net/sunrpc/sched.c

   1 /*
   2  * linux/net/sunrpc/sched.c
   3  *
   4  * Scheduling for synchronous and asynchronous RPC requests.
   5  *
   6  * Copyright (C) 1996 Olaf Kirch, <okir@monad.swb.de>
   7  *
   8  * TCP NFS related read + write fixes
   9  * (C) 1999 Dave Airlie, University of Limerick, Ireland <airlied@linux.ie>
  10  */
  11
  12 #include <linux/module.h>
  13
  14 #include <linux/sched.h>
  15 #include <linux/interrupt.h>
  16 #include <linux/slab.h>
  17 #include <linux/mempool.h>
  18 #include <linux/smp.h>
  19 #include <linux/smp_lock.h>
  20 #include <linux/spinlock.h>
  21 #include <linux/mutex.h>
  22
  23 #include <linux/sunrpc/clnt.h>
  24
  25 #ifdef RPC_DEBUG
  26 #define RPCDBG_FACILITY         RPCDBG_SCHED
  27 #define RPC_TASK_MAGIC_ID       0xf00baa
  28 #endif
  29
  30 /*
  31  * RPC slabs and memory pools
  32  */
  33 #define RPC_BUFFER_MAXSIZE      (2048)
  34 #define RPC_BUFFER_POOLSIZE     (8)
  35 #define RPC_TASK_POOLSIZE       (8)
  36 static struct kmem_cache        *rpc_task_slabp __read_mostly;
  37 static struct kmem_cache        *rpc_buffer_slabp __read_mostly;
  38 static mempool_t        *rpc_task_mempool __read_mostly;
  39 static mempool_t        *rpc_buffer_mempool __read_mostly;
  40
  41 static void                     __rpc_default_timer(struct rpc_task *task);
  42 static void                     rpc_async_schedule(struct work_struct *);
  43 static void                      rpc_release_task(struct rpc_task *task);
  44
  45 /*
  46  * RPC tasks sit here while waiting for conditions to improve.
  47  */
  48 static RPC_WAITQ(delay_queue, "delayq");
  49
  50 /*
  51  * rpciod-related stuff
  52  */
  53 struct workqueue_struct *rpciod_workqueue;
  54
  55 /*
  56  * Disable the timer for a given RPC task. Should be called with
  57  * queue->lock and bh_disabled in order to avoid races within
  58  * rpc_run_timer().
  59  */
  60 static inline void
  61 __rpc_disable_timer(struct rpc_task *task)
  62 {
  63         dprintk("RPC: %5u disabling timer\n", task->tk_pid);
  64         task->tk_timeout_fn = NULL;
  65         task->tk_timeout = 0;
  66 }
  67
  68 /*
  69  * Run a timeout function.
  70  * We use the callback in order to allow __rpc_wake_up_task()
  71  * and friends to disable the timer synchronously on SMP systems
  72  * without calling del_timer_sync(). The latter could cause a
  73  * deadlock if called while we're holding spinlocks...
  74  */
  75 static void rpc_run_timer(struct rpc_task *task)
  76 {
  77         void (*callback)(struct rpc_task *);
  78
  79         callback = task->tk_timeout_fn;
  80         task->tk_timeout_fn = NULL;
  81         if (callback && RPC_IS_QUEUED(task)) {
  82                 dprintk("RPC: %5u running timer\n", task->tk_pid);
  83                 callback(task);
  84         }
  85         smp_mb__before_clear_bit();
  86         clear_bit(RPC_TASK_HAS_TIMER, &task->tk_runstate);
  87         smp_mb__after_clear_bit();
  88 }
  89
  90 /*
  91  * Set up a timer for the current task.
  92  */
  93 static inline void
  94 __rpc_add_timer(struct rpc_task *task, rpc_action timer)
  95 {
  96         if (!task->tk_timeout)
  97                 return;
  98
  99         dprintk("RPC: %5u setting alarm for %lu ms\n",
 100                         task->tk_pid, task->tk_timeout * 1000 / HZ);
 101
 102         if (timer)
 103                 task->tk_timeout_fn = timer;
 104         else
 105                 task->tk_timeout_fn = __rpc_default_timer;
 106         set_bit(RPC_TASK_HAS_TIMER, &task->tk_runstate);
 107         mod_timer(&task->tk_timer, jiffies + task->tk_timeout);
 108 }
 109
 110 /*
 111  * Delete any timer for the current task. Because we use del_timer_sync(),
 112  * this function should never be called while holding queue->lock.
 113  */
 114 static void
 115 rpc_delete_timer(struct rpc_task *task)
 116 {
 117         if (RPC_IS_QUEUED(task))
 118                 return;
 119         if (test_and_clear_bit(RPC_TASK_HAS_TIMER, &task->tk_runstate)) {
 120                 del_singleshot_timer_sync(&task->tk_timer);
 121                 dprintk("RPC: %5u deleting timer\n", task->tk_pid);
 122         }
 123 }
 124
 125 /*
 126  * Add new request to a priority queue.
 127  */
 128 static void __rpc_add_wait_queue_priority(struct rpc_wait_queue *queue, struct rpc_task *task)
 129 {
 130         struct list_head *q;
 131         struct rpc_task *t;
 132
 133         INIT_LIST_HEAD(&task->u.tk_wait.links);
 134         q = &queue->tasks[task->tk_priority];
 135         if (unlikely(task->tk_priority > queue->maxpriority))
 136                 q = &queue->tasks[queue->maxpriority];
 137         list_for_each_entry(t, q, u.tk_wait.list) {
 138                 if (t->tk_cookie == task->tk_cookie) {
 139                         list_add_tail(&task->u.tk_wait.list, &t->u.tk_wait.links);
 140                         return;
 141                 }
 142         }
 143         list_add_tail(&task->u.tk_wait.list, q);
 144 }
 145
 146 /*
 147  * Add new request to wait queue.
 148  *
 149  * Swapper tasks always get inserted at the head of the queue.
 150  * This should avoid many nasty memory deadlocks and hopefully
 151  * improve overall performance.
 152  * Everyone else gets appended to the queue to ensure proper FIFO behavior.
 153  */
 154 static void __rpc_add_wait_queue(struct rpc_wait_queue *queue, struct rpc_task *task)
 155 {
 156         BUG_ON (RPC_IS_QUEUED(task));
 157
 158         if (RPC_IS_PRIORITY(queue))
 159                 __rpc_add_wait_queue_priority(queue, task);
 160         else if (RPC_IS_SWAPPER(task))
 161                 list_add(&task->u.tk_wait.list, &queue->tasks[0]);
 162         else
 163                 list_add_tail(&task->u.tk_wait.list, &queue->tasks[0]);
 164         task->u.tk_wait.rpc_waitq = queue;
 165         queue->qlen++;
 166         rpc_set_queued(task);
 167
 168         dprintk("RPC: %5u added to queue %p \"%s\"\n",
 169                         task->tk_pid, queue, rpc_qname(queue));
 170 }
 171
 172 /*
 173  * Remove request from a priority queue.
 174  */
 175 static void __rpc_remove_wait_queue_priority(struct rpc_task *task)
 176 {
 177         struct rpc_task *t;
 178
 179         if (!list_empty(&task->u.tk_wait.links)) {
 180                 t = list_entry(task->u.tk_wait.links.next, struct rpc_task, u.tk_wait.list);
 181                 list_move(&t->u.tk_wait.list, &task->u.tk_wait.list);
 182                 list_splice_init(&task->u.tk_wait.links, &t->u.tk_wait.links);
 183         }
 184         list_del(&task->u.tk_wait.list);
 185 }
 186
 187 /*
 188  * Remove request from queue.
 189  * Note: must be called with spin lock held.
 190  */
 191 static void __rpc_remove_wait_queue(struct rpc_task *task)
 192 {
 193         struct rpc_wait_queue *queue;
 194         queue = task->u.tk_wait.rpc_waitq;
 195
 196         if (RPC_IS_PRIORITY(queue))
 197                 __rpc_remove_wait_queue_priority(task);
 198         else
 199                 list_del(&task->u.tk_wait.list);
 200         queue->qlen--;
 201         dprintk("RPC: %5u removed from queue %p \"%s\"\n",
 202                         task->tk_pid, queue, rpc_qname(queue));
 203 }
 204
 205 static inline void rpc_set_waitqueue_priority(struct rpc_wait_queue *queue, int priority)
 206 {
 207         queue->priority = priority;
 208         queue->count = 1 << (priority * 2);
 209 }
 210
 211 static inline void rpc_set_waitqueue_cookie(struct rpc_wait_queue *queue, unsigned long cookie)
 212 {
 213         queue->cookie = cookie;
 214         queue->nr = RPC_BATCH_COUNT;
 215 }
 216
 217 static inline void rpc_reset_waitqueue_priority(struct rpc_wait_queue *queue)
 218 {
 219         rpc_set_waitqueue_priority(queue, queue->maxpriority);
 220         rpc_set_waitqueue_cookie(queue, 0);
 221 }
 222
 223 static void __rpc_init_priority_wait_queue(struct rpc_wait_queue *queue, const char *qname, int maxprio)
 224 {
 225         int i;
 226
 227         spin_lock_init(&queue->lock);
 228         for (i = 0; i < ARRAY_SIZE(queue->tasks); i++)
 229                 INIT_LIST_HEAD(&queue->tasks[i]);
 230         queue->maxpriority = maxprio;
 231         rpc_reset_waitqueue_priority(queue);
 232 #ifdef RPC_DEBUG
 233         queue->name = qname;
 234 #endif
 235 }
 236
 237 void rpc_init_priority_wait_queue(struct rpc_wait_queue *queue, const char *qname)
 238 {
 239         __rpc_init_priority_wait_queue(queue, qname, RPC_PRIORITY_HIGH);
 240 }
 241
 242 void rpc_init_wait_queue(struct rpc_wait_queue *queue, const char *qname)
 243 {
 244         __rpc_init_priority_wait_queue(queue, qname, 0);
 245 }
 246 EXPORT_SYMBOL(rpc_init_wait_queue);
 247
 248 static int rpc_wait_bit_killable(void *word)
 249 {
 250         if (fatal_signal_pending(current))
 251                 return -ERESTARTSYS;
 252         schedule();
 253         return 0;
 254 }
 255
 256 #ifdef RPC_DEBUG
 257 static void rpc_task_set_debuginfo(struct rpc_task *task)
 258 {
 259         static atomic_t rpc_pid;
 260
 261         task->tk_magic = RPC_TASK_MAGIC_ID;
 262         task->tk_pid = atomic_inc_return(&rpc_pid);
 263 }
 264 #else
 265 static inline void rpc_task_set_debuginfo(struct rpc_task *task)
 266 {
 267 }
 268 #endif
 269
 270 static void rpc_set_active(struct rpc_task *task)
 271 {
 272         struct rpc_clnt *clnt;
 273         if (test_and_set_bit(RPC_TASK_ACTIVE, &task->tk_runstate) != 0)
 274                 return;
 275         rpc_task_set_debuginfo(task);
 276         /* Add to global list of all tasks */
 277         clnt = task->tk_client;
 278         if (clnt != NULL) {
 279                 spin_lock(&clnt->cl_lock);
 280                 list_add_tail(&task->tk_task, &clnt->cl_tasks);
 281                 spin_unlock(&clnt->cl_lock);
 282         }
 283 }
 284
 285 /*
 286  * Mark an RPC call as having completed by clearing the 'active' bit
 287  */
 288 static void rpc_mark_complete_task(struct rpc_task *task)
 289 {
 290         smp_mb__before_clear_bit();
 291         clear_bit(RPC_TASK_ACTIVE, &task->tk_runstate);
 292         smp_mb__after_clear_bit();
 293         wake_up_bit(&task->tk_runstate, RPC_TASK_ACTIVE);
 294 }
 295
 296 /*
 297  * Allow callers to wait for completion of an RPC call
 298  */
 299 int __rpc_wait_for_completion_task(struct rpc_task *task, int (*action)(void *))
 300 {
 301         if (action == NULL)
 302                 action = rpc_wait_bit_killable;
 303         return wait_on_bit(&task->tk_runstate, RPC_TASK_ACTIVE,
 304                         action, TASK_KILLABLE);
 305 }
 306 EXPORT_SYMBOL(__rpc_wait_for_completion_task);
 307
 308 /*
 309  * Make an RPC task runnable.
 310  *
 311  * Note: If the task is ASYNC, this must be called with
 312  * the spinlock held to protect the wait queue operation.
 313  */
 314 static void rpc_make_runnable(struct rpc_task *task)
 315 {
 316         BUG_ON(task->tk_timeout_fn);
 317         rpc_clear_queued(task);
 318         if (rpc_test_and_set_running(task))
 319                 return;
 320         /* We might have raced */
 321         if (RPC_IS_QUEUED(task)) {
 322                 rpc_clear_running(task);
 323                 return;
 324         }
 325         if (RPC_IS_ASYNC(task)) {
 326                 int status;
 327
 328                 INIT_WORK(&task->u.tk_work, rpc_async_schedule);
 329                 status = queue_work(task->tk_workqueue, &task->u.tk_work);
 330                 if (status < 0) {
 331                         printk(KERN_WARNING "RPC: failed to add task to queue: error: %d!\n", status);
 332                         task->tk_status = status;
 333                         return;
 334                 }
 335         } else
 336                 wake_up_bit(&task->tk_runstate, RPC_TASK_QUEUED);
 337 }
 338
 339 /*
 340  * Prepare for sleeping on a wait queue.
 341  * By always appending tasks to the list we ensure FIFO behavior.
 342  * NB: An RPC task will only receive interrupt-driven events as long
 343  * as it's on a wait queue.
 344  */
 345 static void __rpc_sleep_on(struct rpc_wait_queue *q, struct rpc_task *task,
 346                         rpc_action action, rpc_action timer)
 347 {
 348         dprintk("RPC: %5u sleep_on(queue \"%s\" time %lu)\n",
 349                         task->tk_pid, rpc_qname(q), jiffies);
 350
 351         if (!RPC_IS_ASYNC(task) && !RPC_IS_ACTIVATED(task)) {
 352                 printk(KERN_ERR "RPC: Inactive synchronous task put to sleep!\n");
 353                 return;
 354         }
 355
 356         __rpc_add_wait_queue(q, task);
 357
 358         BUG_ON(task->tk_callback != NULL);
 359         task->tk_callback = action;
 360         __rpc_add_timer(task, timer);
 361 }
 362
 363 void rpc_sleep_on(struct rpc_wait_queue *q, struct rpc_task *task,
 364                                 rpc_action action, rpc_action timer)
 365 {
 366         /* Mark the task as being activated if so needed */
 367         rpc_set_active(task);
 368
 369         /*
 370          * Protect the queue operations.
 371          */
 372         spin_lock_bh(&q->lock);
 373         __rpc_sleep_on(q, task, action, timer);
 374         spin_unlock_bh(&q->lock);
 375 }
 376
 377 /**
 378  * __rpc_do_wake_up_task - wake up a single rpc_task
 379  * @task: task to be woken up
 380  *
 381  * Caller must hold queue->lock, and have cleared the task queued flag.
 382  */
 383 static void __rpc_do_wake_up_task(struct rpc_task *task)
 384 {
 385         dprintk("RPC: %5u __rpc_wake_up_task (now %lu)\n",
 386                         task->tk_pid, jiffies);
 387
 388 #ifdef RPC_DEBUG
 389         BUG_ON(task->tk_magic != RPC_TASK_MAGIC_ID);
 390 #endif
 391         /* Has the task been executed yet? If not, we cannot wake it up! */
 392         if (!RPC_IS_ACTIVATED(task)) {
 393                 printk(KERN_ERR "RPC: Inactive task (%p) being woken up!\n", task);
 394                 return;
 395         }
 396
 397         __rpc_disable_timer(task);
 398         __rpc_remove_wait_queue(task);
 399
 400         rpc_make_runnable(task);
 401
 402         dprintk("RPC:       __rpc_wake_up_task done\n");
 403 }
 404
 405 /*
 406  * Wake up the specified task
 407  */
 408 static void __rpc_wake_up_task(struct rpc_task *task)
 409 {
 410         if (rpc_start_wakeup(task)) {
 411                 if (RPC_IS_QUEUED(task))
 412                         __rpc_do_wake_up_task(task);
 413                 rpc_finish_wakeup(task);
 414         }
 415 }
 416
 417 /*
 418  * Default timeout handler if none specified by user
 419  */
 420 static void
 421 __rpc_default_timer(struct rpc_task *task)
 422 {
 423         dprintk("RPC: %5u timeout (default timer)\n", task->tk_pid);
 424         task->tk_status = -ETIMEDOUT;
 425         rpc_wake_up_task(task);
 426 }
 427
 428 /*
 429  * Wake up the specified task
 430  */
 431 void rpc_wake_up_task(struct rpc_task *task)
 432 {
 433         rcu_read_lock_bh();
 434         if (rpc_start_wakeup(task)) {
 435                 if (RPC_IS_QUEUED(task)) {
 436                         struct rpc_wait_queue *queue = task->u.tk_wait.rpc_waitq;
 437
 438                         /* Note: we're already in a bh-safe context */
 439                         spin_lock(&queue->lock);
 440                         __rpc_do_wake_up_task(task);
 441                         spin_unlock(&queue->lock);
 442                 }
 443                 rpc_finish_wakeup(task);
 444         }
 445         rcu_read_unlock_bh();
 446 }
 447
 448 /*
 449  * Wake up the next task on a priority queue.
 450  */
 451 static struct rpc_task * __rpc_wake_up_next_priority(struct rpc_wait_queue *queue)
 452 {
 453         struct list_head *q;
 454         struct rpc_task *task;
 455
 456         /*
 457          * Service a batch of tasks from a single cookie.
 458          */
 459         q = &queue->tasks[queue->priority];
 460         if (!list_empty(q)) {
 461                 task = list_entry(q->next, struct rpc_task, u.tk_wait.list);
 462                 if (queue->cookie == task->tk_cookie) {
 463                         if (--queue->nr)
 464                                 goto out;
 465                         list_move_tail(&task->u.tk_wait.list, q);
 466                 }
 467                 /*
 468                  * Check if we need to switch queues.
 469                  */
 470                 if (--queue->count)
 471                         goto new_cookie;
 472         }
 473
 474         /*
 475          * Service the next queue.
 476          */
 477         do {
 478                 if (q == &queue->tasks[0])
 479                         q = &queue->tasks[queue->maxpriority];
 480                 else
 481                         q = q - 1;
 482                 if (!list_empty(q)) {
 483                         task = list_entry(q->next, struct rpc_task, u.tk_wait.list);
 484                         goto new_queue;
 485                 }
 486         } while (q != &queue->tasks[queue->priority]);
 487
 488         rpc_reset_waitqueue_priority(queue);
 489         return NULL;
 490
 491 new_queue:
 492         rpc_set_waitqueue_priority(queue, (unsigned int)(q - &queue->tasks[0]));
 493 new_cookie:
 494         rpc_set_waitqueue_cookie(queue, task->tk_cookie);
 495 out:
 496         __rpc_wake_up_task(task);
 497         return task;
 498 }
 499
 500 /*
 501  * Wake up the next task on the wait queue.
 502  */
 503 struct rpc_task * rpc_wake_up_next(struct rpc_wait_queue *queue)
 504 {
 505         struct rpc_task *task = NULL;
 506
 507         dprintk("RPC:       wake_up_next(%p \"%s\")\n",
 508                         queue, rpc_qname(queue));
 509         rcu_read_lock_bh();
 510         spin_lock(&queue->lock);
 511         if (RPC_IS_PRIORITY(queue))
 512                 task = __rpc_wake_up_next_priority(queue);
 513         else {
 514                 task_for_first(task, &queue->tasks[0])
 515                         __rpc_wake_up_task(task);
 516         }
 517         spin_unlock(&queue->lock);
 518         rcu_read_unlock_bh();
 519
 520         return task;
 521 }
 522
 523 /**
 524  * rpc_wake_up - wake up all rpc_tasks
 525  * @queue: rpc_wait_queue on which the tasks are sleeping
 526  *
 527  * Grabs queue->lock
 528  */
 529 void rpc_wake_up(struct rpc_wait_queue *queue)
 530 {
 531         struct rpc_task *task, *next;
 532         struct list_head *head;
 533
 534         rcu_read_lock_bh();
 535         spin_lock(&queue->lock);
 536         head = &queue->tasks[queue->maxpriority];
 537         for (;;) {
 538                 list_for_each_entry_safe(task, next, head, u.tk_wait.list)
 539                         __rpc_wake_up_task(task);
 540                 if (head == &queue->tasks[0])
 541                         break;
 542                 head--;
 543         }
 544         spin_unlock(&queue->lock);
 545         rcu_read_unlock_bh();
 546 }
 547
 548 /**
 549  * rpc_wake_up_status - wake up all rpc_tasks and set their status value.
 550  * @queue: rpc_wait_queue on which the tasks are sleeping
 551  * @status: status value to set
 552  *
 553  * Grabs queue->lock
 554  */
 555 void rpc_wake_up_status(struct rpc_wait_queue *queue, int status)
 556 {
 557         struct rpc_task *task, *next;
 558         struct list_head *head;
 559
 560         rcu_read_lock_bh();
 561         spin_lock(&queue->lock);
 562         head = &queue->tasks[queue->maxpriority];
 563         for (;;) {
 564                 list_for_each_entry_safe(task, next, head, u.tk_wait.list) {
 565                         task->tk_status = status;
 566                         __rpc_wake_up_task(task);
 567                 }
 568                 if (head == &queue->tasks[0])
 569                         break;
 570                 head--;
 571         }
 572         spin_unlock(&queue->lock);
 573         rcu_read_unlock_bh();
 574 }
 575
 576 static void __rpc_atrun(struct rpc_task *task)
 577 {
 578         rpc_wake_up_task(task);
 579 }
 580
 581 /*
 582  * Run a task at a later time
 583  */
 584 void rpc_delay(struct rpc_task *task, unsigned long delay)
 585 {
 586         task->tk_timeout = delay;
 587         rpc_sleep_on(&delay_queue, task, NULL, __rpc_atrun);
 588 }
 589
 590 /*
 591  * Helper to call task->tk_ops->rpc_call_prepare
 592  */
 593 static void rpc_prepare_task(struct rpc_task *task)
 594 {
 595         lock_kernel();
 596         task->tk_ops->rpc_call_prepare(task, task->tk_calldata);
 597         unlock_kernel();
 598 }
 599
 600 /*
 601  * Helper that calls task->tk_ops->rpc_call_done if it exists
 602  */
 603 void rpc_exit_task(struct rpc_task *task)
 604 {
 605         task->tk_action = NULL;
 606         if (task->tk_ops->rpc_call_done != NULL) {
 607                 lock_kernel();
 608                 task->tk_ops->rpc_call_done(task, task->tk_calldata);
 609                 unlock_kernel();
 610                 if (task->tk_action != NULL) {
 611                         WARN_ON(RPC_ASSASSINATED(task));
 612                         /* Always release the RPC slot and buffer memory */
 613                         xprt_release(task);
 614                 }
 615         }
 616 }
 617 EXPORT_SYMBOL(rpc_exit_task);
 618
 619 void rpc_release_calldata(const struct rpc_call_ops *ops, void *calldata)
 620 {
 621         if (ops->rpc_release != NULL) {
 622                 lock_kernel();
 623                 ops->rpc_release(calldata);
 624                 unlock_kernel();
 625         }
 626 }
 627
 628 /*
 629  * This is the RPC `scheduler' (or rather, the finite state machine).
 630  */
 631 static void __rpc_execute(struct rpc_task *task)
 632 {
 633         int             status = 0;
 634
 635         dprintk("RPC: %5u __rpc_execute flags=0x%x\n",
 636                         task->tk_pid, task->tk_flags);
 637
 638         BUG_ON(RPC_IS_QUEUED(task));
 639
 640         for (;;) {
 641                 /*
 642                  * Garbage collection of pending timers...
 643                  */
 644                 rpc_delete_timer(task);
 645
 646                 /*
 647                  * Execute any pending callback.
 648                  */
 649                 if (RPC_DO_CALLBACK(task)) {
 650                         /* Define a callback save pointer */
 651                         void (*save_callback)(struct rpc_task *);
 652
 653                         /*
 654                          * If a callback exists, save it, reset it,
 655                          * call it.
 656                          * The save is needed to stop from resetting
 657                          * another callback set within the callback handler
 658                          * - Dave
 659                          */
 660                         save_callback=task->tk_callback;
 661                         task->tk_callback=NULL;
 662                         save_callback(task);
 663                 }
 664
 665                 /*
 666                  * Perform the next FSM step.
 667                  * tk_action may be NULL when the task has been killed
 668                  * by someone else.
 669                  */
 670                 if (!RPC_IS_QUEUED(task)) {
 671                         if (task->tk_action == NULL)
 672                                 break;
 673                         task->tk_action(task);
 674                 }
 675
 676                 /*
 677                  * Lockless check for whether task is sleeping or not.
 678                  */
 679                 if (!RPC_IS_QUEUED(task))
 680                         continue;
 681                 rpc_clear_running(task);
 682                 if (RPC_IS_ASYNC(task)) {
 683                         /* Careful! we may have raced... */
 684                         if (RPC_IS_QUEUED(task))
 685                                 return;
 686                         if (rpc_test_and_set_running(task))
 687                                 return;
 688                         continue;
 689                 }
 690
 691                 /* sync task: sleep here */
 692                 dprintk("RPC: %5u sync task going to sleep\n", task->tk_pid);
 693                 status = out_of_line_wait_on_bit(&task->tk_runstate,
 694                                 RPC_TASK_QUEUED, rpc_wait_bit_killable,
 695                                 TASK_KILLABLE);
 696                 if (status == -ERESTARTSYS) {
 697                         /*
 698                          * When a sync task receives a signal, it exits with
 699                          * -ERESTARTSYS. In order to catch any callbacks that
 700                          * clean up after sleeping on some queue, we don't
 701                          * break the loop here, but go around once more.
 702                          */
 703                         dprintk("RPC: %5u got signal\n", task->tk_pid);
 704                         task->tk_flags |= RPC_TASK_KILLED;
 705                         rpc_exit(task, -ERESTARTSYS);
 706                         rpc_wake_up_task(task);
 707                 }
 708                 rpc_set_running(task);
 709                 dprintk("RPC: %5u sync task resuming\n", task->tk_pid);
 710         }
 711
 712         dprintk("RPC: %5u return %d, status %d\n", task->tk_pid, status,
 713                         task->tk_status);
 714         /* Release all resources associated with the task */
 715         rpc_release_task(task);
 716 }
 717
 718 /*
 719  * User-visible entry point to the scheduler.
 720  *
 721  * This may be called recursively if e.g. an async NFS task updates
 722  * the attributes and finds that dirty pages must be flushed.
 723  * NOTE: Upon exit of this function the task is guaranteed to be
 724  *       released. In particular note that tk_release() will have
 725  *       been called, so your task memory may have been freed.
 726  */
 727 void rpc_execute(struct rpc_task *task)
 728 {
 729         rpc_set_active(task);
 730         rpc_set_running(task);
 731         __rpc_execute(task);
 732 }
 733
 734 static void rpc_async_schedule(struct work_struct *work)
 735 {
 736         __rpc_execute(container_of(work, struct rpc_task, u.tk_work));
 737 }
 738
 739 struct rpc_buffer {
 740         size_t  len;
 741         char    data[];
 742 };
 743
 744 /**
 745  * rpc_malloc - allocate an RPC buffer
 746  * @task: RPC task that will use this buffer
 747  * @size: requested byte size
 748  *
 749  * To prevent rpciod from hanging, this allocator never sleeps,
 750  * returning NULL if the request cannot be serviced immediately.
 751  * The caller can arrange to sleep in a way that is safe for rpciod.
 752  *
 753  * Most requests are 'small' (under 2KiB) and can be serviced from a
 754  * mempool, ensuring that NFS reads and writes can always proceed,
 755  * and that there is good locality of reference for these buffers.
 756  *
 757  * In order to avoid memory starvation triggering more writebacks of
 758  * NFS requests, we avoid using GFP_KERNEL.
 759  */
 760 void *rpc_malloc(struct rpc_task *task, size_t size)
 761 {
 762         struct rpc_buffer *buf;
 763         gfp_t gfp = RPC_IS_SWAPPER(task) ? GFP_ATOMIC : GFP_NOWAIT;
 764
 765         size += sizeof(struct rpc_buffer);
 766         if (size <= RPC_BUFFER_MAXSIZE)
 767                 buf = mempool_alloc(rpc_buffer_mempool, gfp);
 768         else
 769                 buf = kmalloc(size, gfp);
 770
 771         if (!buf)
 772                 return NULL;
 773
 774         buf->len = size;
 775         dprintk("RPC: %5u allocated buffer of size %zu at %p\n",
 776                         task->tk_pid, size, buf);
 777         return &buf->data;
 778 }
 779 EXPORT_SYMBOL_GPL(rpc_malloc);
 780
 781 /**
 782  * rpc_free - free buffer allocated via rpc_malloc
 783  * @buffer: buffer to free
 784  *
 785  */
 786 void rpc_free(void *buffer)
 787 {
 788         size_t size;
 789         struct rpc_buffer *buf;
 790
 791         if (!buffer)
 792                 return;
 793
 794         buf = container_of(buffer, struct rpc_buffer, data);
 795         size = buf->len;
 796
 797         dprintk("RPC:       freeing buffer of size %zu at %p\n",
 798                         size, buf);
 799
 800         if (size <= RPC_BUFFER_MAXSIZE)
 801                 mempool_free(buf, rpc_buffer_mempool);
 802         else
 803                 kfree(buf);
 804 }
 805 EXPORT_SYMBOL_GPL(rpc_free);
 806
 807 /*
 808  * Creation and deletion of RPC task structures
 809  */
 810 void rpc_init_task(struct rpc_task *task, struct rpc_clnt *clnt, int flags, const struct rpc_call_ops *tk_ops, void *calldata)
 811 {
 812         memset(task, 0, sizeof(*task));
 813         init_timer(&task->tk_timer);
 814         task->tk_timer.data     = (unsigned long) task;
 815         task->tk_timer.function = (void (*)(unsigned long)) rpc_run_timer;
 816         atomic_set(&task->tk_count, 1);
 817         task->tk_client = clnt;
 818         task->tk_flags  = flags;
 819         task->tk_ops = tk_ops;
 820         if (tk_ops->rpc_call_prepare != NULL)
 821                 task->tk_action = rpc_prepare_task;
 822         task->tk_calldata = calldata;
 823         INIT_LIST_HEAD(&task->tk_task);
 824
 825         /* Initialize retry counters */
 826         task->tk_garb_retry = 2;
 827         task->tk_cred_retry = 2;
 828
 829         task->tk_priority = RPC_PRIORITY_NORMAL;
 830         task->tk_cookie = (unsigned long)current;
 831
 832         /* Initialize workqueue for async tasks */
 833         task->tk_workqueue = rpciod_workqueue;
 834
 835         if (clnt) {
 836                 kref_get(&clnt->cl_kref);
 837                 if (clnt->cl_softrtry)
 838                         task->tk_flags |= RPC_TASK_SOFT;
 839         }
 840
 841         BUG_ON(task->tk_ops == NULL);
 842
 843         /* starting timestamp */
 844         task->tk_start = jiffies;
 845
 846         dprintk("RPC:       new task initialized, procpid %u\n",
 847                                 task_pid_nr(current));
 848 }
 849
 850 static struct rpc_task *
 851 rpc_alloc_task(void)
 852 {
 853         return (struct rpc_task *)mempool_alloc(rpc_task_mempool, GFP_NOFS);
 854 }
 855
 856 static void rpc_free_task(struct rcu_head *rcu)
 857 {
 858         struct rpc_task *task = container_of(rcu, struct rpc_task, u.tk_rcu);
 859         dprintk("RPC: %5u freeing task\n", task->tk_pid);
 860         mempool_free(task, rpc_task_mempool);
 861 }
 862
 863 /*
 864  * Create a new task for the specified client.
 865  */
 866 struct rpc_task *rpc_new_task(struct rpc_clnt *clnt, int flags, const struct rpc_call_ops *tk_ops, void *calldata)
 867 {
 868         struct rpc_task *task;
 869
 870         task = rpc_alloc_task();
 871         if (!task)
 872                 goto out;
 873
 874         rpc_init_task(task, clnt, flags, tk_ops, calldata);
 875
 876         dprintk("RPC:       allocated task %p\n", task);
 877         task->tk_flags |= RPC_TASK_DYNAMIC;
 878 out:
 879         return task;
 880 }
 881
 882
 883 void rpc_put_task(struct rpc_task *task)
 884 {
 885         const struct rpc_call_ops *tk_ops = task->tk_ops;
 886         void *calldata = task->tk_calldata;
 887
 888         if (!atomic_dec_and_test(&task->tk_count))
 889                 return;
 890         /* Release resources */
 891         if (task->tk_rqstp)
 892                 xprt_release(task);
 893         if (task->tk_msg.rpc_cred)
 894                 rpcauth_unbindcred(task);
 895         if (task->tk_client) {
 896                 rpc_release_client(task->tk_client);
 897                 task->tk_client = NULL;
 898         }
 899         if (task->tk_flags & RPC_TASK_DYNAMIC)
 900                 call_rcu_bh(&task->u.tk_rcu, rpc_free_task);
 901         rpc_release_calldata(tk_ops, calldata);
 902 }
 903 EXPORT_SYMBOL(rpc_put_task);
 904
 905 static void rpc_release_task(struct rpc_task *task)
 906 {
 907 #ifdef RPC_DEBUG
 908         BUG_ON(task->tk_magic != RPC_TASK_MAGIC_ID);
 909 #endif
 910         dprintk("RPC: %5u release task\n", task->tk_pid);
 911
 912         if (!list_empty(&task->tk_task)) {
 913                 struct rpc_clnt *clnt = task->tk_client;
 914                 /* Remove from client task list */
 915                 spin_lock(&clnt->cl_lock);
 916                 list_del(&task->tk_task);
 917                 spin_unlock(&clnt->cl_lock);
 918         }
 919         BUG_ON (RPC_IS_QUEUED(task));
 920
 921         /* Synchronously delete any running timer */
 922         rpc_delete_timer(task);
 923
 924 #ifdef RPC_DEBUG
 925         task->tk_magic = 0;
 926 #endif
 927         /* Wake up anyone who is waiting for task completion */
 928         rpc_mark_complete_task(task);
 929
 930         rpc_put_task(task);
 931 }
 932
 933 /*
 934  * Kill all tasks for the given client.
 935  * XXX: kill their descendants as well?
 936  */
 937 void rpc_killall_tasks(struct rpc_clnt *clnt)
 938 {
 939         struct rpc_task *rovr;
 940
 941
 942         if (list_empty(&clnt->cl_tasks))
 943                 return;
 944         dprintk("RPC:       killing all tasks for client %p\n", clnt);
 945         /*
 946          * Spin lock all_tasks to prevent changes...
 947          */
 948         spin_lock(&clnt->cl_lock);
 949         list_for_each_entry(rovr, &clnt->cl_tasks, tk_task) {
 950                 if (! RPC_IS_ACTIVATED(rovr))
 951                         continue;
 952                 if (!(rovr->tk_flags & RPC_TASK_KILLED)) {
 953                         rovr->tk_flags |= RPC_TASK_KILLED;
 954                         rpc_exit(rovr, -EIO);
 955                         rpc_wake_up_task(rovr);
 956                 }
 957         }
 958         spin_unlock(&clnt->cl_lock);
 959 }
 960
 961 int rpciod_up(void)
 962 {
 963         return try_module_get(THIS_MODULE) ? 0 : -EINVAL;
 964 }
 965
 966 void rpciod_down(void)
 967 {
 968         module_put(THIS_MODULE);
 969 }
 970
 971 /*
 972  * Start up the rpciod workqueue.
 973  */
 974 static int rpciod_start(void)
 975 {
 976         struct workqueue_struct *wq;
 977
 978         /*
 979          * Create the rpciod thread and wait for it to start.
 980          */
 981         dprintk("RPC:       creating workqueue rpciod\n");
 982         wq = create_workqueue("rpciod");
 983         rpciod_workqueue = wq;
 984         return rpciod_workqueue != NULL;
 985 }
 986
 987 static void rpciod_stop(void)
 988 {
 989         struct workqueue_struct *wq = NULL;
 990
 991         if (rpciod_workqueue == NULL)
 992                 return;
 993         dprintk("RPC:       destroying workqueue rpciod\n");
 994
 995         wq = rpciod_workqueue;
 996         rpciod_workqueue = NULL;
 997         destroy_workqueue(wq);
 998 }
 999
1000 void
1001 rpc_destroy_mempool(void)
1002 {
1003         rpciod_stop();
1004         if (rpc_buffer_mempool)
1005                 mempool_destroy(rpc_buffer_mempool);
1006         if (rpc_task_mempool)
1007                 mempool_destroy(rpc_task_mempool);
1008         if (rpc_task_slabp)
1009                 kmem_cache_destroy(rpc_task_slabp);
1010         if (rpc_buffer_slabp)
1011                 kmem_cache_destroy(rpc_buffer_slabp);
1012 }
1013
1014 int
1015 rpc_init_mempool(void)
1016 {
1017         rpc_task_slabp = kmem_cache_create("rpc_tasks",
1018                                              sizeof(struct rpc_task),
1019                                              0, SLAB_HWCACHE_ALIGN,
1020                                              NULL);
1021         if (!rpc_task_slabp)
1022                 goto err_nomem;
1023         rpc_buffer_slabp = kmem_cache_create("rpc_buffers",
1024                                              RPC_BUFFER_MAXSIZE,
1025                                              0, SLAB_HWCACHE_ALIGN,
1026                                              NULL);
1027         if (!rpc_buffer_slabp)
1028                 goto err_nomem;
1029         rpc_task_mempool = mempool_create_slab_pool(RPC_TASK_POOLSIZE,
1030                                                     rpc_task_slabp);
1031         if (!rpc_task_mempool)
1032                 goto err_nomem;
1033         rpc_buffer_mempool = mempool_create_slab_pool(RPC_BUFFER_POOLSIZE,
1034                                                       rpc_buffer_slabp);
1035         if (!rpc_buffer_mempool)
1036                 goto err_nomem;
1037         if (!rpciod_start())
1038                 goto err_nomem;
1039         return 0;
1040 err_nomem:
1041         rpc_destroy_mempool();
1042         return -ENOMEM;
1043 }