1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
2151
2152
2153
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263
2264
2265
2266
2267
2268
2269
2270
2271
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
2286
2287
2288
2289
2290
2291
2292
2293
2294
2295
2296
2297
2298
2299
2300
2301
2302
2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
2429
2430
2431
2432
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
2445
2446
2447
2448
2449
2450
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
2507
2508
2509
2510
2511
2512
2513
2514
2515
2516
2517
2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
2535
2536
2537
2538
2539
2540
2541
2542
2543
2544
2545
2546
2547
2548
2549
2550
2551
2552
2553
2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
2571
2572
2573
2574
2575
2576
2577
2578
2579
2580
2581
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
2616
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
2628
2629
2630
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
2671
2672
2673
2674
2675
2676
2677
2678
2679
2680
2681
2682
2683
2684
|
/*-
* Copyright (c) 1998 Matthew Dillon,
* Copyright (c) 1994 John S. Dyson
* Copyright (c) 1990 University of Utah.
* Copyright (c) 1982, 1986, 1989, 1993
* The Regents of the University of California. All rights reserved.
*
* This code is derived from software contributed to Berkeley by
* the Systems Programming Group of the University of Utah Computer
* Science Department.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by the University of
* California, Berkeley and its contributors.
* 4. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* New Swap System
* Matthew Dillon
*
* Radix Bitmap 'blists'.
*
* - The new swapper uses the new radix bitmap code. This should scale
* to arbitrarily small or arbitrarily large swap spaces and an almost
* arbitrary degree of fragmentation.
*
* Features:
*
* - on the fly reallocation of swap during putpages. The new system
* does not try to keep previously allocated swap blocks for dirty
* pages.
*
* - on the fly deallocation of swap
*
* - No more garbage collection required. Unnecessarily allocated swap
* blocks only exist for dirty vm_page_t's now and these are already
* cycled (in a high-load system) by the pager. We also do on-the-fly
* removal of invalidated swap blocks when a page is destroyed
* or renamed.
*
* from: Utah $Hdr: swap_pager.c 1.4 91/04/30$
*
* @(#)swap_pager.c 8.9 (Berkeley) 3/21/94
* @(#)vm_swap.c 8.5 (Berkeley) 2/17/94
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include "opt_swap.h"
#include "opt_vm.h"
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/conf.h>
#include <sys/kernel.h>
#include <sys/priv.h>
#include <sys/proc.h>
#include <sys/bio.h>
#include <sys/buf.h>
#include <sys/disk.h>
#include <sys/fcntl.h>
#include <sys/mount.h>
#include <sys/namei.h>
#include <sys/vnode.h>
#include <sys/malloc.h>
#include <sys/resource.h>
#include <sys/resourcevar.h>
#include <sys/sysctl.h>
#include <sys/sysproto.h>
#include <sys/blist.h>
#include <sys/lock.h>
#include <sys/sx.h>
#include <sys/vmmeter.h>
#include <security/mac/mac_framework.h>
#include <vm/vm.h>
#include <vm/pmap.h>
#include <vm/vm_map.h>
#include <vm/vm_kern.h>
#include <vm/vm_object.h>
#include <vm/vm_page.h>
#include <vm/vm_pager.h>
#include <vm/vm_pageout.h>
#include <vm/vm_param.h>
#include <vm/swap_pager.h>
#include <vm/vm_extern.h>
#include <vm/uma.h>
#include <geom/geom.h>
/*
* SWB_NPAGES must be a power of 2. It may be set to 1, 2, 4, 8, or 16
* pages per allocation. We recommend you stick with the default of 8.
* The 16-page limit is due to the radix code (kern/subr_blist.c).
*/
#ifndef MAX_PAGEOUT_CLUSTER
#define MAX_PAGEOUT_CLUSTER 16
#endif
#if !defined(SWB_NPAGES)
#define SWB_NPAGES MAX_PAGEOUT_CLUSTER
#endif
/*
* Piecemeal swap metadata structure. Swap is stored in a radix tree.
*
* If SWB_NPAGES is 8 and sizeof(char *) == sizeof(daddr_t), our radix
* is basically 8. Assuming PAGE_SIZE == 4096, one tree level represents
* 32K worth of data, two levels represent 256K, three levels represent
* 2 MBytes. This is acceptable.
*
* Overall memory utilization is about the same as the old swap structure.
*/
#define SWCORRECT(n) (sizeof(void *) * (n) / sizeof(daddr_t))
#define SWAP_META_PAGES (SWB_NPAGES * 2)
#define SWAP_META_MASK (SWAP_META_PAGES - 1)
struct swblock {
struct swblock *swb_hnext;
vm_object_t swb_object;
vm_pindex_t swb_index;
int swb_count;
daddr_t swb_pages[SWAP_META_PAGES];
};
static struct mtx sw_dev_mtx;
static TAILQ_HEAD(, swdevt) swtailq = TAILQ_HEAD_INITIALIZER(swtailq);
static struct swdevt *swdevhd; /* Allocate from here next */
static int nswapdev; /* Number of swap devices */
int swap_pager_avail;
static int swdev_syscall_active = 0; /* serialize swap(on|off) */
static vm_ooffset_t swap_total;
SYSCTL_QUAD(_vm, OID_AUTO, swap_total, CTLFLAG_RD, &swap_total, 0, "");
static vm_ooffset_t swap_reserved;
SYSCTL_QUAD(_vm, OID_AUTO, swap_reserved, CTLFLAG_RD, &swap_reserved, 0, "");
static int overcommit = 0;
SYSCTL_INT(_vm, OID_AUTO, overcommit, CTLFLAG_RW, &overcommit, 0, "");
/* bits from overcommit */
#define SWAP_RESERVE_FORCE_ON (1 << 0)
#define SWAP_RESERVE_RLIMIT_ON (1 << 1)
#define SWAP_RESERVE_ALLOW_NONWIRED (1 << 2)
int
swap_reserve(vm_ooffset_t incr)
{
return (swap_reserve_by_uid(incr, curthread->td_ucred->cr_ruidinfo));
}
int
swap_reserve_by_uid(vm_ooffset_t incr, struct uidinfo *uip)
{
vm_ooffset_t r, s, max;
int res, error;
static int curfail;
static struct timeval lastfail;
if (incr & PAGE_MASK)
panic("swap_reserve: & PAGE_MASK");
res = 0;
error = priv_check(curthread, PRIV_VM_SWAP_NOQUOTA);
mtx_lock(&sw_dev_mtx);
r = swap_reserved + incr;
if (overcommit & SWAP_RESERVE_ALLOW_NONWIRED) {
s = cnt.v_page_count - cnt.v_free_reserved - cnt.v_wire_count;
s *= PAGE_SIZE;
} else
s = 0;
s += swap_total;
if ((overcommit & SWAP_RESERVE_FORCE_ON) == 0 || r <= s ||
(error = priv_check(curthread, PRIV_VM_SWAP_NOQUOTA)) == 0) {
res = 1;
swap_reserved = r;
}
mtx_unlock(&sw_dev_mtx);
if (res) {
PROC_LOCK(curproc);
UIDINFO_VMSIZE_LOCK(uip);
error = priv_check(curthread, PRIV_VM_SWAP_NORLIMIT);
max = (error != 0) ? lim_cur(curproc, RLIMIT_SWAP) : 0;
if (max != 0 && uip->ui_vmsize + incr > max &&
(overcommit & SWAP_RESERVE_RLIMIT_ON) != 0)
res = 0;
else
uip->ui_vmsize += incr;
UIDINFO_VMSIZE_UNLOCK(uip);
PROC_UNLOCK(curproc);
if (!res) {
mtx_lock(&sw_dev_mtx);
swap_reserved -= incr;
mtx_unlock(&sw_dev_mtx);
}
}
if (!res && ppsratecheck(&lastfail, &curfail, 1)) {
printf("uid %d, pid %d: swap reservation for %jd bytes failed\n",
curproc->p_pid, uip->ui_uid, incr);
}
return (res);
}
void
swap_reserve_force(vm_ooffset_t incr)
{
struct uidinfo *uip;
mtx_lock(&sw_dev_mtx);
swap_reserved += incr;
mtx_unlock(&sw_dev_mtx);
uip = curthread->td_ucred->cr_ruidinfo;
PROC_LOCK(curproc);
UIDINFO_VMSIZE_LOCK(uip);
uip->ui_vmsize += incr;
UIDINFO_VMSIZE_UNLOCK(uip);
PROC_UNLOCK(curproc);
}
void
swap_release(vm_ooffset_t decr)
{
struct uidinfo *uip;
PROC_LOCK(curproc);
uip = curthread->td_ucred->cr_ruidinfo;
swap_release_by_uid(decr, uip);
PROC_UNLOCK(curproc);
}
void
swap_release_by_uid(vm_ooffset_t decr, struct uidinfo *uip)
{
if (decr & PAGE_MASK)
panic("swap_release: & PAGE_MASK");
mtx_lock(&sw_dev_mtx);
if (swap_reserved < decr)
panic("swap_reserved < decr");
swap_reserved -= decr;
mtx_unlock(&sw_dev_mtx);
UIDINFO_VMSIZE_LOCK(uip);
if (uip->ui_vmsize < decr)
printf("negative vmsize for uid = %d\n", uip->ui_uid);
uip->ui_vmsize -= decr;
UIDINFO_VMSIZE_UNLOCK(uip);
}
static void swapdev_strategy(struct buf *, struct swdevt *sw);
#define SWM_FREE 0x02 /* free, period */
#define SWM_POP 0x04 /* pop out */
int swap_pager_full = 2; /* swap space exhaustion (task killing) */
static int swap_pager_almost_full = 1; /* swap space exhaustion (w/hysteresis)*/
static int nsw_rcount; /* free read buffers */
static int nsw_wcount_sync; /* limit write buffers / synchronous */
static int nsw_wcount_async; /* limit write buffers / asynchronous */
static int nsw_wcount_async_max;/* assigned maximum */
static int nsw_cluster_max; /* maximum VOP I/O allowed */
static struct swblock **swhash;
static int swhash_mask;
static struct mtx swhash_mtx;
static int swap_async_max = 4; /* maximum in-progress async I/O's */
static struct sx sw_alloc_sx;
SYSCTL_INT(_vm, OID_AUTO, swap_async_max,
CTLFLAG_RW, &swap_async_max, 0, "Maximum running async swap ops");
/*
* "named" and "unnamed" anon region objects. Try to reduce the overhead
* of searching a named list by hashing it just a little.
*/
#define NOBJLISTS 8
#define NOBJLIST(handle) \
(&swap_pager_object_list[((int)(intptr_t)handle >> 4) & (NOBJLISTS-1)])
static struct mtx sw_alloc_mtx; /* protect list manipulation */
static struct pagerlst swap_pager_object_list[NOBJLISTS];
static uma_zone_t swap_zone;
static struct vm_object swap_zone_obj;
/*
* pagerops for OBJT_SWAP - "swap pager". Some ops are also global procedure
* calls hooked from other parts of the VM system and do not appear here.
* (see vm/swap_pager.h).
*/
static vm_object_t
swap_pager_alloc(void *handle, vm_ooffset_t size,
vm_prot_t prot, vm_ooffset_t offset, struct ucred *);
static void swap_pager_dealloc(vm_object_t object);
static int swap_pager_getpages(vm_object_t, vm_page_t *, int, int);
static void swap_pager_putpages(vm_object_t, vm_page_t *, int, boolean_t, int *);
static boolean_t
swap_pager_haspage(vm_object_t object, vm_pindex_t pindex, int *before, int *after);
static void swap_pager_init(void);
static void swap_pager_unswapped(vm_page_t);
static void swap_pager_swapoff(struct swdevt *sp);
struct pagerops swappagerops = {
.pgo_init = swap_pager_init, /* early system initialization of pager */
.pgo_alloc = swap_pager_alloc, /* allocate an OBJT_SWAP object */
.pgo_dealloc = swap_pager_dealloc, /* deallocate an OBJT_SWAP object */
.pgo_getpages = swap_pager_getpages, /* pagein */
.pgo_putpages = swap_pager_putpages, /* pageout */
.pgo_haspage = swap_pager_haspage, /* get backing store status for page */
.pgo_pageunswapped = swap_pager_unswapped, /* remove swap related to page */
};
/*
* dmmax is in page-sized chunks with the new swap system. It was
* dev-bsized chunks in the old. dmmax is always a power of 2.
*
* swap_*() routines are externally accessible. swp_*() routines are
* internal.
*/
static int dmmax;
static int nswap_lowat = 128; /* in pages, swap_pager_almost_full warn */
static int nswap_hiwat = 512; /* in pages, swap_pager_almost_full warn */
SYSCTL_INT(_vm, OID_AUTO, dmmax,
CTLFLAG_RD, &dmmax, 0, "Maximum size of a swap block");
static void swp_sizecheck(void);
static void swp_pager_async_iodone(struct buf *bp);
static int swapongeom(struct thread *, struct vnode *);
static int swaponvp(struct thread *, struct vnode *, u_long);
static int swapoff_one(struct swdevt *sp, struct ucred *cred);
/*
* Swap bitmap functions
*/
static void swp_pager_freeswapspace(daddr_t blk, int npages);
static daddr_t swp_pager_getswapspace(int npages);
/*
* Metadata functions
*/
static struct swblock **swp_pager_hash(vm_object_t object, vm_pindex_t index);
static void swp_pager_meta_build(vm_object_t, vm_pindex_t, daddr_t);
static void swp_pager_meta_free(vm_object_t, vm_pindex_t, daddr_t);
static void swp_pager_meta_free_all(vm_object_t);
static daddr_t swp_pager_meta_ctl(vm_object_t, vm_pindex_t, int);
/*
* SWP_SIZECHECK() - update swap_pager_full indication
*
* update the swap_pager_almost_full indication and warn when we are
* about to run out of swap space, using lowat/hiwat hysteresis.
*
* Clear swap_pager_full ( task killing ) indication when lowat is met.
*
* No restrictions on call
* This routine may not block.
* This routine must be called at splvm()
*/
static void
swp_sizecheck(void)
{
if (swap_pager_avail < nswap_lowat) {
if (swap_pager_almost_full == 0) {
printf("swap_pager: out of swap space\n");
swap_pager_almost_full = 1;
}
} else {
swap_pager_full = 0;
if (swap_pager_avail > nswap_hiwat)
swap_pager_almost_full = 0;
}
}
/*
* SWP_PAGER_HASH() - hash swap meta data
*
* This is an helper function which hashes the swapblk given
* the object and page index. It returns a pointer to a pointer
* to the object, or a pointer to a NULL pointer if it could not
* find a swapblk.
*
* This routine must be called at splvm().
*/
static struct swblock **
swp_pager_hash(vm_object_t object, vm_pindex_t index)
{
struct swblock **pswap;
struct swblock *swap;
index &= ~(vm_pindex_t)SWAP_META_MASK;
pswap = &swhash[(index ^ (int)(intptr_t)object) & swhash_mask];
while ((swap = *pswap) != NULL) {
if (swap->swb_object == object &&
swap->swb_index == index
) {
break;
}
pswap = &swap->swb_hnext;
}
return (pswap);
}
/*
* SWAP_PAGER_INIT() - initialize the swap pager!
*
* Expected to be started from system init. NOTE: This code is run
* before much else so be careful what you depend on. Most of the VM
* system has yet to be initialized at this point.
*/
static void
swap_pager_init(void)
{
/*
* Initialize object lists
*/
int i;
for (i = 0; i < NOBJLISTS; ++i)
TAILQ_INIT(&swap_pager_object_list[i]);
mtx_init(&sw_alloc_mtx, "swap_pager list", NULL, MTX_DEF);
mtx_init(&sw_dev_mtx, "swapdev", NULL, MTX_DEF);
/*
* Device Stripe, in PAGE_SIZE'd blocks
*/
dmmax = SWB_NPAGES * 2;
}
/*
* SWAP_PAGER_SWAP_INIT() - swap pager initialization from pageout process
*
* Expected to be started from pageout process once, prior to entering
* its main loop.
*/
void
swap_pager_swap_init(void)
{
int n, n2;
/*
* Number of in-transit swap bp operations. Don't
* exhaust the pbufs completely. Make sure we
* initialize workable values (0 will work for hysteresis
* but it isn't very efficient).
*
* The nsw_cluster_max is constrained by the bp->b_pages[]
* array (MAXPHYS/PAGE_SIZE) and our locally defined
* MAX_PAGEOUT_CLUSTER. Also be aware that swap ops are
* constrained by the swap device interleave stripe size.
*
* Currently we hardwire nsw_wcount_async to 4. This limit is
* designed to prevent other I/O from having high latencies due to
* our pageout I/O. The value 4 works well for one or two active swap
* devices but is probably a little low if you have more. Even so,
* a higher value would probably generate only a limited improvement
* with three or four active swap devices since the system does not
* typically have to pageout at extreme bandwidths. We will want
* at least 2 per swap devices, and 4 is a pretty good value if you
* have one NFS swap device due to the command/ack latency over NFS.
* So it all works out pretty well.
*/
nsw_cluster_max = min((MAXPHYS/PAGE_SIZE), MAX_PAGEOUT_CLUSTER);
mtx_lock(&pbuf_mtx);
nsw_rcount = (nswbuf + 1) / 2;
nsw_wcount_sync = (nswbuf + 3) / 4;
nsw_wcount_async = 4;
nsw_wcount_async_max = nsw_wcount_async;
mtx_unlock(&pbuf_mtx);
/*
* Initialize our zone. Right now I'm just guessing on the number
* we need based on the number of pages in the system. Each swblock
* can hold 16 pages, so this is probably overkill. This reservation
* is typically limited to around 32MB by default.
*/
n = cnt.v_page_count / 2;
if (maxswzone && n > maxswzone / sizeof(struct swblock))
n = maxswzone / sizeof(struct swblock);
n2 = n;
swap_zone = uma_zcreate("SWAPMETA", sizeof(struct swblock), NULL, NULL,
NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_NOFREE | UMA_ZONE_VM);
if (swap_zone == NULL)
panic("failed to create swap_zone.");
do {
if (uma_zone_set_obj(swap_zone, &swap_zone_obj, n))
break;
/*
* if the allocation failed, try a zone two thirds the
* size of the previous attempt.
*/
n -= ((n + 2) / 3);
} while (n > 0);
if (n2 != n)
printf("Swap zone entries reduced from %d to %d.\n", n2, n);
n2 = n;
/*
* Initialize our meta-data hash table. The swapper does not need to
* be quite as efficient as the VM system, so we do not use an
* oversized hash table.
*
* n: size of hash table, must be power of 2
* swhash_mask: hash table index mask
*/
for (n = 1; n < n2 / 8; n *= 2)
;
swhash = malloc(sizeof(struct swblock *) * n, M_VMPGDATA, M_WAITOK | M_ZERO);
swhash_mask = n - 1;
mtx_init(&swhash_mtx, "swap_pager swhash", NULL, MTX_DEF);
}
/*
* SWAP_PAGER_ALLOC() - allocate a new OBJT_SWAP VM object and instantiate
* its metadata structures.
*
* This routine is called from the mmap and fork code to create a new
* OBJT_SWAP object. We do this by creating an OBJT_DEFAULT object
* and then converting it with swp_pager_meta_build().
*
* This routine may block in vm_object_allocate() and create a named
* object lookup race, so we must interlock. We must also run at
* splvm() for the object lookup to handle races with interrupts, but
* we do not have to maintain splvm() in between the lookup and the
* add because (I believe) it is not possible to attempt to create
* a new swap object w/handle when a default object with that handle
* already exists.
*
* MPSAFE
*/
static vm_object_t
swap_pager_alloc(void *handle, vm_ooffset_t size, vm_prot_t prot,
vm_ooffset_t offset, struct ucred *cred)
{
vm_object_t object;
vm_pindex_t pindex;
struct uidinfo *uip;
uip = NULL;
pindex = OFF_TO_IDX(offset + PAGE_MASK + size);
if (handle) {
mtx_lock(&Giant);
/*
* Reference existing named region or allocate new one. There
* should not be a race here against swp_pager_meta_build()
* as called from vm_page_remove() in regards to the lookup
* of the handle.
*/
sx_xlock(&sw_alloc_sx);
object = vm_pager_object_lookup(NOBJLIST(handle), handle);
if (object == NULL) {
if (cred != NULL) {
uip = cred->cr_ruidinfo;
if (!swap_reserve_by_uid(size, uip)) {
sx_xunlock(&sw_alloc_sx);
mtx_unlock(&Giant);
return (NULL);
}
uihold(uip);
}
object = vm_object_allocate(OBJT_DEFAULT, pindex);
VM_OBJECT_LOCK(object);
object->handle = handle;
if (cred != NULL) {
object->uip = uip;
object->charge = size;
}
swp_pager_meta_build(object, 0, SWAPBLK_NONE);
VM_OBJECT_UNLOCK(object);
}
sx_xunlock(&sw_alloc_sx);
mtx_unlock(&Giant);
} else {
if (cred != NULL) {
uip = cred->cr_ruidinfo;
if (!swap_reserve_by_uid(size, uip))
return (NULL);
uihold(uip);
}
object = vm_object_allocate(OBJT_DEFAULT, pindex);
VM_OBJECT_LOCK(object);
if (cred != NULL) {
object->uip = uip;
object->charge = size;
}
swp_pager_meta_build(object, 0, SWAPBLK_NONE);
VM_OBJECT_UNLOCK(object);
}
return (object);
}
/*
* SWAP_PAGER_DEALLOC() - remove swap metadata from object
*
* The swap backing for the object is destroyed. The code is
* designed such that we can reinstantiate it later, but this
* routine is typically called only when the entire object is
* about to be destroyed.
*
* This routine may block, but no longer does.
*
* The object must be locked or unreferenceable.
*/
static void
swap_pager_dealloc(vm_object_t object)
{
/*
* Remove from list right away so lookups will fail if we block for
* pageout completion.
*/
if (object->handle != NULL) {
mtx_lock(&sw_alloc_mtx);
TAILQ_REMOVE(NOBJLIST(object->handle), object, pager_object_list);
mtx_unlock(&sw_alloc_mtx);
}
VM_OBJECT_LOCK_ASSERT(object, MA_OWNED);
vm_object_pip_wait(object, "swpdea");
/*
* Free all remaining metadata. We only bother to free it from
* the swap meta data. We do not attempt to free swapblk's still
* associated with vm_page_t's for this object. We do not care
* if paging is still in progress on some objects.
*/
swp_pager_meta_free_all(object);
}
/************************************************************************
* SWAP PAGER BITMAP ROUTINES *
************************************************************************/
/*
* SWP_PAGER_GETSWAPSPACE() - allocate raw swap space
*
* Allocate swap for the requested number of pages. The starting
* swap block number (a page index) is returned or SWAPBLK_NONE
* if the allocation failed.
*
* Also has the side effect of advising that somebody made a mistake
* when they configured swap and didn't configure enough.
*
* Must be called at splvm() to avoid races with bitmap frees from
* vm_page_remove() aka swap_pager_page_removed().
*
* This routine may not block
* This routine must be called at splvm().
*
* We allocate in round-robin fashion from the configured devices.
*/
static daddr_t
swp_pager_getswapspace(int npages)
{
daddr_t blk;
struct swdevt *sp;
int i;
blk = SWAPBLK_NONE;
mtx_lock(&sw_dev_mtx);
sp = swdevhd;
for (i = 0; i < nswapdev; i++) {
if (sp == NULL)
sp = TAILQ_FIRST(&swtailq);
if (!(sp->sw_flags & SW_CLOSING)) {
blk = blist_alloc(sp->sw_blist, npages);
if (blk != SWAPBLK_NONE) {
blk += sp->sw_first;
sp->sw_used += npages;
swap_pager_avail -= npages;
swp_sizecheck();
swdevhd = TAILQ_NEXT(sp, sw_list);
goto done;
}
}
sp = TAILQ_NEXT(sp, sw_list);
}
if (swap_pager_full != 2) {
printf("swap_pager_getswapspace(%d): failed\n", npages);
swap_pager_full = 2;
swap_pager_almost_full = 1;
}
swdevhd = NULL;
done:
mtx_unlock(&sw_dev_mtx);
return (blk);
}
static int
swp_pager_isondev(daddr_t blk, struct swdevt *sp)
{
return (blk >= sp->sw_first && blk < sp->sw_end);
}
static void
swp_pager_strategy(struct buf *bp)
{
struct swdevt *sp;
mtx_lock(&sw_dev_mtx);
TAILQ_FOREACH(sp, &swtailq, sw_list) {
if (bp->b_blkno >= sp->sw_first && bp->b_blkno < sp->sw_end) {
mtx_unlock(&sw_dev_mtx);
sp->sw_strategy(bp, sp);
return;
}
}
panic("Swapdev not found");
}
/*
* SWP_PAGER_FREESWAPSPACE() - free raw swap space
*
* This routine returns the specified swap blocks back to the bitmap.
*
* Note: This routine may not block (it could in the old swap code),
* and through the use of the new blist routines it does not block.
*
* We must be called at splvm() to avoid races with bitmap frees from
* vm_page_remove() aka swap_pager_page_removed().
*
* This routine may not block
* This routine must be called at splvm().
*/
static void
swp_pager_freeswapspace(daddr_t blk, int npages)
{
struct swdevt *sp;
mtx_lock(&sw_dev_mtx);
TAILQ_FOREACH(sp, &swtailq, sw_list) {
if (blk >= sp->sw_first && blk < sp->sw_end) {
sp->sw_used -= npages;
/*
* If we are attempting to stop swapping on
* this device, we don't want to mark any
* blocks free lest they be reused.
*/
if ((sp->sw_flags & SW_CLOSING) == 0) {
blist_free(sp->sw_blist, blk - sp->sw_first,
npages);
swap_pager_avail += npages;
swp_sizecheck();
}
mtx_unlock(&sw_dev_mtx);
return;
}
}
panic("Swapdev not found");
}
/*
* SWAP_PAGER_FREESPACE() - frees swap blocks associated with a page
* range within an object.
*
* This is a globally accessible routine.
*
* This routine removes swapblk assignments from swap metadata.
*
* The external callers of this routine typically have already destroyed
* or renamed vm_page_t's associated with this range in the object so
* we should be ok.
*
* This routine may be called at any spl. We up our spl to splvm temporarily
* in order to perform the metadata removal.
*/
void
swap_pager_freespace(vm_object_t object, vm_pindex_t start, vm_size_t size)
{
VM_OBJECT_LOCK_ASSERT(object, MA_OWNED);
swp_pager_meta_free(object, start, size);
}
/*
* SWAP_PAGER_RESERVE() - reserve swap blocks in object
*
* Assigns swap blocks to the specified range within the object. The
* swap blocks are not zerod. Any previous swap assignment is destroyed.
*
* Returns 0 on success, -1 on failure.
*/
int
swap_pager_reserve(vm_object_t object, vm_pindex_t start, vm_size_t size)
{
int n = 0;
daddr_t blk = SWAPBLK_NONE;
vm_pindex_t beg = start; /* save start index */
VM_OBJECT_LOCK(object);
while (size) {
if (n == 0) {
n = BLIST_MAX_ALLOC;
while ((blk = swp_pager_getswapspace(n)) == SWAPBLK_NONE) {
n >>= 1;
if (n == 0) {
swp_pager_meta_free(object, beg, start - beg);
VM_OBJECT_UNLOCK(object);
return (-1);
}
}
}
swp_pager_meta_build(object, start, blk);
--size;
++start;
++blk;
--n;
}
swp_pager_meta_free(object, start, n);
VM_OBJECT_UNLOCK(object);
return (0);
}
/*
* SWAP_PAGER_COPY() - copy blocks from source pager to destination pager
* and destroy the source.
*
* Copy any valid swapblks from the source to the destination. In
* cases where both the source and destination have a valid swapblk,
* we keep the destination's.
*
* This routine is allowed to block. It may block allocating metadata
* indirectly through swp_pager_meta_build() or if paging is still in
* progress on the source.
*
* This routine can be called at any spl
*
* XXX vm_page_collapse() kinda expects us not to block because we
* supposedly do not need to allocate memory, but for the moment we
* *may* have to get a little memory from the zone allocator, but
* it is taken from the interrupt memory. We should be ok.
*
* The source object contains no vm_page_t's (which is just as well)
*
* The source object is of type OBJT_SWAP.
*
* The source and destination objects must be locked or
* inaccessible (XXX are they ?)
*/
void
swap_pager_copy(vm_object_t srcobject, vm_object_t dstobject,
vm_pindex_t offset, int destroysource)
{
vm_pindex_t i;
VM_OBJECT_LOCK_ASSERT(srcobject, MA_OWNED);
VM_OBJECT_LOCK_ASSERT(dstobject, MA_OWNED);
/*
* If destroysource is set, we remove the source object from the
* swap_pager internal queue now.
*/
if (destroysource) {
if (srcobject->handle != NULL) {
mtx_lock(&sw_alloc_mtx);
TAILQ_REMOVE(
NOBJLIST(srcobject->handle),
srcobject,
pager_object_list
);
mtx_unlock(&sw_alloc_mtx);
}
}
/*
* transfer source to destination.
*/
for (i = 0; i < dstobject->size; ++i) {
daddr_t dstaddr;
/*
* Locate (without changing) the swapblk on the destination,
* unless it is invalid in which case free it silently, or
* if the destination is a resident page, in which case the
* source is thrown away.
*/
dstaddr = swp_pager_meta_ctl(dstobject, i, 0);
if (dstaddr == SWAPBLK_NONE) {
/*
* Destination has no swapblk and is not resident,
* copy source.
*/
daddr_t srcaddr;
srcaddr = swp_pager_meta_ctl(
srcobject,
i + offset,
SWM_POP
);
if (srcaddr != SWAPBLK_NONE) {
/*
* swp_pager_meta_build() can sleep.
*/
vm_object_pip_add(srcobject, 1);
VM_OBJECT_UNLOCK(srcobject);
vm_object_pip_add(dstobject, 1);
swp_pager_meta_build(dstobject, i, srcaddr);
vm_object_pip_wakeup(dstobject);
VM_OBJECT_LOCK(srcobject);
vm_object_pip_wakeup(srcobject);
}
} else {
/*
* Destination has valid swapblk or it is represented
* by a resident page. We destroy the sourceblock.
*/
swp_pager_meta_ctl(srcobject, i + offset, SWM_FREE);
}
}
/*
* Free left over swap blocks in source.
*
* We have to revert the type to OBJT_DEFAULT so we do not accidently
* double-remove the object from the swap queues.
*/
if (destroysource) {
swp_pager_meta_free_all(srcobject);
/*
* Reverting the type is not necessary, the caller is going
* to destroy srcobject directly, but I'm doing it here
* for consistency since we've removed the object from its
* queues.
*/
srcobject->type = OBJT_DEFAULT;
}
}
/*
* SWAP_PAGER_HASPAGE() - determine if we have good backing store for
* the requested page.
*
* We determine whether good backing store exists for the requested
* page and return TRUE if it does, FALSE if it doesn't.
*
* If TRUE, we also try to determine how much valid, contiguous backing
* store exists before and after the requested page within a reasonable
* distance. We do not try to restrict it to the swap device stripe
* (that is handled in getpages/putpages). It probably isn't worth
* doing here.
*/
static boolean_t
swap_pager_haspage(vm_object_t object, vm_pindex_t pindex, int *before, int *after)
{
daddr_t blk0;
VM_OBJECT_LOCK_ASSERT(object, MA_OWNED);
/*
* do we have good backing store at the requested index ?
*/
blk0 = swp_pager_meta_ctl(object, pindex, 0);
if (blk0 == SWAPBLK_NONE) {
if (before)
*before = 0;
if (after)
*after = 0;
return (FALSE);
}
/*
* find backwards-looking contiguous good backing store
*/
if (before != NULL) {
int i;
for (i = 1; i < (SWB_NPAGES/2); ++i) {
daddr_t blk;
if (i > pindex)
break;
blk = swp_pager_meta_ctl(object, pindex - i, 0);
if (blk != blk0 - i)
break;
}
*before = (i - 1);
}
/*
* find forward-looking contiguous good backing store
*/
if (after != NULL) {
int i;
for (i = 1; i < (SWB_NPAGES/2); ++i) {
daddr_t blk;
blk = swp_pager_meta_ctl(object, pindex + i, 0);
if (blk != blk0 + i)
break;
}
*after = (i - 1);
}
return (TRUE);
}
/*
* SWAP_PAGER_PAGE_UNSWAPPED() - remove swap backing store related to page
*
* This removes any associated swap backing store, whether valid or
* not, from the page.
*
* This routine is typically called when a page is made dirty, at
* which point any associated swap can be freed. MADV_FREE also
* calls us in a special-case situation
*
* NOTE!!! If the page is clean and the swap was valid, the caller
* should make the page dirty before calling this routine. This routine
* does NOT change the m->dirty status of the page. Also: MADV_FREE
* depends on it.
*
* This routine may not block
* This routine must be called at splvm()
*/
static void
swap_pager_unswapped(vm_page_t m)
{
VM_OBJECT_LOCK_ASSERT(m->object, MA_OWNED);
swp_pager_meta_ctl(m->object, m->pindex, SWM_FREE);
}
/*
* SWAP_PAGER_GETPAGES() - bring pages in from swap
*
* Attempt to retrieve (m, count) pages from backing store, but make
* sure we retrieve at least m[reqpage]. We try to load in as large
* a chunk surrounding m[reqpage] as is contiguous in swap and which
* belongs to the same object.
*
* The code is designed for asynchronous operation and
* immediate-notification of 'reqpage' but tends not to be
* used that way. Please do not optimize-out this algorithmic
* feature, I intend to improve on it in the future.
*
* The parent has a single vm_object_pip_add() reference prior to
* calling us and we should return with the same.
*
* The parent has BUSY'd the pages. We should return with 'm'
* left busy, but the others adjusted.
*/
static int
swap_pager_getpages(vm_object_t object, vm_page_t *m, int count, int reqpage)
{
struct buf *bp;
vm_page_t mreq;
int i;
int j;
daddr_t blk;
mreq = m[reqpage];
KASSERT(mreq->object == object,
("swap_pager_getpages: object mismatch %p/%p",
object, mreq->object));
/*
* Calculate range to retrieve. The pages have already been assigned
* their swapblks. We require a *contiguous* range but we know it to
* not span devices. If we do not supply it, bad things
* happen. Note that blk, iblk & jblk can be SWAPBLK_NONE, but the
* loops are set up such that the case(s) are handled implicitly.
*
* The swp_*() calls must be made at splvm(). vm_page_free() does
* not need to be, but it will go a little faster if it is.
*/
blk = swp_pager_meta_ctl(mreq->object, mreq->pindex, 0);
for (i = reqpage - 1; i >= 0; --i) {
daddr_t iblk;
iblk = swp_pager_meta_ctl(m[i]->object, m[i]->pindex, 0);
if (blk != iblk + (reqpage - i))
break;
}
++i;
for (j = reqpage + 1; j < count; ++j) {
daddr_t jblk;
jblk = swp_pager_meta_ctl(m[j]->object, m[j]->pindex, 0);
if (blk != jblk - (j - reqpage))
break;
}
/*
* free pages outside our collection range. Note: we never free
* mreq, it must remain busy throughout.
*/
if (0 < i || j < count) {
int k;
vm_page_lock_queues();
for (k = 0; k < i; ++k)
vm_page_free(m[k]);
for (k = j; k < count; ++k)
vm_page_free(m[k]);
vm_page_unlock_queues();
}
/*
* Return VM_PAGER_FAIL if we have nothing to do. Return mreq
* still busy, but the others unbusied.
*/
if (blk == SWAPBLK_NONE)
return (VM_PAGER_FAIL);
/*
* Getpbuf() can sleep.
*/
VM_OBJECT_UNLOCK(object);
/*
* Get a swap buffer header to perform the IO
*/
bp = getpbuf(&nsw_rcount);
bp->b_flags |= B_PAGING;
/*
* map our page(s) into kva for input
*/
pmap_qenter((vm_offset_t)bp->b_data, m + i, j - i);
bp->b_iocmd = BIO_READ;
bp->b_iodone = swp_pager_async_iodone;
bp->b_rcred = crhold(thread0.td_ucred);
bp->b_wcred = crhold(thread0.td_ucred);
bp->b_blkno = blk - (reqpage - i);
bp->b_bcount = PAGE_SIZE * (j - i);
bp->b_bufsize = PAGE_SIZE * (j - i);
bp->b_pager.pg_reqpage = reqpage - i;
VM_OBJECT_LOCK(object);
{
int k;
for (k = i; k < j; ++k) {
bp->b_pages[k - i] = m[k];
m[k]->oflags |= VPO_SWAPINPROG;
}
}
bp->b_npages = j - i;
PCPU_INC(cnt.v_swapin);
PCPU_ADD(cnt.v_swappgsin, bp->b_npages);
/*
* We still hold the lock on mreq, and our automatic completion routine
* does not remove it.
*/
vm_object_pip_add(object, bp->b_npages);
VM_OBJECT_UNLOCK(object);
/*
* perform the I/O. NOTE!!! bp cannot be considered valid after
* this point because we automatically release it on completion.
* Instead, we look at the one page we are interested in which we
* still hold a lock on even through the I/O completion.
*
* The other pages in our m[] array are also released on completion,
* so we cannot assume they are valid anymore either.
*
* NOTE: b_blkno is destroyed by the call to swapdev_strategy
*/
BUF_KERNPROC(bp);
swp_pager_strategy(bp);
/*
* wait for the page we want to complete. VPO_SWAPINPROG is always
* cleared on completion. If an I/O error occurs, SWAPBLK_NONE
* is set in the meta-data.
*/
VM_OBJECT_LOCK(object);
while ((mreq->oflags & VPO_SWAPINPROG) != 0) {
mreq->oflags |= VPO_WANTED;
vm_page_lock_queues();
vm_page_flag_set(mreq, PG_REFERENCED);
vm_page_unlock_queues();
PCPU_INC(cnt.v_intrans);
if (msleep(mreq, VM_OBJECT_MTX(object), PSWP, "swread", hz*20)) {
printf(
"swap_pager: indefinite wait buffer: bufobj: %p, blkno: %jd, size: %ld\n",
bp->b_bufobj, (intmax_t)bp->b_blkno, bp->b_bcount);
}
}
/*
* mreq is left busied after completion, but all the other pages
* are freed. If we had an unrecoverable read error the page will
* not be valid.
*/
if (mreq->valid != VM_PAGE_BITS_ALL) {
return (VM_PAGER_ERROR);
} else {
return (VM_PAGER_OK);
}
/*
* A final note: in a low swap situation, we cannot deallocate swap
* and mark a page dirty here because the caller is likely to mark
* the page clean when we return, causing the page to possibly revert
* to all-zero's later.
*/
}
/*
* swap_pager_putpages:
*
* Assign swap (if necessary) and initiate I/O on the specified pages.
*
* We support both OBJT_DEFAULT and OBJT_SWAP objects. DEFAULT objects
* are automatically converted to SWAP objects.
*
* In a low memory situation we may block in VOP_STRATEGY(), but the new
* vm_page reservation system coupled with properly written VFS devices
* should ensure that no low-memory deadlock occurs. This is an area
* which needs work.
*
* The parent has N vm_object_pip_add() references prior to
* calling us and will remove references for rtvals[] that are
* not set to VM_PAGER_PEND. We need to remove the rest on I/O
* completion.
*
* The parent has soft-busy'd the pages it passes us and will unbusy
* those whos rtvals[] entry is not set to VM_PAGER_PEND on return.
* We need to unbusy the rest on I/O completion.
*/
void
swap_pager_putpages(vm_object_t object, vm_page_t *m, int count,
boolean_t sync, int *rtvals)
{
int i;
int n = 0;
if (count && m[0]->object != object) {
panic("swap_pager_putpages: object mismatch %p/%p",
object,
m[0]->object
);
}
/*
* Step 1
*
* Turn object into OBJT_SWAP
* check for bogus sysops
* force sync if not pageout process
*/
if (object->type != OBJT_SWAP)
swp_pager_meta_build(object, 0, SWAPBLK_NONE);
VM_OBJECT_UNLOCK(object);
if (curproc != pageproc)
sync = TRUE;
/*
* Step 2
*
* Update nsw parameters from swap_async_max sysctl values.
* Do not let the sysop crash the machine with bogus numbers.
*/
mtx_lock(&pbuf_mtx);
if (swap_async_max != nsw_wcount_async_max) {
int n;
/*
* limit range
*/
if ((n = swap_async_max) > nswbuf / 2)
n = nswbuf / 2;
if (n < 1)
n = 1;
swap_async_max = n;
/*
* Adjust difference ( if possible ). If the current async
* count is too low, we may not be able to make the adjustment
* at this time.
*/
n -= nsw_wcount_async_max;
if (nsw_wcount_async + n >= 0) {
nsw_wcount_async += n;
nsw_wcount_async_max += n;
wakeup(&nsw_wcount_async);
}
}
mtx_unlock(&pbuf_mtx);
/*
* Step 3
*
* Assign swap blocks and issue I/O. We reallocate swap on the fly.
* The page is left dirty until the pageout operation completes
* successfully.
*/
for (i = 0; i < count; i += n) {
int j;
struct buf *bp;
daddr_t blk;
/*
* Maximum I/O size is limited by a number of factors.
*/
n = min(BLIST_MAX_ALLOC, count - i);
n = min(n, nsw_cluster_max);
/*
* Get biggest block of swap we can. If we fail, fall
* back and try to allocate a smaller block. Don't go
* overboard trying to allocate space if it would overly
* fragment swap.
*/
while (
(blk = swp_pager_getswapspace(n)) == SWAPBLK_NONE &&
n > 4
) {
n >>= 1;
}
if (blk == SWAPBLK_NONE) {
for (j = 0; j < n; ++j)
rtvals[i+j] = VM_PAGER_FAIL;
continue;
}
/*
* All I/O parameters have been satisfied, build the I/O
* request and assign the swap space.
*/
if (sync == TRUE) {
bp = getpbuf(&nsw_wcount_sync);
} else {
bp = getpbuf(&nsw_wcount_async);
bp->b_flags = B_ASYNC;
}
bp->b_flags |= B_PAGING;
bp->b_iocmd = BIO_WRITE;
pmap_qenter((vm_offset_t)bp->b_data, &m[i], n);
bp->b_rcred = crhold(thread0.td_ucred);
bp->b_wcred = crhold(thread0.td_ucred);
bp->b_bcount = PAGE_SIZE * n;
bp->b_bufsize = PAGE_SIZE * n;
bp->b_blkno = blk;
VM_OBJECT_LOCK(object);
for (j = 0; j < n; ++j) {
vm_page_t mreq = m[i+j];
swp_pager_meta_build(
mreq->object,
mreq->pindex,
blk + j
);
vm_page_dirty(mreq);
rtvals[i+j] = VM_PAGER_OK;
mreq->oflags |= VPO_SWAPINPROG;
bp->b_pages[j] = mreq;
}
VM_OBJECT_UNLOCK(object);
bp->b_npages = n;
/*
* Must set dirty range for NFS to work.
*/
bp->b_dirtyoff = 0;
bp->b_dirtyend = bp->b_bcount;
PCPU_INC(cnt.v_swapout);
PCPU_ADD(cnt.v_swappgsout, bp->b_npages);
/*
* asynchronous
*
* NOTE: b_blkno is destroyed by the call to swapdev_strategy
*/
if (sync == FALSE) {
bp->b_iodone = swp_pager_async_iodone;
BUF_KERNPROC(bp);
swp_pager_strategy(bp);
for (j = 0; j < n; ++j)
rtvals[i+j] = VM_PAGER_PEND;
/* restart outter loop */
continue;
}
/*
* synchronous
*
* NOTE: b_blkno is destroyed by the call to swapdev_strategy
*/
bp->b_iodone = bdone;
swp_pager_strategy(bp);
/*
* Wait for the sync I/O to complete, then update rtvals.
* We just set the rtvals[] to VM_PAGER_PEND so we can call
* our async completion routine at the end, thus avoiding a
* double-free.
*/
bwait(bp, PVM, "swwrt");
for (j = 0; j < n; ++j)
rtvals[i+j] = VM_PAGER_PEND;
/*
* Now that we are through with the bp, we can call the
* normal async completion, which frees everything up.
*/
swp_pager_async_iodone(bp);
}
VM_OBJECT_LOCK(object);
}
/*
* swp_pager_async_iodone:
*
* Completion routine for asynchronous reads and writes from/to swap.
* Also called manually by synchronous code to finish up a bp.
*
* For READ operations, the pages are PG_BUSY'd. For WRITE operations,
* the pages are vm_page_t->busy'd. For READ operations, we PG_BUSY
* unbusy all pages except the 'main' request page. For WRITE
* operations, we vm_page_t->busy'd unbusy all pages ( we can do this
* because we marked them all VM_PAGER_PEND on return from putpages ).
*
* This routine may not block.
*/
static void
swp_pager_async_iodone(struct buf *bp)
{
int i;
vm_object_t object = NULL;
/*
* report error
*/
if (bp->b_ioflags & BIO_ERROR) {
printf(
"swap_pager: I/O error - %s failed; blkno %ld,"
"size %ld, error %d\n",
((bp->b_iocmd == BIO_READ) ? "pagein" : "pageout"),
(long)bp->b_blkno,
(long)bp->b_bcount,
bp->b_error
);
}
/*
* remove the mapping for kernel virtual
*/
pmap_qremove((vm_offset_t)bp->b_data, bp->b_npages);
if (bp->b_npages) {
object = bp->b_pages[0]->object;
VM_OBJECT_LOCK(object);
}
vm_page_lock_queues();
/*
* cleanup pages. If an error occurs writing to swap, we are in
* very serious trouble. If it happens to be a disk error, though,
* we may be able to recover by reassigning the swap later on. So
* in this case we remove the m->swapblk assignment for the page
* but do not free it in the rlist. The errornous block(s) are thus
* never reallocated as swap. Redirty the page and continue.
*/
for (i = 0; i < bp->b_npages; ++i) {
vm_page_t m = bp->b_pages[i];
m->oflags &= ~VPO_SWAPINPROG;
if (bp->b_ioflags & BIO_ERROR) {
/*
* If an error occurs I'd love to throw the swapblk
* away without freeing it back to swapspace, so it
* can never be used again. But I can't from an
* interrupt.
*/
if (bp->b_iocmd == BIO_READ) {
/*
* When reading, reqpage needs to stay
* locked for the parent, but all other
* pages can be freed. We still want to
* wakeup the parent waiting on the page,
* though. ( also: pg_reqpage can be -1 and
* not match anything ).
*
* We have to wake specifically requested pages
* up too because we cleared VPO_SWAPINPROG and
* someone may be waiting for that.
*
* NOTE: for reads, m->dirty will probably
* be overridden by the original caller of
* getpages so don't play cute tricks here.
*/
m->valid = 0;
if (i != bp->b_pager.pg_reqpage)
vm_page_free(m);
else
vm_page_flash(m);
/*
* If i == bp->b_pager.pg_reqpage, do not wake
* the page up. The caller needs to.
*/
} else {
/*
* If a write error occurs, reactivate page
* so it doesn't clog the inactive list,
* then finish the I/O.
*/
vm_page_dirty(m);
vm_page_activate(m);
vm_page_io_finish(m);
}
} else if (bp->b_iocmd == BIO_READ) {
/*
* NOTE: for reads, m->dirty will probably be
* overridden by the original caller of getpages so
* we cannot set them in order to free the underlying
* swap in a low-swap situation. I don't think we'd
* want to do that anyway, but it was an optimization
* that existed in the old swapper for a time before
* it got ripped out due to precisely this problem.
*
* If not the requested page then deactivate it.
*
* Note that the requested page, reqpage, is left
* busied, but we still have to wake it up. The
* other pages are released (unbusied) by
* vm_page_wakeup().
*/
KASSERT(!pmap_page_is_mapped(m),
("swp_pager_async_iodone: page %p is mapped", m));
m->valid = VM_PAGE_BITS_ALL;
KASSERT(m->dirty == 0,
("swp_pager_async_iodone: page %p is dirty", m));
/*
* We have to wake specifically requested pages
* up too because we cleared VPO_SWAPINPROG and
* could be waiting for it in getpages. However,
* be sure to not unbusy getpages specifically
* requested page - getpages expects it to be
* left busy.
*/
if (i != bp->b_pager.pg_reqpage) {
vm_page_deactivate(m);
vm_page_wakeup(m);
} else {
vm_page_flash(m);
}
} else {
/*
* For write success, clear the dirty
* status, then finish the I/O ( which decrements the
* busy count and possibly wakes waiter's up ).
*/
KASSERT((m->flags & PG_WRITEABLE) == 0,
("swp_pager_async_iodone: page %p is not write"
" protected", m));
vm_page_undirty(m);
vm_page_io_finish(m);
if (vm_page_count_severe())
vm_page_try_to_cache(m);
}
}
vm_page_unlock_queues();
/*
* adjust pip. NOTE: the original parent may still have its own
* pip refs on the object.
*/
if (object != NULL) {
vm_object_pip_wakeupn(object, bp->b_npages);
VM_OBJECT_UNLOCK(object);
}
/*
* swapdev_strategy() manually sets b_vp and b_bufobj before calling
* bstrategy(). Set them back to NULL now we're done with it, or we'll
* trigger a KASSERT in relpbuf().
*/
if (bp->b_vp) {
bp->b_vp = NULL;
bp->b_bufobj = NULL;
}
/*
* release the physical I/O buffer
*/
relpbuf(
bp,
((bp->b_iocmd == BIO_READ) ? &nsw_rcount :
((bp->b_flags & B_ASYNC) ?
&nsw_wcount_async :
&nsw_wcount_sync
)
)
);
}
/*
* swap_pager_isswapped:
*
* Return 1 if at least one page in the given object is paged
* out to the given swap device.
*
* This routine may not block.
*/
int
swap_pager_isswapped(vm_object_t object, struct swdevt *sp)
{
daddr_t index = 0;
int bcount;
int i;
VM_OBJECT_LOCK_ASSERT(object, MA_OWNED);
if (object->type != OBJT_SWAP)
return (0);
mtx_lock(&swhash_mtx);
for (bcount = 0; bcount < object->un_pager.swp.swp_bcount; bcount++) {
struct swblock *swap;
if ((swap = *swp_pager_hash(object, index)) != NULL) {
for (i = 0; i < SWAP_META_PAGES; ++i) {
if (swp_pager_isondev(swap->swb_pages[i], sp)) {
mtx_unlock(&swhash_mtx);
return (1);
}
}
}
index += SWAP_META_PAGES;
if (index > 0x20000000)
panic("swap_pager_isswapped: failed to locate all swap meta blocks");
}
mtx_unlock(&swhash_mtx);
return (0);
}
/*
* SWP_PAGER_FORCE_PAGEIN() - force a swap block to be paged in
*
* This routine dissociates the page at the given index within a
* swap block from its backing store, paging it in if necessary.
* If the page is paged in, it is placed in the inactive queue,
* since it had its backing store ripped out from under it.
* We also attempt to swap in all other pages in the swap block,
* we only guarantee that the one at the specified index is
* paged in.
*
* XXX - The code to page the whole block in doesn't work, so we
* revert to the one-by-one behavior for now. Sigh.
*/
static inline void
swp_pager_force_pagein(vm_object_t object, vm_pindex_t pindex)
{
vm_page_t m;
vm_object_pip_add(object, 1);
m = vm_page_grab(object, pindex, VM_ALLOC_NORMAL|VM_ALLOC_RETRY);
if (m->valid == VM_PAGE_BITS_ALL) {
vm_object_pip_subtract(object, 1);
vm_page_lock_queues();
vm_page_activate(m);
vm_page_dirty(m);
vm_page_unlock_queues();
vm_page_wakeup(m);
vm_pager_page_unswapped(m);
return;
}
if (swap_pager_getpages(object, &m, 1, 0) != VM_PAGER_OK)
panic("swap_pager_force_pagein: read from swap failed");/*XXX*/
vm_object_pip_subtract(object, 1);
vm_page_lock_queues();
vm_page_dirty(m);
vm_page_dontneed(m);
vm_page_unlock_queues();
vm_page_wakeup(m);
vm_pager_page_unswapped(m);
}
/*
* swap_pager_swapoff:
*
* Page in all of the pages that have been paged out to the
* given device. The corresponding blocks in the bitmap must be
* marked as allocated and the device must be flagged SW_CLOSING.
* There may be no processes swapped out to the device.
*
* This routine may block.
*/
static void
swap_pager_swapoff(struct swdevt *sp)
{
struct swblock *swap;
int i, j, retries;
GIANT_REQUIRED;
retries = 0;
full_rescan:
mtx_lock(&swhash_mtx);
for (i = 0; i <= swhash_mask; i++) { /* '<=' is correct here */
restart:
for (swap = swhash[i]; swap != NULL; swap = swap->swb_hnext) {
vm_object_t object = swap->swb_object;
vm_pindex_t pindex = swap->swb_index;
for (j = 0; j < SWAP_META_PAGES; ++j) {
if (swp_pager_isondev(swap->swb_pages[j], sp)) {
/* avoid deadlock */
if (!VM_OBJECT_TRYLOCK(object)) {
break;
} else {
mtx_unlock(&swhash_mtx);
swp_pager_force_pagein(object,
pindex + j);
VM_OBJECT_UNLOCK(object);
mtx_lock(&swhash_mtx);
goto restart;
}
}
}
}
}
mtx_unlock(&swhash_mtx);
if (sp->sw_used) {
/*
* Objects may be locked or paging to the device being
* removed, so we will miss their pages and need to
* make another pass. We have marked this device as
* SW_CLOSING, so the activity should finish soon.
*/
retries++;
if (retries > 100) {
panic("swapoff: failed to locate %d swap blocks",
sp->sw_used);
}
pause("swpoff", hz / 20);
goto full_rescan;
}
}
/************************************************************************
* SWAP META DATA *
************************************************************************
*
* These routines manipulate the swap metadata stored in the
* OBJT_SWAP object. All swp_*() routines must be called at
* splvm() because swap can be freed up by the low level vm_page
* code which might be called from interrupts beyond what splbio() covers.
*
* Swap metadata is implemented with a global hash and not directly
* linked into the object. Instead the object simply contains
* appropriate tracking counters.
*/
/*
* SWP_PAGER_META_BUILD() - add swap block to swap meta data for object
*
* We first convert the object to a swap object if it is a default
* object.
*
* The specified swapblk is added to the object's swap metadata. If
* the swapblk is not valid, it is freed instead. Any previously
* assigned swapblk is freed.
*
* This routine must be called at splvm(), except when used to convert
* an OBJT_DEFAULT object into an OBJT_SWAP object.
*/
static void
swp_pager_meta_build(vm_object_t object, vm_pindex_t pindex, daddr_t swapblk)
{
struct swblock *swap;
struct swblock **pswap;
int idx;
VM_OBJECT_LOCK_ASSERT(object, MA_OWNED);
/*
* Convert default object to swap object if necessary
*/
if (object->type != OBJT_SWAP) {
object->type = OBJT_SWAP;
object->un_pager.swp.swp_bcount = 0;
if (object->handle != NULL) {
mtx_lock(&sw_alloc_mtx);
TAILQ_INSERT_TAIL(
NOBJLIST(object->handle),
object,
pager_object_list
);
mtx_unlock(&sw_alloc_mtx);
}
}
/*
* Locate hash entry. If not found create, but if we aren't adding
* anything just return. If we run out of space in the map we wait
* and, since the hash table may have changed, retry.
*/
retry:
mtx_lock(&swhash_mtx);
pswap = swp_pager_hash(object, pindex);
if ((swap = *pswap) == NULL) {
int i;
if (swapblk == SWAPBLK_NONE)
goto done;
swap = *pswap = uma_zalloc(swap_zone, M_NOWAIT);
if (swap == NULL) {
mtx_unlock(&swhash_mtx);
VM_OBJECT_UNLOCK(object);
if (uma_zone_exhausted(swap_zone)) {
printf("swap zone exhausted, increase kern.maxswzone\n");
vm_pageout_oom(VM_OOM_SWAPZ);
pause("swzonex", 10);
} else
VM_WAIT;
VM_OBJECT_LOCK(object);
goto retry;
}
swap->swb_hnext = NULL;
swap->swb_object = object;
swap->swb_index = pindex & ~(vm_pindex_t)SWAP_META_MASK;
swap->swb_count = 0;
++object->un_pager.swp.swp_bcount;
for (i = 0; i < SWAP_META_PAGES; ++i)
swap->swb_pages[i] = SWAPBLK_NONE;
}
/*
* Delete prior contents of metadata
*/
idx = pindex & SWAP_META_MASK;
if (swap->swb_pages[idx] != SWAPBLK_NONE) {
swp_pager_freeswapspace(swap->swb_pages[idx], 1);
--swap->swb_count;
}
/*
* Enter block into metadata
*/
swap->swb_pages[idx] = swapblk;
if (swapblk != SWAPBLK_NONE)
++swap->swb_count;
done:
mtx_unlock(&swhash_mtx);
}
/*
* SWP_PAGER_META_FREE() - free a range of blocks in the object's swap metadata
*
* The requested range of blocks is freed, with any associated swap
* returned to the swap bitmap.
*
* This routine will free swap metadata structures as they are cleaned
* out. This routine does *NOT* operate on swap metadata associated
* with resident pages.
*
* This routine must be called at splvm()
*/
static void
swp_pager_meta_free(vm_object_t object, vm_pindex_t index, daddr_t count)
{
VM_OBJECT_LOCK_ASSERT(object, MA_OWNED);
if (object->type != OBJT_SWAP)
return;
while (count > 0) {
struct swblock **pswap;
struct swblock *swap;
mtx_lock(&swhash_mtx);
pswap = swp_pager_hash(object, index);
if ((swap = *pswap) != NULL) {
daddr_t v = swap->swb_pages[index & SWAP_META_MASK];
if (v != SWAPBLK_NONE) {
swp_pager_freeswapspace(v, 1);
swap->swb_pages[index & SWAP_META_MASK] =
SWAPBLK_NONE;
if (--swap->swb_count == 0) {
*pswap = swap->swb_hnext;
uma_zfree(swap_zone, swap);
--object->un_pager.swp.swp_bcount;
}
}
--count;
++index;
} else {
int n = SWAP_META_PAGES - (index & SWAP_META_MASK);
count -= n;
index += n;
}
mtx_unlock(&swhash_mtx);
}
}
/*
* SWP_PAGER_META_FREE_ALL() - destroy all swap metadata associated with object
*
* This routine locates and destroys all swap metadata associated with
* an object.
*
* This routine must be called at splvm()
*/
static void
swp_pager_meta_free_all(vm_object_t object)
{
daddr_t index = 0;
VM_OBJECT_LOCK_ASSERT(object, MA_OWNED);
if (object->type != OBJT_SWAP)
return;
while (object->un_pager.swp.swp_bcount) {
struct swblock **pswap;
struct swblock *swap;
mtx_lock(&swhash_mtx);
pswap = swp_pager_hash(object, index);
if ((swap = *pswap) != NULL) {
int i;
for (i = 0; i < SWAP_META_PAGES; ++i) {
daddr_t v = swap->swb_pages[i];
if (v != SWAPBLK_NONE) {
--swap->swb_count;
swp_pager_freeswapspace(v, 1);
}
}
if (swap->swb_count != 0)
panic("swap_pager_meta_free_all: swb_count != 0");
*pswap = swap->swb_hnext;
uma_zfree(swap_zone, swap);
--object->un_pager.swp.swp_bcount;
}
mtx_unlock(&swhash_mtx);
index += SWAP_META_PAGES;
if (index > 0x20000000)
panic("swp_pager_meta_free_all: failed to locate all swap meta blocks");
}
}
/*
* SWP_PAGER_METACTL() - misc control of swap and vm_page_t meta data.
*
* This routine is capable of looking up, popping, or freeing
* swapblk assignments in the swap meta data or in the vm_page_t.
* The routine typically returns the swapblk being looked-up, or popped,
* or SWAPBLK_NONE if the block was freed, or SWAPBLK_NONE if the block
* was invalid. This routine will automatically free any invalid
* meta-data swapblks.
*
* It is not possible to store invalid swapblks in the swap meta data
* (other then a literal 'SWAPBLK_NONE'), so we don't bother checking.
*
* When acting on a busy resident page and paging is in progress, we
* have to wait until paging is complete but otherwise can act on the
* busy page.
*
* This routine must be called at splvm().
*
* SWM_FREE remove and free swap block from metadata
* SWM_POP remove from meta data but do not free.. pop it out
*/
static daddr_t
swp_pager_meta_ctl(vm_object_t object, vm_pindex_t pindex, int flags)
{
struct swblock **pswap;
struct swblock *swap;
daddr_t r1;
int idx;
VM_OBJECT_LOCK_ASSERT(object, MA_OWNED);
/*
* The meta data only exists of the object is OBJT_SWAP
* and even then might not be allocated yet.
*/
if (object->type != OBJT_SWAP)
return (SWAPBLK_NONE);
r1 = SWAPBLK_NONE;
mtx_lock(&swhash_mtx);
pswap = swp_pager_hash(object, pindex);
if ((swap = *pswap) != NULL) {
idx = pindex & SWAP_META_MASK;
r1 = swap->swb_pages[idx];
if (r1 != SWAPBLK_NONE) {
if (flags & SWM_FREE) {
swp_pager_freeswapspace(r1, 1);
r1 = SWAPBLK_NONE;
}
if (flags & (SWM_FREE|SWM_POP)) {
swap->swb_pages[idx] = SWAPBLK_NONE;
if (--swap->swb_count == 0) {
*pswap = swap->swb_hnext;
uma_zfree(swap_zone, swap);
--object->un_pager.swp.swp_bcount;
}
}
}
}
mtx_unlock(&swhash_mtx);
return (r1);
}
/*
* System call swapon(name) enables swapping on device name,
* which must be in the swdevsw. Return EBUSY
* if already swapping on this device.
*/
#ifndef _SYS_SYSPROTO_H_
struct swapon_args {
char *name;
};
#endif
/*
* MPSAFE
*/
/* ARGSUSED */
int
swapon(struct thread *td, struct swapon_args *uap)
{
struct vattr attr;
struct vnode *vp;
struct nameidata nd;
int error;
error = priv_check(td, PRIV_SWAPON);
if (error)
return (error);
mtx_lock(&Giant);
while (swdev_syscall_active)
tsleep(&swdev_syscall_active, PUSER - 1, "swpon", 0);
swdev_syscall_active = 1;
/*
* Swap metadata may not fit in the KVM if we have physical
* memory of >1GB.
*/
if (swap_zone == NULL) {
error = ENOMEM;
goto done;
}
NDINIT(&nd, LOOKUP, ISOPEN | FOLLOW | AUDITVNODE1, UIO_USERSPACE,
uap->name, td);
error = namei(&nd);
if (error)
goto done;
NDFREE(&nd, NDF_ONLY_PNBUF);
vp = nd.ni_vp;
if (vn_isdisk(vp, &error)) {
error = swapongeom(td, vp);
} else if (vp->v_type == VREG &&
(vp->v_mount->mnt_vfc->vfc_flags & VFCF_NETWORK) != 0 &&
(error = VOP_GETATTR(vp, &attr, td->td_ucred)) == 0) {
/*
* Allow direct swapping to NFS regular files in the same
* way that nfs_mountroot() sets up diskless swapping.
*/
error = swaponvp(td, vp, attr.va_size / DEV_BSIZE);
}
if (error)
vrele(vp);
done:
swdev_syscall_active = 0;
wakeup_one(&swdev_syscall_active);
mtx_unlock(&Giant);
return (error);
}
static void
swaponsomething(struct vnode *vp, void *id, u_long nblks, sw_strategy_t *strategy, sw_close_t *close, dev_t dev)
{
struct swdevt *sp, *tsp;
swblk_t dvbase;
u_long mblocks;
/*
* If we go beyond this, we get overflows in the radix
* tree bitmap code.
*/
mblocks = 0x40000000 / BLIST_META_RADIX;
if (nblks > mblocks) {
printf("WARNING: reducing size to maximum of %lu blocks per swap unit\n",
mblocks);
nblks = mblocks;
}
/*
* nblks is in DEV_BSIZE'd chunks, convert to PAGE_SIZE'd chunks.
* First chop nblks off to page-align it, then convert.
*
* sw->sw_nblks is in page-sized chunks now too.
*/
nblks &= ~(ctodb(1) - 1);
nblks = dbtoc(nblks);
sp = malloc(sizeof *sp, M_VMPGDATA, M_WAITOK | M_ZERO);
sp->sw_vp = vp;
sp->sw_id = id;
sp->sw_dev = dev;
sp->sw_flags = 0;
sp->sw_nblks = nblks;
sp->sw_used = 0;
sp->sw_strategy = strategy;
sp->sw_close = close;
sp->sw_blist = blist_create(nblks, M_WAITOK);
/*
* Do not free the first two block in order to avoid overwriting
* any bsd label at the front of the partition
*/
blist_free(sp->sw_blist, 2, nblks - 2);
dvbase = 0;
mtx_lock(&sw_dev_mtx);
TAILQ_FOREACH(tsp, &swtailq, sw_list) {
if (tsp->sw_end >= dvbase) {
/*
* We put one uncovered page between the devices
* in order to definitively prevent any cross-device
* I/O requests
*/
dvbase = tsp->sw_end + 1;
}
}
sp->sw_first = dvbase;
sp->sw_end = dvbase + nblks;
TAILQ_INSERT_TAIL(&swtailq, sp, sw_list);
nswapdev++;
swap_pager_avail += nblks;
swap_total += (vm_ooffset_t)nblks * PAGE_SIZE;
swp_sizecheck();
mtx_unlock(&sw_dev_mtx);
}
/*
* SYSCALL: swapoff(devname)
*
* Disable swapping on the given device.
*
* XXX: Badly designed system call: it should use a device index
* rather than filename as specification. We keep sw_vp around
* only to make this work.
*/
#ifndef _SYS_SYSPROTO_H_
struct swapoff_args {
char *name;
};
#endif
/*
* MPSAFE
*/
/* ARGSUSED */
int
swapoff(struct thread *td, struct swapoff_args *uap)
{
struct vnode *vp;
struct nameidata nd;
struct swdevt *sp;
int error;
error = priv_check(td, PRIV_SWAPOFF);
if (error)
return (error);
mtx_lock(&Giant);
while (swdev_syscall_active)
tsleep(&swdev_syscall_active, PUSER - 1, "swpoff", 0);
swdev_syscall_active = 1;
NDINIT(&nd, LOOKUP, FOLLOW | AUDITVNODE1, UIO_USERSPACE, uap->name,
td);
error = namei(&nd);
if (error)
goto done;
NDFREE(&nd, NDF_ONLY_PNBUF);
vp = nd.ni_vp;
mtx_lock(&sw_dev_mtx);
TAILQ_FOREACH(sp, &swtailq, sw_list) {
if (sp->sw_vp == vp)
break;
}
mtx_unlock(&sw_dev_mtx);
if (sp == NULL) {
error = EINVAL;
goto done;
}
error = swapoff_one(sp, td->td_ucred);
done:
swdev_syscall_active = 0;
wakeup_one(&swdev_syscall_active);
mtx_unlock(&Giant);
return (error);
}
static int
swapoff_one(struct swdevt *sp, struct ucred *cred)
{
u_long nblks, dvbase;
#ifdef MAC
int error;
#endif
mtx_assert(&Giant, MA_OWNED);
#ifdef MAC
(void) vn_lock(sp->sw_vp, LK_EXCLUSIVE | LK_RETRY);
error = mac_system_check_swapoff(cred, sp->sw_vp);
(void) VOP_UNLOCK(sp->sw_vp, 0);
if (error != 0)
return (error);
#endif
nblks = sp->sw_nblks;
/*
* We can turn off this swap device safely only if the
* available virtual memory in the system will fit the amount
* of data we will have to page back in, plus an epsilon so
* the system doesn't become critically low on swap space.
*/
if (cnt.v_free_count + cnt.v_cache_count + swap_pager_avail <
nblks + nswap_lowat) {
return (ENOMEM);
}
/*
* Prevent further allocations on this device.
*/
mtx_lock(&sw_dev_mtx);
sp->sw_flags |= SW_CLOSING;
for (dvbase = 0; dvbase < sp->sw_end; dvbase += dmmax) {
swap_pager_avail -= blist_fill(sp->sw_blist,
dvbase, dmmax);
}
swap_total -= (vm_ooffset_t)nblks * PAGE_SIZE;
mtx_unlock(&sw_dev_mtx);
/*
* Page in the contents of the device and close it.
*/
swap_pager_swapoff(sp);
sp->sw_close(curthread, sp);
sp->sw_id = NULL;
mtx_lock(&sw_dev_mtx);
TAILQ_REMOVE(&swtailq, sp, sw_list);
nswapdev--;
if (nswapdev == 0) {
swap_pager_full = 2;
swap_pager_almost_full = 1;
}
if (swdevhd == sp)
swdevhd = NULL;
mtx_unlock(&sw_dev_mtx);
blist_destroy(sp->sw_blist);
free(sp, M_VMPGDATA);
return (0);
}
void
swapoff_all(void)
{
struct swdevt *sp, *spt;
const char *devname;
int error;
mtx_lock(&Giant);
while (swdev_syscall_active)
tsleep(&swdev_syscall_active, PUSER - 1, "swpoff", 0);
swdev_syscall_active = 1;
mtx_lock(&sw_dev_mtx);
TAILQ_FOREACH_SAFE(sp, &swtailq, sw_list, spt) {
mtx_unlock(&sw_dev_mtx);
if (vn_isdisk(sp->sw_vp, NULL))
devname = sp->sw_vp->v_rdev->si_name;
else
devname = "[file]";
error = swapoff_one(sp, thread0.td_ucred);
if (error != 0) {
printf("Cannot remove swap device %s (error=%d), "
"skipping.\n", devname, error);
} else if (bootverbose) {
printf("Swap device %s removed.\n", devname);
}
mtx_lock(&sw_dev_mtx);
}
mtx_unlock(&sw_dev_mtx);
swdev_syscall_active = 0;
wakeup_one(&swdev_syscall_active);
mtx_unlock(&Giant);
}
void
swap_pager_status(int *total, int *used)
{
struct swdevt *sp;
*total = 0;
*used = 0;
mtx_lock(&sw_dev_mtx);
TAILQ_FOREACH(sp, &swtailq, sw_list) {
*total += sp->sw_nblks;
*used += sp->sw_used;
}
mtx_unlock(&sw_dev_mtx);
}
static int
sysctl_vm_swap_info(SYSCTL_HANDLER_ARGS)
{
int *name = (int *)arg1;
int error, n;
struct xswdev xs;
struct swdevt *sp;
if (arg2 != 1) /* name length */
return (EINVAL);
n = 0;
mtx_lock(&sw_dev_mtx);
TAILQ_FOREACH(sp, &swtailq, sw_list) {
if (n == *name) {
mtx_unlock(&sw_dev_mtx);
xs.xsw_version = XSWDEV_VERSION;
xs.xsw_dev = sp->sw_dev;
xs.xsw_flags = sp->sw_flags;
xs.xsw_nblks = sp->sw_nblks;
xs.xsw_used = sp->sw_used;
error = SYSCTL_OUT(req, &xs, sizeof(xs));
return (error);
}
n++;
}
mtx_unlock(&sw_dev_mtx);
return (ENOENT);
}
SYSCTL_INT(_vm, OID_AUTO, nswapdev, CTLFLAG_RD, &nswapdev, 0,
"Number of swap devices");
SYSCTL_NODE(_vm, OID_AUTO, swap_info, CTLFLAG_RD, sysctl_vm_swap_info,
"Swap statistics by device");
/*
* vmspace_swap_count() - count the approximate swap usage in pages for a
* vmspace.
*
* The map must be locked.
*
* Swap usage is determined by taking the proportional swap used by
* VM objects backing the VM map. To make up for fractional losses,
* if the VM object has any swap use at all the associated map entries
* count for at least 1 swap page.
*/
int
vmspace_swap_count(struct vmspace *vmspace)
{
vm_map_t map = &vmspace->vm_map;
vm_map_entry_t cur;
int count = 0;
for (cur = map->header.next; cur != &map->header; cur = cur->next) {
vm_object_t object;
if ((cur->eflags & MAP_ENTRY_IS_SUB_MAP) == 0 &&
(object = cur->object.vm_object) != NULL) {
VM_OBJECT_LOCK(object);
if (object->type == OBJT_SWAP &&
object->un_pager.swp.swp_bcount != 0) {
int n = (cur->end - cur->start) / PAGE_SIZE;
count += object->un_pager.swp.swp_bcount *
SWAP_META_PAGES * n / object->size + 1;
}
VM_OBJECT_UNLOCK(object);
}
}
return (count);
}
/*
* GEOM backend
*
* Swapping onto disk devices.
*
*/
static g_orphan_t swapgeom_orphan;
static struct g_class g_swap_class = {
.name = "SWAP",
.version = G_VERSION,
.orphan = swapgeom_orphan,
};
DECLARE_GEOM_CLASS(g_swap_class, g_class);
static void
swapgeom_done(struct bio *bp2)
{
struct buf *bp;
bp = bp2->bio_caller2;
bp->b_ioflags = bp2->bio_flags;
if (bp2->bio_error)
bp->b_ioflags |= BIO_ERROR;
bp->b_resid = bp->b_bcount - bp2->bio_completed;
bp->b_error = bp2->bio_error;
bufdone(bp);
g_destroy_bio(bp2);
}
static void
swapgeom_strategy(struct buf *bp, struct swdevt *sp)
{
struct bio *bio;
struct g_consumer *cp;
cp = sp->sw_id;
if (cp == NULL) {
bp->b_error = ENXIO;
bp->b_ioflags |= BIO_ERROR;
bufdone(bp);
return;
}
if (bp->b_iocmd == BIO_WRITE)
bio = g_new_bio();
else
bio = g_alloc_bio();
if (bio == NULL) {
bp->b_error = ENOMEM;
bp->b_ioflags |= BIO_ERROR;
bufdone(bp);
return;
}
bio->bio_caller2 = bp;
bio->bio_cmd = bp->b_iocmd;
bio->bio_data = bp->b_data;
bio->bio_offset = (bp->b_blkno - sp->sw_first) * PAGE_SIZE;
bio->bio_length = bp->b_bcount;
bio->bio_done = swapgeom_done;
g_io_request(bio, cp);
return;
}
static void
swapgeom_orphan(struct g_consumer *cp)
{
struct swdevt *sp;
mtx_lock(&sw_dev_mtx);
TAILQ_FOREACH(sp, &swtailq, sw_list)
if (sp->sw_id == cp)
sp->sw_id = NULL;
mtx_unlock(&sw_dev_mtx);
}
static void
swapgeom_close_ev(void *arg, int flags)
{
struct g_consumer *cp;
cp = arg;
g_access(cp, -1, -1, 0);
g_detach(cp);
g_destroy_consumer(cp);
}
static void
swapgeom_close(struct thread *td, struct swdevt *sw)
{
/* XXX: direct call when Giant untangled */
g_waitfor_event(swapgeom_close_ev, sw->sw_id, M_WAITOK, NULL);
}
struct swh0h0 {
struct cdev *dev;
struct vnode *vp;
int error;
};
static void
swapongeom_ev(void *arg, int flags)
{
struct swh0h0 *swh;
struct g_provider *pp;
struct g_consumer *cp;
static struct g_geom *gp;
struct swdevt *sp;
u_long nblks;
int error;
swh = arg;
swh->error = 0;
pp = g_dev_getprovider(swh->dev);
if (pp == NULL) {
swh->error = ENODEV;
return;
}
mtx_lock(&sw_dev_mtx);
TAILQ_FOREACH(sp, &swtailq, sw_list) {
cp = sp->sw_id;
if (cp != NULL && cp->provider == pp) {
mtx_unlock(&sw_dev_mtx);
swh->error = EBUSY;
return;
}
}
mtx_unlock(&sw_dev_mtx);
if (gp == NULL)
gp = g_new_geomf(&g_swap_class, "swap", NULL);
cp = g_new_consumer(gp);
g_attach(cp, pp);
/*
* XXX: Everytime you think you can improve the margin for
* footshooting, somebody depends on the ability to do so:
* savecore(8) wants to write to our swapdev so we cannot
* set an exclusive count :-(
*/
error = g_access(cp, 1, 1, 0);
if (error) {
g_detach(cp);
g_destroy_consumer(cp);
swh->error = error;
return;
}
nblks = pp->mediasize / DEV_BSIZE;
swaponsomething(swh->vp, cp, nblks, swapgeom_strategy,
swapgeom_close, dev2udev(swh->dev));
swh->error = 0;
return;
}
static int
swapongeom(struct thread *td, struct vnode *vp)
{
int error;
struct swh0h0 swh;
vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
swh.dev = vp->v_rdev;
swh.vp = vp;
swh.error = 0;
/* XXX: direct call when Giant untangled */
error = g_waitfor_event(swapongeom_ev, &swh, M_WAITOK, NULL);
if (!error)
error = swh.error;
VOP_UNLOCK(vp, 0);
return (error);
}
/*
* VNODE backend
*
* This is used mainly for network filesystem (read: probably only tested
* with NFS) swapfiles.
*
*/
static void
swapdev_strategy(struct buf *bp, struct swdevt *sp)
{
struct vnode *vp2;
bp->b_blkno = ctodb(bp->b_blkno - sp->sw_first);
vp2 = sp->sw_id;
vhold(vp2);
if (bp->b_iocmd == BIO_WRITE) {
if (bp->b_bufobj)
bufobj_wdrop(bp->b_bufobj);
bufobj_wref(&vp2->v_bufobj);
}
if (bp->b_bufobj != &vp2->v_bufobj)
bp->b_bufobj = &vp2->v_bufobj;
bp->b_vp = vp2;
bp->b_iooffset = dbtob(bp->b_blkno);
bstrategy(bp);
return;
}
static void
swapdev_close(struct thread *td, struct swdevt *sp)
{
VOP_CLOSE(sp->sw_vp, FREAD | FWRITE, td->td_ucred, td);
vrele(sp->sw_vp);
}
static int
swaponvp(struct thread *td, struct vnode *vp, u_long nblks)
{
struct swdevt *sp;
int error;
if (nblks == 0)
return (ENXIO);
mtx_lock(&sw_dev_mtx);
TAILQ_FOREACH(sp, &swtailq, sw_list) {
if (sp->sw_id == vp) {
mtx_unlock(&sw_dev_mtx);
return (EBUSY);
}
}
mtx_unlock(&sw_dev_mtx);
(void) vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
#ifdef MAC
error = mac_system_check_swapon(td->td_ucred, vp);
if (error == 0)
#endif
error = VOP_OPEN(vp, FREAD | FWRITE, td->td_ucred, td, NULL);
(void) VOP_UNLOCK(vp, 0);
if (error)
return (error);
swaponsomething(vp, vp, nblks, swapdev_strategy, swapdev_close,
NODEV);
return (0);
}
|