/*
 * Copyright 2017 The OpenSSL Project Authors. All Rights Reserved.
 *
 * Licensed under the OpenSSL license (the "License").  You may not use
 * this file except in compliance with the License.  You can obtain a copy
 * in the file LICENSE in the source distribution or at
 * https://www.openssl.org/source/license.html
 */

/* ====================================================================
 * Copyright (c) 2017 Oracle and/or its affiliates.  All rights reserved.
 */

#include <assert.h>
#include <openssl/e_os2.h>
#include <string.h>
#include "internal/aria.h"

static const unsigned char sb1[256] = {
    0x63, 0x7c, 0x77, 0x7b, 0xf2, 0x6b, 0x6f, 0xc5,
    0x30, 0x01, 0x67, 0x2b, 0xfe, 0xd7, 0xab, 0x76,
    0xca, 0x82, 0xc9, 0x7d, 0xfa, 0x59, 0x47, 0xf0,
    0xad, 0xd4, 0xa2, 0xaf, 0x9c, 0xa4, 0x72, 0xc0,
    0xb7, 0xfd, 0x93, 0x26, 0x36, 0x3f, 0xf7, 0xcc,
    0x34, 0xa5, 0xe5, 0xf1, 0x71, 0xd8, 0x31, 0x15,
    0x04, 0xc7, 0x23, 0xc3, 0x18, 0x96, 0x05, 0x9a,
    0x07, 0x12, 0x80, 0xe2, 0xeb, 0x27, 0xb2, 0x75,
    0x09, 0x83, 0x2c, 0x1a, 0x1b, 0x6e, 0x5a, 0xa0,
    0x52, 0x3b, 0xd6, 0xb3, 0x29, 0xe3, 0x2f, 0x84,
    0x53, 0xd1, 0x00, 0xed, 0x20, 0xfc, 0xb1, 0x5b,
    0x6a, 0xcb, 0xbe, 0x39, 0x4a, 0x4c, 0x58, 0xcf,
    0xd0, 0xef, 0xaa, 0xfb, 0x43, 0x4d, 0x33, 0x85,
    0x45, 0xf9, 0x02, 0x7f, 0x50, 0x3c, 0x9f, 0xa8,
    0x51, 0xa3, 0x40, 0x8f, 0x92, 0x9d, 0x38, 0xf5,
    0xbc, 0xb6, 0xda, 0x21, 0x10, 0xff, 0xf3, 0xd2,
    0xcd, 0x0c, 0x13, 0xec, 0x5f, 0x97, 0x44, 0x17,
    0xc4, 0xa7, 0x7e, 0x3d, 0x64, 0x5d, 0x19, 0x73,
    0x60, 0x81, 0x4f, 0xdc, 0x22, 0x2a, 0x90, 0x88,
    0x46, 0xee, 0xb8, 0x14, 0xde, 0x5e, 0x0b, 0xdb,
    0xe0, 0x32, 0x3a, 0x0a, 0x49, 0x06, 0x24, 0x5c,
    0xc2, 0xd3, 0xac, 0x62, 0x91, 0x95, 0xe4, 0x79,
    0xe7, 0xc8, 0x37, 0x6d, 0x8d, 0xd5, 0x4e, 0xa9,
    0x6c, 0x56, 0xf4, 0xea, 0x65, 0x7a, 0xae, 0x08,
    0xba, 0x78, 0x25, 0x2e, 0x1c, 0xa6, 0xb4, 0xc6,
    0xe8, 0xdd, 0x74, 0x1f, 0x4b, 0xbd, 0x8b, 0x8a,
    0x70, 0x3e, 0xb5, 0x66, 0x48, 0x03, 0xf6, 0x0e,
    0x61, 0x35, 0x57, 0xb9, 0x86, 0xc1, 0x1d, 0x9e,
    0xe1, 0xf8, 0x98, 0x11, 0x69, 0xd9, 0x8e, 0x94,
    0x9b, 0x1e, 0x87, 0xe9, 0xce, 0x55, 0x28, 0xdf,
    0x8c, 0xa1, 0x89, 0x0d, 0xbf, 0xe6, 0x42, 0x68,
    0x41, 0x99, 0x2d, 0x0f, 0xb0, 0x54, 0xbb, 0x16
};

static const unsigned char sb2[256] = {
    0xe2, 0x4e, 0x54, 0xfc, 0x94, 0xc2, 0x4a, 0xcc,
    0x62, 0x0d, 0x6a, 0x46, 0x3c, 0x4d, 0x8b, 0xd1,
    0x5e, 0xfa, 0x64, 0xcb, 0xb4, 0x97, 0xbe, 0x2b,
    0xbc, 0x77, 0x2e, 0x03, 0xd3, 0x19, 0x59, 0xc1,
    0x1d, 0x06, 0x41, 0x6b, 0x55, 0xf0, 0x99, 0x69,
    0xea, 0x9c, 0x18, 0xae, 0x63, 0xdf, 0xe7, 0xbb,
    0x00, 0x73, 0x66, 0xfb, 0x96, 0x4c, 0x85, 0xe4,
    0x3a, 0x09, 0x45, 0xaa, 0x0f, 0xee, 0x10, 0xeb,
    0x2d, 0x7f, 0xf4, 0x29, 0xac, 0xcf, 0xad, 0x91,
    0x8d, 0x78, 0xc8, 0x95, 0xf9, 0x2f, 0xce, 0xcd,
    0x08, 0x7a, 0x88, 0x38, 0x5c, 0x83, 0x2a, 0x28,
    0x47, 0xdb, 0xb8, 0xc7, 0x93, 0xa4, 0x12, 0x53,
    0xff, 0x87, 0x0e, 0x31, 0x36, 0x21, 0x58, 0x48,
    0x01, 0x8e, 0x37, 0x74, 0x32, 0xca, 0xe9, 0xb1,
    0xb7, 0xab, 0x0c, 0xd7, 0xc4, 0x56, 0x42, 0x26,
    0x07, 0x98, 0x60, 0xd9, 0xb6, 0xb9, 0x11, 0x40,
    0xec, 0x20, 0x8c, 0xbd, 0xa0, 0xc9, 0x84, 0x04,
    0x49, 0x23, 0xf1, 0x4f, 0x50, 0x1f, 0x13, 0xdc,
    0xd8, 0xc0, 0x9e, 0x57, 0xe3, 0xc3, 0x7b, 0x65,
    0x3b, 0x02, 0x8f, 0x3e, 0xe8, 0x25, 0x92, 0xe5,
    0x15, 0xdd, 0xfd, 0x17, 0xa9, 0xbf, 0xd4, 0x9a,
    0x7e, 0xc5, 0x39, 0x67, 0xfe, 0x76, 0x9d, 0x43,
    0xa7, 0xe1, 0xd0, 0xf5, 0x68, 0xf2, 0x1b, 0x34,
    0x70, 0x05, 0xa3, 0x8a, 0xd5, 0x79, 0x86, 0xa8,
    0x30, 0xc6, 0x51, 0x4b, 0x1e, 0xa6, 0x27, 0xf6,
    0x35, 0xd2, 0x6e, 0x24, 0x16, 0x82, 0x5f, 0xda,
    0xe6, 0x75, 0xa2, 0xef, 0x2c, 0xb2, 0x1c, 0x9f,
    0x5d, 0x6f, 0x80, 0x0a, 0x72, 0x44, 0x9b, 0x6c,
    0x90, 0x0b, 0x5b, 0x33, 0x7d, 0x5a, 0x52, 0xf3,
    0x61, 0xa1, 0xf7, 0xb0, 0xd6, 0x3f, 0x7c, 0x6d,
    0xed, 0x14, 0xe0, 0xa5, 0x3d, 0x22, 0xb3, 0xf8,
    0x89, 0xde, 0x71, 0x1a, 0xaf, 0xba, 0xb5, 0x81
};

static const unsigned char sb3[256] = {
    0x52, 0x09, 0x6a, 0xd5, 0x30, 0x36, 0xa5, 0x38,
    0xbf, 0x40, 0xa3, 0x9e, 0x81, 0xf3, 0xd7, 0xfb,
    0x7c, 0xe3, 0x39, 0x82, 0x9b, 0x2f, 0xff, 0x87,
    0x34, 0x8e, 0x43, 0x44, 0xc4, 0xde, 0xe9, 0xcb,
    0x54, 0x7b, 0x94, 0x32, 0xa6, 0xc2, 0x23, 0x3d,
    0xee, 0x4c, 0x95, 0x0b, 0x42, 0xfa, 0xc3, 0x4e,
    0x08, 0x2e, 0xa1, 0x66, 0x28, 0xd9, 0x24, 0xb2,
    0x76, 0x5b, 0xa2, 0x49, 0x6d, 0x8b, 0xd1, 0x25,
    0x72, 0xf8, 0xf6, 0x64, 0x86, 0x68, 0x98, 0x16,
    0xd4, 0xa4, 0x5c, 0xcc, 0x5d, 0x65, 0xb6, 0x92,
    0x6c, 0x70, 0x48, 0x50, 0xfd, 0xed, 0xb9, 0xda,
    0x5e, 0x15, 0x46, 0x57, 0xa7, 0x8d, 0x9d, 0x84,
    0x90, 0xd8, 0xab, 0x00, 0x8c, 0xbc, 0xd3, 0x0a,
    0xf7, 0xe4, 0x58, 0x05, 0xb8, 0xb3, 0x45, 0x06,
    0xd0, 0x2c, 0x1e, 0x8f, 0xca, 0x3f, 0x0f, 0x02,
    0xc1, 0xaf, 0xbd, 0x03, 0x01, 0x13, 0x8a, 0x6b,
    0x3a, 0x91, 0x11, 0x41, 0x4f, 0x67, 0xdc, 0xea,
    0x97, 0xf2, 0xcf, 0xce, 0xf0, 0xb4, 0xe6, 0x73,
    0x96, 0xac, 0x74, 0x22, 0xe7, 0xad, 0x35, 0x85,
    0xe2, 0xf9, 0x37, 0xe8, 0x1c, 0x75, 0xdf, 0x6e,
    0x47, 0xf1, 0x1a, 0x71, 0x1d, 0x29, 0xc5, 0x89,
    0x6f, 0xb7, 0x62, 0x0e, 0xaa, 0x18, 0xbe, 0x1b,
    0xfc, 0x56, 0x3e, 0x4b, 0xc6, 0xd2, 0x79, 0x20,
    0x9a, 0xdb, 0xc0, 0xfe, 0x78, 0xcd, 0x5a, 0xf4,
    0x1f, 0xdd, 0xa8, 0x33, 0x88, 0x07, 0xc7, 0x31,
    0xb1, 0x12, 0x10, 0x59, 0x27, 0x80, 0xec, 0x5f,
    0x60, 0x51, 0x7f, 0xa9, 0x19, 0xb5, 0x4a, 0x0d,
    0x2d, 0xe5, 0x7a, 0x9f, 0x93, 0xc9, 0x9c, 0xef,
    0xa0, 0xe0, 0x3b, 0x4d, 0xae, 0x2a, 0xf5, 0xb0,
    0xc8, 0xeb, 0xbb, 0x3c, 0x83, 0x53, 0x99, 0x61,
    0x17, 0x2b, 0x04, 0x7e, 0xba, 0x77, 0xd6, 0x26,
    0xe1, 0x69, 0x14, 0x63, 0x55, 0x21, 0x0c, 0x7d
};

static const unsigned char sb4[256] = {
    0x30, 0x68, 0x99, 0x1b, 0x87, 0xb9, 0x21, 0x78,
    0x50, 0x39, 0xdb, 0xe1, 0x72, 0x09, 0x62, 0x3c,
    0x3e, 0x7e, 0x5e, 0x8e, 0xf1, 0xa0, 0xcc, 0xa3,
    0x2a, 0x1d, 0xfb, 0xb6, 0xd6, 0x20, 0xc4, 0x8d,
    0x81, 0x65, 0xf5, 0x89, 0xcb, 0x9d, 0x77, 0xc6,
    0x57, 0x43, 0x56, 0x17, 0xd4, 0x40, 0x1a, 0x4d,
    0xc0, 0x63, 0x6c, 0xe3, 0xb7, 0xc8, 0x64, 0x6a,
    0x53, 0xaa, 0x38, 0x98, 0x0c, 0xf4, 0x9b, 0xed,
    0x7f, 0x22, 0x76, 0xaf, 0xdd, 0x3a, 0x0b, 0x58,
    0x67, 0x88, 0x06, 0xc3, 0x35, 0x0d, 0x01, 0x8b,
    0x8c, 0xc2, 0xe6, 0x5f, 0x02, 0x24, 0x75, 0x93,
    0x66, 0x1e, 0xe5, 0xe2, 0x54, 0xd8, 0x10, 0xce,
    0x7a, 0xe8, 0x08, 0x2c, 0x12, 0x97, 0x32, 0xab,
    0xb4, 0x27, 0x0a, 0x23, 0xdf, 0xef, 0xca, 0xd9,
    0xb8, 0xfa, 0xdc, 0x31, 0x6b, 0xd1, 0xad, 0x19,
    0x49, 0xbd, 0x51, 0x96, 0xee, 0xe4, 0xa8, 0x41,
    0xda, 0xff, 0xcd, 0x55, 0x86, 0x36, 0xbe, 0x61,
    0x52, 0xf8, 0xbb, 0x0e, 0x82, 0x48, 0x69, 0x9a,
    0xe0, 0x47, 0x9e, 0x5c, 0x04, 0x4b, 0x34, 0x15,
    0x79, 0x26, 0xa7, 0xde, 0x29, 0xae, 0x92, 0xd7,
    0x84, 0xe9, 0xd2, 0xba, 0x5d, 0xf3, 0xc5, 0xb0,
    0xbf, 0xa4, 0x3b, 0x71, 0x44, 0x46, 0x2b, 0xfc,
    0xeb, 0x6f, 0xd5, 0xf6, 0x14, 0xfe, 0x7c, 0x70,
    0x5a, 0x7d, 0xfd, 0x2f, 0x18, 0x83, 0x16, 0xa5,
    0x91, 0x1f, 0x05, 0x95, 0x74, 0xa9, 0xc1, 0x5b,
    0x4a, 0x85, 0x6d, 0x13, 0x07, 0x4f, 0x4e, 0x45,
    0xb2, 0x0f, 0xc9, 0x1c, 0xa6, 0xbc, 0xec, 0x73,
    0x90, 0x7b, 0xcf, 0x59, 0x8f, 0xa1, 0xf9, 0x2d,
    0xf2, 0xb1, 0x00, 0x94, 0x37, 0x9f, 0xd0, 0x2e,
    0x9c, 0x6e, 0x28, 0x3f, 0x80, 0xf0, 0x3d, 0xd3,
    0x25, 0x8a, 0xb5, 0xe7, 0x42, 0xb3, 0xc7, 0xea,
    0xf7, 0x4c, 0x11, 0x33, 0x03, 0xa2, 0xac, 0x60
};

static const ARIA_u128 c1 = {
    0x51, 0x7c, 0xc1, 0xb7, 0x27, 0x22, 0x0a, 0x94,
    0xfe, 0x13, 0xab, 0xe8, 0xfa, 0x9a, 0x6e, 0xe0
};

static const ARIA_u128 c2 = {
    0x6d, 0xb1, 0x4a, 0xcc, 0x9e, 0x21, 0xc8, 0x20,
    0xff, 0x28, 0xb1, 0xd5, 0xef, 0x5d, 0xe2, 0xb0
};

static const ARIA_u128 c3 = {
    0xdb, 0x92, 0x37, 0x1d, 0x21, 0x26, 0xe9, 0x70,
    0x03, 0x24, 0x97, 0x75, 0x04, 0xe8, 0xc9, 0x0e
};

/*
 * Exclusive or two 128 bit values into the result.
 * It is safe for the result to be the same as the either input.
 */
static void xor128(ARIA_u128 o, const ARIA_u128 x, const ARIA_u128 y)
{
    int i;

    for (i = 0; i < ARIA_BLOCK_SIZE; i++)
        o[i] = x[i] ^ y[i];
}

/*
 * Generalised circular rotate right and exclusive or function.
 * It is safe for the output to overlap either input.
 */
static ossl_inline void rotnr(unsigned int n, ARIA_u128 o, const ARIA_u128 xor,
                              const ARIA_u128 z)
{
    const unsigned int bytes = n / 8, bits = n % 8;
    unsigned int i;
    ARIA_u128 t;

    for (i = 0; i < ARIA_BLOCK_SIZE; i++)
        t[(i + bytes) % ARIA_BLOCK_SIZE] = z[i];
    for (i = 0; i < ARIA_BLOCK_SIZE; i++)
        o[i] = ((t[i] >> bits) |
                (t[i ? i - 1 : ARIA_BLOCK_SIZE - 1] << (8 - bits))) ^ xor[i];
}

/*
 * Circular rotate 19 bits right and xor.
 * It is safe for the output to overlap either input.
 */
static void rot19r(ARIA_u128 o, const ARIA_u128 xor, const ARIA_u128 z)
{
    rotnr(19, o, xor, z);
}

/*
 * Circular rotate 31 bits right and xor.
 * It is safe for the output to overlap either input.
 */
static void rot31r(ARIA_u128 o, const ARIA_u128 xor, const ARIA_u128 z)
{
    rotnr(31, o, xor, z);
}

/*
 * Circular rotate 61 bits left and xor.
 * It is safe for the output to overlap either input.
 */
static void rot61l(ARIA_u128 o, const ARIA_u128 xor, const ARIA_u128 z)
{
    rotnr(8 * ARIA_BLOCK_SIZE - 61, o, xor, z);
}

/*
 * Circular rotate 31 bits left and xor.
 * It is safe for the output to overlap either input.
 */
static void rot31l(ARIA_u128 o, const ARIA_u128 xor, const ARIA_u128 z)
{
    rotnr(8 * ARIA_BLOCK_SIZE - 31, o, xor, z);
}

/*
 * Circular rotate 19 bits left and xor.
 * It is safe for the output to overlap either input.
 */
static void rot19l(ARIA_u128 o, const ARIA_u128 xor, const ARIA_u128 z)
{
    rotnr(8 * ARIA_BLOCK_SIZE - 19, o, xor, z);
}

/*
 * First substitution and xor layer, used for odd steps.
 * It is safe for the input and output to be the same.
 */
static void sl1(ARIA_u128 o, const ARIA_u128 x, const ARIA_u128 y)
{
    unsigned int i;
    for (i = 0; i < ARIA_BLOCK_SIZE; i += 4) {
        o[i    ] = sb1[x[i    ] ^ y[i    ]];
        o[i + 1] = sb2[x[i + 1] ^ y[i + 1]];
        o[i + 2] = sb3[x[i + 2] ^ y[i + 2]];
        o[i + 3] = sb4[x[i + 3] ^ y[i + 3]];
    }
}

/*
 * Second substitution and xor layer, used for even steps.
 * It is safe for the input and output to be the same.
 */
static void sl2(ARIA_u128 o, const ARIA_u128 x, const ARIA_u128 y)
{
    unsigned int i;
    for (i = 0; i < ARIA_BLOCK_SIZE; i += 4) {
        o[i    ] = sb3[x[i    ] ^ y[i    ]];
        o[i + 1] = sb4[x[i + 1] ^ y[i + 1]];
        o[i + 2] = sb1[x[i + 2] ^ y[i + 2]];
        o[i + 3] = sb2[x[i + 3] ^ y[i + 3]];
    }
}

/*
 * Diffusion layer step
 * It is NOT safe for the input and output to overlap.
 */
static void a(ARIA_u128 y, const ARIA_u128 x)
{
    y[ 0] = x[3] ^ x[4] ^ x[6] ^ x[ 8] ^ x[ 9] ^ x[13] ^ x[14];
    y[ 1] = x[2] ^ x[5] ^ x[7] ^ x[ 8] ^ x[ 9] ^ x[12] ^ x[15];
    y[ 2] = x[1] ^ x[4] ^ x[6] ^ x[10] ^ x[11] ^ x[12] ^ x[15];
    y[ 3] = x[0] ^ x[5] ^ x[7] ^ x[10] ^ x[11] ^ x[13] ^ x[14];
    y[ 4] = x[0] ^ x[2] ^ x[5] ^ x[ 8] ^ x[11] ^ x[14] ^ x[15];
    y[ 5] = x[1] ^ x[3] ^ x[4] ^ x[ 9] ^ x[10] ^ x[14] ^ x[15];
    y[ 6] = x[0] ^ x[2] ^ x[7] ^ x[ 9] ^ x[10] ^ x[12] ^ x[13];
    y[ 7] = x[1] ^ x[3] ^ x[6] ^ x[ 8] ^ x[11] ^ x[12] ^ x[13];
    y[ 8] = x[0] ^ x[1] ^ x[4] ^ x[ 7] ^ x[10] ^ x[13] ^ x[15];
    y[ 9] = x[0] ^ x[1] ^ x[5] ^ x[ 6] ^ x[11] ^ x[12] ^ x[14];
    y[10] = x[2] ^ x[3] ^ x[5] ^ x[ 6] ^ x[ 8] ^ x[13] ^ x[15];
    y[11] = x[2] ^ x[3] ^ x[4] ^ x[ 7] ^ x[ 9] ^ x[12] ^ x[14];
    y[12] = x[1] ^ x[2] ^ x[6] ^ x[ 7] ^ x[ 9] ^ x[11] ^ x[12];
    y[13] = x[0] ^ x[3] ^ x[6] ^ x[ 7] ^ x[ 8] ^ x[10] ^ x[13];
    y[14] = x[0] ^ x[3] ^ x[4] ^ x[ 5] ^ x[ 9] ^ x[11] ^ x[14];
    y[15] = x[1] ^ x[2] ^ x[4] ^ x[ 5] ^ x[ 8] ^ x[10] ^ x[15];
}

/*
 * Odd round function
 * Apply the first substitution layer and then a diffusion step.
 * It is safe for the input and output to overlap.
 */
static ossl_inline void FO(ARIA_u128 o, const ARIA_u128 d, const ARIA_u128 rk)
{
	ARIA_u128 y;

	sl1(y, d, rk);
	a(o, y);
}

/*
 * Even round function
 * Apply the second substitution layer and then a diffusion step.
 * It is safe for the input and output to overlap.
 */
static ossl_inline void FE(ARIA_u128 o, const ARIA_u128 d, const ARIA_u128 rk)
{
	ARIA_u128 y;

	sl2(y, d, rk);
	a(o, y);
}

/*
 * Encrypt or decrypt a single block
 * in and out can overlap
 */
static void do_encrypt(ARIA_u128 o, const ARIA_u128 pin, unsigned int rounds,
                       const ARIA_u128 *keys)
{
    ARIA_u128 p;
    unsigned int i;

    memcpy(p, pin, sizeof(p));
    for (i = 0; i < rounds - 2; i += 2) {
        FO(p, p, keys[i]);
        FE(p, p, keys[i + 1]);
    }
    FO(p, p, keys[rounds - 2]);
    sl2(o, p, keys[rounds - 1]);
    xor128(o, o, keys[rounds]);
}

/*
 * Encrypt a single block
 * in and out can overlap
 */
void aria_encrypt(const unsigned char *in, unsigned char *out,
                  const ARIA_KEY *key)
{
    assert(in != NULL && out != NULL && key != NULL);
    do_encrypt(out, in, key->rounds, key->rd_key);
}


/*
 * Expand the cipher key into the encryption key schedule.
 * We short circuit execution of the last two
 * or four rotations based on the key size.
 */
int aria_set_encrypt_key(const unsigned char *userKey, const int bits,
                         ARIA_KEY *key)
{
    const unsigned char *ck1, *ck2, *ck3;
    ARIA_u128 kr, w0, w1, w2, w3;

    if (!userKey || !key)
        return -1;
    memcpy(w0, userKey, sizeof(w0));
    switch (bits) {
    default:
        return -2;
    case 128:
        key->rounds = 12;
        ck1 = c1;
        ck2 = c2;
        ck3 = c3;
        memset(kr, 0, sizeof(kr));
        break;

    case 192:
        key->rounds = 14;
        ck1 = c2;
        ck2 = c3;
        ck3 = c1;
        memcpy(kr, userKey + ARIA_BLOCK_SIZE, sizeof(kr) / 2);
        memset(kr + ARIA_BLOCK_SIZE / 2, 0, sizeof(kr) / 2);
        break;

    case 256:
        key->rounds = 16;
        ck1 = c3;
        ck2 = c1;
        ck3 = c2;
        memcpy(kr, userKey + ARIA_BLOCK_SIZE, sizeof(kr));
        break;
    }

    FO(w3, w0, ck1);	xor128(w1, w3, kr);
    FE(w3, w1, ck2);	xor128(w2, w3, w0);
    FO(kr, w2, ck3);	xor128(w3, kr, w1);

    rot19r(key->rd_key[ 0], w0, w1);
    rot19r(key->rd_key[ 1], w1, w2);
    rot19r(key->rd_key[ 2], w2, w3);
    rot19r(key->rd_key[ 3], w3, w0);

    rot31r(key->rd_key[ 4], w0, w1);
    rot31r(key->rd_key[ 5], w1, w2);
    rot31r(key->rd_key[ 6], w2, w3);
    rot31r(key->rd_key[ 7], w3, w0);

    rot61l(key->rd_key[ 8], w0, w1);
    rot61l(key->rd_key[ 9], w1, w2);
    rot61l(key->rd_key[10], w2, w3);
    rot61l(key->rd_key[11], w3, w0);

    rot31l(key->rd_key[12], w0, w1);
    if (key->rounds > 12) {
        rot31l(key->rd_key[13], w1, w2);
        rot31l(key->rd_key[14], w2, w3);

        if (key->rounds > 14) {
            rot31l(key->rd_key[15], w3, w0);
            rot19l(key->rd_key[16], w0, w1);
        }
    }
    return 0;
}

/*
 * Expand the cipher key into the decryption key schedule.
 */
int aria_set_decrypt_key(const unsigned char *userKey, const int bits,
                         ARIA_KEY *key)
{
    ARIA_KEY ek;
    const int r = aria_set_encrypt_key(userKey, bits, &ek);
    unsigned int i, rounds = ek.rounds;

    if (r == 0) {
        key->rounds = rounds;
        memcpy(key->rd_key[0], ek.rd_key[rounds], sizeof(key->rd_key[0]));
        for (i = 1; i < rounds; i++)
            a(key->rd_key[i], ek.rd_key[rounds - i]);
        memcpy(key->rd_key[rounds], ek.rd_key[0], sizeof(key->rd_key[rounds]));
    }
    return r;
}