blob: 72ca7223050740fb0f7a0c219d713b4a614ea99b [file] [log] [blame]
Inna Palantff3f07a2019-07-11 16:15:26 -07001//===-- LLVMSymbolize.cpp -------------------------------------------------===//
2//
Chih-Hung Hsieh08600532019-12-19 15:55:38 -08003// 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
Inna Palantff3f07a2019-07-11 16:15:26 -07006//
7//===----------------------------------------------------------------------===//
8//
9// Implementation for LLVM symbolization library.
10//
11//===----------------------------------------------------------------------===//
12
13#include "llvm/DebugInfo/Symbolize/Symbolize.h"
14
15#include "SymbolizableObjectFile.h"
16
17#include "llvm/ADT/STLExtras.h"
18#include "llvm/BinaryFormat/COFF.h"
Chris Wailese3116c42021-07-13 14:40:48 -070019#include "llvm/Config/config.h"
Inna Palantff3f07a2019-07-11 16:15:26 -070020#include "llvm/DebugInfo/DWARF/DWARFContext.h"
21#include "llvm/DebugInfo/PDB/PDB.h"
22#include "llvm/DebugInfo/PDB/PDBContext.h"
23#include "llvm/Demangle/Demangle.h"
24#include "llvm/Object/COFF.h"
25#include "llvm/Object/MachO.h"
26#include "llvm/Object/MachOUniversal.h"
Chih-Hung Hsieh08600532019-12-19 15:55:38 -080027#include "llvm/Support/CRC.h"
Inna Palantff3f07a2019-07-11 16:15:26 -070028#include "llvm/Support/Casting.h"
29#include "llvm/Support/Compression.h"
30#include "llvm/Support/DataExtractor.h"
31#include "llvm/Support/Errc.h"
32#include "llvm/Support/FileSystem.h"
33#include "llvm/Support/MemoryBuffer.h"
34#include "llvm/Support/Path.h"
35#include <algorithm>
36#include <cassert>
Inna Palantff3f07a2019-07-11 16:15:26 -070037#include <cstring>
38
Inna Palantff3f07a2019-07-11 16:15:26 -070039namespace llvm {
40namespace symbolize {
41
Chris Wailesbcf972c2021-10-21 11:03:28 -070042template <typename T>
Inna Palantff3f07a2019-07-11 16:15:26 -070043Expected<DILineInfo>
Chris Wailesbcf972c2021-10-21 11:03:28 -070044LLVMSymbolizer::symbolizeCodeCommon(const T &ModuleSpecifier,
Chih-Hung Hsieh08600532019-12-19 15:55:38 -080045 object::SectionedAddress ModuleOffset) {
Chris Wailesbcf972c2021-10-21 11:03:28 -070046
47 auto InfoOrErr = getOrCreateModuleInfo(ModuleSpecifier);
48 if (!InfoOrErr)
49 return InfoOrErr.takeError();
50
51 SymbolizableModule *Info = *InfoOrErr;
52
Inna Palantff3f07a2019-07-11 16:15:26 -070053 // A null module means an error has already been reported. Return an empty
54 // result.
55 if (!Info)
56 return DILineInfo();
57
58 // If the user is giving us relative addresses, add the preferred base of the
59 // object to the offset before we do the query. It's what DIContext expects.
60 if (Opts.RelativeAddresses)
Chih-Hung Hsieh08600532019-12-19 15:55:38 -080061 ModuleOffset.Address += Info->getModulePreferredBase();
Inna Palantff3f07a2019-07-11 16:15:26 -070062
Thiébaud Weksteene40e7362020-10-28 15:03:00 +010063 DILineInfo LineInfo = Info->symbolizeCode(
64 ModuleOffset, DILineInfoSpecifier(Opts.PathStyle, Opts.PrintFunctions),
65 Opts.UseSymbolTable);
Inna Palantff3f07a2019-07-11 16:15:26 -070066 if (Opts.Demangle)
67 LineInfo.FunctionName = DemangleName(LineInfo.FunctionName, Info);
68 return LineInfo;
69}
70
Chih-Hung Hsieh08600532019-12-19 15:55:38 -080071Expected<DILineInfo>
72LLVMSymbolizer::symbolizeCode(const ObjectFile &Obj,
73 object::SectionedAddress ModuleOffset) {
Chris Wailesbcf972c2021-10-21 11:03:28 -070074 return symbolizeCodeCommon(Obj, ModuleOffset);
Chih-Hung Hsieh08600532019-12-19 15:55:38 -080075}
76
77Expected<DILineInfo>
78LLVMSymbolizer::symbolizeCode(const std::string &ModuleName,
79 object::SectionedAddress ModuleOffset) {
Chris Wailesbcf972c2021-10-21 11:03:28 -070080 return symbolizeCodeCommon(ModuleName, ModuleOffset);
Chih-Hung Hsieh08600532019-12-19 15:55:38 -080081}
82
Chris Wailesbcf972c2021-10-21 11:03:28 -070083template <typename T>
84Expected<DIInliningInfo> LLVMSymbolizer::symbolizeInlinedCodeCommon(
85 const T &ModuleSpecifier, object::SectionedAddress ModuleOffset) {
86 auto InfoOrErr = getOrCreateModuleInfo(ModuleSpecifier);
87 if (!InfoOrErr)
Inna Palantff3f07a2019-07-11 16:15:26 -070088 return InfoOrErr.takeError();
89
Chris Wailesbcf972c2021-10-21 11:03:28 -070090 SymbolizableModule *Info = *InfoOrErr;
91
Inna Palantff3f07a2019-07-11 16:15:26 -070092 // A null module means an error has already been reported. Return an empty
93 // result.
94 if (!Info)
95 return DIInliningInfo();
96
97 // If the user is giving us relative addresses, add the preferred base of the
98 // object to the offset before we do the query. It's what DIContext expects.
99 if (Opts.RelativeAddresses)
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800100 ModuleOffset.Address += Info->getModulePreferredBase();
Inna Palantff3f07a2019-07-11 16:15:26 -0700101
102 DIInliningInfo InlinedContext = Info->symbolizeInlinedCode(
Thiébaud Weksteene40e7362020-10-28 15:03:00 +0100103 ModuleOffset, DILineInfoSpecifier(Opts.PathStyle, Opts.PrintFunctions),
104 Opts.UseSymbolTable);
Inna Palantff3f07a2019-07-11 16:15:26 -0700105 if (Opts.Demangle) {
106 for (int i = 0, n = InlinedContext.getNumberOfFrames(); i < n; i++) {
107 auto *Frame = InlinedContext.getMutableFrame(i);
108 Frame->FunctionName = DemangleName(Frame->FunctionName, Info);
109 }
110 }
111 return InlinedContext;
112}
113
Chris Wailesbcf972c2021-10-21 11:03:28 -0700114Expected<DIInliningInfo>
115LLVMSymbolizer::symbolizeInlinedCode(const ObjectFile &Obj,
116 object::SectionedAddress ModuleOffset) {
117 return symbolizeInlinedCodeCommon(Obj, ModuleOffset);
118}
119
120Expected<DIInliningInfo>
121LLVMSymbolizer::symbolizeInlinedCode(const std::string &ModuleName,
122 object::SectionedAddress ModuleOffset) {
123 return symbolizeInlinedCodeCommon(ModuleName, ModuleOffset);
124}
125
126template <typename T>
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800127Expected<DIGlobal>
Chris Wailesbcf972c2021-10-21 11:03:28 -0700128LLVMSymbolizer::symbolizeDataCommon(const T &ModuleSpecifier,
129 object::SectionedAddress ModuleOffset) {
130
131 auto InfoOrErr = getOrCreateModuleInfo(ModuleSpecifier);
132 if (!InfoOrErr)
Inna Palantff3f07a2019-07-11 16:15:26 -0700133 return InfoOrErr.takeError();
134
Chris Wailesbcf972c2021-10-21 11:03:28 -0700135 SymbolizableModule *Info = *InfoOrErr;
Inna Palantff3f07a2019-07-11 16:15:26 -0700136 // A null module means an error has already been reported. Return an empty
137 // result.
138 if (!Info)
139 return DIGlobal();
140
141 // If the user is giving us relative addresses, add the preferred base of
142 // the object to the offset before we do the query. It's what DIContext
143 // expects.
144 if (Opts.RelativeAddresses)
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800145 ModuleOffset.Address += Info->getModulePreferredBase();
Inna Palantff3f07a2019-07-11 16:15:26 -0700146
147 DIGlobal Global = Info->symbolizeData(ModuleOffset);
148 if (Opts.Demangle)
149 Global.Name = DemangleName(Global.Name, Info);
150 return Global;
151}
152
Chris Wailesbcf972c2021-10-21 11:03:28 -0700153Expected<DIGlobal>
154LLVMSymbolizer::symbolizeData(const ObjectFile &Obj,
155 object::SectionedAddress ModuleOffset) {
156 return symbolizeDataCommon(Obj, ModuleOffset);
157}
158
159Expected<DIGlobal>
160LLVMSymbolizer::symbolizeData(const std::string &ModuleName,
161 object::SectionedAddress ModuleOffset) {
162 return symbolizeDataCommon(ModuleName, ModuleOffset);
163}
164
165template <typename T>
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800166Expected<std::vector<DILocal>>
Chris Wailesbcf972c2021-10-21 11:03:28 -0700167LLVMSymbolizer::symbolizeFrameCommon(const T &ModuleSpecifier,
168 object::SectionedAddress ModuleOffset) {
169 auto InfoOrErr = getOrCreateModuleInfo(ModuleSpecifier);
170 if (!InfoOrErr)
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800171 return InfoOrErr.takeError();
172
Chris Wailesbcf972c2021-10-21 11:03:28 -0700173 SymbolizableModule *Info = *InfoOrErr;
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800174 // A null module means an error has already been reported. Return an empty
175 // result.
176 if (!Info)
177 return std::vector<DILocal>();
178
179 // If the user is giving us relative addresses, add the preferred base of
180 // the object to the offset before we do the query. It's what DIContext
181 // expects.
182 if (Opts.RelativeAddresses)
183 ModuleOffset.Address += Info->getModulePreferredBase();
184
185 return Info->symbolizeFrame(ModuleOffset);
186}
187
Chris Wailesbcf972c2021-10-21 11:03:28 -0700188Expected<std::vector<DILocal>>
189LLVMSymbolizer::symbolizeFrame(const ObjectFile &Obj,
190 object::SectionedAddress ModuleOffset) {
191 return symbolizeFrameCommon(Obj, ModuleOffset);
192}
193
194Expected<std::vector<DILocal>>
195LLVMSymbolizer::symbolizeFrame(const std::string &ModuleName,
196 object::SectionedAddress ModuleOffset) {
197 return symbolizeFrameCommon(ModuleName, ModuleOffset);
198}
199
Inna Palantff3f07a2019-07-11 16:15:26 -0700200void LLVMSymbolizer::flush() {
201 ObjectForUBPathAndArch.clear();
202 BinaryForPath.clear();
203 ObjectPairForPathArch.clear();
204 Modules.clear();
205}
206
207namespace {
208
209// For Path="/path/to/foo" and Basename="foo" assume that debug info is in
210// /path/to/foo.dSYM/Contents/Resources/DWARF/foo.
211// For Path="/path/to/bar.dSYM" and Basename="foo" assume that debug info is in
212// /path/to/bar.dSYM/Contents/Resources/DWARF/foo.
Chris Wailesbcf972c2021-10-21 11:03:28 -0700213std::string getDarwinDWARFResourceForPath(const std::string &Path,
214 const std::string &Basename) {
Inna Palantff3f07a2019-07-11 16:15:26 -0700215 SmallString<16> ResourceName = StringRef(Path);
216 if (sys::path::extension(Path) != ".dSYM") {
217 ResourceName += ".dSYM";
218 }
219 sys::path::append(ResourceName, "Contents", "Resources", "DWARF");
220 sys::path::append(ResourceName, Basename);
Thiébaud Weksteene40e7362020-10-28 15:03:00 +0100221 return std::string(ResourceName.str());
Inna Palantff3f07a2019-07-11 16:15:26 -0700222}
223
224bool checkFileCRC(StringRef Path, uint32_t CRCHash) {
225 ErrorOr<std::unique_ptr<MemoryBuffer>> MB =
226 MemoryBuffer::getFileOrSTDIN(Path);
227 if (!MB)
228 return false;
Jeff Vander Stoep247d86b2020-08-11 14:27:44 +0200229 return CRCHash == llvm::crc32(arrayRefFromStringRef(MB.get()->getBuffer()));
Inna Palantff3f07a2019-07-11 16:15:26 -0700230}
231
232bool findDebugBinary(const std::string &OrigPath,
233 const std::string &DebuglinkName, uint32_t CRCHash,
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800234 const std::string &FallbackDebugPath,
Inna Palantff3f07a2019-07-11 16:15:26 -0700235 std::string &Result) {
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800236 SmallString<16> OrigDir(OrigPath);
Inna Palantff3f07a2019-07-11 16:15:26 -0700237 llvm::sys::path::remove_filename(OrigDir);
238 SmallString<16> DebugPath = OrigDir;
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800239 // Try relative/path/to/original_binary/debuglink_name
Inna Palantff3f07a2019-07-11 16:15:26 -0700240 llvm::sys::path::append(DebugPath, DebuglinkName);
241 if (checkFileCRC(DebugPath, CRCHash)) {
Thiébaud Weksteene40e7362020-10-28 15:03:00 +0100242 Result = std::string(DebugPath.str());
Inna Palantff3f07a2019-07-11 16:15:26 -0700243 return true;
244 }
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800245 // Try relative/path/to/original_binary/.debug/debuglink_name
Inna Palantff3f07a2019-07-11 16:15:26 -0700246 DebugPath = OrigDir;
247 llvm::sys::path::append(DebugPath, ".debug", DebuglinkName);
248 if (checkFileCRC(DebugPath, CRCHash)) {
Thiébaud Weksteene40e7362020-10-28 15:03:00 +0100249 Result = std::string(DebugPath.str());
Inna Palantff3f07a2019-07-11 16:15:26 -0700250 return true;
251 }
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800252 // Make the path absolute so that lookups will go to
253 // "/usr/lib/debug/full/path/to/debug", not
254 // "/usr/lib/debug/to/debug"
255 llvm::sys::fs::make_absolute(OrigDir);
256 if (!FallbackDebugPath.empty()) {
257 // Try <FallbackDebugPath>/absolute/path/to/original_binary/debuglink_name
258 DebugPath = FallbackDebugPath;
259 } else {
Inna Palantff3f07a2019-07-11 16:15:26 -0700260#if defined(__NetBSD__)
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800261 // Try /usr/libdata/debug/absolute/path/to/original_binary/debuglink_name
262 DebugPath = "/usr/libdata/debug";
Inna Palantff3f07a2019-07-11 16:15:26 -0700263#else
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800264 // Try /usr/lib/debug/absolute/path/to/original_binary/debuglink_name
265 DebugPath = "/usr/lib/debug";
Inna Palantff3f07a2019-07-11 16:15:26 -0700266#endif
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800267 }
Inna Palantff3f07a2019-07-11 16:15:26 -0700268 llvm::sys::path::append(DebugPath, llvm::sys::path::relative_path(OrigDir),
269 DebuglinkName);
270 if (checkFileCRC(DebugPath, CRCHash)) {
Thiébaud Weksteene40e7362020-10-28 15:03:00 +0100271 Result = std::string(DebugPath.str());
Inna Palantff3f07a2019-07-11 16:15:26 -0700272 return true;
273 }
274 return false;
275}
276
277bool getGNUDebuglinkContents(const ObjectFile *Obj, std::string &DebugName,
278 uint32_t &CRCHash) {
279 if (!Obj)
280 return false;
281 for (const SectionRef &Section : Obj->sections()) {
282 StringRef Name;
Jeff Vander Stoep247d86b2020-08-11 14:27:44 +0200283 if (Expected<StringRef> NameOrErr = Section.getName())
284 Name = *NameOrErr;
285 else
286 consumeError(NameOrErr.takeError());
287
Inna Palantff3f07a2019-07-11 16:15:26 -0700288 Name = Name.substr(Name.find_first_not_of("._"));
289 if (Name == "gnu_debuglink") {
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800290 Expected<StringRef> ContentsOrErr = Section.getContents();
291 if (!ContentsOrErr) {
292 consumeError(ContentsOrErr.takeError());
293 return false;
294 }
295 DataExtractor DE(*ContentsOrErr, Obj->isLittleEndian(), 0);
Jeff Vander Stoep247d86b2020-08-11 14:27:44 +0200296 uint64_t Offset = 0;
Inna Palantff3f07a2019-07-11 16:15:26 -0700297 if (const char *DebugNameStr = DE.getCStr(&Offset)) {
298 // 4-byte align the offset.
299 Offset = (Offset + 3) & ~0x3;
300 if (DE.isValidOffsetForDataOfSize(Offset, 4)) {
301 DebugName = DebugNameStr;
302 CRCHash = DE.getU32(&Offset);
303 return true;
304 }
305 }
306 break;
307 }
308 }
309 return false;
310}
311
312bool darwinDsymMatchesBinary(const MachOObjectFile *DbgObj,
313 const MachOObjectFile *Obj) {
314 ArrayRef<uint8_t> dbg_uuid = DbgObj->getUuid();
315 ArrayRef<uint8_t> bin_uuid = Obj->getUuid();
316 if (dbg_uuid.empty() || bin_uuid.empty())
317 return false;
318 return !memcmp(dbg_uuid.data(), bin_uuid.data(), dbg_uuid.size());
319}
320
Jeff Vander Stoep247d86b2020-08-11 14:27:44 +0200321template <typename ELFT>
Chris Wailese3116c42021-07-13 14:40:48 -0700322Optional<ArrayRef<uint8_t>> getBuildID(const ELFFile<ELFT> &Obj) {
323 auto PhdrsOrErr = Obj.program_headers();
Jeff Vander Stoep247d86b2020-08-11 14:27:44 +0200324 if (!PhdrsOrErr) {
325 consumeError(PhdrsOrErr.takeError());
326 return {};
327 }
328 for (const auto &P : *PhdrsOrErr) {
329 if (P.p_type != ELF::PT_NOTE)
330 continue;
331 Error Err = Error::success();
Chris Wailese3116c42021-07-13 14:40:48 -0700332 for (auto N : Obj.notes(P, Err))
Chris Wailesbcf972c2021-10-21 11:03:28 -0700333 if (N.getType() == ELF::NT_GNU_BUILD_ID &&
334 N.getName() == ELF::ELF_NOTE_GNU)
Jeff Vander Stoep247d86b2020-08-11 14:27:44 +0200335 return N.getDesc();
Thiébaud Weksteene40e7362020-10-28 15:03:00 +0100336 consumeError(std::move(Err));
Jeff Vander Stoep247d86b2020-08-11 14:27:44 +0200337 }
338 return {};
339}
340
341Optional<ArrayRef<uint8_t>> getBuildID(const ELFObjectFileBase *Obj) {
342 Optional<ArrayRef<uint8_t>> BuildID;
343 if (auto *O = dyn_cast<ELFObjectFile<ELF32LE>>(Obj))
344 BuildID = getBuildID(O->getELFFile());
345 else if (auto *O = dyn_cast<ELFObjectFile<ELF32BE>>(Obj))
346 BuildID = getBuildID(O->getELFFile());
347 else if (auto *O = dyn_cast<ELFObjectFile<ELF64LE>>(Obj))
348 BuildID = getBuildID(O->getELFFile());
349 else if (auto *O = dyn_cast<ELFObjectFile<ELF64BE>>(Obj))
350 BuildID = getBuildID(O->getELFFile());
351 else
352 llvm_unreachable("unsupported file format");
353 return BuildID;
354}
355
356bool findDebugBinary(const std::vector<std::string> &DebugFileDirectory,
Chris Wailesbcf972c2021-10-21 11:03:28 -0700357 const ArrayRef<uint8_t> BuildID, std::string &Result) {
Jeff Vander Stoep247d86b2020-08-11 14:27:44 +0200358 auto getDebugPath = [&](StringRef Directory) {
359 SmallString<128> Path{Directory};
360 sys::path::append(Path, ".build-id",
361 llvm::toHex(BuildID[0], /*LowerCase=*/true),
362 llvm::toHex(BuildID.slice(1), /*LowerCase=*/true));
363 Path += ".debug";
364 return Path;
365 };
366 if (DebugFileDirectory.empty()) {
367 SmallString<128> Path = getDebugPath(
368#if defined(__NetBSD__)
Chris Wailesbcf972c2021-10-21 11:03:28 -0700369 // Try /usr/libdata/debug/.build-id/../...
370 "/usr/libdata/debug"
Jeff Vander Stoep247d86b2020-08-11 14:27:44 +0200371#else
Chris Wailesbcf972c2021-10-21 11:03:28 -0700372 // Try /usr/lib/debug/.build-id/../...
373 "/usr/lib/debug"
Jeff Vander Stoep247d86b2020-08-11 14:27:44 +0200374#endif
375 );
376 if (llvm::sys::fs::exists(Path)) {
Thiébaud Weksteene40e7362020-10-28 15:03:00 +0100377 Result = std::string(Path.str());
Jeff Vander Stoep247d86b2020-08-11 14:27:44 +0200378 return true;
379 }
380 } else {
381 for (const auto &Directory : DebugFileDirectory) {
382 // Try <debug-file-directory>/.build-id/../...
383 SmallString<128> Path = getDebugPath(Directory);
384 if (llvm::sys::fs::exists(Path)) {
Thiébaud Weksteene40e7362020-10-28 15:03:00 +0100385 Result = std::string(Path.str());
Jeff Vander Stoep247d86b2020-08-11 14:27:44 +0200386 return true;
387 }
388 }
389 }
390 return false;
391}
392
Inna Palantff3f07a2019-07-11 16:15:26 -0700393} // end anonymous namespace
394
395ObjectFile *LLVMSymbolizer::lookUpDsymFile(const std::string &ExePath,
Chris Wailesbcf972c2021-10-21 11:03:28 -0700396 const MachOObjectFile *MachExeObj,
397 const std::string &ArchName) {
Inna Palantff3f07a2019-07-11 16:15:26 -0700398 // On Darwin we may find DWARF in separate object file in
399 // resource directory.
400 std::vector<std::string> DsymPaths;
401 StringRef Filename = sys::path::filename(ExePath);
Thiébaud Weksteene40e7362020-10-28 15:03:00 +0100402 DsymPaths.push_back(
403 getDarwinDWARFResourceForPath(ExePath, std::string(Filename)));
Inna Palantff3f07a2019-07-11 16:15:26 -0700404 for (const auto &Path : Opts.DsymHints) {
Thiébaud Weksteene40e7362020-10-28 15:03:00 +0100405 DsymPaths.push_back(
406 getDarwinDWARFResourceForPath(Path, std::string(Filename)));
Inna Palantff3f07a2019-07-11 16:15:26 -0700407 }
408 for (const auto &Path : DsymPaths) {
409 auto DbgObjOrErr = getOrCreateObject(Path, ArchName);
410 if (!DbgObjOrErr) {
411 // Ignore errors, the file might not exist.
412 consumeError(DbgObjOrErr.takeError());
413 continue;
414 }
415 ObjectFile *DbgObj = DbgObjOrErr.get();
416 if (!DbgObj)
417 continue;
418 const MachOObjectFile *MachDbgObj = dyn_cast<const MachOObjectFile>(DbgObj);
419 if (!MachDbgObj)
420 continue;
421 if (darwinDsymMatchesBinary(MachDbgObj, MachExeObj))
422 return DbgObj;
423 }
424 return nullptr;
425}
426
427ObjectFile *LLVMSymbolizer::lookUpDebuglinkObject(const std::string &Path,
428 const ObjectFile *Obj,
429 const std::string &ArchName) {
430 std::string DebuglinkName;
431 uint32_t CRCHash;
432 std::string DebugBinaryPath;
433 if (!getGNUDebuglinkContents(Obj, DebuglinkName, CRCHash))
434 return nullptr;
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800435 if (!findDebugBinary(Path, DebuglinkName, CRCHash, Opts.FallbackDebugPath,
436 DebugBinaryPath))
Inna Palantff3f07a2019-07-11 16:15:26 -0700437 return nullptr;
438 auto DbgObjOrErr = getOrCreateObject(DebugBinaryPath, ArchName);
439 if (!DbgObjOrErr) {
440 // Ignore errors, the file might not exist.
441 consumeError(DbgObjOrErr.takeError());
442 return nullptr;
443 }
444 return DbgObjOrErr.get();
445}
446
Jeff Vander Stoep247d86b2020-08-11 14:27:44 +0200447ObjectFile *LLVMSymbolizer::lookUpBuildIDObject(const std::string &Path,
448 const ELFObjectFileBase *Obj,
449 const std::string &ArchName) {
450 auto BuildID = getBuildID(Obj);
451 if (!BuildID)
452 return nullptr;
453 if (BuildID->size() < 2)
454 return nullptr;
455 std::string DebugBinaryPath;
456 if (!findDebugBinary(Opts.DebugFileDirectory, *BuildID, DebugBinaryPath))
457 return nullptr;
458 auto DbgObjOrErr = getOrCreateObject(DebugBinaryPath, ArchName);
459 if (!DbgObjOrErr) {
460 consumeError(DbgObjOrErr.takeError());
461 return nullptr;
462 }
463 return DbgObjOrErr.get();
464}
465
Inna Palantff3f07a2019-07-11 16:15:26 -0700466Expected<LLVMSymbolizer::ObjectPair>
467LLVMSymbolizer::getOrCreateObjectPair(const std::string &Path,
468 const std::string &ArchName) {
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800469 auto I = ObjectPairForPathArch.find(std::make_pair(Path, ArchName));
470 if (I != ObjectPairForPathArch.end())
Inna Palantff3f07a2019-07-11 16:15:26 -0700471 return I->second;
Inna Palantff3f07a2019-07-11 16:15:26 -0700472
473 auto ObjOrErr = getOrCreateObject(Path, ArchName);
474 if (!ObjOrErr) {
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800475 ObjectPairForPathArch.emplace(std::make_pair(Path, ArchName),
476 ObjectPair(nullptr, nullptr));
Inna Palantff3f07a2019-07-11 16:15:26 -0700477 return ObjOrErr.takeError();
478 }
479
480 ObjectFile *Obj = ObjOrErr.get();
481 assert(Obj != nullptr);
482 ObjectFile *DbgObj = nullptr;
483
484 if (auto MachObj = dyn_cast<const MachOObjectFile>(Obj))
485 DbgObj = lookUpDsymFile(Path, MachObj, ArchName);
Jeff Vander Stoep247d86b2020-08-11 14:27:44 +0200486 else if (auto ELFObj = dyn_cast<const ELFObjectFileBase>(Obj))
487 DbgObj = lookUpBuildIDObject(Path, ELFObj, ArchName);
Inna Palantff3f07a2019-07-11 16:15:26 -0700488 if (!DbgObj)
489 DbgObj = lookUpDebuglinkObject(Path, Obj, ArchName);
490 if (!DbgObj)
491 DbgObj = Obj;
492 ObjectPair Res = std::make_pair(Obj, DbgObj);
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800493 ObjectPairForPathArch.emplace(std::make_pair(Path, ArchName), Res);
Inna Palantff3f07a2019-07-11 16:15:26 -0700494 return Res;
495}
496
497Expected<ObjectFile *>
498LLVMSymbolizer::getOrCreateObject(const std::string &Path,
499 const std::string &ArchName) {
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800500 Binary *Bin;
501 auto Pair = BinaryForPath.emplace(Path, OwningBinary<Binary>());
502 if (!Pair.second) {
503 Bin = Pair.first->second.getBinary();
Inna Palantff3f07a2019-07-11 16:15:26 -0700504 } else {
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800505 Expected<OwningBinary<Binary>> BinOrErr = createBinary(Path);
506 if (!BinOrErr)
507 return BinOrErr.takeError();
508 Pair.first->second = std::move(BinOrErr.get());
509 Bin = Pair.first->second.getBinary();
Inna Palantff3f07a2019-07-11 16:15:26 -0700510 }
511
512 if (!Bin)
513 return static_cast<ObjectFile *>(nullptr);
514
515 if (MachOUniversalBinary *UB = dyn_cast_or_null<MachOUniversalBinary>(Bin)) {
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800516 auto I = ObjectForUBPathAndArch.find(std::make_pair(Path, ArchName));
517 if (I != ObjectForUBPathAndArch.end())
Inna Palantff3f07a2019-07-11 16:15:26 -0700518 return I->second.get();
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800519
Inna Palantff3f07a2019-07-11 16:15:26 -0700520 Expected<std::unique_ptr<ObjectFile>> ObjOrErr =
Jeff Vander Stoep247d86b2020-08-11 14:27:44 +0200521 UB->getMachOObjectForArch(ArchName);
Inna Palantff3f07a2019-07-11 16:15:26 -0700522 if (!ObjOrErr) {
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800523 ObjectForUBPathAndArch.emplace(std::make_pair(Path, ArchName),
524 std::unique_ptr<ObjectFile>());
Inna Palantff3f07a2019-07-11 16:15:26 -0700525 return ObjOrErr.takeError();
526 }
527 ObjectFile *Res = ObjOrErr->get();
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800528 ObjectForUBPathAndArch.emplace(std::make_pair(Path, ArchName),
529 std::move(ObjOrErr.get()));
Inna Palantff3f07a2019-07-11 16:15:26 -0700530 return Res;
531 }
532 if (Bin->isObject()) {
533 return cast<ObjectFile>(Bin);
534 }
535 return errorCodeToError(object_error::arch_not_found);
536}
537
538Expected<SymbolizableModule *>
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800539LLVMSymbolizer::createModuleInfo(const ObjectFile *Obj,
540 std::unique_ptr<DIContext> Context,
541 StringRef ModuleName) {
Jeff Vander Stoep247d86b2020-08-11 14:27:44 +0200542 auto InfoOrErr = SymbolizableObjectFile::create(Obj, std::move(Context),
543 Opts.UntagAddresses);
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800544 std::unique_ptr<SymbolizableModule> SymMod;
545 if (InfoOrErr)
546 SymMod = std::move(*InfoOrErr);
Thiébaud Weksteene40e7362020-10-28 15:03:00 +0100547 auto InsertResult = Modules.insert(
548 std::make_pair(std::string(ModuleName), std::move(SymMod)));
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800549 assert(InsertResult.second);
Thiébaud Weksteene40e7362020-10-28 15:03:00 +0100550 if (!InfoOrErr)
551 return InfoOrErr.takeError();
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800552 return InsertResult.first->second.get();
553}
554
555Expected<SymbolizableModule *>
556LLVMSymbolizer::getOrCreateModuleInfo(const std::string &ModuleName) {
557 auto I = Modules.find(ModuleName);
558 if (I != Modules.end())
Inna Palantff3f07a2019-07-11 16:15:26 -0700559 return I->second.get();
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800560
Inna Palantff3f07a2019-07-11 16:15:26 -0700561 std::string BinaryName = ModuleName;
562 std::string ArchName = Opts.DefaultArch;
563 size_t ColonPos = ModuleName.find_last_of(':');
564 // Verify that substring after colon form a valid arch name.
565 if (ColonPos != std::string::npos) {
566 std::string ArchStr = ModuleName.substr(ColonPos + 1);
567 if (Triple(ArchStr).getArch() != Triple::UnknownArch) {
568 BinaryName = ModuleName.substr(0, ColonPos);
569 ArchName = ArchStr;
570 }
571 }
572 auto ObjectsOrErr = getOrCreateObjectPair(BinaryName, ArchName);
573 if (!ObjectsOrErr) {
574 // Failed to find valid object file.
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800575 Modules.emplace(ModuleName, std::unique_ptr<SymbolizableModule>());
Inna Palantff3f07a2019-07-11 16:15:26 -0700576 return ObjectsOrErr.takeError();
577 }
578 ObjectPair Objects = ObjectsOrErr.get();
579
580 std::unique_ptr<DIContext> Context;
581 // If this is a COFF object containing PDB info, use a PDBContext to
582 // symbolize. Otherwise, use DWARF.
583 if (auto CoffObject = dyn_cast<COFFObjectFile>(Objects.first)) {
584 const codeview::DebugInfo *DebugInfo;
585 StringRef PDBFileName;
586 auto EC = CoffObject->getDebugPDBInfo(DebugInfo, PDBFileName);
587 if (!EC && DebugInfo != nullptr && !PDBFileName.empty()) {
588 using namespace pdb;
589 std::unique_ptr<IPDBSession> Session;
Chris Wailese3116c42021-07-13 14:40:48 -0700590
591 PDB_ReaderType ReaderType =
592 Opts.UseDIA ? PDB_ReaderType::DIA : PDB_ReaderType::Native;
Thiébaud Weksteene40e7362020-10-28 15:03:00 +0100593 if (auto Err = loadDataForEXE(ReaderType, Objects.first->getFileName(),
594 Session)) {
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800595 Modules.emplace(ModuleName, std::unique_ptr<SymbolizableModule>());
Inna Palantff3f07a2019-07-11 16:15:26 -0700596 // Return along the PDB filename to provide more context
597 return createFileError(PDBFileName, std::move(Err));
598 }
599 Context.reset(new PDBContext(*CoffObject, std::move(Session)));
600 }
601 }
602 if (!Context)
Thiébaud Weksteene40e7362020-10-28 15:03:00 +0100603 Context = DWARFContext::create(*Objects.second, nullptr, Opts.DWPName);
Chih-Hung Hsieh08600532019-12-19 15:55:38 -0800604 return createModuleInfo(Objects.first, std::move(Context), ModuleName);
Inna Palantff3f07a2019-07-11 16:15:26 -0700605}
606
Chris Wailesbcf972c2021-10-21 11:03:28 -0700607Expected<SymbolizableModule *>
608LLVMSymbolizer::getOrCreateModuleInfo(const ObjectFile &Obj) {
609 StringRef ObjName = Obj.getFileName();
610 auto I = Modules.find(ObjName);
611 if (I != Modules.end())
612 return I->second.get();
613
614 std::unique_ptr<DIContext> Context = DWARFContext::create(Obj);
615 // FIXME: handle COFF object with PDB info to use PDBContext
616 return createModuleInfo(&Obj, std::move(Context), ObjName);
617}
618
Inna Palantff3f07a2019-07-11 16:15:26 -0700619namespace {
620
621// Undo these various manglings for Win32 extern "C" functions:
622// cdecl - _foo
623// stdcall - _foo@12
624// fastcall - @foo@12
625// vectorcall - foo@@12
626// These are all different linkage names for 'foo'.
627StringRef demanglePE32ExternCFunc(StringRef SymbolName) {
628 // Remove any '_' or '@' prefix.
629 char Front = SymbolName.empty() ? '\0' : SymbolName[0];
630 if (Front == '_' || Front == '@')
631 SymbolName = SymbolName.drop_front();
632
633 // Remove any '@[0-9]+' suffix.
634 if (Front != '?') {
635 size_t AtPos = SymbolName.rfind('@');
636 if (AtPos != StringRef::npos &&
Chris Wailese3116c42021-07-13 14:40:48 -0700637 all_of(drop_begin(SymbolName, AtPos + 1), isDigit))
Inna Palantff3f07a2019-07-11 16:15:26 -0700638 SymbolName = SymbolName.substr(0, AtPos);
Inna Palantff3f07a2019-07-11 16:15:26 -0700639 }
640
641 // Remove any ending '@' for vectorcall.
642 if (SymbolName.endswith("@"))
643 SymbolName = SymbolName.drop_back();
644
645 return SymbolName;
646}
647
648} // end anonymous namespace
649
650std::string
651LLVMSymbolizer::DemangleName(const std::string &Name,
652 const SymbolizableModule *DbiModuleDescriptor) {
653 // We can spoil names of symbols with C linkage, so use an heuristic
654 // approach to check if the name should be demangled.
655 if (Name.substr(0, 2) == "_Z") {
656 int status = 0;
Chris Wailesbcf972c2021-10-21 11:03:28 -0700657 char *DemangledName =
658 itaniumDemangle(Name.c_str(), nullptr, nullptr, &status);
Inna Palantff3f07a2019-07-11 16:15:26 -0700659 if (status != 0)
660 return Name;
661 std::string Result = DemangledName;
662 free(DemangledName);
663 return Result;
664 }
665
Inna Palantff3f07a2019-07-11 16:15:26 -0700666 if (!Name.empty() && Name.front() == '?') {
667 // Only do MSVC C++ demangling on symbols starting with '?'.
Jeff Vander Stoep247d86b2020-08-11 14:27:44 +0200668 int status = 0;
669 char *DemangledName = microsoftDemangle(
Thiébaud Weksteene40e7362020-10-28 15:03:00 +0100670 Name.c_str(), nullptr, nullptr, nullptr, &status,
Jeff Vander Stoep247d86b2020-08-11 14:27:44 +0200671 MSDemangleFlags(MSDF_NoAccessSpecifier | MSDF_NoCallingConvention |
672 MSDF_NoMemberType | MSDF_NoReturnType));
673 if (status != 0)
674 return Name;
675 std::string Result = DemangledName;
676 free(DemangledName);
677 return Result;
Inna Palantff3f07a2019-07-11 16:15:26 -0700678 }
Jeff Vander Stoep247d86b2020-08-11 14:27:44 +0200679
Inna Palantff3f07a2019-07-11 16:15:26 -0700680 if (DbiModuleDescriptor && DbiModuleDescriptor->isWin32Module())
681 return std::string(demanglePE32ExternCFunc(Name));
682 return Name;
683}
684
685} // namespace symbolize
686} // namespace llvm