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
2685
2686
2687
2688
2689
2690
2691
2692
2693
2694
2695
2696
2697
2698
2699
2700
2701
2702
2703
2704
2705
2706
2707
2708
2709
2710
2711
2712
2713
2714
2715
2716
2717
2718
2719
2720
2721
2722
2723
2724
2725
2726
2727
2728
2729
2730
2731
2732
2733
2734
2735
2736
2737
2738
2739
2740
2741
2742
2743
2744
2745
2746
2747
2748
2749
2750
2751
2752
2753
2754
2755
2756
2757
2758
2759
2760
2761
2762
2763
2764
2765
2766
2767
2768
2769
2770
2771
2772
2773
2774
2775
2776
2777
2778
2779
2780
2781
2782
2783
2784
2785
2786
2787
2788
2789
2790
2791
2792
2793
2794
2795
2796
2797
2798
2799
2800
2801
2802
2803
2804
2805
2806
2807
2808
2809
2810
2811
2812
2813
2814
2815
2816
2817
2818
2819
2820
2821
2822
2823
2824
2825
2826
2827
2828
2829
2830
2831
2832
2833
2834
2835
2836
2837
2838
2839
2840
2841
2842
2843
2844
2845
2846
2847
2848
2849
2850
2851
2852
2853
2854
2855
2856
2857
2858
2859
2860
2861
2862
2863
2864
2865
2866
2867
2868
2869
2870
2871
2872
2873
2874
2875
2876
2877
2878
2879
2880
2881
2882
2883
2884
2885
2886
2887
2888
2889
2890
2891
2892
2893
2894
2895
2896
2897
2898
2899
2900
2901
2902
2903
2904
2905
2906
2907
2908
2909
2910
2911
2912
2913
2914
2915
2916
2917
2918
2919
2920
2921
2922
2923
2924
2925
2926
2927
2928
2929
2930
2931
2932
2933
2934
2935
2936
2937
2938
2939
2940
2941
2942
2943
2944
2945
2946
2947
2948
2949
2950
2951
2952
2953
2954
2955
2956
2957
2958
2959
2960
2961
2962
2963
2964
2965
2966
2967
2968
2969
2970
2971
2972
2973
2974
2975
2976
2977
2978
2979
2980
2981
2982
2983
2984
2985
2986
2987
2988
2989
2990
2991
2992
2993
2994
2995
2996
2997
2998
2999
3000
3001
3002
3003
3004
3005
3006
3007
3008
3009
3010
3011
3012
3013
3014
3015
3016
3017
3018
3019
3020
3021
3022
3023
|
/*
* Mach Operating System
* Copyright (c) 1991,1990,1989,1988,1987 Carnegie Mellon University.
* Copyright (c) 1993,1994 The University of Utah and
* the Computer Systems Laboratory (CSL).
* All rights reserved.
*
* Permission to use, copy, modify and distribute this software and its
* documentation is hereby granted, provided that both the copyright
* notice and this permission notice appear in all copies of the
* software, derivative works or modified versions, and any portions
* thereof, and that both notices appear in supporting documentation.
*
* CARNEGIE MELLON, THE UNIVERSITY OF UTAH AND CSL ALLOW FREE USE OF
* THIS SOFTWARE IN ITS "AS IS" CONDITION, AND DISCLAIM ANY LIABILITY
* OF ANY KIND FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF
* THIS SOFTWARE.
*
* Carnegie Mellon requests users of this software to return to
*
* Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU
* School of Computer Science
* Carnegie Mellon University
* Pittsburgh PA 15213-3890
*
* any improvements or extensions that they make and grant Carnegie Mellon
* the rights to redistribute these changes.
*/
/*
* File: vm/vm_object.c
* Author: Avadis Tevanian, Jr., Michael Wayne Young
*
* Virtual memory object module.
*/
#include <kern/printf.h>
#include <string.h>
#include <mach/memory_object.h>
#include <vm/memory_object_default.user.h>
#include <vm/memory_object_user.user.h>
#include <machine/vm_param.h>
#include <ipc/ipc_port.h>
#include <ipc/ipc_space.h>
#include <kern/assert.h>
#include <kern/debug.h>
#include <kern/lock.h>
#include <kern/queue.h>
#include <kern/xpr.h>
#include <kern/slab.h>
#include <vm/memory_object.h>
#include <vm/vm_fault.h>
#include <vm/vm_map.h>
#include <vm/vm_object.h>
#include <vm/vm_page.h>
#include <vm/vm_pageout.h>
#if MACH_KDB
#include <ddb/db_output.h>
#endif /* MACH_KDB */
void memory_object_release(
ipc_port_t pager,
pager_request_t pager_request,
ipc_port_t pager_name); /* forward */
void vm_object_deactivate_pages(vm_object_t);
/*
* Virtual memory objects maintain the actual data
* associated with allocated virtual memory. A given
* page of memory exists within exactly one object.
*
* An object is only deallocated when all "references"
* are given up. Only one "reference" to a given
* region of an object should be writeable.
*
* Associated with each object is a list of all resident
* memory pages belonging to that object; this list is
* maintained by the "vm_page" module, but locked by the object's
* lock.
*
* Each object also records the memory object port
* that is used by the kernel to request and write
* back data (the memory object port, field "pager"),
* and the ports provided to the memory manager, the server that
* manages that data, to return data and control its
* use (the memory object control port, field "pager_request")
* and for naming (the memory object name port, field "pager_name").
*
* Virtual memory objects are allocated to provide
* zero-filled memory (vm_allocate) or map a user-defined
* memory object into a virtual address space (vm_map).
*
* Virtual memory objects that refer to a user-defined
* memory object are called "permanent", because all changes
* made in virtual memory are reflected back to the
* memory manager, which may then store it permanently.
* Other virtual memory objects are called "temporary",
* meaning that changes need be written back only when
* necessary to reclaim pages, and that storage associated
* with the object can be discarded once it is no longer
* mapped.
*
* A permanent memory object may be mapped into more
* than one virtual address space. Moreover, two threads
* may attempt to make the first mapping of a memory
* object concurrently. Only one thread is allowed to
* complete this mapping; all others wait for the
* "pager_initialized" field is asserted, indicating
* that the first thread has initialized all of the
* necessary fields in the virtual memory object structure.
*
* The kernel relies on a *default memory manager* to
* provide backing storage for the zero-filled virtual
* memory objects. The memory object ports associated
* with these temporary virtual memory objects are only
* generated and passed to the default memory manager
* when it becomes necessary. Virtual memory objects
* that depend on the default memory manager are called
* "internal". The "pager_created" field is provided to
* indicate whether these ports have ever been allocated.
*
* The kernel may also create virtual memory objects to
* hold changed pages after a copy-on-write operation.
* In this case, the virtual memory object (and its
* backing storage -- its memory object) only contain
* those pages that have been changed. The "shadow"
* field refers to the virtual memory object that contains
* the remainder of the contents. The "shadow_offset"
* field indicates where in the "shadow" these contents begin.
* The "copy" field refers to a virtual memory object
* to which changed pages must be copied before changing
* this object, in order to implement another form
* of copy-on-write optimization.
*
* The virtual memory object structure also records
* the attributes associated with its memory object.
* The "pager_ready", "can_persist" and "copy_strategy"
* fields represent those attributes. The "cached_list"
* field is used in the implementation of the persistence
* attribute.
*
* ZZZ Continue this comment.
*/
struct kmem_cache vm_object_cache; /* vm backing store cache */
/*
* All wired-down kernel memory belongs to a single virtual
* memory object (kernel_object) to avoid wasting data structures.
*/
static struct vm_object kernel_object_store;
vm_object_t kernel_object = &kernel_object_store;
/*
* Virtual memory objects that are not referenced by
* any address maps, but that are allowed to persist
* (an attribute specified by the associated memory manager),
* are kept in a queue (vm_object_cached_list).
*
* When an object from this queue is referenced again,
* for example to make another address space mapping,
* it must be removed from the queue. That is, the
* queue contains *only* objects with zero references.
*
* The kernel may choose to terminate objects from this
* queue in order to reclaim storage. The current policy
* is to permit a fixed maximum number of unreferenced
* objects (vm_object_cached_max).
*
* A simple lock (accessed by routines
* vm_object_cache_{lock,lock_try,unlock}) governs the
* object cache. It must be held when objects are
* added to or removed from the cache (in vm_object_terminate).
* The routines that acquire a reference to a virtual
* memory object based on one of the memory object ports
* must also lock the cache.
*
* Ideally, the object cache should be more isolated
* from the reference mechanism, so that the lock need
* not be held to make simple references.
*/
queue_head_t vm_object_cached_list;
int vm_object_cached_count;
int vm_object_cached_max = 4000; /* may be patched*/
decl_simple_lock_data(,vm_object_cached_lock_data)
#define vm_object_cache_lock() \
simple_lock(&vm_object_cached_lock_data)
#define vm_object_cache_lock_try() \
simple_lock_try(&vm_object_cached_lock_data)
#define vm_object_cache_unlock() \
simple_unlock(&vm_object_cached_lock_data)
/*
* Number of physical pages referenced by cached objects.
* This counter is protected by its own lock to work around
* lock ordering issues.
*/
int vm_object_cached_pages;
decl_simple_lock_data(,vm_object_cached_pages_lock_data)
/*
* Virtual memory objects are initialized from
* a template (see vm_object_allocate).
*
* When adding a new field to the virtual memory
* object structure, be sure to add initialization
* (see vm_object_init).
*/
struct vm_object vm_object_template;
/*
* vm_object_allocate:
*
* Returns a new object with the given size.
*/
static void _vm_object_setup(
vm_object_t object,
vm_size_t size)
{
*object = vm_object_template;
queue_init(&object->memq);
vm_object_lock_init(object);
object->size = size;
}
vm_object_t _vm_object_allocate(
vm_size_t size)
{
register vm_object_t object;
object = (vm_object_t) kmem_cache_alloc(&vm_object_cache);
_vm_object_setup(object, size);
return object;
}
vm_object_t vm_object_allocate(
vm_size_t size)
{
register vm_object_t object;
register ipc_port_t port;
object = _vm_object_allocate(size);
port = ipc_port_alloc_kernel();
if (port == IP_NULL)
panic("vm_object_allocate");
object->pager_name = port;
ipc_kobject_set(port, (ipc_kobject_t) object, IKOT_PAGING_NAME);
return object;
}
/*
* vm_object_bootstrap:
*
* Initialize the VM objects module.
*/
void vm_object_bootstrap(void)
{
kmem_cache_init(&vm_object_cache, "vm_object",
sizeof(struct vm_object), 0, NULL, NULL, NULL, 0);
queue_init(&vm_object_cached_list);
simple_lock_init(&vm_object_cached_lock_data);
/*
* Fill in a template object, for quick initialization
*/
vm_object_template.ref_count = 1;
vm_object_template.size = 0;
vm_object_template.resident_page_count = 0;
vm_object_template.copy = VM_OBJECT_NULL;
vm_object_template.shadow = VM_OBJECT_NULL;
vm_object_template.shadow_offset = (vm_offset_t) 0;
vm_object_template.pager = IP_NULL;
vm_object_template.paging_offset = 0;
vm_object_template.pager_request = PAGER_REQUEST_NULL;
vm_object_template.pager_name = IP_NULL;
vm_object_template.pager_created = FALSE;
vm_object_template.pager_initialized = FALSE;
vm_object_template.pager_ready = FALSE;
vm_object_template.copy_strategy = MEMORY_OBJECT_COPY_NONE;
/* ignored if temporary, will be reset before
* permanent object becomes ready */
vm_object_template.use_shared_copy = FALSE;
vm_object_template.shadowed = FALSE;
vm_object_template.absent_count = 0;
vm_object_template.all_wanted = 0; /* all bits FALSE */
vm_object_template.paging_in_progress = 0;
vm_object_template.can_persist = FALSE;
vm_object_template.internal = TRUE;
vm_object_template.temporary = TRUE;
vm_object_template.alive = TRUE;
vm_object_template.lock_in_progress = FALSE;
vm_object_template.lock_restart = FALSE;
vm_object_template.use_old_pageout = TRUE; /* XXX change later */
vm_object_template.last_alloc = (vm_offset_t) 0;
#if MACH_PAGEMAP
vm_object_template.existence_info = VM_EXTERNAL_NULL;
#endif /* MACH_PAGEMAP */
/*
* Initialize the "kernel object"
*/
_vm_object_setup(kernel_object,
VM_MAX_KERNEL_ADDRESS - VM_MIN_KERNEL_ADDRESS);
/*
* Initialize the "submap object". Make it as large as the
* kernel object so that no limit is imposed on submap sizes.
*/
_vm_object_setup(vm_submap_object,
VM_MAX_KERNEL_ADDRESS - VM_MIN_KERNEL_ADDRESS);
#if MACH_PAGEMAP
vm_external_module_initialize();
#endif /* MACH_PAGEMAP */
}
void vm_object_init(void)
{
/*
* Finish initializing the kernel object.
* The submap object doesn't need a name port.
*/
kernel_object->pager_name = ipc_port_alloc_kernel();
ipc_kobject_set(kernel_object->pager_name,
(ipc_kobject_t) kernel_object,
IKOT_PAGING_NAME);
}
/*
* vm_object_reference:
*
* Gets another reference to the given object.
*/
void vm_object_reference(
register vm_object_t object)
{
if (object == VM_OBJECT_NULL)
return;
vm_object_lock(object);
assert(object->ref_count > 0);
object->ref_count++;
vm_object_unlock(object);
}
/*
* vm_object_deallocate:
*
* Release a reference to the specified object,
* gained either through a vm_object_allocate
* or a vm_object_reference call. When all references
* are gone, storage associated with this object
* may be relinquished.
*
* No object may be locked.
*/
void vm_object_deallocate(
register vm_object_t object)
{
vm_object_t temp;
while (object != VM_OBJECT_NULL) {
/*
* The cache holds a reference (uncounted) to
* the object; we must lock it before removing
* the object.
*/
vm_object_cache_lock();
/*
* Lose the reference
*/
vm_object_lock(object);
if (--(object->ref_count) > 0) {
/*
* If there are still references, then
* we are done.
*/
vm_object_unlock(object);
vm_object_cache_unlock();
return;
}
/*
* See whether this object can persist. If so, enter
* it in the cache, then deactivate all of its
* pages.
*/
if (object->can_persist) {
boolean_t overflow;
/*
* Enter the object onto the queue
* of "cached" objects. Remember whether
* we've caused the queue to overflow,
* as a hint.
*/
queue_enter(&vm_object_cached_list, object,
vm_object_t, cached_list);
overflow = (++vm_object_cached_count > vm_object_cached_max);
vm_object_cached_pages_update(object->resident_page_count);
vm_object_cache_unlock();
vm_object_deactivate_pages(object);
vm_object_unlock(object);
/*
* If we didn't overflow, or if the queue has
* been reduced back to below the specified
* minimum, then quit.
*/
if (!overflow)
return;
while (TRUE) {
vm_object_cache_lock();
if (vm_object_cached_count <=
vm_object_cached_max) {
vm_object_cache_unlock();
return;
}
/*
* If we must trim down the queue, take
* the first object, and proceed to
* terminate it instead of the original
* object. Have to wait for pager init.
* if it's in progress.
*/
object= (vm_object_t)
queue_first(&vm_object_cached_list);
vm_object_lock(object);
if (!(object->pager_created &&
!object->pager_initialized)) {
/*
* Ok to terminate, hang on to lock.
*/
break;
}
vm_object_assert_wait(object,
VM_OBJECT_EVENT_INITIALIZED, FALSE);
vm_object_unlock(object);
vm_object_cache_unlock();
thread_block((void (*)()) 0);
/*
* Continue loop to check if cache still
* needs to be trimmed.
*/
}
/*
* Actually remove object from cache.
*/
queue_remove(&vm_object_cached_list, object,
vm_object_t, cached_list);
vm_object_cached_count--;
assert(object->ref_count == 0);
}
else {
if (object->pager_created &&
!object->pager_initialized) {
/*
* Have to wait for initialization.
* Put reference back and retry
* when it's initialized.
*/
object->ref_count++;
vm_object_assert_wait(object,
VM_OBJECT_EVENT_INITIALIZED, FALSE);
vm_object_unlock(object);
vm_object_cache_unlock();
thread_block((void (*)()) 0);
continue;
}
}
/*
* Take the reference to the shadow object
* out of the object to be destroyed.
*/
temp = object->shadow;
/*
* Destroy the object; the cache lock will
* be released in the process.
*/
vm_object_terminate(object);
/*
* Deallocate the reference to the shadow
* by continuing the loop with that object
* in place of the original.
*/
object = temp;
}
}
boolean_t vm_object_terminate_remove_all = FALSE;
/*
* Routine: vm_object_terminate
* Purpose:
* Free all resources associated with a vm_object.
* In/out conditions:
* Upon entry, the object and the cache must be locked,
* and the object must have no references.
*
* The shadow object reference is left alone.
*
* Upon exit, the cache will be unlocked, and the
* object will cease to exist.
*/
void vm_object_terminate(
register vm_object_t object)
{
register vm_page_t p;
vm_object_t shadow_object;
/*
* Make sure the object isn't already being terminated
*/
assert(object->alive);
object->alive = FALSE;
/*
* Make sure no one can look us up now.
*/
vm_object_remove(object);
vm_object_cache_unlock();
/*
* Detach the object from its shadow if we are the shadow's
* copy.
*/
if ((shadow_object = object->shadow) != VM_OBJECT_NULL) {
vm_object_lock(shadow_object);
assert((shadow_object->copy == object) ||
(shadow_object->copy == VM_OBJECT_NULL));
shadow_object->copy = VM_OBJECT_NULL;
vm_object_unlock(shadow_object);
}
/*
* The pageout daemon might be playing with our pages.
* Now that the object is dead, it won't touch any more
* pages, but some pages might already be on their way out.
* Hence, we wait until the active paging activities have ceased.
*/
vm_object_paging_wait(object, FALSE);
/*
* Clean or free the pages, as appropriate.
* It is possible for us to find busy/absent pages,
* if some faults on this object were aborted.
*/
if ((object->temporary) || (object->pager == IP_NULL)) {
while (!queue_empty(&object->memq)) {
p = (vm_page_t) queue_first(&object->memq);
VM_PAGE_CHECK(p);
if (p->busy && !p->absent)
panic("vm_object_terminate.2 0x%x 0x%x",
object, p);
VM_PAGE_FREE(p);
}
} else while (!queue_empty(&object->memq)) {
p = (vm_page_t) queue_first(&object->memq);
VM_PAGE_CHECK(p);
if (p->busy && !p->absent)
panic("vm_object_terminate.3 0x%x 0x%x", object, p);
vm_page_lock_queues();
VM_PAGE_QUEUES_REMOVE(p);
vm_page_unlock_queues();
if (p->absent || p->private) {
/*
* For private pages, VM_PAGE_FREE just
* leaves the page structure around for
* its owner to clean up. For absent
* pages, the structure is returned to
* the appropriate pool.
*/
goto free_page;
}
if (p->fictitious)
panic("vm_object_terminate.4 0x%x 0x%x", object, p);
if (!p->dirty)
p->dirty = pmap_is_modified(p->phys_addr);
if (p->dirty || p->precious) {
p->busy = TRUE;
vm_pageout_page(p, FALSE, TRUE); /* flush page */
} else {
free_page:
VM_PAGE_FREE(p);
}
}
assert(object->ref_count == 0);
assert(object->paging_in_progress == 0);
/*
* Throw away port rights... note that they may
* already have been thrown away (by vm_object_destroy
* or memory_object_destroy).
*
* Instead of destroying the control and name ports,
* we send all rights off to the memory manager instead,
* using memory_object_terminate.
*/
vm_object_unlock(object);
if (object->pager != IP_NULL) {
/* consumes our rights for pager, pager_request, pager_name */
memory_object_release(object->pager,
object->pager_request,
object->pager_name);
} else if (object->pager_name != IP_NULL) {
/* consumes our right for pager_name */
ipc_port_dealloc_kernel(object->pager_name);
}
#if MACH_PAGEMAP
vm_external_destroy(object->existence_info);
#endif /* MACH_PAGEMAP */
/*
* Free the space for the object.
*/
kmem_cache_free(&vm_object_cache, (vm_offset_t) object);
}
/*
* Routine: vm_object_pager_wakeup
* Purpose: Wake up anyone waiting for IKOT_PAGER_TERMINATING
*/
void
vm_object_pager_wakeup(
ipc_port_t pager)
{
boolean_t someone_waiting;
/*
* If anyone was waiting for the memory_object_terminate
* to be queued, wake them up now.
*/
vm_object_cache_lock();
assert(ip_kotype(pager) == IKOT_PAGER_TERMINATING);
someone_waiting = (pager->ip_kobject != IKO_NULL);
if (ip_active(pager))
ipc_kobject_set(pager, IKO_NULL, IKOT_NONE);
vm_object_cache_unlock();
if (someone_waiting) {
thread_wakeup((event_t) pager);
}
}
/*
* Routine: memory_object_release
* Purpose: Terminate the pager and release port rights,
* just like memory_object_terminate, except
* that we wake up anyone blocked in vm_object_enter
* waiting for termination message to be queued
* before calling memory_object_init.
*/
void memory_object_release(
ipc_port_t pager,
pager_request_t pager_request,
ipc_port_t pager_name)
{
/*
* Keep a reference to pager port;
* the terminate might otherwise release all references.
*/
ip_reference(pager);
/*
* Terminate the pager.
*/
(void) memory_object_terminate(pager, pager_request, pager_name);
/*
* Wakeup anyone waiting for this terminate
*/
vm_object_pager_wakeup(pager);
/*
* Release reference to pager port.
*/
ip_release(pager);
}
/*
* Routine: vm_object_abort_activity [internal use only]
* Purpose:
* Abort paging requests pending on this object.
* In/out conditions:
* The object is locked on entry and exit.
*/
void vm_object_abort_activity(
vm_object_t object)
{
register
vm_page_t p;
vm_page_t next;
/*
* Abort all activity that would be waiting
* for a result on this memory object.
*
* We could also choose to destroy all pages
* that we have in memory for this object, but
* we don't.
*/
p = (vm_page_t) queue_first(&object->memq);
while (!queue_end(&object->memq, (queue_entry_t) p)) {
next = (vm_page_t) queue_next(&p->listq);
/*
* If it's being paged in, destroy it.
* If an unlock has been requested, start it again.
*/
if (p->busy && p->absent) {
VM_PAGE_FREE(p);
}
else {
if (p->unlock_request != VM_PROT_NONE)
p->unlock_request = VM_PROT_NONE;
PAGE_WAKEUP(p);
}
p = next;
}
/*
* Wake up threads waiting for the memory object to
* become ready.
*/
object->pager_ready = TRUE;
vm_object_wakeup(object, VM_OBJECT_EVENT_PAGER_READY);
}
/*
* Routine: memory_object_destroy [user interface]
* Purpose:
* Shut down a memory object, despite the
* presence of address map (or other) references
* to the vm_object.
* Note:
* This routine may be called either from the user interface,
* or from port destruction handling (via vm_object_destroy).
*/
kern_return_t memory_object_destroy(
register
vm_object_t object,
kern_return_t reason)
{
ipc_port_t old_object, old_name;
pager_request_t old_control;
if (object == VM_OBJECT_NULL)
return KERN_SUCCESS;
/*
* Remove the port associations immediately.
*
* This will prevent the memory manager from further
* meddling. [If it wanted to flush data or make
* other changes, it should have done so before performing
* the destroy call.]
*/
vm_object_cache_lock();
vm_object_lock(object);
vm_object_remove(object);
object->can_persist = FALSE;
vm_object_cache_unlock();
/*
* Rip out the ports from the vm_object now... this
* will prevent new memory_object calls from succeeding.
*/
old_object = object->pager;
object->pager = IP_NULL;
old_control = object->pager_request;
object->pager_request = PAGER_REQUEST_NULL;
old_name = object->pager_name;
object->pager_name = IP_NULL;
/*
* Wait for existing paging activity (that might
* have the old ports) to subside.
*/
vm_object_paging_wait(object, FALSE);
vm_object_unlock(object);
/*
* Shut down the ports now.
*
* [Paging operations may be proceeding concurrently --
* they'll get the null values established above.]
*/
if (old_object != IP_NULL) {
/* consumes our rights for object, control, name */
memory_object_release(old_object, old_control,
old_name);
} else if (old_name != IP_NULL) {
/* consumes our right for name */
ipc_port_dealloc_kernel(object->pager_name);
}
/*
* Lose the reference that was donated for this routine
*/
vm_object_deallocate(object);
return KERN_SUCCESS;
}
/*
* vm_object_deactivate_pages
*
* Deactivate all pages in the specified object. (Keep its pages
* in memory even though it is no longer referenced.)
*
* The object must be locked.
*/
void vm_object_deactivate_pages(
register vm_object_t object)
{
register vm_page_t p;
queue_iterate(&object->memq, p, vm_page_t, listq) {
vm_page_lock_queues();
if (!p->busy)
vm_page_deactivate(p);
vm_page_unlock_queues();
}
}
/*
* Routine: vm_object_pmap_protect
*
* Purpose:
* Reduces the permission for all physical
* pages in the specified object range.
*
* If removing write permission only, it is
* sufficient to protect only the pages in
* the top-level object; only those pages may
* have write permission.
*
* If removing all access, we must follow the
* shadow chain from the top-level object to
* remove access to all pages in shadowed objects.
*
* The object must *not* be locked. The object must
* be temporary/internal.
*
* If pmap is not NULL, this routine assumes that
* the only mappings for the pages are in that
* pmap.
*/
boolean_t vm_object_pmap_protect_by_page = FALSE;
void vm_object_pmap_protect(
register vm_object_t object,
register vm_offset_t offset,
vm_size_t size,
pmap_t pmap,
vm_offset_t pmap_start,
vm_prot_t prot)
{
if (object == VM_OBJECT_NULL)
return;
vm_object_lock(object);
assert(object->temporary && object->internal);
while (TRUE) {
if (object->resident_page_count > atop(size) / 2 &&
pmap != PMAP_NULL) {
vm_object_unlock(object);
pmap_protect(pmap, pmap_start, pmap_start + size, prot);
return;
}
{
register vm_page_t p;
register vm_offset_t end;
end = offset + size;
queue_iterate(&object->memq, p, vm_page_t, listq) {
if (!p->fictitious &&
(offset <= p->offset) &&
(p->offset < end)) {
if ((pmap == PMAP_NULL) ||
vm_object_pmap_protect_by_page) {
pmap_page_protect(p->phys_addr,
prot & ~p->page_lock);
} else {
vm_offset_t start =
pmap_start +
(p->offset - offset);
pmap_protect(pmap,
start,
start + PAGE_SIZE,
prot);
}
}
}
}
if (prot == VM_PROT_NONE) {
/*
* Must follow shadow chain to remove access
* to pages in shadowed objects.
*/
register vm_object_t next_object;
next_object = object->shadow;
if (next_object != VM_OBJECT_NULL) {
offset += object->shadow_offset;
vm_object_lock(next_object);
vm_object_unlock(object);
object = next_object;
}
else {
/*
* End of chain - we are done.
*/
break;
}
}
else {
/*
* Pages in shadowed objects may never have
* write permission - we may stop here.
*/
break;
}
}
vm_object_unlock(object);
}
/*
* vm_object_pmap_remove:
*
* Removes all physical pages in the specified
* object range from all physical maps.
*
* The object must *not* be locked.
*/
void vm_object_pmap_remove(
register vm_object_t object,
register vm_offset_t start,
register vm_offset_t end)
{
register vm_page_t p;
if (object == VM_OBJECT_NULL)
return;
vm_object_lock(object);
queue_iterate(&object->memq, p, vm_page_t, listq) {
if (!p->fictitious &&
(start <= p->offset) &&
(p->offset < end))
pmap_page_protect(p->phys_addr, VM_PROT_NONE);
}
vm_object_unlock(object);
}
/*
* Routine: vm_object_copy_slowly
*
* Description:
* Copy the specified range of the source
* virtual memory object without using
* protection-based optimizations (such
* as copy-on-write). The pages in the
* region are actually copied.
*
* In/out conditions:
* The caller must hold a reference and a lock
* for the source virtual memory object. The source
* object will be returned *unlocked*.
*
* Results:
* If the copy is completed successfully, KERN_SUCCESS is
* returned. If the caller asserted the interruptible
* argument, and an interruption occurred while waiting
* for a user-generated event, MACH_SEND_INTERRUPTED is
* returned. Other values may be returned to indicate
* hard errors during the copy operation.
*
* A new virtual memory object is returned in a
* parameter (_result_object). The contents of this
* new object, starting at a zero offset, are a copy
* of the source memory region. In the event of
* an error, this parameter will contain the value
* VM_OBJECT_NULL.
*/
kern_return_t vm_object_copy_slowly(
register
vm_object_t src_object,
vm_offset_t src_offset,
vm_size_t size,
boolean_t interruptible,
vm_object_t *_result_object) /* OUT */
{
vm_object_t new_object;
vm_offset_t new_offset;
if (size == 0) {
vm_object_unlock(src_object);
*_result_object = VM_OBJECT_NULL;
return KERN_INVALID_ARGUMENT;
}
/*
* Prevent destruction of the source object while we copy.
*/
assert(src_object->ref_count > 0);
src_object->ref_count++;
vm_object_unlock(src_object);
/*
* Create a new object to hold the copied pages.
* A few notes:
* We fill the new object starting at offset 0,
* regardless of the input offset.
* We don't bother to lock the new object within
* this routine, since we have the only reference.
*/
new_object = vm_object_allocate(size);
new_offset = 0;
assert(size == trunc_page(size)); /* Will the loop terminate? */
for ( ;
size != 0 ;
src_offset += PAGE_SIZE, new_offset += PAGE_SIZE, size -= PAGE_SIZE
) {
vm_page_t new_page;
vm_fault_return_t result;
while ((new_page = vm_page_alloc(new_object, new_offset))
== VM_PAGE_NULL) {
VM_PAGE_WAIT((void (*)()) 0);
}
do {
vm_prot_t prot = VM_PROT_READ;
vm_page_t _result_page;
vm_page_t top_page;
register
vm_page_t result_page;
vm_object_lock(src_object);
src_object->paging_in_progress++;
result = vm_fault_page(src_object, src_offset,
VM_PROT_READ, FALSE, interruptible,
&prot, &_result_page, &top_page,
FALSE, (void (*)()) 0);
switch(result) {
case VM_FAULT_SUCCESS:
result_page = _result_page;
/*
* We don't need to hold the object
* lock -- the busy page will be enough.
* [We don't care about picking up any
* new modifications.]
*
* Copy the page to the new object.
*
* POLICY DECISION:
* If result_page is clean,
* we could steal it instead
* of copying.
*/
vm_object_unlock(result_page->object);
vm_page_copy(result_page, new_page);
/*
* Let go of both pages (make them
* not busy, perform wakeup, activate).
*/
new_page->busy = FALSE;
new_page->dirty = TRUE;
vm_object_lock(result_page->object);
PAGE_WAKEUP_DONE(result_page);
vm_page_lock_queues();
if (!result_page->active &&
!result_page->inactive)
vm_page_activate(result_page);
vm_page_activate(new_page);
vm_page_unlock_queues();
/*
* Release paging references and
* top-level placeholder page, if any.
*/
vm_fault_cleanup(result_page->object,
top_page);
break;
case VM_FAULT_RETRY:
break;
case VM_FAULT_MEMORY_SHORTAGE:
VM_PAGE_WAIT((void (*)()) 0);
break;
case VM_FAULT_FICTITIOUS_SHORTAGE:
vm_page_more_fictitious();
break;
case VM_FAULT_INTERRUPTED:
vm_page_free(new_page);
vm_object_deallocate(new_object);
vm_object_deallocate(src_object);
*_result_object = VM_OBJECT_NULL;
return MACH_SEND_INTERRUPTED;
case VM_FAULT_MEMORY_ERROR:
/*
* A policy choice:
* (a) ignore pages that we can't
* copy
* (b) return the null object if
* any page fails [chosen]
*/
vm_page_free(new_page);
vm_object_deallocate(new_object);
vm_object_deallocate(src_object);
*_result_object = VM_OBJECT_NULL;
return KERN_MEMORY_ERROR;
}
} while (result != VM_FAULT_SUCCESS);
}
/*
* Lose the extra reference, and return our object.
*/
vm_object_deallocate(src_object);
*_result_object = new_object;
return KERN_SUCCESS;
}
/*
* Routine: vm_object_copy_temporary
*
* Purpose:
* Copy the specified range of the source virtual
* memory object, if it can be done without blocking.
*
* Results:
* If the copy is successful, the copy is returned in
* the arguments; otherwise, the arguments are not
* affected.
*
* In/out conditions:
* The object should be unlocked on entry and exit.
*/
vm_object_t vm_object_copy_delayed(); /* forward declaration */
boolean_t vm_object_copy_temporary(
vm_object_t *_object, /* INOUT */
vm_offset_t *_offset, /* INOUT */
boolean_t *_src_needs_copy, /* OUT */
boolean_t *_dst_needs_copy) /* OUT */
{
vm_object_t object = *_object;
if (object == VM_OBJECT_NULL) {
*_src_needs_copy = FALSE;
*_dst_needs_copy = FALSE;
return TRUE;
}
/*
* If the object is temporary, we can perform
* a symmetric copy-on-write without asking.
*/
vm_object_lock(object);
if (object->temporary) {
/*
* Shared objects use delayed copy
*/
if (object->use_shared_copy) {
/*
* Asymmetric copy strategy. Destination
* must be copied (to allow copy object reuse).
* Source is unaffected.
*/
vm_object_unlock(object);
object = vm_object_copy_delayed(object);
*_object = object;
*_src_needs_copy = FALSE;
*_dst_needs_copy = TRUE;
return TRUE;
}
/*
* Make another reference to the object.
*
* Leave object/offset unchanged.
*/
assert(object->ref_count > 0);
object->ref_count++;
object->shadowed = TRUE;
vm_object_unlock(object);
/*
* Both source and destination must make
* shadows, and the source must be made
* read-only if not already.
*/
*_src_needs_copy = TRUE;
*_dst_needs_copy = TRUE;
return TRUE;
}
if (object->pager_ready &&
(object->copy_strategy == MEMORY_OBJECT_COPY_DELAY)) {
/* XXX Do something intelligent (see temporary code above) */
}
vm_object_unlock(object);
return FALSE;
}
/*
* Routine: vm_object_copy_call [internal]
*
* Description:
* Copy the specified (src_offset, size) portion
* of the source object (src_object), using the
* user-managed copy algorithm.
*
* In/out conditions:
* The source object must be locked on entry. It
* will be *unlocked* on exit.
*
* Results:
* If the copy is successful, KERN_SUCCESS is returned.
* This routine is interruptible; if a wait for
* a user-generated event is interrupted, MACH_SEND_INTERRUPTED
* is returned. Other return values indicate hard errors
* in creating the user-managed memory object for the copy.
*
* A new object that represents the copied virtual
* memory is returned in a parameter (*_result_object).
* If the return value indicates an error, this parameter
* is not valid.
*/
kern_return_t vm_object_copy_call(
vm_object_t src_object,
vm_offset_t src_offset,
vm_size_t size,
vm_object_t *_result_object) /* OUT */
{
vm_offset_t src_end = src_offset + size;
ipc_port_t new_memory_object;
vm_object_t new_object;
vm_page_t p;
/*
* Set the backing object for the new
* temporary object.
*/
assert(src_object->ref_count > 0);
src_object->ref_count++;
vm_object_paging_begin(src_object);
vm_object_unlock(src_object);
/*
* Create a memory object port to be associated
* with this new vm_object.
*
* Since the kernel has the only rights to this
* port, we need not hold the cache lock.
*
* Since we have the only object reference, we
* need not be worried about collapse operations.
*
*/
new_memory_object = ipc_port_alloc_kernel();
if (new_memory_object == IP_NULL) {
panic("vm_object_copy_call: allocate memory object port");
/* XXX Shouldn't panic here. */
}
/* we hold a naked receive right for new_memory_object */
(void) ipc_port_make_send(new_memory_object);
/* now we also hold a naked send right for new_memory_object */
/*
* Let the memory manager know that a copy operation
* is in progress. Note that we're using the old
* memory object's ports (for which we're holding
* a paging reference)... the memory manager cannot
* yet affect the new memory object.
*/
(void) memory_object_copy(src_object->pager,
src_object->pager_request,
src_offset, size,
new_memory_object);
/* no longer hold the naked receive right for new_memory_object */
vm_object_lock(src_object);
vm_object_paging_end(src_object);
/*
* Remove write access from all of the pages of
* the old memory object that we can.
*/
queue_iterate(&src_object->memq, p, vm_page_t, listq) {
if (!p->fictitious &&
(src_offset <= p->offset) &&
(p->offset < src_end) &&
!(p->page_lock & VM_PROT_WRITE)) {
p->page_lock |= VM_PROT_WRITE;
pmap_page_protect(p->phys_addr, VM_PROT_ALL & ~p->page_lock);
}
}
vm_object_unlock(src_object);
/*
* Initialize the rest of the paging stuff
*/
new_object = vm_object_enter(new_memory_object, size, FALSE);
new_object->shadow = src_object;
new_object->shadow_offset = src_offset;
/*
* Drop the reference for new_memory_object taken above.
*/
ipc_port_release_send(new_memory_object);
/* no longer hold the naked send right for new_memory_object */
*_result_object = new_object;
return KERN_SUCCESS;
}
/*
* Routine: vm_object_copy_delayed [internal]
*
* Description:
* Copy the specified virtual memory object, using
* the asymmetric copy-on-write algorithm.
*
* In/out conditions:
* The object must be unlocked on entry.
*
* This routine will not block waiting for user-generated
* events. It is not interruptible.
*/
vm_object_t vm_object_copy_delayed(
vm_object_t src_object)
{
vm_object_t new_copy;
vm_object_t old_copy;
vm_page_t p;
/*
* The user-level memory manager wants to see
* all of the changes to this object, but it
* has promised not to make any changes on its own.
*
* Perform an asymmetric copy-on-write, as follows:
* Create a new object, called a "copy object"
* to hold pages modified by the new mapping
* (i.e., the copy, not the original mapping).
* Record the original object as the backing
* object for the copy object. If the
* original mapping does not change a page,
* it may be used read-only by the copy.
* Record the copy object in the original
* object. When the original mapping causes
* a page to be modified, it must be copied
* to a new page that is "pushed" to the
* copy object.
* Mark the new mapping (the copy object)
* copy-on-write. This makes the copy
* object itself read-only, allowing it
* to be reused if the original mapping
* makes no changes, and simplifying the
* synchronization required in the "push"
* operation described above.
*
* The copy-on-write is said to be assymetric because
* the original object is *not* marked copy-on-write.
* A copied page is pushed to the copy object, regardless
* which party attempted to modify the page.
*
* Repeated asymmetric copy operations may be done.
* If the original object has not been changed since
* the last copy, its copy object can be reused.
* Otherwise, a new copy object can be inserted
* between the original object and its previous
* copy object. Since any copy object is read-only,
* this cannot affect the contents of the previous copy
* object.
*
* Note that a copy object is higher in the object
* tree than the original object; therefore, use of
* the copy object recorded in the original object
* must be done carefully, to avoid deadlock.
*/
/*
* Allocate a new copy object before locking, even
* though we may not need it later.
*/
new_copy = vm_object_allocate(src_object->size);
vm_object_lock(src_object);
/*
* See whether we can reuse the result of a previous
* copy operation.
*/
Retry:
old_copy = src_object->copy;
if (old_copy != VM_OBJECT_NULL) {
/*
* Try to get the locks (out of order)
*/
if (!vm_object_lock_try(old_copy)) {
vm_object_unlock(src_object);
simple_lock_pause(); /* wait a bit */
vm_object_lock(src_object);
goto Retry;
}
/*
* Determine whether the old copy object has
* been modified.
*/
if (old_copy->resident_page_count == 0 &&
!old_copy->pager_created) {
/*
* It has not been modified.
*
* Return another reference to
* the existing copy-object.
*/
assert(old_copy->ref_count > 0);
old_copy->ref_count++;
vm_object_unlock(old_copy);
vm_object_unlock(src_object);
vm_object_deallocate(new_copy);
return old_copy;
}
/*
* The copy-object is always made large enough to
* completely shadow the original object, since
* it may have several users who want to shadow
* the original object at different points.
*/
assert((old_copy->shadow == src_object) &&
(old_copy->shadow_offset == (vm_offset_t) 0));
/*
* Make the old copy-object shadow the new one.
* It will receive no more pages from the original
* object.
*/
src_object->ref_count--; /* remove ref. from old_copy */
assert(src_object->ref_count > 0);
old_copy->shadow = new_copy;
assert(new_copy->ref_count > 0);
new_copy->ref_count++;
vm_object_unlock(old_copy); /* done with old_copy */
}
/*
* Point the new copy at the existing object.
*/
new_copy->shadow = src_object;
new_copy->shadow_offset = 0;
new_copy->shadowed = TRUE; /* caller must set needs_copy */
assert(src_object->ref_count > 0);
src_object->ref_count++;
src_object->copy = new_copy;
/*
* Mark all pages of the existing object copy-on-write.
* This object may have a shadow chain below it, but
* those pages will already be marked copy-on-write.
*/
queue_iterate(&src_object->memq, p, vm_page_t, listq) {
if (!p->fictitious)
pmap_page_protect(p->phys_addr,
(VM_PROT_ALL & ~VM_PROT_WRITE &
~p->page_lock));
}
vm_object_unlock(src_object);
return new_copy;
}
/*
* Routine: vm_object_copy_strategically
*
* Purpose:
* Perform a copy according to the source object's
* declared strategy. This operation may block,
* and may be interrupted.
*/
kern_return_t vm_object_copy_strategically(
register
vm_object_t src_object,
vm_offset_t src_offset,
vm_size_t size,
vm_object_t *dst_object, /* OUT */
vm_offset_t *dst_offset, /* OUT */
boolean_t *dst_needs_copy) /* OUT */
{
kern_return_t result = KERN_SUCCESS; /* to quiet gcc warnings */
boolean_t interruptible = TRUE; /* XXX */
assert(src_object != VM_OBJECT_NULL);
vm_object_lock(src_object);
/* XXX assert(!src_object->temporary); JSB FIXME */
/*
* The copy strategy is only valid if the memory manager
* is "ready".
*/
while (!src_object->pager_ready) {
vm_object_wait( src_object,
VM_OBJECT_EVENT_PAGER_READY,
interruptible);
if (interruptible &&
(current_thread()->wait_result != THREAD_AWAKENED)) {
*dst_object = VM_OBJECT_NULL;
*dst_offset = 0;
*dst_needs_copy = FALSE;
return MACH_SEND_INTERRUPTED;
}
vm_object_lock(src_object);
}
/*
* The object may be temporary (even though it is external).
* If so, do a symmetric copy.
*/
if (src_object->temporary) {
/*
* XXX
* This does not count as intelligent!
* This buys us the object->temporary optimizations,
* but we aren't using a symmetric copy,
* which may confuse the vm code. The correct thing
* to do here is to figure out what to call to get
* a temporary shadowing set up.
*/
src_object->copy_strategy = MEMORY_OBJECT_COPY_DELAY;
}
/*
* The object is permanent. Use the appropriate copy strategy.
*/
switch (src_object->copy_strategy) {
case MEMORY_OBJECT_COPY_NONE:
if ((result = vm_object_copy_slowly(
src_object,
src_offset,
size,
interruptible,
dst_object))
== KERN_SUCCESS) {
*dst_offset = 0;
*dst_needs_copy = FALSE;
}
break;
case MEMORY_OBJECT_COPY_CALL:
if ((result = vm_object_copy_call(
src_object,
src_offset,
size,
dst_object))
== KERN_SUCCESS) {
*dst_offset = 0;
*dst_needs_copy = FALSE;
}
break;
case MEMORY_OBJECT_COPY_DELAY:
vm_object_unlock(src_object);
*dst_object = vm_object_copy_delayed(src_object);
*dst_offset = src_offset;
*dst_needs_copy = TRUE;
result = KERN_SUCCESS;
break;
}
return result;
}
/*
* vm_object_shadow:
*
* Create a new object which is backed by the
* specified existing object range. The source
* object reference is deallocated.
*
* The new object and offset into that object
* are returned in the source parameters.
*/
void vm_object_shadow(
vm_object_t *object, /* IN/OUT */
vm_offset_t *offset, /* IN/OUT */
vm_size_t length)
{
register vm_object_t source;
register vm_object_t result;
source = *object;
/*
* Allocate a new object with the given length
*/
if ((result = vm_object_allocate(length)) == VM_OBJECT_NULL)
panic("vm_object_shadow: no object for shadowing");
/*
* The new object shadows the source object, adding
* a reference to it. Our caller changes his reference
* to point to the new object, removing a reference to
* the source object. Net result: no change of reference
* count.
*/
result->shadow = source;
/*
* Store the offset into the source object,
* and fix up the offset into the new object.
*/
result->shadow_offset = *offset;
/*
* Return the new things
*/
*offset = 0;
*object = result;
}
/*
* The relationship between vm_object structures and
* the memory_object ports requires careful synchronization.
*
* All associations are created by vm_object_enter. All three
* port fields are filled in, as follows:
* pager: the memory_object port itself, supplied by
* the user requesting a mapping (or the kernel,
* when initializing internal objects); the
* kernel simulates holding send rights by keeping
* a port reference;
* pager_request:
* pager_name:
* the memory object control and name ports,
* created by the kernel; the kernel holds
* receive (and ownership) rights to these
* ports, but no other references.
* All of the ports are referenced by their global names.
*
* When initialization is complete, the "initialized" field
* is asserted. Other mappings using a particular memory object,
* and any references to the vm_object gained through the
* port association must wait for this initialization to occur.
*
* In order to allow the memory manager to set attributes before
* requests (notably virtual copy operations, but also data or
* unlock requests) are made, a "ready" attribute is made available.
* Only the memory manager may affect the value of this attribute.
* Its value does not affect critical kernel functions, such as
* internal object initialization or destruction. [Furthermore,
* memory objects created by the kernel are assumed to be ready
* immediately; the default memory manager need not explicitly
* set the "ready" attribute.]
*
* [Both the "initialized" and "ready" attribute wait conditions
* use the "pager" field as the wait event.]
*
* The port associations can be broken down by any of the
* following routines:
* vm_object_terminate:
* No references to the vm_object remain, and
* the object cannot (or will not) be cached.
* This is the normal case, and is done even
* though one of the other cases has already been
* done.
* vm_object_destroy:
* The memory_object port has been destroyed,
* meaning that the kernel cannot flush dirty
* pages or request new data or unlock existing
* data.
* memory_object_destroy:
* The memory manager has requested that the
* kernel relinquish rights to the memory object
* port. [The memory manager may not want to
* destroy the port, but may wish to refuse or
* tear down existing memory mappings.]
* Each routine that breaks an association must break all of
* them at once. At some later time, that routine must clear
* the vm_object port fields and release the port rights.
* [Furthermore, each routine must cope with the simultaneous
* or previous operations of the others.]
*
* In addition to the lock on the object, the vm_object_cache_lock
* governs the port associations. References gained through the
* port association require use of the cache lock.
*
* Because the port fields may be cleared spontaneously, they
* cannot be used to determine whether a memory object has
* ever been associated with a particular vm_object. [This
* knowledge is important to the shadow object mechanism.]
* For this reason, an additional "created" attribute is
* provided.
*
* During various paging operations, the port values found in the
* vm_object must be valid. To prevent these port rights from being
* released, and to prevent the port associations from changing
* (other than being removed, i.e., made null), routines may use
* the vm_object_paging_begin/end routines [actually, macros].
* The implementation uses the "paging_in_progress" and "wanted" fields.
* [Operations that alter the validity of the port values include the
* termination routines and vm_object_collapse.]
*/
vm_object_t vm_object_lookup(
ipc_port_t port)
{
vm_object_t object = VM_OBJECT_NULL;
if (IP_VALID(port)) {
ip_lock(port);
if (ip_active(port) &&
(ip_kotype(port) == IKOT_PAGING_REQUEST)) {
vm_object_cache_lock();
object = (vm_object_t) port->ip_kobject;
vm_object_lock(object);
assert(object->alive);
if (object->ref_count == 0) {
queue_remove(&vm_object_cached_list, object,
vm_object_t, cached_list);
vm_object_cached_count--;
vm_object_cached_pages_update(-object->resident_page_count);
}
object->ref_count++;
vm_object_unlock(object);
vm_object_cache_unlock();
}
ip_unlock(port);
}
return object;
}
vm_object_t vm_object_lookup_name(
ipc_port_t port)
{
vm_object_t object = VM_OBJECT_NULL;
if (IP_VALID(port)) {
ip_lock(port);
if (ip_active(port) &&
(ip_kotype(port) == IKOT_PAGING_NAME)) {
vm_object_cache_lock();
object = (vm_object_t) port->ip_kobject;
vm_object_lock(object);
assert(object->alive);
if (object->ref_count == 0) {
queue_remove(&vm_object_cached_list, object,
vm_object_t, cached_list);
vm_object_cached_count--;
vm_object_cached_pages_update(-object->resident_page_count);
}
object->ref_count++;
vm_object_unlock(object);
vm_object_cache_unlock();
}
ip_unlock(port);
}
return object;
}
void vm_object_destroy(
ipc_port_t pager)
{
vm_object_t object;
pager_request_t old_request;
ipc_port_t old_name;
/*
* Perform essentially the same operations as in vm_object_lookup,
* except that this time we look up based on the memory_object
* port, not the control port.
*/
vm_object_cache_lock();
if (ip_kotype(pager) != IKOT_PAGER) {
vm_object_cache_unlock();
return;
}
object = (vm_object_t) pager->ip_kobject;
vm_object_lock(object);
if (object->ref_count == 0) {
queue_remove(&vm_object_cached_list, object,
vm_object_t, cached_list);
vm_object_cached_count--;
vm_object_cached_pages_update(-object->resident_page_count);
}
object->ref_count++;
object->can_persist = FALSE;
assert(object->pager == pager);
/*
* Remove the port associations.
*
* Note that the memory_object itself is dead, so
* we don't bother with it.
*/
object->pager = IP_NULL;
vm_object_remove(object);
old_request = object->pager_request;
object->pager_request = PAGER_REQUEST_NULL;
old_name = object->pager_name;
object->pager_name = IP_NULL;
vm_object_unlock(object);
vm_object_cache_unlock();
/*
* Clean up the port references. Note that there's no
* point in trying the memory_object_terminate call
* because the memory_object itself is dead.
*/
ipc_port_release_send(pager);
if (old_request != IP_NULL)
ipc_port_dealloc_kernel(old_request);
if (old_name != IP_NULL)
ipc_port_dealloc_kernel(old_name);
/*
* Restart pending page requests
*/
vm_object_abort_activity(object);
/*
* Lose the object reference.
*/
vm_object_deallocate(object);
}
boolean_t vm_object_accept_old_init_protocol = FALSE;
/*
* Routine: vm_object_enter
* Purpose:
* Find a VM object corresponding to the given
* pager; if no such object exists, create one,
* and initialize the pager.
*/
vm_object_t vm_object_enter(
ipc_port_t pager,
vm_size_t size,
boolean_t internal)
{
register
vm_object_t object;
vm_object_t new_object;
boolean_t must_init;
ipc_kobject_type_t po;
restart:
if (!IP_VALID(pager))
return vm_object_allocate(size);
new_object = VM_OBJECT_NULL;
must_init = FALSE;
/*
* Look for an object associated with this port.
*/
vm_object_cache_lock();
for (;;) {
po = ip_kotype(pager);
/*
* If a previous object is being terminated,
* we must wait for the termination message
* to be queued.
*
* We set kobject to a non-null value to let the
* terminator know that someone is waiting.
* Among the possibilities is that the port
* could die while we're waiting. Must restart
* instead of continuing the loop.
*/
if (po == IKOT_PAGER_TERMINATING) {
pager->ip_kobject = (ipc_kobject_t) pager;
assert_wait((event_t) pager, FALSE);
vm_object_cache_unlock();
thread_block((void (*)()) 0);
goto restart;
}
/*
* Bail if there is already a kobject associated
* with the pager port.
*/
if (po != IKOT_NONE) {
break;
}
/*
* We must unlock to create a new object;
* if we do so, we must try the lookup again.
*/
if (new_object == VM_OBJECT_NULL) {
vm_object_cache_unlock();
new_object = vm_object_allocate(size);
vm_object_cache_lock();
} else {
/*
* Lookup failed twice, and we have something
* to insert; set the object.
*/
ipc_kobject_set(pager,
(ipc_kobject_t) new_object,
IKOT_PAGER);
new_object = VM_OBJECT_NULL;
must_init = TRUE;
}
}
if (internal)
must_init = TRUE;
/*
* It's only good if it's a VM object!
*/
object = (po == IKOT_PAGER) ? (vm_object_t) pager->ip_kobject
: VM_OBJECT_NULL;
if ((object != VM_OBJECT_NULL) && !must_init) {
vm_object_lock(object);
if (object->ref_count == 0) {
queue_remove(&vm_object_cached_list, object,
vm_object_t, cached_list);
vm_object_cached_count--;
vm_object_cached_pages_update(-object->resident_page_count);
}
object->ref_count++;
vm_object_unlock(object);
vm_stat.hits++;
}
assert((object == VM_OBJECT_NULL) || (object->ref_count > 0) ||
((object->paging_in_progress != 0) && internal));
vm_stat.lookups++;
vm_object_cache_unlock();
/*
* If we raced to create a vm_object but lost, let's
* throw away ours.
*/
if (new_object != VM_OBJECT_NULL)
vm_object_deallocate(new_object);
if (object == VM_OBJECT_NULL)
return(object);
if (must_init) {
/*
* Copy the naked send right we were given.
*/
pager = ipc_port_copy_send(pager);
if (!IP_VALID(pager))
panic("vm_object_enter: port died"); /* XXX */
object->pager_created = TRUE;
object->pager = pager;
/*
* Allocate request port.
*/
object->pager_request = ipc_port_alloc_kernel();
if (object->pager_request == IP_NULL)
panic("vm_object_enter: pager request alloc");
ipc_kobject_set(object->pager_request,
(ipc_kobject_t) object,
IKOT_PAGING_REQUEST);
/*
* Let the pager know we're using it.
*/
if (internal) {
/* acquire a naked send right for the DMM */
ipc_port_t DMM = memory_manager_default_reference();
/* mark the object internal */
object->internal = TRUE;
assert(object->temporary);
/* default-pager objects are ready immediately */
object->pager_ready = TRUE;
/* consumes the naked send right for DMM */
(void) memory_object_create(DMM,
pager,
object->size,
object->pager_request,
object->pager_name,
PAGE_SIZE);
} else {
/* the object is external and not temporary */
object->internal = FALSE;
object->temporary = FALSE;
/* user pager objects are not ready until marked so */
object->pager_ready = FALSE;
(void) memory_object_init(pager,
object->pager_request,
object->pager_name,
PAGE_SIZE);
}
vm_object_lock(object);
object->pager_initialized = TRUE;
if (vm_object_accept_old_init_protocol)
object->pager_ready = TRUE;
vm_object_wakeup(object, VM_OBJECT_EVENT_INITIALIZED);
} else {
vm_object_lock(object);
}
/*
* [At this point, the object must be locked]
*/
/*
* Wait for the work above to be done by the first
* thread to map this object.
*/
while (!object->pager_initialized) {
vm_object_wait( object,
VM_OBJECT_EVENT_INITIALIZED,
FALSE);
vm_object_lock(object);
}
vm_object_unlock(object);
return object;
}
/*
* Routine: vm_object_pager_create
* Purpose:
* Create a memory object for an internal object.
* In/out conditions:
* The object is locked on entry and exit;
* it may be unlocked within this call.
* Limitations:
* Only one thread may be performing a
* vm_object_pager_create on an object at
* a time. Presumably, only the pageout
* daemon will be using this routine.
*/
void vm_object_pager_create(
register
vm_object_t object)
{
ipc_port_t pager;
if (object->pager_created) {
/*
* Someone else got to it first...
* wait for them to finish initializing
*/
while (!object->pager_initialized) {
vm_object_wait( object,
VM_OBJECT_EVENT_PAGER_READY,
FALSE);
vm_object_lock(object);
}
return;
}
/*
* Indicate that a memory object has been assigned
* before dropping the lock, to prevent a race.
*/
object->pager_created = TRUE;
/*
* Prevent collapse or termination by
* holding a paging reference
*/
vm_object_paging_begin(object);
vm_object_unlock(object);
#if MACH_PAGEMAP
object->existence_info = vm_external_create(
object->size +
object->paging_offset);
assert((object->size + object->paging_offset) >=
object->size);
#endif /* MACH_PAGEMAP */
/*
* Create the pager, and associate with it
* this object.
*
* Note that we only make the port association
* so that vm_object_enter can properly look up
* the object to complete the initialization...
* we do not expect any user to ever map this
* object.
*
* Since the kernel has the only rights to the
* port, it's safe to install the association
* without holding the cache lock.
*/
pager = ipc_port_alloc_kernel();
if (pager == IP_NULL)
panic("vm_object_pager_create: allocate pager port");
(void) ipc_port_make_send(pager);
ipc_kobject_set(pager, (ipc_kobject_t) object, IKOT_PAGER);
/*
* Initialize the rest of the paging stuff
*/
if (vm_object_enter(pager, object->size, TRUE) != object)
panic("vm_object_pager_create: mismatch");
/*
* Drop the naked send right taken above.
*/
ipc_port_release_send(pager);
/*
* Release the paging reference
*/
vm_object_lock(object);
vm_object_paging_end(object);
}
/*
* Routine: vm_object_remove
* Purpose:
* Eliminate the pager/object association
* for this pager.
* Conditions:
* The object cache must be locked.
*/
void vm_object_remove(
vm_object_t object)
{
ipc_port_t port;
if ((port = object->pager) != IP_NULL) {
if (ip_kotype(port) == IKOT_PAGER)
ipc_kobject_set(port, IKO_NULL,
IKOT_PAGER_TERMINATING);
else if (ip_kotype(port) != IKOT_NONE)
panic("vm_object_remove: bad object port");
}
if ((port = object->pager_request) != IP_NULL) {
if (ip_kotype(port) == IKOT_PAGING_REQUEST)
ipc_kobject_set(port, IKO_NULL, IKOT_NONE);
else if (ip_kotype(port) != IKOT_NONE)
panic("vm_object_remove: bad request port");
}
if ((port = object->pager_name) != IP_NULL) {
if (ip_kotype(port) == IKOT_PAGING_NAME)
ipc_kobject_set(port, IKO_NULL, IKOT_NONE);
else if (ip_kotype(port) != IKOT_NONE)
panic("vm_object_remove: bad name port");
}
}
/*
* Global variables for vm_object_collapse():
*
* Counts for normal collapses and bypasses.
* Debugging variables, to watch or disable collapse.
*/
long object_collapses = 0;
long object_bypasses = 0;
int vm_object_collapse_debug = 0;
boolean_t vm_object_collapse_allowed = TRUE;
boolean_t vm_object_collapse_bypass_allowed = TRUE;
/*
* vm_object_collapse:
*
* Collapse an object with the object backing it.
* Pages in the backing object are moved into the
* parent, and the backing object is deallocated.
*
* Requires that the object be locked and the page
* queues be unlocked. May unlock/relock the object,
* so the caller should hold a reference for the object.
*/
void vm_object_collapse(
register vm_object_t object)
{
register vm_object_t backing_object;
register vm_offset_t backing_offset;
register vm_size_t size;
register vm_offset_t new_offset;
register vm_page_t p, pp;
ipc_port_t old_name_port;
if (!vm_object_collapse_allowed)
return;
while (TRUE) {
/*
* Verify that the conditions are right for collapse:
*
* The object exists and no pages in it are currently
* being paged out (or have ever been paged out).
*
* This check is probably overkill -- if a memory
* object has not been created, the fault handler
* shouldn't release the object lock while paging
* is in progress or absent pages exist.
*/
if (object == VM_OBJECT_NULL ||
object->pager_created ||
object->paging_in_progress != 0 ||
object->absent_count != 0)
return;
/*
* There is a backing object, and
*/
if ((backing_object = object->shadow) == VM_OBJECT_NULL)
return;
vm_object_lock(backing_object);
/*
* ...
* The backing object is not read_only,
* and no pages in the backing object are
* currently being paged out.
* The backing object is internal.
*
* XXX It may be sufficient for the backing
* XXX object to be temporary.
*/
if (!backing_object->internal ||
backing_object->paging_in_progress != 0) {
vm_object_unlock(backing_object);
return;
}
/*
* The backing object can't be a copy-object:
* the shadow_offset for the copy-object must stay
* as 0. Furthermore (for the 'we have all the
* pages' case), if we bypass backing_object and
* just shadow the next object in the chain, old
* pages from that object would then have to be copied
* BOTH into the (former) backing_object and into the
* parent object.
*/
if (backing_object->shadow != VM_OBJECT_NULL &&
backing_object->shadow->copy != VM_OBJECT_NULL) {
vm_object_unlock(backing_object);
return;
}
/*
* We know that we can either collapse the backing
* object (if the parent is the only reference to
* it) or (perhaps) remove the parent's reference
* to it.
*/
backing_offset = object->shadow_offset;
size = object->size;
/*
* If there is exactly one reference to the backing
* object, we can collapse it into the parent.
*/
if (backing_object->ref_count == 1) {
if (!vm_object_cache_lock_try()) {
vm_object_unlock(backing_object);
return;
}
/*
* We can collapse the backing object.
*
* Move all in-memory pages from backing_object
* to the parent. Pages that have been paged out
* will be overwritten by any of the parent's
* pages that shadow them.
*/
while (!queue_empty(&backing_object->memq)) {
p = (vm_page_t)
queue_first(&backing_object->memq);
new_offset = (p->offset - backing_offset);
assert(!p->busy || p->absent);
/*
* If the parent has a page here, or if
* this page falls outside the parent,
* dispose of it.
*
* Otherwise, move it as planned.
*/
if (p->offset < backing_offset ||
new_offset >= size) {
VM_PAGE_FREE(p);
} else {
pp = vm_page_lookup(object, new_offset);
if (pp != VM_PAGE_NULL && !pp->absent) {
/*
* Parent object has a real page.
* Throw away the backing object's
* page.
*/
VM_PAGE_FREE(p);
}
else {
if (pp != VM_PAGE_NULL) {
/*
* Parent has an absent page...
* it's not being paged in, so
* it must really be missing from
* the parent.
*
* Throw out the absent page...
* any faults looking for that
* page will restart with the new
* one.
*/
/*
* This should never happen -- the
* parent cannot have ever had an
* external memory object, and thus
* cannot have absent pages.
*/
panic("vm_object_collapse: bad case");
VM_PAGE_FREE(pp);
/*
* Fall through to move the backing
* object's page up.
*/
}
/*
* Parent now has no page.
* Move the backing object's page up.
*/
vm_page_rename(p, object, new_offset);
}
}
}
/*
* Move the pager from backing_object to object.
*
* XXX We're only using part of the paging space
* for keeps now... we ought to discard the
* unused portion.
*/
switch (vm_object_collapse_debug) {
case 0:
break;
case 1:
if ((backing_object->pager == IP_NULL) &&
(backing_object->pager_request ==
PAGER_REQUEST_NULL))
break;
/* Fall through to... */
default:
printf("vm_object_collapse: %p (pager %p, request %p) up to %p\n",
backing_object, backing_object->pager, backing_object->pager_request,
object);
if (vm_object_collapse_debug > 2)
SoftDebugger("vm_object_collapse");
}
object->pager = backing_object->pager;
if (object->pager != IP_NULL)
ipc_kobject_set(object->pager,
(ipc_kobject_t) object,
IKOT_PAGER);
object->pager_initialized = backing_object->pager_initialized;
object->pager_ready = backing_object->pager_ready;
object->pager_created = backing_object->pager_created;
object->pager_request = backing_object->pager_request;
if (object->pager_request != IP_NULL)
ipc_kobject_set(object->pager_request,
(ipc_kobject_t) object,
IKOT_PAGING_REQUEST);
old_name_port = object->pager_name;
if (old_name_port != IP_NULL)
ipc_kobject_set(old_name_port,
IKO_NULL, IKOT_NONE);
object->pager_name = backing_object->pager_name;
if (object->pager_name != IP_NULL)
ipc_kobject_set(object->pager_name,
(ipc_kobject_t) object,
IKOT_PAGING_NAME);
vm_object_cache_unlock();
/*
* If there is no pager, leave paging-offset alone.
*/
if (object->pager != IP_NULL)
object->paging_offset =
backing_object->paging_offset +
backing_offset;
#if MACH_PAGEMAP
assert(object->existence_info == VM_EXTERNAL_NULL);
object->existence_info = backing_object->existence_info;
#endif /* MACH_PAGEMAP */
/*
* Object now shadows whatever backing_object did.
* Note that the reference to backing_object->shadow
* moves from within backing_object to within object.
*/
object->shadow = backing_object->shadow;
object->shadow_offset += backing_object->shadow_offset;
if (object->shadow != VM_OBJECT_NULL &&
object->shadow->copy != VM_OBJECT_NULL) {
panic("vm_object_collapse: we collapsed a copy-object!");
}
/*
* Discard backing_object.
*
* Since the backing object has no pages, no
* pager left, and no object references within it,
* all that is necessary is to dispose of it.
*/
assert(
(backing_object->ref_count == 1) &&
(backing_object->resident_page_count == 0) &&
(backing_object->paging_in_progress == 0)
);
assert(backing_object->alive);
backing_object->alive = FALSE;
vm_object_unlock(backing_object);
vm_object_unlock(object);
if (old_name_port != IP_NULL)
ipc_port_dealloc_kernel(old_name_port);
kmem_cache_free(&vm_object_cache, (vm_offset_t) backing_object);
vm_object_lock(object);
object_collapses++;
}
else {
if (!vm_object_collapse_bypass_allowed) {
vm_object_unlock(backing_object);
return;
}
/*
* If all of the pages in the backing object are
* shadowed by the parent object, the parent
* object no longer has to shadow the backing
* object; it can shadow the next one in the
* chain.
*
* The backing object must not be paged out - we'd
* have to check all of the paged-out pages, as
* well.
*/
if (backing_object->pager_created) {
vm_object_unlock(backing_object);
return;
}
/*
* Should have a check for a 'small' number
* of pages here.
*/
queue_iterate(&backing_object->memq, p,
vm_page_t, listq)
{
new_offset = (p->offset - backing_offset);
/*
* If the parent has a page here, or if
* this page falls outside the parent,
* keep going.
*
* Otherwise, the backing_object must be
* left in the chain.
*/
if (p->offset >= backing_offset &&
new_offset <= size &&
(pp = vm_page_lookup(object, new_offset))
== VM_PAGE_NULL) {
/*
* Page still needed.
* Can't go any further.
*/
vm_object_unlock(backing_object);
return;
}
}
/*
* Make the parent shadow the next object
* in the chain. Deallocating backing_object
* will not remove it, since its reference
* count is at least 2.
*/
vm_object_reference(object->shadow = backing_object->shadow);
object->shadow_offset += backing_object->shadow_offset;
/*
* Backing object might have had a copy pointer
* to us. If it did, clear it.
*/
if (backing_object->copy == object)
backing_object->copy = VM_OBJECT_NULL;
/*
* Drop the reference count on backing_object.
* Since its ref_count was at least 2, it
* will not vanish; so we don't need to call
* vm_object_deallocate.
*/
backing_object->ref_count--;
assert(backing_object->ref_count > 0);
vm_object_unlock(backing_object);
object_bypasses ++;
}
/*
* Try again with this object's new backing object.
*/
}
}
/*
* Routine: vm_object_page_remove: [internal]
* Purpose:
* Removes all physical pages in the specified
* object range from the object's list of pages.
*
* In/out conditions:
* The object must be locked.
*/
unsigned int vm_object_page_remove_lookup = 0;
unsigned int vm_object_page_remove_iterate = 0;
void vm_object_page_remove(
register vm_object_t object,
register vm_offset_t start,
register vm_offset_t end)
{
register vm_page_t p, next;
/*
* One and two page removals are most popular.
* The factor of 16 here is somewhat arbitrary.
* It balances vm_object_lookup vs iteration.
*/
if (atop(end - start) < (unsigned)object->resident_page_count/16) {
vm_object_page_remove_lookup++;
for (; start < end; start += PAGE_SIZE) {
p = vm_page_lookup(object, start);
if (p != VM_PAGE_NULL) {
if (!p->fictitious)
pmap_page_protect(p->phys_addr,
VM_PROT_NONE);
VM_PAGE_FREE(p);
}
}
} else {
vm_object_page_remove_iterate++;
p = (vm_page_t) queue_first(&object->memq);
while (!queue_end(&object->memq, (queue_entry_t) p)) {
next = (vm_page_t) queue_next(&p->listq);
if ((start <= p->offset) && (p->offset < end)) {
if (!p->fictitious)
pmap_page_protect(p->phys_addr,
VM_PROT_NONE);
VM_PAGE_FREE(p);
}
p = next;
}
}
}
/*
* Routine: vm_object_coalesce
* Function: Coalesces two objects backing up adjoining
* regions of memory into a single object.
*
* returns TRUE if objects were combined.
*
* NOTE: Only works at the moment if the second object is NULL -
* if it's not, which object do we lock first?
*
* Parameters:
* prev_object First object to coalesce
* prev_offset Offset into prev_object
* next_object Second object into coalesce
* next_offset Offset into next_object
*
* prev_size Size of reference to prev_object
* next_size Size of reference to next_object
*
* Conditions:
* The object must *not* be locked.
*/
boolean_t vm_object_coalesce(
register vm_object_t prev_object,
vm_object_t next_object,
vm_offset_t prev_offset,
vm_offset_t next_offset,
vm_size_t prev_size,
vm_size_t next_size)
{
vm_size_t newsize;
if (next_object != VM_OBJECT_NULL) {
return FALSE;
}
if (prev_object == VM_OBJECT_NULL) {
return TRUE;
}
vm_object_lock(prev_object);
/*
* Try to collapse the object first
*/
vm_object_collapse(prev_object);
/*
* Can't coalesce if pages not mapped to
* prev_entry may be in use anyway:
* . more than one reference
* . paged out
* . shadows another object
* . has a copy elsewhere
* . paging references (pages might be in page-list)
*/
if ((prev_object->ref_count > 1) ||
prev_object->pager_created ||
(prev_object->shadow != VM_OBJECT_NULL) ||
(prev_object->copy != VM_OBJECT_NULL) ||
(prev_object->paging_in_progress != 0)) {
vm_object_unlock(prev_object);
return FALSE;
}
/*
* Remove any pages that may still be in the object from
* a previous deallocation.
*/
vm_object_page_remove(prev_object,
prev_offset + prev_size,
prev_offset + prev_size + next_size);
/*
* Extend the object if necessary.
*/
newsize = prev_offset + prev_size + next_size;
if (newsize > prev_object->size)
prev_object->size = newsize;
vm_object_unlock(prev_object);
return TRUE;
}
vm_object_t vm_object_request_object(
ipc_port_t p)
{
return vm_object_lookup(p);
}
/*
* Routine: vm_object_name
* Purpose:
* Returns a naked send right to the "name" port associated
* with this object.
*/
ipc_port_t vm_object_name(
vm_object_t object)
{
ipc_port_t p;
if (object == VM_OBJECT_NULL)
return IP_NULL;
vm_object_lock(object);
while (object->shadow != VM_OBJECT_NULL) {
vm_object_t new_object = object->shadow;
vm_object_lock(new_object);
vm_object_unlock(object);
object = new_object;
}
p = object->pager_name;
if (p != IP_NULL)
p = ipc_port_make_send(p);
vm_object_unlock(object);
return p;
}
/*
* Attach a set of physical pages to an object, so that they can
* be mapped by mapping the object. Typically used to map IO memory.
*
* The mapping function and its private data are used to obtain the
* physical addresses for each page to be mapped.
*/
void
vm_object_page_map(
vm_object_t object,
vm_offset_t offset,
vm_size_t size,
vm_offset_t (*map_fn)(void *, vm_offset_t),
void * map_fn_data) /* private to map_fn */
{
int num_pages;
int i;
vm_page_t m;
vm_page_t old_page;
vm_offset_t addr;
num_pages = atop(size);
for (i = 0; i < num_pages; i++, offset += PAGE_SIZE) {
addr = (*map_fn)(map_fn_data, offset);
while ((m = vm_page_grab_fictitious()) == VM_PAGE_NULL)
vm_page_more_fictitious();
vm_object_lock(object);
if ((old_page = vm_page_lookup(object, offset))
!= VM_PAGE_NULL)
{
VM_PAGE_FREE(old_page);
}
vm_page_init(m, addr);
m->private = TRUE; /* don`t free page */
m->wire_count = 1;
vm_page_lock_queues();
vm_page_insert(m, object, offset);
vm_page_unlock_queues();
PAGE_WAKEUP_DONE(m);
vm_object_unlock(object);
}
}
#if MACH_KDB
#include <vm/vm_print.h>
#define printf kdbprintf
boolean_t vm_object_print_pages = FALSE;
/*
* vm_object_print: [ debug ]
*/
void vm_object_print(
vm_object_t object)
{
register vm_page_t p;
register int count;
if (object == VM_OBJECT_NULL)
return;
iprintf("Object 0x%X: size=0x%X",
(vm_offset_t) object, (vm_offset_t) object->size);
printf(", %d references, %d resident pages,", object->ref_count,
object->resident_page_count);
printf(" %d absent pages,", object->absent_count);
printf(" %d paging ops\n", object->paging_in_progress);
indent += 2;
iprintf("memory object=0x%X (offset=0x%X),",
(vm_offset_t) object->pager, (vm_offset_t) object->paging_offset);
printf("control=0x%X, name=0x%X\n",
(vm_offset_t) object->pager_request, (vm_offset_t) object->pager_name);
iprintf("%s%s",
object->pager_ready ? " ready" : "",
object->pager_created ? " created" : "");
printf("%s,%s ",
object->pager_initialized ? "" : "uninitialized",
object->temporary ? "temporary" : "permanent");
printf("%s%s,",
object->internal ? "internal" : "external",
object->can_persist ? " cacheable" : "");
printf("copy_strategy=%d\n", (vm_offset_t)object->copy_strategy);
iprintf("shadow=0x%X (offset=0x%X),",
(vm_offset_t) object->shadow, (vm_offset_t) object->shadow_offset);
printf("copy=0x%X\n", (vm_offset_t) object->copy);
indent += 2;
if (vm_object_print_pages) {
count = 0;
p = (vm_page_t) queue_first(&object->memq);
while (!queue_end(&object->memq, (queue_entry_t) p)) {
if (count == 0) iprintf("memory:=");
else if (count == 4) {printf("\n"); iprintf(" ..."); count = 0;}
else printf(",");
count++;
printf("(off=0x%X,page=0x%X)", p->offset, (vm_offset_t) p);
p = (vm_page_t) queue_next(&p->listq);
}
if (count != 0)
printf("\n");
}
indent -= 4;
}
#endif /* MACH_KDB */
|