This commit is contained in:
2026-06-07 21:22:32 -04:00
parent a086fa6590
commit c9add923be
14 changed files with 413 additions and 46 deletions
+21 -9
View File
@@ -9,29 +9,35 @@ pub struct Asm {
#[derive(Clone, Copy)]
pub enum Instr {
Mov { dst: RegMode, src: RegImm },
Mov { dst: RegMode, src: RegModeImm },
Int(u8),
Call(Symbol),
Ret,
Syscall,
Lea { dst: RegMode, sym: Symbol },
Push(Reg),
Push(RegImm),
Pop(Reg),
}
#[derive(Clone, Copy)]
pub enum RegImm {
pub enum RegModeImm {
Reg(RegMode),
Imm(u64),
}
impl From<RegMode> for RegImm {
#[derive(Clone, Copy)]
pub enum RegImm {
Reg(Reg),
Imm(u64),
}
impl From<RegMode> for RegModeImm {
fn from(value: RegMode) -> Self {
Self::Reg(value)
}
}
impl From<u64> for RegImm {
impl From<u64> for RegModeImm {
fn from(value: u64) -> Self {
Self::Imm(value)
}
@@ -39,7 +45,7 @@ impl From<u64> for RegImm {
mod fns {
use super::*;
pub fn mov(dst: RegMode, src: impl Into<RegImm>) -> Instr {
pub fn mov(dst: RegMode, src: impl Into<RegModeImm>) -> Instr {
Instr::Mov {
dst,
src: src.into(),
@@ -50,9 +56,15 @@ mod fns {
Instr::Lea { dst, sym }
}
pub fn push(reg: RegMode) -> Instr {
assert!(reg.width == BitWidth::B64);
Instr::Push(reg.reg)
pub fn push(reg: impl Into<RegModeImm>) -> Instr {
Instr::Push(match reg.into() {
RegModeImm::Reg(reg) => {
assert_eq!(reg.width, BitWidth::B64);
assert!(!reg.high);
RegImm::Reg(reg.reg)
}
RegModeImm::Imm(imm) => RegImm::Imm(imm),
})
}
pub fn pop(reg: RegMode) -> Instr {
+19 -5
View File
@@ -50,7 +50,7 @@ fn compile_instr(encoder: &mut Encoder, instr: &BInstr) -> Result<(), CompilerMs
impl Encoder {
// assembly
pub fn mov(&mut self, dst: RegMode, src: impl Into<RegImm>) -> Result<(), CompilerMsg> {
pub fn mov(&mut self, dst: RegMode, src: impl Into<RegModeImm>) -> Result<(), CompilerMsg> {
let src = src.into();
let width = dst.width;
if width == BitWidth::B16 {
@@ -59,7 +59,7 @@ impl Encoder {
let dst8 = dst.gt8();
let b64 = width == BitWidth::B64;
let b8 = width == BitWidth::B8;
let src8 = if let RegImm::Reg(src) = src {
let src8 = if let RegModeImm::Reg(src) = src {
src.gt8()
} else {
false
@@ -70,7 +70,7 @@ impl Encoder {
.push(0x40 | dst8 as u8 | ((b64 as u8) << 3) | ((src8 as u8) << 2));
}
match src {
RegImm::Reg(src) => {
RegModeImm::Reg(src) => {
if dst.width != src.width {
return Err("src and dst are not the same size".into());
}
@@ -78,7 +78,7 @@ impl Encoder {
let modrm = 0b11_000_000 | (src.base() << 3) | dst.base();
self.data.push(modrm);
}
RegImm::Imm(imm) => {
RegModeImm::Imm(imm) => {
if imm > width.max() {
return Err("immediate cannot fit in register".into());
}
@@ -122,6 +122,17 @@ impl Encoder {
self.data.push(0x50 | reg.base());
}
pub fn push_imm(&mut self, imm: u64) {
const U8: u64 = 2 << 8;
if let 0..U8 = imm {
self.data.push(0x6a);
self.data.push(imm as u8);
} else {
self.data.push(0x68);
self.data.extend(imm.to_le_bytes());
}
}
pub fn pop(&mut self, reg: Reg) {
if reg.gt8() {
self.data.push(0x41);
@@ -148,7 +159,10 @@ impl Encoder {
Instr::Lea { dst, sym } => self.lea(dst, sym),
Instr::Call(sym) => self.call(sym),
Instr::Ret => self.ret(),
Instr::Push(reg) => self.push(reg),
Instr::Push(val) => match val {
RegImm::Reg(reg) => self.push(reg),
RegImm::Imm(imm) => self.push_imm(imm),
},
Instr::Pop(reg) => self.pop(reg),
}
Ok(())
+1 -1
View File
@@ -8,7 +8,7 @@ pub struct RegMode {
pub high: bool,
}
#[derive(Clone, Copy, PartialEq)]
#[derive(Debug, Clone, Copy, PartialEq)]
pub enum BitWidth {
B64,
B32,
+38 -10
View File
@@ -5,6 +5,11 @@ use crate::{
use std::{fs::OpenOptions, io::Write, os::unix::fs::OpenOptionsExt, process::Command};
pub fn run() {
windows();
// linux();
}
fn linux() {
let mut program = Program::<X86_64>::default();
let text = b"Hello world!\n";
let text_sym = program.ro_data(text);
@@ -39,16 +44,7 @@ pub fn run() {
let linked = program.compile().expect("failed to compile");
let binary = linked.to_elf();
let path = "./x86_64_test";
let mut file = OpenOptions::new()
.create(true)
.write(true)
.truncate(true)
.mode(0o750)
.open(path)
.expect("Failed to create file");
file.write_all(&binary).expect("Failed to write to file");
file.sync_all().expect("Failed to sync file");
drop(file);
write(path, &binary);
println!("running...");
let gdb = false;
let mut proc = if gdb {
@@ -72,3 +68,35 @@ pub fn run() {
std::process::exit(code);
}
}
fn windows() {
let mut program = Program::<X86_64>::default();
let entry = program.func([BInstr::Asm(Asm {
instrs: vec![push(39), pop(rax), Instr::Ret],
})]);
program.entry = Some(entry);
let linked = program.compile().expect("failed to compile");
let binary = linked.to_pe();
let path = "./x86_64_test.exe";
write(path, &binary);
let mut cmd = Command::new("wine");
cmd.arg("x86_64_test");
let mut proc = cmd.spawn().expect("failed to run");
let status = proc.wait().expect("failed to wait");
if let Some(code) = status.code() {
std::process::exit(code);
}
}
fn write(path: &str, binary: &[u8]) {
let mut file = OpenOptions::new()
.create(true)
.write(true)
.truncate(true)
.mode(0o750)
.open(path)
.expect("Failed to create file");
file.write_all(binary).expect("Failed to write to file");
file.sync_all().expect("Failed to sync file");
}
@@ -1,4 +1,4 @@
use crate::backend::Addr;
use crate::backend::{Addr, LinkedProgram, container::as_u8_slice};
#[repr(C)]
pub struct ELF64Header {
@@ -73,7 +73,7 @@ pub enum EType {
Core = 4,
}
// this is currently specialized for riscv64; obviously add params later
// this is currently specialized for x86_64; obviously add params later
pub fn create(program: &[u8], start_offset: u64) -> Vec<u8> {
let pie = true;
let addr_start = if pie { 0 } else { 0x400000 };
@@ -116,14 +116,14 @@ pub fn create(program: &[u8], start_offset: u64) -> Vec<u8> {
section_header_str_idx: 0x0,
};
let mut bytes: Vec<u8> = Vec::new();
unsafe {
bytes.extend(as_u8_slice(&header));
bytes.extend(as_u8_slice(&program_header));
}
bytes.extend(as_u8_slice(&header));
bytes.extend(as_u8_slice(&program_header));
bytes.extend(program);
bytes
}
unsafe fn as_u8_slice<T: Sized>(p: &T) -> &[u8] {
unsafe { core::slice::from_raw_parts((p as *const T) as *const u8, size_of::<T>()) }
impl LinkedProgram<u64> {
pub fn to_elf(&self) -> Vec<u8> {
create(&self.code, self.entry.expect("no start"))
}
}
+38
View File
@@ -0,0 +1,38 @@
pub mod elf;
pub mod pe;
#[derive(Default)]
pub struct ByteEncoder {
pub data: Vec<u8>,
}
trait Pushable {
fn push(self, data: &mut Vec<u8>);
}
impl ByteEncoder {
pub fn push(&mut self, byte: u8) {
self.data.push(byte);
}
pub fn extend(&mut self, iter: impl IntoIterator<Item = u8>) {
self.data.extend(iter);
}
pub fn val<T>(&mut self, val: &T) {
self.data.extend(as_u8_slice(val));
}
pub fn fill<T>(&mut self, pos: usize, val: &T) {
self.data[pos..pos + size_of::<T>()].copy_from_slice(as_u8_slice(val));
}
pub fn pos(&self) -> usize {
self.data.len()
}
pub fn reserve<T: Default>(&mut self) -> usize {
let pos = self.pos();
self.val(&T::default());
pos
}
}
fn as_u8_slice<T: Sized>(p: &T) -> &[u8] {
unsafe { core::slice::from_raw_parts((p as *const T) as *const u8, size_of::<T>()) }
}
+27
View File
@@ -0,0 +1,27 @@
#[derive(Default)]
#[repr(C)]
pub struct DataDirs {
pub export: DataDir,
pub import: DataDir,
pub rsc: DataDir,
pub exception: DataDir,
pub cert: DataDir,
pub base_reloc: DataDir,
pub debug: DataDir,
pub arch: DataDir,
pub global_ptr: DataDir,
pub tls: DataDir,
pub load_config: DataDir,
pub bound_import: DataDir,
pub import_addr: DataDir,
pub delay_import_desc: DataDir,
pub clr_runtime_header: DataDir,
pub reserved: DataDir,
}
#[derive(Default)]
#[repr(C)]
pub struct DataDir {
pub virt_addr_rva: u32,
pub size: u32,
}
+67
View File
@@ -0,0 +1,67 @@
#[repr(C)]
pub struct MZHeader {
pub magic: u16,
pub stuff: [u16; 15 + 4 + 10],
pub lfanew: u32,
}
#[repr(C)]
pub struct PeHeader {
pub magic: u32,
pub machine: u16,
pub num_sections: u16,
pub time_date_stamp: u32,
pub sym_tab_ptr: u32,
pub num_symbols: u32,
pub opt_header_size: u16,
pub characteristics: u16,
}
#[repr(C)]
#[derive(Default)]
pub struct OptHeader64 {
pub magic: u16,
pub major_linker_ver: u8,
pub minor_linker_ver: u8,
pub code_size: u32,
pub init_data_size: u32,
pub uninit_data_size: u32,
pub entry_addr: u32,
pub code_base: u32,
pub image_base: u64,
pub section_align: u32,
pub file_align: u32,
pub major_os_ver: u16,
pub minor_os_ver: u16,
pub major_image_ver: u16,
pub minor_image_ver: u16,
pub major_subsystem_ver: u16,
pub minor_subsystem_ver: u16,
pub win32_ver: u32,
pub image_size: u32,
pub headers_size: u32,
pub checksum: u32,
pub subsystem: u16,
pub dll_characteristics: u16,
pub stack_reserve_size: u64,
pub stack_commit_size: u64,
pub heap_reserve_size: u64,
pub heap_commit_size: u64,
pub loader_flags: u32,
pub num_of_rva_and_sizes: u32,
}
#[repr(C)]
#[derive(Default)]
pub struct Section {
pub name: [u8; 8],
pub virtual_size: u32,
pub virtual_addr: u32,
pub raw_data_size: u32,
pub raw_data_ptr: u32,
pub reloc_ptr: u32,
pub line_num_ptr: u32,
pub num_relocs: u16,
pub num_line_nums: u16,
pub characteristics: u32,
}
+77
View File
@@ -0,0 +1,77 @@
use crate::backend::ByteEncoder;
pub struct Import {
pub name: String,
pub names: Vec<String>,
}
pub fn encode(data: &mut ByteEncoder, imports: &[Import]) -> usize {
let mut names = 0;
for import in imports {
for name in &import.names {
data.extend(ImportLookupEntry::name().bytes());
names += 1;
}
data.extend(ImportLookupEntry::NULL.bytes());
}
let table_addr = data.pos();
for import in imports {
let idt = ImportDirTable {
lookup_table_rva: todo!(),
time_date_stamp: 0,
forwarder_chain: 0,
name_rva: todo!(),
address_table_rva: todo!(),
};
}
for import in imports {
for name in &import.names {
hint_name_entry(data, 0, &name);
}
}
}
#[repr(C)]
pub struct ImportDirTable {
pub lookup_table_rva: u32,
pub time_date_stamp: u32,
pub forwarder_chain: u32,
pub name_rva: u32,
pub address_table_rva: u32,
}
impl ImportDirTable {
pub const NULL: Self = Self {
lookup_table_rva: 0,
time_date_stamp: 0,
forwarder_chain: 0,
name_rva: 0,
address_table_rva: 0,
};
}
#[repr(C)]
pub struct ImportLookupEntry(u64);
impl ImportLookupEntry {
pub const NULL: Self = Self(0);
pub fn name(hint_name_table_rva: u32) -> Self {
assert!(hint_name_table_rva >> 30 == 0);
Self(hint_name_table_rva as u64)
}
pub fn ordinal(ordinal: u16) -> Self {
Self(ordinal as u64 | (1 << 63))
}
pub fn bytes(&self) -> [u8; 8] {
self.0.to_le_bytes()
}
}
pub fn hint_name_entry(out: &mut Vec<u8>, hint: u16, name: &str) {
out.extend(hint.to_le_bytes());
out.extend(name.as_bytes());
out.push(0);
if out.len() % 2 == 1 {
out.push(0);
}
}
+112
View File
@@ -0,0 +1,112 @@
mod data_dir;
mod header;
mod import;
use data_dir::*;
use header::*;
use crate::backend::{ByteEncoder, LinkedProgram, pe::import::Import};
pub fn create(program: &[u8], start_offset: u64) -> Vec<u8> {
let mut data = ByteEncoder::default();
let file_align = 1;
let section_align = 1;
let code_size = program.len() as u32;
let mzheader = MZHeader {
magic: u16::from_ne_bytes(*b"MZ"),
stuff: [0; _],
lfanew: size_of::<MZHeader>() as u32,
};
data.val(&mzheader);
let header = PeHeader {
magic: u32::from_ne_bytes(*b"PE\0\0"),
machine: 0x8664,
num_sections: 1,
time_date_stamp: 0,
sym_tab_ptr: 0,
num_symbols: 0,
opt_header_size: size_of::<OptHeader64>() as u16,
// https://learn.microsoft.com/en-us/windows/win32/debug/pe-format
// executable | can handle >2GB addrs | debug info removed
characteristics: 0x2 | 0x20 | 0x0200,
};
data.val(&header);
let opt_header_pos = data.reserve::<OptHeader64>();
let num_of_rva_and_sizes: u32 = (size_of::<DataDirs>() / size_of::<DataDir>()) as u32;
data.val(&DataDirs::default());
let code_sect_pos = data.reserve::<Section>();
let hdr_size = data.pos() as u32;
let imports = [Import {
name: "kernel32.lib".to_string(),
names: ["GetStdHandle", "WriteFile", "ExitProcess"]
.map(String::from)
.to_vec(),
}];
import::encode(&mut data, &imports);
let code_addr = data.pos() as u32;
data.data.extend(program);
let code_section = Section {
name: *b".text\0\0\0",
virtual_size: code_size,
virtual_addr: hdr_size.next_multiple_of(section_align),
raw_data_size: code_size.next_multiple_of(file_align),
raw_data_ptr: code_addr,
reloc_ptr: 0,
line_num_ptr: 0,
num_relocs: 0,
num_line_nums: 0,
characteristics: 0x60000020,
};
let file_size = data.pos() as u32;
let opt_header = OptHeader64 {
magic: 0x20b,
major_linker_ver: 8,
minor_linker_ver: 0,
code_size: code_size.next_multiple_of(file_align),
init_data_size: 0,
uninit_data_size: 0,
entry_addr: code_addr + start_offset as u32,
code_base: code_addr,
image_base: 0x400000,
section_align,
file_align,
major_os_ver: 4,
minor_os_ver: 0,
major_image_ver: 0,
minor_image_ver: 0,
major_subsystem_ver: 4,
minor_subsystem_ver: 0,
win32_ver: 0,
image_size: file_size.next_multiple_of(section_align),
headers_size: hdr_size.next_multiple_of(file_align),
checksum: 0,
subsystem: 3, // windows CLI app
dll_characteristics: 0x400,
stack_reserve_size: 0x100000,
stack_commit_size: 0x1000,
heap_reserve_size: 0x100000,
heap_commit_size: 0x1000,
loader_flags: 0,
num_of_rva_and_sizes,
};
data.fill(opt_header_pos, &opt_header);
data.fill(code_sect_pos, &code_section);
data.data
}
impl LinkedProgram<u64> {
pub fn to_pe(&self) -> Vec<u8> {
create(&self.code, self.entry.expect("no start"))
}
}
-8
View File
@@ -1,12 +1,4 @@
use crate::backend::elf;
pub struct LinkedProgram<Addr> {
pub code: Vec<u8>,
pub entry: Option<Addr>,
}
impl LinkedProgram<u64> {
pub fn to_elf(&self) -> Vec<u8> {
elf::create(&self.code, self.entry.expect("no start"))
}
}
+2 -2
View File
@@ -1,7 +1,7 @@
mod elf;
mod container;
mod ir;
mod link;
pub use container::*;
pub use ir::*;
pub use link::*;
+3 -3
View File
@@ -56,12 +56,12 @@ pub fn parse_imm(mut s: &str, span: Span) -> Result<u64, CompilerMsg> {
u64::from_str_radix(s, radix).map_err(|_| CompilerMsg::from(("invalid immediate", span)))
}
pub fn parse_rmi(ctx: &mut crate::parser::ParseCtx) -> Result<RegImm, CompilerMsg> {
pub fn parse_rmi(ctx: &mut crate::parser::ParseCtx) -> Result<RegModeImm, CompilerMsg> {
let next = ctx.expect_next()?;
let err = || CompilerMsg::unexpected_token(&next, ctx.span, "a register or immediate");
Ok(match &next {
Token::Ident(ident) => RegImm::Reg(RegMode::parse(ident).ok_or_else(err)?),
Token::Lit(LitTy::Number(num)) => RegImm::Imm(parse_imm(num, ctx.span)?),
Token::Ident(ident) => RegModeImm::Reg(RegMode::parse(ident).ok_or_else(err)?),
Token::Lit(LitTy::Number(num)) => RegModeImm::Imm(parse_imm(num, ctx.span)?),
_ => return Err(err()),
})
}
BIN
View File
Binary file not shown.