git.oblomov.eu Git - linux-2.6/blob - net/sunrpc/cache.c

   1 /*
   2  * net/sunrpc/cache.c
   3  *
   4  * Generic code for various authentication-related caches
   5  * used by sunrpc clients and servers.
   6  *
   7  * Copyright (C) 2002 Neil Brown <neilb@cse.unsw.edu.au>
   8  *
   9  * Released under terms in GPL version 2.  See COPYING.
  10  *
  11  */
  12
  13 #include <linux/types.h>
  14 #include <linux/fs.h>
  15 #include <linux/file.h>
  16 #include <linux/slab.h>
  17 #include <linux/signal.h>
  18 #include <linux/sched.h>
  19 #include <linux/kmod.h>
  20 #include <linux/list.h>
  21 #include <linux/module.h>
  22 #include <linux/ctype.h>
  23 #include <asm/uaccess.h>
  24 #include <linux/poll.h>
  25 #include <linux/seq_file.h>
  26 #include <linux/proc_fs.h>
  27 #include <linux/net.h>
  28 #include <linux/workqueue.h>
  29 #include <linux/mutex.h>
  30 #include <asm/ioctls.h>
  31 #include <linux/sunrpc/types.h>
  32 #include <linux/sunrpc/cache.h>
  33 #include <linux/sunrpc/stats.h>
  34
  35 #define  RPCDBG_FACILITY RPCDBG_CACHE
  36
  37 static int cache_defer_req(struct cache_req *req, struct cache_head *item);
  38 static void cache_revisit_request(struct cache_head *item);
  39
  40 static void cache_init(struct cache_head *h)
  41 {
  42         time_t now = get_seconds();
  43         h->next = NULL;
  44         h->flags = 0;
  45         kref_init(&h->ref);
  46         h->expiry_time = now + CACHE_NEW_EXPIRY;
  47         h->last_refresh = now;
  48 }
  49
  50 struct cache_head *sunrpc_cache_lookup(struct cache_detail *detail,
  51                                        struct cache_head *key, int hash)
  52 {
  53         struct cache_head **head,  **hp;
  54         struct cache_head *new = NULL;
  55
  56         head = &detail->hash_table[hash];
  57
  58         read_lock(&detail->hash_lock);
  59
  60         for (hp=head; *hp != NULL ; hp = &(*hp)->next) {
  61                 struct cache_head *tmp = *hp;
  62                 if (detail->match(tmp, key)) {
  63                         cache_get(tmp);
  64                         read_unlock(&detail->hash_lock);
  65                         return tmp;
  66                 }
  67         }
  68         read_unlock(&detail->hash_lock);
  69         /* Didn't find anything, insert an empty entry */
  70
  71         new = detail->alloc();
  72         if (!new)
  73                 return NULL;
  74         /* must fully initialise 'new', else
  75          * we might get lose if we need to
  76          * cache_put it soon.
  77          */
  78         cache_init(new);
  79         detail->init(new, key);
  80
  81         write_lock(&detail->hash_lock);
  82
  83         /* check if entry appeared while we slept */
  84         for (hp=head; *hp != NULL ; hp = &(*hp)->next) {
  85                 struct cache_head *tmp = *hp;
  86                 if (detail->match(tmp, key)) {
  87                         cache_get(tmp);
  88                         write_unlock(&detail->hash_lock);
  89                         cache_put(new, detail);
  90                         return tmp;
  91                 }
  92         }
  93         new->next = *head;
  94         *head = new;
  95         detail->entries++;
  96         cache_get(new);
  97         write_unlock(&detail->hash_lock);
  98
  99         return new;
 100 }
 101 EXPORT_SYMBOL_GPL(sunrpc_cache_lookup);
 102
 103
 104 static void queue_loose(struct cache_detail *detail, struct cache_head *ch);
 105
 106 static int cache_fresh_locked(struct cache_head *head, time_t expiry)
 107 {
 108         head->expiry_time = expiry;
 109         head->last_refresh = get_seconds();
 110         return !test_and_set_bit(CACHE_VALID, &head->flags);
 111 }
 112
 113 static void cache_fresh_unlocked(struct cache_head *head,
 114                         struct cache_detail *detail, int new)
 115 {
 116         if (new)
 117                 cache_revisit_request(head);
 118         if (test_and_clear_bit(CACHE_PENDING, &head->flags)) {
 119                 cache_revisit_request(head);
 120                 queue_loose(detail, head);
 121         }
 122 }
 123
 124 struct cache_head *sunrpc_cache_update(struct cache_detail *detail,
 125                                        struct cache_head *new, struct cache_head *old, int hash)
 126 {
 127         /* The 'old' entry is to be replaced by 'new'.
 128          * If 'old' is not VALID, we update it directly,
 129          * otherwise we need to replace it
 130          */
 131         struct cache_head **head;
 132         struct cache_head *tmp;
 133         int is_new;
 134
 135         if (!test_bit(CACHE_VALID, &old->flags)) {
 136                 write_lock(&detail->hash_lock);
 137                 if (!test_bit(CACHE_VALID, &old->flags)) {
 138                         if (test_bit(CACHE_NEGATIVE, &new->flags))
 139                                 set_bit(CACHE_NEGATIVE, &old->flags);
 140                         else
 141                                 detail->update(old, new);
 142                         is_new = cache_fresh_locked(old, new->expiry_time);
 143                         write_unlock(&detail->hash_lock);
 144                         cache_fresh_unlocked(old, detail, is_new);
 145                         return old;
 146                 }
 147                 write_unlock(&detail->hash_lock);
 148         }
 149         /* We need to insert a new entry */
 150         tmp = detail->alloc();
 151         if (!tmp) {
 152                 cache_put(old, detail);
 153                 return NULL;
 154         }
 155         cache_init(tmp);
 156         detail->init(tmp, old);
 157         head = &detail->hash_table[hash];
 158
 159         write_lock(&detail->hash_lock);
 160         if (test_bit(CACHE_NEGATIVE, &new->flags))
 161                 set_bit(CACHE_NEGATIVE, &tmp->flags);
 162         else
 163                 detail->update(tmp, new);
 164         tmp->next = *head;
 165         *head = tmp;
 166         detail->entries++;
 167         cache_get(tmp);
 168         is_new = cache_fresh_locked(tmp, new->expiry_time);
 169         cache_fresh_locked(old, 0);
 170         write_unlock(&detail->hash_lock);
 171         cache_fresh_unlocked(tmp, detail, is_new);
 172         cache_fresh_unlocked(old, detail, 0);
 173         cache_put(old, detail);
 174         return tmp;
 175 }
 176 EXPORT_SYMBOL_GPL(sunrpc_cache_update);
 177
 178 static int cache_make_upcall(struct cache_detail *detail, struct cache_head *h);
 179 /*
 180  * This is the generic cache management routine for all
 181  * the authentication caches.
 182  * It checks the currency of a cache item and will (later)
 183  * initiate an upcall to fill it if needed.
 184  *
 185  *
 186  * Returns 0 if the cache_head can be used, or cache_puts it and returns
 187  * -EAGAIN if upcall is pending,
 188  * -ETIMEDOUT if upcall failed and should be retried,
 189  * -ENOENT if cache entry was negative
 190  */
 191 int cache_check(struct cache_detail *detail,
 192                     struct cache_head *h, struct cache_req *rqstp)
 193 {
 194         int rv;
 195         long refresh_age, age;
 196
 197         /* First decide return status as best we can */
 198         if (!test_bit(CACHE_VALID, &h->flags) ||
 199             h->expiry_time < get_seconds())
 200                 rv = -EAGAIN;
 201         else if (detail->flush_time > h->last_refresh)
 202                 rv = -EAGAIN;
 203         else {
 204                 /* entry is valid */
 205                 if (test_bit(CACHE_NEGATIVE, &h->flags))
 206                         rv = -ENOENT;
 207                 else rv = 0;
 208         }
 209
 210         /* now see if we want to start an upcall */
 211         refresh_age = (h->expiry_time - h->last_refresh);
 212         age = get_seconds() - h->last_refresh;
 213
 214         if (rqstp == NULL) {
 215                 if (rv == -EAGAIN)
 216                         rv = -ENOENT;
 217         } else if (rv == -EAGAIN || age > refresh_age/2) {
 218                 dprintk("RPC:       Want update, refage=%ld, age=%ld\n",
 219                                 refresh_age, age);
 220                 if (!test_and_set_bit(CACHE_PENDING, &h->flags)) {
 221                         switch (cache_make_upcall(detail, h)) {
 222                         case -EINVAL:
 223                                 clear_bit(CACHE_PENDING, &h->flags);
 224                                 if (rv == -EAGAIN) {
 225                                         set_bit(CACHE_NEGATIVE, &h->flags);
 226                                         cache_fresh_unlocked(h, detail,
 227                                              cache_fresh_locked(h, get_seconds()+CACHE_NEW_EXPIRY));
 228                                         rv = -ENOENT;
 229                                 }
 230                                 break;
 231
 232                         case -EAGAIN:
 233                                 clear_bit(CACHE_PENDING, &h->flags);
 234                                 cache_revisit_request(h);
 235                                 break;
 236                         }
 237                 }
 238         }
 239
 240         if (rv == -EAGAIN)
 241                 if (cache_defer_req(rqstp, h) != 0)
 242                         rv = -ETIMEDOUT;
 243
 244         if (rv)
 245                 cache_put(h, detail);
 246         return rv;
 247 }
 248 EXPORT_SYMBOL_GPL(cache_check);
 249
 250 /*
 251  * caches need to be periodically cleaned.
 252  * For this we maintain a list of cache_detail and
 253  * a current pointer into that list and into the table
 254  * for that entry.
 255  *
 256  * Each time clean_cache is called it finds the next non-empty entry
 257  * in the current table and walks the list in that entry
 258  * looking for entries that can be removed.
 259  *
 260  * An entry gets removed if:
 261  * - The expiry is before current time
 262  * - The last_refresh time is before the flush_time for that cache
 263  *
 264  * later we might drop old entries with non-NEVER expiry if that table
 265  * is getting 'full' for some definition of 'full'
 266  *
 267  * The question of "how often to scan a table" is an interesting one
 268  * and is answered in part by the use of the "nextcheck" field in the
 269  * cache_detail.
 270  * When a scan of a table begins, the nextcheck field is set to a time
 271  * that is well into the future.
 272  * While scanning, if an expiry time is found that is earlier than the
 273  * current nextcheck time, nextcheck is set to that expiry time.
 274  * If the flush_time is ever set to a time earlier than the nextcheck
 275  * time, the nextcheck time is then set to that flush_time.
 276  *
 277  * A table is then only scanned if the current time is at least
 278  * the nextcheck time.
 279  *
 280  */
 281
 282 static LIST_HEAD(cache_list);
 283 static DEFINE_SPINLOCK(cache_list_lock);
 284 static struct cache_detail *current_detail;
 285 static int current_index;
 286
 287 static const struct file_operations cache_file_operations;
 288 static const struct file_operations content_file_operations;
 289 static const struct file_operations cache_flush_operations;
 290
 291 static void do_cache_clean(struct work_struct *work);
 292 static DECLARE_DELAYED_WORK(cache_cleaner, do_cache_clean);
 293
 294 static void remove_cache_proc_entries(struct cache_detail *cd)
 295 {
 296         if (cd->proc_ent == NULL)
 297                 return;
 298         if (cd->flush_ent)
 299                 remove_proc_entry("flush", cd->proc_ent);
 300         if (cd->channel_ent)
 301                 remove_proc_entry("channel", cd->proc_ent);
 302         if (cd->content_ent)
 303                 remove_proc_entry("content", cd->proc_ent);
 304         cd->proc_ent = NULL;
 305         remove_proc_entry(cd->name, proc_net_rpc);
 306 }
 307
 308 #ifdef CONFIG_PROC_FS
 309 static int create_cache_proc_entries(struct cache_detail *cd)
 310 {
 311         struct proc_dir_entry *p;
 312
 313         cd->proc_ent = proc_mkdir(cd->name, proc_net_rpc);
 314         if (cd->proc_ent == NULL)
 315                 goto out_nomem;
 316         cd->channel_ent = cd->content_ent = NULL;
 317
 318         p = proc_create_data("flush", S_IFREG|S_IRUSR|S_IWUSR,
 319                              cd->proc_ent, &cache_flush_operations, cd);
 320         cd->flush_ent = p;
 321         if (p == NULL)
 322                 goto out_nomem;
 323
 324         if (cd->cache_request || cd->cache_parse) {
 325                 p = proc_create_data("channel", S_IFREG|S_IRUSR|S_IWUSR,
 326                                      cd->proc_ent, &cache_file_operations, cd);
 327                 cd->channel_ent = p;
 328                 if (p == NULL)
 329                         goto out_nomem;
 330         }
 331         if (cd->cache_show) {
 332                 p = proc_create_data("content", S_IFREG|S_IRUSR|S_IWUSR,
 333                                 cd->proc_ent, &content_file_operations, cd);
 334                 cd->content_ent = p;
 335                 if (p == NULL)
 336                         goto out_nomem;
 337         }
 338         return 0;
 339 out_nomem:
 340         remove_cache_proc_entries(cd);
 341         return -ENOMEM;
 342 }
 343 #else /* CONFIG_PROC_FS */
 344 static int create_cache_proc_entries(struct cache_detail *cd)
 345 {
 346         return 0;
 347 }
 348 #endif
 349
 350 int cache_register(struct cache_detail *cd)
 351 {
 352         int ret;
 353
 354         ret = create_cache_proc_entries(cd);
 355         if (ret)
 356                 return ret;
 357         rwlock_init(&cd->hash_lock);
 358         INIT_LIST_HEAD(&cd->queue);
 359         spin_lock(&cache_list_lock);
 360         cd->nextcheck = 0;
 361         cd->entries = 0;
 362         atomic_set(&cd->readers, 0);
 363         cd->last_close = 0;
 364         cd->last_warn = -1;
 365         list_add(&cd->others, &cache_list);
 366         spin_unlock(&cache_list_lock);
 367
 368         /* start the cleaning process */
 369         schedule_delayed_work(&cache_cleaner, 0);
 370         return 0;
 371 }
 372 EXPORT_SYMBOL_GPL(cache_register);
 373
 374 void cache_unregister(struct cache_detail *cd)
 375 {
 376         cache_purge(cd);
 377         spin_lock(&cache_list_lock);
 378         write_lock(&cd->hash_lock);
 379         if (cd->entries || atomic_read(&cd->inuse)) {
 380                 write_unlock(&cd->hash_lock);
 381                 spin_unlock(&cache_list_lock);
 382                 goto out;
 383         }
 384         if (current_detail == cd)
 385                 current_detail = NULL;
 386         list_del_init(&cd->others);
 387         write_unlock(&cd->hash_lock);
 388         spin_unlock(&cache_list_lock);
 389         remove_cache_proc_entries(cd);
 390         if (list_empty(&cache_list)) {
 391                 /* module must be being unloaded so its safe to kill the worker */
 392                 cancel_delayed_work_sync(&cache_cleaner);
 393         }
 394         return;
 395 out:
 396         printk(KERN_ERR "nfsd: failed to unregister %s cache\n", cd->name);
 397 }
 398 EXPORT_SYMBOL_GPL(cache_unregister);
 399
 400 /* clean cache tries to find something to clean
 401  * and cleans it.
 402  * It returns 1 if it cleaned something,
 403  *            0 if it didn't find anything this time
 404  *           -1 if it fell off the end of the list.
 405  */
 406 static int cache_clean(void)
 407 {
 408         int rv = 0;
 409         struct list_head *next;
 410
 411         spin_lock(&cache_list_lock);
 412
 413         /* find a suitable table if we don't already have one */
 414         while (current_detail == NULL ||
 415             current_index >= current_detail->hash_size) {
 416                 if (current_detail)
 417                         next = current_detail->others.next;
 418                 else
 419                         next = cache_list.next;
 420                 if (next == &cache_list) {
 421                         current_detail = NULL;
 422                         spin_unlock(&cache_list_lock);
 423                         return -1;
 424                 }
 425                 current_detail = list_entry(next, struct cache_detail, others);
 426                 if (current_detail->nextcheck > get_seconds())
 427                         current_index = current_detail->hash_size;
 428                 else {
 429                         current_index = 0;
 430                         current_detail->nextcheck = get_seconds()+30*60;
 431                 }
 432         }
 433
 434         /* find a non-empty bucket in the table */
 435         while (current_detail &&
 436                current_index < current_detail->hash_size &&
 437                current_detail->hash_table[current_index] == NULL)
 438                 current_index++;
 439
 440         /* find a cleanable entry in the bucket and clean it, or set to next bucket */
 441
 442         if (current_detail && current_index < current_detail->hash_size) {
 443                 struct cache_head *ch, **cp;
 444                 struct cache_detail *d;
 445
 446                 write_lock(&current_detail->hash_lock);
 447
 448                 /* Ok, now to clean this strand */
 449
 450                 cp = & current_detail->hash_table[current_index];
 451                 ch = *cp;
 452                 for (; ch; cp= & ch->next, ch= *cp) {
 453                         if (current_detail->nextcheck > ch->expiry_time)
 454                                 current_detail->nextcheck = ch->expiry_time+1;
 455                         if (ch->expiry_time >= get_seconds()
 456                             && ch->last_refresh >= current_detail->flush_time
 457                                 )
 458                                 continue;
 459                         if (test_and_clear_bit(CACHE_PENDING, &ch->flags))
 460                                 queue_loose(current_detail, ch);
 461
 462                         if (atomic_read(&ch->ref.refcount) == 1)
 463                                 break;
 464                 }
 465                 if (ch) {
 466                         *cp = ch->next;
 467                         ch->next = NULL;
 468                         current_detail->entries--;
 469                         rv = 1;
 470                 }
 471                 write_unlock(&current_detail->hash_lock);
 472                 d = current_detail;
 473                 if (!ch)
 474                         current_index ++;
 475                 spin_unlock(&cache_list_lock);
 476                 if (ch)
 477                         cache_put(ch, d);
 478         } else
 479                 spin_unlock(&cache_list_lock);
 480
 481         return rv;
 482 }
 483
 484 /*
 485  * We want to regularly clean the cache, so we need to schedule some work ...
 486  */
 487 static void do_cache_clean(struct work_struct *work)
 488 {
 489         int delay = 5;
 490         if (cache_clean() == -1)
 491                 delay = 30*HZ;
 492
 493         if (list_empty(&cache_list))
 494                 delay = 0;
 495
 496         if (delay)
 497                 schedule_delayed_work(&cache_cleaner, delay);
 498 }
 499
 500
 501 /*
 502  * Clean all caches promptly.  This just calls cache_clean
 503  * repeatedly until we are sure that every cache has had a chance to
 504  * be fully cleaned
 505  */
 506 void cache_flush(void)
 507 {
 508         while (cache_clean() != -1)
 509                 cond_resched();
 510         while (cache_clean() != -1)
 511                 cond_resched();
 512 }
 513 EXPORT_SYMBOL_GPL(cache_flush);
 514
 515 void cache_purge(struct cache_detail *detail)
 516 {
 517         detail->flush_time = LONG_MAX;
 518         detail->nextcheck = get_seconds();
 519         cache_flush();
 520         detail->flush_time = 1;
 521 }
 522 EXPORT_SYMBOL_GPL(cache_purge);
 523
 524
 525 /*
 526  * Deferral and Revisiting of Requests.
 527  *
 528  * If a cache lookup finds a pending entry, we
 529  * need to defer the request and revisit it later.
 530  * All deferred requests are stored in a hash table,
 531  * indexed by "struct cache_head *".
 532  * As it may be wasteful to store a whole request
 533  * structure, we allow the request to provide a
 534  * deferred form, which must contain a
 535  * 'struct cache_deferred_req'
 536  * This cache_deferred_req contains a method to allow
 537  * it to be revisited when cache info is available
 538  */
 539
 540 #define DFR_HASHSIZE    (PAGE_SIZE/sizeof(struct list_head))
 541 #define DFR_HASH(item)  ((((long)item)>>4 ^ (((long)item)>>13)) % DFR_HASHSIZE)
 542
 543 #define DFR_MAX 300     /* ??? */
 544
 545 static DEFINE_SPINLOCK(cache_defer_lock);
 546 static LIST_HEAD(cache_defer_list);
 547 static struct list_head cache_defer_hash[DFR_HASHSIZE];
 548 static int cache_defer_cnt;
 549
 550 static int cache_defer_req(struct cache_req *req, struct cache_head *item)
 551 {
 552         struct cache_deferred_req *dreq;
 553         int hash = DFR_HASH(item);
 554
 555         if (cache_defer_cnt >= DFR_MAX) {
 556                 /* too much in the cache, randomly drop this one,
 557                  * or continue and drop the oldest below
 558                  */
 559                 if (net_random()&1)
 560                         return -ETIMEDOUT;
 561         }
 562         dreq = req->defer(req);
 563         if (dreq == NULL)
 564                 return -ETIMEDOUT;
 565
 566         dreq->item = item;
 567
 568         spin_lock(&cache_defer_lock);
 569
 570         list_add(&dreq->recent, &cache_defer_list);
 571
 572         if (cache_defer_hash[hash].next == NULL)
 573                 INIT_LIST_HEAD(&cache_defer_hash[hash]);
 574         list_add(&dreq->hash, &cache_defer_hash[hash]);
 575
 576         /* it is in, now maybe clean up */
 577         dreq = NULL;
 578         if (++cache_defer_cnt > DFR_MAX) {
 579                 dreq = list_entry(cache_defer_list.prev,
 580                                   struct cache_deferred_req, recent);
 581                 list_del(&dreq->recent);
 582                 list_del(&dreq->hash);
 583                 cache_defer_cnt--;
 584         }
 585         spin_unlock(&cache_defer_lock);
 586
 587         if (dreq) {
 588                 /* there was one too many */
 589                 dreq->revisit(dreq, 1);
 590         }
 591         if (!test_bit(CACHE_PENDING, &item->flags)) {
 592                 /* must have just been validated... */
 593                 cache_revisit_request(item);
 594         }
 595         return 0;
 596 }
 597
 598 static void cache_revisit_request(struct cache_head *item)
 599 {
 600         struct cache_deferred_req *dreq;
 601         struct list_head pending;
 602
 603         struct list_head *lp;
 604         int hash = DFR_HASH(item);
 605
 606         INIT_LIST_HEAD(&pending);
 607         spin_lock(&cache_defer_lock);
 608
 609         lp = cache_defer_hash[hash].next;
 610         if (lp) {
 611                 while (lp != &cache_defer_hash[hash]) {
 612                         dreq = list_entry(lp, struct cache_deferred_req, hash);
 613                         lp = lp->next;
 614                         if (dreq->item == item) {
 615                                 list_del(&dreq->hash);
 616                                 list_move(&dreq->recent, &pending);
 617                                 cache_defer_cnt--;
 618                         }
 619                 }
 620         }
 621         spin_unlock(&cache_defer_lock);
 622
 623         while (!list_empty(&pending)) {
 624                 dreq = list_entry(pending.next, struct cache_deferred_req, recent);
 625                 list_del_init(&dreq->recent);
 626                 dreq->revisit(dreq, 0);
 627         }
 628 }
 629
 630 void cache_clean_deferred(void *owner)
 631 {
 632         struct cache_deferred_req *dreq, *tmp;
 633         struct list_head pending;
 634
 635
 636         INIT_LIST_HEAD(&pending);
 637         spin_lock(&cache_defer_lock);
 638
 639         list_for_each_entry_safe(dreq, tmp, &cache_defer_list, recent) {
 640                 if (dreq->owner == owner) {
 641                         list_del(&dreq->hash);
 642                         list_move(&dreq->recent, &pending);
 643                         cache_defer_cnt--;
 644                 }
 645         }
 646         spin_unlock(&cache_defer_lock);
 647
 648         while (!list_empty(&pending)) {
 649                 dreq = list_entry(pending.next, struct cache_deferred_req, recent);
 650                 list_del_init(&dreq->recent);
 651                 dreq->revisit(dreq, 1);
 652         }
 653 }
 654
 655 /*
 656  * communicate with user-space
 657  *
 658  * We have a magic /proc file - /proc/sunrpc/<cachename>/channel.
 659  * On read, you get a full request, or block.
 660  * On write, an update request is processed.
 661  * Poll works if anything to read, and always allows write.
 662  *
 663  * Implemented by linked list of requests.  Each open file has
 664  * a ->private that also exists in this list.  New requests are added
 665  * to the end and may wakeup and preceding readers.
 666  * New readers are added to the head.  If, on read, an item is found with
 667  * CACHE_UPCALLING clear, we free it from the list.
 668  *
 669  */
 670
 671 static DEFINE_SPINLOCK(queue_lock);
 672 static DEFINE_MUTEX(queue_io_mutex);
 673
 674 struct cache_queue {
 675         struct list_head        list;
 676         int                     reader; /* if 0, then request */
 677 };
 678 struct cache_request {
 679         struct cache_queue      q;
 680         struct cache_head       *item;
 681         char                    * buf;
 682         int                     len;
 683         int                     readers;
 684 };
 685 struct cache_reader {
 686         struct cache_queue      q;
 687         int                     offset; /* if non-0, we have a refcnt on next request */
 688 };
 689
 690 static ssize_t
 691 cache_read(struct file *filp, char __user *buf, size_t count, loff_t *ppos)
 692 {
 693         struct cache_reader *rp = filp->private_data;
 694         struct cache_request *rq;
 695         struct cache_detail *cd = PDE(filp->f_path.dentry->d_inode)->data;
 696         int err;
 697
 698         if (count == 0)
 699                 return 0;
 700
 701         mutex_lock(&queue_io_mutex); /* protect against multiple concurrent
 702                               * readers on this file */
 703  again:
 704         spin_lock(&queue_lock);
 705         /* need to find next request */
 706         while (rp->q.list.next != &cd->queue &&
 707                list_entry(rp->q.list.next, struct cache_queue, list)
 708                ->reader) {
 709                 struct list_head *next = rp->q.list.next;
 710                 list_move(&rp->q.list, next);
 711         }
 712         if (rp->q.list.next == &cd->queue) {
 713                 spin_unlock(&queue_lock);
 714                 mutex_unlock(&queue_io_mutex);
 715                 BUG_ON(rp->offset);
 716                 return 0;
 717         }
 718         rq = container_of(rp->q.list.next, struct cache_request, q.list);
 719         BUG_ON(rq->q.reader);
 720         if (rp->offset == 0)
 721                 rq->readers++;
 722         spin_unlock(&queue_lock);
 723
 724         if (rp->offset == 0 && !test_bit(CACHE_PENDING, &rq->item->flags)) {
 725                 err = -EAGAIN;
 726                 spin_lock(&queue_lock);
 727                 list_move(&rp->q.list, &rq->q.list);
 728                 spin_unlock(&queue_lock);
 729         } else {
 730                 if (rp->offset + count > rq->len)
 731                         count = rq->len - rp->offset;
 732                 err = -EFAULT;
 733                 if (copy_to_user(buf, rq->buf + rp->offset, count))
 734                         goto out;
 735                 rp->offset += count;
 736                 if (rp->offset >= rq->len) {
 737                         rp->offset = 0;
 738                         spin_lock(&queue_lock);
 739                         list_move(&rp->q.list, &rq->q.list);
 740                         spin_unlock(&queue_lock);
 741                 }
 742                 err = 0;
 743         }
 744  out:
 745         if (rp->offset == 0) {
 746                 /* need to release rq */
 747                 spin_lock(&queue_lock);
 748                 rq->readers--;
 749                 if (rq->readers == 0 &&
 750                     !test_bit(CACHE_PENDING, &rq->item->flags)) {
 751                         list_del(&rq->q.list);
 752                         spin_unlock(&queue_lock);
 753                         cache_put(rq->item, cd);
 754                         kfree(rq->buf);
 755                         kfree(rq);
 756                 } else
 757                         spin_unlock(&queue_lock);
 758         }
 759         if (err == -EAGAIN)
 760                 goto again;
 761         mutex_unlock(&queue_io_mutex);
 762         return err ? err :  count;
 763 }
 764
 765 static char write_buf[8192]; /* protected by queue_io_mutex */
 766
 767 static ssize_t
 768 cache_write(struct file *filp, const char __user *buf, size_t count,
 769             loff_t *ppos)
 770 {
 771         int err;
 772         struct cache_detail *cd = PDE(filp->f_path.dentry->d_inode)->data;
 773
 774         if (count == 0)
 775                 return 0;
 776         if (count >= sizeof(write_buf))
 777                 return -EINVAL;
 778
 779         mutex_lock(&queue_io_mutex);
 780
 781         if (copy_from_user(write_buf, buf, count)) {
 782                 mutex_unlock(&queue_io_mutex);
 783                 return -EFAULT;
 784         }
 785         write_buf[count] = '\0';
 786         if (cd->cache_parse)
 787                 err = cd->cache_parse(cd, write_buf, count);
 788         else
 789                 err = -EINVAL;
 790
 791         mutex_unlock(&queue_io_mutex);
 792         return err ? err : count;
 793 }
 794
 795 static DECLARE_WAIT_QUEUE_HEAD(queue_wait);
 796
 797 static unsigned int
 798 cache_poll(struct file *filp, poll_table *wait)
 799 {
 800         unsigned int mask;
 801         struct cache_reader *rp = filp->private_data;
 802         struct cache_queue *cq;
 803         struct cache_detail *cd = PDE(filp->f_path.dentry->d_inode)->data;
 804
 805         poll_wait(filp, &queue_wait, wait);
 806
 807         /* alway allow write */
 808         mask = POLL_OUT | POLLWRNORM;
 809
 810         if (!rp)
 811                 return mask;
 812
 813         spin_lock(&queue_lock);
 814
 815         for (cq= &rp->q; &cq->list != &cd->queue;
 816              cq = list_entry(cq->list.next, struct cache_queue, list))
 817                 if (!cq->reader) {
 818                         mask |= POLLIN | POLLRDNORM;
 819                         break;
 820                 }
 821         spin_unlock(&queue_lock);
 822         return mask;
 823 }
 824
 825 static int
 826 cache_ioctl(struct inode *ino, struct file *filp,
 827             unsigned int cmd, unsigned long arg)
 828 {
 829         int len = 0;
 830         struct cache_reader *rp = filp->private_data;
 831         struct cache_queue *cq;
 832         struct cache_detail *cd = PDE(ino)->data;
 833
 834         if (cmd != FIONREAD || !rp)
 835                 return -EINVAL;
 836
 837         spin_lock(&queue_lock);
 838
 839         /* only find the length remaining in current request,
 840          * or the length of the next request
 841          */
 842         for (cq= &rp->q; &cq->list != &cd->queue;
 843              cq = list_entry(cq->list.next, struct cache_queue, list))
 844                 if (!cq->reader) {
 845                         struct cache_request *cr =
 846                                 container_of(cq, struct cache_request, q);
 847                         len = cr->len - rp->offset;
 848                         break;
 849                 }
 850         spin_unlock(&queue_lock);
 851
 852         return put_user(len, (int __user *)arg);
 853 }
 854
 855 static int
 856 cache_open(struct inode *inode, struct file *filp)
 857 {
 858         struct cache_reader *rp = NULL;
 859
 860         nonseekable_open(inode, filp);
 861         if (filp->f_mode & FMODE_READ) {
 862                 struct cache_detail *cd = PDE(inode)->data;
 863
 864                 rp = kmalloc(sizeof(*rp), GFP_KERNEL);
 865                 if (!rp)
 866                         return -ENOMEM;
 867                 rp->offset = 0;
 868                 rp->q.reader = 1;
 869                 atomic_inc(&cd->readers);
 870                 spin_lock(&queue_lock);
 871                 list_add(&rp->q.list, &cd->queue);
 872                 spin_unlock(&queue_lock);
 873         }
 874         filp->private_data = rp;
 875         return 0;
 876 }
 877
 878 static int
 879 cache_release(struct inode *inode, struct file *filp)
 880 {
 881         struct cache_reader *rp = filp->private_data;
 882         struct cache_detail *cd = PDE(inode)->data;
 883
 884         if (rp) {
 885                 spin_lock(&queue_lock);
 886                 if (rp->offset) {
 887                         struct cache_queue *cq;
 888                         for (cq= &rp->q; &cq->list != &cd->queue;
 889                              cq = list_entry(cq->list.next, struct cache_queue, list))
 890                                 if (!cq->reader) {
 891                                         container_of(cq, struct cache_request, q)
 892                                                 ->readers--;
 893                                         break;
 894                                 }
 895                         rp->offset = 0;
 896                 }
 897                 list_del(&rp->q.list);
 898                 spin_unlock(&queue_lock);
 899
 900                 filp->private_data = NULL;
 901                 kfree(rp);
 902
 903                 cd->last_close = get_seconds();
 904                 atomic_dec(&cd->readers);
 905         }
 906         return 0;
 907 }
 908
 909
 910
 911 static const struct file_operations cache_file_operations = {
 912         .owner          = THIS_MODULE,
 913         .llseek         = no_llseek,
 914         .read           = cache_read,
 915         .write          = cache_write,
 916         .poll           = cache_poll,
 917         .ioctl          = cache_ioctl, /* for FIONREAD */
 918         .open           = cache_open,
 919         .release        = cache_release,
 920 };
 921
 922
 923 static void queue_loose(struct cache_detail *detail, struct cache_head *ch)
 924 {
 925         struct cache_queue *cq;
 926         spin_lock(&queue_lock);
 927         list_for_each_entry(cq, &detail->queue, list)
 928                 if (!cq->reader) {
 929                         struct cache_request *cr = container_of(cq, struct cache_request, q);
 930                         if (cr->item != ch)
 931                                 continue;
 932                         if (cr->readers != 0)
 933                                 continue;
 934                         list_del(&cr->q.list);
 935                         spin_unlock(&queue_lock);
 936                         cache_put(cr->item, detail);
 937                         kfree(cr->buf);
 938                         kfree(cr);
 939                         return;
 940                 }
 941         spin_unlock(&queue_lock);
 942 }
 943
 944 /*
 945  * Support routines for text-based upcalls.
 946  * Fields are separated by spaces.
 947  * Fields are either mangled to quote space tab newline slosh with slosh
 948  * or a hexified with a leading \x
 949  * Record is terminated with newline.
 950  *
 951  */
 952
 953 void qword_add(char **bpp, int *lp, char *str)
 954 {
 955         char *bp = *bpp;
 956         int len = *lp;
 957         char c;
 958
 959         if (len < 0) return;
 960
 961         while ((c=*str++) && len)
 962                 switch(c) {
 963                 case ' ':
 964                 case '\t':
 965                 case '\n':
 966                 case '\\':
 967                         if (len >= 4) {
 968                                 *bp++ = '\\';
 969                                 *bp++ = '0' + ((c & 0300)>>6);
 970                                 *bp++ = '0' + ((c & 0070)>>3);
 971                                 *bp++ = '0' + ((c & 0007)>>0);
 972                         }
 973                         len -= 4;
 974                         break;
 975                 default:
 976                         *bp++ = c;
 977                         len--;
 978                 }
 979         if (c || len <1) len = -1;
 980         else {
 981                 *bp++ = ' ';
 982                 len--;
 983         }
 984         *bpp = bp;
 985         *lp = len;
 986 }
 987 EXPORT_SYMBOL_GPL(qword_add);
 988
 989 void qword_addhex(char **bpp, int *lp, char *buf, int blen)
 990 {
 991         char *bp = *bpp;
 992         int len = *lp;
 993
 994         if (len < 0) return;
 995
 996         if (len > 2) {
 997                 *bp++ = '\\';
 998                 *bp++ = 'x';
 999                 len -= 2;
1000                 while (blen && len >= 2) {
1001                         unsigned char c = *buf++;
1002                         *bp++ = '0' + ((c&0xf0)>>4) + (c>=0xa0)*('a'-'9'-1);
1003                         *bp++ = '0' + (c&0x0f) + ((c&0x0f)>=0x0a)*('a'-'9'-1);
1004                         len -= 2;
1005                         blen--;
1006                 }
1007         }
1008         if (blen || len<1) len = -1;
1009         else {
1010                 *bp++ = ' ';
1011                 len--;
1012         }
1013         *bpp = bp;
1014         *lp = len;
1015 }
1016 EXPORT_SYMBOL_GPL(qword_addhex);
1017
1018 static void warn_no_listener(struct cache_detail *detail)
1019 {
1020         if (detail->last_warn != detail->last_close) {
1021                 detail->last_warn = detail->last_close;
1022                 if (detail->warn_no_listener)
1023                         detail->warn_no_listener(detail);
1024         }
1025 }
1026
1027 /*
1028  * register an upcall request to user-space.
1029  * Each request is at most one page long.
1030  */
1031 static int cache_make_upcall(struct cache_detail *detail, struct cache_head *h)
1032 {
1033
1034         char *buf;
1035         struct cache_request *crq;
1036         char *bp;
1037         int len;
1038
1039         if (detail->cache_request == NULL)
1040                 return -EINVAL;
1041
1042         if (atomic_read(&detail->readers) == 0 &&
1043             detail->last_close < get_seconds() - 30) {
1044                         warn_no_listener(detail);
1045                         return -EINVAL;
1046         }
1047
1048         buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
1049         if (!buf)
1050                 return -EAGAIN;
1051
1052         crq = kmalloc(sizeof (*crq), GFP_KERNEL);
1053         if (!crq) {
1054                 kfree(buf);
1055                 return -EAGAIN;
1056         }
1057
1058         bp = buf; len = PAGE_SIZE;
1059
1060         detail->cache_request(detail, h, &bp, &len);
1061
1062         if (len < 0) {
1063                 kfree(buf);
1064                 kfree(crq);
1065                 return -EAGAIN;
1066         }
1067         crq->q.reader = 0;
1068         crq->item = cache_get(h);
1069         crq->buf = buf;
1070         crq->len = PAGE_SIZE - len;
1071         crq->readers = 0;
1072         spin_lock(&queue_lock);
1073         list_add_tail(&crq->q.list, &detail->queue);
1074         spin_unlock(&queue_lock);
1075         wake_up(&queue_wait);
1076         return 0;
1077 }
1078
1079 /*
1080  * parse a message from user-space and pass it
1081  * to an appropriate cache
1082  * Messages are, like requests, separated into fields by
1083  * spaces and dequotes as \xHEXSTRING or embedded \nnn octal
1084  *
1085  * Message is
1086  *   reply cachename expiry key ... content....
1087  *
1088  * key and content are both parsed by cache
1089  */
1090
1091 #define isodigit(c) (isdigit(c) && c <= '7')
1092 int qword_get(char **bpp, char *dest, int bufsize)
1093 {
1094         /* return bytes copied, or -1 on error */
1095         char *bp = *bpp;
1096         int len = 0;
1097
1098         while (*bp == ' ') bp++;
1099
1100         if (bp[0] == '\\' && bp[1] == 'x') {
1101                 /* HEX STRING */
1102                 bp += 2;
1103                 while (isxdigit(bp[0]) && isxdigit(bp[1]) && len < bufsize) {
1104                         int byte = isdigit(*bp) ? *bp-'0' : toupper(*bp)-'A'+10;
1105                         bp++;
1106                         byte <<= 4;
1107                         byte |= isdigit(*bp) ? *bp-'0' : toupper(*bp)-'A'+10;
1108                         *dest++ = byte;
1109                         bp++;
1110                         len++;
1111                 }
1112         } else {
1113                 /* text with \nnn octal quoting */
1114                 while (*bp != ' ' && *bp != '\n' && *bp && len < bufsize-1) {
1115                         if (*bp == '\\' &&
1116                             isodigit(bp[1]) && (bp[1] <= '3') &&
1117                             isodigit(bp[2]) &&
1118                             isodigit(bp[3])) {
1119                                 int byte = (*++bp -'0');
1120                                 bp++;
1121                                 byte = (byte << 3) | (*bp++ - '0');
1122                                 byte = (byte << 3) | (*bp++ - '0');
1123                                 *dest++ = byte;
1124                                 len++;
1125                         } else {
1126                                 *dest++ = *bp++;
1127                                 len++;
1128                         }
1129                 }
1130         }
1131
1132         if (*bp != ' ' && *bp != '\n' && *bp != '\0')
1133                 return -1;
1134         while (*bp == ' ') bp++;
1135         *bpp = bp;
1136         *dest = '\0';
1137         return len;
1138 }
1139 EXPORT_SYMBOL_GPL(qword_get);
1140
1141
1142 /*
1143  * support /proc/sunrpc/cache/$CACHENAME/content
1144  * as a seqfile.
1145  * We call ->cache_show passing NULL for the item to
1146  * get a header, then pass each real item in the cache
1147  */
1148
1149 struct handle {
1150         struct cache_detail *cd;
1151 };
1152
1153 static void *c_start(struct seq_file *m, loff_t *pos)
1154         __acquires(cd->hash_lock)
1155 {
1156         loff_t n = *pos;
1157         unsigned hash, entry;
1158         struct cache_head *ch;
1159         struct cache_detail *cd = ((struct handle*)m->private)->cd;
1160
1161
1162         read_lock(&cd->hash_lock);
1163         if (!n--)
1164                 return SEQ_START_TOKEN;
1165         hash = n >> 32;
1166         entry = n & ((1LL<<32) - 1);
1167
1168         for (ch=cd->hash_table[hash]; ch; ch=ch->next)
1169                 if (!entry--)
1170                         return ch;
1171         n &= ~((1LL<<32) - 1);
1172         do {
1173                 hash++;
1174                 n += 1LL<<32;
1175         } while(hash < cd->hash_size &&
1176                 cd->hash_table[hash]==NULL);
1177         if (hash >= cd->hash_size)
1178                 return NULL;
1179         *pos = n+1;
1180         return cd->hash_table[hash];
1181 }
1182
1183 static void *c_next(struct seq_file *m, void *p, loff_t *pos)
1184 {
1185         struct cache_head *ch = p;
1186         int hash = (*pos >> 32);
1187         struct cache_detail *cd = ((struct handle*)m->private)->cd;
1188
1189         if (p == SEQ_START_TOKEN)
1190                 hash = 0;
1191         else if (ch->next == NULL) {
1192                 hash++;
1193                 *pos += 1LL<<32;
1194         } else {
1195                 ++*pos;
1196                 return ch->next;
1197         }
1198         *pos &= ~((1LL<<32) - 1);
1199         while (hash < cd->hash_size &&
1200                cd->hash_table[hash] == NULL) {
1201                 hash++;
1202                 *pos += 1LL<<32;
1203         }
1204         if (hash >= cd->hash_size)
1205                 return NULL;
1206         ++*pos;
1207         return cd->hash_table[hash];
1208 }
1209
1210 static void c_stop(struct seq_file *m, void *p)
1211         __releases(cd->hash_lock)
1212 {
1213         struct cache_detail *cd = ((struct handle*)m->private)->cd;
1214         read_unlock(&cd->hash_lock);
1215 }
1216
1217 static int c_show(struct seq_file *m, void *p)
1218 {
1219         struct cache_head *cp = p;
1220         struct cache_detail *cd = ((struct handle*)m->private)->cd;
1221
1222         if (p == SEQ_START_TOKEN)
1223                 return cd->cache_show(m, cd, NULL);
1224
1225         ifdebug(CACHE)
1226                 seq_printf(m, "# expiry=%ld refcnt=%d flags=%lx\n",
1227                            cp->expiry_time, atomic_read(&cp->ref.refcount), cp->flags);
1228         cache_get(cp);
1229         if (cache_check(cd, cp, NULL))
1230                 /* cache_check does a cache_put on failure */
1231                 seq_printf(m, "# ");
1232         else
1233                 cache_put(cp, cd);
1234
1235         return cd->cache_show(m, cd, cp);
1236 }
1237
1238 static const struct seq_operations cache_content_op = {
1239         .start  = c_start,
1240         .next   = c_next,
1241         .stop   = c_stop,
1242         .show   = c_show,
1243 };
1244
1245 static int content_open(struct inode *inode, struct file *file)
1246 {
1247         struct handle *han;
1248         struct cache_detail *cd = PDE(inode)->data;
1249
1250         han = __seq_open_private(file, &cache_content_op, sizeof(*han));
1251         if (han == NULL)
1252                 return -ENOMEM;
1253
1254         han->cd = cd;
1255         return 0;
1256 }
1257
1258 static const struct file_operations content_file_operations = {
1259         .open           = content_open,
1260         .read           = seq_read,
1261         .llseek         = seq_lseek,
1262         .release        = seq_release_private,
1263 };
1264
1265 static ssize_t read_flush(struct file *file, char __user *buf,
1266                             size_t count, loff_t *ppos)
1267 {
1268         struct cache_detail *cd = PDE(file->f_path.dentry->d_inode)->data;
1269         char tbuf[20];
1270         unsigned long p = *ppos;
1271         size_t len;
1272
1273         sprintf(tbuf, "%lu\n", cd->flush_time);
1274         len = strlen(tbuf);
1275         if (p >= len)
1276                 return 0;
1277         len -= p;
1278         if (len > count)
1279                 len = count;
1280         if (copy_to_user(buf, (void*)(tbuf+p), len))
1281                 return -EFAULT;
1282         *ppos += len;
1283         return len;
1284 }
1285
1286 static ssize_t write_flush(struct file * file, const char __user * buf,
1287                              size_t count, loff_t *ppos)
1288 {
1289         struct cache_detail *cd = PDE(file->f_path.dentry->d_inode)->data;
1290         char tbuf[20];
1291         char *ep;
1292         long flushtime;
1293         if (*ppos || count > sizeof(tbuf)-1)
1294                 return -EINVAL;
1295         if (copy_from_user(tbuf, buf, count))
1296                 return -EFAULT;
1297         tbuf[count] = 0;
1298         flushtime = simple_strtoul(tbuf, &ep, 0);
1299         if (*ep && *ep != '\n')
1300                 return -EINVAL;
1301
1302         cd->flush_time = flushtime;
1303         cd->nextcheck = get_seconds();
1304         cache_flush();
1305
1306         *ppos += count;
1307         return count;
1308 }
1309
1310 static const struct file_operations cache_flush_operations = {
1311         .open           = nonseekable_open,
1312         .read           = read_flush,
1313         .write          = write_flush,
1314 };