diff options
| author | Rose Hogenson <rosehogenson@posteo.net> | 2024-06-02 09:17:55 -0700 |
|---|---|---|
| committer | Rose Hogenson <rosehogenson@posteo.net> | 2024-06-02 09:17:55 -0700 |
| commit | 22b8b7e2478e3b205c89cc7a700058bfe39175f2 (patch) | |
| tree | ad6cf17b8729ab9af71d4d731162e738f3a7b057 | |
| parent | 90c4eac546d2c950b27a84ef0e390ac33a1a347d (diff) | |
| download | sml-22b8b7e2478e3b205c89cc7a700058bfe39175f2.tar.zst | |
Add strings and IO.
| -rw-r--r-- | bytecode/src/encoding.rs | 44 | ||||
| -rw-r--r-- | bytecode/src/heap.rs | 7 | ||||
| -rw-r--r-- | bytecode/src/main.rs | 132 | ||||
| -rw-r--r-- | codegen.sml | 6 | ||||
| -rw-r--r-- | cps.sml | 9 | ||||
| -rw-r--r-- | elab.sml | 3 | ||||
| -rw-r--r-- | linker.sml | 17 | ||||
| -rw-r--r-- | syntax.sml | 14 |
8 files changed, 201 insertions, 31 deletions
diff --git a/bytecode/src/encoding.rs b/bytecode/src/encoding.rs index 80fcc58..92e6bf6 100644 --- a/bytecode/src/encoding.rs +++ b/bytecode/src/encoding.rs @@ -88,6 +88,28 @@ pub struct If { } #[derive(Debug, Clone, Copy)] +pub struct Read { + pub out: u8, + pub ptr: u8, + pub off: Arg, + pub len: Arg, +} + +#[derive(Debug, Clone, Copy)] +pub struct Write { + pub ptr: u8, + pub off: Arg, + pub len: Arg, +} + +#[derive(Debug, Clone, Copy)] +pub struct WriteErr { + pub ptr: u8, + pub off: Arg, + pub len: Arg, +} + +#[derive(Debug, Clone, Copy)] pub enum Op { Alloc(Alloc), Call, @@ -102,6 +124,9 @@ pub enum Op { Less(Less), Eq(Eq), If(If), + Read(Read), + Write(Write), + WriteErr(WriteErr), } struct Reader<'a> { @@ -241,6 +266,25 @@ impl Op { let target = usize::try_from(itarget)?; Op::If(If { test, target }) } + 14 => { + let out = r.parse_local()?; + let ptr = r.parse_local()?; + let off = r.parse_arg(arg1_const)?; + let len = r.parse_arg(arg2_const)?; + Op::Read(Read { out, ptr, off, len }) + } + 15 => { + let ptr = r.parse_local()?; + let off = r.parse_arg(arg1_const)?; + let len = r.parse_arg(arg2_const)?; + Op::Write(Write { ptr, off, len }) + } + 16 => { + let ptr = r.parse_local()?; + let off = r.parse_arg(arg1_const)?; + let len = r.parse_arg(arg2_const)?; + Op::WriteErr(WriteErr { ptr, off, len }) + } _ => { return Err(Box::from(format!("invalid code {}", code))); } diff --git a/bytecode/src/heap.rs b/bytecode/src/heap.rs index 970db7d..f195613 100644 --- a/bytecode/src/heap.rs +++ b/bytecode/src/heap.rs @@ -132,11 +132,14 @@ impl Heap { Ok(Value(self.buf[p])) } - pub fn poke(&mut self, p: usize, val: Value) -> Result<(), Box<dyn Error>> { + pub fn poke(&mut self, p: usize, off: usize, val: Value) -> Result<(), Box<dyn Error>> { if p >= self.buf.len() { return Err(Box::from("poke: out of range")); } - self.buf[p] = val.repr(); + if off >= self.alloc_size(p) { + return Err(Box::from("poke: out of range")); + } + self.buf[p + off] = val.repr(); Ok(()) } } diff --git a/bytecode/src/main.rs b/bytecode/src/main.rs index 4cba4ef..6ccd4b8 100644 --- a/bytecode/src/main.rs +++ b/bytecode/src/main.rs @@ -6,26 +6,69 @@ use encoding::{Arg, Op}; use heap::Heap; use std::error::Error; use std::ffi::OsString; -use std::io::{stderr, Write}; +use std::io::{stdin, BufWriter, Read, Write}; use value::Value; -struct State { - i: usize, - heap: Heap, +trait ReadArg { + fn read_arg(&self, a: Arg) -> Value; } -impl State { +impl ReadArg for Heap { fn read_arg(&self, a: Arg) -> Value { match a { - Arg::Local(l) => self.heap.locals[usize::from(l)], + Arg::Local(l) => self.locals[usize::from(l)], Arg::Const(c) => c, } } +} + +fn write( + w: &mut impl Write, + heap: &Heap, + ptr: u8, + off: Arg, + len: Arg, +) -> Result<(), Box<dyn Error>> { + let Some(ptr) = heap.locals[usize::from(ptr)].to_pointer() else { + return Err(Box::from("write needs a pointer")); + }; + let Some(ioff) = heap.read_arg(off).to_int() else { + return Err(Box::from("write needs an int")); + }; + let Ok(off) = usize::try_from(ioff) else { + return Err(Box::from("write: off is negative")); + }; + let Some(ilen) = heap.read_arg(len).to_int() else { + return Err(Box::from("write needs an int")); + }; + let Ok(len) = usize::try_from(ilen) else { + return Err(Box::from("write: len is negative")); + }; + for i in 0..len { + let Some(ibyte) = heap.peek(ptr + off + i)?.to_int() else { + return Err(Box::from("write: buffer has non-integer entries")); + }; + let Ok(byte) = u8::try_from(ibyte) else { + return Err(Box::from("write: buffer entry is not between 0 and 255")); + }; + w.write_all(&[byte])?; + } + w.flush()?; + Ok(()) +} + +struct State { + i: usize, + heap: Heap, + stdout: BufWriter<Box<dyn Write>>, + stderr: BufWriter<Box<dyn Write>>, +} +impl State { fn op(&mut self, op: Op) -> Result<(), Box<dyn Error>> { match op { Op::Alloc(op) => { - let Some(i) = self.read_arg(op.size).to_int() else { + let Some(i) = self.heap.read_arg(op.size).to_int() else { return Err(Box::from("alloc needs an int")); }; self.heap.locals[usize::from(op.out)] = Value::from_pointer(self.heap.alloc(i)?); @@ -40,55 +83,55 @@ impl State { let Some(p) = self.heap.locals[usize::from(op.ptr)].to_pointer() else { return Err(Box::from("poke needs a pointer")); }; - self.heap.poke(p + op.offset, self.read_arg(op.val))?; + self.heap.poke(p, op.offset, self.heap.read_arg(op.val))?; } Op::Peek(op) => { - let Some(p) = self.read_arg(op.val).to_pointer() else { + let Some(p) = self.heap.read_arg(op.val).to_pointer() else { return Err(Box::from("peek needs a pointer")); }; self.heap.locals[usize::from(op.out)] = self.heap.peek(p + op.offset)?; } Op::Shuf(op) => { - self.heap.locals[usize::from(op.out)] = self.read_arg(op.val); + self.heap.locals[usize::from(op.out)] = self.heap.read_arg(op.val); } Op::Exit(op) => { - let Some(i) = self.read_arg(op.val).to_int() else { + let Some(i) = self.heap.read_arg(op.val).to_int() else { std::process::exit(255); }; std::process::exit(i as i32); } Op::Add(op) => { - let Some(v1) = self.read_arg(op.val1).to_int() else { + let Some(v1) = self.heap.read_arg(op.val1).to_int() else { return Err(Box::from("add needs an int")); }; - let Some(v2) = self.read_arg(op.val2).to_int() else { + let Some(v2) = self.heap.read_arg(op.val2).to_int() else { return Err(Box::from("add needs an int")); }; self.heap.locals[usize::from(op.out)] = Value::from_int(v1.wrapping_add(v2)); } Op::Sub(op) => { - let Some(v1) = self.read_arg(op.val1).to_int() else { + let Some(v1) = self.heap.read_arg(op.val1).to_int() else { return Err(Box::from("sub needs an int")); }; - let Some(v2) = self.read_arg(op.val2).to_int() else { + let Some(v2) = self.heap.read_arg(op.val2).to_int() else { return Err(Box::from("sub needs an int")); }; self.heap.locals[usize::from(op.out)] = Value::from_int(v1.wrapping_sub(v2)); } Op::Mul(op) => { - let Some(v1) = self.read_arg(op.val1).to_int() else { + let Some(v1) = self.heap.read_arg(op.val1).to_int() else { return Err(Box::from("mul needs an int")); }; - let Some(v2) = self.read_arg(op.val2).to_int() else { + let Some(v2) = self.heap.read_arg(op.val2).to_int() else { return Err(Box::from("mul needs an int")); }; self.heap.locals[usize::from(op.out)] = Value::from_int(v1.wrapping_mul(v2)); } Op::Div(op) => { - let Some(v1) = self.read_arg(op.val1).to_int() else { + let Some(v1) = self.heap.read_arg(op.val1).to_int() else { return Err(Box::from("div needs an int")); }; - let Some(v2) = self.read_arg(op.val2).to_int() else { + let Some(v2) = self.heap.read_arg(op.val2).to_int() else { return Err(Box::from("div needs an int")); }; let Some(res) = v1.checked_div(v2) else { @@ -97,33 +140,70 @@ impl State { self.heap.locals[usize::from(op.out)] = Value::from_int(res); } Op::Less(op) => { - let Some(v1) = self.read_arg(op.val1).to_int() else { + let Some(v1) = self.heap.read_arg(op.val1).to_int() else { return Err(Box::from("less needs an int")); }; - let Some(v2) = self.read_arg(op.val2).to_int() else { + let Some(v2) = self.heap.read_arg(op.val2).to_int() else { return Err(Box::from("less needs an int")); }; let res = if v1 < v2 { 1 } else { 0 }; self.heap.locals[usize::from(op.out)] = Value::from_int(res); } Op::Eq(op) => { - let Some(v1) = self.read_arg(op.val1).to_int() else { + let Some(v1) = self.heap.read_arg(op.val1).to_int() else { return Err(Box::from("eq needs an int")); }; - let Some(v2) = self.read_arg(op.val2).to_int() else { + let Some(v2) = self.heap.read_arg(op.val2).to_int() else { return Err(Box::from("eq needs an int")); }; let res = if v1 == v2 { 1 } else { 0 }; self.heap.locals[usize::from(op.out)] = Value::from_int(res); } Op::If(op) => { - let Some(t) = self.read_arg(op.test).to_int() else { + let Some(t) = self.heap.read_arg(op.test).to_int() else { return Err(Box::from("if needs an int")); }; if t != 0 { self.i = op.target; } } + Op::Read(op) => { + let Some(ptr) = self.heap.locals[usize::from(op.ptr)].to_pointer() else { + return Err(Box::from("read needs a pointer")); + }; + let Some(ioff) = self.heap.read_arg(op.off).to_int() else { + return Err(Box::from("read needs an int")); + }; + let Ok(off) = usize::try_from(ioff) else { + return Err(Box::from("read: off is negative")); + }; + let Some(ilen) = self.heap.read_arg(op.len).to_int() else { + return Err(Box::from("read needs an int")); + }; + let Ok(len) = usize::try_from(ilen) else { + return Err(Box::from("read: len is negative")); + }; + let mut n = 0; + for byte in stdin().lock().bytes().take(len) { + let byte = byte?; + self.heap + .poke(ptr, off + n, Value::from_int(i64::from(byte)))?; + n += 1; + if byte == b'\n' { + break; + } + } + self.heap.locals[usize::from(op.out)] = Value::from_int( + i64::try_from(n) + .expect("we can never read more than len bytes, and len fit in an i64"), + ); + } + Op::Write(op) => { + write(&mut self.stdout, &self.heap, op.ptr, op.off, op.len)?; + } + Op::WriteErr(op) => { + write(&mut self.stderr, &self.heap, op.ptr, op.off, op.len)?; + } } Ok(()) } @@ -138,6 +218,8 @@ fn run() -> Result<(), Box<dyn Error>> { let mut st = State { i: 0, heap: Heap::new(), + stdout: BufWriter::new(Box::new(std::io::stdout().lock())), + stderr: BufWriter::new(Box::new(std::io::stderr().lock())), }; loop { let (n, op) = Op::parse(&prog[st.i..])?; @@ -148,7 +230,7 @@ fn run() -> Result<(), Box<dyn Error>> { fn main() { if let Err(err) = run() { - let _ = writeln!(stderr(), "FAIL: {}", err); + let _ = writeln!(std::io::stderr(), "FAIL: {}", err); std::process::exit(255); } } diff --git a/codegen.sml b/codegen.sml index 9dbee51..67b2d30 100644 --- a/codegen.sml +++ b/codegen.sml @@ -169,6 +169,12 @@ struct @ Syntax.OLabel trueLabel :: go k1 end + | Syntax.CPrimop (Syntax.PRead, [Syntax.VVar ptr, off, len], [res], [k]) => + Syntax.ORead (translate res, translate ptr, translateVal off, translateVal len) :: go k + | Syntax.CPrimop (Syntax.PWrite, [Syntax.VVar ptr, off, len], _, [k]) => + Syntax.OWrite (translate ptr, translateVal off, translateVal len) :: go k + | Syntax.CPrimop (Syntax.PWriteErr, [Syntax.VVar ptr, off, len], _, [k]) => + Syntax.OWriteErr (translate ptr, translateVal off, translateVal len) :: go k | _ => raise Fail ("malformed CPS:\n" ^ Syntax.cexpToString expr) in go expr end @@ -45,7 +45,14 @@ struct Syntax.CApp (f', [x', Syntax.VVar addr])))) end | Syntax.LInt i => cont (Syntax.VInt i) - | Syntax.LString s => cont (Syntax.VString s) + | Syntax.LString s => + let val temp = Gensym.new () + in + Syntax.CRecord + ( [(map (fn c => (Syntax.VInt (Char.ord c), [])) (String.explode s), temp)] + , cont (Syntax.VVar temp) + ) + end | Syntax.LRecord [] => cont (Syntax.VInt 0) | Syntax.LSelect (i, expr) => let val temp = Gensym.new () @@ -10,6 +10,9 @@ struct | "sub" => Syntax.PSub | "mul" => Syntax.PMul | "div" => Syntax.PDiv + | "read" => Syntax.PRead + | "write" => Syntax.PWrite + | "writeErr" => Syntax.PWriteErr | _ => raise Fail ("invalid op: " ^ s) fun enumerate (l : 'a list) : (int * 'a) list = @@ -47,7 +47,6 @@ struct fun writeValue w (Syntax.VVar v) = writeVar w v | writeValue w (Syntax.VLabel l) = writeInt w (getOpt (IntMap.lookup l m, 0)) | writeValue w (Syntax.VInt i) = writeInt w i - | writeValue w (Syntax.VString _) = raise Fail "I don't support strings yet" in case oper of Syntax.OAlloc (r, v) => @@ -108,6 +107,22 @@ struct writeValue w condition ; writeValue w (Syntax.VLabel label)) | Syntax.OLabel _ => () + | Syntax.ORead (r, ptr, off, len) => + (makeOpcode w 14 (isConst off) (isConst len) ; + writeVar w r ; + writeVar w ptr ; + writeValue w off ; + writeValue w len) + | Syntax.OWrite (ptr, off, len) => + (makeOpcode w 15 (isConst off) (isConst len) ; + writeVar w ptr ; + writeValue w off ; + writeValue w len) + | Syntax.OWriteErr (ptr, off, len) => + (makeOpcode w 16 (isConst off) (isConst len) ; + writeVar w ptr ; + writeValue w off ; + writeValue w len) end fun link (program : Syntax.opcode list) : Word8Vector.vector = @@ -47,6 +47,9 @@ struct | PLess | PEq | PIf + | PRead + | PWrite + | PWriteErr datatype lexp = LVar of var @@ -65,7 +68,6 @@ struct VVar of var | VLabel of var | VInt of int - | VString of string datatype cexp = CRecord of ((value * int list) list * var) list * cexp @@ -89,6 +91,9 @@ struct | OEq of var * value * value | OIf of value * var | OLabel of var + | ORead of var * var * value * value + | OWrite of var * value * value + | OWriteErr of var * value * value fun listToString (show : 'a -> string) (l : 'a list) = "[" ^ String.concatWith ", " (map show l) ^ "]" @@ -163,6 +168,9 @@ struct | PLess => "PLess" | PEq => "PEq" | PIf => "PIf" + | PRead => "PRead" + | PWrite => "PWrite" + | PWriteErr => "PWriteErr" fun lexpToStringI (indent : string) (x : lexp) : string = case x of @@ -184,7 +192,6 @@ struct VVar v => "VVar " ^ Int.toString v | VLabel l => "VLabel " ^ Int.toString l | VInt i => "VInt " ^ Int.toString i - | VString s => "VString " ^ quote s fun cexpToStringI (indent : string) (x : cexp) : string = let @@ -216,4 +223,7 @@ struct | OEq (r, v1, v2) => "Var " ^ Int.toString r ^ " = OEq (" ^ valueToString v1 ^ ", " ^ valueToString v2 ^ ")" | OIf (condition, target) => "OIf (" ^ valueToString condition ^ ") goto " ^ Int.toString target | OLabel l => "OLabel " ^ Int.toString l + | ORead (r, ptr, off, len) => "Var " ^ Int.toString r ^ " = ORead (Var " ^ Int.toString ptr ^ ", " ^ valueToString off ^ ", " ^ valueToString len ^ ")" + | OWrite (ptr, off, len) => "OWrite (Var " ^ Int.toString ptr ^ ", " ^ valueToString off ^ ", " ^ valueToString len ^ ")" + | OWriteErr (ptr, off, len) => "OWriteErr (Var " ^ Int.toString ptr ^ ", " ^ valueToString off ^ ", " ^ valueToString len ^ ")" end |
