Add binary number parsing

Signed-off-by: Alek Ratzloff <alekratz@gmail.com>

src/main.rs

@@ -1,7 +1,6 @@
 #![allow(dead_code)]
 
 mod common;
-//mod syn;
 mod vm;
 
 use std::{convert::TryFrom, env, fs, io, process};
@@ -32,10 +31,30 @@ fn main() -> Result<()> {
         }
     };
     let obj = vm::obj::obj::Object::try_from(&ast)?;
-    //println!("{:#?}", obj);
+    //dump(&obj)?;
     let mut vm = vm::vm::Vm::new();
     vm.load_object(obj, 1024 * 1024 * 64)?; // 64mb
     let status = vm.run()?;
-    println!("status: {}", status);
+    println!("{}", status);
+    Ok(())
+}
+
+
+fn dump(obj: &vm::obj::obj::Object) -> Result<()> {
+    use vm::obj::obj::Section;
+    use vm::visit::VisitInst;
+    let mut stdout = io::stdout();
+    for section in &obj.sections {
+        let mut disasm = match section {
+            Section::Code { start, contents, .. } => {
+                vm::disassemble::Disassemble::new(&mut stdout, contents, *start)
+            }
+            Section::Meta { .. } | Section::Data { .. } => continue,
+        };
+        while !disasm.is_done() {
+            disasm.visit_inst()?;
+        }
+        println!();
+    }
     Ok(())
 }

src/vm/disassemble.rs

@@ -0,0 +1,294 @@
+use crate::vm::{
+    error::*,
+    inst::*,
+    mem::MemCursor,
+    reg::*,
+    visit::*,
+    vm::{Addr, HalfWord, Word},
+};
+use std::io::Write;
+
+const WIDTH: usize = 60;
+
+pub struct Disassemble<'w, 'o> {
+    writer: &'w mut dyn Write,
+    cursor: MemCursor<'o>,
+    addr_offset: Addr,
+}
+
+impl<'w, 'o> Disassemble<'w, 'o> {
+    pub fn new(writer: &'w mut dyn Write, content: &'o [u8], addr_offset: Addr) -> Self {
+        Disassemble {
+            writer,
+            cursor: MemCursor::new(content),
+            addr_offset,
+        }
+    }
+
+    pub fn is_done(&self) -> bool {
+        self.cursor.position() >= (self.cursor.get_ref().len() as u64)
+    }
+
+    fn adv(&mut self) -> Result<()> {
+        // note the () - this explicitly clones the cursor
+        let op = self.cursor().next_u16()?;
+        let next = self.cursor.position() + (inst_len(op) as u64);
+        self.cursor.set_position(next);
+        Ok(())
+    }
+
+    fn write_addr(&mut self, addr: Addr) {
+        write!(self.writer, "{:06x} | ", self.addr_offset + addr).unwrap();
+    }
+
+    fn write_bytes(&mut self, bytes: &[u8]) {
+        for b in bytes {
+            write!(self.writer, "{:02x} ", b).unwrap();
+        }
+    }
+
+    fn write_inst_bytes(&mut self, op: InstOp) {
+        let len = inst_len(op);
+        let start = self.cursor.position() as usize;
+        let end = start + len;
+        let bytes = &self.cursor.get_ref()[start..end];
+        self.write_bytes(bytes);
+    }
+
+    fn write_r1_r2_inst(&mut self, addr: Addr, op: InstOp, r1: Reg, r2: Reg) {
+        let len = inst_len(op);
+        let line_width = 6 + 3 + (3 * len);
+        let line_offset = WIDTH - line_width;
+        self.write_addr(addr);
+        self.write_inst_bytes(op);
+
+        let iname = inst_name(op).unwrap();
+        let r1name = reg_name(r1).unwrap().to_lowercase();
+        let r2name = reg_name(r2).unwrap().to_lowercase();
+
+        writeln!(
+            self.writer,
+            "{}| {:>10} %{} %{}",
+            " ".repeat(line_offset),
+            iname,
+            r1name,
+            r2name
+        )
+        .unwrap();
+    }
+
+    fn write_r1_inst(&mut self, addr: Addr, op: InstOp, r1: Reg) {
+        let line_width = 6 + 3 + (3 * inst_len(op));
+        let line_offset = WIDTH - line_width;
+        self.write_addr(addr);
+        self.write_inst_bytes(op);
+
+        let iname = inst_name(op).unwrap();
+        let r1name = reg_name(r1).unwrap().to_lowercase();
+
+        writeln!(
+            self.writer,
+            "{}| {:>10} %{}",
+            " ".repeat(line_offset),
+            iname,
+            r1name,
+        )
+        .unwrap();
+    }
+
+    fn write_r1_imm_inst(&mut self, addr: Addr, op: InstOp, r1: Reg, imm: Word) {
+        let line_width = 6 + 3 + (3 * inst_len(op));
+        let line_offset = WIDTH - line_width;
+        self.write_addr(addr);
+        self.write_inst_bytes(op);
+
+        let iname = inst_name(op).unwrap();
+        let r1name = reg_name(r1).unwrap().to_lowercase();
+
+        writeln!(
+            self.writer,
+            "{}| {:>10} %{} {:#X}",
+            " ".repeat(line_offset),
+            iname,
+            r1name,
+            imm,
+        )
+        .unwrap();
+    }
+}
+
+impl VisitInst for Disassemble<'_, '_> {
+    type Out = ();
+
+    fn cursor(&self) -> MemCursor {
+        self.cursor.clone()
+    }
+
+    fn add(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        self.write_r1_r2_inst(self.cursor.position(), ADD, r1, r2);
+        self.adv()?;
+        Ok(())
+    }
+
+    fn mul(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        self.write_r1_r2_inst(self.cursor.position(), MUL, r1, r2);
+        self.adv()?;
+        Ok(())
+    }
+
+    fn div(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        self.write_r1_r2_inst(self.cursor.position(), DIV, r1, r2);
+        self.adv()?;
+        Ok(())
+    }
+
+    fn mod_(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        self.write_r1_r2_inst(self.cursor.position(), MOD, r1, r2);
+        self.adv()?;
+        Ok(())
+    }
+
+    fn ineg(&mut self, r1: Reg) -> Result<Self::Out> {
+        self.write_r1_inst(self.cursor.position(), INEG, r1);
+        self.adv()?;
+        Ok(())
+    }
+
+    fn and(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        self.write_r1_r2_inst(self.cursor.position(), AND, r1, r2);
+        self.adv()?;
+        Ok(())
+    }
+
+    fn or(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        self.write_r1_r2_inst(self.cursor.position(), OR, r1, r2);
+        self.adv()?;
+        Ok(())
+    }
+
+    fn inv(&mut self, r1: Reg) -> Result<Self::Out> {
+        self.write_r1_inst(self.cursor.position(), INV, r1);
+        self.adv()?;
+        Ok(())
+    }
+
+    fn not(&mut self, r1: Reg) -> Result<Self::Out> {
+        self.write_r1_inst(self.cursor.position(), NOT, r1);
+        self.adv()?;
+        Ok(())
+    }
+
+    fn xor(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        self.write_r1_r2_inst(self.cursor.position(), XOR, r1, r2);
+        self.adv()?;
+        Ok(())
+    }
+
+    fn shl(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        self.write_r1_r2_inst(self.cursor.position(), SHL, r1, r2);
+        self.adv()?;
+        Ok(())
+    }
+
+    fn shr(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        self.write_r1_r2_inst(self.cursor.position(), SHR, r1, r2);
+        self.adv()?;
+        Ok(())
+    }
+
+    fn cmpeq(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        self.write_r1_r2_inst(self.cursor.position(), CMPEQ, r1, r2);
+        self.adv()?;
+        Ok(())
+    }
+
+    fn cmplt(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        self.write_r1_r2_inst(self.cursor.position(), CMPLT, r1, r2);
+        self.adv()?;
+        Ok(())
+    }
+
+    fn jmp(&mut self, r1: Reg) -> Result<Self::Out> {
+        self.write_r1_inst(self.cursor.position(), JMP, r1);
+        self.adv()?;
+        Ok(())
+    }
+
+    fn jz(&mut self, r1: Reg) -> Result<Self::Out> {
+        self.write_r1_inst(self.cursor.position(), JZ, r1);
+        self.adv()?;
+        Ok(())
+    }
+
+    fn jnz(&mut self, r1: Reg) -> Result<Self::Out> {
+        self.write_r1_inst(self.cursor.position(), JNZ, r1);
+        self.adv()?;
+        Ok(())
+    }
+
+    fn load(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        self.write_r1_r2_inst(self.cursor.position(), LOAD, r1, r2);
+        self.adv()?;
+        Ok(())
+    }
+
+    fn regcopy(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        self.write_r1_r2_inst(self.cursor.position(), REGCOPY, r1, r2);
+        self.adv()?;
+        Ok(())
+    }
+
+    fn storeimm64(&mut self, r1: Reg, w1: Word) -> Result<Self::Out> {
+        self.write_r1_imm_inst(self.cursor.position(), STOREIMM64, r1, w1);
+        self.adv()?;
+        Ok(())
+    }
+
+    fn storeimm32(&mut self, r1: Reg, w1: HalfWord) -> Result<Self::Out> {
+        self.write_r1_imm_inst(self.cursor.position(), STOREIMM32, r1, w1 as u64);
+        self.adv()?;
+        Ok(())
+    }
+
+    fn memcopy(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        self.write_r1_r2_inst(self.cursor.position(), MEMCOPY, r1, r2);
+        self.adv()?;
+        Ok(())
+    }
+
+    fn store(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        self.write_r1_r2_inst(self.cursor.position(), STORE, r1, r2);
+        self.adv()?;
+        Ok(())
+    }
+
+    fn halt(&mut self) -> Result<Self::Out> {
+        let line_width = 6 + 3 + 3 + 3;
+        let line_offset = WIDTH - line_width;
+        self.write_addr(self.cursor.position());
+        self.write_inst_bytes(HALT);
+        writeln!(
+            self.writer,
+            "{}| {:>10}",
+            " ".repeat(line_offset),
+            "HALT",
+        ).unwrap();
+        self.adv()?;
+        Ok(())
+    }
+
+    fn nop(&mut self) -> Result<Self::Out> {
+        let line_width = 6 + 3 + 3 + 3;
+        let line_offset = WIDTH - line_width;
+        self.write_addr(self.cursor.position());
+        self.write_inst_bytes(NOP);
+        writeln!(
+            self.writer,
+            "{}| {:>10}",
+            " ".repeat(line_offset),
+            "NOP",
+        ).unwrap();
+        self.adv()?;
+        Ok(())
+    }
+}

src/vm/error.rs

@@ -1,4 +1,4 @@
-use crate::vm::{vm::*, reg::Reg};
+use crate::vm::{inst::InstOp, reg::Reg, vm::*};
 use snafu::Snafu;
 
 #[derive(Snafu, Debug, Clone)]
@@ -7,6 +7,8 @@ pub enum VmError {
     IllegalReg { reg: Reg },
     #[snafu(display("memory address out of bounds: 0x{:016x}", addr))]
     MemOutOfBounds { addr: Addr },
+    #[snafu(display("illegal instruction opcode: 0x{:04x}", op))]
+    IllegalOp { op: InstOp },
 }
 
 pub type Result<T, E = VmError> = std::result::Result<T, E>;

src/vm/mem.rs

@@ -8,6 +8,7 @@ use std::{
 const R1_MASK: u16 = 0b1111_1100_0000_0000;
 const R2_MASK: u16 = 0b0000_0011_1111_0000;
 
+#[derive(Debug, Clone)]
 pub struct MemCursor<'mem> {
     cursor: Cursor<&'mem [u8]>,
 }

src/vm/mod.rs

@@ -1,3 +1,4 @@
+pub mod disassemble;
 pub mod error;
 pub mod flags;
 pub mod inst;
@@ -5,4 +6,5 @@ pub mod mem;
 pub mod obj;
 pub mod reg;
 mod tick;
+pub mod visit;
 pub mod vm;

src/vm/obj/assemble/mod.rs

@@ -38,7 +38,6 @@ impl<'a> Assemble<'a> {
     }
 
     pub fn assemble(&mut self) -> Result<Object> {
-        let mut pos = 0;
         let mut sections = Vec::new();
 
         // gather global symbols
@@ -60,7 +59,6 @@ impl<'a> Assemble<'a> {
         for block in self.ast.iter() {
             let locals = Self::gather_symbols(block, false)?;
             self.symbols.replace_locals(locals);
-
             match block {
                 SectionBlock::Data { org, body } | SectionBlock::Code { org, body } => {
                     let mut bytes = Vec::new();
@@ -79,11 +77,9 @@ impl<'a> Assemble<'a> {
                         }
                     }
                     let (start, end) = match org {
-                        Some(SectionOrg::Start(start)) => (*start, start + bytes.len() as u64),
+                        SectionOrg::Start(start) => (*start, start + bytes.len() as u64),
-                        Some(SectionOrg::Range(start, end)) => (*start, *end),
+                        SectionOrg::Range(start, end) => (*start, *end),
-                        None => (pos, pos + bytes.len() as u64),
                     };
-                    pos = end;
 
                     let section = match block {
                         SectionBlock::Data { .. } => Section::Data {
@@ -122,10 +118,12 @@ impl<'a> Assemble<'a> {
 
     fn gather_symbols(block: &SectionBlock, export: bool) -> Result<HashMap<String, u64>> {
         match block {
-            SectionBlock::Data { body, .. } | SectionBlock::Code { body, .. } => {
+            SectionBlock::Data { org, body, .. } | SectionBlock::Code { org, body, .. } => {
                 let mut exports = HashSet::new();
                 let mut labels = HashMap::new();
-                let mut pos = 0;
+                let mut pos = match org {
+                    SectionOrg::Start(start) | SectionOrg::Range(start, _) => (*start) as usize,
+                };
                 for line in body.iter() {
                     match line {
                         Line::Inst(inst) => {
@@ -191,18 +189,35 @@ impl<'a> Assemble<'a> {
             Inst::Jnz(r1) => builder.op(JNZ).r1(*r1),
             Inst::Load(r1, r2) => builder.op(LOAD).r1(*r1).r2(*r2),
             Inst::Store(r1, r2) => builder.op(STORE).r1(*r1).r2(*r2),
-            Inst::StoreImm(r1, imm) => {
+            Inst::StoreImm(r1, imm) => match imm {
-                let imm = match imm {
+                ImmValue::Number(num) => {
-                    ImmValue::Number(num) => *num,
+                    if *num > (u32::max_value() as u64) {
-                    ImmValue::Label(name) => {
+                        builder.op(STOREIMM64).imm64(*num)
-                        self.symbols.get(name).expect("TODO: value label not found")
+                    } else {
+                        builder.op(STOREIMM32).imm32(*num as u32)
                     }
-                };
-                if imm <= (u32::max_value() as u64) {
-                    builder.op(STOREIMM32).r1(*r1).imm32(imm as u32)
-                } else {
-                    builder.op(STOREIMM64).r1(*r1).imm64(imm)
                 }
+                ImmValue::Label(name) => {
+                    let imm = self.symbols.get(name).expect("TODO: value label not found");
+                    builder.op(STOREIMM64).imm64(imm)
+                }
+            }
+            .r1(*r1),
+            Inst::StoreImm32(r1, imm) => {
+                let imm = self.get_value(imm).expect("TODO : value label not found");
+                let imm_truncated = (imm & 0xFFFFFFFF) as u32;
+                // TODO compile-time warnings
+                if imm != (imm_truncated as u64) {
+                    eprintln!(
+                        "WARNING: immediate 32 bit value being truncated ({:#x} to {:#x})",
+                        imm, imm_truncated
+                    )
+                }
+                builder.op(STOREIMM32).imm32(imm_truncated).r1(*r1)
+            }
+            Inst::StoreImm64(r1, imm) => {
+                let imm = self.get_value(imm).expect("TODO : value label not found");
+                builder.op(STOREIMM64).imm64(imm).r1(*r1)
             }
             Inst::MemCopy(r1, r2) => builder.op(MEMCOPY).r1(*r1).r2(*r2),
             Inst::RegCopy(r1, r2) => builder.op(REGCOPY).r1(*r1).r2(*r2),

src/vm/obj/syn/ast.rs

@@ -3,11 +3,11 @@ use crate::vm::{inst::*, reg::Reg};
 #[derive(Debug, Clone, PartialEq, Eq)]
 pub enum SectionBlock {
     Data {
-        org: Option<SectionOrg>,
+        org: SectionOrg,
         body: Vec<Line>,
     },
     Code {
-        org: Option<SectionOrg>,
+        org: SectionOrg,
         body: Vec<Line>,
     },
     Meta {
@@ -59,6 +59,8 @@ pub enum Inst {
     Load(Reg, Reg),
     Store(Reg, Reg),
     StoreImm(Reg, ImmValue),
+    StoreImm32(Reg, ImmValue),
+    StoreImm64(Reg, ImmValue),
     MemCopy(Reg, Reg),
     RegCopy(Reg, Reg),
 
@@ -101,6 +103,8 @@ impl Inst {
                     STOREIMM64
                 }
             }
+            Inst::StoreImm32(_, _) => STOREIMM32,
+            Inst::StoreImm64(_, _) => STOREIMM64,
             Inst::MemCopy(_, _) => MEMCOPY,
             Inst::RegCopy(_, _) => REGCOPY,
 

src/vm/obj/syn/parser.lalrpop

@@ -23,7 +23,8 @@ Label: String = {
 
 Number: u64 = {
     <s:r"\$[0-9]+"> => u64::from_str(&s[1..]).unwrap(),
-    <s:r"\$0x[0-9a-fA-F]+"> => u64::from_str_radix(&s[3..], 16).unwrap()
+    <s:r"\$0x[0-9a-fA-F]+"> => u64::from_str_radix(&s[3..], 16).unwrap(),
+    <s:r"\$0b[01]+"> => u64::from_str_radix(&s[3..], 2).unwrap(),
 }
 
 Reg: Reg = {
@@ -59,6 +60,8 @@ Inst: Inst = {
     "load" <d:Reg> "," <s:Reg> => Inst::Load(d, s),
     "store" <d:Reg> "," <s:Reg> => Inst::Store(d, s),
     "storeimm" <d:Reg> "," <s:ImmValue> => Inst::StoreImm(d, s),
+    "storeimm32" <d:Reg> "," <s:ImmValue> => Inst::StoreImm32(d, s),
+    "storeimm64" <d:Reg> "," <s:ImmValue> => Inst::StoreImm64(d, s),
     "memcopy" <d:Reg> "," <s:Reg> => Inst::MemCopy(d, s),
     "regcopy" <d:Reg> "," <s:Reg> => Inst::RegCopy(d, s),
     "nop" => Inst::Nop,
@@ -82,11 +85,11 @@ SectionOrg: SectionOrg = {
 }
 
 Section: SectionBlock = {
-    "data" <org:SectionOrg?> "{" <body:Line*> "}" => {
+    "data" <org:SectionOrg> "{" <body:Line*> "}" => {
         SectionBlock::Data { org, body }
     },
 
-    "code" <org:SectionOrg?> "{" <body:Line*> "}" => {
+    "code" <org:SectionOrg> "{" <body:Line*> "}" => {
         SectionBlock::Code { org, body }
     },
 

src/vm/reg.rs

@@ -7,6 +7,15 @@ macro_rules! registers {
         $(
             pub const $variant: Reg = $value;
         )*
+
+        pub fn reg_name(reg: Reg) -> Option<&'static str> {
+            match reg {
+                $(
+                    $value => Some(stringify!($variant)),
+                )*
+                _ => None,
+            }
+        }
     };
 }
 
@@ -91,5 +100,6 @@ registers! {
     R47 = 61,
     R48 = 62,
     R49 = 63,
-    LAST_REG = R49,
 }
+
+pub const LAST_REG: Reg = R49;

src/vm/tick.rs

@@ -1,134 +1,189 @@
-use crate::vm::{error::*, flags::Flags, inst::*, reg::*, vm::*};
+use crate::vm::{error::*, flags::Flags, inst::*, reg::*, vm::*, visit::*, mem::MemCursor};
-use std::io::stdin;
 
 impl Vm {
     pub fn tick(&mut self) -> Result<()> {
-        let mut cursor = self.mem_cursor(self.ip() as usize);
+        let next_ip = self.visit_inst()?;
-        let op = cursor.next_u16()?;
-        let next_ip;
-
-        macro_rules! math_inst {
-            ($mapping:expr) => {{
-                let (r1, r2) = cursor.next_regs()?;
-                next_ip = cursor.position();
-                let value = ($mapping)(self.get_reg_checked(r1)?, self.get_reg_checked(r2)?);
-                self.set_reg(r1, value);
-            }};
-        }
-        match op {
-            ADD => math_inst!(|w1: u64, w2: u64| w1.wrapping_add(w2)),
-            MUL => math_inst!(|w1: u64, w2: u64| w1.wrapping_mul(w2)),
-            DIV => math_inst!(|w1: u64, w2: u64| w1.wrapping_div(w2)),
-            MOD => math_inst!(|w1: u64, w2: u64| w1 % w2),
-            INEG => todo!(),
-            AND => math_inst!(|w1: u64, w2: u64| w1 & w2),
-            OR => math_inst!(|w1: u64, w2: u64| w1 | w2),
-            INV => {
-                let r1 = cursor.next_reg()?;
-                next_ip = cursor.position();
-                let value = self.get_reg_checked(r1)?;
-                self.set_reg(r1, !value);
-            }
-            NOT => {
-                let r1 = cursor.next_reg()?;
-                next_ip = cursor.position();
-                let value = self.get_reg(r1);
-                self.set_reg(r1, (value == 0) as Word);
-            }
-            XOR => math_inst!(|w1: u64, w2: u64| w1 ^ w2),
-            SHL => math_inst!(|w1: u64, w2: u64| w1 << w2),
-            SHR => math_inst!(|w1: u64, w2: u64| w1 >> w2),
-            CMPEQ => {
-                let (r1, r2) = cursor.next_regs()?;
-                next_ip = cursor.position();
-                let cmp = self.get_reg(r1) == self.get_reg(r2);
-                if cmp {
-                    self.insert_flags(Flags::COMPARE);
-                } else {
-                    self.remove_flags(Flags::COMPARE);
-                }
-            }
-            CMPLT => {
-                let (r1, r2) = cursor.next_regs()?;
-                next_ip = cursor.position();
-                let cmp = self.get_reg(r1) < self.get_reg(r2);
-                if cmp {
-                    self.insert_flags(Flags::COMPARE);
-                } else {
-                    self.remove_flags(Flags::COMPARE);
-                }
-            }
-            JMP => {
-                let r1 = cursor.next_reg()?;
-                next_ip = self.get_reg(r1);
-            }
-            JZ => {
-                let r1 = cursor.next_reg()?;
-                if !self.flags().contains(Flags::COMPARE) {
-                    next_ip = self.get_reg(r1);
-                } else {
-                    next_ip = cursor.position();
-                }
-            }
-            JNZ => {
-                let r1 = cursor.next_reg()?;
-                if self.flags().contains(Flags::COMPARE) {
-                    next_ip = self.get_reg(r1);
-                } else {
-                    next_ip = cursor.position();
-                }
-            }
-            LOAD => {
-                let (r1, r2) = cursor.next_regs()?;
-                next_ip = cursor.position();
-                let value = self.load(r2)?;
-                self.set_reg_checked(r1, value)?;
-            }
-            REGCOPY => math_inst!(|_: u64, w2: u64| w2),
-            STOREIMM64 => {
-                let r1 = cursor.next_reg()?;
-                // skip
-                cursor.next_u32()?;
-                let imm = cursor.next_u64()?;
-                next_ip = cursor.position();
-                self.set_reg(r1, imm);
-            }
-            STOREIMM32 => {
-                let r1 = cursor.next_reg()?;
-                let imm = cursor.next_u32()?;
-                next_ip = cursor.position();
-                self.set_reg(r1, imm as u64);
-            }
-            MEMCOPY => {
-                let (r1, r2) = cursor.next_regs()?;
-                next_ip = cursor.position();
-                let value = self.load(r2)?;
-                self.store(r1, value)?;
-            }
-            STORE => {
-                let (r1, r2) = cursor.next_regs()?;
-                next_ip = cursor.position();
-                let value = self.get_reg(r1);
-                self.store(r2, value)?;
-            }
-            HALT => {
-                next_ip = cursor.position();
-                self.insert_flags(Flags::HALT);
-            }
-            NOP => {
-                next_ip = cursor.position();
-            }
-            _ => panic!("unknown instruction opcode: 0x{:04x}", op),
-        }
-
-        println!("op: {:04x} {}", op, inst_name(op).unwrap());
-        println!("ip: {:05x}", self.ip());
-        println!("next_ip: {:05x}", next_ip);
-        let mut _line = String::new();
-        stdin().read_line(&mut _line).unwrap();
-    
         self.set_reg(IP, next_ip);
-
         Ok(())
     }
+
+    fn next_ip(&self) -> Result<Addr> {
+        let ip = self.ip();
+        let op = self.get_inst_op(ip)?;
+        Ok(ip + (inst_len(op) as u64))
+    }
+
+    fn with_regs<F, B>(&mut self, r1: Reg, r2: Reg, mapping: F) -> B
+        where F: FnOnce(Word, Word) -> B
+    {
+        let w1 = self.get_reg(r1);
+        let w2 = self.get_reg(r2);
+        (mapping)(w1, w2)
+    }
+}
+
+impl VisitInst for Vm {
+    type Out = Addr;
+
+    fn cursor(&self) -> MemCursor {
+        self.mem_cursor(self.ip() as usize)
+    }
+
+    fn add(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        let out = self.with_regs(r1, r2, |w1, w2| w1.wrapping_add(w2));
+        self.set_reg(r1, out);
+        self.next_ip()
+    }
+
+    fn mul(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        let out = self.with_regs(r1, r2, |w1, w2| w1.wrapping_mul(w2));
+        self.set_reg(r1, out);
+        self.next_ip()
+    }
+
+    fn div(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        // TODO : check w2 == 0 and throw error/exception
+        let out = self.with_regs(r1, r2, |w1, w2| w1 / w2);
+        self.set_reg(r1, out);
+        self.next_ip()
+    }
+
+    fn mod_(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        // TODO : check w2 == 0 and throw error/exception
+        let out = self.with_regs(r1, r2, |w1, w2| w1 % w2);
+        self.set_reg(r1, out);
+        self.next_ip()
+    }
+
+    fn ineg(&mut self, r1: Reg) -> Result<Self::Out> {
+        let w1 = self.get_reg(r1);
+        self.set_reg(r1, (!w1).wrapping_add(1));
+        self.next_ip()
+    }
+
+    fn and(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        let out = self.with_regs(r1, r2, |w1, w2| w1 & w2);
+        self.set_reg(r1, out);
+        self.next_ip()
+    }
+
+    fn or(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        let out = self.with_regs(r1, r2, |w1, w2| w1 | w2);
+        self.set_reg(r1, out);
+        self.next_ip()
+    }
+
+    fn inv(&mut self, r1: Reg) -> Result<Self::Out> {
+        let w1 = self.get_reg(r1);
+        self.set_reg(r1, !w1);
+        self.next_ip()
+    }
+
+    fn not(&mut self, r1: Reg) -> Result<Self::Out> {
+        let w1 = self.get_reg(r1);
+        self.set_reg(r1, (w1 == 0) as Word);
+        self.next_ip()
+    }
+
+    fn xor(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        let out = self.with_regs(r1, r2, |w1, w2| w1 ^ w2);
+        self.set_reg(r1, out);
+        self.next_ip()
+    }
+
+    fn shl(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        let out = self.with_regs(r1, r2, |w1, w2| w1 << w2);
+        self.set_reg(r1, out);
+        self.next_ip()
+    }
+
+    fn shr(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        let out = self.with_regs(r1, r2, |w1, w2| w1 >> w2);
+        self.set_reg(r1, out);
+        self.next_ip()
+    }
+
+    fn cmpeq(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        let cmp = self.with_regs(r1, r2, |w1, w2| w1 == w2); 
+        if cmp {
+            self.insert_flags(Flags::COMPARE);
+        } else {
+            self.remove_flags(Flags::COMPARE);
+        }
+        self.next_ip()
+    }
+
+    fn cmplt(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        let cmp = self.with_regs(r1, r2, |w1, w2| w1 < w2); 
+        if cmp {
+            self.insert_flags(Flags::COMPARE);
+        } else {
+            self.remove_flags(Flags::COMPARE);
+        }
+        self.next_ip()
+    }
+
+    fn jmp(&mut self, r1: Reg) -> Result<Self::Out> {
+        let addr = self.get_reg(r1);
+        Ok(addr)
+    }
+
+    fn jz(&mut self, r1: Reg) -> Result<Self::Out> {
+        if !self.flags().contains(Flags::COMPARE) {
+            Ok(self.get_reg(r1))
+        } else {
+            self.next_ip()
+        }
+    }
+
+    fn jnz(&mut self, r1: Reg) -> Result<Self::Out> {
+        if self.flags().contains(Flags::COMPARE) {
+            Ok(self.get_reg(r1))
+        } else {
+            self.next_ip()
+        }
+    }
+
+    fn load(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        let value = Vm::load(self, r2)?;
+        self.set_reg_checked(r1, value)?;
+        self.next_ip()
+    }
+
+    fn regcopy(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        let value = self.get_reg_checked(r2)?;
+        self.set_reg_checked(r1, value)?;
+        self.next_ip()
+    }
+
+    fn storeimm64(&mut self, r1: Reg, w1: Word) -> Result<Self::Out> {
+        self.set_reg_checked(r1, w1)?;
+        self.next_ip()
+    }
+
+    fn storeimm32(&mut self, r1: Reg, w1: HalfWord) -> Result<Self::Out> {
+        self.set_reg_checked(r1, w1 as Word)?;
+        self.next_ip()
+    }
+
+    fn memcopy(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        let value = Vm::load(self, r2)?;
+        self.store(r1, value)?;
+        self.next_ip()
+    }
+
+    fn store(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out> {
+        let value = self.get_reg_checked(r1)?;
+        self.store(r2, value)?;
+        self.next_ip()
+    }
+
+    fn halt(&mut self) -> Result<Self::Out> {
+        self.insert_flags(Flags::HALT);
+        self.next_ip()
+    }
+
+    fn nop(&mut self) -> Result<Self::Out> {
+        self.next_ip()
+    }
 }

src/vm/visit.rs

@@ -0,0 +1,96 @@
+use crate::vm::{
+    error::*,
+    inst::*,
+    mem::MemCursor,
+    reg::Reg,
+    vm::{HalfWord, Word},
+};
+
+pub trait VisitInst {
+    type Out;
+
+    fn cursor(&self) -> MemCursor;
+    fn add(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out>;
+    fn mul(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out>;
+    fn div(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out>;
+    fn mod_(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out>;
+    fn ineg(&mut self, r1: Reg) -> Result<Self::Out>;
+    fn and(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out>;
+    fn or(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out>;
+    fn inv(&mut self, r1: Reg) -> Result<Self::Out>;
+    fn not(&mut self, r1: Reg) -> Result<Self::Out>;
+    fn xor(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out>;
+    fn shl(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out>;
+    fn shr(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out>;
+    fn cmpeq(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out>;
+    fn cmplt(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out>;
+    fn jmp(&mut self, r1: Reg) -> Result<Self::Out>;
+    fn jz(&mut self, r1: Reg) -> Result<Self::Out>;
+    fn jnz(&mut self, r1: Reg) -> Result<Self::Out>;
+    fn load(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out>;
+    fn regcopy(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out>;
+    fn storeimm64(&mut self, r1: Reg, w1: Word) -> Result<Self::Out>;
+    fn storeimm32(&mut self, r1: Reg, w1: HalfWord) -> Result<Self::Out>;
+    fn memcopy(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out>;
+    fn store(&mut self, r1: Reg, r2: Reg) -> Result<Self::Out>;
+    fn halt(&mut self) -> Result<Self::Out>;
+    fn nop(&mut self) -> Result<Self::Out>;
+
+    fn visit_inst(&mut self) -> Result<Self::Out> {
+        let mut cursor = self.cursor();
+        //panic!("cursor pos: {}", cursor.position());
+        let op = cursor.next_u16()?;
+
+        macro_rules! r1_r2_inst {
+            ($fun:ident) => {{
+                let (r1, r2) = cursor.next_regs()?;
+                self.$fun(r1, r2)
+            }};
+        }
+        macro_rules! r1_inst {
+            ($fun:ident) => {{
+                let r1 = cursor.next_reg()?;
+                self.$fun(r1)
+            }};
+        }
+
+        match op {
+            ADD => r1_r2_inst!(add),
+            MUL => r1_r2_inst!(mul),
+            DIV => r1_r2_inst!(div),
+            MOD => r1_r2_inst!(mod_),
+            INEG => r1_inst!(ineg),
+            AND => r1_r2_inst!(and),
+            OR => r1_r2_inst!(or),
+            INV => r1_inst!(inv),
+            NOT => r1_inst!(not),
+            XOR => r1_r2_inst!(xor),
+            SHL => r1_r2_inst!(shl),
+            SHR => r1_r2_inst!(shr),
+            CMPEQ => r1_r2_inst!(cmpeq),
+            CMPLT => r1_r2_inst!(cmplt),
+            JMP => r1_inst!(jmp),
+            JZ => r1_inst!(jz),
+            JNZ => r1_inst!(jnz),
+            LOAD => r1_r2_inst!(load),
+            REGCOPY => r1_r2_inst!(regcopy),
+            STOREIMM64 => {
+                let r1 = cursor.next_reg()?;
+                // skip
+                cursor.next_u32()?;
+                let imm = cursor.next_u64()?;
+                self.storeimm64(r1, imm)
+            }
+            STOREIMM32 => {
+                let r1 = cursor.next_reg()?;
+                let imm = cursor.next_u32()?;
+                self.storeimm32(r1, imm)
+            }
+            MEMCOPY => r1_r2_inst!(memcopy),
+            STORE => r1_r2_inst!(store),
+            HALT => self.halt(),
+            NOP => self.nop(),
+            _ => Err(VmError::IllegalOp { op }),
+        }
+    }
+}

src/vm/vm.rs

@@ -1,4 +1,4 @@
-use crate::vm::{error::*, flags::*, mem::*, obj::obj::*, reg::*};
+use crate::vm::{error::*, flags::*, inst::InstOp, mem::*, obj::obj::*, reg::*};
 use byteorder::{WriteBytesExt, LE};
 use std::{io::Cursor, mem};
 
@@ -101,6 +101,11 @@ impl Vm {
         Ok(self.mem_cursor(addr as usize).next_u32().unwrap())
     }
 
+    pub fn get_inst_op(&self, addr: Addr) -> Result<InstOp> {
+        self.check_read(addr, 2)?;
+        Ok(self.mem_cursor(addr as usize).next_u16().unwrap())
+    }
+
     pub fn get_byte(&self, addr: Addr) -> Result<u8> {
         self.check_addr(addr)?;
         Ok(self.mem_cursor(addr as usize).next_u8().unwrap())

test.asm

@@ -1,28 +1,25 @@
-data $0x1000 .. $0x2000 {
-}
-
 code $0x0 {
     main:
-        storeimm %r00, $0xDEAD
+        storeimm32 %r00, $0xDEAD
-        storeimm %r01, $16
+        storeimm64 %r01, $16
 
         shl %r00, %r01
 
-        storeimm %r01, $0xBEEF
+        storeimm32 %r01, $0xBEEF
         or %r00, %r01
 
-        storeimm %r01, $0xDEADBEEF
+        storeimm64 %r01, $0xDEADBEEF
 
         cmpeq %r00, %r01
-        storeimm %r00, failure
+        storeimm32 %r00, failure
-        storeimm %r01, ok
+        storeimm64 %r01, ok
 
         jz %r00
 
         jmp %r01
 
     failure:
-        storeimm %status, $1
+        storeimm32 %status, $1
 
     ok: halt