Commit | Line | Data |
---|---|---|
771b53d0 JA |
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> | |
771b53d0 JA |
12 | #include <linux/percpu.h> |
13 | #include <linux/slab.h> | |
771b53d0 | 14 | #include <linux/rculist_nulls.h> |
43c01fbe | 15 | #include <linux/cpu.h> |
3bfe6106 | 16 | #include <linux/tracehook.h> |
dd47c104 | 17 | #include <uapi/linux/io_uring.h> |
771b53d0 JA |
18 | |
19 | #include "io-wq.h" | |
20 | ||
21 | #define WORKER_IDLE_TIMEOUT (5 * HZ) | |
22 | ||
23 | enum { | |
24 | IO_WORKER_F_UP = 1, /* up and active */ | |
25 | IO_WORKER_F_RUNNING = 2, /* account as running */ | |
26 | IO_WORKER_F_FREE = 4, /* worker on free list */ | |
05c5f4ee | 27 | IO_WORKER_F_BOUND = 8, /* is doing bounded work */ |
771b53d0 JA |
28 | }; |
29 | ||
30 | enum { | |
31 | IO_WQ_BIT_EXIT = 0, /* wq exiting */ | |
771b53d0 JA |
32 | }; |
33 | ||
34 | enum { | |
f95dc207 | 35 | IO_ACCT_STALLED_BIT = 0, /* stalled on hash */ |
771b53d0 JA |
36 | }; |
37 | ||
38 | /* | |
39 | * One for each thread in a wqe pool | |
40 | */ | |
41 | struct io_worker { | |
42 | refcount_t ref; | |
43 | unsigned flags; | |
44 | struct hlist_nulls_node nulls_node; | |
e61df66c | 45 | struct list_head all_list; |
771b53d0 | 46 | struct task_struct *task; |
771b53d0 | 47 | struct io_wqe *wqe; |
36c2f922 | 48 | |
771b53d0 | 49 | struct io_wq_work *cur_work; |
36c2f922 | 50 | spinlock_t lock; |
771b53d0 | 51 | |
eb2de941 JA |
52 | struct completion ref_done; |
53 | ||
d3e9f732 JA |
54 | unsigned long create_state; |
55 | struct callback_head create_work; | |
56 | int create_index; | |
57 | ||
3146cba9 JA |
58 | union { |
59 | struct rcu_head rcu; | |
60 | struct work_struct work; | |
61 | }; | |
771b53d0 JA |
62 | }; |
63 | ||
771b53d0 JA |
64 | #if BITS_PER_LONG == 64 |
65 | #define IO_WQ_HASH_ORDER 6 | |
66 | #else | |
67 | #define IO_WQ_HASH_ORDER 5 | |
68 | #endif | |
69 | ||
86f3cd1b PB |
70 | #define IO_WQ_NR_HASH_BUCKETS (1u << IO_WQ_HASH_ORDER) |
71 | ||
c5def4ab JA |
72 | struct io_wqe_acct { |
73 | unsigned nr_workers; | |
74 | unsigned max_workers; | |
685fe7fe | 75 | int index; |
c5def4ab | 76 | atomic_t nr_running; |
f95dc207 JA |
77 | struct io_wq_work_list work_list; |
78 | unsigned long flags; | |
c5def4ab JA |
79 | }; |
80 | ||
81 | enum { | |
82 | IO_WQ_ACCT_BOUND, | |
83 | IO_WQ_ACCT_UNBOUND, | |
f95dc207 | 84 | IO_WQ_ACCT_NR, |
c5def4ab JA |
85 | }; |
86 | ||
771b53d0 JA |
87 | /* |
88 | * Per-node worker thread pool | |
89 | */ | |
90 | struct io_wqe { | |
f95dc207 JA |
91 | raw_spinlock_t lock; |
92 | struct io_wqe_acct acct[2]; | |
771b53d0 JA |
93 | |
94 | int node; | |
771b53d0 | 95 | |
021d1cdd | 96 | struct hlist_nulls_head free_list; |
e61df66c | 97 | struct list_head all_list; |
771b53d0 | 98 | |
e941894e JA |
99 | struct wait_queue_entry wait; |
100 | ||
771b53d0 | 101 | struct io_wq *wq; |
86f3cd1b | 102 | struct io_wq_work *hash_tail[IO_WQ_NR_HASH_BUCKETS]; |
0e03496d JA |
103 | |
104 | cpumask_var_t cpu_mask; | |
771b53d0 JA |
105 | }; |
106 | ||
107 | /* | |
108 | * Per io_wq state | |
109 | */ | |
110 | struct io_wq { | |
771b53d0 | 111 | unsigned long state; |
771b53d0 | 112 | |
e9fd9396 | 113 | free_work_fn *free_work; |
f5fa38c5 | 114 | io_wq_work_fn *do_work; |
7d723065 | 115 | |
e941894e JA |
116 | struct io_wq_hash *hash; |
117 | ||
fb3a1f6c JA |
118 | atomic_t worker_refs; |
119 | struct completion worker_done; | |
120 | ||
43c01fbe | 121 | struct hlist_node cpuhp_node; |
3bfe6106 | 122 | |
685fe7fe | 123 | struct task_struct *task; |
c7f405d6 PB |
124 | |
125 | struct io_wqe *wqes[]; | |
771b53d0 JA |
126 | }; |
127 | ||
43c01fbe JA |
128 | static enum cpuhp_state io_wq_online; |
129 | ||
f0127254 JA |
130 | struct io_cb_cancel_data { |
131 | work_cancel_fn *fn; | |
132 | void *data; | |
133 | int nr_running; | |
134 | int nr_pending; | |
135 | bool cancel_all; | |
136 | }; | |
137 | ||
3146cba9 | 138 | static bool create_io_worker(struct io_wq *wq, struct io_wqe *wqe, int index); |
83d6c393 | 139 | static void io_wqe_dec_running(struct io_worker *worker); |
3146cba9 JA |
140 | static bool io_acct_cancel_pending_work(struct io_wqe *wqe, |
141 | struct io_wqe_acct *acct, | |
142 | struct io_cb_cancel_data *match); | |
f0127254 | 143 | |
771b53d0 JA |
144 | static bool io_worker_get(struct io_worker *worker) |
145 | { | |
146 | return refcount_inc_not_zero(&worker->ref); | |
147 | } | |
148 | ||
149 | static void io_worker_release(struct io_worker *worker) | |
150 | { | |
151 | if (refcount_dec_and_test(&worker->ref)) | |
eb2de941 | 152 | complete(&worker->ref_done); |
771b53d0 JA |
153 | } |
154 | ||
8418f22a PB |
155 | static inline struct io_wqe_acct *io_get_acct(struct io_wqe *wqe, bool bound) |
156 | { | |
157 | return &wqe->acct[bound ? IO_WQ_ACCT_BOUND : IO_WQ_ACCT_UNBOUND]; | |
158 | } | |
159 | ||
c5def4ab JA |
160 | static inline struct io_wqe_acct *io_work_get_acct(struct io_wqe *wqe, |
161 | struct io_wq_work *work) | |
162 | { | |
8418f22a | 163 | return io_get_acct(wqe, !(work->flags & IO_WQ_WORK_UNBOUND)); |
c5def4ab JA |
164 | } |
165 | ||
958234d5 | 166 | static inline struct io_wqe_acct *io_wqe_get_acct(struct io_worker *worker) |
c5def4ab | 167 | { |
8418f22a | 168 | return io_get_acct(worker->wqe, worker->flags & IO_WORKER_F_BOUND); |
c5def4ab JA |
169 | } |
170 | ||
685fe7fe JA |
171 | static void io_worker_ref_put(struct io_wq *wq) |
172 | { | |
173 | if (atomic_dec_and_test(&wq->worker_refs)) | |
174 | complete(&wq->worker_done); | |
175 | } | |
176 | ||
771b53d0 JA |
177 | static void io_worker_exit(struct io_worker *worker) |
178 | { | |
179 | struct io_wqe *wqe = worker->wqe; | |
771b53d0 | 180 | |
eb2de941 JA |
181 | if (refcount_dec_and_test(&worker->ref)) |
182 | complete(&worker->ref_done); | |
183 | wait_for_completion(&worker->ref_done); | |
771b53d0 | 184 | |
a9a4aa9f | 185 | raw_spin_lock(&wqe->lock); |
83d6c393 | 186 | if (worker->flags & IO_WORKER_F_FREE) |
bf1daa4b | 187 | hlist_nulls_del_rcu(&worker->nulls_node); |
e61df66c | 188 | list_del_rcu(&worker->all_list); |
83d6c393 JA |
189 | preempt_disable(); |
190 | io_wqe_dec_running(worker); | |
191 | worker->flags = 0; | |
192 | current->flags &= ~PF_IO_WORKER; | |
193 | preempt_enable(); | |
a9a4aa9f | 194 | raw_spin_unlock(&wqe->lock); |
771b53d0 | 195 | |
364b05fd | 196 | kfree_rcu(worker, rcu); |
685fe7fe | 197 | io_worker_ref_put(wqe->wq); |
46fe18b1 | 198 | do_exit(0); |
771b53d0 JA |
199 | } |
200 | ||
f95dc207 | 201 | static inline bool io_acct_run_queue(struct io_wqe_acct *acct) |
c5def4ab | 202 | { |
f95dc207 JA |
203 | if (!wq_list_empty(&acct->work_list) && |
204 | !test_bit(IO_ACCT_STALLED_BIT, &acct->flags)) | |
c5def4ab JA |
205 | return true; |
206 | return false; | |
207 | } | |
208 | ||
209 | /* | |
210 | * Check head of free list for an available worker. If one isn't available, | |
685fe7fe | 211 | * caller must create one. |
c5def4ab | 212 | */ |
f95dc207 JA |
213 | static bool io_wqe_activate_free_worker(struct io_wqe *wqe, |
214 | struct io_wqe_acct *acct) | |
c5def4ab JA |
215 | __must_hold(RCU) |
216 | { | |
217 | struct hlist_nulls_node *n; | |
218 | struct io_worker *worker; | |
219 | ||
83d6c393 JA |
220 | /* |
221 | * Iterate free_list and see if we can find an idle worker to | |
222 | * activate. If a given worker is on the free_list but in the process | |
223 | * of exiting, keep trying. | |
224 | */ | |
225 | hlist_nulls_for_each_entry_rcu(worker, n, &wqe->free_list, nulls_node) { | |
226 | if (!io_worker_get(worker)) | |
227 | continue; | |
f95dc207 JA |
228 | if (io_wqe_get_acct(worker) != acct) { |
229 | io_worker_release(worker); | |
230 | continue; | |
231 | } | |
83d6c393 JA |
232 | if (wake_up_process(worker->task)) { |
233 | io_worker_release(worker); | |
234 | return true; | |
235 | } | |
c5def4ab | 236 | io_worker_release(worker); |
c5def4ab JA |
237 | } |
238 | ||
239 | return false; | |
240 | } | |
241 | ||
242 | /* | |
243 | * We need a worker. If we find a free one, we're good. If not, and we're | |
685fe7fe | 244 | * below the max number of workers, create one. |
c5def4ab | 245 | */ |
3146cba9 | 246 | static bool io_wqe_create_worker(struct io_wqe *wqe, struct io_wqe_acct *acct) |
c5def4ab | 247 | { |
c5def4ab JA |
248 | /* |
249 | * Most likely an attempt to queue unbounded work on an io_wq that | |
250 | * wasn't setup with any unbounded workers. | |
251 | */ | |
e6ab8991 PB |
252 | if (unlikely(!acct->max_workers)) |
253 | pr_warn_once("io-wq is not configured for unbound workers"); | |
c5def4ab | 254 | |
94ffb0a2 | 255 | raw_spin_lock(&wqe->lock); |
bc369921 | 256 | if (acct->nr_workers >= acct->max_workers) { |
7a842fb5 HX |
257 | raw_spin_unlock(&wqe->lock); |
258 | return true; | |
94ffb0a2 | 259 | } |
7a842fb5 | 260 | acct->nr_workers++; |
94ffb0a2 | 261 | raw_spin_unlock(&wqe->lock); |
7a842fb5 HX |
262 | atomic_inc(&acct->nr_running); |
263 | atomic_inc(&wqe->wq->worker_refs); | |
264 | return create_io_worker(wqe->wq, wqe, acct->index); | |
c5def4ab JA |
265 | } |
266 | ||
958234d5 | 267 | static void io_wqe_inc_running(struct io_worker *worker) |
c5def4ab | 268 | { |
958234d5 | 269 | struct io_wqe_acct *acct = io_wqe_get_acct(worker); |
c5def4ab JA |
270 | |
271 | atomic_inc(&acct->nr_running); | |
272 | } | |
273 | ||
685fe7fe JA |
274 | static void create_worker_cb(struct callback_head *cb) |
275 | { | |
d3e9f732 | 276 | struct io_worker *worker; |
685fe7fe | 277 | struct io_wq *wq; |
21698274 HX |
278 | struct io_wqe *wqe; |
279 | struct io_wqe_acct *acct; | |
05c5f4ee | 280 | bool do_create = false; |
685fe7fe | 281 | |
d3e9f732 JA |
282 | worker = container_of(cb, struct io_worker, create_work); |
283 | wqe = worker->wqe; | |
21698274 | 284 | wq = wqe->wq; |
d3e9f732 | 285 | acct = &wqe->acct[worker->create_index]; |
a9a4aa9f | 286 | raw_spin_lock(&wqe->lock); |
49e7f0c7 | 287 | if (acct->nr_workers < acct->max_workers) { |
21698274 | 288 | acct->nr_workers++; |
49e7f0c7 HX |
289 | do_create = true; |
290 | } | |
a9a4aa9f | 291 | raw_spin_unlock(&wqe->lock); |
49e7f0c7 | 292 | if (do_create) { |
05c5f4ee | 293 | create_io_worker(wq, wqe, worker->create_index); |
49e7f0c7 HX |
294 | } else { |
295 | atomic_dec(&acct->nr_running); | |
296 | io_worker_ref_put(wq); | |
297 | } | |
d3e9f732 JA |
298 | clear_bit_unlock(0, &worker->create_state); |
299 | io_worker_release(worker); | |
685fe7fe JA |
300 | } |
301 | ||
3146cba9 JA |
302 | static bool io_queue_worker_create(struct io_worker *worker, |
303 | struct io_wqe_acct *acct, | |
304 | task_work_func_t func) | |
685fe7fe | 305 | { |
3146cba9 | 306 | struct io_wqe *wqe = worker->wqe; |
685fe7fe JA |
307 | struct io_wq *wq = wqe->wq; |
308 | ||
309 | /* raced with exit, just ignore create call */ | |
310 | if (test_bit(IO_WQ_BIT_EXIT, &wq->state)) | |
311 | goto fail; | |
d3e9f732 JA |
312 | if (!io_worker_get(worker)) |
313 | goto fail; | |
314 | /* | |
315 | * create_state manages ownership of create_work/index. We should | |
316 | * only need one entry per worker, as the worker going to sleep | |
317 | * will trigger the condition, and waking will clear it once it | |
318 | * runs the task_work. | |
319 | */ | |
320 | if (test_bit(0, &worker->create_state) || | |
321 | test_and_set_bit_lock(0, &worker->create_state)) | |
322 | goto fail_release; | |
685fe7fe | 323 | |
3146cba9 | 324 | init_task_work(&worker->create_work, func); |
d3e9f732 JA |
325 | worker->create_index = acct->index; |
326 | if (!task_work_add(wq->task, &worker->create_work, TWA_SIGNAL)) | |
3146cba9 | 327 | return true; |
d3e9f732 JA |
328 | clear_bit_unlock(0, &worker->create_state); |
329 | fail_release: | |
330 | io_worker_release(worker); | |
685fe7fe JA |
331 | fail: |
332 | atomic_dec(&acct->nr_running); | |
333 | io_worker_ref_put(wq); | |
3146cba9 | 334 | return false; |
685fe7fe JA |
335 | } |
336 | ||
958234d5 | 337 | static void io_wqe_dec_running(struct io_worker *worker) |
c5def4ab JA |
338 | __must_hold(wqe->lock) |
339 | { | |
958234d5 JA |
340 | struct io_wqe_acct *acct = io_wqe_get_acct(worker); |
341 | struct io_wqe *wqe = worker->wqe; | |
c5def4ab | 342 | |
685fe7fe JA |
343 | if (!(worker->flags & IO_WORKER_F_UP)) |
344 | return; | |
345 | ||
f95dc207 | 346 | if (atomic_dec_and_test(&acct->nr_running) && io_acct_run_queue(acct)) { |
685fe7fe JA |
347 | atomic_inc(&acct->nr_running); |
348 | atomic_inc(&wqe->wq->worker_refs); | |
3146cba9 | 349 | io_queue_worker_create(worker, acct, create_worker_cb); |
685fe7fe | 350 | } |
c5def4ab JA |
351 | } |
352 | ||
771b53d0 JA |
353 | /* |
354 | * Worker will start processing some work. Move it to the busy list, if | |
355 | * it's currently on the freelist | |
356 | */ | |
357 | static void __io_worker_busy(struct io_wqe *wqe, struct io_worker *worker, | |
358 | struct io_wq_work *work) | |
359 | __must_hold(wqe->lock) | |
360 | { | |
361 | if (worker->flags & IO_WORKER_F_FREE) { | |
362 | worker->flags &= ~IO_WORKER_F_FREE; | |
363 | hlist_nulls_del_init_rcu(&worker->nulls_node); | |
771b53d0 | 364 | } |
771b53d0 JA |
365 | } |
366 | ||
367 | /* | |
368 | * No work, worker going to sleep. Move to freelist, and unuse mm if we | |
369 | * have one attached. Dropping the mm may potentially sleep, so we drop | |
370 | * the lock in that case and return success. Since the caller has to | |
371 | * retry the loop in that case (we changed task state), we don't regrab | |
372 | * the lock if we return success. | |
373 | */ | |
c6d77d92 | 374 | static void __io_worker_idle(struct io_wqe *wqe, struct io_worker *worker) |
771b53d0 JA |
375 | __must_hold(wqe->lock) |
376 | { | |
377 | if (!(worker->flags & IO_WORKER_F_FREE)) { | |
378 | worker->flags |= IO_WORKER_F_FREE; | |
021d1cdd | 379 | hlist_nulls_add_head_rcu(&worker->nulls_node, &wqe->free_list); |
771b53d0 | 380 | } |
771b53d0 JA |
381 | } |
382 | ||
60cf46ae PB |
383 | static inline unsigned int io_get_work_hash(struct io_wq_work *work) |
384 | { | |
385 | return work->flags >> IO_WQ_HASH_SHIFT; | |
386 | } | |
387 | ||
e941894e JA |
388 | static void io_wait_on_hash(struct io_wqe *wqe, unsigned int hash) |
389 | { | |
390 | struct io_wq *wq = wqe->wq; | |
391 | ||
08bdbd39 | 392 | spin_lock_irq(&wq->hash->wait.lock); |
e941894e JA |
393 | if (list_empty(&wqe->wait.entry)) { |
394 | __add_wait_queue(&wq->hash->wait, &wqe->wait); | |
395 | if (!test_bit(hash, &wq->hash->map)) { | |
396 | __set_current_state(TASK_RUNNING); | |
397 | list_del_init(&wqe->wait.entry); | |
398 | } | |
399 | } | |
08bdbd39 | 400 | spin_unlock_irq(&wq->hash->wait.lock); |
e941894e JA |
401 | } |
402 | ||
f95dc207 | 403 | static struct io_wq_work *io_get_next_work(struct io_wqe_acct *acct, |
0242f642 | 404 | struct io_worker *worker) |
771b53d0 JA |
405 | __must_hold(wqe->lock) |
406 | { | |
6206f0e1 | 407 | struct io_wq_work_node *node, *prev; |
86f3cd1b | 408 | struct io_wq_work *work, *tail; |
e941894e | 409 | unsigned int stall_hash = -1U; |
f95dc207 | 410 | struct io_wqe *wqe = worker->wqe; |
771b53d0 | 411 | |
f95dc207 | 412 | wq_list_for_each(node, prev, &acct->work_list) { |
e941894e JA |
413 | unsigned int hash; |
414 | ||
6206f0e1 JA |
415 | work = container_of(node, struct io_wq_work, list); |
416 | ||
771b53d0 | 417 | /* not hashed, can run anytime */ |
8766dd51 | 418 | if (!io_wq_is_hashed(work)) { |
f95dc207 | 419 | wq_list_del(&acct->work_list, node, prev); |
771b53d0 JA |
420 | return work; |
421 | } | |
422 | ||
60cf46ae | 423 | hash = io_get_work_hash(work); |
e941894e JA |
424 | /* all items with this hash lie in [work, tail] */ |
425 | tail = wqe->hash_tail[hash]; | |
426 | ||
427 | /* hashed, can run if not already running */ | |
428 | if (!test_and_set_bit(hash, &wqe->wq->hash->map)) { | |
86f3cd1b | 429 | wqe->hash_tail[hash] = NULL; |
f95dc207 | 430 | wq_list_cut(&acct->work_list, &tail->list, prev); |
771b53d0 JA |
431 | return work; |
432 | } | |
e941894e JA |
433 | if (stall_hash == -1U) |
434 | stall_hash = hash; | |
435 | /* fast forward to a next hash, for-each will fix up @prev */ | |
436 | node = &tail->list; | |
437 | } | |
438 | ||
439 | if (stall_hash != -1U) { | |
0242f642 JA |
440 | /* |
441 | * Set this before dropping the lock to avoid racing with new | |
442 | * work being added and clearing the stalled bit. | |
443 | */ | |
f95dc207 | 444 | set_bit(IO_ACCT_STALLED_BIT, &acct->flags); |
e941894e JA |
445 | raw_spin_unlock(&wqe->lock); |
446 | io_wait_on_hash(wqe, stall_hash); | |
447 | raw_spin_lock(&wqe->lock); | |
771b53d0 JA |
448 | } |
449 | ||
450 | return NULL; | |
451 | } | |
452 | ||
00ddff43 | 453 | static bool io_flush_signals(void) |
cccf0ee8 | 454 | { |
0b8cfa97 | 455 | if (unlikely(test_thread_flag(TIF_NOTIFY_SIGNAL))) { |
00ddff43 | 456 | __set_current_state(TASK_RUNNING); |
0b8cfa97 | 457 | tracehook_notify_signal(); |
00ddff43 | 458 | return true; |
cccf0ee8 | 459 | } |
00ddff43 | 460 | return false; |
dc026a73 PB |
461 | } |
462 | ||
463 | static void io_assign_current_work(struct io_worker *worker, | |
464 | struct io_wq_work *work) | |
465 | { | |
d78298e7 | 466 | if (work) { |
3bfe6106 | 467 | io_flush_signals(); |
d78298e7 PB |
468 | cond_resched(); |
469 | } | |
dc026a73 | 470 | |
a9a4aa9f | 471 | spin_lock(&worker->lock); |
dc026a73 | 472 | worker->cur_work = work; |
a9a4aa9f | 473 | spin_unlock(&worker->lock); |
dc026a73 PB |
474 | } |
475 | ||
60cf46ae PB |
476 | static void io_wqe_enqueue(struct io_wqe *wqe, struct io_wq_work *work); |
477 | ||
771b53d0 JA |
478 | static void io_worker_handle_work(struct io_worker *worker) |
479 | __releases(wqe->lock) | |
480 | { | |
f95dc207 | 481 | struct io_wqe_acct *acct = io_wqe_get_acct(worker); |
771b53d0 JA |
482 | struct io_wqe *wqe = worker->wqe; |
483 | struct io_wq *wq = wqe->wq; | |
c60eb049 | 484 | bool do_kill = test_bit(IO_WQ_BIT_EXIT, &wq->state); |
771b53d0 JA |
485 | |
486 | do { | |
86f3cd1b | 487 | struct io_wq_work *work; |
f462fd36 | 488 | get_next: |
771b53d0 JA |
489 | /* |
490 | * If we got some work, mark us as busy. If we didn't, but | |
491 | * the list isn't empty, it means we stalled on hashed work. | |
492 | * Mark us stalled so we don't keep looking for work when we | |
493 | * can't make progress, any work completion or insertion will | |
494 | * clear the stalled flag. | |
495 | */ | |
f95dc207 | 496 | work = io_get_next_work(acct, worker); |
771b53d0 JA |
497 | if (work) |
498 | __io_worker_busy(wqe, worker, work); | |
771b53d0 | 499 | |
a9a4aa9f | 500 | raw_spin_unlock(&wqe->lock); |
771b53d0 JA |
501 | if (!work) |
502 | break; | |
58e39319 | 503 | io_assign_current_work(worker, work); |
e941894e | 504 | __set_current_state(TASK_RUNNING); |
36c2f922 | 505 | |
dc026a73 PB |
506 | /* handle a whole dependent link */ |
507 | do { | |
5280f7e5 | 508 | struct io_wq_work *next_hashed, *linked; |
b089ed39 | 509 | unsigned int hash = io_get_work_hash(work); |
dc026a73 | 510 | |
86f3cd1b | 511 | next_hashed = wq_next_work(work); |
c60eb049 PB |
512 | |
513 | if (unlikely(do_kill) && (work->flags & IO_WQ_WORK_UNBOUND)) | |
514 | work->flags |= IO_WQ_WORK_CANCEL; | |
5280f7e5 PB |
515 | wq->do_work(work); |
516 | io_assign_current_work(worker, NULL); | |
dc026a73 | 517 | |
5280f7e5 | 518 | linked = wq->free_work(work); |
86f3cd1b PB |
519 | work = next_hashed; |
520 | if (!work && linked && !io_wq_is_hashed(linked)) { | |
521 | work = linked; | |
522 | linked = NULL; | |
523 | } | |
524 | io_assign_current_work(worker, work); | |
86f3cd1b PB |
525 | if (linked) |
526 | io_wqe_enqueue(wqe, linked); | |
527 | ||
528 | if (hash != -1U && !next_hashed) { | |
e941894e | 529 | clear_bit(hash, &wq->hash->map); |
f95dc207 | 530 | clear_bit(IO_ACCT_STALLED_BIT, &acct->flags); |
e941894e JA |
531 | if (wq_has_sleeper(&wq->hash->wait)) |
532 | wake_up(&wq->hash->wait); | |
a9a4aa9f | 533 | raw_spin_lock(&wqe->lock); |
f462fd36 PB |
534 | /* skip unnecessary unlock-lock wqe->lock */ |
535 | if (!work) | |
536 | goto get_next; | |
a9a4aa9f | 537 | raw_spin_unlock(&wqe->lock); |
7d723065 | 538 | } |
58e39319 | 539 | } while (work); |
7d723065 | 540 | |
a9a4aa9f | 541 | raw_spin_lock(&wqe->lock); |
771b53d0 JA |
542 | } while (1); |
543 | } | |
544 | ||
771b53d0 JA |
545 | static int io_wqe_worker(void *data) |
546 | { | |
547 | struct io_worker *worker = data; | |
f95dc207 | 548 | struct io_wqe_acct *acct = io_wqe_get_acct(worker); |
771b53d0 JA |
549 | struct io_wqe *wqe = worker->wqe; |
550 | struct io_wq *wq = wqe->wq; | |
05c5f4ee | 551 | bool last_timeout = false; |
46fe18b1 | 552 | char buf[TASK_COMM_LEN]; |
771b53d0 | 553 | |
46fe18b1 | 554 | worker->flags |= (IO_WORKER_F_UP | IO_WORKER_F_RUNNING); |
46fe18b1 | 555 | |
685fe7fe | 556 | snprintf(buf, sizeof(buf), "iou-wrk-%d", wq->task->pid); |
46fe18b1 | 557 | set_task_comm(current, buf); |
771b53d0 JA |
558 | |
559 | while (!test_bit(IO_WQ_BIT_EXIT, &wq->state)) { | |
16efa4fc JA |
560 | long ret; |
561 | ||
506d95ff | 562 | set_current_state(TASK_INTERRUPTIBLE); |
e995d512 | 563 | loop: |
a9a4aa9f | 564 | raw_spin_lock(&wqe->lock); |
f95dc207 | 565 | if (io_acct_run_queue(acct)) { |
771b53d0 | 566 | io_worker_handle_work(worker); |
e995d512 | 567 | goto loop; |
771b53d0 | 568 | } |
05c5f4ee JA |
569 | /* timed out, exit unless we're the last worker */ |
570 | if (last_timeout && acct->nr_workers > 1) { | |
767a65e9 | 571 | acct->nr_workers--; |
05c5f4ee JA |
572 | raw_spin_unlock(&wqe->lock); |
573 | __set_current_state(TASK_RUNNING); | |
574 | break; | |
575 | } | |
576 | last_timeout = false; | |
c6d77d92 | 577 | __io_worker_idle(wqe, worker); |
a9a4aa9f | 578 | raw_spin_unlock(&wqe->lock); |
00ddff43 JA |
579 | if (io_flush_signals()) |
580 | continue; | |
16efa4fc | 581 | ret = schedule_timeout(WORKER_IDLE_TIMEOUT); |
dbe1bdbb JA |
582 | if (signal_pending(current)) { |
583 | struct ksignal ksig; | |
584 | ||
585 | if (!get_signal(&ksig)) | |
586 | continue; | |
78f8876c | 587 | break; |
dbe1bdbb | 588 | } |
05c5f4ee | 589 | last_timeout = !ret; |
771b53d0 JA |
590 | } |
591 | ||
771b53d0 | 592 | if (test_bit(IO_WQ_BIT_EXIT, &wq->state)) { |
a9a4aa9f | 593 | raw_spin_lock(&wqe->lock); |
e587227b | 594 | io_worker_handle_work(worker); |
771b53d0 JA |
595 | } |
596 | ||
597 | io_worker_exit(worker); | |
598 | return 0; | |
599 | } | |
600 | ||
771b53d0 JA |
601 | /* |
602 | * Called when a worker is scheduled in. Mark us as currently running. | |
603 | */ | |
604 | void io_wq_worker_running(struct task_struct *tsk) | |
605 | { | |
3bfe6106 | 606 | struct io_worker *worker = tsk->pf_io_worker; |
771b53d0 | 607 | |
3bfe6106 JA |
608 | if (!worker) |
609 | return; | |
771b53d0 JA |
610 | if (!(worker->flags & IO_WORKER_F_UP)) |
611 | return; | |
612 | if (worker->flags & IO_WORKER_F_RUNNING) | |
613 | return; | |
614 | worker->flags |= IO_WORKER_F_RUNNING; | |
958234d5 | 615 | io_wqe_inc_running(worker); |
771b53d0 JA |
616 | } |
617 | ||
618 | /* | |
619 | * Called when worker is going to sleep. If there are no workers currently | |
685fe7fe | 620 | * running and we have work pending, wake up a free one or create a new one. |
771b53d0 JA |
621 | */ |
622 | void io_wq_worker_sleeping(struct task_struct *tsk) | |
623 | { | |
3bfe6106 | 624 | struct io_worker *worker = tsk->pf_io_worker; |
771b53d0 | 625 | |
3bfe6106 JA |
626 | if (!worker) |
627 | return; | |
771b53d0 JA |
628 | if (!(worker->flags & IO_WORKER_F_UP)) |
629 | return; | |
630 | if (!(worker->flags & IO_WORKER_F_RUNNING)) | |
631 | return; | |
632 | ||
633 | worker->flags &= ~IO_WORKER_F_RUNNING; | |
634 | ||
a9a4aa9f | 635 | raw_spin_lock(&worker->wqe->lock); |
958234d5 | 636 | io_wqe_dec_running(worker); |
a9a4aa9f | 637 | raw_spin_unlock(&worker->wqe->lock); |
771b53d0 JA |
638 | } |
639 | ||
3146cba9 JA |
640 | static void io_init_new_worker(struct io_wqe *wqe, struct io_worker *worker, |
641 | struct task_struct *tsk) | |
642 | { | |
643 | tsk->pf_io_worker = worker; | |
644 | worker->task = tsk; | |
645 | set_cpus_allowed_ptr(tsk, wqe->cpu_mask); | |
646 | tsk->flags |= PF_NO_SETAFFINITY; | |
647 | ||
648 | raw_spin_lock(&wqe->lock); | |
649 | hlist_nulls_add_head_rcu(&worker->nulls_node, &wqe->free_list); | |
650 | list_add_tail_rcu(&worker->all_list, &wqe->all_list); | |
651 | worker->flags |= IO_WORKER_F_FREE; | |
652 | raw_spin_unlock(&wqe->lock); | |
653 | wake_up_new_task(tsk); | |
654 | } | |
655 | ||
656 | static bool io_wq_work_match_all(struct io_wq_work *work, void *data) | |
657 | { | |
658 | return true; | |
659 | } | |
660 | ||
661 | static inline bool io_should_retry_thread(long err) | |
662 | { | |
663 | switch (err) { | |
664 | case -EAGAIN: | |
665 | case -ERESTARTSYS: | |
666 | case -ERESTARTNOINTR: | |
667 | case -ERESTARTNOHAND: | |
668 | return true; | |
669 | default: | |
670 | return false; | |
671 | } | |
672 | } | |
673 | ||
674 | static void create_worker_cont(struct callback_head *cb) | |
675 | { | |
676 | struct io_worker *worker; | |
677 | struct task_struct *tsk; | |
678 | struct io_wqe *wqe; | |
679 | ||
680 | worker = container_of(cb, struct io_worker, create_work); | |
681 | clear_bit_unlock(0, &worker->create_state); | |
682 | wqe = worker->wqe; | |
683 | tsk = create_io_thread(io_wqe_worker, worker, wqe->node); | |
684 | if (!IS_ERR(tsk)) { | |
685 | io_init_new_worker(wqe, worker, tsk); | |
686 | io_worker_release(worker); | |
687 | return; | |
688 | } else if (!io_should_retry_thread(PTR_ERR(tsk))) { | |
689 | struct io_wqe_acct *acct = io_wqe_get_acct(worker); | |
690 | ||
691 | atomic_dec(&acct->nr_running); | |
692 | raw_spin_lock(&wqe->lock); | |
693 | acct->nr_workers--; | |
694 | if (!acct->nr_workers) { | |
695 | struct io_cb_cancel_data match = { | |
696 | .fn = io_wq_work_match_all, | |
697 | .cancel_all = true, | |
698 | }; | |
699 | ||
700 | while (io_acct_cancel_pending_work(wqe, acct, &match)) | |
701 | raw_spin_lock(&wqe->lock); | |
702 | } | |
703 | raw_spin_unlock(&wqe->lock); | |
704 | io_worker_ref_put(wqe->wq); | |
66e70be7 | 705 | kfree(worker); |
3146cba9 JA |
706 | return; |
707 | } | |
708 | ||
709 | /* re-create attempts grab a new worker ref, drop the existing one */ | |
710 | io_worker_release(worker); | |
711 | schedule_work(&worker->work); | |
712 | } | |
713 | ||
714 | static void io_workqueue_create(struct work_struct *work) | |
715 | { | |
716 | struct io_worker *worker = container_of(work, struct io_worker, work); | |
717 | struct io_wqe_acct *acct = io_wqe_get_acct(worker); | |
718 | ||
719 | if (!io_queue_worker_create(worker, acct, create_worker_cont)) { | |
720 | clear_bit_unlock(0, &worker->create_state); | |
721 | io_worker_release(worker); | |
66e70be7 | 722 | kfree(worker); |
3146cba9 JA |
723 | } |
724 | } | |
725 | ||
726 | static bool create_io_worker(struct io_wq *wq, struct io_wqe *wqe, int index) | |
3bfe6106 | 727 | { |
46fe18b1 | 728 | struct io_wqe_acct *acct = &wqe->acct[index]; |
3bfe6106 | 729 | struct io_worker *worker; |
46fe18b1 | 730 | struct task_struct *tsk; |
3bfe6106 | 731 | |
8b3e78b5 JA |
732 | __set_current_state(TASK_RUNNING); |
733 | ||
3bfe6106 | 734 | worker = kzalloc_node(sizeof(*worker), GFP_KERNEL, wqe->node); |
3146cba9 | 735 | if (!worker) { |
685fe7fe JA |
736 | fail: |
737 | atomic_dec(&acct->nr_running); | |
a9a4aa9f | 738 | raw_spin_lock(&wqe->lock); |
3d4e4fac | 739 | acct->nr_workers--; |
a9a4aa9f | 740 | raw_spin_unlock(&wqe->lock); |
685fe7fe | 741 | io_worker_ref_put(wq); |
3146cba9 | 742 | return false; |
3bfe6106 | 743 | } |
46fe18b1 | 744 | |
3146cba9 JA |
745 | refcount_set(&worker->ref, 1); |
746 | worker->wqe = wqe; | |
747 | spin_lock_init(&worker->lock); | |
748 | init_completion(&worker->ref_done); | |
46fe18b1 | 749 | |
46fe18b1 JA |
750 | if (index == IO_WQ_ACCT_BOUND) |
751 | worker->flags |= IO_WORKER_F_BOUND; | |
3146cba9 JA |
752 | |
753 | tsk = create_io_thread(io_wqe_worker, worker, wqe->node); | |
754 | if (!IS_ERR(tsk)) { | |
755 | io_init_new_worker(wqe, worker, tsk); | |
756 | } else if (!io_should_retry_thread(PTR_ERR(tsk))) { | |
66e70be7 | 757 | kfree(worker); |
3146cba9 JA |
758 | goto fail; |
759 | } else { | |
760 | INIT_WORK(&worker->work, io_workqueue_create); | |
761 | schedule_work(&worker->work); | |
762 | } | |
763 | ||
764 | return true; | |
771b53d0 JA |
765 | } |
766 | ||
c4068bf8 HD |
767 | /* |
768 | * Iterate the passed in list and call the specific function for each | |
769 | * worker that isn't exiting | |
770 | */ | |
771 | static bool io_wq_for_each_worker(struct io_wqe *wqe, | |
772 | bool (*func)(struct io_worker *, void *), | |
773 | void *data) | |
774 | { | |
775 | struct io_worker *worker; | |
776 | bool ret = false; | |
777 | ||
778 | list_for_each_entry_rcu(worker, &wqe->all_list, all_list) { | |
779 | if (io_worker_get(worker)) { | |
780 | /* no task if node is/was offline */ | |
781 | if (worker->task) | |
782 | ret = func(worker, data); | |
783 | io_worker_release(worker); | |
784 | if (ret) | |
785 | break; | |
786 | } | |
787 | } | |
788 | ||
789 | return ret; | |
790 | } | |
791 | ||
792 | static bool io_wq_worker_wake(struct io_worker *worker, void *data) | |
793 | { | |
46fe18b1 | 794 | set_notify_signal(worker->task); |
c4068bf8 HD |
795 | wake_up_process(worker->task); |
796 | return false; | |
797 | } | |
798 | ||
e9fd9396 | 799 | static void io_run_cancel(struct io_wq_work *work, struct io_wqe *wqe) |
fc04c39b | 800 | { |
e9fd9396 PB |
801 | struct io_wq *wq = wqe->wq; |
802 | ||
fc04c39b | 803 | do { |
fc04c39b | 804 | work->flags |= IO_WQ_WORK_CANCEL; |
5280f7e5 PB |
805 | wq->do_work(work); |
806 | work = wq->free_work(work); | |
fc04c39b PB |
807 | } while (work); |
808 | } | |
809 | ||
86f3cd1b PB |
810 | static void io_wqe_insert_work(struct io_wqe *wqe, struct io_wq_work *work) |
811 | { | |
f95dc207 | 812 | struct io_wqe_acct *acct = io_work_get_acct(wqe, work); |
86f3cd1b PB |
813 | unsigned int hash; |
814 | struct io_wq_work *tail; | |
815 | ||
816 | if (!io_wq_is_hashed(work)) { | |
817 | append: | |
f95dc207 | 818 | wq_list_add_tail(&work->list, &acct->work_list); |
86f3cd1b PB |
819 | return; |
820 | } | |
821 | ||
822 | hash = io_get_work_hash(work); | |
823 | tail = wqe->hash_tail[hash]; | |
824 | wqe->hash_tail[hash] = work; | |
825 | if (!tail) | |
826 | goto append; | |
827 | ||
f95dc207 | 828 | wq_list_add_after(&work->list, &tail->list, &acct->work_list); |
86f3cd1b PB |
829 | } |
830 | ||
713b9825 PB |
831 | static bool io_wq_work_match_item(struct io_wq_work *work, void *data) |
832 | { | |
833 | return work == data; | |
834 | } | |
835 | ||
771b53d0 JA |
836 | static void io_wqe_enqueue(struct io_wqe *wqe, struct io_wq_work *work) |
837 | { | |
c5def4ab | 838 | struct io_wqe_acct *acct = io_work_get_acct(wqe, work); |
94ffb0a2 JA |
839 | unsigned work_flags = work->flags; |
840 | bool do_create; | |
771b53d0 | 841 | |
991468dc JA |
842 | /* |
843 | * If io-wq is exiting for this task, or if the request has explicitly | |
844 | * been marked as one that should not get executed, cancel it here. | |
845 | */ | |
846 | if (test_bit(IO_WQ_BIT_EXIT, &wqe->wq->state) || | |
847 | (work->flags & IO_WQ_WORK_CANCEL)) { | |
70e35125 | 848 | io_run_cancel(work, wqe); |
4fb6ac32 JA |
849 | return; |
850 | } | |
851 | ||
a9a4aa9f | 852 | raw_spin_lock(&wqe->lock); |
86f3cd1b | 853 | io_wqe_insert_work(wqe, work); |
f95dc207 | 854 | clear_bit(IO_ACCT_STALLED_BIT, &acct->flags); |
94ffb0a2 JA |
855 | |
856 | rcu_read_lock(); | |
f95dc207 | 857 | do_create = !io_wqe_activate_free_worker(wqe, acct); |
94ffb0a2 JA |
858 | rcu_read_unlock(); |
859 | ||
a9a4aa9f | 860 | raw_spin_unlock(&wqe->lock); |
771b53d0 | 861 | |
94ffb0a2 | 862 | if (do_create && ((work_flags & IO_WQ_WORK_CONCURRENT) || |
3146cba9 JA |
863 | !atomic_read(&acct->nr_running))) { |
864 | bool did_create; | |
865 | ||
866 | did_create = io_wqe_create_worker(wqe, acct); | |
713b9825 PB |
867 | if (likely(did_create)) |
868 | return; | |
869 | ||
870 | raw_spin_lock(&wqe->lock); | |
871 | /* fatal condition, failed to create the first worker */ | |
872 | if (!acct->nr_workers) { | |
873 | struct io_cb_cancel_data match = { | |
874 | .fn = io_wq_work_match_item, | |
875 | .data = work, | |
876 | .cancel_all = false, | |
877 | }; | |
878 | ||
879 | if (io_acct_cancel_pending_work(wqe, acct, &match)) | |
880 | raw_spin_lock(&wqe->lock); | |
3146cba9 | 881 | } |
713b9825 | 882 | raw_spin_unlock(&wqe->lock); |
3146cba9 | 883 | } |
771b53d0 JA |
884 | } |
885 | ||
886 | void io_wq_enqueue(struct io_wq *wq, struct io_wq_work *work) | |
887 | { | |
888 | struct io_wqe *wqe = wq->wqes[numa_node_id()]; | |
889 | ||
890 | io_wqe_enqueue(wqe, work); | |
891 | } | |
892 | ||
893 | /* | |
8766dd51 PB |
894 | * Work items that hash to the same value will not be done in parallel. |
895 | * Used to limit concurrent writes, generally hashed by inode. | |
771b53d0 | 896 | */ |
8766dd51 | 897 | void io_wq_hash_work(struct io_wq_work *work, void *val) |
771b53d0 | 898 | { |
8766dd51 | 899 | unsigned int bit; |
771b53d0 JA |
900 | |
901 | bit = hash_ptr(val, IO_WQ_HASH_ORDER); | |
902 | work->flags |= (IO_WQ_WORK_HASHED | (bit << IO_WQ_HASH_SHIFT)); | |
771b53d0 JA |
903 | } |
904 | ||
2293b419 | 905 | static bool io_wq_worker_cancel(struct io_worker *worker, void *data) |
62755e35 | 906 | { |
2293b419 | 907 | struct io_cb_cancel_data *match = data; |
62755e35 JA |
908 | |
909 | /* | |
910 | * Hold the lock to avoid ->cur_work going out of scope, caller | |
36c2f922 | 911 | * may dereference the passed in work. |
62755e35 | 912 | */ |
a9a4aa9f | 913 | spin_lock(&worker->lock); |
62755e35 | 914 | if (worker->cur_work && |
2293b419 | 915 | match->fn(worker->cur_work, match->data)) { |
3bfe6106 | 916 | set_notify_signal(worker->task); |
4f26bda1 | 917 | match->nr_running++; |
771b53d0 | 918 | } |
a9a4aa9f | 919 | spin_unlock(&worker->lock); |
771b53d0 | 920 | |
4f26bda1 | 921 | return match->nr_running && !match->cancel_all; |
771b53d0 JA |
922 | } |
923 | ||
204361a7 PB |
924 | static inline void io_wqe_remove_pending(struct io_wqe *wqe, |
925 | struct io_wq_work *work, | |
926 | struct io_wq_work_node *prev) | |
927 | { | |
f95dc207 | 928 | struct io_wqe_acct *acct = io_work_get_acct(wqe, work); |
204361a7 PB |
929 | unsigned int hash = io_get_work_hash(work); |
930 | struct io_wq_work *prev_work = NULL; | |
931 | ||
932 | if (io_wq_is_hashed(work) && work == wqe->hash_tail[hash]) { | |
933 | if (prev) | |
934 | prev_work = container_of(prev, struct io_wq_work, list); | |
935 | if (prev_work && io_get_work_hash(prev_work) == hash) | |
936 | wqe->hash_tail[hash] = prev_work; | |
937 | else | |
938 | wqe->hash_tail[hash] = NULL; | |
939 | } | |
f95dc207 | 940 | wq_list_del(&acct->work_list, &work->list, prev); |
204361a7 PB |
941 | } |
942 | ||
3146cba9 JA |
943 | static bool io_acct_cancel_pending_work(struct io_wqe *wqe, |
944 | struct io_wqe_acct *acct, | |
945 | struct io_cb_cancel_data *match) | |
946 | __releases(wqe->lock) | |
771b53d0 | 947 | { |
6206f0e1 | 948 | struct io_wq_work_node *node, *prev; |
771b53d0 | 949 | struct io_wq_work *work; |
771b53d0 | 950 | |
3146cba9 JA |
951 | wq_list_for_each(node, prev, &acct->work_list) { |
952 | work = container_of(node, struct io_wq_work, list); | |
953 | if (!match->fn(work, match->data)) | |
954 | continue; | |
955 | io_wqe_remove_pending(wqe, work, prev); | |
956 | raw_spin_unlock(&wqe->lock); | |
957 | io_run_cancel(work, wqe); | |
958 | match->nr_pending++; | |
959 | /* not safe to continue after unlock */ | |
960 | return true; | |
961 | } | |
962 | ||
963 | return false; | |
964 | } | |
965 | ||
966 | static void io_wqe_cancel_pending_work(struct io_wqe *wqe, | |
967 | struct io_cb_cancel_data *match) | |
968 | { | |
969 | int i; | |
4f26bda1 | 970 | retry: |
a9a4aa9f | 971 | raw_spin_lock(&wqe->lock); |
f95dc207 JA |
972 | for (i = 0; i < IO_WQ_ACCT_NR; i++) { |
973 | struct io_wqe_acct *acct = io_get_acct(wqe, i == 0); | |
4f26bda1 | 974 | |
3146cba9 JA |
975 | if (io_acct_cancel_pending_work(wqe, acct, match)) { |
976 | if (match->cancel_all) | |
977 | goto retry; | |
978 | return; | |
f95dc207 | 979 | } |
771b53d0 | 980 | } |
a9a4aa9f | 981 | raw_spin_unlock(&wqe->lock); |
f4c2665e PB |
982 | } |
983 | ||
4f26bda1 | 984 | static void io_wqe_cancel_running_work(struct io_wqe *wqe, |
f4c2665e PB |
985 | struct io_cb_cancel_data *match) |
986 | { | |
771b53d0 | 987 | rcu_read_lock(); |
4f26bda1 | 988 | io_wq_for_each_worker(wqe, io_wq_worker_cancel, match); |
771b53d0 | 989 | rcu_read_unlock(); |
771b53d0 JA |
990 | } |
991 | ||
2293b419 | 992 | enum io_wq_cancel io_wq_cancel_cb(struct io_wq *wq, work_cancel_fn *cancel, |
4f26bda1 | 993 | void *data, bool cancel_all) |
771b53d0 | 994 | { |
2293b419 | 995 | struct io_cb_cancel_data match = { |
4f26bda1 PB |
996 | .fn = cancel, |
997 | .data = data, | |
998 | .cancel_all = cancel_all, | |
00bcda13 | 999 | }; |
3fc50ab5 | 1000 | int node; |
771b53d0 | 1001 | |
f4c2665e PB |
1002 | /* |
1003 | * First check pending list, if we're lucky we can just remove it | |
1004 | * from there. CANCEL_OK means that the work is returned as-new, | |
1005 | * no completion will be posted for it. | |
1006 | */ | |
3fc50ab5 JH |
1007 | for_each_node(node) { |
1008 | struct io_wqe *wqe = wq->wqes[node]; | |
771b53d0 | 1009 | |
4f26bda1 PB |
1010 | io_wqe_cancel_pending_work(wqe, &match); |
1011 | if (match.nr_pending && !match.cancel_all) | |
f4c2665e | 1012 | return IO_WQ_CANCEL_OK; |
771b53d0 JA |
1013 | } |
1014 | ||
f4c2665e PB |
1015 | /* |
1016 | * Now check if a free (going busy) or busy worker has the work | |
1017 | * currently running. If we find it there, we'll return CANCEL_RUNNING | |
1018 | * as an indication that we attempt to signal cancellation. The | |
1019 | * completion will run normally in this case. | |
1020 | */ | |
1021 | for_each_node(node) { | |
1022 | struct io_wqe *wqe = wq->wqes[node]; | |
1023 | ||
4f26bda1 PB |
1024 | io_wqe_cancel_running_work(wqe, &match); |
1025 | if (match.nr_running && !match.cancel_all) | |
f4c2665e PB |
1026 | return IO_WQ_CANCEL_RUNNING; |
1027 | } | |
1028 | ||
4f26bda1 PB |
1029 | if (match.nr_running) |
1030 | return IO_WQ_CANCEL_RUNNING; | |
1031 | if (match.nr_pending) | |
1032 | return IO_WQ_CANCEL_OK; | |
f4c2665e | 1033 | return IO_WQ_CANCEL_NOTFOUND; |
771b53d0 JA |
1034 | } |
1035 | ||
e941894e JA |
1036 | static int io_wqe_hash_wake(struct wait_queue_entry *wait, unsigned mode, |
1037 | int sync, void *key) | |
1038 | { | |
1039 | struct io_wqe *wqe = container_of(wait, struct io_wqe, wait); | |
f95dc207 | 1040 | int i; |
e941894e JA |
1041 | |
1042 | list_del_init(&wait->entry); | |
1043 | ||
1044 | rcu_read_lock(); | |
f95dc207 JA |
1045 | for (i = 0; i < IO_WQ_ACCT_NR; i++) { |
1046 | struct io_wqe_acct *acct = &wqe->acct[i]; | |
1047 | ||
1048 | if (test_and_clear_bit(IO_ACCT_STALLED_BIT, &acct->flags)) | |
1049 | io_wqe_activate_free_worker(wqe, acct); | |
1050 | } | |
e941894e | 1051 | rcu_read_unlock(); |
e941894e JA |
1052 | return 1; |
1053 | } | |
1054 | ||
576a347b | 1055 | struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data) |
771b53d0 | 1056 | { |
f95dc207 | 1057 | int ret, node, i; |
771b53d0 JA |
1058 | struct io_wq *wq; |
1059 | ||
f5fa38c5 | 1060 | if (WARN_ON_ONCE(!data->free_work || !data->do_work)) |
e9fd9396 | 1061 | return ERR_PTR(-EINVAL); |
e6ab8991 PB |
1062 | if (WARN_ON_ONCE(!bounded)) |
1063 | return ERR_PTR(-EINVAL); | |
e9fd9396 | 1064 | |
c7f405d6 | 1065 | wq = kzalloc(struct_size(wq, wqes, nr_node_ids), GFP_KERNEL); |
771b53d0 JA |
1066 | if (!wq) |
1067 | return ERR_PTR(-ENOMEM); | |
43c01fbe JA |
1068 | ret = cpuhp_state_add_instance_nocalls(io_wq_online, &wq->cpuhp_node); |
1069 | if (ret) | |
c7f405d6 | 1070 | goto err_wq; |
771b53d0 | 1071 | |
e941894e JA |
1072 | refcount_inc(&data->hash->refs); |
1073 | wq->hash = data->hash; | |
e9fd9396 | 1074 | wq->free_work = data->free_work; |
f5fa38c5 | 1075 | wq->do_work = data->do_work; |
7d723065 | 1076 | |
43c01fbe | 1077 | ret = -ENOMEM; |
3fc50ab5 | 1078 | for_each_node(node) { |
771b53d0 | 1079 | struct io_wqe *wqe; |
7563439a | 1080 | int alloc_node = node; |
771b53d0 | 1081 | |
7563439a JA |
1082 | if (!node_online(alloc_node)) |
1083 | alloc_node = NUMA_NO_NODE; | |
1084 | wqe = kzalloc_node(sizeof(struct io_wqe), GFP_KERNEL, alloc_node); | |
771b53d0 | 1085 | if (!wqe) |
3fc50ab5 | 1086 | goto err; |
0e03496d JA |
1087 | if (!alloc_cpumask_var(&wqe->cpu_mask, GFP_KERNEL)) |
1088 | goto err; | |
1089 | cpumask_copy(wqe->cpu_mask, cpumask_of_node(node)); | |
3fc50ab5 | 1090 | wq->wqes[node] = wqe; |
7563439a | 1091 | wqe->node = alloc_node; |
c5def4ab | 1092 | wqe->acct[IO_WQ_ACCT_BOUND].max_workers = bounded; |
728f13e7 | 1093 | wqe->acct[IO_WQ_ACCT_UNBOUND].max_workers = |
c5def4ab | 1094 | task_rlimit(current, RLIMIT_NPROC); |
e941894e | 1095 | INIT_LIST_HEAD(&wqe->wait.entry); |
f95dc207 JA |
1096 | wqe->wait.func = io_wqe_hash_wake; |
1097 | for (i = 0; i < IO_WQ_ACCT_NR; i++) { | |
1098 | struct io_wqe_acct *acct = &wqe->acct[i]; | |
1099 | ||
1100 | acct->index = i; | |
1101 | atomic_set(&acct->nr_running, 0); | |
1102 | INIT_WQ_LIST(&acct->work_list); | |
1103 | } | |
771b53d0 | 1104 | wqe->wq = wq; |
95da8465 | 1105 | raw_spin_lock_init(&wqe->lock); |
021d1cdd | 1106 | INIT_HLIST_NULLS_HEAD(&wqe->free_list, 0); |
e61df66c | 1107 | INIT_LIST_HEAD(&wqe->all_list); |
771b53d0 JA |
1108 | } |
1109 | ||
685fe7fe | 1110 | wq->task = get_task_struct(data->task); |
685fe7fe JA |
1111 | atomic_set(&wq->worker_refs, 1); |
1112 | init_completion(&wq->worker_done); | |
1113 | return wq; | |
b60fda60 | 1114 | err: |
dc7bbc9e | 1115 | io_wq_put_hash(data->hash); |
43c01fbe | 1116 | cpuhp_state_remove_instance_nocalls(io_wq_online, &wq->cpuhp_node); |
0e03496d JA |
1117 | for_each_node(node) { |
1118 | if (!wq->wqes[node]) | |
1119 | continue; | |
1120 | free_cpumask_var(wq->wqes[node]->cpu_mask); | |
3fc50ab5 | 1121 | kfree(wq->wqes[node]); |
0e03496d | 1122 | } |
43c01fbe | 1123 | err_wq: |
b60fda60 | 1124 | kfree(wq); |
771b53d0 JA |
1125 | return ERR_PTR(ret); |
1126 | } | |
1127 | ||
c80ca470 JA |
1128 | static bool io_task_work_match(struct callback_head *cb, void *data) |
1129 | { | |
d3e9f732 | 1130 | struct io_worker *worker; |
c80ca470 | 1131 | |
3b33e3f4 | 1132 | if (cb->func != create_worker_cb && cb->func != create_worker_cont) |
c80ca470 | 1133 | return false; |
d3e9f732 JA |
1134 | worker = container_of(cb, struct io_worker, create_work); |
1135 | return worker->wqe->wq == data; | |
c80ca470 JA |
1136 | } |
1137 | ||
17a91051 PB |
1138 | void io_wq_exit_start(struct io_wq *wq) |
1139 | { | |
1140 | set_bit(IO_WQ_BIT_EXIT, &wq->state); | |
1141 | } | |
1142 | ||
685fe7fe | 1143 | static void io_wq_exit_workers(struct io_wq *wq) |
afcc4015 | 1144 | { |
685fe7fe JA |
1145 | struct callback_head *cb; |
1146 | int node; | |
1147 | ||
685fe7fe JA |
1148 | if (!wq->task) |
1149 | return; | |
1150 | ||
c80ca470 | 1151 | while ((cb = task_work_cancel_match(wq->task, io_task_work_match, wq)) != NULL) { |
d3e9f732 | 1152 | struct io_worker *worker; |
3b33e3f4 | 1153 | struct io_wqe_acct *acct; |
685fe7fe | 1154 | |
d3e9f732 | 1155 | worker = container_of(cb, struct io_worker, create_work); |
3b33e3f4 JA |
1156 | acct = io_wqe_get_acct(worker); |
1157 | atomic_dec(&acct->nr_running); | |
1158 | raw_spin_lock(&worker->wqe->lock); | |
1159 | acct->nr_workers--; | |
1160 | raw_spin_unlock(&worker->wqe->lock); | |
685fe7fe | 1161 | io_worker_ref_put(wq); |
d3e9f732 JA |
1162 | clear_bit_unlock(0, &worker->create_state); |
1163 | io_worker_release(worker); | |
685fe7fe JA |
1164 | } |
1165 | ||
1166 | rcu_read_lock(); | |
1167 | for_each_node(node) { | |
1168 | struct io_wqe *wqe = wq->wqes[node]; | |
1169 | ||
1170 | io_wq_for_each_worker(wqe, io_wq_worker_wake, NULL); | |
afcc4015 | 1171 | } |
685fe7fe JA |
1172 | rcu_read_unlock(); |
1173 | io_worker_ref_put(wq); | |
1174 | wait_for_completion(&wq->worker_done); | |
3743c172 Z |
1175 | |
1176 | for_each_node(node) { | |
1177 | spin_lock_irq(&wq->hash->wait.lock); | |
1178 | list_del_init(&wq->wqes[node]->wait.entry); | |
1179 | spin_unlock_irq(&wq->hash->wait.lock); | |
1180 | } | |
685fe7fe JA |
1181 | put_task_struct(wq->task); |
1182 | wq->task = NULL; | |
afcc4015 JA |
1183 | } |
1184 | ||
4fb6ac32 | 1185 | static void io_wq_destroy(struct io_wq *wq) |
771b53d0 | 1186 | { |
3fc50ab5 | 1187 | int node; |
771b53d0 | 1188 | |
43c01fbe JA |
1189 | cpuhp_state_remove_instance_nocalls(io_wq_online, &wq->cpuhp_node); |
1190 | ||
e941894e JA |
1191 | for_each_node(node) { |
1192 | struct io_wqe *wqe = wq->wqes[node]; | |
f5d2d23b JA |
1193 | struct io_cb_cancel_data match = { |
1194 | .fn = io_wq_work_match_all, | |
1195 | .cancel_all = true, | |
1196 | }; | |
1197 | io_wqe_cancel_pending_work(wqe, &match); | |
0e03496d | 1198 | free_cpumask_var(wqe->cpu_mask); |
e941894e JA |
1199 | kfree(wqe); |
1200 | } | |
e941894e | 1201 | io_wq_put_hash(wq->hash); |
771b53d0 | 1202 | kfree(wq); |
4fb6ac32 JA |
1203 | } |
1204 | ||
afcc4015 JA |
1205 | void io_wq_put_and_exit(struct io_wq *wq) |
1206 | { | |
17a91051 PB |
1207 | WARN_ON_ONCE(!test_bit(IO_WQ_BIT_EXIT, &wq->state)); |
1208 | ||
685fe7fe | 1209 | io_wq_exit_workers(wq); |
382cb030 | 1210 | io_wq_destroy(wq); |
afcc4015 JA |
1211 | } |
1212 | ||
0e03496d JA |
1213 | struct online_data { |
1214 | unsigned int cpu; | |
1215 | bool online; | |
1216 | }; | |
1217 | ||
43c01fbe JA |
1218 | static bool io_wq_worker_affinity(struct io_worker *worker, void *data) |
1219 | { | |
0e03496d | 1220 | struct online_data *od = data; |
e0051d7d | 1221 | |
0e03496d JA |
1222 | if (od->online) |
1223 | cpumask_set_cpu(od->cpu, worker->wqe->cpu_mask); | |
1224 | else | |
1225 | cpumask_clear_cpu(od->cpu, worker->wqe->cpu_mask); | |
43c01fbe JA |
1226 | return false; |
1227 | } | |
1228 | ||
0e03496d | 1229 | static int __io_wq_cpu_online(struct io_wq *wq, unsigned int cpu, bool online) |
43c01fbe | 1230 | { |
0e03496d JA |
1231 | struct online_data od = { |
1232 | .cpu = cpu, | |
1233 | .online = online | |
1234 | }; | |
43c01fbe JA |
1235 | int i; |
1236 | ||
1237 | rcu_read_lock(); | |
1238 | for_each_node(i) | |
0e03496d | 1239 | io_wq_for_each_worker(wq->wqes[i], io_wq_worker_affinity, &od); |
43c01fbe JA |
1240 | rcu_read_unlock(); |
1241 | return 0; | |
1242 | } | |
1243 | ||
0e03496d JA |
1244 | static int io_wq_cpu_online(unsigned int cpu, struct hlist_node *node) |
1245 | { | |
1246 | struct io_wq *wq = hlist_entry_safe(node, struct io_wq, cpuhp_node); | |
1247 | ||
1248 | return __io_wq_cpu_online(wq, cpu, true); | |
1249 | } | |
1250 | ||
1251 | static int io_wq_cpu_offline(unsigned int cpu, struct hlist_node *node) | |
1252 | { | |
1253 | struct io_wq *wq = hlist_entry_safe(node, struct io_wq, cpuhp_node); | |
1254 | ||
1255 | return __io_wq_cpu_online(wq, cpu, false); | |
1256 | } | |
1257 | ||
fe76421d JA |
1258 | int io_wq_cpu_affinity(struct io_wq *wq, cpumask_var_t mask) |
1259 | { | |
1260 | int i; | |
1261 | ||
1262 | rcu_read_lock(); | |
1263 | for_each_node(i) { | |
1264 | struct io_wqe *wqe = wq->wqes[i]; | |
1265 | ||
1266 | if (mask) | |
1267 | cpumask_copy(wqe->cpu_mask, mask); | |
1268 | else | |
1269 | cpumask_copy(wqe->cpu_mask, cpumask_of_node(i)); | |
1270 | } | |
1271 | rcu_read_unlock(); | |
1272 | return 0; | |
1273 | } | |
1274 | ||
2e480058 JA |
1275 | /* |
1276 | * Set max number of unbounded workers, returns old value. If new_count is 0, | |
1277 | * then just return the old value. | |
1278 | */ | |
1279 | int io_wq_max_workers(struct io_wq *wq, int *new_count) | |
1280 | { | |
1281 | int i, node, prev = 0; | |
1282 | ||
dd47c104 ES |
1283 | BUILD_BUG_ON((int) IO_WQ_ACCT_BOUND != (int) IO_WQ_BOUND); |
1284 | BUILD_BUG_ON((int) IO_WQ_ACCT_UNBOUND != (int) IO_WQ_UNBOUND); | |
1285 | BUILD_BUG_ON((int) IO_WQ_ACCT_NR != 2); | |
1286 | ||
2e480058 JA |
1287 | for (i = 0; i < 2; i++) { |
1288 | if (new_count[i] > task_rlimit(current, RLIMIT_NPROC)) | |
1289 | new_count[i] = task_rlimit(current, RLIMIT_NPROC); | |
1290 | } | |
1291 | ||
1292 | rcu_read_lock(); | |
1293 | for_each_node(node) { | |
bc369921 | 1294 | struct io_wqe *wqe = wq->wqes[node]; |
2e480058 JA |
1295 | struct io_wqe_acct *acct; |
1296 | ||
bc369921 | 1297 | raw_spin_lock(&wqe->lock); |
f95dc207 | 1298 | for (i = 0; i < IO_WQ_ACCT_NR; i++) { |
bc369921 | 1299 | acct = &wqe->acct[i]; |
2e480058 JA |
1300 | prev = max_t(int, acct->max_workers, prev); |
1301 | if (new_count[i]) | |
1302 | acct->max_workers = new_count[i]; | |
1303 | new_count[i] = prev; | |
1304 | } | |
bc369921 | 1305 | raw_spin_unlock(&wqe->lock); |
2e480058 JA |
1306 | } |
1307 | rcu_read_unlock(); | |
1308 | return 0; | |
1309 | } | |
1310 | ||
43c01fbe JA |
1311 | static __init int io_wq_init(void) |
1312 | { | |
1313 | int ret; | |
1314 | ||
1315 | ret = cpuhp_setup_state_multi(CPUHP_AP_ONLINE_DYN, "io-wq/online", | |
0e03496d | 1316 | io_wq_cpu_online, io_wq_cpu_offline); |
43c01fbe JA |
1317 | if (ret < 0) |
1318 | return ret; | |
1319 | io_wq_online = ret; | |
1320 | return 0; | |
1321 | } | |
1322 | subsys_initcall(io_wq_init); |