emitnative.c

/*
 * This file is part of the Micro Python project, http://micropython.org/
 *
 * The MIT License (MIT)
 *
 * Copyright (c) 2013, 2014 Damien P. George
 *
 * Permission is hereby granted, free of charge, to any person obtaining a copy
 * of this software and associated documentation files (the "Software"), to deal
 * in the Software without restriction, including without limitation the rights
 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 * copies of the Software, and to permit persons to whom the Software is
 * furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in
 * all copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
 * THE SOFTWARE.
 */

// Essentially normal Python has 1 type: Python objects
// Viper has more than 1 type, and is just a more complicated (a superset of) Python.
// If you declare everything in Viper as a Python object (ie omit type decls) then
// it should in principle be exactly the same as Python native.
// Having types means having more opcodes, like binary_op_nat_nat, binary_op_nat_obj etc.
// In practice we won't have a VM but rather do this in asm which is actually very minimal.

// Because it breaks strict Python equivalence it should be a completely separate
// decorator.  It breaks equivalence because overflow on integers wraps around.
// It shouldn't break equivalence if you don't use the new types, but since the
// type decls might be used in normal Python for other reasons, it's probably safest,
// cleanest and clearest to make it a separate decorator.

// Actually, it does break equivalence because integers default to native integers,
// not Python objects.

// for x in l[0:8]: can be compiled into a native loop if l has pointer type

#include <stdbool.h>
#include <stdint.h>
#include <stdio.h>
#include <string.h>
#include <assert.h>

#include "mpconfig.h"
#include "nlr.h"
#include "misc.h"
#include "qstr.h"
#include "lexer.h"
#include "parse.h"
#include "obj.h"
#include "emitglue.h"
#include "scope.h"
#include "runtime0.h"
#include "emit.h"
#include "runtime.h"

#if 0 // print debugging info
#define DEBUG_PRINT (1)
#define DEBUG_printf DEBUG_printf
#else // don't print debugging info
#define DEBUG_printf(...) (void)0
#endif

// wrapper around everything in this file
#if (MICROPY_EMIT_X64 && N_X64) \
    || (MICROPY_EMIT_X86 && N_X86) \
    || (MICROPY_EMIT_THUMB && N_THUMB) \
    || (MICROPY_EMIT_ARM && N_ARM)

#if N_X64

// x64 specific stuff

#include "asmx64.h"

#define EXPORT_FUN(name) emit_native_x64_##name

#define REG_RET ASM_X64_REG_RAX
#define REG_ARG_1 ASM_X64_REG_RDI
#define REG_ARG_2 ASM_X64_REG_RSI
#define REG_ARG_3 ASM_X64_REG_RDX
#define REG_ARG_4 ASM_X64_REG_RCX

// caller-save
#define REG_TEMP0 ASM_X64_REG_RAX
#define REG_TEMP1 ASM_X64_REG_RDI
#define REG_TEMP2 ASM_X64_REG_RSI

// callee-save
#define REG_LOCAL_1 ASM_X64_REG_RBX
#define REG_LOCAL_2 ASM_X64_REG_R12
#define REG_LOCAL_3 ASM_X64_REG_R13
#define REG_LOCAL_NUM (3)

#define ASM_PASS_COMPUTE    ASM_X64_PASS_COMPUTE
#define ASM_PASS_EMIT       ASM_X64_PASS_EMIT

#define ASM_T               asm_x64_t
#define ASM_NEW             asm_x64_new
#define ASM_FREE            asm_x64_free
#define ASM_GET_CODE        asm_x64_get_code
#define ASM_GET_CODE_SIZE   asm_x64_get_code_size
#define ASM_START_PASS      asm_x64_start_pass
#define ASM_END_PASS        asm_x64_end_pass
#define ASM_ENTRY           asm_x64_entry
#define ASM_EXIT            asm_x64_exit

#define ASM_LABEL_ASSIGN    asm_x64_label_assign
#define ASM_JUMP            asm_x64_jmp_label
#define ASM_JUMP_IF_REG_ZERO(as, reg, label) \
    do { \
        asm_x64_test_r8_with_r8(as, reg, reg); \
        asm_x64_jcc_label(as, ASM_X64_CC_JZ, label); \
    } while (0)
#define ASM_JUMP_IF_REG_NONZERO(as, reg, label) \
    do { \
        asm_x64_test_r8_with_r8(as, reg, reg); \
        asm_x64_jcc_label(as, ASM_X64_CC_JNZ, label); \
    } while (0)
#define ASM_JUMP_IF_REG_EQ(as, reg1, reg2, label) \
    do { \
        asm_x64_cmp_r64_with_r64(as, reg1, reg2); \
        asm_x64_jcc_label(as, ASM_X64_CC_JE, label); \
    } while (0)
#define ASM_CALL_IND(as, ptr, idx) asm_x64_call_ind(as, ptr, ASM_X64_REG_RAX)

#define ASM_MOV_REG_TO_LOCAL        asm_x64_mov_r64_to_local
#define ASM_MOV_IMM_TO_REG          asm_x64_mov_i64_to_r64_optimised
#define ASM_MOV_ALIGNED_IMM_TO_REG  asm_x64_mov_i64_to_r64_aligned
#define ASM_MOV_IMM_TO_LOCAL_USING(as, imm, local_num, reg_temp) \
    do { \
        asm_x64_mov_i64_to_r64_optimised(as, (imm), (reg_temp)); \
        asm_x64_mov_r64_to_local(as, (reg_temp), (local_num)); \
    } while (false)
#define ASM_MOV_LOCAL_TO_REG        asm_x64_mov_local_to_r64
#define ASM_MOV_REG_REG(as, reg_dest, reg_src) asm_x64_mov_r64_r64((as), (reg_dest), (reg_src))
#define ASM_MOV_LOCAL_ADDR_TO_REG   asm_x64_mov_local_addr_to_r64

#define ASM_LSL_REG(as, reg) asm_x64_shl_r64_cl((as), (reg))
#define ASM_ASR_REG(as, reg) asm_x64_sar_r64_cl((as), (reg))
#define ASM_ADD_REG_REG(as, reg_dest, reg_src) asm_x64_add_r64_r64((as), (reg_dest), (reg_src))
#define ASM_SUB_REG_REG(as, reg_dest, reg_src) asm_x64_sub_r64_r64((as), (reg_dest), (reg_src))

#define ASM_STORE_REG_REG(as, reg_src, reg_base) asm_x64_mov_r64_to_disp((as), (reg_src), (reg_base), 0)
#define ASM_STORE8_REG_REG(as, reg_src, reg_base) asm_x64_mov_r8_to_disp((as), (reg_src), (reg_base), 0)
#define ASM_STORE16_REG_REG(as, reg_src, reg_base) asm_x64_mov_r16_to_disp((as), (reg_src), (reg_base), 0)

#elif N_X86

// x86 specific stuff

#include "asmx86.h"

STATIC byte mp_f_n_args[MP_F_NUMBER_OF] = {
    [MP_F_CONVERT_OBJ_TO_NATIVE] = 2,
    [MP_F_CONVERT_NATIVE_TO_OBJ] = 2,
    [MP_F_LOAD_CONST_INT] = 1,
    [MP_F_LOAD_CONST_DEC] = 1,
    [MP_F_LOAD_CONST_STR] = 1,
    [MP_F_LOAD_CONST_BYTES] = 1,
    [MP_F_LOAD_NAME] = 1,
    [MP_F_LOAD_GLOBAL] = 1,
    [MP_F_LOAD_BUILD_CLASS] = 0,
    [MP_F_LOAD_ATTR] = 2,
    [MP_F_LOAD_METHOD] = 3,
    [MP_F_STORE_NAME] = 2,
    [MP_F_STORE_GLOBAL] = 2,
    [MP_F_STORE_ATTR] = 3,
    [MP_F_OBJ_SUBSCR] = 3,
    [MP_F_OBJ_IS_TRUE] = 1,
    [MP_F_UNARY_OP] = 2,
    [MP_F_BINARY_OP] = 3,
    [MP_F_BUILD_TUPLE] = 2,
    [MP_F_BUILD_LIST] = 2,
    [MP_F_LIST_APPEND] = 2,
    [MP_F_BUILD_MAP] = 1,
    [MP_F_STORE_MAP] = 3,
#if MICROPY_PY_BUILTINS_SET
    [MP_F_BUILD_SET] = 2,
    [MP_F_STORE_SET] = 2,
#endif
    [MP_F_MAKE_FUNCTION_FROM_RAW_CODE] = 3,
    [MP_F_NATIVE_CALL_FUNCTION_N_KW] = 3,
    [MP_F_CALL_METHOD_N_KW] = 3,
    [MP_F_GETITER] = 1,
    [MP_F_ITERNEXT] = 1,
    [MP_F_NLR_PUSH] = 1,
    [MP_F_NLR_POP] = 0,
    [MP_F_NATIVE_RAISE] = 1,
    [MP_F_IMPORT_NAME] = 3,
    [MP_F_IMPORT_FROM] = 2,
    [MP_F_IMPORT_ALL] = 1,
#if MICROPY_PY_BUILTINS_SLICE
    [MP_F_NEW_SLICE] = 3,
#endif
    [MP_F_UNPACK_SEQUENCE] = 3,
    [MP_F_UNPACK_EX] = 3,
    [MP_F_DELETE_NAME] = 1,
    [MP_F_DELETE_GLOBAL] = 1,
};

#define EXPORT_FUN(name) emit_native_x86_##name

#define REG_RET ASM_X86_REG_EAX
#define REG_ARG_1 ASM_X86_REG_ARG_1
#define REG_ARG_2 ASM_X86_REG_ARG_2
#define REG_ARG_3 ASM_X86_REG_ARG_3

// caller-save, so can be used as temporaries
#define REG_TEMP0 ASM_X86_REG_EAX
#define REG_TEMP1 ASM_X86_REG_ECX
#define REG_TEMP2 ASM_X86_REG_EDX

// callee-save, so can be used as locals
#define REG_LOCAL_1 ASM_X86_REG_EBX
#define REG_LOCAL_2 ASM_X86_REG_ESI
#define REG_LOCAL_3 ASM_X86_REG_EDI
#define REG_LOCAL_NUM (3)

#define ASM_PASS_COMPUTE    ASM_X86_PASS_COMPUTE
#define ASM_PASS_EMIT       ASM_X86_PASS_EMIT

#define ASM_T               asm_x86_t
#define ASM_NEW             asm_x86_new
#define ASM_FREE            asm_x86_free
#define ASM_GET_CODE        asm_x86_get_code
#define ASM_GET_CODE_SIZE   asm_x86_get_code_size
#define ASM_START_PASS      asm_x86_start_pass
#define ASM_END_PASS        asm_x86_end_pass
#define ASM_ENTRY           asm_x86_entry
#define ASM_EXIT            asm_x86_exit

#define ASM_LABEL_ASSIGN    asm_x86_label_assign
#define ASM_JUMP            asm_x86_jmp_label
#define ASM_JUMP_IF_REG_ZERO(as, reg, label) \
    do { \
        asm_x86_test_r8_with_r8(as, reg, reg); \
        asm_x86_jcc_label(as, ASM_X86_CC_JZ, label); \
    } while (0)
#define ASM_JUMP_IF_REG_NONZERO(as, reg, label) \
    do { \
        asm_x86_test_r8_with_r8(as, reg, reg); \
        asm_x86_jcc_label(as, ASM_X86_CC_JNZ, label); \
    } while (0)
#define ASM_JUMP_IF_REG_EQ(as, reg1, reg2, label) \
    do { \
        asm_x86_cmp_r32_with_r32(as, reg1, reg2); \
        asm_x86_jcc_label(as, ASM_X86_CC_JE, label); \
    } while (0)
#define ASM_CALL_IND(as, ptr, idx) asm_x86_call_ind(as, ptr, mp_f_n_args[idx], ASM_X86_REG_EAX)

#define ASM_MOV_REG_TO_LOCAL        asm_x86_mov_r32_to_local
#define ASM_MOV_IMM_TO_REG          asm_x86_mov_i32_to_r32
#define ASM_MOV_ALIGNED_IMM_TO_REG  asm_x86_mov_i32_to_r32_aligned
#define ASM_MOV_IMM_TO_LOCAL_USING(as, imm, local_num, reg_temp) \
    do { \
        asm_x86_mov_i32_to_r32(as, (imm), (reg_temp)); \
        asm_x86_mov_r32_to_local(as, (reg_temp), (local_num)); \
    } while (false)
#define ASM_MOV_LOCAL_TO_REG        asm_x86_mov_local_to_r32
#define ASM_MOV_REG_REG(as, reg_dest, reg_src) asm_x86_mov_r32_r32((as), (reg_dest), (reg_src))
#define ASM_MOV_LOCAL_ADDR_TO_REG   asm_x86_mov_local_addr_to_r32

#define ASM_LSL_REG(as, reg) asm_x86_shl_r32_cl((as), (reg))
#define ASM_ASR_REG(as, reg) asm_x86_sar_r32_cl((as), (reg))
#define ASM_ADD_REG_REG(as, reg_dest, reg_src) asm_x86_add_r32_r32((as), (reg_dest), (reg_src))
#define ASM_SUB_REG_REG(as, reg_dest, reg_src) asm_x86_sub_r32_r32((as), (reg_dest), (reg_src))

#define ASM_STORE_REG_REG(as, reg_src, reg_base) asm_x86_mov_r32_to_disp((as), (reg_src), (reg_base), 0)
#define ASM_STORE8_REG_REG(as, reg_src, reg_base) asm_x86_mov_r8_to_disp((as), (reg_src), (reg_base), 0)
#define ASM_STORE16_REG_REG(as, reg_src, reg_base) asm_x86_mov_r16_to_disp((as), (reg_src), (reg_base), 0)

#elif N_THUMB

// thumb specific stuff

#include "asmthumb.h"

#define EXPORT_FUN(name) emit_native_thumb_##name

#define REG_RET ASM_THUMB_REG_R0
#define REG_ARG_1 ASM_THUMB_REG_R0
#define REG_ARG_2 ASM_THUMB_REG_R1
#define REG_ARG_3 ASM_THUMB_REG_R2
#define REG_ARG_4 ASM_THUMB_REG_R3

#define REG_TEMP0 ASM_THUMB_REG_R0
#define REG_TEMP1 ASM_THUMB_REG_R1
#define REG_TEMP2 ASM_THUMB_REG_R2

#define REG_LOCAL_1 ASM_THUMB_REG_R4
#define REG_LOCAL_2 ASM_THUMB_REG_R5
#define REG_LOCAL_3 ASM_THUMB_REG_R6
#define REG_LOCAL_NUM (3)

#define ASM_PASS_COMPUTE    ASM_THUMB_PASS_COMPUTE
#define ASM_PASS_EMIT       ASM_THUMB_PASS_EMIT

#define ASM_T               asm_thumb_t
#define ASM_NEW             asm_thumb_new
#define ASM_FREE            asm_thumb_free
#define ASM_GET_CODE        asm_thumb_get_code
#define ASM_GET_CODE_SIZE   asm_thumb_get_code_size
#define ASM_START_PASS      asm_thumb_start_pass
#define ASM_END_PASS        asm_thumb_end_pass
#define ASM_ENTRY           asm_thumb_entry
#define ASM_EXIT            asm_thumb_exit

#define ASM_LABEL_ASSIGN    asm_thumb_label_assign
#define ASM_JUMP            asm_thumb_b_label
#define ASM_JUMP_IF_REG_ZERO(as, reg, label) \
    do { \
        asm_thumb_cmp_rlo_i8(as, reg, 0); \
        asm_thumb_bcc_label(as, ASM_THUMB_CC_EQ, label); \
    } while (0)
#define ASM_JUMP_IF_REG_NONZERO(as, reg, label) \
    do { \
        asm_thumb_cmp_rlo_i8(as, reg, 0); \
        asm_thumb_bcc_label(as, ASM_THUMB_CC_NE, label); \
    } while (0)
#define ASM_JUMP_IF_REG_EQ(as, reg1, reg2, label) \
    do { \
        asm_thumb_cmp_rlo_rlo(as, reg1, reg2); \
        asm_thumb_bcc_label(as, ASM_THUMB_CC_EQ, label); \
    } while (0)
#define ASM_CALL_IND(as, ptr, idx) asm_thumb_bl_ind(as, ptr, idx, ASM_THUMB_REG_R3)

#define ASM_MOV_REG_TO_LOCAL(as, reg, local_num) asm_thumb_mov_local_reg(as, (local_num), (reg))
#define ASM_MOV_IMM_TO_REG(as, imm, reg) asm_thumb_mov_reg_i32_optimised(as, (reg), (imm))
#define ASM_MOV_ALIGNED_IMM_TO_REG(as, imm, reg) asm_thumb_mov_reg_i32_aligned(as, (reg), (imm))
#define ASM_MOV_IMM_TO_LOCAL_USING(as, imm, local_num, reg_temp) \
    do { \
        asm_thumb_mov_reg_i32_optimised(as, (reg_temp), (imm)); \
        asm_thumb_mov_local_reg(as, (local_num), (reg_temp)); \
    } while (false)
#define ASM_MOV_LOCAL_TO_REG(as, local_num, reg) asm_thumb_mov_reg_local(as, (reg), (local_num))
#define ASM_MOV_REG_REG(as, reg_dest, reg_src) asm_thumb_mov_reg_reg((as), (reg_dest), (reg_src))
#define ASM_MOV_LOCAL_ADDR_TO_REG(as, local_num, reg) asm_thumb_mov_reg_local_addr(as, (reg), (local_num))

#define ASM_LSL_REG_REG(as, reg_dest, reg_shift) asm_thumb_format_4((as), ASM_THUMB_FORMAT_4_LSL, (reg_dest), (reg_shift))
#define ASM_ASR_REG_REG(as, reg_dest, reg_shift) asm_thumb_format_4((as), ASM_THUMB_FORMAT_4_ASR, (reg_dest), (reg_shift))
#define ASM_ADD_REG_REG(as, reg_dest, reg_src) asm_thumb_add_rlo_rlo_rlo((as), (reg_dest), (reg_dest), (reg_src))
#define ASM_SUB_REG_REG(as, reg_dest, reg_src) asm_thumb_sub_rlo_rlo_rlo((as), (reg_dest), (reg_dest), (reg_src))

#define ASM_STORE_REG_REG(as, reg_src, reg_base) asm_thumb_str_rlo_rlo_i5((as), (reg_src), (reg_base), 0)
#define ASM_STORE8_REG_REG(as, reg_src, reg_base) asm_thumb_strb_rlo_rlo_i5((as), (reg_src), (reg_base), 0)
#define ASM_STORE16_REG_REG(as, reg_src, reg_base) asm_thumb_strh_rlo_rlo_i5((as), (reg_src), (reg_base), 0)

#elif N_ARM

// ARM specific stuff

#include "asmarm.h"

#define EXPORT_FUN(name) emit_native_arm_##name

#define REG_RET ASM_ARM_REG_R0
#define REG_ARG_1 ASM_ARM_REG_R0
#define REG_ARG_2 ASM_ARM_REG_R1
#define REG_ARG_3 ASM_ARM_REG_R2
#define REG_ARG_4 ASM_ARM_REG_R3

#define REG_TEMP0 ASM_ARM_REG_R0
#define REG_TEMP1 ASM_ARM_REG_R1
#define REG_TEMP2 ASM_ARM_REG_R2

#define REG_LOCAL_1 ASM_ARM_REG_R4
#define REG_LOCAL_2 ASM_ARM_REG_R5
#define REG_LOCAL_3 ASM_ARM_REG_R6
#define REG_LOCAL_NUM (3)

#define ASM_PASS_COMPUTE    ASM_ARM_PASS_COMPUTE
#define ASM_PASS_EMIT       ASM_ARM_PASS_EMIT

#define ASM_T               asm_arm_t
#define ASM_NEW             asm_arm_new
#define ASM_FREE            asm_arm_free
#define ASM_GET_CODE        asm_arm_get_code
#define ASM_GET_CODE_SIZE   asm_arm_get_code_size
#define ASM_START_PASS      asm_arm_start_pass
#define ASM_END_PASS        asm_arm_end_pass
#define ASM_ENTRY           asm_arm_entry
#define ASM_EXIT            asm_arm_exit

#define ASM_LABEL_ASSIGN    asm_arm_label_assign
#define ASM_JUMP            asm_arm_b_label
#define ASM_JUMP_IF_REG_ZERO(as, reg, label) \
    do { \
        asm_arm_cmp_reg_i8(as, reg, 0); \
        asm_arm_bcc_label(as, ASM_ARM_CC_EQ, label); \
    } while (0)
#define ASM_JUMP_IF_REG_NONZERO(as, reg, label) \
    do { \
        asm_arm_cmp_reg_i8(as, reg, 0); \
        asm_arm_bcc_label(as, ASM_ARM_CC_NE, label); \
    } while (0)
#define ASM_JUMP_IF_REG_EQ(as, reg1, reg2, label) \
    do { \
        asm_arm_cmp_reg_reg(as, reg1, reg2); \
        asm_arm_bcc_label(as, ASM_ARM_CC_EQ, label); \
    } while (0)
#define ASM_CALL_IND(as, ptr, idx) asm_arm_bl_ind(as, ptr, idx, ASM_ARM_REG_R3)

#define ASM_MOV_REG_TO_LOCAL(as, reg, local_num) asm_arm_mov_local_reg(as, (local_num), (reg))
#define ASM_MOV_IMM_TO_REG(as, imm, reg) asm_arm_mov_reg_i32(as, (reg), (imm))
#define ASM_MOV_ALIGNED_IMM_TO_REG(as, imm, reg) asm_arm_mov_reg_i32(as, (reg), (imm))
#define ASM_MOV_IMM_TO_LOCAL_USING(as, imm, local_num, reg_temp) \
    do { \
        asm_arm_mov_reg_i32(as, (reg_temp), (imm)); \
        asm_arm_mov_local_reg(as, (local_num), (reg_temp)); \
    } while (false)
#define ASM_MOV_LOCAL_TO_REG(as, local_num, reg) asm_arm_mov_reg_local(as, (reg), (local_num))
#define ASM_MOV_REG_REG(as, reg_dest, reg_src) asm_arm_mov_reg_reg((as), (reg_dest), (reg_src))
#define ASM_MOV_LOCAL_ADDR_TO_REG(as, local_num, reg) asm_arm_mov_reg_local_addr(as, (reg), (local_num))

// TODO someone please implement lsl and asr
#define ASM_LSL_REG_REG(as, reg_dest, reg_shift) asm_arm_lsl_((as), (reg_dest), (reg_shift))
#define ASM_ASR_REG_REG(as, reg_dest, reg_shift) asm_arm_asr_((as), (reg_dest), (reg_shift))
#define ASM_ADD_REG_REG(as, reg_dest, reg_src) asm_arm_add_reg_reg_reg((as), (reg_dest), (reg_dest), (reg_src))
#define ASM_SUB_REG_REG(as, reg_dest, reg_src) asm_arm_sub_reg_reg_reg((as), (reg_dest), (reg_dest), (reg_src))

// TODO someone please implement str
#define ASM_STORE_REG_REG(as, reg_src, reg_base) asm_arm_str_reg_reg_i5((as), (reg_src), (reg_base), 0)
#define ASM_STORE8_REG_REG(as, reg_src, reg_base) asm_arm_strb_reg_reg_i5((as), (reg_src), (reg_base), 0)
#define ASM_STORE16_REG_REG(as, reg_src, reg_base) asm_arm_strh_reg_reg_i5((as), (reg_src), (reg_base), 0)

#else

#error unknown native emitter

#endif

typedef enum {
    STACK_VALUE,
    STACK_REG,
    STACK_IMM,
} stack_info_kind_t;

// these enums must be distinct and the bottom 2 bits
// must correspond to the correct MP_NATIVE_TYPE_xxx value
typedef enum {
    VTYPE_PYOBJ = 0x00 | MP_NATIVE_TYPE_OBJ,
    VTYPE_BOOL = 0x00 | MP_NATIVE_TYPE_BOOL,
    VTYPE_INT = 0x00 | MP_NATIVE_TYPE_INT,
    VTYPE_UINT = 0x00 | MP_NATIVE_TYPE_UINT,

    VTYPE_PTR = 0x10 | MP_NATIVE_TYPE_UINT, // pointer to word sized entity
    VTYPE_PTR8 = 0x20 | MP_NATIVE_TYPE_UINT,
    VTYPE_PTR16 = 0x30 | MP_NATIVE_TYPE_UINT,
    VTYPE_PTR_NONE = 0x40 | MP_NATIVE_TYPE_UINT,

    VTYPE_UNBOUND = 0x50 | MP_NATIVE_TYPE_OBJ,
    VTYPE_BUILTIN_CAST = 0x60 | MP_NATIVE_TYPE_OBJ,
} vtype_kind_t;

typedef struct _stack_info_t {
    vtype_kind_t vtype;
    stack_info_kind_t kind;
    union {
        int u_reg;
        mp_int_t u_imm;
    };
} stack_info_t;

struct _emit_t {
    int pass;

    bool do_viper_types;

    vtype_kind_t return_vtype;

    mp_uint_t local_vtype_alloc;
    vtype_kind_t *local_vtype;

    mp_uint_t stack_info_alloc;
    stack_info_t *stack_info;

    int stack_start;
    int stack_size;

    bool last_emit_was_return_value;

    scope_t *scope;

    ASM_T *as;
};

emit_t *EXPORT_FUN(new)(mp_uint_t max_num_labels) {
    emit_t *emit = m_new0(emit_t, 1);
    emit->as = ASM_NEW(max_num_labels);
    return emit;
}

void EXPORT_FUN(free)(emit_t *emit) {
    ASM_FREE(emit->as, false);
    m_del(vtype_kind_t, emit->local_vtype, emit->local_vtype_alloc);
    m_del(stack_info_t, emit->stack_info, emit->stack_info_alloc);
    m_del_obj(emit_t, emit);
}

STATIC void emit_native_set_native_type(emit_t *emit, mp_uint_t op, mp_uint_t arg1, qstr arg2) {
    switch (op) {
        case MP_EMIT_NATIVE_TYPE_ENABLE:
            emit->do_viper_types = arg1;
            break;

        default: {
            vtype_kind_t type;
            switch (arg2) {
                case MP_QSTR_object: type = VTYPE_PYOBJ; break;
                case MP_QSTR_bool: type = VTYPE_BOOL; break;
                case MP_QSTR_int: type = VTYPE_INT; break;
                case MP_QSTR_uint: type = VTYPE_UINT; break;
                case MP_QSTR_ptr: type = VTYPE_PTR; break;
                case MP_QSTR_ptr8: type = VTYPE_PTR8; break;
                case MP_QSTR_ptr16: type = VTYPE_PTR16; break;
                default: printf("ViperTypeError: unknown type %s\n", qstr_str(arg2)); return;
            }
            if (op == MP_EMIT_NATIVE_TYPE_RETURN) {
                emit->return_vtype = type;
            } else {
                assert(arg1 < emit->local_vtype_alloc);
                emit->local_vtype[arg1] = type;
            }
            break;
        }
    }
}

STATIC void emit_native_start_pass(emit_t *emit, pass_kind_t pass, scope_t *scope) {
    DEBUG_printf("start_pass(pass=%u, scope=%p)\n", pass, scope);

    emit->pass = pass;
    emit->stack_start = 0;
    emit->stack_size = 0;
    emit->last_emit_was_return_value = false;
    emit->scope = scope;

    // allocate memory for keeping track of the types of locals
    if (emit->local_vtype_alloc < scope->num_locals) {
        emit->local_vtype = m_renew(vtype_kind_t, emit->local_vtype, emit->local_vtype_alloc, scope->num_locals);
        emit->local_vtype_alloc = scope->num_locals;
    }

    // allocate memory for keeping track of the objects on the stack
    // XXX don't know stack size on entry, and it should be maximum over all scopes
    if (emit->stack_info == NULL) {
        emit->stack_info_alloc = scope->stack_size + 50;
        emit->stack_info = m_new(stack_info_t, emit->stack_info_alloc);
    }

    // set default type for return and arguments
    emit->return_vtype = VTYPE_PYOBJ;
    for (mp_uint_t i = 0; i < emit->scope->num_pos_args; i++) {
        emit->local_vtype[i] = VTYPE_PYOBJ;
    }

    // local variables begin unbound, and have unknown type
    for (mp_uint_t i = emit->scope->num_pos_args; i < emit->local_vtype_alloc; i++) {
        emit->local_vtype[i] = VTYPE_UNBOUND;
    }

    // values on stack begin unbound
    for (mp_uint_t i = 0; i < emit->stack_info_alloc; i++) {
        emit->stack_info[i].kind = STACK_VALUE;
        emit->stack_info[i].vtype = VTYPE_UNBOUND;
    }

    ASM_START_PASS(emit->as, pass == MP_PASS_EMIT ? ASM_PASS_EMIT : ASM_PASS_COMPUTE);

    // entry to function
    int num_locals = 0;
    if (pass > MP_PASS_SCOPE) {
        num_locals = scope->num_locals - REG_LOCAL_NUM;
        if (num_locals < 0) {
            num_locals = 0;
        }
        emit->stack_start = num_locals;
        num_locals += scope->stack_size;
    }
    ASM_ENTRY(emit->as, num_locals);

    // initialise locals from parameters
#if N_X64
    for (int i = 0; i < scope->num_pos_args; i++) {
        if (i == 0) {
            ASM_MOV_REG_REG(emit->as, REG_LOCAL_1, REG_ARG_1);
        } else if (i == 1) {
            ASM_MOV_REG_REG(emit->as, REG_LOCAL_2, REG_ARG_2);
        } else if (i == 2) {
            ASM_MOV_REG_REG(emit->as, REG_LOCAL_3, REG_ARG_3);
        } else if (i == 3) {
            asm_x64_mov_r64_to_local(emit->as, REG_ARG_4, i - REG_LOCAL_NUM);
        } else {
            // TODO not implemented
            assert(0);
        }
    }
#elif N_X86
    for (int i = 0; i < scope->num_pos_args; i++) {
        if (i == 0) {
            asm_x86_mov_arg_to_r32(emit->as, i, REG_LOCAL_1);
        } else if (i == 1) {
            asm_x86_mov_arg_to_r32(emit->as, i, REG_LOCAL_2);
        } else if (i == 2) {
            asm_x86_mov_arg_to_r32(emit->as, i, REG_LOCAL_3);
        } else {
            asm_x86_mov_arg_to_r32(emit->as, i, REG_TEMP0);
            asm_x86_mov_r32_to_local(emit->as, REG_TEMP0, i - REG_LOCAL_NUM);
        }
    }
#elif N_THUMB
    for (int i = 0; i < scope->num_pos_args; i++) {
        if (i == 0) {
            ASM_MOV_REG_REG(emit->as, REG_LOCAL_1, REG_ARG_1);
        } else if (i == 1) {
            ASM_MOV_REG_REG(emit->as, REG_LOCAL_2, REG_ARG_2);
        } else if (i == 2) {
            ASM_MOV_REG_REG(emit->as, REG_LOCAL_3, REG_ARG_3);
        } else if (i == 3) {
            asm_thumb_mov_local_reg(emit->as, i - REG_LOCAL_NUM, REG_ARG_4);
        } else {
            // TODO not implemented
            assert(0);
        }
    }

    // TODO don't load r7 if we don't need it
    asm_thumb_mov_reg_i32(emit->as, ASM_THUMB_REG_R7, (mp_uint_t)mp_fun_table);
#elif N_ARM
    for (int i = 0; i < scope->num_pos_args; i++) {
        if (i == 0) {
            ASM_MOV_REG_REG(emit->as, REG_LOCAL_1, REG_ARG_1);
        } else if (i == 1) {
            ASM_MOV_REG_REG(emit->as, REG_LOCAL_2, REG_ARG_2);
        } else if (i == 2) {
            ASM_MOV_REG_REG(emit->as, REG_LOCAL_3, REG_ARG_3);
        } else if (i == 3) {
            asm_arm_mov_local_reg(emit->as, i - REG_LOCAL_NUM, REG_ARG_4);
        } else {
            // TODO not implemented
            assert(0);
        }
    }

    // TODO don't load r7 if we don't need it
    asm_arm_mov_reg_i32(emit->as, ASM_ARM_REG_R7, (mp_uint_t)mp_fun_table);
#else
    #error not implemented
#endif
}

STATIC void emit_native_end_pass(emit_t *emit) {
    if (!emit->last_emit_was_return_value) {
        ASM_EXIT(emit->as);
    }
    ASM_END_PASS(emit->as);

    // 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 == MP_PASS_EMIT) {
        void *f = ASM_GET_CODE(emit->as);
        mp_uint_t f_len = ASM_GET_CODE_SIZE(emit->as);

        // compute type signature
        // note that the lower 2 bits of a vtype are tho correct MP_NATIVE_TYPE_xxx
        mp_uint_t type_sig = emit->return_vtype & 3;
        for (mp_uint_t i = 0; i < emit->scope->num_pos_args; i++) {
            type_sig |= (emit->local_vtype[i] & 3) << (i * 2 + 2);
        }

        mp_emit_glue_assign_native(emit->scope->raw_code, emit->do_viper_types ? MP_CODE_NATIVE_VIPER : MP_CODE_NATIVE_PY, f, f_len, emit->scope->num_pos_args, type_sig);
    }
}

STATIC bool emit_native_last_emit_was_return_value(emit_t *emit) {
    return emit->last_emit_was_return_value;
}

STATIC void adjust_stack(emit_t *emit, mp_int_t stack_size_delta) {
    DEBUG_printf("  adjust_stack; stack_size=%d, delta=%d\n", emit->stack_size, stack_size_delta);
    assert((mp_int_t)emit->stack_size + stack_size_delta >= 0);
    emit->stack_size += stack_size_delta;
    if (emit->pass > MP_PASS_SCOPE && emit->stack_size > emit->scope->stack_size) {
        emit->scope->stack_size = emit->stack_size;
    }
}

STATIC void emit_native_adjust_stack_size(emit_t *emit, mp_int_t delta) {
    // If we are adjusting the stack in a positive direction (pushing) then we
    // need to fill in values for the stack kind and vtype of the newly-pushed
    // entries.  These should be set to "value" (ie not reg or imm) because we
    // should only need to adjust the stack due to a jump to this part in the
    // code (and hence we have settled the stack before the jump).
    for (mp_int_t i = 0; i < delta; i++) {
        stack_info_t *si = &emit->stack_info[emit->stack_size + i];
        si->kind = STACK_VALUE;
        si->vtype = VTYPE_PYOBJ; // XXX we don't know the vtype...
    }
    adjust_stack(emit, delta);
}

STATIC void emit_native_set_source_line(emit_t *emit, mp_uint_t source_line) {
}

/*
STATIC void emit_pre_raw(emit_t *emit, int stack_size_delta) {
    adjust_stack(emit, stack_size_delta);
    emit->last_emit_was_return_value = false;
}
*/

// this must be called at start of emit functions
STATIC void emit_native_pre(emit_t *emit) {
    emit->last_emit_was_return_value = false;
    // settle the stack
    /*
    if (regs_needed != 0) {
        for (int i = 0; i < emit->stack_size; i++) {
            switch (emit->stack_info[i].kind) {
                case STACK_VALUE:
                    break;

                case STACK_REG:
                    // TODO only push reg if in regs_needed
                    emit->stack_info[i].kind = STACK_VALUE;
                    ASM_MOV_REG_TO_LOCAL(emit->as, emit->stack_info[i].u_reg, emit->stack_start + i);
                    break;

                case STACK_IMM:
                    // don't think we ever need to push imms for settling
                    //ASM_MOV_IMM_TO_LOCAL(emit->last_imm, emit->stack_start + i);
                    break;
            }
        }
    }
    */
}

// depth==0 is top, depth==1 is before top, etc
STATIC stack_info_t *peek_stack(emit_t *emit, mp_uint_t depth) {
    return &emit->stack_info[emit->stack_size - 1 - depth];
}

// depth==0 is top, depth==1 is before top, etc
STATIC vtype_kind_t peek_vtype(emit_t *emit, mp_uint_t depth) {
    return peek_stack(emit, depth)->vtype;
}

// pos=1 is TOS, pos=2 is next, etc
// use pos=0 for no skipping
STATIC void need_reg_single(emit_t *emit, int reg_needed, int skip_stack_pos) {
    skip_stack_pos = emit->stack_size - skip_stack_pos;
    for (int i = 0; i < emit->stack_size; i++) {
        if (i != skip_stack_pos) {
            stack_info_t *si = &emit->stack_info[i];
            if (si->kind == STACK_REG && si->u_reg == reg_needed) {
                si->kind = STACK_VALUE;
                ASM_MOV_REG_TO_LOCAL(emit->as, si->u_reg, emit->stack_start + i);
            }
        }
    }
}

STATIC void need_reg_all(emit_t *emit) {
    for (int i = 0; i < emit->stack_size; i++) {
        stack_info_t *si = &emit->stack_info[i];
        if (si->kind == STACK_REG) {
            si->kind = STACK_VALUE;
            ASM_MOV_REG_TO_LOCAL(emit->as, si->u_reg, emit->stack_start + i);
        }
    }
}

STATIC void need_stack_settled(emit_t *emit) {
    DEBUG_printf("  need_stack_settled; stack_size=%d\n", emit->stack_size);
    for (int i = 0; i < emit->stack_size; i++) {
        stack_info_t *si = &emit->stack_info[i];
        if (si->kind == STACK_REG) {
            DEBUG_printf("    reg(%u) to local(%u)\n", si->u_reg, emit->stack_start + i);
            si->kind = STACK_VALUE;
            ASM_MOV_REG_TO_LOCAL(emit->as, si->u_reg, emit->stack_start + i);
        }
    }
    for (int i = 0; i < emit->stack_size; i++) {
        stack_info_t *si = &emit->stack_info[i];
        if (si->kind == STACK_IMM) {
            DEBUG_printf("    imm(" INT_FMT ") to local(%u)\n", si->u_imm, emit->stack_start + i);
            si->kind = STACK_VALUE;
            ASM_MOV_IMM_TO_LOCAL_USING(emit->as, si->u_imm, emit->stack_start + i, REG_TEMP0);
        }
    }
}

// pos=1 is TOS, pos=2 is next, etc
STATIC void emit_access_stack(emit_t *emit, int pos, vtype_kind_t *vtype, int reg_dest) {
    need_reg_single(emit, reg_dest, pos);
    stack_info_t *si = &emit->stack_info[emit->stack_size - pos];
    *vtype = si->vtype;
    switch (si->kind) {
        case STACK_VALUE:
            ASM_MOV_LOCAL_TO_REG(emit->as, emit->stack_start + emit->stack_size - pos, reg_dest);
            break;

        case STACK_REG:
            if (si->u_reg != reg_dest) {
                ASM_MOV_REG_REG(emit->as, reg_dest, si->u_reg);
            }
            break;

        case STACK_IMM:
            ASM_MOV_IMM_TO_REG(emit->as, si->u_imm, reg_dest);
            break;
    }
}

// does an efficient X=pop(); discard(); push(X)
// needs a (non-temp) register in case the poped element was stored in the stack
STATIC void emit_fold_stack_top(emit_t *emit, int reg_dest) {
    stack_info_t *si = &emit->stack_info[emit->stack_size - 2];
    si[0] = si[1];
    if (si->kind == STACK_VALUE) {
        // if folded element was on the stack we need to put it in a register
        ASM_MOV_LOCAL_TO_REG(emit->as, emit->stack_start + emit->stack_size - 1, reg_dest);
        si->kind = STACK_REG;
        si->u_reg = reg_dest;
    }
    adjust_stack(emit, -1);
}

// If stacked value is in a register and the register is not r1 or r2, then
// *reg_dest is set to that register.  Otherwise the value is put in *reg_dest.
STATIC void emit_pre_pop_reg_flexible(emit_t *emit, vtype_kind_t *vtype, int *reg_dest, int not_r1, int not_r2) {
    emit->last_emit_was_return_value = false;
    stack_info_t *si = peek_stack(emit, 0);
    if (si->kind == STACK_REG && si->u_reg != not_r1 && si->u_reg != not_r2) {
        *vtype = si->vtype;
        *reg_dest = si->u_reg;
        need_reg_single(emit, *reg_dest, 1);
    } else {
        emit_access_stack(emit, 1, vtype, *reg_dest);
    }
    adjust_stack(emit, -1);
}

STATIC void emit_pre_pop_discard(emit_t *emit) {
    emit->last_emit_was_return_value = false;
    adjust_stack(emit, -1);
}

STATIC void emit_pre_pop_reg(emit_t *emit, vtype_kind_t *vtype, int reg_dest) {
    emit->last_emit_was_return_value = false;
    emit_access_stack(emit, 1, vtype, reg_dest);
    adjust_stack(emit, -1);
}

STATIC void emit_pre_pop_reg_reg(emit_t *emit, vtype_kind_t *vtypea, int rega, vtype_kind_t *vtypeb, int regb) {
    emit_pre_pop_reg(emit, vtypea, rega);
    emit_pre_pop_reg(emit, vtypeb, regb);
}

STATIC void emit_pre_pop_reg_reg_reg(emit_t *emit, vtype_kind_t *vtypea, int rega, vtype_kind_t *vtypeb, int regb, vtype_kind_t *vtypec, int regc) {
    emit_pre_pop_reg(emit, vtypea, rega);
    emit_pre_pop_reg(emit, vtypeb, regb);
    emit_pre_pop_reg(emit, vtypec, regc);
}

STATIC void emit_post(emit_t *emit) {
}

STATIC void emit_post_top_set_vtype(emit_t *emit, vtype_kind_t new_vtype) {
    stack_info_t *si = &emit->stack_info[emit->stack_size - 1];
    si->vtype = new_vtype;
}

STATIC void emit_post_push_reg(emit_t *emit, vtype_kind_t vtype, int reg) {
    stack_info_t *si = &emit->stack_info[emit->stack_size];
    si->vtype = vtype;
    si->kind = STACK_REG;
    si->u_reg = reg;
    adjust_stack(emit, 1);
}

STATIC void emit_post_push_imm(emit_t *emit, vtype_kind_t vtype, mp_int_t imm) {
    stack_info_t *si = &emit->stack_info[emit->stack_size];
    si->vtype = vtype;
    si->kind = STACK_IMM;
    si->u_imm = imm;
    adjust_stack(emit, 1);
}

STATIC void emit_post_push_reg_reg(emit_t *emit, vtype_kind_t vtypea, int rega, vtype_kind_t vtypeb, int regb) {
    emit_post_push_reg(emit, vtypea, rega);
    emit_post_push_reg(emit, vtypeb, regb);
}

STATIC void emit_post_push_reg_reg_reg(emit_t *emit, vtype_kind_t vtypea, int rega, vtype_kind_t vtypeb, int regb, vtype_kind_t vtypec, int regc) {
    emit_post_push_reg(emit, vtypea, rega);
    emit_post_push_reg(emit, vtypeb, regb);
    emit_post_push_reg(emit, vtypec, regc);
}

STATIC void emit_post_push_reg_reg_reg_reg(emit_t *emit, vtype_kind_t vtypea, int rega, vtype_kind_t vtypeb, int regb, vtype_kind_t vtypec, int regc, vtype_kind_t vtyped, int regd) {
    emit_post_push_reg(emit, vtypea, rega);
    emit_post_push_reg(emit, vtypeb, regb);
    emit_post_push_reg(emit, vtypec, regc);
    emit_post_push_reg(emit, vtyped, regd);
}

STATIC void emit_call(emit_t *emit, mp_fun_kind_t fun_kind) {
    need_reg_all(emit);
    ASM_CALL_IND(emit->as, mp_fun_table[fun_kind], fun_kind);
}

STATIC void emit_call_with_imm_arg(emit_t *emit, mp_fun_kind_t fun_kind, mp_int_t arg_val, int arg_reg) {
    need_reg_all(emit);
    ASM_MOV_IMM_TO_REG(emit->as, arg_val, arg_reg);
    ASM_CALL_IND(emit->as, mp_fun_table[fun_kind], fun_kind);
}

// the first arg is stored in the code aligned on a mp_uint_t boundary
STATIC void emit_call_with_imm_arg_aligned(emit_t *emit, mp_fun_kind_t fun_kind, mp_int_t arg_val, int arg_reg) {
    need_reg_all(emit);
    ASM_MOV_ALIGNED_IMM_TO_REG(emit->as, arg_val, arg_reg);
    ASM_CALL_IND(emit->as, mp_fun_table[fun_kind], fun_kind);
}

STATIC void emit_call_with_2_imm_args(emit_t *emit, mp_fun_kind_t fun_kind, mp_int_t arg_val1, int arg_reg1, mp_int_t arg_val2, int arg_reg2) {
    need_reg_all(emit);
    ASM_MOV_IMM_TO_REG(emit->as, arg_val1, arg_reg1);
    ASM_MOV_IMM_TO_REG(emit->as, arg_val2, arg_reg2);
    ASM_CALL_IND(emit->as, mp_fun_table[fun_kind], fun_kind);
}

// the first arg is stored in the code aligned on a mp_uint_t boundary
STATIC void emit_call_with_3_imm_args_and_first_aligned(emit_t *emit, mp_fun_kind_t fun_kind, mp_int_t arg_val1, int arg_reg1, mp_int_t arg_val2, int arg_reg2, mp_int_t arg_val3, int arg_reg3) {
    need_reg_all(emit);
    ASM_MOV_ALIGNED_IMM_TO_REG(emit->as, arg_val1, arg_reg1);
    ASM_MOV_IMM_TO_REG(emit->as, arg_val2, arg_reg2);
    ASM_MOV_IMM_TO_REG(emit->as, arg_val3, arg_reg3);
    ASM_CALL_IND(emit->as, mp_fun_table[fun_kind], fun_kind);
}

// vtype of all n_pop objects is VTYPE_PYOBJ
// Will convert any items that are not VTYPE_PYOBJ to this type and put them back on the stack.
// If any conversions of non-immediate values are needed, then it uses REG_ARG_1, REG_ARG_2 and REG_RET.
// Otherwise, it does not use any temporary registers (but may use reg_dest before loading it with stack pointer).
STATIC void emit_get_stack_pointer_to_reg_for_pop(emit_t *emit, mp_uint_t reg_dest, mp_uint_t n_pop) {
    need_reg_all(emit);

    // First, store any immediate values to their respective place on the stack.
    for (mp_uint_t i = 0; i < n_pop; i++) {
        stack_info_t *si = &emit->stack_info[emit->stack_size - 1 - i];
        // must push any imm's to stack
        // must convert them to VTYPE_PYOBJ for viper code
        if (si->kind == STACK_IMM) {
            si->kind = STACK_VALUE;
            switch (si->vtype) {
                case VTYPE_PYOBJ:
                    ASM_MOV_IMM_TO_LOCAL_USING(emit->as, si->u_imm, emit->stack_start + emit->stack_size - 1 - i, reg_dest);
                    break;
                case VTYPE_BOOL:
                    if (si->u_imm == 0) {
                        ASM_MOV_IMM_TO_LOCAL_USING(emit->as, (mp_uint_t)mp_const_false, emit->stack_start + emit->stack_size - 1 - i, reg_dest);
                    } else {
                        ASM_MOV_IMM_TO_LOCAL_USING(emit->as, (mp_uint_t)mp_const_true, emit->stack_start + emit->stack_size - 1 - i, reg_dest);
                    }
                    si->vtype = VTYPE_PYOBJ;
                    break;
                case VTYPE_INT:
                case VTYPE_UINT:
                    ASM_MOV_IMM_TO_LOCAL_USING(emit->as, (si->u_imm << 1) | 1, emit->stack_start + emit->stack_size - 1 - i, reg_dest);
                    si->vtype = VTYPE_PYOBJ;
                    break;
                default:
                    // not handled
                    assert(0);
            }
        }

        // verify that this value is on the stack
        assert(si->kind == STACK_VALUE);
    }

    // Second, convert any non-VTYPE_PYOBJ to that type.
    for (mp_uint_t i = 0; i < n_pop; i++) {
        stack_info_t *si = &emit->stack_info[emit->stack_size - 1 - i];
        if (si->vtype != VTYPE_PYOBJ) {
            mp_uint_t local_num = emit->stack_start + emit->stack_size - 1 - i;
            ASM_MOV_LOCAL_TO_REG(emit->as, local_num, REG_ARG_1);