Files
lang/src/arch/x86_64/compile.rs
T
2026-07-19 20:43:49 -04:00

255 lines
8.5 KiB
Rust

use std::collections::{HashMap, HashSet, VecDeque};
use super::*;
use crate::backend::{Body, Func, LibImport, LinkedProgram, SymImport, SymTable, Symbol, VarId};
pub struct Encoder<'a> {
pub code: Code,
pub sym_tab: SymTable<u64>,
pub sym_refs: HashMap<Symbol, Vec<usize>>,
pub program: &'a Program<X86_64>,
pub segs: Vec<SegUses>,
pub vars: HashMap<VarId, Reg>,
}
pub fn compile(p: &Program<X86_64>) -> Result<LinkedProgram<u64>, CompilerMsg> {
let mut encoder = Encoder::new(p);
p.encode_data(&mut encoder.code.bytes, &mut encoder.sym_tab);
for f in &p.funcs {
encoder.func(f);
}
for (pos, sym) in encoder.code.missing.drain(..) {
let info = encoder.program.sym_info(sym);
if info.external {
encoder.sym_refs.entry(sym).or_default().push(pos);
} else {
let addr = encoder
.sym_tab
.get(sym)
.ok_or(CompilerMsg::from(format!("missing symbol {}", info.name)))?;
encoder.code.bytes[pos..pos + 4].copy_from_slice(&addr_offset(pos, addr))
}
}
let imports = p
.external
.iter()
.map(|e| LibImport {
name: e.file.clone(),
syms: e
.syms
.iter()
.map(|&s| SymImport {
name: p.sym_info(s).name.clone(),
usages: encoder.sym_refs.entry(s).or_default().clone(),
})
.collect(),
})
.collect();
Ok(LinkedProgram {
code: encoder.code.bytes,
entry: p.entry.and_then(|e| encoder.sym_tab.get(e)),
imports,
})
}
#[derive(Default)]
pub struct SegUses {
var: HashMap<VarId, VecDeque<usize>>,
reg: [VecDeque<RegUse>; 16],
}
#[derive(Clone, Copy)]
pub struct RegUse {
pos: usize,
// required var, if any
var: Option<VarId>,
}
type BInstr = crate::backend::Instr<X86_64>;
impl<'a> Encoder<'a> {
fn func(&mut self, f: &Func<X86_64>) -> EncodeRes {
let addr = self.code.bytes.len();
self.sym_tab.insert(f.sym, addr as u64);
self.segs.clear();
self.calc_uses(&f.body);
for instr in &f.body {
match instr {
BInstr::Asm(asm) => self.asm(asm)?,
_ => todo!(),
}
}
Ok(())
}
pub fn calc_uses(&mut self, body: &Body<X86_64>) -> (usize, usize) {
let mut pos = 0;
let seg_i = self.segs.len();
self.segs.push(Default::default());
let mut uses = SegUses::default();
macro_rules! push {
($var:ident) => {
push!($var, 0)
};
($var:ident, $pos:expr) => {
uses.var
.entry($var.clone())
.or_default()
.push_back(pos + $pos)
};
}
for instr in body {
match instr {
BInstr::Set { dst, src: _ } => push!(dst),
BInstr::Call { dst, f, args } => {
let conv = &self.program.call_convs[self.program.funcs[f].conv];
push!(dst);
for &reg in conv.scratch() {
uses.reg[reg as usize].push_back(RegUse { pos, var: None });
}
for (i, &arg) in args.iter().enumerate() {
push!(arg);
if let Some(&reg) = conv.param().get(i) {
uses.reg[reg as usize].back_mut().unwrap().var = Some(arg);
}
}
}
BInstr::Copy { dst, src } => {
push!(dst);
push!(src);
}
BInstr::Add { dst, src1, src2 } => {
push!(dst);
push!(src1);
push!(src2);
}
BInstr::If { cond, then, else_ } => {
push!(cond);
pos += 1;
let (seg_i1, len1) = self.calc_uses(then);
let (seg_i2, len2) = self.calc_uses(else_);
// insert closest usages
for (var, poss1) in &self.segs[seg_i1].var {
if let Some(poss2) = self.segs[seg_i2].var.get(var) {
push!(var, poss1[0].min(poss2[0]));
} else {
push!(var, poss1[0]);
}
}
for (var, poss2) in &self.segs[seg_i2].var {
if !self.segs[seg_i1].var.contains_key(var) {
push!(var, poss2[0]);
}
}
pos += len1.max(len2);
continue;
}
BInstr::Loop(instrs) => {
let (seg_i2, len) = self.calc_uses(instrs);
// insert closest usages
for (var, poss) in &self.segs[seg_i2].var {
push!(var, poss[0]);
}
// during register allocation, want to insert the first use of each var
// at len + use, because if it jumps back up, that will be the next
// usage rather than if the loop exits; pad ensures there is room so
// the insertions are always before the rest of the usages (after loop)
let pad = (len * 2).saturating_sub(pos);
pos += len + pad;
continue;
}
BInstr::Break(var) => {
if let Some(var) = var {
push!(var);
}
}
BInstr::Return(var) => {
if let Some(var) = var {
push!(var);
}
}
BInstr::Asm(asm) => {
let mut used = HashSet::new();
for i in &asm.instrs {
if let Some(reg) = i.dst_reg() {
used.insert(reg.reg());
}
}
for &(reg, var) in &asm.args {
push!(var);
used.remove(&reg);
uses.reg[reg as usize].push_back(RegUse {
pos,
var: Some(var),
});
}
for reg in used {
uses.reg[reg as usize].push_back(RegUse { pos, var: None });
}
}
}
pos += 1;
}
self.segs[seg_i] = uses;
return (seg_i, pos);
}
pub fn rvm(&mut self, input: Rvm) -> RegMemKind {
match input {
Rvm::Reg(reg) => reg,
Rvm::Var(var_id) => todo!(),
Rvm::Mem(mem) => todo!(),
}
}
pub fn asm(&mut self, asm: &Asm) -> EncodeRes {
let mut used = HashSet::default();
let mut vars = HashSet::default();
for &instr in &asm.instrs {
if let Some(regw) = instr.dst_reg() {
used.insert(regw.reg());
}
for var in instr.vars() {
vars.insert(var);
}
}
let overlap = used.intersection(&self.vars.values());
for var in &vars {
if let Some(reg) = self.vars.get(var)
&& used.contains(reg)
{}
}
for &instr in &asm.instrs {
match instr {
Instr::Mov { dst, src } => self.code.mov(dst, src)?,
Instr::Push(rvmi) => todo!(),
Instr::Pop(rvm) => todo!(),
Instr::Lea { dst, src } => todo!(),
Instr::Int(code) => self.code.int(code),
Instr::Syscall => self.code.syscall(),
Instr::Call(sym) => self.code.call(sym),
Instr::CallAt(sym) => self.code.call_mem(sym),
Instr::Ret => self.code.ret(),
Instr::Add { dst, src } => todo!(),
Instr::Sub { dst, src } => todo!(),
}
}
Ok(())
}
pub fn new(program: &'a Program<X86_64>) -> Self {
Self {
code: Code::default(),
sym_tab: SymTable::new(program.sym_count()),
sym_refs: Default::default(),
segs: Default::default(),
program,
}
}
}