static void
do_sync_xor(struct page *dest, struct page **src_list, unsigned int offset,
int src_cnt, size_t len, enum async_tx_flags flags,
dma_async_tx_callback cb_fn, void *cb_param)
{
int i;
int xor_src_cnt;
int src_off = 0;
void *dest_buf;
void **srcs = (void **) src_list;
/* reuse the 'src_list' array to convert to buffer pointers */
for (i = 0; i < src_cnt; i++)
srcs[i] = page_address(src_list[i]) + offset;
/* set destination address */
dest_buf = page_address(dest) + offset;
if (flags & ASYNC_TX_XOR_ZERO_DST)
memset(dest_buf, 0, len);
while (src_cnt > 0) {
/* process up to 'MAX_XOR_BLOCKS' sources */
xor_src_cnt = min(src_cnt, MAX_XOR_BLOCKS);
xor_blocks(xor_src_cnt, len, dest_buf, &srcs[src_off]);
/* drop completed sources */
src_cnt -= xor_src_cnt;
src_off += xor_src_cnt;
}
async_tx_sync_epilog(cb_fn, cb_param);
}
static void
do_sync_xor(struct page *dest, struct page **src_list, unsigned int offset,
int src_cnt, size_t len, struct async_submit_ctl *submit)
{
int i;
int xor_src_cnt = 0;
int src_off = 0;
void *dest_buf;
void **srcs;
if (submit->scribble)
srcs = submit->scribble;
else
srcs = (void **) src_list;
/* convert to buffer pointers */
for (i = 0; i < src_cnt; i++)
if (src_list[i])
srcs[xor_src_cnt++] = page_address(src_list[i]) + offset;
src_cnt = xor_src_cnt;
/* set destination address */
dest_buf = page_address(dest) + offset;
if (submit->flags & ASYNC_TX_XOR_ZERO_DST)
memset(dest_buf, 0, len);
while (src_cnt > 0) {
/* process up to 'MAX_XOR_BLOCKS' sources */
xor_src_cnt = min(src_cnt, MAX_XOR_BLOCKS);
xor_blocks(xor_src_cnt, len, dest_buf, &srcs[src_off]);
/* drop
static void
do_sync_xor(struct page *dest, struct page **src_list, unsigned int offset,
unsigned int src_cnt, size_t len, enum async_tx_flags flags,
struct dma_async_tx_descriptor *depend_tx,
dma_async_tx_callback cb_fn, void *cb_param)
{
void *_dest;
int i;
pr_debug("%s: len: %zu\n", __func__, len);
/* reuse the 'src_list' array to convert to buffer pointers */
for (i = 0; i < src_cnt; i++)
src_list[i] = (struct page *)
(page_address(src_list[i]) + offset);
/* set destination address */
_dest = page_address(dest) + offset;
if (flags & ASYNC_TX_XOR_ZERO_DST)
memset(_dest, 0, len);
xor_blocks(src_cnt, len, _dest,
(void **) src_list);
async_tx_sync_epilog(flags, depend_tx, cb_fn, cb_param);
}
/*
* Calculate H(i+1) = Hash(Hi,Mi)
* Where H and M are 32 bytes long
*/
static int hash_step(gost_ctx * c, byte * H, const byte * M)
{
byte U[32], W[32], V[32], S[32], Key[32];
int i;
/* Compute first key */
xor_blocks(W, H, M, 32);
swap_bytes(W, Key);
/* Encrypt first 8 bytes of H with first key */
gost_enc_with_key(c, Key, H, S);
/* Compute second key */
circle_xor8(H, U);
circle_xor8(M, V);
circle_xor8(V, V);
xor_blocks(W, U, V, 32);
swap_bytes(W, Key);
/* encrypt second 8 bytes of H with second key */
gost_enc_with_key(c, Key, H + 8, S + 8);
/* compute third key */
circle_xor8(U, U);
U[31] = ~U[31];
U[29] = ~U[29];
U[28] = ~U[28];
U[24] = ~U[24];
U[23] = ~U[23];
U[20] = ~U[20];
U[18] = ~U[18];
U[17] = ~U[17];
U[14] = ~U[14];
U[12] = ~U[12];
U[10] = ~U[10];
U[8] = ~U[8];
U[7] = ~U[7];
U[5] = ~U[5];
U[3] = ~U[3];
U[1] = ~U[1];
circle_xor8(V, V);
circle_xor8(V, V);
xor_blocks(W, U, V, 32);
swap_bytes(W, Key);
/* encrypt third 8 bytes of H with third key */
gost_enc_with_key(c, Key, H + 16, S + 16);
/* Compute fourth key */
circle_xor8(U, U);
circle_xor8(V, V);
circle_xor8(V, V);
xor_blocks(W, U, V, 32);
swap_bytes(W, Key);
/* Encrypt last 8 bytes with fourth key */
gost_enc_with_key(c, Key, H + 24, S + 24);
for (i = 0; i < 12; i++)
transform_3(S);
xor_blocks(S, S, M, 32);
transform_3(S);
xor_blocks(S, S, H, 32);
for (i = 0; i < 61; i++)
transform_3(S);
memcpy(H, S, 32);
return 1;
}
void break_protocol_1()
{
uint16_t victim_id = 1234;
uint16_t my_id;
uint8_t *message;
size_t len;
uint8_t iv[AES_BLOCK_SIZE], fake_iv[AES_BLOCK_SIZE];
uint8_t mac[AES_BLOCK_SIZE];
uint8_t modified_block[AES_BLOCK_SIZE + 1], delta_block[AES_BLOCK_SIZE + 1]; /* For the null char added in sprintf. Just ignore it. */
my_id = client_get_id();
message = client_generate_request_proto1(&len, iv, mac, my_id, 1000000);
snprintf(modified_block, sizeof modified_block, "from=#{%05hu}&to", victim_id);
xor_blocks(delta_block, modified_block, message, AES_BLOCK_SIZE);
xor_blocks(fake_iv, iv, delta_block, AES_BLOCK_SIZE);
memcpy(message, modified_block, AES_BLOCK_SIZE);
printf("Protocol 1 server response:\n");
server_process_request_proto1(message, len, fake_iv, mac);
free(message);
}
void break_protocol_2()
{
uint16_t target_id = 1234;
uint16_t my_id;
uint8_t *target_message, *my_message;
size_t target_len, my_len;
uint8_t target_mac[AES_BLOCK_SIZE], my_mac[AES_BLOCK_SIZE];
uint8_t *extended_message;
/* This part is more ambiguous. I don't think it can be solved
* without making some hand-waving in the server's parsing.
* We will first sign a valid transaction from our account
* to our own account for 1M spacebucks. We then obtain
* a valid message from the target (presumably obtained by
* sniffing network traffic), and concatenate our own message.
* In order for the MAC to remain the same, the first block of
* our message will be xored with the mac of the target message.
* This "simulates" an IV of 0 for our message, keeping the same
* MAC as a result.
*
* We have (message, mac) pairs (M, t) and (M', t'), and we generate
* the pair (M || (M'[0] ^ t) || M'[1..], t').
*
* Of course, this means there will be junk between the end of
* the target's transaction list and the beggining of ours,
* so a correct parsing implementation would reject that as invalid
* request. For the purpose of the challenge, we will assume the
* parsing implementation is very dumb...
*
* This still has a slight change of failing if a semicolon appears as
* part of the junk data, because the parser will look for the
* transactions based on the semicolons. */
my_id = client_get_id();
my_message = client_generate_request_proto2(&my_len, my_mac, my_id, 1000000);
target_message = client_capture_request_proto2(&target_len, target_mac, target_id);
extended_message = malloc(target_len + my_len);
memcpy(extended_message, target_message, target_len);
xor_blocks(extended_message + target_len, my_message, target_mac, AES_BLOCK_SIZE);
memcpy(extended_message + target_len + AES_BLOCK_SIZE, my_message + AES_BLOCK_SIZE, my_len - AES_BLOCK_SIZE);
printf("Protocol 2 server response:\n");
server_process_request_proto2(extended_message, target_len + my_len, my_mac);
free(extended_message);
free(target_message);
free(my_message);
}
static void
do_sync_xor(struct page *dest, struct page **src_list, unsigned int offset,
int src_cnt, size_t len, struct async_submit_ctl *submit)
{
int i;
int xor_src_cnt = 0;
int src_off = 0;
void *dest_buf;
void **srcs;
if (submit->scribble)
srcs = submit->scribble;
else
srcs = (void **) src_list;
for (i = 0; i < src_cnt; i++)
if (src_list[i])
srcs[xor_src_cnt++] = page_address(src_list[i]) + offset;
src_cnt = xor_src_cnt;
dest_buf = page_address(dest) + offset;
if (submit->flags & ASYNC_TX_XOR_ZERO_DST)
memset(dest_buf, 0, len);
while (src_cnt > 0) {
xor_src_cnt = min(src_cnt, MAX_XOR_BLOCKS);
xor_blocks(xor_src_cnt, len, dest_buf, &srcs[src_off]);
src_cnt -= xor_src_cnt;
src_off += xor_src_cnt;
}
async_tx_sync_epilog(submit);
}
void cn_slow_hash(const void *data, size_t length, char *hash) {
uint8_t long_state[MEMORY];
union cn_slow_hash_state state;
uint8_t text[INIT_SIZE_BYTE];
uint8_t a[AES_BLOCK_SIZE];
uint8_t b[AES_BLOCK_SIZE];
uint8_t c[AES_BLOCK_SIZE];
uint8_t d[AES_BLOCK_SIZE];
size_t i, j;
uint8_t aes_key[AES_KEY_SIZE];
oaes_ctx *aes_ctx;
hash_process(&state.hs, data, length);
memcpy(text, state.init, INIT_SIZE_BYTE);
memcpy(aes_key, state.hs.b, AES_KEY_SIZE);
aes_ctx = (oaes_ctx *) oaes_alloc();
oaes_key_import_data(aes_ctx, aes_key, AES_KEY_SIZE);
for (i = 0; i < MEMORY / INIT_SIZE_BYTE; i++) {
for (j = 0; j < INIT_SIZE_BLK; j++) {
aesb_pseudo_round(&text[AES_BLOCK_SIZE * j], &text[AES_BLOCK_SIZE * j], aes_ctx->key->exp_data);
}
memcpy(&long_state[i * INIT_SIZE_BYTE], text, INIT_SIZE_BYTE);
}
for (i = 0; i < 16; i++) {
a[i] = state.k[ i] ^ state.k[32 + i];
b[i] = state.k[16 + i] ^ state.k[48 + i];
}
for (i = 0; i < ITER / 2; i++) {
/* Dependency chain: address -> read value ------+
* written value <-+ hard function (AES or MUL) <+
* next address <-+
*/
/* Iteration 1 */
j = e2i(a, MEMORY / AES_BLOCK_SIZE);
copy_block(c, &long_state[j * AES_BLOCK_SIZE]);
aesb_single_round(c, c, a);
xor_blocks(b, c);
swap_blocks(b, c);
copy_block(&long_state[j * AES_BLOCK_SIZE], c);
//assert(j == e2i(a, MEMORY / AES_BLOCK_SIZE));
swap_blocks(a, b);
/* Iteration 2 */
j = e2i(a, MEMORY / AES_BLOCK_SIZE);
copy_block(c, &long_state[j * AES_BLOCK_SIZE]);
mul(a, c, d);
sum_half_blocks(b, d);
swap_blocks(b, c);
xor_blocks(b, c);
copy_block(&long_state[j * AES_BLOCK_SIZE], c);
//assert(j == e2i(a, MEMORY / AES_BLOCK_SIZE));
swap_blocks(a, b);
}
memcpy(text, state.init, INIT_SIZE_BYTE);
oaes_key_import_data(aes_ctx, &state.hs.b[32], AES_KEY_SIZE);
for (i = 0; i < MEMORY / INIT_SIZE_BYTE; i++) {
for (j = 0; j < INIT_SIZE_BLK; j++) {
xor_blocks(&text[j * AES_BLOCK_SIZE], &long_state[i * INIT_SIZE_BYTE + j * AES_BLOCK_SIZE]);
aesb_pseudo_round(&text[AES_BLOCK_SIZE * j], &text[AES_BLOCK_SIZE * j], aes_ctx->key->exp_data);
}
}
memcpy(state.init, text, INIT_SIZE_BYTE);
hash_permutation(&state.hs);
extra_hashes[state.hs.b[0] & 3](&state, 200, hash);
oaes_free((OAES_CTX **) &aes_ctx);
}
void cryptonight_hash_ctx_aes_ni(void* output, const void* input, size_t len, struct cryptonight_ctx* ctx) {
hash_process(&ctx->state.hs, (const uint8_t*) input, len);
ctx->aes_ctx = (oaes_ctx*) oaes_alloc();
size_t i, j;
memcpy(ctx->text, ctx->state.init, INIT_SIZE_BYTE);
oaes_key_import_data(ctx->aes_ctx, ctx->state.hs.b, AES_KEY_SIZE);
for (i = 0; likely(i < MEMORY); i += INIT_SIZE_BYTE) {
fast_aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 0], ctx->aes_ctx->key->exp_data);
fast_aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 1], ctx->aes_ctx->key->exp_data);
fast_aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 2], ctx->aes_ctx->key->exp_data);
fast_aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 3], ctx->aes_ctx->key->exp_data);
fast_aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 4], ctx->aes_ctx->key->exp_data);
fast_aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 5], ctx->aes_ctx->key->exp_data);
fast_aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 6], ctx->aes_ctx->key->exp_data);
fast_aesb_pseudo_round_mut(&ctx->text[AES_BLOCK_SIZE * 7], ctx->aes_ctx->key->exp_data);
memcpy(&ctx->long_state[i], ctx->text, INIT_SIZE_BYTE);
}
xor_blocks_dst(&ctx->state.k[0], &ctx->state.k[32], ctx->a);
xor_blocks_dst(&ctx->state.k[16], &ctx->state.k[48], ctx->b);
for (i = 0; likely(i < ITER / 4); ++i) {
/* Dependency chain: address -> read value ------+
* written value <-+ hard function (AES or MUL) <+
* next address <-+
*/
/* Iteration 1 */
j = e2i(ctx->a);
fast_aesb_single_round(&ctx->long_state[j], ctx->c, ctx->a);
xor_blocks_dst(ctx->c, ctx->b, &ctx->long_state[j]);
/* Iteration 2 */
mul_sum_xor_dst(ctx->c, ctx->a, &ctx->long_state[e2i(ctx->c)]);
/* Iteration 3 */
j = e2i(ctx->a);
fast_aesb_single_round(&ctx->long_state[j], ctx->b, ctx->a);
xor_blocks_dst(ctx->b, ctx->c, &ctx->long_state[j]);
/* Iteration 4 */
mul_sum_xor_dst(ctx->b, ctx->a, &ctx->long_state[e2i(ctx->b)]);
}
memcpy(ctx->text, ctx->state.init, INIT_SIZE_BYTE);
oaes_key_import_data(ctx->aes_ctx, &ctx->state.hs.b[32], AES_KEY_SIZE);
for (i = 0; likely(i < MEMORY); i += INIT_SIZE_BYTE) {
xor_blocks(&ctx->text[0 * AES_BLOCK_SIZE], &ctx->long_state[i + 0 * AES_BLOCK_SIZE]);
fast_aesb_pseudo_round_mut(&ctx->text[0 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
xor_blocks(&ctx->text[1 * AES_BLOCK_SIZE], &ctx->long_state[i + 1 * AES_BLOCK_SIZE]);
fast_aesb_pseudo_round_mut(&ctx->text[1 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
xor_blocks(&ctx->text[2 * AES_BLOCK_SIZE], &ctx->long_state[i + 2 * AES_BLOCK_SIZE]);
fast_aesb_pseudo_round_mut(&ctx->text[2 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
xor_blocks(&ctx->text[3 * AES_BLOCK_SIZE], &ctx->long_state[i + 3 * AES_BLOCK_SIZE]);
fast_aesb_pseudo_round_mut(&ctx->text[3 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
xor_blocks(&ctx->text[4 * AES_BLOCK_SIZE], &ctx->long_state[i + 4 * AES_BLOCK_SIZE]);
fast_aesb_pseudo_round_mut(&ctx->text[4 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
xor_blocks(&ctx->text[5 * AES_BLOCK_SIZE], &ctx->long_state[i + 5 * AES_BLOCK_SIZE]);
fast_aesb_pseudo_round_mut(&ctx->text[5 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
xor_blocks(&ctx->text[6 * AES_BLOCK_SIZE], &ctx->long_state[i + 6 * AES_BLOCK_SIZE]);
fast_aesb_pseudo_round_mut(&ctx->text[6 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
xor_blocks(&ctx->text[7 * AES_BLOCK_SIZE], &ctx->long_state[i + 7 * AES_BLOCK_SIZE]);
fast_aesb_pseudo_round_mut(&ctx->text[7 * AES_BLOCK_SIZE], ctx->aes_ctx->key->exp_data);
}
memcpy(ctx->state.init, ctx->text, INIT_SIZE_BYTE);
hash_permutation(&ctx->state.hs);
/*memcpy(hash, &state, 32);*/
extra_hashes[ctx->state.hs.b[0] & 3](&ctx->state, 200, output);
oaes_free((OAES_CTX **) &ctx->aes_ctx);
}
void decryptMessage(unsigned char* messageIn, unsigned char* decryptedMessage)
{
unsigned char buffer[16];
int i, j;
unsigned char tmp;
uint32_t key_schedule[11][4];
int mode = messageIn[12]; // 0,1,2,3
printf("mode = %02x\n", mode);
generate_key_schedule(initial_session_key, key_schedule);
// For M0-M6 we follow the same pattern
for (i = 0; i < 8; i++)
{
// First, copy in the nth block (we must start with the last one)
for (j = 0; j < 16; j++)
{
if (mode == 3)
buffer[j] = messageIn[(0x80-0x10*i)+j];
else if (mode == 2 || mode == 1 || mode == 0)
buffer[j] = messageIn[(0x10*(i+1))+j];
}
// do this permutation and update 9 times. Could this be cycle(), or the reverse of cycle()?
for (j = 0; j < 9; j++)
{
int base = 0x80 - 0x10*j;
//print_block("About to cycle. Buffer is currently: ", buffer);
buffer[0x0] = message_table_index(base+0x0)[buffer[0x0]] ^ message_key[mode][base+0x0];
buffer[0x4] = message_table_index(base+0x4)[buffer[0x4]] ^ message_key[mode][base+0x4];
buffer[0x8] = message_table_index(base+0x8)[buffer[0x8]] ^ message_key[mode][base+0x8];
buffer[0xc] = message_table_index(base+0xc)[buffer[0xc]] ^ message_key[mode][base+0xc];
tmp = buffer[0x0d];
buffer[0xd] = message_table_index(base+0xd)[buffer[0x9]] ^ message_key[mode][base+0xd];
buffer[0x9] = message_table_index(base+0x9)[buffer[0x5]] ^ message_key[mode][base+0x9];
buffer[0x5] = message_table_index(base+0x5)[buffer[0x1]] ^ message_key[mode][base+0x5];
buffer[0x1] = message_table_index(base+0x1)[tmp] ^ message_key[mode][base+0x1];
tmp = buffer[0x02];
buffer[0x2] = message_table_index(base+0x2)[buffer[0xa]] ^ message_key[mode][base+0x2];
buffer[0xa] = message_table_index(base+0xa)[tmp] ^ message_key[mode][base+0xa];
tmp = buffer[0x06];
buffer[0x6] = message_table_index(base+0x6)[buffer[0xe]] ^ message_key[mode][base+0x6];
buffer[0xe] = message_table_index(base+0xe)[tmp] ^ message_key[mode][base+0xe];
tmp = buffer[0x3];
buffer[0x3] = message_table_index(base+0x3)[buffer[0x7]] ^ message_key[mode][base+0x3];
buffer[0x7] = message_table_index(base+0x7)[buffer[0xb]] ^ message_key[mode][base+0x7];
buffer[0xb] = message_table_index(base+0xb)[buffer[0xf]] ^ message_key[mode][base+0xb];
buffer[0xf] = message_table_index(base+0xf)[tmp] ^ message_key[mode][base+0xf];
// Now we must replace the entire buffer with 4 words that we read and xor together
uint32_t word;
uint32_t* block = (uint32_t*)buffer;
block[0] = table_s9[0x000 + buffer[0x0]] ^
table_s9[0x100 + buffer[0x1]] ^
table_s9[0x200 + buffer[0x2]] ^
table_s9[0x300 + buffer[0x3]];
block[1] = table_s9[0x000 + buffer[0x4]] ^
table_s9[0x100 + buffer[0x5]] ^
table_s9[0x200 + buffer[0x6]] ^
table_s9[0x300 + buffer[0x7]];
block[2] = table_s9[0x000 + buffer[0x8]] ^
table_s9[0x100 + buffer[0x9]] ^
table_s9[0x200 + buffer[0xa]] ^
table_s9[0x300 + buffer[0xb]];
block[3] = table_s9[0x000 + buffer[0xc]] ^
table_s9[0x100 + buffer[0xd]] ^
table_s9[0x200 + buffer[0xe]] ^
table_s9[0x300 + buffer[0xf]];
}
// Next, another permute with a different table
buffer[0x0] = table_s10[(0x0 << 8) + buffer[0x0]];
buffer[0x4] = table_s10[(0x4 << 8) + buffer[0x4]];
buffer[0x8] = table_s10[(0x8 << 8) + buffer[0x8]];
buffer[0xc] = table_s10[(0xc << 8) + buffer[0xc]];
tmp = buffer[0x0d];
buffer[0xd] = table_s10[(0xd << 8) + buffer[0x9]];
buffer[0x9] = table_s10[(0x9 << 8) + buffer[0x5]];
buffer[0x5] = table_s10[(0x5 << 8) + buffer[0x1]];
buffer[0x1] = table_s10[(0x1 << 8) + tmp];
tmp = buffer[0x02];
buffer[0x2] = table_s10[(0x2 << 8) + buffer[0xa]];
buffer[0xa] = table_s10[(0xa << 8) + tmp];
tmp = buffer[0x06];
buffer[0x6] = table_s10[(0x6 << 8) + buffer[0xe]];
buffer[0xe] = table_s10[(0xe << 8) + tmp];
tmp = buffer[0x3];
buffer[0x3] = table_s10[(0x3 << 8) + buffer[0x7]];
buffer[0x7] = table_s10[(0x7 << 8) + buffer[0xb]];
buffer[0xb] = table_s10[(0xb << 8) + buffer[0xf]];
buffer[0xf] = table_s10[(0xf << 8) + tmp];
// And finally xor with the previous block of the message, except in mode-2 where we do this in reverse
if (mode == 2 || mode == 1 || mode == 0)
{
if (i > 0)
{
xor_blocks(buffer, &messageIn[0x10*i], &decryptedMessage[0x10*i]); // remember that the first 0x10 bytes are the header
}
else
xor_blocks(buffer, message_iv[mode], &decryptedMessage[0x10*i]);
print_block(" ", &decryptedMessage[0x10*i]);
}
else
{
if (i < 7)
xor_blocks(buffer, &messageIn[0x70 - 0x10*i], &decryptedMessage[0x70 - 0x10*i]);
else
xor_blocks(buffer, message_iv[mode], &decryptedMessage[0x70 - 0x10*i]);
printf("Decrypted message block %02X-%02X:", 0x70 - 0x10*i, 0x70 - 0x10*i+0xf);
print_block(" ", &decryptedMessage[0x70 - 0x10*i]);
}
}
}
/* Restore data:
* We are given:
* A list of 'fds' of the active disks. Some may be '-1' for not-available.
* A geometry: raid_disks, chunk_size, level, layout
* An 'fd' to read from. It is already seeked to the right (Read) location.
* A start and length.
* The length must be a multiple of the stripe size.
*
* We build a full stripe in memory and then write it out.
* We assume that there are enough working devices.
*/
int restore_stripes(int *dest, unsigned long long *offsets,
int raid_disks, int chunk_size, int level, int layout,
int source, unsigned long long read_offset,
unsigned long long start, unsigned long long length,
char *src_buf)
{
char *stripe_buf;
char **stripes = xmalloc(raid_disks * sizeof(char*));
char **blocks = xmalloc(raid_disks * sizeof(char*));
int i;
int rv;
int data_disks = raid_disks - (level == 0 ? 0 : level <= 5 ? 1 : 2);
if (posix_memalign((void**)&stripe_buf, 4096, raid_disks * chunk_size))
stripe_buf = NULL;
if (zero == NULL || chunk_size > zero_size) {
if (zero)
free(zero);
zero = xcalloc(1, chunk_size);
zero_size = chunk_size;
}
if (stripe_buf == NULL || stripes == NULL || blocks == NULL
|| zero == NULL) {
rv = -2;
goto abort;
}
for (i = 0; i < raid_disks; i++)
stripes[i] = stripe_buf + i * chunk_size;
while (length > 0) {
unsigned int len = data_disks * chunk_size;
unsigned long long offset;
int disk, qdisk;
int syndrome_disks;
if (length < len) {
rv = -3;
goto abort;
}
for (i = 0; i < data_disks; i++) {
int disk = geo_map(i, start/chunk_size/data_disks,
raid_disks, level, layout);
if (src_buf == NULL) {
/* read from file */
if (lseek64(source, read_offset, 0) !=
(off64_t)read_offset) {
rv = -1;
goto abort;
}
if (read(source,
stripes[disk],
chunk_size) != chunk_size) {
rv = -1;
goto abort;
}
} else {
/* read from input buffer */
memcpy(stripes[disk],
src_buf + read_offset,
chunk_size);
}
read_offset += chunk_size;
}
/* We have the data, now do the parity */
offset = (start/chunk_size/data_disks) * chunk_size;
switch (level) {
case 4:
case 5:
disk = geo_map(-1, start/chunk_size/data_disks,
raid_disks, level, layout);
for (i = 0; i < data_disks; i++)
blocks[i] = stripes[(disk+1+i) % raid_disks];
xor_blocks(stripes[disk], blocks, data_disks, chunk_size);
break;
case 6:
disk = geo_map(-1, start/chunk_size/data_disks,
raid_disks, level, layout);
qdisk = geo_map(-2, start/chunk_size/data_disks,
raid_disks, level, layout);
if (is_ddf(layout)) {
/* q over 'raid_disks' blocks, in device order.
* 'p' and 'q' get to be all zero
*/
for (i = 0; i < raid_disks; i++)
if (i == disk || i == qdisk)
blocks[i] = (char*)zero;
else
blocks[i] = stripes[i];
syndrome_disks = raid_disks;
} else {
/* for md, q is over 'data_disks' blocks,
* starting immediately after 'q'
*/
for (i = 0; i < data_disks; i++)
blocks[i] = stripes[(qdisk+1+i) % raid_disks];
syndrome_disks = data_disks;
}
qsyndrome((uint8_t*)stripes[disk],
(uint8_t*)stripes[qdisk],
(uint8_t**)blocks,
syndrome_disks, chunk_size);
break;
}
for (i=0; i < raid_disks ; i++)
if (dest[i] >= 0) {
if (lseek64(dest[i],
offsets[i]+offset, 0) < 0) {
rv = -1;
goto abort;
}
if (write(dest[i], stripes[i],
chunk_size) != chunk_size) {
rv = -1;
goto abort;
}
}
length -= len;
start += len;
}
rv = 0;
abort:
free(stripe_buf);
free(stripes);
free(blocks);
return rv;
}
/*******************************************************************************
* Function: save_stripes
* Description:
* Function reads data (only data without P and Q) from array and writes
* it to buf and opcjonaly to backup files
* Parameters:
* source : A list of 'fds' of the active disks.
* Some may be absent
* offsets : A list of offsets on disk belonging
* to the array [bytes]
* raid_disks : geometry: number of disks in the array
* chunk_size : geometry: chunk size [bytes]
* level : geometry: RAID level
* layout : geometry: layout
* nwrites : number of backup files
* dest : A list of 'fds' for mirrored targets
* (e.g. backup files). They are already seeked to right
* (write) location. If NULL, data will be wrote
* to the buf only
* start : start address of data to read (must be stripe-aligned)
* [bytes]
* length - : length of data to read (must be stripe-aligned)
* [bytes]
* buf : buffer for data. It is large enough to hold
* one stripe. It is stripe aligned
* Returns:
* 0 : success
* -1 : fail
******************************************************************************/
int save_stripes(int *source, unsigned long long *offsets,
int raid_disks, int chunk_size, int level, int layout,
int nwrites, int *dest,
unsigned long long start, unsigned long long length,
char *buf)
{
int len;
int data_disks = raid_disks - (level == 0 ? 0 : level <=5 ? 1 : 2);
int disk;
int i;
unsigned long long length_test;
if (!tables_ready)
make_tables();
ensure_zero_has_size(chunk_size);
len = data_disks * chunk_size;
length_test = length / len;
length_test *= len;
if (length != length_test) {
dprintf("Error: save_stripes(): Data are not alligned. EXIT\n");
dprintf("\tArea for saving stripes (length) = %llu\n", length);
dprintf("\tWork step (len) = %i\n", len);
dprintf("\tExpected save area (length_test) = %llu\n",
length_test);
abort();
}
while (length > 0) {
int failed = 0;
int fdisk[3], fblock[3];
for (disk = 0; disk < raid_disks ; disk++) {
unsigned long long offset;
int dnum;
offset = (start/chunk_size/data_disks)*chunk_size;
dnum = geo_map(disk < data_disks ? disk : data_disks - disk - 1,
start/chunk_size/data_disks,
raid_disks, level, layout);
if (dnum < 0) abort();
if (source[dnum] < 0 ||
lseek64(source[dnum], offsets[dnum]+offset, 0) < 0 ||
read(source[dnum], buf+disk * chunk_size, chunk_size)
!= chunk_size)
if (failed <= 2) {
fdisk[failed] = dnum;
fblock[failed] = disk;
failed++;
}
}
if (failed == 0 || fblock[0] >= data_disks)
/* all data disks are good */
;
else if (failed == 1 || fblock[1] >= data_disks+1) {
/* one failed data disk and good parity */
char *bufs[data_disks];
for (i=0; i < data_disks; i++)
if (fblock[0] == i)
bufs[i] = buf + data_disks*chunk_size;
else
bufs[i] = buf + i*chunk_size;
xor_blocks(buf + fblock[0]*chunk_size,
bufs, data_disks, chunk_size);
} else if (failed > 2 || level != 6)
/* too much failure */
return -1;
else {
/* RAID6 computations needed. */
uint8_t *bufs[data_disks+4];
int qdisk;
int syndrome_disks;
disk = geo_map(-1, start/chunk_size/data_disks,
raid_disks, level, layout);
qdisk = geo_map(-2, start/chunk_size/data_disks,
raid_disks, level, layout);
if (is_ddf(layout)) {
/* q over 'raid_disks' blocks, in device order.
* 'p' and 'q' get to be all zero
*/
for (i = 0; i < raid_disks; i++)
bufs[i] = zero;
for (i = 0; i < data_disks; i++) {
int dnum = geo_map(i,
start/chunk_size/data_disks,
raid_disks, level, layout);
int snum;
/* i is the logical block number, so is index to 'buf'.
* dnum is physical disk number
* and thus the syndrome number.
*/
snum = dnum;
bufs[snum] = (uint8_t*)buf + chunk_size * i;
}
syndrome_disks = raid_disks;
} else {
/* for md, q is over 'data_disks' blocks,
* starting immediately after 'q'
* Note that for the '_6' variety, the p block
* makes a hole that we need to be careful of.
*/
int j;
int snum = 0;
for (j = 0; j < raid_disks; j++) {
int dnum = (qdisk + 1 + j) % raid_disks;
if (dnum == disk || dnum == qdisk)
continue;
for (i = 0; i < data_disks; i++)
if (geo_map(i,
start/chunk_size/data_disks,
raid_disks, level, layout) == dnum)
break;
/* i is the logical block number, so is index to 'buf'.
* dnum is physical disk number
* snum is syndrome disk for which 0 is immediately after Q
*/
bufs[snum] = (uint8_t*)buf + chunk_size * i;
if (fblock[0] == i)
fdisk[0] = snum;
if (fblock[1] == i)
fdisk[1] = snum;
snum++;
}
syndrome_disks = data_disks;
}
/* Place P and Q blocks at end of bufs */
bufs[syndrome_disks] = (uint8_t*)buf + chunk_size * data_disks;
bufs[syndrome_disks+1] = (uint8_t*)buf + chunk_size * (data_disks+1);
if (fblock[1] == data_disks)
/* One data failed, and parity failed */
raid6_datap_recov(syndrome_disks+2, chunk_size,
fdisk[0], bufs, 0);
else {
/* Two data blocks failed, P,Q OK */
raid6_2data_recov(syndrome_disks+2, chunk_size,
fdisk[0], fdisk[1], bufs, 0);
}
}
if (dest) {
for (i = 0; i < nwrites; i++)
if (write(dest[i], buf, len) != len)
return -1;
} else {
/* build next stripe in buffer */
buf += len;
}
length -= len;
start += len;
}
return 0;
}
void cryptonight_hash(const char* input, char* output, uint32_t len) {
uint8_t long_state[MEMORY];
union cn_slow_hash_state state;
uint8_t text[INIT_SIZE_BYTE];
uint8_t a[AES_BLOCK_SIZE];
uint8_t b[AES_BLOCK_SIZE];
uint8_t c[AES_BLOCK_SIZE];
uint8_t d[AES_BLOCK_SIZE];
size_t i, j;
uint8_t aes_key[AES_KEY_SIZE];
OAES_CTX* aes_ctx;
hash_process(&state.hs, (const uint8_t*) input, len);
memcpy(text, state.init, INIT_SIZE_BYTE);
memcpy(aes_key, state.hs.b, AES_KEY_SIZE);
aes_ctx = oaes_alloc();
oaes_key_import_data(aes_ctx, aes_key, AES_KEY_SIZE);
for (i = 0; i < MEMORY / INIT_SIZE_BYTE; i++) {
for (j = 0; j < INIT_SIZE_BLK; j++) {
oaes_pseudo_encrypt_ecb(aes_ctx, &text[AES_BLOCK_SIZE * j]);
}
memcpy(&long_state[i * INIT_SIZE_BYTE], text, INIT_SIZE_BYTE);
}
for (i = 0; i < 16; i++) {
a[i] = state.k[i] ^ state.k[32 + i];
b[i] = state.k[16 + i] ^ state.k[48 + i];
}
for (i = 0; i < ITER / 2; i++) {
/* Dependency chain: address -> read value ------+
* written value <-+ hard function (AES or MUL) <+
* next address <-+
*/
/* Iteration 1 */
j = e2i(a, MEMORY / AES_BLOCK_SIZE);
copy_block(c, &long_state[j * AES_BLOCK_SIZE]);
oaes_encryption_round(a, c);
xor_blocks(b, c);
swap_blocks(b, c);
copy_block(&long_state[j * AES_BLOCK_SIZE], c);
assert(j == e2i(a, MEMORY / AES_BLOCK_SIZE));
swap_blocks(a, b);
/* Iteration 2 */
j = e2i(a, MEMORY / AES_BLOCK_SIZE);
copy_block(c, &long_state[j * AES_BLOCK_SIZE]);
mul(a, c, d);
sum_half_blocks(b, d);
swap_blocks(b, c);
xor_blocks(b, c);
copy_block(&long_state[j * AES_BLOCK_SIZE], c);
swap_blocks(a, b);
}
memcpy(text, state.init, INIT_SIZE_BYTE);
oaes_key_import_data(aes_ctx, &state.hs.b[32], AES_KEY_SIZE);
for (i = 0; i < MEMORY / INIT_SIZE_BYTE; i++) {
for (j = 0; j < INIT_SIZE_BLK; j++) {
xor_blocks(&text[j * AES_BLOCK_SIZE],
&long_state[i * INIT_SIZE_BYTE + j * AES_BLOCK_SIZE]);
oaes_pseudo_encrypt_ecb(aes_ctx, &text[j * AES_BLOCK_SIZE]);
}
}
memcpy(state.init, text, INIT_SIZE_BYTE);
hash_permutation(&state.hs);
/*memcpy(hash, &state, 32);*/
extra_hashes[state.hs.b[0] & 3](&state, 200, output);
oaes_free(&aes_ctx);
}
/*
* Calculate H(i+1) = Hash(Hi,Mi)
* Where H and M are 32 bytes long
*/
static int
hash_step(GOSTR341194_CTX *c, unsigned char *H, const unsigned char *M)
{
unsigned char U[32], W[32], V[32], S[32], Key[32];
int i;
/* Compute first key */
xor_blocks(W, H, M, 32);
swap_bytes(W, Key);
/* Encrypt first 8 bytes of H with first key */
Gost2814789_set_key(&c->cipher, Key, 256);
Gost2814789_encrypt(H, S, &c->cipher);
/* Compute second key */
circle_xor8(H, U);
circle_xor8(M, V);
circle_xor8(V, V);
xor_blocks(W, U, V, 32);
swap_bytes(W, Key);
/* encrypt second 8 bytes of H with second key */
Gost2814789_set_key(&c->cipher, Key, 256);
Gost2814789_encrypt(H+8, S+8, &c->cipher);
/* compute third key */
circle_xor8(U, U);
U[31] = ~U[31];
U[29] = ~U[29];
U[28] = ~U[28];
U[24] = ~U[24];
U[23] = ~U[23];
U[20] = ~U[20];
U[18] = ~U[18];
U[17] = ~U[17];
U[14] = ~U[14];
U[12] = ~U[12];
U[10] = ~U[10];
U[8] = ~U[8];
U[7] = ~U[7];
U[5] = ~U[5];
U[3] = ~U[3];
U[1] = ~U[1];
circle_xor8(V, V);
circle_xor8(V, V);
xor_blocks(W, U, V, 32);
swap_bytes(W, Key);
/* encrypt third 8 bytes of H with third key */
Gost2814789_set_key(&c->cipher, Key, 256);
Gost2814789_encrypt(H+16, S+16, &c->cipher);
/* Compute fourth key */
circle_xor8(U, U);
circle_xor8(V, V);
circle_xor8(V, V);
xor_blocks(W, U, V, 32);
swap_bytes(W, Key);
/* Encrypt last 8 bytes with fourth key */
Gost2814789_set_key(&c->cipher, Key, 256);
Gost2814789_encrypt(H+24, S+24, &c->cipher);
for (i = 0; i < 12; i++)
transform_3(S);
xor_blocks(S, S, M, 32);
transform_3(S);
xor_blocks(S, S, H, 32);
for (i = 0; i < 61; i++)
transform_3(S);
memcpy(H, S, 32);
return 1;
}
int check_stripes(struct mdinfo *info, int *source, unsigned long long *offsets,
int raid_disks, int chunk_size, int level, int layout,
unsigned long long start, unsigned long long length, char *name[],
int repair, int failed_disk1, int failed_disk2)
{
/* read the data and p and q blocks, and check we got them right */
char *stripe_buf = xmalloc(raid_disks * chunk_size);
char **stripes = xmalloc(raid_disks * sizeof(char*));
char **blocks = xmalloc(raid_disks * sizeof(char*));
int *block_index_for_slot = xmalloc(raid_disks * sizeof(int));
uint8_t *p = xmalloc(chunk_size);
uint8_t *q = xmalloc(chunk_size);
int *results = xmalloc(chunk_size * sizeof(int));
sighandler_t *sig = xmalloc(3 * sizeof(sighandler_t));
int i;
int diskP, diskQ;
int data_disks = raid_disks - 2;
int err = 0;
extern int tables_ready;
if (!tables_ready)
make_tables();
for ( i = 0 ; i < raid_disks ; i++)
stripes[i] = stripe_buf + i * chunk_size;
while (length > 0) {
int disk;
printf("pos --> %llu\n", start);
err = lock_stripe(info, start, chunk_size, data_disks, sig);
if(err != 0) {
if (err != 2)
unlock_all_stripes(info, sig);
goto exitCheck;
}
for (i = 0 ; i < raid_disks ; i++) {
lseek64(source[i], offsets[i] + start * chunk_size, 0);
read(source[i], stripes[i], chunk_size);
}
err = unlock_all_stripes(info, sig);
if(err != 0)
goto exitCheck;
for (i = 0 ; i < data_disks ; i++) {
int disk = geo_map(i, start, raid_disks, level, layout);
blocks[i] = stripes[disk];
block_index_for_slot[disk] = i;
printf("%d->%d\n", i, disk);
}
qsyndrome(p, q, (uint8_t**)blocks, data_disks, chunk_size);
diskP = geo_map(-1, start, raid_disks, level, layout);
diskQ = geo_map(-2, start, raid_disks, level, layout);
blocks[data_disks] = stripes[diskP];
block_index_for_slot[diskP] = data_disks;
blocks[data_disks+1] = stripes[diskQ];
block_index_for_slot[diskQ] = data_disks+1;
if (memcmp(p, stripes[diskP], chunk_size) != 0) {
printf("P(%d) wrong at %llu\n", diskP, start);
}
if (memcmp(q, stripes[diskQ], chunk_size) != 0) {
printf("Q(%d) wrong at %llu\n", diskQ, start);
}
raid6_collect(chunk_size, p, q, stripes[diskP], stripes[diskQ], results);
disk = raid6_stats(results, raid_disks, chunk_size);
if(disk >= -2) {
disk = geo_map(disk, start, raid_disks, level, layout);
}
if(disk >= 0) {
printf("Error detected at %llu: possible failed disk slot: %d --> %s\n",
start, disk, name[disk]);
}
if(disk == -65535) {
printf("Error detected at %llu: disk slot unknown\n", start);
}
if(repair == 1) {
printf("Repairing stripe %llu\n", start);
printf("Assuming slots %d (%s) and %d (%s) are incorrect\n",
failed_disk1, name[failed_disk1],
failed_disk2, name[failed_disk2]);
if (failed_disk1 == diskQ || failed_disk2 == diskQ) {
char *all_but_failed_blocks[data_disks];
int failed_data_or_p;
int failed_block_index;
if (failed_disk1 == diskQ)
failed_data_or_p = failed_disk2;
else
failed_data_or_p = failed_disk1;
printf("Repairing D/P(%d) and Q\n", failed_data_or_p);
failed_block_index = block_index_for_slot[failed_data_or_p];
for (i=0; i < data_disks; i++)
if (failed_block_index == i)
all_but_failed_blocks[i] = stripes[diskP];
else
all_but_failed_blocks[i] = blocks[i];
xor_blocks(stripes[failed_data_or_p],
all_but_failed_blocks, data_disks, chunk_size);
qsyndrome(p, (uint8_t*)stripes[diskQ], (uint8_t**)blocks, data_disks, chunk_size);
} else {
ensure_zero_has_size(chunk_size);
if (failed_disk1 == diskP || failed_disk2 == diskP) {
int failed_data, failed_block_index;
if (failed_disk1 == diskP)
failed_data = failed_disk2;
else
failed_data = failed_disk1;
failed_block_index = block_index_for_slot[failed_data];
printf("Repairing D(%d) and P\n", failed_data);
raid6_datap_recov(raid_disks, chunk_size, failed_block_index, (uint8_t**)blocks);
} else {
printf("Repairing D and D\n");
int failed_block_index1 = block_index_for_slot[failed_disk1];
int failed_block_index2 = block_index_for_slot[failed_disk2];
if (failed_block_index1 > failed_block_index2) {
int t = failed_block_index1;
failed_block_index1 = failed_block_index2;
failed_block_index2 = t;
}
raid6_2data_recov(raid_disks, chunk_size, failed_block_index1, failed_block_index2, (uint8_t**)blocks);
}
}
err = lock_stripe(info, start, chunk_size, data_disks, sig);
if(err != 0) {
if (err != 2)
unlock_all_stripes(info, sig);
goto exitCheck;
}
lseek64(source[failed_disk1], offsets[failed_disk1] + start * chunk_size, 0);
write(source[failed_disk1], stripes[failed_disk1], chunk_size);
lseek64(source[failed_disk2], offsets[failed_disk2] + start * chunk_size, 0);
write(source[failed_disk2], stripes[failed_disk2], chunk_size);
err = unlock_all_stripes(info, sig);
if(err != 0)
goto exitCheck;
} else if (disk >= 0 && repair == 2) {
printf("Auto-repairing slot %d (%s)\n", disk, name[disk]);
if (disk == diskQ) {
qsyndrome(p, (uint8_t*)stripes[diskQ], (uint8_t**)blocks, data_disks, chunk_size);
} else {
char *all_but_failed_blocks[data_disks];
int failed_block_index = block_index_for_slot[disk];
for (i=0; i < data_disks; i++)
if (failed_block_index == i)
all_but_failed_blocks[i] = stripes[diskP];
else
all_but_failed_blocks[i] = blocks[i];
xor_blocks(stripes[disk],
all_but_failed_blocks, data_disks, chunk_size);
}
err = lock_stripe(info, start, chunk_size, data_disks, sig);
if(err != 0) {
if (err != 2)
unlock_all_stripes(info, sig);
goto exitCheck;
}
lseek64(source[disk], offsets[disk] + start * chunk_size, 0);
write(source[disk], stripes[disk], chunk_size);
err = unlock_all_stripes(info, sig);
if(err != 0)
goto exitCheck;
}
length--;
start++;
}
exitCheck:
free(stripe_buf);
free(stripes);
free(blocks);
free(p);
free(q);
free(results);
return err;
}
static void bitfix_xor32(void *dest, const void *src, size_t count)
{
xor_blocks(1, PAGE_SIZE, dest, (void **)&src);
}
void cryptonight_hash(const char* input, char* output, uint32_t len, int variant, uint64_t height) {
struct cryptonight_ctx *ctx = alloca(sizeof(struct cryptonight_ctx));
hash_process(&ctx->state.hs, (const uint8_t*) input, len);
memcpy(ctx->text, ctx->state.init, INIT_SIZE_BYTE);
memcpy(ctx->aes_key, ctx->state.hs.b, AES_KEY_SIZE);
ctx->aes_ctx = (oaes_ctx*) oaes_alloc();
size_t i, j;
VARIANT1_INIT();
VARIANT2_INIT(ctx->b, ctx->state);
VARIANT4_RANDOM_MATH_INIT(ctx->state);
oaes_key_import_data(ctx->aes_ctx, ctx->aes_key, AES_KEY_SIZE);
for (i = 0; i < MEMORY / INIT_SIZE_BYTE; i++) {
for (j = 0; j < INIT_SIZE_BLK; j++) {
aesb_pseudo_round(&ctx->text[AES_BLOCK_SIZE * j],
&ctx->text[AES_BLOCK_SIZE * j],
ctx->aes_ctx->key->exp_data);
}
memcpy(&ctx->long_state[i * INIT_SIZE_BYTE], ctx->text, INIT_SIZE_BYTE);
}
for (i = 0; i < 16; i++) {
ctx->a[i] = ctx->state.k[i] ^ ctx->state.k[32 + i];
ctx->b[i] = ctx->state.k[16 + i] ^ ctx->state.k[48 + i];
}
for (i = 0; i < ITER / 2; i++) {
/* Dependency chain: address -> read value ------+
* written value <-+ hard function (AES or MUL) <+
* next address <-+
*/
/* Iteration 1 */
j = e2i(ctx->a);
aesb_single_round(&ctx->long_state[j * AES_BLOCK_SIZE], ctx->c, ctx->a);
VARIANT2_SHUFFLE_ADD(ctx->long_state, j * AES_BLOCK_SIZE, ctx->a, ctx->b, ctx->c);
xor_blocks_dst(ctx->c, ctx->b, &ctx->long_state[j * AES_BLOCK_SIZE]);
VARIANT1_1((uint8_t*)&ctx->long_state[j * AES_BLOCK_SIZE]);
/* Iteration 2 */
j = e2i(ctx->c);
uint64_t* dst = (uint64_t*)&ctx->long_state[j * AES_BLOCK_SIZE];
uint64_t t[2];
t[0] = dst[0];
t[1] = dst[1];
VARIANT2_INTEGER_MATH(t, ctx->c);
copy_block(ctx->a1, ctx->a);
VARIANT4_RANDOM_MATH(ctx->a, t, r, ctx->b, ctx->b + AES_BLOCK_SIZE);
uint64_t hi;
uint64_t lo = mul128(((uint64_t*)ctx->c)[0], t[0], &hi);
VARIANT2_2();
VARIANT2_SHUFFLE_ADD(ctx->long_state, j * AES_BLOCK_SIZE, ctx->a1, ctx->b, ctx->c);
((uint64_t*)ctx->a)[0] += hi;
((uint64_t*)ctx->a)[1] += lo;
dst[0] = ((uint64_t*)ctx->a)[0];
dst[1] = ((uint64_t*)ctx->a)[1];
((uint64_t*)ctx->a)[0] ^= t[0];
((uint64_t*)ctx->a)[1] ^= t[1];
VARIANT1_2((uint8_t*)&ctx->long_state[j * AES_BLOCK_SIZE]);
copy_block(ctx->b + AES_BLOCK_SIZE, ctx->b);
copy_block(ctx->b, ctx->c);
}
memcpy(ctx->text, ctx->state.init, INIT_SIZE_BYTE);
oaes_key_import_data(ctx->aes_ctx, &ctx->state.hs.b[32], AES_KEY_SIZE);
for (i = 0; i < MEMORY / INIT_SIZE_BYTE; i++) {
for (j = 0; j < INIT_SIZE_BLK; j++) {
xor_blocks(&ctx->text[j * AES_BLOCK_SIZE],
&ctx->long_state[i * INIT_SIZE_BYTE + j * AES_BLOCK_SIZE]);
aesb_pseudo_round(&ctx->text[j * AES_BLOCK_SIZE],
&ctx->text[j * AES_BLOCK_SIZE],
ctx->aes_ctx->key->exp_data);
}
}
memcpy(ctx->state.init, ctx->text, INIT_SIZE_BYTE);
hash_permutation(&ctx->state.hs);
/*memcpy(hash, &state, 32);*/
extra_hashes[ctx->state.hs.b[0] & 3](&ctx->state, 200, output);
oaes_free((OAES_CTX **) &ctx->aes_ctx);
}
