Modifications

This commit is contained in:
2026-09-07 11:44:42 +00:00
parent 617dbeb988
commit 457e63ad8e
7 changed files with 259 additions and 225 deletions
+13 -14
View File
@@ -9,23 +9,22 @@
mod step_1;
mod step_2;
use crate::{parse::Atom, regify::RegFn};
use crate::{declass::step_1::ClassSequence, regify::RegFn};
/// A class's constructor and its hoisted field initialiser, as fn ids into the
/// flat function list. Read by step 1, consumed by step 2.
#[derive(Clone, Copy)]
struct FieldInit {
ctor: usize,
fields: usize,
}
use step_1::{find_class_seqs, rewrite_class_to_proto};
use step_2::splice_ctor;
pub fn declass(reg_fns: Vec<RegFn>, _atoms: &[Atom]) -> Vec<RegFn> {
// Read the constructor <-> field-initialiser links before step 1 rewrites
// the class builds that carry them.
let field_inits = step_1::field_init_links(&reg_fns);
pub fn declass(mut reg_fns: Vec<RegFn>) -> Vec<RegFn> {
let class_seqs = reg_fns
.iter()
.map(find_class_seqs)
.collect::<Vec<Vec<ClassSequence>>>();
let mut reg_fns: Vec<RegFn> = reg_fns.into_iter().map(step_1::rewrite).collect();
reg_fns
.iter_mut()
.zip(&class_seqs)
.for_each(|(f, s)| rewrite_class_to_proto(f, s));
step_2::splice(&mut reg_fns, &field_inits);
splice_ctor(&mut reg_fns, &class_seqs);
reg_fns
}
+114 -133
View File
@@ -1,19 +1,3 @@
//! Step 1 of declass: rewrite how a class is built, in prototype form.
//!
//! A class-build cluster (declass_algo.txt, step 1):
//!
//! push_const8 #ctor ; define_class @Name
//! ( fclosure8 #m ; define_method @meth )*
//! ( fclosure8 #fields ; set_home_object ; put_loc ; drop )?
//!
//! becomes `fclosure8 #ctor` followed, per method, by
//! `get_field "prototype" ; put_field @meth`; the field-initialiser tail is
//! dropped. Detection (`find_class_seqs`) records the ranges; replacement
//! (`ClassSequence::emit_prototype`) reads operands out of each matched range.
//!
//! Derived classes (`extends`) are unsupported and panic: prototype form would
//! have to flatten the parent's prototype in and lower `super`.
use std::range::Range;
use crate::{
@@ -21,81 +5,124 @@ use crate::{
regify::{RegFn, RegInstr},
};
use super::FieldInit;
pub fn find_class_seqs(reg_fn: &RegFn) -> Vec<ClassSequence> {
let mut ins = reg_fn.code.as_slice();
let mut seqs = vec![];
let mut offset = 0;
/// Read each constructor <-> field-initialiser link off the class builds, as
/// fn ids. Run before `rewrite`, which drops the instructions carrying them.
pub(super) fn field_init_links(fns: &[RegFn]) -> Vec<FieldInit> {
let mut links = vec![];
for f in fns {
for seq in find_class_seqs(&f.code) {
if let Some(link) = seq.field_init(&f.code, &f.cpool) {
links.push(link);
while !ins.is_empty() {
let seq = match_class_seq(ins); // Try to match a class decl sequence
match seq {
// Found class sequence, store it and advance `ins`
Ok((rest, num_methods)) => {
let consumed = ins.len() - rest.len(); // How many instrs were consumed
let seq_instr = &ins[..consumed]; // The class seq's instrs
seqs.push(ClassSequence {
range: Range {
start: offset,
end: offset + consumed,
},
num_methods,
ctor_fn: fn_id(&reg_fn.cpool, as_const(seq_instr[0].imm)),
field_init_fn: find_field_init_fn(seq_instr, num_methods, &reg_fn.cpool),
});
offset += consumed;
ins = rest;
}
}
}
links
}
/// Rewrite every class build in one function to prototype form.
pub(super) fn rewrite(mut reg_fn: RegFn) -> RegFn {
let seqs = find_class_seqs(&reg_fn.code);
// Move the code out of reg_fn so we can consume it while still calling
// reg_fn.fresh_val(); the empty Vec left behind costs nothing.
let code = std::mem::take(&mut reg_fn.code);
let total = code.len();
let mut out = Vec::with_capacity(total);
let mut instrs = code.into_iter();
let mut seqs = seqs.iter();
let mut next_seq = seqs.next();
let mut pos = 0;
while pos < total {
match next_seq {
Some(seq) if seq.range.start == pos => {
let adv = seq.range.end - seq.range.start;
let matched: Vec<RegInstr> = instrs.by_ref().take(adv).collect();
seq.emit_prototype(&mut reg_fn, &mut out, &matched);
pos += adv;
next_seq = seqs.next();
}
_ => {
out.push(instrs.next().unwrap());
pos += 1;
// Did not find sequence, advance `ins` by 1 and try again
Err(rest) => {
let consumed = ins.len() - rest.len(); // Handle partial consumption
let consumed = consumed.max(1);
offset += consumed;
ins = &ins[consumed..];
}
}
}
reg_fn.code = out;
reg_fn
seqs
}
struct ClassSequence {
range: Range<usize>,
pub fn rewrite_class_to_proto(reg_fn: &mut RegFn, class_seqs: &[ClassSequence]) {
// The prototype form is never longer than the class build it replaces, so
// it drops straight over it and every instr outside a class seq stays put.
// The code moves out only so fresh_val() can borrow reg_fn beside it.
let mut code = std::mem::take(&mut reg_fn.code);
let mut buf = vec![]; // Reused across seqs, so it allocates once
for seq in class_seqs {
seq.emit_prototype(&code[seq.range], &mut || reg_fn.fresh_val(), &mut buf);
code[seq.range].swap_with_slice(&mut buf);
}
reg_fn.code = code;
}
pub struct ClassSequence {
pub range: Range<usize>,
pub num_methods: usize,
pub ctor_fn: usize, // Fn id of the class ctor
pub field_init_fn: Option<usize>, // Fn id of the field init, if any
}
/// The field initialiser's fn id, if the class has a field-initialiser tail.
fn find_field_init_fn(
matched: &[RegInstr],
num_methods: usize,
cpool: &[JsValue],
) -> Option<usize> {
let tail = 2 + 2 * num_methods; // The ctor and its methods, before the tail
if matched.len() == tail {
return None; // No field-initialiser tail
}
Some(fn_id(cpool, as_const(matched[tail].imm)))
}
impl ClassSequence {
/// Emit the prototype form, reading operands straight out of `matched` --
/// the class sequence's own instructions.
fn emit_prototype(&self, reg_fn: &mut RegFn, out: &mut Vec<RegInstr>, matched: &[RegInstr]) {
/// Emit the prototype form into the scratchpad `buf`, reading operands
/// straight out of `matched` -- this class sequence's own instructions.
/// `buf` comes back padded with nops to exactly the length of the class
/// build, so the caller can drop it over the original in place.
fn emit_prototype<F>(&self, matched: &[RegInstr], fresh_val: &mut F, buf: &mut Vec<RegInstr>)
where
F: FnMut() -> usize,
{
buf.clear();
// push_const8 #ctor ; define_class -> fclosure8 #ctor (constructor kept)
let (ctor_const, ctor_val) = self.ctor(matched);
out.push(fclosure8(ctor_const, ctor_val));
buf.push(fclosure8(ctor_const, ctor_val));
// each ( fclosure8 #m ; define_method @meth ) -> get_field + put_field
for k in 0..self.num_methods {
let (method_const, method_val, meth) = self.method(matched, k);
out.push(fclosure8(method_const, method_val));
let proto = reg_fn.fresh_val();
out.push(get_field(proto_atom(), proto, ctor_val));
out.push(put_field(ImmOperand::Atom(meth), method_val, proto));
// A class with no methods has no prototype to read
if self.num_methods > 0 {
// The prototype cannot change between the put_fields below, so read
// it once for the whole class instead of once per method
let proto = fresh_val();
buf.push(get_field(proto_atom(), proto, ctor_val));
// each ( fclosure8 #m ; define_method @meth ) -> fclosure8 + put_field
for k in 0..self.num_methods {
let (method_const, method_val, meth) = self.method(matched, k);
buf.push(fclosure8(method_const, method_val));
buf.push(put_field(ImmOperand::Atom(meth), method_val, proto));
}
}
// the fclosure8 #fields ; set_home_object ; put_loc ; drop tail is dropped.
// 2 + 2n emitted against the 2 + 2n matched, less the 4-instr field
// initialiser tail, so the emission always fits. Nop pads the slack; it
// pops and pushes nothing, so it cannot disturb the value flow.
let len = self.range.end - self.range.start;
debug_assert!(buf.len() <= len, "class rewrite outgrew its range");
buf.resize_with(len, nop);
}
/// The constructor's constant and value: push_const8 #ctor ; define_class.
/// The constructor's constant and register ID it was written to
/// push_const8 #ctor ; define_class.
fn ctor(&self, matched: &[RegInstr]) -> (u32, usize) {
(as_const(matched[0].imm), matched[1].dsts[0])
}
@@ -111,64 +138,22 @@ impl ClassSequence {
atom_of(define_method.imm),
)
}
/// The constructor and field-initialiser fn ids, if the class has a
/// field-initialiser tail. `code` is the whole enclosing function, so the
/// range indexes it directly.
fn field_init(&self, code: &[RegInstr], cpool: &[JsValue]) -> Option<FieldInit> {
let base = 2 + 2 * self.num_methods; // constructor + methods, before the tail
if self.range.end - self.range.start == base {
return None; // no field-initialiser tail
}
let ctor_const = as_const(code[self.range.start].imm);
let fields_const = as_const(code[self.range.start + base].imm);
Some(FieldInit {
ctor: fn_id(cpool, ctor_const),
fields: fn_id(cpool, fields_const),
})
}
}
fn find_class_seqs(mut ins: &[RegInstr]) -> Vec<ClassSequence> {
let mut seqs = vec![];
let mut offset = 0;
while !ins.is_empty() {
match match_class_seq(ins) {
Ok((rest, num_methods)) => {
let adv = ins.len() - rest.len();
seqs.push(ClassSequence {
range: Range {
start: offset,
end: offset + adv,
},
num_methods,
});
offset += adv;
ins = rest;
}
Err(_) => {
offset += 1;
ins = &ins[1..];
}
}
}
seqs
}
// ── the grammar, as matchers over the instruction stream ───────────────────
// ─────────────────────────── Class Sequence Grammar ─────────────────────────────────────
// push_const8 #ctor ; define_class @Name <-- Class decl header
// ( fclosure8 #method ; define_method @meth )* <-- Methods
// ( fclosure8 #fields ; set_home_object ; put_loc ; drop )? <-- Field initializer stuff
//
// match fns return Ok(advanced_instrs) or Err(original_instrs)
fn match_class_seq(ins: &[RegInstr]) -> Result<(&[RegInstr], usize), &[RegInstr]> {
let ins = match_one(ins, |i| i.op == OpCode::PushConst8)?;
// A class build is `push_const8 ; define_class`. Derived classes (`extends`)
// are unsupported: prototype form would have to flatten the parent's
// prototype in and lower `super`, neither of which is implemented.
if let Some(i) = ins.first() {
if i.op == OpCode::DefineClass && is_derived(i) {
let is_derived_class = matches!(i.imm, ImmOperand::AtomU8 { arg, .. } if arg & 1 == 1);
if i.op == OpCode::DefineClass && is_derived_class {
panic!("declass: derived classes (extends) are not supported");
}
}
@@ -180,15 +165,7 @@ fn match_class_seq(ins: &[RegInstr]) -> Result<(&[RegInstr], usize), &[RegInstr]
Ok((ins, num_methods))
}
/// A derived class: `define_class` with the derived flag set.
fn is_derived(i: &RegInstr) -> bool {
matches!(i.imm, ImmOperand::AtomU8 { arg, .. } if arg & 1 == 1)
}
// TODO: accessors are not handled. A getter/setter is a `define_method` with a
// non-zero kind in `arg`; this matcher takes it for an ordinary method, so it
// is wrongly lowered to a `put_field` (a data property) instead of an accessor
// property. It should instead make the whole class pass through untouched.
// TODO: accessors are not handled
fn match_method_seq(ins: &[RegInstr]) -> Result<&[RegInstr], &[RegInstr]> {
let ins = match_one(ins, |i| i.op == OpCode::Fclosure8)?;
let ins = match_one(ins, |i| i.op == OpCode::DefineMethod)?;
@@ -215,7 +192,7 @@ fn match_field_init_seq(ins: &[RegInstr]) -> Result<&[RegInstr], &[RegInstr]> {
Ok(ins)
}
// ── matcher primitives ─────────────────────────────────────────────────────
// ───────────────────────────── Matcher Primitives ───────────────────────────────────────
fn match_one(i: &[RegInstr], f: impl Fn(&RegInstr) -> bool) -> Result<&[RegInstr], &[RegInstr]> {
match i.first() {
@@ -250,6 +227,10 @@ fn match_opt<'a>(
// ── emitted instructions ───────────────────────────────────────────────────
fn nop() -> RegInstr {
RegInstr::new(OpCode::Nop, ImmOperand::None, vec![], vec![])
}
fn fclosure8(const_index: u32, dst: usize) -> RegInstr {
RegInstr::new(
OpCode::Fclosure8,
+26 -17
View File
@@ -10,7 +10,8 @@
//! This step takes the initialiser's field writes -- its body minus the
//! this/home-object prologue and the trailing return -- and splices them into
//! the constructor in place of that call, renumbered onto fresh value ids. The
//! initialiser function is left unreferenced.
//! initialiser is left with an empty body. It cannot be dropped from the
//! function list, because a function id is an index into that list.
//!
//! Deficiency, left for step 3: a class with NO instance fields compiles
//! `<class_fields_init>` to `Undefined` rather than a closure, so there is no
@@ -19,7 +20,8 @@
//! time -- a harmless no-op. Step 3 must strip that block when it drops the
//! `<class_fields_init>` capture, or the `get_var_ref_check` would dangle.
use super::FieldInit;
use super::step_1::ClassSequence;
use crate::{
parse::{ImmOperand, opcodes::OpCode},
regify::{RegFn, RegInstr},
@@ -28,21 +30,27 @@ use crate::{
/// The name quickjs gives the hoisted field initialiser.
const FIELDS_INIT: &str = "<class_fields_init>";
pub(super) fn splice(fns: &mut [RegFn], field_inits: &[FieldInit]) {
for &FieldInit { ctor, fields } in field_inits {
splice_one(fns, ctor, fields);
pub fn splice_ctor(fns: &mut [RegFn], class_seqs: &[Vec<ClassSequence>]) {
// Only a class with instance fields has an initialiser to splice.
for seq in class_seqs.iter().flatten() {
if let Some(field_init_fn) = seq.field_init_fn {
splice_one(fns, seq.ctor_fn, field_init_fn);
}
}
}
fn splice_one(fns: &mut [RegFn], ctor: usize, fields: usize) {
// Pull the field writes out of the initialiser (a separate function), then
// hand them to the constructor.
let writes = field_writes(&fns[fields]);
let ctor_fn = &mut fns[ctor];
let Some((start, end)) = call_span(ctor_fn) else {
// Find the call block first. If the constructor does not have one there is
// nothing to splice into, and the initialiser has to keep working.
let Some((start, end)) = call_span(&fns[ctor]) else {
return;
};
// Take the field writes out of the initialiser (a separate function), then
// hand them to the constructor.
let writes = take_field_writes(&mut fns[fields]);
let ctor_fn = &mut fns[ctor];
let writes = renumber(writes, ctor_fn);
ctor_fn.code.splice(start..=end, writes);
}
@@ -50,12 +58,13 @@ fn splice_one(fns: &mut [RegFn], ctor: usize, fields: usize) {
/// The initialiser's field writes: its body minus the this/home-object prologue
/// and the trailing return. They reach `this` through get_loc0, which is `this`
/// in the constructor too, so they carry over unchanged.
fn field_writes(init: &RegFn) -> Vec<RegInstr> {
init.code
.iter()
.filter(|i| !is_prologue(i.op))
.map(|i| RegInstr::new(i.op, i.imm, i.dsts.clone(), i.args.clone()))
.collect()
///
/// This empties the initialiser, which is dead once its writes move into the
/// constructor. The function itself has to stay: a function id is an index into
/// the function list, so removing one would renumber every function after it.
fn take_field_writes(init: &mut RegFn) -> Vec<RegInstr> {
init.code.retain(|i| !is_prologue(i.op));
std::mem::take(&mut init.code)
}
/// Renumber the writes' value ids onto fresh ids in the constructor. Each write
+53 -42
View File
@@ -47,28 +47,31 @@
//! to the class passes; step 3 leaves them untouched. Same for `apply`.
use crate::{
flatten::CaptureFrom,
flatten::{Capture, CaptureFrom},
parse::{AtomRef, ImmOperand, JsValue, opcodes::OpCode},
regify::{RegFn, RegInstr},
};
pub fn declosure(reg_fns: Vec<RegFn>) -> Vec<RegFn> {
// Where each function's captures come from, per fn id. Snapshotted first,
// because step 1 clears the capture lists that step 2 reads.
let capture_sources: Vec<Vec<CaptureFrom>> = reg_fns
.iter()
.map(|f| f.captures.iter().map(|c| c.from).collect())
pub fn declosure(mut reg_fns: Vec<RegFn>) -> Vec<RegFn> {
// Every function's captures, per fn id. Step 2 runs on a parent but needs
// its CHILD's captures, and `declosure_one` only ever holds one function,
// so the cross-function view has to be gathered here. Taking the lists
// moves them -- no per-function allocation -- and leaves every function
// with the empty capture list declosure ends with anyway.
let captures: Vec<Vec<Capture>> = reg_fns
.iter_mut()
.map(|f| std::mem::take(&mut f.captures))
.collect();
reg_fns
.into_iter()
.map(|f| declosure_one(f, &capture_sources))
.map(|f| declosure_one(f, &captures))
.collect()
}
fn declosure_one(reg_fn: RegFn, capture_sources: &[Vec<CaptureFrom>]) -> RegFn {
fn declosure_one(reg_fn: RegFn, captures: &[Vec<Capture>]) -> RegFn {
let reg_fn = take_env_param(reg_fn);
let reg_fn = build_closure_objects(reg_fn, capture_sources);
let reg_fn = build_closure_objects(reg_fn, captures);
route_calls_through_object(reg_fn)
}
@@ -81,18 +84,21 @@ fn take_env_param(mut reg_fn: RegFn) -> RegFn {
reg_fn.arg_count += 1;
let code = std::mem::take(&mut reg_fn.code);
let touches_captures = code.iter().any(|i| capture_access(i).is_some());
let mut out = Vec::with_capacity(code.len() + 1);
let env = reg_fn.fresh_val();
// Load the env once at entry -- only if a capture is actually accessed.
if touches_captures {
out.push(get_arg(env, env_arg));
// Nothing reads a capture, so there is nothing to rewrite. The parameter
// is still added, to keep step 3's call sites uniform.
if !code.iter().any(|i| is_capture_access(i).is_some()) {
reg_fn.code = code;
return reg_fn;
}
// Load the env once at entry.
let mut out = Vec::with_capacity(code.len() + 1);
let env = reg_fn.fresh_val();
out.push(get_arg(env, env_arg));
for instr in code {
match capture_access(&instr) {
match is_capture_access(&instr) {
// A read keeps its destination, so downstream uses do not change.
Some(CaptureAccess::Read(k)) => {
out.push(get_field(int_field(k + 1), instr.dsts[0], env))
@@ -105,7 +111,6 @@ fn take_env_param(mut reg_fn: RegFn) -> RegFn {
}
reg_fn.code = out;
reg_fn.captures.clear();
reg_fn
}
@@ -115,7 +120,7 @@ enum CaptureAccess {
Write(u32),
}
fn capture_access(i: &RegInstr) -> Option<CaptureAccess> {
fn is_capture_access(i: &RegInstr) -> Option<CaptureAccess> {
// The numbered forms fold the slot into the opcode; the rest carry it.
let slot = || match i.imm {
ImmOperand::VarRef(k) => k as u32,
@@ -142,7 +147,7 @@ fn capture_access(i: &RegInstr) -> Option<CaptureAccess> {
/// Replace each `fclosure` with the object that stands for it: the function in
/// field 0, then one field per capture, each loaded from the parent frame.
fn build_closure_objects(mut reg_fn: RegFn, capture_sources: &[Vec<CaptureFrom>]) -> RegFn {
fn build_closure_objects(mut reg_fn: RegFn, captures: &[Vec<Capture>]) -> RegFn {
let code = std::mem::take(&mut reg_fn.code);
let mut out = Vec::with_capacity(code.len());
@@ -153,10 +158,7 @@ fn build_closure_objects(mut reg_fn: RegFn, capture_sources: &[Vec<CaptureFrom>]
};
// The value the rest of the code already uses for this closure.
let closure = instr.dsts[0];
let captures: &[CaptureFrom] = capture_sources
.get(child)
.map(Vec::as_slice)
.unwrap_or(&[]);
let child_captures = &captures[child];
// The function is a plain constant, so it loads with `push_const`.
let load = match instr.op {
@@ -169,26 +171,25 @@ fn build_closure_objects(mut reg_fn: RegFn, capture_sources: &[Vec<CaptureFrom>]
let mut obj = reg_fn.fresh_val();
out.push(object(obj));
// Each define_field returns the object, so the stores chain. The LAST
// store must land in `closure`.
let dst = if captures.is_empty() {
closure
} else {
reg_fn.fresh_val()
};
out.push(define_field(int_field(0), dst, func, obj));
obj = dst;
// Field 0 is the function, fields 1.. are the captures. Each
// define_field returns the object, so the stores chain -- and the LAST
// must land in `closure`.
for k in 0..=child_captures.len() {
let value = if k == 0 {
func
} else {
let value = reg_fn.fresh_val();
out.push(load_capture(child_captures[k - 1].from, value));
value
};
for (k, source) in captures.iter().enumerate() {
let value = reg_fn.fresh_val();
out.push(load_capture(*source, value));
let dst = if k + 1 == captures.len() {
let dst = if k == child_captures.len() {
closure
} else {
reg_fn.fresh_val()
};
out.push(define_field(int_field(k as u32 + 1), dst, value, obj));
out.push(define_field(int_field(k as u32), dst, value, obj));
obj = dst;
}
}
@@ -288,11 +289,21 @@ fn plain_call(i: &RegInstr) -> Option<(OpCode, u16)> {
// ── emitted instructions ───────────────────────────────────────────────────
fn get_arg(dst: usize, arg: usize) -> RegInstr {
RegInstr::new(OpCode::GetArg, ImmOperand::Arg(arg as u16), vec![dst], vec![])
RegInstr::new(
OpCode::GetArg,
ImmOperand::Arg(arg as u16),
vec![dst],
vec![],
)
}
fn get_loc(dst: usize, loc: usize) -> RegInstr {
RegInstr::new(OpCode::GetLoc, ImmOperand::Loc(loc as u16), vec![dst], vec![])
RegInstr::new(
OpCode::GetLoc,
ImmOperand::Loc(loc as u16),
vec![dst],
vec![],
)
}
fn object(dst: usize) -> RegInstr {
+1 -1
View File
@@ -40,7 +40,7 @@ fn main() {
println!("{reg_fns:#?}");
println!("=== after declass ===");
let declassed = declass(reg_fns, &atoms);
let declassed = declass(reg_fns);
println!("{declassed:#?}");
println!("=== after declosure ===");
+2 -5
View File
@@ -2,9 +2,10 @@ use std::fmt::Debug;
use crate::{
flatten::{Capture, FlatFn},
parse::{AtomRef, ImmOperand, JsValue, opcodes::OpCode},
parse::{ImmOperand, JsValue, opcodes::OpCode},
};
#[derive(Clone)]
pub struct RegInstr {
pub dsts: Vec<usize>,
pub op: OpCode,
@@ -37,8 +38,6 @@ impl Debug for RegInstr {
pub struct RegFn {
pub name: String,
pub arg_count: usize,
pub var_count: usize,
pub value_names: Vec<AtomRef>,
pub captures: Vec<Capture>,
pub cpool: Vec<JsValue>,
pub code: Vec<RegInstr>,
@@ -98,8 +97,6 @@ impl RegFn {
Self {
name: flat_fn.name,
arg_count: flat_fn.arg_count,
var_count: flat_fn.var_count,
value_names: flat_fn.value_names,
captures: flat_fn.captures,
cpool: flat_fn.cpool,
code: output,