fs/udf/unicode.c

   1 /*
   2  * unicode.c
   3  *
   4  * PURPOSE
   5  *      Routines for converting between UTF-8 and OSTA Compressed Unicode.
   6  *      Also handles filename mangling
   7  *
   8  * DESCRIPTION
   9  *      OSTA Compressed Unicode is explained in the OSTA UDF specification.
  10  *              http://www.osta.org/
  11  *      UTF-8 is explained in the IETF RFC XXXX.
  12  *              ftp://ftp.internic.net/rfc/rfcxxxx.txt
  13  *
  14  * COPYRIGHT
  15  *      This file is distributed under the terms of the GNU General Public
  16  *      License (GPL). Copies of the GPL can be obtained from:
  17  *              ftp://prep.ai.mit.edu/pub/gnu/GPL
  18  *      Each contributing author retains all rights to their own work.
  19  */
  20
  21 #include "udfdecl.h"
  22
  23 #include <linux/kernel.h>
  24 #include <linux/string.h>       /* for memset */
  25 #include <linux/nls.h>
  26 #include <linux/crc-itu-t.h>
  27 #include <linux/slab.h>
  28
  29 #include "udf_sb.h"
  30
  31 static int udf_translate_to_linux(uint8_t *, int, uint8_t *, int, uint8_t *,
  32                                   int);
  33
  34 static int udf_char_to_ustr(struct ustr *dest, const uint8_t *src, int strlen)
  35 {
  36         if ((!dest) || (!src) || (!strlen) || (strlen > UDF_NAME_LEN - 2))
  37                 return 0;
  38
  39         memset(dest, 0, sizeof(struct ustr));
  40         memcpy(dest->u_name, src, strlen);
  41         dest->u_cmpID = 0x08;
  42         dest->u_len = strlen;
  43
  44         return strlen;
  45 }
  46
  47 /*
  48  * udf_build_ustr
  49  */
  50 int udf_build_ustr(struct ustr *dest, dstring *ptr, int size)
  51 {
  52         int usesize;
  53
  54         if (!dest || !ptr || !size)
  55                 return -1;
  56         BUG_ON(size < 2);
  57
  58         usesize = min_t(size_t, ptr[size - 1], sizeof(dest->u_name));
  59         usesize = min(usesize, size - 2);
  60         dest->u_cmpID = ptr[0];
  61         dest->u_len = usesize;
  62         memcpy(dest->u_name, ptr + 1, usesize);
  63         memset(dest->u_name + usesize, 0, sizeof(dest->u_name) - usesize);
  64
  65         return 0;
  66 }
  67
  68 /*
  69  * udf_build_ustr_exact
  70  */
  71 static int udf_build_ustr_exact(struct ustr *dest, dstring *ptr, int exactsize)
  72 {
  73         if ((!dest) || (!ptr) || (!exactsize))
  74                 return -1;
  75
  76         memset(dest, 0, sizeof(struct ustr));
  77         dest->u_cmpID = ptr[0];
  78         dest->u_len = exactsize - 1;
  79         memcpy(dest->u_name, ptr + 1, exactsize - 1);
  80
  81         return 0;
  82 }
  83
  84 /*
  85  * udf_ocu_to_utf8
  86  *
  87  * PURPOSE
  88  *      Convert OSTA Compressed Unicode to the UTF-8 equivalent.
  89  *
  90  * PRE-CONDITIONS
  91  *      utf                     Pointer to UTF-8 output buffer.
  92  *      ocu                     Pointer to OSTA Compressed Unicode input buffer
  93  *                              of size UDF_NAME_LEN bytes.
  94  *                              both of type "struct ustr *"
  95  *
  96  * POST-CONDITIONS
  97  *      <return>                Zero on success.
  98  *
  99  * HISTORY
 100  *      November 12, 1997 - Andrew E. Mileski
 101  *      Written, tested, and released.
 102  */
 103 int udf_CS0toUTF8(struct ustr *utf_o, const struct ustr *ocu_i)
 104 {
 105         const uint8_t *ocu;
 106         uint8_t cmp_id, ocu_len;
 107         int i;
 108
 109         ocu_len = ocu_i->u_len;
 110         if (ocu_len == 0) {
 111                 memset(utf_o, 0, sizeof(struct ustr));
 112                 return 0;
 113         }
 114
 115         cmp_id = ocu_i->u_cmpID;
 116         if (cmp_id != 8 && cmp_id != 16) {
 117                 memset(utf_o, 0, sizeof(struct ustr));
 118                 pr_err("unknown compression code (%d) stri=%s\n",
 119                        cmp_id, ocu_i->u_name);
 120                 return 0;
 121         }
 122
 123         ocu = ocu_i->u_name;
 124         utf_o->u_len = 0;
 125         for (i = 0; (i < ocu_len) && (utf_o->u_len <= (UDF_NAME_LEN - 3));) {
 126
 127                 /* Expand OSTA compressed Unicode to Unicode */
 128                 uint32_t c = ocu[i++];
 129                 if (cmp_id == 16)
 130                         c = (c << 8) | ocu[i++];
 131
 132                 /* Compress Unicode to UTF-8 */
 133                 if (c < 0x80U)
 134                         utf_o->u_name[utf_o->u_len++] = (uint8_t)c;
 135                 else if (c < 0x800U) {
 136                         if (utf_o->u_len > (UDF_NAME_LEN - 4))
 137                                 break;
 138                         utf_o->u_name[utf_o->u_len++] =
 139                                                 (uint8_t)(0xc0 | (c >> 6));
 140                         utf_o->u_name[utf_o->u_len++] =
 141                                                 (uint8_t)(0x80 | (c & 0x3f));
 142                 } else {
 143                         if (utf_o->u_len > (UDF_NAME_LEN - 5))
 144                                 break;
 145                         utf_o->u_name[utf_o->u_len++] =
 146                                                 (uint8_t)(0xe0 | (c >> 12));
 147                         utf_o->u_name[utf_o->u_len++] =
 148                                                 (uint8_t)(0x80 |
 149                                                           ((c >> 6) & 0x3f));
 150                         utf_o->u_name[utf_o->u_len++] =
 151                                                 (uint8_t)(0x80 | (c & 0x3f));
 152                 }
 153         }
 154         utf_o->u_cmpID = 8;
 155
 156         return utf_o->u_len;
 157 }
 158
 159 /*
 160  *
 161  * udf_utf8_to_ocu
 162  *
 163  * PURPOSE
 164  *      Convert UTF-8 to the OSTA Compressed Unicode equivalent.
 165  *
 166  * DESCRIPTION
 167  *      This routine is only called by udf_lookup().
 168  *
 169  * PRE-CONDITIONS
 170  *      ocu                     Pointer to OSTA Compressed Unicode output
 171  *                              buffer of size UDF_NAME_LEN bytes.
 172  *      utf                     Pointer to UTF-8 input buffer.
 173  *      utf_len                 Length of UTF-8 input buffer in bytes.
 174  *
 175  * POST-CONDITIONS
 176  *      <return>                Zero on success.
 177  *
 178  * HISTORY
 179  *      November 12, 1997 - Andrew E. Mileski
 180  *      Written, tested, and released.
 181  */
 182 static int udf_UTF8toCS0(dstring *ocu, struct ustr *utf, int length)
 183 {
 184         unsigned c, i, max_val, utf_char;
 185         int utf_cnt, u_len, u_ch;
 186
 187         memset(ocu, 0, sizeof(dstring) * length);
 188         ocu[0] = 8;
 189         max_val = 0xffU;
 190         u_ch = 1;
 191
 192 try_again:
 193         u_len = 0U;
 194         utf_char = 0U;
 195         utf_cnt = 0U;
 196         for (i = 0U; i < utf->u_len; i++) {
 197                 /* Name didn't fit? */
 198                 if (u_len + 1 + u_ch >= length)
 199                         return 0;
 200
 201                 c = (uint8_t)utf->u_name[i];
 202
 203                 /* Complete a multi-byte UTF-8 character */
 204                 if (utf_cnt) {
 205                         utf_char = (utf_char << 6) | (c & 0x3fU);
 206                         if (--utf_cnt)
 207                                 continue;
 208                 } else {
 209                         /* Check for a multi-byte UTF-8 character */
 210                         if (c & 0x80U) {
 211                                 /* Start a multi-byte UTF-8 character */
 212                                 if ((c & 0xe0U) == 0xc0U) {
 213                                         utf_char = c & 0x1fU;
 214                                         utf_cnt = 1;
 215                                 } else if ((c & 0xf0U) == 0xe0U) {
 216                                         utf_char = c & 0x0fU;
 217                                         utf_cnt = 2;
 218                                 } else if ((c & 0xf8U) == 0xf0U) {
 219                                         utf_char = c & 0x07U;
 220                                         utf_cnt = 3;
 221                                 } else if ((c & 0xfcU) == 0xf8U) {
 222                                         utf_char = c & 0x03U;
 223                                         utf_cnt = 4;
 224                                 } else if ((c & 0xfeU) == 0xfcU) {
 225                                         utf_char = c & 0x01U;
 226                                         utf_cnt = 5;
 227                                 } else {
 228                                         goto error_out;
 229                                 }
 230                                 continue;
 231                         } else {
 232                                 /* Single byte UTF-8 character (most common) */
 233                                 utf_char = c;
 234                         }
 235                 }
 236
 237                 /* Choose no compression if necessary */
 238                 if (utf_char > max_val) {
 239                         if (max_val == 0xffU) {
 240                                 max_val = 0xffffU;
 241                                 ocu[0] = (uint8_t)0x10U;
 242                                 u_ch = 2;
 243                                 goto try_again;
 244                         }
 245                         goto error_out;
 246                 }
 247
 248                 if (max_val == 0xffffU)
 249                         ocu[++u_len] = (uint8_t)(utf_char >> 8);
 250                 ocu[++u_len] = (uint8_t)(utf_char & 0xffU);
 251         }
 252
 253         if (utf_cnt) {
 254 error_out:
 255                 ocu[++u_len] = '?';
 256                 printk(KERN_DEBUG pr_fmt("bad UTF-8 character\n"));
 257         }
 258
 259         ocu[length - 1] = (uint8_t)u_len + 1;
 260
 261         return u_len + 1;
 262 }
 263
 264 static int udf_CS0toNLS(struct nls_table *nls, struct ustr *utf_o,
 265                         const struct ustr *ocu_i)
 266 {
 267         const uint8_t *ocu;
 268         uint8_t cmp_id, ocu_len;
 269         int i, len;
 270
 271
 272         ocu_len = ocu_i->u_len;
 273         if (ocu_len == 0) {
 274                 memset(utf_o, 0, sizeof(struct ustr));
 275                 return 0;
 276         }
 277
 278         cmp_id = ocu_i->u_cmpID;
 279         if (cmp_id != 8 && cmp_id != 16) {
 280                 memset(utf_o, 0, sizeof(struct ustr));
 281                 pr_err("unknown compression code (%d) stri=%s\n",
 282                        cmp_id, ocu_i->u_name);
 283                 return 0;
 284         }
 285
 286         ocu = ocu_i->u_name;
 287         utf_o->u_len = 0;
 288         for (i = 0; (i < ocu_len) && (utf_o->u_len <= (UDF_NAME_LEN - 3));) {
 289                 /* Expand OSTA compressed Unicode to Unicode */
 290                 uint32_t c = ocu[i++];
 291                 if (cmp_id == 16)
 292                         c = (c << 8) | ocu[i++];
 293
 294                 len = nls->uni2char(c, &utf_o->u_name[utf_o->u_len],
 295                                     UDF_NAME_LEN - 2 - utf_o->u_len);
 296                 /* Valid character? */
 297                 if (len >= 0)
 298                         utf_o->u_len += len;
 299                 else
 300                         utf_o->u_name[utf_o->u_len++] = '?';
 301         }
 302         utf_o->u_cmpID = 8;
 303
 304         return utf_o->u_len;
 305 }
 306
 307 static int udf_NLStoCS0(struct nls_table *nls, dstring *ocu, struct ustr *uni,
 308                         int length)
 309 {
 310         int len;
 311         unsigned i, max_val;
 312         uint16_t uni_char;
 313         int u_len, u_ch;
 314
 315         memset(ocu, 0, sizeof(dstring) * length);
 316         ocu[0] = 8;
 317         max_val = 0xffU;
 318         u_ch = 1;
 319
 320 try_again:
 321         u_len = 0U;
 322         for (i = 0U; i < uni->u_len; i++) {
 323                 /* Name didn't fit? */
 324                 if (u_len + 1 + u_ch >= length)
 325                         return 0;
 326                 len = nls->char2uni(&uni->u_name[i], uni->u_len - i, &uni_char);
 327                 if (!len)
 328                         continue;
 329                 /* Invalid character, deal with it */
 330                 if (len < 0) {
 331                         len = 1;
 332                         uni_char = '?';
 333                 }
 334
 335                 if (uni_char > max_val) {
 336                         max_val = 0xffffU;
 337                         ocu[0] = (uint8_t)0x10U;
 338                         u_ch = 2;
 339                         goto try_again;
 340                 }
 341
 342                 if (max_val == 0xffffU)
 343                         ocu[++u_len] = (uint8_t)(uni_char >> 8);
 344                 ocu[++u_len] = (uint8_t)(uni_char & 0xffU);
 345                 i += len - 1;
 346         }
 347
 348         ocu[length - 1] = (uint8_t)u_len + 1;
 349         return u_len + 1;
 350 }
 351
 352 int udf_get_filename(struct super_block *sb, uint8_t *sname, int slen,
 353                      uint8_t *dname, int dlen)
 354 {
 355         struct ustr *filename, *unifilename;
 356         int len = 0;
 357
 358         filename = kmalloc(sizeof(struct ustr), GFP_NOFS);
 359         if (!filename)
 360                 return 0;
 361
 362         unifilename = kmalloc(sizeof(struct ustr), GFP_NOFS);
 363         if (!unifilename)
 364                 goto out1;
 365
 366         if (udf_build_ustr_exact(unifilename, sname, slen))
 367                 goto out2;
 368
 369         if (UDF_QUERY_FLAG(sb, UDF_FLAG_UTF8)) {
 370                 if (!udf_CS0toUTF8(filename, unifilename)) {
 371                         udf_debug("Failed in udf_get_filename: sname = %s\n",
 372                                   sname);
 373                         goto out2;
 374                 }
 375         } else if (UDF_QUERY_FLAG(sb, UDF_FLAG_NLS_MAP)) {
 376                 if (!udf_CS0toNLS(UDF_SB(sb)->s_nls_map, filename,
 377                                   unifilename)) {
 378                         udf_debug("Failed in udf_get_filename: sname = %s\n",
 379                                   sname);
 380                         goto out2;
 381                 }
 382         } else
 383                 goto out2;
 384
 385         len = udf_translate_to_linux(dname, dlen,
 386                                      filename->u_name, filename->u_len,
 387                                      unifilename->u_name, unifilename->u_len);
 388 out2:
 389         kfree(unifilename);
 390 out1:
 391         kfree(filename);
 392         return len;
 393 }
 394
 395 int udf_put_filename(struct super_block *sb, const uint8_t *sname,
 396                      uint8_t *dname, int flen)
 397 {
 398         struct ustr unifilename;
 399         int namelen;
 400
 401         if (!udf_char_to_ustr(&unifilename, sname, flen))
 402                 return 0;
 403
 404         if (UDF_QUERY_FLAG(sb, UDF_FLAG_UTF8)) {
 405                 namelen = udf_UTF8toCS0(dname, &unifilename, UDF_NAME_LEN);
 406                 if (!namelen)
 407                         return 0;
 408         } else if (UDF_QUERY_FLAG(sb, UDF_FLAG_NLS_MAP)) {
 409                 namelen = udf_NLStoCS0(UDF_SB(sb)->s_nls_map, dname,
 410                                         &unifilename, UDF_NAME_LEN);
 411                 if (!namelen)
 412                         return 0;
 413         } else
 414                 return 0;
 415
 416         return namelen;
 417 }
 418
 419 #define ILLEGAL_CHAR_MARK       '_'
 420 #define EXT_MARK                '.'
 421 #define CRC_MARK                '#'
 422 #define EXT_SIZE                5
 423 /* Number of chars we need to store generated CRC to make filename unique */
 424 #define CRC_LEN                 5
 425
 426 static int udf_translate_to_linux(uint8_t *newName, int newLen,
 427                                   uint8_t *udfName, int udfLen,
 428                                   uint8_t *fidName, int fidNameLen)
 429 {
 430         int index, newIndex = 0, needsCRC = 0;
 431         int extIndex = 0, newExtIndex = 0, hasExt = 0;
 432         unsigned short valueCRC;
 433         uint8_t curr;
 434         const uint8_t hexChar[] = "0123456789ABCDEF";
 435
 436         if (udfName[0] == '.' &&
 437             (udfLen == 1 || (udfLen == 2 && udfName[1] == '.'))) {
 438                 needsCRC = 1;
 439                 newIndex = udfLen;
 440                 memcpy(newName, udfName, udfLen);
 441         } else {
 442                 for (index = 0; index < udfLen; index++) {
 443                         curr = udfName[index];
 444                         if (curr == '/' || curr == 0) {
 445                                 needsCRC = 1;
 446                                 curr = ILLEGAL_CHAR_MARK;
 447                                 while (index + 1 < udfLen &&
 448                                                 (udfName[index + 1] == '/' ||
 449                                                  udfName[index + 1] == 0))
 450                                         index++;
 451                         }
 452                         if (curr == EXT_MARK &&
 453                                         (udfLen - index - 1) <= EXT_SIZE) {
 454                                 if (udfLen == index + 1)
 455                                         hasExt = 0;
 456                                 else {
 457                                         hasExt = 1;
 458                                         extIndex = index;
 459                                         newExtIndex = newIndex;
 460                                 }
 461                         }
 462                         if (newIndex < newLen)
 463                                 newName[newIndex++] = curr;
 464                         else
 465                                 needsCRC = 1;
 466                 }
 467         }
 468         if (needsCRC) {
 469                 uint8_t ext[EXT_SIZE];
 470                 int localExtIndex = 0;
 471
 472                 if (hasExt) {
 473                         int maxFilenameLen;
 474                         for (index = 0;
 475                              index < EXT_SIZE && extIndex + index + 1 < udfLen;
 476                              index++) {
 477                                 curr = udfName[extIndex + index + 1];
 478
 479                                 if (curr == '/' || curr == 0) {
 480                                         needsCRC = 1;
 481                                         curr = ILLEGAL_CHAR_MARK;
 482                                         while (extIndex + index + 2 < udfLen &&
 483                                               (index + 1 < EXT_SIZE &&
 484                                                 (udfName[extIndex + index + 2] == '/' ||
 485                                                  udfName[extIndex + index + 2] == 0)))
 486                                                 index++;
 487                                 }
 488                                 ext[localExtIndex++] = curr;
 489                         }
 490                         maxFilenameLen = newLen - CRC_LEN - localExtIndex;
 491                         if (newIndex > maxFilenameLen)
 492                                 newIndex = maxFilenameLen;
 493                         else
 494                                 newIndex = newExtIndex;
 495                 } else if (newIndex > newLen - CRC_LEN)
 496                         newIndex = newLen - CRC_LEN;
 497                 newName[newIndex++] = CRC_MARK;
 498                 valueCRC = crc_itu_t(0, fidName, fidNameLen);
 499                 newName[newIndex++] = hexChar[(valueCRC & 0xf000) >> 12];
 500                 newName[newIndex++] = hexChar[(valueCRC & 0x0f00) >> 8];
 501                 newName[newIndex++] = hexChar[(valueCRC & 0x00f0) >> 4];
 502                 newName[newIndex++] = hexChar[(valueCRC & 0x000f)];
 503
 504                 if (hasExt) {
 505                         newName[newIndex++] = EXT_MARK;
 506                         for (index = 0; index < localExtIndex; index++)
 507                                 newName[newIndex++] = ext[index];
 508                 }
 509         }
 510
 511         return newIndex;
 512 }