Bytecode work
This commit is contained in:
@@ -14,40 +14,106 @@ import poset:*
|
||||
|
||||
|
||||
adt byte_inst {
|
||||
nop
|
||||
nop,
|
||||
imm: imm,
|
||||
ld: ld,
|
||||
st: st,
|
||||
call,
|
||||
ret: ret
|
||||
}
|
||||
obj imm {
|
||||
var reg: int
|
||||
var val: int
|
||||
}
|
||||
obj ld {
|
||||
var reg: int
|
||||
}
|
||||
obj st {
|
||||
var to_reg: int
|
||||
var from_reg: int
|
||||
}
|
||||
obj ret {
|
||||
var reg: int
|
||||
}
|
||||
|
||||
fun bytecode_to_string(bytecode: ref vector<byte_inst>): string {
|
||||
var res = string()
|
||||
bytecode.for_each(fun(b: byte_inst) {
|
||||
match (b) {
|
||||
byte_inst::nop() res += "nop"
|
||||
}
|
||||
res += "\n"
|
||||
})
|
||||
return res
|
||||
fun to_string(b: byte_inst): string {
|
||||
match (b) {
|
||||
byte_inst::nop() return string("nop")
|
||||
byte_inst::imm(i) return to_string(i.reg) + " = imm " + i.val
|
||||
byte_inst::ld(l) return to_string(l.reg) + " = ld "
|
||||
byte_inst::st(s) return string("st(") + s.to_reg + " <= " + s.from_reg + ")"
|
||||
byte_inst::call() return string("call")
|
||||
byte_inst::ret(r) return string("ret ") + r.reg
|
||||
}
|
||||
return string("Missed byte_inst case in to_string")
|
||||
}
|
||||
|
||||
fun generate_bytecode(name_ast_map: ref map<string, pair<*tree<symbol>,*ast_node>>, ast_to_syntax_in: ref map<*ast_node, *tree<symbol>> ): vector<byte_inst> {
|
||||
fun bytecode_to_string(bytecode: ref vector<basic_block>): string {
|
||||
return string("\n").join(bytecode.map(fun(bb: ref basic_block): string return bb.to_string();))
|
||||
}
|
||||
|
||||
fun basic_block(name: ref string): basic_block {
|
||||
var to_ret.construct(name): basic_block
|
||||
return to_ret
|
||||
}
|
||||
obj basic_block (Object) {
|
||||
var name: string
|
||||
var instructions: vector<byte_inst>
|
||||
|
||||
fun construct(): *basic_block {
|
||||
instructions.construct()
|
||||
name.construct()
|
||||
return this
|
||||
}
|
||||
fun construct(name_in: ref string): *basic_block {
|
||||
instructions.construct()
|
||||
name.copy_construct(&name_in)
|
||||
return this
|
||||
}
|
||||
fun copy_construct(old: *basic_block) {
|
||||
instructions.copy_construct(&old->instructions)
|
||||
name.copy_construct(&old->name)
|
||||
}
|
||||
fun operator=(other: ref basic_block) {
|
||||
destruct()
|
||||
copy_construct(&other)
|
||||
}
|
||||
fun destruct() {
|
||||
instructions.destruct()
|
||||
name.destruct()
|
||||
}
|
||||
fun to_string(): string {
|
||||
var res = name + ":\n"
|
||||
instructions.for_each(fun(b: byte_inst) {
|
||||
res += "\t" + to_string(b) + "\n"
|
||||
})
|
||||
return res
|
||||
}
|
||||
}
|
||||
|
||||
fun generate_bytecode(name_ast_map: ref map<string, pair<*tree<symbol>,*ast_node>>, ast_to_syntax_in: ref map<*ast_node, *tree<symbol>> ): vector<basic_block> {
|
||||
var generator.construct(): bytecode_generator
|
||||
return generator.generate_bytecode(name_ast_map, ast_to_syntax_in)
|
||||
}
|
||||
|
||||
obj bytecode_generator (Object) {
|
||||
var reg_counter: int
|
||||
var id_counter: int
|
||||
var ast_name_map: hash_map<*ast_node, string>
|
||||
var bytecode: vector<byte_inst>
|
||||
var blocks: vector<basic_block>
|
||||
fun construct(): *bytecode_generator {
|
||||
reg_counter = 0
|
||||
id_counter = 0
|
||||
ast_name_map.construct()
|
||||
bytecode.construct()
|
||||
blocks.construct()
|
||||
|
||||
return this
|
||||
}
|
||||
fun copy_construct(old: *bytecode_generator) {
|
||||
reg_counter = old->reg_counter
|
||||
id_counter = old->id_counter
|
||||
ast_name_map.copy_construct(&old->ast_name_map)
|
||||
bytecode.copy_construct(&old->bytecode)
|
||||
blocks.copy_construct(&old->blocks)
|
||||
}
|
||||
fun operator=(other: ref bytecode_generator) {
|
||||
destruct()
|
||||
@@ -55,10 +121,11 @@ obj bytecode_generator (Object) {
|
||||
}
|
||||
fun destruct() {
|
||||
ast_name_map.destruct()
|
||||
bytecode.destruct()
|
||||
blocks.destruct()
|
||||
}
|
||||
fun get_id(): string return to_string(id_counter++);
|
||||
fun generate_bytecode(name_ast_map: map<string, pair<*tree<symbol>,*ast_node>>, ast_to_syntax_in: map<*ast_node, *tree<symbol>> ): vector<byte_inst> {
|
||||
fun get_reg(): int return reg_counter++;
|
||||
fun generate_bytecode(name_ast_map: map<string, pair<*tree<symbol>,*ast_node>>, ast_to_syntax_in: map<*ast_node, *tree<symbol>> ): vector<basic_block> {
|
||||
|
||||
// iterate through asts
|
||||
name_ast_map.for_each(fun(name: string, tree_pair: pair<*tree<symbol>,*ast_node>) {
|
||||
@@ -98,92 +165,149 @@ obj bytecode_generator (Object) {
|
||||
}
|
||||
})
|
||||
})
|
||||
return bytecode
|
||||
return blocks
|
||||
}
|
||||
fun generate_function_definition(node: *ast_node) {
|
||||
bytecode.add(byte_inst::nop())
|
||||
fun generate_function_definition(node: *ast_node): int {
|
||||
blocks.add(basic_block(get_name(node)))
|
||||
generate(node->function.body_statement)
|
||||
return -1
|
||||
}
|
||||
fun generate_declaration_statement(node: *ast_node) {
|
||||
fun generate_declaration_statement(node: *ast_node): int {
|
||||
var identifier = node->declaration_statement.identifier
|
||||
var ident_type = identifier->identifier.type
|
||||
if (identifier->identifier.is_extern)
|
||||
bytecode.add(byte_inst::nop())
|
||||
if (identifier->identifier.is_extern) {
|
||||
}
|
||||
if (node->declaration_statement.expression) {
|
||||
bytecode.add(byte_inst::nop())
|
||||
generate(node->declaration_statement.expression)
|
||||
}
|
||||
return -1
|
||||
}
|
||||
fun generate_assignment_statement(node: *ast_node) {
|
||||
bytecode.add(byte_inst::nop())
|
||||
fun generate_assignment_statement(node: *ast_node): int {
|
||||
var to = generate(node->assignment_statement.to)
|
||||
var from = generate(node->assignment_statement.from)
|
||||
emit_st(to, from)
|
||||
return -1
|
||||
}
|
||||
fun generate_if_statement(node: *ast_node) {
|
||||
bytecode.add(byte_inst::nop())
|
||||
fun generate_if_statement(node: *ast_node): int {
|
||||
generate(node->if_statement.condition)
|
||||
generate(node->if_statement.then_part)
|
||||
if (node->if_statement.else_part)
|
||||
generate(node->if_statement.else_part)
|
||||
return -1
|
||||
}
|
||||
fun generate_while_loop(node: *ast_node) {
|
||||
bytecode.add(byte_inst::nop())
|
||||
fun generate_while_loop(node: *ast_node): int {
|
||||
generate(node->while_loop.condition)
|
||||
generate(node->while_loop.statement)
|
||||
return -1
|
||||
}
|
||||
fun generate_for_loop(node: *ast_node) {
|
||||
bytecode.add(byte_inst::nop())
|
||||
fun generate_for_loop(node: *ast_node): int {
|
||||
if (node->for_loop.init)
|
||||
generate(node->for_loop.init)
|
||||
if (node->for_loop.condition)
|
||||
generate(node->for_loop.condition)
|
||||
if (node->for_loop.update)
|
||||
generate(node->for_loop.update)
|
||||
generate(node->for_loop.body)
|
||||
return -1
|
||||
}
|
||||
fun generate_identifier(node: *ast_node) {
|
||||
bytecode.add(byte_inst::nop())
|
||||
fun generate_identifier(node: *ast_node): int {
|
||||
return emit_ld()
|
||||
}
|
||||
fun generate_return_statement(node: *ast_node) {
|
||||
fun generate_return_statement(node: *ast_node): int {
|
||||
if (node->return_statement.return_value)
|
||||
bytecode.add(byte_inst::nop())
|
||||
emit_ret(generate(node->return_statement.return_value))
|
||||
else
|
||||
bytecode.add(byte_inst::nop())
|
||||
emit_ret(-1)
|
||||
return -1
|
||||
}
|
||||
fun generate_branching_statement(node: *ast_node) {
|
||||
fun generate_branching_statement(node: *ast_node): int {
|
||||
match(node->branching_statement.b_type) {
|
||||
branching_type::break_stmt() bytecode.add(byte_inst::nop())
|
||||
branching_type::continue_stmt() bytecode.add(byte_inst::nop())
|
||||
branching_type::break_stmt() blocks.last().instructions.add(byte_inst::nop())
|
||||
branching_type::continue_stmt() blocks.last().instructions.add(byte_inst::nop())
|
||||
}
|
||||
return -1
|
||||
}
|
||||
fun generate_cast(node: *ast_node) {
|
||||
bytecode.add(byte_inst::nop())
|
||||
fun generate_cast(node: *ast_node): int {
|
||||
return generate(node->cast.value)
|
||||
}
|
||||
fun generate_value(node: *ast_node) {
|
||||
bytecode.add(byte_inst::nop())
|
||||
fun generate_value(node: *ast_node): int {
|
||||
return emit_imm(string_to_num<int>(node->value.string_value))
|
||||
}
|
||||
fun generate_code_block(node: *ast_node) {
|
||||
bytecode.add(byte_inst::nop())
|
||||
fun generate_code_block(node: *ast_node): int {
|
||||
node->code_block.children.for_each(fun(child: *ast_node) generate(child);)
|
||||
return -1
|
||||
}
|
||||
// this generates the function as a value, not the actual function
|
||||
fun generate_function(node: *ast_node) {
|
||||
bytecode.add(byte_inst::nop())
|
||||
fun generate_function(node: *ast_node): int {
|
||||
return emit_imm(-2)
|
||||
}
|
||||
fun generate_function_call(node: *ast_node) {
|
||||
bytecode.add(byte_inst::nop())
|
||||
fun generate_function_call(node: *ast_node): int {
|
||||
node->function_call.parameters.for_each(fun(child: *ast_node) generate(child);)
|
||||
return emit_call()
|
||||
}
|
||||
|
||||
fun generate_compiler_intrinsic(node: *ast_node) {
|
||||
bytecode.add(byte_inst::nop())
|
||||
fun generate_compiler_intrinsic(node: *ast_node): int {
|
||||
blocks.last().instructions.add(byte_inst::nop())
|
||||
return -1
|
||||
}
|
||||
|
||||
fun generate(node: *ast_node) {
|
||||
fun generate(node: *ast_node): int {
|
||||
match (*node) {
|
||||
ast_node::declaration_statement(backing) generate_declaration_statement(node)
|
||||
ast_node::assignment_statement(backing) generate_assignment_statement(node)
|
||||
ast_node::if_statement(backing) generate_if_statement(node)
|
||||
ast_node::while_loop(backing) generate_while_loop(node)
|
||||
ast_node::for_loop(backing) generate_for_loop(node)
|
||||
ast_node::function(backing) generate_function(node)
|
||||
ast_node::function_call(backing) generate_function_call(node)
|
||||
ast_node::compiler_intrinsic(backing) generate_compiler_intrinsic(node)
|
||||
ast_node::code_block(backing) generate_code_block(node)
|
||||
ast_node::return_statement(backing) generate_return_statement(node)
|
||||
ast_node::branching_statement(backing) generate_branching_statement(node)
|
||||
ast_node::cast(backing) generate_cast(node)
|
||||
ast_node::value(backing) generate_value(node)
|
||||
ast_node::identifier(backing) generate_identifier(node)
|
||||
ast_node::declaration_statement(backing) return generate_declaration_statement(node)
|
||||
ast_node::assignment_statement(backing) return generate_assignment_statement(node)
|
||||
ast_node::if_statement(backing) return generate_if_statement(node)
|
||||
ast_node::while_loop(backing) return generate_while_loop(node)
|
||||
ast_node::for_loop(backing) return generate_for_loop(node)
|
||||
ast_node::function(backing) return generate_function(node)
|
||||
ast_node::function_call(backing) return generate_function_call(node)
|
||||
ast_node::compiler_intrinsic(backing) return generate_compiler_intrinsic(node)
|
||||
ast_node::code_block(backing) return generate_code_block(node)
|
||||
ast_node::return_statement(backing) return generate_return_statement(node)
|
||||
ast_node::branching_statement(backing) return generate_branching_statement(node)
|
||||
ast_node::cast(backing) return generate_cast(node)
|
||||
ast_node::value(backing) return generate_value(node)
|
||||
ast_node::identifier(backing) return generate_identifier(node)
|
||||
}
|
||||
error("Bad node")
|
||||
}
|
||||
fun get_name(node: *ast_node): string {
|
||||
var maybe_it = ast_name_map.get_ptr_or_null(node);
|
||||
if (maybe_it)
|
||||
return *maybe_it
|
||||
var result = "name" + get_id()
|
||||
var result = get_ast_name(node) + get_id()
|
||||
if (is_function(node) && node->function.name == "main")
|
||||
result = "main"
|
||||
ast_name_map.set(node, result)
|
||||
return result
|
||||
}
|
||||
fun emit_imm(value: int): int {
|
||||
var i: imm
|
||||
i.reg = get_reg()
|
||||
i.val = value
|
||||
blocks.last().instructions.add(byte_inst::imm(i))
|
||||
return i.reg
|
||||
}
|
||||
fun emit_ld(): int {
|
||||
var l: ld
|
||||
l.reg = get_reg()
|
||||
blocks.last().instructions.add(byte_inst::ld(l))
|
||||
return l.reg
|
||||
}
|
||||
fun emit_st(to_reg: int, from_reg: int): int {
|
||||
var s: st
|
||||
s.to_reg = to_reg
|
||||
s.from_reg = from_reg
|
||||
blocks.last().instructions.add(byte_inst::st(s))
|
||||
return -1
|
||||
}
|
||||
fun emit_ret(reg: int): int {
|
||||
var r: ret
|
||||
r.reg = reg
|
||||
blocks.last().instructions.add(byte_inst::ret(r))
|
||||
return -1
|
||||
}
|
||||
fun emit_call(): int {
|
||||
blocks.last().instructions.add(byte_inst::call())
|
||||
return -1
|
||||
}
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user