1 //===- tools/dsymutil/MachODebugMapParser.cpp - Parse STABS debug maps ----===//
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 #include "BinaryHolder.h"
11 #include "MachOUtils.h"
12 #include "llvm/ADT/Optional.h"
13 #include "llvm/Object/MachO.h"
14 #include "llvm/Support/Path.h"
15 #include "llvm/Support/WithColor.h"
16 #include "llvm/Support/raw_ostream.h"
21 using namespace llvm::dsymutil
;
22 using namespace llvm::object
;
24 class MachODebugMapParser
{
26 MachODebugMapParser(StringRef BinaryPath
, ArrayRef
<std::string
> Archs
,
27 StringRef PathPrefix
= "",
28 bool PaperTrailWarnings
= false, bool Verbose
= false)
29 : BinaryPath(BinaryPath
), Archs(Archs
.begin(), Archs
.end()),
30 PathPrefix(PathPrefix
), PaperTrailWarnings(PaperTrailWarnings
),
31 BinHolder(Verbose
), CurrentDebugMapObject(nullptr) {}
33 /// Parses and returns the DebugMaps of the input binary. The binary contains
34 /// multiple maps in case it is a universal binary.
35 /// \returns an error in case the provided BinaryPath doesn't exist
36 /// or isn't of a supported type.
37 ErrorOr
<std::vector
<std::unique_ptr
<DebugMap
>>> parse();
39 /// Walk the symbol table and dump it.
43 std::string BinaryPath
;
44 SmallVector
<StringRef
, 1> Archs
;
45 std::string PathPrefix
;
46 bool PaperTrailWarnings
;
48 /// Owns the MemoryBuffer for the main binary.
49 BinaryHolder BinHolder
;
50 /// Map of the binary symbol addresses.
51 StringMap
<uint64_t> MainBinarySymbolAddresses
;
52 StringRef MainBinaryStrings
;
53 /// The constructed DebugMap.
54 std::unique_ptr
<DebugMap
> Result
;
55 /// List of common symbols that need to be added to the debug map.
56 std::vector
<std::string
> CommonSymbols
;
58 /// Map of the currently processed object file symbol addresses.
59 StringMap
<Optional
<uint64_t>> CurrentObjectAddresses
;
60 /// Element of the debug map corresponding to the current object file.
61 DebugMapObject
*CurrentDebugMapObject
;
63 /// Holds function info while function scope processing.
64 const char *CurrentFunctionName
;
65 uint64_t CurrentFunctionAddress
;
67 std::unique_ptr
<DebugMap
> parseOneBinary(const MachOObjectFile
&MainBinary
,
68 StringRef BinaryPath
);
71 switchToNewDebugMapObject(StringRef Filename
,
72 sys::TimePoint
<std::chrono::seconds
> Timestamp
);
73 void resetParserState();
74 uint64_t getMainBinarySymbolAddress(StringRef Name
);
75 std::vector
<StringRef
> getMainBinarySymbolNames(uint64_t Value
);
76 void loadMainBinarySymbols(const MachOObjectFile
&MainBinary
);
77 void loadCurrentObjectFileSymbols(const object::MachOObjectFile
&Obj
);
78 void handleStabSymbolTableEntry(uint32_t StringIndex
, uint8_t Type
,
79 uint8_t SectionIndex
, uint16_t Flags
,
82 template <typename STEType
> void handleStabDebugMapEntry(const STEType
&STE
) {
83 handleStabSymbolTableEntry(STE
.n_strx
, STE
.n_type
, STE
.n_sect
, STE
.n_desc
,
87 void addCommonSymbols();
89 /// Dump the symbol table output header.
90 void dumpSymTabHeader(raw_ostream
&OS
, StringRef Arch
);
92 /// Dump the contents of nlist entries.
93 void dumpSymTabEntry(raw_ostream
&OS
, uint64_t Index
, uint32_t StringIndex
,
94 uint8_t Type
, uint8_t SectionIndex
, uint16_t Flags
,
97 template <typename STEType
>
98 void dumpSymTabEntry(raw_ostream
&OS
, uint64_t Index
, const STEType
&STE
) {
99 dumpSymTabEntry(OS
, Index
, STE
.n_strx
, STE
.n_type
, STE
.n_sect
, STE
.n_desc
,
102 void dumpOneBinaryStab(const MachOObjectFile
&MainBinary
,
103 StringRef BinaryPath
);
105 void Warning(const Twine
&Msg
, StringRef File
= StringRef()) {
106 WithColor::warning() << "("
107 << MachOUtils::getArchName(
108 Result
->getTriple().getArchName())
109 << ") " << File
<< " " << Msg
<< "\n";
111 if (PaperTrailWarnings
) {
113 Result
->addDebugMapObject(File
, sys::TimePoint
<std::chrono::seconds
>());
114 if (Result
->end() != Result
->begin())
115 (*--Result
->end())->addWarning(Msg
.str());
120 } // anonymous namespace
122 /// Reset the parser state corresponding to the current object
123 /// file. This is to be called after an object file is finished
125 void MachODebugMapParser::resetParserState() {
126 CommonSymbols
.clear();
127 CurrentObjectAddresses
.clear();
128 CurrentDebugMapObject
= nullptr;
131 /// Commons symbols won't show up in the symbol map but might need to be
132 /// relocated. We can add them to the symbol table ourselves by combining the
133 /// information in the object file (the symbol name) and the main binary (the
135 void MachODebugMapParser::addCommonSymbols() {
136 for (auto &CommonSymbol
: CommonSymbols
) {
137 uint64_t CommonAddr
= getMainBinarySymbolAddress(CommonSymbol
);
138 if (CommonAddr
== 0) {
139 // The main binary doesn't have an address for the given symbol.
142 if (!CurrentDebugMapObject
->addSymbol(CommonSymbol
, None
/*ObjectAddress*/,
143 CommonAddr
, 0 /*size*/)) {
144 // The symbol is already present.
150 /// Create a new DebugMapObject. This function resets the state of the
151 /// parser that was referring to the last object file and sets
152 /// everything up to add symbols to the new one.
153 void MachODebugMapParser::switchToNewDebugMapObject(
154 StringRef Filename
, sys::TimePoint
<std::chrono::seconds
> Timestamp
) {
158 SmallString
<80> Path(PathPrefix
);
159 sys::path::append(Path
, Filename
);
161 auto ObjectEntry
= BinHolder
.getObjectEntry(Path
, Timestamp
);
163 auto Err
= ObjectEntry
.takeError();
164 Warning("unable to open object file: " + toString(std::move(Err
)),
169 auto Object
= ObjectEntry
->getObjectAs
<MachOObjectFile
>(Result
->getTriple());
171 auto Err
= Object
.takeError();
172 Warning("unable to open object file: " + toString(std::move(Err
)),
177 CurrentDebugMapObject
=
178 &Result
->addDebugMapObject(Path
, Timestamp
, MachO::N_OSO
);
179 loadCurrentObjectFileSymbols(*Object
);
182 static std::string
getArchName(const object::MachOObjectFile
&Obj
) {
183 Triple T
= Obj
.getArchTriple();
184 return T
.getArchName();
187 std::unique_ptr
<DebugMap
>
188 MachODebugMapParser::parseOneBinary(const MachOObjectFile
&MainBinary
,
189 StringRef BinaryPath
) {
190 loadMainBinarySymbols(MainBinary
);
191 ArrayRef
<uint8_t> UUID
= MainBinary
.getUuid();
192 Result
= std::make_unique
<DebugMap
>(MainBinary
.getArchTriple(), BinaryPath
, UUID
);
193 MainBinaryStrings
= MainBinary
.getStringTableData();
194 for (const SymbolRef
&Symbol
: MainBinary
.symbols()) {
195 const DataRefImpl
&DRI
= Symbol
.getRawDataRefImpl();
196 if (MainBinary
.is64Bit())
197 handleStabDebugMapEntry(MainBinary
.getSymbol64TableEntry(DRI
));
199 handleStabDebugMapEntry(MainBinary
.getSymbolTableEntry(DRI
));
203 return std::move(Result
);
206 // Table that maps Darwin's Mach-O stab constants to strings to allow printing.
207 // llvm-nm has very similar code, the strings used here are however slightly
208 // different and part of the interface of dsymutil (some project's build-systems
209 // parse the ouptut of dsymutil -s), thus they shouldn't be changed.
210 struct DarwinStabName
{
215 static const struct DarwinStabName DarwinStabNames
[] = {
216 {MachO::N_GSYM
, "N_GSYM"}, {MachO::N_FNAME
, "N_FNAME"},
217 {MachO::N_FUN
, "N_FUN"}, {MachO::N_STSYM
, "N_STSYM"},
218 {MachO::N_LCSYM
, "N_LCSYM"}, {MachO::N_BNSYM
, "N_BNSYM"},
219 {MachO::N_PC
, "N_PC"}, {MachO::N_AST
, "N_AST"},
220 {MachO::N_OPT
, "N_OPT"}, {MachO::N_RSYM
, "N_RSYM"},
221 {MachO::N_SLINE
, "N_SLINE"}, {MachO::N_ENSYM
, "N_ENSYM"},
222 {MachO::N_SSYM
, "N_SSYM"}, {MachO::N_SO
, "N_SO"},
223 {MachO::N_OSO
, "N_OSO"}, {MachO::N_LSYM
, "N_LSYM"},
224 {MachO::N_BINCL
, "N_BINCL"}, {MachO::N_SOL
, "N_SOL"},
225 {MachO::N_PARAMS
, "N_PARAM"}, {MachO::N_VERSION
, "N_VERS"},
226 {MachO::N_OLEVEL
, "N_OLEV"}, {MachO::N_PSYM
, "N_PSYM"},
227 {MachO::N_EINCL
, "N_EINCL"}, {MachO::N_ENTRY
, "N_ENTRY"},
228 {MachO::N_LBRAC
, "N_LBRAC"}, {MachO::N_EXCL
, "N_EXCL"},
229 {MachO::N_RBRAC
, "N_RBRAC"}, {MachO::N_BCOMM
, "N_BCOMM"},
230 {MachO::N_ECOMM
, "N_ECOMM"}, {MachO::N_ECOML
, "N_ECOML"},
231 {MachO::N_LENG
, "N_LENG"}, {0, nullptr}};
233 static const char *getDarwinStabString(uint8_t NType
) {
234 for (unsigned i
= 0; DarwinStabNames
[i
].Name
; i
++) {
235 if (DarwinStabNames
[i
].NType
== NType
)
236 return DarwinStabNames
[i
].Name
;
241 void MachODebugMapParser::dumpSymTabHeader(raw_ostream
&OS
, StringRef Arch
) {
242 OS
<< "-----------------------------------"
243 "-----------------------------------\n";
244 OS
<< "Symbol table for: '" << BinaryPath
<< "' (" << Arch
.data() << ")\n";
245 OS
<< "-----------------------------------"
246 "-----------------------------------\n";
247 OS
<< "Index n_strx n_type n_sect n_desc n_value\n";
248 OS
<< "======== -------- ------------------ ------ ------ ----------------\n";
251 void MachODebugMapParser::dumpSymTabEntry(raw_ostream
&OS
, uint64_t Index
,
252 uint32_t StringIndex
, uint8_t Type
,
253 uint8_t SectionIndex
, uint16_t Flags
,
256 OS
<< '[' << format_decimal(Index
, 6)
259 << format_hex_no_prefix(StringIndex
, 8)
262 << format_hex_no_prefix(Type
, 2) << " (";
264 if (Type
& MachO::N_STAB
)
265 OS
<< left_justify(getDarwinStabString(Type
), 13);
267 if (Type
& MachO::N_PEXT
)
271 switch (Type
& MachO::N_TYPE
) {
272 case MachO::N_UNDF
: // 0x0 undefined, n_sect == NO_SECT
275 case MachO::N_ABS
: // 0x2 absolute, n_sect == NO_SECT
278 case MachO::N_SECT
: // 0xe defined in section number n_sect
281 case MachO::N_PBUD
: // 0xc prebound undefined (defined in a dylib)
284 case MachO::N_INDR
: // 0xa indirect
288 OS
<< format_hex_no_prefix(Type
, 2) << " ";
291 if (Type
& MachO::N_EXT
)
299 << format_hex_no_prefix(SectionIndex
, 2)
302 << format_hex_no_prefix(Flags
, 4)
305 << format_hex_no_prefix(Value
, 16);
307 const char *Name
= &MainBinaryStrings
.data()[StringIndex
];
309 OS
<< " '" << Name
<< "'";
314 void MachODebugMapParser::dumpOneBinaryStab(const MachOObjectFile
&MainBinary
,
315 StringRef BinaryPath
) {
316 loadMainBinarySymbols(MainBinary
);
317 MainBinaryStrings
= MainBinary
.getStringTableData();
318 raw_ostream
&OS(llvm::outs());
320 dumpSymTabHeader(OS
, getArchName(MainBinary
));
322 for (const SymbolRef
&Symbol
: MainBinary
.symbols()) {
323 const DataRefImpl
&DRI
= Symbol
.getRawDataRefImpl();
324 if (MainBinary
.is64Bit())
325 dumpSymTabEntry(OS
, Idx
, MainBinary
.getSymbol64TableEntry(DRI
));
327 dumpSymTabEntry(OS
, Idx
, MainBinary
.getSymbolTableEntry(DRI
));
335 static bool shouldLinkArch(SmallVectorImpl
<StringRef
> &Archs
, StringRef Arch
) {
336 if (Archs
.empty() || is_contained(Archs
, "all") || is_contained(Archs
, "*"))
339 if (Arch
.startswith("arm") && Arch
!= "arm64" && is_contained(Archs
, "arm"))
342 SmallString
<16> ArchName
= Arch
;
343 if (Arch
.startswith("thumb"))
344 ArchName
= ("arm" + Arch
.substr(5)).str();
346 return is_contained(Archs
, ArchName
);
349 bool MachODebugMapParser::dumpStab() {
350 auto ObjectEntry
= BinHolder
.getObjectEntry(BinaryPath
);
352 auto Err
= ObjectEntry
.takeError();
353 WithColor::error() << "cannot load '" << BinaryPath
354 << "': " << toString(std::move(Err
)) << '\n';
358 auto Objects
= ObjectEntry
->getObjectsAs
<MachOObjectFile
>();
360 auto Err
= Objects
.takeError();
361 WithColor::error() << "cannot get '" << BinaryPath
362 << "' as MachO file: " << toString(std::move(Err
))
367 for (const auto *Object
: *Objects
)
368 if (shouldLinkArch(Archs
, Object
->getArchTriple().getArchName()))
369 dumpOneBinaryStab(*Object
, BinaryPath
);
374 /// This main parsing routine tries to open the main binary and if
375 /// successful iterates over the STAB entries. The real parsing is
376 /// done in handleStabSymbolTableEntry.
377 ErrorOr
<std::vector
<std::unique_ptr
<DebugMap
>>> MachODebugMapParser::parse() {
378 auto ObjectEntry
= BinHolder
.getObjectEntry(BinaryPath
);
380 return errorToErrorCode(ObjectEntry
.takeError());
383 auto Objects
= ObjectEntry
->getObjectsAs
<MachOObjectFile
>();
385 return errorToErrorCode(ObjectEntry
.takeError());
388 std::vector
<std::unique_ptr
<DebugMap
>> Results
;
389 for (const auto *Object
: *Objects
)
390 if (shouldLinkArch(Archs
, Object
->getArchTriple().getArchName()))
391 Results
.push_back(parseOneBinary(*Object
, BinaryPath
));
393 return std::move(Results
);
396 /// Interpret the STAB entries to fill the DebugMap.
397 void MachODebugMapParser::handleStabSymbolTableEntry(uint32_t StringIndex
,
399 uint8_t SectionIndex
,
402 if (!(Type
& MachO::N_STAB
))
405 const char *Name
= &MainBinaryStrings
.data()[StringIndex
];
407 // An N_OSO entry represents the start of a new object file description.
408 if (Type
== MachO::N_OSO
)
409 return switchToNewDebugMapObject(Name
, sys::toTimePoint(Value
));
411 if (Type
== MachO::N_AST
) {
412 SmallString
<80> Path(PathPrefix
);
413 sys::path::append(Path
, Name
);
414 Result
->addDebugMapObject(Path
, sys::toTimePoint(Value
), Type
);
418 // If the last N_OSO object file wasn't found, CurrentDebugMapObject will be
419 // null. Do not update anything until we find the next valid N_OSO entry.
420 if (!CurrentDebugMapObject
)
426 // This is a global variable. We need to query the main binary
427 // symbol table to find its address as it might not be in the
428 // debug map (for common symbols).
429 Value
= getMainBinarySymbolAddress(Name
);
432 // Functions are scopes in STABS. They have an end marker that
433 // contains the function size.
434 if (Name
[0] == '\0') {
436 Value
= CurrentFunctionAddress
;
437 Name
= CurrentFunctionName
;
440 CurrentFunctionName
= Name
;
441 CurrentFunctionAddress
= Value
;
450 auto ObjectSymIt
= CurrentObjectAddresses
.find(Name
);
452 // If the name of a (non-static) symbol is not in the current object, we
453 // check all its aliases from the main binary.
454 if (ObjectSymIt
== CurrentObjectAddresses
.end() && Type
!= MachO::N_STSYM
) {
455 for (const auto &Alias
: getMainBinarySymbolNames(Value
)) {
456 ObjectSymIt
= CurrentObjectAddresses
.find(Alias
);
457 if (ObjectSymIt
!= CurrentObjectAddresses
.end())
462 if (ObjectSymIt
== CurrentObjectAddresses
.end()) {
463 Warning("could not find object file symbol for symbol " + Twine(Name
));
467 if (!CurrentDebugMapObject
->addSymbol(Name
, ObjectSymIt
->getValue(), Value
,
469 Warning(Twine("failed to insert symbol '") + Name
+ "' in the debug map.");
474 /// Load the current object file symbols into CurrentObjectAddresses.
475 void MachODebugMapParser::loadCurrentObjectFileSymbols(
476 const object::MachOObjectFile
&Obj
) {
477 CurrentObjectAddresses
.clear();
479 for (auto Sym
: Obj
.symbols()) {
480 uint64_t Addr
= Sym
.getValue();
481 Expected
<StringRef
> Name
= Sym
.getName();
483 // TODO: Actually report errors helpfully.
484 consumeError(Name
.takeError());
487 // The value of some categories of symbols isn't meaningful. For
488 // example common symbols store their size in the value field, not
489 // their address. Absolute symbols have a fixed address that can
490 // conflict with standard symbols. These symbols (especially the
491 // common ones), might still be referenced by relocations. These
492 // relocations will use the symbol itself, and won't need an
493 // object file address. The object file address field is optional
494 // in the DebugMap, leave it unassigned for these symbols.
495 uint32_t Flags
= Sym
.getFlags();
496 if (Flags
& SymbolRef::SF_Absolute
) {
497 CurrentObjectAddresses
[*Name
] = None
;
498 } else if (Flags
& SymbolRef::SF_Common
) {
499 CurrentObjectAddresses
[*Name
] = None
;
500 CommonSymbols
.push_back(*Name
);
502 CurrentObjectAddresses
[*Name
] = Addr
;
507 /// Lookup a symbol address in the main binary symbol table. The
508 /// parser only needs to query common symbols, thus not every symbol's
509 /// address is available through this function.
510 uint64_t MachODebugMapParser::getMainBinarySymbolAddress(StringRef Name
) {
511 auto Sym
= MainBinarySymbolAddresses
.find(Name
);
512 if (Sym
== MainBinarySymbolAddresses
.end())
517 /// Get all symbol names in the main binary for the given value.
518 std::vector
<StringRef
>
519 MachODebugMapParser::getMainBinarySymbolNames(uint64_t Value
) {
520 std::vector
<StringRef
> Names
;
521 for (const auto &Entry
: MainBinarySymbolAddresses
) {
522 if (Entry
.second
== Value
)
523 Names
.push_back(Entry
.first());
528 /// Load the interesting main binary symbols' addresses into
529 /// MainBinarySymbolAddresses.
530 void MachODebugMapParser::loadMainBinarySymbols(
531 const MachOObjectFile
&MainBinary
) {
532 section_iterator Section
= MainBinary
.section_end();
533 MainBinarySymbolAddresses
.clear();
534 for (const auto &Sym
: MainBinary
.symbols()) {
535 Expected
<SymbolRef::Type
> TypeOrErr
= Sym
.getType();
537 // TODO: Actually report errors helpfully.
538 consumeError(TypeOrErr
.takeError());
541 SymbolRef::Type Type
= *TypeOrErr
;
542 // Skip undefined and STAB entries.
543 if ((Type
== SymbolRef::ST_Debug
) || (Type
== SymbolRef::ST_Unknown
))
545 // In theory, the only symbols of interest are the global variables. These
546 // are the only ones that need to be queried because the address of common
547 // data won't be described in the debug map. All other addresses should be
548 // fetched for the debug map. In reality, by playing with 'ld -r' and
549 // export lists, you can get symbols described as N_GSYM in the debug map,
550 // but associated with a local symbol. Gather all the symbols, but prefer
553 MainBinary
.getSymbolTableEntry(Sym
.getRawDataRefImpl()).n_type
;
554 bool Extern
= SymType
& (MachO::N_EXT
| MachO::N_PEXT
);
555 Expected
<section_iterator
> SectionOrErr
= Sym
.getSection();
557 // TODO: Actually report errors helpfully.
558 consumeError(SectionOrErr
.takeError());
561 Section
= *SectionOrErr
;
562 if (Section
== MainBinary
.section_end() || Section
->isText())
564 uint64_t Addr
= Sym
.getValue();
565 Expected
<StringRef
> NameOrErr
= Sym
.getName();
567 // TODO: Actually report errors helpfully.
568 consumeError(NameOrErr
.takeError());
571 StringRef Name
= *NameOrErr
;
572 if (Name
.size() == 0 || Name
[0] == '\0')
574 // Override only if the new key is global.
576 MainBinarySymbolAddresses
[Name
] = Addr
;
578 MainBinarySymbolAddresses
.try_emplace(Name
, Addr
);
584 llvm::ErrorOr
<std::vector
<std::unique_ptr
<DebugMap
>>>
585 parseDebugMap(StringRef InputFile
, ArrayRef
<std::string
> Archs
,
586 StringRef PrependPath
, bool PaperTrailWarnings
, bool Verbose
,
589 return DebugMap::parseYAMLDebugMap(InputFile
, PrependPath
, Verbose
);
591 MachODebugMapParser
Parser(InputFile
, Archs
, PrependPath
, PaperTrailWarnings
,
593 return Parser
.parse();
596 bool dumpStab(StringRef InputFile
, ArrayRef
<std::string
> Archs
,
597 StringRef PrependPath
) {
598 MachODebugMapParser
Parser(InputFile
, Archs
, PrependPath
, false);
599 return Parser
.dumpStab();
601 } // namespace dsymutil