circuitpython/py/binary.c

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

470 lines
15 KiB
C
Raw Permalink Normal View History

/*
* This file is part of the MicroPython project, http://micropython.org/
*
* The MIT License (MIT)
*
* Copyright (c) 2014-2017 Paul Sokolovsky
* Copyright (c) 2014-2019 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.
*/
#include <stdint.h>
#include <stdlib.h>
#include <stddef.h>
#include <string.h>
#include <assert.h>
#include "py/binary.h"
#include "py/smallint.h"
#include "py/objint.h"
#include "py/runtime.h"
// Helpers to work with binary-encoded data
#ifndef alignof
#define alignof(type) offsetof(struct { char c; type t; }, t)
#endif
size_t mp_binary_get_size(char struct_type, char val_type, size_t *palign) {
size_t size = 0;
int align = 1;
switch (struct_type) {
2021-03-15 09:57:36 -04:00
case '<':
case '>':
switch (val_type) {
2021-03-15 09:57:36 -04:00
case 'b':
case 'B':
case 'x':
size = 1;
break;
case 'h':
case 'H':
size = 2;
break;
case 'i':
case 'I':
size = 4;
break;
case 'l':
case 'L':
size = 4;
break;
case 'q':
case 'Q':
size = 8;
break;
#if MICROPY_NONSTANDARD_TYPECODES
case 'P':
case 'O':
case 'S':
size = sizeof(void *);
break;
#endif
case 'f':
2021-03-15 09:57:36 -04:00
size = sizeof(float);
break;
case 'd':
2021-03-15 09:57:36 -04:00
size = sizeof(double);
break;
}
break;
case '@': {
// TODO:
// The simplest heuristic for alignment is to align by value
// size, but that doesn't work for "bigger than int" types,
// for example, long long may very well have long alignment
// So, we introduce separate alignment handling, but having
// formal support for that is different from actually supporting
// particular (or any) ABI.
switch (val_type) {
case BYTEARRAY_TYPECODE:
2021-03-15 09:57:36 -04:00
case 'b':
case 'B':
case 'x':
align = size = 1;
break;
case 'h':
case 'H':
align = alignof(short);
2021-03-15 09:57:36 -04:00
size = sizeof(short);
break;
case 'i':
case 'I':
align = alignof(int);
2021-03-15 09:57:36 -04:00
size = sizeof(int);
break;
case 'l':
case 'L':
align = alignof(long);
2021-03-15 09:57:36 -04:00
size = sizeof(long);
break;
case 'q':
case 'Q':
align = alignof(long long);
2021-03-15 09:57:36 -04:00
size = sizeof(long long);
break;
#if MICROPY_NONSTANDARD_TYPECODES
case 'P':
case 'O':
case 'S':
align = alignof(void *);
size = sizeof(void *);
break;
#endif
case 'f':
align = alignof(float);
2021-03-15 09:57:36 -04:00
size = sizeof(float);
break;
case 'd':
align = alignof(double);
2021-03-15 09:57:36 -04:00
size = sizeof(double);
break;
}
}
}
if (size == 0) {
mp_raise_ValueError(MP_ERROR_TEXT("bad typecode"));
}
if (palign != NULL) {
*palign = align;
}
return size;
}
mp_obj_t mp_binary_get_val_array(char typecode, void *p, size_t index) {
mp_int_t val = 0;
switch (typecode) {
case 'b':
2021-03-15 09:57:36 -04:00
val = ((signed char *)p)[index];
break;
case BYTEARRAY_TYPECODE:
case 'B':
2021-03-15 09:57:36 -04:00
val = ((unsigned char *)p)[index];
break;
case 'h':
2021-03-15 09:57:36 -04:00
val = ((short *)p)[index];
break;
case 'H':
2021-03-15 09:57:36 -04:00
val = ((unsigned short *)p)[index];
break;
case 'i':
2021-03-15 09:57:36 -04:00
return mp_obj_new_int(((int *)p)[index]);
case 'I':
2021-03-15 09:57:36 -04:00
return mp_obj_new_int_from_uint(((unsigned int *)p)[index]);
case 'l':
2021-03-15 09:57:36 -04:00
return mp_obj_new_int(((long *)p)[index]);
case 'L':
2021-03-15 09:57:36 -04:00
return mp_obj_new_int_from_uint(((unsigned long *)p)[index]);
#if MICROPY_LONGINT_IMPL != MICROPY_LONGINT_IMPL_NONE
case 'q':
2021-03-15 09:57:36 -04:00
return mp_obj_new_int_from_ll(((long long *)p)[index]);
case 'Q':
2021-03-15 09:57:36 -04:00
return mp_obj_new_int_from_ull(((unsigned long long *)p)[index]);
#endif
2021-03-15 09:57:36 -04:00
#if MICROPY_PY_BUILTINS_FLOAT
case 'f':
return mp_obj_new_float_from_f(((float *)p)[index]);
case 'd':
return mp_obj_new_float_from_d(((double *)p)[index]);
2021-03-15 09:57:36 -04:00
#endif
#if MICROPY_NONSTANDARD_TYPECODES
// Extension to CPython: array of objects
case 'O':
2021-03-15 09:57:36 -04:00
return ((mp_obj_t *)p)[index];
// Extension to CPython: array of pointers
case 'P':
2021-03-15 09:57:36 -04:00
return mp_obj_new_int((mp_int_t)(uintptr_t)((void **)p)[index]);
#endif
}
return MP_OBJ_NEW_SMALL_INT(val);
}
// The long long type is guaranteed to hold at least 64 bits, and size is at
// most 8 (for q and Q), so we will always be able to parse the given data
// and fit it into a long long.
long long mp_binary_get_int(size_t size, bool is_signed, bool big_endian, const byte *src) {
int delta;
if (!big_endian) {
delta = -1;
src += size - 1;
} else {
delta = 1;
}
unsigned long long val = 0;
if (is_signed && *src & 0x80) {
val = -1;
}
for (uint i = 0; i < size; i++) {
// CIRCUITPY-CHANGE: fix for undefined behavior on left shift
val *= 256;
val |= *src;
src += delta;
}
return val;
}
#define is_signed(typecode) (typecode > 'Z')
mp_obj_t mp_binary_get_val(char struct_type, char val_type, byte *p_base, byte **ptr) {
byte *p = *ptr;
size_t align;
size_t size = mp_binary_get_size(struct_type, val_type, &align);
if (struct_type == '@') {
// Align p relative to p_base
p = p_base + (uintptr_t)MP_ALIGN(p - p_base, align);
#if MP_ENDIANNESS_LITTLE
struct_type = '<';
#else
struct_type = '>';
#endif
}
*ptr = p + size;
long long val = mp_binary_get_int(size, is_signed(val_type), (struct_type == '>'), p);
if (MICROPY_NONSTANDARD_TYPECODES && (val_type == 'O')) {
return (mp_obj_t)(mp_uint_t)val;
2021-03-15 09:57:36 -04:00
#if MICROPY_NONSTANDARD_TYPECODES
} else if (val_type == 'S') {
2021-03-15 09:57:36 -04:00
const char *s_val = (const char *)(uintptr_t)(mp_uint_t)val;
2018-07-28 13:29:47 -04:00
return mp_obj_new_str(s_val, strlen(s_val));
2021-03-15 09:57:36 -04:00
#endif
#if MICROPY_PY_BUILTINS_FLOAT
} else if (val_type == 'f') {
union {
uint32_t i;
float f;
2021-03-15 09:57:36 -04:00
} fpu = {val};
return mp_obj_new_float_from_f(fpu.f);
} else if (val_type == 'd') {
union {
uint64_t i;
double f;
2021-03-15 09:57:36 -04:00
} fpu = {val};
return mp_obj_new_float_from_d(fpu.f);
2021-03-15 09:57:36 -04:00
#endif
} else if (is_signed(val_type)) {
if ((long long)MP_SMALL_INT_MIN <= val && val <= (long long)MP_SMALL_INT_MAX) {
return mp_obj_new_int((mp_int_t)val);
} else {
return mp_obj_new_int_from_ll(val);
}
} else {
if ((unsigned long long)val <= (unsigned long long)MP_SMALL_INT_MAX) {
return mp_obj_new_int_from_uint((mp_uint_t)val);
} else {
return mp_obj_new_int_from_ull(val);
}
}
}
void mp_binary_set_int(size_t val_sz, bool big_endian, byte *dest, mp_uint_t val) {
if (MP_ENDIANNESS_LITTLE && !big_endian) {
memcpy(dest, &val, val_sz);
} else if (MP_ENDIANNESS_BIG && big_endian) {
// only copy the least-significant val_sz bytes
2021-03-15 09:57:36 -04:00
memcpy(dest, (byte *)&val + sizeof(mp_uint_t) - val_sz, val_sz);
} else {
const byte *src;
if (MP_ENDIANNESS_LITTLE) {
2021-03-15 09:57:36 -04:00
src = (const byte *)&val + val_sz;
} else {
2021-03-15 09:57:36 -04:00
src = (const byte *)&val + sizeof(mp_uint_t);
}
while (val_sz--) {
*dest++ = *--src;
}
}
}
void mp_binary_set_val(char struct_type, char val_type, mp_obj_t val_in, byte *p_base, byte **ptr) {
byte *p = *ptr;
size_t align;
size_t size = mp_binary_get_size(struct_type, val_type, &align);
if (struct_type == '@') {
// Align p relative to p_base
p = p_base + (uintptr_t)MP_ALIGN(p - p_base, align);
if (MP_ENDIANNESS_LITTLE) {
struct_type = '<';
} else {
struct_type = '>';
}
}
*ptr = p + size;
mp_uint_t val;
switch (val_type) {
2021-03-15 09:57:36 -04:00
#if MICROPY_NONSTANDARD_TYPECODES
case 'O':
val = (mp_uint_t)val_in;
break;
2021-03-15 09:57:36 -04:00
#endif
#if MICROPY_PY_BUILTINS_FLOAT
case 'f': {
union {
uint32_t i;
float f;
2021-03-15 09:57:36 -04:00
} fp_sp;
fp_sp.f = mp_obj_get_float_to_f(val_in);
val = fp_sp.i;
break;
}
case 'd': {
union {
uint64_t i64;
uint32_t i32[2];
double f;
2021-03-15 09:57:36 -04:00
} fp_dp;
fp_dp.f = mp_obj_get_float_to_d(val_in);
if (MP_BYTES_PER_OBJ_WORD == 8) {
val = fp_dp.i64;
} else {
int be = struct_type == '>';
mp_binary_set_int(sizeof(uint32_t), be, p, fp_dp.i32[MP_ENDIANNESS_BIG ^ be]);
p += sizeof(uint32_t);
val = fp_dp.i32[MP_ENDIANNESS_LITTLE ^ be];
}
break;
}
2021-03-15 09:57:36 -04:00
#endif
default: {
bool signed_type = is_signed(val_type);
#if MICROPY_LONGINT_IMPL != MICROPY_LONGINT_IMPL_NONE
py/obj: Add static safety checks to mp_obj_is_type(). Commit d96cfd13e3a464862c introduced a regression by breaking existing users of mp_obj_is_type(.., &mp_obj_bool). This function (and associated helpers like mp_obj_is_int()) have some specific nuances, and mistakes like this one can happen again. This commit adds mp_obj_is_exact_type() which behaves like the the old mp_obj_is_type(). The new mp_obj_is_type() has the same prototype but it attempts to statically assert that it's not called with types which should be checked using mp_obj_is_type(). If called with any of these types: int, str, bool, NoneType - it will cause a compilation error. Additional checked types (e.g function types) can be added in the future. Existing users of mp_obj_is_type() with the now "invalid" types, were translated to use mp_obj_is_exact_type(). The use of MP_STATIC_ASSERT() is not bulletproof - usually GCC (and other compilers) can't statically check conditions that are only known during link-time (like variables' addresses comparison). However, in this case, GCC is able to statically detect these conditions, probably because it's the exact same object - `&mp_type_int == &mp_type_int` is detected. Misuses of this function with runtime-chosen types (e.g: `mp_obj_type_t *x = ...; mp_obj_is_type(..., x);` won't be detected. MSC is unable to detect this, so we use MP_STATIC_ASSERT_NOT_MSC(). Compiling with this commit and without the fix for d96cfd13e3a464862c shows that it detects the problem. Signed-off-by: Yonatan Goldschmidt <yon.goldschmidt@gmail.com>
2020-01-22 07:34:19 -05:00
if (mp_obj_is_exact_type(val_in, &mp_type_int)) {
// It's a longint.
mp_obj_int_buffer_overflow_check(val_in, size, signed_type);
mp_obj_int_to_bytes_impl(val_in, struct_type == '>', size, p);
return;
}
#endif
{
val = mp_obj_get_int(val_in);
// Small int checking is separate, to be fast.
mp_small_int_buffer_overflow_check(val, size, signed_type);
// zero/sign extend if needed
if (MP_BYTES_PER_OBJ_WORD < 8 && size > sizeof(val)) {
int c = (is_signed(val_type) && (mp_int_t)val < 0) ? 0xff : 0x00;
memset(p, c, size);
if (struct_type == '>') {
p += size - sizeof(val);
}
}
break;
}
}
}
mp_binary_set_int(MIN((size_t)size, sizeof(val)), struct_type == '>', p, val);
}
void mp_binary_set_val_array(char typecode, void *p, size_t index, mp_obj_t val_in) {
switch (typecode) {
2021-03-15 09:57:36 -04:00
#if MICROPY_PY_BUILTINS_FLOAT
case 'f':
((float *)p)[index] = mp_obj_get_float_to_f(val_in);
break;
case 'd':
((double *)p)[index] = mp_obj_get_float_to_d(val_in);
break;
2021-03-15 09:57:36 -04:00
#endif
#if MICROPY_NONSTANDARD_TYPECODES
// Extension to CPython: array of objects
case 'O':
2021-03-15 09:57:36 -04:00
((mp_obj_t *)p)[index] = val_in;
break;
2021-03-15 09:57:36 -04:00
#endif
default: {
size_t size = mp_binary_get_size('@', typecode, NULL);
bool signed_type = is_signed(typecode);
#if MICROPY_LONGINT_IMPL != MICROPY_LONGINT_IMPL_NONE
py/obj: Add static safety checks to mp_obj_is_type(). Commit d96cfd13e3a464862c introduced a regression by breaking existing users of mp_obj_is_type(.., &mp_obj_bool). This function (and associated helpers like mp_obj_is_int()) have some specific nuances, and mistakes like this one can happen again. This commit adds mp_obj_is_exact_type() which behaves like the the old mp_obj_is_type(). The new mp_obj_is_type() has the same prototype but it attempts to statically assert that it's not called with types which should be checked using mp_obj_is_type(). If called with any of these types: int, str, bool, NoneType - it will cause a compilation error. Additional checked types (e.g function types) can be added in the future. Existing users of mp_obj_is_type() with the now "invalid" types, were translated to use mp_obj_is_exact_type(). The use of MP_STATIC_ASSERT() is not bulletproof - usually GCC (and other compilers) can't statically check conditions that are only known during link-time (like variables' addresses comparison). However, in this case, GCC is able to statically detect these conditions, probably because it's the exact same object - `&mp_type_int == &mp_type_int` is detected. Misuses of this function with runtime-chosen types (e.g: `mp_obj_type_t *x = ...; mp_obj_is_type(..., x);` won't be detected. MSC is unable to detect this, so we use MP_STATIC_ASSERT_NOT_MSC(). Compiling with this commit and without the fix for d96cfd13e3a464862c shows that it detects the problem. Signed-off-by: Yonatan Goldschmidt <yon.goldschmidt@gmail.com>
2020-01-22 07:34:19 -05:00
if (mp_obj_is_exact_type(val_in, &mp_type_int)) {
// It's a long int.
mp_obj_int_buffer_overflow_check(val_in, size, signed_type);
mp_obj_int_to_bytes_impl(val_in, MP_ENDIANNESS_BIG,
2021-03-15 09:57:36 -04:00
size, (uint8_t *)p + index * size);
return;
}
#endif
mp_int_t val = mp_obj_get_int(val_in);
// Small int checking is separate, to be fast.
mp_small_int_buffer_overflow_check(val, size, signed_type);
mp_binary_set_val_array_from_int(typecode, p, index, val);
}
}
}
void mp_binary_set_val_array_from_int(char typecode, void *p, size_t index, mp_int_t val) {
switch (typecode) {
case 'b':
2021-03-15 09:57:36 -04:00
((signed char *)p)[index] = val;
break;
case BYTEARRAY_TYPECODE:
case 'B':
2021-03-15 09:57:36 -04:00
((unsigned char *)p)[index] = val;
break;
case 'h':
2021-03-15 09:57:36 -04:00
((short *)p)[index] = val;
break;
case 'H':
2021-03-15 09:57:36 -04:00
((unsigned short *)p)[index] = val;
break;
case 'i':
2021-03-15 09:57:36 -04:00
((int *)p)[index] = val;
break;
case 'I':
2021-03-15 09:57:36 -04:00
((unsigned int *)p)[index] = val;
break;
case 'l':
2021-03-15 09:57:36 -04:00
((long *)p)[index] = val;
break;
case 'L':
2021-03-15 09:57:36 -04:00
((unsigned long *)p)[index] = val;
break;
#if MICROPY_LONGINT_IMPL != MICROPY_LONGINT_IMPL_NONE
case 'q':
2021-03-15 09:57:36 -04:00
((long long *)p)[index] = val;
break;
case 'Q':
2021-03-15 09:57:36 -04:00
((unsigned long long *)p)[index] = val;
break;
#endif
2021-03-15 09:57:36 -04:00
#if MICROPY_PY_BUILTINS_FLOAT
case 'f':
((float *)p)[index] = (float)val;
break;
case 'd':
((double *)p)[index] = (double)val;
break;
2021-03-15 09:57:36 -04:00
#endif
#if MICROPY_NONSTANDARD_TYPECODES
// Extension to CPython: array of pointers
case 'P':
2021-03-15 09:57:36 -04:00
((void **)p)[index] = (void *)(uintptr_t)val;
break;
2021-03-15 09:57:36 -04:00
#endif
}
}