cache.c

   1 /* cache.c: cache management
   2  *
   3  * Copyright (C) 2006-2014 cgit Development Team <cgit@lists.zx2c4.com>
   4  *
   5  * Licensed under GNU General Public License v2
   6  *   (see COPYING for full license text)
   7  *
   8  *
   9  * The cache is just a directory structure where each file is a cache slot,
  10  * and each filename is based on the hash of some key (e.g. the cgit url).
  11  * Each file contains the full key followed by the cached content for that
  12  * key.
  13  *
  14  */
  15
  16 #include "cgit.h"
  17 #include "cache.h"
  18 #include "html.h"
  19 #ifdef HAVE_LINUX_SENDFILE
  20 #include <sys/sendfile.h>
  21 #endif
  22
  23 #define CACHE_BUFSIZE (1024 * 4)
  24
  25 struct cache_slot {
  26         const char *key;
  27         size_t keylen;
  28         int ttl;
  29         cache_fill_fn fn;
  30         int cache_fd;
  31         int lock_fd;
  32         const char *cache_name;
  33         const char *lock_name;
  34         int match;
  35         struct stat cache_st;
  36         int bufsize;
  37         char buf[CACHE_BUFSIZE];
  38 };
  39
  40 /* Open an existing cache slot and fill the cache buffer with
  41  * (part of) the content of the cache file. Return 0 on success
  42  * and errno otherwise.
  43  */
  44 static int open_slot(struct cache_slot *slot)
  45 {
  46         char *bufz;
  47         ssize_t bufkeylen = -1;
  48
  49         slot->cache_fd = open(slot->cache_name, O_RDONLY);
  50         if (slot->cache_fd == -1)
  51                 return errno;
  52
  53         if (fstat(slot->cache_fd, &slot->cache_st))
  54                 return errno;
  55
  56         slot->bufsize = xread(slot->cache_fd, slot->buf, sizeof(slot->buf));
  57         if (slot->bufsize < 0)
  58                 return errno;
  59
  60         bufz = memchr(slot->buf, 0, slot->bufsize);
  61         if (bufz)
  62                 bufkeylen = bufz - slot->buf;
  63
  64         if (slot->key)
  65                 slot->match = bufkeylen == slot->keylen &&
  66                     !memcmp(slot->key, slot->buf, bufkeylen + 1);
  67
  68         return 0;
  69 }
  70
  71 /* Close the active cache slot */
  72 static int close_slot(struct cache_slot *slot)
  73 {
  74         int err = 0;
  75         if (slot->cache_fd > 0) {
  76                 if (close(slot->cache_fd))
  77                         err = errno;
  78                 else
  79                         slot->cache_fd = -1;
  80         }
  81         return err;
  82 }
  83
  84 /* Print the content of the active cache slot (but skip the key). */
  85 static int print_slot(struct cache_slot *slot)
  86 {
  87 #ifdef HAVE_LINUX_SENDFILE
  88         off_t start_off;
  89         int ret;
  90
  91         start_off = slot->keylen + 1;
  92
  93         do {
  94                 ret = sendfile(STDOUT_FILENO, slot->cache_fd, &start_off,
  95                                 slot->cache_st.st_size - start_off);
  96                 if (ret < 0) {
  97                         if (errno == EAGAIN || errno == EINTR)
  98                                 continue;
  99                         return errno;
 100                 }
 101                 return 0;
 102         } while (1);
 103 #else
 104         ssize_t i, j;
 105
 106         i = lseek(slot->cache_fd, slot->keylen + 1, SEEK_SET);
 107         if (i != slot->keylen + 1)
 108                 return errno;
 109
 110         do {
 111                 i = j = xread(slot->cache_fd, slot->buf, sizeof(slot->buf));
 112                 if (i > 0)
 113                         j = xwrite(STDOUT_FILENO, slot->buf, i);
 114         } while (i > 0 && j == i);
 115
 116         if (i < 0 || j != i)
 117                 return errno;
 118         else
 119                 return 0;
 120 #endif
 121 }
 122
 123 /* Check if the slot has expired */
 124 static int is_expired(struct cache_slot *slot)
 125 {
 126         if (slot->ttl < 0)
 127                 return 0;
 128         else
 129                 return slot->cache_st.st_mtime + slot->ttl * 60 < time(NULL);
 130 }
 131
 132 /* Check if the slot has been modified since we opened it.
 133  * NB: If stat() fails, we pretend the file is modified.
 134  */
 135 static int is_modified(struct cache_slot *slot)
 136 {
 137         struct stat st;
 138
 139         if (stat(slot->cache_name, &st))
 140                 return 1;
 141         return (st.st_ino != slot->cache_st.st_ino ||
 142                 st.st_mtime != slot->cache_st.st_mtime ||
 143                 st.st_size != slot->cache_st.st_size);
 144 }
 145
 146 /* Close an open lockfile */
 147 static int close_lock(struct cache_slot *slot)
 148 {
 149         int err = 0;
 150         if (slot->lock_fd > 0) {
 151                 if (close(slot->lock_fd))
 152                         err = errno;
 153                 else
 154                         slot->lock_fd = -1;
 155         }
 156         return err;
 157 }
 158
 159 /* Create a lockfile used to store the generated content for a cache
 160  * slot, and write the slot key + \0 into it.
 161  * Returns 0 on success and errno otherwise.
 162  */
 163 static int lock_slot(struct cache_slot *slot)
 164 {
 165         struct flock lock = {
 166                 .l_type = F_WRLCK,
 167                 .l_whence = SEEK_SET,
 168                 .l_start = 0,
 169                 .l_len = 0,
 170         };
 171
 172         slot->lock_fd = open(slot->lock_name, O_RDWR | O_CREAT,
 173                              S_IRUSR | S_IWUSR);
 174         if (slot->lock_fd == -1)
 175                 return errno;
 176         if (fcntl(slot->lock_fd, F_SETLK, &lock) < 0) {
 177                 int saved_errno = errno;
 178                 close(slot->lock_fd);
 179                 slot->lock_fd = -1;
 180                 return saved_errno;
 181         }
 182         if (xwrite(slot->lock_fd, slot->key, slot->keylen + 1) < 0)
 183                 return errno;
 184         return 0;
 185 }
 186
 187 /* Release the current lockfile. If `replace_old_slot` is set the
 188  * lockfile replaces the old cache slot, otherwise the lockfile is
 189  * just deleted.
 190  */
 191 static int unlock_slot(struct cache_slot *slot, int replace_old_slot)
 192 {
 193         int err;
 194
 195         if (replace_old_slot)
 196                 err = rename(slot->lock_name, slot->cache_name);
 197         else
 198                 err = unlink(slot->lock_name);
 199
 200         if (err)
 201                 return errno;
 202
 203         return 0;
 204 }
 205
 206 /* Generate the content for the current cache slot by redirecting
 207  * stdout to the lock-fd and invoking the callback function
 208  */
 209 static int fill_slot(struct cache_slot *slot)
 210 {
 211         int tmp;
 212
 213         /* Preserve stdout */
 214         tmp = dup(STDOUT_FILENO);
 215         if (tmp == -1)
 216                 return errno;
 217
 218         /* Redirect stdout to lockfile */
 219         if (dup2(slot->lock_fd, STDOUT_FILENO) == -1) {
 220                 close(tmp);
 221                 return errno;
 222         }
 223
 224         /* Generate cache content */
 225         slot->fn();
 226
 227         /* update stat info */
 228         if (fstat(slot->lock_fd, &slot->cache_st)) {
 229                 close(tmp);
 230                 return errno;
 231         }
 232
 233         /* Restore stdout */
 234         if (dup2(tmp, STDOUT_FILENO) == -1) {
 235                 close(tmp);
 236                 return errno;
 237         }
 238
 239         /* Close the temporary filedescriptor */
 240         if (close(tmp))
 241                 return errno;
 242
 243         return 0;
 244 }
 245
 246 /* Crude implementation of 32-bit FNV-1 hash algorithm,
 247  * see http://www.isthe.com/chongo/tech/comp/fnv/ for details
 248  * about the magic numbers.
 249  */
 250 #define FNV_OFFSET 0x811c9dc5
 251 #define FNV_PRIME  0x01000193
 252
 253 unsigned long hash_str(const char *str)
 254 {
 255         unsigned long h = FNV_OFFSET;
 256         unsigned char *s = (unsigned char *)str;
 257
 258         if (!s)
 259                 return h;
 260
 261         while (*s) {
 262                 h *= FNV_PRIME;
 263                 h ^= *s++;
 264         }
 265         return h;
 266 }
 267
 268 static int process_slot(struct cache_slot *slot)
 269 {
 270         int err;
 271
 272         err = open_slot(slot);
 273         if (!err && slot->match) {
 274                 if (is_expired(slot)) {
 275                         if (!lock_slot(slot)) {
 276                                 /* If the cachefile has been replaced between
 277                                  * `open_slot` and `lock_slot`, we'll just
 278                                  * serve the stale content from the original
 279                                  * cachefile. This way we avoid pruning the
 280                                  * newly generated slot. The same code-path
 281                                  * is chosen if fill_slot() fails for some
 282                                  * reason.
 283                                  *
 284                                  * TODO? check if the new slot contains the
 285                                  * same key as the old one, since we would
 286                                  * prefer to serve the newest content.
 287                                  * This will require us to open yet another
 288                                  * file-descriptor and read and compare the
 289                                  * key from the new file, so for now we're
 290                                  * lazy and just ignore the new file.
 291                                  */
 292                                 if (is_modified(slot) || fill_slot(slot)) {
 293                                         unlock_slot(slot, 0);
 294                                         close_lock(slot);
 295                                 } else {
 296                                         close_slot(slot);
 297                                         unlock_slot(slot, 1);
 298                                         slot->cache_fd = slot->lock_fd;
 299                                 }
 300                         }
 301                 }
 302                 if ((err = print_slot(slot)) != 0) {
 303                         cache_log("[cgit] error printing cache %s: %s (%d)\n",
 304                                   slot->cache_name,
 305                                   strerror(err),
 306                                   err);
 307                 }
 308                 close_slot(slot);
 309                 return err;
 310         }
 311
 312         /* If the cache slot does not exist (or its key doesn't match the
 313          * current key), lets try to create a new cache slot for this
 314          * request. If this fails (for whatever reason), lets just generate
 315          * the content without caching it and fool the caller to belive
 316          * everything worked out (but print a warning on stdout).
 317          */
 318
 319         close_slot(slot);
 320         if ((err = lock_slot(slot)) != 0) {
 321                 cache_log("[cgit] Unable to lock slot %s: %s (%d)\n",
 322                           slot->lock_name, strerror(err), err);
 323                 slot->fn();
 324                 return 0;
 325         }
 326
 327         if ((err = fill_slot(slot)) != 0) {
 328                 cache_log("[cgit] Unable to fill slot %s: %s (%d)\n",
 329                           slot->lock_name, strerror(err), err);
 330                 unlock_slot(slot, 0);
 331                 close_lock(slot);
 332                 slot->fn();
 333                 return 0;
 334         }
 335         // We've got a valid cache slot in the lock file, which
 336         // is about to replace the old cache slot. But if we
 337         // release the lockfile and then try to open the new cache
 338         // slot, we might get a race condition with a concurrent
 339         // writer for the same cache slot (with a different key).
 340         // Lets avoid such a race by just printing the content of
 341         // the lock file.
 342         slot->cache_fd = slot->lock_fd;
 343         unlock_slot(slot, 1);
 344         if ((err = print_slot(slot)) != 0) {
 345                 cache_log("[cgit] error printing cache %s: %s (%d)\n",
 346                           slot->cache_name,
 347                           strerror(err),
 348                           err);
 349         }
 350         close_slot(slot);
 351         return err;
 352 }
 353
 354 /* Print cached content to stdout, generate the content if necessary. */
 355 int cache_process(int size, const char *path, const char *key, int ttl,
 356                   cache_fill_fn fn)
 357 {
 358         unsigned long hash;
 359         int i;
 360         struct strbuf filename = STRBUF_INIT;
 361         struct strbuf lockname = STRBUF_INIT;
 362         struct cache_slot slot;
 363         int result;
 364
 365         /* If the cache is disabled, just generate the content */
 366         if (size <= 0 || ttl == 0) {
 367                 fn();
 368                 return 0;
 369         }
 370
 371         /* Verify input, calculate filenames */
 372         if (!path) {
 373                 cache_log("[cgit] Cache path not specified, caching is disabled\n");
 374                 fn();
 375                 return 0;
 376         }
 377         if (!key)
 378                 key = "";
 379         hash = hash_str(key) % size;
 380         strbuf_addstr(&filename, path);
 381         strbuf_ensure_end(&filename, '/');
 382         for (i = 0; i < 8; i++) {
 383                 strbuf_addf(&filename, "%x", (unsigned char)(hash & 0xf));
 384                 hash >>= 4;
 385         }
 386         strbuf_addbuf(&lockname, &filename);
 387         strbuf_addstr(&lockname, ".lock");
 388         slot.fn = fn;
 389         slot.ttl = ttl;
 390         slot.cache_name = filename.buf;
 391         slot.lock_name = lockname.buf;
 392         slot.key = key;
 393         slot.keylen = strlen(key);
 394         result = process_slot(&slot);
 395
 396         strbuf_release(&filename);
 397         strbuf_release(&lockname);
 398         return result;
 399 }
 400
 401 /* Return a strftime formatted date/time
 402  * NB: the result from this function is to shared memory
 403  */
 404 static char *sprintftime(const char *format, time_t time)
 405 {
 406         static char buf[64];
 407         struct tm *tm;
 408
 409         if (!time)
 410                 return NULL;
 411         tm = gmtime(&time);
 412         strftime(buf, sizeof(buf)-1, format, tm);
 413         return buf;
 414 }
 415
 416 int cache_ls(const char *path)
 417 {
 418         DIR *dir;
 419         struct dirent *ent;
 420         int err = 0;
 421         struct cache_slot slot = { NULL };
 422         struct strbuf fullname = STRBUF_INIT;
 423         size_t prefixlen;
 424
 425         if (!path) {
 426                 cache_log("[cgit] cache path not specified\n");
 427                 return -1;
 428         }
 429         dir = opendir(path);
 430         if (!dir) {
 431                 err = errno;
 432                 cache_log("[cgit] unable to open path %s: %s (%d)\n",
 433                           path, strerror(err), err);
 434                 return err;
 435         }
 436         strbuf_addstr(&fullname, path);
 437         strbuf_ensure_end(&fullname, '/');
 438         prefixlen = fullname.len;
 439         while ((ent = readdir(dir)) != NULL) {
 440                 if (strlen(ent->d_name) != 8)
 441                         continue;
 442                 strbuf_setlen(&fullname, prefixlen);
 443                 strbuf_addstr(&fullname, ent->d_name);
 444                 slot.cache_name = fullname.buf;
 445                 if ((err = open_slot(&slot)) != 0) {
 446                         cache_log("[cgit] unable to open path %s: %s (%d)\n",
 447                                   fullname.buf, strerror(err), err);
 448                         continue;
 449                 }
 450                 htmlf("%s %s %10"PRIuMAX" %s\n",
 451                       fullname.buf,
 452                       sprintftime("%Y-%m-%d %H:%M:%S",
 453                                   slot.cache_st.st_mtime),
 454                       (uintmax_t)slot.cache_st.st_size,
 455                       slot.buf);
 456                 close_slot(&slot);
 457         }
 458         closedir(dir);
 459         strbuf_release(&fullname);
 460         return 0;
 461 }
 462
 463 /* Print a message to stdout */
 464 void cache_log(const char *format, ...)
 465 {
 466         va_list args;
 467         va_start(args, format);
 468         vfprintf(stderr, format, args);
 469         va_end(args);
 470 }
 471