summaryrefslogtreecommitdiff
path: root/sys/kern/sys_pipe.c
blob: e50f1b2464e46a3bdf021439d24a6a076e7f1e0d (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
/*	$NetBSD: sys_pipe.c,v 1.21 2001/12/18 08:49:40 chs Exp $	*/

/*
 * Copyright (c) 1996 John S. Dyson
 * 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 immediately at the beginning of the file, without modification,
 *    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.
 * 3. Absolutely no warranty of function or purpose is made by the author
 *    John S. Dyson.
 * 4. Modifications may be freely made to this file if the above conditions
 *    are met.
 *
 * $FreeBSD: src/sys/kern/sys_pipe.c,v 1.82 2001/06/15 20:45:01 jlemon Exp $
 */

/*
 * This file contains a high-performance replacement for the socket-based
 * pipes scheme originally used in FreeBSD/4.4Lite.  It does not support
 * all features of sockets, but does do everything that pipes normally
 * do.
 *
 * Adaption for NetBSD UVM, including uvm_loan() based direct write, was
 * written by Jaromir Dolecek.
 */

/*
 * This code has two modes of operation, a small write mode and a large
 * write mode.  The small write mode acts like conventional pipes with
 * a kernel buffer.  If the buffer is less than PIPE_MINDIRECT, then the
 * "normal" pipe buffering is done.  If the buffer is between PIPE_MINDIRECT
 * and PIPE_SIZE in size, it is fully mapped into the kernel (on FreeBSD,
 * those pages are also wired), and the receiving process can copy it directly
 * from the pages in the sending process.
 *
 * If the sending process receives a signal, it is possible that it will
 * go away, and certainly its address space can change, because control
 * is returned back to the user-mode side.  In that case, the pipe code
 * arranges to copy the buffer supplied by the user process on FreeBSD, to
 * a pageable kernel buffer, and the receiving process will grab the data
 * from the pageable kernel buffer.  Since signals don't happen all that often,
 * the copy operation is normally eliminated.
 * For NetBSD, the pages are mapped read-only, COW for kernel by uvm_loan(),
 * so no explicit handling need to be done, all is handled by standard VM
 * facilities.
 *
 * The constant PIPE_MINDIRECT is chosen to make sure that buffering will
 * happen for small transfers so that the system will not spend all of
 * its time context switching.  PIPE_SIZE is constrained by the
 * amount of kernel virtual memory.
 */

#include <sys/cdefs.h>
__KERNEL_RCSID(0, "$NetBSD: sys_pipe.c,v 1.21 2001/12/18 08:49:40 chs Exp $");

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/proc.h>
#include <sys/fcntl.h>
#include <sys/file.h>
#include <sys/filedesc.h>
#include <sys/filio.h>
#include <sys/ttycom.h>
#include <sys/stat.h>
#include <sys/poll.h>
#include <sys/signalvar.h>
#include <sys/vnode.h>
#include <sys/uio.h>
#include <sys/lock.h>
#ifdef __FreeBSD__
#include <sys/mutex.h>
#include <sys/selinfo.h>
#include <sys/sysproto.h>
#elif defined(__NetBSD__)
#include <sys/select.h>
#include <sys/malloc.h>
#include <sys/mount.h>
#include <sys/syscallargs.h>
#include <uvm/uvm.h>
#include <sys/sysctl.h>
#include <sys/kernel.h>
#endif /* NetBSD, FreeBSD */

#include <sys/pipe.h>

#ifdef __NetBSD__
/*
 * Avoid microtime(9), it's slow. We don't guard the read from time(9)
 * with splclock(9) since we don't actually need to be THAT sure the access
 * is atomic.
 */
#define vfs_timestamp(tv)	(*(tv) = time)
#endif

/*
 * Use this define if you want to disable *fancy* VM things.  Expect an
 * approx 30% decrease in transfer rate.  This could be useful for
 * OpenBSD.
 */
/* #define PIPE_NODIRECT */

/*
 * interfaces to the outside world
 */
#ifdef __FreeBSD__
static int pipe_read __P((struct file *fp, struct uio *uio, 
		struct ucred *cred, int flags, struct proc *p));
static int pipe_write __P((struct file *fp, struct uio *uio, 
		struct ucred *cred, int flags, struct proc *p));
static int pipe_close __P((struct file *fp, struct proc *p));
static int pipe_poll __P((struct file *fp, int events, struct ucred *cred,
		struct proc *p));
static int pipe_kqfilter __P((struct file *fp, struct knote *kn));
static int pipe_stat __P((struct file *fp, struct stat *sb, struct proc *p));
static int pipe_ioctl __P((struct file *fp, u_long cmd, caddr_t data, struct proc *p));

static struct fileops pipeops = {
	pipe_read, pipe_write, pipe_ioctl, pipe_poll, pipe_kqfilter,
	pipe_stat, pipe_close
};

static void	filt_pipedetach(struct knote *kn);
static int	filt_piperead(struct knote *kn, long hint);
static int	filt_pipewrite(struct knote *kn, long hint);

static struct filterops pipe_rfiltops =
	{ 1, NULL, filt_pipedetach, filt_piperead };
static struct filterops pipe_wfiltops =
	{ 1, NULL, filt_pipedetach, filt_pipewrite };
#endif /* FreeBSD */

#ifdef __NetBSD__
static int pipe_read __P((struct file *fp, off_t *offset, struct uio *uio, 
		struct ucred *cred, int flags));
static int pipe_write __P((struct file *fp, off_t *offset, struct uio *uio, 
		struct ucred *cred, int flags));
static int pipe_close __P((struct file *fp, struct proc *p));
static int pipe_poll __P((struct file *fp, int events, struct proc *p));
static int pipe_fcntl __P((struct file *fp, u_int com, caddr_t data,
		struct proc *p));
static int pipe_stat __P((struct file *fp, struct stat *sb, struct proc *p));
static int pipe_ioctl __P((struct file *fp, u_long cmd, caddr_t data, struct proc *p));

static struct fileops pipeops =
    { pipe_read, pipe_write, pipe_ioctl, pipe_fcntl, pipe_poll,
      pipe_stat, pipe_close };
#endif /* NetBSD */

/*
 * Default pipe buffer size(s), this can be kind-of large now because pipe
 * space is pageable.  The pipe code will try to maintain locality of
 * reference for performance reasons, so small amounts of outstanding I/O
 * will not wipe the cache.
 */
#define MINPIPESIZE (PIPE_SIZE/3)
#define MAXPIPESIZE (2*PIPE_SIZE/3)

/*
 * Maximum amount of kva for pipes -- this is kind-of a soft limit, but
 * is there so that on large systems, we don't exhaust it.
 */
#define MAXPIPEKVA (8*1024*1024)
static int maxpipekva = MAXPIPEKVA;

/*
 * Limit for direct transfers, we cannot, of course limit
 * the amount of kva for pipes in general though.
 */
#define LIMITPIPEKVA (16*1024*1024)
static int limitpipekva = LIMITPIPEKVA;

/*
 * Limit the number of "big" pipes
 */
#define LIMITBIGPIPES  32
static int maxbigpipes = LIMITBIGPIPES;
static int nbigpipe = 0;

/*
 * Amount of KVA consumed by pipe buffers.
 */
static int amountpipekva = 0;

static void pipeclose __P((struct pipe *));
static void pipe_free_kmem __P((struct pipe *));
static int pipe_create __P((struct pipe **, int));
static __inline int pipelock __P((struct pipe *, int));
static __inline void pipeunlock __P((struct pipe *));
static __inline void pipeselwakeup __P((struct pipe *, struct pipe *));
static int pipespace __P((struct pipe *, int));

#ifdef __FreeBSD__
#ifndef PIPE_NODIRECT
static int pipe_build_write_buffer __P((struct pipe *wpipe, struct uio *uio));
static void pipe_destroy_write_buffer __P((struct pipe *wpipe));
static int pipe_direct_write __P((struct pipe *wpipe, struct uio *uio));
static void pipe_clone_write_buffer __P((struct pipe *wpipe));
#endif

static vm_zone_t pipe_zone;
#endif /* FreeBSD */

#ifdef __NetBSD__
#ifndef PIPE_NODIRECT
static int pipe_direct_write __P((struct pipe *, struct uio *));
static int pipe_loan_alloc __P((struct pipe *, int));
static void pipe_loan_free __P((struct pipe *));
#endif /* PIPE_NODIRECT */

static struct pool pipe_pool;
#endif /* NetBSD */

/*
 * The pipe system call for the DTYPE_PIPE type of pipes
 */

/* ARGSUSED */
#ifdef __FreeBSD__
int
pipe(p, uap)
	struct proc *p;
	struct pipe_args /* {
		int	dummy;
	} */ *uap;
#elif defined(__NetBSD__)
int
sys_pipe(p, v, retval)
	struct proc *p;
	void *v;
	register_t *retval;
#endif
{
	struct file *rf, *wf;
	struct pipe *rpipe, *wpipe;
	int fd, error;

#ifdef __FreeBSD__
	if (pipe_zone == NULL)
		pipe_zone = zinit("PIPE", sizeof(struct pipe), 0, 0, 4);

	rpipe = wpipe = NULL;
	if (pipe_create(&rpipe, 1) || pipe_create(&wpipe, 1)) {
		pipeclose(rpipe);
		pipeclose(wpipe);
		return (ENFILE);
	}

	error = falloc(p, &rf, &fd);
	if (error) {
		pipeclose(rpipe);
		pipeclose(wpipe);
		return (error);
	}
	fhold(rf);
	p->p_retval[0] = fd;

	/*
	 * Warning: once we've gotten past allocation of the fd for the
	 * read-side, we can only drop the read side via fdrop() in order
	 * to avoid races against processes which manage to dup() the read
	 * side while we are blocked trying to allocate the write side.
	 */
	rf->f_flag = FREAD | FWRITE;
	rf->f_type = DTYPE_PIPE;
	rf->f_data = (caddr_t)rpipe;
	rf->f_ops = &pipeops;
	error = falloc(p, &wf, &fd);
	if (error) {
		struct filedesc *fdp = p->p_fd;

		if (fdp->fd_ofiles[p->p_retval[0]] == rf) {
			fdp->fd_ofiles[p->p_retval[0]] = NULL;
			fdrop(rf, p);
		}
		fdrop(rf, p);
		/* rpipe has been closed by fdrop(). */
		pipeclose(wpipe);
		return (error);
	}
	wf->f_flag = FREAD | FWRITE;
	wf->f_type = DTYPE_PIPE;
	wf->f_data = (caddr_t)wpipe;
	wf->f_ops = &pipeops;
	p->p_retval[1] = fd;

	rpipe->pipe_peer = wpipe;
	wpipe->pipe_peer = rpipe;
	fdrop(rf, p);
#endif /* FreeBSD */

#ifdef __NetBSD__
	rpipe = wpipe = NULL;
	if (pipe_create(&rpipe, 1) || pipe_create(&wpipe, 0)) {
		pipeclose(rpipe);
		pipeclose(wpipe);
		return (ENFILE);
	}

	/*
	 * Note: the file structure returned from falloc() is marked
	 * as 'larval' initially. Unless we mark it as 'mature' by
	 * FILE_SET_MATURE(), any attempt to do anything with it would
	 * return EBADF, including e.g. dup(2) or close(2). This avoids
	 * file descriptor races if we block in the second falloc().
	 */

	error = falloc(p, &rf, &fd);
	if (error)
		goto free2;
	retval[0] = fd;
	rf->f_flag = FREAD;
	rf->f_type = DTYPE_PIPE;
	rf->f_data = (caddr_t)rpipe;
	rf->f_ops = &pipeops;

	error = falloc(p, &wf, &fd);
	if (error)
		goto free3;
	retval[1] = fd;
	wf->f_flag = FWRITE;
	wf->f_type = DTYPE_PIPE;
	wf->f_data = (caddr_t)wpipe;
	wf->f_ops = &pipeops;

	rpipe->pipe_peer = wpipe;
	wpipe->pipe_peer = rpipe;

	FILE_SET_MATURE(rf);
	FILE_SET_MATURE(wf);
	FILE_UNUSE(rf, p);
	FILE_UNUSE(wf, p);
	return (0);
free3:
	FILE_UNUSE(rf, p);
	ffree(rf);
	fdremove(p->p_fd, retval[0]);
free2:
	pipeclose(wpipe);
	pipeclose(rpipe);
#endif /* NetBSD */

	return (error);
}

/*
 * Allocate kva for pipe circular buffer, the space is pageable
 * This routine will 'realloc' the size of a pipe safely, if it fails
 * it will retain the old buffer.
 * If it fails it will return ENOMEM.
 */
static int
pipespace(cpipe, size)
	struct pipe *cpipe;
	int size;
{
	caddr_t buffer;
#ifdef __FreeBSD__
	struct vm_object *object;
	int npages, error;

	npages = round_page(size)/PAGE_SIZE;
	/*
	 * Create an object, I don't like the idea of paging to/from
	 * kernel_object.
	 */
	mtx_lock(&vm_mtx);
	object = vm_object_allocate(OBJT_DEFAULT, npages);
	buffer = (caddr_t) vm_map_min(kernel_map);

	/*
	 * Insert the object into the kernel map, and allocate kva for it.
	 * The map entry is, by default, pageable.
	 */
	error = vm_map_find(kernel_map, object, 0,
		(vm_offset_t *) &buffer, size, 1,
		VM_PROT_ALL, VM_PROT_ALL, 0);

	if (error != KERN_SUCCESS) {
		vm_object_deallocate(object);
		mtx_unlock(&vm_mtx);
		return (ENOMEM);
	}
#endif /* FreeBSD */

#ifdef __NetBSD__
	/*
	 * Allocate pageable virtual address space. Physical memory is allocated
	 * on demand.
	 */
	buffer = (caddr_t) uvm_km_valloc(kernel_map, round_page(size));
	if (buffer == NULL)
		return (ENOMEM);
#endif /* NetBSD */

	/* free old resources if we're resizing */
	pipe_free_kmem(cpipe);
#ifdef __FreeBSD__
	mtx_unlock(&vm_mtx);
	cpipe->pipe_buffer.object = object;
#endif
	cpipe->pipe_buffer.buffer = buffer;
	cpipe->pipe_buffer.size = size;
	cpipe->pipe_buffer.in = 0;
	cpipe->pipe_buffer.out = 0;
	cpipe->pipe_buffer.cnt = 0;
	amountpipekva += cpipe->pipe_buffer.size;
	return (0);
}

/*
 * initialize and allocate VM and memory for pipe
 */
static int
pipe_create(cpipep, allockva)
	struct pipe **cpipep;
	int allockva;
{
	struct pipe *cpipe;
	int error;

#ifdef __FreeBSD__
	*cpipep = zalloc(pipe_zone);
#endif
#ifdef __NetBSD__
	*cpipep = pool_get(&pipe_pool, M_WAITOK);
#endif
	if (*cpipep == NULL)
		return (ENOMEM);

	cpipe = *cpipep;

	/* Initialize */ 
	memset(cpipe, 0, sizeof(*cpipe));
	cpipe->pipe_state = PIPE_SIGNALR;

	if (allockva && (error = pipespace(cpipe, PIPE_SIZE)))
		return (error);

	vfs_timestamp(&cpipe->pipe_ctime);
	cpipe->pipe_atime = cpipe->pipe_ctime;
	cpipe->pipe_mtime = cpipe->pipe_ctime;
#ifdef __NetBSD__
	cpipe->pipe_pgid = NO_PID;
	lockinit(&cpipe->pipe_lock, PRIBIO | PCATCH, "pipelk", 0, 0);
#endif

	return (0);
}


/*
 * lock a pipe for I/O, blocking other access
 */
static __inline int
pipelock(cpipe, catch)
	struct pipe *cpipe;
	int catch;
{
	int error;

#ifdef __FreeBSD__
	while (cpipe->pipe_state & PIPE_LOCK) {
		cpipe->pipe_state |= PIPE_LWANT;
		error = tsleep(cpipe, catch ? (PRIBIO | PCATCH) : PRIBIO,
		    "pipelk", 0);
		if (error != 0)
			return (error);
	}
	cpipe->pipe_state |= PIPE_LOCK;
	return (0);
#endif

#ifdef __NetBSD__
	do {
		error = lockmgr(&cpipe->pipe_lock, LK_EXCLUSIVE, NULL);
	} while (!catch && (error == EINTR || error == ERESTART));
	return (error);
#endif
}

/*
 * unlock a pipe I/O lock
 */
static __inline void
pipeunlock(cpipe)
	struct pipe *cpipe;
{
#ifdef __FreeBSD__
	cpipe->pipe_state &= ~PIPE_LOCK;
	if (cpipe->pipe_state & PIPE_LWANT) {
		cpipe->pipe_state &= ~PIPE_LWANT;
		wakeup(cpipe);
	}
#endif

#ifdef __NetBSD__
	lockmgr(&cpipe->pipe_lock, LK_RELEASE, NULL);
#endif
}

/*
 * Select/poll wakup. This also sends SIGIO to peer connected to
 * 'sigpipe' side of pipe.
 */
static __inline void
pipeselwakeup(selp, sigp)
	struct pipe *selp, *sigp;
{
	if (selp->pipe_state & PIPE_SEL) {
		selp->pipe_state &= ~PIPE_SEL;
		selwakeup(&selp->pipe_sel);
	}
#ifdef __FreeBSD__
	if (sigp && (sigp->pipe_state & PIPE_ASYNC) && sigp->pipe_sigio)
		pgsigio(sigp->pipe_sigio, SIGIO, 0);
	KNOTE(&selp->pipe_sel.si_note, 0);
#endif

#ifdef __NetBSD__
	if (sigp && (sigp->pipe_state & PIPE_ASYNC)
	    && sigp->pipe_pgid != NO_PID){
		struct proc *p;

		if (sigp->pipe_pgid < 0)
			gsignal(-sigp->pipe_pgid, SIGIO);
		else if (sigp->pipe_pgid > 0 && (p = pfind(sigp->pipe_pgid)) != 0)
			psignal(p, SIGIO);
	}
#endif /* NetBSD */
}

/* ARGSUSED */
#ifdef __FreeBSD__
static int
pipe_read(fp, uio, cred, flags, p)
	struct file *fp;
	struct uio *uio;
	struct ucred *cred;
	int flags;
	struct proc *p;
#elif defined(__NetBSD__)
static int
pipe_read(fp, offset, uio, cred, flags)
	struct file *fp;
	off_t *offset;
	struct uio *uio;
	struct ucred *cred;
	int flags;
#endif
{
	struct pipe *rpipe = (struct pipe *) fp->f_data;
	int error;
	size_t nread = 0;
	size_t size;
	size_t ocnt;

	++rpipe->pipe_busy;
	error = pipelock(rpipe, 1);
	if (error)
		goto unlocked_error;

	ocnt = rpipe->pipe_buffer.cnt;

	while (uio->uio_resid) {
		/*
		 * normal pipe buffer receive
		 */
		if (rpipe->pipe_buffer.cnt > 0) {
			size = rpipe->pipe_buffer.size - rpipe->pipe_buffer.out;
			if (size > rpipe->pipe_buffer.cnt)
				size = rpipe->pipe_buffer.cnt;
			if (size > uio->uio_resid)
				size = uio->uio_resid;

			error = uiomove(&rpipe->pipe_buffer.buffer[rpipe->pipe_buffer.out],
					size, uio);
			if (error)
				break;

			rpipe->pipe_buffer.out += size;
			if (rpipe->pipe_buffer.out >= rpipe->pipe_buffer.size)
				rpipe->pipe_buffer.out = 0;

			rpipe->pipe_buffer.cnt -= size;

			/*
			 * If there is no more to read in the pipe, reset
			 * its pointers to the beginning.  This improves
			 * cache hit stats.
			 */
			if (rpipe->pipe_buffer.cnt == 0) {
				rpipe->pipe_buffer.in = 0;
				rpipe->pipe_buffer.out = 0;
			}
			nread += size;
#ifndef PIPE_NODIRECT
		/*
		 * Direct copy, bypassing a kernel buffer.
		 */
		} else if ((size = rpipe->pipe_map.cnt) &&
			   (rpipe->pipe_state & PIPE_DIRECTW)) {
			caddr_t	va;
			if (size > uio->uio_resid)
				size = uio->uio_resid;

			va = (caddr_t) rpipe->pipe_map.kva +
			    rpipe->pipe_map.pos;
			error = uiomove(va, size, uio);
			if (error)
				break;
			nread += size;
			rpipe->pipe_map.pos += size;
			rpipe->pipe_map.cnt -= size;
			if (rpipe->pipe_map.cnt == 0) {
				rpipe->pipe_state &= ~PIPE_DIRECTW;
				wakeup(rpipe);
			}
#endif
		} else {
			/*
			 * detect EOF condition
			 * read returns 0 on EOF, no need to set error
			 */
			if (rpipe->pipe_state & PIPE_EOF)
				break;

			/*
			 * If the "write-side" has been blocked, wake it up now.
			 */
			if (rpipe->pipe_state & PIPE_WANTW) {
				rpipe->pipe_state &= ~PIPE_WANTW;
				wakeup(rpipe);
			}

			/*
			 * Break if some data was read.
			 */
			if (nread > 0)
				break;

			/*
			 * don't block on non-blocking I/O
			 */
			if (fp->f_flag & FNONBLOCK) {
				error = EAGAIN;
				break;
			}

			/*
			 * Unlock the pipe buffer for our remaining processing.
			 * We will either break out with an error or we will
			 * sleep and relock to loop.
			 */
			pipeunlock(rpipe);

			/*
			 * We want to read more, wake up select/poll.
			 */
			pipeselwakeup(rpipe, rpipe->pipe_peer);

			rpipe->pipe_state |= PIPE_WANTR;
			error = tsleep(rpipe, PRIBIO | PCATCH, "piperd", 0);
			if (error != 0 || (error = pipelock(rpipe, 1)))
				goto unlocked_error;
		}
	}
	pipeunlock(rpipe);

	if (error == 0)
		vfs_timestamp(&rpipe->pipe_atime);
unlocked_error:
	--rpipe->pipe_busy;

	/*
	 * PIPE_WANTCLOSE processing only makes sense if pipe_busy is 0.
	 */
	if ((rpipe->pipe_busy == 0) && (rpipe->pipe_state & PIPE_WANTCLOSE)) {
		rpipe->pipe_state &= ~(PIPE_WANTCLOSE|PIPE_WANTW);
		wakeup(rpipe);
	} else if (rpipe->pipe_buffer.cnt < MINPIPESIZE) {
		/*
		 * Handle write blocking hysteresis.
		 */
		if (rpipe->pipe_state & PIPE_WANTW) {
			rpipe->pipe_state &= ~PIPE_WANTW;
			wakeup(rpipe);
		}
	}

	/*
	 * If anything was read off the buffer, signal to the writer it's
	 * possible to write more data. Also send signal if we are here for the
	 * first time after last write.
	 */
	if ((rpipe->pipe_buffer.size - rpipe->pipe_buffer.cnt) >= PIPE_BUF
	    && (ocnt != rpipe->pipe_buffer.cnt || (rpipe->pipe_state & PIPE_SIGNALR))) {
		pipeselwakeup(rpipe, rpipe->pipe_peer);
		rpipe->pipe_state &= ~PIPE_SIGNALR;
	}

	return (error);
}

#ifdef __FreeBSD__
#ifndef PIPE_NODIRECT
/*
 * Map the sending processes' buffer into kernel space and wire it.
 * This is similar to a physical write operation.
 */
static int
pipe_build_write_buffer(wpipe, uio)
	struct pipe *wpipe;
	struct uio *uio;
{
	size_t size;
	int i;
	vm_offset_t addr, endaddr, paddr;

	size = uio->uio_iov->iov_len;
	if (size > wpipe->pipe_buffer.size)
		size = wpipe->pipe_buffer.size;

	endaddr = round_page((vm_offset_t)uio->uio_iov->iov_base + size);
	mtx_lock(&vm_mtx);
	addr = trunc_page((vm_offset_t)uio->uio_iov->iov_base);
	for (i = 0; addr < endaddr; addr += PAGE_SIZE, i++) {
		vm_page_t m;

		if (vm_fault_quick((caddr_t)addr, VM_PROT_READ) < 0 ||
		    (paddr = pmap_kextract(addr)) == 0) {
			int j;

			for (j = 0; j < i; j++)
				vm_page_unwire(wpipe->pipe_map.ms[j], 1);
			mtx_unlock(&vm_mtx);
			return (EFAULT);
		}

		m = PHYS_TO_VM_PAGE(paddr);
		vm_page_wire(m);
		wpipe->pipe_map.ms[i] = m;
	}

/*
 * set up the control block
 */
	wpipe->pipe_map.npages = i;
	wpipe->pipe_map.pos =
	    ((vm_offset_t) uio->uio_iov->iov_base) & PAGE_MASK;
	wpipe->pipe_map.cnt = size;

/*
 * and map the buffer
 */
	if (wpipe->pipe_map.kva == 0) {
		/*
		 * We need to allocate space for an extra page because the
		 * address range might (will) span pages at times.
		 */
		wpipe->pipe_map.kva = kmem_alloc_pageable(kernel_map,
			wpipe->pipe_buffer.size + PAGE_SIZE);
		amountpipekva += wpipe->pipe_buffer.size + PAGE_SIZE;
	}
	pmap_qenter(wpipe->pipe_map.kva, wpipe->pipe_map.ms,
		wpipe->pipe_map.npages);

	mtx_unlock(&vm_mtx);
/*
 * and update the uio data
 */

	uio->uio_iov->iov_len -= size;
	uio->uio_iov->iov_base += size;
	if (uio->uio_iov->iov_len == 0)
		uio->uio_iov++;
	uio->uio_resid -= size;
	uio->uio_offset += size;
	return (0);
}

/*
 * unmap and unwire the process buffer
 */
static void
pipe_destroy_write_buffer(wpipe)
	struct pipe *wpipe;
{
	int i;

	mtx_lock(&vm_mtx);
	if (wpipe->pipe_map.kva) {
		pmap_qremove(wpipe->pipe_map.kva, wpipe->pipe_map.npages);

		if (amountpipekva > maxpipekva) {
			vm_offset_t kva = wpipe->pipe_map.kva;
			wpipe->pipe_map.kva = 0;
			kmem_free(kernel_map, kva,
				wpipe->pipe_buffer.size + PAGE_SIZE);
			amountpipekva -= wpipe->pipe_buffer.size + PAGE_SIZE;
		}
	}
	for (i = 0; i < wpipe->pipe_map.npages; i++)
		vm_page_unwire(wpipe->pipe_map.ms[i], 1);
	mtx_unlock(&vm_mtx);
}

/*
 * In the case of a signal, the writing process might go away.  This
 * code copies the data into the circular buffer so that the source
 * pages can be freed without loss of data.
 */
static void
pipe_clone_write_buffer(wpipe)
	struct pipe *wpipe;
{
	int size;
	int pos;

	size = wpipe->pipe_map.cnt;
	pos = wpipe->pipe_map.pos;
	memcpy((caddr_t) wpipe->pipe_buffer.buffer,
	    (caddr_t) wpipe->pipe_map.kva + pos, size);

	wpipe->pipe_buffer.in = size;
	wpipe->pipe_buffer.out = 0;
	wpipe->pipe_buffer.cnt = size;
	wpipe->pipe_state &= ~PIPE_DIRECTW;

	pipe_destroy_write_buffer(wpipe);
}

/*
 * This implements the pipe buffer write mechanism.  Note that only
 * a direct write OR a normal pipe write can be pending at any given time.
 * If there are any characters in the pipe buffer, the direct write will
 * be deferred until the receiving process grabs all of the bytes from
 * the pipe buffer.  Then the direct mapping write is set-up.
 */
static int
pipe_direct_write(wpipe, uio)
	struct pipe *wpipe;
	struct uio *uio;
{
	int error;

retry:
	while (wpipe->pipe_state & PIPE_DIRECTW) {
		if (wpipe->pipe_state & PIPE_WANTR) {
			wpipe->pipe_state &= ~PIPE_WANTR;
			wakeup(wpipe);
		}
		wpipe->pipe_state |= PIPE_WANTW;
		error = tsleep(wpipe, PRIBIO | PCATCH, "pipdww", 0);
		if (error)
			goto error1;
		if (wpipe->pipe_state & PIPE_EOF) {
			error = EPIPE;
			goto error1;
		}
	}
	wpipe->pipe_map.cnt = 0;	/* transfer not ready yet */
	if (wpipe->pipe_buffer.cnt > 0) {
		if (wpipe->pipe_state & PIPE_WANTR) {
			wpipe->pipe_state &= ~PIPE_WANTR;
			wakeup(wpipe);
		}

		wpipe->pipe_state |= PIPE_WANTW;
		error = tsleep(wpipe, PRIBIO | PCATCH, "pipdwc", 0);
		if (error)
			goto error1;
		if (wpipe->pipe_state & PIPE_EOF) {
			error = EPIPE;
			goto error1;
		}
		goto retry;
	}

	wpipe->pipe_state |= PIPE_DIRECTW;

	error = pipe_build_write_buffer(wpipe, uio);
	if (error) {
		wpipe->pipe_state &= ~PIPE_DIRECTW;
		goto error1;
	}

	error = 0;
	while (!error && (wpipe->pipe_state & PIPE_DIRECTW)) {
		if (wpipe->pipe_state & PIPE_EOF) {
			pipelock(wpipe, 0);
			pipe_destroy_write_buffer(wpipe);
			pipeunlock(wpipe);
			pipeselwakeup(wpipe, wpipe);
			error = EPIPE;
			goto error1;
		}
		if (wpipe->pipe_state & PIPE_WANTR) {
			wpipe->pipe_state &= ~PIPE_WANTR;
			wakeup(wpipe);
		}
		pipeselwakeup(wpipe, wpipe);
		error = tsleep(wpipe, PRIBIO | PCATCH, "pipdwt", 0);
	}

	pipelock(wpipe,0);
	if (wpipe->pipe_state & PIPE_DIRECTW) {
		/*
		 * this bit of trickery substitutes a kernel buffer for
		 * the process that might be going away.
		 */
		pipe_clone_write_buffer(wpipe);
	} else {
		pipe_destroy_write_buffer(wpipe);
	}
	pipeunlock(wpipe);
	return (error);

error1:
	wakeup(wpipe);
	return (error);
}
#endif /* !PIPE_NODIRECT */
#endif /* FreeBSD */

#ifdef __NetBSD__
#ifndef PIPE_NODIRECT
/*
 * Allocate structure for loan transfer.
 */
static int
pipe_loan_alloc(wpipe, npages)
	struct pipe *wpipe;
	int npages;
{
	vsize_t len;

	len = (vsize_t)npages << PAGE_SHIFT;
	wpipe->pipe_map.kva = uvm_km_valloc_wait(kernel_map, len);
	if (wpipe->pipe_map.kva == NULL)
		return (ENOMEM);

	amountpipekva += len;
	wpipe->pipe_map.npages = npages;
	wpipe->pipe_map.pgs = malloc(npages * sizeof(struct vm_page *), M_PIPE,
	    M_WAITOK);
	return (0);
}

/*
 * Free resources allocated for loan transfer.
 */
static void
pipe_loan_free(wpipe)
	struct pipe *wpipe;
{
	vsize_t len;

	len = (vsize_t)wpipe->pipe_map.npages << PAGE_SHIFT;
	uvm_km_free(kernel_map, wpipe->pipe_map.kva, len);
	wpipe->pipe_map.kva = NULL;
	amountpipekva -= len;
	free(wpipe->pipe_map.pgs, M_PIPE);
	wpipe->pipe_map.pgs = NULL;
}

/*
 * NetBSD direct write, using uvm_loan() mechanism.
 * This implements the pipe buffer write mechanism.  Note that only
 * a direct write OR a normal pipe write can be pending at any given time.
 * If there are any characters in the pipe buffer, the direct write will
 * be deferred until the receiving process grabs all of the bytes from
 * the pipe buffer.  Then the direct mapping write is set-up.
 */
static int
pipe_direct_write(wpipe, uio)
	struct pipe *wpipe;
	struct uio *uio;
{
	int error, npages, j;
	struct vm_page **pgs;
	vaddr_t bbase, kva, base, bend;
	vsize_t blen, bcnt;
	voff_t bpos;

retry:
	while (wpipe->pipe_state & PIPE_DIRECTW) {
		if (wpipe->pipe_state & PIPE_WANTR) {
			wpipe->pipe_state &= ~PIPE_WANTR;
			wakeup(wpipe);
		}
		wpipe->pipe_state |= PIPE_WANTW;
		error = tsleep(wpipe, PRIBIO | PCATCH, "pipdww", 0);
		if (error)
			goto error;
		if (wpipe->pipe_state & PIPE_EOF) {
			error = EPIPE;
			goto error;
		}
	}
	wpipe->pipe_map.cnt = 0;	/* transfer not ready yet */
	if (wpipe->pipe_buffer.cnt > 0) {
		if (wpipe->pipe_state & PIPE_WANTR) {
			wpipe->pipe_state &= ~PIPE_WANTR;
			wakeup(wpipe);
		}

		wpipe->pipe_state |= PIPE_WANTW;
		error = tsleep(wpipe, PRIBIO | PCATCH, "pipdwc", 0);
		if (error)
			goto error;
		if (wpipe->pipe_state & PIPE_EOF) {
			error = EPIPE;
			goto error;
		}
		goto retry;
	}

	/*
	 * Handle first PIPE_CHUNK_SIZE bytes of buffer. Deal with buffers
	 * not aligned to PAGE_SIZE.
	 */
	bbase = (vaddr_t)uio->uio_iov->iov_base;
	base = trunc_page(bbase);
	bend = round_page(bbase + uio->uio_iov->iov_len);
	blen = bend - base;
	bpos = bbase - base;

	if (blen > PIPE_DIRECT_CHUNK) {
		blen = PIPE_DIRECT_CHUNK;
		bend = base + blen;
		bcnt = PIPE_DIRECT_CHUNK - bpos;
	} else {
		bcnt = uio->uio_iov->iov_len;
	}
	npages = blen >> PAGE_SHIFT;

	wpipe->pipe_map.pos = bpos;
	wpipe->pipe_map.cnt = bcnt;

	/*
	 * Free the old kva if we need more pages than we have
	 * allocated.
	 */
	if (wpipe->pipe_map.kva && npages > wpipe->pipe_map.npages)
		pipe_loan_free(wpipe);

	/* Allocate new kva. */
	if (wpipe->pipe_map.kva == NULL) {
		error = pipe_loan_alloc(wpipe, npages);
		if (error) {
			goto error;
		}
	}

	/* Loan the write buffer memory from writer process */
	pgs = wpipe->pipe_map.pgs;
	error = uvm_loan(&uio->uio_procp->p_vmspace->vm_map, base, blen,
	    pgs, UVM_LOAN_TOPAGE);
	if (error) {
		pgs = NULL;
		goto cleanup;
	}

	/* Enter the loaned pages to kva */
	kva = wpipe->pipe_map.kva;
	for (j = 0; j < npages; j++, kva += PAGE_SIZE) {
		pmap_kenter_pa(kva, VM_PAGE_TO_PHYS(pgs[j]), VM_PROT_READ);
	}
	pmap_update(pmap_kernel());

	wpipe->pipe_state |= PIPE_DIRECTW;
	while (!error && (wpipe->pipe_state & PIPE_DIRECTW)) {
		if (wpipe->pipe_state & PIPE_EOF) {
			error = EPIPE;
			break;
		}
		if (wpipe->pipe_state & PIPE_WANTR) {
			wpipe->pipe_state &= ~PIPE_WANTR;
			wakeup(wpipe);
		}
		pipeselwakeup(wpipe, wpipe);
		error = tsleep(wpipe, PRIBIO | PCATCH, "pipdwt", 0);
	}

	if (error)
		wpipe->pipe_state &= ~PIPE_DIRECTW;

cleanup:
	pipelock(wpipe, 0);
	if (pgs != NULL) {
		pmap_kremove(wpipe->pipe_map.kva, blen);
		uvm_unloan(pgs, npages, UVM_LOAN_TOPAGE);
	}
	if (error || amountpipekva > maxpipekva)
		pipe_loan_free(wpipe);
	pipeunlock(wpipe);

	if (error) {
		pipeselwakeup(wpipe, wpipe);

		/*
		 * If nothing was read from what we offered, return error
		 * straight on. Otherwise update uio resid first. Caller
		 * will deal with the error condition, returning short
		 * write, error, or restarting the write(2) as appropriate.
		 */
		if (wpipe->pipe_map.cnt == bcnt) {
error:
			wakeup(wpipe);
			return (error);
		}

		bcnt -= wpipe->pipe_map.cnt;
	}

	uio->uio_resid -= bcnt;
	/* uio_offset not updated, not set/used for write(2) */
	uio->uio_iov->iov_base = (char *)uio->uio_iov->iov_base + bcnt;
	uio->uio_iov->iov_len -= bcnt;
	if (uio->uio_iov->iov_len == 0) {
		uio->uio_iov++;
		uio->uio_iovcnt--;
	}

	return (error);
}
#endif /* !PIPE_NODIRECT */
#endif /* NetBSD */

#ifdef __FreeBSD__
static int
pipe_write(fp, uio, cred, flags, p)
	struct file *fp;
	off_t *offset;
	struct uio *uio;
	struct ucred *cred;
	int flags;
	struct proc *p;
#elif defined(__NetBSD__)
static int
pipe_write(fp, offset, uio, cred, flags)
	struct file *fp;
	off_t *offset;
	struct uio *uio;
	struct ucred *cred;
	int flags;
#endif
{
	int error = 0;
	struct pipe *wpipe, *rpipe;

	rpipe = (struct pipe *) fp->f_data;
	wpipe = rpipe->pipe_peer;

	/*
	 * detect loss of pipe read side, issue SIGPIPE if lost.
	 */
	if ((wpipe == NULL) || (wpipe->pipe_state & PIPE_EOF))
		return (EPIPE);

	++wpipe->pipe_busy;

	/*
	 * If it is advantageous to resize the pipe buffer, do
	 * so.
	 */
	if ((uio->uio_resid > PIPE_SIZE) &&
		(nbigpipe < maxbigpipes) &&
#ifndef PIPE_NODIRECT
		(wpipe->pipe_state & PIPE_DIRECTW) == 0 &&
#endif
		(wpipe->pipe_buffer.size <= PIPE_SIZE) &&
		(wpipe->pipe_buffer.cnt == 0)) {

		if ((error = pipelock(wpipe,1)) == 0) {
			if (pipespace(wpipe, BIG_PIPE_SIZE) == 0)
				nbigpipe++;
			pipeunlock(wpipe);
		} else {
			/*
			 * If an error occurred, unbusy and return, waking up
			 * any waiting readers.
			 */ 
			--wpipe->pipe_busy;
			if (wpipe->pipe_busy == 0
			    && (wpipe->pipe_state & PIPE_WANTCLOSE)) {
				wpipe->pipe_state &=
				    ~(PIPE_WANTCLOSE | PIPE_WANTR);
				wakeup(wpipe);
			}

			return (error);
		}
	}

#ifdef __FreeBSD__
	KASSERT(wpipe->pipe_buffer.buffer != NULL, ("pipe buffer gone"));
#endif

	while (uio->uio_resid) {
		int space;

#ifndef PIPE_NODIRECT
		/*
		 * If the transfer is large, we can gain performance if
		 * we do process-to-process copies directly.
		 * If the write is non-blocking, we don't use the
		 * direct write mechanism.
		 *
		 * The direct write mechanism will detect the reader going
		 * away on us.
		 */
		if ((uio->uio_iov->iov_len >= PIPE_MINDIRECT) &&
		    (fp->f_flag & FNONBLOCK) == 0 &&
		    (wpipe->pipe_map.kva || (amountpipekva < limitpipekva))) {
			error = pipe_direct_write(wpipe, uio);

			/*
			 * Break out if error occured, unless it's ENOMEM.
			 * ENOMEM means we failed to allocate some resources
			 * for direct write, so we just fallback to ordinary
			 * write. If the direct write was successful,
			 * process rest of data via ordinary write.
			 */
			if (!error)
				continue;

			if (error != ENOMEM)
				break;
		}
#endif /* PIPE_NODIRECT */

		/*
		 * Pipe buffered writes cannot be coincidental with
		 * direct writes.  We wait until the currently executing
		 * direct write is completed before we start filling the
		 * pipe buffer.  We break out if a signal occurs or the
		 * reader goes away.
		 */
	retrywrite:
		while (wpipe->pipe_state & PIPE_DIRECTW) {
			if (wpipe->pipe_state & PIPE_WANTR) {
				wpipe->pipe_state &= ~PIPE_WANTR;
				wakeup(wpipe);
			}
			error = tsleep(wpipe, PRIBIO | PCATCH, "pipbww", 0);
			if (wpipe->pipe_state & PIPE_EOF)
				break;
			if (error)
				break;
		}
		if (wpipe->pipe_state & PIPE_EOF) {
			error = EPIPE;
			break;
		}

		space = wpipe->pipe_buffer.size - wpipe->pipe_buffer.cnt;

		/* Writes of size <= PIPE_BUF must be atomic. */
		if ((space < uio->uio_resid) && (uio->uio_resid <= PIPE_BUF))
			space = 0;

		if (space > 0) {
			int size;	/* Transfer size */
			int segsize;	/* first segment to transfer */

			if ((error = pipelock(wpipe,1)) != 0)
				break;

			/*
			 * It is possible for a direct write to
			 * slip in on us... handle it here...
			 */
			if (wpipe->pipe_state & PIPE_DIRECTW) {
				pipeunlock(wpipe);
				goto retrywrite;
			}
			/* 
			 * If a process blocked in uiomove, our
			 * value for space might be bad.
			 *
			 * XXX will we be ok if the reader has gone
			 * away here?
			 */
			if (space > wpipe->pipe_buffer.size - 
				    wpipe->pipe_buffer.cnt) {
				pipeunlock(wpipe);
				goto retrywrite;
			}

			/*
			 * Transfer size is minimum of uio transfer
			 * and free space in pipe buffer.
			 */
			if (space > uio->uio_resid)
				size = uio->uio_resid;
			else
				size = space;
			/*
			 * First segment to transfer is minimum of 
			 * transfer size and contiguous space in
			 * pipe buffer.  If first segment to transfer
			 * is less than the transfer size, we've got
			 * a wraparound in the buffer.
			 */
			segsize = wpipe->pipe_buffer.size - 
				wpipe->pipe_buffer.in;
			if (segsize > size)
				segsize = size;

			/* Transfer first segment */

			error = uiomove(&wpipe->pipe_buffer.buffer[wpipe->pipe_buffer.in], 
						segsize, uio);

			if (error == 0 && segsize < size) {
				/* 
				 * Transfer remaining part now, to
				 * support atomic writes.  Wraparound
				 * happened.
				 */
#ifdef DEBUG
				if (wpipe->pipe_buffer.in + segsize != 
				    wpipe->pipe_buffer.size)
					panic("Expected pipe buffer wraparound disappeared");
#endif

				error = uiomove(&wpipe->pipe_buffer.buffer[0],
						size - segsize, uio);
			}
			if (error == 0) {
				wpipe->pipe_buffer.in += size;
				if (wpipe->pipe_buffer.in >=
				    wpipe->pipe_buffer.size) {
#ifdef DEBUG
					if (wpipe->pipe_buffer.in != size - segsize + wpipe->pipe_buffer.size)
						panic("Expected wraparound bad");
#endif
					wpipe->pipe_buffer.in = size - segsize;
				}

				wpipe->pipe_buffer.cnt += size;
#ifdef DEBUG
				if (wpipe->pipe_buffer.cnt > wpipe->pipe_buffer.size)
					panic("Pipe buffer overflow");
#endif
			}
			pipeunlock(wpipe);
			if (error)
				break;
		} else {
			/*
			 * If the "read-side" has been blocked, wake it up now.
			 */
			if (wpipe->pipe_state & PIPE_WANTR) {
				wpipe->pipe_state &= ~PIPE_WANTR;
				wakeup(wpipe);
			}

			/*
			 * don't block on non-blocking I/O
			 */
			if (fp->f_flag & FNONBLOCK) {
				error = EAGAIN;
				break;
			}

			/*
			 * We have no more space and have something to offer,
			 * wake up select/poll.
			 */
			pipeselwakeup(wpipe, wpipe);

			wpipe->pipe_state |= PIPE_WANTW;
			error = tsleep(wpipe, PRIBIO | PCATCH, "pipewr", 0);
			if (error != 0)
				break;
			/*
			 * If read side wants to go away, we just issue a signal
			 * to ourselves.
			 */
			if (wpipe->pipe_state & PIPE_EOF) {
				error = EPIPE;
				break;
			}
		}
	}

	--wpipe->pipe_busy;
	if ((wpipe->pipe_busy == 0) && (wpipe->pipe_state & PIPE_WANTCLOSE)) {
		wpipe->pipe_state &= ~(PIPE_WANTCLOSE | PIPE_WANTR);
		wakeup(wpipe);
	} else if (wpipe->pipe_buffer.cnt > 0) {
		/*
		 * If we have put any characters in the buffer, we wake up
		 * the reader.
		 */
		if (wpipe->pipe_state & PIPE_WANTR) {
			wpipe->pipe_state &= ~PIPE_WANTR;
			wakeup(wpipe);
		}
	}

	/*
	 * Don't return EPIPE if I/O was successful
	 */
	if ((error == EPIPE) && (wpipe->pipe_buffer.cnt == 0)
	    && (uio->uio_resid == 0))
		error = 0;

	if (error == 0)
		vfs_timestamp(&wpipe->pipe_mtime);

	/*
	 * We have something to offer, wake up select/poll.
	 * wpipe->pipe_map.cnt is always 0 in this point (direct write
	 * is only done synchronously), so check only wpipe->pipe_buffer.cnt
	 */
	if (wpipe->pipe_buffer.cnt)
		pipeselwakeup(wpipe, wpipe);

	/*
	 * Arrange for next read(2) to do a signal.
	 */
	wpipe->pipe_state |= PIPE_SIGNALR;

	return (error);
}

/*
 * we implement a very minimal set of ioctls for compatibility with sockets.
 */
int
pipe_ioctl(fp, cmd, data, p)
	struct file *fp;
	u_long cmd;
	caddr_t data;
	struct proc *p;
{
	struct pipe *mpipe = (struct pipe *)fp->f_data;

	switch (cmd) {

	case FIONBIO:
		return (0);

	case FIOASYNC:
		if (*(int *)data) {
			mpipe->pipe_state |= PIPE_ASYNC;
		} else {
			mpipe->pipe_state &= ~PIPE_ASYNC;
		}
		return (0);

	case FIONREAD:
#ifndef PIPE_NODIRECT
		if (mpipe->pipe_state & PIPE_DIRECTW)
			*(int *)data = mpipe->pipe_map.cnt;
		else
#endif
			*(int *)data = mpipe->pipe_buffer.cnt;
		return (0);

#ifdef __FreeBSD__
	case FIOSETOWN:
		return (fsetown(*(int *)data, &mpipe->pipe_sigio));

	case FIOGETOWN:
		*(int *)data = fgetown(mpipe->pipe_sigio);
		return (0);

	/* This is deprecated, FIOSETOWN should be used instead. */
	case TIOCSPGRP:
		return (fsetown(-(*(int *)data), &mpipe->pipe_sigio));

	/* This is deprecated, FIOGETOWN should be used instead. */
	case TIOCGPGRP:
		*(int *)data = -fgetown(mpipe->pipe_sigio);
		return (0);
#endif /* FreeBSD */
#ifdef __NetBSD__
	case TIOCSPGRP:
		mpipe->pipe_pgid = *(int *)data;
		return (0);

	case TIOCGPGRP:
		*(int *)data = mpipe->pipe_pgid;
		return (0);
#endif /* NetBSD */

	}
	return (ENOTTY);
}

int
pipe_poll(fp, events, p)
	struct file *fp;
	int events;
	struct proc *p;
{
	struct pipe *rpipe = (struct pipe *)fp->f_data;
	struct pipe *wpipe;
	int revents = 0;

	wpipe = rpipe->pipe_peer;
	if (events & (POLLIN | POLLRDNORM))
		if ((rpipe->pipe_buffer.cnt > 0) ||
#ifndef PIPE_NODIRECT
		    (rpipe->pipe_state & PIPE_DIRECTW) ||
#endif
		    (rpipe->pipe_state & PIPE_EOF))
			revents |= events & (POLLIN | POLLRDNORM);

	if (events & (POLLOUT | POLLWRNORM))
		if (wpipe == NULL || (wpipe->pipe_state & PIPE_EOF)
		    || (
#ifndef PIPE_NODIRECT
		     ((wpipe->pipe_state & PIPE_DIRECTW) == 0) &&
#endif
		     (wpipe->pipe_buffer.size - wpipe->pipe_buffer.cnt) >= PIPE_BUF))
			revents |= events & (POLLOUT | POLLWRNORM);

	if ((rpipe->pipe_state & PIPE_EOF) ||
	    (wpipe == NULL) ||
	    (wpipe->pipe_state & PIPE_EOF))
		revents |= POLLHUP;

	if (revents == 0) {
		if (events & (POLLIN | POLLRDNORM)) {
			selrecord(p, &rpipe->pipe_sel);
			rpipe->pipe_state |= PIPE_SEL;
		}

		if (events & (POLLOUT | POLLWRNORM)) {
			selrecord(p, &wpipe->pipe_sel);
			wpipe->pipe_state |= PIPE_SEL;
		}
	}

	return (revents);
}

static int
pipe_stat(fp, ub, p)
	struct file *fp;
	struct stat *ub;
	struct proc *p;
{
	struct pipe *pipe = (struct pipe *)fp->f_data;

	memset((caddr_t)ub, 0, sizeof(*ub));
	ub->st_mode = S_IFIFO;
	ub->st_blksize = pipe->pipe_buffer.size;
	ub->st_size = pipe->pipe_buffer.cnt;
	ub->st_blocks = (ub->st_size) ? 1 : 0;
#ifdef __FreeBSD__
	ub->st_atimespec = pipe->pipe_atime;
	ub->st_mtimespec = pipe->pipe_mtime;
	ub->st_ctimespec = pipe->pipe_ctime;
#endif /* FreeBSD */
#ifdef __NetBSD__
	TIMEVAL_TO_TIMESPEC(&pipe->pipe_atime, &ub->st_atimespec)
	TIMEVAL_TO_TIMESPEC(&pipe->pipe_mtime, &ub->st_mtimespec);
	TIMEVAL_TO_TIMESPEC(&pipe->pipe_ctime, &ub->st_ctimespec);
#endif /* NetBSD */
	ub->st_uid = fp->f_cred->cr_uid;
	ub->st_gid = fp->f_cred->cr_gid;
	/*
	 * Left as 0: st_dev, st_ino, st_nlink, st_rdev, st_flags, st_gen.
	 * XXX (st_dev, st_ino) should be unique.
	 */
	return (0);
}

/* ARGSUSED */
static int
pipe_close(fp, p)
	struct file *fp;
	struct proc *p;
{
	struct pipe *cpipe = (struct pipe *)fp->f_data;

#ifdef __FreeBSD__
	fp->f_ops = &badfileops;
	funsetown(cpipe->pipe_sigio);
#endif
	fp->f_data = NULL;
	pipeclose(cpipe);
	return (0);
}

static void
pipe_free_kmem(cpipe)
	struct pipe *cpipe;
{

#ifdef __FreeBSD__
	mtx_assert(&vm_mtx, MA_OWNED);
#endif
	if (cpipe->pipe_buffer.buffer != NULL) {
		if (cpipe->pipe_buffer.size > PIPE_SIZE)
			--nbigpipe;
		amountpipekva -= cpipe->pipe_buffer.size;
#ifdef __FreeBSD__
		kmem_free(kernel_map,
			(vm_offset_t)cpipe->pipe_buffer.buffer,
			cpipe->pipe_buffer.size);
#elif defined(__NetBSD__)
		uvm_km_free(kernel_map,
			(vaddr_t)cpipe->pipe_buffer.buffer,
			cpipe->pipe_buffer.size);
#endif /* NetBSD */
		cpipe->pipe_buffer.buffer = NULL;
	}
#ifndef PIPE_NODIRECT
	if (cpipe->pipe_map.kva != NULL) {
#ifdef __FreeBSD__
		amountpipekva -= cpipe->pipe_buffer.size + PAGE_SIZE;
		kmem_free(kernel_map,
			cpipe->pipe_map.kva,
			cpipe->pipe_buffer.size + PAGE_SIZE);
#elif defined(__NetBSD__)
		pipe_loan_free(cpipe);
#endif /* NetBSD */
		cpipe->pipe_map.cnt = 0;
		cpipe->pipe_map.kva = NULL;
		cpipe->pipe_map.pos = 0;
		cpipe->pipe_map.npages = 0;
	}
#endif /* !PIPE_NODIRECT */
}

/*
 * shutdown the pipe
 */
static void
pipeclose(cpipe)
	struct pipe *cpipe;
{
	struct pipe *ppipe;

	if (!cpipe)
		return;

	pipeselwakeup(cpipe, cpipe);

	/*
	 * If the other side is blocked, wake it up saying that
	 * we want to close it down.
	 */
	while (cpipe->pipe_busy) {
		wakeup(cpipe);
		cpipe->pipe_state |= PIPE_WANTCLOSE | PIPE_EOF;
		tsleep(cpipe, PRIBIO, "pipecl", 0);
	}

	/*
	 * Disconnect from peer
	 */
	if ((ppipe = cpipe->pipe_peer) != NULL) {
		pipeselwakeup(ppipe, ppipe);

		ppipe->pipe_state |= PIPE_EOF;
		wakeup(ppipe);
		ppipe->pipe_peer = NULL;
	}

	/*
	 * free resources
	 */
#ifdef __FreeBSD__
	mtx_lock(&vm_mtx);
	pipe_free_kmem(cpipe);
	/* XXX: erm, doesn't zalloc already have its own locks and
	 * not need the giant vm lock?
	 */
	zfree(pipe_zone, cpipe);
	mtx_unlock(&vm_mtx);
#endif /* FreeBSD */

#ifdef __NetBSD__
	pipe_free_kmem(cpipe);
	(void) lockmgr(&cpipe->pipe_lock, LK_DRAIN, NULL);
	pool_put(&pipe_pool, cpipe);
#endif
}

#ifdef __FreeBSD__
/*ARGSUSED*/
static int
pipe_kqfilter(struct file *fp, struct knote *kn)
{
	struct pipe *cpipe = (struct pipe *)kn->kn_fp->f_data;

	switch (kn->kn_filter) {
	case EVFILT_READ:
		kn->kn_fop = &pipe_rfiltops;
		break;
	case EVFILT_WRITE:
		kn->kn_fop = &pipe_wfiltops;
		cpipe = cpipe->pipe_peer;
		break;
	default:
		return (1);
	}
	kn->kn_hook = (caddr_t)cpipe;
	SLIST_INSERT_HEAD(&cpipe->pipe_sel.si_note, kn, kn_selnext);
	return (0);
}

static void
filt_pipedetach(struct knote *kn)
{
	struct pipe *cpipe = (struct pipe *)kn->kn_fp->f_data;

	SLIST_REMOVE(&cpipe->pipe_sel.si_note, kn, knote, kn_selnext);
}

/*ARGSUSED*/
static int
filt_piperead(struct knote *kn, long hint)
{
	struct pipe *rpipe = (struct pipe *)kn->kn_fp->f_data;
	struct pipe *wpipe = rpipe->pipe_peer;

	kn->kn_data = rpipe->pipe_buffer.cnt;
	if ((kn->kn_data == 0) && (rpipe->pipe_state & PIPE_DIRECTW))
		kn->kn_data = rpipe->pipe_map.cnt;

	if ((rpipe->pipe_state & PIPE_EOF) ||
	    (wpipe == NULL) || (wpipe->pipe_state & PIPE_EOF)) {
		kn->kn_flags |= EV_EOF; 
		return (1);
	}
	return (kn->kn_data > 0);
}

/*ARGSUSED*/
static int
filt_pipewrite(struct knote *kn, long hint)
{
	struct pipe *rpipe = (struct pipe *)kn->kn_fp->f_data;
	struct pipe *wpipe = rpipe->pipe_peer;

	if ((wpipe == NULL) || (wpipe->pipe_state & PIPE_EOF)) {
		kn->kn_data = 0;
		kn->kn_flags |= EV_EOF; 
		return (1);
	}
	kn->kn_data = wpipe->pipe_buffer.size - wpipe->pipe_buffer.cnt;
	if (wpipe->pipe_state & PIPE_DIRECTW)
		kn->kn_data = 0;

	return (kn->kn_data >= PIPE_BUF);
}
#endif /* FreeBSD */

#ifdef __NetBSD__
static int
pipe_fcntl(fp, cmd, data, p)
	struct file *fp;
	u_int cmd;
	caddr_t data;
	struct proc *p;
{
	if (cmd == F_SETFL)
		return (0);
	else
		return (EOPNOTSUPP);
}

/*
 * Handle pipe sysctls.
 */
int
sysctl_dopipe(name, namelen, oldp, oldlenp, newp, newlen)
	int *name;
	u_int namelen;
	void *oldp;
	size_t *oldlenp;
	void *newp;
	size_t newlen;
{
	/* All sysctl names at this level are terminal. */
	if (namelen != 1)
		return (ENOTDIR);		/* overloaded */

	switch (name[0]) {
	case KERN_PIPE_MAXKVASZ:
		return (sysctl_int(oldp, oldlenp, newp, newlen, &maxpipekva));
	case KERN_PIPE_LIMITKVA:
		return (sysctl_int(oldp, oldlenp, newp, newlen, &limitpipekva));
	case KERN_PIPE_MAXBIGPIPES:
		return (sysctl_int(oldp, oldlenp, newp, newlen, &maxbigpipes));
	case KERN_PIPE_NBIGPIPES:
		return (sysctl_rdint(oldp, oldlenp, newp, nbigpipe));
	case KERN_PIPE_KVASIZE:
		return (sysctl_rdint(oldp, oldlenp, newp, amountpipekva));
	default:
		return (EOPNOTSUPP);
	}
	/* NOTREACHED */
}

/*
 * Initialize pipe structs.
 */
void
pipe_init(void)
{
	pool_init(&pipe_pool, sizeof(struct pipe), 0, 0, 0, "pipepl",
		0, NULL, NULL, M_PIPE);
}

#endif /* __NetBSD __ */