summaryrefslogtreecommitdiffstats
path: root/sys/netinet/ip_dummynet.c
blob: a7c0a31c01075c8ec7ad7add7bcd3446d38963be (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
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
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
2151
2152
2153
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263
2264
2265
2266
2267
2268
2269
2270
2271
/*-
 * Copyright (c) 1998-2002 Luigi Rizzo, Universita` di Pisa
 * Portions Copyright (c) 2000 Akamba Corp.
 * 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 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 AUTHOR 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.
 */

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

#define	DUMMYNET_DEBUG

#include "opt_inet6.h"

/*
 * This module implements IP dummynet, a bandwidth limiter/delay emulator
 * used in conjunction with the ipfw package.
 * Description of the data structures used is in ip_dummynet.h
 * Here you mainly find the following blocks of code:
 *  + variable declarations;
 *  + heap management functions;
 *  + scheduler and dummynet functions;
 *  + configuration and initialization.
 *
 * NOTA BENE: critical sections are protected by the "dummynet lock".
 *
 * Most important Changes:
 *
 * 011004: KLDable
 * 010124: Fixed WF2Q behaviour
 * 010122: Fixed spl protection.
 * 000601: WF2Q support
 * 000106: large rewrite, use heaps to handle very many pipes.
 * 980513:	initial release
 *
 * include files marked with XXX are probably not needed
 */

#include <sys/limits.h>
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/malloc.h>
#include <sys/mbuf.h>
#include <sys/kernel.h>
#include <sys/module.h>
#include <sys/priv.h>
#include <sys/proc.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/time.h>
#include <sys/sysctl.h>
#include <sys/taskqueue.h>
#include <net/if.h>
#include <net/netisr.h>
#include <net/route.h>
#include <netinet/in.h>
#include <netinet/in_systm.h>
#include <netinet/in_var.h>
#include <netinet/ip.h>
#include <netinet/ip_fw.h>
#include <netinet/ip_dummynet.h>
#include <netinet/ip_var.h>

#include <netinet/if_ether.h> /* for struct arpcom */

#include <netinet/ip6.h>       /* for ip6_input, ip6_output prototypes */
#include <netinet6/ip6_var.h>

/*
 * We keep a private variable for the simulation time, but we could
 * probably use an existing one ("softticks" in sys/kern/kern_timeout.c)
 */
static dn_key curr_time = 0 ; /* current simulation time */

static int dn_hash_size = 64 ;	/* default hash size */

/* statistics on number of queue searches and search steps */
static long searches, search_steps ;
static int pipe_expire = 1 ;   /* expire queue if empty */
static int dn_max_ratio = 16 ; /* max queues/buckets ratio */

static long pipe_slot_limit = 100; /* Foot shooting limit for pipe queues. */
static long pipe_byte_limit = 1024 * 1024;

static int red_lookup_depth = 256;	/* RED - default lookup table depth */
static int red_avg_pkt_size = 512;      /* RED - default medium packet size */
static int red_max_pkt_size = 1500;     /* RED - default max packet size */

static struct timeval prev_t, t;
static long tick_last;			/* Last tick duration (usec). */
static long tick_delta;			/* Last vs standard tick diff (usec). */
static long tick_delta_sum;		/* Accumulated tick difference (usec).*/
static long tick_adjustment;		/* Tick adjustments done. */
static long tick_lost;			/* Lost(coalesced) ticks number. */
/* Adjusted vs non-adjusted curr_time difference (ticks). */
static long tick_diff;

static int		io_fast;
static unsigned long	io_pkt;
static unsigned long	io_pkt_fast;
static unsigned long	io_pkt_drop;

/*
 * Three heaps contain queues and pipes that the scheduler handles:
 *
 * ready_heap contains all dn_flow_queue related to fixed-rate pipes.
 *
 * wfq_ready_heap contains the pipes associated with WF2Q flows
 *
 * extract_heap contains pipes associated with delay lines.
 *
 */

MALLOC_DEFINE(M_DUMMYNET, "dummynet", "dummynet heap");

static struct dn_heap ready_heap, extract_heap, wfq_ready_heap ;

static int	heap_init(struct dn_heap *h, int size);
static int	heap_insert (struct dn_heap *h, dn_key key1, void *p);
static void	heap_extract(struct dn_heap *h, void *obj);
static void	transmit_event(struct dn_pipe *pipe, struct mbuf **head,
		    struct mbuf **tail);
static void	ready_event(struct dn_flow_queue *q, struct mbuf **head,
		    struct mbuf **tail);
static void	ready_event_wfq(struct dn_pipe *p, struct mbuf **head,
		    struct mbuf **tail);

#define	HASHSIZE	16
#define	HASH(num)	((((num) >> 8) ^ ((num) >> 4) ^ (num)) & 0x0f)
static struct dn_pipe_head	pipehash[HASHSIZE];	/* all pipes */
static struct dn_flow_set_head	flowsethash[HASHSIZE];	/* all flowsets */

static struct callout dn_timeout;

extern	void (*bridge_dn_p)(struct mbuf *, struct ifnet *);

#ifdef SYSCTL_NODE
SYSCTL_NODE(_net_inet_ip, OID_AUTO, dummynet, CTLFLAG_RW, 0, "Dummynet");
SYSCTL_INT(_net_inet_ip_dummynet, OID_AUTO, hash_size,
    CTLFLAG_RW, &dn_hash_size, 0, "Default hash table size");
SYSCTL_LONG(_net_inet_ip_dummynet, OID_AUTO, curr_time,
    CTLFLAG_RD, &curr_time, 0, "Current tick");
SYSCTL_INT(_net_inet_ip_dummynet, OID_AUTO, ready_heap,
    CTLFLAG_RD, &ready_heap.size, 0, "Size of ready heap");
SYSCTL_INT(_net_inet_ip_dummynet, OID_AUTO, extract_heap,
    CTLFLAG_RD, &extract_heap.size, 0, "Size of extract heap");
SYSCTL_LONG(_net_inet_ip_dummynet, OID_AUTO, searches,
    CTLFLAG_RD, &searches, 0, "Number of queue searches");
SYSCTL_LONG(_net_inet_ip_dummynet, OID_AUTO, search_steps,
    CTLFLAG_RD, &search_steps, 0, "Number of queue search steps");
SYSCTL_INT(_net_inet_ip_dummynet, OID_AUTO, expire,
    CTLFLAG_RW, &pipe_expire, 0, "Expire queue if empty");
SYSCTL_INT(_net_inet_ip_dummynet, OID_AUTO, max_chain_len,
    CTLFLAG_RW, &dn_max_ratio, 0,
    "Max ratio between dynamic queues and buckets");
SYSCTL_INT(_net_inet_ip_dummynet, OID_AUTO, red_lookup_depth,
    CTLFLAG_RD, &red_lookup_depth, 0, "Depth of RED lookup table");
SYSCTL_INT(_net_inet_ip_dummynet, OID_AUTO, red_avg_pkt_size,
    CTLFLAG_RD, &red_avg_pkt_size, 0, "RED Medium packet size");
SYSCTL_INT(_net_inet_ip_dummynet, OID_AUTO, red_max_pkt_size,
    CTLFLAG_RD, &red_max_pkt_size, 0, "RED Max packet size");
SYSCTL_LONG(_net_inet_ip_dummynet, OID_AUTO, tick_delta,
    CTLFLAG_RD, &tick_delta, 0, "Last vs standard tick difference (usec).");
SYSCTL_LONG(_net_inet_ip_dummynet, OID_AUTO, tick_delta_sum,
    CTLFLAG_RD, &tick_delta_sum, 0, "Accumulated tick difference (usec).");
SYSCTL_LONG(_net_inet_ip_dummynet, OID_AUTO, tick_adjustment,
    CTLFLAG_RD, &tick_adjustment, 0, "Tick adjustments done.");
SYSCTL_LONG(_net_inet_ip_dummynet, OID_AUTO, tick_diff,
    CTLFLAG_RD, &tick_diff, 0,
    "Adjusted vs non-adjusted curr_time difference (ticks).");
SYSCTL_LONG(_net_inet_ip_dummynet, OID_AUTO, tick_lost,
    CTLFLAG_RD, &tick_lost, 0,
    "Number of ticks coalesced by dummynet taskqueue.");
SYSCTL_INT(_net_inet_ip_dummynet, OID_AUTO, io_fast,
    CTLFLAG_RW, &io_fast, 0, "Enable fast dummynet io.");
SYSCTL_ULONG(_net_inet_ip_dummynet, OID_AUTO, io_pkt,
    CTLFLAG_RD, &io_pkt, 0,
    "Number of packets passed to dummynet.");
SYSCTL_ULONG(_net_inet_ip_dummynet, OID_AUTO, io_pkt_fast,
    CTLFLAG_RD, &io_pkt_fast, 0,
    "Number of packets bypassed dummynet scheduler.");
SYSCTL_ULONG(_net_inet_ip_dummynet, OID_AUTO, io_pkt_drop,
    CTLFLAG_RD, &io_pkt_drop, 0,
    "Number of packets dropped by dummynet.");
SYSCTL_LONG(_net_inet_ip_dummynet, OID_AUTO, pipe_slot_limit,
    CTLFLAG_RW, &pipe_slot_limit, 0, "Upper limit in slots for pipe queue.");
SYSCTL_LONG(_net_inet_ip_dummynet, OID_AUTO, pipe_byte_limit,
    CTLFLAG_RW, &pipe_byte_limit, 0, "Upper limit in bytes for pipe queue.");
#endif

#ifdef DUMMYNET_DEBUG
int	dummynet_debug = 0;
#ifdef SYSCTL_NODE
SYSCTL_INT(_net_inet_ip_dummynet, OID_AUTO, debug, CTLFLAG_RW, &dummynet_debug,
	    0, "control debugging printfs");
#endif
#define	DPRINTF(X)	if (dummynet_debug) printf X
#else
#define	DPRINTF(X)
#endif

static struct task	dn_task;
static struct taskqueue	*dn_tq = NULL;
static void dummynet_task(void *, int);

static struct mtx dummynet_mtx;
#define	DUMMYNET_LOCK_INIT() \
	mtx_init(&dummynet_mtx, "dummynet", NULL, MTX_DEF)
#define	DUMMYNET_LOCK_DESTROY()	mtx_destroy(&dummynet_mtx)
#define	DUMMYNET_LOCK()		mtx_lock(&dummynet_mtx)
#define	DUMMYNET_UNLOCK()	mtx_unlock(&dummynet_mtx)
#define	DUMMYNET_LOCK_ASSERT()	mtx_assert(&dummynet_mtx, MA_OWNED)

static int	config_pipe(struct dn_pipe *p);
static int	ip_dn_ctl(struct sockopt *sopt);

static void	dummynet(void *);
static void	dummynet_flush(void);
static void	dummynet_send(struct mbuf *);
void		dummynet_drain(void);
static ip_dn_io_t dummynet_io;
static void	dn_rule_delete(void *);

/*
 * Heap management functions.
 *
 * In the heap, first node is element 0. Children of i are 2i+1 and 2i+2.
 * Some macros help finding parent/children so we can optimize them.
 *
 * heap_init() is called to expand the heap when needed.
 * Increment size in blocks of 16 entries.
 * XXX failure to allocate a new element is a pretty bad failure
 * as we basically stall a whole queue forever!!
 * Returns 1 on error, 0 on success
 */
#define HEAP_FATHER(x) ( ( (x) - 1 ) / 2 )
#define HEAP_LEFT(x) ( 2*(x) + 1 )
#define HEAP_IS_LEFT(x) ( (x) & 1 )
#define HEAP_RIGHT(x) ( 2*(x) + 2 )
#define	HEAP_SWAP(a, b, buffer) { buffer = a ; a = b ; b = buffer ; }
#define HEAP_INCREMENT	15

static int
heap_init(struct dn_heap *h, int new_size)
{
    struct dn_heap_entry *p;

    if (h->size >= new_size ) {
	printf("dummynet: %s, Bogus call, have %d want %d\n", __func__,
		h->size, new_size);
	return 0 ;
    }
    new_size = (new_size + HEAP_INCREMENT ) & ~HEAP_INCREMENT ;
    p = malloc(new_size * sizeof(*p), M_DUMMYNET, M_NOWAIT);
    if (p == NULL) {
	printf("dummynet: %s, resize %d failed\n", __func__, new_size );
	return 1 ; /* error */
    }
    if (h->size > 0) {
	bcopy(h->p, p, h->size * sizeof(*p) );
	free(h->p, M_DUMMYNET);
    }
    h->p = p ;
    h->size = new_size ;
    return 0 ;
}

/*
 * Insert element in heap. Normally, p != NULL, we insert p in
 * a new position and bubble up. If p == NULL, then the element is
 * already in place, and key is the position where to start the
 * bubble-up.
 * Returns 1 on failure (cannot allocate new heap entry)
 *
 * If offset > 0 the position (index, int) of the element in the heap is
 * also stored in the element itself at the given offset in bytes.
 */
#define SET_OFFSET(heap, node) \
    if (heap->offset > 0) \
	    *((int *)((char *)(heap->p[node].object) + heap->offset)) = node ;
/*
 * RESET_OFFSET is used for sanity checks. It sets offset to an invalid value.
 */
#define RESET_OFFSET(heap, node) \
    if (heap->offset > 0) \
	    *((int *)((char *)(heap->p[node].object) + heap->offset)) = -1 ;
static int
heap_insert(struct dn_heap *h, dn_key key1, void *p)
{
    int son = h->elements ;

    if (p == NULL)	/* data already there, set starting point */
	son = key1 ;
    else {		/* insert new element at the end, possibly resize */
	son = h->elements ;
	if (son == h->size) /* need resize... */
	    if (heap_init(h, h->elements+1) )
		return 1 ; /* failure... */
	h->p[son].object = p ;
	h->p[son].key = key1 ;
	h->elements++ ;
    }
    while (son > 0) {				/* bubble up */
	int father = HEAP_FATHER(son) ;
	struct dn_heap_entry tmp  ;

	if (DN_KEY_LT( h->p[father].key, h->p[son].key ) )
	    break ; /* found right position */
	/* son smaller than father, swap and repeat */
	HEAP_SWAP(h->p[son], h->p[father], tmp) ;
	SET_OFFSET(h, son);
	son = father ;
    }
    SET_OFFSET(h, son);
    return 0 ;
}

/*
 * remove top element from heap, or obj if obj != NULL
 */
static void
heap_extract(struct dn_heap *h, void *obj)
{
    int child, father, max = h->elements - 1 ;

    if (max < 0) {
	printf("dummynet: warning, extract from empty heap 0x%p\n", h);
	return ;
    }
    father = 0 ; /* default: move up smallest child */
    if (obj != NULL) { /* extract specific element, index is at offset */
	if (h->offset <= 0)
	    panic("dummynet: heap_extract from middle not supported on this heap!!!\n");
	father = *((int *)((char *)obj + h->offset)) ;
	if (father < 0 || father >= h->elements) {
	    printf("dummynet: heap_extract, father %d out of bound 0..%d\n",
		father, h->elements);
	    panic("dummynet: heap_extract");
	}
    }
    RESET_OFFSET(h, father);
    child = HEAP_LEFT(father) ;		/* left child */
    while (child <= max) {		/* valid entry */
	if (child != max && DN_KEY_LT(h->p[child+1].key, h->p[child].key) )
	    child = child+1 ;		/* take right child, otherwise left */
	h->p[father] = h->p[child] ;
	SET_OFFSET(h, father);
	father = child ;
	child = HEAP_LEFT(child) ;   /* left child for next loop */
    }
    h->elements-- ;
    if (father != max) {
	/*
	 * Fill hole with last entry and bubble up, reusing the insert code
	 */
	h->p[father] = h->p[max] ;
	heap_insert(h, father, NULL); /* this one cannot fail */
    }
}

#if 0
/*
 * change object position and update references
 * XXX this one is never used!
 */
static void
heap_move(struct dn_heap *h, dn_key new_key, void *object)
{
    int temp;
    int i ;
    int max = h->elements-1 ;
    struct dn_heap_entry buf ;

    if (h->offset <= 0)
	panic("cannot move items on this heap");

    i = *((int *)((char *)object + h->offset));
    if (DN_KEY_LT(new_key, h->p[i].key) ) { /* must move up */
	h->p[i].key = new_key ;
	for (; i>0 && DN_KEY_LT(new_key, h->p[(temp = HEAP_FATHER(i))].key) ;
		 i = temp ) { /* bubble up */
	    HEAP_SWAP(h->p[i], h->p[temp], buf) ;
	    SET_OFFSET(h, i);
	}
    } else {		/* must move down */
	h->p[i].key = new_key ;
	while ( (temp = HEAP_LEFT(i)) <= max ) { /* found left child */
	    if ((temp != max) && DN_KEY_GT(h->p[temp].key, h->p[temp+1].key))
		temp++ ; /* select child with min key */
	    if (DN_KEY_GT(new_key, h->p[temp].key)) { /* go down */
		HEAP_SWAP(h->p[i], h->p[temp], buf) ;
		SET_OFFSET(h, i);
	    } else
		break ;
	    i = temp ;
	}
    }
    SET_OFFSET(h, i);
}
#endif /* heap_move, unused */

/*
 * heapify() will reorganize data inside an array to maintain the
 * heap property. It is needed when we delete a bunch of entries.
 */
static void
heapify(struct dn_heap *h)
{
    int i ;

    for (i = 0 ; i < h->elements ; i++ )
	heap_insert(h, i , NULL) ;
}

/*
 * cleanup the heap and free data structure
 */
static void
heap_free(struct dn_heap *h)
{
    if (h->size >0 )
	free(h->p, M_DUMMYNET);
    bzero(h, sizeof(*h) );
}

/*
 * --- end of heap management functions ---
 */

/*
 * Return the mbuf tag holding the dummynet state.  As an optimization
 * this is assumed to be the first tag on the list.  If this turns out
 * wrong we'll need to search the list.
 */
static struct dn_pkt_tag *
dn_tag_get(struct mbuf *m)
{
    struct m_tag *mtag = m_tag_first(m);
    KASSERT(mtag != NULL &&
	    mtag->m_tag_cookie == MTAG_ABI_COMPAT &&
	    mtag->m_tag_id == PACKET_TAG_DUMMYNET,
	    ("packet on dummynet queue w/o dummynet tag!"));
    return (struct dn_pkt_tag *)(mtag+1);
}

/*
 * Scheduler functions:
 *
 * transmit_event() is called when the delay-line needs to enter
 * the scheduler, either because of existing pkts getting ready,
 * or new packets entering the queue. The event handled is the delivery
 * time of the packet.
 *
 * ready_event() does something similar with fixed-rate queues, and the
 * event handled is the finish time of the head pkt.
 *
 * wfq_ready_event() does something similar with WF2Q queues, and the
 * event handled is the start time of the head pkt.
 *
 * In all cases, we make sure that the data structures are consistent
 * before passing pkts out, because this might trigger recursive
 * invocations of the procedures.
 */
static void
transmit_event(struct dn_pipe *pipe, struct mbuf **head, struct mbuf **tail)
{
	struct mbuf *m;
	struct dn_pkt_tag *pkt;

	DUMMYNET_LOCK_ASSERT();

	while ((m = pipe->head) != NULL) {
		pkt = dn_tag_get(m);
		if (!DN_KEY_LEQ(pkt->output_time, curr_time))
			break;

		pipe->head = m->m_nextpkt;
		if (*tail != NULL)
			(*tail)->m_nextpkt = m;
		else
			*head = m;
		*tail = m;
	}
	if (*tail != NULL)
		(*tail)->m_nextpkt = NULL;

	/* If there are leftover packets, put into the heap for next event. */
	if ((m = pipe->head) != NULL) {
		pkt = dn_tag_get(m);
		/*
		 * XXX Should check errors on heap_insert, by draining the
		 * whole pipe p and hoping in the future we are more successful.
		 */
		heap_insert(&extract_heap, pkt->output_time, pipe);
	}
}

/*
 * the following macro computes how many ticks we have to wait
 * before being able to transmit a packet. The credit is taken from
 * either a pipe (WF2Q) or a flow_queue (per-flow queueing)
 */
#define SET_TICKS(_m, q, p)	\
    ((_m)->m_pkthdr.len * 8 * hz - (q)->numbytes + p->bandwidth - 1) / \
    p->bandwidth;

/*
 * extract pkt from queue, compute output time (could be now)
 * and put into delay line (p_queue)
 */
static void
move_pkt(struct mbuf *pkt, struct dn_flow_queue *q, struct dn_pipe *p,
    int len)
{
    struct dn_pkt_tag *dt = dn_tag_get(pkt);

    q->head = pkt->m_nextpkt ;
    q->len-- ;
    q->len_bytes -= len ;

    dt->output_time = curr_time + p->delay ;

    if (p->head == NULL)
	p->head = pkt;
    else
	p->tail->m_nextpkt = pkt;
    p->tail = pkt;
    p->tail->m_nextpkt = NULL;
}

/*
 * ready_event() is invoked every time the queue must enter the
 * scheduler, either because the first packet arrives, or because
 * a previously scheduled event fired.
 * On invokation, drain as many pkts as possible (could be 0) and then
 * if there are leftover packets reinsert the pkt in the scheduler.
 */
static void
ready_event(struct dn_flow_queue *q, struct mbuf **head, struct mbuf **tail)
{
	struct mbuf *pkt;
	struct dn_pipe *p = q->fs->pipe;
	int p_was_empty;

	DUMMYNET_LOCK_ASSERT();

	if (p == NULL) {
		printf("dummynet: ready_event- pipe is gone\n");
		return;
	}
	p_was_empty = (p->head == NULL);

	/*
	 * Schedule fixed-rate queues linked to this pipe:
	 * account for the bw accumulated since last scheduling, then
	 * drain as many pkts as allowed by q->numbytes and move to
	 * the delay line (in p) computing output time.
	 * bandwidth==0 (no limit) means we can drain the whole queue,
	 * setting len_scaled = 0 does the job.
	 */
	q->numbytes += (curr_time - q->sched_time) * p->bandwidth;
	while ((pkt = q->head) != NULL) {
		int len = pkt->m_pkthdr.len;
		int len_scaled = p->bandwidth ? len * 8 * hz : 0;

		if (len_scaled > q->numbytes)
			break;
		q->numbytes -= len_scaled;
		move_pkt(pkt, q, p, len);
	}
	/*
	 * If we have more packets queued, schedule next ready event
	 * (can only occur when bandwidth != 0, otherwise we would have
	 * flushed the whole queue in the previous loop).
	 * To this purpose we record the current time and compute how many
	 * ticks to go for the finish time of the packet.
	 */
	if ((pkt = q->head) != NULL) {	/* this implies bandwidth != 0 */
		dn_key t = SET_TICKS(pkt, q, p); /* ticks i have to wait */

		q->sched_time = curr_time;
		heap_insert(&ready_heap, curr_time + t, (void *)q);
		/*
		 * XXX Should check errors on heap_insert, and drain the whole
		 * queue on error hoping next time we are luckier.
		 */
	} else		/* RED needs to know when the queue becomes empty. */
		q->q_time = curr_time;

	/*
	 * If the delay line was empty call transmit_event() now.
	 * Otherwise, the scheduler will take care of it.
	 */
	if (p_was_empty)
		transmit_event(p, head, tail);
}

/*
 * Called when we can transmit packets on WF2Q queues. Take pkts out of
 * the queues at their start time, and enqueue into the delay line.
 * Packets are drained until p->numbytes < 0. As long as
 * len_scaled >= p->numbytes, the packet goes into the delay line
 * with a deadline p->delay. For the last packet, if p->numbytes < 0,
 * there is an additional delay.
 */
static void
ready_event_wfq(struct dn_pipe *p, struct mbuf **head, struct mbuf **tail)
{
	int p_was_empty = (p->head == NULL);
	struct dn_heap *sch = &(p->scheduler_heap);
	struct dn_heap *neh = &(p->not_eligible_heap);
	int64_t p_numbytes = p->numbytes;

	DUMMYNET_LOCK_ASSERT();

	if (p->if_name[0] == 0)		/* tx clock is simulated */
		/*
		 * Since result may not fit into p->numbytes (32bit) we
		 * are using 64bit var here.
		 */
		p_numbytes += (curr_time - p->sched_time) * p->bandwidth;
	else {	/*
		 * tx clock is for real,
		 * the ifq must be empty or this is a NOP.
		 */
		if (p->ifp && p->ifp->if_snd.ifq_head != NULL)
			return;
		else {
			DPRINTF(("dummynet: pipe %d ready from %s --\n",
			    p->pipe_nr, p->if_name));
		}
	}

	/*
	 * While we have backlogged traffic AND credit, we need to do
	 * something on the queue.
	 */
	while (p_numbytes >= 0 && (sch->elements > 0 || neh->elements > 0)) {
		if (sch->elements > 0) {
			/* Have some eligible pkts to send out. */
			struct dn_flow_queue *q = sch->p[0].object;
			struct mbuf *pkt = q->head;
			struct dn_flow_set *fs = q->fs;
			uint64_t len = pkt->m_pkthdr.len;
			int len_scaled = p->bandwidth ? len * 8 * hz : 0;

			heap_extract(sch, NULL); /* Remove queue from heap. */
			p_numbytes -= len_scaled;
			move_pkt(pkt, q, p, len);

			p->V += (len << MY_M) / p->sum;	/* Update V. */
			q->S = q->F;			/* Update start time. */
			if (q->len == 0) {
				/* Flow not backlogged any more. */
				fs->backlogged--;
				heap_insert(&(p->idle_heap), q->F, q);
			} else {
				/* Still backlogged. */

				/*
				 * Update F and position in backlogged queue,
				 * then put flow in not_eligible_heap
				 * (we will fix this later).
				 */
				len = (q->head)->m_pkthdr.len;
				q->F += (len << MY_M) / (uint64_t)fs->weight;
				if (DN_KEY_LEQ(q->S, p->V))
					heap_insert(neh, q->S, q);
				else
					heap_insert(sch, q->F, q);
			}
		}
		/*
		 * Now compute V = max(V, min(S_i)). Remember that all elements
		 * in sch have by definition S_i <= V so if sch is not empty,
		 * V is surely the max and we must not update it. Conversely,
		 * if sch is empty we only need to look at neh.
		 */
		if (sch->elements == 0 && neh->elements > 0)
			p->V = MAX64(p->V, neh->p[0].key);
		/* Move from neh to sch any packets that have become eligible */
		while (neh->elements > 0 && DN_KEY_LEQ(neh->p[0].key, p->V)) {
			struct dn_flow_queue *q = neh->p[0].object;
			heap_extract(neh, NULL);
			heap_insert(sch, q->F, q);
		}

		if (p->if_name[0] != '\0') { /* Tx clock is from a real thing */
			p_numbytes = -1;	/* Mark not ready for I/O. */
			break;
		}
	}
	if (sch->elements == 0 && neh->elements == 0 && p_numbytes >= 0 &&
	    p->idle_heap.elements > 0) {
		/*
		 * No traffic and no events scheduled.
		 * We can get rid of idle-heap.
		 */
		int i;

		for (i = 0; i < p->idle_heap.elements; i++) {
			struct dn_flow_queue *q = p->idle_heap.p[i].object;

			q->F = 0;
			q->S = q->F + 1;
		}
		p->sum = 0;
		p->V = 0;
		p->idle_heap.elements = 0;
	}
	/*
	 * If we are getting clocks from dummynet (not a real interface) and
	 * If we are under credit, schedule the next ready event.
	 * Also fix the delivery time of the last packet.
	 */
	if (p->if_name[0]==0 && p_numbytes < 0) { /* This implies bw > 0. */
		dn_key t = 0;		/* Number of ticks i have to wait. */

		if (p->bandwidth > 0)
			t = (p->bandwidth - 1 - p_numbytes) / p->bandwidth;
		dn_tag_get(p->tail)->output_time += t;
		p->sched_time = curr_time;
		heap_insert(&wfq_ready_heap, curr_time + t, (void *)p);
		/*
		 * XXX Should check errors on heap_insert, and drain the whole
		 * queue on error hoping next time we are luckier.
		 */
	}

	/* Fit (adjust if necessary) 64bit result into 32bit variable. */
	if (p_numbytes > INT_MAX)
		p->numbytes = INT_MAX;
	else if (p_numbytes < INT_MIN)
		p->numbytes = INT_MIN;
	else
		p->numbytes = p_numbytes;

	/*
	 * If the delay line was empty call transmit_event() now.
	 * Otherwise, the scheduler will take care of it.
	 */
	if (p_was_empty)
		transmit_event(p, head, tail);
}

/*
 * This is called one tick, after previous run. It is used to
 * schedule next run.
 */
static void
dummynet(void * __unused unused)
{

	taskqueue_enqueue(dn_tq, &dn_task);
}

/*
 * The main dummynet processing function.
 */
static void
dummynet_task(void *context, int pending)
{
	struct mbuf *head = NULL, *tail = NULL;
	struct dn_pipe *pipe;
	struct dn_heap *heaps[3];
	struct dn_heap *h;
	void *p;	/* generic parameter to handler */
	int i;

	DUMMYNET_LOCK();

	heaps[0] = &ready_heap;			/* fixed-rate queues */
	heaps[1] = &wfq_ready_heap;		/* wfq queues */
	heaps[2] = &extract_heap;		/* delay line */

 	/* Update number of lost(coalesced) ticks. */
 	tick_lost += pending - 1;
 
 	getmicrouptime(&t);
 	/* Last tick duration (usec). */
 	tick_last = (t.tv_sec - prev_t.tv_sec) * 1000000 +
 	    (t.tv_usec - prev_t.tv_usec);
 	/* Last tick vs standard tick difference (usec). */
 	tick_delta = (tick_last * hz - 1000000) / hz;
 	/* Accumulated tick difference (usec). */
 	tick_delta_sum += tick_delta;
 
 	prev_t = t;
 
 	/*
 	 * Adjust curr_time if accumulated tick difference greater than
 	 * 'standard' tick. Since curr_time should be monotonically increasing,
 	 * we do positive adjustment as required and throttle curr_time in
 	 * case of negative adjustment.
 	 */
  	curr_time++;
 	if (tick_delta_sum - tick >= 0) {
 		int diff = tick_delta_sum / tick;
 
 		curr_time += diff;
 		tick_diff += diff;
 		tick_delta_sum %= tick;
 		tick_adjustment++;
 	} else if (tick_delta_sum + tick <= 0) {
 		curr_time--;
 		tick_diff--;
 		tick_delta_sum += tick;
 		tick_adjustment++;
 	}

	for (i = 0; i < 3; i++) {
		h = heaps[i];
		while (h->elements > 0 && DN_KEY_LEQ(h->p[0].key, curr_time)) {
			if (h->p[0].key > curr_time)
				printf("dummynet: warning, "
				    "heap %d is %d ticks late\n",
				    i, (int)(curr_time - h->p[0].key));
			/* store a copy before heap_extract */
			p = h->p[0].object;
			/* need to extract before processing */
			heap_extract(h, NULL);
			if (i == 0)
				ready_event(p, &head, &tail);
			else if (i == 1) {
				struct dn_pipe *pipe = p;
				if (pipe->if_name[0] != '\0')
					printf("dummynet: bad ready_event_wfq "
					    "for pipe %s\n", pipe->if_name);
				else
					ready_event_wfq(p, &head, &tail);
			} else
				transmit_event(p, &head, &tail);
		}
	}

	/* Sweep pipes trying to expire idle flow_queues. */
	for (i = 0; i < HASHSIZE; i++)
		SLIST_FOREACH(pipe, &pipehash[i], next)
			if (pipe->idle_heap.elements > 0 &&
			    DN_KEY_LT(pipe->idle_heap.p[0].key, pipe->V)) {
				struct dn_flow_queue *q =
				    pipe->idle_heap.p[0].object;

				heap_extract(&(pipe->idle_heap), NULL);
				/* Mark timestamp as invalid. */
				q->S = q->F + 1;
				pipe->sum -= q->fs->weight;
			}

	DUMMYNET_UNLOCK();

	if (head != NULL)
		dummynet_send(head);

	callout_reset(&dn_timeout, 1, dummynet, NULL);
}

static void
dummynet_send(struct mbuf *m)
{
	struct dn_pkt_tag *pkt;
	struct mbuf *n;
	struct ip *ip;

	for (; m != NULL; m = n) {
		n = m->m_nextpkt;
		m->m_nextpkt = NULL;
		pkt = dn_tag_get(m);
		switch (pkt->dn_dir) {
		case DN_TO_IP_OUT:
			ip_output(m, NULL, NULL, IP_FORWARDING, NULL, NULL);
			break ;
		  case DN_TO_IP_IN :
			ip = mtod(m, struct ip *);
			ip->ip_len = htons(ip->ip_len);
			ip->ip_off = htons(ip->ip_off);
			netisr_dispatch(NETISR_IP, m);
			break;
#ifdef INET6
		case DN_TO_IP6_IN:
			netisr_dispatch(NETISR_IPV6, m);
			break;

		case DN_TO_IP6_OUT:
			ip6_output(m, NULL, NULL, IPV6_FORWARDING, NULL, NULL, NULL);
			break;
#endif
		case DN_TO_IFB_FWD:
			if (bridge_dn_p != NULL)
				((*bridge_dn_p)(m, pkt->ifp));
			else
				printf("dummynet: if_bridge not loaded\n");

			break;
		case DN_TO_ETH_DEMUX:
			/*
			 * The Ethernet code assumes the Ethernet header is
			 * contiguous in the first mbuf header.
			 * Insure this is true.
			 */
			if (m->m_len < ETHER_HDR_LEN &&
			    (m = m_pullup(m, ETHER_HDR_LEN)) == NULL) {
				printf("dummynet/ether: pullup failed, "
				    "dropping packet\n");
				break;
			}
			ether_demux(m->m_pkthdr.rcvif, m);
			break;
		case DN_TO_ETH_OUT:
			ether_output_frame(pkt->ifp, m);
			break;
		default:
			printf("dummynet: bad switch %d!\n", pkt->dn_dir);
			m_freem(m);
			break;
		}
	}
}

/*
 * Unconditionally expire empty queues in case of shortage.
 * Returns the number of queues freed.
 */
static int
expire_queues(struct dn_flow_set *fs)
{
    struct dn_flow_queue *q, *prev ;
    int i, initial_elements = fs->rq_elements ;

    if (fs->last_expired == time_uptime)
	return 0 ;
    fs->last_expired = time_uptime ;
    for (i = 0 ; i <= fs->rq_size ; i++) /* last one is overflow */
	for (prev=NULL, q = fs->rq[i] ; q != NULL ; )
	    if (q->head != NULL || q->S != q->F+1) {
  		prev = q ;
  	        q = q->next ;
  	    } else { /* entry is idle, expire it */
		struct dn_flow_queue *old_q = q ;

		if (prev != NULL)
		    prev->next = q = q->next ;
		else
		    fs->rq[i] = q = q->next ;
		fs->rq_elements-- ;
		free(old_q, M_DUMMYNET);
	    }
    return initial_elements - fs->rq_elements ;
}

/*
 * If room, create a new queue and put at head of slot i;
 * otherwise, create or use the default queue.
 */
static struct dn_flow_queue *
create_queue(struct dn_flow_set *fs, int i)
{
	struct dn_flow_queue *q;

	if (fs->rq_elements > fs->rq_size * dn_max_ratio &&
	    expire_queues(fs) == 0) {
		/* No way to get room, use or create overflow queue. */
		i = fs->rq_size;
		if (fs->rq[i] != NULL)
		    return fs->rq[i];
	}
	q = malloc(sizeof(*q), M_DUMMYNET, M_NOWAIT | M_ZERO);
	if (q == NULL) {
		printf("dummynet: sorry, cannot allocate queue for new flow\n");
		return (NULL);
	}
	q->fs = fs;
	q->hash_slot = i;
	q->next = fs->rq[i];
	q->S = q->F + 1;	/* hack - mark timestamp as invalid. */
	q->numbytes = io_fast ? fs->pipe->bandwidth : 0;
	fs->rq[i] = q;
	fs->rq_elements++;
	return (q);
}

/*
 * Given a flow_set and a pkt in last_pkt, find a matching queue
 * after appropriate masking. The queue is moved to front
 * so that further searches take less time.
 */
static struct dn_flow_queue *
find_queue(struct dn_flow_set *fs, struct ipfw_flow_id *id)
{
    int i = 0 ; /* we need i and q for new allocations */
    struct dn_flow_queue *q, *prev;
    int is_v6 = IS_IP6_FLOW_ID(id);

    if ( !(fs->flags_fs & DN_HAVE_FLOW_MASK) )
	q = fs->rq[0] ;
    else {
	/* first, do the masking, then hash */
	id->dst_port &= fs->flow_mask.dst_port ;
	id->src_port &= fs->flow_mask.src_port ;
	id->proto &= fs->flow_mask.proto ;
	id->flags = 0 ; /* we don't care about this one */
	if (is_v6) {
	    APPLY_MASK(&id->dst_ip6, &fs->flow_mask.dst_ip6);
	    APPLY_MASK(&id->src_ip6, &fs->flow_mask.src_ip6);
	    id->flow_id6 &= fs->flow_mask.flow_id6;

	    i = ((id->dst_ip6.__u6_addr.__u6_addr32[0]) & 0xffff)^
		((id->dst_ip6.__u6_addr.__u6_addr32[1]) & 0xffff)^
		((id->dst_ip6.__u6_addr.__u6_addr32[2]) & 0xffff)^
		((id->dst_ip6.__u6_addr.__u6_addr32[3]) & 0xffff)^

		((id->dst_ip6.__u6_addr.__u6_addr32[0] >> 15) & 0xffff)^
		((id->dst_ip6.__u6_addr.__u6_addr32[1] >> 15) & 0xffff)^
		((id->dst_ip6.__u6_addr.__u6_addr32[2] >> 15) & 0xffff)^
		((id->dst_ip6.__u6_addr.__u6_addr32[3] >> 15) & 0xffff)^

		((id->src_ip6.__u6_addr.__u6_addr32[0] << 1) & 0xfffff)^
		((id->src_ip6.__u6_addr.__u6_addr32[1] << 1) & 0xfffff)^
		((id->src_ip6.__u6_addr.__u6_addr32[2] << 1) & 0xfffff)^
		((id->src_ip6.__u6_addr.__u6_addr32[3] << 1) & 0xfffff)^

		((id->src_ip6.__u6_addr.__u6_addr32[0] << 16) & 0xffff)^
		((id->src_ip6.__u6_addr.__u6_addr32[1] << 16) & 0xffff)^
		((id->src_ip6.__u6_addr.__u6_addr32[2] << 16) & 0xffff)^
		((id->src_ip6.__u6_addr.__u6_addr32[3] << 16) & 0xffff)^

		(id->dst_port << 1) ^ (id->src_port) ^
		(id->proto ) ^
		(id->flow_id6);
	} else {
	    id->dst_ip &= fs->flow_mask.dst_ip ;
	    id->src_ip &= fs->flow_mask.src_ip ;

	    i = ( (id->dst_ip) & 0xffff ) ^
		( (id->dst_ip >> 15) & 0xffff ) ^
		( (id->src_ip << 1) & 0xffff ) ^
		( (id->src_ip >> 16 ) & 0xffff ) ^
		(id->dst_port << 1) ^ (id->src_port) ^
		(id->proto );
	}
	i = i % fs->rq_size ;
	/* finally, scan the current list for a match */
	searches++ ;
	for (prev=NULL, q = fs->rq[i] ; q ; ) {
	    search_steps++;
	    if (is_v6 &&
		    IN6_ARE_ADDR_EQUAL(&id->dst_ip6,&q->id.dst_ip6) &&  
		    IN6_ARE_ADDR_EQUAL(&id->src_ip6,&q->id.src_ip6) &&  
		    id->dst_port == q->id.dst_port &&
		    id->src_port == q->id.src_port &&
		    id->proto == q->id.proto &&
		    id->flags == q->id.flags &&
		    id->flow_id6 == q->id.flow_id6)
		break ; /* found */

	    if (!is_v6 && id->dst_ip == q->id.dst_ip &&
		    id->src_ip == q->id.src_ip &&
		    id->dst_port == q->id.dst_port &&
		    id->src_port == q->id.src_port &&
		    id->proto == q->id.proto &&
		    id->flags == q->id.flags)
		break ; /* found */

	    /* No match. Check if we can expire the entry */
	    if (pipe_expire && q->head == NULL && q->S == q->F+1 ) {
		/* entry is idle and not in any heap, expire it */
		struct dn_flow_queue *old_q = q ;

		if (prev != NULL)
		    prev->next = q = q->next ;
		else
		    fs->rq[i] = q = q->next ;
		fs->rq_elements-- ;
		free(old_q, M_DUMMYNET);
		continue ;
	    }
	    prev = q ;
	    q = q->next ;
	}
	if (q && prev != NULL) { /* found and not in front */
	    prev->next = q->next ;
	    q->next = fs->rq[i] ;
	    fs->rq[i] = q ;
	}
    }
    if (q == NULL) { /* no match, need to allocate a new entry */
	q = create_queue(fs, i);
	if (q != NULL)
	q->id = *id ;
    }
    return q ;
}

static int
red_drops(struct dn_flow_set *fs, struct dn_flow_queue *q, int len)
{
	/*
	 * RED algorithm
	 *
	 * RED calculates the average queue size (avg) using a low-pass filter
	 * with an exponential weighted (w_q) moving average:
	 * 	avg  <-  (1-w_q) * avg + w_q * q_size
	 * where q_size is the queue length (measured in bytes or * packets).
	 *
	 * If q_size == 0, we compute the idle time for the link, and set
	 *	avg = (1 - w_q)^(idle/s)
	 * where s is the time needed for transmitting a medium-sized packet.
	 *
	 * Now, if avg < min_th the packet is enqueued.
	 * If avg > max_th the packet is dropped. Otherwise, the packet is
	 * dropped with probability P function of avg.
	 */

	int64_t p_b = 0;

	/* Queue in bytes or packets? */
	u_int q_size = (fs->flags_fs & DN_QSIZE_IS_BYTES) ?
	    q->len_bytes : q->len;

	DPRINTF(("\ndummynet: %d q: %2u ", (int)curr_time, q_size));

	/* Average queue size estimation. */
	if (q_size != 0) {
		/* Queue is not empty, avg <- avg + (q_size - avg) * w_q */
		int diff = SCALE(q_size) - q->avg;
		int64_t v = SCALE_MUL((int64_t)diff, (int64_t)fs->w_q);

		q->avg += (int)v;
	} else {
		/*
		 * Queue is empty, find for how long the queue has been
		 * empty and use a lookup table for computing
		 * (1 - * w_q)^(idle_time/s) where s is the time to send a
		 * (small) packet.
		 * XXX check wraps...
		 */
		if (q->avg) {
			u_int t = (curr_time - q->q_time) / fs->lookup_step;

			q->avg = (t < fs->lookup_depth) ?
			    SCALE_MUL(q->avg, fs->w_q_lookup[t]) : 0;
		}
	}
	DPRINTF(("dummynet: avg: %u ", SCALE_VAL(q->avg)));

	/* Should i drop? */
	if (q->avg < fs->min_th) {
		q->count = -1;
		return (0);	/* accept packet */
	}
	if (q->avg >= fs->max_th) {	/* average queue >=  max threshold */
		if (fs->flags_fs & DN_IS_GENTLE_RED) {
			/*
			 * According to Gentle-RED, if avg is greater than
			 * max_th the packet is dropped with a probability
			 *	 p_b = c_3 * avg - c_4
			 * where c_3 = (1 - max_p) / max_th
			 *       c_4 = 1 - 2 * max_p
			 */
			p_b = SCALE_MUL((int64_t)fs->c_3, (int64_t)q->avg) -
			    fs->c_4;
		} else {
			q->count = -1;
			DPRINTF(("dummynet: - drop"));
			return (1);
		}
	} else if (q->avg > fs->min_th) {
		/*
		 * We compute p_b using the linear dropping function
		 *	 p_b = c_1 * avg - c_2
		 * where c_1 = max_p / (max_th - min_th)
		 * 	 c_2 = max_p * min_th / (max_th - min_th)
		 */
		p_b = SCALE_MUL((int64_t)fs->c_1, (int64_t)q->avg) - fs->c_2;
	}

	if (fs->flags_fs & DN_QSIZE_IS_BYTES)
		p_b = (p_b * len) / fs->max_pkt_size;
	if (++q->count == 0)
		q->random = random() & 0xffff;
	else {
		/*
		 * q->count counts packets arrived since last drop, so a greater
		 * value of q->count means a greater packet drop probability.
		 */
		if (SCALE_MUL(p_b, SCALE((int64_t)q->count)) > q->random) {
			q->count = 0;
			DPRINTF(("dummynet: - red drop"));
			/* After a drop we calculate a new random value. */
			q->random = random() & 0xffff;
			return (1);	/* drop */
		}
	}
	/* End of RED algorithm. */

	return (0);	/* accept */
}

static __inline struct dn_flow_set *
locate_flowset(int fs_nr)
{
	struct dn_flow_set *fs;

	SLIST_FOREACH(fs, &flowsethash[HASH(fs_nr)], next)
		if (fs->fs_nr == fs_nr)
			return (fs);

	return (NULL);
}

static __inline struct dn_pipe *
locate_pipe(int pipe_nr)
{
	struct dn_pipe *pipe;

	SLIST_FOREACH(pipe, &pipehash[HASH(pipe_nr)], next)
		if (pipe->pipe_nr == pipe_nr)
			return (pipe);

	return (NULL);
}

/*
 * dummynet hook for packets. Below 'pipe' is a pipe or a queue
 * depending on whether WF2Q or fixed bw is used.
 *
 * pipe_nr	pipe or queue the packet is destined for.
 * dir		where shall we send the packet after dummynet.
 * m		the mbuf with the packet
 * ifp		the 'ifp' parameter from the caller.
 *		NULL in ip_input, destination interface in ip_output,
 * rule		matching rule, in case of multiple passes
 */
static int
dummynet_io(struct mbuf **m0, int dir, struct ip_fw_args *fwa)
{
	struct mbuf *m = *m0, *head = NULL, *tail = NULL;
	struct dn_pkt_tag *pkt;
	struct m_tag *mtag;
	struct dn_flow_set *fs = NULL;
	struct dn_pipe *pipe;
	uint64_t len = m->m_pkthdr.len;
	struct dn_flow_queue *q = NULL;
	int is_pipe;
	ipfw_insn *cmd = ACTION_PTR(fwa->rule);

	KASSERT(m->m_nextpkt == NULL,
	    ("dummynet_io: mbuf queue passed to dummynet"));

	if (cmd->opcode == O_LOG)
		cmd += F_LEN(cmd);
	if (cmd->opcode == O_ALTQ)
		cmd += F_LEN(cmd);
	if (cmd->opcode == O_TAG)
		cmd += F_LEN(cmd);
	is_pipe = (cmd->opcode == O_PIPE);

	DUMMYNET_LOCK();
	io_pkt++;
	/*
	 * This is a dummynet rule, so we expect an O_PIPE or O_QUEUE rule.
	 *
	 * XXXGL: probably the pipe->fs and fs->pipe logic here
	 * below can be simplified.
	 */
	if (is_pipe) {
		pipe = locate_pipe(fwa->cookie);
		if (pipe != NULL)
			fs = &(pipe->fs);
	} else
		fs = locate_flowset(fwa->cookie);

	if (fs == NULL)
		goto dropit;	/* This queue/pipe does not exist! */
	pipe = fs->pipe;
	if (pipe == NULL) {	/* Must be a queue, try find a matching pipe. */
		pipe = locate_pipe(fs->parent_nr);
		if (pipe != NULL)
			fs->pipe = pipe;
		else {
			printf("dummynet: no pipe %d for queue %d, drop pkt\n",
			    fs->parent_nr, fs->fs_nr);
			goto dropit;
		}
	}
	q = find_queue(fs, &(fwa->f_id));
	if (q == NULL)
		goto dropit;		/* Cannot allocate queue. */

	/* Update statistics, then check reasons to drop pkt. */
	q->tot_bytes += len;
	q->tot_pkts++;
	if (fs->plr && random() < fs->plr)
		goto dropit;		/* Random pkt drop. */
	if (fs->flags_fs & DN_QSIZE_IS_BYTES) {
		if (q->len_bytes > fs->qsize)
			goto dropit;	/* Queue size overflow. */
	} else {
		if (q->len >= fs->qsize)
			goto dropit;	/* Queue count overflow. */
	}
	if (fs->flags_fs & DN_IS_RED && red_drops(fs, q, len))
		goto dropit;

	/* XXX expensive to zero, see if we can remove it. */
	mtag = m_tag_get(PACKET_TAG_DUMMYNET,
	    sizeof(struct dn_pkt_tag), M_NOWAIT | M_ZERO);
	if (mtag == NULL)
		goto dropit;		/* Cannot allocate packet header. */
	m_tag_prepend(m, mtag);		/* Attach to mbuf chain. */

	pkt = (struct dn_pkt_tag *)(mtag + 1);
	/*
	 * Ok, i can handle the pkt now...
	 * Build and enqueue packet + parameters.
	 */
	pkt->rule = fwa->rule;
	pkt->dn_dir = dir;

	pkt->ifp = fwa->oif;

	if (q->head == NULL)
		q->head = m;
	else
		q->tail->m_nextpkt = m;
	q->tail = m;
	q->len++;
	q->len_bytes += len;

	if (q->head != m)		/* Flow was not idle, we are done. */
		goto done;

	if (q->q_time < curr_time)
		q->numbytes = io_fast ? fs->pipe->bandwidth : 0;
	q->q_time = curr_time;

	/*
	 * If we reach this point the flow was previously idle, so we need
	 * to schedule it. This involves different actions for fixed-rate or
	 * WF2Q queues.
	 */
	if (is_pipe) {
		/* Fixed-rate queue: just insert into the ready_heap. */
		dn_key t = 0;

		if (pipe->bandwidth && m->m_pkthdr.len * 8 * hz > q->numbytes)
			t = SET_TICKS(m, q, pipe);
		q->sched_time = curr_time;
		if (t == 0)		/* Must process it now. */
			ready_event(q, &head, &tail);
		else
			heap_insert(&ready_heap, curr_time + t , q);
	} else {
		/*
		 * WF2Q. First, compute start time S: if the flow was
		 * idle (S = F + 1) set S to the virtual time V for the
		 * controlling pipe, and update the sum of weights for the pipe;
		 * otherwise, remove flow from idle_heap and set S to max(F,V).
		 * Second, compute finish time F = S + len / weight.
		 * Third, if pipe was idle, update V = max(S, V).
		 * Fourth, count one more backlogged flow.
		 */
		if (DN_KEY_GT(q->S, q->F)) { /* Means timestamps are invalid. */
			q->S = pipe->V;
			pipe->sum += fs->weight; /* Add weight of new queue. */
		} else {
			heap_extract(&(pipe->idle_heap), q);
			q->S = MAX64(q->F, pipe->V);
		}
		q->F = q->S + (len << MY_M) / (uint64_t)fs->weight;

		if (pipe->not_eligible_heap.elements == 0 &&
		    pipe->scheduler_heap.elements == 0)
			pipe->V = MAX64(q->S, pipe->V);
		fs->backlogged++;
		/*
		 * Look at eligibility. A flow is not eligibile if S>V (when
		 * this happens, it means that there is some other flow already
		 * scheduled for the same pipe, so the scheduler_heap cannot be
		 * empty). If the flow is not eligible we just store it in the
		 * not_eligible_heap. Otherwise, we store in the scheduler_heap
		 * and possibly invoke ready_event_wfq() right now if there is
		 * leftover credit.
		 * Note that for all flows in scheduler_heap (SCH), S_i <= V,
		 * and for all flows in not_eligible_heap (NEH), S_i > V.
		 * So when we need to compute max(V, min(S_i)) forall i in
		 * SCH+NEH, we only need to look into NEH.
		 */
		if (DN_KEY_GT(q->S, pipe->V)) {		/* Not eligible. */
			if (pipe->scheduler_heap.elements == 0)
				printf("dummynet: ++ ouch! not eligible but empty scheduler!\n");
			heap_insert(&(pipe->not_eligible_heap), q->S, q);
		} else {
			heap_insert(&(pipe->scheduler_heap), q->F, q);
			if (pipe->numbytes >= 0) {	 /* Pipe is idle. */
				if (pipe->scheduler_heap.elements != 1)
					printf("dummynet: OUCH! pipe should have been idle!\n");
				DPRINTF(("dummynet: waking up pipe %d at %d\n",
				    pipe->pipe_nr, (int)(q->F >> MY_M)));
				pipe->sched_time = curr_time;
				ready_event_wfq(pipe, &head, &tail);
			}
		}
	}
done:
	if (head == m && dir != DN_TO_IFB_FWD && dir != DN_TO_ETH_DEMUX &&
	    dir != DN_TO_ETH_OUT) {	/* Fast io. */
		io_pkt_fast++;
		if (m->m_nextpkt != NULL)
			printf("dummynet: fast io: pkt chain detected!\n");
		head = m->m_nextpkt = NULL;
	} else
		*m0 = NULL;		/* Normal io. */

	DUMMYNET_UNLOCK();
	if (head != NULL)
		dummynet_send(head);
	return (0);

dropit:
	io_pkt_drop++;
	if (q)
		q->drops++;
	DUMMYNET_UNLOCK();
	m_freem(m);
	*m0 = NULL;
	return ((fs && (fs->flags_fs & DN_NOERROR)) ? 0 : ENOBUFS);
}

/*
 * Below, the rt_unref is only needed when (pkt->dn_dir == DN_TO_IP_OUT)
 * Doing this would probably save us the initial bzero of dn_pkt
 */
#define	DN_FREE_PKT(_m) do {				\
	m_freem(_m);					\
} while (0)

/*
 * Dispose all packets and flow_queues on a flow_set.
 * If all=1, also remove red lookup table and other storage,
 * including the descriptor itself.
 * For the one in dn_pipe MUST also cleanup ready_heap...
 */
static void
purge_flow_set(struct dn_flow_set *fs, int all)
{
	struct dn_flow_queue *q, *qn;
	int i;

	DUMMYNET_LOCK_ASSERT();

	for (i = 0; i <= fs->rq_size; i++) {
		for (q = fs->rq[i]; q != NULL; q = qn) {
			struct mbuf *m, *mnext;

			mnext = q->head;
			while ((m = mnext) != NULL) {
				mnext = m->m_nextpkt;
				DN_FREE_PKT(m);
			}
			qn = q->next;
			free(q, M_DUMMYNET);
		}
		fs->rq[i] = NULL;
	}

	fs->rq_elements = 0;
	if (all) {
		/* RED - free lookup table. */
		if (fs->w_q_lookup != NULL)
			free(fs->w_q_lookup, M_DUMMYNET);
		if (fs->rq != NULL)
			free(fs->rq, M_DUMMYNET);
		/* If this fs is not part of a pipe, free it. */
		if (fs->pipe == NULL || fs != &(fs->pipe->fs))
			free(fs, M_DUMMYNET);
	}
}

/*
 * Dispose all packets queued on a pipe (not a flow_set).
 * Also free all resources associated to a pipe, which is about
 * to be deleted.
 */
static void
purge_pipe(struct dn_pipe *pipe)
{
    struct mbuf *m, *mnext;

    purge_flow_set( &(pipe->fs), 1 );

    mnext = pipe->head;
    while ((m = mnext) != NULL) {
	mnext = m->m_nextpkt;
	DN_FREE_PKT(m);
    }

    heap_free( &(pipe->scheduler_heap) );
    heap_free( &(pipe->not_eligible_heap) );
    heap_free( &(pipe->idle_heap) );
}

/*
 * Delete all pipes and heaps returning memory. Must also
 * remove references from all ipfw rules to all pipes.
 */
static void
dummynet_flush(void)
{
	struct dn_pipe *pipe, *pipe1;
	struct dn_flow_set *fs, *fs1;
	int i;

	DUMMYNET_LOCK();
	/* Free heaps so we don't have unwanted events. */
	heap_free(&ready_heap);
	heap_free(&wfq_ready_heap);
	heap_free(&extract_heap);

	/*
	 * Now purge all queued pkts and delete all pipes.
	 *
	 * XXXGL: can we merge the for(;;) cycles into one or not?
	 */
	for (i = 0; i < HASHSIZE; i++)
		SLIST_FOREACH_SAFE(fs, &flowsethash[i], next, fs1) {
			SLIST_REMOVE(&flowsethash[i], fs, dn_flow_set, next);
			purge_flow_set(fs, 1);
		}
	for (i = 0; i < HASHSIZE; i++)
		SLIST_FOREACH_SAFE(pipe, &pipehash[i], next, pipe1) {
			SLIST_REMOVE(&pipehash[i], pipe, dn_pipe, next);
			purge_pipe(pipe);
			free(pipe, M_DUMMYNET);
		}
	DUMMYNET_UNLOCK();
}

extern struct ip_fw *ip_fw_default_rule ;
static void
dn_rule_delete_fs(struct dn_flow_set *fs, void *r)
{
    int i ;
    struct dn_flow_queue *q ;
    struct mbuf *m ;

    for (i = 0 ; i <= fs->rq_size ; i++) /* last one is ovflow */
	for (q = fs->rq[i] ; q ; q = q->next )
	    for (m = q->head ; m ; m = m->m_nextpkt ) {
		struct dn_pkt_tag *pkt = dn_tag_get(m) ;
		if (pkt->rule == r)
		    pkt->rule = ip_fw_default_rule ;
	    }
}
/*
 * when a firewall rule is deleted, scan all queues and remove the flow-id
 * from packets matching this rule.
 */
void
dn_rule_delete(void *r)
{
    struct dn_pipe *pipe;
    struct dn_flow_set *fs;
    struct dn_pkt_tag *pkt;
    struct mbuf *m;
    int i;

    DUMMYNET_LOCK();
    /*
     * If the rule references a queue (dn_flow_set), then scan
     * the flow set, otherwise scan pipes. Should do either, but doing
     * both does not harm.
     */
    for (i = 0; i < HASHSIZE; i++)
	SLIST_FOREACH(fs, &flowsethash[i], next)
		dn_rule_delete_fs(fs, r);

    for (i = 0; i < HASHSIZE; i++)
	SLIST_FOREACH(pipe, &pipehash[i], next) {
		fs = &(pipe->fs);
		dn_rule_delete_fs(fs, r);
		for (m = pipe->head ; m ; m = m->m_nextpkt ) {
			pkt = dn_tag_get(m);
			if (pkt->rule == r)
				pkt->rule = ip_fw_default_rule;
		}
	}
    DUMMYNET_UNLOCK();
}

/*
 * setup RED parameters
 */
static int
config_red(struct dn_flow_set *p, struct dn_flow_set *x)
{
	int i;

	x->w_q = p->w_q;
	x->min_th = SCALE(p->min_th);
	x->max_th = SCALE(p->max_th);
	x->max_p = p->max_p;

	x->c_1 = p->max_p / (p->max_th - p->min_th);
	x->c_2 = SCALE_MUL(x->c_1, SCALE(p->min_th));

	if (x->flags_fs & DN_IS_GENTLE_RED) {
		x->c_3 = (SCALE(1) - p->max_p) / p->max_th;
		x->c_4 = SCALE(1) - 2 * p->max_p;
	}

	/* If the lookup table already exist, free and create it again. */
	if (x->w_q_lookup) {
		free(x->w_q_lookup, M_DUMMYNET);
		x->w_q_lookup = NULL;
	}
	if (red_lookup_depth == 0) {
		printf("\ndummynet: net.inet.ip.dummynet.red_lookup_depth"
		    "must be > 0\n");
		free(x, M_DUMMYNET);
		return (EINVAL);
	}
	x->lookup_depth = red_lookup_depth;
	x->w_q_lookup = (u_int *)malloc(x->lookup_depth * sizeof(int),
	    M_DUMMYNET, M_NOWAIT);
	if (x->w_q_lookup == NULL) {
		printf("dummynet: sorry, cannot allocate red lookup table\n");
		free(x, M_DUMMYNET);
		return(ENOSPC);
	}

	/* Fill the lookup table with (1 - w_q)^x */
	x->lookup_step = p->lookup_step;
	x->lookup_weight = p->lookup_weight;
	x->w_q_lookup[0] = SCALE(1) - x->w_q;

	for (i = 1; i < x->lookup_depth; i++)
		x->w_q_lookup[i] =
		    SCALE_MUL(x->w_q_lookup[i - 1], x->lookup_weight);

	if (red_avg_pkt_size < 1)
		red_avg_pkt_size = 512;
	x->avg_pkt_size = red_avg_pkt_size;
	if (red_max_pkt_size < 1)
		red_max_pkt_size = 1500;
	x->max_pkt_size = red_max_pkt_size;
	return (0);
}

static int
alloc_hash(struct dn_flow_set *x, struct dn_flow_set *pfs)
{
    if (x->flags_fs & DN_HAVE_FLOW_MASK) {     /* allocate some slots */
	int l = pfs->rq_size;

	if (l == 0)
	    l = dn_hash_size;
	if (l < 4)
	    l = 4;
	else if (l > DN_MAX_HASH_SIZE)
	    l = DN_MAX_HASH_SIZE;
	x->rq_size = l;
    } else                  /* one is enough for null mask */
	x->rq_size = 1;
    x->rq = malloc((1 + x->rq_size) * sizeof(struct dn_flow_queue *),
	    M_DUMMYNET, M_NOWAIT | M_ZERO);
    if (x->rq == NULL) {
	printf("dummynet: sorry, cannot allocate queue\n");
	return (ENOMEM);
    }
    x->rq_elements = 0;
    return 0 ;
}

static void
set_fs_parms(struct dn_flow_set *x, struct dn_flow_set *src)
{
	x->flags_fs = src->flags_fs;
	x->qsize = src->qsize;
	x->plr = src->plr;
	x->flow_mask = src->flow_mask;
	if (x->flags_fs & DN_QSIZE_IS_BYTES) {
		if (x->qsize > pipe_byte_limit)
			x->qsize = 1024 * 1024;
	} else {
		if (x->qsize == 0)
			x->qsize = 50;
		if (x->qsize > pipe_slot_limit)
			x->qsize = 50;
	}
	/* Configuring RED. */
	if (x->flags_fs & DN_IS_RED)
		config_red(src, x);	/* XXX should check errors */
}

/*
 * Setup pipe or queue parameters.
 */
static int
config_pipe(struct dn_pipe *p)
{
	struct dn_flow_set *pfs = &(p->fs);
	struct dn_flow_queue *q;
	int i, error;

	/*
	 * The config program passes parameters as follows:
	 * bw = bits/second (0 means no limits),
	 * delay = ms, must be translated into ticks.
	 * qsize = slots/bytes
	 */
	p->delay = (p->delay * hz) / 1000;
	/* We need either a pipe number or a flow_set number. */
	if (p->pipe_nr == 0 && pfs->fs_nr == 0)
		return (EINVAL);
	if (p->pipe_nr != 0 && pfs->fs_nr != 0)
		return (EINVAL);
	if (p->pipe_nr != 0) {			/* this is a pipe */
		struct dn_pipe *pipe;

		DUMMYNET_LOCK();
		pipe = locate_pipe(p->pipe_nr);	/* locate pipe */

		if (pipe == NULL) {		/* new pipe */
			pipe = malloc(sizeof(struct dn_pipe), M_DUMMYNET,
			    M_NOWAIT | M_ZERO);
			if (pipe == NULL) {
				DUMMYNET_UNLOCK();
				printf("dummynet: no memory for new pipe\n");
				return (ENOMEM);
			}
			pipe->pipe_nr = p->pipe_nr;
			pipe->fs.pipe = pipe;
			/*
			 * idle_heap is the only one from which
			 * we extract from the middle.
			 */
			pipe->idle_heap.size = pipe->idle_heap.elements = 0;
			pipe->idle_heap.offset =
			    offsetof(struct dn_flow_queue, heap_pos);
		} else
			/* Flush accumulated credit for all queues. */
			for (i = 0; i <= pipe->fs.rq_size; i++)
				for (q = pipe->fs.rq[i]; q; q = q->next)
					q->numbytes = io_fast ? p->bandwidth : 0;

		pipe->bandwidth = p->bandwidth;
		pipe->numbytes = 0;		/* just in case... */
		bcopy(p->if_name, pipe->if_name, sizeof(p->if_name));
		pipe->ifp = NULL;		/* reset interface ptr */
		pipe->delay = p->delay;
		set_fs_parms(&(pipe->fs), pfs);

		if (pipe->fs.rq == NULL) {	/* a new pipe */
			error = alloc_hash(&(pipe->fs), pfs);
			if (error) {
				DUMMYNET_UNLOCK();
				free(pipe, M_DUMMYNET);
				return (error);
			}
			SLIST_INSERT_HEAD(&pipehash[HASH(pipe->pipe_nr)],
			    pipe, next);
		}
		DUMMYNET_UNLOCK();
	} else {				/* config queue */
		struct dn_flow_set *fs;

		DUMMYNET_LOCK();
		fs = locate_flowset(pfs->fs_nr); /* locate flow_set */

		if (fs == NULL) {		/* new */
			if (pfs->parent_nr == 0) { /* need link to a pipe */
				DUMMYNET_UNLOCK();
				return (EINVAL);
			}
			fs = malloc(sizeof(struct dn_flow_set), M_DUMMYNET,
			    M_NOWAIT | M_ZERO);
			if (fs == NULL) {
				DUMMYNET_UNLOCK();
				printf(
				    "dummynet: no memory for new flow_set\n");
				return (ENOMEM);
			}
			fs->fs_nr = pfs->fs_nr;
			fs->parent_nr = pfs->parent_nr;
			fs->weight = pfs->weight;
			if (fs->weight == 0)
				fs->weight = 1;
			else if (fs->weight > 100)
				fs->weight = 100;
		} else {
			/*
			 * Change parent pipe not allowed;
			 * must delete and recreate.
			 */
			if (pfs->parent_nr != 0 &&
			    fs->parent_nr != pfs->parent_nr) {
				DUMMYNET_UNLOCK();
				return (EINVAL);
			}
		}

		set_fs_parms(fs, pfs);

		if (fs->rq == NULL) {		/* a new flow_set */
			error = alloc_hash(fs, pfs);
			if (error) {
				DUMMYNET_UNLOCK();
				free(fs, M_DUMMYNET);
				return (error);
			}
			SLIST_INSERT_HEAD(&flowsethash[HASH(fs->fs_nr)],
			    fs, next);
		}
		DUMMYNET_UNLOCK();
	}
	return (0);
}

/*
 * Helper function to remove from a heap queues which are linked to
 * a flow_set about to be deleted.
 */
static void
fs_remove_from_heap(struct dn_heap *h, struct dn_flow_set *fs)
{
    int i = 0, found = 0 ;
    for (; i < h->elements ;)
	if ( ((struct dn_flow_queue *)h->p[i].object)->fs == fs) {
	    h->elements-- ;
	    h->p[i] = h->p[h->elements] ;
	    found++ ;
	} else
	    i++ ;
    if (found)
	heapify(h);
}

/*
 * helper function to remove a pipe from a heap (can be there at most once)
 */
static void
pipe_remove_from_heap(struct dn_heap *h, struct dn_pipe *p)
{
    if (h->elements > 0) {
	int i = 0 ;
	for (i=0; i < h->elements ; i++ ) {
	    if (h->p[i].object == p) { /* found it */
		h->elements-- ;
		h->p[i] = h->p[h->elements] ;
		heapify(h);
		break ;
	    }
	}
    }
}

/*
 * drain all queues. Called in case of severe mbuf shortage.
 */
void
dummynet_drain(void)
{
    struct dn_flow_set *fs;
    struct dn_pipe *pipe;
    struct mbuf *m, *mnext;
    int i;

    DUMMYNET_LOCK_ASSERT();

    heap_free(&ready_heap);
    heap_free(&wfq_ready_heap);
    heap_free(&extract_heap);
    /* remove all references to this pipe from flow_sets */
    for (i = 0; i < HASHSIZE; i++)
	SLIST_FOREACH(fs, &flowsethash[i], next)
		purge_flow_set(fs, 0);

    for (i = 0; i < HASHSIZE; i++) {
	SLIST_FOREACH(pipe, &pipehash[i], next) {
		purge_flow_set(&(pipe->fs), 0);

		mnext = pipe->head;
		while ((m = mnext) != NULL) {
			mnext = m->m_nextpkt;
			DN_FREE_PKT(m);
		}
		pipe->head = pipe->tail = NULL;
	}
    }
}

/*
 * Fully delete a pipe or a queue, cleaning up associated info.
 */
static int
delete_pipe(struct dn_pipe *p)
{

    if (p->pipe_nr == 0 && p->fs.fs_nr == 0)
	return EINVAL ;
    if (p->pipe_nr != 0 && p->fs.fs_nr != 0)
	return EINVAL ;
    if (p->pipe_nr != 0) { /* this is an old-style pipe */
	struct dn_pipe *pipe;
	struct dn_flow_set *fs;
	int i;

	DUMMYNET_LOCK();
	pipe = locate_pipe(p->pipe_nr);	/* locate pipe */

	if (pipe == NULL) {
	    DUMMYNET_UNLOCK();
	    return (ENOENT);	/* not found */
	}

	/* Unlink from list of pipes. */
	SLIST_REMOVE(&pipehash[HASH(pipe->pipe_nr)], pipe, dn_pipe, next);

	/* Remove all references to this pipe from flow_sets. */
	for (i = 0; i < HASHSIZE; i++)
	    SLIST_FOREACH(fs, &flowsethash[i], next)
		if (fs->pipe == pipe) {
			printf("dummynet: ++ ref to pipe %d from fs %d\n",
			    p->pipe_nr, fs->fs_nr);
			fs->pipe = NULL ;
			purge_flow_set(fs, 0);
		}
	fs_remove_from_heap(&ready_heap, &(pipe->fs));
	purge_pipe(pipe); /* remove all data associated to this pipe */
	/* remove reference to here from extract_heap and wfq_ready_heap */
	pipe_remove_from_heap(&extract_heap, pipe);
	pipe_remove_from_heap(&wfq_ready_heap, pipe);
	DUMMYNET_UNLOCK();

	free(pipe, M_DUMMYNET);
    } else { /* this is a WF2Q queue (dn_flow_set) */
	struct dn_flow_set *fs;

	DUMMYNET_LOCK();
	fs = locate_flowset(p->fs.fs_nr); /* locate set */

	if (fs == NULL) {
	    DUMMYNET_UNLOCK();
	    return (ENOENT); /* not found */
	}

	/* Unlink from list of flowsets. */
	SLIST_REMOVE( &flowsethash[HASH(fs->fs_nr)], fs, dn_flow_set, next);

	if (fs->pipe != NULL) {
	    /* Update total weight on parent pipe and cleanup parent heaps. */
	    fs->pipe->sum -= fs->weight * fs->backlogged ;
	    fs_remove_from_heap(&(fs->pipe->not_eligible_heap), fs);
	    fs_remove_from_heap(&(fs->pipe->scheduler_heap), fs);
#if 1	/* XXX should i remove from idle_heap as well ? */
	    fs_remove_from_heap(&(fs->pipe->idle_heap), fs);
#endif
	}
	purge_flow_set(fs, 1);
	DUMMYNET_UNLOCK();
    }
    return 0 ;
}

/*
 * helper function used to copy data from kernel in DUMMYNET_GET
 */
static char *
dn_copy_set(struct dn_flow_set *set, char *bp)
{
    int i, copied = 0 ;
    struct dn_flow_queue *q, *qp = (struct dn_flow_queue *)bp;

    DUMMYNET_LOCK_ASSERT();

    for (i = 0 ; i <= set->rq_size ; i++)
	for (q = set->rq[i] ; q ; q = q->next, qp++ ) {
	    if (q->hash_slot != i)
		printf("dummynet: ++ at %d: wrong slot (have %d, "
		    "should be %d)\n", copied, q->hash_slot, i);
	    if (q->fs != set)
		printf("dummynet: ++ at %d: wrong fs ptr (have %p, should be %p)\n",
			i, q->fs, set);
	    copied++ ;
	    bcopy(q, qp, sizeof( *q ) );
	    /* cleanup pointers */
	    qp->next = NULL ;
	    qp->head = qp->tail = NULL ;
	    qp->fs = NULL ;
	}
    if (copied != set->rq_elements)
	printf("dummynet: ++ wrong count, have %d should be %d\n",
	    copied, set->rq_elements);
    return (char *)qp ;
}

static size_t
dn_calc_size(void)
{
    struct dn_flow_set *fs;
    struct dn_pipe *pipe;
    size_t size = 0;
    int i;

    DUMMYNET_LOCK_ASSERT();
    /*
     * Compute size of data structures: list of pipes and flow_sets.
     */
    for (i = 0; i < HASHSIZE; i++) {
	SLIST_FOREACH(pipe, &pipehash[i], next)
		size += sizeof(*pipe) +
		    pipe->fs.rq_elements * sizeof(struct dn_flow_queue);
	SLIST_FOREACH(fs, &flowsethash[i], next)
		size += sizeof (*fs) +
		    fs->rq_elements * sizeof(struct dn_flow_queue);
    }
    return size;
}

static int
dummynet_get(struct sockopt *sopt)
{
    char *buf, *bp ; /* bp is the "copy-pointer" */
    size_t size ;
    struct dn_flow_set *fs;
    struct dn_pipe *pipe;
    int error=0, i ;

    /* XXX lock held too long */
    DUMMYNET_LOCK();
    /*
     * XXX: Ugly, but we need to allocate memory with M_WAITOK flag and we
     *      cannot use this flag while holding a mutex.
     */
    for (i = 0; i < 10; i++) {
	size = dn_calc_size();
	DUMMYNET_UNLOCK();
	buf = malloc(size, M_TEMP, M_WAITOK);
	DUMMYNET_LOCK();
	if (size == dn_calc_size())
		break;
	free(buf, M_TEMP);
	buf = NULL;
    }
    if (buf == NULL) {
	DUMMYNET_UNLOCK();
	return ENOBUFS ;
    }
    bp = buf;
    for (i = 0; i < HASHSIZE; i++)
	SLIST_FOREACH(pipe, &pipehash[i], next) {
		struct dn_pipe *pipe_bp = (struct dn_pipe *)bp;

		/*
		 * Copy pipe descriptor into *bp, convert delay back to ms,
		 * then copy the flow_set descriptor(s) one at a time.
		 * After each flow_set, copy the queue descriptor it owns.
		 */
		bcopy(pipe, bp, sizeof(*pipe));
		pipe_bp->delay = (pipe_bp->delay * 1000) / hz;
		/*
		 * XXX the following is a hack based on ->next being the
		 * first field in dn_pipe and dn_flow_set. The correct
		 * solution would be to move the dn_flow_set to the beginning
		 * of struct dn_pipe.
		 */
		pipe_bp->next.sle_next = (struct dn_pipe *)DN_IS_PIPE;
		/* Clean pointers. */
		pipe_bp->head = pipe_bp->tail = NULL;
		pipe_bp->fs.next.sle_next = NULL;
		pipe_bp->fs.pipe = NULL;
		pipe_bp->fs.rq = NULL;

		bp += sizeof(*pipe) ;
		bp = dn_copy_set(&(pipe->fs), bp);
	}

    for (i = 0; i < HASHSIZE; i++)
	SLIST_FOREACH(fs, &flowsethash[i], next) {
		struct dn_flow_set *fs_bp = (struct dn_flow_set *)bp;

		bcopy(fs, bp, sizeof(*fs));
		/* XXX same hack as above */
		fs_bp->next.sle_next = (struct dn_flow_set *)DN_IS_QUEUE;
		fs_bp->pipe = NULL;
		fs_bp->rq = NULL;
		bp += sizeof(*fs);
		bp = dn_copy_set(fs, bp);
	}

    DUMMYNET_UNLOCK();

    error = sooptcopyout(sopt, buf, size);
    free(buf, M_TEMP);
    return error ;
}

/*
 * Handler for the various dummynet socket options (get, flush, config, del)
 */
static int
ip_dn_ctl(struct sockopt *sopt)
{
    int error = 0 ;
    struct dn_pipe *p, tmp_pipe;

    error = priv_check(sopt->sopt_td, PRIV_NETINET_DUMMYNET);
    if (error)
	return (error);

    /* Disallow sets in really-really secure mode. */
    if (sopt->sopt_dir == SOPT_SET) {
#if __FreeBSD_version >= 500034
	error =  securelevel_ge(sopt->sopt_td->td_ucred, 3);
	if (error)
	    return (error);
#else
	if (securelevel >= 3)
	    return (EPERM);
#endif
    }

    switch (sopt->sopt_name) {
    default :
	printf("dummynet: -- unknown option %d", sopt->sopt_name);
	return EINVAL ;

    case IP_DUMMYNET_GET :
	error = dummynet_get(sopt);
	break ;

    case IP_DUMMYNET_FLUSH :
	dummynet_flush() ;
	break ;

    case IP_DUMMYNET_CONFIGURE :
	p = &tmp_pipe ;
	error = sooptcopyin(sopt, p, sizeof *p, sizeof *p);
	if (error)
	    break ;
	error = config_pipe(p);
	break ;

    case IP_DUMMYNET_DEL :	/* remove a pipe or queue */
	p = &tmp_pipe ;
	error = sooptcopyin(sopt, p, sizeof *p, sizeof *p);
	if (error)
	    break ;

	error = delete_pipe(p);
	break ;
    }
    return error ;
}

static void
ip_dn_init(void)
{
	int i;

	if (bootverbose)
		printf("DUMMYNET with IPv6 initialized (040826)\n");

	DUMMYNET_LOCK_INIT();

	for (i = 0; i < HASHSIZE; i++) {
		SLIST_INIT(&pipehash[i]);
		SLIST_INIT(&flowsethash[i]);
	}
	ready_heap.size = ready_heap.elements = 0;
	ready_heap.offset = 0;

	wfq_ready_heap.size = wfq_ready_heap.elements = 0;
	wfq_ready_heap.offset = 0;

	extract_heap.size = extract_heap.elements = 0;
	extract_heap.offset = 0;

	ip_dn_ctl_ptr = ip_dn_ctl;
	ip_dn_io_ptr = dummynet_io;
	ip_dn_ruledel_ptr = dn_rule_delete;

	TASK_INIT(&dn_task, 0, dummynet_task, NULL);
	dn_tq = taskqueue_create_fast("dummynet", M_NOWAIT,
	    taskqueue_thread_enqueue, &dn_tq);
	taskqueue_start_threads(&dn_tq, 1, PI_NET, "dummynet");

	callout_init(&dn_timeout, CALLOUT_MPSAFE);
	callout_reset(&dn_timeout, 1, dummynet, NULL);

	/* Initialize curr_time adjustment mechanics. */
	getmicrouptime(&prev_t);
}

#ifdef KLD_MODULE
static void
ip_dn_destroy(void)
{
	ip_dn_ctl_ptr = NULL;
	ip_dn_io_ptr = NULL;
	ip_dn_ruledel_ptr = NULL;

	DUMMYNET_LOCK();
	callout_stop(&dn_timeout);
	DUMMYNET_UNLOCK();
	taskqueue_drain(dn_tq, &dn_task);
	taskqueue_free(dn_tq);

	dummynet_flush();

	DUMMYNET_LOCK_DESTROY();
}
#endif /* KLD_MODULE */

static int
dummynet_modevent(module_t mod, int type, void *data)
{

	switch (type) {
	case MOD_LOAD:
		if (DUMMYNET_LOADED) {
		    printf("DUMMYNET already loaded\n");
		    return EEXIST ;
		}
		ip_dn_init();
		break;

	case MOD_UNLOAD:
#if !defined(KLD_MODULE)
		printf("dummynet statically compiled, cannot unload\n");
		return EINVAL ;
#else
		ip_dn_destroy();
#endif
		break ;
	default:
		return EOPNOTSUPP;
		break ;
	}
	return 0 ;
}

static moduledata_t dummynet_mod = {
	"dummynet",
	dummynet_modevent,
	NULL
};
DECLARE_MODULE(dummynet, dummynet_mod, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY);
MODULE_DEPEND(dummynet, ipfw, 2, 2, 2);
MODULE_VERSION(dummynet, 1);
OpenPOWER on IntegriCloud