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
103 	vsldoi	mask_32bit,zeroes,v0,4
104 	vsldoi	mask_64bit,zeroes,v0,8
129 1:	lis	r7,MAX_SIZE@h
143 	 * data - 128 / 16 = 8
150 	addi	r7,r7,-1
159 	/* zero v0-v7 which will contain our checksums */
160 	vxor	v0,v0,v0
173 	 * already in v16-v23.
175 	cmpdi	r0,1
251 	 * to complete - first iteration load, second iteration vpmsum, third
259 	vxor	v0,v0,v8
316 	vxor	v0,v0,v8
350 	vxor	v0,v0,v8
366 	vsldoi	v0,v0,zeroes,4
396 	vxor	v16,v0,v8
405 	li	r0,1
408 	bne	1b
425 	lvx	v0,0,r3
435 	VPMSUMW(v0,v16,v0)
444 	/* Now reduce the tail (0 - 112 bytes) */
446 	beq	1f
452 	vxor	v0,v0,v16
453 	bdz	1f
459 	vxor	v0,v0,v16
460 	bdz	1f
466 	vxor	v0,v0,v16
467 	bdz	1f
473 	vxor	v0,v0,v16
474 	bdz	1f
480 	vxor	v0,v0,v16
481 	bdz	1f
487 	vxor	v0,v0,v16
488 	bdz	1f
494 	vxor	v0,v0,v16
497 1:	vxor	v0,v0,v1
502 	vxor	v0,v0,v2
505 	vxor	v0,v0,v4
515 	vsldoi	v1,v0,v0,8
516 	vxor	v0,v0,v1		/* xor two 64 bit results together */
520 	vspltisb v1,1
521 	vsl	v0,v0,v1
524 	vand	v0,v0,mask_64bit
532 	VPMSUMD(v1,v0,const1)	/* ma */
535 	vxor	v0,v0,v1	/* a - qn, subtraction is xor in GF(2) */
539 	 * V0 [ 0 1 2 X ]
540 	 * V0 [ 0 X 2 3 ]
542 	vsldoi	v0,v0,zeroes,8	/* shift result into top 64 bits */
548 	 * our vector registers goes from 0-63 instead of 63-0. We can reflect
551 	vand	v1,v0,mask_32bit	/* bottom 32 bits of a */
555 	vxor	v0,v0,v1		/* a - qn, subtraction is xor in GF(2) */
560 	 * V0 [ 0 1 X 3 ]
561 	 * V0 [ 0 X 2 3 ]
563 	vsldoi	v0,v0,zeroes,4		/* shift result into top 64 bits of */
567 	MFVRD(R3, v0)
584 	ld	r31,-8(r1)
585 	ld	r30,-16(r1)
586 	ld	r29,-24(r1)
587 	ld	r28,-32(r1)
588 	ld	r27,-40(r1)
589 	ld	r26,-48(r1)
590 	ld	r25,-56(r1)
627 	lvx	v0,0,r4
629 	VPERM(v0,v0,v16,byteswap)
630 	vxor	v0,v0,v8	/* xor in initial value */
631 	VPMSUMW(v0,v0,v16)
741 .Lv0:	vxor	v20,v20,v0
743 	vxor	v0,v19,v20