cache.c

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