1 //===- ObjectFile.h - File format independent object file -------*- 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 // This file declares a file format independent ObjectFile class.
11 //===----------------------------------------------------------------------===//
13 #ifndef LLVM_OBJECT_OBJECTFILE_H
14 #define LLVM_OBJECT_OBJECTFILE_H
16 #include "llvm/ADT/DenseMapInfo.h"
17 #include "llvm/ADT/StringRef.h"
18 #include "llvm/ADT/Triple.h"
19 #include "llvm/ADT/iterator_range.h"
20 #include "llvm/BinaryFormat/Magic.h"
21 #include "llvm/MC/SubtargetFeature.h"
22 #include "llvm/Object/Binary.h"
23 #include "llvm/Object/Error.h"
24 #include "llvm/Object/SymbolicFile.h"
25 #include "llvm/Support/Casting.h"
26 #include "llvm/Support/Error.h"
27 #include "llvm/Support/FileSystem.h"
28 #include "llvm/Support/MemoryBuffer.h"
32 #include <system_error>
36 class ARMAttributeParser
;
41 class MachOObjectFile
;
45 class symbol_iterator
;
48 using section_iterator
= content_iterator
<SectionRef
>;
50 /// This is a value type class that represents a single relocation in the list
51 /// of relocations in the object file.
53 DataRefImpl RelocationPimpl
;
54 const ObjectFile
*OwningObject
= nullptr;
57 RelocationRef() = default;
58 RelocationRef(DataRefImpl RelocationP
, const ObjectFile
*Owner
);
60 bool operator==(const RelocationRef
&Other
) const;
64 uint64_t getOffset() const;
65 symbol_iterator
getSymbol() const;
66 uint64_t getType() const;
68 /// Get a string that represents the type of this relocation.
70 /// This is for display purposes only.
71 void getTypeName(SmallVectorImpl
<char> &Result
) const;
73 DataRefImpl
getRawDataRefImpl() const;
74 const ObjectFile
*getObject() const;
77 using relocation_iterator
= content_iterator
<RelocationRef
>;
79 /// This is a value type class that represents a single section in the list of
80 /// sections in the object file.
82 friend class SymbolRef
;
84 DataRefImpl SectionPimpl
;
85 const ObjectFile
*OwningObject
= nullptr;
88 SectionRef() = default;
89 SectionRef(DataRefImpl SectionP
, const ObjectFile
*Owner
);
91 bool operator==(const SectionRef
&Other
) const;
92 bool operator!=(const SectionRef
&Other
) const;
93 bool operator<(const SectionRef
&Other
) const;
97 Expected
<StringRef
> getName() const;
98 uint64_t getAddress() const;
99 uint64_t getIndex() const;
100 uint64_t getSize() const;
101 Expected
<StringRef
> getContents() const;
103 /// Get the alignment of this section as the actual value (not log 2).
104 uint64_t getAlignment() const;
106 bool isCompressed() const;
107 /// Whether this section contains instructions.
109 /// Whether this section contains data, not instructions.
111 /// Whether this section contains BSS uninitialized data.
113 bool isVirtual() const;
114 bool isBitcode() const;
115 bool isStripped() const;
117 /// Whether this section will be placed in the text segment, according to the
118 /// Berkeley size format. This is true if the section is allocatable, and
119 /// contains either code or readonly data.
120 bool isBerkeleyText() const;
121 /// Whether this section will be placed in the data segment, according to the
122 /// Berkeley size format. This is true if the section is allocatable and
123 /// contains data (e.g. PROGBITS), but is not text.
124 bool isBerkeleyData() const;
126 bool containsSymbol(SymbolRef S
) const;
128 relocation_iterator
relocation_begin() const;
129 relocation_iterator
relocation_end() const;
130 iterator_range
<relocation_iterator
> relocations() const {
131 return make_range(relocation_begin(), relocation_end());
133 section_iterator
getRelocatedSection() const;
135 DataRefImpl
getRawDataRefImpl() const;
136 const ObjectFile
*getObject() const;
139 struct SectionedAddress
{
140 const static uint64_t UndefSection
= UINT64_MAX
;
142 uint64_t Address
= 0;
143 uint64_t SectionIndex
= UndefSection
;
146 inline bool operator<(const SectionedAddress
&LHS
,
147 const SectionedAddress
&RHS
) {
148 return std::tie(LHS
.SectionIndex
, LHS
.Address
) <
149 std::tie(RHS
.SectionIndex
, RHS
.Address
);
152 inline bool operator==(const SectionedAddress
&LHS
,
153 const SectionedAddress
&RHS
) {
154 return std::tie(LHS
.SectionIndex
, LHS
.Address
) ==
155 std::tie(RHS
.SectionIndex
, RHS
.Address
);
158 /// This is a value type class that represents a single symbol in the list of
159 /// symbols in the object file.
160 class SymbolRef
: public BasicSymbolRef
{
161 friend class SectionRef
;
165 ST_Unknown
, // Type not specified
173 SymbolRef() = default;
174 SymbolRef(DataRefImpl SymbolP
, const ObjectFile
*Owner
);
175 SymbolRef(const BasicSymbolRef
&B
) : BasicSymbolRef(B
) {
176 assert(isa
<ObjectFile
>(BasicSymbolRef::getObject()));
179 Expected
<StringRef
> getName() const;
180 /// Returns the symbol virtual address (i.e. address at which it will be
182 Expected
<uint64_t> getAddress() const;
184 /// Return the value of the symbol depending on the object this can be an
185 /// offset or a virtual address.
186 uint64_t getValue() const;
188 /// Get the alignment of this symbol as the actual value (not log 2).
189 uint32_t getAlignment() const;
190 uint64_t getCommonSize() const;
191 Expected
<SymbolRef::Type
> getType() const;
193 /// Get section this symbol is defined in reference to. Result is
194 /// end_sections() if it is undefined or is an absolute symbol.
195 Expected
<section_iterator
> getSection() const;
197 const ObjectFile
*getObject() const;
200 class symbol_iterator
: public basic_symbol_iterator
{
202 symbol_iterator(SymbolRef Sym
) : basic_symbol_iterator(Sym
) {}
203 symbol_iterator(const basic_symbol_iterator
&B
)
204 : basic_symbol_iterator(SymbolRef(B
->getRawDataRefImpl(),
205 cast
<ObjectFile
>(B
->getObject()))) {}
207 const SymbolRef
*operator->() const {
208 const BasicSymbolRef
&P
= basic_symbol_iterator::operator *();
209 return static_cast<const SymbolRef
*>(&P
);
212 const SymbolRef
&operator*() const {
213 const BasicSymbolRef
&P
= basic_symbol_iterator::operator *();
214 return static_cast<const SymbolRef
&>(P
);
218 /// This class is the base class for all object file types. Concrete instances
219 /// of this object are created by createObjectFile, which figures out which type
221 class ObjectFile
: public SymbolicFile
{
222 virtual void anchor();
225 ObjectFile(unsigned int Type
, MemoryBufferRef Source
);
227 const uint8_t *base() const {
228 return reinterpret_cast<const uint8_t *>(Data
.getBufferStart());
231 // These functions are for SymbolRef to call internally. The main goal of
232 // this is to allow SymbolRef::SymbolPimpl to point directly to the symbol
233 // entry in the memory mapped object file. SymbolPimpl cannot contain any
234 // virtual functions because then it could not point into the memory mapped
237 // Implementations assume that the DataRefImpl is valid and has not been
238 // modified externally. It's UB otherwise.
239 friend class SymbolRef
;
241 virtual Expected
<StringRef
> getSymbolName(DataRefImpl Symb
) const = 0;
242 Error
printSymbolName(raw_ostream
&OS
,
243 DataRefImpl Symb
) const override
;
244 virtual Expected
<uint64_t> getSymbolAddress(DataRefImpl Symb
) const = 0;
245 virtual uint64_t getSymbolValueImpl(DataRefImpl Symb
) const = 0;
246 virtual uint32_t getSymbolAlignment(DataRefImpl Symb
) const;
247 virtual uint64_t getCommonSymbolSizeImpl(DataRefImpl Symb
) const = 0;
248 virtual Expected
<SymbolRef::Type
> getSymbolType(DataRefImpl Symb
) const = 0;
249 virtual Expected
<section_iterator
>
250 getSymbolSection(DataRefImpl Symb
) const = 0;
252 // Same as above for SectionRef.
253 friend class SectionRef
;
255 virtual void moveSectionNext(DataRefImpl
&Sec
) const = 0;
256 virtual Expected
<StringRef
> getSectionName(DataRefImpl Sec
) const = 0;
257 virtual uint64_t getSectionAddress(DataRefImpl Sec
) const = 0;
258 virtual uint64_t getSectionIndex(DataRefImpl Sec
) const = 0;
259 virtual uint64_t getSectionSize(DataRefImpl Sec
) const = 0;
260 virtual Expected
<ArrayRef
<uint8_t>>
261 getSectionContents(DataRefImpl Sec
) const = 0;
262 virtual uint64_t getSectionAlignment(DataRefImpl Sec
) const = 0;
263 virtual bool isSectionCompressed(DataRefImpl Sec
) const = 0;
264 virtual bool isSectionText(DataRefImpl Sec
) const = 0;
265 virtual bool isSectionData(DataRefImpl Sec
) const = 0;
266 virtual bool isSectionBSS(DataRefImpl Sec
) const = 0;
267 // A section is 'virtual' if its contents aren't present in the object image.
268 virtual bool isSectionVirtual(DataRefImpl Sec
) const = 0;
269 virtual bool isSectionBitcode(DataRefImpl Sec
) const;
270 virtual bool isSectionStripped(DataRefImpl Sec
) const;
271 virtual bool isBerkeleyText(DataRefImpl Sec
) const;
272 virtual bool isBerkeleyData(DataRefImpl Sec
) const;
273 virtual relocation_iterator
section_rel_begin(DataRefImpl Sec
) const = 0;
274 virtual relocation_iterator
section_rel_end(DataRefImpl Sec
) const = 0;
275 virtual section_iterator
getRelocatedSection(DataRefImpl Sec
) const;
277 // Same as above for RelocationRef.
278 friend class RelocationRef
;
279 virtual void moveRelocationNext(DataRefImpl
&Rel
) const = 0;
280 virtual uint64_t getRelocationOffset(DataRefImpl Rel
) const = 0;
281 virtual symbol_iterator
getRelocationSymbol(DataRefImpl Rel
) const = 0;
282 virtual uint64_t getRelocationType(DataRefImpl Rel
) const = 0;
283 virtual void getRelocationTypeName(DataRefImpl Rel
,
284 SmallVectorImpl
<char> &Result
) const = 0;
286 uint64_t getSymbolValue(DataRefImpl Symb
) const;
289 ObjectFile() = delete;
290 ObjectFile(const ObjectFile
&other
) = delete;
292 uint64_t getCommonSymbolSize(DataRefImpl Symb
) const {
293 assert(getSymbolFlags(Symb
) & SymbolRef::SF_Common
);
294 return getCommonSymbolSizeImpl(Symb
);
297 virtual std::vector
<SectionRef
> dynamic_relocation_sections() const {
298 return std::vector
<SectionRef
>();
301 using symbol_iterator_range
= iterator_range
<symbol_iterator
>;
302 symbol_iterator_range
symbols() const {
303 return symbol_iterator_range(symbol_begin(), symbol_end());
306 virtual section_iterator
section_begin() const = 0;
307 virtual section_iterator
section_end() const = 0;
309 using section_iterator_range
= iterator_range
<section_iterator
>;
310 section_iterator_range
sections() const {
311 return section_iterator_range(section_begin(), section_end());
314 /// The number of bytes used to represent an address in this object
316 virtual uint8_t getBytesInAddress() const = 0;
318 virtual StringRef
getFileFormatName() const = 0;
319 virtual Triple::ArchType
getArch() const = 0;
320 virtual SubtargetFeatures
getFeatures() const = 0;
321 virtual void setARMSubArch(Triple
&TheTriple
) const { }
322 virtual Expected
<uint64_t> getStartAddress() const {
323 return errorCodeToError(object_error::parse_failed
);
326 /// Create a triple from the data in this object file.
327 Triple
makeTriple() const;
329 /// Maps a debug section name to a standard DWARF section name.
330 virtual StringRef
mapDebugSectionName(StringRef Name
) const { return Name
; }
332 /// True if this is a relocatable object (.o/.obj).
333 virtual bool isRelocatableObject() const = 0;
335 /// @returns Pointer to ObjectFile subclass to handle this type of object.
336 /// @param ObjectPath The path to the object file. ObjectPath.isObject must
338 /// Create ObjectFile from path.
339 static Expected
<OwningBinary
<ObjectFile
>>
340 createObjectFile(StringRef ObjectPath
);
342 static Expected
<std::unique_ptr
<ObjectFile
>>
343 createObjectFile(MemoryBufferRef Object
, llvm::file_magic Type
);
344 static Expected
<std::unique_ptr
<ObjectFile
>>
345 createObjectFile(MemoryBufferRef Object
) {
346 return createObjectFile(Object
, llvm::file_magic::unknown
);
349 static bool classof(const Binary
*v
) {
350 return v
->isObject();
353 static Expected
<std::unique_ptr
<COFFObjectFile
>>
354 createCOFFObjectFile(MemoryBufferRef Object
);
356 static Expected
<std::unique_ptr
<ObjectFile
>>
357 createXCOFFObjectFile(MemoryBufferRef Object
, unsigned FileType
);
359 static Expected
<std::unique_ptr
<ObjectFile
>>
360 createELFObjectFile(MemoryBufferRef Object
);
362 static Expected
<std::unique_ptr
<MachOObjectFile
>>
363 createMachOObjectFile(MemoryBufferRef Object
,
364 uint32_t UniversalCputype
= 0,
365 uint32_t UniversalIndex
= 0);
367 static Expected
<std::unique_ptr
<WasmObjectFile
>>
368 createWasmObjectFile(MemoryBufferRef Object
);
371 // Inline function definitions.
372 inline SymbolRef::SymbolRef(DataRefImpl SymbolP
, const ObjectFile
*Owner
)
373 : BasicSymbolRef(SymbolP
, Owner
) {}
375 inline Expected
<StringRef
> SymbolRef::getName() const {
376 return getObject()->getSymbolName(getRawDataRefImpl());
379 inline Expected
<uint64_t> SymbolRef::getAddress() const {
380 return getObject()->getSymbolAddress(getRawDataRefImpl());
383 inline uint64_t SymbolRef::getValue() const {
384 return getObject()->getSymbolValue(getRawDataRefImpl());
387 inline uint32_t SymbolRef::getAlignment() const {
388 return getObject()->getSymbolAlignment(getRawDataRefImpl());
391 inline uint64_t SymbolRef::getCommonSize() const {
392 return getObject()->getCommonSymbolSize(getRawDataRefImpl());
395 inline Expected
<section_iterator
> SymbolRef::getSection() const {
396 return getObject()->getSymbolSection(getRawDataRefImpl());
399 inline Expected
<SymbolRef::Type
> SymbolRef::getType() const {
400 return getObject()->getSymbolType(getRawDataRefImpl());
403 inline const ObjectFile
*SymbolRef::getObject() const {
404 const SymbolicFile
*O
= BasicSymbolRef::getObject();
405 return cast
<ObjectFile
>(O
);
409 inline SectionRef::SectionRef(DataRefImpl SectionP
,
410 const ObjectFile
*Owner
)
411 : SectionPimpl(SectionP
)
412 , OwningObject(Owner
) {}
414 inline bool SectionRef::operator==(const SectionRef
&Other
) const {
415 return OwningObject
== Other
.OwningObject
&&
416 SectionPimpl
== Other
.SectionPimpl
;
419 inline bool SectionRef::operator!=(const SectionRef
&Other
) const {
420 return !(*this == Other
);
423 inline bool SectionRef::operator<(const SectionRef
&Other
) const {
424 assert(OwningObject
== Other
.OwningObject
);
425 return SectionPimpl
< Other
.SectionPimpl
;
428 inline void SectionRef::moveNext() {
429 return OwningObject
->moveSectionNext(SectionPimpl
);
432 inline Expected
<StringRef
> SectionRef::getName() const {
433 return OwningObject
->getSectionName(SectionPimpl
);
436 inline uint64_t SectionRef::getAddress() const {
437 return OwningObject
->getSectionAddress(SectionPimpl
);
440 inline uint64_t SectionRef::getIndex() const {
441 return OwningObject
->getSectionIndex(SectionPimpl
);
444 inline uint64_t SectionRef::getSize() const {
445 return OwningObject
->getSectionSize(SectionPimpl
);
448 inline Expected
<StringRef
> SectionRef::getContents() const {
449 Expected
<ArrayRef
<uint8_t>> Res
=
450 OwningObject
->getSectionContents(SectionPimpl
);
452 return Res
.takeError();
453 return StringRef(reinterpret_cast<const char *>(Res
->data()), Res
->size());
456 inline uint64_t SectionRef::getAlignment() const {
457 return OwningObject
->getSectionAlignment(SectionPimpl
);
460 inline bool SectionRef::isCompressed() const {
461 return OwningObject
->isSectionCompressed(SectionPimpl
);
464 inline bool SectionRef::isText() const {
465 return OwningObject
->isSectionText(SectionPimpl
);
468 inline bool SectionRef::isData() const {
469 return OwningObject
->isSectionData(SectionPimpl
);
472 inline bool SectionRef::isBSS() const {
473 return OwningObject
->isSectionBSS(SectionPimpl
);
476 inline bool SectionRef::isVirtual() const {
477 return OwningObject
->isSectionVirtual(SectionPimpl
);
480 inline bool SectionRef::isBitcode() const {
481 return OwningObject
->isSectionBitcode(SectionPimpl
);
484 inline bool SectionRef::isStripped() const {
485 return OwningObject
->isSectionStripped(SectionPimpl
);
488 inline bool SectionRef::isBerkeleyText() const {
489 return OwningObject
->isBerkeleyText(SectionPimpl
);
492 inline bool SectionRef::isBerkeleyData() const {
493 return OwningObject
->isBerkeleyData(SectionPimpl
);
496 inline relocation_iterator
SectionRef::relocation_begin() const {
497 return OwningObject
->section_rel_begin(SectionPimpl
);
500 inline relocation_iterator
SectionRef::relocation_end() const {
501 return OwningObject
->section_rel_end(SectionPimpl
);
504 inline section_iterator
SectionRef::getRelocatedSection() const {
505 return OwningObject
->getRelocatedSection(SectionPimpl
);
508 inline DataRefImpl
SectionRef::getRawDataRefImpl() const {
512 inline const ObjectFile
*SectionRef::getObject() const {
517 inline RelocationRef::RelocationRef(DataRefImpl RelocationP
,
518 const ObjectFile
*Owner
)
519 : RelocationPimpl(RelocationP
)
520 , OwningObject(Owner
) {}
522 inline bool RelocationRef::operator==(const RelocationRef
&Other
) const {
523 return RelocationPimpl
== Other
.RelocationPimpl
;
526 inline void RelocationRef::moveNext() {
527 return OwningObject
->moveRelocationNext(RelocationPimpl
);
530 inline uint64_t RelocationRef::getOffset() const {
531 return OwningObject
->getRelocationOffset(RelocationPimpl
);
534 inline symbol_iterator
RelocationRef::getSymbol() const {
535 return OwningObject
->getRelocationSymbol(RelocationPimpl
);
538 inline uint64_t RelocationRef::getType() const {
539 return OwningObject
->getRelocationType(RelocationPimpl
);
542 inline void RelocationRef::getTypeName(SmallVectorImpl
<char> &Result
) const {
543 return OwningObject
->getRelocationTypeName(RelocationPimpl
, Result
);
546 inline DataRefImpl
RelocationRef::getRawDataRefImpl() const {
547 return RelocationPimpl
;
550 inline const ObjectFile
*RelocationRef::getObject() const {
554 } // end namespace object
556 template <> struct DenseMapInfo
<object::SectionRef
> {
557 static bool isEqual(const object::SectionRef
&A
,
558 const object::SectionRef
&B
) {
561 static object::SectionRef
getEmptyKey() {
562 return object::SectionRef({}, nullptr);
564 static object::SectionRef
getTombstoneKey() {
565 object::DataRefImpl TS
;
566 TS
.p
= (uintptr_t)-1;
567 return object::SectionRef(TS
, nullptr);
569 static unsigned getHashValue(const object::SectionRef
&Sec
) {
570 object::DataRefImpl Raw
= Sec
.getRawDataRefImpl();
571 return hash_combine(Raw
.p
, Raw
.d
.a
, Raw
.d
.b
);
575 } // end namespace llvm
577 #endif // LLVM_OBJECT_OBJECTFILE_H