scummvm/common/endian.h
Yotam Barnoy cdd27ca943 COMMON: changed read/write endian function to use __may_alias__ attribute
This is a better solution for the gcc aliasing problem that happens when aliasing a struct onto something else. What happens is that the compiler assumes no aliasing can happen when -O2 and -O3 are activated, and a call to READ_UINT32() followed by WRITE_UINT32() and another READ_UINT32() will be optimized to return the original read value instead of re-reading.

svn-id: r52480
2010-09-01 12:41:16 +00:00

413 lines
12 KiB
C

/* ScummVM - Graphic Adventure Engine
*
* ScummVM is the legal property of its developers, whose names
* are too numerous to list here. Please refer to the COPYRIGHT
* file distributed with this source distribution.
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version 2
* of the License, or (at your option) any later version.
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
*
* $URL$
* $Id$
*
*/
#ifndef COMMON_ENDIAN_H
#define COMMON_ENDIAN_H
#include "common/scummsys.h"
/**
* \file endian.h
* Endian conversion and byteswap conversion functions or macros
*
* SWAP_BYTES_??(a) - inverse byte order
* SWAP_CONSTANT_??(a) - inverse byte order, implemented as macro.
* Use with compiletime-constants only, the result will be a compiletime-constant aswell.
* Unlike most other functions these can be used for eg. switch-case labels
*
* READ_UINT??(a) - read native value from pointer a
* READ_??_UINT??(a) - read LE/BE value from pointer a and convert it to native
* WRITE_??_UINT??(a, v) - write native value v to pointer a with LE/BE encoding
* TO_??_??(a) - convert native value v to LE/BE
* FROM_??_??(a) - convert LE/BE value v to native
* CONSTANT_??_??(a) - convert LE/BE value v to native, implemented as macro.
* Use with compiletime-constants only, the result will be a compiletime-constant aswell.
* Unlike most other functions these can be used for eg. switch-case labels
*/
// Sanity check
#if !defined(SCUMM_LITTLE_ENDIAN) && !defined(SCUMM_BIG_ENDIAN)
# error No endianness defined
#endif
#define SWAP_CONSTANT_32(a) \
((uint32)((((a) >> 24) & 0x00FF) | \
(((a) >> 8) & 0xFF00) | \
(((a) & 0xFF00) << 8) | \
(((a) & 0x00FF) << 24) ))
#define SWAP_CONSTANT_16(a) \
((uint16)((((a) >> 8) & 0x00FF) | \
(((a) << 8) & 0xFF00) ))
/**
* Swap the bytes in a 32 bit word in order to convert LE encoded data to BE
* and vice versa.
*/
// machine/compiler-specific variants come first, fallback last
// Test for GCC and if the target has the MIPS rel.2 instructions (we know the psp does)
#if defined(__GNUC__) && (defined(__psp__) || defined(_MIPS_ARCH_MIPS32R2) || defined(_MIPS_ARCH_MIPS64R2))
FORCEINLINE uint32 SWAP_BYTES_32(const uint32 a) {
if (__builtin_constant_p(a)) {
return SWAP_CONSTANT_32(a);
} else {
uint32 result;
# if defined(__psp__)
// use special allegrex instruction
__asm__ ("wsbw %0,%1" : "=r" (result) : "r" (a));
# else
__asm__ ("wsbh %0,%1\n"
"rotr %0,%0,16" : "=r" (result) : "r" (a));
# endif
return result;
}
}
// Test for GCC >= 4.3.0 as this version added the bswap builtin
#elif defined(__GNUC__) && (__GNUC__ > 4 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 3))
FORCEINLINE uint32 SWAP_BYTES_32(uint32 a) {
return __builtin_bswap32(a);
}
// test for MSVC 7 or newer
#elif defined(_MSC_VER) && _MSC_VER >= 1300
FORCEINLINE uint32 SWAP_BYTES_32(uint32 a) {
return _byteswap_ulong(a);
}
// generic fallback
#else
inline uint32 SWAP_BYTES_32(uint32 a) {
const uint16 low = (uint16)a, high = (uint16)(a >> 16);
return ((uint32)(uint16)((low >> 8) | (low << 8)) << 16)
| (uint16)((high >> 8) | (high << 8));
}
#endif
/**
* Swap the bytes in a 16 bit word in order to convert LE encoded data to BE
* and vice versa.
*/
// compilerspecific variants come first, fallback last
// Test for GCC and if the target has the MIPS rel.2 instructions (we know the psp does)
#if defined(__GNUC__) && (defined(__psp__) || defined(_MIPS_ARCH_MIPS32R2) || defined(_MIPS_ARCH_MIPS64R2))
FORCEINLINE uint16 SWAP_BYTES_16(const uint16 a) {
if (__builtin_constant_p(a)) {
return SWAP_CONSTANT_16(a);
} else {
uint16 result;
__asm__ ("wsbh %0,%1" : "=r" (result) : "r" (a));
return result;
}
}
#else
inline uint16 SWAP_BYTES_16(const uint16 a) {
return (a >> 8) | (a << 8);
}
#endif
/**
* A wrapper macro used around four character constants, like 'DATA', to
* ensure portability. Typical usage: MKID_BE('DATA').
*
* Why is this necessary? The C/C++ standard does not define the endianess to
* be used for character constants. Hence if one uses multi-byte character
* constants, a potential portability problem opens up.
*
* Fortunately, a semi-standard has been established: On almost all systems
* and compilers, multi-byte character constants are encoded using the big
* endian convention (probably in analogy to the encoding of string constants).
* Still some systems differ. This is why we provide the MKID_BE macro. If
* you wrap your four character constants with it, the result will always be
* BE encoded, even on systems which differ from the default BE encoding.
*
* For the latter systems we provide the INVERSE_MKID override.
*/
#if defined(INVERSE_MKID)
#define MKID_BE(a) SWAP_CONSTANT_32(a)
#else
# define MKID_BE(a) ((uint32)(a))
#endif
// Functions for reading/writing native Integers,
// this transparently handles the need for alignment
#if !defined(SCUMM_NEED_ALIGNMENT)
FORCEINLINE uint16 READ_UINT16(const void *ptr) {
return *(const uint16 *)(ptr);
}
FORCEINLINE uint32 READ_UINT32(const void *ptr) {
return *(const uint32 *)(ptr);
}
FORCEINLINE void WRITE_UINT16(void *ptr, uint16 value) {
*(uint16 *)(ptr) = value;
}
FORCEINLINE void WRITE_UINT32(void *ptr, uint32 value) {
*(uint32 *)(ptr) = value;
}
// test for GCC >= 4.0. these implementations will automatically use CPU-specific
// instructions for unaligned data when they are available (eg. MIPS)
#elif defined(__GNUC__) && (__GNUC__ >= 4)
FORCEINLINE uint16 READ_UINT16(const void *ptr) {
struct Unaligned16 { uint16 val; } __attribute__ ((__packed__, __may_alias__));
return ((const Unaligned16 *)ptr)->val;
}
FORCEINLINE uint32 READ_UINT32(const void *ptr) {
struct Unaligned32 { uint32 val; } __attribute__ ((__packed__, __may_alias__));
return ((const Unaligned32 *)ptr)->val;
}
FORCEINLINE void WRITE_UINT16(void *ptr, uint16 value) {
struct Unaligned16 { uint16 val; } __attribute__ ((__packed__, __may_alias__));
((Unaligned16 *)ptr)->val = value;
}
FORCEINLINE void WRITE_UINT32(void *ptr, uint32 value) {
struct Unaligned32 { uint32 val; } __attribute__ ((__packed__, __may_alias__));
((Unaligned32 *)ptr)->val = value;
}
// use software fallback by loading each byte explicitely
#else
# if defined(SCUMM_LITTLE_ENDIAN)
inline uint16 READ_UINT16(const void *ptr) {
const uint8 *b = (const uint8 *)ptr;
return (b[1] << 8) | b[0];
}
inline uint32 READ_UINT32(const void *ptr) {
const uint8 *b = (const uint8 *)ptr;
return (b[3] << 24) | (b[2] << 16) | (b[1] << 8) | (b[0]);
}
inline void WRITE_UINT16(void *ptr, uint16 value) {
uint8 *b = (uint8 *)ptr;
b[0] = (uint8)(value >> 0);
b[1] = (uint8)(value >> 8);
}
inline void WRITE_UINT32(void *ptr, uint32 value) {
uint8 *b = (uint8 *)ptr;
b[0] = (uint8)(value >> 0);
b[1] = (uint8)(value >> 8);
b[2] = (uint8)(value >> 16);
b[3] = (uint8)(value >> 24);
}
# elif defined(SCUMM_BIG_ENDIAN)
inline uint16 READ_UINT16(const void *ptr) {
const uint8 *b = (const uint8 *)ptr;
return (b[0] << 8) | b[1];
}
inline uint32 READ_UINT32(const void *ptr) {
const uint8 *b = (const uint8 *)ptr;
return (b[0] << 24) | (b[1] << 16) | (b[2] << 8) | (b[3]);
}
inline void WRITE_UINT16(void *ptr, uint16 value) {
uint8 *b = (uint8 *)ptr;
b[0] = (uint8)(value >> 8);
b[1] = (uint8)(value >> 0);
}
inline void WRITE_UINT32(void *ptr, uint32 value) {
uint8 *b = (uint8 *)ptr;
b[0] = (uint8)(value >> 24);
b[1] = (uint8)(value >> 16);
b[2] = (uint8)(value >> 8);
b[3] = (uint8)(value >> 0);
}
# endif
#endif
// Map Funtions for reading/writing BE/LE integers depending on native endianess
#if defined(SCUMM_LITTLE_ENDIAN)
#define READ_LE_UINT16(a) READ_UINT16(a)
#define READ_LE_UINT32(a) READ_UINT32(a)
#define WRITE_LE_UINT16(a, v) WRITE_UINT16(a, v)
#define WRITE_LE_UINT32(a, v) WRITE_UINT32(a, v)
#define FROM_LE_32(a) ((uint32)(a))
#define FROM_LE_16(a) ((uint16)(a))
#define FROM_BE_32(a) SWAP_BYTES_32(a)
#define FROM_BE_16(a) SWAP_BYTES_16(a)
#define TO_LE_32(a) ((uint32)(a))
#define TO_LE_16(a) ((uint16)(a))
#define TO_BE_32(a) SWAP_BYTES_32(a)
#define TO_BE_16(a) SWAP_BYTES_16(a)
#define CONSTANT_LE_32(a) ((uint32)(a))
#define CONSTANT_LE_16(a) ((uint16)(a))
#define CONSTANT_BE_32(a) SWAP_CONSTANT_32(a)
#define CONSTANT_BE_16(a) SWAP_CONSTANT_16(a)
// if the unaligned load and the byteswap take alot instructions its better to directly read and invert
# if defined(SCUMM_NEED_ALIGNMENT) && !defined(__mips__)
inline uint16 READ_BE_UINT16(const void *ptr) {
const uint8 *b = (const uint8 *)ptr;
return (b[0] << 8) | b[1];
}
inline uint32 READ_BE_UINT32(const void *ptr) {
const uint8 *b = (const uint8 *)ptr;
return (b[0] << 24) | (b[1] << 16) | (b[2] << 8) | (b[3]);
}
inline void WRITE_BE_UINT16(void *ptr, uint16 value) {
uint8 *b = (uint8 *)ptr;
b[0] = (uint8)(value >> 8);
b[1] = (uint8)(value >> 0);
}
inline void WRITE_BE_UINT32(void *ptr, uint32 value) {
uint8 *b = (uint8 *)ptr;
b[0] = (uint8)(value >> 24);
b[1] = (uint8)(value >> 16);
b[2] = (uint8)(value >> 8);
b[3] = (uint8)(value >> 0);
}
# else
inline uint16 READ_BE_UINT16(const void *ptr) {
return SWAP_BYTES_16(READ_UINT16(ptr));
}
inline uint32 READ_BE_UINT32(const void *ptr) {
return SWAP_BYTES_32(READ_UINT32(ptr));
}
inline void WRITE_BE_UINT16(void *ptr, uint16 value) {
WRITE_UINT16(ptr, SWAP_BYTES_16(value));
}
inline void WRITE_BE_UINT32(void *ptr, uint32 value) {
WRITE_UINT32(ptr, SWAP_BYTES_32(value));
}
# endif // if defined(SCUMM_NEED_ALIGNMENT)
#elif defined(SCUMM_BIG_ENDIAN)
#define MKID_BE(a) ((uint32)(a))
#define READ_BE_UINT16(a) READ_UINT16(a)
#define READ_BE_UINT32(a) READ_UINT32(a)
#define WRITE_BE_UINT16(a, v) WRITE_UINT16(a, v)
#define WRITE_BE_UINT32(a, v) WRITE_UINT32(a, v)
#define FROM_LE_32(a) SWAP_BYTES_32(a)
#define FROM_LE_16(a) SWAP_BYTES_16(a)
#define FROM_BE_32(a) ((uint32)(a))
#define FROM_BE_16(a) ((uint16)(a))
#define TO_LE_32(a) SWAP_BYTES_32(a)
#define TO_LE_16(a) SWAP_BYTES_16(a)
#define TO_BE_32(a) ((uint32)(a))
#define TO_BE_16(a) ((uint16)(a))
#define CONSTANT_LE_32(a) SWAP_CONSTANT_32(a)
#define CONSTANT_LE_16(a) SWAP_CONSTANT_16(a)
#define CONSTANT_BE_32(a) ((uint32)(a))
#define CONSTANT_BE_16(a) ((uint16)(a))
// if the unaligned load and the byteswap take alot instructions its better to directly read and invert
# if defined(SCUMM_NEED_ALIGNMENT) && !defined(__mips__)
inline uint16 READ_LE_UINT16(const void *ptr) {
const uint8 *b = (const uint8 *)ptr;
return (b[1] << 8) | b[0];
}
inline uint32 READ_LE_UINT32(const void *ptr) {
const uint8 *b = (const uint8 *)ptr;
return (b[3] << 24) | (b[2] << 16) | (b[1] << 8) | (b[0]);
}
inline void WRITE_LE_UINT16(void *ptr, uint16 value) {
uint8 *b = (uint8 *)ptr;
b[0] = (uint8)(value >> 0);
b[1] = (uint8)(value >> 8);
}
inline void WRITE_LE_UINT32(void *ptr, uint32 value) {
uint8 *b = (uint8 *)ptr;
b[0] = (uint8)(value >> 0);
b[1] = (uint8)(value >> 8);
b[2] = (uint8)(value >> 16);
b[3] = (uint8)(value >> 24);
}
# else
inline uint16 READ_LE_UINT16(const void *ptr) {
return SWAP_BYTES_16(READ_UINT16(ptr));
}
inline uint32 READ_LE_UINT32(const void *ptr) {
return SWAP_BYTES_32(READ_UINT32(ptr));
}
inline void WRITE_LE_UINT16(void *ptr, uint16 value) {
WRITE_UINT16(ptr, SWAP_BYTES_16(value));
}
inline void WRITE_LE_UINT32(void *ptr, uint32 value) {
WRITE_UINT32(ptr, SWAP_BYTES_32(value));
}
# endif // if defined(SCUMM_NEED_ALIGNMENT)
#endif // if defined(SCUMM_LITTLE_ENDIAN)
inline uint32 READ_LE_UINT24(const void *ptr) {
const uint8 *b = (const uint8 *)ptr;
return (b[2] << 16) | (b[1] << 8) | (b[0]);
}
inline uint32 READ_BE_UINT24(const void *ptr) {
const uint8 *b = (const uint8 *)ptr;
return (b[0] << 16) | (b[1] << 8) | (b[2]);
}
#endif