apache/support/htcacheclean.c

   1 /* Licensed to the Apache Software Foundation (ASF) under one or more
   2  * contributor license agreements.  See the NOTICE file distributed with
   3  * this work for additional information regarding copyright ownership.
   4  * The ASF licenses this file to You under the Apache License, Version 2.0
   5  * (the "License"); you may not use this file except in compliance with
   6  * the License.  You may obtain a copy of the License at
   7  *
   8  *     http://www.apache.org/licenses/LICENSE-2.0
   9  *
  10  * Unless required by applicable law or agreed to in writing, software
  11  * distributed under the License is distributed on an "AS IS" BASIS,
  12  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  13  * See the License for the specific language governing permissions and
  14  * limitations under the License.
  15  */
  16
  17 /*
  18  * htcacheclean.c: simple program for cleaning of
  19  * the disk cache of the Apache HTTP server
  20  *
  21  * Contributed by Andreas Steinmetz <ast domdv.de>
  22  * 8 Oct 2004
  23  */
  24
  25 #include "apr.h"
  26 #include "apr_lib.h"
  27 #include "apr_strings.h"
  28 #include "apr_file_io.h"
  29 #include "apr_file_info.h"
  30 #include "apr_pools.h"
  31 #include "apr_hash.h"
  32 #include "apr_thread_proc.h"
  33 #include "apr_signal.h"
  34 #include "apr_getopt.h"
  35 #include "apr_ring.h"
  36 #include "apr_date.h"
  37 #include "apr_buckets.h"
  38 #include "../modules/cache/mod_disk_cache.h"
  39
  40 #if APR_HAVE_UNISTD_H
  41 #include <unistd.h>
  42 #endif
  43 #if APR_HAVE_STDLIB_H
  44 #include <stdlib.h>
  45 #endif
  46
  47 /* define the following for debugging */
  48 #undef DEBUG
  49
  50 /*
  51  * Note: on Linux delays <= 2ms are busy waits without
  52  *       scheduling, so never use a delay <= 2ms below
  53  */
  54
  55 #define NICE_DELAY    10000     /* usecs */
  56 #define DELETE_NICE   10        /* be nice after this amount of delete ops */
  57 #define STAT_ATTEMPTS 10        /* maximum stat attempts for a file */
  58 #define STAT_DELAY    5000      /* usecs */
  59 #define HEADER        1         /* headers file */
  60 #define DATA          2         /* body file */
  61 #define TEMP          4         /* temporary file */
  62 #define HEADERDATA    (HEADER|DATA)
  63 #define MAXDEVIATION  3600      /* secs */
  64 #define SECS_PER_MIN  60
  65 #define KBYTE         1024
  66 #define MBYTE         1048576
  67 #define GBYTE         1073741824
  68
  69 #define DIRINFO (APR_FINFO_MTIME|APR_FINFO_SIZE|APR_FINFO_TYPE|APR_FINFO_LINK)
  70
  71 typedef struct _direntry {
  72     APR_RING_ENTRY(_direntry) link;
  73     int type;         /* type of file/fileset: TEMP, HEADER, DATA, HEADERDATA */
  74     apr_time_t htime; /* headers file modification time */
  75     apr_time_t dtime; /* body file modification time */
  76     apr_off_t hsize;  /* headers file size */
  77     apr_off_t dsize;  /* body or temporary file size */
  78     char *basename;   /* file/fileset base name */
  79 } DIRENTRY;
  80
  81 typedef struct _entry {
  82     APR_RING_ENTRY(_entry) link;
  83     apr_time_t expire;        /* cache entry exiration time */
  84     apr_time_t response_time; /* cache entry time of last response to client */
  85     apr_time_t htime;         /* headers file modification time */
  86     apr_time_t dtime;         /* body file modification time */
  87     apr_off_t hsize;          /* headers file size */
  88     apr_off_t dsize;          /* body or temporary file size */
  89     char *basename;           /* fileset base name */
  90 } ENTRY;
  91
  92
  93 static int delcount;    /* file deletion count for nice mode */
  94 static int interrupted; /* flag: true if SIGINT or SIGTERM occurred */
  95 static int realclean;   /* flag: true means user said apache is not running */
  96 static int verbose;     /* flag: true means print statistics */
  97 static int benice;      /* flag: true means nice mode is activated */
  98 static int dryrun;      /* flag: true means dry run, don't actually delete
  99                                  anything */
 100 static int deldirs;     /* flag: true means directories should be deleted */
 101 static int baselen;     /* string length of the path to the proxy directory */
 102 static apr_time_t now;  /* start time of this processing run */
 103
 104 static apr_file_t *errfile;   /* stderr file handle */
 105 static apr_off_t unsolicited; /* file size summary for deleted unsolicited
 106                                  files */
 107 static APR_RING_ENTRY(_entry) root; /* ENTRY ring anchor */
 108
 109 /* short program name as called */
 110 static const char *shortname = "htcacheclean";
 111
 112 #ifdef DEBUG
 113 /*
 114  * fake delete for debug purposes
 115  */
 116 #define apr_file_remove fake_file_remove
 117 static void fake_file_remove(char *pathname, apr_pool_t *p)
 118 {
 119     apr_finfo_t info;
 120
 121     /* stat and printing to simulate some deletion system load and to
 122        display what would actually have happened */
 123     apr_stat(&info, pathname, DIRINFO, p);
 124     apr_file_printf(errfile, "would delete %s" APR_EOL_STR, pathname);
 125 }
 126 #endif
 127
 128 /*
 129  * called on SIGINT or SIGTERM
 130  */
 131 static void setterm(int unused)
 132 {
 133 #ifdef DEBUG
 134     apr_file_printf(errfile, "interrupt" APR_EOL_STR);
 135 #endif
 136     interrupted = 1;
 137 }
 138
 139 /*
 140  * called in out of memory condition
 141  */
 142 static int oom(int unused)
 143 {
 144     static int called = 0;
 145
 146     /* be careful to call exit() only once */
 147     if (!called) {
 148         called = 1;
 149         exit(1);
 150     }
 151     return APR_ENOMEM;
 152 }
 153
 154 /*
 155  * print purge statistics
 156  */
 157 static void printstats(apr_off_t total, apr_off_t sum, apr_off_t max,
 158                        apr_off_t etotal, apr_off_t entries)
 159 {
 160     char ttype, stype, mtype, utype;
 161     apr_off_t tfrag, sfrag, ufrag;
 162
 163     if (!verbose) {
 164         return;
 165     }
 166
 167     ttype = 'K';
 168     tfrag = ((total * 10) / KBYTE) % 10;
 169     total /= KBYTE;
 170     if (total >= KBYTE) {
 171         ttype = 'M';
 172         tfrag = ((total * 10) / KBYTE) % 10;
 173         total /= KBYTE;
 174     }
 175
 176     stype = 'K';
 177     sfrag = ((sum * 10) / KBYTE) % 10;
 178     sum /= KBYTE;
 179     if (sum >= KBYTE) {
 180         stype = 'M';
 181         sfrag = ((sum * 10) / KBYTE) % 10;
 182         sum /= KBYTE;
 183     }
 184
 185     mtype = 'K';
 186     max /= KBYTE;
 187     if (max >= KBYTE) {
 188         mtype = 'M';
 189         max /= KBYTE;
 190     }
 191
 192     apr_file_printf(errfile, "Statistics:" APR_EOL_STR);
 193     if (unsolicited) {
 194         utype = 'K';
 195         ufrag = ((unsolicited * 10) / KBYTE) % 10;
 196         unsolicited /= KBYTE;
 197         if (unsolicited >= KBYTE) {
 198             utype = 'M';
 199             ufrag = ((unsolicited * 10) / KBYTE) % 10;
 200             unsolicited /= KBYTE;
 201         }
 202         if (!unsolicited && !ufrag) {
 203             ufrag = 1;
 204         }
 205         apr_file_printf(errfile, "unsolicited size %d.%d%c" APR_EOL_STR,
 206                         (int)(unsolicited), (int)(ufrag), utype);
 207      }
 208      apr_file_printf(errfile, "size limit %d.0%c" APR_EOL_STR,
 209                      (int)(max), mtype);
 210      apr_file_printf(errfile, "total size was %d.%d%c, total size now "
 211                               "%d.%d%c" APR_EOL_STR,
 212                      (int)(total), (int)(tfrag), ttype, (int)(sum),
 213                      (int)(sfrag), stype);
 214      apr_file_printf(errfile, "total entries was %d, total entries now %d"
 215                               APR_EOL_STR, (int)(etotal), (int)(entries));
 216 }
 217
 218 /*
 219  * delete a single file
 220  */
 221 static void delete_file(char *path, char *basename, apr_pool_t *pool)
 222 {
 223     char *nextpath;
 224     apr_pool_t *p;
 225
 226     if (dryrun) {
 227         return;
 228     }
 229
 230     /* temp pool, otherwise lots of memory could be allocated */
 231     apr_pool_create(&p, pool);
 232     nextpath = apr_pstrcat(p, path, "/", basename, NULL);
 233     apr_file_remove(nextpath, p);
 234     apr_pool_destroy(p);
 235
 236     if (benice) {
 237         if (++delcount >= DELETE_NICE) {
 238             apr_sleep(NICE_DELAY);
 239             delcount = 0;
 240         }
 241     }
 242 }
 243
 244 /*
 245  * delete cache file set
 246  */
 247 static void delete_entry(char *path, char *basename, apr_pool_t *pool)
 248 {
 249     char *nextpath;
 250     apr_pool_t *p;
 251
 252     if (dryrun) {
 253         return;
 254     }
 255
 256     /* temp pool, otherwise lots of memory could be allocated */
 257     apr_pool_create(&p, pool);
 258
 259     nextpath = apr_pstrcat(p, path, "/", basename, CACHE_HEADER_SUFFIX, NULL);
 260     apr_file_remove(nextpath, p);
 261
 262     nextpath = apr_pstrcat(p, path, "/", basename, CACHE_DATA_SUFFIX, NULL);
 263     apr_file_remove(nextpath, p);
 264
 265     apr_pool_destroy(p);
 266
 267     if (benice) {
 268         delcount += 2;
 269         if (delcount >= DELETE_NICE) {
 270             apr_sleep(NICE_DELAY);
 271             delcount = 0;
 272         }
 273     }
 274 }
 275
 276 /*
 277  * walk the cache directory tree
 278  */
 279 static int process_dir(char *path, apr_pool_t *pool)
 280 {
 281     apr_dir_t *dir;
 282     apr_pool_t *p;
 283     apr_hash_t *h;
 284     apr_hash_index_t *i;
 285     apr_file_t *fd;
 286     apr_status_t status;
 287     apr_finfo_t info;
 288     apr_size_t len;
 289     apr_time_t current, deviation;
 290     char *nextpath, *base, *ext, *orig_basename;
 291     APR_RING_ENTRY(_direntry) anchor;
 292     DIRENTRY *d, *t, *n;
 293     ENTRY *e;
 294     int skip, retries;
 295     disk_cache_info_t disk_info;
 296
 297     APR_RING_INIT(&anchor, _direntry, link);
 298     apr_pool_create(&p, pool);
 299     h = apr_hash_make(p);
 300     fd = NULL;
 301     skip = 0;
 302     deviation = MAXDEVIATION * APR_USEC_PER_SEC;
 303
 304     if (apr_dir_open(&dir, path, p) != APR_SUCCESS) {
 305         return 1;
 306     }
 307
 308     while (apr_dir_read(&info, 0, dir) == APR_SUCCESS && !interrupted) {
 309         if (!strcmp(info.name, ".") || !strcmp(info.name, "..")) {
 310             continue;
 311         }
 312         d = apr_pcalloc(p, sizeof(DIRENTRY));
 313         d->basename = apr_pstrcat(p, path, "/", info.name, NULL);
 314         APR_RING_INSERT_TAIL(&anchor, d, _direntry, link);
 315     }
 316
 317     apr_dir_close(dir);
 318
 319     if (interrupted) {
 320         return 1;
 321     }
 322
 323     skip = baselen + 1;
 324
 325     for (d = APR_RING_FIRST(&anchor);
 326          !interrupted && d != APR_RING_SENTINEL(&anchor, _direntry, link);
 327          d=n) {
 328         n = APR_RING_NEXT(d, link);
 329         base = strrchr(d->basename, '/');
 330         if (!base++) {
 331             base = d->basename;
 332         }
 333         ext = strchr(base, '.');
 334
 335         /* there may be temporary files which may be gone before
 336          * processing, always skip these if not in realclean mode
 337          */
 338         if (!ext && !realclean) {
 339             if (!strncasecmp(base, AP_TEMPFILE_BASE, AP_TEMPFILE_BASELEN)
 340                 && strlen(base) == AP_TEMPFILE_NAMELEN) {
 341                 continue;
 342             }
 343         }
 344
 345         /* this may look strange but apr_stat() may return errno which
 346          * is system dependent and there may be transient failures,
 347          * so just blindly retry for a short while
 348          */
 349         retries = STAT_ATTEMPTS;
 350         status = APR_SUCCESS;
 351         do {
 352             if (status != APR_SUCCESS) {
 353                 apr_sleep(STAT_DELAY);
 354             }
 355             status = apr_stat(&info, d->basename, DIRINFO, p);
 356         } while (status != APR_SUCCESS && !interrupted && --retries);
 357
 358         /* what may happen here is that apache did create a file which
 359          * we did detect but then does delete the file before we can
 360          * get file information, so if we don't get any file information
 361          * we will ignore the file in this case
 362          */
 363         if (status != APR_SUCCESS) {
 364             if (!realclean && !interrupted) {
 365                 continue;
 366             }
 367             return 1;
 368         }
 369
 370         if (info.filetype == APR_DIR) {
 371             /* Make a copy of the basename, as process_dir modifies it */
 372             orig_basename = apr_pstrdup(pool, d->basename);
 373             if (process_dir(d->basename, pool)) {
 374                 return 1;
 375             }
 376
 377             /* If asked to delete dirs, do so now. We don't care if it fails.
 378              * If it fails, it likely means there was something else there.
 379              */
 380             if (deldirs && !dryrun) {
 381                 apr_dir_remove(orig_basename, pool);
 382             }
 383             continue;
 384         }
 385
 386         if (info.filetype != APR_REG) {
 387             continue;
 388         }
 389
 390         if (!ext) {
 391             if (!strncasecmp(base, AP_TEMPFILE_BASE, AP_TEMPFILE_BASELEN)
 392                 && strlen(base) == AP_TEMPFILE_NAMELEN) {
 393                 d->basename += skip;
 394                 d->type = TEMP;
 395                 d->dsize = info.size;
 396                 apr_hash_set(h, d->basename, APR_HASH_KEY_STRING, d);
 397             }
 398             continue;
 399         }
 400
 401         if (!strcasecmp(ext, CACHE_HEADER_SUFFIX)) {
 402             *ext = '\0';
 403             d->basename += skip;
 404             /* if a user manually creates a '.header' file */
 405             if (d->basename[0] == '\0') {
 406                 continue;
 407             }
 408             t = apr_hash_get(h, d->basename, APR_HASH_KEY_STRING);
 409             if (t) {
 410                 d = t;
 411             }
 412             d->type |= HEADER;
 413             d->htime = info.mtime;
 414             d->hsize = info.size;
 415             apr_hash_set(h, d->basename, APR_HASH_KEY_STRING, d);
 416             continue;
 417         }
 418
 419         if (!strcasecmp(ext, CACHE_DATA_SUFFIX)) {
 420             *ext = '\0';
 421             d->basename += skip;
 422             /* if a user manually creates a '.data' file */
 423             if (d->basename[0] == '\0') {
 424                 continue;
 425             }
 426             t = apr_hash_get(h, d->basename, APR_HASH_KEY_STRING);
 427             if (t) {
 428                 d = t;
 429             }
 430             d->type |= DATA;
 431             d->dtime = info.mtime;
 432             d->dsize = info.size;
 433             apr_hash_set(h, d->basename, APR_HASH_KEY_STRING, d);
 434         }
 435     }
 436
 437     if (interrupted) {
 438         return 1;
 439     }
 440
 441     path[baselen] = '\0';
 442
 443     for (i = apr_hash_first(p, h); i && !interrupted; i = apr_hash_next(i)) {
 444         void *hvalue;
 445         apr_uint32_t format;
 446
 447         apr_hash_this(i, NULL, NULL, &hvalue);
 448         d = hvalue;
 449
 450         switch(d->type) {
 451         case HEADERDATA:
 452             nextpath = apr_pstrcat(p, path, "/", d->basename,
 453                                    CACHE_HEADER_SUFFIX, NULL);
 454             if (apr_file_open(&fd, nextpath, APR_FOPEN_READ | APR_FOPEN_BINARY,
 455                               APR_OS_DEFAULT, p) == APR_SUCCESS) {
 456                 len = sizeof(format);
 457                 if (apr_file_read_full(fd, &format, len,
 458                                        &len) == APR_SUCCESS) {
 459                     if (format == DISK_FORMAT_VERSION) {
 460                         apr_off_t offset = 0;
 461
 462                         apr_file_seek(fd, APR_SET, &offset);
 463
 464                         len = sizeof(disk_cache_info_t);
 465
 466                         if (apr_file_read_full(fd, &disk_info, len,
 467                                                &len) == APR_SUCCESS) {
 468                             apr_file_close(fd);
 469                             e = apr_palloc(pool, sizeof(ENTRY));
 470                             APR_RING_INSERT_TAIL(&root, e, _entry, link);
 471                             e->expire = disk_info.expire;
 472                             e->response_time = disk_info.response_time;
 473                             e->htime = d->htime;
 474                             e->dtime = d->dtime;
 475                             e->hsize = d->hsize;
 476                             e->dsize = d->dsize;
 477                             e->basename = apr_pstrdup(pool, d->basename);
 478                             break;
 479                         }
 480                         else {
 481                             apr_file_close(fd);
 482                         }
 483                     }
 484                     else if (format == VARY_FORMAT_VERSION) {
 485                         /* This must be a URL that added Vary headers later,
 486                          * so kill the orphaned .data file
 487                          */
 488                         apr_file_close(fd);
 489                         apr_file_remove(apr_pstrcat(p, path, "/", d->basename,
 490                                                     CACHE_DATA_SUFFIX, NULL),
 491                                         p);
 492                         break;
 493                     }
 494                 }
 495                 else {
 496                     apr_file_close(fd);
 497                 }
 498
 499             }
 500             /* we have a somehow unreadable headers file which is associated
 501              * with a data file. this may be caused by apache currently
 502              * rewriting the headers file. thus we may delete the file set
 503              * either in realclean mode or if the headers file modification
 504              * timestamp is not within a specified positive or negative offset
 505              * to the current time.
 506              */
 507             current = apr_time_now();
 508             if (realclean || d->htime < current - deviation
 509                 || d->htime > current + deviation) {
 510                 delete_entry(path, d->basename, p);
 511                 unsolicited += d->hsize;
 512                 unsolicited += d->dsize;
 513             }
 514             break;
 515
 516         /* single data and header files may be deleted either in realclean
 517          * mode or if their modification timestamp is not within a
 518          * specified positive or negative offset to the current time.
 519          * this handling is necessary due to possible race conditions
 520          * between apache and this process
 521          */
 522         case HEADER:
 523             current = apr_time_now();
 524             nextpath = apr_pstrcat(p, path, "/", d->basename,
 525                                    CACHE_HEADER_SUFFIX, NULL);
 526             if (apr_file_open(&fd, nextpath, APR_FOPEN_READ | APR_FOPEN_BINARY,
 527                               APR_OS_DEFAULT, p) == APR_SUCCESS) {
 528                 len = sizeof(format);
 529                 if (apr_file_read_full(fd, &format, len,
 530                                        &len) == APR_SUCCESS) {
 531                     if (format == VARY_FORMAT_VERSION) {
 532                         apr_time_t expires;
 533
 534                         len = sizeof(expires);
 535
 536                         if (apr_file_read_full(fd, &expires, len,
 537                                                &len) == APR_SUCCESS) {
 538
 539                             apr_file_close(fd);
 540
 541                             if (expires < current) {
 542                                 delete_entry(path, d->basename, p);
 543                             }
 544                             break;
 545                         }
 546                     }
 547                 }
 548                 apr_file_close(fd);
 549             }
 550
 551             if (realclean || d->htime < current - deviation
 552                 || d->htime > current + deviation) {
 553                 delete_entry(path, d->basename, p);
 554                 unsolicited += d->hsize;
 555             }
 556             break;
 557
 558         case DATA:
 559             current = apr_time_now();
 560             if (realclean || d->dtime < current - deviation
 561                 || d->dtime > current + deviation) {
 562                 delete_entry(path, d->basename, p);
 563                 unsolicited += d->dsize;
 564             }
 565             break;
 566
 567         /* temp files may only be deleted in realclean mode which
 568          * is asserted above if a tempfile is in the hash array
 569          */
 570         case TEMP:
 571             delete_file(path, d->basename, p);
 572             unsolicited += d->dsize;
 573             break;
 574         }
 575     }
 576
 577     if (interrupted) {
 578         return 1;
 579     }
 580
 581     apr_pool_destroy(p);
 582
 583     if (benice) {
 584         apr_sleep(NICE_DELAY);
 585     }
 586
 587     if (interrupted) {
 588         return 1;
 589     }
 590
 591     return 0;
 592 }
 593
 594 /*
 595  * purge cache entries
 596  */
 597 static void purge(char *path, apr_pool_t *pool, apr_off_t max)
 598 {
 599     apr_off_t sum, total, entries, etotal;
 600     ENTRY *e, *n, *oldest;
 601
 602     sum = 0;
 603     entries = 0;
 604
 605     for (e = APR_RING_FIRST(&root);
 606          e != APR_RING_SENTINEL(&root, _entry, link);
 607          e = APR_RING_NEXT(e, link)) {
 608         sum += e->hsize;
 609         sum += e->dsize;
 610         entries++;
 611     }
 612
 613     total = sum;
 614     etotal = entries;
 615
 616     if (sum <= max) {
 617         printstats(total, sum, max, etotal, entries);
 618         return;
 619     }
 620
 621     /* process all entries with a timestamp in the future, this may
 622      * happen if a wrong system time is corrected
 623      */
 624
 625     for (e = APR_RING_FIRST(&root);
 626          e != APR_RING_SENTINEL(&root, _entry, link) && !interrupted;) {
 627         n = APR_RING_NEXT(e, link);
 628         if (e->response_time > now || e->htime > now || e->dtime > now) {
 629             delete_entry(path, e->basename, pool);
 630             sum -= e->hsize;
 631             sum -= e->dsize;
 632             entries--;
 633             APR_RING_REMOVE(e, link);
 634             if (sum <= max) {
 635                 if (!interrupted) {
 636                     printstats(total, sum, max, etotal, entries);
 637                 }
 638                 return;
 639             }
 640         }
 641         e = n;
 642     }
 643
 644     if (interrupted) {
 645         return;
 646     }
 647
 648     /* process all entries with are expired */
 649     for (e = APR_RING_FIRST(&root);
 650          e != APR_RING_SENTINEL(&root, _entry, link) && !interrupted;) {
 651         n = APR_RING_NEXT(e, link);
 652         if (e->expire != APR_DATE_BAD && e->expire < now) {
 653             delete_entry(path, e->basename, pool);
 654             sum -= e->hsize;
 655             sum -= e->dsize;
 656             entries--;
 657             APR_RING_REMOVE(e, link);
 658             if (sum <= max) {
 659                 if (!interrupted) {
 660                     printstats(total, sum, max, etotal, entries);
 661                 }
 662                 return;
 663             }
 664         }
 665         e = n;
 666     }
 667
 668     if (interrupted) {
 669          return;
 670     }
 671
 672     /* process remaining entries oldest to newest, the check for an emtpy
 673      * ring actually isn't necessary except when the compiler does
 674      * corrupt 64bit arithmetics which happend to me once, so better safe
 675      * than sorry
 676      */
 677     while (sum > max && !interrupted && !APR_RING_EMPTY(&root, _entry, link)) {
 678         oldest = APR_RING_FIRST(&root);
 679
 680         for (e = APR_RING_NEXT(oldest, link);
 681              e != APR_RING_SENTINEL(&root, _entry, link);
 682              e = APR_RING_NEXT(e, link)) {
 683             if (e->dtime < oldest->dtime) {
 684                 oldest = e;
 685             }
 686         }
 687
 688         delete_entry(path, oldest->basename, pool);
 689         sum -= oldest->hsize;
 690         sum -= oldest->dsize;
 691         entries--;
 692         APR_RING_REMOVE(oldest, link);
 693     }
 694
 695     if (!interrupted) {
 696         printstats(total, sum, max, etotal, entries);
 697     }
 698 }
 699
 700 /*
 701  * usage info
 702  */
 703 #define NL APR_EOL_STR
 704 static void usage(void)
 705 {
 706     apr_file_printf(errfile,
 707     "%s -- program for cleaning the disk cache."                             NL
 708     "Usage: %s [-Dvtrn] -pPATH -lLIMIT"                                      NL
 709     "       %s [-nti] -dINTERVAL -pPATH -lLIMIT"                             NL
 710                                                                              NL
 711     "Options:"                                                               NL
 712     "  -d   Daemonize and repeat cache cleaning every INTERVAL minutes."     NL
 713     "       This option is mutually exclusive with the -D, -v and -r"        NL
 714     "       options."                                                        NL
 715                                                                              NL
 716     "  -D   Do a dry run and don't delete anything. This option is mutually" NL
 717     "       exclusive with the -d option."                                   NL
 718                                                                              NL
 719     "  -v   Be verbose and print statistics. This option is mutually"        NL
 720     "       exclusive with the -d option."                                   NL
 721                                                                              NL
 722     "  -r   Clean thoroughly. This assumes that the Apache web server is "   NL
 723     "       not running. This option is mutually exclusive with the -d"      NL
 724     "       option and implies -t."                                          NL
 725                                                                              NL
 726     "  -n   Be nice. This causes slower processing in favour of other"       NL
 727     "       processes."                                                      NL
 728                                                                              NL
 729     "  -t   Delete all empty directories. By default only cache files are"   NL
 730     "       removed, however with some configurations the large number of"   NL
 731     "       directories created may require attention."                      NL
 732                                                                              NL
 733     "  -p   Specify PATH as the root directory of the disk cache."           NL
 734                                                                              NL
 735     "  -l   Specify LIMIT as the total disk cache size limit. Attach 'K'"    NL
 736     "       or 'M' to the number for specifying KBytes or MBytes."           NL
 737                                                                              NL
 738     "  -i   Be intelligent and run only when there was a modification of"    NL
 739     "       the disk cache. This option is only possible together with the"  NL
 740     "       -d option."                                                      NL,
 741     shortname,
 742     shortname,
 743     shortname
 744     );
 745
 746     exit(1);
 747 }
 748 #undef NL
 749
 750 /*
 751  * main
 752  */
 753 int main(int argc, const char * const argv[])
 754 {
 755     apr_off_t max;
 756     apr_time_t current, repeat, delay, previous;
 757     apr_status_t status;
 758     apr_pool_t *pool, *instance;
 759     apr_getopt_t *o;
 760     apr_finfo_t info;
 761     int retries, isdaemon, limit_found, intelligent, dowork;
 762     char opt;
 763     const char *arg;
 764     char *proxypath, *path;
 765
 766     interrupted = 0;
 767     repeat = 0;
 768     isdaemon = 0;
 769     dryrun = 0;
 770     limit_found = 0;
 771     max = 0;
 772     verbose = 0;
 773     realclean = 0;
 774     benice = 0;
 775     deldirs = 0;
 776     intelligent = 0;
 777     previous = 0; /* avoid compiler warning */
 778     proxypath = NULL;
 779
 780     if (apr_app_initialize(&argc, &argv, NULL) != APR_SUCCESS) {
 781         return 1;
 782     }
 783     atexit(apr_terminate);
 784
 785     if (argc) {
 786         shortname = apr_filepath_name_get(argv[0]);
 787     }
 788
 789     if (apr_pool_create(&pool, NULL) != APR_SUCCESS) {
 790         return 1;
 791     }
 792     apr_pool_abort_set(oom, pool);
 793     apr_file_open_stderr(&errfile, pool);
 794     apr_signal(SIGINT, setterm);
 795     apr_signal(SIGTERM, setterm);
 796
 797     apr_getopt_init(&o, pool, argc, argv);
 798
 799     while (1) {
 800         status = apr_getopt(o, "iDnvrtd:l:L:p:", &opt, &arg);
 801         if (status == APR_EOF) {
 802             break;
 803         }
 804         else if (status != APR_SUCCESS) {
 805             usage();
 806         }
 807         else {
 808             switch (opt) {
 809             case 'i':
 810                 if (intelligent) {
 811                     usage();
 812                 }
 813                 intelligent = 1;
 814                 break;
 815
 816             case 'D':
 817                 if (dryrun) {
 818                     usage();
 819                 }
 820                 dryrun = 1;
 821                 break;
 822
 823             case 'n':
 824                 if (benice) {
 825                     usage();
 826                 }
 827                 benice = 1;
 828                 break;
 829
 830             case 't':
 831                 if (deldirs) {
 832                     usage();
 833                 }
 834                 deldirs = 1;
 835                 break;
 836
 837             case 'v':
 838                 if (verbose) {
 839                     usage();
 840                 }
 841                 verbose = 1;
 842                 break;
 843
 844             case 'r':
 845                 if (realclean) {
 846                     usage();
 847                 }
 848                 realclean = 1;
 849                 deldirs = 1;
 850                 break;
 851
 852             case 'd':
 853                 if (isdaemon) {
 854                     usage();
 855                 }
 856                 isdaemon = 1;
 857                 repeat = apr_atoi64(arg);
 858                 repeat *= SECS_PER_MIN;
 859                 repeat *= APR_USEC_PER_SEC;
 860                 break;
 861
 862             case 'l':
 863                 if (limit_found) {
 864                     usage();
 865                 }
 866                 limit_found = 1;
 867
 868                 do {
 869                     apr_status_t rv;
 870                     char *end;
 871
 872                     rv = apr_strtoff(&max, arg, &end, 10);
 873                     if (rv == APR_SUCCESS) {
 874                         if ((*end == 'K' || *end == 'k') && !end[1]) {
 875                             max *= KBYTE;
 876                         }
 877                         else if ((*end == 'M' || *end == 'm') && !end[1]) {
 878                             max *= MBYTE;
 879                         }
 880                         else if ((*end == 'G' || *end == 'g') && !end[1]) {
 881                             max *= GBYTE;
 882                         }
 883                         else if (*end &&        /* neither empty nor [Bb] */
 884                                  ((*end != 'B' && *end != 'b') || end[1])) {
 885                             rv = APR_EGENERAL;
 886                         }
 887                     }
 888                     if (rv != APR_SUCCESS) {
 889                         apr_file_printf(errfile, "Invalid limit: %s"
 890                                                  APR_EOL_STR APR_EOL_STR, arg);
 891                         usage();
 892                     }
 893                 } while(0);
 894                 break;
 895
 896             case 'p':
 897                 if (proxypath) {
 898                     usage();
 899                 }
 900                 proxypath = apr_pstrdup(pool, arg);
 901                 if (apr_filepath_set(proxypath, pool) != APR_SUCCESS) {
 902                     usage();
 903                 }
 904                 break;
 905             } /* switch */
 906         } /* else */
 907     } /* while */
 908
 909     if (o->ind != argc) {
 910          usage();
 911     }
 912
 913     if (isdaemon && (repeat <= 0 || verbose || realclean || dryrun)) {
 914          usage();
 915     }
 916
 917     if (!isdaemon && intelligent) {
 918          usage();
 919     }
 920
 921     if (!proxypath || max <= 0) {
 922          usage();
 923     }
 924
 925     if (apr_filepath_get(&path, 0, pool) != APR_SUCCESS) {
 926         usage();
 927     }
 928     baselen = strlen(path);
 929
 930 #ifndef DEBUG
 931     if (isdaemon) {
 932         apr_file_close(errfile);
 933         apr_proc_detach(APR_PROC_DETACH_DAEMONIZE);
 934     }
 935 #endif
 936
 937     do {
 938         apr_pool_create(&instance, pool);
 939
 940         now = apr_time_now();
 941         APR_RING_INIT(&root, _entry, link);
 942         delcount = 0;
 943         unsolicited = 0;
 944         dowork = 0;
 945
 946         switch (intelligent) {
 947         case 0:
 948             dowork = 1;
 949             break;
 950
 951         case 1:
 952             retries = STAT_ATTEMPTS;
 953             status = APR_SUCCESS;
 954
 955             do {
 956                 if (status != APR_SUCCESS) {
 957                     apr_sleep(STAT_DELAY);
 958                 }
 959                 status = apr_stat(&info, path, APR_FINFO_MTIME, instance);
 960             } while (status != APR_SUCCESS && !interrupted && --retries);
 961
 962             if (status == APR_SUCCESS) {
 963                 previous = info.mtime;
 964                 intelligent = 2;
 965             }
 966             dowork = 1;
 967             break;
 968
 969         case 2:
 970             retries = STAT_ATTEMPTS;
 971             status = APR_SUCCESS;
 972
 973             do {
 974                 if (status != APR_SUCCESS) {
 975                     apr_sleep(STAT_DELAY);
 976                 }
 977                 status = apr_stat(&info, path, APR_FINFO_MTIME, instance);
 978             } while (status != APR_SUCCESS && !interrupted && --retries);
 979
 980             if (status == APR_SUCCESS) {
 981                 if (previous != info.mtime) {
 982                     dowork = 1;
 983                 }
 984                 previous = info.mtime;
 985                 break;
 986             }
 987             intelligent = 1;
 988             dowork = 1;
 989             break;
 990         }
 991
 992         if (dowork && !interrupted) {
 993             if (!process_dir(path, instance) && !interrupted) {
 994                 purge(path, instance, max);
 995             }
 996             else if (!isdaemon && !interrupted) {
 997                 apr_file_printf(errfile, "An error occurred, cache cleaning "
 998                                          "aborted." APR_EOL_STR);
 999                 return 1;
1000             }
1001
1002             if (intelligent && !interrupted) {
1003                 retries = STAT_ATTEMPTS;
1004                 status = APR_SUCCESS;
1005                 do {
1006                     if (status != APR_SUCCESS) {
1007                         apr_sleep(STAT_DELAY);
1008                     }
1009                     status = apr_stat(&info, path, APR_FINFO_MTIME, instance);
1010                 } while (status != APR_SUCCESS && !interrupted && --retries);
1011
1012                 if (status == APR_SUCCESS) {
1013                     previous = info.mtime;
1014                     intelligent = 2;
1015                 }
1016                 else {
1017                     intelligent = 1;
1018                 }
1019             }
1020         }
1021
1022         apr_pool_destroy(instance);
1023
1024         current = apr_time_now();
1025         if (current < now) {
1026             delay = repeat;
1027         }
1028         else if (current - now >= repeat) {
1029             delay = repeat;
1030         }
1031         else {
1032             delay = now + repeat - current;
1033         }
1034
1035         /* we can't sleep the whole delay time here apiece as this is racy
1036          * with respect to interrupt delivery - think about what happens
1037          * if we have tested for an interrupt, then get scheduled
1038          * before the apr_sleep() call and while waiting for the cpu
1039          * we do get an interrupt
1040          */
1041         if (isdaemon) {
1042             while (delay && !interrupted) {
1043                 if (delay > APR_USEC_PER_SEC) {
1044                     apr_sleep(APR_USEC_PER_SEC);
1045                     delay -= APR_USEC_PER_SEC;
1046                 }
1047                 else {
1048                     apr_sleep(delay);
1049                     delay = 0;
1050                 }
1051             }
1052         }
1053     } while (isdaemon && !interrupted);
1054
1055     if (!isdaemon && interrupted) {
1056         apr_file_printf(errfile, "Cache cleaning aborted due to user "
1057                                  "request." APR_EOL_STR);
1058         return 1;
1059     }
1060
1061     return 0;
1062 }