1//===-- sanitizer_symbolizer_libcdep.cpp ----------------------------------===//
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//===----------------------------------------------------------------------===//
8//
9// This file is shared between AddressSanitizer and ThreadSanitizer
10// run-time libraries.
11//===----------------------------------------------------------------------===//
12
13#include "sanitizer_allocator_internal.h"
14#include "sanitizer_internal_defs.h"
15#include "sanitizer_platform.h"
16#include "sanitizer_symbolizer_internal.h"
17
18namespace __sanitizer {
19
20Symbolizer *Symbolizer::GetOrInit() {
21 SpinMutexLock l(&init_mu_);
22 if (symbolizer_)
23 return symbolizer_;
24 symbolizer_ = PlatformInit();
25 CHECK(symbolizer_);
26 return symbolizer_;
27}
28
29// See sanitizer_symbolizer_markup.cpp.
30#if !SANITIZER_SYMBOLIZER_MARKUP
31
32const char *ExtractToken(const char *str, const char *delims, char **result) {
33 uptr prefix_len = internal_strcspn(s: str, reject: delims);
34 *result = (char *)InternalAlloc(size: prefix_len + 1);
35 internal_memcpy(dest: *result, src: str, n: prefix_len);
36 (*result)[prefix_len] = '\0';
37 const char *prefix_end = str + prefix_len;
38 if (*prefix_end != '\0')
39 prefix_end++;
40 return prefix_end;
41}
42
43const char *ExtractInt(const char *str, const char *delims, int *result) {
44 char *buff = nullptr;
45 const char *ret = ExtractToken(str, delims, result: &buff);
46 if (buff) {
47 *result = (int)internal_atoll(nptr: buff);
48 }
49 InternalFree(p: buff);
50 return ret;
51}
52
53const char *ExtractUptr(const char *str, const char *delims, uptr *result) {
54 char *buff = nullptr;
55 const char *ret = ExtractToken(str, delims, result: &buff);
56 if (buff) {
57 *result = (uptr)internal_atoll(nptr: buff);
58 }
59 InternalFree(p: buff);
60 return ret;
61}
62
63const char *ExtractSptr(const char *str, const char *delims, sptr *result) {
64 char *buff = nullptr;
65 const char *ret = ExtractToken(str, delims, result: &buff);
66 if (buff) {
67 *result = (sptr)internal_atoll(nptr: buff);
68 }
69 InternalFree(p: buff);
70 return ret;
71}
72
73const char *ExtractTokenUpToDelimiter(const char *str, const char *delimiter,
74 char **result) {
75 const char *found_delimiter = internal_strstr(haystack: str, needle: delimiter);
76 uptr prefix_len =
77 found_delimiter ? found_delimiter - str : internal_strlen(s: str);
78 *result = (char *)InternalAlloc(size: prefix_len + 1);
79 internal_memcpy(dest: *result, src: str, n: prefix_len);
80 (*result)[prefix_len] = '\0';
81 const char *prefix_end = str + prefix_len;
82 if (*prefix_end != '\0')
83 prefix_end += internal_strlen(s: delimiter);
84 return prefix_end;
85}
86
87SymbolizedStack *Symbolizer::SymbolizePC(uptr addr) {
88 Lock l(&mu_);
89 SymbolizedStack *res = SymbolizedStack::New(addr);
90 auto *mod = FindModuleForAddress(address: addr);
91 if (!mod)
92 return res;
93 // Always fill data about module name and offset.
94 res->info.FillModuleInfo(mod: *mod);
95 for (auto &tool : tools_) {
96 SymbolizerScope sym_scope(this);
97 if (tool.SymbolizePC(addr, stack: res)) {
98 return res;
99 }
100 }
101 return res;
102}
103
104SymbolizedStack* Symbolizer::SymbolizeModuleOffset(const char* module_name,
105 uptr module_offset) {
106 Lock l(&mu_);
107 SymbolizedStack* res = SymbolizedStack::New(addr: module_offset);
108 res->info.FillModuleInfo(mod_name: module_name, mod_offset: module_offset, arch: kModuleArchUnknown);
109 for (auto& tool : tools_) {
110 SymbolizerScope sym_scope(this);
111 if (tool.SymbolizePC(addr: module_offset, stack: res))
112 return res;
113 }
114 return res;
115}
116
117bool Symbolizer::SymbolizeData(uptr addr, DataInfo *info) {
118 Lock l(&mu_);
119 const char *module_name = nullptr;
120 uptr module_offset;
121 ModuleArch arch;
122 if (!FindModuleNameAndOffsetForAddress(address: addr, module_name: &module_name, module_offset: &module_offset,
123 module_arch: &arch))
124 return false;
125 info->Clear();
126 info->module = internal_strdup(s: module_name);
127 info->module_offset = module_offset;
128 info->module_arch = arch;
129 for (auto &tool : tools_) {
130 SymbolizerScope sym_scope(this);
131 if (tool.SymbolizeData(addr, info)) {
132 return true;
133 }
134 }
135 return false;
136}
137
138bool Symbolizer::SymbolizeFrame(uptr addr, FrameInfo *info) {
139 Lock l(&mu_);
140 const char *module_name = nullptr;
141 if (!FindModuleNameAndOffsetForAddress(
142 address: addr, module_name: &module_name, module_offset: &info->module_offset, module_arch: &info->module_arch))
143 return false;
144 info->module = internal_strdup(s: module_name);
145 for (auto &tool : tools_) {
146 SymbolizerScope sym_scope(this);
147 if (tool.SymbolizeFrame(addr, info)) {
148 return true;
149 }
150 }
151 return false;
152}
153
154bool Symbolizer::GetModuleNameAndOffsetForPC(uptr pc, const char **module_name,
155 uptr *module_address) {
156 Lock l(&mu_);
157 const char *internal_module_name = nullptr;
158 ModuleArch arch;
159 if (!FindModuleNameAndOffsetForAddress(address: pc, module_name: &internal_module_name,
160 module_offset: module_address, module_arch: &arch))
161 return false;
162
163 if (module_name)
164 *module_name = module_names_.GetOwnedCopy(str: internal_module_name);
165 return true;
166}
167
168void Symbolizer::Flush() {
169 Lock l(&mu_);
170 for (auto &tool : tools_) {
171 SymbolizerScope sym_scope(this);
172 tool.Flush();
173 }
174}
175
176const char *Symbolizer::Demangle(const char *name) {
177 CHECK(name);
178 Lock l(&mu_);
179 for (auto &tool : tools_) {
180 SymbolizerScope sym_scope(this);
181 if (const char *demangled = tool.Demangle(name))
182 return demangled;
183 }
184 if (const char *demangled = PlatformDemangle(name))
185 return demangled;
186 return name;
187}
188
189bool Symbolizer::FindModuleNameAndOffsetForAddress(uptr address,
190 const char **module_name,
191 uptr *module_offset,
192 ModuleArch *module_arch) {
193 const LoadedModule *module = FindModuleForAddress(address);
194 if (!module)
195 return false;
196 *module_name = module->full_name();
197 *module_offset = address - module->base_address();
198 *module_arch = module->arch();
199 return true;
200}
201
202void Symbolizer::RefreshModules() {
203 modules_.init();
204 fallback_modules_.fallbackInit();
205 RAW_CHECK(modules_.size() > 0);
206 modules_fresh_ = true;
207}
208
209const ListOfModules &Symbolizer::GetRefreshedListOfModules() {
210 if (!modules_fresh_)
211 RefreshModules();
212
213 return modules_;
214}
215
216static const LoadedModule *SearchForModule(const ListOfModules &modules,
217 uptr address) {
218 for (uptr i = 0; i < modules.size(); i++) {
219 if (modules[i].containsAddress(address)) {
220 return &modules[i];
221 }
222 }
223 return nullptr;
224}
225
226const LoadedModule *Symbolizer::FindModuleForAddress(uptr address) {
227 bool modules_were_reloaded = false;
228 if (!modules_fresh_) {
229 RefreshModules();
230 modules_were_reloaded = true;
231 }
232 const LoadedModule *module = SearchForModule(modules: modules_, address);
233 if (module)
234 return module;
235
236 // dlopen/dlclose interceptors invalidate the module list, but when
237 // interception is disabled, we need to retry if the lookup fails in
238 // case the module list changed.
239# if !SANITIZER_INTERCEPT_DLOPEN_DLCLOSE
240 if (!modules_were_reloaded) {
241 RefreshModules();
242 module = SearchForModule(modules: modules_, address);
243 if (module)
244 return module;
245 }
246# endif
247
248 if (fallback_modules_.size()) {
249 module = SearchForModule(modules: fallback_modules_, address);
250 }
251 return module;
252}
253
254// For now we assume the following protocol:
255// For each request of the form
256// <module_name> <module_offset>
257// passed to STDIN, external symbolizer prints to STDOUT response:
258// <function_name>
259// <file_name>:<line_number>:<column_number>
260// <function_name>
261// <file_name>:<line_number>:<column_number>
262// ...
263// <empty line>
264class LLVMSymbolizerProcess final : public SymbolizerProcess {
265 public:
266 explicit LLVMSymbolizerProcess(const char *path)
267 : SymbolizerProcess(path, /*use_posix_spawn=*/SANITIZER_APPLE) {}
268
269 private:
270 bool ReachedEndOfOutput(const char *buffer, uptr length) const override {
271 // Empty line marks the end of llvm-symbolizer output.
272 return length >= 2 && buffer[length - 1] == '\n' &&
273 buffer[length - 2] == '\n';
274 }
275
276 // When adding a new architecture, don't forget to also update
277 // script/asan_symbolize.py and sanitizer_common.h.
278 void GetArgV(const char *path_to_binary,
279 const char *(&argv)[kArgVMax]) const override {
280# if defined(__x86_64h__)
281 const char *const kSymbolizerArch = "--default-arch=x86_64h";
282# elif defined(__x86_64__)
283 const char *const kSymbolizerArch = "--default-arch=x86_64";
284# elif defined(__i386__)
285 const char *const kSymbolizerArch = "--default-arch=i386";
286# elif SANITIZER_LOONGARCH64
287 const char *const kSymbolizerArch = "--default-arch=loongarch64";
288# elif SANITIZER_RISCV64
289 const char *const kSymbolizerArch = "--default-arch=riscv64";
290# elif defined(__aarch64__)
291 const char *const kSymbolizerArch = "--default-arch=arm64";
292# elif defined(__arm__)
293 const char *const kSymbolizerArch = "--default-arch=arm";
294# elif defined(__powerpc64__) && __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
295 const char *const kSymbolizerArch = "--default-arch=powerpc64";
296# elif defined(__powerpc64__) && __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
297 const char *const kSymbolizerArch = "--default-arch=powerpc64le";
298# elif defined(__s390x__)
299 const char *const kSymbolizerArch = "--default-arch=s390x";
300# elif defined(__s390__)
301 const char *const kSymbolizerArch = "--default-arch=s390";
302# else
303 const char *const kSymbolizerArch = "--default-arch=unknown";
304# endif
305
306 const char *const demangle_flag =
307 common_flags()->demangle ? "--demangle" : "--no-demangle";
308 const char *const inline_flag =
309 common_flags()->symbolize_inline_frames ? "--inlines" : "--no-inlines";
310 int i = 0;
311 argv[i++] = path_to_binary;
312 argv[i++] = demangle_flag;
313 argv[i++] = inline_flag;
314 argv[i++] = kSymbolizerArch;
315 argv[i++] = nullptr;
316 CHECK_LE(i, kArgVMax);
317 }
318};
319
320LLVMSymbolizer::LLVMSymbolizer(const char *path, LowLevelAllocator *allocator)
321 : symbolizer_process_(new(*allocator) LLVMSymbolizerProcess(path)) {}
322
323// Parse a <file>:<line>[:<column>] buffer. The file path may contain colons on
324// Windows, so extract tokens from the right hand side first. The column info is
325// also optional.
326static const char *ParseFileLineInfo(AddressInfo *info, const char *str) {
327 char *file_line_info = nullptr;
328 str = ExtractToken(str, delims: "\n", result: &file_line_info);
329 CHECK(file_line_info);
330
331 if (uptr size = internal_strlen(s: file_line_info)) {
332 char *back = file_line_info + size - 1;
333 for (int i = 0; i < 2; ++i) {
334 while (back > file_line_info && IsDigit(c: *back)) --back;
335 if (*back != ':' || !IsDigit(c: back[1]))
336 break;
337 info->column = info->line;
338 info->line = internal_atoll(nptr: back + 1);
339 // Truncate the string at the colon to keep only filename.
340 *back = '\0';
341 --back;
342 }
343 ExtractToken(str: file_line_info, delims: "", result: &info->file);
344 }
345
346 InternalFree(p: file_line_info);
347 return str;
348}
349
350// Parses one or more two-line strings in the following format:
351// <function_name>
352// <file_name>:<line_number>[:<column_number>]
353// Used by LLVMSymbolizer, Addr2LinePool and InternalSymbolizer, since all of
354// them use the same output format.
355void ParseSymbolizePCOutput(const char *str, SymbolizedStack *res) {
356 bool top_frame = true;
357 SymbolizedStack *last = res;
358 while (true) {
359 char *function_name = nullptr;
360 str = ExtractToken(str, delims: "\n", result: &function_name);
361 CHECK(function_name);
362 if (function_name[0] == '\0') {
363 // There are no more frames.
364 InternalFree(p: function_name);
365 break;
366 }
367 SymbolizedStack *cur;
368 if (top_frame) {
369 cur = res;
370 top_frame = false;
371 } else {
372 cur = SymbolizedStack::New(addr: res->info.address);
373 cur->info.FillModuleInfo(mod_name: res->info.module, mod_offset: res->info.module_offset,
374 arch: res->info.module_arch);
375 last->next = cur;
376 last = cur;
377 }
378
379 AddressInfo *info = &cur->info;
380 info->function = function_name;
381 str = ParseFileLineInfo(info, str);
382
383 // Functions and filenames can be "??", in which case we write 0
384 // to address info to mark that names are unknown.
385 if (0 == internal_strcmp(s1: info->function, s2: "??")) {
386 InternalFree(p: info->function);
387 info->function = 0;
388 }
389 if (info->file && 0 == internal_strcmp(s1: info->file, s2: "??")) {
390 InternalFree(p: info->file);
391 info->file = 0;
392 }
393 }
394}
395
396// Parses a two- or three-line string in the following format:
397// <symbol_name>
398// <start_address> <size>
399// <filename>:<column>
400// Used by LLVMSymbolizer and InternalSymbolizer. LLVMSymbolizer added support
401// for symbolizing the third line in D123538, but we support the older two-line
402// information as well.
403void ParseSymbolizeDataOutput(const char *str, DataInfo *info) {
404 str = ExtractToken(str, delims: "\n", result: &info->name);
405 str = ExtractUptr(str, delims: " ", result: &info->start);
406 str = ExtractUptr(str, delims: "\n", result: &info->size);
407 // Note: If the third line isn't present, these calls will set info.{file,
408 // line} to empty strings.
409 str = ExtractToken(str, delims: ":", result: &info->file);
410 str = ExtractUptr(str, delims: "\n", result: &info->line);
411}
412
413void ParseSymbolizeFrameOutput(const char *str,
414 InternalMmapVector<LocalInfo> *locals) {
415 if (internal_strncmp(s1: str, s2: "??", n: 2) == 0)
416 return;
417
418 while (*str) {
419 LocalInfo local;
420 str = ExtractToken(str, delims: "\n", result: &local.function_name);
421 str = ExtractToken(str, delims: "\n", result: &local.name);
422
423 AddressInfo addr;
424 str = ParseFileLineInfo(info: &addr, str);
425 local.decl_file = addr.file;
426 local.decl_line = addr.line;
427
428 local.has_frame_offset = internal_strncmp(s1: str, s2: "??", n: 2) != 0;
429 str = ExtractSptr(str, delims: " ", result: &local.frame_offset);
430
431 local.has_size = internal_strncmp(s1: str, s2: "??", n: 2) != 0;
432 str = ExtractUptr(str, delims: " ", result: &local.size);
433
434 local.has_tag_offset = internal_strncmp(s1: str, s2: "??", n: 2) != 0;
435 str = ExtractUptr(str, delims: "\n", result: &local.tag_offset);
436
437 locals->push_back(element: local);
438 }
439}
440
441bool LLVMSymbolizer::SymbolizePC(uptr addr, SymbolizedStack *stack) {
442 AddressInfo *info = &stack->info;
443 const char *buf = FormatAndSendCommand(
444 command_prefix: "CODE", module_name: info->module, module_offset: info->module_offset, arch: info->module_arch);
445 if (!buf)
446 return false;
447 ParseSymbolizePCOutput(str: buf, res: stack);
448 return true;
449}
450
451bool LLVMSymbolizer::SymbolizeData(uptr addr, DataInfo *info) {
452 const char *buf = FormatAndSendCommand(
453 command_prefix: "DATA", module_name: info->module, module_offset: info->module_offset, arch: info->module_arch);
454 if (!buf)
455 return false;
456 ParseSymbolizeDataOutput(str: buf, info);
457 info->start += (addr - info->module_offset); // Add the base address.
458 return true;
459}
460
461bool LLVMSymbolizer::SymbolizeFrame(uptr addr, FrameInfo *info) {
462 const char *buf = FormatAndSendCommand(
463 command_prefix: "FRAME", module_name: info->module, module_offset: info->module_offset, arch: info->module_arch);
464 if (!buf)
465 return false;
466 ParseSymbolizeFrameOutput(str: buf, locals: &info->locals);
467 return true;
468}
469
470const char *LLVMSymbolizer::FormatAndSendCommand(const char *command_prefix,
471 const char *module_name,
472 uptr module_offset,
473 ModuleArch arch) {
474 CHECK(module_name);
475 int size_needed = 0;
476 if (arch == kModuleArchUnknown)
477 size_needed = internal_snprintf(buffer: buffer_, length: kBufferSize, format: "%s \"%s\" 0x%zx\n",
478 command_prefix, module_name, module_offset);
479 else
480 size_needed = internal_snprintf(
481 buffer: buffer_, length: kBufferSize, format: "%s \"%s:%s\" 0x%zx\n", command_prefix,
482 module_name, ModuleArchToString(arch), module_offset);
483
484 if (size_needed >= static_cast<int>(kBufferSize)) {
485 Report(format: "WARNING: Command buffer too small");
486 return nullptr;
487 }
488
489 return symbolizer_process_->SendCommand(command: buffer_);
490}
491
492SymbolizerProcess::SymbolizerProcess(const char* path, bool use_posix_spawn)
493 : path_(path),
494 input_fd_(kInvalidFd),
495 output_fd_(kInvalidFd),
496 child_stdin_fd_(kInvalidFd),
497 times_restarted_(0),
498 failed_to_start_(false),
499 reported_invalid_path_(false),
500 use_posix_spawn_(use_posix_spawn) {
501 CHECK(path_);
502 CHECK_NE(path_[0], '\0');
503}
504
505SymbolizerProcess::~SymbolizerProcess() {
506 if (child_stdin_fd_ != kInvalidFd)
507 CloseFile(child_stdin_fd_);
508}
509
510static bool IsSameModule(const char *path) {
511 if (const char *ProcessName = GetProcessName()) {
512 if (const char *SymbolizerName = StripModuleName(module: path)) {
513 return !internal_strcmp(s1: ProcessName, s2: SymbolizerName);
514 }
515 }
516 return false;
517}
518
519const char *SymbolizerProcess::SendCommand(const char *command) {
520 if (failed_to_start_)
521 return nullptr;
522 if (IsSameModule(path: path_)) {
523 Report(format: "WARNING: Symbolizer was blocked from starting itself!\n");
524 failed_to_start_ = true;
525 return nullptr;
526 }
527 for (; times_restarted_ < kMaxTimesRestarted; times_restarted_++) {
528 // Start or restart symbolizer if we failed to send command to it.
529 if (const char *res = SendCommandImpl(command))
530 return res;
531 Restart();
532 }
533 if (!failed_to_start_) {
534 Report(format: "WARNING: Failed to use and restart external symbolizer!\n");
535 failed_to_start_ = true;
536 }
537 return nullptr;
538}
539
540const char *SymbolizerProcess::SendCommandImpl(const char *command) {
541 if (input_fd_ == kInvalidFd || output_fd_ == kInvalidFd)
542 return nullptr;
543 if (!WriteToSymbolizer(buffer: command, length: internal_strlen(s: command)))
544 return nullptr;
545 if (!ReadFromSymbolizer())
546 return nullptr;
547 return buffer_.data();
548}
549
550bool SymbolizerProcess::Restart() {
551 if (input_fd_ != kInvalidFd)
552 CloseFile(input_fd_);
553 if (output_fd_ != kInvalidFd)
554 CloseFile(output_fd_);
555 if (child_stdin_fd_ != kInvalidFd) {
556 CloseFile(child_stdin_fd_);
557 child_stdin_fd_ = kInvalidFd; // Don't free in destructor
558 }
559 return StartSymbolizerSubprocess();
560}
561
562bool SymbolizerProcess::ReadFromSymbolizer() {
563 buffer_.clear();
564 constexpr uptr max_length = 1024;
565 bool ret = true;
566 do {
567 uptr just_read = 0;
568 uptr size_before = buffer_.size();
569 buffer_.resize(new_size: size_before + max_length);
570 buffer_.resize(new_size: buffer_.capacity());
571 ret = ReadFromFile(fd: input_fd_, buff: &buffer_[size_before],
572 buff_size: buffer_.size() - size_before, bytes_read: &just_read);
573
574 if (!ret)
575 just_read = 0;
576
577 buffer_.resize(new_size: size_before + just_read);
578
579 // We can't read 0 bytes, as we don't expect external symbolizer to close
580 // its stdout.
581 if (just_read == 0) {
582 Report(format: "WARNING: Can't read from symbolizer at fd %d\n", input_fd_);
583 ret = false;
584 break;
585 }
586 } while (!ReachedEndOfOutput(buffer: buffer_.data(), length: buffer_.size()));
587 buffer_.push_back(element: '\0');
588 return ret;
589}
590
591bool SymbolizerProcess::WriteToSymbolizer(const char *buffer, uptr length) {
592 if (length == 0)
593 return true;
594 uptr write_len = 0;
595 bool success = WriteToFile(fd: output_fd_, buff: buffer, buff_size: length, bytes_written: &write_len);
596 if (!success || write_len != length) {
597 Report(format: "WARNING: Can't write to symbolizer at fd %d\n", output_fd_);
598 return false;
599 }
600 return true;
601}
602
603#endif // !SANITIZER_SYMBOLIZER_MARKUP
604
605} // namespace __sanitizer
606