1 | //! Interface for reading object files. |
2 | //! |
3 | //! ## Unified read API |
4 | //! |
5 | //! The [`Object`] trait provides a unified read API for accessing common features of |
6 | //! object files, such as sections and symbols. There is an implementation of this |
7 | //! trait for [`File`], which allows reading any file format, as well as implementations |
8 | //! for each file format: |
9 | //! [`ElfFile`](elf::ElfFile), [`MachOFile`](macho::MachOFile), [`CoffFile`](coff::CoffFile), |
10 | //! [`PeFile`](pe::PeFile), [`WasmFile`](wasm::WasmFile), [`XcoffFile`](xcoff::XcoffFile). |
11 | //! |
12 | //! ## Low level read API |
13 | //! |
14 | //! The submodules for each file format define helpers that operate on the raw structs. |
15 | //! These can be used instead of the unified API, or in conjunction with it to access |
16 | //! details that are not available via the unified API. |
17 | //! |
18 | //! See the [submodules](#modules) for examples of the low level read API. |
19 | //! |
20 | //! ## Naming Convention |
21 | //! |
22 | //! Types that form part of the unified API for a file format are prefixed with the |
23 | //! name of the file format. |
24 | //! |
25 | //! ## Example for unified read API |
26 | //! ```no_run |
27 | //! use object::{Object, ObjectSection}; |
28 | //! use std::error::Error; |
29 | //! use std::fs; |
30 | //! |
31 | //! /// Reads a file and displays the name of each section. |
32 | //! fn main() -> Result<(), Box<dyn Error>> { |
33 | //! # #[cfg (all(feature = "read" , feature = "std" ))] { |
34 | //! let data = fs::read("path/to/binary" )?; |
35 | //! let file = object::File::parse(&*data)?; |
36 | //! for section in file.sections() { |
37 | //! println!("{}" , section.name()?); |
38 | //! } |
39 | //! # } |
40 | //! Ok(()) |
41 | //! } |
42 | //! ``` |
43 | |
44 | use alloc::borrow::Cow; |
45 | use alloc::vec::Vec; |
46 | use core::{fmt, result}; |
47 | |
48 | #[cfg (not(feature = "std" ))] |
49 | use alloc::collections::btree_map::BTreeMap as Map; |
50 | #[cfg (feature = "std" )] |
51 | use std::collections::hash_map::HashMap as Map; |
52 | |
53 | pub use crate::common::*; |
54 | |
55 | mod read_ref; |
56 | pub use read_ref::*; |
57 | |
58 | mod read_cache; |
59 | pub use read_cache::*; |
60 | |
61 | mod util; |
62 | pub use util::*; |
63 | |
64 | #[cfg (any(feature = "elf" , feature = "macho" ))] |
65 | mod gnu_compression; |
66 | |
67 | #[cfg (any( |
68 | feature = "coff" , |
69 | feature = "elf" , |
70 | feature = "macho" , |
71 | feature = "pe" , |
72 | feature = "wasm" , |
73 | feature = "xcoff" |
74 | ))] |
75 | mod any; |
76 | #[cfg (any( |
77 | feature = "coff" , |
78 | feature = "elf" , |
79 | feature = "macho" , |
80 | feature = "pe" , |
81 | feature = "wasm" , |
82 | feature = "xcoff" |
83 | ))] |
84 | pub use any::*; |
85 | |
86 | #[cfg (feature = "archive" )] |
87 | pub mod archive; |
88 | |
89 | #[cfg (feature = "coff" )] |
90 | pub mod coff; |
91 | |
92 | #[cfg (feature = "elf" )] |
93 | pub mod elf; |
94 | |
95 | #[cfg (feature = "macho" )] |
96 | pub mod macho; |
97 | |
98 | #[cfg (feature = "pe" )] |
99 | pub mod pe; |
100 | |
101 | #[cfg (feature = "wasm" )] |
102 | pub mod wasm; |
103 | |
104 | #[cfg (feature = "xcoff" )] |
105 | pub mod xcoff; |
106 | |
107 | mod traits; |
108 | pub use traits::*; |
109 | |
110 | mod private { |
111 | pub trait Sealed {} |
112 | } |
113 | |
114 | /// The error type used within the read module. |
115 | #[derive (Debug, Clone, Copy, PartialEq, Eq)] |
116 | pub struct Error(pub(crate) &'static str); |
117 | |
118 | impl fmt::Display for Error { |
119 | #[inline ] |
120 | fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { |
121 | f.write_str(self.0) |
122 | } |
123 | } |
124 | |
125 | #[cfg (feature = "std" )] |
126 | impl std::error::Error for Error {} |
127 | #[cfg (all(not(feature = "std" ), core_error))] |
128 | impl core::error::Error for Error {} |
129 | |
130 | /// The result type used within the read module. |
131 | pub type Result<T> = result::Result<T, Error>; |
132 | |
133 | trait ReadError<T> { |
134 | fn read_error(self, error: &'static str) -> Result<T>; |
135 | } |
136 | |
137 | impl<T> ReadError<T> for result::Result<T, ()> { |
138 | fn read_error(self, error: &'static str) -> Result<T> { |
139 | self.map_err(|()| Error(error)) |
140 | } |
141 | } |
142 | |
143 | impl<T> ReadError<T> for result::Result<T, Error> { |
144 | fn read_error(self, error: &'static str) -> Result<T> { |
145 | self.map_err(|_| Error(error)) |
146 | } |
147 | } |
148 | |
149 | impl<T> ReadError<T> for Option<T> { |
150 | fn read_error(self, error: &'static str) -> Result<T> { |
151 | self.ok_or(err:Error(error)) |
152 | } |
153 | } |
154 | |
155 | /// The native executable file for the target platform. |
156 | #[cfg (all( |
157 | unix, |
158 | not(target_os = "macos" ), |
159 | target_pointer_width = "32" , |
160 | feature = "elf" |
161 | ))] |
162 | pub type NativeFile<'data, R = &'data [u8]> = elf::ElfFile32<'data, crate::endian::Endianness, R>; |
163 | |
164 | /// The native executable file for the target platform. |
165 | #[cfg (all( |
166 | unix, |
167 | not(target_os = "macos" ), |
168 | target_pointer_width = "64" , |
169 | feature = "elf" |
170 | ))] |
171 | pub type NativeFile<'data, R = &'data [u8]> = elf::ElfFile64<'data, crate::endian::Endianness, R>; |
172 | |
173 | /// The native executable file for the target platform. |
174 | #[cfg (all(target_os = "macos" , target_pointer_width = "32" , feature = "macho" ))] |
175 | pub type NativeFile<'data, R = &'data [u8]> = |
176 | macho::MachOFile32<'data, crate::endian::Endianness, R>; |
177 | |
178 | /// The native executable file for the target platform. |
179 | #[cfg (all(target_os = "macos" , target_pointer_width = "64" , feature = "macho" ))] |
180 | pub type NativeFile<'data, R = &'data [u8]> = |
181 | macho::MachOFile64<'data, crate::endian::Endianness, R>; |
182 | |
183 | /// The native executable file for the target platform. |
184 | #[cfg (all(target_os = "windows" , target_pointer_width = "32" , feature = "pe" ))] |
185 | pub type NativeFile<'data, R = &'data [u8]> = pe::PeFile32<'data, R>; |
186 | |
187 | /// The native executable file for the target platform. |
188 | #[cfg (all(target_os = "windows" , target_pointer_width = "64" , feature = "pe" ))] |
189 | pub type NativeFile<'data, R = &'data [u8]> = pe::PeFile64<'data, R>; |
190 | |
191 | /// The native executable file for the target platform. |
192 | #[cfg (all(feature = "wasm" , target_arch = "wasm32" , feature = "wasm" ))] |
193 | pub type NativeFile<'data, R = &'data [u8]> = wasm::WasmFile<'data, R>; |
194 | |
195 | /// A file format kind. |
196 | #[derive (Debug, Clone, Copy, PartialEq, Eq, Hash)] |
197 | #[non_exhaustive ] |
198 | pub enum FileKind { |
199 | /// A Unix archive. |
200 | /// |
201 | /// See [`archive::ArchiveFile`]. |
202 | #[cfg (feature = "archive" )] |
203 | Archive, |
204 | /// A COFF object file. |
205 | /// |
206 | /// See [`coff::CoffFile`]. |
207 | #[cfg (feature = "coff" )] |
208 | Coff, |
209 | /// A COFF bigobj object file. |
210 | /// |
211 | /// This supports a larger number of sections. |
212 | /// |
213 | /// See [`coff::CoffBigFile`]. |
214 | #[cfg (feature = "coff" )] |
215 | CoffBig, |
216 | /// A Windows short import file. |
217 | /// |
218 | /// See [`coff::ImportFile`]. |
219 | #[cfg (feature = "coff" )] |
220 | CoffImport, |
221 | /// A dyld cache file containing Mach-O images. |
222 | /// |
223 | /// See [`macho::DyldCache`] |
224 | #[cfg (feature = "macho" )] |
225 | DyldCache, |
226 | /// A 32-bit ELF file. |
227 | /// |
228 | /// See [`elf::ElfFile32`]. |
229 | #[cfg (feature = "elf" )] |
230 | Elf32, |
231 | /// A 64-bit ELF file. |
232 | /// |
233 | /// See [`elf::ElfFile64`]. |
234 | #[cfg (feature = "elf" )] |
235 | Elf64, |
236 | /// A 32-bit Mach-O file. |
237 | /// |
238 | /// See [`macho::MachOFile32`]. |
239 | #[cfg (feature = "macho" )] |
240 | MachO32, |
241 | /// A 64-bit Mach-O file. |
242 | /// |
243 | /// See [`macho::MachOFile64`]. |
244 | #[cfg (feature = "macho" )] |
245 | MachO64, |
246 | /// A 32-bit Mach-O fat binary. |
247 | /// |
248 | /// See [`macho::MachOFatFile32`]. |
249 | #[cfg (feature = "macho" )] |
250 | MachOFat32, |
251 | /// A 64-bit Mach-O fat binary. |
252 | /// |
253 | /// See [`macho::MachOFatFile64`]. |
254 | #[cfg (feature = "macho" )] |
255 | MachOFat64, |
256 | /// A 32-bit PE file. |
257 | /// |
258 | /// See [`pe::PeFile32`]. |
259 | #[cfg (feature = "pe" )] |
260 | Pe32, |
261 | /// A 64-bit PE file. |
262 | /// |
263 | /// See [`pe::PeFile64`]. |
264 | #[cfg (feature = "pe" )] |
265 | Pe64, |
266 | /// A Wasm file. |
267 | /// |
268 | /// See [`wasm::WasmFile`]. |
269 | #[cfg (feature = "wasm" )] |
270 | Wasm, |
271 | /// A 32-bit XCOFF file. |
272 | /// |
273 | /// See [`xcoff::XcoffFile32`]. |
274 | #[cfg (feature = "xcoff" )] |
275 | Xcoff32, |
276 | /// A 64-bit XCOFF file. |
277 | /// |
278 | /// See [`xcoff::XcoffFile64`]. |
279 | #[cfg (feature = "xcoff" )] |
280 | Xcoff64, |
281 | } |
282 | |
283 | impl FileKind { |
284 | /// Determine a file kind by parsing the start of the file. |
285 | pub fn parse<'data, R: ReadRef<'data>>(data: R) -> Result<FileKind> { |
286 | Self::parse_at(data, 0) |
287 | } |
288 | |
289 | /// Determine a file kind by parsing at the given offset. |
290 | pub fn parse_at<'data, R: ReadRef<'data>>(data: R, offset: u64) -> Result<FileKind> { |
291 | let magic = data |
292 | .read_bytes_at(offset, 16) |
293 | .read_error("Could not read file magic" )?; |
294 | if magic.len() < 16 { |
295 | return Err(Error("File too short" )); |
296 | } |
297 | |
298 | let kind = match [magic[0], magic[1], magic[2], magic[3], magic[4], magic[5], magic[6], magic[7]] { |
299 | #[cfg (feature = "archive" )] |
300 | [b'!' , b'<' , b'a' , b'r' , b'c' , b'h' , b'>' , b' \n' ] |
301 | | [b'!' , b'<' , b't' , b'h' , b'i' , b'n' , b'>' , b' \n' ] => FileKind::Archive, |
302 | #[cfg (feature = "macho" )] |
303 | [b'd' , b'y' , b'l' , b'd' , b'_' , b'v' , b'1' , b' ' ] => FileKind::DyldCache, |
304 | #[cfg (feature = "elf" )] |
305 | [0x7f, b'E' , b'L' , b'F' , 1, ..] => FileKind::Elf32, |
306 | #[cfg (feature = "elf" )] |
307 | [0x7f, b'E' , b'L' , b'F' , 2, ..] => FileKind::Elf64, |
308 | #[cfg (feature = "macho" )] |
309 | [0xfe, 0xed, 0xfa, 0xce, ..] |
310 | | [0xce, 0xfa, 0xed, 0xfe, ..] => FileKind::MachO32, |
311 | #[cfg (feature = "macho" )] |
312 | | [0xfe, 0xed, 0xfa, 0xcf, ..] |
313 | | [0xcf, 0xfa, 0xed, 0xfe, ..] => FileKind::MachO64, |
314 | #[cfg (feature = "macho" )] |
315 | [0xca, 0xfe, 0xba, 0xbe, ..] => FileKind::MachOFat32, |
316 | #[cfg (feature = "macho" )] |
317 | [0xca, 0xfe, 0xba, 0xbf, ..] => FileKind::MachOFat64, |
318 | #[cfg (feature = "wasm" )] |
319 | [0x00, b'a' , b's' , b'm' , _, _, 0x00, 0x00] => FileKind::Wasm, |
320 | #[cfg (feature = "pe" )] |
321 | [b'M' , b'Z' , ..] if offset == 0 => { |
322 | // offset == 0 restriction is because optional_header_magic only looks at offset 0 |
323 | match pe::optional_header_magic(data) { |
324 | Ok(crate::pe::IMAGE_NT_OPTIONAL_HDR32_MAGIC) => { |
325 | FileKind::Pe32 |
326 | } |
327 | Ok(crate::pe::IMAGE_NT_OPTIONAL_HDR64_MAGIC) => { |
328 | FileKind::Pe64 |
329 | } |
330 | _ => return Err(Error("Unknown MS-DOS file" )), |
331 | } |
332 | } |
333 | // TODO: more COFF machines |
334 | #[cfg (feature = "coff" )] |
335 | // COFF arm |
336 | [0xc4, 0x01, ..] |
337 | // COFF arm64 |
338 | | [0x64, 0xaa, ..] |
339 | // COFF arm64ec |
340 | | [0x41, 0xa6, ..] |
341 | // COFF x86 |
342 | | [0x4c, 0x01, ..] |
343 | // COFF x86-64 |
344 | | [0x64, 0x86, ..] => FileKind::Coff, |
345 | #[cfg (feature = "coff" )] |
346 | [0x00, 0x00, 0xff, 0xff, 0x00, 0x00, ..] => FileKind::CoffImport, |
347 | #[cfg (feature = "coff" )] |
348 | [0x00, 0x00, 0xff, 0xff, 0x02, 0x00, ..] if offset == 0 => { |
349 | // offset == 0 restriction is because anon_object_class_id only looks at offset 0 |
350 | match coff::anon_object_class_id(data) { |
351 | Ok(crate::pe::ANON_OBJECT_HEADER_BIGOBJ_CLASS_ID) => FileKind::CoffBig, |
352 | _ => return Err(Error("Unknown anon object file" )), |
353 | } |
354 | } |
355 | #[cfg (feature = "xcoff" )] |
356 | [0x01, 0xdf, ..] => FileKind::Xcoff32, |
357 | #[cfg (feature = "xcoff" )] |
358 | [0x01, 0xf7, ..] => FileKind::Xcoff64, |
359 | _ => return Err(Error("Unknown file magic" )), |
360 | }; |
361 | Ok(kind) |
362 | } |
363 | } |
364 | |
365 | /// An object kind. |
366 | /// |
367 | /// Returned by [`Object::kind`]. |
368 | #[derive (Debug, Clone, Copy, PartialEq, Eq, Hash)] |
369 | #[non_exhaustive ] |
370 | pub enum ObjectKind { |
371 | /// The object kind is unknown. |
372 | Unknown, |
373 | /// Relocatable object. |
374 | Relocatable, |
375 | /// Executable. |
376 | Executable, |
377 | /// Dynamic shared object. |
378 | Dynamic, |
379 | /// Core. |
380 | Core, |
381 | } |
382 | |
383 | /// The index used to identify a section in a file. |
384 | #[derive (Debug, Clone, Copy, PartialEq, Eq, Hash)] |
385 | pub struct SectionIndex(pub usize); |
386 | |
387 | impl fmt::Display for SectionIndex { |
388 | fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { |
389 | self.0.fmt(f) |
390 | } |
391 | } |
392 | |
393 | /// The index used to identify a symbol in a symbol table. |
394 | #[derive (Debug, Clone, Copy, PartialEq, Eq, Hash)] |
395 | pub struct SymbolIndex(pub usize); |
396 | |
397 | impl fmt::Display for SymbolIndex { |
398 | fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { |
399 | self.0.fmt(f) |
400 | } |
401 | } |
402 | |
403 | /// The section where an [`ObjectSymbol`] is defined. |
404 | #[derive (Debug, Clone, Copy, PartialEq, Eq, Hash)] |
405 | #[non_exhaustive ] |
406 | pub enum SymbolSection { |
407 | /// The section is unknown. |
408 | Unknown, |
409 | /// The section is not applicable for this symbol (such as file symbols). |
410 | None, |
411 | /// The symbol is undefined. |
412 | Undefined, |
413 | /// The symbol has an absolute value. |
414 | Absolute, |
415 | /// The symbol is a zero-initialized symbol that will be combined with duplicate definitions. |
416 | Common, |
417 | /// The symbol is defined in the given section. |
418 | Section(SectionIndex), |
419 | } |
420 | |
421 | impl SymbolSection { |
422 | /// Returns the section index for the section where the symbol is defined. |
423 | /// |
424 | /// May return `None` if the symbol is not defined in a section. |
425 | #[inline ] |
426 | pub fn index(self) -> Option<SectionIndex> { |
427 | if let SymbolSection::Section(index: SectionIndex) = self { |
428 | Some(index) |
429 | } else { |
430 | None |
431 | } |
432 | } |
433 | } |
434 | |
435 | /// An entry in a [`SymbolMap`]. |
436 | pub trait SymbolMapEntry { |
437 | /// The symbol address. |
438 | fn address(&self) -> u64; |
439 | } |
440 | |
441 | /// A map from addresses to symbol information. |
442 | /// |
443 | /// The symbol information depends on the chosen entry type, such as [`SymbolMapName`]. |
444 | /// |
445 | /// Returned by [`Object::symbol_map`]. |
446 | #[derive (Debug, Default, Clone)] |
447 | pub struct SymbolMap<T: SymbolMapEntry> { |
448 | symbols: Vec<T>, |
449 | } |
450 | |
451 | impl<T: SymbolMapEntry> SymbolMap<T> { |
452 | /// Construct a new symbol map. |
453 | /// |
454 | /// This function will sort the symbols by address. |
455 | pub fn new(mut symbols: Vec<T>) -> Self { |
456 | symbols.sort_by_key(|s| s.address()); |
457 | SymbolMap { symbols } |
458 | } |
459 | |
460 | /// Get the symbol before the given address. |
461 | pub fn get(&self, address: u64) -> Option<&T> { |
462 | let index = match self |
463 | .symbols |
464 | .binary_search_by_key(&address, |symbol| symbol.address()) |
465 | { |
466 | Ok(index) => index, |
467 | Err(index) => index.checked_sub(1)?, |
468 | }; |
469 | self.symbols.get(index) |
470 | } |
471 | |
472 | /// Get all symbols in the map. |
473 | #[inline ] |
474 | pub fn symbols(&self) -> &[T] { |
475 | &self.symbols |
476 | } |
477 | } |
478 | |
479 | /// The type used for entries in a [`SymbolMap`] that maps from addresses to names. |
480 | #[derive (Debug, Clone, Copy, PartialEq, Eq, Hash)] |
481 | pub struct SymbolMapName<'data> { |
482 | address: u64, |
483 | name: &'data str, |
484 | } |
485 | |
486 | impl<'data> SymbolMapName<'data> { |
487 | /// Construct a `SymbolMapName`. |
488 | pub fn new(address: u64, name: &'data str) -> Self { |
489 | SymbolMapName { address, name } |
490 | } |
491 | |
492 | /// The symbol address. |
493 | #[inline ] |
494 | pub fn address(&self) -> u64 { |
495 | self.address |
496 | } |
497 | |
498 | /// The symbol name. |
499 | #[inline ] |
500 | pub fn name(&self) -> &'data str { |
501 | self.name |
502 | } |
503 | } |
504 | |
505 | impl<'data> SymbolMapEntry for SymbolMapName<'data> { |
506 | #[inline ] |
507 | fn address(&self) -> u64 { |
508 | self.address |
509 | } |
510 | } |
511 | |
512 | /// A map from addresses to symbol names and object files. |
513 | /// |
514 | /// This is derived from STAB entries in Mach-O files. |
515 | /// |
516 | /// Returned by [`Object::object_map`]. |
517 | #[derive (Debug, Default, Clone)] |
518 | pub struct ObjectMap<'data> { |
519 | symbols: SymbolMap<ObjectMapEntry<'data>>, |
520 | objects: Vec<ObjectMapFile<'data>>, |
521 | } |
522 | |
523 | impl<'data> ObjectMap<'data> { |
524 | /// Get the entry containing the given address. |
525 | pub fn get(&self, address: u64) -> Option<&ObjectMapEntry<'data>> { |
526 | self.symbols |
527 | .get(address) |
528 | .filter(|entry: &&ObjectMapEntry<'_>| entry.size == 0 || address.wrapping_sub(entry.address) < entry.size) |
529 | } |
530 | |
531 | /// Get all symbols in the map. |
532 | #[inline ] |
533 | pub fn symbols(&self) -> &[ObjectMapEntry<'data>] { |
534 | self.symbols.symbols() |
535 | } |
536 | |
537 | /// Get all objects in the map. |
538 | #[inline ] |
539 | pub fn objects(&self) -> &[ObjectMapFile<'data>] { |
540 | &self.objects |
541 | } |
542 | } |
543 | |
544 | /// A symbol in an [`ObjectMap`]. |
545 | #[derive (Debug, Default, Clone, Copy, PartialEq, Eq, Hash)] |
546 | pub struct ObjectMapEntry<'data> { |
547 | address: u64, |
548 | size: u64, |
549 | name: &'data [u8], |
550 | object: usize, |
551 | } |
552 | |
553 | impl<'data> ObjectMapEntry<'data> { |
554 | /// Get the symbol address. |
555 | #[inline ] |
556 | pub fn address(&self) -> u64 { |
557 | self.address |
558 | } |
559 | |
560 | /// Get the symbol size. |
561 | /// |
562 | /// This may be 0 if the size is unknown. |
563 | #[inline ] |
564 | pub fn size(&self) -> u64 { |
565 | self.size |
566 | } |
567 | |
568 | /// Get the symbol name. |
569 | #[inline ] |
570 | pub fn name(&self) -> &'data [u8] { |
571 | self.name |
572 | } |
573 | |
574 | /// Get the index of the object file name. |
575 | #[inline ] |
576 | pub fn object_index(&self) -> usize { |
577 | self.object |
578 | } |
579 | |
580 | /// Get the object file name. |
581 | #[inline ] |
582 | pub fn object<'a>(&self, map: &'a ObjectMap<'data>) -> &'a ObjectMapFile<'data> { |
583 | &map.objects[self.object] |
584 | } |
585 | } |
586 | |
587 | impl<'data> SymbolMapEntry for ObjectMapEntry<'data> { |
588 | #[inline ] |
589 | fn address(&self) -> u64 { |
590 | self.address |
591 | } |
592 | } |
593 | |
594 | /// An object file name in an [`ObjectMap`]. |
595 | #[derive (Debug, Clone, Copy, PartialEq, Eq, Hash)] |
596 | pub struct ObjectMapFile<'data> { |
597 | path: &'data [u8], |
598 | member: Option<&'data [u8]>, |
599 | } |
600 | |
601 | impl<'data> ObjectMapFile<'data> { |
602 | #[cfg (feature = "macho" )] |
603 | fn new(path: &'data [u8], member: Option<&'data [u8]>) -> Self { |
604 | ObjectMapFile { path, member } |
605 | } |
606 | |
607 | /// Get the path to the file containing the object. |
608 | #[inline ] |
609 | pub fn path(&self) -> &'data [u8] { |
610 | self.path |
611 | } |
612 | |
613 | /// If the file is an archive, get the name of the member containing the object. |
614 | #[inline ] |
615 | pub fn member(&self) -> Option<&'data [u8]> { |
616 | self.member |
617 | } |
618 | } |
619 | |
620 | /// An imported symbol. |
621 | /// |
622 | /// Returned by [`Object::imports`]. |
623 | #[derive (Debug, Clone, Copy, PartialEq, Eq)] |
624 | pub struct Import<'data> { |
625 | library: ByteString<'data>, |
626 | // TODO: or ordinal |
627 | name: ByteString<'data>, |
628 | } |
629 | |
630 | impl<'data> Import<'data> { |
631 | /// The symbol name. |
632 | #[inline ] |
633 | pub fn name(&self) -> &'data [u8] { |
634 | self.name.0 |
635 | } |
636 | |
637 | /// The name of the library to import the symbol from. |
638 | #[inline ] |
639 | pub fn library(&self) -> &'data [u8] { |
640 | self.library.0 |
641 | } |
642 | } |
643 | |
644 | /// An exported symbol. |
645 | /// |
646 | /// Returned by [`Object::exports`]. |
647 | #[derive (Debug, Clone, Copy, PartialEq, Eq)] |
648 | pub struct Export<'data> { |
649 | // TODO: and ordinal? |
650 | name: ByteString<'data>, |
651 | address: u64, |
652 | } |
653 | |
654 | impl<'data> Export<'data> { |
655 | /// The symbol name. |
656 | #[inline ] |
657 | pub fn name(&self) -> &'data [u8] { |
658 | self.name.0 |
659 | } |
660 | |
661 | /// The virtual address of the symbol. |
662 | #[inline ] |
663 | pub fn address(&self) -> u64 { |
664 | self.address |
665 | } |
666 | } |
667 | |
668 | /// PDB information from the debug directory in a PE file. |
669 | #[derive (Debug, Clone, Copy, PartialEq, Eq)] |
670 | pub struct CodeView<'data> { |
671 | guid: [u8; 16], |
672 | path: ByteString<'data>, |
673 | age: u32, |
674 | } |
675 | |
676 | impl<'data> CodeView<'data> { |
677 | /// The path to the PDB as stored in CodeView. |
678 | #[inline ] |
679 | pub fn path(&self) -> &'data [u8] { |
680 | self.path.0 |
681 | } |
682 | |
683 | /// The age of the PDB. |
684 | #[inline ] |
685 | pub fn age(&self) -> u32 { |
686 | self.age |
687 | } |
688 | |
689 | /// The GUID of the PDB. |
690 | #[inline ] |
691 | pub fn guid(&self) -> [u8; 16] { |
692 | self.guid |
693 | } |
694 | } |
695 | |
696 | /// The target referenced by a [`Relocation`]. |
697 | #[derive (Debug, Clone, Copy, PartialEq, Eq, Hash)] |
698 | #[non_exhaustive ] |
699 | pub enum RelocationTarget { |
700 | /// The target is a symbol. |
701 | Symbol(SymbolIndex), |
702 | /// The target is a section. |
703 | Section(SectionIndex), |
704 | /// The offset is an absolute address. |
705 | Absolute, |
706 | } |
707 | |
708 | /// A relocation entry. |
709 | /// |
710 | /// Returned by [`Object::dynamic_relocations`] or [`ObjectSection::relocations`]. |
711 | #[derive (Debug)] |
712 | pub struct Relocation { |
713 | kind: RelocationKind, |
714 | encoding: RelocationEncoding, |
715 | size: u8, |
716 | target: RelocationTarget, |
717 | addend: i64, |
718 | implicit_addend: bool, |
719 | flags: RelocationFlags, |
720 | } |
721 | |
722 | impl Relocation { |
723 | /// The operation used to calculate the result of the relocation. |
724 | #[inline ] |
725 | pub fn kind(&self) -> RelocationKind { |
726 | self.kind |
727 | } |
728 | |
729 | /// Information about how the result of the relocation operation is encoded in the place. |
730 | #[inline ] |
731 | pub fn encoding(&self) -> RelocationEncoding { |
732 | self.encoding |
733 | } |
734 | |
735 | /// The size in bits of the place of the relocation. |
736 | /// |
737 | /// If 0, then the size is determined by the relocation kind. |
738 | #[inline ] |
739 | pub fn size(&self) -> u8 { |
740 | self.size |
741 | } |
742 | |
743 | /// The target of the relocation. |
744 | #[inline ] |
745 | pub fn target(&self) -> RelocationTarget { |
746 | self.target |
747 | } |
748 | |
749 | /// The addend to use in the relocation calculation. |
750 | #[inline ] |
751 | pub fn addend(&self) -> i64 { |
752 | self.addend |
753 | } |
754 | |
755 | /// Set the addend to use in the relocation calculation. |
756 | #[inline ] |
757 | pub fn set_addend(&mut self, addend: i64) { |
758 | self.addend = addend; |
759 | } |
760 | |
761 | /// Returns true if there is an implicit addend stored in the data at the offset |
762 | /// to be relocated. |
763 | #[inline ] |
764 | pub fn has_implicit_addend(&self) -> bool { |
765 | self.implicit_addend |
766 | } |
767 | |
768 | /// Relocation flags that are specific to each file format. |
769 | /// |
770 | /// The values returned by `kind`, `encoding` and `size` are derived |
771 | /// from these flags. |
772 | #[inline ] |
773 | pub fn flags(&self) -> RelocationFlags { |
774 | self.flags |
775 | } |
776 | } |
777 | |
778 | /// A map from section offsets to relocation information. |
779 | /// |
780 | /// This can be used to apply relocations to a value at a given section offset. |
781 | /// This is intended for use with DWARF in relocatable object files, and only |
782 | /// supports relocations that are used in DWARF. |
783 | /// |
784 | /// Returned by [`ObjectSection::relocation_map`]. |
785 | #[derive (Debug, Default)] |
786 | pub struct RelocationMap(Map<u64, RelocationMapEntry>); |
787 | |
788 | impl RelocationMap { |
789 | /// Construct a new relocation map for a section. |
790 | /// |
791 | /// Fails if any relocation cannot be added to the map. |
792 | /// You can manually use `add` if you need different error handling, |
793 | /// such as to list all errors or to ignore them. |
794 | pub fn new<'data, 'file, T>(file: &'file T, section: &T::Section<'file>) -> Result<Self> |
795 | where |
796 | T: Object<'data>, |
797 | { |
798 | let mut map = RelocationMap(Map::new()); |
799 | for (offset, relocation) in section.relocations() { |
800 | map.add(file, offset, relocation)?; |
801 | } |
802 | Ok(map) |
803 | } |
804 | |
805 | /// Add a single relocation to the map. |
806 | pub fn add<'data: 'file, 'file, T>( |
807 | &mut self, |
808 | file: &'file T, |
809 | offset: u64, |
810 | relocation: Relocation, |
811 | ) -> Result<()> |
812 | where |
813 | T: Object<'data>, |
814 | { |
815 | let mut entry = RelocationMapEntry { |
816 | implicit_addend: relocation.has_implicit_addend(), |
817 | addend: relocation.addend() as u64, |
818 | }; |
819 | match relocation.kind() { |
820 | RelocationKind::Absolute => match relocation.target() { |
821 | RelocationTarget::Symbol(symbol_idx) => { |
822 | let symbol = file |
823 | .symbol_by_index(symbol_idx) |
824 | .read_error("Relocation with invalid symbol" )?; |
825 | entry.addend = symbol.address().wrapping_add(entry.addend); |
826 | } |
827 | RelocationTarget::Section(section_idx) => { |
828 | let section = file |
829 | .section_by_index(section_idx) |
830 | .read_error("Relocation with invalid section" )?; |
831 | // DWARF parsers expect references to DWARF sections to be section offsets, |
832 | // not addresses. Addresses are useful for everything else. |
833 | if section.kind() != SectionKind::Debug { |
834 | entry.addend = section.address().wrapping_add(entry.addend); |
835 | } |
836 | } |
837 | _ => { |
838 | return Err(Error("Unsupported relocation target" )); |
839 | } |
840 | }, |
841 | _ => { |
842 | return Err(Error("Unsupported relocation type" )); |
843 | } |
844 | } |
845 | if self.0.insert(offset, entry).is_some() { |
846 | return Err(Error("Multiple relocations for offset" )); |
847 | } |
848 | Ok(()) |
849 | } |
850 | |
851 | /// Relocate a value that was read from the section at the given offset. |
852 | pub fn relocate(&self, offset: u64, value: u64) -> u64 { |
853 | if let Some(relocation) = self.0.get(&offset) { |
854 | if relocation.implicit_addend { |
855 | // Use the explicit addend too, because it may have the symbol value. |
856 | value.wrapping_add(relocation.addend) |
857 | } else { |
858 | relocation.addend |
859 | } |
860 | } else { |
861 | value |
862 | } |
863 | } |
864 | } |
865 | |
866 | #[derive (Debug, Clone, Copy, PartialEq, Eq, Hash)] |
867 | struct RelocationMapEntry { |
868 | implicit_addend: bool, |
869 | addend: u64, |
870 | } |
871 | |
872 | /// A data compression format. |
873 | #[derive (Debug, Clone, Copy, PartialEq, Eq, Hash)] |
874 | #[non_exhaustive ] |
875 | pub enum CompressionFormat { |
876 | /// The data is uncompressed. |
877 | None, |
878 | /// The data is compressed, but the compression format is unknown. |
879 | Unknown, |
880 | /// ZLIB/DEFLATE. |
881 | /// |
882 | /// Used for ELF compression and GNU compressed debug information. |
883 | Zlib, |
884 | /// Zstandard. |
885 | /// |
886 | /// Used for ELF compression. |
887 | Zstandard, |
888 | } |
889 | |
890 | /// A range in a file that may be compressed. |
891 | /// |
892 | /// Returned by [`ObjectSection::compressed_file_range`]. |
893 | #[derive (Debug, Clone, Copy, PartialEq, Eq, Hash)] |
894 | pub struct CompressedFileRange { |
895 | /// The data compression format. |
896 | pub format: CompressionFormat, |
897 | /// The file offset of the compressed data. |
898 | pub offset: u64, |
899 | /// The compressed data size. |
900 | pub compressed_size: u64, |
901 | /// The uncompressed data size. |
902 | pub uncompressed_size: u64, |
903 | } |
904 | |
905 | impl CompressedFileRange { |
906 | /// Data that is uncompressed. |
907 | #[inline ] |
908 | pub fn none(range: Option<(u64, u64)>) -> Self { |
909 | if let Some((offset, size)) = range { |
910 | CompressedFileRange { |
911 | format: CompressionFormat::None, |
912 | offset, |
913 | compressed_size: size, |
914 | uncompressed_size: size, |
915 | } |
916 | } else { |
917 | CompressedFileRange { |
918 | format: CompressionFormat::None, |
919 | offset: 0, |
920 | compressed_size: 0, |
921 | uncompressed_size: 0, |
922 | } |
923 | } |
924 | } |
925 | |
926 | /// Convert to [`CompressedData`] by reading from the file. |
927 | pub fn data<'data, R: ReadRef<'data>>(self, file: R) -> Result<CompressedData<'data>> { |
928 | let data = file |
929 | .read_bytes_at(self.offset, self.compressed_size) |
930 | .read_error("Invalid compressed data size or offset" )?; |
931 | Ok(CompressedData { |
932 | format: self.format, |
933 | data, |
934 | uncompressed_size: self.uncompressed_size, |
935 | }) |
936 | } |
937 | } |
938 | |
939 | /// Data that may be compressed. |
940 | /// |
941 | /// Returned by [`ObjectSection::compressed_data`]. |
942 | #[derive (Debug, Clone, Copy, PartialEq, Eq, Hash)] |
943 | pub struct CompressedData<'data> { |
944 | /// The data compression format. |
945 | pub format: CompressionFormat, |
946 | /// The compressed data. |
947 | pub data: &'data [u8], |
948 | /// The uncompressed data size. |
949 | pub uncompressed_size: u64, |
950 | } |
951 | |
952 | impl<'data> CompressedData<'data> { |
953 | /// Data that is uncompressed. |
954 | #[inline ] |
955 | pub fn none(data: &'data [u8]) -> Self { |
956 | CompressedData { |
957 | format: CompressionFormat::None, |
958 | data, |
959 | uncompressed_size: data.len() as u64, |
960 | } |
961 | } |
962 | |
963 | /// Return the uncompressed data. |
964 | /// |
965 | /// Returns an error for invalid data or unsupported compression. |
966 | /// This includes if the data is compressed but the `compression` feature |
967 | /// for this crate is disabled. |
968 | pub fn decompress(self) -> Result<Cow<'data, [u8]>> { |
969 | match self.format { |
970 | CompressionFormat::None => Ok(Cow::Borrowed(self.data)), |
971 | #[cfg (feature = "compression" )] |
972 | CompressionFormat::Zlib | CompressionFormat::Zstandard => { |
973 | use core::convert::TryInto; |
974 | use std::io::Read; |
975 | let size = self |
976 | .uncompressed_size |
977 | .try_into() |
978 | .ok() |
979 | .read_error("Uncompressed data size is too large." )?; |
980 | let mut decompressed = Vec::new(); |
981 | decompressed |
982 | .try_reserve_exact(size) |
983 | .ok() |
984 | .read_error("Uncompressed data allocation failed" )?; |
985 | |
986 | match self.format { |
987 | CompressionFormat::Zlib => { |
988 | let mut decompress = flate2::Decompress::new(true); |
989 | decompress |
990 | .decompress_vec( |
991 | self.data, |
992 | &mut decompressed, |
993 | flate2::FlushDecompress::Finish, |
994 | ) |
995 | .ok() |
996 | .read_error("Invalid zlib compressed data" )?; |
997 | } |
998 | CompressionFormat::Zstandard => { |
999 | let mut input = self.data; |
1000 | while !input.is_empty() { |
1001 | let mut decoder = match ruzstd::StreamingDecoder::new(&mut input) { |
1002 | Ok(decoder) => decoder, |
1003 | Err( |
1004 | ruzstd::frame_decoder::FrameDecoderError::ReadFrameHeaderError( |
1005 | ruzstd::frame::ReadFrameHeaderError::SkipFrame { |
1006 | length, |
1007 | .. |
1008 | }, |
1009 | ), |
1010 | ) => { |
1011 | input = input |
1012 | .get(length as usize..) |
1013 | .read_error("Invalid zstd compressed data" )?; |
1014 | continue; |
1015 | } |
1016 | x => x.ok().read_error("Invalid zstd compressed data" )?, |
1017 | }; |
1018 | decoder |
1019 | .read_to_end(&mut decompressed) |
1020 | .ok() |
1021 | .read_error("Invalid zstd compressed data" )?; |
1022 | } |
1023 | } |
1024 | _ => unreachable!(), |
1025 | } |
1026 | if size != decompressed.len() { |
1027 | return Err(Error( |
1028 | "Uncompressed data size does not match compression header" , |
1029 | )); |
1030 | } |
1031 | |
1032 | Ok(Cow::Owned(decompressed)) |
1033 | } |
1034 | _ => Err(Error("Unsupported compressed data." )), |
1035 | } |
1036 | } |
1037 | } |
1038 | |