summaryrefslogtreecommitdiffstats
path: root/sys/vm/vm_pageout.c
blob: b7f688732970cad4b27d8592231306d7df4ae18b (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
/*-
 * Copyright (c) 1991 Regents of the University of California.
 * All rights reserved.
 * Copyright (c) 1994 John S. Dyson
 * All rights reserved.
 * Copyright (c) 1994 David Greenman
 * All rights reserved.
 * Copyright (c) 2005 Yahoo! Technologies Norway AS
 * All rights reserved.
 *
 * This code is derived from software contributed to Berkeley by
 * The Mach Operating System project at Carnegie-Mellon University.
 *
 * 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.
 * 3. All advertising materials mentioning features or use of this software
 *    must display the following acknowledgement:
 *	This product includes software developed by the University of
 *	California, Berkeley and its contributors.
 * 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.
 *
 *	from: @(#)vm_pageout.c	7.4 (Berkeley) 5/7/91
 *
 *
 * Copyright (c) 1987, 1990 Carnegie-Mellon University.
 * All rights reserved.
 *
 * Authors: Avadis Tevanian, Jr., Michael Wayne Young
 *
 * Permission to use, copy, modify and distribute this software and
 * its documentation is hereby granted, provided that both the copyright
 * notice and this permission notice appear in all copies of the
 * software, derivative works or modified versions, and any portions
 * thereof, and that both notices appear in supporting documentation.
 *
 * CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS"
 * CONDITION.  CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND
 * FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE.
 *
 * Carnegie Mellon requests users of this software to return to
 *
 *  Software Distribution Coordinator  or  Software.Distribution@CS.CMU.EDU
 *  School of Computer Science
 *  Carnegie Mellon University
 *  Pittsburgh PA 15213-3890
 *
 * any improvements or extensions that they make and grant Carnegie the
 * rights to redistribute these changes.
 */

/*
 *	The proverbial page-out daemon.
 */

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

#include "opt_vm.h"

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/kernel.h>
#include <sys/eventhandler.h>
#include <sys/lock.h>
#include <sys/mutex.h>
#include <sys/proc.h>
#include <sys/kthread.h>
#include <sys/ktr.h>
#include <sys/mount.h>
#include <sys/racct.h>
#include <sys/resourcevar.h>
#include <sys/sched.h>
#include <sys/sdt.h>
#include <sys/signalvar.h>
#include <sys/smp.h>
#include <sys/time.h>
#include <sys/vnode.h>
#include <sys/vmmeter.h>
#include <sys/rwlock.h>
#include <sys/sx.h>
#include <sys/sysctl.h>

#include <vm/vm.h>
#include <vm/vm_param.h>
#include <vm/vm_object.h>
#include <vm/vm_page.h>
#include <vm/vm_map.h>
#include <vm/vm_pageout.h>
#include <vm/vm_pager.h>
#include <vm/vm_phys.h>
#include <vm/swap_pager.h>
#include <vm/vm_extern.h>
#include <vm/uma.h>

/*
 * System initialization
 */

/* the kernel process "vm_pageout"*/
static void vm_pageout(void);
static void vm_pageout_init(void);
static int vm_pageout_clean(vm_page_t m);
static int vm_pageout_cluster(vm_page_t m);
static void vm_pageout_scan(struct vm_domain *vmd, int pass);
static void vm_pageout_mightbe_oom(struct vm_domain *vmd, int page_shortage,
    int starting_page_shortage);

SYSINIT(pagedaemon_init, SI_SUB_KTHREAD_PAGE, SI_ORDER_FIRST, vm_pageout_init,
    NULL);

struct proc *pageproc;

static struct kproc_desc page_kp = {
	"pagedaemon",
	vm_pageout,
	&pageproc
};
SYSINIT(pagedaemon, SI_SUB_KTHREAD_PAGE, SI_ORDER_SECOND, kproc_start,
    &page_kp);

SDT_PROVIDER_DEFINE(vm);
SDT_PROBE_DEFINE(vm, , , vm__lowmem_cache);
SDT_PROBE_DEFINE(vm, , , vm__lowmem_scan);

#if !defined(NO_SWAPPING)
/* the kernel process "vm_daemon"*/
static void vm_daemon(void);
static struct	proc *vmproc;

static struct kproc_desc vm_kp = {
	"vmdaemon",
	vm_daemon,
	&vmproc
};
SYSINIT(vmdaemon, SI_SUB_KTHREAD_VM, SI_ORDER_FIRST, kproc_start, &vm_kp);
#endif


int vm_pages_needed;		/* Event on which pageout daemon sleeps */
int vm_pageout_deficit;		/* Estimated number of pages deficit */
int vm_pageout_wakeup_thresh;
static int vm_pageout_oom_seq = 12;

#if !defined(NO_SWAPPING)
static int vm_pageout_req_swapout;	/* XXX */
static int vm_daemon_needed;
static struct mtx vm_daemon_mtx;
/* Allow for use by vm_pageout before vm_daemon is initialized. */
MTX_SYSINIT(vm_daemon, &vm_daemon_mtx, "vm daemon", MTX_DEF);
#endif
static int vm_max_launder = 32;
static int vm_pageout_update_period;
static int defer_swap_pageouts;
static int disable_swap_pageouts;
static int lowmem_period = 10;
static time_t lowmem_uptime;

#if defined(NO_SWAPPING)
static int vm_swap_enabled = 0;
static int vm_swap_idle_enabled = 0;
#else
static int vm_swap_enabled = 1;
static int vm_swap_idle_enabled = 0;
#endif

static int vm_panic_on_oom = 0;

SYSCTL_INT(_vm, OID_AUTO, panic_on_oom,
	CTLFLAG_RWTUN, &vm_panic_on_oom, 0,
	"panic on out of memory instead of killing the largest process");

SYSCTL_INT(_vm, OID_AUTO, pageout_wakeup_thresh,
	CTLFLAG_RW, &vm_pageout_wakeup_thresh, 0,
	"free page threshold for waking up the pageout daemon");

SYSCTL_INT(_vm, OID_AUTO, max_launder,
	CTLFLAG_RW, &vm_max_launder, 0, "Limit dirty flushes in pageout");

SYSCTL_INT(_vm, OID_AUTO, pageout_update_period,
	CTLFLAG_RW, &vm_pageout_update_period, 0,
	"Maximum active LRU update period");
  
SYSCTL_INT(_vm, OID_AUTO, lowmem_period, CTLFLAG_RW, &lowmem_period, 0,
	"Low memory callback period");

#if defined(NO_SWAPPING)
SYSCTL_INT(_vm, VM_SWAPPING_ENABLED, swap_enabled,
	CTLFLAG_RD, &vm_swap_enabled, 0, "Enable entire process swapout");
SYSCTL_INT(_vm, OID_AUTO, swap_idle_enabled,
	CTLFLAG_RD, &vm_swap_idle_enabled, 0, "Allow swapout on idle criteria");
#else
SYSCTL_INT(_vm, VM_SWAPPING_ENABLED, swap_enabled,
	CTLFLAG_RW, &vm_swap_enabled, 0, "Enable entire process swapout");
SYSCTL_INT(_vm, OID_AUTO, swap_idle_enabled,
	CTLFLAG_RW, &vm_swap_idle_enabled, 0, "Allow swapout on idle criteria");
#endif

SYSCTL_INT(_vm, OID_AUTO, defer_swapspace_pageouts,
	CTLFLAG_RW, &defer_swap_pageouts, 0, "Give preference to dirty pages in mem");

SYSCTL_INT(_vm, OID_AUTO, disable_swapspace_pageouts,
	CTLFLAG_RW, &disable_swap_pageouts, 0, "Disallow swapout of dirty pages");

static int pageout_lock_miss;
SYSCTL_INT(_vm, OID_AUTO, pageout_lock_miss,
	CTLFLAG_RD, &pageout_lock_miss, 0, "vget() lock misses during pageout");

SYSCTL_INT(_vm, OID_AUTO, pageout_oom_seq,
	CTLFLAG_RW, &vm_pageout_oom_seq, 0,
	"back-to-back calls to oom detector to start OOM");

#define VM_PAGEOUT_PAGE_COUNT 16
int vm_pageout_page_count = VM_PAGEOUT_PAGE_COUNT;

int vm_page_max_wired;		/* XXX max # of wired pages system-wide */
SYSCTL_INT(_vm, OID_AUTO, max_wired,
	CTLFLAG_RW, &vm_page_max_wired, 0, "System-wide limit to wired page count");

static boolean_t vm_pageout_fallback_object_lock(vm_page_t, vm_page_t *);
#if !defined(NO_SWAPPING)
static void vm_pageout_map_deactivate_pages(vm_map_t, long);
static void vm_pageout_object_deactivate_pages(pmap_t, vm_object_t, long);
static void vm_req_vmdaemon(int req);
#endif
static boolean_t vm_pageout_page_lock(vm_page_t, vm_page_t *);

/*
 * Initialize a dummy page for marking the caller's place in the specified
 * paging queue.  In principle, this function only needs to set the flag
 * PG_MARKER.  Nonetheless, it wirte busies and initializes the hold count
 * to one as safety precautions.
 */ 
static void
vm_pageout_init_marker(vm_page_t marker, u_short queue)
{

	bzero(marker, sizeof(*marker));
	marker->flags = PG_MARKER;
	marker->busy_lock = VPB_SINGLE_EXCLUSIVER;
	marker->queue = queue;
	marker->hold_count = 1;
}

/*
 * vm_pageout_fallback_object_lock:
 * 
 * Lock vm object currently associated with `m'. VM_OBJECT_TRYWLOCK is
 * known to have failed and page queue must be either PQ_ACTIVE or
 * PQ_INACTIVE.  To avoid lock order violation, unlock the page queues
 * while locking the vm object.  Use marker page to detect page queue
 * changes and maintain notion of next page on page queue.  Return
 * TRUE if no changes were detected, FALSE otherwise.  vm object is
 * locked on return.
 * 
 * This function depends on both the lock portion of struct vm_object
 * and normal struct vm_page being type stable.
 */
static boolean_t
vm_pageout_fallback_object_lock(vm_page_t m, vm_page_t *next)
{
	struct vm_page marker;
	struct vm_pagequeue *pq;
	boolean_t unchanged;
	u_short queue;
	vm_object_t object;

	queue = m->queue;
	vm_pageout_init_marker(&marker, queue);
	pq = vm_page_pagequeue(m);
	object = m->object;
	
	TAILQ_INSERT_AFTER(&pq->pq_pl, m, &marker, plinks.q);
	vm_pagequeue_unlock(pq);
	vm_page_unlock(m);
	VM_OBJECT_WLOCK(object);
	vm_page_lock(m);
	vm_pagequeue_lock(pq);

	/*
	 * The page's object might have changed, and/or the page might
	 * have moved from its original position in the queue.  If the
	 * page's object has changed, then the caller should abandon
	 * processing the page because the wrong object lock was
	 * acquired.  Use the marker's plinks.q, not the page's, to
	 * determine if the page has been moved.  The state of the
	 * page's plinks.q can be indeterminate; whereas, the marker's
	 * plinks.q must be valid.
	 */
	*next = TAILQ_NEXT(&marker, plinks.q);
	unchanged = m->object == object &&
	    m == TAILQ_PREV(&marker, pglist, plinks.q);
	KASSERT(!unchanged || m->queue == queue,
	    ("page %p queue %d %d", m, queue, m->queue));
	TAILQ_REMOVE(&pq->pq_pl, &marker, plinks.q);
	return (unchanged);
}

/*
 * Lock the page while holding the page queue lock.  Use marker page
 * to detect page queue changes and maintain notion of next page on
 * page queue.  Return TRUE if no changes were detected, FALSE
 * otherwise.  The page is locked on return. The page queue lock might
 * be dropped and reacquired.
 *
 * This function depends on normal struct vm_page being type stable.
 */
static boolean_t
vm_pageout_page_lock(vm_page_t m, vm_page_t *next)
{
	struct vm_page marker;
	struct vm_pagequeue *pq;
	boolean_t unchanged;
	u_short queue;

	vm_page_lock_assert(m, MA_NOTOWNED);
	if (vm_page_trylock(m))
		return (TRUE);

	queue = m->queue;
	vm_pageout_init_marker(&marker, queue);
	pq = vm_page_pagequeue(m);

	TAILQ_INSERT_AFTER(&pq->pq_pl, m, &marker, plinks.q);
	vm_pagequeue_unlock(pq);
	vm_page_lock(m);
	vm_pagequeue_lock(pq);

	/* Page queue might have changed. */
	*next = TAILQ_NEXT(&marker, plinks.q);
	unchanged = m == TAILQ_PREV(&marker, pglist, plinks.q);
	KASSERT(!unchanged || m->queue == queue,
	    ("page %p queue %d %d", m, queue, m->queue));
	TAILQ_REMOVE(&pq->pq_pl, &marker, plinks.q);
	return (unchanged);
}

/*
 * vm_pageout_clean:
 *
 * Clean the page and remove it from the laundry.
 * 
 * We set the busy bit to cause potential page faults on this page to
 * block.  Note the careful timing, however, the busy bit isn't set till
 * late and we cannot do anything that will mess with the page.
 */
static int
vm_pageout_cluster(vm_page_t m)
{
	vm_object_t object;
	vm_page_t mc[2*vm_pageout_page_count], pb, ps;
	int pageout_count;
	int ib, is, page_base;
	vm_pindex_t pindex = m->pindex;

	vm_page_lock_assert(m, MA_OWNED);
	object = m->object;
	VM_OBJECT_ASSERT_WLOCKED(object);

	/*
	 * It doesn't cost us anything to pageout OBJT_DEFAULT or OBJT_SWAP
	 * with the new swapper, but we could have serious problems paging
	 * out other object types if there is insufficient memory.  
	 *
	 * Unfortunately, checking free memory here is far too late, so the
	 * check has been moved up a procedural level.
	 */

	/*
	 * Can't clean the page if it's busy or held.
	 */
	vm_page_assert_unbusied(m);
	KASSERT(m->hold_count == 0, ("vm_pageout_clean: page %p is held", m));
	vm_page_unlock(m);

	mc[vm_pageout_page_count] = pb = ps = m;
	pageout_count = 1;
	page_base = vm_pageout_page_count;
	ib = 1;
	is = 1;

	/*
	 * Scan object for clusterable pages.
	 *
	 * We can cluster ONLY if: ->> the page is NOT
	 * clean, wired, busy, held, or mapped into a
	 * buffer, and one of the following:
	 * 1) The page is inactive, or a seldom used
	 *    active page.
	 * -or-
	 * 2) we force the issue.
	 *
	 * During heavy mmap/modification loads the pageout
	 * daemon can really fragment the underlying file
	 * due to flushing pages out of order and not trying
	 * align the clusters (which leave sporatic out-of-order
	 * holes).  To solve this problem we do the reverse scan
	 * first and attempt to align our cluster, then do a 
	 * forward scan if room remains.
	 */
more:
	while (ib && pageout_count < vm_pageout_page_count) {
		vm_page_t p;

		if (ib > pindex) {
			ib = 0;
			break;
		}

		if ((p = vm_page_prev(pb)) == NULL || vm_page_busied(p)) {
			ib = 0;
			break;
		}
		vm_page_test_dirty(p);
		if (p->dirty == 0) {
			ib = 0;
			break;
		}
		vm_page_lock(p);
		if (p->queue != PQ_INACTIVE ||
		    p->hold_count != 0) {	/* may be undergoing I/O */
			vm_page_unlock(p);
			ib = 0;
			break;
		}
		vm_page_unlock(p);
		mc[--page_base] = pb = p;
		++pageout_count;
		++ib;
		/*
		 * alignment boundry, stop here and switch directions.  Do
		 * not clear ib.
		 */
		if ((pindex - (ib - 1)) % vm_pageout_page_count == 0)
			break;
	}

	while (pageout_count < vm_pageout_page_count && 
	    pindex + is < object->size) {
		vm_page_t p;

		if ((p = vm_page_next(ps)) == NULL || vm_page_busied(p))
			break;
		vm_page_test_dirty(p);
		if (p->dirty == 0)
			break;
		vm_page_lock(p);
		if (p->queue != PQ_INACTIVE ||
		    p->hold_count != 0) {	/* may be undergoing I/O */
			vm_page_unlock(p);
			break;
		}
		vm_page_unlock(p);
		mc[page_base + pageout_count] = ps = p;
		++pageout_count;
		++is;
	}

	/*
	 * If we exhausted our forward scan, continue with the reverse scan
	 * when possible, even past a page boundry.  This catches boundry
	 * conditions.
	 */
	if (ib && pageout_count < vm_pageout_page_count)
		goto more;

	/*
	 * we allow reads during pageouts...
	 */
	return (vm_pageout_flush(&mc[page_base], pageout_count, 0, 0, NULL,
	    NULL));
}

/*
 * vm_pageout_flush() - launder the given pages
 *
 *	The given pages are laundered.  Note that we setup for the start of
 *	I/O ( i.e. busy the page ), mark it read-only, and bump the object
 *	reference count all in here rather then in the parent.  If we want
 *	the parent to do more sophisticated things we may have to change
 *	the ordering.
 *
 *	Returned runlen is the count of pages between mreq and first
 *	page after mreq with status VM_PAGER_AGAIN.
 *	*eio is set to TRUE if pager returned VM_PAGER_ERROR or VM_PAGER_FAIL
 *	for any page in runlen set.
 */
int
vm_pageout_flush(vm_page_t *mc, int count, int flags, int mreq, int *prunlen,
    boolean_t *eio)
{
	vm_object_t object = mc[0]->object;
	int pageout_status[count];
	int numpagedout = 0;
	int i, runlen;

	VM_OBJECT_ASSERT_WLOCKED(object);

	/*
	 * Initiate I/O.  Bump the vm_page_t->busy counter and
	 * mark the pages read-only.
	 *
	 * We do not have to fixup the clean/dirty bits here... we can
	 * allow the pager to do it after the I/O completes.
	 *
	 * NOTE! mc[i]->dirty may be partial or fragmented due to an
	 * edge case with file fragments.
	 */
	for (i = 0; i < count; i++) {
		KASSERT(mc[i]->valid == VM_PAGE_BITS_ALL,
		    ("vm_pageout_flush: partially invalid page %p index %d/%d",
			mc[i], i, count));
		vm_page_sbusy(mc[i]);
		pmap_remove_write(mc[i]);
	}
	vm_object_pip_add(object, count);

	vm_pager_put_pages(object, mc, count, flags, pageout_status);

	runlen = count - mreq;
	if (eio != NULL)
		*eio = FALSE;
	for (i = 0; i < count; i++) {
		vm_page_t mt = mc[i];

		KASSERT(pageout_status[i] == VM_PAGER_PEND ||
		    !pmap_page_is_write_mapped(mt),
		    ("vm_pageout_flush: page %p is not write protected", mt));
		switch (pageout_status[i]) {
		case VM_PAGER_OK:
		case VM_PAGER_PEND:
			numpagedout++;
			break;
		case VM_PAGER_BAD:
			/*
			 * Page outside of range of object. Right now we
			 * essentially lose the changes by pretending it
			 * worked.
			 */
			vm_page_undirty(mt);
			break;
		case VM_PAGER_ERROR:
		case VM_PAGER_FAIL:
			/*
			 * If page couldn't be paged out, then reactivate the
			 * page so it doesn't clog the inactive list.  (We
			 * will try paging out it again later).
			 */
			vm_page_lock(mt);
			vm_page_activate(mt);
			vm_page_unlock(mt);
			if (eio != NULL && i >= mreq && i - mreq < runlen)
				*eio = TRUE;
			break;
		case VM_PAGER_AGAIN:
			if (i >= mreq && i - mreq < runlen)
				runlen = i - mreq;
			break;
		}

		/*
		 * If the operation is still going, leave the page busy to
		 * block all other accesses. Also, leave the paging in
		 * progress indicator set so that we don't attempt an object
		 * collapse.
		 */
		if (pageout_status[i] != VM_PAGER_PEND) {
			vm_object_pip_wakeup(object);
			vm_page_sunbusy(mt);
		}
	}
	if (prunlen != NULL)
		*prunlen = runlen;
	return (numpagedout);
}

#if !defined(NO_SWAPPING)
/*
 *	vm_pageout_object_deactivate_pages
 *
 *	Deactivate enough pages to satisfy the inactive target
 *	requirements.
 *
 *	The object and map must be locked.
 */
static void
vm_pageout_object_deactivate_pages(pmap_t pmap, vm_object_t first_object,
    long desired)
{
	vm_object_t backing_object, object;
	vm_page_t p;
	int act_delta, remove_mode;

	VM_OBJECT_ASSERT_LOCKED(first_object);
	if ((first_object->flags & OBJ_FICTITIOUS) != 0)
		return;
	for (object = first_object;; object = backing_object) {
		if (pmap_resident_count(pmap) <= desired)
			goto unlock_return;
		VM_OBJECT_ASSERT_LOCKED(object);
		if ((object->flags & OBJ_UNMANAGED) != 0 ||
		    object->paging_in_progress != 0)
			goto unlock_return;

		remove_mode = 0;
		if (object->shadow_count > 1)
			remove_mode = 1;
		/*
		 * Scan the object's entire memory queue.
		 */
		TAILQ_FOREACH(p, &object->memq, listq) {
			if (pmap_resident_count(pmap) <= desired)
				goto unlock_return;
			if (vm_page_busied(p))
				continue;
			PCPU_INC(cnt.v_pdpages);
			vm_page_lock(p);
			if (p->wire_count != 0 || p->hold_count != 0 ||
			    !pmap_page_exists_quick(pmap, p)) {
				vm_page_unlock(p);
				continue;
			}
			act_delta = pmap_ts_referenced(p);
			if ((p->aflags & PGA_REFERENCED) != 0) {
				if (act_delta == 0)
					act_delta = 1;
				vm_page_aflag_clear(p, PGA_REFERENCED);
			}
			if (p->queue != PQ_ACTIVE && act_delta != 0) {
				vm_page_activate(p);
				p->act_count += act_delta;
			} else if (p->queue == PQ_ACTIVE) {
				if (act_delta == 0) {
					p->act_count -= min(p->act_count,
					    ACT_DECLINE);
					if (!remove_mode && p->act_count == 0) {
						pmap_remove_all(p);
						vm_page_deactivate(p);
					} else
						vm_page_requeue(p);
				} else {
					vm_page_activate(p);
					if (p->act_count < ACT_MAX -
					    ACT_ADVANCE)
						p->act_count += ACT_ADVANCE;
					vm_page_requeue(p);
				}
			} else if (p->queue == PQ_INACTIVE)
				pmap_remove_all(p);
			vm_page_unlock(p);
		}
		if ((backing_object = object->backing_object) == NULL)
			goto unlock_return;
		VM_OBJECT_RLOCK(backing_object);
		if (object != first_object)
			VM_OBJECT_RUNLOCK(object);
	}
unlock_return:
	if (object != first_object)
		VM_OBJECT_RUNLOCK(object);
}

/*
 * deactivate some number of pages in a map, try to do it fairly, but
 * that is really hard to do.
 */
static void
vm_pageout_map_deactivate_pages(map, desired)
	vm_map_t map;
	long desired;
{
	vm_map_entry_t tmpe;
	vm_object_t obj, bigobj;
	int nothingwired;

	if (!vm_map_trylock(map))
		return;

	bigobj = NULL;
	nothingwired = TRUE;

	/*
	 * first, search out the biggest object, and try to free pages from
	 * that.
	 */
	tmpe = map->header.next;
	while (tmpe != &map->header) {
		if ((tmpe->eflags & MAP_ENTRY_IS_SUB_MAP) == 0) {
			obj = tmpe->object.vm_object;
			if (obj != NULL && VM_OBJECT_TRYRLOCK(obj)) {
				if (obj->shadow_count <= 1 &&
				    (bigobj == NULL ||
				     bigobj->resident_page_count < obj->resident_page_count)) {
					if (bigobj != NULL)
						VM_OBJECT_RUNLOCK(bigobj);
					bigobj = obj;
				} else
					VM_OBJECT_RUNLOCK(obj);
			}
		}
		if (tmpe->wired_count > 0)
			nothingwired = FALSE;
		tmpe = tmpe->next;
	}

	if (bigobj != NULL) {
		vm_pageout_object_deactivate_pages(map->pmap, bigobj, desired);
		VM_OBJECT_RUNLOCK(bigobj);
	}
	/*
	 * Next, hunt around for other pages to deactivate.  We actually
	 * do this search sort of wrong -- .text first is not the best idea.
	 */
	tmpe = map->header.next;
	while (tmpe != &map->header) {
		if (pmap_resident_count(vm_map_pmap(map)) <= desired)
			break;
		if ((tmpe->eflags & MAP_ENTRY_IS_SUB_MAP) == 0) {
			obj = tmpe->object.vm_object;
			if (obj != NULL) {
				VM_OBJECT_RLOCK(obj);
				vm_pageout_object_deactivate_pages(map->pmap, obj, desired);
				VM_OBJECT_RUNLOCK(obj);
			}
		}
		tmpe = tmpe->next;
	}

	/*
	 * Remove all mappings if a process is swapped out, this will free page
	 * table pages.
	 */
	if (desired == 0 && nothingwired) {
		pmap_remove(vm_map_pmap(map), vm_map_min(map),
		    vm_map_max(map));
	}

	vm_map_unlock(map);
}
#endif		/* !defined(NO_SWAPPING) */

/*
 * Attempt to acquire all of the necessary locks to launder a page and
 * then call through the clustering layer to PUTPAGES.  Wait a short
 * time for a vnode lock.
 *
 * Requires the page and object lock on entry, releases both before return.
 * Returns 0 on success and an errno otherwise.
 */
static int
vm_pageout_clean(vm_page_t m)
{
	struct vnode *vp;
	struct mount *mp;
	vm_object_t object;
	vm_pindex_t pindex;
	int error, lockmode;

	vm_page_assert_locked(m);
	object = m->object;
	VM_OBJECT_ASSERT_WLOCKED(object);
	error = 0;
	vp = NULL;
	mp = NULL;

	/*
	 * The object is already known NOT to be dead.   It
	 * is possible for the vget() to block the whole
	 * pageout daemon, but the new low-memory handling
	 * code should prevent it.
	 *
	 * We can't wait forever for the vnode lock, we might
	 * deadlock due to a vn_read() getting stuck in
	 * vm_wait while holding this vnode.  We skip the 
	 * vnode if we can't get it in a reasonable amount
	 * of time.
	 */
	if (object->type == OBJT_VNODE) {
		vm_page_unlock(m);
		vp = object->handle;
		if (vp->v_type == VREG &&
		    vn_start_write(vp, &mp, V_NOWAIT) != 0) {
			mp = NULL;
			error = EDEADLK;
			goto unlock_all;
		}
		KASSERT(mp != NULL,
		    ("vp %p with NULL v_mount", vp));
		vm_object_reference_locked(object);
		pindex = m->pindex;
		VM_OBJECT_WUNLOCK(object);
		lockmode = MNT_SHARED_WRITES(vp->v_mount) ?
		    LK_SHARED : LK_EXCLUSIVE;
		if (vget(vp, lockmode | LK_TIMELOCK, curthread)) {
			vp = NULL;
			error = EDEADLK;
			goto unlock_mp;
		}
		VM_OBJECT_WLOCK(object);
		vm_page_lock(m);
		/*
		 * While the object and page were unlocked, the page
		 * may have been:
		 * (1) moved to a different queue,
		 * (2) reallocated to a different object,
		 * (3) reallocated to a different offset, or
		 * (4) cleaned.
		 */
		if (m->queue != PQ_INACTIVE || m->object != object ||
		    m->pindex != pindex || m->dirty == 0) {
			vm_page_unlock(m);
			error = ENXIO;
			goto unlock_all;
		}

		/*
		 * The page may have been busied or held while the object
		 * and page locks were released.
		 */
		if (vm_page_busied(m) || m->hold_count != 0) {
			vm_page_unlock(m);
			error = EBUSY;
			goto unlock_all;
		}
	}

	/*
	 * If a page is dirty, then it is either being washed
	 * (but not yet cleaned) or it is still in the
	 * laundry.  If it is still in the laundry, then we
	 * start the cleaning operation. 
	 */
	if (vm_pageout_cluster(m) == 0)
		error = EIO;

unlock_all:
	VM_OBJECT_WUNLOCK(object);

unlock_mp:
	vm_page_lock_assert(m, MA_NOTOWNED);
	if (mp != NULL) {
		if (vp != NULL)
			vput(vp);
		vm_object_deallocate(object);
		vn_finished_write(mp);
	}

	return (error);
}

/*
 *	vm_pageout_scan does the dirty work for the pageout daemon.
 *
 *	pass 0 - Update active LRU/deactivate pages
 *	pass 1 - Move inactive to cache or free
 *	pass 2 - Launder dirty pages
 */
static void
vm_pageout_scan(struct vm_domain *vmd, int pass)
{
	vm_page_t m, next;
	struct vm_pagequeue *pq;
	vm_object_t object;
	long min_scan;
	int act_delta, addl_page_shortage, deficit, error, maxlaunder, maxscan;
	int page_shortage, scan_tick, scanned, starting_page_shortage;
	int vnodes_skipped;
	boolean_t pageout_ok, queues_locked;

	/*
	 * If we need to reclaim memory ask kernel caches to return
	 * some.  We rate limit to avoid thrashing.
	 */
	if (vmd == &vm_dom[0] && pass > 0 &&
	    (time_uptime - lowmem_uptime) >= lowmem_period) {
		/*
		 * Decrease registered cache sizes.
		 */
		SDT_PROBE0(vm, , , vm__lowmem_scan);
		EVENTHANDLER_INVOKE(vm_lowmem, 0);
		/*
		 * We do this explicitly after the caches have been
		 * drained above.
		 */
		uma_reclaim();
		lowmem_uptime = time_uptime;
	}

	/*
	 * The addl_page_shortage is the number of temporarily
	 * stuck pages in the inactive queue.  In other words, the
	 * number of pages from the inactive count that should be
	 * discounted in setting the target for the active queue scan.
	 */
	addl_page_shortage = 0;

	/*
	 * Calculate the number of pages we want to either free or move
	 * to the cache.
	 */
	if (pass > 0) {
		deficit = atomic_readandclear_int(&vm_pageout_deficit);
		page_shortage = vm_paging_target() + deficit;
	} else
		page_shortage = deficit = 0;
	starting_page_shortage = page_shortage;

	/*
	 * maxlaunder limits the number of dirty pages we flush per scan.
	 * For most systems a smaller value (16 or 32) is more robust under
	 * extreme memory and disk pressure because any unnecessary writes
	 * to disk can result in extreme performance degredation.  However,
	 * systems with excessive dirty pages (especially when MAP_NOSYNC is
	 * used) will die horribly with limited laundering.  If the pageout
	 * daemon cannot clean enough pages in the first pass, we let it go
	 * all out in succeeding passes.
	 */
	if ((maxlaunder = vm_max_launder) <= 1)
		maxlaunder = 1;
	if (pass > 1)
		maxlaunder = 10000;

	vnodes_skipped = 0;

	/*
	 * Start scanning the inactive queue for pages we can move to the
	 * cache or free.  The scan will stop when the target is reached or
	 * we have scanned the entire inactive queue.  Note that m->act_count
	 * is not used to form decisions for the inactive queue, only for the
	 * active queue.
	 */
	pq = &vmd->vmd_pagequeues[PQ_INACTIVE];
	maxscan = pq->pq_cnt;
	vm_pagequeue_lock(pq);
	queues_locked = TRUE;
	for (m = TAILQ_FIRST(&pq->pq_pl);
	     m != NULL && maxscan-- > 0 && page_shortage > 0;
	     m = next) {
		vm_pagequeue_assert_locked(pq);
		KASSERT(queues_locked, ("unlocked queues"));
		KASSERT(m->queue == PQ_INACTIVE, ("Inactive queue %p", m));

		PCPU_INC(cnt.v_pdpages);
		next = TAILQ_NEXT(m, plinks.q);

		/*
		 * skip marker pages
		 */
		if (m->flags & PG_MARKER)
			continue;

		KASSERT((m->flags & PG_FICTITIOUS) == 0,
		    ("Fictitious page %p cannot be in inactive queue", m));
		KASSERT((m->oflags & VPO_UNMANAGED) == 0,
		    ("Unmanaged page %p cannot be in inactive queue", m));

		/*
		 * The page or object lock acquisitions fail if the
		 * page was removed from the queue or moved to a
		 * different position within the queue.  In either
		 * case, addl_page_shortage should not be incremented.
		 */
		if (!vm_pageout_page_lock(m, &next))
			goto unlock_page;
		else if (m->hold_count != 0) {
			/*
			 * Held pages are essentially stuck in the
			 * queue.  So, they ought to be discounted
			 * from the inactive count.  See the
			 * calculation of the page_shortage for the
			 * loop over the active queue below.
			 */
			addl_page_shortage++;
			goto unlock_page;
		}
		object = m->object;
		if (!VM_OBJECT_TRYWLOCK(object)) {
			if (!vm_pageout_fallback_object_lock(m, &next))
				goto unlock_object;
			else if (m->hold_count != 0) {
				addl_page_shortage++;
				goto unlock_object;
			}
		}
		if (vm_page_busied(m)) {
			/*
			 * Don't mess with busy pages.  Leave them at
			 * the front of the queue.  Most likely, they
			 * are being paged out and will leave the
			 * queue shortly after the scan finishes.  So,
			 * they ought to be discounted from the
			 * inactive count.
			 */
			addl_page_shortage++;
unlock_object:
			VM_OBJECT_WUNLOCK(object);
unlock_page:
			vm_page_unlock(m);
			continue;
		}
		KASSERT(m->hold_count == 0, ("Held page %p", m));

		/*
		 * We unlock the inactive page queue, invalidating the
		 * 'next' pointer.  Use our marker to remember our
		 * place.
		 */
		TAILQ_INSERT_AFTER(&pq->pq_pl, m, &vmd->vmd_marker, plinks.q);
		vm_pagequeue_unlock(pq);
		queues_locked = FALSE;

		/*
		 * Invalid pages can be easily freed. They cannot be
		 * mapped, vm_page_free() asserts this.
		 */
		if (m->valid == 0)
			goto free_page;

		/*
		 * If the page has been referenced and the object is not dead,
		 * reactivate or requeue the page depending on whether the
		 * object is mapped.
		 */
		if ((m->aflags & PGA_REFERENCED) != 0) {
			vm_page_aflag_clear(m, PGA_REFERENCED);
			act_delta = 1;
		} else
			act_delta = 0;
		if (object->ref_count != 0) {
			act_delta += pmap_ts_referenced(m);
		} else {
			KASSERT(!pmap_page_is_mapped(m),
			    ("vm_pageout_scan: page %p is mapped", m));
		}
		if (act_delta != 0) {
			if (object->ref_count != 0) {
				vm_page_activate(m);

				/*
				 * Increase the activation count if the page
				 * was referenced while in the inactive queue.
				 * This makes it less likely that the page will
				 * be returned prematurely to the inactive
				 * queue.
 				 */
				m->act_count += act_delta + ACT_ADVANCE;
				goto drop_page;
			} else if ((object->flags & OBJ_DEAD) == 0)
				goto requeue_page;
		}

		/*
		 * If the page appears to be clean at the machine-independent
		 * layer, then remove all of its mappings from the pmap in
		 * anticipation of placing it onto the cache queue.  If,
		 * however, any of the page's mappings allow write access,
		 * then the page may still be modified until the last of those
		 * mappings are removed.
		 */
		if (object->ref_count != 0) {
			vm_page_test_dirty(m);
			if (m->dirty == 0)
				pmap_remove_all(m);
		}

		if (m->dirty == 0) {
			/*
			 * Clean pages can be freed.
			 */
free_page:
			vm_page_free(m);
			PCPU_INC(cnt.v_dfree);
			--page_shortage;
		} else if ((object->flags & OBJ_DEAD) != 0) {
			/*
			 * Leave dirty pages from dead objects at the front of
			 * the queue.  They are being paged out and freed by
			 * the thread that destroyed the object.  They will
			 * leave the queue shortly after the scan finishes, so 
			 * they should be discounted from the inactive count.
			 */
			addl_page_shortage++;
		} else if ((m->flags & PG_WINATCFLS) == 0 && pass < 2) {
			/*
			 * Dirty pages need to be paged out, but flushing
			 * a page is extremely expensive versus freeing
			 * a clean page.  Rather then artificially limiting
			 * the number of pages we can flush, we instead give
			 * dirty pages extra priority on the inactive queue
			 * by forcing them to be cycled through the queue
			 * twice before being flushed, after which the
			 * (now clean) page will cycle through once more
			 * before being freed.  This significantly extends
			 * the thrash point for a heavily loaded machine.
			 */
			m->flags |= PG_WINATCFLS;
requeue_page:
			vm_pagequeue_lock(pq);
			queues_locked = TRUE;
			vm_page_requeue_locked(m);
		} else if (maxlaunder > 0) {
			/*
			 * We always want to try to flush some dirty pages if
			 * we encounter them, to keep the system stable.
			 * Normally this number is small, but under extreme
			 * pressure where there are insufficient clean pages
			 * on the inactive queue, we may have to go all out.
			 */

			if (object->type != OBJT_SWAP &&
			    object->type != OBJT_DEFAULT)
				pageout_ok = TRUE;
			else if (disable_swap_pageouts)
				pageout_ok = FALSE;
			else if (defer_swap_pageouts)
				pageout_ok = vm_page_count_min();
			else
				pageout_ok = TRUE;
			if (!pageout_ok)
				goto requeue_page;
			error = vm_pageout_clean(m);
			/*
			 * Decrement page_shortage on success to account for
			 * the (future) cleaned page.  Otherwise we could wind
			 * up laundering or cleaning too many pages.
			 */
			if (error == 0) {
				page_shortage--;
				maxlaunder--;
			} else if (error == EDEADLK) {
				pageout_lock_miss++;
				vnodes_skipped++;
			} else if (error == EBUSY) {
				addl_page_shortage++;
			}
			vm_page_lock_assert(m, MA_NOTOWNED);
			goto relock_queues;
		}
drop_page:
		vm_page_unlock(m);
		VM_OBJECT_WUNLOCK(object);
relock_queues:
		if (!queues_locked) {
			vm_pagequeue_lock(pq);
			queues_locked = TRUE;
		}
		next = TAILQ_NEXT(&vmd->vmd_marker, plinks.q);
		TAILQ_REMOVE(&pq->pq_pl, &vmd->vmd_marker, plinks.q);
	}
	vm_pagequeue_unlock(pq);

#if !defined(NO_SWAPPING)
	/*
	 * Wakeup the swapout daemon if we didn't cache or free the targeted
	 * number of pages. 
	 */
	if (vm_swap_enabled && page_shortage > 0)
		vm_req_vmdaemon(VM_SWAP_NORMAL);
#endif

	/*
	 * Wakeup the sync daemon if we skipped a vnode in a writeable object
	 * and we didn't cache or free enough pages.
	 */
	if (vnodes_skipped > 0 && page_shortage > vm_cnt.v_free_target -
	    vm_cnt.v_free_min)
		(void)speedup_syncer();

	/*
	 * If the inactive queue scan fails repeatedly to meet its
	 * target, kill the largest process.
	 */
	vm_pageout_mightbe_oom(vmd, page_shortage, starting_page_shortage);

	/*
	 * Compute the number of pages we want to try to move from the
	 * active queue to the inactive queue.
	 */
	page_shortage = vm_cnt.v_inactive_target - vm_cnt.v_inactive_count +
	    vm_paging_target() + deficit + addl_page_shortage;

	pq = &vmd->vmd_pagequeues[PQ_ACTIVE];
	vm_pagequeue_lock(pq);
	maxscan = pq->pq_cnt;

	/*
	 * If we're just idle polling attempt to visit every
	 * active page within 'update_period' seconds.
	 */
	scan_tick = ticks;
	if (vm_pageout_update_period != 0) {
		min_scan = pq->pq_cnt;
		min_scan *= scan_tick - vmd->vmd_last_active_scan;
		min_scan /= hz * vm_pageout_update_period;
	} else
		min_scan = 0;
	if (min_scan > 0 || (page_shortage > 0 && maxscan > 0))
		vmd->vmd_last_active_scan = scan_tick;

	/*
	 * Scan the active queue for pages that can be deactivated.  Update
	 * the per-page activity counter and use it to identify deactivation
	 * candidates.
	 */
	for (m = TAILQ_FIRST(&pq->pq_pl), scanned = 0; m != NULL && (scanned <
	    min_scan || (page_shortage > 0 && scanned < maxscan)); m = next,
	    scanned++) {

		KASSERT(m->queue == PQ_ACTIVE,
		    ("vm_pageout_scan: page %p isn't active", m));

		next = TAILQ_NEXT(m, plinks.q);
		if ((m->flags & PG_MARKER) != 0)
			continue;
		KASSERT((m->flags & PG_FICTITIOUS) == 0,
		    ("Fictitious page %p cannot be in active queue", m));
		KASSERT((m->oflags & VPO_UNMANAGED) == 0,
		    ("Unmanaged page %p cannot be in active queue", m));
		if (!vm_pageout_page_lock(m, &next)) {
			vm_page_unlock(m);
			continue;
		}

		/*
		 * The count for pagedaemon pages is done after checking the
		 * page for eligibility...
		 */
		PCPU_INC(cnt.v_pdpages);

		/*
		 * Check to see "how much" the page has been used.
		 */
		if ((m->aflags & PGA_REFERENCED) != 0) {
			vm_page_aflag_clear(m, PGA_REFERENCED);
			act_delta = 1;
		} else
			act_delta = 0;

		/*
		 * Unlocked object ref count check.  Two races are possible.
		 * 1) The ref was transitioning to zero and we saw non-zero,
		 *    the pmap bits will be checked unnecessarily.
		 * 2) The ref was transitioning to one and we saw zero. 
		 *    The page lock prevents a new reference to this page so
		 *    we need not check the reference bits.
		 */
		if (m->object->ref_count != 0)
			act_delta += pmap_ts_referenced(m);

		/*
		 * Advance or decay the act_count based on recent usage.
		 */
		if (act_delta != 0) {
			m->act_count += ACT_ADVANCE + act_delta;
			if (m->act_count > ACT_MAX)
				m->act_count = ACT_MAX;
		} else
			m->act_count -= min(m->act_count, ACT_DECLINE);

		/*
		 * Move this page to the tail of the active or inactive
		 * queue depending on usage.
		 */
		if (m->act_count == 0) {
			/* Dequeue to avoid later lock recursion. */
			vm_page_dequeue_locked(m);
			vm_page_deactivate(m);
			page_shortage--;
		} else
			vm_page_requeue_locked(m);
		vm_page_unlock(m);
	}
	vm_pagequeue_unlock(pq);
#if !defined(NO_SWAPPING)
	/*
	 * Idle process swapout -- run once per second.
	 */
	if (vm_swap_idle_enabled) {
		static long lsec;
		if (time_second != lsec) {
			vm_req_vmdaemon(VM_SWAP_IDLE);
			lsec = time_second;
		}
	}
#endif
}

static int vm_pageout_oom_vote;

/*
 * The pagedaemon threads randlomly select one to perform the
 * OOM.  Trying to kill processes before all pagedaemons
 * failed to reach free target is premature.
 */
static void
vm_pageout_mightbe_oom(struct vm_domain *vmd, int page_shortage,
    int starting_page_shortage)
{
	int old_vote;

	if (starting_page_shortage <= 0 || starting_page_shortage !=
	    page_shortage)
		vmd->vmd_oom_seq = 0;
	else
		vmd->vmd_oom_seq++;
	if (vmd->vmd_oom_seq < vm_pageout_oom_seq) {
		if (vmd->vmd_oom) {
			vmd->vmd_oom = FALSE;
			atomic_subtract_int(&vm_pageout_oom_vote, 1);
		}
		return;
	}

	/*
	 * Do not follow the call sequence until OOM condition is
	 * cleared.
	 */
	vmd->vmd_oom_seq = 0;

	if (vmd->vmd_oom)
		return;

	vmd->vmd_oom = TRUE;
	old_vote = atomic_fetchadd_int(&vm_pageout_oom_vote, 1);
	if (old_vote != vm_ndomains - 1)
		return;

	/*
	 * The current pagedaemon thread is the last in the quorum to
	 * start OOM.  Initiate the selection and signaling of the
	 * victim.
	 */
	vm_pageout_oom(VM_OOM_MEM);

	/*
	 * After one round of OOM terror, recall our vote.  On the
	 * next pass, current pagedaemon would vote again if the low
	 * memory condition is still there, due to vmd_oom being
	 * false.
	 */
	vmd->vmd_oom = FALSE;
	atomic_subtract_int(&vm_pageout_oom_vote, 1);
}

/*
 * The OOM killer is the page daemon's action of last resort when
 * memory allocation requests have been stalled for a prolonged period
 * of time because it cannot reclaim memory.  This function computes
 * the approximate number of physical pages that could be reclaimed if
 * the specified address space is destroyed.
 *
 * Private, anonymous memory owned by the address space is the
 * principal resource that we expect to recover after an OOM kill.
 * Since the physical pages mapped by the address space's COW entries
 * are typically shared pages, they are unlikely to be released and so
 * they are not counted.
 *
 * To get to the point where the page daemon runs the OOM killer, its
 * efforts to write-back vnode-backed pages may have stalled.  This
 * could be caused by a memory allocation deadlock in the write path
 * that might be resolved by an OOM kill.  Therefore, physical pages
 * belonging to vnode-backed objects are counted, because they might
 * be freed without being written out first if the address space holds
 * the last reference to an unlinked vnode.
 *
 * Similarly, physical pages belonging to OBJT_PHYS objects are
 * counted because the address space might hold the last reference to
 * the object.
 */
static long
vm_pageout_oom_pagecount(struct vmspace *vmspace)
{
	vm_map_t map;
	vm_map_entry_t entry;
	vm_object_t obj;
	long res;

	map = &vmspace->vm_map;
	KASSERT(!map->system_map, ("system map"));
	sx_assert(&map->lock, SA_LOCKED);
	res = 0;
	for (entry = map->header.next; entry != &map->header;
	    entry = entry->next) {
		if ((entry->eflags & MAP_ENTRY_IS_SUB_MAP) != 0)
			continue;
		obj = entry->object.vm_object;
		if (obj == NULL)
			continue;
		if ((entry->eflags & MAP_ENTRY_NEEDS_COPY) != 0 &&
		    obj->ref_count != 1)
			continue;
		switch (obj->type) {
		case OBJT_DEFAULT:
		case OBJT_SWAP:
		case OBJT_PHYS:
		case OBJT_VNODE:
			res += obj->resident_page_count;
			break;
		}
	}
	return (res);
}

void
vm_pageout_oom(int shortage)
{
	struct proc *p, *bigproc;
	vm_offset_t size, bigsize;
	struct thread *td;
	struct vmspace *vm;

	/*
	 * We keep the process bigproc locked once we find it to keep anyone
	 * from messing with it; however, there is a possibility of
	 * deadlock if process B is bigproc and one of it's child processes
	 * attempts to propagate a signal to B while we are waiting for A's
	 * lock while walking this list.  To avoid this, we don't block on
	 * the process lock but just skip a process if it is already locked.
	 */
	bigproc = NULL;
	bigsize = 0;
	sx_slock(&allproc_lock);
	FOREACH_PROC_IN_SYSTEM(p) {
		int breakout;

		PROC_LOCK(p);

		/*
		 * If this is a system, protected or killed process, skip it.
		 */
		if (p->p_state != PRS_NORMAL || (p->p_flag & (P_INEXEC |
		    P_PROTECTED | P_SYSTEM | P_WEXIT)) != 0 ||
		    p->p_pid == 1 || P_KILLED(p) ||
		    (p->p_pid < 48 && swap_pager_avail != 0)) {
			PROC_UNLOCK(p);
			continue;
		}
		/*
		 * If the process is in a non-running type state,
		 * don't touch it.  Check all the threads individually.
		 */
		breakout = 0;
		FOREACH_THREAD_IN_PROC(p, td) {
			thread_lock(td);
			if (!TD_ON_RUNQ(td) &&
			    !TD_IS_RUNNING(td) &&
			    !TD_IS_SLEEPING(td) &&
			    !TD_IS_SUSPENDED(td) &&
			    !TD_IS_SWAPPED(td)) {
				thread_unlock(td);
				breakout = 1;
				break;
			}
			thread_unlock(td);
		}
		if (breakout) {
			PROC_UNLOCK(p);
			continue;
		}
		/*
		 * get the process size
		 */
		vm = vmspace_acquire_ref(p);
		if (vm == NULL) {
			PROC_UNLOCK(p);
			continue;
		}
		_PHOLD(p);
		if (!vm_map_trylock_read(&vm->vm_map)) {
			_PRELE(p);
			PROC_UNLOCK(p);
			vmspace_free(vm);
			continue;
		}
		PROC_UNLOCK(p);
		size = vmspace_swap_count(vm);
		if (shortage == VM_OOM_MEM)
			size += vm_pageout_oom_pagecount(vm);
		vm_map_unlock_read(&vm->vm_map);
		vmspace_free(vm);

		/*
		 * If this process is bigger than the biggest one,
		 * remember it.
		 */
		if (size > bigsize) {
			if (bigproc != NULL)
				PRELE(bigproc);
			bigproc = p;
			bigsize = size;
		} else {
			PRELE(p);
		}
	}
	sx_sunlock(&allproc_lock);
	if (bigproc != NULL) {
		if (vm_panic_on_oom != 0)
			panic("out of swap space");
		PROC_LOCK(bigproc);
		killproc(bigproc, "out of swap space");
		sched_nice(bigproc, PRIO_MIN);
		_PRELE(bigproc);
		PROC_UNLOCK(bigproc);
		wakeup(&vm_cnt.v_free_count);
	}
}

static void
vm_pageout_worker(void *arg)
{
	struct vm_domain *domain;
	int domidx;

	domidx = (uintptr_t)arg;
	domain = &vm_dom[domidx];

	/*
	 * XXXKIB It could be useful to bind pageout daemon threads to
	 * the cores belonging to the domain, from which vm_page_array
	 * is allocated.
	 */

	KASSERT(domain->vmd_segs != 0, ("domain without segments"));
	domain->vmd_last_active_scan = ticks;
	vm_pageout_init_marker(&domain->vmd_marker, PQ_INACTIVE);
	vm_pageout_init_marker(&domain->vmd_inacthead, PQ_INACTIVE);
	TAILQ_INSERT_HEAD(&domain->vmd_pagequeues[PQ_INACTIVE].pq_pl,
	    &domain->vmd_inacthead, plinks.q);

	/*
	 * The pageout daemon worker is never done, so loop forever.
	 */
	while (TRUE) {
		/*
		 * If we have enough free memory, wakeup waiters.  Do
		 * not clear vm_pages_needed until we reach our target,
		 * otherwise we may be woken up over and over again and
		 * waste a lot of cpu.
		 */
		mtx_lock(&vm_page_queue_free_mtx);
		if (vm_pages_needed && !vm_page_count_min()) {
			if (!vm_paging_needed())
				vm_pages_needed = 0;
			wakeup(&vm_cnt.v_free_count);
		}
		if (vm_pages_needed) {
			/*
			 * We're still not done.  Either vm_pages_needed was
			 * set by another thread during the previous scan
			 * (typically, this happens during a level 0 scan) or
			 * vm_pages_needed was already set and the scan failed
			 * to free enough pages.  If we haven't yet performed
			 * a level >= 2 scan (unlimited dirty cleaning), then
			 * upgrade the level and scan again now.  Otherwise,
			 * sleep a bit and try again later.  While sleeping,
			 * vm_pages_needed can be cleared.
			 */
			if (domain->vmd_pass > 1)
				msleep(&vm_pages_needed,
				    &vm_page_queue_free_mtx, PVM, "psleep",
				    hz / 2);
		} else {
			/*
			 * Good enough, sleep until required to refresh
			 * stats.
			 */
			msleep(&vm_pages_needed, &vm_page_queue_free_mtx,
			    PVM, "psleep", hz);
		}
		if (vm_pages_needed) {
			vm_cnt.v_pdwakeups++;
			domain->vmd_pass++;
		} else
			domain->vmd_pass = 0;
		mtx_unlock(&vm_page_queue_free_mtx);
		vm_pageout_scan(domain, domain->vmd_pass);
	}
}

/*
 *	vm_pageout_init initialises basic pageout daemon settings.
 */
static void
vm_pageout_init(void)
{
	/*
	 * Initialize some paging parameters.
	 */
	vm_cnt.v_interrupt_free_min = 2;
	if (vm_cnt.v_page_count < 2000)
		vm_pageout_page_count = 8;

	/*
	 * v_free_reserved needs to include enough for the largest
	 * swap pager structures plus enough for any pv_entry structs
	 * when paging. 
	 */
	if (vm_cnt.v_page_count > 1024)
		vm_cnt.v_free_min = 4 + (vm_cnt.v_page_count - 1024) / 200;
	else
		vm_cnt.v_free_min = 4;
	vm_cnt.v_pageout_free_min = (2*MAXBSIZE)/PAGE_SIZE +
	    vm_cnt.v_interrupt_free_min;
	vm_cnt.v_free_reserved = vm_pageout_page_count +
	    vm_cnt.v_pageout_free_min + (vm_cnt.v_page_count / 768);
	vm_cnt.v_free_severe = vm_cnt.v_free_min / 2;
	vm_cnt.v_free_target = 4 * vm_cnt.v_free_min + vm_cnt.v_free_reserved;
	vm_cnt.v_free_min += vm_cnt.v_free_reserved;
	vm_cnt.v_free_severe += vm_cnt.v_free_reserved;
	vm_cnt.v_inactive_target = (3 * vm_cnt.v_free_target) / 2;
	if (vm_cnt.v_inactive_target > vm_cnt.v_free_count / 3)
		vm_cnt.v_inactive_target = vm_cnt.v_free_count / 3;

	/*
	 * Set the default wakeup threshold to be 10% above the minimum
	 * page limit.  This keeps the steady state out of shortfall.
	 */
	vm_pageout_wakeup_thresh = (vm_cnt.v_free_min / 10) * 11;

	/*
	 * Set interval in seconds for active scan.  We want to visit each
	 * page at least once every ten minutes.  This is to prevent worst
	 * case paging behaviors with stale active LRU.
	 */
	if (vm_pageout_update_period == 0)
		vm_pageout_update_period = 600;

	/* XXX does not really belong here */
	if (vm_page_max_wired == 0)
		vm_page_max_wired = vm_cnt.v_free_count / 3;
}

/*
 *     vm_pageout is the high level pageout daemon.
 */
static void
vm_pageout(void)
{
	int error;
#if MAXMEMDOM > 1
	int i;
#endif

	swap_pager_swap_init();
#if MAXMEMDOM > 1
	for (i = 1; i < vm_ndomains; i++) {
		error = kthread_add(vm_pageout_worker, (void *)(uintptr_t)i,
		    curproc, NULL, 0, 0, "dom%d", i);
		if (error != 0) {
			panic("starting pageout for domain %d, error %d\n",
			    i, error);
		}
	}
#endif
	error = kthread_add(uma_reclaim_worker, NULL, curproc, NULL,
	    0, 0, "uma");
	if (error != 0)
		panic("starting uma_reclaim helper, error %d\n", error);
	vm_pageout_worker((void *)(uintptr_t)0);
}

/*
 * Unless the free page queue lock is held by the caller, this function
 * should be regarded as advisory.  Specifically, the caller should
 * not msleep() on &vm_cnt.v_free_count following this function unless
 * the free page queue lock is held until the msleep() is performed.
 */
void
pagedaemon_wakeup(void)
{

	if (!vm_pages_needed && curthread->td_proc != pageproc) {
		vm_pages_needed = 1;
		wakeup(&vm_pages_needed);
	}
}

#if !defined(NO_SWAPPING)
static void
vm_req_vmdaemon(int req)
{
	static int lastrun = 0;

	mtx_lock(&vm_daemon_mtx);
	vm_pageout_req_swapout |= req;
	if ((ticks > (lastrun + hz)) || (ticks < lastrun)) {
		wakeup(&vm_daemon_needed);
		lastrun = ticks;
	}
	mtx_unlock(&vm_daemon_mtx);
}

static void
vm_daemon(void)
{
	struct rlimit rsslim;
	struct proc *p;
	struct thread *td;
	struct vmspace *vm;
	int breakout, swapout_flags, tryagain, attempts;
#ifdef RACCT
	uint64_t rsize, ravailable;
#endif

	while (TRUE) {
		mtx_lock(&vm_daemon_mtx);
		msleep(&vm_daemon_needed, &vm_daemon_mtx, PPAUSE, "psleep",
#ifdef RACCT
		    racct_enable ? hz : 0
#else
		    0
#endif
		);
		swapout_flags = vm_pageout_req_swapout;
		vm_pageout_req_swapout = 0;
		mtx_unlock(&vm_daemon_mtx);
		if (swapout_flags)
			swapout_procs(swapout_flags);

		/*
		 * scan the processes for exceeding their rlimits or if
		 * process is swapped out -- deactivate pages
		 */
		tryagain = 0;
		attempts = 0;
again:
		attempts++;
		sx_slock(&allproc_lock);
		FOREACH_PROC_IN_SYSTEM(p) {
			vm_pindex_t limit, size;

			/*
			 * if this is a system process or if we have already
			 * looked at this process, skip it.
			 */
			PROC_LOCK(p);
			if (p->p_state != PRS_NORMAL ||
			    p->p_flag & (P_INEXEC | P_SYSTEM | P_WEXIT)) {
				PROC_UNLOCK(p);
				continue;
			}
			/*
			 * if the process is in a non-running type state,
			 * don't touch it.
			 */
			breakout = 0;
			FOREACH_THREAD_IN_PROC(p, td) {
				thread_lock(td);
				if (!TD_ON_RUNQ(td) &&
				    !TD_IS_RUNNING(td) &&
				    !TD_IS_SLEEPING(td) &&
				    !TD_IS_SUSPENDED(td)) {
					thread_unlock(td);
					breakout = 1;
					break;
				}
				thread_unlock(td);
			}
			if (breakout) {
				PROC_UNLOCK(p);
				continue;
			}
			/*
			 * get a limit
			 */
			lim_rlimit_proc(p, RLIMIT_RSS, &rsslim);
			limit = OFF_TO_IDX(
			    qmin(rsslim.rlim_cur, rsslim.rlim_max));

			/*
			 * let processes that are swapped out really be
			 * swapped out set the limit to nothing (will force a
			 * swap-out.)
			 */
			if ((p->p_flag & P_INMEM) == 0)
				limit = 0;	/* XXX */
			vm = vmspace_acquire_ref(p);
			PROC_UNLOCK(p);
			if (vm == NULL)
				continue;

			size = vmspace_resident_count(vm);
			if (size >= limit) {
				vm_pageout_map_deactivate_pages(
				    &vm->vm_map, limit);
			}
#ifdef RACCT
			if (racct_enable) {
				rsize = IDX_TO_OFF(size);
				PROC_LOCK(p);
				racct_set(p, RACCT_RSS, rsize);
				ravailable = racct_get_available(p, RACCT_RSS);
				PROC_UNLOCK(p);
				if (rsize > ravailable) {
					/*
					 * Don't be overly aggressive; this
					 * might be an innocent process,
					 * and the limit could've been exceeded
					 * by some memory hog.  Don't try
					 * to deactivate more than 1/4th
					 * of process' resident set size.
					 */
					if (attempts <= 8) {
						if (ravailable < rsize -
						    (rsize / 4)) {
							ravailable = rsize -
							    (rsize / 4);
						}
					}
					vm_pageout_map_deactivate_pages(
					    &vm->vm_map,
					    OFF_TO_IDX(ravailable));
					/* Update RSS usage after paging out. */
					size = vmspace_resident_count(vm);
					rsize = IDX_TO_OFF(size);
					PROC_LOCK(p);
					racct_set(p, RACCT_RSS, rsize);
					PROC_UNLOCK(p);
					if (rsize > ravailable)
						tryagain = 1;
				}
			}
#endif
			vmspace_free(vm);
		}
		sx_sunlock(&allproc_lock);
		if (tryagain != 0 && attempts <= 10)
			goto again;
	}
}
#endif			/* !defined(NO_SWAPPING) */
OpenPOWER on IntegriCloud