOSection(), DW_SECT_LOC}},
    248       {"debug_line.dwo", {MCOFI.getDwarfLineDWOSection(), DW_SECT_LINE}},
    249       {"debug_abbrev.dwo", {MCOFI.getDwarfAbbrevDWOSection(), DW_SECT_ABBREV}}};
    250 
    251   std::vector<UnitIndexEntry> IndexEntries;
    252   std::vector<UnitIndexEntry> TypeIndexEntries;
    253 
    254   StringMap<uint32_t> Strings;
    255   uint32_t StringOffset = 0;
    256 
    257   uint32_t ContributionOffsets[8] = {};
    258 
    259   for (const auto &Input : Inputs) {
    260     auto ErrOrObj = object::ObjectFile::createObjectFile(Input);
    261     if (!ErrOrObj)
    262       return ErrOrObj.getError();
    263 
    264     IndexEntries.emplace_back();
    265     UnitIndexEntry &CurEntry = IndexEntries.back();
    266 
    267     StringRef CurStrSection;
    268     StringRef CurStrOffsetSection;
    269     StringRef CurTypesSection;
    270     StringRef InfoSection;
    271     StringRef AbbrevSection;
    272 
    273     for (const auto &Section : ErrOrObj->getBinary()->sections()) {
    274       StringRef Name;
    275       if (std::error_code Err = Section.getName(Name))
    276         return Err;
    277 
    278       auto SectionPair =
    279           KnownSections.find(Name.substr(Name.find_first_not_of("._")));
    280       if (SectionPair == KnownSections.end())
    281         continue;
    282 
    283       StringRef Contents;
    284       if (auto Err = Section.getContents(Contents))
    285         return Err;
    286 
    287       if (DWARFSectionKind Kind = SectionPair->second.second) {
    288         auto Index = Kind - DW_SECT_INFO;
    289         if (Kind != DW_SECT_TYPES) {
    290           CurEntry.Contributions[Index].Offset = ContributionOffsets[Index];
    291           ContributionOffsets[Index] +=
    292               (CurEntry.Contributions[Index].Length = Contents.size());
    293         }
    294 
    295         switch (Kind) {
    296         case DW_SECT_INFO:
    297           InfoSection = Contents;
    298           break;
    299         case DW_SECT_ABBREV:
    300           AbbrevSection = Contents;
    301           break;
    302         default:
    303           break;
    304         }
    305       }
    306 
    307       MCSection *OutSection = SectionPair->second.first;
    308       if (OutSection == StrOffsetSection)
    309         CurStrOffsetSection = Contents;
    310       else if (OutSection == StrSection)
    311         CurStrSection = Contents;
    312       else if (OutSection == TypesSection)
    313         CurTypesSection = Contents;
    314       else {
    315         Out.SwitchSection(OutSection);
    316         Out.EmitBytes(Contents);
    317       }
    318     }
    319 
    320     assert(!AbbrevSection.empty());
    321     assert(!InfoSection.empty());
    322     CurEntry.Signature = getCUSignature(AbbrevSection, InfoSection);
    323     addAllTypes(Out, TypeIndexEntries, TypesSection, CurTypesSection, CurEntry,
    324                 ContributionOffsets[DW_SECT_TYPES - DW_SECT_INFO]);
    325 
    326     if (auto Err = writeStringsAndOffsets(Out, Strings, StringOffset,
    327                                           StrSection, StrOffsetSection,
    328                                           CurStrSection, CurStrOffsetSection))
    329       return Err;
    330   }
    331 
    332   if (!TypeIndexEntries.empty()) {
    333     // Lie about there being no info contributions so the TU index only includes
    334     // the type unit contribution
    335     ContributionOffsets[0] = 0;
    336     writeIndex(Out, MCOFI.getDwarfTUIndexSection(), ContributionOffsets,
    337                TypeIndexEntries);
    338   }
    339 
    340   // Lie about the type contribution
    341   ContributionOffsets[DW_SECT_TYPES - DW_SECT_INFO] = 0;
    342   // Unlie about the info contribution
    343   ContributionOffsets[0] = 1;
    344 
    345   writeIndex(Out, MCOFI.getDwarfCUIndexSection(), ContributionOffsets,
    346              IndexEntries);
    347 
    348   return std::error_code();
    349 }
    350 
    351 int main(int argc, char **argv) {
    352 
    353   ParseCommandLineOptions(argc, argv, "merge split dwarf (.dwo) files");
    354 
    355   llvm::InitializeAllTargetInfos();
    356   llvm::InitializeAllTargetMCs();
    357   llvm::InitializeAllTargets();
    358   llvm::InitializeAllAsmPrinters();
    359 
    360   std::string ErrorStr;
    361   StringRef Context = "dwarf streamer init";
    362 
    363   Triple TheTriple("x86_64-linux-gnu");
    364 
    365   // Get the target.
    366   const Target *TheTarget =
    367       TargetRegistry::lookupTarget("", TheTriple, ErrorStr);
    368   if (!TheTarget)
    369     return error(ErrorStr, Context);
    370   std::string TripleName = TheTriple.getTriple();
    371 
    372   // Create all the MC Objects.
    373   std::unique_ptr<MCRegisterInfo> MRI(TheTarget->createMCRegInfo(TripleName));
    374   if (!MRI)
    375     return error(Twine("no register info for target ") + TripleName, Context);
    376 
    377   std::unique_ptr<MCAsmInfo> MAI(TheTarget->createMCAsmInfo(*MRI, TripleName));
    378   if (!MAI)
    379     return error("no asm info for target " + TripleName, Context);
    380 
    381   MCObjectFileInfo MOFI;
    382   MCContext MC(MAI.get(), MRI.get(), &MOFI);
    383   MOFI.InitMCObjectFileInfo(TheTriple, Reloc::Default, CodeModel::Default, MC);
    384 
    385   auto MAB = TheTarget->createMCAsmBackend(*MRI, TripleName, "");
    386   if (!MAB)
    387     return error("no asm backend for target " + TripleName, Context);
    388 
    389   std::unique_ptr<MCInstrInfo> MII(TheTarget->createMCInstrInfo());
    390   if (!MII)
    391     return error("no instr info info for target " + TripleName, Context);
    392 
    393   std::unique_ptr<MCSubtargetInfo> MSTI(
    394       TheTarget->createMCSubtargetInfo(TripleName, "", ""));
    395   if (!MSTI)
    396     return error("no subtarget info for target " + TripleName, Context);
    397 
    398   MCCodeEmitter *MCE = TheTarget->createMCCodeEmitter(*MII, *MRI, MC);
    399   if (!MCE)
    400     return error("no code emitter for target " + TripleName, Context);
    401 
    402   // Create the output file.
    403   std::error_code EC;
    404   raw_fd_ostream OutFile(OutputFilename, EC, sys::fs::F_None);
    405   if (EC)
    406     return error(Twine(OutputFilename) + ": " + EC.message(), Context);
    407 
    408   MCTargetOptions MCOptions = InitMCTargetOptionsFromFlags();
    409   std::unique_ptr<MCStreamer> MS(TheTarget->createMCObjectStreamer(
    410       TheTriple, MC, *MAB, OutFile, MCE, *MSTI, MCOptions.MCRelaxAll,
    411       MCOptions.MCIncrementalLinkerCompatible,
    412       /*DWARFMustBeAtTheEnd*/ false));
    413   if (!MS)
    414     return error("no object streamer for target " + TripleName, Context);
    415 
    416   if (auto Err = write(*MS, InputFiles))
    417     return error(Err.message(), "Writing DWP file");
    418 
    419   MS->Finish();
    420 }
    421