Files
jsrt/compiler/src/regify/walk.rs
T
2026-10-16 08:00:06 +00:00

323 lines
10 KiB
Rust

//! This module converts one basic block's instrs from stack to register form using a simulated stack
//!
//! The output reg form instrs are structured as:
//! `dsts = op(imm, args)`
//!
//! Every instr's pushes allocates new register operands which are pushed on the stack and form the
//! output instr's `dsts`, the instr's pops pop operands from the stack and form the output instr's
//! `args`
//!
//! Each instr's `op` and `imm` is preserved except for some special cases
//!
//! `set_loc` is henceforth used to move data from one operand (local, arg, reg) to another
#![allow(non_snake_case)]
use super::{BasicBlock, Operand, RegInstr};
use crate::libqjs::parse::{ImmOperand, Instr, opcodes::OpCode};
/// Walk a basic block's instructions, converting each instr to register form
pub fn walk_block(
block: &BasicBlock,
code: &[Instr],
stack: &mut Vec<Operand>,
next_reg: &mut usize,
out: &mut [RegInstr],
) {
let mut fresh_reg = || {
let r = *next_reg;
*next_reg += 1;
Operand::Reg(r)
};
for instr_idx in block.start..block.end {
let instr = &code[instr_idx];
let out_instr = &mut out[instr_idx];
if try_handle_named_read(instr, stack, &mut fresh_reg, out_instr) {
continue;
}
if try_handle_named_write(instr, stack, out_instr) {
continue;
}
if try_handle_shuffle(instr, stack, out_instr) {
continue;
}
handle_normal_case(instr, stack, &mut fresh_reg, out_instr);
if instr.op == OpCode::DefineFieldRef {
let prev = std::mem::replace(&mut out[instr_idx - 1], RegInstr::nop());
resolve_location(&prev, &mut out[instr_idx]);
}
}
}
/// Handles a "named read" i.e. a read of a local or arg (and not a stack value)
///
/// We simulate pushing the operand to stack, and replace the instr with `SetLoc`, which
/// is our catch-all for a data-move instr
fn try_handle_named_read(
instr: &Instr,
stack: &mut Vec<Operand>,
mut fresh_reg: impl FnMut() -> Operand,
out: &mut RegInstr,
) -> bool {
use OpCode::*;
let arg = match instr.op {
GetLoc0 => Operand::Local(0),
GetLoc1 => Operand::Local(1),
GetLoc2 => Operand::Local(2),
GetLoc3 => Operand::Local(3),
GetLoc | GetLoc8 | GetLocCheck => match instr.imm {
ImmOperand::Loc(k) => Operand::Local(k as usize),
_ => unreachable!(),
},
GetArg0 => Operand::Arg(0),
GetArg1 => Operand::Arg(1),
GetArg2 => Operand::Arg(2),
GetArg3 => Operand::Arg(3),
GetArg => match instr.imm {
ImmOperand::Arg(k) => Operand::Arg(k as usize),
_ => unreachable!(),
},
_ => return false,
};
let dst = fresh_reg();
// Converted SetLoc, reading from arg, writing to dst
*out = RegInstr {
dsts: vec![dst],
op: SetLoc,
imm: ImmOperand::None,
args: vec![arg],
};
// Simulate stack push
stack.push(dst);
true
}
/// Handles a "named write" i.e. a write of a local or arg (and not a stack value)
///
/// We simulate popping the operand off stack, and replace the instr with `SetLoc`, which
/// is our catch-all for a data-move instr
///
/// A distinction needs to be made between `Set*` and `Put*` since `Set*` instrs leave
/// the value-to-write on the stack and `Put*` pops it off
fn try_handle_named_write(instr: &Instr, stack: &mut Vec<Operand>, out: &mut RegInstr) -> bool {
use OpCode::*;
let (dst, preserved) = match instr.op {
PutLoc0 => (Operand::Local(0), false),
PutLoc1 => (Operand::Local(1), false),
PutLoc2 => (Operand::Local(2), false),
PutLoc3 => (Operand::Local(3), false),
PutLoc | PutLoc8 | PutLocCheck | PutLocCheckInit => match instr.imm {
ImmOperand::Loc(k) => (Operand::Local(k as usize), false),
_ => unreachable!(),
},
PutArg0 => (Operand::Arg(0), false),
PutArg1 => (Operand::Arg(1), false),
PutArg2 => (Operand::Arg(2), false),
PutArg3 => (Operand::Arg(3), false),
SetLoc0 => (Operand::Local(0), true),
SetLoc1 => (Operand::Local(1), true),
SetLoc2 => (Operand::Local(2), true),
SetLoc3 => (Operand::Local(3), true),
SetLoc | SetLoc8 => match instr.imm {
ImmOperand::Loc(k) => (Operand::Local(k as usize), true),
_ => unreachable!(),
},
SetArg0 => (Operand::Arg(0), true),
SetArg1 => (Operand::Arg(1), true),
SetArg2 => (Operand::Arg(2), true),
SetArg3 => (Operand::Arg(3), true),
_ => return false,
};
// Simulate stack pop or peek, depending on instr
let arg = if preserved {
stack.last().unwrap().clone()
} else {
stack.pop().unwrap()
};
// Converted SetLoc, reading from arg, writing to dst
*out = RegInstr {
dsts: vec![dst],
op: SetLoc,
imm: ImmOperand::None,
args: vec![arg],
};
true
}
/// Handles an instr that only moves stack values around: it pops `n` values, and pushes some
/// of them again. No value is made, so `out_instr` becomes `nop`
fn try_handle_shuffle(instr: &Instr, stack: &mut Vec<Operand>, out_instr: &mut RegInstr) -> bool {
let Some((n, pushes)) = shuffle(instr.op) else {
return false;
};
let popped = stack.split_off(stack.len() - n);
stack.extend(pushes.iter().map(|&i| popped[i]));
*out_instr = RegInstr::nop();
true
}
/// The shuffles: how many values an instr pops, and which of them it pushes, in stack order
/// (bottom first). The numbers index the popped values, bottom first
fn shuffle(op: OpCode) -> Option<(usize, &'static [usize])> {
use OpCode::*;
Some(match op {
Drop => (1, &[]), // a ->
Nip => (2, &[1]), // a b -> b
Nip1 => (3, &[1, 2]), // a b c -> b c
Dup => (1, &[0, 0]), // a -> a a
Dup1 => (2, &[0, 0, 1]), // a b -> a a b
Dup2 => (2, &[0, 1, 0, 1]), // a b -> a b a b
Dup3 => (3, &[0, 1, 2, 0, 1, 2]), // a b c -> a b c a b c
Insert2 => (2, &[1, 0, 1]), // obj a -> a obj a
Insert3 => (3, &[2, 0, 1, 2]), // obj prop a -> a obj prop a
Insert4 => (4, &[3, 0, 1, 2, 3]), // this obj prop a -> a this obj prop a
Perm3 => (3, &[1, 0, 2]), // obj a b -> a obj b
Perm4 => (4, &[2, 0, 1, 3]), // obj prop a b -> a obj prop b
Perm5 => (5, &[3, 0, 1, 2, 4]), // this obj prop a b -> a this obj prop b
Swap => (2, &[1, 0]), // a b -> b a
Swap2 => (4, &[2, 3, 0, 1]), // a b c d -> c d a b
Rot3l => (3, &[1, 2, 0]), // x a b -> a b x
Rot3r => (3, &[2, 0, 1]), // a b x -> x a b
Rot4l => (4, &[1, 2, 3, 0]), // x a b c -> a b c x
Rot5l => (5, &[1, 2, 3, 4, 0]), // x a b c d -> a b c d x
// Binds [[HomeObject]], which only `super` reads. `super` is not supported
SetHomeObject => (2, &[0, 1]),
// Makes the key a property key. A key is kept as it is (see infer)
ToPropkey2 => (2, &[0, 1]),
_ => return None,
})
}
/// Handle a normal instruction with simulated stack pushes/pops
fn handle_normal_case(
instr: &Instr,
stack: &mut Vec<Operand>,
mut fresh_reg: impl FnMut() -> Operand,
out_instr: &mut RegInstr,
) {
use OpCode::*;
// op and imm carried as-is
*out_instr = RegInstr {
dsts: vec![],
op: instr.op,
imm: instr.imm,
args: vec![],
};
// Some instrs have extra pops not reflected in info.n_pop
let extra_pops = match instr.imm {
ImmOperand::Npop(npop) => npop as usize,
ImmOperand::NpopU16 { npop, .. } => npop as usize,
_ => 0,
};
let info = instr.op.info();
let n_pop = info.n_pop as usize + extra_pops;
let n_push = info.n_push as usize;
// Simulate this instr's pops, top first
for _ in 0..n_pop {
let x = stack.pop().unwrap();
out_instr.args.push(x);
}
if matches!(instr.op, Call | CallMethod | CallConstructor) {
out_instr.args = call_operands(instr.op, std::mem::take(&mut out_instr.args));
}
// These leave the object they wrote on the stack: the same value, not a new one
if matches!(instr.op, DefineField | DefineFieldRef) {
stack.push(out_instr.args[1]);
return;
}
// Simulate this instr's pushes
for _ in 0..n_push {
let x = fresh_reg();
stack.push(x);
out_instr.dsts.push(x);
}
}
/// Put a call's popped operands in the order of the callee's params: the callee (which is its
/// env), `this`, then the args in source order
///
/// `call_method` has `this` at the bottom of its frame, below the callee. `call_constructor`
/// has the callee at the bottom and `this` above it (see transform/new.rs). `call` has `this`
/// on top, where transform pushed it
fn call_operands(op: OpCode, mut popped: Vec<Operand>) -> Vec<Operand> {
let (callee, this) = match op {
OpCode::CallMethod => {
let this = popped.pop().unwrap();
(popped.pop().unwrap(), this)
}
OpCode::CallConstructor => {
let callee = popped.pop().unwrap();
(callee, popped.pop().unwrap())
}
_ => {
let callee = popped.pop().unwrap();
(callee, popped.remove(0))
}
};
// What remains are the args, last arg first
popped.reverse();
[callee, this].into_iter().chain(popped).collect()
}
/// `define_field_ref x` defines `x` as the location that its value was read from (see
/// transform/closure.rs), and the instr before it is that read. Its value arg becomes the
/// location:
/// - a local or arg: the field is that binding
/// - the reg of an object `o`, for a read `get_field x` of `o`: the field is `o.x`, which is
/// a location itself (a capture that the owner gives to its own child)
///
/// The read itself is then not used, so it becomes `nop`
fn resolve_location(prev: &RegInstr, out_instr: &mut RegInstr) {
let value = out_instr.args[0];
assert_eq!(
prev.dsts,
[value],
"define_field_ref must follow the read of its value"
);
out_instr.args[0] = match prev.op {
OpCode::SetLoc => prev.args[0],
OpCode::GetField => {
assert!(prev.imm.atom() == out_instr.imm.atom());
prev.args[0]
}
op => panic!("define_field_ref after {op:?}, which reads no location"),
};
}