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
|
/*-
* Copyright (c) 1989, 1993
* The Regents of the University of California. All rights reserved.
*
* This code is derived from software contributed to Berkeley by
* Rick Macklem at The University of Guelph.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 4. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* @(#)nfs_bio.c 8.9 (Berkeley) 3/30/95
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include "opt_kdtrace.h"
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/bio.h>
#include <sys/buf.h>
#include <sys/kernel.h>
#include <sys/mbuf.h>
#include <sys/mount.h>
#include <sys/proc.h>
#include <sys/rwlock.h>
#include <sys/vmmeter.h>
#include <sys/vnode.h>
#include <vm/vm.h>
#include <vm/vm_param.h>
#include <vm/vm_extern.h>
#include <vm/vm_page.h>
#include <vm/vm_object.h>
#include <vm/vm_pager.h>
#include <vm/vnode_pager.h>
#include <nfs/nfsproto.h>
#include <nfsclient/nfs.h>
#include <nfsclient/nfsmount.h>
#include <nfsclient/nfsnode.h>
#include <nfs/nfs_kdtrace.h>
static struct buf *nfs_getcacheblk(struct vnode *vp, daddr_t bn, int size,
struct thread *td);
static int nfs_directio_write(struct vnode *vp, struct uio *uiop,
struct ucred *cred, int ioflag);
extern int nfs_directio_enable;
extern int nfs_directio_allow_mmap;
/*
* Vnode op for VM getpages.
*/
int
nfs_getpages(struct vop_getpages_args *ap)
{
int i, error, nextoff, size, toff, count, npages;
struct uio uio;
struct iovec iov;
vm_offset_t kva;
struct buf *bp;
struct vnode *vp;
struct thread *td;
struct ucred *cred;
struct nfsmount *nmp;
vm_object_t object;
vm_page_t *pages;
struct nfsnode *np;
vp = ap->a_vp;
np = VTONFS(vp);
td = curthread; /* XXX */
cred = curthread->td_ucred; /* XXX */
nmp = VFSTONFS(vp->v_mount);
pages = ap->a_m;
count = ap->a_count;
if ((object = vp->v_object) == NULL) {
nfs_printf("nfs_getpages: called with non-merged cache vnode??\n");
return (VM_PAGER_ERROR);
}
if (nfs_directio_enable && !nfs_directio_allow_mmap) {
mtx_lock(&np->n_mtx);
if ((np->n_flag & NNONCACHE) && (vp->v_type == VREG)) {
mtx_unlock(&np->n_mtx);
nfs_printf("nfs_getpages: called on non-cacheable vnode??\n");
return (VM_PAGER_ERROR);
} else
mtx_unlock(&np->n_mtx);
}
mtx_lock(&nmp->nm_mtx);
if ((nmp->nm_flag & NFSMNT_NFSV3) != 0 &&
(nmp->nm_state & NFSSTA_GOTFSINFO) == 0) {
mtx_unlock(&nmp->nm_mtx);
/* We'll never get here for v4, because we always have fsinfo */
(void)nfs_fsinfo(nmp, vp, cred, td);
} else
mtx_unlock(&nmp->nm_mtx);
npages = btoc(count);
/*
* If the requested page is partially valid, just return it and
* allow the pager to zero-out the blanks. Partially valid pages
* can only occur at the file EOF.
*/
VM_OBJECT_LOCK(object);
if (pages[ap->a_reqpage]->valid != 0) {
for (i = 0; i < npages; ++i) {
if (i != ap->a_reqpage) {
vm_page_lock(pages[i]);
vm_page_free(pages[i]);
vm_page_unlock(pages[i]);
}
}
VM_OBJECT_UNLOCK(object);
return (0);
}
VM_OBJECT_UNLOCK(object);
/*
* We use only the kva address for the buffer, but this is extremely
* convienient and fast.
*/
bp = getpbuf(&nfs_pbuf_freecnt);
kva = (vm_offset_t) bp->b_data;
pmap_qenter(kva, pages, npages);
PCPU_INC(cnt.v_vnodein);
PCPU_ADD(cnt.v_vnodepgsin, npages);
iov.iov_base = (caddr_t) kva;
iov.iov_len = count;
uio.uio_iov = &iov;
uio.uio_iovcnt = 1;
uio.uio_offset = IDX_TO_OFF(pages[0]->pindex);
uio.uio_resid = count;
uio.uio_segflg = UIO_SYSSPACE;
uio.uio_rw = UIO_READ;
uio.uio_td = td;
error = (nmp->nm_rpcops->nr_readrpc)(vp, &uio, cred);
pmap_qremove(kva, npages);
relpbuf(bp, &nfs_pbuf_freecnt);
if (error && (uio.uio_resid == count)) {
nfs_printf("nfs_getpages: error %d\n", error);
VM_OBJECT_LOCK(object);
for (i = 0; i < npages; ++i) {
if (i != ap->a_reqpage) {
vm_page_lock(pages[i]);
vm_page_free(pages[i]);
vm_page_unlock(pages[i]);
}
}
VM_OBJECT_UNLOCK(object);
return (VM_PAGER_ERROR);
}
/*
* Calculate the number of bytes read and validate only that number
* of bytes. Note that due to pending writes, size may be 0. This
* does not mean that the remaining data is invalid!
*/
size = count - uio.uio_resid;
VM_OBJECT_LOCK(object);
for (i = 0, toff = 0; i < npages; i++, toff = nextoff) {
vm_page_t m;
nextoff = toff + PAGE_SIZE;
m = pages[i];
if (nextoff <= size) {
/*
* Read operation filled an entire page
*/
m->valid = VM_PAGE_BITS_ALL;
KASSERT(m->dirty == 0,
("nfs_getpages: page %p is dirty", m));
} else if (size > toff) {
/*
* Read operation filled a partial page.
*/
m->valid = 0;
vm_page_set_valid_range(m, 0, size - toff);
KASSERT(m->dirty == 0,
("nfs_getpages: page %p is dirty", m));
} else {
/*
* Read operation was short. If no error
* occured we may have hit a zero-fill
* section. We leave valid set to 0, and page
* is freed by vm_page_readahead_finish() if
* its index is not equal to requested, or
* page is zeroed and set valid by
* vm_pager_get_pages() for requested page.
*/
;
}
if (i != ap->a_reqpage)
vm_page_readahead_finish(m);
}
VM_OBJECT_UNLOCK(object);
return (0);
}
/*
* Vnode op for VM putpages.
*/
int
nfs_putpages(struct vop_putpages_args *ap)
{
struct uio uio;
struct iovec iov;
vm_offset_t kva;
struct buf *bp;
int iomode, must_commit, i, error, npages, count;
off_t offset;
int *rtvals;
struct vnode *vp;
struct thread *td;
struct ucred *cred;
struct nfsmount *nmp;
struct nfsnode *np;
vm_page_t *pages;
vp = ap->a_vp;
np = VTONFS(vp);
td = curthread; /* XXX */
/* Set the cred to n_writecred for the write rpcs. */
if (np->n_writecred != NULL)
cred = crhold(np->n_writecred);
else
cred = crhold(curthread->td_ucred); /* XXX */
nmp = VFSTONFS(vp->v_mount);
pages = ap->a_m;
count = ap->a_count;
rtvals = ap->a_rtvals;
npages = btoc(count);
offset = IDX_TO_OFF(pages[0]->pindex);
mtx_lock(&nmp->nm_mtx);
if ((nmp->nm_flag & NFSMNT_NFSV3) != 0 &&
(nmp->nm_state & NFSSTA_GOTFSINFO) == 0) {
mtx_unlock(&nmp->nm_mtx);
(void)nfs_fsinfo(nmp, vp, cred, td);
} else
mtx_unlock(&nmp->nm_mtx);
mtx_lock(&np->n_mtx);
if (nfs_directio_enable && !nfs_directio_allow_mmap &&
(np->n_flag & NNONCACHE) && (vp->v_type == VREG)) {
mtx_unlock(&np->n_mtx);
nfs_printf("nfs_putpages: called on noncache-able vnode??\n");
mtx_lock(&np->n_mtx);
}
for (i = 0; i < npages; i++)
rtvals[i] = VM_PAGER_ERROR;
/*
* When putting pages, do not extend file past EOF.
*/
if (offset + count > np->n_size) {
count = np->n_size - offset;
if (count < 0)
count = 0;
}
mtx_unlock(&np->n_mtx);
/*
* We use only the kva address for the buffer, but this is extremely
* convienient and fast.
*/
bp = getpbuf(&nfs_pbuf_freecnt);
kva = (vm_offset_t) bp->b_data;
pmap_qenter(kva, pages, npages);
PCPU_INC(cnt.v_vnodeout);
PCPU_ADD(cnt.v_vnodepgsout, count);
iov.iov_base = (caddr_t) kva;
iov.iov_len = count;
uio.uio_iov = &iov;
uio.uio_iovcnt = 1;
uio.uio_offset = offset;
uio.uio_resid = count;
uio.uio_segflg = UIO_SYSSPACE;
uio.uio_rw = UIO_WRITE;
uio.uio_td = td;
if ((ap->a_sync & VM_PAGER_PUT_SYNC) == 0)
iomode = NFSV3WRITE_UNSTABLE;
else
iomode = NFSV3WRITE_FILESYNC;
error = (nmp->nm_rpcops->nr_writerpc)(vp, &uio, cred, &iomode, &must_commit);
crfree(cred);
pmap_qremove(kva, npages);
relpbuf(bp, &nfs_pbuf_freecnt);
if (!error) {
vnode_pager_undirty_pages(pages, rtvals, count - uio.uio_resid);
if (must_commit) {
nfs_clearcommit(vp->v_mount);
}
}
return rtvals[0];
}
/*
* For nfs, cache consistency can only be maintained approximately.
* Although RFC1094 does not specify the criteria, the following is
* believed to be compatible with the reference port.
* For nfs:
* If the file's modify time on the server has changed since the
* last read rpc or you have written to the file,
* you may have lost data cache consistency with the
* server, so flush all of the file's data out of the cache.
* Then force a getattr rpc to ensure that you have up to date
* attributes.
* NB: This implies that cache data can be read when up to
* NFS_ATTRTIMEO seconds out of date. If you find that you need current
* attributes this could be forced by setting n_attrstamp to 0 before
* the VOP_GETATTR() call.
*/
static inline int
nfs_bioread_check_cons(struct vnode *vp, struct thread *td, struct ucred *cred)
{
int error = 0;
struct vattr vattr;
struct nfsnode *np = VTONFS(vp);
int old_lock;
struct nfsmount *nmp = VFSTONFS(vp->v_mount);
/*
* Grab the exclusive lock before checking whether the cache is
* consistent.
* XXX - We can make this cheaper later (by acquiring cheaper locks).
* But for now, this suffices.
*/
old_lock = nfs_upgrade_vnlock(vp);
if (vp->v_iflag & VI_DOOMED) {
nfs_downgrade_vnlock(vp, old_lock);
return (EBADF);
}
mtx_lock(&np->n_mtx);
if (np->n_flag & NMODIFIED) {
mtx_unlock(&np->n_mtx);
if (vp->v_type != VREG) {
if (vp->v_type != VDIR)
panic("nfs: bioread, not dir");
(nmp->nm_rpcops->nr_invaldir)(vp);
error = nfs_vinvalbuf(vp, V_SAVE, td, 1);
if (error)
goto out;
}
np->n_attrstamp = 0;
KDTRACE_NFS_ATTRCACHE_FLUSH_DONE(vp);
error = VOP_GETATTR(vp, &vattr, cred);
if (error)
goto out;
mtx_lock(&np->n_mtx);
np->n_mtime = vattr.va_mtime;
mtx_unlock(&np->n_mtx);
} else {
mtx_unlock(&np->n_mtx);
error = VOP_GETATTR(vp, &vattr, cred);
if (error)
return (error);
mtx_lock(&np->n_mtx);
if ((np->n_flag & NSIZECHANGED)
|| (NFS_TIMESPEC_COMPARE(&np->n_mtime, &vattr.va_mtime))) {
mtx_unlock(&np->n_mtx);
if (vp->v_type == VDIR)
(nmp->nm_rpcops->nr_invaldir)(vp);
error = nfs_vinvalbuf(vp, V_SAVE, td, 1);
if (error)
goto out;
mtx_lock(&np->n_mtx);
np->n_mtime = vattr.va_mtime;
np->n_flag &= ~NSIZECHANGED;
}
mtx_unlock(&np->n_mtx);
}
out:
nfs_downgrade_vnlock(vp, old_lock);
return error;
}
/*
* Vnode op for read using bio
*/
int
nfs_bioread(struct vnode *vp, struct uio *uio, int ioflag, struct ucred *cred)
{
struct nfsnode *np = VTONFS(vp);
int biosize, i;
struct buf *bp, *rabp;
struct thread *td;
struct nfsmount *nmp = VFSTONFS(vp->v_mount);
daddr_t lbn, rabn;
off_t end;
int bcount;
int seqcount;
int nra, error = 0, n = 0, on = 0;
KASSERT(uio->uio_rw == UIO_READ, ("nfs_read mode"));
if (uio->uio_resid == 0)
return (0);
if (uio->uio_offset < 0) /* XXX VDIR cookies can be negative */
return (EINVAL);
td = uio->uio_td;
mtx_lock(&nmp->nm_mtx);
if ((nmp->nm_flag & NFSMNT_NFSV3) != 0 &&
(nmp->nm_state & NFSSTA_GOTFSINFO) == 0) {
mtx_unlock(&nmp->nm_mtx);
(void)nfs_fsinfo(nmp, vp, cred, td);
} else
mtx_unlock(&nmp->nm_mtx);
end = uio->uio_offset + uio->uio_resid;
if (vp->v_type != VDIR &&
(end > nmp->nm_maxfilesize || end < uio->uio_offset))
return (EFBIG);
if (nfs_directio_enable && (ioflag & IO_DIRECT) && (vp->v_type == VREG))
/* No caching/ no readaheads. Just read data into the user buffer */
return nfs_readrpc(vp, uio, cred);
biosize = vp->v_bufobj.bo_bsize;
seqcount = (int)((off_t)(ioflag >> IO_SEQSHIFT) * biosize / BKVASIZE);
error = nfs_bioread_check_cons(vp, td, cred);
if (error)
return error;
do {
u_quad_t nsize;
mtx_lock(&np->n_mtx);
nsize = np->n_size;
mtx_unlock(&np->n_mtx);
switch (vp->v_type) {
case VREG:
nfsstats.biocache_reads++;
lbn = uio->uio_offset / biosize;
on = uio->uio_offset & (biosize - 1);
/*
* Start the read ahead(s), as required.
*/
if (nmp->nm_readahead > 0) {
for (nra = 0; nra < nmp->nm_readahead && nra < seqcount &&
(off_t)(lbn + 1 + nra) * biosize < nsize; nra++) {
rabn = lbn + 1 + nra;
if (incore(&vp->v_bufobj, rabn) == NULL) {
rabp = nfs_getcacheblk(vp, rabn, biosize, td);
if (!rabp) {
error = nfs_sigintr(nmp, td);
return (error ? error : EINTR);
}
if ((rabp->b_flags & (B_CACHE|B_DELWRI)) == 0) {
rabp->b_flags |= B_ASYNC;
rabp->b_iocmd = BIO_READ;
vfs_busy_pages(rabp, 0);
if (nfs_asyncio(nmp, rabp, cred, td)) {
rabp->b_flags |= B_INVAL;
rabp->b_ioflags |= BIO_ERROR;
vfs_unbusy_pages(rabp);
brelse(rabp);
break;
}
} else {
brelse(rabp);
}
}
}
}
/* Note that bcount is *not* DEV_BSIZE aligned. */
bcount = biosize;
if ((off_t)lbn * biosize >= nsize) {
bcount = 0;
} else if ((off_t)(lbn + 1) * biosize > nsize) {
bcount = nsize - (off_t)lbn * biosize;
}
bp = nfs_getcacheblk(vp, lbn, bcount, td);
if (!bp) {
error = nfs_sigintr(nmp, td);
return (error ? error : EINTR);
}
/*
* If B_CACHE is not set, we must issue the read. If this
* fails, we return an error.
*/
if ((bp->b_flags & B_CACHE) == 0) {
bp->b_iocmd = BIO_READ;
vfs_busy_pages(bp, 0);
error = nfs_doio(vp, bp, cred, td);
if (error) {
brelse(bp);
return (error);
}
}
/*
* on is the offset into the current bp. Figure out how many
* bytes we can copy out of the bp. Note that bcount is
* NOT DEV_BSIZE aligned.
*
* Then figure out how many bytes we can copy into the uio.
*/
n = 0;
if (on < bcount)
n = MIN((unsigned)(bcount - on), uio->uio_resid);
break;
case VLNK:
nfsstats.biocache_readlinks++;
bp = nfs_getcacheblk(vp, (daddr_t)0, NFS_MAXPATHLEN, td);
if (!bp) {
error = nfs_sigintr(nmp, td);
return (error ? error : EINTR);
}
if ((bp->b_flags & B_CACHE) == 0) {
bp->b_iocmd = BIO_READ;
vfs_busy_pages(bp, 0);
error = nfs_doio(vp, bp, cred, td);
if (error) {
bp->b_ioflags |= BIO_ERROR;
brelse(bp);
return (error);
}
}
n = MIN(uio->uio_resid, NFS_MAXPATHLEN - bp->b_resid);
on = 0;
break;
case VDIR:
nfsstats.biocache_readdirs++;
if (np->n_direofoffset
&& uio->uio_offset >= np->n_direofoffset) {
return (0);
}
lbn = (uoff_t)uio->uio_offset / NFS_DIRBLKSIZ;
on = uio->uio_offset & (NFS_DIRBLKSIZ - 1);
bp = nfs_getcacheblk(vp, lbn, NFS_DIRBLKSIZ, td);
if (!bp) {
error = nfs_sigintr(nmp, td);
return (error ? error : EINTR);
}
if ((bp->b_flags & B_CACHE) == 0) {
bp->b_iocmd = BIO_READ;
vfs_busy_pages(bp, 0);
error = nfs_doio(vp, bp, cred, td);
if (error) {
brelse(bp);
}
while (error == NFSERR_BAD_COOKIE) {
(nmp->nm_rpcops->nr_invaldir)(vp);
error = nfs_vinvalbuf(vp, 0, td, 1);
/*
* Yuck! The directory has been modified on the
* server. The only way to get the block is by
* reading from the beginning to get all the
* offset cookies.
*
* Leave the last bp intact unless there is an error.
* Loop back up to the while if the error is another
* NFSERR_BAD_COOKIE (double yuch!).
*/
for (i = 0; i <= lbn && !error; i++) {
if (np->n_direofoffset
&& (i * NFS_DIRBLKSIZ) >= np->n_direofoffset)
return (0);
bp = nfs_getcacheblk(vp, i, NFS_DIRBLKSIZ, td);
if (!bp) {
error = nfs_sigintr(nmp, td);
return (error ? error : EINTR);
}
if ((bp->b_flags & B_CACHE) == 0) {
bp->b_iocmd = BIO_READ;
vfs_busy_pages(bp, 0);
error = nfs_doio(vp, bp, cred, td);
/*
* no error + B_INVAL == directory EOF,
* use the block.
*/
if (error == 0 && (bp->b_flags & B_INVAL))
break;
}
/*
* An error will throw away the block and the
* for loop will break out. If no error and this
* is not the block we want, we throw away the
* block and go for the next one via the for loop.
*/
if (error || i < lbn)
brelse(bp);
}
}
/*
* The above while is repeated if we hit another cookie
* error. If we hit an error and it wasn't a cookie error,
* we give up.
*/
if (error)
return (error);
}
/*
* If not eof and read aheads are enabled, start one.
* (You need the current block first, so that you have the
* directory offset cookie of the next block.)
*/
if (nmp->nm_readahead > 0 &&
(bp->b_flags & B_INVAL) == 0 &&
(np->n_direofoffset == 0 ||
(lbn + 1) * NFS_DIRBLKSIZ < np->n_direofoffset) &&
incore(&vp->v_bufobj, lbn + 1) == NULL) {
rabp = nfs_getcacheblk(vp, lbn + 1, NFS_DIRBLKSIZ, td);
if (rabp) {
if ((rabp->b_flags & (B_CACHE|B_DELWRI)) == 0) {
rabp->b_flags |= B_ASYNC;
rabp->b_iocmd = BIO_READ;
vfs_busy_pages(rabp, 0);
if (nfs_asyncio(nmp, rabp, cred, td)) {
rabp->b_flags |= B_INVAL;
rabp->b_ioflags |= BIO_ERROR;
vfs_unbusy_pages(rabp);
brelse(rabp);
}
} else {
brelse(rabp);
}
}
}
/*
* Unlike VREG files, whos buffer size ( bp->b_bcount ) is
* chopped for the EOF condition, we cannot tell how large
* NFS directories are going to be until we hit EOF. So
* an NFS directory buffer is *not* chopped to its EOF. Now,
* it just so happens that b_resid will effectively chop it
* to EOF. *BUT* this information is lost if the buffer goes
* away and is reconstituted into a B_CACHE state ( due to
* being VMIO ) later. So we keep track of the directory eof
* in np->n_direofoffset and chop it off as an extra step
* right here.
*/
n = lmin(uio->uio_resid, NFS_DIRBLKSIZ - bp->b_resid - on);
if (np->n_direofoffset && n > np->n_direofoffset - uio->uio_offset)
n = np->n_direofoffset - uio->uio_offset;
break;
default:
nfs_printf(" nfs_bioread: type %x unexpected\n", vp->v_type);
bp = NULL;
break;
};
if (n > 0) {
error = uiomove(bp->b_data + on, (int)n, uio);
}
if (vp->v_type == VLNK)
n = 0;
if (bp != NULL)
brelse(bp);
} while (error == 0 && uio->uio_resid > 0 && n > 0);
return (error);
}
/*
* The NFS write path cannot handle iovecs with len > 1. So we need to
* break up iovecs accordingly (restricting them to wsize).
* For the SYNC case, we can do this with 1 copy (user buffer -> mbuf).
* For the ASYNC case, 2 copies are needed. The first a copy from the
* user buffer to a staging buffer and then a second copy from the staging
* buffer to mbufs. This can be optimized by copying from the user buffer
* directly into mbufs and passing the chain down, but that requires a
* fair amount of re-working of the relevant codepaths (and can be done
* later).
*/
static int
nfs_directio_write(vp, uiop, cred, ioflag)
struct vnode *vp;
struct uio *uiop;
struct ucred *cred;
int ioflag;
{
int error;
struct nfsmount *nmp = VFSTONFS(vp->v_mount);
struct thread *td = uiop->uio_td;
int size;
int wsize;
mtx_lock(&nmp->nm_mtx);
wsize = nmp->nm_wsize;
mtx_unlock(&nmp->nm_mtx);
if (ioflag & IO_SYNC) {
int iomode, must_commit;
struct uio uio;
struct iovec iov;
do_sync:
while (uiop->uio_resid > 0) {
size = MIN(uiop->uio_resid, wsize);
size = MIN(uiop->uio_iov->iov_len, size);
iov.iov_base = uiop->uio_iov->iov_base;
iov.iov_len = size;
uio.uio_iov = &iov;
uio.uio_iovcnt = 1;
uio.uio_offset = uiop->uio_offset;
uio.uio_resid = size;
uio.uio_segflg = UIO_USERSPACE;
uio.uio_rw = UIO_WRITE;
uio.uio_td = td;
iomode = NFSV3WRITE_FILESYNC;
error = (nmp->nm_rpcops->nr_writerpc)(vp, &uio, cred,
&iomode, &must_commit);
KASSERT((must_commit == 0),
("nfs_directio_write: Did not commit write"));
if (error)
return (error);
uiop->uio_offset += size;
uiop->uio_resid -= size;
if (uiop->uio_iov->iov_len <= size) {
uiop->uio_iovcnt--;
uiop->uio_iov++;
} else {
uiop->uio_iov->iov_base =
(char *)uiop->uio_iov->iov_base + size;
uiop->uio_iov->iov_len -= size;
}
}
} else {
struct uio *t_uio;
struct iovec *t_iov;
struct buf *bp;
/*
* Break up the write into blocksize chunks and hand these
* over to nfsiod's for write back.
* Unfortunately, this incurs a copy of the data. Since
* the user could modify the buffer before the write is
* initiated.
*
* The obvious optimization here is that one of the 2 copies
* in the async write path can be eliminated by copying the
* data here directly into mbufs and passing the mbuf chain
* down. But that will require a fair amount of re-working
* of the code and can be done if there's enough interest
* in NFS directio access.
*/
while (uiop->uio_resid > 0) {
size = MIN(uiop->uio_resid, wsize);
size = MIN(uiop->uio_iov->iov_len, size);
bp = getpbuf(&nfs_pbuf_freecnt);
t_uio = malloc(sizeof(struct uio), M_NFSDIRECTIO, M_WAITOK);
t_iov = malloc(sizeof(struct iovec), M_NFSDIRECTIO, M_WAITOK);
t_iov->iov_base = malloc(size, M_NFSDIRECTIO, M_WAITOK);
t_iov->iov_len = size;
t_uio->uio_iov = t_iov;
t_uio->uio_iovcnt = 1;
t_uio->uio_offset = uiop->uio_offset;
t_uio->uio_resid = size;
t_uio->uio_segflg = UIO_SYSSPACE;
t_uio->uio_rw = UIO_WRITE;
t_uio->uio_td = td;
KASSERT(uiop->uio_segflg == UIO_USERSPACE ||
uiop->uio_segflg == UIO_SYSSPACE,
("nfs_directio_write: Bad uio_segflg"));
if (uiop->uio_segflg == UIO_USERSPACE) {
error = copyin(uiop->uio_iov->iov_base,
t_iov->iov_base, size);
if (error != 0)
goto err_free;
} else
/*
* UIO_SYSSPACE may never happen, but handle
* it just in case it does.
*/
bcopy(uiop->uio_iov->iov_base, t_iov->iov_base,
size);
bp->b_flags |= B_DIRECT;
bp->b_iocmd = BIO_WRITE;
if (cred != NOCRED) {
crhold(cred);
bp->b_wcred = cred;
} else
bp->b_wcred = NOCRED;
bp->b_caller1 = (void *)t_uio;
bp->b_vp = vp;
error = nfs_asyncio(nmp, bp, NOCRED, td);
err_free:
if (error) {
free(t_iov->iov_base, M_NFSDIRECTIO);
free(t_iov, M_NFSDIRECTIO);
free(t_uio, M_NFSDIRECTIO);
bp->b_vp = NULL;
relpbuf(bp, &nfs_pbuf_freecnt);
if (error == EINTR)
return (error);
goto do_sync;
}
uiop->uio_offset += size;
uiop->uio_resid -= size;
if (uiop->uio_iov->iov_len <= size) {
uiop->uio_iovcnt--;
uiop->uio_iov++;
} else {
uiop->uio_iov->iov_base =
(char *)uiop->uio_iov->iov_base + size;
uiop->uio_iov->iov_len -= size;
}
}
}
return (0);
}
/*
* Vnode op for write using bio
*/
int
nfs_write(struct vop_write_args *ap)
{
int biosize;
struct uio *uio = ap->a_uio;
struct thread *td = uio->uio_td;
struct vnode *vp = ap->a_vp;
struct nfsnode *np = VTONFS(vp);
struct ucred *cred = ap->a_cred;
int ioflag = ap->a_ioflag;
struct buf *bp;
struct vattr vattr;
struct nfsmount *nmp = VFSTONFS(vp->v_mount);
daddr_t lbn;
off_t end;
int bcount;
int n, on, error = 0;
KASSERT(uio->uio_rw == UIO_WRITE, ("nfs_write mode"));
KASSERT(uio->uio_segflg != UIO_USERSPACE || uio->uio_td == curthread,
("nfs_write proc"));
if (vp->v_type != VREG)
return (EIO);
mtx_lock(&np->n_mtx);
if (np->n_flag & NWRITEERR) {
np->n_flag &= ~NWRITEERR;
mtx_unlock(&np->n_mtx);
return (np->n_error);
} else
mtx_unlock(&np->n_mtx);
mtx_lock(&nmp->nm_mtx);
if ((nmp->nm_flag & NFSMNT_NFSV3) != 0 &&
(nmp->nm_state & NFSSTA_GOTFSINFO) == 0) {
mtx_unlock(&nmp->nm_mtx);
(void)nfs_fsinfo(nmp, vp, cred, td);
} else
mtx_unlock(&nmp->nm_mtx);
/*
* Synchronously flush pending buffers if we are in synchronous
* mode or if we are appending.
*/
if (ioflag & (IO_APPEND | IO_SYNC)) {
mtx_lock(&np->n_mtx);
if (np->n_flag & NMODIFIED) {
mtx_unlock(&np->n_mtx);
#ifdef notyet /* Needs matching nonblock semantics elsewhere, too. */
/*
* Require non-blocking, synchronous writes to
* dirty files to inform the program it needs
* to fsync(2) explicitly.
*/
if (ioflag & IO_NDELAY)
return (EAGAIN);
#endif
flush_and_restart:
np->n_attrstamp = 0;
KDTRACE_NFS_ATTRCACHE_FLUSH_DONE(vp);
error = nfs_vinvalbuf(vp, V_SAVE, td, 1);
if (error)
return (error);
} else
mtx_unlock(&np->n_mtx);
}
/*
* If IO_APPEND then load uio_offset. We restart here if we cannot
* get the append lock.
*/
if (ioflag & IO_APPEND) {
np->n_attrstamp = 0;
KDTRACE_NFS_ATTRCACHE_FLUSH_DONE(vp);
error = VOP_GETATTR(vp, &vattr, cred);
if (error)
return (error);
mtx_lock(&np->n_mtx);
uio->uio_offset = np->n_size;
mtx_unlock(&np->n_mtx);
}
if (uio->uio_offset < 0)
return (EINVAL);
end = uio->uio_offset + uio->uio_resid;
if (end > nmp->nm_maxfilesize || end < uio->uio_offset)
return (EFBIG);
if (uio->uio_resid == 0)
return (0);
if (nfs_directio_enable && (ioflag & IO_DIRECT) && vp->v_type == VREG)
return nfs_directio_write(vp, uio, cred, ioflag);
/*
* Maybe this should be above the vnode op call, but so long as
* file servers have no limits, i don't think it matters
*/
if (vn_rlimit_fsize(vp, uio, td))
return (EFBIG);
biosize = vp->v_bufobj.bo_bsize;
/*
* Find all of this file's B_NEEDCOMMIT buffers. If our writes
* would exceed the local maximum per-file write commit size when
* combined with those, we must decide whether to flush,
* go synchronous, or return error. We don't bother checking
* IO_UNIT -- we just make all writes atomic anyway, as there's
* no point optimizing for something that really won't ever happen.
*/
if (!(ioflag & IO_SYNC)) {
int nflag;
mtx_lock(&np->n_mtx);
nflag = np->n_flag;
mtx_unlock(&np->n_mtx);
int needrestart = 0;
if (nmp->nm_wcommitsize < uio->uio_resid) {
/*
* If this request could not possibly be completed
* without exceeding the maximum outstanding write
* commit size, see if we can convert it into a
* synchronous write operation.
*/
if (ioflag & IO_NDELAY)
return (EAGAIN);
ioflag |= IO_SYNC;
if (nflag & NMODIFIED)
needrestart = 1;
} else if (nflag & NMODIFIED) {
int wouldcommit = 0;
BO_LOCK(&vp->v_bufobj);
if (vp->v_bufobj.bo_dirty.bv_cnt != 0) {
TAILQ_FOREACH(bp, &vp->v_bufobj.bo_dirty.bv_hd,
b_bobufs) {
if (bp->b_flags & B_NEEDCOMMIT)
wouldcommit += bp->b_bcount;
}
}
BO_UNLOCK(&vp->v_bufobj);
/*
* Since we're not operating synchronously and
* bypassing the buffer cache, we are in a commit
* and holding all of these buffers whether
* transmitted or not. If not limited, this
* will lead to the buffer cache deadlocking,
* as no one else can flush our uncommitted buffers.
*/
wouldcommit += uio->uio_resid;
/*
* If we would initially exceed the maximum
* outstanding write commit size, flush and restart.
*/
if (wouldcommit > nmp->nm_wcommitsize)
needrestart = 1;
}
if (needrestart)
goto flush_and_restart;
}
do {
nfsstats.biocache_writes++;
lbn = uio->uio_offset / biosize;
on = uio->uio_offset & (biosize-1);
n = MIN((unsigned)(biosize - on), uio->uio_resid);
again:
/*
* Handle direct append and file extension cases, calculate
* unaligned buffer size.
*/
mtx_lock(&np->n_mtx);
if (uio->uio_offset == np->n_size && n) {
mtx_unlock(&np->n_mtx);
/*
* Get the buffer (in its pre-append state to maintain
* B_CACHE if it was previously set). Resize the
* nfsnode after we have locked the buffer to prevent
* readers from reading garbage.
*/
bcount = on;
bp = nfs_getcacheblk(vp, lbn, bcount, td);
if (bp != NULL) {
long save;
mtx_lock(&np->n_mtx);
np->n_size = uio->uio_offset + n;
np->n_flag |= NMODIFIED;
vnode_pager_setsize(vp, np->n_size);
mtx_unlock(&np->n_mtx);
save = bp->b_flags & B_CACHE;
bcount += n;
allocbuf(bp, bcount);
bp->b_flags |= save;
}
} else {
/*
* Obtain the locked cache block first, and then
* adjust the file's size as appropriate.
*/
bcount = on + n;
if ((off_t)lbn * biosize + bcount < np->n_size) {
if ((off_t)(lbn + 1) * biosize < np->n_size)
bcount = biosize;
else
bcount = np->n_size - (off_t)lbn * biosize;
}
mtx_unlock(&np->n_mtx);
bp = nfs_getcacheblk(vp, lbn, bcount, td);
mtx_lock(&np->n_mtx);
if (uio->uio_offset + n > np->n_size) {
np->n_size = uio->uio_offset + n;
np->n_flag |= NMODIFIED;
vnode_pager_setsize(vp, np->n_size);
}
mtx_unlock(&np->n_mtx);
}
if (!bp) {
error = nfs_sigintr(nmp, td);
if (!error)
error = EINTR;
break;
}
/*
* Issue a READ if B_CACHE is not set. In special-append
* mode, B_CACHE is based on the buffer prior to the write
* op and is typically set, avoiding the read. If a read
* is required in special append mode, the server will
* probably send us a short-read since we extended the file
* on our end, resulting in b_resid == 0 and, thusly,
* B_CACHE getting set.
*
* We can also avoid issuing the read if the write covers
* the entire buffer. We have to make sure the buffer state
* is reasonable in this case since we will not be initiating
* I/O. See the comments in kern/vfs_bio.c's getblk() for
* more information.
*
* B_CACHE may also be set due to the buffer being cached
* normally.
*/
if (on == 0 && n == bcount) {
bp->b_flags |= B_CACHE;
bp->b_flags &= ~B_INVAL;
bp->b_ioflags &= ~BIO_ERROR;
}
if ((bp->b_flags & B_CACHE) == 0) {
bp->b_iocmd = BIO_READ;
vfs_busy_pages(bp, 0);
error = nfs_doio(vp, bp, cred, td);
if (error) {
brelse(bp);
break;
}
}
if (bp->b_wcred == NOCRED)
bp->b_wcred = crhold(cred);
mtx_lock(&np->n_mtx);
np->n_flag |= NMODIFIED;
mtx_unlock(&np->n_mtx);
/*
* If dirtyend exceeds file size, chop it down. This should
* not normally occur but there is an append race where it
* might occur XXX, so we log it.
*
* If the chopping creates a reverse-indexed or degenerate
* situation with dirtyoff/end, we 0 both of them.
*/
if (bp->b_dirtyend > bcount) {
nfs_printf("NFS append race @%lx:%d\n",
(long)bp->b_blkno * DEV_BSIZE,
bp->b_dirtyend - bcount);
bp->b_dirtyend = bcount;
}
if (bp->b_dirtyoff >= bp->b_dirtyend)
bp->b_dirtyoff = bp->b_dirtyend = 0;
/*
* If the new write will leave a contiguous dirty
* area, just update the b_dirtyoff and b_dirtyend,
* otherwise force a write rpc of the old dirty area.
*
* While it is possible to merge discontiguous writes due to
* our having a B_CACHE buffer ( and thus valid read data
* for the hole), we don't because it could lead to
* significant cache coherency problems with multiple clients,
* especially if locking is implemented later on.
*
* as an optimization we could theoretically maintain
* a linked list of discontinuous areas, but we would still
* have to commit them separately so there isn't much
* advantage to it except perhaps a bit of asynchronization.
*/
if (bp->b_dirtyend > 0 &&
(on > bp->b_dirtyend || (on + n) < bp->b_dirtyoff)) {
if (bwrite(bp) == EINTR) {
error = EINTR;
break;
}
goto again;
}
error = uiomove((char *)bp->b_data + on, n, uio);
/*
* Since this block is being modified, it must be written
* again and not just committed. Since write clustering does
* not work for the stage 1 data write, only the stage 2
* commit rpc, we have to clear B_CLUSTEROK as well.
*/
bp->b_flags &= ~(B_NEEDCOMMIT | B_CLUSTEROK);
if (error) {
bp->b_ioflags |= BIO_ERROR;
brelse(bp);
break;
}
/*
* Only update dirtyoff/dirtyend if not a degenerate
* condition.
*/
if (n) {
if (bp->b_dirtyend > 0) {
bp->b_dirtyoff = min(on, bp->b_dirtyoff);
bp->b_dirtyend = max((on + n), bp->b_dirtyend);
} else {
bp->b_dirtyoff = on;
bp->b_dirtyend = on + n;
}
vfs_bio_set_valid(bp, on, n);
}
/*
* If IO_SYNC do bwrite().
*
* IO_INVAL appears to be unused. The idea appears to be
* to turn off caching in this case. Very odd. XXX
*/
if ((ioflag & IO_SYNC)) {
if (ioflag & IO_INVAL)
bp->b_flags |= B_NOCACHE;
error = bwrite(bp);
if (error)
break;
} else if ((n + on) == biosize) {
bp->b_flags |= B_ASYNC;
(void) (nmp->nm_rpcops->nr_writebp)(bp, 0, NULL);
} else {
bdwrite(bp);
}
} while (uio->uio_resid > 0 && n > 0);
return (error);
}
/*
* Get an nfs cache block.
*
* Allocate a new one if the block isn't currently in the cache
* and return the block marked busy. If the calling process is
* interrupted by a signal for an interruptible mount point, return
* NULL.
*
* The caller must carefully deal with the possible B_INVAL state of
* the buffer. nfs_doio() clears B_INVAL (and nfs_asyncio() clears it
* indirectly), so synchronous reads can be issued without worrying about
* the B_INVAL state. We have to be a little more careful when dealing
* with writes (see comments in nfs_write()) when extending a file past
* its EOF.
*/
static struct buf *
nfs_getcacheblk(struct vnode *vp, daddr_t bn, int size, struct thread *td)
{
struct buf *bp;
struct mount *mp;
struct nfsmount *nmp;
mp = vp->v_mount;
nmp = VFSTONFS(mp);
if (nmp->nm_flag & NFSMNT_INT) {
sigset_t oldset;
nfs_set_sigmask(td, &oldset);
bp = getblk(vp, bn, size, NFS_PCATCH, 0, 0);
nfs_restore_sigmask(td, &oldset);
while (bp == NULL) {
if (nfs_sigintr(nmp, td))
return (NULL);
bp = getblk(vp, bn, size, 0, 2 * hz, 0);
}
} else {
bp = getblk(vp, bn, size, 0, 0, 0);
}
if (vp->v_type == VREG)
bp->b_blkno = bn * (vp->v_bufobj.bo_bsize / DEV_BSIZE);
return (bp);
}
/*
* Flush and invalidate all dirty buffers. If another process is already
* doing the flush, just wait for completion.
*/
int
nfs_vinvalbuf(struct vnode *vp, int flags, struct thread *td, int intrflg)
{
struct nfsnode *np = VTONFS(vp);
struct nfsmount *nmp = VFSTONFS(vp->v_mount);
int error = 0, slpflag, slptimeo;
int old_lock = 0;
ASSERT_VOP_LOCKED(vp, "nfs_vinvalbuf");
if ((nmp->nm_flag & NFSMNT_INT) == 0)
intrflg = 0;
if (intrflg) {
slpflag = NFS_PCATCH;
slptimeo = 2 * hz;
} else {
slpflag = 0;
slptimeo = 0;
}
old_lock = nfs_upgrade_vnlock(vp);
if (vp->v_iflag & VI_DOOMED) {
/*
* Since vgonel() uses the generic vinvalbuf() to flush
* dirty buffers and it does not call this function, it
* is safe to just return OK when VI_DOOMED is set.
*/
nfs_downgrade_vnlock(vp, old_lock);
return (0);
}
/*
* Now, flush as required.
*/
if ((flags & V_SAVE) && (vp->v_bufobj.bo_object != NULL)) {
VM_OBJECT_LOCK(vp->v_bufobj.bo_object);
vm_object_page_clean(vp->v_bufobj.bo_object, 0, 0, OBJPC_SYNC);
VM_OBJECT_UNLOCK(vp->v_bufobj.bo_object);
/*
* If the page clean was interrupted, fail the invalidation.
* Not doing so, we run the risk of losing dirty pages in the
* vinvalbuf() call below.
*/
if (intrflg && (error = nfs_sigintr(nmp, td)))
goto out;
}
error = vinvalbuf(vp, flags, slpflag, 0);
while (error) {
if (intrflg && (error = nfs_sigintr(nmp, td)))
goto out;
error = vinvalbuf(vp, flags, 0, slptimeo);
}
mtx_lock(&np->n_mtx);
if (np->n_directio_asyncwr == 0)
np->n_flag &= ~NMODIFIED;
mtx_unlock(&np->n_mtx);
out:
nfs_downgrade_vnlock(vp, old_lock);
return error;
}
/*
* Initiate asynchronous I/O. Return an error if no nfsiods are available.
* This is mainly to avoid queueing async I/O requests when the nfsiods
* are all hung on a dead server.
*
* Note: nfs_asyncio() does not clear (BIO_ERROR|B_INVAL) but when the bp
* is eventually dequeued by the async daemon, nfs_doio() *will*.
*/
int
nfs_asyncio(struct nfsmount *nmp, struct buf *bp, struct ucred *cred, struct thread *td)
{
int iod;
int gotiod;
int slpflag = 0;
int slptimeo = 0;
int error, error2;
/*
* Commits are usually short and sweet so lets save some cpu and
* leave the async daemons for more important rpc's (such as reads
* and writes).
*/
mtx_lock(&nfs_iod_mtx);
if (bp->b_iocmd == BIO_WRITE && (bp->b_flags & B_NEEDCOMMIT) &&
(nmp->nm_bufqiods > nfs_numasync / 2)) {
mtx_unlock(&nfs_iod_mtx);
return(EIO);
}
again:
if (nmp->nm_flag & NFSMNT_INT)
slpflag = NFS_PCATCH;
gotiod = FALSE;
/*
* Find a free iod to process this request.
*/
for (iod = 0; iod < nfs_numasync; iod++)
if (nfs_iodwant[iod] == NFSIOD_AVAILABLE) {
gotiod = TRUE;
break;
}
/*
* Try to create one if none are free.
*/
if (!gotiod)
nfs_nfsiodnew();
else {
/*
* Found one, so wake it up and tell it which
* mount to process.
*/
NFS_DPF(ASYNCIO, ("nfs_asyncio: waking iod %d for mount %p\n",
iod, nmp));
nfs_iodwant[iod] = NFSIOD_NOT_AVAILABLE;
nfs_iodmount[iod] = nmp;
nmp->nm_bufqiods++;
wakeup(&nfs_iodwant[iod]);
}
/*
* If none are free, we may already have an iod working on this mount
* point. If so, it will process our request.
*/
if (!gotiod) {
if (nmp->nm_bufqiods > 0) {
NFS_DPF(ASYNCIO,
("nfs_asyncio: %d iods are already processing mount %p\n",
nmp->nm_bufqiods, nmp));
gotiod = TRUE;
}
}
/*
* If we have an iod which can process the request, then queue
* the buffer.
*/
if (gotiod) {
/*
* Ensure that the queue never grows too large. We still want
* to asynchronize so we block rather then return EIO.
*/
while (nmp->nm_bufqlen >= 2 * nfs_numasync) {
NFS_DPF(ASYNCIO,
("nfs_asyncio: waiting for mount %p queue to drain\n", nmp));
nmp->nm_bufqwant = TRUE;
error = nfs_msleep(td, &nmp->nm_bufq, &nfs_iod_mtx,
slpflag | PRIBIO,
"nfsaio", slptimeo);
if (error) {
error2 = nfs_sigintr(nmp, td);
if (error2) {
mtx_unlock(&nfs_iod_mtx);
return (error2);
}
if (slpflag == NFS_PCATCH) {
slpflag = 0;
slptimeo = 2 * hz;
}
}
/*
* We might have lost our iod while sleeping,
* so check and loop if nescessary.
*/
goto again;
}
/* We might have lost our nfsiod */
if (nmp->nm_bufqiods == 0) {
NFS_DPF(ASYNCIO,
("nfs_asyncio: no iods after mount %p queue was drained, looping\n", nmp));
goto again;
}
if (bp->b_iocmd == BIO_READ) {
if (bp->b_rcred == NOCRED && cred != NOCRED)
bp->b_rcred = crhold(cred);
} else {
if (bp->b_wcred == NOCRED && cred != NOCRED)
bp->b_wcred = crhold(cred);
}
if (bp->b_flags & B_REMFREE)
bremfreef(bp);
BUF_KERNPROC(bp);
TAILQ_INSERT_TAIL(&nmp->nm_bufq, bp, b_freelist);
nmp->nm_bufqlen++;
if ((bp->b_flags & B_DIRECT) && bp->b_iocmd == BIO_WRITE) {
mtx_lock(&(VTONFS(bp->b_vp))->n_mtx);
VTONFS(bp->b_vp)->n_flag |= NMODIFIED;
VTONFS(bp->b_vp)->n_directio_asyncwr++;
mtx_unlock(&(VTONFS(bp->b_vp))->n_mtx);
}
mtx_unlock(&nfs_iod_mtx);
return (0);
}
mtx_unlock(&nfs_iod_mtx);
/*
* All the iods are busy on other mounts, so return EIO to
* force the caller to process the i/o synchronously.
*/
NFS_DPF(ASYNCIO, ("nfs_asyncio: no iods available, i/o is synchronous\n"));
return (EIO);
}
void
nfs_doio_directwrite(struct buf *bp)
{
int iomode, must_commit;
struct uio *uiop = (struct uio *)bp->b_caller1;
char *iov_base = uiop->uio_iov->iov_base;
struct nfsmount *nmp = VFSTONFS(bp->b_vp->v_mount);
iomode = NFSV3WRITE_FILESYNC;
uiop->uio_td = NULL; /* NULL since we're in nfsiod */
(nmp->nm_rpcops->nr_writerpc)(bp->b_vp, uiop, bp->b_wcred, &iomode, &must_commit);
KASSERT((must_commit == 0), ("nfs_doio_directwrite: Did not commit write"));
free(iov_base, M_NFSDIRECTIO);
free(uiop->uio_iov, M_NFSDIRECTIO);
free(uiop, M_NFSDIRECTIO);
if ((bp->b_flags & B_DIRECT) && bp->b_iocmd == BIO_WRITE) {
struct nfsnode *np = VTONFS(bp->b_vp);
mtx_lock(&np->n_mtx);
np->n_directio_asyncwr--;
if (np->n_directio_asyncwr == 0) {
VTONFS(bp->b_vp)->n_flag &= ~NMODIFIED;
if ((np->n_flag & NFSYNCWAIT)) {
np->n_flag &= ~NFSYNCWAIT;
wakeup((caddr_t)&np->n_directio_asyncwr);
}
}
mtx_unlock(&np->n_mtx);
}
bp->b_vp = NULL;
relpbuf(bp, &nfs_pbuf_freecnt);
}
/*
* Do an I/O operation to/from a cache block. This may be called
* synchronously or from an nfsiod.
*/
int
nfs_doio(struct vnode *vp, struct buf *bp, struct ucred *cr, struct thread *td)
{
struct uio *uiop;
struct nfsnode *np;
struct nfsmount *nmp;
int error = 0, iomode, must_commit = 0;
struct uio uio;
struct iovec io;
struct proc *p = td ? td->td_proc : NULL;
uint8_t iocmd;
np = VTONFS(vp);
nmp = VFSTONFS(vp->v_mount);
uiop = &uio;
uiop->uio_iov = &io;
uiop->uio_iovcnt = 1;
uiop->uio_segflg = UIO_SYSSPACE;
uiop->uio_td = td;
/*
* clear BIO_ERROR and B_INVAL state prior to initiating the I/O. We
* do this here so we do not have to do it in all the code that
* calls us.
*/
bp->b_flags &= ~B_INVAL;
bp->b_ioflags &= ~BIO_ERROR;
KASSERT(!(bp->b_flags & B_DONE), ("nfs_doio: bp %p already marked done", bp));
iocmd = bp->b_iocmd;
if (iocmd == BIO_READ) {
io.iov_len = uiop->uio_resid = bp->b_bcount;
io.iov_base = bp->b_data;
uiop->uio_rw = UIO_READ;
switch (vp->v_type) {
case VREG:
uiop->uio_offset = ((off_t)bp->b_blkno) * DEV_BSIZE;
nfsstats.read_bios++;
error = (nmp->nm_rpcops->nr_readrpc)(vp, uiop, cr);
if (!error) {
if (uiop->uio_resid) {
/*
* If we had a short read with no error, we must have
* hit a file hole. We should zero-fill the remainder.
* This can also occur if the server hits the file EOF.
*
* Holes used to be able to occur due to pending
* writes, but that is not possible any longer.
*/
int nread = bp->b_bcount - uiop->uio_resid;
int left = uiop->uio_resid;
if (left > 0)
bzero((char *)bp->b_data + nread, left);
uiop->uio_resid = 0;
}
}
/* ASSERT_VOP_LOCKED(vp, "nfs_doio"); */
if (p && (vp->v_vflag & VV_TEXT)) {
mtx_lock(&np->n_mtx);
if (NFS_TIMESPEC_COMPARE(&np->n_mtime, &np->n_vattr.va_mtime)) {
mtx_unlock(&np->n_mtx);
PROC_LOCK(p);
killproc(p, "text file modification");
PROC_UNLOCK(p);
} else
mtx_unlock(&np->n_mtx);
}
break;
case VLNK:
uiop->uio_offset = (off_t)0;
nfsstats.readlink_bios++;
error = (nmp->nm_rpcops->nr_readlinkrpc)(vp, uiop, cr);
break;
case VDIR:
nfsstats.readdir_bios++;
uiop->uio_offset = ((u_quad_t)bp->b_lblkno) * NFS_DIRBLKSIZ;
if ((nmp->nm_flag & NFSMNT_RDIRPLUS) != 0) {
error = nfs_readdirplusrpc(vp, uiop, cr);
if (error == NFSERR_NOTSUPP)
nmp->nm_flag &= ~NFSMNT_RDIRPLUS;
}
if ((nmp->nm_flag & NFSMNT_RDIRPLUS) == 0)
error = nfs_readdirrpc(vp, uiop, cr);
/*
* end-of-directory sets B_INVAL but does not generate an
* error.
*/
if (error == 0 && uiop->uio_resid == bp->b_bcount)
bp->b_flags |= B_INVAL;
break;
default:
nfs_printf("nfs_doio: type %x unexpected\n", vp->v_type);
break;
};
if (error) {
bp->b_ioflags |= BIO_ERROR;
bp->b_error = error;
}
} else {
/*
* If we only need to commit, try to commit
*/
if (bp->b_flags & B_NEEDCOMMIT) {
int retv;
off_t off;
off = ((u_quad_t)bp->b_blkno) * DEV_BSIZE + bp->b_dirtyoff;
retv = (nmp->nm_rpcops->nr_commit)(
vp, off, bp->b_dirtyend-bp->b_dirtyoff,
bp->b_wcred, td);
if (retv == 0) {
bp->b_dirtyoff = bp->b_dirtyend = 0;
bp->b_flags &= ~(B_NEEDCOMMIT | B_CLUSTEROK);
bp->b_resid = 0;
bufdone(bp);
return (0);
}
if (retv == NFSERR_STALEWRITEVERF) {
nfs_clearcommit(vp->v_mount);
}
}
/*
* Setup for actual write
*/
mtx_lock(&np->n_mtx);
if ((off_t)bp->b_blkno * DEV_BSIZE + bp->b_dirtyend > np->n_size)
bp->b_dirtyend = np->n_size - (off_t)bp->b_blkno * DEV_BSIZE;
mtx_unlock(&np->n_mtx);
if (bp->b_dirtyend > bp->b_dirtyoff) {
io.iov_len = uiop->uio_resid = bp->b_dirtyend
- bp->b_dirtyoff;
uiop->uio_offset = (off_t)bp->b_blkno * DEV_BSIZE
+ bp->b_dirtyoff;
io.iov_base = (char *)bp->b_data + bp->b_dirtyoff;
uiop->uio_rw = UIO_WRITE;
nfsstats.write_bios++;
if ((bp->b_flags & (B_ASYNC | B_NEEDCOMMIT | B_NOCACHE | B_CLUSTER)) == B_ASYNC)
iomode = NFSV3WRITE_UNSTABLE;
else
iomode = NFSV3WRITE_FILESYNC;
error = (nmp->nm_rpcops->nr_writerpc)(vp, uiop, cr, &iomode, &must_commit);
/*
* When setting B_NEEDCOMMIT also set B_CLUSTEROK to try
* to cluster the buffers needing commit. This will allow
* the system to submit a single commit rpc for the whole
* cluster. We can do this even if the buffer is not 100%
* dirty (relative to the NFS blocksize), so we optimize the
* append-to-file-case.
*
* (when clearing B_NEEDCOMMIT, B_CLUSTEROK must also be
* cleared because write clustering only works for commit
* rpc's, not for the data portion of the write).
*/
if (!error && iomode == NFSV3WRITE_UNSTABLE) {
bp->b_flags |= B_NEEDCOMMIT;
if (bp->b_dirtyoff == 0
&& bp->b_dirtyend == bp->b_bcount)
bp->b_flags |= B_CLUSTEROK;
} else {
bp->b_flags &= ~(B_NEEDCOMMIT | B_CLUSTEROK);
}
/*
* For an interrupted write, the buffer is still valid
* and the write hasn't been pushed to the server yet,
* so we can't set BIO_ERROR and report the interruption
* by setting B_EINTR. For the B_ASYNC case, B_EINTR
* is not relevant, so the rpc attempt is essentially
* a noop. For the case of a V3 write rpc not being
* committed to stable storage, the block is still
* dirty and requires either a commit rpc or another
* write rpc with iomode == NFSV3WRITE_FILESYNC before
* the block is reused. This is indicated by setting
* the B_DELWRI and B_NEEDCOMMIT flags.
*
* If the buffer is marked B_PAGING, it does not reside on
* the vp's paging queues so we cannot call bdirty(). The
* bp in this case is not an NFS cache block so we should
* be safe. XXX
*
* The logic below breaks up errors into recoverable and
* unrecoverable. For the former, we clear B_INVAL|B_NOCACHE
* and keep the buffer around for potential write retries.
* For the latter (eg ESTALE), we toss the buffer away (B_INVAL)
* and save the error in the nfsnode. This is less than ideal
* but necessary. Keeping such buffers around could potentially
* cause buffer exhaustion eventually (they can never be written
* out, so will get constantly be re-dirtied). It also causes
* all sorts of vfs panics. For non-recoverable write errors,
* also invalidate the attrcache, so we'll be forced to go over
* the wire for this object, returning an error to user on next
* call (most of the time).
*/
if (error == EINTR || error == EIO || error == ETIMEDOUT
|| (!error && (bp->b_flags & B_NEEDCOMMIT))) {
int s;
s = splbio();
bp->b_flags &= ~(B_INVAL|B_NOCACHE);
if ((bp->b_flags & B_PAGING) == 0) {
bdirty(bp);
bp->b_flags &= ~B_DONE;
}
if (error && (bp->b_flags & B_ASYNC) == 0)
bp->b_flags |= B_EINTR;
splx(s);
} else {
if (error) {
bp->b_ioflags |= BIO_ERROR;
bp->b_flags |= B_INVAL;
bp->b_error = np->n_error = error;
mtx_lock(&np->n_mtx);
np->n_flag |= NWRITEERR;
np->n_attrstamp = 0;
KDTRACE_NFS_ATTRCACHE_FLUSH_DONE(vp);
mtx_unlock(&np->n_mtx);
}
bp->b_dirtyoff = bp->b_dirtyend = 0;
}
} else {
bp->b_resid = 0;
bufdone(bp);
return (0);
}
}
bp->b_resid = uiop->uio_resid;
if (must_commit)
nfs_clearcommit(vp->v_mount);
bufdone(bp);
return (error);
}
/*
* Used to aid in handling ftruncate() operations on the NFS client side.
* Truncation creates a number of special problems for NFS. We have to
* throw away VM pages and buffer cache buffers that are beyond EOF, and
* we have to properly handle VM pages or (potentially dirty) buffers
* that straddle the truncation point.
*/
int
nfs_meta_setsize(struct vnode *vp, struct ucred *cred, struct thread *td, u_quad_t nsize)
{
struct nfsnode *np = VTONFS(vp);
u_quad_t tsize;
int biosize = vp->v_bufobj.bo_bsize;
int error = 0;
mtx_lock(&np->n_mtx);
tsize = np->n_size;
np->n_size = nsize;
mtx_unlock(&np->n_mtx);
if (nsize < tsize) {
struct buf *bp;
daddr_t lbn;
int bufsize;
/*
* vtruncbuf() doesn't get the buffer overlapping the
* truncation point. We may have a B_DELWRI and/or B_CACHE
* buffer that now needs to be truncated.
*/
error = vtruncbuf(vp, cred, nsize, biosize);
lbn = nsize / biosize;
bufsize = nsize & (biosize - 1);
bp = nfs_getcacheblk(vp, lbn, bufsize, td);
if (!bp)
return EINTR;
if (bp->b_dirtyoff > bp->b_bcount)
bp->b_dirtyoff = bp->b_bcount;
if (bp->b_dirtyend > bp->b_bcount)
bp->b_dirtyend = bp->b_bcount;
bp->b_flags |= B_RELBUF; /* don't leave garbage around */
brelse(bp);
} else {
vnode_pager_setsize(vp, nsize);
}
return(error);
}
|