[Alignment][NFC] Migrate Instructions to Align
[llvm-core.git] / include / llvm / DebugInfo / DWARF / DWARFUnit.h
blob51de114a350665eeb25a476a5707f1b44895ef97
1 //===- DWARFUnit.h ----------------------------------------------*- C++ -*-===//
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 //===----------------------------------------------------------------------===//
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"
27 #include <algorithm>
28 #include <cassert>
29 #include <cstddef>
30 #include <cstdint>
31 #include <map>
32 #include <memory>
33 #include <utility>
34 #include <vector>
36 namespace llvm {
38 class DWARFAbbreviationDeclarationSet;
39 class DWARFContext;
40 class DWARFDebugAbbrev;
41 class DWARFUnit;
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.
48 uint64_t Offset = 0;
49 // Version, address size, and DWARF format.
50 dwarf::FormParams FormParams;
51 uint64_t Length = 0;
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 uint64_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.
65 uint8_t UnitType = 0;
67 // Size as parsed. uint8_t for compactness.
68 uint8_t Size = 0;
70 public:
71 /// Parse a unit header from \p debug_info starting at \p offset_ptr.
72 bool extract(DWARFContext &Context, const DWARFDataExtractor &debug_info,
73 uint64_t *offset_ptr, DWARFSectionKind Kind = DW_SECT_INFO,
74 const DWARFUnitIndex *Index = nullptr,
75 const DWARFUnitIndex::Entry *Entry = nullptr);
76 uint64_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");
90 DWOId = Id;
92 const DWARFUnitIndex::Entry *getIndexEntry() const { return IndexEntry; }
93 uint64_t getTypeHash() const { return TypeHash; }
94 uint64_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 uint8_t getUnitLengthFieldByteSize() const {
101 return dwarf::getUnitLengthFieldByteSize(FormParams.Format);
103 uint64_t getNextUnitOffset() const {
104 return Offset + Length + getUnitLengthFieldByteSize();
108 const DWARFUnitIndex &getDWARFUnitIndex(DWARFContext &Context,
109 DWARFSectionKind Kind);
111 /// Describe a collection of units. Intended to hold all units either from
112 /// .debug_info and .debug_types, or from .debug_info.dwo and .debug_types.dwo.
113 class DWARFUnitVector final : public SmallVector<std::unique_ptr<DWARFUnit>, 1> {
114 std::function<std::unique_ptr<DWARFUnit>(uint64_t, DWARFSectionKind,
115 const DWARFSection *,
116 const DWARFUnitIndex::Entry *)>
117 Parser;
118 int NumInfoUnits = -1;
120 public:
121 using UnitVector = SmallVectorImpl<std::unique_ptr<DWARFUnit>>;
122 using iterator = typename UnitVector::iterator;
123 using iterator_range = llvm::iterator_range<typename UnitVector::iterator>;
125 DWARFUnit *getUnitForOffset(uint64_t Offset) const;
126 DWARFUnit *getUnitForIndexEntry(const DWARFUnitIndex::Entry &E);
128 /// Read units from a .debug_info or .debug_types section. Calls made
129 /// before finishedInfoUnits() are assumed to be for .debug_info sections,
130 /// calls after finishedInfoUnits() are for .debug_types sections. Caller
131 /// must not mix calls to addUnitsForSection and addUnitsForDWOSection.
132 void addUnitsForSection(DWARFContext &C, const DWARFSection &Section,
133 DWARFSectionKind SectionKind);
134 /// Read units from a .debug_info.dwo or .debug_types.dwo section. Calls
135 /// made before finishedInfoUnits() are assumed to be for .debug_info.dwo
136 /// sections, calls after finishedInfoUnits() are for .debug_types.dwo
137 /// sections. Caller must not mix calls to addUnitsForSection and
138 /// addUnitsForDWOSection.
139 void addUnitsForDWOSection(DWARFContext &C, const DWARFSection &DWOSection,
140 DWARFSectionKind SectionKind, bool Lazy = false);
142 /// Add an existing DWARFUnit to this UnitVector. This is used by the DWARF
143 /// verifier to process unit separately.
144 DWARFUnit *addUnit(std::unique_ptr<DWARFUnit> Unit);
146 /// Returns number of all units held by this instance.
147 unsigned getNumUnits() const { return size(); }
148 /// Returns number of units from all .debug_info[.dwo] sections.
149 unsigned getNumInfoUnits() const {
150 return NumInfoUnits == -1 ? size() : NumInfoUnits;
152 /// Returns number of units from all .debug_types[.dwo] sections.
153 unsigned getNumTypesUnits() const { return size() - NumInfoUnits; }
154 /// Indicate that parsing .debug_info[.dwo] is done, and remaining units
155 /// will be from .debug_types[.dwo].
156 void finishedInfoUnits() { NumInfoUnits = size(); }
158 private:
159 void addUnitsImpl(DWARFContext &Context, const DWARFObject &Obj,
160 const DWARFSection &Section, const DWARFDebugAbbrev *DA,
161 const DWARFSection *RS, const DWARFSection *LocSection,
162 StringRef SS, const DWARFSection &SOS,
163 const DWARFSection *AOS, const DWARFSection &LS, bool LE,
164 bool IsDWO, bool Lazy, DWARFSectionKind SectionKind);
167 /// Represents base address of the CU.
168 /// Represents a unit's contribution to the string offsets table.
169 struct StrOffsetsContributionDescriptor {
170 uint64_t Base = 0;
171 /// The contribution size not including the header.
172 uint64_t Size = 0;
173 /// Format and version.
174 dwarf::FormParams FormParams = {0, 0, dwarf::DwarfFormat::DWARF32};
176 StrOffsetsContributionDescriptor(uint64_t Base, uint64_t Size,
177 uint8_t Version, dwarf::DwarfFormat Format)
178 : Base(Base), Size(Size), FormParams({Version, 0, Format}) {}
179 StrOffsetsContributionDescriptor() = default;
181 uint8_t getVersion() const { return FormParams.Version; }
182 dwarf::DwarfFormat getFormat() const { return FormParams.Format; }
183 uint8_t getDwarfOffsetByteSize() const {
184 return FormParams.getDwarfOffsetByteSize();
186 /// Determine whether a contribution to the string offsets table is
187 /// consistent with the relevant section size and that its length is
188 /// a multiple of the size of one of its entries.
189 Expected<StrOffsetsContributionDescriptor>
190 validateContributionSize(DWARFDataExtractor &DA);
193 class DWARFUnit {
194 DWARFContext &Context;
195 /// Section containing this DWARFUnit.
196 const DWARFSection &InfoSection;
198 DWARFUnitHeader Header;
199 const DWARFDebugAbbrev *Abbrev;
200 const DWARFSection *RangeSection;
201 uint64_t RangeSectionBase;
202 /// We either keep track of the location list section or its data, depending
203 /// on whether we are handling a split DWARF section or not.
204 union {
205 const DWARFSection *LocSection;
206 StringRef LocSectionData;
208 const DWARFSection &LineSection;
209 StringRef StringSection;
210 const DWARFSection &StringOffsetSection;
211 const DWARFSection *AddrOffsetSection;
212 uint32_t AddrOffsetSectionBase = 0;
213 bool isLittleEndian;
214 bool IsDWO;
215 const DWARFUnitVector &UnitVector;
217 /// Start, length, and DWARF format of the unit's contribution to the string
218 /// offsets table (DWARF v5).
219 Optional<StrOffsetsContributionDescriptor> StringOffsetsTableContribution;
221 /// A table of range lists (DWARF v5 and later).
222 Optional<DWARFDebugRnglistTable> RngListTable;
224 mutable const DWARFAbbreviationDeclarationSet *Abbrevs;
225 llvm::Optional<object::SectionedAddress> BaseAddr;
226 /// The compile unit debug information entry items.
227 std::vector<DWARFDebugInfoEntry> DieArray;
229 /// Map from range's start address to end address and corresponding DIE.
230 /// IntervalMap does not support range removal, as a result, we use the
231 /// std::map::upper_bound for address range lookup.
232 std::map<uint64_t, std::pair<uint64_t, DWARFDie>> AddrDieMap;
234 using die_iterator_range =
235 iterator_range<std::vector<DWARFDebugInfoEntry>::iterator>;
237 std::shared_ptr<DWARFUnit> DWO;
239 uint32_t getDIEIndex(const DWARFDebugInfoEntry *Die) {
240 auto First = DieArray.data();
241 assert(Die >= First && Die < First + DieArray.size());
242 return Die - First;
245 protected:
246 const DWARFUnitHeader &getHeader() const { return Header; }
248 /// Size in bytes of the parsed unit header.
249 uint32_t getHeaderSize() const { return Header.getSize(); }
251 /// Find the unit's contribution to the string offsets table and determine its
252 /// length and form. The given offset is expected to be derived from the unit
253 /// DIE's DW_AT_str_offsets_base attribute.
254 Expected<Optional<StrOffsetsContributionDescriptor>>
255 determineStringOffsetsTableContribution(DWARFDataExtractor &DA);
257 /// Find the unit's contribution to the string offsets table and determine its
258 /// length and form. The given offset is expected to be 0 in a dwo file or,
259 /// in a dwp file, the start of the unit's contribution to the string offsets
260 /// table section (as determined by the index table).
261 Expected<Optional<StrOffsetsContributionDescriptor>>
262 determineStringOffsetsTableContributionDWO(DWARFDataExtractor &DA);
264 public:
265 DWARFUnit(DWARFContext &Context, const DWARFSection &Section,
266 const DWARFUnitHeader &Header, const DWARFDebugAbbrev *DA,
267 const DWARFSection *RS, const DWARFSection *LocSection,
268 StringRef SS, const DWARFSection &SOS, const DWARFSection *AOS,
269 const DWARFSection &LS, bool LE, bool IsDWO,
270 const DWARFUnitVector &UnitVector);
272 virtual ~DWARFUnit();
274 bool isDWOUnit() const { return IsDWO; }
275 DWARFContext& getContext() const { return Context; }
276 const DWARFSection &getInfoSection() const { return InfoSection; }
277 const DWARFSection *getLocSection() const { return LocSection; }
278 StringRef getLocSectionData() const { return LocSectionData; }
279 uint64_t getOffset() const { return Header.getOffset(); }
280 const dwarf::FormParams &getFormParams() const {
281 return Header.getFormParams();
283 uint16_t getVersion() const { return Header.getVersion(); }
284 uint8_t getAddressByteSize() const { return Header.getAddressByteSize(); }
285 uint8_t getRefAddrByteSize() const { return Header.getRefAddrByteSize(); }
286 uint8_t getDwarfOffsetByteSize() const {
287 return Header.getDwarfOffsetByteSize();
289 uint64_t getLength() const { return Header.getLength(); }
290 uint8_t getUnitType() const { return Header.getUnitType(); }
291 bool isTypeUnit() const { return Header.isTypeUnit(); }
292 uint64_t getNextUnitOffset() const { return Header.getNextUnitOffset(); }
293 const DWARFSection &getLineSection() const { return LineSection; }
294 StringRef getStringSection() const { return StringSection; }
295 const DWARFSection &getStringOffsetSection() const {
296 return StringOffsetSection;
299 void setAddrOffsetSection(const DWARFSection *AOS, uint32_t Base) {
300 AddrOffsetSection = AOS;
301 AddrOffsetSectionBase = Base;
304 /// Recursively update address to Die map.
305 void updateAddressDieMap(DWARFDie Die);
307 void setRangesSection(const DWARFSection *RS, uint64_t Base) {
308 RangeSection = RS;
309 RangeSectionBase = Base;
312 Optional<object::SectionedAddress>
313 getAddrOffsetSectionItem(uint32_t Index) const;
314 Optional<uint64_t> getStringOffsetSectionItem(uint32_t Index) const;
316 DWARFDataExtractor getDebugInfoExtractor() const;
318 DataExtractor getStringExtractor() const {
319 return DataExtractor(StringSection, false, 0);
322 /// Extract the range list referenced by this compile unit from the
323 /// .debug_ranges section. If the extraction is unsuccessful, an error
324 /// is returned. Successful extraction requires that the compile unit
325 /// has already been extracted.
326 Error extractRangeList(uint64_t RangeListOffset,
327 DWARFDebugRangeList &RangeList) const;
328 void clear();
330 const Optional<StrOffsetsContributionDescriptor> &
331 getStringOffsetsTableContribution() const {
332 return StringOffsetsTableContribution;
335 uint8_t getDwarfStringOffsetsByteSize() const {
336 assert(StringOffsetsTableContribution);
337 return StringOffsetsTableContribution->getDwarfOffsetByteSize();
340 uint64_t getStringOffsetsBase() const {
341 assert(StringOffsetsTableContribution);
342 return StringOffsetsTableContribution->Base;
345 const DWARFAbbreviationDeclarationSet *getAbbreviations() const;
347 static bool isMatchingUnitTypeAndTag(uint8_t UnitType, dwarf::Tag Tag) {
348 switch (UnitType) {
349 case dwarf::DW_UT_compile:
350 return Tag == dwarf::DW_TAG_compile_unit;
351 case dwarf::DW_UT_type:
352 return Tag == dwarf::DW_TAG_type_unit;
353 case dwarf::DW_UT_partial:
354 return Tag == dwarf::DW_TAG_partial_unit;
355 case dwarf::DW_UT_skeleton:
356 return Tag == dwarf::DW_TAG_skeleton_unit;
357 case dwarf::DW_UT_split_compile:
358 case dwarf::DW_UT_split_type:
359 return dwarf::isUnitType(Tag);
361 return false;
364 /// Return the number of bytes for the header of a unit of
365 /// UnitType type.
367 /// This function must be called with a valid unit type which in
368 /// DWARF5 is defined as one of the following six types.
369 static uint32_t getDWARF5HeaderSize(uint8_t UnitType) {
370 switch (UnitType) {
371 case dwarf::DW_UT_compile:
372 case dwarf::DW_UT_partial:
373 return 12;
374 case dwarf::DW_UT_skeleton:
375 case dwarf::DW_UT_split_compile:
376 return 20;
377 case dwarf::DW_UT_type:
378 case dwarf::DW_UT_split_type:
379 return 24;
381 llvm_unreachable("Invalid UnitType.");
384 llvm::Optional<object::SectionedAddress> getBaseAddress();
386 DWARFDie getUnitDIE(bool ExtractUnitDIEOnly = true) {
387 extractDIEsIfNeeded(ExtractUnitDIEOnly);
388 if (DieArray.empty())
389 return DWARFDie();
390 return DWARFDie(this, &DieArray[0]);
393 DWARFDie getNonSkeletonUnitDIE(bool ExtractUnitDIEOnly = true) {
394 parseDWO();
395 if (DWO)
396 return DWO->getUnitDIE(ExtractUnitDIEOnly);
397 return getUnitDIE(ExtractUnitDIEOnly);
400 const char *getCompilationDir();
401 Optional<uint64_t> getDWOId() {
402 extractDIEsIfNeeded(/*CUDieOnly*/ true);
403 return getHeader().getDWOId();
405 void setDWOId(uint64_t NewID) { Header.setDWOId(NewID); }
407 /// Return a vector of address ranges resulting from a (possibly encoded)
408 /// range list starting at a given offset in the appropriate ranges section.
409 Expected<DWARFAddressRangesVector> findRnglistFromOffset(uint64_t Offset);
411 /// Return a vector of address ranges retrieved from an encoded range
412 /// list whose offset is found via a table lookup given an index (DWARF v5
413 /// and later).
414 Expected<DWARFAddressRangesVector> findRnglistFromIndex(uint32_t Index);
416 /// Return a rangelist's offset based on an index. The index designates
417 /// an entry in the rangelist table's offset array and is supplied by
418 /// DW_FORM_rnglistx.
419 Optional<uint64_t> getRnglistOffset(uint32_t Index) {
420 if (RngListTable)
421 return RngListTable->getOffsetEntry(Index);
422 return None;
425 Expected<DWARFAddressRangesVector> collectAddressRanges();
427 /// Returns subprogram DIE with address range encompassing the provided
428 /// address. The pointer is alive as long as parsed compile unit DIEs are not
429 /// cleared.
430 DWARFDie getSubroutineForAddress(uint64_t Address);
432 /// getInlinedChainForAddress - fetches inlined chain for a given address.
433 /// Returns empty chain if there is no subprogram containing address. The
434 /// chain is valid as long as parsed compile unit DIEs are not cleared.
435 void getInlinedChainForAddress(uint64_t Address,
436 SmallVectorImpl<DWARFDie> &InlinedChain);
438 /// Return the DWARFUnitVector containing this unit.
439 const DWARFUnitVector &getUnitVector() const { return UnitVector; }
441 /// Returns the number of DIEs in the unit. Parses the unit
442 /// if necessary.
443 unsigned getNumDIEs() {
444 extractDIEsIfNeeded(false);
445 return DieArray.size();
448 /// Return the index of a DIE inside the unit's DIE vector.
450 /// It is illegal to call this method with a DIE that hasn't be
451 /// created by this unit. In other word, it's illegal to call this
452 /// method on a DIE that isn't accessible by following
453 /// children/sibling links starting from this unit's getUnitDIE().
454 uint32_t getDIEIndex(const DWARFDie &D) {
455 return getDIEIndex(D.getDebugInfoEntry());
458 /// Return the DIE object at the given index.
459 DWARFDie getDIEAtIndex(unsigned Index) {
460 assert(Index < DieArray.size());
461 return DWARFDie(this, &DieArray[Index]);
464 DWARFDie getParent(const DWARFDebugInfoEntry *Die);
465 DWARFDie getSibling(const DWARFDebugInfoEntry *Die);
466 DWARFDie getPreviousSibling(const DWARFDebugInfoEntry *Die);
467 DWARFDie getFirstChild(const DWARFDebugInfoEntry *Die);
468 DWARFDie getLastChild(const DWARFDebugInfoEntry *Die);
470 /// Return the DIE object for a given offset inside the
471 /// unit's DIE vector.
473 /// The unit needs to have its DIEs extracted for this method to work.
474 DWARFDie getDIEForOffset(uint64_t Offset) {
475 extractDIEsIfNeeded(false);
476 assert(!DieArray.empty());
477 auto It =
478 llvm::partition_point(DieArray, [=](const DWARFDebugInfoEntry &DIE) {
479 return DIE.getOffset() < Offset;
481 if (It != DieArray.end() && It->getOffset() == Offset)
482 return DWARFDie(this, &*It);
483 return DWARFDie();
486 uint32_t getLineTableOffset() const {
487 if (auto IndexEntry = Header.getIndexEntry())
488 if (const auto *Contrib = IndexEntry->getOffset(DW_SECT_LINE))
489 return Contrib->Offset;
490 return 0;
493 die_iterator_range dies() {
494 extractDIEsIfNeeded(false);
495 return die_iterator_range(DieArray.begin(), DieArray.end());
498 virtual void dump(raw_ostream &OS, DIDumpOptions DumpOpts) = 0;
500 Error tryExtractDIEsIfNeeded(bool CUDieOnly);
502 private:
503 /// Size in bytes of the .debug_info data associated with this compile unit.
504 size_t getDebugInfoSize() const {
505 return Header.getLength() + Header.getUnitLengthFieldByteSize() -
506 getHeaderSize();
509 /// extractDIEsIfNeeded - Parses a compile unit and indexes its DIEs if it
510 /// hasn't already been done
511 void extractDIEsIfNeeded(bool CUDieOnly);
513 /// extractDIEsToVector - Appends all parsed DIEs to a vector.
514 void extractDIEsToVector(bool AppendCUDie, bool AppendNonCUDIEs,
515 std::vector<DWARFDebugInfoEntry> &DIEs) const;
517 /// clearDIEs - Clear parsed DIEs to keep memory usage low.
518 void clearDIEs(bool KeepCUDie);
520 /// parseDWO - Parses .dwo file for current compile unit. Returns true if
521 /// it was actually constructed.
522 bool parseDWO();
525 } // end namespace llvm
527 #endif // LLVM_DEBUGINFO_DWARF_DWARFUNIT_H