feat: init Rust VM
This commit is contained in:
@@ -1 +1,531 @@
|
||||
use std::hash::BuildHasher;
|
||||
|
||||
use bumpalo::Bump;
|
||||
use gc_arena::{Arena, Rootable, arena::CollectionPhase};
|
||||
use ghost_cell::{GhostCell, GhostToken};
|
||||
use hashbrown::{DefaultHashBuilder, HashMap, HashSet, HashTable};
|
||||
use rnix::TextRange;
|
||||
use string_interner::DefaultStringInterner;
|
||||
|
||||
use crate::codegen::{BytecodeContext, InstructionPtr};
|
||||
use crate::downgrade::{Downgrade as _, DowngradeContext};
|
||||
use crate::error::{Error, Result, Source};
|
||||
use crate::ir::{ArgId, Ir, IrKey, IrRef, RawIrRef, StringId, ThunkId, ir_content_eq};
|
||||
use crate::runtime::builtins::new_builtins_env;
|
||||
use crate::store::{DaemonStore, StoreConfig};
|
||||
use crate::value::Symbol;
|
||||
|
||||
mod builtins;
|
||||
mod stack;
|
||||
mod value;
|
||||
mod vm;
|
||||
use vm::{Action, VM};
|
||||
|
||||
pub struct Runtime {
|
||||
bytecode: Vec<u8>,
|
||||
global_env: HashMap<StringId, Ir<'static, RawIrRef<'static>>>,
|
||||
sources: Vec<Source>,
|
||||
store: DaemonStore,
|
||||
spans: Vec<(usize, TextRange)>,
|
||||
thunk_count: usize,
|
||||
strings: DefaultStringInterner,
|
||||
arena: Arena<Rootable![VM<'_>]>,
|
||||
}
|
||||
|
||||
impl Runtime {
|
||||
const COLLECTOR_GRANULARITY: f64 = 1024.0;
|
||||
|
||||
pub fn new() -> Result<Self> {
|
||||
let mut strings = DefaultStringInterner::new();
|
||||
let global_env = new_builtins_env(&mut strings);
|
||||
|
||||
let config = StoreConfig::from_env();
|
||||
let store = DaemonStore::connect(&config.daemon_socket)?;
|
||||
|
||||
Ok(Self {
|
||||
global_env,
|
||||
store,
|
||||
strings,
|
||||
thunk_count: 0,
|
||||
bytecode: Vec::new(),
|
||||
sources: Vec::new(),
|
||||
spans: Vec::new(),
|
||||
arena: Arena::new(|mc| VM::new(mc)),
|
||||
})
|
||||
}
|
||||
|
||||
pub fn eval(&mut self, source: Source) -> Result<crate::value::Value> {
|
||||
let root = self.downgrade(source, None)?;
|
||||
let ip = crate::codegen::compile_bytecode(root.as_ref(), self);
|
||||
self.run(ip)
|
||||
}
|
||||
|
||||
pub fn eval_shallow(&mut self, _source: Source) -> Result<crate::value::Value> {
|
||||
todo!()
|
||||
}
|
||||
|
||||
pub fn eval_deep(&mut self, source: Source) -> Result<crate::value::Value> {
|
||||
// FIXME: deep
|
||||
let root = self.downgrade(source, None)?;
|
||||
let ip = crate::codegen::compile_bytecode(root.as_ref(), self);
|
||||
self.run(ip)
|
||||
}
|
||||
|
||||
pub fn eval_repl(
|
||||
&mut self,
|
||||
source: Source,
|
||||
scope: &HashSet<StringId>,
|
||||
) -> Result<crate::value::Value> {
|
||||
// FIXME: shallow
|
||||
let root = self.downgrade(source, Some(Scope::Repl(scope)))?;
|
||||
let ip = crate::codegen::compile_bytecode(root.as_ref(), self);
|
||||
self.run(ip)
|
||||
}
|
||||
|
||||
pub fn add_binding(
|
||||
&mut self,
|
||||
_ident: &str,
|
||||
_expr: &str,
|
||||
_scope: &mut HashSet<StringId>,
|
||||
) -> Result<crate::value::Value> {
|
||||
todo!()
|
||||
}
|
||||
|
||||
fn downgrade_ctx<'a, 'bump, 'id>(
|
||||
&'a mut self,
|
||||
bump: &'bump Bump,
|
||||
token: GhostToken<'id>,
|
||||
extra_scope: Option<Scope<'a>>,
|
||||
) -> DowngradeCtx<'a, 'id, 'bump> {
|
||||
let Runtime {
|
||||
global_env,
|
||||
sources,
|
||||
thunk_count,
|
||||
strings,
|
||||
..
|
||||
} = self;
|
||||
DowngradeCtx {
|
||||
bump,
|
||||
token,
|
||||
strings,
|
||||
source: sources.last().unwrap().clone(),
|
||||
scopes: [Scope::Global(global_env)].into_iter().chain(extra_scope.into_iter()).collect(),
|
||||
with_scope_count: 0,
|
||||
arg_count: 0,
|
||||
thunk_count,
|
||||
thunk_scopes: vec![ThunkScope::new_in(bump)],
|
||||
}
|
||||
}
|
||||
|
||||
fn downgrade<'a>(&'a mut self, source: Source, extra_scope: Option<Scope<'a>>) -> Result<OwnedIr> {
|
||||
tracing::debug!("Parsing Nix expression");
|
||||
|
||||
self.sources.push(source.clone());
|
||||
|
||||
let root = rnix::Root::parse(&source.src);
|
||||
handle_parse_error(root.errors(), source).map_or(Ok(()), Err)?;
|
||||
|
||||
tracing::debug!("Downgrading Nix expression");
|
||||
let expr = root
|
||||
.tree()
|
||||
.expr()
|
||||
.ok_or_else(|| Error::parse_error("unexpected EOF".into()))?;
|
||||
let bump = Bump::new();
|
||||
GhostToken::new(|token| {
|
||||
let ir = self
|
||||
.downgrade_ctx(&bump, token, extra_scope)
|
||||
.downgrade_toplevel(expr)?;
|
||||
let ir = unsafe { std::mem::transmute::<RawIrRef<'_>, RawIrRef<'static>>(ir) };
|
||||
Ok(OwnedIr { _bump: bump, ir })
|
||||
})
|
||||
}
|
||||
|
||||
fn run(&mut self, ip: InstructionPtr) -> Result<crate::value::Value> {
|
||||
let mut pc = ip.0;
|
||||
loop {
|
||||
let Runtime {
|
||||
bytecode,
|
||||
strings,
|
||||
arena,
|
||||
..
|
||||
} = self;
|
||||
let action =
|
||||
arena.mutate_root(|mc, root| root.run_batch(bytecode, &mut pc, mc, strings));
|
||||
match action {
|
||||
Action::NeedGc => {
|
||||
if self.arena.collection_phase() == CollectionPhase::Sweeping {
|
||||
self.arena.collect_debt();
|
||||
} else if let Some(marked) = self.arena.mark_debt() {
|
||||
marked.start_sweeping();
|
||||
}
|
||||
}
|
||||
Action::Done(done) => {
|
||||
break done;
|
||||
}
|
||||
Action::Continue => (),
|
||||
Action::IoRequest(_) => todo!(),
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
fn parse_error_span(error: &rnix::ParseError) -> Option<rnix::TextRange> {
|
||||
use rnix::ParseError::*;
|
||||
match error {
|
||||
Unexpected(range)
|
||||
| UnexpectedExtra(range)
|
||||
| UnexpectedWanted(_, range, _)
|
||||
| UnexpectedDoubleBind(range)
|
||||
| DuplicatedArgs(range, _) => Some(*range),
|
||||
_ => None,
|
||||
}
|
||||
}
|
||||
|
||||
fn handle_parse_error<'a>(
|
||||
errors: impl IntoIterator<Item = &'a rnix::ParseError>,
|
||||
source: Source,
|
||||
) -> Option<Box<Error>> {
|
||||
for err in errors {
|
||||
if let Some(span) = parse_error_span(err) {
|
||||
return Some(
|
||||
Error::parse_error(err.to_string())
|
||||
.with_source(source)
|
||||
.with_span(span),
|
||||
);
|
||||
}
|
||||
}
|
||||
None
|
||||
}
|
||||
|
||||
struct DowngradeCtx<'ctx, 'id, 'ir> {
|
||||
bump: &'ir Bump,
|
||||
token: GhostToken<'id>,
|
||||
strings: &'ctx mut DefaultStringInterner,
|
||||
source: Source,
|
||||
scopes: Vec<Scope<'ctx>>,
|
||||
with_scope_count: u32,
|
||||
arg_count: u32,
|
||||
thunk_count: &'ctx mut usize,
|
||||
thunk_scopes: Vec<ThunkScope<'id, 'ir>>,
|
||||
}
|
||||
|
||||
fn should_thunk<'id>(ir: IrRef<'id, '_>, token: &GhostToken<'id>) -> bool {
|
||||
!matches!(
|
||||
ir.borrow(token),
|
||||
Ir::Builtin(_)
|
||||
| Ir::Builtins
|
||||
| Ir::Int(_)
|
||||
| Ir::Float(_)
|
||||
| Ir::Bool(_)
|
||||
| Ir::Null
|
||||
| Ir::Str(_)
|
||||
| Ir::Thunk(_)
|
||||
)
|
||||
}
|
||||
|
||||
impl<'ctx, 'id, 'ir> DowngradeCtx<'ctx, 'id, 'ir> {
|
||||
fn new(
|
||||
bump: &'ir Bump,
|
||||
token: GhostToken<'id>,
|
||||
symbols: &'ctx mut DefaultStringInterner,
|
||||
global: &'ctx HashMap<StringId, Ir<'static, RawIrRef<'static>>>,
|
||||
extra_scope: Option<Scope<'ctx>>,
|
||||
thunk_count: &'ctx mut usize,
|
||||
source: Source,
|
||||
) -> Self {
|
||||
Self {
|
||||
bump,
|
||||
token,
|
||||
strings: symbols,
|
||||
source,
|
||||
scopes: std::iter::once(Scope::Global(global))
|
||||
.chain(extra_scope)
|
||||
.collect(),
|
||||
thunk_count,
|
||||
arg_count: 0,
|
||||
with_scope_count: 0,
|
||||
thunk_scopes: vec![ThunkScope::new_in(bump)],
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
impl<'ctx: 'ir, 'id, 'ir> DowngradeContext<'id, 'ir> for DowngradeCtx<'ctx, 'id, 'ir> {
|
||||
fn new_expr(&self, expr: Ir<'ir, IrRef<'id, 'ir>>) -> IrRef<'id, 'ir> {
|
||||
IrRef::new(self.bump.alloc(GhostCell::new(expr)))
|
||||
}
|
||||
|
||||
fn new_arg(&mut self) -> ArgId {
|
||||
self.arg_count += 1;
|
||||
ArgId(self.arg_count - 1)
|
||||
}
|
||||
|
||||
fn maybe_thunk(&mut self, ir: IrRef<'id, 'ir>) -> IrRef<'id, 'ir> {
|
||||
if !should_thunk(ir, &self.token) {
|
||||
return ir;
|
||||
}
|
||||
|
||||
let cached = self
|
||||
.thunk_scopes
|
||||
.last()
|
||||
.expect("no active cache scope")
|
||||
.lookup_cache(ir, &self.token);
|
||||
|
||||
if let Some(id) = cached {
|
||||
return IrRef::alloc(self.bump, Ir::Thunk(id));
|
||||
}
|
||||
|
||||
let id = ThunkId(*self.thunk_count);
|
||||
*self.thunk_count = self.thunk_count.checked_add(1).expect("thunk id overflow");
|
||||
self.thunk_scopes
|
||||
.last_mut()
|
||||
.expect("no active cache scope")
|
||||
.add_binding(id, ir, &self.token);
|
||||
IrRef::alloc(self.bump, Ir::Thunk(id))
|
||||
}
|
||||
|
||||
fn new_sym(&mut self, sym: String) -> StringId {
|
||||
StringId(self.strings.get_or_intern(sym))
|
||||
}
|
||||
|
||||
fn get_sym(&self, id: StringId) -> Symbol<'_> {
|
||||
self.strings.resolve(id.0).expect("no symbol found").into()
|
||||
}
|
||||
|
||||
fn lookup(&self, sym: StringId, span: TextRange) -> Result<IrRef<'id, 'ir>> {
|
||||
for scope in self.scopes.iter().rev() {
|
||||
match scope {
|
||||
&Scope::Global(global_scope) => {
|
||||
if let Some(expr) = global_scope.get(&sym) {
|
||||
let ir = match expr {
|
||||
Ir::Builtins => Ir::Builtins,
|
||||
Ir::Builtin(s) => Ir::Builtin(*s),
|
||||
Ir::Bool(b) => Ir::Bool(*b),
|
||||
Ir::Null => Ir::Null,
|
||||
_ => unreachable!("globals should only contain leaf IR nodes"),
|
||||
};
|
||||
return Ok(self.new_expr(ir));
|
||||
}
|
||||
}
|
||||
&Scope::Repl(repl_bindings) => {
|
||||
if repl_bindings.contains(&sym) {
|
||||
return Ok(self.new_expr(Ir::ReplBinding(sym)));
|
||||
}
|
||||
}
|
||||
Scope::ScopedImport(scoped_bindings) => {
|
||||
if scoped_bindings.contains(&sym) {
|
||||
return Ok(self.new_expr(Ir::ScopedImportBinding(sym)));
|
||||
}
|
||||
}
|
||||
Scope::Let(let_scope) => {
|
||||
if let Some(&expr) = let_scope.get(&sym) {
|
||||
return Ok(self.new_expr(Ir::Thunk(expr)));
|
||||
}
|
||||
}
|
||||
&Scope::Param(param_sym, id) => {
|
||||
if param_sym == sym {
|
||||
return Ok(self.new_expr(Ir::Arg(id)));
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
if self.with_scope_count > 0 {
|
||||
Ok(self.new_expr(Ir::WithLookup(sym)))
|
||||
} else {
|
||||
Err(Error::downgrade_error(
|
||||
format!("'{}' not found", self.get_sym(sym)),
|
||||
self.get_current_source(),
|
||||
span,
|
||||
))
|
||||
}
|
||||
}
|
||||
|
||||
fn get_current_source(&self) -> Source {
|
||||
self.source.clone()
|
||||
}
|
||||
|
||||
fn with_let_scope<F, R>(&mut self, keys: &[StringId], f: F) -> Result<R>
|
||||
where
|
||||
F: FnOnce(&mut Self) -> Result<(bumpalo::collections::Vec<'ir, IrRef<'id, 'ir>>, R)>,
|
||||
{
|
||||
let base = *self.thunk_count;
|
||||
*self.thunk_count = self
|
||||
.thunk_count
|
||||
.checked_add(keys.len())
|
||||
.expect("thunk id overflow");
|
||||
let iter = keys.iter().enumerate().map(|(offset, &key)| {
|
||||
(
|
||||
key,
|
||||
ThunkId(unsafe { base.checked_add(offset).unwrap_unchecked() }),
|
||||
)
|
||||
});
|
||||
self.scopes.push(Scope::Let(iter.collect()));
|
||||
let (vals, ret) = {
|
||||
let mut guard = ScopeGuard { ctx: self };
|
||||
f(guard.as_ctx())?
|
||||
};
|
||||
assert_eq!(keys.len(), vals.len());
|
||||
let scope = self.thunk_scopes.last_mut().expect("no active thunk scope");
|
||||
scope.extend_bindings((base..base + keys.len()).map(ThunkId).zip(vals));
|
||||
Ok(ret)
|
||||
}
|
||||
|
||||
fn with_param_scope<F, R>(&mut self, param: StringId, arg: ArgId, f: F) -> R
|
||||
where
|
||||
F: FnOnce(&mut Self) -> R,
|
||||
{
|
||||
self.scopes.push(Scope::Param(param, arg));
|
||||
let mut guard = ScopeGuard { ctx: self };
|
||||
f(guard.as_ctx())
|
||||
}
|
||||
|
||||
fn with_with_scope<F, R>(&mut self, f: F) -> R
|
||||
where
|
||||
F: FnOnce(&mut Self) -> R,
|
||||
{
|
||||
self.with_scope_count += 1;
|
||||
let ret = f(self);
|
||||
self.with_scope_count -= 1;
|
||||
ret
|
||||
}
|
||||
|
||||
fn with_thunk_scope<F, R>(
|
||||
&mut self,
|
||||
f: F,
|
||||
) -> (
|
||||
R,
|
||||
bumpalo::collections::Vec<'ir, (ThunkId, IrRef<'id, 'ir>)>,
|
||||
)
|
||||
where
|
||||
F: FnOnce(&mut Self) -> R,
|
||||
{
|
||||
self.thunk_scopes.push(ThunkScope::new_in(self.bump));
|
||||
let ret = f(self);
|
||||
(
|
||||
ret,
|
||||
self.thunk_scopes
|
||||
.pop()
|
||||
.expect("no thunk scope left???")
|
||||
.bindings,
|
||||
)
|
||||
}
|
||||
|
||||
fn bump(&self) -> &'ir bumpalo::Bump {
|
||||
self.bump
|
||||
}
|
||||
}
|
||||
|
||||
impl<'id, 'ir, 'ctx: 'ir> DowngradeCtx<'ctx, 'id, 'ir> {
|
||||
fn downgrade_toplevel(mut self, root: rnix::ast::Expr) -> Result<RawIrRef<'ir>> {
|
||||
let body = root.downgrade(&mut self)?;
|
||||
let thunks = self
|
||||
.thunk_scopes
|
||||
.pop()
|
||||
.expect("no thunk scope left???")
|
||||
.bindings;
|
||||
let ir = IrRef::alloc(self.bump, Ir::TopLevel { body, thunks });
|
||||
Ok(ir.freeze(self.token))
|
||||
}
|
||||
}
|
||||
|
||||
struct ThunkScope<'id, 'ir> {
|
||||
bindings: bumpalo::collections::Vec<'ir, (ThunkId, IrRef<'id, 'ir>)>,
|
||||
cache: HashTable<(IrRef<'id, 'ir>, ThunkId)>,
|
||||
hasher: DefaultHashBuilder,
|
||||
}
|
||||
|
||||
impl<'id, 'ir> ThunkScope<'id, 'ir> {
|
||||
fn new_in(bump: &'ir Bump) -> Self {
|
||||
Self {
|
||||
bindings: bumpalo::collections::Vec::new_in(bump),
|
||||
cache: HashTable::new(),
|
||||
hasher: DefaultHashBuilder::default(),
|
||||
}
|
||||
}
|
||||
|
||||
fn lookup_cache(&self, key: IrRef<'id, 'ir>, token: &GhostToken<'id>) -> Option<ThunkId> {
|
||||
let hash = self.hasher.hash_one(IrKey(key, token));
|
||||
self.cache
|
||||
.find(hash, |&(ir, _)| ir_content_eq(key, ir, token))
|
||||
.map(|&(_, id)| id)
|
||||
}
|
||||
|
||||
fn add_binding(&mut self, id: ThunkId, ir: IrRef<'id, 'ir>, token: &GhostToken<'id>) {
|
||||
self.bindings.push((id, ir));
|
||||
let hash = self.hasher.hash_one(IrKey(ir, token));
|
||||
self.cache.insert_unique(hash, (ir, id), |&(ir, _)| {
|
||||
self.hasher.hash_one(IrKey(ir, token))
|
||||
});
|
||||
}
|
||||
|
||||
fn extend_bindings(&mut self, iter: impl IntoIterator<Item = (ThunkId, IrRef<'id, 'ir>)>) {
|
||||
self.bindings.extend(iter);
|
||||
}
|
||||
}
|
||||
|
||||
enum Scope<'ctx> {
|
||||
Global(&'ctx HashMap<StringId, Ir<'static, RawIrRef<'static>>>),
|
||||
Repl(&'ctx HashSet<StringId>),
|
||||
ScopedImport(HashSet<StringId>),
|
||||
Let(HashMap<StringId, ThunkId>),
|
||||
Param(StringId, ArgId),
|
||||
}
|
||||
|
||||
struct ScopeGuard<'a, 'ctx, 'id, 'ir> {
|
||||
ctx: &'a mut DowngradeCtx<'ctx, 'id, 'ir>,
|
||||
}
|
||||
|
||||
impl Drop for ScopeGuard<'_, '_, '_, '_> {
|
||||
fn drop(&mut self) {
|
||||
self.ctx.scopes.pop();
|
||||
}
|
||||
}
|
||||
|
||||
impl<'id, 'ir, 'ctx> ScopeGuard<'_, 'ctx, 'id, 'ir> {
|
||||
fn as_ctx(&mut self) -> &mut DowngradeCtx<'ctx, 'id, 'ir> {
|
||||
self.ctx
|
||||
}
|
||||
}
|
||||
|
||||
struct OwnedIr {
|
||||
_bump: Bump,
|
||||
ir: RawIrRef<'static>,
|
||||
}
|
||||
|
||||
impl OwnedIr {
|
||||
unsafe fn new(ir: RawIrRef<'_>, bump: Bump) -> Self {
|
||||
Self {
|
||||
_bump: bump,
|
||||
ir: unsafe { std::mem::transmute::<RawIrRef<'_>, RawIrRef<'static>>(ir) }
|
||||
}
|
||||
}
|
||||
|
||||
fn as_ref(&self) -> RawIrRef<'_> {
|
||||
self.ir
|
||||
}
|
||||
}
|
||||
|
||||
impl BytecodeContext for Runtime {
|
||||
fn intern_string(&mut self, s: &str) -> StringId {
|
||||
StringId(self.strings.get_or_intern(s))
|
||||
}
|
||||
|
||||
fn register_span(&mut self, range: TextRange) -> u32 {
|
||||
let id = self.spans.len();
|
||||
let source_id = self
|
||||
.sources
|
||||
.len()
|
||||
.checked_sub(1)
|
||||
.expect("current_source not set");
|
||||
self.spans.push((source_id, range));
|
||||
id as u32
|
||||
}
|
||||
|
||||
fn get_code(&self) -> &[u8] {
|
||||
&self.bytecode
|
||||
}
|
||||
|
||||
fn get_code_mut(&mut self) -> &mut Vec<u8> {
|
||||
&mut self.bytecode
|
||||
}
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user