xref: /qemu/block/linux-aio.c (revision 526d7984)
1 /*
2  * Linux native AIO support.
3  *
4  * Copyright (C) 2009 IBM, Corp.
5  * Copyright (C) 2009 Red Hat, Inc.
6  *
7  * This work is licensed under the terms of the GNU GPL, version 2 or later.
8  * See the COPYING file in the top-level directory.
9  */
10 #include "qemu/osdep.h"
11 #include "qemu-common.h"
12 #include "block/aio.h"
13 #include "qemu/queue.h"
14 #include "block/block.h"
15 #include "block/raw-aio.h"
16 #include "qemu/event_notifier.h"
17 #include "qemu/coroutine.h"
18 #include "qapi/error.h"
19 
20 #include <libaio.h>
21 
22 /*
23  * Queue size (per-device).
24  *
25  * XXX: eventually we need to communicate this to the guest and/or make it
26  *      tunable by the guest.  If we get more outstanding requests at a time
27  *      than this we will get EAGAIN from io_submit which is communicated to
28  *      the guest as an I/O error.
29  */
30 #define MAX_EVENTS 128
31 
32 struct qemu_laiocb {
33     BlockAIOCB common;
34     Coroutine *co;
35     LinuxAioState *ctx;
36     struct iocb iocb;
37     ssize_t ret;
38     size_t nbytes;
39     QEMUIOVector *qiov;
40     bool is_read;
41     QSIMPLEQ_ENTRY(qemu_laiocb) next;
42 };
43 
44 typedef struct {
45     int plugged;
46     unsigned int in_queue;
47     unsigned int in_flight;
48     bool blocked;
49     QSIMPLEQ_HEAD(, qemu_laiocb) pending;
50 } LaioQueue;
51 
52 struct LinuxAioState {
53     AioContext *aio_context;
54 
55     io_context_t ctx;
56     EventNotifier e;
57 
58     /* io queue for submit at batch.  Protected by AioContext lock. */
59     LaioQueue io_q;
60 
61     /* I/O completion processing.  Only runs in I/O thread.  */
62     QEMUBH *completion_bh;
63     int event_idx;
64     int event_max;
65 };
66 
67 static void ioq_submit(LinuxAioState *s);
68 
69 static inline ssize_t io_event_ret(struct io_event *ev)
70 {
71     return (ssize_t)(((uint64_t)ev->res2 << 32) | ev->res);
72 }
73 
74 /*
75  * Completes an AIO request (calls the callback and frees the ACB).
76  */
77 static void qemu_laio_process_completion(struct qemu_laiocb *laiocb)
78 {
79     int ret;
80 
81     ret = laiocb->ret;
82     if (ret != -ECANCELED) {
83         if (ret == laiocb->nbytes) {
84             ret = 0;
85         } else if (ret >= 0) {
86             /* Short reads mean EOF, pad with zeros. */
87             if (laiocb->is_read) {
88                 qemu_iovec_memset(laiocb->qiov, ret, 0,
89                     laiocb->qiov->size - ret);
90             } else {
91                 ret = -ENOSPC;
92             }
93         }
94     }
95 
96     laiocb->ret = ret;
97     if (laiocb->co) {
98         /* If the coroutine is already entered it must be in ioq_submit() and
99          * will notice laio->ret has been filled in when it eventually runs
100          * later.  Coroutines cannot be entered recursively so avoid doing
101          * that!
102          */
103         if (!qemu_coroutine_entered(laiocb->co)) {
104             aio_co_wake(laiocb->co);
105         }
106     } else {
107         laiocb->common.cb(laiocb->common.opaque, ret);
108         qemu_aio_unref(laiocb);
109     }
110 }
111 
112 /**
113  * aio_ring buffer which is shared between userspace and kernel.
114  *
115  * This copied from linux/fs/aio.c, common header does not exist
116  * but AIO exists for ages so we assume ABI is stable.
117  */
118 struct aio_ring {
119     unsigned    id;    /* kernel internal index number */
120     unsigned    nr;    /* number of io_events */
121     unsigned    head;  /* Written to by userland or by kernel. */
122     unsigned    tail;
123 
124     unsigned    magic;
125     unsigned    compat_features;
126     unsigned    incompat_features;
127     unsigned    header_length;  /* size of aio_ring */
128 
129     struct io_event io_events[0];
130 };
131 
132 /**
133  * io_getevents_peek:
134  * @ctx: AIO context
135  * @events: pointer on events array, output value
136 
137  * Returns the number of completed events and sets a pointer
138  * on events array.  This function does not update the internal
139  * ring buffer, only reads head and tail.  When @events has been
140  * processed io_getevents_commit() must be called.
141  */
142 static inline unsigned int io_getevents_peek(io_context_t ctx,
143                                              struct io_event **events)
144 {
145     struct aio_ring *ring = (struct aio_ring *)ctx;
146     unsigned int head = ring->head, tail = ring->tail;
147     unsigned int nr;
148 
149     nr = tail >= head ? tail - head : ring->nr - head;
150     *events = ring->io_events + head;
151     /* To avoid speculative loads of s->events[i] before observing tail.
152        Paired with smp_wmb() inside linux/fs/aio.c: aio_complete(). */
153     smp_rmb();
154 
155     return nr;
156 }
157 
158 /**
159  * io_getevents_commit:
160  * @ctx: AIO context
161  * @nr: the number of events on which head should be advanced
162  *
163  * Advances head of a ring buffer.
164  */
165 static inline void io_getevents_commit(io_context_t ctx, unsigned int nr)
166 {
167     struct aio_ring *ring = (struct aio_ring *)ctx;
168 
169     if (nr) {
170         ring->head = (ring->head + nr) % ring->nr;
171     }
172 }
173 
174 /**
175  * io_getevents_advance_and_peek:
176  * @ctx: AIO context
177  * @events: pointer on events array, output value
178  * @nr: the number of events on which head should be advanced
179  *
180  * Advances head of a ring buffer and returns number of elements left.
181  */
182 static inline unsigned int
183 io_getevents_advance_and_peek(io_context_t ctx,
184                               struct io_event **events,
185                               unsigned int nr)
186 {
187     io_getevents_commit(ctx, nr);
188     return io_getevents_peek(ctx, events);
189 }
190 
191 /**
192  * qemu_laio_process_completions:
193  * @s: AIO state
194  *
195  * Fetches completed I/O requests and invokes their callbacks.
196  *
197  * The function is somewhat tricky because it supports nested event loops, for
198  * example when a request callback invokes aio_poll().  In order to do this,
199  * indices are kept in LinuxAioState.  Function schedules BH completion so it
200  * can be called again in a nested event loop.  When there are no events left
201  * to complete the BH is being canceled.
202  */
203 static void qemu_laio_process_completions(LinuxAioState *s)
204 {
205     struct io_event *events;
206 
207     /* Reschedule so nested event loops see currently pending completions */
208     qemu_bh_schedule(s->completion_bh);
209 
210     while ((s->event_max = io_getevents_advance_and_peek(s->ctx, &events,
211                                                          s->event_idx))) {
212         for (s->event_idx = 0; s->event_idx < s->event_max; ) {
213             struct iocb *iocb = events[s->event_idx].obj;
214             struct qemu_laiocb *laiocb =
215                 container_of(iocb, struct qemu_laiocb, iocb);
216 
217             laiocb->ret = io_event_ret(&events[s->event_idx]);
218 
219             /* Change counters one-by-one because we can be nested. */
220             s->io_q.in_flight--;
221             s->event_idx++;
222             qemu_laio_process_completion(laiocb);
223         }
224     }
225 
226     qemu_bh_cancel(s->completion_bh);
227 
228     /* If we are nested we have to notify the level above that we are done
229      * by setting event_max to zero, upper level will then jump out of it's
230      * own `for` loop.  If we are the last all counters droped to zero. */
231     s->event_max = 0;
232     s->event_idx = 0;
233 }
234 
235 static void qemu_laio_process_completions_and_submit(LinuxAioState *s)
236 {
237     aio_context_acquire(s->aio_context);
238     qemu_laio_process_completions(s);
239 
240     if (!s->io_q.plugged && !QSIMPLEQ_EMPTY(&s->io_q.pending)) {
241         ioq_submit(s);
242     }
243     aio_context_release(s->aio_context);
244 }
245 
246 static void qemu_laio_completion_bh(void *opaque)
247 {
248     LinuxAioState *s = opaque;
249 
250     qemu_laio_process_completions_and_submit(s);
251 }
252 
253 static void qemu_laio_completion_cb(EventNotifier *e)
254 {
255     LinuxAioState *s = container_of(e, LinuxAioState, e);
256 
257     if (event_notifier_test_and_clear(&s->e)) {
258         qemu_laio_process_completions_and_submit(s);
259     }
260 }
261 
262 static bool qemu_laio_poll_cb(void *opaque)
263 {
264     EventNotifier *e = opaque;
265     LinuxAioState *s = container_of(e, LinuxAioState, e);
266     struct io_event *events;
267 
268     if (!io_getevents_peek(s->ctx, &events)) {
269         return false;
270     }
271 
272     qemu_laio_process_completions_and_submit(s);
273     return true;
274 }
275 
276 static void laio_cancel(BlockAIOCB *blockacb)
277 {
278     struct qemu_laiocb *laiocb = (struct qemu_laiocb *)blockacb;
279     struct io_event event;
280     int ret;
281 
282     if (laiocb->ret != -EINPROGRESS) {
283         return;
284     }
285     ret = io_cancel(laiocb->ctx->ctx, &laiocb->iocb, &event);
286     laiocb->ret = -ECANCELED;
287     if (ret != 0) {
288         /* iocb is not cancelled, cb will be called by the event loop later */
289         return;
290     }
291 
292     laiocb->common.cb(laiocb->common.opaque, laiocb->ret);
293 }
294 
295 static const AIOCBInfo laio_aiocb_info = {
296     .aiocb_size         = sizeof(struct qemu_laiocb),
297     .cancel_async       = laio_cancel,
298 };
299 
300 static void ioq_init(LaioQueue *io_q)
301 {
302     QSIMPLEQ_INIT(&io_q->pending);
303     io_q->plugged = 0;
304     io_q->in_queue = 0;
305     io_q->in_flight = 0;
306     io_q->blocked = false;
307 }
308 
309 static void ioq_submit(LinuxAioState *s)
310 {
311     int ret, len;
312     struct qemu_laiocb *aiocb;
313     struct iocb *iocbs[MAX_EVENTS];
314     QSIMPLEQ_HEAD(, qemu_laiocb) completed;
315 
316     do {
317         if (s->io_q.in_flight >= MAX_EVENTS) {
318             break;
319         }
320         len = 0;
321         QSIMPLEQ_FOREACH(aiocb, &s->io_q.pending, next) {
322             iocbs[len++] = &aiocb->iocb;
323             if (s->io_q.in_flight + len >= MAX_EVENTS) {
324                 break;
325             }
326         }
327 
328         ret = io_submit(s->ctx, len, iocbs);
329         if (ret == -EAGAIN) {
330             break;
331         }
332         if (ret < 0) {
333             /* Fail the first request, retry the rest */
334             aiocb = QSIMPLEQ_FIRST(&s->io_q.pending);
335             QSIMPLEQ_REMOVE_HEAD(&s->io_q.pending, next);
336             s->io_q.in_queue--;
337             aiocb->ret = ret;
338             qemu_laio_process_completion(aiocb);
339             continue;
340         }
341 
342         s->io_q.in_flight += ret;
343         s->io_q.in_queue  -= ret;
344         aiocb = container_of(iocbs[ret - 1], struct qemu_laiocb, iocb);
345         QSIMPLEQ_SPLIT_AFTER(&s->io_q.pending, aiocb, next, &completed);
346     } while (ret == len && !QSIMPLEQ_EMPTY(&s->io_q.pending));
347     s->io_q.blocked = (s->io_q.in_queue > 0);
348 
349     if (s->io_q.in_flight) {
350         /* We can try to complete something just right away if there are
351          * still requests in-flight. */
352         qemu_laio_process_completions(s);
353         /*
354          * Even we have completed everything (in_flight == 0), the queue can
355          * have still pended requests (in_queue > 0).  We do not attempt to
356          * repeat submission to avoid IO hang.  The reason is simple: s->e is
357          * still set and completion callback will be called shortly and all
358          * pended requests will be submitted from there.
359          */
360     }
361 }
362 
363 void laio_io_plug(BlockDriverState *bs, LinuxAioState *s)
364 {
365     s->io_q.plugged++;
366 }
367 
368 void laio_io_unplug(BlockDriverState *bs, LinuxAioState *s)
369 {
370     assert(s->io_q.plugged);
371     if (--s->io_q.plugged == 0 &&
372         !s->io_q.blocked && !QSIMPLEQ_EMPTY(&s->io_q.pending)) {
373         ioq_submit(s);
374     }
375 }
376 
377 static int laio_do_submit(int fd, struct qemu_laiocb *laiocb, off_t offset,
378                           int type)
379 {
380     LinuxAioState *s = laiocb->ctx;
381     struct iocb *iocbs = &laiocb->iocb;
382     QEMUIOVector *qiov = laiocb->qiov;
383 
384     switch (type) {
385     case QEMU_AIO_WRITE:
386         io_prep_pwritev(iocbs, fd, qiov->iov, qiov->niov, offset);
387         break;
388     case QEMU_AIO_READ:
389         io_prep_preadv(iocbs, fd, qiov->iov, qiov->niov, offset);
390         break;
391     /* Currently Linux kernel does not support other operations */
392     default:
393         fprintf(stderr, "%s: invalid AIO request type 0x%x.\n",
394                         __func__, type);
395         return -EIO;
396     }
397     io_set_eventfd(&laiocb->iocb, event_notifier_get_fd(&s->e));
398 
399     QSIMPLEQ_INSERT_TAIL(&s->io_q.pending, laiocb, next);
400     s->io_q.in_queue++;
401     if (!s->io_q.blocked &&
402         (!s->io_q.plugged ||
403          s->io_q.in_flight + s->io_q.in_queue >= MAX_EVENTS)) {
404         ioq_submit(s);
405     }
406 
407     return 0;
408 }
409 
410 int coroutine_fn laio_co_submit(BlockDriverState *bs, LinuxAioState *s, int fd,
411                                 uint64_t offset, QEMUIOVector *qiov, int type)
412 {
413     int ret;
414     struct qemu_laiocb laiocb = {
415         .co         = qemu_coroutine_self(),
416         .nbytes     = qiov->size,
417         .ctx        = s,
418         .ret        = -EINPROGRESS,
419         .is_read    = (type == QEMU_AIO_READ),
420         .qiov       = qiov,
421     };
422 
423     ret = laio_do_submit(fd, &laiocb, offset, type);
424     if (ret < 0) {
425         return ret;
426     }
427 
428     if (laiocb.ret == -EINPROGRESS) {
429         qemu_coroutine_yield();
430     }
431     return laiocb.ret;
432 }
433 
434 BlockAIOCB *laio_submit(BlockDriverState *bs, LinuxAioState *s, int fd,
435         int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
436         BlockCompletionFunc *cb, void *opaque, int type)
437 {
438     struct qemu_laiocb *laiocb;
439     off_t offset = sector_num * BDRV_SECTOR_SIZE;
440     int ret;
441 
442     laiocb = qemu_aio_get(&laio_aiocb_info, bs, cb, opaque);
443     laiocb->nbytes = nb_sectors * BDRV_SECTOR_SIZE;
444     laiocb->ctx = s;
445     laiocb->ret = -EINPROGRESS;
446     laiocb->is_read = (type == QEMU_AIO_READ);
447     laiocb->qiov = qiov;
448 
449     ret = laio_do_submit(fd, laiocb, offset, type);
450     if (ret < 0) {
451         qemu_aio_unref(laiocb);
452         return NULL;
453     }
454 
455     return &laiocb->common;
456 }
457 
458 void laio_detach_aio_context(LinuxAioState *s, AioContext *old_context)
459 {
460     aio_set_event_notifier(old_context, &s->e, false, NULL, NULL);
461     qemu_bh_delete(s->completion_bh);
462     s->aio_context = NULL;
463 }
464 
465 void laio_attach_aio_context(LinuxAioState *s, AioContext *new_context)
466 {
467     s->aio_context = new_context;
468     s->completion_bh = aio_bh_new(new_context, qemu_laio_completion_bh, s);
469     aio_set_event_notifier(new_context, &s->e, false,
470                            qemu_laio_completion_cb,
471                            qemu_laio_poll_cb);
472 }
473 
474 LinuxAioState *laio_init(Error **errp)
475 {
476     int rc;
477     LinuxAioState *s;
478 
479     s = g_malloc0(sizeof(*s));
480     rc = event_notifier_init(&s->e, false);
481     if (rc < 0) {
482         error_setg_errno(errp, -rc, "failed to to initialize event notifier");
483         goto out_free_state;
484     }
485 
486     rc = io_setup(MAX_EVENTS, &s->ctx);
487     if (rc < 0) {
488         error_setg_errno(errp, -rc, "failed to create linux AIO context");
489         goto out_close_efd;
490     }
491 
492     ioq_init(&s->io_q);
493 
494     return s;
495 
496 out_close_efd:
497     event_notifier_cleanup(&s->e);
498 out_free_state:
499     g_free(s);
500     return NULL;
501 }
502 
503 void laio_cleanup(LinuxAioState *s)
504 {
505     event_notifier_cleanup(&s->e);
506 
507     if (io_destroy(s->ctx) != 0) {
508         fprintf(stderr, "%s: destroy AIO context %p failed\n",
509                         __func__, &s->ctx);
510     }
511     g_free(s);
512 }
513