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