libblkid/src/read.c

   1 /*
   2  * read.c - read the blkid cache from disk, to avoid scanning all devices
   3  *
   4  * Copyright (C) 2001, 2003 Theodore Y. Ts'o
   5  * Copyright (C) 2001 Andreas Dilger
   6  *
   7  * %Begin-Header%
   8  * This file may be redistributed under the terms of the
   9  * GNU Lesser General Public License.
  10  * %End-Header%
  11  */
  12
  13
  14 #include <stdio.h>
  15 #include <ctype.h>
  16 #include <string.h>
  17 #include <time.h>
  18 #include <sys/types.h>
  19 #include <sys/stat.h>
  20 #include <fcntl.h>
  21 #include <unistd.h>
  22 #ifdef HAVE_ERRNO_H
  23 #include <errno.h>
  24 #endif
  25
  26 #include "blkidP.h"
  27
  28 #ifdef HAVE_STDLIB_H
  29 # ifndef _XOPEN_SOURCE
  30 #  define _XOPEN_SOURCE 600     /* for inclusion of strtoull */
  31 # endif
  32 # include <stdlib.h>
  33 #endif
  34
  35 /*
  36  * File format:
  37  *
  38  *      <device [<NAME="value"> ...]>device_name</device>
  39  *
  40  *      The following tags are required for each entry:
  41  *      <ID="id">       unique (within this file) ID number of this device
  42  *      <TIME="sec.usec"> (time_t and suseconds_t) time this entry was last
  43  *                       read from disk
  44  *      <TYPE="type">   (detected) type of filesystem/data for this partition
  45  *
  46  *      The following tags may be present, depending on the device contents
  47  *      <LABEL="label"> (user supplied) label (volume name, etc)
  48  *      <UUID="uuid">   (generated) universally unique identifier (serial no)
  49  */
  50
  51 static char *skip_over_blank(char *cp)
  52 {
  53         while (*cp && isspace(*cp))
  54                 cp++;
  55         return cp;
  56 }
  57
  58 static char *skip_over_word(char *cp)
  59 {
  60         char ch;
  61
  62         while ((ch = *cp)) {
  63                 /* If we see a backslash, skip the next character */
  64                 if (ch == '\\') {
  65                         cp++;
  66                         if (*cp == '\0')
  67                                 break;
  68                         cp++;
  69                         continue;
  70                 }
  71                 if (isspace(ch) || ch == '<' || ch == '>')
  72                         break;
  73                 cp++;
  74         }
  75         return cp;
  76 }
  77
  78 static char *strip_line(char *line)
  79 {
  80         char    *p;
  81
  82         line = skip_over_blank(line);
  83
  84         p = line + strlen(line) - 1;
  85
  86         while (*line) {
  87                 if (isspace(*p))
  88                         *p-- = '\0';
  89                 else
  90                         break;
  91         }
  92
  93         return line;
  94 }
  95
  96 #if 0
  97 static char *parse_word(char **buf)
  98 {
  99         char *word, *next;
 100
 101         word = *buf;
 102         if (*word == '\0')
 103                 return NULL;
 104
 105         word = skip_over_blank(word);
 106         next = skip_over_word(word);
 107         if (*next) {
 108                 char *end = next - 1;
 109                 if (*end == '"' || *end == '\'')
 110                         *end = '\0';
 111                 *next++ = '\0';
 112         }
 113         *buf = next;
 114
 115         if (*word == '"' || *word == '\'')
 116                 word++;
 117         return word;
 118 }
 119 #endif
 120
 121 /*
 122  * Start parsing a new line from the cache.
 123  *
 124  * line starts with "<device" return 1 -> continue parsing line
 125  * line starts with "<foo", empty, or # return 0 -> skip line
 126  * line starts with other, return -BLKID_ERR_CACHE -> error
 127  */
 128 static int parse_start(char **cp)
 129 {
 130         char *p;
 131
 132         p = strip_line(*cp);
 133
 134         /* Skip comment or blank lines.  We can't just NUL the first '#' char,
 135          * in case it is inside quotes, or escaped.
 136          */
 137         if (*p == '\0' || *p == '#')
 138                 return 0;
 139
 140         if (!strncmp(p, "<device", 7)) {
 141                 DBG(READ, ul_debug("found device header: %8s", p));
 142                 p += 7;
 143
 144                 *cp = p;
 145                 return 1;
 146         }
 147
 148         if (*p == '<')
 149                 return 0;
 150
 151         return -BLKID_ERR_CACHE;
 152 }
 153
 154 /* Consume the remaining XML on the line (cosmetic only) */
 155 static int parse_end(char **cp)
 156 {
 157         *cp = skip_over_blank(*cp);
 158
 159         if (!strncmp(*cp, "</device>", 9)) {
 160                 DBG(READ, ul_debug("found device trailer %9s", *cp));
 161                 *cp += 9;
 162                 return 0;
 163         }
 164
 165         return -BLKID_ERR_CACHE;
 166 }
 167
 168 /*
 169  * Allocate a new device struct with device name filled in.  Will handle
 170  * finding the device on lines of the form:
 171  * <device foo=bar>devname</device>
 172  * <device>devname<foo>bar</foo></device>
 173  */
 174 static int parse_dev(blkid_cache cache, blkid_dev *dev, char **cp)
 175 {
 176         char *start, *tmp, *end, *name;
 177         int ret;
 178
 179         if ((ret = parse_start(cp)) <= 0)
 180                 return ret;
 181
 182         start = tmp = strchr(*cp, '>');
 183         if (!start) {
 184                 DBG(READ, ul_debug("blkid: short line parsing dev: %s", *cp));
 185                 return -BLKID_ERR_CACHE;
 186         }
 187         start = skip_over_blank(start + 1);
 188         end = skip_over_word(start);
 189
 190         DBG(READ, ul_debug("device should be %*s",
 191                                (int)(end - start), start));
 192
 193         if (**cp == '>')
 194                 *cp = end;
 195         else
 196                 (*cp)++;
 197
 198         *tmp = '\0';
 199
 200         if (!(tmp = strrchr(end, '<')) || parse_end(&tmp) < 0) {
 201                 DBG(READ, ul_debug("blkid: missing </device> ending: %s", end));
 202         } else if (tmp)
 203                 *tmp = '\0';
 204
 205         if (end - start <= 1) {
 206                 DBG(READ, ul_debug("blkid: empty device name: %s", *cp));
 207                 return -BLKID_ERR_CACHE;
 208         }
 209
 210         name = strndup(start, end - start);
 211         if (name == NULL)
 212                 return -BLKID_ERR_MEM;
 213
 214         DBG(READ, ul_debug("found dev %s", name));
 215
 216         if (!(*dev = blkid_get_dev(cache, name, BLKID_DEV_CREATE))) {
 217                 free(name);
 218                 return -BLKID_ERR_MEM;
 219         }
 220
 221         free(name);
 222         return 1;
 223 }
 224
 225 /*
 226  * Extract a tag of the form NAME="value" from the line.
 227  */
 228 static int parse_token(char **name, char **value, char **cp)
 229 {
 230         char *end;
 231
 232         if (!name || !value || !cp)
 233                 return -BLKID_ERR_PARAM;
 234
 235         if (!(*value = strchr(*cp, '=')))
 236                 return 0;
 237
 238         **value = '\0';
 239         *name = strip_line(*cp);
 240         *value = skip_over_blank(*value + 1);
 241
 242         if (**value == '"') {
 243                 char *p = end = *value + 1;
 244
 245                 /* convert 'foo\"bar'  to 'foo"bar' */
 246                 while (*p) {
 247                         if (*p == '\\') {
 248                                 p++;
 249                                 *end = *p;
 250                         } else {
 251                                 *end = *p;
 252                                 if (*p == '"')
 253                                         break;
 254                         }
 255                         p++;
 256                         end++;
 257                 }
 258
 259                 if (*end != '"') {
 260                         DBG(READ, ul_debug("unbalanced quotes at: %s", *value));
 261                         *cp = *value;
 262                         return -BLKID_ERR_CACHE;
 263                 }
 264                 (*value)++;
 265                 *end = '\0';
 266                 end = ++p;
 267         } else {
 268                 end = skip_over_word(*value);
 269                 if (*end) {
 270                         *end = '\0';
 271                         end++;
 272                 }
 273         }
 274         *cp = end;
 275
 276         return 1;
 277 }
 278
 279 /*
 280  * Extract a tag from the line.
 281  *
 282  * Return 1 if a valid tag was found.
 283  * Return 0 if no tag found.
 284  * Return -ve error code.
 285  */
 286 static int parse_tag(blkid_cache cache, blkid_dev dev, char **cp)
 287 {
 288         char *name = NULL;
 289         char *value = NULL;
 290         int ret;
 291
 292         if (!cache || !dev)
 293                 return -BLKID_ERR_PARAM;
 294
 295         if ((ret = parse_token(&name, &value, cp)) <= 0)
 296                 return ret;
 297
 298         DBG(READ, ul_debug("tag: %s=\"%s\"", name, value));
 299
 300         /* Some tags are stored directly in the device struct */
 301         if (!strcmp(name, "DEVNO"))
 302                 dev->bid_devno = strtoull(value, NULL, 0);
 303         else if (!strcmp(name, "PRI"))
 304                 dev->bid_pri = strtol(value, NULL, 0);
 305         else if (!strcmp(name, "TIME")) {
 306                 char *end = NULL;
 307                 dev->bid_time = strtoull(value, &end, 0);
 308                 if (end && *end == '.')
 309                         dev->bid_utime = strtoull(end + 1, NULL, 0);
 310         } else
 311                 ret = blkid_set_tag(dev, name, value, strlen(value));
 312
 313         return ret < 0 ? ret : 1;
 314 }
 315
 316 /*
 317  * Parse a single line of data, and return a newly allocated dev struct.
 318  * Add the new device to the cache struct, if one was read.
 319  *
 320  * Lines are of the form <device [TAG="value" ...]>/dev/foo</device>
 321  *
 322  * Returns -ve value on error.
 323  * Returns 0 otherwise.
 324  * If a valid device was read, *dev_p is non-NULL, otherwise it is NULL
 325  * (e.g. comment lines, unknown XML content, etc).
 326  */
 327 static int blkid_parse_line(blkid_cache cache, blkid_dev *dev_p, char *cp)
 328 {
 329         blkid_dev dev;
 330         int ret;
 331
 332         if (!cache || !dev_p)
 333                 return -BLKID_ERR_PARAM;
 334
 335         *dev_p = NULL;
 336
 337         DBG(READ, ul_debug("line: %s", cp));
 338
 339         if ((ret = parse_dev(cache, dev_p, &cp)) <= 0)
 340                 return ret;
 341
 342         dev = *dev_p;
 343
 344         while ((ret = parse_tag(cache, dev, &cp)) > 0) {
 345                 ;
 346         }
 347
 348         if (dev->bid_type == NULL) {
 349                 DBG(READ, ul_debug("blkid: device %s has no TYPE",dev->bid_name));
 350                 blkid_free_dev(dev);
 351                 goto done;
 352         }
 353
 354 done:
 355         return ret;
 356 }
 357
 358 /*
 359  * Parse the specified filename, and return the data in the supplied or
 360  * a newly allocated cache struct.  If the file doesn't exist, return a
 361  * new empty cache struct.
 362  */
 363 void blkid_read_cache(blkid_cache cache)
 364 {
 365         FILE *file;
 366         char buf[4096];
 367         int fd, lineno = 0;
 368         struct stat st;
 369
 370         /*
 371          * If the file doesn't exist, then we just return an empty
 372          * struct so that the cache can be populated.
 373          */
 374         if ((fd = open(cache->bic_filename, O_RDONLY|O_CLOEXEC)) < 0)
 375                 return;
 376         if (fstat(fd, &st) < 0)
 377                 goto errout;
 378         if ((st.st_mtime == cache->bic_ftime) ||
 379             (cache->bic_flags & BLKID_BIC_FL_CHANGED)) {
 380                 DBG(CACHE, ul_debug("skipping re-read of %s",
 381                                         cache->bic_filename));
 382                 goto errout;
 383         }
 384
 385         DBG(CACHE, ul_debug("reading cache file %s",
 386                                 cache->bic_filename));
 387
 388         file = fdopen(fd, "r" UL_CLOEXECSTR);
 389         if (!file)
 390                 goto errout;
 391
 392         while (fgets(buf, sizeof(buf), file)) {
 393                 blkid_dev dev;
 394                 unsigned int end;
 395
 396                 lineno++;
 397                 if (buf[0] == 0)
 398                         continue;
 399                 end = strlen(buf) - 1;
 400                 /* Continue reading next line if it ends with a backslash */
 401                 while (end < (sizeof(buf) - 2) && buf[end] == '\\' &&
 402                        fgets(buf + end, sizeof(buf) - end, file)) {
 403                         end = strlen(buf) - 1;
 404                         lineno++;
 405                 }
 406
 407                 if (blkid_parse_line(cache, &dev, buf) < 0) {
 408                         DBG(READ, ul_debug("blkid: bad format on line %d", lineno));
 409                         continue;
 410                 }
 411         }
 412         fclose(file);
 413
 414         /*
 415          * Initially we do not need to write out the cache file.
 416          */
 417         cache->bic_flags &= ~BLKID_BIC_FL_CHANGED;
 418         cache->bic_ftime = st.st_mtime;
 419
 420         return;
 421 errout:
 422         close(fd);
 423         return;
 424 }
 425
 426 #ifdef TEST_PROGRAM
 427
 428 int main(int argc, char**argv)
 429 {
 430         blkid_cache cache = NULL;
 431         int ret;
 432
 433         blkid_init_debug(BLKID_DEBUG_ALL);
 434         if (argc > 2) {
 435                 fprintf(stderr, "Usage: %s [filename]\n"
 436                         "Test parsing of the cache (filename)\n", argv[0]);
 437                 exit(1);
 438         }
 439         if ((ret = blkid_get_cache(&cache, argv[1])) < 0)
 440                 fprintf(stderr, "error %d reading cache file %s\n", ret,
 441                         argv[1] ? argv[1] : blkid_get_cache_filename(NULL));
 442
 443         blkid_put_cache(cache);
 444
 445         return ret;
 446 }
 447 #endif