1/* $NetBSD: strlen.S,v 1.1 2014/09/03 19:34:25 matt Exp $ */ 2 3/*- 4 * Copyright (C) 2001 Martin J. Laubach <mjl@NetBSD.org> 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 3. The name of the author may not be used to endorse or promote products 16 * derived from this software without specific prior written permission. 17 * 18 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 19 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 20 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 21 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 22 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 23 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 24 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 25 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 26 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE 27 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 28 */ 29/*----------------------------------------------------------------------*/ 30 31#include <machine/asm.h> 32 33__RCSID("$NetBSD: strlen.S,v 1.1 2014/09/03 19:34:25 matt Exp $"); 34 35/*----------------------------------------------------------------------*/ 36/* The algorithm here uses the following techniques: 37 38 1) Given a word 'x', we can test to see if it contains any 0 bytes 39 by subtracting 0x01010101, and seeing if any of the high bits of each 40 byte changed from 0 to 1. This works because the least significant 41 0 byte must have had no incoming carry (otherwise it's not the least 42 significant), so it is 0x00 - 0x01 == 0xff. For all other 43 byte values, either they have the high bit set initially, or when 44 1 is subtracted you get a value in the range 0x00-0x7f, none of which 45 have their high bit set. The expression here is 46 (x + 0xfefefeff) & ~(x | 0x7f7f7f7f), which gives 0x00000000 when 47 there were no 0x00 bytes in the word. 48 49 2) Given a word 'x', we can test to see _which_ byte was zero by 50 calculating ~(((x & 0x7f7f7f7f) + 0x7f7f7f7f) | x | 0x7f7f7f7f). 51 This produces 0x80 in each byte that was zero, and 0x00 in all 52 the other bytes. The '| 0x7f7f7f7f' clears the low 7 bits in each 53 byte, and the '| x' part ensures that bytes with the high bit set 54 produce 0x00. The addition will carry into the high bit of each byte 55 iff that byte had one of its low 7 bits set. We can then just see 56 which was the most significant bit set and divide by 8 to find how 57 many to add to the index. 58 This is from the book 'The PowerPC Compiler Writer's Guide', 59 by Steve Hoxey, Faraydon Karim, Bill Hay and Hank Warren. 60*/ 61/*----------------------------------------------------------------------*/ 62 63ENTRY(strlen) 64 65 l.or r12, r3, r0 /* save start */ 66 67 /* Setup constants */ 68 l.movhi r13, 0x7f7f 69 l.movhi r15, 0xfefe 70 l.ori r13, r13, 0x7f7f 71 l.ori r15, r15, 0xfeff 72 731: l.andi r7, r12, 3 /* get low bits of start */ 74 l.sfeqi r7, 0 /* all clear? */ 75 l.bf 3f /* yes, skip alignment */ 76 l.nop /* -- delay slot -- */ 77 78 l.sub r12, r12, r7 /* word align start */ 79 l.lwz r8, 0(r12) /* load data */ 80 l.addi r6, r0, -1 /* r6 = 0xffffffff */ 81 l.slli r5, r7, 3 /* bits to bytes */ 82 l.srl r6, r6, r5 /* clear low (MSB) bytes */ 83 l.xori r6, r6, -1 /* complement */ 84 l.or r8, r8, r6 /* merge with loaded word */ 85 l.j 4f /* and process */ 86 l.nop /* -- delay-slot -- */ 87 882: l.addi r12, r12, 4 /* advance to next word */ 893: l.lwz r8, 0(r12) /* fetch data word */ 90 91 // Step 1: (x + 0xfefefeff) & ~(x | 0x7f7f7f7f) 924: l.or r7, r8, r13 /* t0 = x | 0x7f7f7f7f */ 93 l.xori r6, r7, -1 /* t1 = ~t0 */ 94 l.add r5, r8, r15 /* t2 = x + 0xfefefeff */ 95 l.and r4, r7, r5 /* t3 = t1 & t2 */ 96 l.sfeqi r4, 0 97 l.bf 2b /* no NUL bytes here */ 98 l.nop /* -- delay slot -- */ 99 100 // Step 2: ~(((x & 0x7f7f7f7f) + 0x7f7f7f7f) | x | 0x7f7f7f7f) 101 l.and r7, r8, r13 /* t0 = x & 0x7f7f7f7f */ 102 l.or r6, r8, r13 /* t1 = x | 0x7f7f7f7f */ 103 l.add r5, r7, r13 /* t2 = t0 + 0x7f7f7f7f */ 104 l.or r4, r5, r6 /* t3 = t2 | t1 */ 105 l.xori r4, r4, -1 /* t3 = ~t3 */ 106 107 l.fl1 r5, r4 /* find last bit set */ 108 l.ori r6, r0, 32 /* bits per word */ 109 l.sub r7, r6, r5 /* cvt to leading zeros */ 110 l.srli r8, r7, 3 /* shift to byte count */ 111 112Ldone: 113 l.add r12, r12, r8 /* r12 contains end pointer */ 114 115 /* NOTE: Keep it so this function returns the end pointer 116 in r12, so we can it use from other str* calls (strcat 117 comes to mind */ 118 119 l.sub r11, r12, r3 /* length = end - start */ 120 l.jr lr /* return */ 121 l.nop /* -- delay slot -- */ 122END(strlen) 123/*----------------------------------------------------------------------*/ 124