Skip to content
Snippets Groups Projects
Commit 429d7194 authored by Damien's avatar Damien
Browse files

Initial commit.

parents
No related branches found
No related tags found
No related merge requests found
*.o
CC = gcc
CFLAGS = -Wall -ansi -std=gnu99 -Os #-DNDEBUG
LDFLAGS =
SRC = \
malloc.c \
misc.c \
qstr.c \
lexer.c \
lexerfile.c \
parse.c \
scope.c \
compile.c \
emitcommon.c \
emitcpy.c \
emitbc.c \
asmx64.c \
emitx64v2.c \
emitthumb.c \
asmthumb.c \
runtime.c \
bc.c \
main.c \
SRC_ASM = \
runtime1.s \
OBJ = $(SRC:.c=.o) $(SRC_ASM:.s=.o)
LIB =
PROG = py
$(PROG): $(OBJ)
$(CC) -o $@ $(OBJ) $(LIB) $(LDFLAGS)
runtime.o: runtime.c
$(CC) $(CFLAGS) -O3 -c -o $@ $<
bc.o: bc.c
$(CC) $(CFLAGS) -O3 -c -o $@ $<
parse.o: grammar.h
compile.o: grammar.h
emitcpy.o: emit.h
emitbc.o: emit.h
emitx64.o: emit.h
emitx64v2.o: emit.h
emitthumb.o: emit.h
clean:
/bin/rm $(OBJ)
#include <stdint.h>
#include <stdio.h>
#include <assert.h>
#include <string.h>
#include "misc.h"
#include "machine.h"
#include "asmthumb.h"
#define UNSIGNED_FIT8(x) (((x) & 0xffffff00) == 0)
#define UNSIGNED_FIT16(x) (((x) & 0xffff0000) == 0)
#define SIGNED_FIT8(x) (((x) & 0xffffff80) == 0) || (((x) & 0xffffff80) == 0xffffff80)
#define SIGNED_FIT9(x) (((x) & 0xffffff00) == 0) || (((x) & 0xffffff00) == 0xffffff00)
#define SIGNED_FIT12(x) (((x) & 0xfffff800) == 0) || (((x) & 0xfffff800) == 0xfffff800)
struct _asm_thumb_t {
int pass;
uint code_offset;
uint code_size;
byte *code_base;
byte dummy_data[8];
int next_label;
int max_num_labels;
int *label_offsets;
int num_locals;
uint push_reglist;
uint stack_adjust;
};
asm_thumb_t *asm_thumb_new() {
asm_thumb_t *as;
as = m_new(asm_thumb_t, 1);
as->pass = 0;
as->code_offset = 0;
as->code_size = 0;
as->code_base = NULL;
as->label_offsets = NULL;
as->num_locals = 0;
return as;
}
void asm_thumb_free(asm_thumb_t *as, bool free_code) {
if (free_code) {
m_free(as->code_base);
}
/*
if (as->label != NULL) {
int i;
for (i = 0; i < as->label->len; ++i)
{
Label *lab = &g_array_index(as->label, Label, i);
if (lab->unresolved != NULL)
g_array_free(lab->unresolved, true);
}
g_array_free(as->label, true);
}
*/
m_free(as);
}
void asm_thumb_start_pass(asm_thumb_t *as, int pass) {
as->pass = pass;
as->code_offset = 0;
as->next_label = 1;
if (pass == ASM_THUMB_PASS_1) {
as->max_num_labels = 0;
} else {
if (pass == ASM_THUMB_PASS_2) {
memset(as->label_offsets, -1, as->max_num_labels * sizeof(int));
}
}
}
void asm_thumb_end_pass(asm_thumb_t *as) {
if (as->pass == ASM_THUMB_PASS_1) {
// calculate number of labels need
if (as->next_label > as->max_num_labels) {
as->max_num_labels = as->next_label;
}
as->label_offsets = m_new(int, as->max_num_labels);
} else if (as->pass == ASM_THUMB_PASS_2) {
// calculate size of code in bytes
as->code_size = as->code_offset;
as->code_base = m_new(byte, as->code_size);
printf("code_size: %u\n", as->code_size);
}
/*
// check labels are resolved
if (as->label != NULL)
{
int i;
for (i = 0; i < as->label->len; ++i)
if (g_array_index(as->label, Label, i).unresolved != NULL)
return false;
}
*/
}
// all functions must go through this one to emit bytes
static byte *asm_thumb_get_cur_to_write_bytes(asm_thumb_t *as, int num_bytes_to_write) {
//printf("emit %d\n", num_bytes_to_write);
if (as->pass < ASM_THUMB_PASS_3) {
as->code_offset += num_bytes_to_write;
return as->dummy_data;
} else {
assert(as->code_offset + num_bytes_to_write <= as->code_size);
byte *c = as->code_base + as->code_offset;
as->code_offset += num_bytes_to_write;
return c;
}
}
uint asm_thumb_get_code_size(asm_thumb_t *as) {
return as->code_size;
}
void *asm_thumb_get_code(asm_thumb_t *as) {
// need to set low bit to indicate that it's thumb code
return (void *)(((machine_uint_t)as->code_base) | 1);
}
/*
static void asm_thumb_write_byte_1(asm_thumb_t *as, byte b1) {
byte *c = asm_thumb_get_cur_to_write_bytes(as, 1);
c[0] = b1;
}
*/
static void asm_thumb_write_op16(asm_thumb_t *as, uint op) {
byte *c = asm_thumb_get_cur_to_write_bytes(as, 2);
// little endian
c[0] = op;
c[1] = op >> 8;
}
static void asm_thumb_write_op32(asm_thumb_t *as, uint op1, uint op2) {
byte *c = asm_thumb_get_cur_to_write_bytes(as, 4);
// little endian, op1 then op2
c[0] = op1;
c[1] = op1 >> 8;
c[2] = op2;
c[3] = op2 >> 8;
}
/*
#define IMM32_L0(x) ((x) & 0xff)
#define IMM32_L1(x) (((x) >> 8) & 0xff)
#define IMM32_L2(x) (((x) >> 16) & 0xff)
#define IMM32_L3(x) (((x) >> 24) & 0xff)
static void asm_thumb_write_word32(asm_thumb_t *as, int w32) {
byte *c = asm_thumb_get_cur_to_write_bytes(as, 4);
c[0] = IMM32_L0(w32);
c[1] = IMM32_L1(w32);
c[2] = IMM32_L2(w32);
c[3] = IMM32_L3(w32);
}
*/
// rlolist is a bit map indicating desired lo-registers
#define OP_PUSH_RLIST(rlolist) (0xb400 | (rlolist))
#define OP_PUSH_RLIST_LR(rlolist) (0xb400 | 0x0100 | (rlolist))
#define OP_POP_RLIST(rlolist) (0xbc00 | (rlolist))
#define OP_POP_RLIST_PC(rlolist) (0xbc00 | 0x0100 | (rlolist))
#define OP_ADD_SP(num_words) (0xb000 | (num_words))
#define OP_SUB_SP(num_words) (0xb080 | (num_words))
void asm_thumb_entry(asm_thumb_t *as, int num_locals) {
// work out what to push and how many extra space to reserve on stack
// so that we have enough for all locals and it's aligned an 8-byte boundary
uint reglist;
uint stack_adjust;
if (num_locals < 0) {
num_locals = 0;
}
// don't ppop r0 because it's used for return value
switch (num_locals) {
case 0:
reglist = 0xf2;
stack_adjust = 0;
break;
case 1:
reglist = 0xf2;
stack_adjust = 0;
break;
case 2:
reglist = 0xfe;
stack_adjust = 0;
break;
case 3:
reglist = 0xfe;
stack_adjust = 0;
break;
default:
reglist = 0xfe;
stack_adjust = ((num_locals - 3) + 1) & (~1);
break;
}
asm_thumb_write_op16(as, OP_PUSH_RLIST_LR(reglist));
if (stack_adjust > 0) {
asm_thumb_write_op16(as, OP_SUB_SP(stack_adjust));
}
as->push_reglist = reglist;
as->stack_adjust = stack_adjust;
as->num_locals = num_locals;
}
void asm_thumb_exit(asm_thumb_t *as) {
if (as->stack_adjust > 0) {
asm_thumb_write_op16(as, OP_ADD_SP(as->stack_adjust));
}
asm_thumb_write_op16(as, OP_POP_RLIST_PC(as->push_reglist));
}
int asm_thumb_label_new(asm_thumb_t *as) {
return as->next_label++;
}
void asm_thumb_label_assign(asm_thumb_t *as, int label) {
if (as->pass > ASM_THUMB_PASS_1) {
assert(label < as->max_num_labels);
if (as->pass == ASM_THUMB_PASS_2) {
// assign label offset
assert(as->label_offsets[label] == -1);
as->label_offsets[label] = as->code_offset;
} else if (as->pass == ASM_THUMB_PASS_3) {
// ensure label offset has not changed from PASS_2 to PASS_3
//printf("l%d: (at %d=%ld)\n", label, as->label_offsets[label], as->code_offset);
assert(as->label_offsets[label] == as->code_offset);
}
}
}
// the i8 value will be zero extended into the r32 register!
void asm_thumb_mov_reg_i8(asm_thumb_t *as, uint rlo_dest, int i8) {
assert(rlo_dest < REG_R8);
// movs rlo_dest, #i8
asm_thumb_write_op16(as, 0x2000 | (rlo_dest << 8) | i8);
}
// if loading lo half, the i16 value will be zero extended into the r32 register!
void asm_thumb_mov_i16_to_reg(asm_thumb_t *as, int i16, uint reg_dest, bool load_hi_half) {
assert(reg_dest < REG_R15);
uint op;
if (load_hi_half) {
// movt reg_dest, #i16
op = 0xf2c0;
} else {
// movw reg_dest, #i16
op = 0xf240;
}
asm_thumb_write_op32(as, op | ((i16 >> 1) & 0x0400) | ((i16 >> 12) & 0xf), ((i16 << 4) & 0x7000) | (reg_dest << 8) | (i16 & 0xff));
}
void asm_thumb_mov_reg_i32(asm_thumb_t *as, uint reg_dest, machine_uint_t i32) {
// movw, movt does it in 8 bytes
// ldr [pc, #], dw does it in 6 bytes, but we might not reach to end of code for dw
asm_thumb_mov_i16_to_reg(as, i32, reg_dest, false);
asm_thumb_mov_i16_to_reg(as, i32 >> 16, reg_dest, true);
}
void asm_thumb_mov_reg_i32_optimised(asm_thumb_t *as, uint reg_dest, int i32) {
if (reg_dest < 8 && UNSIGNED_FIT8(i32)) {
asm_thumb_mov_reg_i8(as, reg_dest, i32);
} else if (UNSIGNED_FIT16(i32)) {
asm_thumb_mov_i16_to_reg(as, i32, reg_dest, false);
} else {
asm_thumb_mov_reg_i32(as, reg_dest, i32);
}
}
void asm_thumb_mov_reg_reg(asm_thumb_t *as, uint reg_dest, uint reg_src) {
uint op_lo;
if (reg_src < 8) {
op_lo = reg_src << 3;
} else {
op_lo = 0x40 | ((reg_src - 8) << 3);
}
if (reg_dest < 8) {
op_lo |= reg_dest;
} else {
op_lo |= 0x80 | (reg_dest - 8);
}
asm_thumb_write_op16(as, 0x4600 | op_lo);
}
#define OP_STR_TO_SP_OFFSET(rlo_dest, word_offset) (0x9000 | ((rlo_dest) << 8) | ((word_offset) & 0x00ff))
#define OP_LDR_FROM_SP_OFFSET(rlo_dest, word_offset) (0x9800 | ((rlo_dest) << 8) | ((word_offset) & 0x00ff))
void asm_thumb_mov_local_reg(asm_thumb_t *as, int local_num, uint rlo_src) {
assert(rlo_src < REG_R8);
int word_offset = as->num_locals - local_num - 1;
assert(as->pass < ASM_THUMB_PASS_3 || word_offset >= 0);
asm_thumb_write_op16(as, OP_STR_TO_SP_OFFSET(rlo_src, word_offset));
}
void asm_thumb_mov_reg_local(asm_thumb_t *as, uint rlo_dest, int local_num) {
assert(rlo_dest < REG_R8);
int word_offset = as->num_locals - local_num - 1;
assert(as->pass < ASM_THUMB_PASS_3 || word_offset >= 0);
asm_thumb_write_op16(as, OP_LDR_FROM_SP_OFFSET(rlo_dest, word_offset));
}
void asm_thumb_mov_reg_local_addr(asm_thumb_t *as, uint reg_dest, int local_num) {
assert(0);
// see format 12, load address
asm_thumb_write_op16(as, 0x0000);
}
#define OP_ADD_REG_REG_REG(rlo_dest, rlo_src_a, rlo_src_b) (0x1800 | ((rlo_src_b) << 6) | ((rlo_src_a) << 3) | (rlo_dest))
void asm_thumb_add_reg_reg_reg(asm_thumb_t *as, uint rlo_dest, uint rlo_src_a, uint rlo_src_b) {
asm_thumb_write_op16(as, OP_ADD_REG_REG_REG(rlo_dest, rlo_src_a, rlo_src_b));
}
#define OP_CMP_REG_REG(rlo_a, rlo_b) (0x4280 | ((rlo_b) << 3) | (rlo_a))
void asm_thumb_cmp_reg_reg(asm_thumb_t *as, uint rlo_a, uint rlo_b) {
asm_thumb_write_op16(as, OP_CMP_REG_REG(rlo_a, rlo_b));
}
void asm_thumb_ite_ge(asm_thumb_t *as) {
asm_thumb_write_op16(as, 0xbfac);
}
#define OP_B(byte_offset) (0xe000 | (((byte_offset) >> 1) & 0x07ff))
// this could be wrong, because it should have a range of +/- 16MiB...
#define OP_BW_HI(byte_offset) (0xf000 | (((byte_offset) >> 12) & 0x07ff))
#define OP_BW_LO(byte_offset) (0xb800 | (((byte_offset) >> 1) & 0x07ff))
void asm_thumb_b_label(asm_thumb_t *as, int label) {
if (as->pass > ASM_THUMB_PASS_1) {
int dest = as->label_offsets[label];
int rel = dest - as->code_offset;
rel -= 4; // account for instruction prefetch, PC is 4 bytes ahead of this instruction
if (dest >= 0 && rel <= -4) {
// is a backwards jump, so we know the size of the jump on the first pass
// calculate rel assuming 12 bit relative jump
if (SIGNED_FIT12(rel)) {
asm_thumb_write_op16(as, OP_B(rel));
} else {
goto large_jump;
}
} else {
// is a forwards jump, so need to assume it's large
large_jump:
asm_thumb_write_op32(as, OP_BW_HI(rel), OP_BW_LO(rel));
}
}
}
#define OP_CMP_REG_IMM(rlo, i8) (0x2800 | ((rlo) << 8) | (i8))
// all these bit arithmetics need coverage testing!
#define OP_BEQ(byte_offset) (0xd000 | (((byte_offset) >> 1) & 0x00ff))
#define OP_BEQW_HI(byte_offset) (0xf000 | (((byte_offset) >> 10) & 0x0400) | (((byte_offset) >> 14) & 0x003f))
#define OP_BEQW_LO(byte_offset) (0x8000 | ((byte_offset) & 0x2000) | (((byte_offset) >> 1) & 0x0fff))
void asm_thumb_cmp_reg_bz_label(asm_thumb_t *as, uint rlo, int label) {
assert(rlo < REG_R8);
// compare reg with 0
asm_thumb_write_op16(as, OP_CMP_REG_IMM(rlo, 0));
// branch if equal
if (as->pass > ASM_THUMB_PASS_1) {
int dest = as->label_offsets[label];
int rel = dest - as->code_offset;
rel -= 4; // account for instruction prefetch, PC is 4 bytes ahead of this instruction
if (dest >= 0 && rel <= -4) {
// is a backwards jump, so we know the size of the jump on the first pass
// calculate rel assuming 12 bit relative jump
if (SIGNED_FIT9(rel)) {
asm_thumb_write_op16(as, OP_BEQ(rel));
} else {
goto large_jump;
}
} else {
// is a forwards jump, so need to assume it's large
large_jump:
asm_thumb_write_op32(as, OP_BEQW_HI(rel), OP_BEQW_LO(rel));
}
}
}
#define OP_BLX(reg) (0x4780 | ((reg) << 3))
#define OP_SVC(arg) (0xdf00 | (arg))
#define OP_LDR_FROM_BASE_OFFSET(rlo_dest, rlo_base, word_offset) (0x6800 | (((word_offset) << 6) & 0x07c0) | ((rlo_base) << 3) | (rlo_dest))
void asm_thumb_bl_ind(asm_thumb_t *as, void *fun_ptr, uint fun_id, uint reg_temp) {
/* TODO make this use less bytes
uint rlo_base = REG_R3;
uint rlo_dest = REG_R7;
uint word_offset = 4;
asm_thumb_write_op16(as, 0x0000);
asm_thumb_write_op16(as, 0x6800 | (word_offset << 6) | (rlo_base << 3) | rlo_dest); // ldr rlo_dest, [rlo_base, #offset]
asm_thumb_write_op16(as, 0x4780 | (REG_R9 << 3)); // blx reg
*/
if (0) {
// load ptr to function into register using immediate, then branch
// not relocatable
asm_thumb_mov_reg_i32(as, reg_temp, (machine_uint_t)fun_ptr);
asm_thumb_write_op16(as, OP_BLX(reg_temp));
} else if (1) {
asm_thumb_write_op16(as, OP_LDR_FROM_BASE_OFFSET(reg_temp, REG_R7, fun_id));
asm_thumb_write_op16(as, OP_BLX(reg_temp));
} else {
// use SVC
asm_thumb_write_op16(as, OP_SVC(fun_id));
}
}
#define ASM_THUMB_PASS_1 (1)
#define ASM_THUMB_PASS_2 (2)
#define ASM_THUMB_PASS_3 (3)
#define REG_R0 (0)
#define REG_R1 (1)
#define REG_R2 (2)
#define REG_R3 (3)
#define REG_R4 (4)
#define REG_R5 (5)
#define REG_R6 (6)
#define REG_R7 (7)
#define REG_R8 (8)
#define REG_R9 (9)
#define REG_R10 (10)
#define REG_R11 (11)
#define REG_R12 (12)
#define REG_R13 (13)
#define REG_R14 (14)
#define REG_R15 (15)
#define REG_LR (REG_R14)
#define REG_RET REG_R0
#define REG_ARG_1 REG_R0
#define REG_ARG_2 REG_R1
#define REG_ARG_3 REG_R2
#define REG_ARG_4 REG_R3
typedef struct _asm_thumb_t asm_thumb_t;
asm_thumb_t *asm_thumb_new();
void asm_thumb_free(asm_thumb_t *as, bool free_code);
void asm_thumb_start_pass(asm_thumb_t *as, int pass);
void asm_thumb_end_pass(asm_thumb_t *as);
uint asm_thumb_get_code_size(asm_thumb_t *as);
void *asm_thumb_get_code(asm_thumb_t *as);
void asm_thumb_entry(asm_thumb_t *as, int num_locals);
void asm_thumb_exit(asm_thumb_t *as);
int asm_thumb_label_new(asm_thumb_t *as);
void asm_thumb_label_assign(asm_thumb_t *as, int label);
// argument order follows ARM, in general dest is first
void asm_thumb_mov_reg_i8(asm_thumb_t *as, uint rlo_dest, int i8_src);
void asm_thumb_mov_reg_i32(asm_thumb_t *as, uint reg_dest, machine_uint_t i32_src);
void asm_thumb_mov_reg_i32_optimised(asm_thumb_t *as, uint reg_dest, int i32_src);
void asm_thumb_mov_reg_reg(asm_thumb_t *as, uint reg_dest, uint reg_src);
void asm_thumb_mov_local_reg(asm_thumb_t *as, int local_num_dest, uint rlo_src);
void asm_thumb_mov_reg_local(asm_thumb_t *as, uint rlo_dest, int local_num);
void asm_thumb_mov_reg_local_addr(asm_thumb_t *as, uint reg_dest, int local_num);
void asm_thumb_add_reg_reg_reg(asm_thumb_t *as, uint rlo_dest, uint rlo_src_a, uint rlo_src_b);
void asm_thumb_cmp_reg_reg(asm_thumb_t *as, uint rlo_a, uint rlo_b);
void asm_thumb_ite_ge(asm_thumb_t *as);
void asm_thumb_b_label(asm_thumb_t *as, int label);
void asm_thumb_cmp_reg_bz_label(asm_thumb_t *as, uint rlo, int label);
void asm_thumb_bl_ind(asm_thumb_t *as, void *fun_ptr, uint fun_id, uint reg_temp);
This diff is collapsed.
#define ASM_X64_PASS_1 (1)
#define ASM_X64_PASS_2 (2)
#define ASM_X64_PASS_3 (3)
#define REG_RAX (0)
#define REG_RCX (1)
#define REG_RDX (2)
#define REG_RBX (3)
#define REG_RSP (4)
#define REG_RBP (5)
#define REG_RSI (6)
#define REG_RDI (7)
// condition codes, used for jcc and setcc (desipite their j-name!)
#define JCC_JB (0x2) // below, unsigned
#define JCC_JZ (0x4)
#define JCC_JE (0x4)
#define JCC_JNZ (0x5)
#define JCC_JNE (0x5)
#define JCC_JL (0xc) // less, signed
#define REG_RET REG_RAX
#define REG_ARG_1 REG_RDI
#define REG_ARG_2 REG_RSI
#define REG_ARG_3 REG_RDX
typedef struct _asm_x64_t asm_x64_t;
asm_x64_t* asm_x64_new();
void asm_x64_free(asm_x64_t* as, bool free_code);
void asm_x64_start_pass(asm_x64_t *as, int pass);
void asm_x64_end_pass(asm_x64_t *as);
uint asm_x64_get_code_size(asm_x64_t* as);
void* asm_x64_get_code(asm_x64_t* as);
void asm_x64_nop(asm_x64_t* as);
void asm_x64_push_r64(asm_x64_t* as, int src_r64);
void asm_x64_push_i32(asm_x64_t* as, int src_i32); // will be sign extended to 64 bits
void asm_x64_push_disp(asm_x64_t* as, int src_r32, int src_offset);
void asm_x64_pop_r64(asm_x64_t* as, int dest_r64);
void asm_x64_mov_r64_to_r64(asm_x64_t* as, int src_r64, int dest_r64);
void asm_x64_mov_r32_to_disp(asm_x64_t* as, int src_r32, int dest_r32, int dest_disp);
void asm_x64_mov_disp_to_r32(asm_x64_t* as, int src_r32, int src_disp, int dest_r32);
void asm_x64_mov_i32_to_r64(asm_x64_t* as, int src_i32, int dest_r64);
void asm_x64_mov_i64_to_r64(asm_x64_t* as, int64_t src_i64, int dest_r64);
void asm_x64_mov_i32_to_disp(asm_x64_t* as, int src_i32, int dest_r32, int dest_disp);
void asm_x64_mov_i64_to_r64_optimised(asm_x64_t *as, int64_t src_i64, int dest_r64);
void asm_x64_xor_r64_to_r64(asm_x64_t *as, int src_r64, int dest_r64);
void asm_x64_add_r64_to_r64(asm_x64_t* as, int src_r64, int dest_r64);
void asm_x64_add_i32_to_r32(asm_x64_t* as, int src_i32, int dest_r32);
void asm_x64_sub_r32_from_r32(asm_x64_t* as, int src_r32, int dest_r32);
void asm_x64_sub_i32_from_r32(asm_x64_t* as, int src_i32, int dest_r32);
void asm_x64_shl_r32_by_imm(asm_x64_t* as, int r32, int imm);
void asm_x64_shr_r32_by_imm(asm_x64_t* as, int r32, int imm);
void asm_x64_sar_r32_by_imm(asm_x64_t* as, int r32, int imm);
void asm_x64_cmp_r64_with_r64(asm_x64_t* as, int src_r64_a, int src_r64_b);
void asm_x64_cmp_r32_with_disp(asm_x64_t* as, int src_r32_a, int src_r32_b, int src_disp_b);
void asm_x64_cmp_disp_with_r32(asm_x64_t* as, int src_r32_a, int src_disp_a, int src_r32_b);
void asm_x64_cmp_i32_with_r32(asm_x64_t* as, int src_i32, int src_r32);
void asm_x64_test_r8_with_r8(asm_x64_t* as, int src_r64_a, int src_r64_b);
void asm_x64_setcc_r8(asm_x64_t* as, int jcc_type, int dest_r8);
int asm_x64_label_new(asm_x64_t* as);
void asm_x64_label_assign(asm_x64_t* as, int label);
void asm_x64_jmp_label(asm_x64_t* as, int label);
void asm_x64_jcc_label(asm_x64_t* as, int jcc_type, int label);
void asm_x64_entry(asm_x64_t* as, int num_locals);
void asm_x64_exit(asm_x64_t* as);
void asm_x64_push_arg(asm_x64_t* as, int src_arg_num);
void asm_x64_mov_arg_to_r32(asm_x64_t* as, int src_arg_num, int dest_r32);
void asm_x64_mov_r32_to_arg(asm_x64_t* as, int src_r32, int dest_arg_num);
void asm_x64_mov_local_to_r64(asm_x64_t* as, int src_local_num, int dest_r64);
void asm_x64_mov_r64_to_local(asm_x64_t* as, int src_r64, int dest_local_num);
void asm_x64_mov_local_addr_to_r64(asm_x64_t* as, int local_num, int dest_r64);
void asm_x64_push_local(asm_x64_t* as, int local_num);
void asm_x64_push_local_addr(asm_x64_t* as, int local_num, int temp_r32);
void asm_x64_call_ind(asm_x64_t* as, void* ptr, int temp_r32);
py/bc.c 0 → 100644
#include <stdint.h>
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <assert.h>
#include "misc.h"
#include "machine.h"
#include "runtime.h"
#include "bc.h"
#define DECODE_UINT do { unum = *ip++; if (unum > 127) { unum = ((unum & 0x3f) << 8) | (*ip++); } } while (0)
#define DECODE_QSTR do { qstr = *ip++; if (qstr > 127) { qstr = ((qstr & 0x3f) << 8) | (*ip++); } } while (0)
#define PUSH(val) *--sp = (val)
#define POP() (*sp++)
py_obj_t py_execute_byte_code(byte *code, uint len, py_obj_t *args, uint n_args) {
byte *ip = code;
py_obj_t stack[10];
py_obj_t *sp = &stack[10]; // stack grows down, sp points to top of stack
machine_uint_t unum;
machine_int_t snum;
qstr qstr;
py_obj_t obj1, obj2;
py_obj_t fast0 = NULL, fast1 = NULL, fast2 = NULL, fastn[4] = {NULL, NULL, NULL, NULL};
// init args
for (int i = 0; i < n_args; i++) {
if (i == 0) {
fast0 = args[0];
} else if (i == 1) {
fast1 = args[1];
} else if (i == 2) {
fast2 = args[2];
} else {
assert(i - 3 < 4);
fastn[i - 3] = args[i];
}
}
// execute byte code
for (;;) {
int op = *ip++;
switch (op) {
case PYBC_LOAD_CONST_FALSE:
PUSH(py_const_false);
break;
case PYBC_LOAD_CONST_NONE:
PUSH(py_const_none);
break;
case PYBC_LOAD_CONST_TRUE:
PUSH(py_const_true);
break;
case PYBC_LOAD_CONST_SMALL_INT:
snum = ip[0] | (ip[1] << 8);
if (snum & 0x8000) {
snum |= ~0xffff;
}
ip += 2;
PUSH((py_obj_t)(snum << 1 | 1));
break;
case PYBC_LOAD_CONST_ID:
DECODE_QSTR;
PUSH(rt_load_const_str(qstr)); // TODO
break;
case PYBC_LOAD_CONST_STRING:
DECODE_QSTR;
PUSH(rt_load_const_str(qstr));
break;
case PYBC_LOAD_FAST_0:
PUSH(fast0);
break;
case PYBC_LOAD_FAST_1:
PUSH(fast1);
break;
case PYBC_LOAD_FAST_2:
PUSH(fast2);
break;
case PYBC_LOAD_FAST_N:
DECODE_UINT;
PUSH(fastn[unum - 3]);
break;
case PYBC_LOAD_NAME:
DECODE_QSTR;
PUSH(rt_load_name(qstr));
break;
case PYBC_LOAD_GLOBAL:
DECODE_QSTR;
PUSH(rt_load_global(qstr));
break;
case PYBC_LOAD_ATTR:
DECODE_QSTR;
*sp = rt_load_attr(*sp, qstr);
break;
case PYBC_LOAD_METHOD:
DECODE_QSTR;
sp -= 1;
rt_load_method(sp[1], qstr, sp);
break;
case PYBC_LOAD_BUILD_CLASS:
PUSH(rt_load_build_class());
break;
case PYBC_STORE_FAST_0:
fast0 = POP();
break;
case PYBC_STORE_FAST_1:
fast1 = POP();
break;
case PYBC_STORE_FAST_2:
fast2 = POP();
break;
case PYBC_STORE_FAST_N:
DECODE_UINT;
fastn[unum - 3] = POP();
break;
case PYBC_STORE_NAME:
DECODE_QSTR;
rt_store_name(qstr, POP());
break;
case PYBC_STORE_SUBSCR:
rt_store_subscr(sp[1], sp[0], sp[2]);
sp += 3;
break;
case PYBC_DUP_TOP:
obj1 = *sp;
PUSH(obj1);
break;
case PYBC_DUP_TOP_TWO:
sp -= 2;
sp[0] = sp[2];
sp[1] = sp[3];
break;
case PYBC_POP_TOP:
++sp;
break;
case PYBC_ROT_THREE:
obj1 = sp[0];
sp[0] = sp[1];
sp[1] = sp[2];
sp[2] = obj1;
break;
case PYBC_JUMP:
DECODE_UINT;
ip = code + unum;
break;
case PYBC_POP_JUMP_IF_FALSE:
DECODE_UINT;
if (!rt_is_true(POP())) {
ip = code + unum;
}
break;
case PYBC_SETUP_LOOP:
DECODE_UINT;
break;
case PYBC_POP_BLOCK:
break;
case PYBC_BINARY_OP:
unum = *ip++;
obj2 = POP();
obj1 = *sp;
*sp = rt_binary_op(unum, obj1, obj2);
break;
case PYBC_COMPARE_OP:
unum = *ip++;
obj2 = POP();
obj1 = *sp;
*sp = rt_compare_op(unum, obj1, obj2);
break;
case PYBC_BUILD_LIST:
DECODE_UINT;
obj1 = rt_build_list(unum, sp);
sp += unum - 1;
*sp = obj1;
break;
case PYBC_BUILD_MAP:
DECODE_UINT;
PUSH(rt_build_map(unum));
break;
case PYBC_STORE_MAP:
sp += 2;
rt_store_map(sp[0], sp[-2], sp[-1]);
break;
case PYBC_BUILD_SET:
DECODE_UINT;
obj1 = rt_build_set(unum, sp);
sp += unum - 1;
*sp = obj1;
break;
case PYBC_MAKE_FUNCTION:
DECODE_UINT;
PUSH(rt_make_function_from_id(unum));
break;
case PYBC_CALL_FUNCTION:
DECODE_UINT;
assert((unum & 0xff00) == 0); // n_keyword
// switch on n_positional
if ((unum & 0xff) == 0) {
*sp = rt_call_function_0(*sp);
} else if ((unum & 0xff) == 1) {
obj1 = *sp++; // the single argument
*sp = rt_call_function_1(*sp, obj1);
} else if ((unum & 0xff) == 2) {
obj2 = *sp++; // the second argument
obj1 = *sp++; // the first argument
*sp = rt_call_function_2(*sp, obj1, obj2);
} else {
assert(0);
}
break;
case PYBC_CALL_METHOD:
DECODE_UINT;
assert((unum & 0xff00) == 0); // n_keyword
// switch on n_positional
if ((unum & 0xff) == 0) {
obj1 = *sp++; // the self object (or NULL)
*sp = rt_call_method_1(*sp, obj1);
} else if ((unum & 0xff) == 1) {
obj2 = *sp++; // the first argument
obj1 = *sp++; // the self object (or NULL)
*sp = rt_call_function_2(*sp, obj1, obj2);
} else {
assert(0);
}
break;
case PYBC_RETURN_VALUE:
return *sp;
default:
printf("code %p, offset %u, byte code 0x%02x not implemented\n", code, (uint)(ip - code), op);
assert(0);
return py_const_none;
}
}
}
py/bc.h 0 → 100644
#define PYBC_LOAD_CONST_FALSE (0x10)
#define PYBC_LOAD_CONST_NONE (0x11)
#define PYBC_LOAD_CONST_TRUE (0x12)
#define PYBC_LOAD_CONST_SMALL_INT (0x13) // int
#define PYBC_LOAD_CONST_INT (0x14) // qstr
#define PYBC_LOAD_CONST_DEC (0x15) // qstr
#define PYBC_LOAD_CONST_ID (0x16) // qstr
#define PYBC_LOAD_CONST_BYTES (0x17) // qstr
#define PYBC_LOAD_CONST_STRING (0x18) // qstr
#define PYBC_LOAD_FAST_0 (0x20)
#define PYBC_LOAD_FAST_1 (0x21)
#define PYBC_LOAD_FAST_2 (0x22)
#define PYBC_LOAD_FAST_N (0x23) // uint
#define PYBC_LOAD_NAME (0x24) // qstr
#define PYBC_LOAD_GLOBAL (0x25) // qstr
#define PYBC_LOAD_ATTR (0x26) // qstr
#define PYBC_LOAD_METHOD (0x27) // qstr
#define PYBC_LOAD_BUILD_CLASS (0x28)
#define PYBC_STORE_FAST_0 (0x30)
#define PYBC_STORE_FAST_1 (0x31)
#define PYBC_STORE_FAST_2 (0x32)
#define PYBC_STORE_FAST_N (0x33) // uint
#define PYBC_STORE_NAME (0x34) // qstr
#define PYBC_STORE_GLOBAL (0x35) // qstr
#define PYBC_STORE_ATTR (0x36) // qstr
#define PYBC_STORE_LOCALS (0x37)
#define PYBC_STORE_SUBSCR (0x38)
#define PYBC_DELETE_FAST_N (0x39) // uint
#define PYBC_DELETE_NAME (0x3a) // qstr
#define PYBC_DELETE_GLOBAL (0x3b) // qstr
#define PYBC_DELETE_DEREF (0x3c) // qstr
#define PYBC_DELETE_ATTR (0x3d) // qstr
#define PYBC_DELETE_SUBSCR (0x3e)
#define PYBC_DUP_TOP (0x40)
#define PYBC_DUP_TOP_TWO (0x41)
#define PYBC_POP_TOP (0x42)
#define PYBC_ROT_TWO (0x43)
#define PYBC_ROT_THREE (0x44)
#define PYBC_JUMP (0x45) // pos
#define PYBC_POP_JUMP_IF_TRUE (0x46) // pos
#define PYBC_POP_JUMP_IF_FALSE (0x47) // pos
#define PYBC_JUMP_IF_TRUE_OR_POP (0x48) // pos
#define PYBC_JUMP_IF_FALSE_OR_POP (0x49) // pos
#define PYBC_SETUP_LOOP (0x4a) // pos
#define PYBC_BREAK_LOOP (0x4b) // pos
#define PYBC_CONTINUE_LOOP (0x4c) // pos
#define PYBC_SETUP_WITH (0x4d) // pos
#define PYBC_WITH_CLEANUP (0x4e)
#define PYBC_SETUP_EXCEPT (0x4f) // pos
#define PYBC_SETUP_FINALLY (0x50) // pos
#define PYBC_END_FINALLY (0x51)
#define PYBC_GET_ITER (0x52)
#define PYBC_FOR_ITER (0x53) // pos
#define PYBC_POP_BLOCK (0x54)
#define PYBC_POP_EXCEPT (0x55)
#define PYBC_UNARY_OP (0x60) // byte
#define PYBC_BINARY_OP (0x61) // byte
#define PYBC_COMPARE_OP (0x62) // byte
#define PYBC_BUILD_TUPLE (0x70) // uint
#define PYBC_BUILD_LIST (0x71) // uint
#define PYBC_LIST_APPEND (0x72) // uint
#define PYBC_BUILD_MAP (0x73) // uint
#define PYBC_STORE_MAP (0x74)
#define PYBC_MAP_ADD (0x75) // uint
#define PYBC_BUILD_SET (0x76) // uint
#define PYBC_SET_ADD (0x77) // uint
#define PYBC_BUILD_SLICE (0x78) // uint
#define PYBC_UNPACK_SEQUENCE (0x79) // uint
#define PYBC_UNPACK_EX (0x7a) // uint
#define PYBC_RETURN_VALUE (0x80)
#define PYBC_RAISE_VARARGS (0x81) // uint
#define PYBC_YIELD_VALUE (0x82)
#define PYBC_YIELD_FROM (0x83)
#define PYBC_MAKE_FUNCTION (0x90) // uint
#define PYBC_MAKE_CLOSURE (0x91) // uint?
#define PYBC_CALL_FUNCTION (0x92) // uint
#define PYBC_CALL_FUNCTION_VAR (0x93) // uint
#define PYBC_CALL_FUNCTION_KW (0x94) // uint
#define PYBC_CALL_FUNCTION_VAR_KW (0x95) // uint
#define PYBC_CALL_METHOD (0x96) // uint
#define PYBC_CALL_METHOD_VAR (0x97) // uint
#define PYBC_CALL_METHOD_KW (0x98) // uint
#define PYBC_CALL_METHOD_VAR_KW (0x99) // uint
#define PYBC_IMPORT_NAME (0xe0)
#define PYBC_IMPORT_FROM (0xe1)
#define PYBC_IMPORT_STAR (0xe2)
py_obj_t py_execute_byte_code(byte *code, uint len, py_obj_t *args, uint n_args);
This diff is collapsed.
void py_compile(py_parse_node_t pn);
py/emit.h 0 → 100644
//#define EMIT_DO_CPY
#define EMIT_DO_BC
//#define EMIT_DO_X64
//#define EMIT_DO_THUMB
/* Notes on passes:
* We don't know exactly the opcodes in pass 1 because they depend on the
* closing over of variables (LOAD_CLOSURE, BUILD_TUPLE, MAKE_CLOSURE), which
* depends on determining the scope of variables in each function, and this
* is not known until the end of pass 1.
* As a consequence, we don't know the maximum stack size until the end of pass 2.
* This is problematic for some emitters (x64) since they need to know the maximum
* stack size to compile the entry to the function, and this effects code size.
*/
typedef enum {
PASS_1 = 1, // work out id's and their kind, and number of labels
PASS_2 = 2, // work out stack size and code size and label offsets
PASS_3 = 3, // emit code
} pass_kind_t;
typedef struct _emitter_t emitter_t;
void emit_common_declare_global(pass_kind_t pass, scope_t *scope, qstr qstr);
void emit_common_declare_nonlocal(pass_kind_t pass, scope_t *scope, qstr qstr);
void emit_common_load_id(pass_kind_t pass, scope_t *scope, qstr qstr___class__, emitter_t *emit, qstr qstr);
void emit_common_store_id(pass_kind_t pass, scope_t *scope, emitter_t *emit, qstr qstr);
void emit_common_delete_id(pass_kind_t pass, scope_t *scope, emitter_t *emit, qstr qstr);
emitter_t *emit_new();
void emit_set_native_types(emitter_t *emit, bool do_native_types);
void emit_start_pass(emitter_t *emit, pass_kind_t pass, scope_t *scope);
void emit_end_pass(emitter_t *emit);
bool emit_last_emit_was_return_value(emitter_t *emit);
int emit_get_stack_size(emitter_t *emit);
void emit_set_stack_size(emitter_t *emit, int size);
int emit_label_new(emitter_t *emit);
void emit_label_assign(emitter_t *emit, int l);
void emit_import_name(emitter_t *emit, qstr qstr);
void emit_import_from(emitter_t *emit, qstr qstr);
void emit_import_star(emitter_t *emit);
void emit_load_const_tok(emitter_t *emit, py_token_kind_t tok);
void emit_load_const_small_int(emitter_t *emit, int arg);
void emit_load_const_int(emitter_t *emit, qstr qstr);
void emit_load_const_dec(emitter_t *emit, qstr qstr);
void emit_load_const_id(emitter_t *emit, qstr qstr);
void emit_load_const_str(emitter_t *emit, qstr qstr, bool bytes);
void emit_load_const_verbatim_start(emitter_t *emit);
void emit_load_const_verbatim_int(emitter_t *emit, int val);
void emit_load_const_verbatim_str(emitter_t *emit, const char *str);
void emit_load_const_verbatim_strn(emitter_t *emit, const char *str, int len);
void emit_load_const_verbatim_quoted_str(emitter_t *emit, qstr qstr, bool bytes);
void emit_load_const_verbatim_end(emitter_t *emit);
void emit_load_fast(emitter_t *emit, qstr qstr, int local_num);
void emit_load_name(emitter_t *emit, qstr qstr);
void emit_load_global(emitter_t *emit, qstr qstr);
void emit_load_deref(emitter_t *emit, qstr qstr);
void emit_load_closure(emitter_t *emit, qstr qstr);
void emit_load_attr(emitter_t *emit, qstr qstr);
void emit_load_method(emitter_t *emit, qstr qstr);
void emit_load_build_class(emitter_t *emit);
void emit_store_fast(emitter_t *emit, qstr qstr, int local_num);
void emit_store_name(emitter_t *emit, qstr qstr);
void emit_store_global(emitter_t *emit, qstr qstr);
void emit_store_deref(emitter_t *emit, qstr qstr);
void emit_store_attr(emitter_t *emit, qstr qstr);
void emit_store_locals(emitter_t *emit);
void emit_store_subscr(emitter_t *emit);
void emit_delete_fast(emitter_t *emit, qstr qstr, int local_num);
void emit_delete_name(emitter_t *emit, qstr qstr);
void emit_delete_global(emitter_t *emit, qstr qstr);
void emit_delete_deref(emitter_t *emit, qstr qstr);
void emit_delete_attr(emitter_t *emit, qstr qstr);
void emit_delete_subscr(emitter_t *emit);
void emit_dup_top(emitter_t *emit);
void emit_dup_top_two(emitter_t *emit);
void emit_pop_top(emitter_t *emit);
void emit_rot_two(emitter_t *emit);
void emit_rot_three(emitter_t *emit);
void emit_jump(emitter_t *emit, int label);
void emit_pop_jump_if_true(emitter_t *emit, int label);
void emit_pop_jump_if_false(emitter_t *emit, int label);
void emit_jump_if_true_or_pop(emitter_t *emit, int label);
void emit_jump_if_false_or_pop(emitter_t *emit, int label);
void emit_setup_loop(emitter_t *emit, int label);
void emit_break_loop(emitter_t *emit, int label);
void emit_continue_loop(emitter_t *emit, int label);
void emit_setup_with(emitter_t *emit, int label);
void emit_with_cleanup(emitter_t *emit);
void emit_setup_except(emitter_t *emit, int label);
void emit_setup_finally(emitter_t *emit, int label);
void emit_end_finally(emitter_t *emit);
void emit_get_iter(emitter_t *emit); // tos = getiter(tos)
void emit_for_iter(emitter_t *emit, int label);
void emit_for_iter_end(emitter_t *emit);
void emit_pop_block(emitter_t *emit);
void emit_pop_except(emitter_t *emit);
void emit_unary_op(emitter_t *emit, rt_unary_op_t op);
void emit_binary_op(emitter_t *emit, rt_binary_op_t op);
void emit_compare_op(emitter_t *emit, rt_compare_op_t op);
void emit_build_tuple(emitter_t *emit, int n_args);
void emit_build_list(emitter_t *emit, int n_args);
void emit_list_append(emitter_t *emit, int list_stack_index);
void emit_build_map(emitter_t *emit, int n_args);
void emit_store_map(emitter_t *emit);
void emit_map_add(emitter_t *emit, int map_stack_index);
void emit_build_set(emitter_t *emit, int n_args);
void emit_set_add(emitter_t *emit, int set_stack_index);
void emit_build_slice(emitter_t *emit, int n_args);
void emit_unpack_sequence(emitter_t *emit, int n_args);
void emit_unpack_ex(emitter_t *emit, int n_left, int n_right);
void emit_make_function(emitter_t *emit, scope_t *scope, int n_dict_params, int n_default_params);
void emit_make_closure(emitter_t *emit, scope_t *scope, int n_dict_params, int n_default_params);
void emit_call_function(emitter_t *emit, int n_positional, int n_keyword, bool have_star_arg, bool have_dbl_star_arg);
void emit_call_method(emitter_t *emit, int n_positional, int n_keyword, bool have_star_arg, bool have_dbl_star_arg);
void emit_return_value(emitter_t *emit);
void emit_raise_varargs(emitter_t *emit, int n_args);
void emit_yield_value(emitter_t *emit);
void emit_yield_from(emitter_t *emit);
#include <unistd.h>
#include <stdlib.h>
#include <stdint.h>
#include <stdio.h>
#include <string.h>
#include <assert.h>
#include "misc.h"
#include "lexer.h"
#include "machine.h"
#include "parse.h"
#include "compile.h"
#include "scope.h"
#include "runtime.h"
#include "emit.h"
#include "bc.h"
#ifdef EMIT_DO_BC
struct _emitter_t {
int pass;
int next_label;
int stack_size;
bool last_emit_was_return_value;
scope_t *scope;
int max_num_labels;
uint *label_offsets;
uint code_offset;
uint code_size;
byte *code_base;
byte dummy_data[8];
};
emitter_t *emit_new() {
emitter_t *emit = m_new(emitter_t, 1);
emit->max_num_labels = 0;
emit->label_offsets = NULL;
emit->code_offset = 0;
emit->code_size = 0;
emit->code_base = NULL;
return emit;
}
uint emit_get_code_size(emitter_t* emit) {
return emit->code_size;
}
void* emit_get_code(emitter_t* emit) {
return emit->code_base;
}
void emit_start_pass(emitter_t *emit, pass_kind_t pass, scope_t *scope) {
emit->pass = pass;
emit->next_label = 1;
emit->stack_size = 0;
emit->last_emit_was_return_value = false;
emit->scope = scope;
if (pass == PASS_1) {
scope->unique_code_id = rt_get_new_unique_code_id();
} else if (pass > PASS_1) {
if (emit->label_offsets == NULL) {
emit->label_offsets = m_new(uint, emit->max_num_labels);
}
if (pass == PASS_2) {
memset(emit->label_offsets, -1, emit->max_num_labels * sizeof(uint));
}
}
emit->code_offset = 0;
}
void emit_end_pass(emitter_t *emit) {
// check stack is back to zero size
if (emit->stack_size != 0) {
printf("ERROR: stack size not back to zero; got %d\n", emit->stack_size);
}
if (emit->pass == PASS_1) {
// calculate number of labels need
if (emit->next_label > emit->max_num_labels) {
emit->max_num_labels = emit->next_label;
}
} else if (emit->pass == PASS_2) {
// calculate size of code in bytes
emit->code_size = emit->code_offset;
emit->code_base = m_new(byte, emit->code_size);
printf("code_size: %u\n", emit->code_size);
} else if (emit->pass == PASS_3) {
rt_assign_byte_code(emit->scope->unique_code_id, emit->code_base, emit->code_size, emit->scope->num_params);
}
}
// all functions must go through this one to emit bytes
static byte* emit_get_cur_to_write_bytes(emitter_t* emit, int num_bytes_to_write) {
//printf("emit %d\n", num_bytes_to_write);
if (emit->pass < PASS_3) {
emit->code_offset += num_bytes_to_write;
return emit->dummy_data;
} else {
assert(emit->code_offset + num_bytes_to_write <= emit->code_size);
byte *c = emit->code_base + emit->code_offset;
emit->code_offset += num_bytes_to_write;
return c;
}
}
static void emit_write_byte_1(emitter_t* emit, byte b1) {
byte* c = emit_get_cur_to_write_bytes(emit, 1);
c[0] = b1;
}
static void emit_write_byte_1_byte(emitter_t* emit, byte b1, uint b2) {
assert((b2 & (~0xff)) == 0);
byte* c = emit_get_cur_to_write_bytes(emit, 2);
c[0] = b1;
c[1] = b2;
}
static void emit_write_byte_1_int(emitter_t* emit, byte b1, int num) {
assert((num & (~0x7fff)) == 0 || (num & (~0x7fff)) == (~0x7fff));
byte* c = emit_get_cur_to_write_bytes(emit, 3);
c[0] = b1;
c[1] = num;
c[2] = num >> 8;
}
static void emit_write_byte_1_uint(emitter_t* emit, byte b1, uint num) {
if (num <= 127) { // fits in 0x7f
// fit argument in single byte
byte* c = emit_get_cur_to_write_bytes(emit, 2);
c[0] = b1;
c[1] = num;
} else if (num <= 16383) { // fits in 0x3fff
// fit argument in two bytes
byte* c = emit_get_cur_to_write_bytes(emit, 3);
c[0] = b1;
c[1] = (num >> 8) | 0x80;
c[2] = num;
} else {
// larger numbers not implemented/supported
assert(0);
}
}
static void emit_write_byte_1_qstr(emitter_t* emit, byte b1, qstr qstr) {
emit_write_byte_1_uint(emit, b1, qstr);
}
static void emit_write_byte_1_label(emitter_t* emit, byte b1, int label) {
uint code_offset;
if (emit->pass < PASS_3) {
code_offset = 0;
} else {
code_offset = emit->label_offsets[label];
}
emit_write_byte_1_uint(emit, b1, code_offset);
}
bool emit_last_emit_was_return_value(emitter_t *emit) {
return emit->last_emit_was_return_value;
}
int emit_get_stack_size(emitter_t *emit) {
return emit->stack_size;
}
void emit_set_stack_size(emitter_t *emit, int size) {
if (emit->pass > PASS_1) {
emit->stack_size = size;
}
}
static void emit_pre(emitter_t *emit, int stack_size_delta) {
if (emit->pass > PASS_1) {
emit->stack_size += stack_size_delta;
if (emit->stack_size > emit->scope->stack_size) {
emit->scope->stack_size = emit->stack_size;
}
}
emit->last_emit_was_return_value = false;
}
int emit_label_new(emitter_t *emit) {
return emit->next_label++;
}
void emit_label_assign(emitter_t *emit, int l) {
emit_pre(emit, 0);
if (emit->pass > PASS_1) {
assert(l < emit->max_num_labels);
if (emit->pass == PASS_2) {
// assign label offset
assert(emit->label_offsets[l] == -1);
emit->label_offsets[l] = emit->code_offset;
} else if (emit->pass == PASS_3) {
// ensure label offset has not changed from PASS_2 to PASS_3
assert(emit->label_offsets[l] == emit->code_offset);
//printf("l%d: (at %d)\n", l, emit->code_offset);
}
}
}
void emit_import_name(emitter_t *emit, qstr qstr) {
emit_pre(emit, -1);
emit_write_byte_1_qstr(emit, PYBC_IMPORT_NAME, qstr);
}
void emit_import_from(emitter_t *emit, qstr qstr) {
emit_pre(emit, 1);
emit_write_byte_1_qstr(emit, PYBC_IMPORT_FROM, qstr);
}
void emit_import_star(emitter_t *emit) {
emit_pre(emit, -1);
emit_write_byte_1(emit, PYBC_IMPORT_STAR);
}
void emit_load_const_tok(emitter_t *emit, py_token_kind_t tok) {
emit_pre(emit, 1);
switch (tok) {
case PY_TOKEN_KW_FALSE: emit_write_byte_1(emit, PYBC_LOAD_CONST_FALSE); break;
case PY_TOKEN_KW_NONE: emit_write_byte_1(emit, PYBC_LOAD_CONST_NONE); break;
case PY_TOKEN_KW_TRUE: emit_write_byte_1(emit, PYBC_LOAD_CONST_TRUE); break;
default: assert(0);
}
}
void emit_load_const_small_int(emitter_t *emit, int arg) {
emit_pre(emit, 1);
emit_write_byte_1_int(emit, PYBC_LOAD_CONST_SMALL_INT, arg);
}
void emit_load_const_int(emitter_t *emit, qstr qstr) {
emit_pre(emit, 1);
emit_write_byte_1_qstr(emit, PYBC_LOAD_CONST_INT, qstr);
}
void emit_load_const_dec(emitter_t *emit, qstr qstr) {
emit_pre(emit, 1);
emit_write_byte_1_qstr(emit, PYBC_LOAD_CONST_DEC, qstr);
}
void emit_load_const_id(emitter_t *emit, qstr qstr) {
emit_pre(emit, 1);
emit_write_byte_1_qstr(emit, PYBC_LOAD_CONST_ID, qstr);
}
void emit_load_const_str(emitter_t *emit, qstr qstr, bool bytes) {
emit_pre(emit, 1);
if (bytes) {
emit_write_byte_1_qstr(emit, PYBC_LOAD_CONST_BYTES, qstr);
} else {
emit_write_byte_1_qstr(emit, PYBC_LOAD_CONST_STRING, qstr);
}
}
void emit_load_const_verbatim_start(emitter_t *emit) {
emit_pre(emit, 1);
assert(0);
}
void emit_load_const_verbatim_int(emitter_t *emit, int val) {
assert(0);
}
void emit_load_const_verbatim_str(emitter_t *emit, const char *str) {
assert(0);
}
void emit_load_const_verbatim_strn(emitter_t *emit, const char *str, int len) {
assert(0);
}
void emit_load_const_verbatim_quoted_str(emitter_t *emit, qstr qstr, bool bytes) {
assert(0);
}
void emit_load_const_verbatim_end(emitter_t *emit) {
assert(0);
}
void emit_load_fast(emitter_t *emit, qstr qstr, int local_num) {
assert(local_num >= 0);
emit_pre(emit, 1);
switch (local_num) {
case 0: emit_write_byte_1(emit, PYBC_LOAD_FAST_0); break;
case 1: emit_write_byte_1(emit, PYBC_LOAD_FAST_1); break;
case 2: emit_write_byte_1(emit, PYBC_LOAD_FAST_2); break;
default: emit_write_byte_1_uint(emit, PYBC_LOAD_FAST_N, local_num); break;
}
}
void emit_load_name(emitter_t *emit, qstr qstr) {
emit_pre(emit, 1);
emit_write_byte_1_qstr(emit, PYBC_LOAD_NAME, qstr);
}
void emit_load_global(emitter_t *emit, qstr qstr) {
emit_pre(emit, 1);
emit_write_byte_1_qstr(emit, PYBC_LOAD_GLOBAL, qstr);
}
void emit_load_deref(emitter_t *emit, qstr qstr) {
emit_pre(emit, 1);
assert(0);
}
void emit_load_closure(emitter_t *emit, qstr qstr) {
emit_pre(emit, 1);
assert(0);
}
void emit_load_attr(emitter_t *emit, qstr qstr) {
emit_pre(emit, 0);
emit_write_byte_1_qstr(emit, PYBC_LOAD_ATTR, qstr);
}
void emit_load_method(emitter_t *emit, qstr qstr) {
emit_pre(emit, 0);
emit_write_byte_1_qstr(emit, PYBC_LOAD_METHOD, qstr);
}
void emit_load_build_class(emitter_t *emit) {
emit_pre(emit, 1);
emit_write_byte_1(emit, PYBC_LOAD_BUILD_CLASS);
}
void emit_store_fast(emitter_t *emit, qstr qstr, int local_num) {
assert(local_num >= 0);
emit_pre(emit, -1);
switch (local_num) {
case 0: emit_write_byte_1(emit, PYBC_STORE_FAST_0); break;
case 1: emit_write_byte_1(emit, PYBC_STORE_FAST_1); break;
case 2: emit_write_byte_1(emit, PYBC_STORE_FAST_2); break;
default: emit_write_byte_1_uint(emit, PYBC_STORE_FAST_N, local_num); break;
}
}
void emit_store_name(emitter_t *emit, qstr qstr) {
emit_pre(emit, -1);
emit_write_byte_1_qstr(emit, PYBC_STORE_NAME, qstr);
}
void emit_store_global(emitter_t *emit, qstr qstr) {
emit_pre(emit, -1);
emit_write_byte_1_qstr(emit, PYBC_STORE_GLOBAL, qstr);
}
void emit_store_deref(emitter_t *emit, qstr qstr) {
emit_pre(emit, -1);
assert(0);
}
void emit_store_attr(emitter_t *emit, qstr qstr) {
emit_pre(emit, -2);
emit_write_byte_1_qstr(emit, PYBC_STORE_ATTR, qstr);
}
void emit_store_locals(emitter_t *emit) {
emit_pre(emit, -1);
emit_write_byte_1(emit, PYBC_STORE_LOCALS);
}
void emit_store_subscr(emitter_t *emit) {
emit_pre(emit, -3);
emit_write_byte_1(emit, PYBC_STORE_SUBSCR);
}
void emit_delete_fast(emitter_t *emit, qstr qstr, int local_num) {
assert(local_num >= 0);
emit_pre(emit, 0);
emit_write_byte_1_uint(emit, PYBC_DELETE_FAST_N, local_num);
}
void emit_delete_name(emitter_t *emit, qstr qstr) {
emit_pre(emit, 0);
emit_write_byte_1_qstr(emit, PYBC_DELETE_NAME, qstr);
}
void emit_delete_global(emitter_t *emit, qstr qstr) {
emit_pre(emit, 0);
emit_write_byte_1_qstr(emit, PYBC_DELETE_GLOBAL, qstr);
}
void emit_delete_deref(emitter_t *emit, qstr qstr) {
emit_pre(emit, 0);
emit_write_byte_1_qstr(emit, PYBC_DELETE_DEREF, qstr);
}
void emit_delete_attr(emitter_t *emit, qstr qstr) {
emit_pre(emit, -1);
emit_write_byte_1_qstr(emit, PYBC_DELETE_ATTR, qstr);
}
void emit_delete_subscr(emitter_t *emit) {
emit_pre(emit, -2);
emit_write_byte_1(emit, PYBC_DELETE_SUBSCR);
}
void emit_dup_top(emitter_t *emit) {
emit_pre(emit, 1);
emit_write_byte_1(emit, PYBC_DUP_TOP);
}
void emit_dup_top_two(emitter_t *emit) {
emit_pre(emit, 2);
emit_write_byte_1(emit, PYBC_DUP_TOP_TWO);
}
void emit_pop_top(emitter_t *emit) {
emit_pre(emit, -1);
emit_write_byte_1(emit, PYBC_POP_TOP);
}
void emit_rot_two(emitter_t *emit) {
emit_pre(emit, 0);
emit_write_byte_1(emit, PYBC_ROT_TWO);
}
void emit_rot_three(emitter_t *emit) {
emit_pre(emit, 0);
emit_write_byte_1(emit, PYBC_ROT_THREE);
}
void emit_jump(emitter_t *emit, int label) {
emit_pre(emit, 0);
emit_write_byte_1_label(emit, PYBC_JUMP, label);
}
void emit_pop_jump_if_true(emitter_t *emit, int label) {
emit_pre(emit, -1);
emit_write_byte_1_label(emit, PYBC_POP_JUMP_IF_TRUE, label);
}
void emit_pop_jump_if_false(emitter_t *emit, int label) {
emit_pre(emit, -1);
emit_write_byte_1_label(emit, PYBC_POP_JUMP_IF_FALSE, label);
}
void emit_jump_if_true_or_pop(emitter_t *emit, int label) {
emit_pre(emit, -1);
emit_write_byte_1_label(emit, PYBC_JUMP_IF_TRUE_OR_POP, label);
}
void emit_jump_if_false_or_pop(emitter_t *emit, int label) {
emit_pre(emit, -1);
emit_write_byte_1_label(emit, PYBC_JUMP_IF_FALSE_OR_POP, label);
}
void emit_setup_loop(emitter_t *emit, int label) {
emit_pre(emit, 0);
emit_write_byte_1_label(emit, PYBC_SETUP_LOOP, label);
}
void emit_break_loop(emitter_t *emit, int label) {
emit_pre(emit, 0);
emit_write_byte_1_label(emit, PYBC_BREAK_LOOP, label);
}
void emit_continue_loop(emitter_t *emit, int label) {
emit_pre(emit, 0);
emit_write_byte_1_label(emit, PYBC_CONTINUE_LOOP, label);
}
void emit_setup_with(emitter_t *emit, int label) {
emit_pre(emit, 7);
emit_write_byte_1_label(emit, PYBC_SETUP_WITH, label);
}
void emit_with_cleanup(emitter_t *emit) {
emit_pre(emit, -7);
emit_write_byte_1(emit, PYBC_WITH_CLEANUP);
}
void emit_setup_except(emitter_t *emit, int label) {
emit_pre(emit, 6);
emit_write_byte_1_label(emit, PYBC_SETUP_EXCEPT, label);
}
void emit_setup_finally(emitter_t *emit, int label) {
emit_pre(emit, 6);
emit_write_byte_1_label(emit, PYBC_SETUP_FINALLY, label);
}
void emit_end_finally(emitter_t *emit) {
emit_pre(emit, -1);
emit_write_byte_1(emit, PYBC_END_FINALLY);
}
void emit_get_iter(emitter_t *emit) {
emit_pre(emit, 0);
emit_write_byte_1(emit, PYBC_GET_ITER);
}
void emit_for_iter(emitter_t *emit, int label) {
emit_pre(emit, 1);
emit_write_byte_1_label(emit, PYBC_FOR_ITER, label);
}
void emit_for_iter_end(emitter_t *emit) {
emit_pre(emit, -1);
}
void emit_pop_block(emitter_t *emit) {
emit_pre(emit, 0);
emit_write_byte_1(emit, PYBC_POP_BLOCK);
}
void emit_pop_except(emitter_t *emit) {
emit_pre(emit, 0);
emit_write_byte_1(emit, PYBC_POP_EXCEPT);
}
void emit_unary_op(emitter_t *emit, rt_unary_op_t op) {
emit_pre(emit, 0);
emit_write_byte_1_byte(emit, PYBC_UNARY_OP, op);
}
void emit_binary_op(emitter_t *emit, rt_binary_op_t op) {
emit_pre(emit, -1);
emit_write_byte_1_byte(emit, PYBC_BINARY_OP, op);
}
void emit_compare_op(emitter_t *emit, rt_compare_op_t op) {
emit_pre(emit, -1);
emit_write_byte_1_byte(emit, PYBC_COMPARE_OP, op);
}
void emit_build_tuple(emitter_t *emit, int n_args) {
assert(n_args >= 0);
emit_pre(emit, 1 - n_args);
emit_write_byte_1_uint(emit, PYBC_BUILD_TUPLE, n_args);
}
void emit_build_list(emitter_t *emit, int n_args) {
assert(n_args >= 0);
emit_pre(emit, 1 - n_args);
emit_write_byte_1_uint(emit, PYBC_BUILD_LIST, n_args);
}
void emit_list_append(emitter_t *emit, int list_stack_index) {
assert(list_stack_index >= 0);
emit_pre(emit, -1);
emit_write_byte_1_uint(emit, PYBC_LIST_APPEND, list_stack_index);
}
void emit_build_map(emitter_t *emit, int n_args) {
assert(n_args >= 0);
emit_pre(emit, 1);
emit_write_byte_1_uint(emit, PYBC_BUILD_MAP, n_args);
}
void emit_store_map(emitter_t *emit) {
emit_pre(emit, -2);
emit_write_byte_1(emit, PYBC_STORE_MAP);
}
void emit_map_add(emitter_t *emit, int map_stack_index) {
assert(map_stack_index >= 0);
emit_pre(emit, -2);
emit_write_byte_1_uint(emit, PYBC_MAP_ADD, map_stack_index);
}
void emit_build_set(emitter_t *emit, int n_args) {
assert(n_args >= 0);
emit_pre(emit, 1 - n_args);
emit_write_byte_1_uint(emit, PYBC_BUILD_SET, n_args);
}
void emit_set_add(emitter_t *emit, int set_stack_index) {
assert(set_stack_index >= 0);
emit_pre(emit, -1);
emit_write_byte_1_uint(emit, PYBC_SET_ADD, set_stack_index);
}
void emit_build_slice(emitter_t *emit, int n_args) {
assert(n_args >= 0);
emit_pre(emit, 1 - n_args);
emit_write_byte_1_uint(emit, PYBC_BUILD_SLICE, n_args);
}
void emit_unpack_sequence(emitter_t *emit, int n_args) {
assert(n_args >= 0);
emit_pre(emit, -1 + n_args);
emit_write_byte_1_uint(emit, PYBC_UNPACK_SEQUENCE, n_args);
}
void emit_unpack_ex(emitter_t *emit, int n_left, int n_right) {
assert(n_left >=0 && n_right >= 0);
emit_pre(emit, -1 + n_left + n_right + 1);
emit_write_byte_1_uint(emit, PYBC_UNPACK_EX, n_left | (n_right << 8));
}
void emit_make_function(emitter_t *emit, scope_t *scope, int n_dict_params, int n_default_params) {
assert(n_default_params == 0 && n_dict_params == 0);
emit_pre(emit, 1);
emit_write_byte_1_uint(emit, PYBC_MAKE_FUNCTION, scope->unique_code_id);
}
void emit_make_closure(emitter_t *emit, scope_t *scope, int n_dict_params, int n_default_params) {
assert(0);
emit_pre(emit, -2 - n_default_params - 2 * n_dict_params);
if (emit->pass == PASS_3) {
printf("MAKE_CLOSURE %d\n", (n_dict_params << 8) | n_default_params);
}
}
void emit_call_function(emitter_t *emit, int n_positional, int n_keyword, bool have_star_arg, bool have_dbl_star_arg) {
int s = 0;
if (have_star_arg) {
s += 1;
}
if (have_dbl_star_arg) {
s += 1;
}
emit_pre(emit, -n_positional - 2 * n_keyword - s);
int op;
if (have_star_arg) {
if (have_dbl_star_arg) {
op = PYBC_CALL_FUNCTION_VAR_KW;
} else {
op = PYBC_CALL_FUNCTION_VAR;
}
} else {
if (have_dbl_star_arg) {
op = PYBC_CALL_FUNCTION_KW;
} else {
op = PYBC_CALL_FUNCTION;
}
}
emit_write_byte_1_uint(emit, op, (n_keyword << 8) | n_positional); // TODO make it 2 separate uints
}
void emit_call_method(emitter_t *emit, int n_positional, int n_keyword, bool have_star_arg, bool have_dbl_star_arg) {
int s = 0;
if (have_star_arg) {
s += 1;
}
if (have_dbl_star_arg) {
s += 1;
}
emit_pre(emit, -n_positional - 2 * n_keyword - s);
int op;
if (have_star_arg) {
if (have_dbl_star_arg) {
op = PYBC_CALL_METHOD_VAR_KW;
} else {
op = PYBC_CALL_METHOD_VAR;
}
} else {
if (have_dbl_star_arg) {
op = PYBC_CALL_METHOD_KW;
} else {
op = PYBC_CALL_METHOD;
}
}
emit_write_byte_1_uint(emit, op, (n_keyword << 8) | n_positional); // TODO make it 2 separate uints
}
void emit_return_value(emitter_t *emit) {
emit_pre(emit, -1);
emit->last_emit_was_return_value = true;
emit_write_byte_1(emit, PYBC_RETURN_VALUE);
}
void emit_raise_varargs(emitter_t *emit, int n_args) {
assert(n_args >= 0);
emit_pre(emit, -n_args);
emit_write_byte_1_uint(emit, PYBC_RAISE_VARARGS, n_args);
}
void emit_yield_value(emitter_t *emit) {
emit_pre(emit, 0);
if (emit->pass == PASS_2) {
emit->scope->flags |= SCOPE_FLAG_GENERATOR;
}
emit_write_byte_1(emit, PYBC_YIELD_VALUE);
}
void emit_yield_from(emitter_t *emit) {
emit_pre(emit, -1);
if (emit->pass == PASS_2) {
emit->scope->flags |= SCOPE_FLAG_GENERATOR;
}
emit_write_byte_1(emit, PYBC_YIELD_FROM);
}
#endif // EMIT_DO_BC
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
#include <stdint.h>
#include <stdio.h>
#include <unistd.h>
#include <fcntl.h>
#include "misc.h"
#include "lexer.h"
py_lexer_t *py_lexer_from_file(const char *filename) {
// TODO abstract away file functionality
int fd = open(filename, O_RDONLY);
if (fd < 0) {
printf("cannot open file %s\n", filename);
return NULL;
}
uint size = lseek(fd, 0, SEEK_END);
lseek(fd, 0, SEEK_SET);
char *data = m_new(char, size);
read(fd, data, size);
close(fd);
return py_lexer_from_str_len(filename, data, size, true);
}
0% Loading or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment