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