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