Yabin Cui | 8f62251 | 2015-05-05 19:58:07 -0700 | [diff] [blame] | 1 | /* |
| 2 | * Copyright (C) 2015 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 | |
| 17 | #include "read_elf.h" |
Than McIntosh | f831e6d | 2016-02-01 19:50:20 -0500 | [diff] [blame] | 18 | #include "read_apk.h" |
Yabin Cui | 8f62251 | 2015-05-05 19:58:07 -0700 | [diff] [blame] | 19 | |
| 20 | #include <stdio.h> |
| 21 | #include <string.h> |
Than McIntosh | f831e6d | 2016-02-01 19:50:20 -0500 | [diff] [blame] | 22 | #include <sys/stat.h> |
| 23 | #include <sys/types.h> |
Yabin Cui | 547c60e | 2015-10-12 16:56:05 -0700 | [diff] [blame] | 24 | |
Yabin Cui | 8f62251 | 2015-05-05 19:58:07 -0700 | [diff] [blame] | 25 | #include <algorithm> |
Yabin Cui | 547c60e | 2015-10-12 16:56:05 -0700 | [diff] [blame] | 26 | #include <limits> |
| 27 | |
Elliott Hughes | 66dd09e | 2015-12-04 14:00:57 -0800 | [diff] [blame] | 28 | #include <android-base/file.h> |
| 29 | #include <android-base/logging.h> |
Yabin Cui | 8f62251 | 2015-05-05 19:58:07 -0700 | [diff] [blame] | 30 | |
| 31 | #pragma clang diagnostic push |
| 32 | #pragma clang diagnostic ignored "-Wunused-parameter" |
| 33 | |
| 34 | #include <llvm/ADT/StringRef.h> |
| 35 | #include <llvm/Object/Binary.h> |
| 36 | #include <llvm/Object/ELFObjectFile.h> |
| 37 | #include <llvm/Object/ObjectFile.h> |
| 38 | |
| 39 | #pragma clang diagnostic pop |
| 40 | |
Yabin Cui | 9ba4d94 | 2020-09-08 16:12:46 -0700 | [diff] [blame] | 41 | #include "JITDebugReader.h" |
Yabin Cui | 8f62251 | 2015-05-05 19:58:07 -0700 | [diff] [blame] | 42 | #include "utils.h" |
| 43 | |
Yabin Cui | faa7b92 | 2021-01-11 17:35:57 -0800 | [diff] [blame] | 44 | namespace simpleperf { |
Yabin Cui | a5bdf0a | 2015-06-22 19:41:39 -0700 | [diff] [blame] | 45 | |
Yabin Cui | faa7b92 | 2021-01-11 17:35:57 -0800 | [diff] [blame] | 46 | const static char* ELF_NOTE_GNU = "GNU"; |
| 47 | const static int NT_GNU_BUILD_ID = 3; |
Yabin Cui | 02e2033 | 2020-03-16 19:38:23 -0700 | [diff] [blame] | 48 | |
Yabin Cui | dec43c1 | 2016-07-29 16:40:40 -0700 | [diff] [blame] | 49 | std::ostream& operator<<(std::ostream& os, const ElfStatus& status) { |
| 50 | switch (status) { |
| 51 | case ElfStatus::NO_ERROR: |
| 52 | os << "No error"; |
| 53 | break; |
| 54 | case ElfStatus::FILE_NOT_FOUND: |
| 55 | os << "File not found"; |
| 56 | break; |
| 57 | case ElfStatus::READ_FAILED: |
| 58 | os << "Read failed"; |
| 59 | break; |
| 60 | case ElfStatus::FILE_MALFORMED: |
| 61 | os << "Malformed file"; |
| 62 | break; |
| 63 | case ElfStatus::NO_SYMBOL_TABLE: |
| 64 | os << "No symbol table"; |
| 65 | break; |
| 66 | case ElfStatus::NO_BUILD_ID: |
| 67 | os << "No build id"; |
| 68 | break; |
| 69 | case ElfStatus::BUILD_ID_MISMATCH: |
| 70 | os << "Build id mismatch"; |
| 71 | break; |
| 72 | case ElfStatus::SECTION_NOT_FOUND: |
| 73 | os << "Section not found"; |
| 74 | break; |
| 75 | } |
| 76 | return os; |
Than McIntosh | f831e6d | 2016-02-01 19:50:20 -0500 | [diff] [blame] | 77 | } |
| 78 | |
Yabin Cui | 9e43e9f | 2018-03-09 15:57:13 -0800 | [diff] [blame] | 79 | bool IsValidElfFileMagic(const char* buf, size_t buf_size) { |
Yabin Cui | dec43c1 | 2016-07-29 16:40:40 -0700 | [diff] [blame] | 80 | static const char elf_magic[] = {0x7f, 'E', 'L', 'F'}; |
Yabin Cui | 9e43e9f | 2018-03-09 15:57:13 -0800 | [diff] [blame] | 81 | return (buf_size >= 4u && memcmp(buf, elf_magic, 4) == 0); |
| 82 | } |
| 83 | |
Yabin Cui | 02e2033 | 2020-03-16 19:38:23 -0700 | [diff] [blame] | 84 | ElfStatus IsValidElfFile(int fd, uint64_t file_offset) { |
Yabin Cui | dec43c1 | 2016-07-29 16:40:40 -0700 | [diff] [blame] | 85 | char buf[4]; |
Yabin Cui | 02e2033 | 2020-03-16 19:38:23 -0700 | [diff] [blame] | 86 | if (!android::base::ReadFullyAtOffset(fd, buf, 4, file_offset)) { |
Yabin Cui | dec43c1 | 2016-07-29 16:40:40 -0700 | [diff] [blame] | 87 | return ElfStatus::READ_FAILED; |
| 88 | } |
Yabin Cui | 9e43e9f | 2018-03-09 15:57:13 -0800 | [diff] [blame] | 89 | return IsValidElfFileMagic(buf, 4) ? ElfStatus::NO_ERROR : ElfStatus::FILE_MALFORMED; |
Yabin Cui | dec43c1 | 2016-07-29 16:40:40 -0700 | [diff] [blame] | 90 | } |
| 91 | |
Yabin Cui | f79a008 | 2016-11-14 11:23:14 -0800 | [diff] [blame] | 92 | bool GetBuildIdFromNoteSection(const char* section, size_t section_size, BuildId* build_id) { |
Yabin Cui | 8f62251 | 2015-05-05 19:58:07 -0700 | [diff] [blame] | 93 | const char* p = section; |
| 94 | const char* end = p + section_size; |
| 95 | while (p < end) { |
Yabin Cui | f79a008 | 2016-11-14 11:23:14 -0800 | [diff] [blame] | 96 | if (p + 12 >= end) { |
| 97 | return false; |
| 98 | } |
| 99 | uint32_t namesz; |
| 100 | uint32_t descsz; |
| 101 | uint32_t type; |
| 102 | MoveFromBinaryFormat(namesz, p); |
| 103 | MoveFromBinaryFormat(descsz, p); |
| 104 | MoveFromBinaryFormat(type, p); |
Yabin Cui | a7a0e50 | 2016-06-15 11:49:23 -0700 | [diff] [blame] | 105 | namesz = Align(namesz, 4); |
| 106 | descsz = Align(descsz, 4); |
Yabin Cui | f79a008 | 2016-11-14 11:23:14 -0800 | [diff] [blame] | 107 | if ((type == NT_GNU_BUILD_ID) && (p < end) && (strcmp(p, ELF_NOTE_GNU) == 0)) { |
| 108 | const char* desc_start = p + namesz; |
| 109 | const char* desc_end = desc_start + descsz; |
| 110 | if (desc_start > p && desc_start < desc_end && desc_end <= end) { |
| 111 | *build_id = BuildId(p + namesz, descsz); |
| 112 | return true; |
| 113 | } else { |
| 114 | return false; |
| 115 | } |
Yabin Cui | 8f62251 | 2015-05-05 19:58:07 -0700 | [diff] [blame] | 116 | } |
| 117 | p += namesz + descsz; |
| 118 | } |
| 119 | return false; |
| 120 | } |
| 121 | |
Yabin Cui | dec43c1 | 2016-07-29 16:40:40 -0700 | [diff] [blame] | 122 | ElfStatus GetBuildIdFromNoteFile(const std::string& filename, BuildId* build_id) { |
Yabin Cui | 8f62251 | 2015-05-05 19:58:07 -0700 | [diff] [blame] | 123 | std::string content; |
| 124 | if (!android::base::ReadFileToString(filename, &content)) { |
Yabin Cui | dec43c1 | 2016-07-29 16:40:40 -0700 | [diff] [blame] | 125 | return ElfStatus::READ_FAILED; |
Yabin Cui | 8f62251 | 2015-05-05 19:58:07 -0700 | [diff] [blame] | 126 | } |
Yabin Cui | dec43c1 | 2016-07-29 16:40:40 -0700 | [diff] [blame] | 127 | if (!GetBuildIdFromNoteSection(content.c_str(), content.size(), build_id)) { |
| 128 | return ElfStatus::NO_BUILD_ID; |
Yabin Cui | 8f62251 | 2015-05-05 19:58:07 -0700 | [diff] [blame] | 129 | } |
Yabin Cui | dec43c1 | 2016-07-29 16:40:40 -0700 | [diff] [blame] | 130 | return ElfStatus::NO_ERROR; |
Yabin Cui | 8f62251 | 2015-05-05 19:58:07 -0700 | [diff] [blame] | 131 | } |
| 132 | |
Yabin Cui | faa7b92 | 2021-01-11 17:35:57 -0800 | [diff] [blame] | 133 | bool IsArmMappingSymbol(const char* name) { |
| 134 | // Mapping symbols in arm, which are described in "ELF for ARM Architecture" and |
| 135 | // "ELF for ARM 64-bit Architecture". The regular expression to match mapping symbol |
| 136 | // is ^\$(a|d|t|x)(\..*)?$ |
| 137 | return name[0] == '$' && strchr("adtx", name[1]) != nullptr && |
| 138 | (name[2] == '\0' || name[2] == '.'); |
| 139 | } |
| 140 | |
| 141 | namespace { |
| 142 | |
Yabin Cui | dec43c1 | 2016-07-29 16:40:40 -0700 | [diff] [blame] | 143 | struct BinaryWrapper { |
Yabin Cui | 02e2033 | 2020-03-16 19:38:23 -0700 | [diff] [blame] | 144 | std::unique_ptr<llvm::MemoryBuffer> buffer; |
| 145 | std::unique_ptr<llvm::object::Binary> binary; |
| 146 | llvm::object::ObjectFile* obj = nullptr; |
Yabin Cui | b1a885b | 2016-02-14 19:18:02 -0800 | [diff] [blame] | 147 | }; |
| 148 | |
Yabin Cui | dec43c1 | 2016-07-29 16:40:40 -0700 | [diff] [blame] | 149 | static ElfStatus OpenObjectFile(const std::string& filename, uint64_t file_offset, |
| 150 | uint64_t file_size, BinaryWrapper* wrapper) { |
Yabin Cui | 02e2033 | 2020-03-16 19:38:23 -0700 | [diff] [blame] | 151 | if (!IsRegularFile(filename)) { |
| 152 | return ElfStatus::FILE_NOT_FOUND; |
| 153 | } |
Yabin Cui | 2a53ff3 | 2018-05-21 17:37:00 -0700 | [diff] [blame] | 154 | android::base::unique_fd fd = FileHelper::OpenReadOnly(filename); |
| 155 | if (fd == -1) { |
Yabin Cui | dec43c1 | 2016-07-29 16:40:40 -0700 | [diff] [blame] | 156 | return ElfStatus::READ_FAILED; |
Than McIntosh | f831e6d | 2016-02-01 19:50:20 -0500 | [diff] [blame] | 157 | } |
Yabin Cui | b1a885b | 2016-02-14 19:18:02 -0800 | [diff] [blame] | 158 | if (file_size == 0) { |
| 159 | file_size = GetFileSize(filename); |
| 160 | if (file_size == 0) { |
Yabin Cui | dec43c1 | 2016-07-29 16:40:40 -0700 | [diff] [blame] | 161 | return ElfStatus::READ_FAILED; |
Yabin Cui | b1a885b | 2016-02-14 19:18:02 -0800 | [diff] [blame] | 162 | } |
Than McIntosh | f831e6d | 2016-02-01 19:50:20 -0500 | [diff] [blame] | 163 | } |
Yabin Cui | 02e2033 | 2020-03-16 19:38:23 -0700 | [diff] [blame] | 164 | ElfStatus status = IsValidElfFile(fd, file_offset); |
| 165 | if (status != ElfStatus::NO_ERROR) { |
| 166 | return status; |
| 167 | } |
Yabin Cui | 2a53ff3 | 2018-05-21 17:37:00 -0700 | [diff] [blame] | 168 | auto buffer_or_err = llvm::MemoryBuffer::getOpenFileSlice(fd, filename, file_size, file_offset); |
Yabin Cui | b1a885b | 2016-02-14 19:18:02 -0800 | [diff] [blame] | 169 | if (!buffer_or_err) { |
Yabin Cui | dec43c1 | 2016-07-29 16:40:40 -0700 | [diff] [blame] | 170 | return ElfStatus::READ_FAILED; |
Than McIntosh | f831e6d | 2016-02-01 19:50:20 -0500 | [diff] [blame] | 171 | } |
Yabin Cui | b1a885b | 2016-02-14 19:18:02 -0800 | [diff] [blame] | 172 | auto binary_or_err = llvm::object::createBinary(buffer_or_err.get()->getMemBufferRef()); |
| 173 | if (!binary_or_err) { |
Yabin Cui | dec43c1 | 2016-07-29 16:40:40 -0700 | [diff] [blame] | 174 | return ElfStatus::READ_FAILED; |
Yabin Cui | b1a885b | 2016-02-14 19:18:02 -0800 | [diff] [blame] | 175 | } |
Yabin Cui | 02e2033 | 2020-03-16 19:38:23 -0700 | [diff] [blame] | 176 | wrapper->buffer = std::move(buffer_or_err.get()); |
| 177 | wrapper->binary = std::move(binary_or_err.get()); |
| 178 | wrapper->obj = llvm::dyn_cast<llvm::object::ObjectFile>(wrapper->binary.get()); |
Yabin Cui | dec43c1 | 2016-07-29 16:40:40 -0700 | [diff] [blame] | 179 | if (wrapper->obj == nullptr) { |
| 180 | return ElfStatus::FILE_MALFORMED; |
Yabin Cui | b1a885b | 2016-02-14 19:18:02 -0800 | [diff] [blame] | 181 | } |
Yabin Cui | dec43c1 | 2016-07-29 16:40:40 -0700 | [diff] [blame] | 182 | return ElfStatus::NO_ERROR; |
Than McIntosh | f831e6d | 2016-02-01 19:50:20 -0500 | [diff] [blame] | 183 | } |
| 184 | |
Yabin Cui | 9e43e9f | 2018-03-09 15:57:13 -0800 | [diff] [blame] | 185 | static ElfStatus OpenObjectFileInMemory(const char* data, size_t size, BinaryWrapper* wrapper) { |
| 186 | auto buffer = llvm::MemoryBuffer::getMemBuffer(llvm::StringRef(data, size)); |
Yabin Cui | 0540053 | 2016-03-17 21:18:53 -0700 | [diff] [blame] | 187 | auto binary_or_err = llvm::object::createBinary(buffer->getMemBufferRef()); |
| 188 | if (!binary_or_err) { |
Yabin Cui | dec43c1 | 2016-07-29 16:40:40 -0700 | [diff] [blame] | 189 | return ElfStatus::FILE_MALFORMED; |
Yabin Cui | 0540053 | 2016-03-17 21:18:53 -0700 | [diff] [blame] | 190 | } |
Yabin Cui | 02e2033 | 2020-03-16 19:38:23 -0700 | [diff] [blame] | 191 | wrapper->buffer = std::move(buffer); |
| 192 | wrapper->binary = std::move(binary_or_err.get()); |
| 193 | wrapper->obj = llvm::dyn_cast<llvm::object::ObjectFile>(wrapper->binary.get()); |
Yabin Cui | dec43c1 | 2016-07-29 16:40:40 -0700 | [diff] [blame] | 194 | if (wrapper->obj == nullptr) { |
| 195 | return ElfStatus::FILE_MALFORMED; |
Yabin Cui | 0540053 | 2016-03-17 21:18:53 -0700 | [diff] [blame] | 196 | } |
Yabin Cui | dec43c1 | 2016-07-29 16:40:40 -0700 | [diff] [blame] | 197 | return ElfStatus::NO_ERROR; |
Yabin Cui | 0540053 | 2016-03-17 21:18:53 -0700 | [diff] [blame] | 198 | } |
| 199 | |
Thiébaud Weksteen | 4848ee0 | 2020-10-23 16:06:59 +0200 | [diff] [blame] | 200 | void ReadSymbolTable(llvm::object::symbol_iterator sym_begin, llvm::object::symbol_iterator sym_end, |
| 201 | const std::function<void(const ElfFileSymbol&)>& callback, bool is_arm, |
Yabin Cui | 81d4c49 | 2018-08-09 12:06:10 -0700 | [diff] [blame] | 202 | const llvm::object::section_iterator& section_end) { |
Yabin Cui | 0540053 | 2016-03-17 21:18:53 -0700 | [diff] [blame] | 203 | for (; sym_begin != sym_end; ++sym_begin) { |
Yabin Cui | ec12ed9 | 2015-06-08 10:38:10 -0700 | [diff] [blame] | 204 | ElfFileSymbol symbol; |
Yabin Cui | 0540053 | 2016-03-17 21:18:53 -0700 | [diff] [blame] | 205 | auto symbol_ref = static_cast<const llvm::object::ELFSymbolRef*>(&*sym_begin); |
Yabin Cui | 1b276a3 | 2019-11-18 17:23:27 -0800 | [diff] [blame] | 206 | // Exclude undefined symbols, otherwise we may wrongly use them as labels in functions. |
| 207 | if (symbol_ref->getFlags() & symbol_ref->SF_Undefined) { |
| 208 | continue; |
| 209 | } |
Pirama Arumuga Nainar | 0ea8250 | 2016-09-16 16:53:25 -0700 | [diff] [blame] | 210 | llvm::Expected<llvm::object::section_iterator> section_it_or_err = symbol_ref->getSection(); |
Yabin Cui | 0540053 | 2016-03-17 21:18:53 -0700 | [diff] [blame] | 211 | if (!section_it_or_err) { |
Yabin Cui | ec12ed9 | 2015-06-08 10:38:10 -0700 | [diff] [blame] | 212 | continue; |
| 213 | } |
Yabin Cui | 81d4c49 | 2018-08-09 12:06:10 -0700 | [diff] [blame] | 214 | // Symbols in .dynsym section don't have associated section. |
| 215 | if (section_it_or_err.get() != section_end) { |
| 216 | llvm::StringRef section_name; |
| 217 | if (section_it_or_err.get()->getName(section_name) || section_name.empty()) { |
| 218 | continue; |
| 219 | } |
| 220 | if (section_name == ".text") { |
| 221 | symbol.is_in_text_section = true; |
| 222 | } |
| 223 | } |
Yabin Cui | ec12ed9 | 2015-06-08 10:38:10 -0700 | [diff] [blame] | 224 | |
Pirama Arumuga Nainar | 0ea8250 | 2016-09-16 16:53:25 -0700 | [diff] [blame] | 225 | llvm::Expected<llvm::StringRef> symbol_name_or_err = symbol_ref->getName(); |
Yabin Cui | 0540053 | 2016-03-17 21:18:53 -0700 | [diff] [blame] | 226 | if (!symbol_name_or_err || symbol_name_or_err.get().empty()) { |
| 227 | continue; |
| 228 | } |
| 229 | |
| 230 | symbol.name = symbol_name_or_err.get(); |
| 231 | symbol.vaddr = symbol_ref->getValue(); |
Yabin Cui | 547c60e | 2015-10-12 16:56:05 -0700 | [diff] [blame] | 232 | if ((symbol.vaddr & 1) != 0 && is_arm) { |
Yabin Cui | b378355 | 2015-06-11 11:15:42 -0700 | [diff] [blame] | 233 | // Arm sets bit 0 to mark it as thumb code, remove the flag. |
Yabin Cui | 547c60e | 2015-10-12 16:56:05 -0700 | [diff] [blame] | 234 | symbol.vaddr &= ~1; |
Yabin Cui | b378355 | 2015-06-11 11:15:42 -0700 | [diff] [blame] | 235 | } |
Yabin Cui | 0540053 | 2016-03-17 21:18:53 -0700 | [diff] [blame] | 236 | symbol.len = symbol_ref->getSize(); |
Pirama Arumuga Nainar | 0ea8250 | 2016-09-16 16:53:25 -0700 | [diff] [blame] | 237 | llvm::object::SymbolRef::Type symbol_type = *symbol_ref->getType(); |
Yabin Cui | 0540053 | 2016-03-17 21:18:53 -0700 | [diff] [blame] | 238 | if (symbol_type == llvm::object::SymbolRef::ST_Function) { |
Yabin Cui | b378355 | 2015-06-11 11:15:42 -0700 | [diff] [blame] | 239 | symbol.is_func = true; |
Yabin Cui | 0540053 | 2016-03-17 21:18:53 -0700 | [diff] [blame] | 240 | } else if (symbol_type == llvm::object::SymbolRef::ST_Unknown) { |
Yabin Cui | b378355 | 2015-06-11 11:15:42 -0700 | [diff] [blame] | 241 | if (symbol.is_in_text_section) { |
| 242 | symbol.is_label = true; |
Yabin Cui | 2c17e0d | 2015-06-11 14:57:21 -0700 | [diff] [blame] | 243 | if (is_arm) { |
Yabin Cui | a5bdf0a | 2015-06-22 19:41:39 -0700 | [diff] [blame] | 244 | // Remove mapping symbols in arm. |
| 245 | const char* p = (symbol.name.compare(0, linker_prefix.size(), linker_prefix) == 0) |
| 246 | ? symbol.name.c_str() + linker_prefix.size() |
| 247 | : symbol.name.c_str(); |
| 248 | if (IsArmMappingSymbol(p)) { |
Yabin Cui | 2c17e0d | 2015-06-11 14:57:21 -0700 | [diff] [blame] | 249 | symbol.is_label = false; |
| 250 | } |
Yabin Cui | b378355 | 2015-06-11 11:15:42 -0700 | [diff] [blame] | 251 | } |
| 252 | } |
| 253 | } |
| 254 | |
Yabin Cui | ec12ed9 | 2015-06-08 10:38:10 -0700 | [diff] [blame] | 255 | callback(symbol); |
| 256 | } |
Yabin Cui | ec12ed9 | 2015-06-08 10:38:10 -0700 | [diff] [blame] | 257 | } |
| 258 | |
Yabin Cui | 0540053 | 2016-03-17 21:18:53 -0700 | [diff] [blame] | 259 | template <class ELFT> |
Yabin Cui | 5783fa0 | 2016-07-06 18:29:00 -0700 | [diff] [blame] | 260 | void AddSymbolForPltSection(const llvm::object::ELFObjectFile<ELFT>* elf, |
Yabin Cui | 3e4c595 | 2016-07-26 15:03:27 -0700 | [diff] [blame] | 261 | const std::function<void(const ElfFileSymbol&)>& callback) { |
Yabin Cui | 5783fa0 | 2016-07-06 18:29:00 -0700 | [diff] [blame] | 262 | // We may sample instructions in .plt section if the program |
| 263 | // calls functions from shared libraries. Different architectures use |
| 264 | // different formats to store .plt section, so it needs a lot of work to match |
| 265 | // instructions in .plt section to symbols. As samples in .plt section rarely |
| 266 | // happen, and .plt section can hardly be a performance bottleneck, we can |
| 267 | // just use a symbol @plt to represent instructions in .plt section. |
| 268 | for (auto it = elf->section_begin(); it != elf->section_end(); ++it) { |
| 269 | const llvm::object::ELFSectionRef& section_ref = *it; |
| 270 | llvm::StringRef section_name; |
| 271 | std::error_code err = section_ref.getName(section_name); |
| 272 | if (err || section_name != ".plt") { |
| 273 | continue; |
| 274 | } |
| 275 | const auto* shdr = elf->getSection(section_ref.getRawDataRefImpl()); |
| 276 | if (shdr == nullptr) { |
| 277 | return; |
| 278 | } |
| 279 | ElfFileSymbol symbol; |
| 280 | symbol.vaddr = shdr->sh_addr; |
| 281 | symbol.len = shdr->sh_size; |
| 282 | symbol.is_func = true; |
| 283 | symbol.is_label = true; |
| 284 | symbol.is_in_text_section = true; |
| 285 | symbol.name = "@plt"; |
| 286 | callback(symbol); |
| 287 | return; |
| 288 | } |
| 289 | } |
| 290 | |
Yabin Cui | 5783fa0 | 2016-07-06 18:29:00 -0700 | [diff] [blame] | 291 | template <class ELFT> |
Thiébaud Weksteen | 4848ee0 | 2020-10-23 16:06:59 +0200 | [diff] [blame] | 292 | void CheckSymbolSections(const llvm::object::ELFObjectFile<ELFT>* elf, bool* has_symtab, |
| 293 | bool* has_dynsym) { |
Yabin Cui | e4e13f3 | 2017-12-15 11:36:51 -0800 | [diff] [blame] | 294 | *has_symtab = false; |
| 295 | *has_dynsym = false; |
| 296 | for (auto it = elf->section_begin(); it != elf->section_end(); ++it) { |
| 297 | const llvm::object::ELFSectionRef& section_ref = *it; |
| 298 | llvm::StringRef section_name; |
| 299 | std::error_code err = section_ref.getName(section_name); |
| 300 | if (err) { |
| 301 | continue; |
| 302 | } |
| 303 | if (section_name == ".dynsym") { |
| 304 | *has_dynsym = true; |
| 305 | } else if (section_name == ".symtab") { |
| 306 | *has_symtab = true; |
| 307 | } |
| 308 | } |
| 309 | } |
| 310 | |
Yabin Cui | 02e2033 | 2020-03-16 19:38:23 -0700 | [diff] [blame] | 311 | template <typename T> |
| 312 | class ElfFileImpl {}; |
| 313 | |
| 314 | template <typename ELFT> |
Yabin Cui | 3a88045 | 2020-06-29 16:37:31 -0700 | [diff] [blame] | 315 | class ElfFileImpl<llvm::object::ELFObjectFile<ELFT>> : public ElfFile { |
Yabin Cui | 02e2033 | 2020-03-16 19:38:23 -0700 | [diff] [blame] | 316 | public: |
Yabin Cui | 3a88045 | 2020-06-29 16:37:31 -0700 | [diff] [blame] | 317 | ElfFileImpl(BinaryWrapper&& wrapper, const llvm::object::ELFObjectFile<ELFT>* elf_obj) |
| 318 | : wrapper_(std::move(wrapper)), elf_obj_(elf_obj), elf_(elf_obj->getELFFile()) {} |
Yabin Cui | 02e2033 | 2020-03-16 19:38:23 -0700 | [diff] [blame] | 319 | |
Thiébaud Weksteen | 4848ee0 | 2020-10-23 16:06:59 +0200 | [diff] [blame] | 320 | bool Is64Bit() override { return elf_->getHeader()->getFileClass() == llvm::ELF::ELFCLASS64; } |
Yabin Cui | 4ad10fb | 2020-04-01 15:45:48 -0700 | [diff] [blame] | 321 | |
Thiébaud Weksteen | 4848ee0 | 2020-10-23 16:06:59 +0200 | [diff] [blame] | 322 | llvm::MemoryBuffer* GetMemoryBuffer() override { return wrapper_.buffer.get(); } |
Yabin Cui | 02e2033 | 2020-03-16 19:38:23 -0700 | [diff] [blame] | 323 | |
Yabin Cui | 4ad10fb | 2020-04-01 15:45:48 -0700 | [diff] [blame] | 324 | std::vector<ElfSegment> GetProgramHeader() override { |
| 325 | auto program_headers = elf_->program_headers(); |
| 326 | std::vector<ElfSegment> segments(program_headers.size()); |
| 327 | for (size_t i = 0; i < program_headers.size(); i++) { |
Yabin Cui | 7078c67 | 2020-11-10 16:24:12 -0800 | [diff] [blame] | 328 | const auto& phdr = program_headers[i]; |
Yabin Cui | 4ad10fb | 2020-04-01 15:45:48 -0700 | [diff] [blame] | 329 | segments[i].vaddr = phdr.p_vaddr; |
| 330 | segments[i].file_offset = phdr.p_offset; |
| 331 | segments[i].file_size = phdr.p_filesz; |
| 332 | segments[i].is_executable = |
| 333 | (phdr.p_type == llvm::ELF::PT_LOAD) && (phdr.p_flags & llvm::ELF::PF_X); |
Yabin Cui | 4e4cbbc | 2020-12-14 15:26:19 -0800 | [diff] [blame] | 334 | segments[i].is_load = (phdr.p_type == llvm::ELF::PT_LOAD); |
Yabin Cui | 4ad10fb | 2020-04-01 15:45:48 -0700 | [diff] [blame] | 335 | } |
| 336 | return segments; |
| 337 | } |
| 338 | |
Yabin Cui | 7078c67 | 2020-11-10 16:24:12 -0800 | [diff] [blame] | 339 | std::vector<ElfSection> GetSectionHeader() override { |
| 340 | auto section_headers_or_err = elf_->sections(); |
| 341 | if (!section_headers_or_err) { |
| 342 | return {}; |
| 343 | } |
| 344 | const auto& section_headers = section_headers_or_err.get(); |
| 345 | std::vector<ElfSection> sections(section_headers.size()); |
| 346 | for (size_t i = 0; i < section_headers.size(); i++) { |
| 347 | const auto& shdr = section_headers[i]; |
| 348 | if (auto name = elf_->getSectionName(&shdr); name) { |
| 349 | sections[i].name = name.get(); |
| 350 | } |
| 351 | sections[i].vaddr = shdr.sh_addr; |
| 352 | sections[i].file_offset = shdr.sh_offset; |
Yabin Cui | 16f41ff | 2021-03-23 14:58:25 -0700 | [diff] [blame] | 353 | sections[i].size = shdr.sh_size; |
Yabin Cui | 7078c67 | 2020-11-10 16:24:12 -0800 | [diff] [blame] | 354 | } |
| 355 | return sections; |
| 356 | } |
| 357 | |
Yabin Cui | 3a88045 | 2020-06-29 16:37:31 -0700 | [diff] [blame] | 358 | ElfStatus GetBuildId(BuildId* build_id) override { |
| 359 | llvm::StringRef data = elf_obj_->getData(); |
| 360 | const char* binary_start = data.data(); |
| 361 | const char* binary_end = data.data() + data.size(); |
| 362 | for (auto it = elf_obj_->section_begin(); it != elf_obj_->section_end(); ++it) { |
| 363 | const llvm::object::ELFSectionRef& section_ref = *it; |
| 364 | if (section_ref.getType() == llvm::ELF::SHT_NOTE) { |
| 365 | if (it->getContents(data)) { |
| 366 | return ElfStatus::READ_FAILED; |
| 367 | } |
| 368 | if (data.data() < binary_start || data.data() + data.size() > binary_end) { |
| 369 | return ElfStatus::NO_BUILD_ID; |
| 370 | } |
| 371 | if (GetBuildIdFromNoteSection(data.data(), data.size(), build_id)) { |
| 372 | return ElfStatus::NO_ERROR; |
| 373 | } |
| 374 | } |
| 375 | } |
| 376 | return ElfStatus::NO_BUILD_ID; |
| 377 | } |
| 378 | |
Yabin Cui | 0194703 | 2020-06-30 14:36:46 -0700 | [diff] [blame] | 379 | ElfStatus ParseSymbols(const ParseSymbolCallback& callback) override { |
| 380 | auto machine = elf_->getHeader()->e_machine; |
| 381 | bool is_arm = (machine == llvm::ELF::EM_ARM || machine == llvm::ELF::EM_AARCH64); |
| 382 | AddSymbolForPltSection(elf_obj_, callback); |
| 383 | // Some applications deliberately ship elf files with broken section tables. |
| 384 | // So check the existence of .symtab section and .dynsym section before reading symbols. |
| 385 | bool has_symtab; |
| 386 | bool has_dynsym; |
| 387 | CheckSymbolSections(elf_obj_, &has_symtab, &has_dynsym); |
| 388 | if (has_symtab && elf_obj_->symbol_begin() != elf_obj_->symbol_end()) { |
| 389 | ReadSymbolTable(elf_obj_->symbol_begin(), elf_obj_->symbol_end(), callback, is_arm, |
| 390 | elf_obj_->section_end()); |
| 391 | return ElfStatus::NO_ERROR; |
| 392 | } else if (has_dynsym && elf_obj_->dynamic_symbol_begin()->getRawDataRefImpl() != |
| 393 | llvm::object::DataRefImpl()) { |
| 394 | ReadSymbolTable(elf_obj_->dynamic_symbol_begin(), elf_obj_->dynamic_symbol_end(), callback, |
| 395 | is_arm, elf_obj_->section_end()); |
| 396 | } |
| 397 | std::string debugdata; |
| 398 | ElfStatus result = ReadSection(".gnu_debugdata", &debugdata); |
| 399 | if (result == ElfStatus::SECTION_NOT_FOUND) { |
| 400 | return ElfStatus::NO_SYMBOL_TABLE; |
| 401 | } else if (result == ElfStatus::NO_ERROR) { |
| 402 | std::string decompressed_data; |
| 403 | if (XzDecompress(debugdata, &decompressed_data)) { |
| 404 | auto debugdata_elf = |
| 405 | ElfFile::Open(decompressed_data.data(), decompressed_data.size(), &result); |
| 406 | if (debugdata_elf) { |
| 407 | return debugdata_elf->ParseSymbols(callback); |
| 408 | } |
| 409 | } |
| 410 | } |
| 411 | return result; |
| 412 | } |
| 413 | |
| 414 | void ParseDynamicSymbols(const ParseSymbolCallback& callback) override { |
| 415 | auto machine = elf_->getHeader()->e_machine; |
| 416 | bool is_arm = (machine == llvm::ELF::EM_ARM || machine == llvm::ELF::EM_AARCH64); |
| 417 | ReadSymbolTable(elf_obj_->dynamic_symbol_begin(), elf_obj_->dynamic_symbol_end(), callback, |
| 418 | is_arm, elf_obj_->section_end()); |
| 419 | } |
| 420 | |
| 421 | ElfStatus ReadSection(const std::string& section_name, std::string* content) override { |
| 422 | for (llvm::object::section_iterator it = elf_obj_->section_begin(); |
| 423 | it != elf_obj_->section_end(); ++it) { |
| 424 | llvm::StringRef name; |
| 425 | if (it->getName(name) || name != section_name) { |
| 426 | continue; |
| 427 | } |
| 428 | llvm::StringRef data; |
| 429 | std::error_code err = it->getContents(data); |
| 430 | if (err) { |
| 431 | return ElfStatus::READ_FAILED; |
| 432 | } |
| 433 | *content = data; |
| 434 | return ElfStatus::NO_ERROR; |
| 435 | } |
| 436 | return ElfStatus::SECTION_NOT_FOUND; |
| 437 | } |
| 438 | |
Yabin Cui | 90c3b30 | 2020-07-01 10:09:16 -0700 | [diff] [blame] | 439 | uint64_t ReadMinExecutableVaddr(uint64_t* file_offset) { |
| 440 | bool has_vaddr = false; |
| 441 | uint64_t min_addr = std::numeric_limits<uint64_t>::max(); |
| 442 | for (auto it = elf_->program_header_begin(); it != elf_->program_header_end(); ++it) { |
| 443 | if ((it->p_type == llvm::ELF::PT_LOAD) && (it->p_flags & llvm::ELF::PF_X)) { |
| 444 | if (it->p_vaddr < min_addr) { |
| 445 | min_addr = it->p_vaddr; |
| 446 | *file_offset = it->p_offset; |
| 447 | has_vaddr = true; |
| 448 | } |
| 449 | } |
| 450 | } |
| 451 | if (!has_vaddr) { |
| 452 | // JIT symfiles don't have program headers. |
| 453 | min_addr = 0; |
| 454 | *file_offset = 0; |
| 455 | } |
| 456 | return min_addr; |
| 457 | } |
| 458 | |
Yabin Cui | b60b33c | 2020-07-15 11:58:41 -0700 | [diff] [blame] | 459 | bool VaddrToOff(uint64_t vaddr, uint64_t* file_offset) override { |
| 460 | for (auto ph = elf_->program_header_begin(); ph != elf_->program_header_end(); ++ph) { |
| 461 | if (ph->p_type == llvm::ELF::PT_LOAD && vaddr >= ph->p_vaddr && |
| 462 | vaddr < ph->p_vaddr + ph->p_filesz) { |
| 463 | *file_offset = vaddr - ph->p_vaddr + ph->p_offset; |
| 464 | return true; |
| 465 | } |
| 466 | } |
| 467 | return false; |
| 468 | } |
| 469 | |
Yabin Cui | 02e2033 | 2020-03-16 19:38:23 -0700 | [diff] [blame] | 470 | private: |
| 471 | BinaryWrapper wrapper_; |
Yabin Cui | 3a88045 | 2020-06-29 16:37:31 -0700 | [diff] [blame] | 472 | const llvm::object::ELFObjectFile<ELFT>* elf_obj_; |
Yabin Cui | 02e2033 | 2020-03-16 19:38:23 -0700 | [diff] [blame] | 473 | const llvm::object::ELFFile<ELFT>* elf_; |
| 474 | }; |
| 475 | |
Yabin Cui | 0194703 | 2020-06-30 14:36:46 -0700 | [diff] [blame] | 476 | std::unique_ptr<ElfFile> CreateElfFileImpl(BinaryWrapper&& wrapper, ElfStatus* status) { |
| 477 | if (auto obj = llvm::dyn_cast<llvm::object::ELF32LEObjectFile>(wrapper.obj)) { |
| 478 | return std::unique_ptr<ElfFile>( |
| 479 | new ElfFileImpl<llvm::object::ELF32LEObjectFile>(std::move(wrapper), obj)); |
| 480 | } |
| 481 | if (auto obj = llvm::dyn_cast<llvm::object::ELF64LEObjectFile>(wrapper.obj)) { |
| 482 | return std::unique_ptr<ElfFile>( |
| 483 | new ElfFileImpl<llvm::object::ELF64LEObjectFile>(std::move(wrapper), obj)); |
| 484 | } |
| 485 | *status = ElfStatus::FILE_MALFORMED; |
| 486 | return nullptr; |
| 487 | } |
| 488 | |
Yabin Cui | 02e2033 | 2020-03-16 19:38:23 -0700 | [diff] [blame] | 489 | } // namespace |
| 490 | |
Yabin Cui | b60b33c | 2020-07-15 11:58:41 -0700 | [diff] [blame] | 491 | std::unique_ptr<ElfFile> ElfFile::Open(const std::string& filename) { |
| 492 | ElfStatus status; |
| 493 | auto elf = Open(filename, &status); |
| 494 | if (!elf) { |
| 495 | LOG(ERROR) << "failed to open " << filename << ": " << status; |
| 496 | } |
| 497 | return elf; |
| 498 | } |
| 499 | |
Yabin Cui | 0194703 | 2020-06-30 14:36:46 -0700 | [diff] [blame] | 500 | std::unique_ptr<ElfFile> ElfFile::Open(const std::string& filename, |
| 501 | const BuildId* expected_build_id, ElfStatus* status) { |
Yabin Cui | 02e2033 | 2020-03-16 19:38:23 -0700 | [diff] [blame] | 502 | BinaryWrapper wrapper; |
| 503 | auto tuple = SplitUrlInApk(filename); |
| 504 | if (std::get<0>(tuple)) { |
| 505 | EmbeddedElf* elf = ApkInspector::FindElfInApkByName(std::get<1>(tuple), std::get<2>(tuple)); |
| 506 | if (elf == nullptr) { |
| 507 | *status = ElfStatus::FILE_NOT_FOUND; |
| 508 | } else { |
| 509 | *status = OpenObjectFile(elf->filepath(), elf->entry_offset(), elf->entry_size(), &wrapper); |
| 510 | } |
Yabin Cui | 9ba4d94 | 2020-09-08 16:12:46 -0700 | [diff] [blame] | 511 | } else if (JITDebugReader::IsPathInJITSymFile(filename)) { |
| 512 | size_t colon_pos = filename.rfind(':'); |
| 513 | CHECK_NE(colon_pos, std::string::npos); |
Yabin Cui | e32ed2b | 2020-07-23 15:30:14 -0700 | [diff] [blame] | 514 | // path generated by JITDebugReader: app_jit_cache:<file_start>-<file_end> |
| 515 | uint64_t file_start; |
| 516 | uint64_t file_end; |
| 517 | if (sscanf(filename.data() + colon_pos, ":%" PRIu64 "-%" PRIu64, &file_start, &file_end) != 2) { |
| 518 | *status = ElfStatus::FILE_NOT_FOUND; |
| 519 | return nullptr; |
| 520 | } |
| 521 | *status = |
| 522 | OpenObjectFile(filename.substr(0, colon_pos), file_start, file_end - file_start, &wrapper); |
Yabin Cui | 02e2033 | 2020-03-16 19:38:23 -0700 | [diff] [blame] | 523 | } else { |
| 524 | *status = OpenObjectFile(filename, 0, 0, &wrapper); |
| 525 | } |
Yabin Cui | 0194703 | 2020-06-30 14:36:46 -0700 | [diff] [blame] | 526 | if (*status != ElfStatus::NO_ERROR) { |
| 527 | return nullptr; |
Yabin Cui | 02e2033 | 2020-03-16 19:38:23 -0700 | [diff] [blame] | 528 | } |
Yabin Cui | 0194703 | 2020-06-30 14:36:46 -0700 | [diff] [blame] | 529 | auto elf = CreateElfFileImpl(std::move(wrapper), status); |
| 530 | if (elf && expected_build_id != nullptr && !expected_build_id->IsEmpty()) { |
| 531 | BuildId real_build_id; |
| 532 | *status = elf->GetBuildId(&real_build_id); |
| 533 | if (*status != ElfStatus::NO_ERROR) { |
| 534 | return nullptr; |
| 535 | } |
| 536 | if (*expected_build_id != real_build_id) { |
| 537 | *status = ElfStatus::BUILD_ID_MISMATCH; |
| 538 | return nullptr; |
| 539 | } |
| 540 | } |
| 541 | return elf; |
| 542 | } |
| 543 | |
| 544 | std::unique_ptr<ElfFile> ElfFile::Open(const char* data, size_t size, ElfStatus* status) { |
| 545 | BinaryWrapper wrapper; |
| 546 | *status = OpenObjectFileInMemory(data, size, &wrapper); |
| 547 | if (*status != ElfStatus::NO_ERROR) { |
| 548 | return nullptr; |
| 549 | } |
| 550 | return CreateElfFileImpl(std::move(wrapper), status); |
Yabin Cui | 02e2033 | 2020-03-16 19:38:23 -0700 | [diff] [blame] | 551 | } |
| 552 | |
Yabin Cui | e55f032 | 2020-05-27 16:26:45 -0700 | [diff] [blame] | 553 | } // namespace simpleperf |
| 554 | |
Yabin Cui | e55f032 | 2020-05-27 16:26:45 -0700 | [diff] [blame] | 555 | // LLVM libraries uses ncurses library, but that isn't needed by simpleperf. |
| 556 | // So support a naive implementation to avoid depending on ncurses. |
Thiébaud Weksteen | 4848ee0 | 2020-10-23 16:06:59 +0200 | [diff] [blame] | 557 | __attribute__((weak)) extern "C" int setupterm(char*, int, int*) { |
Yabin Cui | e55f032 | 2020-05-27 16:26:45 -0700 | [diff] [blame] | 558 | return -1; |
| 559 | } |
| 560 | |
Thiébaud Weksteen | 4848ee0 | 2020-10-23 16:06:59 +0200 | [diff] [blame] | 561 | __attribute__((weak)) extern "C" struct term* set_curterm(struct term*) { |
Yabin Cui | e55f032 | 2020-05-27 16:26:45 -0700 | [diff] [blame] | 562 | return nullptr; |
| 563 | } |
| 564 | |
Thiébaud Weksteen | 4848ee0 | 2020-10-23 16:06:59 +0200 | [diff] [blame] | 565 | __attribute__((weak)) extern "C" int del_curterm(struct term*) { |
Yabin Cui | e55f032 | 2020-05-27 16:26:45 -0700 | [diff] [blame] | 566 | return -1; |
| 567 | } |
| 568 | |
Thiébaud Weksteen | 4848ee0 | 2020-10-23 16:06:59 +0200 | [diff] [blame] | 569 | __attribute__((weak)) extern "C" int tigetnum(char*) { |
Yabin Cui | e55f032 | 2020-05-27 16:26:45 -0700 | [diff] [blame] | 570 | return -1; |
| 571 | } |