summaryrefslogtreecommitdiff
path: root/engines/net.c
blob: 0c90e1cac74f77d80dbdae4280c691bda06ec864 (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
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
/*
 * net engine
 *
 * IO engine that reads/writes to/from sockets.
 *
 */
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <signal.h>
#include <errno.h>
#include <assert.h>
#include <netinet/in.h>
#include <netinet/tcp.h>
#include <arpa/inet.h>
#include <netdb.h>
#include <sys/poll.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <sys/socket.h>
#include <sys/un.h>

#include "../fio.h"

struct netio_data {
	int listenfd;
	int use_splice;
	int pipes[2];
	struct sockaddr_in addr;
	struct sockaddr_un addr_un;
};

struct netio_options {
	struct thread_data *td;
	unsigned int port;
	unsigned int proto;
	unsigned int listen;
	unsigned int pingpong;
	unsigned int nodelay;
	unsigned int ttl;
	char * interface;
};

struct udp_close_msg {
	uint32_t magic;
	uint32_t cmd;
};

enum {
	FIO_LINK_CLOSE = 0x89,
	FIO_LINK_OPEN_CLOSE_MAGIC = 0x6c696e6b,
	FIO_LINK_OPEN = 0x98,

	FIO_TYPE_TCP	= 1,
	FIO_TYPE_UDP	= 2,
	FIO_TYPE_UNIX	= 3,
};

static int str_hostname_cb(void *data, const char *input);
static struct fio_option options[] = {
	{
		.name	= "hostname",
		.lname	= "net engine hostname",
		.type	= FIO_OPT_STR_STORE,
		.cb	= str_hostname_cb,
		.help	= "Hostname for net IO engine",
		.category = FIO_OPT_C_ENGINE,
		.group	= FIO_OPT_G_NETIO,
	},
	{
		.name	= "port",
		.lname	= "net engine port",
		.type	= FIO_OPT_INT,
		.off1	= offsetof(struct netio_options, port),
		.minval	= 1,
		.maxval	= 65535,
		.help	= "Port to use for TCP or UDP net connections",
		.category = FIO_OPT_C_ENGINE,
		.group	= FIO_OPT_G_NETIO,
	},
	{
		.name	= "protocol",
		.lname	= "net engine protocol",
		.alias	= "proto",
		.type	= FIO_OPT_STR,
		.off1	= offsetof(struct netio_options, proto),
		.help	= "Network protocol to use",
		.def	= "tcp",
		.posval = {
			  { .ival = "tcp",
			    .oval = FIO_TYPE_TCP,
			    .help = "Transmission Control Protocol",
			  },
			  { .ival = "udp",
			    .oval = FIO_TYPE_UDP,
			    .help = "User Datagram Protocol",
			  },
			  { .ival = "unix",
			    .oval = FIO_TYPE_UNIX,
			    .help = "UNIX domain socket",
			  },
		},
		.category = FIO_OPT_C_ENGINE,
		.group	= FIO_OPT_G_NETIO,
	},
#ifdef CONFIG_TCP_NODELAY
	{
		.name	= "nodelay",
		.type	= FIO_OPT_BOOL,
		.off1	= offsetof(struct netio_options, nodelay),
		.help	= "Use TCP_NODELAY on TCP connections",
		.category = FIO_OPT_C_ENGINE,
		.group	= FIO_OPT_G_NETIO,
	},
#endif
	{
		.name	= "listen",
		.lname	= "net engine listen",
		.type	= FIO_OPT_STR_SET,
		.off1	= offsetof(struct netio_options, listen),
		.help	= "Listen for incoming TCP connections",
		.category = FIO_OPT_C_ENGINE,
		.group	= FIO_OPT_G_NETIO,
	},
	{
		.name	= "pingpong",
		.type	= FIO_OPT_STR_SET,
		.off1	= offsetof(struct netio_options, pingpong),
		.help	= "Ping-pong IO requests",
		.category = FIO_OPT_C_ENGINE,
		.group	= FIO_OPT_G_NETIO,
	},
	{
		.name	= "interface",
		.lname	= "net engine interface",
		.type	= FIO_OPT_STR_STORE,
		.off1	= offsetof(struct netio_options, interface),
		.help	= "Network interface to use",
		.category = FIO_OPT_C_ENGINE,
		.group	= FIO_OPT_G_NETIO,
	},
	{
		.name	= "ttl",
		.lname	= "net engine multicast ttl",
		.type	= FIO_OPT_INT,
		.off1	= offsetof(struct netio_options, ttl),
		.def    = "1",
		.minval	= 0,
		.help	= "Time-to-live value for outgoing UDP multicast packets",
		.category = FIO_OPT_C_ENGINE,
		.group	= FIO_OPT_G_NETIO,
	},
	{
		.name	= NULL,
	},
};

/*
 * Return -1 for error and 'nr events' for a positive number
 * of events
 */
static int poll_wait(struct thread_data *td, int fd, short events)
{
	struct pollfd pfd;
	int ret;

	while (!td->terminate) {
		pfd.fd = fd;
		pfd.events = events;
		ret = poll(&pfd, 1, -1);
		if (ret < 0) {
			if (errno == EINTR)
				break;

			td_verror(td, errno, "poll");
			return -1;
		} else if (!ret)
			continue;

		break;
	}

	if (pfd.revents & events)
		return 1;

	return -1;
}

static int fio_netio_is_multicast(const char *mcaddr)
{
	in_addr_t addr = inet_network(mcaddr);
	if (addr == -1)
		return 0;

	if (inet_network("224.0.0.0") <= addr &&
	    inet_network("239.255.255.255") >= addr)
		return 1;

	return 0;
}


static int fio_netio_prep(struct thread_data *td, struct io_u *io_u)
{
	struct netio_options *o = td->eo;

	/*
	 * Make sure we don't see spurious reads to a receiver, and vice versa
	 */
	if (o->proto == FIO_TYPE_TCP)
		return 0;

	if ((o->listen && io_u->ddir == DDIR_WRITE) ||
	    (!o->listen && io_u->ddir == DDIR_READ)) {
		td_verror(td, EINVAL, "bad direction");
		return 1;
	}

	return 0;
}

#ifdef CONFIG_LINUX_SPLICE
static int splice_io_u(int fdin, int fdout, unsigned int len)
{
	int bytes = 0;

	while (len) {
		int ret = splice(fdin, NULL, fdout, NULL, len, 0);

		if (ret < 0) {
			if (!bytes)
				bytes = ret;

			break;
		} else if (!ret)
			break;

		bytes += ret;
		len -= ret;
	}

	return bytes;
}

/*
 * Receive bytes from a socket and fill them into the internal pipe
 */
static int splice_in(struct thread_data *td, struct io_u *io_u)
{
	struct netio_data *nd = td->io_ops->data;

	return splice_io_u(io_u->file->fd, nd->pipes[1], io_u->xfer_buflen);
}

/*
 * Transmit 'len' bytes from the internal pipe
 */
static int splice_out(struct thread_data *td, struct io_u *io_u,
		      unsigned int len)
{
	struct netio_data *nd = td->io_ops->data;

	return splice_io_u(nd->pipes[0], io_u->file->fd, len);
}

static int vmsplice_io_u(struct io_u *io_u, int fd, unsigned int len)
{
	struct iovec iov = {
		.iov_base = io_u->xfer_buf,
		.iov_len = len,
	};
	int bytes = 0;

	while (iov.iov_len) {
		int ret = vmsplice(fd, &iov, 1, SPLICE_F_MOVE);

		if (ret < 0) {
			if (!bytes)
				bytes = ret;
			break;
		} else if (!ret)
			break;

		iov.iov_len -= ret;
		iov.iov_base += ret;
		bytes += ret;
	}

	return bytes;

}

/*
 * vmsplice() pipe to io_u buffer
 */
static int vmsplice_io_u_out(struct thread_data *td, struct io_u *io_u,
			     unsigned int len)
{
	struct netio_data *nd = td->io_ops->data;

	return vmsplice_io_u(io_u, nd->pipes[0], len);
}

/*
 * vmsplice() io_u to pipe
 */
static int vmsplice_io_u_in(struct thread_data *td, struct io_u *io_u)
{
	struct netio_data *nd = td->io_ops->data;

	return vmsplice_io_u(io_u, nd->pipes[1], io_u->xfer_buflen);
}

/*
 * splice receive - transfer socket data into a pipe using splice, then map
 * that pipe data into the io_u using vmsplice.
 */
static int fio_netio_splice_in(struct thread_data *td, struct io_u *io_u)
{
	int ret;

	ret = splice_in(td, io_u);
	if (ret > 0)
		return vmsplice_io_u_out(td, io_u, ret);

	return ret;
}

/*
 * splice transmit - map data from the io_u into a pipe by using vmsplice,
 * then transfer that pipe to a socket using splice.
 */
static int fio_netio_splice_out(struct thread_data *td, struct io_u *io_u)
{
	int ret;

	ret = vmsplice_io_u_in(td, io_u);
	if (ret > 0)
		return splice_out(td, io_u, ret);

	return ret;
}
#else
static int fio_netio_splice_in(struct thread_data *td, struct io_u *io_u)
{
	errno = EOPNOTSUPP;
	return -1;
}

static int fio_netio_splice_out(struct thread_data *td, struct io_u *io_u)
{
	errno = EOPNOTSUPP;
	return -1;
}
#endif

static int fio_netio_send(struct thread_data *td, struct io_u *io_u)
{
	struct netio_data *nd = td->io_ops->data;
	struct netio_options *o = td->eo;
	int ret, flags = 0;

	do {
		if (o->proto == FIO_TYPE_UDP) {
			struct sockaddr *to = (struct sockaddr *) &nd->addr;

			ret = sendto(io_u->file->fd, io_u->xfer_buf,
					io_u->xfer_buflen, flags, to,
					sizeof(*to));
		} else {
			/*
			 * if we are going to write more, set MSG_MORE
			 */
#ifdef MSG_MORE
			if ((td->this_io_bytes[DDIR_WRITE] + io_u->xfer_buflen <
			    td->o.size) && !o->pingpong)
				flags |= MSG_MORE;
#endif
			ret = send(io_u->file->fd, io_u->xfer_buf,
					io_u->xfer_buflen, flags);
		}
		if (ret > 0)
			break;

		ret = poll_wait(td, io_u->file->fd, POLLOUT);
		if (ret <= 0)
			break;
	} while (1);

	return ret;
}

static int is_udp_close(struct io_u *io_u, int len)
{
	struct udp_close_msg *msg;

	if (len != sizeof(struct udp_close_msg))
		return 0;

	msg = io_u->xfer_buf;
	if (ntohl(msg->magic) != FIO_LINK_OPEN_CLOSE_MAGIC)
		return 0;
	if (ntohl(msg->cmd) != FIO_LINK_CLOSE)
		return 0;

	return 1;
}

static int fio_netio_recv(struct thread_data *td, struct io_u *io_u)
{
	struct netio_data *nd = td->io_ops->data;
	struct netio_options *o = td->eo;
	int ret, flags = 0;

	do {
		if (o->proto == FIO_TYPE_UDP) {
			socklen_t l;
			socklen_t *len = &l;
			struct sockaddr *from;

			if (o->listen) {
				from = (struct sockaddr *) &nd->addr;
				*len = sizeof(nd->addr);
			} else {
				from = NULL;
				len = NULL;
			}

			ret = recvfrom(io_u->file->fd, io_u->xfer_buf,
					io_u->xfer_buflen, flags, from, len);
			if (is_udp_close(io_u, ret)) {
				td->done = 1;
				return 0;
			}
		} else {
			ret = recv(io_u->file->fd, io_u->xfer_buf,
					io_u->xfer_buflen, flags);
		}
		if (ret > 0)
			break;
		else if (!ret && (flags & MSG_WAITALL))
			break;

		ret = poll_wait(td, io_u->file->fd, POLLIN);
		if (ret <= 0)
			break;
		flags |= MSG_WAITALL;
	} while (1);

	return ret;
}

static int __fio_netio_queue(struct thread_data *td, struct io_u *io_u,
			     enum fio_ddir ddir)
{
	struct netio_data *nd = td->io_ops->data;
	struct netio_options *o = td->eo;
	int ret;

	if (ddir == DDIR_WRITE) {
		if (!nd->use_splice || o->proto == FIO_TYPE_UDP ||
		    o->proto == FIO_TYPE_UNIX)
			ret = fio_netio_send(td, io_u);
		else
			ret = fio_netio_splice_out(td, io_u);
	} else if (ddir == DDIR_READ) {
		if (!nd->use_splice || o->proto == FIO_TYPE_UDP ||
		    o->proto == FIO_TYPE_UNIX)
			ret = fio_netio_recv(td, io_u);
		else
			ret = fio_netio_splice_in(td, io_u);
	} else
		ret = 0;	/* must be a SYNC */

	if (ret != (int) io_u->xfer_buflen) {
		if (ret >= 0) {
			io_u->resid = io_u->xfer_buflen - ret;
			io_u->error = 0;
			return FIO_Q_COMPLETED;
		} else {
			int err = errno;

			if (ddir == DDIR_WRITE && err == EMSGSIZE)
				return FIO_Q_BUSY;

			io_u->error = err;
		}
	}

	if (io_u->error)
		td_verror(td, io_u->error, "xfer");

	return FIO_Q_COMPLETED;
}

static int fio_netio_queue(struct thread_data *td, struct io_u *io_u)
{
	struct netio_options *o = td->eo;
	int ret;

	fio_ro_check(td, io_u);

	ret = __fio_netio_queue(td, io_u, io_u->ddir);
	if (!o->pingpong || ret != FIO_Q_COMPLETED)
		return ret;

	/*
	 * For ping-pong mode, receive or send reply as needed
	 */
	if (td_read(td) && io_u->ddir == DDIR_READ)
		ret = __fio_netio_queue(td, io_u, DDIR_WRITE);
	else if (td_write(td) && io_u->ddir == DDIR_WRITE)
		ret = __fio_netio_queue(td, io_u, DDIR_READ);

	return ret;
}

static int fio_netio_connect(struct thread_data *td, struct fio_file *f)
{
	struct netio_data *nd = td->io_ops->data;
	struct netio_options *o = td->eo;
	int type, domain;

	if (o->proto == FIO_TYPE_TCP) {
		domain = AF_INET;
		type = SOCK_STREAM;
	} else if (o->proto == FIO_TYPE_UDP) {
		domain = AF_INET;
		type = SOCK_DGRAM;
	} else if (o->proto == FIO_TYPE_UNIX) {
		domain = AF_UNIX;
		type = SOCK_STREAM;
	} else {
		log_err("fio: bad network type %d\n", o->proto);
		f->fd = -1;
		return 1;
	}

	f->fd = socket(domain, type, 0);
	if (f->fd < 0) {
		td_verror(td, errno, "socket");
		return 1;
	}

#ifdef CONFIG_TCP_NODELAY
	if (o->nodelay && o->proto == FIO_TYPE_TCP) {
		int optval = 1;

		if (setsockopt(f->fd, IPPROTO_TCP, TCP_NODELAY, (void *) &optval, sizeof(int)) < 0) {
			log_err("fio: cannot set TCP_NODELAY option on socket (%s), disable with 'nodelay=0'\n", strerror(errno));
			return 1;
		}
	}
#endif

	if (o->proto == FIO_TYPE_UDP) {
		if (!fio_netio_is_multicast(td->o.filename))
			return 0;

		if (o->interface) {
			struct in_addr interface_addr;
			if (inet_aton(o->interface, &interface_addr) == 0) {
				log_err("fio: interface not valid interface IP\n");
				close(f->fd);
				return 1;
			}
			if (setsockopt(f->fd, IPPROTO_IP, IP_MULTICAST_IF, &interface_addr, sizeof(interface_addr)) < 0) {
				td_verror(td, errno, "setsockopt IP_MULTICAST_IF");
				close(f->fd);
				return 1;
			}
		}
		if (setsockopt(f->fd, IPPROTO_IP, IP_MULTICAST_TTL, &o->ttl, sizeof(o->ttl)) < 0) {
			td_verror(td, errno, "setsockopt IP_MULTICAST_TTL");
			close(f->fd);
			return 1;
		}
		return 0;
	} else if (o->proto == FIO_TYPE_TCP) {
		socklen_t len = sizeof(nd->addr);

		if (connect(f->fd, (struct sockaddr *) &nd->addr, len) < 0) {
			td_verror(td, errno, "connect");
			close(f->fd);
			return 1;
		}
	} else {
		struct sockaddr_un *addr = &nd->addr_un;
		socklen_t len;

		len = sizeof(addr->sun_family) + strlen(addr->sun_path) + 1;

		if (connect(f->fd, (struct sockaddr *) addr, len) < 0) {
			td_verror(td, errno, "connect");
			close(f->fd);
			return 1;
		}
	}

	return 0;
}

static int fio_netio_accept(struct thread_data *td, struct fio_file *f)
{
	struct netio_data *nd = td->io_ops->data;
	struct netio_options *o = td->eo;
	socklen_t socklen = sizeof(nd->addr);
	int state;

	if (o->proto == FIO_TYPE_UDP) {
		f->fd = nd->listenfd;
		return 0;
	}

	state = td->runstate;
	td_set_runstate(td, TD_SETTING_UP);

	log_info("fio: waiting for connection\n");

	if (poll_wait(td, nd->listenfd, POLLIN) < 0)
		goto err;

	f->fd = accept(nd->listenfd, (struct sockaddr *) &nd->addr, &socklen);
	if (f->fd < 0) {
		td_verror(td, errno, "accept");
		goto err;
	}

#ifdef CONFIG_TCP_NODELAY
	if (o->nodelay && o->proto == FIO_TYPE_TCP) {
		int optval = 1;

		if (setsockopt(f->fd, IPPROTO_TCP, TCP_NODELAY, (void *) &optval, sizeof(int)) < 0) {
			log_err("fio: cannot set TCP_NODELAY option on socket (%s), disable with 'nodelay=0'\n", strerror(errno));
			return 1;
		}
	}
#endif

	reset_all_stats(td);
	td_set_runstate(td, state);
	return 0;
err:
	td_set_runstate(td, state);
	return 1;
}

static void fio_netio_udp_close(struct thread_data *td, struct fio_file *f)
{
	struct netio_data *nd = td->io_ops->data;
	struct udp_close_msg msg;
	struct sockaddr *to = (struct sockaddr *) &nd->addr;
	int ret;

	msg.magic = htonl(FIO_LINK_OPEN_CLOSE_MAGIC);
	msg.cmd = htonl(FIO_LINK_CLOSE);

	ret = sendto(f->fd, (void *) &msg, sizeof(msg), MSG_WAITALL, to,
			sizeof(nd->addr));
	if (ret < 0)
		td_verror(td, errno, "sendto udp link close");
}

static int fio_netio_close_file(struct thread_data *td, struct fio_file *f)
{
	struct netio_options *o = td->eo;

	/*
	 * If this is an UDP connection, notify the receiver that we are
	 * closing down the link
	 */
	if (o->proto == FIO_TYPE_UDP)
		fio_netio_udp_close(td, f);

	return generic_close_file(td, f);
}

static int fio_netio_udp_recv_open(struct thread_data *td, struct fio_file *f)
{
	struct netio_data *nd = td->io_ops->data;
	struct udp_close_msg msg;
	struct sockaddr *to = (struct sockaddr *) &nd->addr;
	socklen_t len = sizeof(nd->addr);
	int ret;

	ret = recvfrom(f->fd, (void *) &msg, sizeof(msg), MSG_WAITALL, to, &len);
	if (ret < 0) {
		td_verror(td, errno, "recvfrom udp link open");
		return ret;
	}

	if (ntohl(msg.magic) != FIO_LINK_OPEN_CLOSE_MAGIC ||
	    ntohl(msg.cmd) != FIO_LINK_OPEN) {
		log_err("fio: bad udp open magic %x/%x\n", ntohl(msg.magic),
								ntohl(msg.cmd));
		return -1;
	}

	return 0;
}

static int fio_netio_udp_send_open(struct thread_data *td, struct fio_file *f)
{
	struct netio_data *nd = td->io_ops->data;
	struct udp_close_msg msg;
	struct sockaddr *to = (struct sockaddr *) &nd->addr;
	int ret;

	msg.magic = htonl(FIO_LINK_OPEN_CLOSE_MAGIC);
	msg.cmd = htonl(FIO_LINK_OPEN);

	ret = sendto(f->fd, (void *) &msg, sizeof(msg), MSG_WAITALL, to,
			sizeof(nd->addr));
	if (ret < 0) {
		td_verror(td, errno, "sendto udp link open");
		return ret;
	}

	return 0;
}

static int fio_netio_open_file(struct thread_data *td, struct fio_file *f)
{
	int ret;
	struct netio_options *o = td->eo;

	if (o->listen)
		ret = fio_netio_accept(td, f);
	else
		ret = fio_netio_connect(td, f);

	if (ret) {
		f->fd = -1;
		return ret;
	}

	if (o->proto == FIO_TYPE_UDP) {
		if (td_write(td))
			ret = fio_netio_udp_send_open(td, f);
		else {
			int state;

			state = td->runstate;
			td_set_runstate(td, TD_SETTING_UP);
			ret = fio_netio_udp_recv_open(td, f);
			td_set_runstate(td, state);
		}
	}

	if (ret)
		fio_netio_close_file(td, f);

	return ret;
}

static int fio_netio_setup_connect_inet(struct thread_data *td,
					const char *host, unsigned short port)
{
	struct netio_data *nd = td->io_ops->data;

	if (!host) {
		log_err("fio: connect with no host to connect to.\n");
		if (td_read(td))
			log_err("fio: did you forget to set 'listen'?\n");

		td_verror(td, EINVAL, "no hostname= set");
		return 1;
	}

	nd->addr.sin_family = AF_INET;
	nd->addr.sin_port = htons(port);

	if (inet_aton(host, &nd->addr.sin_addr) != 1) {
		struct hostent *hent;

		hent = gethostbyname(host);
		if (!hent) {
			td_verror(td, errno, "gethostbyname");
			return 1;
		}

		memcpy(&nd->addr.sin_addr, hent->h_addr, 4);
	}

	return 0;
}

static int fio_netio_setup_connect_unix(struct thread_data *td,
					const char *path)
{
	struct netio_data *nd = td->io_ops->data;
	struct sockaddr_un *soun = &nd->addr_un;

	soun->sun_family = AF_UNIX;
	strcpy(soun->sun_path, path);
	return 0;
}

static int fio_netio_setup_connect(struct thread_data *td)
{
	struct netio_options *o = td->eo;

	if (o->proto == FIO_TYPE_UDP || o->proto == FIO_TYPE_TCP)
		return fio_netio_setup_connect_inet(td, td->o.filename,o->port);
	else
		return fio_netio_setup_connect_unix(td, td->o.filename);
}

static int fio_netio_setup_listen_unix(struct thread_data *td, const char *path)
{
	struct netio_data *nd = td->io_ops->data;
	struct sockaddr_un *addr = &nd->addr_un;
	mode_t mode;
	int len, fd;

	fd = socket(AF_UNIX, SOCK_STREAM, 0);
	if (fd < 0) {
		log_err("fio: socket: %s\n", strerror(errno));
		return -1;
	}

	mode = umask(000);

	memset(addr, 0, sizeof(*addr));
	addr->sun_family = AF_UNIX;
	strcpy(addr->sun_path, path);
	unlink(path);

	len = sizeof(addr->sun_family) + strlen(path) + 1;

	if (bind(fd, (struct sockaddr *) addr, len) < 0) {
		log_err("fio: bind: %s\n", strerror(errno));
		close(fd);
		return -1;
	}

	umask(mode);
	nd->listenfd = fd;
	return 0;
}

static int fio_netio_setup_listen_inet(struct thread_data *td, short port)
{
	struct netio_data *nd = td->io_ops->data;
	struct netio_options *o = td->eo;
	struct ip_mreq mr;
	struct sockaddr_in sin;
	int fd, opt, type;

	memset(&sin, 0, sizeof(sin));
	if (o->proto == FIO_TYPE_TCP)
		type = SOCK_STREAM;
	else
		type = SOCK_DGRAM;

	fd = socket(AF_INET, type, 0);
	if (fd < 0) {
		td_verror(td, errno, "socket");
		return 1;
	}

	opt = 1;
	if (setsockopt(fd, SOL_SOCKET, SO_REUSEADDR, (void *) &opt, sizeof(opt)) < 0) {
		td_verror(td, errno, "setsockopt");
		close(fd);
		return 1;
	}
#ifdef SO_REUSEPORT
	if (setsockopt(fd, SOL_SOCKET, SO_REUSEPORT, (void *) &opt, sizeof(opt)) < 0) {
		td_verror(td, errno, "setsockopt");
		close(fd);
		return 1;
	}
#endif

	if (td->o.filename){
		if(o->proto != FIO_TYPE_UDP ||
		   !fio_netio_is_multicast(td->o.filename)) {
			log_err("fio: hostname not valid for non-multicast inbound network IO\n");
			close(fd);
			return 1;
		}

		inet_aton(td->o.filename, &sin.sin_addr);

		mr.imr_multiaddr = sin.sin_addr;
		if (o->interface) {
			if (inet_aton(o->interface, &mr.imr_interface) == 0) {
				log_err("fio: interface not valid interface IP\n");
				close(fd);
				return 1;
			}
		} else {
			mr.imr_interface.s_addr = htonl(INADDR_ANY);
		}
		if (setsockopt(fd, IPPROTO_IP, IP_ADD_MEMBERSHIP, &mr, sizeof(mr)) < 0) {
			td_verror(td, errno, "setsockopt IP_ADD_MEMBERSHIP");
			close(fd);
			return 1;
		}
	}

	nd->addr.sin_family = AF_INET;
	nd->addr.sin_addr.s_addr = sin.sin_addr.s_addr ? sin.sin_addr.s_addr : htonl(INADDR_ANY);
	nd->addr.sin_port = htons(port);

	if (bind(fd, (struct sockaddr *) &nd->addr, sizeof(nd->addr)) < 0) {
		td_verror(td, errno, "bind");
		return 1;
	}

	nd->listenfd = fd;
	return 0;
}

static int fio_netio_setup_listen(struct thread_data *td)
{
	struct netio_data *nd = td->io_ops->data;
	struct netio_options *o = td->eo;
	int ret;

	if (o->proto == FIO_TYPE_UDP || o->proto == FIO_TYPE_TCP)
		ret = fio_netio_setup_listen_inet(td, o->port);
	else
		ret = fio_netio_setup_listen_unix(td, td->o.filename);

	if (ret)
		return ret;
	if (o->proto == FIO_TYPE_UDP)
		return 0;

	if (listen(nd->listenfd, 10) < 0) {
		td_verror(td, errno, "listen");
		nd->listenfd = -1;
		return 1;
	}

	return 0;
}

static int fio_netio_init(struct thread_data *td)
{
	struct netio_options *o = td->eo;
	int ret;

#ifdef WIN32
	WSADATA wsd;
	WSAStartup(MAKEWORD(2,2), &wsd);
#endif

	if (td_random(td)) {
		log_err("fio: network IO can't be random\n");
		return 1;
	}

	if (o->proto == FIO_TYPE_UNIX && o->port) {
		log_err("fio: network IO port not valid with unix socket\n");
		return 1;
	} else if (o->proto != FIO_TYPE_UNIX && !o->port) {
		log_err("fio: network IO requires port for tcp or udp\n");
		return 1;
	}

	if (o->proto != FIO_TYPE_TCP) {
		if (o->listen) {
			log_err("fio: listen only valid for TCP proto IO\n");
			return 1;
		}
		if (td_rw(td)) {
			log_err("fio: datagram network connections must be"
				   " read OR write\n");
			return 1;
		}
		if (o->proto == FIO_TYPE_UNIX && !td->o.filename) {
			log_err("fio: UNIX sockets need host/filename\n");
			return 1;
		}
		o->listen = td_read(td);
	}

	if (o->listen)
		ret = fio_netio_setup_listen(td);
	else
		ret = fio_netio_setup_connect(td);

	return ret;
}

static void fio_netio_cleanup(struct thread_data *td)
{
	struct netio_data *nd = td->io_ops->data;

	if (nd) {
		if (nd->listenfd != -1)
			close(nd->listenfd);
		if (nd->pipes[0] != -1)
			close(nd->pipes[0]);
		if (nd->pipes[1] != -1)
			close(nd->pipes[1]);

		free(nd);
	}
}

static int fio_netio_setup(struct thread_data *td)
{
	struct netio_data *nd;

	if (!td->files_index) {
		add_file(td, td->o.filename ?: "net");
		td->o.nr_files = td->o.nr_files ?: 1;
	}

	if (!td->io_ops->data) {
		nd = malloc(sizeof(*nd));;

		memset(nd, 0, sizeof(*nd));
		nd->listenfd = -1;
		nd->pipes[0] = nd->pipes[1] = -1;
		td->io_ops->data = nd;
	}

	return 0;
}

static void fio_netio_terminate(struct thread_data *td)
{
	kill(td->pid, SIGUSR2);
}

#ifdef CONFIG_LINUX_SPLICE
static int fio_netio_setup_splice(struct thread_data *td)
{
	struct netio_data *nd;

	fio_netio_setup(td);

	nd = td->io_ops->data;
	if (nd) {
		if (pipe(nd->pipes) < 0)
			return 1;

		nd->use_splice = 1;
		return 0;
	}

	return 1;
}

static struct ioengine_ops ioengine_splice = {
	.name			= "netsplice",
	.version		= FIO_IOOPS_VERSION,
	.prep			= fio_netio_prep,
	.queue			= fio_netio_queue,
	.setup			= fio_netio_setup_splice,
	.init			= fio_netio_init,
	.cleanup		= fio_netio_cleanup,
	.open_file		= fio_netio_open_file,
	.close_file		= fio_netio_close_file,
	.terminate		= fio_netio_terminate,
	.options		= options,
	.option_struct_size	= sizeof(struct netio_options),
	.flags			= FIO_SYNCIO | FIO_DISKLESSIO | FIO_UNIDIR |
				  FIO_PIPEIO,
};
#endif

static struct ioengine_ops ioengine_rw = {
	.name			= "net",
	.version		= FIO_IOOPS_VERSION,
	.prep			= fio_netio_prep,
	.queue			= fio_netio_queue,
	.setup			= fio_netio_setup,
	.init			= fio_netio_init,
	.cleanup		= fio_netio_cleanup,
	.open_file		= fio_netio_open_file,
	.close_file		= fio_netio_close_file,
	.terminate		= fio_netio_terminate,
	.options		= options,
	.option_struct_size	= sizeof(struct netio_options),
	.flags			= FIO_SYNCIO | FIO_DISKLESSIO | FIO_UNIDIR |
				  FIO_PIPEIO | FIO_BIT_BASED,
};

static int str_hostname_cb(void *data, const char *input)
{
	struct netio_options *o = data;

	if (o->td->o.filename)
		free(o->td->o.filename);
	o->td->o.filename = strdup(input);
	return 0;
}

static void fio_init fio_netio_register(void)
{
	register_ioengine(&ioengine_rw);
#ifdef CONFIG_LINUX_SPLICE
	register_ioengine(&ioengine_splice);
#endif
}

static void fio_exit fio_netio_unregister(void)
{
	unregister_ioengine(&ioengine_rw);
#ifdef CONFIG_LINUX_SPLICE
	unregister_ioengine(&ioengine_splice);
#endif
}