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 * Copyright 2007 Sun Microsystems, Inc. All rights reserved. 23 * Use is subject to license terms. 24 */ 25 26 #pragma ident "@(#)kiconv.c 1.1 07/10/08 SMI" 27 28 /* 29 * Kernel iconv code conversion functions (PSARC/2007/173). 30 * 31 * Man pages: kiconv_open(9F), kiconv(9F), kiconv_close(9F), and kiconvstr(9F). 32 * Interface stability: Committed. 33 */ 34 35 #include <sys/types.h> 36 #include <sys/param.h> 37 #include <sys/sysmacros.h> 38 #include <sys/systm.h> 39 #include <sys/debug.h> 40 #include <sys/kmem.h> 41 #include <sys/sunddi.h> 42 #include <sys/ksynch.h> 43 #include <sys/modctl.h> 44 #include <sys/byteorder.h> 45 #include <sys/errno.h> 46 #include <sys/kiconv.h> 47 #include <sys/kiconv_latin1.h> 48 49 50 /* 51 * The following macros indicate ids to the correct code conversion mapping 52 * data tables to use. The actual tables are coming from <sys/kiconv_latin1.h>. 53 */ 54 #define KICONV_TBLID_1252 (0x00) 55 #define KICONV_TBLID_8859_1 (0x01) 56 #define KICONV_TBLID_8859_15 (0x02) 57 #define KICONV_TBLID_850 (0x03) 58 59 #define KICONV_MAX_MAPPING_TBLID (0x03) 60 61 /* 62 * The following tables are coming from u8_textprep.c. We use them to 63 * check on validity of UTF-8 characters and their bytes. 64 */ 65 extern const int8_t u8_number_of_bytes[]; 66 extern const uint8_t u8_valid_min_2nd_byte[]; 67 extern const uint8_t u8_valid_max_2nd_byte[]; 68 69 70 /* 71 * The following four functions, open_to_1252(), open_to_88591(), 72 * open_to_885915(), and open_to_850(), are kiconv_open functions from 73 * UTF-8 to corresponding single byte codesets. 74 */ 75 static void * 76 open_to_1252() 77 { 78 kiconv_state_t s; 79 80 s = (kiconv_state_t)kmem_alloc(sizeof (kiconv_state_data_t), KM_SLEEP); 81 s->id = KICONV_TBLID_1252; 82 s->bom_processed = 0; 83 84 return ((void *)s); 85 } 86 87 static void * 88 open_to_88591() 89 { 90 kiconv_state_t s; 91 92 s = (kiconv_state_t)kmem_alloc(sizeof (kiconv_state_data_t), KM_SLEEP); 93 s->id = KICONV_TBLID_8859_1; 94 s->bom_processed = 0; 95 96 return ((void *)s); 97 } 98 99 static void * 100 open_to_885915() 101 { 102 kiconv_state_t s; 103 104 s = (kiconv_state_t)kmem_alloc(sizeof (kiconv_state_data_t), KM_SLEEP); 105 s->id = KICONV_TBLID_8859_15; 106 s->bom_processed = 0; 107 108 return ((void *)s); 109 } 110 111 static void * 112 open_to_850() 113 { 114 kiconv_state_t s; 115 116 s = (kiconv_state_t)kmem_alloc(sizeof (kiconv_state_data_t), KM_SLEEP); 117 s->id = KICONV_TBLID_850; 118 s->bom_processed = 0; 119 120 return ((void *)s); 121 } 122 123 /* 124 * The following four functions, open_fr_1252(), open_fr_88591(), 125 * open_fr_885915(), and open_fr_850(), are kiconv_open functions from 126 * corresponding single byte codesets to UTF-8. 127 */ 128 static void * 129 open_fr_1252() 130 { 131 return ((void *)KICONV_TBLID_1252); 132 } 133 134 static void * 135 open_fr_88591() 136 { 137 return ((void *)KICONV_TBLID_8859_1); 138 } 139 140 static void * 141 open_fr_885915() 142 { 143 return ((void *)KICONV_TBLID_8859_15); 144 } 145 146 static void * 147 open_fr_850() 148 { 149 return ((void *)KICONV_TBLID_850); 150 } 151 152 /* 153 * The following close_to_sb() function is kiconv_close function for 154 * the conversions from UTF-8 to single byte codesets. The close_fr_sb() 155 * is kiconv_close function for the conversions from single byte codesets to 156 * UTF-8. 157 */ 158 static int 159 close_to_sb(void *s) 160 { 161 if (! s || s == (void *)-1) 162 return (EBADF); 163 164 kmem_free(s, sizeof (kiconv_state_data_t)); 165 166 return (0); 167 } 168 169 static int 170 close_fr_sb(void *s) 171 { 172 if ((ulong_t)s > KICONV_MAX_MAPPING_TBLID) 173 return (EBADF); 174 175 return (0); 176 } 177 178 /* 179 * The following is the common kiconv function for conversions from UTF-8 180 * to single byte codesets. 181 */ 182 static size_t 183 kiconv_to_sb(void *kcd, char **inbuf, size_t *inbytesleft, char **outbuf, 184 size_t *outbytesleft, int *errno) 185 { 186 size_t id; 187 size_t ret_val; 188 uchar_t *ib; 189 uchar_t *oldib; 190 uchar_t *ob; 191 uchar_t *ibtail; 192 uchar_t *obtail; 193 uint32_t u8; 194 size_t i; 195 size_t l; 196 size_t h; 197 size_t init_h; 198 int8_t sz; 199 boolean_t second; 200 201 /* Check on the kiconv code conversion descriptor. */ 202 if (! kcd || kcd == (void *)-1) { 203 *errno = EBADF; 204 return ((size_t)-1); 205 } 206 207 /* 208 * Get the table id we are going to use for the code conversion 209 * and let's double check on it. 210 */ 211 id = ((kiconv_state_t)kcd)->id; 212 if (id > KICONV_MAX_MAPPING_TBLID) { 213 *errno = EBADF; 214 return ((size_t)-1); 215 } 216 217 /* If this is a state reset request, process and return. */ 218 if (! inbuf || ! (*inbuf)) { 219 ((kiconv_state_t)kcd)->bom_processed = 0; 220 return ((size_t)0); 221 } 222 223 ret_val = 0; 224 ib = (uchar_t *)*inbuf; 225 ob = (uchar_t *)*outbuf; 226 ibtail = ib + *inbytesleft; 227 obtail = ob + *outbytesleft; 228 229 /* 230 * The inital high value for the binary search we will be using 231 * shortly is a literal constant as of today but to be future proof, 232 * let's calculate it like the following at here. 233 */ 234 init_h = sizeof (to_sb_tbl[id]) / sizeof (kiconv_to_sb_tbl_comp_t) - 1; 235 236 /* 237 * If we haven't checked on the UTF-8 signature BOM character in 238 * the beginning of the conversion data stream, we check it and if 239 * find one, we skip it since we have no use for it. 240 */ 241 if (((kiconv_state_t)kcd)->bom_processed == 0 && (ibtail - ib) >= 3 && 242 *ib == 0xef && *(ib + 1) == 0xbb && *(ib + 2) == 0xbf) 243 ib += 3; 244 ((kiconv_state_t)kcd)->bom_processed = 1; 245 246 while (ib < ibtail) { 247 sz = u8_number_of_bytes[*ib]; 248 if (sz <= 0) { 249 *errno = EILSEQ; 250 ret_val = (size_t)-1; 251 break; 252 } 253 254 /* 255 * If there is no room to write at the output buffer, 256 * issue E2BIG error. 257 */ 258 if (ob >= obtail) { 259 *errno = E2BIG; 260 ret_val = (size_t)-1; 261 break; 262 } 263 264 /* 265 * If it is a 7-bit ASCII character, we don't need to 266 * process further and we just copy the character over. 267 * 268 * If not, we collect the character bytes up to four bytes, 269 * validate the bytes, and binary search for the corresponding 270 * single byte codeset character byte. If we find it from 271 * the mapping table, we put that into the output buffer; 272 * otherwise, we put a replacement character instead as 273 * a non-identical conversion. 274 */ 275 if (sz == 1) { 276 *ob++ = *ib++; 277 continue; 278 } 279 280 /* 281 * Issue EINVAL error if input buffer has an incomplete 282 * character at the end of the buffer. 283 */ 284 if ((ibtail - ib) < sz) { 285 *errno = EINVAL; 286 ret_val = (size_t)-1; 287 break; 288 } 289 290 /* 291 * We collect UTF-8 character bytes and also check if 292 * this is a valid UTF-8 character without any bogus bytes 293 * based on the latest UTF-8 binary representation. 294 */ 295 oldib = ib; 296 u8 = *ib++; 297 second = B_TRUE; 298 for (i = 1; i < sz; i++) { 299 if (second) { 300 if (*ib < u8_valid_min_2nd_byte[u8] || 301 *ib > u8_valid_max_2nd_byte[u8]) { 302 *errno = EILSEQ; 303 ret_val = (size_t)-1; 304 ib = oldib; 305 goto TO_SB_ILLEGAL_CHAR_ERR; 306 } 307 second = B_FALSE; 308 } else if (*ib < 0x80 || *ib > 0xbf) { 309 *errno = EILSEQ; 310 ret_val = (size_t)-1; 311 ib = oldib; 312 goto TO_SB_ILLEGAL_CHAR_ERR; 313 } 314 u8 = (u8 << 8) | ((uint32_t)*ib); 315 ib++; 316 } 317 318 i = l = 0; 319 h = init_h; 320 while (l <= h) { 321 i = (l + h) / 2; 322 if (to_sb_tbl[id][i].u8 == u8) 323 break; 324 else if (to_sb_tbl[id][i].u8 < u8) 325 l = i + 1; 326 else 327 h = i - 1; 328 } 329 330 if (to_sb_tbl[id][i].u8 == u8) { 331 *ob++ = to_sb_tbl[id][i].sb; 332 } else { 333 /* 334 * If we don't find a character in the target 335 * codeset, we insert an ASCII replacement character 336 * at the output buffer and indicate such 337 * "non-identical" conversion by increasing the 338 * return value which is the non-identical conversion 339 * counter if bigger than 0. 340 */ 341 *ob++ = KICONV_ASCII_REPLACEMENT_CHAR; 342 ret_val++; 343 } 344 } 345 346 TO_SB_ILLEGAL_CHAR_ERR: 347 *inbuf = (char *)ib; 348 *inbytesleft = ibtail - ib; 349 *outbuf = (char *)ob; 350 *outbytesleft = obtail - ob; 351 352 return (ret_val); 353 } 354 355 /* 356 * The following is the common kiconv function from single byte codesets to 357 * UTF-8. 358 */ 359 static size_t 360 kiconv_fr_sb(void *kcd, char **inbuf, size_t *inbytesleft, char **outbuf, 361 size_t *outbytesleft, int *errno) 362 { 363 size_t ret_val; 364 uchar_t *ib; 365 uchar_t *ob; 366 uchar_t *ibtail; 367 uchar_t *obtail; 368 size_t i; 369 size_t k; 370 int8_t sz; 371 372 /* Check on the kiconv code conversion descriptor validity. */ 373 if ((ulong_t)kcd > KICONV_MAX_MAPPING_TBLID) { 374 *errno = EBADF; 375 return ((size_t)-1); 376 } 377 378 /* 379 * If this is a state reset request, there is nothing to do and so 380 * we just return. 381 */ 382 if (! inbuf || ! (*inbuf)) 383 return ((size_t)0); 384 385 ret_val = 0; 386 ib = (uchar_t *)*inbuf; 387 ob = (uchar_t *)*outbuf; 388 ibtail = ib + *inbytesleft; 389 obtail = ob + *outbytesleft; 390 391 while (ib < ibtail) { 392 /* 393 * If this is a 7-bit ASCII character, we just copy over and 394 * that's all we need to do for this character. 395 */ 396 if (*ib < 0x80) { 397 if (ob >= obtail) { 398 *errno = E2BIG; 399 ret_val = (size_t)-1; 400 break; 401 } 402 403 *ob++ = *ib++; 404 continue; 405 } 406 407 /* 408 * Otherwise, we get the corresponding UTF-8 character bytes 409 * from the mapping table and copy them over. 410 * 411 * We don't need to worry about if the UTF-8 character bytes 412 * at the mapping tables are valid or not since they are good. 413 */ 414 k = *ib - 0x80; 415 sz = u8_number_of_bytes[to_u8_tbl[(ulong_t)kcd][k].u8[0]]; 416 417 /* 418 * If sz <= 0, that means we don't have any assigned character 419 * at the code point, k + 0x80, of the single byte codeset 420 * which is the fromcode. In other words, the input buffer 421 * has an illegal character. 422 */ 423 if (sz <= 0) { 424 *errno = EILSEQ; 425 ret_val = (size_t)-1; 426 break; 427 } 428 429 if ((obtail - ob) < sz) { 430 *errno = E2BIG; 431 ret_val = (size_t)-1; 432 break; 433 } 434 435 for (i = 0; i < sz; i++) 436 *ob++ = to_u8_tbl[(ulong_t)kcd][k].u8[i]; 437 438 ib++; 439 } 440 441 *inbuf = (char *)ib; 442 *inbytesleft = ibtail - ib; 443 *outbuf = (char *)ob; 444 *outbytesleft = obtail - ob; 445 446 return (ret_val); 447 } 448 449 /* 450 * The following is the common kiconvstr function from UTF-8 to single byte 451 * codesets. 452 */ 453 static size_t 454 kiconvstr_to_sb(size_t id, uchar_t *ib, size_t *inlen, uchar_t *ob, 455 size_t *outlen, int flag, int *errno) 456 { 457 size_t ret_val; 458 uchar_t *oldib; 459 uchar_t *ibtail; 460 uchar_t *obtail; 461 uint32_t u8; 462 size_t i; 463 size_t l; 464 size_t h; 465 size_t init_h; 466 int8_t sz; 467 boolean_t second; 468 boolean_t do_not_ignore_null; 469 470 /* Let's make sure that the table id is within the valid boundary. */ 471 if (id > KICONV_MAX_MAPPING_TBLID) { 472 *errno = EBADF; 473 return ((size_t)-1); 474 } 475 476 ret_val = 0; 477 ibtail = ib + *inlen; 478 obtail = ob + *outlen; 479 do_not_ignore_null = ((flag & KICONV_IGNORE_NULL) == 0); 480 init_h = sizeof (to_sb_tbl[id]) / sizeof (kiconv_to_sb_tbl_comp_t) - 1; 481 482 /* Skip any UTF-8 signature BOM character in the beginning. */ 483 if ((ibtail - ib) >= 3 && *ib == 0xef && *(ib + 1) == 0xbb && 484 *(ib + 2) == 0xbf) 485 ib += 3; 486 487 /* 488 * Basically this is pretty much the same as kiconv_to_sb() except 489 * that we are now accepting two flag values and doing the processing 490 * accordingly. 491 */ 492 while (ib < ibtail) { 493 sz = u8_number_of_bytes[*ib]; 494 if (sz <= 0) { 495 if (flag & KICONV_REPLACE_INVALID) { 496 if (ob >= obtail) { 497 *errno = E2BIG; 498 ret_val = (size_t)-1; 499 break; 500 } 501 502 ib++; 503 goto STR_TO_SB_REPLACE_INVALID; 504 } 505 506 *errno = EILSEQ; 507 ret_val = (size_t)-1; 508 break; 509 } 510 511 if (*ib == '\0' && do_not_ignore_null) 512 break; 513 514 if (ob >= obtail) { 515 *errno = E2BIG; 516 ret_val = (size_t)-1; 517 break; 518 } 519 520 if (sz == 1) { 521 *ob++ = *ib++; 522 continue; 523 } 524 525 if ((ibtail - ib) < sz) { 526 if (flag & KICONV_REPLACE_INVALID) { 527 ib = ibtail; 528 goto STR_TO_SB_REPLACE_INVALID; 529 } 530 531 *errno = EINVAL; 532 ret_val = (size_t)-1; 533 break; 534 } 535 536 oldib = ib; 537 u8 = *ib++; 538 second = B_TRUE; 539 for (i = 1; i < sz; i++) { 540 if (second) { 541 if (*ib < u8_valid_min_2nd_byte[u8] || 542 *ib > u8_valid_max_2nd_byte[u8]) { 543 if (flag & KICONV_REPLACE_INVALID) { 544 ib = oldib + sz; 545 goto STR_TO_SB_REPLACE_INVALID; 546 } 547 548 *errno = EILSEQ; 549 ret_val = (size_t)-1; 550 ib = oldib; 551 goto STR_TO_SB_ILLEGAL_CHAR_ERR; 552 } 553 second = B_FALSE; 554 } else if (*ib < 0x80 || *ib > 0xbf) { 555 if (flag & KICONV_REPLACE_INVALID) { 556 ib = oldib + sz; 557 goto STR_TO_SB_REPLACE_INVALID; 558 } 559 560 *errno = EILSEQ; 561 ret_val = (size_t)-1; 562 ib = oldib; 563 goto STR_TO_SB_ILLEGAL_CHAR_ERR; 564 } 565 u8 = (u8 << 8) | ((uint32_t)*ib); 566 ib++; 567 } 568 569 i = l = 0; 570 h = init_h; 571 while (l <= h) { 572 i = (l + h) / 2; 573 if (to_sb_tbl[id][i].u8 == u8) 574 break; 575 else if (to_sb_tbl[id][i].u8 < u8) 576 l = i + 1; 577 else 578 h = i - 1; 579 } 580 581 if (to_sb_tbl[id][i].u8 == u8) { 582 *ob++ = to_sb_tbl[id][i].sb; 583 } else { 584 STR_TO_SB_REPLACE_INVALID: 585 *ob++ = KICONV_ASCII_REPLACEMENT_CHAR; 586 ret_val++; 587 } 588 } 589 590 STR_TO_SB_ILLEGAL_CHAR_ERR: 591 *inlen = ibtail - ib; 592 *outlen = obtail - ob; 593 594 return (ret_val); 595 } 596 597 /* 598 * The following four functions are entry points recorded at the conv_list[] 599 * defined at below. 600 */ 601 static size_t 602 kiconvstr_to_1252(char *inarray, size_t *inlen, char *outarray, 603 size_t *outlen, int flag, int *errno) 604 { 605 return (kiconvstr_to_sb(KICONV_TBLID_1252, (uchar_t *)inarray, 606 inlen, (uchar_t *)outarray, outlen, flag, errno)); 607 } 608 609 static size_t 610 kiconvstr_to_1(char *inarray, size_t *inlen, char *outarray, 611 size_t *outlen, int flag, int *errno) 612 { 613 return (kiconvstr_to_sb(KICONV_TBLID_8859_1, (uchar_t *)inarray, 614 inlen, (uchar_t *)outarray, outlen, flag, errno)); 615 } 616 617 static size_t 618 kiconvstr_to_15(char *inarray, size_t *inlen, char *outarray, 619 size_t *outlen, int flag, int *errno) 620 { 621 return (kiconvstr_to_sb(KICONV_TBLID_8859_15, (uchar_t *)inarray, 622 inlen, (uchar_t *)outarray, outlen, flag, errno)); 623 } 624 625 static size_t 626 kiconvstr_to_850(char *inarray, size_t *inlen, char *outarray, 627 size_t *outlen, int flag, int *errno) 628 { 629 return (kiconvstr_to_sb(KICONV_TBLID_850, (uchar_t *)inarray, 630 inlen, (uchar_t *)outarray, outlen, flag, errno)); 631 } 632 633 /* 634 * The following is the common kiconvstr function for conversions from 635 * single byte codesets to UTF-8. 636 */ 637 static size_t 638 kiconvstr_fr_sb(size_t id, uchar_t *ib, size_t *inlen, uchar_t *ob, 639 size_t *outlen, int flag, int *errno) 640 { 641 size_t ret_val; 642 uchar_t *ibtail; 643 uchar_t *obtail; 644 size_t i; 645 size_t k; 646 int8_t sz; 647 boolean_t do_not_ignore_null; 648 649 ret_val = 0; 650 ibtail = ib + *inlen; 651 obtail = ob + *outlen; 652 do_not_ignore_null = ((flag & KICONV_IGNORE_NULL) == 0); 653 654 while (ib < ibtail) { 655 if (*ib == '\0' && do_not_ignore_null) 656 break; 657 658 if (*ib < 0x80) { 659 if (ob >= obtail) { 660 *errno = E2BIG; 661 ret_val = (size_t)-1; 662 break; 663 } 664 *ob++ = *ib++; 665 continue; 666 } 667 668 k = *ib - 0x80; 669 sz = u8_number_of_bytes[to_u8_tbl[id][k].u8[0]]; 670 671 if (sz <= 0) { 672 if (flag & KICONV_REPLACE_INVALID) { 673 if ((obtail - ob) < 3) { 674 *errno = E2BIG; 675 ret_val = (size_t)-1; 676 break; 677 } 678 679 /* Save KICONV_UTF8_REPLACEMENT_CHAR. */ 680 *ob++ = 0xef; 681 *ob++ = 0xbf; 682 *ob++ = 0xbd; 683 ret_val++; 684 ib++; 685 686 continue; 687 } 688 689 *errno = EILSEQ; 690 ret_val = (size_t)-1; 691 break; 692 } 693 694 if ((obtail - ob) < sz) { 695 *errno = E2BIG; 696 ret_val = (size_t)-1; 697 break; 698 } 699 700 for (i = 0; i < sz; i++) 701 *ob++ = to_u8_tbl[id][k].u8[i]; 702 703 ib++; 704 } 705 706 *inlen = ibtail - ib; 707 *outlen = obtail - ob; 708 709 return (ret_val); 710 } 711 712 /* 713 * The following four functions are also entry points recorded at 714 * the conv_list[] at below. 715 */ 716 static size_t 717 kiconvstr_fr_1252(char *inarray, size_t *inlen, char *outarray, 718 size_t *outlen, int flag, int *errno) 719 { 720 return (kiconvstr_fr_sb(KICONV_TBLID_1252, (uchar_t *)inarray, 721 inlen, (uchar_t *)outarray, outlen, flag, errno)); 722 } 723 724 static size_t 725 kiconvstr_fr_1(char *inarray, size_t *inlen, char *outarray, 726 size_t *outlen, int flag, int *errno) 727 { 728 return (kiconvstr_fr_sb(KICONV_TBLID_8859_1, (uchar_t *)inarray, 729 inlen, (uchar_t *)outarray, outlen, flag, errno)); 730 } 731 732 static size_t 733 kiconvstr_fr_15(char *inarray, size_t *inlen, char *outarray, 734 size_t *outlen, int flag, int *errno) 735 { 736 return (kiconvstr_fr_sb(KICONV_TBLID_8859_15, (uchar_t *)inarray, 737 inlen, (uchar_t *)outarray, outlen, flag, errno)); 738 } 739 740 static size_t 741 kiconvstr_fr_850(char *inarray, size_t *inlen, char *outarray, 742 size_t *outlen, int flag, int *errno) 743 { 744 return (kiconvstr_fr_sb(KICONV_TBLID_850, (uchar_t *)inarray, 745 inlen, (uchar_t *)outarray, outlen, flag, errno)); 746 } 747 748 /* 749 * The following static vector contains the normalized code names 750 * and their corresponding code ids. They are somewhat arbitrarily ordered 751 * based on marketing data available. A code id could repeat for aliases. 752 * 753 * The vector was generated by using a small utility program called 754 * codeidlistgen.c that you can find from PSARC/2007/173/materials/util/. 755 * 756 * The code ids must be portable, i.e., if needed, you can always generate 757 * the code_list[] again with different code ids. You'll also need to 758 * update the conv_list[] at below. 759 */ 760 #define KICONV_MAX_CODEID_ENTRY 68 761 #define KICONV_MAX_CODEID 42 762 763 static kiconv_code_list_t code_list[KICONV_MAX_CODEID_ENTRY] = { 764 { "utf8", 0 }, 765 { "cp1252", 1 }, 766 { "1252", 1 }, 767 { "iso88591", 2 }, 768 { "iso885915", 3 }, 769 { "cp850", 4 }, 770 { "850", 4 }, 771 { "eucjp", 5 }, 772 { "eucjpms", 6 }, 773 { "cp932", 7 }, 774 { "932", 7 }, 775 { "shiftjis", 8 }, 776 { "pck", 8 }, 777 { "sjis", 8 }, 778 { "gb18030", 9 }, 779 { "gbk", 10 }, 780 { "cp936", 10 }, 781 { "936", 10 }, 782 { "euccn", 11 }, 783 { "euckr", 12 }, 784 { "unifiedhangul", 13 }, 785 { "cp949", 13 }, 786 { "949", 13 }, 787 { "big5", 14 }, 788 { "cp950", 14 }, 789 { "950", 14 }, 790 { "big5hkscs", 15 }, 791 { "euctw", 16 }, 792 { "cp950hkscs", 17 }, 793 { "cp1250", 18 }, 794 { "1250", 18 }, 795 { "iso88592", 19 }, 796 { "cp852", 20 }, 797 { "852", 20 }, 798 { "cp1251", 21 }, 799 { "1251", 21 }, 800 { "iso88595", 22 }, 801 { "koi8r", 23 }, 802 { "cp866", 24 }, 803 { "866", 24 }, 804 { "cp1253", 25 }, 805 { "1253", 25 }, 806 { "iso88597", 26 }, 807 { "cp737", 27 }, 808 { "737", 27 }, 809 { "cp1254", 28 }, 810 { "1254", 28 }, 811 { "iso88599", 29 }, 812 { "cp857", 30 }, 813 { "857", 30 }, 814 { "cp1256", 31 }, 815 { "1256", 31 }, 816 { "iso88596", 32 }, 817 { "cp720", 33 }, 818 { "720", 33 }, 819 { "cp1255", 34 }, 820 { "1255", 34 }, 821 { "iso88598", 35 }, 822 { "cp862", 36 }, 823 { "862", 36 }, 824 { "cp1257", 37 }, 825 { "1257", 37 }, 826 { "iso885913", 38 }, 827 { "iso885910", 39 }, 828 { "iso885911", 40 }, 829 { "tis620", 40 }, 830 { "iso88593", 41 }, 831 { "iso88594", 42 }, 832 }; 833 834 /* 835 * The list of code conversions supported are grouped together per 836 * module which will be loaded as needed. 837 */ 838 #define KICONV_MAX_CONVERSIONS 84 839 840 static kiconv_conv_list_t conv_list[KICONV_MAX_CONVERSIONS] = { 841 /* Embedded code conversions: */ 842 { 843 1, 0, KICONV_EMBEDDED, 844 open_to_1252, kiconv_to_sb, close_to_sb, kiconvstr_to_1252 845 }, 846 { 847 0, 1, KICONV_EMBEDDED, 848 open_fr_1252, kiconv_fr_sb, close_fr_sb, kiconvstr_fr_1252 849 }, 850 { 851 2, 0, KICONV_EMBEDDED, 852 open_to_88591, kiconv_to_sb, close_to_sb, kiconvstr_to_1 853 }, 854 { 855 0, 2, KICONV_EMBEDDED, 856 open_fr_88591, kiconv_fr_sb, close_fr_sb, kiconvstr_fr_1 857 }, 858 { 859 3, 0, KICONV_EMBEDDED, 860 open_to_885915, kiconv_to_sb, close_to_sb, kiconvstr_to_15 861 }, 862 { 863 0, 3, KICONV_EMBEDDED, 864 open_fr_885915, kiconv_fr_sb, close_fr_sb, kiconvstr_fr_15 865 }, 866 { 867 4, 0, KICONV_EMBEDDED, 868 open_to_850, kiconv_to_sb, close_to_sb, kiconvstr_to_850 869 }, 870 { 871 0, 4, KICONV_EMBEDDED, 872 open_fr_850, kiconv_fr_sb, close_fr_sb, kiconvstr_fr_850 873 }, 874 875 /* kiconv_ja module conversions: */ 876 { 0, 5, KICONV_MODULE_ID_JA, NULL, NULL, NULL, NULL }, 877 { 5, 0, KICONV_MODULE_ID_JA, NULL, NULL, NULL, NULL }, 878 { 0, 6, KICONV_MODULE_ID_JA, NULL, NULL, NULL, NULL }, 879 { 6, 0, KICONV_MODULE_ID_JA, NULL, NULL, NULL, NULL }, 880 { 0, 7, KICONV_MODULE_ID_JA, NULL, NULL, NULL, NULL }, 881 { 7, 0, KICONV_MODULE_ID_JA, NULL, NULL, NULL, NULL }, 882 { 0, 8, KICONV_MODULE_ID_JA, NULL, NULL, NULL, NULL }, 883 { 8, 0, KICONV_MODULE_ID_JA, NULL, NULL, NULL, NULL }, 884 885 /* kiconv_sc module conversions: */ 886 { 0, 9, KICONV_MODULE_ID_SC, NULL, NULL, NULL, NULL }, 887 { 9, 0, KICONV_MODULE_ID_SC, NULL, NULL, NULL, NULL }, 888 { 0, 10, KICONV_MODULE_ID_SC, NULL, NULL, NULL, NULL }, 889 { 10, 0, KICONV_MODULE_ID_SC, NULL, NULL, NULL, NULL }, 890 { 0, 11, KICONV_MODULE_ID_SC, NULL, NULL, NULL, NULL }, 891 { 11, 0, KICONV_MODULE_ID_SC, NULL, NULL, NULL, NULL }, 892 893 /* kiconv_ko module conversions: */ 894 { 0, 12, KICONV_MODULE_ID_KO, NULL, NULL, NULL, NULL }, 895 { 12, 0, KICONV_MODULE_ID_KO, NULL, NULL, NULL, NULL }, 896 { 0, 13, KICONV_MODULE_ID_KO, NULL, NULL, NULL, NULL }, 897 { 13, 0, KICONV_MODULE_ID_KO, NULL, NULL, NULL, NULL }, 898 899 /* kiconv_tc module conversions: */ 900 { 0, 14, KICONV_MODULE_ID_TC, NULL, NULL, NULL, NULL }, 901 { 14, 0, KICONV_MODULE_ID_TC, NULL, NULL, NULL, NULL }, 902 { 0, 15, KICONV_MODULE_ID_TC, NULL, NULL, NULL, NULL }, 903 { 15, 0, KICONV_MODULE_ID_TC, NULL, NULL, NULL, NULL }, 904 { 0, 16, KICONV_MODULE_ID_TC, NULL, NULL, NULL, NULL }, 905 { 16, 0, KICONV_MODULE_ID_TC, NULL, NULL, NULL, NULL }, 906 { 0, 17, KICONV_MODULE_ID_TC, NULL, NULL, NULL, NULL }, 907 { 17, 0, KICONV_MODULE_ID_TC, NULL, NULL, NULL, NULL }, 908 909 /* kiconv_emea module conversions: */ 910 { 0, 18, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 911 { 18, 0, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 912 { 0, 19, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 913 { 19, 0, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 914 { 0, 20, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 915 { 20, 0, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 916 { 0, 21, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 917 { 21, 0, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 918 { 0, 22, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 919 { 22, 0, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 920 { 0, 23, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 921 { 23, 0, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 922 { 0, 24, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 923 { 24, 0, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 924 { 0, 25, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 925 { 25, 0, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 926 { 0, 26, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 927 { 26, 0, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 928 { 0, 27, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 929 { 27, 0, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 930 { 0, 28, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 931 { 28, 0, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 932 { 0, 29, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 933 { 29, 0, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 934 { 0, 30, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 935 { 30, 0, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 936 { 0, 31, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 937 { 31, 0, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 938 { 0, 32, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 939 { 32, 0, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 940 { 0, 33, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 941 { 33, 0, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 942 { 0, 34, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 943 { 34, 0, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 944 { 0, 35, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 945 { 35, 0, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 946 { 0, 36, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 947 { 36, 0, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 948 { 0, 37, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 949 { 37, 0, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 950 { 0, 38, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 951 { 38, 0, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 952 { 0, 39, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 953 { 39, 0, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 954 { 0, 40, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 955 { 40, 0, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 956 { 0, 41, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 957 { 41, 0, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 958 { 0, 42, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 959 { 42, 0, KICONV_MODULE_ID_EMEA, NULL, NULL, NULL, NULL }, 960 }; 961 962 /* The list of implemeted and supported modules. */ 963 static kiconv_mod_list_t module_list[KICONV_MAX_MODULE_ID + 1] = { 964 "kiconv_embedded", 0, 965 "kiconv_ja", 0, 966 "kiconv_sc", 0, 967 "kiconv_ko", 0, 968 "kiconv_tc", 0, 969 "kiconv_emea", 0, 970 }; 971 972 /* 973 * We use conv_list_lock to restrict data access of both conv_list[] and 974 * module_list[] as they are tightly coupled critical sections that need to be 975 * dealt together as a unit. 976 */ 977 static kmutex_t conv_list_lock; 978 979 void 980 kiconv_init() 981 { 982 mutex_init(&conv_list_lock, NULL, MUTEX_DEFAULT, NULL); 983 } 984 985 /* 986 * The following is used to check on whether a kiconv module is being 987 * used or not at the _fini() of the module. 988 */ 989 size_t 990 kiconv_module_ref_count(size_t mid) 991 { 992 int count; 993 994 if (mid <= 0 || mid > KICONV_MAX_MODULE_ID) 995 return (0); 996 997 mutex_enter(&conv_list_lock); 998 999 count = module_list[mid].refcount; 1000 1001 mutex_exit(&conv_list_lock); 1002 1003 return (count); 1004 } 1005 1006 /* 1007 * This function "normalizes" a given code name, n, by not including skippable 1008 * characters and folding uppercase letters to corresponding lowercase letters. 1009 * We only fold 7-bit ASCII uppercase characters since the names should be in 1010 * Portable Character Set of 7-bit ASCII. 1011 * 1012 * By doing this, we will be able to maximize the code name matches. 1013 */ 1014 static size_t 1015 normalize_codename(const char *n) 1016 { 1017 char s[KICONV_MAX_CODENAME_LEN + 1]; 1018 size_t i; 1019 1020 if (n == NULL) 1021 return ((size_t)-1); 1022 1023 for (i = 0; *n; n++) { 1024 if (KICONV_SKIPPABLE_CHAR(*n)) 1025 continue; 1026 1027 /* If unreasonably lengthy, we don't support such names. */ 1028 if (i >= KICONV_MAX_CODENAME_LEN) 1029 return ((size_t)-1); 1030 1031 s[i++] = (*n >= 'A' && *n <= 'Z') ? *n - 'A' + 'a' : *n; 1032 } 1033 s[i] = '\0'; 1034 1035 /* With the normalized name, find the corresponding codeset id. */ 1036 for (i = 0; i < KICONV_MAX_CODEID_ENTRY; i++) 1037 if (strcmp(s, code_list[i].name) == 0) 1038 return (code_list[i].id); 1039 1040 /* 1041 * In future time, we will also have a few more lines of code at below 1042 * that will deal with other user-created modules' fromcodes and 1043 * tocodes including aliases in a different vector. For now, we don't 1044 * support that but only the known names to this project at this time. 1045 */ 1046 1047 return ((size_t)-1); 1048 } 1049 1050 /* 1051 * This function called from mod_install() registers supplied code 1052 * conversions. At this point, it does not honor aliases and hence does not 1053 * use nowait data field from the kiconv module info data structure. 1054 */ 1055 int 1056 kiconv_register_module(kiconv_module_info_t *info) 1057 { 1058 size_t mid; 1059 size_t fid; 1060 size_t tid; 1061 size_t i; 1062 size_t j; 1063 kiconv_ops_t *op; 1064 1065 /* Validate the given kiconv module info. */ 1066 if (info == NULL || info->module_name == NULL || 1067 info->kiconv_num_convs == 0 || info->kiconv_ops_tbl == NULL) 1068 return (EINVAL);