aboutsummaryrefslogtreecommitdiff
path: root/sys/kern/uipc_mbuf.c
blob: 3880456bddf7259592b2a64a6deab72759d712ff (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
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
/*-
 * Copyright (c) 1982, 1986, 1988, 1991, 1993
 *	The Regents of the University of California.  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.
 * 4. Neither the name of the University nor the names of its contributors
 *    may be used to endorse or promote products derived from this software
 *    without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE REGENTS 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 REGENTS 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.
 *
 *	@(#)uipc_mbuf.c	8.2 (Berkeley) 1/4/94
 */

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

#include "opt_param.h"
#include "opt_mbuf_stress_test.h"
#include "opt_mbuf_profiling.h"

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/kernel.h>
#include <sys/limits.h>
#include <sys/lock.h>
#include <sys/malloc.h>
#include <sys/mbuf.h>
#include <sys/sysctl.h>
#include <sys/domain.h>
#include <sys/protosw.h>
#include <sys/uio.h>

int	max_linkhdr;
int	max_protohdr;
int	max_hdr;
int	max_datalen;
#ifdef MBUF_STRESS_TEST
int	m_defragpackets;
int	m_defragbytes;
int	m_defraguseless;
int	m_defragfailure;
int	m_defragrandomfailures;
#endif

/*
 * sysctl(8) exported objects
 */
SYSCTL_INT(_kern_ipc, KIPC_MAX_LINKHDR, max_linkhdr, CTLFLAG_RD,
	   &max_linkhdr, 0, "Size of largest link layer header");
SYSCTL_INT(_kern_ipc, KIPC_MAX_PROTOHDR, max_protohdr, CTLFLAG_RD,
	   &max_protohdr, 0, "Size of largest protocol layer header");
SYSCTL_INT(_kern_ipc, KIPC_MAX_HDR, max_hdr, CTLFLAG_RD,
	   &max_hdr, 0, "Size of largest link plus protocol header");
SYSCTL_INT(_kern_ipc, KIPC_MAX_DATALEN, max_datalen, CTLFLAG_RD,
	   &max_datalen, 0, "Minimum space left in mbuf after max_hdr");
#ifdef MBUF_STRESS_TEST
SYSCTL_INT(_kern_ipc, OID_AUTO, m_defragpackets, CTLFLAG_RD,
	   &m_defragpackets, 0, "");
SYSCTL_INT(_kern_ipc, OID_AUTO, m_defragbytes, CTLFLAG_RD,
	   &m_defragbytes, 0, "");
SYSCTL_INT(_kern_ipc, OID_AUTO, m_defraguseless, CTLFLAG_RD,
	   &m_defraguseless, 0, "");
SYSCTL_INT(_kern_ipc, OID_AUTO, m_defragfailure, CTLFLAG_RD,
	   &m_defragfailure, 0, "");
SYSCTL_INT(_kern_ipc, OID_AUTO, m_defragrandomfailures, CTLFLAG_RW,
	   &m_defragrandomfailures, 0, "");
#endif

/*
 * Ensure the correct size of various mbuf parameters.  It could be off due
 * to compiler-induced padding and alignment artifacts.
 */
CTASSERT(sizeof(struct mbuf) == MSIZE);
CTASSERT(MSIZE - offsetof(struct mbuf, m_dat) == MLEN);
CTASSERT(MSIZE - offsetof(struct mbuf, m_pktdat) == MHLEN);

/*
 * m_get2() allocates minimum mbuf that would fit "size" argument.
 */
struct mbuf *
m_get2(int size, int how, short type, int flags)
{
	struct mb_args args;
	struct mbuf *m, *n;

	args.flags = flags;
	args.type = type;

	if (size <= MHLEN || (size <= MLEN && (flags & M_PKTHDR) == 0))
		return (uma_zalloc_arg(zone_mbuf, &args, how));
	if (size <= MCLBYTES)
		return (uma_zalloc_arg(zone_pack, &args, how));

	if (size > MJUMPAGESIZE)
		return (NULL);

	m = uma_zalloc_arg(zone_mbuf, &args, how);
	if (m == NULL)
		return (NULL);

	n = uma_zalloc_arg(zone_jumbop, m, how);
	if (n == NULL) {
		uma_zfree(zone_mbuf, m);
		return (NULL);
	}

	return (m);
}

/*
 * m_getjcl() returns an mbuf with a cluster of the specified size attached.
 * For size it takes MCLBYTES, MJUMPAGESIZE, MJUM9BYTES, MJUM16BYTES.
 */
struct mbuf *
m_getjcl(int how, short type, int flags, int size)
{
	struct mb_args args;
	struct mbuf *m, *n;
	uma_zone_t zone;

	if (size == MCLBYTES)
		return m_getcl(how, type, flags);

	args.flags = flags;
	args.type = type;

	m = uma_zalloc_arg(zone_mbuf, &args, how);
	if (m == NULL)
		return (NULL);

	zone = m_getzone(size);
	n = uma_zalloc_arg(zone, m, how);
	if (n == NULL) {
		uma_zfree(zone_mbuf, m);
		return (NULL);
	}
	return (m);
}

/*
 * Allocate a given length worth of mbufs and/or clusters (whatever fits
 * best) and return a pointer to the top of the allocated chain.  If an
 * existing mbuf chain is provided, then we will append the new chain
 * to the existing one but still return the top of the newly allocated
 * chain.
 */
struct mbuf *
m_getm2(struct mbuf *m, int len, int how, short type, int flags)
{
	struct mbuf *mb, *nm = NULL, *mtail = NULL;

	KASSERT(len >= 0, ("%s: len is < 0", __func__));

	/* Validate flags. */
	flags &= (M_PKTHDR | M_EOR);

	/* Packet header mbuf must be first in chain. */
	if ((flags & M_PKTHDR) && m != NULL)
		flags &= ~M_PKTHDR;

	/* Loop and append maximum sized mbufs to the chain tail. */
	while (len > 0) {
		if (len > MCLBYTES)
			mb = m_getjcl(how, type, (flags & M_PKTHDR),
			    MJUMPAGESIZE);
		else if (len >= MINCLSIZE)
			mb = m_getcl(how, type, (flags & M_PKTHDR));
		else if (flags & M_PKTHDR)
			mb = m_gethdr(how, type);
		else
			mb = m_get(how, type);

		/* Fail the whole operation if one mbuf can't be allocated. */
		if (mb == NULL) {
			if (nm != NULL)
				m_freem(nm);
			return (NULL);
		}

		/* Book keeping. */
		len -= (mb->m_flags & M_EXT) ? mb->m_ext.ext_size :
			((mb->m_flags & M_PKTHDR) ? MHLEN : MLEN);
		if (mtail != NULL)
			mtail->m_next = mb;
		else
			nm = mb;
		mtail = mb;
		flags &= ~M_PKTHDR;	/* Only valid on the first mbuf. */
	}
	if (flags & M_EOR)
		mtail->m_flags |= M_EOR;  /* Only valid on the last mbuf. */

	/* If mbuf was supplied, append new chain to the end of it. */
	if (m != NULL) {
		for (mtail = m; mtail->m_next != NULL; mtail = mtail->m_next)
			;
		mtail->m_next = nm;
		mtail->m_flags &= ~M_EOR;
	} else
		m = nm;

	return (m);
}

/*
 * Free an entire chain of mbufs and associated external buffers, if
 * applicable.
 */
void
m_freem(struct mbuf *mb)
{

	while (mb != NULL)
		mb = m_free(mb);
}

/*-
 * Configure a provided mbuf to refer to the provided external storage
 * buffer and setup a reference count for said buffer.  If the setting
 * up of the reference count fails, the M_EXT bit will not be set.  If
 * successfull, the M_EXT bit is set in the mbuf's flags.
 *
 * Arguments:
 *    mb     The existing mbuf to which to attach the provided buffer.
 *    buf    The address of the provided external storage buffer.
 *    size   The size of the provided buffer.
 *    freef  A pointer to a routine that is responsible for freeing the
 *           provided external storage buffer.
 *    args   A pointer to an argument structure (of any type) to be passed
 *           to the provided freef routine (may be NULL).
 *    flags  Any other flags to be passed to the provided mbuf.
 *    type   The type that the external storage buffer should be
 *           labeled with.
 *
 * Returns:
 *    Nothing.
 */
int
m_extadd(struct mbuf *mb, caddr_t buf, u_int size,
    void (*freef)(struct mbuf *, void *, void *), void *arg1, void *arg2,
    int flags, int type, int wait)
{
	KASSERT(type != EXT_CLUSTER, ("%s: EXT_CLUSTER not allowed", __func__));

	if (type != EXT_EXTREF)
		mb->m_ext.ext_cnt = uma_zalloc(zone_ext_refcnt, wait);

	if (mb->m_ext.ext_cnt == NULL)
		return (ENOMEM);

	*(mb->m_ext.ext_cnt) = 1;
	mb->m_flags |= (M_EXT | flags);
	mb->m_ext.ext_buf = buf;
	mb->m_data = mb->m_ext.ext_buf;
	mb->m_ext.ext_size = size;
	mb->m_ext.ext_free = freef;
	mb->m_ext.ext_arg1 = arg1;
	mb->m_ext.ext_arg2 = arg2;
	mb->m_ext.ext_type = type;
	mb->m_ext.ext_flags = 0;

	return (0);
}

/*
 * Non-directly-exported function to clean up after mbufs with M_EXT
 * storage attached to them if the reference count hits 1.
 */
void
mb_free_ext(struct mbuf *m)
{
	int freembuf;

	KASSERT(m->m_flags & M_EXT, ("%s: M_EXT not set on %p", __func__, m));

	/*
	 * Check if the header is embedded in the cluster.
	 */
	freembuf = (m->m_flags & M_NOFREE) ? 0 : 1;

	switch (m->m_ext.ext_type) {
	case EXT_SFBUF:
		sf_ext_free(m->m_ext.ext_arg1, m->m_ext.ext_arg2);
		break;
	default:
		KASSERT(m->m_ext.ext_cnt != NULL,
		    ("%s: no refcounting pointer on %p", __func__, m));
		/* 
		 * Free attached storage if this mbuf is the only
		 * reference to it.
		 */
		if (*(m->m_ext.ext_cnt) != 1) {
			if (atomic_fetchadd_int(m->m_ext.ext_cnt, -1) != 1)
				break;
		}

		switch (m->m_ext.ext_type) {
		case EXT_PACKET:	/* The packet zone is special. */
			if (*(m->m_ext.ext_cnt) == 0)
				*(m->m_ext.ext_cnt) = 1;
			uma_zfree(zone_pack, m);
			return;		/* Job done. */
		case EXT_CLUSTER:
			uma_zfree(zone_clust, m->m_ext.ext_buf);
			break;
		case EXT_JUMBOP:
			uma_zfree(zone_jumbop, m->m_ext.ext_buf);
			break;
		case EXT_JUMBO9:
			uma_zfree(zone_jumbo9, m->m_ext.ext_buf);
			break;
		case EXT_JUMBO16:
			uma_zfree(zone_jumbo16, m->m_ext.ext_buf);
			break;
		case EXT_NET_DRV:
		case EXT_MOD_TYPE:
		case EXT_DISPOSABLE:
			*(m->m_ext.ext_cnt) = 0;
			uma_zfree(zone_ext_refcnt, __DEVOLATILE(u_int *,
				m->m_ext.ext_cnt));
			/* FALLTHROUGH */
		case EXT_EXTREF:
			KASSERT(m->m_ext.ext_free != NULL,
				("%s: ext_free not set", __func__));
			(*(m->m_ext.ext_free))(m, m->m_ext.ext_arg1,
			    m->m_ext.ext_arg2);
			break;
		default:
			KASSERT(m->m_ext.ext_type == 0,
				("%s: unknown ext_type", __func__));
		}
	}

	if (freembuf)
		uma_zfree(zone_mbuf, m);
}

/*
 * Attach the cluster from *m to *n, set up m_ext in *n
 * and bump the refcount of the cluster.
 */
static void
mb_dupcl(struct mbuf *n, struct mbuf *m)
{

	KASSERT(m->m_flags & M_EXT, ("%s: M_EXT not set on %p", __func__, m));
	KASSERT(!(n->m_flags & M_EXT), ("%s: M_EXT set on %p", __func__, n));

	switch (m->m_ext.ext_type) {
	case EXT_SFBUF:
		sf_ext_ref(m->m_ext.ext_arg1, m->m_ext.ext_arg2);
		break;
	default:
		KASSERT(m->m_ext.ext_cnt != NULL,
		    ("%s: no refcounting pointer on %p", __func__, m));
		if (*(m->m_ext.ext_cnt) == 1)
			*(m->m_ext.ext_cnt) += 1;
		else
			atomic_add_int(m->m_ext.ext_cnt, 1);
	}

	n->m_ext = m->m_ext;
	n->m_flags |= M_EXT;
	n->m_flags |= m->m_flags & M_RDONLY;
}

/*
 * Clean up mbuf (chain) from any tags and packet headers.
 * If "all" is set then the first mbuf in the chain will be
 * cleaned too.
 */
void
m_demote(struct mbuf *m0, int all, int flags)
{
	struct mbuf *m;

	for (m = all ? m0 : m0->m_next; m != NULL; m = m->m_next) {
		KASSERT(m->m_nextpkt == NULL, ("%s: m_nextpkt in m %p, m0 %p",
		    __func__, m, m0));
		if (m->m_flags & M_PKTHDR) {
			m_tag_delete_chain(m, NULL);
			m->m_flags &= ~M_PKTHDR;
			bzero(&m->m_pkthdr, sizeof(struct pkthdr));
		}
		m->m_flags = m->m_flags & (M_EXT | M_RDONLY | M_NOFREE | flags);
	}
}

/*
 * Sanity checks on mbuf (chain) for use in KASSERT() and general
 * debugging.
 * Returns 0 or panics when bad and 1 on all tests passed.
 * Sanitize, 0 to run M_SANITY_ACTION, 1 to garble things so they
 * blow up later.
 */
int
m_sanity(struct mbuf *m0, int sanitize)
{
	struct mbuf *m;
	caddr_t a, b;
	int pktlen = 0;

#ifdef INVARIANTS
#define	M_SANITY_ACTION(s)	panic("mbuf %p: " s, m)
#else
#define	M_SANITY_ACTION(s)	printf("mbuf %p: " s, m)
#endif

	for (m = m0; m != NULL; m = m->m_next) {
		/*
		 * Basic pointer checks.  If any of these fails then some
		 * unrelated kernel memory before or after us is trashed.
		 * No way to recover from that.
		 */
		a = ((m->m_flags & M_EXT) ? m->m_ext.ext_buf :
			((m->m_flags & M_PKTHDR) ? (caddr_t)(&m->m_pktdat) :
			 (caddr_t)(&m->m_dat)) );
		b = (caddr_t)(a + (m->m_flags & M_EXT ? m->m_ext.ext_size :
			((m->m_flags & M_PKTHDR) ? MHLEN : MLEN)));
		if ((caddr_t)m->m_data < a)
			M_SANITY_ACTION("m_data outside mbuf data range left");
		if ((caddr_t)m->m_data > b)
			M_SANITY_ACTION("m_data outside mbuf data range right");
		if ((caddr_t)m->m_data + m->m_len > b)
			M_SANITY_ACTION("m_data + m_len exeeds mbuf space");

		/* m->m_nextpkt may only be set on first mbuf in chain. */
		if (m != m0 && m->m_nextpkt != NULL) {
			if (sanitize) {
				m_freem(m->m_nextpkt);
				m->m_nextpkt = (struct mbuf *)0xDEADC0DE;
			} else
				M_SANITY_ACTION("m->m_nextpkt on in-chain mbuf");
		}

		/* packet length (not mbuf length!) calculation */
		if (m0->m_flags & M_PKTHDR)
			pktlen += m->m_len;

		/* m_tags may only be attached to first mbuf in chain. */
		if (m != m0 && m->m_flags & M_PKTHDR &&
		    !SLIST_EMPTY(&m->m_pkthdr.tags)) {
			if (sanitize) {
				m_tag_delete_chain(m, NULL);
				/* put in 0xDEADC0DE perhaps? */
			} else
				M_SANITY_ACTION("m_tags on in-chain mbuf");
		}

		/* M_PKTHDR may only be set on first mbuf in chain */
		if (m != m0 && m->m_flags & M_PKTHDR) {
			if (sanitize) {
				bzero(&m->m_pkthdr, sizeof(m->m_pkthdr));
				m->m_flags &= ~M_PKTHDR;
				/* put in 0xDEADCODE and leave hdr flag in */
			} else
				M_SANITY_ACTION("M_PKTHDR on in-chain mbuf");
		}
	}
	m = m0;
	if (pktlen && pktlen != m->m_pkthdr.len) {
		if (sanitize)
			m->m_pkthdr.len = 0;
		else
			M_SANITY_ACTION("m_pkthdr.len != mbuf chain length");
	}
	return 1;

#undef	M_SANITY_ACTION
}


/*
 * "Move" mbuf pkthdr from "from" to "to".
 * "from" must have M_PKTHDR set, and "to" must be empty.
 */
void
m_move_pkthdr(struct mbuf *to, struct mbuf *from)
{

#if 0
	/* see below for why these are not enabled */
	M_ASSERTPKTHDR(to);
	/* Note: with MAC, this may not be a good assertion. */
	KASSERT(SLIST_EMPTY(&to->m_pkthdr.tags),
	    ("m_move_pkthdr: to has tags"));
#endif
#ifdef MAC
	/*
	 * XXXMAC: It could be this should also occur for non-MAC?
	 */
	if (to->m_flags & M_PKTHDR)
		m_tag_delete_chain(to, NULL);
#endif
	to->m_flags = (from->m_flags & M_COPYFLAGS) | (to->m_flags & M_EXT);
	if ((to->m_flags & M_EXT) == 0)
		to->m_data = to->m_pktdat;
	to->m_pkthdr = from->m_pkthdr;		/* especially tags */
	SLIST_INIT(&from->m_pkthdr.tags);	/* purge tags from src */
	from->m_flags &= ~M_PKTHDR;
}

/*
 * Duplicate "from"'s mbuf pkthdr in "to".
 * "from" must have M_PKTHDR set, and "to" must be empty.
 * In particular, this does a deep copy of the packet tags.
 */
int
m_dup_pkthdr(struct mbuf *to, struct mbuf *from, int how)
{

#if 0
	/*
	 * The mbuf allocator only initializes the pkthdr
	 * when the mbuf is allocated with m_gethdr(). Many users
	 * (e.g. m_copy*, m_prepend) use m_get() and then
	 * smash the pkthdr as needed causing these
	 * assertions to trip.  For now just disable them.
	 */
	M_ASSERTPKTHDR(to);
	/* Note: with MAC, this may not be a good assertion. */
	KASSERT(SLIST_EMPTY(&to->m_pkthdr.tags), ("m_dup_pkthdr: to has tags"));
#endif
	MBUF_CHECKSLEEP(how);
#ifdef MAC
	if (to->m_flags & M_PKTHDR)
		m_tag_delete_chain(to, NULL);
#endif
	to->m_flags = (from->m_flags & M_COPYFLAGS) | (to->m_flags & M_EXT);
	if ((to->m_flags & M_EXT) == 0)
		to->m_data = to->m_pktdat;
	to->m_pkthdr = from->m_pkthdr;
	SLIST_INIT(&to->m_pkthdr.tags);
	return (m_tag_copy_chain(to, from, how));
}

/*
 * Lesser-used path for M_PREPEND:
 * allocate new mbuf to prepend to chain,
 * copy junk along.
 */
struct mbuf *
m_prepend(struct mbuf *m, int len, int how)
{
	struct mbuf *mn;

	if (m->m_flags & M_PKTHDR)
		mn = m_gethdr(how, m->m_type);
	else
		mn = m_get(how, m->m_type);
	if (mn == NULL) {
		m_freem(m);
		return (NULL);
	}
	if (m->m_flags & M_PKTHDR)
		m_move_pkthdr(mn, m);
	mn->m_next = m;
	m = mn;
	if(m->m_flags & M_PKTHDR) {
		if (len < MHLEN)
			MH_ALIGN(m, len);
	} else {
		if (len < MLEN)
			M_ALIGN(m, len);
	}
	m->m_len = len;
	return (m);
}

/*
 * Make a copy of an mbuf chain starting "off0" bytes from the beginning,
 * continuing for "len" bytes.  If len is M_COPYALL, copy to end of mbuf.
 * The wait parameter is a choice of M_WAITOK/M_NOWAIT from caller.
 * Note that the copy is read-only, because clusters are not copied,
 * only their reference counts are incremented.
 */
struct mbuf *
m_copym(struct mbuf *m, int off0, int len, int wait)
{
	struct mbuf *n, **np;
	int off = off0;
	struct mbuf *top;
	int copyhdr = 0;

	KASSERT(off >= 0, ("m_copym, negative off %d", off));
	KASSERT(len >= 0, ("m_copym, negative len %d", len));
	MBUF_CHECKSLEEP(wait);
	if (off == 0 && m->m_flags & M_PKTHDR)
		copyhdr = 1;
	while (off > 0) {
		KASSERT(m != NULL, ("m_copym, offset > size of mbuf chain"));
		if (off < m->m_len)
			break;
		off -= m->m_len;
		m = m->m_next;
	}
	np = &top;
	top = 0;
	while (len > 0) {
		if (m == NULL) {
			KASSERT(len == M_COPYALL,
			    ("m_copym, length > size of mbuf chain"));
			break;
		}
		if (copyhdr)
			n = m_gethdr(wait, m->m_type);
		else
			n = m_get(wait, m->m_type);
		*np = n;
		if (n == NULL)
			goto nospace;
		if (copyhdr) {
			if (!m_dup_pkthdr(n, m, wait))
				goto nospace;
			if (len == M_COPYALL)
				n->m_pkthdr.len -= off0;
			else
				n->m_pkthdr.len = len;
			copyhdr = 0;
		}
		n->m_len = min(len, m->m_len - off);
		if (m->m_flags & M_EXT) {
			n->m_data = m->m_data + off;
			mb_dupcl(n, m);
		} else
			bcopy(mtod(m, caddr_t)+off, mtod(n, caddr_t),
			    (u_int)n->m_len);
		if (len != M_COPYALL)
			len -= n->m_len;
		off = 0;
		m = m->m_next;
		np = &n->m_next;
	}

	return (top);
nospace:
	m_freem(top);
	return (NULL);
}

/*
 * Returns mbuf chain with new head for the prepending case.
 * Copies from mbuf (chain) n from off for len to mbuf (chain) m
 * either prepending or appending the data.
 * The resulting mbuf (chain) m is fully writeable.
 * m is destination (is made writeable)
 * n is source, off is offset in source, len is len from offset
 * dir, 0 append, 1 prepend
 * how, wait or nowait
 */

static int
m_bcopyxxx(void *s, void *t, u_int len)
{
	bcopy(s, t, (size_t)len);
	return 0;
}

struct mbuf *
m_copymdata(struct mbuf *m, struct mbuf *n, int off, int len,
    int prep, int how)
{
	struct mbuf *mm, *x, *z, *prev = NULL;
	caddr_t p;
	int i, nlen = 0;
	caddr_t buf[MLEN];

	KASSERT(m != NULL && n != NULL, ("m_copymdata, no target or source"));
	KASSERT(off >= 0, ("m_copymdata, negative off %d", off));
	KASSERT(len >= 0, ("m_copymdata, negative len %d", len));
	KASSERT(prep == 0 || prep == 1, ("m_copymdata, unknown direction %d", prep));

	mm = m;
	if (!prep) {
		while(mm->m_next) {
			prev = mm;
			mm = mm->m_next;
		}
	}
	for (z = n; z != NULL; z = z->m_next)
		nlen += z->m_len;
	if (len == M_COPYALL)
		len = nlen - off;
	if (off + len > nlen || len < 1)
		return NULL;

	if (!M_WRITABLE(mm)) {
		/* XXX: Use proper m_xxx function instead. */
		x = m_getcl(how, MT_DATA, mm->m_flags);
		if (x == NULL)
			return NULL;
		bcopy(mm->m_ext.ext_buf, x->m_ext.ext_buf, x->m_ext.ext_size);
		p = x->m_ext.ext_buf + (mm->m_data - mm->m_ext.ext_buf);
		x->m_data = p;
		mm->m_next = NULL;
		if (mm != m)
			prev->m_next = x;
		m_free(mm);
		mm = x;
	}

	/*
	 * Append/prepend the data.  Allocating mbufs as necessary.
	 */
	/* Shortcut if enough free space in first/last mbuf. */
	if (!prep && M_TRAILINGSPACE(mm) >= len) {
		m_apply(n, off, len, m_bcopyxxx, mtod(mm, caddr_t) +
			 mm->m_len);
		mm->m_len += len;
		mm->m_pkthdr.len += len;
		return m;
	}
	if (prep && M_LEADINGSPACE(mm) >= len) {
		mm->m_data = mtod(mm, caddr_t) - len;
		m_apply(n, off, len, m_bcopyxxx, mtod(mm, caddr_t));
		mm->m_len += len;
		mm->m_pkthdr.len += len;
		return mm;
	}

	/* Expand first/last mbuf to cluster if possible. */
	if (!prep && !(mm->m_flags & M_EXT) && len > M_TRAILINGSPACE(mm)) {
		bcopy(mm->m_data, &buf, mm->m_len);
		m_clget(mm, how);
		if (!(mm->m_flags & M_EXT))
			return NULL;
		bcopy(&buf, mm->m_ext.ext_buf, mm->m_len);
		mm->m_data = mm->m_ext.ext_buf;
	}
	if (prep && !(mm->m_flags & M_EXT) && len > M_LEADINGSPACE(mm)) {
		bcopy(mm->m_data, &buf, mm->m_len);
		m_clget(mm, how);
		if (!(mm->m_flags & M_EXT))
			return NULL;
		bcopy(&buf, (caddr_t *)mm->m_ext.ext_buf +
		    mm->m_ext.ext_size - mm->m_len, mm->m_len);
		mm->m_data = (caddr_t)mm->m_ext.ext_buf +
		    mm->m_ext.ext_size - mm->m_len;
	}

	/* Append/prepend as many mbuf (clusters) as necessary to fit len. */
	if (!prep && len > M_TRAILINGSPACE(mm)) {
		if (!m_getm(mm, len - M_TRAILINGSPACE(mm), how, MT_DATA))
			return NULL;
	}
	if (prep && len > M_LEADINGSPACE(mm)) {
		if (!(z = m_getm(NULL, len - M_LEADINGSPACE(mm), how, MT_DATA)))
			return NULL;
		i = 0;
		for (x = z; x != NULL; x = x->m_next) {
			i += x->m_flags & M_EXT ? x->m_ext.ext_size :
			    (x->m_flags & M_PKTHDR ? MHLEN : MLEN);
			if (!x->m_next)
				break;
		}
		z->m_data += i - len;
		m_move_pkthdr(mm, z);
		x->m_next = mm;
		mm = z;
	}

	/* Seek to start position in source mbuf. Optimization for long chains. */
	while (off > 0) {
		if (off < n->m_len)
			break;
		off -= n->m_len;
		n = n->m_next;
	}

	/* Copy data into target mbuf. */
	z = mm;
	while (len > 0) {
		KASSERT(z != NULL, ("m_copymdata, falling off target edge"));
		i = M_TRAILINGSPACE(z);
		m_apply(n, off, i, m_bcopyxxx, mtod(z, caddr_t) + z->m_len);
		z->m_len += i;
		/* fixup pkthdr.len if necessary */
		if ((prep ? mm : m)->m_flags & M_PKTHDR)
			(prep ? mm : m)->m_pkthdr.len += i;
		off += i;
		len -= i;
		z = z->m_next;
	}
	return (prep ? mm : m);
}

/*
 * Copy an entire packet, including header (which must be present).
 * An optimization of the common case `m_copym(m, 0, M_COPYALL, how)'.
 * Note that the copy is read-only, because clusters are not copied,
 * only their reference counts are incremented.
 * Preserve alignment of the first mbuf so if the creator has left
 * some room at the beginning (e.g. for inserting protocol headers)
 * the copies still have the room available.
 */
struct mbuf *
m_copypacket(struct mbuf *m, int how)
{
	struct mbuf *top, *n, *o;

	MBUF_CHECKSLEEP(how);
	n = m_get(how, m->m_type);
	top = n;
	if (n == NULL)
		goto nospace;

	if (!m_dup_pkthdr(n, m, how))
		goto nospace;
	n->m_len = m->m_len;
	if (m->m_flags & M_EXT) {
		n->m_data = m->m_data;
		mb_dupcl(n, m);
	} else {
		n->m_data = n->m_pktdat + (m->m_data - m->m_pktdat );
		bcopy(mtod(m, char *), mtod(n, char *), n->m_len);
	}

	m = m->m_next;
	while (m) {
		o = m_get(how, m->m_type);
		if (o == NULL)
			goto nospace;

		n->m_next = o;
		n = n->m_next;

		n->m_len = m->m_len;
		if (m->m_flags & M_EXT) {
			n->m_data = m->m_data;
			mb_dupcl(n, m);
		} else {
			bcopy(mtod(m, char *), mtod(n, char *), n->m_len);
		}

		m = m->m_next;
	}
	return top;
nospace:
	m_freem(top);
	return (NULL);
}

/*
 * Copy data from an mbuf chain starting "off" bytes from the beginning,
 * continuing for "len" bytes, into the indicated buffer.
 */
void
m_copydata(const struct mbuf *m, int off, int len, caddr_t cp)
{
	u_int count;

	KASSERT(off >= 0, ("m_copydata, negative off %d", off));
	KASSERT(len >= 0, ("m_copydata, negative len %d", len));
	while (off > 0) {
		KASSERT(m != NULL, ("m_copydata, offset > size of mbuf chain"));
		if (off < m->m_len)
			break;
		off -= m->m_len;
		m = m->m_next;
	}
	while (len > 0) {
		KASSERT(m != NULL, ("m_copydata, length > size of mbuf chain"));
		count = min(m->m_len - off, len);
		bcopy(mtod(m, caddr_t) + off, cp, count);
		len -= count;
		cp += count;
		off = 0;
		m = m->m_next;
	}
}

/*
 * Copy a packet header mbuf chain into a completely new chain, including
 * copying any mbuf clusters.  Use this instead of m_copypacket() when
 * you need a writable copy of an mbuf chain.
 */
struct mbuf *
m_dup(struct mbuf *m, int how)
{
	struct mbuf **p, *top = NULL;
	int remain, moff, nsize;

	MBUF_CHECKSLEEP(how);
	/* Sanity check */
	if (m == NULL)
		return (NULL);
	M_ASSERTPKTHDR(m);

	/* While there's more data, get a new mbuf, tack it on, and fill it */
	remain = m->m_pkthdr.len;
	moff = 0;
	p = &top;
	while (remain > 0 || top == NULL) {	/* allow m->m_pkthdr.len == 0 */
		struct mbuf *n;

		/* Get the next new mbuf */
		if (remain >= MINCLSIZE) {
			n = m_getcl(how, m->m_type, 0);
			nsize = MCLBYTES;
		} else {
			n = m_get(how, m->m_type);
			nsize = MLEN;
		}
		if (n == NULL)
			goto nospace;

		if (top == NULL) {		/* First one, must be PKTHDR */
			if (!m_dup_pkthdr(n, m, how)) {
				m_free(n);
				goto nospace;
			}
			if ((n->m_flags & M_EXT) == 0)
				nsize = MHLEN;
		}
		n->m_len = 0;

		/* Link it into the new chain */
		*p = n;
		p = &n->m_next;

		/* Copy data from original mbuf(s) into new mbuf */
		while (n->m_len < nsize && m != NULL) {
			int chunk = min(nsize - n->m_len, m->m_len - moff);

			bcopy(m->m_data + moff, n->m_data + n->m_len, chunk);
			moff += chunk;
			n->m_len += chunk;
			remain -= chunk;
			if (moff == m->m_len) {
				m = m->m_next;
				moff = 0;
			}
		}

		/* Check correct total mbuf length */
		KASSERT((remain > 0 && m != NULL) || (remain == 0 && m == NULL),
		    	("%s: bogus m_pkthdr.len", __func__));
	}
	return (top);

nospace:
	m_freem(top);
	return (NULL);
}

/*
 * Concatenate mbuf chain n to m.
 * Both chains must be of the same type (e.g. MT_DATA).
 * Any m_pkthdr is not updated.
 */
void
m_cat(struct mbuf *m, struct mbuf *n)
{
	while (m->m_next)
		m = m->m_next;
	while (n) {
		if (!M_WRITABLE(m) ||
		    M_TRAILINGSPACE(m) < n->m_len) {
			/* just join the two chains */
			m->m_next = n;
			return;
		}
		/* splat the data from one into the other */
		bcopy(mtod(n, caddr_t), mtod(m, caddr_t) + m->m_len,
		    (u_int)n->m_len);
		m->m_len += n->m_len;
		n = m_free(n);
	}
}

/*
 * Concatenate two pkthdr mbuf chains.
 */
void
m_catpkt(struct mbuf *m, struct mbuf *n)
{

	M_ASSERTPKTHDR(m);
	M_ASSERTPKTHDR(n);

	m->m_pkthdr.len += n->m_pkthdr.len;
	m_demote(n, 1, 0);

	m_cat(m, n);
}

void
m_adj(struct mbuf *mp, int req_len)
{
	int len = req_len;
	struct mbuf *m;
	int count;

	if ((m = mp) == NULL)
		return;
	if (len >= 0) {
		/*
		 * Trim from head.
		 */
		while (m != NULL && len > 0) {
			if (m->m_len <= len) {
				len -= m->m_len;
				m->m_len = 0;
				m = m->m_next;
			} else {
				m->m_len -= len;
				m->m_data += len;
				len = 0;
			}
		}
		if (mp->m_flags & M_PKTHDR)
			mp->m_pkthdr.len -= (req_len - len);
	} else {
		/*
		 * Trim from tail.  Scan the mbuf chain,
		 * calculating its length and finding the last mbuf.
		 * If the adjustment only affects this mbuf, then just
		 * adjust and return.  Otherwise, rescan and truncate
		 * after the remaining size.
		 */
		len = -len;
		count = 0;
		for (;;) {
			count += m->m_len;
			if (m->m_next == (struct mbuf *)0)
				break;
			m = m->m_next;
		}
		if (m->m_len >= len) {
			m->m_len -= len;
			if (mp->m_flags & M_PKTHDR)
				mp->m_pkthdr.len -= len;
			return;
		}
		count -= len;
		if (count < 0)
			count = 0;
		/*
		 * Correct length for chain is "count".
		 * Find the mbuf with last data, adjust its length,
		 * and toss data from remaining mbufs on chain.
		 */
		m = mp;
		if (m->m_flags & M_PKTHDR)
			m->m_pkthdr.len = count;
		for (; m; m = m->m_next) {
			if (m->m_len >= count) {
				m->m_len = count;
				if (m->m_next != NULL) {
					m_freem(m->m_next);
					m->m_next = NULL;
				}
				break;
			}
			count -= m->m_len;
		}
	}
}

/*
 * Rearange an mbuf chain so that len bytes are contiguous
 * and in the data area of an mbuf (so that mtod will work
 * for a structure of size len).  Returns the resulting
 * mbuf chain on success, frees it and returns null on failure.
 * If there is room, it will add up to max_protohdr-len extra bytes to the
 * contiguous region in an attempt to avoid being called next time.
 */
struct mbuf *
m_pullup(struct mbuf *n, int len)
{
	struct mbuf *m;
	int count;
	int space;

	/*
	 * If first mbuf has no cluster, and has room for len bytes
	 * without shifting current data, pullup into it,
	 * otherwise allocate a new mbuf to prepend to the chain.
	 */
	if ((n->m_flags & M_EXT) == 0 &&
	    n->m_data + len < &n->m_dat[MLEN] && n->m_next) {
		if (n->m_len >= len)
			return (n);
		m = n;
		n = n->m_next;
		len -= m->m_len;
	} else {
		if (len > MHLEN)
			goto bad;
		m = m_get(M_NOWAIT, n->m_type);
		if (m == NULL)
			goto bad;
		if (n->m_flags & M_PKTHDR)
			m_move_pkthdr(m, n);
	}
	space = &m->m_dat[MLEN] - (m->m_data + m->m_len);
	do {
		count = min(min(max(len, max_protohdr), space), n->m_len);
		bcopy(mtod(n, caddr_t), mtod(m, caddr_t) + m->m_len,
		  (u_int)count);
		len -= count;
		m->m_len += count;
		n->m_len -= count;
		space -= count;
		if (n->m_len)
			n->m_data += count;
		else
			n = m_free(n);
	} while (len > 0 && n);
	if (len > 0) {
		(void) m_free(m);
		goto bad;
	}
	m->m_next = n;
	return (m);
bad:
	m_freem(n);
	return (NULL);
}

/*
 * Like m_pullup(), except a new mbuf is always allocated, and we allow
 * the amount of empty space before the data in the new mbuf to be specified
 * (in the event that the caller expects to prepend later).
 */
int MSFail;

struct mbuf *
m_copyup(struct mbuf *n, int len, int dstoff)
{
	struct mbuf *m;
	int count, space;

	if (len > (MHLEN - dstoff))
		goto bad;
	m = m_get(M_NOWAIT, n->m_type);
	if (m == NULL)
		goto bad;
	if (n->m_flags & M_PKTHDR)
		m_move_pkthdr(m, n);
	m->m_data += dstoff;
	space = &m->m_dat[MLEN] - (m->m_data + m->m_len);
	do {
		count = min(min(max(len, max_protohdr), space), n->m_len);
		memcpy(mtod(m, caddr_t) + m->m_len, mtod(n, caddr_t),
		    (unsigned)count);
		len -= count;
		m->m_len += count;
		n->m_len -= count;
		space -= count;
		if (n->m_len)
			n->m_data += count;
		else
			n = m_free(n);
	} while (len > 0 && n);
	if (len > 0) {
		(void) m_free(m);
		goto bad;
	}
	m->m_next = n;
	return (m);
 bad:
	m_freem(n);
	MSFail++;
	return (NULL);
}

/*
 * Partition an mbuf chain in two pieces, returning the tail --
 * all but the first len0 bytes.  In case of failure, it returns NULL and
 * attempts to restore the chain to its original state.
 *
 * Note that the resulting mbufs might be read-only, because the new
 * mbuf can end up sharing an mbuf cluster with the original mbuf if
 * the "breaking point" happens to lie within a cluster mbuf. Use the
 * M_WRITABLE() macro to check for this case.
 */
struct mbuf *
m_split(struct mbuf *m0, int len0, int wait)
{
	struct mbuf *m, *n;
	u_int len = len0, remain;

	MBUF_CHECKSLEEP(wait);
	for (m = m0; m && len > m->m_len; m = m->m_next)
		len -= m->m_len;
	if (m == NULL)
		return (NULL);
	remain = m->m_len - len;
	if (m0->m_flags & M_PKTHDR && remain == 0) {
		n = m_gethdr(wait, m0->m_type);
		if (n == NULL)
			return (NULL);
		n->m_next = m->m_next;
		m->m_next = NULL;
		n->m_pkthdr.rcvif = m0->m_pkthdr.rcvif;
		n->m_pkthdr.len = m0->m_pkthdr.len - len0;
		m0->m_pkthdr.len = len0;
		return (n);
	} else if (m0->m_flags & M_PKTHDR) {
		n = m_gethdr(wait, m0->m_type);
		if (n == NULL)
			return (NULL);
		n->m_pkthdr.rcvif = m0->m_pkthdr.rcvif;
		n->m_pkthdr.len = m0->m_pkthdr.len - len0;
		m0->m_pkthdr.len = len0;
		if (m->m_flags & M_EXT)
			goto extpacket;
		if (remain > MHLEN) {
			/* m can't be the lead packet */
			MH_ALIGN(n, 0);
			n->m_next = m_split(m, len, wait);
			if (n->m_next == NULL) {
				(void) m_free(n);
				return (NULL);
			} else {
				n->m_len = 0;
				return (n);
			}
		} else
			MH_ALIGN(n, remain);
	} else if (remain == 0) {
		n = m->m_next;
		m->m_next = NULL;
		return (n);
	} else {
		n = m_get(wait, m->m_type);
		if (n == NULL)
			return (NULL);
		M_ALIGN(n, remain);
	}
extpacket:
	if (m->m_flags & M_EXT) {
		n->m_data = m->m_data + len;
		mb_dupcl(n, m);
	} else {
		bcopy(mtod(m, caddr_t) + len, mtod(n, caddr_t), remain);
	}
	n->m_len = remain;
	m->m_len = len;
	n->m_next = m->m_next;
	m->m_next = NULL;
	return (n);
}
/*
 * Routine to copy from device local memory into mbufs.
 * Note that `off' argument is offset into first mbuf of target chain from
 * which to begin copying the data to.
 */
struct mbuf *
m_devget(char *buf, int totlen, int off, struct ifnet *ifp,
    void (*copy)(char *from, caddr_t to, u_int len))
{
	struct mbuf *m;
	struct mbuf *top = NULL, **mp = &top;
	int len;

	if (off < 0 || off > MHLEN)
		return (NULL);

	while (totlen > 0) {
		if (top == NULL) {	/* First one, must be PKTHDR */
			if (totlen + off >= MINCLSIZE) {
				m = m_getcl(M_NOWAIT, MT_DATA, M_PKTHDR);
				len = MCLBYTES;
			} else {
				m = m_gethdr(M_NOWAIT, MT_DATA);
				len = MHLEN;

				/* Place initial small packet/header at end of mbuf */
				if (m && totlen + off + max_linkhdr <= MLEN) {
					m->m_data += max_linkhdr;
					len -= max_linkhdr;
				}
			}
			if (m == NULL)
				return NULL;
			m->m_pkthdr.rcvif = ifp;
			m->m_pkthdr.len = totlen;
		} else {
			if (totlen + off >= MINCLSIZE) {
				m = m_getcl(M_NOWAIT, MT_DATA, 0);
				len = MCLBYTES;
			} else {
				m = m_get(M_NOWAIT, MT_DATA);
				len = MLEN;
			}
			if (m == NULL) {
				m_freem(top);
				return NULL;
			}
		}
		if (off) {
			m->m_data += off;
			len -= off;
			off = 0;
		}
		m->m_len = len = min(totlen, len);
		if (copy)
			copy(buf, mtod(m, caddr_t), (u_int)len);
		else
			bcopy(buf, mtod(m, caddr_t), (u_int)len);
		buf += len;
		*mp = m;
		mp = &m->m_next;
		totlen -= len;
	}
	return (top);
}

/*
 * Copy data from a buffer back into the indicated mbuf chain,
 * starting "off" bytes from the beginning, extending the mbuf
 * chain if necessary.
 */
void
m_copyback(struct mbuf *m0, int off, int len, c_caddr_t cp)
{
	int mlen;
	struct mbuf *m = m0, *n;
	int totlen = 0;

	if (m0 == NULL)
		return;
	while (off > (mlen = m->m_len)) {
		off -= mlen;
		totlen += mlen;
		if (m->m_next == NULL) {
			n = m_get(M_NOWAIT, m->m_type);
			if (n == NULL)
				goto out;
			bzero(mtod(n, caddr_t), MLEN);
			n->m_len = min(MLEN, len + off);
			m->m_next = n;
		}
		m = m->m_next;
	}
	while (len > 0) {
		if (m->m_next == NULL && (len > m->m_len - off)) {
			m->m_len += min(len - (m->m_len - off),
			    M_TRAILINGSPACE(m));
		}
		mlen = min (m->m_len - off, len);
		bcopy(cp, off + mtod(m, caddr_t), (u_int)mlen);
		cp += mlen;
		len -= mlen;
		mlen += off;
		off = 0;
		totlen += mlen;
		if (len == 0)
			break;
		if (m->m_next == NULL) {
			n = m_get(M_NOWAIT, m->m_type);
			if (n == NULL)
				break;
			n->m_len = min(MLEN, len);
			m->m_next = n;
		}
		m = m->m_next;
	}
out:	if (((m = m0)->m_flags & M_PKTHDR) && (m->m_pkthdr.len < totlen))
		m->m_pkthdr.len = totlen;
}

/*
 * Append the specified data to the indicated mbuf chain,
 * Extend the mbuf chain if the new data does not fit in
 * existing space.
 *
 * Return 1 if able to complete the job; otherwise 0.
 */
int
m_append(struct mbuf *m0, int len, c_caddr_t cp)
{
	struct mbuf *m, *n;
	int remainder, space;

	for (m = m0; m->m_next != NULL; m = m->m_next)
		;
	remainder = len;
	space = M_TRAILINGSPACE(m);
	if (space > 0) {
		/*
		 * Copy into available space.
		 */
		if (space > remainder)
			space = remainder;
		bcopy(cp, mtod(m, caddr_t) + m->m_len, space);
		m->m_len += space;
		cp += space, remainder -= space;
	}
	while (remainder > 0) {
		/*
		 * Allocate a new mbuf; could check space
		 * and allocate a cluster instead.
		 */
		n = m_get(M_NOWAIT, m->m_type);
		if (n == NULL)
			break;
		n->m_len = min(MLEN, remainder);
		bcopy(cp, mtod(n, caddr_t), n->m_len);
		cp += n->m_len, remainder -= n->m_len;
		m->m_next = n;
		m = n;
	}
	if (m0->m_flags & M_PKTHDR)
		m0->m_pkthdr.len += len - remainder;
	return (remainder == 0);
}

/*
 * Apply function f to the data in an mbuf chain starting "off" bytes from
 * the beginning, continuing for "len" bytes.
 */
int
m_apply(struct mbuf *m, int off, int len,
    int (*f)(void *, void *, u_int), void *arg)
{
	u_int count;
	int rval;

	KASSERT(off >= 0, ("m_apply, negative off %d", off));
	KASSERT(len >= 0, ("m_apply, negative len %d", len));
	while (off > 0) {
		KASSERT(m != NULL, ("m_apply, offset > size of mbuf chain"));
		if (off < m->m_len)
			break;
		off -= m->m_len;
		m = m->m_next;
	}
	while (len > 0) {
		KASSERT(m != NULL, ("m_apply, offset > size of mbuf chain"));
		count = min(m->m_len - off, len);
		rval = (*f)(arg, mtod(m, caddr_t) + off, count);
		if (rval)
			return (rval);
		len -= count;
		off = 0;
		m = m->m_next;
	}
	return (0);
}

/*
 * Return a pointer to mbuf/offset of location in mbuf chain.
 */
struct mbuf *
m_getptr(struct mbuf *m, int loc, int *off)
{

	while (loc >= 0) {
		/* Normal end of search. */
		if (m->m_len > loc) {
			*off = loc;
			return (m);
		} else {
			loc -= m->m_len;
			if (m->m_next == NULL) {
				if (loc == 0) {
					/* Point at the end of valid data. */
					*off = m->m_len;
					return (m);
				}
				return (NULL);
			}
			m = m->m_next;
		}
	}
	return (NULL);
}

void
m_print(const struct mbuf *m, int maxlen)
{
	int len;
	int pdata;
	const struct mbuf *m2;

	if (m == NULL) {
		printf("mbuf: %p\n", m);
		return;
	}

	if (m->m_flags & M_PKTHDR)
		len = m->m_pkthdr.len;
	else
		len = -1;
	m2 = m;
	while (m2 != NULL && (len == -1 || len)) {
		pdata = m2->m_len;
		if (maxlen != -1 && pdata > maxlen)
			pdata = maxlen;
		printf("mbuf: %p len: %d, next: %p, %b%s", m2, m2->m_len,
		    m2->m_next, m2->m_flags, "\20\20freelist\17skipfw"
		    "\11proto5\10proto4\7proto3\6proto2\5proto1\4rdonly"
		    "\3eor\2pkthdr\1ext", pdata ? "" : "\n");
		if (pdata)
			printf(", %*D\n", pdata, (u_char *)m2->m_data, "-");
		if (len != -1)
			len -= m2->m_len;
		m2 = m2->m_next;
	}
	if (len > 0)
		printf("%d bytes unaccounted for.\n", len);
	return;
}

u_int
m_fixhdr(struct mbuf *m0)
{
	u_int len;

	len = m_length(m0, NULL);
	m0->m_pkthdr.len = len;
	return (len);
}

u_int
m_length(struct mbuf *m0, struct mbuf **last)
{
	struct mbuf *m;
	u_int len;

	len = 0;
	for (m = m0; m != NULL; m = m->m_next) {
		len += m->m_len;
		if (m->m_next == NULL)
			break;
	}
	if (last != NULL)
		*last = m;
	return (len);
}

/*
 * Defragment a mbuf chain, returning the shortest possible
 * chain of mbufs and clusters.  If allocation fails and
 * this cannot be completed, NULL will be returned, but
 * the passed in chain will be unchanged.  Upon success,
 * the original chain will be freed, and the new chain
 * will be returned.
 *
 * If a non-packet header is passed in, the original
 * mbuf (chain?) will be returned unharmed.
 */
struct mbuf *
m_defrag(struct mbuf *m0, int how)
{
	struct mbuf *m_new = NULL, *m_final = NULL;
	int progress = 0, length;

	MBUF_CHECKSLEEP(how);
	if (!(m0->m_flags & M_PKTHDR))
		return (m0);

	m_fixhdr(m0); /* Needed sanity check */

#ifdef MBUF_STRESS_TEST
	if (m_defragrandomfailures) {
		int temp = arc4random() & 0xff;
		if (temp == 0xba)
			goto nospace;
	}
#endif

	if (m0->m_pkthdr.len > MHLEN)
		m_final = m_getcl(how, MT_DATA, M_PKTHDR);
	else
		m_final = m_gethdr(how, MT_DATA);

	if (m_final == NULL)
		goto nospace;

	if (m_dup_pkthdr(m_final, m0, how) == 0)
		goto nospace;

	m_new = m_final;

	while (progress < m0->m_pkthdr.len) {
		length = m0->m_pkthdr.len - progress;
		if (length > MCLBYTES)
			length = MCLBYTES;

		if (m_new == NULL) {
			if (length > MLEN)
				m_new = m_getcl(how, MT_DATA, 0);
			else
				m_new = m_get(how, MT_DATA);
			if (m_new == NULL)
				goto nospace;
		}

		m_copydata(m0, progress, length, mtod(m_new, caddr_t));
		progress += length;
		m_new->m_len = length;
		if (m_new != m_final)
			m_cat(m_final, m_new);
		m_new = NULL;
	}
#ifdef MBUF_STRESS_TEST
	if (m0->m_next == NULL)
		m_defraguseless++;
#endif
	m_freem(m0);
	m0 = m_final;
#ifdef MBUF_STRESS_TEST
	m_defragpackets++;
	m_defragbytes += m0->m_pkthdr.len;
#endif
	return (m0);
nospace:
#ifdef MBUF_STRESS_TEST
	m_defragfailure++;
#endif
	if (m_final)
		m_freem(m_final);
	return (NULL);
}

/*
 * Defragment an mbuf chain, returning at most maxfrags separate
 * mbufs+clusters.  If this is not possible NULL is returned and
 * the original mbuf chain is left in it's present (potentially
 * modified) state.  We use two techniques: collapsing consecutive
 * mbufs and replacing consecutive mbufs by a cluster.
 *
 * NB: this should really be named m_defrag but that name is taken
 */
struct mbuf *
m_collapse(struct mbuf *m0, int how, int maxfrags)
{
	struct mbuf *m, *n, *n2, **prev;
	u_int curfrags;

	/*
	 * Calculate the current number of frags.
	 */
	curfrags = 0;
	for (m = m0; m != NULL; m = m->m_next)
		curfrags++;
	/*
	 * First, try to collapse mbufs.  Note that we always collapse
	 * towards the front so we don't need to deal with moving the
	 * pkthdr.  This may be suboptimal if the first mbuf has much
	 * less data than the following.
	 */
	m = m0;
again:
	for (;;) {
		n = m->m_next;
		if (n == NULL)
			break;
		if (M_WRITABLE(m) &&
		    n->m_len < M_TRAILINGSPACE(m)) {
			bcopy(mtod(n, void *), mtod(m, char *) + m->m_len,
				n->m_len);
			m->m_len += n->m_len;
			m->m_next = n->m_next;
			m_free(n);
			if (--curfrags <= maxfrags)
				return m0;
		} else
			m = n;
	}
	KASSERT(maxfrags > 1,
		("maxfrags %u, but normal collapse failed", maxfrags));
	/*
	 * Collapse consecutive mbufs to a cluster.
	 */
	prev = &m0->m_next;		/* NB: not the first mbuf */
	while ((n = *prev) != NULL) {
		if ((n2 = n->m_next) != NULL &&
		    n->m_len + n2->m_len < MCLBYTES) {
			m = m_getcl(how, MT_DATA, 0);
			if (m == NULL)
				goto bad;
			bcopy(mtod(n, void *), mtod(m, void *), n->m_len);
			bcopy(mtod(n2, void *), mtod(m, char *) + n->m_len,
				n2->m_len);
			m->m_len = n->m_len + n2->m_len;
			m->m_next = n2->m_next;
			*prev = m;
			m_free(n);
			m_free(n2);
			if (--curfrags <= maxfrags)	/* +1 cl -2 mbufs */
				return m0;
			/*
			 * Still not there, try the normal collapse
			 * again before we allocate another cluster.
			 */
			goto again;
		}
		prev = &n->m_next;
	}
	/*
	 * No place where we can collapse to a cluster; punt.
	 * This can occur if, for example, you request 2 frags
	 * but the packet requires that both be clusters (we
	 * never reallocate the first mbuf to avoid moving the
	 * packet header).
	 */
bad:
	return NULL;
}

#ifdef MBUF_STRESS_TEST

/*
 * Fragment an mbuf chain.  There's no reason you'd ever want to do
 * this in normal usage, but it's great for stress testing various
 * mbuf consumers.
 *
 * If fragmentation is not possible, the original chain will be
 * returned.
 *
 * Possible length values:
 * 0	 no fragmentation will occur
 * > 0	each fragment will be of the specified length
 * -1	each fragment will be the same random value in length
 * -2	each fragment's length will be entirely random
 * (Random values range from 1 to 256)
 */
struct mbuf *
m_fragment(struct mbuf *m0, int how, int length)
{
	struct mbuf *m_new = NULL, *m_final = NULL;
	int progress = 0;

	if (!(m0->m_flags & M_PKTHDR))
		return (m0);

	if ((length == 0) || (length < -2))
		return (m0);

	m_fixhdr(m0); /* Needed sanity check */

	m_final = m_getcl(how, MT_DATA, M_PKTHDR);

	if (m_final == NULL)
		goto nospace;

	if (m_dup_pkthdr(m_final, m0, how) == 0)
		goto nospace;

	m_new = m_final;

	if (length == -1)
		length = 1 + (arc4random() & 255);

	while (progress < m0->m_pkthdr.len) {
		int fraglen;

		if (length > 0)
			fraglen = length;
		else
			fraglen = 1 + (arc4random() & 255);
		if (fraglen > m0->m_pkthdr.len - progress)
			fraglen = m0->m_pkthdr.len - progress;

		if (fraglen > MCLBYTES)
			fraglen = MCLBYTES;

		if (m_new == NULL) {
			m_new = m_getcl(how, MT_DATA, 0);
			if (m_new == NULL)
				goto nospace;
		}

		m_copydata(m0, progress, fraglen, mtod(m_new, caddr_t));
		progress += fraglen;
		m_new->m_len = fraglen;
		if (m_new != m_final)
			m_cat(m_final, m_new);
		m_new = NULL;
	}
	m_freem(m0);
	m0 = m_final;
	return (m0);
nospace:
	if (m_final)
		m_freem(m_final);
	/* Return the original chain on failure */
	return (m0);
}

#endif

/*
 * Copy the contents of uio into a properly sized mbuf chain.
 */
struct mbuf *
m_uiotombuf(struct uio *uio, int how, int len, int align, int flags)
{
	struct mbuf *m, *mb;
	int error, length;
	ssize_t total;
	int progress = 0;

	/*
	 * len can be zero or an arbitrary large value bound by
	 * the total data supplied by the uio.
	 */
	if (len > 0)
		total = min(uio->uio_resid, len);
	else
		total = uio->uio_resid;

	/*
	 * The smallest unit returned by m_getm2() is a single mbuf
	 * with pkthdr.  We can't align past it.
	 */
	if (align >= MHLEN)
		return (NULL);

	/*
	 * Give us the full allocation or nothing.
	 * If len is zero return the smallest empty mbuf.
	 */
	m = m_getm2(NULL, max(total + align, 1), how, MT_DATA, flags);
	if (m == NULL)
		return (NULL);
	m->m_data += align;

	/* Fill all mbufs with uio data and update header information. */
	for (mb = m; mb != NULL; mb = mb->m_next) {
		length = min(M_TRAILINGSPACE(mb), total - progress);

		error = uiomove(mtod(mb, void *), length, uio);
		if (error) {
			m_freem(m);
			return (NULL);
		}

		mb->m_len = length;
		progress += length;
		if (flags & M_PKTHDR)
			m->m_pkthdr.len += length;
	}
	KASSERT(progress == total, ("%s: progress != total", __func__));

	return (m);
}

/*
 * Copy an mbuf chain into a uio limited by len if set.
 */
int
m_mbuftouio(struct uio *uio, struct mbuf *m, int len)
{
	int error, length, total;
	int progress = 0;

	if (len > 0)
		total = min(uio->uio_resid, len);
	else
		total = uio->uio_resid;

	/* Fill the uio with data from the mbufs. */
	for (; m != NULL; m = m->m_next) {
		length = min(m->m_len, total - progress);

		error = uiomove(mtod(m, void *), length, uio);
		if (error)
			return (error);

		progress += length;
	}

	return (0);
}

/*
 * Set the m_data pointer of a newly-allocated mbuf
 * to place an object of the specified size at the
 * end of the mbuf, longword aligned.
 */
void
m_align(struct mbuf *m, int len)
{
#ifdef INVARIANTS
	const char *msg = "%s: not a virgin mbuf";
#endif
	int adjust;

	if (m->m_flags & M_EXT) {
		KASSERT(m->m_data == m->m_ext.ext_buf, (msg, __func__));
		adjust = m->m_ext.ext_size - len;
	} else if (m->m_flags & M_PKTHDR) {
		KASSERT(m->m_data == m->m_pktdat, (msg, __func__));
		adjust = MHLEN - len;
	} else {
		KASSERT(m->m_data == m->m_dat, (msg, __func__));
		adjust = MLEN - len;
	}

	m->m_data += adjust &~ (sizeof(long)-1);
}

/*
 * Create a writable copy of the mbuf chain.  While doing this
 * we compact the chain with a goal of producing a chain with
 * at most two mbufs.  The second mbuf in this chain is likely
 * to be a cluster.  The primary purpose of this work is to create
 * a writable packet for encryption, compression, etc.  The
 * secondary goal is to linearize the data so the data can be
 * passed to crypto hardware in the most efficient manner possible.
 */
struct mbuf *
m_unshare(struct mbuf *m0, int how)
{
	struct mbuf *m, *mprev;
	struct mbuf *n, *mfirst, *mlast;
	int len, off;

	mprev = NULL;
	for (m = m0; m != NULL; m = mprev->m_next) {
		/*
		 * Regular mbufs are ignored unless there's a cluster
		 * in front of it that we can use to coalesce.  We do
		 * the latter mainly so later clusters can be coalesced
		 * also w/o having to handle them specially (i.e. convert
		 * mbuf+cluster -> cluster).  This optimization is heavily
		 * influenced by the assumption that we're running over
		 * Ethernet where MCLBYTES is large enough that the max
		 * packet size will permit lots of coalescing into a
		 * single cluster.  This in turn permits efficient
		 * crypto operations, especially when using hardware.
		 */
		if ((m->m_flags & M_EXT) == 0) {
			if (mprev && (mprev->m_flags & M_EXT) &&
			    m->m_len <= M_TRAILINGSPACE(mprev)) {
				/* XXX: this ignores mbuf types */
				memcpy(mtod(mprev, caddr_t) + mprev->m_len,
				    mtod(m, caddr_t), m->m_len);
				mprev->m_len += m->m_len;
				mprev->m_next = m->m_next;	/* unlink from chain */
				m_free(m);			/* reclaim mbuf */
#if 0
				newipsecstat.ips_mbcoalesced++;
#endif
			} else {
				mprev = m;
			}
			continue;
		}
		/*
		 * Writable mbufs are left alone (for now).
		 */
		if (M_WRITABLE(m)) {
			mprev = m;
			continue;
		}

		/*
		 * Not writable, replace with a copy or coalesce with
		 * the previous mbuf if possible (since we have to copy
		 * it anyway, we try to reduce the number of mbufs and
		 * clusters so that future work is easier).
		 */
		KASSERT(m->m_flags & M_EXT, ("m_flags 0x%x", m->m_flags));
		/* NB: we only coalesce into a cluster or larger */
		if (mprev != NULL && (mprev->m_flags & M_EXT) &&
		    m->m_len <= M_TRAILINGSPACE(mprev)) {
			/* XXX: this ignores mbuf types */
			memcpy(mtod(mprev, caddr_t) + mprev->m_len,
			    mtod(m, caddr_t), m->m_len);
			mprev->m_len += m->m_len;
			mprev->m_next = m->m_next;	/* unlink from chain */
			m_free(m);			/* reclaim mbuf */
#if 0
			newipsecstat.ips_clcoalesced++;
#endif
			continue;
		}

		/*
		 * Allocate new space to hold the copy and copy the data.
		 * We deal with jumbo mbufs (i.e. m_len > MCLBYTES) by
		 * splitting them into clusters.  We could just malloc a
		 * buffer and make it external but too many device drivers
		 * don't know how to break up the non-contiguous memory when
		 * doing DMA.
		 */
		n = m_getcl(how, m->m_type, m->m_flags);
		if (n == NULL) {
			m_freem(m0);
			return (NULL);
		}
		len = m->m_len;
		off = 0;
		mfirst = n;
		mlast = NULL;
		for (;;) {
			int cc = min(len, MCLBYTES);
			memcpy(mtod(n, caddr_t), mtod(m, caddr_t) + off, cc);
			n->m_len = cc;
			if (mlast != NULL)
				mlast->m_next = n;
			mlast = n;
#if 0
			newipsecstat.ips_clcopied++;
#endif

			len -= cc;
			if (len <= 0)
				break;
			off += cc;

			n = m_getcl(how, m->m_type, m->m_flags);
			if (n == NULL) {
				m_freem(mfirst);
				m_freem(m0);
				return (NULL);
			}
		}
		n->m_next = m->m_next;
		if (mprev == NULL)
			m0 = mfirst;		/* new head of chain */
		else
			mprev->m_next = mfirst;	/* replace old mbuf */
		m_free(m);			/* release old mbuf */
		mprev = mfirst;
	}
	return (m0);
}

#ifdef MBUF_PROFILING

#define MP_BUCKETS 32 /* don't just change this as things may overflow.*/
struct mbufprofile {
	uintmax_t wasted[MP_BUCKETS];
	uintmax_t used[MP_BUCKETS];
	uintmax_t segments[MP_BUCKETS];
} mbprof;

#define MP_MAXDIGITS 21	/* strlen("16,000,000,000,000,000,000") == 21 */
#define MP_NUMLINES 6
#define MP_NUMSPERLINE 16
#define MP_EXTRABYTES 64	/* > strlen("used:\nwasted:\nsegments:\n") */
/* work out max space needed and add a bit of spare space too */
#define MP_MAXLINE ((MP_MAXDIGITS+1) * MP_NUMSPERLINE)
#define MP_BUFSIZE ((MP_MAXLINE * MP_NUMLINES) + 1 + MP_EXTRABYTES)

char mbprofbuf[MP_BUFSIZE];

void
m_profile(struct mbuf *m)
{
	int segments = 0;
	int used = 0;
	int wasted = 0;

	while (m) {
		segments++;
		used += m->m_len;
		if (m->m_flags & M_EXT) {
			wasted += MHLEN - sizeof(m->m_ext) +
			    m->m_ext.ext_size - m->m_len;
		} else {
			if (m->m_flags & M_PKTHDR)
				wasted += MHLEN - m->m_len;
			else
				wasted += MLEN - m->m_len;
		}
		m = m->m_next;
	}
	/* be paranoid.. it helps */
	if (segments > MP_BUCKETS - 1)
		segments = MP_BUCKETS - 1;
	if (used > 100000)
		used = 100000;
	if (wasted > 100000)
		wasted = 100000;
	/* store in the appropriate bucket */
	/* don't bother locking. if it's slightly off, so what? */
	mbprof.segments[segments]++;
	mbprof.used[fls(used)]++;
	mbprof.wasted[fls(wasted)]++;
}

static void
mbprof_textify(void)
{
	int offset;
	char *c;
	uint64_t *p;

	p = &mbprof.wasted[0];
	c = mbprofbuf;
	offset = snprintf(c, MP_MAXLINE + 10,
	    "wasted:\n"
	    "%ju %ju %ju %ju %ju %ju %ju %ju "
	    "%ju %ju %ju %ju %ju %ju %ju %ju\n",
	    p[0], p[1], p[2], p[3], p[4], p[5], p[6], p[7],
	    p[8], p[9], p[10], p[11], p[12], p[13], p[14], p[15]);
#ifdef BIG_ARRAY
	p = &mbprof.wasted[16];
	c += offset;
	offset = snprintf(c, MP_MAXLINE,
	    "%ju %ju %ju %ju %ju %ju %ju %ju "
	    "%ju %ju %ju %ju %ju %ju %ju %ju\n",
	    p[0], p[1], p[2], p[3], p[4], p[5], p[6], p[7],
	    p[8], p[9], p[10], p[11], p[12], p[13], p[14], p[15]);
#endif
	p = &mbprof.used[0];
	c += offset;
	offset = snprintf(c, MP_MAXLINE + 10,
	    "used:\n"
	    "%ju %ju %ju %ju %ju %ju %ju %ju "
	    "%ju %ju %ju %ju %ju %ju %ju %ju\n",
	    p[0], p[1], p[2], p[3], p[4], p[5], p[6], p[7],
	    p[8], p[9], p[10], p[11], p[12], p[13], p[14], p[15]);
#ifdef BIG_ARRAY
	p = &mbprof.used[16];
	c += offset;
	offset = snprintf(c, MP_MAXLINE,
	    "%ju %ju %ju %ju %ju %ju %ju %ju "
	    "%ju %ju %ju %ju %ju %ju %ju %ju\n",
	    p[0], p[1], p[2], p[3], p[4], p[5], p[6], p[7],
	    p[8], p[9], p[10], p[11], p[12], p[13], p[14], p[15]);
#endif
	p = &mbprof.segments[0];
	c += offset;
	offset = snprintf(c, MP_MAXLINE + 10,
	    "segments:\n"
	    "%ju %ju %ju %ju %ju %ju %ju %ju "
	    "%ju %ju %ju %ju %ju %ju %ju %ju\n",
	    p[0], p[1], p[2], p[3], p[4], p[5], p[6], p[7],
	    p[8], p[9], p[10], p[11], p[12], p[13], p[14], p[15]);
#ifdef BIG_ARRAY
	p = &mbprof.segments[16];
	c += offset;
	offset = snprintf(c, MP_MAXLINE,
	    "%ju %ju %ju %ju %ju %ju %ju %ju "
	    "%ju %ju %ju %ju %ju %ju %ju %jju",
	    p[0], p[1], p[2], p[3], p[4], p[5], p[6], p[7],
	    p[8], p[9], p[10], p[11], p[12], p[13], p[14], p[15]);
#endif
}

static int
mbprof_handler(SYSCTL_HANDLER_ARGS)
{
	int error;

	mbprof_textify();
	error = SYSCTL_OUT(req, mbprofbuf, strlen(mbprofbuf) + 1);
	return (error);
}

static int
mbprof_clr_handler(SYSCTL_HANDLER_ARGS)
{
	int clear, error;

	clear = 0;
	error = sysctl_handle_int(oidp, &clear, 0, req);
	if (error || !req->newptr)
		return (error);

	if (clear) {
		bzero(&mbprof, sizeof(mbprof));
	}

	return (error);
}


SYSCTL_PROC(_kern_ipc, OID_AUTO, mbufprofile, CTLTYPE_STRING|CTLFLAG_RD,
	    NULL, 0, mbprof_handler, "A", "mbuf profiling statistics");

SYSCTL_PROC(_kern_ipc, OID_AUTO, mbufprofileclr, CTLTYPE_INT|CTLFLAG_RW,
	    NULL, 0, mbprof_clr_handler, "I", "clear mbuf profiling statistics");
#endif