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 void udf_build_ustr_exact(struct ustr *dest, dstring *ptr, int exactsize)
  72 {
  73         memset(dest, 0, sizeof(struct ustr));
  74         dest->u_cmpID = ptr[0];
  75         dest->u_len = exactsize - 1;
  76         memcpy(dest->u_name, ptr + 1, exactsize - 1);
  77 }
  78
  79 /*
  80  * udf_CS0toUTF8
  81  *
  82  * PURPOSE
  83  *      Convert OSTA Compressed Unicode to the UTF-8 equivalent.
  84  *
  85  * PRE-CONDITIONS
  86  *      utf                     Pointer to UTF-8 output buffer.
  87  *      ocu                     Pointer to OSTA Compressed Unicode input buffer
  88  *                              of size UDF_NAME_LEN bytes.
  89  *                              both of type "struct ustr *"
  90  *
  91  * POST-CONDITIONS
  92  *      <return>                >= 0 on success.
  93  *
  94  * HISTORY
  95  *      November 12, 1997 - Andrew E. Mileski
  96  *      Written, tested, and released.
  97  */
  98 int udf_CS0toUTF8(struct ustr *utf_o, const struct ustr *ocu_i)
  99 {
 100         const uint8_t *ocu;
 101         uint8_t cmp_id, ocu_len;
 102         int i;
 103
 104         ocu_len = ocu_i->u_len;
 105         if (ocu_len == 0) {
 106                 memset(utf_o, 0, sizeof(struct ustr));
 107                 return 0;
 108         }
 109
 110         cmp_id = ocu_i->u_cmpID;
 111         if (cmp_id != 8 && cmp_id != 16) {
 112                 memset(utf_o, 0, sizeof(struct ustr));
 113                 pr_err("unknown compression code (%d) stri=%s\n",
 114                        cmp_id, ocu_i->u_name);
 115                 return -EINVAL;
 116         }
 117
 118         ocu = ocu_i->u_name;
 119         utf_o->u_len = 0;
 120         for (i = 0; (i < ocu_len) && (utf_o->u_len <= (UDF_NAME_LEN - 3));) {
 121
 122                 /* Expand OSTA compressed Unicode to Unicode */
 123                 uint32_t c = ocu[i++];
 124                 if (cmp_id == 16)
 125                         c = (c << 8) | ocu[i++];
 126
 127                 /* Compress Unicode to UTF-8 */
 128                 if (c < 0x80U)
 129                         utf_o->u_name[utf_o->u_len++] = (uint8_t)c;
 130                 else if (c < 0x800U) {
 131                         if (utf_o->u_len > (UDF_NAME_LEN - 4))
 132                                 break;
 133                         utf_o->u_name[utf_o->u_len++] =
 134                                                 (uint8_t)(0xc0 | (c >> 6));
 135                         utf_o->u_name[utf_o->u_len++] =
 136                                                 (uint8_t)(0x80 | (c & 0x3f));
 137                 } else {
 138                         if (utf_o->u_len > (UDF_NAME_LEN - 5))
 139                                 break;
 140                         utf_o->u_name[utf_o->u_len++] =
 141                                                 (uint8_t)(0xe0 | (c >> 12));
 142                         utf_o->u_name[utf_o->u_len++] =
 143                                                 (uint8_t)(0x80 |
 144                                                           ((c >> 6) & 0x3f));
 145                         utf_o->u_name[utf_o->u_len++] =
 146                                                 (uint8_t)(0x80 | (c & 0x3f));
 147                 }
 148         }
 149         utf_o->u_cmpID = 8;
 150
 151         return utf_o->u_len;
 152 }
 153
 154 /*
 155  *
 156  * udf_UTF8toCS0
 157  *
 158  * PURPOSE
 159  *      Convert UTF-8 to the OSTA Compressed Unicode equivalent.
 160  *
 161  * DESCRIPTION
 162  *      This routine is only called by udf_lookup().
 163  *
 164  * PRE-CONDITIONS
 165  *      ocu                     Pointer to OSTA Compressed Unicode output
 166  *                              buffer of size UDF_NAME_LEN bytes.
 167  *      utf                     Pointer to UTF-8 input buffer.
 168  *      utf_len                 Length of UTF-8 input buffer in bytes.
 169  *
 170  * POST-CONDITIONS
 171  *      <return>                Zero on success.
 172  *
 173  * HISTORY
 174  *      November 12, 1997 - Andrew E. Mileski
 175  *      Written, tested, and released.
 176  */
 177 static int udf_UTF8toCS0(dstring *ocu, struct ustr *utf, int length)
 178 {
 179         unsigned c, i, max_val, utf_char;
 180         int utf_cnt, u_len, u_ch;
 181
 182         memset(ocu, 0, sizeof(dstring) * length);
 183         ocu[0] = 8;
 184         max_val = 0xffU;
 185         u_ch = 1;
 186
 187 try_again:
 188         u_len = 0U;
 189         utf_char = 0U;
 190         utf_cnt = 0U;
 191         for (i = 0U; i < utf->u_len; i++) {
 192                 /* Name didn't fit? */
 193                 if (u_len + 1 + u_ch >= length)
 194                         return 0;
 195
 196                 c = (uint8_t)utf->u_name[i];
 197
 198                 /* Complete a multi-byte UTF-8 character */
 199                 if (utf_cnt) {
 200                         utf_char = (utf_char << 6) | (c & 0x3fU);
 201                         if (--utf_cnt)
 202                                 continue;
 203                 } else {
 204                         /* Check for a multi-byte UTF-8 character */
 205                         if (c & 0x80U) {
 206                                 /* Start a multi-byte UTF-8 character */
 207                                 if ((c & 0xe0U) == 0xc0U) {
 208                                         utf_char = c & 0x1fU;
 209                                         utf_cnt = 1;
 210                                 } else if ((c & 0xf0U) == 0xe0U) {
 211                                         utf_char = c & 0x0fU;
 212                                         utf_cnt = 2;
 213                                 } else if ((c & 0xf8U) == 0xf0U) {
 214                                         utf_char = c & 0x07U;
 215                                         utf_cnt = 3;
 216                                 } else if ((c & 0xfcU) == 0xf8U) {
 217                                         utf_char = c & 0x03U;
 218                                         utf_cnt = 4;
 219                                 } else if ((c & 0xfeU) == 0xfcU) {
 220                                         utf_char = c & 0x01U;
 221                                         utf_cnt = 5;
 222                                 } else {
 223                                         goto error_out;
 224                                 }
 225                                 continue;
 226                         } else {
 227                                 /* Single byte UTF-8 character (most common) */
 228                                 utf_char = c;
 229                         }
 230                 }
 231
 232                 /* Choose no compression if necessary */
 233                 if (utf_char > max_val) {
 234                         if (max_val == 0xffU) {
 235                                 max_val = 0xffffU;
 236                                 ocu[0] = (uint8_t)0x10U;
 237                                 u_ch = 2;
 238                                 goto try_again;
 239                         }
 240                         goto error_out;
 241                 }
 242
 243                 if (max_val == 0xffffU)
 244                         ocu[++u_len] = (uint8_t)(utf_char >> 8);
 245                 ocu[++u_len] = (uint8_t)(utf_char & 0xffU);
 246         }
 247
 248         if (utf_cnt) {
 249 error_out:
 250                 ocu[++u_len] = '?';
 251                 printk(KERN_DEBUG pr_fmt("bad UTF-8 character\n"));
 252         }
 253
 254         ocu[length - 1] = (uint8_t)u_len + 1;
 255
 256         return u_len + 1;
 257 }
 258
 259 static int udf_CS0toNLS(struct nls_table *nls, struct ustr *utf_o,
 260                         const struct ustr *ocu_i)
 261 {
 262         const uint8_t *ocu;
 263         uint8_t cmp_id, ocu_len;
 264         int i, len;
 265
 266
 267         ocu_len = ocu_i->u_len;
 268         if (ocu_len == 0) {
 269                 memset(utf_o, 0, sizeof(struct ustr));
 270                 return 0;
 271         }
 272
 273         cmp_id = ocu_i->u_cmpID;
 274         if (cmp_id != 8 && cmp_id != 16) {
 275                 memset(utf_o, 0, sizeof(struct ustr));
 276                 pr_err("unknown compression code (%d) stri=%s\n",
 277                        cmp_id, ocu_i->u_name);
 278                 return -EINVAL;
 279         }
 280
 281         ocu = ocu_i->u_name;
 282         utf_o->u_len = 0;
 283         for (i = 0; (i < ocu_len) && (utf_o->u_len <= (UDF_NAME_LEN - 3));) {
 284                 /* Expand OSTA compressed Unicode to Unicode */
 285                 uint32_t c = ocu[i++];
 286                 if (cmp_id == 16)
 287                         c = (c << 8) | ocu[i++];
 288
 289                 len = nls->uni2char(c, &utf_o->u_name[utf_o->u_len],
 290                                     UDF_NAME_LEN - 2 - utf_o->u_len);
 291                 /* Valid character? */
 292                 if (len >= 0)
 293                         utf_o->u_len += len;
 294                 else
 295                         utf_o->u_name[utf_o->u_len++] = '?';
 296         }
 297         utf_o->u_cmpID = 8;
 298
 299         return utf_o->u_len;
 300 }
 301
 302 static int udf_NLStoCS0(struct nls_table *nls, dstring *ocu, struct ustr *uni,
 303                         int length)
 304 {
 305         int len;
 306         unsigned i, max_val;
 307         uint16_t uni_char;
 308         int u_len, u_ch;
 309
 310         memset(ocu, 0, sizeof(dstring) * length);
 311         ocu[0] = 8;
 312         max_val = 0xffU;
 313         u_ch = 1;
 314
 315 try_again:
 316         u_len = 0U;
 317         for (i = 0U; i < uni->u_len; i++) {
 318                 /* Name didn't fit? */
 319                 if (u_len + 1 + u_ch >= length)
 320                         return 0;
 321                 len = nls->char2uni(&uni->u_name[i], uni->u_len - i, &uni_char);
 322                 if (!len)
 323                         continue;
 324                 /* Invalid character, deal with it */
 325                 if (len < 0) {
 326                         len = 1;
 327                         uni_char = '?';
 328                 }
 329
 330                 if (uni_char > max_val) {
 331                         max_val = 0xffffU;
 332                         ocu[0] = (uint8_t)0x10U;
 333                         u_ch = 2;
 334                         goto try_again;
 335                 }
 336
 337                 if (max_val == 0xffffU)
 338                         ocu[++u_len] = (uint8_t)(uni_char >> 8);
 339                 ocu[++u_len] = (uint8_t)(uni_char & 0xffU);
 340                 i += len - 1;
 341         }
 342
 343         ocu[length - 1] = (uint8_t)u_len + 1;
 344         return u_len + 1;
 345 }
 346
 347 int udf_get_filename(struct super_block *sb, uint8_t *sname, int slen,
 348                      uint8_t *dname, int dlen)
 349 {
 350         struct ustr *filename, *unifilename;
 351         int ret;
 352
 353         if (!slen)
 354                 return -EIO;
 355
 356         filename = kmalloc(sizeof(struct ustr), GFP_NOFS);
 357         if (!filename)
 358                 return -ENOMEM;
 359
 360         unifilename = kmalloc(sizeof(struct ustr), GFP_NOFS);
 361         if (!unifilename) {
 362                 ret = -ENOMEM;
 363                 goto out1;
 364         }
 365
 366         udf_build_ustr_exact(unifilename, sname, slen);
 367         if (UDF_QUERY_FLAG(sb, UDF_FLAG_UTF8)) {
 368                 ret = udf_CS0toUTF8(filename, unifilename);
 369                 if (ret < 0) {
 370                         udf_debug("Failed in udf_get_filename: sname = %s\n",
 371                                   sname);
 372                         goto out2;
 373                 }
 374         } else if (UDF_QUERY_FLAG(sb, UDF_FLAG_NLS_MAP)) {
 375                 ret = udf_CS0toNLS(UDF_SB(sb)->s_nls_map, filename,
 376                                    unifilename);
 377                 if (ret < 0) {
 378                         udf_debug("Failed in udf_get_filename: sname = %s\n",
 379                                   sname);
 380                         goto out2;
 381                 }
 382         } else
 383                 BUG();
 384
 385         ret = udf_translate_to_linux(dname, dlen,
 386                                      filename->u_name, filename->u_len,
 387                                      unifilename->u_name, unifilename->u_len);
 388         /* Zero length filename isn't valid... */
 389         if (ret == 0)
 390                 ret = -EINVAL;
 391 out2:
 392         kfree(unifilename);
 393 out1:
 394         kfree(filename);
 395         return ret;
 396 }
 397
 398 int udf_put_filename(struct super_block *sb, const uint8_t *sname,
 399                      uint8_t *dname, int flen)
 400 {
 401         struct ustr unifilename;
 402         int namelen;
 403
 404         if (!udf_char_to_ustr(&unifilename, sname, flen))
 405                 return 0;
 406
 407         if (UDF_QUERY_FLAG(sb, UDF_FLAG_UTF8)) {
 408                 namelen = udf_UTF8toCS0(dname, &unifilename, UDF_NAME_LEN);
 409                 if (!namelen)
 410                         return 0;
 411         } else if (UDF_QUERY_FLAG(sb, UDF_FLAG_NLS_MAP)) {
 412                 namelen = udf_NLStoCS0(UDF_SB(sb)->s_nls_map, dname,
 413                                         &unifilename, UDF_NAME_LEN);
 414                 if (!namelen)
 415                         return 0;
 416         } else
 417                 return 0;
 418
 419         return namelen;
 420 }
 421
 422 #define ILLEGAL_CHAR_MARK       '_'
 423 #define EXT_MARK                '.'
 424 #define CRC_MARK                '#'
 425 #define EXT_SIZE                5
 426 /* Number of chars we need to store generated CRC to make filename unique */
 427 #define CRC_LEN                 5
 428
 429 static int udf_translate_to_linux(uint8_t *newName, int newLen,
 430                                   uint8_t *udfName, int udfLen,
 431                                   uint8_t *fidName, int fidNameLen)
 432 {
 433         int index, newIndex = 0, needsCRC = 0;
 434         int extIndex = 0, newExtIndex = 0, hasExt = 0;
 435         unsigned short valueCRC;
 436         uint8_t curr;
 437
 438         if (udfName[0] == '.' &&
 439             (udfLen == 1 || (udfLen == 2 && udfName[1] == '.'))) {
 440                 needsCRC = 1;
 441                 newIndex = udfLen;
 442                 memcpy(newName, udfName, udfLen);
 443         } else {
 444                 for (index = 0; index < udfLen; index++) {
 445                         curr = udfName[index];
 446                         if (curr == '/' || curr == 0) {
 447                                 needsCRC = 1;
 448                                 curr = ILLEGAL_CHAR_MARK;
 449                                 while (index + 1 < udfLen &&
 450                                                 (udfName[index + 1] == '/' ||
 451                                                  udfName[index + 1] == 0))
 452                                         index++;
 453                         }
 454                         if (curr == EXT_MARK &&
 455                                         (udfLen - index - 1) <= EXT_SIZE) {
 456                                 if (udfLen == index + 1)
 457                                         hasExt = 0;
 458                                 else {
 459                                         hasExt = 1;
 460                                         extIndex = index;
 461                                         newExtIndex = newIndex;
 462                                 }
 463                         }
 464                         if (newIndex < newLen)
 465                                 newName[newIndex++] = curr;
 466                         else
 467                                 needsCRC = 1;
 468                 }
 469         }
 470         if (needsCRC) {
 471                 uint8_t ext[EXT_SIZE];
 472                 int localExtIndex = 0;
 473
 474                 if (hasExt) {
 475                         int maxFilenameLen;
 476                         for (index = 0;
 477                              index < EXT_SIZE && extIndex + index + 1 < udfLen;
 478                              index++) {
 479                                 curr = udfName[extIndex + index + 1];
 480
 481                                 if (curr == '/' || curr == 0) {
 482                                         needsCRC = 1;
 483                                         curr = ILLEGAL_CHAR_MARK;
 484                                         while (extIndex + index + 2 < udfLen &&
 485                                               (index + 1 < EXT_SIZE &&
 486                                                 (udfName[extIndex + index + 2] == '/' ||
 487                                                  udfName[extIndex + index + 2] == 0)))
 488                                                 index++;
 489                                 }
 490                                 ext[localExtIndex++] = curr;
 491                         }
 492                         maxFilenameLen = newLen - CRC_LEN - localExtIndex;
 493                         if (newIndex > maxFilenameLen)
 494                                 newIndex = maxFilenameLen;
 495                         else
 496                                 newIndex = newExtIndex;
 497                 } else if (newIndex > newLen - CRC_LEN)
 498                         newIndex = newLen - CRC_LEN;
 499                 newName[newIndex++] = CRC_MARK;
 500                 valueCRC = crc_itu_t(0, fidName, fidNameLen);
 501                 newName[newIndex++] = hex_asc_upper_hi(valueCRC >> 8);
 502                 newName[newIndex++] = hex_asc_upper_lo(valueCRC >> 8);
 503                 newName[newIndex++] = hex_asc_upper_hi(valueCRC);
 504                 newName[newIndex++] = hex_asc_upper_lo(valueCRC);
 505
 506                 if (hasExt) {
 507                         newName[newIndex++] = EXT_MARK;
 508                         for (index = 0; index < localExtIndex; index++)
 509                                 newName[newIndex++] = ext[index];
 510                 }
 511         }
 512
 513         return newIndex;
 514 }