object/read/

mod.rs

//! Interface for reading object files.
//!
//! ## Unified read API
//!
//! The [`Object`] trait provides a unified read API for accessing common features of
//! object files, such as sections and symbols. There is an implementation of this
//! trait for [`File`], which allows reading any file format, as well as implementations
//! for each file format:
//! [`ElfFile`](elf::ElfFile), [`MachOFile`](macho::MachOFile), [`CoffFile`](coff::CoffFile),
//! [`PeFile`](pe::PeFile), [`WasmFile`](wasm::WasmFile), [`XcoffFile`](xcoff::XcoffFile).
//!
//! ## Low level read API
//!
//! The submodules for each file format define helpers that operate on the raw structs.
//! These can be used instead of the unified API, or in conjunction with it to access
//! details that are not available via the unified API.
//!
//! See the [submodules](#modules) for examples of the low level read API.
//!
//! ## Naming Convention
//!
//! Types that form part of the unified API for a file format are prefixed with the
//! name of the file format.
//!
//! ## Example for unified read API
//!  ```no_run
//! use object::{Object, ObjectSection};
//! use std::error::Error;
//! use std::fs;
//!
//! /// Reads a file and displays the name of each section.
//! fn main() -> Result<(), Box<dyn Error>> {
//! #   #[cfg(all(feature = "read", feature = "std"))] {
//!     let data = fs::read("path/to/binary")?;
//!     let file = object::File::parse(&*data)?;
//!     for section in file.sections() {
//!         println!("{}", section.name()?);
//!     }
//! #   }
//!     Ok(())
//! }
//! ```

use alloc::borrow::Cow;
use alloc::vec::Vec;
use core::{fmt, result};

#[cfg(not(feature = "std"))]
use alloc::collections::btree_map::BTreeMap as Map;
#[cfg(feature = "std")]
use std::collections::hash_map::HashMap as Map;

pub use crate::common::*;

mod read_ref;
pub use read_ref::*;

mod read_cache;
pub use read_cache::*;

mod util;
pub use util::*;

#[cfg(any(feature = "elf", feature = "macho"))]
mod gnu_compression;

#[cfg(any(
    feature = "coff",
    feature = "elf",
    feature = "macho",
    feature = "pe",
    feature = "wasm",
    feature = "xcoff"
))]
mod any;
#[cfg(any(
    feature = "coff",
    feature = "elf",
    feature = "macho",
    feature = "pe",
    feature = "wasm",
    feature = "xcoff"
))]
pub use any::*;

#[cfg(feature = "archive")]
pub mod archive;

#[cfg(feature = "coff")]
pub mod coff;

#[cfg(feature = "elf")]
pub mod elf;

#[cfg(feature = "macho")]
pub mod macho;

#[cfg(feature = "pe")]
pub mod pe;

#[cfg(feature = "wasm")]
pub mod wasm;

#[cfg(feature = "xcoff")]
pub mod xcoff;

mod traits;
pub use traits::*;

mod private {
    pub trait Sealed {}
}

/// The error type used within the read module.
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub struct Error(pub(crate) &'static str);

impl fmt::Display for Error {
    #[inline]
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        f.write_str(self.0)
    }
}

#[cfg(feature = "std")]
impl std::error::Error for Error {}

/// The result type used within the read module.
pub type Result<T> = result::Result<T, Error>;

trait ReadError<T> {
    fn read_error(self, error: &'static str) -> Result<T>;
}

impl<T> ReadError<T> for result::Result<T, ()> {
    fn read_error(self, error: &'static str) -> Result<T> {
        self.map_err(|()| Error(error))
    }
}

impl<T> ReadError<T> for result::Result<T, Error> {
    fn read_error(self, error: &'static str) -> Result<T> {
        self.map_err(|_| Error(error))
    }
}

impl<T> ReadError<T> for Option<T> {
    fn read_error(self, error: &'static str) -> Result<T> {
        self.ok_or(Error(error))
    }
}

/// The native executable file for the target platform.
#[cfg(all(
    unix,
    not(target_os = "macos"),
    target_pointer_width = "32",
    feature = "elf"
))]
pub type NativeFile<'data, R = &'data [u8]> = elf::ElfFile32<'data, crate::endian::Endianness, R>;

/// The native executable file for the target platform.
#[cfg(all(
    unix,
    not(target_os = "macos"),
    target_pointer_width = "64",
    feature = "elf"
))]
pub type NativeFile<'data, R = &'data [u8]> = elf::ElfFile64<'data, crate::endian::Endianness, R>;

/// The native executable file for the target platform.
#[cfg(all(target_os = "macos", target_pointer_width = "32", feature = "macho"))]
pub type NativeFile<'data, R = &'data [u8]> =
    macho::MachOFile32<'data, crate::endian::Endianness, R>;

/// The native executable file for the target platform.
#[cfg(all(target_os = "macos", target_pointer_width = "64", feature = "macho"))]
pub type NativeFile<'data, R = &'data [u8]> =
    macho::MachOFile64<'data, crate::endian::Endianness, R>;

/// The native executable file for the target platform.
#[cfg(all(target_os = "windows", target_pointer_width = "32", feature = "pe"))]
pub type NativeFile<'data, R = &'data [u8]> = pe::PeFile32<'data, R>;

/// The native executable file for the target platform.
#[cfg(all(target_os = "windows", target_pointer_width = "64", feature = "pe"))]
pub type NativeFile<'data, R = &'data [u8]> = pe::PeFile64<'data, R>;

/// The native executable file for the target platform.
#[cfg(all(feature = "wasm", target_arch = "wasm32", feature = "wasm"))]
pub type NativeFile<'data, R = &'data [u8]> = wasm::WasmFile<'data, R>;

/// A file format kind.
#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
#[non_exhaustive]
pub enum FileKind {
    /// A Unix archive.
    ///
    /// See [`archive::ArchiveFile`].
    #[cfg(feature = "archive")]
    Archive,
    /// A COFF object file.
    ///
    /// See [`coff::CoffFile`].
    #[cfg(feature = "coff")]
    Coff,
    /// A COFF bigobj object file.
    ///
    /// This supports a larger number of sections.
    ///
    /// See [`coff::CoffBigFile`].
    #[cfg(feature = "coff")]
    CoffBig,
    /// A Windows short import file.
    ///
    /// See [`coff::ImportFile`].
    #[cfg(feature = "coff")]
    CoffImport,
    /// A dyld cache file containing Mach-O images.
    ///
    /// See [`macho::DyldCache`]
    #[cfg(feature = "macho")]
    DyldCache,
    /// A 32-bit ELF file.
    ///
    /// See [`elf::ElfFile32`].
    #[cfg(feature = "elf")]
    Elf32,
    /// A 64-bit ELF file.
    ///
    /// See [`elf::ElfFile64`].
    #[cfg(feature = "elf")]
    Elf64,
    /// A 32-bit Mach-O file.
    ///
    /// See [`macho::MachOFile32`].
    #[cfg(feature = "macho")]
    MachO32,
    /// A 64-bit Mach-O file.
    ///
    /// See [`macho::MachOFile64`].
    #[cfg(feature = "macho")]
    MachO64,
    /// A 32-bit Mach-O fat binary.
    ///
    /// See [`macho::MachOFatFile32`].
    #[cfg(feature = "macho")]
    MachOFat32,
    /// A 64-bit Mach-O fat binary.
    ///
    /// See [`macho::MachOFatFile64`].
    #[cfg(feature = "macho")]
    MachOFat64,
    /// A 32-bit PE file.
    ///
    /// See [`pe::PeFile32`].
    #[cfg(feature = "pe")]
    Pe32,
    /// A 64-bit PE file.
    ///
    /// See [`pe::PeFile64`].
    #[cfg(feature = "pe")]
    Pe64,
    /// A Wasm file.
    ///
    /// See [`wasm::WasmFile`].
    #[cfg(feature = "wasm")]
    Wasm,
    /// A 32-bit XCOFF file.
    ///
    /// See [`xcoff::XcoffFile32`].
    #[cfg(feature = "xcoff")]
    Xcoff32,
    /// A 64-bit XCOFF file.
    ///
    /// See [`xcoff::XcoffFile64`].
    #[cfg(feature = "xcoff")]
    Xcoff64,
}

impl FileKind {
    /// Determine a file kind by parsing the start of the file.
    pub fn parse<'data, R: ReadRef<'data>>(data: R) -> Result<FileKind> {
        Self::parse_at(data, 0)
    }

    /// Determine a file kind by parsing at the given offset.
    pub fn parse_at<'data, R: ReadRef<'data>>(data: R, offset: u64) -> Result<FileKind> {
        let magic = data
            .read_bytes_at(offset, 16)
            .read_error("Could not read file magic")?;
        if magic.len() < 16 {
            return Err(Error("File too short"));
        }

        let kind = match [magic[0], magic[1], magic[2], magic[3], magic[4], magic[5], magic[6], magic[7]] {
            #[cfg(feature = "archive")]
            [b'!', b'<', b'a', b'r', b'c', b'h', b'>', b'\n']
            | [b'!', b'<', b't', b'h', b'i', b'n', b'>', b'\n'] => FileKind::Archive,
            #[cfg(feature = "macho")]
            [b'd', b'y', b'l', b'd', b'_', b'v', b'1', b' '] => FileKind::DyldCache,
            #[cfg(feature = "elf")]
            [0x7f, b'E', b'L', b'F', 1, ..] => FileKind::Elf32,
            #[cfg(feature = "elf")]
            [0x7f, b'E', b'L', b'F', 2, ..] => FileKind::Elf64,
            #[cfg(feature = "macho")]
            [0xfe, 0xed, 0xfa, 0xce, ..]
            | [0xce, 0xfa, 0xed, 0xfe, ..] => FileKind::MachO32,
            #[cfg(feature = "macho")]
            | [0xfe, 0xed, 0xfa, 0xcf, ..]
            | [0xcf, 0xfa, 0xed, 0xfe, ..] => FileKind::MachO64,
            #[cfg(feature = "macho")]
            [0xca, 0xfe, 0xba, 0xbe, ..] => FileKind::MachOFat32,
            #[cfg(feature = "macho")]
            [0xca, 0xfe, 0xba, 0xbf, ..] => FileKind::MachOFat64,
            #[cfg(feature = "wasm")]
            [0x00, b'a', b's', b'm', _, _, 0x00, 0x00] => FileKind::Wasm,
            #[cfg(feature = "pe")]
            [b'M', b'Z', ..] if offset == 0 => {
                // offset == 0 restriction is because optional_header_magic only looks at offset 0
                match pe::optional_header_magic(data) {
                    Ok(crate::pe::IMAGE_NT_OPTIONAL_HDR32_MAGIC) => {
                        FileKind::Pe32
                    }
                    Ok(crate::pe::IMAGE_NT_OPTIONAL_HDR64_MAGIC) => {
                        FileKind::Pe64
                    }
                    _ => return Err(Error("Unknown MS-DOS file")),
                }
            }
            // TODO: more COFF machines
            #[cfg(feature = "coff")]
            // COFF arm
            [0xc4, 0x01, ..]
            // COFF arm64
            | [0x64, 0xaa, ..]
            // COFF arm64ec
            | [0x41, 0xa6, ..]
            // COFF x86
            | [0x4c, 0x01, ..]
            // COFF x86-64
            | [0x64, 0x86, ..] => FileKind::Coff,
            #[cfg(feature = "coff")]
            [0x00, 0x00, 0xff, 0xff, 0x00, 0x00, ..] => FileKind::CoffImport,
            #[cfg(feature = "coff")]
            [0x00, 0x00, 0xff, 0xff, 0x02, 0x00, ..] if offset == 0 => {
                // offset == 0 restriction is because anon_object_class_id only looks at offset 0
                match coff::anon_object_class_id(data) {
                    Ok(crate::pe::ANON_OBJECT_HEADER_BIGOBJ_CLASS_ID) => FileKind::CoffBig,
                    _ => return Err(Error("Unknown anon object file")),
                }
            }
            #[cfg(feature = "xcoff")]
            [0x01, 0xdf, ..] => FileKind::Xcoff32,
            #[cfg(feature = "xcoff")]
            [0x01, 0xf7, ..] => FileKind::Xcoff64,
            _ => return Err(Error("Unknown file magic")),
        };
        Ok(kind)
    }
}

/// An object kind.
///
/// Returned by [`Object::kind`].
#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
#[non_exhaustive]
pub enum ObjectKind {
    /// The object kind is unknown.
    Unknown,
    /// Relocatable object.
    Relocatable,
    /// Executable.
    Executable,
    /// Dynamic shared object.
    Dynamic,
    /// Core.
    Core,
}

/// The index used to identify a section in a file.
#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
pub struct SectionIndex(pub usize);

impl fmt::Display for SectionIndex {
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        self.0.fmt(f)
    }
}

/// The index used to identify a symbol in a symbol table.
#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
pub struct SymbolIndex(pub usize);

impl fmt::Display for SymbolIndex {
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        self.0.fmt(f)
    }
}

/// The section where an [`ObjectSymbol`] is defined.
#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
#[non_exhaustive]
pub enum SymbolSection {
    /// The section is unknown.
    Unknown,
    /// The section is not applicable for this symbol (such as file symbols).
    None,
    /// The symbol is undefined.
    Undefined,
    /// The symbol has an absolute value.
    Absolute,
    /// The symbol is a zero-initialized symbol that will be combined with duplicate definitions.
    Common,
    /// The symbol is defined in the given section.
    Section(SectionIndex),
}

impl SymbolSection {
    /// Returns the section index for the section where the symbol is defined.
    ///
    /// May return `None` if the symbol is not defined in a section.
    #[inline]
    pub fn index(self) -> Option<SectionIndex> {
        if let SymbolSection::Section(index) = self {
            Some(index)
        } else {
            None
        }
    }
}

/// An entry in a [`SymbolMap`].
pub trait SymbolMapEntry {
    /// The symbol address.
    fn address(&self) -> u64;
}

/// A map from addresses to symbol information.
///
/// The symbol information depends on the chosen entry type, such as [`SymbolMapName`].
///
/// Returned by [`Object::symbol_map`].
#[derive(Debug, Default, Clone)]
pub struct SymbolMap<T: SymbolMapEntry> {
    symbols: Vec<T>,
}

impl<T: SymbolMapEntry> SymbolMap<T> {
    /// Construct a new symbol map.
    ///
    /// This function will sort the symbols by address.
    pub fn new(mut symbols: Vec<T>) -> Self {
        symbols.sort_by_key(|s| s.address());
        SymbolMap { symbols }
    }

    /// Get the symbol before the given address.
    pub fn get(&self, address: u64) -> Option<&T> {
        let index = match self
            .symbols
            .binary_search_by_key(&address, |symbol| symbol.address())
        {
            Ok(index) => index,
            Err(index) => index.checked_sub(1)?,
        };
        self.symbols.get(index)
    }

    /// Get all symbols in the map.
    #[inline]
    pub fn symbols(&self) -> &[T] {
        &self.symbols
    }
}

/// The type used for entries in a [`SymbolMap`] that maps from addresses to names.
#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
pub struct SymbolMapName<'data> {
    address: u64,
    name: &'data str,
}

impl<'data> SymbolMapName<'data> {
    /// Construct a `SymbolMapName`.
    pub fn new(address: u64, name: &'data str) -> Self {
        SymbolMapName { address, name }
    }

    /// The symbol address.
    #[inline]
    pub fn address(&self) -> u64 {
        self.address
    }

    /// The symbol name.
    #[inline]
    pub fn name(&self) -> &'data str {
        self.name
    }
}

impl<'data> SymbolMapEntry for SymbolMapName<'data> {
    #[inline]
    fn address(&self) -> u64 {
        self.address
    }
}

/// A map from addresses to symbol names and object files.
///
/// This is derived from STAB entries in Mach-O files.
///
/// Returned by [`Object::object_map`].
#[derive(Debug, Default, Clone)]
pub struct ObjectMap<'data> {
    symbols: SymbolMap<ObjectMapEntry<'data>>,
    objects: Vec<ObjectMapFile<'data>>,
}

impl<'data> ObjectMap<'data> {
    /// Get the entry containing the given address.
    pub fn get(&self, address: u64) -> Option<&ObjectMapEntry<'data>> {
        self.symbols
            .get(address)
            .filter(|entry| entry.size == 0 || address.wrapping_sub(entry.address) < entry.size)
    }

    /// Get all symbols in the map.
    #[inline]
    pub fn symbols(&self) -> &[ObjectMapEntry<'data>] {
        self.symbols.symbols()
    }

    /// Get all objects in the map.
    #[inline]
    pub fn objects(&self) -> &[ObjectMapFile<'data>] {
        &self.objects
    }
}

/// A symbol in an [`ObjectMap`].
#[derive(Debug, Default, Clone, Copy, PartialEq, Eq, Hash)]
pub struct ObjectMapEntry<'data> {
    address: u64,
    size: u64,
    name: &'data [u8],
    object: usize,
}

impl<'data> ObjectMapEntry<'data> {
    /// Get the symbol address.
    #[inline]
    pub fn address(&self) -> u64 {
        self.address
    }

    /// Get the symbol size.
    ///
    /// This may be 0 if the size is unknown.
    #[inline]
    pub fn size(&self) -> u64 {
        self.size
    }

    /// Get the symbol name.
    #[inline]
    pub fn name(&self) -> &'data [u8] {
        self.name
    }

    /// Get the index of the object file name.
    #[inline]
    pub fn object_index(&self) -> usize {
        self.object
    }

    /// Get the object file name.
    #[inline]
    pub fn object<'a>(&self, map: &'a ObjectMap<'data>) -> &'a ObjectMapFile<'data> {
        &map.objects[self.object]
    }
}

impl<'data> SymbolMapEntry for ObjectMapEntry<'data> {
    #[inline]
    fn address(&self) -> u64 {
        self.address
    }
}

/// An object file name in an [`ObjectMap`].
#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
pub struct ObjectMapFile<'data> {
    path: &'data [u8],
    member: Option<&'data [u8]>,
}

impl<'data> ObjectMapFile<'data> {
    #[cfg(feature = "macho")]
    fn new(path: &'data [u8], member: Option<&'data [u8]>) -> Self {
        ObjectMapFile { path, member }
    }

    /// Get the path to the file containing the object.
    #[inline]
    pub fn path(&self) -> &'data [u8] {
        self.path
    }

    /// If the file is an archive, get the name of the member containing the object.
    #[inline]
    pub fn member(&self) -> Option<&'data [u8]> {
        self.member
    }
}

/// An imported symbol.
///
/// Returned by [`Object::imports`].
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub struct Import<'data> {
    library: ByteString<'data>,
    // TODO: or ordinal
    name: ByteString<'data>,
}

impl<'data> Import<'data> {
    /// The symbol name.
    #[inline]
    pub fn name(&self) -> &'data [u8] {
        self.name.0
    }

    /// The name of the library to import the symbol from.
    #[inline]
    pub fn library(&self) -> &'data [u8] {
        self.library.0
    }
}

/// An exported symbol.
///
/// Returned by [`Object::exports`].
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub struct Export<'data> {
    // TODO: and ordinal?
    name: ByteString<'data>,
    address: u64,
}

impl<'data> Export<'data> {
    /// The symbol name.
    #[inline]
    pub fn name(&self) -> &'data [u8] {
        self.name.0
    }

    /// The virtual address of the symbol.
    #[inline]
    pub fn address(&self) -> u64 {
        self.address
    }
}

/// PDB information from the debug directory in a PE file.
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub struct CodeView<'data> {
    guid: [u8; 16],
    path: ByteString<'data>,
    age: u32,
}

impl<'data> CodeView<'data> {
    /// The path to the PDB as stored in CodeView.
    #[inline]
    pub fn path(&self) -> &'data [u8] {
        self.path.0
    }

    /// The age of the PDB.
    #[inline]
    pub fn age(&self) -> u32 {
        self.age
    }

    /// The GUID of the PDB.
    #[inline]
    pub fn guid(&self) -> [u8; 16] {
        self.guid
    }
}

/// The target referenced by a [`Relocation`].
#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
#[non_exhaustive]
pub enum RelocationTarget {
    /// The target is a symbol.
    Symbol(SymbolIndex),
    /// The target is a section.
    Section(SectionIndex),
    /// The offset is an absolute address.
    Absolute,
}

/// A relocation entry.
///
/// Returned by [`Object::dynamic_relocations`] or [`ObjectSection::relocations`].
#[derive(Debug)]
pub struct Relocation {
    kind: RelocationKind,
    encoding: RelocationEncoding,
    size: u8,
    target: RelocationTarget,
    addend: i64,
    implicit_addend: bool,
    flags: RelocationFlags,
}

impl Relocation {
    /// The operation used to calculate the result of the relocation.
    #[inline]
    pub fn kind(&self) -> RelocationKind {
        self.kind
    }

    /// Information about how the result of the relocation operation is encoded in the place.
    #[inline]
    pub fn encoding(&self) -> RelocationEncoding {
        self.encoding
    }

    /// The size in bits of the place of the relocation.
    ///
    /// If 0, then the size is determined by the relocation kind.
    #[inline]
    pub fn size(&self) -> u8 {
        self.size
    }

    /// The target of the relocation.
    #[inline]
    pub fn target(&self) -> RelocationTarget {
        self.target
    }

    /// The addend to use in the relocation calculation.
    #[inline]
    pub fn addend(&self) -> i64 {
        self.addend
    }

    /// Set the addend to use in the relocation calculation.
    #[inline]
    pub fn set_addend(&mut self, addend: i64) {
        self.addend = addend;
    }

    /// Returns true if there is an implicit addend stored in the data at the offset
    /// to be relocated.
    #[inline]
    pub fn has_implicit_addend(&self) -> bool {
        self.implicit_addend
    }

    /// Relocation flags that are specific to each file format.
    ///
    /// The values returned by `kind`, `encoding` and `size` are derived
    /// from these flags.
    #[inline]
    pub fn flags(&self) -> RelocationFlags {
        self.flags
    }
}

/// A map from section offsets to relocation information.
///
/// This can be used to apply relocations to a value at a given section offset.
/// This is intended for use with DWARF in relocatable object files, and only
/// supports relocations that are used in DWARF.
///
/// Returned by [`ObjectSection::relocation_map`].
#[derive(Debug, Default)]
pub struct RelocationMap(Map<u64, RelocationMapEntry>);

impl RelocationMap {
    /// Construct a new relocation map for a section.
    ///
    /// Fails if any relocation cannot be added to the map.
    /// You can manually use `add` if you need different error handling,
    /// such as to list all errors or to ignore them.
    pub fn new<'data, 'file, T>(file: &'file T, section: &T::Section<'file>) -> Result<Self>
    where
        T: Object<'data>,
    {
        let mut map = RelocationMap(Map::new());
        for (offset, relocation) in section.relocations() {
            map.add(file, offset, relocation)?;
        }
        Ok(map)
    }

    /// Add a single relocation to the map.
    pub fn add<'data: 'file, 'file, T>(
        &mut self,
        file: &'file T,
        offset: u64,
        relocation: Relocation,
    ) -> Result<()>
    where
        T: Object<'data>,
    {
        let mut entry = RelocationMapEntry {
            implicit_addend: relocation.has_implicit_addend(),
            addend: relocation.addend() as u64,
        };
        match relocation.kind() {
            RelocationKind::Absolute => match relocation.target() {
                RelocationTarget::Symbol(symbol_idx) => {
                    let symbol = file
                        .symbol_by_index(symbol_idx)
                        .read_error("Relocation with invalid symbol")?;
                    entry.addend = symbol.address().wrapping_add(entry.addend);
                }
                RelocationTarget::Section(section_idx) => {
                    let section = file
                        .section_by_index(section_idx)
                        .read_error("Relocation with invalid section")?;
                    // DWARF parsers expect references to DWARF sections to be section offsets,
                    // not addresses. Addresses are useful for everything else.
                    if section.kind() != SectionKind::Debug {
                        entry.addend = section.address().wrapping_add(entry.addend);
                    }
                }
                _ => {
                    return Err(Error("Unsupported relocation target"));
                }
            },
            _ => {
                return Err(Error("Unsupported relocation type"));
            }
        }
        if self.0.insert(offset, entry).is_some() {
            return Err(Error("Multiple relocations for offset"));
        }
        Ok(())
    }

    /// Relocate a value that was read from the section at the given offset.
    pub fn relocate(&self, offset: u64, value: u64) -> u64 {
        if let Some(relocation) = self.0.get(&offset) {
            if relocation.implicit_addend {
                // Use the explicit addend too, because it may have the symbol value.
                value.wrapping_add(relocation.addend)
            } else {
                relocation.addend
            }
        } else {
            value
        }
    }
}

#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
struct RelocationMapEntry {
    implicit_addend: bool,
    addend: u64,
}

/// A data compression format.
#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
#[non_exhaustive]
pub enum CompressionFormat {
    /// The data is uncompressed.
    None,
    /// The data is compressed, but the compression format is unknown.
    Unknown,
    /// ZLIB/DEFLATE.
    ///
    /// Used for ELF compression and GNU compressed debug information.
    Zlib,
    /// Zstandard.
    ///
    /// Used for ELF compression.
    Zstandard,
}

/// A range in a file that may be compressed.
///
/// Returned by [`ObjectSection::compressed_file_range`].
#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
pub struct CompressedFileRange {
    /// The data compression format.
    pub format: CompressionFormat,
    /// The file offset of the compressed data.
    pub offset: u64,
    /// The compressed data size.
    pub compressed_size: u64,
    /// The uncompressed data size.
    pub uncompressed_size: u64,
}

impl CompressedFileRange {
    /// Data that is uncompressed.
    #[inline]
    pub fn none(range: Option<(u64, u64)>) -> Self {
        if let Some((offset, size)) = range {
            CompressedFileRange {
                format: CompressionFormat::None,
                offset,
                compressed_size: size,
                uncompressed_size: size,
            }
        } else {
            CompressedFileRange {
                format: CompressionFormat::None,
                offset: 0,
                compressed_size: 0,
                uncompressed_size: 0,
            }
        }
    }

    /// Convert to [`CompressedData`] by reading from the file.
    pub fn data<'data, R: ReadRef<'data>>(self, file: R) -> Result<CompressedData<'data>> {
        let data = file
            .read_bytes_at(self.offset, self.compressed_size)
            .read_error("Invalid compressed data size or offset")?;
        Ok(CompressedData {
            format: self.format,
            data,
            uncompressed_size: self.uncompressed_size,
        })
    }
}

/// Data that may be compressed.
///
/// Returned by [`ObjectSection::compressed_data`].
#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
pub struct CompressedData<'data> {
    /// The data compression format.
    pub format: CompressionFormat,
    /// The compressed data.
    pub data: &'data [u8],
    /// The uncompressed data size.
    pub uncompressed_size: u64,
}

impl<'data> CompressedData<'data> {
    /// Data that is uncompressed.
    #[inline]
    pub fn none(data: &'data [u8]) -> Self {
        CompressedData {
            format: CompressionFormat::None,
            data,
            uncompressed_size: data.len() as u64,
        }
    }

    /// Return the uncompressed data.
    ///
    /// Returns an error for invalid data or unsupported compression.
    /// This includes if the data is compressed but the `compression` feature
    /// for this crate is disabled.
    pub fn decompress(self) -> Result<Cow<'data, [u8]>> {
        match self.format {
            CompressionFormat::None => Ok(Cow::Borrowed(self.data)),
            #[cfg(feature = "compression")]
            CompressionFormat::Zlib | CompressionFormat::Zstandard => {
                use core::convert::TryInto;
                use std::io::Read;
                let size = self
                    .uncompressed_size
                    .try_into()
                    .ok()
                    .read_error("Uncompressed data size is too large.")?;
                let mut decompressed = Vec::new();
                decompressed
                    .try_reserve_exact(size)
                    .ok()
                    .read_error("Uncompressed data allocation failed")?;

                match self.format {
                    CompressionFormat::Zlib => {
                        let mut decompress = flate2::Decompress::new(true);
                        decompress
                            .decompress_vec(
                                self.data,
                                &mut decompressed,
                                flate2::FlushDecompress::Finish,
                            )
                            .ok()
                            .read_error("Invalid zlib compressed data")?;
                    }
                    CompressionFormat::Zstandard => {
                        let mut input = self.data;
                        while !input.is_empty() {
                            let mut decoder = match ruzstd::StreamingDecoder::new(&mut input) {
                                Ok(decoder) => decoder,
                                Err(
                                    ruzstd::frame_decoder::FrameDecoderError::ReadFrameHeaderError(
                                        ruzstd::frame::ReadFrameHeaderError::SkipFrame {
                                            length,
                                            ..
                                        },
                                    ),
                                ) => {
                                    input = input
                                        .get(length as usize..)
                                        .read_error("Invalid zstd compressed data")?;
                                    continue;
                                }
                                x => x.ok().read_error("Invalid zstd compressed data")?,
                            };
                            decoder
                                .read_to_end(&mut decompressed)
                                .ok()
                                .read_error("Invalid zstd compressed data")?;
                        }
                    }
                    _ => unreachable!(),
                }
                if size != decompressed.len() {
                    return Err(Error(
                        "Uncompressed data size does not match compression header",
                    ));
                }

                Ok(Cow::Owned(decompressed))
            }
            _ => Err(Error("Unsupported compressed data.")),
        }
    }
}