use object::write::{Object as Writer, Relocation, StandardSection, Symbol, SymbolSection};
use object::{
Architecture, BinaryFormat, Endianness, RelocationFlags, SectionKind, SymbolFlags, SymbolKind,
SymbolScope, elf,
};
use rucc_target::{Arch, Os, TargetInfo};
use crate::section::{Binding, Data, Object, Place, Reference, Reloc, Text};
const ALIGN: u64 = 16;
#[derive(Debug, Clone, PartialEq, Eq)]
pub enum Error {
Format {
triple: String,
},
Refused {
why: String,
},
}
impl std::fmt::Display for Error {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
match self {
Error::Format { triple } => {
write!(f, "there is no object writer for {triple} in this compiler yet")
}
Error::Refused { why } => {
write!(f, "the object writer refused what it was given: {why}")
}
}
}
}
impl std::error::Error for Error {}
pub fn write(text: &Text, data: &Data, target: &TargetInfo) -> Result<Vec<u8>, Error> {
if target.triple.arch != Arch::X86_64 || target.triple.os == Os::Darwin {
return Err(Error::Format { triple: target.triple.to_string() });
}
let mut obj = Writer::new(BinaryFormat::Elf, Architecture::X86_64, Endianness::Little);
let section = obj.section_id(StandardSection::Text);
obj.append_section_data(section, &text.bytes, ALIGN);
let mut symbols = std::collections::BTreeMap::new();
for func in &text.funcs {
let id = obj.add_symbol(Symbol {
name: func.name.clone().into_bytes(),
value: func.start as u64,
size: func.len as u64,
kind: SymbolKind::Text,
scope: SymbolScope::Linkage,
weak: false,
section: SymbolSection::Section(section),
flags: SymbolFlags::None,
});
symbols.insert(func.name.clone(), id);
}
let mut placed = Vec::with_capacity(data.objects.len());
for object in &data.objects {
let (section, offset) = put(&mut obj, object);
let id = obj.add_symbol(Symbol {
name: object.name.clone().into_bytes(),
value: if object.place == Place::Merged { object.align } else { offset },
size: object.size,
kind: SymbolKind::Data,
scope: scope_of(object.binding),
weak: object.binding == Binding::Weak,
section,
flags: SymbolFlags::None,
});
symbols.insert(object.name.clone(), id);
placed.push((section.id(), offset));
}
let wanted = text.relocs.iter().chain(data.objects.iter().flat_map(|object| &object.relocs));
for reloc in wanted {
if symbols.contains_key(&reloc.symbol) {
continue;
}
let id = obj.add_symbol(Symbol {
name: reloc.symbol.clone().into_bytes(),
value: 0,
size: 0,
kind: SymbolKind::Unknown,
scope: SymbolScope::Dynamic,
weak: false,
section: SymbolSection::Undefined,
flags: SymbolFlags::None,
});
symbols.insert(reloc.symbol.clone(), id);
}
for reloc in &text.relocs {
add(&mut obj, section, 0, reloc, &symbols)?;
}
for (object, &(section, offset)) in data.objects.iter().zip(&placed) {
let Some(section) = section else { continue };
for reloc in &object.relocs {
add(&mut obj, section, offset, reloc, &symbols)?;
}
}
obj.add_section(Vec::new(), b".note.GNU-stack".to_vec(), SectionKind::Metadata);
obj.write().map_err(|why| Error::Refused { why: why.to_string() })
}
fn put(obj: &mut Writer<'_>, object: &Object) -> (SymbolSection, u64) {
let section = match &object.place {
Place::Written => obj.section_id(StandardSection::Data),
Place::ReadOnly => obj.section_id(StandardSection::ReadOnlyData),
Place::Zero => obj.section_id(StandardSection::UninitializedData),
Place::Merged => return (SymbolSection::Common, 0),
Place::Named(name) => {
obj.add_section(Vec::new(), name.clone().into_bytes(), SectionKind::Data)
}
};
let offset = if object.place == Place::Zero {
obj.append_section_bss(section, object.size, object.align)
} else {
obj.append_section_data(section, &object.bytes, object.align)
};
(SymbolSection::Section(section), offset)
}
fn add(
obj: &mut Writer<'_>,
section: object::write::SectionId,
offset: u64,
reloc: &Reloc,
symbols: &std::collections::BTreeMap<String, object::write::SymbolId>,
) -> Result<(), Error> {
let r_type = r_type(reloc.kind)
.ok_or_else(|| Error::Refused { why: format!("no relocation is {:?}", reloc.kind) })?;
obj.add_relocation(
section,
Relocation {
offset: offset + reloc.at as u64,
symbol: symbols[&reloc.symbol],
addend: reloc.addend,
flags: RelocationFlags::Elf { r_type },
},
)
.map_err(|why| Error::Refused { why: why.to_string() })
}
fn scope_of(binding: Binding) -> SymbolScope {
match binding {
Binding::Local => SymbolScope::Compilation,
Binding::Global | Binding::Weak => SymbolScope::Linkage,
}
}
fn r_type(reference: Reference) -> Option<elf::RelocationType> {
Some(match reference {
Reference::Call => elf::R_X86_64_PLT32,
Reference::Data => elf::R_X86_64_PC32,
Reference::Address { bytes: 8 } => elf::R_X86_64_64,
Reference::Address { bytes: 4 } => elf::R_X86_64_32,
Reference::Address { .. } => return None,
})
}
#[cfg(test)]
mod tests {
use super::*;
use object::read::elf::Sym as _;
use object::read::{Object as _, ObjectSection as _, ObjectSymbol as _};
use rucc_target::{Env, Triple};
use crate::section::{Extent, Reloc};
fn target() -> TargetInfo {
TargetInfo::new(Triple::new(Arch::X86_64, Os::Linux, Env::Gnu))
}
fn calling(name: &str) -> Text {
Text {
bytes: vec![0xe8, 0, 0, 0, 0, 0xc3],
funcs: vec![Extent { name: "f".to_owned(), start: 0, len: 6 }],
relocs: vec![Reloc {
at: 1,
symbol: name.to_owned(),
kind: Reference::Call,
addend: -4,
}],
}
}
#[test]
fn the_bytes_come_back_out_of_the_section_they_went_into() {
let text = calling("puts");
let bytes = write(&text, &Data::default(), &target()).expect("an object");
let file = object::File::parse(&bytes[..]).expect("a readable object");
let section = file.section_by_name(".text").expect("a text section");
assert_eq!(section.data().expect("the bytes"), &text.bytes[..]);
}
#[test]
fn a_function_is_a_symbol_that_says_where_it_is_and_how_long_it_is() {
let mut text = calling("puts");
text.funcs.push(Extent { name: "g".to_owned(), start: 16, len: 1 });
text.bytes.resize(17, 0x90);
let bytes = write(&text, &Data::default(), &target()).expect("an object");
let file = object::File::parse(&bytes[..]).expect("a readable object");
let g = file.symbols().find(|s| s.name() == Ok("g")).expect("the second function");
assert_eq!(g.address(), 16);
assert_eq!(g.size(), 1);
assert_eq!(g.kind(), SymbolKind::Text);
assert!(g.is_global(), "a function is global until the machine IR can say otherwise");
}
#[test]
fn a_name_this_file_does_not_define_is_left_for_the_linker_to_find() {
let bytes = write(&calling("puts"), &Data::default(), &target()).expect("an object");
let file = object::File::parse(&bytes[..]).expect("a readable object");
let puts = file.symbols().find(|s| s.name() == Ok("puts")).expect("the callee");
assert!(puts.is_undefined(), "the file does not define it and must not claim to");
}
#[test]
fn a_call_asks_for_the_relocation_a_stub_may_answer_and_a_load_asks_for_the_one_that_may_not() {
for (reference, wanted) in
[(Reference::Call, elf::R_X86_64_PLT32), (Reference::Data, elf::R_X86_64_PC32)]
{
let mut text = calling("puts");
text.relocs[0].kind = reference;
let bytes = write(&text, &Data::default(), &target()).expect("an object");
let file = object::File::parse(&bytes[..]).expect("a readable object");
let section = file.section_by_name(".text").expect("a text section");
let (offset, reloc) = section.relocations().next().expect("one relocation");
assert_eq!(offset, 1);
assert_eq!(reloc.addend(), -4);
assert_eq!(reloc.flags(), RelocationFlags::Elf { r_type: wanted });
}
}
#[test]
fn a_name_wanted_twice_is_one_symbol_rather_than_two() {
let mut text = calling("puts");
text.relocs.push(Reloc {
at: 1,
symbol: "puts".to_owned(),
kind: Reference::Call,
addend: -4,
});
let bytes = write(&text, &Data::default(), &target()).expect("an object");
let file = object::File::parse(&bytes[..]).expect("a readable object");
assert_eq!(file.symbols().filter(|s| s.name() == Ok("puts")).count(), 1);
}
#[test]
fn a_function_that_is_also_called_is_not_a_second_symbol() {
let text = calling("f");
let bytes = write(&text, &Data::default(), &target()).expect("an object");
let file = object::File::parse(&bytes[..]).expect("a readable object");
let mut found = file.symbols().filter(|s| s.name() == Ok("f"));
let f = found.next().expect("the function");
assert!(!f.is_undefined(), "the file defines it");
assert!(found.next().is_none(), "and defines it once");
}
#[test]
fn the_marker_that_says_the_stack_is_not_executable_is_written() {
let bytes = write(&calling("puts"), &Data::default(), &target()).expect("an object");
let file = object::File::parse(&bytes[..]).expect("a readable object");
let note = file.section_by_name(".note.GNU-stack").expect("the marker");
assert!(note.data().expect("no bytes").is_empty());
}
fn variable(name: &str, place: Place) -> Object {
Object {
name: name.to_owned(),
bytes: if place == Place::Zero { Vec::new() } else { vec![1, 0, 0, 0] },
size: 4,
align: 4,
place,
binding: Binding::Global,
relocs: Vec::new(),
}
}
fn holding(object: Object) -> Vec<u8> {
let data = Data { objects: vec![object] };
write(&Text::default(), &data, &target()).expect("an object")
}
#[test]
fn what_a_variable_is_decides_which_section_it_goes_in() {
for (place, wanted) in [
(Place::Written, ".data"),
(Place::ReadOnly, ".rodata"),
(Place::Zero, ".bss"),
(Place::Named(".init_array".to_owned()), ".init_array"),
] {
let bytes = holding(variable("x", place.clone()));
let file = object::File::parse(&bytes[..]).expect("a readable object");
let section = file.section_by_name(wanted).unwrap_or_else(|| panic!("{place:?}"));
assert_eq!(section.size(), 4, "{place:?}");
let carried = section.data().expect("the bytes").len();
assert_eq!(carried, if place == Place::Zero { 0 } else { 4 }, "{place:?}");
}
}
#[test]
fn a_variable_is_a_symbol_that_says_where_it_is_and_how_long_it_is() {
let mut data = Data { objects: vec![variable("first", Place::Written)] };
data.objects.push(Object { align: 16, ..variable("second", Place::Written) });
let bytes = write(&Text::default(), &data, &target()).expect("an object");
let file = object::File::parse(&bytes[..]).expect("a readable object");
let second = file.symbols().find(|s| s.name() == Ok("second")).expect("the second one");
assert_eq!(second.kind(), SymbolKind::Data);
assert_eq!(second.size(), 4);
assert_eq!(second.address(), 16);
}
#[test]
fn the_linkage_a_variable_had_is_the_binding_the_symbol_gets() {
for (binding, global, weak) in [
(Binding::Global, true, false),
(Binding::Local, false, false),
(Binding::Weak, true, true),
] {
let bytes = holding(Object { binding, ..variable("x", Place::Written) });
let file = object::File::parse(&bytes[..]).expect("a readable object");
let x = file.symbols().find(|s| s.name() == Ok("x")).expect("the variable");
assert_eq!(x.is_global(), global, "{binding:?}");
assert_eq!(x.is_weak(), weak, "{binding:?}");
}
}
#[test]
fn a_tentative_definition_asks_the_linker_for_space_rather_than_naming_any() {
let bytes = holding(Object { align: 8, ..variable("x", Place::Merged) });
let file = object::read::elf::ElfFile64::<Endianness>::parse(&bytes[..]).expect("readable");
let x = file.symbols().find(|s| s.name() == Ok("x")).expect("the variable");
assert!(x.is_common(), "the linker merges every definition of this name into one");
assert_eq!(x.size(), 4);
assert_eq!(x.address(), 0);
assert_eq!(x.elf_symbol().st_value(Endianness::Little), 8);
}
#[test]
fn an_address_in_an_image_is_the_address_and_not_a_distance_to_it() {
let object = Object {
bytes: vec![0; 8],
size: 8,
align: 8,
relocs: vec![Reloc {
at: 0,
symbol: "y".to_owned(),
kind: Reference::Address { bytes: 8 },
addend: 16,
}],
..variable("p", Place::Written)
};
let bytes = holding(object);
let file = object::File::parse(&bytes[..]).expect("a readable object");
let section = file.section_by_name(".data").expect("a data section");
let (offset, reloc) = section.relocations().next().expect("one relocation");
assert_eq!(offset, 0);
assert_eq!(reloc.addend(), 16);
assert_eq!(reloc.flags(), RelocationFlags::Elf { r_type: elf::R_X86_64_64 });
let y = file.symbols().find(|s| s.name() == Ok("y")).expect("what it points at");
assert!(y.is_undefined(), "nothing here defines it and the linker is being asked for it");
}
#[test]
fn a_relocation_counts_from_the_start_of_the_section_and_not_of_the_image_it_is_in() {
let mut data = Data { objects: vec![variable("first", Place::Written)] };
data.objects.push(Object {
bytes: vec![0; 16],
size: 16,
align: 8,
relocs: vec![Reloc {
at: 8,
symbol: "y".to_owned(),
kind: Reference::Address { bytes: 8 },
addend: 0,
}],
..variable("second", Place::Written)
});
let bytes = write(&Text::default(), &data, &target()).expect("an object");
let file = object::File::parse(&bytes[..]).expect("a readable object");
let section = file.section_by_name(".data").expect("a data section");
let (offset, _) = section.relocations().next().expect("one relocation");
assert_eq!(offset, 16);
}
#[test]
fn a_platform_this_does_not_write_is_said_so_rather_than_written_as_elf() {
let text = calling("puts");
for triple in [
Triple::new(Arch::Aarch64, Os::Linux, Env::Gnu),
Triple::new(Arch::X86_64, Os::Darwin, Env::Gnu),
] {
let error =
write(&text, &Data::default(), &TargetInfo::new(triple)).expect_err("no writer");
assert!(matches!(error, Error::Format { .. }), "{error:?}");
}
}
}