?? aesopt.h
字號:
/*
-------------------------------------------------------------------------
Copyright (c) 2001, Dr Brian Gladman < >, Worcester, UK.
All rights reserved.
LICENSE TERMS
The free distribution and use of this software in both source and binary
form is allowed (with or without changes) provided that:
1. distributions of this source code include the above copyright
notice, this list of conditions and the following disclaimer;
2. distributions in binary form include the above copyright
notice, this list of conditions and the following disclaimer
in the documentation and/or other associated materials;
3. the copyright holder's name is not used to endorse products
built using this software without specific written permission.
DISCLAIMER
This software is provided 'as is' with no explicit or implied warranties
in respect of its properties, including, but not limited to, correctness
and fitness for purpose.
-------------------------------------------------------------------------
Issue Date: 29/07/2002
This file contains the compilation options for AES (Rijndael) and code
that is common across encryption, key scheduling and table generation.
OPERATION
These source code files implement the AES algorithm Rijndael designed by
Joan Daemen and Vincent Rijmen. The version in aes.c is designed for
block and key sizes of 128, 192 and 256 bits (16, 24 and 32 bytes) while
that in aespp.c provides for block and keys sizes of 128, 160, 192, 224
and 256 bits (16, 20, 24, 28 and 32 bytes). This file is a common header
file for these two implementations and for aesref.c, which is a reference
implementation.
This version is designed for flexibility and speed using operations on
32-bit words rather than operations on bytes. It provides aes_both fixed
and dynamic block and key lengths and can also run with either big or
little endian internal byte order (see aes.h). It inputs block and key
lengths in bytes with the legal values being 16, 24 and 32 for aes.c and
16, 20, 24, 28 and 32 for aespp.c
THE CIPHER INTERFACE
aes_08t (an unsigned 8-bit type)
aes_32t (an unsigned 32-bit type)
aes_fret (a signed 16 bit type for function return values)
aes_good (value != 0, a good return)
aes_bad (value == 0, an error return)
struct aes_ctx (structure for the cipher encryption context)
struct aes_ctx (structure for the cipher decryption context)
aes_rval the function return type (aes_fret if not DLL)
C subroutine calls:
aes_rval aes_blk_len(unsigned int blen, aes_ctx cx[1]);
aes_rval aes_enc_key(const unsigned char in_key[], unsigned int klen, aes_ctx cx[1]);
aes_rval aes_enc_blk(const unsigned char in_blk[], unsigned char out_blk[], const aes_ctx cx[1]);
aes_rval aes_dec_len(unsigned int blen, aes_ctx cx[1]);
aes_rval aes_dec_key(const unsigned char in_key[], unsigned int klen, aes_ctx cx[1]);
aes_rval aes_dec_blk(const unsigned char in_blk[], unsigned char out_blk[], const aes_ctx cx[1]);
IMPORTANT NOTE: If you are using this C interface and your compiler does
not set the memory used for objects to zero before use, you will need to
ensure that cx.s_flg is set to zero before using these subroutine calls.
C++ aes class subroutines:
class AESclass for encryption
class AESclass for decryption
aes_rval len(unsigned int blen = 16);
aes_rval key(const unsigned char in_key[], unsigned int klen);
aes_rval blk(const unsigned char in_blk[], unsigned char out_blk[]);
aes_rval len(unsigned int blen = 16);
aes_rval key(const unsigned char in_key[], unsigned int klen);
aes_rval blk(const unsigned char in_blk[], unsigned char out_blk[]);
The block length inputs to set_block and set_key are in numbers of
BYTES, not bits. The calls to subroutines must be made in the above
order but multiple calls can be made without repeating earlier calls
if their parameters have not changed. If the cipher block length is
variable but set_blk has not been called before cipher operations a
value of 16 is assumed (that is, the AES block size). In contrast to
earlier versions the block and key length parameters are now checked
for correctness and the encryption and decryption routines check to
ensure that an appropriate key has been set before they are called.
COMPILATION
The files used to provide AES (Rijndael) are
a. aes.h for the definitions needed for use in C.
b. aescpp.h for the definitions needed for use in C++.
c. aesopt.h for setting compilation options (also includes common
code).
d. aescrypt.c for encryption and decrytpion, or
e. aescrypt.asm for encryption and decryption using assembler code.
f. aeskey.c for key scheduling.
g. aestab.c for table loading or generation.
The assembler code uses the NASM assembler. The above files provice
block and key lengths of 16, 24 and 32 bytes (128, 192 and 256 bits).
If aescrypp.c and aeskeypp.c are used instead of aescrypt.c and
aeskey.c respectively, the block and key lengths can then be 16, 20,
24, 28 or 32 bytes. However this code has not been optimised to the
same extent and is hence slower (esepcially for the AES block size
of 16 bytes).
To compile AES (Rijndael) for use in C code use aes.h and exclude
the AES_DLL define in aes.h
To compile AES (Rijndael) for use in in C++ code use aescpp.h and
exclude the AES_DLL define in aes.h
To compile AES (Rijndael) in C as a Dynamic Link Library DLL) use
aes.h, include the AES_DLL define and compile the DLL. If using
the test files to test the DLL, exclude aes.c from the test build
project and compile it with the same defines as used for the DLL
(ensure that the DLL path is correct)
CONFIGURATION OPTIONS (here and in aes.h)
a. define BLOCK_SIZE in aes.h to set the cipher block size (16, 24
or 32 for the standard code, or 16, 20, 24, 28 or 32 for the
extended code) or leave this undefined for dynamically variable
block size (this will result in much slower code).
b. set AES_DLL in aes.h if AES (Rijndael) is to be compiled as a DLL
c. You may need to set PLATFORM_BYTE_ORDER to define the byte order.
d. If you want the code to run in a specific internal byte order, then
INTERNAL_BYTE_ORDER must be set accordingly.
e. set other configuration options decribed below.
*/
#ifndef _AESOPT_H
#define _AESOPT_H
/* START OF CONFIGURATION OPTIONS
USE OF DEFINES
Later in this section there are a number of defines that control the
operation of the code. In each section, the purpose of each define is
explained so that the relevant form can be included or excluded by
setting either 1's or 0's respectively on the branches of the related
#if clauses.
*/
/* 1. PLATFORM SPECIFIC INCLUDES */
#if defined( __CRYPTLIB__ ) && !defined( INC_ALL ) && !defined( INC_CHILD )
#include "crypt/aes.h"
#else
#include "aes.h"
#endif
#if defined(__GNUC__) || defined(__GNU_LIBRARY__)
# include <endian.h>
# include <byteswap.h>
#elif defined(__CRYPTLIB__)
# if defined( INC_ALL )
# include "crypt.h"
# elif defined( INC_CHILD )
# include "../crypt.h"
# else
# include "crypt.h"
# endif
# if defined(DATA_LITTLEENDIAN)
# define PLATFORM_BYTE_ORDER AES_LITTLE_ENDIAN
# else
# define PLATFORM_BYTE_ORDER AES_BIG_ENDIAN
# endif
#elif defined(_MSC_VER)
# include <stdlib.h>
#elif !defined(WIN32)
# include <stdlib.h>
# if !defined (_ENDIAN_H)
# include <sys/param.h>
# else
# include _ENDIAN_H
# endif
#endif
/* 2. BYTE ORDER IN 32-BIT WORDS
To obtain the highest speed on processors with 32-bit words, this code
needs to determine the order in which bytes are packed into such words.
The following block of code is an attempt to capture the most obvious
ways in which various environemnts define byte order. It may well fail,
in which case the definitions will need to be set by editing at the
points marked **** EDIT HERE IF NECESSARY **** below.
*/
#define AES_LITTLE_ENDIAN 1234 /* byte 0 is least significant (i386) */
#define AES_BIG_ENDIAN 4321 /* byte 0 is most significant (mc68k) */
#if !defined(PLATFORM_BYTE_ORDER)
#if defined(LITTLE_ENDIAN) || defined(BIG_ENDIAN)
# if defined(LITTLE_ENDIAN) && defined(BIG_ENDIAN)
# if defined(BYTE_ORDER)
# if (BYTE_ORDER == LITTLE_ENDIAN)
# define PLATFORM_BYTE_ORDER AES_LITTLE_ENDIAN
# elif (BYTE_ORDER == BIG_ENDIAN)
# define PLATFORM_BYTE_ORDER AES_BIG_ENDIAN
# endif
# endif
# elif defined(LITTLE_ENDIAN) && !defined(BIG_ENDIAN)
# define PLATFORM_BYTE_ORDER AES_LITTLE_ENDIAN
# elif !defined(LITTLE_ENDIAN) && defined(BIG_ENDIAN)
# define PLATFORM_BYTE_ORDER AES_BIG_ENDIAN
# endif
#elif defined(_LITTLE_ENDIAN) || defined(_BIG_ENDIAN)
# if defined(_LITTLE_ENDIAN) && defined(_BIG_ENDIAN)
# if defined(_BYTE_ORDER)
# if (_BYTE_ORDER == _LITTLE_ENDIAN)
# define PLATFORM_BYTE_ORDER AES_LITTLE_ENDIAN
# elif (_BYTE_ORDER == _BIG_ENDIAN)
# define PLATFORM_BYTE_ORDER AES_BIG_ENDIAN
# endif
# endif
# elif defined(_LITTLE_ENDIAN) && !defined(_BIG_ENDIAN)
# define PLATFORM_BYTE_ORDER AES_LITTLE_ENDIAN
# elif !defined(_LITTLE_ENDIAN) && defined(_BIG_ENDIAN)
# define PLATFORM_BYTE_ORDER AES_BIG_ENDIAN
# endif
#elif 0 /* **** EDIT HERE IF NECESSARY **** */
#define PLATFORM_BYTE_ORDER AES_LITTLE_ENDIAN
#elif 0 /* **** EDIT HERE IF NECESSARY **** */
#define PLATFORM_BYTE_ORDER AES_BIG_ENDIAN
#elif (('1234' >> 24) == '1')
# define PLATFORM_BYTE_ORDER AES_LITTLE_ENDIAN
#elif (('4321' >> 24) == '1')
# define PLATFORM_BYTE_ORDER AES_BIG_ENDIAN
#endif
#endif
#if !defined(PLATFORM_BYTE_ORDER)
# error Please set undetermined byte order (lines 233 or 235 of aesopt.h).
#endif
/* 3. ASSEMBLER SUPPORT
If the assembler code is used for encryption and decryption this file only
provides key scheduling so the following defines are used
*/
#ifdef AES_ASM
#define ENCRYPTION_KEY_SCHEDULE
#define DECRYPTION_KEY_SCHEDULE
#else
/* 4. FUNCTIONS REQUIRED
This implementation provides five main subroutines which provide for
setting block length, setting encryption and decryption keys and for
encryption and decryption. When the assembler code is not being used
the following definition blocks allow the selection of the routines
that are to be included in the compilation.
*/
#if 1
#define ENCRYPTION_KEY_SCHEDULE
#endif
#if 1
#define DECRYPTION_KEY_SCHEDULE
#endif
#if 1
#define ENCRYPTION
#endif
#if 1
#define DECRYPTION
#endif
#endif
/* 5. BYTE ORDER WITHIN 32 BIT WORDS
The fundamental data processing units in Rijndael are 8-bit bytes. The
input, output and key input are all enumerated arrays of bytes in which
bytes are numbered starting at zero and increasing to one less than the
number of bytes in the array in question. This enumeration is only used
for naming bytes and does not imply any adjacency or order relationship
from one byte to another. When these inputs and outputs are considered
as bit sequences, bits 8*n to 8*n+7 of the bit sequence are mapped to
byte[n] with bit 8n+i in the sequence mapped to bit 7-i within the byte.
In this implementation bits are numbered from 0 to 7 starting at the
numerically least significant end of each byte (bit n represents 2^n).
However, Rijndael can be implemented more efficiently using 32-bit
words by packing bytes into words so that bytes 4*n to 4*n+3 are placed
into word[n]. While in principle these bytes can be assembled into words
in any positions, this implementation only supports the two formats in
which bytes in adjacent positions within words also have adjacent byte
numbers. This order is called big-endian if the lowest numbered bytes
in words have the highest numeric significance and little-endian if the
opposite applies.
This code can work in either order irrespective of the order used by the
machine on which it runs. Normally the internal byte order will be set
to the order of the processor on which the code is to be run but this
define can be used to reverse this in special situations
*/
#if 1
#define INTERNAL_BYTE_ORDER PLATFORM_BYTE_ORDER
#elif defined(AES_LITTLE_ENDIAN)
#define INTERNAL_BYTE_ORDER AES_LITTLE_ENDIAN
#elif defined(AES_BIG_ENDIAN)
#define INTERNAL_BYTE_ORDER AES_BIG_ENDIAN
#endif
/* 6. FAST INPUT/OUTPUT OPERATIONS.
On some machines it is possible to improve speed by transferring the
bytes in the input and output arrays to and from the internal 32-bit
variables by addressing these arrays as if they are arrays of 32-bit
words. On some machines this will always be possible but there may
be a large performance penalty if the byte arrays are not aligned on
the normal word boundaries. On other machines this technique will
lead to memory access errors when such 32-bit word accesses are not
properly aligned. The option SAFE_IO avoids such problems but will
often be slower on those machines that support misaligned access
(especially so if care is taken to align the input and output byte
arrays on 32-bit word boundaries). If SAFE_IO is not defined it is
assumed that access to byte arrays as if they are arrays of 32-bit
words will not cause problems when such accesses are misaligned.
*/
#if 1
#define SAFE_IO
#endif
/* 7. LOOP UNROLLING
The code for encryption and decrytpion cycles through a number of rounds
that can be implemented either in a loop or by expanding the code into a
long sequence of instructions, the latter producing a larger program but
one that will often be much faster. The latter is called loop unrolling.
There are also potential speed advantages in expanding two iterations in
a loop with half the number of iterations, which is called partial loop
unrolling. The following options allow partial or full loop unrolling
to be set independently for encryption and decryption
*/
#if 1
#define ENC_UNROLL FULL
#elif 0
#define ENC_UNROLL PARTIAL
#else
#define ENC_UNROLL NONE
#endif
#if 1
#define DEC_UNROLL FULL
#elif 0
#define DEC_UNROLL PARTIAL
#else
#define DEC_UNROLL NONE
#endif
/* 8. FIXED OR DYNAMIC TABLES
When this section is included the tables used by the code are comipled
statically into the binary file. Otherwise they are computed once when
the code is first used.
*/
#if 1
#define FIXED_TABLES
#endif
/* 9. FAST FINITE FIELD OPERATIONS
If this section is included, tables are used to provide faster finite
field arithmetic (this has no effect if FIXED_TABLES is defined).
*/
#if 1
#define FF_TABLES
#endif
/* 10. INTERNAL STATE VARIABLE FORMAT
The internal state of Rijndael is stored in a number of local 32-bit
word varaibles which can be defined either as an array or as individual
names variables. Include this section if you want to store these local
varaibles in arrays. Otherwise individual local variables will be used.
*/
#if 1
#define ARRAYS
#endif
/* In this implementation the columns of the state array are each held in
32-bit words. The state array can be held in various ways: in an array
of words, in a number of individual word variables or in a number of
processor registers. The following define maps a variable name x and
a column number c to the way the state array variable is to be held.
The first define below maps the state into an array x[c] whereas the
second form maps the state into a number of individual variables x0,
x1, etc. Another form could map individual state colums to machine
register names.
*/
#if defined(ARRAYS)
#define s(x,c) x[c]
#else
#define s(x,c) x##c
#endif
/* 11. VARIABLE BLOCK SIZE SPEED
This section is only relevant if you wish to use the variable block
?? 快捷鍵說明
復制代碼
Ctrl + C
搜索代碼
Ctrl + F
全屏模式
F11
切換主題
Ctrl + Shift + D
顯示快捷鍵
?
增大字號
Ctrl + =
減小字號
Ctrl + -