Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 1 | //===-- ObjectFile.h --------------------------------------------*- C++ -*-===// |
| 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 | |
Olivier Deprez | f4ef2d0 | 2021-04-20 13:36:24 +0200 | [diff] [blame] | 9 | #ifndef LLDB_SYMBOL_OBJECTFILE_H |
| 10 | #define LLDB_SYMBOL_OBJECTFILE_H |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 11 | |
| 12 | #include "lldb/Core/FileSpecList.h" |
| 13 | #include "lldb/Core/ModuleChild.h" |
| 14 | #include "lldb/Core/PluginInterface.h" |
| 15 | #include "lldb/Symbol/Symtab.h" |
| 16 | #include "lldb/Symbol/UnwindTable.h" |
| 17 | #include "lldb/Utility/DataExtractor.h" |
| 18 | #include "lldb/Utility/Endian.h" |
| 19 | #include "lldb/Utility/FileSpec.h" |
| 20 | #include "lldb/Utility/UUID.h" |
| 21 | #include "lldb/lldb-private.h" |
| 22 | #include "llvm/Support/VersionTuple.h" |
| 23 | |
| 24 | namespace lldb_private { |
| 25 | |
| 26 | class ObjectFileJITDelegate { |
| 27 | public: |
| 28 | ObjectFileJITDelegate() {} |
| 29 | |
| 30 | virtual ~ObjectFileJITDelegate() {} |
| 31 | |
| 32 | virtual lldb::ByteOrder GetByteOrder() const = 0; |
| 33 | |
| 34 | virtual uint32_t GetAddressByteSize() const = 0; |
| 35 | |
| 36 | virtual void PopulateSymtab(lldb_private::ObjectFile *obj_file, |
| 37 | lldb_private::Symtab &symtab) = 0; |
| 38 | |
| 39 | virtual void PopulateSectionList(lldb_private::ObjectFile *obj_file, |
| 40 | lldb_private::SectionList §ion_list) = 0; |
| 41 | |
| 42 | virtual ArchSpec GetArchitecture() = 0; |
| 43 | }; |
| 44 | |
| 45 | /// \class ObjectFile ObjectFile.h "lldb/Symbol/ObjectFile.h" |
| 46 | /// A plug-in interface definition class for object file parsers. |
| 47 | /// |
| 48 | /// Object files belong to Module objects and know how to extract information |
| 49 | /// from executable, shared library, and object (.o) files used by operating |
| 50 | /// system runtime. The symbol table and section list for an object file. |
| 51 | /// |
| 52 | /// Object files can be represented by the entire file, or by part of a file. |
| 53 | /// An example of a partial file ObjectFile is one that contains information |
| 54 | /// for one of multiple architectures in the same file. |
| 55 | /// |
| 56 | /// Once an architecture is selected the object file information can be |
| 57 | /// extracted from this abstract class. |
| 58 | class ObjectFile : public std::enable_shared_from_this<ObjectFile>, |
| 59 | public PluginInterface, |
| 60 | public ModuleChild { |
| 61 | friend class lldb_private::Module; |
| 62 | |
| 63 | public: |
| 64 | enum Type { |
| 65 | eTypeInvalid = 0, |
Olivier Deprez | f4ef2d0 | 2021-04-20 13:36:24 +0200 | [diff] [blame] | 66 | /// A core file that has a checkpoint of a program's execution state. |
| 67 | eTypeCoreFile, |
| 68 | /// A normal executable. |
| 69 | eTypeExecutable, |
| 70 | /// An object file that contains only debug information. |
| 71 | eTypeDebugInfo, |
| 72 | /// The platform's dynamic linker executable. |
| 73 | eTypeDynamicLinker, |
| 74 | /// An intermediate object file. |
| 75 | eTypeObjectFile, |
| 76 | /// A shared library that can be used during execution. |
| 77 | eTypeSharedLibrary, |
| 78 | /// A library that can be linked against but not used for execution. |
| 79 | eTypeStubLibrary, |
| 80 | /// JIT code that has symbols, sections and possibly debug info. |
| 81 | eTypeJIT, |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 82 | eTypeUnknown |
| 83 | }; |
| 84 | |
| 85 | enum Strata { |
| 86 | eStrataInvalid = 0, |
| 87 | eStrataUnknown, |
| 88 | eStrataUser, |
| 89 | eStrataKernel, |
| 90 | eStrataRawImage, |
| 91 | eStrataJIT |
| 92 | }; |
| 93 | |
Olivier Deprez | f4ef2d0 | 2021-04-20 13:36:24 +0200 | [diff] [blame] | 94 | /// If we have a corefile binary hint, this enum |
| 95 | /// specifies the binary type which we can use to |
| 96 | /// select the correct DynamicLoader plugin. |
| 97 | enum BinaryType { |
| 98 | eBinaryTypeInvalid = 0, |
| 99 | eBinaryTypeUnknown, |
| 100 | eBinaryTypeKernel, /// kernel binary |
| 101 | eBinaryTypeUser, /// user process binary |
| 102 | eBinaryTypeStandalone /// standalone binary / firmware |
| 103 | }; |
| 104 | |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 105 | struct LoadableData { |
| 106 | lldb::addr_t Dest; |
| 107 | llvm::ArrayRef<uint8_t> Contents; |
| 108 | }; |
| 109 | |
| 110 | /// Construct with a parent module, offset, and header data. |
| 111 | /// |
| 112 | /// Object files belong to modules and a valid module must be supplied upon |
| 113 | /// construction. The at an offset within a file for objects that contain |
| 114 | /// more than one architecture or object. |
| 115 | ObjectFile(const lldb::ModuleSP &module_sp, const FileSpec *file_spec_ptr, |
| 116 | lldb::offset_t file_offset, lldb::offset_t length, |
| 117 | const lldb::DataBufferSP &data_sp, lldb::offset_t data_offset); |
| 118 | |
| 119 | ObjectFile(const lldb::ModuleSP &module_sp, const lldb::ProcessSP &process_sp, |
| 120 | lldb::addr_t header_addr, lldb::DataBufferSP &data_sp); |
| 121 | |
| 122 | /// Destructor. |
| 123 | /// |
| 124 | /// The destructor is virtual since this class is designed to be inherited |
| 125 | /// from by the plug-in instance. |
| 126 | ~ObjectFile() override; |
| 127 | |
| 128 | /// Dump a description of this object to a Stream. |
| 129 | /// |
| 130 | /// Dump a description of the current contents of this object to the |
| 131 | /// supplied stream \a s. The dumping should include the section list if it |
| 132 | /// has been parsed, and the symbol table if it has been parsed. |
| 133 | /// |
| 134 | /// \param[in] s |
| 135 | /// The stream to which to dump the object description. |
| 136 | virtual void Dump(Stream *s) = 0; |
| 137 | |
| 138 | /// Find a ObjectFile plug-in that can parse \a file_spec. |
| 139 | /// |
| 140 | /// Scans all loaded plug-in interfaces that implement versions of the |
| 141 | /// ObjectFile plug-in interface and returns the first instance that can |
| 142 | /// parse the file. |
| 143 | /// |
Olivier Deprez | f4ef2d0 | 2021-04-20 13:36:24 +0200 | [diff] [blame] | 144 | /// \param[in] module_sp |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 145 | /// The parent module that owns this object file. |
| 146 | /// |
| 147 | /// \param[in] file_spec |
| 148 | /// A file specification that indicates which file to use as the |
| 149 | /// object file. |
| 150 | /// |
| 151 | /// \param[in] file_offset |
| 152 | /// The offset into the file at which to start parsing the |
| 153 | /// object. This is for files that contain multiple |
| 154 | /// architectures or objects. |
| 155 | /// |
| 156 | /// \param[in] file_size |
| 157 | /// The size of the current object file if it can be determined |
| 158 | /// or if it is known. This can be zero. |
| 159 | /// |
| 160 | /// \see ObjectFile::ParseHeader() |
| 161 | static lldb::ObjectFileSP |
| 162 | FindPlugin(const lldb::ModuleSP &module_sp, const FileSpec *file_spec, |
| 163 | lldb::offset_t file_offset, lldb::offset_t file_size, |
| 164 | lldb::DataBufferSP &data_sp, lldb::offset_t &data_offset); |
| 165 | |
| 166 | /// Find a ObjectFile plug-in that can parse a file in memory. |
| 167 | /// |
| 168 | /// Scans all loaded plug-in interfaces that implement versions of the |
| 169 | /// ObjectFile plug-in interface and returns the first instance that can |
| 170 | /// parse the file. |
| 171 | /// |
Olivier Deprez | f4ef2d0 | 2021-04-20 13:36:24 +0200 | [diff] [blame] | 172 | /// \param[in] module_sp |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 173 | /// The parent module that owns this object file. |
| 174 | /// |
| 175 | /// \param[in] process_sp |
| 176 | /// A shared pointer to the process whose memory space contains |
| 177 | /// an object file. This will be stored as a std::weak_ptr. |
| 178 | /// |
| 179 | /// \param[in] header_addr |
| 180 | /// The address of the header for the object file in memory. |
| 181 | static lldb::ObjectFileSP FindPlugin(const lldb::ModuleSP &module_sp, |
| 182 | const lldb::ProcessSP &process_sp, |
| 183 | lldb::addr_t header_addr, |
| 184 | lldb::DataBufferSP &file_data_sp); |
| 185 | |
Olivier Deprez | f4ef2d0 | 2021-04-20 13:36:24 +0200 | [diff] [blame] | 186 | static size_t |
| 187 | GetModuleSpecifications(const FileSpec &file, lldb::offset_t file_offset, |
| 188 | lldb::offset_t file_size, ModuleSpecList &specs, |
| 189 | lldb::DataBufferSP data_sp = lldb::DataBufferSP()); |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 190 | |
| 191 | static size_t GetModuleSpecifications(const lldb_private::FileSpec &file, |
| 192 | lldb::DataBufferSP &data_sp, |
| 193 | lldb::offset_t data_offset, |
| 194 | lldb::offset_t file_offset, |
| 195 | lldb::offset_t file_size, |
| 196 | lldb_private::ModuleSpecList &specs); |
| 197 | /// Split a path into a file path with object name. |
| 198 | /// |
| 199 | /// For paths like "/tmp/foo.a(bar.o)" we often need to split a path up into |
| 200 | /// the actual path name and into the object name so we can make a valid |
| 201 | /// object file from it. |
| 202 | /// |
| 203 | /// \param[in] path_with_object |
| 204 | /// A path that might contain an archive path with a .o file |
| 205 | /// specified in parens in the basename of the path. |
| 206 | /// |
| 207 | /// \param[out] archive_file |
| 208 | /// If \b true is returned, \a file_spec will be filled in with |
| 209 | /// the path to the archive. |
| 210 | /// |
| 211 | /// \param[out] archive_object |
| 212 | /// If \b true is returned, \a object will be filled in with |
| 213 | /// the name of the object inside the archive. |
| 214 | /// |
| 215 | /// \return |
| 216 | /// \b true if the path matches the pattern of archive + object |
| 217 | /// and \a archive_file and \a archive_object are modified, |
| 218 | /// \b false otherwise and \a archive_file and \a archive_object |
| 219 | /// are guaranteed to be remain unchanged. |
| 220 | static bool SplitArchivePathWithObject( |
Olivier Deprez | f4ef2d0 | 2021-04-20 13:36:24 +0200 | [diff] [blame] | 221 | llvm::StringRef path_with_object, lldb_private::FileSpec &archive_file, |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 222 | lldb_private::ConstString &archive_object, bool must_exist); |
| 223 | |
Olivier Deprez | f4ef2d0 | 2021-04-20 13:36:24 +0200 | [diff] [blame] | 224 | // LLVM RTTI support |
| 225 | static char ID; |
| 226 | virtual bool isA(const void *ClassID) const { return ClassID == &ID; } |
| 227 | |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 228 | /// Gets the address size in bytes for the current object file. |
| 229 | /// |
| 230 | /// \return |
| 231 | /// The size of an address in bytes for the currently selected |
| 232 | /// architecture (and object for archives). Returns zero if no |
| 233 | /// architecture or object has been selected. |
| 234 | virtual uint32_t GetAddressByteSize() const = 0; |
| 235 | |
| 236 | /// Get the address type given a file address in an object file. |
| 237 | /// |
| 238 | /// Many binary file formats know what kinds This is primarily for ARM |
| 239 | /// binaries, though it can be applied to any executable file format that |
| 240 | /// supports different opcode types within the same binary. ARM binaries |
| 241 | /// support having both ARM and Thumb within the same executable container. |
| 242 | /// We need to be able to get \return |
| 243 | /// The size of an address in bytes for the currently selected |
| 244 | /// architecture (and object for archives). Returns zero if no |
| 245 | /// architecture or object has been selected. |
| 246 | virtual AddressClass GetAddressClass(lldb::addr_t file_addr); |
| 247 | |
| 248 | /// Extract the dependent modules from an object file. |
| 249 | /// |
| 250 | /// If an object file has information about which other images it depends on |
| 251 | /// (such as shared libraries), this function will provide the list. Since |
| 252 | /// many executables or shared libraries may depend on the same files, |
| 253 | /// FileSpecList::AppendIfUnique(const FileSpec &) should be used to make |
| 254 | /// sure any files that are added are not already in the list. |
| 255 | /// |
| 256 | /// \param[out] file_list |
| 257 | /// A list of file specification objects that gets dependent |
| 258 | /// files appended to. |
| 259 | /// |
| 260 | /// \return |
| 261 | /// The number of new files that were appended to \a file_list. |
| 262 | /// |
| 263 | /// \see FileSpecList::AppendIfUnique(const FileSpec &) |
| 264 | virtual uint32_t GetDependentModules(FileSpecList &file_list) = 0; |
| 265 | |
| 266 | /// Tells whether this object file is capable of being the main executable |
| 267 | /// for a process. |
| 268 | /// |
| 269 | /// \return |
| 270 | /// \b true if it is, \b false otherwise. |
| 271 | virtual bool IsExecutable() const = 0; |
| 272 | |
| 273 | /// Returns the offset into a file at which this object resides. |
| 274 | /// |
| 275 | /// Some files contain many object files, and this function allows access to |
| 276 | /// an object's offset within the file. |
| 277 | /// |
| 278 | /// \return |
| 279 | /// The offset in bytes into the file. Defaults to zero for |
| 280 | /// simple object files that a represented by an entire file. |
| 281 | virtual lldb::addr_t GetFileOffset() const { return m_file_offset; } |
| 282 | |
| 283 | virtual lldb::addr_t GetByteSize() const { return m_length; } |
| 284 | |
| 285 | /// Get accessor to the object file specification. |
| 286 | /// |
| 287 | /// \return |
| 288 | /// The file specification object pointer if there is one, or |
| 289 | /// NULL if this object is only from memory. |
| 290 | virtual FileSpec &GetFileSpec() { return m_file; } |
| 291 | |
| 292 | /// Get const accessor to the object file specification. |
| 293 | /// |
| 294 | /// \return |
| 295 | /// The const file specification object pointer if there is one, |
| 296 | /// or NULL if this object is only from memory. |
| 297 | virtual const FileSpec &GetFileSpec() const { return m_file; } |
| 298 | |
| 299 | /// Get the ArchSpec for this object file. |
| 300 | /// |
| 301 | /// \return |
| 302 | /// The ArchSpec of this object file. In case of error, an invalid |
| 303 | /// ArchSpec object is returned. |
| 304 | virtual ArchSpec GetArchitecture() = 0; |
| 305 | |
| 306 | /// Gets the section list for the currently selected architecture (and |
| 307 | /// object for archives). |
| 308 | /// |
| 309 | /// Section list parsing can be deferred by ObjectFile instances until this |
| 310 | /// accessor is called the first time. |
| 311 | /// |
| 312 | /// \return |
| 313 | /// The list of sections contained in this object file. |
| 314 | virtual SectionList *GetSectionList(bool update_module_section_list = true); |
| 315 | |
| 316 | virtual void CreateSections(SectionList &unified_section_list) = 0; |
| 317 | |
| 318 | /// Notify the ObjectFile that the file addresses in the Sections for this |
| 319 | /// module have been changed. |
| 320 | virtual void SectionFileAddressesChanged() {} |
| 321 | |
| 322 | /// Gets the symbol table for the currently selected architecture (and |
| 323 | /// object for archives). |
| 324 | /// |
| 325 | /// Symbol table parsing can be deferred by ObjectFile instances until this |
| 326 | /// accessor is called the first time. |
| 327 | /// |
| 328 | /// \return |
| 329 | /// The symbol table for this object file. |
| 330 | virtual Symtab *GetSymtab() = 0; |
| 331 | |
| 332 | /// Perform relocations on the section if necessary. |
| 333 | /// |
| 334 | virtual void RelocateSection(lldb_private::Section *section); |
| 335 | |
| 336 | /// Appends a Symbol for the specified so_addr to the symbol table. |
| 337 | /// |
| 338 | /// If verify_unique is false, the symbol table is not searched to determine |
| 339 | /// if a Symbol found at this address has already been added to the symbol |
| 340 | /// table. When verify_unique is true, this method resolves the Symbol as |
| 341 | /// the first match in the SymbolTable and appends a Symbol only if |
| 342 | /// required/found. |
| 343 | /// |
| 344 | /// \return |
| 345 | /// The resolved symbol or nullptr. Returns nullptr if a |
| 346 | /// a Symbol could not be found for the specified so_addr. |
| 347 | virtual Symbol *ResolveSymbolForAddress(const Address &so_addr, |
| 348 | bool verify_unique) { |
| 349 | // Typically overridden to lazily add stripped symbols recoverable from the |
| 350 | // exception handling unwind information (i.e. without parsing the entire |
| 351 | // eh_frame section. |
| 352 | // |
| 353 | // The availability of LC_FUNCTION_STARTS allows ObjectFileMachO to |
| 354 | // efficiently add stripped symbols when the symbol table is first |
| 355 | // constructed. Poorer cousins are PECoff and ELF. |
| 356 | return nullptr; |
| 357 | } |
| 358 | |
| 359 | /// Detect if this object file has been stripped of local symbols. |
| 360 | /// Detect if this object file has been stripped of local symbols. |
| 361 | /// |
| 362 | /// \return |
| 363 | /// Return \b true if the object file has been stripped of local |
| 364 | /// symbols. |
| 365 | virtual bool IsStripped() = 0; |
| 366 | |
| 367 | /// Frees the symbol table. |
| 368 | /// |
| 369 | /// This function should only be used when an object file is |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 370 | virtual void ClearSymtab(); |
| 371 | |
| 372 | /// Gets the UUID for this object file. |
| 373 | /// |
| 374 | /// If the object file format contains a UUID, the value should be returned. |
| 375 | /// Else ObjectFile instances should return the MD5 checksum of all of the |
| 376 | /// bytes for the object file (or memory for memory based object files). |
| 377 | /// |
| 378 | /// \return |
| 379 | /// The object file's UUID. In case of an error, an empty UUID is |
| 380 | /// returned. |
| 381 | virtual UUID GetUUID() = 0; |
| 382 | |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 383 | /// Gets the file spec list of libraries re-exported by this object file. |
| 384 | /// |
| 385 | /// If the object file format has the notion of one library re-exporting the |
| 386 | /// symbols from another, the re-exported libraries will be returned in the |
| 387 | /// FileSpecList. |
| 388 | /// |
| 389 | /// \return |
| 390 | /// Returns filespeclist. |
| 391 | virtual lldb_private::FileSpecList GetReExportedLibraries() { |
| 392 | return FileSpecList(); |
| 393 | } |
| 394 | |
| 395 | /// Sets the load address for an entire module, assuming a rigid slide of |
| 396 | /// sections, if possible in the implementation. |
| 397 | /// |
| 398 | /// \return |
| 399 | /// Returns true iff any section's load address changed. |
| 400 | virtual bool SetLoadAddress(Target &target, lldb::addr_t value, |
| 401 | bool value_is_offset) { |
| 402 | return false; |
| 403 | } |
| 404 | |
| 405 | /// Gets whether endian swapping should occur when extracting data from this |
| 406 | /// object file. |
| 407 | /// |
| 408 | /// \return |
| 409 | /// Returns \b true if endian swapping is needed, \b false |
| 410 | /// otherwise. |
| 411 | virtual lldb::ByteOrder GetByteOrder() const = 0; |
| 412 | |
| 413 | /// Attempts to parse the object header. |
| 414 | /// |
| 415 | /// This function is used as a test to see if a given plug-in instance can |
| 416 | /// parse the header data already contained in ObjectFile::m_data. If an |
| 417 | /// object file parser does not recognize that magic bytes in a header, |
| 418 | /// false should be returned and the next plug-in can attempt to parse an |
| 419 | /// object file. |
| 420 | /// |
| 421 | /// \return |
| 422 | /// Returns \b true if the header was parsed successfully, \b |
| 423 | /// false otherwise. |
| 424 | virtual bool ParseHeader() = 0; |
| 425 | |
| 426 | /// Returns if the function bounds for symbols in this symbol file are |
| 427 | /// likely accurate. |
| 428 | /// |
| 429 | /// The unwinder can emulate the instructions of functions to understand |
| 430 | /// prologue/epilogue code sequences, where registers are spilled on the |
| 431 | /// stack, etc. This feature relies on having the correct start addresses |
| 432 | /// of all functions. If the ObjectFile has a way to tell that symbols have |
| 433 | /// been stripped and there's no way to reconstruct start addresses (e.g. |
| 434 | /// LC_FUNCTION_STARTS on Mach-O, or eh_frame unwind info), the ObjectFile |
| 435 | /// should indicate that assembly emulation should not be used for this |
| 436 | /// module. |
| 437 | /// |
| 438 | /// It is uncommon for this to return false. An ObjectFile needs to be sure |
| 439 | /// that symbol start addresses are unavailable before false is returned. |
| 440 | /// If it is unclear, this should return true. |
| 441 | /// |
| 442 | /// \return |
| 443 | /// Returns true if assembly emulation should be used for this |
| 444 | /// module. |
| 445 | /// Only returns false if the ObjectFile is sure that symbol |
| 446 | /// addresses are insufficient for accurate assembly emulation. |
| 447 | virtual bool AllowAssemblyEmulationUnwindPlans() { return true; } |
| 448 | |
| 449 | /// Similar to Process::GetImageInfoAddress(). |
| 450 | /// |
| 451 | /// Some platforms embed auxiliary structures useful to debuggers in the |
| 452 | /// address space of the inferior process. This method returns the address |
| 453 | /// of such a structure if the information can be resolved via entries in |
| 454 | /// the object file. ELF, for example, provides a means to hook into the |
| 455 | /// runtime linker so that a debugger may monitor the loading and unloading |
| 456 | /// of shared libraries. |
| 457 | /// |
| 458 | /// \return |
| 459 | /// The address of any auxiliary tables, or an invalid address if this |
| 460 | /// object file format does not support or contain such information. |
| 461 | virtual lldb_private::Address GetImageInfoAddress(Target *target) { |
| 462 | return Address(); |
| 463 | } |
| 464 | |
| 465 | /// Returns the address of the Entry Point in this object file - if the |
| 466 | /// object file doesn't have an entry point (because it is not an executable |
| 467 | /// file) then an invalid address is returned. |
| 468 | /// |
| 469 | /// \return |
| 470 | /// Returns the entry address for this module. |
| 471 | virtual lldb_private::Address GetEntryPointAddress() { return Address(); } |
| 472 | |
| 473 | /// Returns base address of this object file. |
| 474 | /// |
| 475 | /// This also sometimes referred to as the "preferred load address" or the |
| 476 | /// "image base address". Addresses within object files are often expressed |
| 477 | /// relative to this base. If this address corresponds to a specific section |
| 478 | /// (usually the first byte of the first section) then the returned address |
| 479 | /// will have this section set. Otherwise, the address will just have the |
| 480 | /// offset member filled in, indicating that this represents a file address. |
| 481 | virtual lldb_private::Address GetBaseAddress() { |
| 482 | return Address(m_memory_addr); |
| 483 | } |
| 484 | |
| 485 | virtual uint32_t GetNumThreadContexts() { return 0; } |
| 486 | |
| 487 | /// Some object files may have an identifier string embedded in them, e.g. |
| 488 | /// in a Mach-O core file using the LC_IDENT load command (which is |
| 489 | /// obsolete, but can still be found in some old files) |
| 490 | /// |
| 491 | /// \return |
| 492 | /// Returns the identifier string if one exists, else an empty |
| 493 | /// string. |
Olivier Deprez | f4ef2d0 | 2021-04-20 13:36:24 +0200 | [diff] [blame] | 494 | virtual std::string GetIdentifierString () { |
| 495 | return std::string(); |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 496 | } |
| 497 | |
| 498 | /// When the ObjectFile is a core file, lldb needs to locate the "binary" in |
| 499 | /// the core file. lldb can iterate over the pages looking for a valid |
| 500 | /// binary, but some core files may have metadata describing where the main |
| 501 | /// binary is exactly which removes ambiguity when there are multiple |
| 502 | /// binaries present in the captured memory pages. |
| 503 | /// |
| 504 | /// \param[out] address |
| 505 | /// If the address of the binary is specified, this will be set. |
| 506 | /// This is an address is the virtual address space of the core file |
| 507 | /// memory segments; it is not an offset into the object file. |
| 508 | /// If no address is available, will be set to LLDB_INVALID_ADDRESS. |
| 509 | /// |
| 510 | /// \param[out] uuid |
| 511 | /// If the uuid of the binary is specified, this will be set. |
| 512 | /// If no UUID is available, will be cleared. |
| 513 | /// |
Olivier Deprez | f4ef2d0 | 2021-04-20 13:36:24 +0200 | [diff] [blame] | 514 | /// \param[out] type |
| 515 | /// Return the type of the binary, which will dictate which |
| 516 | /// DynamicLoader plugin should be used. |
| 517 | /// |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 518 | /// \return |
| 519 | /// Returns true if either address or uuid has been set. |
Olivier Deprez | f4ef2d0 | 2021-04-20 13:36:24 +0200 | [diff] [blame] | 520 | virtual bool GetCorefileMainBinaryInfo(lldb::addr_t &address, UUID &uuid, |
| 521 | ObjectFile::BinaryType &type) { |
| 522 | address = LLDB_INVALID_ADDRESS; |
| 523 | uuid.Clear(); |
| 524 | return false; |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 525 | } |
| 526 | |
| 527 | virtual lldb::RegisterContextSP |
| 528 | GetThreadContextAtIndex(uint32_t idx, lldb_private::Thread &thread) { |
| 529 | return lldb::RegisterContextSP(); |
| 530 | } |
| 531 | |
| 532 | /// The object file should be able to calculate its type by looking at its |
| 533 | /// file header and possibly the sections or other data in the object file. |
| 534 | /// The file type is used in the debugger to help select the correct plug- |
| 535 | /// ins for the job at hand, so this is important to get right. If any |
| 536 | /// eTypeXXX definitions do not match up with the type of file you are |
| 537 | /// loading, please feel free to add a new enumeration value. |
| 538 | /// |
| 539 | /// \return |
| 540 | /// The calculated file type for the current object file. |
| 541 | virtual Type CalculateType() = 0; |
| 542 | |
| 543 | /// In cases where the type can't be calculated (elf files), this routine |
| 544 | /// allows someone to explicitly set it. As an example, SymbolVendorELF uses |
| 545 | /// this routine to set eTypeDebugInfo when loading debug link files. |
| 546 | virtual void SetType(Type type) { m_type = type; } |
| 547 | |
| 548 | /// The object file should be able to calculate the strata of the object |
| 549 | /// file. |
| 550 | /// |
| 551 | /// Many object files for platforms might be for either user space debugging |
| 552 | /// or for kernel debugging. If your object file subclass can figure this |
| 553 | /// out, it will help with debugger plug-in selection when it comes time to |
| 554 | /// debug. |
| 555 | /// |
| 556 | /// \return |
| 557 | /// The calculated object file strata for the current object |
| 558 | /// file. |
| 559 | virtual Strata CalculateStrata() = 0; |
| 560 | |
| 561 | /// Get the object file version numbers. |
| 562 | /// |
| 563 | /// Many object files have a set of version numbers that describe the |
| 564 | /// version of the executable or shared library. Typically there are major, |
| 565 | /// minor and build, but there may be more. This function will extract the |
| 566 | /// versions from object files if they are available. |
| 567 | /// |
| 568 | /// \return |
| 569 | /// This function returns extracted version numbers as a |
| 570 | /// llvm::VersionTuple. In case of error an empty VersionTuple is |
| 571 | /// returned. |
| 572 | virtual llvm::VersionTuple GetVersion() { return llvm::VersionTuple(); } |
| 573 | |
| 574 | /// Get the minimum OS version this object file can run on. |
| 575 | /// |
| 576 | /// Some object files have information that specifies the minimum OS version |
| 577 | /// that they can be used on. |
| 578 | /// |
| 579 | /// \return |
| 580 | /// This function returns extracted version numbers as a |
| 581 | /// llvm::VersionTuple. In case of error an empty VersionTuple is |
| 582 | /// returned. |
| 583 | virtual llvm::VersionTuple GetMinimumOSVersion() { |
| 584 | return llvm::VersionTuple(); |
| 585 | } |
| 586 | |
| 587 | /// Get the SDK OS version this object file was built with. |
| 588 | /// |
| 589 | /// \return |
| 590 | /// This function returns extracted version numbers as a |
| 591 | /// llvm::VersionTuple. In case of error an empty VersionTuple is |
| 592 | /// returned. |
| 593 | virtual llvm::VersionTuple GetSDKVersion() { return llvm::VersionTuple(); } |
| 594 | |
| 595 | /// Return true if this file is a dynamic link editor (dyld) |
| 596 | /// |
| 597 | /// Often times dyld has symbols that mirror symbols in libc and other |
| 598 | /// shared libraries (like "malloc" and "free") and the user does _not_ want |
| 599 | /// to stop in these shared libraries by default. We can ask the ObjectFile |
| 600 | /// if it is such a file and should be avoided for things like settings |
| 601 | /// breakpoints and doing function lookups for expressions. |
| 602 | virtual bool GetIsDynamicLinkEditor() { return false; } |
| 603 | |
| 604 | // Member Functions |
| 605 | Type GetType() { |
| 606 | if (m_type == eTypeInvalid) |
| 607 | m_type = CalculateType(); |
| 608 | return m_type; |
| 609 | } |
| 610 | |
| 611 | Strata GetStrata() { |
| 612 | if (m_strata == eStrataInvalid) |
| 613 | m_strata = CalculateStrata(); |
| 614 | return m_strata; |
| 615 | } |
| 616 | |
| 617 | // When an object file is in memory, subclasses should try and lock the |
| 618 | // process weak pointer. If the process weak pointer produces a valid |
| 619 | // ProcessSP, then subclasses can call this function to read memory. |
| 620 | static lldb::DataBufferSP ReadMemory(const lldb::ProcessSP &process_sp, |
| 621 | lldb::addr_t addr, size_t byte_size); |
| 622 | |
| 623 | // This function returns raw file contents. Do not use it if you want |
| 624 | // transparent decompression of section contents. |
| 625 | size_t GetData(lldb::offset_t offset, size_t length, |
| 626 | DataExtractor &data) const; |
| 627 | |
| 628 | // This function returns raw file contents. Do not use it if you want |
| 629 | // transparent decompression of section contents. |
| 630 | size_t CopyData(lldb::offset_t offset, size_t length, void *dst) const; |
| 631 | |
| 632 | // This function will transparently decompress section data if the section if |
| 633 | // compressed. |
| 634 | virtual size_t ReadSectionData(Section *section, |
| 635 | lldb::offset_t section_offset, void *dst, |
| 636 | size_t dst_len); |
| 637 | |
| 638 | // This function will transparently decompress section data if the section if |
| 639 | // compressed. Note that for compressed section the resulting data size may |
| 640 | // be larger than what Section::GetFileSize reports. |
| 641 | virtual size_t ReadSectionData(Section *section, |
| 642 | DataExtractor §ion_data); |
| 643 | |
| 644 | bool IsInMemory() const { return m_memory_addr != LLDB_INVALID_ADDRESS; } |
| 645 | |
| 646 | // Strip linker annotations (such as @@VERSION) from symbol names. |
| 647 | virtual llvm::StringRef |
| 648 | StripLinkerSymbolAnnotations(llvm::StringRef symbol_name) const { |
| 649 | return symbol_name; |
| 650 | } |
| 651 | |
| 652 | static lldb::SymbolType GetSymbolTypeFromName( |
| 653 | llvm::StringRef name, |
| 654 | lldb::SymbolType symbol_type_hint = lldb::eSymbolTypeUndefined); |
| 655 | |
| 656 | /// Loads this objfile to memory. |
| 657 | /// |
| 658 | /// Loads the bits needed to create an executable image to the memory. It is |
| 659 | /// useful with bare-metal targets where target does not have the ability to |
| 660 | /// start a process itself. |
| 661 | /// |
| 662 | /// \param[in] target |
| 663 | /// Target where to load. |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 664 | virtual std::vector<LoadableData> GetLoadableData(Target &target); |
| 665 | |
Olivier Deprez | f4ef2d0 | 2021-04-20 13:36:24 +0200 | [diff] [blame] | 666 | /// Creates a plugin-specific call frame info |
| 667 | virtual std::unique_ptr<CallFrameInfo> CreateCallFrameInfo(); |
| 668 | |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 669 | protected: |
| 670 | // Member variables. |
| 671 | FileSpec m_file; |
| 672 | Type m_type; |
| 673 | Strata m_strata; |
| 674 | lldb::addr_t m_file_offset; ///< The offset in bytes into the file, or the |
| 675 | ///address in memory |
| 676 | lldb::addr_t m_length; ///< The length of this object file if it is known (can |
| 677 | ///be zero if length is unknown or can't be |
| 678 | ///determined). |
| 679 | DataExtractor |
| 680 | m_data; ///< The data for this object file so things can be parsed lazily. |
| 681 | lldb::ProcessWP m_process_wp; |
| 682 | const lldb::addr_t m_memory_addr; |
| 683 | std::unique_ptr<lldb_private::SectionList> m_sections_up; |
| 684 | std::unique_ptr<lldb_private::Symtab> m_symtab_up; |
| 685 | uint32_t m_synthetic_symbol_idx; |
| 686 | |
| 687 | /// Sets the architecture for a module. At present the architecture can |
| 688 | /// only be set if it is invalid. It is not allowed to switch from one |
| 689 | /// concrete architecture to another. |
| 690 | /// |
| 691 | /// \param[in] new_arch |
| 692 | /// The architecture this module will be set to. |
| 693 | /// |
| 694 | /// \return |
| 695 | /// Returns \b true if the architecture was changed, \b |
| 696 | /// false otherwise. |
| 697 | bool SetModulesArchitecture(const ArchSpec &new_arch); |
| 698 | |
| 699 | ConstString GetNextSyntheticSymbolName(); |
| 700 | |
| 701 | static lldb::DataBufferSP MapFileData(const FileSpec &file, uint64_t Size, |
| 702 | uint64_t Offset); |
| 703 | |
| 704 | private: |
Olivier Deprez | f4ef2d0 | 2021-04-20 13:36:24 +0200 | [diff] [blame] | 705 | ObjectFile(const ObjectFile &) = delete; |
| 706 | const ObjectFile &operator=(const ObjectFile &) = delete; |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 707 | }; |
| 708 | |
| 709 | } // namespace lldb_private |
| 710 | |
| 711 | namespace llvm { |
| 712 | template <> struct format_provider<lldb_private::ObjectFile::Type> { |
| 713 | static void format(const lldb_private::ObjectFile::Type &type, |
| 714 | raw_ostream &OS, StringRef Style); |
| 715 | }; |
| 716 | |
| 717 | template <> struct format_provider<lldb_private::ObjectFile::Strata> { |
| 718 | static void format(const lldb_private::ObjectFile::Strata &strata, |
| 719 | raw_ostream &OS, StringRef Style); |
| 720 | }; |
| 721 | } // namespace llvm |
| 722 | |
Olivier Deprez | f4ef2d0 | 2021-04-20 13:36:24 +0200 | [diff] [blame] | 723 | #endif // LLDB_SYMBOL_OBJECTFILE_H |