1 //===- DWARFUnit.h ----------------------------------------------*- C++ -*-===//
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
7 //===----------------------------------------------------------------------===//
9 #ifndef LLVM_DEBUGINFO_DWARF_DWARFUNIT_H
10 #define LLVM_DEBUGINFO_DWARF_DWARFUNIT_H
12 #include "llvm/ADT/Optional.h"
13 #include "llvm/ADT/STLExtras.h"
14 #include "llvm/ADT/SmallVector.h"
15 #include "llvm/ADT/StringRef.h"
16 #include "llvm/ADT/iterator_range.h"
17 #include "llvm/BinaryFormat/Dwarf.h"
18 #include "llvm/DebugInfo/DWARF/DWARFDebugInfoEntry.h"
19 #include "llvm/DebugInfo/DWARF/DWARFDebugRangeList.h"
20 #include "llvm/DebugInfo/DWARF/DWARFDebugRnglists.h"
21 #include "llvm/DebugInfo/DWARF/DWARFDie.h"
22 #include "llvm/DebugInfo/DWARF/DWARFFormValue.h"
23 #include "llvm/DebugInfo/DWARF/DWARFRelocMap.h"
24 #include "llvm/DebugInfo/DWARF/DWARFSection.h"
25 #include "llvm/DebugInfo/DWARF/DWARFUnitIndex.h"
26 #include "llvm/Support/DataExtractor.h"
38 class DWARFAbbreviationDeclarationSet;
40 class DWARFDebugAbbrev;
43 /// Base class describing the header of any kind of "unit." Some information
44 /// is specific to certain unit types. We separate this class out so we can
45 /// parse the header before deciding what specific kind of unit to construct.
46 class DWARFUnitHeader {
47 // Offset within section.
49 // Version, address size, and DWARF format.
50 dwarf::FormParams FormParams;
52 uint64_t AbbrOffset = 0;
54 // For DWO units only.
55 const DWARFUnitIndex::Entry *IndexEntry = nullptr;
57 // For type units only.
58 uint64_t TypeHash = 0;
59 uint32_t TypeOffset = 0;
61 // For v5 split or skeleton compile units only.
62 Optional<uint64_t> DWOId;
64 // Unit type as parsed, or derived from the section kind.
67 // Size as parsed. uint8_t for compactness.
71 /// Parse a unit header from \p debug_info starting at \p offset_ptr.
72 bool extract(DWARFContext &Context, const DWARFDataExtractor &debug_info,
73 uint32_t *offset_ptr, DWARFSectionKind Kind = DW_SECT_INFO,
74 const DWARFUnitIndex *Index = nullptr,
75 const DWARFUnitIndex::Entry *Entry = nullptr);
76 uint32_t getOffset() const { return Offset; }
77 const dwarf::FormParams &getFormParams() const { return FormParams; }
78 uint16_t getVersion() const { return FormParams.Version; }
79 dwarf::DwarfFormat getFormat() const { return FormParams.Format; }
80 uint8_t getAddressByteSize() const { return FormParams.AddrSize; }
81 uint8_t getRefAddrByteSize() const { return FormParams.getRefAddrByteSize(); }
82 uint8_t getDwarfOffsetByteSize() const {
83 return FormParams.getDwarfOffsetByteSize();
85 uint64_t getLength() const { return Length; }
86 uint64_t getAbbrOffset() const { return AbbrOffset; }
87 Optional<uint64_t> getDWOId() const { return DWOId; }
88 void setDWOId(uint64_t Id) {
89 assert((!DWOId || *DWOId == Id) && "setting DWOId to a different value");
92 const DWARFUnitIndex::Entry *getIndexEntry() const { return IndexEntry; }
93 uint64_t getTypeHash() const { return TypeHash; }
94 uint32_t getTypeOffset() const { return TypeOffset; }
95 uint8_t getUnitType() const { return UnitType; }
96 bool isTypeUnit() const {
97 return UnitType == dwarf::DW_UT_type || UnitType == dwarf::DW_UT_split_type;
99 uint8_t getSize() const { return Size; }
100 uint32_t getNextUnitOffset() const {
101 return Offset + Length +
102 (FormParams.Format == llvm::dwarf::DwarfFormat::DWARF64 ? 4 : 0) +
103 FormParams.getDwarfOffsetByteSize();
107 const DWARFUnitIndex &getDWARFUnitIndex(DWARFContext &Context,
108 DWARFSectionKind Kind);
110 /// Describe a collection of units. Intended to hold all units either from
111 /// .debug_info and .debug_types, or from .debug_info.dwo and .debug_types.dwo.
112 class DWARFUnitVector final : public SmallVector<std::unique_ptr<DWARFUnit>, 1> {
113 std::function<std::unique_ptr<DWARFUnit>(uint32_t, DWARFSectionKind,
114 const DWARFSection *,
115 const DWARFUnitIndex::Entry *)>
117 int NumInfoUnits = -1;
120 using UnitVector = SmallVectorImpl<std::unique_ptr<DWARFUnit>>;
121 using iterator = typename UnitVector::iterator;
122 using iterator_range = llvm::iterator_range<typename UnitVector::iterator>;
124 DWARFUnit *getUnitForOffset(uint32_t Offset) const;
125 DWARFUnit *getUnitForIndexEntry(const DWARFUnitIndex::Entry &E);
127 /// Read units from a .debug_info or .debug_types section. Calls made
128 /// before finishedInfoUnits() are assumed to be for .debug_info sections,
129 /// calls after finishedInfoUnits() are for .debug_types sections. Caller
130 /// must not mix calls to addUnitsForSection and addUnitsForDWOSection.
131 void addUnitsForSection(DWARFContext &C, const DWARFSection &Section,
132 DWARFSectionKind SectionKind);
133 /// Read units from a .debug_info.dwo or .debug_types.dwo section. Calls
134 /// made before finishedInfoUnits() are assumed to be for .debug_info.dwo
135 /// sections, calls after finishedInfoUnits() are for .debug_types.dwo
136 /// sections. Caller must not mix calls to addUnitsForSection and
137 /// addUnitsForDWOSection.
138 void addUnitsForDWOSection(DWARFContext &C, const DWARFSection &DWOSection,
139 DWARFSectionKind SectionKind, bool Lazy = false);
141 /// Add an existing DWARFUnit to this UnitVector. This is used by the DWARF
142 /// verifier to process unit separately.
143 DWARFUnit *addUnit(std::unique_ptr<DWARFUnit> Unit);
145 /// Returns number of all units held by this instance.
146 unsigned getNumUnits() const { return size(); }
147 /// Returns number of units from all .debug_info[.dwo] sections.
148 unsigned getNumInfoUnits() const {
149 return NumInfoUnits == -1 ? size() : NumInfoUnits;
151 /// Returns number of units from all .debug_types[.dwo] sections.
152 unsigned getNumTypesUnits() const { return size() - NumInfoUnits; }
153 /// Indicate that parsing .debug_info[.dwo] is done, and remaining units
154 /// will be from .debug_types[.dwo].
155 void finishedInfoUnits() { NumInfoUnits = size(); }
158 void addUnitsImpl(DWARFContext &Context, const DWARFObject &Obj,
159 const DWARFSection &Section, const DWARFDebugAbbrev *DA,
160 const DWARFSection *RS, const DWARFSection *LocSection,
161 StringRef SS, const DWARFSection &SOS,
162 const DWARFSection *AOS, const DWARFSection &LS, bool LE,
163 bool IsDWO, bool Lazy, DWARFSectionKind SectionKind);
166 /// Represents base address of the CU.
167 /// Represents a unit's contribution to the string offsets table.
168 struct StrOffsetsContributionDescriptor {
170 /// The contribution size not including the header.
172 /// Format and version.
173 dwarf::FormParams FormParams = {0, 0, dwarf::DwarfFormat::DWARF32};
175 StrOffsetsContributionDescriptor(uint64_t Base, uint64_t Size,
176 uint8_t Version, dwarf::DwarfFormat Format)
177 : Base(Base), Size(Size), FormParams({Version, 0, Format}) {}
178 StrOffsetsContributionDescriptor() = default;
180 uint8_t getVersion() const { return FormParams.Version; }
181 dwarf::DwarfFormat getFormat() const { return FormParams.Format; }
182 uint8_t getDwarfOffsetByteSize() const {
183 return FormParams.getDwarfOffsetByteSize();
185 /// Determine whether a contribution to the string offsets table is
186 /// consistent with the relevant section size and that its length is
187 /// a multiple of the size of one of its entries.
188 Expected<StrOffsetsContributionDescriptor>
189 validateContributionSize(DWARFDataExtractor &DA);
193 DWARFContext &Context;
194 /// Section containing this DWARFUnit.
195 const DWARFSection &InfoSection;
197 DWARFUnitHeader Header;
198 const DWARFDebugAbbrev *Abbrev;
199 const DWARFSection *RangeSection;
200 uint32_t RangeSectionBase;
201 /// We either keep track of the location list section or its data, depending
202 /// on whether we are handling a split DWARF section or not.
204 const DWARFSection *LocSection;
205 StringRef LocSectionData;
207 const DWARFSection &LineSection;
208 StringRef StringSection;
209 const DWARFSection &StringOffsetSection;
210 const DWARFSection *AddrOffsetSection;
211 uint32_t AddrOffsetSectionBase = 0;
214 const DWARFUnitVector &UnitVector;
216 /// Start, length, and DWARF format of the unit's contribution to the string
217 /// offsets table (DWARF v5).
218 Optional<StrOffsetsContributionDescriptor> StringOffsetsTableContribution;
220 /// A table of range lists (DWARF v5 and later).
221 Optional<DWARFDebugRnglistTable> RngListTable;
223 mutable const DWARFAbbreviationDeclarationSet *Abbrevs;
224 llvm::Optional<object::SectionedAddress> BaseAddr;
225 /// The compile unit debug information entry items.
226 std::vector<DWARFDebugInfoEntry> DieArray;
228 /// Map from range's start address to end address and corresponding DIE.
229 /// IntervalMap does not support range removal, as a result, we use the
230 /// std::map::upper_bound for address range lookup.
231 std::map<uint64_t, std::pair<uint64_t, DWARFDie>> AddrDieMap;
233 using die_iterator_range =
234 iterator_range<std::vector<DWARFDebugInfoEntry>::iterator>;
236 std::shared_ptr<DWARFUnit> DWO;
238 uint32_t getDIEIndex(const DWARFDebugInfoEntry *Die) {
239 auto First = DieArray.data();
240 assert(Die >= First && Die < First + DieArray.size());
245 const DWARFUnitHeader &getHeader() const { return Header; }
247 /// Size in bytes of the parsed unit header.
248 uint32_t getHeaderSize() const { return Header.getSize(); }
250 /// Find the unit's contribution to the string offsets table and determine its
251 /// length and form. The given offset is expected to be derived from the unit
252 /// DIE's DW_AT_str_offsets_base attribute.
253 Expected<Optional<StrOffsetsContributionDescriptor>>
254 determineStringOffsetsTableContribution(DWARFDataExtractor &DA);
256 /// Find the unit's contribution to the string offsets table and determine its
257 /// length and form. The given offset is expected to be 0 in a dwo file or,
258 /// in a dwp file, the start of the unit's contribution to the string offsets
259 /// table section (as determined by the index table).
260 Expected<Optional<StrOffsetsContributionDescriptor>>
261 determineStringOffsetsTableContributionDWO(DWARFDataExtractor &DA);
264 DWARFUnit(DWARFContext &Context, const DWARFSection &Section,
265 const DWARFUnitHeader &Header, const DWARFDebugAbbrev *DA,
266 const DWARFSection *RS, const DWARFSection *LocSection,
267 StringRef SS, const DWARFSection &SOS, const DWARFSection *AOS,
268 const DWARFSection &LS, bool LE, bool IsDWO,
269 const DWARFUnitVector &UnitVector);
271 virtual ~DWARFUnit();
273 bool isDWOUnit() const { return IsDWO; }
274 DWARFContext& getContext() const { return Context; }
275 const DWARFSection &getInfoSection() const { return InfoSection; }
276 const DWARFSection *getLocSection() const { return LocSection; }
277 StringRef getLocSectionData() const { return LocSectionData; }
278 uint32_t getOffset() const { return Header.getOffset(); }
279 const dwarf::FormParams &getFormParams() const {
280 return Header.getFormParams();
282 uint16_t getVersion() const { return Header.getVersion(); }
283 uint8_t getAddressByteSize() const { return Header.getAddressByteSize(); }
284 uint8_t getRefAddrByteSize() const { return Header.getRefAddrByteSize(); }
285 uint8_t getDwarfOffsetByteSize() const {
286 return Header.getDwarfOffsetByteSize();
288 uint32_t getLength() const { return Header.getLength(); }
289 uint8_t getUnitType() const { return Header.getUnitType(); }
290 bool isTypeUnit() const { return Header.isTypeUnit(); }
291 uint32_t getNextUnitOffset() const { return Header.getNextUnitOffset(); }
292 const DWARFSection &getLineSection() const { return LineSection; }
293 StringRef getStringSection() const { return StringSection; }
294 const DWARFSection &getStringOffsetSection() const {
295 return StringOffsetSection;
298 void setAddrOffsetSection(const DWARFSection *AOS, uint32_t Base) {
299 AddrOffsetSection = AOS;
300 AddrOffsetSectionBase = Base;
303 /// Recursively update address to Die map.
304 void updateAddressDieMap(DWARFDie Die);
306 void setRangesSection(const DWARFSection *RS, uint32_t Base) {
308 RangeSectionBase = Base;
311 Optional<object::SectionedAddress>
312 getAddrOffsetSectionItem(uint32_t Index) const;
313 Optional<uint64_t> getStringOffsetSectionItem(uint32_t Index) const;
315 DWARFDataExtractor getDebugInfoExtractor() const;
317 DataExtractor getStringExtractor() const {
318 return DataExtractor(StringSection, false, 0);
321 /// Extract the range list referenced by this compile unit from the
322 /// .debug_ranges section. If the extraction is unsuccessful, an error
323 /// is returned. Successful extraction requires that the compile unit
324 /// has already been extracted.
325 Error extractRangeList(uint32_t RangeListOffset,
326 DWARFDebugRangeList &RangeList) const;
329 const Optional<StrOffsetsContributionDescriptor> &
330 getStringOffsetsTableContribution() const {
331 return StringOffsetsTableContribution;
334 uint8_t getDwarfStringOffsetsByteSize() const {
335 assert(StringOffsetsTableContribution);
336 return StringOffsetsTableContribution->getDwarfOffsetByteSize();
339 uint64_t getStringOffsetsBase() const {
340 assert(StringOffsetsTableContribution);
341 return StringOffsetsTableContribution->Base;
344 const DWARFAbbreviationDeclarationSet *getAbbreviations() const;
346 static bool isMatchingUnitTypeAndTag(uint8_t UnitType, dwarf::Tag Tag) {
348 case dwarf::DW_UT_compile:
349 return Tag == dwarf::DW_TAG_compile_unit;
350 case dwarf::DW_UT_type:
351 return Tag == dwarf::DW_TAG_type_unit;
352 case dwarf::DW_UT_partial:
353 return Tag == dwarf::DW_TAG_partial_unit;
354 case dwarf::DW_UT_skeleton:
355 return Tag == dwarf::DW_TAG_skeleton_unit;
356 case dwarf::DW_UT_split_compile:
357 case dwarf::DW_UT_split_type:
358 return dwarf::isUnitType(Tag);
363 /// Return the number of bytes for the header of a unit of
366 /// This function must be called with a valid unit type which in
367 /// DWARF5 is defined as one of the following six types.
368 static uint32_t getDWARF5HeaderSize(uint8_t UnitType) {
370 case dwarf::DW_UT_compile:
371 case dwarf::DW_UT_partial:
373 case dwarf::DW_UT_skeleton:
374 case dwarf::DW_UT_split_compile:
376 case dwarf::DW_UT_type:
377 case dwarf::DW_UT_split_type:
380 llvm_unreachable("Invalid UnitType.");
383 llvm::Optional<object::SectionedAddress> getBaseAddress();
385 DWARFDie getUnitDIE(bool ExtractUnitDIEOnly = true) {
386 extractDIEsIfNeeded(ExtractUnitDIEOnly);
387 if (DieArray.empty())
389 return DWARFDie(this, &DieArray[0]);
392 DWARFDie getNonSkeletonUnitDIE(bool ExtractUnitDIEOnly = true) {
395 return DWO->getUnitDIE(ExtractUnitDIEOnly);
396 return getUnitDIE(ExtractUnitDIEOnly);
399 const char *getCompilationDir();
400 Optional<uint64_t> getDWOId() {
401 extractDIEsIfNeeded(/*CUDieOnly*/ true);
402 return getHeader().getDWOId();
404 void setDWOId(uint64_t NewID) { Header.setDWOId(NewID); }
406 /// Return a vector of address ranges resulting from a (possibly encoded)
407 /// range list starting at a given offset in the appropriate ranges section.
408 Expected<DWARFAddressRangesVector> findRnglistFromOffset(uint32_t Offset);
410 /// Return a vector of address ranges retrieved from an encoded range
411 /// list whose offset is found via a table lookup given an index (DWARF v5
413 Expected<DWARFAddressRangesVector> findRnglistFromIndex(uint32_t Index);
415 /// Return a rangelist's offset based on an index. The index designates
416 /// an entry in the rangelist table's offset array and is supplied by
417 /// DW_FORM_rnglistx.
418 Optional<uint32_t> getRnglistOffset(uint32_t Index) {
420 return RngListTable->getOffsetEntry(Index);
424 Expected<DWARFAddressRangesVector> collectAddressRanges();
426 /// Returns subprogram DIE with address range encompassing the provided
427 /// address. The pointer is alive as long as parsed compile unit DIEs are not
429 DWARFDie getSubroutineForAddress(uint64_t Address);
431 /// getInlinedChainForAddress - fetches inlined chain for a given address.
432 /// Returns empty chain if there is no subprogram containing address. The
433 /// chain is valid as long as parsed compile unit DIEs are not cleared.
434 void getInlinedChainForAddress(uint64_t Address,
435 SmallVectorImpl<DWARFDie> &InlinedChain);
437 /// Return the DWARFUnitVector containing this unit.
438 const DWARFUnitVector &getUnitVector() const { return UnitVector; }
440 /// Returns the number of DIEs in the unit. Parses the unit
442 unsigned getNumDIEs() {
443 extractDIEsIfNeeded(false);
444 return DieArray.size();
447 /// Return the index of a DIE inside the unit's DIE vector.
449 /// It is illegal to call this method with a DIE that hasn't be
450 /// created by this unit. In other word, it's illegal to call this
451 /// method on a DIE that isn't accessible by following
452 /// children/sibling links starting from this unit's getUnitDIE().
453 uint32_t getDIEIndex(const DWARFDie &D) {
454 return getDIEIndex(D.getDebugInfoEntry());
457 /// Return the DIE object at the given index.
458 DWARFDie getDIEAtIndex(unsigned Index) {
459 assert(Index < DieArray.size());
460 return DWARFDie(this, &DieArray[Index]);
463 DWARFDie getParent(const DWARFDebugInfoEntry *Die);
464 DWARFDie getSibling(const DWARFDebugInfoEntry *Die);
465 DWARFDie getPreviousSibling(const DWARFDebugInfoEntry *Die);
466 DWARFDie getFirstChild(const DWARFDebugInfoEntry *Die);
467 DWARFDie getLastChild(const DWARFDebugInfoEntry *Die);
469 /// Return the DIE object for a given offset inside the
470 /// unit's DIE vector.
472 /// The unit needs to have its DIEs extracted for this method to work.
473 DWARFDie getDIEForOffset(uint32_t Offset) {
474 extractDIEsIfNeeded(false);
475 assert(!DieArray.empty());
477 llvm::partition_point(DieArray, [=](const DWARFDebugInfoEntry &DIE) {
478 return DIE.getOffset() < Offset;
480 if (It != DieArray.end() && It->getOffset() == Offset)
481 return DWARFDie(this, &*It);
485 uint32_t getLineTableOffset() const {
486 if (auto IndexEntry = Header.getIndexEntry())
487 if (const auto *Contrib = IndexEntry->getOffset(DW_SECT_LINE))
488 return Contrib->Offset;
492 die_iterator_range dies() {
493 extractDIEsIfNeeded(false);
494 return die_iterator_range(DieArray.begin(), DieArray.end());
497 virtual void dump(raw_ostream &OS, DIDumpOptions DumpOpts) = 0;
499 /// Size in bytes of the .debug_info data associated with this compile unit.
500 size_t getDebugInfoSize() const {
501 return Header.getLength() + 4 - getHeaderSize();
504 /// extractDIEsIfNeeded - Parses a compile unit and indexes its DIEs if it
505 /// hasn't already been done. Returns the number of DIEs parsed at this call.
506 size_t extractDIEsIfNeeded(bool CUDieOnly);
508 /// extractDIEsToVector - Appends all parsed DIEs to a vector.
509 void extractDIEsToVector(bool AppendCUDie, bool AppendNonCUDIEs,
510 std::vector<DWARFDebugInfoEntry> &DIEs) const;
512 /// clearDIEs - Clear parsed DIEs to keep memory usage low.
513 void clearDIEs(bool KeepCUDie);
515 /// parseDWO - Parses .dwo file for current compile unit. Returns true if
516 /// it was actually constructed.
520 } // end namespace llvm
522 #endif // LLVM_DEBUGINFO_DWARF_DWARFUNIT_H