lama_byterun/src/X86.ml

1374 lines
46 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
2024-02-15 10:43:47 +01:00
open SM
2023-09-04 21:43:28 +02:00
2018-03-05 00:54:50 +03:00
(* X86 codegeneration interface *)
2024-01-31 17:48:25 +01:00
module Register : sig
type t
val from_names : l8:string -> l64:string -> t
val from_number : int -> t
val of_8bit : t -> t
val of_64bit : t -> t
val show : t -> string
end = struct
(* Other sizes skipped as they are not used *)
type register_desc = { name8 : string; name64 : string }
type t = string * register_desc
let from_names ~l8 ~l64 = (l64, { name8 = l8; name64 = l64 })
let from_number n =
let name64 = Printf.sprintf "%%r%s" (string_of_int n) in
let name8 = Printf.sprintf "%%r%sb" (string_of_int n) in
(name64, { name8; name64 })
let of_8bit (_, { name8; name64 }) = (name8, { name8; name64 })
let of_64bit (_, { name8; name64 }) = (name64, { name8; name64 })
let show (name, _) = name
end
2024-01-30 18:16:44 +01:00
module Registers : sig
2024-01-31 17:48:25 +01:00
val rax : Register.t
val rdi : Register.t
val rsi : Register.t
val rdx : Register.t
val rcx : Register.t
val rbp : Register.t
val rsp : Register.t
val r8 : Register.t
val r9 : Register.t
val r10 : Register.t
val r11 : Register.t
val r12 : Register.t
val r13 : Register.t
val r14 : Register.t
val r15 : Register.t
val argument_registers : Register.t array
2024-01-30 18:16:44 +01:00
(** All of argument registers are caller-saved *)
2024-01-31 17:48:25 +01:00
val extra_caller_saved_registers : Register.t array
2024-01-30 18:16:44 +01:00
(** Caller saved registers that are not used for arguments *)
end = struct
(* Caller-saved special registers *)
2024-01-31 17:48:25 +01:00
let rax = Register.from_names ~l8:"%al" ~l64:"%rax"
2024-01-30 18:16:44 +01:00
(* Caller-saved special and argument registers *)
2024-01-31 17:48:25 +01:00
let rdx = Register.from_names ~l8:"%dl" ~l64:"%rdx"
2024-01-30 18:16:44 +01:00
(* Caller-saved argument registers *)
2024-01-31 17:48:25 +01:00
let rdi = Register.from_names ~l8:"%dil" ~l64:"%rdi"
let rsi = Register.from_names ~l8:"%sil" ~l64:"%rsi"
let rcx = Register.from_names ~l8:"%cl" ~l64:"%rcx"
let r8 = Register.from_number 8
let r9 = Register.from_number 9
2024-01-30 18:16:44 +01:00
(* Extra caller-saved registers *)
2024-01-31 17:48:25 +01:00
let r10 = Register.from_number 10
let r11 = Register.from_number 11
2024-01-30 18:16:44 +01:00
(* Callee-saved special registers *)
2024-01-31 17:48:25 +01:00
let rbp = Register.from_names ~l8:"%bpl" ~l64:"%rbp"
let rsp = Register.from_names ~l8:"%spl" ~l64:"%rsp"
(* r12-15 registes are calee-saved in X86_64
But we are using them as caller-save for simplicity
This disallows calling Lama code from C
While does not affects C calls from Lama *)
let r12 = Register.from_number 12
let r13 = Register.from_number 13
let r14 = Register.from_number 14
let r15 = Register.from_number 15
2024-01-30 18:16:44 +01:00
let argument_registers = [| rdi; rsi; rdx; rcx; r8; r9 |]
2024-01-31 17:48:25 +01:00
let extra_caller_saved_registers = [| r10; r11; r12; r13; r14 |]
2024-01-30 18:16:44 +01:00
end
2018-03-05 00:54:50 +03:00
(* We need to know the word size to calculate offsets correctly *)
2024-01-26 18:45:03 +01:00
let word_size = 8
2018-03-05 00:54:50 +03:00
2024-03-14 09:01:50 +01:00
type externality = I (**Internal*) | E (**External*)
type data_kind = F (**Function*) | D (**Data*)
type addressed = A (**Address*) | V (**Value*)
2018-03-05 00:54:50 +03:00
(* We need to distinguish the following operand types: *)
2023-09-04 21:43:28 +02:00
type opnd =
2024-01-31 17:48:25 +01:00
| R of Register.t (* hard register *)
2023-09-04 21:43:28 +02:00
| S of int (* a position on the hardware stack *)
2024-03-14 09:01:50 +01:00
| M of
(* a named memory location *)
data_kind
* externality
* addressed
* string
| C of string (* a named constant *)
2023-09-04 21:43:28 +02:00
| L of int (* an immediate operand *)
| I of int * opnd (* an indirect operand with offset *)
2024-01-31 17:48:25 +01:00
let as_register opnd =
match opnd with R r -> r | _ -> failwith "as_register: not a register"
2018-03-05 00:54:50 +03:00
2024-01-30 18:16:44 +01:00
type argument_location = Register of opnd | Stack
2024-01-31 17:48:25 +01:00
let rec show_opnd = function
| R r -> Printf.sprintf "R %s" (Register.show r)
| S i -> Printf.sprintf "S %d" i
| L i -> Printf.sprintf "L %d" i
| I (i, o) -> Printf.sprintf "I %d %s" i (show_opnd o)
2024-03-14 09:01:50 +01:00
| C s -> Printf.sprintf "C %s" s
| M (e, d, a, s) ->
Printf.sprintf "M %s %s %s %s"
(match e with F -> "Function" | D -> "Data")
(match d with I -> "Internal" | E -> "External")
(match a with A -> "Address" | V -> "Value")
s
2024-01-31 17:48:25 +01:00
(* We need to know the word size to calculate offsets correctly *)
2019-09-10 00:46:10 +03:00
(* For convenience we define the following synonyms for the registers: *)
2024-01-30 18:16:44 +01:00
let rax = R Registers.rax
let rdx = R Registers.rdx
let rbp = R Registers.rbp
let rsp = R Registers.rsp
let rdi = R Registers.rdi
let rsi = R Registers.rsi
let rcx = R Registers.rcx
let r8 = R Registers.r8
let r9 = R Registers.r9
let r10 = R Registers.r10
let r11 = R Registers.r11
let r12 = R Registers.r12
let r13 = R Registers.r13
let r14 = R Registers.r14
let r15 = R Registers.r15
2018-03-05 00:54:50 +03:00
2024-03-14 09:01:50 +01:00
(* Value that could be used to fill unused stack locations *)
let filler = M (D, I, A, "filler")
2018-03-05 00:54:50 +03:00
(* Now x86 instruction (we do not need all of them): *)
type instr =
2023-09-04 21:43:28 +02:00
(* 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
2024-03-14 09:01:50 +01:00
(* makes a binary operation; note, the first operand
designates x86 operator, not the source language one *)
2023-09-04 21:43:28 +02:00
| Binop of string * opnd * opnd
(* x86 integer division, see instruction set reference *)
| IDiv of opnd
(* see instruction set reference *)
| Cltd
2024-01-31 17:48:25 +01:00
(* sets a value from flags; the first operand is the
suffix, which determines the value being set, the
the second --- (sub)register name *)
| Set of string * Register.t
2023-09-04 21:43:28 +02:00
(* 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
(* call a function by indirect address *)
| CallI of opnd
(* returns from a function *)
| Ret
(* a label in the code *)
| Label of string
(* a conditional jump *)
| CJmp of string * string
(* a non-conditional jump by a name *)
2023-09-04 21:43:28 +02:00
| Jmp of string
(* a non-conditional jump by indirect address *)
| JmpI of opnd
2023-09-04 21:43:28 +02:00
(* directive *)
| Meta of string
(* 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
| Repmovsl
2018-03-05 00:54:50 +03:00
(* Instruction printer *)
2020-09-10 09:07:38 +03:00
let stack_offset i =
2024-01-30 18:16:44 +01:00
if i >= 0 then (i + 1) * word_size else (-i + 1) * word_size
2023-09-04 21:43:28 +02:00
2018-03-05 00:54:50 +03:00
let show instr =
2020-09-10 09:07:38 +03:00
let rec opnd = function
2024-01-31 17:48:25 +01:00
| R r -> Register.show r
2023-09-04 21:43:28 +02:00
| S i ->
2024-01-26 18:45:03 +01:00
if i >= 0 then Printf.sprintf "-%d(%%rbp)" (stack_offset i)
else Printf.sprintf "%d(%%rbp)" (stack_offset i)
2024-03-14 09:01:50 +01:00
| M (_, I, _, s) -> Printf.sprintf "%s(%%rip)" s
| M (F, E, _, s) -> Printf.sprintf "%s@plt(%%rip)" s
| M (D, E, _, s) -> Printf.sprintf "%s@GOTPCREL(%%rip)" s
| C s -> Printf.sprintf "$%s" s
2023-09-04 21:43:28 +02:00
| L i -> Printf.sprintf "$%d" i
| I (0, x) -> Printf.sprintf "(%s)" (opnd x)
| I (n, x) -> Printf.sprintf "%d(%s)" n (opnd x)
2020-09-10 09:07:38 +03:00
in
2018-03-05 00:54:50 +03:00
let binop = function
| "+" -> "addq"
| "-" -> "subq"
| "*" -> "imulq"
| "&&" -> "andq"
| "!!" -> "orq"
| "^" -> "xorq"
| "cmp" -> "cmpq"
2023-09-04 21:43:28 +02:00
| "test" -> "test"
| _ -> failwith "unknown binary operator"
2018-03-05 00:54:50 +03:00
in
match instr with
2024-01-26 18:45:03 +01:00
| Cltd -> "\tcqo"
2024-01-31 17:48:25 +01:00
| Set (suf, r) ->
Printf.sprintf "\tset%s\t%s" suf (Register.show (Register.of_8bit r))
2024-01-26 18:45:03 +01:00
| IDiv s1 -> Printf.sprintf "\tidivq\t%s" (opnd s1)
2023-09-04 21:43:28 +02:00
| Binop (op, s1, s2) ->
Printf.sprintf "\t%s\t%s,\t%s" (binop op) (opnd s1) (opnd s2)
2024-03-14 09:01:50 +01:00
| Mov ((M (_, _, A, _) as x), y) | Lea (x, y) ->
(* TODO: It looks like a bad design.
Maybe we should introduce eopnd with the boolean if we referenceing an address but not a value *)
Printf.sprintf "\tleaq\t%s,\t%s" (opnd x) (opnd y)
2024-01-26 18:45:03 +01:00
| Mov (s1, s2) -> Printf.sprintf "\tmovq\t%s,\t%s" (opnd s1) (opnd s2)
| Push s -> Printf.sprintf "\tpushq\t%s" (opnd s)
| Pop s -> Printf.sprintf "\tpopq\t%s" (opnd s)
2023-09-04 21:43:28 +02:00
| Ret -> "\tret"
| Call p -> Printf.sprintf "\tcall\t%s" p
2023-09-04 21:43:28 +02:00
| CallI o -> Printf.sprintf "\tcall\t*(%s)" (opnd o)
| Label l -> Printf.sprintf "%s:\n" l
| Jmp l -> Printf.sprintf "\tjmp\t%s" l
| JmpI o -> Printf.sprintf "\tjmp\t*(%s)" (opnd o)
2023-09-04 21:43:28 +02:00
| CJmp (s, l) -> Printf.sprintf "\tj%s\t%s" s l
| Meta s -> Printf.sprintf "%s\n" s
| Dec s -> Printf.sprintf "\tdecq\t%s" (opnd s)
| Or1 s -> Printf.sprintf "\torq\t$0x0001,\t%s" (opnd s)
| Sal1 s -> Printf.sprintf "\tsalq\t%s" (opnd s)
| Sar1 s -> Printf.sprintf "\tsarq\t%s" (opnd s)
2024-01-26 18:45:03 +01:00
| Repmovsl -> Printf.sprintf "\trep movsq\t"
2019-09-10 00:46:10 +03:00
2024-03-14 09:01:50 +01:00
let in_memory = function M _ | S _ | I _ -> true | C _ | R _ | L _ -> false
2024-02-05 18:34:40 +01:00
let big_numeric_literal = function L num -> num > 0xFFFFFFFF | _ -> false
let mov x s =
if x = s then []
2024-02-05 18:34:40 +01:00
else if (in_memory x && in_memory s) || big_numeric_literal x then
[ Mov (x, rax); Mov (rax, s) ]
else [ Mov (x, s) ]
let box n = (n lsl 1) lor 1
2024-01-30 18:16:44 +01:00
(*
Compile binary operation
2018-03-05 00:54:50 +03:00
2024-01-30 18:16:44 +01:00
compile_binop : env -> string -> env * instr list
*)
let compile_binop env op =
2018-03-05 00:54:50 +03:00
let suffix = function
2023-09-04 21:43:28 +02:00
| "<" -> "l"
| "<=" -> "le"
| "==" -> "e"
| "!=" -> "ne"
| ">=" -> "ge"
| ">" -> "g"
| _ -> failwith "unknown operator"
2020-02-13 18:56:27 +03:00
in
2024-01-31 17:48:25 +01:00
let x, y = env#peek2 in
2024-01-30 18:16:44 +01:00
let without_extra op =
2024-01-31 17:48:25 +01:00
let _x, env = env#pop in
(env, op ())
2024-01-30 18:16:44 +01:00
in
let with_rdx op =
if not env#rdx_in_use then
2024-01-31 17:48:25 +01:00
let _x, env = env#pop in
(env, op rdx)
2024-01-30 18:16:44 +01:00
else
let extra, env = env#allocate in
2024-01-31 17:48:25 +01:00
let _extra, env = env#pop in
let _x, env = env#pop in
let code = op rdx in
2024-01-30 18:16:44 +01:00
(env, [ Mov (rdx, extra) ] @ code @ [ Mov (extra, rdx) ])
in
let with_extra op =
let extra, env = env#allocate in
2024-01-31 17:48:25 +01:00
let _extra, env = env#pop in
let _x, env = env#pop in
if in_memory extra then
2024-02-02 15:26:47 +01:00
(env, [ Mov (rdx, extra) ] @ op rdx @ [ Mov (extra, rdx) ])
2024-01-31 17:48:25 +01:00
else (env, op extra)
2024-01-30 18:16:44 +01:00
in
match op with
| "/" ->
2024-01-31 17:48:25 +01:00
with_rdx (fun rdx ->
2024-01-30 18:16:44 +01:00
[
Mov (y, rax);
Sar1 rax;
Binop ("^", rdx, rdx);
Cltd;
Sar1 x;
IDiv x;
Sal1 rax;
Or1 rax;
Mov (rax, y);
])
| "%" ->
2024-01-31 17:48:25 +01:00
with_rdx (fun rdx ->
2024-01-30 18:16:44 +01:00
[
Mov (y, rax);
Sar1 rax;
Cltd;
Sar1 x;
IDiv x;
Sal1 rdx;
Or1 rdx;
Mov (rdx, y);
])
| "<" | "<=" | "==" | "!=" | ">=" | ">" ->
2024-01-31 17:48:25 +01:00
if in_memory x then
with_extra (fun extra ->
2024-01-30 18:16:44 +01:00
[
Binop ("^", rax, rax);
Mov (x, extra);
Binop ("cmp", extra, y);
2024-01-31 17:48:25 +01:00
Set (suffix op, Registers.rax);
2024-01-30 18:16:44 +01:00
Sal1 rax;
Or1 rax;
Mov (rax, y);
])
else
2024-01-31 17:48:25 +01:00
without_extra (fun () ->
2024-01-30 18:16:44 +01:00
[
Binop ("^", rax, rax);
Binop ("cmp", x, y);
2024-01-31 17:48:25 +01:00
Set (suffix op, Registers.rax);
2024-01-30 18:16:44 +01:00
Sal1 rax;
Or1 rax;
Mov (rax, y);
])
| "*" ->
2024-01-31 17:48:25 +01:00
without_extra (fun () ->
if in_memory y then
2024-01-30 18:16:44 +01:00
[
Dec y;
Mov (x, rax);
Sar1 rax;
Binop (op, y, rax);
Or1 rax;
Mov (rax, y);
]
2024-01-31 17:48:25 +01:00
else [ Dec y; Mov (x, rax); Sar1 rax; Binop (op, rax, y); Or1 y ])
2024-01-30 18:16:44 +01:00
| "&&" ->
2024-01-31 17:48:25 +01:00
with_extra (fun extra ->
2024-01-30 18:16:44 +01:00
[
Dec x;
Mov (x, rax);
Binop (op, x, rax);
Mov (L 0, rax);
2024-01-31 17:48:25 +01:00
Set ("ne", Registers.rax);
2024-01-30 18:16:44 +01:00
Dec y;
Mov (y, extra);
Binop (op, y, extra);
Mov (L 0, extra);
2024-01-31 17:48:25 +01:00
Set ("ne", as_register extra);
2024-01-30 18:16:44 +01:00
Binop (op, extra, rax);
2024-01-31 17:48:25 +01:00
Set ("ne", Registers.rax);
2024-01-30 18:16:44 +01:00
Sal1 rax;
Or1 rax;
Mov (rax, y);
])
| "!!" ->
2024-01-31 17:48:25 +01:00
without_extra (fun () ->
2024-01-30 18:16:44 +01:00
[
Mov (y, rax);
Sar1 rax;
Sar1 x;
Binop (op, x, rax);
Mov (L 0, rax);
2024-01-31 17:48:25 +01:00
Set ("ne", Registers.rax);
2024-01-30 18:16:44 +01:00
Sal1 rax;
Or1 rax;
Mov (rax, y);
])
| "+" ->
2024-01-31 17:48:25 +01:00
without_extra (fun () ->
2024-01-30 18:16:44 +01:00
if in_memory x && in_memory y then
[ Mov (x, rax); Dec rax; Binop ("+", rax, y) ]
else [ Binop (op, x, y); Dec y ])
| "-" ->
2024-01-31 17:48:25 +01:00
without_extra (fun () ->
2024-01-30 18:16:44 +01:00
if in_memory x && in_memory y then
[ Mov (x, rax); Binop (op, rax, y); Or1 y ]
else [ Binop (op, x, y); Or1 y ])
| _ ->
failwith (Printf.sprintf "Unexpected pattern: %s: %d" __FILE__ __LINE__)
2024-02-15 10:18:14 +01:00
let safepoint_functions =
[
2024-02-15 10:43:47 +01:00
label "s__Infix_58";
label "substring";
label "clone";
2024-02-15 10:49:20 +01:00
builtin_label "string";
2024-02-15 10:43:47 +01:00
label "stringcat";
label "string";
2024-02-15 10:49:20 +01:00
builtin_label "closure";
builtin_label "array";
builtin_label "sexp";
2024-02-15 10:43:47 +01:00
label "i__Infix_4343"
(* "makeArray"; not required as do not have ptr arguments *)
(* "makeString"; not required as do not have ptr arguments *)
(* "getEnv", not required as do not have ptr arguments *)
2024-02-15 10:18:14 +01:00
(* "set_args", not required as do not have ptr arguments *);
(* Lsprintf, or Bsprintf is an extra dirty hack that works? *)
]
2024-02-15 10:43:47 +01:00
let vararg_functions =
[
(label "printf", 1);
(label "fprintf", 2);
(label "sprintf", 1);
(label "failure", 1);
]
let compile_call env ?fname nargs tail =
2024-02-15 10:18:14 +01:00
let fname =
Option.map
(fun fname ->
match fname.[0] with
2024-02-15 10:49:20 +01:00
| '.' -> builtin_label (String.sub fname 1 (String.length fname - 1))
2024-02-15 10:18:14 +01:00
| _ -> fname)
fname
in
let tail_call_optimization_applicable =
let allowed_function =
2024-03-13 17:07:22 +01:00
match fname with
| Some fname ->
let is_vararg =
Option.is_some @@ List.assoc_opt fname vararg_functions
in
let is_internal = fname.[0] = 'B' in
(not is_internal) && not is_vararg
2024-03-13 17:07:22 +01:00
| None -> true
in
let same_arguments_count = env#nargs = nargs in
2024-02-02 17:24:40 +01:00
tail && allowed_function && same_arguments_count
in
let compile_tail_call env fname nargs =
let rec push_args env acc = function
| 0 -> (env, acc)
| n ->
let x, env = env#pop in
push_args env (mov x (env#loc (Value.Arg (n - 1))) @ acc) (n - 1)
in
let env, pushs = push_args env [] nargs in
let env, jump =
match fname with
| Some fname -> (env, [ Jmp fname ])
| None ->
let closure, env = env#pop in
(env, [ Mov (closure, r15); JmpI r15 ])
in
let _, env = env#allocate in
(env, pushs @ [ Mov (rbp, rsp); Pop rbp ] @ jump)
in
let compile_common_call env fname nargs =
2024-02-15 10:18:14 +01:00
let setup_arguments env nargs =
let rec pop_arguments env acc = function
| 0 -> (env, acc)
| n ->
let x, env = env#pop in
pop_arguments env (x :: acc) (n - 1)
in
2024-02-02 17:24:40 +01:00
let move_arguments args arg_locs =
List.fold_left2
(fun acc arg arg_loc ->
match arg_loc with
| Register r -> Mov (arg, r) :: acc
| Stack -> Push arg :: acc)
[] args arg_locs
in
let env, args = pop_arguments env [] nargs in
let arg_locs, stack_slots = env#arguments_locations (List.length args) in
2024-02-02 17:24:40 +01:00
let setup_args_code = move_arguments args arg_locs in
(stack_slots, env, setup_args_code)
in
let protect_registers env =
let pushr, popr =
List.split @@ List.map (fun r -> (Push r, Pop r)) env#live_registers
in
if env#has_closure then (Push r15 :: pushr, Pop r15 :: popr)
else (pushr, popr)
in
let align_stack saved_registers stack_arguments =
let aligned = (saved_registers + stack_arguments) mod 2 == 0 in
if aligned && stack_arguments = 0 then ([], [])
else if aligned then
([], [ Binop ("+", L (word_size * stack_arguments), rsp) ])
else
2024-03-14 09:01:50 +01:00
( [ Push filler ],
[ Binop ("+", L (word_size * (1 + stack_arguments)), rsp) ] )
in
let call env fname =
match fname with
| Some fname -> (env, [ Call fname ])
| None ->
let closure, env = env#pop in
(env, [ Mov (closure, r15); CallI r15 ])
in
let move_result env =
let y, env = env#allocate in
(env, [ Mov (rax, y) ])
in
let add_printf_count =
2024-02-05 19:18:41 +01:00
match fname with
2024-02-15 10:43:47 +01:00
| Some fname -> (
match List.assoc_opt fname vararg_functions with
| Some n -> [ Mov (L (nargs - n), r11) ]
| None -> [])
2024-02-05 19:18:41 +01:00
| _ -> []
in
2024-02-15 10:18:14 +01:00
let stack_slots, env, setup_args_code = setup_arguments env nargs in
let push_registers, pop_registers = protect_registers env in
let align_prologue, align_epilogue =
align_stack (List.length push_registers) stack_slots
in
let env, call = call env fname in
let env, move_result = move_result env in
( env,
push_registers @ align_prologue @ setup_args_code @ add_printf_count
@ call @ align_epilogue @ List.rev pop_registers @ move_result )
in
2024-02-15 10:18:14 +01:00
let safepoint_call =
match fname with
| Some fname -> List.mem fname safepoint_functions
| None -> false
in
let compile_safe_point_call env fname nargs =
let setup_arguments env nargs =
let rec pop_arguments env acc = function
| 0 -> (env, acc)
| n ->
let x, env = env#pop in
pop_arguments env (x :: acc) (n - 1)
in
let env, args = pop_arguments env [] nargs in
2024-02-20 14:10:31 +01:00
let setup_args_code = List.map (fun arg -> Push arg) @@ List.rev args in
2024-03-13 17:07:22 +01:00
let setup_args_code = setup_args_code @ [ Mov (rsp, rdi) ] in
2024-02-15 10:18:14 +01:00
let setup_args_code =
2024-02-15 10:49:20 +01:00
if fname = builtin_label "closure" then
2024-02-20 14:10:31 +01:00
setup_args_code @ [ Mov (L (box (nargs - 1)), rsi) ]
2024-02-15 10:49:20 +01:00
else if fname = builtin_label "sexp" || fname = builtin_label "array"
2024-02-20 14:10:31 +01:00
then setup_args_code @ [ Mov (L (box nargs), rsi) ]
2024-02-15 10:49:20 +01:00
else setup_args_code
2024-02-15 10:18:14 +01:00
in
(nargs, env, setup_args_code)
in
let protect_registers env =
let pushr, popr =
List.split @@ List.map (fun r -> (Push r, Pop r)) env#live_registers
in
if env#has_closure then (Push r15 :: pushr, Pop r15 :: popr)
else (pushr, popr)
in
let align_stack saved_registers stack_arguments =
let aligned = (saved_registers + stack_arguments) mod 2 == 0 in
if aligned && stack_arguments = 0 then ([], [])
else if aligned then
([], [ Binop ("+", L (word_size * stack_arguments), rsp) ])
else
2024-03-14 09:01:50 +01:00
( [ Push filler ],
2024-02-15 10:18:14 +01:00
[ Binop ("+", L (word_size * (1 + stack_arguments)), rsp) ] )
in
let call env fname = (env, [ Call fname ]) in
let move_result env =
let y, env = env#allocate in
(env, [ Mov (rax, y) ])
in
let stack_slots, env, setup_args_code = setup_arguments env nargs in
let push_registers, pop_registers = protect_registers env in
let align_prologue, align_epilogue =
align_stack (List.length push_registers) stack_slots
in
let env, call = call env fname in
let env, move_result = move_result env in
( env,
push_registers @ align_prologue @ setup_args_code @ call @ align_epilogue
@ List.rev pop_registers @ move_result )
in
if safepoint_call then compile_safe_point_call env (Option.get fname) nargs
else if tail_call_optimization_applicable then
compile_tail_call env fname nargs
else compile_common_call env fname nargs
2024-01-30 18:16:44 +01:00
(* Symbolic stack machine evaluator
compile : env -> prg -> env * instr list
Take an environment, a stack machine program, and returns a pair ---
the updated environment and the list of x86 instructions
*)
let compile cmd env imports code =
(* SM.print_prg code;
flush stdout; *)
2019-09-10 00:46:10 +03:00
let rec compile' env scode =
2018-03-05 00:54:50 +03:00
match scode with
2023-09-04 21:43:28 +02:00
| [] -> (env, [])
2018-03-05 00:54:50 +03:00
| 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' =
2023-09-04 21:43:28 +02:00
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
2023-09-04 21:43:28 +02:00
match instr with
| PUBLIC name -> (env#register_public name, [])
| EXTERN name -> (env#register_extern name, [])
| IMPORT _ -> (env, [])
| CLOSURE (name, closure) ->
2024-03-14 09:01:50 +01:00
let ext = if env#is_external name then E else I in
let address = M (F, ext, A, name) in
2024-01-30 18:16:44 +01:00
let l, env = env#allocate in
let env, push_closure_code =
List.fold_left
(fun (env, code) c ->
let cr, env = env#allocate in
2024-01-31 17:48:25 +01:00
(env, mov (env#loc c) cr @ code))
2024-01-30 18:16:44 +01:00
(env, []) closure
2023-09-04 21:43:28 +02:00
in
2024-01-30 18:16:44 +01:00
let env, call_code =
compile_call env ~fname:".closure"
(1 + List.length closure)
false
2023-09-04 21:43:28 +02:00
in
2024-03-14 09:01:50 +01:00
(env, push_closure_code @ mov address l @ call_code)
2023-09-04 21:43:28 +02:00
| CONST n ->
let s, env' = env#allocate in
(env', [ Mov (L (box n), s) ])
| STRING s ->
2024-03-14 09:01:50 +01:00
let addr, env = env#string s in
2023-09-04 21:43:28 +02:00
let l, env = env#allocate in
let env, call = compile_call env ~fname:".string" 1 false in
2024-03-14 09:01:50 +01:00
(env, mov addr l @ call)
2023-09-04 21:43:28 +02:00
| LDA x ->
let s, env' = (env#variable x)#allocate in
let s', env'' = env'#allocate in
2024-01-26 18:45:03 +01:00
(env'', [ Lea (env'#loc x, rax); Mov (rax, s); Mov (rax, s') ])
2023-09-04 21:43:28 +02:00
| LD x -> (
let s, env' = (env#variable x)#allocate in
( env',
match s with
2024-01-26 18:45:03 +01:00
| S _ | M _ -> [ Mov (env'#loc x, rax); Mov (rax, s) ]
2023-09-04 21:43:28 +02:00
| _ -> [ Mov (env'#loc x, s) ] ))
| ST x -> (
let env' = env#variable x in
let s = env'#peek in
( env',
match s with
2024-01-26 18:45:03 +01:00
| S _ | M _ -> [ Mov (s, rax); Mov (rax, env'#loc x) ]
2023-09-04 21:43:28 +02:00
| _ -> [ Mov (s, env'#loc x) ] ))
| STA -> compile_call env ~fname:".sta" 3 false
2023-09-04 21:43:28 +02:00
| STI -> (
2024-01-30 18:16:44 +01:00
let v, env = env#pop in
let x = env#peek in
( env,
2023-09-04 21:43:28 +02:00
match x with
| S _ | M _ ->
[
2024-01-26 18:45:03 +01:00
Mov (v, rdx);
Mov (x, rax);
Mov (rdx, I (0, rax));
Mov (rdx, x);
2023-09-04 21:43:28 +02:00
]
2024-01-26 18:45:03 +01:00
| _ -> [ Mov (v, rax); Mov (rax, I (0, x)); Mov (rax, x) ] ))
2024-01-30 18:16:44 +01:00
| BINOP op -> compile_binop env op
2023-09-04 21:43:28 +02:00
| LABEL s | FLABEL s | SLABEL s -> (env, [ Label s ])
| JMP l -> ((env#set_stack l)#set_barrier, [ Jmp l ])
| CJMP (s, l) ->
let x, env = env#pop in
( env#set_stack l,
[ Sar1 x; (*!!!*) Binop ("cmp", L 0, x); CJmp (s, l) ] )
| BEGIN (f, nargs, nlocals, closure, args, scopes) ->
let _ =
let is_safepoint = List.mem f safepoint_functions in
let is_vararg =
Option.is_some @@ List.assoc_opt f vararg_functions
in
if is_safepoint || is_vararg then
raise
(Failure
(Printf.sprintf
"Function name %s is reserved for built-in" f))
in
2023-09-04 21:43:28 +02:00
let rec stabs_scope scope =
let names = []
(* List.map
2023-09-04 21:43:28 +02:00
(fun (name, index) ->
Meta
(Printf.sprintf "\t.stabs \"%s:1\",128,0,0,-%d" name
(stack_offset index)))
scope.names *)
2023-09-04 21:43:28 +02:00
in
names
@ (if names = [] then []
2024-01-26 18:45:03 +01:00
else
[
(* Meta
(Printf.sprintf "\t.stabn 192,0,0,%s-%s" scope.blab f); *)
2024-01-26 18:45:03 +01:00
])
2023-09-04 21:43:28 +02:00
@ (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); *)
2023-09-04 21:43:28 +02:00
]
in
let name =
if f.[0] = 'L' then String.sub f 1 (String.length f - 1)
else f
in
env#assert_empty_stack;
let has_closure = closure <> [] in
let env = env#enter f nargs nlocals has_closure in
( env,
[ (* Meta (Printf.sprintf "\t.type %s, @function" name) *) ]
2023-09-04 21:43:28 +02:00
@ (if f = "main" then []
2024-01-26 18:45:03 +01:00
else
[
(* Meta
(Printf.sprintf "\t.stabs \"%s:F1\",36,0,0,%s" name f); *)
2024-01-26 18:45:03 +01:00
]
(* @ List.mapi
2024-01-26 18:45:03 +01:00
(fun i a ->
Meta
(Printf.sprintf "\t.stabs \"%s:p1\",160,0,0,%d" a
((i * 4) + 8)))
args *)
2024-01-26 18:45:03 +01:00
@ List.flatten
@@ List.map stabs_scope scopes)
2023-09-04 21:43:28 +02:00
@ [ Meta "\t.cfi_startproc" ]
@ (if f = cmd#topname then
2024-01-26 18:45:03 +01:00
[
2024-03-14 09:01:50 +01:00
Mov (M (D, I, V, "_init"), rax);
2024-01-26 18:45:03 +01:00
Binop ("test", rax, rax);
CJmp ("z", "_continue");
Ret;
Label "_ERROR";
2024-02-15 10:43:47 +01:00
Call (label "binoperror");
2024-01-26 18:45:03 +01:00
Ret;
Label "_ERROR2";
2024-02-15 10:43:47 +01:00
Call (label "binoperror2");
2024-01-26 18:45:03 +01:00
Ret;
Label "_continue";
2024-03-14 09:01:50 +01:00
Mov (L 1, M (D, I, V, "_init"));
2024-01-26 18:45:03 +01:00
]
else [])
2023-09-04 21:43:28 +02:00
@ [
2024-01-26 18:45:03 +01:00
Push rbp;
2024-01-30 18:16:44 +01:00
Meta "\t.cfi_def_cfa_offset\t8";
Meta "\t.cfi_offset 5, -8";
2024-01-26 18:45:03 +01:00
Mov (rsp, rbp);
2023-09-04 21:43:28 +02:00
Meta "\t.cfi_def_cfa_register\t5";
2024-03-14 09:01:50 +01:00
Binop ("-", C env#lsize, rsp);
2024-01-30 18:16:44 +01:00
Mov (rdi, r12);
Mov (rsi, r13);
Mov (rcx, r14);
Mov (rsp, rdi);
2024-03-14 09:01:50 +01:00
Mov (filler, rsi);
Mov (C env#allocated_size, rcx);
2024-01-30 18:16:44 +01:00
Repmovsl;
Mov (r12, rdi);
Mov (r13, rsi);
Mov (r14, rcx);
2023-09-04 21:43:28 +02:00
]
@ (if f = "main" then
2024-01-26 18:45:03 +01:00
[
2024-03-14 09:01:50 +01:00
(* Align stack as main function is the only function that could be called without alignment. TODO *)
Mov (L 0xF, rax);
2024-02-02 17:38:04 +01:00
Binop ("test", rsp, rax);
CJmp ("z", "_ALIGNED");
2024-03-14 09:01:50 +01:00
Push filler;
2024-02-02 17:38:04 +01:00
Label "_ALIGNED";
(* Initialize gc and arguments *)
2024-01-30 18:16:44 +01:00
Push (R Registers.rdi);
Push (R Registers.rsi);
2024-01-26 18:45:03 +01:00
Call "__gc_init";
2024-01-30 18:16:44 +01:00
Pop (R Registers.rsi);
Pop (R Registers.rdi);
Call "set_args";
2024-01-26 18:45:03 +01:00
]
else [])
@
2023-09-04 21:43:28 +02:00
if f = cmd#topname then
List.map
(fun i -> Call ("init" ^ i))
(List.filter (fun i -> i <> "Std") imports)
else [] )
| END ->
let x, env = env#pop in
env#assert_empty_stack;
let name = env#fname in
( env#leave,
[
2024-01-26 18:45:03 +01:00
Mov (x, rax);
2023-09-04 21:43:28 +02:00
(*!!*)
Label env#epilogue;
2024-01-26 18:45:03 +01:00
Mov (rbp, rsp);
Pop rbp;
2023-09-04 21:43:28 +02:00
]
2024-01-26 18:45:03 +01:00
@ (if name = "main" then [ Binop ("^", rax, rax) ] else [])
2023-09-04 21:43:28 +02:00
@ [
Meta "\t.cfi_restore\t5";
Meta "\t.cfi_def_cfa\t4, 4";
Ret;
2023-09-04 21:43:28 +02:00
Meta "\t.cfi_endproc";
Meta
2024-02-02 17:24:40 +01:00
(* Allocate space for the symbolic stack
Add extra word if needed to preserve alignment *)
2023-09-04 21:43:28 +02:00
(Printf.sprintf "\t.set\t%s,\t%d" env#lsize
2024-02-02 17:24:40 +01:00
(if env#allocated mod 2 == 0 then
2024-01-26 18:45:03 +01:00
env#allocated * word_size
2024-02-02 17:24:40 +01:00
else (env#allocated + 1) * word_size));
2023-09-04 21:43:28 +02:00
Meta
(Printf.sprintf "\t.set\t%s,\t%d" env#allocated_size
env#allocated);
(* Meta (Printf.sprintf "\t.size %s, .-%s" name name); *)
2023-09-04 21:43:28 +02:00
] )
| RET ->
let x = env#peek in
2024-01-26 18:45:03 +01:00
(env, [ Mov (x, rax); Jmp env#epilogue ])
| ELEM -> compile_call env ~fname:".elem" 2 false
| CALL (fname, n, tail) -> compile_call env ~fname n tail
| CALLC (n, tail) -> compile_call env n tail
2023-09-04 21:43:28 +02:00
| SEXP (t, n) ->
let s, env = env#allocate in
let env, code = compile_call env ~fname:".sexp" (n + 1) false in
2024-02-05 19:18:41 +01:00
(env, mov (L (box (env#hash t))) s @ code)
2023-09-04 21:43:28 +02:00
| DROP -> (snd env#pop, [])
| DUP ->
let x = env#peek in
let s, env = env#allocate in
(env, mov x s)
| SWAP ->
let x, y = env#peek2 in
(env, [ Push x; Push y; Pop x; Pop y ])
| TAG (t, n) ->
let s1, env = env#allocate in
let s2, env = env#allocate in
let env, code = compile_call env ~fname:".tag" 3 false in
2024-02-05 19:18:41 +01:00
(env, mov (L (box (env#hash t))) s1 @ mov (L (box n)) s2 @ code)
2023-09-04 21:43:28 +02:00
| ARRAY n ->
let s, env = env#allocate in
let env, code = compile_call env ~fname:".array_patt" 2 false in
2023-09-04 21:43:28 +02:00
(env, [ Mov (L (box n), s) ] @ code)
| PATT StrCmp -> compile_call env ~fname:".string_patt" 2 false
2023-09-04 21:43:28 +02:00
| PATT patt ->
compile_call env
~fname:
(match patt with
| Boxed -> ".boxed_patt"
| UnBoxed -> ".unboxed_patt"
| Array -> ".array_tag_patt"
| String -> ".string_tag_patt"
| Sexp -> ".sexp_tag_patt"
| Closure -> ".closure_tag_patt"
| StrCmp ->
failwith
(Printf.sprintf "Unexpected pattern: StrCmp %s: %d"
__FILE__ __LINE__))
2023-09-04 21:43:28 +02:00
1 false
2024-04-17 11:26:47 +03:00
| LINE line -> env#gen_line
2023-09-04 21:43:28 +02:00
| FAIL ((line, col), value) ->
let v, env = if value then (env#peek, env) else env#pop in
2024-03-14 09:01:50 +01:00
let msg_addr, env = env#string cmd#get_infile in
2024-01-30 18:16:44 +01:00
let vr, env = env#allocate in
let sr, env = env#allocate in
let liner, env = env#allocate in
let colr, env = env#allocate in
let env, code =
compile_call env ~fname:".match_failure" 4 false
in
2024-01-30 18:16:44 +01:00
let _, env = env#pop in
2023-09-04 21:43:28 +02:00
( env,
[
2024-01-30 18:16:44 +01:00
Mov (L col, colr);
Mov (L line, liner);
2024-03-14 09:01:50 +01:00
Mov (msg_addr, sr);
2024-01-30 18:16:44 +01:00
Mov (v, vr);
]
@ code )
2023-09-04 21:43:28 +02:00
| 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
2023-09-04 21:43:28 +02:00
( 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
2023-09-04 21:43:28 +02:00
2024-01-30 18:16:44 +01:00
module AbstractSymbolicStack : sig
type 'a t
type 'a symbolic_location = Stack of int | Register of 'a
val empty : 'a array -> 'a t
val is_empty : _ t -> bool
val live_registers : 'a t -> 'a list
val stack_size : _ t -> int
val allocate : 'a t -> 'a t * 'a symbolic_location
val pop : 'a t -> 'a t * 'a symbolic_location
val peek : 'a t -> 'a symbolic_location
val peek2 : 'a t -> 'a symbolic_location * 'a symbolic_location
end = struct
type 'a symbolic_location = Stack of int | Register of 'a
(* Last allocated position on symbolic stack *)
type stack_state = S of int | R of int | E
type 'a t = stack_state * 'a array
let empty registers = (E, registers)
let next (state, registers) =
let state =
match state with
| S n -> S (n + 1)
| R n when n + 1 = Array.length registers -> S 0
| R n -> R (n + 1)
| E -> R 0
in
(state, registers)
let previos (state, registers) =
let state =
match state with
| S 0 -> R (Array.length registers - 1)
| S n -> S (n - 1)
| R 0 -> E
| R n -> R (n - 1)
| E -> failwith (Printf.sprintf "Empty stack %s: %d" __FILE__ __LINE__)
in
(state, registers)
let location (state, registers) =
match state with
| S n -> Stack n
| R n -> Register registers.(n)
| E -> failwith (Printf.sprintf "Empty stack %s: %d" __FILE__ __LINE__)
let is_empty (state, _) = match state with E -> true | _ -> false
let live_registers (stack, registers) =
match stack with
| S _ -> Array.to_list registers
| R n -> Array.to_list (Array.sub registers 0 (n + 1))
| E -> []
let stack_size (state, _) = match state with S n -> n + 1 | R _ | E -> 0
let allocate state =
let state = next state in
(state, location state)
let pop stack = (previos stack, location stack)
let peek stack = location stack
let peek2 stack = (location stack, location (previos stack))
end
module SymbolicStack : sig
type t
val empty : int -> int -> t
val is_empty : t -> bool
val live_registers : t -> opnd list
val stack_size : t -> int
val allocate : t -> t * opnd
val pop : t -> t * opnd
val peek : t -> opnd
val peek2 : t -> opnd * opnd
end = struct
2024-01-31 17:48:25 +01:00
type t = { state : Register.t AbstractSymbolicStack.t; nlocals : int }
2024-01-30 18:16:44 +01:00
(* romanv: add free args registers? *)
let empty _nargs nlocals =
{
state = AbstractSymbolicStack.empty Registers.extra_caller_saved_registers;
nlocals;
}
let opnd_from_loc v = function
| AbstractSymbolicStack.Register r -> R r
| AbstractSymbolicStack.Stack n -> S (n + v.nlocals)
let is_empty v = AbstractSymbolicStack.is_empty v.state
let live_registers v =
List.map (fun r -> R r) (AbstractSymbolicStack.live_registers v.state)
let stack_size v = AbstractSymbolicStack.stack_size v.state
let allocate v =
let state, loc = AbstractSymbolicStack.allocate v.state in
({ v with state }, opnd_from_loc v loc)
let pop v =
let state, loc = AbstractSymbolicStack.pop v.state in
({ v with state }, opnd_from_loc v loc)
let peek v = opnd_from_loc v (AbstractSymbolicStack.peek v.state)
let peek2 v =
let loc1, loc2 = AbstractSymbolicStack.peek2 v.state in
(opnd_from_loc v loc1, opnd_from_loc v loc2)
end
(* Environment for symbolic stack machine *)
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 =
2023-09-04 21:43:28 +02:00
let chars =
"_abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789'"
in
2024-01-30 18:16:44 +01:00
let argument_registers =
Array.map (fun r -> R r) Registers.argument_registers
in
let num_of_argument_registers = Array.length argument_registers in
2023-09-04 21:43:28 +02:00
(* 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
2024-02-02 17:24:40 +01:00
val globals = S.empty (* a set of global variables *)
2024-03-14 09:01:50 +01:00
val stringm : string M.t = M.empty (* a string map *)
2024-02-02 17:24:40 +01:00
val scount = 0 (* string count *)
2023-09-04 21:43:28 +02:00
val stack_slots = 0 (* maximal number of stack positions *)
2024-02-02 17:24:40 +01:00
val static_size = 0 (* static data size *)
val stack = SymbolicStack.empty 0 0 (* symbolic stack *)
val nargs = 0 (* number of function arguments *)
val locals = [] (* function local variables *)
val fname = "" (* function name *)
val stackmap = M.empty (* labels to stack map *)
val barrier = false (* barrier condition *)
val max_locals_size =
0 (* maximal number of stack position in all functions *)
2023-09-04 21:43:28 +02:00
val has_closure = false
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
2023-09-04 21:43:28 +02:00
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 fname = fname
2023-09-04 21:43:28 +02:00
2019-10-16 21:07:27 +03:00
method leave =
2023-09-04 21:43:28 +02:00
if stack_slots > max_locals_size then {<max_locals_size = stack_slots>}
2018-12-12 12:42:38 +03:00
else self
2019-09-10 00:46:10 +03:00
2024-01-31 17:48:25 +01:00
method show_stack = show_opnd (SymbolicStack.peek 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;
2023-09-04 21:43:28 +02:00
Printf.printf ")\n")
locals;
2018-05-16 09:24:40 +03:00
Printf.printf "END LOCALS\n"
(* Assert empty stack *)
2024-01-30 18:16:44 +01:00
method assert_empty_stack = assert (SymbolicStack.is_empty 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 *)
2023-09-04 21:43:28 +02:00
method set_barrier = {<barrier = true>}
2018-05-16 09:24:40 +03:00
(* drop barrier *)
2023-09-04 21:43:28 +02:00
method drop_barrier = {<barrier = false>}
2019-09-10 00:46:10 +03:00
(* drop stack *)
2024-01-30 18:16:44 +01:00
method drop_stack = {<stack = SymbolicStack.empty nargs static_size>}
2018-05-16 09:24:40 +03:00
(* associates a stack to a label *)
2023-09-04 21:43:28 +02:00
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 *)
2023-09-04 21:43:28 +02:00
method retrieve_stack l =
(*Printf.printf "Retrieving stack for %s\n" l;*)
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 *)
2023-09-04 21:43:28 +02:00
method has_stack l =
(*Printf.printf "Retrieving stack for %s\n" l;*)
M.mem l stackmap
2024-03-14 09:01:50 +01:00
method is_external name = S.mem name externs
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
2024-03-14 09:01:50 +01:00
| Value.Global name ->
let loc_name = "global_" ^ name in
let ext = if self#is_external name then E else I in
M (D, ext, V, loc_name)
| Value.Fun name ->
let ext = if self#is_external name then E else I in
M (F, ext, A, name)
2023-09-04 21:43:28 +02:00
| Value.Local i -> S i
2024-01-30 18:16:44 +01:00
| Value.Arg i when i < num_of_argument_registers -> argument_registers.(i)
| Value.Arg i -> S (-(i - num_of_argument_registers) - 1)
| Value.Access i -> I (word_size * (i + 1), r15)
2023-09-04 21:43:28 +02:00
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 =
2024-01-30 18:16:44 +01:00
let stack, opnd = SymbolicStack.allocate stack in
let stack_slots =
max stack_slots (static_size + SymbolicStack.stack_size stack)
2018-03-05 00:54:50 +03:00
in
2024-01-30 18:16:44 +01:00
(opnd, {<stack_slots; stack>})
2018-03-05 00:54:50 +03:00
(* pops one operand from the symbolic stack *)
2023-09-04 21:43:28 +02:00
method pop =
2024-01-30 18:16:44 +01:00
let stack, opnd = SymbolicStack.pop stack in
(opnd, {<stack>})
2018-03-05 00:54:50 +03:00
2024-01-30 18:16:44 +01:00
(* is rdx register in use *)
method rdx_in_use = nargs > 2
method arguments_locations n =
if n < num_of_argument_registers then
( Array.to_list (Array.sub argument_registers 0 n)
|> List.map (fun r -> Register r),
0 )
else
( (Array.to_list argument_registers |> List.map (fun r -> Register r))
@ List.init (n - num_of_argument_registers) (fun _ -> Stack),
n - num_of_argument_registers )
2018-03-05 00:54:50 +03:00
2018-05-16 09:24:40 +03:00
(* peeks the top of the stack (the stack does not change) *)
2024-01-30 18:16:44 +01:00
method peek = SymbolicStack.peek stack
2018-05-16 09:24:40 +03:00
(* peeks two topmost values from the stack (the stack itself does not change) *)
2024-01-30 18:16:44 +01:00
method peek2 = SymbolicStack.peek2 stack
2018-05-16 09:24:40 +03:00
(* tag hash: gets a hash for a string tag *)
method hash tag =
2023-09-04 21:43:28 +02:00
let h = Stdlib.ref 0 in
for i = 0 to min (String.length tag - 1) 9 do
2023-09-04 21:43:28 +02:00
h := (!h lsl 6) lor String.index chars tag.[i]
2018-05-16 09:24:40 +03:00
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
2023-09-04 21:43:28 +02:00
| Value.Global name -> {<globals = S.add ("global_" ^ name) globals>}
| _ -> 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 =
2023-09-04 21:43:28 +02:00
let n = String.length x in
let buf = Buffer.create (n * 2) in
2019-12-31 00:59:28 +03:00
let rec iterate i =
2023-09-04 21:43:28 +02:00
if i < n then (
2020-01-03 01:38:49 +03:00
(match x.[i] with
2023-09-04 21:43:28 +02:00
| '"' -> Buffer.add_string buf "\\\""
2020-01-03 01:38:49 +03:00
| '\n' -> Buffer.add_string buf "\n"
| '\t' -> Buffer.add_string buf "\t"
2023-09-04 21:43:28 +02:00
| c -> Buffer.add_char buf c);
iterate (i + 1))
2019-12-31 00:59:28 +03:00
in
iterate 0;
Buffer.contents buf
in
let x = escape x in
2024-03-14 09:01:50 +01:00
let name = M.find_opt x stringm in
match name with
| Some name -> (M (D, I, A, name), self)
| None ->
let name = Printf.sprintf "string_%d" scount in
let m = M.add x name stringm in
(M (D, I, A, name), {<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
2023-09-04 21:43:28 +02:00
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
2024-02-15 10:43:47 +01:00
method allocated_size = label (Printf.sprintf "S%s_SIZE" fname)
2023-09-04 21:43:28 +02:00
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 =
2023-09-04 21:43:28 +02:00
{<nargs
; static_size = nlocals
; stack_slots = nlocals
2024-01-30 18:16:44 +01:00
; stack = SymbolicStack.empty nargs nlocals
2023-09-04 21:43:28 +02:00
; fname = f
; 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 *)
2024-02-15 10:43:47 +01:00
method epilogue = label (Printf.sprintf "%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 *)
2024-02-15 10:43:47 +01:00
method lsize = label (Printf.sprintf "%s_SIZE" fname)
2023-09-04 21:43:28 +02:00
2018-04-11 00:47:46 +03:00
(* returns a list of live registers *)
2024-01-30 18:16:44 +01:00
method live_registers =
Array.to_list
(Array.sub argument_registers 0
(min nargs (Array.length argument_registers)))
@ SymbolicStack.live_registers stack
2019-09-10 00:46:10 +03:00
(* generate a line number information for current function *)
2024-04-17 11:26:47 +03:00
method gen_line =
let lab = Printf.sprintf ".L%d" nlabels in
2023-09-04 21:43:28 +02:00
( {<nlabels = nlabels + 1; first_line = false>},
if fname = "main" then
[ (* Meta (Printf.sprintf "\t.stabn 68,0,%d,%s" line lab); *) Label lab ]
2023-09-04 21:43:28 +02:00
else
(if first_line then
[ (* Meta (Printf.sprintf "\t.stabn 68,0,%d,0" line) *) ]
2024-01-26 18:45:03 +01:00
else [])
2023-09-04 21:43:28 +02:00
@ [
(* Meta (Printf.sprintf "\t.stabn 68,0,%d,%s-%s" line lab fname); *)
2023-09-04 21:43:28 +02:00
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 =
2023-09-04 21:43:28 +02:00
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
2023-09-04 21:43:28 +02:00
let data =
[ Meta "\t.data" ]
@ List.map
(fun (s, v) -> Meta (Printf.sprintf "%s:\t.string\t\"%s\"" v s))
env#strings
@ [
2024-01-26 18:45:03 +01:00
Meta "_init:\t.quad 0";
Meta "\t.section __DATA, custom_data, regular, no_dead_strip";
2024-01-26 18:45:03 +01:00
Meta (Printf.sprintf "filler:\t.fill\t%d, 8, 1" env#max_locals_size);
2023-09-04 21:43:28 +02:00
]
@ List.concat
@@ List.map
(fun s ->
[
(* For mach-o STABS format is not supported: Meta
( Printf.sprintf "\t.stabs \"%s:S1\",40,0,0,%s"
2023-09-04 21:43:28 +02:00
(String.sub s (String.length "global_")
(String.length s - String.length "global_"))
s); *)
2024-01-26 18:45:03 +01:00
Meta (Printf.sprintf "%s:\t.quad\t1" s);
2023-09-04 21:43:28 +02:00
])
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))
2023-09-04 21:43:28 +02:00
([
Meta (Printf.sprintf "\t.file \"%s\"" cmd#get_absolute_infile);
(* For mach-o STABS format is not supported: Meta
( Printf.sprintf "\t.stabs \"%s\",100,0,0,.Ltext"
cmd#get_absolute_infile); *)
2023-09-04 21:43:28 +02:00
]
@ globals @ data
@ [
Meta "\t.text";
Label ".Ltext";
(* For mach-o STABS format is not supported: Meta "\t.stabs \"data:t1=r1;0;4294967295;\",128,0,0,0"; *)
2023-09-04 21:43:28 +02:00
]
@ code);
2018-03-05 00:54:50 +03:00
Buffer.contents asm
2020-02-13 18:56:27 +03:00
let get_std_path () =
2023-09-04 21:43:28 +02:00
match Sys.getenv_opt "LAMA" with Some s -> s | None -> Stdpath.path
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
2023-09-04 21:43:28 +02:00
| [] -> 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);
2023-09-04 21:43:28 +02:00
let inc = get_std_path () in
let compiler = "clang" in
let flags = "-pie -target X86_64-apple-macho" in
2019-11-24 02:30:32 +03:00
match cmd#get_mode with
| `Default ->
2023-09-04 21:43:28 +02:00
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;
let gcc_cmdline =
Printf.sprintf "%s %s %s %s %s.s %s %s/runtime.a" compiler flags
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 ->
2023-09-04 21:43:28 +02:00
Sys.command
(Printf.sprintf "%s %s %s -c -g %s.s" compiler flags cmd#get_debug
2023-09-04 21:43:28 +02:00
cmd#basename)
2019-11-24 02:30:32 +03:00
| _ -> invalid_arg "must not happen"