fs/btrfs/async-thread.c

   1 /*
   2  * Copyright (C) 2007 Oracle.  All rights reserved.
   3  *
   4  * This program is free software; you can redistribute it and/or
   5  * modify it under the terms of the GNU General Public
   6  * License v2 as published by the Free Software Foundation.
   7  *
   8  * This program is distributed in the hope that it will be useful,
   9  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  10  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  11  * General Public License for more details.
  12  *
  13  * You should have received a copy of the GNU General Public
  14  * License along with this program; if not, write to the
  15  * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  16  * Boston, MA 021110-1307, USA.
  17  */
  18
  19 #include <linux/version.h>
  20 #include <linux/kthread.h>
  21 #include <linux/list.h>
  22 #include <linux/spinlock.h>
  23 # include <linux/freezer.h>
  24 #include "async-thread.h"
  25
  26 #define WORK_QUEUED_BIT 0
  27 #define WORK_DONE_BIT 1
  28 #define WORK_ORDER_DONE_BIT 2
  29
  30 /*
  31  * container for the kthread task pointer and the list of pending work
  32  * One of these is allocated per thread.
  33  */
  34 struct btrfs_worker_thread {
  35         /* pool we belong to */
  36         struct btrfs_workers *workers;
  37
  38         /* list of struct btrfs_work that are waiting for service */
  39         struct list_head pending;
  40
  41         /* list of worker threads from struct btrfs_workers */
  42         struct list_head worker_list;
  43
  44         /* kthread */
  45         struct task_struct *task;
  46
  47         /* number of things on the pending list */
  48         atomic_t num_pending;
  49
  50         unsigned long sequence;
  51
  52         /* protects the pending list. */
  53         spinlock_t lock;
  54
  55         /* set to non-zero when this thread is already awake and kicking */
  56         int working;
  57
  58         /* are we currently idle */
  59         int idle;
  60 };
  61
  62 /*
  63  * helper function to move a thread onto the idle list after it
  64  * has finished some requests.
  65  */
  66 static void check_idle_worker(struct btrfs_worker_thread *worker)
  67 {
  68         if (!worker->idle && atomic_read(&worker->num_pending) <
  69             worker->workers->idle_thresh / 2) {
  70                 unsigned long flags;
  71                 spin_lock_irqsave(&worker->workers->lock, flags);
  72                 worker->idle = 1;
  73                 list_move(&worker->worker_list, &worker->workers->idle_list);
  74                 spin_unlock_irqrestore(&worker->workers->lock, flags);
  75         }
  76 }
  77
  78 /*
  79  * helper function to move a thread off the idle list after new
  80  * pending work is added.
  81  */
  82 static void check_busy_worker(struct btrfs_worker_thread *worker)
  83 {
  84         if (worker->idle && atomic_read(&worker->num_pending) >=
  85             worker->workers->idle_thresh) {
  86                 unsigned long flags;
  87                 spin_lock_irqsave(&worker->workers->lock, flags);
  88                 worker->idle = 0;
  89                 list_move_tail(&worker->worker_list,
  90                                &worker->workers->worker_list);
  91                 spin_unlock_irqrestore(&worker->workers->lock, flags);
  92         }
  93 }
  94
  95 static noinline int run_ordered_completions(struct btrfs_workers *workers,
  96                                             struct btrfs_work *work)
  97 {
  98         unsigned long flags;
  99
 100         if (!workers->ordered)
 101                 return 0;
 102
 103         set_bit(WORK_DONE_BIT, &work->flags);
 104
 105         spin_lock_irqsave(&workers->lock, flags);
 106
 107         while (!list_empty(&workers->order_list)) {
 108                 work = list_entry(workers->order_list.next,
 109                                   struct btrfs_work, order_list);
 110
 111                 if (!test_bit(WORK_DONE_BIT, &work->flags))
 112                         break;
 113
 114                 /* we are going to call the ordered done function, but
 115                  * we leave the work item on the list as a barrier so
 116                  * that later work items that are done don't have their
 117                  * functions called before this one returns
 118                  */
 119                 if (test_and_set_bit(WORK_ORDER_DONE_BIT, &work->flags))
 120                         break;
 121
 122                 spin_unlock_irqrestore(&workers->lock, flags);
 123
 124                 work->ordered_func(work);
 125
 126                 /* now take the lock again and call the freeing code */
 127                 spin_lock_irqsave(&workers->lock, flags);
 128                 list_del(&work->order_list);
 129                 work->ordered_free(work);
 130         }
 131
 132         spin_unlock_irqrestore(&workers->lock, flags);
 133         return 0;
 134 }
 135
 136 /*
 137  * main loop for servicing work items
 138  */
 139 static int worker_loop(void *arg)
 140 {
 141         struct btrfs_worker_thread *worker = arg;
 142         struct list_head *cur;
 143         struct btrfs_work *work;
 144         do {
 145                 spin_lock_irq(&worker->lock);
 146                 while (!list_empty(&worker->pending)) {
 147                         cur = worker->pending.next;
 148                         work = list_entry(cur, struct btrfs_work, list);
 149                         list_del(&work->list);
 150                         clear_bit(WORK_QUEUED_BIT, &work->flags);
 151
 152                         work->worker = worker;
 153                         spin_unlock_irq(&worker->lock);
 154
 155                         work->func(work);
 156
 157                         atomic_dec(&worker->num_pending);
 158                         /*
 159                          * unless this is an ordered work queue,
 160                          * 'work' was probably freed by func above.
 161                          */
 162                         run_ordered_completions(worker->workers, work);
 163
 164                         spin_lock_irq(&worker->lock);
 165                         check_idle_worker(worker);
 166
 167                 }
 168                 worker->working = 0;
 169                 if (freezing(current)) {
 170                         refrigerator();
 171                 } else {
 172                         set_current_state(TASK_INTERRUPTIBLE);
 173                         spin_unlock_irq(&worker->lock);
 174                         if (!kthread_should_stop())
 175                                 schedule();
 176                         __set_current_state(TASK_RUNNING);
 177                 }
 178         } while (!kthread_should_stop());
 179         return 0;
 180 }
 181
 182 /*
 183  * this will wait for all the worker threads to shutdown
 184  */
 185 int btrfs_stop_workers(struct btrfs_workers *workers)
 186 {
 187         struct list_head *cur;
 188         struct btrfs_worker_thread *worker;
 189
 190         list_splice_init(&workers->idle_list, &workers->worker_list);
 191         while (!list_empty(&workers->worker_list)) {
 192                 cur = workers->worker_list.next;
 193                 worker = list_entry(cur, struct btrfs_worker_thread,
 194                                     worker_list);
 195                 kthread_stop(worker->task);
 196                 list_del(&worker->worker_list);
 197                 kfree(worker);
 198         }
 199         return 0;
 200 }
 201
 202 /*
 203  * simple init on struct btrfs_workers
 204  */
 205 void btrfs_init_workers(struct btrfs_workers *workers, char *name, int max)
 206 {
 207         workers->num_workers = 0;
 208         INIT_LIST_HEAD(&workers->worker_list);
 209         INIT_LIST_HEAD(&workers->idle_list);
 210         INIT_LIST_HEAD(&workers->order_list);
 211         spin_lock_init(&workers->lock);
 212         workers->max_workers = max;
 213         workers->idle_thresh = 32;
 214         workers->name = name;
 215         workers->ordered = 0;
 216 }
 217
 218 /*
 219  * starts new worker threads.  This does not enforce the max worker
 220  * count in case you need to temporarily go past it.
 221  */
 222 int btrfs_start_workers(struct btrfs_workers *workers, int num_workers)
 223 {
 224         struct btrfs_worker_thread *worker;
 225         int ret = 0;
 226         int i;
 227
 228         for (i = 0; i < num_workers; i++) {
 229                 worker = kzalloc(sizeof(*worker), GFP_NOFS);
 230                 if (!worker) {
 231                         ret = -ENOMEM;
 232                         goto fail;
 233                 }
 234
 235                 INIT_LIST_HEAD(&worker->pending);
 236                 INIT_LIST_HEAD(&worker->worker_list);
 237                 spin_lock_init(&worker->lock);
 238                 atomic_set(&worker->num_pending, 0);
 239                 worker->task = kthread_run(worker_loop, worker,
 240                                            "btrfs-%s-%d", workers->name,
 241                                            workers->num_workers + i);
 242                 worker->workers = workers;
 243                 if (IS_ERR(worker->task)) {
 244                         kfree(worker);
 245                         ret = PTR_ERR(worker->task);
 246                         goto fail;
 247                 }
 248
 249                 spin_lock_irq(&workers->lock);
 250                 list_add_tail(&worker->worker_list, &workers->idle_list);
 251                 worker->idle = 1;
 252                 workers->num_workers++;
 253                 spin_unlock_irq(&workers->lock);
 254         }
 255         return 0;
 256 fail:
 257         btrfs_stop_workers(workers);
 258         return ret;
 259 }
 260
 261 /*
 262  * run through the list and find a worker thread that doesn't have a lot
 263  * to do right now.  This can return null if we aren't yet at the thread
 264  * count limit and all of the threads are busy.
 265  */
 266 static struct btrfs_worker_thread *next_worker(struct btrfs_workers *workers)
 267 {
 268         struct btrfs_worker_thread *worker;
 269         struct list_head *next;
 270         int enforce_min = workers->num_workers < workers->max_workers;
 271
 272         /*
 273          * if we find an idle thread, don't move it to the end of the
 274          * idle list.  This improves the chance that the next submission
 275          * will reuse the same thread, and maybe catch it while it is still
 276          * working
 277          */
 278         if (!list_empty(&workers->idle_list)) {
 279                 next = workers->idle_list.next;
 280                 worker = list_entry(next, struct btrfs_worker_thread,
 281                                     worker_list);
 282                 return worker;
 283         }
 284         if (enforce_min || list_empty(&workers->worker_list))
 285                 return NULL;
 286
 287         /*
 288          * if we pick a busy task, move the task to the end of the list.
 289          * hopefully this will keep things somewhat evenly balanced.
 290          * Do the move in batches based on the sequence number.  This groups
 291          * requests submitted at roughly the same time onto the same worker.
 292          */
 293         next = workers->worker_list.next;
 294         worker = list_entry(next, struct btrfs_worker_thread, worker_list);
 295         atomic_inc(&worker->num_pending);
 296         worker->sequence++;
 297
 298         if (worker->sequence % workers->idle_thresh == 0)
 299                 list_move_tail(next, &workers->worker_list);
 300         return worker;
 301 }
 302
 303 /*
 304  * selects a worker thread to take the next job.  This will either find
 305  * an idle worker, start a new worker up to the max count, or just return
 306  * one of the existing busy workers.
 307  */
 308 static struct btrfs_worker_thread *find_worker(struct btrfs_workers *workers)
 309 {
 310         struct btrfs_worker_thread *worker;
 311         unsigned long flags;
 312
 313 again:
 314         spin_lock_irqsave(&workers->lock, flags);
 315         worker = next_worker(workers);
 316         spin_unlock_irqrestore(&workers->lock, flags);
 317
 318         if (!worker) {
 319                 spin_lock_irqsave(&workers->lock, flags);
 320                 if (workers->num_workers >= workers->max_workers) {
 321                         struct list_head *fallback = NULL;
 322                         /*
 323                          * we have failed to find any workers, just
 324                          * return the force one
 325                          */
 326                         if (!list_empty(&workers->worker_list))
 327                                 fallback = workers->worker_list.next;
 328                         if (!list_empty(&workers->idle_list))
 329                                 fallback = workers->idle_list.next;
 330                         BUG_ON(!fallback);
 331                         worker = list_entry(fallback,
 332                                   struct btrfs_worker_thread, worker_list);
 333                         spin_unlock_irqrestore(&workers->lock, flags);
 334                 } else {
 335                         spin_unlock_irqrestore(&workers->lock, flags);
 336                         /* we're below the limit, start another worker */
 337                         btrfs_start_workers(workers, 1);
 338                         goto again;
 339                 }
 340         }
 341         return worker;
 342 }
 343
 344 /*
 345  * btrfs_requeue_work just puts the work item back on the tail of the list
 346  * it was taken from.  It is intended for use with long running work functions
 347  * that make some progress and want to give the cpu up for others.
 348  */
 349 int btrfs_requeue_work(struct btrfs_work *work)
 350 {
 351         struct btrfs_worker_thread *worker = work->worker;
 352         unsigned long flags;
 353
 354         if (test_and_set_bit(WORK_QUEUED_BIT, &work->flags))
 355                 goto out;
 356
 357         spin_lock_irqsave(&worker->lock, flags);
 358         atomic_inc(&worker->num_pending);
 359         list_add_tail(&work->list, &worker->pending);
 360
 361         /* by definition we're busy, take ourselves off the idle
 362          * list
 363          */
 364         if (worker->idle) {
 365                 spin_lock_irqsave(&worker->workers->lock, flags);
 366                 worker->idle = 0;
 367                 list_move_tail(&worker->worker_list,
 368                                &worker->workers->worker_list);
 369                 spin_unlock_irqrestore(&worker->workers->lock, flags);
 370         }
 371
 372         spin_unlock_irqrestore(&worker->lock, flags);
 373
 374 out:
 375         return 0;
 376 }
 377
 378 /*
 379  * places a struct btrfs_work into the pending queue of one of the kthreads
 380  */
 381 int btrfs_queue_worker(struct btrfs_workers *workers, struct btrfs_work *work)
 382 {
 383         struct btrfs_worker_thread *worker;
 384         unsigned long flags;
 385         int wake = 0;
 386
 387         /* don't requeue something already on a list */
 388         if (test_and_set_bit(WORK_QUEUED_BIT, &work->flags))
 389                 goto out;
 390
 391         worker = find_worker(workers);
 392         if (workers->ordered) {
 393                 spin_lock_irqsave(&workers->lock, flags);
 394                 list_add_tail(&work->order_list, &workers->order_list);
 395                 spin_unlock_irqrestore(&workers->lock, flags);
 396         } else {
 397                 INIT_LIST_HEAD(&work->order_list);
 398         }
 399
 400         spin_lock_irqsave(&worker->lock, flags);
 401         atomic_inc(&worker->num_pending);
 402         check_busy_worker(worker);
 403         list_add_tail(&work->list, &worker->pending);
 404
 405         /*
 406          * avoid calling into wake_up_process if this thread has already
 407          * been kicked
 408          */
 409         if (!worker->working)
 410                 wake = 1;
 411         worker->working = 1;
 412
 413         spin_unlock_irqrestore(&worker->lock, flags);
 414
 415         if (wake)
 416                 wake_up_process(worker->task);
 417 out:
 418         return 0;
 419 }