375 	u64 a1 = left >> 32;  in mul_64_64()
377 	u64 b1 = right >> 32;  in mul_64_64()
383 	m2 += (m0 >> 32);  in mul_64_64()
390 	result.m_low = (m0 & 0xffffffffull) | (m2 << 32);  in mul_64_64()
391 	result.m_high = m3 + (m2 >> 32);  in mul_64_64()
411 	unsigned int i, k;  in vli_mult()  local
416 	for (k = 0; k < ndigits * 2 - 1; k++) {  in vli_mult()
419 		if (k < ndigits)  in vli_mult()
422 			min = (k + 1) - ndigits;  in vli_mult()
424 		for (i = min; i <= k && i < ndigits; i++) {  in vli_mult()
427 			product = mul_64_64(left[i], right[k - i]);  in vli_mult()
433 		result[k] = r01.m_low;  in vli_mult()
447 	unsigned int k;  in vli_umult()  local
449 	for (k = 0; k < ndigits; k++) {  in vli_umult()
452 		product = mul_64_64(left[k], right);  in vli_umult()
455 		result[k] = r01.m_low;  in vli_umult()
459 	result[k] = r01.m_low;  in vli_umult()
460 	for (++k; k < ndigits * 2; k++)  in vli_umult()
461 		result[k] = 0;  in vli_umult()
468 	int i, k;  in vli_square()  local
470 	for (k = 0; k < ndigits * 2 - 1; k++) {  in vli_square()
473 		if (k < ndigits)  in vli_square()
476 			min = (k + 1) - ndigits;  in vli_square()
478 		for (i = min; i <= k && i <= k - i; i++) {  in vli_square()
481 			product = mul_64_64(left[i], left[k - i]);  in vli_square()
483 			if (i < k - i) {  in vli_square()
494 		result[k] = r01.m_low;  in vli_square()
538  * for special form moduli: p = 2^k-c, for small c (note the minus sign)
567  * for special form moduli: p = 2^{k-1}+c, for small c (note the plus sign)
568  * where k-1 does not fit into qword boundary by -1 bit (such as 255).
575  * H. Cohen, G. Frey, R. Avanzi, C. Doche, T. Lange, K. Nguyen, F. Vercauteren.
749 	tmp[1] = product[6] << 32;  in vli_mmod_fast_256()
750 	tmp[2] = (product[6] >> 32) | (product[7] << 32);  in vli_mmod_fast_256()
751 	tmp[3] = product[7] >> 32;  in vli_mmod_fast_256()
763 	tmp[0] = (product[4] >> 32) | (product[5] << 32);  in vli_mmod_fast_256()
764 	tmp[1] = (product[5] >> 32) | (product[6] & 0xffffffff00000000ull);  in vli_mmod_fast_256()
766 	tmp[3] = (product[6] >> 32) | (product[4] << 32);  in vli_mmod_fast_256()
770 	tmp[0] = (product[5] >> 32) | (product[6] << 32);  in vli_mmod_fast_256()
771 	tmp[1] = (product[6] >> 32);  in vli_mmod_fast_256()
773 	tmp[3] = (product[4] & 0xffffffff) | (product[5] << 32);  in vli_mmod_fast_256()
780 	tmp[3] = (product[4] >> 32) | (product[5] & 0xffffffff00000000ull);  in vli_mmod_fast_256()
784 	tmp[0] = (product[6] >> 32) | (product[7] << 32);  in vli_mmod_fast_256()
785 	tmp[1] = (product[7] >> 32) | (product[4] << 32);  in vli_mmod_fast_256()
786 	tmp[2] = (product[4] >> 32) | (product[5] << 32);  in vli_mmod_fast_256()
787 	tmp[3] = (product[6] << 32);  in vli_mmod_fast_256()
807 #define SL32OR32(x32, y32) (((u64)x32 << 32) | y32)
826 	tmp[2] = SL32OR32(product[11], (product[10]>>32));	//a22||a21  in vli_mmod_fast_384()
827 	tmp[3] = product[11]>>32;	// 0 ||a23  in vli_mmod_fast_384()
843 	tmp[0] = SL32OR32(product[11], (product[10]>>32));	//a22||a21  in vli_mmod_fast_384()
844 	tmp[1] = SL32OR32(product[6], (product[11]>>32));	//a12||a23  in vli_mmod_fast_384()
845 	tmp[2] = SL32OR32(product[7], (product[6])>>32);	//a14||a13  in vli_mmod_fast_384()
846 	tmp[3] = SL32OR32(product[8], (product[7]>>32));	//a16||a15  in vli_mmod_fast_384()
847 	tmp[4] = SL32OR32(product[9], (product[8]>>32));	//a18||a17  in vli_mmod_fast_384()
848 	tmp[5] = SL32OR32(product[10], (product[9]>>32));	//a20||a19  in vli_mmod_fast_384()
853 	tmp[1] = (product[10]<<32);	//a20|| 0  in vli_mmod_fast_384()
879 	tmp[0] = SL32OR32(product[6], (product[11]>>32));	//a12||a23  in vli_mmod_fast_384()
880 	tmp[1] = SL32OR32(product[7], (product[6]>>32));	//a14||a13  in vli_mmod_fast_384()
881 	tmp[2] = SL32OR32(product[8], (product[7]>>32));	//a16||a15  in vli_mmod_fast_384()
882 	tmp[3] = SL32OR32(product[9], (product[8]>>32));	//a18||a17  in vli_mmod_fast_384()
883 	tmp[4] = SL32OR32(product[10], (product[9]>>32));	//a20||a19  in vli_mmod_fast_384()
884 	tmp[5] = SL32OR32(product[11], (product[10]>>32));	//a22||a21  in vli_mmod_fast_384()
888 	tmp[0] = (product[10]<<32);	//a20|| 0  in vli_mmod_fast_384()
889 	tmp[1] = SL32OR32(product[11], (product[10]>>32));	//a22||a21  in vli_mmod_fast_384()
890 	tmp[2] = (product[11]>>32);	// 0 ||a23  in vli_mmod_fast_384()
899 	tmp[2] = product[11]>>32;	// 0 ||a23  in vli_mmod_fast_384()