aboutsummaryrefslogtreecommitdiff
path: root/sys/netpfil/pf/pf_norm.c
blob: 47407014adf09a7582e6cd950a4e3720d658ad7f (plain) (blame)
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
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
/*-
 * SPDX-License-Identifier: BSD-2-Clause
 *
 * Copyright 2001 Niels Provos <provos@citi.umich.edu>
 * Copyright 2011-2018 Alexander Bluhm <bluhm@openbsd.org>
 * All rights reserved.
 *
 * 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 AUTHOR ``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 AUTHOR 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.
 *
 *	$OpenBSD: pf_norm.c,v 1.114 2009/01/29 14:11:45 henning Exp $
 */

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

#include "opt_inet.h"
#include "opt_inet6.h"
#include "opt_pf.h"

#include <sys/param.h>
#include <sys/kernel.h>
#include <sys/lock.h>
#include <sys/mbuf.h>
#include <sys/mutex.h>
#include <sys/refcount.h>
#include <sys/socket.h>

#include <net/if.h>
#include <net/vnet.h>
#include <net/pfvar.h>
#include <net/if_pflog.h>

#include <netinet/in.h>
#include <netinet/ip.h>
#include <netinet/ip_var.h>
#include <netinet6/ip6_var.h>
#include <netinet/tcp.h>
#include <netinet/tcp_fsm.h>
#include <netinet/tcp_seq.h>

#ifdef INET6
#include <netinet/ip6.h>
#endif /* INET6 */

struct pf_frent {
	TAILQ_ENTRY(pf_frent)	fr_next;
	struct mbuf	*fe_m;
	uint16_t	fe_hdrlen;	/* ipv4 header length with ip options
					   ipv6, extension, fragment header */
	uint16_t	fe_extoff;	/* last extension header offset or 0 */
	uint16_t	fe_len;		/* fragment length */
	uint16_t	fe_off;		/* fragment offset */
	uint16_t	fe_mff;		/* more fragment flag */
};

struct pf_fragment_cmp {
	struct pf_addr	frc_src;
	struct pf_addr	frc_dst;
	uint32_t	frc_id;
	sa_family_t	frc_af;
	uint8_t		frc_proto;
};

struct pf_fragment {
	struct pf_fragment_cmp	fr_key;
#define fr_src	fr_key.frc_src
#define fr_dst	fr_key.frc_dst
#define fr_id	fr_key.frc_id
#define fr_af	fr_key.frc_af
#define fr_proto	fr_key.frc_proto

	/* pointers to queue element */
	struct pf_frent	*fr_firstoff[PF_FRAG_ENTRY_POINTS];
	/* count entries between pointers */
	uint8_t	fr_entries[PF_FRAG_ENTRY_POINTS];
	RB_ENTRY(pf_fragment) fr_entry;
	TAILQ_ENTRY(pf_fragment) frag_next;
	uint32_t	fr_timeout;
	uint16_t	fr_maxlen;	/* maximum length of single fragment */
	u_int16_t	fr_holes;	/* number of holes in the queue */
	TAILQ_HEAD(pf_fragq, pf_frent) fr_queue;
};

struct pf_fragment_tag {
	uint16_t	ft_hdrlen;	/* header length of reassembled pkt */
	uint16_t	ft_extoff;	/* last extension header offset or 0 */
	uint16_t	ft_maxlen;	/* maximum fragment payload length */
	uint32_t	ft_id;		/* fragment id */
};

VNET_DEFINE_STATIC(struct mtx, pf_frag_mtx);
#define V_pf_frag_mtx		VNET(pf_frag_mtx)
#define PF_FRAG_LOCK()		mtx_lock(&V_pf_frag_mtx)
#define PF_FRAG_UNLOCK()	mtx_unlock(&V_pf_frag_mtx)
#define PF_FRAG_ASSERT()	mtx_assert(&V_pf_frag_mtx, MA_OWNED)

VNET_DEFINE(uma_zone_t, pf_state_scrub_z);	/* XXX: shared with pfsync */

VNET_DEFINE_STATIC(uma_zone_t, pf_frent_z);
#define	V_pf_frent_z	VNET(pf_frent_z)
VNET_DEFINE_STATIC(uma_zone_t, pf_frag_z);
#define	V_pf_frag_z	VNET(pf_frag_z)

TAILQ_HEAD(pf_fragqueue, pf_fragment);
TAILQ_HEAD(pf_cachequeue, pf_fragment);
VNET_DEFINE_STATIC(struct pf_fragqueue,	pf_fragqueue);
#define	V_pf_fragqueue			VNET(pf_fragqueue)
RB_HEAD(pf_frag_tree, pf_fragment);
VNET_DEFINE_STATIC(struct pf_frag_tree,	pf_frag_tree);
#define	V_pf_frag_tree			VNET(pf_frag_tree)
static int		 pf_frag_compare(struct pf_fragment *,
			    struct pf_fragment *);
static RB_PROTOTYPE(pf_frag_tree, pf_fragment, fr_entry, pf_frag_compare);
static RB_GENERATE(pf_frag_tree, pf_fragment, fr_entry, pf_frag_compare);

static void	pf_flush_fragments(void);
static void	pf_free_fragment(struct pf_fragment *);
static void	pf_remove_fragment(struct pf_fragment *);
static int	pf_normalize_tcpopt(struct pf_rule *, struct mbuf *,
		    struct tcphdr *, int, sa_family_t);
static struct pf_frent *pf_create_fragment(u_short *);
static int	pf_frent_holes(struct pf_frent *frent);
static struct pf_fragment *pf_find_fragment(struct pf_fragment_cmp *key,
		    struct pf_frag_tree *tree);
static inline int	pf_frent_index(struct pf_frent *);
static int	pf_frent_insert(struct pf_fragment *,
			    struct pf_frent *, struct pf_frent *);
void			pf_frent_remove(struct pf_fragment *,
			    struct pf_frent *);
struct pf_frent		*pf_frent_previous(struct pf_fragment *,
			    struct pf_frent *);
static struct pf_fragment *pf_fillup_fragment(struct pf_fragment_cmp *,
		    struct pf_frent *, u_short *);
static struct mbuf *pf_join_fragment(struct pf_fragment *);
#ifdef INET
static void	pf_scrub_ip(struct mbuf **, uint32_t, uint8_t, uint8_t);
static int	pf_reassemble(struct mbuf **, struct ip *, int, u_short *);
#endif	/* INET */
#ifdef INET6
static int	pf_reassemble6(struct mbuf **, struct ip6_hdr *,
		    struct ip6_frag *, uint16_t, uint16_t, u_short *);
static void	pf_scrub_ip6(struct mbuf **, uint8_t);
#endif	/* INET6 */

#define	DPFPRINTF(x) do {				\
	if (V_pf_status.debug >= PF_DEBUG_MISC) {	\
		printf("%s: ", __func__);		\
		printf x ;				\
	}						\
} while(0)

#ifdef INET
static void
pf_ip2key(struct ip *ip, int dir, struct pf_fragment_cmp *key)
{

	key->frc_src.v4 = ip->ip_src;
	key->frc_dst.v4 = ip->ip_dst;
	key->frc_af = AF_INET;
	key->frc_proto = ip->ip_p;
	key->frc_id = ip->ip_id;
}
#endif	/* INET */

void
pf_normalize_init(void)
{

	V_pf_frag_z = uma_zcreate("pf frags", sizeof(struct pf_fragment),
	    NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, 0);
	V_pf_frent_z = uma_zcreate("pf frag entries", sizeof(struct pf_frent),
	    NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, 0);
	V_pf_state_scrub_z = uma_zcreate("pf state scrubs",
	    sizeof(struct pf_state_scrub),  NULL, NULL, NULL, NULL,
	    UMA_ALIGN_PTR, 0);

	mtx_init(&V_pf_frag_mtx, "pf fragments", NULL, MTX_DEF);

	V_pf_limits[PF_LIMIT_FRAGS].zone = V_pf_frent_z;
	V_pf_limits[PF_LIMIT_FRAGS].limit = PFFRAG_FRENT_HIWAT;
	uma_zone_set_max(V_pf_frent_z, PFFRAG_FRENT_HIWAT);
	uma_zone_set_warning(V_pf_frent_z, "PF frag entries limit reached");

	TAILQ_INIT(&V_pf_fragqueue);
}

void
pf_normalize_cleanup(void)
{

	uma_zdestroy(V_pf_state_scrub_z);
	uma_zdestroy(V_pf_frent_z);
	uma_zdestroy(V_pf_frag_z);

	mtx_destroy(&V_pf_frag_mtx);
}

static int
pf_frag_compare(struct pf_fragment *a, struct pf_fragment *b)
{
	int	diff;

	if ((diff = a->fr_id - b->fr_id) != 0)
		return (diff);
	if ((diff = a->fr_proto - b->fr_proto) != 0)
		return (diff);
	if ((diff = a->fr_af - b->fr_af) != 0)
		return (diff);
	if ((diff = pf_addr_cmp(&a->fr_src, &b->fr_src, a->fr_af)) != 0)
		return (diff);
	if ((diff = pf_addr_cmp(&a->fr_dst, &b->fr_dst, a->fr_af)) != 0)
		return (diff);
	return (0);
}

void
pf_purge_expired_fragments(void)
{
	u_int32_t	expire = time_uptime -
			    V_pf_default_rule.timeout[PFTM_FRAG];

	pf_purge_fragments(expire);
}

void
pf_purge_fragments(uint32_t expire)
{
	struct pf_fragment	*frag;

	PF_FRAG_LOCK();
	while ((frag = TAILQ_LAST(&V_pf_fragqueue, pf_fragqueue)) != NULL) {
		if (frag->fr_timeout > expire)
			break;

		DPFPRINTF(("expiring %d(%p)\n", frag->fr_id, frag));
		pf_free_fragment(frag);
	}

	PF_FRAG_UNLOCK();
}

/*
 * Try to flush old fragments to make space for new ones
 */
static void
pf_flush_fragments(void)
{
	struct pf_fragment	*frag;
	int			 goal;

	PF_FRAG_ASSERT();

	goal = uma_zone_get_cur(V_pf_frent_z) * 9 / 10;
	DPFPRINTF(("trying to free %d frag entriess\n", goal));
	while (goal < uma_zone_get_cur(V_pf_frent_z)) {
		frag = TAILQ_LAST(&V_pf_fragqueue, pf_fragqueue);
		if (frag)
			pf_free_fragment(frag);
		else
			break;
	}
}

/* Frees the fragments and all associated entries */
static void
pf_free_fragment(struct pf_fragment *frag)
{
	struct pf_frent		*frent;

	PF_FRAG_ASSERT();

	/* Free all fragments */
	for (frent = TAILQ_FIRST(&frag->fr_queue); frent;
	    frent = TAILQ_FIRST(&frag->fr_queue)) {
		TAILQ_REMOVE(&frag->fr_queue, frent, fr_next);

		m_freem(frent->fe_m);
		uma_zfree(V_pf_frent_z, frent);
	}

	pf_remove_fragment(frag);
}

static struct pf_fragment *
pf_find_fragment(struct pf_fragment_cmp *key, struct pf_frag_tree *tree)
{
	struct pf_fragment	*frag;

	PF_FRAG_ASSERT();

	frag = RB_FIND(pf_frag_tree, tree, (struct pf_fragment *)key);
	if (frag != NULL) {
		/* XXX Are we sure we want to update the timeout? */
		frag->fr_timeout = time_uptime;
		TAILQ_REMOVE(&V_pf_fragqueue, frag, frag_next);
		TAILQ_INSERT_HEAD(&V_pf_fragqueue, frag, frag_next);
	}

	return (frag);
}

/* Removes a fragment from the fragment queue and frees the fragment */
static void
pf_remove_fragment(struct pf_fragment *frag)
{

	PF_FRAG_ASSERT();
	KASSERT(frag, ("frag != NULL"));

	RB_REMOVE(pf_frag_tree, &V_pf_frag_tree, frag);
	TAILQ_REMOVE(&V_pf_fragqueue, frag, frag_next);
	uma_zfree(V_pf_frag_z, frag);
}

static struct pf_frent *
pf_create_fragment(u_short *reason)
{
	struct pf_frent *frent;

	PF_FRAG_ASSERT();

	frent = uma_zalloc(V_pf_frent_z, M_NOWAIT);
	if (frent == NULL) {
		pf_flush_fragments();
		frent = uma_zalloc(V_pf_frent_z, M_NOWAIT);
		if (frent == NULL) {
			REASON_SET(reason, PFRES_MEMORY);
			return (NULL);
		}
	}

	return (frent);
}

/*
 * Calculate the additional holes that were created in the fragment
 * queue by inserting this fragment.  A fragment in the middle
 * creates one more hole by splitting.  For each connected side,
 * it loses one hole.
 * Fragment entry must be in the queue when calling this function.
 */
static int
pf_frent_holes(struct pf_frent *frent)
{
	struct pf_frent *prev = TAILQ_PREV(frent, pf_fragq, fr_next);
	struct pf_frent *next = TAILQ_NEXT(frent, fr_next);
	int holes = 1;

	if (prev == NULL) {
		if (frent->fe_off == 0)
			holes--;
	} else {
		KASSERT(frent->fe_off != 0, ("frent->fe_off != 0"));
		if (frent->fe_off == prev->fe_off + prev->fe_len)
			holes--;
	}
	if (next == NULL) {
		if (!frent->fe_mff)
			holes--;
	} else {
		KASSERT(frent->fe_mff, ("frent->fe_mff"));
		if (next->fe_off == frent->fe_off + frent->fe_len)
			holes--;
	}
	return holes;
}

static inline int
pf_frent_index(struct pf_frent *frent)
{
	/*
	 * We have an array of 16 entry points to the queue.  A full size
	 * 65535 octet IP packet can have 8192 fragments.  So the queue
	 * traversal length is at most 512 and at most 16 entry points are
	 * checked.  We need 128 additional bytes on a 64 bit architecture.
	 */
	CTASSERT(((u_int16_t)0xffff &~ 7) / (0x10000 / PF_FRAG_ENTRY_POINTS) ==
	    16 - 1);
	CTASSERT(((u_int16_t)0xffff >> 3) / PF_FRAG_ENTRY_POINTS == 512 - 1);

	return frent->fe_off / (0x10000 / PF_FRAG_ENTRY_POINTS);
}

static int
pf_frent_insert(struct pf_fragment *frag, struct pf_frent *frent,
    struct pf_frent *prev)
{
	int index;

	CTASSERT(PF_FRAG_ENTRY_LIMIT <= 0xff);

	/*
	 * A packet has at most 65536 octets.  With 16 entry points, each one
	 * spawns 4096 octets.  We limit these to 64 fragments each, which
	 * means on average every fragment must have at least 64 octets.
	 */
	index = pf_frent_index(frent);
	if (frag->fr_entries[index] >= PF_FRAG_ENTRY_LIMIT)
		return ENOBUFS;
	frag->fr_entries[index]++;

	if (prev == NULL) {
		TAILQ_INSERT_HEAD(&frag->fr_queue, frent, fr_next);
	} else {
		KASSERT(prev->fe_off + prev->fe_len <= frent->fe_off,
		    ("overlapping fragment"));
		TAILQ_INSERT_AFTER(&frag->fr_queue, prev, frent, fr_next);
	}

	if (frag->fr_firstoff[index] == NULL) {
		KASSERT(prev == NULL || pf_frent_index(prev) < index,
		    ("prev == NULL || pf_frent_index(pref) < index"));
		frag->fr_firstoff[index] = frent;
	} else {
		if (frent->fe_off < frag->fr_firstoff[index]->fe_off) {
			KASSERT(prev == NULL || pf_frent_index(prev) < index,
			    ("prev == NULL || pf_frent_index(pref) < index"));
			frag->fr_firstoff[index] = frent;
		} else {
			KASSERT(prev != NULL, ("prev != NULL"));
			KASSERT(pf_frent_index(prev) == index,
			    ("pf_frent_index(prev) == index"));
		}
	}

	frag->fr_holes += pf_frent_holes(frent);

	return 0;
}

void
pf_frent_remove(struct pf_fragment *frag, struct pf_frent *frent)
{
#ifdef INVARIANTS
	struct pf_frent *prev = TAILQ_PREV(frent, pf_fragq, fr_next);
#endif
	struct pf_frent *next = TAILQ_NEXT(frent, fr_next);
	int index;

	frag->fr_holes -= pf_frent_holes(frent);

	index = pf_frent_index(frent);
	KASSERT(frag->fr_firstoff[index] != NULL, ("frent not found"));
	if (frag->fr_firstoff[index]->fe_off == frent->fe_off) {
		if (next == NULL) {
			frag->fr_firstoff[index] = NULL;
		} else {
			KASSERT(frent->fe_off + frent->fe_len <= next->fe_off,
			    ("overlapping fragment"));
			if (pf_frent_index(next) == index) {
				frag->fr_firstoff[index] = next;
			} else {
				frag->fr_firstoff[index] = NULL;
			}
		}
	} else {
		KASSERT(frag->fr_firstoff[index]->fe_off < frent->fe_off,
		    ("frag->fr_firstoff[index]->fe_off < frent->fe_off"));
		KASSERT(prev != NULL, ("prev != NULL"));
		KASSERT(prev->fe_off + prev->fe_len <= frent->fe_off,
		    ("overlapping fragment"));
		KASSERT(pf_frent_index(prev) == index,
		    ("pf_frent_index(prev) == index"));
	}

	TAILQ_REMOVE(&frag->fr_queue, frent, fr_next);

	KASSERT(frag->fr_entries[index] > 0, ("No fragments remaining"));
	frag->fr_entries[index]--;
}

struct pf_frent *
pf_frent_previous(struct pf_fragment *frag, struct pf_frent *frent)
{
	struct pf_frent *prev, *next;
	int index;

	/*
	 * If there are no fragments after frag, take the final one.  Assume
	 * that the global queue is not empty.
	 */
	prev = TAILQ_LAST(&frag->fr_queue, pf_fragq);
	KASSERT(prev != NULL, ("prev != NULL"));
	if (prev->fe_off <= frent->fe_off)
		return prev;
	/*
	 * We want to find a fragment entry that is before frag, but still
	 * close to it.  Find the first fragment entry that is in the same
	 * entry point or in the first entry point after that.  As we have
	 * already checked that there are entries behind frag, this will
	 * succeed.
	 */
	for (index = pf_frent_index(frent); index < PF_FRAG_ENTRY_POINTS;
	    index++) {
		prev = frag->fr_firstoff[index];
		if (prev != NULL)
			break;
	}
	KASSERT(prev != NULL, ("prev != NULL"));
	/*
	 * In prev we may have a fragment from the same entry point that is
	 * before frent, or one that is just one position behind frent.
	 * In the latter case, we go back one step and have the predecessor.
	 * There may be none if the new fragment will be the first one.
	 */
	if (prev->fe_off > frent->fe_off) {
		prev = TAILQ_PREV(prev, pf_fragq, fr_next);
		if (prev == NULL)
			return NULL;
		KASSERT(prev->fe_off <= frent->fe_off,
		    ("prev->fe_off <= frent->fe_off"));
		return prev;
	}
	/*
	 * In prev is the first fragment of the entry point.  The offset
	 * of frag is behind it.  Find the closest previous fragment.
	 */
	for (next = TAILQ_NEXT(prev, fr_next); next != NULL;
	    next = TAILQ_NEXT(next, fr_next)) {
		if (next->fe_off > frent->fe_off)
			break;
		prev = next;
	}
	return prev;
}

static struct pf_fragment *
pf_fillup_fragment(struct pf_fragment_cmp *key, struct pf_frent *frent,
    u_short *reason)
{
	struct pf_frent		*after, *next, *prev;
	struct pf_fragment	*frag;
	uint16_t		total;

	PF_FRAG_ASSERT();

	/* No empty fragments. */
	if (frent->fe_len == 0) {
		DPFPRINTF(("bad fragment: len 0\n"));
		goto bad_fragment;
	}

	/* All fragments are 8 byte aligned. */
	if (frent->fe_mff && (frent->fe_len & 0x7)) {
		DPFPRINTF(("bad fragment: mff and len %d\n", frent->fe_len));
		goto bad_fragment;
	}

	/* Respect maximum length, IP_MAXPACKET == IPV6_MAXPACKET. */
	if (frent->fe_off + frent->fe_len > IP_MAXPACKET) {
		DPFPRINTF(("bad fragment: max packet %d\n",
		    frent->fe_off + frent->fe_len));
		goto bad_fragment;
	}

	DPFPRINTF((key->frc_af == AF_INET ?
	    "reass frag %d @ %d-%d\n" : "reass frag %#08x @ %d-%d\n",
	    key->frc_id, frent->fe_off, frent->fe_off + frent->fe_len));

	/* Fully buffer all of the fragments in this fragment queue. */
	frag = pf_find_fragment(key, &V_pf_frag_tree);

	/* Create a new reassembly queue for this packet. */
	if (frag == NULL) {
		frag = uma_zalloc(V_pf_frag_z, M_NOWAIT);
		if (frag == NULL) {
			pf_flush_fragments();
			frag = uma_zalloc(V_pf_frag_z, M_NOWAIT);
			if (frag == NULL) {
				REASON_SET(reason, PFRES_MEMORY);
				goto drop_fragment;
			}
		}

		*(struct pf_fragment_cmp *)frag = *key;
		memset(frag->fr_firstoff, 0, sizeof(frag->fr_firstoff));
		memset(frag->fr_entries, 0, sizeof(frag->fr_entries));
		frag->fr_timeout = time_uptime;
		frag->fr_maxlen = frent->fe_len;
		frag->fr_holes = 1;
		TAILQ_INIT(&frag->fr_queue);

		RB_INSERT(pf_frag_tree, &V_pf_frag_tree, frag);
		TAILQ_INSERT_HEAD(&V_pf_fragqueue, frag, frag_next);

		/* We do not have a previous fragment, cannot fail. */
		pf_frent_insert(frag, frent, NULL);

		return (frag);
	}

	KASSERT(!TAILQ_EMPTY(&frag->fr_queue), ("!TAILQ_EMPTY()->fr_queue"));

	/* Remember maximum fragment len for refragmentation. */
	if (frent->fe_len > frag->fr_maxlen)
		frag->fr_maxlen = frent->fe_len;

	/* Maximum data we have seen already. */
	total = TAILQ_LAST(&frag->fr_queue, pf_fragq)->fe_off +
		TAILQ_LAST(&frag->fr_queue, pf_fragq)->fe_len;

	/* Non terminal fragments must have more fragments flag. */
	if (frent->fe_off + frent->fe_len < total && !frent->fe_mff)
		goto bad_fragment;

	/* Check if we saw the last fragment already. */
	if (!TAILQ_LAST(&frag->fr_queue, pf_fragq)->fe_mff) {
		if (frent->fe_off + frent->fe_len > total ||
		    (frent->fe_off + frent->fe_len == total && frent->fe_mff))
			goto bad_fragment;
	} else {
		if (frent->fe_off + frent->fe_len == total && !frent->fe_mff)
			goto bad_fragment;
	}

	/* Find neighbors for newly inserted fragment */
	prev = pf_frent_previous(frag, frent);
	if (prev == NULL) {
		after = TAILQ_FIRST(&frag->fr_queue);
		KASSERT(after != NULL, ("after != NULL"));
	} else {
		after = TAILQ_NEXT(prev, fr_next);
	}

	if (prev != NULL && prev->fe_off + prev->fe_len > frent->fe_off) {
		uint16_t precut;

		precut = prev->fe_off + prev->fe_len - frent->fe_off;
		if (precut >= frent->fe_len)
			goto bad_fragment;
		DPFPRINTF(("overlap -%d\n", precut));
		m_adj(frent->fe_m, precut);
		frent->fe_off += precut;
		frent->fe_len -= precut;
	}

	for (; after != NULL && frent->fe_off + frent->fe_len > after->fe_off;
	    after = next) {
		uint16_t aftercut;

		aftercut = frent->fe_off + frent->fe_len - after->fe_off;
		DPFPRINTF(("adjust overlap %d\n", aftercut));
		if (aftercut < after->fe_len) {
			m_adj(after->fe_m, aftercut);
			after->fe_off += aftercut;
			after->fe_len -= aftercut;
			break;
		}

		/* This fragment is completely overlapped, lose it. */
		next = TAILQ_NEXT(after, fr_next);
		pf_frent_remove(frag, after);
		m_freem(after->fe_m);
		uma_zfree(V_pf_frent_z, after);
	}

	/* If part of the queue gets too long, there is not way to recover. */
	if (pf_frent_insert(frag, frent, prev)) {
		DPFPRINTF(("fragment queue limit exceeded\n"));
		goto bad_fragment;
	}

	return (frag);

bad_fragment:
	REASON_SET(reason, PFRES_FRAG);
drop_fragment:
	uma_zfree(V_pf_frent_z, frent);
	return (NULL);
}

static struct mbuf *
pf_join_fragment(struct pf_fragment *frag)
{
	struct mbuf *m, *m2;
	struct pf_frent	*frent, *next;

	frent = TAILQ_FIRST(&frag->fr_queue);
	next = TAILQ_NEXT(frent, fr_next);

	m = frent->fe_m;
	m_adj(m, (frent->fe_hdrlen + frent->fe_len) - m->m_pkthdr.len);
	uma_zfree(V_pf_frent_z, frent);
	for (frent = next; frent != NULL; frent = next) {
		next = TAILQ_NEXT(frent, fr_next);

		m2 = frent->fe_m;
		/* Strip off ip header. */
		m_adj(m2, frent->fe_hdrlen);
		/* Strip off any trailing bytes. */
		m_adj(m2, frent->fe_len - m2->m_pkthdr.len);

		uma_zfree(V_pf_frent_z, frent);
		m_cat(m, m2);
	}

	/* Remove from fragment queue. */
	pf_remove_fragment(frag);

	return (m);
}

#ifdef INET
static int
pf_reassemble(struct mbuf **m0, struct ip *ip, int dir, u_short *reason)
{
	struct mbuf		*m = *m0;
	struct pf_frent		*frent;
	struct pf_fragment	*frag;
	struct pf_fragment_cmp	key;
	uint16_t		total, hdrlen;

	/* Get an entry for the fragment queue */
	if ((frent = pf_create_fragment(reason)) == NULL)
		return (PF_DROP);

	frent->fe_m = m;
	frent->fe_hdrlen = ip->ip_hl << 2;
	frent->fe_extoff = 0;
	frent->fe_len = ntohs(ip->ip_len) - (ip->ip_hl << 2);
	frent->fe_off = (ntohs(ip->ip_off) & IP_OFFMASK) << 3;
	frent->fe_mff = ntohs(ip->ip_off) & IP_MF;

	pf_ip2key(ip, dir, &key);

	if ((frag = pf_fillup_fragment(&key, frent, reason)) == NULL)
		return (PF_DROP);

	/* The mbuf is part of the fragment entry, no direct free or access */
	m = *m0 = NULL;

	if (frag->fr_holes) {
		DPFPRINTF(("frag %d, holes %d\n", frag->fr_id, frag->fr_holes));
		return (PF_PASS);  /* drop because *m0 is NULL, no error */
	}

	/* We have all the data */
	frent = TAILQ_FIRST(&frag->fr_queue);
	KASSERT(frent != NULL, ("frent != NULL"));
	total = TAILQ_LAST(&frag->fr_queue, pf_fragq)->fe_off +
		TAILQ_LAST(&frag->fr_queue, pf_fragq)->fe_len;
	hdrlen = frent->fe_hdrlen;

	m = *m0 = pf_join_fragment(frag);
	frag = NULL;

	if (m->m_flags & M_PKTHDR) {
		int plen = 0;
		for (m = *m0; m; m = m->m_next)
			plen += m->m_len;
		m = *m0;
		m->m_pkthdr.len = plen;
	}

	ip = mtod(m, struct ip *);
	ip->ip_len = htons(hdrlen + total);
	ip->ip_off &= ~(IP_MF|IP_OFFMASK);

	if (hdrlen + total > IP_MAXPACKET) {
		DPFPRINTF(("drop: too big: %d\n", total));
		ip->ip_len = 0;
		REASON_SET(reason, PFRES_SHORT);
		/* PF_DROP requires a valid mbuf *m0 in pf_test() */
		return (PF_DROP);
	}

	DPFPRINTF(("complete: %p(%d)\n", m, ntohs(ip->ip_len)));
	return (PF_PASS);
}
#endif	/* INET */

#ifdef INET6
static int
pf_reassemble6(struct mbuf **m0, struct ip6_hdr *ip6, struct ip6_frag *fraghdr,
    uint16_t hdrlen, uint16_t extoff, u_short *reason)
{
	struct mbuf		*m = *m0;
	struct pf_frent		*frent;
	struct pf_fragment	*frag;
	struct pf_fragment_cmp	 key;
	struct m_tag		*mtag;
	struct pf_fragment_tag	*ftag;
	int			 off;
	uint32_t		 frag_id;
	uint16_t		 total, maxlen;
	uint8_t			 proto;

	PF_FRAG_LOCK();

	/* Get an entry for the fragment queue. */
	if ((frent = pf_create_fragment(reason)) == NULL) {
		PF_FRAG_UNLOCK();
		return (PF_DROP);
	}

	frent->fe_m = m;
	frent->fe_hdrlen = hdrlen;
	frent->fe_extoff = extoff;
	frent->fe_len = sizeof(struct ip6_hdr) + ntohs(ip6->ip6_plen) - hdrlen;
	frent->fe_off = ntohs(fraghdr->ip6f_offlg & IP6F_OFF_MASK);
	frent->fe_mff = fraghdr->ip6f_offlg & IP6F_MORE_FRAG;

	key.frc_src.v6 = ip6->ip6_src;
	key.frc_dst.v6 = ip6->ip6_dst;
	key.frc_af = AF_INET6;
	/* Only the first fragment's protocol is relevant. */
	key.frc_proto = 0;
	key.frc_id = fraghdr->ip6f_ident;

	if ((frag = pf_fillup_fragment(&key, frent, reason)) == NULL) {
		PF_FRAG_UNLOCK();
		return (PF_DROP);
	}

	/* The mbuf is part of the fragment entry, no direct free or access. */
	m = *m0 = NULL;

	if (frag->fr_holes) {
		DPFPRINTF(("frag %d, holes %d\n", frag->fr_id,
		    frag->fr_holes));
		PF_FRAG_UNLOCK();
		return (PF_PASS);  /* Drop because *m0 is NULL, no error. */
	}

	/* We have all the data. */
	frent = TAILQ_FIRST(&frag->fr_queue);
	KASSERT(frent != NULL, ("frent != NULL"));
	extoff = frent->fe_extoff;
	maxlen = frag->fr_maxlen;
	frag_id = frag->fr_id;
	total = TAILQ_LAST(&frag->fr_queue, pf_fragq)->fe_off +
		TAILQ_LAST(&frag->fr_queue, pf_fragq)->fe_len;
	hdrlen = frent->fe_hdrlen - sizeof(struct ip6_frag);

	m = *m0 = pf_join_fragment(frag);
	frag = NULL;

	PF_FRAG_UNLOCK();

	/* Take protocol from first fragment header. */
	m = m_getptr(m, hdrlen + offsetof(struct ip6_frag, ip6f_nxt), &off);
	KASSERT(m, ("%s: short mbuf chain", __func__));
	proto = *(mtod(m, caddr_t) + off);
	m = *m0;

	/* Delete frag6 header */
	if (ip6_deletefraghdr(m, hdrlen, M_NOWAIT) != 0)
		goto fail;

	if (m->m_flags & M_PKTHDR) {
		int plen = 0;
		for (m = *m0; m; m = m->m_next)
			plen += m->m_len;
		m = *m0;
		m->m_pkthdr.len = plen;
	}

	if ((mtag = m_tag_get(PF_REASSEMBLED, sizeof(struct pf_fragment_tag),
	    M_NOWAIT)) == NULL)
		goto fail;
	ftag = (struct pf_fragment_tag *)(mtag + 1);
	ftag->ft_hdrlen = hdrlen;
	ftag->ft_extoff = extoff;
	ftag->ft_maxlen = maxlen;
	ftag->ft_id = frag_id;
	m_tag_prepend(m, mtag);

	ip6 = mtod(m, struct ip6_hdr *);
	ip6->ip6_plen = htons(hdrlen - sizeof(struct ip6_hdr) + total);
	if (extoff) {
		/* Write protocol into next field of last extension header. */
		m = m_getptr(m, extoff + offsetof(struct ip6_ext, ip6e_nxt),
		    &off);
		KASSERT(m, ("%s: short mbuf chain", __func__));
		*(mtod(m, char *) + off) = proto;
		m = *m0;
	} else
		ip6->ip6_nxt = proto;

	if (hdrlen - sizeof(struct ip6_hdr) + total > IPV6_MAXPACKET) {
		DPFPRINTF(("drop: too big: %d\n", total));
		ip6->ip6_plen = 0;
		REASON_SET(reason, PFRES_SHORT);
		/* PF_DROP requires a valid mbuf *m0 in pf_test6(). */
		return (PF_DROP);
	}

	DPFPRINTF(("complete: %p(%d)\n", m, ntohs(ip6->ip6_plen)));
	return (PF_PASS);

fail:
	REASON_SET(reason, PFRES_MEMORY);
	/* PF_DROP requires a valid mbuf *m0 in pf_test6(), will free later. */
	return (PF_DROP);
}
#endif	/* INET6 */

#ifdef INET6
int
pf_refragment6(struct ifnet *ifp, struct mbuf **m0, struct m_tag *mtag)
{
	struct mbuf		*m = *m0, *t;
	struct pf_fragment_tag	*ftag = (struct pf_fragment_tag *)(mtag + 1);
	struct pf_pdesc		 pd;
	uint32_t		 frag_id;
	uint16_t		 hdrlen, extoff, maxlen;
	uint8_t			 proto;
	int			 error, action;

	hdrlen = ftag->ft_hdrlen;
	extoff = ftag->ft_extoff;
	maxlen = ftag->ft_maxlen;
	frag_id = ftag->ft_id;
	m_tag_delete(m, mtag);
	mtag = NULL;
	ftag = NULL;

	if (extoff) {
		int off;

		/* Use protocol from next field of last extension header */
		m = m_getptr(m, extoff + offsetof(struct ip6_ext, ip6e_nxt),
		    &off);
		KASSERT((m != NULL), ("pf_refragment6: short mbuf chain"));
		proto = *(mtod(m, caddr_t) + off);
		*(mtod(m, char *) + off) = IPPROTO_FRAGMENT;
		m = *m0;
	} else {
		struct ip6_hdr *hdr;

		hdr = mtod(m, struct ip6_hdr *);
		proto = hdr->ip6_nxt;
		hdr->ip6_nxt = IPPROTO_FRAGMENT;
	}

	/* The MTU must be a multiple of 8 bytes, or we risk doing the
	 * fragmentation wrong. */
	maxlen = maxlen & ~7;

	/*
	 * Maxlen may be less than 8 if there was only a single
	 * fragment.  As it was fragmented before, add a fragment
	 * header also for a single fragment.  If total or maxlen
	 * is less than 8, ip6_fragment() will return EMSGSIZE and
	 * we drop the packet.
	 */
	error = ip6_fragment(ifp, m, hdrlen, proto, maxlen, frag_id);
	m = (*m0)->m_nextpkt;
	(*m0)->m_nextpkt = NULL;
	if (error == 0) {
		/* The first mbuf contains the unfragmented packet. */
		m_freem(*m0);
		*m0 = NULL;
		action = PF_PASS;
	} else {
		/* Drop expects an mbuf to free. */
		DPFPRINTF(("refragment error %d\n", error));
		action = PF_DROP;
	}
	for (t = m; m; m = t) {
		t = m->m_nextpkt;
		m->m_nextpkt = NULL;
		m->m_flags |= M_SKIP_FIREWALL;
		memset(&pd, 0, sizeof(pd));
		pd.pf_mtag = pf_find_mtag(m);
		if (error == 0)
			ip6_forward(m, 0);
		else
			m_freem(m);
	}

	return (action);
}
#endif /* INET6 */

#ifdef INET
int
pf_normalize_ip(struct mbuf **m0, int dir, struct pfi_kif *kif, u_short *reason,
    struct pf_pdesc *pd)
{
	struct mbuf		*m = *m0;
	struct pf_rule		*r;
	struct ip		*h = mtod(m, struct ip *);
	int			 mff = (ntohs(h->ip_off) & IP_MF);
	int			 hlen = h->ip_hl << 2;
	u_int16_t		 fragoff = (ntohs(h->ip_off) & IP_OFFMASK) << 3;
	u_int16_t		 max;
	int			 ip_len;
	int			 ip_off;
	int			 tag = -1;
	int			 verdict;

	PF_RULES_RASSERT();

	r = TAILQ_FIRST(pf_main_ruleset.rules[PF_RULESET_SCRUB].active.ptr);
	while (r != NULL) {
		r->evaluations++;
		if (pfi_kif_match(r->kif, kif) == r->ifnot)
			r = r->skip[PF_SKIP_IFP].ptr;
		else if (r->direction && r->direction != dir)
			r = r->skip[PF_SKIP_DIR].ptr;
		else if (r->af && r->af != AF_INET)
			r = r->skip[PF_SKIP_AF].ptr;
		else if (r->proto && r->proto != h->ip_p)
			r = r->skip[PF_SKIP_PROTO].ptr;
		else if (PF_MISMATCHAW(&r->src.addr,
		    (struct pf_addr *)&h->ip_src.s_addr, AF_INET,
		    r->src.neg, kif, M_GETFIB(m)))
			r = r->skip[PF_SKIP_SRC_ADDR].ptr;
		else if (PF_MISMATCHAW(&r->dst.addr,
		    (struct pf_addr *)&h->ip_dst.s_addr, AF_INET,
		    r->dst.neg, NULL, M_GETFIB(m)))
			r = r->skip[PF_SKIP_DST_ADDR].ptr;
		else if (r->match_tag && !pf_match_tag(m, r, &tag,
		    pd->pf_mtag ? pd->pf_mtag->tag : 0))
			r = TAILQ_NEXT(r, entries);
		else
			break;
	}

	if (r == NULL || r->action == PF_NOSCRUB)
		return (PF_PASS);
	else {
		r->packets[dir == PF_OUT]++;
		r->bytes[dir == PF_OUT] += pd->tot_len;
	}

	/* Check for illegal packets */
	if (hlen < (int)sizeof(struct ip)) {
		REASON_SET(reason, PFRES_NORM);
		goto drop;
	}

	if (hlen > ntohs(h->ip_len)) {
		REASON_SET(reason, PFRES_NORM);
		goto drop;
	}

	/* Clear IP_DF if the rule uses the no-df option */
	if (r->rule_flag & PFRULE_NODF && h->ip_off & htons(IP_DF)) {
		u_int16_t ip_off = h->ip_off;

		h->ip_off &= htons(~IP_DF);
		h->ip_sum = pf_cksum_fixup(h->ip_sum, ip_off, h->ip_off, 0);
	}

	/* We will need other tests here */
	if (!fragoff && !mff)
		goto no_fragment;

	/* We're dealing with a fragment now. Don't allow fragments
	 * with IP_DF to enter the cache. If the flag was cleared by
	 * no-df above, fine. Otherwise drop it.
	 */
	if (h->ip_off & htons(IP_DF)) {
		DPFPRINTF(("IP_DF\n"));
		goto bad;
	}

	ip_len = ntohs(h->ip_len) - hlen;
	ip_off = (ntohs(h->ip_off) & IP_OFFMASK) << 3;

	/* All fragments are 8 byte aligned */
	if (mff && (ip_len & 0x7)) {
		DPFPRINTF(("mff and %d\n", ip_len));
		goto bad;
	}

	/* Respect maximum length */
	if (fragoff + ip_len > IP_MAXPACKET) {
		DPFPRINTF(("max packet %d\n", fragoff + ip_len));
		goto bad;
	}
	max = fragoff + ip_len;

	/* Fully buffer all of the fragments
	 * Might return a completely reassembled mbuf, or NULL */
	PF_FRAG_LOCK();
	DPFPRINTF(("reass frag %d @ %d-%d\n", h->ip_id, fragoff, max));
	verdict = pf_reassemble(m0, h, dir, reason);
	PF_FRAG_UNLOCK();

	if (verdict != PF_PASS)
		return (PF_DROP);

	m = *m0;
	if (m == NULL)
		return (PF_DROP);

	h = mtod(m, struct ip *);

 no_fragment:
	/* At this point, only IP_DF is allowed in ip_off */
	if (h->ip_off & ~htons(IP_DF)) {
		u_int16_t ip_off = h->ip_off;

		h->ip_off &= htons(IP_DF);
		h->ip_sum = pf_cksum_fixup(h->ip_sum, ip_off, h->ip_off, 0);
	}

	pf_scrub_ip(&m, r->rule_flag, r->min_ttl, r->set_tos);

	return (PF_PASS);

 bad:
	DPFPRINTF(("dropping bad fragment\n"));
	REASON_SET(reason, PFRES_FRAG);
 drop:
	if (r != NULL && r->log)
		PFLOG_PACKET(kif, m, AF_INET, dir, *reason, r, NULL, NULL, pd,
		    1);

	return (PF_DROP);
}
#endif

#ifdef INET6
int
pf_normalize_ip6(struct mbuf **m0, int dir, struct pfi_kif *kif,
    u_short *reason, struct pf_pdesc *pd)
{
	struct mbuf		*m = *m0;
	struct pf_rule		*r;
	struct ip6_hdr		*h = mtod(m, struct ip6_hdr *);
	int			 extoff;
	int			 off;
	struct ip6_ext		 ext;
	struct ip6_opt		 opt;
	struct ip6_frag		 frag;
	u_int32_t		 plen;
	int			 optend;
	int			 ooff;
	u_int8_t		 proto;
	int			 terminal;

	PF_RULES_RASSERT();

	r = TAILQ_FIRST(pf_main_ruleset.rules[PF_RULESET_SCRUB].active.ptr);
	while (r != NULL) {
		r->evaluations++;
		if (pfi_kif_match(r->kif, kif) == r->ifnot)
			r = r->skip[PF_SKIP_IFP].ptr;
		else if (r->direction && r->direction != dir)
			r = r->skip[PF_SKIP_DIR].ptr;
		else if (r->af && r->af != AF_INET6)
			r = r->skip[PF_SKIP_AF].ptr;
#if 0 /* header chain! */
		else if (r->proto && r->proto != h->ip6_nxt)
			r = r->skip[PF_SKIP_PROTO].ptr;
#endif
		else if (PF_MISMATCHAW(&r->src.addr,
		    (struct pf_addr *)&h->ip6_src, AF_INET6,
		    r->src.neg, kif, M_GETFIB(m)))
			r = r->skip[PF_SKIP_SRC_ADDR].ptr;
		else if (PF_MISMATCHAW(&r->dst.addr,
		    (struct pf_addr *)&h->ip6_dst, AF_INET6,
		    r->dst.neg, NULL, M_GETFIB(m)))
			r = r->skip[PF_SKIP_DST_ADDR].ptr;
		else
			break;
	}

	if (r == NULL || r->action == PF_NOSCRUB)
		return (PF_PASS);
	else {
		r->packets[dir == PF_OUT]++;
		r->bytes[dir == PF_OUT] += pd->tot_len;
	}

	/* Check for illegal packets */
	if (sizeof(struct ip6_hdr) + IPV6_MAXPACKET < m->m_pkthdr.len)
		goto drop;

	plen = ntohs(h->ip6_plen);
	/* jumbo payload option not supported */
	if (plen == 0)
		goto drop;

	extoff = 0;
	off = sizeof(struct ip6_hdr);
	proto = h->ip6_nxt;
	terminal = 0;
	do {
		switch (proto) {
		case IPPROTO_FRAGMENT:
			goto fragment;
			break;
		case IPPROTO_AH:
		case IPPROTO_ROUTING:
		case IPPROTO_DSTOPTS:
			if (!pf_pull_hdr(m, off, &ext, sizeof(ext), NULL,
			    NULL, AF_INET6))
				goto shortpkt;
			extoff = off;
			if (proto == IPPROTO_AH)
				off += (ext.ip6e_len + 2) * 4;
			else
				off += (ext.ip6e_len + 1) * 8;
			proto = ext.ip6e_nxt;
			break;
		case IPPROTO_HOPOPTS:
			if (!pf_pull_hdr(m, off, &ext, sizeof(ext), NULL,
			    NULL, AF_INET6))
				goto shortpkt;
			extoff = off;
			optend = off + (ext.ip6e_len + 1) * 8;
			ooff = off + sizeof(ext);
			do {
				if (!pf_pull_hdr(m, ooff, &opt.ip6o_type,
				    sizeof(opt.ip6o_type), NULL, NULL,
				    AF_INET6))
					goto shortpkt;
				if (opt.ip6o_type == IP6OPT_PAD1) {
					ooff++;
					continue;
				}
				if (!pf_pull_hdr(m, ooff, &opt, sizeof(opt),
				    NULL, NULL, AF_INET6))
					goto shortpkt;
				if (ooff + sizeof(opt) + opt.ip6o_len > optend)
					goto drop;
				if (opt.ip6o_type == IP6OPT_JUMBO)
					goto drop;
				ooff += sizeof(opt) + opt.ip6o_len;
			} while (ooff < optend);

			off = optend;
			proto = ext.ip6e_nxt;
			break;
		default:
			terminal = 1;
			break;
		}
	} while (!terminal);

	if (sizeof(struct ip6_hdr) + plen > m->m_pkthdr.len)
		goto shortpkt;

	pf_scrub_ip6(&m, r->min_ttl);

	return (PF_PASS);

 fragment:
	if (sizeof(struct ip6_hdr) + plen > m->m_pkthdr.len)
		goto shortpkt;

	if (!pf_pull_hdr(m, off, &frag, sizeof(frag), NULL, NULL, AF_INET6))
		goto shortpkt;

	/* Offset now points to data portion. */
	off += sizeof(frag);

	/* Returns PF_DROP or *m0 is NULL or completely reassembled mbuf. */
	if (pf_reassemble6(m0, h, &frag, off, extoff, reason) != PF_PASS)
		return (PF_DROP);
	m = *m0;
	if (m == NULL)
		return (PF_DROP);

	pd->flags |= PFDESC_IP_REAS;
	return (PF_PASS);

 shortpkt:
	REASON_SET(reason, PFRES_SHORT);
	if (r != NULL && r->log)
		PFLOG_PACKET(kif, m, AF_INET6, dir, *reason, r, NULL, NULL, pd,
		    1);
	return (PF_DROP);

 drop:
	REASON_SET(reason, PFRES_NORM);
	if (r != NULL && r->log)
		PFLOG_PACKET(kif, m, AF_INET6, dir, *reason, r, NULL, NULL, pd,
		    1);
	return (PF_DROP);
}
#endif /* INET6 */

int
pf_normalize_tcp(int dir, struct pfi_kif *kif, struct mbuf *m, int ipoff,
    int off, void *h, struct pf_pdesc *pd)
{
	struct pf_rule	*r, *rm = NULL;
	struct tcphdr	*th = pd->hdr.tcp;
	int		 rewrite = 0;
	u_short		 reason;
	u_int8_t	 flags;
	sa_family_t	 af = pd->af;

	PF_RULES_RASSERT();

	r = TAILQ_FIRST(pf_main_ruleset.rules[PF_RULESET_SCRUB].active.ptr);
	while (r != NULL) {
		r->evaluations++;
		if (pfi_kif_match(r->kif, kif) == r->ifnot)
			r = r->skip[PF_SKIP_IFP].ptr;
		else if (r->direction && r->direction != dir)
			r = r->skip[PF_SKIP_DIR].ptr;
		else if (r->af && r->af != af)
			r = r->skip[PF_SKIP_AF].ptr;
		else if (r->proto && r->proto != pd->proto)
			r = r->skip[PF_SKIP_PROTO].ptr;
		else if (PF_MISMATCHAW(&r->src.addr, pd->src, af,
		    r->src.neg, kif, M_GETFIB(m)))
			r = r->skip[PF_SKIP_SRC_ADDR].ptr;
		else if (r->src.port_op && !pf_match_port(r->src.port_op,
			    r->src.port[0], r->src.port[1], th->th_sport))
			r = r->skip[PF_SKIP_SRC_PORT].ptr;
		else if (PF_MISMATCHAW(&r->dst.addr, pd->dst, af,
		    r->dst.neg, NULL, M_GETFIB(m)))
			r = r->skip[PF_SKIP_DST_ADDR].ptr;
		else if (r->dst.port_op && !pf_match_port(r->dst.port_op,
			    r->dst.port[0], r->dst.port[1], th->th_dport))
			r = r->skip[PF_SKIP_DST_PORT].ptr;
		else if (r->os_fingerprint != PF_OSFP_ANY && !pf_osfp_match(
			    pf_osfp_fingerprint(pd, m, off, th),
			    r->os_fingerprint))
			r = TAILQ_NEXT(r, entries);
		else {
			rm = r;
			break;
		}
	}

	if (rm == NULL || rm->action == PF_NOSCRUB)
		return (PF_PASS);
	else {
		r->packets[dir == PF_OUT]++;
		r->bytes[dir == PF_OUT] += pd->tot_len;
	}

	if (rm->rule_flag & PFRULE_REASSEMBLE_TCP)
		pd->flags |= PFDESC_TCP_NORM;

	flags = th->th_flags;
	if (flags & TH_SYN) {
		/* Illegal packet */
		if (flags & TH_RST)
			goto tcp_drop;

		if (flags & TH_FIN)
			goto tcp_drop;
	} else {
		/* Illegal packet */
		if (!(flags & (TH_ACK|TH_RST)))
			goto tcp_drop;
	}

	if (!(flags & TH_ACK)) {
		/* These flags are only valid if ACK is set */
		if ((flags & TH_FIN) || (flags & TH_PUSH) || (flags & TH_URG))
			goto tcp_drop;
	}

	/* Check for illegal header length */
	if (th->th_off < (sizeof(struct tcphdr) >> 2))
		goto tcp_drop;

	/* If flags changed, or reserved data set, then adjust */
	if (flags != th->th_flags || th->th_x2 != 0) {
		u_int16_t	ov, nv;

		ov = *(u_int16_t *)(&th->th_ack + 1);
		th->th_flags = flags;
		th->th_x2 = 0;
		nv = *(u_int16_t *)(&th->th_ack + 1);

		th->th_sum = pf_proto_cksum_fixup(m, th->th_sum, ov, nv, 0);
		rewrite = 1;
	}

	/* Remove urgent pointer, if TH_URG is not set */
	if (!(flags & TH_URG) && th->th_urp) {
		th->th_sum = pf_proto_cksum_fixup(m, th->th_sum, th->th_urp,
		    0, 0);
		th->th_urp = 0;
		rewrite = 1;
	}

	/* Process options */
	if (r->max_mss && pf_normalize_tcpopt(r, m, th, off, pd->af))
		rewrite = 1;

	/* copy back packet headers if we sanitized */
	if (rewrite)
		m_copyback(m, off, sizeof(*th), (caddr_t)th);

	return (PF_PASS);

 tcp_drop:
	REASON_SET(&reason, PFRES_NORM);
	if (rm != NULL && r->log)
		PFLOG_PACKET(kif, m, AF_INET, dir, reason, r, NULL, NULL, pd,
		    1);
	return (PF_DROP);
}

int
pf_normalize_tcp_init(struct mbuf *m, int off, struct pf_pdesc *pd,
    struct tcphdr *th, struct pf_state_peer *src, struct pf_state_peer *dst)
{
	u_int32_t tsval, tsecr;
	u_int8_t hdr[60];
	u_int8_t *opt;

	KASSERT((src->scrub == NULL),
	    ("pf_normalize_tcp_init: src->scrub != NULL"));

	src->scrub = uma_zalloc(V_pf_state_scrub_z, M_ZERO | M_NOWAIT);
	if (src->scrub == NULL)
		return (1);

	switch (pd->af) {
#ifdef INET
	case AF_INET: {
		struct ip *h = mtod(m, struct ip *);
		src->scrub->pfss_ttl = h->ip_ttl;
		break;
	}
#endif /* INET */
#ifdef INET6
	case AF_INET6: {
		struct ip6_hdr *h = mtod(m, struct ip6_hdr *);
		src->scrub->pfss_ttl = h->ip6_hlim;
		break;
	}
#endif /* INET6 */
	}


	/*
	 * All normalizations below are only begun if we see the start of
	 * the connections.  They must all set an enabled bit in pfss_flags
	 */
	if ((th->th_flags & TH_SYN) == 0)
		return (0);


	if (th->th_off > (sizeof(struct tcphdr) >> 2) && src->scrub &&
	    pf_pull_hdr(m, off, hdr, th->th_off << 2, NULL, NULL, pd->af)) {
		/* Diddle with TCP options */
		int hlen;
		opt = hdr + sizeof(struct tcphdr);
		hlen = (th->th_off << 2) - sizeof(struct tcphdr);
		while (hlen >= TCPOLEN_TIMESTAMP) {
			switch (*opt) {
			case TCPOPT_EOL:	/* FALLTHROUGH */
			case TCPOPT_NOP:
				opt++;
				hlen--;
				break;
			case TCPOPT_TIMESTAMP:
				if (opt[1] >= TCPOLEN_TIMESTAMP) {
					src->scrub->pfss_flags |=
					    PFSS_TIMESTAMP;
					src->scrub->pfss_ts_mod =
					    htonl(arc4random());

					/* note PFSS_PAWS not set yet */
					memcpy(&tsval, &opt[2],
					    sizeof(u_int32_t));
					memcpy(&tsecr, &opt[6],
					    sizeof(u_int32_t));
					src->scrub->pfss_tsval0 = ntohl(tsval);
					src->scrub->pfss_tsval = ntohl(tsval);
					src->scrub->pfss_tsecr = ntohl(tsecr);
					getmicrouptime(&src->scrub->pfss_last);
				}
				/* FALLTHROUGH */
			default:
				hlen -= MAX(opt[1], 2);
				opt += MAX(opt[1], 2);
				break;
			}
		}
	}

	return (0);
}

void
pf_normalize_tcp_cleanup(struct pf_state *state)
{
	if (state->src.scrub)
		uma_zfree(V_pf_state_scrub_z, state->src.scrub);
	if (state->dst.scrub)
		uma_zfree(V_pf_state_scrub_z, state->dst.scrub);

	/* Someday... flush the TCP segment reassembly descriptors. */
}

int
pf_normalize_tcp_stateful(struct mbuf *m, int off, struct pf_pdesc *pd,
    u_short *reason, struct tcphdr *th, struct pf_state *state,
    struct pf_state_peer *src, struct pf_state_peer *dst, int *writeback)
{
	struct timeval uptime;
	u_int32_t tsval, tsecr;
	u_int tsval_from_last;
	u_int8_t hdr[60];
	u_int8_t *opt;
	int copyback = 0;
	int got_ts = 0;

	KASSERT((src->scrub || dst->scrub),
	    ("%s: src->scrub && dst->scrub!", __func__));

	/*
	 * Enforce the minimum TTL seen for this connection.  Negate a common
	 * technique to evade an intrusion detection system and confuse
	 * firewall state code.
	 */
	switch (pd->af) {
#ifdef INET
	case AF_INET: {
		if (src->scrub) {
			struct ip *h = mtod(m, struct ip *);
			if (h->ip_ttl > src->scrub->pfss_ttl)
				src->scrub->pfss_ttl = h->ip_ttl;
			h->ip_ttl = src->scrub->pfss_ttl;
		}
		break;
	}
#endif /* INET */
#ifdef INET6
	case AF_INET6: {
		if (src->scrub) {
			struct ip6_hdr *h = mtod(m, struct ip6_hdr *);
			if (h->ip6_hlim > src->scrub->pfss_ttl)
				src->scrub->pfss_ttl = h->ip6_hlim;
			h->ip6_hlim = src->scrub->pfss_ttl;
		}
		break;
	}
#endif /* INET6 */
	}

	if (th->th_off > (sizeof(struct tcphdr) >> 2) &&
	    ((src->scrub && (src->scrub->pfss_flags & PFSS_TIMESTAMP)) ||
	    (dst->scrub && (dst->scrub->pfss_flags & PFSS_TIMESTAMP))) &&
	    pf_pull_hdr(m, off, hdr, th->th_off << 2, NULL, NULL, pd->af)) {
		/* Diddle with TCP options */
		int hlen;
		opt = hdr + sizeof(struct tcphdr);
		hlen = (th->th_off << 2) - sizeof(struct tcphdr);
		while (hlen >= TCPOLEN_TIMESTAMP) {
			switch (*opt) {
			case TCPOPT_EOL:	/* FALLTHROUGH */
			case TCPOPT_NOP:
				opt++;
				hlen--;
				break;
			case TCPOPT_TIMESTAMP:
				/* Modulate the timestamps.  Can be used for
				 * NAT detection, OS uptime determination or
				 * reboot detection.
				 */

				if (got_ts) {
					/* Huh?  Multiple timestamps!? */
					if (V_pf_status.debug >= PF_DEBUG_MISC) {
						DPFPRINTF(("multiple TS??\n"));
						pf_print_state(state);
						printf("\n");
					}
					REASON_SET(reason, PFRES_TS);
					return (PF_DROP);
				}
				if (opt[1] >= TCPOLEN_TIMESTAMP) {
					memcpy(&tsval, &opt[2],
					    sizeof(u_int32_t));
					if (tsval && src->scrub &&
					    (src->scrub->pfss_flags &
					    PFSS_TIMESTAMP)) {
						tsval = ntohl(tsval);
						pf_change_proto_a(m, &opt[2],
						    &th->th_sum,
						    htonl(tsval +
						    src->scrub->pfss_ts_mod),
						    0);
						copyback = 1;
					}

					/* Modulate TS reply iff valid (!0) */
					memcpy(&tsecr, &opt[6],
					    sizeof(u_int32_t));
					if (tsecr && dst->scrub &&
					    (dst->scrub->pfss_flags &
					    PFSS_TIMESTAMP)) {
						tsecr = ntohl(tsecr)
						    - dst->scrub->pfss_ts_mod;
						pf_change_proto_a(m, &opt[6],
						    &th->th_sum, htonl(tsecr),
						    0);
						copyback = 1;
					}
					got_ts = 1;
				}
				/* FALLTHROUGH */
			default:
				hlen -= MAX(opt[1], 2);
				opt += MAX(opt[1], 2);
				break;
			}
		}
		if (copyback) {
			/* Copyback the options, caller copys back header */
			*writeback = 1;
			m_copyback(m, off + sizeof(struct tcphdr),
			    (th->th_off << 2) - sizeof(struct tcphdr), hdr +
			    sizeof(struct tcphdr));
		}
	}


	/*
	 * Must invalidate PAWS checks on connections idle for too long.
	 * The fastest allowed timestamp clock is 1ms.  That turns out to
	 * be about 24 days before it wraps.  XXX Right now our lowerbound
	 * TS echo check only works for the first 12 days of a connection
	 * when the TS has exhausted half its 32bit space
	 */
#define TS_MAX_IDLE	(24*24*60*60)
#define TS_MAX_CONN	(12*24*60*60)	/* XXX remove when better tsecr check */

	getmicrouptime(&uptime);
	if (src->scrub && (src->scrub->pfss_flags & PFSS_PAWS) &&
	    (uptime.tv_sec - src->scrub->pfss_last.tv_sec > TS_MAX_IDLE ||
	    time_uptime - state->creation > TS_MAX_CONN))  {
		if (V_pf_status.debug >= PF_DEBUG_MISC) {
			DPFPRINTF(("src idled out of PAWS\n"));
			pf_print_state(state);
			printf("\n");
		}
		src->scrub->pfss_flags = (src->scrub->pfss_flags & ~PFSS_PAWS)
		    | PFSS_PAWS_IDLED;
	}
	if (dst->scrub && (dst->scrub->pfss_flags & PFSS_PAWS) &&
	    uptime.tv_sec - dst->scrub->pfss_last.tv_sec > TS_MAX_IDLE) {
		if (V_pf_status.debug >= PF_DEBUG_MISC) {
			DPFPRINTF(("dst idled out of PAWS\n"));
			pf_print_state(state);
			printf("\n");
		}
		dst->scrub->pfss_flags = (dst->scrub->pfss_flags & ~PFSS_PAWS)
		    | PFSS_PAWS_IDLED;
	}

	if (got_ts && src->scrub && dst->scrub &&
	    (src->scrub->pfss_flags & PFSS_PAWS) &&
	    (dst->scrub->pfss_flags & PFSS_PAWS)) {
		/* Validate that the timestamps are "in-window".
		 * RFC1323 describes TCP Timestamp options that allow
		 * measurement of RTT (round trip time) and PAWS
		 * (protection against wrapped sequence numbers).  PAWS
		 * gives us a set of rules for rejecting packets on
		 * long fat pipes (packets that were somehow delayed
		 * in transit longer than the time it took to send the
		 * full TCP sequence space of 4Gb).  We can use these
		 * rules and infer a few others that will let us treat
		 * the 32bit timestamp and the 32bit echoed timestamp
		 * as sequence numbers to prevent a blind attacker from
		 * inserting packets into a connection.
		 *
		 * RFC1323 tells us:
		 *  - The timestamp on this packet must be greater than
		 *    or equal to the last value echoed by the other
		 *    endpoint.  The RFC says those will be discarded
		 *    since it is a dup that has already been acked.
		 *    This gives us a lowerbound on the timestamp.
		 *        timestamp >= other last echoed timestamp
		 *  - The timestamp will be less than or equal to
		 *    the last timestamp plus the time between the
		 *    last packet and now.  The RFC defines the max
		 *    clock rate as 1ms.  We will allow clocks to be
		 *    up to 10% fast and will allow a total difference
		 *    or 30 seconds due to a route change.  And this
		 *    gives us an upperbound on the timestamp.
		 *        timestamp <= last timestamp + max ticks
		 *    We have to be careful here.  Windows will send an
		 *    initial timestamp of zero and then initialize it
		 *    to a random value after the 3whs; presumably to
		 *    avoid a DoS by having to call an expensive RNG
		 *    during a SYN flood.  Proof MS has at least one
		 *    good security geek.
		 *
		 *  - The TCP timestamp option must also echo the other
		 *    endpoints timestamp.  The timestamp echoed is the
		 *    one carried on the earliest unacknowledged segment
		 *    on the left edge of the sequence window.  The RFC
		 *    states that the host will reject any echoed
		 *    timestamps that were larger than any ever sent.
		 *    This gives us an upperbound on the TS echo.
		 *        tescr <= largest_tsval
		 *  - The lowerbound on the TS echo is a little more
		 *    tricky to determine.  The other endpoint's echoed
		 *    values will not decrease.  But there may be
		 *    network conditions that re-order packets and
		 *    cause our view of them to decrease.  For now the
		 *    only lowerbound we can safely determine is that
		 *    the TS echo will never be less than the original
		 *    TS.  XXX There is probably a better lowerbound.
		 *    Remove TS_MAX_CONN with better lowerbound check.
		 *        tescr >= other original TS
		 *
		 * It is also important to note that the fastest
		 * timestamp clock of 1ms will wrap its 32bit space in
		 * 24 days.  So we just disable TS checking after 24
		 * days of idle time.  We actually must use a 12d
		 * connection limit until we can come up with a better
		 * lowerbound to the TS echo check.
		 */
		struct timeval delta_ts;
		int ts_fudge;


		/*
		 * PFTM_TS_DIFF is how many seconds of leeway to allow
		 * a host's timestamp.  This can happen if the previous
		 * packet got delayed in transit for much longer than
		 * this packet.
		 */
		if ((ts_fudge = state->rule.ptr->timeout[PFTM_TS_DIFF]) == 0)
			ts_fudge = V_pf_default_rule.timeout[PFTM_TS_DIFF];

		/* Calculate max ticks since the last timestamp */
#define TS_MAXFREQ	1100		/* RFC max TS freq of 1Khz + 10% skew */
#define TS_MICROSECS	1000000		/* microseconds per second */
		delta_ts = uptime;
		timevalsub(&delta_ts, &src->scrub->pfss_last);
		tsval_from_last = (delta_ts.tv_sec + ts_fudge) * TS_MAXFREQ;
		tsval_from_last += delta_ts.tv_usec / (TS_MICROSECS/TS_MAXFREQ);

		if ((src->state >= TCPS_ESTABLISHED &&
		    dst->state >= TCPS_ESTABLISHED) &&
		    (SEQ_LT(tsval, dst->scrub->pfss_tsecr) ||
		    SEQ_GT(tsval, src->scrub->pfss_tsval + tsval_from_last) ||
		    (tsecr && (SEQ_GT(tsecr, dst->scrub->pfss_tsval) ||
		    SEQ_LT(tsecr, dst->scrub->pfss_tsval0))))) {
			/* Bad RFC1323 implementation or an insertion attack.
			 *
			 * - Solaris 2.6 and 2.7 are known to send another ACK
			 *   after the FIN,FIN|ACK,ACK closing that carries
			 *   an old timestamp.
			 */

			DPFPRINTF(("Timestamp failed %c%c%c%c\n",
			    SEQ_LT(tsval, dst->scrub->pfss_tsecr) ? '0' : ' ',
			    SEQ_GT(tsval, src->scrub->pfss_tsval +
			    tsval_from_last) ? '1' : ' ',
			    SEQ_GT(tsecr, dst->scrub->pfss_tsval) ? '2' : ' ',
			    SEQ_LT(tsecr, dst->scrub->pfss_tsval0)? '3' : ' '));
			DPFPRINTF((" tsval: %u  tsecr: %u  +ticks: %u  "
			    "idle: %jus %lums\n",
			    tsval, tsecr, tsval_from_last,
			    (uintmax_t)delta_ts.tv_sec,
			    delta_ts.tv_usec / 1000));
			DPFPRINTF((" src->tsval: %u  tsecr: %u\n",
			    src->scrub->pfss_tsval, src->scrub->pfss_tsecr));
			DPFPRINTF((" dst->tsval: %u  tsecr: %u  tsval0: %u"
			    "\n", dst->scrub->pfss_tsval,
			    dst->scrub->pfss_tsecr, dst->scrub->pfss_tsval0));
			if (V_pf_status.debug >= PF_DEBUG_MISC) {
				pf_print_state(state);
				pf_print_flags(th->th_flags);
				printf("\n");
			}
			REASON_SET(reason, PFRES_TS);
			return (PF_DROP);
		}

		/* XXX I'd really like to require tsecr but it's optional */

	} else if (!got_ts && (th->th_flags & TH_RST) == 0 &&
	    ((src->state == TCPS_ESTABLISHED && dst->state == TCPS_ESTABLISHED)
	    || pd->p_len > 0 || (th->th_flags & TH_SYN)) &&
	    src->scrub && dst->scrub &&
	    (src->scrub->pfss_flags & PFSS_PAWS) &&
	    (dst->scrub->pfss_flags & PFSS_PAWS)) {
		/* Didn't send a timestamp.  Timestamps aren't really useful
		 * when:
		 *  - connection opening or closing (often not even sent).
		 *    but we must not let an attacker to put a FIN on a
		 *    data packet to sneak it through our ESTABLISHED check.
		 *  - on a TCP reset.  RFC suggests not even looking at TS.
		 *  - on an empty ACK.  The TS will not be echoed so it will
		 *    probably not help keep the RTT calculation in sync and
		 *    there isn't as much danger when the sequence numbers
		 *    got wrapped.  So some stacks don't include TS on empty
		 *    ACKs :-(
		 *
		 * To minimize the disruption to mostly RFC1323 conformant
		 * stacks, we will only require timestamps on data packets.
		 *
		 * And what do ya know, we cannot require timestamps on data
		 * packets.  There appear to be devices that do legitimate
		 * TCP connection hijacking.  There are HTTP devices that allow
		 * a 3whs (with timestamps) and then buffer the HTTP request.
		 * If the intermediate device has the HTTP response cache, it
		 * will spoof the response but not bother timestamping its
		 * packets.  So we can look for the presence of a timestamp in
		 * the first data packet and if there, require it in all future
		 * packets.
		 */

		if (pd->p_len > 0 && (src->scrub->pfss_flags & PFSS_DATA_TS)) {
			/*
			 * Hey!  Someone tried to sneak a packet in.  Or the
			 * stack changed its RFC1323 behavior?!?!
			 */
			if (V_pf_status.debug >= PF_DEBUG_MISC) {
				DPFPRINTF(("Did not receive expected RFC1323 "
				    "timestamp\n"));
				pf_print_state(state);
				pf_print_flags(th->th_flags);
				printf("\n");
			}
			REASON_SET(reason, PFRES_TS);
			return (PF_DROP);
		}
	}


	/*
	 * We will note if a host sends his data packets with or without
	 * timestamps.  And require all data packets to contain a timestamp
	 * if the first does.  PAWS implicitly requires that all data packets be
	 * timestamped.  But I think there are middle-man devices that hijack
	 * TCP streams immediately after the 3whs and don't timestamp their
	 * packets (seen in a WWW accelerator or cache).
	 */
	if (pd->p_len > 0 && src->scrub && (src->scrub->pfss_flags &
	    (PFSS_TIMESTAMP|PFSS_DATA_TS|PFSS_DATA_NOTS)) == PFSS_TIMESTAMP) {
		if (got_ts)
			src->scrub->pfss_flags |= PFSS_DATA_TS;
		else {
			src->scrub->pfss_flags |= PFSS_DATA_NOTS;
			if (V_pf_status.debug >= PF_DEBUG_MISC && dst->scrub &&
			    (dst->scrub->pfss_flags & PFSS_TIMESTAMP)) {
				/* Don't warn if other host rejected RFC1323 */
				DPFPRINTF(("Broken RFC1323 stack did not "
				    "timestamp data packet. Disabled PAWS "
				    "security.\n"));
				pf_print_state(state);
				pf_print_flags(th->th_flags);
				printf("\n");
			}
		}
	}


	/*
	 * Update PAWS values
	 */
	if (got_ts && src->scrub && PFSS_TIMESTAMP == (src->scrub->pfss_flags &
	    (PFSS_PAWS_IDLED|PFSS_TIMESTAMP))) {
		getmicrouptime(&src->scrub->pfss_last);
		if (SEQ_GEQ(tsval, src->scrub->pfss_tsval) ||
		    (src->scrub->pfss_flags & PFSS_PAWS) == 0)
			src->scrub->pfss_tsval = tsval;

		if (tsecr) {
			if (SEQ_GEQ(tsecr, src->scrub->pfss_tsecr) ||
			    (src->scrub->pfss_flags & PFSS_PAWS) == 0)
				src->scrub->pfss_tsecr = tsecr;

			if ((src->scrub->pfss_flags & PFSS_PAWS) == 0 &&
			    (SEQ_LT(tsval, src->scrub->pfss_tsval0) ||
			    src->scrub->pfss_tsval0 == 0)) {
				/* tsval0 MUST be the lowest timestamp */
				src->scrub->pfss_tsval0 = tsval;
			}

			/* Only fully initialized after a TS gets echoed */
			if ((src->scrub->pfss_flags & PFSS_PAWS) == 0)
				src->scrub->pfss_flags |= PFSS_PAWS;
		}
	}

	/* I have a dream....  TCP segment reassembly.... */
	return (0);
}

static int
pf_normalize_tcpopt(struct pf_rule *r, struct mbuf *m, struct tcphdr *th,
    int off, sa_family_t af)
{
	u_int16_t	*mss;
	int		 thoff;
	int		 opt, cnt, optlen = 0;
	int		 rewrite = 0;
	u_char		 opts[TCP_MAXOLEN];
	u_char		*optp = opts;

	thoff = th->th_off << 2;
	cnt = thoff - sizeof(struct tcphdr);

	if (cnt > 0 && !pf_pull_hdr(m, off + sizeof(*th), opts, cnt,
	    NULL, NULL, af))
		return (rewrite);

	for (; cnt > 0; cnt -= optlen, optp += optlen) {
		opt = optp[0];
		if (opt == TCPOPT_EOL)
			break;
		if (opt == TCPOPT_NOP)
			optlen = 1;
		else {
			if (cnt < 2)
				break;
			optlen = optp[1];
			if (optlen < 2 || optlen > cnt)
				break;
		}
		switch (opt) {
		case TCPOPT_MAXSEG:
			mss = (u_int16_t *)(optp + 2);
			if ((ntohs(*mss)) > r->max_mss) {
				th->th_sum = pf_proto_cksum_fixup(m,
				    th->th_sum, *mss, htons(r->max_mss), 0);
				*mss = htons(r->max_mss);
				rewrite = 1;
			}
			break;
		default:
			break;
		}
	}

	if (rewrite)
		m_copyback(m, off + sizeof(*th), thoff - sizeof(*th), opts);

	return (rewrite);
}

#ifdef INET
static void
pf_scrub_ip(struct mbuf **m0, u_int32_t flags, u_int8_t min_ttl, u_int8_t tos)
{
	struct mbuf		*m = *m0;
	struct ip		*h = mtod(m, struct ip *);

	/* Clear IP_DF if no-df was requested */
	if (flags & PFRULE_NODF && h->ip_off & htons(IP_DF)) {
		u_int16_t ip_off = h->ip_off;

		h->ip_off &= htons(~IP_DF);
		h->ip_sum = pf_cksum_fixup(h->ip_sum, ip_off, h->ip_off, 0);
	}

	/* Enforce a minimum ttl, may cause endless packet loops */
	if (min_ttl && h->ip_ttl < min_ttl) {
		u_int16_t ip_ttl = h->ip_ttl;

		h->ip_ttl = min_ttl;
		h->ip_sum = pf_cksum_fixup(h->ip_sum, ip_ttl, h->ip_ttl, 0);
	}

	/* Enforce tos */
	if (flags & PFRULE_SET_TOS) {
		u_int16_t	ov, nv;

		ov = *(u_int16_t *)h;
		h->ip_tos = tos | (h->ip_tos & IPTOS_ECN_MASK);
		nv = *(u_int16_t *)h;

		h->ip_sum = pf_cksum_fixup(h->ip_sum, ov, nv, 0);
	}

	/* random-id, but not for fragments */
	if (flags & PFRULE_RANDOMID && !(h->ip_off & ~htons(IP_DF))) {
		uint16_t ip_id = h->ip_id;

		ip_fillid(h);
		h->ip_sum = pf_cksum_fixup(h->ip_sum, ip_id, h->ip_id, 0);
	}
}
#endif /* INET */

#ifdef INET6
static void
pf_scrub_ip6(struct mbuf **m0, u_int8_t min_ttl)
{
	struct mbuf		*m = *m0;
	struct ip6_hdr		*h = mtod(m, struct ip6_hdr *);

	/* Enforce a minimum ttl, may cause endless packet loops */
	if (min_ttl && h->ip6_hlim < min_ttl)
		h->ip6_hlim = min_ttl;
}
#endif