1 // SPDX-License-Identifier: GPL-2.0-or-later 2 /* 3 * 4 * Copyright (c) International Business Machines Corp., 2000,2009 5 * Modified by Steve French (sfrench@us.ibm.com) 6 */ 7 #include <linux/fs.h> 8 #include <linux/slab.h> 9 #include "cifs_fs_sb.h" 10 #include "cifs_unicode.h" 11 #include "cifsglob.h" 12 #include "cifs_debug.h" 13 14 int cifs_remap(struct cifs_sb_info *cifs_sb) 15 { 16 int map_type; 17 18 if (cifs_sb->mnt_cifs_flags & CIFS_MOUNT_MAP_SFM_CHR) 19 map_type = SFM_MAP_UNI_RSVD; 20 else if (cifs_sb->mnt_cifs_flags & CIFS_MOUNT_MAP_SPECIAL_CHR) 21 map_type = SFU_MAP_UNI_RSVD; 22 else 23 map_type = NO_MAP_UNI_RSVD; 24 25 return map_type; 26 } 27 28 /* Convert character using the SFU - "Services for Unix" remapping range */ 29 static bool 30 convert_sfu_char(const __u16 src_char, char *target) 31 { 32 /* 33 * BB: Cannot handle remapping UNI_SLASH until all the calls to 34 * build_path_from_dentry are modified, as they use slash as 35 * separator. 36 */ 37 switch (src_char) { 38 case UNI_COLON: 39 *target = ':'; 40 break; 41 case UNI_ASTERISK: 42 *target = '*'; 43 break; 44 case UNI_QUESTION: 45 *target = '?'; 46 break; 47 case UNI_PIPE: 48 *target = '|'; 49 break; 50 case UNI_GRTRTHAN: 51 *target = '>'; 52 break; 53 case UNI_LESSTHAN: 54 *target = '<'; 55 break; 56 default: 57 return false; 58 } 59 return true; 60 } 61 62 /* Convert character using the SFM - "Services for Mac" remapping range */ 63 static bool 64 convert_sfm_char(const __u16 src_char, char *target) 65 { 66 if (src_char >= 0xF001 && src_char <= 0xF01F) { 67 *target = src_char - 0xF000; 68 return true; 69 } 70 switch (src_char) { 71 case SFM_COLON: 72 *target = ':'; 73 break; 74 case SFM_DOUBLEQUOTE: 75 *target = '"'; 76 break; 77 case SFM_ASTERISK: 78 *target = '*'; 79 break; 80 case SFM_QUESTION: 81 *target = '?'; 82 break; 83 case SFM_PIPE: 84 *target = '|'; 85 break; 86 case SFM_GRTRTHAN: 87 *target = '>'; 88 break; 89 case SFM_LESSTHAN: 90 *target = '<'; 91 break; 92 case SFM_SPACE: 93 *target = ' '; 94 break; 95 case SFM_PERIOD: 96 *target = '.'; 97 break; 98 default: 99 return false; 100 } 101 return true; 102 } 103 104 105 /* 106 * cifs_mapchar - convert a host-endian char to proper char in codepage 107 * @target - where converted character should be copied 108 * @src_char - 2 byte host-endian source character 109 * @cp - codepage to which character should be converted 110 * @map_type - How should the 7 NTFS/SMB reserved characters be mapped to UCS2? 111 * 112 * This function handles the conversion of a single character. It is the 113 * responsibility of the caller to ensure that the target buffer is large 114 * enough to hold the result of the conversion (at least NLS_MAX_CHARSET_SIZE). 115 */ 116 static int 117 cifs_mapchar(char *target, const __u16 *from, const struct nls_table *cp, 118 int maptype) 119 { 120 int len = 1; 121 __u16 src_char; 122 123 src_char = *from; 124 125 if ((maptype == SFM_MAP_UNI_RSVD) && convert_sfm_char(src_char, target)) 126 return len; 127 else if ((maptype == SFU_MAP_UNI_RSVD) && 128 convert_sfu_char(src_char, target)) 129 return len; 130 131 /* if character not one of seven in special remap set */ 132 len = cp->uni2char(src_char, target, NLS_MAX_CHARSET_SIZE); 133 if (len <= 0) 134 goto surrogate_pair; 135 136 return len; 137 138 surrogate_pair: 139 /* convert SURROGATE_PAIR and IVS */ 140 if (strcmp(cp->charset, "utf8")) 141 goto unknown; 142 len = utf16s_to_utf8s(from, 3, UTF16_LITTLE_ENDIAN, target, 6); 143 if (len <= 0) 144 goto unknown; 145 return len; 146 147 unknown: 148 *target = '?'; 149 len = 1; 150 return len; 151 } 152 153 /* 154 * cifs_from_utf16 - convert utf16le string to local charset 155 * @to - destination buffer 156 * @from - source buffer 157 * @tolen - destination buffer size (in bytes) 158 * @fromlen - source buffer size (in bytes) 159 * @codepage - codepage to which characters should be converted 160 * @mapchar - should characters be remapped according to the mapchars option? 161 * 162 * Convert a little-endian utf16le string (as sent by the server) to a string 163 * in the provided codepage. The tolen and fromlen parameters are to ensure 164 * that the code doesn't walk off of the end of the buffer (which is always 165 * a danger if the alignment of the source buffer is off). The destination 166 * string is always properly null terminated and fits in the destination 167 * buffer. Returns the length of the destination string in bytes (including 168 * null terminator). 169 * 170 * Note that some windows versions actually send multiword UTF-16 characters 171 * instead of straight UTF16-2. The linux nls routines however aren't able to 172 * deal with those characters properly. In the event that we get some of 173 * those characters, they won't be translated properly. 174 */ 175 int 176 cifs_from_utf16(char *to, const __le16 *from, int tolen, int fromlen, 177 const struct nls_table *codepage, int map_type) 178 { 179 int i, charlen, safelen; 180 int outlen = 0; 181 int nullsize = nls_nullsize(codepage); 182 int fromwords = fromlen / 2; 183 char tmp[NLS_MAX_CHARSET_SIZE]; 184 __u16 ftmp[3]; /* ftmp[3] = 3array x 2bytes = 6bytes UTF-16 */ 185 186 /* 187 * because the chars can be of varying widths, we need to take care 188 * not to overflow the destination buffer when we get close to the 189 * end of it. Until we get to this offset, we don't need to check 190 * for overflow however. 191 */ 192 safelen = tolen - (NLS_MAX_CHARSET_SIZE + nullsize); 193 194 for (i = 0; i < fromwords; i++) { 195 ftmp[0] = get_unaligned_le16(&from[i]); 196 if (ftmp[0] == 0) 197 break; 198 if (i + 1 < fromwords) 199 ftmp[1] = get_unaligned_le16(&from[i + 1]); 200 else 201 ftmp[1] = 0; 202 if (i + 2 < fromwords) 203 ftmp[2] = get_unaligned_le16(&from[i + 2]); 204 else 205 ftmp[2] = 0; 206 207 /* 208 * check to see if converting this character might make the 209 * conversion bleed into the null terminator 210 */ 211 if (outlen >= safelen) { 212 charlen = cifs_mapchar(tmp, ftmp, codepage, map_type); 213 if ((outlen + charlen) > (tolen - nullsize)) 214 break; 215 } 216 217 /* put converted char into 'to' buffer */ 218 charlen = cifs_mapchar(&to[outlen], ftmp, codepage, map_type); 219 outlen += charlen; 220 221 /* charlen (=bytes of UTF-8 for 1 character) 222 * 4bytes UTF-8(surrogate pair) is charlen=4 223 * (4bytes UTF-16 code) 224 * 7-8bytes UTF-8(IVS) is charlen=3+4 or 4+4 225 * (2 UTF-8 pairs divided to 2 UTF-16 pairs) */ 226 if (charlen == 4) 227 i++; 228 else if (charlen >= 5) 229 /* 5-6bytes UTF-8 */ 230 i += 2; 231 } 232 233 /* properly null-terminate string */ 234 for (i = 0; i < nullsize; i++) 235 to[outlen++] = 0; 236 237 return outlen; 238 } 239 240 /* 241 * NAME: cifs_strtoUTF16() 242 * 243 * FUNCTION: Convert character string to unicode string 244 * 245 */ 246 int 247 cifs_strtoUTF16(__le16 *to, const char *from, int len, 248 const struct nls_table *codepage) 249 { 250 int charlen; 251 int i; 252 wchar_t wchar_to; /* needed to quiet sparse */ 253 254 /* special case for utf8 to handle no plane0 chars */ 255 if (!strcmp(codepage->charset, "utf8")) { 256 /* 257 * convert utf8 -> utf16, we assume we have enough space 258 * as caller should have assumed conversion does not overflow 259 * in destination len is length in wchar_t units (16bits) 260 */ 261 i = utf8s_to_utf16s(from, len, UTF16_LITTLE_ENDIAN, 262 (wchar_t *) to, len); 263 264 /* if success terminate and exit */ 265 if (i >= 0) 266 goto success; 267 /* 268 * if fails fall back to UCS encoding as this 269 * function should not return negative values 270 * currently can fail only if source contains 271 * invalid encoded characters 272 */ 273 } 274 275 for (i = 0; len && *from; i++, from += charlen, len -= charlen) { 276 charlen = codepage->char2uni(from, len, &wchar_to); 277 if (charlen < 1) { 278 cifs_dbg(VFS, "strtoUTF16: char2uni of 0x%x returned %d\n", 279 *from, charlen); 280 /* A question mark */ 281 wchar_to = 0x003f; 282 charlen = 1; 283 } 284 put_unaligned_le16(wchar_to, &to[i]); 285 } 286 287 success: 288 put_unaligned_le16(0, &to[i]); 289 return i; 290 } 291 292 /* 293 * cifs_utf16_bytes - how long will a string be after conversion? 294 * @utf16 - pointer to input string 295 * @maxbytes - don't go past this many bytes of input string 296 * @codepage - destination codepage 297 * 298 * Walk a utf16le string and return the number of bytes that the string will 299 * be after being converted to the given charset, not including any null 300 * termination required. Don't walk past maxbytes in the source buffer. 301 */ 302 int 303 cifs_utf16_bytes(const __le16 *from, int maxbytes, 304 const struct nls_table *codepage) 305 { 306 int i; 307 int charlen, outlen = 0; 308 int maxwords = maxbytes / 2; 309 char tmp[NLS_MAX_CHARSET_SIZE]; 310 __u16 ftmp[3]; 311 312 for (i = 0; i < maxwords; i++) { 313 ftmp[0] = get_unaligned_le16(&from[i]); 314 if (ftmp[0] == 0) 315 break; 316 if (i + 1 < maxwords) 317 ftmp[1] = get_unaligned_le16(&from[i + 1]); 318 else 319 ftmp[1] = 0; 320 if (i + 2 < maxwords) 321 ftmp[2] = get_unaligned_le16(&from[i + 2]); 322 else 323 ftmp[2] = 0; 324 325 charlen = cifs_mapchar(tmp, ftmp, codepage, NO_MAP_UNI_RSVD); 326 outlen += charlen; 327 } 328 329 return outlen; 330 } 331 332 /* 333 * cifs_strndup_from_utf16 - copy a string from wire format to the local 334 * codepage 335 * @src - source string 336 * @maxlen - don't walk past this many bytes in the source string 337 * @is_unicode - is this a unicode string? 338 * @codepage - destination codepage 339 * 340 * Take a string given by the server, convert it to the local codepage and 341 * put it in a new buffer. Returns a pointer to the new string or NULL on 342 * error. 343 */ 344 char * 345 cifs_strndup_from_utf16(const char *src, const int maxlen, 346 const bool is_unicode, const struct nls_table *codepage) 347 { 348 int len; 349 char *dst; 350 351 if (is_unicode) { 352 len = cifs_utf16_bytes((__le16 *) src, maxlen, codepage); 353 len += nls_nullsize(codepage); 354 dst = kmalloc(len, GFP_KERNEL); 355 if (!dst) 356 return NULL; 357 cifs_from_utf16(dst, (__le16 *) src, len, maxlen, codepage, 358 NO_MAP_UNI_RSVD); 359 } else { 360 dst = kstrndup(src, maxlen, GFP_KERNEL); 361 } 362 363 return dst; 364 } 365 366 static __le16 convert_to_sfu_char(char src_char) 367 { 368 __le16 dest_char; 369 370 switch (src_char) { 371 case ':': 372 dest_char = cpu_to_le16(UNI_COLON); 373 break; 374 case '*': 375 dest_char = cpu_to_le16(UNI_ASTERISK); 376 break; 377 case '?': 378 dest_char = cpu_to_le16(UNI_QUESTION); 379 break; 380 case '<': 381 dest_char = cpu_to_le16(UNI_LESSTHAN); 382 break; 383 case '>': 384 dest_char = cpu_to_le16(UNI_GRTRTHAN); 385 break; 386 case '|': 387 dest_char = cpu_to_le16(UNI_PIPE); 388 break; 389 default: 390 dest_char = 0; 391 } 392 393 return dest_char; 394 } 395 396 static __le16 convert_to_sfm_char(char src_char, bool end_of_string) 397 { 398 __le16 dest_char; 399 400 if (src_char >= 0x01 && src_char <= 0x1F) { 401 dest_char = cpu_to_le16(src_char + 0xF000); 402 return dest_char; 403 } 404 switch (src_char) { 405 case ':': 406 dest_char = cpu_to_le16(SFM_COLON); 407 break; 408 case '"': 409 dest_char = cpu_to_le16(SFM_DOUBLEQUOTE); 410 break; 411 case '*': 412 dest_char = cpu_to_le16(SFM_ASTERISK); 413 break; 414 case '?': 415 dest_char = cpu_to_le16(SFM_QUESTION); 416 break; 417 case '<': 418 dest_char = cpu_to_le16(SFM_LESSTHAN); 419 break; 420 case '>': 421 dest_char = cpu_to_le16(SFM_GRTRTHAN); 422 break; 423 case '|': 424 dest_char = cpu_to_le16(SFM_PIPE); 425 break; 426 case '.': 427 if (end_of_string) 428 dest_char = cpu_to_le16(SFM_PERIOD); 429 else 430 dest_char = 0; 431 break; 432 case ' ': 433 if (end_of_string) 434 dest_char = cpu_to_le16(SFM_SPACE); 435 else 436 dest_char = 0; 437 break; 438 default: 439 dest_char = 0; 440 } 441 442 return dest_char; 443 } 444 445 /* 446 * Convert 16 bit Unicode pathname to wire format from string in current code 447 * page. Conversion may involve remapping up the six characters that are 448 * only legal in POSIX-like OS (if they are present in the string). Path 449 * names are little endian 16 bit Unicode on the wire 450 */ 451 int 452 cifsConvertToUTF16(__le16 *target, const char *source, int srclen, 453 const struct nls_table *cp, int map_chars) 454 { 455 int i, charlen; 456 int j = 0; 457 char src_char; 458 __le16 dst_char; 459 wchar_t tmp; 460 wchar_t *wchar_to; /* UTF-16 */ 461 int ret; 462 unicode_t u; 463 464 if (map_chars == NO_MAP_UNI_RSVD) 465 return cifs_strtoUTF16(target, source, PATH_MAX, cp); 466 467 wchar_to = kzalloc(6, GFP_KERNEL); 468 469 for (i = 0; i < srclen; j++) { 470 src_char = source[i]; 471 charlen = 1; 472 473 /* check if end of string */ 474 if (src_char == 0) 475 goto ctoUTF16_out; 476 477 /* see if we must remap this char */ 478 if (map_chars == SFU_MAP_UNI_RSVD) 479 dst_char = convert_to_sfu_char(src_char); 480 else if (map_chars == SFM_MAP_UNI_RSVD) { 481 bool end_of_string; 482 483 /** 484 * Remap spaces and periods found at the end of every 485 * component of the path. The special cases of '.' and 486 * '..' are need to be handled because of symlinks. 487 * They are treated as non-end-of-string to avoid 488 * remapping and breaking symlinks pointing to . or .. 489 **/ 490 if ((i == 0 || source[i-1] == '\\') && 491 source[i] == '.' && 492 (i == srclen-1 || source[i+1] == '\\')) 493 end_of_string = false; /* "." case */ 494 else if (i >= 1 && 495 (i == 1 || source[i-2] == '\\') && 496 source[i-1] == '.' && 497 source[i] == '.' && 498 (i == srclen-1 || source[i+1] == '\\')) 499 end_of_string = false; /* ".." case */ 500 else if ((i == srclen - 1) || (source[i+1] == '\\')) 501 end_of_string = true; 502 else 503 end_of_string = false; 504 505 dst_char = convert_to_sfm_char(src_char, end_of_string); 506 } else 507 dst_char = 0; 508 /* 509 * FIXME: We can not handle remapping backslash (UNI_SLASH) 510 * until all the calls to build_path_from_dentry are modified, 511 * as they use backslash as separator. 512 */ 513 if (dst_char == 0) { 514 charlen = cp->char2uni(source + i, srclen - i, &tmp); 515 dst_char = cpu_to_le16(tmp); 516 517 /* 518 * if no match, use question mark, which at least in 519 * some cases serves as wild card 520 */ 521 if (charlen > 0) 522 goto ctoUTF16; 523 524 /* convert SURROGATE_PAIR */ 525 if (strcmp(cp->charset, "utf8") || !wchar_to) 526 goto unknown; 527 if (*(source + i) & 0x80) { 528 charlen = utf8_to_utf32(source + i, 6, &u); 529 if (charlen < 0) 530 goto unknown; 531 } else 532 goto unknown; 533 ret = utf8s_to_utf16s(source + i, charlen, 534 UTF16_LITTLE_ENDIAN, 535 wchar_to, 6); 536 if (ret < 0) 537 goto unknown; 538 539 i += charlen; 540 dst_char = cpu_to_le16(*wchar_to); 541 if (charlen <= 3) 542 /* 1-3bytes UTF-8 to 2bytes UTF-16 */ 543 put_unaligned(dst_char, &target[j]); 544 else if (charlen == 4) { 545 /* 4bytes UTF-8(surrogate pair) to 4bytes UTF-16 546 * 7-8bytes UTF-8(IVS) divided to 2 UTF-16 547 * (charlen=3+4 or 4+4) */ 548 put_unaligned(dst_char, &target[j]); 549 dst_char = cpu_to_le16(*(wchar_to + 1)); 550 j++; 551 put_unaligned(dst_char, &target[j]); 552 } else if (charlen >= 5) { 553 /* 5-6bytes UTF-8 to 6bytes UTF-16 */ 554 put_unaligned(dst_char, &target[j]); 555 dst_char = cpu_to_le16(*(wchar_to + 1)); 556 j++; 557 put_unaligned(dst_char, &target[j]); 558 dst_char = cpu_to_le16(*(wchar_to + 2)); 559 j++; 560 put_unaligned(dst_char, &target[j]); 561 } 562 continue; 563 564 unknown: 565 dst_char = cpu_to_le16(0x003f); 566 charlen = 1; 567 } 568 569 ctoUTF16: 570 /* 571 * character may take more than one byte in the source string, 572 * but will take exactly two bytes in the target string 573 */ 574 i += charlen; 575 put_unaligned(dst_char, &target[j]); 576 } 577 578 ctoUTF16_out: 579 put_unaligned(0, &target[j]); /* Null terminate target unicode string */ 580 kfree(wchar_to); 581 return j; 582 } 583 584 /* 585 * cifs_local_to_utf16_bytes - how long will a string be after conversion? 586 * @from - pointer to input string 587 * @maxbytes - don't go past this many bytes of input string 588 * @codepage - source codepage 589 * 590 * Walk a string and return the number of bytes that the string will 591 * be after being converted to the given charset, not including any null 592 * termination required. Don't walk past maxbytes in the source buffer. 593 */ 594 595 static int 596 cifs_local_to_utf16_bytes(const char *from, int len, 597 const struct nls_table *codepage) 598 { 599 int charlen; 600 int i; 601 wchar_t wchar_to; 602 603 for (i = 0; len && *from; i++, from += charlen, len -= charlen) { 604 charlen = codepage->char2uni(from, len, &wchar_to); 605 /* Failed conversion defaults to a question mark */ 606 if (charlen < 1) 607 charlen = 1; 608 } 609 return 2 * i; /* UTF16 characters are two bytes */ 610 } 611 612 /* 613 * cifs_strndup_to_utf16 - copy a string to wire format from the local codepage 614 * @src - source string 615 * @maxlen - don't walk past this many bytes in the source string 616 * @utf16_len - the length of the allocated string in bytes (including null) 617 * @cp - source codepage 618 * @remap - map special chars 619 * 620 * Take a string convert it from the local codepage to UTF16 and 621 * put it in a new buffer. Returns a pointer to the new string or NULL on 622 * error. 623 */ 624 __le16 * 625 cifs_strndup_to_utf16(const char *src, const int maxlen, int *utf16_len, 626 const struct nls_table *cp, int remap) 627 { 628 int len; 629 __le16 *dst; 630 631 if (!src) 632 return NULL; 633 634 len = cifs_local_to_utf16_bytes(src, maxlen, cp); 635 len += 2; /* NULL */ 636 dst = kmalloc(len, GFP_KERNEL); 637 if (!dst) { 638 *utf16_len = 0; 639 return NULL; 640 } 641 cifsConvertToUTF16(dst, src, strlen(src), cp, remap); 642 *utf16_len = len; 643 return dst; 644 } 645