summaryrefslogtreecommitdiffstats
path: root/sys/netinet/siftr.c
blob: c838d7ae1e1f205cca0713564a6bab5db9c0862e (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
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
/*-
 * Copyright (c) 2007-2009
 * 	Swinburne University of Technology, Melbourne, Australia.
 * Copyright (c) 2009-2010, The FreeBSD Foundation
 * All rights reserved.
 *
 * Portions of this software were developed at the Centre for Advanced
 * Internet Architectures, Swinburne University of Technology, Melbourne,
 * Australia by Lawrence Stewart under sponsorship from the FreeBSD Foundation.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 *
 * THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 */

/******************************************************
 * Statistical Information For TCP Research (SIFTR)
 *
 * A FreeBSD kernel module that adds very basic intrumentation to the
 * TCP stack, allowing internal stats to be recorded to a log file
 * for experimental, debugging and performance analysis purposes.
 *
 * SIFTR was first released in 2007 by James Healy and Lawrence Stewart whilst
 * working on the NewTCP research project at Swinburne University of
 * Technology's Centre for Advanced Internet Architectures, Melbourne,
 * Australia, which was made possible in part by a grant from the Cisco
 * University Research Program Fund at Community Foundation Silicon Valley.
 * More details are available at:
 *   http://caia.swin.edu.au/urp/newtcp/
 *
 * Work on SIFTR v1.2.x was sponsored by the FreeBSD Foundation as part of
 * the "Enhancing the FreeBSD TCP Implementation" project 2008-2009.
 * More details are available at:
 *   http://www.freebsdfoundation.org/
 *   http://caia.swin.edu.au/freebsd/etcp09/
 *
 * Lawrence Stewart is the current maintainer, and all contact regarding
 * SIFTR should be directed to him via email: lastewart@swin.edu.au
 *
 * Initial release date: June 2007
 * Most recent update: September 2010
 ******************************************************/

#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");

#include <sys/param.h>
#include <sys/alq.h>
#include <sys/errno.h>
#include <sys/eventhandler.h>
#include <sys/hash.h>
#include <sys/kernel.h>
#include <sys/kthread.h>
#include <sys/lock.h>
#include <sys/mbuf.h>
#include <sys/module.h>
#include <sys/mutex.h>
#include <sys/pcpu.h>
#include <sys/proc.h>
#include <sys/sbuf.h>
#include <sys/smp.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/sysctl.h>
#include <sys/unistd.h>

#include <net/if.h>
#include <net/if_var.h>
#include <net/pfil.h>

#include <netinet/in.h>
#include <netinet/in_pcb.h>
#include <netinet/in_systm.h>
#include <netinet/in_var.h>
#include <netinet/ip.h>
#include <netinet/tcp_var.h>

#ifdef SIFTR_IPV6
#include <netinet/ip6.h>
#include <netinet6/in6_pcb.h>
#endif /* SIFTR_IPV6 */

#include <machine/in_cksum.h>

/*
 * Three digit version number refers to X.Y.Z where:
 * X is the major version number
 * Y is bumped to mark backwards incompatible changes
 * Z is bumped to mark backwards compatible changes
 */
#define V_MAJOR		1
#define V_BACKBREAK	2
#define V_BACKCOMPAT	4
#define MODVERSION	__CONCAT(V_MAJOR, __CONCAT(V_BACKBREAK, V_BACKCOMPAT))
#define MODVERSION_STR	__XSTRING(V_MAJOR) "." __XSTRING(V_BACKBREAK) "." \
    __XSTRING(V_BACKCOMPAT)

#define HOOK 0
#define UNHOOK 1
#define SIFTR_EXPECTED_MAX_TCP_FLOWS 65536
#define SYS_NAME "FreeBSD"
#define PACKET_TAG_SIFTR 100
#define PACKET_COOKIE_SIFTR 21749576
#define SIFTR_LOG_FILE_MODE 0644
#define SIFTR_DISABLE 0
#define SIFTR_ENABLE 1

/*
 * Hard upper limit on the length of log messages. Bump this up if you add new
 * data fields such that the line length could exceed the below value.
 */
#define MAX_LOG_MSG_LEN 200
/* XXX: Make this a sysctl tunable. */
#define SIFTR_ALQ_BUFLEN (1000*MAX_LOG_MSG_LEN)

/*
 * 1 byte for IP version
 * IPv4: src/dst IP (4+4) + src/dst port (2+2) = 12 bytes
 * IPv6: src/dst IP (16+16) + src/dst port (2+2) = 36 bytes
 */
#ifdef SIFTR_IPV6
#define FLOW_KEY_LEN 37
#else
#define FLOW_KEY_LEN 13
#endif

#ifdef SIFTR_IPV6
#define SIFTR_IPMODE 6
#else
#define SIFTR_IPMODE 4
#endif

/* useful macros */
#define CAST_PTR_INT(X) (*((int*)(X)))

#define UPPER_SHORT(X)	(((X) & 0xFFFF0000) >> 16)
#define LOWER_SHORT(X)	((X) & 0x0000FFFF)

#define FIRST_OCTET(X)	(((X) & 0xFF000000) >> 24)
#define SECOND_OCTET(X)	(((X) & 0x00FF0000) >> 16)
#define THIRD_OCTET(X)	(((X) & 0x0000FF00) >> 8)
#define FOURTH_OCTET(X)	((X) & 0x000000FF)

static MALLOC_DEFINE(M_SIFTR, "siftr", "dynamic memory used by SIFTR");
static MALLOC_DEFINE(M_SIFTR_PKTNODE, "siftr_pktnode",
    "SIFTR pkt_node struct");
static MALLOC_DEFINE(M_SIFTR_HASHNODE, "siftr_hashnode",
    "SIFTR flow_hash_node struct");

/* Used as links in the pkt manager queue. */
struct pkt_node {
	/* Timestamp of pkt as noted in the pfil hook. */
	struct timeval		tval;
	/* Direction pkt is travelling; either PFIL_IN or PFIL_OUT. */
	uint8_t			direction;
	/* IP version pkt_node relates to; either INP_IPV4 or INP_IPV6. */
	uint8_t			ipver;
	/* Hash of the pkt which triggered the log message. */
	uint32_t		hash;
	/* Local/foreign IP address. */
#ifdef SIFTR_IPV6
	uint32_t		ip_laddr[4];
	uint32_t		ip_faddr[4];
#else
	uint8_t			ip_laddr[4];
	uint8_t			ip_faddr[4];
#endif
	/* Local TCP port. */
	uint16_t		tcp_localport;
	/* Foreign TCP port. */
	uint16_t		tcp_foreignport;
	/* Congestion Window (bytes). */
	u_long			snd_cwnd;
	/* Sending Window (bytes). */
	u_long			snd_wnd;
	/* Receive Window (bytes). */
	u_long			rcv_wnd;
	/* Unused (was: Bandwidth Controlled Window (bytes)). */
	u_long			snd_bwnd;
	/* Slow Start Threshold (bytes). */
	u_long			snd_ssthresh;
	/* Current state of the TCP FSM. */
	int			conn_state;
	/* Max Segment Size (bytes). */
	u_int			max_seg_size;
	/*
	 * Smoothed RTT stored as found in the TCP control block
	 * in units of (TCP_RTT_SCALE*hz).
	 */
	int			smoothed_rtt;
	/* Is SACK enabled? */
	u_char			sack_enabled;
	/* Window scaling for snd window. */
	u_char			snd_scale;
	/* Window scaling for recv window. */
	u_char			rcv_scale;
	/* TCP control block flags. */
	u_int			flags;
	/* Retransmit timeout length. */
	int			rxt_length;
	/* Size of the TCP send buffer in bytes. */
	u_int			snd_buf_hiwater;
	/* Current num bytes in the send socket buffer. */
	u_int			snd_buf_cc;
	/* Size of the TCP receive buffer in bytes. */
	u_int			rcv_buf_hiwater;
	/* Current num bytes in the receive socket buffer. */
	u_int			rcv_buf_cc;
	/* Number of bytes inflight that we are waiting on ACKs for. */
	u_int			sent_inflight_bytes;
	/* Number of segments currently in the reassembly queue. */
	int			t_segqlen;
	/* Link to next pkt_node in the list. */
	STAILQ_ENTRY(pkt_node)	nodes;
};

struct flow_hash_node
{
	uint16_t counter;
	uint8_t key[FLOW_KEY_LEN];
	LIST_ENTRY(flow_hash_node) nodes;
};

struct siftr_stats
{
	/* # TCP pkts seen by the SIFTR PFIL hooks, including any skipped. */
	uint64_t n_in;
	uint64_t n_out;
	/* # pkts skipped due to failed malloc calls. */
	uint32_t nskip_in_malloc;
	uint32_t nskip_out_malloc;
	/* # pkts skipped due to failed mtx acquisition. */
	uint32_t nskip_in_mtx;
	uint32_t nskip_out_mtx;
	/* # pkts skipped due to failed inpcb lookups. */
	uint32_t nskip_in_inpcb;
	uint32_t nskip_out_inpcb;
	/* # pkts skipped due to failed tcpcb lookups. */
	uint32_t nskip_in_tcpcb;
	uint32_t nskip_out_tcpcb;
	/* # pkts skipped due to stack reinjection. */
	uint32_t nskip_in_dejavu;
	uint32_t nskip_out_dejavu;
};

static DPCPU_DEFINE(struct siftr_stats, ss);

static volatile unsigned int siftr_exit_pkt_manager_thread = 0;
static unsigned int siftr_enabled = 0;
static unsigned int siftr_pkts_per_log = 1;
static unsigned int siftr_generate_hashes = 0;
/* static unsigned int siftr_binary_log = 0; */
static char siftr_logfile[PATH_MAX] = "/var/log/siftr.log";
static u_long siftr_hashmask;
STAILQ_HEAD(pkthead, pkt_node) pkt_queue = STAILQ_HEAD_INITIALIZER(pkt_queue);
LIST_HEAD(listhead, flow_hash_node) *counter_hash;
static int wait_for_pkt;
static struct alq *siftr_alq = NULL;
static struct mtx siftr_pkt_queue_mtx;
static struct mtx siftr_pkt_mgr_mtx;
static struct thread *siftr_pkt_manager_thr = NULL;
/*
 * pfil.h defines PFIL_IN as 1 and PFIL_OUT as 2,
 * which we use as an index into this array.
 */
static char direction[3] = {'\0', 'i','o'};

/* Required function prototypes. */
static int siftr_sysctl_enabled_handler(SYSCTL_HANDLER_ARGS);
static int siftr_sysctl_logfile_name_handler(SYSCTL_HANDLER_ARGS);


/* Declare the net.inet.siftr sysctl tree and populate it. */

SYSCTL_DECL(_net_inet_siftr);

SYSCTL_NODE(_net_inet, OID_AUTO, siftr, CTLFLAG_RW, NULL,
    "siftr related settings");

SYSCTL_PROC(_net_inet_siftr, OID_AUTO, enabled, CTLTYPE_UINT|CTLFLAG_RW,
    &siftr_enabled, 0, &siftr_sysctl_enabled_handler, "IU",
    "switch siftr module operations on/off");

SYSCTL_PROC(_net_inet_siftr, OID_AUTO, logfile, CTLTYPE_STRING|CTLFLAG_RW,
    &siftr_logfile, sizeof(siftr_logfile), &siftr_sysctl_logfile_name_handler,
    "A", "file to save siftr log messages to");

SYSCTL_UINT(_net_inet_siftr, OID_AUTO, ppl, CTLFLAG_RW,
    &siftr_pkts_per_log, 1,
    "number of packets between generating a log message");

SYSCTL_UINT(_net_inet_siftr, OID_AUTO, genhashes, CTLFLAG_RW,
    &siftr_generate_hashes, 0,
    "enable packet hash generation");

/* XXX: TODO
SYSCTL_UINT(_net_inet_siftr, OID_AUTO, binary, CTLFLAG_RW,
    &siftr_binary_log, 0,
    "write log files in binary instead of ascii");
*/


/* Begin functions. */

static void
siftr_process_pkt(struct pkt_node * pkt_node)
{
	struct flow_hash_node *hash_node;
	struct listhead *counter_list;
	struct siftr_stats *ss;
	struct ale *log_buf;
	uint8_t key[FLOW_KEY_LEN];
	uint8_t found_match, key_offset;

	hash_node = NULL;
	ss = DPCPU_PTR(ss);
	found_match = 0;
	key_offset = 1;

	/*
	 * Create the key that will be used to create a hash index
	 * into our hash table. Our key consists of:
	 * ipversion, localip, localport, foreignip, foreignport
	 */
	key[0] = pkt_node->ipver;
	memcpy(key + key_offset, &pkt_node->ip_laddr,
	    sizeof(pkt_node->ip_laddr));
	key_offset += sizeof(pkt_node->ip_laddr);
	memcpy(key + key_offset, &pkt_node->tcp_localport,
	    sizeof(pkt_node->tcp_localport));
	key_offset += sizeof(pkt_node->tcp_localport);
	memcpy(key + key_offset, &pkt_node->ip_faddr,
	    sizeof(pkt_node->ip_faddr));
	key_offset += sizeof(pkt_node->ip_faddr);
	memcpy(key + key_offset, &pkt_node->tcp_foreignport,
	    sizeof(pkt_node->tcp_foreignport));

	counter_list = counter_hash +
	    (hash32_buf(key, sizeof(key), 0) & siftr_hashmask);

	/*
	 * If the list is not empty i.e. the hash index has
	 * been used by another flow previously.
	 */
	if (LIST_FIRST(counter_list) != NULL) {
		/*
		 * Loop through the hash nodes in the list.
		 * There should normally only be 1 hash node in the list,
		 * except if there have been collisions at the hash index
		 * computed by hash32_buf().
		 */
		LIST_FOREACH(hash_node, counter_list, nodes) {
			/*
			 * Check if the key for the pkt we are currently
			 * processing is the same as the key stored in the
			 * hash node we are currently processing.
			 * If they are the same, then we've found the
			 * hash node that stores the counter for the flow
			 * the pkt belongs to.
			 */
			if (memcmp(hash_node->key, key, sizeof(key)) == 0) {
				found_match = 1;
				break;
			}
		}
	}

	/* If this flow hash hasn't been seen before or we have a collision. */
	if (hash_node == NULL || !found_match) {
		/* Create a new hash node to store the flow's counter. */
		hash_node = malloc(sizeof(struct flow_hash_node),
		    M_SIFTR_HASHNODE, M_WAITOK);

		if (hash_node != NULL) {
			/* Initialise our new hash node list entry. */
			hash_node->counter = 0;
			memcpy(hash_node->key, key, sizeof(key));
			LIST_INSERT_HEAD(counter_list, hash_node, nodes);
		} else {
			/* Malloc failed. */
			if (pkt_node->direction == PFIL_IN)
				ss->nskip_in_malloc++;
			else
				ss->nskip_out_malloc++;

			return;
		}
	} else if (siftr_pkts_per_log > 1) {
		/*
		 * Taking the remainder of the counter divided
		 * by the current value of siftr_pkts_per_log
		 * and storing that in counter provides a neat
		 * way to modulate the frequency of log
		 * messages being written to the log file.
		 */
		hash_node->counter = (hash_node->counter + 1) %
		    siftr_pkts_per_log;

		/*
		 * If we have not seen enough packets since the last time
		 * we wrote a log message for this connection, return.
		 */
		if (hash_node->counter > 0)
			return;
	}

	log_buf = alq_getn(siftr_alq, MAX_LOG_MSG_LEN, ALQ_WAITOK);

	if (log_buf == NULL)
		return; /* Should only happen if the ALQ is shutting down. */

#ifdef SIFTR_IPV6
	pkt_node->ip_laddr[3] = ntohl(pkt_node->ip_laddr[3]);
	pkt_node->ip_faddr[3] = ntohl(pkt_node->ip_faddr[3]);

	if (pkt_node->ipver == INP_IPV6) { /* IPv6 packet */
		pkt_node->ip_laddr[0] = ntohl(pkt_node->ip_laddr[0]);
		pkt_node->ip_laddr[1] = ntohl(pkt_node->ip_laddr[1]);
		pkt_node->ip_laddr[2] = ntohl(pkt_node->ip_laddr[2]);
		pkt_node->ip_faddr[0] = ntohl(pkt_node->ip_faddr[0]);
		pkt_node->ip_faddr[1] = ntohl(pkt_node->ip_faddr[1]);
		pkt_node->ip_faddr[2] = ntohl(pkt_node->ip_faddr[2]);

		/* Construct an IPv6 log message. */
		log_buf->ae_bytesused = snprintf(log_buf->ae_data,
		    MAX_LOG_MSG_LEN,
		    "%c,0x%08x,%zd.%06ld,%x:%x:%x:%x:%x:%x:%x:%x,%u,%x:%x:%x:"
		    "%x:%x:%x:%x:%x,%u,%ld,%ld,%ld,%ld,%ld,%u,%u,%u,%u,%u,%u,"
		    "%u,%d,%u,%u,%u,%u,%u,%u\n",
		    direction[pkt_node->direction],
		    pkt_node->hash,
		    pkt_node->tval.tv_sec,
		    pkt_node->tval.tv_usec,
		    UPPER_SHORT(pkt_node->ip_laddr[0]),
		    LOWER_SHORT(pkt_node->ip_laddr[0]),
		    UPPER_SHORT(pkt_node->ip_laddr[1]),
		    LOWER_SHORT(pkt_node->ip_laddr[1]),
		    UPPER_SHORT(pkt_node->ip_laddr[2]),
		    LOWER_SHORT(pkt_node->ip_laddr[2]),
		    UPPER_SHORT(pkt_node->ip_laddr[3]),
		    LOWER_SHORT(pkt_node->ip_laddr[3]),
		    ntohs(pkt_node->tcp_localport),
		    UPPER_SHORT(pkt_node->ip_faddr[0]),
		    LOWER_SHORT(pkt_node->ip_faddr[0]),
		    UPPER_SHORT(pkt_node->ip_faddr[1]),
		    LOWER_SHORT(pkt_node->ip_faddr[1]),
		    UPPER_SHORT(pkt_node->ip_faddr[2]),
		    LOWER_SHORT(pkt_node->ip_faddr[2]),
		    UPPER_SHORT(pkt_node->ip_faddr[3]),
		    LOWER_SHORT(pkt_node->ip_faddr[3]),
		    ntohs(pkt_node->tcp_foreignport),
		    pkt_node->snd_ssthresh,
		    pkt_node->snd_cwnd,
		    pkt_node->snd_bwnd,
		    pkt_node->snd_wnd,
		    pkt_node->rcv_wnd,
		    pkt_node->snd_scale,
		    pkt_node->rcv_scale,
		    pkt_node->conn_state,
		    pkt_node->max_seg_size,
		    pkt_node->smoothed_rtt,
		    pkt_node->sack_enabled,
		    pkt_node->flags,
		    pkt_node->rxt_length,
		    pkt_node->snd_buf_hiwater,
		    pkt_node->snd_buf_cc,
		    pkt_node->rcv_buf_hiwater,
		    pkt_node->rcv_buf_cc,
		    pkt_node->sent_inflight_bytes,
		    pkt_node->t_segqlen);
	} else { /* IPv4 packet */
		pkt_node->ip_laddr[0] = FIRST_OCTET(pkt_node->ip_laddr[3]);
		pkt_node->ip_laddr[1] = SECOND_OCTET(pkt_node->ip_laddr[3]);
		pkt_node->ip_laddr[2] = THIRD_OCTET(pkt_node->ip_laddr[3]);
		pkt_node->ip_laddr[3] = FOURTH_OCTET(pkt_node->ip_laddr[3]);
		pkt_node->ip_faddr[0] = FIRST_OCTET(pkt_node->ip_faddr[3]);
		pkt_node->ip_faddr[1] = SECOND_OCTET(pkt_node->ip_faddr[3]);
		pkt_node->ip_faddr[2] = THIRD_OCTET(pkt_node->ip_faddr[3]);
		pkt_node->ip_faddr[3] = FOURTH_OCTET(pkt_node->ip_faddr[3]);
#endif /* SIFTR_IPV6 */

		/* Construct an IPv4 log message. */
		log_buf->ae_bytesused = snprintf(log_buf->ae_data,
		    MAX_LOG_MSG_LEN,
		    "%c,0x%08x,%jd.%06ld,%u.%u.%u.%u,%u,%u.%u.%u.%u,%u,%ld,%ld,"
		    "%ld,%ld,%ld,%u,%u,%u,%u,%u,%u,%u,%d,%u,%u,%u,%u,%u,%u\n",
		    direction[pkt_node->direction],
		    pkt_node->hash,
		    (intmax_t)pkt_node->tval.tv_sec,
		    pkt_node->tval.tv_usec,
		    pkt_node->ip_laddr[0],
		    pkt_node->ip_laddr[1],
		    pkt_node->ip_laddr[2],
		    pkt_node->ip_laddr[3],
		    ntohs(pkt_node->tcp_localport),
		    pkt_node->ip_faddr[0],
		    pkt_node->ip_faddr[1],
		    pkt_node->ip_faddr[2],
		    pkt_node->ip_faddr[3],
		    ntohs(pkt_node->tcp_foreignport),
		    pkt_node->snd_ssthresh,
		    pkt_node->snd_cwnd,
		    pkt_node->snd_bwnd,
		    pkt_node->snd_wnd,
		    pkt_node->rcv_wnd,
		    pkt_node->snd_scale,
		    pkt_node->rcv_scale,
		    pkt_node->conn_state,
		    pkt_node->max_seg_size,
		    pkt_node->smoothed_rtt,
		    pkt_node->sack_enabled,
		    pkt_node->flags,
		    pkt_node->rxt_length,
		    pkt_node->snd_buf_hiwater,
		    pkt_node->snd_buf_cc,
		    pkt_node->rcv_buf_hiwater,
		    pkt_node->rcv_buf_cc,
		    pkt_node->sent_inflight_bytes,
		    pkt_node->t_segqlen);
#ifdef SIFTR_IPV6
	}
#endif

	alq_post_flags(siftr_alq, log_buf, 0);
}


static void
siftr_pkt_manager_thread(void *arg)
{
	STAILQ_HEAD(pkthead, pkt_node) tmp_pkt_queue =
	    STAILQ_HEAD_INITIALIZER(tmp_pkt_queue);
	struct pkt_node *pkt_node, *pkt_node_temp;
	uint8_t draining;

	draining = 2;

	mtx_lock(&siftr_pkt_mgr_mtx);

	/* draining == 0 when queue has been flushed and it's safe to exit. */
	while (draining) {
		/*
		 * Sleep until we are signalled to wake because thread has
		 * been told to exit or until 1 tick has passed.
		 */
		mtx_sleep(&wait_for_pkt, &siftr_pkt_mgr_mtx, PWAIT, "pktwait",
		    1);

		/* Gain exclusive access to the pkt_node queue. */
		mtx_lock(&siftr_pkt_queue_mtx);

		/*
		 * Move pkt_queue to tmp_pkt_queue, which leaves
		 * pkt_queue empty and ready to receive more pkt_nodes.
		 */
		STAILQ_CONCAT(&tmp_pkt_queue, &pkt_queue);

		/*
		 * We've finished making changes to the list. Unlock it
		 * so the pfil hooks can continue queuing pkt_nodes.
		 */
		mtx_unlock(&siftr_pkt_queue_mtx);

		/*
		 * We can't hold a mutex whilst calling siftr_process_pkt
		 * because ALQ might sleep waiting for buffer space.
		 */
		mtx_unlock(&siftr_pkt_mgr_mtx);

		/* Flush all pkt_nodes to the log file. */
		STAILQ_FOREACH_SAFE(pkt_node, &tmp_pkt_queue, nodes,
		    pkt_node_temp) {
			siftr_process_pkt(pkt_node);
			STAILQ_REMOVE_HEAD(&tmp_pkt_queue, nodes);
			free(pkt_node, M_SIFTR_PKTNODE);
		}

		KASSERT(STAILQ_EMPTY(&tmp_pkt_queue),
		    ("SIFTR tmp_pkt_queue not empty after flush"));

		mtx_lock(&siftr_pkt_mgr_mtx);

		/*
		 * If siftr_exit_pkt_manager_thread gets set during the window
		 * where we are draining the tmp_pkt_queue above, there might
		 * still be pkts in pkt_queue that need to be drained.
		 * Allow one further iteration to occur after
		 * siftr_exit_pkt_manager_thread has been set to ensure
		 * pkt_queue is completely empty before we kill the thread.
		 *
		 * siftr_exit_pkt_manager_thread is set only after the pfil
		 * hooks have been removed, so only 1 extra iteration
		 * is needed to drain the queue.
		 */
		if (siftr_exit_pkt_manager_thread)
			draining--;
	}

	mtx_unlock(&siftr_pkt_mgr_mtx);

	/* Calls wakeup on this thread's struct thread ptr. */
	kthread_exit();
}


static uint32_t
hash_pkt(struct mbuf *m, uint32_t offset)
{
	uint32_t hash;

	hash = 0;

	while (m != NULL && offset > m->m_len) {
		/*
		 * The IP packet payload does not start in this mbuf, so
		 * need to figure out which mbuf it starts in and what offset
		 * into the mbuf's data region the payload starts at.
		 */
		offset -= m->m_len;
		m = m->m_next;
	}

	while (m != NULL) {
		/* Ensure there is data in the mbuf */
		if ((m->m_len - offset) > 0)
			hash = hash32_buf(m->m_data + offset,
			    m->m_len - offset, hash);

		m = m->m_next;
		offset = 0;
        }

	return (hash);
}


/*
 * Check if a given mbuf has the SIFTR mbuf tag. If it does, log the fact that
 * it's a reinjected packet and return. If it doesn't, tag the mbuf and return.
 * Return value >0 means the caller should skip processing this mbuf.
 */
static inline int
siftr_chkreinject(struct mbuf *m, int dir, struct siftr_stats *ss)
{
	if (m_tag_locate(m, PACKET_COOKIE_SIFTR, PACKET_TAG_SIFTR, NULL)
	    != NULL) {
		if (dir == PFIL_IN)
			ss->nskip_in_dejavu++;
		else
			ss->nskip_out_dejavu++;

		return (1);
	} else {
		struct m_tag *tag = m_tag_alloc(PACKET_COOKIE_SIFTR,
		    PACKET_TAG_SIFTR, 0, M_NOWAIT);
		if (tag == NULL) {
			if (dir == PFIL_IN)
				ss->nskip_in_malloc++;
			else
				ss->nskip_out_malloc++;

			return (1);
		}

		m_tag_prepend(m, tag);
	}

	return (0);
}


/*
 * Look up an inpcb for a packet. Return the inpcb pointer if found, or NULL
 * otherwise.
 */
static inline struct inpcb *
siftr_findinpcb(int ipver, struct ip *ip, struct mbuf *m, uint16_t sport,
    uint16_t dport, int dir, struct siftr_stats *ss)
{
	struct inpcb *inp;

	/* We need the tcbinfo lock. */
	INP_INFO_UNLOCK_ASSERT(&V_tcbinfo);

	if (dir == PFIL_IN)
		inp = (ipver == INP_IPV4 ?
		    in_pcblookup(&V_tcbinfo, ip->ip_src, sport, ip->ip_dst,
		    dport, INPLOOKUP_RLOCKPCB, m->m_pkthdr.rcvif)
		    :
#ifdef SIFTR_IPV6
		    in6_pcblookup(&V_tcbinfo,
		    &((struct ip6_hdr *)ip)->ip6_src, sport,
		    &((struct ip6_hdr *)ip)->ip6_dst, dport, INPLOOKUP_RLOCKPCB,
		    m->m_pkthdr.rcvif)
#else
		    NULL
#endif
		    );

	else
		inp = (ipver == INP_IPV4 ?
		    in_pcblookup(&V_tcbinfo, ip->ip_dst, dport, ip->ip_src,
		    sport, INPLOOKUP_RLOCKPCB, m->m_pkthdr.rcvif)
		    :
#ifdef SIFTR_IPV6
		    in6_pcblookup(&V_tcbinfo,
		    &((struct ip6_hdr *)ip)->ip6_dst, dport,
		    &((struct ip6_hdr *)ip)->ip6_src, sport, INPLOOKUP_RLOCKPCB,
		    m->m_pkthdr.rcvif)
#else
		    NULL
#endif
		    );

	/* If we can't find the inpcb, bail. */
	if (inp == NULL) {
		if (dir == PFIL_IN)
			ss->nskip_in_inpcb++;
		else
			ss->nskip_out_inpcb++;
	}

	return (inp);
}


static inline void
siftr_siftdata(struct pkt_node *pn, struct inpcb *inp, struct tcpcb *tp,
    int ipver, int dir, int inp_locally_locked)
{
#ifdef SIFTR_IPV6
	if (ipver == INP_IPV4) {
		pn->ip_laddr[3] = inp->inp_laddr.s_addr;
		pn->ip_faddr[3] = inp->inp_faddr.s_addr;
#else
		*((uint32_t *)pn->ip_laddr) = inp->inp_laddr.s_addr;
		*((uint32_t *)pn->ip_faddr) = inp->inp_faddr.s_addr;
#endif
#ifdef SIFTR_IPV6
	} else {
		pn->ip_laddr[0] = inp->in6p_laddr.s6_addr32[0];
		pn->ip_laddr[1] = inp->in6p_laddr.s6_addr32[1];
		pn->ip_laddr[2] = inp->in6p_laddr.s6_addr32[2];
		pn->ip_laddr[3] = inp->in6p_laddr.s6_addr32[3];
		pn->ip_faddr[0] = inp->in6p_faddr.s6_addr32[0];
		pn->ip_faddr[1] = inp->in6p_faddr.s6_addr32[1];
		pn->ip_faddr[2] = inp->in6p_faddr.s6_addr32[2];
		pn->ip_faddr[3] = inp->in6p_faddr.s6_addr32[3];
	}
#endif
	pn->tcp_localport = inp->inp_lport;
	pn->tcp_foreignport = inp->inp_fport;
	pn->snd_cwnd = tp->snd_cwnd;
	pn->snd_wnd = tp->snd_wnd;
	pn->rcv_wnd = tp->rcv_wnd;
	pn->snd_bwnd = 0;		/* Unused, kept for compat. */
	pn->snd_ssthresh = tp->snd_ssthresh;
	pn->snd_scale = tp->snd_scale;
	pn->rcv_scale = tp->rcv_scale;
	pn->conn_state = tp->t_state;
	pn->max_seg_size = tp->t_maxseg;
	pn->smoothed_rtt = tp->t_srtt;
	pn->sack_enabled = (tp->t_flags & TF_SACK_PERMIT) != 0;
	pn->flags = tp->t_flags;
	pn->rxt_length = tp->t_rxtcur;
	pn->snd_buf_hiwater = inp->inp_socket->so_snd.sb_hiwat;
	pn->snd_buf_cc = inp->inp_socket->so_snd.sb_cc;
	pn->rcv_buf_hiwater = inp->inp_socket->so_rcv.sb_hiwat;
	pn->rcv_buf_cc = inp->inp_socket->so_rcv.sb_cc;
	pn->sent_inflight_bytes = tp->snd_max - tp->snd_una;
	pn->t_segqlen = tp->t_segqlen;

	/* We've finished accessing the tcb so release the lock. */
	if (inp_locally_locked)
		INP_RUNLOCK(inp);

	pn->ipver = ipver;
	pn->direction = dir;

	/*
	 * Significantly more accurate than using getmicrotime(), but slower!
	 * Gives true microsecond resolution at the expense of a hit to
	 * maximum pps throughput processing when SIFTR is loaded and enabled.
	 */
	microtime(&pn->tval);
}


/*
 * pfil hook that is called for each IPv4 packet making its way through the
 * stack in either direction.
 * The pfil subsystem holds a non-sleepable mutex somewhere when
 * calling our hook function, so we can't sleep at all.
 * It's very important to use the M_NOWAIT flag with all function calls
 * that support it so that they won't sleep, otherwise you get a panic.
 */
static int
siftr_chkpkt(void *arg, struct mbuf **m, struct ifnet *ifp, int dir,
    struct inpcb *inp)
{
	struct pkt_node *pn;
	struct ip *ip;
	struct tcphdr *th;
	struct tcpcb *tp;
	struct siftr_stats *ss;
	unsigned int ip_hl;
	int inp_locally_locked;

	inp_locally_locked = 0;
	ss = DPCPU_PTR(ss);

	/*
	 * m_pullup is not required here because ip_{input|output}
	 * already do the heavy lifting for us.
	 */

	ip = mtod(*m, struct ip *);

	/* Only continue processing if the packet is TCP. */
	if (ip->ip_p != IPPROTO_TCP)
		goto ret;

	/*
	 * If a kernel subsystem reinjects packets into the stack, our pfil
	 * hook will be called multiple times for the same packet.
	 * Make sure we only process unique packets.
	 */
	if (siftr_chkreinject(*m, dir, ss))
		goto ret;

	if (dir == PFIL_IN)
		ss->n_in++;
	else
		ss->n_out++;

	/*
	 * Create a tcphdr struct starting at the correct offset
	 * in the IP packet. ip->ip_hl gives the ip header length
	 * in 4-byte words, so multiply it to get the size in bytes.
	 */
	ip_hl = (ip->ip_hl << 2);
	th = (struct tcphdr *)((caddr_t)ip + ip_hl);

	/*
	 * If the pfil hooks don't provide a pointer to the
	 * inpcb, we need to find it ourselves and lock it.
	 */
	if (!inp) {
		/* Find the corresponding inpcb for this pkt. */
		inp = siftr_findinpcb(INP_IPV4, ip, *m, th->th_sport,
		    th->th_dport, dir, ss);

		if (inp == NULL)
			goto ret;
		else
			inp_locally_locked = 1;
	}

	INP_LOCK_ASSERT(inp);

	/* Find the TCP control block that corresponds with this packet */
	tp = intotcpcb(inp);

	/*
	 * If we can't find the TCP control block (happens occasionaly for a
	 * packet sent during the shutdown phase of a TCP connection),
	 * or we're in the timewait state, bail
	 */
	if (tp == NULL || inp->inp_flags & INP_TIMEWAIT) {
		if (dir == PFIL_IN)
			ss->nskip_in_tcpcb++;
		else
			ss->nskip_out_tcpcb++;

		goto inp_unlock;
	}

	pn = malloc(sizeof(struct pkt_node), M_SIFTR_PKTNODE, M_NOWAIT|M_ZERO);

	if (pn == NULL) {
		if (dir == PFIL_IN)
			ss->nskip_in_malloc++;
		else
			ss->nskip_out_malloc++;

		goto inp_unlock;
	}

	siftr_siftdata(pn, inp, tp, INP_IPV4, dir, inp_locally_locked);

	if (siftr_generate_hashes) {
		if ((*m)->m_pkthdr.csum_flags & CSUM_TCP) {
			/*
			 * For outbound packets, the TCP checksum isn't
			 * calculated yet. This is a problem for our packet
			 * hashing as the receiver will calc a different hash
			 * to ours if we don't include the correct TCP checksum
			 * in the bytes being hashed. To work around this
			 * problem, we manually calc the TCP checksum here in
			 * software. We unset the CSUM_TCP flag so the lower
			 * layers don't recalc it.
			 */
			(*m)->m_pkthdr.csum_flags &= ~CSUM_TCP;

			/*
			 * Calculate the TCP checksum in software and assign
			 * to correct TCP header field, which will follow the
			 * packet mbuf down the stack. The trick here is that
			 * tcp_output() sets th->th_sum to the checksum of the
			 * pseudo header for us already. Because of the nature
			 * of the checksumming algorithm, we can sum over the
			 * entire IP payload (i.e. TCP header and data), which
			 * will include the already calculated pseduo header
			 * checksum, thus giving us the complete TCP checksum.
			 *
			 * To put it in simple terms, if checksum(1,2,3,4)=10,
			 * then checksum(1,2,3,4,5) == checksum(10,5).
			 * This property is what allows us to "cheat" and
			 * checksum only the IP payload which has the TCP
			 * th_sum field populated with the pseudo header's
			 * checksum, and not need to futz around checksumming
			 * pseudo header bytes and TCP header/data in one hit.
			 * Refer to RFC 1071 for more info.
			 *
			 * NB: in_cksum_skip(struct mbuf *m, int len, int skip)
			 * in_cksum_skip 2nd argument is NOT the number of
			 * bytes to read from the mbuf at "skip" bytes offset
			 * from the start of the mbuf (very counter intuitive!).
			 * The number of bytes to read is calculated internally
			 * by the function as len-skip i.e. to sum over the IP
			 * payload (TCP header + data) bytes, it is INCORRECT
			 * to call the function like this:
			 * in_cksum_skip(at, ip->ip_len - offset, offset)
			 * Rather, it should be called like this:
			 * in_cksum_skip(at, ip->ip_len, offset)
			 * which means read "ip->ip_len - offset" bytes from
			 * the mbuf cluster "at" at offset "offset" bytes from
			 * the beginning of the "at" mbuf's data pointer.
			 */
			th->th_sum = in_cksum_skip(*m, ntohs(ip->ip_len),
			    ip_hl);
		}

		/*
		 * XXX: Having to calculate the checksum in software and then
		 * hash over all bytes is really inefficient. Would be nice to
		 * find a way to create the hash and checksum in the same pass
		 * over the bytes.
		 */
		pn->hash = hash_pkt(*m, ip_hl);
	}

	mtx_lock(&siftr_pkt_queue_mtx);
	STAILQ_INSERT_TAIL(&pkt_queue, pn, nodes);
	mtx_unlock(&siftr_pkt_queue_mtx);
	goto ret;

inp_unlock:
	if (inp_locally_locked)
		INP_RUNLOCK(inp);

ret:
	/* Returning 0 ensures pfil will not discard the pkt */
	return (0);
}


#ifdef SIFTR_IPV6
static int
siftr_chkpkt6(void *arg, struct mbuf **m, struct ifnet *ifp, int dir,
    struct inpcb *inp)
{
	struct pkt_node *pn;
	struct ip6_hdr *ip6;
	struct tcphdr *th;
	struct tcpcb *tp;
	struct siftr_stats *ss;
	unsigned int ip6_hl;
	int inp_locally_locked;

	inp_locally_locked = 0;
	ss = DPCPU_PTR(ss);

	/*
	 * m_pullup is not required here because ip6_{input|output}
	 * already do the heavy lifting for us.
	 */

	ip6 = mtod(*m, struct ip6_hdr *);

	/*
	 * Only continue processing if the packet is TCP
	 * XXX: We should follow the next header fields
	 * as shown on Pg 6 RFC 2460, but right now we'll
	 * only check pkts that have no extension headers.
	 */
	if (ip6->ip6_nxt != IPPROTO_TCP)
		goto ret6;

	/*
	 * If a kernel subsystem reinjects packets into the stack, our pfil
	 * hook will be called multiple times for the same packet.
	 * Make sure we only process unique packets.
	 */
	if (siftr_chkreinject(*m, dir, ss))
		goto ret6;

	if (dir == PFIL_IN)
		ss->n_in++;
	else
		ss->n_out++;

	ip6_hl = sizeof(struct ip6_hdr);

	/*
	 * Create a tcphdr struct starting at the correct offset
	 * in the ipv6 packet. ip->ip_hl gives the ip header length
	 * in 4-byte words, so multiply it to get the size in bytes.
	 */
	th = (struct tcphdr *)((caddr_t)ip6 + ip6_hl);

	/*
	 * For inbound packets, the pfil hooks don't provide a pointer to the
	 * inpcb, so we need to find it ourselves and lock it.
	 */
	if (!inp) {
		/* Find the corresponding inpcb for this pkt. */
		inp = siftr_findinpcb(INP_IPV6, (struct ip *)ip6, *m,
		    th->th_sport, th->th_dport, dir, ss);

		if (inp == NULL)
			goto ret6;
		else
			inp_locally_locked = 1;
	}

	/* Find the TCP control block that corresponds with this packet. */
	tp = intotcpcb(inp);

	/*
	 * If we can't find the TCP control block (happens occasionaly for a
	 * packet sent during the shutdown phase of a TCP connection),
	 * or we're in the timewait state, bail.
	 */
	if (tp == NULL || inp->inp_flags & INP_TIMEWAIT) {
		if (dir == PFIL_IN)
			ss->nskip_in_tcpcb++;
		else
			ss->nskip_out_tcpcb++;

		goto inp_unlock6;
	}

	pn = malloc(sizeof(struct pkt_node), M_SIFTR_PKTNODE, M_NOWAIT|M_ZERO);

	if (pn == NULL) {
		if (dir == PFIL_IN)
			ss->nskip_in_malloc++;
		else
			ss->nskip_out_malloc++;

		goto inp_unlock6;
	}

	siftr_siftdata(pn, inp, tp, INP_IPV6, dir, inp_locally_locked);

	/* XXX: Figure out how to generate hashes for IPv6 packets. */

	mtx_lock(&siftr_pkt_queue_mtx);
	STAILQ_INSERT_TAIL(&pkt_queue, pn, nodes);
	mtx_unlock(&siftr_pkt_queue_mtx);
	goto ret6;

inp_unlock6:
	if (inp_locally_locked)
		INP_RUNLOCK(inp);

ret6:
	/* Returning 0 ensures pfil will not discard the pkt. */
	return (0);
}
#endif /* #ifdef SIFTR_IPV6 */


static int
siftr_pfil(int action)
{
	struct pfil_head *pfh_inet;
#ifdef SIFTR_IPV6
	struct pfil_head *pfh_inet6;
#endif
	VNET_ITERATOR_DECL(vnet_iter);

	VNET_LIST_RLOCK();
	VNET_FOREACH(vnet_iter) {
		CURVNET_SET(vnet_iter);
		pfh_inet = pfil_head_get(PFIL_TYPE_AF, AF_INET);
#ifdef SIFTR_IPV6
		pfh_inet6 = pfil_head_get(PFIL_TYPE_AF, AF_INET6);
#endif

		if (action == HOOK) {
			pfil_add_hook(siftr_chkpkt, NULL,
			    PFIL_IN | PFIL_OUT | PFIL_WAITOK, pfh_inet);
#ifdef SIFTR_IPV6
			pfil_add_hook(siftr_chkpkt6, NULL,
			    PFIL_IN | PFIL_OUT | PFIL_WAITOK, pfh_inet6);
#endif
		} else if (action == UNHOOK) {
			pfil_remove_hook(siftr_chkpkt, NULL,
			    PFIL_IN | PFIL_OUT | PFIL_WAITOK, pfh_inet);
#ifdef SIFTR_IPV6
			pfil_remove_hook(siftr_chkpkt6, NULL,
			    PFIL_IN | PFIL_OUT | PFIL_WAITOK, pfh_inet6);
#endif
		}
		CURVNET_RESTORE();
	}
	VNET_LIST_RUNLOCK();

	return (0);
}


static int
siftr_sysctl_logfile_name_handler(SYSCTL_HANDLER_ARGS)
{
	struct alq *new_alq;
	int error;

	if (req->newptr == NULL)
		goto skip;

	/* If old filename and new filename are different. */
	if (strncmp(siftr_logfile, (char *)req->newptr, PATH_MAX)) {

		error = alq_open(&new_alq, req->newptr, curthread->td_ucred,
		    SIFTR_LOG_FILE_MODE, SIFTR_ALQ_BUFLEN, 0);

		/* Bail if unable to create new alq. */
		if (error)
			return (1);

		/*
		 * If disabled, siftr_alq == NULL so we simply close
		 * the alq as we've proved it can be opened.
		 * If enabled, close the existing alq and switch the old
		 * for the new.
		 */
		if (siftr_alq == NULL)
			alq_close(new_alq);
		else {
			alq_close(siftr_alq);
			siftr_alq = new_alq;
		}
	}

skip:
	return (sysctl_handle_string(oidp, arg1, arg2, req));
}


static int
siftr_manage_ops(uint8_t action)
{
	struct siftr_stats totalss;
	struct timeval tval;
	struct flow_hash_node *counter, *tmp_counter;
	struct sbuf *s;
	int i, key_index, ret, error;
	uint32_t bytes_to_write, total_skipped_pkts;
	uint16_t lport, fport;
	uint8_t *key, ipver;

#ifdef SIFTR_IPV6
	uint32_t laddr[4];
	uint32_t faddr[4];
#else
	uint8_t laddr[4];
	uint8_t faddr[4];
#endif

	error = 0;
	total_skipped_pkts = 0;

	/* Init an autosizing sbuf that initially holds 200 chars. */
	if ((s = sbuf_new(NULL, NULL, 200, SBUF_AUTOEXTEND)) == NULL)
		return (-1);

	if (action == SIFTR_ENABLE) {
		/*
		 * Create our alq
		 * XXX: We should abort if alq_open fails!
		 */
		alq_open(&siftr_alq, siftr_logfile, curthread->td_ucred,
		    SIFTR_LOG_FILE_MODE, SIFTR_ALQ_BUFLEN, 0);

		STAILQ_INIT(&pkt_queue);

		DPCPU_ZERO(ss);

		siftr_exit_pkt_manager_thread = 0;

		ret = kthread_add(&siftr_pkt_manager_thread, NULL, NULL,
		    &siftr_pkt_manager_thr, RFNOWAIT, 0,
		    "siftr_pkt_manager_thr");

		siftr_pfil(HOOK);

		microtime(&tval);

		sbuf_printf(s,
		    "enable_time_secs=%jd\tenable_time_usecs=%06ld\t"
		    "siftrver=%s\thz=%u\ttcp_rtt_scale=%u\tsysname=%s\t"
		    "sysver=%u\tipmode=%u\n",
		    (intmax_t)tval.tv_sec, tval.tv_usec, MODVERSION_STR, hz,
		    TCP_RTT_SCALE, SYS_NAME, __FreeBSD_version, SIFTR_IPMODE);

		sbuf_finish(s);
		alq_writen(siftr_alq, sbuf_data(s), sbuf_len(s), ALQ_WAITOK);

	} else if (action == SIFTR_DISABLE && siftr_pkt_manager_thr != NULL) {
		/*
		 * Remove the pfil hook functions. All threads currently in
		 * the hook functions are allowed to exit before siftr_pfil()
		 * returns.
		 */
		siftr_pfil(UNHOOK);

		/* This will block until the pkt manager thread unlocks it. */
		mtx_lock(&siftr_pkt_mgr_mtx);

		/* Tell the pkt manager thread that it should exit now. */
		siftr_exit_pkt_manager_thread = 1;

		/*
		 * Wake the pkt_manager thread so it realises that
		 * siftr_exit_pkt_manager_thread == 1 and exits gracefully.
		 * The wakeup won't be delivered until we unlock
		 * siftr_pkt_mgr_mtx so this isn't racy.
		 */
		wakeup(&wait_for_pkt);

		/* Wait for the pkt_manager thread to exit. */
		mtx_sleep(siftr_pkt_manager_thr, &siftr_pkt_mgr_mtx, PWAIT,
		    "thrwait", 0);

		siftr_pkt_manager_thr = NULL;
		mtx_unlock(&siftr_pkt_mgr_mtx);

		totalss.n_in = DPCPU_VARSUM(ss, n_in);
		totalss.n_out = DPCPU_VARSUM(ss, n_out);
		totalss.nskip_in_malloc = DPCPU_VARSUM(ss, nskip_in_malloc);
		totalss.nskip_out_malloc = DPCPU_VARSUM(ss, nskip_out_malloc);
		totalss.nskip_in_mtx = DPCPU_VARSUM(ss, nskip_in_mtx);
		totalss.nskip_out_mtx = DPCPU_VARSUM(ss, nskip_out_mtx);
		totalss.nskip_in_tcpcb = DPCPU_VARSUM(ss, nskip_in_tcpcb);
		totalss.nskip_out_tcpcb = DPCPU_VARSUM(ss, nskip_out_tcpcb);
		totalss.nskip_in_inpcb = DPCPU_VARSUM(ss, nskip_in_inpcb);
		totalss.nskip_out_inpcb = DPCPU_VARSUM(ss, nskip_out_inpcb);

		total_skipped_pkts = totalss.nskip_in_malloc +
		    totalss.nskip_out_malloc + totalss.nskip_in_mtx +
		    totalss.nskip_out_mtx + totalss.nskip_in_tcpcb +
		    totalss.nskip_out_tcpcb + totalss.nskip_in_inpcb +
		    totalss.nskip_out_inpcb;

		microtime(&tval);

		sbuf_printf(s,
		    "disable_time_secs=%jd\tdisable_time_usecs=%06ld\t"
		    "num_inbound_tcp_pkts=%ju\tnum_outbound_tcp_pkts=%ju\t"
		    "total_tcp_pkts=%ju\tnum_inbound_skipped_pkts_malloc=%u\t"
		    "num_outbound_skipped_pkts_malloc=%u\t"
		    "num_inbound_skipped_pkts_mtx=%u\t"
		    "num_outbound_skipped_pkts_mtx=%u\t"
		    "num_inbound_skipped_pkts_tcpcb=%u\t"
		    "num_outbound_skipped_pkts_tcpcb=%u\t"
		    "num_inbound_skipped_pkts_inpcb=%u\t"
		    "num_outbound_skipped_pkts_inpcb=%u\t"
		    "total_skipped_tcp_pkts=%u\tflow_list=",
		    (intmax_t)tval.tv_sec,
		    tval.tv_usec,
		    (uintmax_t)totalss.n_in,
		    (uintmax_t)totalss.n_out,
		    (uintmax_t)(totalss.n_in + totalss.n_out),
		    totalss.nskip_in_malloc,
		    totalss.nskip_out_malloc,
		    totalss.nskip_in_mtx,
		    totalss.nskip_out_mtx,
		    totalss.nskip_in_tcpcb,
		    totalss.nskip_out_tcpcb,
		    totalss.nskip_in_inpcb,
		    totalss.nskip_out_inpcb,
		    total_skipped_pkts);

		/*
		 * Iterate over the flow hash, printing a summary of each
		 * flow seen and freeing any malloc'd memory.
		 * The hash consists of an array of LISTs (man 3 queue).
		 */
		for (i = 0; i <= siftr_hashmask; i++) {
			LIST_FOREACH_SAFE(counter, counter_hash + i, nodes,
			    tmp_counter) {
				key = counter->key;
				key_index = 1;

				ipver = key[0];

				memcpy(laddr, key + key_index, sizeof(laddr));
				key_index += sizeof(laddr);
				memcpy(&lport, key + key_index, sizeof(lport));
				key_index += sizeof(lport);
				memcpy(faddr, key + key_index, sizeof(faddr));
				key_index += sizeof(faddr);
				memcpy(&fport, key + key_index, sizeof(fport));

#ifdef SIFTR_IPV6
				laddr[3] = ntohl(laddr[3]);
				faddr[3] = ntohl(faddr[3]);

				if (ipver == INP_IPV6) {
					laddr[0] = ntohl(laddr[0]);
					laddr[1] = ntohl(laddr[1]);
					laddr[2] = ntohl(laddr[2]);
					faddr[0] = ntohl(faddr[0]);
					faddr[1] = ntohl(faddr[1]);
					faddr[2] = ntohl(faddr[2]);

					sbuf_printf(s,
					    "%x:%x:%x:%x:%x:%x:%x:%x;%u-"
					    "%x:%x:%x:%x:%x:%x:%x:%x;%u,",
					    UPPER_SHORT(laddr[0]),
					    LOWER_SHORT(laddr[0]),
					    UPPER_SHORT(laddr[1]),
					    LOWER_SHORT(laddr[1]),
					    UPPER_SHORT(laddr[2]),
					    LOWER_SHORT(laddr[2]),
					    UPPER_SHORT(laddr[3]),
					    LOWER_SHORT(laddr[3]),
					    ntohs(lport),
					    UPPER_SHORT(faddr[0]),
					    LOWER_SHORT(faddr[0]),
					    UPPER_SHORT(faddr[1]),
					    LOWER_SHORT(faddr[1]),
					    UPPER_SHORT(faddr[2]),
					    LOWER_SHORT(faddr[2]),
					    UPPER_SHORT(faddr[3]),
					    LOWER_SHORT(faddr[3]),
					    ntohs(fport));
				} else {
					laddr[0] = FIRST_OCTET(laddr[3]);
					laddr[1] = SECOND_OCTET(laddr[3]);
					laddr[2] = THIRD_OCTET(laddr[3]);
					laddr[3] = FOURTH_OCTET(laddr[3]);
					faddr[0] = FIRST_OCTET(faddr[3]);
					faddr[1] = SECOND_OCTET(faddr[3]);
					faddr[2] = THIRD_OCTET(faddr[3]);
					faddr[3] = FOURTH_OCTET(faddr[3]);
#endif
					sbuf_printf(s,
					    "%u.%u.%u.%u;%u-%u.%u.%u.%u;%u,",
					    laddr[0],
					    laddr[1],
					    laddr[2],
					    laddr[3],
					    ntohs(lport),
					    faddr[0],
					    faddr[1],
					    faddr[2],
					    faddr[3],
					    ntohs(fport));
#ifdef SIFTR_IPV6
				}
#endif

				free(counter, M_SIFTR_HASHNODE);
			}

			LIST_INIT(counter_hash + i);
		}

		sbuf_printf(s, "\n");
		sbuf_finish(s);

		i = 0;
		do {
			bytes_to_write = min(SIFTR_ALQ_BUFLEN, sbuf_len(s)-i);
			alq_writen(siftr_alq, sbuf_data(s)+i, bytes_to_write, ALQ_WAITOK);
			i += bytes_to_write;
		} while (i < sbuf_len(s));

		alq_close(siftr_alq);
		siftr_alq = NULL;
	}

	sbuf_delete(s);

	/*
	 * XXX: Should be using ret to check if any functions fail
	 * and set error appropriately
	 */

	return (error);
}


static int
siftr_sysctl_enabled_handler(SYSCTL_HANDLER_ARGS)
{
	if (req->newptr == NULL)
		goto skip;

	/* If the value passed in isn't 0 or 1, return an error. */
	if (CAST_PTR_INT(req->newptr) != 0 && CAST_PTR_INT(req->newptr) != 1)
		return (1);

	/* If we are changing state (0 to 1 or 1 to 0). */
	if (CAST_PTR_INT(req->newptr) != siftr_enabled )
		if (siftr_manage_ops(CAST_PTR_INT(req->newptr))) {
			siftr_manage_ops(SIFTR_DISABLE);
			return (1);
		}

skip:
	return (sysctl_handle_int(oidp, arg1, arg2, req));
}


static void
siftr_shutdown_handler(void *arg)
{
	siftr_manage_ops(SIFTR_DISABLE);
}


/*
 * Module is being unloaded or machine is shutting down. Take care of cleanup.
 */
static int
deinit_siftr(void)
{
	/* Cleanup. */
	siftr_manage_ops(SIFTR_DISABLE);
	hashdestroy(counter_hash, M_SIFTR, siftr_hashmask);
	mtx_destroy(&siftr_pkt_queue_mtx);
	mtx_destroy(&siftr_pkt_mgr_mtx);

	return (0);
}


/*
 * Module has just been loaded into the kernel.
 */
static int
init_siftr(void)
{
	EVENTHANDLER_REGISTER(shutdown_pre_sync, siftr_shutdown_handler, NULL,
	    SHUTDOWN_PRI_FIRST);

	/* Initialise our flow counter hash table. */
	counter_hash = hashinit(SIFTR_EXPECTED_MAX_TCP_FLOWS, M_SIFTR,
	    &siftr_hashmask);

	mtx_init(&siftr_pkt_queue_mtx, "siftr_pkt_queue_mtx", NULL, MTX_DEF);
	mtx_init(&siftr_pkt_mgr_mtx, "siftr_pkt_mgr_mtx", NULL, MTX_DEF);

	/* Print message to the user's current terminal. */
	uprintf("\nStatistical Information For TCP Research (SIFTR) %s\n"
	    "          http://caia.swin.edu.au/urp/newtcp\n\n",
	    MODVERSION_STR);

	return (0);
}


/*
 * This is the function that is called to load and unload the module.
 * When the module is loaded, this function is called once with
 * "what" == MOD_LOAD
 * When the module is unloaded, this function is called twice with
 * "what" = MOD_QUIESCE first, followed by "what" = MOD_UNLOAD second
 * When the system is shut down e.g. CTRL-ALT-DEL or using the shutdown command,
 * this function is called once with "what" = MOD_SHUTDOWN
 * When the system is shut down, the handler isn't called until the very end
 * of the shutdown sequence i.e. after the disks have been synced.
 */
static int
siftr_load_handler(module_t mod, int what, void *arg)
{
	int ret;

	switch (what) {
	case MOD_LOAD:
		ret = init_siftr();
		break;

	case MOD_QUIESCE:
	case MOD_SHUTDOWN:
		ret = deinit_siftr();
		break;

	case MOD_UNLOAD:
		ret = 0;
		break;

	default:
		ret = EINVAL;
		break;
	}

	return (ret);
}


static moduledata_t siftr_mod = {
	.name = "siftr",
	.evhand = siftr_load_handler,
};

/*
 * Param 1: name of the kernel module
 * Param 2: moduledata_t struct containing info about the kernel module
 *          and the execution entry point for the module
 * Param 3: From sysinit_sub_id enumeration in /usr/include/sys/kernel.h
 *          Defines the module initialisation order
 * Param 4: From sysinit_elem_order enumeration in /usr/include/sys/kernel.h
 *          Defines the initialisation order of this kld relative to others
 *          within the same subsystem as defined by param 3
 */
DECLARE_MODULE(siftr, siftr_mod, SI_SUB_SMP, SI_ORDER_ANY);
MODULE_DEPEND(siftr, alq, 1, 1, 1);
MODULE_VERSION(siftr, MODVERSION);
OpenPOWER on IntegriCloud