1/* SPDX-License-Identifier: GPL-2.0 */
2/*
3 * strlen() for PPC32
4 *
5 * Copyright (C) 2018 Christophe Leroy CS Systemes d'Information.
6 *
7 * Inspired from glibc implementation
8 */
9#include <linux/export.h>
10#include <asm/ppc_asm.h>
11#include <asm/cache.h>
12
13	.text
14
15/*
16 * Algorithm:
17 *
18 * 1) Given a word 'x', we can test to see if it contains any 0 bytes
19 *    by subtracting 0x01010101, and seeing if any of the high bits of each
20 *    byte changed from 0 to 1. This works because the least significant
21 *    0 byte must have had no incoming carry (otherwise it's not the least
22 *    significant), so it is 0x00 - 0x01 == 0xff. For all other
23 *    byte values, either they have the high bit set initially, or when
24 *    1 is subtracted you get a value in the range 0x00-0x7f, none of which
25 *    have their high bit set. The expression here is
26 *    (x - 0x01010101) & ~x & 0x80808080), which gives 0x00000000 when
27 *    there were no 0x00 bytes in the word.  You get 0x80 in bytes that
28 *    match, but possibly false 0x80 matches in the next more significant
29 *    byte to a true match due to carries.  For little-endian this is
30 *    of no consequence since the least significant match is the one
31 *    we're interested in, but big-endian needs method 2 to find which
32 *    byte matches.
33 * 2) Given a word 'x', we can test to see _which_ byte was zero by
34 *    calculating ~(((x & ~0x80808080) - 0x80808080 - 1) | x | ~0x80808080).
35 *    This produces 0x80 in each byte that was zero, and 0x00 in all
36 *    the other bytes. The '| ~0x80808080' clears the low 7 bits in each
37 *    byte, and the '| x' part ensures that bytes with the high bit set
38 *    produce 0x00. The addition will carry into the high bit of each byte
39 *    iff that byte had one of its low 7 bits set. We can then just see
40 *    which was the most significant bit set and divide by 8 to find how
41 *    many to add to the index.
42 *    This is from the book 'The PowerPC Compiler Writer's Guide',
43 *    by Steve Hoxey, Faraydon Karim, Bill Hay and Hank Warren.
44 */
45
46_GLOBAL(strlen)
47	andi.   r0, r3, 3
48	lis	r7, 0x0101
49	addi	r10, r3, -4
50	addic	r7, r7, 0x0101	/* r7 = 0x01010101 (lomagic) & clear XER[CA] */
51	rotlwi	r6, r7, 31 	/* r6 = 0x80808080 (himagic) */
52	bne-	3f
53	.balign IFETCH_ALIGN_BYTES
541:	lwzu	r9, 4(r10)
552:	subf	r8, r7, r9
56	and.	r8, r8, r6
57	beq+	1b
58	andc.	r8, r8, r9
59	beq+	1b
60	andc	r8, r9, r6
61	orc	r9, r9, r6
62	subfe	r8, r6, r8
63	nor	r8, r8, r9
64	cntlzw	r8, r8
65	subf	r3, r3, r10
66	srwi	r8, r8, 3
67	add	r3, r3, r8
68	blr
69
70	/* Missaligned string: make sure bytes before string are seen not 0 */
713:	xor	r10, r10, r0
72	orc	r8, r8, r8
73	lwzu	r9, 4(r10)
74	slwi	r0, r0, 3
75	srw	r8, r8, r0
76	orc	r9, r9, r8
77	b	2b
78EXPORT_SYMBOL(strlen)
79