arm/crypto/blake2b-neon-core.S

65 // pointer points to a 32-byte aligned buffer containing a copy of q8 and q9
74 	// (v[2], v[6], v[10], v[14]), and (v[3], v[7], v[11], v[15]).
76 	// a += b + m[blake2b_sigma[r][2*i + 0]];
84 	// d = ror64(d ^ a, 32);
103 	// a += b + m[blake2b_sigma[r][2*i + 1]];
118 	// d = ror64(d ^ a, 16);
133 	// This rotation amount isn't a multiple of 8, so it has to be
134 	// implemented using a pair of shifts, which requires temporary
146 	// (v[2], v[7], v[8], v[13]), and (v[3], v[4], v[9], v[14]).
154 	// a += b + m[blake2b_sigma[r][2*i + 0]];
164 	// d = ror64(d ^ a, 32);
191 	// a += b + m[blake2b_sigma[r][2*i + 1]];
204 	// d = ror64(d ^ a, 16);
249 	// Allocate a 32-byte stack buffer that is 32-byte aligned.
280 	// Therefore we store a copy of the first 32 bytes of the message block
287 	  veor		q7, q7, q15	// v[14..15] = IV[6..7] ^ f[0..1]
295 	_blake2b_round	0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15
296 	_blake2b_round	14, 10, 4, 8, 9, 15, 13, 6, 1, 12, 0, 2, 11, 7, 5, 3
297 	_blake2b_round	11, 8, 12, 0, 5, 2, 15, 13, 10, 14, 3, 6, 7, 1, 9, 4
298 	_blake2b_round	7, 9, 3, 1, 13, 12, 11, 14, 2, 6, 5, 10, 4, 0, 15, 8
299 	_blake2b_round	9, 0, 5, 7, 2, 4, 10, 15, 14, 1, 11, 12, 6, 8, 3, 13
300 	_blake2b_round	2, 12, 6, 10, 0, 11, 8, 3, 4, 13, 7, 5, 15, 14, 1, 9
301 	_blake2b_round	12, 5, 1, 15, 14, 13, 4, 10, 0, 7, 6, 3, 9, 2, 8, 11
302 	_blake2b_round	13, 11, 7, 14, 12, 1, 3, 9, 5, 0, 15, 4, 8, 6, 2, 10
303 	_blake2b_round	6, 15, 14, 9, 11, 3, 0, 8, 12, 2, 13, 7, 1, 4, 10, 5
304 	_blake2b_round	10, 2, 8, 4, 7, 6, 1, 5, 15, 11, 9, 14, 3, 12, 13, 0
305 	_blake2b_round	0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15
306 	_blake2b_round	14, 10, 4, 8, 9, 15, 13, 6, 1, 12, 0, 2, 11, 7, 5, 3 \
319 	veor		q3, q3, q7		// v[6..7] ^= v[14..15]