fix: 结构哈希补全全部变体(消除 code cache 确定性碰撞)

This commit is contained in:
2026-08-10 11:49:50 +08:00
parent 6a58f89e11
commit 7942269be8
6 changed files with 313 additions and 14 deletions
+9
View File
@@ -37,6 +37,15 @@ pub(super) fn hash_class_element(element: &ClassElement, h: &mut rustc_hash::FxH
statement::hash_statement(stmt, h, include_binding_names);
}
}
ClassElement::AccessorProperty(prop) => {
prop.r#static.hash(h);
prop.computed.hash(h);
property::hash_property_key(&prop.key, h, include_binding_names);
if let Some(value) = &prop.value {
expression::hash_expression(value, h, include_binding_names);
}
}
// TS 索引签名变体在 JS 模式(SourceType::unambiguous)下不可达,落入兜底。
_ => {}
});
}
@@ -154,11 +154,89 @@ pub(super) fn hash_expression(expr: &Expression, h: &mut rustc_hash::FxHasher, i
hash_chain_element(&chain.expression, h, include_binding_names);
}
Expression::Super(_) => {}
Expression::NullLiteral(_) => {
1u8.hash(h);
}
Expression::ThisExpression(_) => {
2u8.hash(h);
}
Expression::BigIntLiteral(big) => {
big.value.hash(h);
}
Expression::RegExpLiteral(lit) => {
if let Some(raw) = &lit.raw {
raw.to_string().hash(h);
}
}
Expression::ArrayExpression(arr) => {
(arr.elements.len() as u32).hash(h);
for element in &arr.elements {
match element {
ArrayExpressionElement::SpreadElement(spread) => {
hash_expression(&spread.argument, h, include_binding_names);
}
ArrayExpressionElement::Elision(_) => {
1u8.hash(h);
}
other => {
if let Some(expr) = other.as_expression() {
hash_expression(expr, h, include_binding_names);
}
}
}
}
}
Expression::MetaProperty(mp) => {
mp.meta.name.as_str().hash(h);
mp.property.name.as_str().hash(h);
}
Expression::AwaitExpression(ae) => {
hash_expression(&ae.argument, h, include_binding_names);
}
Expression::YieldExpression(ye) => {
ye.delegate.hash(h);
if let Some(arg) = &ye.argument {
hash_expression(arg, h, include_binding_names);
}
}
Expression::ImportExpression(ie) => {
hash_expression(&ie.source, h, include_binding_names);
if let Some(options) = &ie.options {
hash_expression(options, h, include_binding_names);
}
match &ie.phase {
Some(phase) => std::mem::discriminant(phase).hash(h),
None => 0u8.hash(h),
}
}
Expression::ParenthesizedExpression(p) => {
hash_expression(&p.expression, h, include_binding_names);
}
Expression::TSAsExpression(ts) => {
hash_expression(&ts.expression, h, include_binding_names);
}
Expression::TSSatisfiesExpression(ts) => {
hash_expression(&ts.expression, h, include_binding_names);
}
Expression::TSTypeAssertion(ts) => {
hash_expression(&ts.expression, h, include_binding_names);
}
Expression::TSNonNullExpression(ts) => {
hash_expression(&ts.expression, h, include_binding_names);
}
Expression::TSInstantiationExpression(ts) => {
hash_expression(&ts.expression, h, include_binding_names);
}
Expression::V8IntrinsicExpression(v8) => {
v8.name.name.as_str().hash(h);
(v8.arguments.len() as u32).hash(h);
for arg in &v8.arguments {
if let Some(expr) = arg.as_expression() {
hash_expression(expr, h, include_binding_names);
}
}
}
// JSX 变体在 JS 模式(SourceType::unambiguous)下不可达,落入兜底。
_ => {}
});
}
+3 -2
View File
@@ -5,8 +5,9 @@
//! `compiled_module_hash` 纳入绑定名(精确匹配避免错误复用)。
use oxide_parser::{
BindingPattern, ChainElement, ClassElement, Expression, ForStatementInit, ObjectPropertyKind, PropertyKey,
SimpleAssignmentTarget, Statement,
ArrayExpressionElement, BindingPattern, ChainElement, Class, ClassElement, Declaration, ExportDefaultDeclarationKind,
Expression, ForStatementInit, ForStatementLeft, Function, ImportDeclarationSpecifier, ModuleExportName,
ObjectPropertyKind, PropertyKey, SimpleAssignmentTarget, Statement,
};
use std::hash::Hash;
+119 -11
View File
@@ -53,9 +53,26 @@ pub(super) fn hash_statement(stmt: &Statement, h: &mut rustc_hash::FxHasher, inc
expression::hash_expression(&dw.test, h, include_binding_names);
}
Statement::ForInStatement(fi) => {
hash_for_statement_left(&fi.left, h, include_binding_names);
expression::hash_expression(&fi.right, h, include_binding_names);
hash_statement(&fi.body, h, include_binding_names);
}
Statement::ForOfStatement(fo) => {
fo.r#await.hash(h);
hash_for_statement_left(&fo.left, h, include_binding_names);
expression::hash_expression(&fo.right, h, include_binding_names);
hash_statement(&fo.body, h, include_binding_names);
}
Statement::WithStatement(ws) => {
expression::hash_expression(&ws.object, h, include_binding_names);
hash_statement(&ws.body, h, include_binding_names);
}
Statement::EmptyStatement(_) => {
1u8.hash(h);
}
Statement::DebuggerStatement(_) => {
2u8.hash(h);
}
Statement::SwitchStatement(sw) => {
expression::hash_expression(&sw.discriminant, h, include_binding_names);
for case in &sw.cases {
@@ -71,16 +88,7 @@ pub(super) fn hash_statement(stmt: &Statement, h: &mut rustc_hash::FxHasher, inc
hash_function_declaration(fd, h, include_binding_names);
}
Statement::ClassDeclaration(class) => {
if let Some(id) = &class.id {
id.name.as_str().hash(h);
}
class.super_class.is_some().hash(h);
if let Some(super_class) = &class.super_class {
expression::hash_expression(super_class, h, include_binding_names);
}
for element in &class.body.body {
class::hash_class_element(element, h, include_binding_names);
}
hash_class(class, h, include_binding_names);
}
Statement::ThrowStatement(ts) => {
expression::hash_expression(&ts.argument, h, include_binding_names);
@@ -88,6 +96,67 @@ pub(super) fn hash_statement(stmt: &Statement, h: &mut rustc_hash::FxHasher, inc
Statement::TryStatement(ts) => {
hash_try_statement(ts, h, include_binding_names);
}
Statement::ImportDeclaration(imp) => {
imp.source.value.hash(h);
if let Some(specifiers) = &imp.specifiers {
(specifiers.len() as u32).hash(h);
for spec in specifiers {
match spec {
ImportDeclarationSpecifier::ImportSpecifier(s) => {
hash_module_export_name(&s.imported, h);
if include_binding_names {
s.local.name.as_str().hash(h);
}
}
ImportDeclarationSpecifier::ImportDefaultSpecifier(s) => {
if include_binding_names {
s.local.name.as_str().hash(h);
}
}
ImportDeclarationSpecifier::ImportNamespaceSpecifier(s) => {
if include_binding_names {
s.local.name.as_str().hash(h);
}
}
}
}
}
}
Statement::ExportNamedDeclaration(exp) => {
if let Some(decl) = &exp.declaration {
hash_declaration(decl, h, include_binding_names);
}
(exp.specifiers.len() as u32).hash(h);
for spec in &exp.specifiers {
hash_module_export_name(&spec.local, h);
hash_module_export_name(&spec.exported, h);
}
if let Some(source) = &exp.source {
source.value.hash(h);
}
}
Statement::ExportDefaultDeclaration(exp) => {
match &exp.declaration {
ExportDefaultDeclarationKind::FunctionDeclaration(fd) => {
hash_function_declaration(fd, h, include_binding_names);
}
ExportDefaultDeclarationKind::ClassDeclaration(class) => {
hash_class(class, h, include_binding_names);
}
other => {
if let Some(expr) = other.as_expression() {
expression::hash_expression(expr, h, include_binding_names);
}
}
}
}
Statement::ExportAllDeclaration(exp) => {
if let Some(exported) = &exp.exported {
hash_module_export_name(exported, h);
}
exp.source.value.hash(h);
}
// TS 声明/JSX 等变体在 JS 模式(SourceType::unambiguous)下不可达,落入兜底。
_ => {}
});
}
@@ -125,7 +194,7 @@ fn hash_for_statement(fr: &oxide_parser::ForStatement<'_>, h: &mut rustc_hash::F
}
fn hash_function_declaration(
fd: &oxide_parser::Function<'_>, h: &mut rustc_hash::FxHasher, include_binding_names: bool,
fd: &Function<'_>, h: &mut rustc_hash::FxHasher, include_binding_names: bool,
) {
if include_binding_names {
if let Some(id) = &fd.id {
@@ -147,6 +216,45 @@ fn hash_function_declaration(
}
}
fn hash_class(class: &Class<'_>, h: &mut rustc_hash::FxHasher, include_binding_names: bool) {
if let Some(id) = &class.id {
id.name.as_str().hash(h);
}
class.super_class.is_some().hash(h);
if let Some(super_class) = &class.super_class {
expression::hash_expression(super_class, h, include_binding_names);
}
for element in &class.body.body {
class::hash_class_element(element, h, include_binding_names);
}
}
fn hash_declaration(decl: &Declaration<'_>, h: &mut rustc_hash::FxHasher, include_binding_names: bool) {
match decl {
Declaration::VariableDeclaration(vd) => hash_variable_declaration(vd, h, include_binding_names),
Declaration::FunctionDeclaration(fd) => hash_function_declaration(fd, h, include_binding_names),
Declaration::ClassDeclaration(class) => hash_class(class, h, include_binding_names),
_ => {}
}
}
fn hash_module_export_name(name: &ModuleExportName<'_>, h: &mut rustc_hash::FxHasher) {
match name {
ModuleExportName::IdentifierName(ident) => ident.name.as_str().hash(h),
ModuleExportName::IdentifierReference(ident) => ident.name.as_str().hash(h),
ModuleExportName::StringLiteral(s) => s.value.hash(h),
}
}
fn hash_for_statement_left(left: &ForStatementLeft<'_>, h: &mut rustc_hash::FxHasher, include_binding_names: bool) {
std::mem::discriminant(left).hash(h);
if let ForStatementLeft::VariableDeclaration(decl) = left {
hash_variable_declaration(decl, h, include_binding_names);
} else if let Some(target) = left.as_simple_assignment_target() {
target::hash_simple_assignment_target(target, h, include_binding_names);
}
}
fn hash_try_statement(ts: &oxide_parser::TryStatement<'_>, h: &mut rustc_hash::FxHasher, include_binding_names: bool) {
for s in &ts.block.body {
hash_statement(s, h, include_binding_names);
+12 -1
View File
@@ -22,6 +22,17 @@ pub(super) fn hash_simple_assignment_target(
expression::hash_expression(&member.object, h, include_binding_names);
member.field.name.as_str().hash(h);
}
_ => {}
SimpleAssignmentTarget::TSAsExpression(ts) => {
expression::hash_expression(&ts.expression, h, include_binding_names);
}
SimpleAssignmentTarget::TSSatisfiesExpression(ts) => {
expression::hash_expression(&ts.expression, h, include_binding_names);
}
SimpleAssignmentTarget::TSNonNullExpression(ts) => {
expression::hash_expression(&ts.expression, h, include_binding_names);
}
SimpleAssignmentTarget::TSTypeAssertion(ts) => {
expression::hash_expression(&ts.expression, h, include_binding_names);
}
});
}
@@ -0,0 +1,92 @@
//! 结构哈希确定性碰撞的回归测试:
//! 补全 hash 的 `_ => {}` 兜底漏掉的变体后,不同 AST 不得产生相同哈希。
use oxide_parser::Allocator;
fn compiled(source: &str) -> u64 {
let allocator = Allocator::default();
let program = oxide_parser::parse(&allocator, source).expect("parse failed");
oxide_compiler::compiler::compiled_module_hash(&program)
}
fn structural(source: &str) -> u64 {
let allocator = Allocator::default();
let program = oxide_parser::parse(&allocator, source).expect("parse failed");
oxide_compiler::compiler::structural_hash(&program)
}
#[test]
fn array_literals_distinguished() {
assert_ne!(compiled("[1, 2, 3]"), compiled("[4, 5, 6]"));
assert_ne!(compiled("[1, 2, 3]"), compiled("[1, , 3]"), "elision vs value must differ");
assert_ne!(compiled("[1, 2, ...a]"), compiled("[1, 2, 3]"), "spread vs value must differ");
assert_ne!(compiled("[...a]"), compiled("[...b]"), "different spread sources must differ");
}
#[test]
fn meta_property_distinguished() {
// `new.target` 是 MetaProperty,`a.b` 是普通成员访问,结构不同。
assert_ne!(compiled("function f(){ return new.target }"), compiled("function f(){ return a.b }"));
// 同为 MetaProperty 时 meta/property 名字必须参与哈希(修复前两者哈希相同)。
assert_ne!(
compiled("import './x.js'; function f(){ return import.meta }"),
compiled("import './x.js'; function f(){ return new.target }"),
"meta/property names must be part of the hash"
);
}
#[test]
fn parenthesized_distinguished() {
assert_ne!(compiled("(a)"), compiled("a"), "parenthesized vs bare expression must differ");
assert_ne!(compiled("(a + b)"), compiled("a + b"));
assert_ne!(compiled("(a)"), compiled("(b)"), "inner expression must be hashed");
}
#[test]
fn await_yield_plain_distinguished() {
let await_h = compiled("async function f(){ await x }");
let yield_h = compiled("function* f(){ yield x }");
let plain_h = compiled("function f(){ x }");
assert_ne!(await_h, yield_h);
assert_ne!(await_h, plain_h);
assert_ne!(yield_h, plain_h);
// yield 的 delegate 标志与参数必须参与哈希。
assert_ne!(compiled("function* f(){ yield x }"), compiled("function* f(){ yield* x }"));
assert_ne!(compiled("function* f(){ yield x }"), compiled("function* f(){ yield y }"));
}
#[test]
fn object_literals_distinguished() {
assert_ne!(compiled("var o = { a: 1 }"), compiled("var o = { a: 2 }"));
assert_ne!(compiled("var o = { a: 1 }"), compiled("var o = { b: 1 }"));
assert_ne!(compiled("var o = { a: 1 }"), compiled("var o = { a: 1, b: 2 }"), "property count must be hashed");
}
#[test]
fn function_statement_hash_distinguished() {
assert_ne!(compiled("function f(){ return 1 }"), compiled("function f(){ return 2 }"));
assert_ne!(compiled("function f(){ return 1 }"), compiled("function g(){ return 1 }"));
}
#[test]
fn for_of_with_distinguished() {
assert_ne!(compiled("for (x of xs) { body(x) }"), compiled("for (x of ys) { body(x) }"));
assert_ne!(compiled("for (x of xs) { body(x) }"), compiled("for (x in xs) { body(x) }"));
assert_ne!(compiled("with (a) { f() }"), compiled("with (b) { f() }"));
}
#[test]
fn import_expression_distinguished() {
assert_ne!(compiled("import('./a.js')"), compiled("import('./b.js')"));
assert_ne!(compiled("import('./a.js')"), compiled("import('./a.js', { with: { type: 'json' } })"));
}
#[test]
fn structural_hash_ignores_binding_names() {
// 结构哈希忽略绑定名:仅参数/变量改名后哈希相等。
assert_eq!(structural("function f(x){ return x }"), structural("function f(y){ return y }"));
assert_eq!(structural("var a = 1; a + a"), structural("var b = 1; b + b"));
// 编译模块哈希纳入绑定名:改名的程序哈希不等。
assert_ne!(compiled("function f(x){ return x }"), compiled("function f(y){ return y }"));
}