summaryrefslogtreecommitdiffstats
path: root/sys/kern/vfs_bio.c
blob: 494a53d0c8d33518b518c95eec2e81fd012193c8 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
/*
 * Copyright (c) 1994 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. This work was done expressly for inclusion into FreeBSD.  Other use
 *    is allowed if this notation is included.
 * 5. Modifications may be freely made to this file if the above conditions
 *    are met.
 *
 * $Id$
 */

/*
 * this file contains a new buffer I/O scheme implementing a coherent
 * VM object and buffer cache scheme.  Pains have been taken to make
 * sure that the performance degradation associated with schemes such
 * as this is not realized.
 *
 * Author:  John S. Dyson
 * Significant help during the development and debugging phases
 * had been provided by David Greenman, also of the FreeBSD core team.
 */

#include "opt_bounce.h"

#define VMIO
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/sysproto.h>
#include <sys/kernel.h>
#include <sys/sysctl.h>
#include <sys/proc.h>
#include <sys/vnode.h>
#include <sys/vmmeter.h>
#include <vm/vm.h>
#include <vm/vm_param.h>
#include <vm/vm_prot.h>
#include <vm/vm_kern.h>
#include <vm/vm_pageout.h>
#include <vm/vm_page.h>
#include <vm/vm_object.h>
#include <vm/vm_extern.h>
#include <vm/lock.h>
#include <vm/vm_map.h>
#include <sys/buf.h>
#include <sys/mount.h>
#include <sys/malloc.h>
#include <sys/resourcevar.h>
#include <sys/proc.h>

#include <miscfs/specfs/specdev.h>

static void vfs_update __P((void));
static struct	proc *updateproc;
static struct kproc_desc up_kp = {
	"update",
	vfs_update,
	&updateproc
};
SYSINIT_KT(update, SI_SUB_KTHREAD_UPDATE, SI_ORDER_FIRST, kproc_start, &up_kp)

struct buf *buf;		/* buffer header pool */
struct swqueue bswlist;

int count_lock_queue __P((void));
static void vm_hold_free_pages(struct buf * bp, vm_offset_t from,
		vm_offset_t to);
static void vm_hold_load_pages(struct buf * bp, vm_offset_t from,
		vm_offset_t to);
static void vfs_clean_pages(struct buf * bp);
static void vfs_setdirty(struct buf *bp);
static void vfs_vmio_release(struct buf *bp);

int needsbuffer;

/*
 * Internal update daemon, process 3
 *	The variable vfs_update_wakeup allows for internal syncs.
 */
int vfs_update_wakeup;


/*
 * buffers base kva
 */

/*
 * bogus page -- for I/O to/from partially complete buffers
 * this is a temporary solution to the problem, but it is not
 * really that bad.  it would be better to split the buffer
 * for input in the case of buffers partially already in memory,
 * but the code is intricate enough already.
 */
vm_page_t bogus_page;
static vm_offset_t bogus_offset;

static int bufspace, maxbufspace, vmiospace, maxvmiobufspace,
	bufmallocspace, maxbufmallocspace;

static struct bufhashhdr bufhashtbl[BUFHSZ], invalhash;
static struct bqueues bufqueues[BUFFER_QUEUES];

extern int vm_swap_size;

#define BUF_MAXUSE 16

/*
 * Initialize buffer headers and related structures.
 */
void
bufinit()
{
	struct buf *bp;
	int i;

	TAILQ_INIT(&bswlist);
	LIST_INIT(&invalhash);

	/* first, make a null hash table */
	for (i = 0; i < BUFHSZ; i++)
		LIST_INIT(&bufhashtbl[i]);

	/* next, make a null set of free lists */
	for (i = 0; i < BUFFER_QUEUES; i++)
		TAILQ_INIT(&bufqueues[i]);

	/* finally, initialize each buffer header and stick on empty q */
	for (i = 0; i < nbuf; i++) {
		bp = &buf[i];
		bzero(bp, sizeof *bp);
		bp->b_flags = B_INVAL;	/* we're just an empty header */
		bp->b_dev = NODEV;
		bp->b_rcred = NOCRED;
		bp->b_wcred = NOCRED;
		bp->b_qindex = QUEUE_EMPTY;
		bp->b_vnbufs.le_next = NOLIST;
		TAILQ_INSERT_TAIL(&bufqueues[QUEUE_EMPTY], bp, b_freelist);
		LIST_INSERT_HEAD(&invalhash, bp, b_hash);
	}
/*
 * maxbufspace is currently calculated to support all filesystem blocks
 * to be 8K.  If you happen to use a 16K filesystem, the size of the buffer
 * cache is still the same as it would be for 8K filesystems.  This
 * keeps the size of the buffer cache "in check" for big block filesystems.
 */
	maxbufspace = (nbuf + 8) * DFLTBSIZE;
/*
 * reserve 1/3 of the buffers for metadata (VDIR) which might not be VMIO'ed
 */
	maxvmiobufspace = 2 * maxbufspace / 3;
/*
 * Limit the amount of malloc memory since it is wired permanently into
 * the kernel space.  Even though this is accounted for in the buffer
 * allocation, we don't want the malloced region to grow uncontrolled.
 * The malloc scheme improves memory utilization significantly on average
 * (small) directories.
 */
	maxbufmallocspace = maxbufspace / 20;

	bogus_offset = kmem_alloc_pageable(kernel_map, PAGE_SIZE);
	bogus_page = vm_page_alloc(kernel_object,
			((bogus_offset - VM_MIN_KERNEL_ADDRESS) >> PAGE_SHIFT),
			VM_ALLOC_NORMAL);

}

/*
 * Free the kva allocation for a buffer
 * Must be called only at splbio or higher,
 *  as this is the only locking for buffer_map.
 */
static void
bfreekva(struct buf * bp)
{
	if (bp->b_kvasize == 0)
		return;
		
	vm_map_delete(buffer_map,
		(vm_offset_t) bp->b_kvabase,
		(vm_offset_t) bp->b_kvabase + bp->b_kvasize);

	bp->b_kvasize = 0;

}

/*
 * remove the buffer from the appropriate free list
 */
void
bremfree(struct buf * bp)
{
	int s = splbio();

	if (bp->b_qindex != QUEUE_NONE) {
		TAILQ_REMOVE(&bufqueues[bp->b_qindex], bp, b_freelist);
		bp->b_qindex = QUEUE_NONE;
	} else {
		panic("bremfree: removing a buffer when not on a queue");
	}
	splx(s);
}

/*
 * Get a buffer with the specified data.  Look in the cache first.
 */
int
bread(struct vnode * vp, daddr_t blkno, int size, struct ucred * cred,
    struct buf ** bpp)
{
	struct buf *bp;

	bp = getblk(vp, blkno, size, 0, 0);
	*bpp = bp;

	/* if not found in cache, do some I/O */
	if ((bp->b_flags & B_CACHE) == 0) {
		if (curproc != NULL)
			curproc->p_stats->p_ru.ru_inblock++;
		bp->b_flags |= B_READ;
		bp->b_flags &= ~(B_DONE | B_ERROR | B_INVAL);
		if (bp->b_rcred == NOCRED) {
			if (cred != NOCRED)
				crhold(cred);
			bp->b_rcred = cred;
		}
		vfs_busy_pages(bp, 0);
		VOP_STRATEGY(bp);
		return (biowait(bp));
	}
	return (0);
}

/*
 * Operates like bread, but also starts asynchronous I/O on
 * read-ahead blocks.
 */
int
breadn(struct vnode * vp, daddr_t blkno, int size,
    daddr_t * rablkno, int *rabsize,
    int cnt, struct ucred * cred, struct buf ** bpp)
{
	struct buf *bp, *rabp;
	int i;
	int rv = 0, readwait = 0;

	*bpp = bp = getblk(vp, blkno, size, 0, 0);

	/* if not found in cache, do some I/O */
	if ((bp->b_flags & B_CACHE) == 0) {
		if (curproc != NULL)
			curproc->p_stats->p_ru.ru_inblock++;
		bp->b_flags |= B_READ;
		bp->b_flags &= ~(B_DONE | B_ERROR | B_INVAL);
		if (bp->b_rcred == NOCRED) {
			if (cred != NOCRED)
				crhold(cred);
			bp->b_rcred = cred;
		}
		vfs_busy_pages(bp, 0);
		VOP_STRATEGY(bp);
		++readwait;
	}
	for (i = 0; i < cnt; i++, rablkno++, rabsize++) {
		if (inmem(vp, *rablkno))
			continue;
		rabp = getblk(vp, *rablkno, *rabsize, 0, 0);

		if ((rabp->b_flags & B_CACHE) == 0) {
			if (curproc != NULL)
				curproc->p_stats->p_ru.ru_inblock++;
			rabp->b_flags |= B_READ | B_ASYNC;
			rabp->b_flags &= ~(B_DONE | B_ERROR | B_INVAL);
			if (rabp->b_rcred == NOCRED) {
				if (cred != NOCRED)
					crhold(cred);
				rabp->b_rcred = cred;
			}
			vfs_busy_pages(rabp, 0);
			VOP_STRATEGY(rabp);
		} else {
			brelse(rabp);
		}
	}

	if (readwait) {
		rv = biowait(bp);
	}
	return (rv);
}

/*
 * Write, release buffer on completion.  (Done by iodone
 * if async.)
 */
int
bwrite(struct buf * bp)
{
	int oldflags = bp->b_flags;

	if (bp->b_flags & B_INVAL) {
		brelse(bp);
		return (0);
	}
	if (!(bp->b_flags & B_BUSY))
		panic("bwrite: buffer is not busy???");

	bp->b_flags &= ~(B_READ | B_DONE | B_ERROR | B_DELWRI);
	bp->b_flags |= B_WRITEINPROG;

	if ((oldflags & (B_ASYNC|B_DELWRI)) == (B_ASYNC|B_DELWRI)) {
		reassignbuf(bp, bp->b_vp);
	}

	bp->b_vp->v_numoutput++;
	vfs_busy_pages(bp, 1);
	if (curproc != NULL)
		curproc->p_stats->p_ru.ru_oublock++;
	VOP_STRATEGY(bp);

	/*
	 * Handle ordered writes here.
	 * If the write was originally flagged as ordered,
	 * then we check to see if it was converted to async.
	 * If it was converted to async, and is done now, then
	 * we release the buffer.  Otherwise we clear the
	 * ordered flag because it is not needed anymore.
	 *
 	 * Note that biodone has been modified so that it does
	 * not release ordered buffers.  This allows us to have
	 * a chance to determine whether or not the driver
	 * has set the async flag in the strategy routine.  Otherwise
	 * if biodone was not modified, then the buffer may have been
	 * reused before we have had a chance to check the flag.
	 */

	if ((oldflags & B_ORDERED) == B_ORDERED) {
		int s;
		s = splbio();
		if (bp->b_flags & B_ASYNC)  {
			if ((bp->b_flags & B_DONE)) {
				if ((bp->b_flags & (B_NOCACHE | B_INVAL | B_ERROR | B_RELBUF)) != 0)
					brelse(bp);
				else
					bqrelse(bp);
			}
			splx(s);
			return (0);
		} else {
			bp->b_flags &= ~B_ORDERED;
		}
		splx(s);
	}

	if ((oldflags & B_ASYNC) == 0) {
		int rtval = biowait(bp);

		if (oldflags & B_DELWRI) {
			reassignbuf(bp, bp->b_vp);
		}
		brelse(bp);
		return (rtval);
	}
	return (0);
}

int
vn_bwrite(ap)
	struct vop_bwrite_args *ap;
{
	return (bwrite(ap->a_bp));
}

/*
 * Delayed write. (Buffer is marked dirty).
 */
void
bdwrite(struct buf * bp)
{

	if ((bp->b_flags & B_BUSY) == 0) {
		panic("bdwrite: buffer is not busy");
	}
	if (bp->b_flags & B_INVAL) {
		brelse(bp);
		return;
	}
	if (bp->b_flags & B_TAPE) {
		bawrite(bp);
		return;
	}
	bp->b_flags &= ~(B_READ|B_RELBUF);
	if ((bp->b_flags & B_DELWRI) == 0) {
		bp->b_flags |= B_DONE | B_DELWRI;
		reassignbuf(bp, bp->b_vp);
	}

	/*
	 * This bmap keeps the system from needing to do the bmap later,
	 * perhaps when the system is attempting to do a sync.  Since it
	 * is likely that the indirect block -- or whatever other datastructure
	 * that the filesystem needs is still in memory now, it is a good
	 * thing to do this.  Note also, that if the pageout daemon is
	 * requesting a sync -- there might not be enough memory to do
	 * the bmap then...  So, this is important to do.
	 */
	if( bp->b_lblkno == bp->b_blkno) {
		VOP_BMAP(bp->b_vp, bp->b_lblkno, NULL, &bp->b_blkno, NULL, NULL);
	}

	/*
	 * Set the *dirty* buffer range based upon the VM system dirty pages.
	 */
	vfs_setdirty(bp);

	/*
	 * We need to do this here to satisfy the vnode_pager and the
	 * pageout daemon, so that it thinks that the pages have been
	 * "cleaned".  Note that since the pages are in a delayed write
	 * buffer -- the VFS layer "will" see that the pages get written
	 * out on the next sync, or perhaps the cluster will be completed.
	 */
	vfs_clean_pages(bp);
	bqrelse(bp);
	return;
}

/*
 * Asynchronous write.
 * Start output on a buffer, but do not wait for it to complete.
 * The buffer is released when the output completes.
 */
void
bawrite(struct buf * bp)
{
	bp->b_flags |= B_ASYNC;
	(void) VOP_BWRITE(bp);
}

/*
 * Ordered write.
 * Start output on a buffer, but only wait for it to complete if the
 * output device cannot guarantee ordering in some other way.  Devices
 * that can perform asynchronous ordered writes will set the B_ASYNC
 * flag in their strategy routine.
 * The buffer is released when the output completes.
 */
int
bowrite(struct buf * bp)
{
	bp->b_flags |= B_ORDERED;
	return (VOP_BWRITE(bp));
}

/*
 * Release a buffer.
 */
void
brelse(struct buf * bp)
{
	int s;

	if (bp->b_flags & B_CLUSTER) {
		relpbuf(bp);
		return;
	}
	/* anyone need a "free" block? */
	s = splbio();

	/* anyone need this block? */
	if (bp->b_flags & B_WANTED) {
		bp->b_flags &= ~(B_WANTED | B_AGE);
		wakeup(bp);
	} 

	if (bp->b_flags & B_LOCKED)
		bp->b_flags &= ~B_ERROR;

	if ((bp->b_flags & (B_NOCACHE | B_INVAL | B_ERROR)) ||
	    (bp->b_bufsize <= 0)) {
		bp->b_flags |= B_INVAL;
		bp->b_flags &= ~(B_DELWRI | B_CACHE);
		if (((bp->b_flags & B_VMIO) == 0) && bp->b_vp) {
			if (bp->b_bufsize)
				allocbuf(bp, 0);
			brelvp(bp);
		}
	}

	/*
	 * VMIO buffer rundown.  It is not very necessary to keep a VMIO buffer
	 * constituted, so the B_INVAL flag is used to *invalidate* the buffer,
	 * but the VM object is kept around.  The B_NOCACHE flag is used to
	 * invalidate the pages in the VM object.
	 */
	if (bp->b_flags & B_VMIO) {
		vm_ooffset_t foff;
		vm_object_t obj;
		int i, resid;
		vm_page_t m;
		struct vnode *vp;
		int iototal = bp->b_bufsize;

		vp = bp->b_vp;
		if (!vp) 
			panic("brelse: missing vp");

		if (bp->b_npages) {
			vm_pindex_t poff;
			obj = (vm_object_t) vp->v_object;
			if (vp->v_type == VBLK)
				foff = ((vm_ooffset_t) bp->b_lblkno) << DEV_BSHIFT;
			else
				foff = (vm_ooffset_t) vp->v_mount->mnt_stat.f_iosize * bp->b_lblkno;
			poff = OFF_TO_IDX(foff);
			for (i = 0; i < bp->b_npages; i++) {
				m = bp->b_pages[i];
				if (m == bogus_page) {
					m = vm_page_lookup(obj, poff + i);
					if (!m) {
						panic("brelse: page missing\n");
					}
					bp->b_pages[i] = m;
					pmap_qenter(trunc_page(bp->b_data),
						bp->b_pages, bp->b_npages);
				}
				resid = IDX_TO_OFF(m->pindex+1) - foff;
				if (resid > iototal)
					resid = iototal;
				if (resid > 0) {
					/*
					 * Don't invalidate the page if the local machine has already
					 * modified it.  This is the lesser of two evils, and should
					 * be fixed.
					 */
					if (bp->b_flags & (B_NOCACHE | B_ERROR)) {
						vm_page_test_dirty(m);
						if (m->dirty == 0) {
							vm_page_set_invalid(m, (vm_offset_t) foff, resid);
							if (m->valid == 0)
								vm_page_protect(m, VM_PROT_NONE);
						}
					}
					if (resid >= PAGE_SIZE) {
						if ((m->valid & VM_PAGE_BITS_ALL) != VM_PAGE_BITS_ALL) {
							bp->b_flags |= B_INVAL;
						}
					} else {
						if (!vm_page_is_valid(m,
							(((vm_offset_t) bp->b_data) & PAGE_MASK), resid)) {
							bp->b_flags |= B_INVAL;
						}
					}
				}
				foff += resid;
				iototal -= resid;
			}
		}
		if (bp->b_flags & (B_INVAL | B_RELBUF))
			vfs_vmio_release(bp);
	}
	if (bp->b_qindex != QUEUE_NONE)
		panic("brelse: free buffer onto another queue???");

	/* enqueue */
	/* buffers with no memory */
	if (bp->b_bufsize == 0) {
		bp->b_qindex = QUEUE_EMPTY;
		TAILQ_INSERT_HEAD(&bufqueues[QUEUE_EMPTY], bp, b_freelist);
		LIST_REMOVE(bp, b_hash);
		LIST_INSERT_HEAD(&invalhash, bp, b_hash);
		bp->b_dev = NODEV;
		/*
		 * Get rid of the kva allocation *now*
		 */
		bfreekva(bp);
		if (needsbuffer) {
			wakeup(&needsbuffer);
			needsbuffer=0;
		}
		/* buffers with junk contents */
	} else if (bp->b_flags & (B_ERROR | B_INVAL | B_NOCACHE | B_RELBUF)) {
		bp->b_qindex = QUEUE_AGE;
		TAILQ_INSERT_HEAD(&bufqueues[QUEUE_AGE], bp, b_freelist);
		LIST_REMOVE(bp, b_hash);
		LIST_INSERT_HEAD(&invalhash, bp, b_hash);
		bp->b_dev = NODEV;
		if (needsbuffer) {
			wakeup(&needsbuffer);
			needsbuffer=0;
		}
		/* buffers that are locked */
	} else if (bp->b_flags & B_LOCKED) {
		bp->b_qindex = QUEUE_LOCKED;
		TAILQ_INSERT_TAIL(&bufqueues[QUEUE_LOCKED], bp, b_freelist);
		/* buffers with stale but valid contents */
	} else if (bp->b_flags & B_AGE) {
		bp->b_qindex = QUEUE_AGE;
		TAILQ_INSERT_TAIL(&bufqueues[QUEUE_AGE], bp, b_freelist);
		if (needsbuffer) {
			wakeup(&needsbuffer);
			needsbuffer=0;
		}
		/* buffers with valid and quite potentially reuseable contents */
	} else {
		bp->b_qindex = QUEUE_LRU;
		TAILQ_INSERT_TAIL(&bufqueues[QUEUE_LRU], bp, b_freelist);
		if (needsbuffer) {
			wakeup(&needsbuffer);
			needsbuffer=0;
		}
	}

	/* unlock */
	bp->b_flags &= ~(B_ORDERED | B_WANTED | B_BUSY |
				B_ASYNC | B_NOCACHE | B_AGE | B_RELBUF);
	splx(s);
}

/*
 * Release a buffer.
 */
void
bqrelse(struct buf * bp)
{
	int s;

	s = splbio();


	/* anyone need this block? */
	if (bp->b_flags & B_WANTED) {
		bp->b_flags &= ~(B_WANTED | B_AGE);
		wakeup(bp);
	} 

	if (bp->b_qindex != QUEUE_NONE)
		panic("bqrelse: free buffer onto another queue???");

	if (bp->b_flags & B_LOCKED) {
		bp->b_flags &= ~B_ERROR;
		bp->b_qindex = QUEUE_LOCKED;
		TAILQ_INSERT_TAIL(&bufqueues[QUEUE_LOCKED], bp, b_freelist);
		/* buffers with stale but valid contents */
	} else {
		bp->b_qindex = QUEUE_LRU;
		TAILQ_INSERT_TAIL(&bufqueues[QUEUE_LRU], bp, b_freelist);
		if (needsbuffer) {
			wakeup(&needsbuffer);
			needsbuffer=0;
		}
	}

	/* unlock */
	bp->b_flags &= ~(B_ORDERED | B_WANTED | B_BUSY |
		B_ASYNC | B_NOCACHE | B_AGE | B_RELBUF);
	splx(s);
}

static void
vfs_vmio_release(bp)
	struct buf *bp;
{
	int i;
	vm_page_t m;

	for (i = 0; i < bp->b_npages; i++) {
		m = bp->b_pages[i];
		bp->b_pages[i] = NULL;
		vm_page_unwire(m);
		/*
		 * We don't mess with busy pages, it is
		 * the responsibility of the process that
		 * busied the pages to deal with them.
		 */
		if ((m->flags & PG_BUSY) || (m->busy != 0))
			continue;
			
		if (m->wire_count == 0) {

			if (m->flags & PG_WANTED) {
				m->flags &= ~PG_WANTED;
				wakeup(m);
			}

			/*
			 * If this is an async free -- we cannot place
			 * pages onto the cache queue, so our policy for
			 * such buffers is to avoid the cache queue, and
			 * only modify the active queue or free queue.
			 */
			if ((bp->b_flags & B_ASYNC) == 0) {

			/*
			 * In the case of sync buffer frees, we can do pretty much
			 * anything to any of the memory queues.  Specifically,
			 * the cache queue is free to be modified.
			 */
				if (m->valid) {
					if(m->dirty == 0)
						vm_page_test_dirty(m);
					/*
					 * this keeps pressure off of the process memory
					 */
					if ((vm_swap_size == 0) ||
						(cnt.v_free_count < cnt.v_free_min)) {
						if ((m->dirty == 0) &&
							(m->hold_count == 0)) 
							vm_page_cache(m);
						else
							vm_page_deactivate(m);
					}
				} else if (m->hold_count == 0) {
					vm_page_protect(m, VM_PROT_NONE);
					vm_page_free(m);
				}
			} else {
				/*
				 * If async, then at least we clear the
				 * act_count.
				 */
				m->act_count = 0;
			}
		}
	}
	bufspace -= bp->b_bufsize;
	vmiospace -= bp->b_bufsize;
	pmap_qremove(trunc_page((vm_offset_t) bp->b_data), bp->b_npages);
	bp->b_npages = 0;
	bp->b_bufsize = 0;
	bp->b_flags &= ~B_VMIO;
	if (bp->b_vp)
		brelvp(bp);
}

/*
 * Check to see if a block is currently memory resident.
 */
struct buf *
gbincore(struct vnode * vp, daddr_t blkno)
{
	struct buf *bp;
	struct bufhashhdr *bh;

	bh = BUFHASH(vp, blkno);
	bp = bh->lh_first;

	/* Search hash chain */
	while (bp != NULL) {
		/* hit */
		if (bp->b_vp == vp && bp->b_lblkno == blkno &&
		    (bp->b_flags & B_INVAL) == 0) {
			break;
		}
		bp = bp->b_hash.le_next;
	}
	return (bp);
}

/*
 * this routine implements clustered async writes for
 * clearing out B_DELWRI buffers...  This is much better
 * than the old way of writing only one buffer at a time.
 */
int
vfs_bio_awrite(struct buf * bp)
{
	int i;
	daddr_t lblkno = bp->b_lblkno;
	struct vnode *vp = bp->b_vp;
	int s;
	int ncl;
	struct buf *bpa;
	int nwritten;

	s = splbio();
	/*
	 * right now we support clustered writing only to regular files
	 */
	if ((vp->v_type == VREG) && 
	    (vp->v_mount != 0) && /* Only on nodes that have the size info */
	    (bp->b_flags & (B_CLUSTEROK | B_INVAL)) == B_CLUSTEROK) {
		int size;
		int maxcl;

		size = vp->v_mount->mnt_stat.f_iosize;
		maxcl = MAXPHYS / size;

		for (i = 1; i < maxcl; i++) {
			if ((bpa = gbincore(vp, lblkno + i)) &&
			    ((bpa->b_flags & (B_BUSY | B_DELWRI | B_CLUSTEROK | B_INVAL)) ==
			    (B_DELWRI | B_CLUSTEROK)) &&
			    (bpa->b_bufsize == size)) {
				if ((bpa->b_blkno == bpa->b_lblkno) ||
				    (bpa->b_blkno != bp->b_blkno + ((i * size) >> DEV_BSHIFT)))
					break;
			} else {
				break;
			}
		}
		ncl = i;
		/*
		 * this is a possible cluster write
		 */
		if (ncl != 1) {
			nwritten = cluster_wbuild(vp, size, lblkno, ncl);
			splx(s);
			return nwritten;
		}
	}
	bremfree(bp);
	splx(s);
	/*
	 * default (old) behavior, writing out only one block
	 */
	bp->b_flags |= B_BUSY | B_ASYNC;
	nwritten = bp->b_bufsize;
	(void) VOP_BWRITE(bp);
	return nwritten;
}


/*
 * Find a buffer header which is available for use.
 */
static struct buf *
getnewbuf(int slpflag, int slptimeo, int size, int maxsize)
{
	struct buf *bp;
	int nbyteswritten = 0;
	vm_offset_t addr;

start:
	if (bufspace >= maxbufspace)
		goto trytofreespace;

	/* can we constitute a new buffer? */
	if ((bp = TAILQ_FIRST(&bufqueues[QUEUE_EMPTY]))) {
		if (bp->b_qindex != QUEUE_EMPTY)
			panic("getnewbuf: inconsistent EMPTY queue, qindex=%d",
			    bp->b_qindex);
		bp->b_flags |= B_BUSY;
		bremfree(bp);
		goto fillbuf;
	}
trytofreespace:
	/*
	 * We keep the file I/O from hogging metadata I/O
	 * This is desirable because file data is cached in the
	 * VM/Buffer cache even if a buffer is freed.
	 */
	if ((bp = TAILQ_FIRST(&bufqueues[QUEUE_AGE]))) {
		if (bp->b_qindex != QUEUE_AGE)
			panic("getnewbuf: inconsistent AGE queue, qindex=%d",
			    bp->b_qindex);
	} else if ((bp = TAILQ_FIRST(&bufqueues[QUEUE_LRU]))) {
		if (bp->b_qindex != QUEUE_LRU)
			panic("getnewbuf: inconsistent LRU queue, qindex=%d",
			    bp->b_qindex);
	}
	if (!bp) {
		/* wait for a free buffer of any kind */
		needsbuffer = 1;
		tsleep(&needsbuffer,
			(PRIBIO + 1) | slpflag, "newbuf", slptimeo);
		return (0);
	}

#if defined(DIAGNOSTIC)
	if (bp->b_flags & B_BUSY) {
		panic("getnewbuf: busy buffer on free list\n");
	}
#endif

	/*
	 * We are fairly aggressive about freeing VMIO buffers, but since
	 * the buffering is intact without buffer headers, there is not
	 * much loss.  We gain by maintaining non-VMIOed metadata in buffers.
	 */
	if ((bp->b_qindex == QUEUE_LRU) && (bp->b_usecount > 0)) {
		if ((bp->b_flags & B_VMIO) == 0 ||
			(vmiospace < maxvmiobufspace)) {
			--bp->b_usecount;
			TAILQ_REMOVE(&bufqueues[QUEUE_LRU], bp, b_freelist);
			if (TAILQ_FIRST(&bufqueues[QUEUE_LRU]) != NULL) {
				TAILQ_INSERT_TAIL(&bufqueues[QUEUE_LRU], bp, b_freelist);
				goto start;
			}
			TAILQ_INSERT_TAIL(&bufqueues[QUEUE_LRU], bp, b_freelist);
		}
	}

	/* if we are a delayed write, convert to an async write */
	if ((bp->b_flags & (B_DELWRI | B_INVAL)) == B_DELWRI) {
		nbyteswritten += vfs_bio_awrite(bp);
		if (!slpflag && !slptimeo) {
			return (0);
		}
		goto start;
	}

	if (bp->b_flags & B_WANTED) {
		bp->b_flags &= ~B_WANTED;
		wakeup(bp);
	}
	bremfree(bp);
	bp->b_flags |= B_BUSY;

	if (bp->b_flags & B_VMIO) {
		bp->b_flags &= ~B_ASYNC;
		vfs_vmio_release(bp);
	}

	if (bp->b_vp)
		brelvp(bp);

fillbuf:
	/* we are not free, nor do we contain interesting data */
	if (bp->b_rcred != NOCRED) {
		crfree(bp->b_rcred);
		bp->b_rcred = NOCRED;
	}
	if (bp->b_wcred != NOCRED) {
		crfree(bp->b_wcred);
		bp->b_wcred = NOCRED;
	}

	LIST_REMOVE(bp, b_hash);
	LIST_INSERT_HEAD(&invalhash, bp, b_hash);
	if (bp->b_bufsize) {
		allocbuf(bp, 0);
	}
	bp->b_flags = B_BUSY;
	bp->b_dev = NODEV;
	bp->b_vp = NULL;
	bp->b_blkno = bp->b_lblkno = 0;
	bp->b_iodone = 0;
	bp->b_error = 0;
	bp->b_resid = 0;
	bp->b_bcount = 0;
	bp->b_npages = 0;
	bp->b_dirtyoff = bp->b_dirtyend = 0;
	bp->b_validoff = bp->b_validend = 0;
	bp->b_usecount = 4;

	maxsize = (maxsize + PAGE_MASK) & ~PAGE_MASK;

	/*
	 * we assume that buffer_map is not at address 0
	 */
	addr = 0;
	if (maxsize != bp->b_kvasize) {
		bfreekva(bp);
		
		/*
		 * See if we have buffer kva space
		 */
		if (vm_map_findspace(buffer_map,
			vm_map_min(buffer_map), maxsize, &addr)) {
			bp->b_flags |= B_INVAL;
			brelse(bp);
			goto trytofreespace;
		}
	}

	/*
	 * See if we are below are allocated minimum
	 */
	if (bufspace >= (maxbufspace + nbyteswritten)) {
		bp->b_flags |= B_INVAL;
		brelse(bp);
		goto trytofreespace;
	}

	/*
	 * create a map entry for the buffer -- in essence
	 * reserving the kva space.
	 */
	if (addr) {
		vm_map_insert(buffer_map, NULL, 0,
			addr, addr + maxsize,
			VM_PROT_ALL, VM_PROT_ALL, MAP_NOFAULT);

		bp->b_kvabase = (caddr_t) addr;
		bp->b_kvasize = maxsize;
	}
	bp->b_data = bp->b_kvabase;
	
	return (bp);
}

/*
 * Check to see if a block is currently memory resident.
 */
struct buf *
incore(struct vnode * vp, daddr_t blkno)
{
	struct buf *bp;

	int s = splbio();
	bp = gbincore(vp, blkno);
	splx(s);
	return (bp);
}

/*
 * Returns true if no I/O is needed to access the
 * associated VM object.  This is like incore except
 * it also hunts around in the VM system for the data.
 */

int
inmem(struct vnode * vp, daddr_t blkno)
{
	vm_object_t obj;
	vm_offset_t toff, tinc;
	vm_page_t m;
	vm_ooffset_t off;

	if (incore(vp, blkno))
		return 1;
	if (vp->v_mount == NULL)
		return 0;
	if ((vp->v_object == NULL) || (vp->v_flag & VVMIO) == 0)
		return 0;

	obj = vp->v_object;
	tinc = PAGE_SIZE;
	if (tinc > vp->v_mount->mnt_stat.f_iosize)
		tinc = vp->v_mount->mnt_stat.f_iosize;
	off = blkno * vp->v_mount->mnt_stat.f_iosize;

	for (toff = 0; toff < vp->v_mount->mnt_stat.f_iosize; toff += tinc) {

		m = vm_page_lookup(obj, OFF_TO_IDX(off + toff));
		if (!m)
			return 0;
		if (vm_page_is_valid(m, (vm_offset_t) (toff + off), tinc) == 0)
			return 0;
	}
	return 1;
}

/*
 * now we set the dirty range for the buffer --
 * for NFS -- if the file is mapped and pages have
 * been written to, let it know.  We want the
 * entire range of the buffer to be marked dirty if
 * any of the pages have been written to for consistancy
 * with the b_validoff, b_validend set in the nfs write
 * code, and used by the nfs read code.
 */
static void
vfs_setdirty(struct buf *bp) {
	int i;
	vm_object_t object;
	vm_offset_t boffset, offset;
	/*
	 * We qualify the scan for modified pages on whether the
	 * object has been flushed yet.  The OBJ_WRITEABLE flag
	 * is not cleared simply by protecting pages off.
	 */
	if ((bp->b_flags & B_VMIO) &&
		((object = bp->b_pages[0]->object)->flags & (OBJ_WRITEABLE|OBJ_CLEANING))) {
		/*
		 * test the pages to see if they have been modified directly
		 * by users through the VM system.
		 */
		for (i = 0; i < bp->b_npages; i++)
			vm_page_test_dirty(bp->b_pages[i]);

		/*
		 * scan forwards for the first page modified
		 */
		for (i = 0; i < bp->b_npages; i++) {
			if (bp->b_pages[i]->dirty) {
				break;
			}
		}
		boffset = (i << PAGE_SHIFT);
		if (boffset < bp->b_dirtyoff) {
			bp->b_dirtyoff = boffset;
		}

		/*
		 * scan backwards for the last page modified
		 */
		for (i = bp->b_npages - 1; i >= 0; --i) {
			if (bp->b_pages[i]->dirty) {
				break;
			}
		}
		boffset = (i + 1);
		offset = boffset + bp->b_pages[0]->pindex;
		if (offset >= object->size)
			boffset = object->size - bp->b_pages[0]->pindex;
		if (bp->b_dirtyend < (boffset << PAGE_SHIFT))
			bp->b_dirtyend = (boffset << PAGE_SHIFT);
	}
}

/*
 * Get a block given a specified block and offset into a file/device.
 */
struct buf *
getblk(struct vnode * vp, daddr_t blkno, int size, int slpflag, int slptimeo)
{
	struct buf *bp;
	int s;
	struct bufhashhdr *bh;
	int maxsize;

	if (vp->v_mount) {
		maxsize = vp->v_mount->mnt_stat.f_iosize;
		/*
		 * This happens on mount points.
		 */
		if (maxsize < size)
			maxsize = size;
	} else {
		maxsize = size;
	}

	if (size > MAXBSIZE)
		panic("getblk: size(%d) > MAXBSIZE(%d)\n", size, MAXBSIZE);

	s = splbio();
loop:
	if ((bp = gbincore(vp, blkno))) {
		if (bp->b_flags & B_BUSY) {
			bp->b_flags |= B_WANTED;
			if (bp->b_usecount < BUF_MAXUSE)
				++bp->b_usecount;
			if (!tsleep(bp,
				(PRIBIO + 1) | slpflag, "getblk", slptimeo))
				goto loop;

			splx(s);
			return (struct buf *) NULL;
		}
		bp->b_flags |= B_BUSY | B_CACHE;
		bremfree(bp);
				
		/*
		 * check for size inconsistancies (note that they shouldn't happen
		 * but do when filesystems don't handle the size changes correctly.)
		 * We are conservative on metadata and don't just extend the buffer
		 * but write and re-constitute it.
		 */

		if (bp->b_bcount != size) {
			if ((bp->b_flags & B_VMIO) && (size <= bp->b_kvasize)) {
				allocbuf(bp, size);
			} else {
				bp->b_flags |= B_NOCACHE;
				VOP_BWRITE(bp);
				goto loop;
			}
		}

		if (bp->b_usecount < BUF_MAXUSE)
			++bp->b_usecount;
		splx(s);
		return (bp);
	} else {
		vm_object_t obj;

		if ((bp = getnewbuf(slpflag, slptimeo, size, maxsize)) == 0) {
			if (slpflag || slptimeo) {
				splx(s);
				return NULL;
			}
			goto loop;
		}

		/*
		 * This code is used to make sure that a buffer is not
		 * created while the getnewbuf routine is blocked.
		 * Normally the vnode is locked so this isn't a problem.
		 * VBLK type I/O requests, however, don't lock the vnode.
		 */
		if (!VOP_ISLOCKED(vp) && gbincore(vp, blkno)) {
			bp->b_flags |= B_INVAL;
			brelse(bp);
			goto loop;
		}

		/*
		 * Insert the buffer into the hash, so that it can
		 * be found by incore.
		 */
		bp->b_blkno = bp->b_lblkno = blkno;
		bgetvp(vp, bp);
		LIST_REMOVE(bp, b_hash);
		bh = BUFHASH(vp, blkno);
		LIST_INSERT_HEAD(bh, bp, b_hash);

		if ((obj = vp->v_object) && (vp->v_flag & VVMIO)) {
			bp->b_flags |= (B_VMIO | B_CACHE);
#if defined(VFS_BIO_DEBUG)
			if (vp->v_type != VREG && vp->v_type != VBLK)
				printf("getblk: vmioing file type %d???\n", vp->v_type);
#endif
		} else {
			bp->b_flags &= ~B_VMIO;
		}
		splx(s);

		allocbuf(bp, size);
#ifdef	PC98
		/*
		 * 1024byte/sector support
		 */
#define B_XXX2 0x8000000
		if (vp->v_flag & 0x10000) bp->b_flags |= B_XXX2;
#endif
		return (bp);
	}
}

/*
 * Get an empty, disassociated buffer of given size.
 */
struct buf *
geteblk(int size)
{
	struct buf *bp;
	int s;

	s = splbio();
	while ((bp = getnewbuf(0, 0, size, MAXBSIZE)) == 0);
	splx(s);
	allocbuf(bp, size);
	bp->b_flags |= B_INVAL;
	return (bp);
}


/*
 * This code constitutes the buffer memory from either anonymous system
 * memory (in the case of non-VMIO operations) or from an associated
 * VM object (in the case of VMIO operations).
 *
 * Note that this code is tricky, and has many complications to resolve
 * deadlock or inconsistant data situations.  Tread lightly!!!
 *
 * Modify the length of a buffer's underlying buffer storage without
 * destroying information (unless, of course the buffer is shrinking).
 */
int
allocbuf(struct buf * bp, int size)
{

	int s;
	int newbsize, mbsize;
	int i;

	if (!(bp->b_flags & B_BUSY))
		panic("allocbuf: buffer not busy");

	if (bp->b_kvasize < size)
		panic("allocbuf: buffer too small");

	if ((bp->b_flags & B_VMIO) == 0) {
		caddr_t origbuf;
		int origbufsize;
		/*
		 * Just get anonymous memory from the kernel
		 */
		mbsize = (size + DEV_BSIZE - 1) & ~(DEV_BSIZE - 1);
#if !defined(NO_B_MALLOC)
		if (bp->b_flags & B_MALLOC)
			newbsize = mbsize;
		else
#endif
			newbsize = round_page(size);

		if (newbsize < bp->b_bufsize) {
#if !defined(NO_B_MALLOC)
			/*
			 * malloced buffers are not shrunk
			 */
			if (bp->b_flags & B_MALLOC) {
				if (newbsize) {
					bp->b_bcount = size;
				} else {
					free(bp->b_data, M_BIOBUF);
					bufspace -= bp->b_bufsize;
					bufmallocspace -= bp->b_bufsize;
					bp->b_data = bp->b_kvabase;
					bp->b_bufsize = 0;
					bp->b_bcount = 0;
					bp->b_flags &= ~B_MALLOC;
				}
				return 1;
			}		
#endif
			vm_hold_free_pages(
			    bp,
			    (vm_offset_t) bp->b_data + newbsize,
			    (vm_offset_t) bp->b_data + bp->b_bufsize);
		} else if (newbsize > bp->b_bufsize) {
#if !defined(NO_B_MALLOC)
			/*
			 * We only use malloced memory on the first allocation.
			 * and revert to page-allocated memory when the buffer grows.
			 */
			if ( (bufmallocspace < maxbufmallocspace) &&
				(bp->b_bufsize == 0) &&
				(mbsize <= PAGE_SIZE/2)) {

				bp->b_data = malloc(mbsize, M_BIOBUF, M_WAITOK);
				bp->b_bufsize = mbsize;
				bp->b_bcount = size;
				bp->b_flags |= B_MALLOC;
				bufspace += mbsize;
				bufmallocspace += mbsize;
				return 1;
			}
#endif
			origbuf = NULL;
			origbufsize = 0;
#if !defined(NO_B_MALLOC)
			/*
			 * If the buffer is growing on it's other-than-first allocation,
			 * then we revert to the page-allocation scheme.
			 */
			if (bp->b_flags & B_MALLOC) {
				origbuf = bp->b_data;
				origbufsize = bp->b_bufsize;
				bp->b_data = bp->b_kvabase;
				bufspace -= bp->b_bufsize;
				bufmallocspace -= bp->b_bufsize;
				bp->b_bufsize = 0;
				bp->b_flags &= ~B_MALLOC;
				newbsize = round_page(newbsize);
			}
#endif
			vm_hold_load_pages(
			    bp,
			    (vm_offset_t) bp->b_data + bp->b_bufsize,
			    (vm_offset_t) bp->b_data + newbsize);
#if !defined(NO_B_MALLOC)
			if (origbuf) {
				bcopy(origbuf, bp->b_data, origbufsize);
				free(origbuf, M_BIOBUF);
			}
#endif
		}
	} else {
		vm_page_t m;
		int desiredpages;

		newbsize = (size + DEV_BSIZE - 1) & ~(DEV_BSIZE - 1);
		desiredpages = (round_page(newbsize) >> PAGE_SHIFT);

#if !defined(NO_B_MALLOC)
		if (bp->b_flags & B_MALLOC)
			panic("allocbuf: VMIO buffer can't be malloced");
#endif

		if (newbsize < bp->b_bufsize) {
			if (desiredpages < bp->b_npages) {
				for (i = desiredpages; i < bp->b_npages; i++) {
					/*
					 * the page is not freed here -- it
					 * is the responsibility of vnode_pager_setsize
					 */
					m = bp->b_pages[i];
#if defined(DIAGNOSTIC)
					if (m == bogus_page)
						panic("allocbuf: bogus page found");
#endif
					s = splvm();
					while ((m->flags & PG_BUSY) || (m->busy != 0)) {
						m->flags |= PG_WANTED;
						tsleep(m, PVM, "biodep", 0);
					}
					splx(s);

					bp->b_pages[i] = NULL;
					vm_page_unwire(m);
				}
				pmap_qremove((vm_offset_t) trunc_page(bp->b_data) +
				    (desiredpages << PAGE_SHIFT), (bp->b_npages - desiredpages));
				bp->b_npages = desiredpages;
			}
		} else if (newbsize > bp->b_bufsize) {
			vm_object_t obj;
			vm_offset_t tinc, toff;
			vm_ooffset_t off;
			vm_pindex_t objoff;
			int pageindex, curbpnpages;
			struct vnode *vp;
			int bsize;

			vp = bp->b_vp;

			if (vp->v_type == VBLK)
				bsize = DEV_BSIZE;
			else
				bsize = vp->v_mount->mnt_stat.f_iosize;

			if (bp->b_npages < desiredpages) {
				obj = vp->v_object;
				tinc = PAGE_SIZE;
				if (tinc > bsize)
					tinc = bsize;
				off = (vm_ooffset_t) bp->b_lblkno * bsize;
				curbpnpages = bp->b_npages;
		doretry:
				bp->b_flags |= B_CACHE;
				for (toff = 0; toff < newbsize; toff += tinc) {
					int bytesinpage;

					pageindex = toff >> PAGE_SHIFT;
					objoff = OFF_TO_IDX(off + toff);
					if (pageindex < curbpnpages) {

						m = bp->b_pages[pageindex];
#ifdef VFS_BIO_DIAG
						if (m->pindex != objoff)
							panic("allocbuf: page changed offset??!!!?");
#endif
						bytesinpage = tinc;
						if (tinc > (newbsize - toff))
							bytesinpage = newbsize - toff;
						if ((bp->b_flags & B_CACHE) &&
							!vm_page_is_valid(m,
							(vm_offset_t) ((toff + off) & PAGE_MASK),
							bytesinpage)) {
							bp->b_flags &= ~B_CACHE;
						}
						continue;
					}
					m = vm_page_lookup(obj, objoff);
					if (!m) {
						m = vm_page_alloc(obj, objoff, VM_ALLOC_NORMAL);
						if (!m) {
							VM_WAIT;
							goto doretry;
						}
						/*
						 * Normally it is unwise to clear PG_BUSY without
						 * PAGE_WAKEUP -- but it is okay here, as there is
						 * no chance for blocking between here and vm_page_alloc
						 */
						m->flags &= ~PG_BUSY;
						vm_page_wire(m);
						bp->b_flags &= ~B_CACHE;
					} else if (m->flags & PG_BUSY) {
						s = splvm();
						if (m->flags & PG_BUSY) {
							m->flags |= PG_WANTED;
							tsleep(m, PVM, "pgtblk", 0);
						}
						splx(s);
						goto doretry;
					} else {
						if ((curproc != pageproc) &&
							((m->queue - m->pc) == PQ_CACHE) &&
						    ((cnt.v_free_count + cnt.v_cache_count) <
								(cnt.v_free_min + cnt.v_cache_min))) {
							pagedaemon_wakeup();
						}
						bytesinpage = tinc;
						if (tinc > (newbsize - toff))
							bytesinpage = newbsize - toff;
						if ((bp->b_flags & B_CACHE) &&
							!vm_page_is_valid(m,
							(vm_offset_t) ((toff + off) & PAGE_MASK),
							bytesinpage)) {
							bp->b_flags &= ~B_CACHE;
						}
						vm_page_wire(m);
					}
					bp->b_pages[pageindex] = m;
					curbpnpages = pageindex + 1;
				}
				bp->b_data = (caddr_t) trunc_page(bp->b_data);
				bp->b_npages = curbpnpages;
				pmap_qenter((vm_offset_t) bp->b_data,
					bp->b_pages, bp->b_npages);
				((vm_offset_t) bp->b_data) |= off & PAGE_MASK;
			}
		}
	}
	if (bp->b_flags & B_VMIO)
		vmiospace += bp->b_bufsize;
	bufspace += (newbsize - bp->b_bufsize);
	bp->b_bufsize = newbsize;
	bp->b_bcount = size;
	return 1;
}

/*
 * Wait for buffer I/O completion, returning error status.
 */
int
biowait(register struct buf * bp)
{
	int s;

	s = splbio();
	while ((bp->b_flags & B_DONE) == 0)
		tsleep(bp, PRIBIO, "biowait", 0);
	splx(s);
	if (bp->b_flags & B_EINTR) {
		bp->b_flags &= ~B_EINTR;
		return (EINTR);
	}
	if (bp->b_flags & B_ERROR) {
		return (bp->b_error ? bp->b_error : EIO);
	} else {
		return (0);
	}
}

/*
 * Finish I/O on a buffer, calling an optional function.
 * This is usually called from interrupt level, so process blocking
 * is not *a good idea*.
 */
void
biodone(register struct buf * bp)
{
	int s;

	s = splbio();
	if (!(bp->b_flags & B_BUSY))
		panic("biodone: buffer not busy");

	if (bp->b_flags & B_DONE) {
		splx(s);
		printf("biodone: buffer already done\n");
		return;
	}
	bp->b_flags |= B_DONE;

	if ((bp->b_flags & B_READ) == 0) {
		vwakeup(bp);
	}
#ifdef BOUNCE_BUFFERS
	if (bp->b_flags & B_BOUNCE)
		vm_bounce_free(bp);
#endif

	/* call optional completion function if requested */
	if (bp->b_flags & B_CALL) {
		bp->b_flags &= ~B_CALL;
		(*bp->b_iodone) (bp);
		splx(s);
		return;
	}
	if (bp->b_flags & B_VMIO) {
		int i, resid;
		vm_ooffset_t foff;
		vm_page_t m;
		vm_object_t obj;
		int iosize;
		struct vnode *vp = bp->b_vp;

		if (vp->v_type == VBLK)
			foff = (vm_ooffset_t) DEV_BSIZE * bp->b_lblkno;
		else
			foff = (vm_ooffset_t) vp->v_mount->mnt_stat.f_iosize * bp->b_lblkno;
		obj = vp->v_object;
		if (!obj) {
			panic("biodone: no object");
		}
#if defined(VFS_BIO_DEBUG)
		if (obj->paging_in_progress < bp->b_npages) {
			printf("biodone: paging in progress(%d) < bp->b_npages(%d)\n",
			    obj->paging_in_progress, bp->b_npages);
		}
#endif
		iosize = bp->b_bufsize;
		for (i = 0; i < bp->b_npages; i++) {
			int bogusflag = 0;
			m = bp->b_pages[i];
			if (m == bogus_page) {
				bogusflag = 1;
				m = vm_page_lookup(obj, OFF_TO_IDX(foff));
				if (!m) {
#if defined(VFS_BIO_DEBUG)
					printf("biodone: page disappeared\n");
#endif
					--obj->paging_in_progress;
					continue;
				}
				bp->b_pages[i] = m;
				pmap_qenter(trunc_page(bp->b_data), bp->b_pages, bp->b_npages);
			}
#if defined(VFS_BIO_DEBUG)
			if (OFF_TO_IDX(foff) != m->pindex) {
				printf("biodone: foff(%d)/m->pindex(%d) mismatch\n", foff, m->pindex);
			}
#endif
			resid = IDX_TO_OFF(m->pindex + 1) - foff;
			if (resid > iosize)
				resid = iosize;
			/*
			 * In the write case, the valid and clean bits are
			 * already changed correctly, so we only need to do this
			 * here in the read case.
			 */
			if ((bp->b_flags & B_READ) && !bogusflag && resid > 0) {
				vm_page_set_validclean(m,
					(vm_offset_t) (foff & PAGE_MASK), resid);
			}

			/*
			 * when debugging new filesystems or buffer I/O methods, this
			 * is the most common error that pops up.  if you see this, you
			 * have not set the page busy flag correctly!!!
			 */
			if (m->busy == 0) {
				printf("biodone: page busy < 0, "
				    "pindex: %d, foff: 0x(%x,%x), "
				    "resid: %d, index: %d\n",
				    (int) m->pindex, (int)(foff >> 32),
						(int) foff & 0xffffffff, resid, i);
				if (vp->v_type != VBLK)
					printf(" iosize: %ld, lblkno: %d, flags: 0x%lx, npages: %d\n",
					    bp->b_vp->v_mount->mnt_stat.f_iosize,
					    (int) bp->b_lblkno,
					    bp->b_flags, bp->b_npages);
				else
					printf(" VDEV, lblkno: %d, flags: 0x%lx, npages: %d\n",
					    (int) bp->b_lblkno,
					    bp->b_flags, bp->b_npages);
				printf(" valid: 0x%x, dirty: 0x%x, wired: %d\n",
				    m->valid, m->dirty, m->wire_count);
				panic("biodone: page busy < 0\n");
			}
			--m->busy;
			if ((m->busy == 0) && (m->flags & PG_WANTED)) {
				m->flags &= ~PG_WANTED;
				wakeup(m);
			}
			--obj->paging_in_progress;
			foff += resid;
			iosize -= resid;
		}
		if (obj && obj->paging_in_progress == 0 &&
		    (obj->flags & OBJ_PIPWNT)) {
			obj->flags &= ~OBJ_PIPWNT;
			wakeup(obj);
		}
	}
	/*
	 * For asynchronous completions, release the buffer now. The brelse
	 * checks for B_WANTED and will do the wakeup there if necessary - so
	 * no need to do a wakeup here in the async case.
	 */

	if (bp->b_flags & B_ASYNC) {
		if ((bp->b_flags & B_ORDERED) == 0) {
			if ((bp->b_flags & (B_NOCACHE | B_INVAL | B_ERROR | B_RELBUF)) != 0)
				brelse(bp);
			else
				bqrelse(bp);
		}
	} else {
		bp->b_flags &= ~B_WANTED;
		wakeup(bp);
	}
	splx(s);
}

int
count_lock_queue()
{
	int count;
	struct buf *bp;

	count = 0;
	for (bp = TAILQ_FIRST(&bufqueues[QUEUE_LOCKED]);
	    bp != NULL;
	    bp = TAILQ_NEXT(bp, b_freelist))
		count++;
	return (count);
}

int vfs_update_interval = 30;

static void
vfs_update()
{
	while (1) {
		tsleep(&vfs_update_wakeup, PUSER, "update",
		    hz * vfs_update_interval);
		vfs_update_wakeup = 0;
		sync(curproc, NULL, NULL);
	}
}

static int
sysctl_kern_updateinterval SYSCTL_HANDLER_ARGS
{
	int error = sysctl_handle_int(oidp,
		oidp->oid_arg1, oidp->oid_arg2, req);
	if (!error)
		wakeup(&vfs_update_wakeup);
	return error;
}

SYSCTL_PROC(_kern, KERN_UPDATEINTERVAL, update, CTLTYPE_INT|CTLFLAG_RW,
	&vfs_update_interval, 0, sysctl_kern_updateinterval, "I", "");


/*
 * This routine is called in lieu of iodone in the case of
 * incomplete I/O.  This keeps the busy status for pages
 * consistant.
 */
void
vfs_unbusy_pages(struct buf * bp)
{
	int i;

	if (bp->b_flags & B_VMIO) {
		struct vnode *vp = bp->b_vp;
		vm_object_t obj = vp->v_object;
		vm_ooffset_t foff;

		foff = (vm_ooffset_t) vp->v_mount->mnt_stat.f_iosize * bp->b_lblkno;

		for (i = 0; i < bp->b_npages; i++) {
			vm_page_t m = bp->b_pages[i];

			if (m == bogus_page) {
				m = vm_page_lookup(obj, OFF_TO_IDX(foff) + i);
				if (!m) {
					panic("vfs_unbusy_pages: page missing\n");
				}
				bp->b_pages[i] = m;
				pmap_qenter(trunc_page(bp->b_data), bp->b_pages, bp->b_npages);
			}
			--obj->paging_in_progress;
			--m->busy;
			if ((m->busy == 0) && (m->flags & PG_WANTED)) {
				m->flags &= ~PG_WANTED;
				wakeup(m);
			}
		}
		if (obj->paging_in_progress == 0 &&
		    (obj->flags & OBJ_PIPWNT)) {
			obj->flags &= ~OBJ_PIPWNT;
			wakeup(obj);
		}
	}
}

/*
 * This routine is called before a device strategy routine.
 * It is used to tell the VM system that paging I/O is in
 * progress, and treat the pages associated with the buffer
 * almost as being PG_BUSY.  Also the object paging_in_progress
 * flag is handled to make sure that the object doesn't become
 * inconsistant.
 */
void
vfs_busy_pages(struct buf * bp, int clear_modify)
{
	int i;

	if (bp->b_flags & B_VMIO) {
		vm_object_t obj = bp->b_vp->v_object;
		vm_ooffset_t foff;
		int iocount = bp->b_bufsize;

		if (bp->b_vp->v_type == VBLK)
			foff = (vm_ooffset_t) DEV_BSIZE * bp->b_lblkno;
		else
			foff = (vm_ooffset_t) bp->b_vp->v_mount->mnt_stat.f_iosize * bp->b_lblkno;
		vfs_setdirty(bp);
		for (i = 0; i < bp->b_npages; i++) {
			vm_page_t m = bp->b_pages[i];
			int resid = IDX_TO_OFF(m->pindex + 1) - foff;

			if (resid > iocount)
				resid = iocount;
			if ((bp->b_flags & B_CLUSTER) == 0) {
				obj->paging_in_progress++;
				m->busy++;
			}
			vm_page_protect(m, VM_PROT_NONE);
			if (clear_modify) {
				vm_page_set_validclean(m,
					(vm_offset_t) (foff & PAGE_MASK), resid);
			} else if (bp->b_bcount >= PAGE_SIZE) {
				if (m->valid && (bp->b_flags & B_CACHE) == 0) {
					bp->b_pages[i] = bogus_page;
					pmap_qenter(trunc_page(bp->b_data), bp->b_pages, bp->b_npages);
				}
			}
			foff += resid;
			iocount -= resid;
		}
	}
}

/*
 * Tell the VM system that the pages associated with this buffer
 * are clean.  This is used for delayed writes where the data is
 * going to go to disk eventually without additional VM intevention.
 */
void
vfs_clean_pages(struct buf * bp)
{
	int i;

	if (bp->b_flags & B_VMIO) {
		vm_ooffset_t foff;
		int iocount = bp->b_bufsize;

		if (bp->b_vp->v_type == VBLK)
			foff = (vm_ooffset_t) DEV_BSIZE * bp->b_lblkno;
		else
			foff = (vm_ooffset_t) bp->b_vp->v_mount->mnt_stat.f_iosize * bp->b_lblkno;

		for (i = 0; i < bp->b_npages; i++) {
			vm_page_t m = bp->b_pages[i];
			int resid = IDX_TO_OFF(m->pindex + 1) - foff;

			if (resid > iocount)
				resid = iocount;
			if (resid > 0) {
				vm_page_set_validclean(m,
					((vm_offset_t) foff & PAGE_MASK), resid);
			}
			foff += resid;
			iocount -= resid;
		}
	}
}

void
vfs_bio_clrbuf(struct buf *bp) {
	int i;
	if( bp->b_flags & B_VMIO) {
		if( (bp->b_npages == 1) && (bp->b_bufsize < PAGE_SIZE)) {
			int mask;
			mask = 0;
			for(i=0;i<bp->b_bufsize;i+=DEV_BSIZE)
				mask |= (1 << (i/DEV_BSIZE));
			if( bp->b_pages[0]->valid != mask) {
				bzero(bp->b_data, bp->b_bufsize);
			}
			bp->b_pages[0]->valid = mask;
			bp->b_resid = 0;
			return;
		}
		for(i=0;i<bp->b_npages;i++) {
			if( bp->b_pages[i]->valid == VM_PAGE_BITS_ALL)
				continue;
			if( bp->b_pages[i]->valid == 0) {
				if ((bp->b_pages[i]->flags & PG_ZERO) == 0) {
					bzero(bp->b_data + (i << PAGE_SHIFT), PAGE_SIZE);
				}
			} else {
				int j;
				for(j=0;j<PAGE_SIZE/DEV_BSIZE;j++) {
					if( (bp->b_pages[i]->valid & (1<<j)) == 0)
						bzero(bp->b_data + (i << PAGE_SHIFT) + j * DEV_BSIZE, DEV_BSIZE);
				}
			}
			/* bp->b_pages[i]->valid = VM_PAGE_BITS_ALL; */
		}
		bp->b_resid = 0;
	} else {
		clrbuf(bp);
	}
}

/*
 * vm_hold_load_pages and vm_hold_unload pages get pages into
 * a buffers address space.  The pages are anonymous and are
 * not associated with a file object.
 */
void
vm_hold_load_pages(struct buf * bp, vm_offset_t from, vm_offset_t to)
{
	vm_offset_t pg;
	vm_page_t p;
	int index;

	to = round_page(to);
	from = round_page(from);
	index = (from - trunc_page(bp->b_data)) >> PAGE_SHIFT;

	for (pg = from; pg < to; pg += PAGE_SIZE, index++) {

tryagain:

		p = vm_page_alloc(kernel_object, ((pg - VM_MIN_KERNEL_ADDRESS) >> PAGE_SHIFT),
		    VM_ALLOC_NORMAL);
		if (!p) {
			VM_WAIT;
			goto tryagain;
		}
		vm_page_wire(p);
		pmap_kenter(pg, VM_PAGE_TO_PHYS(p));
		bp->b_pages[index] = p;
		PAGE_WAKEUP(p);
	}
	bp->b_npages = to >> PAGE_SHIFT;
}

void
vm_hold_free_pages(struct buf * bp, vm_offset_t from, vm_offset_t to)
{
	vm_offset_t pg;
	vm_page_t p;
	int index;

	from = round_page(from);
	to = round_page(to);
	index = (from - trunc_page(bp->b_data)) >> PAGE_SHIFT;

	for (pg = from; pg < to; pg += PAGE_SIZE, index++) {
		p = bp->b_pages[index];
		if (p && (index < bp->b_npages)) {
			if (p->busy) {
				printf("vm_hold_free_pages: blkno: %d, lblkno: %d\n",
					bp->b_blkno, bp->b_lblkno);
			}
			bp->b_pages[index] = NULL;
			pmap_kremove(pg);
			vm_page_unwire(p);
			vm_page_free(p);
		}
	}
	bp->b_npages = from >> PAGE_SHIFT;
}
OpenPOWER on IntegriCloud