fs/cifs/cifs_unicode.c

   1 /*
   2  *   fs/cifs/cifs_unicode.c
   3  *
   4  *   Copyright (c) International Business Machines  Corp., 2000,2009
   5  *   Modified by Steve French (sfrench@us.ibm.com)
   6  *
   7  *   This program is free software;  you can redistribute it and/or modify
   8  *   it under the terms of the GNU General Public License as published by
   9  *   the Free Software Foundation; either version 2 of the License, or
  10  *   (at your option) any later version.
  11  *
  12  *   This program is distributed in the hope that it will be useful,
  13  *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
  14  *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
  15  *   the GNU General Public License for more details.
  16  *
  17  *   You should have received a copy of the GNU General Public License
  18  *   along with this program;  if not, write to the Free Software
  19  *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
  20  */
  21 #include <linux/fs.h>
  22 #include <linux/slab.h>
  23 #include "cifs_fs_sb.h"
  24 #include "cifs_unicode.h"
  25 #include "cifs_uniupr.h"
  26 #include "cifspdu.h"
  27 #include "cifsglob.h"
  28 #include "cifs_debug.h"
  29
  30 int cifs_remap(struct cifs_sb_info *cifs_sb)
  31 {
  32         int map_type;
  33
  34         if (cifs_sb->mnt_cifs_flags & CIFS_MOUNT_MAP_SFM_CHR)
  35                 map_type = SFM_MAP_UNI_RSVD;
  36         else if (cifs_sb->mnt_cifs_flags & CIFS_MOUNT_MAP_SPECIAL_CHR)
  37                 map_type = SFU_MAP_UNI_RSVD;
  38         else
  39                 map_type = NO_MAP_UNI_RSVD;
  40
  41         return map_type;
  42 }
  43
  44 /* Convert character using the SFU - "Services for Unix" remapping range */
  45 static bool
  46 convert_sfu_char(const __u16 src_char, char *target)
  47 {
  48         /*
  49          * BB: Cannot handle remapping UNI_SLASH until all the calls to
  50          *     build_path_from_dentry are modified, as they use slash as
  51          *     separator.
  52          */
  53         switch (src_char) {
  54         case UNI_COLON:
  55                 *target = ':';
  56                 break;
  57         case UNI_ASTERISK:
  58                 *target = '*';
  59                 break;
  60         case UNI_QUESTION:
  61                 *target = '?';
  62                 break;
  63         case UNI_PIPE:
  64                 *target = '|';
  65                 break;
  66         case UNI_GRTRTHAN:
  67                 *target = '>';
  68                 break;
  69         case UNI_LESSTHAN:
  70                 *target = '<';
  71                 break;
  72         default:
  73                 return false;
  74         }
  75         return true;
  76 }
  77
  78 /* Convert character using the SFM - "Services for Mac" remapping range */
  79 static bool
  80 convert_sfm_char(const __u16 src_char, char *target)
  81 {
  82         switch (src_char) {
  83         case SFM_COLON:
  84                 *target = ':';
  85                 break;
  86         case SFM_DOUBLEQUOTE:
  87                 *target = '"';
  88                 break;
  89         case SFM_ASTERISK:
  90                 *target = '*';
  91                 break;
  92         case SFM_QUESTION:
  93                 *target = '?';
  94                 break;
  95         case SFM_PIPE:
  96                 *target = '|';
  97                 break;
  98         case SFM_GRTRTHAN:
  99                 *target = '>';
 100                 break;
 101         case SFM_LESSTHAN:
 102                 *target = '<';
 103                 break;
 104         case SFM_SPACE:
 105                 *target = ' ';
 106                 break;
 107         case SFM_PERIOD:
 108                 *target = '.';
 109                 break;
 110         default:
 111                 return false;
 112         }
 113         return true;
 114 }
 115
 116
 117 /*
 118  * cifs_mapchar - convert a host-endian char to proper char in codepage
 119  * @target - where converted character should be copied
 120  * @src_char - 2 byte host-endian source character
 121  * @cp - codepage to which character should be converted
 122  * @map_type - How should the 7 NTFS/SMB reserved characters be mapped to UCS2?
 123  *
 124  * This function handles the conversion of a single character. It is the
 125  * responsibility of the caller to ensure that the target buffer is large
 126  * enough to hold the result of the conversion (at least NLS_MAX_CHARSET_SIZE).
 127  */
 128 static int
 129 cifs_mapchar(char *target, const __u16 *from, const struct nls_table *cp,
 130              int maptype)
 131 {
 132         int len = 1;
 133         __u16 src_char;
 134
 135         src_char = *from;
 136
 137         if ((maptype == SFM_MAP_UNI_RSVD) && convert_sfm_char(src_char, target))
 138                 return len;
 139         else if ((maptype == SFU_MAP_UNI_RSVD) &&
 140                   convert_sfu_char(src_char, target))
 141                 return len;
 142
 143         /* if character not one of seven in special remap set */
 144         len = cp->uni2char(src_char, target, NLS_MAX_CHARSET_SIZE);
 145         if (len <= 0)
 146                 goto surrogate_pair;
 147
 148         return len;
 149
 150 surrogate_pair:
 151         /* convert SURROGATE_PAIR and IVS */
 152         if (strcmp(cp->charset, "utf8"))
 153                 goto unknown;
 154         len = utf16s_to_utf8s(from, 3, UTF16_LITTLE_ENDIAN, target, 6);
 155         if (len <= 0)
 156                 goto unknown;
 157         return len;
 158
 159 unknown:
 160         *target = '?';
 161         len = 1;
 162         return len;
 163 }
 164
 165 /*
 166  * cifs_from_utf16 - convert utf16le string to local charset
 167  * @to - destination buffer
 168  * @from - source buffer
 169  * @tolen - destination buffer size (in bytes)
 170  * @fromlen - source buffer size (in bytes)
 171  * @codepage - codepage to which characters should be converted
 172  * @mapchar - should characters be remapped according to the mapchars option?
 173  *
 174  * Convert a little-endian utf16le string (as sent by the server) to a string
 175  * in the provided codepage. The tolen and fromlen parameters are to ensure
 176  * that the code doesn't walk off of the end of the buffer (which is always
 177  * a danger if the alignment of the source buffer is off). The destination
 178  * string is always properly null terminated and fits in the destination
 179  * buffer. Returns the length of the destination string in bytes (including
 180  * null terminator).
 181  *
 182  * Note that some windows versions actually send multiword UTF-16 characters
 183  * instead of straight UTF16-2. The linux nls routines however aren't able to
 184  * deal with those characters properly. In the event that we get some of
 185  * those characters, they won't be translated properly.
 186  */
 187 int
 188 cifs_from_utf16(char *to, const __le16 *from, int tolen, int fromlen,
 189                 const struct nls_table *codepage, int map_type)
 190 {
 191         int i, charlen, safelen;
 192         int outlen = 0;
 193         int nullsize = nls_nullsize(codepage);
 194         int fromwords = fromlen / 2;
 195         char tmp[NLS_MAX_CHARSET_SIZE];
 196         __u16 ftmp[3];          /* ftmp[3] = 3array x 2bytes = 6bytes UTF-16 */
 197
 198         /*
 199          * because the chars can be of varying widths, we need to take care
 200          * not to overflow the destination buffer when we get close to the
 201          * end of it. Until we get to this offset, we don't need to check
 202          * for overflow however.
 203          */
 204         safelen = tolen - (NLS_MAX_CHARSET_SIZE + nullsize);
 205
 206         for (i = 0; i < fromwords; i++) {
 207                 ftmp[0] = get_unaligned_le16(&from[i]);
 208                 if (ftmp[0] == 0)
 209                         break;
 210                 if (i + 1 < fromwords)
 211                         ftmp[1] = get_unaligned_le16(&from[i + 1]);
 212                 else
 213                         ftmp[1] = 0;
 214                 if (i + 2 < fromwords)
 215                         ftmp[2] = get_unaligned_le16(&from[i + 2]);
 216                 else
 217                         ftmp[2] = 0;
 218
 219                 /*
 220                  * check to see if converting this character might make the
 221                  * conversion bleed into the null terminator
 222                  */
 223                 if (outlen >= safelen) {
 224                         charlen = cifs_mapchar(tmp, ftmp, codepage, map_type);
 225                         if ((outlen + charlen) > (tolen - nullsize))
 226                                 break;
 227                 }
 228
 229                 /* put converted char into 'to' buffer */
 230                 charlen = cifs_mapchar(&to[outlen], ftmp, codepage, map_type);
 231                 outlen += charlen;
 232
 233                 /* charlen (=bytes of UTF-8 for 1 character)
 234                  * 4bytes UTF-8(surrogate pair) is charlen=4
 235                  *   (4bytes UTF-16 code)
 236                  * 7-8bytes UTF-8(IVS) is charlen=3+4 or 4+4
 237                  *   (2 UTF-8 pairs divided to 2 UTF-16 pairs) */
 238                 if (charlen == 4)
 239                         i++;
 240                 else if (charlen >= 5)
 241                         /* 5-6bytes UTF-8 */
 242                         i += 2;
 243         }
 244
 245         /* properly null-terminate string */
 246         for (i = 0; i < nullsize; i++)
 247                 to[outlen++] = 0;
 248
 249         return outlen;
 250 }
 251
 252 /*
 253  * NAME:        cifs_strtoUTF16()
 254  *
 255  * FUNCTION:    Convert character string to unicode string
 256  *
 257  */
 258 int
 259 cifs_strtoUTF16(__le16 *to, const char *from, int len,
 260               const struct nls_table *codepage)
 261 {
 262         int charlen;
 263         int i;
 264         wchar_t wchar_to; /* needed to quiet sparse */
 265
 266         /* special case for utf8 to handle no plane0 chars */
 267         if (!strcmp(codepage->charset, "utf8")) {
 268                 /*
 269                  * convert utf8 -> utf16, we assume we have enough space
 270                  * as caller should have assumed conversion does not overflow
 271                  * in destination len is length in wchar_t units (16bits)
 272                  */
 273                 i  = utf8s_to_utf16s(from, len, UTF16_LITTLE_ENDIAN,
 274                                        (wchar_t *) to, len);
 275
 276                 /* if success terminate and exit */
 277                 if (i >= 0)
 278                         goto success;
 279                 /*
 280                  * if fails fall back to UCS encoding as this
 281                  * function should not return negative values
 282                  * currently can fail only if source contains
 283                  * invalid encoded characters
 284                  */
 285         }
 286
 287         for (i = 0; len && *from; i++, from += charlen, len -= charlen) {
 288                 charlen = codepage->char2uni(from, len, &wchar_to);
 289                 if (charlen < 1) {
 290                         cifs_dbg(VFS, "strtoUTF16: char2uni of 0x%x returned %d\n",
 291                                  *from, charlen);
 292                         /* A question mark */
 293                         wchar_to = 0x003f;
 294                         charlen = 1;
 295                 }
 296                 put_unaligned_le16(wchar_to, &to[i]);
 297         }
 298
 299 success:
 300         put_unaligned_le16(0, &to[i]);
 301         return i;
 302 }
 303
 304 /*
 305  * cifs_utf16_bytes - how long will a string be after conversion?
 306  * @utf16 - pointer to input string
 307  * @maxbytes - don't go past this many bytes of input string
 308  * @codepage - destination codepage
 309  *
 310  * Walk a utf16le string and return the number of bytes that the string will
 311  * be after being converted to the given charset, not including any null
 312  * termination required. Don't walk past maxbytes in the source buffer.
 313  */
 314 int
 315 cifs_utf16_bytes(const __le16 *from, int maxbytes,
 316                 const struct nls_table *codepage)
 317 {
 318         int i;
 319         int charlen, outlen = 0;
 320         int maxwords = maxbytes / 2;
 321         char tmp[NLS_MAX_CHARSET_SIZE];
 322         __u16 ftmp[3];
 323
 324         for (i = 0; i < maxwords; i++) {
 325                 ftmp[0] = get_unaligned_le16(&from[i]);
 326                 if (ftmp[0] == 0)
 327                         break;
 328                 if (i + 1 < maxwords)
 329                         ftmp[1] = get_unaligned_le16(&from[i + 1]);
 330                 else
 331                         ftmp[1] = 0;
 332                 if (i + 2 < maxwords)
 333                         ftmp[2] = get_unaligned_le16(&from[i + 2]);
 334                 else
 335                         ftmp[2] = 0;
 336
 337                 charlen = cifs_mapchar(tmp, ftmp, codepage, NO_MAP_UNI_RSVD);
 338                 outlen += charlen;
 339         }
 340
 341         return outlen;
 342 }
 343
 344 /*
 345  * cifs_strndup_from_utf16 - copy a string from wire format to the local
 346  * codepage
 347  * @src - source string
 348  * @maxlen - don't walk past this many bytes in the source string
 349  * @is_unicode - is this a unicode string?
 350  * @codepage - destination codepage
 351  *
 352  * Take a string given by the server, convert it to the local codepage and
 353  * put it in a new buffer. Returns a pointer to the new string or NULL on
 354  * error.
 355  */
 356 char *
 357 cifs_strndup_from_utf16(const char *src, const int maxlen,
 358                         const bool is_unicode, const struct nls_table *codepage)
 359 {
 360         int len;
 361         char *dst;
 362
 363         if (is_unicode) {
 364                 len = cifs_utf16_bytes((__le16 *) src, maxlen, codepage);
 365                 len += nls_nullsize(codepage);
 366                 dst = kmalloc(len, GFP_KERNEL);
 367                 if (!dst)
 368                         return NULL;
 369                 cifs_from_utf16(dst, (__le16 *) src, len, maxlen, codepage,
 370                                NO_MAP_UNI_RSVD);
 371         } else {
 372                 len = strnlen(src, maxlen);
 373                 len++;
 374                 dst = kmalloc(len, GFP_KERNEL);
 375                 if (!dst)
 376                         return NULL;
 377                 strlcpy(dst, src, len);
 378         }
 379
 380         return dst;
 381 }
 382
 383 static __le16 convert_to_sfu_char(char src_char)
 384 {
 385         __le16 dest_char;
 386
 387         switch (src_char) {
 388         case ':':
 389                 dest_char = cpu_to_le16(UNI_COLON);
 390                 break;
 391         case '*':
 392                 dest_char = cpu_to_le16(UNI_ASTERISK);
 393                 break;
 394         case '?':
 395                 dest_char = cpu_to_le16(UNI_QUESTION);
 396                 break;
 397         case '<':
 398                 dest_char = cpu_to_le16(UNI_LESSTHAN);
 399                 break;
 400         case '>':
 401                 dest_char = cpu_to_le16(UNI_GRTRTHAN);
 402                 break;
 403         case '|':
 404                 dest_char = cpu_to_le16(UNI_PIPE);
 405                 break;
 406         default:
 407                 dest_char = 0;
 408         }
 409
 410         return dest_char;
 411 }
 412
 413 static __le16 convert_to_sfm_char(char src_char, bool end_of_string)
 414 {
 415         __le16 dest_char;
 416
 417         switch (src_char) {
 418         case ':':
 419                 dest_char = cpu_to_le16(SFM_COLON);
 420                 break;
 421         case '"':
 422                 dest_char = cpu_to_le16(SFM_DOUBLEQUOTE);
 423                 break;
 424         case '*':
 425                 dest_char = cpu_to_le16(SFM_ASTERISK);
 426                 break;
 427         case '?':
 428                 dest_char = cpu_to_le16(SFM_QUESTION);
 429                 break;
 430         case '<':
 431                 dest_char = cpu_to_le16(SFM_LESSTHAN);
 432                 break;
 433         case '>':
 434                 dest_char = cpu_to_le16(SFM_GRTRTHAN);
 435                 break;
 436         case '|':
 437                 dest_char = cpu_to_le16(SFM_PIPE);
 438                 break;
 439         case '.':
 440                 if (end_of_string)
 441                         dest_char = cpu_to_le16(SFM_PERIOD);
 442                 else
 443                         dest_char = 0;
 444                 break;
 445         case ' ':
 446                 if (end_of_string)
 447                         dest_char = cpu_to_le16(SFM_SPACE);
 448                 else
 449                         dest_char = 0;
 450                 break;
 451         default:
 452                 dest_char = 0;
 453         }
 454
 455         return dest_char;
 456 }
 457
 458 /*
 459  * Convert 16 bit Unicode pathname to wire format from string in current code
 460  * page. Conversion may involve remapping up the six characters that are
 461  * only legal in POSIX-like OS (if they are present in the string). Path
 462  * names are little endian 16 bit Unicode on the wire
 463  */
 464 int
 465 cifsConvertToUTF16(__le16 *target, const char *source, int srclen,
 466                  const struct nls_table *cp, int map_chars)
 467 {
 468         int i, charlen;
 469         int j = 0;
 470         char src_char;
 471         __le16 dst_char;
 472         wchar_t tmp;
 473         wchar_t *wchar_to;      /* UTF-16 */
 474         int ret;
 475         unicode_t u;
 476
 477         if (map_chars == NO_MAP_UNI_RSVD)
 478                 return cifs_strtoUTF16(target, source, PATH_MAX, cp);
 479
 480         wchar_to = kzalloc(6, GFP_KERNEL);
 481
 482         for (i = 0; i < srclen; j++) {
 483                 src_char = source[i];
 484                 charlen = 1;
 485
 486                 /* check if end of string */
 487                 if (src_char == 0)
 488                         goto ctoUTF16_out;
 489
 490                 /* see if we must remap this char */
 491                 if (map_chars == SFU_MAP_UNI_RSVD)
 492                         dst_char = convert_to_sfu_char(src_char);
 493                 else if (map_chars == SFM_MAP_UNI_RSVD) {
 494                         bool end_of_string;
 495
 496                         if (i == srclen - 1)
 497                                 end_of_string = true;
 498                         else
 499                                 end_of_string = false;
 500
 501                         dst_char = convert_to_sfm_char(src_char, end_of_string);
 502                 } else
 503                         dst_char = 0;
 504                 /*
 505                  * FIXME: We can not handle remapping backslash (UNI_SLASH)
 506                  * until all the calls to build_path_from_dentry are modified,
 507                  * as they use backslash as separator.
 508                  */
 509                 if (dst_char == 0) {
 510                         charlen = cp->char2uni(source + i, srclen - i, &tmp);
 511                         dst_char = cpu_to_le16(tmp);
 512
 513                         /*
 514                          * if no match, use question mark, which at least in
 515                          * some cases serves as wild card
 516                          */
 517                         if (charlen > 0)
 518                                 goto ctoUTF16;
 519
 520                         /* convert SURROGATE_PAIR */
 521                         if (strcmp(cp->charset, "utf8") || !wchar_to)
 522                                 goto unknown;
 523                         if (*(source + i) & 0x80) {
 524                                 charlen = utf8_to_utf32(source + i, 6, &u);
 525                                 if (charlen < 0)
 526                                         goto unknown;
 527                         } else
 528                                 goto unknown;
 529                         ret  = utf8s_to_utf16s(source + i, charlen,
 530                                                UTF16_LITTLE_ENDIAN,
 531                                                wchar_to, 6);
 532                         if (ret < 0)
 533                                 goto unknown;
 534
 535                         i += charlen;
 536                         dst_char = cpu_to_le16(*wchar_to);
 537                         if (charlen <= 3)
 538                                 /* 1-3bytes UTF-8 to 2bytes UTF-16 */
 539                                 put_unaligned(dst_char, &target[j]);
 540                         else if (charlen == 4) {
 541                                 /* 4bytes UTF-8(surrogate pair) to 4bytes UTF-16
 542                                  * 7-8bytes UTF-8(IVS) divided to 2 UTF-16
 543                                  *   (charlen=3+4 or 4+4) */
 544                                 put_unaligned(dst_char, &target[j]);
 545                                 dst_char = cpu_to_le16(*(wchar_to + 1));
 546                                 j++;
 547                                 put_unaligned(dst_char, &target[j]);
 548                         } else if (charlen >= 5) {
 549                                 /* 5-6bytes UTF-8 to 6bytes UTF-16 */
 550                                 put_unaligned(dst_char, &target[j]);
 551                                 dst_char = cpu_to_le16(*(wchar_to + 1));
 552                                 j++;
 553                                 put_unaligned(dst_char, &target[j]);
 554                                 dst_char = cpu_to_le16(*(wchar_to + 2));
 555                                 j++;
 556                                 put_unaligned(dst_char, &target[j]);
 557                         }
 558                         continue;
 559
 560 unknown:
 561                         dst_char = cpu_to_le16(0x003f);
 562                         charlen = 1;
 563                 }
 564
 565 ctoUTF16:
 566                 /*
 567                  * character may take more than one byte in the source string,
 568                  * but will take exactly two bytes in the target string
 569                  */
 570                 i += charlen;
 571                 put_unaligned(dst_char, &target[j]);
 572         }
 573
 574 ctoUTF16_out:
 575         put_unaligned(0, &target[j]); /* Null terminate target unicode string */
 576         kfree(wchar_to);
 577         return j;
 578 }
 579
 580 #ifdef CONFIG_CIFS_SMB2
 581 /*
 582  * cifs_local_to_utf16_bytes - how long will a string be after conversion?
 583  * @from - pointer to input string
 584  * @maxbytes - don't go past this many bytes of input string
 585  * @codepage - source codepage
 586  *
 587  * Walk a string and return the number of bytes that the string will
 588  * be after being converted to the given charset, not including any null
 589  * termination required. Don't walk past maxbytes in the source buffer.
 590  */
 591
 592 static int
 593 cifs_local_to_utf16_bytes(const char *from, int len,
 594                           const struct nls_table *codepage)
 595 {
 596         int charlen;
 597         int i;
 598         wchar_t wchar_to;
 599
 600         for (i = 0; len && *from; i++, from += charlen, len -= charlen) {
 601                 charlen = codepage->char2uni(from, len, &wchar_to);
 602                 /* Failed conversion defaults to a question mark */
 603                 if (charlen < 1)
 604                         charlen = 1;
 605         }
 606         return 2 * i; /* UTF16 characters are two bytes */
 607 }
 608
 609 /*
 610  * cifs_strndup_to_utf16 - copy a string to wire format from the local codepage
 611  * @src - source string
 612  * @maxlen - don't walk past this many bytes in the source string
 613  * @utf16_len - the length of the allocated string in bytes (including null)
 614  * @cp - source codepage
 615  * @remap - map special chars
 616  *
 617  * Take a string convert it from the local codepage to UTF16 and
 618  * put it in a new buffer. Returns a pointer to the new string or NULL on
 619  * error.
 620  */
 621 __le16 *
 622 cifs_strndup_to_utf16(const char *src, const int maxlen, int *utf16_len,
 623                       const struct nls_table *cp, int remap)
 624 {
 625         int len;
 626         __le16 *dst;
 627
 628         len = cifs_local_to_utf16_bytes(src, maxlen, cp);
 629         len += 2; /* NULL */
 630         dst = kmalloc(len, GFP_KERNEL);
 631         if (!dst) {
 632                 *utf16_len = 0;
 633                 return NULL;
 634         }
 635         cifsConvertToUTF16(dst, src, strlen(src), cp, remap);
 636         *utf16_len = len;
 637         return dst;
 638 }
 639 #endif /* CONFIG_CIFS_SMB2 */