extras/volume_id/lib/util.c

   1 /*
   2  * volume_id - reads filesystem label and uuid
   3  *
   4  * Copyright (C) 2005-2007 Kay Sievers <kay.sievers@vrfy.org>
   5  *
   6  *      This program is free software; you can redistribute it and/or modify it
   7  *      under the terms of the GNU General Public License as published by the
   8  *      Free Software Foundation version 2 of the License.
   9  */
  10
  11 #ifndef _GNU_SOURCE
  12 #define _GNU_SOURCE 1
  13 #endif
  14
  15 #ifdef HAVE_CONFIG_H
  16 #  include <config.h>
  17 #endif
  18
  19 #include <stdio.h>
  20 #include <stdlib.h>
  21 #include <unistd.h>
  22 #include <string.h>
  23 #include <errno.h>
  24 #include <ctype.h>
  25 #include <fcntl.h>
  26 #include <sys/stat.h>
  27
  28 #include "libvolume_id.h"
  29 #include "util.h"
  30
  31 /* count of characters used to encode one unicode char */
  32 static int utf8_encoded_expected_len(const char *str)
  33 {
  34         unsigned char c = (unsigned char)str[0];
  35
  36         if (c < 0x80)
  37                 return 1;
  38         if ((c & 0xe0) == 0xc0)
  39                 return 2;
  40         if ((c & 0xf0) == 0xe0)
  41                 return 3;
  42         if ((c & 0xf8) == 0xf0)
  43                 return 4;
  44         if ((c & 0xfc) == 0xf8)
  45                 return 5;
  46         if ((c & 0xfe) == 0xfc)
  47                 return 6;
  48         return 0;
  49 }
  50
  51 /* decode one unicode char */
  52 static int utf8_encoded_to_unichar(const char *str)
  53 {
  54         int unichar;
  55         int len;
  56         int i;
  57
  58         len = utf8_encoded_expected_len(str);
  59         switch (len) {
  60         case 1:
  61                 return (int)str[0];
  62         case 2:
  63                 unichar = str[0] & 0x1f;
  64                 break;
  65         case 3:
  66                 unichar = (int)str[0] & 0x0f;
  67                 break;
  68         case 4:
  69                 unichar = (int)str[0] & 0x07;
  70                 break;
  71         case 5:
  72                 unichar = (int)str[0] & 0x03;
  73                 break;
  74         case 6:
  75                 unichar = (int)str[0] & 0x01;
  76                 break;
  77         default:
  78                 return -1;
  79         }
  80
  81         for (i = 1; i < len; i++) {
  82                 if (((int)str[i] & 0xc0) != 0x80)
  83                         return -1;
  84                 unichar <<= 6;
  85                 unichar |= (int)str[i] & 0x3f;
  86         }
  87
  88         return unichar;
  89 }
  90
  91 /* expected size used to encode one unicode char */
  92 static int utf8_unichar_to_encoded_len(int unichar)
  93 {
  94         if (unichar < 0x80)
  95                 return 1;
  96         if (unichar < 0x800)
  97                 return 2;
  98         if (unichar < 0x10000)
  99                 return 3;
 100         if (unichar < 0x200000)
 101                 return 4;
 102         if (unichar < 0x4000000)
 103                 return 5;
 104         return 6;
 105 }
 106
 107 /* check if unicode char has a valid numeric range */
 108 static int utf8_unichar_valid_range(int unichar)
 109 {
 110         if (unichar > 0x10ffff)
 111                 return 0;
 112         if ((unichar & 0xfffff800) == 0xd800)
 113                 return 0;
 114         if ((unichar > 0xfdcf) && (unichar < 0xfdf0))
 115                 return 0;
 116         if ((unichar & 0xffff) == 0xffff)
 117                 return 0;
 118         return 1;
 119 }
 120
 121 /* validate one encoded unicode char and return its length */
 122 int volume_id_utf8_encoded_valid_unichar(const char *str)
 123 {
 124         int len;
 125         int unichar;
 126         int i;
 127
 128         len = utf8_encoded_expected_len(str);
 129         if (len == 0)
 130                 return -1;
 131
 132         /* ascii is valid */
 133         if (len == 1)
 134                 return 1;
 135
 136         /* check if expected encoded chars are available */
 137         for (i = 0; i < len; i++)
 138                 if ((str[i] & 0x80) != 0x80)
 139                         return -1;
 140
 141         unichar = utf8_encoded_to_unichar(str);
 142
 143         /* check if encoded length matches encoded value */
 144         if (utf8_unichar_to_encoded_len(unichar) != len)
 145                 return -1;
 146
 147         /* check if value has valid range */
 148         if (!utf8_unichar_valid_range(unichar))
 149                 return -1;
 150
 151         return len;
 152 }
 153
 154 void volume_id_set_unicode16(char *str, size_t len, const uint8_t *buf, enum endian endianess, size_t count)
 155 {
 156         unsigned int i, j;
 157         uint16_t c;
 158
 159         j = 0;
 160         for (i = 0; i + 2 <= count; i += 2) {
 161                 if (endianess == LE)
 162                         c = (buf[i+1] << 8) | buf[i];
 163                 else
 164                         c = (buf[i] << 8) | buf[i+1];
 165                 if (c == 0) {
 166                         str[j] = '\0';
 167                         break;
 168                 } else if (c < 0x80) {
 169                         if (j+1 >= len)
 170                                 break;
 171                         str[j++] = (uint8_t) c;
 172                 } else if (c < 0x800) {
 173                         if (j+2 >= len)
 174                                 break;
 175                         str[j++] = (uint8_t) (0xc0 | (c >> 6));
 176                         str[j++] = (uint8_t) (0x80 | (c & 0x3f));
 177                 } else {
 178                         if (j+3 >= len)
 179                                 break;
 180                         str[j++] = (uint8_t) (0xe0 | (c >> 12));
 181                         str[j++] = (uint8_t) (0x80 | ((c >> 6) & 0x3f));
 182                         str[j++] = (uint8_t) (0x80 | (c & 0x3f));
 183                 }
 184         }
 185         str[j] = '\0';
 186 }
 187
 188 static char *usage_to_string(enum volume_id_usage usage_id)
 189 {
 190         switch (usage_id) {
 191         case VOLUME_ID_FILESYSTEM:
 192                 return "filesystem";
 193         case VOLUME_ID_OTHER:
 194                 return "other";
 195         case VOLUME_ID_RAID:
 196                 return "raid";
 197         case VOLUME_ID_DISKLABEL:
 198                 return "disklabel";
 199         case VOLUME_ID_CRYPTO:
 200                 return "crypto";
 201         case VOLUME_ID_UNPROBED:
 202                 return "unprobed";
 203         case VOLUME_ID_UNUSED:
 204                 return "unused";
 205         }
 206         return NULL;
 207 }
 208
 209 void volume_id_set_usage(struct volume_id *id, enum volume_id_usage usage_id)
 210 {
 211         id->usage_id = usage_id;
 212         id->usage = usage_to_string(usage_id);
 213 }
 214
 215 void volume_id_set_label_raw(struct volume_id *id, const uint8_t *buf, size_t count)
 216 {
 217         memcpy(id->label_raw, buf, count);
 218         id->label_raw_len = count;
 219 }
 220
 221 void volume_id_set_label_string(struct volume_id *id, const uint8_t *buf, size_t count)
 222 {
 223         unsigned int i;
 224
 225         memcpy(id->label, buf, count);
 226
 227         /* remove trailing whitespace */
 228         i = strnlen(id->label, count);
 229         while (i--) {
 230                 if (!isspace(id->label[i]))
 231                         break;
 232         }
 233         id->label[i+1] = '\0';
 234 }
 235
 236 void volume_id_set_label_unicode16(struct volume_id *id, const uint8_t *buf, enum endian endianess, size_t count)
 237 {
 238          volume_id_set_unicode16(id->label, sizeof(id->label), buf, endianess, count);
 239 }
 240
 241 void volume_id_set_uuid(struct volume_id *id, const uint8_t *buf, size_t len, enum uuid_format format)
 242 {
 243         unsigned int i;
 244         unsigned int count = 0;
 245
 246         if (len > sizeof(id->uuid_raw))
 247                 len = sizeof(id->uuid_raw);
 248
 249         switch(format) {
 250         case UUID_STRING:
 251                 count = len;
 252                 break;
 253         case UUID_HEX_STRING:
 254                 count = len;
 255                 break;
 256         case UUID_DOS:
 257                 count = 4;
 258                 break;
 259         case UUID_64BIT_LE:
 260         case UUID_64BIT_BE:
 261                 count = 8;
 262                 break;
 263         case UUID_DCE:
 264                 count = 16;
 265                 break;
 266         case UUID_FOURINT:
 267                 count = 35;
 268                 break;
 269         }
 270         memcpy(id->uuid_raw, buf, count);
 271         id->uuid_raw_len = count;
 272
 273         /* if set, create string in the same format, the native platform uses */
 274         for (i = 0; i < count; i++)
 275                 if (buf[i] != 0)
 276                         goto set;
 277         return;
 278
 279 set:
 280         switch(format) {
 281         case UUID_DOS:
 282                 sprintf(id->uuid, "%02X%02X-%02X%02X",
 283                         buf[3], buf[2], buf[1], buf[0]);
 284                 break;
 285         case UUID_64BIT_LE:
 286                 sprintf(id->uuid,"%02X%02X%02X%02X%02X%02X%02X%02X",
 287                         buf[7], buf[6], buf[5], buf[4],
 288                         buf[3], buf[2], buf[1], buf[0]);
 289                 break;
 290         case UUID_64BIT_BE:
 291                 sprintf(id->uuid,"%02X%02X%02X%02X%02X%02X%02X%02X",
 292                         buf[0], buf[1], buf[2], buf[3],
 293                         buf[4], buf[5], buf[6], buf[7]);
 294                 break;
 295         case UUID_DCE:
 296                 sprintf(id->uuid,
 297                         "%02x%02x%02x%02x-%02x%02x-%02x%02x-%02x%02x-%02x%02x%02x%02x%02x%02x",
 298                         buf[0], buf[1], buf[2], buf[3],
 299                         buf[4], buf[5],
 300                         buf[6], buf[7],
 301                         buf[8], buf[9],
 302                         buf[10], buf[11], buf[12], buf[13], buf[14],buf[15]);
 303                 break;
 304         case UUID_HEX_STRING:
 305                 /* translate A..F to a..f */
 306                 memcpy(id->uuid, buf, count);
 307                 for (i = 0; i < count; i++)
 308                         if (id->uuid[i] >= 'A' && id->uuid[i] <= 'F')
 309                                 id->uuid[i] = (id->uuid[i] - 'A') + 'a';
 310                 id->uuid[count] = '\0';
 311                 break;
 312         case UUID_STRING:
 313                 memcpy(id->uuid, buf, count);
 314                 id->uuid[count] = '\0';
 315                 break;
 316         case UUID_FOURINT:
 317                 sprintf(id->uuid,
 318                         "%02x%02x%02x%02x:%02x%02x%02x%02x:%02x%02x%02x%02x:%02x%02x%02x%02x",
 319                         buf[0], buf[1], buf[2], buf[3],
 320                         buf[4], buf[5], buf[6], buf[7],
 321                         buf[8], buf[9], buf[10], buf[11],
 322                         buf[12], buf[13], buf[14],buf[15]);
 323                 break;
 324         }
 325 }
 326
 327 uint8_t *volume_id_get_buffer(struct volume_id *id, uint64_t off, size_t len)
 328 {
 329         ssize_t buf_len;
 330
 331         info("get buffer off 0x%llx(%llu), len 0x%zx", (unsigned long long) off, (unsigned long long) off, len);
 332         /* check if requested area fits in superblock buffer */
 333         if (off + len <= SB_BUFFER_SIZE) {
 334                 if (id->sbbuf == NULL) {
 335                         id->sbbuf = malloc(SB_BUFFER_SIZE);
 336                         if (id->sbbuf == NULL) {
 337                                 dbg("error malloc");
 338                                 return NULL;
 339                         }
 340                 }
 341
 342                 /* check if we need to read */
 343                 if ((off + len) > id->sbbuf_len) {
 344                         info("read sbbuf len:0x%llx", (unsigned long long) (off + len));
 345                         if (lseek(id->fd, 0, SEEK_SET) < 0) {
 346                                 dbg("lseek failed (%s)", strerror(errno));
 347                                 return NULL;
 348                         }
 349                         buf_len = read(id->fd, id->sbbuf, off + len);
 350                         if (buf_len < 0) {
 351                                 dbg("read failed (%s)", strerror(errno));
 352                                 return NULL;
 353                         }
 354                         dbg("got 0x%zx (%zi) bytes", buf_len, buf_len);
 355                         id->sbbuf_len = buf_len;
 356                         if ((size_t)buf_len < off + len) {
 357                                 dbg("requested 0x%zx bytes, got only 0x%zx bytes", len, buf_len);
 358                                 return NULL;
 359                         }
 360                 }
 361
 362                 return &(id->sbbuf[off]);
 363         } else {
 364                 if (len > SEEK_BUFFER_SIZE) {
 365                         dbg("seek buffer too small %d", SEEK_BUFFER_SIZE);
 366                         return NULL;
 367                 }
 368
 369                 /* get seek buffer */
 370                 if (id->seekbuf == NULL) {
 371                         id->seekbuf = malloc(SEEK_BUFFER_SIZE);
 372                         if (id->seekbuf == NULL) {
 373                                 dbg("error malloc");
 374                                 return NULL;
 375                         }
 376                 }
 377
 378                 /* check if we need to read */
 379                 if ((off < id->seekbuf_off) || ((off + len) > (id->seekbuf_off + id->seekbuf_len))) {
 380                         info("read seekbuf off:0x%llx len:0x%zx", (unsigned long long) off, len);
 381                         if (lseek(id->fd, off, SEEK_SET) < 0) {
 382                                 dbg("lseek failed (%s)", strerror(errno));
 383                                 return NULL;
 384                         }
 385                         buf_len = read(id->fd, id->seekbuf, len);
 386                         if (buf_len < 0) {
 387                                 dbg("read failed (%s)", strerror(errno));
 388                                 return NULL;
 389                         }
 390                         dbg("got 0x%zx (%zi) bytes", buf_len, buf_len);
 391                         id->seekbuf_off = off;
 392                         id->seekbuf_len = buf_len;
 393                         if ((size_t)buf_len < len) {
 394                                 dbg("requested 0x%zx bytes, got only 0x%zx bytes", len, buf_len);
 395                                 return NULL;
 396                         }
 397                 }
 398
 399                 return &(id->seekbuf[off - id->seekbuf_off]);
 400         }
 401 }
 402
 403 void volume_id_free_buffer(struct volume_id *id)
 404 {
 405         if (id->sbbuf != NULL) {
 406                 free(id->sbbuf);
 407                 id->sbbuf = NULL;
 408                 id->sbbuf_len = 0;
 409         }
 410         if (id->seekbuf != NULL) {
 411                 free(id->seekbuf);
 412                 id->seekbuf = NULL;
 413                 id->seekbuf_len = 0;
 414         }
 415 }