initial commit

This commit is contained in:
2024-02-08 17:34:21 +01:00
commit 7a8554d313
8 changed files with 1129 additions and 0 deletions

143
src/ints.rs Normal file
View File

@@ -0,0 +1,143 @@
use bytemuck::{Pod, Zeroable};
use core::fmt::Debug;
use core::marker::PhantomData;
use serde::Serialize;
#[derive(Pod, Zeroable, Clone, Copy, Debug, Serialize)]
#[repr(C, packed)]
pub struct BigEndian;
#[derive(Pod, Zeroable, Clone, Copy, Debug, Serialize)]
#[repr(C, packed)]
pub struct LittleEndian;
pub trait Decode: Copy {
type Native: Debug + Copy;
fn to_native(self) -> Self::Native;
}
#[derive(Pod, Zeroable, Clone, Copy)]
#[repr(C, packed)]
pub struct U64<E> {
raw: [u8; 8],
_endian: PhantomData<E>,
}
#[derive(Pod, Zeroable, Clone, Copy)]
#[repr(C, packed)]
pub struct U32<E> {
raw: [u8; 4],
_endian: PhantomData<E>,
}
#[derive(Pod, Zeroable, Clone, Copy)]
#[repr(C, packed)]
pub struct U16<E> {
raw: [u8; 2],
_endian: PhantomData<E>,
}
impl Decode for U64<BigEndian> {
type Native = u64;
fn to_native(self) -> Self::Native {
u64::from_be_bytes(self.raw)
}
}
impl Decode for U64<LittleEndian> {
type Native = u64;
fn to_native(self) -> Self::Native {
u64::from_le_bytes(self.raw)
}
}
impl Decode for U32<BigEndian> {
type Native = u32;
fn to_native(self) -> Self::Native {
u32::from_be_bytes(self.raw)
}
}
impl Decode for U32<LittleEndian> {
type Native = u32;
fn to_native(self) -> Self::Native {
u32::from_le_bytes(self.raw)
}
}
impl Decode for U16<BigEndian> {
type Native = u16;
fn to_native(self) -> Self::Native {
u16::from_be_bytes(self.raw)
}
}
impl Decode for U16<LittleEndian> {
type Native = u16;
fn to_native(self) -> Self::Native {
u16::from_le_bytes(self.raw)
}
}
impl<E> Debug for U64<E>
where
Self: Decode,
{
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
Debug::fmt(&self.to_native(), f)
}
}
impl<E> Debug for U32<E>
where
Self: Decode,
{
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
Debug::fmt(&self.to_native(), f)
}
}
impl<E> Debug for U16<E>
where
Self: Decode,
{
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
Debug::fmt(&self.to_native(), f)
}
}
impl<E> Serialize for U64<E>
where
Self: Decode,
{
fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
where
S: serde::Serializer,
{
format!("{:?}", self.to_native()).serialize(serializer)
}
}
impl<E> Serialize for U32<E>
where
Self: Decode,
{
fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
where
S: serde::Serializer,
{
format!("{:?}", self.to_native()).serialize(serializer)
}
}
impl<E> Serialize for U16<E>
where
Self: Decode,
{
fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
where
S: serde::Serializer,
{
format!("{:?}", self.to_native()).serialize(serializer)
}
}

208
src/main.rs Normal file
View File

@@ -0,0 +1,208 @@
mod ints;
mod output;
mod structs;
use crate::{
ints::BigEndian,
output::Output,
structs::{ElfHeader1, ElfHeader2, SectionHeaderEntry, SymbolTableEntry},
};
use bytemuck::from_bytes;
use clap::{Parser, Subcommand};
use core::str::from_utf8;
use eyre::{bail, eyre, Context};
use ints::{Decode, LittleEndian, U32, U64};
use serde_json::Value;
use std::{fs, mem::size_of, path::PathBuf};
#[derive(Parser)]
struct Opt {
elf: PathBuf,
#[clap(subcommand)]
command: Command,
}
#[derive(Subcommand)]
enum Command {
/// List sections
Sections {},
/// List symbols
Symbols {
#[clap(short, long)]
section: String,
},
}
const HEADER1_LEN: usize = size_of::<ElfHeader1>();
macro_rules! parse_elf {
($endian:ty, $word:ty, $out:expr, $header1:expr, $elf:expr) => {{
//let header1 = $header1;
let out = $out;
const HEADER2_LEN: usize = size_of::<ElfHeader2<$endian, $word>>();
let elf = $elf;
if elf.len() < HEADER1_LEN + HEADER2_LEN {
bail!("not a valid elf file: file too small");
}
let header2: &[u8; HEADER2_LEN] = elf[HEADER1_LEN..][..HEADER2_LEN].try_into()?;
let header2: ElfHeader2<$endian, $word> = *from_bytes(header2);
out.write("header2", &header2)?;
let e_shentsize = usize::from(header2.e_shentsize.to_native());
if e_shentsize != size_of::<SectionHeaderEntry<$endian, $word>>() {
bail!("wrong e_shentsize (0x{e_shentsize:x})");
}
let e_shnum = usize::from(header2.e_shnum.to_native());
let section_header_table_size = e_shentsize * e_shnum;
let section_header_table_start =
usize::try_from(header2.e_shoff.to_native()).wrap_err("e_shoff bigger than usize")?;
let section_header_table_end = section_header_table_start + section_header_table_size;
if elf.len() < section_header_table_end {
bail!("not a valid elf file: section table goes past eof");
} else if elf.len() > section_header_table_end {
bail!("not a valid elf file: section table doesn't end at eof");
}
let mut section_header_entries = vec![];
for entry_start in
(section_header_table_start..section_header_table_end).step_by(e_shentsize)
{
let entry = &elf[entry_start..][..e_shentsize];
let entry: SectionHeaderEntry<$endian, $word> = *from_bytes(entry);
section_header_entries.push(entry);
}
out.write("section_table", &section_header_entries)?;
let names_section_i = usize::from(header2.e_shstrndx.to_native());
let names_section_entry = &section_header_entries
.get(names_section_i)
.ok_or(eyre!("invalid e_shstrndx"))?;
let names_section_start: usize = (names_section_entry.sh_offset.to_native().try_into())
.wrap_err("sh_offset bigger than usize")?;
let names_section_size: usize = (names_section_entry.sh_size.to_native().try_into())
.wrap_err("sh_size bigger than usize")?;
let names_section_end = names_section_start + names_section_size;
let names_section = &elf[names_section_start..names_section_end];
let mut symtab_i = None;
let mut strtab = None;
let mut section_header_entries_with_names = vec![];
for (i, entry) in section_header_entries.iter().enumerate() {
let name_start: usize = (entry.sh_name.to_native())
.try_into()
.wrap_err("sh_name bigger than usize")?;
if name_start > names_section.len() {
bail!("sh_name out of bounds for .names section");
}
let name_end = (name_start..names_section.len())
.find(|&i| names_section[i] == b'\0')
.ok_or(eyre!(".names entry missing null byte"))?;
let name = &names_section[name_start..name_end];
let name = from_utf8(name).wrap_err(eyre!(".names entry not valid utf-8"))?;
let section_start = entry.start().wrap_err_with(|| name.to_string())?;
let section_end = entry.end().wrap_err_with(|| name.to_string())?;
if section_end > elf.len() {
bail!("section {name:?} end is past eof");
}
match name {
".symtab" => symtab_i = Some(i),
".strtab" => {
strtab = Some(&elf[section_start..section_end]);
}
_ => {}
}
let mut entry_json =
serde_json::to_value(entry).wrap_err("failed to serialize section header")?;
let Value::Object(fields) = &mut entry_json else {
unreachable!()
};
fields.insert("name".to_string(), Value::String(name.to_string()));
section_header_entries_with_names.push(entry_json);
}
let _ = out.write("section_table", &section_header_entries_with_names);
if let Some(symtab_i) = symtab_i {
let symtab = &section_header_entries[symtab_i];
const ENTRY_SIZE: usize = size_of::<SymbolTableEntry<$word>>();
if symtab.sh_size.to_native() as usize != ENTRY_SIZE {
bail!(".symtab size not a multiple of symtab entry size");
}
let section_start: usize = symtab.start().wrap_err(".symtab")?;
let section_end: usize = symtab.end().wrap_err(".symtab")?;
for entry in elf[section_start..section_end].chunks_exact(ENTRY_SIZE) {
let entry: &[u8; ENTRY_SIZE] = entry.try_into().unwrap();
let entry: &SymbolTableEntry<$word> = from_bytes(entry);
let name_start: usize = (entry.st_name.to_native().try_into())
.wrap_err("symtab entry st_name too big to fit in usize")?;
let name = (name_start != 0)
.then(|| {
strtab.map(|strtab| {
eprintln!("{name_start} {}", strtab.len());
let name_end = (name_start..strtab.len())
.find(|&i| strtab[i] == b'\0')
.ok_or(eyre!(".strtab entry missing null byte"))?;
let name = &strtab[name_start..name_end];
from_utf8(name).wrap_err(eyre!(
".strtab entry at 0x{name_start:x} is not valid utf-8"
))
})
})
// Option<Option<Result>> to Result<Option>
.flatten()
.transpose()?;
eprintln!("[.symtab] {name:?}: {entry:?}");
}
}
}};
}
fn main() -> eyre::Result<()> {
let opt = Opt::parse();
color_eyre::install()?;
let elf = fs::read(opt.elf).wrap_err("failed to read elf file")?;
if elf.len() < HEADER1_LEN {
bail!("not a valid elf file: file too small");
}
let header1: &[u8; HEADER1_LEN] = elf[..HEADER1_LEN].try_into()?;
let header1: ElfHeader1 = *from_bytes(header1);
let mut out = Output::default();
out.write("header1", &header1)?;
if &header1.ei_magic != b"\x7FELF" {
bail!("not a valid elf file: invalid magic");
}
if header1.ei_version != 1 {
bail!("unknown elf version: 0x{:x}", header1.ei_version);
}
match (header1.ei_class, header1.ei_data) {
(1, 1) => parse_elf!(LittleEndian, U32<LittleEndian>, &mut out, header1, elf),
(2, 1) => parse_elf!(LittleEndian, U64<LittleEndian>, &mut out, header1, elf),
(1, 2) => parse_elf!(BigEndian, U32<BigEndian>, &mut out, header1, elf),
(2, 2) => parse_elf!(BigEndian, U64<BigEndian>, &mut out, header1, elf),
(1 | 2, ei_data) => bail!("unknown e_ident[EI_DATA]: 0x{ei_data:x}"),
(ei_class, ..) => bail!("unknown e_ident[EI_CLASS]: 0x{ei_class:x}"),
}
Ok(())
}

28
src/output.rs Normal file
View File

@@ -0,0 +1,28 @@
use std::collections::BTreeMap;
use eyre::{bail, Context};
use serde::Serialize;
use serde_json::Value;
#[derive(Default)]
pub struct Output {
elements: BTreeMap<&'static str, Value>,
}
impl Output {
pub fn write(&mut self, key: &'static str, v: &impl Serialize) -> eyre::Result<()> {
let v = serde_json::to_value(v).wrap_err("failed to serialize output")?;
if self.elements.insert(key, v).is_some() {
bail!("duplicated output key {key:?}");
}
Ok(())
}
}
impl Drop for Output {
fn drop(&mut self) {
let out =
serde_json::to_string_pretty(&self.elements).expect("can serialize output as json");
println!("{out}",);
}
}

96
src/structs.rs Normal file
View File

@@ -0,0 +1,96 @@
use crate::ints::{Decode, U16, U32};
use bytemuck::{Pod, Zeroable};
use eyre::eyre;
use serde::Serialize;
#[derive(Pod, Zeroable, Clone, Copy, Debug, Serialize)]
#[repr(C, packed)]
pub struct ElfHeader1 {
pub ei_magic: [u8; 4],
pub ei_class: u8,
pub ei_data: u8,
pub ei_version: u8,
pub ei_osabi: u8,
pub ei_abiversion: u8,
#[serde(skip)]
_padding: [u8; 7],
}
#[derive(Pod, Zeroable, Clone, Copy, Debug, Serialize)]
#[repr(C, packed)]
pub struct ElfHeader2<Endian, Word>
where
U16<Endian>: Decode,
U32<Endian>: Decode,
Word: Decode,
{
pub e_type: U16<Endian>,
pub e_machine: U16<Endian>,
pub e_version: U32<Endian>,
pub e_entry: Word,
pub e_phoff: Word,
pub e_shoff: Word,
pub e_flags: U32<Endian>,
pub e_ehsize: U16<Endian>,
pub e_phentsize: U16<Endian>,
pub e_phnum: U16<Endian>,
pub e_shentsize: U16<Endian>,
pub e_shnum: U16<Endian>,
pub e_shstrndx: U16<Endian>,
}
#[derive(Pod, Zeroable, Clone, Copy, Debug, Serialize)]
#[repr(C, packed)]
pub struct SectionHeaderEntry<Endian, Word>
where
U16<Endian>: Decode,
U32<Endian>: Decode,
Word: Decode,
{
pub sh_name: U32<Endian>,
pub sh_type: U32<Endian>,
pub sh_flags: Word,
pub sh_addr: Word,
pub sh_offset: Word,
pub sh_size: Word,
pub sh_link: U32<Endian>,
pub sh_info: U32<Endian>,
pub sh_addralign: Word,
pub sh_entsize: Word,
}
#[derive(Pod, Zeroable, Clone, Copy, Debug, Serialize)]
#[repr(C, packed)]
pub struct SymbolTableEntry<Word: Decode> {
pub st_name: Word,
pub st_value: Word,
pub st_size: Word,
pub st_info: u8,
pub st_other: u8,
pub st_shndx: Word,
}
impl<Endian, Word> SectionHeaderEntry<Endian, Word>
where
U16<Endian>: Decode,
U32<Endian>: Decode,
Word: Decode,
<Word as Decode>::Native: TryInto<usize>,
{
pub fn start(&self) -> eyre::Result<usize> {
self.sh_offset
.to_native()
.try_into()
.map_err(|_| eyre!("sh_offset bigger than usize"))
}
pub fn end(&self) -> eyre::Result<usize> {
let size: usize = self
.sh_size
.to_native()
.try_into()
.map_err(|_| eyre!("sh_size bigger than usize"))?;
Ok(self.start()? + size)
}
}