powerpc/crypto/crc32-vpmsum_core.S

1 /* SPDX-License-Identifier: GPL-2.0-or-later */
16  * 32 bits of 0s to the end - this matches what a CRC does. We just
28 #include <asm/ppc-opcode.h>
66 	std	r31,-8(r1)
67 	std	r30,-16(r1)
68 	std	r29,-24(r1)
69 	std	r28,-32(r1)
70 	std	r27,-40(r1)
71 	std	r26,-48(r1)
72 	std	r25,-56(r1)
101 	vspltisw v0,-1
141 	 * data - 128 / 16 = 8
148 	addi	r7,r7,-1
156 	/* zero v0-v7 which will contain our checksums */
159 	vxor	v2,v2,v2
170 	 * already in v16-v23.
194 	/* xor in initial value */
248 	 * to complete - first iteration load, second iteration vpmsum, third
249 	 * iteration xor.
268 	vxor	v2,v2,v10
321 	vxor	v2,v2,v10
349 	vxor	v2,v2,v10
365 	vsldoi	v2,v2,zeroes,4
373 	/* xor with last 1024 bits */
395 	vxor	v18,v2,v10
424 	lvx	v2,off32,r3
434 	VPMSUMW(v2,v18,v2)
441 	/* Now reduce the tail (0 - 112 bytes) */
493 	/* Now xor all the parallel chunks together */
495 	vxor	v2,v2,v3
499 	vxor	v0,v0,v2
512 	vxor	v0,v0,v1		/* xor two 64 bit results together */
531 	vxor	v0,v0,v1	/* a - qn, subtraction is xor in GF(2) */
544 	 * our vector registers goes from 0-63 instead of 63-0. We can reflect
551 	vxor	v0,v0,v1		/* a - qn, subtraction is xor in GF(2) */
580 	ld	r31,-8(r1)
581 	ld	r30,-16(r1)
582 	ld	r29,-24(r1)
583 	ld	r28,-32(r1)
584 	ld	r27,-40(r1)
585 	ld	r26,-48(r1)
586 	ld	r25,-56(r1)
625 	vxor	v0,v0,v8	/* xor in initial value */
635 	lvx	v2,off32,r4
637 	VPERM(v2,v2,v16,byteswap)
638 	VPMSUMW(v2,v2,v16)
734 .Lv2:	vxor	v20,v20,v2