]>
Commit | Line | Data |
---|---|---|
d354c7ec PB |
1 | /* |
2 | * QEMU block layer thread pool | |
3 | * | |
4 | * Copyright IBM, Corp. 2008 | |
5 | * Copyright Red Hat, Inc. 2012 | |
6 | * | |
7 | * Authors: | |
8 | * Anthony Liguori <aliguori@us.ibm.com> | |
9 | * Paolo Bonzini <pbonzini@redhat.com> | |
10 | * | |
11 | * This work is licensed under the terms of the GNU GPL, version 2. See | |
12 | * the COPYING file in the top-level directory. | |
13 | * | |
14 | * Contributions after 2012-01-13 are licensed under the terms of the | |
15 | * GNU GPL, version 2 or (at your option) any later version. | |
16 | */ | |
d38ea87a | 17 | #include "qemu/osdep.h" |
1de7afc9 PB |
18 | #include "qemu/queue.h" |
19 | #include "qemu/thread.h" | |
10817bf0 | 20 | #include "qemu/coroutine.h" |
c2b38b27 | 21 | #include "trace.h" |
737e150e | 22 | #include "block/thread-pool.h" |
6a1751b7 | 23 | #include "qemu/main-loop.h" |
d354c7ec | 24 | |
b811203c | 25 | static void do_spawn_thread(ThreadPool *pool); |
d354c7ec PB |
26 | |
27 | typedef struct ThreadPoolElement ThreadPoolElement; | |
28 | ||
29 | enum ThreadState { | |
30 | THREAD_QUEUED, | |
31 | THREAD_ACTIVE, | |
32 | THREAD_DONE, | |
d354c7ec PB |
33 | }; |
34 | ||
35 | struct ThreadPoolElement { | |
7c84b1b8 | 36 | BlockAIOCB common; |
b811203c | 37 | ThreadPool *pool; |
d354c7ec PB |
38 | ThreadPoolFunc *func; |
39 | void *arg; | |
19d092cf PB |
40 | |
41 | /* Moving state out of THREAD_QUEUED is protected by lock. After | |
42 | * that, only the worker thread can write to it. Reads and writes | |
43 | * of state and ret are ordered with memory barriers. | |
44 | */ | |
d354c7ec PB |
45 | enum ThreadState state; |
46 | int ret; | |
47 | ||
48 | /* Access to this list is protected by lock. */ | |
49 | QTAILQ_ENTRY(ThreadPoolElement) reqs; | |
50 | ||
0fdb7311 | 51 | /* This list is only written by the thread pool's mother thread. */ |
d354c7ec PB |
52 | QLIST_ENTRY(ThreadPoolElement) all; |
53 | }; | |
54 | ||
b811203c | 55 | struct ThreadPool { |
f7311ccc | 56 | AioContext *ctx; |
c2e50e3d | 57 | QEMUBH *completion_bh; |
b811203c | 58 | QemuMutex lock; |
f7311ccc | 59 | QemuCond worker_stopped; |
900fa208 | 60 | QemuCond request_cond; |
b811203c SH |
61 | QEMUBH *new_thread_bh; |
62 | ||
63 | /* The following variables are only accessed from one AioContext. */ | |
64 | QLIST_HEAD(, ThreadPoolElement) head; | |
65 | ||
66 | /* The following variables are protected by lock. */ | |
67 | QTAILQ_HEAD(, ThreadPoolElement) request_list; | |
68 | int cur_threads; | |
69 | int idle_threads; | |
70 | int new_threads; /* backlog of threads we need to create */ | |
71 | int pending_threads; /* threads created but not running yet */ | |
71ad4713 NSJ |
72 | int min_threads; |
73 | int max_threads; | |
b811203c SH |
74 | }; |
75 | ||
b811203c | 76 | static void *worker_thread(void *opaque) |
d354c7ec | 77 | { |
b811203c SH |
78 | ThreadPool *pool = opaque; |
79 | ||
80 | qemu_mutex_lock(&pool->lock); | |
81 | pool->pending_threads--; | |
82 | do_spawn_thread(pool); | |
d354c7ec | 83 | |
232e9255 | 84 | while (pool->cur_threads <= pool->max_threads) { |
d354c7ec PB |
85 | ThreadPoolElement *req; |
86 | int ret; | |
87 | ||
900fa208 | 88 | if (QTAILQ_EMPTY(&pool->request_list)) { |
b811203c | 89 | pool->idle_threads++; |
900fa208 | 90 | ret = qemu_cond_timedwait(&pool->request_cond, &pool->lock, 10000); |
b811203c | 91 | pool->idle_threads--; |
900fa208 PB |
92 | if (ret == 0 && |
93 | QTAILQ_EMPTY(&pool->request_list) && | |
94 | pool->cur_threads > pool->min_threads) { | |
95 | /* Timed out + no work to do + no need for warm threads = exit. */ | |
96 | break; | |
97 | } | |
98 | /* | |
99 | * Even if there was some work to do, check if there aren't | |
100 | * too many worker threads before picking it up. | |
101 | */ | |
102 | continue; | |
d354c7ec PB |
103 | } |
104 | ||
b811203c SH |
105 | req = QTAILQ_FIRST(&pool->request_list); |
106 | QTAILQ_REMOVE(&pool->request_list, req, reqs); | |
d354c7ec | 107 | req->state = THREAD_ACTIVE; |
b811203c | 108 | qemu_mutex_unlock(&pool->lock); |
d354c7ec PB |
109 | |
110 | ret = req->func(req->arg); | |
111 | ||
d354c7ec | 112 | req->ret = ret; |
19d092cf PB |
113 | /* Write ret before state. */ |
114 | smp_wmb(); | |
115 | req->state = THREAD_DONE; | |
116 | ||
c2e50e3d | 117 | qemu_bh_schedule(pool->completion_bh); |
3c7b72dd | 118 | qemu_mutex_lock(&pool->lock); |
d354c7ec PB |
119 | } |
120 | ||
b811203c | 121 | pool->cur_threads--; |
f7311ccc | 122 | qemu_cond_signal(&pool->worker_stopped); |
900fa208 PB |
123 | |
124 | /* | |
125 | * Wake up another thread, in case we got a wakeup but decided | |
126 | * to exit due to pool->cur_threads > pool->max_threads. | |
127 | */ | |
128 | qemu_cond_signal(&pool->request_cond); | |
f4f71363 | 129 | qemu_mutex_unlock(&pool->lock); |
d354c7ec PB |
130 | return NULL; |
131 | } | |
132 | ||
b811203c | 133 | static void do_spawn_thread(ThreadPool *pool) |
d354c7ec PB |
134 | { |
135 | QemuThread t; | |
136 | ||
137 | /* Runs with lock taken. */ | |
b811203c | 138 | if (!pool->new_threads) { |
d354c7ec PB |
139 | return; |
140 | } | |
141 | ||
b811203c SH |
142 | pool->new_threads--; |
143 | pool->pending_threads++; | |
d354c7ec | 144 | |
4900116e | 145 | qemu_thread_create(&t, "worker", worker_thread, pool, QEMU_THREAD_DETACHED); |
d354c7ec PB |
146 | } |
147 | ||
148 | static void spawn_thread_bh_fn(void *opaque) | |
149 | { | |
b811203c SH |
150 | ThreadPool *pool = opaque; |
151 | ||
152 | qemu_mutex_lock(&pool->lock); | |
153 | do_spawn_thread(pool); | |
154 | qemu_mutex_unlock(&pool->lock); | |
d354c7ec PB |
155 | } |
156 | ||
b811203c | 157 | static void spawn_thread(ThreadPool *pool) |
d354c7ec | 158 | { |
b811203c SH |
159 | pool->cur_threads++; |
160 | pool->new_threads++; | |
d354c7ec PB |
161 | /* If there are threads being created, they will spawn new workers, so |
162 | * we don't spend time creating many threads in a loop holding a mutex or | |
163 | * starving the current vcpu. | |
164 | * | |
165 | * If there are no idle threads, ask the main thread to create one, so we | |
166 | * inherit the correct affinity instead of the vcpu affinity. | |
167 | */ | |
b811203c SH |
168 | if (!pool->pending_threads) { |
169 | qemu_bh_schedule(pool->new_thread_bh); | |
d354c7ec PB |
170 | } |
171 | } | |
172 | ||
c2e50e3d | 173 | static void thread_pool_completion_bh(void *opaque) |
d354c7ec | 174 | { |
c2e50e3d | 175 | ThreadPool *pool = opaque; |
d354c7ec PB |
176 | ThreadPoolElement *elem, *next; |
177 | ||
d354c7ec | 178 | restart: |
b811203c | 179 | QLIST_FOREACH_SAFE(elem, &pool->head, all, next) { |
3391f5e5 | 180 | if (elem->state != THREAD_DONE) { |
d354c7ec PB |
181 | continue; |
182 | } | |
1faa5bb7 SH |
183 | |
184 | trace_thread_pool_complete(pool, elem, elem->common.opaque, | |
185 | elem->ret); | |
186 | QLIST_REMOVE(elem, all); | |
187 | ||
188 | if (elem->common.cb) { | |
19d092cf PB |
189 | /* Read state before ret. */ |
190 | smp_rmb(); | |
3c80ca15 SH |
191 | |
192 | /* Schedule ourselves in case elem->common.cb() calls aio_poll() to | |
193 | * wait for another request that completed at the same time. | |
194 | */ | |
195 | qemu_bh_schedule(pool->completion_bh); | |
196 | ||
19d092cf | 197 | elem->common.cb(elem->common.opaque, elem->ret); |
b7a745dc PL |
198 | |
199 | /* We can safely cancel the completion_bh here regardless of someone | |
200 | * else having scheduled it meanwhile because we reenter the | |
201 | * completion function anyway (goto restart). | |
202 | */ | |
203 | qemu_bh_cancel(pool->completion_bh); | |
204 | ||
8007429a | 205 | qemu_aio_unref(elem); |
d354c7ec PB |
206 | goto restart; |
207 | } else { | |
8007429a | 208 | qemu_aio_unref(elem); |
d354c7ec PB |
209 | } |
210 | } | |
211 | } | |
212 | ||
7c84b1b8 | 213 | static void thread_pool_cancel(BlockAIOCB *acb) |
d354c7ec PB |
214 | { |
215 | ThreadPoolElement *elem = (ThreadPoolElement *)acb; | |
b811203c | 216 | ThreadPool *pool = elem->pool; |
d354c7ec PB |
217 | |
218 | trace_thread_pool_cancel(elem, elem->common.opaque); | |
219 | ||
6e8a355d | 220 | QEMU_LOCK_GUARD(&pool->lock); |
900fa208 | 221 | if (elem->state == THREAD_QUEUED) { |
b811203c | 222 | QTAILQ_REMOVE(&pool->request_list, elem, reqs); |
c2e50e3d | 223 | qemu_bh_schedule(pool->completion_bh); |
3391f5e5 FZ |
224 | |
225 | elem->state = THREAD_DONE; | |
226 | elem->ret = -ECANCELED; | |
d354c7ec | 227 | } |
3391f5e5 | 228 | |
3391f5e5 FZ |
229 | } |
230 | ||
d7331bed | 231 | static const AIOCBInfo thread_pool_aiocb_info = { |
d354c7ec | 232 | .aiocb_size = sizeof(ThreadPoolElement), |
3391f5e5 | 233 | .cancel_async = thread_pool_cancel, |
d354c7ec PB |
234 | }; |
235 | ||
aef04fc7 EGE |
236 | BlockAIOCB *thread_pool_submit_aio(ThreadPoolFunc *func, void *arg, |
237 | BlockCompletionFunc *cb, void *opaque) | |
d354c7ec PB |
238 | { |
239 | ThreadPoolElement *req; | |
aef04fc7 EGE |
240 | AioContext *ctx = qemu_get_current_aio_context(); |
241 | ThreadPool *pool = aio_get_thread_pool(ctx); | |
d354c7ec | 242 | |
0fdb7311 EGE |
243 | /* Assert that the thread submitting work is the same running the pool */ |
244 | assert(pool->ctx == qemu_get_current_aio_context()); | |
245 | ||
d7331bed | 246 | req = qemu_aio_get(&thread_pool_aiocb_info, NULL, cb, opaque); |
d354c7ec PB |
247 | req->func = func; |
248 | req->arg = arg; | |
249 | req->state = THREAD_QUEUED; | |
b811203c | 250 | req->pool = pool; |
d354c7ec | 251 | |
b811203c | 252 | QLIST_INSERT_HEAD(&pool->head, req, all); |
d354c7ec | 253 | |
b811203c | 254 | trace_thread_pool_submit(pool, req, arg); |
d354c7ec | 255 | |
b811203c SH |
256 | qemu_mutex_lock(&pool->lock); |
257 | if (pool->idle_threads == 0 && pool->cur_threads < pool->max_threads) { | |
258 | spawn_thread(pool); | |
d354c7ec | 259 | } |
b811203c SH |
260 | QTAILQ_INSERT_TAIL(&pool->request_list, req, reqs); |
261 | qemu_mutex_unlock(&pool->lock); | |
900fa208 | 262 | qemu_cond_signal(&pool->request_cond); |
d354c7ec PB |
263 | return &req->common; |
264 | } | |
265 | ||
266 | typedef struct ThreadPoolCo { | |
267 | Coroutine *co; | |
268 | int ret; | |
269 | } ThreadPoolCo; | |
270 | ||
271 | static void thread_pool_co_cb(void *opaque, int ret) | |
272 | { | |
273 | ThreadPoolCo *co = opaque; | |
274 | ||
275 | co->ret = ret; | |
b9e413dd | 276 | aio_co_wake(co->co); |
d354c7ec PB |
277 | } |
278 | ||
aef04fc7 | 279 | int coroutine_fn thread_pool_submit_co(ThreadPoolFunc *func, void *arg) |
d354c7ec PB |
280 | { |
281 | ThreadPoolCo tpc = { .co = qemu_coroutine_self(), .ret = -EINPROGRESS }; | |
282 | assert(qemu_in_coroutine()); | |
aef04fc7 | 283 | thread_pool_submit_aio(func, arg, thread_pool_co_cb, &tpc); |
d354c7ec PB |
284 | qemu_coroutine_yield(); |
285 | return tpc.ret; | |
286 | } | |
287 | ||
aef04fc7 | 288 | void thread_pool_submit(ThreadPoolFunc *func, void *arg) |
d354c7ec | 289 | { |
aef04fc7 | 290 | thread_pool_submit_aio(func, arg, NULL, NULL); |
d354c7ec PB |
291 | } |
292 | ||
71ad4713 NSJ |
293 | void thread_pool_update_params(ThreadPool *pool, AioContext *ctx) |
294 | { | |
295 | qemu_mutex_lock(&pool->lock); | |
296 | ||
297 | pool->min_threads = ctx->thread_pool_min; | |
298 | pool->max_threads = ctx->thread_pool_max; | |
299 | ||
300 | /* | |
301 | * We either have to: | |
302 | * - Increase the number available of threads until over the min_threads | |
303 | * threshold. | |
900fa208 | 304 | * - Bump the worker threads so that they exit, until under the max_threads |
71ad4713 NSJ |
305 | * threshold. |
306 | * - Do nothing. The current number of threads fall in between the min and | |
307 | * max thresholds. We'll let the pool manage itself. | |
308 | */ | |
309 | for (int i = pool->cur_threads; i < pool->min_threads; i++) { | |
310 | spawn_thread(pool); | |
311 | } | |
312 | ||
313 | for (int i = pool->cur_threads; i > pool->max_threads; i--) { | |
900fa208 | 314 | qemu_cond_signal(&pool->request_cond); |
71ad4713 NSJ |
315 | } |
316 | ||
317 | qemu_mutex_unlock(&pool->lock); | |
318 | } | |
319 | ||
b811203c SH |
320 | static void thread_pool_init_one(ThreadPool *pool, AioContext *ctx) |
321 | { | |
322 | if (!ctx) { | |
323 | ctx = qemu_get_aio_context(); | |
324 | } | |
325 | ||
326 | memset(pool, 0, sizeof(*pool)); | |
f7311ccc | 327 | pool->ctx = ctx; |
c2e50e3d | 328 | pool->completion_bh = aio_bh_new(ctx, thread_pool_completion_bh, pool); |
b811203c | 329 | qemu_mutex_init(&pool->lock); |
f7311ccc | 330 | qemu_cond_init(&pool->worker_stopped); |
900fa208 | 331 | qemu_cond_init(&pool->request_cond); |
b811203c SH |
332 | pool->new_thread_bh = aio_bh_new(ctx, spawn_thread_bh_fn, pool); |
333 | ||
334 | QLIST_INIT(&pool->head); | |
335 | QTAILQ_INIT(&pool->request_list); | |
71ad4713 NSJ |
336 | |
337 | thread_pool_update_params(pool, ctx); | |
b811203c SH |
338 | } |
339 | ||
f7311ccc SH |
340 | ThreadPool *thread_pool_new(AioContext *ctx) |
341 | { | |
342 | ThreadPool *pool = g_new(ThreadPool, 1); | |
343 | thread_pool_init_one(pool, ctx); | |
344 | return pool; | |
345 | } | |
346 | ||
347 | void thread_pool_free(ThreadPool *pool) | |
348 | { | |
349 | if (!pool) { | |
350 | return; | |
351 | } | |
352 | ||
353 | assert(QLIST_EMPTY(&pool->head)); | |
354 | ||
355 | qemu_mutex_lock(&pool->lock); | |
356 | ||
357 | /* Stop new threads from spawning */ | |
358 | qemu_bh_delete(pool->new_thread_bh); | |
359 | pool->cur_threads -= pool->new_threads; | |
360 | pool->new_threads = 0; | |
361 | ||
362 | /* Wait for worker threads to terminate */ | |
232e9255 | 363 | pool->max_threads = 0; |
900fa208 | 364 | qemu_cond_broadcast(&pool->request_cond); |
f7311ccc | 365 | while (pool->cur_threads > 0) { |
f7311ccc SH |
366 | qemu_cond_wait(&pool->worker_stopped, &pool->lock); |
367 | } | |
368 | ||
369 | qemu_mutex_unlock(&pool->lock); | |
370 | ||
c2e50e3d | 371 | qemu_bh_delete(pool->completion_bh); |
900fa208 | 372 | qemu_cond_destroy(&pool->request_cond); |
f7311ccc SH |
373 | qemu_cond_destroy(&pool->worker_stopped); |
374 | qemu_mutex_destroy(&pool->lock); | |
f7311ccc SH |
375 | g_free(pool); |
376 | } |