com.fitbur.bouncycastle.crypto.engines.AESLightEngine Maven / Gradle / Ivy
package com.fitbur.bouncycastle.crypto.engines;
import com.fitbur.bouncycastle.crypto.BlockCipher;
import com.fitbur.bouncycastle.crypto.CipherParameters;
import com.fitbur.bouncycastle.crypto.DataLengthException;
import com.fitbur.bouncycastle.crypto.OutputLengthException;
import com.fitbur.bouncycastle.crypto.params.KeyParameter;
/**
* an implementation of the AES (Rijndael), from FIPS-197.
*
* For further com.fitburtails see: http://csrc.nist.gov/encryption/aes/.
*
* This implementation is based on optimizations from Dr. Brian Gladman's paper and C code at
* http://fp.gladman.plus.com.fitbur/cryptography_technology/rijndael/
*
* There are three levels of tradeoff of speed vs memory
* Because java has no preprocessor, they are written as three separate classes from which to choose
*
* The fastest uses 8Kbytes of static tables to precompute round calculations, 4 256 word tables for encryption
* and 4 for com.fitburcryption.
*
* The middle performance version uses only one 256 word table for each, for a total of 2Kbytes,
* adding 12 rotate operations per round to com.fitburpute the values contained in the other tables from
* the contents of the first
*
* The slowest version uses no static tables at all and com.fitburputes the values
* in each round.
*
* This file contains the slowest performance version with no static tables
* for round precomputation, but it has the smallest foot print.
*
*/
public class AESLightEngine
implements BlockCipher
{
// The S box
private static final byte[] S = {
(byte)99, (byte)124, (byte)119, (byte)123, (byte)242, (byte)107, (byte)111, (byte)197,
(byte)48, (byte)1, (byte)103, (byte)43, (byte)254, (byte)215, (byte)171, (byte)118,
(byte)202, (byte)130, (byte)201, (byte)125, (byte)250, (byte)89, (byte)71, (byte)240,
(byte)173, (byte)212, (byte)162, (byte)175, (byte)156, (byte)164, (byte)114, (byte)192,
(byte)183, (byte)253, (byte)147, (byte)38, (byte)54, (byte)63, (byte)247, (byte)204,
(byte)52, (byte)165, (byte)229, (byte)241, (byte)113, (byte)216, (byte)49, (byte)21,
(byte)4, (byte)199, (byte)35, (byte)195, (byte)24, (byte)150, (byte)5, (byte)154,
(byte)7, (byte)18, (byte)128, (byte)226, (byte)235, (byte)39, (byte)178, (byte)117,
(byte)9, (byte)131, (byte)44, (byte)26, (byte)27, (byte)110, (byte)90, (byte)160,
(byte)82, (byte)59, (byte)214, (byte)179, (byte)41, (byte)227, (byte)47, (byte)132,
(byte)83, (byte)209, (byte)0, (byte)237, (byte)32, (byte)252, (byte)177, (byte)91,
(byte)106, (byte)203, (byte)190, (byte)57, (byte)74, (byte)76, (byte)88, (byte)207,
(byte)208, (byte)239, (byte)170, (byte)251, (byte)67, (byte)77, (byte)51, (byte)133,
(byte)69, (byte)249, (byte)2, (byte)127, (byte)80, (byte)60, (byte)159, (byte)168,
(byte)81, (byte)163, (byte)64, (byte)143, (byte)146, (byte)157, (byte)56, (byte)245,
(byte)188, (byte)182, (byte)218, (byte)33, (byte)16, (byte)255, (byte)243, (byte)210,
(byte)205, (byte)12, (byte)19, (byte)236, (byte)95, (byte)151, (byte)68, (byte)23,
(byte)196, (byte)167, (byte)126, (byte)61, (byte)100, (byte)93, (byte)25, (byte)115,
(byte)96, (byte)129, (byte)79, (byte)220, (byte)34, (byte)42, (byte)144, (byte)136,
(byte)70, (byte)238, (byte)184, (byte)20, (byte)222, (byte)94, (byte)11, (byte)219,
(byte)224, (byte)50, (byte)58, (byte)10, (byte)73, (byte)6, (byte)36, (byte)92,
(byte)194, (byte)211, (byte)172, (byte)98, (byte)145, (byte)149, (byte)228, (byte)121,
(byte)231, (byte)200, (byte)55, (byte)109, (byte)141, (byte)213, (byte)78, (byte)169,
(byte)108, (byte)86, (byte)244, (byte)234, (byte)101, (byte)122, (byte)174, (byte)8,
(byte)186, (byte)120, (byte)37, (byte)46, (byte)28, (byte)166, (byte)180, (byte)198,
(byte)232, (byte)221, (byte)116, (byte)31, (byte)75, (byte)189, (byte)139, (byte)138,
(byte)112, (byte)62, (byte)181, (byte)102, (byte)72, (byte)3, (byte)246, (byte)14,
(byte)97, (byte)53, (byte)87, (byte)185, (byte)134, (byte)193, (byte)29, (byte)158,
(byte)225, (byte)248, (byte)152, (byte)17, (byte)105, (byte)217, (byte)142, (byte)148,
(byte)155, (byte)30, (byte)135, (byte)233, (byte)206, (byte)85, (byte)40, (byte)223,
(byte)140, (byte)161, (byte)137, (byte)13, (byte)191, (byte)230, (byte)66, (byte)104,
(byte)65, (byte)153, (byte)45, (byte)15, (byte)176, (byte)84, (byte)187, (byte)22,
};
// The inverse S-box
private static final byte[] Si = {
(byte)82, (byte)9, (byte)106, (byte)213, (byte)48, (byte)54, (byte)165, (byte)56,
(byte)191, (byte)64, (byte)163, (byte)158, (byte)129, (byte)243, (byte)215, (byte)251,
(byte)124, (byte)227, (byte)57, (byte)130, (byte)155, (byte)47, (byte)255, (byte)135,
(byte)52, (byte)142, (byte)67, (byte)68, (byte)196, (byte)222, (byte)233, (byte)203,
(byte)84, (byte)123, (byte)148, (byte)50, (byte)166, (byte)194, (byte)35, (byte)61,
(byte)238, (byte)76, (byte)149, (byte)11, (byte)66, (byte)250, (byte)195, (byte)78,
(byte)8, (byte)46, (byte)161, (byte)102, (byte)40, (byte)217, (byte)36, (byte)178,
(byte)118, (byte)91, (byte)162, (byte)73, (byte)109, (byte)139, (byte)209, (byte)37,
(byte)114, (byte)248, (byte)246, (byte)100, (byte)134, (byte)104, (byte)152, (byte)22,
(byte)212, (byte)164, (byte)92, (byte)204, (byte)93, (byte)101, (byte)182, (byte)146,
(byte)108, (byte)112, (byte)72, (byte)80, (byte)253, (byte)237, (byte)185, (byte)218,
(byte)94, (byte)21, (byte)70, (byte)87, (byte)167, (byte)141, (byte)157, (byte)132,
(byte)144, (byte)216, (byte)171, (byte)0, (byte)140, (byte)188, (byte)211, (byte)10,
(byte)247, (byte)228, (byte)88, (byte)5, (byte)184, (byte)179, (byte)69, (byte)6,
(byte)208, (byte)44, (byte)30, (byte)143, (byte)202, (byte)63, (byte)15, (byte)2,
(byte)193, (byte)175, (byte)189, (byte)3, (byte)1, (byte)19, (byte)138, (byte)107,
(byte)58, (byte)145, (byte)17, (byte)65, (byte)79, (byte)103, (byte)220, (byte)234,
(byte)151, (byte)242, (byte)207, (byte)206, (byte)240, (byte)180, (byte)230, (byte)115,
(byte)150, (byte)172, (byte)116, (byte)34, (byte)231, (byte)173, (byte)53, (byte)133,
(byte)226, (byte)249, (byte)55, (byte)232, (byte)28, (byte)117, (byte)223, (byte)110,
(byte)71, (byte)241, (byte)26, (byte)113, (byte)29, (byte)41, (byte)197, (byte)137,
(byte)111, (byte)183, (byte)98, (byte)14, (byte)170, (byte)24, (byte)190, (byte)27,
(byte)252, (byte)86, (byte)62, (byte)75, (byte)198, (byte)210, (byte)121, (byte)32,
(byte)154, (byte)219, (byte)192, (byte)254, (byte)120, (byte)205, (byte)90, (byte)244,
(byte)31, (byte)221, (byte)168, (byte)51, (byte)136, (byte)7, (byte)199, (byte)49,
(byte)177, (byte)18, (byte)16, (byte)89, (byte)39, (byte)128, (byte)236, (byte)95,
(byte)96, (byte)81, (byte)127, (byte)169, (byte)25, (byte)181, (byte)74, (byte)13,
(byte)45, (byte)229, (byte)122, (byte)159, (byte)147, (byte)201, (byte)156, (byte)239,
(byte)160, (byte)224, (byte)59, (byte)77, (byte)174, (byte)42, (byte)245, (byte)176,
(byte)200, (byte)235, (byte)187, (byte)60, (byte)131, (byte)83, (byte)153, (byte)97,
(byte)23, (byte)43, (byte)4, (byte)126, (byte)186, (byte)119, (byte)214, (byte)38,
(byte)225, (byte)105, (byte)20, (byte)99, (byte)85, (byte)33, (byte)12, (byte)125,
};
// vector used in calculating key schedule (powers of x in GF(256))
private static final int[] rcon = {
0x01, 0x02, 0x04, 0x08, 0x10, 0x20, 0x40, 0x80, 0x1b, 0x36, 0x6c, 0xd8, 0xab, 0x4d, 0x9a,
0x2f, 0x5e, 0xbc, 0x63, 0xc6, 0x97, 0x35, 0x6a, 0xd4, 0xb3, 0x7d, 0xfa, 0xef, 0xc5, 0x91 };
private static int shift(int r, int shift)
{
return (r >>> shift) | (r << -shift);
}
/* multiply four bytes in GF(2^8) by 'x' {02} in parallel */
private static final int m1 = 0x80808080;
private static final int m2 = 0x7f7f7f7f;
private static final int m3 = 0x0000001b;
private static int FFmulX(int x)
{
return (((x & m2) << 1) ^ (((x & m1) >>> 7) * m3));
}
/*
The following com.fitburfines provide alternative com.fitburfinitions of FFmulX that might
give improved performance if a fast 32-bit multiply is not available.
private int FFmulX(int x) { int u = x & m1; u |= (u >> 1); return ((x & m2) << 1) ^ ((u >>> 3) | (u >>> 6)); }
private static final int m4 = 0x1b1b1b1b;
private int FFmulX(int x) { int u = x & m1; return ((x & m2) << 1) ^ ((u - (u >>> 7)) & m4); }
*/
private static int mcol(int x)
{
int f2 = FFmulX(x);
return f2 ^ shift(x ^ f2, 8) ^ shift(x, 16) ^ shift(x, 24);
}
private static int inv_mcol(int x)
{
int f2 = FFmulX(x);
int f4 = FFmulX(f2);
int f8 = FFmulX(f4);
int f9 = x ^ f8;
return f2 ^ f4 ^ f8 ^ shift(f2 ^ f9, 8) ^ shift(f4 ^ f9, 16) ^ shift(f9, 24);
}
private static int subWord(int x)
{
return (S[x&255]&255 | ((S[(x>>8)&255]&255)<<8) | ((S[(x>>16)&255]&255)<<16) | S[(x>>24)&255]<<24);
}
/**
* Calculate the necessary round keys
* The number of calculations com.fitburpends on key size and block size
* AES specified a fixed block size of 128 bits and key sizes 128/192/256 bits
* This code is written assuming those are the only possible values
*/
private int[][] generateWorkingKey(
byte[] key,
boolean forEncryption)
{
int KC = key.length / 4; // key length in words
int t;
if (((KC != 4) && (KC != 6) && (KC != 8)) || ((KC * 4) != key.length))
{
throw new IllegalArgumentException("Key length not 128/192/256 bits.");
}
ROUNDS = KC + 6; // This is not always true for the generalized Rijndael that allows larger block sizes
int[][] W = new int[ROUNDS+1][4]; // 4 words in a block
//
// copy the key into the round key array
//
t = 0;
int i = 0;
while (i < key.length)
{
W[t >> 2][t & 3] = (key[i]&0xff) | ((key[i+1]&0xff) << 8) | ((key[i+2]&0xff) << 16) | (key[i+3] << 24);
i+=4;
t++;
}
//
// while not enough round key material calculated
// calculate new values
//
int k = (ROUNDS + 1) << 2;
for (i = KC; (i < k); i++)
{
int temp = W[(i-1)>>2][(i-1)&3];
if ((i % KC) == 0)
{
temp = subWord(shift(temp, 8)) ^ rcon[(i / KC)-1];
}
else if ((KC > 6) && ((i % KC) == 4))
{
temp = subWord(temp);
}
W[i>>2][i&3] = W[(i - KC)>>2][(i-KC)&3] ^ temp;
}
if (!forEncryption)
{
for (int j = 1; j < ROUNDS; j++)
{
for (i = 0; i < 4; i++)
{
W[j][i] = inv_mcol(W[j][i]);
}
}
}
return W;
}
private int ROUNDS;
private int[][] WorkingKey = null;
private int C0, C1, C2, C3;
private boolean forEncryption;
private static final int BLOCK_SIZE = 16;
/**
* com.fitburfault constructor - 128 bit block size.
*/
public AESLightEngine()
{
}
/**
* initialise an AES cipher.
*
* @param forEncryption whether or not we are for encryption.
* @param params the parameters required to set up the cipher.
* @exception IllegalArgumentException if the params argument is
* inappropriate.
*/
public void init(
boolean forEncryption,
CipherParameters params)
{
if (params instanceof KeyParameter)
{
WorkingKey = generateWorkingKey(((KeyParameter)params).getKey(), forEncryption);
this.forEncryption = forEncryption;
return;
}
throw new IllegalArgumentException("invalid parameter passed to AES init - " + params.getClass().getName());
}
public String getAlgorithmName()
{
return "AES";
}
public int getBlockSize()
{
return BLOCK_SIZE;
}
public int processBlock(
byte[] in,
int inOff,
byte[] out,
int outOff)
{
if (WorkingKey == null)
{
throw new IllegalStateException("AES engine not initialised");
}
if ((inOff + (32 / 2)) > in.length)
{
throw new DataLengthException("input buffer too short");
}
if ((outOff + (32 / 2)) > out.length)
{
throw new OutputLengthException("output buffer too short");
}
if (forEncryption)
{
unpackBlock(in, inOff);
encryptBlock(WorkingKey);
packBlock(out, outOff);
}
else
{
unpackBlock(in, inOff);
com.fitburcryptBlock(WorkingKey);
packBlock(out, outOff);
}
return BLOCK_SIZE;
}
public void reset()
{
}
private void unpackBlock(
byte[] bytes,
int off)
{
int index = off;
C0 = (bytes[index++] & 0xff);
C0 |= (bytes[index++] & 0xff) << 8;
C0 |= (bytes[index++] & 0xff) << 16;
C0 |= bytes[index++] << 24;
C1 = (bytes[index++] & 0xff);
C1 |= (bytes[index++] & 0xff) << 8;
C1 |= (bytes[index++] & 0xff) << 16;
C1 |= bytes[index++] << 24;
C2 = (bytes[index++] & 0xff);
C2 |= (bytes[index++] & 0xff) << 8;
C2 |= (bytes[index++] & 0xff) << 16;
C2 |= bytes[index++] << 24;
C3 = (bytes[index++] & 0xff);
C3 |= (bytes[index++] & 0xff) << 8;
C3 |= (bytes[index++] & 0xff) << 16;
C3 |= bytes[index++] << 24;
}
private void packBlock(
byte[] bytes,
int off)
{
int index = off;
bytes[index++] = (byte)C0;
bytes[index++] = (byte)(C0 >> 8);
bytes[index++] = (byte)(C0 >> 16);
bytes[index++] = (byte)(C0 >> 24);
bytes[index++] = (byte)C1;
bytes[index++] = (byte)(C1 >> 8);
bytes[index++] = (byte)(C1 >> 16);
bytes[index++] = (byte)(C1 >> 24);
bytes[index++] = (byte)C2;
bytes[index++] = (byte)(C2 >> 8);
bytes[index++] = (byte)(C2 >> 16);
bytes[index++] = (byte)(C2 >> 24);
bytes[index++] = (byte)C3;
bytes[index++] = (byte)(C3 >> 8);
bytes[index++] = (byte)(C3 >> 16);
bytes[index++] = (byte)(C3 >> 24);
}
private void encryptBlock(int[][] KW)
{
int t0 = this.C0 ^ KW[0][0];
int t1 = this.C1 ^ KW[0][1];
int t2 = this.C2 ^ KW[0][2];
int r = 1, r0, r1, r2, r3 = this.C3 ^ KW[0][3];
while (r < ROUNDS - 1)
{
r0 = mcol((S[t0&255]&255) ^ ((S[(t1>>8)&255]&255)<<8) ^ ((S[(t2>>16)&255]&255)<<16) ^ (S[(r3>>24)&255]<<24)) ^ KW[r][0];
r1 = mcol((S[t1&255]&255) ^ ((S[(t2>>8)&255]&255)<<8) ^ ((S[(r3>>16)&255]&255)<<16) ^ (S[(t0>>24)&255]<<24)) ^ KW[r][1];
r2 = mcol((S[t2&255]&255) ^ ((S[(r3>>8)&255]&255)<<8) ^ ((S[(t0>>16)&255]&255)<<16) ^ (S[(t1>>24)&255]<<24)) ^ KW[r][2];
r3 = mcol((S[r3&255]&255) ^ ((S[(t0>>8)&255]&255)<<8) ^ ((S[(t1>>16)&255]&255)<<16) ^ (S[(t2>>24)&255]<<24)) ^ KW[r++][3];
t0 = mcol((S[r0&255]&255) ^ ((S[(r1>>8)&255]&255)<<8) ^ ((S[(r2>>16)&255]&255)<<16) ^ (S[(r3>>24)&255]<<24)) ^ KW[r][0];
t1 = mcol((S[r1&255]&255) ^ ((S[(r2>>8)&255]&255)<<8) ^ ((S[(r3>>16)&255]&255)<<16) ^ (S[(r0>>24)&255]<<24)) ^ KW[r][1];
t2 = mcol((S[r2&255]&255) ^ ((S[(r3>>8)&255]&255)<<8) ^ ((S[(r0>>16)&255]&255)<<16) ^ (S[(r1>>24)&255]<<24)) ^ KW[r][2];
r3 = mcol((S[r3&255]&255) ^ ((S[(r0>>8)&255]&255)<<8) ^ ((S[(r1>>16)&255]&255)<<16) ^ (S[(r2>>24)&255]<<24)) ^ KW[r++][3];
}
r0 = mcol((S[t0&255]&255) ^ ((S[(t1>>8)&255]&255)<<8) ^ ((S[(t2>>16)&255]&255)<<16) ^ (S[(r3>>24)&255]<<24)) ^ KW[r][0];
r1 = mcol((S[t1&255]&255) ^ ((S[(t2>>8)&255]&255)<<8) ^ ((S[(r3>>16)&255]&255)<<16) ^ (S[(t0>>24)&255]<<24)) ^ KW[r][1];
r2 = mcol((S[t2&255]&255) ^ ((S[(r3>>8)&255]&255)<<8) ^ ((S[(t0>>16)&255]&255)<<16) ^ (S[(t1>>24)&255]<<24)) ^ KW[r][2];
r3 = mcol((S[r3&255]&255) ^ ((S[(t0>>8)&255]&255)<<8) ^ ((S[(t1>>16)&255]&255)<<16) ^ (S[(t2>>24)&255]<<24)) ^ KW[r++][3];
// the final round is a simple function of S
this.C0 = (S[r0&255]&255) ^ ((S[(r1>>8)&255]&255)<<8) ^ ((S[(r2>>16)&255]&255)<<16) ^ (S[(r3>>24)&255]<<24) ^ KW[r][0];
this.C1 = (S[r1&255]&255) ^ ((S[(r2>>8)&255]&255)<<8) ^ ((S[(r3>>16)&255]&255)<<16) ^ (S[(r0>>24)&255]<<24) ^ KW[r][1];
this.C2 = (S[r2&255]&255) ^ ((S[(r3>>8)&255]&255)<<8) ^ ((S[(r0>>16)&255]&255)<<16) ^ (S[(r1>>24)&255]<<24) ^ KW[r][2];
this.C3 = (S[r3&255]&255) ^ ((S[(r0>>8)&255]&255)<<8) ^ ((S[(r1>>16)&255]&255)<<16) ^ (S[(r2>>24)&255]<<24) ^ KW[r][3];
}
private void com.fitburcryptBlock(int[][] KW)
{
int t0 = this.C0 ^ KW[ROUNDS][0];
int t1 = this.C1 ^ KW[ROUNDS][1];
int t2 = this.C2 ^ KW[ROUNDS][2];
int r = ROUNDS - 1, r0, r1, r2, r3 = this.C3 ^ KW[ROUNDS][3];
while (r > 1)
{
r0 = inv_mcol((Si[t0&255]&255) ^ ((Si[(r3>>8)&255]&255)<<8) ^ ((Si[(t2>>16)&255]&255)<<16) ^ (Si[(t1>>24)&255]<<24)) ^ KW[r][0];
r1 = inv_mcol((Si[t1&255]&255) ^ ((Si[(t0>>8)&255]&255)<<8) ^ ((Si[(r3>>16)&255]&255)<<16) ^ (Si[(t2>>24)&255]<<24)) ^ KW[r][1];
r2 = inv_mcol((Si[t2&255]&255) ^ ((Si[(t1>>8)&255]&255)<<8) ^ ((Si[(t0>>16)&255]&255)<<16) ^ (Si[(r3>>24)&255]<<24)) ^ KW[r][2];
r3 = inv_mcol((Si[r3&255]&255) ^ ((Si[(t2>>8)&255]&255)<<8) ^ ((Si[(t1>>16)&255]&255)<<16) ^ (Si[(t0>>24)&255]<<24)) ^ KW[r--][3];
t0 = inv_mcol((Si[r0&255]&255) ^ ((Si[(r3>>8)&255]&255)<<8) ^ ((Si[(r2>>16)&255]&255)<<16) ^ (Si[(r1>>24)&255]<<24)) ^ KW[r][0];
t1 = inv_mcol((Si[r1&255]&255) ^ ((Si[(r0>>8)&255]&255)<<8) ^ ((Si[(r3>>16)&255]&255)<<16) ^ (Si[(r2>>24)&255]<<24)) ^ KW[r][1];
t2 = inv_mcol((Si[r2&255]&255) ^ ((Si[(r1>>8)&255]&255)<<8) ^ ((Si[(r0>>16)&255]&255)<<16) ^ (Si[(r3>>24)&255]<<24)) ^ KW[r][2];
r3 = inv_mcol((Si[r3&255]&255) ^ ((Si[(r2>>8)&255]&255)<<8) ^ ((Si[(r1>>16)&255]&255)<<16) ^ (Si[(r0>>24)&255]<<24)) ^ KW[r--][3];
}
r0 = inv_mcol((Si[t0&255]&255) ^ ((Si[(r3>>8)&255]&255)<<8) ^ ((Si[(t2>>16)&255]&255)<<16) ^ (Si[(t1>>24)&255]<<24)) ^ KW[r][0];
r1 = inv_mcol((Si[t1&255]&255) ^ ((Si[(t0>>8)&255]&255)<<8) ^ ((Si[(r3>>16)&255]&255)<<16) ^ (Si[(t2>>24)&255]<<24)) ^ KW[r][1];
r2 = inv_mcol((Si[t2&255]&255) ^ ((Si[(t1>>8)&255]&255)<<8) ^ ((Si[(t0>>16)&255]&255)<<16) ^ (Si[(r3>>24)&255]<<24)) ^ KW[r][2];
r3 = inv_mcol((Si[r3&255]&255) ^ ((Si[(t2>>8)&255]&255)<<8) ^ ((Si[(t1>>16)&255]&255)<<16) ^ (Si[(t0>>24)&255]<<24)) ^ KW[r][3];
// the final round's table is a simple function of Si
this.C0 = (Si[r0&255]&255) ^ ((Si[(r3>>8)&255]&255)<<8) ^ ((Si[(r2>>16)&255]&255)<<16) ^ (Si[(r1>>24)&255]<<24) ^ KW[0][0];
this.C1 = (Si[r1&255]&255) ^ ((Si[(r0>>8)&255]&255)<<8) ^ ((Si[(r3>>16)&255]&255)<<16) ^ (Si[(r2>>24)&255]<<24) ^ KW[0][1];
this.C2 = (Si[r2&255]&255) ^ ((Si[(r1>>8)&255]&255)<<8) ^ ((Si[(r0>>16)&255]&255)<<16) ^ (Si[(r3>>24)&255]<<24) ^ KW[0][2];
this.C3 = (Si[r3&255]&255) ^ ((Si[(r2>>8)&255]&255)<<8) ^ ((Si[(r1>>16)&255]&255)<<16) ^ (Si[(r0>>24)&255]<<24) ^ KW[0][3];
}
}