summaryrefslogtreecommitdiff
path: root/source3/modules/vfs_aio_fork.c
blob: 54fb4d20151dd4fe346d93972506d3aad7a85df6 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
/*
 * Simulate the Posix AIO using mmap/fork
 *
 * Copyright (C) Volker Lendecke 2008
 * Copyright (C) Jeremy Allison 2010
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
 */

#include "includes.h"
#include "system/filesys.h"
#include "system/shmem.h"
#include "smbd/smbd.h"
#include "smbd/globals.h"
#include "lib/async_req/async_sock.h"
#include "lib/util/tevent_unix.h"

#undef recvmsg

#ifndef MAP_FILE
#define MAP_FILE 0
#endif

struct aio_fork_config {
	bool erratic_testing_mode;
};

struct mmap_area {
	size_t size;
	volatile void *ptr;
};

static int mmap_area_destructor(struct mmap_area *area)
{
	munmap((void *)area->ptr, area->size);
	return 0;
}

static struct mmap_area *mmap_area_init(TALLOC_CTX *mem_ctx, size_t size)
{
	struct mmap_area *result;
	int fd;

	result = talloc(mem_ctx, struct mmap_area);
	if (result == NULL) {
		DEBUG(0, ("talloc failed\n"));
		goto fail;
	}

	fd = open("/dev/zero", O_RDWR);
	if (fd == -1) {
		DEBUG(3, ("open(\"/dev/zero\") failed: %s\n",
			  strerror(errno)));
		goto fail;
	}

	result->ptr = mmap(NULL, size, PROT_READ|PROT_WRITE,
			   MAP_SHARED|MAP_FILE, fd, 0);
	if (result->ptr == MAP_FAILED) {
		DEBUG(1, ("mmap failed: %s\n", strerror(errno)));
		goto fail;
	}

	close(fd);

	result->size = size;
	talloc_set_destructor(result, mmap_area_destructor);

	return result;

fail:
	TALLOC_FREE(result);
	return NULL;
}

enum cmd_type {
	READ_CMD,
	WRITE_CMD,
	FSYNC_CMD
};

static const char *cmd_type_str(enum cmd_type cmd)
{
	const char *result;

	switch (cmd) {
	case READ_CMD:
		result = "READ";
		break;
	case WRITE_CMD:
		result = "WRITE";
		break;
	case FSYNC_CMD:
		result = "FSYNC";
		break;
	default:
		result = "<UNKNOWN>";
		break;
	}
	return result;
}

struct rw_cmd {
	size_t n;
	off_t offset;
	enum cmd_type cmd;
	bool erratic_testing_mode;
};

struct rw_ret {
	ssize_t size;
	int ret_errno;
};

struct aio_child_list;

struct aio_child {
	struct aio_child *prev, *next;
	struct aio_child_list *list;
	pid_t pid;
	int sockfd;
	struct mmap_area *map;
	bool dont_delete;	/* Marked as in use since last cleanup */
	bool busy;
};

struct aio_child_list {
	struct aio_child *children;
	struct tevent_timer *cleanup_event;
};

static void free_aio_children(void **p)
{
	TALLOC_FREE(*p);
}

static ssize_t read_fd(int fd, void *ptr, size_t nbytes, int *recvfd)
{
	struct msghdr msg;
	struct iovec iov[1];
	ssize_t n;
#ifndef HAVE_MSGHDR_MSG_CONTROL
	int newfd;
#endif

#ifdef	HAVE_MSGHDR_MSG_CONTROL
	union {
	  struct cmsghdr	cm;
	  char				control[CMSG_SPACE(sizeof(int))];
	} control_un;
	struct cmsghdr	*cmptr;

	msg.msg_control = control_un.control;
	msg.msg_controllen = sizeof(control_un.control);
#else
#if HAVE_MSGHDR_MSG_ACCTRIGHTS
	msg.msg_accrights = (caddr_t) &newfd;
	msg.msg_accrightslen = sizeof(int);
#else
#error Can not pass file descriptors
#endif
#endif

	msg.msg_name = NULL;
	msg.msg_namelen = 0;
	msg.msg_flags = 0;

	iov[0].iov_base = (void *)ptr;
	iov[0].iov_len = nbytes;
	msg.msg_iov = iov;
	msg.msg_iovlen = 1;

	if ( (n = recvmsg(fd, &msg, 0)) <= 0) {
		return(n);
	}

#ifdef	HAVE_MSGHDR_MSG_CONTROL
	if ((cmptr = CMSG_FIRSTHDR(&msg)) != NULL
	    && cmptr->cmsg_len == CMSG_LEN(sizeof(int))) {
		if (cmptr->cmsg_level != SOL_SOCKET) {
			DEBUG(10, ("control level != SOL_SOCKET"));
			errno = EINVAL;
			return -1;
		}
		if (cmptr->cmsg_type != SCM_RIGHTS) {
			DEBUG(10, ("control type != SCM_RIGHTS"));
			errno = EINVAL;
			return -1;
		}
		memcpy(recvfd, CMSG_DATA(cmptr), sizeof(*recvfd));
	} else {
		*recvfd = -1;		/* descriptor was not passed */
	}
#else
	if (msg.msg_accrightslen == sizeof(int)) {
		*recvfd = newfd;
	}
	else {
		*recvfd = -1;		/* descriptor was not passed */
	}
#endif

	return(n);
}

static ssize_t write_fd(int fd, void *ptr, size_t nbytes, int sendfd)
{
	struct msghdr	msg;
	struct iovec	iov[1];

#ifdef	HAVE_MSGHDR_MSG_CONTROL
	union {
		struct cmsghdr	cm;
		char control[CMSG_SPACE(sizeof(int))];
	} control_un;
	struct cmsghdr	*cmptr;

	ZERO_STRUCT(msg);
	ZERO_STRUCT(control_un);

	msg.msg_control = control_un.control;
	msg.msg_controllen = sizeof(control_un.control);

	cmptr = CMSG_FIRSTHDR(&msg);
	cmptr->cmsg_len = CMSG_LEN(sizeof(int));
	cmptr->cmsg_level = SOL_SOCKET;
	cmptr->cmsg_type = SCM_RIGHTS;
	memcpy(CMSG_DATA(cmptr), &sendfd, sizeof(sendfd));
#else
	ZERO_STRUCT(msg);
	msg.msg_accrights = (caddr_t) &sendfd;
	msg.msg_accrightslen = sizeof(int);
#endif

	msg.msg_name = NULL;
	msg.msg_namelen = 0;

	ZERO_STRUCT(iov);
	iov[0].iov_base = (void *)ptr;
	iov[0].iov_len = nbytes;
	msg.msg_iov = iov;
	msg.msg_iovlen = 1;

	return (sendmsg(fd, &msg, 0));
}

static void aio_child_cleanup(struct tevent_context *event_ctx,
			      struct tevent_timer *te,
			      struct timeval now,
			      void *private_data)
{
	struct aio_child_list *list = talloc_get_type_abort(
		private_data, struct aio_child_list);
	struct aio_child *child, *next;

	TALLOC_FREE(list->cleanup_event);

	for (child = list->children; child != NULL; child = next) {
		next = child->next;

		if (child->busy) {
			DEBUG(10, ("child %d currently active\n",
				   (int)child->pid));
			continue;
		}

		if (child->dont_delete) {
			DEBUG(10, ("Child %d was active since last cleanup\n",
				   (int)child->pid));
			child->dont_delete = false;
			continue;
		}

		DEBUG(10, ("Child %d idle for more than 30 seconds, "
			   "deleting\n", (int)child->pid));

		TALLOC_FREE(child);
		child = next;
	}

	if (list->children != NULL) {
		/*
		 * Re-schedule the next cleanup round
		 */
		list->cleanup_event = event_add_timed(server_event_context(), list,
						      timeval_add(&now, 30, 0),
						      aio_child_cleanup, list);

	}
}

static struct aio_child_list *init_aio_children(struct vfs_handle_struct *handle)
{
	struct aio_child_list *data = NULL;

	if (SMB_VFS_HANDLE_TEST_DATA(handle)) {
		SMB_VFS_HANDLE_GET_DATA(handle, data, struct aio_child_list,
					return NULL);
	}

	if (data == NULL) {
		data = talloc_zero(NULL, struct aio_child_list);
		if (data == NULL) {
			return NULL;
		}
	}

	/*
	 * Regardless of whether the child_list had been around or not, make
	 * sure that we have a cleanup timed event. This timed event will
	 * delete itself when it finds that no children are around anymore.
	 */

	if (data->cleanup_event == NULL) {
		data->cleanup_event = event_add_timed(server_event_context(), data,
						      timeval_current_ofs(30, 0),
						      aio_child_cleanup, data);
		if (data->cleanup_event == NULL) {
			TALLOC_FREE(data);
			return NULL;
		}
	}

	if (!SMB_VFS_HANDLE_TEST_DATA(handle)) {
		SMB_VFS_HANDLE_SET_DATA(handle, data, free_aio_children,
					struct aio_child_list, return False);
	}

	return data;
}

static void aio_child_loop(int sockfd, struct mmap_area *map)
{
	while (true) {
		int fd = -1;
		ssize_t ret;
		struct rw_cmd cmd_struct;
		struct rw_ret ret_struct;

		ret = read_fd(sockfd, &cmd_struct, sizeof(cmd_struct), &fd);
		if (ret != sizeof(cmd_struct)) {
			DEBUG(10, ("read_fd returned %d: %s\n", (int)ret,
				   strerror(errno)));
			exit(1);
		}

		DEBUG(10, ("aio_child_loop: %s %d bytes at %d from fd %d\n",
			   cmd_type_str(cmd_struct.cmd),
			   (int)cmd_struct.n, (int)cmd_struct.offset, fd));

		if (cmd_struct.erratic_testing_mode) {
			/*
			 * For developer testing, we want erratic behaviour for
			 * async I/O times
			 */
			uint8_t randval;
			unsigned msecs;
			/*
			 * use generate_random_buffer, we just forked from a
			 * common parent state
			 */
			generate_random_buffer(&randval, sizeof(randval));
			msecs = randval + 20;
			DEBUG(10, ("delaying for %u msecs\n", msecs));
			smb_msleep(msecs);
		}

		ZERO_STRUCT(ret_struct);

		switch (cmd_struct.cmd) {
		case READ_CMD:
			ret_struct.size = sys_pread(
				fd, (void *)map->ptr, cmd_struct.n,
				cmd_struct.offset);
#if 0
/* This breaks "make test" when run with aio_fork module. */
#ifdef DEVELOPER
			ret_struct.size = MAX(1, ret_struct.size * 0.9);
#endif
#endif
			break;
		case WRITE_CMD:
			ret_struct.size = sys_pwrite(
				fd, (void *)map->ptr, cmd_struct.n,
				cmd_struct.offset);
			break;
		case FSYNC_CMD:
			ret_struct.size = fsync(fd);
			break;
		default:
			ret_struct.size = -1;
			errno = EINVAL;
		}

		DEBUG(10, ("aio_child_loop: syscall returned %d\n",
			   (int)ret_struct.size));

		if (ret_struct.size == -1) {
			ret_struct.ret_errno = errno;
		}

		/*
		 * Close the fd before telling our parent we're done. The
		 * parent might close and re-open the file very quickly, and
		 * with system-level share modes (GPFS) we would get an
		 * unjustified SHARING_VIOLATION.
		 */
		close(fd);

		ret = write_data(sockfd, (char *)&ret_struct,
				 sizeof(ret_struct));
		if (ret != sizeof(ret_struct)) {
			DEBUG(10, ("could not write ret_struct: %s\n",
				   strerror(errno)));
			exit(2);
		}
	}
}

static int aio_child_destructor(struct aio_child *child)
{
	char c=0;

	SMB_ASSERT(!child->busy);

	DEBUG(10, ("aio_child_destructor: removing child %d on fd %d\n",
			child->pid, child->sockfd));

	/*
	 * closing the sockfd makes the child not return from recvmsg() on RHEL
	 * 5.5 so instead force the child to exit by writing bad data to it
	 */
	write(child->sockfd, &c, sizeof(c));
	close(child->sockfd);
	DLIST_REMOVE(child->list->children, child);
	return 0;
}

/*
 * We have to close all fd's in open files, we might incorrectly hold a system
 * level share mode on a file.
 */

static struct files_struct *close_fsp_fd(struct files_struct *fsp,
					 void *private_data)
{
	if ((fsp->fh != NULL) && (fsp->fh->fd != -1)) {
		close(fsp->fh->fd);
		fsp->fh->fd = -1;
	}
	return NULL;
}

static int create_aio_child(struct smbd_server_connection *sconn,
			    struct aio_child_list *children,
			    size_t map_size,
			    struct aio_child **presult)
{
	struct aio_child *result;
	int fdpair[2];
	int ret;

	fdpair[0] = fdpair[1] = -1;

	result = talloc_zero(children, struct aio_child);
	if (result == NULL) {
		return ENOMEM;
	}

	if (socketpair(AF_UNIX, SOCK_STREAM, 0, fdpair) == -1) {
		ret = errno;
		DEBUG(10, ("socketpair() failed: %s\n", strerror(errno)));
		goto fail;
	}

	DEBUG(10, ("fdpair = %d/%d\n", fdpair[0], fdpair[1]));

	result->map = mmap_area_init(result, map_size);
	if (result->map == NULL) {
		ret = errno;
		DEBUG(0, ("Could not create mmap area\n"));
		goto fail;
	}

	result->pid = fork();
	if (result->pid == -1) {
		ret = errno;
		DEBUG(0, ("fork failed: %s\n", strerror(errno)));
		goto fail;
	}

	if (result->pid == 0) {
		close(fdpair[0]);
		result->sockfd = fdpair[1];
		files_forall(sconn, close_fsp_fd, NULL);
		aio_child_loop(result->sockfd, result->map);
	}

	DEBUG(10, ("Child %d created with sockfd %d\n",
			result->pid, fdpair[0]));

	result->sockfd = fdpair[0];
	close(fdpair[1]);

	result->list = children;
	DLIST_ADD(children->children, result);

	talloc_set_destructor(result, aio_child_destructor);

	*presult = result;

	return 0;

 fail:
	if (fdpair[0] != -1) close(fdpair[0]);
	if (fdpair[1] != -1) close(fdpair[1]);
	TALLOC_FREE(result);

	return ret;
}

static int get_idle_child(struct vfs_handle_struct *handle,
			  struct aio_child **pchild)
{
	struct aio_child_list *children;
	struct aio_child *child;

	children = init_aio_children(handle);
	if (children == NULL) {
		return ENOMEM;
	}

	for (child = children->children; child != NULL; child = child->next) {
		if (!child->busy) {
			break;
		}
	}

	if (child == NULL) {
		int ret;

		DEBUG(10, ("no idle child found, creating new one\n"));

		ret = create_aio_child(handle->conn->sconn, children,
					  128*1024, &child);
		if (ret != 0) {
			DEBUG(10, ("create_aio_child failed: %s\n",
				   strerror(errno)));
			return ret;
		}
	}

	child->dont_delete = true;
	child->busy = true;

	*pchild = child;
	return 0;
}

struct aio_fork_pread_state {
	struct aio_child *child;
	ssize_t ret;
	int err;
};

static void aio_fork_pread_done(struct tevent_req *subreq);

static struct tevent_req *aio_fork_pread_send(struct vfs_handle_struct *handle,
					      TALLOC_CTX *mem_ctx,
					      struct tevent_context *ev,
					      struct files_struct *fsp,
					      void *data,
					      size_t n, off_t offset)
{
	struct tevent_req *req, *subreq;
	struct aio_fork_pread_state *state;
	struct rw_cmd cmd;
	ssize_t written;
	int err;
	struct aio_fork_config *config;

	SMB_VFS_HANDLE_GET_DATA(handle, config,
				struct aio_fork_config,
				return NULL);

	req = tevent_req_create(mem_ctx, &state, struct aio_fork_pread_state);
	if (req == NULL) {
		return NULL;
	}

	if (n > 128*1024) {
		/* TODO: support variable buffers */
		tevent_req_error(req, EINVAL);
		return tevent_req_post(req, ev);
	}

	err = get_idle_child(handle, &state->child);
	if (err != 0) {
		tevent_req_error(req, err);
		return tevent_req_post(req, ev);
	}

	ZERO_STRUCT(cmd);
	cmd.n = n;
	cmd.offset = offset;
	cmd.cmd = READ_CMD;
	cmd.erratic_testing_mode = config->erratic_testing_mode;

	DEBUG(10, ("sending fd %d to child %d\n", fsp->fh->fd,
		   (int)state->child->pid));

	/*
	 * Not making this async. We're writing into an empty unix
	 * domain socket. This should never block.
	 */
	written = write_fd(state->child->sockfd, &cmd, sizeof(cmd),
			   fsp->fh->fd);
	if (written == -1) {
		err = errno;

		TALLOC_FREE(state->child);

		DEBUG(10, ("write_fd failed: %s\n", strerror(err)));
		tevent_req_error(req, err);
		return tevent_req_post(req, ev);
	}

	subreq = read_packet_send(state, ev, state->child->sockfd,
				  sizeof(struct rw_ret), NULL, NULL);
	if (tevent_req_nomem(subreq, req)) {
		TALLOC_FREE(state->child); /* we sent sth down */
		return tevent_req_post(req, ev);
	}
	tevent_req_set_callback(subreq, aio_fork_pread_done, req);
	return req;
}

static void aio_fork_pread_done(struct tevent_req *subreq)
{
	struct tevent_req *req = tevent_req_callback_data(
		subreq, struct tevent_req);
	struct aio_fork_pread_state *state = tevent_req_data(
		req, struct aio_fork_pread_state);
	ssize_t nread;
	uint8_t *buf;
	int err;
	struct rw_ret *retbuf;

	nread = read_packet_recv(subreq, talloc_tos(), &buf, &err);
	TALLOC_FREE(subreq);
	if (nread == -1) {
		TALLOC_FREE(state->child);
		tevent_req_error(req, err);
		return;
	}

	state->child->busy = false;

	retbuf = (struct rw_ret *)buf;
	state->ret = retbuf->size;
	state->err = retbuf->ret_errno;
	tevent_req_done(req);
}

static ssize_t aio_fork_pread_recv(struct tevent_req *req, int *err)
{
	struct aio_fork_pread_state *state = tevent_req_data(
		req, struct aio_fork_pread_state);

	if (tevent_req_is_unix_error(req, err)) {
		return -1;
	}
	if (state->ret == -1) {
		*err = state->err;
	}
	return state->ret;
}

struct aio_fork_pwrite_state {
	struct aio_child *child;
	ssize_t ret;
	int err;
};

static void aio_fork_pwrite_done(struct tevent_req *subreq);

static struct tevent_req *aio_fork_pwrite_send(
	struct vfs_handle_struct *handle, TALLOC_CTX *mem_ctx,
	struct tevent_context *ev, struct files_struct *fsp,
	const void *data, size_t n, off_t offset)
{
	struct tevent_req *req, *subreq;
	struct aio_fork_pwrite_state *state;
	struct rw_cmd cmd;
	ssize_t written;
	int err;
	struct aio_fork_config *config;
	SMB_VFS_HANDLE_GET_DATA(handle, config,
				struct aio_fork_config,
				return NULL);

	req = tevent_req_create(mem_ctx, &state, struct aio_fork_pwrite_state);
	if (req == NULL) {
		return NULL;
	}

	if (n > 128*1024) {
		/* TODO: support variable buffers */
		tevent_req_error(req, EINVAL);
		return tevent_req_post(req, ev);
	}

	err = get_idle_child(handle, &state->child);
	if (err != 0) {
		tevent_req_error(req, err);
		return tevent_req_post(req, ev);
	}

	ZERO_STRUCT(cmd);
	cmd.n = n;
	cmd.offset = offset;
	cmd.cmd = WRITE_CMD;
	cmd.erratic_testing_mode = config->erratic_testing_mode;

	DEBUG(10, ("sending fd %d to child %d\n", fsp->fh->fd,
		   (int)state->child->pid));

	/*
	 * Not making this async. We're writing into an empty unix
	 * domain socket. This should never block.
	 */
	written = write_fd(state->child->sockfd, &cmd, sizeof(cmd),
			   fsp->fh->fd);
	if (written == -1) {
		err = errno;

		TALLOC_FREE(state->child);

		DEBUG(10, ("write_fd failed: %s\n", strerror(err)));
		tevent_req_error(req, err);
		return tevent_req_post(req, ev);
	}

	subreq = read_packet_send(state, ev, state->child->sockfd,
				  sizeof(struct rw_ret), NULL, NULL);
	if (tevent_req_nomem(subreq, req)) {
		TALLOC_FREE(state->child); /* we sent sth down */
		return tevent_req_post(req, ev);
	}
	tevent_req_set_callback(subreq, aio_fork_pwrite_done, req);
	return req;
}

static void aio_fork_pwrite_done(struct tevent_req *subreq)
{
	struct tevent_req *req = tevent_req_callback_data(
		subreq, struct tevent_req);
	struct aio_fork_pwrite_state *state = tevent_req_data(
		req, struct aio_fork_pwrite_state);
	ssize_t nread;
	uint8_t *buf;
	int err;
	struct rw_ret *retbuf;

	nread = read_packet_recv(subreq, talloc_tos(), &buf, &err);
	TALLOC_FREE(subreq);
	if (nread == -1) {
		TALLOC_FREE(state->child);
		tevent_req_error(req, err);
		return;
	}

	state->child->busy = false;

	retbuf = (struct rw_ret *)buf;
	state->ret = retbuf->size;
	state->err = retbuf->ret_errno;
	tevent_req_done(req);
}

static ssize_t aio_fork_pwrite_recv(struct tevent_req *req, int *err)
{
	struct aio_fork_pwrite_state *state = tevent_req_data(
		req, struct aio_fork_pwrite_state);

	if (tevent_req_is_unix_error(req, err)) {
		return -1;
	}
	if (state->ret == -1) {
		*err = state->err;
	}
	return state->ret;
}

struct aio_fork_fsync_state {
	struct aio_child *child;
	ssize_t ret;
	int err;
};

static void aio_fork_fsync_done(struct tevent_req *subreq);

static struct tevent_req *aio_fork_fsync_send(
	struct vfs_handle_struct *handle, TALLOC_CTX *mem_ctx,
	struct tevent_context *ev, struct files_struct *fsp)
{
	struct tevent_req *req, *subreq;
	struct aio_fork_fsync_state *state;
	struct rw_cmd cmd;
	ssize_t written;
	int err;
	struct aio_fork_config *config;

	SMB_VFS_HANDLE_GET_DATA(handle, config,
				struct aio_fork_config,
				return NULL);

	req = tevent_req_create(mem_ctx, &state, struct aio_fork_fsync_state);
	if (req == NULL) {
		return NULL;
	}

	err = get_idle_child(handle, &state->child);
	if (err != 0) {
		tevent_req_error(req, err);
		return tevent_req_post(req, ev);
	}

	ZERO_STRUCT(cmd);
	cmd.cmd = FSYNC_CMD;
	cmd.erratic_testing_mode = config->erratic_testing_mode;

	DEBUG(10, ("sending fd %d to child %d\n", fsp->fh->fd,
		   (int)state->child->pid));

	/*
	 * Not making this async. We're writing into an empty unix
	 * domain socket. This should never block.
	 */
	written = write_fd(state->child->sockfd, &cmd, sizeof(cmd),
			   fsp->fh->fd);
	if (written == -1) {
		err = errno;

		TALLOC_FREE(state->child);

		DEBUG(10, ("write_fd failed: %s\n", strerror(err)));
		tevent_req_error(req, err);
		return tevent_req_post(req, ev);
	}

	subreq = read_packet_send(state, ev, state->child->sockfd,
				  sizeof(struct rw_ret), NULL, NULL);
	if (tevent_req_nomem(subreq, req)) {
		TALLOC_FREE(state->child); /* we sent sth down */
		return tevent_req_post(req, ev);
	}
	tevent_req_set_callback(subreq, aio_fork_fsync_done, req);
	return req;
}

static void aio_fork_fsync_done(struct tevent_req *subreq)
{
	struct tevent_req *req = tevent_req_callback_data(
		subreq, struct tevent_req);
	struct aio_fork_fsync_state *state = tevent_req_data(
		req, struct aio_fork_fsync_state);
	ssize_t nread;
	uint8_t *buf;
	int err;
	struct rw_ret *retbuf;

	nread = read_packet_recv(subreq, talloc_tos(), &buf, &err);
	TALLOC_FREE(subreq);
	if (nread == -1) {
		TALLOC_FREE(state->child);
		tevent_req_error(req, err);
		return;
	}

	state->child->busy = false;

	retbuf = (struct rw_ret *)buf;
	state->ret = retbuf->size;
	state->err = retbuf->ret_errno;
	tevent_req_done(req);
}

static int aio_fork_fsync_recv(struct tevent_req *req, int *err)
{
	struct aio_fork_fsync_state *state = tevent_req_data(
		req, struct aio_fork_fsync_state);

	if (tevent_req_is_unix_error(req, err)) {
		return -1;
	}
	if (state->ret == -1) {
		*err = state->err;
	}
	return state->ret;
}

static int aio_fork_connect(vfs_handle_struct *handle, const char *service,
			    const char *user)
{
	int ret;
	struct aio_fork_config *config;
	ret = SMB_VFS_NEXT_CONNECT(handle, service, user);

	if (ret < 0) {
		return ret;
	}

	config = talloc_zero(handle->conn, struct aio_fork_config);
	if (!config) {
		SMB_VFS_NEXT_DISCONNECT(handle);
		DEBUG(0, ("talloc_zero() failed\n"));
		return -1;
	}

	config->erratic_testing_mode = lp_parm_bool(SNUM(handle->conn), "vfs_aio_fork",
						    "erratic_testing_mode", false);
	
	SMB_VFS_HANDLE_SET_DATA(handle, config,
				NULL, struct aio_fork_config,
				return -1);

	/*********************************************************************
	 * How many threads to initialize ?
	 * 100 per process seems insane as a default until you realize that
	 * (a) Threads terminate after 1 second when idle.
	 * (b) Throttling is done in SMB2 via the crediting algorithm.
	 * (c) SMB1 clients are limited to max_mux (50) outstanding
	 *     requests and Windows clients don't use this anyway.
	 * Essentially we want this to be unlimited unless smb.conf
	 * says different.
	 *********************************************************************/
	aio_pending_size = 100;
	return 0;
}

static struct vfs_fn_pointers vfs_aio_fork_fns = {
	.connect_fn = aio_fork_connect,
	.pread_send_fn = aio_fork_pread_send,
	.pread_recv_fn = aio_fork_pread_recv,
	.pwrite_send_fn = aio_fork_pwrite_send,
	.pwrite_recv_fn = aio_fork_pwrite_recv,
	.fsync_send_fn = aio_fork_fsync_send,
	.fsync_recv_fn = aio_fork_fsync_recv,
};

NTSTATUS vfs_aio_fork_init(void);
NTSTATUS vfs_aio_fork_init(void)
{
	return smb_register_vfs(SMB_VFS_INTERFACE_VERSION,
				"aio_fork", &vfs_aio_fork_fns);
}