lama_byterun/src/X86.ml

878 lines
32 KiB
OCaml
Raw Normal View History

2018-05-16 09:24:40 +03:00
open GT
2019-10-14 19:44:33 +03:00
open Language
open SM
2018-03-05 00:54:50 +03:00
(* X86 codegeneration interface *)
(* The registers: *)
2018-03-07 10:13:34 +03:00
let regs = [|"%ebx"; "%ecx"; "%esi"; "%edi"; "%eax"; "%edx"; "%ebp"; "%esp"|]
2018-03-05 00:54:50 +03:00
2019-09-10 00:46:10 +03:00
(* We can not freely operate with all register; only 3 by now *)
2018-03-05 00:54:50 +03:00
let num_of_regs = Array.length regs - 5
(* We need to know the word size to calculate offsets correctly *)
2018-05-16 09:24:40 +03:00
let word_size = 4;;
2018-03-05 00:54:50 +03:00
(* We need to distinguish the following operand types: *)
2019-09-10 00:46:10 +03:00
@type opnd =
2019-10-16 21:07:27 +03:00
| R of int (* hard register *)
| S of int (* a position on the hardware stack *)
2019-12-29 01:12:40 +03:00
| C (* a saved closure *)
2019-10-16 21:07:27 +03:00
| M of string (* a named memory location *)
| L of int (* an immediate operand *)
| I of int * opnd (* an indirect operand with offset *)
2018-05-16 09:24:40 +03:00
with show
2018-03-05 00:54:50 +03:00
2018-11-21 14:23:35 +03:00
let show_opnd = show(opnd)
2019-09-10 00:46:10 +03:00
(* For convenience we define the following synonyms for the registers: *)
2018-03-05 00:54:50 +03:00
let ebx = R 0
let ecx = R 1
let esi = R 2
let edi = R 3
let eax = R 4
let edx = R 5
let ebp = R 6
let esp = R 7
(* Now x86 instruction (we do not need all of them): *)
type instr =
2019-04-10 22:15:08 +03:00
(* copies a value from the first to the second operand *) | Mov of opnd * opnd
(* loads an address of the first operand into the second *) | Lea of opnd * opnd
(* makes a binary operation; note, the first operand *) | Binop of string * opnd * opnd
(* designates x86 operator, not the source language one *)
(* x86 integer division, see instruction set reference *) | IDiv of opnd
(* see instruction set reference *) | Cltd
(* sets a value from flags; the first operand is the *) | Set of string * string
2019-09-10 00:46:10 +03:00
(* suffix, which determines the value being set, the *)
2019-04-10 22:15:08 +03:00
(* the second --- (sub)register name *)
(* pushes the operand on the hardware stack *) | Push of opnd
(* pops from the hardware stack to the operand *) | Pop of opnd
(* call a function by a name *) | Call of string
2019-10-16 01:13:52 +03:00
(* call a function by indirect address *) | CallI of opnd
2019-04-10 22:15:08 +03:00
(* returns from a function *) | Ret
(* a label in the code *) | Label of string
(* a conditional jump *) | CJmp of string * string
(* a non-conditional jump *) | Jmp of string
(* directive *) | Meta of string
2019-04-10 22:15:08 +03:00
(* arithmetic correction: decrement *) | Dec of opnd
(* arithmetic correction: or 0x0001 *) | Or1 of opnd
(* arithmetic correction: shl 1 *) | Sal1 of opnd
(* arithmetic correction: shr 1 *) | Sar1 of opnd
2019-09-10 00:46:10 +03:00
| Repmovsl
2018-03-05 00:54:50 +03:00
(* Instruction printer *)
2020-09-10 09:07:38 +03:00
let stack_offset i =
if i >= 0
then (i+1) * word_size
else 8 + (-i-1) * word_size
2018-03-05 00:54:50 +03:00
let show instr =
2020-09-10 09:07:38 +03:00
let rec opnd = function
| R i -> regs.(i)
| C -> "4(%ebp)"
| S i -> if i >= 0
then Printf.sprintf "-%d(%%ebp)" (stack_offset i)
else Printf.sprintf "%d(%%ebp)" (stack_offset i)
| M x -> x
| L i -> Printf.sprintf "$%d" i
| I (0, x) -> Printf.sprintf "(%s)" (opnd x)
| I (n, x) -> Printf.sprintf "%d(%s)" n (opnd x)
in
2018-03-05 00:54:50 +03:00
let binop = function
| "+" -> "addl"
| "-" -> "subl"
| "*" -> "imull"
| "&&" -> "andl"
| "!!" -> "orl"
| "^" -> "xorl"
| "cmp" -> "cmpl"
| "test" -> "test"
| _ -> failwith "unknown binary operator"
2018-03-05 00:54:50 +03:00
in
match instr with
| Cltd -> "\tcltd"
| Set (suf, s) -> Printf.sprintf "\tset%s\t%s" suf s
| IDiv s1 -> Printf.sprintf "\tidivl\t%s" (opnd s1)
| Binop (op, s1, s2) -> Printf.sprintf "\t%s\t%s,\t%s" (binop op) (opnd s1) (opnd s2)
| Mov (s1, s2) -> Printf.sprintf "\tmovl\t%s,\t%s" (opnd s1) (opnd s2)
2019-04-10 22:15:08 +03:00
| Lea (x, y) -> Printf.sprintf "\tleal\t%s,\t%s" (opnd x) (opnd y)
2018-03-05 00:54:50 +03:00
| Push s -> Printf.sprintf "\tpushl\t%s" (opnd s)
| Pop s -> Printf.sprintf "\tpopl\t%s" (opnd s)
| Ret -> "\tret"
| Call p -> Printf.sprintf "\tcall\t%s" p
2019-10-16 01:13:52 +03:00
| CallI o -> Printf.sprintf "\tcall\t*(%s)" (opnd o)
| Label l -> Printf.sprintf "%s:\n" l
| Jmp l -> Printf.sprintf "\tjmp\t%s" l
| CJmp (s , l) -> Printf.sprintf "\tj%s\t%s" s l
2018-04-11 00:47:46 +03:00
| Meta s -> Printf.sprintf "%s\n" s
| Dec s -> Printf.sprintf "\tdecl\t%s" (opnd s)
| Or1 s -> Printf.sprintf "\torl\t$0x0001,\t%s" (opnd s)
| Sal1 s -> Printf.sprintf "\tsall\t%s" (opnd s)
| Sar1 s -> Printf.sprintf "\tsarl\t%s" (opnd s)
2018-12-12 12:42:38 +03:00
| Repmovsl -> Printf.sprintf "\trep movsl\t"
2019-09-10 00:46:10 +03:00
2018-03-05 00:54:50 +03:00
(* Opening stack machine to use instructions without fully qualified names *)
open SM
(* Symbolic stack machine evaluator
compile : env -> prg -> env * instr list
2018-03-07 10:13:34 +03:00
Take an environment, a stack machine program, and returns a pair --- the updated environment and the list
2018-03-05 00:54:50 +03:00
of x86 instructions
*)
let compile cmd env imports code =
(* SM.print_prg code; *)
2018-05-16 09:24:40 +03:00
flush stdout;
2018-03-05 00:54:50 +03:00
let suffix = function
| "<" -> "l"
| "<=" -> "le"
| "==" -> "e"
| "!=" -> "ne"
| ">=" -> "ge"
| ">" -> "g"
2019-09-10 00:46:10 +03:00
| _ -> failwith "unknown operator"
2020-02-13 18:56:27 +03:00
in
let box n = (n lsl 1) lor 1 in
2019-09-10 00:46:10 +03:00
let rec compile' env scode =
2018-03-05 00:54:50 +03:00
let on_stack = function S _ -> true | _ -> false in
2018-05-16 09:24:40 +03:00
let mov x s = if on_stack x && on_stack s then [Mov (x, eax); Mov (eax, s)] else [Mov (x, s)] in
2020-03-23 00:49:20 +03:00
let callc env n tail =
let tail = tail && env#nargs = n in
if tail
then (
2019-10-16 01:13:52 +03:00
let rec push_args env acc = function
| 0 -> env, acc
| n -> let x, env = env#pop in
2020-03-23 00:49:20 +03:00
if x = env#loc (Value.Arg (n-1))
then push_args env acc (n-1)
else push_args env ((mov x (env#loc (Value.Arg (n-1)))) @ acc) (n-1)
2019-10-16 01:13:52 +03:00
in
2020-03-23 00:49:20 +03:00
let env , pushs = push_args env [] n in
let closure, env = env#pop in
let y , env = env#allocate in
env, pushs @ [Mov (closure, edx);
Mov (I(0, edx), eax);
Mov (ebp, esp);
Pop (ebp)] @
(if env#has_closure then [Pop ebx] else []) @
[Jmp "*%eax"] (* UGLY!!! *)
)
else (
let pushr, popr =
List.split @@ List.map (fun r -> (Push r, Pop r)) (env#live_registers n)
2019-12-29 01:12:40 +03:00
in
2020-03-23 00:49:20 +03:00
let pushr, popr = env#save_closure @ pushr, env#rest_closure @ popr in
let env, code =
let rec push_args env acc = function
| 0 -> env, acc
| n -> let x, env = env#pop in
push_args env ((Push x)::acc) (n-1)
in
let env, pushs = push_args env [] n in
let pushs = List.rev pushs in
let closure, env = env#pop in
let call_closure =
if on_stack closure
then [Mov (closure, edx); Mov (edx, eax); CallI eax]
else [Mov (closure, edx); CallI closure]
in
env, pushr @ pushs @ call_closure @ [Binop ("+", L (word_size * List.length pushs), esp)] @ (List.rev popr)
in
let y, env = env#allocate in env, code @ [Mov (eax, y)]
)
2019-10-16 01:13:52 +03:00
in
2020-03-23 00:49:20 +03:00
let call env f n tail =
2020-04-10 03:15:18 +03:00
let tail = tail && env#nargs = n && f.[0] <> '.' in
let f =
match f.[0] with '.' -> "B" ^ String.sub f 1 (String.length f - 1) | _ -> f
2018-04-30 17:18:41 +03:00
in
2020-03-23 00:49:20 +03:00
if tail
then (
let rec push_args env acc = function
2020-03-23 00:49:20 +03:00
| 0 -> env, acc
| n -> let x, env = env#pop in
if x = env#loc (Value.Arg (n-1))
then push_args env acc (n-1)
else push_args env ((mov x (env#loc (Value.Arg (n-1)))) @ acc) (n-1)
in
let env, pushs = push_args env [] n in
2020-03-23 00:49:20 +03:00
let y, env = env#allocate in
2020-04-10 03:15:18 +03:00
env, pushs @ [Mov (ebp, esp); Pop (ebp)] @ (if env#has_closure then [Pop ebx] else []) @ [Jmp f]
2020-03-23 00:49:20 +03:00
)
else (
let pushr, popr =
List.split @@ List.map (fun r -> (Push r, Pop r)) (env#live_registers n)
in
let pushr, popr = env#save_closure @ pushr, env#rest_closure @ popr in
let env, code =
let rec push_args env acc = function
| 0 -> env, acc
| n -> let x, env = env#pop in
push_args env ((Push x)::acc) (n-1)
in
let env, pushs = push_args env [] n in
let pushs =
match f with
| "Barray" -> List.rev @@ (Push (L (box n))) :: pushs
| "Bsexp" -> List.rev @@ (Push (L (box n))) :: pushs
| "Bsta" -> pushs
| _ -> List.rev pushs
in
env, pushr @ pushs @ [Call f; Binop ("+", L (word_size * List.length pushs), esp)] @ (List.rev popr)
in
2020-03-23 00:49:20 +03:00
let y, env = env#allocate in env, code @ [Mov (eax, y)]
)
2018-04-30 17:18:41 +03:00
in
2018-03-05 00:54:50 +03:00
match scode with
| [] -> env, []
| instr :: scode' ->
2020-01-15 22:33:46 +03:00
let stack = "" (* env#show_stack*) in
(* Printf.printf "insn=%s, stack=%s\n%!" (GT.show(insn) instr) (env#show_stack); *)
2018-03-05 00:54:50 +03:00
let env', code' =
if env#is_barrier
then match instr with
| LABEL s -> if env#has_stack s then (env#drop_barrier)#retrieve_stack s, [Label s] else env#drop_stack, []
| FLABEL s -> env#drop_barrier, [Label s]
| SLABEL s -> env, [Label s]
| _ -> env, []
else
2018-03-05 00:54:50 +03:00
match instr with
2019-11-24 02:30:32 +03:00
| PUBLIC name -> env#register_public name, []
| EXTERN name -> env#register_extern name, []
2021-10-03 17:10:21 +03:00
| IMPORT name -> env, []
2019-11-24 02:30:32 +03:00
2019-12-29 01:12:40 +03:00
| CLOSURE (name, closure) ->
2019-10-16 01:13:52 +03:00
let pushr, popr =
List.split @@ List.map (fun r -> (Push r, Pop r)) (env#live_registers 0)
in
let closure_len = List.length closure in
let push_closure =
List.map (fun d -> Push (env#loc d)) @@ List.rev closure
in
let s, env = env#allocate in
(env,
pushr @
push_closure @
[Push (M ("$" ^ name));
2020-02-13 18:56:27 +03:00
Push (L (box closure_len));
2019-10-16 01:13:52 +03:00
Call "Bclosure";
Binop ("+", L (word_size * (closure_len + 2)), esp);
Mov (eax, s)] @
2019-12-29 01:12:40 +03:00
List.rev popr @ env#reload_closure)
2019-10-16 01:13:52 +03:00
2018-03-08 01:00:01 +03:00
| CONST n ->
let s, env' = env#allocate in
2020-02-13 18:56:27 +03:00
(env', [Mov (L (box n), s)])
2019-09-10 00:46:10 +03:00
2018-04-30 17:18:41 +03:00
| STRING s ->
let s, env = env#string s in
let l, env = env#allocate in
2020-03-23 00:49:20 +03:00
let env, call = call env ".string" 1 false in
2018-04-30 17:18:41 +03:00
(env, Mov (M ("$" ^ s), l) :: call)
2019-04-10 22:15:08 +03:00
| LDA x ->
let s, env' = (env #variable x)#allocate in
let s', env''= env'#allocate in
env'',
[Lea (env'#loc x, eax); Mov (eax, s); Mov (eax, s')]
2019-09-10 00:46:10 +03:00
2018-03-05 00:54:50 +03:00
| LD x ->
2018-11-21 14:23:35 +03:00
let s, env' = (env#variable x)#allocate in
2018-03-05 00:54:50 +03:00
env',
(match s with
2018-04-11 00:47:46 +03:00
| S _ | M _ -> [Mov (env'#loc x, eax); Mov (eax, s)]
| _ -> [Mov (env'#loc x, s)]
2018-05-16 09:24:40 +03:00
)
2019-04-11 17:31:45 +03:00
2019-09-10 00:46:10 +03:00
| ST x ->
2019-04-11 17:31:45 +03:00
let env' = env#variable x in
let s = env'#peek in
env',
(match s with
| S _ | M _ -> [Mov (s, eax); Mov (eax, env'#loc x)]
| _ -> [Mov (s, env'#loc x)]
)
2019-09-10 00:46:10 +03:00
2019-04-10 22:15:08 +03:00
| STA ->
2020-03-23 00:49:20 +03:00
call env ".sta" 3 false
2019-09-10 00:46:10 +03:00
2019-04-11 17:31:45 +03:00
| STI ->
2019-04-10 22:15:08 +03:00
let v, x, env' = env#pop2 in
env'#push x,
2019-04-10 22:15:08 +03:00
(match x with
2019-10-16 21:07:27 +03:00
| S _ | M _ -> [Mov (v, edx); Mov (x, eax); Mov (edx, I (0, eax)); Mov (edx, x)] @ env#reload_closure
| _ -> [Mov (v, eax); Mov (eax, I (0, x)); Mov (eax, x)]
2019-09-10 00:46:10 +03:00
)
| BINOP op ->
2018-03-11 15:41:35 +03:00
let x, y, env' = env#pop2 in
env'#push y,
2023-07-24 14:22:44 +02:00
(* (match op with
|"<" | "<=" | "==" | "!=" | ">=" | ">" ->
[Push (eax);
Push (edx);
Mov (y, eax);
Binop("&&", L(1), eax);
Mov (x, edx);
Binop("&&", L(1), edx);
Binop("cmp", eax, edx);
CJmp ("nz", "_ERROR2");
Pop (edx);
Pop (eax)]
(* | "+" | "-" | "*" | "/" -> *)
| _ ->
[Mov (y, eax);
Binop("&&", L(1), eax);
Binop("cmp", L(0), eax);
CJmp ("z", "_ERROR");
Mov (x, eax);
Binop("&&", L(1), eax);
Binop("cmp", L(0), eax);
CJmp ("z", "_ERROR")]
2023-07-24 14:22:44 +02:00
| _ -> []) @ *)
(match op with
| "/" ->
[Mov (y, eax);
Sar1 eax;
Cltd;
(* x := x >> 1 ?? *)
Sar1 x; (*!!!*)
IDiv x;
Sal1 eax;
Or1 eax;
Mov (eax, y)
]
| "%" ->
2018-03-11 15:41:35 +03:00
[Mov (y, eax);
Sar1 eax;
2018-03-11 15:41:35 +03:00
Cltd;
(* x := x >> 1 ?? *)
Sar1 x; (*!!!*)
2018-03-11 15:41:35 +03:00
IDiv x;
Sal1 edx;
Or1 edx;
Mov (edx, y)
2019-10-16 21:07:27 +03:00
] @ env#reload_closure
2018-03-11 15:41:35 +03:00
| "<" | "<=" | "==" | "!=" | ">=" | ">" ->
(match x with
| M _ | S _ ->
2018-03-07 10:13:34 +03:00
[Binop ("^", eax, eax);
2018-03-05 00:54:50 +03:00
Mov (x, edx);
Binop ("cmp", edx, y);
Set (suffix op, "%al");
Sal1 eax;
Or1 eax;
2018-03-05 00:54:50 +03:00
Mov (eax, y)
2019-10-16 21:07:27 +03:00
] @ env#reload_closure
2018-03-11 15:41:35 +03:00
| _ ->
2018-03-05 00:54:50 +03:00
[Binop ("^" , eax, eax);
Binop ("cmp", x, y);
Set (suffix op, "%al");
Sal1 eax;
2019-09-10 00:46:10 +03:00
Or1 eax;
2018-03-05 00:54:50 +03:00
Mov (eax, y)
]
2018-03-11 15:41:35 +03:00
)
| "*" ->
if on_stack y
then [Dec y; Mov (x, eax); Sar1 eax; Binop (op, y, eax); Or1 eax; Mov (eax, y)]
else [Dec y; Mov (x, eax); Sar1 eax; Binop (op, eax, y); Or1 y]
2018-03-11 15:41:35 +03:00
| "&&" ->
[Dec x; (*!!!*)
Mov (x, eax);
2018-03-11 15:41:35 +03:00
Binop (op, x, eax);
Mov (L 0, eax);
Set ("ne", "%al");
Dec y; (*!!!*)
2018-03-11 15:41:35 +03:00
Mov (y, edx);
Binop (op, y, edx);
Mov (L 0, edx);
Set ("ne", "%dl");
2019-09-10 00:46:10 +03:00
2018-03-11 15:41:35 +03:00
Binop (op, edx, eax);
Set ("ne", "%al");
Sal1 eax;
Or1 eax;
2018-03-11 15:41:35 +03:00
Mov (eax, y)
2019-10-16 21:07:27 +03:00
] @ env#reload_closure
2018-03-11 15:41:35 +03:00
| "!!" ->
[Mov (y, eax);
Sar1 eax;
Sar1 x; (*!!!*)
2018-03-11 15:41:35 +03:00
Binop (op, x, eax);
Mov (L 0, eax);
Set ("ne", "%al");
Sal1 eax;
Or1 eax;
2018-03-11 15:41:35 +03:00
Mov (eax, y)
2019-09-10 00:46:10 +03:00
]
| "+" ->
2019-09-10 00:46:10 +03:00
if on_stack x && on_stack y
then [Mov (x, eax); Dec eax; Binop ("+", eax, y)]
else [Binop (op, x, y); Dec y]
| "-" ->
2019-09-10 00:46:10 +03:00
if on_stack x && on_stack y
then [Mov (x, eax); Binop (op, eax, y); Or1 y]
else [Binop (op, x, y); Or1 y]
2018-03-11 15:41:35 +03:00
)
2020-09-10 09:07:38 +03:00
| LABEL s
| FLABEL s
2020-09-10 09:07:38 +03:00
| SLABEL s -> env, [Label s]
2018-05-16 09:24:40 +03:00
| JMP l -> (env#set_stack l)#set_barrier, [Jmp l]
2019-09-10 00:46:10 +03:00
| CJMP (s, l) ->
let x, env = env#pop in
env#set_stack l, [Sar1 x; (*!!!*) Binop ("cmp", L 0, x); CJmp (s, l)]
2019-09-10 00:46:10 +03:00
2020-09-10 09:07:38 +03:00
| BEGIN (f, nargs, nlocals, closure, args, scopes) ->
let rec stabs_scope scope =
let names =
List.map
(fun (name, index) ->
Meta (Printf.sprintf "\t.stabs \"%s:1\",128,0,0,-%d" name (stack_offset index))
)
scope.names
in
names @
(if names = [] then [] else [Meta (Printf.sprintf "\t.stabn 192,0,0,%s-%s" scope.blab f)]) @
(List.flatten @@ List.map stabs_scope scope.subs) @
(if names = [] then [] else [Meta (Printf.sprintf "\t.stabn 224,0,0,%s-%s" scope.elab f)])
in
let name =
if f.[0] = 'L' then String.sub f 1 (String.length f - 1) else f
in
2019-10-14 19:44:33 +03:00
env#assert_empty_stack;
2019-10-16 01:13:52 +03:00
let has_closure = closure <> [] in
let env = env#enter f nargs nlocals has_closure in
env, [Meta (Printf.sprintf "\t.type %s, @function" name)] @
(if f = "main"
then []
else
[Meta (Printf.sprintf "\t.stabs \"%s:F1\",36,0,0,%s" name f)] @
2020-09-10 09:07:38 +03:00
(List.mapi (fun i a -> Meta (Printf.sprintf "\t.stabs \"%s:p1\",160,0,0,%d" a ((i*4) + 8))) args) @
(List.flatten @@ List.map stabs_scope scopes)
)
@
[Meta "\t.cfi_startproc"] @
(if has_closure then [Push edx] else []) @
(if f = cmd#topname
then
[Mov (M "_init", eax);
Binop ("test", eax, eax);
CJmp ("z", "_continue");
Ret;
Label ("_ERROR");
Call "Lbinoperror";
Ret;
Label ("_ERROR2");
Call "Lbinoperror2";
Ret;
Label "_continue";
Mov (L 1, M "_init");
]
else []
) @
2019-10-16 21:07:27 +03:00
[Push ebp;
Meta ("\t.cfi_def_cfa_offset\t" ^ if has_closure then "12" else "8");
Meta ("\t.cfi_offset 5, -" ^ if has_closure then "12" else "8");
2019-10-16 21:07:27 +03:00
Mov (esp, ebp);
2020-09-04 23:45:57 +03:00
Meta "\t.cfi_def_cfa_register\t5";
2019-10-16 21:07:27 +03:00
Binop ("-", M ("$" ^ env#lsize), esp);
Mov (esp, edi);
Mov (M "$filler", esi);
Mov (M ("$" ^ (env#allocated_size)), ecx);
Repmovsl
] @
(if f = "main"
2020-08-02 23:56:21 +03:00
then [Call "__gc_init"; Push (I (12, ebp)); Push (I (8, ebp)); Call "set_args"; Binop ("+", L 8, esp)]
else []
) @
(if f = cmd#topname
then List.map (fun i -> Call ("init" ^ i)) (List.filter (fun i -> i <> "Std") imports)
else []
)
2019-10-16 01:13:52 +03:00
2019-09-10 00:46:10 +03:00
| END ->
let x, env = env#pop in
2019-12-24 03:59:05 +03:00
env#assert_empty_stack;
2019-10-16 21:07:27 +03:00
let name = env#fname in
env#leave, [
Mov (x, eax); (*!!*)
2019-10-16 21:07:27 +03:00
Label env#epilogue;
Mov (ebp, esp);
2020-09-04 23:45:57 +03:00
Pop ebp;
2019-10-16 21:07:27 +03:00
] @
env#rest_closure @
(if name = "main" then [Binop ("^", eax, eax)] else []) @
2020-09-04 23:45:57 +03:00
[Meta "\t.cfi_restore\t5";
Meta "\t.cfi_def_cfa\t4, 4";
Ret;
Meta "\t.cfi_endproc";
2019-10-16 21:07:27 +03:00
Meta (Printf.sprintf "\t.set\t%s,\t%d" env#lsize (env#allocated * word_size));
Meta (Printf.sprintf "\t.set\t%s,\t%d" env#allocated_size env#allocated);
Meta (Printf.sprintf "\t.size %s, .-%s" name name);
2019-10-16 21:07:27 +03:00
]
2019-09-10 00:46:10 +03:00
| RET ->
let x = env#peek in
2019-04-10 22:15:08 +03:00
env, [Mov (x, eax); Jmp env#epilogue]
2021-10-03 17:10:21 +03:00
| ELEM -> call env ".elem" 2 false
2020-03-23 00:49:20 +03:00
| CALL (f, n, tail) -> call env f n tail
2019-10-15 01:54:57 +03:00
2020-03-23 00:49:20 +03:00
| CALLC (n, tail) -> callc env n tail
2019-10-14 19:44:33 +03:00
2018-05-11 02:40:52 +03:00
| SEXP (t, n) ->
2018-05-16 09:24:40 +03:00
let s, env = env#allocate in
2020-03-23 00:49:20 +03:00
let env, code = call env ".sexp" (n+1) false in
2020-09-01 20:31:34 +03:00
env, [Mov (L (box (env#hash t)), s)] @ code
2018-05-16 09:24:40 +03:00
2018-11-05 18:21:41 +03:00
| DROP ->
snd env#pop, []
2019-09-10 00:46:10 +03:00
2018-05-16 09:24:40 +03:00
| DUP ->
let x = env#peek in
let s, env = env#allocate in
env, mov x s
2019-09-10 00:46:10 +03:00
2018-05-16 09:24:40 +03:00
| SWAP ->
let x, y = env#peek2 in
env, [Push x; Push y; Pop x; Pop y]
2019-09-10 00:46:10 +03:00
2018-11-05 20:17:11 +03:00
| TAG (t, n) ->
let s1, env = env#allocate in
let s2, env = env#allocate in
2020-03-23 00:49:20 +03:00
let env, code = call env ".tag" 3 false in
2020-02-13 18:56:27 +03:00
env, [Mov (L (box (env#hash t)), s1); Mov (L (box n), s2)] @ code
2018-11-06 00:21:38 +03:00
| ARRAY n ->
let s, env = env#allocate in
2020-03-23 00:49:20 +03:00
let env, code = call env ".array_patt" 2 false in
2020-02-13 18:56:27 +03:00
env, [Mov (L (box n), s)] @ code
2018-11-06 00:21:38 +03:00
2020-03-23 00:49:20 +03:00
| PATT StrCmp -> call env ".string_patt" 2 false
2018-11-06 00:21:38 +03:00
| PATT patt ->
2019-10-15 01:54:57 +03:00
call env
2018-11-06 00:21:38 +03:00
(match patt with
| Boxed -> ".boxed_patt"
| UnBoxed -> ".unboxed_patt"
| Array -> ".array_tag_patt"
| String -> ".string_tag_patt"
| Sexp -> ".sexp_tag_patt"
2019-10-16 01:13:52 +03:00
| Closure -> ".closure_tag_patt"
2020-03-23 00:49:20 +03:00
) 1 false
2020-09-04 00:25:07 +03:00
| LINE (line) ->
env#gen_line line
2020-09-04 00:25:07 +03:00
2019-12-26 00:17:34 +03:00
| FAIL ((line, col), value) ->
let v, env = if value then env#peek, env else env#pop in
2019-12-24 03:59:05 +03:00
let s, env = env#string cmd#get_infile in
2021-01-12 01:14:18 +03:00
env, [Push (L (box col)); Push (L (box line)); Push (M ("$" ^ s)); Push v; Call "Bmatch_failure"; Binop ("+", L (4 * word_size), esp)]
2019-12-24 03:59:05 +03:00
2019-10-14 19:44:33 +03:00
| i ->
invalid_arg (Printf.sprintf "invalid SM insn: %s\n" (GT.show(insn) i))
2018-03-05 00:54:50 +03:00
in
let env'', code'' = compile' env' scode' in
2018-05-16 09:24:40 +03:00
env'', [Meta (Printf.sprintf "# %s / % s" (GT.show(SM.insn) instr) stack)] @ code' @ code''
2018-03-05 00:54:50 +03:00
in
compile' env code
2019-09-10 00:46:10 +03:00
(* A set of strings *)
module S = Set.Make (String)
2018-03-05 00:54:50 +03:00
2018-04-30 17:18:41 +03:00
(* A map indexed by strings *)
2019-09-10 00:46:10 +03:00
module M = Map.Make (String)
2018-04-30 17:18:41 +03:00
2018-03-05 00:54:50 +03:00
(* Environment implementation *)
2019-10-16 01:13:52 +03:00
class env prg =
2020-08-25 16:09:43 +03:00
let chars = "_abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789'" in
2018-05-16 09:24:40 +03:00
let make_assoc l i = List.combine l (List.init (List.length l) (fun x -> x + i)) in
let rec assoc x = function [] -> raise Not_found | l :: ls -> try List.assoc x l with Not_found -> assoc x ls in
2018-03-05 00:54:50 +03:00
object (self)
2019-10-16 01:13:52 +03:00
inherit SM.indexer prg
val globals = S.empty (* a set of global variables *)
val stringm = M.empty (* a string map *)
val scount = 0 (* string count *)
val stack_slots = 0 (* maximal number of stack positions *)
2020-03-23 00:49:20 +03:00
2019-10-16 01:13:52 +03:00
val static_size = 0 (* static data size *)
val stack = [] (* symbolic stack *)
2020-03-23 00:49:20 +03:00
val nargs = 0 (* number of function arguments *)
2019-10-16 01:13:52 +03:00
val locals = [] (* function local variables *)
val fname = "" (* function name *)
val stackmap = M.empty (* labels to stack map *)
val barrier = false (* barrier condition *)
2018-12-12 12:42:38 +03:00
val max_locals_size = 0
2019-10-16 01:13:52 +03:00
val has_closure = false
2019-11-24 02:30:32 +03:00
val publics = S.empty
val externs = S.empty
val nlabels = 0
val first_line = true
2019-11-24 02:30:32 +03:00
method publics = S.elements publics
method register_public name = {< publics = S.add name publics >}
method register_extern name = {< externs = S.add name externs >}
2018-12-12 12:42:38 +03:00
method max_locals_size = max_locals_size
2019-10-16 21:07:27 +03:00
method has_closure = has_closure
method save_closure =
if has_closure then [Push edx] else []
method rest_closure =
if has_closure then [Pop edx] else []
2019-09-10 00:46:10 +03:00
2019-10-16 21:07:27 +03:00
method reload_closure =
2019-12-29 01:12:40 +03:00
if has_closure then [Mov (C (*S 0*), edx)] else []
2019-10-16 21:07:27 +03:00
method fname = fname
method leave =
2018-12-12 12:42:38 +03:00
if stack_slots > max_locals_size
then {< max_locals_size = stack_slots >}
else self
2019-09-10 00:46:10 +03:00
2018-05-16 09:24:40 +03:00
method show_stack =
GT.show(list) (GT.show(opnd)) stack
2019-09-10 00:46:10 +03:00
2018-05-16 09:24:40 +03:00
method print_locals =
Printf.printf "LOCALS: size = %d\n" static_size;
List.iter
(fun l ->
Printf.printf "(";
List.iter (fun (a, i) -> Printf.printf "%s=%d " a i) l;
Printf.printf ")\n"
) locals;
Printf.printf "END LOCALS\n"
(* Assert empty stack *)
method assert_empty_stack = assert (stack = [])
2019-09-10 00:46:10 +03:00
2018-05-16 09:24:40 +03:00
(* check barrier condition *)
method is_barrier = barrier
(* set barrier *)
method set_barrier = {< barrier = true >}
2018-05-16 09:24:40 +03:00
(* drop barrier *)
method drop_barrier = {< barrier = false >}
2019-09-10 00:46:10 +03:00
(* drop stack *)
method drop_stack = {< stack = [] >}
2018-05-16 09:24:40 +03:00
(* associates a stack to a label *)
method set_stack l = (*Printf.printf "Setting stack for %s\n" l;*)
{< stackmap = M.add l stack stackmap >}
2019-09-10 00:46:10 +03:00
2018-05-16 09:24:40 +03:00
(* retrieves a stack for a label *)
2018-05-16 16:50:36 +03:00
method retrieve_stack l = (*Printf.printf "Retrieving stack for %s\n" l;*)
2018-05-16 09:24:40 +03:00
try {< stack = M.find l stackmap >} with Not_found -> self
2019-09-10 00:46:10 +03:00
(* checks if there is a stack for a label *)
method has_stack l = (*Printf.printf "Retrieving stack for %s\n" l;*)
M.mem l stackmap
2018-03-05 00:54:50 +03:00
(* gets a name for a global variable *)
2018-04-11 00:47:46 +03:00
method loc x =
2019-10-14 19:44:33 +03:00
match x with
2019-11-29 23:56:03 +03:00
| Value.Global name -> M ("global_" ^ name)
2019-10-16 01:13:52 +03:00
| Value.Fun name -> M ("$" ^ name)
2019-10-16 21:07:27 +03:00
| Value.Local i -> S i
| Value.Arg i -> S (- (i + if has_closure then 2 else 1))
| Value.Access i -> I (word_size * (i+1), edx)
2018-03-05 00:54:50 +03:00
(* allocates a fresh position on a symbolic stack *)
2019-09-10 00:46:10 +03:00
method allocate =
2018-03-05 00:54:50 +03:00
let x, n =
2018-11-21 14:23:35 +03:00
let rec allocate' = function
| [] -> ebx , 0
| (S n)::_ -> S (n+1) , n+2
| (R n)::_ when n < num_of_regs -> R (n+1) , stack_slots
| _ -> S static_size, static_size+1
in
allocate' stack
2018-03-05 00:54:50 +03:00
in
x, {< stack_slots = max n stack_slots; stack = x::stack >}
(* pushes an operand to the symbolic stack *)
method push y = {< stack = y::stack >}
(* pops one operand from the symbolic stack *)
2018-04-11 00:47:46 +03:00
method pop = let x::stack' = stack in x, {< stack = stack' >}
2018-03-05 00:54:50 +03:00
(* pops two operands from the symbolic stack *)
method pop2 = let x::y::stack' = stack in x, y, {< stack = stack' >}
2018-05-16 09:24:40 +03:00
(* peeks the top of the stack (the stack does not change) *)
method peek = List.hd stack
(* peeks two topmost values from the stack (the stack itself does not change) *)
method peek2 = let x::y::_ = stack in x, y
(* tag hash: gets a hash for a string tag *)
method hash tag =
2018-10-23 14:29:30 +03:00
let h = Pervasives.ref 0 in
2018-05-16 09:24:40 +03:00
for i = 0 to min (String.length tag - 1) 4 do
h := (!h lsl 6) lor (String.index chars tag.[i])
done;
2019-09-10 00:46:10 +03:00
!h
2018-11-21 14:23:35 +03:00
(* registers a variable in the environment *)
method variable x =
2019-10-14 19:44:33 +03:00
match x with
2019-11-29 23:56:03 +03:00
| Value.Global name -> {< globals = S.add ("global_" ^ name) globals >}
2019-10-14 19:44:33 +03:00
| _ -> self
2018-03-05 00:54:50 +03:00
2018-04-30 17:18:41 +03:00
(* registers a string constant *)
method string x =
2019-12-31 00:59:28 +03:00
let escape x =
let n = String.length x in
let buf = Buffer.create (n*2) in
let rec iterate i =
if i < n
then (
2020-01-03 01:38:49 +03:00
(match x.[i] with
| '"' -> Buffer.add_string buf "\\\""
| '\n' -> Buffer.add_string buf "\n"
| '\t' -> Buffer.add_string buf "\t"
| c -> Buffer.add_char buf c
);
2019-12-31 00:59:28 +03:00
iterate (i+1)
)
in
iterate 0;
Buffer.contents buf
in
let x = escape x in
2018-04-30 17:18:41 +03:00
try M.find x stringm, self
with Not_found ->
let y = Printf.sprintf "string_%d" scount in
let m = M.add x y stringm in
y, {< scount = scount + 1; stringm = m>}
2019-09-10 00:46:10 +03:00
2020-03-23 00:49:20 +03:00
(* gets number of arguments in the current function *)
method nargs = nargs
2019-09-10 00:46:10 +03:00
(* gets all global variables *)
2019-11-24 02:30:32 +03:00
method globals = S.elements (S.diff globals externs)
2018-03-05 00:54:50 +03:00
2019-09-10 00:46:10 +03:00
(* gets all string definitions *)
2018-04-30 17:18:41 +03:00
method strings = M.bindings stringm
2018-04-11 00:47:46 +03:00
(* gets a number of stack positions allocated *)
2018-12-12 12:42:38 +03:00
method allocated = stack_slots
2019-10-15 01:54:57 +03:00
2018-12-12 12:42:38 +03:00
method allocated_size = Printf.sprintf "LS%s_SIZE" fname
2019-10-14 19:44:33 +03:00
2018-04-11 00:47:46 +03:00
(* enters a function *)
2020-03-23 00:49:20 +03:00
method enter f nargs nlocals has_closure =
{< nargs = nargs; static_size = nlocals; stack_slots = nlocals; stack = []; fname = f; has_closure = has_closure; first_line = true >}
2018-03-05 00:54:50 +03:00
2018-04-11 00:47:46 +03:00
(* returns a label for the epilogue *)
method epilogue = Printf.sprintf "L%s_epilogue" fname
2019-09-10 00:46:10 +03:00
2018-04-11 00:47:46 +03:00
(* returns a name for local size meta-symbol *)
method lsize = Printf.sprintf "L%s_SIZE" fname
2019-10-14 19:44:33 +03:00
2018-04-11 00:47:46 +03:00
(* returns a list of live registers *)
2018-04-30 17:18:41 +03:00
method live_registers depth =
let rec inner d acc = function
| [] -> acc
| (R _ as r)::tl -> inner (d+1) (if d >= depth then (r::acc) else acc) tl
| _::tl -> inner (d+1) acc tl
in
inner 0 [] stack
2019-09-10 00:46:10 +03:00
(* generate a line number information for current function *)
method gen_line line =
let lab = Printf.sprintf ".L%d" nlabels in
{< nlabels = nlabels + 1; first_line = false >},
if fname = "main"
then
[Meta (Printf.sprintf "\t.stabn 68,0,%d,%s" line lab); Label lab]
else
(if first_line then [Meta (Printf.sprintf "\t.stabn 68,0,%d,0" line)] else []) @
[Meta (Printf.sprintf "\t.stabn 68,0,%d,%s-%s" line lab fname); Label lab]
2018-04-11 00:47:46 +03:00
end
2019-09-10 00:46:10 +03:00
2018-03-05 01:01:36 +03:00
(* Generates an assembler text for a program: first compiles the program into
the stack code, then generates x86 assember code, then prints the assembler file
2018-03-05 00:54:50 +03:00
*)
let genasm cmd prog =
let sm = SM.compile cmd prog in
let env, code = compile cmd (new env sm) (fst (fst prog)) sm in
2019-11-24 02:30:32 +03:00
let globals =
List.map (fun s -> Meta (Printf.sprintf "\t.globl\t%s" s)) env#publics
2019-11-24 02:30:32 +03:00
in
2020-02-07 19:41:35 +03:00
let data = [Meta "\t.data"] @
(List.map (fun (s, v) -> Meta (Printf.sprintf "%s:\t.string\t\"%s\"" v s)) env#strings) @
[Meta "_init:\t.int 0";
Meta "\t.section custom_data,\"aw\",@progbits";
Meta (Printf.sprintf "filler:\t.fill\t%d, 4, 1" env#max_locals_size)] @
(List.concat @@
List.map
(fun s -> [Meta (Printf.sprintf "\t.stabs \"%s:S1\",40,0,0,%s" (String.sub s (String.length "global_") (String.length s - String.length "global_")) s);
Meta (Printf.sprintf "%s:\t.int\t1" s)])
env#globals
)
2018-12-12 12:42:38 +03:00
in
2018-03-05 00:54:50 +03:00
let asm = Buffer.create 1024 in
List.iter
(fun i -> Buffer.add_string asm (Printf.sprintf "%s\n" @@ show i))
([Meta (Printf.sprintf "\t.file \"%s\"" cmd#get_absolute_infile);
Meta (Printf.sprintf "\t.stabs \"%s\",100,0,0,.Ltext" cmd#get_absolute_infile)] @
globals @
data @
[Meta "\t.text"; Label ".Ltext"; Meta "\t.stabs \"data:t1=r1;0;4294967295;\",128,0,0,0"] @
code);
2018-03-05 00:54:50 +03:00
Buffer.contents asm
2020-02-13 18:56:27 +03:00
let get_std_path () =
2020-02-16 00:21:15 +03:00
match Sys.getenv_opt "LAMA" with
2020-02-13 18:56:27 +03:00
| Some s -> s
2020-02-16 02:02:54 +03:00
| None -> Stdpath.path
2020-02-13 18:56:27 +03:00
2018-03-05 00:54:50 +03:00
(* Builds a program: generates the assembler file and compiles it with the gcc toolchain *)
2019-11-24 02:30:32 +03:00
let build cmd prog =
let find_objects imports paths =
let module S = Set.Make (String) in
let rec iterate acc s = function
| [] -> acc
| import::imports ->
if S.mem import s
then iterate acc s imports
else
let path, intfs = Interface.find import paths in
iterate
((Filename.concat path (import ^ ".o")) :: acc)
(S.add import s)
((List.map (function `Import name -> name | _ -> invalid_arg "must not happen") @@
List.filter (function `Import _ -> true | _ -> false) intfs) @
imports)
in
2019-12-18 18:44:01 +03:00
iterate [] (S.add "Std" S.empty) imports
in
2019-12-12 17:42:45 +03:00
cmd#dump_file "s" (genasm cmd prog);
cmd#dump_file "i" (Interface.gen prog);
2020-02-13 18:56:27 +03:00
let inc = get_std_path () in
2019-11-24 02:30:32 +03:00
match cmd#get_mode with
| `Default ->
2019-11-29 23:56:03 +03:00
let objs = find_objects (fst @@ fst prog) cmd#get_include_paths in
let buf = Buffer.create 255 in
List.iter (fun o -> Buffer.add_string buf o; Buffer.add_string buf " ") objs;
2020-09-08 01:50:16 +03:00
let gcc_cmdline = Printf.sprintf "gcc %s -m32 %s %s.s %s %s/runtime.a" cmd#get_debug cmd#get_output_option cmd#basename (Buffer.contents buf) inc in
Sys.command gcc_cmdline
2019-11-24 02:30:32 +03:00
| `Compile ->
2020-09-08 01:50:16 +03:00
Sys.command (Printf.sprintf "gcc %s -m32 -c %s.s" cmd#get_debug cmd#basename)
2019-11-24 02:30:32 +03:00
| _ -> invalid_arg "must not happen"