1 | //===- bolt/Rewrite/DWARFRewriter.cpp -------------------------------------===// |
2 | // |
3 | // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. |
4 | // See https://llvm.org/LICENSE.txt for license information. |
5 | // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception |
6 | // |
7 | //===----------------------------------------------------------------------===// |
8 | |
9 | #include "bolt/Rewrite/DWARFRewriter.h" |
10 | #include "bolt/Core/BinaryContext.h" |
11 | #include "bolt/Core/BinaryFunction.h" |
12 | #include "bolt/Core/DIEBuilder.h" |
13 | #include "bolt/Core/DebugData.h" |
14 | #include "bolt/Core/DynoStats.h" |
15 | #include "bolt/Core/ParallelUtilities.h" |
16 | #include "bolt/Rewrite/RewriteInstance.h" |
17 | #include "llvm/ADT/STLExtras.h" |
18 | #include "llvm/ADT/SmallVector.h" |
19 | #include "llvm/ADT/StringRef.h" |
20 | #include "llvm/BinaryFormat/Dwarf.h" |
21 | #include "llvm/CodeGen/AsmPrinter.h" |
22 | #include "llvm/CodeGen/DIE.h" |
23 | #include "llvm/DWARFLinker/Classic/DWARFStreamer.h" |
24 | #include "llvm/DebugInfo/DWARF/DWARFContext.h" |
25 | #include "llvm/DebugInfo/DWARF/DWARFDebugAbbrev.h" |
26 | #include "llvm/DebugInfo/DWARF/DWARFDebugLoc.h" |
27 | #include "llvm/DebugInfo/DWARF/DWARFExpression.h" |
28 | #include "llvm/DebugInfo/DWARF/DWARFFormValue.h" |
29 | #include "llvm/DebugInfo/DWARF/DWARFTypeUnit.h" |
30 | #include "llvm/DebugInfo/DWARF/DWARFUnit.h" |
31 | #include "llvm/MC/MCAsmBackend.h" |
32 | #include "llvm/MC/MCAsmLayout.h" |
33 | #include "llvm/MC/MCObjectWriter.h" |
34 | #include "llvm/MC/MCStreamer.h" |
35 | #include "llvm/Object/ObjectFile.h" |
36 | #include "llvm/Support/Casting.h" |
37 | #include "llvm/Support/CommandLine.h" |
38 | #include "llvm/Support/Debug.h" |
39 | #include "llvm/Support/Endian.h" |
40 | #include "llvm/Support/Error.h" |
41 | #include "llvm/Support/FileSystem.h" |
42 | #include "llvm/Support/LEB128.h" |
43 | #include "llvm/Support/ThreadPool.h" |
44 | #include "llvm/Support/raw_ostream.h" |
45 | #include <algorithm> |
46 | #include <cstdint> |
47 | #include <functional> |
48 | #include <iterator> |
49 | #include <memory> |
50 | #include <optional> |
51 | #include <string> |
52 | #include <unordered_map> |
53 | #include <utility> |
54 | #include <vector> |
55 | |
56 | #undef DEBUG_TYPE |
57 | #define DEBUG_TYPE "bolt" |
58 | |
59 | static void printDie(const DWARFDie &DIE) { |
60 | DIDumpOptions DumpOpts; |
61 | DumpOpts.ShowForm = true; |
62 | DumpOpts.Verbose = true; |
63 | DumpOpts.ChildRecurseDepth = 0; |
64 | DumpOpts.ShowChildren = false; |
65 | DIE.dump(OS&: dbgs(), indent: 0, DumpOpts); |
66 | } |
67 | |
68 | /// Lazily parse DWARF DIE and print it out. |
69 | LLVM_ATTRIBUTE_UNUSED |
70 | static void printDie(DWARFUnit &DU, uint64_t DIEOffset) { |
71 | uint64_t OriginalOffsets = DIEOffset; |
72 | uint64_t NextCUOffset = DU.getNextUnitOffset(); |
73 | DWARFDataExtractor DebugInfoData = DU.getDebugInfoExtractor(); |
74 | DWARFDebugInfoEntry DIEEntry; |
75 | if (DIEEntry.extractFast(U: DU, OffsetPtr: &DIEOffset, DebugInfoData, UEndOffset: NextCUOffset, ParentIdx: 0)) { |
76 | if (const DWARFAbbreviationDeclaration *AbbrDecl = |
77 | DIEEntry.getAbbreviationDeclarationPtr()) { |
78 | DWARFDie DDie(&DU, &DIEEntry); |
79 | printDie(DIE: DDie); |
80 | } else { |
81 | dbgs() << "Failed to extract abbreviation for" |
82 | << Twine::utohexstr(Val: OriginalOffsets) << "\n" ; |
83 | } |
84 | } else { |
85 | dbgs() << "Failed to extract DIE for " << Twine::utohexstr(Val: OriginalOffsets) |
86 | << " \n" ; |
87 | } |
88 | } |
89 | |
90 | using namespace bolt; |
91 | |
92 | /// Take a set of DWARF address ranges corresponding to the input binary and |
93 | /// translate them to a set of address ranges in the output binary. |
94 | static DebugAddressRangesVector |
95 | translateInputToOutputRanges(const BinaryFunction &BF, |
96 | const DWARFAddressRangesVector &InputRanges) { |
97 | DebugAddressRangesVector OutputRanges; |
98 | |
99 | // If the function hasn't changed return the same ranges. |
100 | if (!BF.isEmitted()) { |
101 | OutputRanges.resize(N: InputRanges.size()); |
102 | llvm::transform(Range: InputRanges, d_first: OutputRanges.begin(), |
103 | F: [](const DWARFAddressRange &Range) { |
104 | return DebugAddressRange(Range.LowPC, Range.HighPC); |
105 | }); |
106 | return OutputRanges; |
107 | } |
108 | |
109 | for (const DWARFAddressRange &Range : InputRanges) |
110 | llvm::append_range(C&: OutputRanges, R: BF.translateInputToOutputRange( |
111 | InRange: {Range.LowPC, Range.HighPC})); |
112 | |
113 | // Post-processing pass to sort and merge ranges. |
114 | llvm::sort(C&: OutputRanges); |
115 | DebugAddressRangesVector MergedRanges; |
116 | uint64_t PrevHighPC = 0; |
117 | for (const DebugAddressRange &Range : OutputRanges) { |
118 | if (Range.LowPC <= PrevHighPC) { |
119 | MergedRanges.back().HighPC = |
120 | std::max(a: MergedRanges.back().HighPC, b: Range.HighPC); |
121 | } else { |
122 | MergedRanges.emplace_back(Args: Range.LowPC, Args: Range.HighPC); |
123 | } |
124 | PrevHighPC = MergedRanges.back().HighPC; |
125 | } |
126 | |
127 | return MergedRanges; |
128 | } |
129 | |
130 | /// Similar to translateInputToOutputRanges() but operates on location lists. |
131 | static DebugLocationsVector |
132 | translateInputToOutputLocationList(const BinaryFunction &BF, |
133 | const DebugLocationsVector &InputLL) { |
134 | DebugLocationsVector OutputLL; |
135 | |
136 | // If the function hasn't changed - there's nothing to update. |
137 | if (!BF.isEmitted()) |
138 | return InputLL; |
139 | |
140 | for (const DebugLocationEntry &Entry : InputLL) { |
141 | DebugAddressRangesVector OutRanges = |
142 | BF.translateInputToOutputRange(InRange: {Entry.LowPC, Entry.HighPC}); |
143 | if (!OutRanges.empty() && !OutputLL.empty()) { |
144 | if (OutRanges.front().LowPC == OutputLL.back().HighPC && |
145 | Entry.Expr == OutputLL.back().Expr) { |
146 | OutputLL.back().HighPC = |
147 | std::max(a: OutputLL.back().HighPC, b: OutRanges.front().HighPC); |
148 | OutRanges.erase(CI: OutRanges.begin()); |
149 | } |
150 | } |
151 | llvm::transform(Range&: OutRanges, d_first: std::back_inserter(x&: OutputLL), |
152 | F: [&Entry](const DebugAddressRange &R) { |
153 | return DebugLocationEntry{.LowPC: R.LowPC, .HighPC: R.HighPC, .Expr: Entry.Expr}; |
154 | }); |
155 | } |
156 | |
157 | // Sort and merge adjacent entries with identical locations. |
158 | llvm::stable_sort( |
159 | Range&: OutputLL, C: [](const DebugLocationEntry &A, const DebugLocationEntry &B) { |
160 | return A.LowPC < B.LowPC; |
161 | }); |
162 | DebugLocationsVector MergedLL; |
163 | uint64_t PrevHighPC = 0; |
164 | const SmallVectorImpl<uint8_t> *PrevExpr = nullptr; |
165 | for (const DebugLocationEntry &Entry : OutputLL) { |
166 | if (Entry.LowPC <= PrevHighPC && *PrevExpr == Entry.Expr) { |
167 | MergedLL.back().HighPC = std::max(a: Entry.HighPC, b: MergedLL.back().HighPC); |
168 | } else { |
169 | const uint64_t Begin = std::max(a: Entry.LowPC, b: PrevHighPC); |
170 | const uint64_t End = std::max(a: Begin, b: Entry.HighPC); |
171 | MergedLL.emplace_back(Args: DebugLocationEntry{.LowPC: Begin, .HighPC: End, .Expr: Entry.Expr}); |
172 | } |
173 | PrevHighPC = MergedLL.back().HighPC; |
174 | PrevExpr = &MergedLL.back().Expr; |
175 | } |
176 | |
177 | return MergedLL; |
178 | } |
179 | |
180 | using namespace dwarf_linker; |
181 | using namespace dwarf_linker::classic; |
182 | |
183 | namespace llvm { |
184 | namespace bolt { |
185 | /// Emits debug information into .debug_info or .debug_types section. |
186 | class DIEStreamer : public DwarfStreamer { |
187 | DIEBuilder *DIEBldr; |
188 | DWARFRewriter &Rewriter; |
189 | |
190 | private: |
191 | /// Emit the compilation unit header for \p Unit in the debug_info |
192 | /// section. |
193 | /// |
194 | /// A Dwarf 4 section header is encoded as: |
195 | /// uint32_t Unit length (omitting this field) |
196 | /// uint16_t Version |
197 | /// uint32_t Abbreviation table offset |
198 | /// uint8_t Address size |
199 | /// Leading to a total of 11 bytes. |
200 | /// |
201 | /// A Dwarf 5 section header is encoded as: |
202 | /// uint32_t Unit length (omitting this field) |
203 | /// uint16_t Version |
204 | /// uint8_t Unit type |
205 | /// uint8_t Address size |
206 | /// uint32_t Abbreviation table offset |
207 | /// Leading to a total of 12 bytes. |
208 | void (DWARFUnit &Unit, DIE &UnitDIE, |
209 | unsigned DwarfVersion) { |
210 | |
211 | AsmPrinter &Asm = getAsmPrinter(); |
212 | switchToDebugInfoSection(DwarfVersion); |
213 | |
214 | emitCommonHeader(Unit, UnitDIE, Version: DwarfVersion); |
215 | |
216 | if (DwarfVersion >= 5 && |
217 | Unit.getUnitType() != dwarf::UnitType::DW_UT_compile) { |
218 | std::optional<uint64_t> DWOId = Unit.getDWOId(); |
219 | assert(DWOId && |
220 | "DWOId does not exist and this is not a DW_UT_compile Unit" ); |
221 | Asm.emitInt64(Value: *DWOId); |
222 | } |
223 | } |
224 | |
225 | void (DWARFUnit &Unit, DIE &UnitDIE, uint16_t Version) { |
226 | dwarf::UnitType UT = dwarf::UnitType(Unit.getUnitType()); |
227 | llvm::AsmPrinter &Asm = getAsmPrinter(); |
228 | |
229 | // Emit size of content not including length itself |
230 | Asm.emitInt32(Value: Unit.getHeaderSize() + UnitDIE.getSize() - 4); |
231 | Asm.emitInt16(Value: Version); |
232 | |
233 | // DWARF v5 reorders the address size and adds a unit type. |
234 | if (Version >= 5) { |
235 | Asm.emitInt8(Value: UT); |
236 | Asm.emitInt8(Value: Asm.MAI->getCodePointerSize()); |
237 | } |
238 | |
239 | Asm.emitInt32(Value: 0); |
240 | if (Version <= 4) { |
241 | Asm.emitInt8(Value: Asm.MAI->getCodePointerSize()); |
242 | } |
243 | } |
244 | |
245 | void (DWARFUnit &Unit, DIE &UnitDIE, |
246 | unsigned DwarfVersion) { |
247 | AsmPrinter &Asm = getAsmPrinter(); |
248 | const uint64_t TypeSignature = cast<DWARFTypeUnit>(Val&: Unit).getTypeHash(); |
249 | DIE *TypeDIE = DIEBldr->getTypeDIE(DU&: Unit); |
250 | const DIEBuilder::DWARFUnitInfo &UI = DIEBldr->getUnitInfoByDwarfUnit(DwarfUnit: Unit); |
251 | Rewriter.addGDBTypeUnitEntry( |
252 | Entry: {.UnitOffset: UI.UnitOffset, .TypeHash: TypeSignature, .TypeDIERelativeOffset: TypeDIE->getOffset()}); |
253 | if (Unit.getVersion() < 5) { |
254 | // Switch the section to .debug_types section. |
255 | std::unique_ptr<MCStreamer> &MS = Asm.OutStreamer; |
256 | llvm::MCContext &MC = Asm.OutContext; |
257 | const llvm::MCObjectFileInfo *MOFI = MC.getObjectFileInfo(); |
258 | |
259 | MS->switchSection(Section: MOFI->getDwarfTypesSection(Hash: 0)); |
260 | MC.setDwarfVersion(DwarfVersion); |
261 | } else |
262 | switchToDebugInfoSection(DwarfVersion); |
263 | |
264 | emitCommonHeader(Unit, UnitDIE, Version: DwarfVersion); |
265 | Asm.OutStreamer->emitIntValue(Value: TypeSignature, Size: sizeof(TypeSignature)); |
266 | Asm.emitDwarfLengthOrOffset(Value: TypeDIE ? TypeDIE->getOffset() : 0); |
267 | } |
268 | |
269 | void (DWARFUnit &Unit, DIE &UnitDIE) { |
270 | if (Unit.isTypeUnit()) |
271 | emitTypeUnitHeader(Unit, UnitDIE, DwarfVersion: Unit.getVersion()); |
272 | else |
273 | emitCompileUnitHeader(Unit, UnitDIE, DwarfVersion: Unit.getVersion()); |
274 | } |
275 | |
276 | void emitDIE(DIE &Die) override { |
277 | AsmPrinter &Asm = getAsmPrinter(); |
278 | Asm.emitDwarfDIE(Die); |
279 | } |
280 | |
281 | public: |
282 | DIEStreamer(DIEBuilder *DIEBldr, DWARFRewriter &Rewriter, |
283 | DWARFLinkerBase::OutputFileType OutFileType, |
284 | raw_pwrite_stream &OutFile, |
285 | DWARFLinkerBase::MessageHandlerTy Warning) |
286 | : DwarfStreamer(OutFileType, OutFile, Warning), DIEBldr(DIEBldr), |
287 | Rewriter(Rewriter){}; |
288 | |
289 | using DwarfStreamer::emitCompileUnitHeader; |
290 | |
291 | void emitUnit(DWARFUnit &Unit, DIE &UnitDIE) { |
292 | emitUnitHeader(Unit, UnitDIE); |
293 | emitDIE(Die&: UnitDIE); |
294 | } |
295 | }; |
296 | |
297 | /// Finds attributes FormValue and Offset. |
298 | /// |
299 | /// \param DIE die to look up in. |
300 | /// \param Attrs finds the first attribute that matches and extracts it. |
301 | /// \return an optional AttrInfo with DWARFFormValue and Offset. |
302 | std::optional<AttrInfo> findAttributeInfo(const DWARFDie DIE, |
303 | std::vector<dwarf::Attribute> Attrs) { |
304 | for (dwarf::Attribute &Attr : Attrs) |
305 | if (std::optional<AttrInfo> Info = findAttributeInfo(DIE, Attr)) |
306 | return Info; |
307 | return std::nullopt; |
308 | } |
309 | |
310 | } // namespace bolt |
311 | } // namespace llvm |
312 | |
313 | using namespace llvm; |
314 | using namespace llvm::support::endian; |
315 | using namespace object; |
316 | using namespace bolt; |
317 | |
318 | namespace opts { |
319 | |
320 | extern cl::OptionCategory BoltCategory; |
321 | extern cl::opt<unsigned> Verbosity; |
322 | extern cl::opt<std::string> OutputFilename; |
323 | |
324 | static cl::opt<bool> KeepARanges( |
325 | "keep-aranges" , |
326 | cl::desc( |
327 | "keep or generate .debug_aranges section if .gdb_index is written" ), |
328 | cl::Hidden, cl::cat(BoltCategory)); |
329 | |
330 | static cl::opt<bool> |
331 | DeterministicDebugInfo("deterministic-debuginfo" , |
332 | cl::desc("disables parallel execution of tasks that may produce " |
333 | "nondeterministic debug info" ), |
334 | cl::init(Val: true), |
335 | cl::cat(BoltCategory)); |
336 | |
337 | static cl::opt<std::string> DwarfOutputPath( |
338 | "dwarf-output-path" , |
339 | cl::desc("Path to where .dwo files or dwp file will be written out to." ), |
340 | cl::init(Val: "" ), cl::cat(BoltCategory)); |
341 | |
342 | static cl::opt<bool> |
343 | WriteDWP("write-dwp" , |
344 | cl::desc("output a single dwarf package file (dwp) instead of " |
345 | "multiple non-relocatable dwarf object files (dwo)." ), |
346 | cl::init(Val: false), cl::cat(BoltCategory)); |
347 | |
348 | static cl::opt<bool> CreateDebugNames( |
349 | "create-debug-names-section" , |
350 | cl::desc("Creates .debug_names section, if the input binary doesn't have " |
351 | "it already, for DWARF5 CU/TUs." ), |
352 | cl::init(Val: false), cl::cat(BoltCategory)); |
353 | |
354 | static cl::opt<bool> |
355 | DebugSkeletonCu("debug-skeleton-cu" , |
356 | cl::desc("prints out offsetrs for abbrev and debu_info of " |
357 | "Skeleton CUs that get patched." ), |
358 | cl::ZeroOrMore, cl::Hidden, cl::init(Val: false), |
359 | cl::cat(BoltCategory)); |
360 | |
361 | static cl::opt<unsigned> BatchSize( |
362 | "cu-processing-batch-size" , |
363 | cl::desc( |
364 | "Specifies the size of batches for processing CUs. Higher number has " |
365 | "better performance, but more memory usage. Default value is 1." ), |
366 | cl::Hidden, cl::init(Val: 1), cl::cat(BoltCategory)); |
367 | |
368 | static cl::opt<bool> AlwaysConvertToRanges( |
369 | "always-convert-to-ranges" , |
370 | cl::desc("This option is for testing purposes only. It forces BOLT to " |
371 | "convert low_pc/high_pc to ranges always." ), |
372 | cl::ReallyHidden, cl::init(Val: false), cl::cat(BoltCategory)); |
373 | |
374 | extern cl::opt<std::string> CompDirOverride; |
375 | } // namespace opts |
376 | |
377 | /// If DW_AT_low_pc exists sets LowPC and returns true. |
378 | static bool getLowPC(const DIE &Die, const DWARFUnit &DU, uint64_t &LowPC, |
379 | uint64_t &SectionIndex) { |
380 | DIEValue DvalLowPc = Die.findAttribute(Attribute: dwarf::DW_AT_low_pc); |
381 | if (!DvalLowPc) |
382 | return false; |
383 | |
384 | dwarf::Form Form = DvalLowPc.getForm(); |
385 | bool AddrOffset = Form == dwarf::DW_FORM_LLVM_addrx_offset; |
386 | uint64_t LowPcValue = DvalLowPc.getDIEInteger().getValue(); |
387 | if (Form == dwarf::DW_FORM_GNU_addr_index || Form == dwarf::DW_FORM_addrx || |
388 | AddrOffset) { |
389 | |
390 | uint32_t Index = AddrOffset ? (LowPcValue >> 32) : LowPcValue; |
391 | std::optional<object::SectionedAddress> SA = |
392 | DU.getAddrOffsetSectionItem(Index); |
393 | if (!SA) |
394 | return false; |
395 | if (AddrOffset) |
396 | SA->Address += (LowPcValue & 0xffffffff); |
397 | |
398 | LowPC = SA->Address; |
399 | SectionIndex = SA->SectionIndex; |
400 | } else { |
401 | LowPC = LowPcValue; |
402 | SectionIndex = 0; |
403 | } |
404 | return true; |
405 | } |
406 | |
407 | /// If DW_AT_high_pc exists sets HighPC and returns true. |
408 | static bool getHighPC(const DIE &Die, const uint64_t LowPC, uint64_t &HighPC) { |
409 | DIEValue DvalHighPc = Die.findAttribute(Attribute: dwarf::DW_AT_high_pc); |
410 | if (!DvalHighPc) |
411 | return false; |
412 | if (DvalHighPc.getForm() == dwarf::DW_FORM_addr) |
413 | HighPC = DvalHighPc.getDIEInteger().getValue(); |
414 | else |
415 | HighPC = LowPC + DvalHighPc.getDIEInteger().getValue(); |
416 | return true; |
417 | } |
418 | |
419 | /// If DW_AT_low_pc and DW_AT_high_pc exist sets LowPC and HighPC and returns |
420 | /// true. |
421 | static bool getLowAndHighPC(const DIE &Die, const DWARFUnit &DU, |
422 | uint64_t &LowPC, uint64_t &HighPC, |
423 | uint64_t &SectionIndex) { |
424 | uint64_t TempLowPC = LowPC; |
425 | uint64_t TempHighPC = HighPC; |
426 | uint64_t TempSectionIndex = SectionIndex; |
427 | if (getLowPC(Die, DU, LowPC&: TempLowPC, SectionIndex&: TempSectionIndex) && |
428 | getHighPC(Die, LowPC: TempLowPC, HighPC&: TempHighPC)) { |
429 | LowPC = TempLowPC; |
430 | HighPC = TempHighPC; |
431 | SectionIndex = TempSectionIndex; |
432 | return true; |
433 | } |
434 | return false; |
435 | } |
436 | |
437 | static Expected<llvm::DWARFAddressRangesVector> |
438 | getDIEAddressRanges(const DIE &Die, DWARFUnit &DU) { |
439 | uint64_t LowPC, HighPC, Index; |
440 | if (getLowAndHighPC(Die, DU, LowPC, HighPC, SectionIndex&: Index)) |
441 | return DWARFAddressRangesVector{{LowPC, HighPC, Index}}; |
442 | if (DIEValue Dval = Die.findAttribute(Attribute: dwarf::DW_AT_ranges)) { |
443 | if (Dval.getForm() == dwarf::DW_FORM_rnglistx) |
444 | return DU.findRnglistFromIndex(Index: Dval.getDIEInteger().getValue()); |
445 | |
446 | return DU.findRnglistFromOffset(Offset: Dval.getDIEInteger().getValue()); |
447 | } |
448 | |
449 | return DWARFAddressRangesVector(); |
450 | } |
451 | |
452 | static std::optional<uint64_t> getAsAddress(const DWARFUnit &DU, |
453 | const DIEValue &AttrVal) { |
454 | DWARFFormValue::ValueType Value(AttrVal.getDIEInteger().getValue()); |
455 | if (std::optional<object::SectionedAddress> SA = |
456 | DWARFFormValue::getAsSectionedAddress(Val: Value, Form: AttrVal.getForm(), U: &DU)) |
457 | return SA->Address; |
458 | return std::nullopt; |
459 | } |
460 | |
461 | /// Returns DWO Name to be used. Handles case where user specifies output DWO |
462 | /// directory, and there are duplicate names. Assumes DWO ID is unique. |
463 | static std::string |
464 | getDWOName(llvm::DWARFUnit &CU, |
465 | std::unordered_map<std::string, uint32_t> &NameToIndexMap) { |
466 | std::optional<uint64_t> DWOId = CU.getDWOId(); |
467 | assert(DWOId && "DWO ID not found." ); |
468 | (void)DWOId; |
469 | |
470 | std::string DWOName = dwarf::toString( |
471 | V: CU.getUnitDIE().find(Attrs: {dwarf::DW_AT_dwo_name, dwarf::DW_AT_GNU_dwo_name}), |
472 | Default: "" ); |
473 | assert(!DWOName.empty() && |
474 | "DW_AT_dwo_name/DW_AT_GNU_dwo_name does not exists." ); |
475 | if (!opts::DwarfOutputPath.empty()) { |
476 | DWOName = std::string(sys::path::filename(path: DWOName)); |
477 | auto Iter = NameToIndexMap.find(x: DWOName); |
478 | if (Iter == NameToIndexMap.end()) |
479 | Iter = NameToIndexMap.insert(x: {DWOName, 0}).first; |
480 | DWOName.append(str: std::to_string(val: Iter->second)); |
481 | ++Iter->second; |
482 | } |
483 | DWOName.append(s: ".dwo" ); |
484 | return DWOName; |
485 | } |
486 | |
487 | static std::unique_ptr<DIEStreamer> |
488 | createDIEStreamer(const Triple &TheTriple, raw_pwrite_stream &OutFile, |
489 | StringRef Swift5ReflectionSegmentName, DIEBuilder &DIEBldr, |
490 | DWARFRewriter &Rewriter) { |
491 | |
492 | std::unique_ptr<DIEStreamer> Streamer = std::make_unique<DIEStreamer>( |
493 | args: &DIEBldr, args&: Rewriter, args: DWARFLinkerBase::OutputFileType::Object, args&: OutFile, |
494 | args: [&](const Twine &Warning, StringRef Context, const DWARFDie *) {}); |
495 | Error Err = Streamer->init(TheTriple, Swift5ReflectionSegmentName); |
496 | if (Err) |
497 | errs() |
498 | << "BOLT-WARNING: [internal-dwarf-error]: Could not init DIEStreamer!" |
499 | << toString(E: std::move(Err)) << "\n" ; |
500 | return Streamer; |
501 | } |
502 | |
503 | static DWARFRewriter::UnitMeta |
504 | emitUnit(DIEBuilder &DIEBldr, DIEStreamer &Streamer, DWARFUnit &Unit) { |
505 | DIE *UnitDIE = DIEBldr.getUnitDIEbyUnit(DU: Unit); |
506 | const DIEBuilder::DWARFUnitInfo &U = DIEBldr.getUnitInfoByDwarfUnit(DwarfUnit: Unit); |
507 | Streamer.emitUnit(Unit, UnitDIE&: *UnitDIE); |
508 | uint64_t TypeHash = 0; |
509 | if (DWARFTypeUnit *DTU = dyn_cast_or_null<DWARFTypeUnit>(Val: &Unit)) |
510 | TypeHash = DTU->getTypeHash(); |
511 | return {.Offset: U.UnitOffset, .Length: U.UnitLength, .TUHash: TypeHash}; |
512 | } |
513 | |
514 | static void emitDWOBuilder(const std::string &DWOName, |
515 | DIEBuilder &DWODIEBuilder, DWARFRewriter &Rewriter, |
516 | DWARFUnit &SplitCU, DWARFUnit &CU, |
517 | DWARFRewriter::DWPState &State, |
518 | DebugLocWriter &LocWriter) { |
519 | // Populate debug_info and debug_abbrev for current dwo into StringRef. |
520 | DWODIEBuilder.generateAbbrevs(); |
521 | DWODIEBuilder.finish(); |
522 | |
523 | SmallVector<char, 20> OutBuffer; |
524 | std::shared_ptr<raw_svector_ostream> ObjOS = |
525 | std::make_shared<raw_svector_ostream>(args&: OutBuffer); |
526 | const object::ObjectFile *File = SplitCU.getContext().getDWARFObj().getFile(); |
527 | auto TheTriple = std::make_unique<Triple>(args: File->makeTriple()); |
528 | std::unique_ptr<DIEStreamer> Streamer = createDIEStreamer( |
529 | TheTriple: *TheTriple, OutFile&: *ObjOS, Swift5ReflectionSegmentName: "DwoStreamerInitAug2" , DIEBldr&: DWODIEBuilder, Rewriter); |
530 | DWARFRewriter::UnitMetaVectorType TUMetaVector; |
531 | DWARFRewriter::UnitMeta CUMI = {.Offset: 0, .Length: 0, .TUHash: 0}; |
532 | if (SplitCU.getContext().getMaxDWOVersion() >= 5) { |
533 | for (std::unique_ptr<llvm::DWARFUnit> &CU : |
534 | SplitCU.getContext().dwo_info_section_units()) { |
535 | if (!CU->isTypeUnit()) |
536 | continue; |
537 | DWARFRewriter::UnitMeta MI = |
538 | emitUnit(DIEBldr&: DWODIEBuilder, Streamer&: *Streamer, Unit&: *CU.get()); |
539 | TUMetaVector.emplace_back(args&: MI); |
540 | } |
541 | CUMI = emitUnit(DIEBldr&: DWODIEBuilder, Streamer&: *Streamer, Unit&: SplitCU); |
542 | } else { |
543 | for (std::unique_ptr<llvm::DWARFUnit> &CU : |
544 | SplitCU.getContext().dwo_compile_units()) |
545 | emitUnit(DIEBldr&: DWODIEBuilder, Streamer&: *Streamer, Unit&: *CU.get()); |
546 | |
547 | // emit debug_types sections for dwarf4 |
548 | for (DWARFUnit *CU : DWODIEBuilder.getDWARF4TUVector()) { |
549 | DWARFRewriter::UnitMeta MI = emitUnit(DIEBldr&: DWODIEBuilder, Streamer&: *Streamer, Unit&: *CU); |
550 | TUMetaVector.emplace_back(args&: MI); |
551 | } |
552 | } |
553 | |
554 | Streamer->emitAbbrevs(Abbrevs: DWODIEBuilder.getAbbrevs(), |
555 | DwarfVersion: SplitCU.getContext().getMaxVersion()); |
556 | Streamer->finish(); |
557 | |
558 | std::unique_ptr<MemoryBuffer> ObjectMemBuffer = |
559 | MemoryBuffer::getMemBuffer(InputData: ObjOS->str(), BufferName: "in-memory object file" , RequiresNullTerminator: false); |
560 | std::unique_ptr<object::ObjectFile> Obj = cantFail( |
561 | ValOrErr: object::ObjectFile::createObjectFile(Object: ObjectMemBuffer->getMemBufferRef()), |
562 | Msg: "error creating in-memory object" ); |
563 | |
564 | DWARFRewriter::OverriddenSectionsMap OverriddenSections; |
565 | for (const SectionRef &Secs : Obj->sections()) { |
566 | StringRef Contents = cantFail(ValOrErr: Secs.getContents()); |
567 | StringRef Name = cantFail(ValOrErr: Secs.getName()); |
568 | DWARFSectionKind Kind = |
569 | StringSwitch<DWARFSectionKind>(Name) |
570 | .Case(S: ".debug_abbrev" , Value: DWARFSectionKind::DW_SECT_ABBREV) |
571 | .Case(S: ".debug_info" , Value: DWARFSectionKind::DW_SECT_INFO) |
572 | .Case(S: ".debug_types" , Value: DWARFSectionKind::DW_SECT_EXT_TYPES) |
573 | .Default(Value: DWARFSectionKind::DW_SECT_EXT_unknown); |
574 | if (Kind == DWARFSectionKind::DW_SECT_EXT_unknown) |
575 | continue; |
576 | OverriddenSections[Kind] = Contents; |
577 | } |
578 | if (opts::WriteDWP) |
579 | Rewriter.updateDWP(CU, OverriddenSections, CUMI, TUMetaVector, State, |
580 | LocWriter); |
581 | else |
582 | Rewriter.writeDWOFiles(CU, OverriddenSections, DWOName, LocWriter); |
583 | } |
584 | |
585 | void DWARFRewriter::addStringHelper(DIEBuilder &DIEBldr, DIE &Die, |
586 | const DWARFUnit &Unit, |
587 | DIEValue &DIEAttrInfo, StringRef Str) { |
588 | uint32_t NewOffset = StrWriter->addString(Str); |
589 | if (Unit.getVersion() >= 5) { |
590 | StrOffstsWriter->updateAddressMap(Index: DIEAttrInfo.getDIEInteger().getValue(), |
591 | Address: NewOffset); |
592 | return; |
593 | } |
594 | DIEBldr.replaceValue(Die: &Die, Attribute: DIEAttrInfo.getAttribute(), Form: DIEAttrInfo.getForm(), |
595 | NewValue: DIEInteger(NewOffset)); |
596 | } |
597 | |
598 | using DWARFUnitVec = std::vector<DWARFUnit *>; |
599 | using CUPartitionVector = std::vector<DWARFUnitVec>; |
600 | /// Partitions CUs in to buckets. Bucket size is controlled by |
601 | /// cu-processing-batch-size. All the CUs that have cross CU reference reference |
602 | /// as a source are put in to the same initial bucket. |
603 | static CUPartitionVector partitionCUs(DWARFContext &DwCtx) { |
604 | CUPartitionVector Vec(2); |
605 | unsigned Counter = 0; |
606 | const DWARFDebugAbbrev *Abbr = DwCtx.getDebugAbbrev(); |
607 | for (std::unique_ptr<DWARFUnit> &CU : DwCtx.compile_units()) { |
608 | Expected<const DWARFAbbreviationDeclarationSet *> AbbrDeclSet = |
609 | Abbr->getAbbreviationDeclarationSet(CUAbbrOffset: CU->getAbbreviationsOffset()); |
610 | if (!AbbrDeclSet) { |
611 | consumeError(Err: AbbrDeclSet.takeError()); |
612 | return Vec; |
613 | } |
614 | bool CrossCURefFound = false; |
615 | for (const DWARFAbbreviationDeclaration &Decl : *AbbrDeclSet.get()) { |
616 | for (const DWARFAbbreviationDeclaration::AttributeSpec &Attr : |
617 | Decl.attributes()) { |
618 | if (Attr.Form == dwarf::DW_FORM_ref_addr) { |
619 | CrossCURefFound = true; |
620 | break; |
621 | } |
622 | } |
623 | if (CrossCURefFound) |
624 | break; |
625 | } |
626 | if (CrossCURefFound) { |
627 | Vec[0].push_back(x: CU.get()); |
628 | } else { |
629 | ++Counter; |
630 | Vec.back().push_back(x: CU.get()); |
631 | } |
632 | if (Counter % opts::BatchSize == 0 && !Vec.back().empty()) |
633 | Vec.push_back(x: {}); |
634 | } |
635 | return Vec; |
636 | } |
637 | |
638 | void DWARFRewriter::updateDebugInfo() { |
639 | ErrorOr<BinarySection &> DebugInfo = BC.getUniqueSectionByName(SectionName: ".debug_info" ); |
640 | if (!DebugInfo) |
641 | return; |
642 | |
643 | ARangesSectionWriter = std::make_unique<DebugARangesSectionWriter>(); |
644 | StrWriter = std::make_unique<DebugStrWriter>(args&: BC); |
645 | |
646 | StrOffstsWriter = std::make_unique<DebugStrOffsetsWriter>(); |
647 | |
648 | if (!opts::DeterministicDebugInfo) { |
649 | opts::DeterministicDebugInfo = true; |
650 | errs() << "BOLT-WARNING: --deterministic-debuginfo is being deprecated\n" ; |
651 | } |
652 | |
653 | if (BC.isDWARF5Used()) { |
654 | AddrWriter = std::make_unique<DebugAddrWriterDwarf5>(args: &BC); |
655 | RangeListsSectionWriter = std::make_unique<DebugRangeListsSectionWriter>(); |
656 | DebugRangeListsSectionWriter::setAddressWriter(AddrWriter.get()); |
657 | } else { |
658 | AddrWriter = std::make_unique<DebugAddrWriter>(args: &BC); |
659 | } |
660 | |
661 | if (BC.isDWARFLegacyUsed()) |
662 | LegacyRangesSectionWriter = std::make_unique<DebugRangesSectionWriter>(); |
663 | |
664 | DebugLoclistWriter::setAddressWriter(AddrWriter.get()); |
665 | |
666 | uint32_t CUIndex = 0; |
667 | std::mutex AccessMutex; |
668 | // Needs to be invoked in the same order as CUs are processed. |
669 | auto createRangeLocList = [&](DWARFUnit &CU) -> DebugLocWriter * { |
670 | std::lock_guard<std::mutex> Lock(AccessMutex); |
671 | const uint16_t DwarfVersion = CU.getVersion(); |
672 | if (DwarfVersion >= 5) { |
673 | LocListWritersByCU[CUIndex] = |
674 | std::make_unique<DebugLoclistWriter>(args&: CU, args: DwarfVersion, args: false); |
675 | |
676 | if (std::optional<uint64_t> DWOId = CU.getDWOId()) { |
677 | assert(RangeListsWritersByCU.count(*DWOId) == 0 && |
678 | "RangeLists writer for DWO unit already exists." ); |
679 | auto RangeListsSectionWriter = |
680 | std::make_unique<DebugRangeListsSectionWriter>(); |
681 | RangeListsSectionWriter->initSection(CU); |
682 | RangeListsWritersByCU[*DWOId] = std::move(RangeListsSectionWriter); |
683 | } |
684 | |
685 | } else { |
686 | LocListWritersByCU[CUIndex] = std::make_unique<DebugLocWriter>(); |
687 | } |
688 | return LocListWritersByCU[CUIndex++].get(); |
689 | }; |
690 | |
691 | // Unordered maps to handle name collision if output DWO directory is |
692 | // specified. |
693 | std::unordered_map<std::string, uint32_t> NameToIndexMap; |
694 | |
695 | auto updateDWONameCompDir = [&](DWARFUnit &Unit, DIEBuilder &DIEBldr, |
696 | DIE &UnitDIE) -> std::string { |
697 | DIEValue DWONameAttrInfo = UnitDIE.findAttribute(Attribute: dwarf::DW_AT_dwo_name); |
698 | if (!DWONameAttrInfo) |
699 | DWONameAttrInfo = UnitDIE.findAttribute(Attribute: dwarf::DW_AT_GNU_dwo_name); |
700 | assert(DWONameAttrInfo && "DW_AT_dwo_name is not in Skeleton CU." ); |
701 | std::string ObjectName; |
702 | |
703 | { |
704 | std::lock_guard<std::mutex> Lock(AccessMutex); |
705 | ObjectName = getDWOName(CU&: Unit, NameToIndexMap); |
706 | } |
707 | addStringHelper(DIEBldr, Die&: UnitDIE, Unit, DIEAttrInfo&: DWONameAttrInfo, |
708 | Str: ObjectName.c_str()); |
709 | |
710 | DIEValue CompDirAttrInfo = UnitDIE.findAttribute(Attribute: dwarf::DW_AT_comp_dir); |
711 | assert(CompDirAttrInfo && "DW_AT_comp_dir is not in Skeleton CU." ); |
712 | |
713 | if (!opts::DwarfOutputPath.empty()) { |
714 | if (!sys::fs::exists(Path: opts::DwarfOutputPath)) |
715 | sys::fs::create_directory(path: opts::DwarfOutputPath); |
716 | addStringHelper(DIEBldr, Die&: UnitDIE, Unit, DIEAttrInfo&: CompDirAttrInfo, |
717 | Str: opts::DwarfOutputPath.c_str()); |
718 | } |
719 | return ObjectName; |
720 | }; |
721 | |
722 | DWARF5AcceleratorTable DebugNamesTable(opts::CreateDebugNames, BC, |
723 | *StrWriter); |
724 | DWPState State; |
725 | if (opts::WriteDWP) |
726 | initDWPState(State); |
727 | auto processUnitDIE = [&](DWARFUnit *Unit, DIEBuilder *DIEBlder) { |
728 | // Check if the unit is a skeleton and we need special updates for it and |
729 | // its matching split/DWO CU. |
730 | std::optional<DWARFUnit *> SplitCU; |
731 | std::optional<uint64_t> RangesBase; |
732 | std::optional<uint64_t> DWOId = Unit->getDWOId(); |
733 | if (DWOId) |
734 | SplitCU = BC.getDWOCU(DWOId: *DWOId); |
735 | DebugLocWriter *DebugLocWriter = createRangeLocList(*Unit); |
736 | DebugRangesSectionWriter *RangesSectionWriter = |
737 | Unit->getVersion() >= 5 ? RangeListsSectionWriter.get() |
738 | : LegacyRangesSectionWriter.get(); |
739 | // Skipping CUs that failed to load. |
740 | if (SplitCU) { |
741 | DIEBuilder DWODIEBuilder(BC, &(*SplitCU)->getContext(), DebugNamesTable, |
742 | Unit); |
743 | DWODIEBuilder.buildDWOUnit(U&: **SplitCU); |
744 | std::string DWOName = updateDWONameCompDir( |
745 | *Unit, *DIEBlder, *DIEBlder->getUnitDIEbyUnit(DU: *Unit)); |
746 | |
747 | DebugLoclistWriter DebugLocDWoWriter(*Unit, Unit->getVersion(), true); |
748 | DebugRangesSectionWriter *TempRangesSectionWriter = RangesSectionWriter; |
749 | if (Unit->getVersion() >= 5) { |
750 | TempRangesSectionWriter = RangeListsWritersByCU[*DWOId].get(); |
751 | } else { |
752 | RangesBase = RangesSectionWriter->getSectionOffset(); |
753 | setDwoRangesBase(DWOId: *DWOId, RangesBase: *RangesBase); |
754 | } |
755 | |
756 | updateUnitDebugInfo(Unit&: *(*SplitCU), DIEBldr&: DWODIEBuilder, DebugLocWriter&: DebugLocDWoWriter, |
757 | RangesSectionWriter&: *TempRangesSectionWriter); |
758 | DebugLocDWoWriter.finalize(DIEBldr&: DWODIEBuilder, |
759 | Die&: *DWODIEBuilder.getUnitDIEbyUnit(DU: **SplitCU)); |
760 | if (Unit->getVersion() >= 5) |
761 | TempRangesSectionWriter->finalizeSection(); |
762 | |
763 | emitDWOBuilder(DWOName, DWODIEBuilder, Rewriter&: *this, SplitCU&: **SplitCU, CU&: *Unit, State, |
764 | LocWriter&: DebugLocDWoWriter); |
765 | } |
766 | |
767 | if (Unit->getVersion() >= 5) { |
768 | RangesBase = RangesSectionWriter->getSectionOffset() + |
769 | getDWARF5RngListLocListHeaderSize(); |
770 | RangesSectionWriter->initSection(CU&: *Unit); |
771 | StrOffstsWriter->finalizeSection(Unit&: *Unit, DIEBldr&: *DIEBlder); |
772 | } |
773 | |
774 | updateUnitDebugInfo(Unit&: *Unit, DIEBldr&: *DIEBlder, DebugLocWriter&: *DebugLocWriter, RangesSectionWriter&: *RangesSectionWriter, |
775 | RangesBase); |
776 | DebugLocWriter->finalize(DIEBldr&: *DIEBlder, Die&: *DIEBlder->getUnitDIEbyUnit(DU: *Unit)); |
777 | if (Unit->getVersion() >= 5) |
778 | RangesSectionWriter->finalizeSection(); |
779 | AddrWriter->update(DIEBlder&: *DIEBlder, CUs&: *Unit); |
780 | }; |
781 | |
782 | DIEBuilder DIEBlder(BC, BC.DwCtx.get(), DebugNamesTable); |
783 | DIEBlder.buildTypeUnits(StrOffsetWriter: StrOffstsWriter.get()); |
784 | SmallVector<char, 20> OutBuffer; |
785 | std::unique_ptr<raw_svector_ostream> ObjOS = |
786 | std::make_unique<raw_svector_ostream>(args&: OutBuffer); |
787 | const object::ObjectFile *File = BC.DwCtx->getDWARFObj().getFile(); |
788 | auto TheTriple = std::make_unique<Triple>(args: File->makeTriple()); |
789 | std::unique_ptr<DIEStreamer> Streamer = |
790 | createDIEStreamer(TheTriple: *TheTriple, OutFile&: *ObjOS, Swift5ReflectionSegmentName: "TypeStreamer" , DIEBldr&: DIEBlder, Rewriter&: *this); |
791 | CUOffsetMap OffsetMap = finalizeTypeSections(DIEBlder, Streamer&: *Streamer); |
792 | |
793 | const bool SingleThreadedMode = |
794 | opts::NoThreads || opts::DeterministicDebugInfo; |
795 | if (!SingleThreadedMode) |
796 | DIEBlder.buildCompileUnits(); |
797 | if (SingleThreadedMode) { |
798 | CUPartitionVector PartVec = partitionCUs(DwCtx&: *BC.DwCtx); |
799 | for (std::vector<DWARFUnit *> &Vec : PartVec) { |
800 | DIEBlder.buildCompileUnits(CUs: Vec); |
801 | for (DWARFUnit *CU : DIEBlder.getProcessedCUs()) |
802 | processUnitDIE(CU, &DIEBlder); |
803 | finalizeCompileUnits(DIEBlder, Streamer&: *Streamer, CUMap&: OffsetMap, |
804 | CUs: DIEBlder.getProcessedCUs()); |
805 | } |
806 | } else { |
807 | // Update unit debug info in parallel |
808 | ThreadPoolInterface &ThreadPool = ParallelUtilities::getThreadPool(); |
809 | for (std::unique_ptr<DWARFUnit> &CU : BC.DwCtx->compile_units()) |
810 | ThreadPool.async(F&: processUnitDIE, ArgList: CU.get(), ArgList: &DIEBlder); |
811 | ThreadPool.wait(); |
812 | } |
813 | |
814 | DebugNamesTable.emitAccelTable(); |
815 | |
816 | if (opts::WriteDWP) |
817 | finalizeDWP(State); |
818 | |
819 | finalizeDebugSections(DIEBlder, DebugNamesTable, Streamer&: *Streamer, ObjOS&: *ObjOS, |
820 | CUMap&: OffsetMap); |
821 | updateGdbIndexSection(CUMap&: OffsetMap, NumCUs: CUIndex); |
822 | } |
823 | |
824 | void DWARFRewriter::updateUnitDebugInfo( |
825 | DWARFUnit &Unit, DIEBuilder &DIEBldr, DebugLocWriter &DebugLocWriter, |
826 | DebugRangesSectionWriter &RangesSectionWriter, |
827 | std::optional<uint64_t> RangesBase) { |
828 | // Cache debug ranges so that the offset for identical ranges could be reused. |
829 | std::map<DebugAddressRangesVector, uint64_t> CachedRanges; |
830 | |
831 | uint64_t DIEOffset = Unit.getOffset() + Unit.getHeaderSize(); |
832 | uint64_t NextCUOffset = Unit.getNextUnitOffset(); |
833 | const std::vector<std::unique_ptr<DIEBuilder::DIEInfo>> &DIs = |
834 | DIEBldr.getDIEsByUnit(DU&: Unit); |
835 | |
836 | // Either updates or normalizes DW_AT_range to DW_AT_low_pc and DW_AT_high_pc. |
837 | auto updateLowPCHighPC = [&](DIE *Die, const DIEValue &LowPCVal, |
838 | const DIEValue &HighPCVal, uint64_t LowPC, |
839 | const uint64_t HighPC) { |
840 | dwarf::Attribute AttrLowPC = dwarf::DW_AT_low_pc; |
841 | dwarf::Form FormLowPC = dwarf::DW_FORM_addr; |
842 | dwarf::Attribute AttrHighPC = dwarf::DW_AT_high_pc; |
843 | dwarf::Form FormHighPC = dwarf::DW_FORM_data4; |
844 | const uint32_t Size = HighPC - LowPC; |
845 | // Whatever was generated is not low_pc/high_pc, so will reset to |
846 | // default for size 1. |
847 | if (!LowPCVal || !HighPCVal) { |
848 | if (Unit.getVersion() >= 5) |
849 | FormLowPC = dwarf::DW_FORM_addrx; |
850 | else if (Unit.isDWOUnit()) |
851 | FormLowPC = dwarf::DW_FORM_GNU_addr_index; |
852 | } else { |
853 | AttrLowPC = LowPCVal.getAttribute(); |
854 | FormLowPC = LowPCVal.getForm(); |
855 | AttrHighPC = HighPCVal.getAttribute(); |
856 | FormHighPC = HighPCVal.getForm(); |
857 | } |
858 | |
859 | if (FormLowPC == dwarf::DW_FORM_addrx || |
860 | FormLowPC == dwarf::DW_FORM_GNU_addr_index) |
861 | LowPC = AddrWriter->getIndexFromAddress(Address: LowPC, CU&: Unit); |
862 | |
863 | if (LowPCVal) |
864 | DIEBldr.replaceValue(Die, Attribute: AttrLowPC, Form: FormLowPC, NewValue: DIEInteger(LowPC)); |
865 | else |
866 | DIEBldr.addValue(Die, Attribute: AttrLowPC, Form: FormLowPC, Value: DIEInteger(LowPC)); |
867 | if (HighPCVal) { |
868 | DIEBldr.replaceValue(Die, Attribute: AttrHighPC, Form: FormHighPC, NewValue: DIEInteger(Size)); |
869 | } else { |
870 | DIEBldr.deleteValue(Die, Attribute: dwarf::DW_AT_ranges); |
871 | DIEBldr.addValue(Die, Attribute: AttrHighPC, Form: FormHighPC, Value: DIEInteger(Size)); |
872 | } |
873 | }; |
874 | |
875 | for (const std::unique_ptr<DIEBuilder::DIEInfo> &DI : DIs) { |
876 | DIE *Die = DI->Die; |
877 | switch (Die->getTag()) { |
878 | case dwarf::DW_TAG_compile_unit: |
879 | case dwarf::DW_TAG_skeleton_unit: { |
880 | // For dwarf5 section 3.1.3 |
881 | // The following attributes are not part of a split full compilation unit |
882 | // entry but instead are inherited (if present) from the corresponding |
883 | // skeleton compilation unit: DW_AT_low_pc, DW_AT_high_pc, DW_AT_ranges, |
884 | // DW_AT_stmt_list, DW_AT_comp_dir, DW_AT_str_offsets_base, |
885 | // DW_AT_addr_base and DW_AT_rnglists_base. |
886 | if (Unit.getVersion() == 5 && Unit.isDWOUnit()) |
887 | continue; |
888 | auto ModuleRangesOrError = getDIEAddressRanges(Die: *Die, DU&: Unit); |
889 | if (!ModuleRangesOrError) { |
890 | consumeError(Err: ModuleRangesOrError.takeError()); |
891 | break; |
892 | } |
893 | DWARFAddressRangesVector &ModuleRanges = *ModuleRangesOrError; |
894 | DebugAddressRangesVector OutputRanges = |
895 | BC.translateModuleAddressRanges(InputRanges: ModuleRanges); |
896 | DIEValue LowPCAttrInfo = Die->findAttribute(Attribute: dwarf::DW_AT_low_pc); |
897 | // For a case where LLD GCs only function used in the CU. |
898 | // If CU doesn't have DW_AT_low_pc we are not going to convert, |
899 | // so don't need to do anything. |
900 | if (OutputRanges.empty() && !Unit.isDWOUnit() && LowPCAttrInfo) |
901 | OutputRanges.push_back(Elt: {0, 0}); |
902 | const uint64_t RangesSectionOffset = |
903 | RangesSectionWriter.addRanges(Ranges&: OutputRanges); |
904 | // Don't emit the zero low_pc arange. |
905 | if (!Unit.isDWOUnit() && !OutputRanges.empty() && |
906 | OutputRanges.back().LowPC) |
907 | ARangesSectionWriter->addCURanges(CUOffset: Unit.getOffset(), |
908 | Ranges: std::move(OutputRanges)); |
909 | updateDWARFObjectAddressRanges(Unit, DIEBldr, Die&: *Die, DebugRangesOffset: RangesSectionOffset, |
910 | RangesBase); |
911 | DIEValue StmtListAttrVal = Die->findAttribute(Attribute: dwarf::DW_AT_stmt_list); |
912 | if (LineTablePatchMap.count(x: &Unit)) |
913 | DIEBldr.replaceValue(Die, Attribute: dwarf::DW_AT_stmt_list, |
914 | Form: StmtListAttrVal.getForm(), |
915 | NewValue: DIEInteger(LineTablePatchMap[&Unit])); |
916 | break; |
917 | } |
918 | |
919 | case dwarf::DW_TAG_subprogram: { |
920 | // Get function address either from ranges or [LowPC, HighPC) pair. |
921 | uint64_t Address = UINT64_MAX; |
922 | uint64_t SectionIndex, HighPC; |
923 | DebugAddressRangesVector FunctionRanges; |
924 | if (!getLowAndHighPC(Die: *Die, DU: Unit, LowPC&: Address, HighPC, SectionIndex)) { |
925 | Expected<DWARFAddressRangesVector> RangesOrError = |
926 | getDIEAddressRanges(Die: *Die, DU&: Unit); |
927 | if (!RangesOrError) { |
928 | consumeError(Err: RangesOrError.takeError()); |
929 | break; |
930 | } |
931 | DWARFAddressRangesVector Ranges = *RangesOrError; |
932 | // Not a function definition. |
933 | if (Ranges.empty()) |
934 | break; |
935 | |
936 | for (const DWARFAddressRange &Range : Ranges) { |
937 | if (const BinaryFunction *Function = |
938 | BC.getBinaryFunctionAtAddress(Address: Range.LowPC)) |
939 | FunctionRanges.append(RHS: Function->getOutputAddressRanges()); |
940 | } |
941 | } else { |
942 | if (const BinaryFunction *Function = |
943 | BC.getBinaryFunctionAtAddress(Address)) |
944 | FunctionRanges = Function->getOutputAddressRanges(); |
945 | } |
946 | |
947 | // Clear cached ranges as the new function will have its own set. |
948 | CachedRanges.clear(); |
949 | DIEValue LowPCVal = Die->findAttribute(Attribute: dwarf::DW_AT_low_pc); |
950 | DIEValue HighPCVal = Die->findAttribute(Attribute: dwarf::DW_AT_high_pc); |
951 | if (FunctionRanges.empty()) { |
952 | if (LowPCVal && HighPCVal) |
953 | FunctionRanges.push_back(Elt: {0, HighPCVal.getDIEInteger().getValue()}); |
954 | else |
955 | FunctionRanges.push_back(Elt: {0, 1}); |
956 | } |
957 | |
958 | if (FunctionRanges.size() == 1 && !opts::AlwaysConvertToRanges) { |
959 | updateLowPCHighPC(Die, LowPCVal, HighPCVal, FunctionRanges.back().LowPC, |
960 | FunctionRanges.back().HighPC); |
961 | break; |
962 | } |
963 | |
964 | updateDWARFObjectAddressRanges( |
965 | Unit, DIEBldr, Die&: *Die, DebugRangesOffset: RangesSectionWriter.addRanges(Ranges&: FunctionRanges)); |
966 | |
967 | break; |
968 | } |
969 | case dwarf::DW_TAG_lexical_block: |
970 | case dwarf::DW_TAG_inlined_subroutine: |
971 | case dwarf::DW_TAG_try_block: |
972 | case dwarf::DW_TAG_catch_block: { |
973 | uint64_t RangesSectionOffset = 0; |
974 | Expected<DWARFAddressRangesVector> RangesOrError = |
975 | getDIEAddressRanges(Die: *Die, DU&: Unit); |
976 | const BinaryFunction *Function = |
977 | RangesOrError && !RangesOrError->empty() |
978 | ? BC.getBinaryFunctionContainingAddress( |
979 | Address: RangesOrError->front().LowPC) |
980 | : nullptr; |
981 | DebugAddressRangesVector OutputRanges; |
982 | if (Function) { |
983 | OutputRanges = translateInputToOutputRanges(BF: *Function, InputRanges: *RangesOrError); |
984 | LLVM_DEBUG(if (OutputRanges.empty() != RangesOrError->empty()) { |
985 | dbgs() << "BOLT-DEBUG: problem with DIE at 0x" |
986 | << Twine::utohexstr(Die->getOffset()) << " in CU at 0x" |
987 | << Twine::utohexstr(Unit.getOffset()) << '\n'; |
988 | }); |
989 | if (opts::AlwaysConvertToRanges || OutputRanges.size() > 1) { |
990 | RangesSectionOffset = RangesSectionWriter.addRanges( |
991 | Ranges: std::move(OutputRanges), CachedRanges); |
992 | OutputRanges.clear(); |
993 | } else if (OutputRanges.empty()) { |
994 | OutputRanges.push_back(Elt: {0, RangesOrError.get().front().HighPC}); |
995 | } |
996 | } else if (!RangesOrError) { |
997 | consumeError(Err: RangesOrError.takeError()); |
998 | } else { |
999 | OutputRanges.push_back(Elt: {0, !RangesOrError->empty() |
1000 | ? RangesOrError.get().front().HighPC |
1001 | : 0}); |
1002 | } |
1003 | DIEValue LowPCVal = Die->findAttribute(Attribute: dwarf::DW_AT_low_pc); |
1004 | DIEValue HighPCVal = Die->findAttribute(Attribute: dwarf::DW_AT_high_pc); |
1005 | if (OutputRanges.size() == 1) { |
1006 | updateLowPCHighPC(Die, LowPCVal, HighPCVal, OutputRanges.back().LowPC, |
1007 | OutputRanges.back().HighPC); |
1008 | break; |
1009 | } |
1010 | updateDWARFObjectAddressRanges(Unit, DIEBldr, Die&: *Die, DebugRangesOffset: RangesSectionOffset); |
1011 | break; |
1012 | } |
1013 | case dwarf::DW_TAG_call_site: { |
1014 | auto patchPC = [&](DIE *Die, DIEValue &AttrVal, StringRef Entry) -> void { |
1015 | std::optional<uint64_t> Address = getAsAddress(DU: Unit, AttrVal); |
1016 | const BinaryFunction *Function = |
1017 | BC.getBinaryFunctionContainingAddress(Address: *Address); |
1018 | uint64_t UpdatedAddress = *Address; |
1019 | if (Function) |
1020 | UpdatedAddress = |
1021 | Function->translateInputToOutputAddress(Address: UpdatedAddress); |
1022 | |
1023 | if (AttrVal.getForm() == dwarf::DW_FORM_addrx) { |
1024 | const uint32_t Index = |
1025 | AddrWriter->getIndexFromAddress(Address: UpdatedAddress, CU&: Unit); |
1026 | DIEBldr.replaceValue(Die, Attribute: AttrVal.getAttribute(), Form: AttrVal.getForm(), |
1027 | NewValue: DIEInteger(Index)); |
1028 | } else if (AttrVal.getForm() == dwarf::DW_FORM_addr) { |
1029 | DIEBldr.replaceValue(Die, Attribute: AttrVal.getAttribute(), Form: AttrVal.getForm(), |
1030 | NewValue: DIEInteger(UpdatedAddress)); |
1031 | } else { |
1032 | errs() << "BOLT-ERROR: unsupported form for " << Entry << "\n" ; |
1033 | } |
1034 | }; |
1035 | DIEValue CallPcAttrVal = Die->findAttribute(Attribute: dwarf::DW_AT_call_pc); |
1036 | if (CallPcAttrVal) |
1037 | patchPC(Die, CallPcAttrVal, "DW_AT_call_pc" ); |
1038 | |
1039 | DIEValue CallRetPcAttrVal = |
1040 | Die->findAttribute(Attribute: dwarf::DW_AT_call_return_pc); |
1041 | if (CallRetPcAttrVal) |
1042 | patchPC(Die, CallRetPcAttrVal, "DW_AT_call_return_pc" ); |
1043 | |
1044 | break; |
1045 | } |
1046 | default: { |
1047 | // Handle any tag that can have DW_AT_location attribute. |
1048 | DIEValue LocAttrInfo = Die->findAttribute(Attribute: dwarf::DW_AT_location); |
1049 | DIEValue LowPCAttrInfo = Die->findAttribute(Attribute: dwarf::DW_AT_low_pc); |
1050 | if (LocAttrInfo) { |
1051 | if (doesFormBelongToClass(Form: LocAttrInfo.getForm(), |
1052 | FC: DWARFFormValue::FC_Constant, |
1053 | DwarfVersion: Unit.getVersion()) || |
1054 | doesFormBelongToClass(Form: LocAttrInfo.getForm(), |
1055 | FC: DWARFFormValue::FC_SectionOffset, |
1056 | DwarfVersion: Unit.getVersion())) { |
1057 | uint64_t Offset = LocAttrInfo.getForm() == dwarf::DW_FORM_loclistx |
1058 | ? LocAttrInfo.getDIELocList().getValue() |
1059 | : LocAttrInfo.getDIEInteger().getValue(); |
1060 | DebugLocationsVector InputLL; |
1061 | |
1062 | std::optional<object::SectionedAddress> SectionAddress = |
1063 | Unit.getBaseAddress(); |
1064 | uint64_t BaseAddress = 0; |
1065 | if (SectionAddress) |
1066 | BaseAddress = SectionAddress->Address; |
1067 | |
1068 | if (Unit.getVersion() >= 5 && |
1069 | LocAttrInfo.getForm() == dwarf::DW_FORM_loclistx) { |
1070 | std::optional<uint64_t> LocOffset = Unit.getLoclistOffset(Index: Offset); |
1071 | assert(LocOffset && "Location Offset is invalid." ); |
1072 | Offset = *LocOffset; |
1073 | } |
1074 | |
1075 | Error E = Unit.getLocationTable().visitLocationList( |
1076 | Offset: &Offset, Callback: [&](const DWARFLocationEntry &Entry) { |
1077 | switch (Entry.Kind) { |
1078 | default: |
1079 | llvm_unreachable("Unsupported DWARFLocationEntry Kind." ); |
1080 | case dwarf::DW_LLE_end_of_list: |
1081 | return false; |
1082 | case dwarf::DW_LLE_base_address: { |
1083 | assert(Entry.SectionIndex == SectionedAddress::UndefSection && |
1084 | "absolute address expected" ); |
1085 | BaseAddress = Entry.Value0; |
1086 | break; |
1087 | } |
1088 | case dwarf::DW_LLE_offset_pair: |
1089 | assert( |
1090 | (Entry.SectionIndex == SectionedAddress::UndefSection && |
1091 | (!Unit.isDWOUnit() || Unit.getVersion() == 5)) && |
1092 | "absolute address expected" ); |
1093 | InputLL.emplace_back(Args: DebugLocationEntry{ |
1094 | .LowPC: BaseAddress + Entry.Value0, .HighPC: BaseAddress + Entry.Value1, |
1095 | .Expr: Entry.Loc}); |
1096 | break; |
1097 | case dwarf::DW_LLE_start_length: |
1098 | InputLL.emplace_back(Args: DebugLocationEntry{ |
1099 | .LowPC: Entry.Value0, .HighPC: Entry.Value0 + Entry.Value1, .Expr: Entry.Loc}); |
1100 | break; |
1101 | case dwarf::DW_LLE_base_addressx: { |
1102 | std::optional<object::SectionedAddress> EntryAddress = |
1103 | Unit.getAddrOffsetSectionItem(Index: Entry.Value0); |
1104 | assert(EntryAddress && "base Address not found." ); |
1105 | BaseAddress = EntryAddress->Address; |
1106 | break; |
1107 | } |
1108 | case dwarf::DW_LLE_startx_length: { |
1109 | std::optional<object::SectionedAddress> EntryAddress = |
1110 | Unit.getAddrOffsetSectionItem(Index: Entry.Value0); |
1111 | assert(EntryAddress && "Address does not exist." ); |
1112 | InputLL.emplace_back(Args: DebugLocationEntry{ |
1113 | .LowPC: EntryAddress->Address, |
1114 | .HighPC: EntryAddress->Address + Entry.Value1, .Expr: Entry.Loc}); |
1115 | break; |
1116 | } |
1117 | case dwarf::DW_LLE_startx_endx: { |
1118 | std::optional<object::SectionedAddress> StartAddress = |
1119 | Unit.getAddrOffsetSectionItem(Index: Entry.Value0); |
1120 | assert(StartAddress && "Start Address does not exist." ); |
1121 | std::optional<object::SectionedAddress> EndAddress = |
1122 | Unit.getAddrOffsetSectionItem(Index: Entry.Value1); |
1123 | assert(EndAddress && "Start Address does not exist." ); |
1124 | InputLL.emplace_back(Args: DebugLocationEntry{ |
1125 | .LowPC: StartAddress->Address, .HighPC: EndAddress->Address, .Expr: Entry.Loc}); |
1126 | break; |
1127 | } |
1128 | } |
1129 | return true; |
1130 | }); |
1131 | |
1132 | if (E || InputLL.empty()) { |
1133 | consumeError(Err: std::move(E)); |
1134 | errs() << "BOLT-WARNING: empty location list detected at 0x" |
1135 | << Twine::utohexstr(Val: Offset) << " for DIE at 0x" << Die |
1136 | << " in CU at 0x" << Twine::utohexstr(Val: Unit.getOffset()) |
1137 | << '\n'; |
1138 | } else { |
1139 | const uint64_t Address = InputLL.front().LowPC; |
1140 | DebugLocationsVector OutputLL; |
1141 | if (const BinaryFunction *Function = |
1142 | BC.getBinaryFunctionContainingAddress(Address)) { |
1143 | OutputLL = translateInputToOutputLocationList(BF: *Function, InputLL); |
1144 | LLVM_DEBUG(if (OutputLL.empty()) { |
1145 | dbgs() << "BOLT-DEBUG: location list translated to an empty " |
1146 | "one at 0x" |
1147 | << Die << " in CU at 0x" |
1148 | << Twine::utohexstr(Unit.getOffset()) << '\n'; |
1149 | }); |
1150 | } else { |
1151 | // It's possible for a subprogram to be removed and to have |
1152 | // address of 0. Adding this entry to output to preserve debug |
1153 | // information. |
1154 | OutputLL = InputLL; |
1155 | } |
1156 | DebugLocWriter.addList(DIEBldr, Die&: *Die, AttrInfo&: LocAttrInfo, LocList&: OutputLL); |
1157 | } |
1158 | } else { |
1159 | assert((doesFormBelongToClass(LocAttrInfo.getForm(), |
1160 | DWARFFormValue::FC_Exprloc, |
1161 | Unit.getVersion()) || |
1162 | doesFormBelongToClass(LocAttrInfo.getForm(), |
1163 | DWARFFormValue::FC_Block, |
1164 | Unit.getVersion())) && |
1165 | "unexpected DW_AT_location form" ); |
1166 | if (Unit.isDWOUnit() || Unit.getVersion() >= 5) { |
1167 | std::vector<uint8_t> Sblock; |
1168 | DIEValueList *AttrLocValList; |
1169 | if (doesFormBelongToClass(Form: LocAttrInfo.getForm(), |
1170 | FC: DWARFFormValue::FC_Exprloc, |
1171 | DwarfVersion: Unit.getVersion())) { |
1172 | for (const DIEValue &Val : LocAttrInfo.getDIELoc().values()) { |
1173 | Sblock.push_back(x: Val.getDIEInteger().getValue()); |
1174 | } |
1175 | DIELoc *LocAttr = const_cast<DIELoc *>(&LocAttrInfo.getDIELoc()); |
1176 | AttrLocValList = static_cast<DIEValueList *>(LocAttr); |
1177 | } else { |
1178 | for (const DIEValue &Val : LocAttrInfo.getDIEBlock().values()) { |
1179 | Sblock.push_back(x: Val.getDIEInteger().getValue()); |
1180 | } |
1181 | DIEBlock *BlockAttr = |
1182 | const_cast<DIEBlock *>(&LocAttrInfo.getDIEBlock()); |
1183 | AttrLocValList = static_cast<DIEValueList *>(BlockAttr); |
1184 | } |
1185 | ArrayRef<uint8_t> Expr = ArrayRef<uint8_t>(Sblock); |
1186 | DataExtractor Data( |
1187 | StringRef((const char *)Expr.data(), Expr.size()), |
1188 | Unit.getContext().isLittleEndian(), 0); |
1189 | DWARFExpression LocExpr(Data, Unit.getAddressByteSize(), |
1190 | Unit.getFormParams().Format); |
1191 | uint32_t PrevOffset = 0; |
1192 | DIEValueList *NewAttr; |
1193 | DIEValue Value; |
1194 | uint32_t NewExprSize = 0; |
1195 | DIELoc *Loc = nullptr; |
1196 | DIEBlock *Block = nullptr; |
1197 | if (LocAttrInfo.getForm() == dwarf::DW_FORM_exprloc) { |
1198 | Loc = DIEBldr.allocateDIEValue<DIELoc>(); |
1199 | NewAttr = Loc; |
1200 | Value = DIEValue(LocAttrInfo.getAttribute(), |
1201 | LocAttrInfo.getForm(), Loc); |
1202 | } else if (doesFormBelongToClass(Form: LocAttrInfo.getForm(), |
1203 | FC: DWARFFormValue::FC_Block, |
1204 | DwarfVersion: Unit.getVersion())) { |
1205 | Block = DIEBldr.allocateDIEValue<DIEBlock>(); |
1206 | NewAttr = Block; |
1207 | Value = DIEValue(LocAttrInfo.getAttribute(), |
1208 | LocAttrInfo.getForm(), Block); |
1209 | } else { |
1210 | errs() << "BOLT-WARNING: Unexpected Form value in Updating " |
1211 | "DW_AT_Location\n" ; |
1212 | continue; |
1213 | } |
1214 | |
1215 | for (const DWARFExpression::Operation &Expr : LocExpr) { |
1216 | uint32_t CurEndOffset = PrevOffset + 1; |
1217 | if (Expr.getDescription().Op.size() == 1) |
1218 | CurEndOffset = Expr.getOperandEndOffset(Idx: 0); |
1219 | if (Expr.getDescription().Op.size() == 2) |
1220 | CurEndOffset = Expr.getOperandEndOffset(Idx: 1); |
1221 | if (Expr.getDescription().Op.size() > 2) |
1222 | errs() << "BOLT-WARNING: [internal-dwarf-error]: Unsupported " |
1223 | "number of operands.\n" ; |
1224 | // not addr index, just copy. |
1225 | if (!(Expr.getCode() == dwarf::DW_OP_GNU_addr_index || |
1226 | Expr.getCode() == dwarf::DW_OP_addrx)) { |
1227 | auto Itr = AttrLocValList->values().begin(); |
1228 | std::advance(i&: Itr, n: PrevOffset); |
1229 | uint32_t CopyNum = CurEndOffset - PrevOffset; |
1230 | NewExprSize += CopyNum; |
1231 | while (CopyNum--) { |
1232 | DIEBldr.addValue(Die: NewAttr, V: *Itr); |
1233 | std::advance(i&: Itr, n: 1); |
1234 | } |
1235 | } else { |
1236 | const uint64_t Index = Expr.getRawOperand(Idx: 0); |
1237 | std::optional<object::SectionedAddress> EntryAddress = |
1238 | Unit.getAddrOffsetSectionItem(Index); |
1239 | assert(EntryAddress && "Address is not found." ); |
1240 | assert(Index <= std::numeric_limits<uint32_t>::max() && |
1241 | "Invalid Operand Index." ); |
1242 | const uint32_t AddrIndex = AddrWriter->getIndexFromAddress( |
1243 | Address: EntryAddress->Address, CU&: Unit); |
1244 | // update Index into .debug_address section for DW_AT_location. |
1245 | // The Size field is not stored in IR, we need to minus 1 in |
1246 | // offset for each expr. |
1247 | SmallString<8> Tmp; |
1248 | raw_svector_ostream OSE(Tmp); |
1249 | encodeULEB128(Value: AddrIndex, OS&: OSE); |
1250 | |
1251 | DIEBldr.addValue(Die: NewAttr, Attribute: static_cast<dwarf::Attribute>(0), |
1252 | Form: dwarf::DW_FORM_data1, |
1253 | Value: DIEInteger(Expr.getCode())); |
1254 | NewExprSize += 1; |
1255 | for (uint8_t Byte : Tmp) { |
1256 | DIEBldr.addValue(Die: NewAttr, Attribute: static_cast<dwarf::Attribute>(0), |
1257 | Form: dwarf::DW_FORM_data1, Value: DIEInteger(Byte)); |
1258 | NewExprSize += 1; |
1259 | } |
1260 | } |
1261 | PrevOffset = CurEndOffset; |
1262 | } |
1263 | |
1264 | // update the size since the index might be changed |
1265 | if (Loc) |
1266 | Loc->setSize(NewExprSize); |
1267 | else |
1268 | Block->setSize(NewExprSize); |
1269 | DIEBldr.replaceValue(Die, Attribute: LocAttrInfo.getAttribute(), |
1270 | Form: LocAttrInfo.getForm(), NewValue&: Value); |
1271 | } |
1272 | } |
1273 | } else if (LowPCAttrInfo) { |
1274 | uint64_t Address = 0; |
1275 | uint64_t SectionIndex = 0; |
1276 | if (getLowPC(Die: *Die, DU: Unit, LowPC&: Address, SectionIndex)) { |
1277 | uint64_t NewAddress = 0; |
1278 | if (const BinaryFunction *Function = |
1279 | BC.getBinaryFunctionContainingAddress(Address)) { |
1280 | NewAddress = Function->translateInputToOutputAddress(Address); |
1281 | LLVM_DEBUG(dbgs() |
1282 | << "BOLT-DEBUG: Fixing low_pc 0x" |
1283 | << Twine::utohexstr(Address) << " for DIE with tag " |
1284 | << Die->getTag() << " to 0x" |
1285 | << Twine::utohexstr(NewAddress) << '\n'); |
1286 | } |
1287 | |
1288 | dwarf::Form Form = LowPCAttrInfo.getForm(); |
1289 | assert(Form != dwarf::DW_FORM_LLVM_addrx_offset && |
1290 | "DW_FORM_LLVM_addrx_offset is not supported" ); |
1291 | std::lock_guard<std::mutex> Lock(DWARFRewriterMutex); |
1292 | if (Form == dwarf::DW_FORM_addrx || |
1293 | Form == dwarf::DW_FORM_GNU_addr_index) { |
1294 | const uint32_t Index = AddrWriter->getIndexFromAddress( |
1295 | Address: NewAddress ? NewAddress : Address, CU&: Unit); |
1296 | DIEBldr.replaceValue(Die, Attribute: LowPCAttrInfo.getAttribute(), |
1297 | Form: LowPCAttrInfo.getForm(), NewValue: DIEInteger(Index)); |
1298 | } else { |
1299 | DIEBldr.replaceValue(Die, Attribute: LowPCAttrInfo.getAttribute(), |
1300 | Form: LowPCAttrInfo.getForm(), |
1301 | NewValue: DIEInteger(NewAddress)); |
1302 | } |
1303 | } else if (opts::Verbosity >= 1) { |
1304 | errs() << "BOLT-WARNING: unexpected form value for attribute " |
1305 | "LowPCAttrInfo\n" ; |
1306 | } |
1307 | } |
1308 | } |
1309 | } |
1310 | } |
1311 | if (DIEOffset > NextCUOffset) |
1312 | errs() << "BOLT-WARNING: corrupt DWARF detected at 0x" |
1313 | << Twine::utohexstr(Val: Unit.getOffset()) << '\n'; |
1314 | } |
1315 | |
1316 | void DWARFRewriter::updateDWARFObjectAddressRanges( |
1317 | DWARFUnit &Unit, DIEBuilder &DIEBldr, DIE &Die, uint64_t DebugRangesOffset, |
1318 | std::optional<uint64_t> RangesBase) { |
1319 | |
1320 | if (RangesBase) { |
1321 | // If DW_AT_GNU_ranges_base is present, update it. No further modifications |
1322 | // are needed for ranges base. |
1323 | |
1324 | DIEValue RangesBaseInfo = Die.findAttribute(Attribute: dwarf::DW_AT_GNU_ranges_base); |
1325 | if (!RangesBaseInfo) { |
1326 | RangesBaseInfo = Die.findAttribute(Attribute: dwarf::DW_AT_rnglists_base); |
1327 | } |
1328 | |
1329 | if (RangesBaseInfo) { |
1330 | DIEBldr.replaceValue(Die: &Die, Attribute: RangesBaseInfo.getAttribute(), |
1331 | Form: RangesBaseInfo.getForm(), |
1332 | NewValue: DIEInteger(static_cast<uint32_t>(*RangesBase))); |
1333 | RangesBase = std::nullopt; |
1334 | } |
1335 | } |
1336 | |
1337 | DIEValue LowPCAttrInfo = Die.findAttribute(Attribute: dwarf::DW_AT_low_pc); |
1338 | DIEValue RangesAttrInfo = Die.findAttribute(Attribute: dwarf::DW_AT_ranges); |
1339 | if (RangesAttrInfo) { |
1340 | // Case 1: The object was already non-contiguous and had DW_AT_ranges. |
1341 | // In this case we simply need to update the value of DW_AT_ranges |
1342 | // and introduce DW_AT_GNU_ranges_base if required. |
1343 | // For DWARF5 converting all of DW_AT_ranges into DW_FORM_rnglistx |
1344 | bool NeedConverted = false; |
1345 | |
1346 | if (Unit.getVersion() >= 5 && |
1347 | RangesAttrInfo.getForm() == dwarf::DW_FORM_sec_offset) |
1348 | NeedConverted = true; |
1349 | |
1350 | uint64_t CurRangeBase = 0; |
1351 | if (Unit.isDWOUnit()) { |
1352 | if (std::optional<uint64_t> DWOId = Unit.getDWOId()) |
1353 | CurRangeBase = getDwoRangesBase(DWOId: *DWOId); |
1354 | else |
1355 | errs() << "BOLT-WARNING: [internal-dwarf-error]: DWOId is not found " |
1356 | "for DWO Unit." ; |
1357 | } |
1358 | if (NeedConverted || RangesAttrInfo.getForm() == dwarf::DW_FORM_rnglistx) |
1359 | DIEBldr.replaceValue(Die: &Die, Attribute: dwarf::DW_AT_ranges, Form: dwarf::DW_FORM_rnglistx, |
1360 | NewValue: DIEInteger(DebugRangesOffset)); |
1361 | else |
1362 | DIEBldr.replaceValue(Die: &Die, Attribute: dwarf::DW_AT_ranges, Form: RangesAttrInfo.getForm(), |
1363 | NewValue: DIEInteger(DebugRangesOffset - CurRangeBase)); |
1364 | |
1365 | if (!RangesBase) { |
1366 | if (LowPCAttrInfo && |
1367 | LowPCAttrInfo.getForm() != dwarf::DW_FORM_GNU_addr_index && |
1368 | LowPCAttrInfo.getForm() != dwarf::DW_FORM_addrx) |
1369 | DIEBldr.replaceValue(Die: &Die, Attribute: dwarf::DW_AT_low_pc, Form: LowPCAttrInfo.getForm(), |
1370 | NewValue: DIEInteger(0)); |
1371 | return; |
1372 | } |
1373 | |
1374 | if (!(Die.getTag() == dwarf::DW_TAG_compile_unit || |
1375 | Die.getTag() == dwarf::DW_TAG_skeleton_unit)) |
1376 | return; |
1377 | |
1378 | // If we are at this point we are in the CU/Skeleton CU, and |
1379 | // DW_AT_GNU_ranges_base or DW_AT_rnglists_base doesn't exist. |
1380 | if (Unit.getVersion() <= 4) |
1381 | DIEBldr.addValue(Die: &Die, Attribute: dwarf::DW_AT_GNU_ranges_base, Form: dwarf::DW_FORM_data4, |
1382 | Value: DIEInteger(*RangesBase)); |
1383 | else if (Unit.getVersion() == 5) |
1384 | DIEBldr.addValue(Die: &Die, Attribute: dwarf::DW_AT_rnglists_base, |
1385 | Form: dwarf::DW_FORM_sec_offset, Value: DIEInteger(*RangesBase)); |
1386 | else |
1387 | DIEBldr.addValue(Die: &Die, Attribute: dwarf::DW_AT_rnglists_base, |
1388 | Form: dwarf::DW_FORM_sec_offset, Value: DIEInteger(*RangesBase)); |
1389 | return; |
1390 | } |
1391 | |
1392 | // Case 2: The object has both DW_AT_low_pc and DW_AT_high_pc emitted back |
1393 | // to back. Replace with new attributes and patch the DIE. |
1394 | DIEValue HighPCAttrInfo = Die.findAttribute(Attribute: dwarf::DW_AT_high_pc); |
1395 | if (LowPCAttrInfo && HighPCAttrInfo) { |
1396 | |
1397 | convertToRangesPatchDebugInfo(Unit, DIEBldr, Die, RangesSectionOffset: DebugRangesOffset, |
1398 | LowPCAttrInfo, HighPCAttrInfo, RangesBase); |
1399 | } else if (!(Unit.isDWOUnit() && |
1400 | Die.getTag() == dwarf::DW_TAG_compile_unit)) { |
1401 | if (opts::Verbosity >= 1) |
1402 | errs() << "BOLT-WARNING: cannot update ranges for DIE in Unit offset 0x" |
1403 | << Unit.getOffset() << '\n'; |
1404 | } |
1405 | } |
1406 | |
1407 | void DWARFRewriter::updateLineTableOffsets(const MCAsmLayout &Layout) { |
1408 | ErrorOr<BinarySection &> DbgInfoSection = |
1409 | BC.getUniqueSectionByName(SectionName: ".debug_info" ); |
1410 | ErrorOr<BinarySection &> TypeInfoSection = |
1411 | BC.getUniqueSectionByName(SectionName: ".debug_types" ); |
1412 | assert(((BC.DwCtx->getNumTypeUnits() > 0 && TypeInfoSection) || |
1413 | BC.DwCtx->getNumTypeUnits() == 0) && |
1414 | "Was not able to retrieve Debug Types section." ); |
1415 | |
1416 | // There is no direct connection between CU and TU, but same offsets, |
1417 | // encoded in DW_AT_stmt_list, into .debug_line get modified. |
1418 | // We take advantage of that to map original CU line table offsets to new |
1419 | // ones. |
1420 | std::unordered_map<uint64_t, uint64_t> DebugLineOffsetMap; |
1421 | |
1422 | auto GetStatementListValue = |
1423 | [](const DWARFDie &DIE) -> std::optional<uint64_t> { |
1424 | std::optional<DWARFFormValue> StmtList = DIE.find(Attr: dwarf::DW_AT_stmt_list); |
1425 | if (!StmtList) |
1426 | return std::nullopt; |
1427 | std::optional<uint64_t> Offset = dwarf::toSectionOffset(V: StmtList); |
1428 | assert(Offset && "Was not able to retrieve value of DW_AT_stmt_list." ); |
1429 | return *Offset; |
1430 | }; |
1431 | |
1432 | SmallVector<DWARFUnit *, 1> TUs; |
1433 | for (const std::unique_ptr<DWARFUnit> &CU : BC.DwCtx->info_section_units()) { |
1434 | if (CU->isTypeUnit()) { |
1435 | TUs.push_back(Elt: CU.get()); |
1436 | continue; |
1437 | } |
1438 | const unsigned CUID = CU->getOffset(); |
1439 | MCSymbol *Label = BC.getDwarfLineTable(CUID).getLabel(); |
1440 | if (!Label) |
1441 | continue; |
1442 | |
1443 | std::optional<uint64_t> StmtOffset = |
1444 | GetStatementListValue(CU.get()->getUnitDIE()); |
1445 | if (!StmtOffset) |
1446 | continue; |
1447 | |
1448 | const uint64_t LineTableOffset = Layout.getSymbolOffset(S: *Label); |
1449 | DebugLineOffsetMap[*StmtOffset] = LineTableOffset; |
1450 | assert(DbgInfoSection && ".debug_info section must exist" ); |
1451 | LineTablePatchMap[CU.get()] = LineTableOffset; |
1452 | } |
1453 | |
1454 | for (const std::unique_ptr<DWARFUnit> &TU : BC.DwCtx->types_section_units()) |
1455 | TUs.push_back(Elt: TU.get()); |
1456 | |
1457 | for (DWARFUnit *TU : TUs) { |
1458 | std::optional<uint64_t> StmtOffset = |
1459 | GetStatementListValue(TU->getUnitDIE()); |
1460 | if (!StmtOffset) |
1461 | continue; |
1462 | auto Iter = DebugLineOffsetMap.find(x: *StmtOffset); |
1463 | if (Iter == DebugLineOffsetMap.end()) { |
1464 | // Implementation depends on TU sharing DW_AT_stmt_list with a CU. |
1465 | // Only case that it hasn't been true was for manually modified assembly |
1466 | // file. Adding this warning in case assumption is false. |
1467 | errs() |
1468 | << "BOLT-WARNING: [internal-dwarf-error]: A TU at offset: 0x" |
1469 | << Twine::utohexstr(Val: TU->getOffset()) |
1470 | << " is not sharing " |
1471 | ".debug_line entry with CU. DW_AT_stmt_list for this TU won't be " |
1472 | "updated.\n" ; |
1473 | continue; |
1474 | } |
1475 | TypeUnitRelocMap[TU] = Iter->second; |
1476 | } |
1477 | |
1478 | // Set .debug_info as finalized so it won't be skipped over when |
1479 | // we process sections while writing out the new binary. This ensures |
1480 | // that the pending relocations will be processed and not ignored. |
1481 | if (DbgInfoSection) |
1482 | DbgInfoSection->setIsFinalized(); |
1483 | |
1484 | if (TypeInfoSection) |
1485 | TypeInfoSection->setIsFinalized(); |
1486 | } |
1487 | |
1488 | CUOffsetMap DWARFRewriter::finalizeTypeSections(DIEBuilder &DIEBlder, |
1489 | DIEStreamer &Streamer) { |
1490 | // update TypeUnit DW_AT_stmt_list with new .debug_line information. |
1491 | auto updateLineTable = [&](const DWARFUnit &Unit) -> void { |
1492 | DIE *UnitDIE = DIEBlder.getUnitDIEbyUnit(DU: Unit); |
1493 | DIEValue StmtAttrInfo = UnitDIE->findAttribute(Attribute: dwarf::DW_AT_stmt_list); |
1494 | if (!StmtAttrInfo || !TypeUnitRelocMap.count(x: &Unit)) |
1495 | return; |
1496 | DIEBlder.replaceValue(Die: UnitDIE, Attribute: dwarf::DW_AT_stmt_list, |
1497 | Form: StmtAttrInfo.getForm(), |
1498 | NewValue: DIEInteger(TypeUnitRelocMap[&Unit])); |
1499 | }; |
1500 | |
1501 | // generate and populate abbrevs here |
1502 | DIEBlder.generateAbbrevs(); |
1503 | DIEBlder.finish(); |
1504 | SmallVector<char, 20> OutBuffer; |
1505 | std::shared_ptr<raw_svector_ostream> ObjOS = |
1506 | std::make_shared<raw_svector_ostream>(args&: OutBuffer); |
1507 | const object::ObjectFile *File = BC.DwCtx->getDWARFObj().getFile(); |
1508 | auto TheTriple = std::make_unique<Triple>(args: File->makeTriple()); |
1509 | std::unique_ptr<DIEStreamer> TypeStreamer = |
1510 | createDIEStreamer(TheTriple: *TheTriple, OutFile&: *ObjOS, Swift5ReflectionSegmentName: "TypeStreamer" , DIEBldr&: DIEBlder, Rewriter&: *this); |
1511 | |
1512 | // generate debug_info and CUMap |
1513 | CUOffsetMap CUMap; |
1514 | for (std::unique_ptr<llvm::DWARFUnit> &CU : BC.DwCtx->info_section_units()) { |
1515 | if (!CU->isTypeUnit()) |
1516 | continue; |
1517 | updateLineTable(*CU.get()); |
1518 | emitUnit(DIEBldr&: DIEBlder, Streamer, Unit&: *CU.get()); |
1519 | uint32_t StartOffset = CUOffset; |
1520 | DIE *UnitDIE = DIEBlder.getUnitDIEbyUnit(DU: *CU.get()); |
1521 | CUOffset += CU.get()->getHeaderSize(); |
1522 | CUOffset += UnitDIE->getSize(); |
1523 | CUMap[CU.get()->getOffset()] = {.Offset: StartOffset, .Length: CUOffset - StartOffset - 4}; |
1524 | } |
1525 | |
1526 | // Emit Type Unit of DWARF 4 to .debug_type section |
1527 | for (DWARFUnit *TU : DIEBlder.getDWARF4TUVector()) { |
1528 | updateLineTable(*TU); |
1529 | emitUnit(DIEBldr&: DIEBlder, Streamer&: *TypeStreamer, Unit&: *TU); |
1530 | } |
1531 | |
1532 | TypeStreamer->finish(); |
1533 | |
1534 | std::unique_ptr<MemoryBuffer> ObjectMemBuffer = |
1535 | MemoryBuffer::getMemBuffer(InputData: ObjOS->str(), BufferName: "in-memory object file" , RequiresNullTerminator: false); |
1536 | std::unique_ptr<object::ObjectFile> Obj = cantFail( |
1537 | ValOrErr: object::ObjectFile::createObjectFile(Object: ObjectMemBuffer->getMemBufferRef()), |
1538 | Msg: "error creating in-memory object" ); |
1539 | |
1540 | for (const SectionRef &Section : Obj->sections()) { |
1541 | StringRef Contents = cantFail(ValOrErr: Section.getContents()); |
1542 | StringRef Name = cantFail(ValOrErr: Section.getName()); |
1543 | if (Name.equals(RHS: ".debug_types" )) |
1544 | BC.registerOrUpdateNoteSection(Name: ".debug_types" , Data: copyByteArray(Buffer: Contents), |
1545 | Size: Contents.size()); |
1546 | } |
1547 | return CUMap; |
1548 | } |
1549 | |
1550 | void DWARFRewriter::finalizeDebugSections( |
1551 | DIEBuilder &DIEBlder, DWARF5AcceleratorTable &DebugNamesTable, |
1552 | DIEStreamer &Streamer, raw_svector_ostream &ObjOS, CUOffsetMap &CUMap) { |
1553 | if (StrWriter->isInitialized()) { |
1554 | RewriteInstance::addToDebugSectionsToOverwrite(Section: ".debug_str" ); |
1555 | std::unique_ptr<DebugStrBufferVector> DebugStrSectionContents = |
1556 | StrWriter->releaseBuffer(); |
1557 | BC.registerOrUpdateNoteSection(Name: ".debug_str" , |
1558 | Data: copyByteArray(Buffer: *DebugStrSectionContents), |
1559 | Size: DebugStrSectionContents->size()); |
1560 | } |
1561 | |
1562 | if (StrOffstsWriter->isFinalized()) { |
1563 | RewriteInstance::addToDebugSectionsToOverwrite(Section: ".debug_str_offsets" ); |
1564 | std::unique_ptr<DebugStrOffsetsBufferVector> |
1565 | DebugStrOffsetsSectionContents = StrOffstsWriter->releaseBuffer(); |
1566 | BC.registerOrUpdateNoteSection( |
1567 | Name: ".debug_str_offsets" , Data: copyByteArray(Buffer: *DebugStrOffsetsSectionContents), |
1568 | Size: DebugStrOffsetsSectionContents->size()); |
1569 | } |
1570 | |
1571 | if (BC.isDWARFLegacyUsed()) { |
1572 | std::unique_ptr<DebugBufferVector> RangesSectionContents = |
1573 | LegacyRangesSectionWriter->releaseBuffer(); |
1574 | BC.registerOrUpdateNoteSection(Name: ".debug_ranges" , |
1575 | Data: copyByteArray(Buffer: *RangesSectionContents), |
1576 | Size: RangesSectionContents->size()); |
1577 | } |
1578 | |
1579 | if (BC.isDWARF5Used()) { |
1580 | std::unique_ptr<DebugBufferVector> RangesSectionContents = |
1581 | RangeListsSectionWriter->releaseBuffer(); |
1582 | BC.registerOrUpdateNoteSection(Name: ".debug_rnglists" , |
1583 | Data: copyByteArray(Buffer: *RangesSectionContents), |
1584 | Size: RangesSectionContents->size()); |
1585 | } |
1586 | |
1587 | if (BC.isDWARF5Used()) { |
1588 | std::unique_ptr<DebugBufferVector> LocationListSectionContents = |
1589 | makeFinalLocListsSection(Version: DWARFVersion::DWARF5); |
1590 | if (!LocationListSectionContents->empty()) |
1591 | BC.registerOrUpdateNoteSection( |
1592 | Name: ".debug_loclists" , Data: copyByteArray(Buffer: *LocationListSectionContents), |
1593 | Size: LocationListSectionContents->size()); |
1594 | } |
1595 | |
1596 | if (BC.isDWARFLegacyUsed()) { |
1597 | std::unique_ptr<DebugBufferVector> LocationListSectionContents = |
1598 | makeFinalLocListsSection(Version: DWARFVersion::DWARFLegacy); |
1599 | if (!LocationListSectionContents->empty()) |
1600 | BC.registerOrUpdateNoteSection( |
1601 | Name: ".debug_loc" , Data: copyByteArray(Buffer: *LocationListSectionContents), |
1602 | Size: LocationListSectionContents->size()); |
1603 | } |
1604 | |
1605 | // AddrWriter should be finalized after debug_loc since more addresses can be |
1606 | // added there. |
1607 | if (AddrWriter->isInitialized()) { |
1608 | AddressSectionBuffer AddressSectionContents = AddrWriter->finalize(); |
1609 | BC.registerOrUpdateNoteSection(Name: ".debug_addr" , |
1610 | Data: copyByteArray(Buffer: AddressSectionContents), |
1611 | Size: AddressSectionContents.size()); |
1612 | } |
1613 | |
1614 | Streamer.emitAbbrevs(Abbrevs: DIEBlder.getAbbrevs(), DwarfVersion: BC.DwCtx->getMaxVersion()); |
1615 | Streamer.finish(); |
1616 | |
1617 | std::unique_ptr<MemoryBuffer> ObjectMemBuffer = |
1618 | MemoryBuffer::getMemBuffer(InputData: ObjOS.str(), BufferName: "in-memory object file" , RequiresNullTerminator: false); |
1619 | std::unique_ptr<object::ObjectFile> Obj = cantFail( |
1620 | ValOrErr: object::ObjectFile::createObjectFile(Object: ObjectMemBuffer->getMemBufferRef()), |
1621 | Msg: "error creating in-memory object" ); |
1622 | |
1623 | for (const SectionRef &Secs : Obj->sections()) { |
1624 | StringRef Contents = cantFail(ValOrErr: Secs.getContents()); |
1625 | StringRef Name = cantFail(ValOrErr: Secs.getName()); |
1626 | if (Name.equals(RHS: ".debug_abbrev" )) { |
1627 | BC.registerOrUpdateNoteSection(Name: ".debug_abbrev" , Data: copyByteArray(Buffer: Contents), |
1628 | Size: Contents.size()); |
1629 | } else if (Name.equals(RHS: ".debug_info" )) { |
1630 | BC.registerOrUpdateNoteSection(Name: ".debug_info" , Data: copyByteArray(Buffer: Contents), |
1631 | Size: Contents.size()); |
1632 | } |
1633 | } |
1634 | |
1635 | // Skip .debug_aranges if we are re-generating .gdb_index. |
1636 | if (opts::KeepARanges || !BC.getGdbIndexSection()) { |
1637 | SmallVector<char, 16> ARangesBuffer; |
1638 | raw_svector_ostream OS(ARangesBuffer); |
1639 | |
1640 | auto MAB = std::unique_ptr<MCAsmBackend>( |
1641 | BC.TheTarget->createMCAsmBackend(STI: *BC.STI, MRI: *BC.MRI, Options: MCTargetOptions())); |
1642 | |
1643 | ARangesSectionWriter->writeARangesSection(RangesStream&: OS, CUMap); |
1644 | const StringRef &ARangesContents = OS.str(); |
1645 | |
1646 | BC.registerOrUpdateNoteSection(Name: ".debug_aranges" , |
1647 | Data: copyByteArray(Buffer: ARangesContents), |
1648 | Size: ARangesContents.size()); |
1649 | } |
1650 | |
1651 | if (DebugNamesTable.isCreated()) { |
1652 | RewriteInstance::addToDebugSectionsToOverwrite(Section: ".debug_names" ); |
1653 | std::unique_ptr<DebugBufferVector> DebugNamesSectionContents = |
1654 | DebugNamesTable.releaseBuffer(); |
1655 | BC.registerOrUpdateNoteSection(Name: ".debug_names" , |
1656 | Data: copyByteArray(Buffer: *DebugNamesSectionContents), |
1657 | Size: DebugNamesSectionContents->size()); |
1658 | } |
1659 | } |
1660 | |
1661 | void DWARFRewriter::finalizeCompileUnits(DIEBuilder &DIEBlder, |
1662 | DIEStreamer &Streamer, |
1663 | CUOffsetMap &CUMap, |
1664 | const std::list<DWARFUnit *> &CUs) { |
1665 | DIEBlder.generateAbbrevs(); |
1666 | DIEBlder.finish(); |
1667 | // generate debug_info and CUMap |
1668 | for (DWARFUnit *CU : CUs) { |
1669 | emitUnit(DIEBldr&: DIEBlder, Streamer, Unit&: *CU); |
1670 | const uint32_t StartOffset = CUOffset; |
1671 | DIE *UnitDIE = DIEBlder.getUnitDIEbyUnit(DU: *CU); |
1672 | CUOffset += CU->getHeaderSize(); |
1673 | CUOffset += UnitDIE->getSize(); |
1674 | CUMap[CU->getOffset()] = {.Offset: StartOffset, .Length: CUOffset - StartOffset - 4}; |
1675 | } |
1676 | } |
1677 | |
1678 | // Creates all the data structures necessary for creating MCStreamer. |
1679 | // They are passed by reference because they need to be kept around. |
1680 | // Also creates known debug sections. These are sections handled by |
1681 | // handleDebugDataPatching. |
1682 | namespace { |
1683 | |
1684 | std::unique_ptr<BinaryContext> |
1685 | createDwarfOnlyBC(const object::ObjectFile &File) { |
1686 | return cantFail(ValOrErr: BinaryContext::createBinaryContext( |
1687 | TheTriple: File.makeTriple(), InputFileName: File.getFileName(), Features: nullptr, IsPIC: false, |
1688 | DwCtx: DWARFContext::create(Obj: File, RelocAction: DWARFContext::ProcessDebugRelocations::Ignore, |
1689 | L: nullptr, DWPName: "" , RecoverableErrorHandler: WithColor::defaultErrorHandler, |
1690 | WarningHandler: WithColor::defaultWarningHandler), |
1691 | Logger: {.Out: llvm::outs(), .Err: llvm::errs()})); |
1692 | } |
1693 | |
1694 | StringMap<DWARFRewriter::KnownSectionsEntry> |
1695 | createKnownSectionsMap(const MCObjectFileInfo &MCOFI) { |
1696 | StringMap<DWARFRewriter::KnownSectionsEntry> KnownSectionsTemp = { |
1697 | {"debug_info.dwo" , {MCOFI.getDwarfInfoDWOSection(), DW_SECT_INFO}}, |
1698 | {"debug_types.dwo" , {MCOFI.getDwarfTypesDWOSection(), DW_SECT_EXT_TYPES}}, |
1699 | {"debug_str_offsets.dwo" , |
1700 | {MCOFI.getDwarfStrOffDWOSection(), DW_SECT_STR_OFFSETS}}, |
1701 | {"debug_str.dwo" , {MCOFI.getDwarfStrDWOSection(), DW_SECT_EXT_unknown}}, |
1702 | {"debug_loc.dwo" , {MCOFI.getDwarfLocDWOSection(), DW_SECT_EXT_LOC}}, |
1703 | {"debug_abbrev.dwo" , {MCOFI.getDwarfAbbrevDWOSection(), DW_SECT_ABBREV}}, |
1704 | {"debug_line.dwo" , {MCOFI.getDwarfLineDWOSection(), DW_SECT_LINE}}, |
1705 | {"debug_loclists.dwo" , |
1706 | {MCOFI.getDwarfLoclistsDWOSection(), DW_SECT_LOCLISTS}}, |
1707 | {"debug_rnglists.dwo" , |
1708 | {MCOFI.getDwarfRnglistsDWOSection(), DW_SECT_RNGLISTS}}}; |
1709 | return KnownSectionsTemp; |
1710 | } |
1711 | |
1712 | StringRef getSectionName(const SectionRef &Section) { |
1713 | Expected<StringRef> SectionName = Section.getName(); |
1714 | assert(SectionName && "Invalid section name." ); |
1715 | StringRef Name = *SectionName; |
1716 | Name = Name.substr(Start: Name.find_first_not_of(Chars: "._" )); |
1717 | return Name; |
1718 | } |
1719 | |
1720 | // Exctracts an appropriate slice if input is DWP. |
1721 | // Applies patches or overwrites the section. |
1722 | std::optional<StringRef> updateDebugData( |
1723 | DWARFContext &DWCtx, StringRef SectionName, StringRef SectionContents, |
1724 | const StringMap<DWARFRewriter::KnownSectionsEntry> &KnownSections, |
1725 | MCStreamer &Streamer, DWARFRewriter &Writer, |
1726 | const DWARFUnitIndex::Entry *CUDWOEntry, uint64_t DWOId, |
1727 | std::unique_ptr<DebugBufferVector> &OutputBuffer, |
1728 | DebugRangeListsSectionWriter *RangeListsWriter, DebugLocWriter &LocWriter, |
1729 | const llvm::bolt::DWARFRewriter::OverriddenSectionsMap &OverridenSections) { |
1730 | |
1731 | using DWOSectionContribution = |
1732 | const DWARFUnitIndex::Entry::SectionContribution; |
1733 | auto getSliceData = [&](const DWARFUnitIndex::Entry *DWOEntry, |
1734 | StringRef OutData, DWARFSectionKind Sec, |
1735 | uint64_t &DWPOffset) -> StringRef { |
1736 | if (DWOEntry) { |
1737 | DWOSectionContribution *DWOContrubution = DWOEntry->getContribution(Sec); |
1738 | DWPOffset = DWOContrubution->getOffset(); |
1739 | OutData = OutData.substr(Start: DWPOffset, N: DWOContrubution->getLength()); |
1740 | } |
1741 | return OutData; |
1742 | }; |
1743 | |
1744 | auto SectionIter = KnownSections.find(Key: SectionName); |
1745 | if (SectionIter == KnownSections.end()) |
1746 | return std::nullopt; |
1747 | Streamer.switchSection(Section: SectionIter->second.first); |
1748 | uint64_t DWPOffset = 0; |
1749 | |
1750 | auto getOverridenSection = |
1751 | [&](DWARFSectionKind Kind) -> std::optional<StringRef> { |
1752 | auto Iter = OverridenSections.find(x: Kind); |
1753 | if (Iter == OverridenSections.end()) { |
1754 | errs() |
1755 | << "BOLT-WARNING: [internal-dwarf-error]: Could not find overriden " |
1756 | "section for: " |
1757 | << Twine::utohexstr(Val: DWOId) << ".\n" ; |
1758 | return std::nullopt; |
1759 | } |
1760 | return Iter->second; |
1761 | }; |
1762 | switch (SectionIter->second.second) { |
1763 | default: { |
1764 | if (!SectionName.equals(RHS: "debug_str.dwo" )) |
1765 | errs() << "BOLT-WARNING: unsupported debug section: " << SectionName |
1766 | << "\n" ; |
1767 | return SectionContents; |
1768 | } |
1769 | case DWARFSectionKind::DW_SECT_INFO: { |
1770 | return getOverridenSection(DWARFSectionKind::DW_SECT_INFO); |
1771 | } |
1772 | case DWARFSectionKind::DW_SECT_EXT_TYPES: { |
1773 | return getOverridenSection(DWARFSectionKind::DW_SECT_EXT_TYPES); |
1774 | } |
1775 | case DWARFSectionKind::DW_SECT_STR_OFFSETS: { |
1776 | return getSliceData(CUDWOEntry, SectionContents, |
1777 | DWARFSectionKind::DW_SECT_STR_OFFSETS, DWPOffset); |
1778 | } |
1779 | case DWARFSectionKind::DW_SECT_ABBREV: { |
1780 | return getOverridenSection(DWARFSectionKind::DW_SECT_ABBREV); |
1781 | } |
1782 | case DWARFSectionKind::DW_SECT_EXT_LOC: |
1783 | case DWARFSectionKind::DW_SECT_LOCLISTS: { |
1784 | OutputBuffer = LocWriter.getBuffer(); |
1785 | // Creating explicit StringRef here, otherwise |
1786 | // with implicit conversion it will take null byte as end of |
1787 | // string. |
1788 | return StringRef(reinterpret_cast<const char *>(OutputBuffer->data()), |
1789 | OutputBuffer->size()); |
1790 | } |
1791 | case DWARFSectionKind::DW_SECT_LINE: { |
1792 | return getSliceData(CUDWOEntry, SectionContents, |
1793 | DWARFSectionKind::DW_SECT_LINE, DWPOffset); |
1794 | } |
1795 | case DWARFSectionKind::DW_SECT_RNGLISTS: { |
1796 | assert(RangeListsWriter && "RangeListsWriter was not created." ); |
1797 | OutputBuffer = RangeListsWriter->releaseBuffer(); |
1798 | return StringRef(reinterpret_cast<const char *>(OutputBuffer->data()), |
1799 | OutputBuffer->size()); |
1800 | } |
1801 | } |
1802 | } |
1803 | |
1804 | } // namespace |
1805 | |
1806 | void DWARFRewriter::initDWPState(DWPState &State) { |
1807 | SmallString<0> OutputNameStr; |
1808 | StringRef OutputName; |
1809 | if (opts::DwarfOutputPath.empty()) { |
1810 | OutputName = |
1811 | Twine(opts::OutputFilename).concat(Suffix: ".dwp" ).toStringRef(Out&: OutputNameStr); |
1812 | } else { |
1813 | StringRef ExeFileName = llvm::sys::path::filename(path: opts::OutputFilename); |
1814 | OutputName = Twine(opts::DwarfOutputPath) |
1815 | .concat(Suffix: "/" ) |
1816 | .concat(Suffix: ExeFileName) |
1817 | .concat(Suffix: ".dwp" ) |
1818 | .toStringRef(Out&: OutputNameStr); |
1819 | errs() << "BOLT-WARNING: dwarf-output-path is in effect and .dwp file will " |
1820 | "possibly be written to another location that is not the same as " |
1821 | "the executable\n" ; |
1822 | } |
1823 | std::error_code EC; |
1824 | State.Out = |
1825 | std::make_unique<ToolOutputFile>(args&: OutputName, args&: EC, args: sys::fs::OF_None); |
1826 | const object::ObjectFile *File = BC.DwCtx->getDWARFObj().getFile(); |
1827 | State.TmpBC = createDwarfOnlyBC(File: *File); |
1828 | State.Streamer = State.TmpBC->createStreamer(OS&: State.Out->os()); |
1829 | State.MCOFI = State.Streamer->getContext().getObjectFileInfo(); |
1830 | State.KnownSections = createKnownSectionsMap(MCOFI: *State.MCOFI); |
1831 | MCSection *const StrSection = State.MCOFI->getDwarfStrDWOSection(); |
1832 | |
1833 | // Data Structures for DWP book keeping |
1834 | // Size of array corresponds to the number of sections supported by DWO format |
1835 | // in DWARF4/5. |
1836 | |
1837 | State.Strings = std::make_unique<DWPStringPool>(args&: *State.Streamer, args: StrSection); |
1838 | |
1839 | // Setup DWP code once. |
1840 | DWARFContext *DWOCtx = BC.getDWOContext(); |
1841 | |
1842 | if (DWOCtx) { |
1843 | State.CUIndex = &DWOCtx->getCUIndex(); |
1844 | State.IsDWP = !State.CUIndex->getRows().empty(); |
1845 | } |
1846 | } |
1847 | |
1848 | void DWARFRewriter::finalizeDWP(DWPState &State) { |
1849 | if (State.Version < 5) { |
1850 | // Lie about there being no info contributions so the TU index only includes |
1851 | // the type unit contribution for DWARF < 5. In DWARFv5 the TU index has a |
1852 | // contribution to the info section, so we do not want to lie about it. |
1853 | State.ContributionOffsets[0] = 0; |
1854 | } |
1855 | writeIndex(Out&: *State.Streamer.get(), Section: State.MCOFI->getDwarfTUIndexSection(), |
1856 | ContributionOffsets: State.ContributionOffsets, IndexEntries: State.TypeIndexEntries, |
1857 | IndexVersion: State.IndexVersion); |
1858 | |
1859 | if (State.Version < 5) { |
1860 | // Lie about the type contribution for DWARF < 5. In DWARFv5 the type |
1861 | // section does not exist, so no need to do anything about this. |
1862 | State.ContributionOffsets[getContributionIndex(Kind: DW_SECT_EXT_TYPES, IndexVersion: 2)] = 0; |
1863 | // Unlie about the info contribution |
1864 | State.ContributionOffsets[0] = 1; |
1865 | } |
1866 | writeIndex(Out&: *State.Streamer.get(), Section: State.MCOFI->getDwarfCUIndexSection(), |
1867 | ContributionOffsets: State.ContributionOffsets, IndexEntries: State.IndexEntries, IndexVersion: State.IndexVersion); |
1868 | |
1869 | State.Streamer->finish(); |
1870 | State.Out->keep(); |
1871 | } |
1872 | |
1873 | void DWARFRewriter::updateDWP(DWARFUnit &CU, |
1874 | const OverriddenSectionsMap &OverridenSections, |
1875 | const DWARFRewriter::UnitMeta &CUMI, |
1876 | DWARFRewriter::UnitMetaVectorType &TUMetaVector, |
1877 | DWPState &State, DebugLocWriter &LocWriter) { |
1878 | const uint64_t DWOId = *CU.getDWOId(); |
1879 | MCSection *const StrOffsetSection = State.MCOFI->getDwarfStrOffDWOSection(); |
1880 | assert(StrOffsetSection && "StrOffsetSection does not exist." ); |
1881 | // Skipping CUs that we failed to load. |
1882 | std::optional<DWARFUnit *> DWOCU = BC.getDWOCU(DWOId); |
1883 | if (!DWOCU) |
1884 | return; |
1885 | |
1886 | if (State.Version == 0) { |
1887 | State.Version = CU.getVersion(); |
1888 | State.IndexVersion = State.Version < 5 ? 2 : 5; |
1889 | } else if (State.Version != CU.getVersion()) { |
1890 | errs() << "BOLT-ERROR: incompatible DWARF compile unit versions\n" ; |
1891 | exit(status: 1); |
1892 | } |
1893 | |
1894 | UnitIndexEntry CurEntry = {}; |
1895 | CurEntry.DWOName = dwarf::toString( |
1896 | V: CU.getUnitDIE().find(Attrs: {dwarf::DW_AT_dwo_name, dwarf::DW_AT_GNU_dwo_name}), |
1897 | Default: "" ); |
1898 | const char *Name = CU.getUnitDIE().getShortName(); |
1899 | if (Name) |
1900 | CurEntry.Name = Name; |
1901 | StringRef CurStrSection; |
1902 | StringRef CurStrOffsetSection; |
1903 | |
1904 | // This maps each section contained in this file to its length. |
1905 | // This information is later on used to calculate the contributions, |
1906 | // i.e. offset and length, of each compile/type unit to a section. |
1907 | std::vector<std::pair<DWARFSectionKind, uint32_t>> SectionLength; |
1908 | |
1909 | const DWARFUnitIndex::Entry *CUDWOEntry = nullptr; |
1910 | if (State.IsDWP) |
1911 | CUDWOEntry = State.CUIndex->getFromHash(Offset: DWOId); |
1912 | |
1913 | bool StrSectionWrittenOut = false; |
1914 | const object::ObjectFile *DWOFile = |
1915 | (*DWOCU)->getContext().getDWARFObj().getFile(); |
1916 | |
1917 | DebugRangeListsSectionWriter *RangeListssWriter = nullptr; |
1918 | if (CU.getVersion() == 5) { |
1919 | assert(RangeListsWritersByCU.count(DWOId) != 0 && |
1920 | "No RangeListsWriter for DWO ID." ); |
1921 | RangeListssWriter = RangeListsWritersByCU[DWOId].get(); |
1922 | } |
1923 | auto AddType = [&](unsigned int Index, uint32_t IndexVersion, uint64_t Offset, |
1924 | uint64_t Length, uint64_t Hash) -> void { |
1925 | UnitIndexEntry TUEntry = CurEntry; |
1926 | if (IndexVersion < 5) |
1927 | TUEntry.Contributions[0] = {}; |
1928 | TUEntry.Contributions[Index].setOffset(Offset); |
1929 | TUEntry.Contributions[Index].setLength(Length); |
1930 | State.ContributionOffsets[Index] += |
1931 | TUEntry.Contributions[Index].getLength32(); |
1932 | State.TypeIndexEntries.insert(KV: std::make_pair(x&: Hash, y&: TUEntry)); |
1933 | }; |
1934 | for (const SectionRef &Section : DWOFile->sections()) { |
1935 | std::unique_ptr<DebugBufferVector> OutputData; |
1936 | StringRef SectionName = getSectionName(Section); |
1937 | Expected<StringRef> ContentsExp = Section.getContents(); |
1938 | assert(ContentsExp && "Invalid contents." ); |
1939 | std::optional<StringRef> TOutData = updateDebugData( |
1940 | DWCtx&: (*DWOCU)->getContext(), SectionName, SectionContents: *ContentsExp, KnownSections: State.KnownSections, |
1941 | Streamer&: *State.Streamer, Writer&: *this, CUDWOEntry, DWOId, OutputBuffer&: OutputData, |
1942 | RangeListsWriter: RangeListssWriter, LocWriter, OverridenSections); |
1943 | if (!TOutData) |
1944 | continue; |
1945 | |
1946 | StringRef OutData = *TOutData; |
1947 | if (SectionName == "debug_types.dwo" ) { |
1948 | State.Streamer->emitBytes(Data: OutData); |
1949 | continue; |
1950 | } |
1951 | |
1952 | if (SectionName.equals(RHS: "debug_str.dwo" )) { |
1953 | CurStrSection = OutData; |
1954 | } else { |
1955 | // Since handleDebugDataPatching returned true, we already know this is |
1956 | // a known section. |
1957 | auto SectionIter = State.KnownSections.find(Key: SectionName); |
1958 | if (SectionIter->second.second == DWARFSectionKind::DW_SECT_STR_OFFSETS) |
1959 | CurStrOffsetSection = OutData; |
1960 | else |
1961 | State.Streamer->emitBytes(Data: OutData); |
1962 | unsigned int Index = |
1963 | getContributionIndex(Kind: SectionIter->second.second, IndexVersion: State.IndexVersion); |
1964 | uint64_t Offset = State.ContributionOffsets[Index]; |
1965 | uint64_t Length = OutData.size(); |
1966 | if (CU.getVersion() >= 5 && |
1967 | SectionIter->second.second == DWARFSectionKind::DW_SECT_INFO) { |
1968 | for (UnitMeta &MI : TUMetaVector) |
1969 | MI.Offset += State.DebugInfoSize; |
1970 | |
1971 | Offset = State.DebugInfoSize + CUMI.Offset; |
1972 | Length = CUMI.Length; |
1973 | State.DebugInfoSize += OutData.size(); |
1974 | } |
1975 | CurEntry.Contributions[Index].setOffset(Offset); |
1976 | CurEntry.Contributions[Index].setLength(Length); |
1977 | State.ContributionOffsets[Index] += |
1978 | CurEntry.Contributions[Index].getLength32(); |
1979 | } |
1980 | |
1981 | // Strings are combined in to a new string section, and de-duplicated |
1982 | // based on hash. |
1983 | if (!StrSectionWrittenOut && !CurStrOffsetSection.empty() && |
1984 | !CurStrSection.empty()) { |
1985 | writeStringsAndOffsets(Out&: *State.Streamer.get(), Strings&: *State.Strings.get(), |
1986 | StrOffsetSection, CurStrSection, |
1987 | CurStrOffsetSection, Version: CU.getVersion()); |
1988 | StrSectionWrittenOut = true; |
1989 | } |
1990 | } |
1991 | CompileUnitIdentifiers CUI{.Signature: DWOId, .Name: CurEntry.Name.c_str(), |
1992 | .DWOName: CurEntry.DWOName.c_str()}; |
1993 | auto P = State.IndexEntries.insert(KV: std::make_pair(x&: CUI.Signature, y&: CurEntry)); |
1994 | if (!P.second) { |
1995 | Error Err = buildDuplicateError(PrevE: *P.first, ID: CUI, DWPName: "" ); |
1996 | errs() << "BOLT-ERROR: " << toString(E: std::move(Err)) << "\n" ; |
1997 | return; |
1998 | } |
1999 | |
2000 | // Handling TU |
2001 | const unsigned Index = getContributionIndex( |
2002 | Kind: State.IndexVersion < 5 ? DW_SECT_EXT_TYPES : DW_SECT_INFO, |
2003 | IndexVersion: State.IndexVersion); |
2004 | for (UnitMeta &MI : TUMetaVector) |
2005 | AddType(Index, State.IndexVersion, MI.Offset, MI.Length, MI.TUHash); |
2006 | } |
2007 | |
2008 | void DWARFRewriter::writeDWOFiles( |
2009 | DWARFUnit &CU, const OverriddenSectionsMap &OverridenSections, |
2010 | const std::string &DWOName, DebugLocWriter &LocWriter) { |
2011 | // Setup DWP code once. |
2012 | DWARFContext *DWOCtx = BC.getDWOContext(); |
2013 | const uint64_t DWOId = *CU.getDWOId(); |
2014 | const DWARFUnitIndex *CUIndex = nullptr; |
2015 | bool IsDWP = false; |
2016 | if (DWOCtx) { |
2017 | CUIndex = &DWOCtx->getCUIndex(); |
2018 | IsDWP = !CUIndex->getRows().empty(); |
2019 | } |
2020 | |
2021 | // Skipping CUs that we failed to load. |
2022 | std::optional<DWARFUnit *> DWOCU = BC.getDWOCU(DWOId); |
2023 | if (!DWOCU) { |
2024 | errs() << "BOLT-WARNING: [internal-dwarf-error]: CU for DWO_ID " |
2025 | << Twine::utohexstr(Val: DWOId) << " is not found.\n" ; |
2026 | return; |
2027 | } |
2028 | |
2029 | std::string CompDir = CU.getCompilationDir(); |
2030 | |
2031 | if (!opts::DwarfOutputPath.empty()) |
2032 | CompDir = opts::DwarfOutputPath.c_str(); |
2033 | else if (!opts::CompDirOverride.empty()) |
2034 | CompDir = opts::CompDirOverride; |
2035 | |
2036 | SmallString<16> AbsolutePath; |
2037 | sys::path::append(path&: AbsolutePath, a: CompDir); |
2038 | sys::path::append(path&: AbsolutePath, a: DWOName); |
2039 | |
2040 | std::error_code EC; |
2041 | std::unique_ptr<ToolOutputFile> TempOut = |
2042 | std::make_unique<ToolOutputFile>(args&: AbsolutePath, args&: EC, args: sys::fs::OF_None); |
2043 | |
2044 | const DWARFUnitIndex::Entry *CUDWOEntry = nullptr; |
2045 | if (IsDWP) |
2046 | CUDWOEntry = CUIndex->getFromHash(Offset: DWOId); |
2047 | |
2048 | const object::ObjectFile *File = |
2049 | (*DWOCU)->getContext().getDWARFObj().getFile(); |
2050 | std::unique_ptr<BinaryContext> TmpBC = createDwarfOnlyBC(File: *File); |
2051 | std::unique_ptr<MCStreamer> Streamer = TmpBC->createStreamer(OS&: TempOut->os()); |
2052 | const MCObjectFileInfo &MCOFI = *Streamer->getContext().getObjectFileInfo(); |
2053 | StringMap<KnownSectionsEntry> KnownSections = createKnownSectionsMap(MCOFI); |
2054 | |
2055 | DebugRangeListsSectionWriter *RangeListssWriter = nullptr; |
2056 | if (CU.getVersion() == 5) { |
2057 | assert(RangeListsWritersByCU.count(DWOId) != 0 && |
2058 | "No RangeListsWriter for DWO ID." ); |
2059 | RangeListssWriter = RangeListsWritersByCU[DWOId].get(); |
2060 | |
2061 | // Handling .debug_rnglists.dwo separately. The original .o/.dwo might not |
2062 | // have .debug_rnglists so won't be part of the loop below. |
2063 | if (!RangeListssWriter->empty()) { |
2064 | std::unique_ptr<DebugBufferVector> OutputData; |
2065 | if (std::optional<StringRef> OutData = updateDebugData( |
2066 | DWCtx&: (*DWOCU)->getContext(), SectionName: "debug_rnglists.dwo" , SectionContents: "" , KnownSections, |
2067 | Streamer&: *Streamer, Writer&: *this, CUDWOEntry, DWOId, OutputBuffer&: OutputData, |
2068 | RangeListsWriter: RangeListssWriter, LocWriter, OverridenSections)) |
2069 | Streamer->emitBytes(Data: *OutData); |
2070 | } |
2071 | } |
2072 | |
2073 | for (const SectionRef &Section : File->sections()) { |
2074 | std::unique_ptr<DebugBufferVector> OutputData; |
2075 | StringRef SectionName = getSectionName(Section); |
2076 | if (SectionName == "debug_rnglists.dwo" ) |
2077 | continue; |
2078 | Expected<StringRef> ContentsExp = Section.getContents(); |
2079 | assert(ContentsExp && "Invalid contents." ); |
2080 | if (std::optional<StringRef> OutData = updateDebugData( |
2081 | DWCtx&: (*DWOCU)->getContext(), SectionName, SectionContents: *ContentsExp, KnownSections, |
2082 | Streamer&: *Streamer, Writer&: *this, CUDWOEntry, DWOId, OutputBuffer&: OutputData, RangeListsWriter: RangeListssWriter, |
2083 | LocWriter, OverridenSections)) |
2084 | Streamer->emitBytes(Data: *OutData); |
2085 | } |
2086 | Streamer->finish(); |
2087 | TempOut->keep(); |
2088 | } |
2089 | |
2090 | void DWARFRewriter::addGDBTypeUnitEntry(const GDBIndexTUEntry &&Entry) { |
2091 | std::lock_guard<std::mutex> Lock(DWARFRewriterMutex); |
2092 | if (!BC.getGdbIndexSection()) |
2093 | return; |
2094 | GDBIndexTUEntryVector.emplace_back(args: Entry); |
2095 | } |
2096 | |
2097 | void DWARFRewriter::updateGdbIndexSection(CUOffsetMap &CUMap, uint32_t NumCUs) { |
2098 | if (!BC.getGdbIndexSection()) |
2099 | return; |
2100 | |
2101 | // See https://sourceware.org/gdb/onlinedocs/gdb/Index-Section-Format.html |
2102 | // for .gdb_index section format. |
2103 | |
2104 | StringRef GdbIndexContents = BC.getGdbIndexSection()->getContents(); |
2105 | |
2106 | const char *Data = GdbIndexContents.data(); |
2107 | |
2108 | // Parse the header. |
2109 | const uint32_t Version = read32le(P: Data); |
2110 | if (Version != 7 && Version != 8) { |
2111 | errs() << "BOLT-ERROR: can only process .gdb_index versions 7 and 8\n" ; |
2112 | exit(status: 1); |
2113 | } |
2114 | |
2115 | // Some .gdb_index generators use file offsets while others use section |
2116 | // offsets. Hence we can only rely on offsets relative to each other, |
2117 | // and ignore their absolute values. |
2118 | const uint32_t CUListOffset = read32le(P: Data + 4); |
2119 | const uint32_t CUTypesOffset = read32le(P: Data + 8); |
2120 | const uint32_t AddressTableOffset = read32le(P: Data + 12); |
2121 | const uint32_t SymbolTableOffset = read32le(P: Data + 16); |
2122 | const uint32_t ConstantPoolOffset = read32le(P: Data + 20); |
2123 | Data += 24; |
2124 | |
2125 | // Map CUs offsets to indices and verify existing index table. |
2126 | std::map<uint32_t, uint32_t> OffsetToIndexMap; |
2127 | const uint32_t CUListSize = CUTypesOffset - CUListOffset; |
2128 | const uint32_t TUListSize = AddressTableOffset - CUTypesOffset; |
2129 | const unsigned NUmCUsEncoded = CUListSize / 16; |
2130 | unsigned MaxDWARFVersion = BC.DwCtx->getMaxVersion(); |
2131 | unsigned NumDWARF5TUs = |
2132 | getGDBIndexTUEntryVector().size() - BC.DwCtx->getNumTypeUnits(); |
2133 | bool SkipTypeUnits = false; |
2134 | // For DWARF5 Types are in .debug_info. |
2135 | // LLD doesn't generate Types CU List, and in CU list offset |
2136 | // only includes CUs. |
2137 | // GDB 11+ includes only CUs in CU list and generates Types |
2138 | // list. |
2139 | // GDB 9 includes CUs and TUs in CU list and generates TYpes |
2140 | // list. The NumCUs is CUs + TUs, so need to modify the check. |
2141 | // For split-dwarf |
2142 | // GDB-11, DWARF5: TU units from dwo are not included. |
2143 | // GDB-11, DWARF4: TU units from dwo are included. |
2144 | if (MaxDWARFVersion >= 5) |
2145 | SkipTypeUnits = !TUListSize ? true |
2146 | : ((NUmCUsEncoded + NumDWARF5TUs) == |
2147 | BC.DwCtx->getNumCompileUnits()); |
2148 | |
2149 | if (!((CUListSize == NumCUs * 16) || |
2150 | (CUListSize == (NumCUs + NumDWARF5TUs) * 16))) { |
2151 | errs() << "BOLT-ERROR: .gdb_index: CU count mismatch\n" ; |
2152 | exit(status: 1); |
2153 | } |
2154 | DenseSet<uint64_t> OriginalOffsets; |
2155 | for (unsigned Index = 0, Units = BC.DwCtx->getNumCompileUnits(); |
2156 | Index < Units; ++Index) { |
2157 | const DWARFUnit *CU = BC.DwCtx->getUnitAtIndex(index: Index); |
2158 | if (SkipTypeUnits && CU->isTypeUnit()) |
2159 | continue; |
2160 | const uint64_t Offset = read64le(P: Data); |
2161 | Data += 16; |
2162 | if (CU->getOffset() != Offset) { |
2163 | errs() << "BOLT-ERROR: .gdb_index CU offset mismatch\n" ; |
2164 | exit(status: 1); |
2165 | } |
2166 | |
2167 | OriginalOffsets.insert(V: Offset); |
2168 | OffsetToIndexMap[Offset] = Index; |
2169 | } |
2170 | |
2171 | // Ignore old address table. |
2172 | const uint32_t OldAddressTableSize = SymbolTableOffset - AddressTableOffset; |
2173 | // Move Data to the beginning of symbol table. |
2174 | Data += SymbolTableOffset - CUTypesOffset; |
2175 | |
2176 | // Calculate the size of the new address table. |
2177 | uint32_t NewAddressTableSize = 0; |
2178 | for (const auto &CURangesPair : ARangesSectionWriter->getCUAddressRanges()) { |
2179 | const SmallVector<DebugAddressRange, 2> &Ranges = CURangesPair.second; |
2180 | NewAddressTableSize += Ranges.size() * 20; |
2181 | } |
2182 | |
2183 | // Difference between old and new table (and section) sizes. |
2184 | // Could be negative. |
2185 | int32_t Delta = NewAddressTableSize - OldAddressTableSize; |
2186 | |
2187 | size_t NewGdbIndexSize = GdbIndexContents.size() + Delta; |
2188 | |
2189 | // Free'd by ExecutableFileMemoryManager. |
2190 | auto *NewGdbIndexContents = new uint8_t[NewGdbIndexSize]; |
2191 | uint8_t *Buffer = NewGdbIndexContents; |
2192 | |
2193 | write32le(P: Buffer, V: Version); |
2194 | write32le(P: Buffer + 4, V: CUListOffset); |
2195 | write32le(P: Buffer + 8, V: CUTypesOffset); |
2196 | write32le(P: Buffer + 12, V: AddressTableOffset); |
2197 | write32le(P: Buffer + 16, V: SymbolTableOffset + Delta); |
2198 | write32le(P: Buffer + 20, V: ConstantPoolOffset + Delta); |
2199 | Buffer += 24; |
2200 | |
2201 | using MapEntry = std::pair<uint32_t, CUInfo>; |
2202 | std::vector<MapEntry> CUVector(CUMap.begin(), CUMap.end()); |
2203 | // Need to sort since we write out all of TUs in .debug_info before CUs. |
2204 | std::sort(first: CUVector.begin(), last: CUVector.end(), |
2205 | comp: [](const MapEntry &E1, const MapEntry &E2) -> bool { |
2206 | return E1.second.Offset < E2.second.Offset; |
2207 | }); |
2208 | // Writing out CU List <Offset, Size> |
2209 | for (auto &CUInfo : CUVector) { |
2210 | // Skipping TU for DWARF5 when they are not included in CU list. |
2211 | if (!OriginalOffsets.count(V: CUInfo.first)) |
2212 | continue; |
2213 | write64le(P: Buffer, V: CUInfo.second.Offset); |
2214 | // Length encoded in CU doesn't contain first 4 bytes that encode length. |
2215 | write64le(P: Buffer + 8, V: CUInfo.second.Length + 4); |
2216 | Buffer += 16; |
2217 | } |
2218 | |
2219 | // Rewrite TU CU List, since abbrevs can be different. |
2220 | // Entry example: |
2221 | // 0: offset = 0x00000000, type_offset = 0x0000001e, type_signature = |
2222 | // 0x418503b8111e9a7b Spec says " triplet, the first value is the CU offset, |
2223 | // the second value is the type offset in the CU, and the third value is the |
2224 | // type signature" Looking at what is being generated by gdb-add-index. The |
2225 | // first entry is TU offset, second entry is offset from it, and third entry |
2226 | // is the type signature. |
2227 | if (TUListSize) |
2228 | for (const GDBIndexTUEntry &Entry : getGDBIndexTUEntryVector()) { |
2229 | write64le(P: Buffer, V: Entry.UnitOffset); |
2230 | write64le(P: Buffer + 8, V: Entry.TypeDIERelativeOffset); |
2231 | write64le(P: Buffer + 16, V: Entry.TypeHash); |
2232 | Buffer += sizeof(GDBIndexTUEntry); |
2233 | } |
2234 | |
2235 | // Generate new address table. |
2236 | for (const std::pair<const uint64_t, DebugAddressRangesVector> &CURangesPair : |
2237 | ARangesSectionWriter->getCUAddressRanges()) { |
2238 | const uint32_t CUIndex = OffsetToIndexMap[CURangesPair.first]; |
2239 | const DebugAddressRangesVector &Ranges = CURangesPair.second; |
2240 | for (const DebugAddressRange &Range : Ranges) { |
2241 | write64le(P: Buffer, V: Range.LowPC); |
2242 | write64le(P: Buffer + 8, V: Range.HighPC); |
2243 | write32le(P: Buffer + 16, V: CUIndex); |
2244 | Buffer += 20; |
2245 | } |
2246 | } |
2247 | |
2248 | const size_t TrailingSize = |
2249 | GdbIndexContents.data() + GdbIndexContents.size() - Data; |
2250 | assert(Buffer + TrailingSize == NewGdbIndexContents + NewGdbIndexSize && |
2251 | "size calculation error" ); |
2252 | |
2253 | // Copy over the rest of the original data. |
2254 | memcpy(dest: Buffer, src: Data, n: TrailingSize); |
2255 | |
2256 | // Register the new section. |
2257 | BC.registerOrUpdateNoteSection(Name: ".gdb_index" , Data: NewGdbIndexContents, |
2258 | Size: NewGdbIndexSize); |
2259 | } |
2260 | |
2261 | std::unique_ptr<DebugBufferVector> |
2262 | DWARFRewriter::makeFinalLocListsSection(DWARFVersion Version) { |
2263 | auto LocBuffer = std::make_unique<DebugBufferVector>(); |
2264 | auto LocStream = std::make_unique<raw_svector_ostream>(args&: *LocBuffer); |
2265 | auto Writer = |
2266 | std::unique_ptr<MCObjectWriter>(BC.createObjectWriter(OS&: *LocStream)); |
2267 | |
2268 | for (std::pair<const uint64_t, std::unique_ptr<DebugLocWriter>> &Loc : |
2269 | LocListWritersByCU) { |
2270 | DebugLocWriter *LocWriter = Loc.second.get(); |
2271 | auto *LocListWriter = llvm::dyn_cast<DebugLoclistWriter>(Val: LocWriter); |
2272 | |
2273 | // Filter out DWARF4, writing out DWARF5 |
2274 | if (Version == DWARFVersion::DWARF5 && |
2275 | (!LocListWriter || LocListWriter->getDwarfVersion() <= 4)) |
2276 | continue; |
2277 | |
2278 | // Filter out DWARF5, writing out DWARF4 |
2279 | if (Version == DWARFVersion::DWARFLegacy && |
2280 | (LocListWriter && LocListWriter->getDwarfVersion() >= 5)) |
2281 | continue; |
2282 | |
2283 | // Skipping DWARF4/5 split dwarf. |
2284 | if (LocListWriter && LocListWriter->getDwarfVersion() <= 4) |
2285 | continue; |
2286 | std::unique_ptr<DebugBufferVector> CurrCULocationLists = |
2287 | LocWriter->getBuffer(); |
2288 | *LocStream << *CurrCULocationLists; |
2289 | } |
2290 | |
2291 | return LocBuffer; |
2292 | } |
2293 | |
2294 | void DWARFRewriter::convertToRangesPatchDebugInfo( |
2295 | DWARFUnit &Unit, DIEBuilder &DIEBldr, DIE &Die, |
2296 | uint64_t RangesSectionOffset, DIEValue &LowPCAttrInfo, |
2297 | DIEValue &HighPCAttrInfo, std::optional<uint64_t> RangesBase) { |
2298 | uint32_t BaseOffset = 0; |
2299 | dwarf::Form LowForm = LowPCAttrInfo.getForm(); |
2300 | dwarf::Attribute RangeBaseAttribute = dwarf::DW_AT_GNU_ranges_base; |
2301 | dwarf::Form RangesForm = dwarf::DW_FORM_sec_offset; |
2302 | |
2303 | if (Unit.getVersion() >= 5) { |
2304 | RangeBaseAttribute = dwarf::DW_AT_rnglists_base; |
2305 | RangesForm = dwarf::DW_FORM_rnglistx; |
2306 | } else if (Unit.getVersion() < 4) { |
2307 | RangesForm = dwarf::DW_FORM_data4; |
2308 | } |
2309 | bool IsUnitDie = Die.getTag() == dwarf::DW_TAG_compile_unit || |
2310 | Die.getTag() == dwarf::DW_TAG_skeleton_unit; |
2311 | if (!IsUnitDie) |
2312 | DIEBldr.deleteValue(Die: &Die, Attribute: LowPCAttrInfo.getAttribute()); |
2313 | // In DWARF4 for DW_AT_low_pc in binary DW_FORM_addr is used. In the DWO |
2314 | // section DW_FORM_GNU_addr_index is used. So for if we are converting |
2315 | // DW_AT_low_pc/DW_AT_high_pc and see DW_FORM_GNU_addr_index. We are |
2316 | // converting in DWO section, and DW_AT_ranges [DW_FORM_sec_offset] is |
2317 | // relative to DW_AT_GNU_ranges_base. |
2318 | if (LowForm == dwarf::DW_FORM_GNU_addr_index) { |
2319 | // Ranges are relative to DW_AT_GNU_ranges_base. |
2320 | uint64_t CurRangeBase = 0; |
2321 | if (std::optional<uint64_t> DWOId = Unit.getDWOId()) { |
2322 | CurRangeBase = getDwoRangesBase(DWOId: *DWOId); |
2323 | } |
2324 | BaseOffset = CurRangeBase; |
2325 | } else { |
2326 | // In DWARF 5 we can have DW_AT_low_pc either as DW_FORM_addr, or |
2327 | // DW_FORM_addrx. Former is when DW_AT_rnglists_base is present. Latter is |
2328 | // when it's absent. |
2329 | if (IsUnitDie) { |
2330 | if (LowForm == dwarf::DW_FORM_addrx) { |
2331 | const uint32_t Index = AddrWriter->getIndexFromAddress(Address: 0, CU&: Unit); |
2332 | DIEBldr.replaceValue(Die: &Die, Attribute: LowPCAttrInfo.getAttribute(), |
2333 | Form: LowPCAttrInfo.getForm(), NewValue: DIEInteger(Index)); |
2334 | } else { |
2335 | DIEBldr.replaceValue(Die: &Die, Attribute: LowPCAttrInfo.getAttribute(), |
2336 | Form: LowPCAttrInfo.getForm(), NewValue: DIEInteger(0)); |
2337 | } |
2338 | } |
2339 | // Original CU didn't have DW_AT_*_base. We converted it's children (or |
2340 | // dwo), so need to insert it into CU. |
2341 | if (RangesBase) |
2342 | DIEBldr.addValue(Die: &Die, Attribute: RangeBaseAttribute, Form: dwarf::DW_FORM_sec_offset, |
2343 | Value: DIEInteger(*RangesBase)); |
2344 | } |
2345 | |
2346 | uint64_t RangeAttrVal = RangesSectionOffset - BaseOffset; |
2347 | if (Unit.getVersion() >= 5) |
2348 | RangeAttrVal = RangesSectionOffset; |
2349 | // HighPC was conveted into DW_AT_ranges. |
2350 | // For DWARF5 we only access ranges through index. |
2351 | |
2352 | DIEBldr.replaceValue(Die: &Die, Attribute: HighPCAttrInfo.getAttribute(), NewAttribute: dwarf::DW_AT_ranges, |
2353 | Form: RangesForm, NewValue: DIEInteger(RangeAttrVal)); |
2354 | } |
2355 | |