progress: new assembler works, instruction set fully implemented with call/ret and push/pop, bf.dsa works which means the dsa assembly language works reliably. still a few bugs to fix. might be able to squeeze out a tiny bit more performance

This commit is contained in:
2026-03-09 03:24:20 +00:00
parent fc972b9b7b
commit e01a9f2808
63 changed files with 4975 additions and 1579 deletions
+399
View File
@@ -0,0 +1,399 @@
use common::prelude::*;
use crate::assembler::Token;
use crate::assembler::model::{Node, Opcode};
use crate::{assembler::AssembleError, expect_token};
fn log(message: &str) {
println!("\x1b[32mINFO:\x1b[0m {message}");
}
pub fn codegen(nodes: Vec<Node>) -> Result<Vec<Instruction>, AssembleError> {
let mut instructions = vec![];
for node in nodes {
println!("{:?}", node);
instructions.push(build_instruction(&node)?);
}
log("Assembly Successful ✅");
Ok(instructions)
}
fn build_instruction(node: &Node) -> Result<Instruction, AssembleError> {
let opcode = node.opcode();
let args = node.args();
match opcode {
Opcode::Nop => Ok(Instruction::nop()),
Opcode::Mov => ins_mov(opcode, &args),
Opcode::CMov => ins_cmov(opcode, &args),
Opcode::Ldb
| Opcode::Ldw
| Opcode::Ldh
| Opcode::Ldbs
| Opcode::Ldhs
| Opcode::Stb
| Opcode::Stw
| Opcode::Sth => ins_ldx_stx(opcode, &args),
Opcode::Lli | Opcode::Lui => ins_load_imm(opcode, &args),
Opcode::Push | Opcode::Pop => ins_stack(opcode, &args),
Opcode::Ieq | Opcode::Ine | Opcode::Igt | Opcode::Ige | Opcode::Ile | Opcode::Ilt => {
ins_comparison(opcode, &args)
}
Opcode::Jmp | Opcode::Call | Opcode::Jic | Opcode::Jnc => {
ins_jump_unconditional(opcode, &args)
}
Opcode::Jez | Opcode::Jnz => ins_jump_conditional(opcode, &args),
Opcode::Shl | Opcode::Shr => ins_bitshift(opcode, &args),
Opcode::Add
| Opcode::Sub
| Opcode::And
| Opcode::Or
| Opcode::Xor
| Opcode::Nand
| Opcode::Nor
| Opcode::Xnor => ins_arithmetic(opcode, &args),
Opcode::AddI | Opcode::SubI => ins_imm_arithmetic(opcode, &args),
Opcode::Not => ins_not(&args),
Opcode::Int => ins_interrupt(&args),
Opcode::Ret => Ok(Instruction::ret()),
Opcode::IRet => Ok(Instruction::irt()),
Opcode::Hlt => Ok(Instruction::hlt()),
Opcode::Data => build_data_instruction(&args),
Opcode::Segment => build_segment_instruction(&args),
Opcode::Db
| Opcode::Dh
| Opcode::Dw
| Opcode::Resb
| Opcode::Resh
| Opcode::Resw
| Opcode::Lwi
| Opcode::Include
| Opcode::Func
| Opcode::Return => Err(AssembleError::InvalidArg),
}
}
fn ins_mov(
opcode: Opcode,
args: &[crate::assembler::model::Token],
) -> Result<Instruction, AssembleError> {
let Some(src_token) = args.first() else {
return Err(AssembleError::MissingArgument(0));
};
let Some(dest_token) = args.get(1) else {
return Err(AssembleError::MissingArgument(1));
};
let src = expect_token!(src_token, Register)?;
let dest = expect_token!(dest_token, Register)?;
match opcode {
Opcode::Mov => Ok(Instruction::mov(src, dest)),
_ => unreachable!(),
}
}
fn ins_cmov(
opcode: Opcode,
args: &[crate::assembler::model::Token],
) -> Result<Instruction, AssembleError> {
let Some(src_token) = args.first() else {
return Err(AssembleError::MissingArgument(0));
};
let Some(dest_token) = args.get(1) else {
return Err(AssembleError::MissingArgument(1));
};
let Some(cmp_token) = args.get(2) else {
return Err(AssembleError::MissingArgument(1));
};
let src = expect_token!(src_token, Register)?;
let dest = expect_token!(dest_token, Register)?;
let cmp = expect_token!(cmp_token, Register)?;
match opcode {
Opcode::CMov => Ok(Instruction::cmov(src, dest, cmp)),
_ => unreachable!(),
}
}
fn ins_ldx_stx(
opcode: Opcode,
args: &[crate::assembler::model::Token],
) -> Result<Instruction, AssembleError> {
let Some(src_token) = args.first() else {
return Err(AssembleError::MissingArgument(0));
};
let Some(dest_token) = args.get(1) else {
return Err(AssembleError::MissingArgument(1));
};
let Some(offset_token) = args.get(2) else {
return Err(AssembleError::MissingArgument(2));
};
let src = expect_token!(src_token, Register)?;
let dest = expect_token!(dest_token, Register)?;
let offset = expect_token!(offset_token, Immediate)?;
match opcode {
Opcode::Ldb => Ok(Instruction::ldb(src, dest, offset as u16)),
Opcode::Ldbs => Ok(Instruction::ldbs(src, dest, offset as u16)),
Opcode::Ldh => Ok(Instruction::ldh(src, dest, offset as u16)),
Opcode::Ldhs => Ok(Instruction::ldhs(src, dest, offset as u16)),
Opcode::Ldw => Ok(Instruction::ldw(src, dest, offset as u16)),
Opcode::Stb => Ok(Instruction::stb(src, dest, offset as u16)),
Opcode::Sth => Ok(Instruction::sth(src, dest, offset as u16)),
Opcode::Stw => Ok(Instruction::stw(src, dest, offset as u16)),
_ => unreachable!(),
}
}
fn ins_stack(
opcode: Opcode,
args: &[crate::assembler::model::Token],
) -> Result<Instruction, AssembleError> {
let Some(reg_token) = args.first() else {
return Err(AssembleError::MissingArgument(0));
};
let reg = expect_token!(reg_token, Register)?;
match opcode {
Opcode::Push => Ok(Instruction::push(reg)),
Opcode::Pop => Ok(Instruction::pop(reg)),
_ => unreachable!(),
}
}
fn ins_load_imm(
opcode: Opcode,
args: &[crate::assembler::model::Token],
) -> Result<Instruction, AssembleError> {
let Some(value_token) = args.first() else {
return Err(AssembleError::MissingArgument(0));
};
let Some(dest_token) = args.get(1) else {
return Err(AssembleError::MissingArgument(1));
};
let value = expect_token!(value_token, Immediate)?;
let dest = expect_token!(dest_token, Register)?;
match opcode {
Opcode::Lli => Ok(Instruction::lli(dest, value as u16)),
Opcode::Lui => Ok(Instruction::lui(dest, (value >> 16) as u16)),
_ => unreachable!(),
}
}
fn ins_jump_unconditional(
opcode: Opcode,
args: &[crate::assembler::model::Token],
) -> Result<Instruction, AssembleError> {
let Some(addr_imm_token) = args.first() else {
return Err(AssembleError::MissingArgument(0));
};
// addr_reg is optional; default to Register::Zero if missing
let addr_reg_token = args.get(1).unwrap_or(&Token::Register(Register::Zero));
let addr_imm = expect_token!(addr_imm_token, Immediate)?;
let addr_reg = expect_token!(addr_reg_token, Register)?;
match opcode {
Opcode::Jmp => Ok(Instruction::jmp(addr_reg, addr_imm as u16)),
Opcode::Jic => Ok(Instruction::jic(addr_reg, addr_imm as u16)),
Opcode::Jnc => Ok(Instruction::jnc(addr_reg, addr_imm as u16)),
Opcode::Call => Ok(Instruction::call(addr_reg, addr_imm as u16)),
_ => unreachable!(),
}
}
fn ins_jump_conditional(
opcode: Opcode,
args: &[crate::assembler::model::Token],
) -> Result<Instruction, AssembleError> {
let Some(condition_token) = args.first() else {
return Err(AssembleError::MissingArgument(0));
};
let Some(addr_imm_token) = args.get(1) else {
return Err(AssembleError::MissingArgument(1));
};
let addr_reg_token = if let Some(token) = args.get(2) {
token
} else {
&Token::Register(Register::Zero)
};
let condition = expect_token!(condition_token, Register)?;
let addr_imm = expect_token!(addr_imm_token, Immediate)?;
let addr_reg = expect_token!(addr_reg_token, Register)?;
Ok(match opcode {
Opcode::Jez => Instruction::jez(condition, addr_reg, addr_imm as u16),
Opcode::Jnz => Instruction::jnz(condition, addr_reg, addr_imm as u16),
_ => unreachable!(),
})
}
fn ins_comparison(
opcode: Opcode,
args: &[crate::assembler::model::Token],
) -> Result<Instruction, AssembleError> {
let Some(left_token) = args.first() else {
return Err(AssembleError::MissingArgument(0));
};
let Some(right_token) = args.get(1) else {
return Err(AssembleError::MissingArgument(1));
};
let Some(dest_token) = args.get(2) else {
return Err(AssembleError::MissingArgument(2));
};
let left = expect_token!(left_token, Register)?;
let right = expect_token!(right_token, Register)?;
let dest = expect_token!(dest_token, Register)?;
Ok(match opcode {
Opcode::Ieq => Instruction::ieq(left, right, dest),
Opcode::Ine => Instruction::ine(left, right, dest),
Opcode::Igt => Instruction::igt(left, right, dest),
Opcode::Ige => Instruction::ige(left, right, dest),
Opcode::Ile => Instruction::ile(left, right, dest),
Opcode::Ilt => Instruction::ilt(left, right, dest),
_ => unreachable!(),
})
}
fn ins_bitshift(
opcode: Opcode,
args: &[crate::assembler::model::Token],
) -> Result<Instruction, AssembleError> {
let Some(src_reg) = args.first() else {
return Err(AssembleError::MissingArgument(0));
};
let Some(r_shamt) = args.get(1) else {
return Err(AssembleError::MissingArgument(0));
};
let Some(i_shamt) = args.get(2) else {
return Err(AssembleError::MissingArgument(1));
};
let Some(dest_reg) = args.get(3) else {
return Err(AssembleError::MissingArgument(1));
};
let src = expect_token!(src_reg, Register)?;
let r_shamt = expect_token!(r_shamt, Register)?;
let i_shamt = expect_token!(i_shamt, Immediate)? as u8;
let dest = expect_token!(dest_reg, Register)?;
Ok(match opcode {
Opcode::Shl => Instruction::shl(src, r_shamt, dest, i_shamt),
Opcode::Shr => Instruction::shr(src, r_shamt, dest, i_shamt),
_ => unreachable!(),
})
}
fn ins_arithmetic(
opcode: Opcode,
args: &[crate::assembler::model::Token],
) -> Result<Instruction, AssembleError> {
let Some(left_token) = args.first() else {
return Err(AssembleError::MissingArgument(0));
};
let Some(right_token) = args.get(1) else {
return Err(AssembleError::MissingArgument(1));
};
let Some(dest_token) = args.get(2) else {
return Err(AssembleError::MissingArgument(2));
};
let left = expect_token!(left_token, Register)?;
let right = expect_token!(right_token, Register)?;
let dest = expect_token!(dest_token, Register)?;
Ok(match opcode {
Opcode::Add => Instruction::add(left, right, dest),
Opcode::Sub => Instruction::sub(left, right, dest),
Opcode::And => Instruction::and(left, right, dest),
Opcode::Or => Instruction::or(left, right, dest),
Opcode::Xor => Instruction::xor(left, right, dest),
Opcode::Nand => Instruction::nand(left, right, dest),
Opcode::Nor => Instruction::nor(left, right, dest),
Opcode::Xnor => Instruction::xnor(left, right, dest),
_ => unreachable!(),
})
}
fn ins_imm_arithmetic(
opcode: Opcode,
args: &[crate::assembler::model::Token],
) -> Result<Instruction, AssembleError> {
let Some(reg_token) = args.first() else {
return Err(AssembleError::MissingArgument(0));
};
let Some(immediate_token) = args.get(1) else {
return Err(AssembleError::MissingArgument(1));
};
let Some(dest_token) = args.get(2) else {
return Err(AssembleError::MissingArgument(2));
};
let reg = expect_token!(reg_token, Register)?;
let immediate = expect_token!(immediate_token, Immediate)? as u16;
let dest = expect_token!(dest_token, Register)?;
Ok(match opcode {
Opcode::AddI => Instruction::addi(reg, dest, immediate),
Opcode::SubI => Instruction::subi(reg, dest, immediate),
_ => unreachable!(),
})
}
fn ins_not(args: &[crate::assembler::model::Token]) -> Result<Instruction, AssembleError> {
let Some(reg_token) = args.first() else {
return Err(AssembleError::MissingArgument(0));
};
let Some(dest_token) = args.get(1) else {
return Err(AssembleError::MissingArgument(1));
};
let src = expect_token!(reg_token, Register)?;
let dest = expect_token!(dest_token, Register)?;
Ok(Instruction::not(src, dest))
}
fn ins_interrupt(args: &[crate::assembler::model::Token]) -> Result<Instruction, AssembleError> {
let Some(code_token) = args.first() else {
return Err(AssembleError::MissingArgument(0));
};
let code = expect_token!(code_token, Immediate)? as u8;
Ok(Instruction::int(code))
}
fn build_data_instruction(
args: &[crate::assembler::model::Token],
) -> Result<Instruction, AssembleError> {
let Some(immediate_token) = args.first() else {
return Err(AssembleError::MissingArgument(0));
};
let immediate = expect_token!(immediate_token, Immediate)?;
Ok(Instruction::data(immediate))
}
fn build_segment_instruction(
args: &[crate::assembler::model::Token],
) -> Result<Instruction, AssembleError> {
let Some(immediate_token) = args.first() else {
return Err(AssembleError::MissingArgument(0));
};
let immediate = expect_token!(immediate_token, Immediate)?;
Ok(Instruction::data(
// mask the first 6 bits for the opcode
// this is really deprecated tbh. may remove in future.
Opcode::Segment as u32 | (immediate & 0x05FFFFFF),
))
}