aboutsummaryrefslogtreecommitdiffstats
path: root/gpxe/src/net/tcp.c
blob: da8e87b4b4360b0fe90d3d6a7c4268ee3541f233 (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
#include <string.h>
#include <stdlib.h>
#include <stdio.h>
#include <assert.h>
#include <errno.h>
#include <byteswap.h>
#include <gpxe/timer.h>
#include <gpxe/iobuf.h>
#include <gpxe/malloc.h>
#include <gpxe/retry.h>
#include <gpxe/refcnt.h>
#include <gpxe/xfer.h>
#include <gpxe/open.h>
#include <gpxe/uri.h>
#include <gpxe/tcpip.h>
#include <gpxe/tcp.h>

/** @file
 *
 * TCP protocol
 *
 */

/** A TCP connection */
struct tcp_connection {
	/** Reference counter */
	struct refcnt refcnt;
	/** List of TCP connections */
	struct list_head list;

	/** Data transfer interface */
	struct xfer_interface xfer;
	/** Data transfer interface closed flag */
	int xfer_closed;

	/** Remote socket address */
	struct sockaddr_tcpip peer;
	/** Local port, in network byte order */
	unsigned int local_port;

	/** Current TCP state */
	unsigned int tcp_state;
	/** Previous TCP state
	 *
	 * Maintained only for debug messages
	 */
	unsigned int prev_tcp_state;
	/** Current sequence number
	 *
	 * Equivalent to SND.UNA in RFC 793 terminology.
	 */
	uint32_t snd_seq;
	/** Unacknowledged sequence count
	 *
	 * Equivalent to (SND.NXT-SND.UNA) in RFC 793 terminology.
	 */
	uint32_t snd_sent;
	/** Send window
	 *
	 * Equivalent to SND.WND in RFC 793 terminology
	 */
	uint32_t snd_win;
	/** Current acknowledgement number
	 *
	 * Equivalent to RCV.NXT in RFC 793 terminology.
	 */
	uint32_t rcv_ack;
	/** Most recent received timestamp
	 *
	 * Equivalent to TS.Recent in RFC 1323 terminology.
	 */
	uint32_t ts_recent;
	/** Timestamps enabled */
	int timestamps;

	/** Transmit queue */
	struct list_head queue;
	/** Retransmission timer */
	struct retry_timer timer;
};

/**
 * List of registered TCP connections
 */
static LIST_HEAD ( tcp_conns );

/* Forward declarations */
static struct xfer_interface_operations tcp_xfer_operations;
static void tcp_expired ( struct retry_timer *timer, int over );
static int tcp_rx_ack ( struct tcp_connection *tcp, uint32_t ack,
			uint32_t win );

/**
 * Name TCP state
 *
 * @v state		TCP state
 * @ret name		Name of TCP state
 */
static inline __attribute__ (( always_inline )) const char *
tcp_state ( int state ) {
	switch ( state ) {
	case TCP_CLOSED:		return "CLOSED";
	case TCP_LISTEN:		return "LISTEN";
	case TCP_SYN_SENT:		return "SYN_SENT";
	case TCP_SYN_RCVD:		return "SYN_RCVD";
	case TCP_ESTABLISHED:		return "ESTABLISHED";
	case TCP_FIN_WAIT_1:		return "FIN_WAIT_1";
	case TCP_FIN_WAIT_2:		return "FIN_WAIT_2";
	case TCP_CLOSING_OR_LAST_ACK:	return "CLOSING/LAST_ACK";
	case TCP_TIME_WAIT:		return "TIME_WAIT";
	case TCP_CLOSE_WAIT:		return "CLOSE_WAIT";
	default:			return "INVALID";
	}
}

/**
 * Dump TCP state transition
 *
 * @v tcp		TCP connection
 */
static inline __attribute__ (( always_inline )) void
tcp_dump_state ( struct tcp_connection *tcp ) {

	if ( tcp->tcp_state != tcp->prev_tcp_state ) {
		DBGC ( tcp, "TCP %p transitioned from %s to %s\n", tcp,
		       tcp_state ( tcp->prev_tcp_state ),
		       tcp_state ( tcp->tcp_state ) );
	}
	tcp->prev_tcp_state = tcp->tcp_state;
}

/**
 * Dump TCP flags
 *
 * @v flags		TCP flags
 */
static inline __attribute__ (( always_inline )) void
tcp_dump_flags ( struct tcp_connection *tcp, unsigned int flags ) {
	if ( flags & TCP_RST )
		DBGC ( tcp, " RST" );
	if ( flags & TCP_SYN )
		DBGC ( tcp, " SYN" );
	if ( flags & TCP_PSH )
		DBGC ( tcp, " PSH" );
	if ( flags & TCP_FIN )
		DBGC ( tcp, " FIN" );
	if ( flags & TCP_ACK )
		DBGC ( tcp, " ACK" );
}

/***************************************************************************
 *
 * Open and close
 *
 ***************************************************************************
 */

/**
 * Bind TCP connection to local port
 *
 * @v tcp		TCP connection
 * @v port		Local port number, in network-endian order
 * @ret rc		Return status code
 *
 * If the port is 0, the connection is assigned an available port
 * between 1024 and 65535.
 */
static int tcp_bind ( struct tcp_connection *tcp, unsigned int port ) {
	struct tcp_connection *existing;
	static uint16_t try_port = 1023;

	/* If no port specified, find the first available port */
	if ( ! port ) {
		while ( try_port ) {
			try_port++;
			if ( try_port < 1024 )
				continue;
			if ( tcp_bind ( tcp, htons ( try_port ) ) == 0 )
				return 0;
		}
		DBGC ( tcp, "TCP %p could not bind: no free ports\n", tcp );
		return -EADDRINUSE;
	}

	/* Attempt bind to local port */
	list_for_each_entry ( existing, &tcp_conns, list ) {
		if ( existing->local_port == port ) {
			DBGC ( tcp, "TCP %p could not bind: port %d in use\n",
			       tcp, ntohs ( port ) );
			return -EADDRINUSE;
		}
	}
	tcp->local_port = port;

	DBGC ( tcp, "TCP %p bound to port %d\n", tcp, ntohs ( port ) );
	return 0;
}

/**
 * Open a TCP connection
 *
 * @v xfer		Data transfer interface
 * @v peer		Peer socket address
 * @v local		Local socket address, or NULL
 * @ret rc		Return status code
 */
static int tcp_open ( struct xfer_interface *xfer, struct sockaddr *peer,
		      struct sockaddr *local ) {
	struct sockaddr_tcpip *st_peer = ( struct sockaddr_tcpip * ) peer;
	struct sockaddr_tcpip *st_local = ( struct sockaddr_tcpip * ) local;
	struct tcp_connection *tcp;
	unsigned int bind_port;
	int rc;

	/* Allocate and initialise structure */
	tcp = zalloc ( sizeof ( *tcp ) );
	if ( ! tcp )
		return -ENOMEM;
	DBGC ( tcp, "TCP %p allocated\n", tcp );
	xfer_init ( &tcp->xfer, &tcp_xfer_operations, &tcp->refcnt );
	tcp->prev_tcp_state = TCP_CLOSED;
	tcp->tcp_state = TCP_STATE_SENT ( TCP_SYN );
	tcp_dump_state ( tcp );
	tcp->snd_seq = random();
	INIT_LIST_HEAD ( &tcp->queue );
	tcp->timer.expired = tcp_expired;
	memcpy ( &tcp->peer, st_peer, sizeof ( tcp->peer ) );

	/* Bind to local port */
	bind_port = ( st_local ? st_local->st_port : 0 );
	if ( ( rc = tcp_bind ( tcp, bind_port ) ) != 0 )
		goto err;

	/* Start timer to initiate SYN */
	start_timer_nodelay ( &tcp->timer );

	/* Attach parent interface, transfer reference to connection
	 * list and return
	 */
	xfer_plug_plug ( &tcp->xfer, xfer );
	list_add ( &tcp->list, &tcp_conns );
	return 0;

 err:
	ref_put ( &tcp->refcnt );
	return rc;
}

/**
 * Close TCP connection
 *
 * @v tcp		TCP connection
 * @v rc		Reason for close
 *
 * Closes the data transfer interface.  If the TCP state machine is in
 * a suitable state, the connection will be deleted.
 */
static void tcp_close ( struct tcp_connection *tcp, int rc ) {
	struct io_buffer *iobuf;
	struct io_buffer *tmp;

	/* Close data transfer interface */
	xfer_nullify ( &tcp->xfer );
	xfer_close ( &tcp->xfer, rc );
	tcp->xfer_closed = 1;

	/* If we are in CLOSED, or have otherwise not yet received a
	 * SYN (i.e. we are in LISTEN or SYN_SENT), just delete the
	 * connection.
	 */
	if ( ! ( tcp->tcp_state & TCP_STATE_RCVD ( TCP_SYN ) ) ) {

		/* Transition to CLOSED for the sake of debugging messages */
		tcp->tcp_state = TCP_CLOSED;
		tcp_dump_state ( tcp );

		/* Free any unsent I/O buffers */
		list_for_each_entry_safe ( iobuf, tmp, &tcp->queue, list ) {
			list_del ( &iobuf->list );
			free_iob ( iobuf );
		}

		/* Remove from list and drop reference */
		stop_timer ( &tcp->timer );
		list_del ( &tcp->list );
		ref_put ( &tcp->refcnt );
		DBGC ( tcp, "TCP %p connection deleted\n", tcp );
		return;
	}

	/* If we have not had our SYN acknowledged (i.e. we are in
	 * SYN_RCVD), pretend that it has been acknowledged so that we
	 * can send a FIN without breaking things.
	 */
	if ( ! ( tcp->tcp_state & TCP_STATE_ACKED ( TCP_SYN ) ) )
		tcp_rx_ack ( tcp, ( tcp->snd_seq + 1 ), 0 );

	/* If we have no data remaining to send, start sending FIN */
	if ( list_empty ( &tcp->queue ) ) {
		tcp->tcp_state |= TCP_STATE_SENT ( TCP_FIN );
		tcp_dump_state ( tcp );
	}
}

/***************************************************************************
 *
 * Transmit data path
 *
 ***************************************************************************
 */

/**
 * Calculate transmission window
 *
 * @v tcp		TCP connection
 * @ret len		Maximum length that can be sent in a single packet
 */
static size_t tcp_xmit_win ( struct tcp_connection *tcp ) {
	size_t len;

	/* Not ready if we're not in a suitable connection state */
	if ( ! TCP_CAN_SEND_DATA ( tcp->tcp_state ) )
		return 0;

	/* Length is the minimum of the receiver's window and the path MTU */
	len = tcp->snd_win;
	if ( len > TCP_PATH_MTU )
		len = TCP_PATH_MTU;

	return len;
}

/**
 * Process TCP transmit queue
 *
 * @v tcp		TCP connection
 * @v max_len		Maximum length to process
 * @v dest		I/O buffer to fill with data, or NULL
 * @v remove		Remove data from queue
 * @ret len		Length of data processed
 *
 * This processes at most @c max_len bytes from the TCP connection's
 * transmit queue.  Data will be copied into the @c dest I/O buffer
 * (if provided) and, if @c remove is true, removed from the transmit
 * queue.
 */
static size_t tcp_process_queue ( struct tcp_connection *tcp, size_t max_len,
				  struct io_buffer *dest, int remove ) {
	struct io_buffer *iobuf;
	struct io_buffer *tmp;
	size_t frag_len;
	size_t len = 0;

	list_for_each_entry_safe ( iobuf, tmp, &tcp->queue, list ) {
		frag_len = iob_len ( iobuf );
		if ( frag_len > max_len )
			frag_len = max_len;
		if ( dest ) {
			memcpy ( iob_put ( dest, frag_len ), iobuf->data,
				 frag_len );
		}
		if ( remove ) {
			iob_pull ( iobuf, frag_len );
			if ( ! iob_len ( iobuf ) ) {
				list_del ( &iobuf->list );
				free_iob ( iobuf );
			}
		}
		len += frag_len;
		max_len -= frag_len;
	}
	return len;
}

/**
 * Transmit any outstanding data
 *
 * @v tcp		TCP connection
 * @v force_send	Force sending of packet
 * 
 * Transmits any outstanding data on the connection.
 *
 * Note that even if an error is returned, the retransmission timer
 * will have been started if necessary, and so the stack will
 * eventually attempt to retransmit the failed packet.
 */
static int tcp_xmit ( struct tcp_connection *tcp, int force_send ) {
	struct io_buffer *iobuf;
	struct tcp_header *tcphdr;
	struct tcp_mss_option *mssopt;
	struct tcp_timestamp_padded_option *tsopt;
	void *payload;
	unsigned int flags;
	size_t len = 0;
	size_t seq_len;
	size_t app_win;
	size_t rcv_win;

	/* If retransmission timer is already running, do nothing */
	if ( timer_running ( &tcp->timer ) )
		return 0;

	/* Calculate both the actual (payload) and sequence space
	 * lengths that we wish to transmit.
	 */
	if ( TCP_CAN_SEND_DATA ( tcp->tcp_state ) ) {
		len = tcp_process_queue ( tcp, tcp_xmit_win ( tcp ),
					  NULL, 0 );
	}
	seq_len = len;
	flags = TCP_FLAGS_SENDING ( tcp->tcp_state );
	if ( flags & ( TCP_SYN | TCP_FIN ) ) {
		/* SYN or FIN consume one byte, and we can never send both */
		assert ( ! ( ( flags & TCP_SYN ) && ( flags & TCP_FIN ) ) );
		seq_len++;
	}
	tcp->snd_sent = seq_len;

	/* If we have nothing to transmit, stop now */
	if ( ( seq_len == 0 ) && ! force_send )
		return 0;

	/* If we are transmitting anything that requires
	 * acknowledgement (i.e. consumes sequence space), start the
	 * retransmission timer.  Do this before attempting to
	 * allocate the I/O buffer, in case allocation itself fails.
	 */
	if ( seq_len )
		start_timer ( &tcp->timer );

	/* Allocate I/O buffer */
	iobuf = alloc_iob ( len + MAX_HDR_LEN );
	if ( ! iobuf ) {
		DBGC ( tcp, "TCP %p could not allocate data buffer\n", tcp );
		return -ENOMEM;
	}
	iob_reserve ( iobuf, MAX_HDR_LEN );

	/* Fill data payload from transmit queue */
	tcp_process_queue ( tcp, len, iobuf, 0 );

	/* Estimate window size */
	rcv_win = ( ( freemem * 3 ) / 4 );
	if ( rcv_win > TCP_MAX_WINDOW_SIZE )
		rcv_win = TCP_MAX_WINDOW_SIZE;
	app_win = xfer_window ( &tcp->xfer );
	if ( rcv_win > app_win )
		rcv_win = app_win;
	rcv_win &= ~0x03; /* Keep everything dword-aligned */

	/* Fill up the TCP header */
	payload = iobuf->data;
	if ( flags & TCP_SYN ) {
		mssopt = iob_push ( iobuf, sizeof ( *mssopt ) );
		mssopt->kind = TCP_OPTION_MSS;
		mssopt->length = sizeof ( *mssopt );
		mssopt->mss = htons ( TCP_MSS );
	}
	if ( ( flags & TCP_SYN ) || tcp->timestamps ) {
		tsopt = iob_push ( iobuf, sizeof ( *tsopt ) );
		memset ( tsopt->nop, TCP_OPTION_NOP, sizeof ( tsopt->nop ) );
		tsopt->tsopt.kind = TCP_OPTION_TS;
		tsopt->tsopt.length = sizeof ( tsopt->tsopt );
		tsopt->tsopt.tsval = ntohl ( currticks() );
		tsopt->tsopt.tsecr = ntohl ( tcp->ts_recent );
	}
	tcphdr = iob_push ( iobuf, sizeof ( *tcphdr ) );
	memset ( tcphdr, 0, sizeof ( *tcphdr ) );
	tcphdr->src = tcp->local_port;
	tcphdr->dest = tcp->peer.st_port;
	tcphdr->seq = htonl ( tcp->snd_seq );
	tcphdr->ack = htonl ( tcp->rcv_ack );
	tcphdr->hlen = ( ( payload - iobuf->data ) << 2 );
	tcphdr->flags = flags;
	tcphdr->win = htons ( rcv_win );
	tcphdr->csum = tcpip_chksum ( iobuf->data, iob_len ( iobuf ) );

	/* Dump header */
	DBGC ( tcp, "TCP %p TX %d->%d %08lx..%08lx           %08lx %4zd",
	       tcp, ntohs ( tcphdr->src ), ntohs ( tcphdr->dest ),
	       ntohl ( tcphdr->seq ), ( ntohl ( tcphdr->seq ) + seq_len ),
	       ntohl ( tcphdr->ack ), len );
	tcp_dump_flags ( tcp, tcphdr->flags );
	DBGC ( tcp, "\n" );

	/* Transmit packet */
	return tcpip_tx ( iobuf, &tcp_protocol, &tcp->peer, NULL,
			  &tcphdr->csum );
}

/**
 * Retransmission timer expired
 *
 * @v timer	Retry timer
 * @v over	Failure indicator
 */
static void tcp_expired ( struct retry_timer *timer, int over ) {
	struct tcp_connection *tcp =
		container_of ( timer, struct tcp_connection, timer );
	int graceful_close = TCP_CLOSED_GRACEFULLY ( tcp->tcp_state );

	DBGC ( tcp, "TCP %p timer %s in %s\n", tcp,
	       ( over ? "expired" : "fired" ), tcp_state ( tcp->tcp_state ) );

	assert ( ( tcp->tcp_state == TCP_SYN_SENT ) ||
		 ( tcp->tcp_state == TCP_SYN_RCVD ) ||
		 ( tcp->tcp_state == TCP_ESTABLISHED ) ||
		 ( tcp->tcp_state == TCP_FIN_WAIT_1 ) ||
		 ( tcp->tcp_state == TCP_TIME_WAIT ) ||
		 ( tcp->tcp_state == TCP_CLOSE_WAIT ) ||
		 ( tcp->tcp_state == TCP_CLOSING_OR_LAST_ACK ) );

	if ( over || graceful_close ) {
		/* If we have finally timed out and given up, or if
		 * this is the result of a graceful close, terminate
		 * the connection
		 */
		tcp->tcp_state = TCP_CLOSED;
		tcp_dump_state ( tcp );
		tcp_close ( tcp, -ETIMEDOUT );
	} else {
		/* Otherwise, retransmit the packet */
		tcp_xmit ( tcp, 0 );
	}
}

/**
 * Send RST response to incoming packet
 *
 * @v in_tcphdr		TCP header of incoming packet
 * @ret rc		Return status code
 */
static int tcp_xmit_reset ( struct tcp_connection *tcp,
			    struct sockaddr_tcpip *st_dest,
			    struct tcp_header *in_tcphdr ) {
	struct io_buffer *iobuf;
	struct tcp_header *tcphdr;

	/* Allocate space for dataless TX buffer */
	iobuf = alloc_iob ( MAX_HDR_LEN );
	if ( ! iobuf ) {
		DBGC ( tcp, "TCP %p could not allocate data buffer\n", tcp );
		return -ENOMEM;
	}
	iob_reserve ( iobuf, MAX_HDR_LEN );

	/* Construct RST response */
	tcphdr = iob_push ( iobuf, sizeof ( *tcphdr ) );
	memset ( tcphdr, 0, sizeof ( *tcphdr ) );
	tcphdr->src = in_tcphdr->dest;
	tcphdr->dest = in_tcphdr->src;
	tcphdr->seq = in_tcphdr->ack;
	tcphdr->ack = in_tcphdr->seq;
	tcphdr->hlen = ( ( sizeof ( *tcphdr ) / 4 ) << 4 );
	tcphdr->flags = ( TCP_RST | TCP_ACK );
	tcphdr->win = htons ( TCP_MAX_WINDOW_SIZE );
	tcphdr->csum = tcpip_chksum ( iobuf->data, iob_len ( iobuf ) );

	/* Dump header */
	DBGC ( tcp, "TCP %p TX %d->%d %08lx..%08lx           %08lx %4d",
	       tcp, ntohs ( tcphdr->src ), ntohs ( tcphdr->dest ),
	       ntohl ( tcphdr->seq ), ( ntohl ( tcphdr->seq ) ),
	       ntohl ( tcphdr->ack ), 0 );
	tcp_dump_flags ( tcp, tcphdr->flags );
	DBGC ( tcp, "\n" );

	/* Transmit packet */
	return tcpip_tx ( iobuf, &tcp_protocol, st_dest,
			  NULL, &tcphdr->csum );
}

/***************************************************************************
 *
 * Receive data path
 *
 ***************************************************************************
 */

/**
 * Identify TCP connection by local port number
 *
 * @v local_port	Local port (in network-endian order)
 * @ret tcp		TCP connection, or NULL
 */
static struct tcp_connection * tcp_demux ( unsigned int local_port ) {
	struct tcp_connection *tcp;

	list_for_each_entry ( tcp, &tcp_conns, list ) {
		if ( tcp->local_port == local_port )
			return tcp;
	}
	return NULL;
}

/**
 * Parse TCP received options
 *
 * @v tcp		TCP connection
 * @v data		Raw options data
 * @v len		Raw options length
 * @v options		Options structure to fill in
 */
static void tcp_rx_opts ( struct tcp_connection *tcp, const void *data,
			  size_t len, struct tcp_options *options ) {
	const void *end = ( data + len );
	const struct tcp_option *option;
	unsigned int kind;

	memset ( options, 0, sizeof ( *options ) );
	while ( data < end ) {
		option = data;
		kind = option->kind;
		if ( kind == TCP_OPTION_END )
			return;
		if ( kind == TCP_OPTION_NOP ) {
			data++;
			continue;
		}
		switch ( kind ) {
		case TCP_OPTION_MSS:
			options->mssopt = data;
			break;
		case TCP_OPTION_TS:
			options->tsopt = data;
			break;
		default:
			DBGC ( tcp, "TCP %p received unknown option %d\n",
			       tcp, kind );
			break;
		}
		data += option->length;
	}
}

/**
 * Handle TCP received SYN
 *
 * @v tcp		TCP connection
 * @v seq		SEQ value (in host-endian order)
 * @v options		TCP options
 * @ret rc		Return status code
 */
static int tcp_rx_syn ( struct tcp_connection *tcp, uint32_t seq,
			struct tcp_options *options ) {

	/* Synchronise sequence numbers on first SYN */
	if ( ! ( tcp->tcp_state & TCP_STATE_RCVD ( TCP_SYN ) ) ) {
		tcp->rcv_ack = seq;
		if ( options->tsopt )
			tcp->timestamps = 1;
	}

	/* Ignore duplicate SYN */
	if ( ( tcp->rcv_ack - seq ) > 0 )
		return 0;

	/* Mark SYN as received and start sending ACKs with each packet */
	tcp->tcp_state |= ( TCP_STATE_SENT ( TCP_ACK ) |
			    TCP_STATE_RCVD ( TCP_SYN ) );

	/* Acknowledge SYN */
	tcp->rcv_ack++;

	return 0;
}

/**
 * Handle TCP received ACK
 *
 * @v tcp		TCP connection
 * @v ack		ACK value (in host-endian order)
 * @v win		WIN value (in host-endian order)
 * @ret rc		Return status code
 */
static int tcp_rx_ack ( struct tcp_connection *tcp, uint32_t ack,
			uint32_t win ) {
	size_t ack_len = ( ack - tcp->snd_seq );
	size_t len;
	unsigned int acked_flags;

	/* Ignore duplicate or out-of-range ACK */
	if ( ack_len > tcp->snd_sent ) {
		DBGC ( tcp, "TCP %p received ACK for [%08lx,%08lx), "
		       "sent only [%08lx,%08lx)\n", tcp, tcp->snd_seq,
		       ( tcp->snd_seq + ack_len ), tcp->snd_seq,
		       ( tcp->snd_seq + tcp->snd_sent ) );
		return -EINVAL;
	}

	/* Acknowledge any flags being sent */
	len = ack_len;
	acked_flags = ( TCP_FLAGS_SENDING ( tcp->tcp_state ) &
			( TCP_SYN | TCP_FIN ) );
	if ( acked_flags )
		len--;

	/* Update SEQ and sent counters, and window size */
	tcp->snd_seq = ack;
	tcp->snd_sent = 0;
	tcp->snd_win = win;

	/* Stop the retransmission timer */
	stop_timer ( &tcp->timer );

	/* Remove any acknowledged data from transmit queue */
	tcp_process_queue ( tcp, len, NULL, 1 );
		
	/* Mark SYN/FIN as acknowledged if applicable. */
	if ( acked_flags )
		tcp->tcp_state |= TCP_STATE_ACKED ( acked_flags );

	/* Start sending FIN if we've had all possible data ACKed */
	if ( list_empty ( &tcp->queue ) && tcp->xfer_closed )
		tcp->tcp_state |= TCP_STATE_SENT ( TCP_FIN );

	return 0;
}

/**
 * Handle TCP received data
 *
 * @v tcp		TCP connection
 * @v seq		SEQ value (in host-endian order)
 * @v iobuf		I/O buffer
 * @ret rc		Return status code
 *
 * This function takes ownership of the I/O buffer.
 */
static int tcp_rx_data ( struct tcp_connection *tcp, uint32_t seq,
			 struct io_buffer *iobuf ) {
	size_t already_rcvd;
	size_t len;
	int rc;

	/* Ignore duplicate or out-of-order data */
	already_rcvd = ( tcp->rcv_ack - seq );
	len = iob_len ( iobuf );
	if ( already_rcvd >= len ) {
		free_iob ( iobuf );
		return 0;
	}
	iob_pull ( iobuf, already_rcvd );
	len -= already_rcvd;

	/* Deliver data to application */
	if ( ( rc = xfer_deliver_iob ( &tcp->xfer, iobuf ) ) != 0 )
		return rc;

	/* Acknowledge new data */
	tcp->rcv_ack += len;
	return 0;
}

/**
 * Handle TCP received FIN
 *
 * @v tcp		TCP connection
 * @v seq		SEQ value (in host-endian order)
 * @ret rc		Return status code
 */
static int tcp_rx_fin ( struct tcp_connection *tcp, uint32_t seq ) {

	/* Ignore duplicate or out-of-order FIN */
	if ( ( tcp->rcv_ack - seq ) > 0 )
		return 0;

	/* Mark FIN as received and acknowledge it */
	tcp->tcp_state |= TCP_STATE_RCVD ( TCP_FIN );
	tcp->rcv_ack++;

	/* Close connection */
	tcp_close ( tcp, 0 );

	return 0;
}

/**
 * Handle TCP received RST
 *
 * @v tcp		TCP connection
 * @v seq		SEQ value (in host-endian order)
 * @ret rc		Return status code
 */
static int tcp_rx_rst ( struct tcp_connection *tcp, uint32_t seq ) {

	/* Accept RST only if it falls within the window.  If we have
	 * not yet received a SYN, then we have no window to test
	 * against, so fall back to checking that our SYN has been
	 * ACKed.
	 */
	if ( tcp->tcp_state & TCP_STATE_RCVD ( TCP_SYN ) ) {
		if ( ( tcp->rcv_ack - seq ) > 0 )
			return 0;
	} else {
		if ( ! ( tcp->tcp_state & TCP_STATE_ACKED ( TCP_SYN ) ) )
			return 0;
	}

	/* Abort connection */
	tcp->tcp_state = TCP_CLOSED;
	tcp_dump_state ( tcp );
	tcp_close ( tcp, -ECONNRESET );

	return -ECONNRESET;
}

/**
 * Process received packet
 *
 * @v iobuf		I/O buffer
 * @v st_src		Partially-filled source address
 * @v st_dest		Partially-filled destination address
 * @v pshdr_csum	Pseudo-header checksum
 * @ret rc		Return status code
  */
static int tcp_rx ( struct io_buffer *iobuf,
		    struct sockaddr_tcpip *st_src,
		    struct sockaddr_tcpip *st_dest __unused,
		    uint16_t pshdr_csum ) {
	struct tcp_header *tcphdr = iobuf->data;
	struct tcp_connection *tcp;
	struct tcp_options options;
	size_t hlen;
	uint16_t csum;
	uint32_t start_seq;
	uint32_t seq;
	uint32_t ack;
	uint32_t win;
	unsigned int flags;
	size_t len;
	int rc;

	/* Sanity check packet */
	if ( iob_len ( iobuf ) < sizeof ( *tcphdr ) ) {
		DBG ( "TCP packet too short at %zd bytes (min %zd bytes)\n",
		      iob_len ( iobuf ), sizeof ( *tcphdr ) );
		rc = -EINVAL;
		goto discard;
	}
	hlen = ( ( tcphdr->hlen & TCP_MASK_HLEN ) / 16 ) * 4;
	if ( hlen < sizeof ( *tcphdr ) ) {
		DBG ( "TCP header too short at %zd bytes (min %zd bytes)\n",
		      hlen, sizeof ( *tcphdr ) );
		rc = -EINVAL;
		goto discard;
	}
	if ( hlen > iob_len ( iobuf ) ) {
		DBG ( "TCP header too long at %zd bytes (max %zd bytes)\n",
		      hlen, iob_len ( iobuf ) );
		rc = -EINVAL;
		goto discard;
	}
	csum = tcpip_continue_chksum ( pshdr_csum, iobuf->data, iob_len ( iobuf ));
	if ( csum != 0 ) {
		DBG ( "TCP checksum incorrect (is %04x including checksum "
		      "field, should be 0000)\n", csum );
		rc = -EINVAL;
		goto discard;
	}
	
	/* Parse parameters from header and strip header */
	tcp = tcp_demux ( tcphdr->dest );
	start_seq = seq = ntohl ( tcphdr->seq );
	ack = ntohl ( tcphdr->ack );
	win = ntohs ( tcphdr->win );
	flags = tcphdr->flags;
	tcp_rx_opts ( tcp, ( ( ( void * ) tcphdr ) + sizeof ( *tcphdr ) ),
		      ( hlen - sizeof ( *tcphdr ) ), &options );
	iob_pull ( iobuf, hlen );
	len = iob_len ( iobuf );

	/* Dump header */
	DBGC ( tcp, "TCP %p RX %d<-%d           %08lx %08lx..%08lx %4zd",
	       tcp, ntohs ( tcphdr->dest ), ntohs ( tcphdr->src ),
	       ntohl ( tcphdr->ack ), ntohl ( tcphdr->seq ),
	       ( ntohl ( tcphdr->seq ) + len +
		 ( ( tcphdr->flags & ( TCP_SYN | TCP_FIN ) ) ? 1 : 0 ) ), len);
	tcp_dump_flags ( tcp, tcphdr->flags );
	DBGC ( tcp, "\n" );

	/* If no connection was found, send RST */
	if ( ! tcp ) {
		tcp_xmit_reset ( tcp, st_src, tcphdr );
		rc = -ENOTCONN;
		goto discard;
	}

	/* Handle ACK, if present */
	if ( flags & TCP_ACK ) {
		if ( ( rc = tcp_rx_ack ( tcp, ack, win ) ) != 0 ) {
			tcp_xmit_reset ( tcp, st_src, tcphdr );
			goto discard;
		}
	}

	/* Handle SYN, if present */
	if ( flags & TCP_SYN ) {
		tcp_rx_syn ( tcp, seq, &options );
		seq++;
	}

	/* Handle RST, if present */
	if ( flags & TCP_RST ) {
		if ( ( rc = tcp_rx_rst ( tcp, seq ) ) != 0 )
			goto discard;
	}

	/* Handle new data, if any */
	tcp_rx_data ( tcp, seq, iobuf );
	seq += len;

	/* Handle FIN, if present */
	if ( flags & TCP_FIN ) {
		tcp_rx_fin ( tcp, seq );
		seq++;
	}

	/* Update timestamp, if present and applicable */
	if ( ( seq == tcp->rcv_ack ) && options.tsopt )
		tcp->ts_recent = ntohl ( options.tsopt->tsval );

	/* Dump out any state change as a result of the received packet */
	tcp_dump_state ( tcp );

	/* Send out any pending data.  If peer is expecting an ACK for
	 * this packet then force sending a reply.
	 */
	tcp_xmit ( tcp, ( start_seq != seq ) );

	/* If this packet was the last we expect to receive, set up
	 * timer to expire and cause the connection to be freed.
	 */
	if ( TCP_CLOSED_GRACEFULLY ( tcp->tcp_state ) ) {
		tcp->timer.timeout = ( 2 * TCP_MSL );
		start_timer ( &tcp->timer );
	}

	return 0;

 discard:
	/* Free received packet */
	free_iob ( iobuf );
	return rc;
}

/** TCP protocol */
struct tcpip_protocol tcp_protocol __tcpip_protocol = {
	.name = "TCP",
	.rx = tcp_rx,
	.tcpip_proto = IP_TCP,
};

/***************************************************************************
 *
 * Data transfer interface
 *
 ***************************************************************************
 */

/**
 * Close interface
 *
 * @v xfer		Data transfer interface
 * @v rc		Reason for close
 */
static void tcp_xfer_close ( struct xfer_interface *xfer, int rc ) {
	struct tcp_connection *tcp =
		container_of ( xfer, struct tcp_connection, xfer );

	/* Close data transfer interface */
	tcp_close ( tcp, rc );

	/* Transmit FIN, if possible */
	tcp_xmit ( tcp, 0 );
}

/**
 * Check flow control window
 *
 * @v xfer		Data transfer interface
 * @ret len		Length of window
 */
static size_t tcp_xfer_window ( struct xfer_interface *xfer ) {
	struct tcp_connection *tcp =
		container_of ( xfer, struct tcp_connection, xfer );

	/* Not ready if data queue is non-empty.  This imposes a limit
	 * of only one unACKed packet in the TX queue at any time; we
	 * do this to conserve memory usage.
	 */
	if ( ! list_empty ( &tcp->queue ) )
		return 0;

	/* Return TCP window length */
	return tcp_xmit_win ( tcp );
}

/**
 * Deliver datagram as I/O buffer
 *
 * @v xfer		Data transfer interface
 * @v iobuf		Datagram I/O buffer
 * @v meta		Data transfer metadata, or NULL
 * @ret rc		Return status code
 */
static int tcp_xfer_deliver_iob ( struct xfer_interface *xfer,
				  struct io_buffer *iobuf,
				  struct xfer_metadata *meta __unused ) {
	struct tcp_connection *tcp =
		container_of ( xfer, struct tcp_connection, xfer );

	/* Enqueue packet */
	list_add_tail ( &iobuf->list, &tcp->queue );

	/* Transmit data, if possible */
	tcp_xmit ( tcp, 0 );

	return 0;
}

/** TCP data transfer interface operations */
static struct xfer_interface_operations tcp_xfer_operations = {
	.close		= tcp_xfer_close,
	.vredirect	= ignore_xfer_vredirect,
	.window		= tcp_xfer_window,
	.alloc_iob	= default_xfer_alloc_iob,
	.deliver_iob	= tcp_xfer_deliver_iob,
	.deliver_raw	= xfer_deliver_as_iob,
};

/***************************************************************************
 *
 * Openers
 *
 ***************************************************************************
 */

/** TCP socket opener */
struct socket_opener tcp_socket_opener __socket_opener = {
	.semantics	= SOCK_STREAM,
	.family		= AF_INET,
	.open		= tcp_open,
};

char TCP_SOCK_STREAM[1];

/**
 * Open TCP URI
 *
 * @v xfer		Data transfer interface
 * @v uri		URI
 * @ret rc		Return status code
 */
static int tcp_open_uri ( struct xfer_interface *xfer, struct uri *uri ) {
	struct sockaddr_tcpip peer;

	/* Sanity check */
	if ( ! uri->host )
		return -EINVAL;

	memset ( &peer, 0, sizeof ( peer ) );
	peer.st_port = htons ( uri_port ( uri, 0 ) );
	return xfer_open_named_socket ( xfer, SOCK_STREAM,
					( struct sockaddr * ) &peer,
					uri->host, NULL );
}

/** TCP URI opener */
struct uri_opener tcp_uri_opener __uri_opener = {
	.scheme		= "tcp",
	.open		= tcp_open_uri,
};