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