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
3024
3025
3026
3027
3028
3029
3030
3031
3032
3033
3034
3035
3036
3037
3038
3039
3040
3041
3042
3043
3044
3045
3046
3047
3048
3049
3050
3051
3052
3053
3054
3055
3056
3057
3058
3059
3060
3061
3062
3063
3064
3065
3066
3067
3068
3069
3070
3071
3072
3073
3074
3075
3076
3077
3078
3079
3080
3081
3082
3083
3084
3085
3086
3087
3088
3089
3090
3091
3092
3093
3094
3095
3096
3097
3098
3099
3100
3101
3102
3103
3104
3105
3106
3107
3108
3109
3110
3111
3112
3113
3114
3115
3116
3117
3118
3119
3120
3121
3122
3123
3124
3125
3126
3127
3128
3129
3130
3131
3132
3133
3134
3135
3136
3137
3138
3139
3140
3141
3142
3143
3144
3145
3146
3147
3148
3149
3150
3151
3152
3153
3154
3155
3156
3157
3158
3159
3160
3161
3162
3163
3164
3165
3166
3167
3168
3169
3170
3171
3172
3173
3174
3175
3176
3177
3178
3179
3180
3181
3182
3183
3184
3185
3186
3187
3188
3189
3190
3191
3192
3193
3194
3195
3196
3197
3198
3199
3200
3201
3202
3203
3204
3205
3206
3207
3208
3209
3210
3211
3212
3213
3214
3215
3216
3217
3218
3219
3220
3221
3222
3223
3224
3225
3226
3227
3228
3229
3230
3231
3232
3233
3234
3235
3236
3237
3238
3239
3240
3241
3242
3243
3244
3245
3246
3247
3248
3249
3250
3251
3252
3253
3254
3255
3256
3257
3258
3259
3260
3261
3262
3263
3264
3265
3266
3267
3268
3269
3270
3271
3272
3273
3274
3275
3276
3277
3278
3279
3280
3281
3282
3283
3284
3285
3286
3287
3288
3289
3290
3291
3292
3293
3294
3295
3296
3297
3298
3299
3300
3301
3302
3303
3304
3305
3306
3307
3308
3309
3310
3311
3312
3313
3314
3315
3316
3317
3318
3319
|
/*-
* Copyright (c) 2002-2005, 2009 Jeffrey Roberson <jeff@FreeBSD.org>
* Copyright (c) 2004, 2005 Bosko Milekic <bmilekic@FreeBSD.org>
* Copyright (c) 2004-2006 Robert N. M. Watson
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice unmodified, 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.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``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 AUTHOR 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.
*/
/*
* uma_core.c Implementation of the Universal Memory allocator
*
* This allocator is intended to replace the multitude of similar object caches
* in the standard FreeBSD kernel. The intent is to be flexible as well as
* effecient. A primary design goal is to return unused memory to the rest of
* the system. This will make the system as a whole more flexible due to the
* ability to move memory to subsystems which most need it instead of leaving
* pools of reserved memory unused.
*
* The basic ideas stem from similar slab/zone based allocators whose algorithms
* are well known.
*
*/
/*
* TODO:
* - Improve memory usage for large allocations
* - Investigate cache size adjustments
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
/* I should really use ktr.. */
/*
#define UMA_DEBUG 1
#define UMA_DEBUG_ALLOC 1
#define UMA_DEBUG_ALLOC_1 1
*/
#include "opt_ddb.h"
#include "opt_param.h"
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/kernel.h>
#include <sys/types.h>
#include <sys/queue.h>
#include <sys/malloc.h>
#include <sys/ktr.h>
#include <sys/lock.h>
#include <sys/sysctl.h>
#include <sys/mutex.h>
#include <sys/proc.h>
#include <sys/sbuf.h>
#include <sys/smp.h>
#include <sys/vmmeter.h>
#include <vm/vm.h>
#include <vm/vm_object.h>
#include <vm/vm_page.h>
#include <vm/vm_param.h>
#include <vm/vm_map.h>
#include <vm/vm_kern.h>
#include <vm/vm_extern.h>
#include <vm/uma.h>
#include <vm/uma_int.h>
#include <vm/uma_dbg.h>
#include <machine/vmparam.h>
#include <ddb/ddb.h>
/*
* This is the zone and keg from which all zones are spawned. The idea is that
* even the zone & keg heads are allocated from the allocator, so we use the
* bss section to bootstrap us.
*/
static struct uma_keg masterkeg;
static struct uma_zone masterzone_k;
static struct uma_zone masterzone_z;
static uma_zone_t kegs = &masterzone_k;
static uma_zone_t zones = &masterzone_z;
/* This is the zone from which all of uma_slab_t's are allocated. */
static uma_zone_t slabzone;
static uma_zone_t slabrefzone; /* With refcounters (for UMA_ZONE_REFCNT) */
/*
* The initial hash tables come out of this zone so they can be allocated
* prior to malloc coming up.
*/
static uma_zone_t hashzone;
/* The boot-time adjusted value for cache line alignment. */
static int uma_align_cache = 64 - 1;
static MALLOC_DEFINE(M_UMAHASH, "UMAHash", "UMA Hash Buckets");
/*
* Are we allowed to allocate buckets?
*/
static int bucketdisable = 1;
/* Linked list of all kegs in the system */
static LIST_HEAD(,uma_keg) uma_kegs = LIST_HEAD_INITIALIZER(&uma_kegs);
/* This mutex protects the keg list */
static struct mtx uma_mtx;
/* Linked list of boot time pages */
static LIST_HEAD(,uma_slab) uma_boot_pages =
LIST_HEAD_INITIALIZER(&uma_boot_pages);
/* This mutex protects the boot time pages list */
static struct mtx uma_boot_pages_mtx;
/* Is the VM done starting up? */
static int booted = 0;
/* Maximum number of allowed items-per-slab if the slab header is OFFPAGE */
static u_int uma_max_ipers;
static u_int uma_max_ipers_ref;
/*
* This is the handle used to schedule events that need to happen
* outside of the allocation fast path.
*/
static struct callout uma_callout;
#define UMA_TIMEOUT 20 /* Seconds for callout interval. */
/*
* This structure is passed as the zone ctor arg so that I don't have to create
* a special allocation function just for zones.
*/
struct uma_zctor_args {
char *name;
size_t size;
uma_ctor ctor;
uma_dtor dtor;
uma_init uminit;
uma_fini fini;
uma_keg_t keg;
int align;
u_int32_t flags;
};
struct uma_kctor_args {
uma_zone_t zone;
size_t size;
uma_init uminit;
uma_fini fini;
int align;
u_int32_t flags;
};
struct uma_bucket_zone {
uma_zone_t ubz_zone;
char *ubz_name;
int ubz_entries;
};
#define BUCKET_MAX 128
struct uma_bucket_zone bucket_zones[] = {
{ NULL, "16 Bucket", 16 },
{ NULL, "32 Bucket", 32 },
{ NULL, "64 Bucket", 64 },
{ NULL, "128 Bucket", 128 },
{ NULL, NULL, 0}
};
#define BUCKET_SHIFT 4
#define BUCKET_ZONES ((BUCKET_MAX >> BUCKET_SHIFT) + 1)
/*
* bucket_size[] maps requested bucket sizes to zones that allocate a bucket
* of approximately the right size.
*/
static uint8_t bucket_size[BUCKET_ZONES];
/*
* Flags and enumerations to be passed to internal functions.
*/
enum zfreeskip { SKIP_NONE, SKIP_DTOR, SKIP_FINI };
#define ZFREE_STATFAIL 0x00000001 /* Update zone failure statistic. */
#define ZFREE_STATFREE 0x00000002 /* Update zone free statistic. */
/* Prototypes.. */
static void *obj_alloc(uma_zone_t, int, u_int8_t *, int);
static void *page_alloc(uma_zone_t, int, u_int8_t *, int);
static void *startup_alloc(uma_zone_t, int, u_int8_t *, int);
static void page_free(void *, int, u_int8_t);
static uma_slab_t keg_alloc_slab(uma_keg_t, uma_zone_t, int);
static void cache_drain(uma_zone_t);
static void bucket_drain(uma_zone_t, uma_bucket_t);
static void bucket_cache_drain(uma_zone_t zone);
static int keg_ctor(void *, int, void *, int);
static void keg_dtor(void *, int, void *);
static int zone_ctor(void *, int, void *, int);
static void zone_dtor(void *, int, void *);
static int zero_init(void *, int, int);
static void keg_small_init(uma_keg_t keg);
static void keg_large_init(uma_keg_t keg);
static void zone_foreach(void (*zfunc)(uma_zone_t));
static void zone_timeout(uma_zone_t zone);
static int hash_alloc(struct uma_hash *);
static int hash_expand(struct uma_hash *, struct uma_hash *);
static void hash_free(struct uma_hash *hash);
static void uma_timeout(void *);
static void uma_startup3(void);
static void *zone_alloc_item(uma_zone_t, void *, int);
static void zone_free_item(uma_zone_t, void *, void *, enum zfreeskip,
int);
static void bucket_enable(void);
static void bucket_init(void);
static uma_bucket_t bucket_alloc(int, int);
static void bucket_free(uma_bucket_t);
static void bucket_zone_drain(void);
static int zone_alloc_bucket(uma_zone_t zone, int flags);
static uma_slab_t zone_fetch_slab(uma_zone_t zone, uma_keg_t last, int flags);
static uma_slab_t zone_fetch_slab_multi(uma_zone_t zone, uma_keg_t last, int flags);
static void *slab_alloc_item(uma_zone_t zone, uma_slab_t slab);
static uma_keg_t uma_kcreate(uma_zone_t zone, size_t size, uma_init uminit,
uma_fini fini, int align, u_int32_t flags);
static inline void zone_relock(uma_zone_t zone, uma_keg_t keg);
static inline void keg_relock(uma_keg_t keg, uma_zone_t zone);
void uma_print_zone(uma_zone_t);
void uma_print_stats(void);
static int sysctl_vm_zone_count(SYSCTL_HANDLER_ARGS);
static int sysctl_vm_zone_stats(SYSCTL_HANDLER_ARGS);
SYSINIT(uma_startup3, SI_SUB_VM_CONF, SI_ORDER_SECOND, uma_startup3, NULL);
SYSCTL_PROC(_vm, OID_AUTO, zone_count, CTLFLAG_RD|CTLTYPE_INT,
0, 0, sysctl_vm_zone_count, "I", "Number of UMA zones");
SYSCTL_PROC(_vm, OID_AUTO, zone_stats, CTLFLAG_RD|CTLTYPE_STRUCT,
0, 0, sysctl_vm_zone_stats, "s,struct uma_type_header", "Zone Stats");
/*
* This routine checks to see whether or not it's safe to enable buckets.
*/
static void
bucket_enable(void)
{
if (cnt.v_free_count < cnt.v_free_min)
bucketdisable = 1;
else
bucketdisable = 0;
}
/*
* Initialize bucket_zones, the array of zones of buckets of various sizes.
*
* For each zone, calculate the memory required for each bucket, consisting
* of the header and an array of pointers. Initialize bucket_size[] to point
* the range of appropriate bucket sizes at the zone.
*/
static void
bucket_init(void)
{
struct uma_bucket_zone *ubz;
int i;
int j;
for (i = 0, j = 0; bucket_zones[j].ubz_entries != 0; j++) {
int size;
ubz = &bucket_zones[j];
size = roundup(sizeof(struct uma_bucket), sizeof(void *));
size += sizeof(void *) * ubz->ubz_entries;
ubz->ubz_zone = uma_zcreate(ubz->ubz_name, size,
NULL, NULL, NULL, NULL, UMA_ALIGN_PTR,
UMA_ZFLAG_INTERNAL | UMA_ZFLAG_BUCKET);
for (; i <= ubz->ubz_entries; i += (1 << BUCKET_SHIFT))
bucket_size[i >> BUCKET_SHIFT] = j;
}
}
/*
* Given a desired number of entries for a bucket, return the zone from which
* to allocate the bucket.
*/
static struct uma_bucket_zone *
bucket_zone_lookup(int entries)
{
int idx;
idx = howmany(entries, 1 << BUCKET_SHIFT);
return (&bucket_zones[bucket_size[idx]]);
}
static uma_bucket_t
bucket_alloc(int entries, int bflags)
{
struct uma_bucket_zone *ubz;
uma_bucket_t bucket;
/*
* This is to stop us from allocating per cpu buckets while we're
* running out of vm.boot_pages. Otherwise, we would exhaust the
* boot pages. This also prevents us from allocating buckets in
* low memory situations.
*/
if (bucketdisable)
return (NULL);
ubz = bucket_zone_lookup(entries);
bucket = zone_alloc_item(ubz->ubz_zone, NULL, bflags);
if (bucket) {
#ifdef INVARIANTS
bzero(bucket->ub_bucket, sizeof(void *) * ubz->ubz_entries);
#endif
bucket->ub_cnt = 0;
bucket->ub_entries = ubz->ubz_entries;
}
return (bucket);
}
static void
bucket_free(uma_bucket_t bucket)
{
struct uma_bucket_zone *ubz;
ubz = bucket_zone_lookup(bucket->ub_entries);
zone_free_item(ubz->ubz_zone, bucket, NULL, SKIP_NONE,
ZFREE_STATFREE);
}
static void
bucket_zone_drain(void)
{
struct uma_bucket_zone *ubz;
for (ubz = &bucket_zones[0]; ubz->ubz_entries != 0; ubz++)
zone_drain(ubz->ubz_zone);
}
static inline uma_keg_t
zone_first_keg(uma_zone_t zone)
{
return (LIST_FIRST(&zone->uz_kegs)->kl_keg);
}
static void
zone_foreach_keg(uma_zone_t zone, void (*kegfn)(uma_keg_t))
{
uma_klink_t klink;
LIST_FOREACH(klink, &zone->uz_kegs, kl_link)
kegfn(klink->kl_keg);
}
/*
* Routine called by timeout which is used to fire off some time interval
* based calculations. (stats, hash size, etc.)
*
* Arguments:
* arg Unused
*
* Returns:
* Nothing
*/
static void
uma_timeout(void *unused)
{
bucket_enable();
zone_foreach(zone_timeout);
/* Reschedule this event */
callout_reset(&uma_callout, UMA_TIMEOUT * hz, uma_timeout, NULL);
}
/*
* Routine to perform timeout driven calculations. This expands the
* hashes and does per cpu statistics aggregation.
*
* Returns nothing.
*/
static void
keg_timeout(uma_keg_t keg)
{
KEG_LOCK(keg);
/*
* Expand the keg hash table.
*
* This is done if the number of slabs is larger than the hash size.
* What I'm trying to do here is completely reduce collisions. This
* may be a little aggressive. Should I allow for two collisions max?
*/
if (keg->uk_flags & UMA_ZONE_HASH &&
keg->uk_pages / keg->uk_ppera >= keg->uk_hash.uh_hashsize) {
struct uma_hash newhash;
struct uma_hash oldhash;
int ret;
/*
* This is so involved because allocating and freeing
* while the keg lock is held will lead to deadlock.
* I have to do everything in stages and check for
* races.
*/
newhash = keg->uk_hash;
KEG_UNLOCK(keg);
ret = hash_alloc(&newhash);
KEG_LOCK(keg);
if (ret) {
if (hash_expand(&keg->uk_hash, &newhash)) {
oldhash = keg->uk_hash;
keg->uk_hash = newhash;
} else
oldhash = newhash;
KEG_UNLOCK(keg);
hash_free(&oldhash);
KEG_LOCK(keg);
}
}
KEG_UNLOCK(keg);
}
static void
zone_timeout(uma_zone_t zone)
{
zone_foreach_keg(zone, &keg_timeout);
}
/*
* Allocate and zero fill the next sized hash table from the appropriate
* backing store.
*
* Arguments:
* hash A new hash structure with the old hash size in uh_hashsize
*
* Returns:
* 1 on sucess and 0 on failure.
*/
static int
hash_alloc(struct uma_hash *hash)
{
int oldsize;
int alloc;
oldsize = hash->uh_hashsize;
/* We're just going to go to a power of two greater */
if (oldsize) {
hash->uh_hashsize = oldsize * 2;
alloc = sizeof(hash->uh_slab_hash[0]) * hash->uh_hashsize;
hash->uh_slab_hash = (struct slabhead *)malloc(alloc,
M_UMAHASH, M_NOWAIT);
} else {
alloc = sizeof(hash->uh_slab_hash[0]) * UMA_HASH_SIZE_INIT;
hash->uh_slab_hash = zone_alloc_item(hashzone, NULL,
M_WAITOK);
hash->uh_hashsize = UMA_HASH_SIZE_INIT;
}
if (hash->uh_slab_hash) {
bzero(hash->uh_slab_hash, alloc);
hash->uh_hashmask = hash->uh_hashsize - 1;
return (1);
}
return (0);
}
/*
* Expands the hash table for HASH zones. This is done from zone_timeout
* to reduce collisions. This must not be done in the regular allocation
* path, otherwise, we can recurse on the vm while allocating pages.
*
* Arguments:
* oldhash The hash you want to expand
* newhash The hash structure for the new table
*
* Returns:
* Nothing
*
* Discussion:
*/
static int
hash_expand(struct uma_hash *oldhash, struct uma_hash *newhash)
{
uma_slab_t slab;
int hval;
int i;
if (!newhash->uh_slab_hash)
return (0);
if (oldhash->uh_hashsize >= newhash->uh_hashsize)
return (0);
/*
* I need to investigate hash algorithms for resizing without a
* full rehash.
*/
for (i = 0; i < oldhash->uh_hashsize; i++)
while (!SLIST_EMPTY(&oldhash->uh_slab_hash[i])) {
slab = SLIST_FIRST(&oldhash->uh_slab_hash[i]);
SLIST_REMOVE_HEAD(&oldhash->uh_slab_hash[i], us_hlink);
hval = UMA_HASH(newhash, slab->us_data);
SLIST_INSERT_HEAD(&newhash->uh_slab_hash[hval],
slab, us_hlink);
}
return (1);
}
/*
* Free the hash bucket to the appropriate backing store.
*
* Arguments:
* slab_hash The hash bucket we're freeing
* hashsize The number of entries in that hash bucket
*
* Returns:
* Nothing
*/
static void
hash_free(struct uma_hash *hash)
{
if (hash->uh_slab_hash == NULL)
return;
if (hash->uh_hashsize == UMA_HASH_SIZE_INIT)
zone_free_item(hashzone,
hash->uh_slab_hash, NULL, SKIP_NONE, ZFREE_STATFREE);
else
free(hash->uh_slab_hash, M_UMAHASH);
}
/*
* Frees all outstanding items in a bucket
*
* Arguments:
* zone The zone to free to, must be unlocked.
* bucket The free/alloc bucket with items, cpu queue must be locked.
*
* Returns:
* Nothing
*/
static void
bucket_drain(uma_zone_t zone, uma_bucket_t bucket)
{
void *item;
if (bucket == NULL)
return;
while (bucket->ub_cnt > 0) {
bucket->ub_cnt--;
item = bucket->ub_bucket[bucket->ub_cnt];
#ifdef INVARIANTS
bucket->ub_bucket[bucket->ub_cnt] = NULL;
KASSERT(item != NULL,
("bucket_drain: botched ptr, item is NULL"));
#endif
zone_free_item(zone, item, NULL, SKIP_DTOR, 0);
}
}
/*
* Drains the per cpu caches for a zone.
*
* NOTE: This may only be called while the zone is being turn down, and not
* during normal operation. This is necessary in order that we do not have
* to migrate CPUs to drain the per-CPU caches.
*
* Arguments:
* zone The zone to drain, must be unlocked.
*
* Returns:
* Nothing
*/
static void
cache_drain(uma_zone_t zone)
{
uma_cache_t cache;
int cpu;
/*
* XXX: It is safe to not lock the per-CPU caches, because we're
* tearing down the zone anyway. I.e., there will be no further use
* of the caches at this point.
*
* XXX: It would good to be able to assert that the zone is being
* torn down to prevent improper use of cache_drain().
*
* XXX: We lock the zone before passing into bucket_cache_drain() as
* it is used elsewhere. Should the tear-down path be made special
* there in some form?
*/
for (cpu = 0; cpu <= mp_maxid; cpu++) {
if (CPU_ABSENT(cpu))
continue;
cache = &zone->uz_cpu[cpu];
bucket_drain(zone, cache->uc_allocbucket);
bucket_drain(zone, cache->uc_freebucket);
if (cache->uc_allocbucket != NULL)
bucket_free(cache->uc_allocbucket);
if (cache->uc_freebucket != NULL)
bucket_free(cache->uc_freebucket);
cache->uc_allocbucket = cache->uc_freebucket = NULL;
}
ZONE_LOCK(zone);
bucket_cache_drain(zone);
ZONE_UNLOCK(zone);
}
/*
* Drain the cached buckets from a zone. Expects a locked zone on entry.
*/
static void
bucket_cache_drain(uma_zone_t zone)
{
uma_bucket_t bucket;
/*
* Drain the bucket queues and free the buckets, we just keep two per
* cpu (alloc/free).
*/
while ((bucket = LIST_FIRST(&zone->uz_full_bucket)) != NULL) {
LIST_REMOVE(bucket, ub_link);
ZONE_UNLOCK(zone);
bucket_drain(zone, bucket);
bucket_free(bucket);
ZONE_LOCK(zone);
}
/* Now we do the free queue.. */
while ((bucket = LIST_FIRST(&zone->uz_free_bucket)) != NULL) {
LIST_REMOVE(bucket, ub_link);
bucket_free(bucket);
}
}
/*
* Frees pages from a keg back to the system. This is done on demand from
* the pageout daemon.
*
* Returns nothing.
*/
static void
keg_drain(uma_keg_t keg)
{
struct slabhead freeslabs = { 0 };
uma_slab_t slab;
uma_slab_t n;
u_int8_t flags;
u_int8_t *mem;
int i;
/*
* We don't want to take pages from statically allocated kegs at this
* time
*/
if (keg->uk_flags & UMA_ZONE_NOFREE || keg->uk_freef == NULL)
return;
#ifdef UMA_DEBUG
printf("%s free items: %u\n", keg->uk_name, keg->uk_free);
#endif
KEG_LOCK(keg);
if (keg->uk_free == 0)
goto finished;
slab = LIST_FIRST(&keg->uk_free_slab);
while (slab) {
n = LIST_NEXT(slab, us_link);
/* We have no where to free these to */
if (slab->us_flags & UMA_SLAB_BOOT) {
slab = n;
continue;
}
LIST_REMOVE(slab, us_link);
keg->uk_pages -= keg->uk_ppera;
keg->uk_free -= keg->uk_ipers;
if (keg->uk_flags & UMA_ZONE_HASH)
UMA_HASH_REMOVE(&keg->uk_hash, slab, slab->us_data);
SLIST_INSERT_HEAD(&freeslabs, slab, us_hlink);
slab = n;
}
finished:
KEG_UNLOCK(keg);
while ((slab = SLIST_FIRST(&freeslabs)) != NULL) {
SLIST_REMOVE(&freeslabs, slab, uma_slab, us_hlink);
if (keg->uk_fini)
for (i = 0; i < keg->uk_ipers; i++)
keg->uk_fini(
slab->us_data + (keg->uk_rsize * i),
keg->uk_size);
flags = slab->us_flags;
mem = slab->us_data;
if (keg->uk_flags & UMA_ZONE_VTOSLAB) {
vm_object_t obj;
if (flags & UMA_SLAB_KMEM)
obj = kmem_object;
else if (flags & UMA_SLAB_KERNEL)
obj = kernel_object;
else
obj = NULL;
for (i = 0; i < keg->uk_ppera; i++)
vsetobj((vm_offset_t)mem + (i * PAGE_SIZE),
obj);
}
if (keg->uk_flags & UMA_ZONE_OFFPAGE)
zone_free_item(keg->uk_slabzone, slab, NULL,
SKIP_NONE, ZFREE_STATFREE);
#ifdef UMA_DEBUG
printf("%s: Returning %d bytes.\n",
keg->uk_name, UMA_SLAB_SIZE * keg->uk_ppera);
#endif
keg->uk_freef(mem, UMA_SLAB_SIZE * keg->uk_ppera, flags);
}
}
static void
zone_drain_wait(uma_zone_t zone, int waitok)
{
/*
* Set draining to interlock with zone_dtor() so we can release our
* locks as we go. Only dtor() should do a WAITOK call since it
* is the only call that knows the structure will still be available
* when it wakes up.
*/
ZONE_LOCK(zone);
while (zone->uz_flags & UMA_ZFLAG_DRAINING) {
if (waitok == M_NOWAIT)
goto out;
mtx_unlock(&uma_mtx);
msleep(zone, zone->uz_lock, PVM, "zonedrain", 1);
mtx_lock(&uma_mtx);
}
zone->uz_flags |= UMA_ZFLAG_DRAINING;
bucket_cache_drain(zone);
ZONE_UNLOCK(zone);
/*
* The DRAINING flag protects us from being freed while
* we're running. Normally the uma_mtx would protect us but we
* must be able to release and acquire the right lock for each keg.
*/
zone_foreach_keg(zone, &keg_drain);
ZONE_LOCK(zone);
zone->uz_flags &= ~UMA_ZFLAG_DRAINING;
wakeup(zone);
out:
ZONE_UNLOCK(zone);
}
void
zone_drain(uma_zone_t zone)
{
zone_drain_wait(zone, M_NOWAIT);
}
/*
* Allocate a new slab for a keg. This does not insert the slab onto a list.
*
* Arguments:
* wait Shall we wait?
*
* Returns:
* The slab that was allocated or NULL if there is no memory and the
* caller specified M_NOWAIT.
*/
static uma_slab_t
keg_alloc_slab(uma_keg_t keg, uma_zone_t zone, int wait)
{
uma_slabrefcnt_t slabref;
uma_alloc allocf;
uma_slab_t slab;
u_int8_t *mem;
u_int8_t flags;
int i;
mtx_assert(&keg->uk_lock, MA_OWNED);
slab = NULL;
#ifdef UMA_DEBUG
printf("slab_zalloc: Allocating a new slab for %s\n", keg->uk_name);
#endif
allocf = keg->uk_allocf;
KEG_UNLOCK(keg);
if (keg->uk_flags & UMA_ZONE_OFFPAGE) {
slab = zone_alloc_item(keg->uk_slabzone, NULL, wait);
if (slab == NULL) {
KEG_LOCK(keg);
return NULL;
}
}
/*
* This reproduces the old vm_zone behavior of zero filling pages the
* first time they are added to a zone.
*
* Malloced items are zeroed in uma_zalloc.
*/
if ((keg->uk_flags & UMA_ZONE_MALLOC) == 0)
wait |= M_ZERO;
else
wait &= ~M_ZERO;
/* zone is passed for legacy reasons. */
mem = allocf(zone, keg->uk_ppera * UMA_SLAB_SIZE, &flags, wait);
if (mem == NULL) {
if (keg->uk_flags & UMA_ZONE_OFFPAGE)
zone_free_item(keg->uk_slabzone, slab, NULL,
SKIP_NONE, ZFREE_STATFREE);
KEG_LOCK(keg);
return (NULL);
}
/* Point the slab into the allocated memory */
if (!(keg->uk_flags & UMA_ZONE_OFFPAGE))
slab = (uma_slab_t )(mem + keg->uk_pgoff);
if (keg->uk_flags & UMA_ZONE_VTOSLAB)
for (i = 0; i < keg->uk_ppera; i++)
vsetslab((vm_offset_t)mem + (i * PAGE_SIZE), slab);
slab->us_keg = keg;
slab->us_data = mem;
slab->us_freecount = keg->uk_ipers;
slab->us_firstfree = 0;
slab->us_flags = flags;
if (keg->uk_flags & UMA_ZONE_REFCNT) {
slabref = (uma_slabrefcnt_t)slab;
for (i = 0; i < keg->uk_ipers; i++) {
slabref->us_freelist[i].us_refcnt = 0;
slabref->us_freelist[i].us_item = i+1;
}
} else {
for (i = 0; i < keg->uk_ipers; i++)
slab->us_freelist[i].us_item = i+1;
}
if (keg->uk_init != NULL) {
for (i = 0; i < keg->uk_ipers; i++)
if (keg->uk_init(slab->us_data + (keg->uk_rsize * i),
keg->uk_size, wait) != 0)
break;
if (i != keg->uk_ipers) {
if (keg->uk_fini != NULL) {
for (i--; i > -1; i--)
keg->uk_fini(slab->us_data +
(keg->uk_rsize * i),
keg->uk_size);
}
if (keg->uk_flags & UMA_ZONE_VTOSLAB) {
vm_object_t obj;
if (flags & UMA_SLAB_KMEM)
obj = kmem_object;
else if (flags & UMA_SLAB_KERNEL)
obj = kernel_object;
else
obj = NULL;
for (i = 0; i < keg->uk_ppera; i++)
vsetobj((vm_offset_t)mem +
(i * PAGE_SIZE), obj);
}
if (keg->uk_flags & UMA_ZONE_OFFPAGE)
zone_free_item(keg->uk_slabzone, slab,
NULL, SKIP_NONE, ZFREE_STATFREE);
keg->uk_freef(mem, UMA_SLAB_SIZE * keg->uk_ppera,
flags);
KEG_LOCK(keg);
return (NULL);
}
}
KEG_LOCK(keg);
if (keg->uk_flags & UMA_ZONE_HASH)
UMA_HASH_INSERT(&keg->uk_hash, slab, mem);
keg->uk_pages += keg->uk_ppera;
keg->uk_free += keg->uk_ipers;
return (slab);
}
/*
* This function is intended to be used early on in place of page_alloc() so
* that we may use the boot time page cache to satisfy allocations before
* the VM is ready.
*/
static void *
startup_alloc(uma_zone_t zone, int bytes, u_int8_t *pflag, int wait)
{
uma_keg_t keg;
uma_slab_t tmps;
keg = zone_first_keg(zone);
/*
* Check our small startup cache to see if it has pages remaining.
*/
mtx_lock(&uma_boot_pages_mtx);
if ((tmps = LIST_FIRST(&uma_boot_pages)) != NULL) {
LIST_REMOVE(tmps, us_link);
mtx_unlock(&uma_boot_pages_mtx);
*pflag = tmps->us_flags;
return (tmps->us_data);
}
mtx_unlock(&uma_boot_pages_mtx);
if (booted == 0)
panic("UMA: Increase vm.boot_pages");
/*
* Now that we've booted reset these users to their real allocator.
*/
#ifdef UMA_MD_SMALL_ALLOC
keg->uk_allocf = uma_small_alloc;
#else
keg->uk_allocf = page_alloc;
#endif
return keg->uk_allocf(zone, bytes, pflag, wait);
}
/*
* Allocates a number of pages from the system
*
* Arguments:
* bytes The number of bytes requested
* wait Shall we wait?
*
* Returns:
* A pointer to the alloced memory or possibly
* NULL if M_NOWAIT is set.
*/
static void *
page_alloc(uma_zone_t zone, int bytes, u_int8_t *pflag, int wait)
{
void *p; /* Returned page */
*pflag = UMA_SLAB_KMEM;
p = (void *) kmem_malloc(kmem_map, bytes, wait);
return (p);
}
/*
* Allocates a number of pages from within an object
*
* Arguments:
* bytes The number of bytes requested
* wait Shall we wait?
*
* Returns:
* A pointer to the alloced memory or possibly
* NULL if M_NOWAIT is set.
*/
static void *
obj_alloc(uma_zone_t zone, int bytes, u_int8_t *flags, int wait)
{
vm_object_t object;
vm_offset_t retkva, zkva;
vm_page_t p;
int pages, startpages;
uma_keg_t keg;
keg = zone_first_keg(zone);
object = keg->uk_obj;
retkva = 0;
/*
* This looks a little weird since we're getting one page at a time.
*/
VM_OBJECT_LOCK(object);
p = TAILQ_LAST(&object->memq, pglist);
pages = p != NULL ? p->pindex + 1 : 0;
startpages = pages;
zkva = keg->uk_kva + pages * PAGE_SIZE;
for (; bytes > 0; bytes -= PAGE_SIZE) {
p = vm_page_alloc(object, pages,
VM_ALLOC_INTERRUPT | VM_ALLOC_WIRED);
if (p == NULL) {
if (pages != startpages)
pmap_qremove(retkva, pages - startpages);
while (pages != startpages) {
pages--;
p = TAILQ_LAST(&object->memq, pglist);
vm_page_lock_queues();
vm_page_unwire(p, 0);
vm_page_free(p);
vm_page_unlock_queues();
}
retkva = 0;
goto done;
}
pmap_qenter(zkva, &p, 1);
if (retkva == 0)
retkva = zkva;
zkva += PAGE_SIZE;
pages += 1;
}
done:
VM_OBJECT_UNLOCK(object);
*flags = UMA_SLAB_PRIV;
return ((void *)retkva);
}
/*
* Frees a number of pages to the system
*
* Arguments:
* mem A pointer to the memory to be freed
* size The size of the memory being freed
* flags The original p->us_flags field
*
* Returns:
* Nothing
*/
static void
page_free(void *mem, int size, u_int8_t flags)
{
vm_map_t map;
if (flags & UMA_SLAB_KMEM)
map = kmem_map;
else if (flags & UMA_SLAB_KERNEL)
map = kernel_map;
else
panic("UMA: page_free used with invalid flags %d", flags);
kmem_free(map, (vm_offset_t)mem, size);
}
/*
* Zero fill initializer
*
* Arguments/Returns follow uma_init specifications
*/
static int
zero_init(void *mem, int size, int flags)
{
bzero(mem, size);
return (0);
}
/*
* Finish creating a small uma keg. This calculates ipers, and the keg size.
*
* Arguments
* keg The zone we should initialize
*
* Returns
* Nothing
*/
static void
keg_small_init(uma_keg_t keg)
{
u_int rsize;
u_int memused;
u_int wastedspace;
u_int shsize;
KASSERT(keg != NULL, ("Keg is null in keg_small_init"));
rsize = keg->uk_size;
if (rsize < UMA_SMALLEST_UNIT)
rsize = UMA_SMALLEST_UNIT;
if (rsize & keg->uk_align)
rsize = (rsize & ~keg->uk_align) + (keg->uk_align + 1);
keg->uk_rsize = rsize;
keg->uk_ppera = 1;
if (keg->uk_flags & UMA_ZONE_REFCNT) {
rsize += UMA_FRITMREF_SZ; /* linkage & refcnt */
shsize = sizeof(struct uma_slab_refcnt);
} else {
rsize += UMA_FRITM_SZ; /* Account for linkage */
shsize = sizeof(struct uma_slab);
}
keg->uk_ipers = (UMA_SLAB_SIZE - shsize) / rsize;
KASSERT(keg->uk_ipers != 0, ("keg_small_init: ipers is 0"));
memused = keg->uk_ipers * rsize + shsize;
wastedspace = UMA_SLAB_SIZE - memused;
/*
* We can't do OFFPAGE if we're internal or if we've been
* asked to not go to the VM for buckets. If we do this we
* may end up going to the VM (kmem_map) for slabs which we
* do not want to do if we're UMA_ZFLAG_CACHEONLY as a
* result of UMA_ZONE_VM, which clearly forbids it.
*/
if ((keg->uk_flags & UMA_ZFLAG_INTERNAL) ||
(keg->uk_flags & UMA_ZFLAG_CACHEONLY))
return;
if ((wastedspace >= UMA_MAX_WASTE) &&
(keg->uk_ipers < (UMA_SLAB_SIZE / keg->uk_rsize))) {
keg->uk_ipers = UMA_SLAB_SIZE / keg->uk_rsize;
KASSERT(keg->uk_ipers <= 255,
("keg_small_init: keg->uk_ipers too high!"));
#ifdef UMA_DEBUG
printf("UMA decided we need offpage slab headers for "
"keg: %s, calculated wastedspace = %d, "
"maximum wasted space allowed = %d, "
"calculated ipers = %d, "
"new wasted space = %d\n", keg->uk_name, wastedspace,
UMA_MAX_WASTE, keg->uk_ipers,
UMA_SLAB_SIZE - keg->uk_ipers * keg->uk_rsize);
#endif
keg->uk_flags |= UMA_ZONE_OFFPAGE;
if ((keg->uk_flags & UMA_ZONE_VTOSLAB) == 0)
keg->uk_flags |= UMA_ZONE_HASH;
}
}
/*
* Finish creating a large (> UMA_SLAB_SIZE) uma kegs. Just give in and do
* OFFPAGE for now. When I can allow for more dynamic slab sizes this will be
* more complicated.
*
* Arguments
* keg The keg we should initialize
*
* Returns
* Nothing
*/
static void
keg_large_init(uma_keg_t keg)
{
int pages;
KASSERT(keg != NULL, ("Keg is null in keg_large_init"));
KASSERT((keg->uk_flags & UMA_ZFLAG_CACHEONLY) == 0,
("keg_large_init: Cannot large-init a UMA_ZFLAG_CACHEONLY keg"));
pages = keg->uk_size / UMA_SLAB_SIZE;
/* Account for remainder */
if ((pages * UMA_SLAB_SIZE) < keg->uk_size)
pages++;
keg->uk_ppera = pages;
keg->uk_ipers = 1;
keg->uk_flags |= UMA_ZONE_OFFPAGE;
if ((keg->uk_flags & UMA_ZONE_VTOSLAB) == 0)
keg->uk_flags |= UMA_ZONE_HASH;
keg->uk_rsize = keg->uk_size;
}
static void
keg_cachespread_init(uma_keg_t keg)
{
int alignsize;
int trailer;
int pages;
int rsize;
alignsize = keg->uk_align + 1;
rsize = keg->uk_size;
/*
* We want one item to start on every align boundary in a page. To
* do this we will span pages. We will also extend the item by the
* size of align if it is an even multiple of align. Otherwise, it
* would fall on the same boundary every time.
*/
if (rsize & keg->uk_align)
rsize = (rsize & ~keg->uk_align) + alignsize;
if ((rsize & alignsize) == 0)
rsize += alignsize;
trailer = rsize - keg->uk_size;
pages = (rsize * (PAGE_SIZE / alignsize)) / PAGE_SIZE;
pages = MIN(pages, (128 * 1024) / PAGE_SIZE);
keg->uk_rsize = rsize;
keg->uk_ppera = pages;
keg->uk_ipers = ((pages * PAGE_SIZE) + trailer) / rsize;
keg->uk_flags |= UMA_ZONE_OFFPAGE | UMA_ZONE_VTOSLAB;
KASSERT(keg->uk_ipers <= uma_max_ipers,
("keg_small_init: keg->uk_ipers too high(%d) increase max_ipers",
keg->uk_ipers));
}
/*
* Keg header ctor. This initializes all fields, locks, etc. And inserts
* the keg onto the global keg list.
*
* Arguments/Returns follow uma_ctor specifications
* udata Actually uma_kctor_args
*/
static int
keg_ctor(void *mem, int size, void *udata, int flags)
{
struct uma_kctor_args *arg = udata;
uma_keg_t keg = mem;
uma_zone_t zone;
bzero(keg, size);
keg->uk_size = arg->size;
keg->uk_init = arg->uminit;
keg->uk_fini = arg->fini;
keg->uk_align = arg->align;
keg->uk_free = 0;
keg->uk_pages = 0;
keg->uk_flags = arg->flags;
keg->uk_allocf = page_alloc;
keg->uk_freef = page_free;
keg->uk_recurse = 0;
keg->uk_slabzone = NULL;
/*
* The master zone is passed to us at keg-creation time.
*/
zone = arg->zone;
keg->uk_name = zone->uz_name;
if (arg->flags & UMA_ZONE_VM)
keg->uk_flags |= UMA_ZFLAG_CACHEONLY;
if (arg->flags & UMA_ZONE_ZINIT)
keg->uk_init = zero_init;
if (arg->flags & UMA_ZONE_REFCNT || arg->flags & UMA_ZONE_MALLOC)
keg->uk_flags |= UMA_ZONE_VTOSLAB;
/*
* The +UMA_FRITM_SZ added to uk_size is to account for the
* linkage that is added to the size in keg_small_init(). If
* we don't account for this here then we may end up in
* keg_small_init() with a calculated 'ipers' of 0.
*/
if (keg->uk_flags & UMA_ZONE_REFCNT) {
if (keg->uk_flags & UMA_ZONE_CACHESPREAD)
keg_cachespread_init(keg);
else if ((keg->uk_size+UMA_FRITMREF_SZ) >
(UMA_SLAB_SIZE - sizeof(struct uma_slab_refcnt)))
keg_large_init(keg);
else
keg_small_init(keg);
} else {
if (keg->uk_flags & UMA_ZONE_CACHESPREAD)
keg_cachespread_init(keg);
else if ((keg->uk_size+UMA_FRITM_SZ) >
(UMA_SLAB_SIZE - sizeof(struct uma_slab)))
keg_large_init(keg);
else
keg_small_init(keg);
}
if (keg->uk_flags & UMA_ZONE_OFFPAGE) {
if (keg->uk_flags & UMA_ZONE_REFCNT)
keg->uk_slabzone = slabrefzone;
else
keg->uk_slabzone = slabzone;
}
/*
* If we haven't booted yet we need allocations to go through the
* startup cache until the vm is ready.
*/
if (keg->uk_ppera == 1) {
#ifdef UMA_MD_SMALL_ALLOC
keg->uk_allocf = uma_small_alloc;
keg->uk_freef = uma_small_free;
#endif
if (booted == 0)
keg->uk_allocf = startup_alloc;
}
/*
* Initialize keg's lock (shared among zones).
*/
if (arg->flags & UMA_ZONE_MTXCLASS)
KEG_LOCK_INIT(keg, 1);
else
KEG_LOCK_INIT(keg, 0);
/*
* If we're putting the slab header in the actual page we need to
* figure out where in each page it goes. This calculates a right
* justified offset into the memory on an ALIGN_PTR boundary.
*/
if (!(keg->uk_flags & UMA_ZONE_OFFPAGE)) {
u_int totsize;
/* Size of the slab struct and free list */
if (keg->uk_flags & UMA_ZONE_REFCNT)
totsize = sizeof(struct uma_slab_refcnt) +
keg->uk_ipers * UMA_FRITMREF_SZ;
else
totsize = sizeof(struct uma_slab) +
keg->uk_ipers * UMA_FRITM_SZ;
if (totsize & UMA_ALIGN_PTR)
totsize = (totsize & ~UMA_ALIGN_PTR) +
(UMA_ALIGN_PTR + 1);
keg->uk_pgoff = UMA_SLAB_SIZE - totsize;
if (keg->uk_flags & UMA_ZONE_REFCNT)
totsize = keg->uk_pgoff + sizeof(struct uma_slab_refcnt)
+ keg->uk_ipers * UMA_FRITMREF_SZ;
else
totsize = keg->uk_pgoff + sizeof(struct uma_slab)
+ keg->uk_ipers * UMA_FRITM_SZ;
/*
* The only way the following is possible is if with our
* UMA_ALIGN_PTR adjustments we are now bigger than
* UMA_SLAB_SIZE. I haven't checked whether this is
* mathematically possible for all cases, so we make
* sure here anyway.
*/
if (totsize > UMA_SLAB_SIZE) {
printf("zone %s ipers %d rsize %d size %d\n",
zone->uz_name, keg->uk_ipers, keg->uk_rsize,
keg->uk_size);
panic("UMA slab won't fit.");
}
}
if (keg->uk_flags & UMA_ZONE_HASH)
hash_alloc(&keg->uk_hash);
#ifdef UMA_DEBUG
printf("UMA: %s(%p) size %d(%d) flags %d ipers %d ppera %d out %d free %d\n",
zone->uz_name, zone, keg->uk_size, keg->uk_rsize, keg->uk_flags,
keg->uk_ipers, keg->uk_ppera,
(keg->uk_ipers * keg->uk_pages) - keg->uk_free, keg->uk_free);
#endif
LIST_INSERT_HEAD(&keg->uk_zones, zone, uz_link);
mtx_lock(&uma_mtx);
LIST_INSERT_HEAD(&uma_kegs, keg, uk_link);
mtx_unlock(&uma_mtx);
return (0);
}
/*
* Zone header ctor. This initializes all fields, locks, etc.
*
* Arguments/Returns follow uma_ctor specifications
* udata Actually uma_zctor_args
*/
static int
zone_ctor(void *mem, int size, void *udata, int flags)
{
struct uma_zctor_args *arg = udata;
uma_zone_t zone = mem;
uma_zone_t z;
uma_keg_t keg;
bzero(zone, size);
zone->uz_name = arg->name;
zone->uz_ctor = arg->ctor;
zone->uz_dtor = arg->dtor;
zone->uz_slab = zone_fetch_slab;
zone->uz_init = NULL;
zone->uz_fini = NULL;
zone->uz_allocs = 0;
zone->uz_frees = 0;
zone->uz_fails = 0;
zone->uz_fills = zone->uz_count = 0;
zone->uz_flags = 0;
keg = arg->keg;
if (arg->flags & UMA_ZONE_SECONDARY) {
KASSERT(arg->keg != NULL, ("Secondary zone on zero'd keg"));
zone->uz_init = arg->uminit;
zone->uz_fini = arg->fini;
zone->uz_lock = &keg->uk_lock;
zone->uz_flags |= UMA_ZONE_SECONDARY;
mtx_lock(&uma_mtx);
ZONE_LOCK(zone);
LIST_FOREACH(z, &keg->uk_zones, uz_link) {
if (LIST_NEXT(z, uz_link) == NULL) {
LIST_INSERT_AFTER(z, zone, uz_link);
break;
}
}
ZONE_UNLOCK(zone);
mtx_unlock(&uma_mtx);
} else if (keg == NULL) {
if ((keg = uma_kcreate(zone, arg->size, arg->uminit, arg->fini,
arg->align, arg->flags)) == NULL)
return (ENOMEM);
} else {
struct uma_kctor_args karg;
int error;
/* We should only be here from uma_startup() */
karg.size = arg->size;
karg.uminit = arg->uminit;
karg.fini = arg->fini;
karg.align = arg->align;
karg.flags = arg->flags;
karg.zone = zone;
error = keg_ctor(arg->keg, sizeof(struct uma_keg), &karg,
flags);
if (error)
return (error);
}
/*
* Link in the first keg.
*/
zone->uz_klink.kl_keg = keg;
LIST_INSERT_HEAD(&zone->uz_kegs, &zone->uz_klink, kl_link);
zone->uz_lock = &keg->uk_lock;
zone->uz_size = keg->uk_size;
zone->uz_flags |= (keg->uk_flags &
(UMA_ZONE_INHERIT | UMA_ZFLAG_INHERIT));
/*
* Some internal zones don't have room allocated for the per cpu
* caches. If we're internal, bail out here.
*/
if (keg->uk_flags & UMA_ZFLAG_INTERNAL) {
KASSERT((zone->uz_flags & UMA_ZONE_SECONDARY) == 0,
("Secondary zone requested UMA_ZFLAG_INTERNAL"));
return (0);
}
if (keg->uk_flags & UMA_ZONE_MAXBUCKET)
zone->uz_count = BUCKET_MAX;
else if (keg->uk_ipers <= BUCKET_MAX)
zone->uz_count = keg->uk_ipers;
else
zone->uz_count = BUCKET_MAX;
return (0);
}
/*
* Keg header dtor. This frees all data, destroys locks, frees the hash
* table and removes the keg from the global list.
*
* Arguments/Returns follow uma_dtor specifications
* udata unused
*/
static void
keg_dtor(void *arg, int size, void *udata)
{
uma_keg_t keg;
keg = (uma_keg_t)arg;
KEG_LOCK(keg);
if (keg->uk_free != 0) {
printf("Freed UMA keg was not empty (%d items). "
" Lost %d pages of memory.\n",
keg->uk_free, keg->uk_pages);
}
KEG_UNLOCK(keg);
hash_free(&keg->uk_hash);
KEG_LOCK_FINI(keg);
}
/*
* Zone header dtor.
*
* Arguments/Returns follow uma_dtor specifications
* udata unused
*/
static void
zone_dtor(void *arg, int size, void *udata)
{
uma_klink_t klink;
uma_zone_t zone;
uma_keg_t keg;
zone = (uma_zone_t)arg;
keg = zone_first_keg(zone);
if (!(zone->uz_flags & UMA_ZFLAG_INTERNAL))
cache_drain(zone);
mtx_lock(&uma_mtx);
LIST_REMOVE(zone, uz_link);
mtx_unlock(&uma_mtx);
/*
* XXX there are some races here where
* the zone can be drained but zone lock
* released and then refilled before we
* remove it... we dont care for now
*/
zone_drain_wait(zone, M_WAITOK);
/*
* Unlink all of our kegs.
*/
while ((klink = LIST_FIRST(&zone->uz_kegs)) != NULL) {
klink->kl_keg = NULL;
LIST_REMOVE(klink, kl_link);
if (klink == &zone->uz_klink)
continue;
free(klink, M_TEMP);
}
/*
* We only destroy kegs from non secondary zones.
*/
if ((zone->uz_flags & UMA_ZONE_SECONDARY) == 0) {
mtx_lock(&uma_mtx);
LIST_REMOVE(keg, uk_link);
mtx_unlock(&uma_mtx);
zone_free_item(kegs, keg, NULL, SKIP_NONE,
ZFREE_STATFREE);
}
}
/*
* Traverses every zone in the system and calls a callback
*
* Arguments:
* zfunc A pointer to a function which accepts a zone
* as an argument.
*
* Returns:
* Nothing
*/
static void
zone_foreach(void (*zfunc)(uma_zone_t))
{
uma_keg_t keg;
uma_zone_t zone;
mtx_lock(&uma_mtx);
LIST_FOREACH(keg, &uma_kegs, uk_link) {
LIST_FOREACH(zone, &keg->uk_zones, uz_link)
zfunc(zone);
}
mtx_unlock(&uma_mtx);
}
/* Public functions */
/* See uma.h */
void
uma_startup(void *bootmem, int boot_pages)
{
struct uma_zctor_args args;
uma_slab_t slab;
u_int slabsize;
u_int objsize, totsize, wsize;
int i;
#ifdef UMA_DEBUG
printf("Creating uma keg headers zone and keg.\n");
#endif
mtx_init(&uma_mtx, "UMA lock", NULL, MTX_DEF);
/*
* Figure out the maximum number of items-per-slab we'll have if
* we're using the OFFPAGE slab header to track free items, given
* all possible object sizes and the maximum desired wastage
* (UMA_MAX_WASTE).
*
* We iterate until we find an object size for
* which the calculated wastage in keg_small_init() will be
* enough to warrant OFFPAGE. Since wastedspace versus objsize
* is an overall increasing see-saw function, we find the smallest
* objsize such that the wastage is always acceptable for objects
* with that objsize or smaller. Since a smaller objsize always
* generates a larger possible uma_max_ipers, we use this computed
* objsize to calculate the largest ipers possible. Since the
* ipers calculated for OFFPAGE slab headers is always larger than
* the ipers initially calculated in keg_small_init(), we use
* the former's equation (UMA_SLAB_SIZE / keg->uk_rsize) to
* obtain the maximum ipers possible for offpage slab headers.
*
* It should be noted that ipers versus objsize is an inversly
* proportional function which drops off rather quickly so as
* long as our UMA_MAX_WASTE is such that the objsize we calculate
* falls into the portion of the inverse relation AFTER the steep
* falloff, then uma_max_ipers shouldn't be too high (~10 on i386).
*
* Note that we have 8-bits (1 byte) to use as a freelist index
* inside the actual slab header itself and this is enough to
* accomodate us. In the worst case, a UMA_SMALLEST_UNIT sized
* object with offpage slab header would have ipers =
* UMA_SLAB_SIZE / UMA_SMALLEST_UNIT (currently = 256), which is
* 1 greater than what our byte-integer freelist index can
* accomodate, but we know that this situation never occurs as
* for UMA_SMALLEST_UNIT-sized objects, we will never calculate
* that we need to go to offpage slab headers. Or, if we do,
* then we trap that condition below and panic in the INVARIANTS case.
*/
wsize = UMA_SLAB_SIZE - sizeof(struct uma_slab) - UMA_MAX_WASTE;
totsize = wsize;
objsize = UMA_SMALLEST_UNIT;
while (totsize >= wsize) {
totsize = (UMA_SLAB_SIZE - sizeof(struct uma_slab)) /
(objsize + UMA_FRITM_SZ);
totsize *= (UMA_FRITM_SZ + objsize);
objsize++;
}
if (objsize > UMA_SMALLEST_UNIT)
objsize--;
uma_max_ipers = MAX(UMA_SLAB_SIZE / objsize, 64);
wsize = UMA_SLAB_SIZE - sizeof(struct uma_slab_refcnt) - UMA_MAX_WASTE;
totsize = wsize;
objsize = UMA_SMALLEST_UNIT;
while (totsize >= wsize) {
totsize = (UMA_SLAB_SIZE - sizeof(struct uma_slab_refcnt)) /
(objsize + UMA_FRITMREF_SZ);
totsize *= (UMA_FRITMREF_SZ + objsize);
objsize++;
}
if (objsize > UMA_SMALLEST_UNIT)
objsize--;
uma_max_ipers_ref = MAX(UMA_SLAB_SIZE / objsize, 64);
KASSERT((uma_max_ipers_ref <= 255) && (uma_max_ipers <= 255),
("uma_startup: calculated uma_max_ipers values too large!"));
#ifdef UMA_DEBUG
printf("Calculated uma_max_ipers (for OFFPAGE) is %d\n", uma_max_ipers);
printf("Calculated uma_max_ipers_slab (for OFFPAGE) is %d\n",
uma_max_ipers_ref);
#endif
/* "manually" create the initial zone */
args.name = "UMA Kegs";
args.size = sizeof(struct uma_keg);
args.ctor = keg_ctor;
args.dtor = keg_dtor;
args.uminit = zero_init;
args.fini = NULL;
args.keg = &masterkeg;
args.align = 32 - 1;
args.flags = UMA_ZFLAG_INTERNAL;
/* The initial zone has no Per cpu queues so it's smaller */
zone_ctor(kegs, sizeof(struct uma_zone), &args, M_WAITOK);
#ifdef UMA_DEBUG
printf("Filling boot free list.\n");
#endif
for (i = 0; i < boot_pages; i++) {
slab = (uma_slab_t)((u_int8_t *)bootmem + (i * UMA_SLAB_SIZE));
slab->us_data = (u_int8_t *)slab;
slab->us_flags = UMA_SLAB_BOOT;
LIST_INSERT_HEAD(&uma_boot_pages, slab, us_link);
}
mtx_init(&uma_boot_pages_mtx, "UMA boot pages", NULL, MTX_DEF);
#ifdef UMA_DEBUG
printf("Creating uma zone headers zone and keg.\n");
#endif
args.name = "UMA Zones";
args.size = sizeof(struct uma_zone) +
(sizeof(struct uma_cache) * (mp_maxid + 1));
args.ctor = zone_ctor;
args.dtor = zone_dtor;
args.uminit = zero_init;
args.fini = NULL;
args.keg = NULL;
args.align = 32 - 1;
args.flags = UMA_ZFLAG_INTERNAL;
/* The initial zone has no Per cpu queues so it's smaller */
zone_ctor(zones, sizeof(struct uma_zone), &args, M_WAITOK);
#ifdef UMA_DEBUG
printf("Initializing pcpu cache locks.\n");
#endif
#ifdef UMA_DEBUG
printf("Creating slab and hash zones.\n");
#endif
/*
* This is the max number of free list items we'll have with
* offpage slabs.
*/
slabsize = uma_max_ipers * UMA_FRITM_SZ;
slabsize += sizeof(struct uma_slab);
/* Now make a zone for slab headers */
slabzone = uma_zcreate("UMA Slabs",
slabsize,
NULL, NULL, NULL, NULL,
UMA_ALIGN_PTR, UMA_ZFLAG_INTERNAL);
/*
* We also create a zone for the bigger slabs with reference
* counts in them, to accomodate UMA_ZONE_REFCNT zones.
*/
slabsize = uma_max_ipers_ref * UMA_FRITMREF_SZ;
slabsize += sizeof(struct uma_slab_refcnt);
slabrefzone = uma_zcreate("UMA RCntSlabs",
slabsize,
NULL, NULL, NULL, NULL,
UMA_ALIGN_PTR,
UMA_ZFLAG_INTERNAL);
hashzone = uma_zcreate("UMA Hash",
sizeof(struct slabhead *) * UMA_HASH_SIZE_INIT,
NULL, NULL, NULL, NULL,
UMA_ALIGN_PTR, UMA_ZFLAG_INTERNAL);
bucket_init();
#if defined(UMA_MD_SMALL_ALLOC) && !defined(UMA_MD_SMALL_ALLOC_NEEDS_VM)
booted = 1;
#endif
#ifdef UMA_DEBUG
printf("UMA startup complete.\n");
#endif
}
/* see uma.h */
void
uma_startup2(void)
{
booted = 1;
bucket_enable();
#ifdef UMA_DEBUG
printf("UMA startup2 complete.\n");
#endif
}
/*
* Initialize our callout handle
*
*/
static void
uma_startup3(void)
{
#ifdef UMA_DEBUG
printf("Starting callout.\n");
#endif
callout_init(&uma_callout, CALLOUT_MPSAFE);
callout_reset(&uma_callout, UMA_TIMEOUT * hz, uma_timeout, NULL);
#ifdef UMA_DEBUG
printf("UMA startup3 complete.\n");
#endif
}
static uma_keg_t
uma_kcreate(uma_zone_t zone, size_t size, uma_init uminit, uma_fini fini,
int align, u_int32_t flags)
{
struct uma_kctor_args args;
args.size = size;
args.uminit = uminit;
args.fini = fini;
args.align = (align == UMA_ALIGN_CACHE) ? uma_align_cache : align;
args.flags = flags;
args.zone = zone;
return (zone_alloc_item(kegs, &args, M_WAITOK));
}
/* See uma.h */
void
uma_set_align(int align)
{
if (align != UMA_ALIGN_CACHE)
uma_align_cache = align;
}
/* See uma.h */
uma_zone_t
uma_zcreate(char *name, size_t size, uma_ctor ctor, uma_dtor dtor,
uma_init uminit, uma_fini fini, int align, u_int32_t flags)
{
struct uma_zctor_args args;
/* This stuff is essential for the zone ctor */
args.name = name;
args.size = size;
args.ctor = ctor;
args.dtor = dtor;
args.uminit = uminit;
args.fini = fini;
args.align = align;
args.flags = flags;
args.keg = NULL;
return (zone_alloc_item(zones, &args, M_WAITOK));
}
/* See uma.h */
uma_zone_t
uma_zsecond_create(char *name, uma_ctor ctor, uma_dtor dtor,
uma_init zinit, uma_fini zfini, uma_zone_t master)
{
struct uma_zctor_args args;
uma_keg_t keg;
keg = zone_first_keg(master);
args.name = name;
args.size = keg->uk_size;
args.ctor = ctor;
args.dtor = dtor;
args.uminit = zinit;
args.fini = zfini;
args.align = keg->uk_align;
args.flags = keg->uk_flags | UMA_ZONE_SECONDARY;
args.keg = keg;
/* XXX Attaches only one keg of potentially many. */
return (zone_alloc_item(zones, &args, M_WAITOK));
}
static void
zone_lock_pair(uma_zone_t a, uma_zone_t b)
{
if (a < b) {
ZONE_LOCK(a);
mtx_lock_flags(b->uz_lock, MTX_DUPOK);
} else {
ZONE_LOCK(b);
mtx_lock_flags(a->uz_lock, MTX_DUPOK);
}
}
static void
zone_unlock_pair(uma_zone_t a, uma_zone_t b)
{
ZONE_UNLOCK(a);
ZONE_UNLOCK(b);
}
int
uma_zsecond_add(uma_zone_t zone, uma_zone_t master)
{
uma_klink_t klink;
uma_klink_t kl;
int error;
error = 0;
klink = malloc(sizeof(*klink), M_TEMP, M_WAITOK | M_ZERO);
zone_lock_pair(zone, master);
/*
* zone must use vtoslab() to resolve objects and must already be
* a secondary.
*/
if ((zone->uz_flags & (UMA_ZONE_VTOSLAB | UMA_ZONE_SECONDARY))
!= (UMA_ZONE_VTOSLAB | UMA_ZONE_SECONDARY)) {
error = EINVAL;
goto out;
}
/*
* The new master must also use vtoslab().
*/
if ((zone->uz_flags & UMA_ZONE_VTOSLAB) != UMA_ZONE_VTOSLAB) {
error = EINVAL;
goto out;
}
/*
* Both must either be refcnt, or not be refcnt.
*/
if ((zone->uz_flags & UMA_ZONE_REFCNT) !=
(master->uz_flags & UMA_ZONE_REFCNT)) {
error = EINVAL;
goto out;
}
/*
* The underlying object must be the same size. rsize
* may be different.
*/
if (master->uz_size != zone->uz_size) {
error = E2BIG;
goto out;
}
/*
* Put it at the end of the list.
*/
klink->kl_keg = zone_first_keg(master);
LIST_FOREACH(kl, &zone->uz_kegs, kl_link) {
if (LIST_NEXT(kl, kl_link) == NULL) {
LIST_INSERT_AFTER(kl, klink, kl_link);
break;
}
}
klink = NULL;
zone->uz_flags |= UMA_ZFLAG_MULTI;
zone->uz_slab = zone_fetch_slab_multi;
out:
zone_unlock_pair(zone, master);
if (klink != NULL)
free(klink, M_TEMP);
return (error);
}
/* See uma.h */
void
uma_zdestroy(uma_zone_t zone)
{
zone_free_item(zones, zone, NULL, SKIP_NONE, ZFREE_STATFREE);
}
/* See uma.h */
void *
uma_zalloc_arg(uma_zone_t zone, void *udata, int flags)
{
void *item;
uma_cache_t cache;
uma_bucket_t bucket;
int cpu;
/* This is the fast path allocation */
#ifdef UMA_DEBUG_ALLOC_1
printf("Allocating one item from %s(%p)\n", zone->uz_name, zone);
#endif
CTR3(KTR_UMA, "uma_zalloc_arg thread %x zone %s flags %d", curthread,
zone->uz_name, flags);
if (flags & M_WAITOK) {
WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, NULL,
"uma_zalloc_arg: zone \"%s\"", zone->uz_name);
}
/*
* If possible, allocate from the per-CPU cache. There are two
* requirements for safe access to the per-CPU cache: (1) the thread
* accessing the cache must not be preempted or yield during access,
* and (2) the thread must not migrate CPUs without switching which
* cache it accesses. We rely on a critical section to prevent
* preemption and migration. We release the critical section in
* order to acquire the zone mutex if we are unable to allocate from
* the current cache; when we re-acquire the critical section, we
* must detect and handle migration if it has occurred.
*/
zalloc_restart:
critical_enter();
cpu = curcpu;
cache = &zone->uz_cpu[cpu];
zalloc_start:
bucket = cache->uc_allocbucket;
if (bucket) {
if (bucket->ub_cnt > 0) {
bucket->ub_cnt--;
item = bucket->ub_bucket[bucket->ub_cnt];
#ifdef INVARIANTS
bucket->ub_bucket[bucket->ub_cnt] = NULL;
#endif
KASSERT(item != NULL,
("uma_zalloc: Bucket pointer mangled."));
cache->uc_allocs++;
critical_exit();
#ifdef INVARIANTS
ZONE_LOCK(zone);
uma_dbg_alloc(zone, NULL, item);
ZONE_UNLOCK(zone);
#endif
if (zone->uz_ctor != NULL) {
if (zone->uz_ctor(item, zone->uz_size,
udata, flags) != 0) {
zone_free_item(zone, item, udata,
SKIP_DTOR, ZFREE_STATFAIL |
ZFREE_STATFREE);
return (NULL);
}
}
if (flags & M_ZERO)
bzero(item, zone->uz_size);
return (item);
} else if (cache->uc_freebucket) {
/*
* We have run out of items in our allocbucket.
* See if we can switch with our free bucket.
*/
if (cache->uc_freebucket->ub_cnt > 0) {
#ifdef UMA_DEBUG_ALLOC
printf("uma_zalloc: Swapping empty with"
" alloc.\n");
#endif
bucket = cache->uc_freebucket;
cache->uc_freebucket = cache->uc_allocbucket;
cache->uc_allocbucket = bucket;
goto zalloc_start;
}
}
}
/*
* Attempt to retrieve the item from the per-CPU cache has failed, so
* we must go back to the zone. This requires the zone lock, so we
* must drop the critical section, then re-acquire it when we go back
* to the cache. Since the critical section is released, we may be
* preempted or migrate. As such, make sure not to maintain any
* thread-local state specific to the cache from prior to releasing
* the critical section.
*/
critical_exit();
ZONE_LOCK(zone);
critical_enter();
cpu = curcpu;
cache = &zone->uz_cpu[cpu];
bucket = cache->uc_allocbucket;
if (bucket != NULL) {
if (bucket->ub_cnt > 0) {
ZONE_UNLOCK(zone);
goto zalloc_start;
}
bucket = cache->uc_freebucket;
if (bucket != NULL && bucket->ub_cnt > 0) {
ZONE_UNLOCK(zone);
goto zalloc_start;
}
}
/* Since we have locked the zone we may as well send back our stats */
zone->uz_allocs += cache->uc_allocs;
cache->uc_allocs = 0;
zone->uz_frees += cache->uc_frees;
cache->uc_frees = 0;
/* Our old one is now a free bucket */
if (cache->uc_allocbucket) {
KASSERT(cache->uc_allocbucket->ub_cnt == 0,
("uma_zalloc_arg: Freeing a non free bucket."));
LIST_INSERT_HEAD(&zone->uz_free_bucket,
cache->uc_allocbucket, ub_link);
cache->uc_allocbucket = NULL;
}
/* Check the free list for a new alloc bucket */
if ((bucket = LIST_FIRST(&zone->uz_full_bucket)) != NULL) {
KASSERT(bucket->ub_cnt != 0,
("uma_zalloc_arg: Returning an empty bucket."));
LIST_REMOVE(bucket, ub_link);
cache->uc_allocbucket = bucket;
ZONE_UNLOCK(zone);
goto zalloc_start;
}
/* We are no longer associated with this CPU. */
critical_exit();
/* Bump up our uz_count so we get here less */
if (zone->uz_count < BUCKET_MAX)
zone->uz_count++;
/*
* Now lets just fill a bucket and put it on the free list. If that
* works we'll restart the allocation from the begining.
*/
if (zone_alloc_bucket(zone, flags)) {
ZONE_UNLOCK(zone);
goto zalloc_restart;
}
ZONE_UNLOCK(zone);
/*
* We may not be able to get a bucket so return an actual item.
*/
#ifdef UMA_DEBUG
printf("uma_zalloc_arg: Bucketzone returned NULL\n");
#endif
item = zone_alloc_item(zone, udata, flags);
return (item);
}
static uma_slab_t
keg_fetch_slab(uma_keg_t keg, uma_zone_t zone, int flags)
{
uma_slab_t slab;
mtx_assert(&keg->uk_lock, MA_OWNED);
slab = NULL;
for (;;) {
/*
* Find a slab with some space. Prefer slabs that are partially
* used over those that are totally full. This helps to reduce
* fragmentation.
*/
if (keg->uk_free != 0) {
if (!LIST_EMPTY(&keg->uk_part_slab)) {
slab = LIST_FIRST(&keg->uk_part_slab);
} else {
slab = LIST_FIRST(&keg->uk_free_slab);
LIST_REMOVE(slab, us_link);
LIST_INSERT_HEAD(&keg->uk_part_slab, slab,
us_link);
}
MPASS(slab->us_keg == keg);
return (slab);
}
/*
* M_NOVM means don't ask at all!
*/
if (flags & M_NOVM)
break;
if (keg->uk_maxpages && keg->uk_pages >= keg->uk_maxpages) {
keg->uk_flags |= UMA_ZFLAG_FULL;
/*
* If this is not a multi-zone, set the FULL bit.
* Otherwise slab_multi() takes care of it.
*/
if ((zone->uz_flags & UMA_ZFLAG_MULTI) == 0)
zone->uz_flags |= UMA_ZFLAG_FULL;
if (flags & M_NOWAIT)
break;
msleep(keg, &keg->uk_lock, PVM, "keglimit", 0);
continue;
}
keg->uk_recurse++;
slab = keg_alloc_slab(keg, zone, flags);
keg->uk_recurse--;
/*
* If we got a slab here it's safe to mark it partially used
* and return. We assume that the caller is going to remove
* at least one item.
*/
if (slab) {
MPASS(slab->us_keg == keg);
LIST_INSERT_HEAD(&keg->uk_part_slab, slab, us_link);
return (slab);
}
/*
* We might not have been able to get a slab but another cpu
* could have while we were unlocked. Check again before we
* fail.
*/
flags |= M_NOVM;
}
return (slab);
}
static inline void
zone_relock(uma_zone_t zone, uma_keg_t keg)
{
if (zone->uz_lock != &keg->uk_lock) {
KEG_UNLOCK(keg);
ZONE_LOCK(zone);
}
}
static inline void
keg_relock(uma_keg_t keg, uma_zone_t zone)
{
if (zone->uz_lock != &keg->uk_lock) {
ZONE_UNLOCK(zone);
KEG_LOCK(keg);
}
}
static uma_slab_t
zone_fetch_slab(uma_zone_t zone, uma_keg_t keg, int flags)
{
uma_slab_t slab;
if (keg == NULL)
keg = zone_first_keg(zone);
/*
* This is to prevent us from recursively trying to allocate
* buckets. The problem is that if an allocation forces us to
* grab a new bucket we will call page_alloc, which will go off
* and cause the vm to allocate vm_map_entries. If we need new
* buckets there too we will recurse in kmem_alloc and bad
* things happen. So instead we return a NULL bucket, and make
* the code that allocates buckets smart enough to deal with it
*/
if (keg->uk_flags & UMA_ZFLAG_BUCKET && keg->uk_recurse != 0)
return (NULL);
for (;;) {
slab = keg_fetch_slab(keg, zone, flags);
if (slab)
return (slab);
if (flags & (M_NOWAIT | M_NOVM))
break;
}
return (NULL);
}
/*
* uma_zone_fetch_slab_multi: Fetches a slab from one available keg. Returns
* with the keg locked. Caller must call zone_relock() afterwards if the
* zone lock is required. On NULL the zone lock is held.
*
* The last pointer is used to seed the search. It is not required.
*/
static uma_slab_t
zone_fetch_slab_multi(uma_zone_t zone, uma_keg_t last, int rflags)
{
uma_klink_t klink;
uma_slab_t slab;
uma_keg_t keg;
int flags;
int empty;
int full;
/*
* Don't wait on the first pass. This will skip limit tests
* as well. We don't want to block if we can find a provider
* without blocking.
*/
flags = (rflags & ~M_WAITOK) | M_NOWAIT;
/*
* Use the last slab allocated as a hint for where to start
* the search.
*/
if (last) {
slab = keg_fetch_slab(last, zone, flags);
if (slab)
return (slab);
zone_relock(zone, last);
last = NULL;
}
/*
* Loop until we have a slab incase of transient failures
* while M_WAITOK is specified. I'm not sure this is 100%
* required but we've done it for so long now.
*/
for (;;) {
empty = 0;
full = 0;
/*
* Search the available kegs for slabs. Be careful to hold the
* correct lock while calling into the keg layer.
*/
LIST_FOREACH(klink, &zone->uz_kegs, kl_link) {
keg = klink->kl_keg;
keg_relock(keg, zone);
if ((keg->uk_flags & UMA_ZFLAG_FULL) == 0) {
slab = keg_fetch_slab(keg, zone, flags);
if (slab)
return (slab);
}
if (keg->uk_flags & UMA_ZFLAG_FULL)
full++;
else
empty++;
zone_relock(zone, keg);
}
if (rflags & (M_NOWAIT | M_NOVM))
break;
flags = rflags;
/*
* All kegs are full. XXX We can't atomically check all kegs
* and sleep so just sleep for a short period and retry.
*/
if (full && !empty) {
zone->uz_flags |= UMA_ZFLAG_FULL;
msleep(zone, zone->uz_lock, PVM, "zonelimit", hz/100);
zone->uz_flags &= ~UMA_ZFLAG_FULL;
continue;
}
}
return (NULL);
}
static void *
slab_alloc_item(uma_zone_t zone, uma_slab_t slab)
{
uma_keg_t keg;
uma_slabrefcnt_t slabref;
void *item;
u_int8_t freei;
keg = slab->us_keg;
mtx_assert(&keg->uk_lock, MA_OWNED);
freei = slab->us_firstfree;
if (keg->uk_flags & UMA_ZONE_REFCNT) {
slabref = (uma_slabrefcnt_t)slab;
slab->us_firstfree = slabref->us_freelist[freei].us_item;
} else {
slab->us_firstfree = slab->us_freelist[freei].us_item;
}
item = slab->us_data + (keg->uk_rsize * freei);
slab->us_freecount--;
keg->uk_free--;
#ifdef INVARIANTS
uma_dbg_alloc(zone, slab, item);
#endif
/* Move this slab to the full list */
if (slab->us_freecount == 0) {
LIST_REMOVE(slab, us_link);
LIST_INSERT_HEAD(&keg->uk_full_slab, slab, us_link);
}
return (item);
}
static int
zone_alloc_bucket(uma_zone_t zone, int flags)
{
uma_bucket_t bucket;
uma_slab_t slab;
uma_keg_t keg;
int16_t saved;
int max, origflags = flags;
/*
* Try this zone's free list first so we don't allocate extra buckets.
*/
if ((bucket = LIST_FIRST(&zone->uz_free_bucket)) != NULL) {
KASSERT(bucket->ub_cnt == 0,
("zone_alloc_bucket: Bucket on free list is not empty."));
LIST_REMOVE(bucket, ub_link);
} else {
int bflags;
bflags = (flags & ~M_ZERO);
if (zone->uz_flags & UMA_ZFLAG_CACHEONLY)
bflags |= M_NOVM;
ZONE_UNLOCK(zone);
bucket = bucket_alloc(zone->uz_count, bflags);
ZONE_LOCK(zone);
}
if (bucket == NULL) {
return (0);
}
#ifdef SMP
/*
* This code is here to limit the number of simultaneous bucket fills
* for any given zone to the number of per cpu caches in this zone. This
* is done so that we don't allocate more memory than we really need.
*/
if (zone->uz_fills >= mp_ncpus)
goto done;
#endif
zone->uz_fills++;
max = MIN(bucket->ub_entries, zone->uz_count);
/* Try to keep the buckets totally full */
saved = bucket->ub_cnt;
slab = NULL;
keg = NULL;
while (bucket->ub_cnt < max &&
(slab = zone->uz_slab(zone, keg, flags)) != NULL) {
keg = slab->us_keg;
while (slab->us_freecount && bucket->ub_cnt < max) {
bucket->ub_bucket[bucket->ub_cnt++] =
slab_alloc_item(zone, slab);
}
/* Don't block on the next fill */
flags |= M_NOWAIT;
}
if (slab)
zone_relock(zone, keg);
/*
* We unlock here because we need to call the zone's init.
* It should be safe to unlock because the slab dealt with
* above is already on the appropriate list within the keg
* and the bucket we filled is not yet on any list, so we
* own it.
*/
if (zone->uz_init != NULL) {
int i;
ZONE_UNLOCK(zone);
for (i = saved; i < bucket->ub_cnt; i++)
if (zone->uz_init(bucket->ub_bucket[i], zone->uz_size,
origflags) != 0)
break;
/*
* If we couldn't initialize the whole bucket, put the
* rest back onto the freelist.
*/
if (i != bucket->ub_cnt) {
int j;
for (j = i; j < bucket->ub_cnt; j++) {
zone_free_item(zone, bucket->ub_bucket[j],
NULL, SKIP_FINI, 0);
#ifdef INVARIANTS
bucket->ub_bucket[j] = NULL;
#endif
}
bucket->ub_cnt = i;
}
ZONE_LOCK(zone);
}
zone->uz_fills--;
if (bucket->ub_cnt != 0) {
LIST_INSERT_HEAD(&zone->uz_full_bucket,
bucket, ub_link);
return (1);
}
#ifdef SMP
done:
#endif
bucket_free(bucket);
return (0);
}
/*
* Allocates an item for an internal zone
*
* Arguments
* zone The zone to alloc for.
* udata The data to be passed to the constructor.
* flags M_WAITOK, M_NOWAIT, M_ZERO.
*
* Returns
* NULL if there is no memory and M_NOWAIT is set
* An item if successful
*/
static void *
zone_alloc_item(uma_zone_t zone, void *udata, int flags)
{
uma_slab_t slab;
void *item;
item = NULL;
#ifdef UMA_DEBUG_ALLOC
printf("INTERNAL: Allocating one item from %s(%p)\n", zone->uz_name, zone);
#endif
ZONE_LOCK(zone);
slab = zone->uz_slab(zone, NULL, flags);
if (slab == NULL) {
zone->uz_fails++;
ZONE_UNLOCK(zone);
return (NULL);
}
item = slab_alloc_item(zone, slab);
zone_relock(zone, slab->us_keg);
zone->uz_allocs++;
ZONE_UNLOCK(zone);
/*
* We have to call both the zone's init (not the keg's init)
* and the zone's ctor. This is because the item is going from
* a keg slab directly to the user, and the user is expecting it
* to be both zone-init'd as well as zone-ctor'd.
*/
if (zone->uz_init != NULL) {
if (zone->uz_init(item, zone->uz_size, flags) != 0) {
zone_free_item(zone, item, udata, SKIP_FINI,
ZFREE_STATFAIL | ZFREE_STATFREE);
return (NULL);
}
}
if (zone->uz_ctor != NULL) {
if (zone->uz_ctor(item, zone->uz_size, udata, flags) != 0) {
zone_free_item(zone, item, udata, SKIP_DTOR,
ZFREE_STATFAIL | ZFREE_STATFREE);
return (NULL);
}
}
if (flags & M_ZERO)
bzero(item, zone->uz_size);
return (item);
}
/* See uma.h */
void
uma_zfree_arg(uma_zone_t zone, void *item, void *udata)
{
uma_cache_t cache;
uma_bucket_t bucket;
int bflags;
int cpu;
#ifdef UMA_DEBUG_ALLOC_1
printf("Freeing item %p to %s(%p)\n", item, zone->uz_name, zone);
#endif
CTR2(KTR_UMA, "uma_zfree_arg thread %x zone %s", curthread,
zone->uz_name);
if (zone->uz_dtor)
zone->uz_dtor(item, zone->uz_size, udata);
#ifdef INVARIANTS
ZONE_LOCK(zone);
if (zone->uz_flags & UMA_ZONE_MALLOC)
uma_dbg_free(zone, udata, item);
else
uma_dbg_free(zone, NULL, item);
ZONE_UNLOCK(zone);
#endif
/*
* The race here is acceptable. If we miss it we'll just have to wait
* a little longer for the limits to be reset.
*/
if (zone->uz_flags & UMA_ZFLAG_FULL)
goto zfree_internal;
/*
* If possible, free to the per-CPU cache. There are two
* requirements for safe access to the per-CPU cache: (1) the thread
* accessing the cache must not be preempted or yield during access,
* and (2) the thread must not migrate CPUs without switching which
* cache it accesses. We rely on a critical section to prevent
* preemption and migration. We release the critical section in
* order to acquire the zone mutex if we are unable to free to the
* current cache; when we re-acquire the critical section, we must
* detect and handle migration if it has occurred.
*/
zfree_restart:
critical_enter();
cpu = curcpu;
cache = &zone->uz_cpu[cpu];
zfree_start:
bucket = cache->uc_freebucket;
if (bucket) {
/*
* Do we have room in our bucket? It is OK for this uz count
* check to be slightly out of sync.
*/
if (bucket->ub_cnt < bucket->ub_entries) {
KASSERT(bucket->ub_bucket[bucket->ub_cnt] == NULL,
("uma_zfree: Freeing to non free bucket index."));
bucket->ub_bucket[bucket->ub_cnt] = item;
bucket->ub_cnt++;
cache->uc_frees++;
critical_exit();
return;
} else if (cache->uc_allocbucket) {
#ifdef UMA_DEBUG_ALLOC
printf("uma_zfree: Swapping buckets.\n");
#endif
/*
* We have run out of space in our freebucket.
* See if we can switch with our alloc bucket.
*/
if (cache->uc_allocbucket->ub_cnt <
cache->uc_freebucket->ub_cnt) {
bucket = cache->uc_freebucket;
cache->uc_freebucket = cache->uc_allocbucket;
cache->uc_allocbucket = bucket;
goto zfree_start;
}
}
}
/*
* We can get here for two reasons:
*
* 1) The buckets are NULL
* 2) The alloc and free buckets are both somewhat full.
*
* We must go back the zone, which requires acquiring the zone lock,
* which in turn means we must release and re-acquire the critical
* section. Since the critical section is released, we may be
* preempted or migrate. As such, make sure not to maintain any
* thread-local state specific to the cache from prior to releasing
* the critical section.
*/
critical_exit();
ZONE_LOCK(zone);
critical_enter();
cpu = curcpu;
cache = &zone->uz_cpu[cpu];
if (cache->uc_freebucket != NULL) {
if (cache->uc_freebucket->ub_cnt <
cache->uc_freebucket->ub_entries) {
ZONE_UNLOCK(zone);
goto zfree_start;
}
if (cache->uc_allocbucket != NULL &&
(cache->uc_allocbucket->ub_cnt <
cache->uc_freebucket->ub_cnt)) {
ZONE_UNLOCK(zone);
goto zfree_start;
}
}
/* Since we have locked the zone we may as well send back our stats */
zone->uz_allocs += cache->uc_allocs;
cache->uc_allocs = 0;
zone->uz_frees += cache->uc_frees;
cache->uc_frees = 0;
bucket = cache->uc_freebucket;
cache->uc_freebucket = NULL;
/* Can we throw this on the zone full list? */
if (bucket != NULL) {
#ifdef UMA_DEBUG_ALLOC
printf("uma_zfree: Putting old bucket on the free list.\n");
#endif
/* ub_cnt is pointing to the last free item */
KASSERT(bucket->ub_cnt != 0,
("uma_zfree: Attempting to insert an empty bucket onto the full list.\n"));
LIST_INSERT_HEAD(&zone->uz_full_bucket,
bucket, ub_link);
}
if ((bucket = LIST_FIRST(&zone->uz_free_bucket)) != NULL) {
LIST_REMOVE(bucket, ub_link);
ZONE_UNLOCK(zone);
cache->uc_freebucket = bucket;
goto zfree_start;
}
/* We are no longer associated with this CPU. */
critical_exit();
/* And the zone.. */
ZONE_UNLOCK(zone);
#ifdef UMA_DEBUG_ALLOC
printf("uma_zfree: Allocating new free bucket.\n");
#endif
bflags = M_NOWAIT;
if (zone->uz_flags & UMA_ZFLAG_CACHEONLY)
bflags |= M_NOVM;
bucket = bucket_alloc(zone->uz_count, bflags);
if (bucket) {
ZONE_LOCK(zone);
LIST_INSERT_HEAD(&zone->uz_free_bucket,
bucket, ub_link);
ZONE_UNLOCK(zone);
goto zfree_restart;
}
/*
* If nothing else caught this, we'll just do an internal free.
*/
zfree_internal:
zone_free_item(zone, item, udata, SKIP_DTOR, ZFREE_STATFREE);
return;
}
/*
* Frees an item to an INTERNAL zone or allocates a free bucket
*
* Arguments:
* zone The zone to free to
* item The item we're freeing
* udata User supplied data for the dtor
* skip Skip dtors and finis
*/
static void
zone_free_item(uma_zone_t zone, void *item, void *udata,
enum zfreeskip skip, int flags)
{
uma_slab_t slab;
uma_slabrefcnt_t slabref;
uma_keg_t keg;
u_int8_t *mem;
u_int8_t freei;
int clearfull;
if (skip < SKIP_DTOR && zone->uz_dtor)
zone->uz_dtor(item, zone->uz_size, udata);
if (skip < SKIP_FINI && zone->uz_fini)
zone->uz_fini(item, zone->uz_size);
ZONE_LOCK(zone);
if (flags & ZFREE_STATFAIL)
zone->uz_fails++;
if (flags & ZFREE_STATFREE)
zone->uz_frees++;
if (!(zone->uz_flags & UMA_ZONE_VTOSLAB)) {
mem = (u_int8_t *)((unsigned long)item & (~UMA_SLAB_MASK));
keg = zone_first_keg(zone); /* Must only be one. */
if (zone->uz_flags & UMA_ZONE_HASH) {
slab = hash_sfind(&keg->uk_hash, mem);
} else {
mem += keg->uk_pgoff;
slab = (uma_slab_t)mem;
}
} else {
/* This prevents redundant lookups via free(). */
if ((zone->uz_flags & UMA_ZONE_MALLOC) && udata != NULL)
slab = (uma_slab_t)udata;
else
slab = vtoslab((vm_offset_t)item);
keg = slab->us_keg;
keg_relock(keg, zone);
}
MPASS(keg == slab->us_keg);
/* Do we need to remove from any lists? */
if (slab->us_freecount+1 == keg->uk_ipers) {
LIST_REMOVE(slab, us_link);
LIST_INSERT_HEAD(&keg->uk_free_slab, slab, us_link);
} else if (slab->us_freecount == 0) {
LIST_REMOVE(slab, us_link);
LIST_INSERT_HEAD(&keg->uk_part_slab, slab, us_link);
}
/* Slab management stuff */
freei = ((unsigned long)item - (unsigned long)slab->us_data)
/ keg->uk_rsize;
#ifdef INVARIANTS
if (!skip)
uma_dbg_free(zone, slab, item);
#endif
if (keg->uk_flags & UMA_ZONE_REFCNT) {
slabref = (uma_slabrefcnt_t)slab;
slabref->us_freelist[freei].us_item = slab->us_firstfree;
} else {
slab->us_freelist[freei].us_item = slab->us_firstfree;
}
slab->us_firstfree = freei;
slab->us_freecount++;
/* Zone statistics */
keg->uk_free++;
clearfull = 0;
if (keg->uk_flags & UMA_ZFLAG_FULL) {
if (keg->uk_pages < keg->uk_maxpages) {
keg->uk_flags &= ~UMA_ZFLAG_FULL;
clearfull = 1;
}
/*
* We can handle one more allocation. Since we're clearing ZFLAG_FULL,
* wake up all procs blocked on pages. This should be uncommon, so
* keeping this simple for now (rather than adding count of blocked
* threads etc).
*/
wakeup(keg);
}
if (clearfull) {
zone_relock(zone, keg);
zone->uz_flags &= ~UMA_ZFLAG_FULL;
wakeup(zone);
ZONE_UNLOCK(zone);
} else
KEG_UNLOCK(keg);
}
/* See uma.h */
void
uma_zone_set_max(uma_zone_t zone, int nitems)
{
uma_keg_t keg;
ZONE_LOCK(zone);
keg = zone_first_keg(zone);
keg->uk_maxpages = (nitems / keg->uk_ipers) * keg->uk_ppera;
if (keg->uk_maxpages * keg->uk_ipers < nitems)
keg->uk_maxpages += keg->uk_ppera;
ZONE_UNLOCK(zone);
}
/* See uma.h */
void
uma_zone_set_init(uma_zone_t zone, uma_init uminit)
{
uma_keg_t keg;
ZONE_LOCK(zone);
keg = zone_first_keg(zone);
KASSERT(keg->uk_pages == 0,
("uma_zone_set_init on non-empty keg"));
keg->uk_init = uminit;
ZONE_UNLOCK(zone);
}
/* See uma.h */
void
uma_zone_set_fini(uma_zone_t zone, uma_fini fini)
{
uma_keg_t keg;
ZONE_LOCK(zone);
keg = zone_first_keg(zone);
KASSERT(keg->uk_pages == 0,
("uma_zone_set_fini on non-empty keg"));
keg->uk_fini = fini;
ZONE_UNLOCK(zone);
}
/* See uma.h */
void
uma_zone_set_zinit(uma_zone_t zone, uma_init zinit)
{
ZONE_LOCK(zone);
KASSERT(zone_first_keg(zone)->uk_pages == 0,
("uma_zone_set_zinit on non-empty keg"));
zone->uz_init = zinit;
ZONE_UNLOCK(zone);
}
/* See uma.h */
void
uma_zone_set_zfini(uma_zone_t zone, uma_fini zfini)
{
ZONE_LOCK(zone);
KASSERT(zone_first_keg(zone)->uk_pages == 0,
("uma_zone_set_zfini on non-empty keg"));
zone->uz_fini = zfini;
ZONE_UNLOCK(zone);
}
/* See uma.h */
/* XXX uk_freef is not actually used with the zone locked */
void
uma_zone_set_freef(uma_zone_t zone, uma_free freef)
{
ZONE_LOCK(zone);
zone_first_keg(zone)->uk_freef = freef;
ZONE_UNLOCK(zone);
}
/* See uma.h */
/* XXX uk_allocf is not actually used with the zone locked */
void
uma_zone_set_allocf(uma_zone_t zone, uma_alloc allocf)
{
uma_keg_t keg;
ZONE_LOCK(zone);
keg = zone_first_keg(zone);
keg->uk_flags |= UMA_ZFLAG_PRIVALLOC;
keg->uk_allocf = allocf;
ZONE_UNLOCK(zone);
}
/* See uma.h */
int
uma_zone_set_obj(uma_zone_t zone, struct vm_object *obj, int count)
{
uma_keg_t keg;
vm_offset_t kva;
int pages;
keg = zone_first_keg(zone);
pages = count / keg->uk_ipers;
if (pages * keg->uk_ipers < count)
pages++;
kva = kmem_alloc_nofault(kernel_map, pages * UMA_SLAB_SIZE);
if (kva == 0)
return (0);
if (obj == NULL) {
obj = vm_object_allocate(OBJT_DEFAULT,
pages);
} else {
VM_OBJECT_LOCK_INIT(obj, "uma object");
_vm_object_allocate(OBJT_DEFAULT,
pages, obj);
}
ZONE_LOCK(zone);
keg->uk_kva = kva;
keg->uk_obj = obj;
keg->uk_maxpages = pages;
keg->uk_allocf = obj_alloc;
keg->uk_flags |= UMA_ZONE_NOFREE | UMA_ZFLAG_PRIVALLOC;
ZONE_UNLOCK(zone);
return (1);
}
/* See uma.h */
void
uma_prealloc(uma_zone_t zone, int items)
{
int slabs;
uma_slab_t slab;
uma_keg_t keg;
keg = zone_first_keg(zone);
ZONE_LOCK(zone);
slabs = items / keg->uk_ipers;
if (slabs * keg->uk_ipers < items)
slabs++;
while (slabs > 0) {
slab = keg_alloc_slab(keg, zone, M_WAITOK);
if (slab == NULL)
break;
MPASS(slab->us_keg == keg);
LIST_INSERT_HEAD(&keg->uk_free_slab, slab, us_link);
slabs--;
}
ZONE_UNLOCK(zone);
}
/* See uma.h */
u_int32_t *
uma_find_refcnt(uma_zone_t zone, void *item)
{
uma_slabrefcnt_t slabref;
uma_keg_t keg;
u_int32_t *refcnt;
int idx;
slabref = (uma_slabrefcnt_t)vtoslab((vm_offset_t)item &
(~UMA_SLAB_MASK));
keg = slabref->us_keg;
KASSERT(slabref != NULL && slabref->us_keg->uk_flags & UMA_ZONE_REFCNT,
("uma_find_refcnt(): zone possibly not UMA_ZONE_REFCNT"));
idx = ((unsigned long)item - (unsigned long)slabref->us_data)
/ keg->uk_rsize;
refcnt = &slabref->us_freelist[idx].us_refcnt;
return refcnt;
}
/* See uma.h */
void
uma_reclaim(void)
{
#ifdef UMA_DEBUG
printf("UMA: vm asked us to release pages!\n");
#endif
bucket_enable();
zone_foreach(zone_drain);
/*
* Some slabs may have been freed but this zone will be visited early
* we visit again so that we can free pages that are empty once other
* zones are drained. We have to do the same for buckets.
*/
zone_drain(slabzone);
zone_drain(slabrefzone);
bucket_zone_drain();
}
/* See uma.h */
int
uma_zone_exhausted(uma_zone_t zone)
{
int full;
ZONE_LOCK(zone);
full = (zone->uz_flags & UMA_ZFLAG_FULL);
ZONE_UNLOCK(zone);
return (full);
}
int
uma_zone_exhausted_nolock(uma_zone_t zone)
{
return (zone->uz_flags & UMA_ZFLAG_FULL);
}
void *
uma_large_malloc(int size, int wait)
{
void *mem;
uma_slab_t slab;
u_int8_t flags;
slab = zone_alloc_item(slabzone, NULL, wait);
if (slab == NULL)
return (NULL);
mem = page_alloc(NULL, size, &flags, wait);
if (mem) {
vsetslab((vm_offset_t)mem, slab);
slab->us_data = mem;
slab->us_flags = flags | UMA_SLAB_MALLOC;
slab->us_size = size;
} else {
zone_free_item(slabzone, slab, NULL, SKIP_NONE,
ZFREE_STATFAIL | ZFREE_STATFREE);
}
return (mem);
}
void
uma_large_free(uma_slab_t slab)
{
vsetobj((vm_offset_t)slab->us_data, kmem_object);
page_free(slab->us_data, slab->us_size, slab->us_flags);
zone_free_item(slabzone, slab, NULL, SKIP_NONE, ZFREE_STATFREE);
}
void
uma_print_stats(void)
{
zone_foreach(uma_print_zone);
}
static void
slab_print(uma_slab_t slab)
{
printf("slab: keg %p, data %p, freecount %d, firstfree %d\n",
slab->us_keg, slab->us_data, slab->us_freecount,
slab->us_firstfree);
}
static void
cache_print(uma_cache_t cache)
{
printf("alloc: %p(%d), free: %p(%d)\n",
cache->uc_allocbucket,
cache->uc_allocbucket?cache->uc_allocbucket->ub_cnt:0,
cache->uc_freebucket,
cache->uc_freebucket?cache->uc_freebucket->ub_cnt:0);
}
static void
uma_print_keg(uma_keg_t keg)
{
uma_slab_t slab;
printf("keg: %s(%p) size %d(%d) flags %d ipers %d ppera %d "
"out %d free %d limit %d\n",
keg->uk_name, keg, keg->uk_size, keg->uk_rsize, keg->uk_flags,
keg->uk_ipers, keg->uk_ppera,
(keg->uk_ipers * keg->uk_pages) - keg->uk_free, keg->uk_free,
(keg->uk_maxpages / keg->uk_ppera) * keg->uk_ipers);
printf("Part slabs:\n");
LIST_FOREACH(slab, &keg->uk_part_slab, us_link)
slab_print(slab);
printf("Free slabs:\n");
LIST_FOREACH(slab, &keg->uk_free_slab, us_link)
slab_print(slab);
printf("Full slabs:\n");
LIST_FOREACH(slab, &keg->uk_full_slab, us_link)
slab_print(slab);
}
void
uma_print_zone(uma_zone_t zone)
{
uma_cache_t cache;
uma_klink_t kl;
int i;
printf("zone: %s(%p) size %d flags %d\n",
zone->uz_name, zone, zone->uz_size, zone->uz_flags);
LIST_FOREACH(kl, &zone->uz_kegs, kl_link)
uma_print_keg(kl->kl_keg);
for (i = 0; i <= mp_maxid; i++) {
if (CPU_ABSENT(i))
continue;
cache = &zone->uz_cpu[i];
printf("CPU %d Cache:\n", i);
cache_print(cache);
}
}
#ifdef DDB
/*
* Generate statistics across both the zone and its per-cpu cache's. Return
* desired statistics if the pointer is non-NULL for that statistic.
*
* Note: does not update the zone statistics, as it can't safely clear the
* per-CPU cache statistic.
*
* XXXRW: Following the uc_allocbucket and uc_freebucket pointers here isn't
* safe from off-CPU; we should modify the caches to track this information
* directly so that we don't have to.
*/
static void
uma_zone_sumstat(uma_zone_t z, int *cachefreep, u_int64_t *allocsp,
u_int64_t *freesp)
{
uma_cache_t cache;
u_int64_t allocs, frees;
int cachefree, cpu;
allocs = frees = 0;
cachefree = 0;
for (cpu = 0; cpu <= mp_maxid; cpu++) {
if (CPU_ABSENT(cpu))
continue;
cache = &z->uz_cpu[cpu];
if (cache->uc_allocbucket != NULL)
cachefree += cache->uc_allocbucket->ub_cnt;
if (cache->uc_freebucket != NULL)
cachefree += cache->uc_freebucket->ub_cnt;
allocs += cache->uc_allocs;
frees += cache->uc_frees;
}
allocs += z->uz_allocs;
frees += z->uz_frees;
if (cachefreep != NULL)
*cachefreep = cachefree;
if (allocsp != NULL)
*allocsp = allocs;
if (freesp != NULL)
*freesp = frees;
}
#endif /* DDB */
static int
sysctl_vm_zone_count(SYSCTL_HANDLER_ARGS)
{
uma_keg_t kz;
uma_zone_t z;
int count;
count = 0;
mtx_lock(&uma_mtx);
LIST_FOREACH(kz, &uma_kegs, uk_link) {
LIST_FOREACH(z, &kz->uk_zones, uz_link)
count++;
}
mtx_unlock(&uma_mtx);
return (sysctl_handle_int(oidp, &count, 0, req));
}
static int
sysctl_vm_zone_stats(SYSCTL_HANDLER_ARGS)
{
struct uma_stream_header ush;
struct uma_type_header uth;
struct uma_percpu_stat ups;
uma_bucket_t bucket;
struct sbuf sbuf;
uma_cache_t cache;
uma_klink_t kl;
uma_keg_t kz;
uma_zone_t z;
uma_keg_t k;
char *buffer;
int buflen, count, error, i;
mtx_lock(&uma_mtx);
restart:
mtx_assert(&uma_mtx, MA_OWNED);
count = 0;
LIST_FOREACH(kz, &uma_kegs, uk_link) {
LIST_FOREACH(z, &kz->uk_zones, uz_link)
count++;
}
mtx_unlock(&uma_mtx);
buflen = sizeof(ush) + count * (sizeof(uth) + sizeof(ups) *
(mp_maxid + 1)) + 1;
buffer = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
mtx_lock(&uma_mtx);
i = 0;
LIST_FOREACH(kz, &uma_kegs, uk_link) {
LIST_FOREACH(z, &kz->uk_zones, uz_link)
i++;
}
if (i > count) {
free(buffer, M_TEMP);
goto restart;
}
count = i;
sbuf_new(&sbuf, buffer, buflen, SBUF_FIXEDLEN);
/*
* Insert stream header.
*/
bzero(&ush, sizeof(ush));
ush.ush_version = UMA_STREAM_VERSION;
ush.ush_maxcpus = (mp_maxid + 1);
ush.ush_count = count;
if (sbuf_bcat(&sbuf, &ush, sizeof(ush)) < 0) {
mtx_unlock(&uma_mtx);
error = ENOMEM;
goto out;
}
LIST_FOREACH(kz, &uma_kegs, uk_link) {
LIST_FOREACH(z, &kz->uk_zones, uz_link) {
bzero(&uth, sizeof(uth));
ZONE_LOCK(z);
strlcpy(uth.uth_name, z->uz_name, UTH_MAX_NAME);
uth.uth_align = kz->uk_align;
uth.uth_size = kz->uk_size;
uth.uth_rsize = kz->uk_rsize;
LIST_FOREACH(kl, &z->uz_kegs, kl_link) {
k = kl->kl_keg;
uth.uth_maxpages += k->uk_maxpages;
uth.uth_pages += k->uk_pages;
uth.uth_keg_free += k->uk_free;
uth.uth_limit = (k->uk_maxpages / k->uk_ppera)
* k->uk_ipers;
}
/*
* A zone is secondary is it is not the first entry
* on the keg's zone list.
*/
if ((z->uz_flags & UMA_ZONE_SECONDARY) &&
(LIST_FIRST(&kz->uk_zones) != z))
uth.uth_zone_flags = UTH_ZONE_SECONDARY;
LIST_FOREACH(bucket, &z->uz_full_bucket, ub_link)
uth.uth_zone_free += bucket->ub_cnt;
uth.uth_allocs = z->uz_allocs;
uth.uth_frees = z->uz_frees;
uth.uth_fails = z->uz_fails;
if (sbuf_bcat(&sbuf, &uth, sizeof(uth)) < 0) {
ZONE_UNLOCK(z);
mtx_unlock(&uma_mtx);
error = ENOMEM;
goto out;
}
/*
* While it is not normally safe to access the cache
* bucket pointers while not on the CPU that owns the
* cache, we only allow the pointers to be exchanged
* without the zone lock held, not invalidated, so
* accept the possible race associated with bucket
* exchange during monitoring.
*/
for (i = 0; i < (mp_maxid + 1); i++) {
bzero(&ups, sizeof(ups));
if (kz->uk_flags & UMA_ZFLAG_INTERNAL)
goto skip;
if (CPU_ABSENT(i))
goto skip;
cache = &z->uz_cpu[i];
if (cache->uc_allocbucket != NULL)
ups.ups_cache_free +=
cache->uc_allocbucket->ub_cnt;
if (cache->uc_freebucket != NULL)
ups.ups_cache_free +=
cache->uc_freebucket->ub_cnt;
ups.ups_allocs = cache->uc_allocs;
ups.ups_frees = cache->uc_frees;
skip:
if (sbuf_bcat(&sbuf, &ups, sizeof(ups)) < 0) {
ZONE_UNLOCK(z);
mtx_unlock(&uma_mtx);
error = ENOMEM;
goto out;
}
}
ZONE_UNLOCK(z);
}
}
mtx_unlock(&uma_mtx);
sbuf_finish(&sbuf);
error = SYSCTL_OUT(req, sbuf_data(&sbuf), sbuf_len(&sbuf));
out:
free(buffer, M_TEMP);
return (error);
}
#ifdef DDB
DB_SHOW_COMMAND(uma, db_show_uma)
{
u_int64_t allocs, frees;
uma_bucket_t bucket;
uma_keg_t kz;
uma_zone_t z;
int cachefree;
db_printf("%18s %8s %8s %8s %12s\n", "Zone", "Size", "Used", "Free",
"Requests");
LIST_FOREACH(kz, &uma_kegs, uk_link) {
LIST_FOREACH(z, &kz->uk_zones, uz_link) {
if (kz->uk_flags & UMA_ZFLAG_INTERNAL) {
allocs = z->uz_allocs;
frees = z->uz_frees;
cachefree = 0;
} else
uma_zone_sumstat(z, &cachefree, &allocs,
&frees);
if (!((z->uz_flags & UMA_ZONE_SECONDARY) &&
(LIST_FIRST(&kz->uk_zones) != z)))
cachefree += kz->uk_free;
LIST_FOREACH(bucket, &z->uz_full_bucket, ub_link)
cachefree += bucket->ub_cnt;
db_printf("%18s %8ju %8jd %8d %12ju\n", z->uz_name,
(uintmax_t)kz->uk_size,
(intmax_t)(allocs - frees), cachefree,
(uintmax_t)allocs);
}
}
}
#endif
|