Fix overflow in multiplication in Saber
Cette révision appartient à :
Parent
3d05000b2c
révision
c59580a355
@ -47,7 +47,7 @@ static void karatsuba_simple(const uint16_t *a_1, const uint16_t *b_1, uint16_t
|
||||
|
||||
acc7 = acc5 + acc6; //b01
|
||||
acc8 = acc1 + acc2; //a01
|
||||
d01[i + j] = d01[i + j] + acc7 * acc8;
|
||||
d01[i + j] = d01[i + j] + (uint16_t)(acc7 * (uint64_t)acc8);
|
||||
//--------------------------------------------------------
|
||||
|
||||
acc7 = b_1[j + 2 * KARATSUBA_N / 4]; //b2
|
||||
|
@ -47,7 +47,7 @@ static void karatsuba_simple(const uint16_t *a_1, const uint16_t *b_1, uint16_t
|
||||
|
||||
acc7 = acc5 + acc6; //b01
|
||||
acc8 = acc1 + acc2; //a01
|
||||
d01[i + j] = d01[i + j] + acc7 * acc8;
|
||||
d01[i + j] = d01[i + j] + (uint16_t)(acc7 * (uint64_t)acc8);
|
||||
//--------------------------------------------------------
|
||||
|
||||
acc7 = b_1[j + 2 * KARATSUBA_N / 4]; //b2
|
||||
|
@ -47,7 +47,7 @@ static void karatsuba_simple(const uint16_t *a_1, const uint16_t *b_1, uint16_t
|
||||
|
||||
acc7 = acc5 + acc6; //b01
|
||||
acc8 = acc1 + acc2; //a01
|
||||
d01[i + j] = d01[i + j] + acc7 * acc8;
|
||||
d01[i + j] = d01[i + j] + (uint16_t)(acc7 * (uint64_t)acc8);
|
||||
//--------------------------------------------------------
|
||||
|
||||
acc7 = b_1[j + 2 * KARATSUBA_N / 4]; //b2
|
||||
|
Chargement…
Référencer dans un nouveau ticket
Block a user