SAFE public projects git trees. - safe/jmp/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 <linux/pagemap.h>
  31 #include <asm/ioctls.h>
  32 #include <linux/sunrpc/types.h>
  33 #include <linux/sunrpc/cache.h>
  34 #include <linux/sunrpc/stats.h>
  35 #include <linux/sunrpc/rpc_pipe_fs.h>
  36
  37 #define  RPCDBG_FACILITY RPCDBG_CACHE
  38
  39 static int cache_defer_req(struct cache_req *req, struct cache_head *item);
  40 static void cache_revisit_request(struct cache_head *item);
  41
  42 static void cache_init(struct cache_head *h)
  43 {
  44         time_t now = get_seconds();
  45         h->next = NULL;
  46         h->flags = 0;
  47         kref_init(&h->ref);
  48         h->expiry_time = now + CACHE_NEW_EXPIRY;
  49         h->last_refresh = now;
  50 }
  51
  52 static inline int cache_is_expired(struct cache_detail *detail, struct cache_head *h)
  53 {
  54         return  (h->expiry_time < get_seconds()) ||
  55                 (detail->flush_time > h->last_refresh);
  56 }
  57
  58 struct cache_head *sunrpc_cache_lookup(struct cache_detail *detail,
  59                                        struct cache_head *key, int hash)
  60 {
  61         struct cache_head **head,  **hp;
  62         struct cache_head *new = NULL, *freeme = NULL;
  63
  64         head = &detail->hash_table[hash];
  65
  66         read_lock(&detail->hash_lock);
  67
  68         for (hp=head; *hp != NULL ; hp = &(*hp)->next) {
  69                 struct cache_head *tmp = *hp;
  70                 if (detail->match(tmp, key)) {
  71                         if (cache_is_expired(detail, tmp))
  72                                 /* This entry is expired, we will discard it. */
  73                                 break;
  74                         cache_get(tmp);
  75                         read_unlock(&detail->hash_lock);
  76                         return tmp;
  77                 }
  78         }
  79         read_unlock(&detail->hash_lock);
  80         /* Didn't find anything, insert an empty entry */
  81
  82         new = detail->alloc();
  83         if (!new)
  84                 return NULL;
  85         /* must fully initialise 'new', else
  86          * we might get lose if we need to
  87          * cache_put it soon.
  88          */
  89         cache_init(new);
  90         detail->init(new, key);
  91
  92         write_lock(&detail->hash_lock);
  93
  94         /* check if entry appeared while we slept */
  95         for (hp=head; *hp != NULL ; hp = &(*hp)->next) {
  96                 struct cache_head *tmp = *hp;
  97                 if (detail->match(tmp, key)) {
  98                         if (cache_is_expired(detail, tmp)) {
  99                                 *hp = tmp->next;
 100                                 tmp->next = NULL;
 101                                 detail->entries --;
 102                                 freeme = tmp;
 103                                 break;
 104                         }
 105                         cache_get(tmp);
 106                         write_unlock(&detail->hash_lock);
 107                         cache_put(new, detail);
 108                         return tmp;
 109                 }
 110         }
 111         new->next = *head;
 112         *head = new;
 113         detail->entries++;
 114         cache_get(new);
 115         write_unlock(&detail->hash_lock);
 116
 117         if (freeme)
 118                 cache_put(freeme, detail);
 119         return new;
 120 }
 121 EXPORT_SYMBOL_GPL(sunrpc_cache_lookup);
 122
 123
 124 static void cache_dequeue(struct cache_detail *detail, struct cache_head *ch);
 125
 126 static void cache_fresh_locked(struct cache_head *head, time_t expiry)
 127 {
 128         head->expiry_time = expiry;
 129         head->last_refresh = get_seconds();
 130         set_bit(CACHE_VALID, &head->flags);
 131 }
 132
 133 static void cache_fresh_unlocked(struct cache_head *head,
 134                                  struct cache_detail *detail)
 135 {
 136         if (test_and_clear_bit(CACHE_PENDING, &head->flags)) {
 137                 cache_revisit_request(head);
 138                 cache_dequeue(detail, head);
 139         }
 140 }
 141
 142 struct cache_head *sunrpc_cache_update(struct cache_detail *detail,
 143                                        struct cache_head *new, struct cache_head *old, int hash)
 144 {
 145         /* The 'old' entry is to be replaced by 'new'.
 146          * If 'old' is not VALID, we update it directly,
 147          * otherwise we need to replace it
 148          */
 149         struct cache_head **head;
 150         struct cache_head *tmp;
 151
 152         if (!test_bit(CACHE_VALID, &old->flags)) {
 153                 write_lock(&detail->hash_lock);
 154                 if (!test_bit(CACHE_VALID, &old->flags)) {
 155                         if (test_bit(CACHE_NEGATIVE, &new->flags))
 156                                 set_bit(CACHE_NEGATIVE, &old->flags);
 157                         else
 158                                 detail->update(old, new);
 159                         cache_fresh_locked(old, new->expiry_time);
 160                         write_unlock(&detail->hash_lock);
 161                         cache_fresh_unlocked(old, detail);
 162                         return old;
 163                 }
 164                 write_unlock(&detail->hash_lock);
 165         }
 166         /* We need to insert a new entry */
 167         tmp = detail->alloc();
 168         if (!tmp) {
 169                 cache_put(old, detail);
 170                 return NULL;
 171         }
 172         cache_init(tmp);
 173         detail->init(tmp, old);
 174         head = &detail->hash_table[hash];
 175
 176         write_lock(&detail->hash_lock);
 177         if (test_bit(CACHE_NEGATIVE, &new->flags))
 178                 set_bit(CACHE_NEGATIVE, &tmp->flags);
 179         else
 180                 detail->update(tmp, new);
 181         tmp->next = *head;
 182         *head = tmp;
 183         detail->entries++;
 184         cache_get(tmp);
 185         cache_fresh_locked(tmp, new->expiry_time);
 186         cache_fresh_locked(old, 0);
 187         write_unlock(&detail->hash_lock);
 188         cache_fresh_unlocked(tmp, detail);
 189         cache_fresh_unlocked(old, detail);
 190         cache_put(old, detail);
 191         return tmp;
 192 }
 193 EXPORT_SYMBOL_GPL(sunrpc_cache_update);
 194
 195 static int cache_make_upcall(struct cache_detail *cd, struct cache_head *h)
 196 {
 197         if (!cd->cache_upcall)
 198                 return -EINVAL;
 199         return cd->cache_upcall(cd, h);
 200 }
 201
 202 static inline int cache_is_valid(struct cache_detail *detail, struct cache_head *h)
 203 {
 204         if (!test_bit(CACHE_VALID, &h->flags))
 205                 return -EAGAIN;
 206         else {
 207                 /* entry is valid */
 208                 if (test_bit(CACHE_NEGATIVE, &h->flags))
 209                         return -ENOENT;
 210                 else
 211                         return 0;
 212         }
 213 }
 214
 215 /*
 216  * This is the generic cache management routine for all
 217  * the authentication caches.
 218  * It checks the currency of a cache item and will (later)
 219  * initiate an upcall to fill it if needed.
 220  *
 221  *
 222  * Returns 0 if the cache_head can be used, or cache_puts it and returns
 223  * -EAGAIN if upcall is pending and request has been queued
 224  * -ETIMEDOUT if upcall failed or request could not be queue or
 225  *           upcall completed but item is still invalid (implying that
 226  *           the cache item has been replaced with a newer one).
 227  * -ENOENT if cache entry was negative
 228  */
 229 int cache_check(struct cache_detail *detail,
 230                     struct cache_head *h, struct cache_req *rqstp)
 231 {
 232         int rv;
 233         long refresh_age, age;
 234
 235         /* First decide return status as best we can */
 236         rv = cache_is_valid(detail, h);
 237
 238         /* now see if we want to start an upcall */
 239         refresh_age = (h->expiry_time - h->last_refresh);
 240         age = get_seconds() - h->last_refresh;
 241
 242         if (rqstp == NULL) {
 243                 if (rv == -EAGAIN)
 244                         rv = -ENOENT;
 245         } else if (rv == -EAGAIN || age > refresh_age/2) {
 246                 dprintk("RPC:       Want update, refage=%ld, age=%ld\n",
 247                                 refresh_age, age);
 248                 if (!test_and_set_bit(CACHE_PENDING, &h->flags)) {
 249                         switch (cache_make_upcall(detail, h)) {
 250                         case -EINVAL:
 251                                 clear_bit(CACHE_PENDING, &h->flags);
 252                                 cache_revisit_request(h);
 253                                 if (rv == -EAGAIN) {
 254                                         set_bit(CACHE_NEGATIVE, &h->flags);
 255                                         cache_fresh_locked(h, get_seconds()+CACHE_NEW_EXPIRY);
 256                                         cache_fresh_unlocked(h, detail);
 257                                         rv = -ENOENT;
 258                                 }
 259                                 break;
 260
 261                         case -EAGAIN:
 262                                 clear_bit(CACHE_PENDING, &h->flags);
 263                                 cache_revisit_request(h);
 264                                 break;
 265                         }
 266                 }
 267         }
 268
 269         if (rv == -EAGAIN) {
 270                 if (cache_defer_req(rqstp, h) < 0) {
 271                         /* Request is not deferred */
 272                         rv = cache_is_valid(detail, h);
 273                         if (rv == -EAGAIN)
 274                                 rv = -ETIMEDOUT;
 275                 }
 276         }
 277         if (rv)
 278                 cache_put(h, detail);
 279         return rv;
 280 }
 281 EXPORT_SYMBOL_GPL(cache_check);
 282
 283 /*
 284  * caches need to be periodically cleaned.
 285  * For this we maintain a list of cache_detail and
 286  * a current pointer into that list and into the table
 287  * for that entry.
 288  *
 289  * Each time clean_cache is called it finds the next non-empty entry
 290  * in the current table and walks the list in that entry
 291  * looking for entries that can be removed.
 292  *
 293  * An entry gets removed if:
 294  * - The expiry is before current time
 295  * - The last_refresh time is before the flush_time for that cache
 296  *
 297  * later we might drop old entries with non-NEVER expiry if that table
 298  * is getting 'full' for some definition of 'full'
 299  *
 300  * The question of "how often to scan a table" is an interesting one
 301  * and is answered in part by the use of the "nextcheck" field in the
 302  * cache_detail.
 303  * When a scan of a table begins, the nextcheck field is set to a time
 304  * that is well into the future.
 305  * While scanning, if an expiry time is found that is earlier than the
 306  * current nextcheck time, nextcheck is set to that expiry time.
 307  * If the flush_time is ever set to a time earlier than the nextcheck
 308  * time, the nextcheck time is then set to that flush_time.
 309  *
 310  * A table is then only scanned if the current time is at least
 311  * the nextcheck time.
 312  *
 313  */
 314
 315 static LIST_HEAD(cache_list);
 316 static DEFINE_SPINLOCK(cache_list_lock);
 317 static struct cache_detail *current_detail;
 318 static int current_index;
 319
 320 static void do_cache_clean(struct work_struct *work);
 321 static DECLARE_DELAYED_WORK(cache_cleaner, do_cache_clean);
 322
 323 static void sunrpc_init_cache_detail(struct cache_detail *cd)
 324 {
 325         rwlock_init(&cd->hash_lock);
 326         INIT_LIST_HEAD(&cd->queue);
 327         spin_lock(&cache_list_lock);
 328         cd->nextcheck = 0;
 329         cd->entries = 0;
 330         atomic_set(&cd->readers, 0);
 331         cd->last_close = 0;
 332         cd->last_warn = -1;
 333         list_add(&cd->others, &cache_list);
 334         spin_unlock(&cache_list_lock);
 335
 336         /* start the cleaning process */
 337         schedule_delayed_work(&cache_cleaner, 0);
 338 }
 339
 340 static void sunrpc_destroy_cache_detail(struct cache_detail *cd)
 341 {
 342         cache_purge(cd);
 343         spin_lock(&cache_list_lock);
 344         write_lock(&cd->hash_lock);
 345         if (cd->entries || atomic_read(&cd->inuse)) {
 346                 write_unlock(&cd->hash_lock);
 347                 spin_unlock(&cache_list_lock);
 348                 goto out;
 349         }
 350         if (current_detail == cd)
 351                 current_detail = NULL;
 352         list_del_init(&cd->others);
 353         write_unlock(&cd->hash_lock);
 354         spin_unlock(&cache_list_lock);
 355         if (list_empty(&cache_list)) {
 356                 /* module must be being unloaded so its safe to kill the worker */
 357                 cancel_delayed_work_sync(&cache_cleaner);
 358         }
 359         return;
 360 out:
 361         printk(KERN_ERR "nfsd: failed to unregister %s cache\n", cd->name);
 362 }
 363
 364 /* clean cache tries to find something to clean
 365  * and cleans it.
 366  * It returns 1 if it cleaned something,
 367  *            0 if it didn't find anything this time
 368  *           -1 if it fell off the end of the list.
 369  */
 370 static int cache_clean(void)
 371 {
 372         int rv = 0;
 373         struct list_head *next;
 374
 375         spin_lock(&cache_list_lock);
 376
 377         /* find a suitable table if we don't already have one */
 378         while (current_detail == NULL ||
 379             current_index >= current_detail->hash_size) {
 380                 if (current_detail)
 381                         next = current_detail->others.next;
 382                 else
 383                         next = cache_list.next;
 384                 if (next == &cache_list) {
 385                         current_detail = NULL;
 386                         spin_unlock(&cache_list_lock);
 387                         return -1;
 388                 }
 389                 current_detail = list_entry(next, struct cache_detail, others);
 390                 if (current_detail->nextcheck > get_seconds())
 391                         current_index = current_detail->hash_size;
 392                 else {
 393                         current_index = 0;
 394                         current_detail->nextcheck = get_seconds()+30*60;
 395                 }
 396         }
 397
 398         /* find a non-empty bucket in the table */
 399         while (current_detail &&
 400                current_index < current_detail->hash_size &&
 401                current_detail->hash_table[current_index] == NULL)
 402                 current_index++;
 403
 404         /* find a cleanable entry in the bucket and clean it, or set to next bucket */
 405
 406         if (current_detail && current_index < current_detail->hash_size) {
 407                 struct cache_head *ch, **cp;
 408                 struct cache_detail *d;
 409
 410                 write_lock(&current_detail->hash_lock);
 411
 412                 /* Ok, now to clean this strand */
 413
 414                 cp = & current_detail->hash_table[current_index];
 415                 for (ch = *cp ; ch ; cp = & ch->next, ch = *cp) {
 416                         if (current_detail->nextcheck > ch->expiry_time)
 417                                 current_detail->nextcheck = ch->expiry_time+1;
 418                         if (!cache_is_expired(current_detail, ch))
 419                                 continue;
 420
 421                         *cp = ch->next;
 422                         ch->next = NULL;
 423                         current_detail->entries--;
 424                         rv = 1;
 425                         break;
 426                 }
 427
 428                 write_unlock(&current_detail->hash_lock);
 429                 d = current_detail;
 430                 if (!ch)
 431                         current_index ++;
 432                 spin_unlock(&cache_list_lock);
 433                 if (ch) {
 434                         if (test_and_clear_bit(CACHE_PENDING, &ch->flags))
 435                                 cache_dequeue(current_detail, ch);
 436                         cache_revisit_request(ch);
 437                         cache_put(ch, d);
 438                 }
 439         } else
 440                 spin_unlock(&cache_list_lock);
 441
 442         return rv;
 443 }
 444
 445 /*
 446  * We want to regularly clean the cache, so we need to schedule some work ...
 447  */
 448 static void do_cache_clean(struct work_struct *work)
 449 {
 450         int delay = 5;
 451         if (cache_clean() == -1)
 452                 delay = round_jiffies_relative(30*HZ);
 453
 454         if (list_empty(&cache_list))
 455                 delay = 0;
 456
 457         if (delay)
 458                 schedule_delayed_work(&cache_cleaner, delay);
 459 }
 460
 461
 462 /*
 463  * Clean all caches promptly.  This just calls cache_clean
 464  * repeatedly until we are sure that every cache has had a chance to
 465  * be fully cleaned
 466  */
 467 void cache_flush(void)
 468 {
 469         while (cache_clean() != -1)
 470                 cond_resched();
 471         while (cache_clean() != -1)
 472                 cond_resched();
 473 }
 474 EXPORT_SYMBOL_GPL(cache_flush);
 475
 476 void cache_purge(struct cache_detail *detail)
 477 {
 478         detail->flush_time = LONG_MAX;
 479         detail->nextcheck = get_seconds();
 480         cache_flush();
 481         detail->flush_time = 1;
 482 }
 483 EXPORT_SYMBOL_GPL(cache_purge);
 484
 485
 486 /*
 487  * Deferral and Revisiting of Requests.
 488  *
 489  * If a cache lookup finds a pending entry, we
 490  * need to defer the request and revisit it later.
 491  * All deferred requests are stored in a hash table,
 492  * indexed by "struct cache_head *".
 493  * As it may be wasteful to store a whole request
 494  * structure, we allow the request to provide a
 495  * deferred form, which must contain a
 496  * 'struct cache_deferred_req'
 497  * This cache_deferred_req contains a method to allow
 498  * it to be revisited when cache info is available
 499  */
 500
 501 #define DFR_HASHSIZE    (PAGE_SIZE/sizeof(struct list_head))
 502 #define DFR_HASH(item)  ((((long)item)>>4 ^ (((long)item)>>13)) % DFR_HASHSIZE)
 503
 504 #define DFR_MAX 300     /* ??? */
 505
 506 static DEFINE_SPINLOCK(cache_defer_lock);
 507 static LIST_HEAD(cache_defer_list);
 508 static struct list_head cache_defer_hash[DFR_HASHSIZE];
 509 static int cache_defer_cnt;
 510
 511 static int cache_defer_req(struct cache_req *req, struct cache_head *item)
 512 {
 513         struct cache_deferred_req *dreq, *discard;
 514         int hash = DFR_HASH(item);
 515
 516         if (cache_defer_cnt >= DFR_MAX) {
 517                 /* too much in the cache, randomly drop this one,
 518                  * or continue and drop the oldest below
 519                  */
 520                 if (net_random()&1)
 521                         return -ENOMEM;
 522         }
 523         dreq = req->defer(req);
 524         if (dreq == NULL)
 525                 return -ENOMEM;
 526
 527         dreq->item = item;
 528
 529         spin_lock(&cache_defer_lock);
 530
 531         list_add(&dreq->recent, &cache_defer_list);
 532
 533         if (cache_defer_hash[hash].next == NULL)
 534                 INIT_LIST_HEAD(&cache_defer_hash[hash]);
 535         list_add(&dreq->hash, &cache_defer_hash[hash]);
 536
 537         /* it is in, now maybe clean up */
 538         discard = NULL;
 539         if (++cache_defer_cnt > DFR_MAX) {
 540                 discard = list_entry(cache_defer_list.prev,
 541                                      struct cache_deferred_req, recent);
 542                 list_del_init(&discard->recent);
 543                 list_del_init(&discard->hash);
 544                 cache_defer_cnt--;
 545         }
 546         spin_unlock(&cache_defer_lock);
 547
 548         if (discard)
 549                 /* there was one too many */
 550                 discard->revisit(discard, 1);
 551
 552         if (!test_bit(CACHE_PENDING, &item->flags)) {
 553                 /* must have just been validated... */
 554                 cache_revisit_request(item);
 555                 return -EAGAIN;
 556         }
 557         return 0;
 558 }
 559
 560 static void cache_revisit_request(struct cache_head *item)
 561 {
 562         struct cache_deferred_req *dreq;
 563         struct list_head pending;
 564
 565         struct list_head *lp;
 566         int hash = DFR_HASH(item);
 567
 568         INIT_LIST_HEAD(&pending);
 569         spin_lock(&cache_defer_lock);
 570
 571         lp = cache_defer_hash[hash].next;
 572         if (lp) {
 573                 while (lp != &cache_defer_hash[hash]) {
 574                         dreq = list_entry(lp, struct cache_deferred_req, hash);
 575                         lp = lp->next;
 576                         if (dreq->item == item) {
 577                                 list_del_init(&dreq->hash);
 578                                 list_move(&dreq->recent, &pending);
 579                                 cache_defer_cnt--;
 580                         }
 581                 }
 582         }
 583         spin_unlock(&cache_defer_lock);
 584
 585         while (!list_empty(&pending)) {
 586                 dreq = list_entry(pending.next, struct cache_deferred_req, recent);
 587                 list_del_init(&dreq->recent);
 588                 dreq->revisit(dreq, 0);
 589         }
 590 }
 591
 592 void cache_clean_deferred(void *owner)
 593 {
 594         struct cache_deferred_req *dreq, *tmp;
 595         struct list_head pending;
 596
 597
 598         INIT_LIST_HEAD(&pending);
 599         spin_lock(&cache_defer_lock);
 600
 601         list_for_each_entry_safe(dreq, tmp, &cache_defer_list, recent) {
 602                 if (dreq->owner == owner) {
 603                         list_del_init(&dreq->hash);
 604                         list_move(&dreq->recent, &pending);
 605                         cache_defer_cnt--;
 606                 }
 607         }
 608         spin_unlock(&cache_defer_lock);
 609
 610         while (!list_empty(&pending)) {
 611                 dreq = list_entry(pending.next, struct cache_deferred_req, recent);
 612                 list_del_init(&dreq->recent);
 613                 dreq->revisit(dreq, 1);
 614         }
 615 }
 616
 617 /*
 618  * communicate with user-space
 619  *
 620  * We have a magic /proc file - /proc/sunrpc/<cachename>/channel.
 621  * On read, you get a full request, or block.
 622  * On write, an update request is processed.
 623  * Poll works if anything to read, and always allows write.
 624  *
 625  * Implemented by linked list of requests.  Each open file has
 626  * a ->private that also exists in this list.  New requests are added
 627  * to the end and may wakeup and preceding readers.
 628  * New readers are added to the head.  If, on read, an item is found with
 629  * CACHE_UPCALLING clear, we free it from the list.
 630  *
 631  */
 632
 633 static DEFINE_SPINLOCK(queue_lock);
 634 static DEFINE_MUTEX(queue_io_mutex);
 635
 636 struct cache_queue {
 637         struct list_head        list;
 638         int                     reader; /* if 0, then request */
 639 };
 640 struct cache_request {
 641         struct cache_queue      q;
 642         struct cache_head       *item;
 643         char                    * buf;
 644         int                     len;
 645         int                     readers;
 646 };
 647 struct cache_reader {
 648         struct cache_queue      q;
 649         int                     offset; /* if non-0, we have a refcnt on next request */
 650 };
 651
 652 static ssize_t cache_read(struct file *filp, char __user *buf, size_t count,
 653                           loff_t *ppos, struct cache_detail *cd)
 654 {
 655         struct cache_reader *rp = filp->private_data;
 656         struct cache_request *rq;
 657         struct inode *inode = filp->f_path.dentry->d_inode;
 658         int err;
 659
 660         if (count == 0)
 661                 return 0;
 662
 663         mutex_lock(&inode->i_mutex); /* protect against multiple concurrent
 664                               * readers on this file */
 665  again:
 666         spin_lock(&queue_lock);
 667         /* need to find next request */
 668         while (rp->q.list.next != &cd->queue &&
 669                list_entry(rp->q.list.next, struct cache_queue, list)
 670                ->reader) {
 671                 struct list_head *next = rp->q.list.next;
 672                 list_move(&rp->q.list, next);
 673         }
 674         if (rp->q.list.next == &cd->queue) {
 675                 spin_unlock(&queue_lock);
 676                 mutex_unlock(&inode->i_mutex);
 677                 BUG_ON(rp->offset);
 678                 return 0;
 679         }
 680         rq = container_of(rp->q.list.next, struct cache_request, q.list);
 681         BUG_ON(rq->q.reader);
 682         if (rp->offset == 0)
 683                 rq->readers++;
 684         spin_unlock(&queue_lock);
 685
 686         if (rp->offset == 0 && !test_bit(CACHE_PENDING, &rq->item->flags)) {
 687                 err = -EAGAIN;
 688                 spin_lock(&queue_lock);
 689                 list_move(&rp->q.list, &rq->q.list);
 690                 spin_unlock(&queue_lock);
 691         } else {
 692                 if (rp->offset + count > rq->len)
 693                         count = rq->len - rp->offset;
 694                 err = -EFAULT;
 695                 if (copy_to_user(buf, rq->buf + rp->offset, count))
 696                         goto out;
 697                 rp->offset += count;
 698                 if (rp->offset >= rq->len) {
 699                         rp->offset = 0;
 700                         spin_lock(&queue_lock);
 701                         list_move(&rp->q.list, &rq->q.list);
 702                         spin_unlock(&queue_lock);
 703                 }
 704                 err = 0;
 705         }
 706  out:
 707         if (rp->offset == 0) {
 708                 /* need to release rq */
 709                 spin_lock(&queue_lock);
 710                 rq->readers--;
 711                 if (rq->readers == 0 &&
 712                     !test_bit(CACHE_PENDING, &rq->item->flags)) {
 713                         list_del(&rq->q.list);
 714                         spin_unlock(&queue_lock);
 715                         cache_put(rq->item, cd);
 716                         kfree(rq->buf);
 717                         kfree(rq);
 718                 } else
 719                         spin_unlock(&queue_lock);
 720         }
 721         if (err == -EAGAIN)
 722                 goto again;
 723         mutex_unlock(&inode->i_mutex);
 724         return err ? err :  count;
 725 }
 726
 727 static ssize_t cache_do_downcall(char *kaddr, const char __user *buf,
 728                                  size_t count, struct cache_detail *cd)
 729 {
 730         ssize_t ret;
 731
 732         if (copy_from_user(kaddr, buf, count))
 733                 return -EFAULT;
 734         kaddr[count] = '\0';
 735         ret = cd->cache_parse(cd, kaddr, count);
 736         if (!ret)
 737                 ret = count;
 738         return ret;
 739 }
 740
 741 static ssize_t cache_slow_downcall(const char __user *buf,
 742                                    size_t count, struct cache_detail *cd)
 743 {
 744         static char write_buf[8192]; /* protected by queue_io_mutex */
 745         ssize_t ret = -EINVAL;
 746
 747         if (count >= sizeof(write_buf))
 748                 goto out;
 749         mutex_lock(&queue_io_mutex);
 750         ret = cache_do_downcall(write_buf, buf, count, cd);
 751         mutex_unlock(&queue_io_mutex);
 752 out:
 753         return ret;
 754 }
 755
 756 static ssize_t cache_downcall(struct address_space *mapping,
 757                               const char __user *buf,
 758                               size_t count, struct cache_detail *cd)
 759 {
 760         struct page *page;
 761         char *kaddr;
 762         ssize_t ret = -ENOMEM;
 763
 764         if (count >= PAGE_CACHE_SIZE)
 765                 goto out_slow;
 766
 767         page = find_or_create_page(mapping, 0, GFP_KERNEL);
 768         if (!page)
 769                 goto out_slow;
 770
 771         kaddr = kmap(page);
 772         ret = cache_do_downcall(kaddr, buf, count, cd);
 773         kunmap(page);
 774         unlock_page(page);
 775         page_cache_release(page);
 776         return ret;
 777 out_slow:
 778         return cache_slow_downcall(buf, count, cd);
 779 }
 780
 781 static ssize_t cache_write(struct file *filp, const char __user *buf,
 782                            size_t count, loff_t *ppos,
 783                            struct cache_detail *cd)
 784 {
 785         struct address_space *mapping = filp->f_mapping;
 786         struct inode *inode = filp->f_path.dentry->d_inode;
 787         ssize_t ret = -EINVAL;
 788
 789         if (!cd->cache_parse)
 790                 goto out;
 791
 792         mutex_lock(&inode->i_mutex);
 793         ret = cache_downcall(mapping, buf, count, cd);
 794         mutex_unlock(&inode->i_mutex);
 795 out:
 796         return ret;
 797 }
 798
 799 static DECLARE_WAIT_QUEUE_HEAD(queue_wait);
 800
 801 static unsigned int cache_poll(struct file *filp, poll_table *wait,
 802                                struct cache_detail *cd)
 803 {
 804         unsigned int mask;
 805         struct cache_reader *rp = filp->private_data;
 806         struct cache_queue *cq;
 807
 808         poll_wait(filp, &queue_wait, wait);
 809
 810         /* alway allow write */
 811         mask = POLL_OUT | POLLWRNORM;
 812
 813         if (!rp)
 814                 return mask;
 815
 816         spin_lock(&queue_lock);
 817
 818         for (cq= &rp->q; &cq->list != &cd->queue;
 819              cq = list_entry(cq->list.next, struct cache_queue, list))
 820                 if (!cq->reader) {
 821                         mask |= POLLIN | POLLRDNORM;
 822                         break;
 823                 }
 824         spin_unlock(&queue_lock);
 825         return mask;
 826 }
 827
 828 static int cache_ioctl(struct inode *ino, struct file *filp,
 829                        unsigned int cmd, unsigned long arg,
 830                        struct cache_detail *cd)
 831 {
 832         int len = 0;
 833         struct cache_reader *rp = filp->private_data;
 834         struct cache_queue *cq;
 835
 836         if (cmd != FIONREAD || !rp)
 837                 return -EINVAL;
 838
 839         spin_lock(&queue_lock);
 840
 841         /* only find the length remaining in current request,
 842          * or the length of the next request
 843          */
 844         for (cq= &rp->q; &cq->list != &cd->queue;
 845              cq = list_entry(cq->list.next, struct cache_queue, list))
 846                 if (!cq->reader) {
 847                         struct cache_request *cr =
 848                                 container_of(cq, struct cache_request, q);
 849                         len = cr->len - rp->offset;
 850                         break;
 851                 }
 852         spin_unlock(&queue_lock);
 853
 854         return put_user(len, (int __user *)arg);
 855 }
 856
 857 static int cache_open(struct inode *inode, struct file *filp,
 858                       struct cache_detail *cd)
 859 {
 860         struct cache_reader *rp = NULL;
 861
 862         if (!cd || !try_module_get(cd->owner))
 863                 return -EACCES;
 864         nonseekable_open(inode, filp);
 865         if (filp->f_mode & FMODE_READ) {
 866                 rp = kmalloc(sizeof(*rp), GFP_KERNEL);
 867                 if (!rp)
 868                         return -ENOMEM;
 869                 rp->offset = 0;
 870                 rp->q.reader = 1;
 871                 atomic_inc(&cd->readers);
 872                 spin_lock(&queue_lock);
 873                 list_add(&rp->q.list, &cd->queue);
 874                 spin_unlock(&queue_lock);
 875         }
 876         filp->private_data = rp;
 877         return 0;
 878 }
 879
 880 static int cache_release(struct inode *inode, struct file *filp,
 881                          struct cache_detail *cd)
 882 {
 883         struct cache_reader *rp = filp->private_data;
 884
 885         if (rp) {
 886                 spin_lock(&queue_lock);
 887                 if (rp->offset) {
 888                         struct cache_queue *cq;
 889                         for (cq= &rp->q; &cq->list != &cd->queue;
 890                              cq = list_entry(cq->list.next, struct cache_queue, list))
 891                                 if (!cq->reader) {
 892                                         container_of(cq, struct cache_request, q)
 893                                                 ->readers--;
 894                                         break;
 895                                 }
 896                         rp->offset = 0;
 897                 }
 898                 list_del(&rp->q.list);
 899                 spin_unlock(&queue_lock);
 900
 901                 filp->private_data = NULL;
 902                 kfree(rp);
 903
 904                 cd->last_close = get_seconds();
 905                 atomic_dec(&cd->readers);
 906         }
 907         module_put(cd->owner);
 908         return 0;
 909 }
 910
 911
 912
 913 static void cache_dequeue(struct cache_detail *detail, struct cache_head *ch)
 914 {
 915         struct cache_queue *cq;
 916         spin_lock(&queue_lock);
 917         list_for_each_entry(cq, &detail->queue, list)
 918                 if (!cq->reader) {
 919                         struct cache_request *cr = container_of(cq, struct cache_request, q);
 920                         if (cr->item != ch)
 921                                 continue;
 922                         if (cr->readers != 0)
 923                                 continue;
 924                         list_del(&cr->q.list);
 925                         spin_unlock(&queue_lock);
 926                         cache_put(cr->item, detail);
 927                         kfree(cr->buf);
 928                         kfree(cr);
 929                         return;
 930                 }
 931         spin_unlock(&queue_lock);
 932 }
 933
 934 /*
 935  * Support routines for text-based upcalls.
 936  * Fields are separated by spaces.
 937  * Fields are either mangled to quote space tab newline slosh with slosh
 938  * or a hexified with a leading \x
 939  * Record is terminated with newline.
 940  *
 941  */
 942
 943 void qword_add(char **bpp, int *lp, char *str)
 944 {
 945         char *bp = *bpp;
 946         int len = *lp;
 947         char c;
 948
 949         if (len < 0) return;
 950
 951         while ((c=*str++) && len)
 952                 switch(c) {
 953                 case ' ':
 954                 case '\t':
 955                 case '\n':
 956                 case '\\':
 957                         if (len >= 4) {
 958                                 *bp++ = '\\';
 959                                 *bp++ = '0' + ((c & 0300)>>6);
 960                                 *bp++ = '0' + ((c & 0070)>>3);
 961                                 *bp++ = '0' + ((c & 0007)>>0);
 962                         }
 963                         len -= 4;
 964                         break;
 965                 default:
 966                         *bp++ = c;
 967                         len--;
 968                 }
 969         if (c || len <1) len = -1;
 970         else {
 971                 *bp++ = ' ';
 972                 len--;
 973         }
 974         *bpp = bp;
 975         *lp = len;
 976 }
 977 EXPORT_SYMBOL_GPL(qword_add);
 978
 979 void qword_addhex(char **bpp, int *lp, char *buf, int blen)
 980 {
 981         char *bp = *bpp;
 982         int len = *lp;
 983
 984         if (len < 0) return;
 985
 986         if (len > 2) {
 987                 *bp++ = '\\';
 988                 *bp++ = 'x';
 989                 len -= 2;
 990                 while (blen && len >= 2) {
 991                         unsigned char c = *buf++;
 992                         *bp++ = '0' + ((c&0xf0)>>4) + (c>=0xa0)*('a'-'9'-1);
 993                         *bp++ = '0' + (c&0x0f) + ((c&0x0f)>=0x0a)*('a'-'9'-1);
 994                         len -= 2;
 995                         blen--;
 996                 }
 997         }
 998         if (blen || len<1) len = -1;
 999         else {
1000                 *bp++ = ' ';
1001                 len--;
1002         }
1003         *bpp = bp;
1004         *lp = len;
1005 }
1006 EXPORT_SYMBOL_GPL(qword_addhex);
1007
1008 static void warn_no_listener(struct cache_detail *detail)
1009 {
1010         if (detail->last_warn != detail->last_close) {
1011                 detail->last_warn = detail->last_close;
1012                 if (detail->warn_no_listener)
1013                         detail->warn_no_listener(detail, detail->last_close != 0);
1014         }
1015 }
1016
1017 /*
1018  * register an upcall request to user-space and queue it up for read() by the
1019  * upcall daemon.
1020  *
1021  * Each request is at most one page long.
1022  */
1023 int sunrpc_cache_pipe_upcall(struct cache_detail *detail, struct cache_head *h,
1024                 void (*cache_request)(struct cache_detail *,
1025                                       struct cache_head *,
1026                                       char **,
1027                                       int *))
1028 {
1029
1030         char *buf;
1031         struct cache_request *crq;
1032         char *bp;
1033         int len;
1034
1035         if (atomic_read(&detail->readers) == 0 &&
1036             detail->last_close < get_seconds() - 30) {
1037                         warn_no_listener(detail);
1038                         return -EINVAL;
1039         }
1040
1041         buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
1042         if (!buf)
1043                 return -EAGAIN;
1044
1045         crq = kmalloc(sizeof (*crq), GFP_KERNEL);
1046         if (!crq) {
1047                 kfree(buf);
1048                 return -EAGAIN;
1049         }
1050
1051         bp = buf; len = PAGE_SIZE;
1052
1053         cache_request(detail, h, &bp, &len);
1054
1055         if (len < 0) {
1056                 kfree(buf);
1057                 kfree(crq);
1058                 return -EAGAIN;
1059         }
1060         crq->q.reader = 0;
1061         crq->item = cache_get(h);
1062         crq->buf = buf;
1063         crq->len = PAGE_SIZE - len;
1064         crq->readers = 0;
1065         spin_lock(&queue_lock);
1066         list_add_tail(&crq->q.list, &detail->queue);
1067         spin_unlock(&queue_lock);
1068         wake_up(&queue_wait);
1069         return 0;
1070 }
1071 EXPORT_SYMBOL_GPL(sunrpc_cache_pipe_upcall);
1072
1073 /*
1074  * parse a message from user-space and pass it
1075  * to an appropriate cache
1076  * Messages are, like requests, separated into fields by
1077  * spaces and dequotes as \xHEXSTRING or embedded \nnn octal
1078  *
1079  * Message is
1080  *   reply cachename expiry key ... content....
1081  *
1082  * key and content are both parsed by cache
1083  */
1084
1085 #define isodigit(c) (isdigit(c) && c <= '7')
1086 int qword_get(char **bpp, char *dest, int bufsize)
1087 {
1088         /* return bytes copied, or -1 on error */
1089         char *bp = *bpp;
1090         int len = 0;
1091
1092         while (*bp == ' ') bp++;
1093
1094         if (bp[0] == '\\' && bp[1] == 'x') {
1095                 /* HEX STRING */
1096                 bp += 2;
1097                 while (isxdigit(bp[0]) && isxdigit(bp[1]) && len < bufsize) {
1098                         int byte = isdigit(*bp) ? *bp-'0' : toupper(*bp)-'A'+10;
1099                         bp++;
1100                         byte <<= 4;
1101                         byte |= isdigit(*bp) ? *bp-'0' : toupper(*bp)-'A'+10;
1102                         *dest++ = byte;
1103                         bp++;
1104                         len++;
1105                 }
1106         } else {
1107                 /* text with \nnn octal quoting */
1108                 while (*bp != ' ' && *bp != '\n' && *bp && len < bufsize-1) {
1109                         if (*bp == '\\' &&
1110                             isodigit(bp[1]) && (bp[1] <= '3') &&
1111                             isodigit(bp[2]) &&
1112                             isodigit(bp[3])) {
1113                                 int byte = (*++bp -'0');
1114                                 bp++;
1115                                 byte = (byte << 3) | (*bp++ - '0');
1116                                 byte = (byte << 3) | (*bp++ - '0');
1117                                 *dest++ = byte;
1118                                 len++;
1119                         } else {
1120                                 *dest++ = *bp++;
1121                                 len++;
1122                         }
1123                 }
1124         }
1125
1126         if (*bp != ' ' && *bp != '\n' && *bp != '\0')
1127                 return -1;
1128         while (*bp == ' ') bp++;
1129         *bpp = bp;
1130         *dest = '\0';
1131         return len;
1132 }
1133 EXPORT_SYMBOL_GPL(qword_get);
1134
1135
1136 /*
1137  * support /proc/sunrpc/cache/$CACHENAME/content
1138  * as a seqfile.
1139  * We call ->cache_show passing NULL for the item to
1140  * get a header, then pass each real item in the cache
1141  */
1142
1143 struct handle {
1144         struct cache_detail *cd;
1145 };
1146
1147 static void *c_start(struct seq_file *m, loff_t *pos)
1148         __acquires(cd->hash_lock)
1149 {
1150         loff_t n = *pos;
1151         unsigned hash, entry;
1152         struct cache_head *ch;
1153         struct cache_detail *cd = ((struct handle*)m->private)->cd;
1154
1155
1156         read_lock(&cd->hash_lock);
1157         if (!n--)
1158                 return SEQ_START_TOKEN;
1159         hash = n >> 32;
1160         entry = n & ((1LL<<32) - 1);
1161
1162         for (ch=cd->hash_table[hash]; ch; ch=ch->next)
1163                 if (!entry--)
1164                         return ch;
1165         n &= ~((1LL<<32) - 1);
1166         do {
1167                 hash++;
1168                 n += 1LL<<32;
1169         } while(hash < cd->hash_size &&
1170                 cd->hash_table[hash]==NULL);
1171         if (hash >= cd->hash_size)
1172                 return NULL;
1173         *pos = n+1;
1174         return cd->hash_table[hash];
1175 }
1176
1177 static void *c_next(struct seq_file *m, void *p, loff_t *pos)
1178 {
1179         struct cache_head *ch = p;
1180         int hash = (*pos >> 32);
1181         struct cache_detail *cd = ((struct handle*)m->private)->cd;
1182
1183         if (p == SEQ_START_TOKEN)
1184                 hash = 0;
1185         else if (ch->next == NULL) {
1186                 hash++;
1187                 *pos += 1LL<<32;
1188         } else {
1189                 ++*pos;
1190                 return ch->next;
1191         }
1192         *pos &= ~((1LL<<32) - 1);
1193         while (hash < cd->hash_size &&
1194                cd->hash_table[hash] == NULL) {
1195                 hash++;
1196                 *pos += 1LL<<32;
1197         }
1198         if (hash >= cd->hash_size)
1199                 return NULL;
1200         ++*pos;
1201         return cd->hash_table[hash];
1202 }
1203
1204 static void c_stop(struct seq_file *m, void *p)
1205         __releases(cd->hash_lock)
1206 {
1207         struct cache_detail *cd = ((struct handle*)m->private)->cd;
1208         read_unlock(&cd->hash_lock);
1209 }
1210
1211 static int c_show(struct seq_file *m, void *p)
1212 {
1213         struct cache_head *cp = p;
1214         struct cache_detail *cd = ((struct handle*)m->private)->cd;
1215
1216         if (p == SEQ_START_TOKEN)
1217                 return cd->cache_show(m, cd, NULL);
1218
1219         ifdebug(CACHE)
1220                 seq_printf(m, "# expiry=%ld refcnt=%d flags=%lx\n",
1221                            cp->expiry_time, atomic_read(&cp->ref.refcount), cp->flags);
1222         cache_get(cp);
1223         if (cache_check(cd, cp, NULL))
1224                 /* cache_check does a cache_put on failure */
1225                 seq_printf(m, "# ");
1226         else
1227                 cache_put(cp, cd);
1228
1229         return cd->cache_show(m, cd, cp);
1230 }
1231
1232 static const struct seq_operations cache_content_op = {
1233         .start  = c_start,
1234         .next   = c_next,
1235         .stop   = c_stop,
1236         .show   = c_show,
1237 };
1238
1239 static int content_open(struct inode *inode, struct file *file,
1240                         struct cache_detail *cd)
1241 {
1242         struct handle *han;
1243
1244         if (!cd || !try_module_get(cd->owner))
1245                 return -EACCES;
1246         han = __seq_open_private(file, &cache_content_op, sizeof(*han));
1247         if (han == NULL) {
1248                 module_put(cd->owner);
1249                 return -ENOMEM;
1250         }
1251
1252         han->cd = cd;
1253         return 0;
1254 }
1255
1256 static int content_release(struct inode *inode, struct file *file,
1257                 struct cache_detail *cd)
1258 {
1259         int ret = seq_release_private(inode, file);
1260         module_put(cd->owner);
1261         return ret;
1262 }
1263
1264 static int open_flush(struct inode *inode, struct file *file,
1265                         struct cache_detail *cd)
1266 {
1267         if (!cd || !try_module_get(cd->owner))
1268                 return -EACCES;
1269         return nonseekable_open(inode, file);
1270 }
1271
1272 static int release_flush(struct inode *inode, struct file *file,
1273                         struct cache_detail *cd)
1274 {
1275         module_put(cd->owner);
1276         return 0;
1277 }
1278
1279 static ssize_t read_flush(struct file *file, char __user *buf,
1280                           size_t count, loff_t *ppos,
1281                           struct cache_detail *cd)
1282 {
1283         char tbuf[20];
1284         unsigned long p = *ppos;
1285         size_t len;
1286
1287         sprintf(tbuf, "%lu\n", cd->flush_time);
1288         len = strlen(tbuf);
1289         if (p >= len)
1290                 return 0;
1291         len -= p;
1292         if (len > count)
1293                 len = count;
1294         if (copy_to_user(buf, (void*)(tbuf+p), len))
1295                 return -EFAULT;
1296         *ppos += len;
1297         return len;
1298 }
1299
1300 static ssize_t write_flush(struct file *file, const char __user *buf,
1301                            size_t count, loff_t *ppos,
1302                            struct cache_detail *cd)
1303 {
1304         char tbuf[20];
1305         char *ep;
1306         long flushtime;
1307         if (*ppos || count > sizeof(tbuf)-1)
1308                 return -EINVAL;
1309         if (copy_from_user(tbuf, buf, count))
1310                 return -EFAULT;
1311         tbuf[count] = 0;
1312         flushtime = simple_strtoul(tbuf, &ep, 0);
1313         if (*ep && *ep != '\n')
1314                 return -EINVAL;
1315
1316         cd->flush_time = flushtime;
1317         cd->nextcheck = get_seconds();
1318         cache_flush();
1319
1320         *ppos += count;
1321         return count;
1322 }
1323
1324 static ssize_t cache_read_procfs(struct file *filp, char __user *buf,
1325                                  size_t count, loff_t *ppos)
1326 {
1327         struct cache_detail *cd = PDE(filp->f_path.dentry->d_inode)->data;
1328
1329         return cache_read(filp, buf, count, ppos, cd);
1330 }
1331
1332 static ssize_t cache_write_procfs(struct file *filp, const char __user *buf,
1333                                   size_t count, loff_t *ppos)
1334 {
1335         struct cache_detail *cd = PDE(filp->f_path.dentry->d_inode)->data;
1336
1337         return cache_write(filp, buf, count, ppos, cd);
1338 }
1339
1340 static unsigned int cache_poll_procfs(struct file *filp, poll_table *wait)
1341 {
1342         struct cache_detail *cd = PDE(filp->f_path.dentry->d_inode)->data;
1343
1344         return cache_poll(filp, wait, cd);
1345 }
1346
1347 static int cache_ioctl_procfs(struct inode *inode, struct file *filp,
1348                               unsigned int cmd, unsigned long arg)
1349 {
1350         struct cache_detail *cd = PDE(inode)->data;
1351
1352         return cache_ioctl(inode, filp, cmd, arg, cd);
1353 }
1354
1355 static int cache_open_procfs(struct inode *inode, struct file *filp)
1356 {
1357         struct cache_detail *cd = PDE(inode)->data;
1358
1359         return cache_open(inode, filp, cd);
1360 }
1361
1362 static int cache_release_procfs(struct inode *inode, struct file *filp)
1363 {
1364         struct cache_detail *cd = PDE(inode)->data;
1365
1366         return cache_release(inode, filp, cd);
1367 }
1368
1369 static const struct file_operations cache_file_operations_procfs = {
1370         .owner          = THIS_MODULE,
1371         .llseek         = no_llseek,
1372         .read           = cache_read_procfs,
1373         .write          = cache_write_procfs,
1374         .poll           = cache_poll_procfs,
1375         .ioctl          = cache_ioctl_procfs, /* for FIONREAD */
1376         .open           = cache_open_procfs,
1377         .release        = cache_release_procfs,
1378 };
1379
1380 static int content_open_procfs(struct inode *inode, struct file *filp)
1381 {
1382         struct cache_detail *cd = PDE(inode)->data;
1383
1384         return content_open(inode, filp, cd);
1385 }
1386
1387 static int content_release_procfs(struct inode *inode, struct file *filp)
1388 {
1389         struct cache_detail *cd = PDE(inode)->data;
1390
1391         return content_release(inode, filp, cd);
1392 }
1393
1394 static const struct file_operations content_file_operations_procfs = {
1395         .open           = content_open_procfs,
1396         .read           = seq_read,
1397         .llseek         = seq_lseek,
1398         .release        = content_release_procfs,
1399 };
1400
1401 static int open_flush_procfs(struct inode *inode, struct file *filp)
1402 {
1403         struct cache_detail *cd = PDE(inode)->data;
1404
1405         return open_flush(inode, filp, cd);
1406 }
1407
1408 static int release_flush_procfs(struct inode *inode, struct file *filp)
1409 {
1410         struct cache_detail *cd = PDE(inode)->data;
1411
1412         return release_flush(inode, filp, cd);
1413 }
1414
1415 static ssize_t read_flush_procfs(struct file *filp, char __user *buf,
1416                             size_t count, loff_t *ppos)
1417 {
1418         struct cache_detail *cd = PDE(filp->f_path.dentry->d_inode)->data;
1419
1420         return read_flush(filp, buf, count, ppos, cd);
1421 }
1422
1423 static ssize_t write_flush_procfs(struct file *filp,
1424                                   const char __user *buf,
1425                                   size_t count, loff_t *ppos)
1426 {
1427         struct cache_detail *cd = PDE(filp->f_path.dentry->d_inode)->data;
1428
1429         return write_flush(filp, buf, count, ppos, cd);
1430 }
1431
1432 static const struct file_operations cache_flush_operations_procfs = {
1433         .open           = open_flush_procfs,
1434         .read           = read_flush_procfs,
1435         .write          = write_flush_procfs,
1436         .release        = release_flush_procfs,
1437 };
1438
1439 static void remove_cache_proc_entries(struct cache_detail *cd)
1440 {
1441         if (cd->u.procfs.proc_ent == NULL)
1442                 return;
1443         if (cd->u.procfs.flush_ent)
1444                 remove_proc_entry("flush", cd->u.procfs.proc_ent);
1445         if (cd->u.procfs.channel_ent)
1446                 remove_proc_entry("channel", cd->u.procfs.proc_ent);
1447         if (cd->u.procfs.content_ent)
1448                 remove_proc_entry("content", cd->u.procfs.proc_ent);
1449         cd->u.procfs.proc_ent = NULL;
1450         remove_proc_entry(cd->name, proc_net_rpc);
1451 }
1452
1453 #ifdef CONFIG_PROC_FS
1454 static int create_cache_proc_entries(struct cache_detail *cd)
1455 {
1456         struct proc_dir_entry *p;
1457
1458         cd->u.procfs.proc_ent = proc_mkdir(cd->name, proc_net_rpc);
1459         if (cd->u.procfs.proc_ent == NULL)
1460                 goto out_nomem;
1461         cd->u.procfs.channel_ent = NULL;
1462         cd->u.procfs.content_ent = NULL;
1463
1464         p = proc_create_data("flush", S_IFREG|S_IRUSR|S_IWUSR,
1465                              cd->u.procfs.proc_ent,
1466                              &cache_flush_operations_procfs, cd);
1467         cd->u.procfs.flush_ent = p;
1468         if (p == NULL)
1469                 goto out_nomem;
1470
1471         if (cd->cache_upcall || cd->cache_parse) {
1472                 p = proc_create_data("channel", S_IFREG|S_IRUSR|S_IWUSR,
1473                                      cd->u.procfs.proc_ent,
1474                                      &cache_file_operations_procfs, cd);
1475                 cd->u.procfs.channel_ent = p;
1476                 if (p == NULL)
1477                         goto out_nomem;
1478         }
1479         if (cd->cache_show) {
1480                 p = proc_create_data("content", S_IFREG|S_IRUSR|S_IWUSR,
1481                                 cd->u.procfs.proc_ent,
1482                                 &content_file_operations_procfs, cd);
1483                 cd->u.procfs.content_ent = p;
1484                 if (p == NULL)
1485                         goto out_nomem;
1486         }
1487         return 0;
1488 out_nomem:
1489         remove_cache_proc_entries(cd);
1490         return -ENOMEM;
1491 }
1492 #else /* CONFIG_PROC_FS */
1493 static int create_cache_proc_entries(struct cache_detail *cd)
1494 {
1495         return 0;
1496 }
1497 #endif
1498
1499 int cache_register(struct cache_detail *cd)
1500 {
1501         int ret;
1502
1503         sunrpc_init_cache_detail(cd);
1504         ret = create_cache_proc_entries(cd);
1505         if (ret)
1506                 sunrpc_destroy_cache_detail(cd);
1507         return ret;
1508 }
1509 EXPORT_SYMBOL_GPL(cache_register);
1510
1511 void cache_unregister(struct cache_detail *cd)
1512 {
1513         remove_cache_proc_entries(cd);
1514         sunrpc_destroy_cache_detail(cd);
1515 }
1516 EXPORT_SYMBOL_GPL(cache_unregister);
1517
1518 static ssize_t cache_read_pipefs(struct file *filp, char __user *buf,
1519                                  size_t count, loff_t *ppos)
1520 {
1521         struct cache_detail *cd = RPC_I(filp->f_path.dentry->d_inode)->private;
1522
1523         return cache_read(filp, buf, count, ppos, cd);
1524 }
1525
1526 static ssize_t cache_write_pipefs(struct file *filp, const char __user *buf,
1527                                   size_t count, loff_t *ppos)
1528 {
1529         struct cache_detail *cd = RPC_I(filp->f_path.dentry->d_inode)->private;
1530
1531         return cache_write(filp, buf, count, ppos, cd);
1532 }
1533
1534 static unsigned int cache_poll_pipefs(struct file *filp, poll_table *wait)
1535 {
1536         struct cache_detail *cd = RPC_I(filp->f_path.dentry->d_inode)->private;
1537
1538         return cache_poll(filp, wait, cd);
1539 }
1540
1541 static int cache_ioctl_pipefs(struct inode *inode, struct file *filp,
1542                               unsigned int cmd, unsigned long arg)
1543 {
1544         struct cache_detail *cd = RPC_I(inode)->private;
1545
1546         return cache_ioctl(inode, filp, cmd, arg, cd);
1547 }
1548
1549 static int cache_open_pipefs(struct inode *inode, struct file *filp)
1550 {
1551         struct cache_detail *cd = RPC_I(inode)->private;
1552
1553         return cache_open(inode, filp, cd);
1554 }
1555
1556 static int cache_release_pipefs(struct inode *inode, struct file *filp)
1557 {
1558         struct cache_detail *cd = RPC_I(inode)->private;
1559
1560         return cache_release(inode, filp, cd);
1561 }
1562
1563 const struct file_operations cache_file_operations_pipefs = {
1564         .owner          = THIS_MODULE,
1565         .llseek         = no_llseek,
1566         .read           = cache_read_pipefs,
1567         .write          = cache_write_pipefs,
1568         .poll           = cache_poll_pipefs,
1569         .ioctl          = cache_ioctl_pipefs, /* for FIONREAD */
1570         .open           = cache_open_pipefs,
1571         .release        = cache_release_pipefs,
1572 };
1573
1574 static int content_open_pipefs(struct inode *inode, struct file *filp)
1575 {
1576         struct cache_detail *cd = RPC_I(inode)->private;
1577
1578         return content_open(inode, filp, cd);
1579 }
1580
1581 static int content_release_pipefs(struct inode *inode, struct file *filp)
1582 {
1583         struct cache_detail *cd = RPC_I(inode)->private;
1584
1585         return content_release(inode, filp, cd);
1586 }
1587
1588 const struct file_operations content_file_operations_pipefs = {
1589         .open           = content_open_pipefs,
1590         .read           = seq_read,
1591         .llseek         = seq_lseek,
1592         .release        = content_release_pipefs,
1593 };
1594
1595 static int open_flush_pipefs(struct inode *inode, struct file *filp)
1596 {
1597         struct cache_detail *cd = RPC_I(inode)->private;
1598
1599         return open_flush(inode, filp, cd);
1600 }
1601
1602 static int release_flush_pipefs(struct inode *inode, struct file *filp)
1603 {
1604         struct cache_detail *cd = RPC_I(inode)->private;
1605
1606         return release_flush(inode, filp, cd);
1607 }
1608
1609 static ssize_t read_flush_pipefs(struct file *filp, char __user *buf,
1610                             size_t count, loff_t *ppos)
1611 {
1612         struct cache_detail *cd = RPC_I(filp->f_path.dentry->d_inode)->private;
1613
1614         return read_flush(filp, buf, count, ppos, cd);
1615 }
1616
1617 static ssize_t write_flush_pipefs(struct file *filp,
1618                                   const char __user *buf,
1619                                   size_t count, loff_t *ppos)
1620 {
1621         struct cache_detail *cd = RPC_I(filp->f_path.dentry->d_inode)->private;
1622
1623         return write_flush(filp, buf, count, ppos, cd);
1624 }
1625
1626 const struct file_operations cache_flush_operations_pipefs = {
1627         .open           = open_flush_pipefs,
1628         .read           = read_flush_pipefs,
1629         .write          = write_flush_pipefs,
1630         .release        = release_flush_pipefs,
1631 };
1632
1633 int sunrpc_cache_register_pipefs(struct dentry *parent,
1634                                  const char *name, mode_t umode,
1635                                  struct cache_detail *cd)
1636 {
1637         struct qstr q;
1638         struct dentry *dir;
1639         int ret = 0;
1640
1641         sunrpc_init_cache_detail(cd);
1642         q.name = name;
1643         q.len = strlen(name);
1644         q.hash = full_name_hash(q.name, q.len);
1645         dir = rpc_create_cache_dir(parent, &q, umode, cd);
1646         if (!IS_ERR(dir))
1647                 cd->u.pipefs.dir = dir;
1648         else {
1649                 sunrpc_destroy_cache_detail(cd);
1650                 ret = PTR_ERR(dir);
1651         }
1652         return ret;
1653 }
1654 EXPORT_SYMBOL_GPL(sunrpc_cache_register_pipefs);
1655
1656 void sunrpc_cache_unregister_pipefs(struct cache_detail *cd)
1657 {
1658         rpc_remove_cache_dir(cd->u.pipefs.dir);
1659         cd->u.pipefs.dir = NULL;
1660         sunrpc_destroy_cache_detail(cd);
1661 }
1662 EXPORT_SYMBOL_GPL(sunrpc_cache_unregister_pipefs);
1663