1/* 2 * CDDL HEADER START 3 * 4 * The contents of this file are subject to the terms of the 5 * Common Development and Distribution License (the "License"). 6 * You may not use this file except in compliance with the License. 7 * 8 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE 9 * or http://www.opensolaris.org/os/licensing. 10 * See the License for the specific language governing permissions 11 * and limitations under the License. 12 * 13 * When distributing Covered Code, include this CDDL HEADER in each 14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE. 15 * If applicable, add the following below this CDDL HEADER, with the 16 * fields enclosed by brackets "[]" replaced with your own identifying 17 * information: Portions Copyright [yyyy] [name of copyright owner] 18 * 19 * CDDL HEADER END 20 */ 21 22/* 23 * Copyright 2008 Sun Microsystems, Inc. All rights reserved. 24 * Use is subject to license terms. 25 */ 26 27#pragma ident "%Z%%M% %I% %E% SMI" 28 29 .file "%M%" 30 31/* 32 * strncpy(s1, s2) 33 * 34 * Copy string s2 to s1, truncating or null-padding to always copy n bytes 35 * return s1. 36 * 37 * Fast assembler language version of the following C-program for strncpy 38 * which represents the `standard' for the C-library. 39 * 40 * char * 41 * strncpy(char *s1, const char *s2, size_t n) 42 * { 43 * char *os1 = s1; 44 * 45 * n++; 46 * while ((--n != 0) && ((*s1++ = *s2++) != '\0')) 47 * ; 48 * if (n != 0) 49 * while (--n != 0) 50 * *s1++ = '\0'; 51 * return (os1); 52 * } 53 */ 54 55#include <sys/asm_linkage.h> 56 57 ! strncpy works similarly to strcpy, except that n bytes of s2 58 ! are copied to s1. If a null character is reached in s2 yet more 59 ! bytes remain to be copied, strncpy will copy null bytes into 60 ! the destination string. 61 ! 62 ! This implementation works by first aligning the src ptr and 63 ! performing small copies until it is aligned. Then, the string 64 ! is copied based upon destination alignment. (byte, half-word, 65 ! word, etc.) 66 67 ENTRY(strncpy) 68 69 .align 32 70 subcc %g0, %o2, %o4 ! n = -n 71 bz .doneshort ! if n == 0, done 72 cmp %o2, 7 ! n < 7 ? 73 add %o1, %o2, %o3 ! src = src + n 74 blu .shortcpy ! n < 7, use byte-wise copy 75 add %o0, %o2, %o2 ! dst = dst + n 76 andcc %o1, 3, %o5 ! src word aligned ? 77 bz .wordaligned ! yup 78 save %sp, -0x40, %sp ! create new register window 79 sub %i5, 4, %i5 ! bytes until src aligned 80 nop ! align loop on 16-byte boundary 81 nop ! align loop on 16-byte boundary 82 83.alignsrc: 84 ldub [%i3 + %i4], %i1 ! src[] 85 stb %i1, [%i2 + %i4] ! dst[] = src[] 86 inccc %i4 ! src++, dst++, n-- 87 bz .done ! n == 0, done 88 tst %i1 ! end of src reached (null byte) ? 89 bz,a .bytepad ! yes, at least one byte to pad here 90 add %i2, %i4, %l0 ! need single dest pointer for fill 91 inccc %i5 ! src aligned now? 92 bnz .alignsrc ! no, copy another byte 93 .empty 94 95.wordaligned: 96 add %i2, %i4, %l0 ! dst 97 sethi %hi(0x01010101), %l1 ! Alan Mycroft's magic1 98 sub %i2, 4, %i2 ! adjust for dest pre-incr in cpy loops 99 or %l1, %lo(0x01010101),%l1! finish loading magic1 100 andcc %l0, 3, %g1 ! destination word aligned ? 101 bnz .dstnotaligned ! nope 102 sll %l1, 7, %i5 ! create Alan Mycroft's magic2 103 104.storeword: 105 lduw [%i3 + %i4], %i1 ! src dword 106 addcc %i4, 4, %i4 ! n += 4, src += 4, dst += 4 107 bcs .lastword ! if counter wraps, last word 108 andn %i5, %i1, %g1 ! ~dword & 0x80808080 109 sub %i1, %l1, %l0 ! dword - 0x01010101 110 andcc %l0, %g1, %g0 ! ((dword - 0x01010101) & ~dword & 0x80808080) 111 bz,a .storeword ! no zero byte if magic expression == 0 112 stw %i1, [%i2 + %i4] ! store word to dst (address pre-incremented) 113 114 ! n has not expired, but src is at the end. we need to push out the 115 ! remaining src bytes and then start padding with null bytes 116 117.zerobyte: 118 add %i2, %i4, %l0 ! pointer to dest string 119 srl %i1, 24, %g1 ! first byte 120 stb %g1, [%l0] ! store it 121 sub %g1, 1, %g1 ! byte == 0 ? -1 : byte - 1 122 sra %g1, 31, %g1 ! byte == 0 ? -1 : 0 123 andn %i1, %g1, %i1 ! if byte == 0, start padding with null bytes 124 srl %i1, 16, %g1 ! second byte 125 stb %g1, [%l0 + 1] ! store it 126 and %g1, 0xff, %g1 ! isolate byte 127 sub %g1, 1, %g1 ! byte == 0 ? -1 : byte - 1 128 sra %g1, 31, %g1 ! byte == 0 ? -1 : 0 129 andn %i1, %g1, %i1 ! if byte == 0, start padding with null bytes 130 srl %i1, 8, %g1 ! third byte 131 stb %g1, [%l0 + 2] ! store it 132 and %g1, 0xff, %g1 ! isolate byte 133 sub %g1, 1, %g1 ! byte == 0 ? -1 : byte - 1 134 sra %g1, 31, %g1 ! byte == 0 ? -1 : 0 135 andn %i1, %g1, %i1 ! if byte == 0, start padding with null bytes 136 stb %i1, [%l0 + 3] ! store fourth byte 137 addcc %i4, 8, %g0 ! number of pad bytes < 8 ? 138 bcs .bytepad ! yes, do simple byte wise fill 139 add %l0, 4, %l0 ! dst += 4 140 andcc %l0, 3, %l1 ! dst offset relative to word boundary 141 bz .fillaligned ! dst already word aligned 142 143 ! here there is a least one more byte to zero out: otherwise we would 144 ! have exited through label .lastword 145 146 sub %l1, 4, %l1 ! bytes to align dst to word boundary 147.makealigned: 148 stb %g0, [%l0] ! dst[] = 0 149 addcc %i4, 1, %i4 ! n-- 150 bz .done ! n == 0, we are done 151 addcc %l1, 1, %l1 ! any more byte needed to align 152 bnz .makealigned ! yup, pad another byte 153 add %l0, 1, %l0 ! dst++ 154 nop ! pad to align copy loop below 155 156 ! here we know that there at least another 4 bytes to pad, since 157 ! we don't get here unless there were >= 8 bytes to pad to begin 158 ! with, and we have padded at most 3 bytes suring dst aligning 159 160.fillaligned: 161 add %i4, 3, %i2 ! round up to next word boundary 162 and %i2, -4, %l1 ! pointer to next word boundary 163 and %i2, 4, %i2 ! word count odd ? 4 : 0 164 stw %g0, [%l0] ! store first word 165 addcc %l1, %i2, %l1 ! dword count == 1 ? 166 add %i4, %i2, %i4 ! if word count odd, n -= 4 167 bz .bytepad ! if word count == 1, pad bytes left 168 add %l0, %i2, %l0 ! bump dst if word count odd 169 170.fillword: 171 addcc %l1, 8, %l1 ! count -= 8 172 stw %g0, [%l0] ! dst[n] = 0 173 stw %g0, [%l0 + 4] ! dst[n+4] = 0 174 add %l0, 8, %l0 ! dst += 8 175 bcc .fillword ! fill words until count == 0 176 addcc %i4, 8, %i4 ! n -= 8 177 bz .done ! if n == 0, we are done 178 .empty 179 180.bytepad: 181 and %i4, 1, %i2 ! byte count odd ? 1 : 0 182 stb %g0, [%l0] ! store first byte 183 addcc %i4, %i2, %i4 ! byte count == 1 ? 184 bz .done ! yup, we are done 185 add %l0, %i2, %l0 ! bump pointer if odd 186 187.fillbyte: 188 addcc %i4, 2, %i4 ! n -= 2 189 stb %g0, [%l0] ! dst[n] = 0 190 stb %g0, [%l0 + 1] ! dst[n+1] = 0 191 bnz .fillbyte ! fill until n == 0 192 add %l0, 2, %l0 ! dst += 2 193 194.done: 195 ret ! done 196 restore %i0, %g0, %o0 ! restore reg window, return dst 197 198 ! this is the last word. It may contain null bytes. store bytes 199 ! until n == 0. if null byte encountered, continue 200 201.lastword: 202 sub %i4, 4, %i4 ! undo counter pre-increment 203 add %i2, 4, %i2 ! adjust dst for counter un-bumping 204 205 srl %i1, 24, %g1 ! first byte 206 stb %g1, [%i2 + %i4] ! store it 207 inccc %i4 ! n-- 208 bz .done ! if n == 0, we're done 209 sub %g1, 1, %g1 ! byte == 0 ? -1 : byte - 1 210 sra %g1, 31, %g1 ! byte == 0 ? -1 : 0 211 andn %i1, %g1, %i1 ! if byte == 0, start padding with null 212 srl %i1, 16, %g1 ! second byte 213 stb %g1, [%i2 + %i4] ! store it 214 inccc %i4 ! n-- 215 bz .done ! if n == 0, we're done 216 and %g1, 0xff, %g1 ! isolate byte 217 sub %g1, 1, %g1 ! byte == 0 ? -1 : byte - 1 218 sra %g1, 31, %g1 ! byte == 0 ? -1 : 0 219 andn %i1, %g1, %i1 ! if byte == 0, start padding with null 220 srl %i1, 8, %g1 ! third byte 221 stb %g1, [%i2 + %i4] ! store it 222 inccc %i4 ! n-- 223 bz .done ! if n == 0, we're done 224 and %g1, 0xff, %g1 ! isolate byte 225 sub %g1, 1, %g1 ! byte == 0 ? -1 : byte - 1 226 sra %g1, 31, %g1 ! byte == 0 ? -1 : 0 227 andn %i1, %g1, %i1 ! if byte == 0, start padding with null 228 ba .done ! here n must be zero, we are done 229 stb %i1, [%i2 + %i4] ! store fourth byte 230 231.dstnotaligned: 232 cmp %g1, 2 ! dst half word aligned? 233 be .storehalfword2 ! yup, store half word at a time 234 .empty 235.storebyte: 236 lduw [%i3 + %i4], %i1 ! x = src[] 237 addcc %i4, 4, %i4 ! src += 4, dst += 4, n -= 4 238 bcs .lastword ! if counter wraps, last word 239 andn %i5, %i1, %g1 ! ~x & 0x80808080 240 sub %i1, %l1, %l0 ! x - 0x01010101 241 andcc %l0, %g1, %g0 ! ((x - 0x01010101) & ~x & 0x80808080) 242 bnz .zerobyte ! end of src found, may need to pad 243 add %i2, %i4, %l0 ! dst (in pointer form) 244 srl %i1, 24, %g1 ! %g1<7:0> = 1st byte; half-word aligned now 245 stb %g1, [%l0] ! store first byte 246 srl %i1, 8, %g1 ! %g1<15:0> = bytes 2, 3 247 sth %g1, [%l0 + 1] ! store bytes 2, 3 248 ba .storebyte ! next word 249 stb %i1, [%l0 + 3] ! store fourth byte 250 nop 251 nop 252 253.storehalfword: 254 lduw [%i3 + %i4], %i1 ! x = src[] 255.storehalfword2: 256 addcc %i4, 4, %i4 ! src += 4, dst += 4, n -= 4 257 bcs .lastword ! if counter wraps, last word 258 andn %i5, %i1, %g1 ! ~x & 0x80808080 259 sub %i1, %l1, %l0 ! x - 0x01010101 260 andcc %l0, %g1, %g0 ! ((x -0x01010101) & ~x & 0x8080808080) 261 bnz .zerobyte ! x has zero byte, handle end cases 262 add %i2, %i4, %l0 ! dst (in pointer form) 263 srl %i1, 16, %g1 ! %g1<15:0> = bytes 1, 2 264 sth %g1, [%l0] ! store bytes 1, 2 265 ba .storehalfword ! next dword 266 sth %i1, [%l0 + 2] ! store bytes 3, 4 267 268.shortcpy: 269 ldub [%o3 + %o4], %o5 ! src[] 270 stb %o5, [%o2 + %o4] ! dst[] = src[] 271 inccc %o4 ! src++, dst++, n-- 272 bz .doneshort ! if n == 0, done 273 tst %o5 ! src[] == 0 ? 274 bnz,a .shortcpy ! nope, next byte 275 nop ! empty delay slot 276 277.padbyte: 278 stb %g0, [%o2 + %o4] ! dst[] = 0 279.padbyte2: 280 addcc %o4, 1, %o4 ! dst++, n-- 281 bnz,a .padbyte2 ! if n != 0, next byte 282 stb %g0, [%o2 + %o4] ! dst[] = 0 283 nop ! align label below to 16-byte boundary 284 285.doneshort: 286 retl ! return from leaf 287 nop ! empty delay slot 288 SET_SIZE(strncpy) 289