sander/java.rs
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

broke through the wall. most unsafe gone. internal mutability gone. better design for classloading

Browse source
This commit is contained in:
Shautvast 2023-11-05 07:51:47 +01:00
parent fa10621c56
commit ceeec215e4
17 changed files with 1335 additions and 1302 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

BIN
jmods/java.base.jar
View file

Binary file not shown.

775
src/class.rs
View file

@ -1,126 +1,10 @@
use std::cell::{RefCell, UnsafeCell}; use std::collections::{HashMap, LinkedList};
use std::collections::HashMap;
use std::fmt;
use std::rc::Rc;
use std::sync::Arc;
use anyhow::Error; use crate::class::ObjectRef::*;
use log::info;
use once_cell::sync::Lazy;
use crate::classloader::{CpEntry, load_class}; pub type ClassId = usize;
use crate::heap::{ObjectRef};
use crate::io::{find_class, read_bytecode, read_u16};
use crate::vm::Vm;
//trying to be ready for multithreaded as much as possible, using Arc's and all, but it will still require (a lot of) extra work #[derive(Debug, Clone)]
pub static mut CLASSDEFS: Lazy<HashMap<String, Arc<RefCell<Class>>>> = Lazy::new(|| HashMap::new());
//TODO add mutex..
pub static mut CLASSES: Lazy<HashMap<String, Value>> = Lazy::new(|| HashMap::new()); //TODO add mutex..
// gets the Class from cache, or reads it from classpath,
// then parses the binary data into a Class struct
// Vm keeps ownership of the class and hands out Arc references to it
pub(crate) fn get_class(
vm: &mut Vm,
class_name: &str,
) -> Result<Arc<RefCell<Class>>, Error> {
info!("get_class {}", class_name);
unsafe {
let class = CLASSDEFS.entry(class_name.into()).or_insert_with(|| {
// println!("read class {} ", class_name);
let resolved_path = find_class(&vm.classpath, class_name).unwrap();
let bytecode = read_bytecode(resolved_path).unwrap();
let class = load_class(bytecode).unwrap();
Arc::new(RefCell::new(class))
});
let clone = class.clone();
let inited = class.borrow().inited;
if !inited {
// not sure why I have to create the clones first
let clone2 = class.clone();
let clone3 = class.clone();
let clone4 = class.clone();
let mut some_class = class.clone();
if class_name != "java/lang/Class" {
let klazz = get_class(vm, "java/lang/Class")?;
let mut class_instance = Vm::new_instance(klazz);
class_instance.set(&"java/lang/Class".to_owned(), &"name".to_owned(), Value::Utf8(class_name.into()));
CLASSES.insert(class_name.into(), Value::Ref(unsafe_ref(ObjectRef::Object(Box::new(class_instance)))));
}
// must not enter here twice!
clone2.borrow_mut().inited = true;
let mut supers = vec![];
if class_name != "java/lang/Class" {
loop {
let super_class_name = some_class
.clone()
.borrow()
.super_class_name
.as_ref()
.map(|n| n.to_owned());
{
if let Some(super_class_name) = super_class_name {
if let Ok(super_class) = get_class(vm, &super_class_name) {
supers.push(super_class.clone());
some_class = super_class.clone();
clone4.borrow_mut().super_class = Some(super_class);
} else {
break;
}
} else {
break;
}
}
}
}
Class::initialize_fields(clone3, supers);
let clinit = clone2.borrow().methods.contains_key("<clinit>()V");
let name = &clone2.borrow().name.to_owned();
if clinit {
vm.execute_special(name, "<clinit>()V", vec![]).unwrap();
}
}
Ok(clone)
}
}
/// the class definition as read from the class file + derived values
// TODO implement call to static initializers
#[derive(Debug)]
pub struct Class {
pub minor_version: u16,
pub major_version: u16,
pub constant_pool: Rc<HashMap<u16, CpEntry>>,
pub access_flags: u16,
pub name: String,
pub super_class_name: Option<String>,
pub super_class: Option<Type>,
pub super_classes: Vec<Type>,
pub interface_indices: Vec<u16>,
pub interfaces: Vec<Class>,
pub fields: HashMap<String, Field>,
pub methods: HashMap<String, Rc<Method>>,
pub attributes: HashMap<String, AttributeType>,
pub inited: bool,
// lookup index and type from the name
pub(crate) object_field_mapping: HashMap<String, HashMap<String, TypeIndex>>,
pub(crate) static_field_mapping: HashMap<String, HashMap<String, TypeIndex>>,
// static fields
pub(crate) static_data: Vec<Value>,
}
#[derive(Debug)]
pub(crate) struct TypeIndex { pub(crate) struct TypeIndex {
pub type_name: String, pub type_name: String,
pub index: usize, pub index: usize,
@ -135,43 +19,23 @@ impl TypeIndex {
} }
} }
// could move classdef into here. Saves one hashmap lookup
// have to look at ownership though
#[derive(Debug, Clone)]
pub struct Class {
pub id: ClassId,
pub initialized: bool,
pub name: String,
pub superclass: Option<ClassId>,
pub parents: LinkedList<ClassId>,
pub interfaces: Vec<ClassId>,
// lookup index and type from the name
pub(crate) object_field_mapping: HashMap<String, HashMap<String, TypeIndex>>,
pub(crate) static_field_mapping: HashMap<String, HashMap<String, TypeIndex>>,
// pub(crate) static_field_data: Vec<Value>,
}
impl Class { impl Class {
pub fn new(
minor_version: u16,
major_version: u16,
constant_pool: Rc<HashMap<u16, CpEntry>>,
access_flags: u16,
this_class: u16,
super_class_index: u16,
interface_indices: Vec<u16>,
fields: HashMap<String, Field>,
methods: HashMap<String, Rc<Method>>,
attributes: HashMap<String, AttributeType>,
) -> Self {
let name = Class::class_name(this_class, constant_pool.clone()).unwrap();
let super_class_name = Class::class_name(super_class_index, constant_pool.clone());
Self {
major_version,
minor_version,
constant_pool,
access_flags,
name,
super_class_name,
super_class: None, // has to be instantiated later, because it involves classloading. maybe not store it here
super_classes: vec![],
interface_indices,
interfaces: vec![], // same
fields,
methods,
attributes,
inited: false,
object_field_mapping: HashMap::new(),
static_field_mapping: HashMap::new(),
static_data: vec![],
}
}
pub(crate) fn n_object_fields(&self) -> usize { pub(crate) fn n_object_fields(&self) -> usize {
self.object_field_mapping self.object_field_mapping
.iter() .iter()
@ -179,448 +43,8 @@ impl Class {
.reduce(|acc, e| acc + e) .reduce(|acc, e| acc + e)
.unwrap() .unwrap()
} }
pub(crate) fn n_static_fields(&self) -> usize {
self.static_field_mapping
.iter()
.map(|(_, v)| v.len())
.reduce(|acc, e| acc + e)
.unwrap()
}
// Create a mapping per field(name) to an index in the storage vector that contains the instance data.
// When a field is stored, first the index will be looked up, using the qualified name (from the FieldRef)
// The qualified name is the combination of class name and field name.
// The class name is needed as part of the key to separate class from superclass fields
// (duplicates in the singular field name are allowed).
// This way `this.a` can be differentiated from `super.a`.
//
// this method looks up this and super classes and calls map_fields for each.
pub fn initialize_fields(class: Arc<RefCell<Class>>, super_classes: Vec<Arc<RefCell<Class>>>) {
let mut this_field_mapping = HashMap::new();
let mut static_field_mapping = HashMap::new();
let mut object_field_map_index: usize = 0;
let mut static_field_map_index: usize = 0;
Class::add_field_mappings(
&mut this_field_mapping,
&mut static_field_mapping,
class.clone(),
&mut object_field_map_index,
&mut static_field_map_index,
);
class.borrow_mut().object_field_mapping = this_field_mapping;
class.borrow_mut().static_field_mapping = static_field_mapping;
let static_data = Class::set_field_data(class.clone());
class.borrow_mut().static_data = static_data;
}
/// for all static and non-static fields on the class compute an index
/// the result of this function is that the class object contains mappings
/// from the field name to the index. This index will be used to store the
/// actual data later in a Vector.
fn add_field_mappings(
this_field_mapping: &mut HashMap<String, HashMap<String, TypeIndex>>,
static_field_mapping: &mut HashMap<String, HashMap<String, TypeIndex>>,
class: Arc<RefCell<Class>>,
object_field_map_index: &mut usize,
static_field_map_index: &mut usize,
) {
let (o, s) = Class::map_fields(
class.clone(),
object_field_map_index,
static_field_map_index,
);
let borrow = class.borrow();
let name = &borrow.name;
this_field_mapping.insert(name.to_owned(), o);
static_field_mapping.insert(name.to_owned(), s);
// // same for super class
// if let Some(super_class) = class.borrow().super_class.as_ref() {
// Class::add_field_mappings(
// this_field_mapping,
// static_field_mapping,
// super_class.clone(),
// object_field_map_index,
// static_field_map_index,
// );
// }
for c in &class.borrow().super_classes {
Class::add_field_mappings(
this_field_mapping,
static_field_mapping,
c.clone(),
object_field_map_index,
static_field_map_index,
);
}
}
// part of the initialize procedure
/// here the actual indices are created
fn map_fields(
class: Arc<RefCell<Class>>,
object_field_map_index: &mut usize,
static_field_map_index: &mut usize,
) -> (
HashMap<String, TypeIndex>,
HashMap<String, TypeIndex>,
) {
let mut this_fields = HashMap::new(); //fields in class are stored per class and every superclass.
let mut static_fields = HashMap::new(); //fields in class are stored per class and every superclass.
for (name, field) in &class.borrow().fields {
if field.is(Modifier::Static) {
static_fields.insert(
name.to_owned(),
TypeIndex::new(field.type_of().to_owned(), *static_field_map_index),
);
*static_field_map_index += 1;
} else {
this_fields.insert(
name.to_owned(),
TypeIndex::new(field.type_of().to_owned(), *object_field_map_index),
); //name => (type,index)
*object_field_map_index += 1;
}
}
(this_fields, static_fields)
}
/// the bytecode version
pub fn get_version(&self) -> (u16, u16) {
(self.major_version, self.minor_version)
}
/// get a method by signature
pub fn get_method(&self, name: &str) -> Option<&Rc<Method>> {
self.methods.get(name)
}
/// get the class name
fn class_name(
super_class_index: u16,
constant_pool: Rc<HashMap<u16, CpEntry>>,
) -> Option<String> {
if super_class_index == 0 {
None
} else if let CpEntry::ClassRef(name_index) = constant_pool.get(&super_class_index).unwrap()
{
if let CpEntry::Utf8(name) = constant_pool.get(name_index).unwrap() {
Some(name.to_owned())
} else {
None
}
} else {
None
}
}
/// creates default values for every field, ie null for objects, 0 for integers etc
/// this is the step before the constructor/static initializer can be called to set hardcoded
/// or computed values.
pub(crate) fn set_field_data(class: Arc<RefCell<Class>>) -> Vec<Value> {
let mut field_data = vec![Value::Null; class.borrow().n_static_fields()];
for (_, this_class) in &class.borrow().static_field_mapping {
for (_name, type_index) in this_class {
let value = match type_index.type_name.as_str() {
"Z" => Value::BOOL(false),
"B" => Value::I32(0),
"S" => Value::I32(0),
"I" => Value::I32(0),
"J" => Value::I64(0),
"F" => Value::F32(0.0),
"D" => Value::F64(0.0),
_ => Value::Null,
};
// println!("{} = {:?}", name, value);
field_data[type_index.index] = value.into();
}
}
field_data
}
// convienence methods for data from the constantpool
pub fn cp_field_ref(&self, index: &u16) -> (&u16, &u16) {
if let CpEntry::Fieldref(class_index, name_and_type_index) =
self.constant_pool.get(index).unwrap()
{
(class_index, name_and_type_index)
} else {
unreachable!("should be field")
}
}
/// both methodRef and InterfaceMethodRef
/// returns (class_index, name_and_type_index)
pub fn cp_method_ref(&self, index: &u16) -> (&u16, &u16) {
if let CpEntry::MethodRef(class_index, name_and_type_index)
| CpEntry::InterfaceMethodref(class_index, name_and_type_index) =
self.constant_pool.get(index).unwrap()
{
(class_index, name_and_type_index)
} else {
unreachable!("should be method")
}
}
pub fn cp_class_ref(&self, index: &u16) -> &u16 {
if let CpEntry::ClassRef(name_index) = self.constant_pool.get(index).unwrap() {
name_index
} else {
unreachable!("should be class entry")
}
}
pub fn cp_utf8(&self, index: &u16) -> &String {
if let CpEntry::Utf8(utf8) = self.constant_pool.get(index).unwrap() {
utf8
} else {
unreachable!("should be utf8 entry")
}
}
pub fn cp_name_and_type(&self, index: &u16) -> (&u16, &u16) {
if let CpEntry::NameAndType(name_index, type_index) = self.constant_pool.get(index).unwrap()
{
(name_index, type_index)
} else {
unreachable!("should be name_and_type entry")
}
}
} }
unsafe impl Send for Class {}
unsafe impl Sync for Class {}
pub struct Method {
pub(crate) constant_pool: Rc<HashMap<u16, CpEntry>>,
access_flags: u16,
name_index: u16,
descriptor_index: u16,
pub(crate) attributes: HashMap<String, AttributeType>,
}
impl fmt::Debug for Method {
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
write!(
f,
"Method {{access_flags: {}, name_index: {}, descriptor_index: {}, attributes: {:?} }}",
self.access_flags, self.name_index, self.descriptor_index, self.attributes
)
}
}
impl Method {
pub fn new(
constant_pool: Rc<HashMap<u16, CpEntry>>,
access_flags: u16,
name_index: u16,
descriptor_index: u16,
attributes: HashMap<String, AttributeType>,
) -> Self {
Method {
constant_pool,
access_flags,
name_index,
descriptor_index,
attributes,
}
}
pub fn name(&self) -> String {
let mut full_name = String::new();
if let CpEntry::Utf8(s) = &self.constant_pool.get(&self.name_index).unwrap() {
full_name.push_str(s);
}
if let CpEntry::Utf8(s) = &self.constant_pool.get(&self.descriptor_index).unwrap() {
full_name.push_str(s);
}
full_name
}
pub fn is(&self, modifier: Modifier) -> bool {
let m = modifier as u16;
(self.access_flags & m) == m
}
}
pub struct Field {
constant_pool: Rc<HashMap<u16, CpEntry>>,
access_flags: u16,
pub(crate) name_index: u16,
descriptor_index: u16,
attributes: HashMap<String, AttributeType>,
_index: u16,
}
impl fmt::Debug for Field {
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
write!(
f,
"Field {{access_flags: {}, name_index: {}, descriptor_index: {}, attributes: {:?} }}",
self.access_flags, self.name_index, self.descriptor_index, self.attributes
)
}
}
impl Field {
pub fn new(
constant_pool: Rc<HashMap<u16, CpEntry>>,
access_flags: u16,
name_index: u16,
descriptor_index: u16,
attributes: HashMap<String, AttributeType>,
field_index: u16,
) -> Self {
Field {
constant_pool,
access_flags,
name_index,
descriptor_index,
attributes,
_index: field_index,
}
}
pub fn is(&self, modifier: Modifier) -> bool {
let modifier = modifier as u16;
self.access_flags & modifier == modifier
}
pub fn name(&self) -> &String {
if let CpEntry::Utf8(utf8) = &self.constant_pool.get(&self.name_index).unwrap() {
return utf8;
}
unreachable!()
}
pub fn type_of(&self) -> &String {
if let CpEntry::Utf8(s) = &self.constant_pool.get(&self.descriptor_index).unwrap() {
return s;
}
panic!()
}
}
const _MODIFIERS: [(Modifier, &str); 12] = [
(Modifier::Public, "public "),
(Modifier::Private, "private "),
(Modifier::Protected, "protected "),
(Modifier::Static, "static "),
(Modifier::Final, "final "),
(Modifier::Synchronized, "synchronized "),
(Modifier::Volatile, "volatile "),
(Modifier::Transient, "transient "),
(Modifier::Native, "native "),
(Modifier::Abstract, "abstract"),
(Modifier::Strict, "strict"),
(Modifier::Synthetic, "synthetic"),
];
pub enum Modifier {
Public = 0x0001,
Private = 0x0002,
Protected = 0x0004,
Static = 0x0008,
Final = 0x0010,
Synchronized = 0x0020,
Volatile = 0x0040,
Transient = 0x0080,
Native = 0x0100,
Abstract = 0x0400,
Strict = 0x0800,
Synthetic = 0x1000,
}
//TODO implement more types
#[derive(Debug)]
pub enum AttributeType {
ConstantValue(u16),
Code(Box<MethodCode>),
StackMapTable,
BootstrapMethods,
NestHost,
NestMembers,
PermittedSubclasses,
Exceptions,
InnerClasses,
EnclosingMethod,
Synthetic,
Signature,
Record,
SourceFile,
LineNumberTable,
LocalVariableTable,
LocalVariableTypeTable,
SourceDebugExtension,
Deprecated,
RuntimeVisibleAnnotations,
RuntimeInvisibleAnnotations,
RuntimeVisibleParameterAnnotations,
RuntimeInvisibleParameterAnnotations,
RuntimeVisibleTypeAnnotations,
RuntimeInvisibleTypeAnnotations,
AnnotationDefault,
MethodParameters,
Module,
ModulePackages,
ModuleMainClass,
}
#[derive(Debug)]
pub struct Exception {
pub start_pc: u16,
pub end_pc: u16,
pub handler_pc: u16,
pub catch_type: u16,
}
impl Exception {
pub fn read(code: &[u8], index: &mut usize) -> Self {
Self {
start_pc: read_u16(code, index),
end_pc: read_u16(code, index),
handler_pc: read_u16(code, index),
catch_type: read_u16(code, index),
}
}
}
#[derive(Debug)]
pub struct MethodCode {
_max_stack: u16,
_max_locals: u16,
pub(crate) opcodes: Vec<u8>,
_exception_table: Vec<Exception>,
_code_attributes: HashMap<String, AttributeType>,
}
impl MethodCode {
pub(crate) fn new(
_max_stack: u16,
_max_locals: u16,
code: Vec<u8>,
_exception_table: Vec<Exception>,
_code_attributes: HashMap<String, AttributeType>,
) -> Self {
Self {
_max_stack,
_max_locals,
opcodes: code,
_exception_table,
_code_attributes,
}
}
}
#[derive(Debug, Clone)] #[derive(Debug, Clone)]
pub enum Value { pub enum Value {
@ -636,7 +60,7 @@ pub enum Value {
BOOL(bool), BOOL(bool),
CHAR(char), CHAR(char),
// objects and arrays // objects and arrays
Ref(UnsafeRef), Ref(ObjectRef),
// special object // special object
Utf8(String), Utf8(String),
} }
@ -651,7 +75,7 @@ impl Value {
} }
} }
pub fn into_object(self) -> UnsafeRef { pub fn into_object(self) -> ObjectRef {
if let Value::Ref(v) = self { if let Value::Ref(v) = self {
v v
} else { } else {
@ -660,22 +84,145 @@ impl Value {
} }
} }
pub type UnsafeRef = Arc<UnsafeCell<ObjectRef>>; #[derive(Debug, Clone)]
pub enum ObjectRef {
pub fn unsafe_ref(object: ObjectRef) -> UnsafeRef { ByteArray(Vec<i8>),
return Arc::new(UnsafeCell::new(object)); ShortArray(Vec<i16>),
IntArray(Vec<i32>),
LongArray(Vec<i64>),
FloatArray(Vec<f32>),
DoubleArray(Vec<f64>),
BooleanArray(Vec<bool>),
CharArray(Vec<char>),
StringArray(Vec<String>),
ObjectArray(ClassId, Vec<ObjectRef>),
Object(Object),
//Box necessary??
Class(Class),
} }
// pub fn unsafe_val(val: Value) -> UnsafeValue { impl ObjectRef {
// return Arc::new(UnsafeCell::new(val)); pub fn get_array_length(&self) -> usize {
match self {
ByteArray(d) => d.len(),
ShortArray(d) => d.len(),
IntArray(d) => d.len(),
LongArray(d) => d.len(),
FloatArray(d) => d.len(),
DoubleArray(d) => d.len(),
BooleanArray(d) => d.len(),
CharArray(d) => d.len(),
StringArray(d) => d.len(),
ObjectArray(_, d) => d.len(),
_ => unreachable!("not an array")
}
}
}
impl ObjectRef {
pub fn new_object_array(class: &Class, size: usize) -> Self {
ObjectArray(class.id, Vec::with_capacity(size))
}
pub fn new_int_array(size: usize) -> Self {
IntArray(Vec::with_capacity(size))
}
pub fn new_byte_array(d: Vec<u8>) -> Self {
ByteArray(into_vec_i8(d))
}
}
fn into_vec_i8(v: Vec<u8>) -> Vec<i8> {
let mut v = std::mem::ManuallyDrop::new(v);
// then, pick apart the existing Vec
let p = v.as_mut_ptr();
let len = v.len();
let cap = v.capacity();
// finally, adopt the data into a new Vec
unsafe { Vec::from_raw_parts(p as *mut i8, len, cap) }
}
#[derive(Debug, Clone)]
pub struct Object {
// locked: bool,
// hashcode: i32,
pub class_id: ClassId,
pub data: Vec<Value>,
} //arrays
// object, not array
impl Object {
pub fn new(class: &Class) -> Self {
let instance_data = Object::init_fields(class);
Self {
class_id: class.id,
data: instance_data,
}
}
// initializes all non-static fields to their default values
pub(crate) fn init_fields(class: &Class) -> Vec<Value> {
let mut field_data = Vec::with_capacity(class.n_object_fields());
for (_, fields) in &class.object_field_mapping {
for (_, type_index) in fields {
let value = match type_index.type_name.as_str() {
"Z" => Value::BOOL(false),
"B" => Value::I32(0),
"S" => Value::I32(0),
"I" => Value::I32(0),
"J" => Value::I64(0),
"F" => Value::F32(0.0),
"D" => Value::F64(0.0),
_ => Value::Null,
};
field_data.push(value.into());
}
}
field_data
}
pub fn set(&mut self, class: &Class, declared_type: &str, field_name: &str, value: Value) {
let type_index = class
.object_field_mapping
.get(declared_type)
.unwrap()
.get(field_name)
.unwrap();
self.data[type_index.index] = value;
}
pub fn get(&mut self, instancedef: &Class, declared_type: &String, field_name: &String) -> &Value {
let type_index = instancedef
.object_field_mapping
.get(declared_type)
.unwrap()
.get(field_name)
.unwrap();
&self.data[type_index.index]
}
// fn get_field_name(&self, cp_index: &u16) -> &str {
// if let CpEntry::Utf8(name) = self.class.constant_pool.get(cp_index).unwrap() {
// return name;
// }
// panic!()
// }
}
// impl Debug for Object {
// fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
// // let fields: Vec<String> = self.data.unwrap().iter().map(|(k)| {
// // // let mut r: String = self.get_field_name(k).into();
// // // r.push(':');
// // // r.push_str(format!("{:?}").as_str());
// // // r
// // }
// // ).collect();
// write!(f, "{}", self.class.name)
// }
// } // }
pub fn type_ref(class: Class) -> Type {
return Arc::new(RefCell::new(class));
}
pub type Type = Arc<RefCell<Class>>;
unsafe impl Send for Value {}
unsafe impl Sync for Value {}

367
src/classloader/classdef.rs Normal file
View file

@ -0,0 +1,367 @@
use std::collections::HashMap;
use std::fmt;
use std::fmt::{Debug, Formatter};
use std::rc::Rc;
use crate::classloader::io::read_u16;
/// This is the class representation when the bytecode had just been loaded.
pub struct ClassDef {
pub minor_version: u16,
pub major_version: u16,
pub constant_pool: Rc<HashMap<u16, CpEntry>>,
pub access_flags: u16,
this_class: u16,
pub super_class: Option<u16>,
pub interfaces: Vec<u16>,
pub fields: HashMap<String, Field>,
pub methods: HashMap<String, Method>,
pub attributes: HashMap<String, AttributeType>,
}
impl Debug for ClassDef{
fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
write!(f, "{}", self.cp_class_name(&self.this_class))
}
}
impl ClassDef {
pub fn new(
minor_version: u16,
major_version: u16,
constant_pool: Rc<HashMap<u16, CpEntry>>,
access_flags: u16,
this_class: u16,
super_class: Option<u16>,
interfaces: Vec<u16>,
fields: HashMap<String, Field>,
methods: HashMap<String, Method>,
attributes: HashMap<String, AttributeType>,
) -> Self {
Self {
major_version,
minor_version,
constant_pool,
access_flags,
this_class,
super_class,
interfaces,
fields,
methods,
attributes,
}
}
/// the bytecode version
pub fn version(&self) -> (u16, u16) {
(self.major_version, self.minor_version)
}
/// get the class name
pub fn name(&self) -> &str {
self.cp_class_name(&self.this_class)
}
pub fn get_method(&self, name: &str) -> Option<&Method> {
self.methods.get(name)
}
pub fn cp_field_ref(&self, index: &u16) -> (&u16, &u16) {
if let CpEntry::Fieldref(class_index, name_and_type_index) =
self.constant_pool.get(index).unwrap()
{
(class_index, name_and_type_index)
} else {
unreachable!("should be field")
}
}
/// both methodRef and InterfaceMethodRef
/// returns (class_index, name_and_type_index)
pub fn cp_method_ref(&self, index: &u16) -> (&u16, &u16) {
if let CpEntry::MethodRef(class_index, name_and_type_index)
| CpEntry::InterfaceMethodref(class_index, name_and_type_index) =
self.constant_pool.get(index).unwrap()
{
(class_index, name_and_type_index)
} else {
unreachable!("should be method")
}
}
pub fn cp_class_name(&self, index: &u16) -> &String {
let cr = self.cp_class_ref(index);
self.cp_utf8(cr)
}
pub fn cp_class_ref(&self, index: &u16) -> &u16 {
if let CpEntry::ClassRef(name_index) = self.constant_pool.get(index).unwrap() {
name_index
} else {
unreachable!("should be class entry")
}
}
pub fn cp_utf8(&self, index: &u16) -> &String {
if let CpEntry::Utf8(utf8) = self.constant_pool.get(index).unwrap() {
utf8
} else {
unreachable!("should be utf8 entry")
}
}
pub fn cp_name_and_type(&self, index: &u16) -> (&u16, &u16) {
if let CpEntry::NameAndType(name_index, type_index) = self.constant_pool.get(index).unwrap()
{
(name_index, type_index)
} else {
unreachable!("should be name_and_type entry")
}
}
}
#[derive(Debug)]
pub enum CpEntry {
Utf8(String),
Integer(i32),
Float(f32),
Long(i64),
Double(f64),
ClassRef(u16),
// (utf8)
StringRef(u16),
// (utf8)
Fieldref(u16, u16),
// (class, name_and_type)
MethodRef(u16, u16),
// (class, name_and_type)
InterfaceMethodref(u16, u16),
// (class, name_and_type)
NameAndType(u16, u16),
// (name, descriptor)
MethodHandle(u8, u16),
MethodType(u16),
InvokeDynamic(u16, u16),
}
pub enum Modifier {
Public = 0x0001,
Private = 0x0002,
Protected = 0x0004,
Static = 0x0008,
Final = 0x0010,
Synchronized = 0x0020,
Volatile = 0x0040,
Transient = 0x0080,
Native = 0x0100,
Abstract = 0x0400,
Strict = 0x0800,
Synthetic = 0x1000,
}
//TODO implement more types
#[derive(Debug)]
pub enum AttributeType {
ConstantValue(u16),
Code(Box<MethodCode>),
StackMapTable,
BootstrapMethods,
NestHost,
NestMembers,
PermittedSubclasses,
Exceptions,
InnerClasses,
EnclosingMethod,
Synthetic,
Signature,
Record,
SourceFile,
LineNumberTable,
LocalVariableTable,
LocalVariableTypeTable,
SourceDebugExtension,
Deprecated,
RuntimeVisibleAnnotations,
RuntimeInvisibleAnnotations,
RuntimeVisibleParameterAnnotations,
RuntimeInvisibleParameterAnnotations,
RuntimeVisibleTypeAnnotations,
RuntimeInvisibleTypeAnnotations,
AnnotationDefault,
MethodParameters,
Module,
ModulePackages,
ModuleMainClass,
}
#[derive(Debug)]
pub struct Exception {
pub start_pc: u16,
pub end_pc: u16,
pub handler_pc: u16,
pub catch_type: u16,
}
impl Exception {
pub fn read(code: &[u8], index: &mut usize) -> Self {
Self {
start_pc: read_u16(code, index),
end_pc: read_u16(code, index),
handler_pc: read_u16(code, index),
catch_type: read_u16(code, index),
}
}
}
#[derive(Debug)]
pub struct MethodCode {
_max_stack: u16,
_max_locals: u16,
pub(crate) opcodes: Vec<u8>,
_exception_table: Vec<Exception>,
_code_attributes: HashMap<String, AttributeType>,
}
impl MethodCode {
pub(crate) fn new(
_max_stack: u16,
_max_locals: u16,
code: Vec<u8>,
_exception_table: Vec<Exception>,
_code_attributes: HashMap<String, AttributeType>,
) -> Self {
Self {
_max_stack,
_max_locals,
opcodes: code,
_exception_table,
_code_attributes,
}
}
}
pub(crate) const _MODIFIERS: [(Modifier, &str); 12] = [
(Modifier::Public, "public "),
(Modifier::Private, "private "),
(Modifier::Protected, "protected "),
(Modifier::Static, "static "),
(Modifier::Final, "final "),
(Modifier::Synchronized, "synchronized "),
(Modifier::Volatile, "volatile "),
(Modifier::Transient, "transient "),
(Modifier::Native, "native "),
(Modifier::Abstract, "abstract"),
(Modifier::Strict, "strict"),
(Modifier::Synthetic, "synthetic"),
];
pub struct Field {
constant_pool: Rc<HashMap<u16, CpEntry>>,
access_flags: u16,
pub(crate) name_index: u16,
descriptor_index: u16,
attributes: HashMap<String, AttributeType>,
_index: u16,
}
impl fmt::Debug for Field {
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
write!(
f,
"Field {{access_flags: {}, name_index: {}, descriptor_index: {}, attributes: {:?} }}",
self.access_flags, self.name_index, self.descriptor_index, self.attributes
)
}
}
impl Field {
pub fn new(
constant_pool: Rc<HashMap<u16, CpEntry>>,
access_flags: u16,
name_index: u16,
descriptor_index: u16,
attributes: HashMap<String, AttributeType>,
field_index: u16,
) -> Self {
Field {
constant_pool,
access_flags,
name_index,
descriptor_index,
attributes,
_index: field_index,
}
}
pub fn is(&self, modifier: Modifier) -> bool {
let modifier = modifier as u16;
self.access_flags & modifier == modifier
}
pub fn name(&self) -> &String {
if let CpEntry::Utf8(utf8) = &self.constant_pool.get(&self.name_index).unwrap() {
return utf8;
}
unreachable!()
}
pub fn type_of(&self) -> &String {
if let CpEntry::Utf8(s) = &self.constant_pool.get(&self.descriptor_index).unwrap() {
return s;
}
panic!()
}
}
pub struct Method {
pub(crate) constant_pool: Rc<HashMap<u16, CpEntry>>,
access_flags: u16,
name_index: u16,
descriptor_index: u16,
pub(crate) attributes: HashMap<String, AttributeType>,
}
impl fmt::Debug for Method {
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
write!(
f,
"Method {{access_flags: {}, name_index: {}, descriptor_index: {}, attributes: {:?} }}",
self.access_flags, self.name_index, self.descriptor_index, self.attributes
)
}
}
impl Method {
pub fn new(
constant_pool: Rc<HashMap<u16, CpEntry>>,
access_flags: u16,
name_index: u16,
descriptor_index: u16,
attributes: HashMap<String, AttributeType>,
) -> Self {
Method {
constant_pool,
access_flags,
name_index,
descriptor_index,
attributes,
}
}
pub fn name(&self) -> String {
let mut full_name = String::new();
if let CpEntry::Utf8(s) = &self.constant_pool.get(&self.name_index).unwrap() {
full_name.push_str(s);
}
if let CpEntry::Utf8(s) = &self.constant_pool.get(&self.descriptor_index).unwrap() {
full_name.push_str(s);
}
full_name
}
pub fn is(&self, modifier: Modifier) -> bool {
let m = modifier as u16;
(self.access_flags & m) == m
}
}

30
src/io.rs → src/classloader/io.rs
View file

@ -1,6 +1,12 @@
use std::fs::{self};
use anyhow::{anyhow, Error}; use anyhow::{anyhow, Error};
use std::fs::{self, File};
use std::io::Read; #[cfg(target_family = "unix")]
pub const PATH_SEPARATOR: char = ':';
#[cfg(target_family = "windows")]
pub const PATH_SEPARATOR: char = ';';
/// resolves the actual path where the class file is found /// resolves the actual path where the class file is found
/// for std lib there is a special case that resolves to the jmod /// for std lib there is a special case that resolves to the jmod
@ -38,25 +44,7 @@ pub fn find_class(classpath: &Vec<String>, class_name: &str) -> Result<String, E
Err(anyhow!("Class not found {}", class_name)) Err(anyhow!("Class not found {}", class_name))
} }
/// reads the binary class file from file path or archive
/// and returns the byte array as Vec
pub fn read_bytecode(name: String) -> Result<Vec<u8>, Error> {
let mut buffer;
if name.contains('#') {
let parts: Vec<&str> = name.split('#').collect();
let archive_file = File::open(parts[0])?;
let mut archive_zip = zip::ZipArchive::new(archive_file)?;
let mut entry = archive_zip.by_name(parts[1])?;
buffer = vec![0; entry.size() as usize];
entry.read_exact(&mut buffer)?;
} else {
let mut f = File::open(&name)?;
let metadata = fs::metadata(&name)?;
buffer = vec![0; metadata.len() as usize];
f.read_exact(&mut buffer)?;
}
Ok(buffer)
}
// methods to read values from big-endian binary data // methods to read values from big-endian binary data

70
src/classloader.rs → src/classloader/mod.rs
View file

@ -1,11 +1,46 @@
use crate::class::{AttributeType, Class, Exception, Field, Method, MethodCode};
use crate::io::{read_bytes, read_f32, read_f64, read_i32, read_i64, read_u16, read_u32, read_u8};
use anyhow::Error;
use std::collections::HashMap; use std::collections::HashMap;
use std::fs;
use std::fs::File;
use std::io::Read;
use std::rc::Rc; use std::rc::Rc;
use anyhow::Error;
use crate::classloader::io::{find_class, read_bytes, read_f32, read_f64, read_i32, read_i64, read_u16, read_u32, read_u8};
use crate::classloader::classdef::{AttributeType, ClassDef, CpEntry, Exception, Field, Method, MethodCode};
pub(crate) mod classdef;
pub(crate) mod io;
pub(crate) fn get_classdef(classpath: &Vec<String>, class_name: &str) -> Result<ClassDef,Error> {
println!("read class {} ", class_name);
let resolved_path = find_class(classpath, class_name)?;
let bytecode = read_bytecode(resolved_path)?;
load_class(bytecode)
}
/// reads the binary class file from file path or archive
/// and returns the byte array as Vec
fn read_bytecode(name: String) -> Result<Vec<u8>, Error> {
let mut buffer;
if name.contains('#') {
let parts: Vec<&str> = name.split('#').collect();
let archive_file = File::open(parts[0])?;
let mut archive_zip = zip::ZipArchive::new(archive_file)?;
let mut entry = archive_zip.by_name(parts[1])?;
buffer = vec![0; entry.size() as usize];
entry.read_exact(&mut buffer)?;
} else {
let mut f = File::open(&name)?;
let metadata = fs::metadata(&name)?;
buffer = vec![0; metadata.len() as usize];
f.read_exact(&mut buffer)?;
}
Ok(buffer)
}
// The native classoader // The native classoader
pub fn load_class(bytecode: Vec<u8>) -> Result<Class, Error> { fn load_class(bytecode: Vec<u8>) -> Result<ClassDef, Error> {
let pos = &mut 0; let pos = &mut 0;
check_magic(&bytecode, pos); check_magic(&bytecode, pos);
let minor_version = read_u16(&bytecode, pos); let minor_version = read_u16(&bytecode, pos);
@ -22,13 +57,11 @@ pub fn load_class(bytecode: Vec<u8>) -> Result<Class, Error> {
); );
cp_index += 1; cp_index += 1;
} }
let constant_pool = Rc::new(constant_pool); let constant_pool = Rc::new(constant_pool);
let access_flags = read_u16(&bytecode, pos); let access_flags = read_u16(&bytecode, pos);
let this_class = read_u16(&bytecode, pos); let this_class = read_u16(&bytecode, pos);
let super_class = read_u16(&bytecode, pos); let super_class = read_u16(&bytecode, pos);
let super_class = if super_class != 0 { Some(super_class) } else { None };
let interfaces_count = read_u16(&bytecode, pos); let interfaces_count = read_u16(&bytecode, pos);
let mut interfaces = vec![]; let mut interfaces = vec![];
for _ in 0..interfaces_count { for _ in 0..interfaces_count {
@ -46,7 +79,7 @@ pub fn load_class(bytecode: Vec<u8>) -> Result<Class, Error> {
let mut methods = HashMap::new(); let mut methods = HashMap::new();
for _ in 0..methods_count { for _ in 0..methods_count {
let m = read_method(constant_pool.clone(), pos, &bytecode); let m = read_method(constant_pool.clone(), pos, &bytecode);
methods.insert(m.name(), Rc::new(m)); methods.insert(m.name(), m);
} }
let attributes_count = read_u16(&bytecode, pos); let attributes_count = read_u16(&bytecode, pos);
@ -60,7 +93,7 @@ pub fn load_class(bytecode: Vec<u8>) -> Result<Class, Error> {
} }
} }
Ok(Class::new( Ok(ClassDef::new(
minor_version, minor_version,
major_version, major_version,
constant_pool, constant_pool,
@ -224,7 +257,6 @@ fn read_attribute(
*index += attribute_length; *index += attribute_length;
if let CpEntry::Utf8(s) = &constant_pool.get(&attribute_name_index).unwrap() { if let CpEntry::Utf8(s) = &constant_pool.get(&attribute_name_index).unwrap() {
// println!("Att [{}]", s);
return match s.as_str() { return match s.as_str() {
"ConstantValue" => { "ConstantValue" => {
assert_eq!(info.len(), 2); assert_eq!(info.len(), 2);
@ -274,6 +306,7 @@ fn read_attribute(
"Signature" => Some(("".into(), AttributeType::Signature)), //stub "Signature" => Some(("".into(), AttributeType::Signature)), //stub
"NestHost" => Some(("".into(), AttributeType::NestHost)), //stub "NestHost" => Some(("".into(), AttributeType::NestHost)), //stub
"EnclosingMethod" => Some(("".into(), AttributeType::EnclosingMethod)), //stub "EnclosingMethod" => Some(("".into(), AttributeType::EnclosingMethod)), //stub
"PermittedSubclasses" => Some(("".into(), AttributeType::PermittedSubclasses)), //stub
//TODO more actual attribute implementations //TODO more actual attribute implementations
_ => None, _ => None,
}; };
@ -281,20 +314,3 @@ fn read_attribute(
None None
} }
#[derive(Debug)]
pub enum CpEntry {
Utf8(String),
Integer(i32),
Float(f32),
Long(i64),
Double(f64),
ClassRef(u16), // (utf8)
StringRef(u16), // (utf8)
Fieldref(u16, u16), // (class, name_and_type)
MethodRef(u16, u16), // (class, name_and_type)
InterfaceMethodref(u16, u16), // (class, name_and_type)
NameAndType(u16, u16), // (name, descriptor)
MethodHandle(u8, u16),
MethodType(u16),
InvokeDynamic(u16, u16),
}

319
src/classmanager.rs Normal file
View file

@ -0,0 +1,319 @@
use std::collections::{HashMap, LinkedList};
use once_cell::sync::Lazy;
use crate::class::{Class, ClassId, ObjectRef, TypeIndex, Value::*, Value};
use crate::class::Object;
use crate::classloader;
use crate::classloader::classdef::{ClassDef, Modifier};
use crate::classloader::io::PATH_SEPARATOR;
use crate::vm::Vm;
static mut CLASSMANAGER: Lazy<ClassManager> = Lazy::new(|| ClassManager::new());
pub fn init() {
unsafe {
CLASSMANAGER.classes.clear();
CLASSMANAGER.names.clear();
CLASSMANAGER.classdefs.clear();
CLASSMANAGER.class_objects.clear();
CLASSMANAGER.static_class_data.clear();
}
}
pub fn set_classpath(classpath: &str) {
unsafe {
CLASSMANAGER.set_classpath(classpath);
}
}
pub fn get_class_by_id(id: ClassId) -> Option<&'static Class> {
unsafe {
CLASSMANAGER.get_class_by_id(id)
}
}
pub fn classdef_name(id: &ClassId) -> Option<String> {
unsafe {
CLASSMANAGER.classdef_name(id)
}
}
pub fn get_classid(name: &str) -> &ClassId {
unsafe {
CLASSMANAGER.get_classid(name)
}
}
pub fn get_classdef(id: &ClassId) -> &ClassDef {
unsafe {
CLASSMANAGER.get_classdef(id)
}
}
pub fn load_class_by_name(name: &str) {
unsafe {
CLASSMANAGER.load_class_by_name(name)
}
}
pub fn get_class_by_name(name: &str) -> Option<&Class> {
unsafe {
CLASSMANAGER.get_class_by_name(name)
}
}
pub fn add_class(name: &str) -> ClassId {
unsafe {
CLASSMANAGER.add_class(name)
}
}
pub fn get_static(id: &ClassId, index: usize) -> Value {
unsafe {
CLASSMANAGER.static_class_data.get(id).unwrap()[index].clone()
}
}
pub fn set_static(id: &ClassId, index: usize, value: Value) {
unsafe {
CLASSMANAGER.static_class_data.get_mut(id).unwrap()[index] = value;
}
}
pub fn get_classobject(id: &ClassId) -> Option<&Value> {
unsafe {
CLASSMANAGER.class_objects.get(id)
}
}
//TODO less pubs
pub struct ClassManager {
static_class_data: HashMap<ClassId, Vec<Value>>,
// sequence for passing new classIds
current_id: ClassId,
// the classpath
classpath: Vec<String>,
//references to classdefs, ie the static class info
pub classdefs: HashMap<ClassId, ClassDef>,
// references to the runtime class
pub classes: HashMap<ClassId, Class>,
pub names: HashMap<String, ClassId>,
pub class_objects: HashMap<ClassId, Value>,
vm: Vm,
}
impl ClassManager {
pub fn new() -> Self {
Self {
static_class_data: HashMap::new(),
current_id: 0,
classdefs: HashMap::new(),
classes: HashMap::new(),
class_objects: HashMap::new(),
names: HashMap::new(),
classpath: vec![],
vm:Vm::new_internal(),
}
}
fn set_classpath(&mut self, classpath: &str) {
self.classpath = classpath
.split(PATH_SEPARATOR)
.map(|s| s.into())
.collect();
}
fn get_class_by_id(&mut self, id: ClassId) -> Option<&Class> {
if !self.classes.contains_key(&id) {
let name = self.classdef_name(&id);
if name.is_some() {
self.add_class(&name.unwrap());
}
}
self.classes.get(&id)
}
fn classdef_name(&self, id: &ClassId) -> Option<String> {
self.classdefs.get(id).map(|c| c.name().to_owned()) //drops borrow to self here
}
fn get_classid(&self, name: &str) -> &ClassId {
self.names.get(name).unwrap()
}
fn get_classdef(&self, id: &ClassId) -> &ClassDef {
self.classdefs.get(&id).unwrap()
}
fn load_class_by_name(&mut self, name: &str) {
let id = self.names.get(name);
match id {
Some(id) => if self.classes.get(id).is_none() {
self.add_class(name);
}
None => { self.add_class(name); }
}
}
fn get_class_by_name(&self, name: &str) -> Option<&Class> {
let id = self.names.get(name);
self.classes.get(id.unwrap())
}
fn add_class(&mut self, name: &str) -> ClassId {
let this_classid = self.load(name);
let this_classdef = self.classdefs.get(&this_classid).unwrap();
//compute indices to fields
let mut object_field_mapping = HashMap::new();
let mut static_field_mapping = HashMap::new();
let object_field_map_index: &mut usize = &mut 0;
let static_field_map_index: &mut usize = &mut 0;
let mut current_id = Some(this_classid);
let mut current_classdef;
let mut parents = LinkedList::new();
while let Some(c) = current_id {
parents.push_front(current_id.unwrap());
current_classdef = self.classdefs.get(&c).unwrap();
Self::add_fields_for_this_or_parents(&mut object_field_mapping, &mut static_field_mapping, object_field_map_index, static_field_map_index, current_classdef);
current_id = current_classdef.super_class.as_ref()
.map(|i| current_classdef.cp_class_name(i).to_owned())
.map(|n| *self.names.get(&n).unwrap());
}
//handrolled references to superclass and interfaces
let superclass_id = this_classdef.super_class.as_ref()
.map(|i| this_classdef.cp_class_name(i).to_owned())
.map(|n| *self.names.get(&n).unwrap());
let interface_ids: Vec<ClassId> = this_classdef.interfaces.iter()
.map(|i| this_classdef.cp_class_name(i).to_owned())
.map(|n| *self.names.get(n.as_str()).unwrap())
.collect();
self.static_class_data.insert(this_classid, Self::set_field_data(&static_field_mapping));
self.names.get(name)
.and_then(|id|
self.classes.insert(*id, Class {
id: *id,
initialized: false,
name: name.into(),
superclass: superclass_id,
parents,
interfaces: interface_ids,
object_field_mapping,
static_field_mapping,
// static_field_data: static_values,
}));
if name != "java/lang/Class" {
let cls = self.get_class_by_name("java/lang/Class").unwrap();
let mut instance = Object::new(cls);
instance.set(cls, "java/lang/Class", "name", Value::Utf8(name.into()));
let instance = Ref(ObjectRef::Object(instance));
self.class_objects.insert(this_classid, instance);
}
let clinit = this_classdef.methods.contains_key("<clinit>()V");
if clinit {
self.vm.execute_special(&mut vec![],name, "<clinit>()V", vec![]).unwrap();
}
this_classid
}
fn add_fields_for_this_or_parents(object_field_mapping: &mut HashMap<String, HashMap<String, TypeIndex>>,
static_field_mapping: &mut HashMap<String, HashMap<String, TypeIndex>>,
object_field_map_index: &mut usize, static_field_map_index:
&mut usize,
current_classdef: &ClassDef) {
let mut instance_field_mappings: HashMap<String, TypeIndex> = HashMap::new();
let mut static_field_mappings: HashMap<String, TypeIndex> = HashMap::new();
for (field_name, field) in &current_classdef.fields {
if !field.is(Modifier::Static) {
instance_field_mappings.insert(field_name.to_owned(),
TypeIndex::new(field.type_of().to_owned(), *object_field_map_index));
*object_field_map_index += 1;
} else {
static_field_mappings.insert(field_name.to_owned(),
TypeIndex::new(field.type_of().to_owned(), *static_field_map_index));
*static_field_map_index += 1;
}
}
object_field_mapping.insert(current_classdef.name().to_owned(), instance_field_mappings);
static_field_mapping.insert(current_classdef.name().to_owned(), static_field_mappings);
}
/// loads the class and recursively its dependencies
fn load(&mut self, name: &str) -> ClassId {
let (id, mut classes_to_load) = self.load_class_and_deps(name);
while !classes_to_load.is_empty() {
if let Some(classname) = classes_to_load.pop() {
classes_to_load.append(&mut self.load_class_and_deps(classname.as_str()).1);
}
}
id
}
/// loads the class and returns it's dependencies
fn load_class_and_deps(&mut self, name: &str) -> (ClassId, Vec<String>) {
let id = *self.names.entry(name.to_string()).or_insert_with(|| {
self.current_id += 1;
self.current_id
});
let classdef = self.classdefs
.entry(id)
.or_insert_with(|| classloader::get_classdef(&self.classpath, name).expect("ClassNotFound"));
(self.current_id, inspect_dependencies(classdef))
}
pub(crate) fn set_field_data(field_mapping: &HashMap<String, HashMap<String, TypeIndex>>) -> Vec<Value> {
let mut field_data = vec![Null; n_fields(field_mapping)];
for (_, this_class) in field_mapping {
for (_name, type_index) in this_class {
let value = match type_index.type_name.as_str() {
"Z" => BOOL(false),
"B" => I32(0),
"S" => I32(0),
"I" => I32(0),
"J" => I64(0),
"F" => F32(0.0),
"D" => F64(0.0),
_ => Null,
};
// println!("{} = {:?}", name, value);
field_data[type_index.index] = value.into();
}
}
field_data
}
}
pub(crate) fn n_fields(field_mapping: &HashMap<String, HashMap<String, TypeIndex>>) -> usize {
field_mapping
.iter()
.map(|(_, v)| v.len())
.reduce(|acc, e| acc + e)
.unwrap()
}
pub(crate) fn inspect_dependencies(classdef: &ClassDef) -> Vec<String> {
let mut classes_to_load: Vec<String> = vec![];
if let Some(superclass) = &classdef.super_class {
classes_to_load.push(classdef.cp_class_name(superclass).into());
}
for interface in &classdef.interfaces {
classes_to_load.push(classdef.cp_class_name(interface).into());
}
classes_to_load
}

190
src/heap.rs
View file

@ -1,190 +0,0 @@
use std::cell::{RefCell, UnsafeCell};
use std::fmt;
use std::fmt::{Debug, Formatter};
use std::sync::Arc;
use ObjectRef::{BooleanArray, CharArray, DoubleArray, FloatArray, LongArray, ShortArray};
use crate::class::{Class, Type, Value};
use crate::heap::ObjectRef::{ByteArray, IntArray, ObjectArray, StringArray};
// can contain object or array
pub enum ObjectRef {
ByteArray(Vec<i8>),
ShortArray(Vec<i16>),
IntArray(Vec<i32>),
LongArray(Vec<i64>),
FloatArray(Vec<f32>),
DoubleArray(Vec<f64>),
BooleanArray(Vec<bool>),
CharArray(Vec<char>),
StringArray(Vec<String>),
ObjectArray(Type, Vec<Arc<UnsafeCell<ObjectRef>>>),
Object(Box<Object>),//Box necessary??
Class(Arc<RefCell<Class>>),
}
impl ObjectRef {
pub fn get_array_length(&self) -> usize {
match self {
ByteArray(d) => d.len(),
ShortArray(d) => d.len(),
IntArray(d) => d.len(),
LongArray(d) => d.len(),
FloatArray(d) => d.len(),
DoubleArray(d) => d.len(),
BooleanArray(d) => d.len(),
CharArray(d) => d.len(),
StringArray(d) => d.len(),
ObjectArray(_, d) => d.len(),
_ => unreachable!("not an array")
}
}
}
fn into_vec_i8(v: Vec<u8>) -> Vec<i8> {
let mut v = std::mem::ManuallyDrop::new(v);
// then, pick apart the existing Vec
let p = v.as_mut_ptr();
let len = v.len();
let cap = v.capacity();
// finally, adopt the data into a new Vec
unsafe { Vec::from_raw_parts(p as *mut i8, len, cap) }
}
impl ObjectRef {
pub fn new_object_array(class: Type, size: usize) -> Self {
ObjectArray(class, Vec::with_capacity(size))
}
pub fn new_int_array(size: usize) -> Self {
IntArray(Vec::with_capacity(size))
}
pub fn new_byte_array(d: Vec<u8>) -> Arc<UnsafeCell<Self>> {
Arc::new(UnsafeCell::new(ByteArray(into_vec_i8(d))))
}
}
impl Debug for ObjectRef {
fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
match self {
BooleanArray(d) => write!(f, "[Z;{}]", d.len()),
ByteArray(d) => write!(f, "[B;{}]", d.len()),
CharArray(d) => write!(f, "[C;{}]", d.len()),
DoubleArray(d) => write!(f, "[D;{}]", d.len()),
FloatArray(d) => write!(f, "[F;{}]", d.len()),
IntArray(d) => write!(f, "[I;{}]", d.len()),
LongArray(d) => write!(f, "[J;{}]", d.len()),
ObjectArray(t, d) => write!(f, "[L{};{}]", t.borrow().name, d.len()),
ShortArray(d) => write!(f, "[S;{}]", d.len()),
StringArray(d) => write!(f, "[S;{}]", d.len()),
ObjectRef::Object(r) => write!(f, "{}{{ {:?} }}", r.class.borrow().name, r.data),
ObjectRef::Class(s) => write!(f, "Class {:?}", s.borrow().name),
}
}
}
// trying to implement efficient object instance storage
pub struct Object {
// locked: bool,
// hashcode: i32,
pub class: Arc<RefCell<Class>>,
pub data: Vec<Value>,
} //arrays
unsafe impl Send for Object {}
unsafe impl Sync for Object {}
// object, not array
impl Object {
pub fn new(class: Arc<RefCell<Class>>) -> Self {
let instance_data = Object::init_fields(class.clone());
Self {
class,
data: instance_data,
}
}
// initializes all non-static fields to their default values
pub(crate) fn init_fields(class: Arc<RefCell<Class>>) -> Vec<Value> {
let mut field_data = Vec::with_capacity(class.borrow().n_object_fields());
for (_, fields) in &class.borrow().object_field_mapping {
for (_, type_index) in fields {
let value = match type_index.type_name.as_str() {
"Z" => Value::BOOL(false),
"B" => Value::I32(0),
"S" => Value::I32(0),
"I" => Value::I32(0),
"J" => Value::I64(0),
"F" => Value::F32(0.0),
"D" => Value::F64(0.0),
_ => Value::Null,
};
field_data.push(value.into());
}
}
field_data
}
pub fn set(&mut self, class_name: &String, field_name: &String, value: Value) {
let borrow = self.class.borrow();
let type_index = borrow
.object_field_mapping
.get(class_name)
.unwrap()
.get(field_name)
.unwrap();
self.data[type_index.index] = value;
}
pub fn get(&mut self, class_name: &String, field_name: &String) -> &Value {
let borrow = self.class.borrow();
let type_index = borrow
.object_field_mapping
.get(class_name)
.unwrap()
.get(field_name)
.unwrap();
&self.data[type_index.index]
}
// fn get_field_name(&self, cp_index: &u16) -> &str {
// if let CpEntry::Utf8(name) = self.class.constant_pool.get(cp_index).unwrap() {
// return name;
// }
// panic!()
// }
}
impl fmt::Debug for Object {
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
// let fields: Vec<String> = self.data.unwrap().iter().map(|(k)| {
// // let mut r: String = self.get_field_name(k).into();
// // r.push(':');
// // r.push_str(format!("{:?}").as_str());
// // r
// }
// ).collect();
write!(f, "{}", self.class.borrow().name)
}
}
// will using Arc's enable a GC-less heap????
pub(crate) struct Heap {
objects: Vec<Arc<UnsafeCell<ObjectRef>>>,
}
impl Heap {
pub fn new() -> Self {
Self { objects: vec![] }
}
pub(crate) fn new_object(&mut self, object: Arc<UnsafeCell<ObjectRef>>) {
self.objects.push(object);
}
}

9
src/lib.rs
View file

@ -1,8 +1,5 @@
mod classloader;
pub mod classmanager;
pub mod class; pub mod class;
pub mod classloader;
pub mod heap;
pub mod io;
pub mod opcodes;
pub mod vm; pub mod vm;
pub mod heap;
pub mod native;

18
src/main.rs
View file

@ -1,17 +1,11 @@
use std::io::Error; use java_rs::classmanager::set_classpath;
use java_rs::vm::Vm; use java_rs::vm::Vm;
fn main() {
let mut stackframes = Vec::new();
fn main() -> Result<(), Error> { let mut vm = Vm::new(&mut stackframes);
// TODO cmdline args set_classpath("/Users/Shautvast/dev/java/tests");
// TODO build index for package -> jarfile?
let mut vm = Vm::new("tests");
// let main_class = "Inheritance";
let main_class = "testclasses.Main"; let main_class = "testclasses.Main";
vm.execute_static( &mut stackframes, main_class, "main([Ljava/lang/String;)V", vec![])
vm.execute_static( main_class, "main([Ljava/lang/String;)V", vec![])
.unwrap(); .unwrap();
Ok(())
} }

23
src/vm/array.rs
View file

@ -1,16 +1,17 @@
use anyhow::{anyhow, Error}; use anyhow::{anyhow, Error};
use crate::class::Value::{self,*};
use crate::heap::ObjectRef;
pub(crate) unsafe fn array_load(index: Value, arrayref: Value) -> Result<Value, Error> { use crate::class::ObjectRef;
if let I32(index) = &index { use crate::class::Value::{self, *};
let index = *index as usize;
pub(crate) fn array_load(index: Value, arrayref: Value) -> Result<Value, Error> {
if let I32(index) = index {
let index = index as usize;
if let Null = arrayref { if let Null = arrayref {
return Err(anyhow!("NullpointerException")); return Err(anyhow!("NullpointerException"));
} }
if let Ref(objectref) = arrayref { if let Ref(objectref) = arrayref {
match &*objectref.get() { match objectref {
ObjectRef::ByteArray(array) => { ObjectRef::ByteArray(array) => {
return Ok(I32(array[index] as i32)); return Ok(I32(array[index] as i32));
} }
@ -53,14 +54,14 @@ pub(crate) unsafe fn array_load(index: Value, arrayref: Value) -> Result<Value,
panic!() panic!()
} }
pub(crate) unsafe fn array_store(value: Value, index: Value, arrayref: &mut Value) -> Result<(), Error> { pub(crate) fn array_store(value: Value, index: Value, arrayref: Value) -> Result<(), Error> {
if let Null = &*arrayref { if let Null = arrayref {
return Err(anyhow!("NullpointerException")); return Err(anyhow!("NullpointerException"));
} }
if let I32(index) = index { if let I32(index) = index {
if let Ref(ref mut objectref) = arrayref { if let Ref(mut objectref) = arrayref {
match &mut *objectref.get() { match objectref {
ObjectRef::ByteArray(ref mut array) => { ObjectRef::ByteArray(ref mut array) => {
if let I32(value) = value { if let I32(value) = value {
// is i32 correct? // is i32 correct?
@ -91,7 +92,7 @@ pub(crate) unsafe fn array_store(value: Value, index: Value, arrayref: &mut Valu
unreachable!() unreachable!()
} }
} }
ObjectRef::CharArray(ref mut array) => { ObjectRef::CharArray(ref mut array) => unsafe{
if let I32(value) = value { if let I32(value) = value {
array[index as usize] = char::from_u32_unchecked(value as u32); array[index as usize] = char::from_u32_unchecked(value as u32);
} else { } else {

4
src/vm/mod.rs
View file

@ -2,4 +2,6 @@ mod vm;
pub use vm::Vm; pub use vm::Vm;
mod operations; mod operations;
mod stack; mod stack;
mod array; mod array;
mod opcodes;
mod native;

42
src/native.rs → src/vm/native.rs
View file

@ -1,52 +1,52 @@
#![allow(non_snake_case)] #![allow(non_snake_case)]
use std::ptr::hash;
use anyhow::Error; use anyhow::Error;
use log::{debug, info}; use log::{info};
use once_cell::sync::Lazy; use once_cell::sync::Lazy;
use crate::class::{get_class, unsafe_ref, Value}; use crate::class::{ObjectRef, Value};
use crate::class::ObjectRef::Object;
use crate::class::Value::Void; use crate::class::Value::Void;
use crate::heap::ObjectRef; use crate::classmanager;
use crate::heap::ObjectRef::Object; use crate::vm::stack::StackFrame;
use crate::vm::Vm; use crate::vm::Vm;
pub fn invoke_native(vm: &mut Vm, class_name: &String, method_name: &String, _args: Vec<Value>) -> Result<Value,Error> { pub fn invoke_native(vm: &mut Vm, stackframes: &mut Vec<StackFrame>, class_name: &String, method_name: &str, _args: Vec<Value>) -> Result<Value, Error> {
info!("native {}.{}", class_name, method_name); info!("native {}.{}", class_name, method_name);
match class_name.as_str() { match class_name.as_str() {
"java/lang/Class" => java_lang_class(vm, method_name), "java/lang/Class" => java_lang_class(vm, method_name),
"jdk/internal/util/SystemProps$Raw" => jdk_internal_util_SystemProps_Raw(vm, method_name), "jdk/internal/util/SystemProps$Raw" => jdk_internal_util_SystemProps_Raw(vm, stackframes, method_name),
_ => Ok(Void) _ => Ok(Void)
} }
} }
fn java_lang_class(_vm: &mut Vm, method_name: &String) -> Result<Value,Error> { fn java_lang_class(_vm: &Vm, method_name: &str) -> Result<Value, Error> {
Ok(match method_name.as_str() { Ok(match method_name {
"desiredAssertionStatus0(Ljava/lang/Class;)Z" => Value::BOOL(false), "desiredAssertionStatus0(Ljava/lang/Class;)Z" => Value::BOOL(false),
_ => Void _ => Void
}) })
} }
fn jdk_internal_util_SystemProps_Raw(vm: &mut Vm,method_name: &String) -> Result<Value,Error> { fn jdk_internal_util_SystemProps_Raw(vm: &mut Vm, stackframes: &mut Vec<StackFrame>, method_name: &str) -> Result<Value, Error> {
match method_name.as_str() { match method_name {
"platformProperties()[Ljava/lang/String;" => systemProps(), "platformProperties()[Ljava/lang/String;" => systemProps(),
"cmdProperties()Ljava/util/HashMap;" => cmdProps(vm), //TODO ability to instantiate classes here "cmdProperties()Ljava/util/HashMap;" => cmdProps(vm, stackframes), //TODO ability to instantiate classes here
"vmProperties()[Ljava/lang/String;" => cmdProps(vm), "vmProperties()[Ljava/lang/String;" => cmdProps(vm, stackframes),
_ => Ok(Void) _ => Ok(Void)
} }
} }
fn cmdProps(vm: &mut Vm,) -> Result<Value,Error> { fn cmdProps(vm: &mut Vm, stackframes: &mut Vec<StackFrame>) -> Result<Value, Error> {
let hashmap_class = get_class(vm, "java/util/HashMap")?; classmanager::load_class_by_name("java/util/HashMap");
let hashmap_class = classmanager::get_class_by_name("java/util/HashMap").unwrap();
let hashmap = Vm::new_instance(hashmap_class); let hashmap = Vm::new_instance(hashmap_class);
let hashmap = Value::Ref(unsafe_ref(Object(Box::new(hashmap)))); let hashmap = Value::Ref(Object(hashmap));
vm.execute_special("java/util/HashMap", "<init>()V", vec![hashmap.clone()]); vm.execute_special(stackframes, "java/util/HashMap", "<init>()V", vec![hashmap.clone()]);
unsafe {debug!("hashmap {:?}", *hashmap.into_object().get());}
panic!() panic!()
} }
fn systemProps() -> Result<Value,Error> { fn systemProps() -> Result<Value, Error> {
unsafe { unsafe {
let props: Lazy<Vec<String>> = Lazy::new(|| { let props: Lazy<Vec<String>> = Lazy::new(|| {
let mut vec: Vec<String> = Vec::new(); let mut vec: Vec<String> = Vec::new();
@ -87,7 +87,7 @@ fn systemProps() -> Result<Value,Error> {
if let Ok(https_proxy) = std::env::var("https_proxy") { if let Ok(https_proxy) = std::env::var("https_proxy") {
vec.push(https_proxy.to_owned()); vec.push(https_proxy.to_owned());
vec.push(https_proxy); vec.push(https_proxy);
}else { } else {
vec.push("".to_owned()); vec.push("".to_owned());
vec.push("".to_owned()); vec.push("".to_owned());
} }
@ -131,6 +131,6 @@ fn systemProps() -> Result<Value,Error> {
vec vec
}); });
Ok(Value::Ref(unsafe_ref(ObjectRef::StringArray(props.to_vec())))) Ok(Value::Ref(ObjectRef::StringArray(props.to_vec())))
} }
} }

0
src/opcodes.rs → src/vm/opcodes.rs
View file

39
src/vm/operations.rs
View file

@ -1,29 +1,27 @@
use std::cell::RefCell;
use std::collections::HashMap; use std::collections::HashMap;
use std::rc::Rc;
use std::sync::Arc;
use anyhow::Error; use anyhow::Error;
use crate::class::{Class, get_class, Value}; use crate::class::{Class, Value};
use crate::classloader::CpEntry; use crate::classloader::classdef::CpEntry;
use crate::vm::Vm; use crate::classmanager;
use crate::vm::vm::{Invocation, MethodSignature}; use crate::vm::vm::{Invocation, MethodSignature};
/// the place for opcode implementations that are a bit long /// the place for opcode implementations that are a bit long
// GET_STATIC opcode // GET_STATIC opcode
pub(crate) fn get_static(vm: &mut Vm, this_class: Arc<RefCell<Class>>, field_index: u16) -> Result<Value,Error> { pub(crate) fn get_static(this_class: &Class, field_index: u16) -> Result<Value, Error> {
let this_class = this_class.borrow(); let classdef = classmanager::get_classdef(&this_class.id);
let (class_index, field_name_and_type_index) = let (class_index, field_name_and_type_index) =
this_class.cp_field_ref(&field_index); // all these unwraps are safe as long as the class is valid classdef.cp_field_ref(&field_index); // all these unwraps are safe as long as the class is valid
let (name_index, _) = let (name_index, _) =
this_class.cp_name_and_type(field_name_and_type_index); classdef.cp_name_and_type(field_name_and_type_index);
let field_name = this_class.cp_utf8(name_index); let field_name = classdef.cp_utf8(name_index);
let that_class_name_index = this_class.cp_class_ref(class_index); let that_class_name_index = classdef.cp_class_ref(class_index);
let that_class_name = this_class.cp_utf8(that_class_name_index); let that_class_name = classdef.cp_utf8(that_class_name_index);
let that_class = get_class(vm, that_class_name.as_str())?; classmanager::load_class_by_name(that_class_name);
let that_class = that_class.borrow(); let that_class = classmanager::get_class_by_name(that_class_name).unwrap();
let type_index = that_class let type_index = that_class
.static_field_mapping .static_field_mapping
@ -32,10 +30,10 @@ pub(crate) fn get_static(vm: &mut Vm, this_class: Arc<RefCell<Class>>, field_ind
.get(field_name) .get(field_name)
.unwrap(); // safe because field must be there .unwrap(); // safe because field must be there
Ok(that_class.static_data[type_index.index].clone()) Ok(classmanager::get_static(&this_class.id, type_index.index))
} }
pub(crate) fn get_name_and_type(cp: Rc<HashMap<u16, CpEntry>>, index: u16) -> Option<MethodSignature> { pub(crate) fn get_name_and_type(cp: &HashMap<u16, CpEntry>, index: u16) -> Option<MethodSignature> {
if let CpEntry::NameAndType(method_name_index, signature_index) = cp.get(&index).unwrap() { if let CpEntry::NameAndType(method_name_index, signature_index) = cp.get(&index).unwrap() {
if let CpEntry::Utf8(method_name) = cp.get(method_name_index).unwrap() { if let CpEntry::Utf8(method_name) = cp.get(method_name_index).unwrap() {
if let CpEntry::Utf8(signature) = cp.get(signature_index).unwrap() { if let CpEntry::Utf8(signature) = cp.get(signature_index).unwrap() {
@ -48,16 +46,17 @@ pub(crate) fn get_name_and_type(cp: Rc<HashMap<u16, CpEntry>>, index: u16) -> Op
} }
None None
} }
pub(crate) fn get_signature_for_invoke(cp: &Rc<HashMap<u16, CpEntry>>, index: u16) -> Option<Invocation> {
pub(crate) fn get_signature_for_invoke(cp: &HashMap<u16, CpEntry>, index: u16) -> Option<Invocation> {
if let CpEntry::MethodRef(class_index, name_and_type_index) if let CpEntry::MethodRef(class_index, name_and_type_index)
| CpEntry::InterfaceMethodref(class_index, name_and_type_index) = cp.get(&index).unwrap() | CpEntry::InterfaceMethodref(class_index, name_and_type_index) = cp.get(&index).unwrap()
{ {
if let Some(method_signature) = get_name_and_type(Rc::clone(&cp), *name_and_type_index) { if let Some(method_signature) = get_name_and_type(cp, *name_and_type_index) {
if let CpEntry::ClassRef(class_name_index) = cp.get(class_index).unwrap() { if let CpEntry::ClassRef(class_name_index) = cp.get(class_index).unwrap() {
if let CpEntry::Utf8(class_name) = cp.get(class_name_index).unwrap() { if let CpEntry::Utf8(class_name) = cp.get(class_name_index).unwrap() {
return Some(Invocation::new( return Some(Invocation::new(
class_name.into(), class_name.into(),
method_signature) method_signature)
); );
} }
} }

6
src/vm/stack.rs
View file

@ -1,8 +1,10 @@
use anyhow::Error; use anyhow::Error;
use log::debug;
use crate::class::Value; use crate::class::Value;
#[derive(Debug)] #[derive(Debug)]
pub(crate) struct StackFrame { pub struct StackFrame {
pub(crate) at: String, pub(crate) at: String,
pub(crate) data: Vec<Value>, pub(crate) data: Vec<Value>,
} }
@ -17,12 +19,14 @@ impl StackFrame {
} }
pub(crate) fn push(&mut self, val: Value) { pub(crate) fn push(&mut self, val: Value) {
debug!("push {:?}", val);
self.data.push(val); self.data.push(val);
} }
pub(crate) fn len(&self) -> usize { pub(crate) fn len(&self) -> usize {
self.data.len() self.data.len()
} }
pub(crate) fn pop(&mut self) -> Result<Value, Error> { pub(crate) fn pop(&mut self) -> Result<Value, Error> {
Ok(self.data.pop().unwrap()) Ok(self.data.pop().unwrap())
} }

717
src/vm/vm.rs
View file

File diff suppressed because it is too large Load diff

28
tests/class_tests.rs
View file

@ -1,18 +1,20 @@
mod test { mod test {
use java_rs::class::Value; use java_rs::class::{ObjectRef, Value};
use java_rs::heap::ObjectRef; use java_rs::classmanager::set_classpath;
use java_rs::vm1::Vm; use java_rs::vm::Vm;
#[test] #[test]
fn if_cmp() { fn if_cmp() {
let mut vm = Vm::new("tests"); let mut stackframes = Vec::new();
let ret = vm.execute_virtual("testclasses.IfCmp", "i_is_1()Z", vec![]).unwrap(); let mut vm = Vm::new(&mut stackframes);
set_classpath("/Users/Shautvast/dev/java/tests");
let ret = vm.execute_virtual(&mut stackframes,"testclasses.IfCmp", "i_is_1()Z", vec![]).unwrap();
unsafe { unsafe {
if let Value::I32(b) = *ret.get() { if let Value::I32(b) = ret {
// internally a boolean is an int // internally a boolean is an int
assert_eq!(0, b); assert_eq!(0, b);
} else { } else {
println!("{:?}", *ret.get()); println!("{:?}", ret);
assert!(false) assert!(false)
} }
} }
@ -20,18 +22,20 @@ mod test {
#[test] #[test]
fn consts() { fn consts() {
let mut vm = Vm::new("tests"); let mut stackframes = Vec::new();
let mut vm = Vm::new(&mut stackframes);
set_classpath("/Users/Shautvast/dev/java/tests");
let ret = vm let ret = vm
.execute_static("testclasses.Const", "hello()Ljava/lang/String;", vec![]) .execute_static(&mut stackframes, "testclasses.Const", "hello()Ljava/lang/String;", vec![])
.unwrap(); .unwrap();
unsafe { unsafe {
if let Value::Ref(s) = &*ret.get() { if let Value::Ref(s) = ret {
// internally a boolean is an int // internally a boolean is an int
if let ObjectRef::Object(a) = &*s.get() { if let ObjectRef::Object(a) = s {
println!("{:?}", a); println!("{:?}", a);
} }
} else { } else {
println!("{:?}", *ret.get()); println!("{:?}", ret);
assert!(false) assert!(false)
} }
} }