blob: 9ade8b4ebca9c4463ed756628614a10933557bad [file] [log] [blame]
Florian Mayerd12a6de2019-08-01 10:36:11 +01001/*
2 * Copyright (C) 2019 The Android Open Source Project
3 *
4 * Licensed under the Apache License, Version 2.0 (the "License");
5 * you may not use this file except in compliance with the License.
6 * You may obtain a copy of the License at
7 *
8 * http://www.apache.org/licenses/LICENSE-2.0
9 *
10 * Unless required by applicable law or agreed to in writing, software
11 * distributed under the License is distributed on an "AS IS" BASIS,
12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13 * See the License for the specific language governing permissions and
14 * limitations under the License.
15 */
16
Florian Mayerbd88ff82019-12-18 16:36:37 +000017#include "src/profiling/symbolizer/local_symbolizer.h"
Florian Mayerd12a6de2019-08-01 10:36:11 +010018
Florian Mayer6d1f0ae2020-07-21 08:59:29 +010019#include <fcntl.h>
20
Primiano Tucci58d2dc62021-06-24 16:03:24 +010021#include <cinttypes>
Florian Mayer6d1f0ae2020-07-21 08:59:29 +010022#include <memory>
Lalit Maganti4e2303c2023-03-29 15:28:36 +010023#include <optional>
Joshua Gilpatrickcd349942020-11-11 15:18:37 -080024#include <sstream>
Florian Mayer6d1f0ae2020-07-21 08:59:29 +010025#include <string>
26#include <vector>
27
28#include "perfetto/base/build_config.h"
29#include "perfetto/base/compiler.h"
30#include "perfetto/base/logging.h"
Primiano Tucciab293f52020-12-08 11:46:52 +010031#include "perfetto/ext/base/file_utils.h"
Florian Mayer6d1f0ae2020-07-21 08:59:29 +010032#include "perfetto/ext/base/scoped_file.h"
Hector Dearmanaa025742021-01-14 13:55:22 +000033#include "perfetto/ext/base/string_utils.h"
Daniele Di Proietto25f80692021-09-17 10:14:41 +000034#include "src/profiling/symbolizer/elf.h"
Daniele Di Proiettocb426002023-02-16 12:14:38 +000035#include "src/profiling/symbolizer/filesystem.h"
Joshua Gilpatrickcd349942020-11-11 15:18:37 -080036#include "src/profiling/symbolizer/scoped_read_mmap.h"
Florian Mayer6d1f0ae2020-07-21 08:59:29 +010037
38namespace perfetto {
39namespace profiling {
40
Florian Mayer98f7a932020-12-11 16:31:19 +000041// TODO(fmayer): Fix up name. This suggests it always returns a symbolizer or
42// dies, which isn't the case.
Florian Mayer6d1f0ae2020-07-21 08:59:29 +010043std::unique_ptr<Symbolizer> LocalSymbolizerOrDie(
44 std::vector<std::string> binary_path,
45 const char* mode) {
46 std::unique_ptr<Symbolizer> symbolizer;
47
48 if (!binary_path.empty()) {
49#if PERFETTO_BUILDFLAG(PERFETTO_LOCAL_SYMBOLIZER)
50 std::unique_ptr<BinaryFinder> finder;
51 if (!mode || strncmp(mode, "find", 4) == 0)
52 finder.reset(new LocalBinaryFinder(std::move(binary_path)));
53 else if (strncmp(mode, "index", 5) == 0)
54 finder.reset(new LocalBinaryIndexer(std::move(binary_path)));
55 else
56 PERFETTO_FATAL("Invalid symbolizer mode [find | index]: %s", mode);
57 symbolizer.reset(new LocalSymbolizer(std::move(finder)));
58#else
59 base::ignore_result(mode);
60 PERFETTO_FATAL("This build does not support local symbolization.");
61#endif
62 }
63 return symbolizer;
64}
65
66} // namespace profiling
67} // namespace perfetto
68
Florian Mayer6d1f0ae2020-07-21 08:59:29 +010069#if PERFETTO_BUILDFLAG(PERFETTO_LOCAL_SYMBOLIZER)
Florian Mayerd12a6de2019-08-01 10:36:11 +010070#include "perfetto/ext/base/string_splitter.h"
Hector Dearman5f069382019-09-16 10:24:01 +010071#include "perfetto/ext/base/string_utils.h"
Florian Mayerd12a6de2019-08-01 10:36:11 +010072#include "perfetto/ext/base/utils.h"
73
Florian Mayer4c955702020-06-10 20:37:38 +020074#include <signal.h>
Florian Mayerd12a6de2019-08-01 10:36:11 +010075#include <sys/stat.h>
76#include <sys/types.h>
Joshua Gilpatrickcd349942020-11-11 15:18:37 -080077
78#if PERFETTO_BUILDFLAG(PERFETTO_OS_WIN)
Florian Mayer79eb2122020-12-10 17:09:09 +000079constexpr const char* kDefaultSymbolizer = "llvm-symbolizer.exe";
80#else
81constexpr const char* kDefaultSymbolizer = "llvm-symbolizer";
Joshua Gilpatrickcd349942020-11-11 15:18:37 -080082#endif
Florian Mayerd12a6de2019-08-01 10:36:11 +010083
84namespace perfetto {
Florian Mayerbd88ff82019-12-18 16:36:37 +000085namespace profiling {
Florian Mayerd12a6de2019-08-01 10:36:11 +010086
Joshua Gilpatrickcd349942020-11-11 15:18:37 -080087std::vector<std::string> GetLines(
88 std::function<int64_t(char*, size_t)> fn_read) {
Florian Mayerd12a6de2019-08-01 10:36:11 +010089 std::vector<std::string> lines;
Joshua Gilpatrickcd349942020-11-11 15:18:37 -080090 char buffer[512];
91 int64_t rd = 0;
92 // Cache the partial line of the previous read.
93 std::string last_line;
94 while ((rd = fn_read(buffer, sizeof(buffer))) > 0) {
95 std::string data(buffer, static_cast<size_t>(rd));
96 // Create stream buffer of last partial line + new data
97 std::stringstream stream(last_line + data);
98 std::string line;
99 last_line = "";
100 while (std::getline(stream, line)) {
101 // Return from reading when we read an empty line.
102 if (line.empty()) {
103 return lines;
104 } else if (stream.eof()) {
105 // Cache off the partial line when we hit end of stream.
106 last_line += line;
107 break;
108 } else {
109 lines.push_back(line);
110 }
Florian Mayerd12a6de2019-08-01 10:36:11 +0100111 }
Joshua Gilpatrickcd349942020-11-11 15:18:37 -0800112 }
113 if (rd == -1) {
114 PERFETTO_ELOG("Failed to read data from subprocess.");
115 }
Florian Mayerd12a6de2019-08-01 10:36:11 +0100116 return lines;
Florian Mayer39179192019-09-09 16:15:22 +0100117}
118
Joshua Gilpatrickcd349942020-11-11 15:18:37 -0800119namespace {
Florian Mayerd12a6de2019-08-01 10:36:11 +0100120bool InRange(const void* base,
121 size_t total_size,
122 const void* ptr,
123 size_t size) {
124 return ptr >= base && static_cast<const char*>(ptr) + size <=
125 static_cast<const char*>(base) + total_size;
126}
127
128template <typename E>
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100129std::optional<uint64_t> GetLoadBias(void* mem, size_t size) {
Florian Mayere7228a72020-10-22 18:18:01 +0100130 const typename E::Ehdr* ehdr = static_cast<typename E::Ehdr*>(mem);
131 if (!InRange(mem, size, ehdr, sizeof(typename E::Ehdr))) {
132 PERFETTO_ELOG("Corrupted ELF.");
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100133 return std::nullopt;
Florian Mayere7228a72020-10-22 18:18:01 +0100134 }
135 for (size_t i = 0; i < ehdr->e_phnum; ++i) {
136 typename E::Phdr* phdr = GetPhdr<E>(mem, ehdr, i);
137 if (!InRange(mem, size, phdr, sizeof(typename E::Phdr))) {
138 PERFETTO_ELOG("Corrupted ELF.");
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100139 return std::nullopt;
Florian Mayere7228a72020-10-22 18:18:01 +0100140 }
141 if (phdr->p_type == PT_LOAD && phdr->p_flags & PF_X) {
142 return phdr->p_vaddr - phdr->p_offset;
143 }
144 }
145 return 0u;
146}
147
148template <typename E>
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100149std::optional<std::string> GetBuildId(void* mem, size_t size) {
Florian Mayerd12a6de2019-08-01 10:36:11 +0100150 const typename E::Ehdr* ehdr = static_cast<typename E::Ehdr*>(mem);
151 if (!InRange(mem, size, ehdr, sizeof(typename E::Ehdr))) {
152 PERFETTO_ELOG("Corrupted ELF.");
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100153 return std::nullopt;
Florian Mayerd12a6de2019-08-01 10:36:11 +0100154 }
155 for (size_t i = 0; i < ehdr->e_shnum; ++i) {
156 typename E::Shdr* shdr = GetShdr<E>(mem, ehdr, i);
157 if (!InRange(mem, size, shdr, sizeof(typename E::Shdr))) {
158 PERFETTO_ELOG("Corrupted ELF.");
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100159 return std::nullopt;
Florian Mayerd12a6de2019-08-01 10:36:11 +0100160 }
161
162 if (shdr->sh_type != SHT_NOTE)
163 continue;
164
165 auto offset = shdr->sh_offset;
166 while (offset < shdr->sh_offset + shdr->sh_size) {
167 typename E::Nhdr* nhdr =
168 reinterpret_cast<typename E::Nhdr*>(static_cast<char*>(mem) + offset);
169
170 if (!InRange(mem, size, nhdr, sizeof(typename E::Nhdr))) {
171 PERFETTO_ELOG("Corrupted ELF.");
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100172 return std::nullopt;
Florian Mayerd12a6de2019-08-01 10:36:11 +0100173 }
174 if (nhdr->n_type == NT_GNU_BUILD_ID && nhdr->n_namesz == 4) {
175 char* name = reinterpret_cast<char*>(nhdr) + sizeof(*nhdr);
176 if (!InRange(mem, size, name, 4)) {
177 PERFETTO_ELOG("Corrupted ELF.");
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100178 return std::nullopt;
Florian Mayerd12a6de2019-08-01 10:36:11 +0100179 }
180 if (memcmp(name, "GNU", 3) == 0) {
181 const char* value = reinterpret_cast<char*>(nhdr) + sizeof(*nhdr) +
182 base::AlignUp<4>(nhdr->n_namesz);
183
184 if (!InRange(mem, size, value, nhdr->n_descsz)) {
185 PERFETTO_ELOG("Corrupted ELF.");
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100186 return std::nullopt;
Florian Mayerd12a6de2019-08-01 10:36:11 +0100187 }
188 return std::string(value, nhdr->n_descsz);
189 }
190 }
191 offset += sizeof(*nhdr) + base::AlignUp<4>(nhdr->n_namesz) +
192 base::AlignUp<4>(nhdr->n_descsz);
193 }
194 }
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100195 return std::nullopt;
Florian Mayerd12a6de2019-08-01 10:36:11 +0100196}
197
Florian Mayer39179192019-09-09 16:15:22 +0100198std::string SplitBuildID(const std::string& hex_build_id) {
199 if (hex_build_id.size() < 3) {
200 PERFETTO_DFATAL_OR_ELOG("Invalid build-id (< 3 char) %s",
201 hex_build_id.c_str());
202 return {};
203 }
204
205 return hex_build_id.substr(0, 2) + "/" + hex_build_id.substr(2);
206}
207
Florian Mayer6d1f0ae2020-07-21 08:59:29 +0100208bool IsElf(const char* mem, size_t size) {
209 if (size <= EI_MAG3)
210 return false;
211 return (mem[EI_MAG0] == ELFMAG0 && mem[EI_MAG1] == ELFMAG1 &&
212 mem[EI_MAG2] == ELFMAG2 && mem[EI_MAG3] == ELFMAG3);
213}
214
Florian Mayere7228a72020-10-22 18:18:01 +0100215struct BuildIdAndLoadBias {
216 std::string build_id;
217 uint64_t load_bias;
218};
219
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100220std::optional<BuildIdAndLoadBias> GetBuildIdAndLoadBias(const char* fname,
221 size_t size) {
Daniele Di Proiettoee909282021-09-20 18:26:16 +0000222 static_assert(EI_CLASS > EI_MAG3, "mem[EI_MAG?] accesses are in range.");
Daniele Di Proiettocb426002023-02-16 12:14:38 +0000223 if (size <= EI_CLASS)
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100224 return std::nullopt;
Daniele Di Proiettocb426002023-02-16 12:14:38 +0000225 ScopedReadMmap map(fname, size);
Joshua Gilpatrickcd349942020-11-11 15:18:37 -0800226 if (!map.IsValid()) {
Florian Mayer6d1f0ae2020-07-21 08:59:29 +0100227 PERFETTO_PLOG("mmap");
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100228 return std::nullopt;
Florian Mayer6d1f0ae2020-07-21 08:59:29 +0100229 }
230 char* mem = static_cast<char*>(*map);
231
Daniele Di Proiettocb426002023-02-16 12:14:38 +0000232 if (!IsElf(mem, size))
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100233 return std::nullopt;
Florian Mayer6d1f0ae2020-07-21 08:59:29 +0100234
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100235 std::optional<std::string> build_id;
236 std::optional<uint64_t> load_bias;
Florian Mayer6d1f0ae2020-07-21 08:59:29 +0100237 switch (mem[EI_CLASS]) {
238 case ELFCLASS32:
Daniele Di Proiettocb426002023-02-16 12:14:38 +0000239 build_id = GetBuildId<Elf32>(mem, size);
240 load_bias = GetLoadBias<Elf32>(mem, size);
Florian Mayere7228a72020-10-22 18:18:01 +0100241 break;
Florian Mayer6d1f0ae2020-07-21 08:59:29 +0100242 case ELFCLASS64:
Daniele Di Proiettocb426002023-02-16 12:14:38 +0000243 build_id = GetBuildId<Elf64>(mem, size);
244 load_bias = GetLoadBias<Elf64>(mem, size);
Florian Mayere7228a72020-10-22 18:18:01 +0100245 break;
Florian Mayer6d1f0ae2020-07-21 08:59:29 +0100246 default:
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100247 return std::nullopt;
Florian Mayer6d1f0ae2020-07-21 08:59:29 +0100248 }
Florian Mayere7228a72020-10-22 18:18:01 +0100249 if (build_id && load_bias) {
250 return BuildIdAndLoadBias{*build_id, *load_bias};
251 }
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100252 return std::nullopt;
Florian Mayer6d1f0ae2020-07-21 08:59:29 +0100253}
254
Florian Mayere7228a72020-10-22 18:18:01 +0100255std::map<std::string, FoundBinary> BuildIdIndex(std::vector<std::string> dirs) {
256 std::map<std::string, FoundBinary> result;
Daniele Di Proiettocb426002023-02-16 12:14:38 +0000257 WalkDirectories(std::move(dirs), [&result](const char* fname, size_t size) {
258 char magic[EI_MAG3 + 1];
259 // Scope file access. On windows OpenFile opens an exclusive lock.
260 // This lock needs to be released before mapping the file.
261 {
262 base::ScopedFile fd(base::OpenFile(fname, O_RDONLY));
263 if (!fd) {
264 PERFETTO_PLOG("Failed to open %s", fname);
265 return;
Joshua Gilpatrickcd349942020-11-11 15:18:37 -0800266 }
Daniele Di Proiettocb426002023-02-16 12:14:38 +0000267 ssize_t rd = base::Read(*fd, &magic, sizeof(magic));
268 if (rd != sizeof(magic)) {
269 PERFETTO_PLOG("Failed to read %s", fname);
270 return;
271 }
272 if (!IsElf(magic, static_cast<size_t>(rd))) {
273 PERFETTO_DLOG("%s not an ELF.", fname);
274 return;
Joshua Gilpatrickcd349942020-11-11 15:18:37 -0800275 }
276 }
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100277 std::optional<BuildIdAndLoadBias> build_id_and_load_bias =
Daniele Di Proiettocb426002023-02-16 12:14:38 +0000278 GetBuildIdAndLoadBias(fname, size);
279 if (build_id_and_load_bias) {
280 result.emplace(build_id_and_load_bias->build_id,
281 FoundBinary{fname, build_id_and_load_bias->load_bias});
282 }
283 });
Florian Mayer6d1f0ae2020-07-21 08:59:29 +0100284 return result;
285}
286
Florian Mayerd12a6de2019-08-01 10:36:11 +0100287} // namespace
288
Florian Mayer4dd5a1f2020-07-14 13:00:51 +0100289bool ParseLlvmSymbolizerLine(const std::string& line,
290 std::string* file_name,
291 uint32_t* line_no) {
292 size_t col_pos = line.rfind(':');
293 if (col_pos == std::string::npos || col_pos == 0)
294 return false;
295 size_t row_pos = line.rfind(':', col_pos - 1);
296 if (row_pos == std::string::npos || row_pos == 0)
297 return false;
298 *file_name = line.substr(0, row_pos);
299 auto line_no_str = line.substr(row_pos + 1, col_pos - row_pos - 1);
300
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100301 std::optional<int32_t> opt_parsed_line_no = base::StringToInt32(line_no_str);
Florian Mayer4dd5a1f2020-07-14 13:00:51 +0100302 if (!opt_parsed_line_no || *opt_parsed_line_no < 0)
303 return false;
304 *line_no = static_cast<uint32_t>(*opt_parsed_line_no);
305 return true;
306}
307
Florian Mayer6d1f0ae2020-07-21 08:59:29 +0100308BinaryFinder::~BinaryFinder() = default;
309
310LocalBinaryIndexer::LocalBinaryIndexer(std::vector<std::string> roots)
311 : buildid_to_file_(BuildIdIndex(std::move(roots))) {}
312
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100313std::optional<FoundBinary> LocalBinaryIndexer::FindBinary(
Florian Mayer6d1f0ae2020-07-21 08:59:29 +0100314 const std::string& abspath,
315 const std::string& build_id) {
316 auto it = buildid_to_file_.find(build_id);
317 if (it != buildid_to_file_.end())
318 return it->second;
319 PERFETTO_ELOG("Could not find Build ID: %s (file %s).",
320 base::ToHex(build_id).c_str(), abspath.c_str());
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100321 return std::nullopt;
Florian Mayer6d1f0ae2020-07-21 08:59:29 +0100322}
323
324LocalBinaryIndexer::~LocalBinaryIndexer() = default;
325
326LocalBinaryFinder::LocalBinaryFinder(std::vector<std::string> roots)
327 : roots_(std::move(roots)) {}
328
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100329std::optional<FoundBinary> LocalBinaryFinder::FindBinary(
Florian Mayerd12a6de2019-08-01 10:36:11 +0100330 const std::string& abspath,
331 const std::string& build_id) {
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100332 auto p = cache_.emplace(abspath, std::nullopt);
Florian Mayerd12a6de2019-08-01 10:36:11 +0100333 if (!p.second)
334 return p.first->second;
335
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100336 std::optional<FoundBinary>& cache_entry = p.first->second;
Florian Mayerd12a6de2019-08-01 10:36:11 +0100337
338 for (const std::string& root_str : roots_) {
339 cache_entry = FindBinaryInRoot(root_str, abspath, build_id);
340 if (cache_entry)
341 return cache_entry;
342 }
Florian Mayerfdc86d62019-09-20 11:23:55 +0100343 PERFETTO_ELOG("Could not find %s (Build ID: %s).", abspath.c_str(),
344 base::ToHex(build_id).c_str());
Florian Mayerd12a6de2019-08-01 10:36:11 +0100345 return cache_entry;
346}
347
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100348std::optional<FoundBinary> LocalBinaryFinder::IsCorrectFile(
Florian Mayere7228a72020-10-22 18:18:01 +0100349 const std::string& symbol_file,
350 const std::string& build_id) {
Primiano Tucciab293f52020-12-08 11:46:52 +0100351 if (!base::FileExists(symbol_file)) {
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100352 return std::nullopt;
Florian Mayere7228a72020-10-22 18:18:01 +0100353 }
Daniele Di Proiettocb426002023-02-16 12:14:38 +0000354 // Openfile opens the file with an exclusive lock on windows.
Daniele Di Proiettob2b829c2024-02-19 17:23:34 +0000355 std::optional<size_t> size = base::GetFileSize(symbol_file);
356 if (!size.has_value()) {
357 PERFETTO_PLOG("Failed to get file size %s", symbol_file.c_str());
358 return std::nullopt;
359 }
Daniele Di Proiettocb426002023-02-16 12:14:38 +0000360
Daniele Di Proiettob2b829c2024-02-19 17:23:34 +0000361 if (*size == 0) {
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100362 return std::nullopt;
Daniele Di Proiettocb426002023-02-16 12:14:38 +0000363 }
364
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100365 std::optional<BuildIdAndLoadBias> build_id_and_load_bias =
Daniele Di Proiettob2b829c2024-02-19 17:23:34 +0000366 GetBuildIdAndLoadBias(symbol_file.c_str(), *size);
Florian Mayere7228a72020-10-22 18:18:01 +0100367 if (!build_id_and_load_bias)
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100368 return std::nullopt;
Florian Mayere7228a72020-10-22 18:18:01 +0100369 if (build_id_and_load_bias->build_id != build_id) {
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100370 return std::nullopt;
Florian Mayere7228a72020-10-22 18:18:01 +0100371 }
372 return FoundBinary{symbol_file, build_id_and_load_bias->load_bias};
Florian Mayerd12a6de2019-08-01 10:36:11 +0100373}
374
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100375std::optional<FoundBinary> LocalBinaryFinder::FindBinaryInRoot(
Florian Mayerd12a6de2019-08-01 10:36:11 +0100376 const std::string& root_str,
377 const std::string& abspath,
378 const std::string& build_id) {
379 constexpr char kApkPrefix[] = "base.apk!";
380
381 std::string filename;
382 std::string dirname;
383
384 for (base::StringSplitter sp(abspath, '/'); sp.Next();) {
Florian Mayer39179192019-09-09 16:15:22 +0100385 if (!dirname.empty())
386 dirname += "/";
387 dirname += filename;
Florian Mayerd12a6de2019-08-01 10:36:11 +0100388 filename = sp.cur_token();
389 }
390
391 // Return the first match for the following options:
392 // * absolute path of library file relative to root.
393 // * absolute path of library file relative to root, but with base.apk!
394 // removed from filename.
395 // * only filename of library file relative to root.
396 // * only filename of library file relative to root, but with base.apk!
397 // removed from filename.
Florian Mayer39179192019-09-09 16:15:22 +0100398 // * in the subdirectory .build-id: the first two hex digits of the build-id
399 // as subdirectory, then the rest of the hex digits, with ".debug"appended.
400 // See
401 // https://fedoraproject.org/wiki/RolandMcGrath/BuildID#Find_files_by_build_ID
Florian Mayerd12a6de2019-08-01 10:36:11 +0100402 //
Florian Mayer39179192019-09-09 16:15:22 +0100403 // For example, "/system/lib/base.apk!foo.so" with build id abcd1234,
404 // is looked for at
Florian Mayerd12a6de2019-08-01 10:36:11 +0100405 // * $ROOT/system/lib/base.apk!foo.so
406 // * $ROOT/system/lib/foo.so
407 // * $ROOT/base.apk!foo.so
408 // * $ROOT/foo.so
Florian Mayer39179192019-09-09 16:15:22 +0100409 // * $ROOT/.build-id/ab/cd1234.debug
Florian Mayerd12a6de2019-08-01 10:36:11 +0100410
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100411 std::optional<FoundBinary> result;
Florian Mayere7228a72020-10-22 18:18:01 +0100412
Florian Mayerd12a6de2019-08-01 10:36:11 +0100413 std::string symbol_file = root_str + "/" + dirname + "/" + filename;
Florian Mayere7228a72020-10-22 18:18:01 +0100414 result = IsCorrectFile(symbol_file, build_id);
415 if (result) {
416 return result;
417 }
Florian Mayerd12a6de2019-08-01 10:36:11 +0100418
Hector Dearmanaa025742021-01-14 13:55:22 +0000419 if (base::StartsWith(filename, kApkPrefix)) {
Daniele Di Proiettofc25f7e2022-02-08 20:13:39 +0000420 symbol_file = root_str + "/" + dirname + "/" +
421 filename.substr(sizeof(kApkPrefix) - 1);
Florian Mayere7228a72020-10-22 18:18:01 +0100422 result = IsCorrectFile(symbol_file, build_id);
423 if (result) {
424 return result;
425 }
Florian Mayerd12a6de2019-08-01 10:36:11 +0100426 }
427
428 symbol_file = root_str + "/" + filename;
Florian Mayere7228a72020-10-22 18:18:01 +0100429 result = IsCorrectFile(symbol_file, build_id);
430 if (result) {
431 return result;
432 }
Florian Mayerd12a6de2019-08-01 10:36:11 +0100433
Hector Dearmanaa025742021-01-14 13:55:22 +0000434 if (base::StartsWith(filename, kApkPrefix)) {
Daniele Di Proiettofc25f7e2022-02-08 20:13:39 +0000435 symbol_file = root_str + "/" + filename.substr(sizeof(kApkPrefix) - 1);
Florian Mayere7228a72020-10-22 18:18:01 +0100436 result = IsCorrectFile(symbol_file, build_id);
437 if (result) {
438 return result;
439 }
Florian Mayerd12a6de2019-08-01 10:36:11 +0100440 }
441
Hector Dearman5f069382019-09-16 10:24:01 +0100442 std::string hex_build_id = base::ToHex(build_id.c_str(), build_id.size());
Florian Mayer39179192019-09-09 16:15:22 +0100443 std::string split_hex_build_id = SplitBuildID(hex_build_id);
444 if (!split_hex_build_id.empty()) {
445 symbol_file =
446 root_str + "/" + ".build-id" + "/" + split_hex_build_id + ".debug";
Florian Mayere7228a72020-10-22 18:18:01 +0100447 result = IsCorrectFile(symbol_file, build_id);
448 if (result) {
449 return result;
450 }
Florian Mayer39179192019-09-09 16:15:22 +0100451 }
452
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100453 return std::nullopt;
Florian Mayerd12a6de2019-08-01 10:36:11 +0100454}
455
Florian Mayer6d1f0ae2020-07-21 08:59:29 +0100456LocalBinaryFinder::~LocalBinaryFinder() = default;
457
Florian Mayer79eb2122020-12-10 17:09:09 +0000458LLVMSymbolizerProcess::LLVMSymbolizerProcess(const std::string& symbolizer_path)
Joshua Gilpatrickcd349942020-11-11 15:18:37 -0800459 :
460#if PERFETTO_BUILDFLAG(PERFETTO_OS_WIN)
Florian Mayer79eb2122020-12-10 17:09:09 +0000461 subprocess_(symbolizer_path, {}) {
Joshua Gilpatrickcd349942020-11-11 15:18:37 -0800462}
463#else
Florian Mayer79eb2122020-12-10 17:09:09 +0000464 subprocess_(symbolizer_path, {"llvm-symbolizer"}) {
Joshua Gilpatrickcd349942020-11-11 15:18:37 -0800465}
466#endif
Florian Mayerd12a6de2019-08-01 10:36:11 +0100467
468std::vector<SymbolizedFrame> LLVMSymbolizerProcess::Symbolize(
469 const std::string& binary,
470 uint64_t address) {
471 std::vector<SymbolizedFrame> result;
Primiano Tucci934f8df2022-11-28 18:35:00 +0000472 base::StackString<1024> buffer("\"%s\" 0x%" PRIx64 "\n", binary.c_str(),
473 address);
474 if (subprocess_.Write(buffer.c_str(), buffer.len()) < 0) {
Florian Mayerd12a6de2019-08-01 10:36:11 +0100475 PERFETTO_ELOG("Failed to write to llvm-symbolizer.");
476 return result;
477 }
Joshua Gilpatrickcd349942020-11-11 15:18:37 -0800478 auto lines = GetLines([&](char* read_buffer, size_t buffer_size) {
479 return subprocess_.Read(read_buffer, buffer_size);
480 });
Florian Mayerd12a6de2019-08-01 10:36:11 +0100481 // llvm-symbolizer writes out records in the form of
482 // Foo(Bar*)
483 // foo.cc:123
484 // This is why we should always get a multiple of two number of lines.
485 PERFETTO_DCHECK(lines.size() % 2 == 0);
486 result.resize(lines.size() / 2);
487 for (size_t i = 0; i < lines.size(); ++i) {
Florian Mayer519f32e2019-08-01 18:18:39 +0100488 SymbolizedFrame& cur = result[i / 2];
489 if (i % 2 == 0) {
490 cur.function_name = lines[i];
491 } else {
Florian Mayer4dd5a1f2020-07-14 13:00:51 +0100492 if (!ParseLlvmSymbolizerLine(lines[i], &cur.file_name, &cur.line)) {
Florian Mayer519f32e2019-08-01 18:18:39 +0100493 PERFETTO_ELOG("Failed to parse llvm-symbolizer line: %s",
494 lines[i].c_str());
495 cur.file_name = "";
496 cur.line = 0;
497 }
498 }
Florian Mayerd12a6de2019-08-01 10:36:11 +0100499 }
Florian Mayercf9eadc2019-08-05 11:59:35 +0100500
501 for (auto it = result.begin(); it != result.end();) {
502 if (it->function_name == "??")
503 it = result.erase(it);
504 else
505 ++it;
506 }
Florian Mayerd12a6de2019-08-01 10:36:11 +0100507 return result;
508}
Florian Mayer04d1ace2019-08-01 11:59:53 +0100509std::vector<std::vector<SymbolizedFrame>> LocalSymbolizer::Symbolize(
Florian Mayerd12a6de2019-08-01 10:36:11 +0100510 const std::string& mapping_name,
511 const std::string& build_id,
Florian Mayere7228a72020-10-22 18:18:01 +0100512 uint64_t load_bias,
Florian Mayer04d1ace2019-08-01 11:59:53 +0100513 const std::vector<uint64_t>& addresses) {
Lalit Maganti4e2303c2023-03-29 15:28:36 +0100514 std::optional<FoundBinary> binary =
Florian Mayer6d1f0ae2020-07-21 08:59:29 +0100515 finder_->FindBinary(mapping_name, build_id);
Florian Mayerd12a6de2019-08-01 10:36:11 +0100516 if (!binary)
517 return {};
Florian Mayere7228a72020-10-22 18:18:01 +0100518 uint64_t load_bias_correction = 0;
519 if (binary->load_bias > load_bias) {
520 // On Android 10, there was a bug in libunwindstack that would incorrectly
521 // calculate the load_bias, and thus the relative PC. This would end up in
522 // frames that made no sense. We can fix this up after the fact if we
523 // detect this situation.
524 load_bias_correction = binary->load_bias - load_bias;
525 PERFETTO_LOG("Correcting load bias by %" PRIu64 " for %s",
526 load_bias_correction, mapping_name.c_str());
527 }
Florian Mayer04d1ace2019-08-01 11:59:53 +0100528 std::vector<std::vector<SymbolizedFrame>> result;
529 result.reserve(addresses.size());
530 for (uint64_t address : addresses)
Florian Mayere7228a72020-10-22 18:18:01 +0100531 result.emplace_back(llvm_symbolizer_.Symbolize(
532 binary->file_name, address + load_bias_correction));
Florian Mayer04d1ace2019-08-01 11:59:53 +0100533 return result;
Florian Mayerd12a6de2019-08-01 10:36:11 +0100534}
535
Florian Mayer79eb2122020-12-10 17:09:09 +0000536LocalSymbolizer::LocalSymbolizer(const std::string& symbolizer_path,
537 std::unique_ptr<BinaryFinder> finder)
538 : llvm_symbolizer_(symbolizer_path), finder_(std::move(finder)) {}
539
540LocalSymbolizer::LocalSymbolizer(std::unique_ptr<BinaryFinder> finder)
541 : LocalSymbolizer(kDefaultSymbolizer, std::move(finder)) {}
542
Florian Mayerd12a6de2019-08-01 10:36:11 +0100543LocalSymbolizer::~LocalSymbolizer() = default;
544
Florian Mayerbd88ff82019-12-18 16:36:37 +0000545} // namespace profiling
Florian Mayerd12a6de2019-08-01 10:36:11 +0100546} // namespace perfetto
Primiano Tucci41af34f2019-10-01 13:09:22 +0100547
548#endif // PERFETTO_BUILDFLAG(PERFETTO_LOCAL_SYMBOLIZER)