From 1b6a04ca5504955c571d1c97504fb45ea0befee4 Mon Sep 17 00:00:00 2001 From: Valentin Popov Date: Mon, 8 Jan 2024 01:21:28 +0400 Subject: Initial vendor packages Signed-off-by: Valentin Popov --- vendor/object/src/read/macho/file.rs | 781 +++++++++++++++++++++++++++++++++++ 1 file changed, 781 insertions(+) create mode 100644 vendor/object/src/read/macho/file.rs (limited to 'vendor/object/src/read/macho/file.rs') diff --git a/vendor/object/src/read/macho/file.rs b/vendor/object/src/read/macho/file.rs new file mode 100644 index 0000000..0e04477 --- /dev/null +++ b/vendor/object/src/read/macho/file.rs @@ -0,0 +1,781 @@ +use alloc::vec::Vec; +use core::fmt::Debug; +use core::{mem, str}; + +use crate::read::{ + self, Architecture, ComdatKind, Error, Export, FileFlags, Import, NoDynamicRelocationIterator, + Object, ObjectComdat, ObjectKind, ObjectMap, ObjectSection, ReadError, ReadRef, Result, + SectionIndex, SubArchitecture, SymbolIndex, +}; +use crate::{endian, macho, BigEndian, ByteString, Endian, Endianness, Pod}; + +use super::{ + DyldCacheImage, LoadCommandIterator, MachOSection, MachOSectionInternal, MachOSectionIterator, + MachOSegment, MachOSegmentInternal, MachOSegmentIterator, MachOSymbol, MachOSymbolIterator, + MachOSymbolTable, Nlist, Section, Segment, SymbolTable, +}; + +/// A 32-bit Mach-O object file. +/// +/// This is a file that starts with [`macho::MachHeader32`], and corresponds +/// to [`crate::FileKind::MachO32`]. +pub type MachOFile32<'data, Endian = Endianness, R = &'data [u8]> = + MachOFile<'data, macho::MachHeader32, R>; +/// A 64-bit Mach-O object file. +/// +/// This is a file that starts with [`macho::MachHeader64`], and corresponds +/// to [`crate::FileKind::MachO64`]. +pub type MachOFile64<'data, Endian = Endianness, R = &'data [u8]> = + MachOFile<'data, macho::MachHeader64, R>; + +/// A partially parsed Mach-O file. +/// +/// Most of the functionality of this type is provided by the [`Object`] trait implementation. +#[derive(Debug)] +pub struct MachOFile<'data, Mach, R = &'data [u8]> +where + Mach: MachHeader, + R: ReadRef<'data>, +{ + pub(super) endian: Mach::Endian, + pub(super) data: R, + pub(super) header_offset: u64, + pub(super) header: &'data Mach, + pub(super) segments: Vec>, + pub(super) sections: Vec>, + pub(super) symbols: SymbolTable<'data, Mach, R>, +} + +impl<'data, Mach, R> MachOFile<'data, Mach, R> +where + Mach: MachHeader, + R: ReadRef<'data>, +{ + /// Parse the raw Mach-O file data. + pub fn parse(data: R) -> Result { + let header = Mach::parse(data, 0)?; + let endian = header.endian()?; + + // Build a list of segments and sections to make some operations more efficient. + let mut segments = Vec::new(); + let mut sections = Vec::new(); + let mut symbols = SymbolTable::default(); + if let Ok(mut commands) = header.load_commands(endian, data, 0) { + while let Ok(Some(command)) = commands.next() { + if let Some((segment, section_data)) = Mach::Segment::from_command(command)? { + let segment_index = segments.len(); + segments.push(MachOSegmentInternal { segment, data }); + for section in segment.sections(endian, section_data)? { + let index = SectionIndex(sections.len() + 1); + sections.push(MachOSectionInternal::parse(index, segment_index, section)); + } + } else if let Some(symtab) = command.symtab()? { + symbols = symtab.symbols(endian, data)?; + } + } + } + + Ok(MachOFile { + endian, + data, + header_offset: 0, + header, + segments, + sections, + symbols, + }) + } + + /// Parse the Mach-O file for the given image from the dyld shared cache. + /// This will read different sections from different subcaches, if necessary. + pub fn parse_dyld_cache_image<'cache, E: Endian>( + image: &DyldCacheImage<'data, 'cache, E, R>, + ) -> Result { + let (data, header_offset) = image.image_data_and_offset()?; + let header = Mach::parse(data, header_offset)?; + let endian = header.endian()?; + + // Build a list of sections to make some operations more efficient. + // Also build a list of segments, because we need to remember which ReadRef + // to read each section's data from. Only the DyldCache knows this information, + // and we won't have access to it once we've exited this function. + let mut segments = Vec::new(); + let mut sections = Vec::new(); + let mut linkedit_data: Option = None; + let mut symtab = None; + if let Ok(mut commands) = header.load_commands(endian, data, header_offset) { + while let Ok(Some(command)) = commands.next() { + if let Some((segment, section_data)) = Mach::Segment::from_command(command)? { + // Each segment can be stored in a different subcache. Get the segment's + // address and look it up in the cache mappings, to find the correct cache data. + let addr = segment.vmaddr(endian).into(); + let (data, _offset) = image + .cache + .data_and_offset_for_address(addr) + .read_error("Could not find segment data in dyld shared cache")?; + if segment.name() == macho::SEG_LINKEDIT.as_bytes() { + linkedit_data = Some(data); + } + let segment_index = segments.len(); + segments.push(MachOSegmentInternal { segment, data }); + + for section in segment.sections(endian, section_data)? { + let index = SectionIndex(sections.len() + 1); + sections.push(MachOSectionInternal::parse(index, segment_index, section)); + } + } else if let Some(st) = command.symtab()? { + symtab = Some(st); + } + } + } + + // The symbols are found in the __LINKEDIT segment, so make sure to read them from the + // correct subcache. + let symbols = match (symtab, linkedit_data) { + (Some(symtab), Some(linkedit_data)) => symtab.symbols(endian, linkedit_data)?, + _ => SymbolTable::default(), + }; + + Ok(MachOFile { + endian, + data, + header_offset, + header, + segments, + sections, + symbols, + }) + } + + /// Return the section at the given index. + #[inline] + pub(super) fn section_internal( + &self, + index: SectionIndex, + ) -> Result<&MachOSectionInternal<'data, Mach>> { + index + .0 + .checked_sub(1) + .and_then(|index| self.sections.get(index)) + .read_error("Invalid Mach-O section index") + } + + pub(super) fn segment_internal( + &self, + index: usize, + ) -> Result<&MachOSegmentInternal<'data, Mach, R>> { + self.segments + .get(index) + .read_error("Invalid Mach-O segment index") + } + + /// Returns the endianness. + pub fn endian(&self) -> Mach::Endian { + self.endian + } + + /// Returns the raw data. + pub fn data(&self) -> R { + self.data + } + + /// Returns the raw Mach-O file header. + pub fn raw_header(&self) -> &'data Mach { + self.header + } + + /// Return the `LC_BUILD_VERSION` load command if present. + pub fn build_version(&self) -> Result>> { + let mut commands = self + .header + .load_commands(self.endian, self.data, self.header_offset)?; + while let Some(command) = commands.next()? { + if let Some(build_version) = command.build_version()? { + return Ok(Some(build_version)); + } + } + Ok(None) + } +} + +impl<'data, Mach, R> read::private::Sealed for MachOFile<'data, Mach, R> +where + Mach: MachHeader, + R: ReadRef<'data>, +{ +} + +impl<'data, 'file, Mach, R> Object<'data, 'file> for MachOFile<'data, Mach, R> +where + 'data: 'file, + Mach: MachHeader, + R: 'file + ReadRef<'data>, +{ + type Segment = MachOSegment<'data, 'file, Mach, R>; + type SegmentIterator = MachOSegmentIterator<'data, 'file, Mach, R>; + type Section = MachOSection<'data, 'file, Mach, R>; + type SectionIterator = MachOSectionIterator<'data, 'file, Mach, R>; + type Comdat = MachOComdat<'data, 'file, Mach, R>; + type ComdatIterator = MachOComdatIterator<'data, 'file, Mach, R>; + type Symbol = MachOSymbol<'data, 'file, Mach, R>; + type SymbolIterator = MachOSymbolIterator<'data, 'file, Mach, R>; + type SymbolTable = MachOSymbolTable<'data, 'file, Mach, R>; + type DynamicRelocationIterator = NoDynamicRelocationIterator; + + fn architecture(&self) -> Architecture { + match self.header.cputype(self.endian) { + macho::CPU_TYPE_ARM => Architecture::Arm, + macho::CPU_TYPE_ARM64 => Architecture::Aarch64, + macho::CPU_TYPE_ARM64_32 => Architecture::Aarch64_Ilp32, + macho::CPU_TYPE_X86 => Architecture::I386, + macho::CPU_TYPE_X86_64 => Architecture::X86_64, + macho::CPU_TYPE_MIPS => Architecture::Mips, + macho::CPU_TYPE_POWERPC => Architecture::PowerPc, + macho::CPU_TYPE_POWERPC64 => Architecture::PowerPc64, + _ => Architecture::Unknown, + } + } + + fn sub_architecture(&self) -> Option { + match ( + self.header.cputype(self.endian), + self.header.cpusubtype(self.endian), + ) { + (macho::CPU_TYPE_ARM64, macho::CPU_SUBTYPE_ARM64E) => Some(SubArchitecture::Arm64E), + _ => None, + } + } + + #[inline] + fn is_little_endian(&self) -> bool { + self.header.is_little_endian() + } + + #[inline] + fn is_64(&self) -> bool { + self.header.is_type_64() + } + + fn kind(&self) -> ObjectKind { + match self.header.filetype(self.endian) { + macho::MH_OBJECT => ObjectKind::Relocatable, + macho::MH_EXECUTE => ObjectKind::Executable, + macho::MH_CORE => ObjectKind::Core, + macho::MH_DYLIB => ObjectKind::Dynamic, + _ => ObjectKind::Unknown, + } + } + + fn segments(&'file self) -> MachOSegmentIterator<'data, 'file, Mach, R> { + MachOSegmentIterator { + file: self, + iter: self.segments.iter(), + } + } + + fn section_by_name_bytes( + &'file self, + section_name: &[u8], + ) -> Option> { + // Translate the "." prefix to the "__" prefix used by OSX/Mach-O, eg + // ".debug_info" to "__debug_info", and limit to 16 bytes total. + let system_name = if section_name.starts_with(b".") { + if section_name.len() > 15 { + Some(§ion_name[1..15]) + } else { + Some(§ion_name[1..]) + } + } else { + None + }; + let cmp_section_name = |section: &MachOSection<'data, 'file, Mach, R>| { + section + .name_bytes() + .map(|name| { + section_name == name + || system_name + .filter(|system_name| { + name.starts_with(b"__") && name[2..] == **system_name + }) + .is_some() + }) + .unwrap_or(false) + }; + + self.sections().find(cmp_section_name) + } + + fn section_by_index( + &'file self, + index: SectionIndex, + ) -> Result> { + let internal = *self.section_internal(index)?; + Ok(MachOSection { + file: self, + internal, + }) + } + + fn sections(&'file self) -> MachOSectionIterator<'data, 'file, Mach, R> { + MachOSectionIterator { + file: self, + iter: self.sections.iter(), + } + } + + fn comdats(&'file self) -> MachOComdatIterator<'data, 'file, Mach, R> { + MachOComdatIterator { file: self } + } + + fn symbol_by_index( + &'file self, + index: SymbolIndex, + ) -> Result> { + let nlist = self.symbols.symbol(index.0)?; + MachOSymbol::new(self, index, nlist).read_error("Unsupported Mach-O symbol index") + } + + fn symbols(&'file self) -> MachOSymbolIterator<'data, 'file, Mach, R> { + MachOSymbolIterator { + file: self, + index: 0, + } + } + + #[inline] + fn symbol_table(&'file self) -> Option> { + Some(MachOSymbolTable { file: self }) + } + + fn dynamic_symbols(&'file self) -> MachOSymbolIterator<'data, 'file, Mach, R> { + MachOSymbolIterator { + file: self, + index: self.symbols.len(), + } + } + + #[inline] + fn dynamic_symbol_table(&'file self) -> Option> { + None + } + + fn object_map(&'file self) -> ObjectMap<'data> { + self.symbols.object_map(self.endian) + } + + fn imports(&self) -> Result>> { + let mut dysymtab = None; + let mut libraries = Vec::new(); + let twolevel = self.header.flags(self.endian) & macho::MH_TWOLEVEL != 0; + if twolevel { + libraries.push(&[][..]); + } + let mut commands = self + .header + .load_commands(self.endian, self.data, self.header_offset)?; + while let Some(command) = commands.next()? { + if let Some(command) = command.dysymtab()? { + dysymtab = Some(command); + } + if twolevel { + if let Some(dylib) = command.dylib()? { + libraries.push(command.string(self.endian, dylib.dylib.name)?); + } + } + } + + let mut imports = Vec::new(); + if let Some(dysymtab) = dysymtab { + let index = dysymtab.iundefsym.get(self.endian) as usize; + let number = dysymtab.nundefsym.get(self.endian) as usize; + for i in index..(index.wrapping_add(number)) { + let symbol = self.symbols.symbol(i)?; + let name = symbol.name(self.endian, self.symbols.strings())?; + let library = if twolevel { + libraries + .get(symbol.library_ordinal(self.endian) as usize) + .copied() + .read_error("Invalid Mach-O symbol library ordinal")? + } else { + &[] + }; + imports.push(Import { + name: ByteString(name), + library: ByteString(library), + }); + } + } + Ok(imports) + } + + fn exports(&self) -> Result>> { + let mut dysymtab = None; + let mut commands = self + .header + .load_commands(self.endian, self.data, self.header_offset)?; + while let Some(command) = commands.next()? { + if let Some(command) = command.dysymtab()? { + dysymtab = Some(command); + break; + } + } + + let mut exports = Vec::new(); + if let Some(dysymtab) = dysymtab { + let index = dysymtab.iextdefsym.get(self.endian) as usize; + let number = dysymtab.nextdefsym.get(self.endian) as usize; + for i in index..(index.wrapping_add(number)) { + let symbol = self.symbols.symbol(i)?; + let name = symbol.name(self.endian, self.symbols.strings())?; + let address = symbol.n_value(self.endian).into(); + exports.push(Export { + name: ByteString(name), + address, + }); + } + } + Ok(exports) + } + + #[inline] + fn dynamic_relocations(&'file self) -> Option { + None + } + + fn has_debug_symbols(&self) -> bool { + self.section_by_name(".debug_info").is_some() + } + + fn mach_uuid(&self) -> Result> { + self.header.uuid(self.endian, self.data, self.header_offset) + } + + fn relative_address_base(&self) -> u64 { + 0 + } + + fn entry(&self) -> u64 { + if let Ok(mut commands) = + self.header + .load_commands(self.endian, self.data, self.header_offset) + { + while let Ok(Some(command)) = commands.next() { + if let Ok(Some(command)) = command.entry_point() { + return command.entryoff.get(self.endian); + } + } + } + 0 + } + + fn flags(&self) -> FileFlags { + FileFlags::MachO { + flags: self.header.flags(self.endian), + } + } +} + +/// An iterator for the COMDAT section groups in a [`MachOFile64`]. +pub type MachOComdatIterator32<'data, 'file, Endian = Endianness, R = &'data [u8]> = + MachOComdatIterator<'data, 'file, macho::MachHeader32, R>; +/// An iterator for the COMDAT section groups in a [`MachOFile64`]. +pub type MachOComdatIterator64<'data, 'file, Endian = Endianness, R = &'data [u8]> = + MachOComdatIterator<'data, 'file, macho::MachHeader64, R>; + +/// An iterator for the COMDAT section groups in a [`MachOFile`]. +/// +/// This is a stub that doesn't implement any functionality. +#[derive(Debug)] +pub struct MachOComdatIterator<'data, 'file, Mach, R = &'data [u8]> +where + Mach: MachHeader, + R: ReadRef<'data>, +{ + #[allow(unused)] + file: &'file MachOFile<'data, Mach, R>, +} + +impl<'data, 'file, Mach, R> Iterator for MachOComdatIterator<'data, 'file, Mach, R> +where + Mach: MachHeader, + R: ReadRef<'data>, +{ + type Item = MachOComdat<'data, 'file, Mach, R>; + + #[inline] + fn next(&mut self) -> Option { + None + } +} + +/// A COMDAT section group in a [`MachOFile32`]. +pub type MachOComdat32<'data, 'file, Endian = Endianness, R = &'data [u8]> = + MachOComdat<'data, 'file, macho::MachHeader32, R>; + +/// A COMDAT section group in a [`MachOFile64`]. +pub type MachOComdat64<'data, 'file, Endian = Endianness, R = &'data [u8]> = + MachOComdat<'data, 'file, macho::MachHeader64, R>; + +/// A COMDAT section group in a [`MachOFile`]. +/// +/// This is a stub that doesn't implement any functionality. +#[derive(Debug)] +pub struct MachOComdat<'data, 'file, Mach, R = &'data [u8]> +where + Mach: MachHeader, + R: ReadRef<'data>, +{ + #[allow(unused)] + file: &'file MachOFile<'data, Mach, R>, +} + +impl<'data, 'file, Mach, R> read::private::Sealed for MachOComdat<'data, 'file, Mach, R> +where + Mach: MachHeader, + R: ReadRef<'data>, +{ +} + +impl<'data, 'file, Mach, R> ObjectComdat<'data> for MachOComdat<'data, 'file, Mach, R> +where + Mach: MachHeader, + R: ReadRef<'data>, +{ + type SectionIterator = MachOComdatSectionIterator<'data, 'file, Mach, R>; + + #[inline] + fn kind(&self) -> ComdatKind { + unreachable!(); + } + + #[inline] + fn symbol(&self) -> SymbolIndex { + unreachable!(); + } + + #[inline] + fn name_bytes(&self) -> Result<&[u8]> { + unreachable!(); + } + + #[inline] + fn name(&self) -> Result<&str> { + unreachable!(); + } + + #[inline] + fn sections(&self) -> Self::SectionIterator { + unreachable!(); + } +} + +/// An iterator for the sections in a COMDAT section group in a [`MachOFile32`]. +pub type MachOComdatSectionIterator32<'data, 'file, Endian = Endianness, R = &'data [u8]> = + MachOComdatSectionIterator<'data, 'file, macho::MachHeader32, R>; +/// An iterator for the sections in a COMDAT section group in a [`MachOFile64`]. +pub type MachOComdatSectionIterator64<'data, 'file, Endian = Endianness, R = &'data [u8]> = + MachOComdatSectionIterator<'data, 'file, macho::MachHeader64, R>; + +/// An iterator for the sections in a COMDAT section group in a [`MachOFile`]. +/// +/// This is a stub that doesn't implement any functionality. +#[derive(Debug)] +pub struct MachOComdatSectionIterator<'data, 'file, Mach, R = &'data [u8]> +where + Mach: MachHeader, + R: ReadRef<'data>, +{ + #[allow(unused)] + file: &'file MachOFile<'data, Mach, R>, +} + +impl<'data, 'file, Mach, R> Iterator for MachOComdatSectionIterator<'data, 'file, Mach, R> +where + Mach: MachHeader, + R: ReadRef<'data>, +{ + type Item = SectionIndex; + + fn next(&mut self) -> Option { + None + } +} + +/// A trait for generic access to [`macho::MachHeader32`] and [`macho::MachHeader64`]. +#[allow(missing_docs)] +pub trait MachHeader: Debug + Pod { + type Word: Into; + type Endian: endian::Endian; + type Segment: Segment; + type Section: Section; + type Nlist: Nlist; + + /// Return true if this type is a 64-bit header. + /// + /// This is a property of the type, not a value in the header data. + fn is_type_64(&self) -> bool; + + /// Return true if the `magic` field signifies big-endian. + fn is_big_endian(&self) -> bool; + + /// Return true if the `magic` field signifies little-endian. + fn is_little_endian(&self) -> bool; + + fn magic(&self) -> u32; + fn cputype(&self, endian: Self::Endian) -> u32; + fn cpusubtype(&self, endian: Self::Endian) -> u32; + fn filetype(&self, endian: Self::Endian) -> u32; + fn ncmds(&self, endian: Self::Endian) -> u32; + fn sizeofcmds(&self, endian: Self::Endian) -> u32; + fn flags(&self, endian: Self::Endian) -> u32; + + // Provided methods. + + /// Read the file header. + /// + /// Also checks that the magic field in the file header is a supported format. + fn parse<'data, R: ReadRef<'data>>(data: R, offset: u64) -> read::Result<&'data Self> { + let header = data + .read_at::(offset) + .read_error("Invalid Mach-O header size or alignment")?; + if !header.is_supported() { + return Err(Error("Unsupported Mach-O header")); + } + Ok(header) + } + + fn is_supported(&self) -> bool { + self.is_little_endian() || self.is_big_endian() + } + + fn endian(&self) -> Result { + Self::Endian::from_big_endian(self.is_big_endian()).read_error("Unsupported Mach-O endian") + } + + fn load_commands<'data, R: ReadRef<'data>>( + &self, + endian: Self::Endian, + data: R, + header_offset: u64, + ) -> Result> { + let data = data + .read_bytes_at( + header_offset + mem::size_of::() as u64, + self.sizeofcmds(endian).into(), + ) + .read_error("Invalid Mach-O load command table size")?; + Ok(LoadCommandIterator::new(endian, data, self.ncmds(endian))) + } + + /// Return the UUID from the `LC_UUID` load command, if one is present. + fn uuid<'data, R: ReadRef<'data>>( + &self, + endian: Self::Endian, + data: R, + header_offset: u64, + ) -> Result> { + let mut commands = self.load_commands(endian, data, header_offset)?; + while let Some(command) = commands.next()? { + if let Ok(Some(uuid)) = command.uuid() { + return Ok(Some(uuid.uuid)); + } + } + Ok(None) + } +} + +impl MachHeader for macho::MachHeader32 { + type Word = u32; + type Endian = Endian; + type Segment = macho::SegmentCommand32; + type Section = macho::Section32; + type Nlist = macho::Nlist32; + + fn is_type_64(&self) -> bool { + false + } + + fn is_big_endian(&self) -> bool { + self.magic() == macho::MH_MAGIC + } + + fn is_little_endian(&self) -> bool { + self.magic() == macho::MH_CIGAM + } + + fn magic(&self) -> u32 { + self.magic.get(BigEndian) + } + + fn cputype(&self, endian: Self::Endian) -> u32 { + self.cputype.get(endian) + } + + fn cpusubtype(&self, endian: Self::Endian) -> u32 { + self.cpusubtype.get(endian) + } + + fn filetype(&self, endian: Self::Endian) -> u32 { + self.filetype.get(endian) + } + + fn ncmds(&self, endian: Self::Endian) -> u32 { + self.ncmds.get(endian) + } + + fn sizeofcmds(&self, endian: Self::Endian) -> u32 { + self.sizeofcmds.get(endian) + } + + fn flags(&self, endian: Self::Endian) -> u32 { + self.flags.get(endian) + } +} + +impl MachHeader for macho::MachHeader64 { + type Word = u64; + type Endian = Endian; + type Segment = macho::SegmentCommand64; + type Section = macho::Section64; + type Nlist = macho::Nlist64; + + fn is_type_64(&self) -> bool { + true + } + + fn is_big_endian(&self) -> bool { + self.magic() == macho::MH_MAGIC_64 + } + + fn is_little_endian(&self) -> bool { + self.magic() == macho::MH_CIGAM_64 + } + + fn magic(&self) -> u32 { + self.magic.get(BigEndian) + } + + fn cputype(&self, endian: Self::Endian) -> u32 { + self.cputype.get(endian) + } + + fn cpusubtype(&self, endian: Self::Endian) -> u32 { + self.cpusubtype.get(endian) + } + + fn filetype(&self, endian: Self::Endian) -> u32 { + self.filetype.get(endian) + } + + fn ncmds(&self, endian: Self::Endian) -> u32 { + self.ncmds.get(endian) + } + + fn sizeofcmds(&self, endian: Self::Endian) -> u32 { + self.sizeofcmds.get(endian) + } + + fn flags(&self, endian: Self::Endian) -> u32 { + self.flags.get(endian) + } +} -- cgit v1.2.3