174 	uint64_t t0, t1, t2, t3, t4, cc;  in f255_sub()  local
191 	t4 = 1 + (uint64_t)(z >> 64);  in f255_sub()
199 	cc = (38 & -t4) + (19 & -(t3 >> 63));  in f255_sub()
218 	uint64_t t0, t1, t2, t3, t4;  in f255_sub()
225 	(void)_subborrow_u64(k, 1, 0, &t4);  in f255_sub()
231 	(void)_subborrow_u64(k, t4, 0, &t4);  in f255_sub()
239 	t4 = (38 & -t4) + (19 & -(t3 >> 63));  in f255_sub()
241 	k = _addcarry_u64(0, t0, t4, &d[0]);  in f255_sub()
258 	uint64_t t0, t1, t2, t3, t4, t5, t6, t7, th;  in f255_mul()  local
271 	t4 = (uint64_t)(z >> 64);  in f255_mul()
283 		+ (unsigned __int128)t4 + (z >> 64);  in f255_mul()
284 	t4 = (uint64_t)z;  in f255_mul()
294 		+ (unsigned __int128)t4 + (z >> 64);  in f255_mul()
295 	t4 = (uint64_t)z;  in f255_mul()
305 		+ (unsigned __int128)t4 + (z >> 64);  in f255_mul()
306 	t4 = (uint64_t)z;  in f255_mul()
322 	 * 2^255. The low one will be in t0..t3; the middle one in t4..t7.  in f255_mul()
329 	t5 = (t5 << 1) | (t4 >> 63);  in f255_mul()
330 	t4 = (t4 << 1) | (t3 >> 63);  in f255_mul()
334 	 * Multiply the middle part (t4..t7) by 19. We truncate it to  in f255_mul()
337 	z = (unsigned __int128)t4 * 19;  in f255_mul()
338 	t4 = (uint64_t)z;  in f255_mul()
352 	 *   t4..t7 fits on 255 bits.  in f255_mul()
355 	z = (unsigned __int128)t0 + (unsigned __int128)t4  in f255_mul()
381 	uint64_t t0, t1, t2, t3, t4, t5, t6, t7, th;  in f255_mul()
395 	(void)_addcarry_u64(k, h3, 0, &t4);  in f255_mul()
400 	k = _addcarry_u64(k, _umul128(a[1], b[3], &h3), t4, &t4);  in f255_mul()
404 	k = _addcarry_u64(k, t4, h2, &t4);  in f255_mul()
409 	k = _addcarry_u64(k, _umul128(a[2], b[2], &h2), t4, &t4);  in f255_mul()
413 	k = _addcarry_u64(k, t4, h1, &t4);  in f255_mul()
418 	k = _addcarry_u64(k, _umul128(a[3], b[1], &h1), t4, &t4);  in f255_mul()
422 	k = _addcarry_u64(0, t4, h0, &t4);  in f255_mul()
434 	 * 2^255. The low one will be in t0..t3; the middle one in t4..t7.  in f255_mul()
441 	t5 = (t5 << 1) | (t4 >> 63);  in f255_mul()
442 	t4 = (t4 << 1) | (t3 >> 63);  in f255_mul()
446 	 * Multiply the middle part (t4..t7) by 19. We truncate it to  in f255_mul()
449 	t4 = _umul128(t4, 19, &h0);  in f255_mul()
464 	 *   t4..t7 fits on 255 bits.  in f255_mul()
467 	k = _addcarry_u64(0, t0, t4, &t0);  in f255_mul()
471 	t4 = k;  in f255_mul()
476 	(void)_addcarry_u64(k, t4, 0, &t4);  in f255_mul()
478 	th = (t4 << 1) + (t3 >> 63);  in f255_mul()
532 	uint64_t t0, t1, t2, t3, t4, h0, h1, h2, h3;  in f255_mul_a24()
542 	(void)_addcarry_u64(k, h3, 0, &t4);  in f255_mul_a24()
544 	t4 = (t4 << 1) + (t3 >> 63);  in f255_mul_a24()
546 	k = _addcarry_u64(0, t0, 19 * t4, &t0);  in f255_mul_a24()
551 	t4 = 19 & -(t3 >> 63);  in f255_mul_a24()
553 	k = _addcarry_u64(0, t0, t4, &d[0]);  in f255_mul_a24()