fs/io-wq.c

   1 // SPDX-License-Identifier: GPL-2.0
   2 /*
   3  * Basic worker thread pool for io_uring
   4  *
   5  * Copyright (C) 2019 Jens Axboe
   6  *
   7  */
   8 #include <linux/kernel.h>
   9 #include <linux/init.h>
  10 #include <linux/errno.h>
  11 #include <linux/sched/signal.h>
  12 #include <linux/mm.h>
  13 #include <linux/sched/mm.h>
  14 #include <linux/percpu.h>
  15 #include <linux/slab.h>
  16 #include <linux/kthread.h>
  17 #include <linux/rculist_nulls.h>
  18 #include <linux/fs_struct.h>
  19 #include <linux/task_work.h>
  20 #include <linux/blk-cgroup.h>
  21 #include <linux/audit.h>
  22 #include <linux/cpu.h>
  23
  24 #include "../kernel/sched/sched.h"
  25 #include "io-wq.h"
  26
  27 #define WORKER_IDLE_TIMEOUT     (5 * HZ)
  28
  29 enum {
  30         IO_WORKER_F_UP          = 1,    /* up and active */
  31         IO_WORKER_F_RUNNING     = 2,    /* account as running */
  32         IO_WORKER_F_FREE        = 4,    /* worker on free list */
  33         IO_WORKER_F_FIXED       = 8,    /* static idle worker */
  34         IO_WORKER_F_BOUND       = 16,   /* is doing bounded work */
  35 };
  36
  37 enum {
  38         IO_WQ_BIT_EXIT          = 0,    /* wq exiting */
  39         IO_WQ_BIT_ERROR         = 1,    /* error on setup */
  40 };
  41
  42 enum {
  43         IO_WQE_FLAG_STALLED     = 1,    /* stalled on hash */
  44 };
  45
  46 /*
  47  * One for each thread in a wqe pool
  48  */
  49 struct io_worker {
  50         refcount_t ref;
  51         unsigned flags;
  52         struct hlist_nulls_node nulls_node;
  53         struct list_head all_list;
  54         struct task_struct *task;
  55         struct io_wqe *wqe;
  56
  57         struct io_wq_work *cur_work;
  58         spinlock_t lock;
  59
  60         struct rcu_head rcu;
  61         struct mm_struct *mm;
  62 #ifdef CONFIG_BLK_CGROUP
  63         struct cgroup_subsys_state *blkcg_css;
  64 #endif
  65         const struct cred *cur_creds;
  66         const struct cred *saved_creds;
  67         struct files_struct *restore_files;
  68         struct nsproxy *restore_nsproxy;
  69         struct fs_struct *restore_fs;
  70 };
  71
  72 #if BITS_PER_LONG == 64
  73 #define IO_WQ_HASH_ORDER        6
  74 #else
  75 #define IO_WQ_HASH_ORDER        5
  76 #endif
  77
  78 #define IO_WQ_NR_HASH_BUCKETS   (1u << IO_WQ_HASH_ORDER)
  79
  80 struct io_wqe_acct {
  81         unsigned nr_workers;
  82         unsigned max_workers;
  83         atomic_t nr_running;
  84 };
  85
  86 enum {
  87         IO_WQ_ACCT_BOUND,
  88         IO_WQ_ACCT_UNBOUND,
  89 };
  90
  91 /*
  92  * Per-node worker thread pool
  93  */
  94 struct io_wqe {
  95         struct {
  96                 raw_spinlock_t lock;
  97                 struct io_wq_work_list work_list;
  98                 unsigned long hash_map;
  99                 unsigned flags;
 100         } ____cacheline_aligned_in_smp;
 101
 102         int node;
 103         struct io_wqe_acct acct[2];
 104
 105         struct hlist_nulls_head free_list;
 106         struct list_head all_list;
 107
 108         struct io_wq *wq;
 109         struct io_wq_work *hash_tail[IO_WQ_NR_HASH_BUCKETS];
 110 };
 111
 112 /*
 113  * Per io_wq state
 114   */
 115 struct io_wq {
 116         struct io_wqe **wqes;
 117         unsigned long state;
 118
 119         free_work_fn *free_work;
 120         io_wq_work_fn *do_work;
 121
 122         struct task_struct *manager;
 123         struct user_struct *user;
 124         refcount_t refs;
 125         struct completion done;
 126
 127         struct hlist_node cpuhp_node;
 128
 129         refcount_t use_refs;
 130 };
 131
 132 static enum cpuhp_state io_wq_online;
 133
 134 static bool io_worker_get(struct io_worker *worker)
 135 {
 136         return refcount_inc_not_zero(&worker->ref);
 137 }
 138
 139 static void io_worker_release(struct io_worker *worker)
 140 {
 141         if (refcount_dec_and_test(&worker->ref))
 142                 wake_up_process(worker->task);
 143 }
 144
 145 /*
 146  * Note: drops the wqe->lock if returning true! The caller must re-acquire
 147  * the lock in that case. Some callers need to restart handling if this
 148  * happens, so we can't just re-acquire the lock on behalf of the caller.
 149  */
 150 static bool __io_worker_unuse(struct io_wqe *wqe, struct io_worker *worker)
 151 {
 152         bool dropped_lock = false;
 153
 154         if (worker->saved_creds) {
 155                 revert_creds(worker->saved_creds);
 156                 worker->cur_creds = worker->saved_creds = NULL;
 157         }
 158
 159         if (current->files != worker->restore_files) {
 160                 __acquire(&wqe->lock);
 161                 raw_spin_unlock_irq(&wqe->lock);
 162                 dropped_lock = true;
 163
 164                 task_lock(current);
 165                 current->files = worker->restore_files;
 166                 current->nsproxy = worker->restore_nsproxy;
 167                 task_unlock(current);
 168         }
 169
 170         if (current->fs != worker->restore_fs)
 171                 current->fs = worker->restore_fs;
 172
 173         /*
 174          * If we have an active mm, we need to drop the wq lock before unusing
 175          * it. If we do, return true and let the caller retry the idle loop.
 176          */
 177         if (worker->mm) {
 178                 if (!dropped_lock) {
 179                         __acquire(&wqe->lock);
 180                         raw_spin_unlock_irq(&wqe->lock);
 181                         dropped_lock = true;
 182                 }
 183                 __set_current_state(TASK_RUNNING);
 184                 kthread_unuse_mm(worker->mm);
 185                 mmput(worker->mm);
 186                 worker->mm = NULL;
 187         }
 188
 189 #ifdef CONFIG_BLK_CGROUP
 190         if (worker->blkcg_css) {
 191                 kthread_associate_blkcg(NULL);
 192                 worker->blkcg_css = NULL;
 193         }
 194 #endif
 195         if (current->signal->rlim[RLIMIT_FSIZE].rlim_cur != RLIM_INFINITY)
 196                 current->signal->rlim[RLIMIT_FSIZE].rlim_cur = RLIM_INFINITY;
 197         return dropped_lock;
 198 }
 199
 200 static inline struct io_wqe_acct *io_work_get_acct(struct io_wqe *wqe,
 201                                                    struct io_wq_work *work)
 202 {
 203         if (work->flags & IO_WQ_WORK_UNBOUND)
 204                 return &wqe->acct[IO_WQ_ACCT_UNBOUND];
 205
 206         return &wqe->acct[IO_WQ_ACCT_BOUND];
 207 }
 208
 209 static inline struct io_wqe_acct *io_wqe_get_acct(struct io_wqe *wqe,
 210                                                   struct io_worker *worker)
 211 {
 212         if (worker->flags & IO_WORKER_F_BOUND)
 213                 return &wqe->acct[IO_WQ_ACCT_BOUND];
 214
 215         return &wqe->acct[IO_WQ_ACCT_UNBOUND];
 216 }
 217
 218 static void io_worker_exit(struct io_worker *worker)
 219 {
 220         struct io_wqe *wqe = worker->wqe;
 221         struct io_wqe_acct *acct = io_wqe_get_acct(wqe, worker);
 222
 223         /*
 224          * If we're not at zero, someone else is holding a brief reference
 225          * to the worker. Wait for that to go away.
 226          */
 227         set_current_state(TASK_INTERRUPTIBLE);
 228         if (!refcount_dec_and_test(&worker->ref))
 229                 schedule();
 230         __set_current_state(TASK_RUNNING);
 231
 232         preempt_disable();
 233         current->flags &= ~PF_IO_WORKER;
 234         if (worker->flags & IO_WORKER_F_RUNNING)
 235                 atomic_dec(&acct->nr_running);
 236         if (!(worker->flags & IO_WORKER_F_BOUND))
 237                 atomic_dec(&wqe->wq->user->processes);
 238         worker->flags = 0;
 239         preempt_enable();
 240
 241         raw_spin_lock_irq(&wqe->lock);
 242         hlist_nulls_del_rcu(&worker->nulls_node);
 243         list_del_rcu(&worker->all_list);
 244         if (__io_worker_unuse(wqe, worker)) {
 245                 __release(&wqe->lock);
 246                 raw_spin_lock_irq(&wqe->lock);
 247         }
 248         acct->nr_workers--;
 249         raw_spin_unlock_irq(&wqe->lock);
 250
 251         kfree_rcu(worker, rcu);
 252         if (refcount_dec_and_test(&wqe->wq->refs))
 253                 complete(&wqe->wq->done);
 254 }
 255
 256 static inline bool io_wqe_run_queue(struct io_wqe *wqe)
 257         __must_hold(wqe->lock)
 258 {
 259         if (!wq_list_empty(&wqe->work_list) &&
 260             !(wqe->flags & IO_WQE_FLAG_STALLED))
 261                 return true;
 262         return false;
 263 }
 264
 265 /*
 266  * Check head of free list for an available worker. If one isn't available,
 267  * caller must wake up the wq manager to create one.
 268  */
 269 static bool io_wqe_activate_free_worker(struct io_wqe *wqe)
 270         __must_hold(RCU)
 271 {
 272         struct hlist_nulls_node *n;
 273         struct io_worker *worker;
 274
 275         n = rcu_dereference(hlist_nulls_first_rcu(&wqe->free_list));
 276         if (is_a_nulls(n))
 277                 return false;
 278
 279         worker = hlist_nulls_entry(n, struct io_worker, nulls_node);
 280         if (io_worker_get(worker)) {
 281                 wake_up_process(worker->task);
 282                 io_worker_release(worker);
 283                 return true;
 284         }
 285
 286         return false;
 287 }
 288
 289 /*
 290  * We need a worker. If we find a free one, we're good. If not, and we're
 291  * below the max number of workers, wake up the manager to create one.
 292  */
 293 static void io_wqe_wake_worker(struct io_wqe *wqe, struct io_wqe_acct *acct)
 294 {
 295         bool ret;
 296
 297         /*
 298          * Most likely an attempt to queue unbounded work on an io_wq that
 299          * wasn't setup with any unbounded workers.
 300          */
 301         WARN_ON_ONCE(!acct->max_workers);
 302
 303         rcu_read_lock();
 304         ret = io_wqe_activate_free_worker(wqe);
 305         rcu_read_unlock();
 306
 307         if (!ret && acct->nr_workers < acct->max_workers)
 308                 wake_up_process(wqe->wq->manager);
 309 }
 310
 311 static void io_wqe_inc_running(struct io_wqe *wqe, struct io_worker *worker)
 312 {
 313         struct io_wqe_acct *acct = io_wqe_get_acct(wqe, worker);
 314
 315         atomic_inc(&acct->nr_running);
 316 }
 317
 318 static void io_wqe_dec_running(struct io_wqe *wqe, struct io_worker *worker)
 319         __must_hold(wqe->lock)
 320 {
 321         struct io_wqe_acct *acct = io_wqe_get_acct(wqe, worker);
 322
 323         if (atomic_dec_and_test(&acct->nr_running) && io_wqe_run_queue(wqe))
 324                 io_wqe_wake_worker(wqe, acct);
 325 }
 326
 327 static void io_worker_start(struct io_wqe *wqe, struct io_worker *worker)
 328 {
 329         allow_kernel_signal(SIGINT);
 330
 331         current->flags |= PF_IO_WORKER;
 332
 333         worker->flags |= (IO_WORKER_F_UP | IO_WORKER_F_RUNNING);
 334         worker->restore_files = current->files;
 335         worker->restore_nsproxy = current->nsproxy;
 336         worker->restore_fs = current->fs;
 337         io_wqe_inc_running(wqe, worker);
 338 }
 339
 340 /*
 341  * Worker will start processing some work. Move it to the busy list, if
 342  * it's currently on the freelist
 343  */
 344 static void __io_worker_busy(struct io_wqe *wqe, struct io_worker *worker,
 345                              struct io_wq_work *work)
 346         __must_hold(wqe->lock)
 347 {
 348         bool worker_bound, work_bound;
 349
 350         if (worker->flags & IO_WORKER_F_FREE) {
 351                 worker->flags &= ~IO_WORKER_F_FREE;
 352                 hlist_nulls_del_init_rcu(&worker->nulls_node);
 353         }
 354
 355         /*
 356          * If worker is moving from bound to unbound (or vice versa), then
 357          * ensure we update the running accounting.
 358          */
 359         worker_bound = (worker->flags & IO_WORKER_F_BOUND) != 0;
 360         work_bound = (work->flags & IO_WQ_WORK_UNBOUND) == 0;
 361         if (worker_bound != work_bound) {
 362                 io_wqe_dec_running(wqe, worker);
 363                 if (work_bound) {
 364                         worker->flags |= IO_WORKER_F_BOUND;
 365                         wqe->acct[IO_WQ_ACCT_UNBOUND].nr_workers--;
 366                         wqe->acct[IO_WQ_ACCT_BOUND].nr_workers++;
 367                         atomic_dec(&wqe->wq->user->processes);
 368                 } else {
 369                         worker->flags &= ~IO_WORKER_F_BOUND;
 370                         wqe->acct[IO_WQ_ACCT_UNBOUND].nr_workers++;
 371                         wqe->acct[IO_WQ_ACCT_BOUND].nr_workers--;
 372                         atomic_inc(&wqe->wq->user->processes);
 373                 }
 374                 io_wqe_inc_running(wqe, worker);
 375          }
 376 }
 377
 378 /*
 379  * No work, worker going to sleep. Move to freelist, and unuse mm if we
 380  * have one attached. Dropping the mm may potentially sleep, so we drop
 381  * the lock in that case and return success. Since the caller has to
 382  * retry the loop in that case (we changed task state), we don't regrab
 383  * the lock if we return success.
 384  */
 385 static bool __io_worker_idle(struct io_wqe *wqe, struct io_worker *worker)
 386         __must_hold(wqe->lock)
 387 {
 388         if (!(worker->flags & IO_WORKER_F_FREE)) {
 389                 worker->flags |= IO_WORKER_F_FREE;
 390                 hlist_nulls_add_head_rcu(&worker->nulls_node, &wqe->free_list);
 391         }
 392
 393         return __io_worker_unuse(wqe, worker);
 394 }
 395
 396 static inline unsigned int io_get_work_hash(struct io_wq_work *work)
 397 {
 398         return work->flags >> IO_WQ_HASH_SHIFT;
 399 }
 400
 401 static struct io_wq_work *io_get_next_work(struct io_wqe *wqe)
 402         __must_hold(wqe->lock)
 403 {
 404         struct io_wq_work_node *node, *prev;
 405         struct io_wq_work *work, *tail;
 406         unsigned int hash;
 407
 408         wq_list_for_each(node, prev, &wqe->work_list) {
 409                 work = container_of(node, struct io_wq_work, list);
 410
 411                 /* not hashed, can run anytime */
 412                 if (!io_wq_is_hashed(work)) {
 413                         wq_list_del(&wqe->work_list, node, prev);
 414                         return work;
 415                 }
 416
 417                 /* hashed, can run if not already running */
 418                 hash = io_get_work_hash(work);
 419                 if (!(wqe->hash_map & BIT(hash))) {
 420                         wqe->hash_map |= BIT(hash);
 421                         /* all items with this hash lie in [work, tail] */
 422                         tail = wqe->hash_tail[hash];
 423                         wqe->hash_tail[hash] = NULL;
 424                         wq_list_cut(&wqe->work_list, &tail->list, prev);
 425                         return work;
 426                 }
 427         }
 428
 429         return NULL;
 430 }
 431
 432 static void io_wq_switch_mm(struct io_worker *worker, struct io_wq_work *work)
 433 {
 434         if (worker->mm) {
 435                 kthread_unuse_mm(worker->mm);
 436                 mmput(worker->mm);
 437                 worker->mm = NULL;
 438         }
 439
 440         if (mmget_not_zero(work->identity->mm)) {
 441                 kthread_use_mm(work->identity->mm);
 442                 worker->mm = work->identity->mm;
 443                 return;
 444         }
 445
 446         /* failed grabbing mm, ensure work gets cancelled */
 447         work->flags |= IO_WQ_WORK_CANCEL;
 448 }
 449
 450 static inline void io_wq_switch_blkcg(struct io_worker *worker,
 451                                       struct io_wq_work *work)
 452 {
 453 #ifdef CONFIG_BLK_CGROUP
 454         if (!(work->flags & IO_WQ_WORK_BLKCG))
 455                 return;
 456         if (work->identity->blkcg_css != worker->blkcg_css) {
 457                 kthread_associate_blkcg(work->identity->blkcg_css);
 458                 worker->blkcg_css = work->identity->blkcg_css;
 459         }
 460 #endif
 461 }
 462
 463 static void io_wq_switch_creds(struct io_worker *worker,
 464                                struct io_wq_work *work)
 465 {
 466         const struct cred *old_creds = override_creds(work->identity->creds);
 467
 468         worker->cur_creds = work->identity->creds;
 469         if (worker->saved_creds)
 470                 put_cred(old_creds); /* creds set by previous switch */
 471         else
 472                 worker->saved_creds = old_creds;
 473 }
 474
 475 static void io_impersonate_work(struct io_worker *worker,
 476                                 struct io_wq_work *work)
 477 {
 478         if ((work->flags & IO_WQ_WORK_FILES) &&
 479             current->files != work->identity->files) {
 480                 task_lock(current);
 481                 current->files = work->identity->files;
 482                 current->nsproxy = work->identity->nsproxy;
 483                 task_unlock(current);
 484                 if (!work->identity->files) {
 485                         /* failed grabbing files, ensure work gets cancelled */
 486                         work->flags |= IO_WQ_WORK_CANCEL;
 487                 }
 488         }
 489         if ((work->flags & IO_WQ_WORK_FS) && current->fs != work->identity->fs)
 490                 current->fs = work->identity->fs;
 491         if ((work->flags & IO_WQ_WORK_MM) && work->identity->mm != worker->mm)
 492                 io_wq_switch_mm(worker, work);
 493         if ((work->flags & IO_WQ_WORK_CREDS) &&
 494             worker->cur_creds != work->identity->creds)
 495                 io_wq_switch_creds(worker, work);
 496         if (work->flags & IO_WQ_WORK_FSIZE)
 497                 current->signal->rlim[RLIMIT_FSIZE].rlim_cur = work->identity->fsize;
 498         else if (current->signal->rlim[RLIMIT_FSIZE].rlim_cur != RLIM_INFINITY)
 499                 current->signal->rlim[RLIMIT_FSIZE].rlim_cur = RLIM_INFINITY;
 500         io_wq_switch_blkcg(worker, work);
 501 #ifdef CONFIG_AUDIT
 502         current->loginuid = work->identity->loginuid;
 503         current->sessionid = work->identity->sessionid;
 504 #endif
 505 }
 506
 507 static void io_assign_current_work(struct io_worker *worker,
 508                                    struct io_wq_work *work)
 509 {
 510         if (work) {
 511                 /* flush pending signals before assigning new work */
 512                 if (signal_pending(current))
 513                         flush_signals(current);
 514                 cond_resched();
 515         }
 516
 517 #ifdef CONFIG_AUDIT
 518         current->loginuid = KUIDT_INIT(AUDIT_UID_UNSET);
 519         current->sessionid = AUDIT_SID_UNSET;
 520 #endif
 521
 522         spin_lock_irq(&worker->lock);
 523         worker->cur_work = work;
 524         spin_unlock_irq(&worker->lock);
 525 }
 526
 527 static void io_wqe_enqueue(struct io_wqe *wqe, struct io_wq_work *work);
 528
 529 static void io_worker_handle_work(struct io_worker *worker)
 530         __releases(wqe->lock)
 531 {
 532         struct io_wqe *wqe = worker->wqe;
 533         struct io_wq *wq = wqe->wq;
 534
 535         do {
 536                 struct io_wq_work *work;
 537 get_next:
 538                 /*
 539                  * If we got some work, mark us as busy. If we didn't, but
 540                  * the list isn't empty, it means we stalled on hashed work.
 541                  * Mark us stalled so we don't keep looking for work when we
 542                  * can't make progress, any work completion or insertion will
 543                  * clear the stalled flag.
 544                  */
 545                 work = io_get_next_work(wqe);
 546                 if (work)
 547                         __io_worker_busy(wqe, worker, work);
 548                 else if (!wq_list_empty(&wqe->work_list))
 549                         wqe->flags |= IO_WQE_FLAG_STALLED;
 550
 551                 raw_spin_unlock_irq(&wqe->lock);
 552                 if (!work)
 553                         break;
 554                 io_assign_current_work(worker, work);
 555
 556                 /* handle a whole dependent link */
 557                 do {
 558                         struct io_wq_work *old_work, *next_hashed, *linked;
 559                         unsigned int hash = io_get_work_hash(work);
 560
 561                         next_hashed = wq_next_work(work);
 562                         io_impersonate_work(worker, work);
 563
 564                         old_work = work;
 565                         linked = wq->do_work(work);
 566
 567                         work = next_hashed;
 568                         if (!work && linked && !io_wq_is_hashed(linked)) {
 569                                 work = linked;
 570                                 linked = NULL;
 571                         }
 572                         io_assign_current_work(worker, work);
 573                         wq->free_work(old_work);
 574
 575                         if (linked)
 576                                 io_wqe_enqueue(wqe, linked);
 577
 578                         if (hash != -1U && !next_hashed) {
 579                                 raw_spin_lock_irq(&wqe->lock);
 580                                 wqe->hash_map &= ~BIT_ULL(hash);
 581                                 wqe->flags &= ~IO_WQE_FLAG_STALLED;
 582                                 /* skip unnecessary unlock-lock wqe->lock */
 583                                 if (!work)
 584                                         goto get_next;
 585                                 raw_spin_unlock_irq(&wqe->lock);
 586                         }
 587                 } while (work);
 588
 589                 raw_spin_lock_irq(&wqe->lock);
 590         } while (1);
 591 }
 592
 593 static int io_wqe_worker(void *data)
 594 {
 595         struct io_worker *worker = data;
 596         struct io_wqe *wqe = worker->wqe;
 597         struct io_wq *wq = wqe->wq;
 598
 599         io_worker_start(wqe, worker);
 600
 601         while (!test_bit(IO_WQ_BIT_EXIT, &wq->state)) {
 602                 set_current_state(TASK_INTERRUPTIBLE);
 603 loop:
 604                 raw_spin_lock_irq(&wqe->lock);
 605                 if (io_wqe_run_queue(wqe)) {
 606                         __set_current_state(TASK_RUNNING);
 607                         io_worker_handle_work(worker);
 608                         goto loop;
 609                 }
 610                 /* drops the lock on success, retry */
 611                 if (__io_worker_idle(wqe, worker)) {
 612                         __release(&wqe->lock);
 613                         goto loop;
 614                 }
 615                 raw_spin_unlock_irq(&wqe->lock);
 616                 if (signal_pending(current))
 617                         flush_signals(current);
 618                 if (schedule_timeout(WORKER_IDLE_TIMEOUT))
 619                         continue;
 620                 /* timed out, exit unless we're the fixed worker */
 621                 if (test_bit(IO_WQ_BIT_EXIT, &wq->state) ||
 622                     !(worker->flags & IO_WORKER_F_FIXED))
 623                         break;
 624         }
 625
 626         if (test_bit(IO_WQ_BIT_EXIT, &wq->state)) {
 627                 raw_spin_lock_irq(&wqe->lock);
 628                 if (!wq_list_empty(&wqe->work_list))
 629                         io_worker_handle_work(worker);
 630                 else
 631                         raw_spin_unlock_irq(&wqe->lock);
 632         }
 633
 634         io_worker_exit(worker);
 635         return 0;
 636 }
 637
 638 /*
 639  * Called when a worker is scheduled in. Mark us as currently running.
 640  */
 641 void io_wq_worker_running(struct task_struct *tsk)
 642 {
 643         struct io_worker *worker = kthread_data(tsk);
 644         struct io_wqe *wqe = worker->wqe;
 645
 646         if (!(worker->flags & IO_WORKER_F_UP))
 647                 return;
 648         if (worker->flags & IO_WORKER_F_RUNNING)
 649                 return;
 650         worker->flags |= IO_WORKER_F_RUNNING;
 651         io_wqe_inc_running(wqe, worker);
 652 }
 653
 654 /*
 655  * Called when worker is going to sleep. If there are no workers currently
 656  * running and we have work pending, wake up a free one or have the manager
 657  * set one up.
 658  */
 659 void io_wq_worker_sleeping(struct task_struct *tsk)
 660 {
 661         struct io_worker *worker = kthread_data(tsk);
 662         struct io_wqe *wqe = worker->wqe;
 663
 664         if (!(worker->flags & IO_WORKER_F_UP))
 665                 return;
 666         if (!(worker->flags & IO_WORKER_F_RUNNING))
 667                 return;
 668
 669         worker->flags &= ~IO_WORKER_F_RUNNING;
 670
 671         raw_spin_lock_irq(&wqe->lock);
 672         io_wqe_dec_running(wqe, worker);
 673         raw_spin_unlock_irq(&wqe->lock);
 674 }
 675
 676 static bool create_io_worker(struct io_wq *wq, struct io_wqe *wqe, int index)
 677 {
 678         struct io_wqe_acct *acct = &wqe->acct[index];
 679         struct io_worker *worker;
 680
 681         worker = kzalloc_node(sizeof(*worker), GFP_KERNEL, wqe->node);
 682         if (!worker)
 683                 return false;
 684
 685         refcount_set(&worker->ref, 1);
 686         worker->nulls_node.pprev = NULL;
 687         worker->wqe = wqe;
 688         spin_lock_init(&worker->lock);
 689
 690         worker->task = kthread_create_on_node(io_wqe_worker, worker, wqe->node,
 691                                 "io_wqe_worker-%d/%d", index, wqe->node);
 692         if (IS_ERR(worker->task)) {
 693                 kfree(worker);
 694                 return false;
 695         }
 696         kthread_bind_mask(worker->task, cpumask_of_node(wqe->node));
 697
 698         raw_spin_lock_irq(&wqe->lock);
 699         hlist_nulls_add_head_rcu(&worker->nulls_node, &wqe->free_list);
 700         list_add_tail_rcu(&worker->all_list, &wqe->all_list);
 701         worker->flags |= IO_WORKER_F_FREE;
 702         if (index == IO_WQ_ACCT_BOUND)
 703                 worker->flags |= IO_WORKER_F_BOUND;
 704         if (!acct->nr_workers && (worker->flags & IO_WORKER_F_BOUND))
 705                 worker->flags |= IO_WORKER_F_FIXED;
 706         acct->nr_workers++;
 707         raw_spin_unlock_irq(&wqe->lock);
 708
 709         if (index == IO_WQ_ACCT_UNBOUND)
 710                 atomic_inc(&wq->user->processes);
 711
 712         refcount_inc(&wq->refs);
 713         wake_up_process(worker->task);
 714         return true;
 715 }
 716
 717 static inline bool io_wqe_need_worker(struct io_wqe *wqe, int index)
 718         __must_hold(wqe->lock)
 719 {
 720         struct io_wqe_acct *acct = &wqe->acct[index];
 721
 722         /* if we have available workers or no work, no need */
 723         if (!hlist_nulls_empty(&wqe->free_list) || !io_wqe_run_queue(wqe))
 724                 return false;
 725         return acct->nr_workers < acct->max_workers;
 726 }
 727
 728 /*
 729  * Iterate the passed in list and call the specific function for each
 730  * worker that isn't exiting
 731  */
 732 static bool io_wq_for_each_worker(struct io_wqe *wqe,
 733                                   bool (*func)(struct io_worker *, void *),
 734                                   void *data)
 735 {
 736         struct io_worker *worker;
 737         bool ret = false;
 738
 739         list_for_each_entry_rcu(worker, &wqe->all_list, all_list) {
 740                 if (io_worker_get(worker)) {
 741                         /* no task if node is/was offline */
 742                         if (worker->task)
 743                                 ret = func(worker, data);
 744                         io_worker_release(worker);
 745                         if (ret)
 746                                 break;
 747                 }
 748         }
 749
 750         return ret;
 751 }
 752
 753 static bool io_wq_worker_wake(struct io_worker *worker, void *data)
 754 {
 755         wake_up_process(worker->task);
 756         return false;
 757 }
 758
 759 /*
 760  * Manager thread. Tasked with creating new workers, if we need them.
 761  */
 762 static int io_wq_manager(void *data)
 763 {
 764         struct io_wq *wq = data;
 765         int node;
 766
 767         /* create fixed workers */
 768         refcount_set(&wq->refs, 1);
 769         for_each_node(node) {
 770                 if (!node_online(node))
 771                         continue;
 772                 if (create_io_worker(wq, wq->wqes[node], IO_WQ_ACCT_BOUND))
 773                         continue;
 774                 set_bit(IO_WQ_BIT_ERROR, &wq->state);
 775                 set_bit(IO_WQ_BIT_EXIT, &wq->state);
 776                 goto out;
 777         }
 778
 779         complete(&wq->done);
 780
 781         while (!kthread_should_stop()) {
 782                 if (current->task_works)
 783                         task_work_run();
 784
 785                 for_each_node(node) {
 786                         struct io_wqe *wqe = wq->wqes[node];
 787                         bool fork_worker[2] = { false, false };
 788
 789                         if (!node_online(node))
 790                                 continue;
 791
 792                         raw_spin_lock_irq(&wqe->lock);
 793                         if (io_wqe_need_worker(wqe, IO_WQ_ACCT_BOUND))
 794                                 fork_worker[IO_WQ_ACCT_BOUND] = true;
 795                         if (io_wqe_need_worker(wqe, IO_WQ_ACCT_UNBOUND))
 796                                 fork_worker[IO_WQ_ACCT_UNBOUND] = true;
 797                         raw_spin_unlock_irq(&wqe->lock);
 798                         if (fork_worker[IO_WQ_ACCT_BOUND])
 799                                 create_io_worker(wq, wqe, IO_WQ_ACCT_BOUND);
 800                         if (fork_worker[IO_WQ_ACCT_UNBOUND])
 801                                 create_io_worker(wq, wqe, IO_WQ_ACCT_UNBOUND);
 802                 }
 803                 set_current_state(TASK_INTERRUPTIBLE);
 804                 schedule_timeout(HZ);
 805         }
 806
 807         if (current->task_works)
 808                 task_work_run();
 809
 810 out:
 811         if (refcount_dec_and_test(&wq->refs)) {
 812                 complete(&wq->done);
 813                 return 0;
 814         }
 815         /* if ERROR is set and we get here, we have workers to wake */
 816         if (test_bit(IO_WQ_BIT_ERROR, &wq->state)) {
 817                 rcu_read_lock();
 818                 for_each_node(node)
 819                         io_wq_for_each_worker(wq->wqes[node], io_wq_worker_wake, NULL);
 820                 rcu_read_unlock();
 821         }
 822         return 0;
 823 }
 824
 825 static bool io_wq_can_queue(struct io_wqe *wqe, struct io_wqe_acct *acct,
 826                             struct io_wq_work *work)
 827 {
 828         bool free_worker;
 829
 830         if (!(work->flags & IO_WQ_WORK_UNBOUND))
 831                 return true;
 832         if (atomic_read(&acct->nr_running))
 833                 return true;
 834
 835         rcu_read_lock();
 836         free_worker = !hlist_nulls_empty(&wqe->free_list);
 837         rcu_read_unlock();
 838         if (free_worker)
 839                 return true;
 840
 841         if (atomic_read(&wqe->wq->user->processes) >= acct->max_workers &&
 842             !(capable(CAP_SYS_RESOURCE) || capable(CAP_SYS_ADMIN)))
 843                 return false;
 844
 845         return true;
 846 }
 847
 848 static void io_run_cancel(struct io_wq_work *work, struct io_wqe *wqe)
 849 {
 850         struct io_wq *wq = wqe->wq;
 851
 852         do {
 853                 struct io_wq_work *old_work = work;
 854
 855                 work->flags |= IO_WQ_WORK_CANCEL;
 856                 work = wq->do_work(work);
 857                 wq->free_work(old_work);
 858         } while (work);
 859 }
 860
 861 static void io_wqe_insert_work(struct io_wqe *wqe, struct io_wq_work *work)
 862 {
 863         unsigned int hash;
 864         struct io_wq_work *tail;
 865
 866         if (!io_wq_is_hashed(work)) {
 867 append:
 868                 wq_list_add_tail(&work->list, &wqe->work_list);
 869                 return;
 870         }
 871
 872         hash = io_get_work_hash(work);
 873         tail = wqe->hash_tail[hash];
 874         wqe->hash_tail[hash] = work;
 875         if (!tail)
 876                 goto append;
 877
 878         wq_list_add_after(&work->list, &tail->list, &wqe->work_list);
 879 }
 880
 881 static void io_wqe_enqueue(struct io_wqe *wqe, struct io_wq_work *work)
 882 {
 883         struct io_wqe_acct *acct = io_work_get_acct(wqe, work);
 884         int work_flags;
 885         unsigned long flags;
 886
 887         /*
 888          * Do early check to see if we need a new unbound worker, and if we do,
 889          * if we're allowed to do so. This isn't 100% accurate as there's a
 890          * gap between this check and incrementing the value, but that's OK.
 891          * It's close enough to not be an issue, fork() has the same delay.
 892          */
 893         if (unlikely(!io_wq_can_queue(wqe, acct, work))) {
 894                 io_run_cancel(work, wqe);
 895                 return;
 896         }
 897
 898         work_flags = work->flags;
 899         raw_spin_lock_irqsave(&wqe->lock, flags);
 900         io_wqe_insert_work(wqe, work);
 901         wqe->flags &= ~IO_WQE_FLAG_STALLED;
 902         raw_spin_unlock_irqrestore(&wqe->lock, flags);
 903
 904         if ((work_flags & IO_WQ_WORK_CONCURRENT) ||
 905             !atomic_read(&acct->nr_running))
 906                 io_wqe_wake_worker(wqe, acct);
 907 }
 908
 909 void io_wq_enqueue(struct io_wq *wq, struct io_wq_work *work)
 910 {
 911         struct io_wqe *wqe = wq->wqes[numa_node_id()];
 912
 913         io_wqe_enqueue(wqe, work);
 914 }
 915
 916 /*
 917  * Work items that hash to the same value will not be done in parallel.
 918  * Used to limit concurrent writes, generally hashed by inode.
 919  */
 920 void io_wq_hash_work(struct io_wq_work *work, void *val)
 921 {
 922         unsigned int bit;
 923
 924         bit = hash_ptr(val, IO_WQ_HASH_ORDER);
 925         work->flags |= (IO_WQ_WORK_HASHED | (bit << IO_WQ_HASH_SHIFT));
 926 }
 927
 928 struct io_cb_cancel_data {
 929         work_cancel_fn *fn;
 930         void *data;
 931         int nr_running;
 932         int nr_pending;
 933         bool cancel_all;
 934 };
 935
 936 static bool io_wq_worker_cancel(struct io_worker *worker, void *data)
 937 {
 938         struct io_cb_cancel_data *match = data;
 939         unsigned long flags;
 940
 941         /*
 942          * Hold the lock to avoid ->cur_work going out of scope, caller
 943          * may dereference the passed in work.
 944          */
 945         spin_lock_irqsave(&worker->lock, flags);
 946         if (worker->cur_work &&
 947             !(worker->cur_work->flags & IO_WQ_WORK_NO_CANCEL) &&
 948             match->fn(worker->cur_work, match->data)) {
 949                 send_sig(SIGINT, worker->task, 1);
 950                 match->nr_running++;
 951         }
 952         spin_unlock_irqrestore(&worker->lock, flags);
 953
 954         return match->nr_running && !match->cancel_all;
 955 }
 956
 957 static inline void io_wqe_remove_pending(struct io_wqe *wqe,
 958                                          struct io_wq_work *work,
 959                                          struct io_wq_work_node *prev)
 960 {
 961         unsigned int hash = io_get_work_hash(work);
 962         struct io_wq_work *prev_work = NULL;
 963
 964         if (io_wq_is_hashed(work) && work == wqe->hash_tail[hash]) {
 965                 if (prev)
 966                         prev_work = container_of(prev, struct io_wq_work, list);
 967                 if (prev_work && io_get_work_hash(prev_work) == hash)
 968                         wqe->hash_tail[hash] = prev_work;
 969                 else
 970                         wqe->hash_tail[hash] = NULL;
 971         }
 972         wq_list_del(&wqe->work_list, &work->list, prev);
 973 }
 974
 975 static void io_wqe_cancel_pending_work(struct io_wqe *wqe,
 976                                        struct io_cb_cancel_data *match)
 977 {
 978         struct io_wq_work_node *node, *prev;
 979         struct io_wq_work *work;
 980         unsigned long flags;
 981
 982 retry:
 983         raw_spin_lock_irqsave(&wqe->lock, flags);
 984         wq_list_for_each(node, prev, &wqe->work_list) {
 985                 work = container_of(node, struct io_wq_work, list);
 986                 if (!match->fn(work, match->data))
 987                         continue;
 988                 io_wqe_remove_pending(wqe, work, prev);
 989                 raw_spin_unlock_irqrestore(&wqe->lock, flags);
 990                 io_run_cancel(work, wqe);
 991                 match->nr_pending++;
 992                 if (!match->cancel_all)
 993                         return;
 994
 995                 /* not safe to continue after unlock */
 996                 goto retry;
 997         }
 998         raw_spin_unlock_irqrestore(&wqe->lock, flags);
 999 }
1000
1001 static void io_wqe_cancel_running_work(struct io_wqe *wqe,
1002                                        struct io_cb_cancel_data *match)
1003 {
1004         rcu_read_lock();
1005         io_wq_for_each_worker(wqe, io_wq_worker_cancel, match);
1006         rcu_read_unlock();
1007 }
1008
1009 enum io_wq_cancel io_wq_cancel_cb(struct io_wq *wq, work_cancel_fn *cancel,
1010                                   void *data, bool cancel_all)
1011 {
1012         struct io_cb_cancel_data match = {
1013                 .fn             = cancel,
1014                 .data           = data,
1015                 .cancel_all     = cancel_all,
1016         };
1017         int node;
1018
1019         /*
1020          * First check pending list, if we're lucky we can just remove it
1021          * from there. CANCEL_OK means that the work is returned as-new,
1022          * no completion will be posted for it.
1023          */
1024         for_each_node(node) {
1025                 struct io_wqe *wqe = wq->wqes[node];
1026
1027                 io_wqe_cancel_pending_work(wqe, &match);
1028                 if (match.nr_pending && !match.cancel_all)
1029                         return IO_WQ_CANCEL_OK;
1030         }
1031
1032         /*
1033          * Now check if a free (going busy) or busy worker has the work
1034          * currently running. If we find it there, we'll return CANCEL_RUNNING
1035          * as an indication that we attempt to signal cancellation. The
1036          * completion will run normally in this case.
1037          */
1038         for_each_node(node) {
1039                 struct io_wqe *wqe = wq->wqes[node];
1040
1041                 io_wqe_cancel_running_work(wqe, &match);
1042                 if (match.nr_running && !match.cancel_all)
1043                         return IO_WQ_CANCEL_RUNNING;
1044         }
1045
1046         if (match.nr_running)
1047                 return IO_WQ_CANCEL_RUNNING;
1048         if (match.nr_pending)
1049                 return IO_WQ_CANCEL_OK;
1050         return IO_WQ_CANCEL_NOTFOUND;
1051 }
1052
1053 struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data)
1054 {
1055         int ret = -ENOMEM, node;
1056         struct io_wq *wq;
1057
1058         if (WARN_ON_ONCE(!data->free_work || !data->do_work))
1059                 return ERR_PTR(-EINVAL);
1060
1061         wq = kzalloc(sizeof(*wq), GFP_KERNEL);
1062         if (!wq)
1063                 return ERR_PTR(-ENOMEM);
1064
1065         wq->wqes = kcalloc(nr_node_ids, sizeof(struct io_wqe *), GFP_KERNEL);
1066         if (!wq->wqes)
1067                 goto err_wq;
1068
1069         ret = cpuhp_state_add_instance_nocalls(io_wq_online, &wq->cpuhp_node);
1070         if (ret)
1071                 goto err_wqes;
1072
1073         wq->free_work = data->free_work;
1074         wq->do_work = data->do_work;
1075
1076         /* caller must already hold a reference to this */
1077         wq->user = data->user;
1078
1079         ret = -ENOMEM;
1080         for_each_node(node) {
1081                 struct io_wqe *wqe;
1082                 int alloc_node = node;
1083
1084                 if (!node_online(alloc_node))
1085                         alloc_node = NUMA_NO_NODE;
1086                 wqe = kzalloc_node(sizeof(struct io_wqe), GFP_KERNEL, alloc_node);
1087                 if (!wqe)
1088                         goto err;
1089                 wq->wqes[node] = wqe;
1090                 wqe->node = alloc_node;
1091                 wqe->acct[IO_WQ_ACCT_BOUND].max_workers = bounded;
1092                 atomic_set(&wqe->acct[IO_WQ_ACCT_BOUND].nr_running, 0);
1093                 if (wq->user) {
1094                         wqe->acct[IO_WQ_ACCT_UNBOUND].max_workers =
1095                                         task_rlimit(current, RLIMIT_NPROC);
1096                 }
1097                 atomic_set(&wqe->acct[IO_WQ_ACCT_UNBOUND].nr_running, 0);
1098                 wqe->wq = wq;
1099                 raw_spin_lock_init(&wqe->lock);
1100                 INIT_WQ_LIST(&wqe->work_list);
1101                 INIT_HLIST_NULLS_HEAD(&wqe->free_list, 0);
1102                 INIT_LIST_HEAD(&wqe->all_list);
1103         }
1104
1105         init_completion(&wq->done);
1106
1107         wq->manager = kthread_create(io_wq_manager, wq, "io_wq_manager");
1108         if (!IS_ERR(wq->manager)) {
1109                 wake_up_process(wq->manager);
1110                 wait_for_completion(&wq->done);
1111                 if (test_bit(IO_WQ_BIT_ERROR, &wq->state)) {
1112                         ret = -ENOMEM;
1113                         goto err;
1114                 }
1115                 refcount_set(&wq->use_refs, 1);
1116                 reinit_completion(&wq->done);
1117                 return wq;
1118         }
1119
1120         ret = PTR_ERR(wq->manager);
1121         complete(&wq->done);
1122 err:
1123         cpuhp_state_remove_instance_nocalls(io_wq_online, &wq->cpuhp_node);
1124         for_each_node(node)
1125                 kfree(wq->wqes[node]);
1126 err_wqes:
1127         kfree(wq->wqes);
1128 err_wq:
1129         kfree(wq);
1130         return ERR_PTR(ret);
1131 }
1132
1133 bool io_wq_get(struct io_wq *wq, struct io_wq_data *data)
1134 {
1135         if (data->free_work != wq->free_work || data->do_work != wq->do_work)
1136                 return false;
1137
1138         return refcount_inc_not_zero(&wq->use_refs);
1139 }
1140
1141 static void __io_wq_destroy(struct io_wq *wq)
1142 {
1143         int node;
1144
1145         cpuhp_state_remove_instance_nocalls(io_wq_online, &wq->cpuhp_node);
1146
1147         set_bit(IO_WQ_BIT_EXIT, &wq->state);
1148         if (wq->manager)
1149                 kthread_stop(wq->manager);
1150
1151         rcu_read_lock();
1152         for_each_node(node)
1153                 io_wq_for_each_worker(wq->wqes[node], io_wq_worker_wake, NULL);
1154         rcu_read_unlock();
1155
1156         wait_for_completion(&wq->done);
1157
1158         for_each_node(node)
1159                 kfree(wq->wqes[node]);
1160         kfree(wq->wqes);
1161         kfree(wq);
1162 }
1163
1164 void io_wq_destroy(struct io_wq *wq)
1165 {
1166         if (refcount_dec_and_test(&wq->use_refs))
1167                 __io_wq_destroy(wq);
1168 }
1169
1170 struct task_struct *io_wq_get_task(struct io_wq *wq)
1171 {
1172         return wq->manager;
1173 }
1174
1175 static bool io_wq_worker_affinity(struct io_worker *worker, void *data)
1176 {
1177         struct task_struct *task = worker->task;
1178         struct rq_flags rf;
1179         struct rq *rq;
1180
1181         rq = task_rq_lock(task, &rf);
1182         do_set_cpus_allowed(task, cpumask_of_node(worker->wqe->node));
1183         task->flags |= PF_NO_SETAFFINITY;
1184         task_rq_unlock(rq, task, &rf);
1185         return false;
1186 }
1187
1188 static int io_wq_cpu_online(unsigned int cpu, struct hlist_node *node)
1189 {
1190         struct io_wq *wq = hlist_entry_safe(node, struct io_wq, cpuhp_node);
1191         int i;
1192
1193         rcu_read_lock();
1194         for_each_node(i)
1195                 io_wq_for_each_worker(wq->wqes[i], io_wq_worker_affinity, NULL);
1196         rcu_read_unlock();
1197         return 0;
1198 }
1199
1200 static __init int io_wq_init(void)
1201 {
1202         int ret;
1203
1204         ret = cpuhp_setup_state_multi(CPUHP_AP_ONLINE_DYN, "io-wq/online",
1205                                         io_wq_cpu_online, NULL);
1206         if (ret < 0)
1207                 return ret;
1208         io_wq_online = ret;
1209         return 0;
1210 }
1211 subsys_initcall(io_wq_init);