tc_aria 0.1.0

//! Portable ARIA key schedule and round functions.

use tc_zeroize::Zeroize;

use crate::BLOCK_BYTES;

pub(crate) const MAX_ROUND_KEYS: usize = 17;
pub(crate) type RoundKeys = [[u8; BLOCK_BYTES]; MAX_ROUND_KEYS];

const C: [[u8; BLOCK_BYTES]; 3] = [
    [
        0x51, 0x7c, 0xc1, 0xb7, 0x27, 0x22, 0x0a, 0x94, 0xfe, 0x13, 0xab, 0xe8, 0xfa, 0x9a, 0x6e,
        0xe0,
    ],
    [
        0x6d, 0xb1, 0x4a, 0xcc, 0x9e, 0x21, 0xc8, 0x20, 0xff, 0x28, 0xb1, 0xd5, 0xef, 0x5d, 0xe2,
        0xb0,
    ],
    [
        0xdb, 0x92, 0x37, 0x1d, 0x21, 0x26, 0xe9, 0x70, 0x03, 0x24, 0x97, 0x75, 0x04, 0xe8, 0xc9,
        0x0e,
    ],
];

const SB1: [u8; 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,
];

const SB2: [u8; 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,
];

const SB3: [u8; 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,
];

const SB4: [u8; 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,
];

/// Expands a 16-, 24- or 32-byte key into round keys and returns them with
/// the round count; decryption keys are the encryption keys transformed for
/// the inverse cipher. The caller has validated the key length.
/// Variable time: the expansion runs the round function, which looks up
/// S-boxes with secret data.
pub(super) fn key_schedule(for_encryption: bool, key: &[u8]) -> (RoundKeys, usize) {
    let key_len_index = (key.len() >> 3) - 2;
    let ck1 = &C[key_len_index];
    let ck2 = &C[(key_len_index + 1) % 3];
    let ck3 = &C[(key_len_index + 2) % 3];

    let mut w0 = [0u8; BLOCK_BYTES];
    let mut w1;
    let mut w2;
    let mut w3;

    w0.copy_from_slice(&key[..BLOCK_BYTES]);
    w1 = w0;
    fo(&mut w1, ck1);
    for (destination, source) in w1.iter_mut().zip(&key[BLOCK_BYTES..]) {
        *destination ^= *source;
    }

    w2 = w1;
    fe(&mut w2, ck2);
    xor(&mut w2, &w0);

    w3 = w2;
    fo(&mut w3, ck3);
    xor(&mut w3, &w1);

    let rounds = 12 + key_len_index * 2;
    let mut round_keys = [[0u8; BLOCK_BYTES]; MAX_ROUND_KEYS];

    round_keys[0] = key_schedule_round(&w0, &w1, 19);
    round_keys[1] = key_schedule_round(&w1, &w2, 19);
    round_keys[2] = key_schedule_round(&w2, &w3, 19);
    round_keys[3] = key_schedule_round(&w3, &w0, 19);

    round_keys[4] = key_schedule_round(&w0, &w1, 31);
    round_keys[5] = key_schedule_round(&w1, &w2, 31);
    round_keys[6] = key_schedule_round(&w2, &w3, 31);
    round_keys[7] = key_schedule_round(&w3, &w0, 31);

    round_keys[8] = key_schedule_round(&w0, &w1, 67);
    round_keys[9] = key_schedule_round(&w1, &w2, 67);
    round_keys[10] = key_schedule_round(&w2, &w3, 67);
    round_keys[11] = key_schedule_round(&w3, &w0, 67);

    round_keys[12] = key_schedule_round(&w0, &w1, 97);
    if rounds > 12 {
        round_keys[13] = key_schedule_round(&w1, &w2, 97);
        round_keys[14] = key_schedule_round(&w2, &w3, 97);
        if rounds > 14 {
            round_keys[15] = key_schedule_round(&w3, &w0, 97);
            round_keys[16] = key_schedule_round(&w0, &w1, 109);
        }
    }

    if !for_encryption {
        round_keys[..=rounds].reverse();
        for round_key in &mut round_keys[1..rounds] {
            diffuse(round_key);
        }
    }

    w0.zeroize();
    w1.zeroize();
    w2.zeroize();
    w3.zeroize();
    (round_keys, rounds)
}

/// Transforms one block with expanded round keys.
/// Variable time: the substitution layers look up S-boxes with secret data.
pub(super) fn process_block(
    round_keys: &RoundKeys,
    rounds: usize,
    input: &[u8; BLOCK_BYTES],
    output: &mut [u8; BLOCK_BYTES],
) {
    let mut state = *input;
    let mut round = 0;

    while round < rounds - 2 {
        fo(&mut state, &round_keys[round]);
        round += 1;
        fe(&mut state, &round_keys[round]);
        round += 1;
    }

    fo(&mut state, &round_keys[round]);
    round += 1;
    xor(&mut state, &round_keys[round]);
    round += 1;
    substitute_layer_2(&mut state);
    xor(&mut state, &round_keys[round]);

    *output = state;
    state.zeroize();
}

fn key_schedule_round(
    word: &[u8; BLOCK_BYTES],
    rotated_word: &[u8; BLOCK_BYTES],
    rotation: usize,
) -> [u8; BLOCK_BYTES] {
    let offset = rotation >> 3;
    let right = rotation & 7;
    let left = 8 - right;
    let mut round_key = [0u8; BLOCK_BYTES];

    let mut high = usize::from(rotated_word[15 - offset]);
    for (index, destination) in round_key.iter_mut().enumerate() {
        let low = usize::from(rotated_word[index.wrapping_sub(offset) & 0x0f]);
        *destination = (((high << left) | (low >> right)) as u8) ^ word[index];
        high = low;
    }

    round_key
}

fn fo(state: &mut [u8; BLOCK_BYTES], round_key: &[u8; BLOCK_BYTES]) {
    xor(state, round_key);
    substitute_layer_1(state);
    diffuse(state);
}

fn fe(state: &mut [u8; BLOCK_BYTES], round_key: &[u8; BLOCK_BYTES]) {
    xor(state, round_key);
    substitute_layer_2(state);
    diffuse(state);
}

fn substitute_layer_1(state: &mut [u8; BLOCK_BYTES]) {
    for chunk in state.chunks_exact_mut(4) {
        chunk[0] = SB1[usize::from(chunk[0])];
        chunk[1] = SB2[usize::from(chunk[1])];
        chunk[2] = SB3[usize::from(chunk[2])];
        chunk[3] = SB4[usize::from(chunk[3])];
    }
}

fn substitute_layer_2(state: &mut [u8; BLOCK_BYTES]) {
    for chunk in state.chunks_exact_mut(4) {
        chunk[0] = SB3[usize::from(chunk[0])];
        chunk[1] = SB4[usize::from(chunk[1])];
        chunk[2] = SB1[usize::from(chunk[2])];
        chunk[3] = SB2[usize::from(chunk[3])];
    }
}

fn diffuse(state: &mut [u8; BLOCK_BYTES]) {
    let [
        x0,
        x1,
        x2,
        x3,
        x4,
        x5,
        x6,
        x7,
        x8,
        x9,
        x10,
        x11,
        x12,
        x13,
        x14,
        x15,
    ] = *state;
    state[0] = x3 ^ x4 ^ x6 ^ x8 ^ x9 ^ x13 ^ x14;
    state[1] = x2 ^ x5 ^ x7 ^ x8 ^ x9 ^ x12 ^ x15;
    state[2] = x1 ^ x4 ^ x6 ^ x10 ^ x11 ^ x12 ^ x15;
    state[3] = x0 ^ x5 ^ x7 ^ x10 ^ x11 ^ x13 ^ x14;
    state[4] = x0 ^ x2 ^ x5 ^ x8 ^ x11 ^ x14 ^ x15;
    state[5] = x1 ^ x3 ^ x4 ^ x9 ^ x10 ^ x14 ^ x15;
    state[6] = x0 ^ x2 ^ x7 ^ x9 ^ x10 ^ x12 ^ x13;
    state[7] = x1 ^ x3 ^ x6 ^ x8 ^ x11 ^ x12 ^ x13;
    state[8] = x0 ^ x1 ^ x4 ^ x7 ^ x10 ^ x13 ^ x15;
    state[9] = x0 ^ x1 ^ x5 ^ x6 ^ x11 ^ x12 ^ x14;
    state[10] = x2 ^ x3 ^ x5 ^ x6 ^ x8 ^ x13 ^ x15;
    state[11] = x2 ^ x3 ^ x4 ^ x7 ^ x9 ^ x12 ^ x14;
    state[12] = x1 ^ x2 ^ x6 ^ x7 ^ x9 ^ x11 ^ x12;
    state[13] = x0 ^ x3 ^ x6 ^ x7 ^ x8 ^ x10 ^ x13;
    state[14] = x0 ^ x3 ^ x4 ^ x5 ^ x9 ^ x11 ^ x14;
    state[15] = x1 ^ x2 ^ x4 ^ x5 ^ x8 ^ x10 ^ x15;
}

fn xor(state: &mut [u8; BLOCK_BYTES], value: &[u8; BLOCK_BYTES]) {
    for (destination, source) in state.iter_mut().zip(value) {
        *destination ^= *source;
    }
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn diffusion_is_an_involution() {
        let original = [
            0x00, 0x11, 0x22, 0x33, 0x44, 0x55, 0x66, 0x77, 0x88, 0x99, 0xaa, 0xbb, 0xcc, 0xdd,
            0xee, 0xff,
        ];
        let mut state = original;
        diffuse(&mut state);
        diffuse(&mut state);
        assert_eq!(state, original);
    }

    #[test]
    fn s_box_pairs_are_inverses() {
        for value in 0u8..=u8::MAX {
            assert_eq!(SB3[usize::from(SB1[usize::from(value)])], value);
            assert_eq!(SB1[usize::from(SB3[usize::from(value)])], value);
            assert_eq!(SB4[usize::from(SB2[usize::from(value)])], value);
            assert_eq!(SB2[usize::from(SB4[usize::from(value)])], value);
        }
    }
}