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
  19 #define CACHE_BUFSIZE (1024 * 4)
  20
  21 struct cache_slot {
  22         const char *key;
  23         int keylen;
  24         int ttl;
  25         cache_fill_fn fn;
  26         void *cbdata;
  27         int cache_fd;
  28         int lock_fd;
  29         const char *cache_name;
  30         const char *lock_name;
  31         int match;
  32         struct stat cache_st;
  33         struct stat lock_st;
  34         int bufsize;
  35         char buf[CACHE_BUFSIZE];
  36 };
  37
  38 /* Open an existing cache slot and fill the cache buffer with
  39  * (part of) the content of the cache file. Return 0 on success
  40  * and errno otherwise.
  41  */
  42 static int open_slot(struct cache_slot *slot)
  43 {
  44         char *bufz;
  45         int bufkeylen = -1;
  46
  47         slot->cache_fd = open(slot->cache_name, O_RDONLY);
  48         if (slot->cache_fd == -1)
  49                 return errno;
  50
  51         if (fstat(slot->cache_fd, &slot->cache_st))
  52                 return errno;
  53
  54         slot->bufsize = xread(slot->cache_fd, slot->buf, sizeof(slot->buf));
  55         if (slot->bufsize < 0)
  56                 return errno;
  57
  58         bufz = memchr(slot->buf, 0, slot->bufsize);
  59         if (bufz)
  60                 bufkeylen = bufz - slot->buf;
  61
  62         slot->match = bufkeylen == slot->keylen &&
  63             !memcmp(slot->key, slot->buf, bufkeylen + 1);
  64
  65         return 0;
  66 }
  67
  68 /* Close the active cache slot */
  69 static int close_slot(struct cache_slot *slot)
  70 {
  71         int err = 0;
  72         if (slot->cache_fd > 0) {
  73                 if (close(slot->cache_fd))
  74                         err = errno;
  75                 else
  76                         slot->cache_fd = -1;
  77         }
  78         return err;
  79 }
  80
  81 /* Print the content of the active cache slot (but skip the key). */
  82 static int print_slot(struct cache_slot *slot)
  83 {
  84         ssize_t i;
  85
  86         i = lseek(slot->cache_fd, slot->keylen + 1, SEEK_SET);
  87         if (i != slot->keylen + 1)
  88                 return errno;
  89
  90         while((i = xread(slot->cache_fd, slot->buf, sizeof(slot->buf))) > 0)
  91                 i = xwrite(STDOUT_FILENO, slot->buf, i);
  92
  93         if (i < 0)
  94                 return errno;
  95         else
  96                 return 0;
  97 }
  98
  99 /* Check if the slot has expired */
 100 static int is_expired(struct cache_slot *slot)
 101 {
 102         if (slot->ttl < 0)
 103                 return 0;
 104         else
 105                 return slot->cache_st.st_mtime + slot->ttl*60 < time(NULL);
 106 }
 107
 108 /* Check if the slot has been modified since we opened it.
 109  * NB: If stat() fails, we pretend the file is modified.
 110  */
 111 static int is_modified(struct cache_slot *slot)
 112 {
 113         struct stat st;
 114
 115         if (stat(slot->cache_name, &st))
 116                 return 1;
 117         return (st.st_ino != slot->cache_st.st_ino ||
 118                 st.st_mtime != slot->cache_st.st_mtime ||
 119                 st.st_size != slot->cache_st.st_size);
 120 }
 121
 122 /* Close an open lockfile */
 123 static int close_lock(struct cache_slot *slot)
 124 {
 125         int err = 0;
 126         if (slot->lock_fd > 0) {
 127                 if (close(slot->lock_fd))
 128                         err = errno;
 129                 else
 130                         slot->lock_fd = -1;
 131         }
 132         return err;
 133 }
 134
 135 /* Create a lockfile used to store the generated content for a cache
 136  * slot, and write the slot key + \0 into it.
 137  * Returns 0 on success and errno otherwise.
 138  */
 139 static int lock_slot(struct cache_slot *slot)
 140 {
 141         slot->lock_fd = open(slot->lock_name, O_RDWR|O_CREAT|O_EXCL,
 142                              S_IRUSR|S_IWUSR);
 143         if (slot->lock_fd == -1)
 144                 return errno;
 145         if (xwrite(slot->lock_fd, slot->key, slot->keylen + 1) < 0)
 146                 return errno;
 147         return 0;
 148 }
 149
 150 /* Release the current lockfile. If `replace_old_slot` is set the
 151  * lockfile replaces the old cache slot, otherwise the lockfile is
 152  * just deleted.
 153  */
 154 static int unlock_slot(struct cache_slot *slot, int replace_old_slot)
 155 {
 156         int err;
 157
 158         if (replace_old_slot)
 159                 err = rename(slot->lock_name, slot->cache_name);
 160         else
 161                 err = unlink(slot->lock_name);
 162
 163         if (err)
 164                 return errno;
 165
 166         return 0;
 167 }
 168
 169 /* Generate the content for the current cache slot by redirecting
 170  * stdout to the lock-fd and invoking the callback function
 171  */
 172 static int fill_slot(struct cache_slot *slot)
 173 {
 174         int tmp;
 175
 176         /* Preserve stdout */
 177         tmp = dup(STDOUT_FILENO);
 178         if (tmp == -1)
 179                 return errno;
 180
 181         /* Redirect stdout to lockfile */
 182         if (dup2(slot->lock_fd, STDOUT_FILENO) == -1)
 183                 return errno;
 184
 185         /* Generate cache content */
 186         slot->fn(slot->cbdata);
 187
 188         /* Restore stdout */
 189         if (dup2(tmp, STDOUT_FILENO) == -1)
 190                 return errno;
 191
 192         /* Close the temporary filedescriptor */
 193         if (close(tmp))
 194                 return errno;
 195
 196         return 0;
 197 }
 198
 199 /* Crude implementation of 32-bit FNV-1 hash algorithm,
 200  * see http://www.isthe.com/chongo/tech/comp/fnv/ for details
 201  * about the magic numbers.
 202  */
 203 #define FNV_OFFSET 0x811c9dc5
 204 #define FNV_PRIME  0x01000193
 205
 206 unsigned long hash_str(const char *str)
 207 {
 208         unsigned long h = FNV_OFFSET;
 209         unsigned char *s = (unsigned char *)str;
 210
 211         if (!s)
 212                 return h;
 213
 214         while(*s) {
 215                 h *= FNV_PRIME;
 216                 h ^= *s++;
 217         }
 218         return h;
 219 }
 220
 221 static int process_slot(struct cache_slot *slot)
 222 {
 223         int err;
 224
 225         err = open_slot(slot);
 226         if (!err && slot->match) {
 227                 if (is_expired(slot)) {
 228                         if (!lock_slot(slot)) {
 229                                 /* If the cachefile has been replaced between
 230                                  * `open_slot` and `lock_slot`, we'll just
 231                                  * serve the stale content from the original
 232                                  * cachefile. This way we avoid pruning the
 233                                  * newly generated slot. The same code-path
 234                                  * is chosen if fill_slot() fails for some
 235                                  * reason.
 236                                  *
 237                                  * TODO? check if the new slot contains the
 238                                  * same key as the old one, since we would
 239                                  * prefer to serve the newest content.
 240                                  * This will require us to open yet another
 241                                  * file-descriptor and read and compare the
 242                                  * key from the new file, so for now we're
 243                                  * lazy and just ignore the new file.
 244                                  */
 245                                 if (is_modified(slot) || fill_slot(slot)) {
 246                                         unlock_slot(slot, 0);
 247                                         close_lock(slot);
 248                                 } else {
 249                                         close_slot(slot);
 250                                         unlock_slot(slot, 1);
 251                                         slot->cache_fd = slot->lock_fd;
 252                                 }
 253                         }
 254                 }
 255                 if ((err = print_slot(slot)) != 0) {
 256                         cache_log("[cgit] error printing cache %s: %s (%d)\n",
 257                                   slot->cache_name,
 258                                   strerror(err),
 259                                   err);
 260                 }
 261                 close_slot(slot);
 262                 return err;
 263         }
 264
 265         /* If the cache slot does not exist (or its key doesn't match the
 266          * current key), lets try to create a new cache slot for this
 267          * request. If this fails (for whatever reason), lets just generate
 268          * the content without caching it and fool the caller to belive
 269          * everything worked out (but print a warning on stdout).
 270          */
 271
 272         close_slot(slot);
 273         if ((err = lock_slot(slot)) != 0) {
 274                 cache_log("[cgit] Unable to lock slot %s: %s (%d)\n",
 275                           slot->lock_name, strerror(err), err);
 276                 slot->fn(slot->cbdata);
 277                 return 0;
 278         }
 279
 280         if ((err = fill_slot(slot)) != 0) {
 281                 cache_log("[cgit] Unable to fill slot %s: %s (%d)\n",
 282                           slot->lock_name, strerror(err), err);
 283                 unlock_slot(slot, 0);
 284                 close_lock(slot);
 285                 slot->fn(slot->cbdata);
 286                 return 0;
 287         }
 288         // We've got a valid cache slot in the lock file, which
 289         // is about to replace the old cache slot. But if we
 290         // release the lockfile and then try to open the new cache
 291         // slot, we might get a race condition with a concurrent
 292         // writer for the same cache slot (with a different key).
 293         // Lets avoid such a race by just printing the content of
 294         // the lock file.
 295         slot->cache_fd = slot->lock_fd;
 296         unlock_slot(slot, 1);
 297         if ((err = print_slot(slot)) != 0) {
 298                 cache_log("[cgit] error printing cache %s: %s (%d)\n",
 299                           slot->cache_name,
 300                           strerror(err),
 301                           err);
 302         }
 303         close_slot(slot);
 304         return err;
 305 }
 306
 307 /* Print cached content to stdout, generate the content if necessary. */
 308 int cache_process(int size, const char *path, const char *key, int ttl,
 309                   cache_fill_fn fn, void *cbdata)
 310 {
 311         unsigned long hash;
 312         int len, i;
 313         char filename[1024];
 314         char lockname[1024 + 5];  /* 5 = ".lock" */
 315         struct cache_slot slot;
 316
 317         /* If the cache is disabled, just generate the content */
 318         if (size <= 0) {
 319                 fn(cbdata);
 320                 return 0;
 321         }
 322
 323         /* Verify input, calculate filenames */
 324         if (!path) {
 325                 cache_log("[cgit] Cache path not specified, caching is disabled\n");
 326                 fn(cbdata);
 327                 return 0;
 328         }
 329         len = strlen(path);
 330         if (len > sizeof(filename) - 10) { /* 10 = "/01234567\0" */
 331                 cache_log("[cgit] Cache path too long, caching is disabled: %s\n",
 332                           path);
 333                 fn(cbdata);
 334                 return 0;
 335         }
 336         if (!key)
 337                 key = "";
 338         hash = hash_str(key) % size;
 339         strcpy(filename, path);
 340         if (filename[len - 1] != '/')
 341                 filename[len++] = '/';
 342         for(i = 0; i < 8; i++) {
 343                 sprintf(filename + len++, "%x",
 344                         (unsigned char)(hash & 0xf));
 345                 hash >>= 4;
 346         }
 347         filename[len] = '\0';
 348         strcpy(lockname, filename);
 349         strcpy(lockname + len, ".lock");
 350         slot.fn = fn;
 351         slot.cbdata = cbdata;
 352         slot.ttl = ttl;
 353         slot.cache_name = filename;
 354         slot.lock_name = lockname;
 355         slot.key = key;
 356         slot.keylen = strlen(key);
 357         return process_slot(&slot);
 358 }
 359
 360 /* Return a strftime formatted date/time
 361  * NB: the result from this function is to shared memory
 362  */
 363 char *sprintftime(const char *format, time_t time)
 364 {
 365         static char buf[64];
 366         struct tm *tm;
 367
 368         if (!time)
 369                 return NULL;
 370         tm = gmtime(&time);
 371         strftime(buf, sizeof(buf)-1, format, tm);
 372         return buf;
 373 }
 374
 375 int cache_ls(const char *path)
 376 {
 377         DIR *dir;
 378         struct dirent *ent;
 379         int err = 0;
 380         struct cache_slot slot;
 381         char fullname[1024];
 382         char *name;
 383
 384         if (!path) {
 385                 cache_log("[cgit] cache path not specified\n");
 386                 return -1;
 387         }
 388         if (strlen(path) > 1024 - 10) {
 389                 cache_log("[cgit] cache path too long: %s\n",
 390                           path);
 391                 return -1;
 392         }
 393         dir = opendir(path);
 394         if (!dir) {
 395                 err = errno;
 396                 cache_log("[cgit] unable to open path %s: %s (%d)\n",
 397                           path, strerror(err), err);
 398                 return err;
 399         }
 400         strcpy(fullname, path);
 401         name = fullname + strlen(path);
 402         if (*(name - 1) != '/') {
 403                 *name++ = '/';
 404                 *name = '\0';
 405         }
 406         slot.cache_name = fullname;
 407         while((ent = readdir(dir)) != NULL) {
 408                 if (strlen(ent->d_name) != 8)
 409                         continue;
 410                 strcpy(name, ent->d_name);
 411                 if ((err = open_slot(&slot)) != 0) {
 412                         cache_log("[cgit] unable to open path %s: %s (%d)\n",
 413                                   fullname, strerror(err), err);
 414                         continue;
 415                 }
 416                 printf("%s %s %10lld %s\n",
 417                        name,
 418                        sprintftime("%Y-%m-%d %H:%M:%S",
 419                                    slot.cache_st.st_mtime),
 420                        slot.cache_st.st_size,
 421                        slot.buf);
 422                 close_slot(&slot);
 423         }
 424         closedir(dir);
 425         return 0;
 426 }
 427
 428 /* Print a message to stdout */
 429 void cache_log(const char *format, ...)
 430 {
 431         va_list args;
 432         va_start(args, format);
 433         vfprintf(stderr, format, args);
 434         va_end(args);
 435 }
 436