Skip to content

Commit

Permalink
Remove some manual copy loops
Browse files Browse the repository at this point in the history
  • Loading branch information
@sebastinas
sebastinas committed Sep 21, 2023
1 parent e908a2c commit 06063a6
Showing 1 changed file with 46 additions and 85 deletions.
131 changes: 46 additions & 85 deletions faest_aes.c
View file
Original file line number Diff line number Diff line change
Expand Up @@ -168,25 +168,20 @@ static void aes_key_schedule_forward_128(const bf128_t* v, uint8_t Mtag, uint8_t
return;
}

const unsigned int lambda = params->faest_param.lambda;
const unsigned int R = params->faest_param.R;
const unsigned int Nwd = params->faest_param.Nwd;
const unsigned int lambdaBytes = lambda / 8;
const unsigned int lambda = params->faest_param.lambda;
const unsigned int R = params->faest_param.R;
const unsigned int Nwd = params->faest_param.Nwd;

for (uint32_t i = 0; i < lambdaBytes; i++) {
for (uint32_t j = 0; j < 8; j++) {
bf_out[(i * 8) + j] = v[i * 8 + j];
}
}
memcpy(bf_out, v, lambda * sizeof(bf128_t));

// Step: 4
uint32_t i_wd = lambda;
// Step: 5..10
for (uint32_t j = Nwd; j < 4 * (R + 1); j++) {
if ((j % Nwd) == 0 || (Nwd > 6 && (j % Nwd) == 4)) {
for (uint32_t i = (j * 32); i <= ((j * 32) + 31); i++, i_wd++) {
bf_out[i] = v[i_wd];
}
// copy all at once
memcpy(bf_out + j * 32, v + i_wd, sizeof(bf128_t) * 32);
i_wd += 32;
} else {
for (uint32_t i = 0; i < 32; i++) {
bf_out[(32 * j) + i] = bf128_add(bf_out[32 * (j - Nwd) + i], bf_out[32 * (j - 1) + i]);
Expand Down Expand Up @@ -239,18 +234,14 @@ static void aes_key_schedule_backward_128(const bf128_t* v, const bf128_t* Vk, u
}
}

bf128_t bf_y_tilde[8];
for (uint32_t i = 0; i < 8; ++i) {
bf_y_tilde[i] = bf128_add(bf128_add(bf_x_tilde[(i + 7) % 8], bf_x_tilde[(i + 5) % 8]),
bf_x_tilde[(i + 2) % 8]);
}
bf_y_tilde[0] = bf128_add(bf_y_tilde[0], bf128_mul_bit(bf_mkey_times_delta, minus_mtag));
bf_y_tilde[2] = bf128_add(bf_y_tilde[2], bf128_mul_bit(bf_mkey_times_delta, minus_mtag));

// TODO: get rid of this copy
for (uint32_t i = 0; i < 8; i++) {
bf_out[(8 * j) + i] = bf_y_tilde[i];
bf_out[i + 8 * j] = bf128_add(bf128_add(bf_x_tilde[(i + 7) % 8], bf_x_tilde[(i + 5) % 8]),
bf_x_tilde[(i + 2) % 8]);
}
bf_out[0 + 8 * j] =
bf128_add(bf_out[0 + 8 * j], bf128_mul_bit(bf_mkey_times_delta, minus_mtag));
bf_out[2 + 8 * j] =
bf128_add(bf_out[2 + 8 * j], bf128_mul_bit(bf_mkey_times_delta, minus_mtag));
c = c + 1;

if (c == 4) {
Expand Down Expand Up @@ -528,17 +519,14 @@ static void aes_enc_backward_128(const bf128_t* bf_x, const bf128_t* bf_xk, uint
// Step: 6
if (j < (R - 1)) {
// Step: 7
for (uint32_t i = 0; i < 8; i++) {
bf_x_tilde[i] = bf_x[ird + i];
}
memcpy(bf_x_tilde, bf_x + ird, sizeof(bf_x_tilde));
} else {
bf128_t bf_xout[8];
// Step: 10
for (uint32_t i = 0; i < 8; ++i) {
// Step: 11
bf_xout[i] = bf128_mul_bit(factor, get_bit(out[(ird - 128 * (R - 1)) / 8], i));
bf128_t bf_xout = bf128_mul_bit(factor, get_bit(out[(ird - 128 * (R - 1)) / 8], i));
// Step: 12
bf_x_tilde[i] = bf128_add(bf_xout[i], bf_xk[128 + ird + i]);
bf_x_tilde[i] = bf128_add(bf_xout, bf_xk[128 + ird + i]);
}
}
// Step: 13..17
Expand Down Expand Up @@ -733,25 +721,19 @@ static void aes_key_schedule_forward_192(const bf192_t* v, uint8_t Mtag, uint8_t
return;
}

const unsigned int lambda = params->faest_param.lambda;
const unsigned int R = params->faest_param.R;
const unsigned int Nwd = params->faest_param.Nwd;
const unsigned int lambdaBytes = lambda / 8;
const unsigned int lambda = params->faest_param.lambda;
const unsigned int R = params->faest_param.R;
const unsigned int Nwd = params->faest_param.Nwd;

for (uint32_t i = 0; i < lambdaBytes; i++) {
for (uint32_t j = 0; j < 8; j++) {
bf_out[(i * 8) + j] = v[i * 8 + j];
}
}
memcpy(bf_out, v, lambda * sizeof(bf192_t));

// Step: 4
uint32_t i_wd = lambda;
// Step: 5..10
for (uint32_t j = Nwd; j < 4 * (R + 1); j++) {
if ((j % Nwd) == 0 || (Nwd > 6 && (j % Nwd) == 4)) {
for (uint32_t i = (j * 32); i <= ((j * 32) + 31); i++, i_wd++) {
bf_out[i] = v[i_wd];
}
memcpy(bf_out + j * 32, v + i_wd, sizeof(bf192_t) * 32);
i_wd += 32;
} else {
for (uint32_t i = 0; i < 32; i++) {
bf_out[(32 * j) + i] = bf192_add(bf_out[32 * (j - Nwd) + i], bf_out[32 * (j - 1) + i]);
Expand Down Expand Up @@ -803,18 +785,14 @@ static void aes_key_schedule_backward_192(const bf192_t* v, const bf192_t* Vk, u
}
}

bf192_t bf_y_tilde[8];
for (uint32_t i = 0; i < 8; ++i) {
bf_y_tilde[i] = bf192_add(bf192_add(bf_x_tilde[(i + 7) % 8], bf_x_tilde[(i + 5) % 8]),
bf_x_tilde[(i + 2) % 8]);
}
bf_y_tilde[0] = bf192_add(bf_y_tilde[0], bf192_mul_bit(bf_mkey_times_delta, minus_mtag));
bf_y_tilde[2] = bf192_add(bf_y_tilde[2], bf192_mul_bit(bf_mkey_times_delta, minus_mtag));

// TODO: get rid of this copy
for (uint32_t i = 0; i < 8; i++) {
bf_out[(8 * j) + i] = bf_y_tilde[i];
bf_out[i + 8 * j] = bf192_add(bf192_add(bf_x_tilde[(i + 7) % 8], bf_x_tilde[(i + 5) % 8]),
bf_x_tilde[(i + 2) % 8]);
}
bf_out[0 + 8 * j] =
bf192_add(bf_out[0 + 8 * j], bf192_mul_bit(bf_mkey_times_delta, minus_mtag));
bf_out[2 + 8 * j] =
bf192_add(bf_out[2 + 8 * j], bf192_mul_bit(bf_mkey_times_delta, minus_mtag));
c = c + 1;

if (c == 4) {
Expand Down Expand Up @@ -1099,17 +1077,14 @@ static void aes_enc_backward_192(const bf192_t* bf_x, const bf192_t* bf_xk, uint
// Step: 6
if (j < (R - 1)) {
// Step: 7
for (uint32_t i = 0; i < 8; i++) {
bf_x_tilde[i] = bf_x[ird + i];
}
memcpy(bf_x_tilde, bf_x + ird, sizeof(bf_x_tilde));
} else {
bf192_t bf_xout[8];
// Step: 10
for (uint32_t i = 0; i < 8; ++i) {
// Step: 11
bf_xout[i] = bf192_mul_bit(factor, get_bit(out[(ird - 128 * (R - 1)) / 8], i));
bf192_t bf_xout = bf192_mul_bit(factor, get_bit(out[(ird - 128 * (R - 1)) / 8], i));
// Step: 12
bf_x_tilde[i] = bf192_add(bf_xout[i], bf_xk[128 + ird + i]);
bf_x_tilde[i] = bf192_add(bf_xout, bf_xk[128 + ird + i]);
}
}
// Step: 13..17
Expand Down Expand Up @@ -1312,26 +1287,19 @@ static void aes_key_schedule_forward_256(const bf256_t* v, uint8_t Mtag, uint8_t
return;
}

const unsigned int lambda = params->faest_param.lambda;
const unsigned int R = params->faest_param.R;
const unsigned int Nwd = params->faest_param.Nwd;
const unsigned int lambdaBytes = lambda / 8;
const unsigned int lambda = params->faest_param.lambda;
const unsigned int R = params->faest_param.R;
const unsigned int Nwd = params->faest_param.Nwd;

// TODO: replace with one memcpy
for (uint32_t i = 0; i < lambdaBytes; i++) {
for (uint32_t j = 0; j < 8; j++) {
bf_out[(i * 8) + j] = v[i * 8 + j];
}
}
memcpy(bf_out, v, sizeof(bf256_t) * lambda);

// Step: 4
uint32_t i_wd = lambda;
// Step: 5..10
for (uint32_t j = Nwd; j < 4 * (R + 1); j++) {
if ((j % Nwd) == 0 || (Nwd > 6 && (j % Nwd) == 4)) {
for (uint32_t i = (j * 32); i <= ((j * 32) + 31); i++, i_wd++) {
bf_out[i] = v[i_wd];
}
memcpy(bf_out + j * 32, v + i_wd, sizeof(bf256_t) * 32);
i_wd += 32;
} else {
for (uint32_t i = 0; i < 32; i++) {
bf_out[(32 * j) + i] = bf256_add(bf_out[32 * (j - Nwd) + i], bf_out[32 * (j - 1) + i]);
Expand Down Expand Up @@ -1383,18 +1351,14 @@ static void aes_key_schedule_backward_256(const bf256_t* v, const bf256_t* Vk, u
}
}

bf256_t bf_y_tilde[8];
for (uint32_t i = 0; i < 8; ++i) {
bf_y_tilde[i] = bf256_add(bf256_add(bf_x_tilde[(i + 7) % 8], bf_x_tilde[(i + 5) % 8]),
bf_x_tilde[(i + 2) % 8]);
}
bf_y_tilde[0] = bf256_add(bf_y_tilde[0], bf256_mul_bit(bf_mkey_times_delta, minus_mtag));
bf_y_tilde[2] = bf256_add(bf_y_tilde[2], bf256_mul_bit(bf_mkey_times_delta, minus_mtag));

// TODO: get rid of this copy
for (uint32_t i = 0; i < 8; i++) {
bf_out[(8 * j) + i] = bf_y_tilde[i];
bf_out[i + 8 * j] = bf256_add(bf256_add(bf_x_tilde[(i + 7) % 8], bf_x_tilde[(i + 5) % 8]),
bf_x_tilde[(i + 2) % 8]);
}
bf_out[0 + 8 * j] =
bf256_add(bf_out[0 + 8 * j], bf256_mul_bit(bf_mkey_times_delta, minus_mtag));
bf_out[2 + 8 * j] =
bf256_add(bf_out[2 + 8 * j], bf256_mul_bit(bf_mkey_times_delta, minus_mtag));
c = c + 1;

if (c == 4) {
Expand Down Expand Up @@ -1697,17 +1661,14 @@ static void aes_enc_backward_256(const bf256_t* bf_x, const bf256_t* bf_xk, uint
// Step: 6
if (j < (R - 1)) {
// Step: 7
for (uint32_t i = 0; i < 8; i++) {
bf_x_tilde[i] = bf_x[ird + i];
}
memcpy(bf_x_tilde, bf_x + ird, sizeof(bf_x_tilde));
} else {
bf256_t bf_xout[8];
// Step: 10
for (uint32_t i = 0; i < 8; ++i) {
// Step: 11
bf_xout[i] = bf256_mul_bit(factor, get_bit(out[(ird - 128 * (R - 1)) / 8], i));
bf256_t bf_xout = bf256_mul_bit(factor, get_bit(out[(ird - 128 * (R - 1)) / 8], i));
// Step: 12
bf_x_tilde[i] = bf256_add(bf_xout[i], bf_xk[128 + ird + i]);
bf_x_tilde[i] = bf256_add(bf_xout, bf_xk[128 + ird + i]);
}
}
// Step: 13..17
Expand Down

0 comments on commit 06063a6

Please sign in to comment.