1 /* Handle general operations.
2 Copyright (C) 1997-2023 Free Software Foundation, Inc.
3 This file is part of the GNU C Library.
4
5 The GNU C Library is free software; you can redistribute it and/or
6 modify it under the terms of the GNU Lesser General Public
7 License as published by the Free Software Foundation; either
8 version 2.1 of the License, or (at your option) any later version.
9
10 The GNU C Library is distributed in the hope that it will be useful,
11 but WITHOUT ANY WARRANTY; without even the implied warranty of
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 Lesser General Public License for more details.
14
15 You should have received a copy of the GNU Lesser General Public
16 License along with the GNU C Library; if not, see
17 <https://www.gnu.org/licenses/>. */
18
19 #include <aio.h>
20 #include <assert.h>
21 #include <errno.h>
22 #include <limits.h>
23 #include <pthreadP.h>
24 #include <stdlib.h>
25 #include <unistd.h>
26 #include <sys/param.h>
27 #include <sys/stat.h>
28 #include <sys/time.h>
29 #include <aio_misc.h>
30
31 #if !PTHREAD_IN_LIBC
32 /* The available function names differ outside of libc. (In libc, we
33 need to use hidden aliases to avoid the PLT.) */
34 # define __pread __libc_pread
35 # define __pthread_attr_destroy pthread_attr_destroy
36 # define __pthread_attr_init pthread_attr_init
37 # define __pthread_attr_setdetachstate pthread_attr_setdetachstate
38 # define __pthread_cond_signal pthread_cond_signal
39 # define __pthread_cond_timedwait pthread_cond_timedwait
40 # define __pthread_getschedparam pthread_getschedparam
41 # define __pthread_setschedparam pthread_setschedparam
42 # define __pwrite __libc_pwrite
43 #endif
44
45 #ifndef aio_create_helper_thread
46 # define aio_create_helper_thread __aio_create_helper_thread
47
48 extern inline int
49 __aio_create_helper_thread (pthread_t *threadp, void *(*tf) (void *), void *arg)
50 {
51 pthread_attr_t attr;
52
53 /* Make sure the thread is created detached. */
54 __pthread_attr_init (&attr);
55 __pthread_attr_setdetachstate (&attr, PTHREAD_CREATE_DETACHED);
56
57 int ret = __pthread_create (threadp, &attr, tf, arg);
58
59 __pthread_attr_destroy (&attr);
60 return ret;
61 }
62 #endif
63
64 static void add_request_to_runlist (struct requestlist *newrequest);
65
66 /* Pool of request list entries. */
67 static struct requestlist **pool;
68
69 /* Number of total and allocated pool entries. */
70 static size_t pool_max_size;
71 static size_t pool_size;
72
73 /* We implement a two dimensional array but allocate each row separately.
74 The macro below determines how many entries should be used per row.
75 It should better be a power of two. */
76 #define ENTRIES_PER_ROW 32
77
78 /* How many rows we allocate at once. */
79 #define ROWS_STEP 8
80
81 /* List of available entries. */
82 static struct requestlist *freelist;
83
84 /* List of request waiting to be processed. */
85 static struct requestlist *runlist;
86
87 /* Structure list of all currently processed requests. */
88 static struct requestlist *requests;
89
90 /* Number of threads currently running. */
91 static int nthreads;
92
93 /* Number of threads waiting for work to arrive. */
94 static int idle_thread_count;
95
96
97 /* These are the values used to optimize the use of AIO. The user can
98 overwrite them by using the `aio_init' function. */
99 static struct aioinit optim =
100 {
101 20, /* int aio_threads; Maximal number of threads. */
102 64, /* int aio_num; Number of expected simultaneous requests. */
103 0,
104 0,
105 0,
106 0,
107 1,
108 0
109 };
110
111
112 /* Since the list is global we need a mutex protecting it. */
113 pthread_mutex_t __aio_requests_mutex = PTHREAD_RECURSIVE_MUTEX_INITIALIZER_NP;
114
115 /* When you add a request to the list and there are idle threads present,
116 you signal this condition variable. When a thread finishes work, it waits
117 on this condition variable for a time before it actually exits. */
118 pthread_cond_t __aio_new_request_notification = PTHREAD_COND_INITIALIZER;
119
120
121 /* Functions to handle request list pool. */
122 static struct requestlist *
123 get_elem (void)
124 {
125 struct requestlist *result;
126
127 if (freelist == NULL)
128 {
129 struct requestlist *new_row;
130 int cnt;
131
132 assert (sizeof (struct aiocb) == sizeof (struct aiocb64));
133
134 if (pool_size + 1 >= pool_max_size)
135 {
136 size_t new_max_size = pool_max_size + ROWS_STEP;
137 struct requestlist **new_tab;
138
139 new_tab = (struct requestlist **)
140 realloc (pool, new_max_size * sizeof (struct requestlist *));
141
142 if (new_tab == NULL)
143 return NULL;
144
145 pool_max_size = new_max_size;
146 pool = new_tab;
147 }
148
149 /* Allocate the new row. */
150 cnt = pool_size == 0 ? optim.aio_num : ENTRIES_PER_ROW;
151 new_row = (struct requestlist *) calloc (cnt,
152 sizeof (struct requestlist));
153 if (new_row == NULL)
154 return NULL;
155
156 pool[pool_size++] = new_row;
157
158 /* Put all the new entries in the freelist. */
159 do
160 {
161 new_row->next_prio = freelist;
162 freelist = new_row++;
163 }
164 while (--cnt > 0);
165 }
166
167 result = freelist;
168 freelist = freelist->next_prio;
169
170 return result;
171 }
172
173
174 void
175 __aio_free_request (struct requestlist *elem)
176 {
177 elem->running = no;
178 elem->next_prio = freelist;
179 freelist = elem;
180 }
181
182
183 struct requestlist *
184 __aio_find_req (aiocb_union *elem)
185 {
186 struct requestlist *runp = requests;
187 int fildes = elem->aiocb.aio_fildes;
188
189 while (runp != NULL && runp->aiocbp->aiocb.aio_fildes < fildes)
190 runp = runp->next_fd;
191
192 if (runp != NULL)
193 {
194 if (runp->aiocbp->aiocb.aio_fildes != fildes)
195 runp = NULL;
196 else
197 while (runp != NULL && runp->aiocbp != elem)
198 runp = runp->next_prio;
199 }
200
201 return runp;
202 }
203
204
205 struct requestlist *
206 __aio_find_req_fd (int fildes)
207 {
208 struct requestlist *runp = requests;
209
210 while (runp != NULL && runp->aiocbp->aiocb.aio_fildes < fildes)
211 runp = runp->next_fd;
212
213 return (runp != NULL && runp->aiocbp->aiocb.aio_fildes == fildes
214 ? runp : NULL);
215 }
216
217
218 void
219 __aio_remove_request (struct requestlist *last, struct requestlist *req,
220 int all)
221 {
222 assert (req->running == yes || req->running == queued
223 || req->running == done);
224
225 if (last != NULL)
226 last->next_prio = all ? NULL : req->next_prio;
227 else
228 {
229 if (all || req->next_prio == NULL)
230 {
231 if (req->last_fd != NULL)
232 req->last_fd->next_fd = req->next_fd;
233 else
234 requests = req->next_fd;
235 if (req->next_fd != NULL)
236 req->next_fd->last_fd = req->last_fd;
237 }
238 else
239 {
240 if (req->last_fd != NULL)
241 req->last_fd->next_fd = req->next_prio;
242 else
243 requests = req->next_prio;
244
245 if (req->next_fd != NULL)
246 req->next_fd->last_fd = req->next_prio;
247
248 req->next_prio->last_fd = req->last_fd;
249 req->next_prio->next_fd = req->next_fd;
250
251 /* Mark this entry as runnable. */
252 req->next_prio->running = yes;
253 }
254
255 if (req->running == yes)
256 {
257 struct requestlist *runp = runlist;
258
259 last = NULL;
260 while (runp != NULL)
261 {
262 if (runp == req)
263 {
264 if (last == NULL)
265 runlist = runp->next_run;
266 else
267 last->next_run = runp->next_run;
268 break;
269 }
270 last = runp;
271 runp = runp->next_run;
272 }
273 }
274 }
275 }
276
277
278 /* The thread handler. */
279 static void *handle_fildes_io (void *arg);
280
281
282 /* User optimization. */
283 void
284 __aio_init (const struct aioinit *init)
285 {
286 /* Get the mutex. */
287 __pthread_mutex_lock (&__aio_requests_mutex);
288
289 /* Only allow writing new values if the table is not yet allocated. */
290 if (pool == NULL)
291 {
292 optim.aio_threads = init->aio_threads < 1 ? 1 : init->aio_threads;
293 assert (powerof2 (ENTRIES_PER_ROW));
294 optim.aio_num = (init->aio_num < ENTRIES_PER_ROW
295 ? ENTRIES_PER_ROW
296 : init->aio_num & ~(ENTRIES_PER_ROW - 1));
297 }
298
299 if (init->aio_idle_time != 0)
300 optim.aio_idle_time = init->aio_idle_time;
301
302 /* Release the mutex. */
303 __pthread_mutex_unlock (&__aio_requests_mutex);
304 }
305
306
307 /* The main function of the async I/O handling. It enqueues requests
308 and if necessary starts and handles threads. */
309 struct requestlist *
310 __aio_enqueue_request (aiocb_union *aiocbp, int operation)
311 {
312 int result = 0;
313 int policy, prio;
314 struct sched_param param;
315 struct requestlist *last, *runp, *newp;
316 int running = no;
317
318 if (operation == LIO_SYNC || operation == LIO_DSYNC)
319 aiocbp->aiocb.aio_reqprio = 0;
320 else if (aiocbp->aiocb.aio_reqprio < 0
321 #ifdef AIO_PRIO_DELTA_MAX
322 || aiocbp->aiocb.aio_reqprio > AIO_PRIO_DELTA_MAX
323 #endif
324 )
325 {
326 /* Invalid priority value. */
327 __set_errno (EINVAL);
328 aiocbp->aiocb.__error_code = EINVAL;
329 aiocbp->aiocb.__return_value = -1;
330 return NULL;
331 }
332
333 /* Compute priority for this request. */
334 __pthread_getschedparam (__pthread_self (), &policy, ¶m);
335 prio = param.sched_priority - aiocbp->aiocb.aio_reqprio;
336
337 /* Get the mutex. */
338 __pthread_mutex_lock (&__aio_requests_mutex);
339
340 last = NULL;
341 runp = requests;
342 /* First look whether the current file descriptor is currently
343 worked with. */
344 while (runp != NULL
345 && runp->aiocbp->aiocb.aio_fildes < aiocbp->aiocb.aio_fildes)
346 {
347 last = runp;
348 runp = runp->next_fd;
349 }
350
351 /* Get a new element for the waiting list. */
352 newp = get_elem ();
353 if (newp == NULL)
354 {
355 __pthread_mutex_unlock (&__aio_requests_mutex);
356 __set_errno (EAGAIN);
357 return NULL;
358 }
359 newp->aiocbp = aiocbp;
360 newp->waiting = NULL;
361
362 aiocbp->aiocb.__abs_prio = prio;
363 aiocbp->aiocb.__policy = policy;
364 aiocbp->aiocb.aio_lio_opcode = operation;
365 aiocbp->aiocb.__error_code = EINPROGRESS;
366 aiocbp->aiocb.__return_value = 0;
367
368 if (runp != NULL
369 && runp->aiocbp->aiocb.aio_fildes == aiocbp->aiocb.aio_fildes)
370 {
371 /* The current file descriptor is worked on. It makes no sense
372 to start another thread since this new thread would fight
373 with the running thread for the resources. But we also cannot
374 say that the thread processing this descriptor shall immediately
375 after finishing the current job process this request if there
376 are other threads in the running queue which have a higher
377 priority. */
378
379 /* Simply enqueue it after the running one according to the
380 priority. */
381 last = NULL;
382 while (runp->next_prio != NULL
383 && runp->next_prio->aiocbp->aiocb.__abs_prio >= prio)
384 {
385 last = runp;
386 runp = runp->next_prio;
387 }
388
389 newp->next_prio = runp->next_prio;
390 runp->next_prio = newp;
391
392 running = queued;
393 }
394 else
395 {
396 running = yes;
397 /* Enqueue this request for a new descriptor. */
398 if (last == NULL)
399 {
400 newp->last_fd = NULL;
401 newp->next_fd = requests;
402 if (requests != NULL)
403 requests->last_fd = newp;
404 requests = newp;
405 }
406 else
407 {
408 newp->next_fd = last->next_fd;
409 newp->last_fd = last;
410 last->next_fd = newp;
411 if (newp->next_fd != NULL)
412 newp->next_fd->last_fd = newp;
413 }
414
415 newp->next_prio = NULL;
416 last = NULL;
417 }
418
419 if (running == yes)
420 {
421 /* We try to create a new thread for this file descriptor. The
422 function which gets called will handle all available requests
423 for this descriptor and when all are processed it will
424 terminate.
425
426 If no new thread can be created or if the specified limit of
427 threads for AIO is reached we queue the request. */
428
429 /* See if we need to and are able to create a thread. */
430 if (nthreads < optim.aio_threads && idle_thread_count == 0)
431 {
432 pthread_t thid;
433
434 running = newp->running = allocated;
435
436 /* Now try to start a thread. */
437 result = aio_create_helper_thread (&thid, handle_fildes_io, newp);
438 if (result == 0)
439 /* We managed to enqueue the request. All errors which can
440 happen now can be recognized by calls to `aio_return' and
441 `aio_error'. */
442 ++nthreads;
443 else
444 {
445 /* Reset the running flag. The new request is not running. */
446 running = newp->running = yes;
447
448 if (nthreads == 0)
449 {
450 /* We cannot create a thread in the moment and there is
451 also no thread running. This is a problem. `errno' is
452 set to EAGAIN if this is only a temporary problem. */
453 __aio_remove_request (last, newp, 0);
454 }
455 else
456 result = 0;
457 }
458 }
459 }
460
461 /* Enqueue the request in the run queue if it is not yet running. */
462 if (running == yes && result == 0)
463 {
464 add_request_to_runlist (newp);
465
466 /* If there is a thread waiting for work, then let it know that we
467 have just given it something to do. */
468 if (idle_thread_count > 0)
469 __pthread_cond_signal (&__aio_new_request_notification);
470 }
471
472 if (result == 0)
473 newp->running = running;
474 else
475 {
476 /* Something went wrong. */
477 __aio_free_request (newp);
478 aiocbp->aiocb.__error_code = result;
479 __set_errno (result);
480 newp = NULL;
481 }
482
483 /* Release the mutex. */
484 __pthread_mutex_unlock (&__aio_requests_mutex);
485
486 return newp;
487 }
488
489
490 static void *
491 handle_fildes_io (void *arg)
492 {
493 pthread_t self = __pthread_self ();
494 struct sched_param param;
495 struct requestlist *runp = (struct requestlist *) arg;
496 aiocb_union *aiocbp;
497 int policy;
498 int fildes;
499
500 __pthread_getschedparam (self, &policy, ¶m);
501
502 do
503 {
504 /* If runp is NULL, then we were created to service the work queue
505 in general, not to handle any particular request. In that case we
506 skip the "do work" stuff on the first pass, and go directly to the
507 "get work off the work queue" part of this loop, which is near the
508 end. */
509 if (runp == NULL)
510 __pthread_mutex_lock (&__aio_requests_mutex);
511 else
512 {
513 /* Hopefully this request is marked as running. */
514 assert (runp->running == allocated);
515
516 /* Update our variables. */
517 aiocbp = runp->aiocbp;
518 fildes = aiocbp->aiocb.aio_fildes;
519
520 /* Change the priority to the requested value (if necessary). */
521 if (aiocbp->aiocb.__abs_prio != param.sched_priority
522 || aiocbp->aiocb.__policy != policy)
523 {
524 param.sched_priority = aiocbp->aiocb.__abs_prio;
525 policy = aiocbp->aiocb.__policy;
526 __pthread_setschedparam (self, policy, ¶m);
527 }
528
529 /* Process request pointed to by RUNP. We must not be disturbed
530 by signals. */
531 if ((aiocbp->aiocb.aio_lio_opcode & 127) == LIO_READ)
532 {
533 if (sizeof (off_t) != sizeof (off64_t)
534 && aiocbp->aiocb.aio_lio_opcode & 128)
535 aiocbp->aiocb.__return_value =
536 TEMP_FAILURE_RETRY (__pread64 (fildes, (void *)
537 aiocbp->aiocb64.aio_buf,
538 aiocbp->aiocb64.aio_nbytes,
539 aiocbp->aiocb64.aio_offset));
540 else
541 aiocbp->aiocb.__return_value =
542 TEMP_FAILURE_RETRY (__pread (fildes,
543 (void *)
544 aiocbp->aiocb.aio_buf,
545 aiocbp->aiocb.aio_nbytes,
546 aiocbp->aiocb.aio_offset));
547
548 if (aiocbp->aiocb.__return_value == -1 && errno == ESPIPE)
549 /* The Linux kernel is different from others. It returns
550 ESPIPE if using pread on a socket. Other platforms
551 simply ignore the offset parameter and behave like
552 read. */
553 aiocbp->aiocb.__return_value =
554 TEMP_FAILURE_RETRY (read (fildes,
555 (void *) aiocbp->aiocb64.aio_buf,
556 aiocbp->aiocb64.aio_nbytes));
557 }
558 else if ((aiocbp->aiocb.aio_lio_opcode & 127) == LIO_WRITE)
559 {
560 if (sizeof (off_t) != sizeof (off64_t)
561 && aiocbp->aiocb.aio_lio_opcode & 128)
562 aiocbp->aiocb.__return_value =
563 TEMP_FAILURE_RETRY (__pwrite64 (fildes, (const void *)
564 aiocbp->aiocb64.aio_buf,
565 aiocbp->aiocb64.aio_nbytes,
566 aiocbp->aiocb64.aio_offset));
567 else
568 aiocbp->aiocb.__return_value =
569 TEMP_FAILURE_RETRY (__pwrite (fildes, (const void *)
570 aiocbp->aiocb.aio_buf,
571 aiocbp->aiocb.aio_nbytes,
572 aiocbp->aiocb.aio_offset));
573
574 if (aiocbp->aiocb.__return_value == -1 && errno == ESPIPE)
575 /* The Linux kernel is different from others. It returns
576 ESPIPE if using pwrite on a socket. Other platforms
577 simply ignore the offset parameter and behave like
578 write. */
579 aiocbp->aiocb.__return_value =
580 TEMP_FAILURE_RETRY (write (fildes,
581 (void *) aiocbp->aiocb64.aio_buf,
582 aiocbp->aiocb64.aio_nbytes));
583 }
584 else if (aiocbp->aiocb.aio_lio_opcode == LIO_DSYNC)
585 aiocbp->aiocb.__return_value =
586 TEMP_FAILURE_RETRY (fdatasync (fildes));
587 else if (aiocbp->aiocb.aio_lio_opcode == LIO_SYNC)
588 aiocbp->aiocb.__return_value =
589 TEMP_FAILURE_RETRY (fsync (fildes));
590 else
591 {
592 /* This is an invalid opcode. */
593 aiocbp->aiocb.__return_value = -1;
594 __set_errno (EINVAL);
595 }
596
597 /* Get the mutex. */
598 __pthread_mutex_lock (&__aio_requests_mutex);
599
600 if (aiocbp->aiocb.__return_value == -1)
601 aiocbp->aiocb.__error_code = errno;
602 else
603 aiocbp->aiocb.__error_code = 0;
604
605 /* Send the signal to notify about finished processing of the
606 request. */
607 __aio_notify (runp);
608
609 /* For debugging purposes we reset the running flag of the
610 finished request. */
611 assert (runp->running == allocated);
612 runp->running = done;
613
614 /* Now dequeue the current request. */
615 __aio_remove_request (NULL, runp, 0);
616 if (runp->next_prio != NULL)
617 add_request_to_runlist (runp->next_prio);
618
619 /* Free the old element. */
620 __aio_free_request (runp);
621 }
622
623 runp = runlist;
624
625 /* If the runlist is empty, then we sleep for a while, waiting for
626 something to arrive in it. */
627 if (runp == NULL && optim.aio_idle_time >= 0)
628 {
629 struct timespec now;
630 struct timespec wakeup_time;
631
632 ++idle_thread_count;
633 __clock_gettime (CLOCK_REALTIME, &now);
634 wakeup_time.tv_sec = now.tv_sec + optim.aio_idle_time;
635 wakeup_time.tv_nsec = now.tv_nsec;
636 if (wakeup_time.tv_nsec >= 1000000000)
637 {
638 wakeup_time.tv_nsec -= 1000000000;
639 ++wakeup_time.tv_sec;
640 }
641 __pthread_cond_timedwait (&__aio_new_request_notification,
642 &__aio_requests_mutex,
643 &wakeup_time);
644 --idle_thread_count;
645 runp = runlist;
646 }
647
648 if (runp == NULL)
649 --nthreads;
650 else
651 {
652 assert (runp->running == yes);
653 runp->running = allocated;
654 runlist = runp->next_run;
655
656 /* If we have a request to process, and there's still another in
657 the run list, then we need to either wake up or create a new
658 thread to service the request that is still in the run list. */
659 if (runlist != NULL)
660 {
661 /* There are at least two items in the work queue to work on.
662 If there are other idle threads, then we should wake them
663 up for these other work elements; otherwise, we should try
664 to create a new thread. */
665 if (idle_thread_count > 0)
666 __pthread_cond_signal (&__aio_new_request_notification);
667 else if (nthreads < optim.aio_threads)
668 {
669 pthread_t thid;
670 pthread_attr_t attr;
671
672 /* Make sure the thread is created detached. */
673 __pthread_attr_init (&attr);
674 __pthread_attr_setdetachstate (&attr,
675 PTHREAD_CREATE_DETACHED);
676
677 /* Now try to start a thread. If we fail, no big deal,
678 because we know that there is at least one thread (us)
679 that is working on AIO operations. */
680 if (__pthread_create (&thid, &attr, handle_fildes_io, NULL)
681 == 0)
682 ++nthreads;
683 }
684 }
685 }
686
687 /* Release the mutex. */
688 __pthread_mutex_unlock (&__aio_requests_mutex);
689 }
690 while (runp != NULL);
691
692 return NULL;
693 }
694
695
696 /* Free allocated resources. */
697 #if !PTHREAD_IN_LIBC
698 __attribute__ ((__destructor__)) static
699 #endif
700 void
701 __aio_freemem (void)
702 {
703 size_t row;
704
705 for (row = 0; row < pool_size; ++row)
706 free (pool[row]);
707
708 free (pool);
709 }
710
711
712 /* Add newrequest to the runlist. The __abs_prio flag of newrequest must
713 be correctly set to do this. Also, you had better set newrequest's
714 "running" flag to "yes" before you release your lock or you'll throw an
715 assertion. */
716 static void
717 add_request_to_runlist (struct requestlist *newrequest)
718 {
719 int prio = newrequest->aiocbp->aiocb.__abs_prio;
720 struct requestlist *runp;
721
722 if (runlist == NULL || runlist->aiocbp->aiocb.__abs_prio < prio)
723 {
724 newrequest->next_run = runlist;
725 runlist = newrequest;
726 }
727 else
728 {
729 runp = runlist;
730
731 while (runp->next_run != NULL
732 && runp->next_run->aiocbp->aiocb.__abs_prio >= prio)
733 runp = runp->next_run;
734
735 newrequest->next_run = runp->next_run;
736 runp->next_run = newrequest;
737 }
738 }
739
740 #if PTHREAD_IN_LIBC
741 versioned_symbol (libc, __aio_init, aio_init, GLIBC_2_34);
742 # if OTHER_SHLIB_COMPAT (librt, GLIBC_2_1, GLIBC_2_34)
743 compat_symbol (librt, __aio_init, aio_init, GLIBC_2_1);
744 # endif
745 #else /* !PTHREAD_IN_LIBC */
746 weak_alias (__aio_init, aio_init)
747 #endif /* !PTHREAD_IN_LIBC */