1
1
mirror of https://github.com/henrydcase/pqc.git synced 2024-11-25 08:51:26 +00:00

Fix an avx2/gf2x.c buffer overflow

This commit is contained in:
John M. Schanck 2020-09-07 18:06:37 -04:00 committed by Kris Kwiatkowski
parent cb7675a9b6
commit d5fd7d6d0c
6 changed files with 18 additions and 26 deletions

View File

@ -511,9 +511,8 @@ static void TOOM3Mult(__m256i *Out, const uint64_t *A, const uint64_t *B) {
_mm256_storeu_si256(&U2_256[i], W3[i] ^ _mm256_loadu_si256(&U2_256[i]));
}
for (int32_t i = 0 ; i < 6 * T_TM3_3W_256 - 2 ; i++) {
uint64_t *out64 = ((uint64_t *)Out) + (i << 2);
_mm256_storeu_si256((__m256i *)out64, ro256[i]);
for (int32_t i = 0 ; i < 2 * VEC_N_SIZE_256 + 1 ; i++) {
_mm256_storeu_si256(&Out[i], ro256[i]);
}
}

View File

@ -418,6 +418,7 @@ static void TOOM3Mult(__m256i *Out, const uint64_t *A, const uint64_t *B) {
//W1 = W2 * W3
karat_mult_64( W1, W2, W3);
//W0 =(U1 + U2*x)*x ; W4 =(V1 + V2*x)*x (SIZE = T_TM3_3W_256 !)
int64_t *U1_64 = ((int64_t *) U1);
int64_t *U2_64 = ((int64_t *) U2);
@ -455,6 +456,7 @@ static void TOOM3Mult(__m256i *Out, const uint64_t *A, const uint64_t *B) {
W4[i] ^= V0[i];
}
karat_mult_64(tmp, W3, W2);
for (int32_t i = 0 ; i < 2 * (T_TM3_3W_256) ; i++) {
@ -462,7 +464,6 @@ static void TOOM3Mult(__m256i *Out, const uint64_t *A, const uint64_t *B) {
}
karat_mult_64( W2, W0, W4);
//W4 = U2 * V2 ; W0 = U0 * V0
karat_mult_64(W4, U2, V2);
karat_mult_64(W0, U0, V0);
@ -550,14 +551,12 @@ static void TOOM3Mult(__m256i *Out, const uint64_t *A, const uint64_t *B) {
_mm256_storeu_si256(&U2_256[i], W3[i] ^ _mm256_loadu_si256(&U2_256[i]));
}
for (int32_t i = 0 ; i < 6 * T_TM3_3W_256 - 2 ; i++) {
uint64_t *out64 = ((uint64_t *)Out) + (i << 2);
_mm256_storeu_si256((__m256i *)out64, ro256[i]);
for (int32_t i = 0 ; i < 2 * VEC_N_SIZE_256 + 1 ; i++) {
_mm256_storeu_si256(&Out[i], ro256[i]);
}
}
/**
* @brief Multiply two polynomials modulo \f$ X^n - 1\f$.
*

View File

@ -519,8 +519,7 @@ static void TOOM3Mult(__m256i *Out, const uint64_t *A, const uint64_t *B) {
}
for (int32_t i = 0 ; i < 6 * T_TM3_3W_256 - 2 ; i++) {
uint64_t *out64 = ((uint64_t *)Out) + (i << 2);
_mm256_storeu_si256((__m256i *)out64, ro256[i]);
_mm256_storeu_si256(&Out[i], ro256[i]);
}
}
@ -715,9 +714,8 @@ static void TOOM3RecMult(__m256i *Out, const uint64_t *A, const uint64_t *B) {
ro256[3 + 5 * T_TM3R_3W_256] ^= W3[3 + 2 * T_TM3R_3W_256];
for (int32_t i = 0 ; i < 6 * T_TM3R_3W_256 - 2 ; i++) {
uint64_t *out64 = ((uint64_t *)Out) + (i << 2);
_mm256_storeu_si256((__m256i *)out64, ro256[i]);
for (int32_t i = 0 ; i < 2 * VEC_N_SIZE_256 + 1 ; i++) {
_mm256_storeu_si256(&Out[i], ro256[i]);
}
}

View File

@ -511,9 +511,8 @@ static void TOOM3Mult(__m256i *Out, const uint64_t *A, const uint64_t *B) {
_mm256_storeu_si256(&U2_256[i], W3[i] ^ _mm256_loadu_si256(&U2_256[i]));
}
for (int32_t i = 0 ; i < 6 * T_TM3_3W_256 - 2 ; i++) {
uint64_t *out64 = ((uint64_t *)Out) + (i << 2);
_mm256_storeu_si256((__m256i *)out64, ro256[i]);
for (int32_t i = 0 ; i < 2 * VEC_N_SIZE_256 + 1 ; i++) {
_mm256_storeu_si256(&Out[i], ro256[i]);
}
}

View File

@ -418,6 +418,7 @@ static void TOOM3Mult(__m256i *Out, const uint64_t *A, const uint64_t *B) {
//W1 = W2 * W3
karat_mult_64( W1, W2, W3);
//W0 =(U1 + U2*x)*x ; W4 =(V1 + V2*x)*x (SIZE = T_TM3_3W_256 !)
int64_t *U1_64 = ((int64_t *) U1);
int64_t *U2_64 = ((int64_t *) U2);
@ -455,6 +456,7 @@ static void TOOM3Mult(__m256i *Out, const uint64_t *A, const uint64_t *B) {
W4[i] ^= V0[i];
}
karat_mult_64(tmp, W3, W2);
for (int32_t i = 0 ; i < 2 * (T_TM3_3W_256) ; i++) {
@ -462,7 +464,6 @@ static void TOOM3Mult(__m256i *Out, const uint64_t *A, const uint64_t *B) {
}
karat_mult_64( W2, W0, W4);
//W4 = U2 * V2 ; W0 = U0 * V0
karat_mult_64(W4, U2, V2);
karat_mult_64(W0, U0, V0);
@ -550,14 +551,12 @@ static void TOOM3Mult(__m256i *Out, const uint64_t *A, const uint64_t *B) {
_mm256_storeu_si256(&U2_256[i], W3[i] ^ _mm256_loadu_si256(&U2_256[i]));
}
for (int32_t i = 0 ; i < 6 * T_TM3_3W_256 - 2 ; i++) {
uint64_t *out64 = ((uint64_t *)Out) + (i << 2);
_mm256_storeu_si256((__m256i *)out64, ro256[i]);
for (int32_t i = 0 ; i < 2 * VEC_N_SIZE_256 + 1 ; i++) {
_mm256_storeu_si256(&Out[i], ro256[i]);
}
}
/**
* @brief Multiply two polynomials modulo \f$ X^n - 1\f$.
*

View File

@ -519,8 +519,7 @@ static void TOOM3Mult(__m256i *Out, const uint64_t *A, const uint64_t *B) {
}
for (int32_t i = 0 ; i < 6 * T_TM3_3W_256 - 2 ; i++) {
uint64_t *out64 = ((uint64_t *)Out) + (i << 2);
_mm256_storeu_si256((__m256i *)out64, ro256[i]);
_mm256_storeu_si256(&Out[i], ro256[i]);
}
}
@ -715,9 +714,8 @@ static void TOOM3RecMult(__m256i *Out, const uint64_t *A, const uint64_t *B) {
ro256[3 + 5 * T_TM3R_3W_256] ^= W3[3 + 2 * T_TM3R_3W_256];
for (int32_t i = 0 ; i < 6 * T_TM3R_3W_256 - 2 ; i++) {
uint64_t *out64 = ((uint64_t *)Out) + (i << 2);
_mm256_storeu_si256((__m256i *)out64, ro256[i]);
for (int32_t i = 0 ; i < 2 * VEC_N_SIZE_256 + 1 ; i++) {
_mm256_storeu_si256(&Out[i], ro256[i]);
}
}