Compare commits

...

10 commits

Author SHA1 Message Date
Shautvast
ccf7426086 finally picked it up again, one bytecode at a time 2024-05-16 17:42:04 +02:00
Shautvast
c271881a79 busy implementing hashcode 2023-12-07 17:20:19 +01:00
Shautvast
b714827f24 bugfix increment needs to be i16 not u16 2023-12-07 15:27:30 +01:00
Shautvast
a7d772c093 added opcodes 2023-12-06 17:52:59 +01:00
Shautvast
2149c8d3de bugfix pop order 2023-12-06 14:39:33 +01:00
Shautvast
7b6a01dfda added more dups 2023-12-02 11:54:18 +01:00
Shautvast
bab0e0d8c2 some arithmetic and some bitwise opcodes 2023-12-02 09:28:17 +01:00
Shautvast
d9985df822 primitiveClasses, fix arrays 2023-11-27 22:12:31 +01:00
Shautvast
dae02cec32 fixed jumping 2023-11-26 19:08:52 +01:00
Shautvast
715fa29cfd almost there, it seems... 2023-11-26 15:57:54 +01:00
17 changed files with 1514 additions and 308 deletions

View file

@ -10,10 +10,7 @@ pub(crate) struct TypeIndex {
impl TypeIndex {
pub(crate) fn new(type_name: String, index: usize) -> Self {
Self {
type_name,
index,
}
Self { type_name, index }
}
}
@ -23,14 +20,21 @@ pub struct Class {
pub initialized: bool,
pub name: String,
pub superclass: Option<ClassId>,
pub parents: LinkedList<ClassId>,
pub all_superclasses: LinkedList<ClassId>, // all superclasses in a flat list
pub interfaces: Vec<ClassId>,
pub all_interfaces: Vec<ClassId>, // all interfaces and their parents in a flat list
// lookup index and type from the name of the declared class and then field
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> // moved to classmanager
}
impl PartialEq for Class {
fn eq(&self, other: &Self) -> bool {
self.id == other.id
}
}
impl Class {
/// gets the number of non-static fields on the class
pub(crate) fn n_object_fields(&self) -> usize {

View file

@ -94,6 +94,7 @@ impl ClassDef {
}
}
// gets the class name if the index is a class ref index
pub fn cp_class_name(&self, index: &u16) -> &String {
let cr = self.cp_class_ref(index);
self.cp_utf8(cr)
@ -318,13 +319,13 @@ impl Field {
}
}
pub(crate) struct Method {
pub struct Method {
pub(crate) constant_pool: Rc<HashMap<u16, CpEntry>>,
access_flags: u16,
pub access_flags: u16,
name_index: u16,
descriptor_index: u16,
pub(crate) attributes: HashMap<String, AttributeType>,
pub code: Vec<Opcode>,
pub(crate) code: Vec<Opcode>,
}
impl Debug for Method {

View file

@ -1,8 +1,8 @@
use std::fs::{self};
use anyhow::{anyhow, Error};
use crate::vm::opcodes::Opcode;
use crate::vm::opcodes::Opcode::*;
use anyhow::{anyhow, Error};
#[cfg(target_family = "unix")]
pub const PATH_SEPARATOR: char = ':';
@ -17,7 +17,6 @@ pub const PATH_SEPARATOR: char = ';';
/// * [jar/zip]#[package_path]/[class].class
/// * [dir]/[package_path]/[class].class
pub fn find_class(classpath: &Vec<String>, class_name: &str) -> Result<String, Error> {
let class_name = &class_name.to_owned().replace(".", "/");
if class_name.starts_with("java")
|| class_name.starts_with("sun/")
|| class_name.starts_with("com/sun/")
@ -45,7 +44,6 @@ pub fn find_class(classpath: &Vec<String>, class_name: &str) -> Result<String, E
Err(anyhow!("Class not found {}", class_name))
}
// methods to read values from big-endian binary data
pub(crate) fn read_u8(data: &[u8], pos: &mut usize) -> u8 {
@ -73,6 +71,15 @@ pub(crate) fn read_u16(data: &[u8], pos: &mut usize) -> u16 {
)
}
pub(crate) fn read_i16(data: &[u8], pos: &mut usize) -> i16 {
*pos += 2;
i16::from_be_bytes(
data[*pos - 2..*pos]
.try_into()
.expect("slice with incorrect length"),
)
}
pub(crate) fn read_i32(data: &[u8], pos: &mut usize) -> i32 {
*pos += 4;
i32::from_be_bytes(
@ -119,8 +126,9 @@ pub(crate) fn read_f64(data: &[u8], pos: &mut usize) -> f64 {
}
pub(crate) fn read_tableswitch(data: &[u8], pos: &mut usize) -> Tableswitch {
while read_u8(data, pos) == 0 {}
*pos -= 1;
while *pos % 4 != 0 {
*pos += 1;
}
let default = read_i32(data, pos);
let low = read_i32(data, pos);
let high = read_i32(data, pos);
@ -128,19 +136,28 @@ pub(crate) fn read_tableswitch(data: &[u8], pos: &mut usize) -> Tableswitch {
for _ in low..=high {
offsets.push(read_i32(data, pos));
}
Tableswitch { default, low, high, offsets }
Tableswitch {
default,
low,
high,
offsets,
}
}
pub(crate) fn read_lookupswitch(data: &[u8], pos: &mut usize) -> Lookupswitch {
while read_u8(data, pos) == 0 {}
*pos -= 1;
while *pos % 4 != 0 {
*pos += 1;
}
let default = read_i32(data, pos);
let npairs = read_i32(data, pos);
let mut match_offset_pairs = vec![];
for _ in 0..npairs {
match_offset_pairs.push((read_i32(data, pos), read_i32(data, pos)));
}
Lookupswitch { default, match_offset_pairs }
Lookupswitch {
default,
match_offset_pairs,
}
}
pub(crate) fn read_wide_opcode(data: &[u8], pos: &mut usize) -> Opcode {
@ -161,7 +178,9 @@ pub(crate) fn read_wide_opcode(data: &[u8], pos: &mut usize) -> Opcode {
57 => WIDE_DSTORE(index),
58 => WIDE_ASTORE(index),
169 => WIDE_RET(index),
_ => { unreachable!("unknown opcode for WIDE") }
_ => {
unreachable!("unknown opcode for WIDE")
}
}
}
}

View file

@ -7,13 +7,17 @@ use std::rc::Rc;
use anyhow::Error;
use log::debug;
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};
use crate::classloader::code_parser::parse_code;
use crate::classloader::classdef::{
AttributeType, ClassDef, CpEntry, Exception, Field, Method, MethodCode,
};
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::opcode_parser::parse_opcodes;
pub mod classdef;
pub(crate) mod io;
mod code_parser;
mod opcode_parser;
pub(crate) fn get_classdef(classpath: &Vec<String>, class_name: &str) -> Result<ClassDef, Error> {
debug!("read class {} ", class_name);
@ -64,7 +68,11 @@ fn load_class(bytecode: Vec<u8>) -> Result<ClassDef, Error> {
let access_flags = read_u16(&bytecode, pos);
let this_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 super_class = if super_class != 0 {
Some(super_class)
} else {
None
};
let interfaces_count = read_u16(&bytecode, pos);
let mut interfaces = vec![];
for _ in 0..interfaces_count {
@ -96,8 +104,6 @@ fn load_class(bytecode: Vec<u8>) -> Result<ClassDef, Error> {
}
}
Ok(ClassDef::new(
minor_version,
major_version,
@ -232,6 +238,7 @@ fn read_method(
) -> Method {
let access_flags = read_u16(bytecode, index);
let name_index = read_u16(bytecode, index);
let name = constant_pool.get(&name_index);
let descriptor_index = read_u16(bytecode, index);
let attributes_count = read_u16(bytecode, index);
@ -242,9 +249,8 @@ fn read_method(
}
}
let code =
if let Some(AttributeType::Code(code)) = attributes.get("Code") {
parse_code(&code.opcodes)
let code = if let Some(AttributeType::Code(code)) = attributes.get("Code") {
parse_opcodes(&code.opcodes)
} else {
vec![]
};
@ -255,7 +261,7 @@ fn read_method(
name_index,
descriptor_index,
attributes,
code
code,
)
}
@ -326,4 +332,3 @@ fn read_attribute(
}
None
}

View file

@ -1,39 +1,50 @@
use std::collections::HashMap;
use log::debug;
use std::collections::BTreeMap;
use crate::classloader::io::{read_i32, read_lookupswitch, read_tableswitch, read_u16, read_u8, read_wide_opcode};
use crate::classloader::io::{
read_i16, read_i32, read_lookupswitch, read_tableswitch, read_u16, read_u8, read_wide_opcode,
};
use crate::vm::opcodes::Opcode::{self, *};
pub(crate) fn parse_code(opcodes: &[u8]) -> Vec<Opcode> {
let mut code: HashMap<u16, (u16, Opcode)> = HashMap::new();
pub(crate) fn parse_opcodes(opcodes: &[u8]) -> Vec<Opcode> {
let mut code: BTreeMap<u16, (u16, Opcode)> = BTreeMap::new();
let mut c = 0;
let mut opcode_index: u16 = 0;
// debug!("len {:?}", opcodes.len());
let mut index_before_read: u16;
while c < opcodes.len() {
index_before_read = c as u16;
let opcode = get_opcode(opcodes, &mut c);
code.insert(c as u16, (opcode_index, opcode));
code.insert(index_before_read, (opcode_index, opcode));
opcode_index += 1;
}
let code2 = code.clone(); //clone to look up
// for jumps, map index of opcode as u8 to index of opcode as enum
code.into_iter().map(|(_, (_, opcode))|
match opcode {
IFNULL(goto) => {
debug!("goto {:?}", goto);
debug!("{:?}", code2);
IFNULL(code2.get(&goto).unwrap().0)
}
IFNONNULL(goto) => {
debug!("goto {:?}", &goto);
debug!("{:?}", code2);
IFNONNULL(code2.get(&goto).unwrap().0)
}
debug!("{:?}", code);
code.into_iter()
.map(|(_, (_, opcode))| match opcode {
IFNULL(goto) => IFNULL(code2.get(&goto).unwrap().0),
IFNONNULL(goto) => IFNONNULL(code2.get(&goto).unwrap().0),
IF_ICMPEQ(goto) => IF_ICMPEQ(code2.get(&goto).unwrap().0),
IF_ICMPNE(goto) => IF_ICMPNE(code2.get(&goto).unwrap().0),
IF_ICMPGT(goto) => IF_ICMPGT(code2.get(&goto).unwrap().0),
IF_ICMPGE(goto) => IF_ICMPGE(code2.get(&goto).unwrap().0),
IF_ICMPLT(goto) => IF_ICMPLT(code2.get(&goto).unwrap().0),
IF_ICMPLE(goto) => IF_ICMPLE(code2.get(&goto).unwrap().0),
IFEQ(goto) => IFEQ(code2.get(&goto).unwrap().0),
IFNE(goto) => IFNE(code2.get(&goto).unwrap().0),
IFGT(goto) => IFGT(code2.get(&goto).unwrap().0),
IFGE(goto) => IFGE(code2.get(&goto).unwrap().0),
IFLT(goto) => IFLT(code2.get(&goto).unwrap().0),
IFLE(goto) => IFLE(code2.get(&goto).unwrap().0),
GOTO(goto) => GOTO(code2.get(&goto).unwrap().0),
IF_ACMPEQ(goto) => GOTO(code2.get(&goto).unwrap().0),
IF_ACMPNE(goto) => GOTO(code2.get(&goto).unwrap().0),
//TODO more jump instructions
_ => opcode
}
).collect()
_ => opcode,
})
.collect()
}
fn get_opcode(opcodes: &[u8], c: &mut usize) -> Opcode {
@ -162,6 +173,8 @@ fn get_opcode(opcodes: &[u8], c: &mut usize) -> Opcode {
121 => LSHL,
122 => ISHR,
123 => LSHR,
124 => IUSHR,
125 => LUSHR,
126 => IAND,
127 => LAND,
128 => IOR,
@ -189,26 +202,27 @@ fn get_opcode(opcodes: &[u8], c: &mut usize) -> Opcode {
150 => FCMPG,
151 => DCMPL,
152 => DCMPG,
153 => IFEQ(read_u16(opcodes, c)),
154 => IFNE(read_u16(opcodes, c)),
155 => IFLT(read_u16(opcodes, c)),
156 => IFGE(read_u16(opcodes, c)),
157 => IFGT(read_u16(opcodes, c)),
158 => IFLE(read_u16(opcodes, c)),
159 => IF_ICMPEQ(read_u16(opcodes, c)),
160 => IF_ICMPNE(read_u16(opcodes, c)),
161 => IF_ICMPLT(read_u16(opcodes, c)),
162 => IF_ICMPGE(read_u16(opcodes, c)),
163 => IF_ICMPGT(read_u16(opcodes, c)),
164 => IF_ICMPLE(read_u16(opcodes, c)),
165 => IF_ACMPEQ(read_u16(opcodes, c)),
166 => IF_ACMPNE(read_u16(opcodes, c)),
167 => GOTO(read_u16(opcodes, c)),
168 => JSR(read_u16(opcodes, c)),
153 => IFEQ(offset(opcodes, c)),
154 => IFNE(offset(opcodes, c)),
155 => IFLT(offset(opcodes, c)),
156 => IFGE(offset(opcodes, c)),
157 => IFGT(offset(opcodes, c)),
158 => IFLE(offset(opcodes, c)),
159 => IF_ICMPEQ(offset(opcodes, c)),
160 => IF_ICMPNE(offset(opcodes, c)),
161 => IF_ICMPLT(offset(opcodes, c)),
162 => IF_ICMPGE(offset(opcodes, c)),
163 => IF_ICMPGT(offset(opcodes, c)),
164 => IF_ICMPLE(offset(opcodes, c)),
165 => IF_ACMPEQ(offset(opcodes, c)),
166 => IF_ACMPNE(offset(opcodes, c)),
167 => GOTO(offset(opcodes, c)),
168 => JSR(offset(opcodes, c)),
169 => RET(read_u8(opcodes, c)),
170 => TABLESWITCH(read_tableswitch(opcodes, c)),
171 => LOOKUPSWITCH(read_lookupswitch(opcodes, c)),
172 => IRETURN,
173 => LRETURN,
174 => FRETURN,
175 => DRETURN,
176 => ARETURN,
@ -242,21 +256,19 @@ fn get_opcode(opcodes: &[u8], c: &mut usize) -> Opcode {
195 => MONITOREXIT,
196 => WIDE(Box::new(read_wide_opcode(opcodes, c))),
197 => MULTIANEWARRAY(read_u16(opcodes, c), read_u8(opcodes, c)),
198 => {
let j = read_u16(opcodes, c);
debug!("ifnull {}",*c as u16 + j - 3);
IFNULL(*c as u16 + j - 3)
}
199 => {
let j = read_u16(opcodes, c);
debug!("ifnonnull {} ", *c as u16 + j - 3);
IFNONNULL(*c as u16 + j - 3)
}
198 => IFNULL(offset(opcodes, c)),
199 => IFNONNULL(offset(opcodes, c)),
200 => GOTOW(read_i32(opcodes, c)),
201 => JSR_W(read_i32(opcodes, c)),
_ => panic!("{}", opcode_u8),
};
// debug!("{}: {:?}", c, opcode);
opcode
}
fn offset(opcodes: &[u8], c: &mut usize) -> u16 {
let j = read_i16(opcodes, c);
// debug!("JUMP TO {} + {}",c, j);
(*c as i16 + j - 3) as u16
}

View file

@ -8,12 +8,13 @@ use once_cell::sync::Lazy;
use crate::class::{Class, ClassId, TypeIndex};
use crate::classloader;
use crate::classloader::classdef::{ClassDef, Method, Modifier};
use crate::vm::object::{Object, ObjectRef};
use crate::value::Value;
use crate::value::Value::*;
use crate::vm::object::{Object, ObjectRef};
use crate::vm::runtime::Vm;
static PRIMITIVES: Lazy<Vec<&str>> = Lazy::new(|| vec!["B", "S", "I", "J", "F", "D", "Z", "J", "C"]);
static PRIMITIVES: Lazy<Vec<&str>> =
Lazy::new(|| vec!["B", "S", "I", "J", "F", "D", "Z", "J", "C"]);
//TODO less pubs
pub struct ClassManager {
@ -99,11 +100,11 @@ impl ClassManager {
if !PRIMITIVES.contains(&type_name.as_str()) {
type_name = type_name[1..type_name.len()].to_owned();
}
let id = self.get_or_new_id(name);
let id = self.get_or_new_id(name.into());
if !self.class_objects.contains_key(&id) {
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()));
instance.set(cls, "java/lang/Class", "name", Utf8(name.into()));
let instance = Ref(ObjectRef::Object(Rc::new(RefCell::new(instance))));
self.class_objects.insert(id, instance);
@ -112,9 +113,11 @@ impl ClassManager {
// in cache?
let id = self.names.get(&type_name);
match id {
Some(id) => if self.classes.get(id).is_none() {
Some(id) => {
if self.classes.get(id).is_none() {
self.add_class(&type_name);
}
}
None => {
self.add_class(&type_name);
}
@ -135,6 +138,7 @@ impl ClassManager {
}
pub fn get_class_by_name(&self, name: &str) -> Option<&Class> {
debug!("get class by name {}", name);
let id = self.names.get(name);
self.classes.get(id.unwrap())
}
@ -166,42 +170,62 @@ impl ClassManager {
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);
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()
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()
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()
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();
// initial values for static fields (before static init)
self.static_class_data.insert(this_classid, Self::set_field_data(&static_field_mapping));
let mut all_interfaces = Vec::new();
self.get_all_interfaces(&interface_ids, &mut all_interfaces);
self.classes.insert(this_classid, Class {
// initial values for static fields (before static init)
self.static_class_data
.insert(this_classid, Self::set_field_data(&static_field_mapping));
self.classes.insert(
this_classid,
Class {
id: this_classid,
initialized: false,
name: name.into(),
superclass: superclass_id,
parents,
all_superclasses: parents,
interfaces: interface_ids,
all_interfaces,
object_field_mapping,
static_field_mapping,
});
},
);
// add a new Class instance
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()));
instance.set(cls, "java/lang/Class", "name", Utf8(name.into()));
let instance = Ref(ObjectRef::Object(Rc::new(RefCell::new(instance))));
self.class_objects.insert(this_classid, instance);
@ -215,22 +239,38 @@ impl ClassManager {
this_classid
}
fn get_all_interfaces(&self, ids: &[ClassId], output: &mut Vec<ClassId>) {
for id in ids {
output.push(*id);
let interfaces = self.classes.get(id).map(|c| c.interfaces.to_vec());
if let Some(intf) = interfaces {
self.get_all_interfaces(&intf, output);
}
}
}
/// like described above
fn add_fields_for_this_or_parents(object_field_mapping: &mut HashMap<String, HashMap<String, TypeIndex>>,
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) {
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));
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_mappings.insert(
field_name.to_owned(),
TypeIndex::new(field.type_of().to_owned(), *static_field_map_index),
);
*static_field_map_index += 1;
}
}
@ -252,25 +292,28 @@ impl ClassManager {
}
/// loads the class and returns it's dependencies
fn load_class_and_deps(&mut self, name: &str) -> (ClassId, Vec<String>) {
debug!("load {}", name);
let id = self.get_or_new_id(name);
fn load_class_and_deps(&mut self, class_name: &str) -> (ClassId, Vec<String>) {
debug!("load {}", class_name);
let class_name = class_name.to_owned().replace(".", "/");
let id = self.get_or_new_id(class_name.clone());
let classdef = self.classdefs
.entry(id)
.or_insert_with(|| classloader::get_classdef(&self.classpath, name).expect("ClassNotFound"));
let classdef = self.classdefs.entry(id).or_insert_with(|| {
classloader::get_classdef(&self.classpath, class_name.as_str()).expect("ClassNotFound")
});
(id, inspect_dependencies(classdef))
}
fn get_or_new_id(&mut self, name: &str) -> ClassId {
let id = *self.names.entry(name.to_string()).or_insert_with(|| {
fn get_or_new_id(&mut self, name: String) -> ClassId {
let id = *self.names.entry(name).or_insert_with(|| {
self.current_id += 1;
self.current_id
});
id
}
pub(crate) fn set_field_data(field_mapping: &HashMap<String, HashMap<String, TypeIndex>>) -> Vec<Value> {
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 {
@ -344,27 +387,100 @@ mod test {
// give class C a fields called value
let mut c_fields = HashMap::new();
c_fields.insert("value1".to_owned(), Field::new(constant_pool.clone(), 0, 9, 5, HashMap::new(), 0));
c_fields.insert("value2".to_owned(), Field::new(constant_pool.clone(), 0, 10, 5, HashMap::new(), 0));
c_fields.insert(
"value1".to_owned(),
Field::new(constant_pool.clone(), 0, 9, 5, HashMap::new(), 0),
);
c_fields.insert(
"value2".to_owned(),
Field::new(constant_pool.clone(), 0, 10, 5, HashMap::new(), 0),
);
// Class needs a public (non-static) field called name
let mut class_fields = HashMap::new();
class_fields.insert("name".to_owned(), Field::new(constant_pool.clone(), 1, 2, 5, HashMap::new(), 0));
class_fields.insert(
"name".to_owned(),
Field::new(constant_pool.clone(), 1, 2, 5, HashMap::new(), 0),
);
let mut classdefs = HashMap::new();
classdefs.insert(1, ClassDef::new(0, 0, constant_pool.clone(), 0, 0, None, vec![], c_fields, HashMap::new(), HashMap::new()));
classdefs.insert(
1,
ClassDef::new(
0,
0,
constant_pool.clone(),
0,
0,
None,
vec![],
c_fields,
HashMap::new(),
HashMap::new(),
),
);
// preload java.lang.String
classdefs.insert(2, ClassDef::new(0, 0, constant_pool.clone(), 0, 6, None, vec![], HashMap::new(), HashMap::new(), HashMap::new()));
classdefs.insert(
2,
ClassDef::new(
0,
0,
constant_pool.clone(),
0,
6,
None,
vec![],
HashMap::new(),
HashMap::new(),
HashMap::new(),
),
);
// preload java.lang.Class
classdefs.insert(3, ClassDef::new(0, 0, constant_pool, 0, 8, None, vec![], class_fields, HashMap::new(), HashMap::new()));
classdefs.insert(
3,
ClassDef::new(
0,
0,
constant_pool,
0,
8,
None,
vec![],
class_fields,
HashMap::new(),
HashMap::new(),
),
);
let mut classes = HashMap::new();
let mut class_field_mapping = HashMap::new();
let mut fields_declared_by_java_lang_class = HashMap::new();
fields_declared_by_java_lang_class.insert("name".to_owned(), TypeIndex { type_name: "java/lang/String".into(), index: 0 });
class_field_mapping.insert("java/lang/Class".to_owned(), fields_declared_by_java_lang_class);
classes.insert(3, Class { id: 3, initialized: false, name: "".into(), superclass: None, parents: LinkedList::new(), interfaces: vec![], object_field_mapping: class_field_mapping, static_field_mapping: HashMap::new() });
fields_declared_by_java_lang_class.insert(
"name".to_owned(),
TypeIndex {
type_name: "java/lang/String".into(),
index: 0,
},
);
class_field_mapping.insert(
"java/lang/Class".to_owned(),
fields_declared_by_java_lang_class,
);
classes.insert(
3,
Class {
id: 3,
initialized: false,
name: "".into(),
superclass: None,
all_superclasses: LinkedList::new(),
interfaces: vec![],
all_interfaces: vec![],
object_field_mapping: class_field_mapping,
static_field_mapping: HashMap::new(),
},
);
let mut cm = ClassManager {
static_class_data: HashMap::new(),
@ -379,7 +495,25 @@ mod test {
let c_id = cm.add_class("C");
let loaded_class = cm.classes.get(&c_id).unwrap();
assert_eq!(0, loaded_class.object_field_mapping.get("C").unwrap().get("value1").unwrap().index);
assert_eq!(1, loaded_class.object_field_mapping.get("C").unwrap().get("value2").unwrap().index);
assert_eq!(
0,
loaded_class
.object_field_mapping
.get("C")
.unwrap()
.get("value1")
.unwrap()
.index
);
assert_eq!(
1,
loaded_class
.object_field_mapping
.get("C")
.unwrap()
.get("value2")
.unwrap()
.index
);
}
}

View file

@ -1,5 +1,5 @@
mod class;
pub mod classloader;
pub mod classmanager;
mod value;
mod class;
pub mod vm;

View file

@ -1,7 +1,10 @@
use java_rs::vm::runtime::Vm;
fn main() {
let vm = Vm::new();
vm.run("/Users/Shautvast/dev/java/tests", "testclasses.Main", "main([Ljava/lang/String;)V");
let mut vm = Vm::new();
vm.run(
"/Users/Shautvast/dev/java.rs/tests",
"testclasses/Main",
"main([Ljava/lang/String;)V",
);
}

View file

@ -1,4 +1,5 @@
use crate::vm::object::ObjectRef;
use std::ops::{Add, AddAssign};
#[derive(Debug, Clone)]
pub enum Value {
@ -12,14 +13,36 @@ pub enum Value {
F32(f32),
F64(f64),
BOOL(bool),
CHAR(char),
CHAR(i32),
// objects and arrays
Ref(ObjectRef),
// special object
Utf8(String),
}
pub enum ComputationalType {
C1 = 1,
C2 = 2,
}
impl Value {
pub fn category_as_u8(&self) -> u8 {
self.category() as u8
}
pub fn category(&self) -> ComputationalType {
match self {
Value::Void
| Value::Null
| Value::I32(_)
| Value::F32(_)
| Value::BOOL(_)
| Value::CHAR(_)
| Value::Ref(_)
| Value::Utf8(_) => ComputationalType::C1,
Value::I64(_) | Value::F64(_) => ComputationalType::C2,
}
}
// panics if not correct type
pub fn into_i32(self) -> i32 {
if let Value::I32(v) = self {
@ -29,8 +52,42 @@ impl Value {
}
}
pub fn into_i64(self) -> i64 {
if let Value::I64(v) = self {
v
} else {
panic!("{:?} is not I64", self);
}
}
pub fn into_f32(self) -> f32 {
if let Value::F32(v) = self {
v
} else {
panic!("{:?} is not F32", self);
}
}
pub fn into_f64(self) -> f64 {
if let Value::F64(v) = self {
v
} else {
panic!("{:?} is not F64", self);
}
}
pub fn into_object(self) -> ObjectRef {
if let Value::Ref(v) = self {
if let Value::Null = self {
ObjectRef::Null
} else if let Value::Ref(v) = self {
v
} else {
panic!("{:?}", self);
}
}
pub fn into_utf8(self) -> String {
if let Value::Utf8(v) = self {
v
} else {
panic!();

View file

@ -1,15 +1,15 @@
use anyhow::{anyhow, Error};
use crate::vm::object::ObjectRef::*;
use anyhow::{anyhow, Error};
use crate::value::Value;
use crate::value::Value::*;
use crate::vm::object::ObjectRef;
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 Value::Null = arrayref {
return Err(anyhow!("NullpointerException"));
}
if let Ref(objectref) = arrayref {
@ -50,6 +50,9 @@ pub(crate) fn array_load(index: Value, arrayref: Value) -> Result<Value, Error>
Object(_) => {
panic!("should be array")
} //throw error?
ObjectRef::Null => {
return Ok(Value::Null);
}
}
}
}
@ -57,7 +60,7 @@ pub(crate) fn array_load(index: Value, arrayref: Value) -> Result<Value, Error>
}
pub(crate) fn array_store(value: Value, index: Value, arrayref: Value) -> Result<(), Error> {
if let Null = arrayref {
if let Value::Null = arrayref {
return Err(anyhow!("NullpointerException"));
}
@ -94,9 +97,9 @@ pub(crate) fn array_store(value: Value, index: Value, arrayref: Value) -> Result
unreachable!()
}
}
CharArray(ref mut array) => unsafe {
CharArray(ref mut array) => {
if let I32(value) = value {
array[index as usize] = char::from_u32_unchecked(value as u32);
array[index as usize] = value
} else {
unreachable!()
}
@ -136,7 +139,7 @@ pub(crate) fn array_store(value: Value, index: Value, arrayref: Value) -> Result
unreachable!()
}
}
Object(_) | Class(_) => {} //throw error?
Object(_) | Class(_) | ObjectRef::Null => {} //throw error?
}
}
} else {

View file

@ -1,4 +1,5 @@
mod array;
mod native;
pub(crate) mod object;
pub(crate) mod opcodes;
pub mod runtime;

243
src/vm/native.rs Normal file
View file

@ -0,0 +1,243 @@
#![allow(non_snake_case)]
use std::cell::RefCell;
use std::collections::HashMap;
use std::rc::Rc;
use anyhow::Error;
use log::debug;
use once_cell::sync::Lazy;
use crate::classmanager::ClassManager;
use crate::value::Value;
use crate::value::Value::{Utf8, Void, I32};
use crate::vm::object::ObjectRef::Object;
use crate::vm::object::{self, Array, ObjectRef};
use crate::vm::runtime::Stackframe;
const primitive_name_classes: Lazy<HashMap<&str, &str>> = Lazy::new(|| {
let mut mapping = HashMap::new();
mapping.insert("B", "byte");
mapping.insert("S", "short");
mapping.insert("I", "int");
mapping.insert("J", "long");
mapping.insert("F", "float");
mapping.insert("D", "double");
mapping.insert("C", "char");
mapping.insert("Z", "boolean");
mapping
});
static mut PRIMITIVE_CLASSES: Lazy<HashMap<String, Value>> = Lazy::new(|| HashMap::new());
pub fn invoke_native(
class_manager: &mut ClassManager,
class_name: &str,
method_name: &str,
args: Vec<Value>,
) -> Result<Value, Error> {
debug!("native {}.{}", class_name, method_name);
match class_name {
"java/lang/Class" => java_lang_Class(class_manager, method_name, args),
"java/lang/Object" => java_lang_Object(method_name, args),
"java/lang/System" => java_lang_System(method_name),
"jdk/internal/misc/Unsafe" => jdk_internal_misc_Unsafe(method_name),
"jdk/internal/util/SystemProps$Raw" => {
jdk_internal_util_SystemProps_Raw(class_manager, method_name)
}
_ => unimplemented!("{}", class_name),
}
}
fn java_lang_Class(
class_manager: &mut ClassManager,
method_name: &str,
args: Vec<Value>,
) -> Result<Value, Error> {
Ok(match method_name {
"desiredAssertionStatus0(Ljava/lang/Class;)Z" => Value::BOOL(false),
"getPrimitiveClass(Ljava/lang/String;)Ljava/lang/Class;" => {
get_primitive_class(class_manager, args)
}
_ => Void,
})
}
fn java_lang_Object(method_name: &str, args: Vec<Value>) -> Result<Value, Error> {
Ok(match method_name {
"hashCode()I" => {
if let Value::Ref(object_ref) = args.get(0).unwrap() {
if let Object(p) = object_ref {
I32(p.borrow().id as i32) // system hashcode
} else {
I32(0) //todo implement for arrays
}
} else {
unreachable!()
}
}
_ => {
unimplemented!("{}", method_name);
}
})
}
fn java_lang_System(method_name: &str) -> Result<Value, Error> {
Ok(match method_name {
_ => Void,
})
}
fn get_primitive_class(class_manager: &mut ClassManager, args: Vec<Value>) -> Value {
if let Utf8(primitive) = args.get(0).unwrap().to_owned() {
unsafe {
PRIMITIVE_CLASSES
.entry(primitive.clone())
.or_insert_with(|| {
let cls = class_manager.get_class_by_name("java/lang/Class").unwrap();
let mut instance = object::Object::new(cls);
instance.set(
cls,
"java/lang/Class",
primitive_name_classes.get(primitive.as_str()).unwrap(),
Utf8("name".into()),
);
Value::Ref(Object(Rc::new(RefCell::new(instance))))
});
}
}
Value::Null
}
fn jdk_internal_misc_Unsafe(method_name: &str) -> Result<Value, Error> {
Ok(match method_name {
"arrayBaseOffset0(Ljava/lang/Class;)I" => I32(0), //TODO surely this is not right
"arrayIndexScale0(Ljava/lang/Class;)I" => I32(0), //TODO surely this is not right
_ => Void,
})
}
fn jdk_internal_util_SystemProps_Raw(
class_manager: &mut ClassManager,
method_name: &str,
) -> Result<Value, Error> {
match method_name {
"platformProperties()[Ljava/lang/String;" => platformProperties(),
"cmdProperties()Ljava/util/HashMap;" => cmdProps(class_manager), //TODO ability to instantiate classes here
"vmProperties()[Ljava/lang/String;" => vmProperties(),
_ => Ok(Void),
}
}
fn cmdProps(class_manager: &mut ClassManager) -> Result<Value, Error> {
class_manager.load_class_by_name("java/util/HashMap");
let hashmap_class = class_manager
.get_class_by_name("java/util/HashMap")
.unwrap();
let hashmap = Value::Ref(Object(Rc::new(RefCell::new(object::Object::new(
hashmap_class,
))))); // this is convoluted
Stackframe::new(vec![hashmap.clone()]).run(class_manager, hashmap_class.id, "<init>()V");
Ok(hashmap)
}
fn vmProperties() -> Result<Value, Error> {
let props: Lazy<Vec<String>> = Lazy::new(|| {
let vec: Vec<String> = Vec::new();
//TODO insert some values
vec
});
Ok(Value::Ref(ObjectRef::StringArray(Array::from_vec(
props.to_vec(),
))))
}
fn platformProperties() -> Result<Value, Error> {
let props: Lazy<Vec<String>> = Lazy::new(|| {
let mut vec: Vec<String> = Vec::new();
//TODO set correct values
vec.push("display_country".into()); //null in jdk21
vec.push("display_language".into()); //null in jdk21
vec.push("display_script".into()); //null in jdk21
vec.push("display_variant".into()); //null in jdk21
vec.push("UTF-8".into());
{
#[cfg(target_family = "unix")]
vec.push("/".into());
#[cfg(target_family = "windows")]
vec.push("\\");
}
vec.push("format_country".into()); //null in jdk21
vec.push("format_language".into()); //null in jdk21
vec.push("format_script".into()); //null in jdk21
vec.push("format_variant".into()); //null in jdk21
vec.push("ftp_nonProxyHosts".into());
if let Ok(ftp_proxy) = std::env::var("ftp_proxy") {
vec.push(ftp_proxy.to_owned()); //TODO
vec.push(ftp_proxy);
} else {
vec.push("".to_owned());
vec.push("".to_owned());
}
vec.push("http_nonProxyHosts".into());
if let Ok(http_proxy) = std::env::var("http_proxy") {
vec.push(http_proxy.to_owned());
vec.push(http_proxy); //TODO
} else {
vec.push("".to_owned());
vec.push("".to_owned());
}
if let Ok(https_proxy) = std::env::var("https_proxy") {
vec.push(https_proxy.to_owned());
vec.push(https_proxy);
} else {
vec.push("".to_owned());
vec.push("".to_owned());
}
vec.push(std::env::temp_dir().display().to_string());
{
#[cfg(target_family = "unix")]
vec.push("\n".into());
#[cfg(target_family = "windows")]
vec.push("\r\n");
}
vec.push(whoami::platform().to_string());
vec.push(whoami::devicename());
vec.push("os_version".into());
{
#[cfg(target_family = "unix")]
vec.push(":".into());
#[cfg(target_family = "windows")]
vec.push(";".into());
}
vec.push("socksNonProxyHosts".into());
vec.push("socksProxyHost".into());
vec.push("socksProxyPort".into());
vec.push("UTF-8".into());
vec.push("UTF-8".into());
vec.push("sun_arch_abi".into());
vec.push("sun_arch_data_model".into());
vec.push("sun_cpu_endian".into()); //null in jdk21
vec.push("sun_cpu_isalist".into()); //null in jdk21
vec.push("sun_io_unicode_encoding".into()); //null in jdk21
vec.push("sun_jnu_encoding".into()); //null in jdk21
vec.push("sun_os_patch_level".into()); //null in jdk21
if let Ok(curdir) = std::env::current_dir() {
vec.push(curdir.display().to_string());
}
let home = std::env::home_dir().unwrap();
vec.push(home.display().to_string());
vec.push(whoami::username());
vec.push("FIXED_LENGTH".into());
vec
});
Ok(Value::Ref(ObjectRef::StringArray(Array::from_vec(
props.to_vec(),
))))
}

View file

@ -1,25 +1,100 @@
use std::cell::RefCell;
use std::rc::Rc;
use crate::class::{Class, ClassId};
use crate::value::Value;
use crate::vm::object::ObjectRef::*;
use log::debug;
use rand::random;
use crate::class::{Class, ClassId};
use crate::vm::object::ObjectRef::*;
use crate::value::Value;
use std::cell::RefCell;
use std::fmt::{Debug, Formatter, Pointer};
use std::rc::Rc;
#[derive(Debug, Clone)]
#[derive(Clone, PartialEq)]
pub struct Array<T>
where
T: Clone,
{
id: u32,
data: Vec<T>,
}
impl<T> Array<T>
where
T: Clone,
{
pub fn new(init: T, size: usize) -> Self {
Self {
id: random::<u32>() >> 1,
data: vec![init; size],
}
}
pub fn from_vec(vec: Vec<T>) -> Self {
Self {
id: random::<u32>() >> 1,
data: vec,
}
}
pub fn len(&self) -> usize {
self.data.len()
}
}
impl<T> std::ops::Index<usize> for Array<T>
where
T: Clone,
{
type Output = T;
fn index(&self, index: usize) -> &Self::Output {
&self.data[index]
}
}
impl<T> std::ops::IndexMut<usize> for Array<T>
where
T: Clone,
{
fn index_mut(&mut self, index: usize) -> &mut Self::Output {
&mut self.data[index]
}
}
#[derive(Clone, PartialEq)]
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>),
ByteArray(Array<i8>),
//maybe use arrays?
ShortArray(Array<i16>),
IntArray(Array<i32>),
LongArray(Array<i64>),
FloatArray(Array<f32>),
DoubleArray(Array<f64>),
BooleanArray(Array<bool>),
CharArray(Array<i32>),
StringArray(Array<String>),
ObjectArray(ClassId, Vec<ObjectRef>),
Object(Rc<RefCell<Object>>),
Class(Class),
Null,
}
impl Debug for ObjectRef {
fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
let name = match self {
ByteArray(a) => format!("[B;{}", a.len()),
ShortArray(a) => format!("[S;{}", a.len()),
IntArray(a) => format!("[I;{}", a.len()),
LongArray(a) => format!("[J;{}", a.len()),
FloatArray(a) => format!("[F;{}", a.len()),
DoubleArray(a) => format!("[D;{}", a.len()),
BooleanArray(a) => format!("[Z;{}", a.len()),
CharArray(a) => format!("[C;{}", a.len()),
StringArray(a) => format!("[Ljava/lang/String;{}]", a.len()),
ObjectArray(_, a) => format!("[[Lx;{}]", a.len()),
Object(_) => "L".into(),
Class(_) => "Class".into(),
ObjectRef::Null => "null".into(),
};
write!(f, "{}", name)
}
}
impl ObjectRef {
@ -35,7 +110,7 @@ impl ObjectRef {
CharArray(d) => d.len(),
StringArray(d) => d.len(),
ObjectArray(_, d) => d.len(),
_ => unreachable!("not an array {:?}", self)
_ => unreachable!("not an array {:?}", self),
}
}
}
@ -53,29 +128,29 @@ pub enum ArrayType {
impl ObjectRef {
pub fn new_object_array(class: &Class, size: usize) -> Self {
ObjectArray(class.id, Vec::with_capacity(size))
ObjectArray(class.id, vec![ObjectRef::Null; size])
}
pub fn new_array(arraytype: u8, size: usize) -> Self {
match arraytype {
8 => ByteArray(Vec::with_capacity(size)),
9 => ShortArray(Vec::with_capacity(size)),
10 => IntArray(Vec::with_capacity(size)),
11 => LongArray(Vec::with_capacity(size)),
6 => FloatArray(Vec::with_capacity(size)),
7 => DoubleArray(Vec::with_capacity(size)),
4 => BooleanArray(Vec::with_capacity(size)),
5 => CharArray(Vec::with_capacity(size)),
_ => unreachable!("impossible array type")
8 => ByteArray(Array::new(0, size)),
9 => ShortArray(Array::new(0, size)),
10 => IntArray(Array::new(0, size)),
11 => LongArray(Array::new(0, size)),
6 => FloatArray(Array::new(0_f32, size)),
7 => DoubleArray(Array::new(0_f64, size)),
4 => BooleanArray(Array::new(false, size)),
5 => CharArray(Array::new(0, size)),
_ => unreachable!("impossible array type"),
}
}
pub fn new_int_array(size: usize) -> Self {
IntArray(Vec::with_capacity(size))
IntArray(Array::new(0, size))
}
pub fn new_byte_array(d: Vec<u8>) -> Self {
ByteArray(into_vec_i8(d))
ByteArray(Array::from_vec(into_vec_i8(d)))
}
}
@ -101,12 +176,18 @@ pub struct Object {
pub data: Vec<Value>,
}
impl PartialEq for Object {
fn eq(&self, other: &Self) -> bool {
self.id == other.id
}
}
// object, not array
impl Object {
pub fn new(class: &Class) -> Self {
let instance_data = Object::init_fields(class);
Self {
id: random(),
id: random::<u32>() >> 1,
class_id: class.id,
data: instance_data,
}
@ -135,7 +216,13 @@ impl Object {
field_data
}
pub fn set(&mut self, runtime_type: &Class, declared_type: &str, field_name: &str, value: Value) {
pub fn set(
&mut self,
runtime_type: &Class,
declared_type: &str,
field_name: &str,
value: Value,
) {
debug!("set {:?}.{}", runtime_type.name, field_name);
let type_index = runtime_type
.object_field_mapping
@ -153,7 +240,10 @@ impl Object {
.unwrap()
.get(field_name)
.unwrap();
debug!("get {:?}:{}.{}:{} @{}", runtime_type, declared_type, field_name, type_index.type_name, type_index.index);
debug!(
"get {:?}:{}.{}:{} @{}",
runtime_type, declared_type, field_name, type_index.type_name, type_index.index
);
debug!("from data {:?}", self.data);
self.data.get(type_index.index).unwrap()
}

View file

@ -85,6 +85,8 @@ pub(crate) enum Opcode {
LSHL,
ISHR,
LSHR,
IUSHR,
LUSHR,
IAND,
LAND,
IOR,
@ -134,6 +136,7 @@ pub(crate) enum Opcode {
TABLESWITCH(Tableswitch),
LOOKUPSWITCH(Lookupswitch),
IRETURN,
LRETURN,
FRETURN,
DRETURN,
ARETURN,
@ -162,7 +165,4 @@ pub(crate) enum Opcode {
IFNONNULL(u16),
GOTOW(i32),
JSR_W(i32),
}

File diff suppressed because it is too large Load diff

View file

@ -1,18 +1,23 @@
mod test {
use java_rs::vm::runtime::Vm;
#[test]
fn if_cmp() {
let vm = Vm::new();
vm.run("/Users/Shautvast/dev/java/tests", "testclasses.IfCmp", "i_is_1()Z");
vm.run(
"/Users/Shautvast/dev/java/tests",
"testclasses.IfCmp",
"i_is_1()Z",
);
}
#[test]
fn consts() {
let vm = Vm::new();
vm.run("/Users/Shautvast/dev/java/tests", "testclasses.Const", "hello()Ljava/lang/String;")
vm.run(
"/Users/Shautvast/dev/java/tests",
"testclasses.Const",
"hello()Ljava/lang/String;",
)
}
}

View file

@ -1,7 +1,7 @@
mod test {
use java_rs::classloader::classdef::{Method, Modifier};
use std::collections::HashMap;
use std::rc::Rc;
use java_rs::classloader::classdef::{Method, Modifier};
#[test]
fn access_flags() {
@ -11,13 +11,10 @@ mod test {
0,
0,
HashMap::new(),
vec![],
);
assert!(m.is(Modifier::Public));
assert!(m.is(Modifier::Static));
assert!(!m.is(Modifier::Private));
}
}