ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/libev/ev_iouring.c
Revision: 1.26
Committed: Wed Aug 10 16:50:05 2022 UTC (23 months, 1 week ago) by root
Content type: text/plain
Branch: MAIN
Changes since 1.25: +1 -1 lines
Log Message:
silence some of the more annoying warnings

File Contents

# Content
1 /*
2 * libev linux io_uring fd activity backend
3 *
4 * Copyright (c) 2019-2020 Marc Alexander Lehmann <libev@schmorp.de>
5 * All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without modifica-
8 * tion, are permitted provided that the following conditions are met:
9 *
10 * 1. Redistributions of source code must retain the above copyright notice,
11 * this list of conditions and the following disclaimer.
12 *
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR IMPLIED
18 * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MER-
19 * CHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO
20 * EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPE-
21 * CIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
22 * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
23 * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
24 * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTH-
25 * ERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
26 * OF THE POSSIBILITY OF SUCH DAMAGE.
27 *
28 * Alternatively, the contents of this file may be used under the terms of
29 * the GNU General Public License ("GPL") version 2 or any later version,
30 * in which case the provisions of the GPL are applicable instead of
31 * the above. If you wish to allow the use of your version of this file
32 * only under the terms of the GPL and not to allow others to use your
33 * version of this file under the BSD license, indicate your decision
34 * by deleting the provisions above and replace them with the notice
35 * and other provisions required by the GPL. If you do not delete the
36 * provisions above, a recipient may use your version of this file under
37 * either the BSD or the GPL.
38 */
39
40 /*
41 * general notes about linux io_uring:
42 *
43 * a) it's the best interface I have seen so far. on linux.
44 * b) best is not necessarily very good.
45 * c) it's better than the aio mess, doesn't suffer from the fork problems
46 * of linux aio or epoll and so on and so on. and you could do event stuff
47 * without any syscalls. what's not to like?
48 * d) ok, it's vastly more complex, but that's ok, really.
49 * e) why two mmaps instead of one? one would be more space-efficient,
50 * and I can't see what benefit two would have (other than being
51 * somehow resizable/relocatable, but that's apparently not possible).
52 * f) hmm, it's practically undebuggable (gdb can't access the memory, and
53 * the bizarre way structure offsets are communicated makes it hard to
54 * just print the ring buffer heads, even *iff* the memory were visible
55 * in gdb. but then, that's also ok, really.
56 * g) well, you cannot specify a timeout when waiting for events. no,
57 * seriously, the interface doesn't support a timeout. never seen _that_
58 * before. sure, you can use a timerfd, but that's another syscall
59 * you could have avoided. overall, this bizarre omission smells
60 * like a ยต-optimisation by the io_uring author for his personal
61 * applications, to the detriment of everybody else who just wants
62 * an event loop. but, umm, ok, if that's all, it could be worse.
63 * (from what I gather from the author Jens Axboe, it simply didn't
64 * occur to him, and he made good on it by adding an unlimited number
65 * of timeouts later :).
66 * h) initially there was a hardcoded limit of 4096 outstanding events.
67 * later versions not only bump this to 32k, but also can handle
68 * an unlimited amount of events, so this only affects the batch size.
69 * i) unlike linux aio, you *can* register more then the limit
70 * of fd events. while early verisons of io_uring signalled an overflow
71 * and you ended up getting wet. 5.5+ does not do this anymore.
72 * j) but, oh my! it had exactly the same bugs as the linux aio backend,
73 * where some undocumented poll combinations just fail. fortunately,
74 * after finally reaching the author, he was more than willing to fix
75 * this probably in 5.6+.
76 * k) overall, the *API* itself is, I dare to say, not a total trainwreck.
77 * once the bugs ae fixed (probably in 5.6+), it will be without
78 * competition.
79 */
80
81 /* TODO: use internal TIMEOUT */
82 /* TODO: take advantage of single mmap, NODROP etc. */
83 /* TODO: resize cq/sq size independently */
84
85 #include <sys/timerfd.h>
86 #include <sys/mman.h>
87 #include <poll.h>
88 #include <stdint.h>
89
90 #define IOURING_INIT_ENTRIES 32
91
92 /*****************************************************************************/
93 /* syscall wrapdadoop - this section has the raw api/abi definitions */
94
95 #include <linux/fs.h>
96 #include <linux/types.h>
97
98 /* mostly directly taken from the kernel or documentation */
99
100 struct io_uring_sqe
101 {
102 __u8 opcode;
103 __u8 flags;
104 __u16 ioprio;
105 __s32 fd;
106 union {
107 __u64 off;
108 __u64 addr2;
109 };
110 __u64 addr;
111 __u32 len;
112 union {
113 __kernel_rwf_t rw_flags;
114 __u32 fsync_flags;
115 __u16 poll_events;
116 __u32 sync_range_flags;
117 __u32 msg_flags;
118 __u32 timeout_flags;
119 __u32 accept_flags;
120 __u32 cancel_flags;
121 __u32 open_flags;
122 __u32 statx_flags;
123 __u32 fadvise_advice;
124 };
125 __u64 user_data;
126 union {
127 __u16 buf_index;
128 __u16 personality;
129 __u64 __pad2[3];
130 };
131 };
132
133 struct io_uring_cqe
134 {
135 __u64 user_data;
136 __s32 res;
137 __u32 flags;
138 };
139
140 struct io_sqring_offsets
141 {
142 __u32 head;
143 __u32 tail;
144 __u32 ring_mask;
145 __u32 ring_entries;
146 __u32 flags;
147 __u32 dropped;
148 __u32 array;
149 __u32 resv1;
150 __u64 resv2;
151 };
152
153 struct io_cqring_offsets
154 {
155 __u32 head;
156 __u32 tail;
157 __u32 ring_mask;
158 __u32 ring_entries;
159 __u32 overflow;
160 __u32 cqes;
161 __u64 resv[2];
162 };
163
164 struct io_uring_params
165 {
166 __u32 sq_entries;
167 __u32 cq_entries;
168 __u32 flags;
169 __u32 sq_thread_cpu;
170 __u32 sq_thread_idle;
171 __u32 features;
172 __u32 resv[4];
173 struct io_sqring_offsets sq_off;
174 struct io_cqring_offsets cq_off;
175 };
176
177 #define IORING_FEAT_SINGLE_MMAP 0x00000001
178 #define IORING_FEAT_NODROP 0x00000002
179 #define IORING_FEAT_SUBMIT_STABLE 0x00000004
180
181 #define IORING_SETUP_CQSIZE 0x00000008
182 #define IORING_SETUP_CLAMP 0x00000010
183
184 #define IORING_OP_POLL_ADD 6
185 #define IORING_OP_POLL_REMOVE 7
186 #define IORING_OP_TIMEOUT 11
187 #define IORING_OP_TIMEOUT_REMOVE 12
188
189 #define IORING_REGISTER_EVENTFD 4
190 #define IORING_REGISTER_EVENTFD_ASYNC 7
191 #define IORING_REGISTER_PROBE 8
192
193 #define IO_URING_OP_SUPPORTED 1
194
195 struct io_uring_probe_op {
196 __u8 op;
197 __u8 resv;
198 __u16 flags;
199 __u32 resv2;
200 };
201
202 struct io_uring_probe
203 {
204 __u8 last_op;
205 __u8 ops_len;
206 __u16 resv;
207 __u32 resv2[3];
208 struct io_uring_probe_op ops[0];
209 };
210
211 /* relative or absolute, reference clock is CLOCK_MONOTONIC */
212 struct iouring_kernel_timespec
213 {
214 int64_t tv_sec;
215 long long tv_nsec;
216 };
217
218 #define IORING_TIMEOUT_ABS 0x00000001
219
220 #define IORING_ENTER_GETEVENTS 0x01
221
222 #define IORING_OFF_SQ_RING 0x00000000ULL
223 #define IORING_OFF_SQES 0x10000000ULL
224
225 #define IORING_FEAT_SINGLE_MMAP 0x00000001
226 #define IORING_FEAT_NODROP 0x00000002
227 #define IORING_FEAT_SUBMIT_STABLE 0x00000004
228
229 inline_size
230 int
231 evsys_io_uring_setup (unsigned entries, struct io_uring_params *params)
232 {
233 return ev_syscall2 (SYS_io_uring_setup, entries, params);
234 }
235
236 inline_size
237 int
238 evsys_io_uring_enter (int fd, unsigned to_submit, unsigned min_complete, unsigned flags, const sigset_t *sig, size_t sigsz)
239 {
240 return ev_syscall6 (SYS_io_uring_enter, fd, to_submit, min_complete, flags, sig, sigsz);
241 }
242
243 inline_size
244 int
245 evsys_io_uring_register (unsigned int fd, unsigned int opcode, void *arg, unsigned int nr_args)
246 {
247 return ev_syscall4 (SYS_io_uring_register, fd, opcode, arg, nr_args);
248 }
249
250 /*****************************************************************************/
251 /* actual backend implementation */
252
253 /* we hope that volatile will make the compiler access this variables only once */
254 #define EV_SQ_VAR(name) *(volatile unsigned *)((char *)iouring_ring + iouring_sq_ ## name)
255 #define EV_CQ_VAR(name) *(volatile unsigned *)((char *)iouring_ring + iouring_cq_ ## name)
256
257 /* the index array */
258 #define EV_SQ_ARRAY ((unsigned *)((char *)iouring_ring + iouring_sq_array))
259
260 /* the submit/completion queue entries */
261 #define EV_SQES ((struct io_uring_sqe *) iouring_sqes)
262 #define EV_CQES ((struct io_uring_cqe *)((char *)iouring_ring + iouring_cq_cqes))
263
264 inline_speed
265 int
266 iouring_enter (EV_P_ ev_tstamp timeout)
267 {
268 int res;
269
270 EV_RELEASE_CB;
271
272 res = evsys_io_uring_enter (iouring_fd, iouring_to_submit, 1,
273 timeout > EV_TS_CONST (0.) ? IORING_ENTER_GETEVENTS : 0, 0, 0);
274
275 assert (("libev: io_uring_enter did not consume all sqes", (res < 0 || res == iouring_to_submit)));
276
277 iouring_to_submit = 0;
278
279 EV_ACQUIRE_CB;
280
281 return res;
282 }
283
284 /* TODO: can we move things around so we don't need this forward-reference? */
285 static void
286 iouring_poll (EV_P_ ev_tstamp timeout);
287
288 static
289 struct io_uring_sqe *
290 iouring_sqe_get (EV_P)
291 {
292 unsigned tail;
293
294 for (;;)
295 {
296 tail = EV_SQ_VAR (tail);
297
298 if (ecb_expect_true (tail + 1 - EV_SQ_VAR (head) <= EV_SQ_VAR (ring_entries)))
299 break; /* whats the problem, we have free sqes */
300
301 /* queue full, need to flush and possibly handle some events */
302
303 #if EV_FEATURE_CODE
304 /* first we ask the kernel nicely, most often this frees up some sqes */
305 int res = iouring_enter (EV_A_ EV_TS_CONST (0.));
306
307 ECB_MEMORY_FENCE_ACQUIRE; /* better safe than sorry */
308
309 if (res >= 0)
310 continue; /* yes, it worked, try again */
311 #endif
312
313 /* some problem, possibly EBUSY - do the full poll and let it handle any issues */
314
315 iouring_poll (EV_A_ EV_TS_CONST (0.));
316 /* iouring_poll should have done ECB_MEMORY_FENCE_ACQUIRE for us */
317 }
318
319 /*assert (("libev: io_uring queue full after flush", tail + 1 - EV_SQ_VAR (head) <= EV_SQ_VAR (ring_entries)));*/
320
321 return EV_SQES + (tail & EV_SQ_VAR (ring_mask));
322 }
323
324 inline_size
325 void
326 iouring_sqe_submit (EV_P_ struct io_uring_sqe *sqe)
327 {
328 unsigned idx = sqe - EV_SQES;
329
330 EV_SQ_ARRAY [idx] = idx;
331 ECB_MEMORY_FENCE_RELEASE;
332 ++EV_SQ_VAR (tail);
333 /*ECB_MEMORY_FENCE_RELEASE; /* for the time being we assume this is not needed */
334 ++iouring_to_submit;
335 }
336
337 /*****************************************************************************/
338
339 /* when the timerfd expires we simply note the fact,
340 * as the purpose of the timerfd is to wake us up, nothing else.
341 * the next iteration should re-set it.
342 */
343 static void
344 iouring_tfd_cb (EV_P_ struct ev_io *w, int revents)
345 {
346 iouring_tfd_to = EV_TSTAMP_HUGE;
347 }
348
349 /* called for full and partial cleanup */
350 ecb_cold
351 static void
352 iouring_internal_destroy (EV_P)
353 {
354 close (iouring_tfd);
355 close (iouring_fd);
356
357 if (iouring_ring != MAP_FAILED) munmap (iouring_ring, iouring_ring_size);
358 if (iouring_sqes != MAP_FAILED) munmap (iouring_sqes, iouring_sqes_size);
359
360 if (ev_is_active (&iouring_tfd_w))
361 {
362 ev_ref (EV_A);
363 ev_io_stop (EV_A_ &iouring_tfd_w);
364 }
365 }
366
367 ecb_cold
368 static int
369 iouring_internal_init (EV_P)
370 {
371 struct io_uring_params params = { 0 };
372 uint32_t sq_size, cq_size;
373
374 params.flags = IORING_SETUP_CLAMP;
375
376 iouring_to_submit = 0;
377
378 iouring_tfd = -1;
379 iouring_ring = MAP_FAILED;
380 iouring_sqes = MAP_FAILED;
381
382 if (!have_monotonic) /* cannot really happen, but what if11 */
383 return -1;
384
385 iouring_fd = evsys_io_uring_setup (iouring_entries, &params);
386
387 if (iouring_fd < 0)
388 return -1;
389
390 if ((~params.features) & (IORING_FEAT_NODROP | IORING_FEAT_SINGLE_MMAP | IORING_FEAT_SUBMIT_STABLE))
391 return -1; /* we require the above features */
392
393 /* TODO: remember somehow whether our queue size has been clamped */
394
395 sq_size = params.sq_off.array + params.sq_entries * sizeof (unsigned);
396 cq_size = params.cq_off.cqes + params.cq_entries * sizeof (struct io_uring_cqe);
397
398 iouring_ring_size = sq_size > cq_size ? sq_size : cq_size;
399 iouring_sqes_size = params.sq_entries * sizeof (struct io_uring_sqe);
400
401 iouring_ring = mmap (0, iouring_ring_size, PROT_READ | PROT_WRITE,
402 MAP_SHARED | MAP_POPULATE, iouring_fd, IORING_OFF_SQ_RING);
403 iouring_sqes = mmap (0, iouring_sqes_size, PROT_READ | PROT_WRITE,
404 MAP_SHARED | MAP_POPULATE, iouring_fd, IORING_OFF_SQES);
405
406 if (iouring_ring == MAP_FAILED || iouring_sqes == MAP_FAILED)
407 return -1;
408
409 iouring_sq_head = params.sq_off.head;
410 iouring_sq_tail = params.sq_off.tail;
411 iouring_sq_ring_mask = params.sq_off.ring_mask;
412 iouring_sq_ring_entries = params.sq_off.ring_entries;
413 iouring_sq_flags = params.sq_off.flags;
414 iouring_sq_dropped = params.sq_off.dropped;
415 iouring_sq_array = params.sq_off.array;
416
417 iouring_cq_head = params.cq_off.head;
418 iouring_cq_tail = params.cq_off.tail;
419 iouring_cq_ring_mask = params.cq_off.ring_mask;
420 iouring_cq_ring_entries = params.cq_off.ring_entries;
421 iouring_cq_overflow = params.cq_off.overflow;
422 iouring_cq_cqes = params.cq_off.cqes;
423
424 iouring_tfd_to = EV_TSTAMP_HUGE;
425
426 iouring_tfd = timerfd_create (CLOCK_MONOTONIC, TFD_CLOEXEC);
427
428 if (iouring_tfd < 0)
429 return -1;
430
431 return 0;
432 }
433
434 ecb_cold
435 static void
436 iouring_fork (EV_P)
437 {
438 iouring_internal_destroy (EV_A);
439
440 while (iouring_internal_init (EV_A) < 0)
441 ev_syserr ("(libev) io_uring_setup");
442
443 fd_rearm_all (EV_A);
444
445 ev_io_stop (EV_A_ &iouring_tfd_w);
446 ev_io_set (EV_A_ &iouring_tfd_w, iouring_tfd, EV_READ);
447 ev_io_start (EV_A_ &iouring_tfd_w);
448 }
449
450 /*****************************************************************************/
451
452 static void
453 iouring_modify (EV_P_ int fd, int oev, int nev)
454 {
455 if (oev)
456 {
457 /* we assume the sqe's are all "properly" initialised */
458 struct io_uring_sqe *sqe = iouring_sqe_get (EV_A);
459 sqe->opcode = IORING_OP_POLL_REMOVE;
460 sqe->fd = fd;
461 /* Jens Axboe notified me that user_data is not what is documented, but is
462 * some kind of unique ID that has to match, otherwise the request cannot
463 * be removed. Since we don't *really* have that, we pass in the old
464 * generation counter - if that fails, too bad, it will hopefully be removed
465 * at close time and then be ignored. */
466 sqe->addr = (uint32_t)fd | ((__u64)(uint32_t)anfds [fd].egen << 32);
467 sqe->user_data = (uint64_t)-1;
468 iouring_sqe_submit (EV_A_ sqe);
469
470 /* increment generation counter to avoid handling old events */
471 ++anfds [fd].egen;
472 }
473
474 if (nev)
475 {
476 struct io_uring_sqe *sqe = iouring_sqe_get (EV_A);
477 sqe->opcode = IORING_OP_POLL_ADD;
478 sqe->fd = fd;
479 sqe->addr = 0;
480 sqe->user_data = (uint32_t)fd | ((__u64)(uint32_t)anfds [fd].egen << 32);
481 sqe->poll_events =
482 (nev & EV_READ ? POLLIN : 0)
483 | (nev & EV_WRITE ? POLLOUT : 0);
484 iouring_sqe_submit (EV_A_ sqe);
485 }
486 }
487
488 inline_size
489 void
490 iouring_tfd_update (EV_P_ ev_tstamp timeout)
491 {
492 ev_tstamp tfd_to = mn_now + timeout;
493
494 /* we assume there will be many iterations per timer change, so
495 * we only re-set the timerfd when we have to because its expiry
496 * is too late.
497 */
498 if (ecb_expect_false (tfd_to < iouring_tfd_to))
499 {
500 struct itimerspec its;
501
502 iouring_tfd_to = tfd_to;
503 EV_TS_SET (its.it_interval, 0.);
504 EV_TS_SET (its.it_value, tfd_to);
505
506 if (timerfd_settime (iouring_tfd, TFD_TIMER_ABSTIME, &its, 0) < 0)
507 assert (("libev: iouring timerfd_settime failed", 0));
508 }
509 }
510
511 inline_size
512 void
513 iouring_process_cqe (EV_P_ struct io_uring_cqe *cqe)
514 {
515 int fd = cqe->user_data & 0xffffffffU;
516 uint32_t gen = cqe->user_data >> 32;
517 int res = cqe->res;
518
519 /* user_data -1 is a remove that we are not atm. interested in */
520 if (cqe->user_data == (uint64_t)-1)
521 return;
522
523 assert (("libev: io_uring fd must be in-bounds", fd >= 0 && fd < anfdmax));
524
525 /* documentation lies, of course. the result value is NOT like
526 * normal syscalls, but like linux raw syscalls, i.e. negative
527 * error numbers. fortunate, as otherwise there would be no way
528 * to get error codes at all. still, why not document this?
529 */
530
531 /* ignore event if generation doesn't match */
532 /* other than skipping removal events, */
533 /* this should actually be very rare */
534 if (ecb_expect_false (gen != (uint32_t)anfds [fd].egen))
535 return;
536
537 if (ecb_expect_false (res < 0))
538 {
539 /*TODO: EINVAL handling (was something failed with this fd)*/
540
541 if (res == -EBADF)
542 {
543 assert (("libev: event loop rejected bad fd", res != -EBADF));
544 fd_kill (EV_A_ fd);
545 }
546 else
547 {
548 errno = -res;
549 ev_syserr ("(libev) IORING_OP_POLL_ADD");
550 }
551
552 return;
553 }
554
555 /* feed events, we do not expect or handle POLLNVAL */
556 fd_event (
557 EV_A_
558 fd,
559 (res & (POLLOUT | POLLERR | POLLHUP) ? EV_WRITE : 0)
560 | (res & (POLLIN | POLLERR | POLLHUP) ? EV_READ : 0)
561 );
562
563 /* io_uring is oneshot, so we need to re-arm the fd next iteration */
564 /* this also means we usually have to do at least one syscall per iteration */
565 anfds [fd].events = 0;
566 fd_change (EV_A_ fd, EV_ANFD_REIFY);
567 }
568
569 /* called when the event queue overflows */
570 ecb_cold
571 static void
572 iouring_overflow (EV_P)
573 {
574 /* we have two options, resize the queue (by tearing down
575 * everything and recreating it, or living with it
576 * and polling.
577 * we implement this by resizing the queue, and, if that fails,
578 * we just recreate the state on every failure, which
579 * kind of is a very inefficient poll.
580 * one danger is, due to the bios toward lower fds,
581 * we will only really get events for those, so
582 * maybe we need a poll() fallback, after all.
583 */
584 /*EV_CQ_VAR (overflow) = 0;*/ /* need to do this if we keep the state and poll manually */
585
586 fd_rearm_all (EV_A);
587
588 /* we double the size until we hit the hard-to-probe maximum */
589 if (!iouring_max_entries)
590 {
591 iouring_entries <<= 1;
592 iouring_fork (EV_A);
593 }
594 else
595 {
596 /* we hit the kernel limit, we should fall back to something else.
597 * we can either poll() a few times and hope for the best,
598 * poll always, or switch to epoll.
599 * TODO: is this necessary with newer kernels?
600 */
601
602 iouring_internal_destroy (EV_A);
603
604 /* this should make it so that on return, we don't call any uring functions */
605 iouring_to_submit = 0;
606
607 for (;;)
608 {
609 backend = epoll_init (EV_A_ 0);
610
611 if (backend)
612 break;
613
614 ev_syserr ("(libev) iouring switch to epoll");
615 }
616 }
617 }
618
619 /* handle any events in the completion queue, return true if there were any */
620 static int
621 iouring_handle_cq (EV_P)
622 {
623 unsigned head, tail, mask;
624
625 head = EV_CQ_VAR (head);
626 ECB_MEMORY_FENCE_ACQUIRE;
627 tail = EV_CQ_VAR (tail);
628
629 if (head == tail)
630 return 0;
631
632 /* it can only overflow if we have events, yes, yes? */
633 if (ecb_expect_false (EV_CQ_VAR (overflow)))
634 {
635 iouring_overflow (EV_A);
636 return 1;
637 }
638
639 mask = EV_CQ_VAR (ring_mask);
640
641 do
642 iouring_process_cqe (EV_A_ &EV_CQES [head++ & mask]);
643 while (head != tail);
644
645 EV_CQ_VAR (head) = head;
646 ECB_MEMORY_FENCE_RELEASE;
647
648 return 1;
649 }
650
651 static void
652 iouring_poll (EV_P_ ev_tstamp timeout)
653 {
654 /* if we have events, no need for extra syscalls, but we might have to queue events */
655 /* we also clar the timeout if there are outstanding fdchanges */
656 /* the latter should only happen if both the sq and cq are full, most likely */
657 /* because we have a lot of event sources that immediately complete */
658 /* TODO: fdchacngecnt is always 0 because fd_reify does not have two buffers yet */
659 if (iouring_handle_cq (EV_A) || fdchangecnt)
660 timeout = EV_TS_CONST (0.);
661 else
662 /* no events, so maybe wait for some */
663 iouring_tfd_update (EV_A_ timeout);
664
665 /* only enter the kernel if we have something to submit, or we need to wait */
666 if (timeout || iouring_to_submit)
667 {
668 int res = iouring_enter (EV_A_ timeout);
669
670 if (ecb_expect_false (res < 0))
671 if (errno == EINTR)
672 /* ignore */;
673 else if (errno == EBUSY)
674 /* cq full, cannot submit - should be rare because we flush the cq first, so simply ignore */;
675 else
676 ev_syserr ("(libev) iouring setup");
677 else
678 iouring_handle_cq (EV_A);
679 }
680 }
681
682 inline_size
683 int
684 iouring_init (EV_P_ int flags)
685 {
686 iouring_entries = IOURING_INIT_ENTRIES;
687 iouring_max_entries = 0;
688
689 if (iouring_internal_init (EV_A) < 0)
690 {
691 iouring_internal_destroy (EV_A);
692 return 0;
693 }
694
695 ev_io_init (&iouring_tfd_w, iouring_tfd_cb, iouring_tfd, EV_READ);
696 ev_set_priority (&iouring_tfd_w, EV_MINPRI);
697 ev_io_start (EV_A_ &iouring_tfd_w);
698 ev_unref (EV_A); /* watcher should not keep loop alive */
699
700 backend_modify = iouring_modify;
701 backend_poll = iouring_poll;
702
703 return EVBACKEND_IOURING;
704 }
705
706 inline_size
707 void
708 iouring_destroy (EV_P)
709 {
710 iouring_internal_destroy (EV_A);
711 }
712